1 10 6 https://cpw.cvlcollections.org/files/original/3adca87aaf9abe042e2fc4c4271ad14c.pdf 7fad827db0996ca75568cd5ca84d0d9f PDF Text Text 1 Colorado Division of Wildlife Wildlife Research Report July 1998 JOB PROGRESS REPORT state of Colorado cost center 3430 Project No. W-153-R-11 Mammals Program Work Package No. __0_6_6-2______ Preble's Meadow Jumping Mouse conservation Task No. : Preble's Meadow Jumping Mouse conservation Period Covered: Author: July 1, 1997 - June 30, 1998 Tanya Shenk ABSTRACT A draft conservation plan, entitled 'Conservation Assessment and Preliminary Conservation Strategy for Preble's Meadow Jumping Mouse (Zapus hudsonius preblei)' was completed and submitted to the .USFWS on January 9, 1998 during the Open Comment Period for the Proposed Listing of Preble's meadow jumping mouse as endangered. The conservation plan was divided into two parts. The first part, a conservation assessment summarized and evaluated available information on the taxonomy, distribution, and ecology of z. h. preblei and identified potential threats to the conservation of the mouse. The second part, a preliminary conservation strategy outlines the goals and objectives of a conservation strategy for z. h. preblei and summarized prioritized research needs to provide necessary information to develop sound conservation strategies. From this list of prioritized research needs two study plans were completed and research begun on both starting in May 1998. These research plans are designed to investigate (1) temporal and spatial variation in the demography of Preble's meadow jumping mouse and (2) habitat use and distribution of Preble's meadow jumping mouse in Larimer and Weld Counties, Colorado. ��3 Preble's Meadow Jumping Mouse Conservation Plan Tanya Shenk P.H. OBJECTIVE Develop and implement a conservation plan for Preble's meadow jumping mouse in Colorado. SEGMENT OBJECTIVES 1. Complete a literature search for all known, relevant information to develop a conservation plan for Preble's meadow jumping mouse. 2. Complete an inventory of current research being conducted on Preble's meadow jumping mouse. 3. Evaluate known information on Preble's meadow jumping mouse in the context of providing information necessary to develop sound strategies for conservation of Preble's meadow jumping mouse. 4. Define and prioritize needed research components to develop sound strategies for conservation of Preble's meadow jumping mouse. s. Begin study design for research to address a high priority need to develop sound strategies for conservation of Preble's meadow jumping mouse. 6. Complete a draft conservation plan for Preble's meadow jumping mouse. RESULTS AND DISCUSSIQH 1. Complete a literature search for all known, relevant information to develop a conservation plan for Preble's meadow jumping mouse. A literature search was completed, the information was summarized and presented in the draft conservation plan (see Appendix A). 2. Complete an inventory of current research being conducted on Preble's meadow jumping mouse. There are four research projects being conducted on z. h. preblei, with a fifth planned to begin spring of 1998 (Table 1). Each of these projects is designed to provide further information on the ecology or demography of Preble's meadow jumping mouse. However, if research methodologies were standardized and coordinated across all the projects, comparability and quality of the data would be greatly enhanced. Such a coordinated effort would maximize data quantity, quality, and comparability over varying conditions throughout the range of z. h. preblei in the most effective and efficient way possible. Comparable information gained across the ecological range of z. h. preblei would then provide more useful information to develop sound management strategies for conservation of the mouse. To achieve such a coordinated research effort all project leaders must agree to follow data collection protocols, establish 'ownership' of data and future publications, and work cooperatively to share equipment and technical assistance during peak data collection periods. To facilitate such a mutual effort will require the participation of all �4 project leaders conducting field studies of z. h. preblei as well as cooperative investigators, and specialized data analysts with expertise in the study design, analysis, and interpretation of mark-recapture and telemetry data. Therefore, I organized a Preble's Meadow Jumping Mouse Research Working Group with the following objective: To explore the idea of coordination of research efforts across multiple principal investigators to enhance comparability and quality of data across the range of z. h. preblei for use in developing sound management strategies for conservation of the mouse. Such an effort would maximize data qu~ntity, quality, and comparability over varying conditions throughout the range of z. h. preblei in the most effective and efficient way possible. To make this work, all principle investigators must agree to follow data collection protocols, establish 'ownership' of data and future publications, and work cooperatively on the possible sharing of equipment and technical assistance during peak data collection periods. This working group needs to involve all interested principle investigators involved in field studies of z. h. preblei as well as cooperative investigators, and specialized data analysts with expertise in the study design, analysis, and interpretation of mark-recapture and telemetry data. The first meeting of the working group (November 12, 1997) was a success with all principle investigators supportive of the idea of a coordinated, cooperative research program. The second meeting of the Research Working Group is scheduled for February 4, 1998 where we will outline what research questions will be addressed at which sites and draft standardized protocols for data collection. A proposed agenda for the Research Working Group includes the following tasks, task leaders, and a timetable to initiate a cooperative research effort: Task Task Leader, Affiliation Date Identify all potential participants: project leaders, cooperating investigators, data analysts. Prioritize research questions. T. Shenk, Colorado Division of Wildlife January 1998 All project leaders* Identify sites to best address prioritized research needs. Design studies specific to research question to be addressed at each site. All project leaders January 1998 January 1998 JanuaryFebruary 1998 All project leaders, cooperating investigators, data analysts All project leaders March 1998 Evaluate needs at each site to conduct specified research. March 1998 Identify discrepancies between All project leaders needs and available funds, equipment, and personnel currently allocated for each site. Attempt to balance discrepancies. March 1998 All project leaders, cooperating investigators * To date: M. Bakeman, c. Meaney, T. Ryon, R. Schorr, T. Shenk �5 3. Evaluate known information on Preble's meadow jumping mouse in the context of providing information necessary to develop sound strategies for conservation of Preble's meadow jumping mouse. Known information on the ecology of Preble's meadow jumping mouse was summarized and presented in the Conservation Plan (see Appendix A). 4. Define and prioritize needed research components to develop sound strategies for conservation of Preble's meadow jumping mouse. The following outline is a complete list of research needs to provide information to develop sound management strategies for conservation of z. h. preblei. Research needs are not listed in order of priority, but listed in a logical manner to provide completion of needs. Demographic studies: Information on the population dynamics of Preble's meadow jumping mouse is necessary to determine which areas support populations where the rate of population growth is~ 1. Key parameters to estimate include: Survival: • estimates of survival, including ► annual ► over-summer ► over-hibernation • investigate possible factors affecting survival, including ► weight sex age abundance (i.e., density dependent response) habitat features: stream reach, vegetation composition weather ► predation ► disease Approach: mark-recapture techniques ► ► ► ► ► Recruitment: • estimate recruitment (as an alternative to reproductive parameters listed below) • investigate possible factors affecting recruitment, including ► weather ► habitat features Approach: mark-recapture techniques Population structure: • estimate sex ratios • estimate age ratios Approach: mark-recapture techniques Abundance: • estimate abundance investigate factors affecting abundance, including ► habitat features (see under habitat use) Approach: mark-recapture techniques for closed populations • Immigration, emigration: • estimate rates of immigration and emigration �6 • investigate possible factors affecting immigration, emigration habitat features (see under habitat use) ► abundance (i.e., density-dependent response) Approach: estimate rates from mark-recapture data, identify existence with radio telemetry ► Reproduction: • • • • • estimate number of litters per year estimate number of young per litter estimate age at first reproduction estimate juvenile survival investigate possible factors affecting reproduction, including ► habitat features: food availability, nest site availability, cover availability ► abundance (i.e., density-dependent response) ► weather Approach: from telemetry work Dispersal is a key process in metapopulation theory and to maintain genetic diversity between isolated populations. Key parameters to evaluate include: Dispersal Studies: Population parameters • who disperses • time of dispersal • estimate rate Approach: estimate rate with mark-recapture, document who and when from both telemetry and mark-recapture data Habitat parameters • • • through what habitat end point descriptions (disperses from what to what) landscape features (connectivity with other riparian strips, corridor use, overland use) Approach: document where from both telemetry and mark-recapture data Conservation of z. h. preblei should maintain populations throughout the range of its natural variation and to try to identify ecological limits for the subspecies. Key concerns include: Distribution Studies: Range-wide distribution: • conduct trapping surveys to better define the eastern boundary of the range of z. h. preblei • conduct trapping surveys in areas of potential sympatry of z. h. preblei with z. princeps Approach: determine from trapping surveys Habitat Studies: To identify and define habitat requirements of z. h. preblei studies should be conducted to address the following: Habitat use: • • • estimate distances traveled: daily, seasonally describe landscape features (connectivity with other potential sites, geology) determine seasonal use �7 • • • • • describe hibernacula describe nest sites estimate distance to nearest open water from other habitats used describe hydrology (water quality, flow) in areas of use evaluate effects of abundance on habitat used (i.e., density dependent responses) Approach: estimate from telemetry work Physiological Studies: Physiological studies will provide information on the mechanisms driving habitat selection. Physiological requirements: • estimate dependency of z. h. preblei on open water • determine energetic requirements to survive hibernation Approach: estimate from laboratory studies Systematic Studies: To better define the relationship of z. h. preblei to other subspecies of z. hudsonius and other species of Zapus the following studies should continue. Molecular systematic relationships: • further explore genetic relationships among different z. h. preblei populations • further explore genetic relationships among different subspecies of z. hudsonius • further explore genetic relationships among different species of Zapus. Approach: explore through laboratory studies Systematic relationships: • link genetic relationships to systematic studies of z. hudsonius. Approach: explore through museum studies Community Studies: Composition of the community where z. h. preblei occur could help explain ecological tolerances of the subspecies, providing insight to the mechanisms determining its distribution. Small mammal assemblages: • comparison of small mammal assemblages in areas where populations of z. h. preblei occur and areas where they do not ► species composition ► relative abundance Approach: estimate from trapping surveys There are currently five research projects on z. h. preblei planned to begin spring of 1998. Each of these projects is designed to provide further information on the ecology or demography of Preble's meadow jumping mouse. However, if research methodologies were standardized and coordinated across all the projects, comparability and quality of the data would be greatly enhanced. Such a coordinated effort would maximize data quantity, quality, and comparability over varying conditions throughout the range of z. h. preblei in the most effective and efficient way possible. Comparable information gained across the ecological range of z. h. preblei would then provide more useful information for use in developing sound management strategies for �8 conservation of the mouse. Therefore, a research group was formed including all principal investigators of the five studies as well as other interested personnel. This research group has developed standardized protocols to be followed in all studies. These include protocols on techniques, data collection, and data analyses for (1) mark-recapture, (2) radio-telemetry, (3) habitat use, and (4) genetic sampling. 5. Begin study design for research co address a high priority need co develop sound strategies for conservation of Preble's meadow jumping mouse. There are four components of z. h. preblei ecology that are currently unknown and yet key to any sound conservation strategy for the subspecies. These are (1) detailed demographic studies estimating survival and reproduction and determining the factors influencing each of the parameters, (2) detailed studies evaluating movements and dispersal habitat of individuals within and among populations, (3) detailed studies to define hibernation needs, primarily descriptions of suitable hibernacula criteria and food requirements for sufficient fat storage prior to immergence, and (4) better defined distributions of the mouse. The following describes the research I will be conducting beginning Spring 1998 to address these needs. Demography Study: Information on the population dynamics of Preble's meadow jumping mouse is necessary to determine which areas support populations where the rate of population growth is stable or suggests an increasing population. Key parameters to estimate include survival (annual, over-summer, and over-hibernation), reproduction, dispersal rates, and density. A combination of mark-recapture techniques and radio-telemetry studies will be used to estimate these parameters and evaluate possible factors affecting such parameters. Intensive trapping efforts will be conducted in June and August of 1998 and in May of 1999 within a given population. All animals captured will be permanently marked with PIT (passive integrated transponders) tags. Select mice will also be fitted with radio transmitters. Survival estimates and evaluation of possible factors affecting survival, including weight, sex, age, abundance (i.e., density dependent response), habitat features (stream reach, vegetational composition), and weather will be obtained from data collected from mark-recapture efforts. Dispersal is a key process in meta-population theory and to maintain genetic diversity between isolated populations. The key objectives are to estimate rate of dispersal and to determine who disperses, time of dispersal, and describe suitable dispersal habitat. These estimates will be obtained from analyses of both the markrecapture data and from following radio-collared individuals. Estimates of immigration and emigration rates will be obtained from data collected during the mark-recapture efforts as well as an evaluation of possible factors affecting these rates including habitat features and abundance (i.e., density-dependent response). Data on who disperses, when dispersal occurs, and dispersal habitat will be obtained by following radio-collared animals. From the mark-recapture data collected during the trapping sessions, abundance and density estimates can also be made. Other ancillary information which can be obtained from data collected during �9 the tapping sessions include estimates of sex and age ratios within the populations sampled. Reproduction parameters such as number of litters per year, number of young per litter, and age at first reproduction will be estimated from radio-collared females. An analysis of variance will be conducted to evaluate possible factors affecting reproduction, including habitat features such as food availability, nest site availability, and cover availability as well as the possible effects of abundance (i.e., density-dependent response) and weather. A complete study plan entitled "Temporal and spatial variation in the demography of Preble's meadow jumping mouse (Zapus hudsonius preblei) is included in Appendix B. Distribution, Habitat Use, and Monitoring Study: Conservation of z. h. preblei should maintain populations throughout the range of its natural variation and to try to identify ecological limits for the subspecies. Key concerns include range-wide distribution, habitat use, and population persistence at a given site. To address these concerns a study will be designed where a sampling frame of suitable habitat will be constructed. The starting point for developing the sampling frame will be to determine the feasible distributional range of z. h. preblei. A conservative approach will be used for elevational and directional limitations. Once the potential distribution map is developed, all but potentially suitable habitat within the feasible range of the subspecies will be eliminated. From this sampling frame, random sites will be selected to conduct trapping surveys. Therefore, all successful trapping sites will provide further information on the distribution of the mouse. Information on habitat will be collected at all sites surveyed, regardless of trapping success for Preble's meadow jumping mouse. Comparisons of habitat variables from both successful and unsuccessful sites will be used to better define suitable habitat for the subspecies. Habitat variables recorded will include both site level characteristics such as vegetational composition, cover, and composition of ·other small mammal species·as well as landscape level characteristics including connectivity with other potential sites, geology, hydrology, and distance to development. Data collected on movements of radiotelemetered mice will also provide information on seasonal use of different habitats, dispersal corridor and end point habitat characteristics, and descriptions of nest sites and hibernacula. Repeated annual visits to the randomly selected sites will provide information on population persistence at a given site. Such a monitoring scheme will be necessary for evaluating the continued status of the mouse as well potentially evaluating the efficacy of any conservation planning efforts. Genetic tissue samples will also be collected from mice captured at these new locations. Future genetic analyses should help to better define the relationship among (1) different populations of z. h. preblei, (2) other subspecies of z. hudsonius and (3) other species of Zapus. A complete study plan entitled "Habitat use and distribution of Preble"s meadow jumping mouse (Zapus hudsonius preblei) in Larimer and Weld Counties, Colorado" is included in Appendix C. 6. Complete a draft conservation plan for Preble's meadow jumping mouse. The draft conservation plan, entitled 'Conservation Assessment �10 and Preliminary Conservation Strategy for Preble's Meadow Jumping Mouse (Zapus hudsonius preblei)' was completed and is attached as Appendix A. The document was submitted to the USFWS during the Open Comment Period for the Proposed Listing of Preble's meadow jumping mouse as endangered. Prepared by ~---_5/J_ ~ a Shenk �Table 1. Current and proposed research projects for Preble's meadow jumping mouse. Marking method Study location Principle Investigate r Pit tags Parameters to estimate Telemetr Surviva Reproducti Recruitme Abundanc Diapers Habita Populatio Stomach t Uset on nt y 1 e al n content structure analyses Rocky Flats Tom Ryon X X X X X AOA* Air Force Academy Rob Schorr X X X X X AOA X X AOA X X AOA Boulder Open Carron Meaney Space Lyons X Mark Bakeman Cherokee Park? Tanya Shenk X X X X X X X X AOA Douglas Cty (Plum Creek)? Tanya Shenk X X X X X X X X AOA * AOA As opportunity arises I-' I-' ��13 APPENDIX A CONSERVATION ASSESS:MENT AND PRELIMINARY CONSERVATION STRATEGY FOR PREBLE'S MEADOW JUMPING MOUSE (Zapus hudsonius preblei) prepared by Tanya Shenk Colorado Division of Wildlife 317 West Prospect Fort Collins, Colorado 80526 January 1998 �14 EXECUTIVE SUMMARY In 1994 a petition was submitted to the U. S. Fish and Wildlife Service from the Biodiversity Legal Foundation (USFWS 1997a) to list Preble's meadow jumping mouse (Zapus hudsonius preblei) under the Endangered Species Act (ESA). A Proposed Rule to list Preble's meadow jumping mouse as endangered under the ESA was submitted by the U. S. Fish and Wildlife Service on March 25, 1997 (USFWS 1997a). Following the Proposed Rule to list the species as endangered, the Colorado Division of Wildlife agreed to develop a species conservation assessment and preliminary conservation strategy which would provide a framework for conservation efforts. This document is the result of that agreement between the Colorado Division of Wildlife and The U.S. Fish and Wildlife Service. Therefore, the purpose of this document is to summarize the current known ecology and status of Z. h. preblei, to outline goals for conservation of the subspecies, to prioritize what information is most needed to develop sound conservation strategies to meet those goals, and to suggest future research required to provide such information. This document will provide the Colorado Division of Wildlife with a prioritized list of research needs to direct ecologists and managers toward obtaining the necessary information for developing, implementing, and evaluating strategies for conserving the Preble's meadow jumping mouse. Should the Final Decision list the species as either threatened or endangered, this document could also provide scientific information for any Habitat Conservation Plan(s) to be developed under the ESA. Any conservation plan for Z. h. preblei should address what is needed for recovery of the subspecies, including how much habitat is needed and where, what information and conservation actions are needed, whether restorations are needed, and what biological information is needed to manage or conserve the mouse. Some of the basic information required to define these needs includes the distribution of the mouse and its habitat, population dynamics (survival, reproduction, and dispersal), minimal habitat requirements, genetic variation between and among populations, physiology, hibernation requirements, and resilience of populations to human alteration of their habitat. This document attempts to summarize what is currently known on each of these topics, prioritizes what information is still most needed to develop conservation strategies based on the conservation goals, and suggests future research required to provide missing information. A review of the studies conducted on Preble's meadow jumping mouse shows that there is insufficient information to fully address defining range-wide ecological requirements, limiting factors, limits of species tolerance, or population status. Most work to date has focused on geographic distribution (presence or absence of Z. h. preb/ei), taxonomy, and habitat descriptions of sites where mice have and have not been captured. Although available information is limited, it is clear that measures must be taken to begin developing a conservation strategy for Z. h. preblei. The primary objectives necessary to develop a successful conservation strategy for Preble's meadow jumping mouse are: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. Document the present distribution of Z. h. preblei. Identify populations of Z. h. preblei where the rate of population growth ~ 1. Protect populations of Z. h. preblei where the rate of population growth~ 1. Maintain current range of natural variability of Z. h. preblei. Identify ecological requirements for sustaining viable populations of Z. h. preblei throughout its range of natural variability. Protect habitats to sustain existing populations of Z. h. preblei where the rate of population growth~. 1. Promote protection and management of habitat for conservation of Z. h. preblei in all currently or recently occupied habitat, and in habitat suitable for restoration of mouse populations. Monitor the status of populations of Z. h. preblei throughout its known range to detect changes in local distribution. Identify threats to the conservation of Z. h. preblei. Eliminate or minimize threats to conservation of Z. h. preblei. �15 11. 12. 13. Integrate Preble's meadow jumping mouse conservation strategy objectives with management and habitat objectives of other Front Range riparian species. Promote scientific management of Preble's meadow jumping mouse. Promote public support for conservation efforts and scientific management of Preble's meadow jumping mouse through public education. In order to achieve the stated objectives outlined above, the following actions must occur: 1. 2. 3. 4. 5. 6. 7. 8. 9. Conduct population studies to estimate and determine what factors influence demographic parameters and rate of population change. Conduct studies to better define the ecological requirements of Z. h. preblei (food, cover, water, hibernation requirements, etc.). Conduct research to evaluate the effects of potential threats on survival, reproduction, dispersal, and abundance of Preble's meadow jumping mouse. Implement habitat and population management practices that emphasize the conservation of Preble's meadow jumping mouse throughout the natural variability of its range. Develop survey protocols to monitor trends in the distribution of Z. h. preblei and protocols to store and analyze survey data. Conduct trapping surveys to better define the boundaries of the range of Z h. preblei. Implement research on Preble's meadow jumping mouse biology to better understand the mechanisms driving the ecological requirements. Further explore genetic relationships among different Z. h. preblei populations, among different subspecies of Z. hudsonius, and among different species of Zapus. Develop educational programs to promote positive public support for Preble's meadow jumping mouse conservation. The criterion to be used to evaluate the success of this program will be the maintenance of local selfsustaining populations of Preble's meadow jumping mouse which are geographically distributed throughout the range of the subspecies and throughout the natural variation of its ecological requirements. The document is divided into two parts. The first part, a conservation assessment will ( 1) summarize available information on the taxonomy, distribution, and ecology of Z. h. preblei; and (2) identify potential threats to the conservation of the mouse. The second part, a preliminmy conservation strategy, will (1) outline the goals and objectives of a conservation strategy for Z. h. preblei; and (2) summarize prioritized research needs to provide necessmy information to develop sound conservation strategies. ACKNOWLEDGMENTS This document could not have been prepared without generous help from all members of the 'Z. h. preblei Conservation Document Working Group' (Mark E. Bakeman, Carron A. Meaney, Chris Pague, and Peter Plage). Members gave freely of their time and expertise and readily shared their data, anecdotal observations, impressions, and scientific interpretations. This working group truly exemplified how scientists should interact when working toward a common goal. Additional scientific contributions came from David M. Armstrong, Alan B. Franklin, Lawrence A. Riggs, Thomas R Ryon, Robert Schorr, Rick Schroeder, and Bruce Wunder. Technical support during this effort was provided by Alan B. Franklin, David Lovell, Francie Pusatari, and Michael Wunder. Logistical support was provided by Steven Nesta, Douglas Robotham, Judy Sheppard, and John Stover. �16 CONSERVATION ASSESSMENT INTRODUCTION The following is a summary of information collected primarily from the five studies conducted on Z. h. preb/ei. The longest and most varied studies have been conducted on the Rocky Flats Environmental Technology Site (I 990 through the present) including distribution studies, habitat surveys, location ofhibernacula, fat requirements for hibernation, and genetic studies conducted by M. Bakeman, A. Deans, F. Harrington, T. Ryon, R Stoecker, and B. Wunder. Range-wide distributional and vegetation surveys have been conducted on sites other than Rocky Flats Environmental Technology Site by (1) Meaney et al. (1995, 1996, 1997), funded by the Colorado Division of Wildlife, (2) City of Boulder Open Space and Boulder County Open Space, (3) Ensight Technical Services of Boulder, Colorado funded by the Colorado Department of Transportation, (4) the Colorado Natural Heritage Program, (5) U. S. Forest Service, Medicine Bow National Forest, Wyoming, and (6) numerous other environmental consulting firms. An evaluation of historical Z. h. preblei sites was conducted by Ryon (1996). Genetics studies were conducted in 1995 by B. Wunder and in 1996 and 1997 by Biosphere Genetics, Inc. Where data were not available for Z. h. preblei, information from studies of other subspecies of Z. hudsonius was provided. In particular, studies conducted by Quimby (1951) and Whitaker (1963) provided substantial information on the species. Caution must be used in extrapolating information reported for other subspecies of Z. hudsonius to Z. h. preblei, although, such information is useful as a general framework from which to work until such information becomes available for Z. h. preblei itself Possible threats to the conservation of Preble's meadow jumping mouse were derived from both general principles of conservation biology and information specific to Zapus hudsonius in general, or the subspecies itself TAXONOMY Description Quimby (1951) described Z. hudsonius as follows: 'A mouse-like rodent with greatly enlarged hind feet and an exceptionally long tail. The forelegs are relatively short. The ears are somewhat conspicuous. The body is clothed in moderately long, somewhat dense hair of a rather coarse texture and several colors. The dorsal portions are marked by a broad stripe of brownish hairs many of which are tipped with black giving the region a grayish-black appearance. The sides are bright yellowishorange, whereas the underparts and feet are white. The tail is bicolor, dark above and light below, and sparsely covered with hair which is longer on the terminal part. The mammae are eight, and quite prominent in lactating females. The male genitalia are inconspicuous except during the breeding season when the scrotal sac becomes enlarged. The testes enlarge and may be either abdominal, inguinal, or scrotal during this period.' The skull of Z. hudsonius is small and light with a narrower braincase and smaller molars than in Z. princeps (from Fitzgerald et al. 1994 p. 18). However, Fitzgerald et al. (1994) urge caution in distinguishing the two species ·or Zapus in Colorado. Such caution is further warranted by the recent conflicting identifications of mice based on genetic characteristics (see Genetics section below). E. A. Preble (1899) first collected the specimen that is, now, the holotype for Z. h. preblei at Loveland, Colorado in 1895 and assigned it taxonomically to Z. h. campestris. Krutzsch (1954) revised the taxonomy of Z. hudsonius and assigned the meadow jumping mice from Colorado and southeastern Wyoming to their own distinct subspecies. Krutzch (1954) described this subspecies as having fewer black-tipped dorsal hairs and a less distinct dorsal band; smaller cranial measurements; a narrower interorbital constriction; smaller, less inflated auditory bullae; narrower incisive foramina; and a more inflated frontal region (Fig. 1) than Z. h. campestris. Krutzch named this subspecies Z. h. preblei in honor of the collector of the holotype (and previous reviewer of genus). �17 Figure 1. Preble's meadow jumping mouse (Z. h. preblei). Note the long tail curves around handlers index finger and the dark dorsal stripe. Also note the large hind foot and small first digits on front and rear feet (insert). Photographs by Norm Clippinger. Measurements for meadow jumping mice in Colorado are: total length 187-255 mm; length of tail 108-155 mm; length ofhindfoot 28-35 mm (Fitzgerald et al. 1994). Body weights of meadow jumping mice are variable, not only for different animals, but for the same individual depending on activity or season (Quimby 1951). Such variability arises from sex, age, reproductive status, preparation for entering hibernation, and condition on emergence from hibernation. Mean body weights reported for adult Preble's meadow jumping mice are 19g (n = 37; Meaney et al. 1996), 23.5g (n = 5, se = 1.2g; Colorado Natural Heritage Program, unpublished data);22.7g (n = 30, se = 0.1g; M. Bakeman unpublished data 1997), 20.3 g (n = 26, se = 0.5g; Meaney et al. 1997). Two species of ectoparasites have been identified on Preble's meadow jumping mice (R Schorr, personal communication). These includes fleas (Megabothris abantis, (identified by K. Gage of the Center for Disease Control) and a larval form of a tick, either Ixodes sculptus or L kingi. Genetics The family Zapodidae (jumping mice) consists of small to medium-sized mice with enlarged hind feet and exceptionally long tails. Four living genera are recognized in this family, two of which, Zapus and Napaeozapus, are found in North America (Hall 1981). There are three living species of the genus Zapus: Z. trinotatus (Pacific jumping mouse), Z. princeps (western jumping mouse), and Z. hudsonius (meadow jumping mouse). Z. h. preblei is one of twelve living subspecies of the species Z. hudsonius (Fig. 2a, Krutzsch 1954, Hafner et al. 1981). Z. h. preblei was first described by Krutzsch (1954) from a specimen collected by E. A. Preble in 1895 near Loveland, Colorado. Genetics work using DNA sequencing of the mitochondrial DNA non-coding or re D-loop .tE region was conducted by Biosphere Genetics, Inc., to help determine whether populations of Z. h. preblei constitute one or more distinct evolutionary units. Genetic samples were collected from livetrapped mice, following a protocol specified by Biosphere Genetics, Inc., adapted and updated from the U.S. Fish and Wildlife Service standardized protocol. Sampling in 1996 and 1997 by all field crews (project leaders M. Bakeman, C. Meaney, T. Ryon, B. Stoecker, Colorado Natural Heritage Program) �18 was performed on meadow jumping mice presumed to be Z. h. preblei, yielding samples from 72 individual mice. Twenty genetic samples were also prepared from specimens provided either directly or indirectly by five different museums and university cooperators. Samples analyzed from livetrapped mice came from 23 locations in Colorado and two locations in Wyoming (Table 1). Samples analyzed from specimens provided by museums and cooperators represented reference material for species and subspecies from Colorado, Indiana, Minnesota, Nebraska, New Mexico, and Wyoming (Table 1; Fig. 2a and b). Preliminary DNA sequencing work in 1996 using a 550 basepair fragment of the mitochondrial DNA non-coding region examined 16 individuals from seven locations in Colorado and found two populations from Las Animas County (Lake Dorothey area) to be distinct from five other populations which ranged from Boulder County south to El Paso County. As more samples from additional trapping locations and museum specimens were brought in to the analysis in 1997, a smaller segment of the non-coding region included within the original 550 base-pair fragment was used to avoid problems with non-specificity of one of the primers. DNA sequences were obtained for 433 basepair-long fragments in all samples analyzed in both years. Phylogenetic relationships among the samples were inferred from maximum parsimony analysis. Strength of these phylogentic relationships were evaluated using bootstrap analysis (see Riggs et al. 1997, Appendix A for detailed methodology). Analysis of the mitochondrial DNA sequence data indicates that mice sampled from southeastern Albany County, Wyoming, south along the Front Range of the Rocky Mountains to western Las Animas County, Colorado (Purgatoire Campground, San Isabel National Forest), form a coherent genetic group (Fig. 3, L. Riggs et al. 1997). This group of samples are distinct from samples obtained from mice from three other populations. Genetic samples from mice captured in the Dorothey Lakes area of southern Colorado (Las Animas County) group together and are most closely allied with · Z. h. luteus, a subspecies described from New Mexico (Riggs et al. 1997). The single genetic sample collected from Weld County, Colorado (Lone Tree Creek) and six samples obtained from Warren Air Force Base in Laramie County, Wyoming are most similar to reference samples of Z. princeps from Colorado (Larimer County) and New Mexico (Taos County) (Riggs et al. 1997). The sequence data indicate that the samples from specimens identified as Z. h. luteus and samples from specimens of Z. princeps are more closely allied with each other than either is with the samples defining the Z. h. preblei gro.up. This closer alliance of the samples of Z. h. luteus with Z. princeps rather than Z. h. preblei conflicts with results ofHafher et al. (1981), based on a combination of pelage, morphologic, and genetic data, which support a closer alliance with Z. hudsonius. Phylogenetic analysis indicates that the group of samples referred to as Z. h. preblei cannot at this point be distinguished clearly from four reference samples of Z. h. campestris from Weston County, Wyoming (two samples), Custer County, South Dakota (two samples), one sample of Z. h. pallidus from Garden County, Nebraska, or from two samples of Z. h. intermedius from an unspecified county in Minnesota (Fig. 3, L. Riggs et al. 1997). A suggestion from the analyses is that Z. hudsonius from Indiana (presumed to be Z. h. americanus) may have shared a common ancestor with the progenitors of samples from Z. h. luteus and Z. princeps more recently than with the Z. h. preblei group and perhaps other subspecies described for the north central states (Riggs et al. 1997). A more complete biosystematic evaluation of jumping mice is needed to clarify and further refine relationships among populations of the group referred to as Z. h. preblei as well as to other subspecies and species of the genus 7Apus. Such·an evaluation requires detailed analyses of pelage, morphometric, and genetic data from sufficient numbers of individuals to adequately represent the populations of interest. However, the mitochondrial DNA non-coding (D-loop) sequence data available at this time are consistent with the view that a geographically contiguous set of populations previously recognized as Preble's meadow jumping mouse form a homogenous group recognizably distinct from other nearby populations and from another geographically-adjacent species of the genus (Riggs et al. 1997). Therefore, given the genetic data available, Riggs et al. (I 997) conclude Preble's meadow jumping mouse is a distinct population of Z. hudsonius. �a. l. 2. 3. 4. S. 6. 7. 8. 9. 10. II. 12. Z h. acadicvs Z h. a/cccnsls Z h. amerlcanus Z h. campestrls Z h. canadensts Z h. hudsonlus Z h. lntennedius Z h. la,Jas Z h. pallidus Z h. pnblei Z h. tenellus Z h. luteus b. ... Figure 2. Range of Z. hudsonius and its 12 subspecies (a) and Z. princeps (b) (modified from Hall 1981). Stars .indicate locations where reference samples were collected from specimens at either museums or from university cooperators for genetic analysis. Reference samples for Z. h. luteus include two from Otero County, New Mexico, and two samples from Sandoval County, New Mexico. Reference samples for Z. h. campestris include two from Weston County, Wyoming, and two samples from Custer County, South Dakota. Two reference samples of Z. h. intermedius came from Minnesota. Two samples of either Z. h. americanus or Z. h. intermedius came from Indiana. One sample of Z. h. pallidus came from Garden County, Nebraska. Reference samples for Z. prlnceps princeps include one sample from Taos County, New Mexico, and four samples from Larimer County, Colorado. �20 Table 1. List of specimens from which samples were taken for genetic analyses. Each specimen is identified by a genetic code name, general location, and origin of the specimen. Code Location BAD BOS BVC CCF CCR CGA CZP DC DMNH Origin of Specimen Badwater Creek, Natrona County, WY hudsonius South Boulder Creek, Boulder County, CO Beaver Creek, El Paso County, CO Warren Air Force Base, Laramie County, WY Chicorica Creek, Lake Dorothey State Park, Las Animas Co., CO Medicine Bow NF, Albany County, WY Neota Creek, Larimer County, CO U. S. Air Force Academy, El Paso County, CO San Isabel National Forest, Las Animas County, CO ZAHU East Plum Creek, Douglas County, CO Hay Gulch, Elbert County, CO Coal Creek, Jefferson County, CO Lake De Smet near Buffalo, Johnson County, WY hudsonius Lone Pine Creek, Larimer County, CO Lone Tree Creek, Weld County, CO Marshall Road, Dry Creek Ditch #2, Boulder County, CO Rabbit Creek, Larimer County, CO Rocky Flats Environmental Technology Site, Jefferson County, CO Roxborough State Park, Douglas County, CO Smith Creek, U. S. Air Force Academy, El Paso County, CO St. Vrain, Boulder County Open Space, Boulder County, CO West Fork Schwacheim Creek, Lake Dorothey, Las Animas Co., CO Woodhouse Ranch, Douglas County, CO West Plum Creek, Douglas County, CO Monument Creek, El Paso County, CO Ralston Creek, Jefferson County, CO Indiana ZHCSDC Custer County, SD EPC HAY JCC LDS LPC LTC ~ RBC RF ROX SCR STV WFS WHR WPC WRN WRP ZHCWYW Weston County, WY ZHLNMO Otero County, NM ZHLNMS Sandoval County, NM ZHPNEC Cherry County, NE ZHPNEG Garden County, NE ZPPCOG Grand County, CO ZPPNMf Taos County, NM Museum specimen identified as Z. Z. h. preblei survey Z. h. preblei survey Z. h. preblei survey Museum specimen identified as Zapus Z. h. preblei survey From B. Wunder identified as Z. princeps Z. h. preblei survey Museum specimen identified as Z. princeps princeps Z. h. preblei survey Z. h. preblei survey Z. h. preblei survey Museum specimen identified as Z. Z. h. preblei survey Z. h. preblei survey Z. h. preblei survey Z. h. preblei survey Z. h. preblei survey Z. h. preblei survey Z. h. preblei survey Z. h. preblei survey Museum specimen identified as Zapus Z. h. preblei survey Z. h. preblei survey Z. h. preblei survey Z. h. preblei survey Museum specimen, may be either Z. h. intermedius or Z. h. americanus Museum specimen identified as Z. h. campestris Museum specimen identified as Z. h. campestris Museum specimen identified as Z. h. luteus Museum specimen identified as Z. h. luteus Museum specimen identified as Z. h. pallidus Museum specimen identified as Z. h. pallidus Museum specimen identified as Z. p. princeps Museum specimen identified as Z. p. princeps �21 MRD97&lMNH9716 ROX9701 80S9706 EPC9701 ZHCSOC2--- 70 LOS1 - - RSS1076 OZP1 OCR ZH.N Figure 3. Graphical representation of the genetic relationships among the single most-representative individuals from eachpopulation or reference group ofjumping mice sampled as determined by Riggs ct al. (1997). The more branching that occurs between the center circle and the edge of the circle, the more dissimilar that sample, or group of samples is from the rest of the samples. The numbers on the lines emanating from the center circle are bootstrap values. Bootstrap values greater than 90 indicate those relationships which are strongly supported. Each specimen location can be derived by matching the sample code with those in Table 1. Numbers following the alphabetical code in this figure denote the sample number used in the analyses from that location. �22 DISTRIBUTION The meadow jumping mouse (Z. hudsonius) is broadly distributed across North America from the Atlantic to Pacific coasts, extending south into the United States to Alabama and Georgia and west across the Great Plains to the base of the Rocky Mountains (Fig. 2a). In general, it is a common inhabitant of moist, grassy and herbaceous fields. Eleven living subspecies have been described (Whitaker 1972). Hafner et al. {1981) describe a twelfth subspecies, Z. h. luteus. Z. h. preblei occurs only in eastern Colorado and southeastern Wyoming (Krutzsch 1954, Long 1965, Armstrong 1972). From its limited ecological and geographic distribution, Fitzgerald et al. (1994) suggest it is an Ice Age relict, once widespread in tallgrass prairie across the eastern plains of Colorado but now restricted to scattered localities on the Colorado Piedmont. Similar relict populations of meadow jumping mice in the White Mountains of Arizona and the Sacramento Mountains and Rio Grande Valley of New Mexico are described as the subspecies Z. h. luteus (Hafner et al. 1981). Numerous surveys have been conducted since 1990 to establish the current distribution of Preble's meadow jumping mouse {Fig. 4). Surveys have been funded by the U. S. Fish and Wildlife Service, Rocky Flats Environmental Technology Site, Colorado Division of Wildlife, U. S. Air Force Academy, Warren Air Force Base, Colorado Department of Transportation, City of Boulder Open Space and Real Estate Department, and Boulder County Parks and Open Space Department (Armstrong et al. 1997, P. Plage, personal communication). From 1990 to 1997 such surveys have yielded captures of Preble's meadow jumping mouse, based on field identification and supported by the genetic analyses, at the following sites: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. Lone Pine Creek, Larimer County, Colorado Rabbit Creek, Larimer County, Colorado St. Vrain Creek and associated tributaries in Boulder County, Colorado City of Boulder Open Space, along South Boulder Creek and its tributaries, Boulder County, Colorado Coal Creek, Jefferson County, Colorado Rocky Flats Environmental Technology Site, Jefferson County, Colorado White Ranch P,ark, Ralston Creek, Jefferson County, Colorado Plum Creek drainages including Indian Creek, West Plum Creek, and East Plum Creek, Douglas County, Colorado Roxborough State Park, Douglas County, Colorado Hay Creek, Elbert County, Colorado Monument Creek on the U. S. Air Force Academy {AFAC) and tributaries of Monument Creek off the AFAC including Smith Creek, Pine Creek, and Jackson Creek, El Paso County, Colorado Medicine Bow National Forest, Albany County, Wyoming Two more sites yielded mice that were identified as Z. h. preblei in the field but were genetically found to be more closely allied with Z. princeps (see Genetics section). These sites include: I. 2. Warren Air Force Base, Laramie County, Wyoming Lone Tree Creek, Weld County, Colorado Similarity of the habitat at these sites compared to those where Z. h. preblei (as determined genetically) were found suggests the possibility of areas of sympatry or parapatry between the two species of Zapus. According to Fitzgerald et al. (1994) the distributions of Z. princeps (Fig. 2b) and Z. hudsonius (Fig. 2a) do overlap. The area of overlap occurs in eastern Wyoming. The distribution of Z. hudsonius in Colorado is now known to be larger than shown in Fitzgerald et al. (I 994). The boundaries are currently as far south as Las Animas County (based on genetically identified specimens of Z. h. �23 Platte Goshen □ Albany □ □ □ □ • Wyoming Laramie Weld Colorado Boulder 0 Adams Denver Arap~hoe • Elbert □ El Paso r:fl □ Pueblo Las Animas a New Mexico Figure 4. Distn"bution of historical occurrence (o) of Z. h. preb/ei, sites trapped since 1990 where Z. h. preblei were found(•). sites trapped since 1990 whereZ. h. preb/ei were not found(□). sites where mice were found and identified as Z. h. preb/ei in the field but were genetically identified as Z. princeps {[!I), and sites where mice were found and identified as Z. princeps but were genetically found to be Z. h. preblei (D). �24 preblei). Captures of Z. princeps and Z. hudsonius (as identified in the museum) have occurred as close as 8 miles within the same drainage (Armstrong 1972). Z. princeps were reported captured in 1981 (Olson and Knopf 1988) at the Lone Pine site in Larimer County, Colorado, where Z. h. preblei were captured this year. Because neither specimens or genetic samples were taken in the 1981 study, identification of those mice will remain in question. The discrepancy may be explained by field misidentification or Meaney et al. (I 997) also suggest this discrepancy might be explained by displacement of Z. princeps with Z. h. preblei sometime in the 16 intervening years between trapping efforts. Although assumed to have different ecological requirements, genetic evidence presented here suggests further investigation of possible distributional overlap between Z. h. preblei and Z. princeps. Several notable results from the genetic analysis of specimens acquired from the Denver Museum of Natural History may affect the location of the southern distribution of Z. h. preblei. One site yielded mice identified as Z. p. princeps in the field but were found to be genetically more closely allied with Z. h. preblei. This site is the most southern location to date of Z. h. preblei and is located at: I. Purgatoire Campground, San Isabel National Forest, Las Animas County, Colorado Also from Las Animas County were mice collected and identified as Z. h. luteus (Jones 1996), an identification also supported by the genetic analysis. These mice were from: 1. 2. Lake Dorothey State Wildlife Area, Chicorica Creek, Las Animas County, Colorado Lake Dorothey State Wildlife Area, West Fork Schwacheim Creek, Las Animas County, Colorado These sites suggest a more northerly distribution of Z. h. luteus, currently known only from limited areas in New Mexico and Arizona (Hafner et al. 1981). Identifying the southern boundary of Z. h. preblei clearly needs further study. Trapping surveys have also provided evidence that a number of historically occupied sites are no longer inhabited by the mouse. Ryon (I 996) trapped eight sites of historically occupied sites with no captures of Z. h. preblei. The eight historic sites had suffered either direct disturbances or increasing isolation as a result of past land use, affecting vegetation structure and composition (Ryon 1997). Thus, although the perimeter of the range of Z. h. preblei may or may not have remained the same, specific locations within the range have definitely changed and may have decreased. The pattern of successful and unsuccessful capture sites (Fig. 4) also suggests that development of the Denver metropolitan area may have created a north-south severing of the range of Z. h. preblei. ECOLOGY The ecology of Preble's meadow jumping mouse has not been studied in detail. However, from the limited information available there appear to be a number of similarities in its natural history and ecological requirements to populations of Z. hudsonius studied in more detail in the eastern and midwestern United States. Therefore, select information available on eastern subspecies of Z. hudsonius is presented below to provide some framework for and possible limitations in ecological tolerances for Z. h. preblei. However, it is critical to acknowledge the limitation and possible misleading effects of extrapolating biological information, natural history, and ecological requirements from one subspecies to another, across the geographic range of a species. Demography Reproduction: Meadow jumping mice have been observed to produce up to three litters per season (Whitaker 1963). Breeding peaks appear to occur in early to mid-June and August with a possible third litter in September (Whitaker 1963). Juvenile Z. h. preblei have been observed in June, August, and September (Meaney et al. 1996, 1997, PTI 1996a, M. Bakeman unpublished data, T. Ryon unpublished data), suggesting two litters per year. Z. hudsonius typically have litters of 5-6 young per �25 litter (Quimby 1951 ). Age of first reproduction is unknown for Z. h. preblei, however, females of Z. hudsonius have been observed to give birth at 3 months (i.e., females born in June have been observed to give birth in August of the same year). Gestation period is approximately 18 days (Quimby 1951). Young remain dependent on the female for approximately 18 days (Quimby 1951). No evidence of male parental care exists for Z. hudsonius (Whitaker 1963). Survival: No information exists on survival rates for populations of Z. h. preblei. Whitaker (1963) observed an over-winter loss of 67% in a New York population of meadow jumping mice. Most of this loss was.assumed to be from insufficient fat stores to survive hibernation. Besides insufficient fat storage prior to hibernation, other observed mortality factors in Z. hudsonius include predation (Whitaker 1963, Poly and Boucher 1997} and cannibalism (Sheldon 1934). Other assumed mortality factors for Z. h. preblei include starvation, exposure, and disease. Population Structure: Armstrong et al. (1997) reported an overall sex ratio for all captured Preble's meadow jumping mice of 51. 6 males: 48.4 females; approximately 86. 0% of captures were identified as adults. However, Armstrong et al. (1997) suggested that these data be interpreted with caution because of possible differences in field techniques. Longevity: Very few individuals of Z. h. preblei have been permanently marked. Therefore, recapture information, necessary to determine longevity, is minimal. Several recaptures have yielded adults surviving through two years, indicating a longevity of at least three years (T. Ryon, unpublished data). One recapture history from Rocky Flats Environmental Technology Site recorded an adult male in 1996, two years after it was first captured as an adult in 1994, indicating survival of at least four years (PTI 1996b). Quimby (1951) found that only a low percentage of Z. hudsonius lived two years or more, but gave two records of mice that lived for at least two years under natural conditions. Whitaker (1963) reported a female living at least two years. Dispersal.: Dispersal information will be key to any conservation strategy designed for Preble's meadow jumping mouse. Key factors include (1) which segment of the population disperses, (2) when do they disperse, (3) through what habitat do they disperse, (4) how far will individuals disperse (i.e., what is the maximum distance that separates adjacent populations) and (5) how critical is dispersal (both into and out of a population) to the persistence of a given population. The only data available on dispersal and/or movement for Z. h. preblei are from marked mice at Rocky Flats Environmental Technology Site (T. Ryon unpublished data). Two mice, an adult female and an adult male, were observed approximately 1.6 kilometers from previous locations (incidences occurred separately). Each of the locations were in the same drainage (Woman Creek). Population Persistence: For purposes here, population persistence is defined as the presence of Z. h. preblei at the same site for multiple years. The majority of recent locations of meadow jumping mice have been the result of survey efforts that focused only on determining presence of Z. h. preblei. Once survey protocols (USFWS 1997b) are met [i.e., a minimum of 400 trapnights (one trap set for one night = 1 trapnight) conducted] sites are typically not revisited eliminating the possibility of determining population persistence at these sites. Thus, these and other sites not yet surveyed may or may not have persistent populations. Areas where trapping was conducted over multiple years as part of further research, yielded three sites in Colorado that support persistent populations: Rocky Flats Environmental Technology Site, the U. S. Air Force Academy near Colorado Springs, and Boulder Open Space on South Boulder Creek. Besides establishing the presence of Z. h. preblei over multiple years, other considerations of persistence include (1) presence in consecutive years versus a 'blinking' in and out of a population as would be expected in a metapopulation [a set of local populations which interact via dispersal of individuals moving among populations and where local extinctions and recolonizations occur (Levins �26 1970)], (2) how populations are sustained in a given area (e.g., source or sink populations), (3) maximum abundance supportable by a given area, and (4) fluctuations in abundance of a population over years. If Z. h. preblei exist in areas as part of a metapopulation or as a series of source-sink populations it would be critical to conserve and protect all key sub-population areas as well as critical habitat for dispersal. Detailed population studies, including estimating and determining factors affecting survival, reproduction, and dispersal rates, must be conducted to determine if Z. h. preblei occurs as either metapopulations or in source-sink populations. There have been no such studies conducted on Z. h. preblei or Z. hudsonius elsewhere to provide any supporting or refuting evidence of this hypothesis. Consistently low abundance in a given area due to limiting ecological requirements (e.g., size of area of suitable habitat) or fluctuations in population abundance, including years of low abundance, must be considered in any conservation strategies for Z. h. preblei because of the threat of complete loss of the population due to catastrophic or extreme environmental conditions. Some evidence exists for such fluctuations in population abundance at Rocky Flats Environmental Technology Site. Trapping surveys on the Woman Creek drainage yielded the following captures of Z. h. preblei: seven in 1993 (EG&G 1993), zero in 1994 (T. Ryon, unpublished data), one in 1995 (T. Ryon, unpublished data), two in 1996 (PTI 1996a), and 33 in 1997 (T. Ryon, unpublished data). Trapping effort was consistent from 1995-1997. Repeated trapping surveys conducted over different years intermittently yielded successful captures of Z. h. preblei along the St. Vrain Creek and lower Coal Creek (three in 1989, zero in 1992, zero in 1994, zero in 1996, one in 1997, M. Bakeman, unpublished data). If protocols in each year were the same, these data also suggest populations may undergo fluctuations in abundance. Hibernation Jumping mice of the genus Zapus are true hibernators, spending much of their lives in hibernation. Meadow jumping mice spend approximately 7 months (~210 days) per year in hibernation (Quimby 1951) whereas estimates for Z. princeps indicate that some populations (e.g., in the western mountains of Utah) spend up to 300 days per year in hibernation (Cranford 1983). Males emerge prior to females (Bailey 1923, Bailey 1929, Hamilton 1935, Quimby 1951, Whitaker 1963) with the earliest annual recorded dates for Z. hudsonius males being April 25-May 16 and females May 4-26. Trapping surveys for Z. h. preblei were not designed to provide estimates for dates of immergence or emergence. However, such dates might be approximated by earliest spring and latest fall capture dates. The earliest spring capture date recorded for an adult male was May 5 in 1993 at Rocky Flats Environmental Technology Site (M. Bakeman unpublished data) and for an adult female, May 21 in 1996 also at Rocky Flats Environmental Technology Site (PTI 1996b). Latest fall capture date for an adult male was October 10 in 1994 at South Boulder Creek (ERO Resources 1995) and September 28 in 1995 at the Van Fleet Parcel on South Boulder Creek (Armstrong et al. 1997). A juvenile male was captured as late as October 26 and a female juvenile on October 27 both in 1995 at Rocky Flats Environmental Technology Site (M. Bakeman, unpublished data). Whitaker (1963) reported a 67% loss of individuals over hibernation and that average body mass of individuals emerging from hibernation was greater than the average for mice entering hibernation. Because no mice are known to store food in their hibemacula, this indicates that the lighter individuals died during hibernation and only those entering with higher masses survived. All the energy they use during hibernation and the periodic arousals (the energetically most expensive part of hibernation) must be the fat they carry into hibernation (B. Wunder, personal communication). Thus, the ability to put on sufficient fat for overwinter survival during hibernation is a critical factor in the life history of these mice. Jumping mice hibernate in underground burrows (Quimby 1951, Whitaker 1963). They are excellent burrowers and create their own hibemacula Meadow jumping mice are generally solitary hibernators, however, there have been occurrences of more than one mouse found in a single hibemaculum. One hibemaculum, located on Rocky Flats Environmental Technology Site, used by Z. �27 h. preblei has been located (Armstrong et al. 1997). This site was 9m above a creek bed (Walnut Creek); it had a thick cover of chokecherry (Prunus virginiana) and snowberry (Symphoricarpos spp.), the mouse was found in a leaf litter nest 30cm beneath the ground in coarse textured soil (Armstrong et al. 1997). Four possible hibemacula were located by tracking radio-telemetered mice at the U. S. Air Force Academy in fall 1997. These sites are located 7, 12, 29, and 31 m from a creek bed (R Schorr, personal communication). There was no consistency among sites in aspect (N/NW, SISE, E, and none [level ground]). Three sites were in vegetation dominated by coyote willow (Salix exigua), one site was in vegetation dominated by snowberry and mullein (Verbascum thapsus). However, all four hibemacula appear to be below coyote willows. These four U. S. Air Force Academy sites have not been disturbed to protect any hibernating mice and therefore are only possible hibemacula because there is no confirmation a mouse is actually hibernating there. Confirmation of a true hibemaculum cannot be made until a chamber, or nest is located. These sites may also possibly be locations of radios discarded by the mice or dead mice. Behavior Jumping mice, including Preble's meadow jumping mouse, are primarily nocturnal or crepuscular but can be observed during the day. Unlike most other nocturnal small mammals Preble's meadow jumping mice do not spend the day underground. The mice can be observed during the day, often under shrubs attempting to remain still (M. Bakeman, T. Ryon, R Schorr personal communication). The shrubs may provide either cover from predators or thermo~regulation. As their common name suggests, the saltatorial ability ofjumping mice is well developed. Although longer jumps have been observed the typical behavior is an initial jump of 60 - 90cm followed by rapid jumps of approximately 30cm (Quimby 1951). The more common mode of locomotion, however, is to crawl through or under grass and other vegetation flattening their bodies close to the ground and proceeding quadrupedally (Quimby 1951). Similar jumping and crawling behavior has been observed in Preble's meadow jumping mice as well as running, jumping, and climbing through shrub canopies (M. Bakeman, C. Meaney, T. Ryon, R Schorr personal communication). Given their preponderance for riparian habitats, it is not surprising to find that the mice are also strong swimmers (M. Bakeman, C. Meaney, R Schorr, personal communication). Both swimming and jumping abilities apparently serve more commonly as mechanisms for retreat from either predation or perceived threats rather than as typical mechanisms for locomotion. The mouse is also adapted for digging and creates nests of grass, leaves, and woody material (Harrington et al. 1996). All nests found at the Rocky Flats Technology Site, with the exception of the hibemaculum, were above-ground, generally at the base of willow (Salix spp.) clumps (M. Bakeman, personal communication). There does not appear to be any social structure among individuals within a population of jumping mice. No evidence exists for parental care by the male or any interaction between males and females after breeding occurs. Adults of the species Z. hudsonius are essentially solitary individuals. Except when young, jumping mice are usually silent. On occasion adults have been heard to emit chirps and other sounds (Quimby 1951, Whitaker 1963). No observations have been made of vocalizations by Preble's meadow jumping mice. Drumming noises can be produced by vibrating the tail rapidly against a surface (Svihla and Svihla 1933, Sheldon 1938, Whitaker 1963). Tail rattling, appearing to be a nervous reaction when placed in a capture bag, has been observed in Preble's meadow jumping mouse (M. Bakeman, personal communication). Quimby (1951) and Whitaker (1963) reported that mice in the genus Zapus often washed their feet, faces, and especially their tails. The tail was grasped in the forepaws, and passed completely through the mouth, whereas the hands and feet were washed by means of the forepaws. Bailey (1926) found that jumping mice would reach as high as it could on grass stems, bite them off and pull them to the ground, repeating the procedure until the head was reached and eaten. This process would result in a pile of pieces of grass stem, often with the rachis and glumes on top. Whitaker (1963) found similar behavior in his study of meadow jumping mice in New York. M. �28 Bakeman (personal communication) found piles of pieces of grass stem in areas where Preble's meadow jumping mice were known to occur, however these piles may also have been created by individuals of the species Microtus. Food Preferences Specific food habits are unknown for Preble's meadow jumping mouse. Armstrong et al. (I 997) summarized what is currently known about food habits of meadow jumping mice in general as follows: Studies of food habits in central and eastern United States indicate they are governed by • availability more than preference (Whitaker 1963). Grass seeds of several species are probably the most important component of the diet, and mice will shift to those species that have available seed. Invertebrates and fungi are also readily eaten. Mice feed on both adult and larval invertebrates, especially Coleoptera (beetles). Invertebrate feeding is very important in the spring as mice emerge from hibernation, and may consist of half of the diet at that time. Mice also feed on various species of fungi, which are often encountered during burrowing activity. As the growing season progresses, graminoid seeds dominate the diet. There is no reason to believe food habitats of one subspecies should differ greatly from those of other subspecies. This belief is further supported by the indirect evidence available that food habits of Z. h. preblei are similar to that described above (i.e., the observation of the piles of grass stems observed in areas where the subspecies is known to occur [see Behavior section]). • As true hibernators, meadow jumping mice do not cache food in their hibernacula Therefore, it is assumed they require high quality food for high fat accumulation prior to immergence. Preble's meadow jumping mice have been observed to increase body weight by 10% in the 2-3 weeks prior to immergence. Laboratory studies show the mouse can gain mass at rates up to 1.0 grams per day (B. Wunder, personal communication). This ability to increase fat reserves at such a rate is used by the mice for preparation to enter hibernation. However, for the mice to successfully gain enough fat prior to hibernation to ensure a high probability of survival throughout hibernation, food of sufficient quality must be available. No specific information is available on what foods Preble's meadow jumping mice eat to meet these ecological requirements. However, from late August through early September, when adults begin to gain weight, there are many species of graminoids that have available seed. In most years, grasshoppers are also readily available and probably dominate invertebrate biomass in most habitats. Preble's meadow jumping mice are most often found near open water. Dependency on open water has not been established, however work is currently in progress to establish if such a need exists (B. Wunder, personal communication). Based on trapping records at Rocky Flats Environmental Technology Site, trapping success at some sites declines dramatically after creekflow ceases, and it is suspected there may be a shift in site use because of the absence of open water (M. Bakeman personal communication). Habitat Use Landscape scale: The habitat matrix within the range of Z. h. preblei is mixed grasslands adjacent to the Colorado Front Range along the Piedmont and along the base of the Laramie Mountains in Wyoming and extends to the Colorado plains. Within this matrix, Preble's ~eadow jumping mice occur along stream drainages that contain patches of suitable vegetation. Suitable habitat appears to have at least two major components. The first component is a supply of open water, at least in part of the active season (M. Bakeman, C. Meaney, personal communication). Secondly, areas where Preble's meadow jumping mouse has been found have dense cover (M. Bakeman, C. Meaney, personal communication). �29 If Preble's meadow jumping mouse behaves as a metapopulation in the classical sense of a set of local populations linked by infrequent dispersal then habitat includes not just one area of suitable habitation but also areas suitable for nearby mouse populations. These suitable areas must also be linked by dispersal habitat. If the mice are dependent on dense riparian habitat for dispersal as well as for areas to reproduce, persistence of discrete populations would require a mosaic of suitable discrete riparian patches interconnected with dispersal corridors of similarly dense riparian vegetation. If mouse populations function in a source (populations where growth rate~ 1) and sink (those populations where growth rate< 1, maintained through immigration) system, it will be critical to identify and protect those populations serving as sources. Thus, for a source-sink population critical habitat will include those areas that support source population, dispersal habitat to sink areas will be less critical. If local mouse populations are functionally discrete, such a mosaic of interconnected areas of suitable habitat would provide a buffer for local, and source, populations against deleterious stochastic events by providing the opportunity for local population failures to be 'rescued' by immigration from other populations. Areas of suitable habitat must also provide requirements to survive throughout the life cycle. These requirements must provide necessities for both the active period and hibernation periods. During the active p·eriod suitable habitat must provide requirements for daily survival, reproductive activities (breeding, nesting, and rearing of young to independence), and dispersal. The hibernation period requires sufficient food supplies to assure fat storage prior to hibernation and suitable hibemacula (see Hibernation above). Habitat providing all seasonal and life cycle requirements may or may not occur in a single contiguous area If not in a contiguous area, habitat patches must occur in a mosaic of usable areas where suitable corridors exist for seasonal movement among sites. Site sca/,e: Based on studies of Z. h. preblei and Z. hudsonius elsewhere, Z. h. preblei apparently occurs mostly in undergrowth consisting of grasses, forbs, or both in open wet meadows and riparian corridors, or where tall shrubs and low trees form an overstory and provide adequate cover (Armstrong et al. 1997). Meadow jumping mice are widespread in abandoned grassy fields, but is often more abundant in thick vegetation along ponds, streams, and marshes or in rank herbaceous vegetation of wooded areas (Whitaker 1963). Preble's meadow jumping mice have been trapped in natural riparian areas as well as areas altered by anthropogenic influence including ditches and wetlands adjacent to interstate highways, cement-lined ditches with tall cover, ditches along driveways and moderate road use, and moderate cattle grazing (M. Bakeman, personal communication). The majority of sites where Z. h. preblei have been found consist of multistoried cover but the species composing the cover vary greatly (Armstrong et al. 1997, Meaney et al. 1997) . Vegetation composition of the dense cover varies considerably and includes both native and non-native species (Meaney et al. 1996, 1997, Armstrong et al. 1997, M. Bakeman, personal communication). The herbaceous understory can be primarily grasses or forbs or a mixture of the two. Few of the sites, however, are dominated by fewer than two understory species. The tall shrub canopy at most sites is willow (several species), although scrub oak, birch, and alder occur in sites south of the Palmer Divide (Armstrong et al. 1997). Ponderosa pine is the most common tree at higher elevations. The mouse appears to tolerate weedy or exotic species in areas that are structurally diverse and species rich; nearly every successful site contained Canada thistle (Armstrong et al. 1997). Thus, the mouse does not appear to have an affinity toward any single plant species but instead favors sites that are structurally diverse and provide adequate cover and food throughout its life cycle. Preble's meadow jumping mouse mice are typically not found in upland areas away from riparian habitats but are most often captured where either ground water daylights to seep springs or on main water channels (M. Bakeman, T. Ryon, personal communication) suggesting a dependence on open water, at least during their active periods. Movement of Z. h. preblei on Woman Creek at Rocky Flats Technology Site suggests mice move along corridors of shrub cover, generally Salix exigua (PTI 1996a, T. Ryon, unpublished data), suggesting dispersal habitat is similar to habitats used for other activities. �30 SMALL MAMMAL ASSEMBLAGE Preble's meadow jumping mice are only one component of the small mammal community inhabiting areas where they were captured. These mice were more often found at sites with high species richness and abundance of small mammals (Armstrong et al. 1997, Meaney et al. 1997). The variety and relative abundance of other small mammalian species trapped during surveys for Preble's meadow jumping mouse provide some framework for comparison of small mammal assemblages at sites where Z. h. preblei occurred and sites where no Preble's meadow jumping mice were captured (Table 2). All trapped sites were either historical sites of known occurrence or apparently suitable habitat for Z. h. preblei, providing a common basis for comparison. Three small mammal species (Spermophilus variegatus, Peromyscus nasutus, and Rattus norvegicus) were not captured where Z. h. preblei occurred but were captured in unsuccessful sites. However, total capture occurrences of these three species were only one or two sites and the species comprised an extremely small percentage of the species caught in those areas (Table 2) providing too little information for speculating on possible interactions between these species and Z. h. preblei. Table 2. Small mammal community analysis from sites where Preble's meadow jumping mice (Zapus hudsonius preblei) were captured (n = 28) and not-captured (n = 35). Reported here are number of occUITences a given species was found in areas with and without Preble's meadow jumping mice captures (n and mean percentage and standard error (se) of capture that given species contributed to total small mammal captures at a given site. Data are summarized from Armstrong et al. (1997) and Meaney et al. (1997). All sites were either areas of historical occUITence of Preble's meadow jumeing mice {n = 82 or OCCUITed in areas of aeearently suitable habitat. 0) Z. h. preblei Not Captured Z. h. preblei Captured Species % Capture mean (se) no % Capture mean (se) no Zapus hudsonius prehlei 28 6.65 (1.38) 35 Spennophilus variegatus 0 0 2.0 Peromyscus nasutus 0 0 3.0 Rattus norvegicus 0 0 2 9.5 (8.2) l.0(-) 0 0 Mustela sp. Chaeotodipus hispidus 2 1.4 (0.2) 5 3.0 (0.8) Mus musculus 3 5.8 (4.6) 17 18.4 (6.0) Sorex cinereus 4 1.1 (0.7) 5 2.0 (0.5) Microtus longicaudus 5 15.8 (6.3) 3 9.6 (4.3) Neotoma mexicana 5 6.1(1.7) 8 12.5 (3.6) Reithrodontomys megalotis 5 3.6(1.4) II 3.7 (0.6) Microtus spp. 8 18.3 (6.4) 4 5.7 (4.3) Microtus ochrogaster 18 16.0 (3.7) 25 13.0(3.5) Microtus pennsylvanicus 21 26.2 (4.4) 32 25.8 (4.4) Peromyscus maniculatus 27 50.4 (3.4) 32 54.8 (5.2) �31 The higher occurrence and percent composition of capture of Mus musculus, the house mouse, in areas where Z. h. preblei was not caught might suggest degradation of habitat for Z. h. preblei in those areas or possibly competition between the two species (Ryon 1996). The house mouse, thrives in areas of human habitation and croplands, but also occurs in abandoned fields and ditch banks where they may displace native rodents (Fitzgerald et al. 1994). Ryon (1996) also reported the presence of domestic cats (Fe/is catus) at sites where Z. h. preblei historically occurred but were not found during his study. Contrarily, C. Miller (personal communication) reported house cats along South Boulder Creek where Preble's meadow jumping mice are known to occur. Caution must be used when percentage of capture is used as an index.to the composition and relative abundance of small mammals within a given area Captures may be influenced by trap placement, bait, season, or trapping protocol. Percentage of capture for a given species may also be biased because of either trap-shy or trap-happy individuals within a species or because some species, in general, tend to be easier or harder to trap. Preble's meadow jumping mice appear to favor 'clean' traps. Once traps have been used and soiled by individuals of other species, the probability of capturing a meadow jumping mouse in that trap appears to decrease (M. Bakeman, A. Deans, F. Harrington, T. Ryon, personal communication). For example, of the 60 captures of Z. h. preblei by M. Bakeman (unpublished data) in 1997 all were captured in traps that had only previously been occupied by Preble's meadow jumping mice. Evidence to the contrary, however, also exists. On at least one occasion, R Schorr (personal communication) caught a Preble's meadow jumping mouse in a trap previously occupied by Sorex sp. If one does assume no trapping biases occurred, where captured, Z. hudsonius was one of the less common species in the areas trapped (Table 2). If such rarity in areas of occurrence is genuine then this combined with both its limited distribution and absence from historically occupied areas supports the need for immediate protection of the subspecies and its habitat. POSSIBLE THREATS TO CONSERVATION •The range of Z. h. preblei corresponds largely to the rapidly developing ~ront Range Urban Corridor that runs from Colorado Springs, Colorado to Laramie, Wyoming. Therefore, possible threats to Z. h. preblei survival may include the widespread destruction and modification of riparian corridors and wet meadow habitats by human land uses (USFWS 1997 a). Agricultural, residential, commercial, industrial, and recreational development are likely the cause of such impacts. Specific activities such as water diversion, stream channelization, and sand, gravel and aggregate mining have impacted habitats required by the mouse Other activities such as overgrazing and development of recreational trails may also impact Preble's meadow jumping mouse habitat. Armstrong et al. (1997) summarized results of habitat studies conducted on Z. h. preblei, including the following summary of disturbance elements. Over the past 100 years the [riparian habitat used by Z. h. preblei] has become increasingly urbanized as influenced by grazing, water diversions, wetland conversion, and real estate development. Compton and Hugie (1993) identified agricultural, residential, and commercial development as habitat impacts and suspected grazing as having a negative influence on meadow jumping mouse habitat. Ryon ( 1996) identified alterations to habitat at eight historic meadow jumping mouse sites. These alterations include water diversion, highway construction, gravel mining, grazing, and real estate development. Although suitable habitat exists within this landscape unit, the anthropogenic influence is changing the unit as a whole and fragmenting and degrading riparian areas in general. The effect of these land use practices on mouse distribution is poorly understood. Several investigators have been puzzled at the lack of jumping mice at sites with seemingly well structured habitat, and the presence of mice at a few sites with less than optimal conditions. Land use histoiy of an area may provide clues as to whether mice are at a site or if they can be �32 restored. The reviewers of the [Report on Habitat Findings of the Preble's Meadow Jwnping Mouse, edited by M. Bakeman] list six potential impacts that were thought to play a role in distribution of Z. h. preblei including trails, grazing, mining, development, haying, and riparian hydrology. In order to threaten conservation of Z. h. preblei a feature or features of their ecological requirements must be altered in such a way as to make the area unusable to the mouse or limit its usefulness to such an extent as to deplete the areas' potential for supporting a viable population of Z. h. preblei. Therefore, a threat is defined as any activity that_has the potential to negatively alter any ecological requirements of Z. h. preblei. Effects of the potential threats, listed below, may be permanent or temporary. Management strategies may be developed to offset or eliminate effects from various activities. However, such management strategies are only as good as our knowledge of the full complement of ecological requirements necessary to maintain a viable population of Z. h. preblei. The following defines such ecological requirements, provides specific examples of activities that could negatively alter these requirements, and suggests how these negative impacts would affect the mouse. Vegetation composition and structure Vegetation composition and structure affects many of the ecological requirements of meadow jumping mice including food, cover, nest sites, and suitable dispersal corridors. Vegetation composition in areas inhabited by Preble's meadow jumping mouse must include species that would provide not only daily nutritional requirements but also high quality foods necessary for the quick, high fat storage that occurs prior to hibernation. Cover may provide concealment from predators, areas for thermoregulation, and raw material for above-ground nests. Above ground nests appear to be associated with woody (tree and shrub) structures (M. Bakeman, personal communication). Dispersal habitat must provide sufficient cover and food to assure successful dispersal from one area to another. Activities which could threaten vegetation structure include: grazing, development (including residential, commercial, industrial, and recreational), fire, sand, gravel and aggregate mining. Grazing may alter vegetation structure by either decreasing or eliminating sufficient tree, shrub, and/or tall grass cover. Heavy grazing is especially hard on woody plants. Species composition may also be altered by grazing, eliminating critical food resources. Development of an area may impact species composition, abundance, and density. Developments such as cutbanks would eliminate all vegetation. Short-term effects of fire may be deleterious if above-ground nests (especially with young) are destroyed. Longterm effects may be minimal, if the population can withstand minor losses. Mining operations for aggregate material generally eliminates most vegetation, affecting both structure and composition. Possible deleterious effects on populations of Z. h. preblei include decreased survival, reproduction, or ability to disperse. Riparian Hydrology Hydrology includes physical presence of open water, water quality, seasonality of available open water and amount of water flow. Activities which could threaten suitable hydrologic regimes for populations of Z. h. preblei include water diversion projects, water management projects including irrigation regimes and ditch maintenance, and pollution. These activities could result in loss or increase of surface water supplies, alter flooding events which may be necessary to maintain vegetation composition and structure, and produce siltation in catchment areas. Such activities do not actually have to occur on areas of suitable Z. h. preblei habitat but could occur upstream or downstream of the site. Mining operations can indirectly affect habitat by altering downstream hydrological flows of both surface water and groundwater. Urbanization may affect water quality and stream flows through siltation of catchment areas, presence of pesticides in water, increased nutrient load, and vegetation composition. At Rocky Flats Technology Site, Preble's meadow jumping mice are most often captured where either ground �33 water daylights to seep springs or on main channels fed by seep springs (T. Ryon, personal communication) suggesting a dependence on open water, at least during their active periods. Possible deleterious effects on populations of Z. h. preblei if riparian hydrology is altered include decreased survival or reproduction. H abitaJ structure Habitat structure includes size of suitable area, connectivity of suitable sites to other suitable sites through corridors, and inclusion of all ecological requirements within a given area Thus, altering habitat structure could result in fragmentation of a suitable site, isolation. of a site to other suitable sites, degradation of a suitable site, or total loss of useable habitat Activities which could threaten habitat structure include: aggregate mining, grazing, real estate development, pollution, and fire. Movement of Preble's meadow jumping mice on Woman Creek at Rocky Flats Technology Site suggests mice move along corridors to patches of the most suitable habitat. Most captures occur in areas of shrub cover, generally Salix exigua (PTI 1996a, T. Ryon, personal communication). Possible deleterious effects on populations of Z. h. preblei include decreased abundance due to a decrease in suitable habitat, loss of inter-connectivity of sites that may be necessary to maintain viable populations either in a source-sink or metapopulation situation, and may also result in decreased gene flow through isolation. Distribution Distribution can be defined on several scales: range perimeter, distribution within the range, and distribution within connected suitable habitat. As discussed above (see Distribution section), based on a comparison of currently known sites and historical sites of occurrence of Z h. preblei it is clear distribution (within the historical range) has already been altered. Activities which could alter distribution of Z. h. preblei include real estate development and sand, gravel and aggregate mining by severely degrading or completely eliminating required habitat. Possible deleterious effects on Z. h. preblei include loss of populations throughout the complete ecological range of suitable habitat and further fragmentation within the range of Z h. preblei. These effects could (I) eliminate critical populations of Z. h. preblei to maintain possible metapopulations or source populations or (2) restrict gene flow among populations. Geomorphology Geomorphology is defined as the description of features of the earth's surface as explained by the underlying dynamic and structural geology. Important geomorphological features that may help delineate habitat for Z. h. preblei include stream and floodplain geometry. Steepness of stream banks may affect vegetation composition ans structure. Underground structure along stream banks may provide critical hibemacula sites. Activities that would affect geomorphology include development and mining of sand, gravel, and aggregate material. Preble's meadow jumping mice have rarely been captured in steep-sided drainages (M. Bakeman, personal communication). Possible deleterious affects of altering the geomorphology in areas used by Preble's meadow jumping mouse include decreased survival due, possibly, to insufficient availability of suitable hibemacula Animal Community Composition Animal community composition includes the ocuurance and abundance of all other species in Preble's meadow jumping mouse habitat. The primary activity that may affect populations of Z. h. preblei is development. In particular, the introduction of the house mouse and the house cat may alter the suitability of a site for Preble's meadow jumping mouse; the house cat as a new and effective predator, the house mouse as a possible �34 competitor for resources. Other possible effects of altering the animal community is the introduction of new diseases and parasites to Preble's meadow jumping mouse. The possible deleterious effects of altering the animal community within areas of habitat for Preble's meadow jumping mouse include decreased survival through increased predation, disease, parasites, or competition for resources. CONSERVATION STRATEGY INTRODUCTION This preliminary conservation strategy describes the current status and the goal, objectives, strategies, and research that should be implemented to maintain and restore viable populations of Z. hudsonius throughout its range and to conserve and manage Preble's meadow jumping mouse habitat. This is intended to be a preliminary planning effort based on the best scientific data available to date. This strategy may also serve as a starting point for more comprehensive conservation planning efforts regarding Z. h. preblei. • This strategy is designed to be implemented through the cooperation of state, federal, and municipal government agencies, and involve partnerships with private conservation organizations, individuals, and landowners. The outlined conservation strategy objectives are not necessarily listed in order of priority, but in a logical manner that provides for completion of information needs prior to addressing specific actions. CURRENT STATUS On March 25, 1997 the U. S. Fish and Wildlife Service (USFWS) published a proposed rule in the Federal Register (62 FR 14093) to list Preble's meadow jumping mouse (Z. h. preblei) as 'endangered' under the Federal Endangered Species Act (ESA). Final ruling on the proposed listing is scheduled by March 25, 1998. There are three possible outcomes depending on progress of conservation efforts and knowledge gained prior to the decision date: list as endangered, list as threatened, or withdraw the proposal. The primary reason for the proposed listing of Preble's meadow jumping mouse under Section 4 of the ESA is 'present or threatened destruction, or modification ofits habitat or range' as well as lack of meaningful protection. The listing decision will be made by the U. S. Fish and Wildlife Service solely on the basis of the best scientific data available. As stated in USWFS 1997b: "All Federal agencies have responsibility under section 7(a)(4) of the ESA to protect proposed endangered and threatened species and habitats on which they depend. For projects where a Federal nexus exists (Federal permit, Federal funding, projects on Federal land) and there is potential affect on Z. h. preblei or its habitat, the Federal action agency should contact the USFWS. In addition, the USFWS encourages all Federal agencies to review their properties and projects and make funds available to conduct Z. h. preblei surveys in all potential habitat. It is also imperative to make project proponents aware of the presence of Z. h. preblei should it be listed prior to the time all actions related to any project are completed. Section 9(a)(l) of the ESA prohibits the take (i.e., harass, harm, pursue, hunt, shoot, kill, wound, trap, capture, or collect, or to attempt to engage in any such conduct) of federally endangered or threatened species, except as provided in sections 6(g)(2) and 10 of the ESA. If Z. h. preb/ei is listed while a project is impacting it or its habitats, the ESA would enter into effect to protect Z. h. preblei. A project occurring in poor or marginal habitat but having potential to disrupt a travel corridor (such as a road crossing a creek) is also of concern as well as the question of secondary impacts from proposed projects. Projects removed from potential Z. h. preblei habitat that have the potential to adversely impact the habitat may also require a survey. For example, a residential or commercial development upslope from a creek supporting potential Z. h. preblei habitat may significantly increase runoff or otherwise impact the hydrology, and thereby the habitat present on the creek." Therefore, even prior to a decision on the proposed listing, there is a �35 great need to provide information on the ecology and ecological requirements necessary for conservation of Z. h. preblei. The species Z. hudsonius was designated as Colorado nongame wildlife (Colorado Division of Wildlife Regulations, Chapter 10, Article N, #1004 A 6), which provides a legal protection against taking. It is also a Colorado Division of Wildlife Species of Special Concern, which is an administrative classification rather than a legal one. However, there are current efforts within the Colorado Division of Wildlife to list the subspecies as State Endangered (J. Sheppard, personal communication), defined as any species or subspecies of native wildlife whose prospects for survival or recruitment within the state are in immediate jeopardy. The Colorado Natural Heritage Program (1997) lists the mouse as an S2 species: imperiled in the state because of rarity (6 to 20 occurrences) or because of other factors demonstrably making it very vulnerable to extirpation from the state. Conservation of Preble's meadow jumping mouse was also listed as a specific task to be addressed by both the State of Colorado and the Department of the Interior in the Memorandum of Agreement (94SMU-058), signed by Governor Roy Romer and Interior Secretary Bruce Babbitt on November 29, 1995, between the State of Colorado and the Department of the Interior concerning 'Programs to Manage Colorado's Declining Native Species.' The Wyoming Natural Diversity Database (Fertig 1997) ranks Z. h. preblei as S 1: critically imperiled in the state because of extreme rarity (five or fewer extant occurrences or very few remaining individuals) or because of some factor of the subspecies life history that makes it vulnerable to extinction. The Wyoming Department of Game and Fish (1998) ranks the subspecies as nongame through their Nongame and Wildlife Habitat Protection Programs. GOAL The goal of this conservation strategy for Preble's meadow jumping mouse should work towards the sustainability, protection, and restoration of Z. h. preblei populations and habitats on both private and public lands to provide the spatial, genetic, and demographic structure needed to promote long-term species viability and provide species management flexibility. Specific objectives include the following. OBJECTIVES I. 2. Document the present distribution of Z. h. preblei. This will require the following: • Conduct trapping surveys to better define the boundaries of the range of Z. h. preblei.; conduct trapping surveys in areas of potential sympatry of Z. h. • preblei with Z. princeps; evaluate effect of small mammal species richness and composition on the distribution and/or abundance of Z. h. preblei. • Further explore genetic relationships among different Z. h. preblei populations, among different subspecies of Z. hudsonius, and among different species of Z:O.pus. Identify populations of Z. h. preblei where the rate of population growth <!: 1. This will require population studies to estimate the following parameters: • Survival (annual, over-summer, over-hibernation), and identity of factors affecting survival [e.g., weight, sex, age, abundance (i.e., density dependent Z. h. preblei response), habitat features (e.g., stream reach, vegetation composition), weather, predation, disease]. • Reproductive and developmental parameters (i.e., number oflitters per year, number of young per litter, age at first reproduction, juvenile survival), and identity of factors affecting reproduction [e.g., habitat features (i.e., food quality, nest site availability, cover availability), abundance (i.e., density dependent response), weather]. • Recruitment, and identity of factors affecting recruitment (e.g., weather, habitat features). • Immigration and emigration rates, and identity of factors affecting immigration, emigration (e.g., age, habitat features, abundance). �36 3. 4. 5. • Population structure (sex and age ratios) • Dispersal parameters (rate, who disperses, time of dispersal) • Abundance, and identity of factors affecting abundance (e.g., habitat features). • Rate of population change. Protect populations of Z. h. preblei where the rate of population growth is ::1: I. Maintain current range of natural variability of Z. h. preblei. Identify ecological requirements for sustaining viable populations of Z. h. preblei throughout its range of natural variability. This will require research to address the following: • Estimate habitat use [distances traveled (daily, seasonally), landscape features (connectivity with other potential sites, geology), seasonal use, hibemacula, nest sites, distance to nearest open water, hydrology (water quality, flow), abundance (i.e., density dependent responses), and cover]. • Dispersal habitat [end point descriptions (disperses from what to what), landscape features (connectivity with other riparian strips, corridor use, overland use]. • Physiology (e.g., estimate dependency of Z. h. preblei on open water, food habits, effects of varying water quality on Z. h. preb/ei physiology) . • 6. 7. 8. 9. 10. 11. 12. 13. Protect habitats to sustain or restore populations of Z. h. preblei with high where the rate of population growth ::1: 1. Promote protection, management, and possible restoration of habitat for conservation of Z. h. preblei in all currently or recently occupied and suitable habitat. Monitor the status of Z. h. preblei populations throughout its range to detect changes in local distribution. Identify threats to the conservation of Z. h. preblei. Eliminate or minimize threats to Z. h. preblei conservation. Integrate Preble's meadow jumping mouse conservation strategy objectives with management and habitat objectives of other Front Range riparian species. Promote scientific management of Preble's meadow jumping mouse. Promote public support for Z. h. preblei conservation efforts and scientific management of Preble's meadow jumping mouse through public education. RESEARCH NEEDS There are currently four research projects being conducted on Z. h. preblei, with a fifth planned to begin spring of 1998. Each of these projects is designed to provide further information on the ecology or demography of Preble's meadow jumping mouse. However, if research methodologies were standardized and coordinated across all the projects, comparability and quality of the data would be greatly enhanced. Such a coordinated effort would maximize data quantity, quality, and comparability over varying conditions throughout the range of Z. h. preblei in the most effective and efficient way possible. Comparable information gained across the ecological range of Z. h. preblei would then provide more useful information for use in developing sound management strategies for conservation of the mouse. To achieve such a coordinated research effort all project leaders must agree to follow data collection protocols, establish 'ownership' of data and future publications, and work cooperatively on the possible sharing of equipment and technical assistance during peak data collection periods. To facilitate such a mutual effort will require the participation of all project leaders conducting field studies of Z. h. preblei as well as cooperative investigators, and specialized data analysts with expertise in the study design, analysis, and interpretation of mark-recapture and telemetry data A proposed agenda including tasks, task leaders, and a timetable to initiate a cooperative research effort include: �37 Task Task Leader, Affiliation Date Identify all potential participants: project leaders, cooperating investigators, data analysts. T. Shenk, Colorado Division of Wildlife January 1998 Prioritize research questions. All project leaders* January 1998 Identify sites to best address prioritized research needs. All project leaders January 1998 Design studies specific to research question to be addressed at each site. All project leaders, cooperating investigators, data analysts January-February 1998 Evaluate needs at each site to conduct specified research. All project leaders March 1998 Identify discrepancies between needs and available funds, equipment, and personnel currently allocated for each site. All project leaders March 1998 Attempt to balance discrepancies. All project leaders, cooperating investigators March 1998 * To date: M. Bakeman, C. Meaney, T. Ryon, R. Schorr, T. Shenk The following outline lists research needs to provide information to develop sound management strategies for conservation of Z. h. preblei. Research needs are not listed in order of priority, but listed in a logical manner to provide completion of needs. There are three components of Z. h. preblei ecology that are currently unknown and yet key to any sound conservation strategy for the subspecies. These are (I) detailed demographic studies estimating survival and reproduction and determining the factors influencing each of the parameters, (2) detailed studies evaluating movements and dispersal habitat of individuals within and among populations, and (3) detailed studies to define hibernation needs, primarily descriptions of suitable hibernacula criteria and food requirements for sufficient fat storage prior to immergence. Demographic studies: Information on the population dynamics of Preble's meadow jumping mouse is necessary to determine which areas support populations where the rate of population growth is <!: 1. Key parameters to estimate include: Survival: • estimates of survival, including ► annual ► over-summer ► over-hibernation • investigate possible factors affecting survival, including ► weight ► sex ► age ► abundance (i.e., density dependent response) ► habitat features: stream reach, vegetation composition ► weather ► predation ► disease Approach: mark-recapture techniques �38 Recruitment: • estimate recruitment (as an alternative to reproductive parameters listed below) • investigate possible factors affecting recruitment, including ► weather ► habitat features Approach: mark-recapture techniques Population structure: • estimate sex ratios • estimate age ratios Approach: mark-recapture techniques Abundance: • estimate abundance • investigate factors affecting abundance, including ► habitat features (see under habitat use) Approach: mark-recapture techniques for closed populations Immigration, emigration: • estimate rates of immigration and emigration • investigate possible factors affecting immigration, emigration ► habitat features (see under habitat use) ► abundance (i.e., density-dependent response) Approach: estimate rates from mark-recapture data, identify existence with radio telemetry Reproduction: • estimate number of litters per year • estimate number of young per litter • estimate age at first reproduction • estimate juvenile survival • investigate possible factors affecting reproduction, including ► habitat features: food availability, nest site availability, cover availability ► abundance (i.e., density-dependent response) ► weather Approach: from telemetry work Dispersal Studies: Dispersal is a key process in metapopulation theory and to maintain genetic diversity between isolated populations. Key parameters to evaluate include: Population parameters • who disperses • time of dispersal • estimate rate Approach: estimate rate with mark-recapture, document who and when from both telemetry and mark-recapture data Habitat parameters • through what habitat • end point descriptions (disperses from what to what) • landscape features (connectivity with other riparian strips, corridor use, overland use) Approach: document where from both telemetry and mark-recapture data �39 Distribution Studies: Conservation of Z. h. preblei should maintain populations throughout the range of its natural variation and to try to identify ecological limits for the subspecies. Key concerns include: Range-wide distribution: • conduct trapping surveys to better define the eastern boundary of the range of Z. h. preblei • conduct trapping surveys in areas of potential sympatry of Z. h. preblei with Z. princeps Approach: determine from trapping surveys Habitat Studies: To identify and define habitat requirements of Z. h. preblei studies should be conducted to address the following: Habitat use: • estimate distances traveled: daily, seasonally • describe landscape features (connectivity with other potential sites, geology) • determine seasonal use • describe hibernacula • describe nest sites • estimate distance to nearest open water from other habitats used • describe hydrology (water quality, flow) in areas of use • evaluate effects of abundance on habitat used (i.e., density dependent responses) Approach: estimate from telemetry work Physiological Studies: Physiological studies will provide information on the mechanisms driving habitat selection. Physiological, requirements: • estimate dependency of Z. h. preblei on open water • determine energetic requirements to survive hibernation Approach: estimate from laboratoiy studies Systematic Studies: To better define the relationship of Z. h. preblei to other subspecies of Z. hudsonius and other species of Zapus the following studies should continue. Molecular systematic relationships: • further explore genetic relationships among different Z. h. preblei populations • further explore genetic relationships among different subspecies of Z. hudsonius • further explore genetic relationships among different species of Zapus. Approach: explore through laboratoiy studies Systematic relationships: • link genetic relationships to systematic studies of Z. hudsonius. Approach: explore through museum studies Community Studies: Composition of the community where Z. h. preblei occur could help explain ecological tolerances of the subspecies, providing insight to the mechanisms determining its distribution. Small mammal assemblages: • comparison of small mammal assemblages in areas where populations of Z. h. preblei occur and areas where they do not ► species composition ► relative abundance Approach: estimate from trapping surveys �40 LITERATURE CITED Armstrong, D. M. 1972. Distribution of mammals in Colorado. University of Kansas, Museum of Natural History Monograph 3:1-415. Armstrong, D. M., M. E. Bakeman, A Deans, C. A Meaney, and T. R Ryon. 1997. Conclusions and recommendations in: Report on habitat findings on the Preble's meadow jumping mouse. Edited by M. E. Bakeman. Report to USFWS and Colorado Division of Wildlife. Bailey, B. 1929. Mammals of Sherburne County, Minnesota. Journal ofMammalogy 10:153-164. Bailey, V. 1923. Mammals of the District of Columbia Proceedings of the Biological Society. Washington 36:103-138. Bailey, V. 1926. A biological survey ofNorth Dakota. U.S. Department of Agriculture North American Fauna No. 49:117-119. Colorado Division of Wildlife. 1997. Colorado's endangered, threatened, special concern, undetermined status and candidate species -Terrestrial species. Draft Report for the Colorado Division of Wildlife. Colorado Natural Heritage Program. 1997. Colorado's Natural Heritage: Rare and imperiled animals, plants, and plant communities. Volume III, No. 1. Unpublished Report for the Colorado Natural Heritage Program. Compton, S. A, and RD. Hugie. 1993. Status report on z.apus hudsonius preblei, a candidate endangered species. Pioneer Environmental Services, Inc. Report for the USFWS. Logan, Utah Logan, Utah. Cranford, J. A 1983. Ecological strategies of a small hibernator, the western jumping mouse Zapus princeps. Canadian Journal of Zoology 61:232-240. EG&G. 1993. Report of Findings: 2nd Year Survey for the Preble's meadow jumping mouse. Prepared by Stoecker Environmental Consultants for ESCO Associates, Inc., Rocky Flats Environmental Technology Site, Jefferson County, Colorado ERO Resources. 1995. Environmental review of South Boulder Creek Management Area Prepared for City of Boulder Real Estate/Open Space Department. Prepared by ERO Resources Crp., Denver, Colorado in association with Stoecker Ecological Consultants, Boulder, Colorado. Fertig, W. 1997. Wyoming plant and animal species of special concern. Wyoming Natural Diversity Database. Laramie, Wyoming. Fitzgerald, J.P., C. A Meaney, and D. M. Armstrong. 1994. Mammals of Colorado. Denver Museum of Natural History, University Press of Colorado. Niwot, Colorado. Hafuer, D. J., K. E. Petersen, and T. L. Yates. 1981. Evolutionary relationships of jumping mice (genus Zapus) of the southwestern United States. Journal ofMammalogy 62:501-512. Hall, E. R 1981. The mammals of North America John Wiley and Sons, Inc., New York, New York, 2 volumes. Hamilton, W. J., Jr. 1935. Habits of jumping mice. American Midland Naturalist 16:187-200. Harrington, F. A., A Deans, M. E. Bakeman, and B. J. Bevirt. 1996. Recent studies of Preble's meadow jumping mouse at the Rocky Flats Site, Colorado. Journal of the Colorado-Wyoming Academy of Science 28(1 ): Abstract 18. Krutzsch, P.H. 1954. North American jumping mice (genus Zapus). University of Kansas Publications, Museum of Natural History 7:349-472. Levins, R 1970. Extinction. Lectures in Mathematical Life Sciences 2:75-107. Long, C. A 1965. The mammals ofWyoming. University of Kansas Publications, Museum of Natural History, 14:493-758. Meaney, C. A and N. W. Clippinger. 1995. A survey ofpreble's meadow jumping mouse (Zapus hudsonius preblei) in Colorado. Report prepared for the Colorado division of Wildlife. Meaney, C. A, N. W. Clippinger, A Deans, and M. OShea-Stone. 1996. Second year survey for Preble's meadow jumping mouse (Zapus hudsonius preblei) in Colorado. Report prepared for the Colorado Division of Wildlife. �41 Meaney, C. A., A. Deans, N. W. Clippinger, M. Rider, N. Daly, and M. O'Shea-Stone. 1997. Third year survey for Preble's meadow jumping mouse (Zapus hudsonius preb/ei) in Colorado. Report prepared for the Colorado Division of Wildlife. Olson, T. E., and F. L. Knopf 1988. Patterns of relative diversity within riparian small mammal communities, Platte River watershed, Colorado. Pg. 379-386 in: Proceedings of the symposium: Management of amphibians, reptiles and small mammals in North America Flagstaff, Arizona U.S. Forest Service, General Technical Report RM-166. Poly, W. J., and C. E. Boucher. 1997. Record of a creek chub preying on a jumping mouse in Bruffey Creek, West Virginia Brimleyana 24: 29-32. PTI Environmental Services. 1996a Preble's Meadow Jumping Mouse Study at Rocky Flats Environmental Technology Site, Annual Report 1996. Final. Rocky Flats Environmental Technology Site, Golden, Colorado. PTI Environmental Services. 1996b. Preble's Meadow Jumping Mouse Study at Rocky Flats Environmental Technology Site, Spring 1996. Final. Rocky Flats Environmental Technology Site, Golden, Colorado. Quimby, D. C. 1951. The life history and ecology of the jumping mouse, Zapus hudsonius. Ecological Monographs 21:61-95. Riggs, L.A., J.M. Dempey, and C. Orrego. 1997. Evaluating distinctness and evolutionary significance of Preble's meadow jumping mouse: Phylogeography of mitochondrial DNA noncoding region variation. Final Report for the Colorado Division of Wildlife. Denver, Colorado. Ryon, T. R 1996. Evaluation of historical capture sites of the Preble's meadow jumping mouse in Colorado, final report. MS Thesis. University of Denver, Denver, Colorado. Ryon, T. R 1997. The evaluation of historic capture sites of the Preble's meadow jumping mouse in Colorado. in Report on habitat findings of the Preble's meadow jumping mouse. Edited by M. E. Bakeman. Report to USFWS and Colorado Division of Wildlife. Sheldon, C. 1934. Studies on the life histories of Zapus and Napaeozapus in Nova Scotia Journal of Mammalogy 15:290-300. Sheldon, C. 1938. Vermont jumping mice of the genus Zapus. Journal ofMammalogy 19:324-332. Svihla, A. and R D. Svihla 1933. Notes on the jumping mouse, Zapus trinotatus trinotatus Rhoads. Journal ofMammalogy 14:131-134. USFWS. 1997a Proposal to list the Preble's meadow jumping mouse as an endangered species. USFWS 50 CFR part 17. USFWS. 1997b. Interim survey guidelines for Preble's meadow jumping mouse. USFWS. Denver, Colorado. Whitaker, J. 0., Jr. 1963. A study of the meadow jumping mouse, Zapus hudsonius (Zimmerman), in cental New York. Ecological Monographs 33:3. Whitaker, J. 0., Jr. 1972. Zapus hudsonius. Mammalian Species 11:1-7. Wyoming Game and Fish Department. 1998. Native species status classification system. Cheyenne, Wyoming. ��43 APPENDIX A Evaluating distinctness and evolutionary significance of Preble's meadow jumping Biosphere mouse: Phylogeography of mitochondrial DNA non-coding region variation. Final genetics Report to the Colorado Division of Wildlife inc Reply to: Box 9528, Berkeley, CA 94709 Tel: 510/531-4848 Fax: 510/530-0580 Email: bgioigc.org EVALUATING DISTINCTNESS & EVOLtrrIONARY SIGNIFICANCE OF PREBLE'S MEADOW JUMPING MOUSE: PIIYLOGEOGRAPHY OF MITOCHONDRIAL DNA NON-CODING REGION VARIATION FINAL REPORT DECEMBER 8, 1997 Submitted by Lawrence A. Riggs, Ph.D. John M. Dempcy, M.A. Cristian Orrego, Ph.D. Biosphere Genetics, Inc. Berkeley, CA Submitted to Judy Sheppard -Terrestrial Section Colorado Division ofWddlife Denver,CO �I • 44 Evaluating Distinctness & Evolutionary Significance of Preble's Meadow Jumping Mouse: Phylogeography of Mitochondrial DNA Non-Coding Region Variation Lawrence A. Riggs, John M. Dempcy, and Cristian Orrego SUMMARY We have generated molecular genetic data fotJ1:!~tal_Qf92.jup1ping mice including 71 sampled from 20 populations in Colorado and four populations in Wyoming, and 21 specimens representing three reference groups of closely related tua. DNA sequences obtained for a.433 basepair (bp)-long portion of the mitochondrial DNA non-coding region (sometimes referred to as the D-loop) in all samples indiQate that a group of populations ranging from southeastern Albany County, Wyoming, south along the Front Range of the Rocky Mountains to western Las Animas County, Colorado, fonn a coherent group, both genetically and geographically. This group, which we refer to as the "Preble's group" is distinct from four of five other populations sampled by live trapping or from museum specimens and may be somewhat differentiated from the fifth of these. The Preble's group is also clearly distinct from two of the three reference groups and less markedly differentiated from the third. Phylogenetic analyses based on these data are still at a preliminary stage and indicate that the Preble's group of populations is most closely allied with, and not strongly differentiated from, samples of Z. h. intennedius from Minnesota. Of the four populations that appear to be distinct from the Preble's group, two sampled near the Colorado-New Mexico border are closely allied with museum specimens identified as Z. h. luteus from New Mexico. Two others from the vicinity of Cheyenne, Wyoming, are most similar to representative samples of Z. princeps from western Larimer Co., Colorado and are next most closely allied with a Z. princeps specimen from northern New Mexico. A suggestion from our analysis is that Z. hudsonius from Indiana (possibly either Z. h. americanus or Z. h. intennedius based on distribution) may be the most ancestral of the populations and taxa sampled and may have shared a common ancestor with progenitors of fonns presently known as .Z: princeps princeps and Z. h. luteus. BACKGROUND • Recent applications of molecular genetic methods in systematics, population genetics, and conservation biology have demonstrated that variation in DNA markers can • be highly infonnative in the evaluation of genetic distinctness and evolutionary significance, two important determinants in decisions regarding application of the Federal Endangered Species Act of 1973 as amended and interpreted by the U. S. Congress. The study descnoed here sought to discoyer whether and how molecular data might support and help objectify the view, based on other more traditional criteria, of the Preble's mouse as an evolutionary unit distinct from other species and subspecies in the genus 7,apus. I, �45 Preblc's DNA study Final Report/12-8-97 Work funded by the Colorado Division of Wtldlife (hereafter CDW) and performed by Biosphere Genetics, Inc. has evaluated the ability of three different approaches to quantifying variation in the DNA molecule to provide reliable and infonnative data addressing issues of key importance in the decision of whether or not to list the Preble,s mouse. Previous reports to CDW have summarized our finding that RAPD (randomly amplified polymorphic DNA) markers are not sufficiently repeatable for use in the context of this study. Methods to assay variation in ~other class of genetic markers found in nuclear DNA called microsatellites or simple sequence repeats, although •potentially promising, are not yet sufficiently well developed and validated for zapodids to apply at this time. A third approach, DNA sequencing of the mitochondrial DNA noncoding region which includes the D-loop, emerged from preliminary work as the method most appropriate and informative, in combination with other more traditional criteria, for the determination ofwhether populations of the Preble's meadow jumping mouse (Zapus hudsonius preblei) constitute one or more distinct evolutionary units with significance warranting protection under the Federal Endangered Species Act. Results based on sequence variation in the mitochondrial DNA non-coding region assayed in a subsample of the 54 mice presumed to be Preble's obtained during the summer of 1996 indicated: (1) that 7 populations sampled at sites ranging from Longmont to Colorado Springs in Colorado varied relativ~ly little from one another,, (2) that two other populations sampled in Las Animas County near Lake Dorothey were similar to one another but differed quite markedly from the first group, and (3) that none of the individuals sampled in Laramie 'l ~ County in Wyoming could be assayed using the techniques successfully applied to the other samples, suggesting that this population may also differ to some degree from those in Colorado. Field surveys of the occurrence of the Preble"s meadow jumping mouse conducted by several individuals and organizations during the summer of 1996 were extended to additional sites in the summer of 1997. CDW spo0$0red some of these efforts and coordinated the collection, recording and· delivery of tissue samples for DNA analyses performed by Biosphere Genetics, Inc., also under contract with the Division. In addition, the U. S. Air Force supported DNA analysis of samples _obtained from trapping efforts during the· summer of 1996 on the Warren Air Force Base and the inclusion of reference material representing two additional subspecies of Zapus hudsonius,, Z. k palustris, and Z. k campestris. • METHODS Tissue Sampling and Storage A sampling protocol and data forms for using in obtaining samples for.DNA analysis were developed for CDW and distnouted to all parties participating in the coordinated trapping effort. The disseminated protocol, adapted and updated from the •U.S. Fish and Wtldlife Service standardized protocol, descnoed handling and instrument cleansing procedures to minimize contamination of the sample with human or other mammalian.DNA Very small samples of tissue were obtained from live-trapped �I. 46 Prcble's DNA study Final Rci,ort/12-8-97 • specimens of mpz(s by using a tagging punch (National Band and Tag Co., Newport, KY) to remove one or more plugs of tissue from the outer portion of each mouse,s ear. In some cases, ear tissue was obtained by clipping or notching the ear with scissors. Tissue plugs or pieces from a given animal were placed in a cryogenic screw-cap vial filled 1/31/2 full with 95% ethanol and labeled with an alpha-numeric code intended to uniquely specify trapping location and the individual animal caught. Hair samples were collected by some participants and placed either in a separate labeled vial or ~ the same vial containine the ear tissue. Samples were conveyed to CDW and accumulatec;{ into lots for deliveiy to BGL When ·samples were transported from the field or shipped by common canier, they were at ambient temperature. Upon receipt at BGI, samples were kept refiigerated at 4°C until tlie time of DNA extraction. Specimens obtained from field surveys and miscellaneous museum specimens used to expand coverage of the potential range of the taxa examined in this study are listed in Table 1 of the Appendix to this report. Reference specimen tissues used for comparison with field-sampledjumping mice were provided from four separate museum collections (see Appendix I, Table 2) as toes clipped from museum specimens or as small pieces of internal organs maintained iri the museum,s frozen collection, usually at-70°C. Toe clips were shipped at ambient temperature and refiigerated at 4°C upon receipt. Frozen specimens were shipped on dry ice and extracted upon receipt with any remaining tissue being transferred to ethanol for further storage. Reference specimens are listed in Table 2 of the Appendix to this report. DNA Extraction & Purification We extracted DNA from plugs or pieces of ear tissue using the Hair Lysis Buffer (HLB) method similar to that of Greer et al. (1995). The tissue was rinsed, either with a jet of triple-distilled water or by agitation in a drop of distilled water on Parafilm®, then placed in a volume ofHLB and incub~ted at S6° C with periodic agitation for 10-12 hours. In most cases, this process completely digested the sample leaving only a trace of particulate matter suspended or settled in the bottom of the tube. With the particulates centrifuged to the bottom, aliquots of the extract were withdrawn for fractionation and for purification prior to use in DNA amplifications. Raw DNA extracts were run out on 1.5% agarose gels and stained with ethidium bromide to quantify the DNA by comparison with staining intensity of Lambda Hind m fragments run on the same gel. We used Prep-AGenen.1 (Bio-Rad, Hercules, CA) to purify the DNA preparations before sequencing. Amplification of Mitochondrial DNA Non-coding Region Sequences We began our examination of the mitochondrial non-coding region using the primers designated as ECO-THR (L15996) and BAM-TDKD (Hl6401) to amplify an approximately SSO bp-long segment between the threonine tRNA gene near the •cytochrome b end of the non-coding region and the TDKD site approximately half way· through the non-coding region. 1 This primer combination, first used to study human 1 The ECO- and BAM- portions of these primer designations refer to.the addition of restriction enzyme sequences at the 5' end of the primer sequence. These "tails" are not essential for the present application and we will use primer designations with and without this notation interchangeably in this report. Biosphere Genetics,. Inc. l>age 3 �47 Final Report/12-8-97 Preblc's DNA study populations (Vigilant et al. 1989) had worked well in studies of other mammals and typically revealed amounts of sequence variation appropriate to differentiating subspecies and geographically isolated populations. Because the non-coding region includes a structure descnoed as a displacement or "D" loop known from work on human and cattle DNA, we will occasionally refer to the region of DNA studied here as the D-loop. Amplifications with the above primers were performed in 12.5 µI reaction volumes using 1 µI oftemplate DNA, 1 unit of the DNA polymerase, Klentaq-1 (Ab Peptides, St. Lo~ MO), and other reaction components in standard proportions. A Teg~e PH~2 . thetmocylcer was used to implement a program using an initial denaturation at 94°C for 3 minutes followed by 42 cycies with denaturing at 94° for 45 seconds., annealing at 55° for 1 minutes, and extension at 72° for 1 minutes followed by a final extension at 72° for 5 minutes. Pro4ucts were fractionated on 1.5% agarose gels. ~plification ofthe 550 bp D-loop fragment was successful, but primer dimer and secondary products also appeared, sometimes in considerable amounts. We experimented with a variety of modifications to obtain cleaner product (prim~ reduction, additions ofMgCh at various levels, alternative forms ofDNA polymerase, etc) and were ultimately able to obtain product with a • minimum"ofthese additional elements ii1 many, but not all, samples of mice presumed to be Z h. preblei. Following the optimization effort, revised reaction conditions were used to assess D-loop fragment amplification success with all new DNA extracts in 12.S µI reactions, then applied with slight modifications to the thermocyler program to amplify targeted products in SO µI reaction volumes to obtain sufficient quantities for automated sequencing. Two or three SO µI reactions were run for each sample, depending on prior estimates of the amount of template available in individual DNA extractions~ The quantity and quality of product generated by each reaction was assessed by agarose gel fractionation. Reaction volumes were combined when necessary, then cleaned with the solid-phase reversiole immobilization method (DeAngelis et a 1995) using carboxyl coated magnetic particles (PerSeptive BioSystems Inc, Cam6ridge, MA). Oeaned, resuspended DNA products were then dried down to a pellet mthe bottom of a 1.5 ml eppie tube for shipping by overnight service at ambient temperature to automated sequencing s~ce. an Resolution of Amplification Problems Some DNA extra~ most notably those of Z. princeps and of putative Z k , preblei populations sampled at the Warren Air Force Base in Wyoming, did not amplify well or at all with the ECO-THR/BAM-TDKD primer combination. We first suspected that this OCCWTecl because the TDKD primer designed from sequence information on the human D-loop was not finding sufficient complementarity on the heterologous primer site in at least some 7.ap,IS populations or taxa. Conventional wisdom suggested that the ECO-nm. primer site should be-highly conserved in vertebrates. Based on this reasolling, we designed two ~ew primers for the middle of the D-loop. We examined D-loop sequences available from Genbank for rat (Rattus norwegicus), mouse (Mus musculus), vole (Clethrionomysglareolus), gopher (Thomomys bottae), and kangaroo rat (Dipodomys ordii) as well as our own sequence data for ?.opus. Sequences for the Paee4 �48 Final Report/12-8-97 · relevant portion of the D-loop for Mus, Clethrio11omys and Zapus were used to identify a primer site most likely to be reasonably conserved in muroid and dipodoid rodents. The primers designed for this new site are designated PRDL L-15738 (forward primer) and PRDL H-15720 (reverse primer). We had these primers synthesized (Operon Technologies, Alameda, CA) for testing and possible use in resolving the difficulties mentioned above. The relative positions and priming directions of all primers are shown in Figure 1. We used PRD¼.!!_-1S~~0 iq coµibination with ECO-THR. in an attempt to amplify those samples that had not.amplified with the ECO-THR/BAM-TDKD primer combination. We also attempted to _amplify the entire D-loop by using ECO-THR. in combination with PHE-rev H29, a primer previously designed for general use in amplifying portions of the D-loop in mammals (A. Gavazzi, unpublished). Fmally we ..test~ PRDL L-1S738 in combination with PHE-rev H29 for the pwpose of generating D. loop fragments spanning the TDKD site for automated sequencing of this site in 'Zapus. :resting of these primers on 7.apus DNA extracts revealed that: 1) the ECO-THR/PRDL H-1S720 primer combination amplified in only two ~f20 samples and did nothing to solve the previous difficulties encountered with the ECO-THR/BAM-TDKD combination; 2) the ECO-THR/PHE-rev H29 combination produced a fragment ca. 1500 bp long, apparently spanning the D-loop successfully in some cases, but not with the DNA extracts that had been previously problematic, and 3) the PRDL L-15738/PHE-rev H29 combination produced multiple bands in some cases and no amplification in others. We inferred from these results that our original problem in amplifying some Zapus samples, most notably those assigned to. Z. princeps and the samples from the Warren Air Force Base, was not with the lDKD site, but rather with annealing of the ECO-THR primer designed for a conserved section of the human threonine tRNA gene to the het~rologous sequence in 7Apus. Reexamination of the sequences successfully obtained in preliminaly work on 'Zapus suggested that a primer site that would be more highly-conserved between 7.apus and other mammals might be present internal to the one primed by ECO-THR. Further testing showed that an available primer designated PROC, for priming of a sequence on the praline tRNA gene, when used in combination with the BAM-TDKD primer, was uniformly very successful in amplifying a 433 bp fragment (excluding primer sequences) of the D-loop. While we have been reluctant to give up on efforts to access -information about variation existing in and around the THR site, the PROC/BAM-TDKD primer combination was clearly the tool of choice for compiling the dataset required for this study within the time frame required. The results presented in this report are for the 433· bp segment of the D-loop falling between the proximal ends of these primer ~ites. ca. Automated Sequencing DNA sequencing of both the S50 bp and 433 bp fragments of the D-loop generated as descnoed above was perfonned by The University of Maine DNA Sequencing Core Facility using an ABI model 373A Stretch DNA sequencer. Results of . one.sequencing pass in each direction were received by e-mail as attached files in ABI •format. Pherograms in these files were viewed with the program Chromas (v. 1.4, Co1:1or Biosphere Genetics. Inc. Pages �49 Preblc's DNA study 1HR PROC Final Rcport/12-8-97 PRDLL15738 ldtNAua! lcRNAno! _I_______N_•_oo_&a_sepaa_·_ _ _ _ _ _ _! ,cRNA,_! ~ PRD H-1S720 ~ IDKD PHEA. Fig. 1. Relative positions of the primers used in this study for amplification of the mitochondrial DNA non-coding region. which includes the D-loop. Primers nm. and IDKD were from Vigilant ct al. (1995), PHE-rev 1129 was from Anita Oavazti and primers prdl L-15738 and prdl H-1S720 arc two new primers designed for this study to work with .2apus and other rodents. Primers arc used in pair-wise combinations to amplify segments of DNA lying between their positions on this simplified map. The relative positions of coding and non-coding regions arc indicated by labeled areas. McCarthey, Griffith University, Brisbane, Queensland, Australia) and machine-assigned sequences were exported as text files for use in subsequent analysis. Data Analysis Text files containing the two sequencing reads generated from opposite ends of the amplified fragment were entered into the program MacDNASIS (v. 1, Hitachi Sofl:ware Engineering Co., Ltd., San Bruno, CA) and combined to obuun a consensus sequence for the amplified region of each individual mouse sampled. Sequ~®S obtained from both THR-IDKD and PROC-Ti>KD amplifications were used as available. However, since the more inclusive THR-TDKD sequences were not available for individuals assigned to ·z princeps and samples ultimately found to be more closely related to Z princeps than to the Preble's samples, and since the longer sequences were of variable quality at the beginning of the 5' to 3' read in a number of the Preble's samples, we~ only the 433 bp sequence available in all samples in all subsequent data analysis. Consensus sequences were aligned manually using the multiple sequence option of MacDNASIS. The program MacClade (v. 3, Maddison, 1992) was used to create· a data matrix of character states for 58 variable sites in the aligned sequences for phylogenetic analyses. Phylogenetic trees were generated using the parsimony algorithm in PAUP (v. 3.1.1, Swofford, 1993). To expedite review of various options for the rooting of trees; one individual was selected as most representative of each population (fewest changes frQm the most commonly shared state at the SS variable sites treated as cbaracters in the .DNA sequence) and used in the depiction of,nost-parsimonious trees and in bootstrap analyses. Bootstrap analyses used the 50% majority rule with 100 replications. Biosphcn: Genetics. Inc. Page6 �sircble's DNA study Final Report/12-8-97 Strict £8§1i8~ iii ~ lffii Iii I • ·---. ~ mi ™ 00 ~OSN I II m§ I I I I I ~ -r.= ========= ~--~~ . __ _ _ _ _ _ _ _ _ _r--_- _ -_ -_-_ -_ -_ -_ --t I I Figure ·2. Strict consensµs tree, identical to_ the single most parsimonious tree generated using the "branch and bound" option of PAUP v. 3.1.1 (Swofford 1993) on.sequences for all samples analyzed. Length of this shortest tree is 61 evolutioruuy steps; consistency index (Cl)= 0.869. Although not supported by bootstrap analysis in the conte.u of phylogenetic analysis, sub-groupings of populations considered here to be part of the Preble's group suggest that population subdivision may be discemable by more appropriate population-analytic methods. Biosphere Genetics. Inc. Pagc7 ,. �51 Final Report/12.S-97 Preble's DNA study RESULTS Figure 2 illustrates phylogenetic relationships among all samples included in the study to date, inferred from maximum parsimony analysis (length = 61 steps, consistency index [CI] =0.869). Bootstrap analysis for the entire dataset was not practical for this report. However, bootstrap analyses were performed for various outgroup scenarios using one individual from each population as described above. No outgroup was designated for the analysis shown in Figure 2. ·-·The important relationships in Figure 2 are perhaps more easily seen in the unrooted, circular fonnat tree generated using one representative individual from each population (Figure 3). •No meaning is attached to the length ofbranches hi this tree. The 19 branches that join the baseline in the center directly (11 o'clock to S o'clock on the diagram) and four more samples that ·are shown grouped together and joining the baseline with• a bootstrap value of 53 (indicating little support for treating this group of samples as distinct from the other 19) form a relatively homogenous group. Also not strongly diff'erenti~ed is the group of three r~p~enting reference samples of Z. h. intennedius from Minnesota and Z. k campestris from South Dakota as well as an individual museum . specimen designated as only Z. hudsonius from Johnson County, Wyoming. Bootstrap values indicate strong support for reco~g samples of Z. princeps, Z. k luteus, and populations affiliated with them in the diagram as distinct from the Preble's group. Further observations following from the analysis are discussed in the following section. DISCUSSION Preliminary data analyses conducted before all new populations sampled during the summer of 1997 and reference material had been analyzed ex_a.mined three options for the rooting of the phylogenetic tree. When the Indiana population of Z. hudsonius was assigned as the outgroup (one most parsimonious tree, length·= 63, CI= 0.921), the strong relationships in the tree were the same ones seen in the tree generated when no outgroup was assigned (Figures 2 and 3). Essentially, populations ofjumping mice ranging from southeastern Albany County in Wyoming south along the Front Range of the Rocky Mountains to western Las Animas County in Colorado were seen to fonn a coherent group within which separate populations or groups of populations were not strongly differentiated. Designating Z. princeps as the outgroup resulted in Indiana Z. . hudsonius appearing as the sister group to the assemblage of presumed Preble's populations with the populations from the Colorado-New Mexico border· and the Warren Air Force Base joining the tree at an intennediate level. Bootstrap values for the resulting tree were uniformly high, but we have not yet resolved doubts about whether Z. prlnceps can properly be treated as ancestral to Z. hudsonius. Finally, designating Z. hudsonius intennedius from Minnesota as the outgroup in earlier analyses resulted in a tree in which all the branches to the Preble's popuiations samples joined the baseline as undifferentiated from the outgrqup. Biosphere Genetics. Inc. Pages �52 Prcblc's DNA study Final Report/12-8-97 MRD97<flMNH9716 ROX9701 CGA1 80S9706 ·--- EPC9701 LOSt - - RF96100 RSS1076 -------------- Figure 3. Bootstrap consensus tree (50% majority rule, 100 replications), based on analysis of single most-representative individual from each population or reference group, obtained with no outgroup designated. ijootstrap values are shown next to relevant branches in the tree. Bootstrap values greater than 90% indicate those relationships which are strongly supported. • • Bwsphcrc Genetics. Inc. ·' • �Preble's DNA study Final Report/12-8-97 There are several notable observations for individual population samples among our findings based-on analyses completed in late November, 1997. We summarize these here in list form with population/sample codes in parentheses referring to designations in the tables of the Appendix: 1. The six jumping mice caught at the Warren Air Force Base (CCF) in the summer of 1996 (identified by field workers as Preble's) and an individual jumping mouse caught in Weld County, Colorado just south of Cheyenne, Wyoming (LTC), are indistinguishable from reference samples ofZ. princeps (CZP) provided by Dr. B~ce Wunder from Larimer County, Colorado. Bootstrap values indicate that this group is, in tum, most closely affiliated with, but distinct from, a sample of Z. princeps from Taos·county, New Mexico. • 2. The eastem-m~st sample identified in the field as Z. h. preb/ei, a single individual from Elbert County, Colorado (HAY), is confirmed to belong in the Preble's group by DNA sequence obtained from a sample of three hairs. 3. Jumping mice caught at two sites in the Lake Dorothey area on the Colorado-New Mexi~ border (CCR and WFS), suspected on the basis of morphological traits to be similar to Z. h. luteus (Cheryl A Jones, personal communication), are confirmed to be so on the basis of the non-coding region sequence data. There is also a reasonably strong indication (bootstrap value·= 88) that these two populations and the Z. h. luteus samples are, together, likely to have shared a common ancestor with progenitors of Z. princeps reference samples and the populations sampled north (CCF) and south (LTC) of Cheyenne, Wyoming. This finding is in contrast to the conclusions ofHafher et al. (1981) (but see next comment). 4. A suggestion from our analysis (not strongly supported by bootstrap analysis) is that Z. hudsonius from Indiana (possibly either Z. h. americanus or Z. h. intennedius based on distnl>ution) may be the most ancestral of the populations and taxa sampled and may have shared a common ancestor with progenitors of i'ornis presently known as Z. princeps princeps and Z. h. luteus. We speculate that the forms recognized as subspecies of Z. princeps and Z. hudsonius present today in the Rocky Mountain and Northern Great Plains region may be the derivatives of two separate range expansions occurring at different times during the Pleistocene. 5. Two ·rererence samples obtained from the Denver Museum of Natural Histoiy, originally identified as Z. princeps from the San Isabel National Forest-in. western Las Animas County, appear on the basi$ of non-coding region sequence data to belong to the Preble's group. 6. A single jumping mouse specimen obtained by Chris Garber from the Medicine Bow National Forest in Wyoming, identified as a Preble's mouse by Dr. David Annstrong on the basis of his analysis of poorly preserved skeletal material, is confirmed to be a Preble's by DNA sequencing from a piece of preserved skin. 7. One reference collection sample identified as Z. h. pallidus from Garden County, Nebraska (ZHPNEG), and two samples identified as .Z: h. campestris from ~eston Biosphere Genetics. Inc. Page 10 53 �I. 54 Preble's DNA study FinalRcport/12-8-97 County, Wyoming (ZHPWYW), are indistinguishable from other samples in the Preble's group in the present analysis. 8. Two reference collection samples identified as Z. h. campestris from Custer County, South Dakota, are found to be most similar to a museum specimen identified only to species level as Z. hudsonius from the vicinity of Buffalo in Johnson County, Wyoming. • Further Researc~ Needs ·--Additional sample.materials and analyses may be required to achieve a rigorous view of the relationships of the group referred to here as Preble's to other subspecies and species of the genus Zapus in North America.- The choice ofan appropriate outgroup remains unresolved, and could benefit from further input from those knowledgeable about evolution and morphology of the Zapodids in North America and elsewhere. Based on extensive work on limb myology of the Dipodoidea (Stein, 1990), the superfamily to • which the genus and the rest of the family Zapodidae belong, the genus Sicista (the birch mice), represented by species in Europe and Asia, would appear to be an appropriate outgrouj>. However, divergence between 7.apus and Sicista may be too great to be effectively measured by variation in the D-loop. Another North American genus, Napeozapus, has been suggested as a possible outgroup for this analysis. However, slightly greater adaptation for saltatorial ability and a higher tooth count in Napeozapus than in Zapus would suggest that the former is derived rather than ancestral to the latter. Sequence information for the cytochrome b gene potentially forthcoming from other researchers (T. Yates, personal communication) may be helpful in sorting out relationships at the species and higher taxonomic levels and may be a useful complement to the noncoding region sequence data in elucidating relationships potentially traceable to • Pleistocene glaciation events. We have begun to examine a portion of the cytochrome b gene and the neighboring threonine gene in a separate study. . The DNA sequence results reported here indicate the ~eed for a reexamination of external and skeletal morphology in Zapus. .In at least a few cases (DMNH, LDS), the sequence data suggest that .even species level identifications cannot always be made unambiguously in the field. Patterns of morphological variation across the range of the species and subspecies of interest may be such that more careful examination of museum accessions is to be recommended as well. One promising dental character that may distinguish Z. hudsonius·from Z. princeps is the presence of a deep anteromedian fold 'in the anteroconid of the molar ml in the former as descnoed by Klingener (1963) and applied in an analysis of cave floor faunal remains by Hafner (1993). It will be interesting to learn what this character may say about the taxonomic identity of the. samples referenced above and about specimens available in museum collections for the areas near the northeQt and sotithem boundaries of the Preble's mouse range. • Further analysis of molecular markers at the species, subspecies, and population levels is needed to confirm the findings suggested by this study and to further delineate patterns of variation bearing on conservation and management ofZ. h. preblei. We have material in hand to continue to expand sample sizes for a number of the populations already swveyed and our initial contacts with several museums in the context of the Biosphere Genetics, Inc. Page 11 �Preblc's DNA study Final Rcport/12-8-97 55 present study indicate that sample sizes of comparative taxa can also be increased for at least some parts of the relevant subspecis ranges. Because there has been relatively little recent suivey activity in Wyoming and only a few reference specimens from museum collections were included in this study, the northern extent of populations assignable to the Preble's group is poorly defined. New trapping suivey initiatives and DNA analysis of more material available from museums would address this situation. We also recommend that a nuclear DNA dataset be generated to complement the mitochondrial DNA dataset. A portion of our work not reported here has evaluated two classes ofDN~ markers potentially appropriate to tills task and b ~ applications of ipicrosatellite methods to development of markers likelyto be applicable both to phylogenetic analysis of relationships at the species and subspecies level and to population genetic analyses supporting assessment and monitoring of population differentiation and gene flow, genetic ·parameters of population viability, and mating system analysis. • We expect that further examination of the results obtained in this study, both by ourselves and others, will lead to a more comprehensive definition of research needs and priorities along two possibly divergent paths, one inclined to address fundamental issues of systematics and biology, and the other focused more sharply on the near-term requirements that may be imposed by the outcome of the pending decision on the status of the Preble's mouse by the U.S. Fish and Wddlife-Service. We urge that planning attention be devoted to providing for both ongoing research and for integration of assessment and monitoring activities into any prospective recovery effort.. Molecular ecology has much to offer in both areas of endeavor. . Conclusion Based on resuits.to date, we conclude that mitochondrial DNA non-coding region (D-loop) sequence data appear consistent with the view that a geographically contiguous set of populations previously recognized as the Preble's mea<Jow jumping mouse (Z. h. • preblei) form a homogenous group recogniz.ably distinct from other nearby populations and from another geographically-adjacent species of the genus. Biosphere Genetics. Inc. Page 12 �56 Preble's DNA study Final Report/12-8-97 REFERENCES CITED DeAngelis, M.M., D. G. Wang, and T. L. Hawkins. 1995. Solid-phase reversible immobilization for the isolation of PCR products. Nucleic Acids Research 23{22):4742-4743. Greer, C.E., C.M Wheeler, and M.M. Manos. 1995. PCR ampijfication from Paraffinembedded tissues: Sample preparation ~d the effects of fixation. pp .99-112, in: PCR Primer, A Laboratory Manual, C.W. Dieffenbach and G.S. D.veksler (eds.. ) CHSL Press, New York, 1995. Hafner, D.J. 1993. Reinterpretation of the Wiconsinan mammalian fauna and paleoenvironment ofthe Edwards Plateau, Texas. J. Mamm. 74(1):162-167. Hafner, D.J., K.E. Petersen, and T.L. Yates. 1981. Evolutionary relationships ofjumping mice (genus Zap11s) of the Southwestern United States. J. Mamm. 62(3):501-512. Klingener. D. 1963. Dental evolution of Za.pus. I. Mamm. 44:248-260. Maddison, W.P. and D.R. Maddison. 1992. MacClade v. 3: analysis of phylogeny and character evolution. Sinauer Associates, Inc. Sunderland, Massachusetts. Stein, B. R. 1990. Limb myology and phylogenetic relationships in the superfamily Dipodoidea (birch mice, jumping mice, arid jerboas). Z. zool. Syst. Evolut.-forsch. 28:299-314. Swofford, D.L. 1993. PAUP - Phylogenetic Analysis Using Parsimony- version 3.1.1. Computer program distributed by the Illinois Natural His~ory Survey. Campaign, Illinois. Vigilant, L., R. Pennington, H. Harpending, T.D. Kocher, and AC. Wilson. 1989. Mitochondrial DNA sequences in single hairs from a southern African population. Proc. Natl. Acad. Sci. USA, 86:9350-9354. Walsh, P.S., D.A. Metzger, and R. Higushi. 1991. Chelex® 100 as a·medium for simple extraction of DNA for PCR-based typing from forensic material. Biotechniques. 10(4):506-513. Biosphere Genetics. Inc. Page 13 �57 Appendix Tables in this appendix contain information on the populations and species samples included in • this study. All samples received ·for which DNA was extracted are listed.· Those for which DNA. sequence data was successfully obtained for the analyses reported in. the main document are indicated as explained in each table. Additions and corrections were still being received as the final draft of this report was going to press. Questions about incomplete or inaccurate data should be referred to the Colorado Division ofWddlife, Terrestrial Section, Attn: Judy Sheppard. A "°"'"nrliv A �Table 1. Sampling locations for jumping mice analyzed as put~tive Preble's mice as {Zapus hudsonius prcblei). This table includes both samples obtained by live-trapping during the summers of 1996 and 1997, and samples from museum specimens not identified to subspecies or otherwise designated as reference material (included in Table 2). Numbers of samples for which DNA sequence data were successfully obtained in time for the analyses presented in the main body of the report arc shown in parentheses under the number of samples processed. The individuals included in the analyses arc indicated·by an.asterix in the column headed "Original sample #s." S1D1plo Codo BAD BOS Locality/site name & desaiption or other soorce informltkin B■dwttcrCredc, Natrona Co., Wyoming . 'oNIAl rein.and SoUlh BoalderCredc. near intcnedion with Eat Boulder Dltdi, on City ofBoulder Open Sptee. Boulder Co;. CO. 5300 ft. Lousivi'Uo Quadrangle, T J S, R 70 W No. Original • -1011. 1 m CU13469• 13 B0S9702 (1) • BOS9703 BOS9704 B0S9105 BOS9706• 8089707 BO89708 BOS9709 BO89710 BOS9711 BO89712 BO89715 BO89716 4 (2) BVC Bemr Credc, 2 1/2 milea SW of Monument, El Pllo Co., CO. 7000 ft. Palmer Lake quadrangle. T 11 S, R 68 W CCF Crow Credc, Wamn Air Form Bue, Laramie 6 Co., WY. [elmitionanclquadrangloNIA) T14N,R67W (6) CCR Crlo:,rica Credc, Lako Dorothay Stlto W"ddlifo Area, LIi Anhnn Co., co DC CCF9604• 6 (4) N Brandl, Middle Fode. Lodgq,olo Credc. Polo Mt. Umt, Modi.cine Bow National Foreat, ce. 2 ml. S, 15 ml. B.Laramie, Aibeny Co., WY, 7600ft. -•o•NIA1 Deaclman Credc, U.S. Air Foroo Mademy, El 1 (1) PuoCo..CO, (2) {elevation and qu■dnmglo NIA) CCF9602• CC~3• [Elev.-?) CGA BVC9701 • BVC9702 BVC9703• BVC9704• CCF960l• 5 CCF9605* ·CCF9606• 432• 433• 503 504 .. 505• 506• [Specimen tag] 97DC02N 97DC03N .97DCQ2s• 97DCOSS• 97DC07S 1/4-1/4 Section NIA Lat-Loog coordinirtcs NIA 3, W 1/2 3, Wl/2 3, Wl/2 3, Wl/2 3, W 1/2 3, Wl/2 3, Wl/2 3, Wl/2 3, Wl/2 3, Wl/2 3, Wl/2 3, W lf.2 3, W 1/2 39"59' 30" N . 105°12'55" w 29.NE 1/4 29,NEl/4 29,NE 1/4 29.NE 1/4 27 27 27 27 27 27 NIA 39"03'27.2" 104°53'49.8" NIA 39"00'02" 104°21'39" UTM Coords. NIA 4818450.00, 4427090.00 4811450.00, 4427090,00 4818450.00, 4427090.00 4818450.00, 4427090.00 4818450.00, 4427090.00 4818450.00, 4427090.00 4818450.00. 4427090.00 4818450.00, 4427090.00 4818450.00, 4427090.00 4818450.00, 4427090.00 4818450.00, 4427090.00 4818450.00, 4427090.00 4818450.00, 4427090.00 Collettor(1) Spoosorin&'Participating 0rl!llnizlrtion(1) P.Robfntcn&E. ·Andenon C. Meaney & N. Clippinger Univ. of Colorado Museum. Boulder. CO Colorado Division of Wildlife Carron Mean.:y, Consultant i i 5088970.00, 4323168.00 5088970.00, 4323168.00 5088970.00, 4323168.00 5088970.00, 4323168.00 NIA N. Clippinger NIA C.AJones C. Flemming C.Paguo • Colorado Division of Wildlife Carron Meanc:y. Consultant Colorado Natural Heritage Program U. S. Air Force Denver Mll!eum of Natural History ' TlSN R71 W sec 11 NIA NIA NIA NIA C.O■rber (collector) D. Aimstra:tg 4318360.00. 513700.00 4318360.00. 513700.00 4318360.00, 513700.00 4318380.00, 513650.00 4318300.00, 513650.00 P.Sd!uennan, s. Aslc, University Museum, Univ. ofColorado Colorado Naturill Heritage Program J. Hobert Paee A-2 U1 (X) �i Table 1 continued (2nd page). EPC Eat PJmn Creek. 6500 ft. Doagtu Co., CO. DawlanButtoquadranglo, T9S,R67W BAY JCC IDS I.PC Hay Ouldl, tn'butaryto Running Crock near Mer, Elbat Co., CO. 6,225ft. Cabin Ouldl 'e, T7S.R64W. Coll Creek. R.lnlom'Edwmk Homeateld Ranch, Jeffinon Co., co. (Bin.•?) IEdondo-~- 00) T 2 S. R 70 W Lib De Snxtnear Baflitlo. Johnson Co., WY. [elmdlon and quadranglo NIA] I 3 (1) BPC9701• BPC9702 EPCf703 9,E 1/2 9,E 1/2 9,E 1/2 I (I) HAY-1• E 1/2 NIA 4 (3) 9601* 9602• 9603 9604• CU15132• 7,8,17,18 [Information neededftom collecton] [Information needod) C.Flcmming C.Plgue NIA NIA NIA LPC970l• LPC9702• LPC9703• LPC9704* LPC970.5 LPC9706 LPC9707 LPC9708 LPC9709 LPC9710 LTC-1* 4,SW 114 9NW,NW 4,SW 1/4 9NW,NW 4;SW1/4 9NW,NW 4,SW 114 9NW,NW 4,SWl/4 9NWNW NENE,SE SE 40046'11.0" 10.5°21'36.7" 4696010.00, 45132560.00 4696010.00, 45132560.00 4696010.00, 45132560.00 4696010.00, 45132560.00 4696010.00, 45132560.00 4696010.00, 4.5132.560.00 4696010.00, 45132560.00 4696010.00, 4.5132560.00 4696010.00, 45132560.00 4696010.00. 45132560.00 5063300.00, 45333900.00 A.Crockett J.MMerritt D. Armstronst A. Deans MRD970l• 16,NW 1/4 39"58'1.6" 105°13'.59.7'' 4798400.00, 44242200.00 A. Deans 21,NW 1/4 18, SE 114 1.5,NW 1/4 10,SW 1/4 10,SWt/4 10, SW 1/4 10,SW 1/4 10.SWl/4 40049'2.5.6" 105°21"8.6" 471.5010.00, 4.5212170.00 471.5010.00, 4.5212170.00 471.5010.00, 4.5212170.00 4702850.00, 4.5194120.00 471.5010.00, 4.5212170.00 4715010.00, 4.5212170.00 4702850.00, 4.5212170.00 47028.50.00, 4.5212170.00 A. Dean's Colorado Division of Wildlife Carron t..-1.:an,:y. Consuhant T.Ryon PTI Environmental Services I (I) Lem PinoCreek. Lowerai«okeo State WildHfl, Area. Llrlmer Co., 00. 6200 ft. Livamoro Mountain quadrangle, T 10 N. R 71 10 (4) Lem Tree Creek. at 1-25, llOl1han podim Weld Co., CO. 5,900 ft. Carr Soulhwest quadrangle, T. ll-12N.R.67W. ManhaD Road, Dry Crock Ditdl #2, bctwem Mmball road and South BoulderCredc. Boulder Co., CO. 5700 ft. Louimllo • TlS.R70W Rabbit Creek. Lower C!erokeo SWA CL 9 mi. NW Uwnnoro, La1'.im«Co. 00. 6300 ft. Livamoro Mountain quadranglo ,T 10 N, R 71 1 (1) 8 (4) RBC9701° RBC9702• RBC9703• RBC9704• RBC970.5 RBC9706 RBC9707 RBC9708 Rocky Flm &tvha11nmtal Tedmology Sito, 1cfflnon Co. 00. 5780ft. Loufmllo, CO, T2 S, R 70 W 10 (6) ZAHl196-62• ZAHU96-63• w LTC MRD . RBC t (l) Atmendbc A 40057' 5.8" N 104"55'29.3" W 5090040.00, 4347802? 5090040.00, 4347802? 5090040.00, 4347802? 509.0040.00. 4347802? 540660.00, 4368470.00 (?) ADeana A.Dims A.Deana . w RF 39"16'46.3" N 104°53'44.2" w NIA ~· ZAHU96-6.5• HU96-100• HU96-10l• RF-97-102 RF-97-103 RF-97-104 RF-97-10.5 11,.5W 11,.5W 11,.5W 11.,w Colorado Division of Wildlife Canon Meaney. Consuhant Colorado Division of Wildlife Carron Meanev. Consuhant Colorado Natural Heritage Program Univ. of Colorado Museum. Boulder.CO Colorado Division of Wildlifo Carron M~ey. Consuhant Colorado Division of Wildlifo Carron Meanev. Consuhant Colorado Division of Wildlife Carron M.:aney. Consuhant l11 4832800.00, 44148500.00 4832800.00, 44148.500.00 4832800.00, 44148.500.00 4832800.00, 44148.500.00 ID Pa!.'c A-J �°'0 Table I, continued (3rd page). Roxborough Statol'mknearjundim of Willow Creek and UttJo wmow Croelc, Dougln Co., CO. 7300 ft. Xnsl« quadrangle, T 7 S, R 69 2 (2) SCR Smllh Croelc, U.S. Air For00 Acadmly, El Paso Co., CO. 6,860 ft. Monummt quadrangle, T12S,R67W 6 (6) STV &. Vraln R., Boulder County Open Spaco, Hygiene, BoutderCo., CO. ,,060 ft. Hygiene quadrangle, T 3 N. R 70 W 6 (5) WFS Wat.POik Sdiachheim Croelc, Lak, Dorothcy State Wildlife Area. Las Anlmu Co., CO [Elev.•?) JOuadnmltlo • ?t ITwmrn>,/RnVSeo • ?l Woodhomc Randt, Douglu Co., CO. 5,760 ft. Kass1« quandrang1e. T 7 S, R68 W 4 (4) ROX w WHR 7 (5) ROX970l• 24,E 1/2 ROX9702• 24, E 1/2 SCRI• SCR2• SCRJ• SCRS* SCR6* SCR7STVl STV2• STV3• STV4• STV5• STV6• 530• 531* 532* 533• WHRl• wmu•- 39° 25'44.s'' 105°03'48.9" 39° 26'10.l" 105°04'24.0" 4945290.00, 43643880.00 6NE,SE 114 NIA 36, SE 1/4 AI>eam Colorado Division of Wildlife Carron Meaney. Consuhant NIA C. Meaney Colorado Division ofWidlifc NIA NIA C. Meaney Colorado Division of Wildlife NIA 37"00'25" 104"22'32" NIA C.A.Jonea Dcnva- Museum of Natural History 36, NE 114, SWl/4 NIA NIA C.Meancyand N. Clippinger Colorado Division of Wildlife 36, SW 1/4, Elt'2 [Need info. from collector] [Infonnstion needed) ? ? NIA NIA 4309600.00, 5139400.00 4309580.00. 5139400.00 4309550.00, 5139400.00 4309SS0.OO, S139400.00 P.Sdtuerman S.Ask J. Hobert 04774210.00. 44084950.00 04774210.0Q. 440845>50.00 04774210.00. 44084950.00 JanPdcrsoo 4936880.00, 43651170.00 WHR3 WHR4* WHR5• WPC Well l'1um Croelc, Dougln Co., CO {Detailed dc,c:ription needed from collector] [Elav.•?) [Quadnmglo • ?). T 7 S, R 68 W WRN Monummt Creek near cx1ofluaicnrith Pinc Croelc, N. tide of Woodman ROid, Et Paso Co., co. WRP WhiteRmdll'm. Ralltm Croelc,oJfHwy.U norlh o!Ooldm. Jeft'encn Co., Colorado, 5740 ft. Ratstm Butt.es quadnmgtc, T 2 S. R._70 W Appendix A 8 (8) s (1) 3 (2) WHR6* WHR7 WPCl• WPC2• WPCJ* .WPC4• WPC5* WPC6* WPC7• WPC8• 97WRN130 97WRN138• 97WRN142 97WRN143 9'7WRN75 WRffl25 WRm41• WRm49• 31 Colorado Natural Heritage Program Page A-4 �Table 2. Reference specimens ofjumping mice provided by museums and other investigators, used as representatives of recognized species and subspecies of the genus Zapus for comparison with specimens listed in Table 1. Original 11mplo Specimen information provided by IOUr00 Individual or imtftuden ~Creek.FR 1'6 off'Col(? Hwy 14, LarfmerCo.CO. 10,000ft. Aalgnedto Zap,11 prtncq,, No. (4) CZPB• CZP9' DMNH Pmptoiro Campground, 262.S m. San babel National Porat, Lu Anlmn Co.. Colorado. Aalffll'!dto 7..tmtt, Drlnt:.DI at DMNH 2 (1) CZP12' CZP13' 7916' 7917 RSS Zaprt, httd,omr11 amplca provided by Bnaco 2 (2) Sample Code CZP ZHLNMO Wunder, Anoka County. Minnelcta. maine, NB 1/4 ofNB, Sec. l, T31N, R23W Presumed to bo Z. h. lntmrtdtit1 Zaprt, httd,omr11 mq,lca provided by Bruce Wunder, origin in Indiana; may be Z. h. tnrmn.dml or Z. h. ammcamu Zapr,1 hffdl0fffl11 can,patrll ftom Culta' Co., 81), provided by tho Mulcmn ofNatural m.iirv. Unfv. ofKanm Zaprt, httdltmltt1 can,patrt1 ftom Weston Co.. WY,providedbytho MulCUID ofNatural m..trv. Unfv. of'Kanm Zaprt, httd,omr11 (lttter11) ftom Otero Co., NM ZHLNMS Zaprt, httd,omr11 (lt1ter11) from Sandoval Co., ZAHU ZHCSDC ZHCWYW Fmton Lake, NM ZHPNBC Zaprt, httdionhll pc,llldt11 from Cieny Co., NB ZHPNEO Zap,11 ln1dionh11 pc,11ldt11 from Oanlm Co., NB ZPPCOG ZPPNMT Zaprt, prtncq,1 prtnc,p1 ftom Ptannigan C-. Onmd "-" CO Zaprt, prtncq,1 prtnc,p1 ftom Taos Co., NM nos. 1/4-1/4 Section NIA Lit-Long UTM Collecton'Sourco Coopenting Institution Coordinatea NIA NIA B. Wunda- Colorado State University NIA 37°15'00" N 105°6'30"W NIA C.AJones Daiver Museum of Natural History RSS 1076' RSS1077' NIA NIA NIA B. WundaRobatSikes ElmerBumey Colorado Stat.: University University of Minnesota 2 (2) ZAHU-253' ZAHU-300• NIA NIA NIA Colorado Stat.: University 2 (2) 109994' 109995• NIA NIA NIA B. Wunder collected by John WhiUalcec R. Timm T.Holmes 2 42469' 42470• NIA NIA NIA (2) R. Timm T.Holmes Natural History Museum, Univ. of Kansas, La\vrmcc 2 (2) NK87&• NK879• NIA NIA NIA T.Yates B.Gannon. 2 (2) NK3835' NK3837' NIA NIA NIA T.Y~ B.qannon 2. . -87043 87044 115762' 2 m 115763 CU14912 2 (0) CU14913 NK811 2 (1) NK814' NIA NIA NIA NIA NIA NIA NIA NIA NIA R. Timm T.Holmea R. Timm T.Holmes AdeQueiroz NIA NIA NIA Museum of Southwestern Biology, Univ. ofNew Mexico, Albuouemue. NM Museum of Southwestern Biology, Univ. ofNew Mexico. Albuouemue, NM Natural History Museum. Univ. ofKansas. Lawrmcc Natural History Museum, Univ. of Kansas, Lawrmcc Univ. of Colorado Museum. Boulder, CO Museum of Southwestern Biology, Univ. ofNew Mexico, AJbum,,.,.,,.,,. NM 4 (O} D. Annstrcn2 T. Yates B.Gannon . Natural History Musc:um. Univ. ofKansas, Lawrence NIA• not available for this report. In some cases the indicated information may be obtained from the person or institution supplying the material analyzed or can be inferred from other available information. ..... °' Am,cndixA ��63 I State of ___C ....o"""l,...o=ra...,d...,o.___ __ Project No. W-153-R-l l Work Plan No. _ _..,._0=66=2=---Task No. _ _ _ _--=-3_ _ __ APPENDIXB 11 STUDY PLAN Cost Center 3430 Mammals Research Conservation ofPreble's meadowjumpin2 mouse Temporal and spatial variation in the demography of Preble's meadow jumping mouse (Zapus hudsonius preblei) TEMPORAL AND SPATIAL VARIATION IN THE DEMOGRAPHY OF PREBLE'S MEADOW JUMPING MOUSE (Zapus hudsonius preblei) A. NEED On May 12, 1998 the U.S. Fish and Wildlife Service (USFWS) published a final rule in the Federal Register ( 63 FR 26517) to list Preble's meadow jumping mouse (Zapus hudsonius preblei) as 'threatened' under the Federal Endangered Species Act (ESA) of 1973, as amended. Recovery goals for Preble's meadow jumping mouse should work towards the sustainability, protection, and restoration of Z. h. preblei populations and habitats on both private and public lands to provide the spatial, genetic, and demographic structure needed to promote long-term species viability and provide species management flexibility. Recovery efforts for the subspecies will be most effective ifreliable information is available on the basic ecology of the subspecies and this information used to design recovery efforts such as Habitat Conservation Plans. A review of studies conducted on Preble's meadow jumping mouse shows that there is insufficient information to fully address defining rangewide ecological requirements, limiting factors, limits of species tolerance, or population status (Shenk 1998). Most work to date has focused on geographic distribution (presence or absence of Z. h. preb/ei), taxonomy, and habitat descriptions of sites where mice have and have not been captured. For Preble's meadow jumping mouse in particular, information on dispersal, habitat use, and population dynamics is most needed to identify minimal ecological requirements of the subspecies. Monitoring, by way of estimating demographic parameters (such as survival, abundance, and reproduction), of a single population over time provides an opportunity to document demography of a species, estimate temporal fluctuations in demography, and gain insights into the temporal variation inherent in demographic parameters. Monitoring of multiple populations over time and over multiple geographic locations provides the opportunity to gain further understanding of population processes and detaches time effects from spatial effects. Population monitoring activities do not constitute scientific experiments, in the spirit of manipulation of salient ecological variables, however, replication of monitoring activities for natural populations over long periods of time and in diverse geographic locations can lead to insights into population processes (Cook and Campbell 1979). These insights can then be translated into hypotheses useful for predicting changes in population demography resulting from either natural perturbations (e.g., flooding events) or anthropogenic modifications (e.g., gravel mining). Experimentation would then be required to test these hypotheses and establish cause and effect. There are three components of Z. h. preblei ecology that are currently unknown and yet key to any sound conservation strategy for the subspecies. These are (I) detailed demographic studies estimating survival, reproduction, immigration, emigration, and abundance and determining the factors influencing each of the parameters, (2) detailed studies evaluating movements and dispersal habitat of individuals within and among populations, and (3) detailed studies to define hibernation needs, primarily descriptions of suitable hibemacula criteria and food requirements for sufficient fat storage prior to immergence. The overall objective of this study is to provide estimates for all three of these �64 needs from three different populations of Preble's meadow jumping mouse. Thus, providing the opportunity to estimate spatial variation in the demography of the species. The three study populations occur in areas of different habitat matrices. The first study area limits animals within the population to a single drainage. The second population occurs in an area that combines a tributary and main drainage. The third population occurs in an area with both a tributary and main drainage as well as a series of ponds and irrigation ditches available to the animals. Continuing the study for multiple years will provide the opportunity to estimate temporal variation in the demography of the subspecies. B. OBJECTIVES Specific objectives for this study on Preble's meadow jumping mouse include: I. 2. 3. 4. 5. 6. 7. Evaluate temporal and spatial variation in daily and seasonal movements of Preble's meadow jumping mouse. Estimate and evaluate temporal and spatial variation of over-summer, over-winter, and annual survival rates for Preble's meadow jumping mouse. Estimate and evaluate temporal and spatial variation of reproduction, immigration, and emigration for Preble's meadow jumping mouse.. Estimate and evaluate spatial variation in abundance and density for Preble's meadow jumping mice. Evaluate temporal and spatial variation of habitat use in Preble's meadow jumping mouse. Document temporal and spatial differences in composition of foods consumed by Preble's meadow jumping mice. Collect genetic tissue samples for future analyses to document and compare variation within and among populations of Preble's meadow jumping mouse. C. EXPECTED RESULTS There are three components of Z. h. preblei ecology that are currently unknown and yet key to any sound conservation strategy for the subspecies. These are (I) detailed demographic studies estimating survival, reproduction, immigration, emigration, and abundance and determining the factors influencing each of the parameters, (2) detailed studies evaluating movements and dispersal habitat of individuals within and among populations, and (3) detailed studies to define hibernation needs, primarily descriptions of suitable hibemacula criteria and food requirements for sufficient fat storage prior to immergence. The overall objective of this study is to provide estimates for all three of these needs from three different populations of Preble's meadow jumping mouse providing the opportunity to estimation spatial variation in the demography of the species. Continuing the study for multiple years will provide the opportunity to estimate temporal variation in the demography of the subspecies Performance Indicators I. Provide descriptions of movements of mice for three Preble's meadow jumping mouse populations. 2. Provide estimates of survival and reproduction for three Preble's meadow jumping mouse populations. 3. Provide estimates of immigration, emigration, and abundance for three Preble's meadow jumping mouse populations. 4. Describe over-summer foods consumed by three Preble's meadow jumping mouse populations. 5. Quantify habitat characteristics of three sites where Preble's meadow jumping mouse are known to occur. 6. Analyze and publish results of research �65 D. APPROACH Study Site Selection All three study sites selected are areas where Preble's meadow jumping mouse (PMJM) has been found in the last two years. To evaluate spatial differences in movement of PMJM, sites were selected that provided a variety of habitat matrices available to the mouse. The first site selected, CDOW Maytag property, has one primary water source available to the mouse. This water source is East Plum Creek. Therefore we predict mice will restrict their movements to up and down this single drainage. The second site, Colorado Open Lands Pine Cliff Ranch, provides both a tributary (Garber Creek) and a main stem drainage (West Plum Creek). This provides the opportunity to investigate whether PMJM will move overland to move from one drainage to another or if they are restricted to moving only along riparian corridors. The third study site, CDOW Woodhouse Property and Dupont Property, provides an area containing a tributary (Indian Creek),a main stem drainage (West Plum Creek), and a series of ponds and irrigation ditches scattered throughout the property. This provides an even greater opportunity to investigate how much the mice will use upland areas or if they restrict their movements strictly to riparian corridors. Each of these unique habitat matrices provides the opportunity to estimate spatial variation in nightly and seasonal movements of PMJM. Movement Study Background Movement and dispersal pattern information will be key to any conservation strategy designed for Preble's meadow jumping mouse. Key factors include (1) which segment of the population disperses, (2) when do they disperse, (3) through what habitat do they disperse, (4) how far will individuals disperse (i.e., what is the maximum distance that separates adjacent populations) and (5) how critical is dispersal (both into and out of a population) to the persistence of a given population. The only currently available data on dispersal and/or movement for Z. h. preblei are from marked mice at Rocky Flats Environmental Technology Site (T. Ryon unpublished data). Two mice, an adult female and an adult male, were observed approximately 1.6 kilometers from previous locations (incidences occurred separately). Each of the locations were in the same drainage (Woman Creek). Objectives The PMJM movement study is designed to describe nightly and seasonal movement patterns of PMJM and to describe habitats used by PMJM. Specific objectives include: 1. 2. 3. 4. 5. Describe nightly movements of PMJM. Evaluate difference in nightly movements as they relate to sex, age, and habitat available to the animals. Describe 30-day (or life of the radio transmitter) interval movements of PMJM. Evaluate difference in 30-day (or life of the radio transmitter) interval movements as they relate to sex, age, and habitat available to the animals. Describe seasonal movements of PMJM. Evaluate difference in seasonal movements as they relate to sex, age, and habitat available to the animals. Describe habitats where mice occur: movement corridors, end point descriptions (disperses from what to what), and landscape features (connectivity with other riparian strips). Estimate the mean amount of time PMJM spend in each available habitat. Approach Movement data will be documented primarily from locations of radio-tagged mice from each of the three study populations. To place radio transmitters on the animals, mice will be captured in Sherman live traps. Mice weighing> 18 grams will be fitted with MD-2C, I-gram radio transmitters supplied by Holohill Systems Ltd. (used successfully on PMJM by R Schorr, personal �66 communication). A maximum of 30 mice at each study site will be radio-tagged. The following procedures will be used to capture and attach the transmitters to the mice. Trapping session details Timing of surveys 1. Four 7-day trapping sessions will be conducted during the following weeks: June 2-9, 1998; July 21-28, 1998; September 8-15, 1998; and June 2-9, 1999. 2. Due to the nocturnal nature of PMJM, traps will be set between 19:00hrs and 21 :OOhrs and checked as early as possible in the morning beginning at 5:00hrs (to reduce stress and the potential for predation on trapped animals). Time required to complete the traplines will vary depending on how many animals are caught. Trapping protocols 1. Trappers will be advised to follow the Center for Disease Control's Hantavirus instructions and recommendations when dealing with rodents which include: a Baseline blood serum samples will be collected and stored at Poudre Valley Hospital for all surveyors. b. Respirators are available for use by surveyors if they request their use. c. Surgical gloves will be used at all times when handling rodents. d. All equipment will be rinsed first with a I 0% bleach solution, then water after use. e. Traps and any equipment that may have rodent feces or urine on it will not be transported in the interior of a vehicle. f Traps will be cleaned after each use by a mouse. g. If a surveyor becomes ill with hantavirus symptoms they will report immediately to a medical facility. 2. Trappers must be in possession of both a federal and state collecting permit. Trapping methodology 1. Small mammal Sherman live traps {folding and non-folding) will be used to conduct the trapping sessions. 2. Traps will be set in two parallel lines of trap stations (1 trap per station) on either side of the drainage. Trap stations will be 5 meters apart for a total of 250 meters; the parallel transects will be IO meters apart unless extent of habitat, terrain topography, or stream hydrology do not allow. 3. Location of transects will be recorded on field data sheets and identified on 7½ minute topographic maps. Locations will also be recorded to the nearest 5 meters in UTM coordinates using a Trimble Geo-Explorer GPS. 4. In case of windy conditions or large number of trap-tampering predators (i.e., raccoons, foxes, coyotes, etc.), traps will be secured to the ground with hoops of heavy-gauge malleable wire, stakes, or with other materials that can effectively secure/immobilize the traps. 5. A small (~I inch) ball of polyester quilt or wool (fleece) will be placed in each trap as nesting/bedding material. 6. Baiting materials will be Manna Pro Sweet 3-way Livestock feed which contains no animal matter. Ingredients include flaked barley, flaked corn, flaked oats, and cane molasses. Peanut butter will be used to stick the bait to the trap. 7. Checking of the traps will be conducted by two surveyors; one person will handle the traps and animals captured, the other person will record the data Handling of captured animals 1. All animal captures will be recorded. �67 2. 3. 4. 5. If an animal has been captured in a trap, a ziplock plastic bag will be placed over the end of the trap. The trap will be opened allowing the animal to fall into the plastic bag. The animal will be identified to species while in the plastic bag. If the animal is not a PMJM, identification of the animal will be recorded and the animal set free. Each captured PMJM will be scanned to detect the presence/absence of a PIT tag and/or radio-collar. If a PIT tag or radio-collar is detected and the mouse has been captured at that same site within the 7-day trapping effort, identification of the mouse will be recorded and the mouse will be released. If the animal is a PMJM and no PIT tag or radio collar is detected the PMJM will be anesthetized for further processing, as follows. All PMJM will be PIT-tagged. If the PMJM weighs > 18 g a radio-collar will also be put on the animal Anesthesia The following protocol has been used successfully on PMJM by R Schorr (personal communication). I. Measure I ml ofMetofane (methoxyflurane) onto one cotton ball. 2. Place ball into ziplock bag and seal (to keep Metofane fumes in bag). 3. Lay bag still and move away. This should minimize stress for the mouse and the time the mouse struggles in the bag, decreasing the amount of time required for the metofane to work. 4. After the animal stops movement, wait I minute before removing it from the baggie. The animal should remain anesthetized for 2-3 minutes. 5. Time animal is exposed to Metofane and reaction to the anesthesia will be recorded. PIT tagging the animal The following protocol has been used successfully on PMJM (C. Meaney, personal communication). I. Each PMJM will have a PIT tag inserted above the shoulder blades by lifting the skin on its back and inserting the needle with. the PIT tag under their skin and injecting the tag. Verification of the PIT tag identification number will be made before insertion into the mouse by running the PIT tag scanner across it. 2. Skin behind the opening will then be pinched to prevent emergence of the tag. 3. Surgical glue will then be applied to the opening to prevent the PIT tag from exiting the body and prevent infection of the mouse. 4. A PIT tag scanner will be held over the mouse to again verify PIT tag identification number and that it still works after the insertion procedure (note: PIT tag will be scanned twice during application to mouse, or once if mouse was previously PIT tagged). 5. PIT tag identification number will be recorded on the field data form and the mouse released. 6. If, at any time during the handling of the PMJM the animal appears to be severely stressed (dramatic changes in heartbeat, respiration and responsiveness or gums turning blue) the animal will be released immediately. Collaring the animal The following protocol has been used successfully on PMJM (R Schorr, personal communication). I. Make sure transmitter is functioning before collaring the animal with the transmitter by tuning the receiver to the transmitter frequency to make sure there is a signal. 2. Slide crimp and then tubing onto the antenna �68 3. 4. 5. 6. 7. 8. Guide antenna through the channel of the transmitter. As antenna exits the transmitter, guide it through the crimp. Keep loop made by antenna large enough to fit over the mouse's head. Once over the neck of the animal, pull antenna to reduce collar size, making sure rubber tubing is on the dorsal side of the neck. The purpose of the rubber tubing is to prevent damage to the collar and prevent the person attaching the collar from cinching the collar too tight. Make sure collar can rotate freely around the animals neck, but not loose enough to be removed. Pinch metal crimp to hold the antenna in the desired collar position. Approximately 2 inches of antenna will be pointing out behind the animal. Measurements 1. Each PMJM will be weighed while in the bag, recording the weight in grams using a Pesola spring balance. 2. Sex, age Guvenile, adult), and reproductive condition (pregnant, lactating or nonbreeding if it is female; for males the position of the testes will indicate if in breeding status or non-breeding status) will be noted. 3. Each PMJM will be measured (total length, length of body, length of hind foot - heel to distal end of claws) in millimeters. Measurements of total body and tail length will be taken with the mouse on a firm surface to minimize error. 4. Capture of every PMJM will be documented by taking a photograph of the mouse on first capture. If the mouse has been captured previously as noted by the detection of a PIT tag, no photograph needs to be taken. 5. If there are feces in the trap where a PMJM was captured, the feces will be collected in a plastic bag labeled with date and location of the site. Fecal samples will be kept cool until returned to the CDOW office where they will be frozen. Fecal samples will be analyzed by the Composition Analysis Laboratory, Inc, 622 ½ Whedbee, Fort Collins, CO for content. Handling of trap mortalities and injuries I. All trap mortalities will be recorded on a 'Trap Mortality' data form which includes information on species, potential duration of time spent in the trap, any information available on cause of death, etc. 2. All animals found dead in a trap will be double-bagged in a plastic bags and placed in a cooler with ice. Specimens should be frozen as soon as possible and deposited the CDOW freezer at the office in Fort Collins. All specimens will be sent to the Colorado State University Diagnostic Laboratory for detection of any disease and/or other conditions that may have contributed to cause of death. All PMJM specimens will be returned to the CDOW. 3. A museum card will be completed and attached to each PMJM specimen and given to Cheri Jones, Curator ofMammalogy at the Denver Museum of Natural History, for study skins and tissue storage once the Diagnostic Laboratory has completed their evaluation. 4. If an animal is severely injured (e.g., severed limb, large lacerations) it will be euthanized by soaking a cotton ball in Metofane (methoxyflurane) and placing the cotton ball and the mouse in a ziplock baggie until the animal stops breathing. All specimens will then go through steps 1-3 above. 5. If an animal appears to be only slightly injured (e.g., broken tail, small laceration) the animal will be released to the wild. If an animal appears to be cold stressed attempts will be made to warm it by holding it in the surveyors hands and/or against their body. If the animal appears to be heat stressed, isopropyl alcohol will be applied with a cotton swab to the ears, arm pits, and feet to cool it down. �69 Training of field crews 1. All field technicians will be required to spend a full day at the Denver Museum of Natural History Zoology Department viewing and handling specimens of all small mammal species likely to be captured in Douglas County. 2. All field technicians will be provided with a small mammal field guide and a key specifically designed to list the species most likely to be captured on the trapping sites. 3. All field technicians will participate in 4 days of practice trapping at the Frank State Wildlife Area from May 26-29. During these trapping sessions all technicians will learn how to place, bait and set traps. Field identification will be practiced on all animals caught each day and verified by crew leaders. All field technicians will also practice handling, measuring, implanting PIT tags, and taking genetic tissue samples from deer mice (Peromyscus manicu/atus) until they are competent in this procedure. Data collection Once transmitters are in place, locations of individual mice will be made nightly. Because very little is known about the movements of PMJM the pilot protocols will be established. As we learn more about PMJM movements, protocols will be adjusted to maximize efficiency and data collection. To describe nightly movements the following parameters will be estimated: 1. Minimum and maximum distances moved each night animal is followed. 2. Mean minimum and maximum distances moved each night for all mice. 3. Minimum and maximum distances moved away from the stream center each night by individual mice. 4. Mean minimum and maximum distances moved away from the stream center each night for all mice. To describe 30-day (or life of the radio transmitter) movements the following parameters will be estimated: Minimum and maximum distances moved each 30-day (or life of the radio transmitter) 1. interval animal is followed. 2. Mean minimum and maximum distances moved each 30-day (or life of the radio transmitter) for all mice. Minimum and maximum distances moved away from the stream center each 30-day (or 3. life of the radio transmitter) interval by individual mice. Mean minimum and maximum distances moved away from the stream center each 304. day (or life of the radio transmitter) interval for all mice. Locations: 1. A minimum of 6 locations per individual mouse per night will be made. As many mice as possible will be tracked on a given night as is feasible to assure the minimum 6 locations per individual. Each individual mouse should be tracked a minimum of 3 nights per week. 2. The 6 locations per night per individual will be scattered throughout the night to maximize information learned about nightly movements (i.e., 8 locations taken within a single hour contains less information than 8 locations taken one each hour). 3. Three bearings will be taken for each location from pre-determined way stations to estimate location of the mouse. 4. Daytime locations will be taken as time permits. �70 Demography Study Background Information on the population dynamics of Preble's meadow jumping mouse is necessary to determine which areas support viable populations. To begin to evaluate the viability of a population information on key demographic parameters must be obtained. Below is a summary of what is know to date on the demography of Preble's meadow jumping mouse. Abundance: There is no information on range-wide abundance of Preble's meadow jumping mouse. However, from trapping surveys it appears that where the subspecies does occur it exists in low densities and thus one of the rarer components of the small mammal assemblage present. Reproduction: Meadow jumping mice have been observed to produce up to three litters per season (Whitaker 1963). Breeding peaks appear to occur in early to mid-June and August with a possible third litter in September (Whitaker 1963). Juvenile Z. h. preb/ei have been observed in June, August, and September (Meaney et al. 1996, 1997, PTI 1996a, M. Bakeman unpublished data, T. Ryon unpublished data), suggesting two litters per year. Z. hudsonius typically have litters of 5-6 young per litter (Quimby 1951 ). Age of first reproduction is unknown for Z. h. preblei, however, females of Z. hudsonius have been observed to give birth at 3 months (i.e., females born in June have been observed to give birth in August of the same year). Gestation period is approximately 18 days (Quimby 1951 ). Young remain dependent on the female for approximately 18 days (Quimby 1951). No evidence of male parental care exists for Z. hudsonius (Whitaker 1963). Survival: No information exists on survival rates for populations of Z. h. preblei. Whitaker (1963) reported a 67% loss of individuals over hibernation and that average body mass of individuals emerging from hibernation was greater than the average for mice entering hibernation. Because no mice are known to store food in their hibernacula, this indicates that the lighter individuals died during hibernation and only those entering with higher masses survived. All the energy they use during hibernation and the periodic arousals (the energetically most expensive part of hibernation) must be the fat they carry into hibernation (B. Wunder, personal communication). Thus, the ability to put on sufficient fat for overwinter survival during hibernation is a critical factor in the life history of these mice. Besides insufficient fat storage prior to hibernation, other observed mortality factors in Z. hudsonius include predation (Whitaker 1963, Poly and Boucher 1997) and cannibalism (Sheldon 1934). Other assumed mortality factors for Z. h. preblei include starvation, exposure, and disease. Population Structure: Armstrong et al. (1997) reported an overall sex ratio for all captured Preble's meadow jumping mice of 51. 6 males: 48. 4 females; approximately 86. 0% of captures were identified as adults. However, Armstrong et al. (1997) suggested that these data be interpreted with caution because of possible differences in field techniques. Longevity: Very few individuals of Z. h. preb/ei have been permanently marked. Therefore, recapture information, necessary to determine longevity, is minimal. Several recaptures have yielded adults surviving through two years, indicating a longevity of at least three years (T. Ryon, unpublished data). One recapture history from Rocky Flats Environmental Technology Site recorded an adult male in 1996, two years after it was first captured as an adult in 1994, indicating survival of at least four years {PTI 1996b). Quimby (1951) found that only a low percentage of Z. hudsonius lived two years or more, but gave two records of mice that lived for at least two years under natural conditions. Whitaker (1963) reported a female living at least two years. Objectives Objectives of the demography study are to: 1. Estimate abundance of PMJM at each of three study populations. 2. Estimate over-summer, over-winter, and annual survival of PMJM at each of three study populations. 3. Estimate temporary emigration of PMJM at each of three study populations. 4. Estimate immigration of marked PMJM back into each of three study populations. 5. Evaluate the affect of weight, sex, age, abundance (i.e., density dependent response), �71 6. and habitat features such as stream reach, vegetation composition and density on survival, reproduction, abundance, temporary emigration, and immigration of marked animals back into three study populations of PMJM. Estimate age and sex ratios of PMJM at each of three study populations. Approach Mark-recapture estimation techniques will be used to estimate abundance, over-summer survival, over-winter survival, temporary emigration and immigration of marked animals back in to the three study populations of PMJM. Abundance: Abundance will be estimated using Pollock's Robust Design (see Kendall et al. 1997, 1995 and Kendall and Nichols 1995 for detail). The robust design is a combination of the CormackJolly-Seber (CJS)(Cormack 1964, Jolly 1965, Seber 1965) live recapture model and the closed capture models. The key difference from the CJS model is that instead of just one capture occasion between survival intervals multiple capture occasions are used. These occasions are close together in time allowing the assumption that no mortality or emigration occurs during theses short time intervals. The closely spaced encounter occasions are termed "trapping sessions" and each trapping session can be viewed as a closed capture survey. Four 7-day trapping sessions will be conducted during the following weeks: June 2-9, 1998; July 21-28, 1998; September 8-15, 1998; and June 2-9, 1999. Survival: By using the estimate of the probability that an animal is captured at least once from the trapping sessions designed to estimate abundance, survival between the longer intervals can be estimated. Four 7-day trapping sessions will be conducted during the following weeks: June 2-9, 1998; July 21-28, 1998; September 8-15, 1998; and June 2-9, 1999. Thus, estimates of over-summer survival, over-hibernation survival, annual survival and survival from June-July, July-August, and August-September can be made. Temporary emigration and immigration: The longer intervals between trapping sessions also allows estimation of temporary emigration from the trapping area, and immigration of marked animals back to the trapping area Reproduction: Reproductive parameters will be estimated by following radio-tagged females to nest sites. At each nest found the following will be recorded: date, number of young in each nest and number oflitters observed for individual females throughout the year Habitat Use Study Background The habitat matrix within the range of Z. h. preblei is mixed grasslands adjacent to the Colorado Front Range along the Piedmont and along the base of the Laramie Mountains in Wyoming and extends to the Colorado plains. Within this matrix, Preble's meadow jumping mice occur along stream drainages that contain patches of suitable vegetation. Suitable habitat appears to have at least two major components. The first component is a supply of open water, at least in part of the active season (M. Bakeman, C. Meaney, personal communication). Secondly, areas where Preble's meadow jumping mouse has been found have dense cover (M. Bakeman, C. Meaney, personal communication). If Preble's meadow jumping mouse behaves as a metapopulation in the classical sense of a set of local populations linked by infrequent dispersal then habitat includes not just one area of suitable habitation but also areas suitable for nearby mouse populations. These suitable areas must also be linked by dispersal habitat. If the mice are dependent on dense riparian habitat for dispersal as well as for areas to reproduce, persistence of discrete populations would require a mosaic of suitable discrete riparian patches interconnected with dispersal corridors of similarly dense riparian vegetation. If mouse populations function in a source (populations where growth rate~ 1) and sink (those populations where �72 growth rate < 1, maintained through immigration) system, it will be critical to identify and protect those populations serving as sources. Thus, for a source-sink population critical habitat will include those areas that support source population, dispersal habitat to sink areas will be less critical. If local mouse populations are functionally discrete, such a mosaic of interconnected areas of suitable habitat would provide a buffer for local, and source, populations against deleterious stochastic events by providing the opportunity for local population failures to be 'rescued' by immigration from other populations. Areas of suitable habitat must also provide requirements to survive throughout the life cycle. These requirements must provide necessities for both the active period and hibernation periods. During the active period suitable habitat must provide requirements for daily survival, reproductive activities (breeding, nesting, and rearing of young to independence), and dispersal. The hibernation period requires sufficient food supplies to assure fat storage prior to hibernation and suitable hibernacula Habitat providing all seasonal and life cycle requirements may or may not occur in a single contiguous area If not in a contiguous area, habitat patches must occur in a mosaic of usable areas where suitable corridors exist for seasonal movement among sites. Based on studies of Z. h. preblei and Z hudsonius elsewhere, Z. h. preblei apparently occurs mostly in undergrowth consisting of grasses, forbs, or both in open wet meadows and riparian corridors, or where tall shrubs and low trees form an overstory and provide adequate cover (Armstrong et al. 1997). Meadow jumping mice are widespread in abandoned grassy fields, but is often more abundant in thick vegetation along ponds, streams, and marshes or in rank herbaceous vegetation of wooded areas (Whitaker 1963). Preble's meadow jumping mice have been trapped in natural riparian areas as well as areas altered by anthropogenic influence including ditches and wetlands adjacent to interstate highways, cement-lined ditches with tall cover, ditches along driveways and moderate road use, and moderate cattle grazing (M. Bakeman, personal communication). The majority of sites where Z. h. preblei have been found consist of multistoried cover but the species composing the cover vary greatly (Armstrong et al. 1997, Meaner et al .. 1997). Vegetation composition of the dense cover varies considerably and includes both native and non-native species (Meaner et al. 1996, 1997, Armstrong et al. 1997, M. Bakeman, personal communication). The herbaceous understory can be primarily grasses or forbs or a mixture of the two. Few of the sites, however, are dominated by fewer than two understory species. The tall shrub canopy at most sites is willow (several species), although scrub oak, birch, and alder occur in sites south of the Palmer Divide (Armstrong et al. 1997). Ponderosa pine is the most common tree at higher elevations. The mouse appears to tolerate weedy or exotic species in areas that are structurally diverse and species rich; nearly every successful site contained Canada thistle (Armstrong et al. 1997). Thus, the mouse does not appear to have an affinity toward any single plant species but instead favors sites that are structurally diverse and provide adequate cover and food throughout its life cycle. Preble's meadow jumping mouse mice are typically not found in upland areas away from riparian habitats but are most often captured where either ground water daylights to seep springs or on main water channels (M. Bakeman, T. Ryon, personal communication) suggesting a dependence on open water, at least during their active periods. Movement of Z. h. preblei on Woman Creek at Rocky Flats Technology Site suggests mice move along corridors of shrub cover, generally Salix exigua (PTI 1996a, T. Ryon, unpublished data), suggesting dispersal habitat is similar to habitats used for other activities. Jumping mice of the genus Zapus are true hibernators, spending much of their lives in hibernation. Meadow jumping mice spend approximately 7 months (~210 days) per year in hibernation (Quimby 1951) whereas estimates for Z. princeps indicate that some populations (e.g., in the western mountains of Utah) spend up to 300 days per year in hibernation (Cranford 1983). Males emerge prior to females (Bailey 1923, Bailey 1929, Hamilton 1935, Quimby 1951, Whitaker 1963) with the earliest annual recorded dates for Z. hudsonius males being April 25-May 16 and females May 4-26. Latest fall capture date for an adult male was October 10 in 1994 at South Boulder Creek (ERO Resources 1995) and September 28 in 1995 at the Van Fleet Parcel on South Boulder Creek (Armstrong et al. 1997). A juvenile male was captured as late as October 26 and a female juvenile on �73 October 27 both in 1995 at Rocky Flats Environmental Technology Site (M. Bakeman, unpublished data). Jumping mice hibernate in underground burrows (Quimby 1951, Whitaker 1963). They are excellent burrowers and create their own hibernacula Meadow jumping mice are generally solitary hibernators, however, there have been occurrences of more than one mouse found in a single hibernaculum. One hibernaculum, located on Rocky Flats Environmental Technology Site, used by Z. h. preblei has been located (Armstrong et al. 1997). This site was 9m above a creek bed (Walnut Creek); it had a thick cover of chokecherry (Prunus virginiana) and snowberry (Symphoricarpos spp.), the mouse was found in a leaf litter nest 30cm beneath the ground in coarse textured soil (Armstrong et al. 1997). Four possible hibernacula were located by tracking radio-telemetered mice at the U. S. Air Force Academy in fall 1997. These sites are located 7, 12, 29, and 31m from a creek bed (R Schorr, personal communication). There was no consistency among sites in aspect (N/NW, SISE, E, and none [level ground]). Three sites were in vegetation dominated by coyote willow (Salix exigua), one site was in vegetation dominated by snowberry and mullein (Verbascum thapsus). However, all four hibernacula appear to be below coyote willows. These four U. S. Air Force Academy sites have not been disturbed to protect any hibernating mice and therefore are only possible hibernacula because there is no confirmation a mouse is actually hibernating there. Confirmation of a true hibernaculum cannot be made until a chamber, or nest is located. These sites may also possibly be locations ofradios discarded by the mice or dead mice. Objective The objective of the habitat study is to identify and refine habitat requirements of Z. h. preblei, including hibernation sites, and to determine if they influenced any of the demographic parameters that will be estimated in this study. Approach The general approach is to measure both site specific and landscape features at each of the three population study sites to document the extent of spatial variation. These quantitative measures of spatial variation will be used in the analyses to determine if they influenced any of the demographic parameters that will be estimated in this study. The following habitat characteristics will be measured and recorded for each site. Micro-site habitat characteristics 1. Cover will be measured at 20 random locations along the 250 meter sampled stream stretch. Mean cover and standard error of the mean will be estimated. Cover will be estimated using a vegetation profile board (Nudds 1977) that allows for an assessment of visual obstruction in 0.5 meter vertical intervals above ground. The board will be 1.5 meters high and 30.48 cm wide. The board is marked in alternate black and white colors at 0.5 meter intervals. Horizontal cover is assessed in each interval by viewing the board from 15 meters away in a randomly chosen direction. The percentage of each interval concealed by vegetation is recorded as 0-20, 21-40, 41-60, 61-80, and 81-100% estimated concealment. 2. Vegetation species composition and richness will be estimated using the Modified-Whittaker nested vegetation sampling method (Stohlgren et al. 1995). A modified Whittaker plot is 20meters x 50-meters with 10 lm2 and 2 10m2 subplots arranged systematically around the perimeter and 1 100m2 subplot centered in the inside of the 20 meter by 50 meter plot. Species composition and percent cover of each species is recorded for each subplot. 3. Soil samples will be taken at each site for a hydrometer textural analysis (percent sand, silt and clay). Samples will be collected using a soil probe. Samples will be taken from 0-12 inches, and 12+ to 24 inches. The 0-12 inch sample will be taken first, removing the probe and soil. Sample will be placed in a labeled ziplock bag. The probe will then be placed in the same hole for the 12+ - 24 inch probe. With that soil sample being placed in a separately labeled ziplock bag. The hydrometer textural analysis will be conducted at the CSU Soils Laboratory. �74 4. 5. 6. 7. Mean stream width will be estimated by measuring stream width at 30 locations along the entire site sampled. The 30 locations will be stratified equidistant from each other to cover the entire stream stretch in increments of site stream length/20. Describe potential hibernacula: upland vegetation, take soil sample at site. Describe nest sites (what they are made of, placement). Compare parameter estimates from 1-6 for the three population study sites. Landscape habitat characteristics The following landscape habitat characteristics will be measured using either GIS, topographic maps, or aerial infrared photography. 1. Estimate distance to nearest human habitation and human disturbance. 2. Estimate connectivity of sites to other streams. 3. Estimate total length of stream stretch at each site and total length of stream stretch with suitable habitat at each site. 4. Compare parameter estimates from 1-3 for the three population study sites. Hibernation site measurements Hibernation sites will be identified by following radio-collared mice in late September and early October. Once a radio-signal remains stationary for 7-10 nights we will assume we have located a hibernaculum. The following measurements will be taken at each hibernation site: 1. 2. 3. 4. Distance ofhibernaculum to stream. Vertical elevation from hibernaculum to stream. Soil sample (as described above for the Habitat Use Study). Vegetation species richness within a 5-m radius. Density of vegetation (measured as described above for the Habitat Use Study). Composition of Seasonal Food Consumption Study Background Specific food habits are unknown for Preble's meadow jumping mouse. Armstrong et al. (I 997) summarized what is currently known about food habits of meadow jumping mice in general as follows: Studies of food habits in central and eastern United States indicate they are governed by availability more than preference (Whitaker 1963). Grass seeds of several species are probably the most important component of the diet, and mice will shift to those species that have available seed. Invertebrates and fungi are also readily eaten. Mice feed on both adult and larval invertebrates, especially Coleoptera (beetles). Invertebrate feeding is very important in the spring as mice emerge from hibernation, and may consist of half of the diet at that time. Mice also feed on various species of fungi, which are often encountered during burrowing activity. As the growing season progresses, graminoid seeds dominate the diet. There is no reason to believe food habitats of one subspecies should differ greatly from those of other subspecies. This belief is further supported by the indirect evidence available that food habits of Z. h. preblei are similar to that described above (i.e., the observation of the piles of grass stems observed in areas where the subspecies is known to occur). As true hibernators, meadow jumping mice do not cache food in their hibernacula Therefore, it is assumed they require high quality food for high fat accumulation prior to immergence. Preble's meadow jumping mice have been observed to increase body weight by 10% in the 2-3 weeks prior to immergence. Laboratory studies show the mouse can gain mass at rates up to 1.0 grams per day (B. Wunder, personal communication). This ability to increase fat reserves at such a rate is used by the �75 mice for preparation to enter hibernation. However, for the mice to successfully gain enough fat prior to hibernation to ensure a high probability of survival throughout hibernation, food of sufficient quality must be available. No specific information is available on what foods Preble's meadow jumping mice eat to meet these ecological requirements. However, from late August through early September, when adults begin to gain weight, there are many species of graminoids that have available seed. In most years, grasshoppers are also readily available and probably dominate invertebrate biomass in most habitats. Objective The objective of this study is to document seasonal changes in foods consumed by Preble's meadow jumping mice. Approach Fecal samples from traps where PMJM are captured during the four trapping sessions will be collected and analyzed for composition and percentage of each discrete food type identified in the total sample. Changes in foods composition by either food type and or percent occurrence will be analyzed for spatial and temporal differences. Spatial differences will be defined as differences among the three populations studied. Temporal differences will be defined as differences among the three trapping sessions (June, July, September). Molecular Systematic Study Background The family Zapodidae Gumping mice) consists of small to medium-sized mice with enlarged hind feet and exceptionally long tails. Four living genera are recognized in this family, two of which, Zapus and Napaeozapus, are found in North America (Hall 1981). There are three living species of the genus Zapus: Z. trinotatus (Pacific jumping mouse), Z. princeps (western jumping mouse), and Z. hudsonius (meadow jumping mouse). Z. h. preblei is one of twelve living subspecies of the species Z. hudsonius (Krutzsch 1954, Hafner et al. 1981). Z. h. preblei was first described by Krutzsch (1954) from a specimen collected by E. A. Preble in 1895 near Loveland, Colorado. Quimby (1951) described the species Z. hudsonius as follows: 'A mouse-like rodent with greatly enlarged hind feet and an exceptionally long tail. The forelegs are relatively short. The ears are somewhat conspicuous. The body is clothed in moderately long, somewhat dense hair of a rather coarse texture and several colors. The dorsal portions are marked by a broad stripe of brownish hairs many of which are tipped with black giving the region a grayish-black appearance. The sides are bright yellowish-orange, whereas the underparts and feet are white. The tail is bicolor, dark above and light below, and sparsely covered with hair which is longer on the terminal part. The mammae are eight, and quite prominent in lactating females. The male genitalia are inconspicuous except during the breeding season when the scrotal sac becomes enlarged. The testes enlarge and may be either abdominal, inguinal, or scrotal during this period.' The skull of Z. hudsonius is small and light with a narrower braincase and smaller molars than in Z. princeps (from Fitzgerald et al. 1994 p. 18). However, Fitzgerald et al. (1994) urge caution in distinguishing the two species of Zapus in Colorado. Such caution is further warranted by the recent conflicting identifications of mice based on genetic characteristics. Analysis of mitochondrial DNA sequence data from 92 individual mice indicates that mice sampled from southeastern Albany County, Wyoming, south along the Front Range of the Rocky Mountains to western Las Animas County, Colorado (Purgatoire Campground, San Isabel National Forest}, form a coherent genetic group (Riggs et al. 1997). This group of samples are distinct from samples obtained from mice from three other populations. Genetic samples from mice captured in the Dorothey Lakes area of southern Colorado (Las Animas County) group together and are most closely allied with Z. h. luteus, a subspecies described from New Mexico (Riggs et al. 1997). The single genetic sample collected from Weld County, Colorado (Lone Tree Creek) and six samples obtained �76 from Warren Air Force Base in Laramie County, Wyoming are most similar to reference samples of Z. princeps from Colorado (Larimer County) and New Mexico (Taos County) (Riggs et al. 1997). The sequence data indicate that the samples from specimens identified as Z. h. luteus and samples from specimens of Z. princeps are more closely allied with each other than either is with the samples defining the Z. h. preblei group. This closer alliance of the samples of Z. h. luteus with Z. princeps rather than Z. h. preblei conflicts with results ofHafuer et al. (1981), based on a combination of pelage, morphologic, and genetic data, which support a closer alliance with Z. hudsonius. A more complete biosystematic evaluation of jumping mice is needed to clarify and further refine relationships among populations of the group referred to as Z. h. preblei as well as to other subspecies and species of the genus 'Zapus. Such an evaluation requires detailed analyses of pelage, morphometric, and genetic data from sufficient numbers of individuals to adequately represent the populations of interest. However, the mitochondrial DNA non-coding (D-loop) sequence data available at this time are consistent with the view that a geographically contiguous set of populations previously recognized as Preble's meadow jumping mouse form a homogenous group recognizably distinct from other nearby populations and from another geographically-adjacent species of the genus (Riggs et al. 1997). Therefore, given the genetic data available, Riggs et al. (1997) conclude Preble's meadow jumping mouse is a distinct population of Z. hudsonius. Objective The objective of the genetic component of this study is to document genetic variation within and among the three study site populations. Efforts are underway to secure funding for this study. Approach Genetic tissue samples will be collected from at least 10 and not more than 30 PMJM captured at each study site. A sample of at least 10 individuals from a population will provide enough information to document the genetic variation within the population, samples sizes in excess of 30 contribute little to documenting the genetic variation within the population (T. Quinn, personal communication). Genetic tissue sampling: collection ofear tissue samples using ear punch tool The following protocol for collection of genetic tissue samples has been used successfully on PMJM (M. Bakeman, C. Meaney, T. Ryon, R Schorr, personal communication). Before checking traps, ear punch tools will be cleaned by immersing them in a 10% bleach solution for a few minutes, then rinsing thoroughly with clean water. Tools will be dried thoroughly. A small screw-cap container will be useful to contain the bleach solution and to receive used ear punch tools. A sports bottle or other container with a squirt or pop-up top may be used for rinsing. Any previously used containers will be washed thoroughly by hand or in the dishwasher and not used for other purposes (e.g., drinking) while in the field. 1. A fresh pair of clean latex gloves will be used when handling each mouse. 2. With a clean ear punch tool, 2 tissue plugs will be obtained from the mouse's ear. If there is excessive bleeding, gentle pressure will be applied to the site with a small , dry piece of cotton for 1 minute. Punch may be applied to deliver the plug as a small circlet of tissue (about the size of the head of a pin) onto a finger of your gloved hand. Placement of the punch may be done in any way that is convenient. 3. Place the finger tip tightly over the end of an opened sample vial and shake to wash the tissue plug into the ethanol. 4. Repeat until all three tissue plugs have been collected into the same sample vial. Inspect the tube to see that all plugs have made it into the ethanol. 5. Close the sample vial--screw the cap on firmly and check that there is no leakage of ethanol when the tube is inverted. Label both the vial and the corresponding record on the data sheet (see item 6 below) with a. unique identifier as follows. Use a seven-place alpha-numeric code composed of the following: �77 - A three-letter designator for the survey location (e.g., WPC = West Plum Creek, GCR = Garber Creek). A number beginning with two digits indicating the year (e.g., 98) followed by 2 digits specifying the individual trapped, numbered sequentially. Example: WPC9801 = first animal sampled at West Plum Creek in 1998. 6. Note: The combined seven-place alpha-numeric code needs to be a unique identifier for an individual mouse across all locations and sites being trapped in studies coordinated by the Colorado Division of Wildlife. Record information for the individual sampled on the data sheet provided. Repetitive entries may be completed before or after the field session but do not allow much time to elapse before doing this--what is obvious to you at the time may not be so obvious later on. Information for each sample needs to be complete and unambiguous if the sample itself is to be useful. After returning from the field, samples will be kept in a cool place or refrigerated until delivery to the CDOW office at 317 West Prospect, Fort Collins, CO where they will be kept in the freezer for analyses. E. LOCATION CDOW Maytag Property is located seven miles south of Castle Rock, Colorado along East Plum Creek. CDOW Woodhouse Property is located on Indian Creek, a tributary of Plum Creek near Louviers, Colorado. The Dupont Property is located west and south of the town of Louviers and contains the eastern stretch of Indian Creek and its confluence with Plum Creek. Pine Cliff Ranch, owned by Colorado Open Lands, is located on Garber Creek and extends eastward to its confluence with West Plum Creek, 5 miles west of Sedalia Training sessions for all field technicians and crew leaders will take place at the Frank State Wildlife Area and Kodak State Wildlife Area located in Weld County, Colorado. Data analyses and office work will be conducted at the CDOW Research Center, 317 West Prospect, Fort Collins, Colorado. F. SCHEDULE April-May April April-May May June 2-9 June-July July July 21-28 July-Aug Sep 8 -15 Sept-Oct, Sept-Oct. December April 1999 Complete Study Plan for a demography study of Preble's meadow jumping mouse Select study sites Purchase equipment for summer 1998 field season Advertize for, hire, and train 4 technicians Conduct first trapping session Collect movement data Collect site and landscape scale habitat data Conduct second trapping session Collect movement data Conduct third trapping survey Collect movement data Collect data on possible hibernation sites Complete preliminary report for Preble's meadow jumping mouse technical working group Complete Final report for 1998 summer field season �78 F. PERSONNEL Tanya Shenk Gary C. White 2 Crew Leaders 2 Field Technicians G. Principal Investigator Statistical Consultant CDOW temporary technicians CNHP temporary technicians BUDGET Operating Expenses radio transmitters (150 @$150) radio receiver (3 @ $2565 ) pit tag scanner (6@$600) pit tags (200 @$10.00) 3 portable communication radios & battery recharger maps computer software, upgrades office supplies photographic equipment, expenses field equipment misc. hardware Personnel 2 field technician for 4 months each 2 crew leader for 6 months $22,500 . $ 7,695 $ 3,600 $ 2,000 $ 2,100 $ 500 $ 1,500 $ 300 $ 500 $ 800 $ 500 $12,000 $30,000 Travel 30,000 miles @0.12 per mile vehicle rental: 3 vehicles, 4 mo. @$120 per month overnight travel 30 days@$75 per day TOTAL $ 3,600 $ 1,440 $ 2,250 $75,365 �79 I. LITERATURE CITED Armstrong, D. M., M. E. Bakeman, A. Deans, C. A. Meaney, and T. R Ryon. 1997. Conclusions and recommendations in: Report on habitat findings on the Preble's meadow jumping mouse. Edited by M. E. Bakeman. Report to USFWS and Colorado Division of Wildlife. Bailey, B. 1929. Mammals of Sherburne County, Minnesota. Journal ofMammalogy 10:153-164. Bailey, V. 1923. Mammals of the District of Columbia Proceedings of the Biological Society. Washington 36:103-138. Cook, T. D. and D. C. Campbell. 1979. Quasi-experimentation: design and analysis issues for field settings. Houghton-Mifflin, Boston. Cormack, RM. 1964. Estimates of survival from the sightings of marked animals. Biometrika 51 :429-438. ERO Resources. 1995. Environmental review of South Boulder Creek Management Area Prepared for City of Boulder Real Estate/Open Space Department. Prepared by ERO Resources Crp., Denver, Colorado in association with Stoecker Ecological Consultants, Boulder, Colorado. Fitzgerald, J.P., C. A. Meaney, and D. M. Armstrong. 1994. Mammals of Colorado. Denver Museum of Natural History, University Press of Colorado. Niwot, Colorado. Hafner, D. J., K. E. Petersen, and T. L. Yates. 1981. Evolutionary relationships of jumping mice (genus Zapus) of the southwestern United States. Journal ofMammalogy 62:501-512. Hall, E. R 1981. The mammals of North America John Wiley and Sons, Inc., New York, New York, 2 volumes. Hamilton, W. J., Jr. 1935. Habits of jumping mice. American Midland Naturalist 16:187-200. Jolly, G. M. 1965. Explicit estimates from capture-recapture data with both death and immigration stochastic model. Biometrika 52:225-247. Jones, C. A. 1996. Mammals of the James John and Lake Dorothey State Wildlife Areas. Final Report, submitted to the Colorado Division of Wildlife and Colorado Natural Areas Program. Kendall, W. L., and J. D. Nichols. 1995. On the use of secondary capture-recapture samples to estimate temporary emigration and breeding proportions. Journal of Applied Statistics.22:751762. Kendall, W. L., K. H. Pollock, and C. Brownie. 1995. A likelihood-based approach to capturerecapture estimation of demographic parameters under the robust design. Biometrics 51 :293308. Kendall, W. L., J. D. Nichols, and J.E. Hines. 1997. Estimating temporary emigration using capturerecapture data with Pollock's robust design. Ecology 78:563-578. Krutzsch, P.H. 1954. North American jumping mice (genus Zapus). University of Kansas Publications, Museum of Natural History 7:349-472. Meaney, C. A., N. W. Clippinger, A. Deans, and M. OShea-Stone. 1996. Second year survey for Preble's meadow jumping mouse (Zapus hudsonius preblei) in Colorado. Report prepared for the Colorado Division of Wildlife. Meaney, C. A., A. Deans, N. W. Clippinger, M. Rider, N. Daly, and M. O'Shea-Stone. 1997. Third year survey for Preble's meadow jumping mouse (Zapus hudsonius preblei) in Colorado. Report prepared for the Colorado Division of Wildlife. Nudds, T. D. 1977. Quantifying the vegetative structure of wildlife cover. Wildlife Society Bulletin 5:113-117. Poly, W. J., and C. E. Boucher. 1997. Record of a creek chub preying on a jumping mouse in Bruffey Creek, West Virginia Brimleyana 24: 29-32. PTI Environmental Services. 1996a Preble's Meadow Jumping Mouse Study at Rocky Flats Environmental Technology Site, Annual Report 1996. Final. Rocky Flats Environmental Technology Site, Golden, Colorado. �80 PTI Environmental Services. 1996b. Preble's Meadow Jumping Mouse Study at Rocky Flats Environmental Technology Site, Spring 1996. Final. Rocky Flats Environmental Technology Site, Golden, Colorado. Quimby, D. C. 1951. The life history and ecology of the jumping mouse, 7.apus hudsonius. Ecological Monographs 21:61-95. Riggs, L.A., J.M. Dempey, and C. Orrego. 1997. Evaluating distinctness and evolutionary significance of Preble's meadow jumping mouse: Phylogeography of mitochondrial DNA noncoding region variation. Final Report for the Colorado Division of Wildlife. Denver, Colorado. Seber, G. A. F. 1965. A note on the multiple recapture census. Biometrika 52:249-259. Sheldon, C. 1934. Studies on the life histories of 7.apus and Napaeozapus in Nova Scotia Journal of Mammalogy 15:290-300. Shenk, T. M. 1998. Conservation assessment and preliminary conservation strategy for Preble's meadow jumping mouse (7.apus hudsonius preblei). Special Report. Colorado Division of Wildlife (in review). Stohlgren,T. J., M. B. Falkner, and L. D. Schell. 1995. A Modified-Whittaker nested vegetation sampling method. Vegetatio 117: 113-121. USFWS. 1997. Interim survey guidelines for Preble's meadow jumping mouse. USFWS. Denver, Colorado. Whitaker, J. 0., Jr. 1963. A study of the meadow jumping mouse, 7.apus hudsonius (Zimmerman), in cental New York. Ecological Monographs 33:3. �81 APPENDIXC STUDY PLAN State of Colorado ProjectNo. W-153-R-ll Work Plan No._....,0=6=6=2_ _ __ Task No. ______2_ _ _ __ Cost Center 3430 Mammals Research Conservation of Preble's meadow jumping mouse (.Zaous hudsonius oreblei) Habitat use and distribution of Preble's meadow jumping mouse (Zapus hudsonius preblei) in Larimer and Weld Counties, Colorado HABITAT USE AND DISTRIBUTION OF PREBLE'S MEADOW JUMPING MOUSE (Zapus hudsonius preblei) IN LARIMER AND WELD COUNTIES, COLORADO A. NEED On May 12, 1998 the U. S. Fish and Wildlife Service (USFWS) published a final rule in the Federal Register (63 FR 26517) to list Preble's meadow jumping mouse (Zapus hudsonius preblei) as 'threatened' under the Federal Endangered Species Act (ESA) of 1973, as amended. Scarcity of suitable habitat presumably limits current distribution of Preble's meadow jumping mouse and thus, maintenance of quality habitat has been identified by the USFWS (63 FR 26517) as the principal conservation goal. Although Meaney et al. (1997) reported an improved ability to recognize suitable habitat for Preble's meadow jumping mouse, a more refined and complete definition of potentially suitable habitat for the mouse does not exist. Because the protection of potentially suitable habitat for Preble's meadow jumping mouse may occur under the ESA, as well as protection of known locations where the subspecies occurs, the definition of potentially suitable habitat, as determined by the USFWS, will directly influence site specific regulatory procedures. Ideally, a definition of potentially suitable habitat for the subspecies would identify areas where the Preble's meadow jumping mouse could survive and reproduce in sufficient numbers to sustain populations throughout its range of natural variability over an extended length of time. During the active period of the life cycle of the mouse, suitable habitat must provide requirements for daily survival, reproductive activities (breeding, nesting, and rearing of young to independence}, and dispersal. The hibernation period requires habitat with sufficient food supplies to assure fat storage prior to hibernation and hibernacula sites. Habitat providing all seasonal and life cycle requirements may or may not occur in a single contiguous area If not in a contiguous area, habitat patches must occur in a mosaic of usable areas where suitable corridors exist for seasonal movement among sites. Because very little is known about the ecological requirements of the subspecies, potentially suitable habitat is currently defined only as areas of well-developed, dense herbaceous vegetation consisting of a variety of grasses, forbs and thick shrubs in close proximity to open water. This definition is vague and possibly incomplete. Very few surveys for determining the presence or absence of Preble's meadow jumping mouse have been conducted in Larimer or Weld Counties, Colorado. Of the surveys nine conducted since 1990, only three sites yielded captures of Preble's meadow jumping mouse. Therefore, very little information is known about habitat use or current distribution of the subspecies in these two counties. Quantifying and comparing both landscape and site specific characteristics of habitats where mice are found and not found in these two counties would further our understanding of habitat use by the subspecies, particularly in the northern half of its probable range. Any new locations where Preble's �82 meadow jumping mouse are found during this study would contribute to the range-wide distribution map currently available for the subspecies. Repeated annual visits to these same randomly selected sites would also provide information on population persistence at a given site. Such a monitoring scheme would be necessary for evaluating the continued status of the mouse at each site, providing further information as to the suitability of the habitat for long-term persistence of the population. Both Larimer and Weld Counties include habitat that is currently perceived to be outside the ecological limitations of Preble's meadow jumping mouse. Larimer County provides the opportunity to explore elevational limitations, currently thought to be 7400 feet (2260m) (USFWS 1997); Weld County extends beyond the currently believed eastern boundary of the subspecies. Both of these boundaries will be challenged by selecting survey sites within the two counties beyond the perceived ecological limits of the subspecies. Genetic tissue samples will be collected on all Preble's meadow jumping mice captured to assist identification through DNA analysis, particularly for those animals captured outside the currently perceived ecological limits, and to provide information for future studies to better define the relationships among different populations of Z. h. preblei, other subspecies of Z. hudsonius and other species of Zapus. Thus, the primary objective of this study is to quantify and define habitats used and identify ecological limitations of the subspecies in Larimer and Weld Counties, Colorado. Such information could be used by the USFWS to further refine the current definition of potentially suitable habitat for Preble's meadow jumping mouse. B. OBJECTIVES Specific objectives for this study on Preble's meadow jumping mouse (PMJM) include: 1. Define and quantify habitat characteristics at sites where PMJM was found and sites where PMJM was not found. 2. Identify elevational and eastern boundary limitations for PMJM in Larimer and Weld Counties, Colorado. 3. Describe the distribution of the PMJM in Larimer and Weld Counties, Colorado. 4. Select sites and establish monitoring protocols for determining long-term trends in populations of PMJM. 5. Collect genetic tissue samples to assure identification of PMJM captured during this study and for future genetic analyses to better define the relationship of populations of Z. h. preblei to each other, other subspecies of Z. hudsonius and other species of Zapus. C. EXPECTED RESULTS Conservation of Z. h. preblei should include maintaining populations of the subspecies throughout the range of its natural variation and to try to identify ecological limits for the subspecies. Key concerns include range-wide distribution, habitat use, and population persistence at a given site. This study will address these concerns. Comparisons of habitat variables from both successful and unsuccessful sites will be used to better define suitable habitat for the subspecies. Habitat variables recorded will include both site level characteristics such as vegetational composition, cover, and composition of other small mammal species as well as landscape level characteristics including connectivity with other potential sites, geology, hydrology, and distance to development. Repeated annual visits to the randomly selected sites will provide information on population persistence and site fidelity of individuals at the given sites. Such a monitoring scheme will be necessary for evaluating the continued status of the mouse at given sites and provide further information on the suitability of the �83 habitat to sustain populations of PMJM over the long term. Genetic tissue samples collected from PMJM captured will be used to confirm identification of the mice and for future analyses to better define the relationship among (1) different populations of Z. h. preblei, (2) other subspecies of Z. hudsonius and (3) other species of Zapus. Performa.nce Indicators 1. Provide a definition of and quantify habitat characteristics at sites where PMJM was found and sites where PMJM was not found in Larimer and Weld Counties, Colorado. 2. Update the known distribution of PMJM in Larimer and Weld Counties, Colorado. 3. Provide monitoring sites and protocols for determining persistence trends of PMJM at a given site. 4. Genetically identify all PMJM captured. 5. Complete a genetic tissue sample collection for future genetic analyses to better define the relationship of populations of Z. h. preblei to each other, other subspecies of Z. hudsonius and other species of Zapus. 6. Analyze and publish results of research. 7. Update and modify the CDOW conservation plan for PMJM. D. APPROACH Distribution Study Background The meadow jumping mouse (Z. hudsonius) is broadly distributed across North America from the Atlantic to Pacific coasts, extending south into the United States to Alabama and Georgia and west across the Great Plains to the base of the Rocky Mountains. In general, it is a common inhabitant of moist, grassy and herbaceous fields. Eleven living subspecies have been described (Whitaker 1972). Hafner et al. (1981) describe a twelfth subspecies, Z. h. luteus. Z. h. preblei occurs only in eastern Colorado and southeastern Wyoming (Krutzsch 1954, Long 1965, Armstrong 1972). From its limited ecological and geographic distribution, Fitzgerald et al. (1994) suggest it is an Ice Age relict, once widespread in tallgrass prairie across the eastern plains of Colorado but now restricted to scattered localities on the Colorado Piedmont. Similar relict populations of meadow jumping mice in the White Mountains of Arizona and the Sacramento Mountains and Rio Grande Valley of New Mexico are described as the subspecies Z. h. luteus (Hafner et al. 1981). Numerous surveys have been conducted since 1990 to establish the current distribution of Preble's meadow jumping mouse. Surveys have been funded by the U. S. Fish and Wildlife Service, Rocky Flats Environmental Technology Site, Colorado Division of Wildlife, U.S. Air Force Academy, Warren Air Force Base, Colorado Department of Transportation, City of Boulder Open Space, and City of Boulder Greenways Program (Armstrong et al. 1997, P. Plage, personal communication). From 1990 to 1997 such surveys have yielded captures of Preble's meadow jumping mouse, based on field identification and supported by the genetic analyses, at the following sites: 1. 2. 3. 4. Lone Pine Creek, Larimer County, Colorado Rabbit Creek, Larimer County, Colorado St. Vrain Creek and associated tributaries in Boulder County, Colorado City of Boulder Open Space, along South Boulder Creek and its tributaries, Boulder County, Colorado �84 5. 6. 7. 8. 9. 10. 11. 12. Coal Creek, Jefferson County, Colorado Rocky Flats Environmental Technology Site, Jefferson County, Colorado White Ranch Park, Ralston Creek, Jefferson County, Colorado Plum Creek drainages including Indian Creek, West Plum Creek, and East Plum Creek, Douglas County, Colorado Roxborough State Park, Douglas County, Colorado Hay Creek, Elbert County, Colorado Monument Creek on the U. S. Air Force Academy (AF AC) and tributaries of Monument Creek off the AFAC including Smith Creek, Pine Creek, and Jackson Creek, El Paso County, Colorado Medicine Bow National Forest, Albany County, Wyoming Two more sites yielded mice that were identified as Z. h. preblei in the field but were genetically found to be more closely allied with the western jumping mouse, Z. princeps. These sites include: 1. 2. Warren Air Force Base, Laramie County, Wyoming Lone Tree Creek, Weld County, Colorado Similarity of the habitat at these two sites compared to those where Z. h. preblei (as determined genetically) were found suggests the possibility of areas of sympatry or parapatry between the two species of Za.pus. According to Fitzgerald et al. (1994) the distributions of Z. princeps and Z. hudsonius do overlap. The area of overlap occurs in eastern Wyoming. The distribution of Z. hudsonius in Colorado is now known to be larger than shown in Fitzgerald et al. (1994). The boundaries are currently as far south as Las Animas County (based on genetically identified specimens of Z. h. preblei). Captures of Z. princeps and Z. hudsonius (as identified in the museum) have occurred as close as eight miles of one another within the same drainage (Armstrong 1972). Z. princeps were reported captured in 1981 (Olson and Knopf 1988) at the Lone Pine site in Larimer County, Colorado, where Z. h. preblei were captured this year. Because neither specimens or genetic samples were taken in the 1981 study, identification of those mice will remain in question. The discrepancy may be explained by field misidentification or Meaney et al. (1997) also suggest this discrepancy might be explained by displacement of Z. princeps with Z. h. preblei sometime in the sixteen intervening years between trapping efforts. Although assumed to have different ecological requirements, genetic evidence presented here suggests further investigation of possible distributional overlap between Z. h. preblei and Z. princeps. Several notable results from the genetic analysis of specimens acquired from the Denver Museum of Natural History may affect the location of the southern distribution of Z. h. preblei. One site yielded mice identified as Z. p. princeps in the field but were found to be genetically more closely allied with Z. h. preblei. This site is the most southern location to date of Z. h. preblei and is located at: 1. Purgatoire Campground, San Isabel National Forest, Las Animas County, Colorado Also from Las Animas County were mice collected and identified as Z. h. luteus (Jones 1996), an identification also supported by the genetic analysis. These mice were from: 1. 2. Lake Dorothey State Wildlife Area, Chicorica Creek, Las Animas County, Colorado Lake Dorothey State Wildlife Area, West Fork Schwacheim Creek, Las Animas County, Colorado These sites suggest a more northerly distribution of Z. h. luteus, currently known only from limited areas in New Mexico and Arizona (Hafner et al. 1981 ). Identifying the southern boundary of Z. h. preblei clearly needs further study. �85 Objective Conservation of Z. h. preblei should require maintaining populations of the subspecies throughout the range of its natural variation and to try to identify ecological limits for the subspecies. Specific objectives for the distributional component of this study are to (1) clarify the distribution of the subspecies in Larimer and Weld Counties, Colorado, and (2) define and quantify the amount of suitable habitat within Larimer and Weld Counties. Approach To meet these objectives, this study will include (I) constructing a sampling frame of potentially suitable habitat within the area of interest (Larimer and Weld Counties, Colorado) and (2) conducting trapping surveys to determine presence or absence at 30 sites randomly selected from the sampling frame. Specific approaches to each of these tasks follow. Sampling.frame A sampling frame will be developed to delineate potentially suitable habitat for PMJM from unsuitable habitat. To produce a preliminary map displaying existing potential habitat for Preble's meadow jumping mouse would require at least three primary layers of ecological information. Two initial layers, hydrology and vegetative cover would provide a map displaying all areas where these two ecological requirements of Preble's meadow jumping mouse co-occur. The hydrology layer must be in enough detail to provide locations of intermittent and small order streams as well as identify water source (e.g., irrigation ditches, stream, seep). Degree of density of vegetative cover is sufficient for initial mapping efforts. Demarcation of shrub, trees and grasses will suffice in these initial efforts rather than species identification. However, not all combinations of vegetative cover and water provide sufficient habitat to support Preble's meadow jumping mouse. Thus, mapping potential habitat in such a way would produce a map displaying far more potential habitat than exists. For example, an extremely critical ecological requirement for the survival of the mouse, that to date we have very little information for, is potential hibemacula sites. As a possible index to hibemacula habitat, a third GIS layer, indicating the presence of alluvial deposits may provide some insight in to hibernation requirements. Areas where alluvial deposits co-occur with the presence of water and vegetative cover may provide a more realistic map of potentially suitable habitat for the mouse. Our ability to survey a site selected at random will be restricted to those areas where we are able to obtain permission to access the land. We should be able to survey any randomly selected site on public lands and thus our inference should be unrestricted. Such readily available access will probably not occur on privately owned or leased lands, thus, restricting inferences made on private lands. By stratifying the sampling frame, this lack of access on private lands will not restrict inferences that can be made on public lands. Inference on private lands will depend on the percent of access we get from private landowners. The sampling frame will be developed from the 1:24,000 scale Hydrographic GIS layer developed by the CDOW (Reese Tietje, unpublished data). From this sampling frame a three-stage sampling design will be used to select the 30 random sites for doing PMJM surveys. Primary sampling units will be 3rd order streams. Of the 150 3rd order streams that exist within the sampling frame, 30 will be selected on public lands, 30 on private lands. The secondary sampling unit will be the stream stretch selected within the primary (3rd order stream) unit. The tertiary sampling unit will be the actual sites along the stream stretch selected at the secondary sampling stage that will be sampled. Each of the primary, secondary, and tertiary sampling units will be selected using simple random sampling. If the secondary sampling unit does not yield potentially suitable habitat (i.e., no dense vegetation) secondary sampling units will continue to be selected at random until one yields potentially suitable habitat. Determination of suitable habitat will be made by field site visits. Total stream length of all the secondary sampling units will be recorded as well as total stream length containing potentially suitable habitat. From these measurements an estimate of the probability �86 of potentially suitable habitat occurring within the primary sampling unit will be made for each area surveyed. A mean estimate, over all sites surveyed, of the probability of a primary sampling unit having potentially suitable habitat will then be made. Because of the random selection of primary sampling units, inference can be made as to the probability of potentially suitable habitat existing on all 3rd order streams within the sampling frame. By recording presence or absence of PMJM within the tertiary sampling units we can estimate the probability of PMJM occurring within potentially suitable habitat. Because of the random selection of sampling units in each of the two strata {public and private lands) we can then estimate the probability of PMJM occurring in potentially suitable habitat within both public and private lands. Trapping surveys Trapping surveys to determine presence/absence of PMJM will primarily be conducted following protocols defined by the USFWS (1997). Additional guidelines have also been developed to accommodate the needs of this project. A brief summary of these guidelines follow. Landowner contact 1. To request permission to conduct surveys on private property, landowners will first be notified by letter. The letter will briefly describe the project and outline specifically how they might be affected if PMJM were found on their property. The letter will be followed by a phone call to determine if permission will be granted by the landowner for access to the property or not. 2. No private property will be surveyed without prior consent of the landowner. 3. To request permission to conduct surveys on public land, the agency (e.g., USFS, State Forest, etc.) responsible will be contacted. Timing ofsurveys I. Trapping surveys will be conducted within the period from June 1 to September 15, 1998. These dates correspond to dates when the mouse is known to be out of hibernation. 2. Due to the nocturnal nature of PMJM, traps will be set between l 900hrs and 21 00hrs and checked as early as possible in the morning beginning at 500hrs (to reduce stress and the potential for predation on trapped animals). Time required to complete the traplines will vary depending on how many animals are caught. 3. Absence of PMJM at a site will be defined as no captures of PMJM from a minimum of three consecutive nights and 750 trapnights (a trap night is defined as the sum total number of traps available each night). 4. Presence of PMJM at a site will be defined as at least one capture of PMJM at the site. However, to accommodate sample size requirements for the monitoring and genetics study (see below) trapping efforts will continue until at least 10 individuals are tagged with Passive Integrated Transponders (PIT) tags. Survey protocols 1. Surveyors will be advised to follow the Center for Disease Control's Hantavirus instructions and recommendations when dealing with rodents which include: a Baseline blood serum samples will be collected and stored at Poudre Valley Hospital for all surveyors. b. Respirators are available for use by surveyors if they request their use. c. Surgical gloves will be used at all times when handling rodents. d. All equipment will be rinsed first with a 10% bleach solution, then water after use. e. Traps and any equipment that may have rodent feces or urine on it will not be transported in the interior of a vehicle. �87 2. 3. f. Traps will be cleaned after each use by a mouse. g. If a surveyor becomes ill with hantavirus symptoms they will report immediately to a medical facility. Surveyors must be in possession of both a federal and state collecting permit. If PMJM is found at previously undocumented sites these locations must be reported within one day to the USFWS. Survey methodology I. Small mammal Sherman live traps (folding and non-folding) will be used to conduct the surveys. 2. Traps will be set in two parallel lines of trap stations (I trap per station) on either side of the drainage. Trap stations will be 5 meters apart for a total of 250 meters; the parallel transects will be 10 meters apart unless extent of habitat, terrain topography, or stream hydrology do not allow. 3. Location of transects will be recorded on field data sheets (see Appendix B) and identified on 7½ minute topographic maps. Locations will also be recorded to the nearest 5 meters in UTM coordinates using a Trimble Geo-Explorer GPS. 4. In case of windy conditions or large number of trap-tampering predators (i.e., raccoons, foxes, coyotes, etc.), traps will be secured to the ground with hoops of heavy-gauge malleable wire, stakes, or with other materials that can effectively secure/immobilize the traps. 5. A small (~I inch) ball of polyester quilt or wool (fleece) will be placed in each trap as nesting/bedding material. 6. Baiting materials will be Manna Pro Sweet 3-way Livestock feed which contains no animal matter. Ingredients include flaked barley, flaked com, flaked oats, and cane molasses. Peanut butter will be used to stick the bait to the trap. 7. Checking of the traps will be conducted by two surveyors; one person will handle the traps and animals captured, the other person will record the data Handling ofcaptured animals I. All animal captures will be recorded (see Appendix A, Data Forms). I. If an animal has been captured in a trap, a ziplock plastic bag will be placed over the end of the trap. The trap will be opened allowing the animal to fall into the plastic bag. The animal will be identified to species while in the plastic bag. 2. If the animal is not a PMJM, identification of the animal will be recorded and the animal set free. 3. Each captured PMJM will be scanned to detect the presence/absence of a PIT tag. If a PIT tag is detected and the mouse has been captured at that same site within the 4-day trapping effort, identification of the mouse will be recorded and the mouse will be released. 4. If the animal is a PMJM and no PIT tag is detected.the PMJM will be weighed while in the bag, recording the weight in grams using a Pesola spring balance. 5. Sex, age (juvenile, adult), and reproductive condition (pregnant, lactating or nonlactating, inactive if it is female; for males the position of the testes - scrotal, inguinal, or abdominal) will be noted. 6. Each PMJM will be measured (total length, length of body, length ofhind foot-heel to distal end of claws) in millimeters. Measurements of total body and tail length will be taken with the mouse on a firm surface to minimize error. 7. Capture of every PMJM will be documented by taking two photographs of the mouse on first capture. If the mouse has been captured previously as noted by the detection of a PIT tag, no photograph needs to be taken. �88 8. 9. 10. 11. 12. 13. 14. Each PMJM will have a PIT tag inserted above the shoulder blades by lifting the skin on its back and inserting the needle with the PIT tag under their skin and injecting the tag. Verification of the PIT tag identification number will be made before insertion into the mouse by running the PIT tag scanner across it. Skin behind the opening will then be pinched to prevent emergence of the tag. Surgical glue will then be applied to the opening to prevent the PIT tag from exiting the body and prevent infection of the mouse. A PIT tag scanner will be held over the mouse to again verify PIT tag identification number and that it still works after the insertion procedure (note: PIT tag will be scanned twice during application to mouse, or once if mouse was previously PIT tagged) .• PIT tag identification number will be recorded on the field data form and the mouse released. If there are feces in the trap where a PMJM was captured, the feces will be collected in a plastic bag labeled with date and location of the site. Fecal samples will be kept cool until returned to the CDOW office where they will be frozen. Fecal samples will be analyzed by the Composition Analysis Laboratory, Inc, 622 ½ Whedbee, Fort Collins, CO for content. If, at any time during the handling of the PMJM the animal appears to be severely stressed (dramatic changes in respiration, heartbeat or responsiveness or gums turning blue) the animal will be released immediately. Handling of trap mortalities and injuries 1. All trap mortalities will be recorded on a 'Trap Mortality' data form which includes information on species, potential duration of time spent in the trap, any information available on cause of death, etc. (see Appendix A, Data Formsf 2. All animals found dead in a trap will be double-bagged in a plastic bags and placed in a cooler with ice. Specimens should be frozen as soon as possible and deposited the CDOW freezer at the office in Fort Collins. All specimens will be sent to the Colorado State University Diagnostic Laboratory for detection of any disease and/or other conditions that may have contributed to cause of death. All PMJM specimens will be returned to the CDOW. 3. A museum card will be completed and attached to each PMJM specimen and given to Cheri Jones, Curator of Mammalogy at the Denver Museum of Natural History, for study skins and tissue storage once the Diagnostic Laboratory has completed their evaluation. 4. If an animal is severely injured it will be euthanized by soaking a cotton ball in Metofane (methoxyflurane) and placing the cotton ball and the mouse in a ziplock bag until the animal stops breathing. All specimens will then go through steps 1-3 above. 5. If an animal appears to be only slightly injured (e.g., broken tail, small laceration) the animal will be released to the wild. If an animal appears to be cold stressed attempts will be made to warm it by holding it in the surveyors hands and/or against their body. If the animal appears to be heat stressed, isopropyl alcohol will be applied with a cotton swab to the ears, arm pits, and feet to cool it down. Training offield crews 1. 2. All field technicians will be required to spend a full day at the Denver Museum of Natural History Zoology Department viewing and handling specimens of all small mammal species likely to be captured in Larimer and Weld Counties. All field technicians will be provided with a small mammal field guide and a key specifically designed to list the species most likely to be captured on the trapping sites. �89 3. All field technicians will participate in 4 days of practice trapping at the Frank State Wildlife Area and/or the Air Force Academy from May 26-29. During these trapping sessions all technicians will learn how to place, bait and set traps. Field identification will be practiced on all animals caught each day and verified by crew leaders. All field technicians will also practice handling, measuring, implanting PIT tags, and taking genetic tissue samples from deer mice (Peromyscus maniculatus) until they are competent in this procedure. Habitat Use Study Background Site scale: Based on studies of Z. h. preblei and Z. hudsonius elsewhere, Z. h. preblei apparently occurs mostly in undergrowth consisting of grasses, forbs, or both in open wet meadows and riparian corridors, or where tall shrubs and low trees form an overstory and provide adequate cover (Armstrong et al. 1997). Meadow jumping mice are widespread in abandoned grassy fields, but is often more abundant in thick vegetation along ponds, streams, and marshes or in rank herbaceous vegetation of wooded areas (Whitaker 1963). Preble's meadow jumping mice have been trapped in natural riparian areas as well as areas altered by anthropogenic influence including ditches and wetlands adjacent to interstate highways, cement-lined ditches with tall cover, ditches along driveways and moderate road use, and moderate cattle grazing (M. Bakeman, personal communication). The majority of sites where Z. h. preblei have been found consist of multistoried cover but the species composing the cover vary greatly (Armstrong et al. 1997, Meaney et al. 1997). Vegetation composition of the dense cover varies considerably and includes both native and non-native species (Meaney et al. 1996, 1997, Armstrong et al. 1997, M. Bakeman, personal communication). The herbaceous understory can be primarily grasses or forbs or a mixture of the two. Few of the sites, however, are dominated by fewer than two understory species. The tall shrub canopy at most sites is willow (Salix spp.), although scrub oak (Quercus gambelli), birch (Betula spp.), and alder (A/nus spp.) occur in sites south of the Palmer Divide (Armstrong et al. 1997). Ponderosa pine (Pinus ponderosa) is the most common tree at higher elevations. The mouse appears to tolerate weedy or exotic species in areas that are structurally diverse and species rich; nearly every successful site contained Canada thistle (Cirsium arvense) (Armstrong et al. 1997). Thus, the mouse does not appear to have an affinity toward any single plant species but instead favors sites that are structurally diverse and provide adequate cover and food throughout its life cycle. Preble's meadow jumping mouse mice are typically not found in upland areas away from riparian habitats but are most often captured where either ground water surfaces to seep springs or on main water channels (M. Bakeman, T. Ryon, personal communication) suggesting a dependence on open water, at least during their active periods. Movement of Z. h. preblei on Woman Creek at Rocky Flats Technology Site suggests mice move along corridors of shrub cover, generally Salix exigua (PTI 1996a, T. Ryon, unpublished data), suggesting dispersal habitat is similar to habitats used for other activities. Small mammal assemblage: Preble's meadow jumping mice are only one component of the small mammal community inhabiting areas where they were captured. These mice were more often found at sites with high species richness and small mammal abundance (Armstrong et al. 1997, Meaney et al. 1997). The variety and relative abundance of other small mammalian species trapped during surveys for Preble's meadow jumping mouse provide some framework for comparison of small mammal assemblages at sites where Z. h. preblei occurred and sites where none were captured. All trapped sites were either historical sites of known occurrence or apparently suitable habitat for Z. h. preblei, providing a common basis for comparison. Three small mammal species (Spermophilus variegatus, Peromyscus nasutus, and Rattus norvegicus) were not captured where Z. h. preblei occurred but were captured in unsuccessful sites. However, captures of these three species only occurred at one or two sites and the species comprised an extremely small percentage of the species �90 caught in those areas providing too little information for speculating on possible interactions between these species and Z. h. preblei. The higher occurrence and percent composition of capture of Mus musculus, the house mouse, in areas where Z. h. preb/ei was not caught might suggest degradation of habitat for Z. h. preblei in those areas or possibly competition between the two species (Ryon 1996). The house mouse, thrives in areas of human habitation and croplands, but also occurs in abandoned fields and ditch banks where they may displace native rodents (Fitzgerald et al. 1994). Ryon (1996) also reported the presence of domestic cats (Fe/is catus) at sites where Z. h. preblei historically occurred but were not found during his study. Contrarily, C. Miller (personal communication) reported house cats along South Boulder Creek where Preble's meadow jumping mice are known to occur. Landscape scale: Areas of suitable habitat must provide requirements for PMJM to survive throughout its life cycle. These requirements must provide necessities for both the active period and hibernation periods. During the active period suitable habitat must provide requirements for daily survival, reproductive activities (breeding, nesting, and rearing of young to independence), and dispersal. The hibernation period requires sufficient food supplies to assure fat storage prior to hibernation and suitable hibernacula Habitat providing all seasonal and life cycle requirements may or may not occur in a single contiguous area If not in a contiguous area, habitat patches must occur in a mosaic of usable areas where suitable corridors exist for seasonal movement among sites. The habitat matrix within the range of Z. h. preblei is mixed grasslands adjacent to the Colorado Front Range along the Piedmont and along the base of the Laramie Mountains in Wyoming and extends to the Colorado plains. Within this matrix, Preble's meadow jumping mice occur along stream drainages that contain patches of suitable vegetation. Suitable habitat appears to have at least two major components. The first component is a supply of open water, at least in part of the active season (M. Bakeman, C. Meaney, personal communication). Secondly, areas where Preble's meadow jumping mouse has been found have dense cover (M. Bakeman, C. Meaney, personal communication). If Preble's meadow jumping mouse behaves as a metapopulation in the classical sense [a set of local populations which interact via dispersal of individuals moving among populations and where local extinctions and recolonizations occur (Levins 1970)] then habitat includes not just one area of suitable habitation but also areas suitable for nearby mouse populations. These suitable areas must also be linked by dispersal habitat. If the mice are dependent on dense riparian habitat for dispersal, as well as for areas to reproduce, persistence of discrete populations would require a mosaic of suitable discrete riparian patches interconnected with dispersal corridors of similarly dense riparian vegetation. If mouse populations function in a source (populations where growth rate~ 1) and sink (those populations where growth rate < 1, maintained through immigration) system, it will be critical to identify and protect those populations serving as sources. Thus, for a source-sink population critical habitat will include those areas that support source population, dispersal habitat to sink areas will be less critical. If local mouse populations are functionally discrete, such a mosaic of interconnected areas of suitable habitat would provide a buffer for local, and source, populations against deleterious stochastic events by providing the opportunity for local population failures to be 'rescued' by immigration from other populations. Objective The objective of the habitat study is to identify and refine ecological requirements of Z. h. preblei in Larimer and Weld Counties, Colorado. Approach The general approach is to measure both site specific and landscape features of 30 sites selected at random to be surveyed for the presence of PMJM. A comparison of sites where PMJM are found and are not found will then be made in an attempt to refine our current understanding of the ecological requirements of the subspecies. The following habitat characteristics will be measured and recorded for each site. �91 Micro-site habitat characteristics 1. Cover will be measured at 20 random locations along the 250 meter sampled stream stretch. Mean cover and standard error of the mean will be estimated. Cover will be estimated using a vegetation profile board (Nudds 1977) that allows for an assessment of visual obstruction in 0.5 meter vertical intervals above ground. The board will be 1.5 meters high and 30.48 cm wide. The board is marked in alternate black and white colors at 0.5 meter intervals. Horizontal cover is assessed in each interval by viewing the board from 15 meters away in a randomly chosen direction. The percentage of each interval concealed by vegetation is recorded as 0-20, 21-40, 41-60, 61-80, and 81-100% estimated concealment. 2. Vegetation species composition and richness will be estimated using the Modified-Whittaker nested vegetation sampling method (Stohlgren et al. 1995). A modified Whittaker plot is 20meters x 50-meters with 10 lm2 and 2 10m2 subplots arranged systematically around the perimeter and 1 100m2 subplot centered in the inside of the 20 meter by 50 meter plot. Species composition and percent cover (optical estimate) of each species is recorded for each subplot. 3. Soil samples will be taken at each site for a hydrometer textural analysis (percent sand, silt and clay). Samples will be collected using a soil probe. Samples will be taken from 0-12 inches, and 12+ to 24 inches. The 0-12 inch sample will be taken first, removing the probe and soil. Sample will be placed in a labeled ziplock bag. The probe will then be placed in the same hole for the 12+ - 24 inch probe, with that soil sample being placed in a separately labeled ziplock bag. The hydrometer textural analysis will be conducted at the CSU Soils Laboratory. 4. Mean stream width will be estimated by measuring stream width at 30 locations along the entire site sampled. The 30 locations will be stratified equidistant from each other to cover the entire stream stretch in increments of site stream length/30. 5. Compare parameter estimates from 1-4 for sites where PMJM were captured and sites where PMJM were not captured. Landscape habitat characteristics The following landscape habitat characteristics will be measured using either GIS, topographic maps, or aerial infrared photography. 1. Estimate distance to nearest human habitation and human disturbance. 2. Estimate connectivity of sites to other streams from I :24,000 topographic maps. 3. Estimate total length of stream stretch at each site and total length of stream stretch with suitable habitat at each site. 4. Compare parameter estimates from 1-3 for sites where PMJM were captured and sites where PMJM were not captured. Monitoring Study Background For purposes here, population persistence is defined as the presence of Z. h. preblei at the same site for multiple years. The majority of recent locations of meadow jumping mice have been the result of survey efforts that focused only on determining presence of Z. h. preblei. Once survey protocols (USFWS 1997) are met [i.e., a minimum of 400 trapnights (one trap set for one night= I trapnight) conducted] sites are typically not revisited eliminating the possibility of determining population persistence at these sites. Thus, these and other sites not yet surveyed may or may not have persistent populations. Areas where trapping was conducted over multiple years as part of further research, yielded three sites in Colorado that support persistent populations: Rocky Flats Environmental Technology Site, the U. S. Air Force Academy near Colorado Springs, and Boulder Open Space on South Boulder Creek. Besides establishing the presence of Z. h. preblei over multiple years, other considerations of persistence include (I) presence in consecutive years versus a 'blinking' in and out of a population as would be expected in a metapopulation, (2) how populations are sustained in a given area (e.g., source or sink populations), (3) maximum abundance supportable by a given area, and (4) fluctuations in �92 abundance of a population over years. If Z. h. preblei exist in areas as part of a metapopulation or as a series of source-sink populations it would be critical to conserve and protect all key sub-population areas as well as critical habitat for dispersal. Detailed population studies, including estimating and determining factors affecting survival, reproduction, and dispersal rates, must be conducted to determine if Z. h. preblei occurs as either metapopulations or in source-sink populations. There have been no such studies conducted on Z. h. preblei or Z. hudsonius elsewhere to provide any supporting or refuting evidence for either metapopulation or source-sink structure. Consistently low abundance in a given area due to limiting ecological requirements (e.g., size of area of suitable habitat) or fluctuations in population abundance, including years of low abundance, must be considered in any conservation strategies for Z. h. preblei because of the threat of complete loss of the population due to catastrophic or extreme environmental conditions. Some evidence exists for such fluctuations in population abundance at Rocky Flats Environmental Technology Site. Trapping surveys on the Woman Creek drainage yielded the following captures of Z. h. preblei: seven in 1993 (EG&G 1993), zero in 1994 (T. Ryon, unpublished data), one in 1995 (T. Ryon, unpublished data), two in 1996 (PTI 1996a), and 33 in 1997 (T. Ryon, unpublished data). Trapping effort was consistent from 1995-1997. Repeated trapping surveys conducted over different years intermittently yielded successful captures of Z. h. preblei along the St. Vrain Creek and lower Coal Creek (three in 1989, zero in 1992, zero in 1994, zero in 1996, one in 1997, M. Bakeman, unpublished data). If protocols in each year were the same, these data also suggest populations may undergo fluctuations in abundance. Objective The objective of the monitoring study is to establish a monitoring protocol to evaluate the status (i.e., presence or absence) of Z. h. preblei populations at specific sites and to detect changes in local distribution from year to year. Detection of PIT-tagged individuals from year to year will also provide evidence for site fidelity and/or movements of individual mice from one area to another. Approach Trapping surveys will be repeated each year at 20 randomly selected sites of those surveyed. Of the sites surveyed, monitoring sites will be established from a random selection often sites where no PMJM are captured and 10 sites where PMJM are captured in 1998. These 20 sites will be surveyed every year thereafter. 1. 3. 4. 5. 6. 7. Trapping surveys will be conducted as described in the Distribution Study. Habitat will be evaluated at all sites as described in the Habitat Use Study and comparisons made to results from previous years. Each PMJM captured will be PIT-tagged as per the protocols outlined in the Distribution Study. Population persistence at a given site will be evaluated as the number of years when a given site hadPMJM. Recaptures of PIT-tagged individuals will be mapped from year to year to provide evidence for site fidelity and/or movements of individual mice from one area to another. Detailed protocols for long-term data analyses will be designed once a state-wide, long-term monitoring program is fully developed. Molecular Systematic Study Background The family Zapodidae Gumping mice) consists of small to medium-sized mice with enlarged hind feet and exceptionally long tails. Four living genera are recognized in this family, two of which, 'Zapus and Napaeozapus, are found in North America (Hall 1981). There are three living species of the genus 'Zapus: Z. trinotatus (Pacific jumping mouse), Z. princeps (western jumping mouse), and Z. hudsonius (meadow jumping mouse). Z. h. preblei is one of twelve living subspecies of the species Z. �93 hudsonius (Krutzsch 1954, Hafner et al. 1981). Z. h. preblei was first described by Krutzsch (1954) from a specimen collected by E. A. Preble in 1895 near Loveland, Colorado. Quimby (I 951) described the species Z. hudsonius as follows: 'A mouse-like rodent with greatly enlarged hind feet and an exceptionally long tail. The forelegs are relatively short. The ears are somewhat conspicuous. The body is clothed in moderately long, somewhat dense hair of a rather coarse texture and several colors. The dorsal portions are marked by a broad stripe of brownish hairs many of which are tipped with black giving the region a grayish-black appearance. The sides are bright yellowish-orange, whereas the underparts and feet are white. The tail is bicolor, dark above and light below, and sparsely covered with hair which is longer on the terminal part. The mammae are eight, and quite prominent in lactating females. The male genitalia are inconspicuous except during the breeding season when the scrotal sac becomes enlarged. The testes enlarge and may be either abdominal, inguinal, or scrotal during this period.' The skull of Z. hudsonius is small and light with a narrower braincase and smaller molars than in Z. princeps (from Fitzgerald et al. 1994 p. 18). However, Fitzgerald et al. (I 994) urge caution in distinguishing the two species of Zapus in Colorado. Such caution is further warranted by the recent conflicting identifications of mice based on genetic characteristics. Analysis of mitochondrial DNA sequence data from 92 individual mice indicates that mice sampled from southeastern Albany County, Wyoming, south along the Front Range of the Rocky Mountains to western Las Animas County, Colorado (Purgatoire Campground, San Isabel National Forest), form a coherent genetic group (Riggs et al. 1997). This group of samples are distinct from samples obtained from mice from three other populations. Genetic samples from mice captured in the Dorothey Lakes area of southern Colorado (Las Animas County) group together and are most closely allied with Z. h. luteus, a subspecies described from New Mexico (Riggs et al. 1997). The single genetic sample collected from Weld County, Colorado (Lone Tree Creek) and six samples obtained from Warren Air Force Base in Laramie County, Wyoming are most similar to reference samples of Z. princeps from Colorado (Larimer County) and New Mexico (Taos County) (Riggs et al. 1997). The sequence data indicate that the samples from specimens identified as Z. h. luteus and samples from specimens of Z. princeps are more closely allied with each other than either is with the samples defining the Z. h. preblei group. This closer alliance of the samples of Z. h. luteus with Z. princeps rather than Z. h. preblei conflicts with results ofHafuer et al. (1981), based on a combination of pelage, morphologic, and genetic data, which support a closer alliance with Z. hudsonius. A more complete biosystematic evaluation of jumping mice is needed to clarify and further refine relationships among populations of the group referred to as Z. h. preblei as well as to other subspecies and species of the genus Zapus. Such an evaluation requires detailed analyses of pelage, morphometric, and genetic data from sufficient numbers of individuals to adequately represent the populations of interest. However, the mitochondrial DNA non-coding (D-loop) sequence data available at this time are consistent with the view that a geographically contiguous set of populations previously recognized as Preble's meadow jumping mouse form a homogenous group recognizably distinct from other nearby populations and from another geographically-adjacent species of the genus (Riggs et al. 1997). Therefore, given the genetic data available, Riggs et al. (I 997) conclude Preble's meadow jumping mouse is a distinct population of Z. hudsonius. Objective Because of the uncertainty of the identification of the mice captured in Weld County, Colorado, and because currently perceived ecological boundaries of PMJM will be challenged in the survey effort, the objective of the genetic component of this study is to genetically identify the PMJM captured during this study. Therefore, genetic tissue samples will be collected from each PMJM captured to confirm identification of the animal. These genetic tissue samples could also provide further data to better define the relationships of different populations of Z. h. preblei as well as to other subspecies of Z. hudsonius and other species of Zapus through molecular systematic relationships and to link these genetic relationships to systematic studies of Z. hudsonius. Efforts are underway to secure funding for these studies. �94 Approach Genetic tissue samples will be collected from every PMJM captured during the study. To date, positive identification of the individual to subspecies cannot be made using only genetic information. However, results of the DNA analysis combined with the morphometric data that will also be collected (e.g., body length, tail length) and the photograph of the individual should provide sufficient information to support the field identification to subspecies. As established by the USFWS (1997) Survey Guidelines, presence of PMJM is established when only one individual is captured. Thus, we could stop our trapping efforts after the first PMJM capture. However, because we would also like to provide sufficient genetic data to more fully explore the relationships of different populations of Z h. preblei we will attempt to collect genetic tissue samples from a minimum of IO individuals per successful survey site. A sample of at least IO individuals from a population will provide enough information to document the genetic variation within the population (T. Quinn, personal communication). Genetic tissue sampling: collection ofear tissue samples using ear punch tool The following protocol has been used on PMJM (M. Bakeman, C. Meaney, T. Ryon, R Schorr, personal communication). Before checking traps, clean the ear punch tools by immersing them in a I 0% bleach solution for a few minutes, then rinsing thoroughly with clean water. Dry tools thoroughly. A small screw-cap container will be useful to contain the bleach solution and to receive used ear punch tools. A sports bottle or other container with a squirt or pop-up top may be used for rinsing. Wash any previously used containers thoroughly by hand or in the dishwasher and do not use for other purposes (e.g., drinking) while in the field. I. Use a fresh pair of clean latex gloves when handling each mouse. 2. With a clean ear punch tool, obtain 2 tissue plugs from the mouse's ear. If there is excessive bleeding, gentle pressure will be applied to the injured area for approximately I minute. Punch may be applied to deliver the plug as a small circlet of tissue (about the size of the head of a pin) onto a finger of your gloved hand. [Placement of the punch may be done in any way that is convenient or useful as an ear mark in the context of your own work. A consistent location on one ear can help identify recaptured mice that have already been sampled.] 3. Place the finger tip tightly over the end of an opened sample vial and shake to wash the tissue plug into the ethanol. 4. Repeat until all three tissue plugs have been collected into the same sample vial. Inspect the tube to see that all plugs have made it into the ethanol. 5. Close the sample vial--screw the cap on firmly and check that there is no leakage of ethanol when the tube is inverted. Label both the vial and the corresponding record on the data sheet (see item 6 below) with a unique identifier as follows. Use a seven-place alpha-numeric code composed of the following: - A three-letter designator for the survey location (e.g., STV = St. Vrain, SCR = Smith Creek). A number beginning with two digits indicating the year (e.g., 98) followed by 2 digits specifying the individual trapped, numbered sequentially. Example: STV9801 = first animal sampled at St. Vrain in 1998. 6. Note: The combined seven-place alpha-numeric code needs to be a unique identifier for an individual mouse across all locations and sites being trapped in studies coordinated by the Colorado Division of Wildlife. Record information for the individual sampled on the data sheet provided. Be sure to include all data requested. Repetitive entries may be completed before or after the field session but we �95 recommend not allowing much time to elapse before doing this--what is obvious to you at the time may not be so obvious to us, or to you, later on. Information for each sample needs to be complete and unambiguous if the sample itself is to be useful. After returning from the field, samples will be kept in a cool place or refrigerated until delivery to the CDOW office at 317 West Prospect, Fort Collins, CO where they will be kept in the freezer for analyses. E. LOCATION Trapping surveys will be conducted at 30 sites selected at random from the sampling frame developed for Larimer and Weld Counties, Colorado. No survey site will occur at elevations greater than 7600 feet and not east of the UTM Easting Coordinate of 602000. The molecular systematic studies will be conducted in genetic laboratories using the genetic tissue samples collected in the field (ear punches). Training sessions for all field technicians and crew leaders will take place at the Frank State Wildlife Area located in Larimer County, Colorado 2.3 miles east ofl-25 on Highway 392 and 0.5 miles south on County Road 13. Data analyses and office work will be conducted at the CDOW Research Center, 317 West Prospect, Fort Collins, Colorado. F. SCHEDULE March 1998 April 1998 May 1998 May 1998 May 1998 June-September 1998 October 1998 Oct 1998-March 1999 December 1998 Aprill999 Aprill999 G. Purchase equipment for summer 1998 field season Complete sampling frame and select trapping sites Complete Study Plan for a distribution study of Preble's meadow jumping mouse Advertize for, hire, and train technicians Train field crews Conduct trapping surveys, collect site habitat data Conduct GIS analyses of the sites surveyed Conduct analyses Provide USFWS with a preliminary report of the results of the study Complete analyses for 1998 season. Submit Final Report for the 1998 season to the CDOW and USFWS PERSONNEL Tanya Shenk Gary C. White I CDOWTFTE 4 CNHP temporary employees Principal Investigator Statistical Consultant Field Crew leader Field Technicians �96 H. BUDGET The following is a budget to conduct habitat use and distributional information for 30 sites to be completed in summer 1998. Operating Expenses PIT Tags PIT tag scanners PIT tag software genetic sampling kit expenses traps topographic maps photographic equipment, expenses laboratory analyses (fecal, soil) camera trap supplies misc. equipment (gloves, cotton, bait, etc.) $ $ $ $ $ $ $ $ $ $ Personnel 4 field technician for 3.5 months each 1 crew leader for 5 months GIS contract work Statistical consultant $24,000 $24,000 $ 4,000 $ 3,000 Travel 30,000 miles @0.12 per mile vehicle rental: 3 vehicles, 4 mo. @$120 per month overnight travel 30 days @$75 per day $ $ $ TOTAL $80,290 I. 3,000 1,800 300 500 2,000 500 3,500 1,200 5,000 200 3,600 1,440 2,250 LITERATURE CITED Armstrong, D. M. 1972. Distribution of mammals in Colorado. University of Kansas, Museum of Natural History Monograph 3:1-415. Armstrong, D. M., M. E. Bakeman, A. Deans, C. A. Meaney, and T. R Ryon. 1997. Conclusions and recommendations in: Report on habitat findings on the Preble's meadow jumping mouse. Edited by M. E. Bakeman. Report to USFWS and Colorado Division of Wildlife. EG&G. 1993. Report of Findings: 2nd Year Survey for the Preble's meadow jumping mouse. Prepared by Stoecker Environmental Consultants for ESCO Associates, Inc., Rocky Flats Environmental Technology Site, Jefferson County, Colorado Fitzgerald, J.P., C. A. Meaney, and D. M. Armstrong. 1994. Mammals of Colorado. Denver Museum of Natural History, University Press of Colorado. Niwot, Colorado. Hafner, D. J., K. E. Petersen, and T. L. Yates. 1981. Evolutionary relationships of jumping mice (genus Zapus) of the southwestern United States. Journal ofMammalogy 62:501-512. �97 Hall, E. R 1981. The mammals of North America John Wiley and Sons, Inc., New York, New York, 2 volumes. Jones, C. A. 1996. Mammals of the James John and Lake Dorothey State Wildlife Areas. Final Report, submitted to the Colorado Division of Wildlife and Colorado Natural Areas Program. Krutzsch, P.H. 1954. North American jumping mice (genus Zapus). University of Kansas Publications, Museum of Natural History 7:349-472. Levins, R 1970. Extinction. Lectures in Mathematical Life Sciences 2:75-107. Long, C. A. 1965. The mammals of Wyoming. University of Kansas Publications, Museum of Natural History, 14:493-758. Meaney, C. A., N. W. Clippinger, A. Deans, and M. OShea-Stone. 1996. Second year survey for Preble's meadow jumping mouse (Zapus hudsonius preblei) in Colorado. Report prepared for the Colorado Division of Wildlife. Meaney, C. A., A. Deans, N. W. Clippinger, M. Rider, N. Daly, and M. O'Shea-Stone. 1997. Third year survey for Preble's meadow jumping mo.use (:lapus hudsonius preblei) in Colorado. Report prepared for the Colorado Division of Wildlife. Nudds, T. D. 1977. Quantifying the vegetative structure of wildlife cover. Wildlife Society Bulletin 5:113-117. Olson, T. E., and F. L. Knopf. 1988. Patterns of relative diversity within riparian small mammal communities, Platte River watershed, Colorado. Pg. 379-386 in: Proceedings of the symposium: Management of amphibians, reptiles and small mammals in North America. Flagstaff, Arizona U. S. Forest Service, General Technical Report RM-166. PTI Environmental Services. 1996a Preble's Meadow Jumping Mouse Study at Rocky Flats Environmental Technology Site, Annual Report 1996. Final. Rocky Flats Environmental Technology Site, Golden, Colorado. Quimby, D. C. 1951. The life history and ecology of the jumping mouse, Zapus hudsonius. Ecological Monographs 21:61-95. Riggs, L. A., J. M. Dempey, and C. Orrego. 1997. Evaluating distinctness and evolutionary significance of Preble's meadow jumping mouse: Phylogeography of mitochondrial DNA noncoding region variation Final Report for the Colorado Division of Wildlife. Denver, Colorado. Ryon, T. R 1996. Evaluation of historical capture sites of the Preble's meadow jumping mouse in Colorado, final report. MS Thesis. University of Denver, Denver, Colorado. Stohlgren,T. J., M. B. Falkner, and L. D. Schell. 1995. A Modified-Whittaker nested vegetation sampling method. Vegetatio 117: 113-121. USFWS. 1997a Proposal to list the Preble's meadow jumping mouse as an endangered species. USFWS 50 CFR part 17. �98 USFWS. 1997b. Interim survey guidelines for Preble's meadow jumping mouse. USFWS. Denver, Colorado. Whitaker, J. 0., Jr. 1963. A study of the meadow jumping mouse, Zapus hudsonius (Zimmerman), in cental New York. Ecological Monographs 33:3. Whitaker, J. 0., Jr. 1972. Zapus hudsonius. Mammalian Species 11:1-7. �0 Entcrcu 0 Checked SMALL MAMMAL TRAPPING FORM DATE_ _ _/_· _J___ _ SITEI.D(DRAINAGE). _ _ _ _ _ _ _ _ _ _ _ TOPOQUAD. NAME _ _ _ _ _ _ _ _ _ _UlME _ _ _ _ _ _ _ _ _ _ _UTMN_ _ _ _ _ _ _ _ _ _ _ PAGE HANDLER_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ TRAP TYPE _ _ _ _ _ _ _ _ _ _ _ __ TEMPERATURE _ _ _ __ OBS NUM Tnp NO. OF TRAPS _ _ _ _ _ _ SAMPLE SITE _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ TIME START _ _ _ _ TIMEFINISH _ _ _ __ CLOUD COVER (0-8) _ __ WIND _ _ _ _ _ _ __ COMMONNAME OF RECORDER._ _ _ _ _ _ _ _ _ _ _ _ _ __ [NJ/ TOTAL BAG MORT WT(I) WT(g). TAIL LENGTII B°"1 Length HIND FOOT (111111) (mm) (mm) PRECIPITATION _ _ _ _ _ _ _ _ _ _ __ EAR (mm) SEX REP. #Ear COND Plugs PITT1g# RADIO PJM SIie FREQ Phoeo Pbollo COMMENTS ~~ &~ "r1Z 0 t:, ~ ~ > '°ID COMMENTS:_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ �100 Zapus hudsonius preblei Injury/Mortality Documentation Found dead Found severely injured, euthanized Slightly injured, returned to wild Died during handling Date/Time:------------------------------Location: _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ Weather Conditions: - - - - - - - - - - - - - - - - - - - - - - - - - - Approximate Time Trap Set:._ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ Time Trap Checked:._ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ Field Technician(s) Present: Information: Species: _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ PIT TAG Number: - - - - - - - - - - - - - - - - - - - - - - - - - - Weight (g): - - - - - - - - - - - - - - - - - - - - - - - - - Total Body Length (mm): - - - - - - - - - - - - - - - - - - - - - - Tail Length ( m m ) : - - - - - - - - - - - - - - - - - - - - - - - - - Hindfoot Length (mm): - - - - - - - - - - - - - - - - - - - - - - Ear Length ( m m ) : - - - - - - - - - - - - - - - - - - - - - - - - Sex: _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ Reproductive Condition(s): - - - - - - - - - - - - - - - - - - - - - - - - Description of Injury: - - - - - - - - - - - - - - - - - - - - - - - - - - Details of Probable Reasons for Injury or Mortality: Signature of Technician(s): �1 Colorado Division of Wildlife Wildlife Research Report July 1999 JOB PROGRESS REPORT State of ----..-=..a---=="'------Colorado P roj ect No. _ _ _W..:..:--=15:;..;:3;_-R=:.;::...;-1=2'---_ _ __ Work Package No. _0=6=6=2'--------Task No. _ _ _ _-=-1_ _ _ _ _ _ __ Cost Center 3430 Mammals Program Preble's Meadow Jumping Mouse Conservation Develop Conservation Plan for Preble's Meadow Jumping Mouse Period Covered: July 1, 1998 - June 30, 1999 Author: Tanya M. Shenk ·Personnel: J. Brinker, J. Eussen, M. Miller, M. Sivert, M. Wild, CDOW, K Burnham, G. White, CSU ABSTRACT Preliminary analysis of the movement data collected on radio-collared PMJM indicate (1) maximum movements of> 1 mile, (2) greater use of upland habitats than previously assumed, (3) general site fidelity to both daytime nesting sites and nighttime feeding sites, (4) seasonal shifts in movement patterns, (5) use of both perennial and intermittent tributaries adjacent to the capture drainage, and (6) long distance movements in September to new locations, presumably in preparation for hibernation. Mouse locations were categorized by the distance criteria used in the USFWS Proposed 4-D Rule to provide the greatest amount of information to evaluate if the proposed regulations would be sufficient to ensure protection of PMJM. Given the high proportion of locations within both the proposed buffer zone (150-300 feet) and outside any area of protection(> 300 feet), data from this study do not support the regulations regarding the 300 foot protection zone from the center of the capture stream. Analyses were not completed using the criteria of protection up to 300 feet from the edge of contiguous wetlands because a delineation of contiguous wetlands for any of the three study sites is not yet available. Once these contiguous wetlands are mapped, similar analyses will be conducted using edge of the wetland rather than stream center for perpendicular distance measurements. Eight possible hibemacula were located during this study. Five of the eight mice using these possible hibemacula traveled ;;,: 90 meters from the center of their typical September night time locations. Vegetation characteristics of the eight sites located during this study are similar to other hibemacula described for PMJM. Natural mortality factors documented during this study included predation by house cats, garter snakes, rattlesnakes, and fox as well as accidents by drowning and road kill. High percentages of arthropods and endogenous fungus were found in the 85 PMJM fecal samples analyzed thus far. The combination of shifts in both general mouse movements, individual mouse movements, and diet provide strong circumstantial evidence that PMJM may be selecting for or require specific seasonal diets. A total of39 sites were surveyed for PMJM in Larimer and Weld Counties, Colorado with a total of 16,671 trap-nights completed. Of the 39 sites survey, Zapus spp. were captured at 21 �2 locations. A total of 71 unique individuals were captured in the 21 locations, with recaptures of 14 of those individuals. All sites where Zapus spp. were captured were located in Larimer County. lbree soil samples from 0-12 inches and 12-24 inches were collected at all sites where possible. Soils at some sites were too rocky to collect the 12-24 inch soil samples. Nine fecal samples were collected from traps where Zapus spp. were captured. Initial result indicate that arthropods, fungus, and willows are important dietary components of the Zapus diet. Measurements of habitat characteristics at both the site and landscape levels were completed. Genetic tissue samples were collected from 68 of the 71 jumping mice captured. Analyses of these tissue samples is pending. A total of 186 individual mice were captured and PIT-tagged at three study sites (73 mice at Maytag Property, 77 mice at PineCliffRanch, and 36 mice at Woodhouse Ranch) during the 1998 field season. These mice were captured over three different trapping sessions with an effort of 17,330 trapnights. Genetic tissue samples were collect!;ld for at least 30 individual mice at each of the three study sites. Density estimates were expected to increase as the summer progressed to account for the birth pulses in late June, and late July-August. This was the trend observed at Maytag. PMJM densities at Woodhouse Ranch increased from June to July but did not increase further during the September trapping session. Mean PMJM density over all sites and sessions was 40.5 (range 16.9-79.0) mice per . kilometer of stream stretch. Defining summer as June 1 - October 5, over-summer survival was estimated as 0.36 (se = 0.056) over all three study sites. Temporary emigration and immigration rates were estimated but both had extremely high variances associated with those estimates. Thus, these estimates cannot be used, with any confidence, to provide information on movements of mice into and out of these populations. Very little new information was gained during this study on reproductive parameters of PMJM. Most adult mice captured exhibited evidence of active reproductive behavior, either pregnancy, lactation, or enlarged genitalia However, we were unable to loca~e any breeding nests. Juvenile mice were not captured during the June trapping session. Juvenile mice were captured at all three sites during both the July and September trapping sessions. Preliminary results from the demography, movement and distribution studies of PMJM have suggested the need to continue with research currently being addressed in these three studies. However, the preliminary results also suggest further research be conducted on (1) use of upland habitat by PMJM, (2) refining water requirements of PMJM (i.e., do they require stream habitat or are wetland areas sufficient), (3) range-wide distributional boundaries (e.g., elevation restrictions), and (4) investigating areas of potential sympatry or hibridization with Z. princeps (western jumping mouse). The demography and movement studies will be modified during the summer 1999 field season to further investigate upland use and water requirements. This report includes results from only the first year of a multi-year project to follow individually marked PMJM through time. Collection of more data, and more years of data, will improve our ability to evaluate demographic parameters, movement patterns, and habitat use and evaluate how they vary across space and time. �3 PREBLES'S MEADOW JUMPING MOUSE CONSERVATION Tanya M. Shenk P.N. OBJECTIVE Develop and implement a conservation plan for Preble's meadow jumping mouse in Colorado. SEGMENT OBJECTIVES 1. Evaluate movement data of Preble's meadow jumping mouse as it relates to landscape features at two study sites. 2. Conduct presence/absence surveys for Preble's meadow jumping mouse at 30 sites in Larimer and Weld Counties, Colorado. 3. Refine and prioritize needed research components to develop sound strategies for conservation of Preble's meadow jumping mouse. 4. Analyze data and prepare a Federal Aid Job Progress Report. Status Report I, Evaluate movement data ofPreble's meadow jumping mouse as it relates to landscape features at two study sites. Movement data of PMJM were collected on radio-collared mice to evaluate the effects of landscape features on movement patterns. Preliminary analysis of the movement data is presented as a draft manuscript, Movement patterns ofPreble's meadow jumping mouse (Zapus hudsonius preblei) as they vary across time and space, in Appendix A In summary, movement data collected on radio-collared PMJM indicate (I) maximum movements of> I mile, (2) greater use of upland habitats than previously assumed, (3) general site fidelity to both daytime nesting sites and nighttime feeding sites, (4) seasonal shifts in movement patterns, (5) use of both perennial and intermittent tributaries adjacent to the capture drainage, and (6) long distance movements in September to new locations, presumably in preparation for hibernation. 2. Conduct presence/absence surveys for Preble's meadow jumping mouse at 30 sites in Larimer and Weld Counties, Colorado. a. Presence/absence surveys Preliminary analysis of the PMJM survey data is presented as a draft manuscript, Habitat use and distribution of Preble's meadow jumping mouse (Zapus hudsonius preblei) in Larimer and Weld Counties, Colorado, in Appendix B. In summary, a total of39 sites were surveyed for PMJM in Larimer and Weld Counties, Colorado with a total of 16,671 trap-nights completed. Of the 39 sites survey, Zapus spp. were �4 captured at 21 locations. A total of 71 unique individuals were captured in the 21 locations, with recaptures of 14 of those individuals. All sites where Zapus spp. were captured were located in Larimer County. Three soil samples from 0-12 inches and 12-24 inches were collected at all sites where possible. Soils at some sites were too rocky to collect the 12-24 inch soil samples. Nine fecal samples were collected from traps where Zapus spp. were captu{ed. Initial result indicate that arthropods, fungus, and willows are important dietary components of the Zapus diet. Measurements of habitat characteristics at both the site and landscape levels were completed. Genetic tissue samples were collected from 68 of the 71 jumping mice captured. b. Abundance, over-summer survival, over-winter, annual survival estimates and population age and sex structure ofPreble 's meadow jumping mouse populations at two study sites. Preliminary analysis of the demography study ofPMJM is presented as a draft manuscript, Temporal and spatial variation in the demography ofPreble's meadow jumping mouse (Zapus hudsonius preblei), in Appendix C. In summary, a total of 186 individual mice were captured and PIT-tagged at three study sites. Mean PMJM density over all sites and sessions was 40.5 (range 16.9-79.0) mice per kilometer of stream stretch. Defining summer as June 1 - October 5, over-summer survival was estimated as 0.36 (se = 0.056) over all three study sites. Temporary emigration and immigration rates were estimated but both had extremely high variances associated with those estimates. Thus, these estimates cannot be used, with any confidence, to provide information on movements of mice into and out of these populations. Very little new information was gained during this study on reproductive parameters of PMJM. Most adult mice captured exhibited evidence of active reproductive behavior, either pregnancy, lactation, or enlarged genitalia However, we were unable to locate any breeding nests. Juvenile mice were not captured during the June trapping session. Juvenile mice were captured at all three sites during both the July and September trapping sessions, indicating two birth pulses. 4. Refine and prioritize needed research components to develop sound strategies for conservation ofPreble's meadow jumping mouse. Preliminary results from the demography, movement and distribution studies of PMJM have suggested the need to continue with research currently being addressed in these three studies. However, the preliminary results also suggest further research be conducted on (1) use of upland habitat by PMJM, (2) refining water requirements of PMJM (i.e., do they require stream habitat or are wetland areas sufficient), (3) range-wide distributional boundaries (e.g., elevation restrictions), and (4) investigating areas of potential sympatry or hibridization with Z. princeps (western jumping mouse). The demography and movement studies will be modified during the summer 1999 field season to further investigate upland use and water requirements. 5. Analyze data and prepare a Federal Aid Job Progress Report. A Federal Aid Job Progress Report was submitted to the CDOW on August 16, 1999. �5 APPENDIX A MOVEMENT PATTERNS OF PREBLE'S MEADOW JUMPING MOUSE (Zapus hudsonius preblei) As THEY VARY ACROSS TIME AND SPACE Tanya Shenk and Maile M. Sivert Abstract Preliminary analysis of the movement data collected on radio-collared Preble's meadow jumping mice (PMJM) indicate (I) maximum movements of> I mile, (2) greater use of upland habitats than previously assumed, (3) general site fidelity to both daytime nesting sites and nighttime feeding sites, (4) seasonal shifts in movement patterns, (5) use of both perennial and intermittent tributaries adjacent to the capture drainage, and (6) long distance movements in September to new locations, presumably in preparation for hibernation. Mouse locations were categorized by the distance criteria used in the USFWS Proposed 4-D Rule to provide the greatest amount of information to evaluate if the proposed regulations would be sufficient to ensure protection of PMJM. Given the high proportion oflocations within both the proposed buffer zone (150-300 feet) and outside any area of protection(> 300 feet), data from this study do not support the regulations regarding the 300 foot protection zone from the center of the capture stream. Analyses were not completed using the criteria of protection up to 300 feet from the edge of contiguous wetlands because a delineation of contiguous wetlands for any of the three study sites is not yet available. Once these contiguous wetlands are mapped, similar analyses will be conducted using edge of the wetland rather than stream center for perpendicular distance measurements. Eight possible hibernacula were located during this study. Five of the eight mice using these possible hibernacula traveled :?: 90 meters from the center of their typical September night time locations. Vegetation characteristics of the eight sites located during this study are similar to other hibernacula described for PMJM. Natural mortality factors documented during this study included predation by house cats, garter snakes, rattlesnakes, and fox as well as accidents by drowning and road kill. High percentages of arthropods and endogenous fungus were found in the 85 PMJM fecal samples analyzed thus far. The combination of shifts in both general mouse movements, individual mouse movements, and diet provide strong circumstantial evidence that PMJM may be selecting for or require specific seasonal diets. These data are limited to the first year of a multi-year study and thus annual variation in mouse movement patterns cannot yet be addressed. Continuation of this study, at least through the 1999 field season will provide information on annual variation in PMJM movement patterns. This is an interim progress report. Further analyses will be conducted on these data and data collected during the 1999 field season. llltroduction Movement and dispersal pattern information will be key to any conservation strategy designed for Preble's meadow jumping mouse (PMJM). Documenting daily and seasonal movement patterns of PMJM will provide information on habitats used by the mice and on the relative configuration and juxtaposition of these habitats. Configuration of habitats include vegetation and size of the areas used. Juxtaposition includes relative locations of different habitats used such as nest sites, feeding areas, and movement corridors connecting those areas. Key dispersal factors to document for PMJM include (I) which segment of the population disperses, (2) when do they disperse, (3) through what habitat do they disperse, (4) how far will individuals disperse (i.e., what is the maximum distance that separates adjacent populations) and (5) how critical is dispersal (both into and out of a population) to the persistence of a given population. �6 Areas of suitable habitat must provide requirements to survive throughout the life cycle. These requirements must provide necessities for both the active period and hibernation periods. During the active period suitable habitat must provide requirements for daily survival, reproductive activities (breeding, nesting, and rearing of young to independence), and dispersal. The hibernation period requires sufficient food supplies to assure fat storage prior to hib~mation and suitable hibernacula Habitat providing all seasonal and life cycle requirements may or may not occur in a single contiguous area If not in a contiguous area, habitat patches must occur in a mosaic of usable areas where suitable -corridors exist for seasonal movement among sites. The habitat matrix within the range of Z. h. preblei is mixed grasslands adjacent to the Colorado Front Range along the Piedmont and along the base of the Laramie Mountains in Wyoming and extends to the Colorado plains. Within this matrix, PMJM occur along stream drainages that contain patches of suitable vegetation. Suitable habitat appears to have at least two major components. The first component is a supply of open water, at least in part of the active season (M. Bakeman, C. Meaney, personal communication). Secondly, areas where PMJM has been found have dense cover (M. Bakeman, C. Meaney, personal communication). Based on studies of Z. h. preb/ei and Z. hudsonius elsewhere, Z. h. preblei apparently occurs mostly in undergrowth consisting of grasses, forbs, or both in open wet meadows and riparian corridors, or where tall shrubs and low trees form an overstory and provide adequate cover (Armstrong et al. 1997). Meadow jumping mice are widespread in abandoned grassy fields, but are often more abundant in thick vegetation along ponds, streams, and marshes or in rank herbaceous vegetation of wooded areas (Whitaker 1963). The mouse does not appear to have an affinity toward any single plant species but instead favors sites that are structurally diverse and provide adequate cover and food throughout its life cycle. PMJM are typically not found in upland areas away from riparian habitats but are most often captured where either ground water becomes visible as either seep springs or as main water channels (M. Bakeman,T. Ryon, personal communication) suggesting a dependence on open water, at least during their active periods. PMJM have been trapped in natural riparian areas as well as areas altered by anthropogenic influence including ditches and wetlands adjacent to interstate highways, cement-lined ditches with tall cover, ditches along driveways and moderate road use, and moderate cattle grazing (M. Bakeman, personal communication). The only currently available data on dispersal and/or movement for Z. h. preblei are from marked mice at Rocky Flats Environmental Technology Site (T. Ryon unpublished data). Two mice, an adult female and an adult male, were observed approximately 1.6 kilometers from previous locations (incidences occurred separately). Each of the locations were in the same drainage (Woman Creek). Movement of Z. h. preblei on Woman Creek at Rocky Flats Technology Site suggests mice move along corridors of shrub cover, generally Salix exigua (PTI 1996a, T. Ryon, unpublished data), suggesting dispersal habitat is similar to habitats used for other activities. If PMJM occurs as metapopulations in the classical sense of a set oflocal populations linked by infrequent dispersal then habitat includes not just one area of suitable habitation but also areas suitable for nearby mouse populations. These suitable areas must also be linked by dispersal habitat. If the mice are dependent on dense riparian habitat for dispersal as well as for areas to reproduce, persistence of discrete populations would require a mosaic of suitable discrete riparian patches interconnected with dispersal corridors of similarly dense riparian vegetation. If mouse populations function in a source (populations where growth rate :2: 1) and sink (those populations where growth rate< 1, maintained through immigration) system, it will be critical to identify and protect those populations serving as sources. Thus, for a source-sink population critical habitat will include those areas that support source population, dispersal habitat to sink areas will be less critical. If local mouse populations are functionally discrete, such a mosaic of interconnected areas of suitable habitat would provide a buffer �7 for local, and source, populations against deleterious stochastic events by providing the opportunity for local population failures to be 'rescued' by immigration from other populations. To begin to understand PMJM movement, dispersal, and habitat use we monitored radiocollared mice at three different study areas. Study areas were selected to cover a variety of habitat configurations to evaluate spatial variation in movement patterns of PMJM. To address temporal variation in PMJM movement, dispersal, and habitat use we will' continue this study for at least one more field season at the same three field sites. This is an interim report, after one field season of data collection. Study sites To evaluate spatial differences in movements of PMJM, the three sites selected provided a variety of habitat matrices available to the mouse. The first site selected, the Maytag Property, has one primary water source available to the mouse. This water source is East Plum Creek. Therefore, we predicted mice would restrict their movements to up and down this single drainage. The second site, PineCliffRanch, has both a tributary (Garber Creek) and a main stem drainage (West Plum Creek). This provided the opportunity to investigate whether PMJM will use upland areas to move from one drainage to another or if they are restricted to only moving along riparian corridors. The third study site, Woodhouse Ranch, provided an area with a tributary (Indian Creek) and a series of ponds and irrigation ditches within 0.5 kilometers of the drainage. The configuration at Woodhouse Ranch provided an even greater opportunity to investigate how much the mice use upland areas or if they restrict their movements strictly to riparian corridors. By replicating the same methodologies at each of these three unique habitat matrices we can begin to estimate spatial variation in nightly and seasonal movements of PMJM. Objectives This PMJM movement study is designed to describe nightly and seasonal movement patterns of PMJM and to describe habitats used by PMJM. These movement patterns will be described for three different study areas to evaluate spatial variation, and over two different years at the same three study areas to evaluate temporal variation. Specific objectives include: 1. Describe nightly movements of PMJM. Evaluate difference in nightly movements as they relate to sex, age, and habitat available to the animals. I. Describe 30-day (or life of the radio transmitter) interval movements of PMJM. Evaluate difference in 30-day (or life of the radio transmitter) interval movements as they relate to sex, age, and habitat available to the animals. 2. Describe seasonal movements of PMJM. Evaluate difference in seasonal movements as they relate to sex, age, and habitat available to the animals. 3. Describe habitats where mice occur: movement corridors, end point descriptions (i.e., movement from what to what), and landscape features (connectivity with other riparian areas and other habitats used). 4. Estimate the mean amount of time PMJM spend in each available habitat. 5. Evaluate spatial and temporal variation in movement patterns of PMJM. The objective of the habitat use study is to identify and refine habitat requirements of Z. h. preblei, including hibernation sites, and to determine if they influenced any of the demographic parameters that will be estimated in this and complementary studies (see Shenk and Sivert 1999). �8 Metl,ods Movement data were documented primarily from locations of radio-collared mice from each of the three study populations. To place radio transmitters on the animals, mice were captured in Sherman live traps. Mice weighing> 18 grams were fitted with either :MD-2C, I-gram radio transmitters supplied by Holohill Systems Ltd. (used successfully on PMJM by R Schorr, personal communication) or I-gram radio transmitters supplied by Advanced Telemetry Systems (ATS). Three trapping sessions were conducted during the following weeks: June 2-9, 1998; July 21Aug 6, 1998; and September 8-15, 1998. Trappers were advised to follow the Center for Disease Control's Hantavirus instructions and recommendations when dealing with rodents. Due to the nocturnal nature of PMJM, traps were set between 19:00hrs and 21:00hrs and checked as early as possible in the morning beginning at 5:00hrs (to reduce stress and the potential for predation on trapped animals). Time required to complete the traplines varied depending on how many animals were caught. Small mammal Sherman live traps (folding and non-folding) were used to conduct the trapping sessions. Traps were set in two parallel lines of trap stations (1 trap per station) on either side of the drainage. Trap stations were 5 meters apart for a total of 250 meters; the parallel transects were 10 meters apart unless extent of habitat, terrain topography, or stream hydrology do not allow. Location of transects were recorded to the nearest 5 meters in UTM coordinates using a Trimble Geo-Explorer GPS. A small (~l inch) ball of polyester quilt material was placed in each trap as nesting/bedding material. Baiting material was Manna Pro Sweet 3-way Livestock feed which contains no animal matter; ingredients include flaked barley, flaked com, flaked oats, and cane molasses. Peanut butter was used to stick the bait to the trap. Traps were checked by two surveyors. All animal captures were recorded. If an animal was captured in a trap, a ziplock plastic bag was placed over the end of the trap. The trap was opened allowing the animal to fall into the plastic bag. The animal was identified to species while in the plastic bag. If the animal was not a PMJM, identification of the animal was recorded and the animal set free. Each captured PMJM was checked to detect the presence/absence of a PIT-tag and/or radio-collar. If a PIT-tag or radio-collar was detected and the mouse had been captured at that same site within the same trapping session, identification of the mouse was recorded and the mouse was released. If the animal was a PMJM and no PIT-tag or radio collar was detected the PMJM was anesthetized for further processing, as follows. All PMJM were PIT-tagged for a complementary demography study on PMJM at these same three study sites (see Shenk and Sivert 1999 for details). If the PMJM weighed> 18 g a radio-collar was also put on the animal. Each PMJM was weighed while in the bag, the weight recorded in grams using a Pesola spring balance. Sex, age (juvenile, adult), and reproductive condition (pregnant, lactating or non-breeding if it is female; for males the position of the testes indicated if in breeding status or non-breeding status) was noted. Each PMJM w~ measured (total length, length of body, length of hind foot-heel to distal end of claws) in millimeters. Mice were placed in a plexiglass photobox to reduce handling stress while taking the photograph. If the mouse had been captured previously as noted by the detection of a PITtag, no photograph was taken. If there were feces in the trap where a PMJM was captured, the fecal material was collected in a plastic bag labeled with date, location of the site, and animal PIT-tag number. Fecal samples were kept cool until returned to the CDOW office where they were frozen. Salt was added to each specimen bag for preservation. Once the field season was over all fecal samples were analyzed by the Composition Analysis Laboratory, Inc, 622 ½ Whedbee, Fort Collins, CO for content. All trap mortalities were recorded on a 'Trap Mortality' data form which included information on species, potential duration of time spent in the trap, and any information available to help determine cause of death. All animals found dead were double-bagged in a plastic bag and placed in a cooler with ice. Specimens were frozen as soon as possible and deposited in the CDOW freezer at the office . ·. . : �9 in Fort Collins. A museum card was completed and attached to each PMJM specimen and the specimen and card were given to Cheri Jones, Curator of Mammalogy at the Denver Museum of Natural History, for study skins and tissue storage. If an animal was severely injured (e.g., severed limb, large lacerations) it was euthanized by soaking a cotton ball in Metofane (methoxyflurane) and placing the cotton ball and the mouse in a ziplock bag until the animal stopped breathing. If an animal appeared to be only slightly injured (e.g., broken tail, small laceration) the animal was released to the wild. If an animal appeared to be coldstressed attempts were made to warm it by holding it in the surveyors hands and/or against their body. If the animal appeared to be heat stressed, isopropyl alcohol was applied with a cotton swab to the ears, arm pits, and feet to cool it down. During the last trapping session homeopathic first aid treatments were used in an attempt to minimize stress and improve recovery of each PMJM. Homeopathic first aid kits and instruction was provided by WildAgain Wildlife Rehabilitation, Evergreen, Colorado. PIT-tagging and radio-collaring Each PMJM was individually marked with a Passive Integrated Transponder (PIT-tag). PITtags are electromagnetic, glass-encased tags that emit a passive signal (125 kHz) that can be decoded by a portable reader. Destron-Fearing PIT-tags were used on all mice. We used portable readers from Biomark to read the PIT-tags. All mice were anesthetized to PIT-tag them. Anesthesia procedures followed protocols successfully used on PMJM before (R. Schorr personal communication). Mice were anesthetized by placing them and a cotton ball with 1 ml ofMetofane (methoxyflurane) into a sealed ziplock bag (to keep Metofane fumes in bag). The bag was then lain aside to minimize stress for the mouse and the time the mouse struggles in the bag. The less agitated the mouse is in the bag the less time required for the metofane to take affect. The mouse was observed at all times to make sure the mouse did not position itself such that the cotton ball was in direct contact with its face, which might cause the mouse to have a severe reaction to the anesthesia. After the animal stopped moving, the handlers waited one minute before removing the mouse from the bag. The animal typically remained anesthetized for 2-3 minutes once outside the bag. Each PMJM was PIT-tagged using a protocol successfully used by C. Meaney (personal communication). Each PMJM had a PIT-tag inserted above the shoulder blades by lifting the skin on its back and inserting the individually sterilized needle with the PIT-tag under their skin and injecting the tag. Verification of the PIT-tag identification number was made before and after insertion into the mouse by running the PIT-tag scanner across it. The skin behind the opening was then pinched to prevent emergence of the tag. PIT-tag identification number was recorded on the field data form. Each radio transmitter was checked to ensure proper functioning before collaring the animal with the transmitter by tuning the receiver to the transmitter frequency to make sure there was a signal. The radio antenna was fed through a small piece of plastic tubing and then back through the radio to make a collar of the plastic coated antenna. The loop made by the antenna was kept large enough to fit over the mouse's head. Once over the neck of the animal, the antenna was pulled to reduce collar size, making sure rubber tubing was on the dorsal side of the neck. The purpose of the rubber tubing was to prevent damage to the collar and prevent the person attaching the collar from cinching the collar too tight. The collar was tested to be sure it could rotate freely around the animals' neck, but was not loose enough to be removed. The metal crimp was flattened to hold the antenna in the desired collar position. Approximately two inches of antenna was left to extend out behind the animal. If, at any time during the handling of the PMJM the animal appeared to be severely stressed (dramatic changes in heartbeat, respiration and responsiveness or gums turning blue) first aid was administered and, once recovered, released without further processing. �IO Movement data Once transmitters were in place, locations of individual mice were made nightly. Because very little is known about the movements of PMJM pilot protocols were established and then modified. Final protocols included tracking up to eight individual mice per person per night. A minimum of six locations per individual mouse per night were made. Each individual mouse was tracked a minimum of three nights per week. The six locations per night per individual were scattered throughout the night to maximize information learned about nightly movements (i.e., eight locations taken within a single hour contains less information than eight locations taken one each hour). Each location was determined by locating the mouse and recording the position as a file in a Trimble Geo-Explorer GPS. Each GPS location (file) was differentially corrected for 2-5 meter accuracy. Daytime locations were taken as time permitted. To describe nightly movements the following parameters were estimated: (I) minimum and maximum distances moved each night animal was followed, (2) mean minimum and maximum distances moved each night for all mice, (3) minimum and maximum distances moved away from the stream center each night by individual mice, and (4) mean minimum and maximum distances moved away from the stream center each night for all mice. To describe 30-day (or life of the radio transmitter) movements the following parameters were estimated:(!) minimum and maximum distances moved each 30-day (or life of the radio transmitter) interval animal was followed, (2) mean minimum and maximum distances moved each 30-day (or life of the radio transmitter) for all mice, (3) minimum and maximum distances moved away from the stream center each 30-day (or life of the radio transmitter) interval by individual mice, (4) mean minimum and maximum distances moved away from the stream center each 30-day (or life of the radio transmitter) interval for all mice. Mortality factors for any mouse found dead was recorded Both site specific and landscape features were recorded at each of the three population study sites to document the extent of spatial variation. These quantitative measures of spatial variation were used in the analyses to determine if they influenced any of the movement parameters estimated in this study. The following habitat characteristics were measured and recorded for each site. ~ Site characteristics Cover was measured at 20 random locations along the 250 meter sampled stream stretch. Mean cover and standard error of the mean was estimated. Cover was estimated using a vegetation profile board (Nudds 1977) that allowed for an assessment of visual obstruction in 0.5 meter vertical intervals above ground. The board was 2.0 meters high and 30.48 cm wide. The board was marked in alternate black and white colors at 0.5 meter intervals. Horizontal cover was assessed in each interval by viewing the board from 15 meters away in a randomly chosen direction. The percentage of each interval concealed by vegetation was recorded as 0-20, 21-40, 41-60, 61-80, and 81-100% estimated concealment. Vegetation species composition and richness was estimated using the Modified-Whittaker nested vegetation sampling method (Stohlgren et al. 1995). A modified Whittaker plot is 20-meters x 50-meters with 10 I m2 and 2 10m2 subplots arranged systematically around the perimeter and 1 100m2 subplot centered in the inside of the 20 meter by 50 meter plot. Species composition and percent cover of each species was recorded for each subplot. Soil samples were taken at each site for a hydrometer textural analysis (percent sand, silt and clay). Samples were collected using a soil probe. Samples were taken from 0-12 inches, and 12+ to 24 inches. The 0-12 inch sample was taken first, removing the probe and soil. Samples were placed in a labeled ziplock bag. The probe was then placed in the same hole for the 12+ - 24 inch probe with that soil sample being placed in a separately labeled ziplock bag. The hydrometer textural analysis was conducted at the Colorado State University Soils Laboratory. �11 Mean stream width was estimated by measuring stream width at 30 locations along the entire site sampled. The 30 locations were stratified equidistant from each other to cover the entire stream stretch in increments of site stream length/30. Each nest site was described including distance from the stream and vegetation. Landscape characteristics The following landscape habitat characteristics were measured using either GIS, topographic maps, or aerial infrared photography. 1. Distance to nearest human habitation and human disturbance. 2. Connectivity of sites to other streams. 3. Total length of stream stretch at each site and total length of stream stretch with suitable habitat at each site. Hibernation site measurements Hibernation sites were identified by following radio-collared mice in late September and early October. Once a radio-signal remained stationary for 2-10 nights we assumed we located a hibernaculum. The following measurements were taken at each hibernation site: 1. 2. 3. 4. Distance ofhibernaculum to stream. Vertical elevation gain from stream to hibernaculum. A soil sample was taken as described above. Vegetation species richness within a 5-m radius. Density of vegetation was measured as described above. Results PIT-tagging and radio-tracking effort A total of 186 individual mice were captured and PIT-tagged at the three study sites (Table 1: 73 mice at Maytag Property, 77 mice at PineCliffRanch, and 36 mice at Woodhouse Ranch). These mice were captured over three different trapping session with an effort of 17,330 trap-nights. A total of 125 radio-collars were put on mice over the three trapping sessions (Table 1: 48 at Maytag property, 4 7 at PineCliff Ranch, and 3 0 at Woodhouse Ranch). A total of 62 females and 63 males were radiocollared over the field season. Initial protocols called for using triangulation methods (simultaneous compass bearing readings from three different locations) to determine mouse locations. Triangulation was used during the month of June. Locations were estimated in late June from these data and found to have large variances associated with most locations. Field protocols were changed for the latter two tracking sessions to walk-up locations. A walk-up location required technicians to use the signal strength indicator on the receiver to determine when they were within 2-7 meters of the mouse. GPS location files were collected each time a mouse was located. File name was recorded as well as the UTM coordinates displayed on the GPS. These files were later differentially corrected to provide a positional accuracy of 2-5 meters. Due to the large variances on the mouse locations collected during June these locations are not reported here. Further analyses will be conducted on these data to see if any information on PMJM movements can be gained. The walk-up method was first evaluated to ensure we were not causing the mice to move in response to us. Given the consistency of individual mouse locations throughout the night and on a day to day basis we felt the method was being employed in such a way that we were not affecting mouse movement patterns. However, it should be noted that T. Ryon (personal communication) observed mice moving in response to him when he tried to conduct similar walk-up locations. �12 Daily movements Percent of PMJM locations, as determined from radio-tracking, at each of three distance categories were calculated for each study site and the latter two radio-tracking sessions (Table 2). The majority of locations at Maytag, PineCliff, and Woodhouse were within 46m of stream center for both tracking periods (July-August and September-October). For all areas and tracking periods except Maytag during the last tracking session, 90% of PMJM movements were within 91m of either side of the center of the capture drainage. Mouse movements at Maytag during September-October resulted in 32.6% of the locations being> 91m from the stream center. Analyses were also conducted on mice captured on PineCliff Ranch separating mice by drainage of initial capture. These analyses were conducted to address questions that might arise when connecting drainages. occur in areas known to have PMJM, but only one drainage has been surveyed. Mice initially captured on Garber Creek showed similar movement patterns as far as distance from center of capture drainage (i.e., Garber Creek) as mice from both Maytag and Woodhouse (Table 2). Mice initially captured on West Plum Creek showed movements more often> 91m from center of the capture drainage (Table 2). Seasonal movement The distribution of mouse locations at Maytag Property for the July-August tracking session is different from the distribution of PMJM locations for the September-October tracking session. The pattern shifts from heavy use along East Plum Creek to the north of the trapline to a more concentrated distribution either side of the trap line. The more concentrated use of the areas east and west of the trap line also resulted in locations being further from the center of the creek. The northern end of the trap line is largely vegetated by willows with the remainder of the trap line more diversely vegetated. Mouse movements were more concentrated along the drainages at PineCliff Ranch during the September-October tracking sessions than during the July-August tracking sessions. Vegetation along both West Plum Creek and Garber Creek along the trapline was primarily willow. At Woodhouse Ranch, the distribution of mouse locations also shifted between tracking sessions. Mouse movements were more concentrated along the southern half of the trap line in September-October as compared to the more even distribution of mouse locations recorded for JulyAugust The northern end of the trap line is dominated by willows, the middle and southern end of the trapline is in more diverse riparian habitat. Eight mice were captured, radio-collared, and tracked for two tracking sessions, one mouse was captured, radio-collared, and tracked for all three tracking sessions (Table 3). Other mice were captured during more than one trapping session but radio collars were not replaced either because all the radio collars had been put on other mice or because the physical condition of the animal was such that we decided not to replace the radio-collar. Such conditions included mice who had significant hair worn off the neck from the previous radio collar. In no incidence did we find open wounds caused by radio collars. Of the nine mice radio-collared for more than one session, four provided information to evaluate seasonal changes in areas and habitats use between the July-August tracking period and the September-October tracking period. One mouse, a male from Maytag was observed only once during the September-October tracking session, providing no information concerning seasonal changes in movement patterns. Movement data from June have not been fully analyzed and are not included in this report. Thus, changes in areas used by individual mice between June and either the July-August or September-October tracking session are not summarized here. A female at Maytag exhibited a seasonal movement shift, small sample size prohibited any evaluation of the movement patterns of the Maytag male (n = l in September-October). There does not appear to be any shift in areas used by the male tracked during both latter sessions at Pine Cliff Ranch, �13 however, sample sizes were small (n = 17, 14). Two females at Woodhouse Ranch exhibited a shift in areas used between July-August and September-October. However, caution should be used in interpreting these results because of low sample sizes in the latter tracking session (n = 14, 6). Nest sites ,. Numerous daytime nest sites were located at all thee study sites. Nest sites were made of tightly woven vegetation, located on the ground. Nest material included leaves, grass, and small sticks. There was only one small entrance hole on each of the nests found. When observing mice in their nests the mice would peer out of the hole and remain motionless as long as we were present. No breeding nests were located, however there were several locations at each of the study sites where adult females returned to repeatedly. Each of these sites were in patches of extremely dense vegetation or underneath a large downed log. We did not disturb these areas because we did not want to cause failure of the nest. Hibemacula Eight potential hibernacula were located, at least one at each study site (Table 4). Mortality factors Mortalities factors of radio-collared mice included predation by rattlesnake, garter snake, fox, and house cat (Table 5). Four probable predations were also noted and identified by finding tightly crimped (i.e., no possibility of the mouse having slipped the collar over its head) radio-collars lying on the ground. Two accidental deaths were documented, a road kill and drowning. Mortality factors also included trapping and/or handling mortalities and unknown causes. Fecal analysis The amount of bait found in each of the fecal samples was quantified as either 100%, abundant (> 70%), trace (< 10%) or 0%. The following summaries are made after eliminating all samples of 100% bait, and then only classifying the fecal sample contents other than bait. Fecal analyses indicate a seasonal shift in diets. The most common item found in the fecal samples during the June trapping session at all three sites were arthropods. This was true for three of the five June samples collected at Maytag (Table 6), eight of ten June samples at PineCliff (Table 7), and four of five June samples collected at Woodhouse (Table 8). Other common items included endogenous fungus (1) and seed (1) at Maytag; endogenous fungus (2) at PineCliff,; and Poa (1) at Woodhouse. During the July 21-August 4 trapping session the most common items in the fecal samples at Maytag were arthropod (2), endogenous fungus (1), pollen (1), and Carex (1). During the same trapping period, the most common items in five of eight samples collected at PineCliff were endogenous fungus, the other three samples having a majority of arthropod (1 ), moss (1 ), and pollen (1 ). The two samples from Woodhouse during this trapping session were composed primarily of either mushroom or seed. On August 13, nine samples were collected during an extra trapping session (on the same trapline as the other trapping sessions occur) with the most common items in the fecal samples being moss (4), endogenous fungus (3) or pollen (2). All three samples collected from PMJM captured at a back drainage on August 23n1 at Woodhouse Ranch were primarily fungus. The September trapping session at Maytag yielded samples with majority fecal contents of arthropod (2), moss (2), pollen (2), endogenous fungus (1 ), and seed (1 ). Ten samples from PineCliff during September had majority fecal contents of arthropod (3), endogenous fungus (3), and seed (3). Eight of nine samples taken from Woodhouse contained primarily arthropods, with one a trace of seed. �14 Discussion Preliminary analysis of the movement data collected on radio-collared Preble's meadow jumping mice (PMJM) indicate (1) maximum movements of> 1 mile, (2) greater use of upland habitats than previously assumed, (3) general site fidelity to both daytime nesting sites and nighttime feeding sites, (4) seasonal shifts in movement patterns, (5) use of both perennial and intermittent tributaries adjacent to the capture drainage, and (6) long distance'movements in September to new locations, presumably in preparation for hibernation. Initial evaluation of PMJM movement data focused on maximum distances mice moved perpendicular to the capture drainage. This focus resulted from criteria used in the USFWS Draft 4-D Ruling (1998) to provide special regulations for the mouse. The 4-D Rule defines a Mouse Protection Area (MPA) as "extending 300 feet on each side of the stream measured from the centerline, or 300 feet from the exterior boundary of any contiguous wetlands, whichever is further." The 4-D Rule also states "the basis for the 300-foot standard is that mice have been documented to regularly move up to 150 feet from streams and wetlands. The remaining 150-foot zone serves as a buffer zone to avoid disturbance of PMJM habitat associated with human activities. We believe that this zone will encompass the normal home range of the Preble's and will provide an adequate buffer from adjoining development." Mouse locations were categorized by the distance criteria used in the 4-D Rule to provide the greatest amount of information to evaluate if the proposed regulations would be sufficient to ensure protection of PMJM. Given the high proportion oflocations within both the buffer zone (150-300 feet) and outside any area of protection(> 300 feet), data from this study do not support the regulations regarding the 300 foot buffer from the center of the capture stream. Analyses were not completed using the criteria of protection up to 300 feet from the edge of contiguous wetlands because a delineation of contiguous wetlands for any of the three study sites is not yet available. Once these contiguous wetlands are mapped, similar analyses will be conducted using edge of the wetland rather than stream center for perpendicular distance measurements. To address the issue of whether or not connected but untrapped tributaries should be included in the designation of MP A's, the more detailed analyses of mouse movements at PineCiiffRanch were conducted. The issue of whether or not Garber Creek, a tributary of West Plum Creek, would be included in that particular MP A is not at issue since both the main drainage, West Plum Creek and the tributary, Garber Creek have been trapped in that area with mice captured on both drainages. Rather, the analysis was conducted to test how well the proposed regulation to protect only 300 foot from the center of the capture stream would have 'encompassed the normal home range of the Preble's and will pro.vide an adequate buffer from adjoining development." It is clear from the results that the regulation would have failed to accomplish its objective, with up to 47% of the mouse locations falling beyond the 300 foot demarcation when only the capture drainage was considered. Use of tributaries was also an issue at the Maytag Property. Two mice moved away from East Plum Creek in September and focused their movements ~300 meters from their previous locations, up a dry drainage dominated by upland grasses and gambel oak. These two mice continued to use this new area and were not observed again near East Plum Creek for at least two weeks. Normal radiofailure (i.e., batteries failed after ~4 weeks) after this time did not allow us to determine if the mice ever returned to East Plum Creek or if they hibernated in their new location. No tributaries were available to mice at Woodhouse Ranch. In summary, movement patterns of mice at the Maytag Property suggest use of tributaries, and at PineCliffRanch extensive use of a tributary. Jumping mice of the genus Zapus are true hibernators, spending much of their lives in hibernation. Meadow jumping mice spend approximately 7 months (~210 days) per year in hibernation (Quimby 1951) whereas estimates for Z. princeps indicate that some populations (e.g., in the western mountains of Utah) spend up to 300 days per year in hibernation (Cranford 1983). �15 Jumping mice hibernate in underground burrows (Quimby 1951, Whitaker 1963). They are excellent burrowers and create their own hibernacula Meadow jumping mice are generally solitary hibernators, however, there have been occurrences of more than one mouse found in a single hibernaculum. Eight possible hibernacula were located during this study. The greatest perpendicular distance from the center of the main drainage at the site (East Plum Creek for Maytag Property, West Plum Creek for PineCliffRanch, and Indian Creek for Woodhouse Rari'ch) was 341 meters. However, if tributaries were considered, greatest distance from center of the nearest drainage to any of the eight sites. was 78 meters. Five of the eight mice using these possible hibernacula traveled ~ 90 meters from the center of their typical September night time locations. One mouse, a female at Maytag moved 750 meters to a possible hibernacula Vegetation characteristics of the eight sites located during this study (Table 4) are similar to other hibernacula described for PMJM. One confirmed hibernaculum, located on Rocky Flats Environmental Technology Site, used by Z. h. preblei has been located {Armstrong et al. 1997). This site was 9m above a creek bed (Walnut Creek); it had a thick cover of chokecherry (Prunus virginiana) and snowberry (Symphoricarpos spp.), the mouse was found in a leaf litter nest 30cm beneath the ground in coarse textured soil {Armstrong et al. 1997). Four possible hibernacula were located by tracking radio-telemetered mice at the U. S. Air Force Academy in fall 1997. These sites are located 7, 12, 29, and 3 lm from a creek bed (R Schorr, unpublished data). There was no consistency among sites in aspect. Three sites were in vegetation dominated by coyote willow (Salix exigua), one site was in vegetation dominated by snowberry and mullein (Verbascum thapsus). However, all four hibernacula appeared to be below coyote willows. The eight sites located during this study and the four U. S. Air Force Academy sites were not disturbed to protect any hibernating mice and therefore are only possible hibernacula because there is no confirmation a mouse actually hibernated there. Confirmation of a true hibernaculum cannot be made until a chamber, or nest is located. The other explanation for these collar locations might be either locations of radios discarded by the mice or dead mice carried underground by a predator. Location or more hibernation sites was limited primarily by normal battery failure of the radio transmitters before mice went into hibernation. Prior to the 1998 field season, natural mortality factors reported for Z. hudsonius included only insufficient fat storage prior to hibernation (Whitaker 1963 ), predation (Whitaker 1963, Poly and Boucher 1997, R. Schorr unpublished data) and cannibalism (Sheldon 1934). Other assumed natural mortality factors for Z. h. preblei included starvation, exposure, and disease. This study further reports confirmation of predation by house cats, garter snakes, rattlesnakes, and fox. Other documented natural mortality factors now include accidents by drowning and road kill. The high percentage of arthropods and endogenous fungus found in the fecal samples provides a new aspect to evaluating PMJM habitat requirements. When considering habitats used by PMJM, or in trying to predict habitats that might be suitable for the subspecies, consideration should be given as to whether those habitats could support the arthropods and fungus apparently being selected for by the mouse. Arthropods are a food source high in protein and fat which would benefit mice emerging from hibernation and mice preparing for hibernation. Whitaker (1963) reported a 67% loss of individuals over hibernation and that average body mass of individuals emerging from hibernation was greater than the average for mice entering hibernation. Because no mice are known to store food in their hibernacula, this indicates that the lighter individuals died during hibernation and only those entering with higher masses survived. All the energy they use during hibernation and the periodic arousals (the energetically most expensive part of hibernation) must be the fat they carry into hibernation (B. Wunder, personal communication). The ability to put on sufficient fat for overwinter survival during hibernation is a critical factor in the life history of these mice. Thus, appropriate and sufficient food sources must be available to the mouse to meet these nutritional requirements. �16 The apparent seasonal shift in mouse movements between the July-August and September tracking sessions may be a result of diet switching. The broader diets suggested by the fecal analyses from samples collected during July and August possibly represent both a wider availability of suitable foods and the ability of PMJM to exploit these resources. It might also suggest a need to exploit these other resources to provide the mouse with necessary food requirements for breeding. Because mice tracked during each session were not generally the same mice there is a possibility these apparent movement shifts might only be different areas used by different mice. The probability of this being the case is small for two reasons. The first is the low density of mice in the stream (Shenk and Sivert 1999). Given that, the proportion of mice followed during each session is a high proportion of the mice in the population. Thus, each subset of mice should be representative of the population. Secondly, evidence from the four mice followed during both the latter tracking sessions also exhibited movement shifts between sessions. The combination of shifts in both general mouse movements, individual mouse movements, and diet provide strong circumstantial evidence that PMJM may be selecting for or require specific seasonal diets. If this is the case, all these requirements must be considered and provided for to ensure conservation of the subspecies. A detailed literature search and further studies need to be conducted to investigate the possible implications of these dietary patterns. These data are limited to the first year of a multi-year study and thus annual variation in mouse movement patterns cannot yet be addressed. Continuation of this study, at least through the 1999 field season will provide information on annual variation in PMJM movement patterns. This study looked at PMJM movements at only three sites. These sites were selected based on known presence of PMJM. Other sites, perhaps because of different habitat configurations or sites of poorer quality may require mice to move further from the creek or may allow mice to remain closer to the creek in order to obtain all life requirements. However, these study sites were specifically selected to address spatial variation in movement patterns of PMJM due to different spatial configuration and juxtaposition of habitats. And although all study sites were different they could all be considered within the general description of what is consider typical habitat for PMJM. It should also be noted that we generally trapped along the creek. Thus, mice captured and subsequently radio-collared and followed were those mice that use the area near the most prominent drainage. If there are mice that do not regularly use the main channel and thus were not available for capture there would be a bias in the data towards fewer observations 300 feet away from the creek. To test for such a bias, further research needs to be conducted trying to capture mice further from the creek and compare their movements and locations of those movements in relation to distance from the creek. We plan to test for this bias during the 1999 field season at Maytag Property and Woodhouse Ranch. This is an interim progress report of a multi-year study. Further analyses will be conducted on these data and data collected during the 1999 field season. Literature Cited Armstrong, D. M., M. E. Bakeman, A. Deans, C. A. Meaney, and T. R. Ryon. 1997. Conclusions and recommendations in: Report on habitat findings on the Preble's meadow jumping mouse. Edited by M. E. Bakeman. Report to USFWS and Colorado Division of Wildlife. Bailey, B. 1929. Mammals of Sherburne County, Minnesota Journal ofMammalogy 10:153-164. Bailey, V. 1923. Mammals of the District of Columbia Proceedings of the Biological Society. Washington 36:103-138. Cook, T. D. and D. C. Campbell. 1979. Quasi-experimentation: design and analysis issues for field settings. Houghton-Miffiin, Boston. �17 Cormack, R. M. 1964. Estimates of survival from the sightings of marked animals. Biometrika 51:429-438. ERO Resources. 1995. Environmental review of South Boulder Creek Management Area Prepared for City of Boulder Real Estate/Open Space Department. Prepared by ERO Resources Crp., Denver, Colorado in association with Stoecker Ecological Consultants, Boulder, Colorado. Fitzgerald, J.P., C. A. Meaney, and D. M. Armstrong. 1994. Mammals of Colorado. Denver Museum of Natural History, University Press of Colorado. Niwot, Colorado. Hafner, D. J., K. E. Petersen, and T. L. Yates. 1981. Evolutionary relationships of jumping mice (genus Zapus) of the southwestern United States. Journal ofMammalogy 62:501-512. Hall, E. R. 1981. The mammals of North America. John Wiley and Sons, Inc., New York, New York, 2 volumes. Hamilton, W. J., Jr. 1935. Habits of jumping mice. American Midland Naturalist 16:187-200. Jolly, G. M. 1965. Explicit estimates from capture-recapture data with both death and immigration stochastic model. Biometrika 52:225-247. Jones, C. A. 1996. Mammals of the James John and Lake Dorothey State Wildlife Areas. Final Report, submitted to the Colorado Division of Wildlife and Colorado Natural Areas Program. Kendall, W. L., and J. D. Nichols. 1995. On the use of secondary capture-recapture samples to estimate temporary emigration and breeding proportions. Journal of Applied Statistics.22:751762. Kendall, W. L., K. H. Pollock, and C. Brownie. 1995. A likelihood-based approach to capturerecapture estimation of demographic parameters under the robust design. Biometrics 51 :293308. Kendall, W. L., J. D. Nichols, and J.E. Hines. 1997. Estimating temporary emigration using capturerecapture data with Pollock's robust design. Ecology 78:563-578. Krutzsch, P.H. 1954. North American jumping mice (genus Zapus). University of Kansas Publications, Museum of Natural History 7:349-472. Meaney, C. A., N. W. Clippinger, A. Deans, and M. OShea-Stone. 1996. Second year survey for Preble's meadow jumping mouse (Zapus hudsonius preblei) in Colorado. Report prepared for the Colorado Division of Wildlife. Meaney, C. A., A. Deans, N. W. Clippinger, M. Rider, N. Daly, and M. O'Shea-Stone. 1997. Third year survey for Preble's meadow jumping mouse (Zapus hudsonius preblei) in Colorado. Report prepared for the Colorado Division of Wildlife. Nudds, T. D. 1977. Quantifying the vegetative structure of wildlife cover. Wildlife Society Bulletin 5:113-117. Poly, W. J., and C. E. Boucher. 1997. Record of a creek chub preying on a jumping mouse in Bruffey Creek, West Vjrginia Brimleyana 24: 29-32. PTI Environmental Services. 1996a Preble's Meadow Jumping Mouse Study at Rocky Flats Environmental Technology Site, Annual Report 1996. Final. Rocky Flats Environmental Technology Site, Golden, Colorado. PTI Environmental Services. 1996b. Preble's Meadow Jumping Mouse Study at Rocky Flats Environmental Technology Site, Spring 1996. Final. Rocky Flats Environmental Technology Site, Golden, Colorado. Quimby, D. C. 1951. The life history and ecology of the jumping mouse, Zapus hudsonius. Ecological Monographs 21:61-95. Riggs, L. A., J. M. Dempey, and C. Orrego. 1997. Evaluating distinctness and evolutionary significance of Preble's meadow jumping mouse: Phylogeography of mitochondrial DNA noncoding region variation. Final Report for the Colorado Division of Wildlife. Denver, Colorado. �18 Seber, G. A. F. I 965. A note on the multiple recapture census. Biometrika 52:249-259. Sheldo~ C. 1934. Studies on the life histories of Zapus and Napaeozapus in Nova Scotia Journal of Mammalogy 15:290-300. Shenk, T. M. 1998. Conservation assessment and preliminary conservation strategy for Preble's meadow jumping mouse (Zapus hudsonius preblei). Colqrado Division of Wildlife FY 199798 Annual Report. ... • Shenk, T. M. and M. M. Sivert. 1999. Temporal and spatial variation in the demography of Preble's meadow jumping mouse (Zapus hudsonius preblei). Colorado Division of Wildlife JanuaryMarch 1999 Quarterly Report. Stohlgren,T. J., M. B. Falkner, and L. D. Schell. 1995. A Modified-Whittaker nested vegetation sampling method. Vegetatio 117: 113-121. USFWS. 1997. Interim survey guidelines for Preble's meadow jumping mouse. USFWS. Denver, Colorado. Whitaker, J. 0., Jr. 1963. A study of the meadow jumping mouse, Zapus hudsonius (Zimmerman), in cental New York. Ecological Monographs 33 :3. Table 1. Number of Preble's meadow jumping mice captured and radio-collared at each study site durin~ each traeein~ session. # CaQturedt Site Trapping session Trap nights Maytag June Maytag # Radio-collared female (%) male(%) total female male total 1949 9(64) 5 (36) 14 9 5 14 July 2370 15 (54) 13 (46) 28 ll 5 16 Maytag September 3256 18 (43) 24 (57) 42 IO 8 18 PineCliff June 1095 6 (32) 13 (68) 19 5 13 18 PineCliff July 1603 8(50) 8(50) 16 6 7 13 PineCliff September 1544 13 (28) 33(72) 46 4 12 16 Woodhouse June 1525 4 (57) 3 (43) 7 2 3 5 Woodhouse July 2420 9 (64) 5(36) 14 8 4 12 Woodhouse September 1568 8 (44) IO (56) 18 7 6 13 17330 90 114 204 62 63 125 TOTAlS t Some captured animals were PIT-tagged during previous trapping sessions �19 Table 2. Percent locations of Preble's meadow jumping mice at each of three distance categories for each study site and radio-tracking session. PineCliffRanch locations are also divided into percent locations for each distance category from center of the stream of initial capture for mice captured on either West Plum Creek or Garber Creek. % locations at each distance from center of main Site Tracking period N Maytag Jul-Aug 1998 Maytag <46m (150 feet) 46 - 91 m 150-300 feet >91 m (300 feet) 901 75.6 21.0 3.4 Sep-Oct 1998 900 47.8 19.6 32.6 PineCliff Ranch Jul-Aug 1998 302 86.1 11.9 2.0 PineCliff Ranch Sep-Oct 1998 520 81.2 9.2 9.6 Woodhouse Ranch Jul-Aug 1998 377 91.S 8.5 0.0 Woodhouse Ranch Sep-Oct 1998 533 67.6 32.1 0.4 Pine Cliff Ranch Garber Creek Jul-Aug 1998 210 68.1 10.S 21.4 Pine Cliff Ranch Garber Creek Sep-Oct 1998 499 75.6 9.8 14.6 Pine Cliff Ranch West Plum Creek Jul-Aug 1998 92 32.6 20.7 46.7 Pine Cliff Ranch West Plum Creek Sep-Oct 1998 21 52.4 0.0 47.6 Table 3. Preble's meadow jumping mice radio-collared and tracked (X) for more than one tracking session. Location of capture, sex, identification, and number of locations (n) for each mouse during each tracking session is noted. Tracking Session Site Sex PIT-tag# Maytag F Maytag Junei July-Aug n Sep-Oct 4140371625 X X 25 * F 4141435234 X Maytag F 4140753620 X Maytag F 41412B4A2D X Maytag M 4141241836 Woodhouse F 41413£610B Woodhouse F 41413E7B07 PineCliff F 4141125D66 PineCliff M 41413D4C04 1 Locational data from June is not included in this report. * Mouse captured but radio-collar not replaced. X X n X 52 X 82 103 X 105 X 93 X X 78 X 14 X 66 X s X 13 X 18 X IS * �20 Table 4. Possible hibemacula for Preble's meadow jumping mouse at three study areas in Douglas County, Colorado. Sex, beginning date of stationary signal, duration of stationary signal, distances from the main drainage (East Plum Creek at Maytag Property, West Plum Creek at PineCliffRanch, or Indian Creek at Woodhouse), nearest tributary, and center of night time locations are presented. Primary vegetation at the site is also noted. Distance (m) from: Site Sex Date Days Main Drainage Tributary Center Night Location Vegetation Woodhouse M 9/22 12 40 NIA 91 skunkbush, sumac, hawthome,chokecherry Woodhouse M 9/27 9 33 NIA 67 clematis vine, snowberry Woodhouse M 9129 5 23 NIA 116 narrowleaf cottonwood, Bromus spp., chokecherry PineCliff M 9117 5 210 JO 29 Salix spp. PineCliff M 9/23 11 35 40 90 Salix spp., alysium PineCliff M 9/23 11 120 35 70 Salix spp., grass PineCliff M 9/25 9 100 78 72 snowberry, Salix spp., alysium, thistle PineCliff F 9/25 9 70 l05 l05 snowberry, Salix spp. PineCliff M 9125 9 341 0 Salix spp., cottonwood, snowberry Maytag F 9/23 14 120 750 gambel oak, chokecherry, alysium NIA �Table 5. Known and probable mortalities of Preble's meadow jumping mice from Maytag Property, PineCliffRanch, and Woodhouse Ranch collected during the 1998 field season (June I-October 31 ). Probable cause of death, date radio-collars located, description of how what was found, sex of the animal 2 and dis12osition of s12ecimen noted. Site where found Mortality factor Date How obtained Sex Current location of specimen Maytag handling stress June 8 died during processing of the mouse, never came out of anesthesia female DMNH Maytag handling stress July 27 died from anesthesia shock female DMNH Maytag trap mortality June 10 trap mortality male Maytag trap mortality: heat stress Sept. 10 trap mortality, probably over-heated female DMNH DMNH Maytag predation July 2 tracked radio-collar, found collar in fox scat female no body, scat at CDOW Maytag predation Sept. 21 radio-collar tracked to a rattlesnake, waited and picked up snake scat with the radiocollar in it female no body, scat at CDOW Maytag road kill Sept. 15 tracked radio-collar; found highly decomposed, unsalvageable body covered with ants, side of road-probably road kill male no body Maytag unknown Aug. 1 tracked radio-collar; found highly decomposed, unsalvageable body male no body PineCliff possible predation Aug. 6 possible predation; crimp still tightly on collar when retrieved female no body PineCliff possible predation Aug. 12 tightly crimped collar found on ground, no part of mouse detected; suspect predation since movement detected earlier female no body PincCliff possible predation Sept. 13 possible predation; crimp still tightly on collar when retrieved male no body PineCliff drowning Aug. 6 radio-tracked collar; found mouse in water female rnv:IN-:::I Woodhouse handling stress June 5 died during handling while trying to readjust radio-collar male Woodhouse radio-collar: suffocation/starvation Aug. 6 radio-tracked collar, found dead mouse; collar may have been too tight female DMNH DMNH Woodhouse predation Aug. 11 radio-collar found in snake scat female no body, scat at CDOW Woodhouse predation Sept. 23 radio-collar tracked to a garter snake, waited and found collar in snake scat female no body, scat at CDOW Woodhouse predation Sept. 29 radio-collar tracked to a house cat, waited and found collar in house cat scat female no body, scat at CDOW Woodhouse possible predation Aug. I tightly crimped radio-collar found in exposed area female no body Woodhouse unknown Aug. 7 radio-tracked collar, found dead mouse (skeleton) on shrub female DMNH Woodhouse unknown Sept. 15 radio-tracked collar and found dead mouse male DMNH ts) �N Table 6. Percent contents of fecal samples collected from traps where Preble's meadow jumping mice were captured. All samples were collected N from Maytag Property during the 1998 field season. Several samples are from the same animal captured on different dates. Amount of bait in each samp 1e was quant1"fid o o t e samp l:),'T'fi e e as l00¾'A'f. 0 or ab un dant (703/cfh > or trace amount (103/cfth < e samp1:), e or Oo/cfh oO o o t e samp1e was b. rut. endogenous VerAgro- EquiLesunknown mushID Date Bait arthropod fungus seed moss IPOilen Poa Helianthus Salix 'querella bascum Carex ovron setum flower forb room 4l405E402 Jun 3 A 76 414l25476E 4141382614 Jun 3 T Jun4 T 64 7 ·4140271625 Jun 8 T 4 .4l4119674D Jun 11 A 91 4l407E5E35 4141106852 4141287948 Jul 21 Jul 21 0 100 16 Jul 28 100 41417Al807 • 4141297F0F Jul 28 Jul 28 100 A 4141407328 Jul 28 T 4141360708 41411E6879 4141283D24 4l4l3A026A Jul 29 Aug 3 Aug 13 A T UNKNOWN 4140371625 4140205478 4141187C76 4141284A2D . 4141241836 4140173A6D 414125476 • 4140753620 Aug 13 Aug 13 Aug 13 Aug 13 Aug 13 Aug 13 Aug 13 Aug 13 Seo 7 Seo 9 Sep 9 0 A A A T T T T T A A 92 23 13 16 4 l 96 9 4 82 18 2 47 30 4 3 17 96 16 25 7 46 11 29 67 59 12 60 2 46 3 25 20 6 98 4 41412CC443 41417CIE42 41411E6879 4141284A2D Sep IO Sep IO T Seo 11 Seo 12 A A 73 20 4140754707 Sen 12 T 24 21 12 3 16 5 ·" 2 5 2 43 16 14 90 11 6 100 100 100 T A 414164363C 24 89 78 11 5 84 30 19 84 4 65 5 5 27 80 7 35 34 �Table 7. Percent contents of fecal samples collected from traps where Preble's meadow jumping mice were captured. All samples were collected from PineCliffRan~h during the 1998 field season. Several samples are from the same animal captured on different dates. Amount of bait in each sample was quantified as 100%, 'A' for abundant (> 70% of the sample 'T' for trace amount(< 10% of the sample), or 0% of the sample was bait. ID UNKNOWN 4140761237 4141 IA3D4B 4140755157 UNKNOWN UNKNOWN 4140755157 UNKNOWN UNKNOWN 41411Alll4 4141714141 41413D4C04 4141505F54 ·.414125D66 413F6F7EOA 4140784Bl4 41411AEC12 41412F2433 41412F5E45 41413C604F .414176064A 4141656001 ;414138666D 41415B4748 41412F2F62 4141370528 4141350155 41415E434A 4141491F02 4140502204 Date Bait arthropod seed 100 Jun3 A Jun4 0 96 4 Jun4 A ,, 97 Jun4 A T Jun4 A Jun4 A 88 Jun4 8 T Jun 7 T T T Jun 8 A Jun 8 T 88 6 Jul 11 A 4 Jul 21 0 13 2 Jul 21 0 37 Jul 21 100 Jul 21 T 8 Jul 21 T 8 Jul 22 100 Jul 22 A Jul 24 A 41 Jul 24 A T 3 7 Seo9 T 3 8 Seo9 Seo 11 A 100 14 Seo 11 T 6 Sep 14 A 96 4 75 Seo 14 T 22 Sep 15 A 83 Sep 15 A T Sep 16 A Sep 16 T 17 endogenous Chenopodium grass Eleagnus Helianthus lilv oollen Salsola seed comoosite Festuca funl!US moss Bromus Poa seed 4 96 3 T 6 6 4 88 T 0 6 96 85 63 84 92 8 100 100 18 90 86 74 41 3 6 3 5 12 83 �· Ta,ble 8. Percent contents of fecal samples collected from traps where Preble's meadow jumping mice were captured. All samples were collected from Woodhouse Ranch during the 1998 field season. Several samples are from the same animal captured on different dates. Amount of bait in • each sample was quantified as 100%, 'A' for abundant(> 70% of the sample), 'T' for trace amount(< 10% of the sample), or 0% of the sample . was bait PIT-tag# Date Bait arthropod 41407B2575 Jun 3 A 100 414l3E610B Jun 3 A 100 41413E610B Jun4 A 100 41407B2575 Jun4 T 100 414147354F Jun4 T 24 UNKNOWN Jul 27 T 41413E610B Aug4 T 4141076B41 Aug4 100 4141250646 Aug23 0 12 41404A7D58 Aug23 T 4 '414128F2607 Aug23 0 41415D402E Aug23 100 41406A1049 Sep 9 A 100 4141313E6F Sep 9 A 74 4141350C2A Sep 10 JOO endogenous fungus mushroom Poa seed Agropyron Sporobolus seed Chenopodium composite Equisetum flower 76 50 50 23 77 88 96 89 11 26 T 414133085 Sep 10 A UNKNOWN Sep 11 0 100 414173462B Sep 11 T 70 4143P5772B Sep 11 0 45 4141611413 Sep 12 0 98 41414A6B6F Sep 12 0 88 4140241303 Sep 12 0 88 30 33 7 15 2 12 2 10 �25 APPENDIXB HABITAT USE AND DISTRIBUTION OF PREBLE'S MEADOW JUMPING MOUSE (Zapus hudsonius preblei) IN LARIMER AND WELD COUNTIES, COLORADO Tanya M. Shenk and James T. Eussen Abstract A total of39 sites were surveyed for Preble's meadow jumping mouse in Larimer and Weld Counties, Colorado with a total of 16,671 trap-nights completed. Of the 39 sites survey, Zapus spp. were captured at 21 locations. A total of 71 unique individuals were captured in the 21 locations, with recaptures of 14 of those individuals. All sites where Zapus spp. were captured were located in Larimer County. Three soil samples from 0-12 inches and 12-24 inches were collected at all sites where possible. Soils at some sites were too rocky to collect the 12-24 inch soil samples. Nine fecal samples were collected from traps where Zapus spp. were captured. Initial result indicate that arthropods, fungus, and willows are important dietary components of the Zapus diet. Two trap mortalities of Zapus spp. occurred. Both specimens were saved and delivered to Cheri Jones of the Denver Museum of Natural History on October 26, 1998. Measurements of habitat characteristics at both the site and landscape levels were completed. Genetic tissue samples were collected from 68 of the 71 jumping mice captured. Analyses of these tissue samples is pending. Introduction The meadow jumping mouse (Z. hudsonius) is broadly distributed across North America from the Atlantic to Pacific coasts, extending south into the United States to Alabama and Georgia and west across the Great Plains to the base of the Rocky Mountains. In general, it is a common inhabitant of moist, grassy and herbaceous fields. Eleven living subspecies have been described (Whitaker 1972). Hafner et al. (1981) describe a twelfth subspecies, Z. h. luteus. Z. h. preblei occurs only in eastern Colorado and southeastern Wyoming (Krutzsch 1954, Long 1965, Armstrong 1972). From its limited ecological and geographic distribution, Fitzgerald et al. (1994) suggest it is an Ice Age relict, once widespread in tallgrass prairie across the eastern plains of Colorado but now restricted to scattered localities on the Colorado Piedmont. Similar relict populations of meadow jumping mice in the White Mountains of Arizona and the Sacramento Mountains and Rio Grande Valley of New Mexico are described as the subspecies Z. h. luteus (Hafner et al. 1981 ). On May 12, 1998 the U. S. Fish and Wildlife Service (USFWS) published a final rule in the Federal Register (63 FR 26517) to list Preble's meadow jumping mouse (Zapus hudsonius preblei) as 'threatened' under the Federal Endangered Species Act (ESA) of 1973, as amended. Scarcity of suitable habitat presumably limits current distribution of Preble's meadow jumping mouse (PMJM) and thus, maintenance of quality habitat has been identified by the USFWS (63 FR 26517) as the principal conservation goal. Although Meaney et al. ( 1997) reported an improved ability to recognize suitable habitat for PMJM, a more refined and complete definition of potentially suitable habitat for the mouse does not exist. Because the protection of potentially suitable habitat for PMJM may occur under the ESA, as well as protection of known locations where the subspecies occurs, the definition of potentially suitable habitat, as determined by the USFWS, will directly influence site specific regulatory procedures. Ideally, a definition of potentially suitable habitat for the subspecies would identify areas where the PMJM could survive and reproduce in sufficient numbers to sustain populations throughout its range of natural variability over an extended length of time. During the active period of the life cycle of the mouse, suitable habitat must provide requirements for daily survival, reproductive activities (breeding, nesting,. and rearing of young to independence), and dispersal. The hibernation period . .. . . ~ • . • �26 requires habitat with sufficient food supplies to assure fat storage prior to hibernation and hibernacula sites. Habitat providing all seasonal and life cycle requirements may or may not occur in a single contiguous area If not in a contiguous area, habitat patches must occur in a mosaic of usable areas where suitable corridors exist for seasonal movement among sites. Because very little is known about the ecological requirements of the subspecies, potentially suitable habitat is currently defined only as areas of well-developed, dense herbaceous vegetation consisting of a variety of grasses, forbs and thick shrubs in close proximity to open water. This definition is vague and possibly incomplete. Numerous surveys have been conducted since 1990 to establish the current distribution of PMJM. However, very few surveys for determining the presence or absence of PMJM have been conducted in Larimer or Weld Counties, Colorado. Of the surveys conducted up to 1997, only three sites yielded captures of PMJM in either Larimer or Weld Counties. In Larimer County, two sites (Lone Pine and Rabbit Creeks) yielded captures of PMJM, based on field identification and supported by genetic analyses conducted to date (Riggs 1998). One mouse was captured on Lone Tree Creek in Weld County that was identified as Z. h. preblei in the field but was genetically found to be more . closely allied with the western jumping mouse, Z. princeps (Riggs 1998). Therefore, very little information is known about habitat use or current distribution of the subspecies in these two counties. Both Larimer and Weld Counties include habitat that is currently perceived to be outside the ecological limits of PMJM. Larimer County provides the opportunity to explore elevational limitations, currently thought to be 7400 feet (2260m) (USFWS 1997); Weld County extends beyond the currently believed eastern boundary of the subspecies. Both of these boundaries will be challenged by selecting survey sites within the two counties beyond the perceived ecological limits of the subspecies. Genetic tissue samples will be collected on all PMJM captured to assist identification through DNA analysis, particularly for those animals captured outside the currently perceived ecological limits, and to provide information for future studies to better define the relationships among different populations of Z. h. preblei, other subspecies of Z. hudsonius and other species of Zapus. Quantifying and comparing both landscape and site specific characteristics of habitats where PMJM are found and not found in these two counties would further our understanding of habitat use by the subspecies, particularly in the northern half of its probable range. Any new locations where PMJM are found during this study would contribute to the range-wide distribution map currently available for the subspecies. Repeated annual visits to these same randomly selected sites would also provide information on population persistence at a given site. Such a monitoring scheme would be necessary for evaluating the continued status of the mouse at each site, providing further information as to the suitability of the habitat for long-term persistence of the population. Thus, the primary objective of this study is to quantify and define habitats used and identify ecological limits of the subspecies in Larimer and Weld Counties, Colorado. Such information could be used by the USFWS to further refine the current definition of potentially suitable habitat for PMJM. Objectives Conservation of Z. h. preblei should require maintaining populations of the subspecies throughout the range of its natural variation and to try to identify ecological limits for the subspecies. Specific objectives for the distributional component of this study are to (1) clarify the distribution of the subspecies in Larimer and Weld Counties, Colorado, and (2) define and quantify the amount of suitable habitat within Larimer and Weld Counties. The objective of the habitat study is to identify and refine ecological requirements of Z. h. preblei in Larimer and Weld Counties, Colorado. Because of the uncertainty of the identification of the mouse captured in Weld County, Colorado, and because currently perceived ecological boundaries of PMJM will be challenged in the survey effort, the objective of the genetic component of this study is to genetically identify all Zapus spp captured during this study. Therefore, genetic tissue samples will be . collected from each jumping mouse captured to confirm identification of the animal. These genetic �27 tissue samples could also provide further data to better define the relationships of different populations of Z. h. preblei as well as to other subspecies of Z. hudsonius and other species of 7.apus through molecular systematic relationships and to link these genetic relationships to systematic studies of Z. hudsonius. Specific objectives for this study on PMJM include: . 1. Define and quantify habitat characteristics at sites wllere PMJM was found and sites where PMJM was not found. 2. Identify elevational and eastern boundary limitations for PMJM in Larimer and Weld Counties, Colorado. 3. Describe the distribution of PMJM in Larimer and Weld Counties, Colorado. 4. Select sites and establish monitoring protocols for determining long-term trends in populations of PMJM. 5. Collect genetic tissue samples to assure identification of jumping mice captured during this study and for future genetic analyses to better define the. relationship of populations of Z. h. preblei to each other, other subspecies of Z. hudsonius and other species of 7.apus. Study sites Trapping surveys will be conducted at a minimum of30 sites selected at random from the sampling frame developed for Larimer and Weld Counties, Colorado. No survey site will occur at elevations greater than 7600 feet and not east of the UTM Easting Coordinate of 602000. The molecular systematic studies will be conducted in genetic laboratories using the genetic tissue samples collected in the field (ear punches). Methods For simplicity, the methods described below assume all jumping mice captured are PMJM. To meet these objectives, this study included (1) constructing a sampling frame of potentially suitable habitat within the area of interest (Larimer and Weld Counties, Colorado) and (2) conducting trapping surveys to determine presence or absence at a minimum of 30 sites randomly selected from the sampling frame. Specific approaches to each of these tasks follow. Sampling frame A sampling frame was developed to delineate potentially suitable habitat for PMJM from unsuitable habitat. To produce a preliminary map displaying existing potential habitat for PMJM would require at least three primary layers of ecological information. Two initial layers, hydrology and vegetative cover would provide a map displaying all areas where these two ecological requirements of PMJM co-occur. The hydrology layer must be in enough detail to provide locations of intermittent and small order streams as well as identify water source (e.g., irrigation ditches, stream, seep). Degree of density of vegetative cover is sufficient for initial mapping efforts. Demarcation of shrub, trees and grasses will suffice in these initial efforts rather than species identification. However, not all combinations of vegetative cover and water provide sufficient habitat to support PMJM. Thus, mapping potential habitat as described above would produce a map displaying far more potential habitat than exists. For example, an extremely critical ecological requirement for the survival of the mouse, that to date we have very little information for, is potential hibernacula sites. As a possible index to hibernacula habitat, a third GIS layer, indicating the presence of alluvial deposits may provide some insight in to hibernation requirements. Areas where alluvial deposits co-occur with the presence of water and vegetative cover may provide a more realistic map of potentially suitable habitat for the mouse. Our ability to survey a site selected at random would be restricted to those areas where we are _able to bbtain permission to access th_e land. We shou_ld_be able_to survey any randomly se_lected site �28 on public lands and thus our inference should be unrestricted. Such readily available access would probably not occur on privately owned or leased lands, thus, restricting inferences made on private lands. By stratifying the sampling frame, this lack of access on private lands would not restrict inferences that can be made on public lands. Inference on private lands will depend on the percent of access we get from private landowners. The sampling frame was developed from the 1:24,000 sc'ale Hydrographic GIS layer developed by the CDOW (Reese Tietje, unpublished data). From this sampling frame a three-stage sampling design was used to select 40 random sites (to ensure meeting our objective of a minimum of 30) for conducting PMJM surveys. Primary sampling units were 3rd order streams. Of the 150 3rd order streams that exist within the sampling frame, 40 were selected on public lands, 40 on private lands. The secondary sampling unit was the stream stretch selected within the primary (3rd order stream) unit. The tertiary sampling unit was the actual sites along the stream stretch selected at the secondary sampling stage that were sampled. Each of the primary, secondary, and tertiary sampling units were selected using simple random sampling. If the secondary sampling unit did not yield potentially suitable habitat (i.e., no dense vegetation) secondary sampling units continued to be selected at random until one yielded potentially suitable habitat. Determination of suitable habitat was made by field site visits. Total stream length of all the secondary sampling units were recorded as well as total stream length containing potentially suitable habitat. From these measurements an estimate of the probability of potentially suitable habitat occurring within the primary sampling unit can be made for each area surveyed. A mean estimate, over all sites surveyed, of the probability of a primary sampling unit having potentially suitable habitat was made. Because of the random selection of primary sampling units, inference can be made as to the probability of potentially suitable habitat existing on all 3rd order streams within the sampling frame. By recording presence or absence of PMJM within the tertiary sampling units we can estimate the probability of PMJM occurring within potentially suitable habitat. Because of the random selection of sampling units in each of the two strata (public and private lands) we can then estimate the probability of PMJM occurring in potentially suitable habitat within both public and private lands. Trapping surveys Trapping surveys to determine presence/absence of PMJM were primarily conducted following protocols defined by the USFWS (1997). Additional guidelines have also been developed to accommodate the needs of this project. Trapping surveys were conducted within the period from June I to September 15, 1998. These dates correspond to dates when PMJM is known to be out of hibernation. Trappers were_advised to follow the Center for Disease Control's Hantavirus instructions and recommendations when dealing with rodents. Due to the nocturnal nature of PMJM, traps were set between 19:00hrs and 21 :00hrs and checked as early as possible in the morning beginning at 5:00hrs to reduce stress and the potential for predation on trapped animals. Time required to complete the traplines varied depending on how many animals were caught. Small mammal Sherman live traps (folding and non-folding) were used to conduct the trapping sessions. Traps were set in two parallel lines of trap stations (I trap per station) on either side of the drainage. Trap stations were 5 meters apart for a total of 250 meters; the parallel transects were 10 meters apart unless extent of habitat, terrain topography, or stream hydrology did not allow. Location of transects were recorded on field data sheets and 7.5° topographic maps. A small (~l inch) ball of polyester quilt was placed in each trap as nesting/bedding material. Baiting material was Manna Pro Sweet 3-way Livestock feed which contains no animal matter; ingredients include flaked barley, flaked corn, flaked oats, and cane molasses. Peanut butter was used to stick the bait to the trap. �29 Traps were checked by two surveyors. All animal captures were recorded. If an animal had been captured in a trap, a ziplock plastic bag was placed over the end of the trap. The trap was opened allowing the animal to fall into the plastic bag. The animal was identified to species while in the plastic bag. If the animal was not a PMJM, identification of the animal was recorded and the animal set free without further handling. Each captured PMJM was checked to detect the presence/absence of a PITtag (See below). If a PIT-tag was detected and the mouse had been captured at that same site within the same trapping session, identification of the mouse was recorded and the mouse was released. If the animal was a PMJM and no PIT-tag was detected the PMJM was anesthetized for further processing, as follows. All PMJM were PIT-tagged. Each PMJM was weighed while in the bag, recording the weight in grams using a Pesola spring balance. Sex, age (juvenile, adult), and reproductive condition (pregnant, lactating or nonbreeding if it was female; for males the position of the testes indicated if in breeding status or nonbreeding status) was noted. Each PMJM was measured (total length, length of body, length of hind foot - heel to distal end of claws) in millimeters. Capture of every PMJM was documented by taking a photograph of the mouse on first capture. Mice were placed in a plexiglass photobox to reduce handling stress while taking the photograph. If the mouse had been captured previously as noted by the detection of a PIT-tag, no photograph was taken. If there were feces in the trap where a PMJM was captured, the fecal material was collected in a plastic bag labeled with date, location of the site, and animal PIT-tag number. Fecal samples were kept cool until returned to the CDOW office where they were frozen. Salt was added to each specimen bag for preservation. Once the field season was over all fecal samples were analyzed by the Composition Analysis Laboratory, Inc, 622 ½ Whedbee, Fort Collins, CO for content. All trap mortalities were recorded on a 'Trap Mortality' data form which included information on species, potential duration of time spent in the trap, and any information available to help determine cause of death. All animals found dead were double-bagged in a plastic bag and placed in a cooler with ice. Specimens were frozen as soon as possible and deposited in the CDOW freezer at the office in Fort Collins. A museum card was completed and attached to each PMJM specimen and the specimen and card were given to Cheri Jones, Curator of Mammalogy at the Denver Museum of Natural History, for study skins and tissue storage. If an animal was severely injured (e.g., severed limb, large lacerations) it was euthanized by soaking a cotton ball in Metofane (methoxyflurane) and placing the cotton ball and the mouse in a ziplock bag until the animal stopped breathing. If an animal appeared to be only slightly injured (e.g., broken tail, small laceration) the animal was released to the wild. If an animal appeared to be coldstressed attempts were made to warm it by holding it in the surveyors hands and/or against their body. If the animal appeared to be heat stressed, isopropyl alcohol was applied with a cotton swab to the ears, arm pits, and feet to cool it down. PIT-tagging Each PMJM was individually marked with a Passive Integrated Transponder (PIT-tag). PITtags are electromagnetic, glass-encased tags that emit a passive signal (125 kHz) that can be decoded by a portable reader. Destron-Fearing PIT-tags were used on all mice. We used portable readers from Biomark to read the PIT-tags. All mice were anesthetized to PIT-tag them. Anesthesia procedures followed protocols successfully used on PMJM before (R. Schorr personal communication). Mice were anesthetized by placing them and a cotton ball with 1 ml ofMetofane (methoxyflurane) into a sealed ziplock bag (to keep Metofane fumes in bag). The bag was then lain aside to minimize stress for the mouse and the time the mouse struggles in the bag. The less agitated the mouse is in the bag the less time required for the metofane to take affect. The mouse was observed at all times to make sure the mouse did not .. position itself such that the c_otton ball was in direct cont~ct with its face, which might c~use the mouse �30 to have a severe reaction to the anesthesia After the animal stopped moving, the handlers waited one minute before removing the mouse from the bag. The animal typically remained anesthetized for 2-3 minutes once outside the bag. Each PMJM was PIT-tagged using a protocol successfully used by C. Meaney (personal communication). Each PMJM had a PIT-tag inserted above the s~oulder blades by lifting the skin on its back and inserting the individually sterilized needle with the PIT-tag under their skin and injecting the tag. Verification of the PIT-tag identification number was made before and after insertion into the mouse by running the PIT-tag scanner across it. The skin behind the opening was then pinched to prevent emergence of the tag. PIT-tag identification number was recorded on the field data form. If, at any time during the handling of the PMJM the animal appeared to be severely stressed (dramatic changes in heartbeat, respiration and responsiveness or gums turning blue) first aid was administered and, once recovered, released without further processing. Absence of PMJM at a site was defined as no captures of PMJM from a minimum of four consecutive nights and 400 trap nights ( a trap night is defined as the sum total number of traps available each night). Presence of PMJM at a site was defined as at least one capture of PMJM at the site. However, to accommodate sample size requirements for the monitoring and genetics study (see below) trapping efforts continued until at least 10 individuals were tagged with PIT-tags. Habitat Use The general approach was to measure both site specific and landscape features of at least 30 sites selected at random to be surveyed for the presence of PMJM. A comparison of sites where PMJM were found and were not found was then made in an attempt to refine our current understanding of the ecological requirements of the subspecies. The following habitat characteristics were measured and recorded for each site. Site characteristics Cover was measured at 20 random locations along the 250 meter sampled stream stretch. Mean cover and standard error of the mean will be estimated. Cover was estimated using a vegetation profile board (Nudds 1977) that allows for an assessment of visual obstruction in 0.5 meter vertical intervals above ground. The board was 2.0 meters high and 30.48 cm wide. The board was marked in alternate black and white colors at 0.25 (0.50) meter intervals. Horizontal cover was assessed in each interval by viewing the board from 15 meters away in a randomly chosen direction. The percentage of each interval concealed by vegetation was recorded as 0-20, 21-40, 41-60, 61-80, and 81-100% estimated concealment. Vegetation species composition and richness was estimated using the Modified-Whittaker nested vegetation sampling method (Stohlgren et al. 1995). A modified Whittaker plot is 20-meters x 50-meters with 10 lm2 and 2 10m2 subplots arranged systematically around the perimeter and 1 100m2 subplot centered in the inside of the 20 meter by 50 meter plot. Species composition and percent cover (optical estimate) of each species was recorded for each subplot. Soil samples were taken randomly at each site for a hydrometer textural analysis (percent sand, silt and clay). Samples were collected using an AMS 24" soil recovery probe. Samples were taken from 0-12 inches, and 12+ to 24 inches. The 0-12 inch sample was taken first, removing the probe and soil. Samples were placed in a labeled ziplock bag. The probe was then placed in the same hole for the 12+ - 24 inch probe, with that soil sample being placed in a separately labeled ziplock bag. The hydrometer textural analysis will be conducted at the CSU Soils Laboratory. Mean stream width was estimated by measuring stream width at 30 locations along the entire site sampled. The 30 locations were stratified equidistant from each other to cover the entire stream stretch in increments of site stream length/30. �31 The above listed parameter estimates will be compared for sites where PMJM were. captured and sites where PMJM were not captured. Landscape characteristics The following landscape habitat characteristics will be measured and compared for sites where PMJM were captured and sites where PMJM were not captured using either GIS, topographic maps, or aerial infrared photography. 1. Distance to nearest human habitation and human disturbance. 2. Connectivity of sites to other streams from 1:24,000 topographic maps. 3. Total length of stream stretch at each site and total length of stream stretch with suitable habitat at each site. Molecular Systematics Genetic tissue samples were collected from all PMJM captured during the study. To date, positive identification of the individual to subspecies cannot be made using only genetic information. However, results of the DNA analysis combined with the morphometric data that will also be collected (e.g., body length, tail length) and the photograph of the individual should provide sufficient information to support the field identification to subspecies. As established by the USFWS (1997) Survey Guidelines, presence of PMJM is established when only one individual is captured. Thus, we could stop our trapping efforts after the first PMJM capture. However, because we would also like to provide sufficient genetic data to more fully explore the relationships of different populations of Z. h. preblei we will attempt to collect genetic tissue samples from a minimum of 10 individuals per successful survey site. A sample of at least 10 individuals from a population will provide enough information to document the genetic variation within the population {T. Quinn, personal communication). Genetic tissue sampling Genetic tissue samples were collected from every PMJM captured at each survey site. The following protocol was used based on previous success with PMJM (M. Bakeman, C. Meaney, T. Ryon, R. Schorr, personal communication). A fresh pair of clean latex gloves were used to handle each mouse. With a clean ear punch tool, one tissue plug was obtained from each mouse ear. If there was excessive bleeding, gentle pressure was applied to the injured area for approximately one minute. Each ear plug was placed in a vial of 95% ethanol. Both the vial and the corresponding record on the data sheet were labeled with a unique identifier as follows. A seven-place alpha-numeric code was composed as follows: a three-letter designator for the survey location ( e.g., MYT = Maytag Property); a number beginning with two digits indicating the year (e.g-., 98); followed by 2 digits specifying the individual trapped, numbered sequentially. For example: MYT9801 = first animal sampled at Maytag Property in 1998. After returning from the field, all samples were put in a cool place or refrigerated until delivery to the CDOW office at 317 West Prospect, Fort Collins, CO where they are being held in the freezer for future analyses. Ear punch tools were cleaned after each use by immersing them in a 10% bleach solution for a few minutes, rinsing them thoroughly with clean water, and then dried thoroughly to prevent rusting. Dull ear punch were discarded and replaced with new sharp punches to ensure the quickest, cleanest cut possible. Results Trapping results Thirty-nine sites were surveyed for PMJM in Larimer and Weld Counties, Colorado (Fig. 1), with a total survey effort of 16,671 trap-nights. Of the 39 sites survey, Zapus spp. were captured at 21 survey sites (Table 1). A total of 71 unique. individuals-were captured in the 21 locations, wit}:l_· �32 recaptures of 14 of those individuals. All sites where jumping mice were captured were located in Larimer County. Two trap mortalities of Zapus spp. occurred. Both specimens were saved and delivered to Cheri Jones of the Denver Museum of Natural History on October 26, 1998. Numbers of other small mammal species captures was recorded for each site surveyed (Table < 2). Nine fecal samples were collected from traps where Zapus spp. were captured. Results of eight samples indicate the jumping mice at the capture sites in Larimer County feed on arthropods and fungus which occurred in 50% (4) of the samples, followed by willows, pollen, and moss occurring in 38% (3), 25% (2) and 25% (2) of samples respectively. Other food items found included, moss, horsetail, unknown flowers, and unknown seeds (Table 3). Habitat characteristics Measurements of habitat characteristics at both the site and landscape levels were completed. No single habitat feature could be identified as unique to either sites where jumping mice were captured or sites where jumping mice were not captured. Kentucky blue grass (Poa pratensis), snowberry (Symphoricarpos spp), several species of willow (Salix spp.), and Mountain Alder were most abundant in areas occupied by Zapus spp. Horsetail (Equisetum spp), wild rose (R.osa spp), and cottonwood were also common. Dominant vegetation at survey sites with negative trapping results for jumping mice included wheat grass {Agropyron spp), Canary reed grass (Phalaris angustifolia), thistle, and willows (Table 4). Russian olive (E/aeaqnus angustifolia) occurred in 6 (33%) of the areas where no jumping mice were found, while absent in areas where jumping mice were captured. Jumping mice were captured in vegetation adjacent to large rivers, small streams, ponds, ditches, and seeps. Mean stream width varied considerably for both sites where jumping mice were captured (0.3-24.Sm) and where no jumping mice were captured (0.0-20m)(Table 4). No jumping mice were found in locations surveyed east of Interstate 25(125) during 4,716 nights of trapping effort. Habitat features along certain stretches of Willow Creek within the Pawnee National Grasslands and near the town of Briggsdale are similar to areas where jumping mice were found west ofl25. However, these sites are extremely isolated and the surrounding landscape is in either rangeland or agriculture production. From late June through late August, 22 locations in Larimer County were surveyed. Zapus spp were found in 20 of these 22 sites. The exceptions being the effluent below Seaman Reservoir on the Poudre River and Soldier Canyon west ofHorsetooth Reservoir. Below Seaman Reservoir just prior to and during trapping efforts, livestock (primarily cattle) were allowed to graze along this section of the Poudre River. Habitat composition along Soldier Canyon is similar to that of Arthur's Rock ~3/4 miles to the south (occupied by Zapus spp). However water is permanent along the Arthur's Rock drainage but flows intermittently through Soldiers Canyon. Only one jumping mouse was captured west ofl25 after August 24 during 5,382 nights of trapping effort at nine different sites. All locations trapped after August 24 were within drainages known to be occupied by Zapus spp., including four sites along the Big Thompson River (this study), two along the Poudre River (this study), and one each along Rabbit Creek (Meaney et al. 1997), Bull Creek (this study) and Lone Tree Creek (Meaney et al. 1997). The surrounding areas that were trapped along Bull Creek and Lone Tree Creek had been extensively grazed, leaving only isolated patches of vegetation over most of the site, however all other locations were similar in habitat composition to known PMJM locations. Three soil samples from 0-12 inches and 12-24 inches were collected at all sites where possible. Soils at some sites were too rocky to collect the 12-24 inch soil samples. A total of 104 012inch samples and 46 12-24 inch soil samples were delivered on October 21, 1998 to the Colorado State University-Soil Composition lab for soil-textural analyses. To date-these analyses have not beeri. compJeted. �33 Genetic sampling Genetic tissue samples were collected from 68 of the 71 jumping mice captured. Analyses of these tissue samples is pending. Discussion The habitat matrix within the range of Z. h. preblei is mixed grasslands adjacent to the Colorado Front Range along the Piedmont and along the base ofti.e Laramie Mountains in Wyoming and extends to the Colorado plains. Within this matrix, PMJM occur along stream drainages that contain patches of suitable vegetation. Suitable habitat appears to have at least two major components. The first component is a supply of open water, at least in part of the active season (M. Bakeman, C. Meaney, personal communication). Secondly, areas where PMJM has been found have dense cover (M. Bakeman, C. Meaney, personal communication). Based on studies of Z. h. preb/ei and Z. hudsonius elsewhere, Z. h. preblei apparently occurs mostly in undergrowth consisting of grasses, forbs, or both in open wet meadows and riparian corridors, or where tall shrubs and low trees form an overstory and provide adequate cover (Armstrong et al. 1997). Meadow jumping mice are widespread in abandoned grassy fields, but are often more abundant in thick vegetation along ponds, streams, and marshes or in rank herbaceous vegetation of wooded areas (Whitaker 1963). Vegetation at sites where Zapus spp were captured in this study were structurally diverse. No single plant species dominated areas occupied or unoccupied by Zapus spp. Such diversity in a multistory cover has been reported for the majority of other sites where Z. h. preblei have been found (Armstrong et al. 1997, Meaney et al. 1997). Vegetation composition of the dense cover varied considerably and included both native and non-native species as was found by other studies (Meaney et al. 1996, 1997, Armstrong et al. 1997, M. Bakeman, personal communication). Results from this study continue to support the habitat descriptions provided by Armstrong et al. (1997) which follows: The herbaceous understory is primarily grasses or forbs or a mixture of the two. Few of the sites, however, are dominated by fewer than two understory species. The tall shrub canopy at most sites is willow (Salix spp.), although scrub oak (Quercus gambelli), birch (Betula spp.), and alder (A/nus spp.) occur in sites south of the Palmer Divide Ponderosa pine (Pinus ponderosa) is the most common tree at higher elevations. The mouse appears to tolerate weedy or exotic species in areas that are structurally diverse and species rich; nearly every successful site contained Canada thistle (Cirsium arvense ). Thus, the mouse does not appear to have an affinity toward any single plant species but instead favors sites that are structurally diverse and provide adequate cover and food throughout its life cycle. PMJM have been trapped in natural riparian areas as well as areas altered by anthropogenic influence including ditches and wetlands adjacent to interstate highways, cement-lined ditches with tall cover, ditches along driveways and moderate road use, and moderate cattle grazing (M. Bakeman, personal communication). Open water was present at all sites surveyed in this study, with stream width varying across locations with and without jumping mice. Jumping mice were captured in 21 new locations in both the Poudre River and the Big Thompson River drainages in Larimer County. No jumping mice were captured at any survey site located in Weld County, however there may still be some sites within and adjacent to the Pawnee National Grasslands that may support PMJM (M. Ball, personal communication). Fish Creek, north of Bull Creek is the only location trapped after the 24th of August where jumping mice occurred. Poor trap success in late summer especially in areas along the Big Thompson and Poudre Rivers, including locations near Cherokee Park SWA, may be influenced in part by jumping mice entering hibernation. Elevation at these sites ranged from 5800 ft along the Poudre River to 7600 ft along Bull Creek north of Cherokee Park, much higher then the maximum elevation in D_ouglas (Shen~.and S.i':ert 1999,a, l 999b.) and Boulder. Countie~_(M.eaney_et al 19_99) \1/,heren_o r_edu~tion..in trap success • • •• . ... " could be detect_~d. . . . . .. •; �34 Food habit studies using scat analysis indicate jumping mice captured in Larimer County fed on several different food types. This is similar to Shenk and Sivert (1999b) in Douglas County, Colorado who also reported several food items in Z. h. preblei scat. It appears jumping mice captured in Larimer County fed on willows, with willow present in 38% of the scats examined. This is in contrast to Shenk and Sivert (1999b) who report virtually no willows in the scat examined from 85 fecal samples collected at three sites in Douglas County. ,, PMJM are only one component of the small mammal community inhabiting areas where they were captured. These mice were more often found at sites with high species richness and small mammal abundance (Table 2). The highest occurrences of Mus musculus captures occurred on sites where jumping mice were not found. This is similar to results summarized in Shenk (1998). The presence oflarge numbers of house mice might suggest degradation of habitat for Z. h. preblei in those areas or possibly competition between the two species (Ryon 1996). Jumping mice were also not found in areas where large numbers of captures of deer mice (> 115) occurred. These preliminary analyses have treated all the jumping mice captured the same. However, the truth may be they are all PMJM, all western jumping mice, or a mixture of the two. Once the genetic study is completed and species identifications confirmed these analyses will need to be reevaluated in light of that information. If some of the mice captured are identified as western jumping mice it would be most beneficial to compare areas of occupancy by this species to areas occupied by PMJM and to areas potentially occupied by both. According to Fitzgerald et al. (1994) the distributions of Z. princeps and Z. hudsonius do overlap. The area of overlap occurs in eastern Wyoming. The distribution of Z. hudsonius in Colorado is now known to be larger than shown in Fitzgerald et al. (1994). The boundaries are currently as far south as Las Animas County, based on genetically identified specimens of Z. h. preblei (Riggs 1998). Captures of Z. princeps and Z. hudsonius (as identified in the museum) have occurred as close as eight miles of one another within the same drainage (Armstrong 1972). Z. princeps were reported captured in 1981 (Olson and Knopf 1988) at the Lone Pine site in Larimer County, Colorado, where Z. h. preblei were captured by Menaey et al. (1997). Because neither specimens or genetic samples were taken in the 1981 study, identification of those mice will remain in question. The discrepancy may be explained by field misidentification or Meaney et al. (1997) also suggest this discrepancy might be explained by displacement of Z. princeps with Z. h. preb/ei sometime in the sixteen intervening years between trapping efforts. Although assumed to have different ecological requirements, genetic evidence presented by Riggs (1998) suggests further investigation of possible distributional overlap between Z. h. preb/ei and Z. princeps. These preliminary results from this study suggests further research should be conducted to continue evaluating range-wide distributional boundaries (e.g., elevation restrictions). Further studies should also be conducted to investigate areas of potential sympatry or hibridization of PMJM with Z. princeps (western jumping mouse). Literature Cited Armstrong, D. M. 1972. Distribution of mammals in Colorado. University of Kansas, Museum of Natural History Monograph 3:1-415. Armstrong, D. M., M. E. Bakeman, A. Deans, C. A. Meaney, and T. R. Ryon. 1997. Conclusions and recommendations in: Report on habitat findings on the Preble's meadow jumping mouse. Edited by M. E. Bakeman. Report to USFWS and Colorado Division of Wildlife. EG&G. 1993. Report of Findings: 2nd Year Survey for the Preble's meadow jumping mouse. Prepared by Stoecker Environmental Consultants for ESCO Associates, Inc., Rocky Flats Environmental Technology Site, Jefferson County, Colorado Fitzgerald, J. P., C. A. Meaney, and D. M. Armstrong. 1994. Mammals of Colorado. Denver Museum of Natural History, University Press of Colorado. Niwot, Colorado. H::uner, D. J., K, E. Petersen, .and T. L. Yates. 198 L Evolutionary relationships of jumping mice .,. . . . .. (genµs µipus) of the s,outhwestern U~ited States. Journ.~ of_M!iJllmalogy 6~:501-51.2 ... �35 Hall, E. R 1981. The mammals of North America John Wiley and Sons, Inc., New York, New York, 2 volumes. Jones, C. A. 1996. Mammals of the James John and Lake Dorothey State Wildlife Areas. Final Report, submitted to the Colorado Division of Wildlife and Colorado Natural Areas Program. Krutzsch, P.H. 1954. North American jumping mice (genus Zapus). University of Kansas Publications, Museum of Natural History 7:349-472. •• Levins, R 1970. Extinction. Lectures in Mathematical Life Sciences 2:75-107. Long, C. A. 1965. The mammals of Wyoming. University of Kansas Publications, Museum of Natural History, 14:493-758. Meaney, C. A., N. W. Clippinger, A. Deans, and M. OShea-Stone. 1996. Second year survey for Preble's meadow jumping mouse (Zapus hudsonius preblei) in Colorado. Report prepared for the Colorado Division of Wildlife. Meaney, C. A., A. Deans, N. W. Clippinger, M. Rider, N. Daly, and M. O'Shea-Stone. 1997. Third year survey for Preble's meadow jumping mouse (Zapus hudsonius preblei) in Colorado. Report prepared for the Colorado Division of Wildlife. Nudds, T. D. 1977. Quantifying the vegetative structure of wildlife cover. Wildlife Society Bulletin 5:113-117. Olson, T. E., and F. L. Knopf 1988. Patterns ofrelative diversity within riparian small mammal communities, Platte River watershed, Colorado. Pg. 379-386 in: Proceedings of the symposium: Management of amphibians, reptiles and small mammals in North America Flagstaff, Arizona U.S. Forest Service, General Technical Report RM-166. PTI Environmental Services. 1996a Preble's Meadow Jumping Mouse Study at Rocky Flats Environmental Technology Site, Annual Report 1996. Final. Rocky Flats Environmental Technology Site, Golden, Colorado. • Quimby, D. C. 1951. The life history and ecology of the jumping mouse, Zapus hudsonius. Ecological Monographs 21:61-95. Riggs, L. A., J. M. Dempey, and C. Orrego. 1997. Evaluating distinctness and evolutionary significance of Preble's meadow jumping mouse: Phylogeography of mitochondrial DNA noncoding region variation. Final Report for the Colorado Division of Wildlife. Denver, Colorado. Ryon, T. R 1996. Evaluation of historical capture sites of the Preble's meadow jumping mouse in Colorado, final report. MS Thesis. University of Denver, Denver, Colorado. Shenk, T. M. 1998. Conservation assessment and preliminary conservation strategy for Preble's meadow jumping mouse (Zapus hudsonius preblei). Colorado Division of Wildlife FY 1997-98 Annual Report. Shenk, T. M. and M. M. Sivert. 1999. Temporal and spatial variation in the demography of Preble's meadow jumping mouse (Zapus hudsonius preblei). Colorado Division of Wildlife JanuaryMarch 1999 Quarterly Report. Shenk, T. M. and Sivert M. 1999. Movement patterns of Preble's meadow jumping mouse (Zapus hudsonius preblei) as they vary across time and space. Colorado Division of Wildlife January March 1999 Quarterly Report. Stohlgren,T. J., M. B. Falkner, and L. D. Schell. 1995. A Modified-Whittaker nested vegetation sampling method. Vegetatio 117:113-121. USFWS. 1997a Proposal to list the Preble's meadow jumping mouse as an endangered species. USFWS 50 CFR part 17. USFWS. 1997b. Interim survey guidelines for Preble's meadow jumping mouse. USFWS. Denver, Colorado. Whitaker, J. 0., Jr. 1963. A study of the meadow jumping mouse, Zapus hudsonius (Zimmerman), in cental New York. Ecological Monographs 33:3. Whitaker, J, 0., Jr. -1972. Zapus-.,hudsonius. M~malian Species.1.1:l-I .. _,. �36 Table I. Survey site locations and trapping results for presence ofjumping mice (Zapus spp.) in Larimer and Weld Counties, Colorado. Positive identification to species is pending. Number Weights (grams) General Description ofLocation 3 total 3 females 21, 23.5 unnamed drainage to north of Arthur's Rock drainage 7480 feet 13 total 6 females 7males 20, 21.5, 28, 22, 26.6, 19, 23, 22, 25, 20.5, 23, 22, 23 west of Prairie Divide Road E485875 N4498154 5120 feet 2 total I male I female 18.5, 19 near Watson Lake, northwest of Laporte 7nt987/9/98 E467500 N4502140 6320 feet 12 total 8 females 4 males 21.5,25, 19,20, 24, 21, 19, 21, 23.5, 21, 23 drainage runs along west side of Stove Prairie Road into the Poudre River Meadow Creek 7/10/987/13/98 E467149 N4524532 6680 feet 3 total 3 females 17,22.5,25.5 along Cherokee Park Road Young Gulch 7/12/987/15/98 E470750 N4502380 E470650 N4503800 6240 feet I total I female 2 total 2 female 22 tributary of Poudre River Dates trapped U1M coordinates Elevation Arthur's Rock 6/23/986/26/98 E485250 N4490700 5520 feet Bull Creek 6/28/987/1/98 E458594 N4525510 Poudre River 7ll/987/10/98 Skin Gulch Drainage ofZapus spp.found 5840 feet 33, 23.5 Sevenmile Creek 7/12/987/15/98 E450611 N4505834 7000 feet 2 total I female 1 male 24, 24.8 tributary of Poudre River N. Fork Poudre River 7/207/22/98 E468100 N 4527485 6440 feet 2 total I female I male 20, 22.5 Cherokee Park, drainage above Halligan Reservoir Pendergrass Creek 7/267/29/98 E461380 N 4501300 7000 feet I total I male 28 tributary ofS. Fork of the Poudre River Buck Gulch 8/3-4/98 E463575 N4502360 6400 feet 2 total 2 females 10, 23.5 tributary of Poudre River just east of Big Narrows Lakey Canyon 8/3-4/98 E466550 N 4491360 7120 feet ~ total 28.5, 22, 30, 13 tributary of Buckhorn Creek Little Bear Gulch 8/5-7/98 E 475650 N 4483745 6600 feet 2 total 2 female 29, 16 Buckhorn Mtn. area Bear Gulch 8/6-7/98 E 472500 N 4483550 7200 feet 1 total I female 19.5 tributary of Buckhorn Creek N. Fork, Big Thompson 8/9-11/98 E 462950 N 4478090 7080 feet 1 total 1 female 24.5 by Glen Haven N. Fork Poudre River 8/9-11/98 E 478692 N4516030 5900 feet 7 total 3 female 3 male _l escaped 22, 17.5, 17.5, 18, 17, 28, Robert's property 3 females 1 male �37 Drainage Dates trapped U1M Elevation Number ofZapus spp.found Weights (grams) General Description of Location coordinates North Poudre Canal 8/17-18/98 E478000 N 4518600 6000 feet 4 total 4 male 23, I 9, 13, 18 irrigation ditch and pond - Knight's property North Fork ofPoudre 8/18-19/98 E472935 N 4523400 6100 feet I total I female 28.5 Phantom Canyon Little Thompson River 8/19-20/98 E468500 N 4459400 6600 feet I total I male 17.5 northwest of Pinewood Sprngs, off Hwy 36 Cedar Creek 8/21-22/98 E 477097 N 4477899 6080 feet 5 total 3 female 20, 28, 18.5, 33, 15 2male Sheep Creek 8/21-22/98 E470451 N 4490818 6700 feet I total I male 18.5 tributary of Buckhorn Creek Fish Creek 9/9-10/98 E462800 N 4537100 7400 feet I total I female 18.5 tributary of Fish Creek 6/2-6/5/98 E 547250 N 4522450 5160 feet 0 E 502350 N 4506750 5180 feet drainage Geary Creek Windsor Ditch 6n-6n0t98 Geary Creek, Pawnee National Grasslands 0 Windsor ditch, Cobb LakeSWA Willow Creek · 6/116/14/98 E 537260 N 4521900 5200 feet Willow Creek 6/126/15/98 E 545310 N 4516687 5060 feet 0 Willow Creek, Pawnee National Grasslands Crow Creek 6/166/19/98 E 555711 N 4499260 4820 feet 0 Crow Creek, Pawnee National Grasslands Owl Creek 6/236/26/98 E 522550 N 4523050 5450 feet 0 Owl Creek, Pawnee National Grasslands Soldier Canyon 6/28-7/1/98 E 484350 N 4492950 5520-5600 feet 0 Soldier Canyon drainage, Lory State 0 Willow Creek, Pawnee National Grasslands Park Seaman Reservoir 7/237/26/98 E480I00 N 4505680 5400 feet 0 effiuent of Seaman Reservoir, North Fork of the Poudre River South Fork Rabbit Creek 8/248/27/98 E 470150 N 4515100 6334 feet 0 South Fork Rabbit Creek, Cherokee Park Banner Lakes 8/248/26/98 E 537850 N4435190 5030 feet 0 Banner Lakes SWA Big Thompson River 8/27-30/98 E497210 N 4470130 4870 feet 0 Big Thompson River at Simpson Ponds SWA South Side Ditch 9/1-4/98 E 484350 N 4473690 5100 feet 0 South Side Ditch, Big . ThompsoJJ. Canyon SWA �38 Dates trapped UfM Big Thompson River 9/1-4/98 Long Gulch Elevation Number ofZapus spp.found E467150 N4473120 6800 feet 0 Big Thompson River 9/1-9/4/98 E465850 N4473450 7100 feet 0 Long Gulch, Big Thompson drainage Lone Tree Creek 9/9-9/12/98 E506850 N4536000 E 506320 N 4535400 5900 feet 0 Long Tree Creek, Meadow Springs Ranch Bull Creek 9/129/15/98 E462150 N 4534150 7600 feet 0 Bull Creek South Fork Poudre River 9/139/16/98 E462250 N4503650 6400 feet 0 South Fork Poudre River Poudre River 9/139/16/98 E470955 N4504350 5800 feet 0 Poudre River Drainage coordinates Weights (grams) General Description ofLocation �•Table 2. Numbers of other small mammal captures at the 39 sites surveyed for Preble's meadow jumping mouse in Larimer and Weld Counties2 Colorado. S2ecies codes are listed below the table. Mumu Other Drainage PMJM Sos2 Sxs2 Tase Tahu Chhi Rese Perna Petr Pese Neme Mise • Arthur's Rock : Bull Creek • Poudre River at Watson Lake • Skin Gulch Meadow Creek Youn Gulch Seven Mile Creek North Fork Poudre River • •Pendergrass Creek Buck Gulch Lake Can on : Little Bear Gulch •. Bear Gulch : North Fork Big ThomQson • North Fork of Poudre River North Poudre Canal North Fork of Poudre River • Little ThomQson River Cedar Creek She Creek •• Fish Creek drainage Ge Creek Windsor Ditch • Willow Creek Willow Creek Crow Creek Owl Creek .•. Soldier Can on Seamen Reservoir South Fork Rabbit Creek . Banner Lakes 2 3 13 2 12 3 3 2 2 1 2 4 2 27 1 34 5 43 25 3 1 22 6 5 3 106 113 114 11 5 14 15 46 15 6 23 2 8 11 5 8 8 3 2 7 4 4 5 1 9 22 2 2 1 5 1 2 0 0 0 0 0 0 43 14 8 0 9 3 65 15 5 2 5 15 3 37 16 1 6 g 0 0 1 79 2 7 2 0 1 2 4 3 ,. 1 2 6 2 8 8 2 30 174 12 211 176 93 13 110 32 39 28 166 4 3 1 1 13 w. 6 '° �~ 0 PMJM Drainage Big Thompson River • South Side Ditch Big Thompson River Lon Gulch Lone Tree Creek 0 0 0 0 0 Bull Creek ., South Fork Poudre River 0 0 Poudre River Sosp Sysp Tasp Tahu 2 Chhi Resp Mumu IO 25 59 8 13 11 104 67 5 22 14 4 0 :' Sosp =Sorex spp. • Chhi =Clzaetodipus lzispidus Petr =Peromyscus troei ·: Mumu =Mus musculus Neme 16 31 1 1 3 ·' Pesp 28 12 186 83 174 1 4 Perna Petr 6 3 Misp Other 21 81 9 72 Tasp =Tamiasciurus hudsonicus Perna =Peromyscus maniculatus Neme= Neotoma mexicana Sysp =Sylvilagus spp. Resp =Reithrodontomys spp. Pesp =Peromyscus spp. Misp =Microtis spp. •. Table 3. Percent contents of fecal samples collected from traps where Preble's meadow jumping mice were captured. All samples were · collected during the 1998 field season. Amount of bait in each sample was quantified as 100%, 'A' for abundant(> 70% of the sample), T for trace amount (< 10% of the sample), or 0% of the sample was bait. endogenous Date Salix seed moss pollen Equisetum flower Salsola Location bait arthropods fungus PIT-tag ID# 41412PH3D : 4 I405A3326 41405C68l8 41414A5459 Juvenile Juvenile 41413A7757 4141310654 Juvenile Seven Mile Creek Seven Mile Creek North Fork Poudre North Fork Poudre Sheep Creek Buck Gulch Buck Gulch Bear Gulch Fish Creek July 14 July 18 July21 Aug 11 Aug22 Aug4 Aug4 Aug7 Sept 10 T T A T A A T A 0 2 89 11 53 9 89 41 6 78 91 15 100 T 7 9 7 7 100 79 7 �41 Table 4. Mean stream width and dominant understory, shrub, and overstory vegetation at each of the sites surveyed for Preble's meadow jumping in Larimer and Weld County, Colorado during 1998. Mean Stream wid~(m) Drainage:!: PMJM X Dominant Overstory se(X) Grass/Forb Shrub Tree Bluegrass/Horsetail Bluegrass AltaFescue Bluegrass/Canada Thistle Bluegrass/Canada thistle Wild Plum Willow Willow Snowbeny/ Chock Cherry Elm Mountain Alder Cottonwood Snowbeny Willow Wax.current Mountain alder Arthur's Rock Bull Creek Poudre R. at Watson Lake y y y 0.77 1.75 24.47 1.01 0.2 0.54 Skin Gulch y 0.28 5.15 Meadow Creek l.77 0.97 Young Gulch Seven Mile Creek y y y 1.22 3.38 1.06 1.14 North Fork Poudre River Pendergrass Creek Buck Gulch Lakey Canyon Little Bear Gulch Bear Gulch North Fork Big Thompson y y y y y y y 8.30 l.72 0.52 0.84 0.46 0.53 6.93 3.03 0.53 0.26 0.18 0.26 1.9 l.73 North Fork of Poudre River y y 12.50 11.97 6.97 y y y y y 8.30 2.93 2.07 2.58 0.49 1.43 1.82 1.1 0.4 2.12 0.00 7.07 9.20 0.91 5.74 2.41 Willow Creek Crow Creek N N N N N l.78 0.89 Owl Creek N 8.00 3.37 2.40 Wheatgrass Wheatgrass Canada Thistle Canada Thistle 0.39 Cone Flower North Poudre Canal North Fork of Poudre River Little Thompson River Cedar Creek Sheep Creek Fish Creek drainage Geary Creek Windsor Ditch Willow Creek Soldier Canyon Seamen Reservoir South Fork Rabbit Creek Banner Lakes Big Thompson River South Side Ditch Big Thompson River Long Gulch Lone Tree Creek N N N 1.20 9.87 0.52 5.20 10.27 14.17 15.53 1.18 3.84 0.20 8.53 1.24 0.42 0.79 2.41 Bluegrass Bluegrass Bluegrass/ wheatgrass Columbine Black-eyed susan Aster Bluegrass Bluegrass/ yarrow Bluegrass Bluegrass/Canada thistle Bluegrass Bluegrass/ Canada thistle Canada thistle Bluegrass Blue grama Canada thistle Wheatgrass Bluegrass/ Dandelion Cheatgrass Wheatgrass Canary Reed Grass Canary Reed Grass Canary Reed Grass Canada Thistle Horsetail Yarrow Cottonwood Ponderosa pine Juniper Willow Mountain alder Snowbeny Aspen/ maple Wax.current Juniper Wild rose/ wild grape Spruce Wild raspberry Aspen Willow Mountain alder Snowberry Mountain alder Willow Willow Cottonwood Cottonwood Snowbeny Willow Willow Wild rose Willow Mountain alder Cottonwood Ponderosa pine Mountain alder Ponderosa pine Cottonwood (Absent) Willow (Absent) Wax.Current Wax.Current Willow Willow Willow Willow Willow (Absent) (Absent) (Absent) Elm Russian Olive Cottonwood Ponderosa pine Mountain Alder Russian Olive Russian Olive Cottonwood Mountain Alder Ponderosa Pine Willow N 0.2 N 6.57 Snowbeny Willow N 0.38 N 0.25 Willow N Willow l.86 Snowberry N 0.1 Wild Rose Bull Creek N Juniper 1.18 Goosebcny South Fork Poudre River N ERR Golden Banner Snowbeny/ Wild Rose Ponderosa Pine Poudrc River N 20.00 Common Lupine Choke Cherry Juniwr i Drainages ar~ list~d in the same order ~s i~ Table 1 so that exa~t locations of each site could be obtained by using both tables. .. . ._ . .··.. �• II ■ l'=,-b'--tf-r----~~~>c-f'-J art ~ollins~ 1---1L • ■ Zapus spp. Found Site Trapped No Zapus spp. Found D County Line Nstreams /V Roads Figure 1. 1998 survey site locations and trapping results for presence of Preble's meadow jumping mouse (PMJM) in Larimer and Weld Counties, Colorado. Locations where PMJM were found are indicated by a red circle, locations where PMJM were not found are indicated by a green circle. �43 APPENDIXC TEMPORAL AND SPATIAL VARIATION IN THE DEMOGRAPHY OF PREBLE'S MEADOW JUMPING MOUSE (Zapus hudsonius preblei) Tanya M. Shenk and Maile M. Sivert Abstract A total of 186 individual mice were captured and PIT-tagged at three study sites (73 mice at Maytag Property, 77 mice at PineCliffRanch, and 36 mice at Woodhouse Ranch). These mice were captured over three different trapping sessions with an effort of 17,330 trap-nights. Genetic tissue samples were collected for at least 30 individual mice at each of the three study sites. Density estimates were expected to increase as the summer progressed to account for the birth pulses in late June, and late July-August. This was the trend observed· at Maytag. PMJM densities at Woodhouse Ranch increased from June to July but did not increase further during the September trapping session. Mean PMJM density over all sites and sessions was 40.5 (range 16.9-79.0) mice per kilometer of stream stretch. Highest densities occurred at PineCliff, where the vegetation is primarily willow and both a mainstem and tributary were used by mice. Lowest densities occurred at Woodhouse Ranch where the riparian vegetation has fewer willow but was dense with other riparian vegetation. Vegetation at Maytag provided some areas of dense willow with the remaining areas being of moderate density of riparian vegetation. The lower density of PMJM reported for Woodhouse Ranch might be explained by the composition and densities of other small mammals at that site. Woodhouse Ranch had the highest captures of both house mice and voles of the three sites studied. Defining summer as June 1 - October 5, over-summer survival was estimated as 0.36 (se = 0.056) over all three study sites. Temporary emigration and immigration rates were estimated but both had extremely high variances associated with those estimates. Thus, these estimates cannot be used, with any confidence, to provide information on movements of mice into and out of these populations. Very little new information was gained during this study on reproductive parameters of PMJM. Most adult mice captured exhibited evidence of active reproductive behavior, either pregnancy, lactation, or enlarged genitalia However, we were unable to locate any breeding nests. Juvenile mice were not captured during the June trapping session. Juvenile mice were captured at all three sites during both the July and September trapping sess10ns. This report includes results from only the first year of a multi-year project to follow individually marked PMJM through time. Collection of more data, and more years of data, will improve our ability to evaluate demographic parameters and estimates of how they vary across space and time. Introduction On May 12, 1998 the U.S. Fish and Wildlife Service (USFWS) published a final rule in the Federal Register (63 FR 26517) to list Preble's meadow jumping mouse (Zapus hudsonius preblei) as 'threatened' under the Federal Endangered Species Act (ESA) of 1973, as amended. Recovery goals for Preble's meadow jumping mouse (PMJM) should work towards the sustainability, protection, and restoration of Z. h. preblei populations and habitats on both private and public lands to provide the spatial, genetic, and demographic structure needed to promote long-term species viability and provide species management flexibility. Recovery efforts for the subspecies will be most effective if reliable information is available on the basic ecology of the subspecies and this information used to design recovery efforts such as Habitat Conservation Plans. A review of studies conducted on PMJM shows that there is insufficient information to fully address defining range-wide ecological requirements, .- . limiti~g _factors, .l.imits. of speci~s toler:µice, _oi:_popula~ion st~Ws (Shenk 199_8) .. Most work to date _has �44 focused on geographic distribution (presence or absence of Z. h. preblei), taxonomy, and habitat descriptions of sites where mice have and have not been captured. For PMJM in particular, information on dispersal, habitat use, and population dynamics is most needed to identify minimal ecological requirements of the subspecies. Monitoring, by way of estimating demographic parameters (such as survival, abundance, and reproduction), of a single population over time provides an opportunity to document demography of a species, estimate temporal fluctuations in demography, and gain insights into the temporal variation inherent in demographic parameters. Monitoring of multiple populations over time and over multiple geographic locations provides the opportunity to gain further understanding of population processes and detaches time effects from spatial effects. Population monitoring activities do not constitute scientific experiments, in the spirit of manipulation of salient ecological variables, however, replication of monitoring activities for natural populations over long periods of time and in diverse geographic locations can lead to insights into population processes (Cook and Campbell 1979). These insights can then be translated into hypotheses useful for predicting changes in population demography resulting from either natural perturbations (e.g., flooding events) or anthropogenic modifications (e.g., gravel mining). Experimentation would then be required to test these hypotheses and establish cause and effect. The primary objective of this study is to investigate spatial and temporal variation in the demography of PMJM. Demographic parameters to be estimated include survival, reproduction, temporary emigration, immigration, population structure, and density. These demographic parameters will be estimated from individually marked animals from geographically distinct populations. To evaluate spatial differences in the demography of PMJM, populations were selected from three sites that provided a variety of habitat matrices available to the mouse. Multiple years of conducting the study at those same sites will provide an estimate of the temporal variation in demography of PMJM. Study sites Demography of PMJM was evaluated for three populations. All three populations selected were located in areas where PMJM had previously been found. To evaluate spatial differences in the demography of PMJM, populations were selected on sites that provided a variety of habitat matrices available to the mouse. The first site selected, Colorado Division of Wildlife (CDOW) Maytag Property, has one primary water source available to the mouse. This water source is East Plum Creek. The second site, Colorado Open Lands PineCliffRanch, provides both a tributary (Garber Creek) and a main stem drainage (West Plum Creek). The third study site, CDOW Woodhouse Ranch provides an area containing a tributary (Indian Creek) and a series of ponds and irrigation ditches scattered throughout the property. Objectives Objectives of the demography study are to: 1. Estimate abundance and density of PMJM for each of three study populations. 2. Estimate over-summer, over-winter, and annual survival of PMJM at each of three study populations. 3. Estimate temporary emigration of PMJM at each of three study populations. 4. Estimate immigration of marked PMJM back into each of three study populations. 5. Estimate reproduction of PMJM at-each of three study populations. 6. Evaluate the affect of weight, sex, age, abundance (i.e., density dependent response), and habitat features such as stream reach, vegetation composition and density on survival, reproduction, abundance, temporary emigration, and immigration of marked animals back into three study populations of PMJM. .7, J;stimate _1;1ge ~d.se~ ratios 9f PMJM at e~Gh of three stµdy pop4lati.ons .. .. . . .- . ~ . . . .. �45 Met/,ods Trapping Three trapping sessions were conducted during the following weeks: June 2-9, 1998; July 21August 5, 1998; and September 8-15, 1998. Trappers were advised to follow the Center for Disease Control's Hantavirus instructions and recommendations when dealing with rodents. Due to the nocturnal nature of PMJM, traps were set between 19:00hrs an<l'21:00hrs and checked as early as possible in the morning beginning at 5:00hrs to reduce stress and the potential for predation on trapped animals. Time required to complete the traplines varied depending on how many animals were caught. Small mammal Sherman live traps (folding and non-folding) were used to conduct the trapping sessions. Traps were set in two parallel lines of trap stations (1 trap per station) on either side of the drainage. Trap stations were 5 meters apart for a total of 250 meters; the parallel transects were 10 meters apart unless extent of habitat, terrain topography, or stream hydrology did not allow. Location of transects were recorded on field data sheets. Each trap location was also recorded to the nearest 5 meters in UTM coordinates using a Trimble Geo-Explorer GPS. A small (~1 inch) ball of polyester quilt was placed in each trap as nesting/bedding material. Baiting material was Manna Pro Sweet 3-way Livestock feed which contains no animal matter; ingredients include flaked barley, flaked com, flaked oats, and cane molasses. Peanut butter was used to stick the bait to the trap. Traps were checked by two surveyors. All animal captures were recorded. If an animal had been captured in a trap, a ziplock plastic bag was placed over the end of the trap .. The trap was opened allowing the animal to fall into the plastic bag. The animal was identified to species while in the plastic bag. If the animal was not a PMJM, identification of the animal was recorded and the animal set free without further handling. Each captured PMJM was checked to detect the presence/absence of a PITtag and/or radio-collar. If a PIT-tag or radio-collar was detected and the mouse had been captured at that same site within the same trapping session, identification of the mouse was recorded and the mouse was released. If the animal. was a PMJM and no PIT-tag or radio collar was detected the PMJM was anesthetized for further processing, as follows. All PMJM were PIT-tagged. If the PMJM weighed> 18 g a radio-collar was also put on the animal for movement data collection (see Shenk and Sivert 1999). Each PMJM was weighed while in the bag, recording the weight in grams using a Pesola spring balance. Sex, age (juvenile, adult), and reproductive condition (pregnant, lactating or nonbreeding if it was female; for males the position of the testes indicated if in breeding status or nonbreeding status) was noted. Each PMJM was measured (total length, length of body, length of hind foot - heel to distal end of claws) in millimeters. Capture of every PMJM was documented by taking a photograph of the mouse on first capture. Mice were placed in a plexiglass photobox to reduce handling stress to take a photograph. If the mouse had been captured previously as noted by the detection of a PIT-tag, no photograph was taken. If there were feces in the trap where a PMJM was captured, the fecal material was collected in a plastic bag labeled with date, location of the site, and animal PIT-tag number. Fecal samples were kept cool until returned to the CDOW office where they were frozen. Salt was added to each specimen bag for preservation. Once the field season was over all fecal samples were analyzed by the Composition Analysis Laboratory, Inc, 622 ½ Whedbee, Fort Collins, CO for content. These analyses were for a complementary study on food habits and movements of PMJM at the same three study areas (see Shenk and Sivert 1999 for details). All trap mortalities were recorded on a 'Trap Mortality' data form which included information on species, potential duration of time spent in the trap, and any information available to help determine cause of death. All animals found dead were double-bagged in a plastic bag and placed in a cooler with ice. Specimens were frozen as soon as possible and deposited in the CDOW freezer at the office in.Fort Collins .. A museum card was.. completed and attached-to each PMJMspecimen.and the. �46 specimen and card were given to Cheri Jones, Curator of Mammalogy at the Denver Museum of Natural History, for study skins and tissue storage. If an animal was severely injured (e.g., severed limb, large lacerations) it was euthanized by soaking a cotton ball in Metofane (methoxyflurane) and placing the cotton ball and the mouse in a ziplock bag until the animal stopped breathing. If an animal appeared to be only slightly injured (e.g., broken tail, small laceration) the animal was released to the wild.' If an animal appeared to be coldstressed attempts were made to warm it by holding it in the surveyors hands and/or against their body. If the animal appeared to be heat stressed, isopropyl alcohol was applied with a cotton swab to the ears, arm pits, and feet to cool it down. During the last trapping session homeopathic first aid treatments were used in an attempt to minimize stress and improve recovery of each PMJM. Homeopathic first aid kits and instruction was provided by WildAgain Wildlife Rehabilitation, Evergreen, Colorado. PIT-tagging Each PMJM was individually marked with a Passive Integrated Transponder (PIT-tag). PITtags are electromagnetic, glass-encased tags that emit a passive signal (125 kHz) that can be decoded by a portable reader. Destron-Fearing PIT-tags were used on all mice. We used portable readers from · Biomark to read the PIT-tags. All mice were anesthetized to PIT-tag them. Anesthesia procedures followed protocols successfully used on PMJM before (R. Schorr personal communication). Mice were anesthetized by placing them and a cotton ball with 1 ml ofMetofane (methoxyflurane) into a sealed ziplock bag (to keep Metofane fumes in bag). The bag was then lain aside to minimize stress for the mouse and the time the mouse struggles in the bag. The less agitated the mouse is in the bag the less time required for the metofane to take affect. The mouse was observed at all times to make sure the mouse did not position itself such that the cotton ball was in direct contact with its face, which might cause the mouse to have a severe reaction to the anesthesia After the animal stopped moving, the handlers waited one minute before removing the mouse from the bag. The animal typically remained anesthetized for 2-3 minutes once outside the bag. Each PMJM was PIT-tagged using a protocol successfully used by C. Meaney (personal communication). Each PMJM had a PIT-tag inserted above the shoulder blades by lifting the skin on its back and inserting the individually sterilized needle with the PIT-tag under their skin and injecting the tag. Verification of the PIT-tag identification number was made before and after insertion into the mouse by running the PIT-tag scanner across it. The skin behind the opening was then pinched to prevent emergence of the tag. PIT-tag identification number was recorded on the field data form. If, at any time during the handling of the PMJM the animal appeared to be severely stressed (dramatic changes in heartbeat, respiration and responsiveness or gums turning blue) first aid was administered and, once recovered, released without further processing. Genetic tissue sampling Genetic tissue samples were collected from the first 30 PMJM captured at each study site. The following protocol was used based on previous success with PMJM (M. Bakeman, C. Meaney, T. Ryon, R. Schorr, personal communication). A fresh pair of clean latex gloves were used to handle each mouse. With a clean ear punch tool, one tissue plug was obtained from each mouse ear. If there was excessive bleeding, gentle pressure was applied to the injured area for approximately one minute. Each ear plug was placed in a vial of 95% ethanol. Both the vial and the corresponding record on the data sheet were labeled with a unique identifier as follows. A seven-place alpha-numeric code was composed as follows: a three-letter designator for the survey location (e.g., MYT = Maytag Property); a number beginning with two digits •. . . -indicating the year-.(e.g., 98); followed by 2- digits specifying the individual trapped, numbered . . . . -~~quentt<J:lly,. fot e.~ample: MYT_980J ;= fir~t ari~al sampled,at Maytag ProP.~rty)~ J9~_8. �47 After returning from the field, all samples were put in a cool place or refrigerated until delivery to the CDOW office at 317 West Prospect, Fort Collins, CO where they are being held in the freezer for future analyses. Ear punch tools were cleaned after each use by immersing them in a 10% bleach solution for a few minutes, rinsing them thoroughly with clean water, and then dried thoroughly to prevent rusting. Dull ear punch were discarded and replaced with new sharp punches to ensure the quickest, cleanest cut possible. ' Demographic parameter estimation Mark-recapture estimation techniques were used to estimate abundance, over-summer survival, over-winter survival, temporary emigration, and immigration of marked animals back in to the three study populations of PMJM. Abundance: Abundance was estimated using Pollock's Robust Design (see Kendall et al. 1997, 1995 and Kendall and Nichols 1995 for detail) in Program MARK (White and Burnham 1999). The robust design is a combination of the Cormack-Jolly-Seber (CJS)(Cormack 1964, Jolly 1965, Seber 1965) live recapture model and the closed capture models. The key difference from the CJS model is that instead of just one capture occasion between survival intervals multiple capture occasions are used. These occasions are close together in time allowing the assumption that no mortality or emigration occurs during these short time intervals. The closely spaced encounter occasions are termed "trapping sessions" and each trapping session is viewed as a closed capture survey whereby abundance can be estimated. Three 7-day trapping sessions were conducted during the following weeks: June 2-9, 1998; July 21-28, 1998; and September 8-15, 1998. Abundance estimates were calculated for each trapping session. Survival: By using the estimate of the probability that an animal is captured at least once from the trapping sessions designed to estimate abundance, survival between the longer intervals was estimated. Analyses were conducted to estimate over-summer survival, and survival from June-July and August-September. A fourth 7-day trapping session to be conducted June 1-8, 1999 will provide information to estimate over-hibernation and annual survival. Temporary emigration and immigration: The longer intervals between trapping sessions also allows estimation of temporary emigration from the trapping area, and immigration of marked animals back to the trapping area using Pollock's Robust Design. Reproduction: Reproductive parameters were estimated by evaluating the reproductive status and age of captured PMJM. Results Trapping and PIT-tagging Effort A total of 186 tndividual PMJM were captured from the three study areas during the three 1998 trapping sessions ·(Table 1: 73 at Maytag Property, 77 at PineCliff Ranch, and 36 at Woodhouse Ranch). These mice were captured over three different trapping sessions with an effort of 17,330 trapnights. Every PMJM captured was PIT-tagged, providing each mouse with a unique, permanent, lifetime identification marker. Other small mammals captured during the trapping sessions included Peromyscus spp., Mexican wood rat (Neotoma mexicana), house mouse (Mus musculus), vole (Microtis spp.), western harvest mouse (Reithrodontomys megalotis), hispid pocket mouse (Chaetodipus hispidus), and shrew (Sorex spp.). House mice were captured at all three study areas during either June or September (Tables 2, 3). Species and relative numbers of captures within each of those species were similar for Maytag Property and PineCliffRanch (Tables 2,3). Species richness and relative number of captures between these sites were also similar in June and September with the exception of an increase in vole captures at both sites in September. Most of these common species were also captured at Woocihouse Rqnch during both June-::. and September 2,3). However, hispid pocket mice an_d.. . . . . . .. . ·... ·........ ··. .· . -~ ... •.. . . . . . ,-, . . (Taples . ' .. . ... : . ·. -· . . .- .. - ~- ... . . . . . . . ·• .., . : : ;.. �48 Mexican woodrats were also captured at Woodhouse Ranch and not at either of the other two sites. Numbers of house mice and vole captures at Woodhouse Ranch were greater than at either Maytag or PineCliff during June and September, with numbers of captures for both species increasing in the September trapping session (Tables 2,3). Age and Sex Ratios As expected, no juvenile PMJM were captured during the June trapping session as it was too soon after hibernation for any litters to have been born and/or if born the juveniles would still be restricted to their nests. Juvenile mice were captured during both the July and September trapping sessions at all three sites. Juveniles as small as I 0g were captured. Sex ratios may be biased towards males, however further analyses need to be completed to confirm this hypothesis (Table I). Genetic tissue sampling Genetic tissue samples were collected from 40 individual mice at Maytag, 30 mice at PineCliff Ranch, and 32 mice at Woodhouse Ranch. Additional samples were collected from four mice captured at a back drainage on the Woodhouse Ranch. Samples are currently being stored in the cooler at the CDOW Research Center in Fort Collins, Colorado. A Request for Proposals was submitted through the Colorado Department of Natural Resources. A committee is currently evaluating the three proposal that were submitted. Samples from these three populations will be used in studies to evaluate genetic uniqueness of PMJM populations. Abundance and Density Mark-recapture analysis methodologies were used to estimate abundance at each of the three study sites for each of the three trapping sessions (Table 4). Abundance estimates and length of the trapline were used to estimate an unadjusted density of mice per kilometer of stream stretch. Mice that typically don't use the area of stream stretch where the traps were placed could be trapped because they were attracted to the area by the artificial food source. Including these mice in the density estimates for the stream stretch covered by the trapline would artificially increase density estimates. Therefore, an adjustment was made to the density estimates. Locations of each individual mouse were classified as either within the stream stretch that was trapped or above or below the stream stretch where traps were placed. If greater than 50% oflocations of an individual mouse were off the trapline the mouse was categorized as an 'off-trapline' mouse; if less than or 50% of the locations were within the trapline the mouse was categorized as an 'on-trapline' mouse. The percent of off-line mice for each trapping session and each site was calculated and a mean adjustment parameter (p) estimated. This adjustment parameter is sensitive to the length of the trapline. The shorter the trapline the greater the adjustment would have to be. Regression models were run and AIC used to select the best regression model to estimate p by trapline length. These adjustment parameters are simply an estimated percent of mice that should be considered resident mice along the stream stretch where traps were placed. The density estimates were then decreased to percentage p. Use of the adjustment parameter provided a more realistic estimate of density of mice per kilometer of stream stretch. Without the adjustment densities would be inflated. •·.· ·•.·· ·:·_. -· Survival Survival rate was estimated using the mark-recapture estimator for Pollock's Robust Design in Program MARK. The best fitting model combined data from all three study sites over all three trapping sessions. Survival was estimated as 0.75 (se = 0.033) for a five week period. Extrapolating this survival rate across the summer results in an over-summer (June 1- October 5, 18 weeks) survival rate of 0.355 (se = 0.056). No estimates of over-winter survival can be made until trapping sessions .occur in S.ummer.19.99~-, ...,. . - . ·. . ;_-. <,_ ,.... • •__ .... , .... , . . ...•...,_ ... _: .• �49 Temporary emigration and immigration Temporary emigration rate and immigration was estimated using the mark-recapture estimator for Pollock's Robust Design in Program MARK. The best fitting model combined data from all three study sites over all three trapping sessions for these two parameters. Temporary emigration was estimated as 0.64 (se = 2.21), immigration was estimated as 0.45,, (se = 89.94). Reproduction No estimates of reproduction could be made. We located what we believed to be nest sites at several locations. We believe these sites to be nest locations with young because we followed radiocollared adult females (see Shenk and Sivert 1999) to these sites on numerous morning during breeding season. However, all sites were in very thick vegetation and we were not able to visually observe a nest with young. We felt it to be in the best interest of the mouse to not disturb the areas which might possibly cause failure of the nest. Discussion Information on the population dynamics of PMJM is necessary to determine which areas support viable populations. To begin to evaluate the viability of a population information on key demographic parameters must be obtained. Conducting studies on individually marked animals provides the greatest insight on the demography of a population. In general, estimated sex ratios from this study are comparable to those found by Armstrong et al. ( 1997) who reported an overall sex ratio for all captured PMJM of 51. 6 males: 48.4 females; approximately 86.0% of captures were identified as adults. There is a possible male sex bias at PineCliffRanch. However, small sample sizes and possible trapping biases by sex may explain the discrepancy. Density estimates were expected to increase as the summer progressed to account for the birth pulses in late June, and late July-August. This was the trend observed at Maytag. PMJM densities at Woodhouse Ranch increased from June to July but did not increase further during the September trapping session. Mean PMJM density over all sites and sessions was 40.5 (range 16.9-79.0) mice per kilometer of stream stretch. Highest densities occurred at PineCliff, where the vegetation is primarily willow and both a mainstem and tributary were used by mice. Lowest densities occurred at Woodhouse Ranch where the riparian vegetation has fewer willow but was dense with other riparian vegetation. Vegetation at Maytag provided some areas of dense willow with the remaining areas being of moderate density of riparian vegetation. The lower density of PMJM reported for Woodhouse Ranch might be explained by the composition and densities of other small mammals at that site. Woodhouse Ranch had the highest captures of both house mice and voles of the three sites studied. Defining summer as June 1 - October 5, over-summer survival was estimated as 0.36 (se = 0.056) over all three study sites. Meaney et al. (1999) report a one-month summer survival rate of 78%. Extrapolating Meaney et al.'s (1999) estimate over our summer period(~ four months) would result in a similar over-summer survival rate estimate of 36%. Prior to studies conducted in 1998 no information existed on survival rates for populations of Z. h. preblei although Whitaker (1963) reported a 67% loss of Z. hudsonius over hibernation. Temporary emigration and immigration rates were estimated but both had extremely high variances associated with those estimates. Thus these estimates cannot be used, with any confidence, to provide information on movements of mice into and out of these populations. Very little new information was gained during this study on reproductive parameters of PMJM. Most adult mice captured exhibited evidence of active reproductive behavior, either pregnancy, lactation, or enlarged genitalia. However, we were unable to locate any breeding nests. Juvenile mice were not captured during the June trapping session. Juvenile mice were captured at all three sites during .both the. Ju_ly .wd .S_eptember trappi1.1g sessions, .. Giv.e_n .t.h~t Jm~e.d.ing pea,k.s appe_~ to, 9.c_<;,UJ: to. .. . •. ·'···· -· •. . . . . . ·•. _.,, ... �50 early to mid-June and August with a possible third litter in September (Whitaker 1963) this was not unexpected and agrees with previous observations (Meaney et al. 1996, 1997, PTI 1996a, M. Bakeman unpublished data, T. Ryon unpublished data). This report includes results from only the first year of a multi-year project to follow individually marked PMJM through time. Collection of more data, and more_ years of data, will improve our ability to evaluate demographic parameters and estimates of how they vary across space and time. ... ·.· ..• Literatllre Cited Armstrong, D. M., M. E. Bakeman, A Deans, C. A Meaney, and T. R Ryon. 1997. Conclusions and recommendations in: Report on habitat findings on the Preble's meadow jumping mouse. Edited by M. E. Bakeman. Report to USFWS and Colorado Division of Wildlife. Bailey, B. 1929. Mammals of Sherburne County, Minnesota Journal ofMammalogy 10:153-164. Bailey, V. 1923. Mammals of the District of Columbia Proceedings of the Biological Society. Washington 36:103-138. Cook, T. D. and D. C. Campbell. 1979. Quasi-experimentation: design and analysis issues for field settings. Houghton-Mifflin, Boston. Cormack, RM. 1964. Estimates of survival from the sightings of marked animals. Biometrika 51:429-438. ERO Resources. 1995. Environmental review of South Boulder Creek Management Area Prepared for City of Boulder Real Estate/Open Space Department. Prepared by ERO Resources Crp., Denver, Colorado in association with Stoecker Ecological Consultants, Boulder, Colorado. Fitzgerald, J.P., C. A Meaney, and D. M. Armstrong. 1994. Mammals of Colorado. Denver Museum of Natural History, University Press of Colorado. Niwot, Colorado. Hafuer, D. J., K. E. Petersen, and T. L. Yates; 1981. Evolutionary relationships of jumping mice (genus Zapus) of the southwestern United States. Journal ofMammalogy 62:501-512. Hall, E. R 1981. The mammals ofNorth America John Wiley and Sons, Inc., New York, New York, 2 volumes. Hamilton, W. J., Jr. 1935. Habits of jumping mice. American Midland Naturalist 16:187-200. Jolly, G. M. 1965. Explicit estimates from capture-recapture data with both death and immigration stochastic model. Biometrika 52:225-247. Jones, C. A 1996. Mammals of the James John and Lake Dorothey State Wildlife Areas. Final Report, submitted to the Colorado Division of Wildlife and Colorado Natural Areas Program. Kendall, W. L., and J. D. Nichols. 1995. On the use of secondary capture-recapture samples to estimate temporary emigration and breeding proportions. Journal of Applied Statistics.22:751762. Kendall, W. L., K. H. Pollock, and C. Brownie. 1995. A likelihood-based approach to capturerecapture estimation of demographic parameters under the robust design. Biometrics 51 :293308. Kendall, W. L., J. D. Nichols, and J.E. Hines. 1997. Estimating temporary emigration using capturerecapture data with Pollock's robust design. Ecology 78:563-578. Krutzsch, P.H. 1954. North American jumping mice (genus Zapus). University of Kansas Publications, Museum of Natural History 7:349-472. Meaney, C. A, N. W. Clippinger, A Deans, and M. OShea-Stone. 1996. Second year survey for Preble's rrieadow jumping mouse (Zapus hudsonius preblei) in Colorado. Report prepared for the Colorado Division of Wildlife. Meaney, C. A, A Deans, N. W. Clippinger, M. Rider, N. Daly, and M. O'Shea-Stone. 1997. Third year survey for Preble's meadow jumping mouse (Zapus hudsonius preblei) in Colorado. . ._., .-. Report.pre,paredfor.theColorado.piyisiooofW.ildlife.... .. : . .., ,,. .- . . . ...._.. , .. .. . . .: . . . : · . ... ..•· . ·- .~ •. ~ ,• .. : �51 Meaney, C. A., A. Ruggles, B. Lubow, N. W. Clippinger, and A. Deans. 1999. Preliminary results: Second year study of the impact of trails on small mammals and population estimates for Preble's meadow jumping mice on City of Boulder Open Space. Report for Greenways Program, City of Boulder Transportation Department, City of Boulder Open Space, and Environmental Protection Agency, Region 8. . Nudds, T. D. 1977. Quantifying the vegetative structure ofwildllfe cover. Wildlife Society Bulletin 5:113-117. Poly, W. J., and C. E. Boucher. 1997. Record of a creek chub preying on a jumping mouse in Bruffey Creek, West Virginia Brimleyana 24: 29-32. PTI Environmental Services. 1996a Preble's Meadow Jumping Mouse Study at Rocky Flats Environmental Technology Site, Annual Report 1996. Final. Rocky Flats Environmental Technology Site, Golden, Colorado. PTI Environmental Services. 1996b. Preble's Meadow Jumping Mouse Study at Rocky Flats Environmental Technology Site, Spring 1996. Final. Rocky Flats Environmental Technology Site, Golden, Colorado. Quimby, D. C. 1951. The life history and ecology of the jumping mouse, Zapus hudsonius. Ecological Monographs 21:61-95. Riggs, L. A., J. M. Dempey, and C. Orrego. 1997. Evaluating distinctness and evolutionary significance of Preble's meadow jumping mouse: Phylogeography of mitochondrial DNA noncoding region variation. Final Report for the Colorado Division of Wildlife. Denver, Colorado. Seber, G. A. F. 1965. A note on the multiple recapture census. Biometrika 52:249-259. Sheldon, C. 1934. Studies on the life histories of Zapus and Napaeozapus in Nova Scotia Journal of Mammalogy 15:290-300. Shenk, T. M. 1998. Conservation assessment and preliminary conservation strategy for Preble's meadow jumping mouse (Zapus hudsonius preblei). Colorado Division of Wildlife FY199798 Annual Report. Shenk, T. M. and Sivert M. 1999. Movement patterns of Preble's meadow jumping mouse (Zapus hudsonius preblei) as they vary across time and space. Colorado Division of Wildlife January - March 1999 Quarterly Report. Stohlgren,T. J., M. B. Falkner, and L. D. Schell. 1995. A Modified-Whittaker nested vegetation sampling method. Vegetatio 117:113-121. USFWS. 1997. Interim survey guidelines for Preble's meadow jumping mouse. USFWS. Denver, Colorado. Whitaker, J. 0., Jr. 1963. A study of the meadow jumping mouse, Zapus hudsonius (Zimmerman), in cental New York. Ecological Monographs 33:3. . .-:- • • •· !.. • . . ~. .. . ·.• • . . -··: . .... .. ··. :· .. · ..... ·-· ···:· ·._: .... :- ... -.~ .· -. . . �52 Table 1. Number of Preble's meadow jumping mice PIT-tagged and captured at each of the three study sites for each of the three trapping sessions. Trapping effort is recorded as number of trap nights for each session at each site. # Capturedt # PIT-tagged Site Trapping session Trap nights Female Male Unknown Total Female(%) Male(%) Total Maytag Jun 1949 9 5 0 14 9 (64) 5 (36) 14 Maytag Jul 2370 13 13 0 26 15 (54) 13 (46) 28 Maytag Sep 3256 12 20 1 33 18 (43) 24 (57) 42 PineClitf Jun 1095 6 13 0 19 6 (32) 13 (68) 19 PineClitf Jul 1603 7 8 0 15 8 (50) 8 (50) 16 PineClitf Sep 1544 12 31 0 43 13 (28) 33(72) 46 Woodhouse Jun 1525 4 3 0 7 4 (57) 3 (43) 7 Woodhouse Jul 2420 8 5 0 13 9 (64) 5(36) 14 Woodhouse Sep 1568 6 9 1 16 8 (44) IO (56) 18 17330 77 107 2 186 90 114 204 TOTALS t Some captured animals were PIT-tagged during previous trapping sessions Table 2. Number of other small mammal species captured during the June 1998 trapping session at Maytag ProEerty, PineCliffRanch, and Woodhouse Ranch. Peromyscus Date Site seeMaytag 27 Jun2 42 Jun 3 Maytag Jun4 Maytag 48 Jun 5 Maytag 34 53 Jun 7 Maytag Jun 8 Maytag 39 Maytag 41 Jun 11 Maytag 45 Jun 12 _ 36 Jun 13 Maytag Jun2 Pine Cliff 36 Jun 3 Pine Cliff Jun 4 Pine Cliff 39 Jun 5 Pine Cliff 34 22 Jun 7 Pine Cliff Jun8 Pine Cliff 35 Woodhouse 13 Jun2 Woodhouse 23 Jun 3 Woodhouse 21 Jun4 Woodhouse 12 Jun 5 Woodhouse 13 Jun 7 Woodhouse 18 Jun 8 Jun 14 Woodhouse 16 • )u;i"15· •• Woodhouse ·s . ..~. ..... ;- .... -. •. ·-· '·.• Reithrodontomys Chaetodipus megalotis hispidus Microtis Harvest mice 4 1 0 2 2 2 0 0 6 3 0 0 0 I 4 o· • Hispid pocket mice 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 3 1 0 0 ff .... : . . spp. voles 0 0 0 2 4 0 Sorex spp. shrews 0 0 4 3 0 0 0 2 0 7 0 2 5 7 3 9 14 6 _.... ·. .. ••.. ··· 0 0 0 0 I 0 0 0 0 0 0 0 0 0 0 0 0 0 1 Mus Zapus Neatoma muscu/us hudsonius mexicana House preblei Mexican mice PMJM woodrat 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 IO 3 5 4 3 3 5 5 0 0 0 _....... 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 4 2 I 4 2 6 3 3 0 -.· .•...- ··- • •. () .• ::.- . . .~ '! • �53 Table 3. Number of other small mammal species captured during the September 1998 trapping session at Maytag ProEerty, PineCliffRanch, and Woodhouse Ranch. ReithroPero- dontomys Chaetodipus Mus myscus megalotis hispidus Microtis Sorex musculus ,·' Hispid house spp. harvest spp. spp. Date Site mice pocket mice voles shrews mice Sep9 Maytag 30 0 0 0 5 1 Maytag 39 Sep 10 0 0 4 0 1 Sep 11 Maytag 35 4 0 4 0 0 Sep 12 Maytag 35 6 0 8 0 0 31 Sep 13 Maytag 0 0 10 2 Sep 14 Maytag 30 3 0 9 2 0 Maytag Sep 15 36 2 0 17 2 2 Sep 16 Maytag 31 3 0 16 1 0 Sep9 Pinecliff 15 0 2 0 0 19 Sep 10 Pinecliff 0 0 4 0 Sep 11 Pinecliff 16 0 0 7 1 0 Sep 12 Pinecliff 21 0 0 7 0 0 Sep 13 Pinecliff 18 0 0 II 0 0 Sep 14 Pinecliff 17 0 0 8 0 0 Sep 15 Pinecliff 16 0 0 12 0 Sep 16 Pinecliff 22 0 0 16 1 0 Sep9 Woodhouse 15 0 1 47 2 7 Woodhouse 15 Sep 10 0 0 60 15 Sep 11 Woodhouse 18 70 0 0 21 Sep 12 Woodhouse 18 0 73 2 18 Sep 13 Woodhouse 13 0 67 0 13 Sep 14 Woodhouse 19 0 2 80 0 15 17 Sep 15 Woodhouse 0 3 73 2 14 Zapus hudsonius preb/ei PMJM 12 19 11 8 8 11 9 8 11 8 13 7 12 8 7 14 7 7 6 6 3 Neatoma mexicana Mexican woodrat 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 Table 4. Stream reach abundance (N) and density estimates for Preble's meadow jumping mouse (PMJM) from three sites in Douglas County, Colorado for three trapping sessions. Both adjusted and unadjusted density estimates (PMJM per km of stream stretch) are reported. Adjustments were made to the density estimates to account for the EOSitive bias introduced when animals are attracted to a traEline. Trapping session ii Maytag Jun 18 Maytag Jul 31 Maytag Sep 44 PineCliff Jun PineCliff Lower Upper (m) 2.7 15 21 550 32.7 0.80 26.0 11.4 20 42 608 51.0 0.81 41.4 1.8 42 46 494 89.l 0.78 69.3 30 5.7 24 36 490 61.2 0.78 47.5 Jul 17 0.9 16 18 504 33.7 0.78 26.3 PineCliff Sep 51 3.3 48 54 504 101.2 0.78 79.0 Woodhouse Jun 11 3.8 7 15 510 21.6 0.78 16.9 Woodhouse Jul 20 5.2 15 25 516 38.8 0.79 30.4 Woodhouse Se_Il ]7 2J ~74 J4.8 Q,8Q 20 . . se(N) :-.-·· __3.1 ... ·. Trapline Length Adjusted Density Unadjusted Density (PMJM/km) Site .... ; •·.- ... -_ .. 95% Confidence Interval p (PMJM/ km) 28 ....... -·- . �1 Colorado Division of Wildlife Wildlife Research Annual Report July 2000 JOB PROGRESS REPORT State of Colorado Cost Center 3430 Project No. W-153-R-13 Mammals Program Work Package No. --~0'""""6-"-6=-2_ _ _ _ __ Preble's Meadow Jumping Mouse Conservation - Task No. Develop Conservation Plan for Preble's Meadow Jumping Mouse 1 Period Covered: July 1, 1999 - June 30, 2000 Author: Tanya M. Shenk and Gary C. White --------------------------- ABSTRACT The second field season of a three year study on the demography and movement patterns of Prebles meadow jumping mouse (Zapus hudsonius preblei) was completed. A total of 175 individual mice were captured on stream-side transects at three study sites (60 mice at Maytag Property, 60 mice at PineCliff Ranch, and 55 mice at Woodhouse Ranch) during the 1999 field season. These captures included 34 mice PIT-tagged in 1998 (14 at Maytag Property, 12 at PineCli:ffRanch, and 8 at Woodhouse Ranch). These mice were captured over three different trapping sessions with an effort of 19,222 trap-nights. Most adult mice captured exhibited evidence of active reproductive behavior, either pregnancy, lactation, or enlarged genitalia. We were able to locate one potential maternal nest. Juvenile mice were not captured during the June trapping session. Juvenile mice were captured at all three sites during both the July and September trapping sessions. Natural mortality factors documented during both years include predation by house cats, garter snakes, rattlesnakes, yellow-bellied racers, bullfrogs, weasel, and fox as well as accidents by drowning and ro.µl kill. In 1999 mice were also found in torpor throughout the summer, often in very exposed areas frequently resulting in death by either exposure or predation. Preliminary analysis of the movement data collected on radio-collared PMJM in 1999 support results found in 1998. In particular we were able to document again in this second field season that PMJM exhibit (1) greater use of upland habitats than previously assumed, (2) general site :fidelity to both daytin1e nesting sites and nighttime feeding sites, (3) seasonal shifts in movement patterns, and (4) use of both perennial and intermittent tributaries adjacent to the capture drainage. Detail~ vegetation maps for all three study sites were created th:-ough intensive field mapping. These maps will be used in conjunction with the movement data to further refine habitat use information. Thirteen possible hibernacula were located in 1999. Tirree sites were located at Woodhouse Ranch and ten possible hibernacula at Pine Cliff Ranch. In contrast to 1998 data, these potential hibernation sites were located closer to the stream. However, at both Pine Cliff and Woodhouse Ranches stream banks rise out of the floodplain in closer proximity to the stream than at Maytag Property where the potential hibernation sites located in 1998 were located long distances from the stream center. Vegetation characteristics of the 21 sites located during both 1998 and 1999 are similar to �2 other hibernacula described for PMJM. By cooperating with other researchers PMJM densities (mice/km of stream) were estimated from nine study areas during June 1998 and June 1999, providing a total of 15 study area x year combinations. With these limited data available, 68% of the variation in PMJM density was explained by a model that includes riparian shrub and tree cover (ha/km stream). These results suggest that habitat quality of PMJM can be predicted by the riparian shrub and tree cover available on a site. Preliminary results from the demography, movement and distribution studies of PMJM have suggested the need to continue with research questions currently being addressed. However, the preliminary results also suggest further research be conducted on (1) use of upland and riparian habitat by PMJM, (2) refining water requirements of PMJM (i.e., do they require stream habitat or are wetland areas sufficient?), (3) range-wide distributional boundaries (e.g., elevation restrictions), and (4) investigating areas of potential syrnpatry or hibridization with the western jumping mouse (Z. princeps). The demography and movement studies were modified for the summer 2000 field season to further investigate upland use and water requirements, including a habitat modification experiment. This report includes results from the first two years of a multi-year project to follow individually marked PMJM through time. Collection of more data, and more years of data, will improve our ability to evaluate demographic parameters, movement patterns, and habitat use and evaluate how they vary across space and time. �3 DEVELOP CONSERVATION PLAN FOR PREBLE'S MEADOW JUMPING MOUSE Tanya M. Shenk an" Gary C. White P. N. OBJECTIVE Conservation of Preble's meadow jumping mouse (Zapus hudsonius preblei) in Colorado. SEGMENT OBJECTIVES FY99-00 1. Estimate abundance, over-summer survival, annual survival, and population age and sex structure of Preble's meadow jumping mouse at three study sites. 2.. Evaluate movement data of Preble's meadow jumping mouse as it relates to landscape features at three study sites. 3. Refine and prioritize needed research components to develop sound strategies for conservation of Preble's meadow jumping mouse. 4. Develop study plan to evaluate meta-population dynamics in Preble's meadow jumping mouse (to be funded in FY00-01). 5. Prepare a Federal Aid Job Progress Report. PERFORMANCE INDICATORS FY99-00 1. 2. 3. 4. Second year survival rate estimates for three populations of PMJM. Second year abundance and density estimates for three populations of PMJM. Second year movement patterns for three populations of PMJM. Study plan for PMJM meta-population study. INTRODUCTION On May 12, 1998 the U.S. Fish and Wildlife Service (USFWS) published a final rule in the Federal Register (63 FR 26517) to list Preble's meadow jumping mouse (Zapus hudsonius preblei) as 'threatened' under the Federal Endangered Species Act (ESA) of 1973, as amended. Recovery goals for Preble's meadow jumping mouse (PMJM) should work towards the sustainability;. protection, and restoration of Z. h. preblei populations and habitats on both private and public lands to provide the spatial, genetic, and demographic structure needed to promote long-term species viability and provide species management flexibility. Recovery efforts for the subspecies will be most effective if reliable information is available on the basic ecology of the subspecies and this information used tc.> design recovery efforts such as Habitat Conservation Plans. A review of studies conducted on PMJM shows that there is insufficient information to fully address defining-range-wide ecological requirements, limiting factors, limits of species tolerance, or population status (Shenk 1998). Most work to date has focused on geographic distribution (presence or absence of Z. h. preblei), taxonomy, and habitat descriptions of sites where mice have and have not been captured. For PMJM in particular, information on dispersal, habitat use, and population dynamics is most needed to identify minimal ecological requirements of the subspecies. Monitoring, by way of estimating demographic parameters (such as survival, abundance, and reproduction), of a single population over time provides an opportunity to document demography of a �4 species, estimate temporal fluctuations in demography, and gain insights into the temporal variation inherent in demographic parameters. Monitoring of multiple populations over time and over multiple geographic locations provides the opportunity to gain further understanding of population processes and detaches time effects from spatial e:f;fects. Population monitoring activities do not constitute scientific experiments, in the spirit of manipulation of salient ecological variables, however, replication of monitoring activities for natural populations over long periods of time and in diverse geographic locations can lead to insights into population processes (Cook and Campbell 1979). These insights can then be translated into hypotheses useful for predicting changes in population demography resulting from either natural perturbations (e.g., flooding events) or anthropogenic modifications (e.g., gravel mining). Experimentation would then be required to test these hypotheses and establish cause and effect. Movement and dispersal pattern information will be key to any conservation strategy designed for PMJM. Documenting daily and seasonal movement patterns of PMJM will provide information on habitats used by the mice and on the relative configuration and juxtaposition of these habitats. Configuration of habitats include vegetation and size of the areas used. Juxtaposition includes relative locations of different habitats used such as nest sites, feeding areas, and movement corridors connecting those areas. Key dispersal factors to document for PMJM include (1) which segment of the population disperses, (2) when do they disperse, (3) through what habitat do they disperse, (4) how far will individuals disperse (i.e., what is the maximum distance that separates adjacent populations) and (5) how critical is dispersal (both into and out of a population) to the persistence of a given population. Areas of suitable habitat must provide requirements to survive throughout the life cycle. These requirements must provide necessities for both the active period and hibernation periods. During the active period suitable habitat must provide requirements for daily survival, reproductive activities (breeding, nesting, and rearing of young to independence), and dispersal. The hibernation period requires sufficient food supplies to assure fat storage prior to hibernation and suitable hibernacula. Habitat providing all seasonal and life cycle requirements may or may not occur in a single contiguous area. If not in a contiguous area, habitat patches must occur in a mosaic of usable areas where suitable corridors exist for seasonal movement among sites. The habitat matrix within the range of Z. h. preblei is mixed grasslands_adjacent to the Colorado Front Range along the Piedmont and along the base of the Laramie Mountains in Wyoming and extends to the Colorado plains. Within this matrix, PMJM occur along stream drainages that contain patches of suitable vegetation. Suitable habitat appears to have at least two major compop.ents. The first component is a supply of open water, at least in part of the active season. Secondly, areas where PMJM has been found have dense cover. Based on studies of Z. h. preblei and Z. hudsonius elsewhere, Z. h. preblei apparently occurs mostly in undergrowth consisting of grasses, forbs, or both in open wet meadows and riparian corridors, or where tall shrubs and low trees form an overstory and provide adequate cover (Armstrong et al. 1997). Meadow jumping mice are widespread in abandoned grassy fields, but are often more abundant in thick vegetation along ponds, streams, and marshes or in rank herbaceous vegetation of wooded areas (Whitaker 1963). The mouse does not appear to have an affinity toward any single plant species but instead favors sites that are structurally diverse and provide adequate cover and food throughout its life cycle. PMJM are typically not found in upland areas away from riparian habitats but are most often captured where either ground water becomes visible as either seep springs or as main water channels (M. Bakeman, T. Ryon, personal communication) suggesting a dependence on open water, at least during their active periods. PMJM have been trapped in natural riparian areas as well as areas altered by anthropogenic influence including ditches and wetlands adjacent to interstate highways, cement-lined ditches with tall cover, ditches along driveways and moderate road. use, and moderate cattle grazing. If PMJM occurs as metapopulations in the classical sense of a set of local populations linked by infrequent dispersal then habitat includes not just one area of suitable habitation but also areas suitable for nearby mouse populations. These suitable areas must also be linked by dispersal habitat. If the mice are �5 dependent on dense riparian habitat for dispersal as well as for areas to reproduce, persistence of discrete populations would require a mosaic of suitable discrete riparian patches interconnected with dispersal corridors of similarly dense riparian vegetation. If mouse populations function in a source (populations where growth rate~ 1) and sink (those populations where growth rate< 1, maintained through immigration) system, it will be critical to identify and protect those populations serving as sources. Thus, for a source-sink population critical habitat will include those areas that support source population, dispersal habitat to sink areas will be less critical. If local mouse populations are functionally discrete, such a mosaic of interconnected areas of suitable habitat would provide a buffer for local, and source, populations against deleterious stochastic events by providing the opportunity for local population failures to be 'rescued' by immigration from other populations. The primary objective of this study is to investigate spatial and temporal variation in the demography and movement patterns of PMJM. Demographic parameters to be estimated include survival, reproduction, temporary emigration, immigration, population structure, and density. These demographic parameters will be estimated from individually marked animals from geographically distinct populations. To evaluate spatial differences in the demography of PMJM, populations were selected from three sites that provided a variety of habitat matrices available to the mouse. To begin to understand PMJM movement, dispersal, and habitat use we monitored radio-collared mice at three different study areas .. Study areas were selected to cover a variety of habitat configurations to evaluate spatial variation: in movement patterns of PMJM. Multiple years of conducting the study at those same sites will provide an estimate of the temporal variation in demography and movement patterns of PMJM. STUDY SITES Demography and movement patterns of PMJM are being evaluated for three populations. All three populations selected were located in areas where PMJM had previously been found. To evaluate spatial differences in the demography and movement patterns of PMJM, populations were selected on sites that provided a variety of habitat matrices available to the mouse. The first site selected, Colorado Division of Wildlife (CDOW) Maytag Property, has one primary water source available to the mouse. This water source is East Plum Creek. The second site, PineCliff Ranch, provides both a tributary (Garber Creek) and a main stem drainage (West Plum Creek). This provided the opportunity to investigate whether PMJM will use upland areas to move from one drainage to another or if they are restricted to only moving along riparian corridors. The third study site, CDOW Woodhouse Ranch provides an area containing a tributary (Indian Creek) and a series of ponds and irrigation ditches scattered throughout the property. The configuration at Woodhouse Ranch provided an even greater opportunity to investigate how much the mice use upland areas or if they restrict their movements strictly to riparian corridors. By replicating the same methodologies at each of these three unique habitat matrices we can begin to estimate spatial variation in nightly and seasonai movements of PMJM. • ,, OBJECTIVES Objectives of the demography study are to: 1. Estimate abundance and density of PMJM for each of three study populations. 2. Estimate over-summer, over-winter, and annual survival of PMJM at each of three study populations. 3. Estimate temporary emigration of PMJM at each of three study populations. 4. Estimate immigration of marked PMJM back into each of three study populations. 5. Estimate reproduction of PMJM at each of three study populations. 6. Evaluate the affect of weight, sex, age, abundance (i.e., density dependent response), and habitat features such as stream reach, vegetation composition and density on survival, reproduction, abundance, temporary emigration, and immigration of marked animals back into three study populations of PMJM. �6 7. Estimate age and sex ratios of PMJM at each of three study populations. The PMJM movement study is designed to describe nightly and seasonal movement patterns of PMJM and to describe habitats used by PMJM. These movement patterns will be described for three different study areas to evaluate spatial variation, and over three different years at the same three study areas to evaluate temporal variation. Specific objectives include: 1. Describe nightly movements of PMJM. Evaluate difference in nightly movements as they relate to sex, age, and habitat available to the animals. 2. Describe 30-day (or life of the radio transmitter) interval movements of PMJM. Evaluate difference in 30-day (or life of the radio transmitter) interval movements as they relate to sex, age, and habitat available to the animals. 3. Describe seasonal movements of PMJM. Evaluate difference in seasonal movements as they relate to sex, age, and habitat available to the animals. 4. Describe habitats where mice occur: movement corridors, end point descriptions (i.e., movement from what to what), and landscape features (connectivity with other riparian areas and other habitats used). 5. Estimate the mean amount of time PMJM spend in each available habitat. 6. Evaluate spatial and temporal variation in movement patterns of PMJM. The objective of the habitat use study is to identify and refine habitat requirements of Z. h. preblei, including hibernation sites, and to determine if they influenced any of the demographic parameters that will be estimated in this and complementary studies (see Shenk and Sivert 1999a, 1999b). METHODS See Shenk and Sivert (1999a, 1999b) for field methods. RESULTS Demography Study Trapping and PJT-tagging Effort A total of 186 individual PMJM were captured from the three study areas during the three 1998 trapping sessions (fable 1: 73 at Maytag Property, 77 at PineCliffRanch, and 36 at Woodhouse Ranch). These mice were captured over three different trapping sessions with an effort of 17,330 trap-nights. Every PMJM captured was PIT-tagged, providing each mouse with a unique, permanent, life-time identification marker. A total of 175 individual mice were captured on stream-side transects at three study sites during the 1999 field season including 34 mice PIT-tagged in 1998 (fable 1). These mice were captured over three different trapping sessions with an effort of 19,222 trap-nights. Other small mammals captured during the trapping sessions included Peromyscus spp., Mexican wood rat (Neotoma mericana), house mouse (Mus musculus), vole (Microtis spp.), western harvest mouse (Reithrodontomys megalotis), hispid pocket mouse (Chaetodipus hispidus), and shrew (Sorex spp.). See Shenk and Sivert (1999a) for details of 1998 field season. Reproduction During both 1998 and 1999, most adult mice captured exhibited evidence of active reproductive behavior, either pregnancy, lactation, or enlarged genitalia. We were able to locate one potential maternal nest in 1999. Juvenile mice were not captured during either June trapping sessions. Juvenile mice were captured at all three sites during both the July and September trapping sessions in both years. No estimates of reproduction could be made. �7 Survival Survival rate was estimated using the mark-recapture estimator for Pollock's Robust Design (see Kendall et al. 1995, 1997 and Kendall and Nichols 1995 for detail) in Program MARK (White and Burnham 1999). The robust design is a combination of the Cormack-Jolly-Seber (CJS)(Cormack 1964, Jolly 1965, Seber 1965) live recapture model and the closed capture models. The best fitting model combined data from all three study sites over all three trapping sessions in 1998. Survival was estimated as 0. 75 (se = 0.033) for a five week period in 1998. Extrapolating this survival rate across the summer results in an over-summer (June 1- October 5, 18 weeks) survival rate of 0355 (se = 0.056). No estimates of over-winter or summer survival in 1999 have been completed. Temporary emigration and immigration Temporary emigration rate and immigration was estimated using the mark-recapture estimator for Pollock's Robust Design in Program MARK. The best fitting model combined data from all three study sites over all three trapping sessions for these two parameters. Temporary emigration was estimated as 0.64 (se = 2.21), immigration was estimated as 0.45 (se = 89.94). Age and Sex Ratios As expected, no juvenile PMJM were captured during either June trapping session as it was too soon after hibernation for any litters to have been born and/or if born the juveniles would still be restricted to their nests. Juvenile mice were captured during both the July and September trapping sessions at all three sites in both years. Juveniles as small as 10g were captured. Sex ratios may be biased towards males, however further analyses need to be completed to confirm this hypothesis (Table 1). Density ,., Mark-recapture analysis was used to estimate abundance ( N) at each of the three study sites (Maytag Property, Woodhouse Ranch, and Pine Cliff Ranch) for each of the three trapping sessions in 1998 and for June 1999 (Table 2) using Program MARK (White and Burnham 1999) and Program CAPTURE (Otis et al. 1978, White et al. 1981). These abundance estimates and length of the trapline were used to estimate an unadjusted density of mice per kilometer of stream stretch. However, mice that typically don't use the area of stream stretch where the traps were placed could be· trapped because they were attracted to the area by the artificial food source. Including these mice in the density estimates for the stream stretch covered by the trapline would artificially increase density estimates. Therefore, an adjustment parameter was estimated to calibrate density estimates for a more realistic estimate of PMJM per kilometer of stream length. Without the adjustment densities would be inflated. In collaboration with other PMJM researchers (M. Bakeman, C. Meaney, T. Ryon, and R. Schorr) population estimates ( N) for a specified length of stream were determined with capture-r~pture techniques using Program MARK (White and Burnham 1999) and Program CAPTURE (Otis et al. 1978, White et al. 1981) for nine study areas over two years for early season trapping (June) only. To make these estimates comparable across study areas with unequal lengths of stream reaches trapped, mouse density (mice/km of stream) must be computed. However, traps tend to attract mice from some unknown distance, so that naive density estimate of N divided by stream length is biased high. To remove this bias, radio-tracking data were used to estimate the proportion of tiine (p) radio-collared mice spent within the original trapline once the traps were removed. Data from six study areas (Table 3) provided by this study and the study conducted by T. Ryon at Rocky Flats were used to estimate this correction factor (p) for population estimates from linear traplines or grids. Only these study areas had radio-collared mice available from which to estimate the correction factor. Corrections were applied to all study areas with the function relatingp to trapliri.e length (L) developed from these data. �8 The Michaelis-Menton model was fit to the observed p value: L; P; = - - - L;+2BSW where L; is the length of the trapline for grid i. Nonlinear least squares were used to minimize the sum of squared E; values to obtain an estimate of the unknown parameter, BSW (i.e., boundary strip width). Observations were weighted by the number of radi~llared mice providing the observed value of pi. Because the variance of a proportion is a function of the value of the proportion, a logistic transformation was used to stabilize the variances of the E; across the range of the P; values: Li + 2BSW Li 1 - ----Li + 2BSW Predicted values ofp (p) are obtained from either fitted model for a particular trap line length (L) by substituting the estimate of BSW into the predictive equation: A L p = L + 2BSW. Estimates of BSW for the two estimation methods are shown in Table 4, predicted values are shown in Table 3, and the fit of the two fitted equations is shown in Figure 1. The parameter estimate from the logistic transformation model was used to correct the population estimates, i.e., A L p=-----. L + 2x4I.5446 This model was preferred because of the variance stabilization provided by the logistic transformation. However, as shown in Figure 1, the difference in the predictions of the two fitted equations is not biologically important. To obtain an adjusted number of mice on a trapline or trapping grid, the estimated population on the grid (N) from mark-recapture estimation is multiplied by the correction (ft) from the above equation to give the adjusted number: The variance of Nadj is computed as the variance of a product using the delta method: �9 where var(p) = 4L 2 var(BSW) . (L + 2BSW)4 Similarly, density (mice per unit of length of stream) is computed. as the adjusted. population size divided by the length of the trapline or the original trapline or grid population estimate over the adjusted. length of the trapline: ·" Nadj N L L + 2BSW D=--=---- The variance of D is computed. using the delta method: D = var(Naqj) = _ _ _ 1 ___-var(N) + N2 var(BSW) . 2 4 2 L (L + 2 *BSW) (L + 2 *BSW) A Note that Var(BSW) = SE(BSW)2 with SE(BSW) given in Table 4. •As an example of the variance of p, a trapline length of 500m results in SE(p) = Jvar(p) = 0.02696. Confidence intervals on the fitted function of pare shown in Figure 2, computed. as ft±l.96xSE(p). An alternative approach to computing-confidence intervals on p that we would have thought would work better than the ±1.96SE(p) intervals shown in Figure 2 is to compute the confidence interval on the logit(p), and then back.;.transform the interval endpoints: 1 1 1 + exp(-(Iogit(p) - 1.96SE(logit(p))]' 1 + exp(-Vogit(p) + 1.96SEQogit(p))] where SE(logit(p)) = SE(BSW) = 9.1676/41.5446 = 0.22067 for the data presented. here. After BSW • -· computing confidence intervals by both methods, the results were only negligibly different, with the •· maximum difference in the interval lengths being only 0.00111 over the range of L from 100 to 2000m. Riparian Vegetation Cover Data 1'.otal area of riparian shrubs,trees, herbaceous cover, non-vegetated, and open water were computed for each $tlldy area from riparian vegetation maps (fable 5). These riparian maps were developed from photo interpretation of infrared aerial photography at a resolution of 1:24,000. Areas of riparian vegetation were calculated when patches of vegetation were greater than 24 m wide. To estimate area of each riparian vegetation classification for each study area, stream length for each study area was delineated on the riparian vegetation map. A 200-m buffer was then generated on either si~ of the study area stream reach. Within this 200-m buffer total area (ha) of patches with dominant vegetation classified as riparian �herbaceous, riparian shrub, riparian tree, non-vegetated, and open water were calculated using Arcinfo (ESRI 1987). No ground truthing of vegetation classification was performed. Statistical Analysis ofPMJM Density and Riparian Vegetation Cover The adjusted density and areas of riparian vegetative cover per km of stream (fable 5) were used to assess the relationship between PMJM density (mice/km of stream) and riparian vegetation cover. Estimates for multiple years for study areas were treated as separate estimates so that the year-to-year variation in PMJM density could be evaluated. An AN OVA of PMJM density for study area and year effects suggested no differences between years (1998: = 35.08, SE= 8.85; 1999: = 31.04, SE= 6.29; P = 0.511) or across study areas (P = 0.245). The variation in PMJM density across study areas is what we want to ex.plain with riparian vegetation cover differences across study areas. Individually, none of the vegetation cover variables were important predictors of PMJM density (fable 6). When model selection following Burnham and Anderson (1998) was performed across all models involving the riparian shrub, tree, herbaceous, non-vegetated, and open water cover (ha/km stream ) variables, the best AICc model was shrub and tree cover (fable 7) ex.plaining 68% of the variation in PMJM density. The next best model included open water as well as shrub and tree cover, explaining 71 % of the variation of PMJM density. The combination of riparian shrub and tree cover appears to be an important set of predictors for PMJM density based on the study areas included in this analysis. The parameter estimates for the best AICc model are shown in Table 8, and a residual plot in Figure 3. All riparian vegetation cover variables have positive slope parameters, indicating a positive influence on PMJM density over the range of the vegetation cover variables considered in this analysis. x x Movement Study Radio-tracking effort . A total of 125 radio-collars were put on mice over the three trapping sessions in summer 1998 (fable 9: 48 at Maytag Property, 47 at PineCli:ffRanch, and 30 at Woodhouse Ranch). A total of 62 females and 63 males were radio-collared over the 1998 field season. A total of 138 radio-collars were put on mice over the three trapping sessions in summer 1999 (fable 9: 51 at Maytag Property, 52 at PineCliffRanch, and 35 at Woodhouse Ranch). A total of 57 females and 81 males were radio-collared and tracked in 1999. Daily movements The general daily movement pattern throughout the active season was for mice to become active at dusk, remain active through the night, and to return to a day nest location at dawn. Most of our tracking was confined to night time movements; however we did make a greater effort in 1999 to confirm the mice were primarily in day nests during daylight hours. This was true for most daytime observations, however on occasion mice were active in the daytime as well. Percent of PMJM locations, as determined from radio-tracking, at each of three distance categories were calculated for each study site and the latter two radio-tracking sessions in 1998 (see Shenk and Sivert 1999b for details). The majority oflocations at Maytag, PineCliff: and Woodhouse were within 46m of stream center for both tracking periods (July-August and September-October). For all areas and tracking periods except Maytag during the last tracking session, 90% of PMJM movements were within 9 lm of either side of the center of the capture drainage. Mouse movements at Maytag during September-October resulted in 32.6% of the locations being> 91m from the stream center. Similar patterns were observed for 1999. • During nigh.time activity periods more than one mouse would often be found foraging in the same area. These foraging areas were often the same areas used by the same mice on numerous consecutive nights �11 Seasonal movement Field methods used in June 1998 did not allow us to document accurate June movement patterns for that year. The distribution of mouse locations in 1998 at Maytag Property for the July-August tracking session is different from the distribution of PMJM locations for the September-October tracking session. In 1998, the pattern shifted from heavy use along East Plum Creek to the north of the trapline to a more concentrated distribution either side of the trapline. The more concentrated use of the areas east and west of the trapline also resulted in locations being further from the center of the creek. The northern end of the trapline is largely vegetated by willows with the remainder of the trapline more diversely vegetated. In 1999 we were able to document movements for all three tracking sessions. Greatest movement away from the center of the stream occurred in the August session. In 1998, mouse movements were more concentrated along the drainages at PineCliff Ranch during the September-October tracking sessions than during the July-August tracking sessions. Vegetation along both West Plum Creek and Garber Creek along the trapline was primarily willow. In 1999, greatest movements from either stream center were documented during the August tracking session. At Woodhouse Ranch, the distribution of mouse locations also shifted between tracking sessions. Mouse movements during June 1999 were the most diverse of all tracking sessions. Mouse movements were more concentrated along the southern half of the trapline in September-October as compared to the more even distribution of mouse locations recorded for July-August in both 1998 and 1999. The northern end of the trapline is dominated by willows, the middle and southern end of the trapline is in more diverse riparian habitat. In September 1999 we had fewer mice radio-collared at each of the three sites because we were not able to trap mice. We assume the majority of the mice had already hibernated by the time we started our last trapping effort on September 9, 1999. Over the two-year study, 27 PMJM were captured, radio-collared, and tracked for more than one tracking session (fable 10) . Other mice were captured during more than one trapping session but radio collars were not replaced either because all the radio collars had been put on other mice or because the physical condition of the animal was such that we decided not to replace the radio-collar. Such conditions included mice who had significant hair worn off the neck from the previous radio collar. In no incidence did we find open wounds caused by radio collars. Of the 27 mice radio-collared for more than one session, we have only analyzed the mice captured in 1998. Of those, four provided information to evaluate seasonal changes in areas and habitats use between the July-August tracking period and the September-October tracking period: One mouse, a male from Maytag was observed only once during the September-October tracking session, providing no information concerning seasonal changes in movement patterns. Thus, changes in areas used by individual mice between June and either the July-August or September-October tracking session are not summarized here. A female at Maytag exhibited a seasonal movement shift, small sample size prohibited any evaluation of the movement patterns of the Maytag male (n = 1 in September-October). There does not appear to be any shift in areas used by the male tracked during both latter sessions at Pine Cliff Ranch, however, sample sizes were small (n = 17, 14). Two females at Woodhouse Ranch exhibited a shift in areas used between July-August and September-October. However, caution should be used in interpreting these results because of low sample sizes in the latter tracking session (n = 14, 6). Annual variation in movement patterns Preliminary comparisons of mouse movements from 1998 to-1999 for tracking sessions August and September were completed. Mouse movements during August tracking sessions were similar at Maytag Property. More mice were captured on the southern half of the Maytag Property in 1999 and this is reflected in the greater nUiilber of locations in the southern half of the property for that year. The same area located outside a 100m strip from the stream center was used both years during this tracking session and was comprised primarily of a seep area. August movements at Pine cliff Ranch were further from the �12 stream in 1999. Mice were also found using an ephemeral stream in August at PineCliff Ranch in 1999. 1bis stream was dry at the time. Similar mouse movements were observed at Woodhouse Ranch over both August tracking sessions although movements were more concentrated in the north in 1998 and more concentrated in the south in 1999; Temporal comparisons of mouse movements in the September tracking sessions are not entirely comparable because fewer mice were followed in the 1999 September tracking session because we were not able to capture mice. We assume a large portion of the mice in 1999 were already hibernating when we started our September trapping and radi~llaring session. We did not observe the greater movements away from the stream center at Maytag in 1999 that we observed there in 1998, however we probably missed most of the movements to hibernation sites. Movement patterns at PineClifffRanch and Woodhouse Ranch were similar for both September tracking sessions with movements being more concentrated near the stream. The same feeding hotspot at Woodhouse Ranch was used both years in September. Habitat Use Detailed vegetation maps were created using Global Positioning Systems. Combining these vegetation maps and locations of mice we will be able to describe in further detail habitat use. These analyses are not yet completed. Nest sites Numerous daytime nest sites were located at all three study sites from both years. Most nest sites were made of tightly woven vegetation, located on the ground. However, we also located underground daytime nests in summer 1999. Nest material included leaves, grass, and small sticks. There was only one small entrance hole on each of the nests found. When observing mice in their nests the mice would peer out of the hole and remain motionless as long as we were present. One possible maternal nest was located in 1999. This possible maternal nest was underground, made of densely packed grasses with the burrow lined as well. There also were several locations at each of the study sites where adult females returned to repeatedly. Each of these sites were in patches of extremely dense vegetation or underneath a large downed log. Hibernacula Potential hibernation sites were located by noting stationary radio-telemetry signals over repeated nights. These signals were coming from underground with no evidence of predation or slipped collars. Eight potential hibernacula were located, at least one at each study site in 1998. In 1999 a total of 13 potential hibernation sites were located. Three were located at Woodhouse Ranch (I female, 2 males) and IO sites at PineCliffRanch (2 females, 8 males). In contrast to 1998 data, these potential hibernation sites were located closer to the stream. However, at both Pine Cliff and Woodhouse Ranches stream banks rise out of the floodplain in closer proximity to the stream than at Maytag Property where the potential hibernation sites located in 1998 were located long distances from the stream center. Vegetation characteristics of the eight sites located during both 1998 and 1999 are similar to other hibernacula described for PMJM. Hibernation sites are considered potential until the sites can be dug up to confirm a hibernation chamber. One male mouse was dug up at Pine Cliff Ranch in 1999 to confirm the site as a true hibernation location and to gain information on the hibernation nest and chamber. The mouse was in the chamber and was not aroused when the earth was dug out around him. Once we finished our assessment of the nest the soil was put back in place, leaving the mouse in the chamber. The remaining hibernation sites will be dug up in June 2000 so as not to disturb other hibernating mice. We hope to confirm these sites as hibernation sites then and will document features of the hibernation sites at that time. �13 Mortality factors Mortalities factors of radicxollared mice in 1998 included predation by rattlesnake, garter snake, fox, and house cat. Four probable predations were also noted and identified by finding tightly crimped (i.e., no possibility of the mouse having slipped the collar over its head) radio-collars lying on the ground. Two accidental deaths were documented, a road kill and drowning. Mortality factors also included trapping and/or handling mortalities and unknown causes. In 1999 mortality factors also included predation by bullfrogs, weasel, and yellow-bellied racer (Table 11). Nine mice also died once they entered a state of torpor above ground. These mice either succumbed to cold or were predated on. Fecal analysis The amount of bait found in each of the fecal samples was quantified as either 100%, abundant(> 70%), trace(< 10%) or 0%. The following summaries are made after eliminating all samples of 100% bait, and then only classifying the fecal sample contents other than bait. Fecal analyses indicate a seasonal shift in diets. The most common item found in the fecal samples during the June trapping session at all three sites for 1998 were arthropods. This was true for three of the five June samples collected at Maytag, eight often June samples at PineCliff, and four of five June samples collected at Woodhouse. Other common items included endogenous fungus (1) and seed (1) at Maytag; endogenous fungus (2) at PineCliff,; and Poa (1) at Woodhouse. During the July 21-August 4, 1998 trapping session the most common items in the fecal samples at Maytag were arthropod (2), endogenous fungus (1), pollen (1), and Carex (1). During the same trapping period, the most common items in five of eight samples collected at PineCliff were endogenous fungus, the other three samples having a majority of arthropod (1), moss (1), and pollen (1). The two samples from Woodhouse during this trapping session were composed primarily of either mushroom or seed. On August 13, 1998, nine samples were collected during an extra trapping session (on the same trapline as the other trapping sessions occur) with the most common items in the fecal samples being moss (4), endogenous fungus (3) or pollen (2). All three samples collected from PMJM captured at a back drainage on August 23, 1998 at Woodhouse Ranch were primarily fungus. The September 1998 trapping session at Maytag yielded samples with majority fecal contents of arthropod (2), moss (2), pollen (2), endogenous fungus (1), and seed (1). Ten samples from PineCliff during September 1998 had majority fecal contents of arthropod (3), endogenous fungus (3), and seed (3). Eight of nine samples taken from Woodhouse contained primarily arthropods, with one a trace of seed. A total of 329 fecal samples were collected during the 1999 field season. Analyses of these samples have not been completed. We also collected 16 stomach samples from dead mice. Analyses of these stomach samples are not yet completed. DISCUSSION Information on the population dynamics of PMJM is necessary to determine which areas and habitats support viable populations. To begin to evaluate the viability of a population information on key demographic parameters must be obtained. Conducting studies on individually marked animals provides the greatest insight on the demography of a population. In general, estimated sex ratios from this study are comparable to those found by Armstrong et al. (1997) who reported an overall sex ratio for all captured PMJM of51.6 males: 48.4 females; approximately 86.0% of captures were identified as adults. There is a possible male sex bias at PineCliff Ranch. However, small sample sizes and possible trapping biases by sex may explain the discrepancy. Density estimates were expected to increase as the summer progressed to account for the birth pulses in late June, and late July-August. In 1998, this was the trend observed at Maytag. PMJM densities at Wcxx:lhouse Ranch increased from June to July but did not increase further during the September trapping session. Highest densities occurred at PineCliff, where the vegetation is primarily willow and both a �14 mainstem and tributary were used by mice. Lowest densities occurred at Woodhouse Ranch where the riparian vegetation has fewer willow but was dense with other riparian vegetation. Vegetation at Maytag provided some areas of dense willow with the remaining areas being of moderate density of riparian vegetation. The lower density of PMJM reported for Woodhouse Ranch might be explained by the composition and densities of other small mammals at that site. Woodhouse Ranch had the highest captures of both house mice and voles of the three sites studied. Defining summer as June 1 - October 5, over-summer survival was estimated as 0.36 (se = 0.056) over all three study sites. Meaney et al. (1999) report a one-month summer survival rate of 78%. Extrapolating Meaney et al. 's ( 1999) estimate over our summer period (- four months) would result in a similar oversummer survival rate estimate of 36%. Prior to studies conducted in 1998 no information existed on survival rates for populations of Z. h. preblei although Whitaker (1963) reported a 67% loss of Z. hudsonius over hibernation. Temporary emigration and immigration rates were estimated but both had extremely high variances associated with those estimates. Thus these estimates cannot be used, with any confidence, to provide information on movements of mice into and out of these populations. Very little new information was gained during this study on reproductive parameters of PMJM. Most adult mice captured exhibited evidence of active reproductive behavior, either pregnancy, lactation, or enlarged genitalia. Juvenile mice were not captured during the June trapping session. Juvenile mice were captured at all three sites during both the July and September trapping sessions. Given that breeding pea.ks appear to occur in early to mid-June and August with a possible third litter in September (Whitaker 1963) this was not unexpected and agrees with previous observations (Meaney et al. 1996, 1997, PTI 1996a, M. Bakeman unpublished data, T. Ryon unpublished data) Preliminary analysis of the movement data collected on radio-collared PMJM in 1999 supported results found in 1998. In particular we were able to document again in this second field season that PMJM exhibit (1) greater use of upland habitats than previously assumed, (2) general site fidelity to both daytime nesting sites and nighttime feeding sites, (3) seasonal shifts in movement patterns, and (4) use of both perennial and intermittent tributaries adjacent to the capture drainage. Jumping mice of the genus Zapus are true hibernators, spending much of their lives in hibernation. Meadow jumping mice spend approximately 7 months (-210 days) per year in hibernation (Quimby 1951) whereas estimates for Z. princeps indicate that some populations (e.g., in the western mountains of Utah) spend up to 300 days per year in hibernation (Cranford 1983). Jumping m.i,ce hibernate in underground burrows (Quimby 1951, Whitaker 1963). They are excellent burrowers and create their own hibemacula. Meadow jumping mice are generally solitary hibernators, however, there have been occurrences of more than one mouse found in a single hibemaculum. Eight possible hibemacula were located during 1998. Five of the eight mice using these possible hibernacula traveled ~ 90 meters from the center of their typical September night time locations. Of the 13 possible hibemacula located in 1999, three sites were located at Woodhouse Ranch and ten possible hibemacula at Pine Cliff Ranch. In contrast to 1998 data, these potential hibernation sites were located closer to the stream. However, at both Pine Cliff andWoodhouse Ranches stream banks rise out of the floodplain in closer proximity to the stream than at Maytag Property where the potential hibernation sites located in 1998 were located long distances from the stream center. Vegetation characteristics of all the potential hibemacula located over both years are similar to other hibemacula described for PMJM. One confirmed hibernaculum, located on Rocky Flats Environmental Technology Site, used by Z. h. preblei has been located (Armstrong et al. 1997). This site was 9m above a creek bed (Walnut Creek); it had a thick cover of chokecherry (Prunus virginiana) and snowberry (Symphoricarpos spp.), the mouse was found in a leaf litter nest 30cm beneath the ground in coarse textured soil (Armstrong et al. 1997). Four possible hibemacula were located by tracking radiotelemetered mice at the U. S. Air Force Academy in fall 1997. These sites are located 7, 12, 29, and 3 lm from a creek bed (R. Schorr, unpublished data). There was no consistency among sites in aspect. Three sites were in vegetation dominated by coyote willow (Salix exigua), one site was in vegetation dominated by snowberry and mullein (Verbascum thapsus). However, all four hibernacula appeared to be below �15 coyote willows. The eight sites located during this study and the four U. S. Air Force Academy sites were not disturbed to protect any hibernating mice and therefore are only possible hibernacula because there is no confirmation a mouse actually hibernated there. Confirmation of a true hibernaculum cannot be made until a chamber, or nest is located. The other explanation for these collar locations might be either locations of radios discarded by the mice or dead mice carried underground by a predator. Location or more hibernation sites was limited primarily by normal battery failure of the radio transmitters before mice went into hibernation. Prior to the 1998 field season, natural mortality factors reported for Z. hudsonius included only insufficient fat storage prior to hibernation (Whitaker 1963), predation (Whitaker 1963, Poly and Boucher 1997, R Schorr unpublished data) and cannibalism (Sheldon 1934). Other assumed natural mortality factors for Z h. preblei included starvation, exposure, and disease. Natural mortality factors documented during this study included predation by house cats, garter snakes, rattlesnakes, yellow-bellied racers, bullfrogs, weasel, and fox as well as accidents by drowning and road kill. In 1999.mice were also found in torpor throughout the summer, often in very exposed areas frequently resulting in death by either exposure or predation. Use of ephemeral drainages was observed at both Maytag Property (in l 998) and Pine Cliff Ranch (in 1999), Two mice moved away from East Plum Creek at Maytag Property in September 1998 and focused their movements ~300 meters from their previous locations, up a dry drainage dominated by upland grasses and gambel oak. These two mice continued to use this new area and were not observed again near East Plum Creek for at least two weeks. Normal radio-failure (i.e., batteries failed after ~4 weeks) after this time did not allow us to determine if the mice ever returned to East Plum Creek or if they hibernated in their new location. Use of the ephemeral drainage at Pine Cliff Ranch occurred in the August tracking session. The high percentage of arthropods and endogenous fungus found in the fecal samples provides a new aspect to evaluating PMJM habitat requirements. When considering habitats used by PMJM, or in trying to predict habitats that might be suitable for the subspecies, consideration should be given as to whether those habitats could support the arthropods and fungus apparently being selected for by the mouse. Arthropods are a food source high in protein and fat which would benefit mice emerging from hibernation and mice preparing for hibernation. Whitaker ( 1963) reported a 67% loss of individuals over hibernation and that average body mass of individuals emerging from hibernation was greater than the average for mice entering hibernation. Because no mice are known to store food in their hibernacula, this indicates that the lighter individuals died during hibernation and only those entering with higher masses survived. All the energy they use during hibernation and the periodic arousals (the energetically most expensive part of hibernation) must be the fat they carry into hibernation (B. Wunder, personal communication). The ability to put on sufficient fat for overwinter survival during hibernation is a critical factor in the life history of these mice. Thus, appropriate and sufficient food sources must be available to the mouse to meet these nutritional requirements. The apparent seasonal shift in mouse movements between the July-August and September tracking sessions may be a result of diet switching. The broader diets suggested by the fecal analyses from samples collected during July and August possibly represent both a wider availability of suitable foods and the ability of PMJM to exploit these resources. It might also suggest a need to exploit these other resources to provide the. mouse with necessary food requirements for breeding. Because mice tracked during each session were not generally the same mice there is a possibility these apparent movement shifts might only be different areas used by different mice. The probability of this being the case is small for two reasqns. The first is the low density of mice in the stream (Shenk and Sivert 1999). Given that, the proportion of mice followed during each session is a high p:,;-oportion of the mice in the population. Thus, each subset of mice should be representative of the population. Secondly, evidence from the four mice followed during both the latter tracking sessions also exhibited movement shifts between sessions. The combination of shifts in both general mouse movements, individual mouse movements, and diet provide strong circumstantial evidence that PMJM may be selecting for or require specific seasonal diets. If �16 this is the case, all these requirements must be considered and provided for to ensure conservation of the subspecies. A detailed literature search and further studies need to be conducted to investigate the possible implications of these dietary patterns. In comparing movement patterns from 1998 to 1999 we were able to evaluate annual variation in daily and seasonal movement patterns of PMJM. General patterns that emerged from the comparison of the two years included (1) similar areas being used by the mice over both years, (2) greater use of upland habitats than previously assumed, (2) general site fidelity to both daytime nesting sites and nighttime feeding sites, (3) seasonal shifts in movement patterns, and (4) use of both perennial and intermittent tributaries adjacent to the capture drainage. This study looked at PMJM movements at only three sites. These sites were selected based on known presence of PMJM. Other sites, perhaps because of different habitat configurations or sites of poorer quality may require mice to move further from the creek or may allow mice to remain closer to the creek in order to obtain all life requirements. However, these study sites were specifically selected to address spatial variation in movement patterns of PMJM due to different spatial configuration and juxtaposition of habitats. And although all study sites were different they could all be considered within the general description of what is consider typical habitat for PMJM. It should also be noted that we generally trapped along the creek. Thus, mice captured and subsequently radio-collared and followed were those mice that use the area near the most prominent drainage. If there are mice that do not regularly use the main channel and thus were not available for capture there would be a bias in the data towards fewer observations 300 feet away from the creek. To test for such a bias, traps were placed 300 m from the stream center at Maytag and Woodhouse Properties during suminer 1999 as part of a Master's Project conducted by a student ay Colorado State University. Capture of mice in transects placed away from the stream lends further support for use of habitats further from the creek than initially thought. Mice captured 300 meters from the stream moved near their capture site and between the capture site and the stream. The prevailing idea of what constitutes quality PMJM habitat is amount of riparian shrubs. Therefore, we focused our analysis on describing the relationship between PMJM densities and amount of various riparian vegetation cover, including riparian shrubs. This analysis is not meant to identify all critical components of PMJM habitat. For example, if a necessary component of PMJM habitat is open water, but not necessarily in any large amount, this analysis would not detect this relationship. The best model describing the relationship between PMJM density and riparian vegetation cover included shrub and tree cover, the next best model included open water as well as shrub and tree cover. Tree cover may also reflect shrub and herbaceous cover, because the tree canopy likely may be covering these other understory vegetation classes . Thus, results from this analysis appear to support the hypothesis that riparian shrubs are an important component of PMJM habitat. However, because the study areas included in this analysis were not randomly selected from all possible PMJM habitats, the conclusions reported here must be treated with some caution. To truly assess the relationship between PMJM density and riparian vegetation would require estimating densities from a random selection of areas presumed to have suitable PMJM habitat. Analysis of how the riparian vegetation characteristics at these study sites relates to their PMJM densities would then provide an unbiased estimate of the relationship. With.the limited data available, 68% of the variation in PMJM density is explained by a model that includes riparian shrub and tree cover (ha/km stream), as identified by the vegetation mapping techniques used in this study. These results suggest that habitat quality of PMJM can be predicted by the shrub and tree cover available on a site. This report includes results from only the first two years of a multi-year project to follow individually marked PMJM through time. Further analyses of these data, collection of more data, and more years of data, will continue to improve our ability to evaluate demographic parameters estimates and movement patterns of PMJM and of how they vary across space and time. Preliminary results from the demography, movement and distribution studies of PMJM have suggested the need to continue with research currently �17 being addressed in these three studies. However, the preliminary results also suggest further research be conducted on (1) use of upland habitat by PMJM, (2) refining water requirements of PMJM (i.e., do they require stream habitat or are wetland areas sufficient), (3) range-wide distributional boundaries (e.g., elevation restrictions), and (4) investigating areas of potential sympatry or hibridization with Z. princeps (western jumping mouse). The demography and movement studies will be modified during the summer 2000 field season to further investigate upland use and water requirements. Acknowledgments Additional data on PMJM densities for the vegetation analysis were provided by Mark Bakeman, Rob Schorr, Carron Meaney, and Tom Ryon. Bruce Lubow provided helpful comments on the first drafts of the adjusted PMJM density correction method, and computed the estimates ofPMJM population size for several of the study areas included in this analysis. Seth McClean with CDOW provided the riparian vegetation data. Mark Bakeman and Tom Ryon provided useful comments on earlier drafts of the vegetation analysis. Literature Cited Armstrong, D. M., M. E. Bakeman, A. Deans, C. A. Meaney, and T. R. Ryon. 1997. Conclusions and recommendations in: Report on habitat findings on the Preble's meadow jumping mouse. Edited by M. E. Bakeman. Report to USFWS and Colorado Division of Wildlife. Burnham, K. P., and D.R. Anderson. 1998. Model selection and inference: a practical informationtheoretic approach. Springer-Verlag, New York, New York, USA. 353pp. Cook, T. D. and D. C. Campbell. 1979. Quasi-experimentation: design and analysis issues for field settings. Houghton-Mifllin, Boston. Cormack, R. M. 1964. Estimates of survival from the sightings of marked animals. Biometrika 51:429438. Cranford, J. A. 1983. Ecological strategies of a small hibernator, -the western jumping mouse Zapus princeps. Canadian Journal of Zoology 61:232-240. ESRI. 1987. ARC/INFO users manual. Environmental Systems Research Institute. Redlands, California, USA. Jolly, G. M. 1965. Explicit estimates from capture-recapture data with both death and immigration stochastic model. Biometrika 52:225-247. Kendall, W. L., and J. D. Nichols. 1995. On the use of secondary capture-recapture samples to estimate temporary emigration and breeding proportions. Journal of Applied Statistics.22:751-762. Kendall, W. L., K. H. Pollock, and C. Brownie. 1995. A likelihood-based approach to capture-recapture estimation of demographic parameters under the robust design. Biometrics 51:293-308. Kendall, W. L., J. D. Nichols, and J.E. Hines. 1997. Estimating tempqrary emigration using capturerecapture data with Pollock's robust design. Ecology 78:563-578. Krutzsch, P.H. 1954. North American jumping mice (genus Zapus). University of Kansas Publications, Museum ofNatural History 7:349-472. Meaney, C. A., N. W. Clippinger, A. Deans, and M. OShea-Stone. 1996. Second year survey for Preble's meadow jumping mouse (Zapus hudsonius preblei) in Colorado. Report prepared for the Colorado Division of Wildlife. Meaney, C. A., A. Deans, N. W. Clippinger, M. Rider, N. Daly, and M. O'Shea-Stone. 1997. Third year survey for Preble's meadow jumping mouse (Zapus hudsonius preblei) in Colorado. Report prepared for the Colorado Division of Wildlife. Meaney, C. A., A. Ruggles, B. Lubow, N. W. Clippinger, and A. Deans. 1999. Preliminary results: Second year study of the impact of trails on small mammals and population estimates for Preble's meadow jumping mice on City of Boulder Open Space. Report for Greenways Program, City of Boulder Transportation Department, City of Boulder Open Space, and Environmental Protection Agency, Region 8. �18 Otis, D. L., K. P. Burnham, G. C. White, and D. R Anderson. 1978. Statistical inference from capture data on closed animal populations. Wildlife Monograph 62:1-135. Poly, W. J., and C. E. Boucher. 1997. Record ofa creek chub preying on a jumping mouse in Bruffey Creek, West Virginia. Brimleyana 24: 29-32. PTI Environmental Services. 1996a. Preble's Meadow Jumping Mouse Study at Rocky Flats . Environmental Technology Site, Annual Report 1996. Final. Rocky Flats Environmental Technology Site, Golden, Colorado. PTI Environmental Services. 1996b. Preble's Meadow Jumping Mouse Study at Rocky Flats Environmental Technology Site, Spring 1996. Final. Rocky Flats Environmental Technology Site, Golden, Colorado. Quimby, D. C. 1951. The life history and ecology of the jumping mouse, 7.apus hudsonius. Ecological Monographs 21 :61-95. Seber, G. A. F. 1965. A note on the multiple recapture census. Biometrika 52:249-259. Sheldon, C. 1934. Studies on the life histories of Zapus and Napaeozapus in Nova Scotia. Journal of Mammalogy 15:290-300. Shenk, T. M. 1998. Conservation assessment and preliminary conservation strategy for Preble's meadow jumping mouse (Zapus hudsonius preblei). Colorado Division of Wildlife FY1997-98 Annual Report. Shenk, T. M. and Sivert M. 1999a. Temporal and spatial variation in the demography of preble's meadow jumping mouse (Zapus hudsonius preblei). Colorado Division of Wildlife. Annual Report 1999. Shenk, T. M. and Sivert M. 1999b. Movement patterns of Preble's meadow jumping mouse (Zapus hudsonius preblei) as they vary across time and space. Colorado Division of Wildlife. Annual Report. Whitaker, J. 0., Jr. 1963. A study of the meadow jumping mouse, Zapus hudsonius (Zimmerman), in cental New York. Ecological Monographs 33:3. White, G. C., D. R Anderson, K. P. Burnham, and D. L. Otis. 1982. Capture-recapture and removal methods for sampling closed populations. LA-8787-NERP, Los Alamos National Laboratory, Los Alamos, New Mexico, USA 235 pp. White, G. C. and K. P. Burnham.1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 Supplement: 120-138. Table 1. Total number of individual Preble's meadow jumping mice (7.apus-hudsonius preblei) captured in 1998, new individuals captured in 1999, recaptures in 1999 from 1998, and total number of individual mice caetured in 1999 at three studr sites. New captures 1999 Total individuals captured 1998 Site 1999 from 1998 Recaptures in Total individuals captured 1999 F M u Total F M Total F M Total F M Total Maytag Property 34 38 1 73 19 27 46 4 10 14 23 37 60 PineCliff Ranch 25 52 0 77 19 29 48 4 8 12 23 37 60 Woodhouse 17 18 1 36 17 30 47 3 5 8 20 35 55 TOTAL 75 108 2 185 55 86 141 11 23 34 66 109 175 �19 -- Table 2. Stream reach abundance estimates (N) for Preble's meadow jumping mouse (PMJM) from three sites in Douglas County, Colorado for three traeeing sessions in 1998 and for June 1999. 95% Confidence Interval Trapping Session Site Maytag Maytag Maytag PineCliff PineCliff PineCliff Woodhouse Woodhouse Woodhouse Maytag Woodhouse PineCliff R se(N) June98 July98 September98 Junc98 July98 September98 June98 July98 September98 June99 June99 June99 18 31 44 30 17 51 11 20 20 35 32 28 2.7 11.4 1.8 5.7 0.9 3.3 3.8 5.2 3.1 3.8 3.2 2.1 Lower Upper 15 20 42 24 16 48 7 15 17 24 56 52 21 42 46 36 18 54 15 25 23 39 69 61 Trapline Length Adjusted Density (m) (PMJM/km) 550 608 494 490 504 504 510 516 574 1130 504 503 26.2 41.3 69.5 47.7 26.3 78.9 16.8 30.7 27.8 31.2 62.6 56.5 Table 3. Data on the proportion of locations (p) of Preble's meadow jumping mouse locations within the area defined by parallel lines running through either end of the trap line, sorted into ascending order by traeline length. Site Year Session Meanp SE(p) Weighted Fit Logistic Trapline No. Length Weighted Fit Mice Online Walnut Creek 1999 May/June 0.946 0.0040 0.761 0.796 325 5 A-upper Woodhouse 1999 0.629 0.0843 0.800 0.831 June 409 24 Woodhouse 1999 July 0.902 0.0532 0.800 0.831 409 19 Woodhouse 1999 Sept 409 1.000 0.0000 0.800 0.831 6 Walnut Creek 1999 May/June 490 0.308 0.827 0.855 1 Lower Maytag 1998 Sept 0.954 0.829 494 16 0.0385 0.856 Walnut Creek 1999 May/June 0.828 0.830 500 1 0.858 B-series PineCliff 1999 0.841 0.0670 0.831 0.858 June 503 12 PineCliff 1999 Sept 503 0.924 0.0464 0.831 0.858 15 PineCliff 1999 July 0.899 0.831 503 22 0.0372 0.858 PineCliff 1998 July 0.805 0.831 504 14 0'.0750 0.858 PineCliff 1998 0.671 0.831 Sept 504 16 0.0964 0.858 Woodhouse 1998 July 12 0.817 0.0712 0.835 0.861 516 Woodhouse 1998 Sept 14 0.883 0.0291 0.849 0.874 574 Maytag 1998 July 0.719 0.0744 0.856 0'.880 608 16 Maytag 1999 Sept 0.969 0.1664 0.917 0.932 1130 11 Maytag 1999 1130 26 0.908 0.0364 0.917 0.932 July Maytag 1999 7 0.873 0.917 0.932 June 1130 0.1339 �20 Table 4. Estimates of boundary strip width (BSW) for nonlinear weighted least squares fit and nonlinear weighted least squares fit with the logistic transformation. Method Estimate of BSW SE of estimate Nonlinear weighted least squares 51.1241 10.1033 Nonlinear weighted least squares with logistic transformation 41.5446 9.1676 Table 5. Data used to evaluate the relationship between Preble's meadow jumping mouse density and riparian vegetation cover. Adjusted Area (ha/km) Adjusted SE Stream InvestiStudy A..rea Year Density Adjusted Length HerbaNonOpen Tree gator (mice/km) Density Shrub (km) ceous Vegetated Water Shenk Maytag 1998 32.52 5.84 0.2619 2.4643 2.8571 1.3452 2.0357 0.84 Shenk Maytag 1999 29.04 3.05 0.1549 1.8239 1.9296 1.0000 1.7817 1.42 Shenk Pinecli.ff 1998 57.35 18.39 5.4324 3.3649 1.2432 0.0000 1.5946 0.74 Shenk Pinecli.ff 1999 53.79 5.52 5.4324 3.3649 1.2432 0.0000 1.5946 0.74 Shenk Woodhouse 1998 15.52 3.14 2.7600 0.0000 1.7200 0.0000 0.0000 0.50 Shenk Woodhouse 1999 48.52 3.52 2.7600 0.0000 1.7200 0.0000 0.0000 0.50 Bakeman Dirty Woman Cr. 1998 20.46 7.29 2.9190 1.3646 0.2987 0.0861 0.0354 3.95 Bakeman Dirty Woman Cr. 1999 6.89 2.11 1.3769 1.8396 0.4403 0.0000 0.0522 2.68 Bakeman Castle Rock 1999 41.70 26.91 3.8086 0.1584 0.2145 1.0066 1.1683 3.03 Schorr Monument Cr. 1998 67.08 13.34 5.9363 0.0000 0.4968 0.8535 0.0000 1.57 Schorr Monument Cr. 1999 28.75 8.52 5.9363 0.0000 0.4968 0.8535 0.0000 1.57 Meany South Boulder Cr. 1998 48.80 10.60 0.1640 27.4270 3.8629 0.0000 0.1933 4.45 Meany South Boulder Cr. 1999 34.50 7.75 0.1640 27.4270 3.8629 0.0000 0.1933 4.45 Ryon Walnut Cr. 1999 5.10 3.34 0.1231 1.0865 0.0416 0.0000 1.0266 6.01 Ryon Rock Cr. 1998 3.80 0.36 0.3815 0.9505 0.0000 0.0786 0.0219 14.13 Table 6. Results of linear regressions of Preble' s meadow jumping mouse density (mice/km stream ) predicted by riparian vegetation cover (ha/km stream) for 15 study area and year combinations from Table 5. Variable Shrub Herbaceous Tree Non-vegetated Total Cover Open Water Intercept 30.22 31.79 30.80 31.51 31.23 35.84 p R2 0.06 0.67 2.18 0.06 0.638 0.656 0.495 . 0.729 0.596 -2.35 .0.469 0.018 0.016 0.037 0.011 0.022 0.041 Slope 0.63 --· �21 Table 7. Summary of AICc model selection for the riparian vegetation cover (ha/km stream) variables Shrub, Tree, Herbaceous, Non-Vegetated, and Open Water predicting Prebles' meadow jumping mouse density (mice/km stream). AICc, ~-AICc, and Akaike Weights follow definitions in Burnham and Anderson (1998). Akaike R2 AICc ~-AICc Variables in Model Wei hts 83.5040 0.6831 0.0000 0.59584 Shrub Tree 86.6162 3.1122 0.7143 0.12570 Shrub Tree Open Water 87.4205 3.9165 0.08408 0.6986 Shrub Tree Non-Vegetated 88.1465 4.6425 0.6836 0.05848 Shrub Tree Herbaceous 89.0245 5.5205 0.5421 0.03770 Shrub Herbaceous 89.3639 5.8599 0.6569 0.03182 Shrub Herbaceous Open Water 90.9029 7.3989 0.6198 0.01474 Shrub Herbaceous Non-Vegetated 91.1034 7.5994 0.3216 0.01333 Shrub 91.8340 8.3300 0.7258 0.00925 Shrub Tree Herbaceous Open Water 92.3015 8.7975 0.7171 0.00732 Shrub Tree Non-Vegetated Open Water 92.8597 9.3557 0.7064 0.00554 Shrub Tree Herbaceous Non-Vegetated 93.5258 10.0218 0.3819 0.00397 Shrub Open Water 94.1043 10.6003 0.6810 0.00297 Shrub Herbaceous Non-Vegetated Open Water 94.5170 0.3396 11.0130 0.00242 Shrub Non-Vegetated 95.4143 11.9103 0.0957 0.00154 Tree 0.0462 96.2135 12.7095 0.00104 Open Water 96.3496 0.0375 12.8456 0.00097 Herbaceous 96.3621 12.8581 0.0367 0.00096 Non-Vegetated 0.3828 98.1688 14.6648 0.00039 Shrub Non-Vegetated Open Water 0.1358 98.5520 15.0480 0.00032 Tree Non-Vegetated 98.7044 15.2004 0.1270 0.00030 Tree Open Water 98.8480 15.3440 0.7345 0.00028 Shrub Tree Herbaceous Non-Vegetated Open Water 0.1059 99.0626 15.5586 0.00025 Herbaceous Non-Vegetated 0.1035 99.1022 15.5982 0.00024 Tree Herbaceous 99.1919 15.6879 0.0982 0.00023 Herbaceous Open Water 99.8370 0.0585 16.3330 0.00017 Non-Vegetated Open Water 0.1459 103.0432 19.5392 0.00003 Tree Non-Vegetated Open Water 0.1358 103.2184 19.7144 0.00003 Tree Herbaceous Non-Vegetated 103.3266 19.8226 0.00003 0.1296 Herbaceous Non-Vegetated Open Water 0.1273 103.3666 19.8626 0.00003 Tree Herbaceous Open Water 108.8576 25.3536 0.1470 0.00000 Tree Herbaceous Non-Vegetated Open Water Table 8. Parameter estimates for best AI Cc model explaining PMJM density (mice/km stream length) from vegetative cover variables (ha/km stream length). Standard Pr> jtj Variable t statistic Estimate Error 0.892 Intercept 0.14 0.9853 7.1047 Shrub Tree 7.2342 10.1308 1.5339 2.7381 4.72 3.70 <0.001 0.()03 �22 Table 9. Number of Preble's meadow jumping mice radi~llared at each study site during each trapping session for each l'.ear. 1999 1998 Site Session Trap nights Females Trap Nights Females -Males Tot.al Males Tot.al Maytag Jun Maytag Jul Maytag Sep PineCliff Jun PineCliff Jul PineCliff Sep Woodhouse Jun Woodhouse Jul Sep Woodhouse TOTALS 1949 2370 3256 1095 1603 1544 1525 2420 1568 17330 9 11 10 5 6 4 2 8 7 62 5 5 8 13 7 12 3 4 6 63 14 16 18 18 13 16 5 12 13 125 2604 3191 4341 742 1442 1921 1651 1215 2114 19222 6 10 3 7 6 7 7 8 3 57 12 13 7 .5 10 17 9 5 3 81 18 23 10 12 16 24 16 13 6 138 Table 10. Preble's meadow jumping mice radi~llared. and tracked for more than one tracking session. Location of capture, sex, identification, and number of locations for each mouse during each tracking session is noted. 1998 1999 PIT-tag# Site Sex Sep-Oct Jul-Aug Jun Jul-Aug Sep-Oct Maytag 414lf87C76 60 24 F 105 Maytag F 41412B4A2D 103 Maytag F 413E296246 73 7 Maytag F 4141276D78 76 3 82 Maytag F 4140753620 1 93 · Maytag M 4141241836 1 Maytag 4140754707 19 M 70 4135354E37 Maytag M 37 40 Maytag 41357B5A0E 38 M 49 61 13 Maytag M 414130663A 144 11 65 Maytag M 41411E6879 Maytag 4141445465 M 52 74 4141554B28 PineCliff F 44 21 PineCliff M 41413D4C04 18 15 4141144535 PineCliff M 21 38 PineCliff 413E02020F M 22 9 PineCliff M 414037184D 45 37 87 6 PineCliff M 4141714141 12 47 Woodhouse F 41413E610B 78 14 Woodhouse F 41413E7B07 5 66 63 16 Woodhouse F 41414C721F 39 Woodhouse F 41412A2508 84 93 Woodhouse 4141611413 42 8? M 46 Woodhouse M 4141074013 6 Woodhouse M 4141252D66 63?NC 64 Woodhouse M 41412D3509 59 ?NC Woodhouse 75 M 41406A1049 9 �23 Table 11. Causes of mortality of Preble's meadow jumping mice from Maytag Property, PineCliff Ranch, and Woodhouse Ranch collected during; the 1998 and 1999 field season (June I-October 31). 1998 1999 Cause of Total Mortality M.aytag PineCliff M.aytag Woodhouse PineCliff Woodhouse Predation Bullfrog Garter snake Rattlesnake 0 0 1 1 l 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 3 0 1 0 0 0 0 Yellowbellied racer Weasel Fox House cat Raptor Unknown Exposure Drowning Road.Kill Handlingffrap 0 Unknown TOTAL 1 8 1 4 5 0 0 0 2 0 0 0 0 0 1 0 0 l 0 0 1 0 1 0 0 0 1 0 0 0 2 2 1 0 0 19 4 9 0 5 2 0 1 6 3 0 2 3 24 13 17 1 7 2 0 0 0 2 2 8 4 5 3 0 0 2 0 1 ,--... 8 ell 5 1 1 1 4 1 9 17 74 ■ +---..-------------------------------------------------------1 ■ 0.9 -----=-----------===== --- Q.) < 0.8 --+-----~-----1- Q.) i::::: ·.§< 1--< i::::: 0 0.6 0 ·...... 0.5 ~ ,.D 0 1--< ~ ■ +------------------------------ b .......-< ■ 0.7 ■ -- - 0.4 - + - - - - - - - - I 0.3 200 --------------------------1 ■ - Observed Data Weighted Least Squares Weighted Least Squares Logistic 400 600 I I I 800 1000 1200 Trapline Length Figure 1. Observed estimates ofp with the fit of the weighted nonlinear least squares model, and the logistic transformed nonlinear least squares model. �24 1 0.9 0.8 ~ .-9 0.7 ~ 0.6 0.5 ---······························· 0.4 0 500 1000 1500 Trapline Length (m) 2000 Figure 2. Predicted p with 95% confidence intervals as a function oftrapline length in meters. 20 - . - - - - - - - - - - - - ~ • onument Creek •C~ouse 10 r✓.l 1 0 ·~ -10 r✓.l -20 • Maytag •Walnut Creek • Rock Creek • South Boulde Creek • Pinec. • Maytag • Pinecliff • Dirty Woman Creek . • South Boulder Creek • Dirty Woman Creek • Monument Creek •Woodhouse -30 - - - . + - - - + - - + - - - + - - - - - - - - + - - - - - + - - I 0 10 20 30 40 50 60 70 PMJM Density (mice/km) Figure 3. Residuals for the best AICc model explaining PMJM density (mice/km stream length) from the standardiz.ed vegetation cover variables shrubs and trees (ha/km stream length). � https://cpw.cvlcollections.org/files/original/757faa42748d04dc7c7887433dd58292.pdf dc54b75eddbf976e8545043ce4dbfe20 PDF Text Text 335 JOB PROGRESS REPORT State of Colorado Project No • .....e.W~-~1~5~3~-_R,.,__-~s'--------- Mammals Research Work Plan No. _ ___,l~0~A,.,__ _ _ _ _ _ __ _ Kit Fox Studies Job No. Kit Fox Status in Colorado 1 Period Covered: July 1, 1992 to June 30, 1993 Author: J.P. Fitzgerald and M. .Link, Univ. Northern Colorado, Greeley Personnel: T. Beck, c. Parmeter, (volunteers M. Reddy, T. Hugo from University of Northern Colorado). ABSTRACT Since 9 March 1992, 9 kit fox have been trapped in 2,725 trap nights of effort including 1,141 trap nights in the current reporting period. All individuals (5 males and 4 females) have been captured NE of Montrose in Delta and Montrose counties in desert-shrub/greasewood habitat. Three males and 4 females have been radio-collared and are periodically being monitored. One additional kit fox has been observed while spotlighting in Browns Park in extreme northwestern Moffat county . Den sites used by foxes are in steeper, rougher terrain than reported in the literature . Few kit fox have been reported harvested by trappers responding to the annual survey questionnaire distributed by the Colorado Division of Wildlife. Contacts with ranchers, land management agency personnel, CDOW employees, and local trappers have resulted in only three reports of kit fox in western Colorado. Live trapping efforts in areas from which kit fox have been historically reported have not resulted in any captures or visual observations. �337 KIT FOX (WLPES MACROTIS) DISTRIBUTION AND ABUNDANCE IN WESTERN COLORADO James P. Fitzgerald and Michelle Link P.N. Objective Document the geographic distribution and relative abundance of kit fox in western Colorado. Segment Objectives 1. 2. 3. 4. 5. Identify geographic extent of kit fox distribution in western Colorado. Determine habitat use by the species and general prey species abundance in areas used by kit fox. Compile general harvest information. Test different survey techniques in areas with kit fox populations. Identify suitable kit fox populations for more intensive ecological research. Methods Methods are generally the same as previously reported (Link and Beck 1992). However, captured kit fox are now being collared with transmitters (Model No. HLPM 2180 LD - Wildlife Materials, Inc.) for monitoring purposes. Results and Discussion Determination of Distribution Since the project began 9 March 1992 a total of 2,794 trap nights of effort have been expended (Table 1) resulting in capture of 9 kit fox. Seven of the captured fox, 3 males and 4 females have been radio-collared. Two of the females have been recaptured and their collars have been replaced. One adult female captured on 31 May 1992, rehabilitated in captivity with a broken jaw, and radioed and released to the wild in late summer 1992, was found dead from injuries (probably inflicted by a coyote) in mid December 1992. All animals have been captured in a 2590 ha area of Peach Valley in Montrose and Delta counties. Live trapping (1,210 trap nights) and spotlighting efforts (54 hours) during the fiscal year were concentrated in the Big Gypsum, Disappointment, Paradox and Peach valleys in southwestern Colorado and the Rangely-Dinosaur and Brown's Park areas of northwestern Colorado. Because of the capture of kit fox in the Peach Valley area in May and June of 1992 the area continued to receive periodic, concentrated search effort during the present reporting period. One animal identified by Link as a kit fox was observed by spotlight in the Browns Park area in late June 1993 but no animals were captured during 144 trap nights of effort in that location. Concerted effort will be expended in that area in 1993-94 to attempt to capture and verify presence of fox in Moffat County. Of particular interest is whether animals in Moffat County are "Swift or Kit" foxes as it is possible that "Swift" fox could occupy the "Red Desert" area of Wyoming and filter south into Moffat County rather than represent extensions of Kit fox populations from the Colorado River drainage of Utah. Sampling has included trapping in a number of different habitat types including: sagebrush-grasslands; margins of pinyon-juniper woodland, and desert-shrub/greasewood areas. To date, kit fox have only been captured in the desert-shrub/greasewood type common to the Peach Valley area. Much of the area occupied by kit fox in Peach Valley is dominated by mat saltbush (Atriplex corruaata) on the uplands and ridges with greasewood (Sarcobatus �338 Table 1. Areas searched, dates of search, and trap nights of effort spent in trapping, kit fox study, western Colorado, 1992 to June 30, 93. County and Area Montezuma Co. McElmo Canyon San Miguel Co. McIntyre Canyon Disappointment Valley Big Gypsum Valley Montrose Co. Paradox Valley Dates Searched Total Trap Nights Effort Fox Captures 100 Trap Nights 9/9-12/92 60 0 3/5-8/93 77 0 8/3-9/92 141 0 7/5-8/92; 7/29-8/1/92 174 0 8/10-20/92 141 0 168 0 192 836 0 1 Montrose co., Delta & Mesa co. Sinbad Valley 3/15-22/92 Delta to Grand 4/28-5/16/92 Junction Various - May to Present Peach Valley V Mesa co. Dolores River Valley-Gateway to Utah border 3/6-12/92 114 0 Colo. Natl. Mon. 4/20-23/92 78 0 N.W. Grand Junct. & Rabbit Valley Various - 1992-93 528 0 Rio Blanco Co. Rangely/Dinosaur 6/16-19/93 72 0 Moffat Co. Brown's Park 6/22-28/93 144 0 vermiculatus), Nuttall saltbush (A. nuttallii), shadscale (A. confertifolia, and sagebrush (Artemisia .!ll2.:.) occurring in more localized sites. A considerable amount of the Peach Valley area occupied by foxes is on lands used heavily by off road vehicles (ORV's). Habitat Use and Relationship of Prey Abundance to Kit Fox Abundance Determining the diets of kit foxes in the Peach Valley area will be part of the 93-94 work effort. Link has collected numerous kit fox scat samples from Peach Valley. Remains of white-tailed prairie dogs (Cynomys leucurus), cottontail (Sylvilagus audubonii), and cricetid mice (Peromyscus) have also been found at kit fox dens. No usable relationship appears to exist between numbers of lagomorphs spotlighted at night and distribution of kit fox. An average of .2 lagomorphs (mostly cottontails) have been recorded per road km in over 916 km surveyed (range 1.6 to 0 animals/km). In Peach Valley the average has also been .2/km in a total of 68 km censused. Lagomorph populations have been very low across V �339 western Colorado with highest numbers observed in the Sinbad Valley, Mack and Brown's Park areas. Den sites located in the Peach Valley area are not similar to those reported by other workers (Egoscue 1962, O'Neal et al. 1986, McGrew 1977) who have found most dens on relatively flat ground with multiple burrow entrances. out of 13 dens being used in Peach Valley, 4 are located in deep, dry washes, 4 are on the sides of hills with good drainage, two are half way up hills in washes, two are on top of ridges at 5,600 feet, and one is on the side of a large u-shaped bowl. Most dens are within 300 m of a well-traveled road, jeep trail, or pathway (ORV, horse). Use of these "atypical" den sites has led us to revise search efforts to focus on more broken terrain than suggested by the literature. General Harvest Information/ Other Reports of Kit Fox Table 2 indicates reported harvest data for kit fox from western COlorado. Take is estimated from trapper survey data and numbers may include other fox species. The numbers of animals trapped in Montezuma county in 1981, and reports of kit fox from La Plata county are doubtful. Table 2. Estimated harvest of kit fox from western Colorado based on trapper questionnaire returns to the Colorado Division of Wildlife, 1975-1992. Year \.__I County Delta Garfield Gunnison La Plata Mesa Moffat Montezuma Montrose Rio Blanco Totals 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 2 5 2 3 6 2 3 1 13 33 5 2 2 10 7 1 2 33 41 3 192 198 26 4 10 5 5 4 41 13 6 2 Link has made a concerted effort to question ranchers, agency personnel, or local trappers regarding their knowledge of kit fox. Few reports have been made. The district wildlife manager (DWM) in the Mack/Grand Junction West area reported seeing young kit fox in an arroyo north of Mack in 1988. Extensive trapping has not resulted in any captures in that area. The Fish and Wildlife Service at Browns Park National Wildlife Refuge had no records of kit foxes, but one employee (Brad Pech) reported that trappers had taken kit fox in that area several years ago. Link did spotlight an animal believed to be a kit fox in that area but none were trapped. The DWM (Divergie) and the Chevron Oil Company's on-site environmentalist (Sellars) at Rangely have never observed or heard of kit foxes in that area even after extensive spotlighting and surveying for Black-footed Ferrets. Ranchers (Tozer Bros.) from McElmo Canyon, Montezuma County, have not observed any kit fox and trapping efforts did not produce captures in 1992 in that area, although there are past records of kit fox in that area (Egoscue 1964). Likewise, trapping efforts in other areas reported to have kit fox by Miller and McCoy (1965) including the Colorado National Monument and near the Utah state line in the Grand Valley) have not resulted in any captures. The BLM ranger (Sering) assigned to the Delta and Montrose area reported to Link that he observed a kit fox on Flattop Mountain south of the Peach Valley area in June of this year. The Flattop area will be trapped and surveyed in 1993-94. �340 Test sampling techniques in areas with Kit Fox The Peach Valley area is the only site with a known population of kit fox. Those animals have been located by daytime sightings at dens close to roads or by live-trapping. Trapping success in Peach Valley is low with one capture per 100 trap nights effort. Two of the radio-collared animals (both females) have been recaptured 5 months after their first capture allowing for replacement of their radios. Males appear to be more difficult to recapture using present techniques. V Scent stations baited with standard lure (scented predator survey discs with FAS, Pocatello Supply Depot) were placed at 10 sites in Peach Valley in June 1992. No kit fox tracks were identified at any stations. The scent station surveys were abandoned because it was more efficient for one person to work a trap line than to maintain a station line. Spotlighting has resulted in observations of 2 kit fox in Peach Valley and 1 in Browns Park. Identify suitable kit fox populations for more intensive research The Peach Valley area bas from 4-8 adult fox based on monitoring of radioed animals. Since these animals represent the only kit fox presently known to occur in western Colorado the decision was made to radio-collar all animals captured on the site and to increase efforts to better understand habits of the animals in that area with the expectation that it may enhance efforts to locate other kit fox populations. A temporary worker (Parmeter) was hired from mid-January to mid-April to monitor fox in Peach Valley and to try winter trapping in the Grand Junction and McIntyre Canyon areas. Attempts to radiotrack animals at night in the Peach Valley area by both Link and Parmeter have not been effective for determining distances ranged from dens or areas being hunted. The foxes are aware of the researcher and their behaviors seem to be more intent in evading the tracker rather than on night hunting. In the 93-94 field season attempts will be made to track some radioed animals using teams of trackers stationed at high points of ground rather than a single individual. ~~~ti/res P. Fitzgerald Contractor, UNC and Michelle Link Graduate Research Assistant V �1 93 Colorado Division of Wildlife Wildlife Research Report June 30 1994 JOB PROGRESS REPORT State of -----~C~o~l~o~r~a~d;o=---Project No. W-153-R-7 Mammals Research Work Plan No. 10A Kit Fox Studies Job No. 1 Kit Fox (Vulpes macrotis) Status in Colorado Period Covered: Author: July 1, 1993 - June 30, 1994 J.P . Fitzgerald Personnel: J.P. Fitzgerald, M. Link, T. Verbeck, A. Anderson, L. Dent, J. Eussen, J. Prather, M. Reddy, D. Watson, T. Beck, B. Gill Abstract A total of 1754 trap nights of effort were expended in the project year to capture 18 kit fox. Trapping efforts included parts of Moffat, Mesa, Garfield, Delta, and Montrose counties. All captures were made in Montrose and Delta counties in Peach Valley (8), East of Montrose (9), or North of Delta (1). A reliable report of a "kit or swift" fox was received from U.S. Fish and Wildlife Service and BLM personnal in the Vermillion Creek area in northern Moffat County. Trapping to date in northern Moffat County has not resulted in any fox captures. Monitoring of radio-collared kit foxes in the Peach Valley study area in Montrose and Delta county continued. Four (F18, FS, M13, MS) of 6 radio-collared adults are still alive in that population. On 26 March and 13 April carcasses of two kit fox, M12 and an untagged animal of unknown sex, were recovered in TS0N, R9W, NE corner Section 8 within 100 m of each other. The cause of death could not be determined for either animal. East of Montrose, nine kit fox (4 adult females, 1 adult male, 1 male pup, 3 female pups) were captured and either radio-collared or ear tagged. Gill and field crew personnel have taken photographs of animals in that population. In March, the Wildlife Commission closed the season on kit fox and imposed trapping regulations for the Peach Valley area. The Director of the Division of Wildlife placed the species on the state special concern list. �195 KIT FOX (VULPES MACROTIS) STATUS IN COLORADO James P. Fitzgerald P. N. Objective Document the geographic distribution and relative abundance of kit fox in Western Colorado. Segment Objectives 1. Continue to monitor radio-collared foxes in the Peach Valley Area. 2. Write up_project results (M.A. thesis - Michelle Link in progress). 3. Begin extensive summer trapping effort. Methods Methods continue to be similar to those reported previously (Fitzgerald and Link 1993, Fitzgerald and-Verbeck 1993).·Field Personnel; At the start of the project year Michelle Link was replaced by Tom Verbeck to take over the kit fox field research. Verbeck worked from mid-August to January. Allan Anderson monitored fox populations in Peach Valley from March through late May. Verbeck resigned from the project in April. Three field crews: Matt Reddy-David Watson; Lonnie Dent-Jim Bussen; and Jake Prather (alone) began work on the project in late May. Results and Discussion Trapping Efforts: Between July 1, 1993 and June 30, 1994 a total of 1754 trap nights of effort have resulted in a total of 20 captures of 18 individual kit fox. All captures were made in Montrose or Delta County. Trapping efforts and locations are summarized in Table 1. In areas with known kit fox populations (Peach Valley, North of Delta, and Montrose East) a total of 1012 trap nights of effort-were expended in trapping the 18 animals captured (56 trap nights/fox). The capture rate of foxes in.traps placed near(< 200 m) known fox dens is much higher. The 18 individual animals were captured in a total of 45 trap nights, a capture rate of 2. 5 trap nights per ·fox. It appears that the methods we are employing catch kit foxes efficiently if foxes are present. It appears kit fox populations across the areas surveyed are very low resulting in many trap nights of effort with no captures. In comparing this years trapping effort to efforts by Link, she captured 9 kit fox during 2,725 trap nights of effort, with all 9 animals captured during 836 trap nights (93 trap nights/fox) of effort in Peach Valley (Fitzgerald and Link 1993). • The SW Regional Office of CDOW has hired Mr. Doug.McCauley, a taxidermist from Delta, to work with the project crew to find additional kit fox populations. Mr. McCauley believes that many more kit foxes exist than our trapping results would indicate. The understanding with Mr. McCaughley is that he will try to locate kit fox populations using predator calling and night lighting. He has to show either the project crew or someone from the SW Regional Office the exact location of occupied dens and verify the presence of kit fox in order to be paid his fee. To date he has not taken the crew out but has indicated to Dent and Eussen that he knows of 2 or 3 active dens on the Colorado-Utah border. We are still trying to schedule a time when he can go out with us. �196 Table 1. Areas searched, dates of searches, and trap nights of effort, kit fox study, western Colorado, July 1, 1993 - June 30, 1994. Area Searched and County • Dates Searched Number of Trap Nights Peach Valley Montrose Co. 8-29 to 12-6-93 4-5 to 4-24-94 232 139 0 North of Delta Delta Co. 11-19-93 to 1-6-94 5-27 to 5-30-94 245 39 0 284/1 fox Montrose East Montrose Co. 5-27 to 6-18-94 357 9 357/9 foxes West Gunnison Gorge. Delta Co. 12-14 to 12-16-93 60 0 60/0 SEW of Wells Gulch. Delta Co. 6-21 to 6-27-94 230 0 230/0 Brown's Park Moffat co. 10-5 to 10-12-93 151 0 151/0 Grand Junction NW, Mesa Co. 10-20 to 10-28-93 129 0 129/0 172 0 172/0 1754 18 1754/18 DeBeque/Parachute 5-30 to 6-7-93 Mesa/Garfield Co's Totals Captures Total Trap Nights and Success by Area 8 371/8 foxes l V Trapping and Monitoring Effort - Peach Valley Population: Verbeck from 29 August to 6 December conducted 232 trap nights of effort in Peach Valley resulting in 8 captures of foxes: M13 (captured twice), Fl4, MlS, M16, MB, F18, and an animal that escaped before it could be sexed or marked (Table 2). Animals which were recaptures (F14, MlS, MB, F18) were given new radio-collars and ear tags. Invariably ear tags were torn out on most of the foxes recaptured. Anderson conducted 139 trap nights of effort in Peach Valley from 5 April to 24 April with no animals captured and had difficulty picking up radio signals from collared animals .. Anderson picked up radio-signals from 5 ( Fl8, M12, MS, FS, M13) of the 6 adults radioed in the area. However, he was unable to locate their precise den sites because of field conditions and problems with radio-receivers. on 26 March, Anderson recovered the carcass of Ml2 within a hundred meters of its den but could not verify cause of death. He recovered the carcass of a kit fox that had no ear tags or collar about the same distance from that same den on 13 April. Again it was not possible to estimate cause of death. F14 has been missing since January unless she is represented by the carcass (no radio-collar found) located near Ml2 - they were both trying to pair in December and January. Field crews verified presence of M13, Fl8, MS on 19· June in T51N, R9W, S29,30. This is 4-6 Jan north of locat~ons Verbeck found them ·using in late fall ·and early· winter (Fitzgerald and Verbeck 1993). They also reported picking up good signals in TS0N, R9W, S4 from 150.037 radio - a radio frequency that Anderson attributed to Ml2 found dead on 26 March. This error has to be investigated and resolved, and attempts made to recapture and recollar an;mals in the valley. Field crews report that Ml3 is with an apparently unmarked female in a den with 4 pups close to the water tank in lower (north) Peach Valley. They have also seen several V �197 \...-I foxes crossing the road at night in that same part of Peach Valley. Attempts will be made later this summer to capture and radio those animals. Montrose East Population: Anderson on 23 May received word of a family of foxes along the Landfill Road (Bostwick Park Road) in T49N, RSW, NE7 and began to observe animals in early morning and late afternoons. He estimated as many as 12 animals in the area including pups of the year. He showed the summer field crews the site on 27 May. Since 28 May a total of 9 kit fox have been trapped and marked in that population (Table 3). The population includes several adult females which may represent yearling animals that did not breed. The whelping den included pups from apparently two different litters based on size differences and estimated numbers of pups (7-8). The Montrose East population is located southeast of Flattop Mesa. Habitat is similar to Peach Valley and there is nothing to preclude mixing of the two populations which are about 12 km apart. However, trapping efforts from Flattop Mesa north to upper Peach Valley have not resulted in any other fox captures. The large numbers of females present in the population are encouraging provided sufficient males are available for mating in 1994-95. Several enthusiastic local residents are aware of this population and watch it carefully attempting to protect animals from being shot since they are located close to a heavily used county road. Table 2. Kit fox trapped in Peach Valley, Montrose County, Colorado, 1992-July 1, 1994 and dates of last sightings/radio-signals. CR=Collar replaced; ETR = Ear tags replaced. ~- \...,,I Capture or Date ·Located Sex 5/14/92 . M 5/31/92 11/23/92 F 6/25/92 Ear Tag Number Weight Kg Radio Collar Frequency Location Fate 2.3 not collared T51N/R9W/S29 Unknown 1 2.0 150.238 T51~/R9W/S29 Tl5S/R94W/S27 Killed by Coyote M 4 2.2 not collared T50N/R9W/Sl6 Unknown 9/28/92 M 2 2.7 150.309 T50N/R9W/S9 Unknown 9/28/92 3/1/93 1/3/94 4/6/94 F 5 2.5 150.338 CR150.956 T50N/R9W/S9 T50N/R9W/Sl7 T50N/R9W/S17 T50N/R9W/S9 Signal Weak 9/28/92 3/2/93 9/22/93 1/6/94 F 2.6 150.379 CR150.850 CR150.889 T50N/R9W/S9 T50N/R9W/S9 T50N/R9W/S16 T50N/R9W/Sl7 Unknown 2/23/93 .10/26/93 12/6/93 1/6/94 3/9/94 5/22/94 6/19/94 F 7 2.5 150.638 ETR18 2.6 CRlS0.594 T51N/R9W/S29 T50N/R9W/S5 TSON/R9W/SS TSON/R9W/S5 T50N/R9W/S16 TSON/R9W/S18 T51N/R9W/S29 Alive 2/23/93 10/26/93 12/6/93 1/6/94 3/26/94 M 6 ETR14 8 2.4 3.0 150.468 3.1 CRlS0.189 T51N/R9W/S29 TSON/R9W/S5 TSON/R9W/S5 TSON/R9W/SS TSON/R9W/S5 �198 Table 2 cont. 5/21/94 6/19/94 4/21/93 9/24/93 10/23/93 TS0N/R9W/S5 TS1N/R9W/S30 12 ETRlS M 1/6/94 3/26/94 9/20/93 11/9/93 12/14/93 3/31/94 13 M 2.5 2.8 2.7 150.708 CR150.037 150.813 5/21/94 6/19/94 9/30/93 16 M 2.3 150.499 11/22/93 Alive T50N/R9W/S22 TS0N/R9W/S8 TS0N/R9W/S17 TS0N/R9W/S16 TS0N/R9W/S8 TS0N/R9W/S16 Dead T50N/R9W/S5 T51N/R9W/S29 Alive T50N/R9W/S17 TS0N/R9W/S17 Unknown TS0N/R9W/S16 T50N/R9W/S16 TS0N/R9W/S5 Table 3. Kit foxes trapped in the Montrose East population, Montrose County, 1994. NL= non-lactating females not pups of the year; L = lactating. Location Fate Weight Kg Radio Collar Frequency F(NL) 21 recaptured 2.5 150.947 ·T49N/R9W/S7 F(NL) 22 2.5 150.403 T49N/R9W/S7 Alive 5/29/94 F(NL) 23 2.25 150.33.8 T49N/R9W/S7 Alive 5/29/94 M pup 24 1.75 not-collared T49N/R9W/S7 . Alive 5/30/94 F pup Ll76/Rl77 1.5 not-collared T49N/R9W/S7 Alive 5/30/94 F pup Ll78/R179 1.45 not-collared T49N/R9W/S7 Alive 5/31/94 6/7/94 F (L) Ll80/Rl81 2.3 recaptured 150.940 T49N/R9W/S7 T49N/R9W/S7 Alive 6/1/94 F pup Ll83/Rl84 1.8 not-collared T49N/R9W/S7 Alive 6/3/94 M Ad not-collared T49N/R9W/S12 Alive Capture or Date Located sex 5/29/94 5/30/94 5/29/94 Ear Tag Number L185/R186 3.0 Alive V Use of Baits: Table 4 summarizes the baits used in traps which captured kit fox. Verbeck made a detailed analysis of what baits were used during his 817 trap nights of effort and we are following that procedure this field season. Anderson conducted 139 trap nights of effort in Peach Valley with no successful captures during the month of April. During that period he tried a variety of baits including commercial fox scent, sardines, commercial raccoon-fox scent, and dead chicks. Be also put out carcasses of dead white leghorn chickens near several dens in the Valley with foxes paying little attention to such carrion. Since late May crews have conducted an additional 798 trap nights using a variety of baits. During V �199 trapping efforts at East Montrose, emphasis was given to taking as many animals as possible in as few a trap nights as necessary. As a result of that effort many traps were baited with multiple baits consisting of road killed cottontails and/or prairie dogs and turkeys with the idea being that the foxes might respond best to a variety of fresh killed meats, a technique that worked well. Although baits from local road-killed mammals and birds might work better than turkeys or turkey-lure combinations the numbers of traps being run precludes exclusive use of such local carrion. Table 4. Summary of kit fox captures/recaptures and types of baits used. May 14, 1992 to June 30, 1994. Investigator/Year Link/ParmetQ.r 92 to July 93 Verbeck Aug 93 to Jan 94 Present Crews April - Present Baits cottontail Cottontail/Turkey Cottontail/Turkey/P. Dog Prairie Dog/Turkey Prairie Dog Turkey Turkey/Commercial Lure Pheasant Turkey/Big Game Commercial Lure or w Sardines Chicken/commercial Lure Total Trap Nights Effort l/<50 5/77 3/41 5/70 0/39 4/10 0/6 6/>2,600 2/<200 0/408 2/679 1/32 0/66 0/100 0/39 2,725 817 937 Other Activity: Based on recommendations made by Fitzgerald and presented to CDOW personnel 11/15/93, the Colorado Wildlife commission on 3/10/94 closed the season on kit fox in Colorado and imposed trap and trapping restrictions for the Peach Valley, area. The Director of the CDOW on 22 March, placed the kit fox on the list of species of "special concern." The agency is presently contemplating future management, including whether it should be listed as "thr~atened." A meeting of CDOW personnel and the contractor will take place in late summer 1994 to discuss these issues. Link is finishing her thesis incorporating some of Verbeck' s findings into the paper. Gill has photographed animals in the Montrose Bast group. �253 Colorado Division of Wildlife Wildlife Research Report July 1995 JOB PROGRESS REPORT State of Colorado Project No. W-153-R-7 Mammals Research Work Plan No. 10A Kit Fox Studies Job No. Period Covered: Kit Fox (Vulpes macrotis) status in Colorado 1 July 1, 1994 to June 30, 1995 Author: _J.P. Fitzgerald Personnel: J.P. Fitzgerald, L. Dent, J. Eussen, M. Link, J. Prather, M. Reddy, D. Watson, R. Basagoitia, S. Boyle, R. Hays, T . Beck, B. Gill., J. Olterman ABSTRACT In FY1994-95, 2326 trap nights of effort resulted in 46 captures of 31 individual kit foxes . Eighteen were animals not previously captured. Since work began in March 1992, 6805 trap nights of effort have resulted in capture of 38 kit foxes : 10 adult males, 7 juvenile males, 14 adult females, and 6 juvenile females. Hundreds of square kilometers of what appears to be suitable habitat is unoccupied or occupied at low levels. Trapping was conducted in parts of Mesa, Garfield, Delta, Montrose, and Moffat counties. Captures were made in Mesa and Garfield Counties (10 kit fox), and in Montrose and Delta Counties (21 foxes). Most captures were in the Peach Valley and Montrose East areas. No captures were made in Brown's Park, Moffat County in 171 trap nights of effort. Attempts to locate foxes during the winter using ATV's were not successful due to lack of snow. Of 37 collared or ear tagged foxes caught since 1992, the fate of 23 is unknown, 7 are dead, 7 are alive. Movement of 3 foxes between Montrose East and Peach Valley was documented. Home ranges of 2 Peach Valley and Montrose East foxes were estimated to be from 1.5 to 7.7 km • Foxes used an average of 3.7 dens per individual during the tracking period. Reproductive success is low. Of 14 adult females captured since 1992 only 6 litters of pups have been produced. The Wildlife Commission expanded the area with trap restrictions to portions of Mesa, Garfield, and western Delta counties. Plans for the next fiscal year are to complete the search effort, finalize reports, and make recommendations for any continued research on the species. m~nm1~111,1~ i~m BDOWD1 □ 765 �255 KIT FOX (WLPES MACROTIS) STATUS IN COLORADO James P. Fitzgerald P. N. Objective Document the geographic distribution and relative abundance of kit fox in Western Colorado. Segment Objectives 1. 2. 3. 4. Continue to monitor radio-collared foxes in Montrose, Delta, Mesa and Garfield counties. Continue search for new kit fox populations. Complete write-up of earlier field study effort (M. Link Thesis). Complete plans for 1995-96 field efforts. Methods Field methods for live-trapping were similar to those reported previously (Fitzgerald and Link 1993, Fitzgerald and Verbeck 1993). Live-trapping effort was concentrated in the Gunnison and Colorado River drainages. Boyle radiotracked foxes in fall and winter in the Montrose East and Peach Valley areas to estimate home ranges and nightly movements. Boyle, Basagoitia, and Hays searched for fox tracks in fall and winter months using ATV's. Dent, Eussen and Hatch have continued the search effort and monitored radioed animals during this spring and early summer. Results and Discussion LIVE TRAPPING EFFORTS: A total of 2326 trap nights of effort resulted in 46 captures of 31 different individuals (Table 1). Sixteen were animals captured for the first time. Table 1. Trapping effort and numbers of individual kit foxes captured by county and area, 1 July 94 to 30 June 95, westcentral Colorado. Trap Nights Captures County and Area Mesa/Garfield West and North of Grand Junction Mesa/Delta SE of Grand Junction to Delta Montrose/Delta Delta to SE of Montrose Moffat Browns Park-Vermillion Creek Totals 1125 8 528 2 502 21 171 0 2,326 31 Since May 1992 we have captured, marked, and released 38 individual kit foxes. Baits have usually been road-killed prairie dogs, cottontails, and ground squirrels supplemented with turkey chicks. A commercial attractor scent (chicken and fish, Rob Erickson's on Target A.D.C.) has been used in conjunction with the baits since mid-May 1995 in all traps. Since March 1992 trapping efficiency and capture rates have increased (Table 2) although much of the increase is probably the result of more concentrated effort in areas where foxes are known to be present. Seventy percent of �256 Table 2. Trapping effectiveness and capture rates, March 1992 to July 1, 1995, Kit Fox survey western Colorado. Trap Period Trap Nights # Individual Total Fox TNE/Indiv TNE/All Captures* Fox Captured Fox Capt*. Mar 92-June 93 Peach Valley only 2725 836 9 9 9 9 303 93 303 July 93-June 94 1754 18 21 97 88 July 94-June 95 2326 31 46 75 50 6805 38** 75 179 91 Total 93 *TNE = Trap nights of effort, all captures includes recaptures. ** Total of all individuals captured during the entire project. captures have been in the months of May, July, August and September, when recaptures that occur within 2 weeks of the last capture are excluded (trap happy animals were often caught on consecutive days) (Table 3). Table 3. Trapping success by age, sex and month of capture, 1992 - July 1, 1995, for foxes captured or recaptured when recaptures occurred at least 2 weeks post capture. No captures were made in October or November. Jan Feb Mar Jul Aug Sep Dec Month Apr May Jun Total Male Adult Juv Female Adult Juv Unknown Tot 3 1 0 1 3 1 1 1 1 3 4 1 3 3 2 3 2. 4 3 2 4 8 3 11 12 4 0 1 17 6 5 0 1 4 2 l 6 9 1 23 10 1 57 Of 37 foxes (10 adult males, 7 juvenile males, 13 adult females, 6 juvenile females) captured and collared or ear tagged we have recaptured 14 of them (4 adult males, 2 juvenile males, 6 adult females, 2 juvenile females) at least once (range 1-6). Of 23 animals captured once, 5 were adult males, 6 adult females, 6 juvenile males, 5 juvenile females, and 1 sex unknown. Forty-three percent of our adult animals have been recaptured compared to 31% of the juveniles. In late spring-early summer 1995 we have not captured any foxes in almost 400 trap nights of effort including areas known to have foxes. This may reflect increasing trap wariness as well as a possible decline in populations. Different trapping methods, including selective use of padded jaw traps will be tried in some of the areas where we are finding sign of foxes but not making any catches. Mesa/Garfield (Wand N of Grand Junction): Eight kit foxes were captured during 732 trap nights of effort in July and August (Fig 1, Table 4). An additional 393 trap nights of effort in May and June 1995 have not resulted in any captures. These 8 fox are the first captured in 3 field seasons and 2059 total trap nights of effort since 1992. Seven of the 8 foxes were radio-collared, none have been located since April 1995. �257 Fig 1. Length of radio-contact and status of collared kit foxes, Mesa and Garfield Counties, July 1 1994 - June 30 1995. Month and Year J A 1994 O N D J S 1995 F M A M J Location Rabbit Valley M (194/195) 151.244* F (196/197) 151.221 F (198/199) 151.034 Prairie Canyon/Baxter Pass U (153/154) 151.082 X------------------------ unknown x--- unknown x- unknown X- unknown Corcoran Point F(Jl26/127) 150.834 M(l30/131) 150.468 F(l32/133) 150.638 x---------------------- unknown x--------------------------- unknown x--------------------------- unknown Cheney Res. M(JlOl/102) x- unknown * Ear tags in ( ) followed by radio-collar frequency \--I \--I Three of the 8 animals (1M,2F) were captured in Rabbit Valley close to the Utah border. Neither female has been located since September 1994. The male was last located on 10 April 95. An animal of unknown sex was captured and radio-collared on 21 Aug in Prairie Canyon, Garfield County. It was accompanied by another fox which was not captured. The collared animal has not been located since its date of capture. Four foxes (2M, 2F), 2 of them juveniles (lM, lF), were captured north of Grand Junction near Corcoran Point. Three of them were radio-collared. Two (Ml50.468, Fl50.638) were last located 9 April 95. The third has not been located since 12 Dec. Two other pups were observed but not captured at Corcoran Point, Basagoitia photographed kit foxes visiting guzzlers in this same area in the summer of 1994. Field crews this spring found fresh kit fox sign at Prairie canyon and Corcoran Point but did not capture any foxes. The radio-collars on the 7 animals in this portion of the study region are close to their expected battery life and may no longer be working. Hundreds of square kilometers of what appears to be suitable range is unoccupied or inhabited by scattered individuals difficult to detect and capture. Mesa/Delta Counties (Grand Junction to Delta): \--I Two animals, 1 juvenile male and 1 adult female were captured near Cheney Reservoir (near the Mesa-Delta County line) in 528 trap nights of effort. The female broke her jaw in the trap and died during rehabilitation efforts. The male has not been located since it was collared. A juvenile male (MlS0.980, ear tag 17) captured 22 Nov 93 by Verbeck in T14S/R95W/S33, northeast of Delta north of the Gunnison River has not been located since its capture. These 3 foxes are the only captures made in this section of the study area in 1597 trap nights of effort since 1992. Basagoitia reported on 5 Sept 94 seeing a live fox along Highway 50 south of Cheney Reservoir. On 3 Sept he observed 2 dead kit foxes on the highway in Delta county, 1, 2 km NW of Adobe Flats reservoir the other by the Alkali Flats road. Basagoitia photographed a kit fox at a guzzler in Wells Gulch during the summer of 1994. Basagoitia and Hays spent considerable time live-trapping and searching for fox tracks and sign using ATV's from Jan-April. They placed 4 road killed deer as baits but found no evidence of fox visiting the carcasses. Effort suggests few foxes �258 Table 4. Locations of radioed foxes in the lower Gunnison and Colorado River Valleys, Mesa and Garfield Counties July 1, 1994 to July 1, 1995. V Sex Ear Tag Number Radio Collar Frequency M 194/195 151.244 F F 196/197 198/199 151.221 151.034 Prairie Canyon/Baxter Pass 8/21/94 U 153/154 151.082 T7S R105W S12 150.834 TlOS RlOOW 35 TlOS RlOOW 35 TlOS RlOOW 35 TlOS RlOOW 35 TlOS RlOOW 35 TlOS RlOOW 35 TlOS RlOOW 35 TlOS RlOOW 35 TlOS RlOOW 35 TlOS RlOOW 35 TlOS RlOOW 35 TlOS RlOOW 35 Capture Site Date, and Date of Last Radio-Contact Rabbit Valley 8/2/94 12/19/94 4/10/95 8/2/94 8/10/94 9/9/94 Corcoran Point 7/15/94 12/19/94 3/20/95 7/15/94 7/17/94 12/19/94 3/20/95 4/9/95 7/19/94 12/19/94 3/20/95 4/9/95 Cheney Res. 7/11/94 7/13/94 F(J) 126/127 M(J) M 128/129 130-131 F 132-133 M(J) 101-102 150.851 Jaw Broken - Died in Rehab F(A) not radio-collared 150.468 150.638 Location TlOS Rl04W S20 TlOS Rl04W S20 TlOS Rl04W Sl7 TlOS Rl04W 12 TlOS R104W 13 TlOS R104W 13 T13S R98W Tl3S R98W V 30 30? live in this expanse of the Gunnison River basin with many kilometers of suitable habitat unoccupied or occupied at very low densities. Peach Valley and Montrose East Populations: We have spent 2240 trap nights of effort in the region from Delta south of the Gunnison River to south and east of Montrose since the study began in 1992. This is the center of our only population of foxes and the site of most of our radio-tracking effort. Twenty-seven of the 38 individual kit foxes have been taken from this area. During the 94-95 reporting period, in 452 trap nights of effort, a total of 8 new foxes were captured. Four were captured in Peach Valley: adult female (FlSl.083), adult male (Ml51.257) and 2 female pups too small to radio-collar (they were collared with pink nylon collars). At Montrose East, 3 adult (MlSl.160, MlSl.107, MlSl.042) and 1 juvenile (Ml51.131) male were captured. In Peach Valley 11 adult kit foxes (SF, 6M) have been captured and radioed since 31 May 1992 (Fig 2, Table 5). Five of those animals, 3 females and 2 males are dead. Two were killed by coyotes, the cause of death of the others is unknown. The status of 4 animals is unknown. Two animals, male 151.095 and female 151.083) are still alive. Two radio-collared males (Ml51.209, Ml51.056) have moved from Montrose East to Peach Valley. These 4 animals are the only radio-collared animals known to be alive in Peach Valley. V �259 Fig 2. 1995. Length of radio-contact with collared kit foxes, Peach Valley, 1992- * = Animals that have moved to Peach Valley from Montrose East. 1992 MJSN 1993 JMMJSN 1994 JFMAMJJASOND 1995 JFMAMJ Animal F0.238 x-------Dead X-Unknown M0.309 F0.873 x--------------------- -------------------------------Alive F0.889 X----------------------------------------------Dead Fl.009 X-------------------------------------------------Dead Ml.095 X--------------------------------------------------Alive M0.037 X----------------------Dead Ml.177 X--------------------------------Unknown M0.499 x--Unknown X------------Unknown Fl.083 Ml.257 x--------Dead Ml.209* X--------------------Alive Ml.056* X--------------------Alive Table 5. Kit fox radio-collared in Peach Valley, 1992 - July 1, 1995, and dates of last sightings/radio-signals. New radio-collar frequencies are shown as they are replaced. ETR = ear tag replaced. ~ ~ ~ ~ Capture or Date Located Sex 5/31/92 11/23/92 F 9/28/92 M 2 2.7 9/28/92 4/6/94 4/14/95 6/28/95 F 5 2.5 9/28/92 3/2/93 9/22/93 1/6/94 3/15/95 F Location Fate T51N/R9W/S29 T15S/R94W/S27 Dead (C) 150.309 T50N/R9W/S9 Unknown 150.338 150.956 150.873 T50N/R9W/S9 T50N/R9W/S9 T51N,R9W/S32 T51N,R9W/S32 Alive 150.379 150.850 150.889 T50N/R9W/S9 T50N/R9W/S9 T50N/R9W/S9 T50N/R9W/S17 T49N/R8W/S? Dead T51N/R9W/S29 T50N/R9W/S5 T50N/R9W/S10 T50N/R9W/S5 T51N,R9W,S29 Dead (C) T51N/R9W/S29 TS0N/R9W/S5 T50N/R9W/S10 T51N/R9W/S29 T51N/R9W/S29 Alive T50N/R9W/S22 T50N/R9W/S8 T50N/R9W/S8 Dead Ear Tag Number Weight Kg Radio Collar Frequency 1 2.0 150.238 6 2.4 (moved from PV to SE of Montrose) 2/23/93 12/6/93 9/5/94 2/28/95 4/14/95 F 2/23/93 12/6/93 9/9/94 1/25/95 6/28/95 M 4/21/93 9/24/93 3/26/94 M 7 ETR18 8 2.5 2.6 2.0 3.0 3.1 150.468 150.189 151.095 ETR147/148 12 ETR15 150.638 150.594 151.009 2.5 2.8 150.708 150.037 �260 Table 5 continued. 9/20/93 6/19/94 1/27/95 M 13 9/30/93 11/22/93 M 7/14/94 9/7/94 12/19/94 1/27/95 F 190/191 7/18/94 9/9/94 11/18/94 M(J) 192/193 2.7 ETR46/47 16 150.813 151.177 2.3 150.499 T50N/R9W/S16 T51N/R9W/S29 T51N/R9W/S29 Unknown T50N/R9W/S17 T50N/R9W/S17 Unknown 151.083 T51N/R9W/S31 T51N/R9W/S31 T51N/R9W/S29,30 T51N/R9W/S29 Unknown 151.257 TS0N/R9W/S10 T50N/R9W/S10 T50N/R9W/S21 Dead T51N/R9W/S29 M 30/151 151.209 (First captured on 8/26/94 in Montrose East as Ml51.107) T51N/R9W/S32 4/18/95 T51N/R9W/S32 6/29/95 Alive 151.056 T50N/R9W/S9 M 26/27 (First captured on 8/24/94 in Montrose East as Ml51.131) T50N/R9W/S9 6/20/95 Alive 1.9 1.6 V 1/27/95 4/13/95 * Dead (C); killed by coyote V Adult female (Fl50.873) was captured by Dent in mid-April. He reported she was lactating but that an injury to her right rear leg was causing her to limp. She is still alive but not believed to have any pups. Dent recovered the remains of female 151.009 in Peach Valley on 14 April and judged she had been killed by a coyote. Peach Valley female (FlS0.889) first captured 28 Sept 1992 was recovered dead on 15 Mar 1995, 19 Jan south of her capture area and south of the Montrose East population. The area her carcass was recovered in had been trapped by field crews in the summer of 1994 with no fresh fox sign reported. Trapping by Hays in the area the carcass was recovered in did not yield any other foxes. V The male foxes alive in Peach Valley include 2 animals (MlSl.209, MlSl.056) that moved from the Montrose East site to Peach Valley after early January 1995. Basagoitia and Hays on 27 Jan captured a male with only l ear tag (30) and no radio-collar they recollared the animal (Ml51.209). This animal was radio-collared (Ml51.107) and ear tagged 30/29 on 26 Aug in Montrose East by Boyle. Between 10 Jan and 27 Jan he moved from Montrose East to Peach Valley and lost his collar. Dent recovered collar 151.107 on 18 April close to the Water Tank in Peach Valley not far from the area Hays and Basagoitia recaptured and collared the fox as Ml51.209. Male 151.056 moved to Peach Valley from Montrose East between 10 Jan and 13 April when Dent recaptured him in Peach Valley. He replaced radio-collar 151.131 with a new unit (Ml51.056). Boyle first captured this fox in Montrose East on 25 Aug. The third male in Peach Valley, Ml51.095 is an animal that has stayed in the Valley since first captured on 23 Feb 1993. In 1994, the Montrose East population included at least 13 animals (4 ~dult females, 3 adult males, 3 female pups, 3 male pups) clustered in a 5knt' area (Fig 3, Table 6). As of 28 June 1995, 5 of 11 animals collared in 1994 are alive. Three, Fl51.288, Fl51.037, Ml51.237, are at the Montrose East site. Two males (MlSl.056, Ml51.209) have moved to Peach Valley. Two are dead, the fate of 4 is unknown. Olterman has made several flights over the study area ti V �261 trying to locate radio-collared foxes. He last flew on 27 June and will be making other flights in the next few weeks to attempt to locate missing animals. Brown's Park, Moffat County: In the project year we spent 171 trap nights of effort in Brown's Park with no captures. Clait Braun of the Colorado Division of Wildlife reported seeing a kit or swift fox near Vermillion Creek in July 1994. Dent in late August spotlighted a small fox in that area but could not make an identification. These observations along with reports in 1993 indicate kit or swift foxes live in Moffat County. Three summers of trap effort with a total of 616 trap nights have not resulted in any captures. Figure 3. Length of radio contact and fate of radio-collared kit foxes, Montrose East population, May 1994 to July 1, 1995. Month and Year and Area M J Animal ID Fl51.288 Fl51.146 Fl51.037 Fl50.037 Fl50.940 Fl51.019 ( .020) Ml51.237 Fl51.056 Ml51.160 Ml51.209 Ml51.042 X---------------------------------------Alive X------------------------------Unknown X-------------------------Unknown X------------------------------Unknown x------Dead x---------------Unknown (Collar recovered) X-----------------------------------Alive X----------------------------Alive in PV X--------------Unknown X----------------------------Alive in PV x-----Dead J 1994 A S O N D J 1995 F M A M J Table 6. Kit foxes trapped in the Montrose East population in 1994 and status on July 1 1995. NL = non-lactating females; L = lactating. Capture or Date Located Sex Ear Tag Number Weight Kg Radio Collar Frequency 5/29/94 8/26/94 12/22/94 F(NL) 21/32 2.5 2.4 150.947 2.8 151.288 150.403 ETR22/38 2.5 2.4 2.3 151.146 23 2.3 150.338 1/10/95 4/17/95 F(NL) 6/27/95 ~/ ~ 5/29/94 8/25/94 1/10/95 F(NL) 22/28 5/29/94 8/28/94 9/15/94 F(NL) 5/29/94 M pup (recaptured on 7/7/95 - now Fl51.037) 24 1.75 not-collared Location Fate T49N/R8W/S7 T49N/R8W/S7 T49N/R8W/S6,7 T49N/R8W/S6 T49N/R8W/S7 T49N/R8W/S7 Alive T49N/R8W/S7 T49N/R8W/S7 T49N/R8W/S7 Unknown T49N/R8W/S7 T49N/R8W/S7 T49N/R9W/S12 Alive T49N/R9W/S7 Unknown • �262 V Table 6 continued. 5/30/94 8/25/94 F pup 176/177 1.5 not-collared 2.3 150.037 1/10/95 V T49N/R8W/S7 T49N/R8W/S6 T49N/R9W/S7 Unknown 5/30/94 F pup 178/179 1.4 not-collared T49N/R8W/S7 Unknown 5/31/94 6/7/94 7/28/94 F (L) 180/181 2.3 150.940 T49N/R8W/S7 T49N/R8W/S7 T49N/R8W/S9 Dead 5/3/94 8/28/94 9/16/94 11/11/94 F pup 182/183 M Ad 8/25/94 12/22/94 M Pup 26/27 1/10/95 not-collared 2.2 151.019( .020) Collar recovered no trace of carcass 6/3/94 8/25/94 1/10/95 4/18/95 6/27/95 4/13/95 6/27/95 1.8 184/185 3.0 not-collared 2.5 150.708 151.237 2.0 151.131 Moved to Peach Valley 151.056 = T49N/R8W/S7 T49N/R8W/S7 T49N/R8W/S8 T49N/R8W/S8 Unknown T49N/RSW/S7 T49N/R8W/S7 T49N/R8W/S7 T49N/R8W/S7 T49N/R8W/S6 Alive T49N/RSW/S7 T49N/RSW/S7 T49N/R8W/S7 T50N/R9W/S9 T50N/R9W/S9 Alive 8/27/94 12/29/94 1/4/95 M Pup 33/34 1.5 151.160 T49N/R8W/S7 T49N/R8W/S6,7 T49N/RSW/12 Unknown 8/26/94 12/22/94 1/10/95 1/27/95 4/18/95 M Ad 29/30 2.4 151.107 T49N/R9W/S7 151.209 T51N/R9W/S29 T51N/R9W/S32 Alive T49N/R9W/S7 T49N/R8W/S8 T49N/R8W/S18 Dead(C) 8/27/94 11/9/94 11/22/94 Moved to Peach Valley M Ad 48/49 2.3 151.042 T49N/R8,9W/S7,12 T49N/R9W/S12,7 V Behaviors of foxes in Montrose East and Peach Valley: A report on kit fox activity patterns is appended to this report (Boyle 1995). Boyle estimated home range size for 7 kit foxes in the Montrose East population and 2 animals in Peach Valley (Table 7). Animals at Montrose East averaged 3.6km2 (range 1.5-5.9), those in Peach Valley 7.5-7.7km2 • Home range sizes may reflect habitat and prey abundance differences between the 2 sites. Three animals have moved from Montrose East to Peach Valley (Ml51.209, M151.056) or out of Peach Valley (F150.889). This is the first evidence of foxes moving between the 2 areas. Boyle (1995) reported fox tracks north of Flattop Mesa between Peach Valley and Montrose East and did not believe they were made by collared animals. Scattered individuals may inhabit the area between these 2 sites but to date we have not captured any and several landowners have refused permission to cross their properties. Most radio-collared animals have not moved far from their original capture locations (Table 5, 6). The eight foxes which we have had radio-collared for the greatest length of time (Table 8) show little movement from their original V �263 Table 7. Estimated home range size for nine kit foxes, Peach Valley and Montrose East populations, October-January 1994-1995. From Boyle (1995). Animal No. Age/Sex Montrose East 150.037 150.403 150.947 150.708 151.131 151.160 151.107 Ad Ad Ad Peach Valley 151.009 151.083 Ad F Ad F sites of capture. predation. Juv F F F M Juv M Juv M Juv M Home Range Kmi # of Locations 56 5.9 4.7 1.5 4.7 2.3 2.5 3.9 53 50 36 38 36 7.7 7.5 22 18 52 Fidelity towards home ranges may reduce the probability for Home ranges of animals in the Montrose East study area overlapped and the individuals using different dens varied over the field season (Boyle 1995). Boyle reported 9 individual foxes used an average of 3.7 dens (range 2-6) from late September to mid-January. A den complex adjacent to the main road through the study area was continuously used by 2-4 foxes from late November to January 10. Previous work by Verbeck (Fitzgerald and Verbeck 1993) and Link (1995) agree with Boyle's observation of frequent den changes with a tendency to favor some dens more than others. Boyle (1995) estimated total minimum distance travelled and straight-line maximum distances traveled during single night movements of 5 individual foxes in the Montrose East group. He estimated total minimum distance to average 6.2 Jan and maximum straight line distance to average 2.0 km. Movements of the observer in tracking foxes may have caused them to increase their movements more than might occur if undisturbed. Link, Beck and Dent (personal communications) reported kit foxes being radio-tracked on foot were aware of their presence and moved rapidly away from them. In the narrow Montrose East Valley it would be hard to avoid harassing the foxes when tracking their movements. Boyle's effort confirms our working assumption that foxes do not travel more than a few Jan's in their nightly forays and traps have to be set close together and left out for several days. Reproductive Success: In December Boyle reported transitory pairing of F150.947 and M 151.131, followed by F150.947 then denning with F150.403. Dent in April could not locate any paired animals but did trap a lactating female. Field crews working in late spring and summer have not located any dens with pups. Since May 1992, we have captured 14 adult females at times of the year when reproductive status could be determined. Four have shown evidence of lactation or contained uterine scars (Fl50.238). In Montrose East in spring and summer 1994, 4 of 5 adult females were non-lactating. This field season the only female (150.873) in Peach Valley does not appear to have pups. In Montrose East female 151.288 did not show signs of pregnancy or lactation in mid April. Female 150.338 (captured on 7 July 1995) shows no sign of bearing pups. Since 1992 we know of only 6 litters of pups, 3 in Peach Valley each with 2 pups, 2 in Montrose East with 7-8 pups using a single den, and 1 at Corcoran Point with 4 pups. The low numbers of adult females that are reproducing may be due to a number of different causes including: lack of �264 Table 8. Estimated size of range of 8 kit foxes radio-collared for at least 7 months, Peach Valley and Montrose East populations. Fox# Months Radioed Estimated Range (km2) Peach Valley FlS0.873 FlSl.009 FlS0.889 MlSl.095 MlSl.177 28 16 8 7 Montrose East FlSl.288 FlSl.037 MlSl.237 13 7+ 12+ 3 5 33 25+ 29+ 7 8 8 4 males, lack of adequate food, disturbance from field crews radio-tracking them, or, in the case of the Montrose East animals high population density in a limited spacial area. other Activity: Link has completed her M.A. Thesis detailing the first 2 years of the project. Copies are enclosed with this report. The Wildlife Commission acted in September 1994 to expand the area with trapping restrictions to protect kit fox. Field plans have been completed for the 95-96 year: 1. We will continue to radio-collar animals captured in Peach Valley, Montrose East or at Corcoran Point but will not radio individuals taken elsewhere. All captured animals will be ear-tagged. 2. We will conclude our extensive trapping efforts with work to concentrate on the 64 km area along the base of the Book Cliffs from Corcoran Point to the Utah border and the 38 km long area at the base of Grand Mesa from Cheney Reservoir to Wells Gulch. 3. We will spend additional time in Brown's Park trying to resolve if kit or swift foxes live there. 4. From September 1995 through June 1996 we will use a monthly aerial search as the primary method for monitoring radioed animals in the Gunnison and Colorado river drainages. s. Depending on weather we will conduct a limited amount of snow-tracking during the winter of 95-96. 6. By April, 1996 we will complete reports and recommendations for continued research on kit fox in western Colorado. V V By: James P. Fitzgerald Contractor, University of Northern Colorado. V �313 Colorado Division of Wildlife Wildlife Wildlife Research Report July 1996 JOB PROGRESS REPORT State of Colorado Project No. W-153-R-7 Mammals Research Work Plan No. lOA c!9rt Fox studies Job No. Kit fox (Vulpes macrotis) Status in Colorado Period Covered: July 1, 1995 to June 30, 1996 Author: J.P. Fitzgerald Personnel: J . P. Fitzgerald, J. Eussen, c. Hatch, J. Prather, S . Lechman, L. Dent, D. Finley, J. Olterman, T. Beck. ABSTRACT In 1368 trap nights of effort 8 new kit foxes were captured. Five of them were from a family group (1 adult female, 2 female pups, 2 male pups) captured in July 1995 in Peach Valley. An adult male was captured at Cheney Reservoir and a juvenile male pup and adult female were captured at Corcoran Point. The total number of individual kit foxes trapped since 1992 is 47 with 33 of them from the Peach Valley-Montrose East complex. A total of 17 individual kit foxes were radio-tracked during 1995-96. Nine (53%) of the 17 animals died during the study period. Four deaths were attributed to coyotes, 2 to drowning, 1 to an automobile, and 2 from undetermined causes . Two litters of pups were born in the Montrose East area in 1996 with female F23 being killed by a coyote after her pups came above ground . . The fate of those p~ps is unknown. One other female living between Peach Valley and Montrose East may have young. J. Eussen will be working on kit fox food habits as a thesis project in 1996. The draft final report of project activities was completed and reviewed by T. Beck. Revisions are being made and the final report wlll be delivered by August 1. i]ID1i~Dm1 1 I BDOW010805 �315 KIT FOX (VULPES MACROTIS) STAms IR COI,9RADO James P. Fitzgerald P, N, QBJECTIYE Document the geographic distribution and relative abundance of kit fox in Western Colorado. SEGMENT QBJECTIYES 1. Continue to monitor radio-collared foxes in Montrose, Delta, Mesa and Garfield.counties. 2. Continue search for new kit fox populations. 3. Compl~te draft of final project report. METHODS Field methods for live-trapping were similar to those reported previously (Fitzgerald and Link 1993, Fitzgerald and Verbeck 1993). Limited trapping was also conducted using 1.5, soft-catch, leg-hold traps at baited dirt-hole sets. Trapping efforts were concentrated in the Gunnison and Colorado River Drainages in Mesa, Garfield, Delta, and Montrose counties and in northwestern Moffat county. Field personnel continued to monitor and document locatio~s of radio-collared foxes using searches with a fixed wing aircraft piloted by J. Olterman, or by ground tracking. RESULTS AND DJSCQSSIQH Live ~rapping Bffort From July 1, 1995 to June 30, .1996 we conducted 1368 trap nights of effort resulting in capture of 8 new foxes (Table 1.). All trapping was done in Moffat County.or in the Gunnison River-Lower Colorado River drainages in Mesa, Garfield, Delta and Montrose counties. Two foxes were captured at Corcoran Point, a juvenile male and-an adult female. An adult male was captured at Cheney Reservoir. In Peach Valley, 2 juvenile females, 2 juvenile males, and an adult female were captured at a whelping den (Table 2). Field crews again. failed to capture any foxes in Moffat County, although scat believed to be from kit or swift foxes was discovered and a CDOW biologist (C. Braun) reported seeing a swift or kit fox crossing the road near Vermillion Creek. The total number of individual kit foxes trapped since the study began in 1992 is 47. T~irty-three of the captures have been made in the Peach Valley-Montrose East complex. �316 Table 1. Trapping effort and numbers of individual kit foxes captured by county and area, July 1 1995-June 30, 1996. County/Area Trap Nights New· Captures· Mesa-Garfield Counties Prairie Canyon to Corcoran Point Areas 347 2 Mesa-Delta Counties Cheney Reservoir to -East of Del ta Airport 493 1 Montrose-Delta Counties Peach Valley-Montrose East Complex 368 5 Moffat County Browns Park-Vermillion . Creek 160 0 Location, sex, age class, ear-tag·number and radio-collar frequency for previously uncaptured kit foxes, 1995-96 fiscal year. Table 2. Location Sex/Ear Tag Radio-collar Date Captured Status M(J)310 F(A)311 150.189 8/12/95 151.-186 11/3/95 Unknown Dead M(A)312 151.083 12/4/95 Dead 150.029 150.728 150.004 150.920 151.009 7/28/95 7/26/95 7/.27/95 7/26/95 7/27/95 Alive Dead Dead Dead Alive V V V Corcoran Point Cheney Reservoir Peach Valley F(A)309 F(J)306 F(J)308 M(J)305 M(J)307 V �317 Monitoring of Radio-collared Foxes During the fiscal year, field crews periodically located radioed animals during ground searches. Jim Olterman also made a number of flights to locate animals. A total of 17 foxes (9 males, 8 females) were located and monitored during 1995-96, including 2 male and 2 female pups marked in July 1995, three of which have died (Table 3). Two litters of pups came above ground in May 1996 in Montrose East, one litter (F23's) had 4 pups, the other litter (F2l's) has an unknown number of young. F309 living between Peach Valley and Montrose East is also believed to have pups but the whelping den has not been located. Nine animals died during the year with deaths of 4 attributed to coyotes (F306, F308, F23, M312). Two died from probable drowning (M30, F311), 1 was road killed (M305) and two (M26, F309) died from undetermined causes. ~her Activity The first draft of the final report wae finished and reviewed by T. Beck. It is being revised and will be.submitted by the end-o~ August pending receipt of some additional information from the Montrose Office· of the BLM. The state of Utah has been contacted to see whether UNC can do some trapping in late summer along the Utah-Colorado border to assess kit fox populations in Utah that might provide migrants to Colorado stocks. J~ Eussen is collecting kit fox scat and will be doing a food habits analysis for a M.A. thesis using his materials and material collected by other field workers. By: James P. Fitzgerald, Contractor, University of Northern Colorado �318 Table 3. Kit foxes alive during the 1995-96 fiscal year and their status as of June 30 1996. Site & Date of Capture/ Tracking Corcoran Point 8/12/95 11/2/95 3/17/96 5/16/96 Sex/Age Ear Tag# Radio-collar Frequency JM 310 150.189 n.c. Location T9S,RlOOW,S25 T9S,RlOOW,S34 TlOS,RlOOW,SlO signal too weak to locate Status as of 30 June Dead Cheney Reservoir 12/4/95 AM 312 151.083 5/16/96 moved to Wells Gulch 6/18/96 T3S~R2E,S24 ·T4S, R3E, S13 T4S,R3E,S13 Dead Delta Airport 11/19/95 T14S,R96W,S36 Unknown AF 311 151.186 collar at private residence AM 33* 151.160 Peach Valley 9/28/92 AF 5 150.338 4/6/94 • 150.956 4/14/94 150.873 8/15/95 10/11/95 12/4/95 moved to Alkali Gulch 8/26/94 12/22/94 1/10/95 1/27/95 10/11/95 2/14/96 6/18/96 8/24/94 1/10/95 4/13/95 10/11/95 2/14/96 4/9/96 7/26/95 2/14/96 4/9/96 AM 30 151.107 moved to PV from ME 30/151 151.209 collar in Selig ~anal JM 26 151.131 moved to PV from ME 151.056 JF 306 150.728 V Unknown TlN,RlW,SlO TlOS,RlOOW,SlO T1N,R1E,S29 TlN,RlW,S36 11/3/95 3/17/96 5/16/96 6/18/96 V V V TSON,R9W,S9 T50N,R9W,S9 T51N,R9W,S32 T51N,R9W,S7 T51N,R9w,s29:..32 Tl5S,R97W,Sl Unknown T49N,R9W,S7 T49N,R9W,S12 T49N, R9W, S_12 T51N,R9W,S29 T51N,R9W,S29-32 T51N,R9W,S30 TS1N,R9W,S29 Dead T49N,R8W,S7 T49N,R8W,S7 T50N,R9W,S9 T51N,R9W,S29 T50N,R9W,S16 TSON,R9W,S16 Dead TSON,R9W,S16 TSON,R9W,Sl6 TSON,R9W,S16 Dead V V �Table 3 (continued) Site & Date of Capture/ Tracking Sex/Age Ear Tag# Radio-collar Frequency 7/26/95 12/25/95 JM 305 150.920 7/27/95 2/14/96 4/9/96 JF 308 150.004 moved to ME moved back to PV JM.307 7/27/95 5/16/96 6/18/96 7/28/95 2/14/96 • 6/18/96 n. c. 151.009 AF 309 150.029 moved to ME moved between ME and PV .Montrose East 5/29/94 8/26/94 4/17/95 2/14/96 6/18/96 AF 21 150.947 52/52 151.288 150.098 150.403 Location T50N,R9W,S16 Tl5S,R95W,S33 Status as of 30 June Dead T50N,R9W,Sl6 T49N,RSW,S24 T51N,R9W,S30 Dead T50N,R9W,Sl~ · T49N,R9W,S6 T49N,R9W,S6 Alive T50N,R9W,S16 T49N,R8W,S9 T49N,R8W,S3 Dead T49N,R8W,S7 T49N,R8W,S7 T49N,R8W,S7 T49N,RSW,S7 T49N,R9W,S6 Alive T49N,RSW,S7 T49N,R8W,S7 T49N,RSW,S7 Unknown T49N,R8W,S7 T49N,R8W,S7 T49N,R8W,S5 T49N,R8W,S7 T49N,R8W,S9 Dead 5/29/94 8/25/94 1/10/95 AF 22 5/29/94 9/15/94 7/7/95 2/14/96 6/18/96 AF 23 6/3/94 AM 184 n. c. 150.708 151.237 T49N,RSW,S7 T49N,RSW,S7 T49N,RSW,S7 TS0N,R9W,S18T50N:, R9w·, Sl6 T50N,R9W,S26 T49N,R9W,Sll-12 Alive JM 33 151.160 moved ~orth of Delta Airport T49N,RSW,S7 T49N,RSW,S6-7 T49N,RSW,S12 T14S,R69W,S36 moved to Alkali Gulch Tl5S,R97W,Sl 151.146 151.042 8/25/94 4/18/95 10/11/95 2/14/96 5/16/96 6/18/96 8/27/94 12/29/94 1/4/95 10/11/95 11/19/95 12/4/95 150.338 T14S,R96W,S36 Unknown ������������������������������������������������������������101 Colorado Division of Wildlife Wildlife Research Report July 1998 SEGMENT NARRATIVE State of Project No. Work Package No. Colorado W-153-R-ll 0663 Task No. cost center 3430 Mammals Program Species of Special Concern/Species at Risk conservation Kit fox conservation Period covered: July 1, 1997 - June 30, 1998 Author: T.D.I. Beck Personnel: T. Beck, G. Bock, D. Coven, J. Garner, R. Gill, M. McLain, P. Schnurr, L. Willmarth; CDOW ABSTRACT The Program Narrative was modified to more clearly outline the multiple stages of the development of a conservation strategy. The initial phase of biological assessment was begun this segment. Sixty-three cameras, triggered by active-infra-red sensors, were deployed throughout the Uncompahgre and Gunnison valleys. Seven photos of kit fox were obtained; all are believed to be different individuals. Four of the photos were at or near dens initially located by ground searches. In addition, 2 active kit fox dens were located; one of which had 2 individuals and the other only one. Of the 3 dens with pairs present, none appeared to have young-of-the-year. All but two of the photos and sightings were in the Peach Valley-Montrose East habitat area. This isolated area of habitat is only 90 km2 in area. The Gunnison Valley area comprises 316 km2 of habitat, of which approximately 50\ is deemed marginally suitable because of lack of den sites. No efforts were made to trap and collar kit fox during the whelping season. �103 KIT FOX (VULPES HACROTIS) STATUS IN COLORADO Thomas D. I. Beck P,N, Objective Develop and implement a conservation strategy to recover and conserve kit fox populations in Colorado. segment Objectives 1. Develop a conservation strategy to recover and conserve kit fox in western Colorado. 2. Survey for presence of fox in Gunnison valley area with infra-red activated cameras. 3. Capture and radio-collar adult kit fox in both the Uncompahgre and Gunnison valley areas. METHODS AND MATERIALS The earlier version of the Program Narrative was modified to more clearly outline the progressive stages of the conservation strategy development. The modified document was submitted in May 1998. Ground searches for kit fox sign and dens were conducted in the Uncompahgre and Gunnison valleys on 40 days during April-June 1998. Remote 35mm cameras, activated by active-infra-red sensors, were deployed in areas with fresh kit fox sign, areas with old fox dens, and likely travel areas. The camera system was developed by TrailMaster (Lenexa, KS) and was essentially the same system as used by Beck on black bear (Ursus americanus) studies (Beck 1995). The distance from the camera to the infra-red transmitter was 2 m; camera height above ground was 15-30 cm. A pipe was driven into the ground 10 cm in front of the infra-red transmitter and one of several baits (beaver meat, chicken flesh lure) were placed down in the pipe to minimize bird conflict. The recorder unit was programmed so that the infra-red beam had to be broken for 0.15 sec to be recorded ( value of P = 3) and only a 6 sec delay between successive pictures. Color print film, ASA 400, in rolls of 24 and 36 exposures were used. Cameras were left at locations for varying periods, depending on area, work schedule, and need for cameras in other areas. Observations of active kit fox dens were conducted at dusk on 9 evenings, in hopes of documenting presence of pups. Observations were made with binoculars or a night-vision scope at distances of 30 to 100 m. Maps of areal extent of salt desert shrub communities were obtained from CDOW Habitat Section. Also, 15-minute quad maps were used for plotting camera locations and all fresh fox sign. �104 RESULTS ARD DISCUSSION Program Narrative Modification The revised Program Narrative describes 5 phases in the development of a conservation strategy: Biological Assessment, Socio-Economic Assessment, Political Assessment, Conservation Planning and Implementation, and Conservation Evaluation. The development of a formal Conservation strategy will occur in Phase IV. It is believed that by doing the biological, social, and political assessments as precursors to the formal plan, support for the plan will be more wide-based and solid (Clark et al. 1994). Biological Assessment Analysis of the work reported by Fitzgerald (1996) suggested that summer was the period of lowest trap success, and the majority of search effort had been conducted in the summer. Thus it was deemed prudent to resurvey much of the available kit fox habitat during spring and fall seasons, utilizing a different technology than live-trapping. Ground searches for kit fox dens and spoor were conducted on 40 days during April-June, 1998 by either one or two researchers throughout the 406 km2 of habitat in the Gunnison and Uncompahgre valleys. Only 4 active dens were located, all in den complexes previously identified in the Uncompahgre Valley. Three of the dens supported kit fox pairs while the other had a single fox. Cameras were set near 3 of the 4 dens. The den in the Montrose land fill (kit fox pair) was not equipped with a camera because of the high human traffic flow by the den. The camera at the single den malfunctioned so no photos were made. We obtained 2 photos of kit fox at each of the other 2 dens. A total of 63 camera sets were made, for periods ranging from 7 to 28 days. Thirty-five of the sets were for periods of 10-15 days. Three kit fox photos were obtained in areas where we did not find active fox dens or spoor. Subsequent searches also resulted in no active dens found. Based on camera and ground searches, 8 kit fox were located in the Uncompahgre Valley and 2 in the Gunnison Valley just west of the Delta airport. Photos were obtained from all active den sites where cameras operated correctly; thus supporting the use of cameras as an effective survey tool. Based on 9 evening observations and 8 daytime examinations of den sites it did not appear that any of the 3 kit fox pairs produced any pups in Spring, 1998. Four natal dens of red fox (Vulpes vulpes) were located in the saltbush habitats in what appeared to be suitable kit fox habitat. The camera surveys also produced photos of the following: badger (3), house cat (3), dog (1), raccoon (7), cottontail rabbit (76), prairie dog (3), antelope ground squirrel (16), striped skunk (3), domestic cattle (6), magpies (44), horned lark (9), meadow lark (1), and a snake (1). Sixteen cameras produced no photos, 3 had mechanical malfunctions, and 2 were disturbed and made inoperable by people. The Uncompahgre Valley area is comprised of approximately 90 km2 of suitable kit fox habitat; arranged in an oblong pattern 30 km N-S with an average width of less than 4 km. Eight of the observed kit fox were in this area. This area is separated from the Gunnison Valley habitats by agricultural land, housing developments, a highway, and a river; resulting in a 7-9 km long obstacle course to inhibit dispersal. Based on the range of kit fox densities summarized by White and Garrott (1997) (0.16-0.7/km2 ) this area of habitat �105 would likely only support 14-63 kit foxes. this lower bound. Current density appears to be near The Gunnison Valley area is approximately 316 km2 of apparently suitable kit fox habitat. Only 2 kit fox were located in this region; both at the same camera and likely a pair. They were at the southern tip of the habitat area. Approximately 50% of the habitat is characterized by a volcanic rock cap which severely limits den sites. No old kit fox dens were located in these formations and only a few badger and coyote dens. It appears that the lack of old dens limits the value of this region. The northernmost section of possible habitat in the Gunnison Valley lies between Kannah Creek and the Colorado River. It is approximately 74 km2 in area but separated from the upper valley by extensive irrigated farm lands and new housing developments. No fresh or old kit fox sign was discovered in this area. It was decided not to trap and collar adult foxes during the whelping season because of the added stress of this activity to the individual foxes (Cypher 1997). The small number of kit foxes present seems to justify minimum intrusion during this period. Literature cited Beck, T.D.I. 1995. Development of black bear inventory techniques. Wildlife, Fed. Aid. Rep. W-153-R-8. llpp. Colo. Div. Clark, T.W., R.P. Reading, A.L. Clarke. Eds. 1994. Endangered Species Recovery:Finding the Lessons, Improving the Process. Island Press, Washington, D.C. 450 pp. Cypher, B.L. 1997. Effects of radiocollars on San Joaquin Kit Foxes. Wildl. Manage. 61(4):1412-1423. J. Fitzgerald, J.P. 1996. Status and distribution of the kit fox (Vulpes Colo. Div. Wildlife, Final Fed. Aid. Rep. 153-R-7. 78 pp. macrotis) in Western Colorado. White, P.J. and R.A. Garrott. 1997. Factors regulating kit fox populations. Can. J. Zool. 75:1982-1988. Prepared by Thomas Beck Wildlife Researcher w- �55 Colorado Division of Wildlife Wildlife Research Report July 1999 JOB PROGRESS REPORT State of Colorado Project No. W-153-R-12 Work Package No. __0-=-66=3;;......__ _ _ __ Task No. - - - - ~ 1 ------- Cost Center 3430 Mammals Program Species of Special Concern/Species at Risk Conservation Kit fox Conservation Period Covered: July 1, 1998,. June 30, 1999 Author: T. D. I. Beck Personnel: T. Beck, D. Coven, P. Creeden, V. Graham, M McLain, P. Schnurr, B. Sommerville, L. Willmarth; CDOW ABSTRACT Biological assessment work was expanded to the Colorado River Valley in this segment Fiftyfour cameras, triggered by active-infra-red sensors, were deployed throughout a nearly 800 km2 area for 12-30 days each. No photos of kit fox were obtained. Surveys of kit fox dens active in 1998 found only one den active in 1999; however, the pair apparently did not have any young again this year. A combination of GAP mapping and ground mapping was used to estimate the habitat potential for restoration of kit fox in western Colorado. Three distinct areas were identified: the Colorado, Gunnison, and Uncompahgre Valleys. The Colorado valley is isolated from the other two by urban development and irrigated agriculture. The Colorado River Valley is the largest contiguous area (685 km2) and has the highest percentage of public land Thus, it is likely the best area for kit fox restoration efforts. Successful restoration of kit fox throughout the entire 1310 km2 of suitable kit fox habitat in Colorado could result in kit fox populations varying from 182-728; based on densities reported in the literature. Primary negative factors to be addressed are housing developments, greater isolation of habitats, and red fox pioneering into salt desert shrub communities. �57 KIT FOX (VULPES MACROTIS) STATUS IN COLORADO Thomas D. I. Beck P.N. Objective Develop and implement a conservation strategy to recover and conserve kit fox populations in Colorado. Segment Objectives 1. 2. 3. Conduct the biological assessment work necessary to develop an effective conservation strategy for the conservation of kit fox in western Colorado. Survey for presence of kit fox in Gunnison and ·uncompahgre Valley areas with infra-red activated cameras. Capture and radio-collar adult and juvenile kit fox in both the Uncompahgre and Gunnison valley areas. METHODS AND MATERIALS Plans for field work were modified significantly in July in response to the dearth of kit fox activity found in the Gunnison and Uncompahgre Valleys. Additionally, much oftheiandscape in the 2 valleys are either marginally suitable or becoming unsuitable through a combination of natural features and human development. Therefore, emphasis was shifted to better map the available habitat and conduct kit fox surveys in the Colorado River Valley. Ground searches for kit fox sign and dens were conducted in the Colorado River Valley on 40 days during September-November 1998. Remote 35mm cameras, activated by active-infra-red sensors, were deployed in areas with fresh kit fox sign, areas with old fox dens, and likely travel areas. The camera system was developed by TrailMaster (Lenexa, KS) and was essentially the same system as used by Beck on black bear (Ursus americanus) studies (Beck 1995). The distance from the camera to the infra-red transmitter was 2 m~ camera height above ground was 15-30_ cm. A pipe was driven into the ground 10 cm in front of the infra-red transmitter and one of several baits (beaver meat, chicken flesh lure) were placed down in the pipe to minimize bird conflict. The recorder unit was programmed so that the infra-red beam had to be broken for 0.15 sec to be recorded ( value of P = 3) and only a 6 sec delay between successive pictures. Color print film, _ASA 400, in rolls of24 and 36 exposures were used. Cameras were left at locations for varying periods of 12-30 days, depending on area, work schedule, and need for cameras in other areas. Observations of an active kit fox den were conducted at dusk on 3 evenings in May 1999, in hopes of documenting presence of pups. Observations were made with binoculars or a night-vision scope at distances of30 to 100 m. Historic den sites in the Uncompahgre Valley were visited in May 1999 to check for kit fox activity. Maps of areal extent of salt desert shrub communities were obtained from CDOW Habitat Section. Acreage of salt desert shrub and irrigated lands were estimated from the GAP maps. The land ownership of the primary habitat (salt desert shrub) was also estimated from the GAP maps. Onsite field mapping was conducted to better assess the utility of much of the salt desert shrub component on private land since much of this land has been undergoing significant housing development. Also, 15-minute quad maps were used for plotting camera locations and all fresh fox sign. �58 RES ULTS AND DISCUSSION Biological Assessment Kit Fox Distribution and Abundance - Analysis of the work reported by Fitzgerald (1996) suggested that summer was the period of lowest trap success, and the majority of search effort had been conducted in the summer. Thus it was deemed prudent to resurvey much of the available kit fox habitat during spring and fall seasons, utilizing a different technology than live-trapping. Ground searches for kit fox dens and spoor were conducted during spring 1998 in the Gunnison and Uncompahgre Valleys. Similar searches were conducted in the Colorado River Valley during September-November 1998. Fifty-four cameras were placed for periods of 12-30 days. Two camera units were stolen. No kit fox were photographed at the 52 sites spread throughout the approximately 800 km2 Colorado River valley. No active fox dens were located during ground searches. A variety of wildlife was photographed: cottontail rabbits (158),jackrabbits (7), antelope ground squirrel (2), coyote (6), badger (6), rock squirrel (6), bobcat (1), chipmunk (1), pronghorn (1), magpie (31 ), owl (1 ), unk. birds (3 ). Frequent rains and cold temperatures resulted in the loss of 38 images during the last 2 weeks of work because the plexiglass dust cover frosted over; causing a frosty blur on the photograph. In May, 1999 a dead male kit fox was recovered along a paved road in Colorado National Monument by DWM Paul Creeden. It was freshly killed when retrieved about 0430. The area is in the pinon-juniper vegetation community, approximately 500 m higher than the desert floor. About 3 hours was spent on a ground search in the area looking for any denning activity or otl;ter sign but none was found It was not the normal vegetation or terrain for kit fox. All kit fox dens-active in 1998 were checked in May 1999 for fox activity. The Montrose landfill den again had a pair of kit fox. This pair was observed on 3 evenings in May but there was no indication of pups. Volunteer observers who work at the landfill reported no pup activity during the summer. None of the other known kit fox dens checked had any indication of kit fox activity. Additionally, 3 of the red fox (Vulpes vulpes) dens active in 1998 which were located in the salt desert shrub community were also active in 1999. Prey Abundance - Comparative prey surveys were initially to be conducted in Fall 1998. However, based on the low numbers of kit fox located in Spring 1998 it was decided to expand general surveys to the Colorado River Valley instead Also, preliminary mapping suggested the Uncompahgre and Gunnison Valleys provided only modest opportunity for kit fox restoration because of limited habitat. In addition, the Colorado Legislature debated a bill in the 1999 session which could have restricted the authority of the CDOW to implement kit fox augmentation. This bill (HB 99-1299) was passed in late April and signed into law in May. However, final wording did not cover the kit fox program (in contrast to original language). However, it was decided not to do extensive prey base work pending resolution of this issue. Potential Habitat The western valleys of Colorado which historically supported kit fox were divided into 3 distinct valley segments: Uncompahgre, Gunnison, Colorado. The Gunnison Valley area was subdivided into 3 segments because the central segment is a region with basalt cap rock close to the surface and den sites are extremely limited. Even badger diggings are uncommon in this reach; which is also the area where 3 roadkill kit fox have been documented in the past 5 years. �59 Based on the GAP vegetation maps, there is an estimated 54 7 km2 of salt desert shrub in the Uncompahgre Valley unit (Fig. 1). This unit also has the highest area of irrigated farm land (528 km2), all of which historically was salt desert shrub. It also has a high proportion of existing salt desert shrub in private land status (45.8%). Ground mapping was conducted to exclude peripheral areas of salt desert shrub as well as areas which have been developed for human housing. Radio telemetry work on kit fox in this valley strongly indicated the kit foxes did not venture into the developed desert areas but remained in a relatively small core area Thus, the suitable kit fox habitat for the near future is more likely about 140 km2. This area is separated from the Gunnison Valley habitats by agricultural land, housing developments, a highway, and a river; resulting in a 7-9 km long obstacle course to inhibit dispersal. Based on the range of kit fox densities summarized by White and Garrott (1997) (0.160. 7/km2) this area of habitat would likely only support 22-98 kit foxes. The Gunnison Valley area has approximately 609 km2 of salt desert shrub vegetation (Fig. 2). Only 2 kit fox were located in this region in 1998, 5 were trapped here during 1993-1995, and 3 road kills have been documented. The East Gunnison area (179 km2) only has about 18 km2 of irrigated land but does have some residential development and the irrigated land may serve as a barrier to animals moving east of Surface Creek. Thus, suitable kit fox habitat for recovery is closer to 120 km2, of which 66% is in public ownership. The Central Gunnison area (250 km2) has about 10 km2 of irrigated farmland. This is the area of volcanic rock near the surface which appears to severely limit underground dens. Most of this land is in public ownership (81°/o) and housing development is less a problem here. Thus, about 225 km2of land is marginally suitable for kit fox recovery. Because of the lack of den sites for escapement, this area will likely always be a mortality sink for neighboring populations of kit fox. A small area near Cheney Reservoir has the more typical soil formations found in the salt desert shrub areas and did support a kit fox family in 1994-95. No kit fox activity has been recorded there since. The area is private land and the current owner has been unsuccessful at attempts to subdivide for housing, primarily because of issues of potable water supply. Unsolicited information from local varmint hunters suggests that red fox have been colonizing this area rapidly during the past 2 years. Earlier work on kit fox surveys in this area did not document red fox but both dens and red foxes were observed here in spring of 1998. The North Gunnison area abuts the Colorado River and Grand Junction urban area to the north. This area is highly impacted by housing development throughout the salt desert shrub type. Based on the GAP maps, there are 180 km2 of salt desert shrub in this unit, of which 48% is private. There is an additional 59 km2 of irrigated farm land. No kit fox have been located in this area by any of our surveys this decade. Suitable area for kit fox restoration in the near future is probably closer to 140 km2. Total suitable habitat for kit fox in the Gunnison Valley is approximately 385 km2. It is unlikely this area would support high densities of kit fox because of the denning problem in the Central area Thus, a projection of 50-150 kit fox possible should restoration be successful seems to be the best possible outcome. The housing development in the north coupled with the den problems in the central suggest that the East unit may be the best area for kit fox populations, which could then support 25-85 kit fox. The Colorado River Valley supports the largest intact area of salt desert shrub, about 797 km2 based on the GAP map. Most (85%) is in public ownership. Irrigated lands amount to 319 km2, mostly in the eastern end (Fig. 3). Ten kit fox were captured here in 1994 and 1995. An active den with a single kit fox was discovered in 1997 near Fruita However, the lack of any photographs of kit fox or sign at historic kit fox dens in 1998 strongly suggests the current density of kit fox in this valley is low. This unit appears to have the lowest probability for loss of kit fox habitat and will likely have at least 685 km 2 of suitable habitat for a long period. Based on kit fox densities reported in White and Garrott (1997), this area could potentially support 110-480 kit fox if fully occupied. �60 Capture Efforts It was decided not to trap and collar adult foxes during the whelping season because of the added stress of this activity to the individual foxes (Cypher 1997). The small number of kit foxes present seems to justify minimum intrusion during this period. There have been no juveniles located in 1998 or 1999 to tag. Literature Cited Beck, T. D. I. 1995. Development of black bear inventory techniques. Colo. Div. Wildlife, Fed. Aid. Rep. W-153-R-8. l lpp. Cypher, B. L. 1997. Effects ofradiocollars on San Joaquin Kit Foxes. J. Wildl. Manage. 61(4):14121423. Fitzgerald, J.P. 1996. Status and distribution of the kit fox (Vulpes macrotis) in Western Colorado. Colo. Div. Wildlife, Final Fed. Aid. Rept. W-153-R-7. 78 pp. White, P. J. and R A Garrott. 1997. Factors regulating kit fox populations. Can. J. Zool. 75: 19821988. �61 N A GREAT OUTDOORS C0L0llAD0 • Colorado Division of Wildlife May, 1999 e:::::! Irrigated Crop Type Ill Selected Vegetation Types• O 1 2 Kilometers ~ • Desert Shrub Saltbush Fans & Rats Greasewood Fans & Flats Figure 1. Selected vegetation types in the Uncompahgre Plateau. O 1 2 ~ 3 4 Miles i,i.iiiiij �N A O'I t-.J ,,,subUnlt Boundaries c;J Irrigated Crop Type ,,,.. f lm1ll Sefected Vegetation Types• • Desert Shrub Saltbush Fans & Flats ats / Grease~d Fans & Fla:_; ~ o 1 2 Kilometers ~ 01234Mlles ,.....-.J G!UAl" OUl"D00Jl5 Co LORAnO • Colorado Division of Wildlife May, 1999 Figure 2. Selected vegetation types in the Gunnison River Valley. DELTA �N A 0 Irrigated Crop Type 11111111 Selected Vegetation Types• Desert Shrub Saltbush Fans & Flats Greasewood Fans & Flats \ 0 1 2 Kilometers ~ -- 1 2 3 4 Miles a GREAT OUTD00ll5. COLORAnO (_ '- 0 0 of Wildlife 99 Figure 3. Selected vegetation types in the Colorado River Valley. �25 Colorado Division of Wildlife Wildlife Research Report July 2000 JOB PROGRESS REPORT State of Colorado Cost Center 3430 Project No. W-153-R-13 Mammals Research Program Work Package No. ___0"""6=-a6=3_ _ _ _ __ Task No. Kit Fox Conservation Kit Fox Augmentation Study 1 Period covered: July I, 1999 - June 30, 2000 Author: T.D.I. Beck Personnel: T. Beck, G. Byrne, CDOW; E. Everett, B. Miller; Denver Zoological Foundation. ABSTRACT Relative abundance of kit fox prey surveys were developed as a written protocol. However, administrative problems prevented conduct of the late-summer 1999 field surveys. Spring 2000 surveys which had been planned were deferred because of organizational changes within CDOW relative to management of threatened and endangered species. The new organizational unit is to develop a prioritization process for allocating resources among species and habitats and it was believed prudent to defer further field work pending this process since active kit fox augmentation had not begun. Surveys of historic kit fox dens in Uncompahgre Valley did not find evidence of pup production in 1999 and no active kit fox dens were located in spring 2000. Data on historic range rehabilitation projects in the shadscale desert communities of western Colorado was compiled. �27 KIT FOX AUGMENTATION STUDY Thomas D. I. Beck SEGMENT OBJECTIVES I . Compare relative abundance of potential kit fox prey between Uncompahgre and Colorado river valleys. 2. Capture as many kit fox as possible in Uncompahgre Valley for DNA work and pup dispersal. 3. Develop a kit fox reintroduction plan for Mesa County. METHODS AND MATERIALS Sampling protocols were developed for comparing relative abundance of small rodents and cottontail rabbits at 2 sites in each valley. Small rodent surveys were to use a 200-trap concentric trapping web. Snap traps were selected rather than live trapping because of concerns about hantavirus prevalence in these areas. Because such trapping removes captured individuals, trapping was to be conducted during late summer; presumably the time of highest animal density. Thus it would be unlikely to create a local depleted zone. Museum collections from these areas of Colorado are scarce so all trapped animals would have been saved for museum collections. Rabbit abundance was to be meas_ured by a combination of mark-resight estimation and spotlight counts. Hopefully the spotlight counts could be indexed to population estimations to provide a long-term monitoring index. Rabbit population monitoring would be conducted in early-spring; presumably the time of lowest animal abundance. Selection of only 2 sites per valley probably would not accurately represent the variation to be expected throughout the valley. However, this was the most that could be done with manpower and budget. Weekly precipitation records were collected from 6 stations maintained by NOAA in the Uncompahgre and Colorado river valleys in Colorado (Montrose, Olathe, Delta, Grand Junction, Fruita, Loma) for the growing season (March-October) for the period 1970-1998. Not all stations had data for each year. Historic kit fox dens were visited in July and August 1999 to survey for adult fox presence and pup •production. Historic dens and den areas were visited in April 2000 to examine for kit fox use. Records compiled by Ron Kufeld, CDOW retired, were searched to document all range restoration projects •conducted in the low elevation desert valleys. This data base was believed to be complete for the period 1960-1996. During other field activities, visits were made to most of the range project sites to see if any noticeable differences could be seen from surrounding desert vegetation. RESULTS AND DISCUSSIONS Because of administrative delays in allocating budgets and temporary FTE' s, the trapping webs could not be set and run prior to the graduate student having to attend classes. Thus the late-summer 1999 prey studies were deferred to spring of 2000. During late-fall 1999 the Colorado Div. of Wildlife altered its organizational structure and created a new section to deal with Species Conservation; specifically species �28 currently on either a federal or state list of threatened or endangered species or a species in decline with potential for listing. Thus lead role in kit fox restoration shifted to this new section. Discussions among administrators resulted in the decision to defer further field studies until the new section could develop a prioritiz.a.tion process for addressing the species conservation needs. Thus prey abundance surveys were deferred during spring 2000; awaiting the development of the new section priorities. All material collected during our kit fox research surveys has been copied to the new section. Den site surveys in summer of 1999 did not produce any evidence of pup production. There was only one active den found in spring 1999 and for the third consecutive year no pups were produced at this den. No active dens were located in spring 2000 in a survey of historic den sites. Extensive surveys were not conducted in areas where we had previously failed to find active kit fox dens based on the rationale that this declining population will not produce significant dispersal. Based on the lack of finding any active dens, no trapping operations were conducted in 2000. Earlier surveys based on trapping and infra-red photography clearly indicate that trapping rarely produced kit fox captures when field surveys could not find evidence of their presence prior to trapping. The growing season precipitation data was examined to see if any seasonal patterns emerged. The variability within a year and among sites was extremely high and no patterns by year or area were apparent. This is likely a reflection of the nature of the summer precipitation events, which are part of a regional monsoon pattern dominated by afternoon thunderstorms. These storm events are quite local in scope yet capable of delivering 2-3 cm of rain in less than an hour. Casual surveys of range rehabilitation sites indicated no difference from surrounding areas. Thus, when developing augmentation plans no special accord needs to be given to these historic management sites. No formal augmentation plan was developed pending results from the new organizational unit on priorities. However, the material needed to do so is on file. Contacts have been made with 2 surrounding states (Utah and Arizona) for sources of kit fox should augmentation move forward. Representatives of both states were confident of being able to provide kit foxes in adequate numbers. � https://cpw.cvlcollections.org/files/original/156cff7c1f3a7693c3dff9a140318378.pdf bddc59ce29c1e34a4c62424d2f469fd0 PDF Text Text 107 Colorado Division of Wildlife Wildlife Research Report July 1998 JOB PROGRESS REPORT State of _____c...o....,.l..,o""'r..,a.,,d..,o"------ cost center 3430 -_1...s...J...-...B...-_.1....1____ : Mammals Program Project No. _ _....,w.... Work Package No. Species of special concern 0663 Lynx conservation Recovery Task No. Period Covered: July 1, 1997 - June 30, 1998 Author: D. F. Reed ABSTRACT A conservation strategy/recovery plan for lynx in Colorado was prepared and the draft revised January 12, 1998 (Chapter 1 in Seidel et al. 1998). This conservation strategy to recover and conserve lynx populations in Colorado provided general background on the species and generally identified the habitat goals and objectives of identifying potential habitat, identifying linkage zones, selecting suitable habitats, refining habitat suitability models, and refining habitat protection, as well as, conservation actions to be taken. Subsequently, more specific protocols were developed to assess potential lynx habitat based on the species' primary prey, the snowshoe hare.• The first protocol was for the winter track survey which was conducted during the winter. The second was for the snowshoe hare pellet counts (Krebs plots) which were begun toward the end of the segment. The routes selected for the winter track surveys were based upon accessibility and transportation mode (snowmobile, cross-country skis, etc.), and hence, were not randomly selected. Furthermore, sample sizes (routes per forest types) were small. Conclusions about representativeness should be made with caution. �109 LYNX CONSERVATION/RECOVERY Dale F. Reed P,N, OBJECTIVE Develop a conservation strategy to recover and conserve lynx populations in Colorado. SEGMENT OBJECTIVES 1. Prepare a conservation strategy/recovery plan for lynx in Colorado. 2. Assess potential habitat for lynx reintroduction. STUDY AREA The study area includes the forests (Aspen, Douglas Fir, Lodgepole Pine, mixed conifer, mixed forest, Ponderosa Pine, and Spruce-fir) and deciduous oak above 7,500 ft throughout the north-south western half of Colorado. METHODS AND MATERIALS The methods used in preparing the "Conservation Strategy/Recovery Plan" (Seidel et al. 1998) involved extensive literature review, writing, rewriting, editing, and coordination with the numerous authors, species experts, and agencies participating in its completion. The methods used in assessing potential habitat for lynx (Felis lynx) involved 1) a winter track survey where snowshoe hare (Lepus americanus) tracks (crossings and parallel movements) were counted in snow 24-28 hours after snowfalls (described by Byrne 1998) and 2) counts of snowshoe hare pellets (Kreb's plots; Krebs et al. 1987) via randomly selected points in the forest types and deciduous oak (see Study Area) as determined by statewide GAP data. RESULTS Results as in the "Conservation Strategy/Recovary Plan" are summarized in the following Executive Summary (Seidel et al. 1998:vii-x): EXECUTIVE SUMMARY Endangered Species Since life began on this planet many species have come and gone through natural changes in physical and biological conditions. Since these extinctions occur naturally why should we spend money and effort to conserve species that are nearing this end? How do they benefit society if restored? Congress addressed these questions in the preamble of the Endangered Species Act of 1973, "recognizing that endangered species of fish, wildlife, and plants are of aesthetic, ecological, educational, historical, recreational, and scientific value to the Nation and its people." Congress further stated its intent to protect and conserve the ecosystems and habitats. The State of Colorado �110 has adopted similar protections for species of special concern. The primary force driving the loss of these species is habitat destruction or human exploitation. While we do not know the causes for the decline in lynx (Lynx canadensis) in the Southern Rocky Mountains (SRM) human activity is at least partly responsible. We have the knowledge to conserve and reestablish these species. We also have the responsibility. History and Distribution Lynx historically occurred, at low densities, as forest carnivores in the SRM. There have been 12 investigations reported in Colorado since 1979 to document the presence of lynx or wolverine. Although the presence of both species has been suspected but not confirmed, and no viable populations have been found. There have been no field studies in the southern ranges in Wyoming or in New Mexico. Wyoming conducted a survey in 1986 of reported sightings but few of these reports were documented (Reeve et al. 1986). This draft does not include information on the Wyoming and New Mexico portions of the SRM. The final strategy will include a coverage of these areas. Any future decisions regarding lynx that affect neighboring states will require further investigation and coordination. Current Status At this time the lynx is classified as a federal "candidate" species by the Fish and Wildlife Service. In Colorado the lynx is classified as state endangered species and in New Mexico are protected species. Wyoming classifies lynx as Native Species Status Category II (Oakleaf et al. 1996). The Forest Service classifies lynx as a "sensitive species." Conservation Strategy On July 8, 1997, representatives of the U.S. Forest Service, U.S. Fish and Wildlife Service and the Colorado Division of Wildlife (CDOW) met to discuss a cooperative program for the conservation and reestablishment of lynx and wolverine in Colorado. On August 4, 1997, those agencies plus the National Park Service signed a letter agreeing to jointly prepare "A Candidate Conservation Strategy for Lynx and Wolverine in Colorado." The Colorado Division of Wildlife is the lead agency for these state listed species. Species Goal The goal of this strategy is the conservation and reestablishment of lynx within their former ranges by establishing populations which reprodu~e sufficiently to allow emigration into unoccupied suitable habitats. If this is accomplished, it would permit the downlisting of these species by the states from endangered. Both species are considered in this document due to expected economies of scale that will be realized in the habitat assessment, acquisition, and monitoring portions of this strategy. This document could serve as the basis for future recovery efforts in Colorado. Risks to the Species We can only speculate on the history of lynx populations in Colorado. Some speculate that trapping, hunting, and poisoning played a significant role in population reductions. If so, Amendment 14 approved by the voters of Colorado in 1996 greatly reduced that threat to future populations because it strongly restricts the use of poisons, leghold, kill-type and snare trapping devices in the State of Colorado. Removal of this historical source of mortality would enhance the probability of success for conservation and/or potential future reintroductions of lynx. A second primary mortality factor for lynx populations reported from other sites within their distribution is mortality of kittens or kits either through starvation or disturbance during rearing. �111 These are factors that can be addressed by specific management practices. Most of the potential habitat for these species in the SRM occurs on public land with the majority on National Forest System lands. Many National Forest lands in Colorado are designated and managed as wilderness. It is currently not known how much of that area has potential to support lynx populations. Habitat Goals for Lynx The first goal is to describe and map potential habitats for lynx in the SRM. The second goal is to protect those habitats that may be important for whatever lynx may still exist in the state and potential habitats for future reintroduction or reestablishment efforts. Habitat Evaluation Actions for Lynx Lynx foraging habitat requirements are inextricably linked to habitat requirements and population distribution of snowshoe hares (Lepus americanus) because that species is a staple prey item. Therefore habitats with abundant snowshoe hare populations would be considered the primary criterion for selecting potential locations for lynx reintroductions. The first step in the strategy will be to assess the potential habitats both for foraging and denning potential using GIS vegetation mapping. Key areas will be identified that contain at least the minimum requirements for species survival. The criteria used for this selection will come from the significant body of knowledge that exists for lynx habitats in other regions (Table 1). These same criteria have been used to develop potential habitat management guidelines that will be recommended to land managers until information obtained from monitoring reintroduced individuals suggests changes to those recommendations. Once key blocks of potential habitat are identified, the CDOW intends to conduct two surveys to detect snowshoe hare occurrence and estimate densities to validate or adjust initial habitat assumptions. Data from both surveys will be used to identify the 2 habitats with the greatest potential to support Colorado lynx. The actual reintroduction sites will be identified from the field surveys and modified or confirmed from advice from a peer review panel of qualified scientists with recognized expertise in lynx ecology and management. Reestablishment of lynx Reestablishment of viable lynx populations in Colorado requires reintroductions. A reintroduction would involve assessing potential release sites for habitat suitability, radio-marking lynx prior to release, and intensive monitoring of radio-marked animals. Lynx populations in Canada and Alaska currently are increasing toward peak population levels (Fontana pers. commun. 1997). Lynx selected for reintroduction otherwise would be killed and sold as pelts to be traded as fur. By paying a compensatory price, the CDOW can obtain lynx for reintroduction at a reasonable price and increase their chances of survival. When lynx are at the lower end of their population cycle, population levels would be insufficient to provide sufficient animals for reintroduction. Since wild lynx populations cycle every 10 to 12 years, there is a narrow 4- to 5-year window of opportunity to reintroduce lynx in Colorado's most suitable habitats. It is estimated that lynx would be needed to be released each year ( in each of 2 habitats each year) for 2 successive years to establish several viable breeding populations. Species Monitoring Radio-marked lynx will be intensively monitored by frequent relocations both from the air and the ground. Recorded data would include habitat selection patterns, seasonal home ranges, reproductive success, survival, and probable cause of death when mortalities are detected. This information would be �112 evaluated critically to determine if additional releases of these species into unoccupied habitats is warranted before additional releases are attempted. Revision Any strategies for conserving lynx in the SRM must be regarded as tentative because very little is known about their specific ecology and habitat requirements in the SRM. As the body of knowledge increases and new information is gained from post-release monitoring of both lynx and wolverine, the Conservation Strategy would be revised. Formal lynx and wolverine status reviews would be scheduled periodically (e.g., annually) by the Lynx and Wolverine Conservation Strategy Team to revise the lynx/wolverine Conservation Strategy as needed. Budget A draft budget is included in Appendix B. It is estimated that the cost to complete the first 3 years of this Conservation Strategy could approximate $2.5 million. The Division of Wildlife has committed to funding $700,000, leaving $1.8 million to be acquired from other sources. Habitat Protection Following the pre-release habitat suitability surveys, it will be necessary to develop interim guidelines to delineate, preserve and protect lynx habitats on public lands that are deemed critical to the conservation of these species. These interim habitat protection guidelines periodically would be revised as additional information surfaces from the post-release monitoring of radiomarked individuals. Guidelines will be based on the best scientific and economic information available; will conform to existing laws and regulations; and will be subjected to further peer and public review prior to implementation. These guidelines will be developed and amended into this document at a later date subject to concurrence by all signatories. Results of the winter track surveys were reported by Byrne (1998). Counts of snowshoe hare pellets (Kreb's plots; Krebs et al. 1987) were only begun toward the end of this segment - hence the results from this second method will not be available until the next segment. LITERATURE CITED Byrne, G. 1998. A Colorado winter track survey for snowshoe hares and other species. Colo. Div. Wildl. 35pp. Krebs, c. J., B. s. Gilbert, s. Boutin, and R. Boonstra 1987. Estimation of snowshoe hare density from turd transects. Can. J. Zool. 65:565-567. Seidel, J., B. Andree, s. Berlinger, K. Buell, G. Byrne, B. Gill, D. Kenvin, and D. Reed. 1998. Draft strategy for the conservation and reestablishment of lynx and wolverine in the southern Rocky Mountains. U.S. Forest service, National Park Service, U.S. Fish and Wildlife Service, and Colorado Division of Wildlife. 115pp. ~· Prepar:;~:2J/4&5JEU/ Dale F. Reed Wildlife Researcher �66 �65 Coiorado Division of Wildlife Wildlife Research Report July 1999 JOB FINAL REPORT State of ------=-=~=="------Colorado Project No. ----'-W-'---"1""'"53"'""--=-R"--""'12~---Work Package No. _0~6"-'6'""3_ _ _ _ _ __ Task No. _ _ _ _2 " ' - - - - - - - - - Cost Center 3430 Mammals Program Species of Special Concern Lynx Conservation/Recovery Period Covered: July 1, 1998 - June 30, 1999 Author: D. F. Reed ABSTRACT Snowshoe hare pellet counts (Krebs plots) were analyzed and a Division report prepared. BDOWD14185 �67 LYNX CONSERVATION/RECOVERY Dale F. Reed P.N. OBJECTIVE Task 2 - Develop a conservation strategy to recover and conserve lynx populations in Colorado. SEGMENT OBJECTIVES 1. Analyze data and prepare report. STUDY AREA The study area includes the forests (Aspen. Douglas Fir, Lodgepole Pine, mixed conifer, mixed forest, Ponderosa Pine, and Spruce-fir) and deciduous oak above 7,500 ft throughout the north-south western half of Colorado as reported by Reed (1998). METIIODS AND MATERIALS The methods used in assessing potential habitat for lynx (Fe/is lynx) involved counts of snowshoe hare pellets (Kreb' s plots; Krebs et al. 1987) via randomly selected points in forest types and deciduous oak as determined by statewide GAP data and as reported by Reed (I 998). RESULTS The results are in the report titled "Snowshoe hare density/distribution estimates - potential habitat for reintroduced lynx in Colorado" (Reed and Kindler 1999). LITERATURE CITED Krebs, C. J., B. S. Gilbert, S. Boutin, and R. Boonstra. 1987. Estimation of snowshoe hare density from turd transects. Can. J. Zool. 65:565-567. Reed, D. F. 1998. Lynx conservation/recovery. Colo. Div. Wildl. Res. Rep. July, 107-l 12pp. Reed, D. F., and J. Kindler. 1999. Snowshoe hare density/distribution estimates - potential habitat for reintroduced lynx in Colorado. Fort Collins, CO :Colo. Div. ofWildl. Division report; no. (in review). Prepared by _ _ _ _ _ _ _ __ Dale F. Reed � https://cpw.cvlcollections.org/files/original/2eb8d87cd168378fbddb2cf59080f9d6.pdf b14610cb2f685095678a3ec48ea998fa PDF Text Text 47 Colorado Division of Wildlife Wildlife Research Report July 2000 JOB PROGRESS REPORT State of _ _ _ _ ____.C._.o=lo=rad=o=--------Project No. ________ W__--=-1~53__-___ R----1---4_______ Mammals Research Work Package No. _ _---'0'-'6a..a.7..a;..0_ _ _ _ __ Lynx Conservation Task No. 2 -------------- Lynx Veterinary Services and Diagnostics Period Covered: July 1, 1999 - June 30, 2000. Author: M. A. Wild. Personnel: T. Shenk, S. Dieterich, H. Dieterich, T. R. Spraker, D. H. Gould. ABSTRACT Fifty-six lynx (34 adult females, 19 adult males, 2 juvenile females, 1 juvenile male) were received from British Columbia, the Yukon, and Alaska during January - April 2000. Lynx arrived in generally good condition with the exception of one female lynx that required euthanasia due to an extensive foot lesion from trapping. All lynx were anesthetized (Telazol 2-3 mg/kg IM) for physical examination and identification at least twice during the =::3 week acclimation period at the captive holding facility. During captivity lynx maintained or improved body condition. Using abdominal radiographs, we diagnosed one female as pregnant and three others were suspicious. Seven lynx had lesions to 1-2 toes that required treatment. One of these lynx also had two fractured metacarpal bones that required splinting. All recovered and were released in Spring 2000. One lynx from the 1999 release was recaptured in March 2000. The lynx was emaciated but no other abnormalities were determined. He was released in May 2000 after rehabilitation. Fifteen mortalities were recorded in free-ranging lynx during FY2000. Eight of the carcasses were in advanced stages of decay (or only the collar was found) and cause of death could not be determined; however, in three of these cases we were able to rule out starvation as cause of death. Two lynx died from gunshot wounds, two from trauma, and one from predation. One trauma case and the predation case were in poor body condition. One male kitten died from starvation about 7 weeks postrelease. A female yearling died of pneumonic plague in Hinsdale County, Colorado. �48 �49 LYNX VETERINARY SERVICES AND DIAGNOSTICS Margaret A. Wild P. N. OBJECTIVES I. Provide veterinary care and diagnostic services for reintroduced lynx. SEGMENT OBJECTIVES I. Provide veterinary care and diagnostic services for reintroduced lynx. METHODS AND MATERIALS Lynx were trapped in British Columbia (BC) and the Yukon, Canada, and Alaska then transported by truck and/or airline to holding pens at the Frisco Creek Wildlife Hospital and Rehabilitation Center, Del Norte, Colorado. Care of lynx during captivity was based on the Husbandry and Management Protocol (Attachment I of Addendum A). Post-release, diagnostic and forensic services were provided to support ongoing research and law enforcement efforts. Carcasses were transported to the Colorado State Diagnostic Laboratory, Fort Collins, for examination by a board certified veterinary pathologist and a wildlife veterinarian. Complete post-mortem examination was conducted following the Necropsy Protocol (Attachment 2 of Addendum A). RESULTS AND DISCUSSION Results of the 1999 lynx reintroduction are summarized in Addendum A. An additional 56 lynx were received January-April 2000. Lynx appeared healthy on arrival with no transportation-related injuries. Captive management and husbandry techniques were similar to those reported in 1999; however, all lynx were held at least 3 weeks for acclimation and fattening prior to release. Releases occurred in early April through late May. Daily clinical assessment suggested that lynx adjusted well to confinement in our isolated holding facility. Daily feed intake was not quantified this year however consumption appeared similar to the average I 0-15% of body weight/day over a weekly basis that was observed in 1999. Health Assessment We anesthetized each lynx with about 2.5-3 mg/kg body weight Telazol IM for initial examination and sample collection (as described in Attachment 1 of Addendum A). One lynx (BC00F 17) was euthanized upon initial exam due to extensive foot lesions. The other 55 lynx were anesthetized again with about 2-2.5 mg/kg Telazol IM for placement of a radiocollar just prior to release. In general, lynx arrived in good condition. Upon arrival, females from British Columbia (BC; n = 9) and The Yukon (n = 20) averaged 8.86 kg (SE 0.28) and 8. 72 kg (SE= 0.22), respectively. At examination prior to release, body weights had increased to 10.23 kg (SE= 0.19) and 10.28 kg (SE= 0.13) for BC and Yukon females, respectively (Fig. I). Adult females from Alaska (n = 4) averaged 9.82 kg (SE= 0.67) at arrival and 11.91 (SE= 0.63) prior to release; however, one of these was known pregnant and three were suspected to be pregnant. Body weight change was less marked in male lynx. Upon arrival, adult males from BC (n = 9), The Yukon (n = �50 6), and Alaska (n = 4) averaged 12.2 kg (SE 0.53), 10.64 kg (SE= 0.32), and 11.23 (SE= 0.32) respectively. At examination prior to release, body weights were 12.93 kg (SE= 0.45), 12.13 kg (SE= 0.17), and 13.27 (SE= 0.63) for BC, Yukon, and Alaska males, respectively (Fig. I). One male kitten from The Yukon weighed 6.2 kg on arrival and 7.87 kg prior to release. Two female kittens from Alaska averaged 6.89 kg (SE= 0.34) on arrival and 9.17 kg (SE= 0.08) prior to release. Using radiographic examination of the growth plates of the distal radius and ulna as an indicator of age (Nava 1970), we determined that three juveniles were received (YK00M5, AK.00Fl, and AK.00F4). Apparently, the body length measurements that we collected in 1999 to distinguish kittens from adults proved useful in field evaluation oflynx and minimized the number of kittens that we received. This year we also adopted a new approach for estimating age of lynx based on percent pulp cavity in the canine tooth. This method has been used to age canine teeth collected from lynx carcasses (K. Poole, pers. comm.), but we attempted to collect the infonnation in situ using skull radiographs. Evaluation of the method is underway. Several injuries to lynx were found. Eight lynx had fractured toes or toes previously amputated by veterinarians in BC or The Yukon. Damage to the foot of one of these lynx (BC00F 17) was severe and involved the middle three toes on a front paw. Due to the poor prognosis, we euthanized this lynx at the time of initial veterinary examination. Of the remaining seven lynx, three lynx had one toe amputated (one of these lynx also had two fractured metacarpal bones that required splinting), two lynx had one toe amputated and an additional toe damaged, one lynx had two toes amputated, and one lynx had two fractured toes that were ankylosed but were not treated. Five other lynx had minor lacerations on the legs or face that healed without complications. Recapture On 24 March 2000, we recaptured a lynx (AK.99M9) released in 1999. Field observations by the lynx monitoring crew suggested that the lynx was severely emaciated. Live trapping the lynx failed, so we darted the lynx with Telazol (3 mg/kg) using the Dan-Inject CO2 pistol. Physical examination revealed severe emaciation (6 kg). It can be assumed that the lynx would have died if not recaptured. Blood work was unremarkable with the exception of a low titer to Toxoplasmosis; however, this was unlikely the cause of the abnormality. No underlying disease condition was found that would explain the debilitation. The lynx responded well to supportive care at the holding facility and was released in May 2000. Post-mortem Examination Fifteen mortalities were recorded in free-ranging lynx during FY2000 (Table 1; one of these was included in Addendum A as well). Eight of the carcasses were in advanced stages of decay (or only the collar was found) and cause of death could not be determined; however, in three of these cases we were able to rule out starvation as cause of death based on the presence oflipid in bone marrow. Two lynx died from gunshot wounds, two from trauma, and one from predation (apparently from a bobcat). One trauma case and the predation case were in poor body condition. One male kitten died from starvation about 7 weeks post-release. A female yearling died of pneumonic plague. The carcass was recovered in Hinsdale County, Colorado. Necropsy findings showed an acute fibrinous pneumonia. Plague was diagnosed by fluorescent antibody test and isolation of Yersinia pestis from lung and spleen samples. After this diagnosis, we retrieved bone marrow samples to test for plague in six other lynx that had died from unknown causes in 1999 and 2000. Fluorescent antibody test was negative on all of these cases. Plague in rodents is not uncommon in southwestern Colorado. This lynx most likely consumed a rodent or rabbit infected with plague. �51 LITERATURE CITED Nava, J. 1970. The reproductive biology of the Alaska lynx. M.S. Thesis. Univ. Alas., Fairbanks. 14lpp. Table l. Summary of mortalities in free-ranging lynx during FY 2000. Date1 Animal ID Sex Age State of Carcass Cause of Death 08/26/99 09/15/99 10/22/99 11/03/99 11/17/99 11/24/99 01/26/00 01/29/00 05/23/00 05/23/00 06/08/00 06/15/00 06/15/00 06/22/00 07/29/00 AK.99F18 AK.99F10 BC99-02 AK.99F27 AK.99M6 AK.99F15 YK99F4 AK.99Mll BCOOF3 YKOOM5 YK.99F3 AKOOF4 YK.99M6 AK.99Fl3 YKOOF17 F F 2 2 4 1 5 3 Good Poor Poor Good Good Good Good Minimal remains Good Good Poor Collar only Collar only Minimal remains Poor Trauma, emaciation2 Unknown, not starvation2 Unknown, not starvation2 Shot Shot Blunt trauma Predation, emaciation Unknown Pneumonic plague Starvation2 Unknown, not starvation2 Unknown Unknown Unknown Unknown2 M F M F F 5 M 3 1 0 3 0 4 1 2 F M F F M F F 1 Date mortality confirmed, date of death may have been earlier. 2 Plague negative on FA of bone marrow 14 12 'bl) c 10 i ·- 8 ~ 6 "'O 0 4 >. CQ I I ii d) 2 0 BC YK Females AK BC YK AK Males Fig. 1. Initial body weight (stippled) and body weight at release (solid) of female and male lynx reintroduced in 2000. �52 ADDENDUM A Status Report on the Health of Lynx Reintroduced to Colorado by Margaret A. Wild, DVM Colorado Division of Wildlife With additional information from Herman Dieterich, DVM, DACVS and Susan Dieterich Frisco Creek Wildlife Hospital and Rehabilitation Center For the 30-31 August 1999 CLAWS/LAT Meeting GOAL To assure optimal health of lynx reintroduced into Colorado. OBJECTIVES Objectives of the health program were: 1. 2. 3. 4. To protect the health of wild and domestic animals in Colorado. To optimize health oflynx released into Colorado. To optimiz.e welfare oflynx during transport and holding. To provide diagnostic and forensic services in the case of mortalities. MATERIALS AND METHODS Lynx were trapped in British Columbia (BC) and the Yukon, Canada, and Alaska then transported by truck and/or airline to holding pens at the Frisco Creek Wildlife Hospital and Rehabilitation Center, Del Norte, Colorado. Care of lynx during captivity was based on the Husbandry and Management Protocol (Attachment 1). Post-release, diagnostic and forensic services were provided to support ongoing research and law enforcement efforts. Carcasses were transported to the Colorado State Diagnostic Laboratory, Fort Collins, for examination by a board certified veterinary pathologist and a wi_ldlife veterinarian. Complete post-mortem examination was conducted following the Necropsy Protocol (Attachment 2). PRELIMINARY RES ULTS AND DISCUSSION Captive Management and Husbandry Forty-two lynx were received at the holding facility between 29 January and 14 April 1999. All appeared healthy on arrival with no transportation-related injuries. Individual lynx were held at the facility for varying lengths of time based on assigned release protocol. Lynx from BC were held <1- 7 wk (median 6.5 wk), from the Yukon 3 - 10 wk (median 8 wk), and from Alaska 3 - 6 wk (median 5) prior to release. Daily clinical assessment suggested that most lynx adjusted well to confinement in our isolated holding �53 facility. However, at least three lynx never adjusted to confinement; two paced the pen causing erosions on the palmar and plantar surfaces of the pads (AKFX99 and AKF499) and one appeared distressed and remained in relatively poor body condition (AKF1599). For other animals, clinical observations suggesting acclimation to confinement were supported by daily feed intake and body weight data. Lynx consumed on average about 10-15% of their body weight/day over a weekly average (Fig. 1) despite the fact that highly palatable feed items were not offered ad libitum. Lynx often fasted rather than consume less palatable items such as the prepared feline diet, fish, or beaver and engorged themselves on highly palatable items such as wild rabbits (up to >2 kg/day). Most lynx gained considerable body weight while in captivity (Fig. 2). In 1999, our initial plan was to hold lynx in captivity for as short a period as possible. We assumed that the animals would be distressed in confinement and would have a greater chance of survival if released immediately to the wild. However, we found that most lynx appeared to adjust to confinement without distress. Further, this period in confinement was likely beneficial to the animals in giving them time to acclimate to Colorado and to gain body weight prior to release. Health Assessment We anesthetized all lynx at least once for examination, and most were anesthetized again for placement of a radiocollar prior to release. We used 3-6 mg/kg body weight Telazol IM to perform 84 anesthetic events on 42 lynx. Of the 79 anesthesias induced with a single injection, mean induction time was 4.3 min (SE 0.2 min; range 2-10 min). Five other anesthesias required supplementation and mean induction time was extended to 16.2 min (SE 7.2 min; range 7-29 min). Full recovery was achieved in all cases on average 91 min (range 27-188 min) after initial Telazol injection. All parameters stayed within acceptable limits; however, lynx receiving 5-6 mg/kg were markedly deeper and oxygen saturation frequently dropped to 70 80%. Supplemental oxygen was administered in these cases. In general, lynx arrived in good condition. Lynx from the Yukon were in markedly better body condition than those from other sites (Fig. 2). Using radiographic examination of the growth plates of the distal radius and ulna as an indicator of age (Nava 1970), we determined that at least six juveniles were received (two from BC, one from the Yukon, and three from Alaska). Total length of the two juveniles from BC was markedly less than that of other lynx from BC Guveniles 78 cm, range in adults 87-97.5 cm). The juvenile from the Yukon was correctly identified using the body length criterion of Slough ( 1996); the juvenile was 90.2 cm while the adults were all well above Slough's 90.5 cm cut-off for adults (range 96105 cm). Identification of juveniles from Alaska was more difficult, likely in part because evaluation occurred later in the year. Three juveniles were identified based on lack of closure of the growth plate; these lynx ranged from 86-89.5 cm in length. Lynx classified as adults ranged from 89.5-106.5 cm. Given these findings, body length can likely be used as an indicator to classify juveniles; however, site-specific cut-off points for body length will be required (i.e., while Slough's classification is appropriate in the Yukon, it appears less reliable for BC and potentially Alaska). We also used radiography to diagnose advanced pregnancy (>45 days) in lynx. Six lynx (AKF299, AKF399, AKF499, AKF899, AKF1099, AKF1799) had fetuses with calcified skeletons present and two other lynx had radiographic evidence suggesting earlier pregnancy (AKF599, YKF599). Of these pregnant females, three have died as of 28 August 1999 (one starvation, one hit-by-car, one undetermined cause). Five lynx had lesions or fractures of one to three toes that required medical treatment or surgical amputation. Two lynx (BC99-8 and AKF499) had minor infections on one toe of a front foot that were treated medically. These animals recovered fully prior to release. Two lynx had fractures and/or freeze �54 damage to two toes on a front foot, with one toe amputated from each lynx (AKf 1099, AKf 1899). The remaining lynx had fractures and freeze damage to three toes on a hind foot that required amputation of all three toes (BC99-15). Of the three lynx released with toes amputated, two are alive and one has died (cause of death unknown, although animal was slightly emaciated and had traumatic injuries) as of 28 August 1999. Sample collection, treatments, and identification were as described in the Husbandry and Management Protocol; however, eartags were placed only in lynx from BC. An incidental finding observed in the majority oflynx from the Yukon and Alaska were pinpoint to 2 mm slightly raised lesions on the oral mucosa, primarily the ventral surface of the tongue. Biopsy samples were submitted to Colorado State Diagnostic Laboratory for histopathology. Results indicated the plaques were caused by a slightly thickened epidermal layer, with no specific lesions present. Otherwise, lynx remained healthy in captivity with the exception of one adult male (YKM 199) and one adult female (BC99-06). Lynx YKM199 exhibited paresis and muscle weakness of about I wk duration. Results of diagnostic procedures were unremarkable, with the exception of apparently enlarged kidneys on radiographs. The lynx was euthanized 29 April 1999 and a thorough post-mortem examination was performed. No gross or histologic lesions were found (kidneys were normal despite radiographic interpretation). Kidney lead levels were normal. Serology for Felv/FIV, CDV, FIP and Toxoplasma gondii were negative. Rabies and parasitology exam were negative. The cause of paresis in this lynx was undetermined; however, infectious diseases that could have affected other lynx in the holding facility were excluded from the list of possible causes. Lynx BC99-06 was diagnosed with an abdominal hernia after recapture. The lynx was initially released under protocol 1, then recaptured in a debilitated, emaciated state. The lynx was nursed back to health on a diet slowly increasing in amount. Although no adverse effects were observed during the rehabilitation, an abdominal hernia developed. The rent in the abdominal wall was surgically corrected without complication. Post-mortem Examination In addition to the one lynx euthanized and examined prior to release, as of 28 August 1999 we have examined 10 mortalities (Table I). Starvation was diagnosed as the cause of death in five lynx based on extreme emaciation as evidenced by lack of depot fat, muscle atrophy, and serous atrophy of bone marrow. Three other lynx that died acutely (one shot, two hit-by-car), were in good body condition. Cause of death in two lynx was undetermined. One lynx (AKF899) had undergone severe autolysis and dehydration prior to examination. No internal organs or bone marrow remained for examination. Another lynx (AKF1899) was in a moderate state of autolysis, however, a complete examination was still possible. Gross examination of this animal revealed slight emaciation and traumatic injuries. Histologic examination of tissues is pending. All carcasses have been tested for rabies and internal and external parasites. Rabies examination has been negative in all cases. Incidental findings on parasitology have included Trichinella, Trichuris, coccidia, Toxascaris, Taenia, giardia, and rodent fleas. LITERATURE CITED Nava, J. 1970. The reproductive biology of the Alaska lynx. M.S. Thesis. Univ. Alas., Fairbanks. 14lpp. Slough, B. G. 1996. Estimating lynx population age ratio with pelt-length data. Wild.I. Soc. Bull. 24:495499. �55 Attachment 1 Husbandry and Management of Captive Lynx at Frisco Creek Wildlife Hospital and Rehabilitation Center Husbandry Lynx will be held singly or in pairs in holding pens (about 5 m2) with attached nest boxes (about 0.7 m2) at the Frisco Creek Wildlife Hospital and Rehabilitation Center. Two units, each with 10 pens, are available. Pens are fully enclosed by fencing and have concrete floors. Straw or hay bedding will be placed in the nest box as well as in two additional piles in the pen. Pens will be cleaned daily. Fresh water, snow, and feed will also be provided ad libitum daily. Feed remaining in the pen after 24 hr will be inspected for quality and discarded if rejected by the animal or if unfit for consumption. Quantity of feed provided and removed will be measured daily. Type of feed will vary, but will consist primarily of domestic, cottontail, and jackrabbits, deer and elk meat, and poultry. Additional feeds may include a prepared feline diet, fish, wild birds, and small mammals (excluding plague vector species, i.e., prairie dogs). Head, kidneys, and urinary bladder will be removed from domestic rabbits prior to feeding to prevent transfer of Encephalitozoon cuniculi. Health and behavior oflynx will be observed twice daily. Individual animal records will be maintained and abnormalities reported to attending veterinarians. Site Security Animal housing units will remain under double lock except when access is required for animal care or other approved activities. Access to the holding pens will be limited to caretakers and visitors with approval from the Colorado Division of Wildlife (CDOW) Program Manager. Alarms and other security devices may be installed as needed. Handling When handling or transportation is required, lynx will be placed in their individual nest box or alternatively, a skykennel. If required, the lynx will be transferred from their nest box to a custom squeeze cage where they can be hand-injected with an anesthetic or other treatments. Lynx will be anesthetized by, or under the direct supervision of, a veterinarian with 5 mg/kg Telazol, IM. Temperature, pulse, respiration rate, oxygen saturation, and anesthetic depth will be assessed throughout the anesthetic event. A record for each anesthetic event will be completed (see attached). Supplemental oxygen will be provided by mask as needed. Lynx will not be returned to their pens until recovered from anesthesia. Examination and Animal Identification Lynx will receive a brief visual examination by a veterinarian upon arrival at the holding facility. Abnormalities will be noted and appropriate care provided. After 1-7 days rest, each lynx will be anesthetized for thorough examination. We will obtain a body weight and morphometric measurements, assess body condition, estimate age (based on tooth wear), take a photograph of the face, collect hair and whole blood for DNA analysis (R.Ramey, University of Colorado, Boulder) and blood for archiving, administer penicillin prophylactically (22,000 U/kg, SC), perform a thorough physical examination, and apply needed treatments and markings for identification. The physical examination will include close inspection of the condition of toes, claws, and teeth. We will use radiology to augment diagnosis of injuries, identify juveniles (based on lack of closure of the growth plate of the distal radius and ulna), and confirm late pregnancy. Other diagnostic samples will be collected as needed if abnormalities are found. If �56 written documentation is not available to confirm that required anthelmintic treatments have been provided, we will administer praziquantel (5 mg/kg SC), ivermectin (0.3 mg/kg SC), and topical dusting with carbaryl. Lynx will be identified with two subcutaneously placed transponders (one on dorsal midline, the other in the intermandibular area) and an eartag. A radiocollar will be placed on each lynx about 2-14 days prior to release. If additional anesthesias are required for placement of radiocollar, treatment, etc., determination of body weight and other required procedures may be repeated as well. Treatment and Euthanasia Medical treatments will be provided by, or under the direct supervision of, the attending veterinarians as needed. Surgery will be performed by a board certified veterinary surgeon. If euthanasia is required, ideally it will be performed with the lynx under anesthesia using an overdose of barbiturates. A diagnostic necropsy will be performed on all mortalities by a board certified veterinary pathologist. �57 ANIMAL CAPTURE RECORD Species: -L< nx- - - - - - - - - - - - - - - - - - - ~ I I 99 Date: Name of investigator(s): Sex: M 0 F 0 Est □ Actual O Age: _ _ Weight: _ _ __ Est □ Actual O Body Condition: Animal No. Excel Good O 0 Fair D Poor O Purpose of Capture: Pre-release exam. -----------------------------------Location of Capture: _J)ieterich 's Holding Pens Ambient Temperature: _ _ _ _ Weather Conditions: ____________________ Drug Administration Time Drug Method Dose Location lvermectin Praziquantel Penicillin Event Times Start Procedures: Immobilized: Recovered: Finish Procedures: Vital Signs Time Temp Pulse Respiration Results of Physical Exam: Body Measurements: Samples Taken: Blood- EDTA Blood-RU Hi Collar Freq Markings D D D D D Transponder Number BodvLenath Tai{Lenath cm cm cm cm cm Ear Tag Location Number Color Left Right Comments: Colorado Division of WIidiife Revision Date: 08124/00 �58 Attachment 2 Protocol for post-mortem sampling of lynx Contacts: Dr. Margaret Wild (CDOW) Dr. Dan Gould (CSUDL) All information releases and contact with press need to be through the CDOW. Margaret Wild will coordinate. The objectives of the post-mortem examination are to: 1) determine the cause of death and document with evidence, 2) collect samples for a variety of research projects, 3) archive samples for future reference (research or forensic). The gross necropsy and histology will be performed by, or under the lead and direct supervision of, a board certified veterinary pathologist. Preferably, Drs. Margaret Wild and/or Tanya Shenk of the Colorado Division of Wildlife (CDOW) will also be present. In general, the protocol will follow standard procedures used for thorough post-mortem examination and sample collection for histopathology and diagnostic testing. Some other data/samples listed below will be routinely collected for research, forensics, and archiving. Additional data/samples may be collected based on the circumstances of the death (e.g., photographs, video, radiographs, bullet recovery, samples for toxicology or other diagnostic tests, etc.). Procedures: 1. 2. 3. 4. 5. 6. Remove the radio collar (animal ID will be the number written on the collar) Determine the body weight and note general body condition Note condition of the claws and teeth Collect external parasites and submit for ID If the carcass is in good condition, please skin and save the pelt (freeze) Remove and save PIT tags (about 1.5mm x 12mm rod-shaped transponder~ne under chin, the other on dorsal midline between the scapulae) 7. Conduct necropsy 8. Collect samples of all major organs including long bone, muscle, spinal cord, etc. (formalin) 9. Collect samples of all major organs including long bone, muscle, spinal cord, etc. (freeze) 10. Collect a 3cm x 3cm section of muscle (freeze) 11. Note condition of bone marrow in femur or other long bone 12. Collect a sample ofvas defcrens and testis (bag with a moist gauze and refrigerate) 13. Bag gut contents that could be used to determine diet (freeze) 14. Collect a fecal sample and submit for parasitology 15. Collect brain. Submit half for rabies exam; save other half in formalin 16. After examination, place the skull in one bag, the pelt in another bag, and other remains in a separate bag (freeze) The CDOW will retain all samples and carcass remains with the exception of: tissues in formalin for histopathology, brain for rabies exam, feces for parasitology, external parasites for ID, other diagnostic samples. �59 Table 1. Mortality of lynx reintroduced to Colorado, February-August 1999. Animal ID YKM0199 BC99-01 BC99-09 BC99-07 BC99-08 AKM2399 AKF0499 BC99-06 AKF1799 AKF0899 AKF1899 Sex M M F F F M F F F F F Age Protocol Date of Death Weeks Out Cause of Death 2 NIA NIA 0 1 04129199 02124199 2 1 02/26199 3 3 1 03116199 6 0 2 7 1 3 04110199 06118199 4 1 3 06112/99 s Undetennined Starvation Starvation Starvation Starvation Shot Starvation 2 3 07/19199 7 HBC 2 3 07124199 11 HBC s 3 07131199 12 2 3 08126199 15 Undetennined Trawna., emaciation 3 1600 - 1400 Q) 1200 >. 800 bl) -- ~ 1000 C =-@ 0 600 ~ 400 200 0 3110 3117 3/24 I-._ 3131 4/7 Males -a- Females 4114 4/21 * Pregnant 4128 SIS I Fig. 1. Average daily feed intake of captive lynx reported weekly from 10 March through 5 May 1999. �60 16 14 ~ 12 rn 10 ~ 8 ~ 6 ,:Q 4 I 0 2 0 YK BC I□ Arrival Wt Im Release Wt AK I Fig. 2a. Body weight of adult male lynx from British Columbia (BC; n = 5 arrival. n = 3 release). Yukon (YK; n = 7 arrival, n = 5 release), and Alaska (AK; n = 7). 12 ~ .l=bO ·-Q) ~ >"'0 0 ,:Q 10 8 6 4 2 0 BC YK AK I□ Arrival Wt II Release Wt I Fig. 2b. Body mass of juvenile female lynx from British Columbia (BC. n = 1), Yukon (YK, n = 1). and Alaska (AK, n = 3). �61 14 12 'bi> 10 c ell ~ ~ 0 ~ 8 6 4 2 0 BC-O YK-P YK-0 BC-P AK-0 AK-P Fig 2c. Body mass of adult female lynx from British Columbia (BC; open n =4 arrival, n = 1, release; pregnant n = 0), Yukon (YK; open n = 2 arrival, n = 1 release; pregnant n = 1), and Alaska (AK; open n = 3; pregnant n = 7). 6 5 1-4 4 Q.) 1 3 2 1 0 Shot Starve Trauma Cause of Death I111 !Vele ■ Female Fig. 3. Lynx mortality by cause through August 1999. J HBC �62 4 3 ~ .0 e 2 z 1 0 Feb 111 Starve Mar Apr ■ Unk May IJJ Shot Fig. 4. Lynx mortality by month through August 1999. Jun IIIHBC Jul Aug ■ Trauma � https://cpw.cvlcollections.org/files/original/68d69b2c7ab7e63303dc55591394cc63.pdf 2301eb2cbeb1cd2229a2ee93ca4c09ab PDF Text Text 63 Colorado Division of Wildlife Wildlife Research Report July 2000 JOB FINAL REPORT State of Colorado Cost Center 3430 Project No. W-153-R-13 Mammals Research Program Work Package No. ------"-0___ 67"'--0"--_ _ _ _ __ Task No. Lynx Conservation Assessing Abundance of Snowshoe Hares 3 Period Covered: July 1, 1999 -June 30, 2000 Authors: D. F. Reed Personnel: D. F Reed, G. Byrne, T. Shenk, J. Kindler, P. Albers, T. Black, H. McNally, S. Znamenacek, K. Buell, S. King, G. Patton, J. Zahratka, L. Uphham, B. Andree, J. Hicks, ·L. Green, K. Wright, B. Heicher, P. Jones, D. Homan, R. Adams, S. Wait, D. Kenvin, B. Woodward, D. Boyer, K. Cordove, B. Ochs, T. McLean, A. Bellotti, K. Schrollt, G. Cross, F. Kelffuer, G. Madison, L. Dwyer, D. Swanson, D. Brownne, F. Pusateri, M. Wunder. ABSTRACT Data on hare density for this report were derived from pellet surveys using methods describe by Krebs et al. 1987. Samples were distributed among 5 blocks selected by biologist who believed them to be suitable for sustaining lynx populations. Sample sizes of plot arrays ranged from 62-239 with the most plot arrays occurring in block 5 (n = 239). Density calculations yielded 0.44 bares per ha for block 5 and lower densities for the remaining 4 blocks. Hare pellet densities were highest in white fir habitats; followed by limber pine and Englemann spruce habitats. Pellet densities were lowest in Gambel's oak and ponderosa pine habitats, but these were also habitats which were not intensively sampled. The most relevant results in this report are estimates of snowshoe density, the type of overstory in which . the highest densities occurred (Engleman Spruce, based on those with adequate sample size) and where the highest densities occurred (National Forests/wilderness areas and counties in Block 5; Figs. 4-8, 9, 10, 12 in Appendix 1). These indicate marginal densities for supporting reintroduced lynx according to the literature and, professionals working with these species. Granted that most of the work and most of the .professionals working in the field are in the northern boreal forests where snowshoe populations cycle dramatically. Poole (1995) reported that concomitant with the first full winter of low bare density in the Northwest Territories (91-92), all resident lynx died or dispersed in a 135-km2 study area. Marginal habitat conditions for hares probably result in a scarcity of prey and may explain the relatively large home ranges (and hence their behavioral adaptations toward dispersal) of the lynx (Koehler 1990, Roloff 1997). Demographic characteristics of lynx, if any successfully establish home ranges, may be representative of lynx populations along the southern periphery of their range where habitat conditions are marginal for snowshoe hares. �64 �65 SNOWSHOE HARE DENSITY/DISTRIBUTION ESTIMATES AND POTENTIAL RELEASE SITES FOR REINTRODUCING LYNX IN COLORADO D.F. Reed P.N. OBJECTIVES Analyze, summarize, and organize snowshoe hare survey data into a comprehensive final report for release to the public (Division Report No. 20). SEGMENT OBJECTIVES l. Analyze, summarize, and organize snowshoe hare survey data into a comprehensive final report for release to the public (Division Report No. 20). RESULTS Results of the analysis of snowshoe hare pellet sampling are detailed in the attached Appendix 1. Division staff contracted with West, Inc. to evaluate and review the adequacy of both the methodology and the analysis of these data and that report is contained in Appendix 2 �66 �67 APPENDIX 1. SNOWSHOE HARE DENSITY/DISTRIBUTION ESTIMATES AND POTENTIAL RELEASE SITES FOR REINTRODUCING LYNX IN COLORADO Dale F. Reed FOREWORD Native wildlife are valuable resources in Colorado. Such values are manifested in several ways. Some people enjoy seeing or photographing non-hunted species or get satisfaction from merely knowing that they exist as important components of the ecosystem. During the 1980s and 1990s interest increased in the concept of ecosystem management and in maintaining or enhancing biodiversity. Furthermore, interest in enhancing the numbers of or in reintroducing species such as the lynx (Felis lynx) gained momentum in the late 1990s. To that end, a reintroduction oflynx was planned for the winter and spring of 1999 and a prerequisite to such an effort should involve examining the distribution and abundance of its principal prey, the snowshoe hare (Lepus americanus). TABLE OF CONTENTS INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 METHODS ................................................................... . Krebs Plots ................................................................... . Habitat Attributes at Plot Sites ..................................................... . Model Probability Surface ........................................................ . Potential Release Sites ........................................................... . RESULTS AND DISCUSSION ................................................... . Krebs Plots .................................................................... . Habitat Attributes at Plot Sites ..................................................... . Model Probability Surface ........................................................ . Potential Release Sites ........................................................... . CONCLUSIONS/SUMMARY .................................................... . ACKNOWLEDGMENTS ........................................................ . LITERATURE CITED .......................................................... . TABLES 1 Number of points and plot arrays ................................................. . 2 Wyoming and Colorado hare densities .............................................. . 3 Probability of suitable hare habitat ................................................ . FIGURES l GAP Vegetation Classification and Survey Blocks 1-5 .................................. . 2 Random Sample Points Generated for Krebs plots ..................................... . 3 Krebs Plot Points Completed .................................................... . 4 Map of Snowshoe Hare Density by National Forest .................................... . 5 Graph of Snowshoe Hare Density by National Forest .................................. . 6 Map of Snowshoe Hare Density by County .......................................... . 7 Graph of Snowshoe Hare Density by County ......................................... . 8 Map of Snowshoe Hare Density by Block ........................................... . �68 9 Graph of Snowshoe Hare Density by Block .......................................... . IO Snowshoe Hare Density by Block Using "Old" Equation ............................... . 11 Snowshoe Hare Density by Elevation ............................................. . 12 Snowshoe Hare Density by Primary Overstory ...................................... . 13 Snowshoe Hare Density by Forest Structure ........................................ . 14 Snowshoe Hare Density by Primary Understory .................. : ................... . 15 Snowshoe Hare Density by Understory Density ...................................... . 16 Snowshoe Hare Density by Slope ................................................ . 17 Snowshoe Hare Density by Aspect ............................................... . 18 Snowshoe Hare Density by Domestic Grazing ....................................... . 19 Model Probability Surface for Blocks 1-5 .......................................... . 20 Predicted Areas with Greater than 80% Probability of Suitable Hare Habitat in Block 5 ........ . APPENDICIES A Krebs Pellet Plot Field Form ..................................................... . B Selected Data from Block 1 ..................................................... . C Selected Data from Block 2 ..................................................... . D Selected Data from Block 3 ..................................................... . E Selected Data from Block 4 ..................................................... . F Selected Data from Block 5 ...................................................... . G Genus-species Understory Abbreviations in Fig. 14 ................................... . H Summary by Hare Densities per Block, Primary Overstory, National Forest/Wilderness Areas, and by County ...................................................................... . �69 SNOWSHOE HARE DENSITY/DISTRIBUTION ESTIMATES AND POTENTIAL RELEASE SITES FOR REINTRODUCING LYNX IN COLORADO INTRODUCTION Based on a renewed interest in the lynx and the possibility of reintroducing the animal in Colorado, four agencies including the U.S. Forest service, National Park Service, U.S. Fish and Wildlife Service, and Colorado Division of Wildlife prepared a draft strategy for the conservation and reestablishment of lynx (Fe/is lynx) in the southern Rocky Mountains (Seidel et al. 1998). This draft outlined generally two methods for assessing snowshoe hare (Lepus americanus) habitat use or presence and abundance or population trends as recommended by Weaver (1997). The first method was winter track surveys and the second was counts of snowshoe hare pellets (fecal droppings). Tracks in the snow have long been used to note habitat use and the presence of animals (Forrest 1988, Halfpenny 1986). However, only careful approaches or methods may yield definitive density or population estimates (Becker et al. 1998). A winter track survey was completed during the winter months of 1998 and reported by Byrne (1998). • Counts of snowshoe hare fecal pellets have been used to estimate population trends or density (Angerbjorn 1983, Krebs et al. 1987, Wolff 1982). Angerbjorn (1983) estimated density in Sweden from counts of square-shaped quadrats of 0.1 m2 and obtained log-log regression: log (pellets)= 1.02 log (hare density)+ 0.44, where hare density is per hectare (ha). He found a close relationship between pellet counts and hare numbers (r2 = 0.91). Conversely, Wolff (1982:141) indicated that "...pellet counts can be used to indicate habitat use or population trends, but not to estimate population numbers." Krebs et al. (1987) found a relationship between mean pellet counts and mean hare density on 50 quadrats of0.155 m2 (5.08 x 305 cm). A modified version of his equation, log10 (hares/ha) = ((log 10 (pellets) - 0.2359] x 2.303) + 0.2773, was used by Slough and Mowat (1996). Since then, the equation has been modified several times. The purpose of this document is to report on snowshoe hare pellet counts, the results of estimating hare densities across major habitat types and associated selected habitat characteristics, and an approach for estimating hare habitat as it relates to providing the primary prey for reintroduced lynx. METHODS Krebs Plots Five major core areas were delineated for the Kreb's pelletplot counts (Fig. 1). The blocks were selected based on large contiguous parcels of land that appeared to have suitable habitat (oak, aspen, coniferous forest type or a mixture of these types). The vegetation data is from the Colorado GAP Analysis project (Thompson et al. 1993). Natural or manmade barriers, such as Interstate 70, further define the block boundaries. Small isolated blocks of suitable habitat such as Greenhorn Mountain, the Roan Plateau, Sangre de Cristos, and the Uncomphagre Plateau were not considered in the hare data collection and the GIS maps. Random points for each of the five blocks were generated using a program modified after Zack (pers. com., J. Zack) which created a grid for each vegetation type with random numbers between 0 and 100,000. A sub-program checked the number of cells containing identical values, and aggregated these until a total was reached that was equal or greater than the required number of random points. The centroids of the cells with values lower than this cut-off value were then plotted as an Arc/Info coverage. �70 The number of random points generated across the vegetation types (Fig. 2) were about 200 for blocks l, 4, and 5, and 100 for blocks 2 and 3 (in some cases random point program generated a few extra points). Initially, every 4th random point ploted on map (BLM Surface Mangement Status 1:100,000 scale) overlays were selected to insure wide distribution and randomness in case all points could not be completed. According to the protocol used, once the location was determined by use of a GPS unit (matching UTM coordinates generated for a given point as listed on the overlay) or map orienteering, a random bearing (0360 deg) was chosen, and a rectangular array of 10 Krebs plots was set (A), each 30 m apart (rectangular array= 30 m x 120 m, with 5 plots per side in direction and then in opposite direction of the bearing). Additionally, a second array (B) was selected at a randomly chosen bearing and distance (10-190 m inclusive) from the completion point (about 30 m from the beginning of the first array) of the first array (A). Hence, pellet counts were made for two fairly close areas near each randomly selected point. These pellets were tallied as either new or old and their means calculated. New and old pellets were differentiated by color, lighter tan being judged "new" (estimated from current season), and darker gray being judged "old" (estimated from before current season). The mean number of pellets (2 sets of 10, i.e. n = 20 plots for each randomly selected point [Fig. 31) was entered into a published equation to produce an estimated number of hares per hectare. This equation was a modified version of Krebs et al. (1987) (log 10 hares/ha= {[(0.8127 x log 10 pellets) - 0.2359] x 2.303} + 0.2773, as in Slough and Mowat 1996). Later, the equation was modified (per. com., K. Poole, 9 Nov 98). Still later, it was modified again (per. com., C. Krebs, 29 Sep 98 and 24 Nov 98). This third iteration, lo& (hares/ha)= (0.888962 x 10& [pellets] - 1.203391) x 1.567026, was recommended by Charles Krebs (per. com., C. Krebs, 24 Nov 98) and the one ultimately used for the overall analyses. It was based on his review of his data (collected samples from the Yukon: 1976-1996, n = 86, calculated from annual pellet counts in June and average densities of hares from the previous spring and autumn estimated by markrecapture including adding a correction for bias [Sprugel 1983]). Habitat Attributes at Plot Sites Categories other than those related to the plot locations (Name of Property, Game Management Unit [GMU], County, Lynx Survey Block Number, General Location Description, Land Status, and UTM coordinates) recorded on the field forms (Appendix A) included: - elevation - overstory (primary, secondary, and tertiary) - vegetative structure (e.g. mature, old growth) - understory vegetation (primary, secondary, and tertiary) - understory vegetation density (estimated visual obstruction of vegetation from IO m in front of a vertical I x 6 ft [0.30 x 1.83 m] white panel, marked off in ft2 [0.30 m2], where highest value of 6.0 = no visual obstruction, and to be recorded with photograph) - slope - aspect - overstory and understory distribution (patchy or continuous ) - domestic grazing presence or absence. Model Probability Surface A logistic model was used to generate a continuous surface representing the probability of any geographic location supporting snowshoe hare habitat (pers. com., M. Wunder). The results from the Krebs plot effort were examined spatially using chi-squared analysis to produce parameter estimates for the model. �71 Parameter values were estimated for each of vegetation (primary cover type), elevation, slope, and aspect using the Krebs plot coverage, the vegetation coverage from the GAP analysis project, and a digital elevation model (DEM) for the study area. The DEM was used to generate separate grids for elevation, slope and aspect. Parameter estimates with P-values less than 0.05 were then used in the logistic model, which was then run using the pertinent grids. The model equation follows: 1/l+exp(Ao + Aa *aspect+ Ae *elevation+ Av* vegetation), where Ao is the overall parameter estimate, Aa is the estimate value for aspect, Ae is that for elevation, and Av is that for vegetation. Release Sites Initially, release sites were selected based on the model probability surface, but for practical reasons of access and managing the media and number of observers during the releases, private land areas abutting potential lynx habitat or proximity to the Weminuche Wilderness were given high priority. Potential sites were inspected and coordinated with landowners and caretakers where appropriate. RESULTS AND DISCUSSION Krebs Plots Paired pellet plot arrays and single plot arrays were completed at 303 and 5 randomly selected points, respectively. Sample sizes of the number of plot arrays ranged from 62-239 (Table 1, Appendices B-F) with the most plot arrays (n = 239) completed in block 5 (Appendix F). Distribution of hare densities by National Forest (Figs. 4-5), by county (Figs. 6-7), and by block (Fig. 8), suggest that block 5 had the highest hare density. Density calculations yielded 0.44 hares per ha for block 5 and lower densities for the remaining blocks (Fig. 9). This mean of hares per ha in block 5 results from using the latest regression equation from Krebs as discussed under Krebs Plots in Methods. The initial calculation using an.earlier equation (Slough and Mowat [ 1996]) yielded a mean of almost 1.2 hares per ha (Fig. 10). Using different equations and possibly using them incorrectly yields substantially different results. Also, initial analysis of block 5 data using a SAS program used in the Northwest Territories (per. com., K. Poole, 14 Sep 98) was corrected as were the analyses of data from 2 areas over 2 years in Wyoming, where lynx were present (per. com., K. Poole, 9 Sep 98; per. com. T. Laurion, 18 May 99) (Table 2). Our calculations using the latest regression equation from Krebs (per. com., C. Krebs, 27 Oct 98 and 24 Nov 98), matched these same values (0.817, 0.856, 0.694, and 1.429, respectively) (Table 2). Thus, using the most current accepted calculations, the Wyoming areas having lynx present, had hare densities ranging from 0.817-1.429, similar to some other densities reported (0.73/ ha, Dolbeer and Clark 1975; 1.40/ha, Lima 1998; 2.40/ha, Meslow and Keith 1968), but subtantially lower than others (12.3/ha, Bailey et al. 1986), and all higher than that considered the lower "edge" for suppporting lynx (0.5/ha; C. Krebs, per. comm., 24 Nov 98). Dolbeer and Clark's (1975) 0.73 hares per ha might be considered a model for habitat in Colorado, but another study using 175 permanent plots, reports lower densities (0.053/ha in lodgepole pine and 0.068/ha in spruce/fir) (Shivery 1997). This causes concern that the density of our 0.44 hares per ha in block 5 may not be sufficient as a primary prey base to support lynx. Part of the problem, however, is that our data were collected randomly across 11 primary overstory types. The Wyoming data were collected from plots ~lected in coniferous forest types. �72 It is generally accepted that some of the primary overstory types (probably the 3 non-conifer types [aspen {Populus tremuloides}, willow {Salix spp}, and Gamble's oak {Quercus gambelii} or mixed mountain shrub], ponderosa pine (Pinus ponderosa), and bristlecone pine (Pinus aristata) might be expected to have low hare densities. Apparently this was the case based on our small sample sizes. Most of our samples were taken in 6 conifer types described below. Hence, the problem remains - we apparently have a relatively low density of hares if compared to the Wyoming data, but is it comparable? Given that Wyoming's plots were taken non-randomly in optimal habitats, the extent to which optimal habitat considerations could induce bias is unknown. It is possible that if we had used our method of selecting random samples in Wyoming that their means would have been lower. Similarly, it is possible that ifwe had sampled randomly in "optimal habitats" (whatever those are in Colorado) that our means would have been higher. In either case, how much lower, or higher, is simply unknown. This probably limits the utility of comparing these 2 sets of data. Habitat Attributes at Plot Sites Attributes from the plot locations as prescribed by data collection (Appendix A) across densities, included elevation (Fig. 11 ), 11 primary overstory or canopy types, - Lodgepole pine (Pinus contorta) - Englemann spruce (Picea engelmannli) - Subalpine fir (Abies lasiocarpa) - Douglas fir (Pseudotsuga menziesli) - Ponderosa pine - Limber pine (Pinus jlexi/is) - Bristlecone pine - White fir (Abies concolor) -Aspen -Willow - Gamble's oak or mixed mountain shrub (Fig. 12), structure (Fig. 13), primary understory vegetation (Fig. 14), vegetation density indices (Fig. 15), slope (Fig. 16), aspect (Fig. 17), and grazing (Fig. 18). The relationships between hare density and some of these attributes, i.e. elevation, structure, primary understory, aspect, and grazing, yielded relatively predictable results. Forest types preferred by hares in Colorado occur at higher elevations (Fig. 11). Limber pine and white fir show the highest hare densities, but with very small sample sizes these densities are probably not represen~ve (Fig. 12). Conversely, Englemann spruce had a sample of 220, lodgepole pine 108, subalpine fir 64, and Douglas fir 48 (Fig. 12). Willow had only one sample where it was considered the primary overstory and is not included in Fig. 12. For structure, most of the plots completed were in mature or old growth. The small sample of sapling/pole (n = 49 of 606 records, with only about 50% of the 49 having any pellets) and the even more negligible sample sizes ofGrass/Forb and Shrub/Seedling, essentially yield zero densities (Fig. 13). Selected understory species appear to be associated with higher hare densities than others (Fig. 14), but another measurement, the vegetation density indices, may suggest that the amount and height of the understory is limited (Fig. 15 and 376 photographs appended to data records). Most of the samples (245 and 97 of 428 records) occurred in the 5.0-6.0 and 4.0-4.9 categories, respectively, where little understory obscured the panel (Fig. 15). The value for <1.0 is unlikely to be representative. Other workers have suggested that predation on snowshoe hares may be heaviest in habitats that have little understory (Sievert and Keith 1985), that snowshoe hares avoided open understories and greater understory �73 density provided both escape and thermal cover (Litvaitis et al. 1985), and that height ofunderstory was important (Wolfe et al. [1982] found a strong correlation between hare use and cover densities at heights of 1. 0-2.5 m above ground, and that vegetation types in which cover densities above snow level [ 1. 0-1.5 m] were at least 40%, accounted for 85% of winter use by hares). Hare density shows a bimodal response to slope (Fig. 16), but upon inspection of photographs it appears that at least one field person consistently over-estimated degree of slope, probably skewing these data. In relation to aspect, hares probably avoided the wanner, dryer, southern exposures (Fig. 17). Hare density means from plots judged to have domestic grazing or not (Fig. 18) were predictably no different (0.358 vs 0.342, respectively). How recent or the extent of any grazing was not estimated. Judging vegetation distribution for overstory and understory as patchy or continuous (Appendix A) was largely dependent on "scale" and was therefore not analyzed. Similarly, "Old" pellets were not analyzed because their age or persistence could not be determined. Generally, these pellets occurred in low numbers and were noted in 402 of 613 records. Cottontail (Sylvilagus nuttalli) and jackrabbit (Lepus townsendi) pellets were noted in a low number of records and were not analyzed. Some overlap or abutting of habitat boundaries might be expected between hares and mountain cottontails, but doubtfully with jackrabbits. One could conclude, that if you counted jackrabbit pellets, you were in the "wrong" (i.e., non-hare) habitat. Model Probability Surface The probability values from the model were naturally distributed across fiv~ range classes (pers. com., M. Wunder). These were represented as zero percent probability, one to twenty percent, twenty to eighty, eighty to ninety-three and ninety-three to one hundred percent probability (Fig. 19). The most relevant results include the areas that were modeled as greater than 80 percent probability of hosting snowshoe hare habitat. Block 5 included the largest surface area and proportion of area that was modeled greater than 80 percent probability. Areas within a radius of 3.2 km from potential lynx release sites that were predicted with greater than 80% probability to provide suitable snowshoe hare habitat were delineated (Fig. 20). The distribution and amount of the area is shown in Table 3. It should be noted, however, that raw values for area may not be the best (and certainly is not the only) criterion for drawing conclusions about the suitability of an area for snowshoe hare or lynx. We did not develop any method for examining the distribution of this area for spatial context. If and when this issue is addressed in the future, it should be acknowledged that there are many varied factors involved with such an analysis. Release Sites Hence, one of the sites north of the Weminuche Wilderness, Humphreys (Goose Creek), was ultimately chosen for the initial release February 3-4, 1999. Similarly, other sites northwest, north, south, and southwest of the Weminuche Wilderness Area, were ultimately chosen for later releases in March. Release sites in Block 5 located northwest, north, and east (n = 4) and south and southwest (n = 4) of the Weminuche Wilderness were initially described using a predicted probability surface of snowshoe hare habitat (per. comm., M. Wunder) and examined using the following criteria: - ownership, public and private lands where permission for egress was obtained - access, where at least snowmobiles could be used to gain access - seclusion, where public access is generally not available - habitat near release site should have a range of 1,000-10,000 ha of continuous or at least only lightly broken conifer forests �74 - connectivity, where forest segments are sufficiently contiguous to allow for traveling lynx to avoid open areas The 4 sites northwest, north, and east of the Weminuche Wilderness Area were ranked as follows: -Humphreys (Goose Creek) - Red Mountain Ck - Rio Grand Reservoir, and - Big Meadows Similarly, the 4 sites located south and southwest of the Weminuche Wilderness Area were ranked as follows: - Weminuche Valley - Beaver Meadows - Endlich Mesa, and - Middle Mtn/Bear Ck CONCLUSIONS/SUMMARY The most relevant results in this report are the estimates of snowshoe hare density, the type of overstory in which the highest densities occurred (Engleman Spruce, based on those with adequate sample size) and where the highest densities occurred (National Forests/wilderness areas and counties in Block 5; Figs. 4-8, 9, 10, 12, and Appendix H). These indicate marginal densities for supporting reintroduced lynx according to the literature and professionals working with these species. Granted that most of the work and most of the professionals working in this field are in the northern boreal forests where snowshoe hare populations cycle dramatically. Poole (1995) reported that concomitant with the first full winter of low hare density in the North West Territories (91-92), all resident lynx died or dispersed in a 135-km2 study area. Marginal habitat conditions for hares probably result in a scarcity of prey and may explain the relatively large home ranges (and hence their behavioral adaptations toward dispersal) of the lynx (Koehler 1990, Roloff 1997). Demographic characteristics of lynx, if any successfully establish home ranges, may be representative of lynx populations along the southern periphery of their range where habitat conditions are mariginal for snowshoe hares. ACKNOWLEDGMENTS We especially thank Pam Albers, Travis Black, Heath McNally, and Steven Znamenacek for the completion of most of the Krebs plots. Other agency field assistance came from Steve King, Rocky Mountain National Park; Gary Patton, U.S. Fish and Wildlife; Kit Buell and Jennifer Zahratka, U.S. Forest Service; and Lee Upham and crews, Bureau of Land Management. Field work by Division personnel included Bill Andree, Jim Hicks, Larry Green, Kevin Wright, Bill Heicher, Paul Jones, Doug Homan, Rick Adams, Scot Wait, Dave Kenvin, Dale Reed, Gene Byrne, and Brent Woodward. Gene Byrne organized and coordinated the field work. Work by volunteers included Don Boyer, K. Cordova, Brett Ochs, Travis McLean, Amy Bellotti, Kathrin Schrott, Gretchen Cross, Francis Kleffner, Griffin Madison, Lynn Dwyer, and Dawson Swanson. Geographic Information Systems (GIS) support and cartography provided by Jon Kindler. Dawn Brownne assisted with data entry. Francie Pusateri provided advice and converted Visual dBase files to EXCEL. Mike Wunder, Natural Heritage Program, generated the model probability surface and associated figures. We also thank Tom Beck, Rich Reading, and Tanya Shenk for reviewing the manuscript. Work reported here was funded mostly by Division GoCo funds and from funds from Vail Associates. �75 LITERATURE CITED Angerbjorn, A. 1983. Reliability of pellet counts as density estimates of mountain hares. Finn. Game Res. 41:13-20. Bailey, T. N., E. E. Bangs, M. F. Portner, J. C. Malloy, and R. J. McAvinchey. 1986. An apparent overexploitated lynx population on the Kenai Peninsula, Alaska. J. Wildl. Manage. 50:279-290. Becker, E. F., M.A. Spindler, and T. 0. Osborne. 1998. A population estimator based on network sampling of tracks in the snow. J. Wildl. Manage. 62:968-977. Byrne, G. 1998. A Colorado winter track survey for snowshoe hares and other species. Colo. Div. Wildl. 35pp. Dolbeer, R. A., and W.R. Clark. 1975. Population ecology of snowshoe hares in the central Rocky Mountains. J. Wildl. Manage. 39:535-549. Forrest, L. R. 1988. Field guide to tracking animals in the snow. Stackpole Books, Harrisburg, PA. 185 pp. • Halfpenny, J. 1986. A field guide to mammal tracking in North America. Johnson Publishing Co., Boulder, CO. 161 pp. Koehler, G. M. 1990. Population and habitat characteristics oflynx and snowshoe hare in north central Washington. Can. J. Zool. 68:845-851. Krebs, C. J., B. S. Gilbert, S. Boutin, and R. Boonstra. 1987. Estimation of snowshoe hare density from turd transects. Can. J. Zool. 65:565-567. Lima, S. L. 1998. Nonlethal effects in the ecology of predatorprey interactions. BioSci. 48:25-34. Litvaitis, J. A., J. A. Sherburne, and J. A. Bissonnette. 1985. Influence of understory characteristics on snowshoe hare habitat use and density. J. Wildl. Manage. 49:866-873. Meslow, E. C., and L.B. Keith. 1968. Demographic parameters ofa snowshoe hare population. J. Wildl. Manage. 32(4):812-834. Poole, K. G. 1995. Spatial organization of a lynx population. Can. J. Zool. 73:632-641. Roloff, G. J., and J.B. Haufler. 1997. Establishing population viability planning objectives based on habitat potentials. Wildl. Soc. Bull. 25:895-904. Seidel, J., B. Andree, S. Berlinger, K. Buell, G. Byrne, B. Gill, D. Kenvin, and D. Reed. 1998. Draft strategy for the conservation and reestablishment of lynx and wolverine in the southern Rocky Mountains. U.S. Forest service, National Park Service, U.S. Fish and Wildlife Service, and Colorado Division of Wildlife. 115pp. Shively, K. 1997. Snowshoe hare density in Vail, Colorado; implications for lynx reintroduction. Unpub. Rep.:1-7. Sievert, P.R., and L.B. Keith. 1985. Survival of snowshoe hares at a geographic range boundary. J. Wildl. Manage. 49:854-866. Slough, B. G., and G. Mowat. 1996. Lynx population dynamics in an untrapped refugium. J. Wildl. Manage. 60:946-961. Sprugel, D. G. 1983. Correcting for bias in log-transformed allometric equations. Ecol. 64:209-210. Thompson, T. G., W. A. Reiners, and D. L. Schrupp. 1993. Colorado gap analysis vegetation classification (draft). Unpub. Rpt. Weaver, J. L. 1997. Reconnaissance oflynx habitat in Colorado. Wildl. Conserv. Soc. lOpp. Wolff, J. 0. 1982. Snowshoe hare. Pages 140-141 in CRC handbook of census methods for terrestrial vertebrates. Edited by D E Davis. CRC Press, Boca Raton, FL. Wolff, J. 0., N. V. Debyln, C. S. Winchell, and T. R. McCabe. 1982. Snowshoe hare cover relationships in northern Utah. J. Wildl. Manage. 46:662-670. �76 Table 1. Block numbers, number of points generated, points completed (2 plot arrays for each point with 5 exceptions where only 1 array was completed), and plot arrays completed ( 10 plots for each array). Block No. of Points No. of Points No. of Plot Arrays Generated Completed (2 x points completed) 1 2 3 4 5 Total 200 100 100 200 200 800 57 31 33.5 64.5 119.5 305.5 114 62 67 129 239 611 Table 2. Mean number of hare pellets per Krebs plot and calculated hare densities per ha in 2 areas over 2 years in Wyoming where l)'TIX occurred and in Block 5 of this study. Wyoming8 Colorado Area DUBOIS97 DUBOIS98 MERNA97 MERN98 BLOCK5 Pellets/plot Hares/ha 1.86 0.82 (0.64-1.05)b 1.96 0.86 (0.67-1.09) 1.55 0.69 (0.54-0.89) 3.49 1.43 (1.11-1.84) 0.94 0.44 (0.35-0.56) •Data from T. Laurion, calculations by K. Poole. b95% CI. Table 3. Study areas or blocks, areas with greater than 80 percent probability of providing suitable hare habitat, and the proportion of the study areas involved. Study Area/Block Area (ha) with > 80% Probability of Proportion (%) of Study Area Providing Suitable Hare Habitat 1 2 3 4 5 Total 460,753 97,920 110,233 587,903 658,258 1 915 067 20.3 11.7 15.7 21.4 24.1 93.2 �1 I~ /L~-1 Dociduou•Oak ■ Aapon II ~ 1111) l'l!illllZr,I s--·tc· r•-•·IC lllllliilfl Douglaa Fir Lodgepole Pino - 0 0 .. ■ ■ ~<.:onifor (me. Blue Sptueo) u ~ 100 40 IO r.a.. I -, ----.J---- I I ,- - - - - - I I I :------' I I - - - -1- - - - - I I ~J I I_ - - ... - - - ....:i ....:i �--..:i 00 F,~. 2- ......*.,,.,,,, Study Area BoundAly Sample Point , / ' ,, , County Boundaiy ~ Intemato Higb.Wlly 100 I I :- - - - - - - - - - - - - - - - - - - --------- I - - - .,. __ ..,,,.. I 'r , ... ···········-··········------·· , I - - - - - 1- - - - - _,_ - - - - - - - - - - - �............................ SNOWSHOE HARE DENSITY SA * Sample Point ##•,...,#>' , f;·I~. !J , /\ '/ Study Alea Boundaiy County Boundaiy /V Intemate Highway I * ,,,, ] "1c I, ** -,,_, * * I I I I I I ,_ - - - - ,.•• I ista : I I I I -----------~ __ ,_J I - - - - - - - - - -1- - - - - - ' ✓ I I I 'rI ,-- ' ·---·····----------·--······· 80 ,- - - - - - - - I I *• 40 Mio, - - - - - - - - - - - - - -I 'L, * ,, ' 0 100 I I I I ,,_J-,lc--- *' ,.,.;,., u ...... ,0 0 --,------ I /~, '...,I I - - I - - - - - 1- - - - - ~- - - - - - - - - - - - I �00 0 SNOWSHOE HARE DENSITY BY NATIONAL FOREST D 0 Hares/ha - ----------- - 0.3 • 0.399 Harea/ha ~ 0.001 • .099 Ham/ha - 0.4 - 0.499 Harea/ha I ll 0.1 • 0.199 Harea/ha - >0.5 Harea/ha 0.2 • 0.299 Harea/ha 0 0 -- ,- .. - - - 40 Ma.. +---------,~ I I ~ I I I I I I I I I -- I----------1 ~------------------ I I :______, I I ----------1------ I --7-----I I I ~-~----------T----------- I , ~, I I I I I 5 • ·---·-··········--···-------- I I I I '.J I I_ - - - - - - - I I - - - I - - - - - 1- - - - - _,_ - - - - - - - - - - I �:c A) a (II ;,. "CJ CD "'I ':r A) 0 0 ...... 0 ~ ~ (,) 0 0 :i:,.. en 0 a> 0 ...... 9 00 9 <O ARAPAHO NF BUFFALO PEAKS WILDERNESS GRAND MESA NF GUNNISON NF LAGARITA WILDERNESS AREA 6' ~ RIOGRANDE NF (II ~ ROCKY MOUNTAIN ii NATIONAL PARK CD i Al CQ CD :::c m a C ::s en CD ~ tr '< "'I ROOSEVELT NF (II )> ii? A) r ROUTT NF "Tl 0 i SAN !SABEL NF SANJUAN NF UNCOMPAHGRE NF WEST ELK WILDERNESS WHITE RIVER NF l.r1 18 �00 N SNOWSHOE HARB DENSITY BY COUNTY D OHatell'ha 0.3 - 0.399 Hare,/ha G 0.001 - .099 HIU'ea/ha Ill 0A - OA99 Hare,/ha flil 0.1 - 0.199 Hatell'ha - >0.$ Hare,/ha 1111 0.2 - 0.299 Hares/ha 0- 0 ~ 0 40 Mi,e I 1---~ I I I I I I ----- .. I I I I I :- - - - - - - - - - - - - - - I I I I , _ _ _ _ _ _1 I I I I I I I - - - - - - - - - -1- - - - - - --,------ I I I I /- -,- - - • • - • - - - - T - - - - - - - - • I I Pueblo , ;,-. ' f f f I I I ,. I.,,. )- I I '•I ' I I r' ' 1('" , ... I I ., __ ... ,,. ,- '.J I I_ - - - - - - - -- I -----1----1 '------------ �0 0 0 0 C> N :i,.. <J> 0 0 0 0 0 0 0 0 0 0 0 0 0 co 0 0 0 C> 0 0 0 iv 0 0 0 :I: ... II) (I) VJ 1:1 ... (I) ::r II) :c $1) ., CD CJ CD :::l LA 0 0 C :J ~ en ~ O" '< C) 0 C: :::l ~ ---0 £8 �---- SNOWSHOE HARB DENSITY (HARES/HA) 0<0.1 0.10. Ucmct,n 50 0 0 1---~ DENVER -- -,- - - --- - -- -- ' I I I I I I I I I -- I---------- 1 I I I -----------------I : ______1 ' I - - - - - - - - -' I I I I I I I I I I ~ontro - - - - - - - - . -1- - - - - - ..... ~:. \ I I I 'r--------------' \ , --,------ J- -,- - - - - " - " - - - T - - - - - - - - - • • • I \ Pueblo 'I ' . . \,-,1 \ r ...... ' . ' \ ' ... - , - -,---------~ I I '.J I , ,_ - - - - - - - -------- - - - - - -- - - - - - - - - - - - 1· ~,. / ... ,' \~ \ ) I 'r' ·:!'-~~: 5 Durango !'i; ············-··········--···· ., ~' - ... ,. I 00 -"" �85 0.60 0 50 0.40 C1l .s::. ... Q) Q. VJ ... 0.30 Q) C1l :I: 0.20 0.10 0.00 2 Fig. 9. bars). 4 5 Mean hare density (w/ SD) by block (samp~e sizes same as in Fig. 10). 2 Fig. 10. 3 3 4 Mean hare density by block using Kreb's oriqinal equation (sample sizes above 5 �86 3 25 ~ 2 "' ,E e ., 1/) ~ :!' 1.5 ·;;; ., C Cl e r"' 0.5 0 <7,000 FT. 8,000 • 8,999 7,000 • 7,999 9,000 • 9,999 10,000 • 10,999 11,000-11,999 >=12,000 El~vation Ranges Hares per ha z w Q'. Q'. ~ a: ii: a: 0 z :5 C) ii: w w Q. a:__. 0 0 ::::> 0 i1J __. m 0 < C) 0 w Cl) < z w 0 w __. ICl) ii: m Cl) 0 z z w __. < m w Cl) 0 ::::> Q. C) 0 __. < Cl) ::::E w z a: < Cl) Q'. w z Zw <o ::::E ::> Woc __. Q. ~ Cl) w z 0 Q. Fig.12. Hare density by primary overstory (sample sizes above bars). Q'. ii: w I- I ~ w z a: Q'. w m ::::E ::::; �87 Hare Density by Forest Structure 16 ,---,-:-=----,----.---,---, ~'. ,[ 1.4 1.2 ... ~ 1/) ., :. :S :i!- 0.8 'iii ., C C ; 0.6 ::c 0.4 0.2 SHRUB/SEEDLING GRASS/FORS SAPLING/POLE MATURE STAND OLD GROWTH ~orest Structure 1.2 08 "' a. 0 6 .. 1/) ~ ::c 04 0.2 AMAL Fig. 14 ARCO ARUV CA spp JU spp QUGA SHCA SYAL SYOC Hare density by selected primary understories (number of records n = 10 or>, and pellets n = 1 or>; abbr. see Appendix G). VACA VASC �88 OS Hares per ha 0.3 0.1 0 5.0-6.0 4.0-4.9 3.0-3.9 2.0-2.9 < 1.0 1.0-1.9 Fig. 15. Vegetation density indices by hare deniity. .; 2 ~., :. ;;_ ~ 1.5 'iii ., C: 0 ~ "' :c 0.5 0 < 10 10 - 19 20-29 30-39 40- 49 Slope Range (degrees) 50 -59 60-69 >= 70 �89 Hare Density by Aspect (Area 5) NORTH NORTHEAST EAST SOUTHEAST • SOUTH SOUTHWEST Aspect 0.4 0.35 0.3 0.25 0.2 0.15 0.1 a.as 0 Fig. 1a. Absence or presence of grazing (1st and 2nd bar, respectively) versus hare density. WEST NORTHWEST �rig. 1 CJ. Study area-wide view showing predicted probability surface of snowshoe hare nabitat, study area boundaries, roads, and municipality boundaries. Variable Ao Aa As Ae Av . . - - - - . , r - - : - - - - - - · ·····-· - ............... __ \0 0 Parameter p 0.0001 15.6638 0.2300 0.0503 0.8573 0.0641 0.0001 -0.7106 -1.4909 0.0130 i--.----_J ---~,~ .. I Model Surface = 1/1 +exp(Ao+Aa* Aspect+Ae"Elevation+Av*Vegetation) ,.J ' .. •\_,/ Highways Study Areas Predicted Probability -0 -1-20% 1 ·'?~~I 20-80% .--180-93% · - 93-100% ' , .. :<:. 'p-,1 Snowshoe hare data used in this map were collected by the CDOW, USFS, NPS, USFWS, biologists and numerous volunteers. \ J-{ait,y,.:'§, ..~ ~ The data were compiled and provided by CDOW staff The predictive model was constructed by CNHP staff ' ' '~\ \ e Co; Oll ;;,i'P ····-·••-···--- ---···----·-······--' �Fig. 20. Areas within a radius of 3.2 km from potential lynx release sites that are predicted with greater than 80% probability to provide suitable snowshoe hare habitat. • Potential release sites 80-100%Probability within two mile radius Wilderness and Research Natural Areas Predicted Probability of Suitable Habitat D . -0 -1-20% 1 :_._;· 120-80% ~80-93% - - 93-100% I I. ! Hectares Name Endlich Mesa 14 63 Weminuche Valley 596 Be aver Meadows 469 Big Meadows 2108 ----------------Rio Grande____ Reservo1___..__________ 1421 _ i Humphreys _ _ _ ____..__·"_--•________§_~_J ------.----- Mosca Divide Middle Mtn/Tuckervi Coal Bank/Lime _____,_____ Middle Mt~/Bear Cr 977 : Wi 11 i a rn s Cr TH 921 : West Fork San Juan 912 ----~---- ---------·--·--- i I ----------·--·-··---- Red Mountain Creek ----'------···· Transfer Pk. ---------Four mile TH 817. ---·- ----~ 794 656 ---------------------Middle Fork Piedra 610 ---------East Fork Piedra 431 ---------------Pl um ta w 120 ••••• . ; -····--··-·- - - - - - - - - - - - - - 0 40 Kilometers s . -·· - - · - - · - · • - · · - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - �92 Colo.--ado Di VI•sion of Wldlife - Krebs Pellet Plot Survey Form Krebs Plot# · - Observer(s): Slope: (in Degrees) lime: Date: Name of Property GMU: Lynx Suivey Block# (Adual) County: Start Location from? D General Location Description: D GPS Map D D Aspect 0 UTMY: UTMX: Zone: D D 12 13 App QH..J,.x A N D s □ E D □ Patchy □ Continuous Understory □ Patchy D Continuous KREBS PLOT DATA: s,_,, ____ New Old New Old J~bbit New Old -1 - -- l - - - 2 - - - - -j - -- 3 - - -! - -- _j - -- - -- I 4 - - D - - - • GPS □ 5 - - - 6 - - - 7 - - - Secondary OVerstory: 8 - - Tertiaty OVerstcxy: 9 - Vegetative Stn.icture Class: 10 _ feet other Prima,y Overatory: GraS$IFOrb L Shrublseedrmg Sapling/Pole Mature OldGl"OINth L Underatory Vegetative Type: Common Name 1 (Primary) 2 (8econdary) 3 (Tertiary) Grazed D KrebsFonn Yes □ 7-8-98 No i - -- - ~1 - -- - - - - -- - - - - - -- - - - - - I - , ,' OVerstory Type: other - 1 D Cott<intail Elevation (MSL): From: Topo - NE WN SE SW overstay J ·NAD27 D V\GS84 [J NAD83 i Rat 0--5 Moderate 6-20 Moderately steep 21-40 Steep 41 degrees or more Vegetation Distribution: land status COde: t D D D D w C D Datum: Density Index Photo Taken?· D Yes □ No -.. r- L ,. Land Status Codes: ,_ G ~e Private - P U.S. F. S. -F BLM-8 ~ State,LaAd Board - S OON-D National Park - N other- O (Explain) Overstory Veg. Type Codes: Lodgepole Pine - L mann S(:)(1!ce - S ~ Su pine Fir - F • lasFir-D ~ P erosa Pine - P Umber Pine - X Bristteoone Pine - B WliteFir-W Asoen-A WIICMt-Y Gamble's Oak or mixed mountain Shrub - C Comments (Continue on back if needed): _ _ _ _ _ _ _ _ __ �I KREBSPL LSBNUM UTMX FC-21-A FC-21-B FC-95-A FC-95-B FC-97-A FC-97-B FC-101-A 1 1 1 1 1 FC-101-B FC-178-A FC-178-B 1 1 1 1 1 1 1 1 1 1 1 1 VA-62-A VA-62-B VA-132-A VA-132-B VA-144-A VA-144-B VA-149-A VA-149-B VA-150-A VA-150-B VA-155-A VA-155-B VA-161-A VA-161-B VA-163-A VA-163-B EP-40-A EP-40-B VA-164-A VA-164-B VA-135-B VA-153-A VA-153-B VA-135-A WA-77-B WA-81-A WA-81-B WA-73-A WA-73-B WA-75-A WA-75-B VA-166-A VA-166-B DW-13-A DW-13-B DW-49-A 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 446555 446510 448682 448660 428211 428210 425389 UTMY 453580 397240 397326 392830 373296 373376 392913 403816 SN9 MEAN SN10 1 1 1 I 4489480 4505757 L 4505920 L 4373370 S 4372950 S 4424300 A 394760 4424351 L 371270 4410075 F 371161 4410046 F 406232 4405739 L 5 4 2 6 1 3 1 (, .. ' 1 4405872 S 4405295 L 4405428 L 4394020 L App~dix (J 2 425240 450931 450870 388866 389100 394594 406200 393331 393518 406653 406432 405438 405303 403862 403935 453565 SNS SN3 SN2 PRIMOVE SN1 4488703 F 4488640 S 4492238 S 4492180 S 4490776 S 4490840 S 4489448 0 1 1 1 1 1 1 1 43.9.3960 A 43!!6943 S 4386987 S 4384857 F 4364789 S 4442252 F 4442167 L 4381807 F 4381750 F 4419021 F 4399796 F 4399898 L 4419011 L 4528065 L 1 4538167 L 4531982 L 4532088 A 4376295 L 4376415 L 4374855 A 1 1 1 449594 4374855 A 434487 4398253 S 1 1 5 3 •1 2 407484 4521979 S 407495 4522070 S 406978 4538343 L 407057 405250 405123 406867 407088 449594 4 1 21 2 1 1 7 ·2 4 1 1 1 2 1 •1 1 1 1 1 1 1 2 1 1 1 3 1 3 2 1 4 1 2 10 1 1 3 4 1 16 4 4 1 . Page 1 of3 2 KREB'SN HARE DEN!- ' 0.4 -0.55929 0.2 -0.80394 0.1 -1.0486 0.1 • -1.0486 0 0 0.3 -0.66083 0 0 0 0 0.9 -0.27306 1.2 -0.17152 0 0 0 0 0.1 -1.0486 0 0 0.1 -1.0486 0 0 0 0 0.1 -1.0486 0.2 -0.80394 0.1 -1.0486 0.1 -1.0486 0.2 -0.80394 0 0 0 0 0 0 0 0 0.6 -0.41617 0.1 -1.0486 0.5 -0.48052 0 0 0.6 -0.41617 0.9 -0.27306 2.6 0.101391 1.2 -0.17152 0 0 0.1 -1.0486 0.1 -1.0486 0.1 -1.0486 0.1 -1.0486 0.7 -0.36176 0.3 -0.66083 0 0 0.5 -0.48052 2.1 0.026007 2.5 0.087548 0.9 -0.27306 0.275876311 0.1570576~ 0.08941368E 0. 08941368E ( 0.21836012~ ( ( 0.53326447< 0.67372666! ( ( 0.08941368( ( 0. 089413681 ( ( 0. 089413681 0.1570576! 0.089413681 0.089413681 0.1570576! I I I I 0.3835557: 0.08941368 1 0.33073167 0.3835557 0.53326447 1.26296421 0.67372666 0.08941368 0.08941368 0.08941368 0.08941368 0.43474840 0.21836012 0.33073167 1.06171345 1.22334116 0.53326M7 w �KREBSPL LSBNUM UTMX DW-49-B DW-50-A DW-50-B DW-53--A DW-53--B DW-56-A DW-56-B DW-145-Jl DW-145-8 DW-157-A DW-157-8 EP-30-A EP-30-B EP-33--A EP-33--8 EP-118-A EP-118-8 EP-38-A EP-38-B EP-180-A EP-180-8 EP-9-A EP-9-8 EP-31-A EP-31-B EP-26-A EP-26-8 EP-6-A EP-6-B EP-27-A EP-27-8 EP-1-A EP-1-B EP-4-A EP-4-8 EP-28-A EP-28-8 EP-24-A EP-24-B EP-32-A EP-32-B EP-11-A EP-11-B EP-8-A EP-8-8 EP-36-A 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 434500 426188 426270 435584 435605 454053 453998 451826 451826 461484 461435 444784 444662 451484 451463 451984 452014 450684 450679 450158 450001 442308 442240 445384 445432 452684 452726 437509 437449 438284 438338 427480 427373 422217 422068 425884 426471 431484 431517 432284 432314 425062 424950 433478 433573 429484 UTMY 4398210 S 4396472 S 4396380 S 4388779 S 4388725 S 4379173 S 4379265 S 4410001 L 4410131 L 4390954 L 4390970 L 4470479 S 4470409 X 4460879 X 4460801 X 4448579 F 4448550 F 4448279 F 4448511 F 4474204 L 4478203 F 4465163 S 4464949 F 4470179 L 4470199 L 447.13879 L 4478919 L 4471239 S 4471169 S 4476679 S 4476732 S 4472833 L 4472806 L 4470980 S 4470689 S 4474379 S 4473939 L 4482579 S 4482788 S 4461379 L 4461296 S SN3 SN2 PRIMOVE SN1 3 1 SN5 SN4 SN6 SN7 SN8 SN9 SN10 1 MEAN 3 1 1 1 2 1 8 1 1 1 1 1 1 2 3 7 1 0 1 0 1 1 3 5 1 2 4 1 1 2 1 1 4 1 1 3 8 1 2 2 4 3 1 ~ ' 1 2 1 1 1 1 1 1 2 1 2 2 4 1 1 1 4 0 1 1 2 2 1 4 1 4 1 1 1 1 2 1 1 1 2 1 1 1 2 1 2 3 1 1 2 1 1 3 1 .1 1 14 2 0 1 2 1 3 2 1 1 3 1 1 1 1 1 2 2 1 16 1 3 4464125 L 4464089 L 4451783 L 4451435 F 4452379 L Page2of3 KREB'S N HARE DENSI 0.8 -0.31463 0.5 -0.48052 0.1 -1.0486 1 -0.23587 0.1 -1.0486 0.5 -0.48052 0.8 -0.31463 0.3 -0.66083 0.6 -0.41617 0.3 -0.66083 1 -0.23587 2.1 0.026007 1.5 -0.09275 1.6 -0.06998 1 -0.23587 0.1 -1.0486 0.4 -0.55929 0.2 -0.80394 0 0 0.4 -0.55929 0 0 0.1 -1.0486 0 0 0.9 -0.27306 1 -0.23587 0.3 -0.66083 0.1 -1.0486 0.4 -0.55929 0.1 -1.0486 0.3 -0.66083 0.4 -0.55929 0.1 -1.0486 -1.0486 0.1 0.2 -0.80394 0.1 -1.0486 1.2 -0.17152 2.4 0.073139 0.2 -0.80394 1.6 -0.06998 0.1 -1.0486 0 0 0.3 -0.66083 0 0 0 0 0 0 0 0 0.48458460£ 0.33073167E 0.08941368E 0.58093962: 0.08941368E 0.33073167E 0.48458460£ 0.218360121 0.3835557c 0.218360121 0.58093962: 1.0617134~ 0.807690771 0.85118668: 0.58093962: 0.08941368E 0.275876311 0.15705761 ( 0.275876311 ( 0.08941368E ( 0.53326447t 0.58093962: 0.218360121 0.08941368E 0.275876311 0.08941368( 0.21836012', 0.275876311 0.08941368E 0. 08941368( 0.15705761 0.08941368( 0.67372666: 1.18342010·, 0.15705761 0.85118668( 0.08941368E ( 0.21836012'. ( ( ( ( �KREBSPL LSBNUM UTMX EP-36-8 EP-35-A EP-35-B SS-103-A SS-103-B SS-107-A SS-107-B SS-111-A SS-111-B S$115-A SS-115-B WA•TT•A SS-37-A SS-37-B VA-11-A VA-11-B VA-12-A VA-12-B VA-153-A VA-153-BVA-154-A VA-154-B 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 429442 434484 434201 406384 408211 386984 387083 413184 413242 406384 406348 403884 401384 401418 366284 365940 387284 387580 373284 373100 367984 368100 UTMY PRIMOVE SN1 4452678 L 4454279 L 4454346 L 4460379 S 4482059 S 4469379 S 4469472 S 4466179 L 4466143 L 4460379 L 4460391 L 4529079 F 4450579 S 4450424 S 4409979 A 4409800 S 4380379 S 4380600 S ~874 S 4399720 S 4399379 S 4398800 S SN3 SN2 SN4 SN5 SNS SN7 SNS SN9 SN10 MEAN 1 1 3 2 1 1 1 1 1 1 t 1 1 2 2 1 1 2 1 2 1 1 1 1 2 3 5 1 4 2 2 1 1 1 1 3 3 2 KREB'SN HARE DENSll 0 0 0 0 0 0 0 0 0.1 -1.0486 0 0 0 0 0 0 0.1 -1.0486 0 0 0 0 0.3 -0.66083 0 0 0.3 -0.66083 0 0 1.1 -0.20223 0.7 -0.36176 0.2 -0.80394 1.6 -0.06998 0.9 -0.27306 0.6 -0.41617 0.3 -0.66083 ( ( ( ( 0.08941368{ ( ( ( 0.089413681 I I 0.21836012 I 0.21836012 I 0.62772865 0.43474840: 0.1570576' 0.851186681 0.533264470.3835557; 0.21836012 \0 V, Page 3 of 3 �I KREBSPL LSBNUM UTMX CG-12-A CG-12-8 CG-78-A CG-78-B SS-21-A SS-21-B SS-24-A SS-24-B SS-29-A SS-29-B SS-31-A SS-31-B SS-33-A SS-33-B SS-37-A SS-37-B SS-91-A SS-91-B VA-41-A VA-41-B W-5-A W-5-B W-13-A W-13-B W-53-A W-53-B W-81-A W-81-B W-82-A W-82-B SS-35-A SS-35-B S$-65-A SS-65-B WA-45-A WA-45-B CR-73-A CR-73-B CR-46-A CR-46-B CR-49-A CR-49-B WA-85-A WA-85-B SS-66-A SS-66-B 2 2 2 2 2 2 2 2 2 2 2 2 UTMY 4516720 F 4516636 L 4518850 F 4518851 F 4470720 A 4470747 A 4451320 L 4451400 L 4439720 L 4439703 L 4437620 L 4437718 L 4432620 L 4432676 L 4429320 L 4429389 L 4467220 A 4467128 A 442-0619 L 4425591 L 4533917 L 453.3875 w 4513120 F 451"3062 F 4509020 L 45013942 A 4514220 A 4514433 F 4514120 A 4513955 L 4430920 F 4430965 A 4475620 S 4475833 S 4521319 S 4521281 S 4530820 A 2 2 2 323599 323546 322068 322094 367299 367408 351299 351255 355299 355435 363099 363097 361899 361730 342699 342599 366082 365986 345207 345219 336521 336407 331599 331657 345899 345745 347099 347250 331599 331514 345099 345191 355899 355886 346477 346543 330499 2 2 2 2 2 2 2 2 2 330582 4530639 L 310699 4518920 S 310548 4518948 S 313599 4518520 S 313790 4518470 S 333799 4482873 A 333714 4492646 A 352650 4468880 A 352523 4468800 A 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 SN2 PRIMOVE SN1 SNS SN3 SN9 AfpluY\,J,·-1. C. SN10 MEAN 1 1 1 2 1 5 :.! 1 1 1 1 1 1 2 2 3 41 1 1 13 2 1 .. , 2 1 1 4 1 6 3 2 2 9 1 3 1 2 4 2 1 9 1 11 2 2 9 2 2 1 1 1 9 2 1 1 2 2 4 2 Page 1 of 2 1 3 2 3 KREB'$ N HARE DENS~ 0.1 0 0.2 0.1 0 0 1.7 0 0.4 0 0 0 0.1 0.1 0.5 0 0.9 2.7 0 0 5.4 2.2 0.1 1.1 0 0 1.3 1.4 0 0 0 0 0 0 0.4 0.1 0 0 0.1 0 0 0 0.5 1.2 0.2 0 -1.0486 0 -0.80394 -1.0486 0 0 -0.04858 0 -0.55929 0 0 0 -1.0486 -1.0486 -0.48052 0 -0.27306 0.114712 0 0 0.359367 0.042427 -1.0486 -0.20223 0 0 -0.14326 -0.11711 0 0 0 0 0 0 -0.55929 -1.0486 0 0 -1.0486 0 0 0 -0.48052 -0.17152 -0.80394 0 0.08941368€ C 0.15705769 0.08941368€ C C 0.8941761 OE C 0.275876311 C C C 0.08941368€ 0.08941368€ o.33073167e C 0.533264474 1.302302811 C C 2.287532027 1.102623281 0.089413686 0.627728653 0 0 0.719011504 0. 76364 7967 C C C C C C 0.275876311 0.08941368€ C C 0.089413686 C C C 0.330731678 0.673726665 0.15705769 (\ �l KREBSPLLSBNUM UTMX SS-69-A SS-69-B WA•1·A WA-1·8 WA-17-A WA-17-B CR-50-A CR-50-8 CR-83-A SS-25-A SS-25-B WA-61-A WA-61-8 WA-86-A WA-9-A WA-9-B 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 355658 355565 334299 334285 361099 361201 303378 3Q3492 304699 359099 359206 353099 353099 345895 365099 365120 UTMY PRIMOVESN1 4461691 S 4461595 S 4540420 S 4540385 L 4497920 F 4497971 F 4514532 S 4514404 S 4508720 A 4450020 L 4450122 L 4493520 S 4493520 S 4491620 A 4528020 L 4527890 L SN3 SN2 SN4 SNS SN6 SN7 SNS SN9 SN10 MEAN 1 2 1 1 1 1 1 2 1 2 3 c.;.·.. 1 KREB'$ N HARE DENSIT 0 0.1 0.2 0.3 0 0.1 0 0.1 0 0 0 0 0.2 0.1 0.3 0.3 0 -1.0486 -0.80394 -0.66083 0 -1.0486 0 -1.0486 0 0 0 0 -0.80394 -1.0486 -0.66083 -0.66083 0 0. 089413686 0.15705769 0.218360121 C 0.089413686 C 0.08941368€ C C C C 0.1570576~ 0.08941368E 0.218360121 0.218360121 l,O Page 2 of 2 -..J �• KREBSPL LSBNUM UTMX GS-9-A GS-9-B GS-13-A GS-13-B GS-57-A GS-57-B GS-60-A GS-60-B GS-61-A GS-61-B GS-65-A GS-65-B GS-67-A GS-67-B GS-93-A GS-93-B GS-98-A GS-98-B M-25-A M-25-B M-71-A M-71-B M-73-A M-73-B M-1-A M-1-B M-74-A M-74-B M-77-A M-77-B M-79-A M-79-B VA-19-A VA-19-B ME-29-A ME-29-B ME-33-A ME-33-B ME-85-A ME-85-B ME-26-A ME-26-B ME-81-A ME-81-B ME-21-A ME-21-B 3 .3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 323290 323297 276237 276431 274782 274773 296592 296642 297492 297601 290492 290300 276892 276966 311992 311826 290592 290500 321592 321584 298192 298256 287292 287246 308292 308099 313192 313305 324792 324962 323992 324030 329547 329434 319157 319160 290592 290682 313870 313932 314992 315015 313492 313407 314792 314733 UTMY PRIMOVE SN1 4412469 F 4412272 F 4395669 0 4395598 0 4408689 S 4408444 S 4403498 S 4403428 S 4402598 S 4402735 S 4399198 S 4399250 S 4396587 A 4396513 0 440!i098 S 4405172 S 4397498 S 4397850 S 444-2998 S 444~917 S 445TT98 A 4457786 A 4455598 A 4455648 A 4448798 A 4448695 A 4454298 A 4454069 A 4452690 L 4452684 L 4450598 A 4450468 A 4416879 A 4416904 F 4435523 F 4435606 F 4432198 A 4432190 F 4440307 A 4440419 F 4441498 S 4441563 S 4448198 F 4448093 A 4446098 S 4446242 S SN4 SN3 SN2 SN9 SN10 MEAN Atf'a.~d tx D 1 1 4 4 2 5 3 1 1 8 1 3 1 3 L .. ' 7 1 1 1 1 1 1 2 3 3 10 13 1 3 1 3 1 5 5 1 1 4 5 2 1 1 27 23 1 1 1 2 2 2 1 3 1 2 2 2 5 1 1 1 1 1 1 1 1 1 6 1 1 2 1 1 1 1 2 Page 1 of 2 IO KREB'S N HARE DENSIT'I 00 0.1 -1.0486 0 0 0 0 0 0 0.2 -0.80394 0.5 -0.48052 0.6 -0.41617 0 0 0.9 -0.27306 1.3 -0.14326 0 0 0.4 -0.55929 0 0 0 0 0.9 -0.27306 0 0 0.1 -1.0486 0.5 -0.48052 0.3 -0.66083 1.3 -0.14326 1.3 -0.14326 0 0 0.7 -0.36176 0 0 0 0 0 0 0 0 0 0 4 0.253441 4 0.253441 0 0 0 0 0.6 -0.41617 0.4 -0.55929 0 0 0.1 -1.0486 0.1 -1.0486 0.1 -1.0486 0.5 -0.48052 1.6 -0.06998 0.3 -0.66083 0.2 -0.80394 0.3 -0.66083 0.1 -1.0486 0.1 -1.0486 0.3 -0.66083 0.08941368€ C C C 0.15705769 0.330731678 0.38355578 C 0. 5332644 74 0.719011504 C 0.275876311 C C 0.533264474 C 0.08941368€ 0.33073167t 0.218360121 0.719011504 0.719011504 C 0.434748402 C C C C C 1. 79242672: 1. 79242672: C C 0.3835557f 0.275876311 C 0.08941368€ 0.08941368E 0.08941368E 0.33073167( 0.85118668: 0.218360121 0.15705761 0.218360121 0.08941368E 0.08941368€ 0.218360121 �KREBSPLLSBNUM UTMX GS-53-A GS-53-B GS-69-A Gs-69-B GS-17-A GS-17-B GS-41-A GS-41-B GS-42-A GS-42-B GS-43-A GS-43-B ME-45-A ME-45-B ME-46-A ME-46-B ME-89-A ME-89-B GS-49-A ME-30-A ME-30-B 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 315807 315741 303319 303421 283207 283246 319767 319706 320543 320495 3204n 320499 295792 295928 291692 291663 295600 295532 268390 325792 325748 UTMY PRIMOVESN1 4412384 F 4412364 F 4391172 S 4391426 S 4400690 S 4400732 A 4422156 F 4422201 F 4421864 F 4421868 F 4421531 F 4421391 F 44;.:w98 S 4420845 S 4420398 S 4420254 S 44f6432 A 44~957 A 4415600 A 4433698 F 4433677 L SN3 SN2 SN4 SNS SN6 SN7 SNS SN9 2 2 8 1 2 1 1 0 1 8 2 8 1 SN10 MEAN 1 3 7 5 3 3 1 10 14 4 3 0 1 2 1 3 3 7 1 3 2 5 1 3 3 4 4 1 1 1 3 1 1 5 3 5 2 8 3 6 2 1 4 1 1 -. 1 1 6 KREB'$ N HARE DENSl1" 1.6 -0.06998 0 0 -1.0486 0.1 0.3 -0.66083 0.3 -0.66083 0 0 3.5 0.20631 3.2 0.17468 1.6 -0.06998 3.3 0.185541 2.3 0.058117 2.1 0.026007 0 0 0 0 0 0 0 0 0 0 0 0 0.1 -1.0486 0.1 -1.0486 0.6 -0.41617 0.851186685 0 0.089413686 0.218360121 0.218360121 0 1.60808792 1.495133686 0.851186685 1.532996908 1.143186215 1.061713454 01 0 0 0 0 0 0.089413686 0.089413686 0.38355578 l,C) l,C) Page 2 of 2 �KREBSPL LSBNUM UTMX UTMY PRIMOVE SN1 PA-12-A PA-12-B GU-82-A GU-82-B GU-86-A GU-86-B SA-14-A SA-14-8 SA-64-A SA-64-B CA-25-A CA-25-B CA-27-A CA-27-B CA-30-A CA-30-B CA-32-A CA-32-B CA-38-A CA-38-B CA-46-A CA-46-B 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 308071 4273055 D 308068 • 4273008 D 372071 4301755 L 371956 4301612 L 370471 4289955 L 370421 4290005 L 370371 4230655 P 370487 4230576 P 374471 4235455 L 374271 4235529 L 291683 4359468 A 291929 435a523 A 292637 4358699 A 292438 4358853 A 260735 4356491 A 260659 4356410 A 282106 435;!554 0 282058 4352570 0 293676 4335150 A 293766 4335189 A 253361 4321074 A 253422 4320992 A CA-119-A CA-119-B CA-132-A CA-132-B CA-141-A CA-141-B GU-61-A GU-61-B GU-80-A GU-80-B GU-82-A GU-82-B GU-86-A GU-86-B GU-89-A 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 283084 4357784 S 283109 4357727 S 307856 4337843 D 307950 4337810 D 313455 4326347 D 313440 4326279 D 412171 4308255 P 412146 4308040 P 398571 4312655 P 398581 4312777 P 372071 4301755 L 371956 4301612 L 370471 4289955 L 370421 4290005 L 394371 4280155 L 394552 4280140 L 372671 4274155 L 372600 4274150 L 364671 4269255 D 364685 4269700 D 292071 4292555 F 292034 4292573 F 290071 4291055 A 290001 4290862 A GU-89-B GU-91-A GU-91-B GU-92-A GU-92-B PA-19-A PA-19-B PA-56-A PA-56-B 4 4 SN3 SN2 SN8 3 0 2 4 18 2 1 0 0 2 6 0 1 0 0 0 0 0 2 1 Apf Q.M. d I.)( 0 SN10 MEAN 3 t 2 0 0 0 SN9 0 0 1 5 12 4 0 0 0 0 1 3 2 0 0 0 0 0 0 0 3 < • 1 6 1 1 1 2 5 1 3 1 1 6 1 2 1 2 3 3 1 6 7 2 2 1 1 1 3 4 18 2 1 6 1 2 1 2 1 1 2 1 2 2 4 1 5 1 3 12 4 33 1 2 1 1 Page 1 of 3 4 3 1 2 1 1 3 1 KREB'S N HARE DENSl°G: ,-., 0.38355578 0.38355578 0.627728653 1.495133686 0.978772092 0.434 748402 0.15705769 0.275876311 0 0 0 0 0.218360121 0 0 0 0 0. 089413686 0 0 0 0 0.330731678 0.673726665 0.894176106 0.434748402 0 0.38355578 0.484584609 0.15705769 0.533264474 0.484584609 0.627728653 1.495133686 0.978772092 0.434748402 1. 792426725 0.218360121 C C C C 0.38355578 0.15705769 C 0 0.6 -0.41617 0.6 -0.41617 1.1 -0.20223 3.2 0.17468 1.9 -0.00932 0.7 -0.36176 0.2 -0.80394 0.4 -0.55929 0 0 0 0 0 0 0 0 0.3 -0.66083 0 0 0 0 0 0 0 0 0.1 ·1.0486 0 0 0 0 0 0 0 0 0.5 -0.48052 1.2 -0.17152 1.7 -0.04858 0.7 -0.36176 0 0 0.6 -0.41617 0.8 -0.31463 0.2 -0.80394 0.9 -0.27306 0.8 -0.31463 1.1 -0.20223 3.2 0.17468 1.9 -0.00932 0.7 -0.36176 4 0.253441 0.3 -0.66083 0 0 0 0 0 0 0 0 0.6 -0.41617 0.2 -0.80394 0 0 0 0 �I KREBSPLLSBNUM UTMX SA-65-A SA-65-B SA-69-A SA-69-8 SA-67-A SA-67-8 SA-100-A SA-100-B SA-102-A SA-102-B SA-104-A SA-104-B SA-107-A SA-107-B SA-108-A SA-108-B SA-109-A SA-109-B SA-110-A SA-110-B VA-1-A VA-1-B VA-2-A VA-2-8 CA-33-A CA-33-B LE-3-A LE-3-8 CA-133-A CA-133-B LE-24-A LE-24-B LE-116-A LE-116-8 VA+A-2 VA-111-A VA-113-A CA-17-A- CA-17-B CA-123-A CA-123-B GU-52-A GU-52-B GIJ.157•A GIJ.157•8 GIJ.15a.A 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 ·4 4 4 4 4 4 4 4 4 4 4 4 4 4 UTMY PRIMOVE SN1 373071 373233 390271 390070 368013 367745 370371 370242 374271 374148 362171 362268 368771 368724 365571 365543 351871 351662 351530 351524 337267 337329 351577 351544 302269 302330 436930 436945 245402 245436 350683 350801 360122 360057 337271 356671 357371 296459 4235355 A 4235373 A 4230055 D 4229851 D 4231272 D 4231280 A 4246555 P 4246541 L 4245655 L 4245672 L 4241955 L 4241864 L 4236155 L 42~_6361 L 4234755 D 4234940 S 4231455 A 4231491 D 4229214 D 4229297 D 4380741 L 4380577 L 4374462 L 4374495 L 4351331 4351244 A 3397400 S 3397800 S 4335106 A 4335245 F 4370138 L 4369892 A 4371050 S 4370936 S 4380755 L 4379855 L 4378955 S 4373781 S 296390 315656 315720 327571 327548 348771 348674 365371 4373800 S 4351061 L 4350930 F 4296255 A 4296315 A 4303555 S 4303630 S 4309155 L SN2 SN4 SN3 SN5 SN6 SN7 SN9 SN8 SN10 MEAN 1 1 2 1 0 7 2 4 1 2 1 1 1 1 1 2 4 1 C, , 4 1 13 2 3 2 1 2 1 1 1 2 5 2 11 1 1 1 5 1 1 0 1 1 5 1 2 4 0 1 2 6 1 1 1 3 1 3 4 1 3 1 1 1 Page2of3 3 KREB'S N HARE DENSIT'° 0 0 0.2 -0.80394 1.6 -0.06998 0.3 -0.66083 0 0 0 0 0 0 0 0 0.1 -1.0486 0.1 -1.0486 0.1 -1.0486 0.2 -0.80394 0.3 -0.66083 0.4 -0.55929 0.4 -0.55929 0 0 0.3 -0.66083 0.6 -0.41617 2.4 0.073139 1.4 -0.11711 0.1 -1.0486 0.8 -0.31463 0 0 0.1 -1.0486 0 0 0 0 0 0 0 0 1.1 -0.20223 0 0 0 0 0 0 ·O 0 0.5 -0.48052 1.6 -0.06998 0.1 -1.0486 0 0 0.4 -0.55929 0.1 -1.0486 0.7 -0.36176 0.2 -0.80394 0 0 0 0 0 0 0 0 0 0 ( 0.1570576~ 0.85118668~ 0.218360121 ( ( ( ( 0.08941368( 0.08941368( 0.08941368( 0.1570576~ 0.21836012', 0.27587631, 0.27587631' ( 0.21836012' 0.38355571 1.18342010' 0.76364796; 0.08941368( 0.48458460( ( 0.08941368( ( ( ( ( 0.62772865 I I I I 0.33073167: 0.85118668: 0.089413681 I 0.27587631 0.089413681 0.43474840: 0.1570576' I I I I .... -- �- I KREBSPLLSBNUM UTMX GU-153-8 LE-9-A LE-9-8 LE-73-A LE-73-B LE•74•A LE-74-B LE-78-A LE-78-B LE-128-A LE-128-B LE-129-A LE-129-B LE-144-A LE•144-B PA-18-A PA-18-B PA•22·A PA-22-B PA-20-A PA-20-B PA-21·A PA-21·8 PA-186-A PA-186-B PA·187•A PA·187•B SA-106-A SA-106-B LE-29-A LE•29-B LE-120-A LE-120-B C0-148-B LE-5-A LE-5-B CR-83-8 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 365558 412871 412756 377871 377906 370877 370998 407171 407263 373571 373292 368871 369228 371171 371104 320871 320822 288076 288225 293771 293804 294371 294368 290271 290372 288471 288585 360209 360219 346471 346471 350771 350771 761178 380700 380520 304199 UTMY SN2 PRIMOVESN1 4309087 L 4317355 P 4317207 P 4334155 L 4334211 L 4333181 S 4333200 S 4318255 S 4318427 S 4347255 F 4346979 F 434.7155 F 4347084 F 4324855 L 4324853 L 4304655 S 4304619 S 4266178 S 4266146 A 4275055 F 4275216 F 4270255 S 4270108 S 4271155 S 4271208 S 4270055 S 4270239 S 4237602 S 4237494 S 4357055 L 4357055 L 4354155 L 4354155 L 4148888 S 4361960 S 4362240 S 4509001 A SN4 SN3 SNS SN6 SN7 SNS SN9 SN10 MEAN 1 1 1 1 1 2 1 1 1 3 2 1 1 L,,·• 1 1 1 1 6 1 1 2 2 2 1 1 1 1 1 1 1 1 1 3 2 1 1 2 4 1 1 1 1 1 2 1 1 1 1 8 5 1 1 Page 3 of 3 3 1 1 7 KREB'S N HARE DENSll:0 0.1 0 0 0.1 0.2 0.3 0.2 0.1 0.3 0.2 0 0 0.1 0.5 0.7 0.1 0 0.2 0.7 0.3 0.4 0.5 0.2 0.4 0.5 0.2 0.3 0.2 0 0 0 0 0 2.3 0.2 0.2 0 -1.0486 0 0 -1.0486 -0.80394 -0.66083 -0.80394 -1.0486 -0.66083 ·0.80394 0 0 ·1.0486 -0.48052 -0.36176 -1.0486 0 -0.80394 -0.36176 -0.66083 -0.55929 -0.48052 -0.80394 -0.55929 -0.48052 -0.80394 -0.66083 -0.80394 0 0 0 0 0 0.058117 -0.80394 -0.80394 0 N 0. 08941 368( ( ( 0.08941368( 0.1570576! 0.21836012~ 0.1570576! 0.08941368( 0.21836012' 0.1570576! ( ( 0.08941368( 0.330731671 0.43474840; 0.08941368( ( 0.1570576! 0.43474840; 0.21836012" 0.27587631" 0.330731671 0.1570576! 0.27587631 • 0.330731671 0.1570576! 0.21836012" 0.1570576! ( ( ( ( ( 1.14318621 ! 0.1570576! 0.1570576! ( �UTMY AN-152-A AN-152-B AN-156-A AN-156-B AN-157-A AN-157-B AN-158-A AN-158-B AN-160-A AN-160-B AN-168-B AN-168-A C0-14-A C0-148 C0-187-A C0-187-B DE-8-A DE-8-8 DE-13-A DE-13-B DE-49-A DE-49-B DE-54-A DE-54-B DE-89-B DE-89-A DE-146-A DE-146-B DU-30-A DU-30-B DU-192-A DU-192-B M0-3-A M0-3-B M0-28-A M0-28-8 M0-29-A M0-29-B M0-64-A M0-64-8 M0-66-A M0-66-B DU-186-A DU-186-8 DU-190-A DU-190-B 4145971 S 4143036 S 4142726 D 4142692 D 41'4'0671 S 4140586 S 4139791 S 4139726 S 4137071 S 4137124 S 4105024 S 4104968 S 4146928A 4146893 A 4145794 A 4145810 A 4189271 D 4189201 D 4151461 S 4151379 S 4205671 S 4205646 S 4158215 P 4158145 P 4203561 B 4203671 B 4152177 S 4152211 S 4135828 0 4135981 0 4134560 P 4134607 P. 4234397 A 4234334 A 4221691 0 4221649 0 4220445 A 4220443 A 4231212 D 4231125 D 4228274 S 4228172 S 4147354 0 4147273 P 4138467 D 4138570 D 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 370326 370357 367132 367087 369726 369772 366889 366884 371126 371201 375941 375935 747483 747553 748557 748500 364326 364304 370832 370796 378026 377952 368766 368729 364807 364826 335236 334987 319412 319396 251960 251911 277574 277510 249194 249047 258872 258968 277527 277463 277150 277120 254345 254522 265782 265683 'ffi SN2 PRIMOVE SN1 KREBSPL LSBNUM UTMX 1 SN3 1 3 1 4 4 2 1 0 1 0 0 0 0 1 2 SNS 3 2 2 6 5 0 0 3 0 1 0 A,o p~d ,·x F 1 5 2 0 0 2 3 0 3 1 0 1 0 1 0 0 1 2 0 0 0 4 0 0 0 1 1 2 5 l 1 0 0 0 11 5 2 0 0 1 3 11 10 2 2 0 0 4 . 0 1 2 1 1 1 0 0 1 0 0 1 1 0 4 3 6 23 SN9 0 0 19 0 1 1 0 0 2 2 0 0 2 1 SN10 MEAN 4 7 5 2 7 5 14 2 12 0 0 1 1 1 2 13 7 0 0 0 0 3 0 6 1 0 0 0 2 0 1 0 Page 1 of 6 KREB'S N HARE DENSITt 0.1 -1.0486 0.1 -1.0486 2.5 0.087548 5.1 0.339192 1.5 -0.09275 0.7 -0.36176 3.5 0.20631 2.5 0.087548 0.4 -0.55929 0.7 -0.36176 2.5 0.087548 3 0.1519 0 0 0 0 0 0 0 0 3.5 0.20631 1.3 -0.14326 0.2 -0.80394 1.3 -0.14326 0.4 -0.55929 0.2 -0.80394 0.4 -0.55929 0.7 -0.36176 0.1 -1.0486 0.1 -1.0486 0.2 -0.80394 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.4 -0.55929 0.8 -0.31463 0 0 0 0 0 0 0 0 0.08941368( 0.08941368( 1.2233411 SL 2.18369704! 0.80769077'. 0.43474840: 1.6080879: 1.22334116' 0.27587631' 0.43474840: 1.22334116' 1.41873187< ( ( ( ( 1.6080879: 0.71901150• 0.1570576! 0.71901150, 0.27587631 0.1570576! 0.27587631 0.43474840: 0.089413681 0.089413681 0.1570576' I ' 0.27587631 0.48458460 0 �I KREBSPLLSBNUM UTMX DU-193-A DU-193-B M0--2-B M0--2-A M0--60-A M()..60-B M0--61-A M0--61-B M()..65-A M0--65-B M0--68--A M0--68--B M0--79-A M0--79-B M0--179-A M0--179-B DE-144-A DE-144-B AN-167-A AN-167-B DU-33-A DU-33-B M0--4-B M0--4-A M0--26-A M0--26-B M0--72-A M0--72-B M0--77-A M0--77-B M()..80-B M0--80-A SA-1-A SA-1-B SA-39-A SA-39-B SA-41-A SA-41-B 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 SA-69-A SA-69-B SA-76-A SA-76-B SA-78--A SA-78--B S1-96-A 5 Sl-96-B 5 5 5 5 5 5 5 UTMY SN2 PRIMOVESN1 SN4 SN3 SNS SN6 SN7 SN8 SN9 SN10 MEAN 318011 4134176 P 318116 4134283 P 268285 42;¼~607 A 268326 4238671 A 284257 4241267 A 284212 4241345 A 280720 280683 286117 280095 281510 281561 242213 4234378 S 4234438 S 4229070 S 4229119 S 4225540 S 4225578 S 4211663 S 242282 4211746 S 241604 241618 373926 373904 379909 379825 268426 268396 271352 271426 265626 265631 276226 276156 271726 271755 251688 251626 '337326 337333 332426 332433 339126 339218 4222825 A 4222749 A 4152871 S 4152938 S 4105045 S 4105078 S 4136471 P 4136238 P 4232051 A 4232071 A 4235571 0 4235639 0 4222071 S 4222085 S 4215171 S 4215157 S 4211206 S 4211171 S 4244871 F 4244929 A 4230371 L 4230234 L 4235071 D 4234944 D 331926 332041 341126 341115 327073 327039 267926 267859 4224571 L 4224576 L 4217171 S 4217057 S 4213018 F 4213095 F 4196571 S 4196596 S 3 1 1 0 4 3 4 6 1 3 1 2 0 6 2 1 4 6 7 2 8 2 2 0 1 1 3 1 3 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 2 2 3 4 2 1 1 1 9 0 0 1 1 1 2 0 1 1 2 0 1 1 1 3 0 0 9 0 8 11 2 0 1 1 7 2 14 8 4 16 0 1 0 0 0 0 6 5 1 0 0 0 0 0 1 3 1 0 5 6 0 2 6 3 5 L > 4 7 1 13 23 12 1 3 2 6 2 4 2 8 0 5 1 7 0 3 1 1 0 0 0 2 0 0 0 0 2 0 3 7 7 0 5 9 0 0 1 1 2 0 2 3 0 0 0 0 0 0 0 0 0 0 0 0 3 2 1 2 1 1 1 2 3 3 13 1 10 Page 2 of6 7 2 2 2 6 1 1 1 5 5 2 2 1 2 3 2 2 3 4 6 1 KREB'S N HARE DENSIT' 0 0 0 0 0 0 0 0 0 0 0 0 1.2 -0.17152 2.1 0.026007 0.4 -0.55929 0.4 -0.55929 0.9 -0.27306 1.1 -0.20223 2.6 0.101391 2.2 0.042427 0 0 0 0 2.9 0.139934 5.3 0.35277 0.7 -0.36176 0.9 -0.27306 0 0 0 0 0 0 0 0 0 0 0 0 3.5 0.20631 2.2 0.042427 2 0.008786 3.2 0.17468 3.3 0.185541 0.9 -0.27306 1.2 -0.17152 0 0 0 0 0 0 0 0 0.3 -0.66083 0.2 -0.80394 1.8 -0.0284 1.1 -0.20223 1.4 -0.11711 1.4 -0.11711 1.9 -0.00932 1.9 -0.00932 3.1 0.163474 ( ( ( ( ( ( 0.67372666'. 1.06171345L 0.275876311 0.275876311 0.53326447.: o.62772865c 1.26296421; 1.10262328' ( ( 1.38017557( 2.25304338( 0.43474840: 0.53326447 1 ( ( ( ( ( ( 1.6080879: 1.10262328 • 1.02043701 I 1.49513368{ 1.53299690! 0.53326447< 0.67372666! ( ( ( I 0.21836012 0.1570576! 0.93669425( 0.62772865: 0. 76364 796' 0.76364796', 0.97877209: 0.97877209: 1.4570481 e· �KREBSPLLSBNUM UTMX SI-102-A SI-102-B SI-107-A St-107-B SI-120-A S1-120-B DE-111-A DE-111-B DE-112-A DE-112-B DE-138-A DE-138-B SI-50-A S1-50-B SI-86-A SI-86-B SI·110-A S1·110-B SI-113-A SI-113-B SI-122-A Sl-122-B SI-126-A SI-126-B DC-140-A DC-140-B DC-141-A DC-141-B SI·12-A S1-12-B SI-185-A . SI-185-B DU-147-A DU-147-B DU-145-A DU-145-B DU-151-A DU-151-B DU-154-A DU-154-B DU-189-A DU-189-B SI-123-A SI-123-B OE-132-A DE-132-B 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 UTMY 260726 41aa211 o 260809 4193273 D 274226 4188771 S 274147 4188681 S 317826 4171771 S 317901 4171806 S 344226 4185671 A 344198 4185621 A 350326 4183271 S 350264 4183331 S 351826 4157371 S 351853 4157311 S 298726 4201771 S 298652 4201741 S 308226 4206871 S 308294 4206850 S 293526 4186371 A 293451 4186413 A 294626 4182671 S 294582 4182594 S 316126 4171271 S 316056 4171236 S 319126 4168071 S 319118 4168086 S 761926 4156371 F 761750 4156280 F 761426 4156271 F 761350 4156350 F 247730 4156170 A 247800 4156200 A 253126 4155871 P 253110 • 4155730 D 261419 4150956 F 261451 4150964 S 245026 4152171 D 244980 4152260 D 248713 4146127 D 248830 4146020 D 281790 4144654 A 281712 4144508 A 257530 4140571 D 257150 4140565 D 254400 4170897 A 254369 4170937 D 334612 4165099 S 334624 4165200 S SN4 SN3 SN2 PRIMOVESN1 SN5 SNS SN7 SNS SN9 SN10 MEAN 4 2 7 7 4 3 1 2 2 1 1 2 2 1 8 7 1 8 4 5 5 10 1 3 6 4 6 3 5 1 8 5 16 4 6 18 1 1 2 2 3 3 5 1 3 2 14 3 3 9 5 2 8 1 4 5 7 2 5 1 8 1 6 3 1 2 5 4 8 5 ,. 2 1 2 10 5 9 3 1 3 12 7 26 7 11 3 4 4 7 3 2 5 1 4 9 14 6 7 6 19 1 7 4 8 2 7 11 12 2 4 6 3 2 1 1 4 3 1 1 19 5 3 1 1 3 2 1 3 1 2 3 3 1 11 1 1 1 2 1 1 3 1 1 1 1 3 4 I 2 8 1 1 Page 3 of 6 2 KREB'S N HARE DENSIT'i 0.4 -0.55929 0.7 -0.36176 0.7 -0.36176 3.3 0.185541 0.9 -0.27306 1.6 -0.06998 1.4 -0.11711 2.1 0.026007 4.3 0.278968 3.2 0.17468 1.6 -0.06998 2.4 0.073139 3.6 0.216253 7.3 0.465777 2.8 0.127548 4.5 0.295014 0 0 0 0 1.4 -0.11711 2.4 0.073139 2 0.008786 3.8 0.235337 1.9 -0.00932 3.3 0.185541 0 0 0 0 0 0 0 0 3.8 0.235337 1.7 -0.04858 0.3 -0.66083 0.4 -0.55929 1.8 -0.0284 0.5 -0.48052 0 0 0 0 0 0 0 0 0.1 -1.0486 0.3 -0.66083 0.9 -0.27306 0.4 -0.55929 0 0 0.2 -0.80394 0.1 -1.0486 1.4 -0.11711 0.275876311 0.434748402 0.43474840, 1.53299690f 0.53326447t 0.85118668: 0.76364796i 1.06171345': 1.90093784i 1.49513368€ 0.85118668: 1.183420101 1.645330m 2.92264992' 1.34136943: 1.97248834\ ( ( 0.76364796; 1.18342010' 1.02043701\ 1.71924109: 0.97877209: 1.532996901 I I I I 1.71924109 0.894176101 0.21836012 0.27587631 0.93669425 1 0.33073167, 0.08941368 0.21836012 0.53326447 0.27587631 0.1570576 0.08941368 0.7636479fr-, �KREBSPLLSBNUM UTMX SI-121-A SI-121-B DC-136-A DC-136-B AN-163-A AN-163-B AN-198-A AN-198-B AN-199-A AN-199-B AN-200-A AN-200-8 AN-208-A AN-208-B AN-210-A AN-210-8 DU-149-A DU-149-B AN-18-A AN-18-B AN-21-A AN-21-B AN-23-A AN-23-8 AN-20-A AN-20-8 AN-31-A AN-31-8 AN-37-A AN-37-B AN-38-A AN-38-B AN-56-A AN-56-B AN-153-A AN-153-B C0-15-A C0-15-B C0-148-A DC-118-A DC-118-B DC-119-A DC-119-B DC-100-A DC-100-8 OC-103-A 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 UTMY SN2 PRIMOVESN1 302378 302370 763726 763840 339226 339400 329626 329550 331126 331200 331426 331550 339626 339610 336626 336381 315226 315200 374826 374869 379849 379852 363426 363412 348526 348459 346626 346544 389326 389319 358726 368634 372726 372684 337526 337416 755508 755418 4171441 S 4171435 S 4161571 S 4161650 S 4127771 0 4127800 F 4128671 P 4128600 P 4126371 P 4126400 P 4125971 P 4126000 P 4107771 0 4107779 0 4105271 P 4105136 P 4147571 0 4147900 0 4114271 A 4114352 A 4108329 A 4108424 A 4096771 A 4096848 A 4112171 A 4112107 A 4102771 0 4102762 0 4100371 P 4100270 P 4096271 S 4096289 S 4112271 0 4112332 0 4145471 F 4145472 F 4143632 A 4143617 A 761083 763232 763168 763333 763262 746543 746470 751667 4148897 S 4173781 S 417;~814 S 4171687 S 4171605 S 4192851 S 41si824 s SN3 '. SNS SN4 2 1 2 SN6 SN? ,:• 1 3 1 2 2 1 1 3 5 6 3 7 11 3 5 1 2 MEAN 6 8 2 SN10 1 8 (. 8 SN9 4 1 3 2 2 SN8 3 5 6 3 2 4191077 S .. Page 4 of 6 2 1 4 1 2 KREB'S N HARE DENSI 1 -0.23587 0.2 -0.80394 1.1 -0.20223 1 -0.23587 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.4 -0.55929 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1.1 -0.20223 0 0 0 0 0.6 -0.41617 2.1 0.026007 0 0 0 0 2.4 0.073139 0 0 0 0 0.5 -0.48052 0.6 -0.41617 1 -0.23587 1.6 -0.06998 0 0 0.580939625 0.15705769 0.627728653 0.580939625 0 0 0 0 0 0 0 0 0 0 0 0 0 0.275876311 0 0 0 0 0 0 0 0 0 0 0 0 0 0.627728653 0 0 0.38355578 1.061713454 0 0 1.183420101 0 0 0.330731678 0.38355578 0.580939625 0.85118668" �KREBSPLLSBNUM UTMX DC-103-B OC-124-A DC-124-B DC-128-A DC-128-8 DE-142-A DE-142-8 DN-98-A DN-98-B DN-99-A ON-99-8 DU-35-A DU-35-B DU-194-A DU-194·8 DU-195-A DU-195-B DU-197-A DU-197-8 DU-201-A DU-201-B DU-204-A DU-204-B OC-9-A DC-9-B DC-97-A OC-97-B DC-101-A OC-101-B DC-104-A DC-104-B DC·108-A OC-108-B Sl-51-A S1-51-B Sl-87-A S1-87-B S1·90-B S1-90-A S1-106-A SI-106-B SI-109-A S1-109-B SI·114-A SI-114-B SI-11-A 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 UTMY SN2 PRIMOVESN1 751699 760062 760116 756231 756142 341726 341679 358626 358674 359126 359063 299926 299861 274126 274131 265726 265795 304326 304386 298826 298965 303126 303137 211626 211613 233026 232894 229826 229820 233926 233885 232026 232125 293005 293126 293826 293746 271226 271226 4191024 S 4169573 D 4169674 D 4166799 P 4166761 P 4154871 S 4154753 S 4195171 S 4195215 S 4194671 S 4194666 S 4130371 P 4130316 P 4132771 P 4132686 P 4132271 P 4132317 P 4130571 P 4130495 P 4124971 D 4125051 D 4119771 P 4119857 P 4181971 A 4181996 A 4196471 S 4196480 S 4193471 S 4193380 S 4190971 S 4190960 S 4188571 S 4188570 S 4199502 S 4199371 S 4206271 S 4206331 S 4203471 S 4203471 S 238966 238966 243126 243560 259926 260012 288426 4189716 S 4189971 S 4187271 S 4187052 S 4181871 S 4181977 S 4157071 A 9 1 5 1 6 1 1 3 ' SN4 SN3 5 4 SN5 3 1 7 9 SNS 2 5 5 SN? 1 3 8 SN8 SN9 4 8 1 8 SN10 MEAN 5 1 2 9 6 7 2 C 0 0 7 2 3 0 6 9 0 5 2 1 3 3 1 1 8 1 3 3 2 5 5 2 24 10 3 5 5 3 1 0 6 7 0 5 5 3 3 6 8 15 1 8 7 2 2 3 5 1 7 1 3 9 8 5 9 2 7 3 7 4 1 4 9 12 2 5 2 1 1 1 31 8 7 1 3 3 4 1 2 2 3 4 2 6 I 5 1 2 1 4 Page 5 of 6 2 4 3 6 4 7 1 2 KREB'S N HARE DENSIT'l 0 0 0 0 0 0 0 0 0 0 0 0 2.3 0.058117 1.9 -0.00932 2.8 0.127548 3.5 0.20631 2.8 0.127548 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 0.008786 2.1 0.026007 3.4 0.196078 1.6 -0.06998 3.1 0.163474 4.7 0.310363 2.2 0.042427 1.7 -0.04858 2.2 0.042427 1.9 ·0.00932 2.4 0.073139 2.1 0.026007 2.2 0.042427 1.6 -0.06998 2.5 0.087548 3.3 0.185541 0 0 0 0 2.4 0.073139 1.8 -0.0284 0 0 0 0 0 0 0 0 1.143186215 0.978772092 1.341369433 1.60808792 1.341369433 0 0 C C C C C C C C C C C C 1.0204370H 1.06171345L 1.57064583; 0.85118668'. 1.45704818\ 2.04344557! 1.10262328 • 0.89417610{ 1.10262328 • 0.97877209; 1.1834201 o· 1.06171345' 1. 10262328 • 0.85118668! 1.22334116, 1.53299690! ( ( 1.1834201 o· 0.93669425{ .E -.....i �l KREBSPLLSBNUM UTMX Sl-11-B Sl-125-A Sl-125-B Sl-129-A St-129-B Sl-133-A Sl-133-B Sl-134-A Sl-134-B 5 5 5 5 5 5 5 5 5 288420 276100 276250 264326 264400 287326 287150 277726 277550 UTMY PRIMOVESN1 4156972 A 4167900 A 4167850 A 4166571 D 4166500 D 4164071 S 4164100 S 4162971 D 4163000 D SN2 SNS SN4 SN3 SN6 SN7 SN8 2 7 SN9 SN10 MEAN 5 7 3 3 1 1 9 2 2 9 Page 6 of 6 1 3 KREB'S N HARE DENSITY~ 0.2 1.2 0.7 0.6 1.1 0 0.5 0.2 1 -0.80394 -0.17152 -0.36176 -0.41617 -0.20223 0 -0.48052 -0.80394 -0.23587 0.15705769 0.673726665 0.434748402 0.38355578 0.627728653 0 0.330731678 0.15705769 0.580939625 �109 Appendix G Genus-species Abbr. Name Scientific Common AMAL Amelanchier alnifolia Serviceberry ARCO Arnica cordifolia Heart-leaved Arnica ARUV Arctostaphylos uvaursi Kinnikinnik CA spp Carex spp Sedge JU spp Juniperus spp Juniper Quercus gambelii Gambel's Oak QUGA " SHCA Shepherdia canadensis Buffaloberry SYAL Symphoricarpos albus Snowberry SYOC symphoricarpos occidentalis Snowberry VA spp Vaccinium spp Blueberry VASC Vaccinium scoparium Broom Huckleberry �110 Appendix I! SUMMARY BY FOREST ---- r· J_ - - ___j_ J --- --- j ---- ___ SAMPLE SIZE HARES PER HA j ------- - -- ------ -- --- ------ -------- -- - - - - -- --- 32 03142 ARAPAHO NF------- ------ -----0.5089 ____1_l BUFFALO PEAKS WILDERNESS 21 0.1046 GRAND MESA NF --------------------~ 63 0.3259 GUNNISON NF --7 0.8712 LAGARITA WILDERNESS AREA -------89 0.6679 RIOGRANDE NF 31 0.2744 ROCKY MOUNTAIN NATIONAL PARK 22 0.2158 ROOSEVELT NF 83 0.2114 ROUTT NF 16 0.2815 SAN !SABEL NF 92 0.3546 SANJUAN NF 45 0.7032 UNCOMPAHGRE NF 4 0.1352 WEST ELK WILDERNESS 94 0.2632 WHITE RIVER NF I SAMPLE SIZE HARES PER HA 95% Cl SUMMARY BY AR~I --- --------- ---· ---- ---- - ---- ------ ---- ------- - - - - -- - I " 114 62 67 129 239 0.29 0.27 0.41 0.28 0.56 0.16 0.14 0.22 0.16 0.35 AREA 1 AREA2 AREA3 AREA4 AREAS I SUMMARY BY PRIMARY OVERSTORY 0.2200 0.1900 0.3000 0.2100 0.4400 SAMPLE SIZE HARES PER HA I I 105 2 48 64 108 16 41 220 1 3 ASPEN BRISTLECONE PINE DOUGLAS FIR SUBALPINE FIR LODGEPOLE PINE GAMBLE'S OAK PONDEROSA PINE ENGLEMANN SPRUCE WHITE PINE UMBER PINE 0.1558 0.0894 0.3351 0.3301 0.2981 0.0056 0.0737 0.5163 1.1026 0.7466 I SUMMARY BY COUNTY -- ·1· I _ _ _ j _ __ - ARCHULETA ------1 BOULDER CHAFFEE ... CLEAR CREEK CONEJOS DELTA DOLORES EAGLE GARFIELD GILPIN GRAND GUNNISON HINSDALE JACKSON LA PLATA LAKE --- -- --- LARIMER MESA MINERAL MOFFAT MONTEZUMA OURAY I PARK PITKIN RIO GRANDE RIO BLANCO ROUTT SAGUACHE SANJUAN SANMIGUEL SARATOGA SUMMIT ---------·- ---------- . - - ---- ---------- - - - - - - - - - - -- - - - - - - - -- -------·- -- ----- - - ---- -------- - SAMPLE SIZE HARES PER HA - ----- - - - - - - - - - ----- - --------- - ~ - - --- 33 ~ - --- ----- - -------- ---- ----- i--------- -I------- - - - 6 -~ ------- 17 2 18 35 36 2 31 59 18 8 28 - 8 36 8 18 6 14 16 8 16 14 21 52 45 24 2 2 14 ----- 0.0438 0.1659 0.6117 0.3635 0.3642 0 0.6749 0.2941 0.3096 0.3010 0.1955 0.3163 0.9531 0.0769 0.3114 0.1931 0.2840 0.2152 0.7937 0.0874 0.0000 0.2864 0.4260 0.1393 0.9766 0.3316 0.1942 0.4143 0.5622 1.1828 0.1877 0.0888 �111 Review of the habitat assessment study of snowshoe hares by the Colorado Division of Wildlife • Shay Howlin Lyman McDonald West, Inc. 2003 Central Avenue Cheyenne, Wyoming 8200 I (307) 634-1756 February 14, 2000 In our examination of the issues surrounding the snowshoe hare pellet collection in 1998 we have identified 4 major concerns in the use of this data. Here we attempt to clarify these concerns and offer our analysis of their severity as they relate to the use of the pellet data as an index to sgowshoe hare abundance. CONCERNS OF REVIEWERS • Failure to validate or accurately duplicate Krebs' protocol This criticism is based on the premise that the division of wildlife (DOW) snowshoe bare pellet study intended to use the linear models developed by Krebs et al. (1996) to estimate snowshoe bare density based on the number of snowshoe hare pellets. Krebs' model for estimating the density of bares was developed in Canada with different snowshoe bare ecology and habitat. And the field protocol used by Krebs to develop the density estimation formula specified the removal of pellets from plots at a set time period before counting pellets. The DOW study protocol differed from Krebs' protocol in that they did not clear the study plots of old pellets berore conducting pellet surveys. The biologists conducting the surveys for the DOW were trained to age pellets and could identify last year versus current year pellets. In our view, the use of current year pellets can provide an index to abundance of snowshoe hares. The analysis of the DOW data does not need to validate or use Krebs' formula if only an index to abundance is attempted. • Failure to follow DOW survey protocol The DOW generated a random sample of survey points in each of the five study areas. The number of random points generated for the survey was roughly proportional to the size of the area. Recognizing the logistical problems associated with visiting all the random points, the DOW field crews reduced the number ofplots·to visit by selecting every fourth point from the original random sample (call this subset the reduced sample). DOW and BLM biologists surveyed additional points from the original random sample when conveniem. The resuiting dataset contains points in the reduced sample visited by the field crews, and points in the original sample visited by biologists opportunistically on the way to other points or other work. Since the resulting sample contains a �112 disproportionate number of points in block five, reviewers of the study have noted that the blocks with higher sampling intensity could result in increased chances of finding pockets of modest hare abundance. The fact that neither every point in the original random sample nor every point in the reduced sample was not surveyed resulted in a nonrandom sample of actual data. Biases associated with the selection of the plots sampled (i.e. closest to roads), makes the use of the data for an unbiased index to snowshoe hare abundance questionable. The tables below show the percent of original plots and the percent of the subset of plots that were surveyed in each block and were entered into the dataset called newform.xls (obtained from G. Byrne on December 28, 1999). Percentage of Krebs' plots sampled from original random sample. Block Number l 2 3 4 5 Number of Original Percent of plots sample size original visited sample 55 200 27.5 32 100 32.0 34 100 34.0 64 200 32.0 121 200 60.5 Percentage of Krebs' plots sampled from the reduced sample(every fourth plot from the ori inal sam le . Block Number l 2 3 4 5 Number of Reduced Percent of plots sample size reduced visited le 18 so 36.0 20 25 80.0 22 25 88.0 20 50 40.0 31 so 62.0 • Time frame of surveys The DOW conducted the snowshoe hare pellet surveys throughout the summer of 1998. One reviewer of the study contends that plots that were sampled later in the season had the potential to have more pellets, because there was more time for pellets to be dropped onto the plots and there are more hares due to the addition of hares to the population through parturition. From Figure I, it appears that each block had an even distribution of plots visited throughout the study period. If there was a time affect influencing the number of pellets on a plot, it appears it would have approximately the same influence in every block. • • Definition of suitable habitat The DOW protocol details the placement of the Krebs' plots once the randomly selected point is located. Plots were moved from the starting point to the nearest suitable �113 habitat (forest, willow, or mountain shrub habitat). A random number table was used to locate directions and distances to move the point to suitable habitat. If the first randomly selected direction or distance did not locate suitable habitat~ another direction and distance was chosen.: ' This protocol effectively reduces the area surveyed during the study to the "suitable habitat" in the study area. Field personnel made subjective determinations of suitable versus nonsuitable habitat, because a rigorous definition of suitable habitat is not easy to follow in the field. The actual size of the area that had a chance of being sampled is unknown and difficult to define. The size of this area could potentially be determined by the delineation of suitable habitat on the GIS vegetation coverage if definitions were consistent throughout the study. Figure I. Date of Division of Wildlife sampling of plots by block. - .. ..,..._ -······"· ---··- - - - - - ... - -· ... .. .. - ...... - - - - .. - ... -- • ~ 0 0 m 3 • • • • • •• --- - .. - • • 1 -..,_,-,-"T""""iy.::,MT""'T--,-T•-r-..--?-•-.-..-.-........,--,--..-,!!!•~-~H.;:,-~,-•::;,:-.y-r"•-.-'•--;..:;MT"""T"M~Nr'•'r••=-;:•=•MrrN'-re=,--r--,- Date CONCLUSIONS The failure of the DOW snowshoe hare study to validate or follow Krebs' protocol for the estimation of snowshoe hare density prohibits the use of the data in Krebs' estimation formulas. We do not feel this criticism discounts the use of the DOW data for an index of snowshoe hare abundance in the study area. The criticisms surrounding the time component of the study also do not seem to pose a critical problem when producing estimates of hare abundance by block. And if an estimate of suitable habitat can be made using the GIS coverage of the vegetation, the index of abundance can incorporate the amount of area surveyed. �114 Our review of the data revealed problems with the quality of the data. There were twelve missing dates in the dataset, and two dates that were 9 and 49 years before the bulk of the other observations. There were other simple errors in the plot identification variable that indicates the dataset was not verified. We recommend the data be checked for quality control-quality assurance before any analyses are conducted. If a truly random sample of Krebs' plots could be identified, we would recommend an analysis of the pellet data using only these plots. But neither all the points in the entire original sample were visited, nor were all the plots identified in the DOW biologists subset sample visited. We feel it is impossible to identify a sample of plots that were selected and visited that do not contain biases associated with the time and logistical constraints oflocating and sampling the pre-selected Krebs' plots. Therefore, we do not think this data is capable of producing an index of snowshoe hare abundance that is unbiased. The fieldwork associated with the collection of this data has provided information that can be used to design a study of snowshoe hare abundance that will be adequate to produce estimates by block. The study provides an estimate of the plot to plot variance in the number of pellets in the study area, as well as an idea of the number of randomly placed Krebs' plots that C&l1 be surveyed in a season. LITERATURE CITED Krebs, C.J., B.S. Gilbert, S. Boutin and R. Boonstra. 1987. Estimation of snowshoe hare population density from turd transects. Canadian Journal of Zoology. 65: 565567. � https://cpw.cvlcollections.org/files/original/6e3556eb1f4b3ee22038a42da446f254.pdf 50c421566fe5ea4f2eb4abb9de71bac1 PDF Text Text 113 Colorado Division of Wildlife Wildlife Research Report July 1998 JOB PROGRESS REPORT Colorado state of cost center 3430 Project No. _,_,W~--1~5~3--~R~--l_l~----- Mammals. Research Work Package No. _ _~P~B~B~O,_______ : Black-footed Ferret Recovery Task No. __....;.._ _,,,_________ Monitoring and Managing Disease in Black-footed Ferrets Period Covered: Author: July 1, 1997 - June 30, 1998. M.A. Wild and K. T. Castle Personnel: E. Wheeler. ABSTRACT We prepared a Program Narrative describing research to support black-footed ferret (Mustela nigripes) recovery efforts in Colorado. The black-footed ferret is a federally listed endangered species in the United States. Blackfooted ferrets have been extirpated from Colorado, but are scheduled to be reintroduced to the wild at the Little Snake Management Area (LSMA) in Moffat County, Colorado in 1999. Our research can be sub-divided into two broad sections: disease monitoring and flea control as a tool to manage sylvatic plague in prairie dogs and black-footed ferrets. Disease monitoring will be performed using collection of carnivores (primarily coyotes) from the LSMA twice per year. Carnivores will be examined post-mortem for signs of active disease and serology will be performed to test for exposure to diseases potentially threatening black-footed ferrets. Results of testing in FY97/98 show that toxoplasmosis and tularemia are present at LSMA, but potential impact on black-footed ferrets is unknown. Two potentially devastating diseases, canine distemper and plague, are also present at LSMA. Although prevalence of titers to canine distemper virus (CDV) in coyotes appears low at LSMA, an increasing trend.in prevalence is suggested .. Titers to plague (Yersina pestis) were found in 65% and 48% of coyotes tested in the summer and winter collections, respectively. Titers were present in both adult and juvenile animals, suggesting ongoing plague activity in some areas. The second section of our Program Narrative describes experiments with captive white-tailed prairie dogs to evaluate insect growth regulators (IGRs) on control of prairie dog fleas (genus Oropsylla). Preliminary work to establish a captive colony of >30 white-tailed prairie dogs and to develop techniques to rear Oropsylla in an insectary have been initiated this fiscal year. ��115 MONITORING AND MANAGING DISEASE IN BLACK-FOOTED FERRETS Margaret A. Wild and Kevin T. Castle P. N. OBJECTIVES 1. Monitor enzootic disease activity threatening survival of black-footed ferrets reintroduced into the LSMA. 2. Write a Final Program Narrative describing an experiment to develop and evaluate management techniques to minimize disease-related mortality of reintroduced black-footed ferrets. SEGMENT OBJECTIVES 1. Determine the level of activity of canine distemper virus, Aleutian disease, toxoplasmosis, tularemia, and leptospirosis in carnivores and plague in prairie dogs (using coyotes as sentinels) in the LSMA by sampling ~20 coyotes in summer and winter. 2. Write a Final Program Narrative describing an experiment to develop and evaluate management techniques to minimize disease-related mortality of reintroduced black-footed ferrets. METHODS AND MATERIALS Carnivore Disease Survey Infectious diseases can severely impact the success of black-footed ferret (Mustela nigripes) reintroduction efforts. As part of the black-footed ferret reintroduction protocol, we monitored disease activity in carnivores in the Little Snake Management Area (LSMA), Colorado. Coyotes (Canis latrans) were collected during a summer and winter sampling period. Post-mortem examination and sample collection was performed as described in the Planning Program Narrative (Wild 1997). In addition to serologic assay for plague (Yersina pestis) using the passive hemagglutination inhibition test (PHA/PHI), sera collected in the winter sampling were also tested using the plague ELISA test. Proposed Research Experiments We compiled data and performed literature searches and interviews to prepare a Program Narrative. We initiated work on Phase I--Pilot Study as per the Program Narrative and captured white-tailed prairie dogs using techniques described in the attached protocol (Attachment 1). Black-footed Ferret Reintroduction We assisted in preparation of the Black-footed ferret allocation request submitted to the US Fish and Wildlife Service by the Colorado-Utah blackfooted ferret recovery working team in 1997 and 1998. �116 RESULTS AND DISCUSSION Carnivore Disease Survey With the assistance of USDA Wildlife Services and the Bureau of Land Management (BLM) we collected 20 coyotes from the LSMA between 21-25 July 1997 and 21 coyotes between 10-11 February 1998. Coyotes were collected using a combination of calling and aerial gunning. Death occurred rapidly and the collection technique was adequate, but much less efficient in summer than when used in the winter months. Coyotes were collected from various locations in the >4700 mi 2 management area; however, we focused on the Powderwash area and near the site of the pre-conditioning pens. No lesions indicative of active disease were noted on gross examination of carcasses. We found no serologic evidence of exposure to leptospirosis (serovars canicola, grippe, hardjo, ictero, and pomona). Coyotes were not tested for Aleutian Disease (a mustelid disease). Serologic titers were positive to toxoplasmosis in two coyotes (one· adult, one juvenile) collected in the winter sampling. Fifty percent of the coyotes sampled during the summer had positive titers to tularemia, while <5% of those sampled in winter were positive. Positive titers could have been associated with predation on rabbits, prairie dogs, or other rodents or exposure to ectoparasites. Duration of titers to tularemia are likely of short duration (<6 mo). The impact of tularemia or toxoplasmosis on blackfooted ferrets is not currently known; however, ferrets and humans are susceptible to these diseases. Although a relatively small proportion of coyotes had positive titers (~1:16) to canine distemper virus (CDV), data collected over the last 18 mo suggest a trend toward increasing prevalence of CDV at the LSMA (Fig. 1). Because CDV is a serious disease threat to blackfooted ferrets, a priority in the coming year will be to monitor the trend of CDV prevalence. Likewise, plague activity on the LSMA is of concern. In the summer collection, both juvenile (7/13) and adult (5/7) animals showed evidence of exposure to plague (titers >1:8; Fig. 2). Seropositive juveniles indicates that some active plague is continuing in the area. Samples collected in the winter were tested using the standard PHA/PHI test and also an ELISA test. The ELISA test may be more specific than the PHA/PHI test; however, it has not been fully validated and accepted (H. Edwards, pers. commun.). Results of the standard PHA/PHI test indicate about half (10/21) of the animals samples had positive plague titers (Fig. 2). Alternatively, results of the ELISA test suggest that 81% (17/21) of animals samples were positive. Anecdotal evidence suggests that localized outbreaks of plague may be occurring in some areas of the LSMA, while plague activity may be lesser in the other areas (M. Albee, pers. commun.). This localization of plague activity is not uncommon. Proposed Research Experiments Given the results of these initial surveys, plague appears to be the most significant disease threat currently to black-footed ferrets reintroduced into the LSMA. Further, it is now more apparent that research emphasis needs to be placed on management and control of plague in prairie dogs and black-footed ferrets. The Program Narrative describing this research is attached (Attachment 2). To support this research, we captured 34 white-tailed prairie dogs from the Arapaho National Wildlife Refuge (ANWR), Colorado. Initially, four adult males were captured in April to evaluate methodology. Thiry juveniles were then captured in June. Spring/summer capture is required because white-tailed prairie dogs hibernate. Further, juveniles can be easily differentiated from �117 adults in early summer (June). One prairie dog died from trauma at FWRF, but all prairie dogs remained healthy during the quarantine period. The housing and care protocol (see Program Narrative) appears sufficient for maintaining prairie dogs. The white-tailed prairie dogs are more fractious than anticipated; however, training continues in an attempt to habituate the animals to handling. We also initiated investigation of maintaining prairie dog fleas (.Oropsylla spp.) in artificial insectaries. Flea mortality has been high immediately after capture from ANWR and transport to Fort Collins. We will continue to modify our protocols to increase survival. Fleas that survive transport have been successfully maintained in the rearing medium. Pinky (neonatal) mice are provided to the fleas for a 24-hr period every 2-3 days as a blood source. Fleas have been observed to lay eggs, but the numbers (a few/day) are, as expected, low. Black-footed Ferret Reintroduction The U.S. Fish and Wildlife Service approved an allocation of black-footed ferrets for reintroduction into the LSMA in fall 1997; however, due to delay in publishing the final rule on experimental designation of the joint Colorado-Utah reintroduction sites, black-footed ferrets were not received this year. The Colorado-Utah site has received conditional approval for allocation of 20 black-footed ferrets later in 1998. LITERATURE CITED Wild, M.A. 1997. Monitoring and managing disease in black-footed ferrets. Colorado Div. Wildl. Res. Rep., 0880-1, Jul 1996 - Jun 1997, Fort Collins. Prepared by M~tJLJ Magaret A. Wild Wildlife Researcher ��119 ATTACHMENT 1 PROTOCOL FOR LIVE-TRAPPING PRAIRIE DOGS Prepared by: Kevin T. Castle, M.S. 17 March 1998 The following procedures are based on my personal experience with trapping a variety of mammals, and on the prairie dog trapping experiences of Hoogland (1995) and A. Schwartz (pers. commun.). Hoogland reported only 12 trap-related deaths in over 10,000 captures, and Schwartz observed only 5 deaths in approximately 1,700 captures; the average mortality rate is approximately 0.1%, so trapping mortality is unlikely if these procedures are followed. If an animal is severely injured (fracture, severe laceration), such that its ability to survive is markedly impaired, or it would suffer undue misery and stress, it will be euthanized. Euthanasia will be performed using an overdose of inhalant anesthetic, by placing the prairie dog in an induction chamber with a halothane-soaked gauze. If two trapping-related deaths occur, trapping will cease and the protocol will be reviewed and modified. Minor injuries or metabolic imbalances will be treated by, or under the supervision of, a veterinarian. One or two single or double-door Tomahawk traps (6" x 6" x 24") are placed near active burrows; single door traps seem to selectively capture juvenile prairie dogs, whereas double-door traps will capture both juveniles and adults. Traps are baited with rolled oats and/or "horse sweet mixture"; horse sweet mixture is sticky, and therefore tends to stay in place, rather than roll or blow away. During a 1-2 week "prebaiting" session, the traps are wired closed (to prevent unwanted, accidental entry), and bait is spread around the traps and burrows. Prebaiting allows the prairie dogs to become accustomed to having strange objects in their environment, and lets them associate the traps with the presence of food. In addition, the traps "weather" and take on a more natural scent; this is especially important for new, unused traps that may smell of oils used in the manufacturing process. After the prebaiting session, traps are set (opened and baited) just prior to sunset, and checked (minimally) at noon and at dusk the following day. It is better to set the traps at night rather than in the morning, to avoid disturbing any animals that become active early. Once disturbed, it may take from 15 minutes to several hours for an animal to re-emerge from its burrow, and any other animals may become more wary. Prairie dogs are diurnal, and are typically not seen above ground in inclement weather, so it is unlikely that any will be captured at night or when temperatures are very cold or very hot. Nonetheless, traps are not set during rain or snow, or when daytime ambient temperature is expected be below 20°F or above 90°F. Prairie dogs should be handled by experienced mammal trappers, wearing heavy leather or welder's gloves. Upon capture, each animal is placed into a coneshaped nylon or canvas bag for non-chemical immobilization. They are sexed and weighed, and age is determined (Hoogland 1995; cox and Franklin 1990). A numbered monel ear tag is attached to each animal. Released animals should be tagged also, to minimize the handling of animals recaptured at a later date. Animals are then transferred to individual mesh-bottom holding cages (minimum size= 18" x 18" x 18"), and the cages are covered for transport. At least part of the holding cage should remain covered at all times, to provide a hiding place for the animal; if possible, a nest box containing bedding material (e.g. "Bed-o'-cobs", care Fresh, hay) should be provided. Rodent chow, �120 alfalfa cubes, and water are provided ad lib. if transport/holding time will exceed 2 h. Animals held in the field should be kept indoors if possible, but minimally need to be protected from the elements and from potential predators. They should be checked every 3-4 h the first 1-2 days after capture, to assess health and acclimation to captivity, and at least once per day thereafter. Literature Cited: Hoogland, J. L. 1995. The black-tailed prairie dog: social life of a burrowing animal. University of Chicago Press, Chicago, 557 pp. �121 ATTACHMENT 2 PROGRAM NARRATIVE State of Colorado cost center 3430 Project No. W-153-R Mammals Program Black-footed Ferret conservation Work Package No._-P-8~8~0,..__ _ _ __ Task No. ______________ : Monitoring and Managing Disease in Black-footed Ferrets A. NEED The mission of the Colorado Division of Wildlife (CDOW) is to "perpetuate the wildlife resources of the state and provide people the .opportunity to enjoy them" (State of Colorado, 1994). Further, the CDOW Long Range Plan established a goal directing the Division to "cooperate with federal, state, county and local government agencies, private landowners and other organizations in evaluating the potential for restoration of extirpated species" (State of Colorado, 1994; Goal 8). The black-footed ferret (Hustela nigripes) is a federally listed endangered species in the United States. Black-footed ferrets have been extirpated from Colorado, but are scheduled to be reintroduced to the wild at the Little Snake Management Area (LSMA) in Moffat County, Colorado in 1999. Coyote Basin, Utah and White River Resource Area, Colorado are also under consideration for blackfooted ferret reintroduction efforts in the near future. The overall goal of ferret reintroduction is to contribute to the national black-footed ferret recovery plan objectives (U. s. Fish and Wildlife Service, 1988). The specific reintroduction objective for LSMA is to sustain a natural breeding population of 30 adult black-footed ferrets. success of ferret reintroduction programs in Wyoming, south Dakota, and Arizona have been hampered by multiple factors. Mortality rates, primarily from predation and disease, have been high (D. Biggins, pers. commun.; P. Marinari, pers. commun.). It has become apparent that improved methods for preconditioning ferrets prior to release, predator management, and disease management are prerequisite to success of future reintroduction efforts. Here, we will begin investigating the potential for impact of diseases on black-footed ferrets in the LSMA reintroduction site. Disease, primarily canine distemper and sylvatic plague, has been a significant factor limiting captive as well as free-ranging black-footed ferret populations. Mortality of black-footed ferrets from infection with canine distemper or sylvatic plague is extremely high (Williams et al., 1988; Williams et al., 1994). Further, mortality from sylvatic plague in prairie dogs can markedly impact prey base for black-footed ferrets. Before black-footed ferrets are reintroduced to the site, data on these potentially devastating diseases must be collected and unique plans for their management devised. B. OBJECTIVES 1. Monitor enzootic disease activity threatening survival of black-footed ferrets reintroduced into the LSMA. �122 2. C. Develop techniques to manage plague in the LSMA using insect growth regulators applied orally to prairie dogs. EXPECTED RESULTS AND BENEFITS Disease Monitoring Improved techniques for disease management will likely be required for the successful reintroduction of the extirpated black-footed ferret to Colorado. We must monitor enzootic disease activity to determine immediate risks to black-footed ferrets. Monitoring will also provide information on the epizootiology of diseases that will assist in formulating management plans for black-footed ferrets and may also aid in understanding population dynamics of other inhabitants. of the short grass prairie. Performance Indicator: Monitor disease status of ~40 carnivores in the LSMA annually and based on these findings make disease management recommendations to the Colorado-Utah black-footed ferret recovery working team. Flea control We believe that the proposed research will provide a novel and effective means of controlling plague in prairie dogs and black-footed ferrets. Sylvatic plague is transmitted primarily via the bite of infected fleas. Control of flea populations can be used to break the cycle and reduce the occurrence of disease. Products, such as lufenuron and pyriproxyfen, administered orally to pet animals have effectively controlled fleas (Blagburn et al, 1994; Hink et al., 1994; Palma et al., 1993). If these products are efficacious and safe in prairie dogs, and to black-footed ferrets which consume treated prairie dogs, then they could be used to control fleas, and plague, at reintroduction sites. Further, these products could be used as an alternative method of wildlife management in areas where sylvatic plague is a threat to public health. Performance Indicator: Provide a novel technique for the management of plague in the LSMA to protect the endangered black-footed ferret. D. APPROACH Disease Monitoring Rationale: Infectious diseases can severely impact the success of black-footed ferret reintroduction efforts. Canine distemper and sylvatic plague pose the greatest threats to black-footed ferret survival (Williams et al, 1988; Williams et al, 1994). Wild canids and mustelids, as well as domestic dogs, are susceptible to canine distemper and can serve as reservoirs for infection (Williams, 1982). Plague is maintained primarily in rodent populations, such as prairie dogs (Quan, 1982). Therefore, prior to reintroduction, status of diseases such as canine distemper and plague in sympatric wildlife populations must be determined to assess the risk to black-footed ferrets. We will monitor activity of canine distemper in carnivores, primarily coyotes (Canis latrans), by determining serologic titers and performing postmortem examination. Titers are useful in ascertaining if carnivores have recently been exposed to the virus. Postmortem examination is required to determine if carnivores have active infection with canine distemper. We will monitor plague activity in prairie dogs using coyotes as sentinels. Although coyotes are rarely involved in plague transmission, serologic testing of coyotes for titers to plague can �123 help define the extent of plague activity in rodent prey populations (Thomas and Hughes, 1992). Moreover, by sampling juvenile animals in the summer, recent plague activity can be identified. This survey of carnivores, combined with collaborative work by the Bureau of Land Management investigating changes in prairie dog numbers, will provide important information on activity of diseases that are potentially devastating to black-footed ferrets. Disease management activities will be driven by these findings. Methods: We will use survey of carnivores as a tool to determine status of canine distemper and sylvatic plague in LSMA. Surveys will be performed annually in January-March and July-September. Each survey will consist of a minimum sample of 20 coyotes. Aerial gunning will be used as the primary means of sample collection. Summer collections will focus on juvenile animals. Animals will be located by personnel in a fixed wing aircraft. Personnel from USDA Wildlife Services skilled in coyote control will fly over and shoot animals at close range (about 10-45 m) using a 12 g shotgun loaded with copper coated BB's or #4 buck shot. Multiple shots will be fired at the head of the animal. Ground crews will locate and recover the animal carcass. If the animal was not fatally wounded, ground crews will euthanatize animals with a shot to the neck or chest. Location of carcass collection will be recorded within 1/4 section. Surveys will be supplemented with additional samples collected opportunistically from carcasses of coyotes and other carnivores, including badger (Taxidea taxus), red fox (Vulpes vulpes), and striped skunk (Mephitis mephitis), killed as a result of hunting or by unintentional vehicular collision. Diagnostic techniques will include collection of blood samples for serology, collection of lower jaw for age determination, and postmortem examination to detect evidence of active disease. Serum will be harvested from blood samples, frozen, and submitted to Wyoming state Diagnostic Laboratory (WSVL, Laramie) for determination of titers to canine distemper using the serum neutralization test and to plague (Yersinia pestis) using the passive hemagglutination inhibition test. Titers to toxoplasmosis, tularemia, Aleutian disease (mustelids only), and leptospirosis will also be determined using samples collected during the initial carnivore survey. Gross necropsy will be performed by or under the direct supervision of a veterinarian. Routine tissue collection will include: samples of gross lesions, kidney, urinary bladder, lung, and brain placed in 10% buffered formalin; samples of lung, small intestine, and brain frozen; and samples of lesions stored fresh for immediate analysis. Samples of parasites will also be collected opportunistically; however, all badgers with dermatitis will be sampled for Filaria taxideae. Fixed tissues will be submitted to Dr. E. s. Williams, WSVL for histologic examination. Frozen samples will be stored by Colorado Division of Wildlife for analysis as needed (i.e., as indicated by histologic·or serologic findings). Fresh samples will be submitted to WSVL or Colorado State Veterinary Diagnostic Laboratory (CSVDL, Fort Collins). Parasites will be stored and submitted to WSVL or CSVDL as needed. Additional blood samples will be collected for serology as available from carcasses harvested by local hunters and from unintentional vehicular collision. Serum samples will be handled as previously described. Analysis: A titer of ~1:16 will be considered positive for antibodies against canine distemper virus or for antibodies against Yersinia pestis. We will use Fisher's exact test (or extensions) to statistically analyze the prevalence of antibodies among age classes and over time. �124 Flea Control Rationale: Mortality of black-footed ferrets from infection with sylvatic plague is extremely high (Williams et al., 1994). Further, mortality from sylvatic plague in prairie dogs can markedly impact prey base for black-footed ferrets. Plague has severely hampered reintroduction efforts in other states (P. Marinari, pers. commun.) and preliminary data suggest that plague activity is present in some areas of the LSMA (M. Albee, unpub. data). Successful reintroduction will likely require management techniques to control plague. Plague i~ maintained primarily in rodent populations, such as prairie dogs, where it is transmitted via the bite of an infected flea. Fleas of the genera Opisocroscis and Oropsylla have been associated with plague transmission in prairie dogs (Ubico et al., 1988). Insecticide dusts have been applied to prairie dogs burrows in an attempt to kill fleas and thus control plague. Carbary! has been the most commonly used insecticide; however, application is labor intensive and activity is short-lived (Barnes, 1993). Permethrin has been shown effective up to 84 days after application, but the authors warn that application at the recommended dosage rate would be highly laborious (Beard et al., 1992). Recently developed compounds used to control fleas in pet animals offer a promising alternative to insecticide dusts. Although topically applied flea adulticides, such as imidacloprid, are not feasible for use in wildlife, insect growth regulators (IGRs) with ovicidal and/or larvicidal activity could be delivered orally to free-ranging animals via bait. The IGR lufenuron is a benzoylphenylurea derivative which inhibits formation of chitin in the exoskeleton of insects (Cohen, 1987). A single oral dose of lufenuron has been shown effective in controlling the cat flea (Ccenocephalides felis felis) for at least 30 days in treated dogs (Hink et al., 1994) and cats (Blagburn et al., 1994). Davis (1997) reported a significant reduction in fleas on ground squirrels (Spermophilus beecheyi) that had been treated with lufenuron. Another IGR, pyriproxifen, exerts activity by mimicking natural insect juvenile hormones. Pyriproxifen has been shown effective in vitro in inhibiting normal egg production for 80 hr post-treatment (Palma et al., 1993) and may have other longer lasting means of control as well (T. Miller, pers. commun.). If these compounds proved efficacious over a period of time(~ 1 mo) when administered orally to prairie dogs, they could be applied over large areas of prairie dog habitat via a treated bait. However, prior to management application, the products should be evaluated in a controlled laboratory setting to determine an effective dose, duration of efficacy, product safety, and to formulate an acceptable bait carrier. If results of laboratory tests are positive, the products should be tested in a controlled field study to determine efficacy under natural conditions. Without this evaluation, the management potential of IGRs for controlling plague and protecting the health of the endangered blackfooted ferret, as well as human health, will be difficult to discern. Our work will be divided into three phases. Phase I will be pilot work to establish a colony of captive prairie dogs and to develop an environment suitable for flea survival and reproduction both on the prairie dogs (and in their bedding) and in an artificial medium. Before we can evaluate the efficacy of IGRs, we must be sure that fleas will survive and reproduce on our captive prairie dogs. Further, we will need to artificially rear prairie dog fleas for infestation of prairie dogs and assessment of egg viability. These techniques are not currently available (Barnes, 1993). Phase II will evaluate efficacy and safety of lufenuron and pyriproxifen fed to captive prairie dogs. Finally, in Phase III we will evaluate potentially efficacious IGRs (from phase II) in a limited field application. White-tailed prairie dogs (Cynomys leucurus) will be used as the study animal because this is the species of prairie dog present at the LSMA black-footed ferret reintroduction site. �125 Phase !--Pilot Study Methods: Capture of prairie dogs. White-tailed prairie dogs will be captured at the Arapaho National Wildlife Refuge, Colorado. One or two single or double-door Tomahawk traps (15 cm x 15 cm x 60 cm) will be placed near active burrows. Traps will be baited with horse sweet mixture (sweetened grain mixture). During a 1-2 week prebaiting session, the traps will be wired closed (to prevent unwanted, accidental entry), and bait will be spread around the traps and burrows. After the prebaiting session, traps will be set (opened and baited) just prior to sunset, and checked (minimally) at noon and at dusk the following day. Prairie dogs are diurnal, and are typically not seen above ground in inclement weather, so it is unlikely that any will be captured at night or when temperatures are very cold or very hot. Nonetheless, traps will not be set during rain or snow, or when daytime ambient temperature is expected be below -6 C or above 32 c. Prairie dogs will be handled by experienced mammal trappers, wearing heavy leather or welder's gloves. Upon capture, each animal will be physically restrained in a cone-shaped nylon or canvas bag. They will be sexed and weighed, and age determined (Hoogland, 1995; Cox and Franklin, 1990). A numbered monel ear tag will be attached to each animal. The animals will then be transferred to an individual mesh-bottom holding cage (minimum size= 45 cm x 45 cm x 45 cm) with nest box and covered for transport. Feed (e.g., rodent chow, alfalfa cubes) and water will be provided ad lib if transport/holding time will exceed 2 hr. care and housing of prairie dogs. Prairie dogs will be maintained indoors in individual cages (45 cm x 90 cm x 45 cm) with nest boxes at the Foothills Wildlife Research Facility (FWRF). Animals will be quarantined for 14 days upon return to FWRF to assess disease (plague) status. Due to our study needs, prairie dogs will not be treated to control external parasites. During the quarantine period, all personnel will be required to wear room-specific gloves and coveralls when in the room, whether observing animals or cleaning cages. Personnel will be informed of the clinical signs of plague in prairie dogs and humans and instructed to monitor themselves and their clothing for fleas. If plague is diagnosed in a prairie dog, all animals in the building will be euthanized with an overdose of halothane inhalant anesthetic. Our custom cage design is based on specific study needs (providing the prairie dogs with sufficient living space, while allowing us to rear and collect fleas), discussions with M. Metzger, and the published work of Hilton (1971). Individual prairie dog cages will be constructed of wood and wire fencing, with "no-see-um" netting over the top and covering potential flea escape routes. Half of the cage serves as a nest box, and the other half serves as a feeding area. Two hinged doors allow access to each half of the cage for cleaning or animal handling. The nest box is a 40 x 28 x 23 cm plastic container fitted with a wooden lid. The nest box has a 10 cm diameter hole in one wall to provide access to the feeding area; a 10 cm diameter piece of PVC tubing joins the nest box to the feeding area. The access hole can be covered to confine the animal to either side of the cage as necessary. Previous work on ground squirrels (Hilton, 1971) suggests that the animals are unlikely to defecate/urinate in the nest box, and Hoogland (1995) noted that there was little fecal matter in the burrows and nests of black-tailed prairie dogs he studied. Our preliminary work shows that defecation in the nest box is minimal, and that urination is very rare; we therefore do not expect the nest boxes to become soiled during experiments. �126 The floor of the feeding area consists of 1 cm mesh wire fencing, to allow feces, urine, and spilled food to fall through to a pan below. This mesh size prevents the animal's feet from becoming wedged in the floor, but may encourage the animal to spend most of its time in the nest box when not feeding. It is important for us to keep the animal in the nest box as much as possible, in order to enhance flea reproduction; many fleas of the genus Oropsylla are "nest fleas" and are associated with the host's body only when feeding (Hilton, 1971; M. Metzger, pers. comm.). Confining the prairie dogs and fleas to the nest box will also facilitate collection of fleas and.their eggs. A food dish and water bottle will be wired to the side of the feeding area. A 2.5 cm deep, removable, galvanized sheet metal pan will be placed under the entire cage to catch food, feces, urine, and any fleas that escape the nest box. All prairie dogs will be checked at least once per day. Fresh feed (rodent chow, alfalfa cubes) and water will be provided daily, and cage bottoms will be cleaned of feces each day. Pans will be cleaned and bedding will be replaced as needed, but a minimum of 2 times per week. Rooms (or nest boxes) will be maintained at approximately 24 C and 75% relative humidity to ensure flea survival and reproduction (M. Metzger, pers. commun.; Metzger and Rust, 1997). Two south-facing windows in the animal room should provide sufficient light for these fossorial animals, and will allow us to maintain the animals on a natural photoperiod. We do not anticipate any prairie dog mortality due to capture or experimental procedures. If an animal becomes sick or injured, and recovery is not likely, it will be euthanized with an overdose of halothane inhalant anesthetic. Pilot study. We will capture and house 3-5 prairie dogs for initial evaluation of our methods. Prairie dogs will be observed daily and weighed at least once per week, to help assess health and acclimation to captivity. If two or more animals do not adjust to our captive conditions (they refuse food and/or water or appear sick) we will review and modify our procedures. All prairie dogs that die in captivity will undergo a full post-mortem examination. Minor adjustments to husbandry may be made as needed to meet animal needs. One objective of the pilot study is to determine if prairie dogs can be trained to allow repeated grooming by handlers, without being chemically immobilized. Hoogland (1995) successfully groomed black-tailed prairie dogs without chemical immobilization, and one of us (KTC) repeatedly handled captive black-tailed prairie dogs that were distracted with a treat, such as a carrot. We will therefore attempt to handle each animal on a daily basis, in order to establish a routine whereby they become accustomed to being groomed. If we can train the animals to allow grooming, chemical immobilization will not be necessary. If chemical immobilization appears necessary, we will test the following methods to determine which to use in Phase II: 1) animals will be placed in an induction chamber and anesthetized with isoflurane. The animals will be removed from the chamber within 15 seconds after cessation of movement and anesthesia will be maintained via a mask; or 2) animals will be placed into a restraining cone and injected intramuscularly (IM) with a combination of ketamine (30 mg/kg) and diazepam (0.5 mg/kg) (Kreeger, 1997). During anesthesia, animals will be monitored visually for respiration, heartbeat, and signs of distress. Trained or immobilized prairie dogs will be groomed with a fine-toothed flea comb for 3 minutes; grooming will take place up to 3 times per week. After immobilization, each animal will be returned to its cage, and closely observed until it regains normal functions. �127 Concurrent to establishing the captive prairie dog colony, we will develop a technique to artificially rear prairie dog fleas of the genus Oropsylla. Although prairie dogs can harbor several genera of fleas, we have selected to maintain only Oropsylla because they are among the most common fleas of prairie dogs and they have been implicated in the transmission of plague (Ubico et al., 1988). Fleas will be collected by combing trapped prairie dogs and by swabbing of burrows (Barnes et al., 1972). Fleas will be placed in 500 ml glass jars each containing about 125 g of larva-rearing media consisting of wheast (Red Star Biologicals), dried beef blood (Monfort Biologicals), powdered dog chow, and sand. In the laboratory, live fleas will be anesthetized with CO2 and identified by comparison to a reference collection of preserved fleas. Because fleas are sensitive to changes in temperature and humidity, they will be maintained in an incubator at about 23-24 c and ~70% relative humidity (Metzger, pers. commun.). Appropriate relative humidity will be maintained using a super-saturated solution of potassium chloride (Winston and Bates, 1960). A natural photoperiod will be approximated within the incubator through the use of fluorescent lights and a timer. Fleas will be monitored daily for survival and reproduction. Modifications will be made as necessary to optimize flea performance. To determine adequacy of the cage environment for flea survival and reproduction, we will infest prairie dogs with about 20-50 male and female fleas. We will collect and count fleas from prairie dogs and their bedding at weekly intervals. Fleas will be collected from prairie dogs by combing (while distracted with a treat or while under anesthesia) for 3 min and from bedding by sifting and visual inspection. Fleas will be returned to the prairie dog after counting. Collections, and if needed reinfestations, will be continued until successful colonization occurs. If a severe allergic reaction or other health problem associated with flea infestation occurs, the prairie dog will be removed from the study and provided veterinary care or euthanized. Analysis: Evaluation of husbandry techniques and artificial flea rearing procedures will be subjective. No statistical analyses will be performed. Phase II--Laboratory Evaluation of IGRs Methods: Prairie dogs will be captured and maintained in captivity as described in Phase I. Laboratory evaluation of pyriproxifen and lufenuron will be divided into four experiments--efficacy dose titration, efficacy of controlling flea infestations in a simulated burrow environment, product safety, and bait formulation. Detailed Study Plans will be written for each of these experiments based on results of the Pilot Study; however, general procedures will be as follows. Efficacy dose titration. In this experiment we will determine an effective dose of pyriproxifen and of lufenuron to interrupt reproduction in prairie dog fleas and monitor duration of the products' efficacy. Prairie dogs (n = 21) will be blocked by sex and randomly divided into two treatment groups and a control group. In the first experiment, lufenuron will be evaluated at two dose rates. Treatment group 1 (n = 7) will receive 15 mg/kg and treatment group 2 (n = 7) will receive 45 mg/kg lufenuron PO. The control group (n = 7) will receive excipient suspension without lufenuron PO. After a sufficient washout period, the experiment will be repeated using pyriproxifen at anticipated dosages of 50 mg/kg and 100 mg/kg PO for treatment groups 1 and 2, respectively. Prairie dogs will be infested with fleas prior to treatment and periodically after treatment to maintain flea infestations. We will collect flea eggs prior to treatment and at 5 weekly intervals and determine egg �128 viability by rearing in artificial media. Number of eggs hatching in a specific time period (to be determined in Phase I) will be determined and compared between groups and over time. We may also collect blood from prairie dogs during the weekly samplings if an assay to determine serum levels of test product can be developed. Development of a successful artificial rearing system for prairie dog fleas is prerequisite to the performance of this experiment. If such a protocol is not developed, we will attempt to determine an effective dose using the methods described below for evaluation of efficacy in a simulated burrow environment. In this case, duration of efficacy will be more difficult to determine. Efficacy of controlling flea infestations in a simulated burrow environment. If results of the efficacy dose titration experiment suggest that a single oral dose of pyriproxifen and/or lufenuron is effective in controlling reproduction of prairie dog fleas for at least 4 wk, we will further evaluate the product(s) in a simulated burrow environment. The purpose of this experiment is to evaluate the ability of the test products to control prairie dog flea populations over an extended period. Prairie dogs (n = 20) will be blocked by sex and divided into treatment and control groups. They will be infested with fleas and treated monthly with pyriproxifen or lufenuron. Adult fleas on the animal and in the cage will be counted at 1-2 wk intervals for 12 wk. After counting, fleas will be returned to the animal. Number of fleas recovered will be compared between treatment groups and over time. Product safety. We will determine safety of an overdose of pyriproxifen and/or lufenuron. Prairie dogs (n = 20) will be divided into a treatment and control group. Prairie dogs in the two groups will be maintained similarly except the treatment group will receive an amount of product equal to that contained in the maximum calculated daily intake of treated bait. Overdose feeding will continue for a 3 day period. Health, attitude, and weight change of prairie dogs in the treatment and control groups will be monitored during, and for 1 wk following, this treatment. Based on type of action, these products should be safe in mammals (Blagburn et al., 1995; T. Miller, pers. commun.); however, if animals become moribund, they will be euthanatized. All dead animals will receive a complete postmortem examination. If the overdose of test product is safe in prairie dogs, product safety will be tested in black-footed ferrets or black-footed ferret x Siberian polecat hybrids. Ferrets maintained at other research facilities will be used as test animals and will be divided into a treatment and control group. Prairie dogs treated with three times the maximum daily dosage of test product (as described above) will be fed to ferrets in the treatment group for 7 days. Health, attitude, and weight change of ferrets in the treatment and control groups will be monitored during, and for 1 wk following, this treatment. Bait formulation. When an effective dose of pyriproxifen and/or lufenuron is determined, we will begin investigating formulation of a bait carrier. Because of differences in the chemical properties of the test products, unique formulations will likely be required for each product. Due to its high palatability, we will initially attempt to incorporate the test product into horse sweet feed (sweetened grain mixture); however, incorporation into a pelleted feed may be required to adequately deliver the product. Palatability of treated bait will be evaluated in feed-deprived (hungry) prairie dogs as well as those fed ad libitum maintenance diets (satiated). Blood samples may be collected from prairie dogs to determine serum level of test product after bait consumption. �129 Analysis: Efficacy dose titration. Efficacy of test product will be determined by comparing developmental success of eggs produced by fleas exposed to treated prairie dogs vs. eggs produced by fleas exposed to control group prairie dogs. These formulas will be used in efficacy determination: Developmental success= number emergent adult fleas x 100 number eggs collected Percentage efficacy= mean developmental success (control)- mean develqpmental success (treated) x 100 mean developmental success (control) Mean developmental success values for eggs collected from prairie dogs in each of the treatment groups will be analyzed using analysis of variance. Efficacy in a simulated burrow environment. Efficacy of test product will be determined by comparing number of fleas recovered from treatment and control prairie dogs and their cages on each sampling day using analysis of variance. Significant reductions in numbers of fleas on treated prairie dogs would signify interruption of the flea life cycle attributable to treatment with the test product. Product safety and Bait formulation: Evaluation of product safety and acceptability of bait will be subjective. No statistical analyses will be performed. Phase III-Field Evaluation of IGRs Methqds: Based on results of laboratory evaluations, we will evaluate pyriproxifen and/or lufenuron in a limited field application. Only product(s) that significantly reduced flea numbers on treated hosts for ~1 mo and that showed no adverse effects on captive prairie dogs and ferrets will be field tested. We will identify similar, yet geographically distinct, colonies of healthy prairie dogs for evaluation of the test product(s). Treatment colony(s) will receive bait fortified with the test compound while the control colony will receive bait only. Treatments will be applied at monthly intervals during the summer. Observations will be made to determine prairie dog numbers and consumption of bait by prairie dogs and non-target species. Flea indices (number of fleas per individual and number of fleas per burrow) will be determined by counting and identifying fleas collected by combing live-trapped prairie dogs and by swabbing burrows (Barnes et al., 1972). Blood samples may also be collected from prairie dogs to determine serum level of test product. Collections will be made at monthly intervals beginning ~l mo prior to initiation of treatment, continuing through the active season of prairie dogs (September), and resuming the following summer. Flea indices will be compared between treatment and control colonies and over time. Analysis: Efficacy of test product will be determined by comparing flea indices of control and treatment colonies at each sampling using analysis of variance. Significant reductions in numbers of fleas on prairie dogs at treated sites would signify interruption of the flea life cycle attributable to treatment with the test product. �130 schedule: Objective 1. Disease survey 2. Phase I--Pilot study 3. Phase II--Laboratory study 4. Phase III- Field study 5. Analyze data, report results E. Fiscal Year 1998-2002 1998-1999 1999-2000 2000-2001 2001-2002 LOCATION Planning, captive animal research, data analysis, and reporting will be carried out at the COlorado Division of Wildlife's Foothills Wildlife Research Facility, 4330 w. Laporte Ave., Fort Collins, Colorado. Carnivore disease survey will be conducted at the Little Snake Management Area, Moffat County, Colorado. Capture of prairie dogs to establish a captive colony will be at Arapaho National Wildlife Refuge, Jackson County, COlorado. Location of Phase III (field study) is yet to be determined, but will be reported in the detailed Study Plan for that experiment. F. ESTIMATED COST Fiscal Year Estimated costs 1998 1999 2000 2001 2002 G. $13,000 20,500 21,400 21,400 13,000 llE ReW,1irements PFTE 0.5 0.5 0.5 0.5 0.5 PERSONNEL Margaret A. Wild1 Kevin T. Castle1 Thomas A. Miller2 Craig Parks3 co-principal investigator Co-principal investigator CO-investigator CO-investigator Colorado Division of Wildlife, Fort Collins, COlorado Virbac, Inc., Fort Worth, Texas 3 Novartis Animal Health U.S., Greensboro, North Carolina 1 2 TFTE 0.25 0 0.5 0.5 0 �131 H. LITERATURE CITED Barnes, A. M. 1993. A review of plague and its relevance to prairie dog populations and the black-footed ferret. Proceedings of the symposium on the management of prairie dog complexes for the reintroduction of the blackfooted ferret. J. L. Oldemeyer, D. E. Biggins, B. J. Miller, and R. Crete, eds. 96pp. Barnes, A. M., L. J. Ogden, and E.G. Campos. 1972. Control of the plague vector, Opisocrostis hirsutus, by treatment of prairie dog (Cynomys ludovicianus) burrows with 2% carbaryl dust. Journal of Medical Entomology 9: 330-333. Beard, M. L., s. T. Rose, A. M. Barnes, and J. A. Montenieri. 1992. Control of Oropsylla hirsuta, a plague vector, by treatment of prairie dog burrows with 0.5% permethrin dust. Journal of Medical Entomology 29: 25-29. Blagburn, B. L., J. L. Vaughan, D.S. Lindsay, and G. L. Tebbitt. 1994. Efficacy dosage titration of lufenuron against developmental stages of fleas (Ctenocephalides felis felis) in cats. American Journal of Veterinary Research 55: 98-101. Blagburn, B. L., C. M. Hendrix, J. L. Vaughan, D. s. Lindsay, ands. H. Barnett. 1995. Efficacy of lufenuron against developmental stages of fleas (Ctenocephalides felis felis) in dogs housed in simulated home environments. American Journal of Veterinary Research 56: 464-467. Cohen, E. 1987. Interference with chitin biosynthesis in insects. In Chitin and benzoylphenyl ureas, series entomologica, vol. 38, J.E. Wright and A. Retnakaran (eds), Dr. W. Junk, Publishers, Boston, pp. 33-42. Cox, M. K. and W. L. Franklin. 1990. Premolar gap technique for aging live black-tailed prairie dogs. Journal of Wildlife Management 54:143-146. Davis, R. M. 1997. Use of an orally administered insect development inhibitor (lufenuron) as a flea control agent in the California ground squirrel, Spermophilus beecheyi. Fourth International Symposium on Ectoparasites of Pets, pp 31-35. Hilton, D. F. J. 1971. A method for rearing fleas of ground squirrels. Transactions of the Royal Society of Tropical Medicine and Hygiene 66: 188189. Hink, w. F., M. Zakson, ands. Barnett. 1994. Evaluation of a single oral dose of lufenuron to control flea infestations in dogs. American Journal of Veterinary Research 55: 822-824. Hoogland, J. L. 1995. The black-tailed prairie dog: social life of a burrowing animal. University of Chicago Press, Chicago, 557 pp. Kreeger, T. J. 1997. Handbook of Wildlife Chemical Immobilization. International Wildlife Veterinary Services, Inc., Laramie, WY. 342 pp. Metzger, M. E. and M. K. Rust. 1997. Effect of temperature on cat flea (Siphonaptera: Pulicidae) development and overwintering. Journal of Medical Entomology 34: 173-178. �132 Palma, K. G., s. M. Meola, and R. w. Meola. 1993. Mode of action of pyriproxifen and methoprene on eggs of Ctenocephalides felis (Siphonaptera: Pulicidae). Journal of Medical Entomology 30: 421-426. Quan, T. J. 1982. Plague. In Diseases of wildlife in Wyoming, E.T. Thorne, N. Kingston, w. R. Jolley, and R. c. Bergstrom (eds), Wyoming Game and Fish Department, Cheyenne, pp. 67-71. Rust, M. K. 1992. Influence of photoperiod on egg production ·of cat fleas (Siphonaptera: Pulicidae) infesting cats. Journal of Medical Entomology 29: 242-245. state of Colorado. 1994. Long Range Plan. Division of Wildlife. 32pp. Department of Natural Resources, Thomas, c. U., and P. E. Hughes. 1992. Plague surveillance by serological testing of coyotes (Canis latrans) in Los Angeles County, California. Journal of Wildlife Diseases 28:610-613. Ubico, S. R., G. O. Maupin, K. A. Fagerstone, and R. G. McLean. 1988. A plague epizootic in the white-tailed prairie dogs (Cynomys leucurus) of Meeteetse, Wyoming. Journal of Wildlife Diseases 24: 399-406. u. s. Fish and Wildlife Service. 1988. Black-footed Ferret Recovery Plan. 87pp. Williams, E. s. 1982. Canine distemper. In Diseases of wildlife in Wyoming. E.T. Thorne, N. Kingston, W.R. Jolley, and R. c. Bergstrom (eds), Wyoming Game and Fish Department, Cheyenne, pp. 10-13. Williams, E. S., K. Mills, D.R. Kwiatkowski, E.T. Thorne, and A. BoergerFields. 1994. Plague in a black-footed ferret (Hustela nigripes). Journal of Wildlife Diseases 30:581-585. Williams, E. S., E.T. Thorne, M. J. G. Appel, and D.·W. Belitsky. 1988. Canine distemper in black-footed ferrets (Hustela nigripes) from Wyoming. Journal of Wildlife Diseases 24:385-398. Winston, P. w., and D. H. Bates. 1960. saturated solutions for the control of humidity in biological research. Ecology 41: 232-237. �133 SEGMENT NARRATIVE state of Colorado Project No. W-153-R Work Package No. __0_8-8-0'------Task No. cost center 3430 Mammals Program Black-footed Ferret conservation Monitoring and Managing Disease in Black-footed Ferrets Planning Program Narrative Objectives: 1. Monitor enzootic disease activity threatening survival of black-footed ferrets reintroduced into the LSMA. 2. Develop techniques to manage plague in the LSMA using insect growth regulators applied orally to prairie dogs. Segment Objectives: 1. Determine the level of activity of canine distemper in carnivores and plague in prairie dogs (using coyotes as sentinels) in the LSMA by sampling ~40 coyotes. 2. Establish a colony of captive white-tailed prairie dogs that can sustain infestation with fleas of the genus Oropsylla. 3. Develop techniques to artificially rear fleas of the genus Oropsylla. 4. Determine an effective dose of pyriproxifen and of lufenuron to control fleas of the genus Oropsylla on prairie dogs. Estimated segment costs= $46,540 Personal Services Permanent employees (0.5 FTE) Temporary employees Contracts Total Personal Services Operating supplies and services Travel expenses capital eq,uipment $26,040 0 10.000 $36,040 $ 8,300 2,200 0 Total Operating $10,500 Total Costs $46,540 �69 Colorado Division of Wildlife Wildlife Research Report July 1999 JOB PROGRESS REPORT ------~~----- State of Colorado Project No. _ _ _ _W'------1=5-=-3--=-R___--=-l=-2_ _ __ Work Package No . _..;;.0""'"88~0'--_ _ _ __ Task No. ______I.___ _ _ _ _ __ Mammals Research Black-footed Ferret Recovery Monitoring and Managing Disease in Black-footed Ferrets Period Covered: July 1, 1998 - June 30, 1999. Authors: .M. A. Wild and K. T. Castle Personnel: E. Wheeler. ABSTRACT The black-footed ferret is a federally listed endangered speci~s in the United States. Blackfooted ferrets-have been extirpated from Colorado, but are scheduled to be reintroduced to the wild at the Little Snake Management Area (LSMA) in Moffat County, Colorado in the near future. Our research can be sub-divided into two broad sections: disease monitoring in the proposed release area and flea control as a tool to manage sylvatic plague in prairie dogs and black-footed ferrets. Disease monitoring was performed using collection of carnivores (primarily coyotes) from the LSMA in July 1998 and February 1999. Two potentially devastating diseases, canine distemper and plague, are present at LSMA. Prevalence of titers to canine distemper virus (CDV) in coyotes are low ~10%). Prev~ence of titers to plague (Yersina pestis) were found in 60% and 20% of coyotes tested in the summer and winter collections, respectively. Titers were present in both adult and juvenile animals, suggesting ongoing plague activity in some areas. We began investigations into the bioavailability of lufenuron to orally dosed prairie dogs. A test dose of300 mg/kg body weight was determined and administered to 30 captive prairie dogs. Prairie dogs were divided into two groups (one group torpid, the other non-torpid). Blood samples for lufenuron assay were collected pre-treatment, at 1 wk postdosing, then at 2-wk intervals through week 9 post-dosing. No adverse affects to lufenuron were observed. Lufenuron assay is pending. We also attempted to establish a colony of fleas of the genus Dropsy/la for artificial infestations on captive prairie dogs. Although we were unsuccessful in establishing self-sustaining colonies of 0. tuberculata (the prairie dog flea) from wild populations, we did establish a thriving colony ofinsectary-reared 0. montana (the ground squirrel flea). These fleas will be used in future investigations into the efficacy of lufenuron to control fleas on captive prairie dogs. f '=ii WH t~ BDOW014186 ��71 MONITORING AND MANAGING DISEASE IN BLACK-FOOTED FERRETS Margaret A. Wild and Kevin T. Castle P. N. OBJECTNES 1. Monitor enzootic disease activity threatening survival of black-footed ferrets reintroduced into the LSMA. 2. Develop techniques to manage plague in the LSMA using insect growth regulators applied orally to prairie dogs. SEGMENT OBJECTNES 1. Determine the level of activity of canine distemper virus in carnivores and plague in prairie dogs (using coyotes as sentinels) in the LSMA by sampling >40 coyotes. \"' 2. Establish a colony of captive white-tailed prairie dogs that can sustain infestations ~eas__of.the genus Oropsylla. 3. Develop techniques to artificially rear fleas of the genus Oropsylla. 4. Determine an effective dose ofpyriproxifen and oflufenuron to control fleas of the genus Oropsylla on prairie dogs. METHODS AND MATERIALS Carnivore Disease Survey Infectious diseases can severely impact the success of black-footed ferret (Mustela nigripes) reintroduction efforts. As part of the black-footed ferret reintroduction protocol, we monitored disease activity in carnivores in the Little Snake Management Area (LSMA), Colorado. Coyotes (Canis latrans) were collected.in July 1998 and February 1999. Post-mortem examination and sample collection was performed as described in the Program Narrative (Wild 1998). Black-footed Ferret Reintroduction We assisted in preparation of the black-footed ferret allocation request submitted to the US Fish and Wildlife Service by the Colorado-Utah black-footed ferret recovery working team in 1999. We prepared a protocol for the care of captive black-footed ferrets at the LSMA {Attachment 1). Consultation on health matters and veterinary care were.provided for captive black-footed ferrets. Artificial Rearing of Oropsylla fleas In summer 1998 and spring 1999 we collected Oropsylla spp. fleas from prairie dog burrows at Arapaho National Wildlife Refuge. We attempted to maintain and propagate these wild fleas under �72 insectary conditions (Wild 1998). Additionally, we attempted to produce self-sustaining populations of 0. tubercu/ata within the nest boxes of four quarantined prairie dogs. To enhance flea survival, we placed a wooden box fitted with a screen top into each nest box. The wooden "flea refuge" allowed fleas to jump on or off the animal at will, and kept the prairie dog from stepping on fleas and eggs. We placed "bed-o-cobs" bedding material on the refuge floor in order to provide hiding places for the fleas. Each prairie dog was infested with up to 60 adult fleas (about 3:2 females:males). To count fleas, we anesthetized and groomed each animal periodically, and noted the number of surviving fleas found on the animal and in its nest box. Because of the decline in 0. tuberculata availability in the field and the low survivorship and reproduction of that species in the lab, we decided to explore the use oflaboratory-reared 0. montana for experimental purposes. 0. montana is generally found on ground squirrels, such as Spermophilus beecheyi (California ground squirrel); it should be able to reproduce when fed on the closely-related prairie dogs. Other researchers have maintained lab colonies of 0. montana for a number of years, and have found that the species readily reproduces when fed in artificial systems, and when fed on captive neonatal or adult rodents. While we would rather utilize true prairie dog fleas, such as 0. tuberculata, in our experiments, the use oflab-reared 0 montana offers some advantages over 0. tubercu/ata: 1) 0. montana is readily available; 2) the temperature and humidity preferences of the species are wellcharacterized, so it reproduces well in an insectary; and 3) lab-reared 0. montana are plague-free. In summer 1999 we acquired a small colony of about 400 0. montana from the Centers for Disease Control (CDC), Fort Collins. These fleas were maintained under insectary conditions (Wild 1998). Flea Control In Prairie Dogs Thirty-three white-tailed prairie dogs (Cynomys leucurus) collected in June 1998 from Arapaho National Wildlife Refuge were maintained at Foothills Wildlife Research Facility (FWRF). Details of animal husbandry and health are reported by Wild (1999). Initially, we planned to evaluate two insect growth regulators in captive prairie dogs: lufenuron and pyriproxyfen. However, after reviewing the literature, contacting product manufacturers, and investigating methods to measure efficacy, we have determined that lufenuron holds the most promise for effective application. Lufenuron is longer lasting than pyriproxyfen (Palma et al. 1993, Hink et al. 1994), researchers with Novartis (the manufacturer oflufenuron) are interested in collaboration, and a blood assay is available to measure lufenuron but not pyriproxyfen. We performed two studies to begin evaluating lufenuron in white-tailed prairie dogs. First, a pilot study to determine bioavailabilty oflufenuron administered orally to captive prairie dogs at three dosages: 20 mg/kg, 60 mg/kg, and 300 mg/kg (n = I at each dosage). Blood samples were collected for lufenuron assay pr~treatment and 1, 7, 28, 42, and 56 days post-dosing. Lufenuron assay using HPLC was performed by En-Cas Laboratories. Results of this pilot study were used in planning the following study to determined the seurm profile oflufenuron during active and torpid periods in orallydosed white-tailed prairie dogs: Bioavailability of lufenuron administered orally to captive white-tailed prairie dogs (Cynomys leucurus) Kevin T. Castle 1, Margaret A. Wild 1, and S. Craig Parks2 1 Colorado Division of Wildlife, 317 W. Prospect, Ft. Collins, CO 80526 2 Novartis Animal Health, P. 0. Box 26402, Greensboro, NC 27404 �73 Introduction Mortality of black-footed ferrets (Mustela nigripes) from infection with sylvatic plague is extremely high (Williams et al., 1994). Further, mortality from sylvatic plague in prairie dogs (Cynomys spp.) can markedly impact the prey base of black-footed ferrets. Plague has severely hampered reintroduction efforts in other states (P. Marinari, pers. comm.) and preliminary data suggest that plague activity is present in some areas of the Little Snake Management Area (Wild, unpub. data) where black-footed ferrets are scheduled to be released in late 1999. Successful reintroduction will likely require management techniques to control·plague. We believe that the proposed research will provide a novel and effective means of controlling plague in prairie dogs and black-footed ferrets. Plague is maintained primarily in rodent populations, where it is typically transmitted via the • bite of an infected flea Fleas of the genera Opisocrostis and Dropsy/la have been associated with plague transmission in prairie dogs (Ubico et al., 1988). Insecticide dusts have been applied to prairie dogs burrows in an attempt to kill fleas and thus control plague. Carbaryl has been the most commonly used insecticide; however, application is labor intensive and activity is short-lived (Barnes, 1993). Permethrin has been shown effective up to 84 days after· application (Beard et al., 1992), but the authors warn that application at the recommended dosage rate would be highly laborious. Recently developed compounds used to control fleas in pet animals offer a promising alternative to insecticide dusts. Although topically applied flea adulticides, such as imidacloprid, are not feasible for use in wildlife, insect growth regulators (IGRs) with ovicidal and/or larvicidal activity could be delivered orally to free-ranging animals via bait. The IGR lufenuron is a benzoylphenylurea derivative which inhibits formation of chitin in the exoskeleton of insects (Cohen, 1987). A single oral dose oflufenuron has been shown effective in controlling the cat flea (Ctenocephalidesfelisfelis) for at least 30 days in treated cats (Blagburn et al., I 994) and dogs (Hink et al., 1994). Davis (1997) reported a significant reduction in fleas on free-ranging ground squirrels (Spermophi/us beecheyi) that had been treated with lufenuron. Iflufenuron proves efficacious over a long period of time~ I mo) when administered orally to prairie dogs, it could be applied over large areas of prairie dog habitat via a treated bait. Prior to management application, lufenuron should be evaluated in a controlled laboratory setting to determine an effective dose, duration of efficacy, product safety, and to formulate an acceptable bait carrier. If results of laboratory tests are positive, lufenuron should be tested in a controlled field study to determine efficacy under natural conditions. Without this evaluation, the management potential ofIGRs for controlling plague and protecting the health of the endangered blackfooted ferret, as well as human health, will be difficult to discern. We have proposed studies to test the efficacy oflufenuron for controlling fleas on artificiallyinfested captive white-tailed prairie dogs (Cynomys leucurns; Wild and Castle, 1998); however, laboratory rearing of fleas has proven difficult thus far. Further, because Dropsy/la and Dpistocrostis fleas are most readily collected in spring and early summer, our lufenuron-efficacy work is limited to the summer months. As an alternative, preliminary, means oflufenuron evaluation, we will conduct a bioavailability study oflufenuron administered orally to captive white-tailed prairie dogs. Prescribed oral doses of lufenuron for domestic cats and dogs are 30 mg/kg and IO mg/kg body weight, respectively. The concentration oflufenuron in blood of protected animals is somewhat variable, but protection seems to be conferred to dogs and cats when blood levels reach at least 50-100 parts per billion (B. Blagburn, pers. comm.). No data are available regarding the bioavailability of ingested lufenuron in any wild rodent species; however, assay results are pending from a recent pilot study (Castle and Wild, 1998). Based on results from this pilot study and assuming 100 ppb as the efficacious blood level, we will determine the test dose for this experiment. Although the correlation between blood levels oflufenuron and efficacy of flea control in prairie dogs will need to be determined in future experiments, this preliminary work will provide insights into: 1) between-animal variability �74 and 2) features of the decay curve oflufenu~on (e.g. peak concentration and duration oflevels over 100 ppb) in the blood of prairie dogs. Because white-tailed prairie dogs are spontaneous, obligate hibernators, they exhibit marked changes in activity and physiologic function seasonally (Harlow, 1997). These activity and physiologic changes may influence lufenuron bioavailability. Throughout much of their range, adult white-tailed prairie dogs enter hibernation in August or early September, and juveniles begin hibernation in late September or early October. Before hibernating, they store large amounts of body fat, to provide energy during their winter-long fast (Harlow, 1997). It is possible that as fat is mobilized during hibernation and upon arousal, lipid-soluble materials that were ingested and stored during late summer and fall may also be mobilized. Because lufenuron is highly lipid-soluble (Blagburn et al., 1994), white-tailed prairie dogs fed treated bait prior to hibernation could potentially store the compound over . winter, and therefore be protected during winter and when they arouse in spring. If prairie dogs were thus protected, flea population increases could be pre-empted, and management of plague outbreaks would be enhanced. To test this hypothesis, we will compare blood lufenuron levels in active vs torpid prairie dogs in a controlled laboratory setting. To summarize, the objective of this study is to determine the serum profile oflufenuron during active and torpid periods in orally-dosed white-tailed prairie dogs. These data will be used in the design of future experiments to evaluate lufenuron in captive and free-ranging prairie dogs, and will provide insights on the between-animal and seasonal variation that is to be expected. When combined with efficacy information, blood assay for lufenuron would provide a useful tool for evaluating lufenuron treatment programs involving free-ranging animals. Methods Care and housing ofprairie dog All captive prairie dogs will be pair-housed in cages under the same conditions described in Castle and Wild, 1998. Prairie dog cages (46 x 91 x 46 cm) are constructed of wood and wire fencing. Half of the cage serves as a nest box, and the other half serves as a feeding area Two hinged doors allow access to each half of the cage for cleaning or animal handling. The nest box is a 40 x 28 x 23 cm plastic container· fitted with a wooden lid. The nest box has a 10 cm diameter hole in one wall to provide access to the feeding area; a 10 cm diameter piece of PVC pipe joins the nest b.ox to the feeding area The access hole can be covered to confine the animals to either side of the cage as necessary. Our preliminary work shows that defecation and urination in the nest box are minimal, so we do not expect the nest boxes to become soiled during experiments. The floor of the feeding area consists of 1cm mesh wire fencing to allow feces, urine, and spilled food to fall through to a newspaper-lined metal pan below. This mesh size prevents the animal's feet from becoming wedged in the floor, but may encourage the animal to spend most of its time in the nest box when not feeding. Newspaper in the metal pan will be changed daily if soiling occurs. Food (Teklad Rodent Blocks) will be provided ad libitum except as noted below. Water will be available at all times. Prairie dog cages will be kept in two similar buildings. Temperature in each building is thermostatically controlled, and artificial lighting is controlled with timers. Health and attitude of prairie dogs will be assessed daily during the experimental period. Body weight will be measured at 1-2 week intervals while each animal is anesthetized for blood collection. �75 Experimental Design Thirty captive white-tailed prairie_ dogs (16 males and 14 females) less than one year old will be blocked by sex and randomly divided into two treatment groups (n = 13 each) and two control groups (n = 2 each). Treatment group 1 and one control group (active-group animals) will be housed in a building under conditions that inhibit hibernation (14L: 1OD photoperiod plus natural lighting through windows, ambient temperatures above 15 °C). Treatment group 2 and a second control group (torpid-group animals) will be housed in a second building under conditions that promote hibernation (1 0L: 14D photoperiod, ambient temperature of 5-6 °C). Intermittent fasting of up to 2 d per 7 d period will also be used in some individuals to promote hibernation if temperature and photoperiod changes prove insufficient. Our goal is to expose the two groups to the most extreme conditions known to inhibit or promote torpor. We do not expect food deprivation to harm the torpid-group animals, because food intake between torpor bouts is minimal (Harlow, 1997). We have determined that these experimental conditions allow us to inhibit or promote hibernation in a majority of our white-tailed prairie dogs; however, despite our efforts, some animals housed under hibernation-promoting conditions remain active, while some housed under hibernationinhibiting conditions become torpid. We will therefore assess each animal daily, and record its activity level. Active-group prairie dogs that become torpid will be aroused by gentle shaking and rubbing; torpid-group animals that remain active will not be disturbed. We will assess the duration of torpor in torpid-group animals by placing a small amount of sawdust on the back of each torpid individual; the presence of sawdust on the back at subsequent checks will be interpreted to mean that the individual was torpid the entire period (Harlow, 1997). Dosing Methods Prairie dogs will be fasted for 24 h, then weighed prior to dosing; water will be available during the fast. After the fasting period, each individual in a cage will be confined to one-half of the cage by blocking off the nest box access hole, and temporarily removing the nest box. Each animal in treatment groups 1 and 2 will be fed a bolus dose of lufenuron (20-300 mg/kg dose, pending pilot trial results) mixed thoroughly in a highly palatable bait (ground rat chow and molasses). Control animals will receive the bait only. The bait will be offered in each animal's feeding dish. We previously determined that fasted prairie dogs consume up to 2 g of such a bait within about 30 min, and will consume up to 13 gin less than 12 h. We will observe the progress of bait ingestion in each animal to determine when the dose is ingested. We will weigh the bait/lufenuron mixture before and after the prairie dogs are dosed, in order to calculate the actual dose ingested. After both animals in a cage have consumed their dose, the nest box will be returned and they will again be pair-housed. Blood Collection We will collect 3 ml of blood from each prairie dog prior to lufenuron dosing, one week after dosing, then at 2 week intervals through week 9 post-dosing, or until torpor bouts are no longer observed. Anesthesia is required for blood collection. We have determined that prairie dogs respond very well to isoflurane anesthesia. Induction and recovery are quick (approximately 3-5 min for each), and we have detected no adverse effects. All prairie dogs will be aroused from torpor prior to anesthesia, in order to assure sufficient blood pressure for blood collection. Prairie dogs_will be placed into a plastic chamber for induction at a concentration of 4% isoflurane in oxygen, then anesthesia will be maintained via mask at about 2-2.5% isoflurane. Heart rate and respiration will be monitored during anesthesia. Blood will be collected by jugular venipuncture and placed into a glass vacutainer blood tube (without anticoagulant). Alternatively, if we are unable to collect an adequate blood sample �76 peripherally, we will collect blood from the vena cava or directly from the heart. Although cardiac puncture is considered to be a safe procedure for collection oflarge volumes of blood from laboratory rodents (CCAC, 1984), due to the increased risks involved with cardiac puncture (5-10% mortality expected; J. Wimsatt, pers. comm.), we will use this technique only to obtain critical samples. Serum will be harvested and frozen within 4 h of collection. Serum lufenuron concentration will be determined by HPLC at Novartis Laboratories. Product safety Based on mode of action, chitin inhibitors should be safe in mammals, even at very high doses (Blagbum et al., 1995; T. Miller, pers. comm.). Cats have been fed more than 5 times the prescribed dose oflufenuron without showing adverse side effects (B. Blagburn, pers. comm.). The high dose to which our prairie dogs may be exposed (300 mg/kg) is based upon our estimate of the maximum amount of bait a prairie dog might eat in a day in the field (40-60 g) and a reasonable lufenuron concentration in the bait (5 mg/g). We did not detect adverse effects from lufenuron administration during our pilot study, when 3 prairie dogs received doses of20-300 mg/kg. We do not anticipate any prairie dog mortality due to the dosing regime or blood collection. However, if an animal becomes sick or injured, and recovery is not likely, it will be euthanized with an overdose of inhalant anesthetic. A complete post-mortem examination will be performed on any animal that may die during the experimental period. Data Analysis We will compare lufenuron blood concentrations of the torpid- and active-group prairie dogs over time by conducting a profile analysis of the lufenuron decay curves. In addition, we will use analysis of covariance to compare: 1) peak concentrations obtained in each group, and 2) the duration of concentrations above 100 ppb in each group. The number of days an animal was observed to be active will serve as the covariate in our analyses. Group responses will be considered significantly different at p 0.05. Literature Cited Barnes, A. M. 1993. A review of plague and its relevance to prairie dog populations and the blackfooted ferret. Proceedings of the symposium on the management of prairie dog complexes for the reintroduction of the black-footed ferret. J. L. Oldemeyer, D. E. Biggins, B. J. Miller, and R Crete, eds. 96pp-. • Beard, M. L., S. T. Rose, A. M. Barnes, and J. A. Montenieri. 1992. Control of Oropsylla hirsuta, a plague vector, by treatment of prairie dog burrows with 0.5% permethrin dust. Journal of Medical Entomology 29:25-29. Blagbum, B. L., J. L. Vaughan, D.S. Lindsay, and G. L. Tebbitt. 1994. Efficacy dosage titration of lufenuron against developmental stages of fleas (Ctenocephalidesfelisfelis) in cats. American Journal of Veterinary Research 55:98-101. Blagburn, B. L., C. M. Hendrix, J. L. Vaughan, D. S. Lindsay, and S. H. Barnett. 1995. Efficacy of lufenuron against developmental stages of fleas (Ctenocephalidesfelisfelis) in dogs housed in simulated home environments. American Journal of Veterinary Research 56:464-467. Castle, K. T. and M. A. Wild. 1998. Safety and efficacy of pyriproxifen and lufenuron for the control of prairie dog fleas: a pilot study. Colorado Division of Wildlife Animal Care and Use Committee Study Plan. Canadian Council on Animal Care (CCAC). 1984. Guide to the care and use of experimental animals, Vol. 2. Ottawa, Ont., Canada �77 Cohen, E. 1987. Interference with chitin biosynthesis in insects. In Chitin and benzoylphenyl ureas, series entomologica, vol. 38, J.E. Wright and A. Retnakaran (eds), Dr. W. Junk, Publishers, Boston, pp.33-42. Davis, R. M. 1997. Use of an orally administered insect development inhibitor (lufenuron) as a flea control agent in the California ground squirrel, Spermophilus beecheyi. Fourth International Symposium on Ectoparasites of Pets, pp. 31-35. Harlow, H. J. 1997. Winter body fat, food consumption, and nonshivering thermogenesis of representative spontaneous and facultative hibernators: the white-tailed prairie dog and blacktailed prairie dog. Journal of Thermal Biology 22:21-30. Hink, W. F., M. Zakson, and S. Barnett. 1994. Evaluation of a single oral dose of Iufenuron to control flea infestations in dogs. American Journal of Veterinary Research 55:822-824. Ubico, S. R, G. 0. Maupin, K. A. Fagerstone, and R. G. McLean. 1988. A plague epizootic in the white-tailed prairie dogs (Cynomys leucurus) of Meeteetse, Wyoming. Journal of Wildlife Diseases 24:399-406. Williams, E. S.; K. Mills, D. R Kwiatkowski, E.T. Thome, and A. Boerger-Fields. 1994. Plague in a black-footed ferret (Mustela nigripes). Journal of Wildlife Diseases 30:581-585. Wild, M. A. and K. T. Castle 1998. Monitoring and managing disease in black-footed ferrets. Colorado Division of Wildlife Program Narrative, Project# W-153-R RESULTS AND DISCUSSION Carnivore Disease Survey With the assistance of USDA Wildlife Services and the Bureau of Land Management (BLM) we collected 19 coyotes and 4 badgers from the LSMA between 27-30 July 1998 and 20 coyotes between 11-12 February 1999. Coyotes were collected using a combination of calling and aerial gunning. Death occurred rapidly and the collection technique was adequate, but much less efficient in summer than in the winter months. Coyotes were collected from various locations in the >4 700 mi 2 management area; however, we focused on the Powderwash area and near the site of the preconditioning pens. No lesions indicative of active disease were noted on gross examination of carcasses. We found no serologic evidence of exposure to leptospirosis serovars canicola, grippo, hardjo, and pomona; however, low positive titers to serovar ictero were found in nine coyotes (five in July and four in February). No serologic titers to toxoplasmosis were found. Coyotes were not tested for Aleutian Disease (a mustelid disease), but all four badgers were sei-onegative for the disease. All badgers and 20% of coyotes sampled during the summer had positive titers to tularemia, while 10% of the coyotes sampled in winter were positive. Positive titers could have been associated with predation on rabbits, prairie dogs, or other rodents or exposure to ectoparasites. Duration of titers to tularemia are likely of short duration (<6 mo). The impact of tularemia or leptospirosis on black-footed ferrets is not currently known; however, ferrets and humans are susceptible to these diseases. Canine distemper virus (CDV) and plague are serious threats to survival of black-footed ferrets. The prevalence of coyotes with serological titers to CDV was .:S,10%; markedly lower than in previous years (Fig. 1). However, plague activity continues in LSMA as evidenced by titers in sampled coyotes (Fig. 2). Black-footed Ferret Reintroduction Nineteen black-footed ferrets (8 adult females, 5 juvenile females, 6 juvenile males) were received from the National Black-footed Ferret Conservation Center in November 1998 and placed in breeding pens at LSMA. Seventeen ferrets remained in good health with minimal health treatment �78 required. One adult female ferret died of a biliary adenocarcinoma and one juvenile female is missing and presumed dead. Ten female ferrets were paired with males and five of these gave birth. One litter was apparently consumed by the female when <l day of age. Toe· other four litters yielded 13 pups. Artificial Rearing of Dropsy/la fleas Although fleas had been abundant in the spring of 1998, in July fleas seemed to disappear from the field. Burrow swabs consistently returned 0, 1, or 2 fleas per burrow. Because most burrows had no fleas, the ratio of fleas/burrow dropped to less than 0.5. The cause of decline is unclear. One possibility is that because many burrows were filled in by prairie dogs escaping the summer heat, the nest areas that house a majority of fleas were no longer accessible from the surface. It is also possible that the flea populations experienced a real decrease during the summer, and that decrease was reflected in swabbing success. Our attempts to artificially rear Oropsylla fleas during summer 1998 met with limited success. We had successful completion of the life cycle within our rearing jars, but only about 4 adult fleas were produced from eggs laid in captivity. Our overall success was limited by low flea survival during transport to the lab, and the relatively low numbers of fleas available for collection in late summer when our facilities were readied for use (virtually no fleas were available in the field after 30 June). By the end of October 1998, no fleas were alive in our insectary. We resumed field collections of fleas on 5 April 1999, when snow cover had receded and prairie dog burrows were open. We inade five more trips to the field through the middle of May. A total of 1645 fleas (906 of these 0. tuberculata) were collected. We successfully increased flea survival during transport from the field to the lab by keeping fleas under cooler and more humid conditions. Most fleas collected were alive when they arrived at the lab. Unfortunately, adult flea mortality in the insectary was high, despite our attempts to keep the fleas well-fed and under optimal temperature and humidity conditions. Reproduction within the insectary was therefore limited. We collected 8 larva from the insectary jars, and placed them into a separate dish containing rearing media; none of the larva pupated. Attempts to produce self-sustaining populations of 0. tuberculata on prairie dogs were also disappointing. While a few fleas survived for 21-24 days on the prairie dogs, it became clear that selfsustaining populations were not possible under our conditions. Our lack of success was most likely due to our inability to properly mimic natural burrow conditions required by fleas (constant temperature and high humidity). We attempted to simulate burrow conditions by attaching insulation to each nest box (to maintain temperature), and by placing a screen-covered petri dish of moist vermiculite into the flea refuge (to increase humidity). We were able to keep temperatures fairly stable, but were not able to maintain high relative humidity within the nest boxes. Survival and reproduction of 0. montana in the insectary has been high. We will use 0. montana in future experiments investigating flea control in captive prairie dogs. Flea Control In Prairie Dogs In the pilot study we observed no adverse affects with any of the lufenuron dosages. Previous researchers have shown lufenuron to be efficacious for controlling fleas in cats and dogs when serum concentration is 50-100 parts per billion (ppb) (C. Parks, Pers. comm.). Our results indicate that, as expected, baseline samples contained no lufenuron, but that one day after dosing, blood concentration was at or above efficacious levels at all treatment levels. After one week, blood concentrations had decreased considerably; serum concentration in the animals dosed with 20 mg/kg and 60 mg/kg had fallen below the efficacious level, while the concentration in the animal fed 300 mg/kg remained well above the efficacious level at 220 ppb. Analyses of remaining blood samples are pending. �79 In January 1999, we performed the experiment titled "Bioavailability oflufenuron administered orally to captive white-tailed prairie dogs". All prairie dogs except for two non-torpid group animals ingested >95% of their bait within 18 h; all remaining bait was removed and weighed after 18 h. We observed all prairie dogs twice daily (8:00 am. and 4:00 p.m.) during the experiment for overall health and to quantify length of torpor bouts. Animals that were torpid for two consecutive observations were considered to have been torpid the entire time between observations, unless a small amount of sawdust placed on the back of torpid animals had been displaced. Animals that were torpid at one observation time then awake the next were considered to have awakened at the midway point between observations. Torpid-group animals found hibernating were allowed to remain torpid, while non-torpid group animals were awakened if found hibernating. All torpid animals were awakened prior to anesthesia and blood collection. The duration of the study was 1512 hours; torpid group animals were torpid for an average of 406 h (range= 0 - 936 h), while non-torpid group animals averaged 7.5 h torpid (range= 0 - 36 h). We detected no detrimental effects of lufenuron in any prairie dog, nor were any animals harmed by repeated anesthesia and blood collections. HPLC analysis for blood lufenuron concentrations is pending. LITERATURE CITED Hink, W. F., M. Zakson, and S. Barnett. 1994. Evaluation of a single oral dose oflufenuron to control flea infestations in dogs. Am. J. Vet. Res. 55:822-824. Palma, K. G., S. M. Meola, and R W. Meola 1993. Mode of action of pyriproxifen and methoprene on eggs ofCtenocephalides felis. J. Med. Entomol. 30:421-426. Wild, M. A 1998. Monitoring and managing disease in black-footed ferrets. Colorado Div. Wildl. Res. Rep., 0880-1, Jul 1997 -Jun 1998, Fort Collins. Wild, M. A 1999. Animal and pen support services for mammals research. Colorado Div. Wildl. Res. Rep., 8160-1, Jul 1998 - Jun 1999, Fort Collins. 25 20 uQ,) ~ 15 ro U) 'Q,) ..o 10 E ~ z 5 . . . .. . . . 1997-W ...... 1997-S 1998-W ■ •• 1998-S 1999-W Sample period 8 CDV positive □ CDV negative I Fig. 1. Prevalence of exposure to canine distemper virus (CDV) in coyotes from the Little Snake Management Area, Colorado, from winter 1997 through winter 1999. All positive coyotes were adults with the exception of one juvenile in summer 1998. Data from winter 1997 (age class unknown) provided by M. Albee. �80 A 20 15 .... Q) e 10 ::J z 5 0+-------~1997-W 1997-S 1998-W 1998-S 1999-W Sampling period B 20 15 .... e 10 Q) :::, z 5 1997-W 1997-S 1998-W 1998-S 1999-W Sampling period I mPlague-Pos £3 Plague-Neg I Fig. 2. Prevalence of exposure to plague in coyotes (A= juveniles~ B = adults) from the Little Snake Management Area, Colorado, from winter 1997 through winter 1999. Data from winter 1997 (age class unknown) provided by M. Albee. �81 ATIACHMENT I HUSBANDRY PROTOCOL FOR CARE OF CAPTIVE BLACK-FOOTED FERRETS AT THE LIITLE SNAKE MANAGEMENT AREA Pen Construction Breeding pens will be constructed over established prairie dogs towns. If an active prairie dog colony is not present, prairie dogs from plague-free areas will be reintroduced to the pen site prior to construction. Individual pens will be about 1500-5000 ft2 and constructed in clusters of four (four-plex design). Generally, females will be housed in three of the pens and a male in the fourth pen. Each pen will contain a vault capable of holding a buried insulated wooden nest box (about 12" x 20" x 16"), which will be covered with a roof to provide shade and to keep runoff away from the nest box. In the females' pens, the vault will be contained a sub-enclosure, or whelping pen, about 100 ft2 in size. A perimeter fence will be constructed around each four-plex or group offour-plexes. Tubing and traps will be placed at intervals along the fence to protect and capture potential escapees. Additionally, prior to initial introduction of breeding-age ferrets, the burrows will be checked using a "smoke test" and then radio-collared ferrets will be placed in the pens to further check pen sec~ty: Site Security Breeding pens will be located in a remote area with access limited to authorized personnel. Visitors will be allowed only on a limited basis and when accompanied ~y-site personnel. Human activity near the pens will be limited strictly to ferret husbandry and management practices. No_dogs will be allowed in the pen area Use of motorized vehicles in the pen ar~a, will be minimized. The perimeter fence will serve to contain escaped ferrets and as a deterrent to entrance of wild and feral carnivores; however, other means of carnivore control (nonlethal and lethal) may be necessary to protect ferrets. Biosafety protocols will include the use of site specific coveralls and boots by personnel handling ferrets or entering ferret pens. Treatment of shoe soles with a bacteriocidal/virucidal solution (e.g., Roccal-D) may substitute for designated footwear. Personnel will not enter ferret pens or handle ferrets if they are experiencing flu symptoms (fever, chills, body aches, respiratory signs); caretakers with cold symptoms should wear a mask and gloves when working around ferrets. Ideally, caretakers should receive flu shots annually. Only prairie dogs from the immediate vicinity of the pen site or those completing quarantine-will be provided to ferrets. Personnel will not handle carnivores or prairie dogs ( except those completing the quarantine period) during the work day prior to entering the pen site unless they shower and change clothes. Pets of site personnel should be vaccinated annually against canine distemper. General Husbandry Ferrets will be checked daily at dawn or dusk by trained caretakers. Caretakers should minimize contact with ferrets, but should observe ferrets directly or with binoculars to assess health status daily. Abnormalities will be noted in animal records and, if necessary, the attending veterinarian notified. If a ferret is not observed (directly or via sign, e.g., consumed food) for 3 consecutive days, a trap (attached to a nest box via tubing) will be placed in the pen. Traps will be checked at intervals appropriate to assure that the ferret is not adversely affected by temperature, weather, or restraint time �82 (generally, every 2-6 hr). Trapped ferrets will be weighed and examined grossly. Additional examination and treatment will be administered if necessary under the direction of the attending veterinarian. Body weights will be obtained monthly (and anytime the ferret is handled for other reasons) to monitor health status and to evaluate feeding programs. Ferrets will be trapped, weighed, briefly examined, and released qack to the pen. Transponder function will be checked at least once monthly and transponders will be replaced (with the ferret under anesthesia) if needed. Nest boxes will be provided only during periods when the ferret is confined to the whelping cage, e.g .. , the breeding and whelping period or during required confinement to prevent escape or enhance treatment. Nest boxes will be cleaned every other day ( excepting during breeding period when male and female are housed together and immediately prior to and following parturition) using pen-specific instruments, bedding, and trash can when kits are present or if a contagious disease is suspected. Wood shavings (not cedar) will be placed about 3-4 inches deep on one side of the nest box and about l inch deep of al phi-dry will be placed on the other side. A minimum-maximum thermometer will be placed on the roof of one nest box in each four-plex. Temperatures will be checked during nest box cleaning or daily during periods of extreme weather. Modifications to the nest box insulation may be needed if temperatures in the nest box are extreme. Maintenance diet will include live prairie dogs supplemented with frozen prairie dog and rabbit carcasses, hamsters, and dry ferret food (Totally Ferret). Dry ferret food will be used only a supplement if insufficient supplies of prairie dogs are available as food. Dry ferret food will not be provided as a sole feed source, but as a supplement to the meat diet. Prairie dogs will be collected, quarantined, and processed as described in the protocol by Marinari and Williams (1998). Rabbits will be procured from breeders and processed as prairie dogs; however, additionally the head and kidneys will be removed. Dry feed will be stored in rodent-proof containers prior to feeding. Whenever possible, live prairie dogs will be provided to ferrets; however, we will avoid offering large, aggressive prairie dogs and any prairie dog surviving > 3 days will be trapped and removed from the ferret pen. These prairie dogs will be euthanized and fed as meat. Female ferrets will be provided with. an average of 100 g feed/day and males will receive 125 g feed/d; however, ferrets may not be provided fresh feed daily (e.g., if a 500 g prairie dog is provided). Amount of feed provided to individuals will vary based on assessment of physical condition. Frozen prairie dogs will be thawed about 48 hr in a refrigerator prior to feeding. Feed may at times be provided in a trap (with the doors wired open) for training purposes. Unconsumed meat left in the pen will be discarded after 24 hr. Water will be injected into prairie dog carcasses during the winter or when available water is minimal. A water source will be provided, but may be frozen during the winter. If snow cover is not present, fresh water will be provided daily. Waterers will be cleaned weekly or more frequently if needed. Records will be maintained to document feed offered and estimate feed consumed by ferrets in each pen. During gestation, whelping, and lactation, modifications of the maintenance diet will be required. About 2 wk following breeding, feed will be increased 1.33 times the base diet. About 4 wk following breeding, feed will be increased to I .66 times the base diet. From whelping to weaning, feed will be offered ad libitum. Ideally, only live prairie dogs and hamsters should be offered, but if unfeasible, ferrets will be supplemented with prairie dog or rabbit meat or dry feed initially, then when kits begin to kill prey, only live prairie dogs (about 150 g/ferret) will be offered until release. as Breeding Management Adult males and females will be housed in separate pens except during the breeding period. The breeding period will vary annually based on weather, so multiple handling of ferrets may be required to assess reproductive status. Beginning about 1 February, or when signs of breeding behavior are noted (increased digging, marking, etc), males will be trapped weekly for breeding �83 soundness examination. With ferrets restrained in handling cages, the testes will be measured (width, depth) and tested for firmness. After testes are firm for 1 mo, males will be considered ready for breeding. When males are showing signs of reproductive activity (testes firm and about >20 mm width or depth), females will be trapped for signs of estrous. The females will be restrained in a handling cage and the vulva measured. If the vulva is small (<2 mm x 3 mm), the female will be trapped and measured again in about 7 days. If the vulva is beginning to swell, the female will be handled again at 3 day intervals to monitor the trend in change of the vulva Vaginal cytology may be used when the vulva shows a trend toward increasing size (generally to about 4 x 7 mm). When vaginal cytology indicates >90% comified cells, we will wait 5 days, then introduce the male to the females whelping pen (male and female will be restricted to whelping pen). The male and female will be monitored remotely using a pen camera or by watching from a distance and listening for vocalizations to determine compatibility and breeding activity. Breeding activity is assumed if chuckling is heard, copulation observed, and/or a neck stain is present on the female. In this case, the male will be removed 3 days after introduction. Function of the females transponder will be checked after copulation and replaced if necessary. If breeding activity is not observed, the male may be left with the female up to 7 days; however, if the male and female show no interest in one another (in separate areas of pen, no neck stain), the male may be removed after 24 hr and a new male introduced 1-2 days later. The male will be removed immediately if it shows aggressive behavior that threatens the female. Males will be given 3-4 days rest after each breeding. A male may be brought from a separate fourplex to breed a female if the resident male is incompatible or unavailable. The female will be trapped 7-10 days following suspected breeding for vaginal cytology. If cells remain highly cornified, breeding likely did not occur and the female will be re-paired. Records will be kept on results of breeding soundness exams, vaginal cytology, pairings, and breedings. The female will be restricted to the whelping area beginning 35 days post-breeding. A waterer will be placed in the whelping area and bedding material placed in the nest box. Vitamin K will be supplemented to the females diet at the rate of 5 drops/day from 5 days_prior to whelping to day 35 following whelping. On day 42 post-breeding, we will begin listening daily for sound from the nest box. If sounds are heard, the animals will be left undisturbed for 5 days. On day 5, the nest box will be cleaned using pen-specific bedding, instruments, and trash can. The nest box will be cleaned again on day 10 and the door to the whelping area opened. The nest box will be cleaned at 2 day intervals until the female moves the kits from the nest box. Prairie dogs and hamsters will be the only feed offered until kit release. If females do not whelp a second breeding will be attempted. About 10 days following the calculated whelping date we will again begin monitoring vaginal cytology. When 90% cornified cells are present, a male will be paired with the female as described previously. Kits will be trapped on day 90 for physical examination, marking with transponders, and vaccination against canine distemper virus (if vaccine available). Kits will be trapped subsequently for booster vaccines if needed. Care of Sick Animals Health status of ferrets will be checked daily (or as possible). Healthy ferrets should be active, consume feed, have a clean haircoat, and be free of physical abnormalities (lameness, cuts, swellings, etc.). The pen floor should also be examined for abnormal feces, blood (not from prey item), vo,mit, etc. Any abnormalities will be noted on the daily observation forms and the attending veterinarian notified. If there is any likelihood that the condition could be contagious, the caretaker will disinfect their hands and boots and ideally change coveralls before proceeding to other pens. If an animal is known to be sick, that animal will be cared for last (or by an alternative caretaker) using good sanitation techniques and any equipment used in the pen will be disinfected. Sick animals may be �84 confined to the nest box and whelping pen to aid in care and treatment. Alternatively, critically ill animals may be placed indoors in quarantine for treatment. If coccidiosis is suspected (lack of appetite, mustard-like feces) and the attending veterinarian cannot be reached, treatment can be initiated using 50 mg/kg Albon orally on day 1 followed by 25 mg/kg Albon orally daily for about 7 days. Fresh water must be available at all times when animals are treated with Albon (parenteral fluid therapy may be needed as well). Specimens (feces, discharges) should be obtained opportunistically if possible (using good sanitation practices) and placed in a whirlpack and refrigerated until further instructions are obtained. If a ferret is found dead, it should be double or triple bagged (e.g., in a ziplock or trash bag), labeled with animal ID, pen, and date and refrigerated. The carcass will be boxed in a cooler and submitted via overnight Federal Express to Dr. Beth Williams, Wyoming State Veterinary Laboratory for necropsy. The facility supervisor, attending veterinarian, and BFFCC should be notified for further instructions. Prairie dogs found sick or dead in the pen area should also be collected and submitted for necropsy. Caretakers should wear gloves and practice good sanitation measures whenever dealing with sick or dead animals. Release We will attempt to maintain 20 breeding animals of various ages: ideally, 5 female and 1 male I-year-olds, 5 female and 2 male 2-year-olds, 5 female and 1 male 3-year-old, and 1 male 4-year-old. Animals will be redistributed to pens each fall to maximize diversity in breeding pairs. Older adults and excess young of the y~ar will be released each fall when young are about 20 weeks of age. Prior to release, all ferrets will be fitted with radiocollars. �115 Colorado Division of Wildlife Wildlife Research Report July 2000 JOB PROGRESS REPORT State of ----~~=~-----Colorado Project No. W-153-R-13 Mammals Research Work Package No. ___ 0_88_0_ _ _ _ _ __ Task No. 1 Black-footed Ferret Recovery Monitoring and Managing Disease in Black-footed Ferrets Period Covered: July. 1, 1999 - June 30, 2000. Authors: M.A. Wild and K. T. Castle Personnel: E. Wheeler, E. Schmal, and S. Kasven. \ ) ABSTRACT \ ) The black-footed ferret is a federally listed endangered species in the United States. Black-footed ferrets have been extirpated from Colorado, but were scheduled to be reintroduced to Moffat County, Colorado in 1999. However, due to high plague activity and low prairie dog densities at the Little Snake Management Area (LSMA), black-footed ferret reintroduction was postponed indefinitely at the site. The secondary site at Coyote Basin, Utah was readied and reintroduction of ferrets was performed there in 1999. The Wolf Creek site in Moffat County, Colorado is also being readied for reintroduction of ferrets, likely in 2001. Our work in support of black-footed ferret reintroduction can be sub-divided into three broad sections: disease monitoring in the proposed release areas in Colorado, care of black-footed ferrets, and flea control as a tool to manage sylvatic plague in prairie dogs and black-footed ferrets. Disease monitoring was performed using collection of coyotes from LSMA in July 1999 and from the Wolf Creek site in February 2000. Two potentially devastating diseases, canine distemper and plague, are present at LSMA. Although prevalence of positive titers to canine distemper virus (CDV) in coyotes ·have been relatively low over the last 3 years (:S33%), the prevalence of positive titers to plague (Yersina pestis) have been high (up to 89% positive). Titers were present in both adult and juvenile animals, suggesting ongoing plague activity in at least some sections of the management area. Samples collected from coyotes at the Wolf Creek site indicated substantially lower disease activity than at LSMA. Prevalence of positive titers to plague and to CDV was 7% of samples collected in February 2000. In general, captive black-footed ferrets maintained at the LSMA breeding and preconditioning pens remained healthy. One ferret died and another was presumed dead in the burrow system after a severe hailstorm. Of 13 kits born at the pens in spring 1999, 12 survived to weaning in fall 1999. These 12 kits in addition to 50 other black-footed ferrets were released at the Coyote Basin site in fall 1999. Twenty ferrets were maintained in the pens overwinter and produced 11 kits in spring 2000. We summarized research results and presented a paper titled "Dose titration and safety of lufenuron fed to captive white-tailed prairie dogs (Cynomys leucurus)" at the 2000 meeting of the �116 Wildlife Disease Association, Jackson, Wyoming. We also performed a trial to test' the efficacy of lufenuron in controlling fleas on captive prairie dogs. One week post-dosing, fleas that fed on prairie dog treated with 500 mg/kg lufenuron prcxiuced a lower proportion (p < 0.05) of viable rggs than those fed on control prairie dogs (mean 0.16 vs. 0.39, respectively). However, the proportion of viable eggs prcxiuced during week 2 post-dosing was low for each group (0.24 treatment vs. 0.17 control). By week 4 postdosing, egg prcxiuction had fallen to nearly zero in each group; however, the three eggs produced by fleas fed on treated prairie dogs were all viable. Although further, similarly controlled investigation is warranted, preliminary results suggest that given current techniques, the likelihood of lufenuron limiting fleas populations and breaking the plague cycle is not extremely promising. �117 MONITORING AND MANAGING DISEASE IN BLACK-FOOTED FERRETS Margaret A. Wild and Kevin T. Castle P. N. OBJECTIVES 1. Monitor disease activity threatening survival of black-footed ferrets reintroduced into the Little Snake Management Area (LSMA). 2. Develop techniques to manage plague in the LSMA using insect growth regulators applied orally to prairie dogs. SEGMENT OBJECTIVES 1. Provide ;ete~ care to captive and reintroduced black-footed ferrets. 2. Monitor and manage plague activity in LSMA. METHODS AND MATERIALS Carnivore Disease Survey Infectious diseases can severely impact the success of black-footed ferret (}Juste/a nigripes) reintroduction efforts. As part of the black-footed ferret reintroduction protocol, we monitored disease activity in carnivores at proposed ferret reintroduction sites: in July 1999 at the Little Snake Management Area (LSMA), Colorado and in February 2000 at the Wolf Creek Management Area (WCMA), Colorado. Coyotes (Canis latrans) were collected for post-mortem examination and samples collected as described in the Program Narrative (Wild and Castle 1998). Black-footed Ferret Reintroduction and Veterinary Care I assisted in preparation of the black-footed ferret allocation request submitted to the US Fish and Wildlife Service by the Colorado-Utah black-footed ferret recovery working team in 2000. I provided veterinary care and consultation on health matters for captive black-footed ferrets and black-footed ferret releases. Animal care was performed in accordance with the established protocol (Wild and Castle 1999). Flea Control In Prairie Dogs We completed the experiment "Bioavailability of lufenuron administered orally to captive white-tailed prairie dogs (Cynomys leucurus)" and performed a follow-up experiment "Efficacy of lufenuron for the control of fleas in white-tailed prairie dogs (Cynomy leucurus)". These studies were outlined in Wild and Castle 1998. The detailed study plan for the lufenuron bioavailability study and e:fp.cacy trial are reported in Wild and Castle 1999 and in Attachment 1, respectively. �118 RESULTS AND DISCUSSION Carnivore Disease Survey With the assistance of USDA Wildlife Services and the Bureau of Land Management (BLM) we collected 16 coyotes from the LSMA between 26-30 July 1999 and 14 coyotes from the Wolf Creek site on 8-9 February 2000. Coyotes were collected using a combination of calling and aerial gunning. Further collections were not possible due to weather constraints and aircraft availability. No lesions indicative of active disease were noted on gross examination of carcasses. Of the coyotes collected from the LSMA, 31 % (5/16) had positive titers to plague (Fig. I) using the standard HA/HI test while one additional coyote was positive using the ELISA test. Interestingly, all juveniles sampled this summer (n = 9) were negative to plague. Because the juvenile coyotes have consumed (sampled) prairie dogs from this summer only, lack of exposure may indicate that the prevalence of plague in prairie dogs in the area is declining. If this is the case, titers in adults would likely be from exposure in previous years. Alternatively, the prey base of the coyotes may have shifted to species that are less commonly infected with plague (e.g., rabbits) if the density of prairie dogs has been greatly reduced by plague. Forty-three percent of coyotes sampled (6/14 usable samples) had positive titers to tularemia. In contrast to plague, all positive titers to tularemia were observed in juvenile coyotes. As previously observed, tularemia appears to elicit a serologic response of short duration in coyotes. The impact of tularemia on black-footed ferrets and prairie dogs is unknown and warrants investigation. A titer to canine distemper virus (CDV) was found in only one of 14 serum samples tested (Fig. 2). Samples collected from coyotes at the Wolf Creek site indicate substantially lower disease activity than at LSMA. A 9-year-old male coyote had positive titers to plague and to CDV, but all other coyotes (n = 13) were negative to plague and CDV (7% positive; Fig. 3). The adult male and one additional juvenile coyote also had titers to tularemia (14% positive). Black-footed Ferret Reintroduction and Veterinary Care In general, captive black-footed ferrets maintained at LSMA remained healthy. One adult female was treated for an apparent mite infection and secondary bacterial pyoderma. Unfortunately, samples to confirm this diagnosis could not be collected prior to treatment; however, the ferret responded quickly and completely to therapy with ivermectin and amoxicillin. Additional cases of crusty skin have been successfully treated with ivermectin but confirmation of the etiology has not yet been made. A I-yr-old male was found dead and a 4-yr-old female was missing and presumed dead after a severe hailstorm. Necropsy results revealed death from blunt trauma. The death(s) occurred in the new pen complex where shallow burrow systems may have been flooded forcing ferrets to the surface in the severe weather. To avoid this problem in the future, additional above ground shelter will be provided. Of 13 kits born at the pens in spring 1999, 12 survived to weaning in fall 1999. One kit disappeared and is presumed dead in the burrow system. Based on results of carnivore disease monitoring over the past 3 years and prairie dog inventories performed in summer 1999, LSMA was determined to be currently unsuitable habitat for the release of black-footed ferrets. Prairie dog inventories performed by Utah Division of Wildlife showed insufficient densities of prairie dogs to support black-footed ferret reintroduction. As a result, ferrets were not released into LSMA but instead were released into Coyote Basin, Utah. Continued monitoring will determine when (if) LSMA can support reintroduction of black-footed ferrets. A secondary site in Colorado, (Wolf Creek) is also being readied for reintroduction of black-footed ferrets in fall 2000 or 2001. _ _, �119 The 12 kits produced onsite, in addition to three 1-yr-old males, and five 3-yr-old females from the LSMA pens were released at Coyote Basin in November 1999. Additionally, 19 kits and five 3-yr-old females from other captive breeding sites were pre-conditioned at the LSMA prior to release at Coyote Basin. The reintroduction was further supplemented with 28 ferrets released immediately upon arrival from other captive breeding sites without pre-conditioning at LSMA. Prior to release, ferrets were trapped for routine examination, treatment, and identification (Wild and Castle 1999) and a health certificate was issued for each individual. All appeared healthy except for the presence of ectoparasites (ticks, mites, fleas). Individual ferrets were treated with ivermectin and pen dusting was advised. In an attempt to meet our ideal age and sex structure of captive black-footed ferrets (Wild and Castle 1999), we retained seven ferrets and supplemented the population with 13 additional ferrets from other captive breeding facilities. Seven black-footed ferrets (five 1-yr-old females and two males) were retained in captivity at the LSMA pens. In November 1999, six adult females,'three female kits, and four male kits were added to this breeding group bringing the total number of ferrets at the LSMA pens to 20. Ferrets were maintained under the standard care protocol (Wild and Castle 1999). Females were paired with males in spring 2000, and four litters resulted. One litter was apparently consumed by the female when about 1 day of age. The other three litters yielded 11 kits. Flea Control In Prairie Dogs We presented results of the lufenuron dose titration study at the 2000 meeting of the Wildlife Disease Association, Jackson, Wyoming. The abstract of that presentation read: DOSE-TITRATION AND SAFETY OF LUFENURON FED TO CAPTIVE WHITE-TAILED PRAIRIE IX>GS (CYNOMYS LEUCURUS) KEVIN T. CASTLE Colorado Division of Wildlife, 317 W. Prospect St. Fort Collins, CO, 80521, MARGARET A. WILD, Colorado Division of Wildlife, 317 W. Prospect St. Fort Collins, CO, 80521, and S. CRAIG PARKS, Novartis Animal Health, P.O. Box 26402, Greensboro, NC 27404. Plague is a zoonotic disease that impacts populations of prairie dogs (Cynomys spp.) and other species, such as black-footed ferrets, that rely on them for food and shelter. Yerstnia pestis, the etiological agent of plague, is transmitted primarily by the bite of an infective flea. Recently developed compounds used to control fleas in pet animals offer a promising alternative to insecticide dusts for the control of fleas in wild rodents. Lufenuron is a lipid-soluble insect growth regulator with ovicidal and larvicidal activity. Lufenuron is efficacious for controlling fleas in cats and dogs at blood concentrations above 50-100 parts per billion (ppb). A single oral dose oflufenuron has been shown to be effective in controlling the cat flea (Ctenocephaltdes felts felts) for at least 30 days in treated cats and dogs. To date, there have been no studies conducted to detennine the duration oflufenuron blood concentrations in any prairie dog species. We compared lufenuron blood concentrations in white-tailed prairie dogs (C. leucurus) during periods of activity (nontorpid group) and hibernation (torpid group) during January-March 1999. We hypothesized that if high serum concentrations of lufenuron could be maintained over winter during hibernation or for > 1 mo in active prairie dogs, the compound may be effective for use in breaking the plague cycle. Thirty captive WTPD were fed 300 rrig/kg lufenuron; half the animals were allowed to become torpid, while the other half were kept awake. All animals remained healthy throughout the 9 week study period. Prairie dogs in the active group gained weight, while those in the torpid group lost weight over the 9 weeks. Blood was drawn from each animal prior to dosing, one week after dosing, then every other week until week 9 post-dosing. Serum was harvested and tested by HPLC for lufenuron concentration. Blood lufenuron concentration did not differ between the groups one week post-dosing. Concentration in both groups decreased over time, but the �120 concentration in torpid animals declined at a more gradual rate; after weeks 3, 5, and 7, lufenuron levels in torpid WfPD were significantly higher than levels in nontorpid WfPD. After nine weeks, blood levels were again similar, and had approached the limit of detection (10 ppb). Blood levels in nontorpid WfPD declined to <50 ppb after 3 weeks, while levels in torpid WfPD declined to <50 ppb after 7 weeks. Future studies will be required to determine efficacy of lufenuron in controlling fleas on WfPD. If effective blood concentrations are similar to dogs and cats, however, frequent dosing would be required to control flea numbers on prairie dogs and thus break the plague cycle. Results from this experiment indicated that blood concentrations of lufenuron did not reach initial levels as high as anticipated, nor were the concentrations maintained above 50 ppb for as long as anticipated (Fig. 4). The most likely cause of these low concentrations was poor absorption of the drug by prairie dogs. lbis may be due to the difference in gut morphology between rodents and carnivores. Alternatively, other aspects of pharmacokenetics may have been responsible for the low serum levels in prairie dogs despite dosing at rates 10-30 times higher than those recommended for cats and dogs, respectively. We followed up the dose titration experiment with an experiment to test the efficacy of oral lufenuron to control fleas on captive prairie dogs. Based on data from the bioavailability study, we increased the lufenuron dose to 500 mg/kg. Serum samples from the study have been submitted for lufenuron assay. Results are pending. Interpretation of efficacy results will rely on these serum lufenuron levels; however, we were able to make some preliminary comparisons between performance of fleas fed on treatment and control prairie dogs. One week post-dosing, fleas that fed on treated prairie dog produced a lower proportion (p < 0.05) of viable eggs than those fed on control prairie dogs (mean 0.16 vs. 039, respectively). Unfortunately, the proportion of viable eggs produced during week 2 post-dosing was low for each group (0.24 treatment vs. 0.17 control). By week 4 post-dosing, egg production had fallen to nearly zero in each group; however, the three eggs produced by fleas fed on treated prairie dogs were all viable. We are uncertain of the cause of the dramatic reduction in egg production and viability observed during the course of the experiment. Anecdotal reports suggest that flea production may be influenced seasonally (or at least cyclically) despite attempts to maintain controlled environmental conditions in the insectary (Metzger, Pers. Comm.). Regardless, it is unfortunate that serum levels oflufenuron decreased more rapidly than we had expected and that data did not support the hypothesis that lufenuron would be an effective means to significantly reduce flea production over the summer. Although further, similarly controlled investigation is warranted, preliminary results suggest that given current techniques, the likelihood of lufenuron limiting fleas populations and breaking the plague cycle is not extremely promising. Therefore, experiments into efficacy of controlling flea infestations in simulated burrow environments and in the field (described in Wild and Castle 1998) will not be performed. LITERATURE CITED Wild, M.A. and K. T. Castle. 1998. Monitoring and managing disease in black-footed ferrets. Colorado Div. Wildl. Res. Rep., 0880-1, Jul 1997 - Jun 1998, Fort Collins. Wild, M.A. and K. T. Castle. 1999. Monitoring and managing disease in black-footed ferrets. Colorado Div. Wildl. Res. Rep., 0880-1, Jul 1998 - Jun 1999, Fort Collins. �121 Attachment l Kevin T. Castle, Margaret A. Wild, and S. Craig Parks Colorado Division of Wildlife Foothills Wildlife Research Facility (KTC and MAW) Novartis Animal Health, Greensboro, NC (SCP) Introduction Black-footed ferret (Mustela nigripes) recovery plans call for the reintroduction of ferrets to sites characterized by the presence of viable populations of prairie dogs (Cynomys spp.), which provide food and shelter for ferrets. Unfortunately, prairie dogs inhabiting many potential reintroduction sites carry fleas that can serve as vectors ofYersinia pestis, the causative agent of plague (Ubico et al, 1988). Prairie dogs and ferrets are both highly susceptible to plague (Barnes, 1993; Williams et al., 1994) so the chances of successful ferret reintroduction will be enhanced if the numbers of fleas infesting a prairie dog colony can be significantly reduced. Lufenuron is a benzoylphenylurea derivative which inhibits formation of chitin in the exoskeleton of insects (Cohen, 1987). A single oral dose oflufenuron has been shown effective in controlling the cat flea (Ctenocephalides felts felis) for at least 30 days in treated cats (Blagburn et al., 1994) and dogs (Hink et al., 1994; Blagburn et al., 1995). No studies on the efficacy oflufenuron have been conducted with prairie dogs; however, Davis ( 1997) reported a significant reduction in fleas on free-ranging ground squirrels (Spermophilus beecheyi) that had been treated with lufenuron. We are performing a series of experiments to test the applicability oflufenuron to control fleas in captive, and ultimately wild, white-tailed prairie dogs (Cynomys leucurus). Thus far in pilot studies we have determined standard husbandry, maintenance, and handling protocols for captive prairie dogs and determined a test dose oflufenuron based upon a bioavailability study. We have also attempted to establish an insectary colony of a flea species (0ropsylla tuberculata) that naturally infests prairie dogs and their burrows. Oropsylla fleas are among the most common prairie dog fleas, and have been implicated in the transmission of plague in prairie dogs (Ubico et al., 1988). Fleas of this genus are nest fleas that infest the host only to obtain a blood meal. At other times the fleas select microenvironments within in the burrow system that are conducive to successful reproduction and survival. In pilot studies we were unable to develop methods to successfully maintain and produce self-sustaining populations of 0. tuberculata in an insectary. However, a related species, 0. montana, which naturally infests ground squirrels (e.g. Spermophilus beechyi) has been successfully maintained in the insectary using methods of M. Metzger and K. Gage (pers. comm). These fleas will feed on prairie dogs and successfully reproduce after ingesting a blood meal. Because of its similarity to 0. tuberculata, we will use 0. montana as a model to determine if flea reproduction can be controlled in lufenuron-treated prairie dogs. A chambered-flea system has been used to test on-host viability and fecundity of fleas on cats (Thomas et al. 1996) and laboratory mice (D. Engelthaller, pers. comm.). In this system, fleas are contained in a chamber attached to the host, and can obtain a blood meal through a mesh screen. This system has several advantages over other methods of artificial infestation. First, because a known number of adults can be placed into the chamber, flea mortality is easily noted, and survivorship can be determined. Second, sex ratios can be adjusted to maximize egg production. Third, eggs can be recovered readily, to determine viability. Fourth, the environment in the chamber will be warm and humid, and therefore conducive to adult and egg survival. Finally, fleas will not be able to leave the prairie dog, and therefore will not be able to infest caretakers or other animals. �122 In this study we will: 1) develop techniques for artificially infesting white-tailed prairie dogs with chambered populations of fleas, 2) test the flea control efficacy oflufenuron fed to white-tailed prairie dogs artificially infested with fleas, and 3) compare efficacy of flea control with serum lufenuron levels. Our working hypothesis is that fleas which feed on lufenuron-dosed prairie dogs will have reduced survival of eggs and larvae compared to fleas which feed on control animals that receive no lufenuron. Methods Prairie Dog Maintenance We will use 31 captive white-tailed prairie dogs maintained at the Foothills Wildlife Research Facility in Ft. Collins in our experiment. Prairie dogs will be housed singly (if used in experiments) or in pairs (if used for flea colony maintenance) in custom-designed cages (100 cm x 500 cm x 600 cm; Wild and Castle 1998). Prairie dogs will be observed daily, and will have ad libitum access to Teklad Rodent Blocks and water. Windows will provide a natural photoperiod, and a combination of timed heaters and air conditioners will be used to keep ambient temperature between 10 and 25° C. Twenty prairie dogs (10 males and 10 females) will be blocked by sex and randomly divided equally into a treatment group and a control group for the experiment. The treatment group will receive lufenuron while the control group will not. Because all animals cannot be housed in one building due to space limitations, the groups will be split evenly between two separate but similar buildings, to help control for any interbuilding differences (e.g. temperature, humidity) that may exist. The sample size of20 prairie dogs will allow us to detect a ~91 % reduction in the number of eggs that successfully hatch in the treated group given alpha = 0.10 and beta = 0 .90. An additional 11 captive prairie dogs will be maintained under similar conditions, but will not participate in the experiment. Instead, they will be used in the maintenance of the flea colony (see below). Lufenuron Dosing Prairie dogs in the experimental group will be fasted for 24 h prior to dosing; water will be available during the fast. After the fasting period (day 0), each experimental animal will be weighed, then offered a bolus dose oflufenuron (300 mg/kg body mass). Lufenuron will be mixed thoroughly in approximately 10 g of highly palatable bait (ground rat chow and molasses); control animals will receive bait only. We previously determined that a majority of fasted prairie dogs consume about 90-95% of their dose in less than 12 h. We will observe the progress of bait ingestion in each animal to determine when the dose is ingested. Remaining bait will be removed after 24 h. We will weigh the bait/lufenuron mixture before and after the prairie dogs are dosed, in order to calculate the actual dose ingested. After dosing, normal feeding will resume. Flea Infestation An initial stock of laboratory-reared, disease-free fleas (0. montana) will be obtained from Dr. Kenneth Gage at the Centers for Disease Control and Prevention (CDC) in Ft. Collins. Fleas will be housed in 500 ml glass jars containing larva-rearing media, to allow the population to be self-sustaining . Rearing media consists ofwheast (Red Star Biologicals), dried beef blood (Monfort Biologicals), powdered dog chow, and sand. Fleas will be maintained in an incubator at about 22-23° C and ~70% relative humidity (M. Metzger, pers. comm.) to ensure optimal reproduction. A natural photoperiod will be approximated within the incubator using fluorescent lights and a timer. Adult fleas must ingest a blood meal in order to reproduce. We will use two methods to provide blood meals to fleas held in the insectary for propagation of the colony. The first 1:11ethod will follow an established protocol (Castle and Wild 1998) to provide blood to the adult fleas in each rearing chamber, using neonatal rodents. Briefly, when fleas are in need of a blood meal (1-2 times per week), we will obtain �123 neonatal rodents from a private colony. The neonates will be placed into the insectaries with fleas for up to 24 h; previous work has shown that over 80% of the neonates are alive after 24 h. No food or water will be provided for the neonates, as they are strictly dependent on nursing. After feeding by the fleas, neonates will be euthanized by an overdose of inhalant anesthetic. Once per week, for 7 weeks, we will utilize the 11 non-experimental prairie dogs as blood sources, using the chambered flea technique described below. While the use of neonatal rodents is an efficient, approved method of providing blood to fleas, neonatal rodents are not always available from private colonies. Prairie dogs will therefore serve the dual purposes of providing blood meals when neonates are unavailable, and minimizing the number of neonatal rodents sacrificed. One week prior to study initiation, and on study days 2, 7, 14, 21, 28, 35, and 42 we will place flea chambers on experimental prairie dogs. Fleas feeding on treated prairie dogs will potentially be exposed to lufenuron from this blood meal. W.e will collect 50 adult female and 30 ad~lt male fleas from the insectary and place them"into a chamber (2.5 cm diameter). Each chamber will be-attached to a prairie dog so the fleas can obtain a blood meal. To attach the chambers, prairie dogs will be anesthetized using isoflurane delivered by a vaporizer. Respiration and depth of anesthesia will be monitored. A patch of fur on the dorsal thorax caudal to the shoulders will be shaved. A flea chamber will be placed on the skin, and taped into place using Elastikon and Vet-rap. The prairie dog will then be placed in a 30 cm x 20 cm x 20 cm holding box to recover from anesthesia, and will be monitored while the chamber is attached. Chambers will remain on each prairie dog for 30 min. At the end of the feeding time, the prairie dog will again be anesthetized with isoflurane, and the tape and chamber will be removed. Prairie dogs will be returned to their cages after recovery from anesthesia. Fleas will be observed for evidence of feeding by observation under a 10-20x ,;nicroscope, and blood-filled fleas will be placed in plastic vials and put into our insectary for egg recovery. 0. montana fleas typically lay eggs within 2-3 days of a blood meal (M. Metzger, pers. comm.), so adult dishes will be monitored every day for 7 days to monitor egg production. Eggs will be removed from the adult dish and placed in new vials inside the insectary; they will be observed every day for larval emergence. The total number of eggs produced by fleas from each prairie dog will be recorded, as will the number of larva that emerge each ·day. Unfed adult fleas will be returned to the insectary (prior to lufenuron dosing) or preserved in alcohol (after lufenuron dosing). Blood Collection Concentration of lufenuron in the blood may be an indicator of efficacy of flea control. To determine the relationship between blood lufenuron concentration and egg viability and larval development, we will collect 3 ml of blood from each anesthetized prairie dog prior to chamber attachment on each sampling day. Blood will be collected by jugular venipuncture and placed into a glass vacutainer blood tube without anticoagulant. Alternatively, ifwe are unable to collect an adequate blood sample peripherally, we will collect blood from the vena cava or directly from the heart. Although cardiac puncture is considered to be a safe procedure for collection of large volumes of blood from laboratory rodents (CCAC, 1984), due to the increased risks involved with cardiac puncture, we will use this technique only to obtain critical samples. Serum will be harvested and frozen within 4 h of collection. Serum lufenuron concentration will i>C? determined by HPLC at Novartis Laboratories. Based on our pilot studies and other literature (Blagbum et al. 1994, 1995; Thomas et al. 1996) we do not anticipate any prairie dog mortality due to the flea infestations, lufenuron dosing regime or blood collection, but if a severe allergic reaction or other health problem associated with our procedures occurs, the prairie dog will be removed from the study and provided veterinary care or euthanized with an overdose of inhalant anesthetic or barbiturate. A complete post-mortem examination will be performed on any animal that may die during the experimental period. �124 Data Analysis Efficacy of lufenuron will be determined by comparing developmental success of eggs produced by fleas exposed to treated prairie dogs vs. eggs produced by fleas exposed to control group prairie dogs. These formulas will be used in efficacy determination: Developmental success = number of larvae hatched x 100 number of eggs collected Percentage efficacy=mean developmental success (control) - mean developmental success (treated) x 100 mean developmental success (control) Mean developmental success values for eggs collected from prairie dogs at each sampling period will be analyzed using analysis of covariance, using blood lufenuron concentration at each time step as the covariate. Group responses will be considered significantly different ifp < 0.10. Literature Cited Barnes, A. M. 1993. A review of plague and its relevance to prairie dog populations and the black-footed ferret. Proceedings of the symposium on the management of prairie dog complexes for the reintroduction of the black-footed ferret. J. L. Oldemeyer, D. E. Biggins, B. J. Miller, and R. Crete, eds. 96pp. Blagburn, B. L., J. L. Vaughan, D.S. Lindsay, and G. L. Tebbitt. 1994. Efficacy dosage titration of lufenuron against developmental stages of fleas (Ctenocephalides felis felis) in cats. American Journal of Veterinary Research 55: 98-101. Blagburn, B. L., C. M. Hendrix, J. L. Vaughan, D.S. Lindsay, and S. H. Barnett. 1995. Efficacy of lufenuron against developmental stages of fleas (Ctenocephalides felis felis) in dogs housed in simulated home environments. American Journal of Veterinary Research 56: 464-467. Castle, K. T. and M.A. Wild. 1998. Protocol for the use of neonatal rodents as a blood source for insectary-reared fleas. Colorado Division of Wildlife Animal Care and Use Committee Study Plan. Canadian Council on Animal Care (CCAC). 1984. Guide to the care and use of experimental animals, Vol. 2. Ottawa, Ont., Canada. Cohen, E. 1987. Interference with chitin biosynthesis in insects. In Chitin and benzoylphenyl ureas, series entomologica, vol. 38, J.E. Wright and A. Retnakaran (eds), Dr. W. Junk, Publishers, Boston, pp.3342. Davis, R. M. 1997. Use of an orally administered insect development inhibitor (lufenuron) as a flea control agent in the California ground squirrel, Spermophilus beecheyi. Fourth International Symposium on Ectoparasites of Pets, pp. 31-35. Hilton, D. F. J. 1971. A method for rearing fleas of ground squirrels. Transactions of the Royal Society of Tropical Medicine and Hygiene. 66: 188-189. Hink, W. F., M. Zakson, and S. Barnett. 1994. Evaluation of a single oral dose oflufenuron to control flea infestations in dogs. American Journal of Veterinary Research 55: 822-824. Thomas, R. E., L. Wallenfels, and I. Popeil. 1996. On-host viability and fecundity of Ctenocephalides felis (Siphonaptera: Pulicidae), using a novel chambered flea technique. Journal of Medical Entomology 33: 250-256. Ubico, S. R., G. 0. Maupin, K. A. Fagerstone, and R. G. McLean. 1988. A plague epizootic in the whitetailed prairie dogs (Cynomys leucurus) of Meeteetse, Wyoming. Journal of Wildlife Diseases 24: 399-406. Williams, E. S., K. Mills, D.R. Kwiatkowski, E.T. Thome, and A. Boerger-Fields. 1994. Plague in a black-footed ferret (Mustela nigripes). Journal ofWildlife Diseases 30:581-585. Wild, M.A. and K. T. Castle 1998. Monitoring and managing disease in black-footed ferrets. Colorado Division of Wildlife Program Narrative, Project# W-153-R. \ ·\,,,_) �125 20 ~ - - - - - - - - - - - - - i ■ Plague-Neg>----------~ □ Plague-Pos 15 L. (1) ..0 E 10 ::::, z 5 1997-W 1997-S 1998-W 1998-S 1999-W 1999-S Fig. la. Prevalence of exposure to plague in juvenile coyotes from the Little Snake Management Area, Colorado, from winter 1997 through summer 1999. II Plague-Neg 20 □ Plague-Pas 15 Q) .0 E 10 :::J z 5 0 -+------'----'-~--'----'----'----------'-----.J'---..L.---'-----'-----'-----'---J 1997-W 1997-S 1998-W 1998-S 1999-W 1999-S Fig. lb. Prevalence of exposure to plague in adult coyotes from the Little Snake Management Area, Colorado, from winter 1997 through summer 1999. Data from winter 1997 (age class unknown) provided by M Albee. �126 25 - . - - - - - - i □ CDV positive II CDV negative 20 t> 15 J 10 5 0 1997-W 1997-S 1998-W 1998-S 1999-W 1999-S Fig. 2. Prevalence of exposure to canine distemper virus (CDV) in coyotes from the Little Snake Management Area, Colorado,' from winter i 997 through smnmer 1999. All positive coyotes were adults with the exception of one juvenile in summer 1998. Data from winter 1997 (age class unknown) provided by M. Albee. 20 , - - - - - - - - - - - , Ill Negative f - - - - - - - , □ Positive 15 .BS 10 z= 5 Plague Fig. 3. Prevalence of exposure to plague and canine distemper virus in adult coyotes from the Wolf Creek site, Colorado in winter 2000. 200 C: 0 175 "ia = 8 1-, 150 . _J25 C: .0 8 ~()() e= 75 C: 50 = 25 ca ...:I • . • • •• • 0 -t-----,--------.-------.----,,---,----,----r----r---, 2 3 4 5 6 7 8 9 10 Week Post-Dosing Fig. 4. Mean serum lufenuron concentrations of active ( ♦) and torpid (■) prairie dogs orally dosed with 300 mg/kg lufenuron. � https://cpw.cvlcollections.org/files/original/e26ad9d611c82f13cd9dee29ba29b8bb.pdf 4efbf3f5e4e8b5886e2515e7e8d4363d PDF Text Text 215 \ I ,/ Colorado Division of Wildlife Wildlife Research Report July 2000 I \ JOB PROGRESS REPORT State of _ _ _ _ _-=C=o=lo=ra=d=o'------- Cost Center 3430 Project _ _ _ _.....W-'---"-1=5=-3-"""'R=--=13~---- Mammals Research Work Package ---"""'3::....:0:....:;0""'1_ _ _ _ __ Deer Management Study No. -------=RMA==-=------- Technical Support for Deer Population Management at the Rocky MoW1tain Arsenal National Wildlife Refuge, Denver Colorado Period Covered: July 1, 1999 - June 30, 2000 Author: Dan L. Baker \ _) Personnel: M.A. Wild, T. M. Nett, J.C. Ritchie, D. Finley, M. Conner, L. Wheeler ABSTRACT We provided technical support for the management of mule and white-tailed deer populations at the Rocky Mountain National Wildlife Refuge, Denver, Colorado. Deer population demographics information was summarized from 1995-1999 and used to develop a simulation model to assist wildlife managers in evaluating alternative strategies for population control. As an alternative to annual culling of overabundant deer, we developed and tested a fertility control technology in captive mule deer. This contraceptive agent , was demonstrated to be 100 % effective in preventing pregnancy when delivered prior to the breeding season. Effective duration of the contraceptive was approximately 153 days. No negative effects on the nutrition, physiology, and general health of the treated animals were observed, however, breeding behavior of treated females was significantly (P :S 0.001) different from untreated co~trols. \ ) J m~i11~1i1mrnmnr BDOW016784 ��217 TECHNICAL SUPPORT FOR DEER POPULATION MANAGEMENT AT THE ROCKY MOUNTAIN ARSENAL NATIONAL WILDLIFE REFUGE, DENVER, COLORADO Dan L. Baker P. N. OBJECTIVES l. Provide technical support for the management of mule deer, white-tailed deer populations and their habitats at the Rocky Mountain Arsenal National Wildlife Refuge (RMA). 2. Develop a practical and acceptable technology to inhibit reproduction in mule deer and whitetailed deer populations at the RMA. 3. Develop a simulation model to predict impacts of alternative population control strategies at RMA. 4. Demonstrate the feasibility of population control strategies in a field application. SEGMENT OBJECTIVES l. Provide professional consultation relevant to the development of a deer management plan at the Rocky Mountain Arsenal National Wildlife Refuge. I 2. Develop and test a practical and acceptable technology to inhibit reproduction in captive mule deer. 3. Evaluate the effects of contraceptive technology on mule deer nutrition, physiology, general health, and breeding behavior. INTRODUCTION Deer Population Management Plan The Rocky Mountain Arsenal National Wildlife Refuge (RMA) has the potential to become one of the premier wildlife viewing areas in the United States. However, realizing that potential depends on wise management. In particular, the mule deer population contained within the boundary fence must be regulated in balance with the resources available. Experience with enclosed deer populations elsewhere has revealed that unregulated population increases will lead relentlessly to degradation of habitat, to widespread starvation, and eventually to catastrophic declines in animal numbers. Professional culling and fertility control offer alternative management strategies for regulating deer populations at the RMA. However, until fertility control technology is demonstrated to be a feasible and efficacious method for controlling free-ranging wild ungulate populations, annual culling of mule deer is the only viable management option available to resource biologist.at the RMA. Using either culling or contraceptives to meet deer population objectives will require, wildlife managers to choose specific tactics of treatment. Choices must be made on the number and age to treat, frequency of treatment, species, sex, etc. Decisions on the best tactics will depend on comparing the effects of �218 alternative management actions on population behavior. To assist in these decisions, we developed an interactive simulation model of deer population dynamics to allow managers to evaluate alternative treatment regimes on simulated populations before applying them to real animals. Critical to model performance is knowledge of population demographics of the RMA deer herds and requires information on sex and age composition, recruitment, pregnancy rates, fetal sex ratio, and estimates of population density. 1bis information together with knowledge of the habitat resources available to deer will provide a sound biological basis for management of deer populations at the RMA. Deer Fertility Control Contraception offers a viable alternative to culling as means oflong-term control of overabundant wild ungulates. During the last decade, research aimed at developing effective contraceptives for free-ranging wildlife populations has accelerated. These efforts have resulted in development and testing of a wide variety of potential contraceptive agents (Kirkpatrick and Turner 1985, Warren et al. 1995). One of the most promising new non-steroidal, non-vaccine, approaches to contraception involves synthetic analogs of gonadotropin-releasing hormone (GnRH). GnRH is a molecule produced in the hypothalamus of the brain. It directs specific cells in the pituitary gland to synthesize and secrete two important reproductive hormones; follicle stimulating hormone (FSH) and luteinizing hormone (LH). These latter two hormones, known as gonadotrophs, control the proper functioning of the ovaries in the female and testes in the male. Inhibition of ovulation caused by chronic administration of GnRH analogs has been successful in several species, including dogs (Vickery et al. 1989), cattle (Herschler and Vickery 1981), sheep (McNeilly and Fraser 1987), white-tailed deer (Becker and Katz 1995), and elk (Baker and Nett, unpublished data). Evidence from studies confirm that sustained release is the most effective approach for temporarily suppressing pituitary-gonadal function. The practicality of this approach, however, is dependent upon development of a long-acting, slow-release analog that can be remotely delivered. Recently, a practical mode of administration using subcutaneous implants has overcome the need for constant mechanical infusion of the analog. Slow release formulations of superactive GnRH analogs are now commercially available and have been shown to be effective in suppressing the pituitary-ovarian axis for up to 6 months in a variety of mammalian species (Fraser et al. 1987). To our knowledge, only limited investigations have been conducted to evaluate the effectiveness ofGnRH analogs in wild ungulates (Becker and Katz 1995, Baker and Nett 1998 unpublished data), however, the minimum dose of GnRH analog required for maximum pituitary stimulation is known for mule deer (Baker et al. 1995) and the minimum effective duration of controlled release GnRH-agonist implants has been determined to be at least 130 days in captive elk (Baker and Nett 1999, unpublished data). Additional research is needed to determine the effectiveness of these contraceptive agents in preventing pregnancy in mule deer and their maximum effective duration. Research is also needed to identify any nutritional, physiological or behavioral side-effects that may result from treatment. Thus, the objectives of our research are: 1) To evaluate the effectiveness and effective duration of GnRH analog in preventing pregnancy in mule deer. 2) To evaluate the effects ofGnRH analog on nutrition, physiology, general health and breeding behavior of captive mule deer. �219 We will test the null hypothesis ofno effect of GnRH analog on luteinizing hormone (LH) levels, pregnancy rates, social behavior, and general health of captive female mule deer. MATERIALS AND METHODS Deer Population Management Plan. The primary deer population management objective for the RMA is to maintain a healthy and productive mule and white-tailed deer population that is in balance with its habitat resources. Specific objectives include: 1) Manage total deer population levels at approximately 50% of habitat carrying capacity with a mule deer : white-tailed deer population ratio of 2: 1. The total winter population goals will be 550-650 mule deer and 250-350 white-tailed deer. 2) Manage sex and age composition of deer to achieve a high proportion of mature males relative to adult females. The sex and age composition goals will be 90 to 120 bucks:100 does with at least 50% of males older than 2 years of age. The status (population siz.e, condition, general health) of the deer population at the RMA was evaluated on an annual basis from 1995 to 1999 using the following demographic information. This information was incorporated into a culling/fertility contrbl simulation model that allowed RMA biologist to evaluate alternative population control strategies for meeting deer population objectives. Census. A total count of all deer on the RMA was conducted each winter from 1989 to 1998. Procedures were standardized for all counts by 1) conducting at least two consecutive aerial surveys each winter, 2) conducting surveys after trees and shrubs had dropped their leaves, 3) conducting surveys when snow cover was at least 95-100 %, 4) using the same 2 observers whenever possible, and 5) using a Bell Soloy helicopter that was flown at an average ground speed of 60 km/hr at altitudes from 24 to 35 m above ground. The helicopter flew transects approximately 400 m apart beginning on the south perimeter and preceding north along east to west transects. Observers searched areas on their respective sides of the helicopter. Temperature, visibility, and percentage snow cover were recorded for each flight. Deer were tallied by species but not classified by sex and age. Sex and Age Classification. Sex, age, and species composition of deer at the RMA were sampled each November. Three permanent survey routes were established by dividing the RMA into 3 separate geographical areas and identifying a permanent route within each area. Each route followed established roads and was chosen to observe the maximum number of deer within the area. The length of each route was determined by the driving distance that could be covered by one vehicle during the period 0700-1200. One pair of previously trained observers sampled each of the three areas each day~ the same pairs of observers were consistent across all days. Classifications were conducted at the peak of the breeding season (approx. Nov 22) to increase the probability of observing all age classes. Observations were made from a vehicle using 8 x 50 binoculars and 10-60x. spotting scopes. No classifications were attempted for deer that were farther than 200 m from the observer or for deer that were bedded. Deer were classified by species as either mule or white-tailed deer, and sex, and age as immature buck, mature buck, doe, fawn, and abnormal buck (Dasman and Taber 1956). Each year, prior to conducting surveys, observers independently classified the same deer from as many groups of deer as required to consistently obtain identical classifications. �220 Reproductive Parameters. We detennined pregnancy rates, fetal rates, fetal sex ratio, and average conception date for adult female mule deer collected from 1995 to 1999. Deer were shot through the spine in the cervical region with a high powered rifle at distances ranging from 30 - 75 m. Reproductive tracts were removed and examined for evidence of fetuses. Does lacking identifiable embryos or fetuses after January I were regarded as current breeding failures. If corpora lutea were present, we used only those does in which there was gross evidence, through presence of embryos or pigmented degenerating corpora lutea, that the doe either had been or was currently pregnant. During 1994 we measured fetal growth rates to estimate breeding dates. Adult does were collected at 4 time periods during gestation: January 25, February 16, March 9, and April 19. -Fetuses from each doe were removed in the field and transported to the laboratory for determination of sex and age. Fetal foreheadrump (mm), hind leg length (mm), and body mass (g) measurements were made according to methods described by Armstrong (1950). Measurements of twins and triplets were averaged to provide a single estimate of fetal age for each doe. We estimated the age of each fetus by using the linear regression of forehead-rump length (mm) (Y) and age (days) (X) established from known-age mule deer fetuses (Y = 124.6 + 3.2lx) (Hudson and Browman 1959). Culling/Fertility Control Model. We developed an interactive simulation model to allow RMA biologist to evaluate alternative mule deer culling strategies. This model combined extant knowledge of mule deer population biology with measured population parameters at the RMA to predict the proportion of male and female deer that need to be removed each year to accomplish management objectives. Table 1. Principal parameters in culling simulation model for mule deer at RMA. Reference Definition Value Measured Whittaker, 1992 Adult/yearling female survival 0.90 No Whittaker, 1992 Adult/yearling male survival 0.90 No Bartmann et al., 1992 Fawn survival coefficient a: 1.20 No Bartmann et al., 1992 Fawn survival coefficient P -0.053 No Fetal sex ratio(% males) 0.47-0.62 Yes December recruitment 0.59-0.72 Yes Deer Fertility Control Mule deer (OdocoUeus hemionus hemionus) are polyestrus owlators that exhibit highly seasonal patterns of reproduction that are controlled by photoperiod regimens. The onset of the breeding season occurs during decreasing daily photoperiods of autumn and is preceded by a period of deep anestrous in summer (Plotka et al.1977). The first ovulation of the breeding season is usually preceded by one or more silent ovulations associated with the formation of short-lived corpora lutea that serve to synchronize the first overt estrus within a herd (Thomas and McT.Cowan 1975). In temperate North America, the majority of conceptions occur in late November, but recurrent estrous cycles of 24 -28 days are possible through March if females fail to conceive (Knox et al. 1988). In early spring, coincidental with increasing day length, reproductive cycles cease and females remain anestrous until October. For pregnant females, parturition generally occurs in late May or early June, after a gestation period of about 200 days (Anderson and Medin 1966). Experimental Animals. We conducted two experiments to evaluate GnRH analog in captive tame and wild female mule deer. Experiment I was a highly controlled experiment using IO tame, adult females and 2 wild, adult male mule deer. Experiment 2 attempted to simulate the appli~on of contraceptive treatments �221 to free-ranging mule deer at the RMA. This effort involved 9 wild, adult female mule deer and 5 wild, adult male mule deer. These studies were conducted from November 3, 1999 until August 15, 2000 at the Colorado Division of Wildlife's Foothills Wildlife Research Facility (FWRF), Fort Collins, Colorado. Experiment 1: Protocol Reproductive Status. Before randomly assigning animals to treatment groups, we detennined the reproductive status of all does by measuring serum progesterone levels at weekly intervals beginning November 10. For this sampling, deer were moved from 5 ha pastures to individual isolation pens, sedated (40-100 mg xylazine (100 mg/ml), IM) and blood samples collected (5ml). Animals were reversed with yohimbine (0 .125 mg/kg, IV) and returned to paddocks the same day. Sedation of deer was done in order to minimize potential adrenal secretion of exogenous progesterone during handling and blood sample collections (Plotka et al.1983, Asher et al. 1989). Treatments. Following fertility evaluations, 5 does were randomly selected to receive an analog of GnRH (LUPRON) and 5 does were assigned to a control group. Treatment does \,\'.ere given a subcutaneous implant containing 10 mg of LUPRON one week prior to being exposed to male deer. Based on previous studies with mule deer (Baker et al. 1994), and elk (Baker and Nett 1999), approximately 3-4 animals per treatment are sufficient to establish statistically significant differences among treatment means. Hormonal Evaluation. Prior to application of the contraceptive treatment, we measured the LH response of each doe in both LUPRON and CONTROL groups to a challenge dose of GnRH analog (1 ml/50 kg BW, IV) . Results from this trial provided a pretreatment baseline for comparison to future posttreatment LH responses. This and succeeding LH 1challenge trials were conducted as follows: On Day 1 of the trial, mule deer were moved from 5 ha pastures to individual isolation pens, sedated (4-6 ml, 5: 1 ketamine (lOOmg/ml):xylazine hydrochloride (100 mg/ml, IM), and fitted nonsurgically with indwelling jugular catheters. Animals were be reversed with yohimbine (0.125 mg/kg, IV). On Day 2, we administered GnRH analog (1µ /50 kg BW) through the cannula and collected blood samples (5 ml) at 0, 60, 120, 180, 240, 300, 360, and 480 minutes postinjection. Following the last blood collection, catheters were removed and each animal given Naxel (3 ml/deer, (50 mg/ml, I.V.). Animals were then returned to 5 ha pastures. Serum was stored at - 20 °c until analyzed for LH. The duration of contraceptive effectiveness was assessed by conducting GnRH challenge trials each month from November 1999 to April 2000. Hormonal Assays: After collection, blood was held at 4 °C for 24 h until serum was obtained by centrifugation. Serum progesterone concentrations was determined using RIA procedures (Niswender 1973). Sensitivity of the progesterone assay was 0.12 ng/ml. Female mule deer with progesterone levels above 1 ng/ml were considered reproductively active (Plotka et al. 1977). 1990). Mule deer with progesterone levels below 1 ng/ml were sampled 7 days later. If after three consecutive blood collections, progesterone levels remained lower than 1 ng/ml the animal was be removed from the experiment. Serum concentrations of LH were quantified by means of ovine LH RIA (Niswender et al. 1969). The limit of sensitivity of the assay was 0.4 ng/ml. Response Measurements. The effects of LUPRON on female mule deer were assessed by comparing responses of treated and control deer by the following methods: a) Hormonal response. Responsiveness of the pituitary to GnRH challenge were determined in three ways: 1) maximum LH (ng/ml) response achieved postinjection minus baseline, 2) time (hr) required to reach maximum LH, and 3) total amount ofLH secreted (ng/ml/min). b) Pregnancy rates. We assessed contraceptive effectiveness by determining the pregnancy rates in treated and control deer. A single blood sample (10 ml) was taken via jugular venipuncture from each animal for �222 pregnancy- specific protein B (PSPB) analysis approximately 60, 90, and 180 days post-conception (Willard et al. 1998). Animal handling and blood collections for PSPB follpwed methods previously described for hormonal assessment and were collected in conjunction with these measurements. Neonates born to any experimental animal were incorporated into the resident FWRF mule deer herd with the exception of fawns born after August 15. Because late born fawns may experience higher than nonnal over-winter mortality, and potentially compromise body condition and short-term fertility of the dam, neonates were removed from the dam and either bottle-raised or euthanized according to ACUC procedures. c) General health. Knowledge of the effects ofGnRH analogs on nutrition, body weight dynamics, blood chemistry and general health of mule deer is unknown. Previous experiments with domestic livestock (Herschler and Vickery 1981), companion animals (Vickery et al. 1989), white-tailed deer (Becker and Katz 1995), and elk (Baker and Nett 1999) have reported no measurable short-term physiological effects. We evaluated these potential side-effects by monitoring body weight, blood chemistry, hematology and overall health of all female deer. We collected blood (10 ml) for blood chemistry and hematology prior to treatment, and at 30, 90, and 180 days posttreatment. We weighed deer at monthly intervals and monitored feed intake on a weekly basis .. d) Breeding behavior. The effects of the GnRH analogs on breeding behavior of mule deer are not known. However, if gonadotroph cells are desensitized by long-acting synthetic GnRH analogs, down-regulation and diminished LH and FSH secretion can be induced. Reduced secretion of LH and FSH could decrease or disrupt ovarian function, estrus cycles and secondary sex characteristics. Furthermore, if pituitary responsiveness returns following depletion of GnRH analog implants, treated females could resume estrous cycles and become pregnant late in the l1reeding season. This, in turn, could have survival consequences for both the pregnant female and her neonate. We evaluated these potential effects by monitoring maintenance and sexual behavior of male and female mule deer during the nonnal breeding period and at the end of the predicted depletion of the GnRH analog (March-April). Each animal in each treatment and the breeding males was individually identified using color-coded neck bands or ear tags. We tested the null hypothesis that the frequency of sexual interactions between males and treated females would be similar to those of males and untreated females. We tested this hypothesis using focal-animal sampling procedures and discriminant function analysis (Lehner 1987). The exact experimental design depended on time-of-day effects, number of observers, length of required sampling period, number of bucks in pasture, and the number of bucks observable by one observer. Statistical All;Olysis. We analyzed data using least squares ANOVA for General Linear Models and the SAS Interactive Matrix Language. Response to contraceptive treatments was analyzed with two-way factorial analysis of variance for a randomized complete block design with repeated measures structure. Levels of GnRH analog were treatments; individual animals were blocks. Factors in the analysis were treatment and time. Treatment effects were tested using the animal-within-treatment variance as the error term. Time was treated as a within subject effect using a multivariate approach to repeated measures. We used orthogonal contrast to test for differences among individual means (Morrison 1976, Miller 1966). Experiment 2 : Protocol Treatments. Nine captive wild adult female deer (previously captured and currently being used in a companion study under an approved ACUC protocol) were systematically assigned to either the control (4 animals) or LUPRON treatment (5 animals). During October 21 - November 1, in association with routine FWRF management procedures, control females plus 5 wild adult male mule deer were sedated (0.075 mg/kg medetomidine and 2 mg/mg ketamine or 4.4 mg/kg Telazol and 2.2 mg/kg xylazine) via projectile �223 darts, and individually marked. Male deer were marked with color-coded ear tags and females with colorcoded neck bands. All deer were reversed using atipamezol (0.38 mg.kg) or yohimbine (0.15 mg/kg). On November 3, treatment females were captured and individually marked (following the same tranquilization procedures as previously described), and given a subcutaneous implant of LUPRON (10 mg/animal). Based on previous studies with mule deer (Baker et al. 1994), and elk ( Baker and Nett 1999) approximately 3-4 animals/treatment are adequate to detect statistically significant differences between treatment means. Response Measurements. The effects of LUPRON on female mule deer were assessed by comparing the responses of treated and control deer by the following methods: a). Fawning rates. We determined contraceptive effectiveness by comparing the fawning rates of treated and control deer. Beginning in early June, all does were observed for behavioral and clinical evidence of parturition. In addition, fawning paddocks were searched daily for neonates. Any treatment female that had not fawned by August 15, was tested for pregnancy using PSPB analysis (Willard et al. 1998). Any fawn born after August 15 was removed from the dam and either bottle-raised or euthanized according to ACUC procedures. b) Breeding behavior. Procedures for comparing breeding behavior of control and treated does was similar to those described in Experiment 1. Statistical Analysis. Differences among fawning rates between control and treated females was analyzed using unpaired ttest (p < 0.05) (Steel and Torrie 1960). ! RESULTS AND DISCUSSION Deer Population Management Census. Mule and white-tailed deer populations generally increased from 1989 to 1998 with relatively large fluctuations in total numbers from year to year, particularly for white-tailed deer (Table 2). Mule deer numbers increased by 75 .6 % and white-tailed deer by 77.1 % during the eight year measurement period. For mule deer, most of the increase in population size occurred between 1990 and 1993. During this time period, the population grew at a rate of about 20% per year. After annual culling was implemented in 1994, the annual growth rate of the population was reduced to approximately 4% per year. Presently (1998), mule deer numbers (637) are within the population objective set for this species (550650). The white-tailed deer population at the RMA has shown large fluctuations in size from year to year. It's not known if these annual fluctuations are due to actual changes in population numbers, measurement error, or a combination of these factors. Regardless, white-tailed deer appear to be increasing with the greatest increase occurring between 1997 and 1998 (+ 44.9 %). The current (1998) population size (318) is within the management objective set for this species (250-350) and is near the population ratio objective of 2: 1 mule deer:white-tailed deer. Due to lack of snow cover no winter counts were conducted in 1999. Biologist at the RMA generally believe that the total counts of white-tailed deer are less reliable than those for mule deer because of the different habitat types preferred by this species and differences in behavior during helicopter surveys. Mule deer are generally more gregarious in winter than white-tailed deer and found in more open habitats where they are more easily observed and accurately counted. They appear to be less fearful of helicopter flights and tend to stay in one place during surveys, thus minimizing duplicate counts. In contrast, white-tailed deer are more solitary and generally found in dense vegetation near streams �224 Table 2. Total counts of mule and white-tailed deer at the RMA during 1989 to 1999. Mule deer White-tailed deer Year Number Number %Change %Change 1990 154 73 1991 200 +23.0 72 -1.4 1992 286 +30.0 35 -51.4 1993 397 84 +58.3 +27.9 1994 1995 550 +62.3 +27.8 223 1996 501 168 -24.6 - 8.9 1997 657 175 +4.0 +23.7 1998 632 318 +45.0 -3.8 1999 No winter counts and woodlands. They are more easily frightened by low the flying helicopter and tend to run when being counted. For population management purposes. we assumed that helicopter surveys accounted for 90-95% of the total mule deer population and 80-85% of the total white-tailed deer population. Ser and Age Classification. We determined the species, sex. and age composition of deer at the RMA during November 1996 to 1999 (fable 3). An average of 466 (se ± 58) individual mule deer and 153 (se ± 22) white-tailed deer were observed and classified each year during this period. Mule Deer: Bucks. The proportion of male mule deer in the population remained relatively constant ( x = 42%, se ± 1.8) from 1996 to 1998 (fable 3). However, the composition of young and mature males in the herd changed over this same time period. The proportion of young male mule deer increased from 8% of the total population in 1996 to 17% in 1998 while mature males declined from 35% to 23%. Most of the decline occurred between 1996 and 1997. The increase in young males in the population is also reflected in the total male mule deer population. The ratio of young males:mature males increased from 26 to 72:100 mature males between 1996 and 1998 (fable 3). Surprisingly, the increase in young males occurred despite selective culling of this cohort each year. Clearly, the trend over the last three years has been an increasing proportion of young males and a decreasing proportion of mature males. Reasons for this trend are unknown. Currently, mature males comprise 57 % of the total male population which meets the objective for this segment of the population. However. management intervention may be needed to reverse this trend in order to meet current deer population objectives. Mule Deer: Does. The proportion of adult does in the population has remained relatively constant (x = 31%, se ± 3.1) during 1996 to 1998 (fable 3). Buck:doe ratios declined by 20% from 1996 to 1998 in large part due selective annual culling of young males. The 1998 buck:doe ratio is within management objects for this segment of the population. Mule Deer: Fawns. The proportion of fawns in the RMA mule deer population remained relatively stable (x = 23 %, se ± 0.67) from 1996 to 1998 (fable 3). However, mean fawn:doe ratios over this time period have steadily declined (27 %) suggesting a substantial reduction in fawn recruitment into the population. Reasons for this decline are unknown and may need to be addressed in the near future. �225 Table 3. Sex and age comeosition of the mule deer eoeulation at RMA during November 1996 to 1998. Mule Deer 1998 1996 1997 Means Mean S.E. S.E. S.E. Mean Category: 71.3 Young bucks 36.3 1.7 87.3 6.8 4.4 Mature bucks Adult does Fawns Abnormal bucks Animals observed Groues observed Ratios: Bucks: 100 does Fawns:100 does Young bucks: 100 mature bucks 139.0 124.3 98.3 3.3 401.3 109.3 10.6 11.8 4.7 0.9 25.7 11.3 150.0 197.0 140.3 6.3 581.0 247.6 1.7 2.0 5.8 2.2 7.0 11.7 98.0 153.0 91.6 3.3 417.3 182.6 9.6 7.8 10.0 0.9 27.2 14.7 142 81 23 8 120 71 30 3 113 59 76 4 25 1 58 4 72 6 White-tailed Deer: Bucks. Unlike male mule deer, the proportion of male white-tailed deer in the total population increased from 30% 1996 to 42% 1998 (fable 4). Almost all of the increase occurred from 1997 to 1998. Similar to male mule deef, this increase was primarily associated with an increase in the proportion of young bucks and a decrease in mature bucks in the total buck population. As of November 1998, mature bucks comprised approximately 62% of the bucks in the total buck population. White-tailed Deer: Does. The proportion of adult white-tailed does in the RMA herd was relatively constant from 1996 to 1998 (x = 43%, se ± 1.5) (fable 4). However, with the apparent increase in bucks over this same time period the buck:doe ratio increased dramatically from 65 bucks: 100 does in 1996 to 103 bucks:100 does in 1998. White-tailed Deer Fawns. The proportion of fawns in the total population increased for 22% in 1996 to 27% in 1997 and 1998 (fable 4). Fawn:doe ratios fluctuated from between 40 and 65 fawns:100 does during 1996 to 1998. Reproductive Parameters Pregnancy Rates: Reproductive data was collected from 164 adult female mule deer from 1995-1999 (fable 5). Mean pregnancy rate during this period was 91.4 %(se ± 4.0) and ranged from a low of 80 % in 1999 to a high of 100 % in 1997 and 1998. Average fetal rate for pregnant does was 1.72 fetuses/doe (se ± 0.05), and 1.54 fetuses/doe (se ± 0.09) for all does combined. Fetal Growth Rates and Estimated Conception Dates: We observed a significant (P = 0.001) linear relationship between fetal age and fetal body length (Fig. 1). The fetal growth rate for mule deer at the RMA was similar (P = 0.65) to the growth rates reported for known age mule deer fetuses (Hudson and Browman 1959) and greater (P = 0.01) than the growth rates for mule deer fetuses in the Piceance Basin, Colorado (Bartmann 1986). A common use of fetal growth rate is to estimate breeding dates. The growth rate for mule deer fetuses derived from data of Hudson and Browman (1959) was applied to fetal data collected from the RMA. �226 Linear regression of estimated breeding date on collection dates was calculated to estimate the breeding season (Fig. 2). The slope of this regression line was not significantly different from 0 (P = 0.034) indicating that fetal length measurements are a predictable indicator of conception dates for the period of gestation sampled. However, it should be noted that this relationship was established for adult female mule deer in good to excellent body condition. Other studies have shown that estimated breeding dates will vary during the gestation period depending on winter severity and its effect on fetal growth rates, particularly during the last trimester of pregnancy (Bartmann 1986). Based on calculated conception dates, we estimate that 95% of the breeding occurred between November 21 and December 1. Conception dates ranged from November 14 to December 7. The breeding season for does at RMA appears to be somewhat earlier (November 25 - December 15) and shorter in duration than the breeding season reported for mule deer in the Cache la Poudre drainage of north-central Colorado and from studies in west central Colorado (Anderson and Medin 1966). Table 4. Sex and age composition of the white-tailed deer population at RMA during November 1996 to 1998. Mule Deer 1998 1996 1997 Mean• S.E. Mean S.E. S.E. Mean Category: 21.0 6.5 Young bucks 6.0 20.7 3.2 1.5 33.7 34.0 2.9 Mature bucks 10.4 39.7 3.5 53.0 2.08 Adult does 61.3 83.7 6.8 9.3 36.3 Fawns 29.7 52.7 7.4 3.9 1.ft 0.0 0.0 0.0 0.0 0.0 0.0 Abnormal bucks 130.3 6.01 131.7 197.0 17.3 Animals observed 24.3 73.3 4.8 46.0 6.9 105.3 3.3 GrouES observed Ratios: 103 8 Bucks:100 does 65 15 2 71 42 2 63 8 Fawn: 100 does 47 6 Young bucks : 20 1 65 22 100 mature bucks 51 3 Fetal Sex Ratios: We examined 251 mule deer fetuses for evidence of sex during 1995 to 1999 (fable 6). Of these, 140 were males and 114 were females resulting in a pooled fetal sex ratio of 56% males and 44% females. The pooled fetal sex ratio for mule deer at the RMA is within the range of values reported elsewhere for Rocky Mountain mule deer (Medin and Anderson 1979, Robinette et al. 1955) and was not significantly different from unity (P = 0.23). However, fetal sex ratios deviated from 50:50 in 2 out of 5 years and favored males over females (P = 0.023). In only 1 year out of 5 did the fetal sex ratio favor females. Disease: 1998. During August 1998, epiz.ootic hemorrhagic disease virus (EHD) was diagnosed as the probable cause of death in 19 white-tailed deer and 21 mule deer at the RMA. However, the extent of deer mortality from this disease is unknown and was probably much greater than the 40 individual deer that were actually found. Culling Model. Professional culling of adult mule deer at the RMA was implemented on a prescribed basis beginning in November 1994. Prior to 1994, the mule deer population was increasing at a rate of approximately 20 % per year. Since 1994, annual culling has reduced the population growth rate to 4% per year and allowed resource managers to meet deer management objectives for this herd. To meet these �227 Table 5. Pregnancy and fetal rates of adult female mule deer collected at the RMA during 1995 to 1999. Fetal Rate <fetuses/doe} Year 1995 1996 1997 1998 1999 Mean SE n Sample Size 39 46 26 38 15 Pregnancy Rate(%) 92.3 84.9 100.0 100.0 80.0 91.4 4.0 5 Pregnant Does 1.61 1.59 1.88 1.63 1.91 1.72 0.07 5 Table 6. Fetal sex ratio for adult mule deer on the RMA during 1995 to 1999. Year Sample size Males(%) 1995 58 60.4 1996 62 62.9 1997 49 53.2 1998 62 47.4 1999 20 60.0 Total Does 1.48 1.34 1.88 1.63 1.53 1.54 0.09 5 Females(%) 39.6 37.1 47.8 52.5 40.0 objectives has required that resource mapagers remove approximately 10- 15% of the adult females and 610% of the young males from the popul!tion each year. These estimates vary each year depending on productivity of the population, changes in sex and age ratios, and deer habitat resources. The mathematical model used to estimate the number of animals that must be culled annually to maintain a target, steady state population appears to provide a reasonably accurate representation of mule deer population dynamics at the RMA. The efficacy of this model, however, is dependent on intensive monitoring of the population and adapting management actions to a changing environment. Deer Fertility Control Pregnancy Rates and Hormonal Responses In both Experiment 1 and 2, the LUPRON subcutaneous implant was 100% effective in preventing pregnancy in all female mule deer (n = 9) that were treated. Untreated control deer (n = 10) all delivered normal fawns. Duration of effectiveness varied among deer and ranged from 118 days to 210 days. Serum progesterone (P4) concentrations of treated females remained below 0.1 ng/ml for at least 118 days posttreatment (Fig. 1). Serum P4 levels in two out of three does remained below 0.1 ng/ml for at least 174 days and for at least 210 days in one doe. Administration of the depot formulation of LUPRON resulted in a significant (P :!> 0.001) decline in serum luteinizing hormone (LH) concentration after 47 days posttreatment (Fig. 2). Similar to P4 levels, LH remained below 1 ng/ml until 153 days posttreatment. These results suggest that GnRH induced interruption of reproductive cycles and pregnancy is associated with diminished progesterone and LH production, implying defective function of the corpus luteum. These luteolytic effects have been attributed to the well recognized desensitizing actions of elevated LH levels on ovarian LH receptors. �228 Breeding Behavior We collected data for 59 sampling periods: 24 morning periods totaling 37.0 hrs, 18 mid-day periods totaling 36.0 hrs, and 17 evening periods totaling 52.2 hrs, for an overall total of 125.2 hrs of observation. Time of day, date, and the interactions were not significant effects and were.dropped from the ANOVA model. The behavior rate of does treated with Lupron SQ was not different than for control does for copulatory or general breeding behaviors (fable 3). However, male pre-copulatory and female precopulatory behavior rates were higher for the Lupron SQ group compared to the control group (fable 3). If Lupron had no treatment effect, then the difference between behavior rates of control and treatment groups will equal zero. The difference in behavior rates for does treated with Lupron SQ and control does was not equal to zero for male pre-copulatory behaviors and was borderline significant at a 0.10 alphalevel for female pre-copulatory behaviors (fable 4). Behavior rates were almost identical for the control and Lupron SQ groups for copulatory and general breeding behavior categories. However, male and female pre-copulatory behavior rates for the Lupron SQ group were 2.0-5.5 times higher than for the control group. It is unclear why the Lupron SQ group had higher pre-copulatory be~vior rates, but similar rates of copulatory behavior. It is possible that sampling variation is responsible, especially given the small sample sizes. The rate of male pre-copulatory behavior was significantly higher toward the Lupron SQ group than the control group. Male pre-copulatory behavior rates were 2 times higher toward the Lupron SQ group. Although the rate of female pre-copulatory behavior was 5.5 times higher for the Lupron SQ group compared to the control group, this difference was not significant. 6 --- Deer K95 - -0- - Deer G96 84 118 -......-- DeerN93 5 ~ E oi .s "<t a.. 4 3 E :l ai Cl) 2 0 .... ' ' '' ' ',"' '' " Pretrmt 47 153 174 210 Days Posttreatment Figure 1. Serum progesterone (P4) levels in female mule deer before and after treatment with a subcutaneous implant containing 10 mg ofLUPRON. Each line and symbol represents an individual deer. �229 As with the elk, rates of general breeding behavior were similar for the 2 treatment groups. Although deer do not have a harem breeding system, the general breeding behaviors may be more concerned with dominance and represent indirect reproductive behaviors. ---- Deer K95 -¾- --0-- DeerG96 Deer N93 30 4 ~ E CD .s 20 I ...I . E ::, cii en Q .:,,: t'G Cl) \ 10 (l_ \ \ \ ·\ \ \ \ \ \ \ \ 'I:\ 'I' 0 Pretrmt 47 84 118 153 174 210 Days Posttreatment Figure 2. Serum luteinizing hormone (LH) levels in female mule deer before and after treatment with a subcutaneous implant containing 10 mg ofLUPRON. Each line and symbol represents an individual deer. Table 3. Mean rate of behavior and standard error for control and treatment deer groups; mean and standard error were estimated using type ill least-squares. Behavior category Treatment Group Copulatory Control Lupron SQ Male Pre-Copulatory Control Lupron SQ Female Pre-Copulatory Control Lupron SQ General Breeding Control Lupron SQ Mean SE (# behviors/h r') 0.010 0.011 0.007 0.010 0.466 0.944 0.124 0.010 0.055 0.004 0.028 0.107 0.029 0.021 0.111 0.101 �230 Table 4. Difference in deer behavior rates(# behaviors/hr) between control and treatment groups, standard error of the difference, and P-value of the difference (the probability that the behavior rates were equal); statistics were estimated using type III least-squares. Behavior Category Treatment Group Copulatory Control - Lupron SQ Male Pre-Copulatory Control - Lupron SQ Female Pre-Copulatory Control - Lupron SQ General Breeding Control - Lupron SQ Mean Difference SE P-Value -0.001 0.013 0.944 -0.478 0.162 0.003 -0.045 0.030 0.129 -0.004 0.035 0.900 LITERATURE CITED Adam, C. L., C. E. Moir, and T. Atkinson. 1985. Plasma concentrations of progesterone in female red deer (Cervus elaphus) during the breeding season, pregnancy, and anoestrus. J. Reprod. Fertil. 74:631-636. Anderson, A. E., and D. E. Medin. 1966. The breeding season in migratory mule deer. Colorado Division of Wildlife, Game Information &aflet No. 60. Armstrong, RA. 1950. Fetal development of the northern white-tailed deer (Odocoileus virgin/anus borealis, Miller). American Midland Naturalist 43:650-666. Asch, RH., F. J. Rojas, T. R Tice, and A. V. Schally. 1985. Studies of a controlled- release microcapsule formulation of an LH-RH agonist in the rhesus monkey menstrual cycle. Int. J. Fert. 30:19-26. Asher, G. W. 1985. Oestrous cycle and breeding season of farmed fallow deer, Dama dama. J. Reprod. Fertil. 75:521-529. Baker, D. L., M. W. Miller, and T. M. Nett. 1995. Gonadotropin-releasing hormone analog-induced patterns ofluteinizing hormone secretion in female wapiti (Cervus elaphus nelsoni) during the breeding season, anestrus, and pregnancy. Biol. ofReprod. 52:1193 - 1197. Bartmann, RM. 1986. Growth rates of mule deer fetuses under different winter conditions. Journal of Wildlife ~ement 46:245-248. Bartmann, RM., G. C. White, and L. C. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monographs No. 121. 39pp. Becker, S. E., Katz, L. S. 1995. Effects of gonadotropin - releasing hormone agonist on serum LH concentrations in female white-tailed deer. Small Ruminant Research 18:145-150. Casper, RF., and S.S. C. Yen. 1979. Induction ofluteolysis in the human with a long-acting analog of luteinizing hormone-releasing fuctor. Science 205:408-410. Clayton, R. N. 1982. GnRH modulation of its own pituitary receptors: evidence of biphasic regulation. Endocrinology 111:152-161. Cole, G. F. 1971. An ecological rationale for the natural or artificial regulation of native ungulates in parks. Transactions of the North American Wildlife and Natural Resources Conference 36:417426. Concannon, P. W., and V. N. Meyers-Wallen. 1991. Current and proposed methods for contraception and termination of pregnancy in dogs and cats. J. Am. Vet. Med. Assoc. 198:1214-1225. �231 Curlewis, J. D., A. S. I. Loudon, and A. P. M. Coleman. 1988. Oestrous cycles and the breeding season of the Pere David's deer hind (Elaphurus davidianus). J. Reprod. Fert. 82: 119-126. Dasmann, R. F., and R. D. Taber. 1956. Determining structure in Columbian black-tailed deer populations. Journal of Wildlife Management 20:78-83. Decker, D. J., and Gavin. 1987. Public attitudes toward a suburban deer herd. Wildlife Society Bulletin 16:53-75. Diamond, J. 1992. Must we shoot deer to save nature? Natural History August: 2-8. Fraser, H. M. 1983. Effect of treatment for one year with a luteinizing hormone-releasing hormone agonist on ovarian, thyroidal, and adrenal function and menstruation in the stumpta.iled monkey (Macaca arctoides). Endocrinology 112:245-253. __, M., J. Sandow, H. Seidel, and W. von Rechenberg. 1987. An implant of a gonadotropin releasing hormone agonist (burserelin) which suppresses ovarian function in the macaque for 3-5 months. Acta Endocrinologica 115: 521-427. Garrot, R. A. 1995. Effective management of free-ranging ungulate populations using contraception. Wildlife Society Bulletin. 23:445-452. Guinness, F. E., Lincoln, G. A., and Short, R. V. 1971. The reproductive cycle of the female red deer, Cervus elaphus L. J. Reprod. FertiL 27: 427-438. Hess, K. 1993. Rocky times in Rocky Mountain National Park. University. Press of Colorado, Niwot, Colorado. Herschler, R. C., and B. H. Vickery. 1981. The effects of LHRH ethylamide on the estrous cycle, weight gain, and feed efficiency in feedlot heifer. Amer. J. of Vet. Res. 42: 1405- 1408. Hobbs, N. T., D. L. Baker, and R. B. Gill. 1999. A general theory describing effects of fertility control on populations of ungulates. Journal ofWildlife Management (in press). Hone, J. 1992. Rate of increase and fertility control. J. Applied Ecology 29:695-698. Hudson, P., and L. G. Ludvig. 1959. Embryonic and fetal development of mule deer. Journal of Wildlife Management 23:295-304. Houston, D. B. 1971. Ecosystems of national parks. Science 172:648-651. __, 1982. The Northern Yellowstone Elk: Ecology and Management. MacMillan Publishing Company,. New York, New York. Jewell, P.A., and S. Holt. 1981. Problems in management oflocallyabundant wild animals. Academic Press. 32lpp. Jopson, N. B., M. W. Fisher, and J.M. Suttie. 1990. Plasma progesterone concentrations in cycling and ovariectomized red deer hinds: the effect of progesterone supplementation and adrenal stimulation. Animal Reprod. Sci. 23 :61-73. Kelly, R. W., K. P. McNatty, G. H. Moore, D. Ross, and M. Gibb. 1982. Plasma concentrations ofLH, prolactin, oestradiol and progesterone in female red deer during pregnancy. Journal of Reproduction and Fertility 64:475-483. Kirkpatrick, J. F., and J. W. Turner, Jr. 1985. Chemical fertility control 31!d wildlife management. Bioscience 35:485-491. Lehner, P. N. 1987. Design and execution of animal behavior research: an overview. Journal of Animal Science 65: 1213-1219. McCullough, D.R., K. W. Jennings, N. B. Gates, B. G. Elliot, and J.E. Didonato. 1997. Overabundant deer populations in California. Wildlife Society Bulletin 25:478-483. McNeilly, A. S., and H. M. Fraser. 1987. Effect of gonadotropin-releasing hormone agonist-induced suppression of LH and FSH on follicle growth and corpus luteum function in the ewe. J. Endocrinology 115:273-282. Medin, D. E., and A. E. Anderson. 1979. Modeling the dynamics ofa Colorado mule deer population. Wildlife Monographs No. 68. 77pp. Miller, M. W., M. A. Wild, and E. S. Williams. 1998. Epidemiology of chronic wasting disease in captive rocky mountain elk. Journal of Wildlife Diseases 34:532-538. �232 Miller, R.G. 1966. Simultaneous statistical inference. New York: McGraw-Hill Book Co. 152-168. Morrison, J. A. 1960. Characteristics of estrus in captive elk. Behavior 16:84-92. Morrison, D. F. 1976. Multivariate statistical methods. New York: McGraw-Hill Book Co. 145-194. Niswender, G.D., L. E. Reichert, Jr., A. R. Midgley, Jr., and A. V. Nalbandov. 1969. Radioimmunoassay for bovine and ovine luteinizing hormone. Endocrinology 84: 1166-1173. _ _ . 1973. Influence of the site of conjugation on the specificity of antibodies to progesterone. Steroids 22:413-424. O'Bryan, M. K., and D.R. McCullough. 1985. Survival of black-tailed deer following relocation in California. Journal ofWildlife Management 49:115-119. Robinette, W. L., and J. S. Gashwiler. 1955. Fertility of mule deer in Utah. Journal ofWildlife Management 19:115-136. Sandow, J. 1982. Inhibition of pituitary and testicular function by LHRH analogs. Pages 19-39 In: Jeffcoate S. L. and Sandlier (eds). Progress towards a male contraceptive. Wiley and Sons, Chichester. Singer, F. J., L. C. Zeigenfuss, R. G. Cates, and D. T. Barnett. 1998. Elk, multiple factors, and persistence of willows in national parks. Wildlife Society Bulletin 26:419-428. Spraker, T. R., M. W. Miller, E. S. Williams, D. M. Getzy, W. J. Adrian, G. G. Schoonveld, R. A. Spowart, K. I. O'Rourke, J.M. Miller, and P.A. Merz. 1997. Spongiform encephalopathy in free-ranging mule deer (Odocoileus hemionus), white-tailed deer (Odocoi/eus virginianus), and Rocky Mountain elk (Cervus elaphus nelsoni) in northcentral Colorado. Journal of Wildlife Diseases 33:1- 6. Vale, W., C. River, M. Brown, J. Leppaluoto, N. Ling, M. Monaham, and J. River. 1976. Pharmacology of hypothalamic peptides. Clio. Endocrin. 5(suppl): 261-273. Wagner, F. H., R. Foresta, R. B. Gill, D.R. McMullough, M. P. Pelton, W: F. Porter, and H. Salwasser. 1995. Wildlife policies in the U. S. National Parks. Island Press. Washington, D. C. 242pp. Warren, R. J., L. M. White, W.R. Lance. 1995. Management of urban deer populations with contraceptives: practicality and agency concerns. Wildlife Society Bulletin 23:441-444. White, C. A., C. E. Olmsted, and C. E. Kay. 1998. Aspen, elk, and fire in the Rocky Mountain national parks of North America. Wildlife Society Bulletin 26:449-462. Whittaker, D. G. 1995. Patterns of coexistence for sympatric mule and white-tailed deer on Rocky Mountain Arsenal, Colorado. Ph.D. Dissertation, University of Willard, S. T., R. G. Sasser, J.C. Gillespie, J. T. Jaques, T. H. Welsh, Jr., and R. D. Randel. 1994. • Methods for pregnancy determination and the effects of body condition on pregnancy status in Rocky Mountain elk. Theriogenology 42:1095-1102. Wright, R. G. 1988. Wildlife issues in national parks. Pp. 168-196 in W. J. Chandler (ed) Audubon Wildlife Report, National Audubon Society. _ _ 1992. Wildlife management and research in the national parks. University of Illinois Press. 224pp. Prepared by _ _ _ _ __ DanL. Baker Research Biologist � https://cpw.cvlcollections.org/files/original/9bdc01ae5ab39a2280ec012171d96662.pdf 113ce9dd666345a864ca310d412f6187 PDF Text Text 135 Colorado Division of Wildlife Wildlife Research Report July 1998 JOB PROGRESS REPORT State of Colorado cost center 3430 Project No. W-153-R-ll Mammals Program 3001 Deer conservation Work Package No. Task N o . - - - - - - - - ~ - - - - - Period Covered: Author: Experimental Deer Inventory July 1, 1997 - June 30, 1998. R. M. Bartmann, T. M.. Pojar. Personnel: R. Arant, L. Bennett, G. Bock, D. c. Bowden, D. coven, M. Leslie, D. Masden, K. Miller, J. Olterman, M. Potter, G. Schoonveld, H. Spear, s. Steinert, J. Thomson, B. Watkins, G. c. White, and D. Younkin. Abstract An inventory system was developed to enable more precise monitoring of mule deer (Odocoileus hemionus) population status. It was implemented in 2 Data Analysis Units (DAU) in late 1997. A spreadsheet model was used to identify key parameters for model inputs. Allocation of sampling effort among the surveys required to obtain parameter estimates was based on annual variability associated with those estimates and sensitivity of the model to change in those estimates. Key parameters are· annual doe survival~ overwinter fawn survival, fawn:doe ratio and proportion of does in the population obtained from age/sex surveys, population size, and harvest. In D4, the fawn:doe ratio estimate was 58.8:100. None of 30 radiocollared does died during the segment and 10 of 38 fawns died for a 74% survival. In D19, the fawn:doe ratio estimate was 34.2:100. Five of 31 does died for an 84% survival and 20 of 39 fawns died for a 49% survival. A quadrat census to estimate population size could not be done in either DAU. �137 EXPERIMENTAL DEER INVENTORY Richard M. Bartmann and Thomas M. Pojar P. N. OBJECTIVES 1. Develop and test an experimental deer inventory system in 5 DAU's in western Colorado to determine efficacy in monitoring deer population status. 2. Publish results in a peer-reviewed scientific journal. SEGMENT OBJECTIVES 1. Estimate the annual doe survival rate in DAU's D-4 (Red Feather) and D-19 (Uncompahgre). 2. Estimate the over-winter fawn survival rate in DAU's D-4 (Red Feather) and D-19 (Uncompahgre). 3. Estimate the December fawn:doe and buck:doe ratios in DAU's D-4 (Red Feather) and D-19 (Uncompahgre). 4. Estimate winter deer density in DAU's D-4 (Red Feather) and D-19 (Uncompahgre). 5. Develop a spreadsheet model for predicting deer population performance in DAU's D-4 (Red Feather) and D-19 (Uncompahgre). 6. Analyze data and prepare an annual Federal Aid Job Progress report. INTRODUCTION During the early 1990's, mule deer populations in much of Colorado and the west began declining. The Colorado Division of Wildlife (CDOW) has collected data on deer populations for many years. These data consist primarily of harvest estimates collected state-wide every year, age/sex classifications made in most Game Management Units (GMU) every 1 or more years, and population estimates made in a few GMU's or DAU's infrequently. These data were assembled to try and determine if and when a deer decline occurred and its magnitude (Colo. Div. Wildl., unpubl. rep.). It was concluded the occurrence of a large-scale deer decline might be inferred but evidence was weak. This was partly because the data were inconsistent in the timing and manner in which they were collected, and also because additional key data on survival rates were not available. Thus, a new deer monitoring system was needed that would enable assessing deer population performance in a more timely and precise manner. STUDY AREAS It was originally planned to establish a new deer inventory system in 5 DAU's around the state. Later, it was decided that for evaluation purposes, 3 would suffice. Because of financial constraints, the system was implemented in 2 �138 DAU's the first year with the third DAU to be included the second year. The 2 DAU's selected for initial implementation of a new deer inventory system were D-4 ( Red Feather) in the northern front range and D-19 (Uncompahgre) in southwest Colorado. DAU D-4 consists of GMU's 7, 8, 9, 19, and 191. Winter range is a mountain shrub type with Ponderosa pine (Pinus ponderosa)- Douglas fir (Pseudotsuga menziesii) overstory in many places. DAU D-19 consists of GMU's 61 and 62. Winter range is primarily a pinyon (Pinus edulis)-juniper (Juniperus osteosperma)-sagebrush (Artemisia tridentata) type with considerable oakbrush (Quercus gambelli) interspersed. METHODS survival Basic methods for both DAU's were the same, although timing of their implementation varied. Annual doe survival and over-winter fawn survival were estimated from a sample of radio-collared animals. Seventy radiocollars were available the first winter and allocated to 30 does and 40 fawns. Deer were captured in early winter by netgunning from a helicopter. Annual doe survival was estimated for the period 1 December to 30 November. over-winter fawn survival estimated from 1 December to 15 June when they were considered adults. Capture locations were originally randomized throughout the winter range in each DAU, but problems developed with private land access and finding deer. Consequently, the helicopter crew was sent to accessible areas where deer were known present and instructed to put out a certain number of doe and fawn radiocollars. Each radiocollar had a mortality mode set for a 3-4-hour delay with frequencies in the 148-151 MHz range. Fawn collars were a drop-off type so they could be retrieved and reused. This precluded getting survival information on yearling females from 15 June to 30 November. Data from Piceance Basin indicates this is not a critical period from a survival standpoint, except during the hunting season, and hunting mortalities are censored anyway. Radiocollared deer were monitored from the air a minimum of once every 2 weeks from December through June. Each mortality was checked as soon as possible to try and determine cause of death. A~e/sex classifications Age/sex classifications were used to estimate fawn recruitment to December and post-season buck:doe ratios. A random-route survey was already in place in D4 and was used without modification for this study. Fifteen existing census quadrats were randomly selected and a route passing through each quadrat devised. The route was flown with a Bell Soloy helicopter using a GPS unit for navigating from point to point. In D-19, a random-route survey was designed similar to that in D-4. Each GMU was divided from north to south into strata at 1,000-meter intervals along UTM coordinates. Two existing census quadrats were randomly selected in each strata within a GMU and, starting at the top of the GMU, a line was drawn to connect quadrats in a zig-zag pattern from north to south. The process was then repeated for the other GMU. �139 Deer in groups encountered along the flight path were classed as fawns, does, bucks, or unknown. Estimates of fawn:doe and buck:doe ratios were based on groups using the estimator developed by Bowden et al. (1984). Population Size Deer population size was estimated from counts of deer on 0.65-km2 quadrats using procedures described by Kufeld et al. (1980). A quadrat census system was established in D-4 in 1985 and has been flown 7 times, the last in January 1997. The current design includes 98 quadrats in 15 strata, but the original data on the total sample area and strata poundaries could not be found. There was not enough time to gather the needed data to define a new total sample area and re-stratify, so the quadrat census was to be flown as in the past. The exception was that a Bell Soley helicopter was to be used instead of a Jet Ranger. Snow conditions along the front range are undependable. When adequate snow does occur to provide a good counting background, it usually lasts only a few days. Helicopters need to be scheduled in advance, so the chance of encountering good counting conditions was minuscule. Therefore, it was decided the D-4 survey would be flown regardless of snow cover. A quadrat census was established in GMU 62 in 1976. It was expanded to include GMU 61 in 1981. There are 346 quadrats in 13 strata which took -70 hours to survey. Only 30 hours were allocated for the census, so the number of quadrats was reduced to 140 by random selection. Population Model The POPII model is currently used to help manage deer populations. The model is complex and requires numerous parameter estimates with no supporting data. As a result, people are free to manipulate various parameters to try and make the model fit their preconceived ideas. A much simpler model is needed for which key inputs are derived from sample-based estimates. The new model will help determine the key types of data required and how to allocate costs and effort to obtain them. RESULTS Doe survival P.=.i: Thirty does were captured and radiocollared from 4-7 January 1998. The delay from the desired late November period was caused by contract problems. Monitoring was done from a fixed-wing aircraft about once every 2 weeks through June 1998. No does died during that period (Table 1). 12=.l.2.: Does were captured and radiocollared from 15-20 December 1997. Here again contract problems delayed the process. The first two does captured were given fawn radiocollars by mistake. Consequently, 31 does were finally radiocollared. Monitoring from a fixed-wing aircraft was done weekly the first month, but was increased to twice weekly the rest of the winter and spring to try and locate mortalities quicker. As of 30 June, 5 of the 31 does had died for a survival estimate of 84%. Annual doe survival is monitored until 30 November, so this rate could go lower. �140 Table 1. Fates of mule deer does and fawns radiocollared in DAU's D-4 (Red Feather) and D-19 {Uncompahgre) during the 1997-98 winter. D-4 Fate D-19 Does Fawns Does Fawns 30 28 26 19 Coyote 3 3 10 Lion 1 1 1 Survived Bobcat Unknown Predator 1 2 1 Starvation 1 1 Accident 1 1 Undetermined 3 5 Fawn survival ~ : Thirty-eight fawns were captured and radiocollared during the same period as for does. Two collars were temporarily misplaced and not put out. Ten fawns died for an over-winter survival rate of 74%. ~ : A doe radiocollar was erroneously placed on a fawn. This, together with the 2 fawn collars placed on does, resulted in only 39 fawns being radiocollared. Twenty fawns died by 15 June for an over-winter survival rate of 49%. The fawn that received a doe collar was 1 of the mortalities which eliminated the need to recapture it to prevent choking when it got older. Age/Sex Classifications D.=_4: The classification survey was flown 16-18 December. There were 950 deer classified for a fawn:doe ratio of 58.8 (95% CI= 52.9-65.2) and buck:doe ratio of 23.8 {95% CI= 19.4-28.6). ~ : The classification survey was flown 11-13 December. The random routes only took about 5 hours to complete with 358 deer classified. To use more of the 15 hours allocated for the survey, a non-random route was flown through the longer north-south length of each GMU. This produced another 831 deer in abo_ut •the same amount of time as for the random routes. For all data combined, the fawn:doe ratio estimate was 34.2 {95% CI= 30.937.7) and the buck:doe ratio estimate was 11.1 (95~ CI= 8.9-13.4). Comparing results from 2 route types, the fawn:doe ratio estimate was marginally higher (35.3 vs. 31.8) while the buck:doe ratio was nearly twice as high (12.8 vs. 7.0) on the non-random routes. Thus, the observer's knowledge about where to find large numbers of deer could bias ratio estimates, particularly for bucks. Population size D.=_4: contract problems prevented getting a helicopter before the mid-January cutoff date for conducting the survey, so it was not flown. �141 D.=.J..2.: Adequate snow conditions did not occur until late January. A population estimate has not been made since 1994, so it was decided to do the survey even though it was past the 15 January cutoff. Work was started 27 January, but weather and other problems prevented the survey from being completed in a reasonable time span, so it was canceled on 12 February after 63 quadrats had been flown. A valid population estimate cannot be made based on this incomplete data. Population Model Development of a population model to identify key inputs for refining the deer inventory system was begun with cooperators from Colorado State University. Deer population dynamics are much more complicated than the model portrays, but routine measurement of the wide array of inputs required for a more complicated model is unrealistic. Thus, it is a reasonable trade-off between what we can measure from a practical standpoint and what is needed to predict deer population status for management purposes. The model concentrates on predicting changes in the adult doe portion of the population as this is the key to evaluating population status. The model has only 2 age classes (fawns and adults) assigned to 3 categories (fawns, does, and bucks). Fawns are initially recruited to the population on 1 December when the fawn:doe ratio is estimated from age/sex classifications. The model contains 4 parameters that are year-specific: annual buck survival, annual doe survival, over-winter fawn survival, and December fawn recruitment. Buck survival has no influence on population status unless their numbers drop low enough to affect breeding; a phenomenon yet to be documented in hunted populations. Therefore, there is no real need for an estimate of buck survival. An indication of status can be derived from the buck:doe ratio estimate obtained from age/sex surveys, and the number of new bucks recruited is estimated from male fawns that survive to become yearlings. Thus, only the last 3 parameters need to be considered for the model. The strategy for managing a mule deer population is to estimate the doe segment so harvest can be adjusted to maintain the December doe population at some desired level. To do this, 3 more parameters have to be estimated in addition to the 3 mentioned above: total population size, the proportion of the population that is does, and doe harvest. A harvest estimate is already available from another source and fawn recruitment and the proportion of does in the population are obtained from the same survey, so we only need to consider 4 surveys to obtain 5 parameter estimates. The problem now comes down to how to allocate costs and effort to these 4 surveys to derive estimates of the 5 parameters that will result in the most precise prediction of the December doe population or, alternatively, the annual rate of population increase. Either will indicate the number of does that must be harvested to maintain the desired doe population level. This process is ongoing and will be completed in the next segment. �142 LITERATURE CITED Bowden, D. c., A. E. Anderson, and D. E. Medin. 1984. Sampling plans for mule deer sex and age ratios. Journal of Wildlife Management. 48:500-509. Kufeld, R. c., J. H. Olterman, and D. c. Bowden. 1980. A helicopter quadrat census for mule deer on Uncompahgre Plateau, Colorado. Journal of Wildlife Management. 44:632-639. /? ~repared by ~ ~~\;£:::;______ WildLi.fe Researche~ ~%~ Wildlife Researcher ��85 Colorado Division of Wildlife Wildlife Research Report July 1999 JOB FINAL REPORT State of: Colorado Project No. W-153-R-12 Work Package No._3~0~0~1~-----Task No. I Cost Center 3430 Mammals Program Deer Conservation Experimental Deer Inventory Period Covered: July 1, 1998 - June 30, 1999 Authors: R B. Gill, and T. M. Pojar Personnel: G. Schoonveld, S. Steinert, J. Olterman, C. Wagner, B. Watkins, R Bartmann, G. White ABSTRACT Over-winter (Nov-May) survival rates were estimated for 3 Data Analysis Units: Red Feather (D-4), Middle Park (D-9), and Uncompahgre (D-19). Overall 3 study areas, doe survival averaged 91.7% and fawn survival averaged 79.4%. Population estimates were obtained for D-9 in late January, 1999 by flying square mile quadrats allocated according to a stratified random sampling scheme. Deer density averaged 12.93 deer/km2 and projected to a total population of 11,016 deer± 2,337 (90% CI). Estimates of the ratios of bucks and fawns:100 does were obtained from helicopter counts conducted in December 1998. The buck:doe ratio was estimated to be 27 bucks: I 00 does. The fawn:doe ratio was estimated to be 62 fawns: I 00 does. �87 EXPERIMENTAL DEER INVENTORY R. Bruce Gill and Thomas M. Pojar P.N. OBJECTIVES I. Develop and test an experimental deer inventory system in 5 DAUs in western Colorado to determine efficacy in monitoring deer populations. 2. Publish results in a peer-reviewed scientific journal. SEGMENT NARRATIVES I. Estimate the annual doe survival rate in DAUs D-4 (Red Feather), D-9 (Middle Park), and D-19 (Uncompahgre). 2. Estimate over-winter fawn survival rate in DAUs d-4 (Red Feather), D-9 (Middle Park), and D19 (Uncompahgre). 3. Estimate the December fawn:doe and buck:doe ratios in DAU D-9 (Middle Park). 4. Estimate winter density in DAU D-9 (Middle Park). 5. Develop a spreadsheet model for predicting deer population performance in DAUs D-4 (Red Feather), D-9 (Middle Park), and D-19 (Uncompahgre). 6. Analyze data and prepare an annual Federal Aid Job Progress Report. INTRODUCTION Initial plans called for the development and testing of experimental inventory systems on 5 deer _Data Analysis Units (DAUs) across Colorado's important deer herds. Manpower and financial constraints limited the scope to 3 DAUs: Red Feather (D-4), Middle Park (D-9), and Uncompahgre (D-19). Measurements on each DAU were to continue for a period of 3-5 years before deciding whether to expand implementation of the experimental deer inventory system. However, the project's principal investigator, Richard M. Hartmann, retired during the segment and his vacancy remains unfilled. Consequently, it was decided to terminate the R&D component of the inventory system at the end of the current segment. This report, therefore, constitutes the Final Report of activities accomplished under Work Package 3001, Task I. STUDY AREAS Study aras D-4 and D-19 were described previously (Hartmann and Pojar 1998). Study area D-9 is situated in the headwaters area of the the Colorado River drainage encompassing areas below approximately 2,600 min elevation in Grand and Summit Counties. Winter range is primarily �88 sagebrush steppe vegetation type characterized by low growing(< 1 m) shrubs. Sagebrush species (Artemisia tridentata tridentata andArtemisia tridentata vaseyana) predominate with other shrub species such as rubber rabbitbrush (Chrysothamnus nauseosus), sticky-flowered rabbitbrush (Chrysothamnus viscidifloris), mountain snowberry (Symphoricarpus oreophilus), bitterbrush (Purshia tridentata), serviceberry {Amelanchier alnifo/ia), and chokecherry (Prunus virginiana)locally abundant (scientific names ·according to Weber 1976). Mule deer winter over approximately 850 km2 below 2600 m in elevation. METHODS Bartmann and Pojar (1998) thorougly described the methods used in this investigation, so they will not be repeated here. RESULTS Doe Survival Across all 3 study areas, doe survival during winter 1998-99 averaged 91. 7% with 121 of 132 does surving from November 1998 through May 31, 1999. This compares to an over-winter survival rate of 90.1 % for adult does throughout the previous winter (Bartmann 1998). No single source of mortality stood out as a primary {Table 1) which was similar to results from 1997-98. Fawn Survival In general, fawn survival throughout winter 1998-99 increased compared to 1997-98. In 199899, 127 of 160 radio-collared fawns survived from November 1, 1998 through May 31, 1999 for an overran survival rate of 79. 4 %. In 1997-98, 61. 0% of radio-collared fawns overall survived the winter. Sex and Age Composition Classification counts of mule deer in Middle Park Data Analysis Unit (D-9) were conducted during the period December 22-24, 1998 and included a total of912 deer, of which 15.9% were bucks, 32. 0% were fawns, and 52.1 % were does (Table 2). The ratio of bucks per 100 does computed to 30.5 ±4.8 bucks: 100 does and 61.5 ±7.5 fawns:100 does. Population Density Middle Park Data Analysis Unit (D-9) deer census quadrats were flown during the fourth week in January, 1998. Counting conditions and deer distribution were both ideal. Overall, 2,415 deer were tallied on 56 sample quadrats. Number of deer/km2 averaged 12.93 with a standard error of 4.00. When projected over the entire winter range ofD-9, the deer population was estimated at 11,016 deer ±2,337 (90% {Table 3). en �89 Table l. Fates of mule deer does and fawns radiocollared in DAUs D-4 (Red Feather), D-9 (Middle Park}, and D-19 (Uncomeahgre} during the 1998-99 winter. Red Feather Fate Uncompahgre Middle Park Does Fawns Does Fawns Does Fawns Total Radiocollared S2 so 40 so 40 60 Survived 4S 40 39 44 37 43 Coyote predation l 4 0 l l 7 2 0 0 0 2 1 0 3 0 2 2 0 0 0 0 0 0 0 0 0 0 2 Accidents 0 0 0 0 0 Unknown 2 0 0 2 4 10 1 6 J 17 Mtn. lion predation Other predators 0 Road kills 0 Hunter kills 2 Starvation 0 7 Total Mortality 1Total does not equal 100% because of rounding error. Table 2.- Results of mule deer classification counts in Middle Park Data Analysis Unit (D-9), December 22-24, 1998. GMU I-yr bucks 2-yr bucks Adult bucks Total bucks Does Fawns Total 12/23/98 27 6 6 ·3 15 66 44 12S 12/22-24/98 37 27 23 19 69 17S 98 342 12/23-24/98 181 6 5 2 13 44 17 74 12/23-24/98 28 7 12 3 22 130 92 244 12/24/98 18 11 10 5 26 60 41 127 -. - <ih Date Totals 32 145 47S 292 D-9 Swrunary 1-yr bucks: 100 does 2-yr bucks:100 does Adult bucks:100 does Total bucks:100 does Fawns:100 does Ratio estimates 12.0 11.8 6.7 30.S 61.S Lower90%CI 9.3 9.1 4.8 2S.9 54.3 Upper9O%CI 14.8 14.6 8.8 35.5 69.4 S7 56 �90 Table 3. Deer quadrat census results for Middle Park Data Analysis Unit (D-9), January 1998. Stratum Strata Deer Counted (km2) Population Estimate N n Total Dccr/km2 SE Total ±_90% 80.3 28.5 582 20.43 12.34 1640 629 2 222.7 10.4 13 1.25 3.17 280 449 3 189.1 46.6 . 730 15.66 7.64 2961 918 4 121.7 15.5 144 9.27 9.84 1128 761 5 72.5 15.5 378 24.33 26.41 1764 1216 6 88.1 15.5 494 31.79 24.59 2799 1375 7 77.7 12.9 74 5.71 5.40 444 266 Overall 852.1 145.0 2415 12.93 4.00 11016 2337 Spreadsheet Models A spreadsheet population model was developed in a Quatro Pro spreadsheet format for Red Feather (D-4), Middle Park (D-9), and Uncompahgre (D-19) Data Analysis Units. Disk copies of spreadsheets for each DAU were sent to Area Biologists responsible for managing each DAU. Copies of the spreadsheet fonnat may be obtained by writing to: Dr. Gary C. White 211 B JVK Wagar Bldg. Department of Fishery and Wildlife Biology Colorado State University Fort Collins, CO 80523 LITERATURE CITED Bartmann, R. M., and T. M. Pojar. 1998. Experimental deer inventory. Colorado Division of Wildlife. Wildlife Research Report, July, 1998, Part II: 135-142. Weber, W. A. I 976. Rocky mountain flora. Colorado Associated University Press. Boulder, CO. Prepared by _ _ _ _ _ _ _ _ _ __ R. Bruce Gill Wildlife Research Leader Thomas M. Pojar Wildlife Researcher � https://cpw.cvlcollections.org/files/original/9d51d60d346754558d312f84eb413df6.pdf d9368aa01bdf206ea3dd3a0f9b262bd5 PDF Text Text 145 DEER REPRODUCTION ASSESSMENT Richard M. Bartmann and Thomas M. Pojar P, N, OBJECTIVES 1. Estimate the variation in pregnancy-specific protein B (PSPB) levels in mule deer does in Game Management Unit (GMU) 19 (Poudre). 2. Develop a study plan to compare current mule deer reproductive rates to historic data to determine if differences exist that might contribute to the recent decline in deer populations in Colorado. SEGMENT OBJECTIVES 1. Estimate the variation in PSPB levels in mule deer does during 2 winter periods in GMU 19. 2. Test for differences in pregnancy rates of mule deer does between the 2 winter sample periods. 3. Develop a study plan to compare current and historic reproductive rates of mule deer in at least 2 areas of Colorado. 4. Analyze data and prepare an annual Federal Aid Job Progress report. INTRODUCTION An assessment of mule deer population status during the early to mid-1990's indicated that many populations across western Colorado had declined to various extents. However, the cause(s) was not identified. One indication of a problem was detection of a significant decline in December fawn:doe ratios in some DAU's over the past 20+ years. This suggested a possible problem with reproduction or neonatal mortality. A logical starting point to try and identify a problem was the breeding season--were does being bred? This was also the easiest aspect to investigate because it did not require killing deer to obtain samples. STUDY AREAS GMU 19 was selected for study because there was historic data for comparison. Winter range along the northern front range is a mountain shrub type with Ponderosa pine (Pinus ponderosa)-Douglas fir (Pseudocsuga menziesii) overstory. METHODS Pregnancy detection was via blood analysis to determine levels of PSPB. PSPB levels in blood have been used to detect pregnancy with a high degree of accuracy in various species including mule deer (Wood et al. 1986). Detection times post-conception have varied among species, but is generally considered reliable after about 24 days with cattle (Sasser et al. 1986). Therefore, delaying blood sample collection until early January was considered to provide an adequate interval post-breeding for detecting PSPB. Delaying the second set of collections until early February would detect pregnancies from �146 conceptions during the second estrus and provide an indication of the extent of late breeding. From 15-20 does (~1-year-old) were to be captured with Clover traps in early January and again in early February. However, helicopter netgunning of deer for survival work in DAU D-4, of which GMU 19 is part, was delayed until early January, so it was decided to collect blood samples from does captured for that project. Blood samples were submitted to BioTracking in Moscow, Idaho to assess levels of PSPB to determine pregnancy status. RESULTS Blood samples were obtained from 29 does captured 4-7 January 1998. Twentyseven were considered pregnant and 2 not pregnant. One doe was at the cut-off point of 93% (93.1%) and may have been a late breeder. The range in values for the pregnant does was 84.3-93.l with mean 88.5 and SE 0.36. Values for the 2 non-pregnant does were 102.6 and 102.8. The 27 pregnancies yielded a 93% pregnancy rate with all except possibly 1 apparently bred during the first estrus. Thus, late breeding did not appear a problem so no collections were done in February. The 93% pregnancy rate is barely significantly lower (P = 0.062) than the 100% rate derived from data for 49 does collected in GMU 19 from January-May 196165 (Anderson 1965a, 1965b, 1966). However, it is still considered a high rate with the conclusion that conception is apparently not a problem in DAU D-4. A next step would be to determine fetal rates, but this would involve euthanizing a large number of does during mid to late winter. One option is to have hunters collect does, but hunting seasons that late in winter are not well accepted. Therefore, no further work is planned to evaluate reproductive status until a satisfactory collection method is found for estimating fetal rates. LITERATURE CITED Anderson, A. E. 1965a. Reproductive studies. Colo. Dep. Game, Fish and Parks, Game Res. Rep. January:165-195. Anderson, A. E. 1965b. Reproductive studies. Colo. Dep. Game, Fish and Parks, Game Res. Rep. January:522-543. Anderson, A. E. 1966. Reproductive studies. Colo. Dep. Game, Fish and Parks, Game Res. Rep. January:275-307. Sasser, R. G., C. A. Ruder, K. A. Ivani, J.E. Butler, and w. C. Hamilton. 1986. Detection of pregnancy by radioimmunoassay of a novel pregnancyspecific protein in serum of cows and a profile of serum concentrations during gestation. Biol. Reprod. 35:936-942. Wood, A. K., R. E. Short, A. Darling, G. L. Dusek, R. G. Sasser, and C. A. Ruder. 1986. Serum assays for detecting pregnancy in mule and whitetailed deer. Journ:il of Wildlife Management. 50:684-687. ,,.71__ L - Prepared by .J(~~~,-_hJ:.~~d~;_:z.iiB~artm;;;zt.~~;n~n=··-=···=····==···=·:--:==::::::==-==WildLi.fe Researche~ Wildlife Researcher � https://cpw.cvlcollections.org/files/original/ebc33e7482d03efc32e7a2da5171b65d.pdf 3f9484b94f4f64460ae4c5468366fd41 PDF Text Text 147 Colorado Division Wildlife Research July 1998 of Wildlife Report JOB PROGRESS REPORT State of Project Work No. Package Colorado Cost Center W-153-R-ll Mammals No. ~3~0~0~1L_ Task No. Period _ 3430 Program Deer Conservation Monitoring and Managing Chronic Wasting Disease in Deer Covered: July 1, 1997 - June 30, 1998 Authors: M. W. Miller Personnel: S. Berry, K. Larsen, Wheeler, M. A. Wild, K. 1. O'Rourke, T. R. Spraker, and E. S. Williams S. Tracy, E. ABSTRACT Deer from throughout Colorado were examined for occurrence of chronic wasting disease using a combination of targeted surveys and harvest/road-kill surveys. We continued to develop and modify a statewide targeted surveillance program for acquiring, examining, and reporting on CWO suspects submitted from Colorado. Between June 1997 and May 1998, 6 chronic wasting disease (CWO)· cases were diagnosed among 16 "suspect" deer submitted from known endemic portions of northeastern Colorado; CWO was not diagnosed in any of 7 additional "suspect" deer submitted from elsewhere in Colorado. All confirmed CWO cases originated in game management units (GMUs) where the disease had been detected previously. Harvest and road-kill surveys were used to estimate CWO prevalence in enzootic management units. About 4.2% of deer harvested in Larimer County data analysis units (DAUs) (D4 or D10) and about 1.1% of deer harvested in South Platte River bottom units (DAU D44 plus GMUs 87, 90, 93, and 95) tested positive for CWO via immunostaining; none of the 171 deer harvested or culled in other select GMUs outside known enzootic areas tested positive for CWO. Prevalence estimates may be somewhat liberal because subclinical cases where either histopathological lesions or anti-PrP immunostaining reactions in brain tissue were included; of the 44 cases identified, ~none appeared to be suffering from clinical CWO and only 26 of 44 (59%) "positive" deer showed both histopathological lesions and immunostaining. In contrast to earlier observations derived from targeted surveillance data, CWO prevalence among male and female mule deer did not differ (P = 0.86). Age distribution of CWO-positive mule deer did not differ from sex-specific age distributions of negative animals. These data support the belief that CWO epidemiology is driven primarily by lateral transmission. �148 Requiring head submissions increased survey sample sizes for deer >4-fold compared to voluntary submissions in 1996. Based on harvest estimates from surveyed GMUs, compliance with mandatory submission regulations was about 75%. Unstaffed barrel sites still appear to be the most efficient method for sample collection~ Data from both targeted surveillance and surveys indicate that eastern Larimer County remains the most significant focus of CWO in Colorado, although some natural spread may be occurring both southward and eastward. Targeted ,surveillance of clinical suspects appears to be the most sensitive approach for initially detecting CWO in deer (and elk) populations throughout Colorado, and should be continued statewide. Once detected, combinations of harvest and road-kill surveys will be employed to estimate prevalence, monitor prevalence trends, and compare prevalence among DAUs. 'Because CWO appears to be more prevalent in deer than in elk in endemic areas, it follows that harvest surveys in high-risk areas should focus on deer rather than elk to be most effective in confirming absence of CWO in nonendemic areas. Cwo affected about 25% (5/25) of the adult (~1-yr-old) mule deer and about 46% (5/11) of the adult white-tailed deer in resident Foothills Wildlife Research Facility herds during the last biological year; CWO accounted for 83% of adult mule deer mortality and 100% of adult white-tailed deer mortality. Resident mule deer also represented the source of infection (treatment) in an ongoing 10-yr study of cattle susceptibility to CWO. No signs of neurological disease were observed in any of the 12 calves (subjects) or 12 mule deer fawns from the Rocky Mountain Arsenal National Wildlife Refuge (contact controls) housed with naturally~infected resident deer since July 1997; one calf died from gastrointestinal obstruction and bloat about 2 wk after entering the deer paddocks, and one fawn that apparently failed to adjust to captivity died about 1 mo after capture, presumably from hypothermia secondary to malnutrition. �149 MONITORING AND MANAGING CHRONIC WASTING DISEASE IN DEER M. W. Miller P. N. OBJECTIVES (1) Design, conduct, and report results of: (a) targeted surveillance to estimate and detect changes in distribution of chronic wasting disease (CWO) in free-ranging deer populations; and (b) harvest or road-kill surveys to estimate and detect prevalence of CWO in enzootic deer populations. (2) Design, captive changes in conduct, and report results of experimental studies using deer naturally or experimentally infected with cwo. SEGMENT OBJECTIVES (1) Conduct and report results of targeted surveillance to estimate and detect changes in distribution of CWO in free-ranging deer populations statewide. (2) Conduct and report results of harvest surveys to estimate prevalence of CWO in DAUs D4, D10( +GMU 29), and D44 (+ GMUs 87, 93, and 95). (3) Continue experimental evaluation natural contact exposure. of cattle (4) Observe deer. of naturally-occurring epizootiological features susceptibility to cwo via CWO in captive INTRODUCTION Chronic wasting disease (CWO) affects native deer and elk, causing behavioral changes and progressive loss of body condition that invariably lead to the death of affected animals (Williams and Young 1992). Neither the causative agent nor its mode of transmission have been identified. There are no tests currently available for diagnosing CWo in live animals, and postmortem tests require microscopic examination of brain tissue. There are no known treatments for cwo. Previous attempts to eradicate CWO from research facilities failed on at least 2 occasions (Williams and Young 1992; Miller et al., 1998). Although similar in some respects to other transmissible spongiform encephalopathies that affect domestic sheep (scrapie) and cattle (bovine spongiform encephalopathy; "mad cow disease"), there is no evidence suggesting cwo can be naturally transmitted to domestic livestock, or that scrapie or SSE can be transmitted to native cervids. Moreover, there is no evidence suggesting that cwo presents a threat to human health. "Chronic wasting disease" was first recognized by biologists in the 1960's as a disease syndrome of captive deer held in wildlife research facilities in Ft. Collins, CO, and was subsequently recognized in captive deer, and later in captive elk, in wildlife research facilities near Ft. Collins, Kremmling, and Meeker, CO and Wheatland, WY (Williams and Young 1980, 1982). Since 1981, CWO has also been diagnosed in free-ranging mule deer, white-tailed deer, and elk from northcentral Colorado; most of these diagnoses have been made since 1990 �150 (Spraker et al. 1997). Although CWO was first diagnosed in captive cervids, the original source of CWO is unknown; whether CWO in captive cervids really preceded CWO in wild cervids, or vice versa, is equally uncertain (Spraker et al. 1997). At present, the known world-wide distribution of CWO in wild cervids appears to be limited to northeastern Colorado and southeastern Wyoming. In Colorado, free-ranging CWO cases have primarily originated from along the Front Range near Estes Park and west of Ft. Collins-Loveland. Cases were diagnosed in 6 different game management units (GMUs)(191, 9, 19, 20, 94, and 96) prior to 1996, but two GMUs (19, 20) yielded about 85% of the documented cases; the affected GMUs comprise portions of 3 deer (04, 010, 044) and 2 elk (E4, E9) data analysis units (DAUs). There is no evidence that wild deer or elk outside northeastern Colorado are infected with CWO. The significance of CWO and its impacts on native deer and elk populations are unclear. Simulation models of CWO dynamics predict that this disease could cause significant declines in affected deer and elk populations (Miller and McCarty, unpublished data). In light of CWO's potential impacts on wildlife resources and the difficulties inherent in eliminating CWO from captive or wild cervid populations once established, it seems most prudent to assume CWO could adversely affect native deer and elk populations and manage to reduce its occurrence and prevent its further spread. A more complete understanding of CWO is fundamental to developing a comprehensive management program. In 1996, ongoing surveillance efforts were enhanced to provide a better tool for estimating and monitoring changes in CWO prevalence in enzootic areas and for potentially detecting emergence of CWO in new areas. Reliable estimates of CWO prevalence are particularly critical to detecting"trends, predicting potential impacts of disease on long-term population performance, and assessing efficacy of management interventions; moreover, such data are needed to guide policy decisions and to provide information to hunters and other publics. Ultimately, surveillance data will "be the foundation of an adaptive resource management plan for CWO in deer and elk; that plan will provide a mechanism for incorporating new knowledge gained through surveys, modeling, and experimental studies into a continuously evolving management program formulated to reduce the occurrence of CWO and minimize the risk of its spread to other native deer and elk populations in Colorado. MATERIALS AND METHODS Surveillance We monitored deer populations throughout Colorado for occurrence of CWO using a combination of targeted surveillance and harvest or road-kill surveys. These were organized and conducted as follows: Targeted (= clinical disease) surveillance: Deer showing clinical signs consistent with those seen in chronic wasting disease were collected by field personnel statewide and brain tissues examined for evidence of spongiform encephalopathy. The "suspect case" profile was defined as follows: • • Species: Age: mule deer white-tailed ~ 18 months deer �151 • Signs: emaciated and abnormal behavior &/or indifference to human activity &/or increased salivation &/or tremor, stumbling, incoordination &/or difficulty or inefficiency in chewing/swallowing increased drinking and urination &/or Where possible, submissions were subjected to complete necropsy; in some situations, only heads were available for examination and sampling. In all cases, histopathology of brain tissue (Williams and Young 1993) was used to diagnose CWO; in some cases, immunohistochemistry or other ancillary tests were used to confirm or support diagnoses. Harvest surveys: In order to obtain reliable estimates of CWO prevalence that will serve as a basis for monitoring responses to management interventions, we continued conducting harvest surveys on select deer populations. During the 1997-1998 hunting seasons, fresh brain and select lymphatic tissues were collected from deer harvested in enzootic GMUs; deer harvested or culled in other select GMUs throughout Colorado were also sampled as negative controls. Brain tissues were examined at the Colorado State University Diagnostic Laboratory for histopathological lesions (Williams and Young, 1993)or anti-PrP immunostaining reactions (O'Rourke et al., 1998) consistent with CWO infection. Because sample sizes for most individual GMUs were too small to provide reliable prevalence estimates by GMU, we pooled data by DAU for comparisons within and among species. We estimated specificity of imunohistochemistry using data from deer harvested outside known enzootic areas. Ages were estimated via replacement (4-6-mo-old, 16-18-mo-old, ~28-moold) and cementum annuli using first incisors from all positive deer and a random sample of negative deer (103 males and 100 females) harvested in four GMUs (9, 191, 19, 20); using these data, we compared sex-specific age frequency distributions for CWO-affected and unaffected male and female mule deer. Epizootiological Studies Epizootology of naturally-occurring CWO in captive mule deer and white-tailed ~ (Miller and Wild): Naturally-occurring CWO was a sporadic disease of resident FWRF mule deer prior to 1985 (Williams and Young, 1992), and also has occurred sporadically since 1994; no cases had been observed in resident white-tailed deer since their addition to FWRF in 1993. We maintained and observed 25 ~1-yr-old mule deer and 11 ~l-yr-old white-tailed deer in paddocks at CDOW's Foothills Wildlife Research Facility (FWRF) during July 1997-June 1998; 10 fawns were also born and recruited into the resident mule deer herd. Deer received natural forage, pelleted rations (high energy supplement and "browser" diet), and alfalfa hay; water and mineralized salt were available ad libitum. All deer were evaluated daily for clinical signs of CWO (and other health problems) in conjunction with routine feeding and handling activities. Resident mule deer also represented the source of infection (= "treatment") in an ongoing study of cattle susceptibility to CWO (see below). Cattle susceptibility to CWO (Williams and Miller): We continued monitoring cattle for clinical evidence of natural CWO transmission as part of a coordinated interagency effort to study cattle susceptibility to CWO. Twelve 4-mo-old calves purchased from a private ranch located near Sheridan, WY, were placed in paddocks at the FWRF with naturally-infected captive mule deer in July 1997 (see above for description of resident deer herd). Twelve 4-mo-old mule deer fawns were captured at the Rocky Mountain Arsenal National Wildlife �152 Refuge (RMANWR) in late September and placed in the same paddocks as contact controls for natural transmission (extensive ongoing surveillance of RMANWR deer has continued to confirm that this population is free of CWO). In a separate but related study, 20 additional 6-mo-old mule deer fawns were captured at the RMANWR in early December. Each fawn was given a single oral dose of about 5 g of fresh brain tissue homogenate from captive mule deer with clinical CWO. Fawns were then released into a separate paddock physically removed from the contact transmission study. These experimentally-infected fawns will serve two purposes: as controls for domestic calves orally inoculated with a single oral dose of about 50 g of this same CWO brain homogenate (Williams et al., 1998), and as subjects of a study on the pathogenesis of CWO in mule deer (Appendix A). RESULTS AND DISCUSSION Surveillance Targeted (= clinical disease) surveillance: Between June 1997 and May 1998, 6 chronic wasting disease (CWO) cases were diagnosed among 16 "suspect" deer submitted from known endemic portions of northeastern Colorado; CWO was not diagnosed in any of 7 additional "suspect" deer submitted from elsewhere in Colorado. All confirmed CWO cases originated in game management units (GMUs) where the disease had been detected previously. Males and females were represented equally. All CWO cases were observed and submitted during November-February, well within the October-April timeframe of most clinical case submissions (Spraker et al., 1997; Miller, 1997). Encephalitis, intoxication, neoplasia, and pneumonia were diagnosed among the 17 suspects not suffering from CWO; for most (9/17) non-CWO cases, no definitive explanation for clinical signs and condition could be determined from the samples submitted. Harvest surveys: We sampled and examined 1421 deer harvested in enzootic GMUs during 1997 archery, muzzleloader, and rifle seasons (Table 1). About 4.2% .(39/933) of deer harvested in Larimer bounty DAUs (04 and 010) tested. positive for CWO via immunostaining. Four GMUs (9, 191, 19, 20) yielded all of the positive deer detected in Larimer County units (04 and 010) (Table 1; Fig. 1); among these units, prevalence ranged from about 3-16%. Prevalence in sympatric elk populations (0-0.2%; Miller, 1998) was much lower than in deer (P < 0.009). Based on data from 1997 archery, muzzleloader, and rifle seasons, combined estimated mean CWO prevalence in 04 and 010 deer populations (4.2%) appeared to be essentially unchanged from 1995 (5.9%) and 1996 (5.7%); for the 3 GMUs (191, 19, 20) where sufficient samples sizes were available for 3 successive years (1995-1997), prevalence has not differed over time (P ~ 0.3). CWO was detected in about 1.1% (5/449) of deer harvested in South Platte River bottom units (DAU 044 plus GMUs 87, 90, 93, and 95·) (Table 1). Among river bottom and adjacent plains units, positive deer were harvested in GMUs 91, 95, 951, and 96 (Table 1; Fig. 1). None of the 171 deer harvested or culled in other select GMUs (mainly and 104) outside known enzootic areas tested positive for CWO. 66, 67, �153 rg 00o o o o • MD PositiveMdpos.txt • ELK PositiveElkpos.txt .•. WTD PositiveWtdpos.txt WTD NegativeWtdneg.tx o MD NegativeMdneg.txt RiverColo riv CountiesCounties A . /\I o \1---' ___, s Figure 1. About 4.2% of deer harvested in larimer County data analysis units (OAUs) (04 or 010) and about 1.1% of deer harvested in South Platte River bottom units (OAU D44 plus GMUs 87,90, 93, and 95) tested positive for CWO via immunostaining. Most positive cases were from eastern larimer County. As reorted previously (Miller, 1997), the foregoing prevalence estimates may be somewhat liberal because the definition of "positive" included subclinical cases where either histopathological lesions or anti-PrP immunostaining reactions in brain tissue were observed. Of the 44 cases identified, none appeared to be suffering from clinical cwo. Twenty-six of the 44 (59%) positive deer showed both histopathological lesions and immunostaining; the other 18 were classified as positive solely on the basis of immunostaining reactions. Although no known "false positives" were identified among the 171 deer examined from outside known enzootic DAUs (specificity ~0.979), further evaluation of both sensitivity and specificity of existing diagnostic techniques still appears warranted. In contrast to observations derived from targeted surveillance data (Spraker et al., 1997; Miller, 1997), CWO prevalence among male (5.5%) and female (4.8%) mule deer did not differ (P = 0.86) for animals harvested in four Larimer County GMUs (9, 191, 19, 20). Antlerless permit numbers issued for 1997 failed to produce targeted sample sizes of 250 adult (~1 yr) does/DAU: we received 195 usable samples from 04 and 151 from 010, including samples from archery, muzzleloader, and rifle seasons. Additional does will be examined during 1998 to increase sample sizes for comparison of prevalence between sexes. Age distribution of CWO-positive mule deer did not differ from respective sexspecific age distributions of negative animals (Fig. 2). These data support the belief (Miller et al., 1998) that CWO epidemiology is driven primarily by lateral transmission. If maternal transmission were the primary mode of transmission (as is believed to be the case in scrapie of domestic sheep), then an age distribution skewed toward younger aged animals would be expected, particularly among does where older-aged animals are more abundant. In light of an estimated 18-24 mo incubation period for CWO in deer (Williams et al., 1998), the rarity of preclinical disease in yearlings (-16-17-mo-old deer) �154 appears to be the strongest argument against maternal transmission being the most common route for CWO transmission. Although survival rates for male mule deer are substantially lower than for females, CWO appears equally prevalent among males and females and uniformly distributed among respective age classes. These data may reflect the potential inefficacy of random culling in reducing CWO prevalence, although relatively high female densities and the potential for female-male transmission could confound such interpretations. Requiring head submissions increased survey sample sizes >4-fold compared to voluntary submissions in 1996: in 04/010, we collected 938 deer heads, as compared to -200 collected in 1996. Limited licenses also may have aided in increasing deer submissions. Based on harvest estimates from surveyed GMUs (COOW, unpubl. data), compliance with mandatory submission regulations was about 75%. Unstaffed barrel sites still appear to be the most efficient method for sample collection. Epizootiologieal Studies Epidemiology of naturally-occurring CWO in captive mule deer and whitetailed deer (Miller and Wild): Five mule deer and 5 white-tailed deer developed clinical cwo during the last biological year (May, 1997-April 1998) (Fig. 3); two additional mule deer cases and one additiona white-tailed deer case occurred during May-June 1998. In all, CWO affected about 20% (5/25) of the adult (~l-yr-old) mule deer and about 46% (5/11) of the adult white-tailed deer in resident FWRF herds during the last biological year; CWO accounted for 83% of adult mule deer mortality and 100% of adult white-tailed deer mortality. ~ CWO-poalilv. D N.D.1I •.• ~ ...•• r::: • ~• :::J 1:r 7-'D ••,D B. Female mule d.er Age cia •• Rgure 2. Age frequency distributions did not differ between CWO-positive and negative (A) male or (8) female mule deer sampled via harvest surveys. A.ilule du, · ." '0 -·,." z , •• , , •• 7 , •• 3 · ... '0 .! ,." z V •• r Since mule deer were reintroduced into Figure 3. Both mule deer and white-tailed deer herds FWRF in late 1990, 17 of 59 (29%) resident at CDOW's Foothills Wildlife Research Facility animals that survived to >1 yr of age are now infected with chronic wasting disease. have succumbed to CWO. The index case in the most recent epizootic occurred in October 1994; 4-5 clinical cases Annual clinical have occurred in each subsequent biological year (Fig. 3A). prevalence has ranged from 2-20% (Fig. 4). Generating comparable epizootic �155 behavior in a forecasting model (Miller and Mccarty, unpubl. data) requires a transmission coefficient (~) of about 3.5 infectious contacts/infectious animal/yr (Fig. 4); two assumptions of this spreadsheet model, no environmental source of infection and a l2-mo incubation period prior to onset of clinical disease, seem somewhat questionable in light of recent experiences and may need to be reevaluated. _a-wd,.."..DldMr 50 40 ~ 30 fI) '0 [EJ~~oI" 25 20 15 ~ B r:::: '0 ~ fI) ..D E 20 ::::J Z 10 ~ l:. Based on observations made over the last 5 10 12 mo, CWO apparently can be an explosive disease 'in white-tailed deer populations. It follows that the relative rarity of o o 1995 1994 1996 1997 CWO in free-ranging white-tailed deer is Year probably more a function infrequent exposure rather than natural species resistance; white-tailed deer are relatively rare in the foothills of Figure 4. Modeled CWO dynamics eastern Larimer county where CWO is most simulated trends observed in FWRF deer prevalent. The source of infection for since 1994 when relatively high transmission our captive white-tailed deer herd is coefficients (13 = 3.5) were used. unclear. One possibility is introduction via two breeding males (C93, W93) that were intermittently housed with CWO-infected mule deer males; another possibility is transmission from infected mule deer via fenceline contact or environmental contamination. The essentially simultaneous occurrence of CWO in two does that had never been housed with mule deer seems to support the latter alternative. Cases in two additional animals 9 and 11 mo after the index case could be consistent with either a point source of exposure and variable incubation period or lateral transmission with a relatively short (-12 mo) incubation period. Data on pathogenesis, incubation periods, agent shedding, potential environmental sources of infection, and the relationship between infectiousness and onset of clinical signs for CWO in both white-tailed and mule deer are clearly needed to improve understanding of CWO epizootiology in both captive and free-ranging deer. Cattle susceptibility to CWO (Williams and Miller): One calf died from gastrointestinal obstruction and bloat about 2 wk after entering the deer paddocks, but the other 11 remained healthy throughout the first year of this 10-yr experiment. Similarly, 11 of the 12 control fawns from RMANWR remained healthy; one fawn that apparently failed to adjust to captivity died about 1 mo after capture, presumably from hypothermia secondary to malnutrition. Seven of the 25 >1-yr-old naturally-exposed resident FWRF deer developed clinical CWO and died or were euthanized during the first 12 mo of the study, thereby ensuring calves and control deer received some exposure to CWO. ACKNOWLEDGMENTS The statewide CWO monitoring and surveillance program described here relies heavily on efforts of dedicated field personnel throughout the Colorado Division of Wildlife, and truly represents a division-wide effort to improve our understanding and management of this important disease problems. In addition to those specifically listed, we collectively thank all of those �156 regional and area biologists, district and area wildlife managers, volunteers, deer and elk hunters, and others who assisted by submitting suspect cases, harvested animals, or road-killed animals throughout the year. LITERATURE CITED Miller, M. W. 1997. Monitoring. and managing wildlife chronic wasting disease in Colorado. in Wildlife Research Report, Mammals Research, Federal Aid Projects, Job Progress Report, Project W-153-R-10, WP2, J17. Colorado Division of Wildlife, Fort Collins, Colorado, USA, pp. 37-46. 1998. Monitoring and managing wildlife chronic wasting disease in elk. in Wildlife Research Report, Mammals Research, Federal Aid Projects, Job Progress Report, Project W-153-R-11, WP3002, T3. Colorado Division of Wildlife, Fort Collins, Colorado, USA, in press. , M. A. Wild, and E. S. Williams. 1998. Epizootiology of chronic --wasting disease in captive Rocky Mountain elk. J. Wildl. Dis. 34: 532-538. O'Rourke, K. I., T. V. Baszler, J. M. Miller, T. R. Spraker, I. SadlerRiggleman, and D. P. Knowles. 1998. Monoclonal antibody F89/160.1.5 defines a conserved epitope on the ruminant prion protein. J. Clin. Microbiol. 36: 1750-1755. Spraker, T. R., M. W. Miller, E. S. Williams, D. M. Getzy, W. J. Adrian, G. G. Schoonveld, R. A. Spowart, K. I. O'Rourke, J. M. Miller, and P. A. Merz. 1997. spongiform encephalopathy in free-ranging mule deer (Odocoileus hemionus), white-tailed deer (Odocoileus virginianus), and Rocky Mountain elk (Cervus elaphus nelsoni) in northcentral Colorado. J. Wildl. Dis. 33:1-6. Williams, E. S., and S. Young. 1980. Chronic Wasting disease of captive mule deer: A spongiform encephalopathy. Journal of Wildlife Diseases 16: 89-98. , and 1982. Spongiform encephalopathy --Journal of Wildlife Diseases 18: 465-471. __ ~' and Scientifique of Rocky Mountain elk. 1992. Spongiform encephalopathies in Cervidae. Revue et Technique Office International des Epizooties 11: 551-567. , and 1993. Neuropathology of chronic wasting disease in mule --deer (Odocoileus hemionus) and elk (Cervus elaphus nelsoni). Veterinary Pathology 30: 36-45. , M. W. Miller, T. J. Kreeger, H. Van Campen, and T. R. Spraker. 1998. --Susceptibility of cattle to cervid spongiform ~ncephalopathy. unpublished grant proposal, submitted to USDA, CSREES, Competitive Grants Program, 88 pp. c Wildlife Research Veterinarian National Research Initiative �157 Table 1. Results of 1997 CWD harvest surveys -- archery, muzzleloader, & rifle seasons. DEER DAU GMU # Examined # Positive D4 7 29 0 8 160 0 191 212 10 9 74 12 19 123 6 Total 598 28 20 335 11 Total 335 11 29 39 0 Total 39 0 87 37 0 90 1 0 91 54 1 . 92 41 0 93 53 0 94 55 0 95 97 2 951 56 1 96 55 1 Total 449 5 DI0 Boulder Plains Prevalence (95% CI) 0.047 (0.031-0.067) 0.033 (0.016-0.058) 0 (0-0.09) 0.011 (0.004-0.025) �158 �159 AJ;lpendix A STUDY PLAN Pathogenesis Elizabeth of Chronic Wasting S. Williams Disease and Michael in Mule Deer W. Miller Need Chronic wasting disease (CWO) is a transmissible spongiform encephalopathy (TSE)of native North American deer and elk (Williams and Young, 1980, 1982, 1992). The neuropathology of clinical CWO is well-described (Williams and Young, 1980, 1982, 1992, 1993; Spraker et al., 1997), but CWO pathogenesis remains largely unstudied. Among clinical CWO cases, the parasympathetic nucleus of the vagus and the olfactory stria are most consistently and severely affected (Williams and Young, 1993). Typical but less abundant lesions occur in the parasympathetic nucleus of the vagus, and sometimes the olfactory stria, in apparently subclinical CWO cases detected during harvest surveys (E. S. Williams, unpubl. data; T. R. Spraker, pers. comm.). Brain and various lymphoid tissues often stain intensively with anti-prion protein (PrP) immunostaining in clinical cases (Williams and Young, 1992; Spraker et al., 1997; E. S. Williams, unpubl. data; T. R. Spraker, pers. comm.). Immunostaining of brain and select lymphoid tissues, particularly tonsil, are usually positive in subclinical cases. Positive staining of tonsil and brain tissue can occur in the absence of spongiform lesions. Lymphoid Prpsc propagation apparently precedes development of detectable neurological lesions in scrapie (Schreuder et al., 1996, van Keulen et al., 1995, 1996), but not bovine spongiform encephalopathy (Wells et al., 1996). The presence of Prpsc in central nervous system and peripheral tissues has lead to development of immunohistochemistry (Miller et al., 1993; van Keulen et al., 1995, 1996; Schreuder et al., 1996) and immunoblotting (Farquhar et al., 1989; Ikegami et al., 1991; Race et al., 1992) as preclinical and clinical diagnostic tests for scrapie in sheep. These antemortem tests are based on biopsy and examination of lymphoid tissues, typically tonsil, lymph node, and spleen for Prpsc• Positive staining of lymphoid and salivary gland tissues in the absence of either lesions or staining in brain tissue has also been observed in deer and elk, but interpretation of these observations remains equivocal. Improved understanding of CWO pathogenesis would clearly facilitate the interpretation of diagnostic findings in subclinical CWO suspects detected via harvest surveys. Such understanding would also enhance interpretation of data being gathered in studies designed to evaluate antemortem diagnostic tests for CWO. Moreover, reliable pathogenesis data may offer valuable insights into initiation, duration, and routes of agent shedding, as well as other important aspects of CWO epizootiology. Here, we propose to study the pathogenesis of CWO in mule deer after oral exposure to infectious brain material. Objectives The specific objectives of this study are to: (1) describe the pathogenesis of CWO in mule deer after oral exposure to infectious material using histopathology, immunohistochemistry, and Western blot analyses; and (2) compare susceptibility, pathogenesis, and incubation periods between male and female deer. Additionally, animals challenged in this study will serve as controls for ongoing cattle challenge trials (Williams et al., 1997). �160 Materials and Methods We will study the pathogenesis of CWO in mule deer after oral exposure to infectious brain material. Twenty 5- to 6-mo-old mule deer fawns (10 males, 10 females) will be captured from the Rocky Mountain Arsenal National Wildlife Refuge (RMANWR) and transported to the Colorado Division of Wildlife's Foothills Wildlife Research Facility (FWRF) in Fort Collins, Colorado. Previous surveillance work has revealed that deer residing at RMANWR are presently unaffected by CWO, thereby ensuring fawns will not be exposed to CWO until inoculated experimentally. For capture, fawns will be anesthetized with tiletamine HCl and zolazepam (Telazol~; 5 mg/kg) and xylazine HCl (2.5 mg/kg) or carfentanil HCl (0.03 mg/kg) and xylazine (1 mg/kg) delivered intramuscularly (1M) via projectile syringe; if carfentanil is used, anesthesia will be antagonized with naltrexone HCl (100 mg/mg carfentanil) delivered intravenously (IV)(25%) and subcutaneously (SC) (75%) after initial handling and processing. Additional xylazine (20-100 mg IV or 1M) will be administered as needed to keep fawns sedated until they arrive at FWRF. (See appendix for detailed capture protocols.) Upon arrival at FWRF, residual sedation will be antagonized with IV yohimbine HCl (0.25 mg/kg). Once fawns are able to swallow effectively, each will receive about 5 g of homogenized brain material pool collected from mule deer previously diagnosed with spongiform encephalopathy; presence of scrapieassociated fibrils in this homogenate was previously confirmed via negativestain electron microscopy (E. S. Williams, unpubl. data). This homogenate is the same being used for oral and intracranial cattle challenges (Williams et al., 1997). Homogenate will be deposited into the posterior oropharynx using a modified syringe. Once fawns have swallowed the homogenate, they will be released into a 3 ha paddock. Alfalfa hay, pelleted supplemental diets (highenergy and "browser" rations), mineralized salt blocks, and water will be provided ad libitum. All fawns will be observed daily by animal caretakers and evaluated at least monthly by an attending veterinarian for signs of CWO. To study CWO pathogenesis, two fawns' (one male, one female) will be randomly sacrificed at 3, 6, 12, 18, 24, 30, 36, or 42 mo after oral challenge. At the time of sacrifice, fawns will be anesthetized with tiletamine HCl and zolazepam (Telazol~; 5 mg/kg) and xylazine Hel (2.5 mg/kg) delivered via projectile syringe. We will collect blood, saliva, feces, and cerebrospinal fluid from each fawn, then administer about 400 mEq KCl intravenously to induce cardiac arrest. The incubation period of CWO after oral challenge is unknown; however, because clinical CWO developed in mule deer 18 to 24 mo after intracranial challenge (E. S. Williams, unpubl. data), we anticipate that most study animals surviving >24 mo will develop Cwo 24 to 36 mo after oral challenge. Deer developing severe clinical CWO (characteristic behavioral changes accompanied by estimated >20% weight loss) will be euthanized as described above. Two age-matched control deer (one male, one female) will be collected from the RMANWR on the same schedule as challenged deer are sacrificed. Control deer will be anesthetized in the field as described above, sampled, and euthanized with KCl; if field anesthesia becomes infeasible, control deer will be killed by shooting in the neck with a highpowered (~0.223 cal) rifle. Saliva, feces, blood, and cerebrospinal fluid will be stored at -70 C for potential evaluation via Western blot or infectivity studies. Carcasses will be transported to the Wyoming State Veterinary Laboratory (WSVL) for complete necropsy and tissue sampling. Samples collected will include brain, spinal �161 cord, and numerous other tissues (Table 1). These tissues will be divided and subsamples fixed in 10% neutral phosphate-buffered formalin or in PLP solution (periodate, lysine, paraformaldehyde) at a concentration of 2% paraformaldehyde (van Keulen et al., 1996); additional subsamples will be stored unfixed at -70 C for studies of Prpsc accumulation. Carcasses will be incinerated once sampling is completed. Tissues will be examined via histopathology, immunocytochemistry, Western blot, and/or negative-stain electron microscopy. Histopathology will be conducted as previously described (Williams and Young, 1993). Central nervous system tissues will be removed within 4 hr following euthanasia. Brains will be sagittally sectioned and half fixed in 10% neutral phosphate-buffered formalin. Samples of spinal cord and other tissues (Table 1) will be similarly fixed. The other half of the brain and portions of the spinal cord will be frozen at -70 C. Paraffin blocks will be prepared from olfactory tubercle and cortex; cerebral cortex (frontal, parietal, temporal, and occipital lobes); basal ganglia; three levels of thalamus; two levels of mesencephalon; three levels of pons and cerebellum; medulla oblongata at the obex; medulla caudal to the obex; and multiple levels of spinal cord to include cervical, thoracic, and lumbar regions. Tissues will be sectioned at 5-6 ~m and stained with hematoxylin and eosin, Bodian's silver stain, luxol fast blue stains, and glial fibrillary acidic protein (Dako, Carpinteria, California) immunocytochemistry for astrocytosis as appropriate. Lesions will be compared to those previously described for CWO in mule deer and elk and other natural TSEs of animals (Williams and Young, 1993; Hadlow, 1996). We will test for Prpsc in formalin- or PLP solution-fixed, paraffin embedded tissues using minor modifications of the technique of van Keulen et al. (1995). We will use a polyclonal rabbit antiserum against ME7 scrapie strain passaged in mice (Rubenstein et al., 1986) (Department of Virology, New York State Office of Mental Retardation and Developmental Disabilities, Staten Island, New York) or mouse monoclonal antibody against ovine Prpsc (F89/160.1.5; O'Rourke et al., 1997) (USDA/ARS, Pullman, Washington). Briefly, paraffin embedded tissue sections will be deparaffinized, treated with 99% formic acid for 30 minutes, rinsed with PBS, and autoclaved for 10 minutes at 122 C in 10 x Automation buffer. Sections will be incubated in 4% normal goat serum for 20 minutes, incubated overnight at 4 C in primary antibody at 1:1500 dilution, rinsed, and incubated for 30 minutes with biotinylated anti-rabbit or anti-mouse IgG (Vector Laboratories, Burlingame, California). Slides will be rinsed, and ABC reagent (Vectastain Elite ABC kit, Vector Laboratories) applied for 30 minutes, rinsed and incubated with AEC substrate solution (Dako Laboratories, Carpinteria, california) for 10 minutes. Slides will be counter stained with Harris hematoxylin, rinsed, "blued" with lithium carbonate, rinsed again, and coversli~d. Positive and negative brain sections, normal rabbit serum, and PBS will be used as controls. Adequate primary antibody is available for the entire study. L Western blot assays will be conducted with modifications of the techniques of Rubenstein et al. (1986) and Collinge et al. (1996). Brain or lymphoid tissues will be homogenized in Tris buffer saline (TBS) at 50 mg/500 ~l followed by a 5 minute low speed centrifugation. 100 ~l supernatant is combined with 6 ~l 22% sarcosyl in TBS, mixed, 4 ~l (5mg/ml) proteinase K is added, mixed again, then incubated at 37 C for 1-2 hours. Aliquots will be collected for protein concentration determination using Pierce's Micro BCA protein assay. Equal volume of concentrated sample buffer (120 roM Tris-HCl pH 6.8, 142 roM BME, 200 roM DTT, 4% SDS, 0.02% BPB, 20% glycerol) will be added and the sample boiled for 5 minutes. Hot samples will be loaded on Mini-Vertical 4%/12% SDS-PAGE �162 (Laemmli). The samples will be electrophoresed at 100 v for 1.5-2 hours with BioRad's Protein SOS-PAGE molecular weight markers. A semidry transfer to 0.45 ~m pure nitrocellulose will be conducted using Towbin buffer according to manufacturer's instructions. Post-transfer, the gel will be Coomassie Blue G250 stained while the membrane is floated on TBST20 (10 roM Tris pH 8.0, 150 roM NaCI, 0.1% Tween2o) until wet, then submerged and rinsed. The membrane will be incubated in 1% NGS -TBST2o for 30 - 60 minutes at room temperature, washed two-three times for 10 minutes each in TBST2o, and TBST20 1:5,000 diluted primary antibody (polyc1onal rabbit or monoclonal mouse anti-Prpsc) added to the blot and incubated 2 hours at room temperature or overnight at 4 C. The antibody will be removed, the membrane washed two-three times for 10 minutes each, and secondary antibody (Promega's goat anti-rabbit conjugated alkaline phosphatase) diluted in TBST20 (per manufacturer's instructions) added and incubated 2 hours at room temperature. The membrane will be again washed two to three times followed by three 2 minute washes in distilled water and then a 5 minute TBS wash done twice. Promega's BCIP/NBT color development solution (made according to manufacturer's instructions) will be added and development monitored. The reaction will be stopped by several distilled water rinses. The membrane and stained gel will be recorded via CCO camera, digitized, and quantitated (Bio-Rad Molecular Analysis System). Scrapie-associated fibrils will be purified using a modification of the techniques of Hilmert and Oiringer (1984). One to two g of brain will be homogenized in 10% sarcosyl (pH 7.4) and N-octano1 added for a 30 minute incubation at room temperature. The homogenate will be centrifuged for 10 minutes at 3,000 x G (JA-20 rotor, J-21 Beckman centrifuge) and supernatant collected and centrifuged for 30 minutes at 22,000 X G. The supernatant will then be transferred to Beckman quick seal tubes and centrifuged at 215,000 X G (75T rotor, L8-80 Beckman centrifuge) for 2 hours. The pellet will be resuspended in 1% sarcosyl and 10% NaCI, stirred for 60 minutes at 37 C, and microfuged for 15 minutes. The pellet will then be resuspended -in 1% sarcosyl, 10% NaCI, and 5 ~g proteinase Klml added and stirred for 2 hours at 37 C. The suspension will be microfuged for 10 minutes. Negative staining will be conducted according to Scott et ale (1987, 1990). The pellet will be resuspended in 50 ~l distilled water and a 300 mesh plastic coated, carbon stabilized grid floated on a drop of suspension for 10 seconds, blotted and negative stained with 2% potassium phosphotungstate, pH 6.6. Some additional modifications of the procedures for SAF detection (Stack et al., 1995; 1996) are currently being tested at WSVL. We will describe distribution of Prpsc and microscopic lesions as determined by each of the foregoing assays, and relate these to both the chronological progression of clinical CWO and the potential shedding of agent from infected deer. Because our study is largely descriptive, no statistical analyses of pathology data will be attempted. Among fawns surviving long enough to develop clinical CWO, mean incubation times for males and females will be compared using Student's t-test (a = 0.1), and the proportions of males and females developing clinical cwo will be compared using Fisher's exact probability test (a 0.1). = Literature Cited Collinge, J., K. C. L. Sidle, J. Meads, J. Ironside, and A. F. Hill. 1966. Molecular analysis of prion strain variation and the aetiology of 'new variant' CJO. Nature 383: 685-690. Farquhar, C. F., R. A. Somerville, and L. A. Ritchie. 1989. Post-mortem immunodiagnosis of scrapie and bovine spongiform encephalopathy. Journal of Virological Methods 24: 215-222. �163 Hadlow, W. J. 1996. Differing neurohistologic images of scrapie, transmissible mink encephalopathy, and chronic wasting disease of mule deer and elk. In Bovine spongiform encephalopathy The BSE dilemma, Vol. C. J. Gibbs, Jr. (ed.). Springer-Verlag, New York, New York, pp. 122-137. Hilmert, H., and H. Diringer. 1984. A rapid and efficient method to enrich SAF-protein from scrapie brains of hamsters. Biosciences Reports 4: 165170. Ikegami, Y., M. Ito, H. Isomura, E. Momotani, K. Sasaki, Y. Muramatsu, N. Ishiguro, and M. Shinagawa. 1991. Pre-clinical and clinical diagnosis of scrapie by detection of PrP protein in tissues of sheep. Veterinary Record 128: 271-275. Miller, J. M., A. L. Jenny, W. D. Taylor, R. F. Marsh, R. Rubenstein, and R. E. Race. 1993. Immunohistochemical detection of prion protein in sheep with scrapie. Journal of Veterinary Diagnostic Investigation 5: 309-316. Race, R. E., D. Ernst, A. L. Jenny, W. D. Taylor, D. Sotton, and B. Caughhey. 1992. Diagnostic implications of detection of proteinase K-resistant protein in spleen, lymph nodes, and brain of sheep. American Journal of Veterinary Research 53: 883-889. Rubenstein, R., R. J. Kascsak, P. A. Merz, M. C. Papini, R. I. Carp, N. K. Robakis, and H. M. Wisniewski. 1986. Detection of scrapie-associated fibril (SAF) proteins using anti-SAF antibodies in non-purified tissue preparations. Journal of General Virology 67: 671-681. Schreuder, B. E. C., L. M. J. van Keulen, M. E. W. Vromans, J. P. M. Langveld, and M. A. Smits. 1996. Preclinical test for prion diseases. Nature 381: 563. Scott, A. C., S. H. Done, C. Venables, and M. Dawson. 1987. Detection of scrapie-associated fibrils as an aid to the diagnosis of natural sheep scrapie. Veterinary Record 120: 280-281. Scott, A. C., G. A. H. Wells, M. J. Stack, H. White, and M. Dawson. 1990. Bovine spongiform encephalopathy: Detection and quantitation of fibrils, fibril protein (PrP) and vacuolation in brain. Veterinary Microbiology 23: 295-305. Spraker, T. R., M. W. Miller, E. S. Williams, D. M. Getzy, W. J. Adrian, G. G. Schoonveld, R. A. Spowart, K. I. O'Rourke, J. M. Miller, and P. A. Merz. 1997. Spongiform encephalopathy in free-ranging mule deer (Odocoileus hemionus), white-tailed deer (0. virgianus), and Rocky Mountain elk (Cervus elaphus nelsoni) in northcentral Colorado. Journal of Wildlife Diseases 33: 1-6. Stack, M. J., A. M. Aldrich, A. D. Kitching, and A. C. Scott. 1995. Comparative study of electron microscopal techniques for the detection of scrapie-associated fibrils. Research in Veterinary Science 59: 247-254. Stack, M. J., A. M. Aldrich, A. D. Kitching, and A. C. Scott. 1996. Comparison of biochemical extraction techniques for the detection of scrapieassociated fibrils in the central nervous system of sheep naturally affected with scrapie. Journal of comparative Pathology 115: 175-184. van Keulen, L. J. M., B. E. C. Schreuder, R. H. Meloen, M. Poe len-van den Berg, G. Mooij-Harkes, M. E. W. Vromans, and J. P. M. Langerveld. 1995. Immunohistochemical detection and localization of prion protein in bran tissue of sheep with natural scrapie. Veterinary Pathology 32: 299-308. van Keulen, L. M. J., B. E. C. Schreuder, R. H. Meloen, G. Mooij-Harkes; M. E. W. Vromans, and J. P. M. Langveld. 1996. Immunohistochemical detection of prion protein in lymphoid tissues of sheep with natural scrapie. Journal of Clinical Microbiology 34:1288-1231. Wells, G. A. H., M. Dawson, S. A. C. Hawkins, A. R. Austin, R. B. Green, I. Dexter, M. W. Horigan, and M. M. Simmons. 1996. Preliminary observations on the pathogenesis of experimental bovine spongiform encephalopathy. In �164 Bovine spongiform encephalopathy The BSE Dilemma, Vol. C. J. Gibbs, Jr. (ed.). Springer-Verlag, New York, New York, pp. 28-44. Williams, E. S. and S. Young. 1980. Chronic wasting disease of captive mule deer: A spongiform encephalopathy. Journal of Wildlife Diseases 16: 89-98. Williams, E. S. and S. Young. 1982. Spongiform encephalopathy of Rocky Mountain elk. Journal of Wildlife Diseases 18: 463-471. Williams, E. S., and S. Young. 1992. Spongiform encephalopathies of Cervidae. Scientific and Technical Review Office of International Epizootics 11: 551-567. Williams, E. S. and S. Young. 1993. Neuropathology of chronic wasting disease of mule deer (Odocoileus hemionus) and elk (Cervus elaphus nelsoni). Veterinary Pathology 30: 36-45. Williams, E. S., M. W. Miller, T. J. Kreeger, and H. Van Campen. 1997. Susceptibility of cattle to cervid spongiform encephalopathy. Unpublished study plan, 66 pp. Table 1. Tissues to be collected and detection of Prpsc. Nervous Tissue brain (multiple levels) pituitary CSF dura spinal cord (cervical, thoracic, lumbar) dorsal root ganglia trigeminal ganglia stellate ganglia sciatic nerve radial nerve Muscle Tissue diaphragm semitendinosus muscle triceps muscle longissimus dorsi muscle Alimentary Tissue tongue submandibular lymph node parotid salivary gland esophagus rumen omasum abomasum duodenum distal ileum and Peyer's patches spiral colon feces pancreas liver from deer exposed to CWO for histopathology Lymphoid Tissue spleen thymus tonsil submandibular lymph node retropharyngeal lymph node bronchial lymph node mediastinal lymph node hepatic lymph node mesenteric lymph nodes ileocecal lymph node hepatic lymph node superficial cervical lymph node popliteal lymph node Other Tissues kidney urine adrenal gland .lung nasal mucosa left ventricle blood (serum, buffy bone marrow skin (head) bone (rib) coat, clot) �91 Colorado Division of Wildlife Wildlife Research Report July 1999 JOB PROGRESS REPORT State of __,..--!:C~o~lo::.!.r=ad...,o,,-_ Project No. __ -..l..W!--...:..1=.:53:!...-~R~-~12~_ Work Package No. _---"'-30;:....;0::..,:1'-_ Task No. 3 ------~--------- Period Covered: Cost Center 3430 Mammals Program Deer Management Monitoring and Managing Disease in Deer Chronic Wasting July 1, 1998 - June 30, 1999 Authors: M. W. Miller, C. T. Larsen, and J. Gross Personnel: R Kahn. M. Leslie, K. I. O'Rourke, Williams T. R Spraker, E. Wheeler, M. A. Wild, and E. S. ABSTRACf . Deer from throughout Colorado were examined for occurrence of chronic wasting disease using a combination of targeted surveys and harvest/road-kill surveys. Between June 1998 and May 1999, 8 chronic wasting disease (CWD) cases were diagnosed among 25 "suspect" deer submitted from known endemic portions of northeastern Colorado; CWD was not diagnosed in any of 10 additional "suspect" deer submitted from elsewhere in Colorado. All confirmed CWD cases originated in game management units (GMUs) where the disease had been detected previously. We sampled over 1800 harvested or road-killed deer to randomly survey for CWD in select DAUs. Immunohistochemistry (llIC)-based prevalence in Larimer County (5%; 39/780) and South Platte River bottom/plains (1.6%; 5/315) DADs were unchanged from previous years. We detected no evidence ofCWD in any of369 samples from Middle Park, or in any of over 300 samples from other GMUs outside northeastern Colorado. Late doe hunts in Larimer County DAUs provided an additional 212 samples for comparison of CWD prevalence between sexes; based on data compiled over the last 2 years, CWD prevalence does not differ between male and female mule deer (P = 0.72). Since June 1998, CWD affected 8 of 29 adult (z l-yr-old) mule deer in the resident Foothills Wildlife Research Facility herd. No signs of neurological disease were observed in any of the 11 cattle (subjects) or 11 mule deer from the Rocky Mountain Arsenal National Wildlife Refuge (RMANWR)(contact controls) housed with naturally-infected resident deer since July 1997. Using mc, Prpres was detected in brain tissue (medulla oblongata at the obex) from 1 of2 mule deer experimentally inoculated with a 5 g oral dose of brain tissue homogenate from CWD-infected deer 6 or 9 mo earlier. One of two deer examined 16 mo after inoculation (Pl) had lesions of spongiform encephalopathy, and both were mC-positive. Two deer examined 3 rno PI and 2 examined 12 rno PI were negative, as were all 10 uninoculated controls collected from RMANWR at the same intervals. One of the 8 remaining deer inoculated in December 1997 began �92 showing early clinical signs of CWD ~ 15 mo PI; a second began showing early signs 2-3 wks later, and subtle signs were noted in at least 4 of 8 inoculated deer alive 18 mo PI. An epidemic model of was developed to simulate the dynamics of chronic wasting disease in mule deer populations. As seen in earlier modeling exercises, simulations resulted in projections wherein either the disease or the deer population was eliminated; stable coexistence of chronic wasting disease could not be achieved in simulated populations. Productivity (i.e., harvest) was reduced in populations by very low rates of prevalence. Spread of CWD within a simulated population was highly sensitive to transmission rate, and very small decreases in transmission efficiency resulted in notable decreases in prevalence. Simulated test and slaughter programs revealed the importance of initiating control while CWD prevalence was low « 0.05). Low rates of test and slaughter (e.g., < 20% of the infected population) effectively eliminated wasting disease in simulated populations if control measures were initiated while prevalence was low (i.e., 0.01), but the likelihood of control diminished rapidly as disease prevalence increased. Test and slaughter programs will require an effort sustained over many decades to ensure elimination of chronic wasting disease. A statewide plan for monitoring and managing CWD in deer and elk was drafted. The plan has been distributed for internal and limited external review, and will be finalized by 30 September 1999. cwn �93 MONITORING AND MANAGING CHRONIC WASTING DISEASE IN DEER M. W. Miller and C. T. Larsen P. N. OBJECTIVES '"1. Design, conduct, and report results of: a. targeted surveillance to estimate and detect changes in distribution of chronic wasting disease (CWD) in free-ranging deer populations; and b. harvest or road-kill surveys to estimate and detect changes in prevalence of CWD in enzootic deer populations. ,2. Design, conduct, and report results of experimental studies using captive deer naturally or experimentally infected with CWD. AGREEMENT OBJECTIVES 1. Conduct and report results of targeted surveillance to estimate and detect changes in distribution of CWD in free-ranging deer populations statewide. 2. Conduct and report results of harvest surveys to estimate prevalence ofCWD in DAUs D4, D9 (GMUs 18,28,37,371), DI0( +GMU 29), and D44 (+ GMUs 87, 90, 93, arid 95). 3. Continue an experiment evaluating cattle susceptibility to CWD via natural contact exposure. 4. Continue to study pathogenesis ofCWD in mule deer. 5. Develop an epidemic model ofCWD dynamics in deer populations. 6. Observe epidemiology of naturally-occurring CWD in captive deer. MATERIALS AND METHODS Surveillance We monitored deer populations throughout Colorado for occurrence of CWD using a combination of targeted surveillance and harvest or road-kill surveys. These were organized and conducted as follows: �94 Targeted (= clinical disease) surveillance: Deer showing clinical signs consistent with those seen in chronic wasting disease were collected by field personnel statewide and brain tissues examined for evidence of spongiform encephalopathy. The "suspect case" profile was defined as follows: • Species: mule deer white-tailed deer • Age: ::::18 months • Signs: emaciated and abnormal behavior &lor indifference to human activity &lor increased salivation &lor tremor, stumbling, incoordination &lor difficulty or inefficiency in chewing/swallowing increased drinking and urination &lor Where possible, submissions were subjected to complete necropsy; in some situations, only heads were available for examination and sampling. In all cases, histopathology of brain tissue (Williams and Young 1993) was used to diagnose CWO; in some cases, immunohistochemistry (ll:IC) or other ancillary tests were used to confirm or support diagnoses. Harvest surveys: In order to obtain reliable estimates of CWO prevalence that will serve as a basis for monitoring responses to management interventions, we continued conducting harvest surveys on select deer populations. During the 1998-1999 hunting seasons, fresh brain and select lymphatic tissues were collected from deer harvested in enzootic GMUs; deer harvested or culled in other select GMUs throughout Colorado were also sampled as negative controls. Brain tissues were examined at the Colorado State University Diagnostic Laboratory for anti-PrP immunostaining reactions (O'Rourke et al., 1998) and histopathological lesions (Williams and Young, 1993) consistent with CWO infection. Epizootiological Studies Epizootology of naturally-occurring CWO in captive mule deer and white-tailed deer: Naturallyoccurring CWO was a sporadic disease of resident FWRF mule deer prior to 1985 (Williams and Young, 1992), and also has occurred sporadically since 1994; no cases had been observed in resident white-tailed deer since their addition to FWRF in 1993. We maintained and observed 29 ~ l-yr-old mule deer and 5 ~ l-yr-old white-tailed deer in paddocks at CDOW's Foothiils Wildlife Research Facility (FWRF) during May-September 1998. Deer received natural forage, pelleted rations (high energy supplement and "browser" diet), and alfalfa hay; water and mineralized salt were available ad libitum. All deer were evaluated daily for clinical signs of CWO (and other health problems) in conjunction with routine feeding and handling activities. Resident mule deer also represented the source of infection (= "treatment") in an ongoing study of cattle susceptibility to CWO (see below). Cattle susceptibility to CWO (Williams and Miller): We continued monitoring cattle for clinical evidence of natural CWO transmission as part of a coordinated interagency effort to study cattle susceptibility to CWO. Twelve 4-mo-old calves purchased from a private ranch located near Sheridan, WY, were placed in paddocks at the FWRF with naturally-infected captive mule deer in July 1997 (see above for description of resident deer herd). Twelve 4-mo-old mule deer fawns were captured at the Rocky Mountain Arsenal National Wildlife Refuge (RMANWR) in late September and placed in the �95 same paddocks as contact controls for natural transmission (extensive ongoing surveillance of RMANWR deer has continued to confirm that this population is free of CWD). CWD Pathogenesis in mule deer (Williams and Miller): In a separate but related study, 20 additional 6mo-old mule deer fawns were captured at the RMANWR in early December. Each fawn was given a single oral dose of about 5 g of fresh brain tissue homogenate from captive mule deer with clinical CWD. Fawns were then released into a separate paddock physically removed from the contact transmission study. These experimentally-infected fawns serve two purposes: as controls for domestic calves orally inoculated with a single oral dose of about 50 g of this same CWD brain homogenate (Williams et al., 1998), and as subjects of a study on the pathogenesis of CWD in mule deer. We have randomly sacrificed 2 experimentally-infected deer at 3,6,9, 12, and 16 mo after inoculation; we also collected 2 age- and sex-matched control deer from RMANWR at each time step. Complete necropsies were performed and samples collected as described in the original study plan. . Select sections of brain tissue (obex) were examined after staining with hematoxylin and eosin (H&E) or anti-PrP immunostain (F89/160.1. 5; O'Rourke et al., 1998); examination of other tissues is pending. Epidemic modeling (Gross and Miller): We developed a mechanistic model to simulate the dynamics of chronic wasting disease in mule deer populations. The model projected age-specific disease dynamics, changes in population size, and control strategies involving selective removal of infected animals or changes in rates of disease transmission. Model parameters were estimated from observations of infected and uninfected deer in Colorado and Monte Carlo techniques were used to evaluate likely responses. Appendix A provides a more detailed description of our model. CWD Monitoring and Management Plan A statewide plan for monitoring and managing CWD in deer and elk was drafted. RESULTS AND DISCUSSION Surveillance Targeted (= clinical disease) surveillance: Between June 1998 and May 1999, 8 chronic wasting disease (CWD) cases were diagnosed among 25 "suspect" deer submitted from known endemic .portions of northeastern Colorado; CWD was not diagnosed in any of 10 additional "suspect" deer submitted from elsewhere in Colorado. All CWD cases confirmed in 1998-1999 were mule deer, and all originated in game management units (GMUs) where the disease had been detected previously. Harvest surveys: We sampled over 1800 harvested or road-killed deer to randomly survey for CWD in select DADs. ruC-based prevalence in Larimer County (5%; 39/780) and South Platte River bottom/plains (1.6%; 5/315) DADs were unchanged from previous years. We detected no evidence of CWD in any of369 samples from Middle Park, or in any of over 300 samples from other GMUs outside northeastern Colorado. Late doe hunts in Larimer County DADs provided an additional 212 samples for comparison of CWD prevalence between sexes; based on data compiled over the last 2 years, CWD prevalence does not differ between male and female mule deer (P = 0.72). �96 Survey data gathered during 1996-1999 indicate that CWD is generally more prevalent and more widely distributed in mule deer than in white-tailed deer in northeastern Colorado (Fig. 1). However, limited data from white-tailed deer harvested near the mouth of the Big Thompson Canyon west of Loveland suggest CWD could become quite prevalent among white-tailed deer (Fig. IB). In light of this potential and the apparent spread of CWD eastward along the Big Thompson/South Platte River corridor in both mule deer and white-tailed deer (Fig. 1), natural spread and eventual infection of more densely populated white-tailed deer habitats in the midwestern and eastern US should be anticipated in the coming decades unless management interventions are successfully identified and implemented. A.Muledeer Epizootiological Studies Epidemiology of naturally-occurring CWD in captive mule deer and whitetailed deer: Since June 1998, CWD affected 8 of 29 adult (~ l-yr-old) mule deer in the resident Foothills Wildlife Research Facility herd (Fig. 2A). Two other adult mule deer died between June and May; enterotoxemia was diagnosed in one, and pasteurellosis in the other. Although neither showed neuropathology consistent with CWD, 1 had IHC staining consistent with preclinical CWD. Resident mule deer also represented the source of infection (treatment) in an ongoing 10-yr study of cattle susceptibility to CWD (see below). One of 5 white-tailed deer remaining in our resident FWRF herd developed CWD and died in June 1999. In all, CWD has claimed 7 of 11 captive white-tailed deer held at FWRF since 1993 (Fig. 2B). One of the 4 remaining individuals is showing early signs of CWD. I B. White-tailed deer Figure 1. Estimated CWD prevalence (%) among subpopulations of (A) mule deer and (B) white-tailed deer in northeastern Colorado. Data were aggregated via known deer movement patterns. Prevalence estimates are based on IHC reactions only (IHC+ltotal). Evaluation of tonsillar biopsies from captive deer are still pending. Lack of results has precluded evaluation of tonsillar biopsy as a potential antemortem test for CWD in deer. Cattle susceptibility to CWD (Williams and Miller): No signs of neurological disease were observed in any of the II cattle (subjects) or II mule deer from the Rocky Mountain Arsenal National Wildlife Refuge (RMANWR)(contact controls) housed with naturally-infected resident deer since July 1997. Neither of2 contact control deer that died during October-June showed neuropathology consistent with CWD, although both suffered from chronic weight loss; malnutrition probably contributed to at least I of these deaths. Fifteen of the 36 > l-yr-old naturally-exposed resident FWRF deer developed clinical �97 CWD and died or were euthanized during the first 24 mo of this study, thereby ensuring calves and control deer have received considerable exposure to CWD. Similarly, all 11 cattle exposed to a single 50 g oral dose of brain tissue homogenate from CWD-infected deer remained healthy 21 mo postinoculation (through 30 June 1999) (E. S. Williams, pers. comm.), as did all cattle exposed to CWD via intracerebral inoculation 22 mo ago (R. 1. Cutlip, pers. commun.). A. lD j - ".dee, ,._.-. :JJ ,-- •• : : :: o ~ 2l i 10 :: :: r-': : ::: • :: :::: :::: :::: ::: !---:: :: r'1i i~H:: ~~ ~i ~~ :: :: :: !_.-:: :: :: ! !; H ~~ ~. .: :. : 'III' 8. WlitR-t.Ied dHr lD .., i Xl o t 2) z Pathogenesis of CWD in mule deer (Williams and Miller): rcs Using ruc, Prp was detected in brain tissue (medulla oblongata at the obex) from 1 of2 muIe deer experimentally inoculated with CWD 6 or 9 mo earlier. One of two deer examined 16 mo after inocuIation (PI) had lesions of spongiform Figure 2. Chronic wasting encephalopathy, and both were ruC-positive. Two deer disease continued to affect both mule deer and whiteexamined 3 rno PI and 2 examined 12 mo PI were negative, as tailed deer held at FWRF. were all 10 uninoculated controls collected from RMANWR at the same intervals. One of the 8 remaining deer inoculated in December 1997 began showing early clinical signs of CWD ~ 15 mo PI; a second began showing early signs 2-3 wks later, and subtle signs were noted in at least 4 of 8 inocuIated deer alive 18 mo PI. ResuIts from examination of lymphoid tissues are pending. Epidemic modeling (Gross and Miller): Simulations ofCWD dynamics that used a broad range of parameter values resulted in projections in which either the disease or the deer population was eliminated. We did not identify a set of realistic parameters that resulted in stable coexistence of chronic wasting disease in deer populations. Population productivity (i.e., harvest) was reduced in populations by very low rates of prevalence due to the combined effects of a reduction in per-capita production and a decrease in population density. Spread of the disease was highly sensitive to the rate of transmission, and a very small decreases in the efficiency of transmission resuIted in notable decreases in rate of spread of the disease. Simulatedjest and slaughter programs revealed the importance of initiating control activities while prevalence of the disease was low « 0.05). Low rates of test and slaughter (e.g., < 20% of the infected population) effectively eliminated wasting disease in simulated populations if control measures were initiated while prevalence was low (i.e., 0.01), but the likelihood of control diminished rapidly as disease prevalence increased. Test and slaughter programs that examine will require an effort sustained over many decades to ensure elimination of chronic wasting disease. CWD Monitoring and Management Plan The draft monitoring and management plan (Appendix B) has been distributed for internal and limited external review, and will be finalized by 30 September 1999. Surveillance activities for 1998-1999, as well as surveillance planned for 1999-2000, reflect prioroties established in the monitoring portion of this plan. �98 ACKNOWLEDGMENTS . The statewide CWD monitoring and surveillance program described here relies heavily on efforts of dedicated field personnel throughout the Colorado Division of Wildlife, and truly represents a division-wide effort to improve our understanding and management of this important disease problems. In addition to those specifically listed, we collectively thank all of those regional and area biologists, district and area wildlife managers, volunteers, deer hunters, and others who assisted by submitting suspect cases, harvested animals, or road-killed animals throughout the year. LITERATURE CITED K. 1, T. V. Baszler, 1. M. Miller, T. R Spraker, 1 Sadler-Riggleman, and D. P. Knowles. 1998. Monoclonal antibody F89/160.1.5 defines a conserved epitope on the ruminant prion protein. }. Clin. Microbiol. 36:1750-1755. Williams, E. S., and S. Young. 1993. Neuropathology of chronic wasting disease in mule deer (Odocoileus hemionus) and elk (Cervus elaphus nelsoni). Veterinary Pathology 30:36-45. _--" M. W. Miller, T. 1. Kreeger, H. Van Campen, and T. R Spraker. 1998. Susceptibility of cattle O'Rourke, to cervid spongiform encephalopathy. unpublished grant proposal, submitted to USDA, CSREES, National Research Initiative Competitive Grants Program, 88 pp. Prepared by _ Michael W. Miller Wildlife Research Veterinarian �99 Appendix A CWO: Simulating Chronic Wasting Disease User's Manual and Model Documentation Version 1.0 June 1999 John E. Gross Natural Resource Ecology Laboratory Colorado State University Ft. Collins, CO USA 80523-1499 Email: JohnG@NREL.ColoState.edu �100 CWD: Simulating Chronic Wasting Disease Summary CWO is an individual-based model that represents the most important processes determining the dynamics of animal populations over periods of decades or centuries. It simulates reproduction, disease dynamics, and the interaction of the animals with environmental characteristics that can change through time. Each animal in the population is explicitly represented and individual characteristics including age, sex, and disease state are tracked for the life of an individual. The model simulates 1 to 3 disease period, and includes the capacity to model a wide variety of harvest scenarios. Files are written that facilitate analyses of population responses to disease, harvest, or environmental variation, including changes in disease prevalence, disease-caused mortality, harvest, and changes in population age/sex structure. CWO has a flexible structure and there are no inherent practical limitations to the . number of individuals that can be simulated during a run. Practical limits to the number of individuals are set only by the capabilities of the computer on which the model is run. Introduction To make informed decisions on management of large ungulates, biologists need to evaluate the consequences of alternative management actions on population processes. . Management' actions can include hunting by the public or agency personnel, vaccination, selective harvest, habitat modification, or a combination of these actions. The consequences of these decisions depend on a complex of factors, and it is thus difficult to evaluate the relative merits of alternative management plans. CWO was designed to simulate management actions that might be used in attempts to eliminate CWO or to inhibit its spread. The model simulates each individual in the population and explicitly represents reproduction, natural mortality, disease transmission, non-selective harvests (Le., hunting), and selective harvests (e.g., test and slaughter). The use of an individual-based model is particularly appropriate for CWO because stochastic effects are important in small populations, or where a small number of individuals have a large effect on population processes. This situation clearly applies to CWO, where the behavior of a small number of infectious individuals can have a profound influence on long-term dynamics of a much . larger population. �101 An Overview of Model Structure and General Features Animal populations are simulated within a landscape that potentially can consist of a single or multiple patches. Each individual is simulated and has attributes including a sex, age, disease state, and membership in a local population. Mortality and fecundity are stage specific processes, and each individual is assigned to a specific stage based on its age. Mortality is density-independent, and fecundity (recruitment) responds to density only when density exceeds a threshold value. CWO differs from other population simulators by . incorporating disease, modeling both males and females, and by including functions that permit age and sex-specific patterns of removal. Because the model is individual-based, it is appropriate for representing the dynamics of small populations as well as large. In general, the number of patches or individuals that can be simulated are limited only by computer memory and processor speed. In reality, it is much easier to generate complex model dynamics than It is to understand the output, and most practical limitations will result from' an inability to analyze and understand output. It is essential to run many simple simulations to ensure that you have a complete understanding of model behavior before introducing spatial structure or other complexities. i Patch Attributes and Dynamics A patch is an area of the landscape that can support a local population. A Patch has the attributes of area, current quality (see below), maximum quality, mean environmental quality (from 0-1), maximum environmental quality (0-1), deviation from mean quality, and time since disturbance. Overall patch quality is quantified in a single variable ("forage") calculated from a user-defined function that transforms current forage to quality multiplier' (from 0-1; Fig. 2). This "forage" quality multiplier is used by the local population each year to determine the ability of the patch to support a population of animals (see below). Patch dynamics are simulated by predicting "forage" as a function of time since disturbance. The function that determines "forage" is calculated from a linear interpolation between forage and time values that are user-defined. Patch attributes can dramatically affect population processes through densitydependent effects on recruitment. If you wish to remove density-dependent effects from the model and control populations through harvest, set the patch size to a number much larger than mean population size and set animal density (in the *.spc file) to a number greater than one. Recruitment and Death The likelihood of recruitment changes with animal age and varies from year to year. Yearly variation is represented by an "environmental quality" multiplier that modifies the average recruitment rates. Each year, the model selects a random deviate from a Beta distribution that represents the "environmental quality" for the current year. The distribution of "environmental quality" is defined by the mean and standard deviation (entered by the user), and if the standard deviation is set to 0 there is no annual variation in the overall rate of recruitment. Each year the model loops through all females and compares the probability that a female of that age recruits an offspring into the population to a uniform random deviate. If the random deviate is less than the probability, recruitment occurs. Animals are culled from the population at stage-specific rates once per year. Mortality rates are independent of animal density and are subject only to stochasticity due to random events and to acute catastrophes. Stage-specific mortality rates are input by the user, and �102 for each individual, these rates are compared to a uniform random deviate to determine whether that particular individual dies during the year. Animal density dependent can effect population dynamics by reducing the probability of recruiting into the population. There is no effect of density on survivorship. Population density affects recruitment only when it exceeds a density threshold. Above the density threshold, the probability of recruitment declines linearly to 0 at a user-defined upper recruitment bound (Fig. ). In most situations this model is used to simulate, density plays no role in population dynamics because populations are controlled primarily by harvest. Under these conditions, its best to set the patch size sufficiently large to ensure that a density function is never invoked. Animal and Patch Aging The age of both animals and patches are incremented at the end of summer, after dispersal and prior to mating. At this stage, any patch manipulations are input and patch disturbance occurs. New patch statistics are generated for each patch, to be used in the following year's simulation. Data Output Model output is generated at the end of each year. The frequency of model output to the screen is specific by the user, while file output is generated every year. Output files formats are designed for analysis by statistical packages (e.g., SAS, SPSS) and are fixed format. The year is incremented immediately before model output and before patch dynamics are simulated. Thus year 0 output records are initial conditions. Running a Simulation To run a simulation, it is first necessary to build a set of input files that contain the variables and parameters used in the simulation (Table 1). The overall model run (which may consist of a large number separate Monte Carlo simulations) is controlled by a file with the extension .run. This file contains the names of input files that define life history characteristics, patch attributes, disease dynamics, and other run-time information described in Table 1 and below. This section describes the input files necessary to conduct a model run. The appendices to this manual include sample input file of each type. Input Files A simulation is controlled by variables and parameters specified in files that define the characteristics of the. animal species and landscape, and by the a file that contains information specific to a particular run (e.g., the number of years to simulate, names of input files, and what information is to be written to output files). In general, lines in an input files begin with an integer that tells the program what the rest of the line contains. This is followed by a one word descriptor (e.g., a series of characters and/or numbers that contain no spaces) of that line, followed by one or more parameters. It is essential that the one word descriptor has no spaces, and that there is at lease one space between the descriptor and the parameter(s). Input files are summarized in Table 1. Summary of parameters contained in each input file. 1. *.run Simulation run parameter file: Contains the names of files with parameters to be used in a specific set of runs, the number of runs to be simulated, the length of �103 each run, and which output files are to be written. It contains a variable for the frequency of output to the screen and a delay interval so that screen output can be more easily viewed. Screen output is slow; you can dramatically speed model execution by increasing the interval at which output is printed to the screen. This is the"master" file that and it deals with the highest level of model control. Code: Inputs::ReadRUNFile 2. .. *.pch Patch attributes include the size of each patch, maximum quality of the patch (0-1.0), a mean quality and the standard deviation of the mean patch quality, an initial forage (which remains unchanged in the 'absence of disturbance). Specified changes to patch quality and/or size in a given year can be used to simulate prescribed bums or other management protocols. Code: LandScape::ReadPCHFile 3.:''. *.spc Species characteristics: sex ratio at birth (m:f), number of age/stage classes for males and females, ages associated with each class, litter size, stage specific mortality, birth schedule, adult male and female body size, maximum density. The density threshold is estimated from the maximum density of animals, patch quality, and patch area. This file also specifies a function that relates patch quality (summarized into one variable, called "forage", for convenience) to a quality multiplier. Linear interpolation is used to estimate quality for values intermediate to those entered. Code: MammaIData::ReadSPCFile 4. *.pop Population parameters: This file contains the initial 'size and composition (age, sex) of populations. Code: AnimaIPopulation::ReadPOPFileO 5. *.hrv Harvest. File with parameters that determine the rules used to implement harvests. This includes the frequency of harvest (every year, every other year, etc.), threshold population sizes for determining the size of harvest, rules that govern the number of animals harvested, and the age and sex composition of the harvested population. 6. *.dis Disease:. File with parameters that determine disease dynamics. Includes parameters for transmission, and transition from incubating to infected and infected to dead. ' Model Functions and Details This section describes the structure of specific model functions, the implementation functions, and sources of data used to parameterize the model. of model Patch Dynamics Patches are areas that potentially support a local population, and they are characterized by their area (_area) and quality (_qualify, 0-1), constrained to a maximum for each patch (_maxQualify). Patch quality defines the ability of the patch to support animals relative to their maximum density. For example, if the maximum density for a species is 10 and the quality of a particular patch 0.1, that patch �104 Table 1. Input file extensions and the types of data they contain. file name type of data example file parameters *.pch patch data size, maximum quality, time since disturbance *.spc species attributes sex ratio at birth, stage-specific fecundity and mortality, maximum *.pop local population data number of local populations, age and sex of the members in each *.run simulation run control # of years of the simulation, names of input and output files, output *.hrv harvest information frequency, number harvested, and composition of harvest *.dis disease parameters transmission and transition proabilities I will support 1 animal per unit of area (area is generally expressed in krrf). The quality of each patch can be affected by environmental stochasticity, which is characterized by a normal random variate with a specified mean LmeanEnvirQualify, 0:-1) and standard deviation LsdEnvirQualify). The quality of a patch during any given year is calculated from it's current forage (see Landscape, _forage), constrained to a _maxForage level specific to each patch, and the quality associated with a given value of forage: _qualify = f(minLforage,_maxForage)) and the effects of environmental _qualify = minLmaxQualify, stochasticity (if any): _qualify * norma/_randomLmeanEnvirQualify, _sdEnvirQuality). The function that calculates quality as a function of _forage is defined by the user. and linear interpolation is used to evaluate points between those defined in the input file. If current forage is greater than that defined by the user, the _forage value at the maximum _gTSinceDisfurb is used; e.g .• no values are extrapolated beyond the data. These are combined to determine the population size at which reproduction is inhibited, _densifyThresSize, as a function of patch quality and the maximum density of animals that the highest-quality patch could support. _densifyThresSize _qualify. = _maxDensify *_area * 50 100 150 200 250 300 S50 ••00 Forage This design has been implemented to Fig. 2. User-defined function relating quality to forage. reflect the observation that population size may be restricted by forage availability, which changes over time. or other habitat features (perhaps escape terrain) that do not change. The quality of any single patch is therefore constrained by both a "food" -type variable Lforage) and other patch characteristics LmaxQualify). Code: LocaIPop:SetDensThresSizeO; LocalPop:DensThresSizeO �105 Species Characteristics Each species to be modeled must be described by life history characteristics including the number of different stages for males and females LnMaleStages, _nFemaleStages), inclusive ages of each stage LmStageMinAge, _'StageMinAge), maximum number of young LmaxYoung), the probability of having O-_maxYoung (_pYoungm, stage-specific birth and death probabilities (_pRecruitj, _pMMortality;, _pFMortality;. where I is stage). Density Dependence Density dependence influences the probability of dispersal and recruitment of new individuals into the population. The probability of recruiting one or more young is stage-specific and is determined by user-input values. This probability is then modified to reflect the effects of density and environmental stochasticity, To calculate density-dependent effects, it is necessary to specify. a species-specific maximum density ·LmaxDensity, #/unit-area), the density a species achieves in optimal habitat.. The units of patch size and ...;,maxDensity (e.g., ha) must be the same. The density at which recruitment falls to zero is defined by _upperPopBound, a decimal value greater than 1 used to determine the population size at which the probability of recruiting an individual falls to zero in the absence of environmental stochasticity (Figure XX). When the population size in patch I is greater then _pop Ttuessize; the age-specific probability of recruitment (_pRecruif) is modified to reflect the density-dependence: _pRecruitj .' ( = _pRecruit j popSizej l-----------------------------------~--(l+_upperPopBoundj) ) _densThresSizej Effects of Density on Recruitment The probability of recruitment of a new individual can be further influenced by environmental stochasticity, which is constrained to the range of _envirQualMin andt: environmental Quality = maxL envirQualMin, nonnalrandomLmeanEnvirQualify,_ _pRecruitj 11-----------------,. l sdEnvirQuality)) = _pRecruitj *environmental Quality Fig. 3. DensitY effects, where A = _densThresSize Taken together, these functions account for annual.stochastic and B = _densThresSize • _upperPopBound fluctuations in habitat quality that affect "carrying capacity", and yearly fluctuations in recruitment that result from abiotic or biotic sources (weather, predation, etc.). -. Code: AnimaIPopulation::BreedAndRecruitO, LocaIPop::SetDensThresSizeO; LocaIPop::DensThresSizeO Data: MammaIData._envirQuaIMean, _envirQuaISD, _probRecruitD Mortality Natural mortality occurs once per year (Figure 1), and is a stage-specific function. Stage and sexspecific mortality rates are. user-specified parameters in the *.spc file. For each individual, a uniform random deviate (0-1) is compared to the probability of death. Code: AnimaIPopulation::CUII Aging At the end of each time loop, the age of each individuals is incremented by 1 and an appropriate adjustment is made to _tSinceDisfurbance for each patch and local population. Code: AnimaIPopulation:: AgeO; LocaIPop::SAgeO; LandScape:: SetTimeO �106 Model Output Model output is written to standard ASCII text files that are created in the current directory, typically the directory containing the input and executable files. By default, any existing files of with the same name are erased upon opening, except for the error log file. Output file formats are described below. Any errors that occur during the run are printed to the screen and error file, and the run will be aborted when a severe error is detected. Some common errors in input parameter values are automatically corrected (e.g., an out-of-range parameter), in which case a message is written to the error log file informing the user of any corrections. All simulation result files are formatted to facilitate input to programs for further analysis in statistics, graphics, or spreadsheet programs. Variables in output files are either space or comma delimited. File output ceases when an entire population goes extinct, except for data written to the summary output file. Thus if a population goes extinct in year 23 of a 50 year run, files will generally contain 24 years of data (year 0 + 23 years). The summary file will be filled with 0 values for the full length of the run (year 0 + 50 years). All output file names begin with the same prefix, which can be specified in the *.run file or which will default to "cwd". All output file names have an ".asc" extension except the error file. Output Files Produced by Every Run _pavg.as<;: - 1 line for each year of run. Mean and standard deviation of the number of susceptibles, incubating, infectious, and total population size. _ sum.asc - 1 line for. each year of each run. Number of susceptibles, incubating, infectious, and harvested. CWD _err. txt (if an error is detected). Error information Information about any errors detected during a run are printed to this file. Errors are assigned to categories Warning, Error, and Fatal Error. Warning messages signal an error than can be automatically corrected within the program or conditions that are unlikely to affect program results. Errors are more severe, and may be corrections to parameters that are frequently entered incorrectly or a condition within the program that may be irregular. Depending on the compilation, the program may abort upon detecting an error (this is a compiler switch, and cannot be changed by the user). The program always aborts when a Fatal Error is detected. Fatal errors are usually the result of an incorrect input file format, invalid parameter, or insufficient hard disk space or memory. Table 2. Output file names and contents. Files in italics are always produced, others are produced only when specified by a switching variable in the *.run file. Files are comma-delimited. File Columns Information cwd err.txt text Produced or appended to if an error is reported. _age I-many run, year, patch number, males by age (O-Age max), females by age (0Age max). Age max is specified in the *.spc file. disfxhs 1-5 run, year, patch number, period, disease-caused deaths by sex and age class -hrv l-many run, year, patch, population size category, males by age (O-MaxAge), females by age (O-MaxAge) _pavg l-many year, mean and standard deviation of susceptible, incubating, infectious, and total number of animals _prevAge l-many run, year, patch number, susceptible males by age category, susceptible females by age category, incubating by sex (m.f) and age category, infectious by sex (m, f) and age category sum 1-5 run, year, patch number, number of susceptible, incubating, infectious, and harvested animals ts 1-9 run, year, patch number, test&slaughter index (O=off, I=on), population size, prevalence before t&s, prevalence after t&s, males removed, females removed - (produced only when t&s is on) �107 Sample input files and formats General File Format Used by CWD The general format of CWD input files is to begin each line with an integer number that indicates the content of one or more lines of the input file. "99" is always used to indicate a comment and the program disregards all information on a line following the "99". In general, the integer is followed by one word (i.e., a series of characters and/or numbers that contains no spaces) describing the following parameter(s) or variable(s) .. The descriptive word can contain underscores, upper and lower case letters, numbers, or any other characters except blank spaces or tabs. For situations where there are a: variable number of input lines (e.g., the age structure of an initial population), lines may begin with a' data value not preceded by an integer indicating the line content. Sections of input than can include a variable number of lines always end with a delimiter, which is usually a number less than zero. It is essential that these file formats be followed, and users are very strongly encouraged to modify existing input files rather than create a new file. Virtually all errors in model runs can be traced to a mistake in an input file. Many (but not all) common mistakes in input files will be detected by the program and reported to the error file (CWD _err.txt). Model Run Data (*.run) "The *.run file is the first file read by CWD and it determines program operations at the highest levelwhich modules are used, the names of other input files, which files output files are to be written. _A value of 1 or greater turns on optional functions or optional outputs and the value "0" turns these ',' function off. If Random = 0, the same random number seed is used at the beginning of all runs. nlls option is used for debugging and Random should generally be set to "1". A delay can be used if you wish to slow the program in order to more easily view screen output. 99 cwd.run 99 -2 PCHFileName: cwd_inpt.pch 3 SPCFile: cwd_inpt.spc 4 POPFile: cwdjnpt.pop 5 Harvest: 1 cwd_inpt.hrv 6 Disease: 1 cwd_inpt.dis 9 Test&SI: 1 cwdjnpt.tst 99 -11 RunYears: 100 12 Runs: 250 14 ScreenOutputFreq: 50 99 - if random = 0 use same seed for all runs; if random = 1 use a different seed. 15 Random: 1 99 _ all output files will have the following "prefixed" to their names 99 _ E.g., the file beta_2_11_14_age.asc would be produced with this line. In addition, 99 _ any sensitivity variables are added to the file name (see below). 99 -16 OutputFilePrefix: beta_2_11_14 99 -21 PopAgeOut: 1 _age 22 PatchesOut: o _pch �108 23 DoDisOutput: 1 dis 29 doHarvestOutput: 1 harv 9999 -- line 31: on/off prevalence level to begin output, prevalence level to end output 99 so the following results is output when the prevalence level is >= 0.03 and <= 0.50. 991 .01 .50 31 doPrevAgeOutput_startPrev_endPrev: 1 disDths 32 doDisDeathsByPeriodOutput 9999 101 = sensitivity analysis. 99 Variables are repeated for each variable, so the number of 99 values on the line MUST be divisible by 4. 99 var_number smallest_val largest_val step 99 -101 sens vars 123.4.8 .05 Lines 21 -31: Any integer of 1 or greater tufns output on. 31: These files can be very large, thus they are produced only when prevalence in a patch is equal to or within the range of the starting and ending value. Patch Parameters (*.pclz) 99 filename: cwd_21 May.pch 10000 3 PatchArea: 4 MaxQuality: 1 5 MeanEnvQual: 1 6 sdEnvQual: .0 7 maxForage: 100 100 8 initialForage: PatchArea: Area of patch used by the local population. Set to a large value unless you want density.•.dependent population regulation to operate. MaxQuality: This value reflects the maximum density that a particular patch can support, relative to the species-specific maximum density. The species-specific maximum density is specified in the *.spc input file. MeanEnvQual: Mean quality of the patch (must b~ less than MaxQuality). sdEnvQuality: Standard deviation of the mean quality. If set to 0, there will be no annual variation in the value of the density threshold due to differences in patch quality. maxForage: Maximum forage value that will ever be obtained on a specific patch. initialForage: Forage value at start of the run �109 Harvest Populations are normally controlled by harvest of animals. The rate of harvest can vary depending on population size, which can be categorized as small, normal, or large. Input variables are used to determine threshold sizes between these categories. For each population size category, input variables specify age-specific rates of harvest. There are no limits to the number of age categories that can be specified, but if ages within a category overlap, the last value in the input file will be used. If you want harvest to be independent ot age, specify the same rate for all ages. i. Annual rates of harvest are determined each year by selecting the appropriate mean harvest rate from the input file (lines 11-17) and then modifying this rate to account for annual variation. Mean rates for all age groups are multiplied by a random deviate with mean 1 and CV from line 10. Thus annual variation for all age groups are correlated. 99 CWD for mule deer 99 Rule 1 has proportion harvested for low, med, and high pop numbers, by sex and age 1 harv_rule 1 10 CV_on_percent_harv .2 99 _ FEMALES min_age max_age portion_harvested. Ages are inclusive 11 F_Iow 0 20 .0 12 F_med 0 0 .01 12 F_med 1 2 .02 12 F_med 3 20 .07 13 F_hgh 0 0 .03 13 F_hgh 1 2 .05 13 F_hgh 3 20 .10 99 - MALES 15 M low 0 20 .0 16 M med 0 0 .02 16 M med 1 1 .06 16 M_med 2 20 .20 17 M_hgh 0 0 _..03 17 M_hgh 1 1 .09 17 M_hgh 2 20 .27 99 -- repeat lines 21 & 22 for each local population. 21 localPop 1 22 threshold numbers 500 1000 . �110 Test and Slaughter Test and slaughter is implemented by specifying the efficacy of the program for males and females. Efficacy is the product of the portion of the population tested and the likelihood that a positive animal will be detected. Variables in the input file determine whether the program is implemented from the beginning of the program, or after other criteria are met. A test and slaughter program can begin at a specific year of a model run, or after the population has achieved a specified prevalence rate. Efficacy is the product of the proportion of the population tested and ability to detect the disease. The variable on line 3 (periods to detect) is the number of periods of incubation that must elapse before a disease can be detected in an exposed individual. If there were 2 periods per year and this parameter were set to 1, an exposed individual would not be detected until they had incubated for a full period, equivalent to about 6 months. If you assume that rate of exposure is constant, one could interpret this to suggest that the average incubation time before detection would then be about 9 months, or one full period and (on average) half of another period. However, since test and slaughter occurs only once per year (after harvest and the second disease period), individuals infected during disease period 2 (falilwinter) will not be detected until the following year unless periods of detection is set to O. Infectious individuals are always detected. 99 cwd_test_ and_ slaughter.tst 99 3 periods_of_incubation_before_detection 4 local_pop_tested 1 5 Efficacy .5 6 Start_year 10 7 Start_prevalence .05 3 �III Sensitivity Analysis The model provides the user with very broad options to conduct sensitivity analysis. Todo so, specify on line 1.01in the run file the variable number (from the table below), the initial value, the maximum value, and the step (amount for the initial value to be incremented for each set of runs). Some variables (will in the future) also require that a local population be specified. When a sensitivity analysis is conducted, the names of all output files reflect this by adding the number of the variable and the step (starting with 0) of variable value to the file name. These values are added after the file name prefix specified on line 16 in the run file and before the normal file suffix. For example: lirie 16 in run file: 16 file_name: 24jan_larimer line 101 in run file: 101 sens_vars 121 .01 .10 .01 Which means: do runs with maternal transmission rates from 0.01 to .1 by .01. Ten sets of runswill be simulated, with 10 output files of each type (e.g., _pavg, _sum, etc.). The summary output files, for example, would be: 24jan_larimer_:_121_0_sum.asc 24jan_larimer_121_1_sum.asc 24jan_larimer_121_9_sum.asc If additional sensitivity variables are added, these are added to the file name in the general format: prefix_var#_step_var#_step ... suffix. There are no internal limits to the number of sensitivity variables, but since all levels are fully crossed, you can easily generate hundreds or even thousands of output files. The difficulties in analyzing such results are substantial. �112 Sensitivit~variables available for anal~sis. MammalData MammalData MammalData MammalData MammalData MammalData MammalData Number Variable name value or Mult # values 101 female survival multiplier 3 102 male survival multiplier 3 103 recruitment multiplier 3 104 sex ratio at birth value 3 105 mean environmental value 3 quality 106 std environmental quality value 3 121 maternal transmission value 3 122 prob. infected recruits value 3 123 period 1 beta value 3 124 period 2 beta value 3 125 period 3 beta value 3 beta for all periods 126 value 3 AnimalPop 21 periods for test detection value 3 LocalPopulation 51 LocalPopulation 61 LocalPoQulation 62 test efficacy mean - both value harvest lower threshold value harvest uQQerthreshold value 3 3 3 Class Affected MammalData MammalData MammalData MammalData MammalData �113 AppendixB Chronic Wasting Disease in Deer and Elk: DRAFT Colorado Statewide Monitoring and Experimental Management Plan M. W. Miller and R.H. Kahn Terrestrial Wildlife Resources, Colorado Division of Wildlife . Background Chronic wasting disease (CWD) is a transmissible spongiform encephalopathy (TSE) of native deer (Odocoileus spp.) and elk (Cervus elaphus nelsoni) characterized by behavioral changes and progressive loss of body condition that invariably lead to the death of affected animals (Williams and Young 1992). Neither the causative agent nor its mode of transmission have been identified. There are no tests currently available for diagnosing CWD in live animals, and extant postmortem tests require microscopic examination of brain tissue. There are no known treatments for CWD. Previous attempts to eradicate CWD from research facilities failed on at least 2 occasions (Williams and Young 1992; Miller et al. 1998). Although similar in some respects to other TSEs that affect domestic sheep (scrapie) and cattle (bovine spongiform encephalopathy; BSE), existing data indicate CWD cannot be naturally transmitted to domestic livestock, and that scrapie and BSE cannot be transmitted to native cervids. Moreover, available data indicate that CWD does not present a threat to human health. "Chronic wasting disease" was first recognized by biologists in the 1960's as a disease syndrome of captive deer held in wildlife research facilities in Ft. Collins, CO. This disease was subsequently recognized in captive deer, and later in captive elk, in wildlife research facilities near Ft. Collins, Kremmling, and Meeker, CO and Wheatland, WY (Williams and Young 1980, 1982), as well as at in least two zoological collections (Williams and Young, 1992). More recently, CWD has been diagnosed in captive elk residing in one game ranch in Saskatchewan (Spinato, pers. comm.), two ranches in South Dakota (Holland, 1997), one ranch in Nebraska (W. Cunningham, pers. comm.), and one ranch in Oklahoma (L. Detwiler, pers. comm.). Since 1981, over 100 cases of clinical and preclinical CWD also have been diagnosed in free-ranging mule deer (0. hemionus), white-tailed deer (0. virginianus), and elk from northeastern Colorado; most of these diagnoses have been made since 1990 (Spraker et al. 1997; Miller, 1997). At present, the known world-wide distribution of CWD in wild cervids appears to be limited to northeastern Colorado and southeastern Wyoming (Miller, 1997; Williams et al., 1998). Although CWD was first diagnosed in captive cervids, the original source of CWD in either captive cervids or free-ranging cervids is unknown; whether CWD in captive cervids really preceded CWD in wild cervids, or vice versa, is equally uncertain (Spraker et al. 1997). In Colorado, free-ranging CWD cases in deer and elk have originated from throughout the northeastern portion of'the state. Game Management Units (GMUs) yielding infected deer or elk include 7,8,9, 19, 191,20,29,91,93,94,95,951, and 96. Although cases have come from 13 different GMUs, two Data Analysis Units (D4, DI0) have yielded over 85% of the documented cases. Based on targeted surveillance data and select harvest surveys, wild deer or elk populations in other parts of Colorado are probably not infected with CWD (Miller, 1997; 1998). Both targeted and harvest surveys indicate GMUs 9, 191, 19, and 20 are the main foci ofCWD in Colorado; estimated prevalence in all four GMUs is ~3% (Miller, 1997; 1998; M.W. Miller, unpubl. data). Data from harvest surveys indicate that CWD is relatively rare in other parts of northeastern Colorado, probably affecting ~2% of the deer in GMUs 7, 8, 29, 91, 93, 94, and 96 (Fig. 1). Similarly, harvest surveys indicate <1 % of the elk in DAUs E4 and E9 are infected with CWD (Miller, 1997; 1998; M. W. Miller, unpubl. data). For deer populations in the four most heavily infected eastern Larimer County GMUs, no real trend in prevalence (increasing or decreasing) can be discerned from data available to date. In the absence of historical (20-30 years ago) prevalence data or reliable estimates of transmission rates, it is �114 also unclear whether overall incidence ofCWD in northcentral Colorado DAUs is stable or increasing, or whether short-term observations can accurately forecast long-term trends. Based on prevalence estimates and preliminary results of simulation modeling to predict dynamics and impacts of CWD on affected deer populations, it appears CWD was introduced into northern Larimer County over 30 yrs ago (M.W. Miller and C.W. McCarty, unpubl. data). Beyond Larimer County, the South Platte River corridor may be the single most predictable avenue for spread of CWD. Kufeld and Bowden (1995) reported that a proportion of South Platte River bottom deer were highly mobile; observed deer movements included some to and from eastern Larimer County along the Cache la Poudre and South Platte Rivers. These movement patterns provide a plausible mechanism for the apparent emergence of CWD in South Platte River GMUs detected recently through targeted surveillance and harvest surveys. Other likely routes for deer emigrating from Larimer County have not been identified, and may be less predictable. Altitudinal movements of some elk subpopulations in the southern part of Larimer County are also somewhat predictable (Bear, 1982). Although there is potential for elk to cross the Continental Divide through Rocky Mountain National Park, the relatively low prevalence of CWD among elk diminishes the likelihood of its spread via their movements. The significance of CWD and its impacts on native deer or elk populations have not been determined. Preliminary results of simulation modeling suggest that, sustained at 5% prevalence as observed in the four most heavily infected GMUs, CWD could impact wild deer herds and lead to population declines (M.W. Miller and C.W. McCarty, unpubl. data). Data on CWD prevalence in . female deer are particularly critical to predicting potential impacts of disease on long-term population performance and assessing efficacy of management interventions. Preliminary comparisons made in conjunction with 1997 and 1998 surveys showed no differences in CWD prevalence between male and . female mule deer harvested in Larimer County GMUs (M.W. Miller, 1998; unpubl. data); these data were comparable to results of simulation models where male and female deer were assumed to be equally susceptible to CWD. In the absence of data to the contrary, and considering the difficulties inherent in eliminating CWD from captive or wild cervid populations once established, it seems most prudent to assume CWD could adversely affect native deer or elk populations and manage to reduce its occurrence and prevent its further spread in Colorado. Unfortunately, there is considerable uncertainty in how to manage CWD in free-ranging wildlife, or whether such an endeavor could even be successful. Because a more complete understanding of CWD is fundamental to developing comprehensive management programs, the Colorado Division of Wildlife (CD OW) needs to further understanding about CWD and its management through surveillance and experimental management. Data from completed, ongoing, and proposed research, both basic and applied, will serve as the foundation of an adaptive resource management plan for CWD in deer and elk. This plan will provide a mechanism for incorporating new knowledge gained through surveys, modeling, research studies, and management experiments into a continuously evolving management program for CWD. STATEWIDE MONITORING PROGRAM Goals: 1. Provide reliable estimates of CWD distribution and prevalence to serve as a basis for management decisions and public information. 2. Provide information to improve understanding of CWD epidemiology. 3. Continue developing efficient and reliable techniques for detecting and monitoring CWD in freeranging populations. �115 Strategy: Reliable estimates of CWD prevalence in wild deer and elk populations are needed to guide policy decisions and monitor efficacy of management efforts. We will continue to monitor deer and elk populations throughout Colorado for occurrence of chronic wasting disease using a combination of extensive and intensive approaches. These have been organized and conducted as follows: Targeted (= clinical disease) surveillance: Ongoing statewide "targeted" surveillance (i.e., submissions of "suspect" deer & elk) will be used to determine & monitor changes in CWD distribution. Based on·. experiences in Colorado and elsewhere, this appears to be the most sensitive approach for determining CWD distribution & detecting range extensions (Miller, 1997). A program for acquiring, examining, reporting on, and summarizing "suspect" CWD cases occurring throughout Colorado has been in place since 1990 (Miller, 1997). This program has served to increase ongoing surveillance efforts by field personnel statewide by encouraging submission of carcasses from deer or elk showing clinical signs resembling CWD. A formalized process for submitting cases has been developed; this process includes criteria for acceptable submissions, submission forms, handling instructions, and a system for networking submissions within CDOW and among the 3 Colorado State Veterinary Diagnostic Laboratories and the Wyoming State Veterinary Laboratory. Under this program, deer and elk showing clinical signs consistent with those seen in chronic wasting disease are collected by field personnel statewide; a "suspect case" profile has been defined (Table 1). Brain tissues are subsequently examined for evidence ofspongiform encephalopathy. Where possible, whole carcasses will be submitted for complete necropsy; at minimum, heads from suspect animals will be submitted for examination and sampling. Ancillary diagnostics, including histopathology (Williams and Y oung,1993) and immunohistochemistry of brain tissue (Spraker et a1., 1997), will be performed on all suspect cases; other ancillary tests may also be used to confirmor support diagnoses. Preliminary examination and/or test results and final reports will be faxed to CDOWs Wildlife Research Center. Pertinent data from preliminary and final reports will be entered into a permanent database, and copies of reports will be filed as well as sent to appropriate field personnel. Results from targeted surveillance will be used to identify new potential foci of CWD infection for further evaluation via harvest or road-kill surveys. Harvest and road-kill surveys: Surveys of harvested or, in some areas, road-killed deer and elk will be conducted in endemic and high-risk DAUs/GMUs to estimate and monitor CWD prevalence. We will continue conducting annual surveys in DAUs or GMUs where CWD prevalence is >2% ("endemic foci") to obtain reliable-estimates of prevalence to use as a basis for monitoring natural trends, as well as responses to management interventions. Annual surveys will be designed to provide an estimated prevalence within ±2% at a 95% level of confidence for affected populations where true prevalence is 1%. In addition to annual surveys of endemic foci, we will also continue conducting a series of surveys for CWD in other select deer and elk populations throughout Colorado over the next 5 yrs, with a goal of determining statewide distribution of CWD. Our surveillance strategy is as follows: Areas newly identified as infected via targeted surveillance (e.g., GMU 29, South Platte River corridor GMUs) will be surveyed sufficiently to obtain a reliable prevalence estimate. In areas where estimated prevalence is <1 %, CWD will be regarded as a sporadic disease and populations will be monitored at ~5-yr intervals to detect increases in prevalence. Areas where estimated prevalence is >2% will be regarded as new endemic foci, and will be monitored annually to track changes in prevalence as described above. We also plan to continue conducting surveys of at least 5 other select deer populations where CWD has not been reported �116 previously. Target areas will include both high-risk (e.g., Middle Park, Piceance Basin) and low risk (e.g., San Luis Valley, Gunnison, southern Front Range) populations. Surveys will be designed such that the probability of failure to detect at least 1 case of CWD in an apparently-unaffected deer or elk population will be ,::::0.01even if herd prevalence is 1%. Finally, we will attempt to gather a sufficient number of random samples from DAUs statewide (n-l,OOO) to determine whether or not CWD should be generally regarded as endemic in Colorado's deer populations. A tentative schedule for the foregoing surveillance is outlined in Table 2. Harvest surveys will be conducted using techniques developed in Colorado since 1991. Because regulatory requirements for head submissions increase sample sizes >4-fold, submissions will be mandatory in most GMUs where surveys are being conducted. Limited licenses may also be used in select GMUs to aid in increasing compliance. Unstaffed barrel sites appear to be the most efficient method for sample collection, and will continue to be used as the primary method for conducting harvest surveys. Brainstem (obex) from ~ I-yr-old deer and elk will be.collected for CWD testing, and immunohistochemistry (IHC) using USDA monoclonal antibody F89/160.1.5 (O'Rourke et al., 1998) will be used as the primary screening tool. Reported prevalence estimates will continue to be based on numbers ofIHC:·positive cases; because IHC appears more sensitive than histopathology in detecting preclinical CWD cases, IHC-based prevalence estimates should provide a relatively unbiased measure of disease trends over time and allow more direct comparison between field data and simulation model predictions. EXPERIMENTAL MANAGEMENT There is no precedent for attempting to manage a TSE in free-ranging wildlife. Moreover, the biological need for such management is unclear at present. Programs for managing or eliminating TSEs of domestic livestock have proven only marginally successful, and the lack of obvious success in eradicating scrapie or BSE from endemic countries, compounded by epidemiological differences between CWD and other TSEs, make such programs rather poor models for prospective CWD management. Several fundainental features of CWD epidemiology are understood poorly, if at all; in particular, the influences of host (e.g., population density, intraspecific vs. interspecific transmission, genetics) and environmental factors (e.g., contamination, reservoirs, weather) on CWD dynamics remain undescribed. In light of the uncertainties associated with its epidemiology and ecology, we believe CWD management should be approached as an experiment, thereby allowing us to learn as we . take actions intended to reduce prevalence and limit distribution. The CDOW's initial approaches for attempting to control CWD in free-ranging deer and elk were outlined in an action plan for CWD drafted in 1995 (Miller et al., 1995; Appendix A). In addition to calling for a comprehensive public information campaign, this plan identified tactics for limiting distribution and reducing prevalence of CWD based on contemporary knowledge of the problem. Prompt removal of affected animals and enforcement of feeding prohibitions were recommended to help reduce CWD prevalence and transmission. Policies and regulations were also recommended, and subsequently instituted, to prevent wider distribution of CWD via human activities (e.g., trans locations, rehabilitation, game ranching). The plan also identified a need for more extensive surveillance to gather data on CWD distribution and prevalence. As more data on CWD distribution and prevalence have been gathered over the last 2.5 yrs, it has become apparent that CWD is more widely distributed and more prevalent in northeastern Colorado (and elsewhere) than initially believed. It follows that policy, goals, and strategies for CWD management in Colorado should be reevaluated in the context of new data, as well as changes in social and political attitudes toward animal TSEs. �117 Management policy: At present, the CDOW has no stated policy related to the management of chronic wasting disease in wild deer andelk. The lack ofa clear policy statement has led to internal confusion and disagreement over how to resolve conflicts with other policies and Long Range Plan objectives (e.g., increase recreational opportunities for deer and elk hunters). Recommendation: We recommend adoption of the following policy to guide decisions on deer and elk population management. in DAUs where CWD is endemic: Pl. The Colorado Division of Wildlife is committed to minimizing the threat that chronic wasting disease (CWD) poses to Colorado's native deer and elk resources. In DAUs where CWD is endemic, goals of reducing the occurrence and the further spread of CWD will serve as the primary basis for setting management objectives. Prospective management goals: We have. identified four overarching goals that could serve as a basis for decisions related to managing CWD: 1. "natural regulation", 2. containment, 3. reduction, 4. eradication. These goals represent a continuum of possible levels of intervention, ranging from benign. rs neglect to aggressive action. There are clear advantages and disadvantages associated with each of .. these prospective management goals. Although each progressive level offers more complete control of the problem, better control is accompanied by ever-increasing technical, fiscal, and political challenges, as outlined below: l."Naturai regulation": Directing no specific management intervention toward CWD beyond culling of reported clinical cases represents the status quo approach to both disease arid deer/elk population management. It could be argued that the ongoing activities described above constitute an attempt to manage CWD. Although the impacts of this approach on CWD prevalence in endemic foci remain undetermined, prevalence in endemic foci appears to have remained stable over at least the last J yrs. This approach offers the fewest technical challenges, may be regarded by some as "safest'unlight of the myriad of uncertainties in CWD epidemiology, and in the shortterm would be most sparing oflocal resources. Unfortunately, simulation models forecast that problems with CWD will likely get worse in affected populations if left largely unmanaged, the disease could become more widely distributed, and there are tangible potential impacts on wildlife resources, recreational opportunities, and revenues. In addition, perceived "inaction" may be unacceptable to some sportsmen's groups, agricultural interests, and perhaps the general public, and could lead to the loss of management authority by CDOW. Finally, opting for "natural regulation" seems biologically irresponsible. 2. Containment: Managing primarily to prevent spread of CWD from endemic foci would require additional knowledge of the mechanisms and probable routes of CWD dissemination, but many of the important strategic components needed to support this strategy are already in place. The likelihood of trans locating infected animals from endemic areas to other areas in Colorado or �118 elsewhere was significantly reduced by regulations adopted in 1996 (ch 14 ref). Between existing data from previous movement studies (e.g., Kufeld et al., 1989; Kufeld and Bowden, 1995) and data from studies initiated more recently, major natural emigration corridors from eastern Larimer County could probably be identified and subsequently might be interrupted via intensive culling operations. This approach, if successful, would limit the magnitude of the CWD problem while largely sparing local resources. This alternative may be more acceptable to sportsmen's groups and the general public than to agricultural interests, but is probably biologically justifiable. However, such an approach would require an essentially infinite long-term commitment and yet will not eliminate CWD entirely; prevalence could conceivably increase in endemic areas. There would likely be some local impacts on resources and recreational opportunities, perhaps accompanied by local public resistance. 3. Reduction: It may be possible to manage affected deer or elk populations to reduce CWD prevalence in endemic foci. A goal of prevalence reduction clearly demonstrates intent to limit the magnitude of the problem. It seems likely that this. approach would also provide for containment of CWD. Attempting to reduce CWD prevalence might improve "consumer confidence" among hunters in endemic GMUs. This alternative may be the most widely acceptable among sportsmen's groups, agricultural interests, and the general public in terms of responsible disease and resource management. It is probably biologically justifiable in view of projected long-term effects ofCWD on infected populations and the potential for CWD to spread from endemic foci if left unmanaged. As with the foregoing goal of disease containment, prevalence reduction will require a long-term commitment and may not eliminate CWD from endemic areas; depending on specific tactics employed, prevalence could even increase despite management efforts. This approach would likely cause more severe impacts on local resources and recreational opportunities, and consequently would be more likely to foster local public resistance to proposed management actions. 4. Eradication: Eradication is probably the most desirable goal of any attempt to manage CWD. Whether it would be feasible to completely eliminate CWD from Colorado remains highly questionable. Eradicating CWD would be the best means of restoring for "consumer confidence" among northeastern Colorado deer and elk hunters, and would presumably nullify concerns of traditional and alternative livestock interests about the potential for transmission to privately-owned animals. Perhaps most importantly, eliminating CWD would be the most effective way to preempt threats of its eventual spread to other native deer and elk populations in Colorado and elsewhere. Although desirable in concept, CWD eradication is probably infeasible given the complexities and uncertainties of its epidemiology. Eradication of CWD would require long-term commitment, and probably would exaet catastrophic impacts on resources and recreational opportunities in affected areas; broader ecological impacts (e.g., on predator-prey balances) could also be severe. Public resistance is likely to emerge, at least locally, as details of such a management program emerge. Finally, it is questionable whether the potential ecological costs of CWD eradication are biologically justifiable in light of the uncertainty about long-term impacts of the disease itself. Recommendations: Based on current understanding of CWD epidemiology, prevalence, and distribution in free-ranging deer in northeastern Colorado, we believe eradication is an extreme and unjustifiable management goal at this time. In light of the magnitude of prevalence and distribution CWD has reached in Larimer County deer populations under historical management regimes, however, continuing to rely on current management approaches seems equally unjustifiable. Consequently, we recommend an intermediate approach with two goals for managing CWD in deer and elk: �119 Gl. limit distribution ofCWD occurs. to no more than the 5 deer DAUs and 2 elk DAUs where it already G2. reduce average CWD prevalence among deer and elk to <1 % in each endemic DAU and <2% in each endemic GMU; and . '~ We believe achieving the foregoing management goals will serve to protect Colorado's deer and elk resources, restore confidence in the quality of deer and elk harvested in northeastern Colorado, and ameliorate political pressures from agricultural constituencies concerned about potential for transmission of CWD to domestic livestock or privately owned wildlife. Prospective management strategies: Effective strategies for managing CWD or any other TSE in free-ranging wildlife have never been identified or evaluated. Many conventional disease management options are precluded from consideration because vaccines, therapeutics, and live-animal tests for CWD are presently unavailable (Table 3). Contrary to experiences with domestic sheep (Goldmann et al., 1994), available data indicate relatively uniform genetic susceptibility to CWD among deer (K. I. O'Rourke and M. W. Miller, unpulb. data); consequently, genetic selection would probably have no impact on prevalence even if tools forits implementation were available. Similarly, controlling reproduction alone probably would be ineffective even if practical tools were available because maternal transmission alone is unlikely to be sustaining CWD prevalence at levels currently observed in some affected deer .populations. All remaining options we can identify to reduce CWD prevalence (Table 3) involve some form of aggressive population control directed at diminishing or preempting-disease transmission -- themagnitude and duration of such control will be driven by management objectives and population responses. Based on current understanding ofCWD epidemiology (Miller 1997, Miller et al., 1998, M.W. Miller, unpubl. data), we believe that strategies capable of lowering CWD transmission rates by reducing numbers of infected animals and point sources of environmental contamination (e.g., feeders) should be most effective in lowering prevalence; although the precise mechanisms of transmission have not been identified, those details will probably have little influence on the resolution of CWD management at the population level. General models predict that random culling should be less effective than selective culling in reducing disease prevalence (Barlow, 1996; McCarty and Miller, 1998). However, field data indicate that management strategies based on culling clinical suspects probably will miss a large proportion of the infected (and potentially infectious) individuals in a population; moreover, there is no practical way to apply "test and slaughter" regimes without an antemortem diagnostic-test for preclinical CWD. Accelerated culling of early clinical cases, either by humans or via predators, may help reduce CWD transmission and prevalence. Alternatively, if the probability of CWD transmission is density-dependent, then reducing deer or elk densities in endemic areas should lead to reduced CWD prevalence; to date, the hypothesis of density-dependent disease transmission remains untested. Obvious strategies for limiting CWD distribution remain equally elusive. The mechanisms driving the spread of CWD among deer and elk populations are even less apparent than those driving disease dynamics. It is possible that CWD distribution to date has been influenced by regular and/or random movements of deer and, less commonly, elk. Previous studies (Kufeld et al., 1989; Kufeld and Bowden, 1995; S. Steinert, pers. comm.; others?) have shown that proportions of both foothills and river bottom deer populations in northeastern Colorado are either migratory or mobile; the movement patterns documented in these studies provide at least a partial explanation for observed CWD distribution. It is also possible that CWD influences its own distribution by changing behaviors, range ..... �120 fidelity, and movement pattens of affected animals; observations of repetitive pacing in captive deer and elk affected by CWD could be manifested as extensive wandering movements of free-ranging individuals. Whether or not movement patterns or dispersal rates are density-dependent remains unclear; either way, reducing population size should also reduce the probability of affected individuals canying CWD to new areas. (However, we do recognize the possibility that at some lower threshold reduced density may actually promote deer and elk movements, particularly during the breeding season). Recommendations: Based on current uncertainty about how to effectively disrupt the processes that influence CWD epidemiology, prevalence, and distribution in free-ranging deer in northeastern Colorado, we recommend initiating controlled management experiments directed toward testing one (or both) of the following hypotheses: HI. CWD transmission and prevalence infoothills and river bottom deer populations can be diminished by a) selective culling; b) increasing natural mortality via mountain lion predation; OR c) reducing deer density in affected populations. H2. CWD distribution in foothills and river bottom habitats can be diminished by a) selective culling; . b) increasing natural mortality via mountain lion predation; OR c) reducing deer density in affected populations. We further recommend that the experiment(s) be designed and conducted in a way that facilitates achievement of two short-term management objectives: MI. to stem dispersal ofCWD via deer movements along the South Platte River corridor (GMUs 94, 951,96,91, and 92). M2. to reduce CWD prevalence among deer inhabiting the four most severely affected GMUs (9, 191, 19,20). Because CWD prevalence is low «1%) in elk, we recommend that populations in DAUs E4 and E9 should be maintained at or below current densities and monitored via harvest surveys at least every 5 yrs to detect potential increases in prevalence that.might warrant more aggressive management action. :. Design of Management Experiment We recommend using an adaptive resource management approach (Walters, 198_) to test candidate strategies for reducing CWD prevalence and distribution. Based on initial evaluations of these candidate strategies (Table 4), more detailed management experiments will be developed; as part of that process, decisions will be made on which, if any, approaches to pursue and where CWD management will be attempted. Tactics for accomplishing management approaches (e.g., harvest, selective culling, ground/aerial gunning, changes in predator harvest, etc.) will follow from internal consensus on which tactics are viable under the social and political constraints imposed by the area(s) selected for study. Prospective management tactics: Tactics supporting selected management strategies: �121 - harvest? (role of public hunting?) - shift emphasis from recreational opportunity to population management - modifications of GMU/ "management area" boundaries to focus mgt & distribute pressure - season structurellength & bag limits- reg changes - schedule - culling? - ground vs aerial vs capture & kill options - public vs agency c; - private land issues - fertility control? (probably not for reductions, but maybe to maintain) - agent, delivery, cost? - reduce predator quotas? (close seasons?) - predator introductions? - poisoning? - other? , '; . <" Issues surrounding CWD management - Should public hunting continue in endemic areas? - What are public desires/expectations related to CWD management? - Can we impose intensive population management on private lands? - What are reasonable expectations for funding and long-term agency/political/public - Others? commitment? Outstanding tasks (partial list) Human dimensions -- Public service -Funding (PBE)-- For both internal and external publics, seek consensus on: what direction to take, which strategies are most viablelleast objectionable, etc., impacts on recreational opportunity, trade-offs Seek/gain cooperation, access, etc. (follows from lID issues above) Estimate support required for evaluation of management activities, as well as fiscal impacts of candidate management approaches LITERATURE CITED Goldmann, W., N. Hunter, G. Smith, 1. Foster, and 1. Hope. 1994. PrP genotype and agent effects in scrapie: change in.allelic interaction with different isolates of agent in sheep, a natural host of scrapie. 1. Gen. Virol. 75:989-995. Miller, M. W. 1997. Monitoring and managing chronic wasting disease in Colorado. Pages _-_ in Wildlife Research Report, Mammals Research, Federal Aid Projects, Job Progress Report, Project W-153-R-I0, Work Plan 2, Job 17. Colorado Division of Wildlife; Fort Collins, Colorado, USA. (in press) Miller, M. W. 1998. Monitoring and managing chronic wasting disease in deer. Pages _-_ in Wildlife Research Report, Mammals Research, Federal Aid Projects, Job Progress Report, Project W-153-R-l1, Work Package 3001, Task 3. Colorado Division of Wildlife, Fort Collins, Colorado, USA. (in press) Miller, M. W. 1998. Monitoring and managing chronic wasting disease in elk. Pages _-_ in Wildlife Research Report, Mammals Research, Federal Aid Projects, Job Progress Report, Project W-153-R-ll, Work Package 3002, Task 3. Colorado Division of Wildlife, Fort Collins, Colorado, USA. (in press) �122 Miller, M. W., M. A. Wild, and E. S. Williams. 1998. Epidemiology of chronic wasting disease in captive Rocky Mountain elk. J. Wildl. Dis. 34:532-536. Spraker, T. R., M. W. Miller, E. S. Williams, D. M. Getzy, W. J. Adrian, G. G. Schoonveld, R. A. Spowart, K. 1. O'Rourke, J. M. Miller, and P. A. Merz. 1997. Spongiform encephalopathy in free-ranging mule deer (Odocoileus hemionus), white-tailed deer (Odocoileus virginianus), and Rocky Mountain elk (Cervus elaphus nelsoni) in northcentral Colorado. J. Wildl. Dis. 33:1-6. Williams, E. S., and S. Young. 1980. Chronic Wasting disease of captive mule deer: A spongiform encephalopathy. Journal of Wildlife Diseases 16:89-98. and __ . 1982. Spongiform encephalopathy of Rocky Mountain elk. Journal of Wildlife Diseases 18: 465-471. and __ . 1992. Spongiform encephalopathies in Cervidae. Revue Scientifique et Technique Office International des Epizooties 11:551-567. ---' and __ . 1993. Neuropathology of chronic wasting disease in mule deer (Odocoileus hemionus) and elk (Cervus elaphus nelsoni). Veterinary Pathology 30:36-45. _-J _-J Table 1. Profiles used in chronic wasting disease targeted surveillance. • Species: • Age: mule deer white-tailed deer elk 2:. 18 months • Signs: emaciated and abnormal behavior &lor indifference to human activity &lor increased salivation &lor tremor, stumbling, incoordination &lor difficulty or inefficiency in chewing/swallowing increased drinking and urination &lor �123 Table 2. Tentative schedule for systematic statewide chronic wasting disease surveys of deer populations using harvest and/or road-kill samples. . CWD Status GMUs Years surveyed Endemic foci (:::::2%) 9, 19, 191,20 1996-2006 Endemic (~1%) .29 90,91,92,93 94 95,951,96 1997-2000 1997-1998;2004-2006 1997-2000 1996-1998,2004-2006 "High Risk" 87 :}8, 28, 37, 371 6, 16, 17, 161, 171 22 1997-2000 1998 1999-2001 1999 Other areas . 104 66,67 83 "Uncompahgre" "Southern Front Range" Random Statewide ..r- -. ~·i .. 1996-2000 1997-1998 1994-1993, 1998-1999 2000 2001 2002-2003 [Note: Survey targets and schedules subject to changes influenced by new data and availability of resources supporting CWD surveillance activities.] Table 3. Prospective strategies for chronic wasting disease management. Strategies for limiting distribution - preclude human-caused translocations (mostly done) - interrupt migration corridors (fencing?) - change habitat (improve/degrade to influence animal distribution?) - alter migratory behavior (how?) - reduce density (relationship between density & distribution untested) - other? Strategies for reducing prevalence - vaccination (no vaccine available) - treatment (no effective therapeutic drugs available) - genetic selection (evidence of uniform susceptibility) - fertility control (maternal transmission probably a minor contribution) - eliminate clinical suspects (misses carriers; ineffective to date) - test & slaughter (no reliable, practical live animal test) - selective culling (what criteria? mechanism?) - foster natural mortality (e.g., predation) to promote selective culling - reduce density (relationship between density & transmission untested) - other? �124 Table 4. Initial CWD management tactics to be evaluated via field studies and/or simulation modeling. Candidate tactic Schedule Selective culling via observation (± baiting) Modified "test and slaughter" (model/field) Selective culling via predation (modeling) Density reduction (modeling) Fertility control (modeling) 2/99-4/99 5/99-4/00 5/99-9/99 5/99-9/99 5/99-9/99 EXPERMIENTAL PLAN MANAGEMENT PROPOSAL -- NOT CURRENTLY INCLUDED IN DRAFT In all, eight management areas will be included in this before-and-after-controlled-intervention (BACI)experiment (Fig._); management areas will be paired, with four areas targeted for density reduction (50% reduction over a 2-yr period, then maintained at 50% of pretreatment density for 5 yr)' and four held under status quo management regimes (i.e., discourage feeding, eliminate suspects, etc.) as control areas (Table _). We will stagger the start of management treatments to minimize confounding effects of temporal variation on CWD prevalence and distribution (Table _). Deer density and CWD prevalence will be estimated annually in each management area, and mean CWD prevalences before and after management intervention will be compared to test density-dependent transmission hypotheses and assess the efficacy of treatment in reducing CWD prevalence. Similarly, CWD prevalence and distribution in GMUs adjacent to treatment and control areas will be compared to test hypotheses on density-dependent dispersal and assess the efficacy of treatment in stemming CWD dispersal. Table _. Treatment assignments and start dates for areas included in CWD management experiment. Location Treatment Control Start date northern Larimer County GMU9 GMU 191 1999-2000 southern Larimer County GMU 1ge/20ea GMU 19w120wb 2000-2001 South Platte River" bottom/plains GMU91 GMU92 1999-2000 South Platte River bottom/plains GMU96 GMU951 2000-2001 b GMU 19e120e will be the portions ofGMUs 19 and 20 east of the Pingree Park Road and ??? GMU 19w/20w will be the portions of GMUs 19 and 20 west of the Pingree Park Road and ??? 1 This preliminary treatment regime is subject to modification guided by simulation modeling exercises that will be completed later this year. �181 -..:':, \ "" Colorado Division of Wildlife Wildlife Research Report July 2000 / \ JOB PROGRESS REPORT State of Colorado Cost Center 3430 Project No. W-153-R-13 Mammals Program Work Package No. __ ..:::.3.:::.;00~1~ Task No. 10 _ Deer Management Monitoring and Managing Chronic Wasting Disease in Deer Period Covered: July 1, 1999 - June 30, 2000 Authors: M. W. Miller, C. T. Larsen, M. Conner, and R. Kahn Personnel: J. Gross, C. Krumm, M. Leslie, E. Myers, K. I. O'Rourke, T. R. Spraker, E. Wheeler, M. A. Wild, and E. S. Williams ABSTRACT Deer from throughout Colorado were examined for occurrence of chronic wasting disease using a combination of targeted surveys and harvest/road-kill surveys. Between June 1999 and May 2000, 11 chronic wasting disease(CWD)cases were diagnosed among 22 suspect deer submitted from known endemic portions of northeastern Colorado. Thirteen additional suspect deer submitted from elsewhere in Colorado were all CWD-negative. All confirmed CWD cases originated in game management units (GMUs) where the disease had been detected previously. We sampled about 1450 harvested or road-killed deer during October-December to randomly survey for CWD in select data analysis units (DADs). Immunohistochemistry (llIC)-based prevalence in Larimer County (6.4%) and South Platte River bottom/plains DADs (0%) were comparable to previous years' estimates. All brain samples from deer harvested in North Park (n = 103), the Piceance Basin (n = 408), and the Uncompahgre Plateau (11 = 209) tested negative. Overall, 1355 mule deer harvested or road-killed outside northeastern Colorado have all tested negative for CWD; these data, combined with those from targeted surveillance, support the contention that CWD is not distributed throughout the state. Since June 1999, CWD affected 11 of 21 adult (;;.:l-yr-old) mule deer and 2 of 5 adult white-tailed deer in the resident Foothills Wildlife Research Facility (FWRF) herds. No signs of neurological disease were observed in any of the 11 cattle (subjects) housed with naturally-infected resident mule deer since July 1997, but 3 of 11 mule deer from the Rocky Mountain Arsenal National Wildlife Refuge (RMANWR)(contact controls) held since October 1997 developed clinical CWD and were euthanized between November 1999 and June 2000. �182 Using ruc, Prl'?" was detected in brain tissue (medulla oblongata at the obex) from 2 mule deer experimentally inoculated with a 5 g oral dose of brain tissue homogenate from CWO-infected deer 20 mo earlier. One of the two deer examined 20 mo after inoculation (PI) also had gross lesions suggestive of I CWO and microscopic lesions of spongiform encephalopathy. Uninoculated controls collected from RMANWR at the same time were negative. Five of the 6 remaining inoculated deer died of clinical CWO or its sequelae 20-26 mo PI. Movements of radiocollared deer appeared related to spatial patterns in CWD prevalence described previously. Deer captured in GMU 9, where CWO prevalence is relatively high, moved north into areas of Wyoming with equally high prevalence; deer from GMU 191, located west and adjacent to GMU 9 and having relatively low levels of CWD prevalence, moved to the west into areas of even lower prevalence. Despite their proximity, there was no overlap in winter range for deer in GMU 9 and deer in GMU 191, and almost no overlap in summer range «1 %). Overall, deer movements appeared related to patterns of prevalence in DAD D4. It follows that migration and dispersal patterns of mule deer may explain, at least in part, spatial patterns of CWD prevalence in northcentral Colorado and southcentral Wyoming. Western blotting appears to have potential application as a tool in studying CWD transmission mechanisms. Preliminary evaluation revealed comparable sensitivity between Western blotting and ruc on both brain and lymphoid tissues. Although PrPCWD was not detected in saliva or buffy coat samples from infected deer, further development and evaluation of these and other potential media for CWD transmission appear warranted. Recommendations on a statewide approaches for monitoring and managing CWD in deer and elk, including policy, goals, and management objectives, were reviewed and finalized in November 1999; specific plans for CWD management will be included in individual DAU plans to be drafted in the next segment. Based on these recommendations, 3 interim management actions were undertaken: First, deer and elk population objectives for all DADs in the endemic area were reset to current (1999) levels regardless of previous longterm objectives established in DAD plans. Antlerless harvest levels were then set to hold populations at 1999 levels or, in cases where populations currently exceed DAU objectives, lower populations to objective within 2 years. Second, a more aggressive program for culling clinical suspects was instituted; this program enlists private landowners to assist in identifying and removing sick deer throughout the endemic area. Third, a management experiment to test relationships between CWD prevalence and deer density was planned and public information and regulation development processes supporting this experiment were initiated; that experiment will probably begin during winter 2000-2001. Surveillance activities for 19992000, as well as surveillance planned for 2000-2001, reflect priorities established in the monitoring portion of this plan. .j �183 MONITORING AND MANAGING CHRONIC WASTING DISEASE IN DEER M. W. Miller, C. T. Larsen, M. M. Conner, and R. Kahn P. N. OBJECTIVES ( I) Design, conduct, and report results of: (a) targeted surveillance to estimate and detect changes in distribution of chronic wasting disease (CWD) in free-ranging deer populations; and (b) harvest or road-kill surveys to estimate and detect changes in prevalence of CWD in endemic deer populations. (2) Design, conduct, and report results of experimental studies using captive deer naturally or experimentally infected with CWD. AGREEMENT OBJECTIVES (1) Conduct and report results of targeted surveillance to estimate and detect changes in distribution of CWD in free-ranging deer populations statewide. (2) Conduct and report results of harvest surveys to estimate prevalence ofCWD in DAUs D3, D4, D 10( +GMU 29), and D44 (+ GMUs 87 and 95), as well as select GMUs in the Piceance Basin and Uncompahgre Plateau. (3) Continue an experiment evaluating cattle susceptibility to CWD via natural contact exposure. (4) Continue to study pathogenesis ofCWD in mule deer. (5) Conduct studies to determine factors contributing to occurrence and natural spread of CWD in northeastern Colorado. (6) Evaluate select strategies for managing deer to limit distribution and occurrence of CWD . MATERIALS AND METHODS Surveillance We monitored deer populations throughout Colorado for occurrence of CWD using a combination of targeted surveillance and harvest or road-kill surveys. These were organized and conducted as follows: Targeted (= clinical. disease) surveillance: Deer showing clinical signs consistent with those seen in chronic wasting disease were collected by field personnel statewide and brain tissues examined for evidence of spongiform encephalopathy. The "suspect case" profile was defined as follows: • Species: • Age: mule deer white-tailed deer 2: 18 months �184 • Signs: emaciated and abnormal behavior &lor indifference to human activity &lor increased salivation &lor tremor, stumbling, incoordination &lor difficulty or inefficiency in chewing/swallowing &lor increased drinking and urination Where possible, submissions were subjected to complete necropsy; in some situations, only heads were available for examination and sampling. In all cases, histopathology of brain tissue (Williams and Young 1993) was used to diagnose CWD; in some cases, immunohistochemistry (lliC) or other ancillary tests were used to confirm or support diagnoses. Harvest surveys: In order to obtain reliable estimates of CWD prevalence that will serve as a basis for monitoring responses to management interventions, we continued conducting harvest surveys on select deer populations. During the 1999-2000 hunting seasons, fresh brain and select lymphatic tissues were collected from deer harvested in endemic game management units (GMUs). Deer harvested or culled in other select data analysis units (DAUs) throughout Colorado were also sampled as negative controls; in particular, deer populations in North Park (DAU D3), the Piceance Basin, and Uncompahgre Plateau were foci of surveillance activities. Brain tissues were examined at the Colorado State University Diagnostic Laboratory for anti-PrP inununostaining reactions (O'Rourke et at, 1998) and histopathological lesions (Williams and Young, 1993) consistent with CWD infection. Epidemiological Studies Epidemiology of naturally-occurring CWD in captive mule deer and white-tailed deer: Naturally-occurring CWD was a sporadic disease of resident FWRF mule deer prior to 1985 (Williams and Young, 1992), and also has occurred sporadically since 1994; no cases had been observed in resident white-tailed deer since their addition to FWRF in 1993. We maintained and observed 21 ~ l-yr-old mule deer and 5 ~ l-yr-old white-tailed deer in paddocks at CDOW's Foothills Wildlife Research Facility (FWRF) during this segment. Deer received natural forage, pelleted rations (high energy supplement and "browser" diet), and alfalfa hay; water and mineralized salt were available ad libitum. All deer were evaluated daily for clinical signs ofCWD (and other health problems) in conjunction with routine feeding and handling activities. Resident mule deer also represented the source of infection (= "treatment") in an ongoing study of cattle susceptibility to CWD (see below). Cattle susceptibility to CWD (Williams and Miller): We continued monitoring cattle for clinical evidence of natural CWD transmission as part of a coordinated interagency effort to study cattle susceptibility to CWD. Twelve 4-mo-old calves purchased from a private ranch located near Sheridan, WY, were placed in paddocks at the FWRF with naturally-infected captive mule deer in July 1997 (see above for description of resident deer herd). Twelve 4-mo-old mule deer fawns were captured at the Rocky Mountain Arsenal National Wildlife Refuge (RMANWR) in late September and placed in the same paddocks as contact controls for natural transmission (extensive ongoing surveillance of RMANWR deer has continued to confirm that this population is free of CWD). CWD Pathogenesis in mule deer (Williams and Miller): In a separate but related study, 20 additional 6-moold mule deer fawns were captured at the RMANWR in early December 1997. Each fawn was given a single oral dose of about 5 g of fresh brain tissue homogenate from captive mule deer with clinical CWD. Fawns were then released into a separate paddock physically removed from the contact transmission study. These experimentally-infected fawns serve two purposes: as controls for domestic calves orally inoculated with a single oral dose of about 50 g of this same CWD brain homogenate (Williams et al., 1998), and as subjects of a study on the pathogenesis of CWD in mule deer. �185 We have randomly sacrificed 2 experimentally-infected deer at 3,6,9, 12, 16, and 20 mo after inoculation; we also collected 2 age- and sex-matched control deer from RMANWR at each time step. Complete necropsies were performed and samples collected as described in the original study plan. Select sections of brain tissue (obex) were examined after staining with hematoxylin and eosin (H&E) or anti-PrP immunostain (F89/160 .1.5; O'Rourke et al., 1998); examination of other tissues is pending. Spatial epidemiology of CWO in free-ranging mule deer (Conner and Miller): Dispersal and/or migration movements could be related to the spatial distribution and spread of CWO. To determine whether mule deer movements from areas of high CWO prevalence are related to levels of CWO prevalence in surrounding areas and/or mule deer populations, we captured and radiocollared mule deer in 2 areas of high CWO prevalence. On 9 December 1999 and 10 December Number 1999,42 deer were captured in game management unit Age Radiocollared Area Sex (GMU) 9 near Virginia Dale, and from 7 January 2000 to Male 14 Fawn 29 February 2000, 22 deer were captured in GMU20 near GMU9 4 Yearling Masonville (Table 1). Because mule deer typically 2 Adult disperse between 12 and 30 months of age, we focused our 18 Female Fawn capture efforts on fawns and yearlings (6-30 months of 0 Yearling age) in order to gather data on both dispersal and 4 Adult 42 Total migration movements. We collected location data on deer radiocollared from this study, as well as from deer 6 Male Fawn GMU20 radiocolIared for an ongoing survival study in DAU 4. 4 Yearling Telemetered mule deer were located every 6-8 weeks. 1 Adult From preliminary data, we developed graphical 6 Female Fawn Yearling 3 information system maps that included locations of 2 Adult summer and winter range and migration movements. We 22 Total also calculated overwinter survival rates in accordance Table 1. Sex and age composition of mule deer with protocols from an ongoing study of deer survival in radiocollared during winter 1999-2000 in GMUs 9 DAU D4 as a basis for comparing vital rates for deer in and 20. areas of high and low CWO prevalence. Utility of Western blotting in detection ofPrPCWD in excreta (Larsen and Miller): Mechanisms for transmission of CWO remain undescribed. Circumstantial evidence suggests either direct or indirect animal-to-animal transmission occurs, most likely via agent excreted in some combination of saliva, feces, urine, or placental tissues (Miller et al., 2000). However, progress in understanding CWO transmission routes has been hampered by lack of tools for detecting PrPCWD in non-tissue media. W~ conducted preliminary evaluations of Western blotting as a technique for detecting PrPCWD in both tissue and excreta samples. Details of Western blot methods are described in detail elsewhere (Larsen and Miller, 2000). Monitoring and Management Recommendations for statewide monitoring and management of CWO in deer and elk were finalized and adopted. RESULTS AND DISCUSSION Surveillance Targeted (= clinical disease) surveillance: Between June 1999 and May 2000, 11 chronic wasting disease (CWO) cases were diagnosed among 22 «suspect" deer submitted from known endemic portions of northeastern Colorado; CWO was not diagnosed in any of 13 additional «suspect" deer submitted from �186 elsewhere in Colorado. All CWD cases confirmed in 1999-2000 were mule deer, and all originated in game management units (GMUs) where the disease had been detected previously. Harvest surveys: We sampled about 1450 harvested or road-killed deer to randomly survey for CWD in select DAUs. ruC-based prevalence in Larimer County mule deer (6.4%; 311487) was similar to prevalence observed in previous years. In contrast, no cases were detected among 60 mule deer harvested from South Platte River bottom/plains DAUs. Improvements in ruc techniques that allow identification of very early preclinical cases (based on strong staining in lymphoid tissue even in the absence of staining in brain tissue) should aid in reducing the bias in our prevalence estimates; we anticipate that these techniques will be validated prior to the 2000-2001 survey season. Based on preliminary results, estimated prevalence will likely increase by about 25-35% when these very early cases are included; the latter will likely be much closer to true prevalence rates in infected populatiuons. We detected no evidence ofCWD in any of the brain samples from North Park (0/105), Piceance Basin (0/423), or Uncompahgre Plateau (0/205) GMUs, or in any of the other 150 samples from GMUs outside northeastern Colorado. Overall, 1355 mule deer harvested or road-killed outside northeastern Colorado have all tested negative for CWD; these data, combined with those from targeted surveillance, support the contention that CWD is not distributed throughout the state. Epidemiological Studies Epidemiology of naturally-occurring CWD in captive mule deer and white-tailed deer: Between June 1999 and May 2000, CWD affected 11 of21 adult (z l-yr-old) mule deer in the resident Foothills Wildlife Research Facility herd. Resident mule deer also represented the source of infection (treatment) in an ongoing 10-yr study of cattle susceptibility to CWO (see below). Two of 5 white-tailed deer remaining in our resident FWRF herd developed CWD during June-August 1999. The 3 remaining white-tailed deer were euthanized in September 1999 - 2 of the 3 showed mildmoderate spongiform encephalopathy on histopathology. In all, CWD claimed 8 of 11 captive white-tailed deer held at FWRF between 1993 and 1999 and, based on the findings in the deer euthanized in September, would have probably eliminated the entire population by 2001. Examinations of tonsillar biopsies from captive deer are still pending. Lack of results has precluded evaluation of tonsillar biopsy as a potential antemortem test for CWD in deer. Cattle susceptibility to CWD (Williams and Miller): No signs of neurological disease were observed in any of the 11 cattle (subjects) housed withnaturally-infected resident deer since July 1997. However, 3 of9 mule deer from the Rocky Mountain Arsenal National Wildlife Refuge (RMANWR)(contact controls) developed clinical CWD between July 1999 and June 2000. Twenty-three of the 44 > l-yr-old naturallyexposed resident FWRF deer developed clinical CWO and diedor were euthanized during the first 36 mo of this study, thereby ensuring calves and control deer have received considerable exposure to CWD. All 11 cattle exposed to a single 50 g oral dose of brain tissue homogenate from CWD-infected deer remained healthy 33 mo postinoculation (through 30 June 2000) (E. S. Williams, pers. comm.). However, 3 of 14 cattle exposed to CWD via intracerebral inoculation died or were euthanized 22-27 mo postinoculation after showing varying combinations of neurological signs and weight loss; mc staining and Western blot results were consistent with probable CWD infection, although it is unclear whether infection with the CWD agent actually caused the clinical signs displayed by the 3 affected calves (R. J. Cutlip and A. Hamir , pers. comm.). �187 Pathogenesis of CWD in mule deer (Williams and Miller): Using IDC, Prpres was detected in brain tissue (medulla oblongata at the obex) from 1 of 2 mule deer experimentally inoculated with CWD 6 or 9 mo earlier. One of2 deer examined 16 or 20 mo after inoculation (PQ had lesions ofspongiform encephalopathy, and both were IHC-positive at each sampling. Two deer examined 3 mo PI and 2 examined 12 mo PI were negative, as were all 10 uninoculated controls collected from RMANWR at the same intervals. One of the 8 remaining deer inoculated in December 1997 began showing early clinical signs of CWO ~15 mo PI and died ~6 mo later; a second began showing early signs 2-3 wks later, and subtle signs were noted in at least 4 of 8 inoculated deer alive 18 mo PI. Of the 6 inoculated deer surviving >20 mo PI, 5 died or were euthanized in end-stage clinical CWD 20-26 mo PI; the remaining animal showed no appreciable clinical signs or lesions, and only minimal IHC staining in brain tissue, when euthanized 26 mo PI. Spatial epidemiology of CWO in free-ranging mule deer (Conner and Miller): . TIlls year served as a pilot study in which movements of mule deer were described. Most (86%) of the deer captured in GMU 9 and GMU 20 were fawns and yearlings (Table 1). From preliminary data, we mapped and visually compared locations of summer and winter range and migration movements _ SJmmer Heme Ranges SUmmer GMU20 in areas of high CWD C]GMJ aty prevalence (GMU 9, GMU N/l../Hiltl-y 20 20) to and areas oflower prevalence (GMU 191, 10 0 10 20 Kilanelers GMU 19) (Fig. 1). Movements of deer showed a remarkably clear Figure 2. Sununer and winter home range and movement patterns from winter to relationship to prevalence sununer home range for mule deer radiocollared in GMUs 9,191, 19, and 20. patterns. Deer in GMU 9, . where CWD prevalence is relatively high, moved north into areas of Wyoming with equally high prevalence. Deer in GMU 191, located west and adjacent to GMU 9 and having relatively low levels of CWO prevalence, moved to the west into areas of even lower prevalence. Despite their proximity, there was no overlap in winter range for deer in GMU 9 and deer in GMU 191, and almost no overlap in summer range 1%). Overall, deer movements appeared related to patterns of prevalence in DAU 4. It follows that migration patterns of mule deer may explain, at least in part, spatial patterns of CWD prevalence in northcentral Colorado and southcentral Wyoming. ~ WlnI ••• Home Ranges ~~~- « Overwinter fawn and adult survival was .estimated for 1 December to 15 June (Table 2). Survival did not differ betwee~ high prevalence (GMU 9) and lower prevalence GMUs in DAU D4 (GMUs 191 and 19); that is, all confidence intervals almost fully overlapped. TIlls is not surprising, because clinical CWO is �188 rare in the younger age classes that dominated our sample. Unfortunately, our sample sizes were insufficient to accurately estimate age- and sex-specific survival rates within GMUs, and thus comparisons of adult survival rates between high and low prevalence GMUs was precluded. We anticipate that cumulative data gathered over the next several years will eventually allow for such comparisons. Suvival Year 1900-00 Period 10 Dec-15 June Wrtering twa 06.U4 GMJ9 GMJ19 GMJ191 GMJ2) Sample Suvival Size Estimate 0.823 137 45 0.764 2) 0.612 0.940 52 2) 0.850 g)%CI 0.748 0.899 0.630 0.00) 0.378 0.843 0.874 1.000 o.eso 1.000 Table 2. Kaplan-Meier survival estimates for mule deer wintering in GMU 9, GMU 19, GMU 191, and GMU 20. Trap mortalities were excluded from survival calculations. Three deer radiocollared prior to 1999 were present in GMU 9, thus sample size was 45 rather than the 42 captured this year. Utility of Western blotting in detection ofPrPCWD in excreta (Larsen and Miller): Western blotting successfully detected PrPCWD in both brain and tonsil tissues from CWD-infected deer (fable 3, Fig. 2); all negative control samples tested negative. Sensitivity of Western blotting in these tissues appeared similar to mc. In contrast, we failed to demonstrate PrPCWD in either saliva or buffy coat extracted from blood. Whether PrPCWD does not occur in these media, or whether it is simply present at concentrations that fall below assay detection limits remains unclear. Tissue- and media-specific modifications in sample preparation procedures probably will be needed before Western blotting can be applied reliably to various materials of interest. In particular, modifications should focus on ways of concentrating the relatively small quantities of PrPCWD that may occur in excreta and environmental samples. Provided such obstacles can be overcome, Western blotting could have a variety of applications in studying CWD epidemiology. Monitoring and Management Recommendations for statewide monitoring and management of CWD (Appendix A) were finalized and adopted. Specific recommendations for policy, as well as management goals and objectives were as follows: Policy The Colorado Division of Wildlife is committed to minimizing the threat that chronic wasting disease (CWD) poses to Colorado's native deer and elk resources. In data analysis units (DAUs) where CWD is endemic, goals of reducing the occurrence and preventing the further spread of CWD will serve as the primary basis for setting management objectives. Goals The CDOW's goals for CWD management will be 1) to limit distribution ofCWD to no more than the 5 deer data analysis units (DAUs) and 2 elk DAUs where it already occurs and 2) to reduce average CWD prevalence among deer and elk to <1% in each endemic DAU and <2% in each endemic game management unit (GMU). Short-term objectives 1. Manage to prevent dispersal ofCWD beyond endemic DAUs via mule deer, white-tailed deer, and elk movements. 2. Manage to reduce CWD prevalence among deer inhabiting the three most severely affected GMUs (9,191,19) by ~50% within 10 years (by 2011). . 3. Manage to reduce CWD prevalence among deer and elk inhabiting all endemic DAUs to 51% within 20 years (by 2021). Based on these recommendations, 3 interim management actions were undertaken: First, deer and elk population objectives for all DAUs in the endemic area were reset to current (1999) levels regardless of previous long-term objectives established in DAU plans. Antlerless harvest levels were set to hold �189 populations at 1999 levels or, in cases where populations currently exceed DAU objectives, lower populations to those objective levels within 2 years. Second, a more aggressive program for culling clinical suspects was instituted; this program enlists private landowners to assist in identifying and removing sick deer throughout the endemic area. Third, a management experiment to test relationships between CWD prevalence and deer density was planned and public information and regulation development processes supporting this experiment were initiated; that experiment will probably begin during winter 2000-2001 (see details in Appendix B). Surveillance activities for 1999-2000, as well as surveillance planned for 2000-2001, reflect priorities established in the monitoring portion of this plan. ACKNOWLEDGMENTS The statewide CWD monitoring and surveillance program described here relies heavily on efforts of dedicated field personnel throughout the Colorado Division of Wildlife, and truly represents a division-wide effort to improve our understanding and management of this important disease problems. In addition to those specifically listed, we collectively thank all of those regional and area biologists, district and area wildlife managers, volunteers, deer hunters, and others who assisted by submitting suspect cases, harvested animals, or road-killed animals throughout the year. LITERATURE CITED Larsen, C. T., and M. W. Miller. 2000. Evaluation of Western blotting as an investigation tool for chronic wasting disease in mule deer. Colorado division of Wildlife, unpublished report. ' Miller, M. W., E. S. Williams, C. W. McCarty, T. R. Spraker, T. 1. Kreeger, C. T. Larsen, and E. T. Thome. 2000. Epizootiology of chr6nic wasting disease in free-ranging cervids in Colorado and Wyoming. Journal of Wildlife Diseases 38: 676-690. O'Rourke, K. I., T. V. Baszler, J. M. Miller, T. R. Spraker, I. Sadler-Riggleman, and D. P. Knowles. 1998. Monoclonal antibody F89/160.l.5 defines a conserved epitope on the ruminant prion protein. 1. Clin. Microbiol. 36: 1750-1755. Williams, E. S., and S. Young. 1993. Neuropathology of chronic wasting disease in mule deer (Odocoileus hemionus) and elk (Cervus elaphus nelsoni). Veterinary Pathology 30: 36-45. ---' M. W. Miller, T. 1. Kreeger, H. Van Campen, and T. R. Spraker. 1998. Susceptibility of cattle to cervid spongiform encephalopathy. unpublished grant proposal, submitted to USDA, CSREES, National Research Initiative Competitive Grants Program, 88 pp. Prepared by Michael W. Miller Wildlife Research Veterinarian _ ��191 Appendix A Chronic Wasting Disease in Colorado Deer and Elk: Recommendations for Statewide Monitoring and Experimental Management Planning M W. Miller and R. H Kahn Terrestrial Wildlife Resources, Colorado Division of Wildlife Executive Summary Chronic wasting disease (CWO) is a transmissible spongiform encephalopathy (TSE) of native deer and elk that is endemic throughout northeastern Colorado and southeastern Wyoming. Estimated infection rates range from <1-15% in deer and s 1% in elk residing in northeastern Colorado game management units (GMUs). There is no precedent for attempting to manage a TSE in free-ranging wildlife, and the biological imperative for such management is not entirely clear. However, it seems most biologically responsible to assume that CWO adversely affects native deer and elk populations and consequently manage to prevent its further spread and reduce its occurrence in Colorado. The CDOW has been actively monitoring CWO distribution and prevalence for over 10 years. Ongoing surveillance programs should be continued to provide data on responses to management and public information on CWO. Future monitoring should focus on reconfirming the limited distribution of CWO in Colorado and on measuring responses to prospective disease management programs. At present, the CDOW has no stated po~cy on managing CWO in free-ranging deer and elk. Lack of clear policy has led to internal confusion, disagreement, and conflicts with other policies and Long Range Plan objectives (e.g., increase recreational opportunities for deer and elk hunters). Because CWO currently affects <5% of Colorado's deer and elk resources and recreational opportunities but threatens far more substantial resources both within and outside Colorado, the following policy should be adopted to guide decisions on deer and elk population management in data analysis units (DADs) where CWO is endemic: The Colorado Division of Wildlife is committed to minimizing the threat that chronic wasting disease (CWD) poses to Colorado's native deer and elk resources. In data analysis units (DAUs) where CWD is endemic, goals of reducing the occurrence and preventing the further spread of CWD will serve as the primary basis for setting management objectives. Based on current understanding of CWO in free-ranging deer and elk, eradication seems an extreme and unjustifiable management goal at this time. Considering the prevalence and distribution CWO has reached under historical management regimes, however, continuing to rely on current management approaches seems equally unjustifiable. Consequently, two intermediate goals for managing CWO in deer and elk should be adopted: The CDOW's goals for CWD management will be 1) to limit distribution ofCWD to no more than the 5 deer data analysis units (DAUs) and 2 elk DAUs where it already occurs and 2) to reduce average CWD prevalence among deer and elk to <1% in each endemic_DAU and <2% in each endemic game management unit (GMU). Achieving the foregoing management goals should protect Colorado's deer and elk resources, restore confidence in the quality of deer and elk harvested in northeastern Colorado, and address agricultural constituencies' concerns about risk of potential transmission to domestic livestock or privately-owned wildlife. To reach these goals, changes in population management should be made via revision ofDAU �192 plans for the 5 deer DADs and 2 elk DADs where CWD is endemic by 1 January 2001 in order to achieve three short-term management objectives: 1. Manage to prevent dispersal ofCWD beyond endemic DAUs via mule deer, white-tailed deer, and elk movements. 2. Manage to reduce CWD prevalence among deer inhabiting the three most severely affected GMUs (9, 191, 19) by ~50% within 10 years (by 2011). 3. Manage to reduce CWD prevalence among deer and elk inhabiting all endemic DAUs to 51% within 20 years (by 2021). Developing these plans will necessarily involve broader public input than typical DAD planning processes in order to properly balance statewide concerns over CWD and its management with local concerns about the impacts of proposed management actions. In light of uncertainties about CWD epidemiology and ecology, plans should be approached as experiments that provide a mechanism for learning while attempting disease management. Because CWD management will be experimental, individual DAD plans should be coordinated through CDOW's Terrestrial Resources Program to ensure that elements of uniformity, design, control, and implementation are identified and achieved. Plans for CWD management should be reviewed and, where necessary, revised every 5 years as part of an adaptive strategy for achieving management objectives. Specific strategies and tactics for accomplishing CWD management objectives should be identified in individual DAD plans. Many conventional disease management options (most notably, vaccines, therapeutics, and live-animal tests for CWD) are presently unavailable; all other prospective strategies identified to reduce CWD prevalence involve some form of aggressive culling or population control. It is likely that current season structure, bag limits, and/or administrative boundaries may be too inflexible to achieve management objectives. Some candidate tactics for CWD management may require modification or temporary suspension of Commission regulations, as well as considerable human dimensions efforts. Managing CWD represents an unprecedented challenge to CDOW's biologists, field personnel, public service representatives, and administrators. However, rising to meet this challenge is vastly preferable to a legacy of inaction that allows CWD to become more widespread at the expense of valuable wildlife resources in Colorado and elsewhere. Background Chronic wasting disease (CWD) is a transmissible spongiform encephalopathy (TSE) of native deer (Odocoileus spp.) and elk (Cervus elaphus nelsoni) characterized by behavioral changes and progressive loss of body condition that invariably leadto the death ofaffected animals (Williams and Young 1992). Neither the causative agent nor its mode of transmission have been fully identified; however, CWD is suspected to be caused by aunique strain ofprion transmitted from animal-to-animal via some combination of saliva, feces, urine, or other tissues or fluids (Miller et al. 1998, 2000; Williams et al. 2000). Both sexes and all age classes show relatively uniform susceptibility (Miller et al. 2000, Williams et al. 2000). There are no valid tests currently available for diagnosing CWD in live animals, and extant postmortem tests require microscopic examination of brain tissue. There are no known treatments for CWD. Previous attempts to eradicate CWD from research facilities failed on at least 2 occasions (Williams and Young 1992, Miller et al. 1998). Although similar in some respects to other TSEs that affect domestic sheep (scrapie) and cattle (bovine spongiform encephalopathy; BSE), existing and unpublished experimental data indicate CWD cannot be naturally transmitted to domestic livestock, and that scrapie and BSE cannot be naturally transmitted to native cervids. Moreover, available data indicate that CWD does not present a threat to human health. �193 "Chronic wasting disease" was first recognized by biologists in the 1960's as a disease syndrome of captive deer held in wildlife research facilities in Ft. Collins, CO. This disease was subsequently recognized in captive deer, and later in captive elk, in wildlife research facilities near Ft. Collins, Kremmling, and Meeker, CO and Wheatland, WY (Williams and Young 1980, 1982), as well as at in least two zoological collections (Williams and Young, 1992). Since 1996, CWD also has been diagnosed in captive elk residing in two game ranches in Saskatchewan, Canada, as well as five ranches in South Dakota, two ranches in Nebraska, and one ranch each in Oklahoma, Colorado, and Montana (Fig. 1). Over 100 cases of clinical CWD also have been diagnosed in free-ranging mule deer (0. hemionus), white-tailed deer (0. virginianusi, and elk from northeastern Colorado and southeastern Wyoming since 1981 (Miller et al. 2000). At present, the known world-wide distribution of CWD in wild cervids appears to be limited to northeastern Colorado and southeastern Wyoming (Miller et al. 2000). Although CWD was first diagnosed in captive cervids, the original source of CWD in either captive cervids or free-ranging cervids is unknown; whether CWD in captive cervids really preceded CWD in wild cervids, or vice versa, is equally uncertain (Spraker et al. 1997). Moreover, the relationship (if any) between CWD cases in privately-owned Figure 1. Distribution of chronic wasting disease in captive and freeelk industry and cases in ranging deer and elk (November 1999). free-ranging and research animals is unclear. A.MuJec2er In Colorado, free-ranging CWD cases in deer and elk have originated from throughout the northeastern portion of the state. Game management units (GMUs) yielding infected deer or elk include 7, 8,9, 19, 191, 20,29,91,93,94,95, 951, and 96. Although B. White-tailed ~er cases have come from 13 different GMUs, two data analysis units (04, 010) have yielded over 85% of the documented cases. Harvest surveys indicate GMUs 9,191, and 19 are the most extensively infected GMUs in Colorado; estimated Figure 2. Estimated CWD prevalence (%) among subpopulations of (A) mule deer and (B) prevalence ranges from white-tailed deer in northeastern Colorado. Data were aggregated on the basis of deer about 6-15% (Fig. 2) movement patterns. Prevalence estimates are based on immunohistochemistry reactions only (lliC+/total). (Miller et al. 2000). I �194 Data from harvest surveys indicate that CWD is less prevalent in GMUs surrounding this core area, affecting about 1-4% of the deer in GMUs 7,8,20,29,91,93,94,95,951, and 96 (Fig. 2). Similarly, harvest surveys indicate <1 % of the elk in DAUs E4 and E9 are infected with CWD (Miller et al. 2000). Based on targeted surveillance data and select harvest surveys, wild deer or elk populations in other parts of Colorado and Wyoming are probably not infected with CWD (Miller et al. 2000); similarly, surveillance data gathered to date indicate CWO has not yet spread to South Dakota, Nebraska, or Kansas (A. Jenny, USDAIAPHISNS/NVSL, pers. comm.). For infected deer populations northeastern Colorado, CWO prevalence appears to be stable; however, because small increases would be difficult to detect over only a few years it is doubtful that our short-term observations accurately reflect or forecast long-term trends. In the absence of historical (20-30 years ago) prevalence data or reliable estimates of transmission rates, changes in overall incidence or distribution of CWO in northcentral Colorado DAUs are probably best projected via modeling. Based on field data and results of simulation modeling to predict dynamics and impacts of CWO on affected deer populations, it appears CWO was introduced into northern Larimer County over 30 years ago (Miller et al. 2000, Gross and Miller 2000) and has increased in both prevalence and distribution since that time. Beyond the Larimer County foothills, the South Platte River bottom corridor appears to be the single most important and predictable avenue for spread ofCWD (Fig. 2). Kufeld and Bowden (1995) reported that a proportion of South Platte River bottom deer were highly mobile; observed deer movements included some to and from eastern Larimer County along the Cache la Poudre and South Platte Rivers. These movement patterns provide a plausible mechanism for the extent of CWD' s distribution across South Platte River bottom GMUs (Fig. 2) (Miller et al. 2000). Other likely routes for deer emigrating from eastern Larimer County have not been completely identified, and may be less predictable or less commonly used. Altitudinal movements of some elk subpopulations in the southern part of Larimer County are also somewhat predictable (Bear 1982). Although there is potential for elk to cross the Continental Divide through Rocky Mountain National Park, the relatively low prevalence of CWO among elk may diminish the likelihood of its spread via their movements. The significance of CWO and its impacts on native deer or elk populations have not been determined . . Preliminary results of simulation modeling suggest that, sustained at >5% prevalence as observed in the most heavily infected GMUs, CWO could impact wild deer herds and lead to population declines (Miller et al. 2000, Gross and Miller 2000). Analyses of harvest and other population data have not detected evidence of population depression attributable to CWD, but existing inventory data for deer in northeastern Colorado are probably insufficient to detect such trends (M. Conner, pers. comm.). In the absence of data to the contrary, and considering the difficulties inherent in eliminating CWO from captive or wild cervid populations once established, it seems most prudent to assume that CWO adversely affects native deer or elk populations and manage to prevent its further spread and reduce its OCcurrence in Colorado. . Unfortunately, there is considerable uncertainty in how to manage CWO in free-ranging wildlife, or whether such an endeavor could even be successful (Gross and Miller 2000). Because a more complete understanding of CWO is fundamental to developing comprehensive management programs, the Colorado Division of Wildlife (CDOW) needs to further understanding about CWO and its management through surveillance and experimental management. Data from completed, ongoing, and proposed research, both basic and applied, will serve as the foundation of a series of adaptive resource management plans for CWO in deer and elk. These plans will provide mechanisms for incorporating new knowledge gained through surveys, modeling, research studies, and management experiments into a continuously evolving management program for CWO. �195 STATEWIDE MONITORING PROGRAM The CDOW has been actively monitoring CWD distribution and prevalence for over 10 years. Ongoing surveillance programs should be continued in order to provide data on responses to management attempts, as well as public information on CWD. The recommended goals and strategies for statewide monitoring are as follows: . Goals: 1. Provide reliable estimates of CWD distribution and prevalence to serve as a basis for management decisions and public information. 2. Provide information to improve understanding ofCWD epidemiology. 3. Continue developing efficient and reliable techniques for detecting and monitoring CWD in free-ranging populations. Strategy: Reliable estimates of CWD prevalence in wild deer and elk populations are needed to guide policy decisions and monitor efficacy of management efforts. The CDOW should continue to monitor deer and elk populations throughout Colorado for occurrence of chronic wasting disease using a combination of extensive and intensive approaches. These have been organized and conducted as follows: Targeted (= clinical disease) surveillance: Ongoing statewide "targeted" surveillance (i.e., submissions of "suspect" deer & elk) will be used to determine & monitor changes in CWD distribution. Based on experiences in Colorado and elsewhere, this appears to be the most sensitive approach for determining CWD distribution and detecting range extensions (Miller et al. 2000). A program for acquiring, examining, reporting on, and summarizing "suspect" CWD cases occurring throughout Colorado has been in place since 1990. This program has served to increase ongoing surveillance efforts by field personnel statewide by encouraging submission of carcasses from deer or elk showing clinical signs resembling CWD. A formalized process for submitting cases has been developed; this process includes criteria for acceptable submissions, submission forms, handling instructions, and a system for networking submissions within CDOW and among the 3 Colorado State Veterinary Diagnostic Laboratories and the Wyoming State Veterinary Laboratory. Under this program, deer and elk showing clinical signs consistent with those seen in chronic wasting disease are collected by field personnel statewide; a "suspect case" profile has been defined (Table 1). Brain tissues are subsequently examined for evidence of spongiform encephalopathy. Where possible, whole carcasses will continue to be submitted for complete necropsy; at minimum, heads from suspect animals will be submitted for examination and sampling. Histopathology (Williams and Young, 1993) and immunohistochemistry of brain tissue (Spraker et al., 1997) will be performed on all suspect cases; other ancillary tests may also be used to confirm or support diagnoses. Pertinent data from preliminary and final reports will be entered into a permanent database, and copies of reports will be filed as well as sent to appropriate field personnel. Results from targeted surveillance should be used to identify new potential foci of CWD infection for further evaluation via harvest or road-kill surveys. Harvest and road-kill surveys: Surveys of harvested or, in some areas, road-killed deer and elk should be conducted in endemic, high-risk, and outlying DAUs/GMUs to estimate and monitor CWD prevalence. We recommend that initial surveys be conducted in all DAUs/GMUs where CWO is detected via targeted surveillance to estimate prevalence, as has been done to date. We also recommend continuing to conduct regular surveys at 1- to 5-year intervals in DAUs or GMUs where CWD is endemic to obtain reliable �196 estimates of prevalence to use as a basis for monitoring natural trends or responses to management interventions. Prevalence should be monitored annually in DAUs/GMUs where experimental management programs are implemented. Surveys should be designed to provide an estimated prevalence within ±2% at a 95% level of confidence for affected populations where true prevalence is 1%; we recognize that it may be necessary to pool data from successive years in order to achieve desired sample sizes. In addition to surveying endemic foci, CDOW should also continue conducting a series of surveys for CWD in other select deer and elk populations throughout Colorado over the next 2 years, with a goal of determining whether CWD is more widespread than suggested by targeted surveillance data. We recommend the following surveillance strategy: Areas newly identified as infected via targeted surveillance will be surveyed sufficiently to obtain a reliable prevalence estimate. Those areas where CWD is not being managed or where estimated prevalence is < 1% will be regarded as endemic and populations will be monitored at -5-year intervals to measure long-term trends in prevalence. Areas where experimental CWD management strategies are being tested (both treatment and control areas) will be monitored annually to track changes in prevalence and evaluate management efficacy. We also recommend conducting surveys of at least 2 other select deer populations where CWD has not been reported previously. Target areas remaining include one high-risk and one low risk population (fable 2). Surveys will be designed such that the probability of failure to detect at least I case of CWD in an apparentlyunaffected deer or elk population will be :sO.0 I even if herd prevalence is 1%. Finally, we will attempt to gather a sufficient number of random samples from DAUs statewide (n-l,OOO) to determine whether or not CWD should be generally regarded as endemic in Colorado's deer populations. Harvest surveys should continue to be conducted using techniques developed and evaluated in Colorado since 1991. With the advent of limited licensing for deer hunting statewide, mandatory submission requirements in GMUs where surveys are being conducted can be dropped after 1999. Limited licenses may be sought in select GMUs to aid in increasing compliance where elk surveillance is needed. Unstaffed barrel sites appear to be the most efficient method for sample collection, and should continue to be used as the primary method for conducting harvest surveys. Brainstem (medulla oblongata at the obex) from <!: 1year-old deer and elk will be collected for CWD testing, and immunohistochemistry (lliC) using USDA monoclonal antibody F89/160.1.5 (O'Rourke et al., 1998) should be used as the primary screening tool. Reported prevalence estimates will continue to be based on numbers of IRC-positive cases; because mc appears more sensitive than histopathology in detecting preclinical CWD cases, mC-based prevalence estimates should provide a relatively unbiased measure of disease trends over time and allow more direct comparison between field data and simulation model predictions. We also recommend continuing to evaluate lymphoid tissue (e.g., tonsil) IRC as a means of increasing survey sensitivity and as a tool for staging disease progress and monitoring responses to management attempts. EXPERIMENTAL MANAGEMENT There is no precedent for attempting to manage a TSE in free-ranging wildlife. Moreover, the biological imperative for such management is not entirely clear at present. Programs for managing or eliminating TSEs of domestic livestock have proven only marginally successful, and the lack of obvious success in eradicating scrapie or BSE from endemic countries, compounded by epidemiological differences between CWD and other TSEs, make such programs rather poor models for prospective CWD management. Several fundamental features of CWD epidemiology are understood poorly, if at all; in particular, the influences of host (e.g., population density, intraspecific vs. interspecific transmission, genetics) and environmental factors (e.g., contamination, reservoirs, weather) on CWD dynamics remain unclear. In light of the uncertainties associated with its epidemiology and ecology, we believe CWD management should be approached as an experiment, thereby allowing us to learn as we take actions intended to reduce prevalence and limit distribution. �197 . . The CDOW's initial approaches for attempting to control CWD in free-ranging deer and elk were outlined in an action plan for CWD drafted in 1995 (Miller et al. 1995; Appendix A). In addition to calling for a comprehensive public information campaign, this plan identified tactics for limiting distribution and reducing prevalence of CWD based on contemporary knowledge of the problem. Prompt removal of affected animals and enforcement of feeding prohibitions were recommended to help reduce CWD prevalence and transmission. Policies and regulations were also recommended, and subsequently instituted, to prevent wider distribution ofCWD via human activities (e.g., translocations, rehabilitation, game ranching). The plan also identified a need for more extensive surveillance to gather data on CWD distribution and prevalence. As more data on CWD distribution and prevalence have been gathered over the last 4 years, it has become apparent that CWD is more widely distributed and more prevalent in northeastern Colorado (and elsewhere) than initially believed (Miller et al. 2000). It follows that policy, goals, and strategies for experimental CWD management in Colorado should be reevaluated in the context of new data, as well as changes in social and political attitudes toward animal TSEs. Management policy: At present, the CDOW has no stated policy related to the management of chronic wasting disease in wild deer and elk. The lack of a clear policy statement has led to internal confusion and disagreement over how to resolve conflicts with other policies and Long Range Plan objectives (e.g., increase recreational opportunities for deer and elk hunters). . Recommendation: Because CWD currently affects a relatively small portion of Colorado's deer and elk resources «5% of both statewide populations and statewide harvests) but shows the potential to spread and thereby threaten more substantial resources both within and outside Colorado, we recommend adoption of the following policy to guide decisions on deer and elk population management in DAUs where CWD is endemic: PI. The Colorado Division of Wildlife is committed to minimizing the threat that chronic wasting disease (CWD) poses to Colorado's native deer and elk resources. In DAUs where CWD is endemic, goals of reducing the occurrence and preventing the further spread of CWD will serve as the primary basis for setting management objectives. The foregoing policy, if adopted, should provide clear direction and a solid foundation for developing goals, objectives, and strategies managing CWD, as detailed below. In the absence of this or a similar policy, CDOW's ability to develop and implement a comprehensive approach for managing CWD will be severely compromised. Prospective management goals and objectives: We have identified four overarching goals that could serve as a basis for decisions related to managing CWD: 1. modified "natural regulation", 2. containment, 3. reduction, 4. eradication. These goals represent a continuum of possible levels of intervention, ranging from benign neglect to aggressive action. There are clear advantages and disadvantages associated with each of these prospective management goals. Although each progressive level offers more complete control of the problem, better control is accompanied by ever-increasing technical, fiscal, and political challenges, as outlined below: �198 1. Modified "natural regulation": Directing no specific management intervention toward CWD beyond culling of reported clinical cases and enforcing feeding and rehabilitation regulations represents the status quo approach to both disease and affected deer and elk population management. It could be argued that the ongoing activities described above, combined with ongoing changes in habitats and movement corridors effected by development, collectively constitute an attempt to manage CWD. Although the impacts of this approach on CWD prevalence in endemic foci remain undetermined, prevalence in endemic foci appears to have remained stable over at least the last 3 years. This approach offers the fewest technical challenges, may be regarded by some as "safest" in light of the myriad of uncertainties in CWD epidemiology, and in the short-term would be most sparing of local resources. Based on current culling rates reflected by case submissions, <10% of clinical CWD cases are being detected and removed from affected populations annually in Larimer County (and fewer elsewhere). Unfortunately, simulation models forecast that problems with CWD will likely get worse in affected populations managed in this manner (Gross and Miller 2000). Continuing existing management practices will likely allow CWD to become more prevalent and widely distributed, with tangible potential impacts on wildlife resources, recreational opportunities, and revenues. In addition, perceived "inaction" may be unacceptable to some sportsmen's groups, agricultural interests, and perhaps the general public, and could lead to the loss of management authority by CDOW. Finally, opting for "natural regulation" of an emerging disease problem seems biologically irresponsible. 2. Containment: Managing primarily to prevent spread of CWD from endemic foci would require additional knowledge of the mechanisms and probable routes of CWD dissemination, but many of the important strategic components needed to support this strategy are already in place. The likelihood of translocating infected animals from endemic areas to other areas in Colorado or elsewhere was significantly reduced by Commission'regulations adopted in 1996. Between existing data from previous movement studies (e.g., Kufeld et al., 1989; Kufeld and Bowden, 1995) and data from studies initiated more recently, major natural emigration corridors from eastern Larimer County could probably be identified and subsequently might be interrupted via intensive culling operations. This approach, if successful, would limit the magnitude of the CWD problem while largely sparing local resources. This alternative may be more acceptable to sportsmen's groups and the general public than to agricultural interests, but is probably biologically justifiable. However, such an approach would require an essentially infinite long-term commitment and yet will not eliminate CWD entirely; prevalence could conceivably increase in endemic areas. There would likely be some local impacts on resources and recreational opportunities, perhaps accompanied by local public resistance. 3. Reduction: It may be possible to manage affected deer or elk populations to reduce CWD prevalence in endemic foci (Gross and Miller 2000). A goal of prevalence reduction clearly demonstrates intent to limit the magnitude of the problem. It seems likely that this approach would also provide for containment ofCWD. Attempting to reduce CWD prevalence might improve "consumer confidence" among hunters in endemic GMUs. This alternative may be the most widely acceptable among sportsmen's groups, agricultural interests, and the general public in terms of responsible disease and resource management. It is probably biologically justifiable in view of projected long-term effects of CWD on infected populations and the potential for CWD to spread from endemic foci if left unmanaged. As with the foregoing goal of disease containment, prevalence reduction will require a long-term commitment and may not eliminate CWD from endemic areas; however, early, aggressive intervention may increase the likelihood of management success (Gross and Miller 2000). Depending on specific tactics employed, prevalence could increase despite management efforts. This approach would likely cause more severe impacts on local resources and recreational opportunities, and consequently would be more likely to foster local public resistance to proposed management actions. 4. Eradication: Eradication is probably the most desirable goal of any attempt to manage CWD. Whether it would be feasible to completely eliminate CWD from Colorado remains highly questionable. �199 Eradicating CWD would be the best means of restoring for "consumer confidence" among northeastern Colorado deer and elk hunters, and would presumably nullify concerns of traditional and alternative livestock interests about the potential for transmission to privately-owned animals. Perhaps most importantly, eliminating CWD would be the most effective way to preempt threats of its eventual spread to other native deer and elk populations in Colorado and elsewhere. Although desirable in concept, CWD eradication is probably infeasible given the complexities and uncertainties of its epidemiology. Eradication of CWD would require long-term commitment, and probably would exact catastrophic impacts on resources and recreational opportunities in affected areas; broader ecological impacts (e.g., on predator-prey balances) also could be severe. Public resistance is likely to emerge, at least locally, as details of such a management program emerge. Finally, it is questionable whether the potential ecological costs of CWD eradication are biologically justifiable in light of the uncertainty about longterm impacts of the disease itself. Recommendations: Based on current understanding of CWD epidemiology, prevalence, and distribution in free-ranging deer and elk in northeastern Colorado, we believe eradication is an extreme and unjustifiable management goal at this time. In light of the magnitude of prevalence and distribution CWD has reached in Larimer County deer populations under historical management regimes, however, continuing to rely on current management approaches seems equally unjustifiable. Consequently, we recommend an intermediate approach with two goals for managing CWD in deer and elk: Gl. limit distribution occurs; and of CWD to no more than the 5 deer DAUs and 2 elk DAUs where it already G2. reduce average CWD prevalence among deer and elk to <1% in each endemic DAU and <2% in each endemic GMU. We believe achieving the foregoing management goals will serve to protect Colorado's deer and elk resources, restore confidence in the quality of deer and elk harvested in northeastern Colorado, and ameliorate political pressures from agricultural constituencies concerned about potential for transmission of CWD to domestic livestock or privately-owned wildlife. We further recommend DAU plans for the 5 deer DAUs (04, D5, DI0, D27, and D44) and 2 elk DAUs (E4, E9) where CWD is endemic be revised by I January 2001 to reflect changes in population management sufficient to achieve three short-term management objectives: Ml. to prevent further dispersal ofCWD beyond endemic DAUs via mule deer, white-tailed deer, and elk movements; and M2. to reduce CWD prevalence among deer inhabiting the three most severely affected GMUs (9, 191, 19) by ~50% within 10 years (by 2011); and M3. to reduce CWD prevalence among deer and elk inhabiting all endemic DAUs to ~1 % within 20 years (by 2021). Strategies for accomplishing management objectives will be identified in individual DAU plans. Developing these plans will necessarily involve broader public input than typical DAU planning processes in order to properly represent statewide concerns over CWO and its management; their development will entail collaborative efforts of Terrestrial Resources, Public Service, and Human Dimensions staffs. Because we regard any attempt to manage chronic wasting disease as experimental, we also recommend coordination through CDOW's Terrestrial Resources Program to ensure that elements of uniformity, design, and control are sufficient to implement management strategies and evaluate their effects. These plans should be �200 collectively reviewed, evaluated, and revised every 5 years as part of an adaptive process for achieving management goals. Where applicable, regulation adoption processes for implementing these DAU plans should be independent of the 5-year season structure process. Prospective management strategies: Prescribing specific management strategies is beyond the scope or intent of this document. However, we believe a clear understanding prospective strategies and options (or lack thereof) is a necessary prerequisite to adoption of the foregoing policy, goals, and objectives for managing CWD in northeastern Colorado. Consequently, we offer the following overview as a context for the decisions on CWD management recommended here. Effective strategies for managing CWD or any other TSE in free-ranging wildlife have never been identified or evaluated. Many conventional disease management options are precluded from consideration. Most notably, vaccines, therapeutics, and live-animal tests for CWD are presently unavailable (Table 3). Contrary to experiences with domestic sheep (Dawson et al. 1998), available data indicate relatively uniform genetic susceptibility to CWD among deer (K. I. O'Rourke and M. W. Miller, unpubl. data); consequently, genetic selection would probably have no impact on prevalence even if tools for its implementation were available. Similarly, controlling reproduction alone probably would be ineffective even if practical tools were available because maternal transmission alone is unlikely to be sustaining CWD prevalence at levels currently observed in some affected deer populations (Miller et al. 2000, Gross and Miller 2000). All other prospective options for reducing CWD prevalence that have been identified to date (Table 3) involve some form of aggressive culling br population control directed at diminishing or preempting disease transmission - the magnitude and duration of such control will be driven by GMU- and DAU-specific CWD prevalence, management objectives, and population responses. Based on current understanding of CWD epidemiology (Miller et al. 1998,2000; Gross and Miller 2000), strategies capable oflowering CWD transmission rates by reducing numbers of infected animals. and point sources of environmental contamination (e.g., feeders) should be most effective in lowering prevalence; although the precise mechanisms of transmission have not been identified, those details will probably have little influence on the resolution of CWD management at the population level. General models predict that random culling should be less effective than selective culling in reducing disease prevalence (Barlow 1996, McCarty and Miller 1998, Gross and Miller 2000). However, field data and pilot work indicate that management strategies based on culling clinical suspects probably will miss a large proportion of the infected (and potentially infectious) individuals in a population; moreover, there is no practical way to apply "test and slaughter" regimes without an antemortem diagnostic test for preclinical CWD. Accelerated culling of early clinical cases, either by humans or via predators, may help reduce CWD transmission and prevalence. Alternatively, if the probability ofCWD transmission is density-dependent, then reducing deer or elk densities in endemic areas should lead to reduced CWD prevalence; to date, .the hypothesis of densitydependent disease transmission remains untested. Obvious strategies for limiting CWD distribution remain equally elusive. The mechanisms driving the spread of CWO among deer and elk populations are even less apparent than those driving disease dynamics. It is possible that CWD distribution to date has been influenced by regular and/or random movements of deer and, less commonly, elk. Previous studies (Kufeld et al. 1989, Kufeld and Bowden 1995, S. Steinert pers. comm.) have shown that proportions of both foothills and river bottom deer populations in northeastern Colorado are either migratory or mobile; the movement patterns documented in these studies provide at least a partial explanation for observed CWO distribution. It is also possible that CWD influences its own distribution by changing behaviors, range fidelity, and movement pattens of affected animals; observations of repetitive pacing in captive deer and elk affected by CWO could be manifested as extensive wandering movements of free-ranging individuals. Whether or not movement �201 patterns or dispersal rates are density-dependent remains unclear; either way, reducing population size should also reduce the probability of affected individuals carrying CWD to new areas. (However, we do recognize the possibility that at some lower threshold reduced density may actually promote deer and elk movements, particularly during the breeding season). Experiments to Manage CWD: Based on current uncertainty about how to effectively disrupt the processes that influence CWD epidemiology, prevalence, and distribution in free-ranging deer in northeastern Colorado, we recommend initiating controlled management experiments directed toward testing one hypotheses about how best to manage CWD. Prospective testable hypotheses include: HI. CWD distribution in foothills and river bottom habitats can be diminished by a) selective culling; b) increasing natural mortality via mountain lion predation; OR c) reducing deer density in affected populations. H2. CWD transmission and prevalence in foothills and river bottom deer populations can be diminished by a) selective culling; b) increasing natural mortality via mountain lion predation; OR c) reducing deer density in affected populations. I We recommend using an adaptive resource management approach (Holling ~978, Walters and Holling 1990) to test candidate strategies for reducing CWD prevalence and distribution. Based on initial evaluations of these candidate strategies, more detailed management experiments can be developed in conjunction with the DAU planing process; as part of that process, decisions will be made on which, ifany, approaches to pursue and where CWD management will be attempted. Tactics for accomplishing management approaches (e.g., harvest, selective culling, ground/aerial gunning, changes in predator harvest, etc.) will follow from internal consensus on which tactics are viable under the social and political constraints imposed by the area(s) selected for study. All proposals should be review and approved through appropriate internal and public processes. However, because it is likely that conventional recreation-based season structures, bag limits, and/or administrative boundaries may be insufficient to accommodate experimental approaches for managing CWD, the processes for changing regulations supporting management experiments should be independent of the 5-year season structure process in order to maximize flexibility in this adaptive management approach. Prospective management tactics: Specific tactical recommendations employed in experimental CWD management should follow from the DAU planning process. As a general consideration, however, it is likely that current season structure, bag limits, and/or administrative boundaries may be too inflexible to accommodate experimental approaches for managing CWD that may be proposed for evaluation. Many candidate tactics (Table 4) for such management will require modification or temporary suspension of Commission regulations, as well as considerable human dimensions efforts. Within the confines of the broad policy, goals, and objectives provided, we believe specific tactics are best identified by a combination of Staff and field collaborators on a local level in conjunction with the DAU planning process and adopted as a package along with specific DAU plans for managing CWD. Developing comprehensive, objective-oriented DAU plans to manage CWD represents an unprecedented challenge to CDOW's biologists, field personnel, public service representatives, and administrators. �202 However, we believe rising to meet this challenge is vastly preferable to a legacy of inaction that will allow a potentially manageable wildlife health problem to become widespread at the expense of valuable wildlife resources in Colorado and elsewhere. LITERATURE CITED Barlow, N.D. 1996. The ecology of wildlife diesase control: simple models revisited. 1. Appl. Ecol. 33: 303-314. Bear, G. D. 1989. Seasonal distribution and population characteristics of elk in Estes Valley, Colorado. Colorado Division of Wildlife, Fort Collins, Colorado. Special Report Number 65: 1- 14. Dawson, M. L. 1. Hoinville, B. D. Hosie, and N. Hunter. 1998. Guidance on the use ofPrP genotyping as an aid to the control of clinical scrapie. Vet. Rec. 142:623-625. " Gross, J. E., and M. W. Miller. 2000. Chronic wasting disease in mule deer: A model of disease dynamics, control options, and population consequences. 1. Wildl. Manage. (in reView). Holling, C. S. 1978. Adaptive environmental assessment and management. John Wiley and Sons, London, England. Kufeld, R. C. and D. C. Bowden. 1995. Mule deer and white-tailed deer inhabiting eastern Colorado plains river bottoms. Colorado Division of Wildlife, Fort Collins, Colorado. Technical Publication Number 41: 1- 58. Kufeld, R. C., D. C. Bowden, and D.L. Schrupp,. 1989. Distribution and movement of female mule deer in the Rocky Mountain foothills. J. Wildl. Manage. 53: 871-877. McCarty, C. W., and M. W. Miller 1998. A versatile model of disease transmission applied to forecasting bovine tuberculosis dynamics in white-tailed deer populations. J. Wildl. Dis. 34: 722-730. Miller, M. W. 1999. Monitoring and managing chronic wasting disease in deer. Pages _-_ in Wildlife Research Report, Manunals Research, Federal Aid Projects, Job Progress Report, Project W-153-R-12, Work Package 3001, Task 3. Colorado Division of Wildlife, Fort Collins, Colorado, USA. (in press) Miller, M. W. 1999. Monitoring and managing chronic wasting disease in elk. Pages _-_ in Wildlife Research Report, Manunals Research, Federal Aid Projects, Job Progress Report, Project W-153-R-12, Work Package 3002, Task 3. Colorado Division of Wildlife, Fort Collins, Colorado, USA. (in press) Miller, M. W., M. A. Wild, and E. S. Williams. 1998. Epidemiology of chronic wasting disease in captive Rocky Mountain elk. J. Wildl. Dis. 34:532-536. Miller, M. W., E. S. Williams, C. W. McCarty, T. R. Spraker, T. 1. Kreeger, and E. T. Thome. 2000. Epidemiology of chronic wasting disease in free-ranging cervids. J. Wildl. Dis. (in review). Spraker, T. R., M. W. Miller, E. S. Williams, D. M. Getzy, W. 1. Adrian, G. G. Schoonveld, R. A. Spowart, K. I. O'Rourke, 1. M. Miller, and P. A. Merz. 1997. Spongiform encephalopathy in freeranging mule deer (Odocoileus hemionus), white-tailed deer (Odocoileus virginianus), and Rocky Mountain elk (Cervus elaphus nelsoni) in northcentral Colorado. J. Wildl. Dis. 33:1-6. Walters, C. J., and C. S. Holling. 1990. Large-scale management experiments and learning by doing. Ecology 71: 2060-2068. Williams, E. S., and S. Young. 1980. Chronic Wasting disease of captive mule deer: A spongiform encephalopathy. Journal of Wildlife Diseases 16: 89-98. ~ and __ . 1982. Spongiform encephalopathy of Rocky Mountain elk. Journal of Wildlife Diseases 18: 465-471. ~ and .; __ . 1992. Spongiform encephalopathies in Cervidae. Revue Scientifique et Technique Office International des Epizooties 11: 551-567. _~ and __ . 1993. Neuropathology of chronic wasting disease in mule deer (Odocoileus hemionus) and elk (Cervus elaphus nelsoni). Veterinary Pathology 30: 36-45. �203 Table 1. Profiles used in chronic wasting disease targeted surveillance. • Species: • Age: • Signs: mule deer white-tailed deer elk ::::18 months emaciated and abnormal behavior &lor indifference to human activity &lor increased salivation &lor tremor, stumbling, incoordination &lor difficulty or inefficiency in chewing/swallowing &lor increased drinking and urination Table 2. Tentative schedule for systematic statewide chronic wasting disease surveys of deer populations using harvest and/or road-kill samples. . CWD Status GMUs Years surveyed Endemic 7,8,9, 19, 191,20 29,91,93,94,95,951,96 1996-1999 1997-1999 "High Risk" 87, 88, 89, 90, 92 104 18,28,37,371 6, 16, 17, 161, 171 22 97, 98, 99, 100, 101 1997-2000 1996-2000 1998 1999-2001 1999 2000-2001 Other areas 66,67 83 1997-1998 1993-1994, 1998 1999 2000-2001 2001-2003 "Uncompahgre" "Southern Front Range" Random Statewide [Note: Survey targets and schedules subject to changes influenced by new data and availability of resources supporting CWD surveillance activities.] �204 Table 3. Prospective strategies for chronic wasting disease management, as well as real or potential limitations of these strategies. Strategies for limiting distribution - vaccination (no vaccine available) - preclude human-caused trans locations (mostly done) - interrupt migration corridors (fencing, local population suppression?) - change habitat (improve/degrade to influence animal distribution?) - alter migratory behavior (selective culling?) - reduce density (relationship between density & distribution untested) Strategies for reducing prevalence - vaccination (no vaccine available) - treatment (no effective therapeutic drugs available) - genetic selection (evidence of uniform susceptibility) - fertility control (maternal transmission probably a minor contribution; no practical tools) - eliminate clinical suspects (ineffective to date; intensive effort necessary) - test & slaughter (no reliable, practical live animal test) - selective culling (what criteria? mechanism?) - foster natural mortality (e.g., predation) to promote selective culling - reduce density (relationship between density & transmission untested) Table 4. Prospective management tactics that could be needed to support selected management strategies, as well as potential issues or limitations ito specific tactics. Harvest (role of public hunting in disease management?) - separate species management for mule deer and white-tailed deer - shift emphasis from recreational opportunity to population management - modify GMU/ "management area" boundaries to focus management & distribute pressure - expand or modify season structure/length - increase bag limits - bounty/incentive programs Culling (public acceptance, access, efficiency?) - ground vs. aerial vs. capture & kill options - use of agency vs. public agents for culling operations - private land issues - bounty/incentive programs Fertility control (probably not useful for reductions, but maybe to maintain) - agent availability and environmental safety - delivery, cost Increase predation (livestock and public safety issues) - reduce predator quotas (close seasons?) - predator introductions Poisoning (specificity, environmental impacts, public acceptance?) �205 Appendix B Chronic Wasting Disease in Colorado Deer and Elk: Recommendations for Experimental Management in GMU 9 M W. Miller and R. H Kahn Terrestrial Wildlife Resources, Colorado Division of Wildlife Executive Summary Chronic wasting disease (CWO) is a transmissible spongiform encephalopathy (TSE) of native deer and elk that is endemic throughout northeastern Colorado and southeastern Wyoming. Estimated infection rates range from <1-15% in deer and 1% in elk residing in northeastern Colorado game management units (GMUs). There is no precedent for attempting to manage a TSE in free-ranging wildlife, and the biological imperative for such management is not entirely clear. However, it seems most biologically responsible to assume that CWO adversely affects native deer and elk populations and consequently manage to prevent its further spread and reduce its occurrence in Colorado. The Colorado Division of Wildlife is committed to minimizing the threat that CWO poses to Colorado's native deer and elk resources. In data analysis units (DAUs) where CWO is endemic, goals of reducing the occurrence and preventing the further spread of CWO currently serve as the primary basis for setting management objectives. Based on current understanding of CWO in free-ranging deer and elk, eradication seems an extreme and unjustifiable management goal at this time. Considering the prevalence and .. distribution CWO has reached under historical management regimes, however, continuing to rely on current management approaches seems equally unjustifiable. Specific strategies and tactics for accomplishing CWO management objectives need to be identified in individual DAU plans. However, many conventional disease management options (most notably, vaccines, therapeutics, and live-animal tests for CWO) are presently unavailable; all other prospective strategies identified to reduce CWO prevalence involve some form of aggressive culling or population control. Here, we propose a management experiment in GMU 9 to evaluate the efficacy of density reduction on CWO prevalence over time. Because current season structure and bag limits are too inflexible to achieve management objectives, we recommend draft language for modifying Commission regulations to support achievement of experimental management objectives. Managing CWO represents an unprecedented. challenge to CDOW's biologists, field personnel, public service representatives, and administrators, as well as to sportsmen, landowners, and other affected publics. However, rising to meet this challenge is vastly preferable to a legacy of inaction that allows CWO to become more widespread at the expense of valuable wildlife resources in Colorado and elsewhere. �206 Background Chronic wasting disease (CWD) is a transmissible spongiform encephalopathy (fSE) of native deer (Odocoileus spp.) and elk (Cervus elaphus nelsoni) characterized by behavioral changes and progressive loss of body condition that invariably lead to the death of affected animals (Williams and Young 1992). Neither the causative agent nor its mode of transmission have been fully identified; however, CWD is suspected to be caused by a unique strain ofprion transmitted from animal-to-animal via some combination of saliva, feces, urine, or other tissues or fluids (Miller et al. 1998, 2000; Williams et al. 2000). Both sexes and all age classes show relatively uniform susceptibility (Miller et al. 2000, Williams et al. 2000). There are no valid tests currently available for diagnosing CWD in live animals, and extant postmortem tests require microscopic examination of brain tissue. There are no known treatments for CWD. Previous attempts to eradicate CWD from research facilities failed on at least 2 occasions (Williams and Young 1992, Miller et al. 1998). Although similar in some respects to other TSEs that affect domestic sheep (scrapie) and cattle (bovine spongiform encephalopathy; BSE), existing and unpublished experimental data indicate CWO cannot be naturally transmitted to domestic livestock, and that scrapie and BSE cannot be naturally transmitted to native cervids. Moreover, available data indicate that CWD does not present a threat to human health. "Chronic wasting disease" was first recognized by biologists in the 1960's as a disease syndrome of captive deer held in wildlife research facilities in Ft. Collins, CO. This disease was subsequently recognized in captive deer, and later in captive elk, in wildlife research facilities near Ft. Collins, Kremmling, and Meeker, CO and Wheatland, WY (Williams and Young 1980, 1982), as well as at in least two zoological collections (Williams and Young, 1992). Since 1996, CWD also has been diagnosed in captive elk residing in two game ranches in Saskatchewan, Canada, as well as five ranches in South Dakota, two ranches in Nebraska, two ranches in Colorado, and' one ranch each in Oklahoma and Montana. Over 100 cases of clinical CWO also have been diagnosed in free-ranging mule deer (0. hemionus), white-tailed deer (0. virginianus), and elk from northeastern Colorado and southeastern Wyoming since 1981 (Miller et al. 2000). At present, the known distribution of CWD in wild cervids appears to be limited to northeastern Colorado and southeastern Wyoming (Miller et al. 2000). Although CWO was first diagnosed in captive cervids, the original source of CWO in either captive cervids or free-ranging cervids is unknown; whether CWO in captive cervids really preceded CWD in wild cervids, or vice versa, is equally uncertain (Spraker et al. 1997). Moreover, the relationship (if any) between CWD cases in privately-owned elk industry and cases in free-ranging and research animals is unclear. In Colorado, free-ranging CWO cases in deer and elk have originated from throughout the northeastern portion of the state. Game management units (GMUs) yielding infected deer or elk include 7,8,9, 19, 191, 20,29,91,93,94,95,951, and 96. Although cases have come from 13 different GMUs~ two data analysis units (04, DI0) have yielded over .85% of the documented cases. Harvest surveys indicate GMUs 9, 191, and 19 are the most extensively infected GMUs in Colorado; estimated prevalence ranges from about 615% (Miller et al. 2000). Data from harvest surveys indicate that CWD is less prevalent in GMUs surrounding this core area, affecting about 1-4% of the deer in GMUs 7, 8, 20, 29, 91, 93, 94, 95, 951, and 96. Similarly, harvest surveys indicate <1 % of the elk in DAUs E4 and E9 are infected with CWO (Miller et al. 2000). Based on targeted surveillance data and select harvest surveys, wild deer or elk populations in other parts of Colorado and Wyoming are probably not infected with CWO (Miller et al. 2000). Similarly, surveillance data gathered to date indicate CWO has not yet spread to South Dakota, Nebraska, or Kansas (A. Jenny, USDA-APHIS-VS-NVSL, pers. comm.). For infected deer populations northeastern Colorado, overall CWO prevalence appears to be stable; however, because small increases would be difficult to detect over only a few years it is doubtful that our short-term observations accurately reflect or forecast long-term trends. In the absence of historical (20-30 years ago) prevalence data or reliable estimates of transmission rates, changes in overall incidence or distribution of CWO in northcentral Colorado DAUs are probably best projected via modeling. Based on �207 field data and results of simulation modeling, it appears that CWD was introduced into northern Larimer County over 30 years ago (Miller et al. 2000, Gross and Miller 2000) and has increased in both prevalence and distribution since that time. The significance of CWD and its impacts on native deer or elk populations have not been determined. Preliminary results of simulation modeling suggest that, sustained at >5% prevalence as observed in the most heavily infected GMUs, CWD could impact wild deer herds and lead to population declines (Miller et al. 2000, Gross and Miller 2000). Analyses of harvest and other population data have not detected evidence of population depression attributable to CWD, but existing inventory data for deer in northeastern Colorado are probably insufficient to detect such trends. In the absence of data to the contrary, it seems most prudent to assume that CWD adversely affects native deer or elk populations and manage to prevent its further spread and reduce its occurrence in Colorado. Unfortunately, there is considerable uncertainty in how to manage CWD in free-ranging wildlife, or whether such an endeavor could even be successful (Gross and Miller 2000). Because a more complete understanding of CWD is fundamental to developing comprehensive management programs, the CDOW needs to further understanding about CWD and its management through surveillance and experimental management. EXPERIMENTAL MANAGEMENT There is no precedent for attempting to manage a TSE in free-ranging wildlife. Moreover, the biological imperative for such management is not entirely clear at present. Programs for managing or eliminating TSEs of domestic livestock have proven only marginally successful, and the lack of obvious success in eradicating scrapie or BSE from endemic countries, compounded by epidemiological differences between CWD and other TSEs, make such programs rather poor models for prospective CWD management. Several fundamental features ofCWD epidemiology are understood poorly, if at all; in particular, the influences of host (e.g., population density, intraspecific vs. interspecific transmission, genetics) and environmental factors (e.g., contamination, reservoirs, weather) on CWD dynamics remain unclear. In light of the uncertainties associated with its epidemiology and ecology, we believe CWD management should be approached as an experiment, thereby allowing us to learn as we take actions intended to reduce prevalence and limit distribution. The CDOW's initial approaches for attempting to control CWD in free-ranging deer and elk were outlined in an action plan for CWD drafted in 1995 (Miller et al. 1995). In addition to calling for a comprehensive public information campaign, this plan identified tactics for limiting distribution and reducing prevalence of CWD based on contemporary knowledge of the problem. Prompt removal of affected animals and enforcement of feeding prohibitions were recommended to help reduce CWD prevalence and transmission. Policies and regulations were also recommended, and subsequently instituted, to prevent wider distribution ofCWD via human activities (e.g., translocations, rehabilitation, game ranching). The plan also identified a need for more extensive surveillance to gather data on CWD distribution and prevalence. As more data on CWD distribution and prevalence have been gathered over the last 4 years, it has become apparent that CWD is more widely distributed and more prevalent in northeastern Colorado (and elsewhere) than initially believed (Miller et al. 2000). It follows that policy, goals, and strategies for experimental CWD management in Colorado should be reevaluated in the context of new data, as well as changes in social and political attitudes toward animal TSEs. The CDOW is committed to minimizing the threat CWD poses to Colorado's native deer and elk resources. In data analysis units (DAUs) where CWD is endemic, goals of reducing the occurrence and preventing the further spread of CWD currently serve as the primary basis for setting management objectives. Based on current understanding of CWD in free-ranging deer and elk, eradication seems an extreme and unjustifiable �208 management goal at this time. Considering the prevalence and distribution CWD has reached under historical management regimes, however, continuing to rely on current management approaches seems equally unjustifiable. Prospective management strategies: A clear understanding of prospective strategies and options (or lack thereof) is a necessary prerequisite to adoption of policy, goals, and objectives for managing CWD in northeastern Colorado. Consequently, we offer the following overview as a context for the recommendations on CWD. management described here. Effective strategies for managing CWD or any other TSE in free-ranging wildlife have never been identified or evaluated. Many conventional disease management options are precluded from consideration. Most notably, vaccines, therapeutics, and live-animal tests for CWD are presently unavailable (Table 1). Contrary to experiences with domestic sheep (Dawson et al. 1998), available data indicate relatively uniform genetic susceptibility to CWD among deer (K. I. O'Rourke and M. W. Miller, unpubl. data); consequently, genetic selection would probably have no impact on prevalence even if tools for its implementation were available. Similarly, controlling reproduction alone probably would be ineffective even if practical tools were available because maternal transmission alone is unlikely to be sustaining CWD prevalence at levels currently observed in some affected deer populations (Miller et al. 2000, Gross and Miller 2000). All other prospective options for reducing CWD prevalence that have been identified to date (Table 1) involve some form of aggressive culling or population control directed at diminishing or preempting disease transmission. The magnitude and duration of such control will be driven by GMU- and DAU-specific CWD prevalence, management objectives, and population responses. Based on current understanding of CWD epidemiology (Miller et al. 1998,2000; Gross and Miller 2000), strategies that lower CWD transmission rates by reducing numbers of infected animals and point sources of environmental contamination (e.g., feeders) should be most effective in lowering prevalence. (Although the precise mechanisms of transmission have not been identified, those details will probably have little influence on the resolution ofCWD management at the population level.) General models predict that random culling should be less effective than selective culling in reducing disease prevalence (Barlow 1996, McCarty and Miller 1998, Gross and Miller 2000). However, field data and pilot work indicate that management strategies based on culling clinical suspects probably will miss a large proportion of the infected (and potentially infectious) individuals in a population. Moreover, there is no practical way to apply "test and slaughter" regimes without an antemortem diagnostic test for preclinical CWD. Accelerated culling of early clinical cases, either by humans or via predators, may help reduce CWD transmission and prevalence. Alternatively, if the probability of CWD transmission is density-dependent, then reducing deer or elk densities in endemic areas should lead to reduced CWD prevalence; to date, the hypothesis of density- . dependent disease transmission remains untested. Obvious strategies for limiting CWD distribution remain equally elusive. The mechanisms driving the spread of CWD among deer and elk populations are even less apparent than those driving disease dynamics. It is possible that CWD distribution to date has been influenced by regular and/or random movements of deer and, less commonly, elk. Previous studies (Kufeld et al. 1989, Kufeld and Bowden 1995, S. Steinert pers. comm.) have shown that proportions of both foothills and river bottom deer populations in northeastern Colorado are either migratory or mobile. The movement patterns documented in these studies provide at least a partial explanation for observed CWD distribution. It is also possible that CWD influences its own distribution by changing behaviors, range fidelity, and movement patterns of affected animals; observations of repetitivepacing in captive deer and elk affected by CWD could be manifested as extensive wandering movements of free-ranging individuals. Whether or not movement patterns or dispersal rates are density-dependent remains unclear; either way, reducing population size should also reduce the probability of affected individuals carrying CWD to new areas. (However, we do �209 recognize the possibility that at some lower threshold reduced population density may actually promote deer and elk movements, particularly during the breeding season). Experiments to manage CWD: Based on current uncertainty about how to effectively disrupt the processes that influence CWD epidemiology, prevalence, and distribution in free-ranging deer in northeastern Colorado, we have recommend initiating controlled management experiments directed toward testing one hypotheses about how best to manage CWD. Prospective testable hypotheses include: HI. CWD distribution in foothills deer populations can be diminished by selective culling and by reducing deer density in affected populations. H2. CWD transmission and prevalence in foothills deer populations can be diminished by selective culling and by reducing deer density in affected populations. We recommend using an adaptive resource management approach (Holling 1978, Walters and Holling 1990) to test candidate strategies for reducing CWD prevalence and distribution. Based on initial evaluations of these candidate strategies, larger management experiments can be developed in conjunction with the DAU planing process. Prospective management tactics: Specific tactical recommendations employed in experimental CWD management should follow from the DAU planning process. As a general consideration, however, it is likely that current season structure, bag limits, and/or administrative boundaries may be too inflexible to accommodate experimental approaches for managing CWD that may be proposed for evaluation. Many candidate tactics (Table 2) for such management will require modification or temporary suspension of Commission regulations, as well as considerable human dimensions efforts. Within the confines of the broad policy, goals, and objectives provided, we believe specific tactics are best identified by a combination of staff and field collaborators on a local level in conjunction with the DAU planning process and adopted as a package along with specific DAU plans for managing CWD. Developing comprehensive, objective-oriented DAU plans to manage CWD represents an unprecedented challenge to CDOW's biologists, field personnel, public service representatives, and administrators. However, we believe rising to meet this challenge is vastly preferable to a legacy of inaction that will allow a potentially manageable wildlife health problem to become widespread at the expense of valuable wildlife resources in Colorado and elsewhere. Experimental management of CWD in GMU 9: Within endemic portions of northeastern Colorado, CWD prevalence is highest in GMU 9. Of 174 harvested deer from GMU 9 sampled during 1996-1999,29 (16.7%) tested positive for CWD (Miller et al. 2000); because diagnostic tests currently in use underestimate true CWD prevalence by a factor of about 0.8 (Miller et al. 2000), true prevalence in GMU 9 is likely closer to 20%. Highest densities of infected deer and highest local prevalence rates occur primarily in the northwestern portion of GMU 9 and coincide with areas of highest wintering deer densities in this GMU (Fig. 1). Although sample sizes for individual �210 CWO Density (No. PoslKm2) and locations of CWO Positive Mule Deer years are small and some year-to-year Clinical Cases and Hunter Survey Samples 1997-1999 variation probably occurs, CWD prevalence N among bucks harvested in GMU 9 has shown an increasing trend over the last 3 years (15.8% in 1997, 12.2% in 1998,20% in 1999), Computer models parameterized to reflect this observed 3-year trend forecast that CWD prevalence in GMU 9 will double in the ""'NH""" ••. next 10 years in the absence of some change in Ntransmission or population dynamics (Fig. 2). 6atv~""'.o Under modest assumptions of densityIIIRO.1.0.15 dependent CWD transmission, models predict that dramatic (e.g., 50% of current levels) reductions in deer density could prevent further increases in prevalence (Fig. 2). If transmission GMJ 9 KrigedON;) Pr8talenceEstinBtes: is more strongly density-dependent than 2x2 Km ard 100 NeaJestNeigl"bors assumed, then prevalence should actually decrease in response to density reductions. Conversely, if CWD transmission is not influenced by deer density, then models forecast that prevalence will continue to rise despite density reductions (Fig. 2). A l.ocII:lonsd~lcdPosiMs tiIrttGt SUN6f O-O 25 0.025·0.OS o.oS.0.1 Po!:itiw5INo. PosMn2) i 0.15002 0.2-0.25 • Understanding the relationship (or lack thereof) between deer density and CWD prevalence trends is fundamentally important to future decisions on CWD management throughout northeastern Colorado. In order to further this understanding, we recommend an experimental reduction in deer Figure 1.Highest densities of infected deer (upper map) densities in GMU 9 to assess the effects of such and highest local prevalence rates (lower map) occur management intervention on CWD prevalence. primarily in the northwestern portion of GMU 9. These We have selected GMU 9 as a study area heavily infected areas coincide with areas of highest primarily because CWD prevalence is high wintering deer densities in GMU 9. enough to allow reliable measurement and detection of trends over time. In addition, the resident deer population in GMU 9 is relatively small (about 1000 head) and insular. Our experimental management plan calls for reducing deer density in GMU 9 by about 50% over the next 2-3 years (20002002) and holding 'deer density at this reduced level for at least 5 years (2003-2007). CWD prevalence will be monitored via sampling of harvested deer; in addition; we will monitor radiocollared deer in GMU 9 and adjacent GMUs to examine effects of lowered density on home ranges and movement patterns. Prevalence trends will be compared to trends in surrounding GMUs. In cooperation with the Wyoming Game and Fish Department, we will also compare data from GMU 9 to data from a Wyoming "hunt area" (HA 64) where current CWO prevalence trends are very similar to those in GMU 9. Because density will not be reduced in HA 64 over the next several years, this population will serve as a control for our management experiment. Virtually all deer ranges in GMU 9 are in private ownership. Consequently, we have enlisted local landowners as partners in this proposed management experiment. Based on input from GMU 9 landowners, we recommend that density reductions be accomplished primarily via extended, limited access hunting seasons; in some cases, culling may be needed to augment removals via harvest. Draft language for regulations to establish these landowner-controlled hunts follows: �211 Draft regulation for experimental chronic wasting disease management in GMU 9 Chapter 2 Article VIII #250 E. Experimental Seasons for Chronic Wasting Disease Management a. All landowners with fee title or exclusive lease to more than (20 acres) of land located in GMU 9 are eligible to receive unlimited numbers of vouchers that may be used to purchase either-sex deer licenses validfor the time period 1 November 2000 through 28 February 2001. Licenses are valid only in GMU 9. All deer harvested under this experimental season shall have the heads submitted to the Division of Wildlife for chronic wasting disease testing. All other regulations are in effect for this season, including but not limited to method of take, tagging, hunting hours and care of game meat. Experimental season licenses will be considered additional licenses. Quotas or limits on geographic distribution of vouchers will be established by CDOW Terrestrial Resources staff, in consultation with Area personnel, in order to achieve uniform density reduction throughout GMU 9. Population trends will be monitored by both aerial inventory and modeling. LITERATURE CITED Barlow, N.D. 1996. The ecology of wildlife disease control: simple models revisited. J. Appl. Ecol. 33: 303-314. Bear, G. D. 1989. Seasonal distribution and population characteristics of elk in Estes Valley, Colorado. Colorado Division of Wildlife, Fort Collins, Colorado. Special Report Number 65: 1- 14. Dawson, M. L. 1. Hoinville, B. D. Hosie, and N. Hunter. 1998. Guidance on the use ofPrP genotyping as an aid to the control of clinical scrapie. Vet. Rec. 142:623-625. Gross, J. E., and M. W. Miller. 2000. Chronic wasting disease in mule deer: A model of disease dynamics, control options, and population consequences. J. Wildl. Manage. (in review). Holling, C. S. 1978. Adaptive environmental assessment and management. John Wiley and Sons, London, England. Kufeld, R. C. and D. C. Bowden. 1995. Mule deer and white-tailed deer inhabiting eastern Colorado plains river bottoms. Colorado Division of Wildlife, Fort Collins, Colorado. Technical Publication Number 41: 1- 58. Kufeld, R. C., D. C. Bowden, and D.L. Schrupp,. 1989. Distribution and movement of female mule deer in the Rocky Mountain foothills. 1. Wildl. Manage. 53: 871-877. McCarty, C. W., and M. W. Miller 1998. Aversatile model of disease transmission applied to forecasting bovine tuberculosis dynamics in white-tailed deer populations. J. Wildl. Dis. 34: 722-730. Miller, M. W. 1999. Monitoring and managing chronic wasting disease in deer. Pages _-_ in Wildlife Research Report, Mammals Research, Federal Aid Projects, Job Progress Report, Project W-153-R-12, Work Package 3001, Task 3. Colorado Division of Wildlife, Fort Collins, Colorado, USA. (in press) Miller, M. W. 1999. Monitoring and managing chronic wasting disease in elk. Pages _-_ in Wildlife Research Report, Mammals Research, Federal Aid Projects, Job Progress Report, Project W-153-R-12, Work Package 3002, Task 3. Colorado Division of Wildlife, Fort Collins, Colorado, USA. (in press) Miller, M. W., G. G. Schoonveld, B. Smith, R. A. Spowart, and R. H. Kahn. 1995. Chronic wasting disease - proposed action plan. Unpubished report, Colorado Division of Wildlife, Fort Collins, Colorado, 16 pp. Miller, M. W., M. A. Wild, and E. S. Williams. 1998. Epidemiology of chronic wasting disease in captive Rocky Mountain elk. J. Wildl. Dis. 34:532-536. �212 Miller, M. W., E. S. Williams, C. W. McCarty, T. R. Spraker, T. J. Kreeger, C. T. Larsen, and E. T. Thome. 2000. Epizootiology of chronic wasting disease in free-ranging cervids in Colorado and Wyoming. J. Wildl. Dis. 36: (in press). Spraker, T. R., M. W. Miller, E. S. Williams, D. M. Getzy, W. 1. Adrian, G. G. Schoonveld, R. A. Spowart, K. I. O'Rourke, 1. M. Miller, and P. A. Merz. 1997. Spongiform encephalopathy in freeranging mule deer (Odocoileus hemionus), white-tailed deer (Odocoileus virginianus), and Rocky Mountain elk (Cervus elaphus nelson i) in northcentral Colorado. J. Wildl. Dis. 33:1-6. Walters, C. 1., and C. S. Holling. 1990. Large-scale management experiments and learning by doing. Ecology 71: 2060-2068. Williams, E. S.; and S. Young. 1980. Chronic Wasting disease of captive mule deer: A spongiform encephalopathy. Journal of Wildlife Diseases 16: 89-98. ____:> and __ . 1982. Spongiform encephalopathy of Rocky Mountain elk. Journal of Wildlife Diseases 18: 465-471. ____:> and __ . 1992. Spongiform encephalopathies in Cervidae. Revue Scientifique et Technique Office International des Epizooties 11: 551-567. ____:> and __ . 1993. Neuropathology of chronic wasting disease in mule deer (Odocoileus hemionus) and elk (Cervus elaphus nelsoni). Veterinary Pathology 30: 36-45. �213 Table 1. Prospective strategies for chronic wasting disease management (and real or potential limitations of these strategies). Strategies for limiting distribution - vaccination (no vaccine available) - preclude human-caused translocations (mostly done) - interrupt migration corridors (fencing, local population suppression?) - change habitat (improve/degrade to influence animal distribution?) - alter migratory behavior (selective culling?) - reduce density (relationship between density & distribution untested) Strategies for reducing prevalence - vaccination (no vaccine available) - treatment (no effective therapeutic drugs available) - genetic selection (evidence of uniform susceptibility) - fertility control (maternal transmission probably a minor contribution; no practical tools) - eliminate clinical suspects (ineffective to date; intensive effort necessary) - test & slaughter (no reliable, practical live animal test) - selective culling (what criteria? mechanism?) - foster natural mortality (e.g., predation) to promote selective culling - reduce density (relationship between density & transmission untested) Table 2. Prospective management tactics that could be needed to support selected management strategies, as well as potential issues or limitations specific tactics. to Harvest (role of public hunting in disease management?) - separate species management for mule deer and white-tailed deer - shift emphasis from recreational opportunity to population management modify GMU/ "management area" boundaries to focus management & distribute pressure - expand or modify season structurellength - increase bag limits - bounty/incentive programs Culling (public acceptance, access, efficiency?) - ground vs. aerial vs. capture & kill options - use of agency vs. public agents for culling operations - private land issues bounty/incentive programs Fertilitv control (probably not useful for reductions, but maybe to maintain) - agent availability and environmental safety - delivery, cost Increase predation (livestock and public safety issues) - reduce predator quotas (close seasons?) - predator introductions Poisoning (specificity, environmental impacts, public acceptance � https://cpw.cvlcollections.org/files/original/e6e06495c7e2a9214e255a329de730a0.pdf 2e4de18739ba3882f2c6408e487cf4f1 PDF Text Text 135 \ Colorado Division of Wildlife Wildlife Research Report Ju1y 2000 I " JOB PROGRESS REPORT State of Colorado Cost Center 3430 Project No. W-153-R-13 Mammals Research Program Deer Conservation Work Package No. ---"3'"""0..;;..0-=-1_ _ _ __ Task No. 4 Effects of Habitat Enrichment on Mule Deer Recruitment and Survival Rates Period Covered: July 1, 1999 - June 30, 2000 Authors: C. J. Bishop and G. C. White Personnel: L. H. Carpenter, D. Coven, D. J. Freddy, R. B. Gill, R. Harthan, J. SaZina, B. E. Watkins, and B. Welch ) ABSTRACT A Program Narrative {Appendix A) was developed to test the effects of habitat enrichment on mule deer recruitment and survival rates on the Uncompahgre Plateau in southwest Colorado. Post hoc analyses were completed to evaluate the effects of buck harvest on age and sex ratios (Appendix B), and to roughly estimate the total number of deer in Colorado (Appendix C). ) �136 I \ �137 EFFECTS OF HABITAT ENRICHMENT ON MULE DEER RECRUITMENT AND SURVIVAL RA TES C. J. Bishop and G. C. White P. N. OBJECTIVES 1. To conduct a one-year pilot study to assess the logistical feasibility of the proposed study and to gather preliminary data to improve the study's efficiency and experimental design. 2. To determine experimentally whether enhancing mule deer nutrition during winter and early spring by supplemental feeding increases December fawn:doe ratios and overwinter fawn survival. 3. To determine experimentally to what extent habitat treatments replicate the effect of enhanced nutrition from supplemental feeding. SEGMENT OBJECTIVES 1. Prepare a Program Narrative. 2. Plan field logistics and purchase equipment and supplies in preparation for initial field work beginning in FY 2000-01. INTRODUCTION Mule deer numbers apparently declined during the 1990 •s throughout much of the West, and have clearly decreased since the peak population levels documented in the 1940's-60's (Gill et al. 1999, Unsworth et al. 1999). Biologists and sportsmen alike have concerns as to what factors may be responsible for declining population trends. Although previous and current research indicates that multiple interacting factors are responsible, habitat and predation have received the focus of attention. A number of studies have evaluated whether predator control increases deer survival, yet results are highly variable (Connolly 1981, Ballard et al. 1999). Together, predator control studies with adequate rigor indicate that predation effects on mule deer are variable as a result of time-specific and site-specific factors. Studies which have demonstrated deer population responses to predator control treatments have failed to determine whether predation is ultimately more limiting than habitat. Numerous research studies have evaluted mule deer habitat quality, but virtually no studies have documented population responses to habitat improvements. In many areas where declining deer numbers are of concern, predation is common yet habitat quality appears to have declined. The question remains as to whether predation or habitat is more limiting to mule deer in these situations, and whether habitat quality can be improved for the benefit of deer. We designed a field experiment to measure deer population responses to habitat enrichment treatments. We will conduct the study on the Uncompahgre Plateau, where several predator species are present in abundant numbers. Predator numbers will not be manipulated in any way. Habitat enrichment treatments will consist of supplemental feed provided to deer during the winter. If December fawn:doe ratios and overwinter fawn survival improve as a direct result o(the supplemental feed, then we can presume that deer nutrition is ultimately more limiting than predation. The field experiment also incorporates habitat manipulation treatments, which will consist of prescribed fire or mechanical techniques to set back succession ofpinyon-juniper habitat in an effort to improve the vigor and quality of winter habitat for mule deer. Deer population responses will be measured in relation to the habitat manipulations in the same manner as the supplemental feed. Thus, the experiment allows us to detennine whether nutritional quality �138 of habitat is ultimately more limiting than predation in a late-seral pinyon-juniper/sagebrush landscape, and if so, whether habitat can be effectively improved for mule deer. There have been several challenges to implementing this study. First, using supplemental feed in a field study sends a contradictory message to the public given the Division of Wildlife's policy on winter feeding. Second, the study is very expensive and therefore requires considerable Division support to implement. Third, logistical concerns raise the issue of whether the study can be carried out appropriately, such that response variables are representative of treatment differences and measured without bias yet with adequate precision. Finally, some personnel have questioned the need for such a study, particularly given the expense. In result, we gave a number of presentations explaining the study in detail to generate widespread support throughout the Division. We also decided to conduct a pilot study in the first year to address logistical concerns and to allow more time to generate necessary funding. MATERIALS AND METHODS Program Narrative Development We developed a Program Narrative (Appendix A), which addresses both the I-year pilot study as well as the complete 4-6 year study. The P. N. has not been peer-reviewed at this point, but will be peer-reviewed in early FY 2000-01 prior to the pilot study. The P. N. will be modified as necessary using results from the pilot study, and once again peer-reviewed assuming there are significant changes. We based the P. N. on an initial study proposal submitted by Gary White and Len Carpenter. We conducted a thorough literature review and discussed study methodologies with numerous Division personnel to refine field methods. We spent time in the field with the local Area Biologist, Bruce Watkins, and District Wildlife Manager, Dale Coven, assessing the Uncompahgre Plateau for potential treatment and control areas. We worked closely with the Uncompahgre Ecosystem Restoration Project (UERP) committee to select treatment/control sites in the context of proposed habitat manipulation areas. Uncompahgre Ecosystem Restoration Project The Uncompahgre Ecosystem Restoration Project (UERP) was initiated by the Division of Wildlife after former Director John Mumma allocated $500,000 of capital construction funds to habitat improvements for the benefit of mule deer. UERP is comprised of individuals from DOW, U.S. Forest Service, Bureau of Land Management (BLM), and the Colorado West Public Lands Partnership (PLP). Since the initial $500,000, more funds have been obtained for UERP by the other agencies and PLP. Mike McLain (AWM) and Bruce Watkins (Area Biologist) have represented DOW's interests in UERP from the onset, while Jim Gamer (Habitat Biologist) and I joined the effort this past winter. UERP's mission is to improve ecosystem health on the Uncompahgre Plateau through an integrated effort of the various agencies and groups involved. A primary goal is to increase species diversity, age diversity, and the quality of habitat communities on the Plateau in order to improve the distribution and quality of mule deer winter range. To accomplish this goal, UERP is using a landscape approach to treat habitat using various techniques, primarily prescribed fire and mechanical treatments. All treatments will be re-seeded with shrub-forb-grass mixtures in an attempt to prevent invasive weeds from establishing and to increase forage diversity for mule deer. Deer population responses on one of these habitat treatment areas will be intensively monitored as part of our study's experimental design. The various other habitat treatments will be implemented in a pattern of treatment and control areas such that long-term deer responses can be monitored using the current mule deer population monitoring system on the Uncompahgre Plateau. UERP is currently in the process of hiring a technical coordinator and a policy/education coordinator to facilitate project implementation. The technical coordinator will assist with project design, implementation, and evaluation; provide technical expertise and budget oversight; assist with grant writing and production �139 of project reports; and ensure that proposed projects comply with state and federal regulations. The policy/education coordinator will provide information concerning the project and generate support from government agencies, stakeholders, and the general public. In essence, the policy/education coordinator will serve a classic "information and education" role for UERP. Once the coordinators are hired, we will begin identifying specific treatment sites and proceed with actual habitat treatments. UERP's current timeline should enable some habitat manipulations to be completed in 2-3 years as needed for our habitat enrichment study. NEPA Requirements Since much of the field work for our habitat enrichment study will be conducted on BLM lands, we are subject to the requirements of the National Environmental Policy Act (NEPA). There are 2 types of experimental treatments in our study: (l) supplemental feeding and (2) habitat manipulations. We submitted our P. N. along with a detailed description of the supplemental feeding protocols to the Uncompahgre Field Office of the BLM. They determined that the feeding portion of our study meets the Categorical Exclusion requirement of NEPA, thereby granting approval to proceed without further review. As explained above, all habitat manipulations will be performed by UERP, which assumes the responsibility for obtaining NEPA approval and any other regulatory clearances. Presentations We presented our habitat enrichment study to both Division personnel and other groups throughout the year. For Division personnel, our goal was to explain the study and address a variety of concerns that had been raised. It was our intention to gain support for the study from the various leaders in the Wildlife Programs Branch, as well as managers and biologists in the West Region arid particularly those in Area 18 . .. ·For non-Division groups, we discussed the various issues affecting mule deer populations, and described the habitat enrichment study in light of how the Division of Wildlife is attempting to address mule deer concerns. We gave presentations at the following meetings, workshops, etc.: ♦ DWM Training Session (for incoming DWMs) - 1/5/2000, Denver • West Section Terrestrial Biologist Staff Meeting- 2/7/2000, Grand Junction ♦ CSU Students/Faculty, Wildlife Field Studies Class Trip - 3/11/2000, Uncompahgre Plateau ♦ Wildlife Programs Branch Meeting- 3/14/2000, Denver • Hunter Education Instructors, Big Game Workshop-4/15/2000, Montrose • Hunter Education Instructors, Masters Workshop- 5/6/2000, Winter Parle ♦ Colorado Elementary and Secondary Education Teachers, Teacher Education Program 6/23/2000, Ridgway State Park/Uncompahgre Plateau ♦ Area 18 Staff Meeting- 7/27/2000, Lone Cone Cabin The following presentation was also given, but was unrelated to the habitat enrichment study: ♦ Large Mammal Capture Techniques and Radio-Telemetry, CSU Wildlife Management Short Course - 3/29/2000, Fort Collins Other Activities Mule Deer Legislative Report Along with Mammals Research Leader, Bruce Gill, and other members of the research staff, I assisted in the writing, editing, and publication of "Declining Mule Deer Populations in Colorado: Reasons and Responses". The report was written for and submitted to the Colorado Legislature at their request. The report was distributed widely via hard copies and the internet. I was responsible for responding to a number of comments the Division received concerning the report. I presented the infonnation contained in the report in several of the presentations listed above. �140 Mule Deer Analyses I completed several analyses/data summaries of statewide mule deer population data. I evaluated the effects of limited buck harvest on mule deer sex and age ratios using data from quality (limited-harvest) Game Management Units (GMUs). I also developed a rough estimate of total deer numbers in Colorado by extrapolating deer density estimates from quadrat surveys to the total amount of delineated winter range in Colorado. In doing so, I summarized the square miles of delineated mule deer winter range by GMU and DAU throughout the state. These reports were distributed widely within the Division as a resource for management biologists. I also attempted to "back-model" statewide mule deer numbers to obtain an approximate estimate of Colorado's deer population in the 1950's. This daunting task has not been completed, although considerable effort was put forth. Different models were developed by changing assumptions; however, considerable refinement would be necessary before releasing any of the models. While completing this task, I compiled monthly weather data from approximately 80 Colorado weather stations from 1940-1999. Data were obtained from the W estem Regional Climate Center (www.wrcc.dri.edu/index.html). These spreadsheets, which contain various statewide summaries, are available. RESULTS AND DISCUSSION Habitat Enrichment Study The various presentations we gave were very effective in gaining greater support for the habitat enrichment study, particularly with Western slope managers and biologists. By openly discussing the major concerns, and deciding to pursue a pilot study in the first year, most DOW personnel are generally supportive of the study. Various personnel in the Montrose area have offered their assistance, and we have developed a good, cooperative working relationship with Bruce Watkins, the Area Biologist. Presentations given to non-Division groups were received very well. The different groups were supportive and interested in our proposed study as well as the Division's overall efforts to manage mule deer in light of declining numbers. Our presentations and other "I&E" efforts were necessary to create a favorable climate prior to initiating the study. It is important to note, however, that no presentations were given to the more outspoken critics of DOW. Budget concerns have subsided given that we have submitted requests for outside grants to fund a portion of the study, and because additional dollars may become available.in the research section during the next 2 or 3 years. Largely as a result ofUERP, a good relationship has been developed with the Uncompahgre Field Office of the BLM as well as Forest Service personnel. The BLM has offered to store the supplemental feed in their compound, and have been very cooperative in making the field work possible. They reviewed our study proposal to ensure compliance with NEPA, and spoke with grazing permitees in the area to inform them of the proposed study. These proactive efforts should prevent or at least alleviate any problems when field work begins. We selected several tentative treatment/control sites (experimental units) to be used for the study. We initially planned to identify a number of potential experimental units, and then randomly select which would be used in the study and which would be treatments and controls. However, given the logistical challenges of this study and the need to maintain some level of homogeneity among experimental units, random allocation is not realistic. In particular, feed treatment areas must be selected carefully to avoid feeding more deer than we can afford, and to avoid feeding significant numbers of elk. We are currently planning on distributing supplemental feed in the following 2 areas: (1) The Colona Tract of the Billy Creek State Wildlife Area (2) BLM Land adjacent to Shavano Valley as defined by the following: �141 Within Dry Creek Basin Quadrangle (USGS 7.5 Minute), will primarily include Section 6 in T. 48 N.R. 10 W. and Section 1 in T. 48 N.-R. 11 W. (38°26'00"-38°27'30" N Lat., 108°00'00"-108°02'30" W Long.). However, may also include Section 7 in T. 48 N .-R. 10 W. and Sections 2, 10, 11, 12, 13, 14, 15 in T. 48 N.-R. 11 W., depending on deer distribution and movements. This overall area roughly includes 38°25'00" - 38°27'30" Latitude and 108°00'00" - 108°04'30" Longitude. We are currently planning to use one of the following areas as a control area throughout the study (they are too close together to both be used): (1) Sims Mesa as defined by the following: Within Government Springs Quadrangle (USGS 7.5 Minute), will approximately include Sections 31, 32 in T. 48 N.-R. 9 W., and Sections 5, 6, 7, 8, 17, 18 in T. 47 N.-R. 9 W. (~38°19'30" - 38°22'30" N Lat. and 107°52'30" - 107°54'00" W Long.) (2) Government Springs as defined by the following: Within Colona and Government Springs Quadrangles (USGS 7.5 Minute), will approximately include Sections 9, 10, 15, 16, 17, 20, 21, 22 in T. 47 N.-R. 9 W. (~38°18'30" - 38°21 '00" N Lat. and 107°50'00" - 107°53'00" W Long.) There is little else to report concerning the study, other than we are in the process of purchasing supplies and equipment, and preparing for field work. We have purchased and received 65 radio-collars to be used in the pilot study, along with 15 additional collars from the Uncompahgre population monitoring work that will be used in conjunction with the study. We have purchased radio-telemetry equipment and are in the process of planning storage for and purchasing the supplemental feed. We are also working on drop-nets that will be used along with helicopter netguns to capture deer. Again, the Program Narrative explaining our detailed study plan is in Appendix A. Mule Deer Report The legislative report, "Declining Mule Deer Populations in Colorado: Reasons and Responses", has proven to be a beneficial document for explaining the complex array of factors currently affecting mule deer populations. During our presentations at 2 Hunter Education Workshops this spring, we presented and provided copies of the report to all instructors in attendance. We received feedback from these instructors as well as various other constituents responding to the report via email. The report is apparently upsetting to some individuals who strongly believe predator control is the only solution for reversing declining deer numbers, but for most others the report has provided a heightened awareness of issues surrounding mule deer management. Aside from predation, buck harvest is the other factor generating considerable debate. It would appear that certain sportsmen evaluate the fitness of deer populations in terms of trophy buck opportunities while hunting. This perception confuses objective evaluations of the effects oflow buck:doe ratios on declining deer populations. Clearly, low buck numbers impact hunter success rates, particularly for trophy bucks, yet existing data does not suggest that buck:doe ratios have had a significant impact on fawn production and recruitment. Harvest regulations represent a trade-off between quality hunting and hunting opportunity. However, currently there is no scientific basis for restricting buck harvest solely for the purpose of increasing fawn:doe ratios. In fact, if habitat is limiting many of our deer herds, it could have the opposite effect on fawn recruitment. Mule Deer Data Analyses Limited buck harvest in quality ,GMUs has resulted in modest increases in buck:doe ratios but has had no effect on fawn:doe ratios. The analysis and results are included in Appendix B. By extrapolating observed deer density estimates from quadrat surveys to the total amount of winter range in Colorado, I estimated approximately 575,000 - 580,000 deer in Colorado, which is very similar to the current estimate of deer numbers based on DAU population models. The analysis and results are located in Appendix C. �142 LITERATURE CITED Ballard, W. B, D. Lutz, T. W. Keegan, L. H. Carpenter, and J.C. deVos, Jr. 1999. Deer-predator relationships: a review of recent North American studies with emphasis on mule and black-tailed deer. Unpublished manuscript. Connolly, G. E. 1981. Limiting factors and population regulation. -Pages 245-285 in 0. C. Wallmo, editor. Mule and black-tailed deer of North America. University of Nebraska Press, Lincoln, Nebraska, USA. Gill, R. B., T. D. I. Beck, C. J. Bishop, D. J. Freddy, N. T. Hobbs, R.H. Kahn, M. W. Miller, T. M. Pojar, and G. C. White. 1999. Declining mule deer populations in Colorado: reasons and responses. A report to the Colorado Legislature. Colorado Division of Wildlife, Denver, Colorado, USA. Unsworth, J. W., D. F. Pac, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. Prepared by _ _ _ _ _ _ _ _ __ Chad Bishop Mammals Researcher �143 APPENDIX A PROGRAM NARRATIVE FY 2000-01 - 2006-07 State of -----=C_,._ol=o=rad=o_ _ _ __ Project No. _ _ _ _W-'-'--~15:;..:3a...-=R=--------Work Package No. ---"3-"0"""0-=--1_ _ _ __ Study No. _ _ _ _ _ _4 ~ - - - - - Cost Center 3430 Mammals Research Program Deer Conservation Effects of Habitat Enrichment on Mule Deer Recruitment and Survival Rates A. NEED Background Mule deer (Odocoileus hemionus hemionus) are a popular wildlife species with the public throughout their range. Sportsmen place high value on mule deer as a harvestable big game species, and outdoor enthusiasts of all kinds enjoy watching and photographing deer. The Colorado Division of Wildlife, along with other state wildlife agencies in the West, depend on revenue generated from mule deer license sales. Given such popularity and economic importance, an objective of western wildlife agencies has been to maintain healthy, productive deer populations. Unfortunately, mule deer numbers fluctuate widely, primarily in response to climatic and habitat conditions (Gill et al. 1999). Wildlife managers have had little success attempting to maintain stable populations over a long time period. Mule deer numbers peaked in the middle part of the 20th century, which for better or worse, set the standard that many people use to evaluate the status of current deer populations. Since the mid to late 1900's, deer numbers have generally declined to the present. In particular, two widespread declines have been documented; one occurred during the 1960 's through mid-1970 's (Workman and Low 1976), and the second occurred during the 1990's (Unsworth et al. 1999). The recent decline in mule deer populations has caused widespread concern as to the causative factors. Although reasons for declining populations are not clear and have been intensely debated, research indicates multiple interacting factors are responsible. These factors include habitat change, weather, urban development, predation, disease, competition with elk, among others (Gill et al. 1999). Ironically, many of these same factors were raised as reasons for the decline during the 1970's crisis. Great strides have been made during the past several decades to better understand the various aspects of mule deer ecology and population performance. The challenge facing managers is that there are no simple management solutions for increasing deer populations. In particular, harvest manipulations alone are rarely sufficient to direct population levels. By and large, sportsmen have suggested that predation is the major cause of declining deer numbers while biologists have implicated decreasing habitat quality as the primary cause. Management actions addressing either factor will be expensive; thus, high quality data should be obtained before any such management program is initiated. Predation Predator control efforts have been conducted on numerous occasions to evaluate whether predators may be limiting mule deer recruitment, and if so, whether predator control is a viable management strategy. Coyotes have been the primary target of such studies. Connolly (1981) summarized 12 studies assessing whether coyotes negatively impacted deer populations, which when taken together, did not provide a definitive answer as to whether coyotes are a major limiting factor of mule deer. Some of these studies found no benefit to controlling coyotes, and suggested that habitat quality was ultimately the limiting factor (Murie 1940, Leopold et al. 1951, Robinette et al. 1977). Other studies �144 were largely inconclusive (Smith and LeCount 1976, Robinette et al. 1977). The remaining studies found coyote control to be potentially beneficial (Hom 1941, Brown 1961, McMichael 1970, Knowles 1976, Schladweiler 1976, Austin et al. 1977, Robinette et al. 1977, Trainer et al. 1978, Lemos et al. 1978), but given lack of proper experimental controls in many cases, the results were largely speculative. Further, coyotes were primarily controlled by poisoning, which is no longer a possible management strategy for reducing coyote numbers. Trainer et al. (1978) and Lemos et al. (1978) suggested that coyotes were responsible for poor fawn survival in Oregon, but questioned whether aerial shooting of coyotes could effectively reduce coyote densities. Since Connolly's (1981) review of predation studies, several more predator control studies have been conducted. Hartmann et al. ( 1992) found no benefit to coyote control in a deer population in northwest Colorado, and provided strong experimental evidence that coyote predation was almost entirely compensatory. In response to the recent deer decline, several more predator control studies were initiated in various western states. The Idaho Department of Fish and Game and Montana Fish, Wildlife, and Parks each began independent predator control experiments in 1997 (Gill et al. 1999). Preliminary results from both studies indicate very little benefit from predator control, which has been primarily aimed at coyotes. To this point in each study, cost/benefit analyses render coyote control as an ineffective strategy for improving fawn recruitment. The Utah Division of Wildlife Resources implemented a coyote control management plan in 1997 (Gill et al. 1999). Coyotes were controlled at varying intensities in 15 herd units during 1997 and 1998. Winter fawn:doe ratios were higher in 1997 and 1998 than in 1995 and 1996 when coyotes were not controlled. However, this increase could not be attributed to the coyote control efforts. The experimental design was flawed, and fawn:doe ratios increased in Utah regardless of whether coyotes were controlled. In New Mexico, biologists from the Jicarilla Apache Reservation suggested that mule deer numbers increased from 1991 through 1999 as a result of predator control (Gill et al. 1999). However, the experimental design was flawed as well as the methods used to estimate deer numbers. Deer population counts were related only to number of hours flown while counting deer, and winter fawn:doe ratios did not improve despite coyote control. Connolly ( 1981) drew several conclusions regarding predator control management which are briefly summarized here: 1) careful localized study is required to determine whether observed predation is impacting deer numbers; 2) coyote and/or lion predation has not been documented as the principle cause of a mule deer population decline; 3) many predation studies may not have effectively controlled predators; 4) predator control is expensive yet cost/benefit analyses are lacking; 5) predator control may benefit deer populations that are below carrying capacity, but will not reverse deer declines that are a result of habitat deterioration. We drew several other reasonable conclusions from both past and present predator control studies: 1) predator control studies must incorporate experimental rigor in order to obtain useful results; 2) under circumstances when coyotes may be limiting deer populations, it appears very difficult and potentially cost prohibitive to effectively lower coyote densities on a management scale with currently available control techniques; 3) predator control will be ineffective in mule deer populations that are at or near carrying capacity (limited by habitat quality), where fawn survival is ultimately a function of density-dependence and not predation; 4) in predator control experiments, we typically cannot determine with much certainty whether habitat quality is ultimately limiting fawn recruitment, even when the predator control treatments are ineffective. Habitat Quality The importance of habitat quality and quantity for sustaining mule deer populations is well understood. In fact, various studies have provided considerable information regarding the components of quality habitat for deer; these studies include habitat selection (Ordway and Krausman 1986, Carson and Peek 1987, Kufeld et al. 1988, Bodurtha et al. 1989, Thomas and Irby 1991,"Armleder et al. 1994), diet composition/selection (Wallmo et al. 1973, Deschamp et al. 1979, Hartmann 1983, Campbell and �145 Johnson 1983, Gill et al. 1983, Kasworm et al. 1984, Sowell et al. 1985, Austin et al. 1994), nutrient analyses of forage (Bissell and Strong 1955, Smith 1957, Trout and Thiessen 1973, Urness et al. 1977, Tueller 1979, Bartmann 1983, Welch 1989, Austin et al. 1994), foraging preference theory (Nudds 1980, Hanley 1982, White 1983, Spalinger and Hobbs 1992, Parker et al. 1996, Hanley 1997), and energetics (Freddy 1984, Parker et al. 1984, Wickstrom et al. 1984, Hobbs 1989). It is generally accepted that most wildlife populations are limited by habitat in terms of carrying capacity, which incorporates all of the above components. At least theoretically, a deer population at carrying capacity cannot expand beyond its current siz.e, which makes carrying capacity a very useful parameter. However, given the complex, non-static relationships among biotic, physical, and edaphic factors of the environment, it can be very difficult and time consuming to quantify and track carrying capacity. In fact, carrying capacity of winter range continually changes as a function of the duration and severity of winter (Wallmo et al. 1977). Based on the concept of carrying capacity and its limiting factors, Wallmo et al. (1977) provided a model for evaluating deer habitat by quantifying the quality and quantity of forage in relation to mule deer requirements. Using our accumulated knowledge of deer habitat to qualitatively evaluate current Colorado deer ranges, it appears that habitat quality, and thus carrying capacity, has declined in some of these ranges as a result of habitat changes over time. Succession and disturbance are the principle components of habitat change. Generally speaking, mule deer are better adapted to earlier-seral habitats, particularly those that are part of a larger habitat complex in various successional stages (Bailey 1984). Disturbance-free communities will eventually succeed to later-seral or climax-type habitats, which have a lower carrying capacity for deer because of declining forage diversity and vigor. Thus, periodic disturbance may often be necessary for maintaining healthy deer populations. Fire suppression is believed to have fostered the expansion of late-seral, decadent habitats throughout many areas in the West. A particular concern has been the encroachment ofpinyon (Pinus spp.)-juniper (Juniperus spp.) into sagebrush (Artemisia spp.) habitats (Blackbum and Tueller 1970, Burkhardt and Tisdale 1976, Miller et al. 1999). Controlled bums and mechanical treatments have been used to restore various habitats to early-seral conditions (Bunting 1996, Riggs et al. 1996, Erskine and Goodrich 1999, Sorensen 1999, Stevens 1999). Following such habitat disturbances, deer have been found to prefer the disturbed habitat over the undisturbed habitat (Keay and Peek 1980, Farmer 1995). Prescribed fire in northern California improved black-tailed deer reproduction, which was considerably higher than in untreated habitat (Taber 1956, Taber and Dasmann 1958). Hobbs and Spowart (1984) increased the·nutrition in winter diets of mule deer following a prescribed fire in Colorado, which enabled greater foraging selectivity by deer. However, reintroducing fire into later-seral communities can potentially have negative impacts, given the increasing presence of exotic weeds. Some rangeland fires have perpetuated invasive weeds and substantially reduced browse species as a result (Loft and Menke 1990,"Clements and Young 1997, Bishop 1998). Thus, habitat treatments must be carefully planned and followed with reseeding to achieve desired objectives (Sheley et al. 1996, Goodrich and Rooks 1999, Svejcar 1999). Qualitative evidence suggests that habitat has declined in overall quality and quantity at the same time deer numbers have declined throughout the West. Clements and Young (1997) linked a declining deer population to habitat changes associated with grazing management, fire frequency, and invasion of exotic annual grasses. Similarly, Gill et al. (1999) discussed declining deer numbers in Colorado in relation to residential development, fire suppression, excessive herbivory from livestock and ungulates, and invasion of exotic plants. Various habitats have inevitably declined in productivity as a result of these factors, yet the impacts on deer populations have not been quantified. To better understand the implications of habitat quality and carrying capacity, we need to quantify how deer populations respond to changes in habitat. For management purposes, we need to improve our ability to actively manage habitat for the benefit of mule deer (Boyd et al. 1986). �146 Habitat vs. Predation There are two overwhelming problems in current deer management regarding predators and habitats. First, state agencies are becoming increasingly focused on predator control as a management solution, typically as a result of public and political pressure, without regard to site-specific factors affecting various deer populations or feasibility of predator control. Second, given the focus on predator control, research and management studies have not addressed whether improving habitat quality is a viable management solution for increasing deer numbers. The first problem may lead to unjustified and poorly planned predator control management at unrealistically large scales, as was done by the Utah Division of Wildlife Resources, which failed to effectively reduce predator numbers and therefore wasted valuable management dollars. Low fawn survival and declines in total deer numbers are not justification for predator control unless the deer population(s) are below habitat carrying capacity and predation has been identified as a major limiting factor (Connolly 1981, Ballard et al. 1999). If predator control is justified, it will not be beneficial for deer unless predator densities are sufficiently reduced, which has only been achieved with carefully planned control efforts over relatively small areas (Ballard et al. 1999). The bottom line is that experimental predator control studies cast a shadow of doubt on the utility of predator control as a large-scale management strategy for reversing declining trends in deer numbers and fawn recruitment. The question remains as to whether or not habitat quality is more of a limiting factor than predation in many of our deer herds. This question has rarely been rigorously addressed in past research and needs more attention given the current debate surrounding wildlife agencies across the West. The prevailing political tide is focused on predation as the overriding factor, yet there is no evidence to support that predators ar~ having an impact on Colorado's deer herds. We must determine whether habitat or predation is the more limiting factor before initiating a large-scale deer management program. The other question is whether fawn survival and recruitment can be increased by enhancing habitat quality in areas where habitat is perceived to be limiting. Much attention has been focused on predator control, which for numerous reasons previously discussed, appears to have limited applicability as a large-scale management strategy. However, virtually no studies have adequately assessed whether we can effectively improve habitat quality for deer to increase fawn recruitment and ultimately population productivity. lbis question has significant management implications for mule deer. Many biologists believe that the overall drop in deer numbers from the middle of the 20th century to the present is a result of declines in both habitat quality and quantity. Although predation may have slowed population recovery at various times and places, deer numbers are ultimately limited by habitat (Connolly 1981). It follows that improvements to habitat quality will provide a much greater long-term benefit to mule deer populations. Also, habitat improvements are self-sustaining for many years following treatment whereas predator control must be continually applied on an annual basis to have a long-term effect. Our proposed study will assess whether nutrition is ultimately more limiting than predation in a deer population with currently low fawn recruitment, and whether habitats can be manipulated to increase fawn survival and recruitment. The greatest management concern over mule deer is focused on herds below population objectives with low fawn recruitment. It is therefore sensible to select one of these herds to evaluate whether habitat/nutrition or predation is more limiting to the deer population. If habitat quality is identified as the ultimate problem, we must then determine whether we can improve the deer herd by manipulating habitat. Although various habitat complexes have been manipulated to benefit wildlife, deer responses have not been appropriately measured to determine if the treated habitat actually improved deer populations. Assessments of treated habitat have traditionally comprised arialyses of pellet-group counts, habitat selection, or forage selection pre- and post-treatment or between adjacent treated and untreated areas (Wallmo 1969, Wallmo et al. 1972, Keay and Peek 1980, Hobbs and Spowart 1984, Farmer 1995). These studies establish that deer used the treated habitat more than untreated habitat, or in some cases demonstrate that diet quality was greater in the treated �147 habitat, but they do not determine whether fawn survival or recruitment actually increased. Given the expense of treating habitat, we must ensure that such treatments will improve deer productivity to justify the cost. Mule deer management decisions with a biological basis incorporate knowledge as to what factors are limiting a·deer herd. Management decisions with a political basis are typically made without regard to biological considerations. The question is whether future deer management decisions will be based on political desires or biological data. The former may or may not benefit deer populations whereas the latter will provide a basis for increasing mule deer numbers as well as sport harvest. Although we must incorporate political decisions into our management programs, the Division of Wildlife will ultimately be held responsible for the future status of mule deer populations. Our only means of accountability is to ensure that our management decisions have scientific merit. This study allows the opportunity for the Division to obtain information necessary for making biologically justifiable management decisions for mule deer. B. OBJECTIVES 1. To conduct a one-year pilot study to assess the logistical feasibility of the proposed study herein and to gather preliminary data to improve the study's efficiency and experimental design. 2. To determine experimentally whether enhancing mule deer nutrition during winter and early spring by supplemental feeding increases December fawn:doe ratios and overwinter fawn survival. 3. To determine experimentally to what extent habitat treatments replicate the effect of enhanced nutrition from supplemental feeding. Null Hypotheses: a. December fawn:doe ratios of deer groups that receive supplemental feed are not different from fawn:doe ratios of deer groups that do not receive supplemental-feed. b. Overwinter survival of fawns that receive supplemental feed is not different from survival of fawns that do not receive supplemental feed. c. December fawn:doe ratios of deer groups that utiliz.e habitat treatment areas are not different from fawn:doe ratios of deer groups that occupy untreated (control) areas. d. Overwinter survival of fawns that utilize habitat treatment areas is not different from survival of fawns that occupy untreated (control) areas. E. EXPECTED RESULTS There is an urgent need for information regarding the causative factors limiting mule deer populations that are currently below DAU (data analysis unit) objectives. This study will evaluate whether habitat or predation is the more limiting factor to fawn survival and recruitment on the Uncompahgre Plateau (DAU D-19), where December fawn:doe ratios have declined over the past 15 years. We will accomplish this by monitoring deer responses in relation to supplemental feed, which will be provided as a nutrition enhancement. We know that predators are killing fawns on the Uncompahgre Plateau (Pojar 2000, B. E. Watkins unpublished data). If this observed predation represents additive mortality, then the supplemental feed should not improve December fawn recruitment or overwinter fawn survival. In other words, it would suggest that predators are killing fawns regardless of their nutritional state. On the other hand, if the supplemental feed causes a beneficial response in the deer population, then predation is largely compensatory and not the main underlying problem. So if fawn survival and recruitment increases as a result of enhanced nutrition, without any manipulation of predator populations, we can assume that habitat is ultimately more limiting than predators. �148 If nutrition is found to be a critical problem, we need to evaluate management approaches for improving habitat quality. Given available information, we do not know whether habitat treatments benefit mule deer populations, even when they're designed to do so. This study will provide an experimental assessment of the effectiveness of habitat manipulations for improving deer populations. Results from the study will provide a quantitative basis for deciding which habitats to treat and how to treat them, by monitoring deer population parameters in response to various treatment techniques. The Division of Wildlife needs this information to plan management strategies for improving population productivity in Colorado deer herds. F. APPROACH 1. Pilot Study The primary purpose of the pilot study is to address a variety of logistical concerns that could impede both the quality and intent of the proposed study. If the pilot study runs smoothly or with only minor problems, then any necessary revisions will be made to the Program Narrative and we will pursue the actual study in the following year. On the other hand, if the pilot study reveals acute problems in proposed methodology that would prevent us from adequately testing our null hypotheses, then we will reevaluate the full-scale study. If possible, the Program Narrative will be modified to address these problems without compromising our experimental approach. Otherwise, a new experimental approach will be developed to address the same objectives, in which case we would not be able to pursue a full-scale study in the following fiscal year. The one-year pilot study will be very similar to the first year of the complete study described below, except that it will only comprise two experimental units and one response variable. The two experimental units will consist of a supplemental feed treatment area and a control area (Fig. 1). The single response variable will be December fawn:doe ratios, which will be measured the December following the pilot winter's treatments. We will capture and radio-collar 40 adult female mule deer in both the feed treatment area and the control area. This Program Narrative addresses the full-scale experiment, with specific references to the pilot study as necessary. 2. Experimental Approach a. Experimental Units Four areas on the Uncompahgre Plateau will be selected to create 4 experimental units (A-D) (Fig. 1). Treatments will be randomly assigned to the experimental units. The following criteria will be used to select experimental units: 1.) Deer densities (~50-80 deer/mi2): areas will be selected where deer densities are sufficient to meet sample size requirements within the experimental unit, while simultaneously selecting areas that will only require feeding less than ~500 animals during a normal winter 2.) Buffer zones: areas will be selected such that experimental units are separated by several miles of non-treatment area (buffers) to prevent deer from occupying more than one experimental unit 3.) Similarity: areas will be selected that comprise relatively similar habitat complexes and deer densities that are representative of the overall area 4.) Elk populations: areas will be selected that will minimize the number of elk present during normal winters Units B and C will be used alone during the pilot study, with Unit B receiving the supplemental feed treatment. In the full-scale study, Unit A will serve as a control area, that will not receive �149 any type of treatment throughout the study. Units Band C will be used to create a cross-over design with supplemental feeding. Deer will be fed on Unit B during the first 2 winters of the study, but not fed during winters 3 and 4. Deer will not be fed on Unit C during winters 1 and 2, but will be fed during years 3 and 4. Unit D will receive a habitat manipulation, which will consist of fire or mechanical treatment, or possibly a combination of both. Figure 1. Schematic representation of experimental units and their associated treatments. The pilot study will incorporate a supplemental feed treatment area and a control area. Supplemental feeding cross-over design will end after 4 years (not including the pilot study); monitoring in the habitat manipulation experimental unit and paired control area will likely continue for 2 additional years. Experimental units will be similar sized when initially selected. The mean deer density is approximately 50-80 deer/mi2 on the portion of the Uncompahgre Plateau where the study will be conducted. On average, 5 mi2 should contain anywhere from 250 to 400 deer, which is probably the minimum necessary to capture 40 does and 40 fawns from separate groups. Thus we will attempt to select experimental units that are at least 5 mi2 in size. The actual size of experimental units will ~ary, however, in relation to a particular treatment. Experimental units receiving the supplemental feed will need to remain relatively small to successfully feed all collared animals without feeding more than a total of ~500 deer. Maintaining the collared deer in a relatively small experimental unit should not be a problem because they will likely have smaller home ranges as a result of the feed. The habitat manipulation experimental unit will need to be larger to successfully treat enough habitat in a mosaic pattern where untreated habitat tracts remain within the unit. We will of course have to carefully monitor deer to ensure that they spend considerable time in the treated habitat. Assuming the manipulated habitat confers an advantage over adjacent untreated habitats, deer will likely demonstrate strong treatment site fidelity to capitalize on the more nutritious forage. Control areas can be even larger without impairing or biasing the experiment, and in the absence of feed or nutritionally enhanced forage, deer will probably utilize larger home ranges. Success of the study in terms of experimental unit size primarily depends on intervening space (buffer zone) between units. Theoretically, a control unit could be several times larger than a supplemental feed unit as long as the control animals do not enter any of the other experimental units. b. Response Variables The primary response variable will be fawn:doe ratios measured during the December following the previous winter's treatments. In other words, the fawns counted during December age classification will have been born the summer following the winter treatments, �150 and classified when they are 6 months of age. Fawn:doe ratios will be measured in Units B and C during the pilot year, and measured in each experimental unit during the initial 5 years of the full-scale study. That is, prior to the first winter treatment of feeding on Unit B in the pilot study, adult females will be radio-collared during the last half of November, and fawn:doe ratios will be estimated from the collared sample to test the procedure and to obtain initial pretreatment data. Ratios will then be measured in each of the following 5 Decembers to evaluate treatments in both the pilot and full-scale studies. Measurements will be continued for more than 5 years only on Units A and D to adequately evaluate the impact of the habitat manipulation on fawn:doe ratios. The second response variable will be overwinter fawn survival, measured directly from radio-collared fawns during the same winter as the treatments. Survival will be measured in each experimental unit for the first 4 winters of the full-scale study. Survival measurements will be continued for several more years in Units A and D to document the long-term impacts of the treatment. Overwinter fawn survival will not be measured during the pilot study. 3. Sample Size/Power Analysis a. Fawn:Doe Ratios The sampling unit of the proposed study is the same as the experimental units .. The primary response variable is the mean fawn:doe ratios of the radio-collared does wintering on the experimental unit of interest. We desire to detect an effect size, i.e., an increase in fawn:doe ratios in response to the treatments, in the range of 15 to 20 fawns per 100 does. These values are based on simple population models with overwinter fawn survival of 0.444, adult female survival of0.853, and December fawn:doe ratios of 66 fawns per 100 does to obtain a stationary population (Unsworth et al. 1999). The fawn:doe ratio for the Uncompahgre Plateau during the 1990's averaged approximately 40 fawns per 100 does (Fig. 2). 90 80 u, Q) 70 0 60 0 0 50 u, 40 0 ..... C: ~ co 30 20 LL 10 0 1982 1984 1986 1988 1990 1992 1994 1996 Year Figure 2. Post-season fawn:doe ratios for deer in DAU D-19, Uncompahgre Plateau, in southwest Colorado. 1998 �151 Fawn:doe ratios of the deer groups containing radio-collared animals will be the main source of variation in the study. Groups may contain non-reproductive females such as yearlings or barren old-age animals, and productivity per reproducing female varies. Based on surveys of DAU D-19 from 1992-96, the standard deviation of the fawn:doe ratio for groups with at least one adult female was 57, with a mean of 41. (Recent classifcation data from D-19 was not incorporated into the standard deviation estimate because fawn:doe ratios are now measured using quadrat classification as opposed to group classification.) Using an expected standard deviation of 57, the standard error of the mean fawn:doe ratio for 40 radio-collared does is 57/(40 1~ = 9.0, which is the expected standard deviation of measured fawn:doe ratios on each of the experimental units. We assessed power of the proposed experiment using SAS Analyst®. We used a two-sample I-test with a sample size of 4, representing the years of the study where treatment effects will be measured. The power of the design to detect an increase of20 fawns per 100 does is about 0.87 (Fig. 3). Another way to interpret this power estimate is to consider that the confidence interval on the treatment effect, e.g., effect of feeding, would not include zero 87% of the time for replications of the entire experiment. One source of variation has been neglected: sampling variance of the mean fawn:doe ratio of the 40 radio-collared does. This variance can be reduced by repeated helicopter surveys to estimate the fawn:doe ratio of the deer groups containing radio-collared does. Spatial variation between the 4 experimental units will also exist, but will be accounted for in the design and removed with the unit effect. Likewise, the temporal variation across years was included in the standard deviation of 57, but will be accounted for in the design by the year effect. Thus, we feel the power estimated in Figure 3 is reasonable for the proposed study. 1 0.9 0.8 0.7 L. Q) 0.6 ~ 0 a.. 0.5 0.4 0.3 0.2 0.1 -- -----... --- Increase in Fawns:100 Does: -- ·······10 ----15 .... - - - - -20 - - 25 --30 - 3 5 ...... ..... -40 ... - 2 3 4 Sample Size (years) 5 6 Figure 3. Expected power of detecting fill increase from 10 to 40 fawns per 100 does with 40 adult females collared per experimental writ, 8Ild a stfilldard deviation of the age ratio of 57. A power of 0.87 is slightly higher than the commonly accepted power of 0.80. A sample size of 35 radio-collared does in each experimental unit would achieve a power of approximately 0.80 using the previously described power calculations. However, it is likely that some fraction of the radio-collared does in December of one year will not be alive to be sampled in December the following year. Additionally, a few of the radio-collared animals may not remain on the experimental unit where they were initially radio-collared. Animals that switch experimental units during mid-winter will be discarded from the study because they would not �152 represent either of the treatment areas that they inhabited. Thus, 40 radio-collared does per experimental unit provides a slight cushion for collared doe losses from one December to the next. The habitat manipulation experimental unit along with the control unit will be measured for more than 4 years because it is likely that deer will respond more slowly and less dramatically than with the supplemental feed treatment. Thus, given >4 years of measurements, additional power will be obtained in the evaluation of habitat manipulations. b. Overwinter Fawn Survival A sample size of 40 fawns per experimental unit per year will provide power of O.81 to detect a difference of O.15 in survival between 2 experimental units if survival on the control unit is 0.40. We expect to see an increase in fawn survival (effect size) of approximately O.15, because this is the difference measured in the density reduction experiment conducted by White and Hartmann (1998). 4. Procedures a. Capture Methods Deer will be captured using helicopter net guns (Barrett et al. 1982, van Reenen 1982) and baited drop nets (Ramsey 1968, Schmidt et al. 1978). Deer in the experimental unit receiving the supplemental feed will be captured primarily with drop nets, where the bait will be partially composed of the supplemental feed. Deer receiving the supplemental feed treatments may also be captured via anesthesia from a projected syringe (dart) using the following procedure. Feeding will be initiated in the appropriate experimental unit. We will monitor feed sites from a nearby blind, and use a dart gun/projectile syringe to anesthetize deer that accept the feed. Deer will be anesthetiz.ed with tiletamine HCI and zolazepam (Telazol®, 5 mg/kg) and xylazine HCI (2.5 mg/kg) delivered intramuscularly. There are two primary advantages of using baited drop nets and dart gunning to capture deer in the supplemental feed treatment areas. First, deer will have demonstrated a willingness to consume the feed prior to being captured, which increases the likelihood that these animals will utilize the feed through the winter. Second, we will be able to capture the desired number of deer over a smaller geographical area than with net gunning. If the 40 does and 40 fawns are captured over too large of an area, it will be very difficult to successfully feed all 80 animals on a daily basis while attempting to feed no more than a total of ~500 animals. Deer will be fitted with leather radio collars equipped with mortality sensors, which after remaining motionless for 4 hours, increase the pulse rate of received signals. Permanent collars will be placed on females, while temporary collars will be placed on fawns. To make collars temporary, one end of the collar will be cut in half and r~ttached using rubber surgical tubing; fawns will shed the collars after approximately 6 months. Fawn collars will be reused annually, which will reduce costs after the initial year of the full-scale study. Doe collars will be equipped with a rectangular piece of flexible plastic (Ritchey® neck band material) stitched to the top-side of the collar. The piece of flexible plastic will be engraved with a unique identifier than can be visually identified and interpreted from the air. The plastic identifier will be used to ensure that each doe is accurately and individually identified during helicopter age and sex classification each December. A uniquely numbered ear tag will be placed on both does and fawns. We will record the weight of each fawn, and collect blood samples from both does and fawns whenever feasible to evaluate disease prevalence. �153 b. Measurement ofFawn:Doe Ratios Approximately 40 adult females will be captured and radio-collared in Units B and C from late November through mid December prior to the pilot study. The following winter, 40 adult females will be radio-collared in Units A and D, and additional females will be captured in Units B and C to maintain a sample size of 40 collared does in each experimental unit. This will result in a total sample size of 160 does during the full-scale study. Radio-collared animals will be monitored through the winter to determine survival and fidelity to a particular experimental unit. Does receiving the supplemental feed will be monitored to ensure that they actually consume the feed on a regular basis. The December following each winter, the group of deer with each radio-collared doe will be located by radio-tracking from a helicopter. Each deer group containing the collared doe will be counted and classified by age and sex. The procedure may be repeated if necessary to reduce the sampling variance of the estimates of fawn:doe ratios for each experimental unit. Since ratio estimates will be based on individual deer groups, we will attempt to capture each of the 40 does from separate groups of deer scattered throughout the experimental unit. Multiple collared does in the same group of deer during December classification flights will reduce our functional sample size. Additional deer will be radio-collared each December to maintain sample sizes at 40 collared does per experimental unit. c. Overwinter Fawn Survival Approximately 40 fawns (6 months old) will be captured and radio-collared during late November through mid December ofeach winter on each experimental unit during the fullscale study. Fawns will not be monitored during the pilot study. We will avoid capturing sibling fawns to ensure that independent samples are used to evaluate overwinter fawn survival. Survival will be directly measured through the winter by radio-monitoring collared fawns to determine fate (live/mortality). Fawns will also be mo_nitored to determine fidelity to the experimental units and to ensure that animals receiving the supplemental feed actually consume the feed. d. Feeding The supplemental feed provided to the deer will consist of pelleted rations developed by Baker and Hobbs (1985). Feed will be periodically transported to the experimental unit receiving the treatment, where the feed will be stored on site in a large steel bin. Pellets will be distributed daily via snowmobiles or ATV's along established trails throughout the experimental unit to provide a food source for the entire deer population in the treatment unit. This will prevent dominant animals from restricting access to the food supply because of the large area over which pellets will be distributed. Pellets will be supplied ad libitum such that a small residual remains when the next distribution of food is provided. Collared deer will be carefully monitored to ensure that treatment deer are consuming the feed whereas non-treatment deer are not. Deer that do not regularly consume the feed will be withdrawn from the sample. Any non-treatment deer that find and utilize the feed will likewise be removed from the sample. Keeping the experimental units spatially separate should minimize the latter problem. The pelleted ration will be commercially produced in the form of 2x lx0.5-cm wafers (Baker and Hobbs 1985). Feed constituents (i.e. digestibility, protein, gross energy etc.) vastly exceed those of typical winter range deer diets; exact constituent values are provided by Baker and Hobbs (1985). When provided ad libitum, the feed should allow deer to meet or exceed �154 nutritional requirements for growth and maintenance (Ullrey et al. 1967, Verme and illlrey 1972, Thompson et al. 1973, Smith et al. 1975, Baker et al. 1979, Holter et al. 1979). The basis for feeding such high quality pellets is to ensure that the treatment (enhanced nutrition) is effectively delivered to the deer. Our intent is not to determine the exact level of nutrition necessary to increase fawn recruitment, but rather to determine if nutrition is a limiting factor to recruitment. If nutrition is in fact limiting, we will rely on the habitat manipulation treatment to evaluate what exactly can be done via management to increase fawn survival and recruitment. e. Habitat Manipulations Habitat will be manipulated in experimental unit D through collaboration with the Uncompahgre Ecosystem Restoration Project (UERP) committee, which comprises personnel from the Division of Wildlife, U.S. Forest Service, Bureau of ~d Management, and the Colorado West Public Lands Partnership. The UERP committee is using an experimental landscape approach to manipulate various habitats in a mosaic pattern throughout the Uncompahgre Plateau. We will focus our intensive deer monitoring on one of these habitat manipulations that will be conducted in experimental unit D. The vast majority of deer winter range on the Uncompahgre Plateau comprises pinyon pine (Pinus edulis)-juniper (Juniperus spp.) habitat. Considerations and techniques for manipulating pinyon-juniper habitats have been evaluated in various areas throughout the West (Monsen and Stevens 1999), including considerations for benefiting wildlife (Fairchild 1999, Miller 1999). We will treat habitat primarily using fire (Erskine and Goodrich 1999, Goodrich and Barber 1999, Roberts 1999) and mechanical treatments such as rollerchopping (Sorensen 1999). The type of treatment used in the experimental unit will be based on several factors, which include topography, seral stage of the habitat, soil type, moisture regime, and information from previous studies. Regardless of the technique employed, treated area will be reseeded with a grass-forb-shrub mixture that is beneficial for mule deer. Reseeding will also be used to prevent invasive weeds from establishing in the newly disturbed areas (Sheley et al. 1996, Goodrich and Rooks 1999, Svejcar 1999). The proximate objective of the habitat manipulations is to raise the overall forage quality on treated areas in order to accomplish the ultimate objective of increased fawn recruitment. In order to meet minimum maintenance requirements, deer must consume diets containing approximately 50% digestibility (Ammann et al. 1973) and 5-7% crude protein (Einarson 1946, Dietz 1965, Murphy and Coates 1966, Holter et al. 1979). Most winter range browse ranges from 25-50% digestibility and 5-10% crude protein (Welch 1989), which is why deer rely heavily on pre-winter fat and protein stores in order to survive the winter. However, because the nutritional value of winter forage is on the borderline of satisfying minimum maintenance requirements, relatively modest increases in digestibility and protein should be beneficial for both fawn survival as well as doe productivity. The majority of forage currently available to deer during winter on the Uncompahgre Plateau consists of older-aged pinyonjuniper and decadent sagebrush (Artemisia spp.), and oakbrush (Quercus gambelli) at the higher winter range elevations. The pinyon-juniper overstory has eliminated most understory forbs, grasses and shrubs. Thus, plant species diversity is very limited, plant communities predominantly comprise older shrubs, and even after "spring green-up" occurs, the quantity of green forage is lacking throughout much of the winter range. Given these field observations, we expect to increase the nutritional quality of winter range habitat by achieving the following objectives with habitat treatments: 1.) Return portions of late-seral pinyon-juniper habitats to earlier successional stages in a mosaic pattern across the landscape �155 2.) Restore vigor to decadent sagebrush communities 3.) Increase the quantity and diversity of other winter browse species (primarily through reseeding): mountain mahogany (Cercocarpus montanus), serviceberry (Amelanchier alnifolia), bitterbrush (Purshia tridentata), gray rabbitbrush (Chrysothamnus nauseosus), green rabbitbrush (C. viscidiflorus) ... 4.) Increase the extent of younger-aged, mixed shrub communities (by accomplishing the first 3 objectives) 5.) Create a productive understory by increasing the quantity and diversity of various forbs and grasses (by freeing up nutrient resources and reseeding) Radio-tracking will be used to monitor deer use of the newly treated areas during each winter following the treatment. We do not necessarily expect manipulated habitats to receive extensive deer use the first winter after treatment, particularly following fire treatment where considerable biomass is removed from the site. Depending on snow conditions, responding vegetation may or may not be available to the animals. In short, it may require 2 or 3 years before treated habitat confers an advantage over untreated habitat for mule deer during winter. However, if December fawn:doe ratios and overwinter fawn survival do not increase after several years, and we verify that collared animals are using the treated habitat, we will quantify the nutritional quality of various forage species sampled from both the treatment and control units. Nutrition will be evaluated based on both in vitro dry matter digestibility (Tilley and Terry 1963) and crude protein (A.O.A.C. 1984). If nutritional quality of forage is not greater in tlie treatment unit than in the control unit, then we will reevaluate our techniques used to manipulate habitats. Treatment techniques used by the UERP committee will then be modified in an attempt to better manipulate habitat for mule deer. If, on the other hand,· the nutritional quality of forage is significantly greater in the treatment area than in the control area, then we can assume one of two things. First, the increase in nutrition was not of sufficient magnitude to increase fawn survival or December recruitment. Or secondly, the deer did not spend enough time feeding in the treated habitat to benefit from the increased nutrition. This could occur if deer perceived an increased level of wlnerability from the changes in habitat structure. We would attempt to differentiate between these two scenarios with intensive radio-monitoring. f Statistical Methods We will test for differences between experimental units and years using the following statistical model: Yi.Jc=µ+ t:;- +A+ af3J1c + eifJ7iJ, whereyiJc= fawn:doe ratio for the ith deer group in treatment combinationjk; i = 1, 2, ... , nJ/c (deer groups);/= 1, 2, 3, 4 experimental units (control, supplemental feed, habitat manipulation); k= 1, 2, 3, 4 (6) years; a/Jic= interactions among experimental units and years; and ei(Jk) = random error associated withyiJ1<. As stated earlier, we will obtain baseline data on each experimental unit by measuring fawn:doe ratios the December prior to implementing the first treatments. By not implementing the habitat manipulation until the second year of the study, two years of baseline data on Unit D will provide an estimate of the differences between this unit and the others. Likewise, the cross-over design on the feeding treatment allows estimation of the year effects on Units Band C, relative to the overall control Unit A (Fig. 1). A similar model will be used to analyze overwinter fawn survival, but a logit-link :function will be used in place of the identity link function in the above general linear model. �156 We will test for differences in fawn weights using a similar mcxlel as above, except the response variable will be fawn mass instead of fawn:doe ratios. Various studies have closely linked December fawn mass to survival probability (Bartmann et al. 1992, Bishop 1998, Unsworth et al. 1999). Fawn mass analyses may help us describe or better understand observed changes (or lack thereof) in population parameters as a result of the treatments. At the minimum, mean December fawn mass from each experimental unit will allow us to evaluate whether fawns are entering the winter treatments with equal survival probability. At least during the first year of the study, the mean mass of fawns collared in December should be similar among the various experimental units, assuming fawns are randomly captured. It is possible, however, that in successive years December fawn mass might vary among the experimental units as a result of previous winter's treatment effects. Although we cannot directly link 6 month old fawns to the previous winter's experimental units, it is certainly possible that many of those fawns' mothers occupied the same treatment areas the winter before given home range site fidelity. For example, fawns collared during the second winter in Experimental Unit B (which received the feed the winter before) could be heavier than fawns in Unit A (control). If so, it would suggest that not only has the probability of survival to December increased, but given the overall greater early winter mass, probability of overwinter survival has increased as well. In short, increased doe nutrition may not only increase December fawn recruitment, but yearling recruitment as well. G. LOCATION OF WORK Research will be conducted on the Uncompahgre Plateau in southwest Colorado, which comprises DAU D-19 and GMUs (game management units) 61 and 62. H. WORK ASSIGNED TO: 1. 2. 3. 4. Chad J. Bishop, Mammals Researcher, Colorado Division of Wildlife Gary C. White, Professor of Wildlife Biology, Colorado State University R. Bruce Gill, Mammals Research Leader, Colorado Division of Wildlife Len H. Carpenter, Southwest Field Representative, Wildlife Management Institute E. RESOURCE REQUIREMENTS Fiscal Year Equipment and Sue12lies Rental Services Service Contracts Vehicles FfE Requirements PFfE TFfE Total Costs 2000-01 $28,300 $45,200 $50,000 $11,046 1.00 0.5 $201,548 2001-02 $87,850 $147,900 $50,000 $11,046 1.00 1.00 $374,408 2002-03 $57,850 $116,400 $50,000 $11,046 1.00 1.00 $312,908 2003-04 $57,850 $116,400 $50,000 $11,046 1.00 1.00 $312,908 2004-05 $57,850 $116,400 $50,000 $11,046 1.00 1.00 $312,908 2005-06 $37,600 $65,200 $50,000 $11,046 1.00 0.5 $230,848 2006-07 $17,600 $50,000 $50,000 $11,046 1.00 0.5 $195,648 �157 F. REPORT DUE DATES FY2000-01 Pilot Results/Revised Program Narrative FY2001-02 Progress Report FY2002-03 Progress Report FY2003-04 Progress _Report *FY2004-05 Progress Report FY2005-06 Progress Report *FY2006-07 Completion Report 8/1/01 8/1/02 8/1/03 8/1/04 8/1/05 8/1/06 8/1/07 *Supplemental Feed portion of experiment will be completed in FY2004-05. Monitoring of habitat manipulation experimental unit and control area will likely continue for 2 more years unless corresponding hypotheses can be adequately tested beforehand. We estimate to finish the completion report for the entire study following FY2006-07. G. LITERATURE CITED Ammann, A. P., R. L. Cowan, C. L. Mothershead, and B. R. Baumgardt. 1973. Dry matter and energy intake in relation to digestibility in white-tailed deer. Journal of Wildlife Management 37:195-201. Armleder, H. M., M. J. Waterhouse, D. G. Keisker, and R. J. Dawson. 1994. Winter habitat use by mule deer in the central interior of British Columbia. Canadian Journal of Zoology 72: 1721-1725. Association of Official Analytical Chemists (A.O.A.C.). 1984. Official methods of analysis. Fourteenth edition. Association of Official Analytical Chemists, Washington D.C., USA. Austin, D. D., R. Stevens, K. R. Jorgensen, and P. J. Urness. 1994. Preferences of mule deer for 16 grasses found on lntermountain winter ranges. Journal of Range Management 47:308-311. Austin, D. D., P. J. Urness, and M. L. Wolfe. 1977. The influence of predator control on two adjacent wintering deer herds. Great Basin Naturalist 37:101-102. Bailey, J. A. 1984. Principles of wildlife management. John Wiley & Sons, New York, New York, USA. Baker, D. L., and N. T. Hobbs. 1985. Emergency feeding of mule deer during winter: tests of a supplemental ration. Journal of Wildlife Management 49:934-942. Baker, D. L., D. E. Johnson, L. H. Carpenter, 0. C. Wallmo, and R. B. Gill. 1979. Energy requirements of mule deer fawns in winter. Journal of Wildlife Management 43:162-169. Ballard, W. B., D. Lutz, T. W. Keegan, L. H. Carpenter, and J.C. deVos, Jr. 1999. Deer-predator relationships: a review of recent North American studies with emphasis on mule and black-tailed deer. Unpublished manuscript. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108-114. Bartmann, R. M. 1983. Composition and quality of mule deer diets on pinyon-juniper winter range, Colorado. Journal of Range Management 36:534-541. Bartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monographs 121:5-39. Bishop, C. J. 1998. Mule deer fawn mortality and habitat use, and the nutritional quality ofbitterbrush and cheatgrass in southwest Idaho. Thesis, University ofldaho, Moscow, Idaho, USA. Bissell, H. D., and H. Strong. 1955. The crude protein variations in the browse diet of California deer. California Fish and Grune 41: 145-155. Blackburn, W. H., and P. T. Tueller. 1970. Pinyon and juniper invasion in black sagebrush communities in east-central Nevada. Ecology 51:841-848. Bodurtha, T. S., J.M. Peek, and J. L. Lauer. 1989. Mule deer habitat use related to succession in a bunchgrass community. Journal of Wildlife Management 53:314-319. �158 Boyd, R. J., A. Y. Cooperrider, P. C. Lent, and J. A. Bailey. 1986. Ungulates. Pages 519-564 in A. Y. Cooperrider, R. J. Boyd, and H. R. Stuart, editors. Inventory and monitoring of wildlife habitat. U.S. Department of the Interior, Bureau of Land Management Service Center, Denver, Colorado, USA. Brown, E. R. 1961. The black-tailed deer of western Washington. Biological Bulletin 13. Washington Department of Game, Olympia, Washington, USA. Bunting, S. C. 1996. The use and role of fire in protected areas. Pages 277-301 in R. G. Wright, editor. National parks and protected areas: Their role in environmental protection. Blackwell Press, Cambridge, Massachusetts, USA. Burkhardt, J. W., and E.W. Tisdale. 1976. Causes ofjuniper invasion in southwestern Idaho. Ecology 57:472-484. Campbell, E.G., and R. L. Johnson. 1983. Food habits of mountain goats, mule deer, and cattle on Chopaka Mountain, Washington, 1977-1980. Journal of Range Management 36:488-491. Carson, R. G., and J.M. Peek. 1987. Mule deer habitat selection patterns in northcentral Washington. Journal of Wildlife Management 51:46-51. Clements, C. D., and J. A. Young. 1997. A viewpoint: Rangeland health and mule deer habitat. Journal of Range Management 50:129-138. Connolly, G. E. 1981. Limiting factors and population regulation. Pages 245-285 in 0. C. Wallmo, editor. Mule and black-tailed deer of North America. University of Nebraska Press, Lincoln, Nebraska, USA. Deschamp, J. A., P. J. Urness, and D. D. Austin. 1979. Summer diets of mule deer from lodgepole pine habitats. Journal of Wildlife Management 43:154-161. Dietz, D. R 1965. Deer nutrition research in range management. Transactions of the North American Wildlife and Natural Resources Conference 30:274-285. Einarsen, A. S. 1946. Crude protein determination of deer food as an applied management technique. Transactions of the North American Wildlife and Natural Resources Conference 11:309-312. Erskine, I., and S. Goodrich. 1999. Applying fire to pinyon-juniper communities of the Green River Corridor, Daggett County, Utah. Pages 315-316 in S. B. Monsen and R Stevens, compilers. Proceedings: ecology and management of pinyon-juniper communities within the Interior West. U.S.D.A. Forest Service, Rocky Mountain Research Station, RMRS-P-9. Fairchild, J. A. 1999. Pinyon-juniper chaining design·guidelines for big game winter range enhancement projects. Pages 278-280 in S. B. Monsen and R Stevens, compilers. Proceedings: ecology and management ofpinyon-juniper communities within the Interior West. U.S.D.A. Forest Service, Rocky Mountain Research Station, RMRS-P-9. Farmer, M. E. 1995. The effect of anchor chaining pinyon-juniper woodland on watershed values and big game animals in central Utah. Thesis, Brigham Young University, Provo, Utah, USA. Freddy, D. J. 1984. Heart rates for activities of mule deer at pasture. Journal of Wildlife Management 48:962-969. Gill, R B., T. D. I. Beck, C. J. Bishop, D. J. Freddy, N. T. Hobbs, RH. Kahn, M. W. Miller, T. M. Pojar, and G. C. White. 1999. Declining mule deer populations in Colorado: reasons and responses. A report to the Colorado Legislature. Colorado Division of Wildlife, Denver, Colorado, USA. Gill, R. B., L. H. Carpenter, R. M. Hartmann, D. L. Baker, and G. G. Schoonveld. 1983. Fecal analysis to estimate mule deer diets. Journal of Wildlife Management 47:902-915. Goodrich, S., and B. Barber. 1999. Return interval for pinyon-juniper following fire in the Green River Corridor, near Dutch John, Utah. Pages 391-393 in S. B. Monsen and R. Stevens, compilers. Proceedings: ecology and management ofpinyon-juniper communities within the Interior West. U.S.D.A. Forest Service, Rocky Mountain Research Station, RMRS-P-9. Goodrich, S., and D. Rooks. 1999. Control of weeds at a pinyon-juniper site by seeding grasses. Pages 403-407 in S. B. Monsen and R Stevens, compilers. Proceedings: ecology and management of pinyon-juniper communities within the Interior West. U.S.D.A. Forest Service, Rocky Mountain Research Station, RMRS-P-9. �159 Hanley, T. A. 1982. The nutritional basis for food selection by ungulates. Journal of Range Management 35:146-151. Hanley, T. A. 1997. A nutritional view of understanding and complexity in the problem of diet selection by deer (Cervidae). Oikos 79:209-218. Hobbs, N. T. 1989. Linking energy balance to survival in mule deer: development and test of a simulation model. Wildlife Monographs 101:5-39. Hobbs, N. T., and R. A. Spowart. 1984. Effects of prescribed fire on nutrition of mountain sheep and mule deer during winter and spring. Journal of Wildlife Management 48:551-560. Holter, J.B., H. H. Hayes, and S. H. Smith. 1979. Protein requirement of yearling white-tailed deer. Journal of Wildlife Management 43:872-879. Hom, E. E. 1941. Some coyote-wildlife relationships. Transactions of the North American Wildlife Conference 6:283-287. Kasworm, W. F., L. R. Irby, and H.B. Ihsle Pac. 1984. Diets of ungulates using winter ranges in northcentral Montana. Journal of Range Management 37:67-71. Keay, J. A., and J. M. Peek. 1980. Relationships between fires and winter habitat of deer in Idaho. Journal of Wildlife Management 44:372-380. Knowles, C. J. 1976. Observations of coyote predation on mule deer and white-tailed deer in the Missouri River breaks, 1975-1976. Pages 117-138 in Montana deer studies. Progress Report, P-R Project W120-R-7. Montana Fish and Game Department, Helena, Montana, USA. • Kufeld, R. C., D. C. Bowden, and D. L. Schrupp. 1988. Habitat selection and activity patterns of female mule deer in the Front Range, Colorado. Journal of Range Management 41:515-522. Lemos, J.C., M. L. Mcinnis, C. E. Trainer, W. J. Castillo, and R. E. Anglin. 1978. Steens Mountain mule deer population study: coyote population evaluation. Job Progress Report, P-R Project W-70-R8, Subproject F, Study VII, Job 5. Oregon Department of Fish and Wildlife, Portland, Oregon, USA. • Leopold, A. S., T. Riney, R. McCain, and L. Tevis, Jr. 1951. The Jawbone deer herd. Bulletin no. 4. California Department of Fish and Game, Sacramento, California, USA. Loft, E. R., and J. W. Menke. 1990. Evaluations of fire effects on mule deer habitats in Lassen County. Hill Bill Contract No. FGlC-2090, California Department of Fish and Game, Sacramento,California, USA. McMichael, T. J. 1970. Rate of predation on deer fawn mortality. Pages 77-83 in Wildlife research in Arizona, 1969-70. Arizona Game and Fish Department, Phoenix, Arizona, USA. Miller, R. F. 1999. Managing western juniper for wildlife. Pages 89-97 in Range field day 1999 Progress Report, Juniper woodlands: history, ecology, and management. Agricultural Experiment Station, Oregon State University, Special Report 1002. Corvallis, Oregon, USA. Miller, R. F., T. J. Svejcar, and J. A. Rose. 1999. Conversion of shrub steppe to juniper woodland. Pages 385-390 in S. B. Monsen and R. Stevens, compilers. Proceedings: ecology and management of pinyon-:juniper communities within the Interior West. U.S.D.A. Forest Service, Rocky Mountain Research Station, RMRS-P-9. Monsen, S. B., and R. Stevens, compilers. 1999. Proceedings: ecology and management of pin.yon-juniper communities within the Interior West. U.S.D.A. Forest Service, Rocky Mountain Research Station, RMRS-P-9. Murie, A. 1940. Ecology of the coyote in the Yellowstone. Fauna National Parks Bulletin no. 4. GPO, Washington, D.C., USA. Murphy, D. A., and J. A. Coates. 1966. Effects of dietary protein on deer. Transactions of the North American Wildlife and Natural Resources Conference 31:129-139. Nudds, T. D. 1980. Forage "preference": theoretical considerations of diet"selection by deer. Journal of Wildlife Management 44:735-740. Ordway, L. L., and P.R. Krausman. 1986. Habitat use by desert mule deer. Journal of Wildlife Management 50:677-683. �160 Parker, K. L., M. P. Gillingham, T. A. Hanley, and C. T. Robbins. 1996. Foraging efficiency: energy expenditure versus energy gain in free-ranging black-tailed deer. Canadian Journal of Zoology 74:442-450. Parker, K. L., C. T. Robbins, and T. A. Hanley. 1984. Energy expenditures for locomotion by mule deer and elk. Journal of Wildlife Management 48:474-488. Pojar, T. M. 2000. Mule deer fawn survival study 1999; Uncompahgre Plateau and Middle Park, Colorado. Unpublished data report, Fort Collins, Colorado, USA. Ramsey, C. W. 1968. A drop-net deer trap. Journal of Wildlife Management 32:187-190. Riggs, R., S. C. Bunting, and S. Daniels. 1996. Prescribed fire. Pages 295-319 in P.R. Krausman, editor. Rangeland wildlife. Society for Range Management, Denver, Colorado, USA. Roberts, T. C., Jr. 1999. The budgetary, ecological, and managerial impacts of pinyon-juniper and cheatgrass fires. Pages 400-402 in S. B. Monsen and R. Stevens, compilers. Proceedings: ecology and management of pinyon-juniper communities within the Interior West. U.S.D.A. Forest Service, Rocky Mountain Research Station, RMRS-P-9. Robinette, W. L., N. V. Hancock, and D. A. Jones. 1977. The Oak Creek mule deer herd in Utah. Resource Publication 77-15. Utah Division of Wildlife, Salt Lake City, Utah, USA. Schladweiler, P. 1976. Effects of coyote predation of big game populations in Montana. Report, P-R Project W-120-R-7, Study NG-47.1. Montana Department of Fish and Game, Helena, Montana, USA. Schmidt, R L., W. H. Rutherford, and F. M. Bodenham. 1978. Colorado bighorn sheep-trapping techniques. Wildlife Society Bulletin 6:159-163. Sheley, R. L., T. J. Svejcar, and B. D. Maxwell. 1996. A theoretical framework for developing successional weed management strategies on rangeland. Weed Technology 10:766-773. Smith, A. D. 1957. Nutritive value of some browse plants in winter. Journal of Range Management 10:162-164. Smith, RH., and A. LeCount. 1976. Factors affecting survival of mule deer fawns. Final Report, P-R Project W-78-R, WP2, J 4. Arizona Game and Fish Department, Phoenix, Arizona, USA. Smith, S. H., J.B. Holter, H. H. Hayes, and H. Silver. 1975. Protein requirement of white-tailed deer fawns. Journal of Wildlife Management 39:582-589. Sorensen, D. 1999. Advantages and effectiveness ofrollerchopping. Page 293 in S. B. Monsen and R Stevens, compilers. Proceedings: ecology and management of pinyon-juniper communities within the Interior West. U.S.D.A. Forest Service, Rocky Mountain Research Station, RMRS-P-9. Sowell, B. F., B. H. Koerth, and F. C. Bryant. 1985. Seasonal nutrient estimates of mule deer diets in the Texas Panhandle. Journal of Range Management 38:163-167. Spalinger, D. E., and N. T. Hobbs. 1992. Mechanisms of foraging in mammalian herbivores: new models of functional response. American Naturalist 140:325-348. Stevens, R 1999. Mechanical chaining and seeding. Pages 281-284 in S. B. Monsen and R. Stevens, compilers. Proceedings: ecology and management of pinyon-juniper communities within the Interior West. U.S.D.A. Forest Service, Rocky Mountain Research Station, RMRS-P-9. Svejcar, T. 1999. hnplications of weedy species in management and restoration of pinyon and juniper woodlands. Pages 394-396 in S. B. Monsen and R Stevens, compilers. Proceedings: ecology and management of pinyon-juniper communities within the Interior West. U.S.D.A. Forest Service, Rocky Mountain Research Station, RMRS-P-9. Taber, R. D. 1956. Deer nutrition and population dynamics in the North Coast Range of California. Transactions of the North American Wildlife Conference 21: 159-172. Taber, R. D., and RF. Dasmann. 1958. The black-tailed deer of the chaparral: Its life history and management in the North Coast Range of California Game Bulletin 8. California Department of Fish and Game, Sacramento, California, USA. Thomas, T. R, and L. R. Irby. 1991. Winter habitat use by mule deer with access to wheat fields and planted forb-grassland. Wildlife Society Bulletin 19:155-162. �161 Thompson, C. B., J.B. Holter, H. H. Hayes, H. Silver, and W. E. Urban, Jr. 1973. Nutrition ofwhitetailed deer. I. Energy requirements offawns. Journal ofWildlife Management 37:301-311. Tilley, J.M. A., and R. A. Terry. 1963. A two-stage technique for the in vitro digestion of forage crops. Journal of the British Grassland Society 18:104-111. Trainer, C. E., M. L. Mcinnis, J.C. Lemos, W. J. Castillo, and R. E. Anglin. 1978. Steens Mountain mule deer population study: fawn survival. Job Progress Report, P-R Project W-70-R-8, Subproject F, Study VI, Job 2. Oregon Department of Fish and Wildlife, Portland, Oregon, USA. Trout, L. E., and J. L. Thiessen. 1973. Physical condition and range relationships of the Owyhee deer herd. Idaho Department of Fish and Game, Federal Aid to Wildlife Restoration, Job Completion Report, Project W-141-R-2. Tueller, P. T. 1979. Food habits and nutrition of mule deer on Nevada ranges. Agricultural Experiment Station, University of Nevada, Reno, Nevada, USA. Ullrey, D. E., W. G. Youatt, H. E. Johnson, L. D. Fay, and B. L. Bradley. 1967. Protein requirement of white-tailed deer fawns. Journal of Wildlife Management 31 :679-685. Unsworth, J. W., D. F. Pac, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. Urness, P. J., A. D. Smith, and R. K. Watkins. 1977. Comparison ofin vivo and in vitro dry matter digestibility of mule deer forages. Journal of Range Management 30:119-121. van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-421 in L. Nielsen, J.C. Haigh, and M. E. Fowler, editors. Chemical immobiliz.ation of North American wildlife. Wisconsin Humane Society, Milwaukee, Wisconsin, USA. Verme, L. J., and D. E. Ullrey. 1972. Feeding and nutrition of deer. Pages 275-291 in D. C. Church, editor. Digestive physiology and nutrition of ruminants. Volume 3 - Practical Nutrition. D. C. Church, Corvallis, Oregon, USA. Wallmo, 0. C. 1969. Response of deer to alternate-strip clearcutting of lodgepole pine and spruce-fir timber in Colorado. U.S. Forest Service, Rocky Mountain Forest and Range Experiment Station, Research Note RM-141. Wallmo, 0. C., L. H. Carpenter, W. L. Regelin, R. B. Gill, and D. L. Baker. 1977. Evaluation of deer habitat on a nutritional basis. Journal of Range Management 30:122-127. Wallmo, 0. C., R. B. Gill, L. H. Carpenter, and D. W. Reichert. 1973. Accuracy of field estimates of deer food habits. Journal of Wildlife Management 37:556-562. Wallmo, 0. C., W. L. Regelin, and D. W. Reichert. 1972. Forage use by mule deer relative to logging in Colorado. Journal ofWildlife Management 36:1025-1033. Welch, B. L. 1989. Nutritive value of shrubs. Pages 405-424 in C. M. McKell, editor. The biology and utilization of shrubs. Academic Press, San Diego, California, USA. White, G. C., and R. M. Bartmann. 1998. Effect of density reduction on overwinter survival of freeranging mule deer fawns. Journal ofWildlife Management 62:214-225. White, R. G. 1983. Foraging patterns and their multiplier effects on productivity of northern ungulates. Oikos 40:377-384. Wickstrom, M. L., C. T. Robbins, T. A. Hanley, D. E. Spalinger, and S. M. Parish. 1984. Food intake and foraging energetics of elk and mule deer. Journal of Wildlife Management 48:1285-1301. Workman, G. W., and J.B. Low, editors. 1976. Mule deer decline in the West: a symposium. Utah State University and Utah Agricultural Experiment Station, Logan, Utah, USA. �162 �163 APPENDIXB DRAFT EFFECTS OF LIMITED BUCK HARVEST ON MULE DEER SEX AND AGE RATIOS An A Posteriori Analysis Chad Bishop, Mammals Researcher Objective Determine whether significant reductions in buck harvest have improved buck:doe ratios and/or fawn:doe ratios in Colorado. Data Set lbis analysis focuses on mule deer populations occupying mountainous regions of Colorado, or those populations west ofl-25. Over the past 10 years, buck harvest restrictions were implemented in a number of Game Management Units (GMUs) within several deer Data Analysis Units (DAUs) in western Colorado (Table 1). Table 1. Deer DAUs and inclusive GMUs in western Colorado where buck harvest limitations were 1990 D-1 201 238 47 ,❖:-:-.-: 'ttltfNz'·:'\@f#@(tji 1992 D-14 44 666 157 1995 D-1 1, 2 123 21 21, 30 1219 595 i;E ,-4~™1e1iF-@!1@M?l,:::::il 1995 D-11 I used these various units to look at buck:doe and fawn:doe ratios an equal number of years before and after the harvest limitation was implemented. DAUs D-1 and D-11 were excluded from all analyses due to insufficient age and sex ratio data. Thus, my analysis of the effects of limited buck harvest is based on DAUs D-3, D-o, D-14, and D-19(GMU 61). I also looked at buck:doe and fawn:doe ratios in units where harvest limitations were not implemented prior to 1999. Again, I only used those DAUs with adequate sex and age ratio data available, which include: D-2, D-4, D-7, D-8, D-9, D-12, D-13, D-18, D-19(GMU62), D-20, D-21, D-22, D-24, D-25, D29, D-30, D-39, D-40, D-42, D-43, and D-51. I obtained nearly all harvest data and age and sex ratio data from the DEAMAN32 database. Data from 1997 and 1998 that was not available on my version ofDEAMAN was obtained from John Ellengberger. �164 Analysis Basic Methodology and Justification As now apparent, my primary goal was to determine whether a set of sex or age ratios prior to a buck harvest restriction differed from a set of corresponding ratios following the restriction. It seems logical, then, to think of a harvest treatment with two groups, or two treatment levels: pre- and post-harvest restriction (unlimited vs. limited buck harvest). I can see a couple ways to approach this analysis. One approach is to treat each ratio estimate as a sample observation, and then calculate a standard mean and variance of the ratios for each treatment category. In other words, follow standard analysis of variance procedures using only the ratio estimates themselves. However, each sex or age ratio estimate for a GMU or DAU has an associated standard error. Our confidence in the accuracy of any given ratio depends largely on the effort put forth for that particular classification. If we ignore this, then each measured ratio is assumed to be an equally accurate measure of the true value. Ideally, my analysis should incorporate the associated variance of each ratio estimate used to avoid bias. I therefore took two approaches: (1) I used Program CONTRAST to analyze age and sex ratio estimates and associated standard errors prior to and after the harvest restriction and (2) I used PROC GLM in SAS and did a standard analysis of variance using only the ratio estimates. Justification for these analyses follows. Program CONTRAST (Sauer and Williams 1989, JWM 53:137-142) is a generalized program designed to evaluate differences in survival based on survival rate estimates and associated standard errors. It is applicable to any type of rate estimates. Similar to my analysis here, it is desirable when estimates and associated standard errors derived from data are available, as opposed to having the raw data. It allows testing various hypotheses without ignoring the variance of estimates. As I alluded to above, the problem with analysis of variance is that the variances are ignored. I considered converting the age and sex ratios to rates, or proportions, which would be more similar to how survival rates are expressed. A typical buck:doe ratio (bucks/does) or fawn:doe ratio (fawns/does) can be expressed as a proportion: bucks/(bucks + does) or fawns/(fawns + does). However, the standard error calculated for an age or sex proportion, based on a binomial distribution, can be much more biased than the standard error estimator for an age or sex ratio. lbis occurs because of behavioral patterns in deer group formation, and resulting sampling schemes used to estimate age/sex ~tios where groups of deer are the sampling unit (Bowden et al. 1984, JWM 48:500-509). Thus, I chose to use the ratios and associated SE's. The chisquare statistic used in CONTRAST appears to be equally valid for analyzing age and sex ratio data. The same assumptions for survival or recovery rates within the confines of this statistic can be made for age and sex ratios given that the associated variance estimates are known. My second approach involved a standard analysis of variance, or multivariate analysis of variance, using buck:doe and fawn:doe ratios as the dependent variables. Although this analysis ignores the variance of each ratio estimate, it is not that different from many of the analyses we do. In many cases, samples obtained from field measur~inents have an associated error that we do not quantify. In other words, it is not uncommon for raw data values to only be an approximation of the true value. Quantifying the digestibility of forage is an example. Our measured digestibility of a plant sample is only an estimate of the true value, yet quantifying an appropriate variance for each individual sample is difficult to do. Based on this line of reasoning, I felt it appropriate to do an analysis of variance using only the ratio estimates as raw data points. It follows that some caution should be used in interpreting the data since we know beforehand that the accuracy of ratio estimates vary, yet this variance is not being accounted for. Program CONTRAST In my analysis of sex and age ratios, I chose to use an equal number of years before and after implementation of the harvest restriction. I began by analyzing each DAU individually because harvest limitations were implemented at various times. I then incorporated each of the DAUs into one analysis by �165 using buck:doe and fawn:doe ratios for 4 years prior to harvest limitation and 4 years following the limitation. This was decided based on the DAU with the least number of years of post-treatment data, D3. Thus, for this analysis of combined data. the 4 years pre- and post-harvest treatment were not necessarily the same (D-3: 1991-94/1995-98; D-6: 1987-90/1991-94; D-14 and D-19: 1988-91/1992-95). Buck:doe ratios increased following limited harvest in each DAU except for D-19, where buck:doe ratios declined (fable 2). Conversely, fawn:doe ratios declined following implementation of limited harvest in each of the DAUs (fable 3). Table 2. Results of analyses contrasting buck:doe ratios pre- and post-implementation of limited harvest ._ ' •. - =--- CO, = Pm-.:m .- C • - _z •• ,r.: .>:. ·:, ..... ·.• D-3 1991-98 18.9 26.4 D-14 1985-98 18.7 23.0 6.22 .013 D-3, D-6, D-14, D-19(GMU 61) 4yrsPre-& Post-Limit 15.8 18.3 6.10 .014 Table 3. Results of analyses contrasting fawn:doe ratios pre- and post-implementation oflimited harvest in ~Vi;;;W DAL~ i l l ~ ~ 4 : a ( ' u~ Pi~~ CO~==nL.\ST (~:~ ~:f "WHJm~~ l~), D-3 1991-98 D-14 1985-98 D-3, D-6, D-14, D-19(GMU61) 4 yrs Pre-& Post-Limit 56.1 48.1 59.4 45.3 95.07 <.001 (Multivariate) Analysis of Variance using PROC GLM in SAS DAUs D-3, D-6, D-14, D-19(GMU 61) I analyzed the same data using PROC GLM in SAS without incorporating the associated variance estimates. Buck:doe and fawn:doe ratios were the dependent variables while buck harvest was the independent variable. Buck harvest was treated as a categorical variable with two levels: (I) pre-harvest limitation and (2) post-harvest limitation. Results are reported for buck:doe and fawn:doe ratios separately based on the respective ANOVA's. However, each significant result reported from an AN OVA is protected by a significant overall MANOVA test. Failing to incorporate the standard error with each ratio estimate apparently resulted in a less efficient analysis. The change in buck:doe ratios following the harvest treatment was only significant in DAU D-3 (P = 0.020). The other comparisons among buck:doe ratios pre- and post-harvest treatment were not �166 significant (Table 4). Declines in fawn:doe ratios following the harvest restriction were significant in 3 of the analyses while not significant in DAUs D-3 and D-14 (Table 5). Table 4. Results of analyses contrasting buck:doe ratios pre- and post-implementation oflimited harvest in ~e·,r~t"~ DAU:!l. in lilii"i;:;~te,m Ql)ki,radl;l! mii:ini ~, ooaly~.i!.I ofv,arian~ mSAS_ D-3 1991-98 1;;;:;:;:;sMEi~Bf§J!m;;;.;;~;£f~:;;:;:;:;;:;112;_ D-14 18.9 26.4 11.39 .020 =~;i;t:;:·:::-=::#1111-=-?:!J!!!;tt~·:::!~t~iJ:~r··m_Yim'..='._=:=1:;~i:~·=mi@====t=mt~}·tfi~;:=::-:-:---:-:tti@:i:w=·\\,·.=.:.~=:5:~:mmmm}t;:t; 1985-98 18.7 23.0 1.61 .237 ;i-i·l·itllC, .:-,;;:;,:'.=;::.:1-::;:~J,l·:·1l·l·l:l·::1l·l1•gi1i,l;;!;;l~:-l·l·11;!-i!:•:-J-:-l-1i·l:·:·~1l-··!·11~w.:···l·l·l·l·:·l::::·ili l·l::·:·l·l····-11!1l::-:-:-:-:-:1J·il ~i-l•l-l-l-l-i:·J,·~:-~11il-1:-l-·-·_:.J.ili :·i·:·:·1:·:·11ii-i-l i-1l-l·J,·1~1l·l·l·li•l-i-i-it;llilil·li~l-i·l·l l ·l··- D-3, D-6, D-14, D-19(GMU 61) 4 yrs Pre- & Post-Limit 15.8 18.3 0.88 .357 Table 5. Results of analyses contrasting fawn:doe ratios pre- and post-implementation oflimited harvest . i,~1 ~v~r.a.l DAU~ i,!11 W~i;)f"fl C'\cilo:r~uk, ttiii!l'!:g :1m ~rn:thi~i~ ofl,i'.,!!;fiii"...-ic.:~ :i,rn S.ASL D-3 1991-98 56.1 48.1 3.42 .124 === :·!:·l·!::l:l:!··:::::!!l:l:!j·:::·!·!··:!··:I!!!!'~~,-!-::!·!·ll!!!·l~!:l·1!!~1~J.1::! D-14 1985-98 ~iinaiiiiiijiiii~ii~mifo1: .,. . D-3, D-6, D-14, D-19(GMU 61) 4 yrs Pre- & Post-Limit' 59.1 51.6 2.46 .151 45.3 15.87 <.001 ::i1iijiil1l!1:~ff-ii!iii1:i·i:i:i:-;~:iii·t,~-5 9 _4 Results from the chi-square analysis (Program CONTRAS1) are probably more informative because more infonnation was provided by incorporating the standard errors of ratio estimates. Nonetheless, both analyses indicate that buck:doe ratios, overall, have modestly increased following restricted harvest; and fawn:doe ratios have significantly declined despite the conservative buck harvest. All DA Us west ofl-25 with sufficient age and sex ratio data available Up to this point I have not included any data from units without harvest restrictions prior to 1999. My goal here is to compare buck:doe and fawn:doe ratios from the units with restricted harvest to those where harvest was not restricted. I once again used a multivariate analysis of variance with buck:doe and fawn:doe ratios as the independent variables, and as before, I will only report the univariate statistics. Since I combined all of the data into one analysis, I adopted the same procedure I used previously in selecting what data should be used: 4 years prior to and 4 years following the harvest limitation (based on the fact that D-3 only has 4 years of post-treatment data). Thus, the same data will be used for DA Us D3, D-6, D-14, and D-19(GMU 61) as used in the previous combined analyses. For all DAUs without harvest limitations, I used age and sex ratio data from 1991-98. I used 2 independent variables. The first is whether or not buck harvest was limited in each DAU prior to 1999: (1) yes or (2) no. The second �167 variable is a categorical time variable: (I) 1991-94 (or 4 yrs pre-harvest treatment for DA Us D-6, 14, 19) and (2) 1995-98 (or 4 yrs post-harvest treatment for DAUs D-6, 14, 19). Tp.us the model I used is: YJ11c = M + QJ + flx + !dl;1c + .iiutcJ where i = I,2, ... , n11 (replicates),j = I, 2 (DAU with, or without, harvest limitation), and k = I, 2 (199194/pre-harvest treatment or 1995-98/post-harvest treatment). My principal concern with this analysis is to determine whether there is a significant interaction. If limited harvest had a significant effect on buck:doe ratios, I would suspect ratios to increase from the first time category to the second time category for the DA Us where limited harvest was implemented. Conversely, I would suspect ratios to remain constant or decline from the first time category to the second time category for the DA Us where harvest was not limited. There were no significant differences among buck:doe ratios. Buck:doe ratios increased in the harvestrestricted DA Us after harvest was limited while ratios remained constant in the unlimited harvest DAUs, but the interaction was not significant (P = 0.348) (Fig. I). Fawn:doe ratios declined significantly from the pre-harvest treatment to the post-harvest treatment time period (P < 0.001). The interaction was significant (P = 0.006), with fawn:doe ratios declining much more dramatically in the limited harvest DAUs than in the unlimited harvest DAUs (Fig. 2). This indicates that buck harvest is probably of little concern in terms of fawn recruitment. Other factors are clearly affecting the deer populations in the DAUs where limited harvest was implemented. - DAUs without Limited Harvest • • • • • DAUs with Limited Harvest 20 18.3 •••••• 0 ~ 18 0:: Cl) 0 0 ~ 0 :::, m 16 17.3 - - - - -..-.~..:-:.....--··-- 17.5 ••••••• 15.8 14 Interaction not significant (P = .348) C: ca Cl) ~ •••••• 12 10 1991-94 1995-98 (Pre-Harvest Treatment} (Post-Harvest Treabnent} Figure 1. Mean buck:doe ratios in DA Us with and without harvest restrictions implemented prior to 1999. 1991-94 represents pre-harvest treatment while 1995-98 represents post-harvest treatment. �168 - • • • • • DA Us with Limited Harvest DAUs without Limited Harvest 70 .....ca0 Cl'.:'. 65 60 59.4 55 53.6 Q.) 0 0 C ~ ca 50 C ca Q.) 45 ~ 40 •• •• - •• •• LL •• •• •• •• •• . .. •• •• •• -49.8 •• •• •• 45.3 Significant Interaction ( P = .006) 35 1991-94 (Pre-Harvest Treatment) 1995-98 (Post-Harvest Treatment) Figure 2. Mean fawn:doe ratios in DA Us with and without harvest restrictions implemented prior to I 999. 1991-94 represents pre-harvest treatment while 1995-98 represents post-harvest treatment. Discussion and Conclusions Limited harvest in DAUs D-3, D-6, D-14, and D-19(GMU 61) have resulted in modest increases in buck:doe ratios overall. However, buck:doe ratios in D-19 (GMU 61) declined despite conservative harvest. There are several reasons why this may have occurred. First of all, GMUs 61 and 62, which together comprise DAU D-19, likely represent a single deer population, at least on summer range. Managing only half of the DAU, or half of this deer population, for limited harvest may have lessened the likelihood that buck:doe ratios would respond positively to the reduced harvest. Other scenarios could also be presented as reasons why differential harvest management in D-19 could influence the observed decline in buck:doe ratios in GMU 61. On a different note, declining buck:doe ratios in 61 could be another indicator of a much larger problem affecting the Uncompahgre deer herd. This is especially true when the declining fawn: doe ratio is taken into account. It is important to point out here that the 1999 buck:doe ratio estimate in GMU 61 was 21.3, which is much higher than has been observed over a number of years. The 1999 buck:doe ratio in GMU 62 dramatically increased as well (27 bucks/100 does). Interestingly, this coincided with the 1999 harvest limitation in GMU 62, which is the first time both GMUs in D-19 have been managed for conservative harvest. When I removed D-19- GMU 61 from the combined analysis of limited harvest units, there was of course a greater increase in buck:doe ratios. With D-19-0-MU 61 removed from the analysis of all DAU data, the interaction between lirnited/unlimi~ed harvest DAUs and pre/post harvest treatment was more pronounced, although not significant (P = 0.106). Limited buck harvest has had no effect on fu.wn:doe ratios. In fact, fu.wn:doe ratios have declined more rapidly in DAUs with limited harvest than those with unlimited harvest. Clearly, other factors of much more importance are affecting these populations. Based on this analysis, it-is very difficult to argue that �169 declining buck:doe ratios are a primary cause for declining fawn recruitment and declining deer numbers. If anything, low buck:doe ratios may be yet another indicator of declining herd productivity. In conclusion, substantial limitations on buck harvest is likely an effective management strategy for increasing the number of bucks in a population. However, buck:doe ratios have increased slowly in these limited harvest units. The judgment as to the effectiveness of this strategy depends on how much hunter opportunity is sacrificed. It seems logical to manage a segment of our DAUs for limited harvest, thereby increasing success rates of those hunters that draw a license while also increasing their chances for shooting a trophy animal. The remaining DAUs should then be managed for much more liberal buck harvest to provide as much hunter opportunity as possible. In other words, a balance needs to be achieved between providing quality hunting (in terms of success rates and trophy buck numbers) and hunting opportunity (in terms of hunter days afield). This will become an increasingly difficult management challenge in the future as Colorado's human population continues to grow along with further inevitable development of habitat. Conversely, limited buck harvest does not appear to be an effective management strategy for improving fawn recruitment or population productivity. For this objective, our management and research efforts must focus on other variables. Graphs presenting the basic data from limited harvest DAUs DAU D-3, GMUs: 6 16 161 17 171 Age and Sex Ratios Before and After Limited Harvest, 1991-98 . . Buck Harvest en a, ~ lfll Fawns:100 Does la Bucks:100 Does 600 500 70 488 56.1 48.1 ca ::c .:.::: u :s m 'ii :s C C < ci> ~ 60 400 50 en 0 ; cu ~ )( -< a, 40 300 u, a, C) 30 C :s 200 20 ::c 10 D.. en 100 0 Before Limited Harvest (1991-94) After Limited Harvest (1995-98) 0 0 �170 DAU D-6, GMU 10 Age and Sex Ratios Before and After Limited Harvest, 1983-98 Ill Buck Harvest -., CD ~ - Fawns:100 Does b~~,d Bucks:100 Does 600 80 522 68.5 70 500 60 C'CJ :::c .x 400 50 u :::, m ci :::, C C <( Cl > <( 300 40 30 200 20 0 .:: ;;_ ,c CD u, CD Cl -., <( C :::, :::c 0 100 0 ., 10 Before Limited Harvest (1983-90) After Limited Harvest (1991-98) a. 0 DAU D-14, GMU 44 Age and Sex Ratios Before and After Limited Harvest, 1985-98 Ill Buck Harvest -., •- Fawns:100 Does Mffil'fl Bucks:100 Does 70 800 700 666 59.1 60 CD ~ C'CJ 600 .x 500 C C <( Cl > <( 50 a:: 40 - 400 ·30 300 ,c u, CD Cl -...., <( C :::, 20 200 10 100 0 C'CJ CD :::, m ci :::, 0 .:: :::c u ., Before Limited Harvest (1985-91) After Limited Harvest (1992-98) 0 :::c 0 a. �171 DAU D-19, GMU 61 Age and Sex Ratios Before and After Limited Harvest, 1985-98 - Buck Harvest en Cl> ~ 1200 - Fawns:100 Does b::::::::::::d Bucks:100 Does 60 53.8 50 1000 C'CS C'CS :::c ~ 40 800 33.3 u :::, m 'ii :::, C C <( CJ) ~ en 0 .::: )( Cl> en 600 30 400 20 200 10 0 0:: Cl> CJ) <( C :::, :::c en 0 a. 0 Before Limited Harvest (1985-91) After Limited Harvest (1992-98) DAU's: D-3, D-6, D-14, D-19(GMU 61) Age and Sex Ratios 4 Years Before and 4 Years After Limited Harvest 1111 Buck Harvest Ill Fawns:100 Does EEi] Bucks:100 Does en Cl> ~ C'CS :::c 2800 70 2400 60 2000 50 0:: 40 - 45.3 m 'ii :::, C C <( C'CS )( 1600 en Cl> CJ) - 30 <( 800 20 :::c 400 10 a. 1200 C :::, 0 en Q ~ .:: Cl> ~ u :::, en 0 Before Limited Harvest After Limited Harvest 0 0 �172 �173 APPENDIXC DRAFf ESTIMATE OF COLORADO'S CURRENT MULE DEER NUMBERS How Many Deer Are There? Chad Bishop, Mammals Researcher Objective Estimate the current number of mule deer in the state of Colorado by extrapolating deer density estimates from quadrat population counts to the estimated amount of winter range in Colorado. Background Our current estimate of the total number of mule deer in the state of Colorado is derived from population models for each deer data analysis unit (DAU). The total number of deer is estimated by summing the individual population projections for each DAU. This is a rough estimate given that a number of the DAU models are built from insufficient data. We cannot appropriately measure all relevant population parameters in each DAU due to budget and logistical constraints. Given this inherent limitation, our current ( 1998) post-hunt population estimate of ~526,000 deer is based on the best information available. However, a number of our constituents believe this figure is too high. In essence, I believe they question our ability to predict total deer numbers from models. Population models can be complex and difficult to understand, particularly to those who lack a statistical or quantitative background. So it comes as no surprise that some sportsmen would doubt our model estimates, particularly when those estimates differ from their own preconceived notions. The purpose of this analysis was to estimate statewide mule deer numbers by extrapolating deer density estimates from DAUs where we have post-hunt population counts to other similar deer winter ranges in the state. Specifically, the number of deer per square mile measured in certain DAUs was applied to the total square miles of deer winter range in Colorado to estimate a total number of deer. In this manner, no models are used and the estimates are based on direct observations of deer/mi2. This procedure also allows us to evaluate whether our current model estimate of deer numbers is reasonable. There are definite weaknesses to this approach. First, estimates of deer density are absolutely relative to the amount of area designated as winter range. In other words, the legitimacy of the analysis depends on how accurately mule deer winter range has been delineated, and more so, whether the standards used to designate winter range were consistent from one biologist to the next around the state. This is a formidable task since the area used by a deer population during winter varies as a function of several environmental variables. Fortunately, similar criteria were used by biologists throughout the state to delineate mule deer winter range, which provided a degree of consistency from one DAU to the next. Another concern is that there is no rigorous way to evaluate whether deer densities measured in one DAU are representative of adjacent DAUs. As with the population modeling approach, this is an inherent limitation since we cannot measure all population parameters in every DAU. The analysis presented here is another approach to estimating the total number of deer in Colorado given the information we have available. The primary benefit of this analysis is that the public should be able to easily understand how the estimate was derived. �174 Methods Winter Range Area The total area (mi 2) of winter range in Colorado was determined for each GMU and DAU from the current mule deer winter range spatial data located on the Colorado Natural Diversity Information Source (NDIS) internet site: http://ndis.nrel.colostate.edu/ndis/ftp html site/statewide ftp.html (file: mule_deerst.tgz). The spatial data layers contained in the file, corresponding to mule deer seasonal activity ranges, were developed through a combined effort among DOW's WRIS (Wildlife Resource Information System) Biologists and Area Terrestrial Biologists. The winter range layer was developed according to the following criteria: 'lhat part of the overall range where 90% of the individuals are located during the average 5 winters out of 10 from the first heavy snowfall to spring green-up, or during a site specific period of winter as defined for each DAU". I used PC Arc/Info to combine the mule deer winter range spatial layer with the GMU-DAU spatial data layer (also obtained from the NDIS website), and then used the combined data layer in ArcView to estimate square miles of deer winter range by GMU and DAU (Fig. 1, Table 1). }o~,, l_E.-,.,,,.__;.:,:_ir--,....,......_· .. ~ · ,: gJ ·., ~~t~: D-47 D-46 D-28 Rm Winter Range Figure 1. Spatial distribution and area of delineated mule deer winter range by deer DAU in Colorado. �175 Table 1. Delineated mule deer winter range (W.R.) in square miles by GMU and DAU in Colorado. Deer Deer DAU GMU W.R DAU GMU (mi2) 1 2 201 Total 3 301 4 441 5 14 214 Total 6 16 161 17 171 Total 7 8 9 19 191 Total 87 88 89 90 93 95 97 98 99 100 101 102 Total 10 11 211 12 13 131 231 22 23 24 Total 32 1033 125 1190 533 334 156 64 32 1 2 1122 190 40 119 85 38 473 31 138 189 138 291 787 70 211 275 319 286 263 347 1038 741 298 522 373 4744 615 581 295 94 89 9 19 750 175 1 2014 15 35 36 45 Total 18 181 27 28 37 371 Total 20 21 30 Total 41 421 42 Total 43 47 471 Total 44 48 481 56 561 Total 49 57 58 581 59 Total 39 46 51 461 Total 40 61 62 Total W.R Deer DAU GMU W.R Deer DAU GMU (mi2) (mi2) 53 111 54 167 74 m@t».:t!HM 55 551 0 353 Total 60 85 70 96 71 69 711 83 83 Total 66 9 67 425 543 Total 68 608 397 681 1005 Total 118 29 38 104 317 Total All 539 143 72 73 13 Total 0 75 156 751 128 77 33 771 91 78 126 51 Total 301 83 218 85 851 255 140 492 568 Total All 278 2024 69 42 84 47 86 861 250 78 Total 80 459 81 488 626 Total 76 683 79 1308 Total .124 280 114 500 185 501 299 Total 310 63 1012 64 7 65 329 Total 1347 it.w.Uti 31 205 32 228 Total 433 33 257 25 230 26 487 34 227 Total 198 91 425 92 0 94 951 0 248 96 248 Total 301 150 102 302 t88.N All 317 105 1171 277 106 930 Total 264 511 512 116 1309 Total 0 ,@I».5tm 411 52 171 521 921 153 Total 151 lOfStm 74 1589 741 231 Total 260 491 63 252 '315 W.R (mi2) 11 219 490 230 81 185 266 261 126 387 187 77 62 31 170 63 72 92 87 234 547 37 227 212 658 96 28 124 62 107 63 232 33 467 499 �176 Mule Deer Density Estimates Conceptually, it is simple to estimate the total number of deer statewide by extrapolating several deer density estimates to the total amount of Colorado winter range: calculate a mean deer density and multiply it by the total square miles of winter range. However, we can be fairly certain that deer density varies across Colorado, and so using a single mean density may not be the best approach. Ideally, deer DA Us should be grouped according to ecological similarities such that deer density is somewhat similar across all winter range in the DAU grouping. Density estimates measured in one DAU would then be representative of the other DA Us in the group. The inherent limitation is that density estimates are needed for at least one of the group's DA Us. Thus, these DAU groupings are dependent on data availability. Mule deer density estimates were obtained from the DEAMAN database (Table 2). These estimates are not corrected for sightability bias, and therefore should represent conservative estimates of the true deer density. I took a several tiered approach to estimate statewide deer numbers from the existing density data. My first objective was to delineate groups of DA Us based on ecological similarities, where at least one DAU in the group had some density estimates available over the past 15 years (Fig. 2). The 15-year time limit was arbitrarily set based largely on available data. For each DAU with density estimates available, I calculated a high, low, and overall mean for the 15-year period. The high mean was calculated from the upper half of recorded density estimates while the low mean was calculated from the lower half of density estimates. The overall mean was naturally calculated using all density estimates recorded in the DAU during the 15 years. In several of these DA Us, only 2 density estimates were available during the 15-year period (D-2, D-6, D-12, and D-42). D-2 and D-12 were the only source of density data for their respective DAU groups, which is a limitation of this first objective analysis. DA Us D-6 and D-42 were in the same DAU group as D-7, which had 6 density estimates available. Density estimates from D-42, which were much higher than those from D-6 and D-7, were applied only to winter range in D-42 and not incorporated into the mean for the DAU group. D-50, which incorporates the Air Force Academy, had only one density estimate available. It was not grouped with any other DAUs, though, because it typically has had higher deer densities than areas adjacent to it. Figure 2 depicts these DAU groupings, and also shows which DA Us provided density data to estimate deer numbers. Table 2 lists the actual density data. This approach enabled me to keep DAU groupings small and more ecologically meaningful, to maximize my use of density data available in DEAMAN, and to estimate a minimum and maximum number of deer for the past 15-year period. The primary drawbacks of this approach were twofold: 1) only limited data was available for some DA Us ·and 2) resulting estimates of total deer numbers do not represent current numbers. My second objective was to use density estimates from only those DA Us with recent data(~ 1995), which primarily corresponded to the intensive deer population monitoring areas (D-4: Red Feather, D-9: Middle Parle, D-16: Cripple Creek, and D-19: Uncompahgre Plateau) (Fig. 3). D-24 in southwest . Colorado also had recent density data available. This required reducing the number of DAU groups, with each group now containing more DA Us. Again, I attempted to group DA Us based on ecological similarities given the obvious constraint that most DAUs do not have any recent data available. The purpose of this analysis was to obtain a current estimate of total deer numbers using recent density estimates while maintaining DAU groupings that are at least somewhat similar from a habitat and deer ecology standpoint. My final objective was to apply a single mean deer density, calculated from recent estimates only(~ 1995), to the total winter range west ofl-25 (Fig. 4). This approach also provides a current estimate of deer numbers in Colorado, but removes any DAU groupings. In all of these analyses, I have focused only on deer populations west ofl-25. There are no density estimates from quadrat counts available for deer populations east ofl-25, yet we would anticipate deer densities to be quite different in a plains landscape. Therefore, in DA Us east ofl-25 that have mule deer winter range delineated, I applied density estimates from the current DAU population models. This was �177 done only to complete my estimate of statewide deer numbers based on delineated winter range. I realize that applying density estimates from biologists' population models directly biases my analysis as an independent estimate. But again, it was only done east ofl-25, which represents a small portion of the state's total mule deer numbers. Also, there is no winter range delineated in deer DAUs 28, 33, 45, 46, 47, and 48, although mule deer are present. I did not include these DA Us in any of my estimates of total deer numbers. Table 2. Mule deer density estimates (deer/mi2) measured during quadrat population counts in various DAUs. Estimates were obtained from the DEAMAN 3.2 database. Year 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 D-2 13.88 Low Mean High Mean Overall Mean Recent(:2: 1995) 13.88 73.46 43.67 D-4 D-6 28.35 32.14 31.17 28.53 46.73 Density Estimate bv DAU" D-19b D-9 D-12 D-16 7.55 25.83 21.51 D-24 D-42c D-50<1 22.78 9.89 35.73 47.64 37.97 25.18 70.09 30.82 10.86 19.16 27.07 17.02 73.46 D-7 32.67 32.57 32.23 149.16 71.96 12.57 8.33 6.96 3.45 46.83 17.58 24.40 14.21 5.19 22.36 24.66 22.28 33.48 19.02 3.5 25.38 35.57 30.47 22.32 17.02 27.07 22.05 6.25 34.40 20.33 20.34 31.29 26.49 29.07 34.17 46.83 47.64 47.24 4.35 9.43 7.40 4.35 19.45 34.17 22.39 34.17 18.30 26.80 22.55 19.02 71.96 149.16 110.56 70.09 70.09 70.09 8Density estimates from all DAUs included in the table were used in the first objective analysis. Only those DAUs with recent (:2: 1995) density data were used in the second and third objective analyses. 1,n D-19, the delineated winter range was modified prior to the 1998 quadrat count to make it more representative of observed wintering deer distribution, which decreased the size of the winter range by 446 mi2. Thus, the density estimate measured in 1998 was based on a smaller winter range area than the densities measured prior to 1998. "For the first objective analysis, I assigned D-42 to the same DAU group as D-6 and D-7 (Fig. 2). However, because density estimates for D-42 were much higher than those for D-6 and D-7, they were applied only to winter range in D-42. Density estimates from D-6 and D-7 were used to calculate a mean deer density for this particular DAU group. dFor the first objective analysis, D-50 was not grouped with any other DAUs (Fig. 2) because it typically has had higher deer densities than areas adjacent to it. The single density estimate for D-50 was applied only to winter range in D-50. �178 Figure 2. First-level grouping of deer DAUs shown by like-patterned blocks. Density estimates based on quadrat counts were obtained from the labeled DAUs to estimate the number of deer in each DAU block. Current population models were used to derive density estimates east ofI-25. Figure 3. Second-level grouping of deer DAUs shown by like-patterned blocks. Density estimates based on quadrat counts were obtained from the labeled DAUs to estimate the number of deer in each DAU block. Current population models were used to derive density estimates east ofI-25. �179 j ~-iO M'U:li!!! ,Oe~r :R,ii!in-0,,;---··· ---~-..c:,.J \ \ ~~=•~ h:!r ! II I -~ ~W7d Figure 4. Third-level grouping of deer DAUs shown by like-colored blocks. Density estimates based on quadrat counts were obtained from the labeled DAUs to estimate the total number of deer in all DAUs west of 1-25. Current population models were used to derive density estimates east ofl-25. Results First Obiective My first estimates of statewide deer numbers incorporated measured density estimates from the past 15 years. I used the low mean density estimates from each DAU to calculate a minimum mean estimate of statewide deer numbers: 421,894 deer (408,486 west"ofl-25). I used the high mean density estimates from each DAU to calculate a maximum mean estimate of statewide deer numbers during this 15-year time period: 807,472 deer (794,065 west ofl-25). My final calculation incorporated the overall mean density estimates from each DAU over the 15 years: 608,608 deer (595,200 west of 1-25). Second and Third Obiectives Using DAU groupings (Fig. 3) based on recent(:.!: 1995) density data, the estimate of total deer numbers currently in Colorado is 575,912 deer (562,505 west of 1-25)). Applying a single mean density estimate to all DAUs west ofl-25, the estimate of total deer numbers currently in Colorado is 581,331 deer (567,923 west ofl-25). Discussion The purpose of this "analysis" was to develop an independent estimate of total deer numbers in the state of Colorado that did not incorporate any modeling, but rather application of measured density estimates obtained from direct observation during aerial quadrat counts. The analysis could certainly have been done· differently, particularly in regards to how the DAUs were grouped together; but it appears that any reasonable method using the available density estimates would have resulted in relatively similar figures. First Obiective I used this 15-year data set to incorporate density estimates from a greater number ofDAUs, and to place upper and lower bounds on the possible deer numbers in the state. Using a mean of the lowest density �180 estimates measured during the past 15 years in various DA Us, we would still estimate there to be at least 400,000 deer in Colorado. Conversely, using a mean of the highest density estimates measured, we wouldn't expect the total population to have been much more than 800,000 deer in the recent past. Second and Third Ob;ectives The second and third objectives dealt with recent density estimates, which was my primary focus. I estimated there to be approximately 575,000- 580,,000 deer currently in Colorado, and approximately 560,000- 570,000 deer west ofl-25. The 1998 estimate of total deer numbers based on summation of individual DAU population models is approximately 525,000 deer (actual estimate: 526,406), and approximately 495,000 deer west ofl-25 (actual estimate: 496,603). The estimates are surprisingly similar given the large degree of error involved with either method of estimating Colorado's total deer numbers. In my perspective, these 2 independent estimates provide the Division of Wildlife with greater confidence that 500,000 to 600,000 deer is a reasonable estimate of the state's total mule deer population. At the very least, this analysis suggests that our population models are not drastically overestimating deer numbers, which is what some of our constituents have claimed. As I stated in the methods section, I did not include several southeast Colorado DAUs (D-28, D-33, D-45, D-46, D-4 7, D-48) in my statewide analysis because winter range had not been delineated. The total number of deer from 1998 population models for these DA Us is 16,695 deer. To make the DAU model estimates more comparable to my estimate, I subtracted these deer from the statewide total (526,406 16,695 = 509,711). Thus, with the southeast plains DAUs excluded, population models indicate approximately 510,000 deer, as compared to my estimate of 575,000-580,000 deer. This further suggests that we are not overestimating deer numbers using DAU population models. I also used another approach to evaluate the legitimacy of our population models. I calculated a mean deer density per square mile of winter range using the DAU model estimate of total deer numbers. This provides another check on the accuracy of the DAU modeling approach because the resulting mean deer density should be in the ball park of density estimates that have actually been measured from quadrat counts. There are 32,230 square miles of deer winter range delineated in Colorado (which excludes the southeast plains DA Us described above). The 1998 DAU model estimates would indicate a statewide mean density of 16 deer per square mile (510,000 deer/32,230 m.i2). This is clearly a reasonable density estimate, which lends further credence to the accuracy of the DAU population models. I also calculated the mean deer density for DAUs west ofl-25 only. There are 26,070 square miles of deer winter range delineated west ofl-25, and 1998 DAU population models indicate approximately 495,000 deer west ofl25. This gives a mean deer density of 19 deer per square mile for DAUs west ofl-25, which is also quite reasonable. The mean deer density measured from recent(~ 1995) quadrat counts in several deer DA Us west ofl-25 (D-4, 9, 16, 19, 24) is 22 deer/m.i2 . Summary Our current (1998) DAU population model estimate of statewide deer numbers is roughly 525,000 deer (526,406). When the southeast plains deer DAUs are excluded (D-28, 33, 45, 46, 47, 48), this estimate reduces to approximately 510,000 deer (509,711). My estimate, calculated by extrapolating measured deer densities to the total amount of deer winter range, which also excludes these southeast plains DAUs, is 575,000-580,000 deer. Considering only DAUs west ofl-25, our population models indicate there are approximately 495,000 deer (496,603)~ while my estimate using measured deer densities in select DAUs indicates there are approximately 560,000-570,000 deer west ofl-25. This latter comparison is more appropriate because the estimates are completely independent of one another. The DAU population models indicate that the mean deer density on winter range west ofl-25 is 19 deer/mi2. The mean deer density measured from recent(~ 1995) quadrat counts in several deer DAUs west ofl-25 (D-4, 9, 16, 19, 24) is 22 deer/m.i2 . The Division's current estimate of deer numbers from DAU population models is reasonable, if not conservative. It appears justifiable to assume there are roughly 500,000 to 600,000 mule deer in Colorado. � https://cpw.cvlcollections.org/files/original/617928ca0c2442937a5a268eef5ccce6.pdf 99de95b5e4bd63e6fa8862aab289f4f0 PDF Text Text 125 Colorado Division of Wildlife Wildlife Research Report July 1999 JOB PROGRESS REPORT Smteof __ ~ ~C~o~lo~r=ad~o~ _ Cost Center 3430 Mammals Program Project No. W..:...:..,_-1=5=3--=-R=-_",1=-2_,.....-_ Work Package No. ___,3::....:0=0~1 _ Deer Management Task No. -'4'-_ Regulation of Mule Deer and Elk Population Growth by F ertilitv Control Period Covered: July 1, 1998 - June 30, 1999 -Author: Dan L. Baker Personnel: T. M. Nett, M. A. Wild ABSTRACI' We conducted preliminary investigations to evaluate the effectiveness and side effects of two contraceptive agents in captive mule deer and elk. We treated female mule deer with a permanent fertility control agent (GnRH-PAP) and female elk with a temporary, reversible contraceptive. In mule deer, serum luteinizing hormone (LH) levels were reduced by 55-78% for at least 6 months following treatment. For elk, serum LH concentrations were reduced to baseline levels for 90 days posttreatment. Body weight dynamics, blood chemistry, hematology, and social behavior of female mule deer and elk appeared normal compared to untreated animals. �127 REGULATION OF MULE DEER POPULATION GROWTH BY FERTILITY CONTROL . Dan L. Baker P. N. OBJECTIVES 1. To develop a practical and acceptable technology to inhibit reproduction in mammalian species which cause damage or constitute a significant public nuisance. 2. To demonstrate the feasibility of such technology in a field application. 3. To predict population impacts of alternative contraceptive regimens using simulation modeling. SEGMENT OBJECTIVES 1. To develop and test GnRH-toxin conjugate in captive mule deer and elk. 2. To develop a si~ulation populations. model to evaluate the ability of contraceptives to regulate large ungulate INTRODUCTION Segment Objective 1: GnRH-ToxinlAgonist Controlling the abundance of animals is fundamental to contemporary wildlife management. This is particularly true for wild ungulates. The most compelling motivation for regulating ungulate numbers is that overabundance causes problems that can be biological, economical, or political in scope (Jewel and Holt, 1981). Resolving these problems requires controlling population growth. Wild ungulate populations have traditionally been regulated by influencing death rates using controlled harvest or CUlling. However, there are an increasing number of circumstances where these traditional methods are not feasible. The use of fertility control to decrease birth rates is one of the most promising approaches to the long-term control of overabundant wild ungulates. During the past decade, research aimed at developing effective contraceptives for free-ranging wildlife populations has accelerated. These efforts have resulted in development and testing of a wide variety of potential contraceptive agents (Kirkpatrick and Turner 1985, Warren et aI. 1995). One of the most promising new non-steroidal, non-vaccine, approaches to contraception involves synthetic analogs of gonadotropin-releasing hormone (GnRH). GnRH is a molecule produced in the hypothalamus of the brain. It directs specific cells in the pituitary gland to synthesize and secrete two important reproductive hormones; follicle stimulating hormone (FSH) and luteinizing hormone (LH). These latter two hormones, known as gonadotrophs, control the proper functioning of the ovaries in the female and testes in the male. Analogs of GnRH have the potential to either permanently or temporarily inhibit reproduction. For most free-ranging wild ungulate applications, permanent sterilization or a combination of permanent sterilization and culling have been proposed as the most efficacious approaches to population management (Hone 1992, Garrot 1995, Hobbs et al. 1999, in press). For this application, �128 superactive analogs ofGnRH are coupled to a cytotoxin. The GnRH-toxin conjugate specifically targets the gonadotroph cells and permanently inhibits the ability of the cell to secrete FSH and LH. This approach has several potential advantages over other methods of contraception. These include: 1) a single treatment should permanently sterilize an animal 2) the same treatment should be effective in both males and females and in different mammalian species 3) GnRH-toxin conjugate will be metabolized from the body within a few days of treatment 4) the proteinaceous nature of GnRH-toxin conjugate eliminates the possibility of passage through the food chain. 5) the small volume required for effective contraception would facilitate microencapsulation and administration by syringe dart or biodegradable projectiles. In other situations where wildlife managers need to maintain flexibility in the use of fertility control, reversible contraception may be desirable. Examples of these situations include 1) wild ungulate populations exposed to periodic, severe, unanticipated winter mortality, 2) populations with low genetic variability, 3) populations that cannot be effectively monitored, 4) populations where public attitudes are opposed to permanent contraception, or 5) populations where non-lethal hunting recreation is a primary management objective. In these situations, superactive analogs of GnRH without the toxin subunit would be more appropriate. The inhibitory actions of long-term GnRH analog agonist on the ovulatory cycle of humans and other mammals is well-established (Casper and Yen 1979, Fraser 1983, Fraser et al. 1987, Concannon et al. 1991). Constant administration of high .doses of GnRH agonist results in down regulation of the pituitary GnRH receptors and suppression ofsecretionofLH andESH. Continued· treatment suppresses LH secretion, preventing the maintenance of normal.luteal function, and thus prevents viable pregnancy. Inhibition of ovulation caused by chronic administration of GnRH agonist has been successful in several species, including dogs (Vickery et al. 1989), cattle (Herschler and Vickery 1981), sheep (McNeilly and Fraser 1987), white-tailed deer (Becker and Katz 1995), and elk (Baker and Nett, 1999). Evidence from studies on pituitary receptors and gonadotropin content in experimental animals treated by long-term infusion ofGnRH agonist shows that sustained release is themost effective approach for temporarily suppressing pituitary-gonadal function (Clayton 1982, Sandow 1982). The practicality of this approach, however, is dependent upon development of a long-acting, slow-release preparation of agonist that can be remotely delivered. Recently, a practical mode of administration using subcutaneous implants has overcome the need for constant mechanical infusion of the analog. Slow release formulations of superactive GnRH agonist are now commercially available and have been shown to be effective in suppressing the pituitary ovarian axis for up to 6 months in a variety of mammalian species (Fraser et al. 1987, Asch et al. 1985). To our knowledge, only limited investigations have been conducted with either of these fertility control techniques on wild ungulates (Baker 1994, Baker et al. 1995, Becker and Katz 1995). Thus the specific objectives of these investigations were: 1) To evaluate the effectiveness and duration ofa single dose application of GnRH-toxin conjugate and GnRH - Agonist in preventing normal production of reproductive hormones in captive mule deer and elk. 2) To evaluate the effects of these contraceptives on the general health, blood chemistry, and hematology in mule deer and elk. �129 Segment Objective 2: SimulationModeling Using either culling or contraceptives to meet deer population objectives will require, wildlife managers to choose specific tactics of treatment Choices must be made on the number and age to treat, frequency of treatment, species, sex, etc, Decisions on the best tactics will depend on comparing the effects of alternative management actions on population behavior. To assist in these decisions, we developed an interactive simulation model of deer population dynamics to allow managers to evaluate alternative treatment regimes on simulated populations before applying them to real animals. Critical to model performance is knowledge of population demographics of the RMA deer herd and requires information on sex and age composition, recruitment, pregnancy rates, fetal sex ratio, and estimates of population density. This information together with knowledge of the habitat resources available to deer will provide a sound biological basis for management of deer populations at the RMA. METHODS AND MATERIALS GnRH- Toxin! Agonist Experiment 1: Evaluation of GnRH-PAP in mule. deer. Previously, we conducted controlled experiments with captive mule deer to determine the most effective dose of GnRH analog and the season of the year when treatments would be most effective in preventing fertility (Baker 1997). Here, we evaluate the effectiveness of GnRH..,P AP in preventing normal production of luteinizing hormone (LH) and the duration of effectiveness. We will evaluate contraceptive effectiveness in 4 ovariectomized mule deer. . Protocol for Ovariectomy in Mule Deer Tonic secretion of pituitary LH is the result of an interplay between a stimulatory input from the brain and an inhibitory feedback from the gonads. In the intact female, estradiol secreted from the gonads is a potent negative feedback hormone on LH secretion during anestrus (Goodman and Karsch 1980). Since measurement ofLH secretion is the primary indicator of GnRH-toxin conjugate effectiveness, it is imperative that female mule deer in this experiment be ovariectomized. Ovariectomies of mule deer were conducted at FWRF during the week of May 11, 1998. Deer were isolated and fasted for about 24 hr prior to surgery to alleviate regurgitation and aspiration of rumen contents. On the day of surgery, anesthesia was induced with intramuscular (IM) administration of 100 mg xylazine w/o 500 mg ketamine depending on response of the animal. Deer were then intubated and surgical anesthesia was maintained using a rebreathing circuit with isoflurane. Anesthetized deer were prepared for surgery by clipping hair in the abdominal area. They were then carried to a designated surgery area., placed in right lateral recumbency, and the surgical site prepared using standard surgical scrub. The surgical area was prepared with sterile towels and a surgical drape. Ovariectomy was performed via mid-ventral laparotomy and require 20 - 30 min per animal. Surgeons were attired in a sterile gown, mask, cap, disposable shoe covers and sterile gloves. Surgical assistants wore cap, mask, and disposable shoe covers. The ovarian artery was ligated prior to removal of the ovary. The incision was closed using a continuous suture and the skin closed with interrupted mattress stitches. We reversed anesthesia with yohimbine at a dose of 0.2 mg/kg (N). To minimize infection, deer were given ceftiofur sodium (1.1 mg/kg IV) administered perioperatively. Phenylbutazone (4 mglkg) was administered orally when deer were partially recovered from anesthesia �130 and every 48 hr for up to one week, if needed. Animals were placed in isolation pens for 24 hr and were observed every 5 min during recovery from anesthesia. This procedure followed ARBL Standard Operating Procedure #1 - Protocol for OvariectomylLuteectomy in Ewes and was modified for use in mule deer at FWRF. GnRH-toxin Conjugate Protocol Approximately 6 weeks following ovariectomy, GnRH-toxin was administered. Four ovariectomized deer were moved from 5 ha pastures to individual isolation pens, sedated with xylazine (100 mg 1M), and administered IV an optimum dose of GnRH-toxin (3 J-lgl50 kgBW). Deer were then placed in individual isolation pens and fitted nonsurgically with indwelling jugular catheters. Indwelling catheters remained in deer for up to 6 weeks and were checked and flushed daily. Deer remained in individual isolation pens for the first 6 weeks of the trial. This minimized tranquilization and general handling of deer and allowed daily evaluation of intake and general health of experimental animals. After 6 weeks, catheters were removed and deer were returned to 5 ha pastures. For subsequent trials, deer were removed from 5 ha pastures one day prior to GnRH trials and fitted with indwelling jugular catheters. Sampling was conducted the next day, catheters removed and deer returned to 5 ha pastures. One week after GnRH-toxin treatment, GnRH analog challenge trials were conducted. Experimental animals were moved to individual isolation pens and 3 J-lgl50 kg BW GnRI:i analog (Baker 1997) was administered through the indwelling cannula Blood samples (5ml) were collected at 0,30,60,90, 120, 180,240,300, and 360 min postinjection. After collections, blood was held at 4 C for 24 hours and serum obtained by centrifugation. Serum was stored at -20 C until analyzed. GnRH ' analog challenge trials were repeated each weekfor 6 weeks, then twice monthly for 3 months, then _' once monthly for 1 year. . Serum concentrations ofLH were quantified by means of ovine LH RIA (Niswender et al. 1969). The limit of sensitivity of the assay is 0.4 nglml. Response of the pituitary to GnRH-toxin conjugate was be assessed by 1) maximum LH response achieved postinjection, 2) total amount ofLH secreted (nglml/min); estimated by calculating the area under the LH curve, and 3) an LH response of < 2.5 ng/ml following GnRH challenge was be considered sufficiently low to prevent ovulation. Experiment 2: Evaluation of GnRH- Agonist (Lupron) in elk. Here, we conducted a pilot experiment to evaluate the effectiveness of GnRH agonist analog in suppressing ovarian function in elk. We determined the most effective dose and evaluated the duration of effectiveness. We also used these studies to evaluate the safety and side effects (if any) of treatment on a small number of animals before proceeding to a larger investigation. The information collected in the pilot study was used to assess individual variation in treatment responses and to conduct statistical power analysis to increase efficiency of future research efforts, We conducted controlled experiments with adult, tame female elk at the Colorado Division of Wildlife's Foothills Wildlife Research Facility, Fort Collins, Colorado during November 1998 to March 1999, Objective: To determine the amount of GnRH agonist analog that will induce half-maximal release of LH in female elk during estrus and evaluate duration of effectiveness. Design: We observed the LH response offemale elk to three levels ofGnRH agonist (0, 19,39, and 78 J-lglhour) in a completely randomized design with two animals per level. Based on results from previous studies, we chose a range of doses that should include the biologically effective range for elk. �131 We evaluated the effective duration of treatments by monitoring ovarian function throughout the remainder of the breeding season (Nov - Apr). Dosage Trial - This experiment was conducted during the breeding season to insure that the pituitary gland was at its most active state when stimulated by the GnRH agonist. On Day 1 of the experiment, 9 female elk were moved from 5 ha pastures to individual isolation pens, sedated with xylazine hydrochloride (50-100 mg/animal, IM), fitted nonsurgically with indwelling jugular catheters and administered subcutaneously one of three experimental doses of GnRH agonist analog. The sampling period began at 1000 h. Blood samples (5 ml) were collected at 0,60, 120, 180,240,300,360,480 min, 12,24,36,48,84, and 240 h postinjection. Animals remained in isolation pens during the 10 day trial. Elk were observed daily for side effects of treatments (if any) and catheters flushed with sterile saline solution. Following the last blood collection, catheters were removed and animals returned to 5 ha paddocks. After collection, blood was held at 4 -c for 24 h until serum was obtained by centrifugation. Serum was then stored at - 20°C until analyzed for LH. Serum concentrations ofLH was quantified by means of ovine LH RIA (Niswender et al. 1969). The limit of sensitivity of the assay was 0.4 ng/ml. Analysis: Responsiveness of the pituitary to GnRH agonist analog challenge was assessed in four ways: 1) maximum LH response achieved postinjection minus baseline 2) time required to reach maximum LH 3) total amount ofLH secreted (ng/mllmin) 4) the amount ofLH required to induce half maximal release ofLH (EDso). Response to treatments was analyzed with a one way analysis of variance for a completely random design with 3 levels of dose as treatments and 3 individual animals as replications. Simulation Model We developed an interactive simulation model to allow RMA biologist to evaluate alternative mule deer culling strategies. This model combines extant knowledge of mule deer population biology with measured population parameters at the RMA to predict the proportion of male and female deer that need to be removed each year to accomplish management objectives for this herd. Table 1. Principal parameters in culling simulation model for mule deer at RMA. Definition Value Measured Adult/yearling female survival Adult/yearling male survival Fawn survival coefficient a Fawn survival coefficient 13 Fetal sex ratio ('Yo males) December recruitment 0.90 0.90 1.20 -0.053 0.47-0.62 0.59-0.72 No No No No Yes Yes Reference Whittaker, 1992 Whittaker, 1992 Bartmann et aI., 1992 Bartmann et aI., 1992 �132 RESULTS AND DISCUSSION GnRH Toxin/Agonist Experiment 1: Evaluation of GnRH-PAP in mule deer. Mean serum LH concentrations were reduced in all experimental animals compared to pretreatment levels following treatment with GnRH-PAP (Fig 1). Mean levels ofLH (ng/ml) were initially reduced 72% of pretreatment levels after 15 days, then increased to about 51% (se ± 0.75) after 33 days and have remained relatively unchanged through 100 days posttreatment. Body weight dynamics, blood chemistry, hematology and social behavior of treated deer appear normal compared to non-treated female deer. 125 100 I - ...J c: II) E 75 C"d II) •... II) •... a. - 50 0 ~ 25 o Pre Toxin 33 15 48 100 Days Figure 1. Luteinizing hormone release in ovariectomized female mule deer treated with GnRH-PAP during the breeding season. Experiment 2: Evaluation of GnRH-Agonist in Rocky Mountain Elk. Administration of GnRH-Agonist implants resulted in a prompt and expected increase in serum LH levels in all female elk. We observed an average peak LH response of 15.52 ng/ml (se ± 0.94) which occurred approximately 3.5 h posttreatment. Peak response (ng/ml) and time to peak (h) were not different (p > 0.01) among treatment levels. Serum LH levels returned to baseline after 12 hours for all elk. Subsequent challenge with an intravenous injection of 1 f.,l GnRHI50 kg BW at 35,75,110, and 130 days posttreatment resulted in a significant (p < 0.02) increase in LH levels of control elk compared to elk treated with GnRH-Agonist implants (Fig. 2). We did not observe a significant (p < 0.001) treatment difference among doses of GnRH-Agonist. Lack of treatment effect was not due to high variation among animals but instead to small differences among treatments responses. Serum LH levels of treated elk remained at baseline (0.34 ng/ml) for the duration of the experiment (130 days). These findings are consistent with other evidence for desensitization of the pituitary gonadotrophs due �133 --f- Control - -is - Low --0-- Mad ...+... High 25 20 E 0; c 15 :r: ...J ~ w 10 a.. 5 o o 35 75 110 130 TIME (DAYS) Figure 2. Effects of subcutaneous administration otGnRH-Agonist on serum LH concentrations in elk. Doses were: control = 0, low = 32.5 mg, medium = 65 mg, and high = 130 mg. to prolonged and sustained stimulation. The results of this study demonstrate that one subcutaneous implant containing at least 32.5 mg of GnRH-Agonist can suppress pituitary-ovarian function for at least 90 days in elk. Simulation Model . Professional culling of adult mule deer at the RMA was implemented on a prescribed basis beginning in November 1994. Prior to 1994, the mule deer population was increasing at a rate of approximately 20 % per year. Since 1994, annual culling has reduced the population growth rate to 4% per year and allowed resource managers to meet deer management objectives for this herd. To meet these objectives has required that resource managers remove approximately 10- 15% of the adult females and 6-10% of the young males from the population each year. These estimates vary each year depending on productivity of the population, changes in sex and age ratios, and deer habitat resources. The mathematical model used to estimate the number of animals that must be culled annually to maintain a target, steady state population appears to provide a reasonably accurate representation of mule deer population dynamics at the RMA.· The efficacy of this model, however, is dependent on intensive monitoring of the population and adapting management actions to a changing environment. LITERATURE CITED Asch, R H, F. J. Rojas, T. R Tice, and A. V. Schally. 1985. Studies of a controlled - release microcapsule formulation of an LH-RH agonist in the rhesus monkey menstrual cycle. International J. Fertility 56:78-81. Baker, D. L. 1997. Regulation of mule deer population growth by fertility control: laboratory, field, and simulation experiments. Pages 81- 87 in Wildlife Research Report, Mammals Research, �134 Federal Aid Projects, Job Progress Report, Project W-153-R-4, SP1, Jl. Colorado Division of Wildlife, Fort Collins, Colorado, USA. Baker, D. L., and N. T. Hobbs. 1996. Regulation of mule deer population growth by fertility control: laboratory, field, and simulation experiments. Pages 113 -148 in Wildlife Research Report, Mammals Research, Federal Aid Projects, Job Progress Report, Project W-153-R-4, SPl, J1. Colorado Division of Wildlife, Fort Collins, Colorado, USA. Baker, D. L, M. W. Miller, and T. M. Nett. 1995. Gonadotropin-releasing hormone analog-induced patterns of luteinizing hormone secretion in female wapiti (Cervus elaphus nelson i) during the breeding season, anestrus, and pregnancy. Bio. Reprod. 52: 1193-1197. Becker, S. E., Katz, L. S. 1995. Effects of gonadotropin - releasing hormone agonist on serum LH concentrations in female white-tailed deer. Small Ruminant Research 18: 145-150. Casper, R. F., and S. S. C. Yen. 1979. Induction of luteolysis in the human with a long-acting analog of luteinizing hormone-releasing factor. Science 205 :408-41 O. Clayton, R. N. 1982. GnRH modulation of its own pituitary receptors: evidence ofbiphasic regulation. Endocrinology 111: 152-161. Concannon, P. W., and V. N. Meyers-Wallen, 1991. Current and proposed methods for contraception and termination of pregnancy in dogs and cats. 1. Am. Vet. Med. Assoc. 198:1214-1225. Fraser, H. M. 1983. Effect of treatment for one year with a luteinizing hormone-releasing hormone agonist on ovarian, thyroidal, and adrenal function and menstruation in the stump tailed monkey (Macaca arctoides). Endocrinology 112:245-253. __ , M., 1. Sandow, H. Seidel, and W. von Rechenberg. 1987. An implant of a gonadotropin releasing hormone agonist (burserelin) which suppresses ovarian function in the macaque for 3-5 months. Acta Endocrinological 15: 521-427. Garrot, R. A. 1995. Effective management of free-ranging ungulate populations using contraception. Wildlife Society Bulletin 23:445-452. Goodman, R. L., and F. 1. Karsch. 1980. Pulsatile secretion of luteinizing hormone: differential suppression by ovarian steroids. Endocrinology 107:1286-1290. Herschler, R. C., and B. H. Vickery. 1981. The effects ofLHRH ethylamide on the estrous cycle, weight gain, and feed efficiency in feedlot heifer. Amer. 1. of Vet. Res. 42: 1405-1408. Hobbs, N. T., D. L. Baker, and R. B. Gill. 1999. A general theory describing effects offertility control on populations of ungulates. Journal of Wildlife Management (in press). Hone, 1. 1992. Rate of increase and fertility control. 1. Applied Ecology 29:695-698. Jewell, P. A., and S. Holt. 1981. Problems in management oflocally abundant wild animals. Academic Press. 321pp. Kirkpatrick, 1. F., and 1. W. Turner, Jr. 1985. Chemical fertility control and wildlife management. Bioscience 35:485-491. McNeilly, A. S., and H. M. Fraser. 1987. Effect of gonadotropin-releasing hormone agonist-induced suppression ofLH and FSH on follicle growth and corpus luteum function in the ewe. 1. Endocrinology 115:273-282. Niswender, G. D., L. E. Reichert, Jr., A. R. Midgley, Jr., and A. V. Nalbandov. 1969. Radioimmunoassay for bovine and ovine luteinizing hormone. Endocrinology 84: 1166-1173. ___ . 1973. Influence of the site of conjugation on the specificity of antibodies to progesterone. Steroids 22:413-424. Sandow,1. 1982. Inhibition of pituitary and testicular function by LHRH analogs. Pages 19-39 in Jeffcoate S. L. and Sandlier (eds). Progress towards a male contraceptive. Wiley and Sons, Chichester. Warren, R. 1., L. M. White, W. R. Lance. 1995. Management of urban deer populations with contraceptives: practicality and agency concerns. Wildlife Society Bulletin 23: 441-444. � https://cpw.cvlcollections.org/files/original/1040e3b2c3c260e7bf139285e9591ed7.pdf 693594c505c12c29068f49a3d60674b9 PDF Text Text 135 Colorado Division of Wildlife Wildlife Research Report July 1999 JOB PROGRESS REPORT Smteof ~C~o~lo~r~ad~o~ _ ,Project No. W.!.!--1~5~3~-R~-..!..12~ __ Work Package No. __ ~3~0:..:::0..!..1 _ -=5:..__ _ Task No.' Cost Center 3430 Mammals Program Deer Conservation Investigating Factors Contributing to Declining Mule Deer Numbers Period Covered: July 1 1998 - June 30, 1999 Authors: T. M. Pojar and W. F. Andelt '" Personnel.. R Arant, D. L. Baker, B. Banulis, D. Hartmann, G. Bock, D. C. Bowden, M. Caddy, D. Coven, B. Devries, B. Diamond, B. Dreher, 1. Ellenberger, V. Graham, D. Gustine, P. Hayden, G. Hust, C. Krumm, A Larsen, K. Larsen, S. Larsen, D. Masden, D. Matiatos, M W. Miller, C. Oliver. J. Olterman, M. Potter, E. Scott, L. Spicer, D. Steele, C. Wagner,:a. Watkins, M. Wild, M. Zeeman, S. Znamenacek. ABSTRACT Long-term data indicate that fawn:doe ratios have shown a significant and consistent decline over the past 20 years in most deer management units in Colorado as well as in most of the western states, This has resulted in a general decline in mule deer (Odocoileus hemionus) density and diminished recreational potential from this major resource. It is important for the Division of Wildlife to elucidate the factors contributing to the reduced doe:fawn radios and the declining deer populations. Reproduction and neonatal survival are 2 important components of population recruitment. Therefore, 2 studies were initiated during this segment to investigate pregnancy and fetal rates of adult does and the survival of neonatal fawns. Both of these studies were done on the Uncompahgre Plateau and, in addition, a supplemental neonate fawn survival study was done in Middle park. Thirty-seven of 40 (93%) adult does on the Uncompahgre Plateau were detected pregnant with transrectal ultrasound. The proportion of does detected pregnant with ultrasound did not differ (P = 0.817) from the proportion of does pregnant (91.4% of303 does) in previous Colorado studies. The average number of fetuses/doe, 1.70, (n=40) for adult does on the Uncompahgre Plateau was not less (Z = 0.340, P = 0.367) than the fetal rate of 1.74 (n=276) in previous Colorado studies. We therefore conclude that does are getting bred and producing fetuses at normal rates for the species and that this component of , the reproductive cycle is most likely not responsible for declining December fawn:doe ratios and, 'consequently, declining deer populations. The field operation was done during February 5-8, 1999; a draft of the manuscript to be submitted to a peer reviewed publication is in Appendix I. The objective of the fawn survival investigation was to put radios on 60 neonatal fawns on the Uncompahgre and 20 neonatal fawns in Middle Park. By June 30, 1999, we had put out 50 and 14 radios on fawns on the �136 Uncompahgre and in Middle Park, respectively for a total of 64 radioed fawns. First fawns were captured on June 9 in both areas and fawn capture peaked during June 13-24 with 76% of the total captured during this time; 19 fawns (30%) were captured during June 22-24. The first mortality was recorded on June 13 with the peak of mortality (50% of total) during June 28-July 12. Causes of mortality when divided between predation, sickness/starvation, and unknown was 47%, 47%, and 6%, respectively. As of this date (August 18, 1999) 30 of64 fawns have died for a mortality rate of 46.9%, Fawns were monitored every day (except weekends) from capture with the exception of the period of July 5-27 when they were monitored twice a day. After July 27 they were again monitored daily except on weekends. Eleven of the 14 fawns that died of causes other that predation (or unknown causes) have been necropsied thus far; all had in common severe thymic atrophy indicating a compromised immune system and chronic stress. It is difficult to discern if the stressor(s) predisposed them to the variety of other pathogens that appeared to be factors in their deaths. Some of the disease agents identified were: Cryptosporidia, Pasteurella, E. Coli, BVD, and a hemorrhagic disease (Blue Tongue or EHD). It should be noted that mortality attributed to predation is probably liberal because the fawn's sickness (including diarrhea in many cases) and weakened condition probably predisposed them to olfactory or visual location by a predator. Positively identifying scavenging vs. predation was difficult even when they were monitored once or twice a day because dense vegetation precluded tracking for evidence of pursuit and kill sites as can be done with snow cover. �137 MULE DEER PREGNANCY AND FETAL RATES William F. Andelt and Thomas M. Pojar P. N. OBJECTNES 1. Estimate pregnancy and fetal rates of adult mule deer does on the Uncompahgre Plateau during the 1998-1999 reproductive cycle. 2. Publish results in a peer-reviewed scientific journal (Appendix I). SEGMENT OBJECTNES 1. Estimate accuracy of an ultrasound system for detecting' pregnancy and number of fetuses in mule deer does maintained at the Colorado Division of Wildlife's Foothills Research Facility in Northwest Fort Collins and in mule deer does collected on the USFWS Rocky Mountain Arsenal. 2. Estimate pregnancy and fetal rates of adult mule deer in DAU D-19 (Uncompahgre). 3. Compare pregnancy and fetal rates of adult mule deer to historic rates in Colorado. 4. Analyze data and prepare an annual Federal Aid Job Progress report. RESULTS See Appendix I for the Research Program Narrative and a draft of the manuscript to be submitted to a peer reviewed publication. �138 �139 lMPACT OF PREDATION AND VEGETATIVE COVER ON MULE DEER FAWN SURVIVAL Thomas M. Pojar and William F. Andelt P. N. OBJECTIVES 1. Identify agents of neonatal mule deer fawn mortality from birth to 6 months of age. SEGMENT OBJECTIVES 1. Capture and radio collar 30-35 neonatal fawns each on the Uncompahgre Plateau (D-19), Middle Park (D-9), and Red Feathers (D-4). 2. Measure vegetative density and height at fawn bed sites. 3. Measure distance between successive bed sites for individual fawns. 4. Compare fawn survival and causes of mortality among 3 vegetatively and ecologically different study areas. 5. Correlate height and density of vegetation at fawn birth sites and bed sites with percent of fawns killed by predators. 6. Compare height and density of vegetation at fawn bed sites and random sites. INTRODUCTION The Research Program Narrative is in Appendix II. Because of budget redirection to accomplish the estimation of pregnancy and fetal rates via ultrasound the objectives outlined in the PN were modified to fit within budget, personnel, and logistic constraints. Major changes include: 1. Limit the fawn collaring effort to the Uncompahgre and Middle Park rather than attempt to do it on 3 study areas. Because of the size and diversity of vegetative communities on the Uncompahgre and because of intense political interest in this area, the target sampling intensity was increased to 60 fawns. The target sample in Middle park was reduced to 20 fawns due to the redirection of resources. 2. Abandon the idea of measuring vegetation at bed sites because of evidence that persistent disturbance of bed sites reduces survival of young ungulates (Philips 1998). 3. The logistics of placing an adequate crew in remote fawning areas and the unknown of whether or not fawns could be captured in sufficient numbers given deer density, terrain, and vegetative cover encountered negated many of the objectives in the PN. In essence, this investigation was reduced to placing emphasis on determining if fawns can be captured under these conditions and, if so, monitor their survival and causes of mortality. Summer fawn survival and causes of mortality are 2 important factors impacting deer population performance. STUDY AREAS There are 2 study areas involved in this investigation, the Uncompahgre Plateau and Middle Park. The Uncompahgre Plateau includes Game Management Units 61 and 62. It is a high elevation plateau, oriented northwest-southeast, formed by a structural uplift that is characterized by steep canyons in both directions from the divide ridge that runs the length of the plateau. The soils are primarily shallow with areas of rich deeper soils that support aspen (Populus tremuloides) communities at intermediate to higher elevations. High elevations have extensive stands of spruce-fir �140 (Picea-Abies) and fir-aspen (Abies-Populus tremuloides). The crest of the plateau peaks at about 9,500 ft (2,898 m) and declines in both directions through oakbrush (Quercus gambellii), pinyonjuniper (Pinus edulis-Juniperus osteosperma), and large sagebrush (Artemisia tridentata) parks as the elevation decreases. A thorough description of the physiography, geography, climate, and vegetation can be found in Anderson et. aI (1992) and Kufeld (1979). Middle Park is a high mountain park with similar plant communities as the Uncompahgre Plateau at similar elevations. A description of the area can be found in Tiedeman et. aI (1987). MElHODS Radio collared does were used to locate fawning areas although no special effort was made to capture the fawns of radioed does. The strategy for capturing fawns involved observing does for behavioral and physical signs, such as udder development, of having fawned. If it was determined that these indicators suggested that the doe had fawned, the area was searched for the fawn(s). Once located, the fawn was approached from behind (out of direct eyesight) and restrained for weighing, measuring, and putting on the radio collar. ·The collars used were expandable with elastic and loops sewed with cotton thread to accommodate growth and to drop off at about 6 months; they weighed about 65 g. RESULTS A total of 64 fawns were captured and radio collared; 50 on the Uncomphagre Plateau and 14 in Middle Park. The first fawns were captured on June 9 with 76% of the total captures made during June 13-24; the peak 3-day period of capture was June 22-24 when 30% of the captures were made (Figure 1). Total fawn mortality as of this writing (August 18) is 30 with 14 from a condition that is generally described as sickness/starvation (S/S), 14 from predation, and 2 from unknown causes where only the collar was found (Table 1). The 14 that died from SIS were all collected, frozen or stored on ice, and later necropsied at the Colorado State University Veterinary Teaching Hospital in Fort Collins. Results thus far are preliminary with some of the lab tests still pending and that is the reason for categorizing these deaths as general SIS. Some disease agents have been identified but it is not known if these were the cause of the fawns deaths. A common condition in all of the fawns is that they had atrophied thymus glands indicating chronic stress and probable immunodeficiency which could have predisposed them to succumbing to the pathogens that they encountered. Some of the pathogens 25 20 (/) c: i& u, -- 15 .c 10 0 •... Cl> E ::::J Z 5 6/1-6/3 6rT-6/9 6/13-6/15 6/19-6/21 6125-6/27 6/4--616 6/10-6/12 6/16-6/18 6/22-6124 6/28-6/30 Date Figure 1. Dates of fawn capture combined Uncompahgre Plateau and Middle Park. from 2 areas in Colorado. 1999- �141 identified are: the parasite Crhyptosporidia, Pasterrella, E. Coli, Bovine Virus Diarrhea (BVD), and a hemorrhagic disease (Blue Tongue or Epizootic Hemorrhagic Disease (EHD». In addition to the 30 dead radioed fawns, we found 2 stillborns, 2 other intact fawn carcasses, and one adult doe carcass. One of the stillborns was fresh enough to necropsy and was small {1.5 kg) and also had an atrophied thymus. The lab results on the adult doe are not complete but she showed some signs of a hemorrhagic condition; the other carcases were in stages of decomposition that precluded materials collection. Determining if a fawn died from some other cause or was killed by a predator is difficult. Because the fawns were monitored on a daily or twice daily schedule, all carcasses that were found partially eaten were attributed to predation; some errors were undoubtedly made by doing this which would inflate the deaths attributed to predation. Vegetation was too thick to discern tracks of a chase and capture as is possible during winter with snow cover, so unless the predation was actually witnessed, there will always be doubt as to the whether or not predation was the cause of death. In 2 instances, only the collar was found with no other tracks or supporting evidence of predation. These are categorized as "unknown" but are likely predator kills. It is unlikely that the collars were slipped because they fit snugly around the neck. One of these collars had a triangular rip that could have been caused by a predator tooth (or barbed wire), the other was without marks or blood so it's categorization is even less certain. With nearly half of the fawns dying of SIS it is probable that some portion of the fawns eaten by a predator were in some stage of "sic knessl starvation" resulting in them being more susceptible to discovery by a predator ~ Several of the fawns that died had diarrhea which would have broken their .shield of scentlessness that protects them from predators early in life. Other fawns that were in a diseased or weaken state may have lacked normal hiding or escape behavior that would have predisposed them to either visual or olfactory location by a predator in the vicinity. The classical evidence that an animal was alive when bitten (i.e. predated) by a predator is subcutaneous hemorrhage; this, however, does not establish that the prey was in a healthy state when preyed upon. The peak of mortalities from all causes was during June 28-July 11 with 16 of30 (53%) dying during this 14-day period (Figure 2); 5 from SIS and 11 from predation. One fawn died on the 2nd day after capture and was judged to be a day old at capture. It was found to have no milk in it's abomasum and had a small amount of forage in it's rumen both of which indicate that it had not nursed and was starving. It is not possible to determine if this fawn was abandoned because of handling or if the dam was not providing milk; a doe was in the area (and acting maternal) during capture and 2 days later when the fawn carcass was recovered. If the fawn was truly a day old when captured and had not nursed by that time it is possible that there was a bonding or lactation problem on the part of the doe. 5 CI) Q) E (ij 4 t 0 ~ c: 3 3: -.8 <IS u.. 2 0 E 1 ::s z 0 .•.... N 0 .•... CD 00 .•.... fb .•.... C{i CD ~ g ch C:! <0 .•... N ~ ,..._ 0 .•... 00 .•.... fb .•... -.. ,..._ j:::: Dates ..,. . ~ ~ ,..._ g, 00 ~ It) £2 00 .•... .•.... , ~ 00 Figure 2. Neonatal fawn mortality from all causes for the Uncompahgre and Middle Park, Colorado, 1999. ,..._ .•... &b .•.... -.. 00 Plateau �142 All of the other SIS occurred on the 3rd, or later, day after capture. Six fawns died between the 3ni and 5th day after capture and the rest of them (77%) died after the 6th day of capture (Figure 3). It is interesting to note that there were 2 somewhat long time spans where there were no mortalities from any cause - 6 days, July 14-19, and 17 days, July 26- August 13. It seems that if there were constant predator pressure on neonatal fawns as a food source and that sickness were not a factor in predator location of fawns, then there would not be these long spans of no mortality on the marked fawns. However, it is possible that just by chance alone, predators did not encounter a marked fawn during these time periods. One radio was verified to have failed and It is believed that 2 other radios ceased to work because extensive ground and aerial searches failed to make signal contact. A set of twins that were radioed recently disappeared after domestic sheep were herded into the area they were inhabiting; no signal was obtained with a ground search of the area so an aerial search will be made. 12,------------- -, 10 CI.I ~ tIS u.. '0 8 6 ~ g :z: 4 2 <0 r= 1 Days From Capture Figure 3. Days from capture that fawns died from any cause. Total for fawns captured on the Uncompahgre Plateau and Middle Par1<.Colorado. 1999. CONCLUSIONS The following conclusions can be made from this investigation. 1. It is possible to capture neonatal fawns under the conditions found on the Uncompahgre Plateau and in Middle Park. Success in capturing fawns is dependent mostly on doe density which dictates the number of does that the capture crew can encounter during a search. Road network is also important because of the greater mobility of the capture crew and, therefore, a greater probability of encountering does .. And finally, experience of the capture crew is a major factor in the success of finding fawns. Dense vegetative cover and roadless areas make sighting of does and capture of fawns unproductive for reasonable sample sizes. 2. There is evidence that there are stress factors impacting the 2 mule deer populations that were sampled. Atrophied thymus glands of all the intact carcasses recovered along with a variety of pathogenic agents identified suggests that the stressor(s) resulted in a suppressed immune response in the fawns making them susceptible to encountered pathogens. It has been demonstrated that thymus gland size in mule deer is seasonally cyclic with the peak during summer (good nutrition) and the trough during winter (limited resources) (Anderson et aI. 1974). Ozoga and Verme (1978:794) have demonstrated the relation between thymus size and dietary plane in white-tailed deer (0. Virginianus) fawns in controlled tests. They also noted that the thymus glands offawns dying of disease or malnutrition within a month of birth were "extremely small" compared to fawns dying of other causes (accidents). �143 LITERATORE CITED Anderson, A E., D. C. Bowden, and D. M. Kattner. 1992. The puma on Uncompahgre Plateau, Colorado. Technical Publication No. 40, Colorado Division of Wildlife. Anderson, A E., D. E. Medin, and D. C. Bowden. 1974. Growth and morphometry of the carcass, selected bones, organs, and glands of mule deer. Wildlife Monographs, No. 39. Kufeld, R C. 1979. History and current status of the mule deer population of the east side of the Uncompahgre Plateau. Division Report No. 11, Colorado Division of Wildlife. Ozoga, J. J. and L. J. Verme. 1978. The thymus gland as a nutritional status indicator in deer. Journal of Wildlife Management, 42:791-798. Philips, G. E. 1998. Effects of human-induced disturbance during calving season on reproductive success of elk in the upper Eagle River valley. Ph. D. Dissertation, Colorado State University, Fort Collins. .Tiedeman, J. A, R E. Francis, C. Terwilliger, Jr., and L. H. Carpenter. 1987. Shrub-steppe habitat types of Middle Park, Colorado. USDA Forest Service Research Paper RM-273, Rocky Mountain Forest and Range Experiment Station, Fort Collins. Table 1. Neonatal fawns captured and radio collared on the Uncompahgre Plateau and Middle Park, Colorado, 1999. Radio frequencies in the 149-150 KHz band are Uncompahgre fawns and 173 KHz radios are Middle Park fawns. ID Date Sex Weight (kg) Hindfoot Date Probable 149.410/99 06/1611999 M 3.5 24.4 07/0111999 Predation 150.010/99 0611511999 F 0.0 24.9 150.020/99 06/0911999 F 0.0 24.7 150.030/99 06114/1999 F 0.0 27.3 06/2911999 Sick/Starve 150.040/99 06114/1999 F 0.0 27.6 07/0611999 Sick/Starve 150.050/99 06/1511999 F 0.0 27.9 150.070/99 0611711999 M 3.8 26.0 08/1811999 Predation 150.080/99 06/2211999 F 4.4 27.0 06118/1999 Sick/Starve 06/29/1999 Predation 150.090/99 06/2111999 F 5.9 27.6 150.100/99 06/2111999 M 5.2 26.4 150.110/99 06114/1999 U 2.3 0.0 150.120/99 06/1711999 F 5.0 26.0 150.130/99 0611411999 F 2.5 24.8 150.140/99 06110/1999 F 3.3 24.4 150.150/99 06/1711999 M 5.3 26.7 07/2011999 Predation 150.160/99 06/2211999 F 0.0 26.0 06/3011999 Predation 150.170/99 0611511999 F 3.7 26.0 150.180/99 06/2111999 F 6.1 29.2 07/2511999 Sick/Starve 07/0711999 Sick/Starve 150.190/99 0611511999 M 4.0 26.0 150.200/99 06/2211999 M 3.3 24.1 150.210/99 06/2411999 F 4.4 26.4 150.220/99 06/1611999 M 3.8 24.4 150.230/99 06/2211999 F 5.3 28.6 150.270/99 06/23/1999 M 5.6 26.4 �144 Weight (kg) Hindfoot Date Probable M 3.1 25.0 06118/1999 Sick/Starve M 4.6 26.0 0611611999 M 0.0 26.7 150.310/99 06/2811999 F 5.6 28.6 150.320/99 06123/1999 M 6.0 27.9 150.340/99 06/2311999 F 5.6 27.6 06/30/1999 Sick/Starve 150.350/99 0611711999 M 0.0 28.7 06/22/1999 Sick/Starve 150.360/99 06/2411999 M 6.0 26.7 150.370/99 06/2311999 M 6.0 27.3 150.380/99 0612411999 M 2.4 23.2 07/1111999 Predation 150.390/99 06/2411999 F 3.8 26.7 07/0111999 Predation 150.400/99 06/2211999 M 5.0 27.3 150.420/99 06128/1999 F 4.2 26.2 150.430/99 06124/1999 M 5.3 26.7 0711111999 Unknown 0811611999 Predation ID Date Sex 150.280al99 06/15/1999 150.280b/99 06/28/1999 150.290/99 150.440/99 06/2511999 M 6.0 29.2 150.45Q/99 06115/1999 F 3.5 25.7 150.460/99 06124/1999 M 3.1 23.5 150.480/99 06/3011999 F 4.2 24.8 08/16/1999 Sick/Starve 150.510/99 0611611999 F 3.6 24.1 07/0111999 Predation 150.520/99 06116/1999 F 3.5 24.8 07/0811999 Predation 150.530/99 06114/1999 M 3.6 25.4 150.540/99 06/29/1999 M 4.5 26.0 150.570199 06/24/1999 F 2.5 21.9 0711311999 Sick/Starve 150.600/99 06/2911999 M 6.0 28.9 150.610/99 06/18/1999 M 3.6 25.4 07/23/1999 Sick/Starve 150.620/99 06/30/1999 M 4.5 24.4 07/0511999 Predation 173.510/99 06/0911999 M 4.5 26.7 0611311999 Sick/Starve 173.540/99 0611111999 F 3.8 26.0 07/2211999 Predation 173.550/99 0612411999 F 4.7 27.3 173.560/99 06/2111999 F 4.1 26.0 173.590/99 06/2211999 M 5.5 28.6 173.600/99 06/2111999 F 4.1 26.0 173.610/99 06/2111999 M 5.1 28.6 06/25/1999 Sick/Starve 07/0811999 Predation -. 173.630/99 06/2111999 F 3.8 25.4 173.640al99 06/22/1999 M 4.5 26.7 06/2511999 Sick/Starve 173.640b/99 06/30/1999 M 4.2 25.4 08/16/1999 Predation 173.660/99 06/29/1999 M 6.5 27.9 173.670/99 06/2511999 F 6.0 29.2 173.690/99 06/19/1999 U 0.0 0.0 07/0511999 Unknown 173.700/99 06/1811999 M 4.7 27.0 'Due to the preliminary nature of this report, the causes of death have been put into only 3 categories, sick/starve, predation, and unknown. The unknown category includes 2 radios that were found that had no other evidence present. It is suspected that these were actually predator kills because the collars fit snug enough that slipping the collar was unlikely, in addition, one of the collars had a small triangular rip in it that could be consistent with a predator tooth tear. �145 APPENDIX I Research Program Narrative and Draft Manuscript MULE DEER PREGNANCY AND FETAL RATES ON THE UNCOMPAHGRE PLATEAU January 12, 1999 Principal Investigators: WILLIAM F. ANDELT Department of Fishery and Wildlife Biology Colorado State University Fort Collins, CO 80523 970-491-7093 mOMAS M. POJAR Colorado Division of Wildlife 317 W. Prospect Fort Collins, CO 80526 970-472-4308 �146 �147 PROGRAM NARRATIVE State of. ___;C::::;o::::..:l~or:..::a:::::d~o_ Project No .. W..!.!._-~15~3::....-.:!:.::R,__ _ Work Package No. ---,3~0~0~1,-_ Task No. -=5'--_ Cost Center 3430 Mammals Program Deer Conservation Mule Deer Pregnancy and Fetal Rates on the Uncompahgre Plateau NEED Fawn:doe (f:d) ratios obtained in December on the Uncompahgre Plateau (Game Management Units 61 and 62, Data Analysis Unit (DAU) D-19) have declined (t = -3.39, P = 0.004) by an average of 1.8 fawns: 100 does per year from 1982-1998. December ratios ranged from a high of79f:l00d in 1982 to a low of 32f: 1OOdin 1996 and 34f: 100d in 1997; there were 51.9f: 1OOdin 1998. Besides low December f:d ratios, over-winter (December - May) fawn survival on the Uncompahgre Plateau was 49% during the winter of 1997-1998 (Bartmann and Pojar 1998a). The low December f:d ratios and poor over-winter survival of fawns for the Uncompahgre deer herd is of concern to the public and game managers. It is unknown if the low December ratios are due to a failure to breed, reduced fetal production, resorption/abortion offetuses, or low summer and fall fawn survival (Bartmann 1998). Pregnancy and Fetal Rates Pregnancy rates are key information to determine if does are breeding. Recent data show that the pregnancy rate during January 1998 of93.0% for 29 does ~ 1 year old in the Red Feather DAU (Bartmann and Pojar 1998b) was similar to 92.0% of 163 does ~ 2 years old in the same area during 1961-1964 (Medin and Anderson 1979). A 94.8% pregnancy rate for does ~ 2 years old (n = 114) and a 94.0% pregnancy rate for does> 1 year old (n = 134) in Middle Park (DAU D-9) was reported by Gill (1971) during 1969-197l. Other pregnancy rates and locations reported in the literature are as follows: 100% for does ~ 2 years old (n = 18) on the Forbes-Trinchera Ranch, Colorado (Freddy 1988), 89% (n = 47) during 1973 and 82% (n = 83) during 1978 in the Piceance Basin, Colorado (Bartmann 1998), and 94% of mule deer does> 1 year old in a California study (Salwasser et al. 1978). The 1998 data collected in Colorado appears to be normal and does not indicate that pregnancy rate is a contributing factor to the low f:d ratios, however these data were not collected in the Uncompahgre DAU where December f:d ratios were lowest. Fetal rates, the number of fetuses per pregnant doe, that are in a normal range for the species would indicate that nutritional and other needs of breeding does are being met. The fetal rate for does > 2 years old was 1.83 (n = 41) in the Cache la Poudre River drainage from 1961-1965 (Medin and Anderson 1979), 1.82 (n = 114) in Middle Park (Gill 1971), 1.89 (n = 18) on the Forbes-Trinchera Ranch (Freddy 1988), and l.46 (n = 61) and 1.65 (n = 37) in the Piceance Basin (G. C. White, Colorado State University, personal communication). Salwasser et al. (1978) reported a fetal rate of l.62 for 2 and 3 year old mule deer does (n = 47) and a rate of l.85 for does> 4 years old (n = 34) in California To help ascertain the cause of low fawn:doe ratios on the Uncompahgre Plateau of Colorado, an investigation to estimate: a) the proportion of ~ 2 year old does that become pregnant and b) the number of fetuses produced is needed. Estimates of pregnancy and fetal rates will indicate if these critical reproductive parameters are within the normal range for the species and will contribute to the understanding of the causes of the observed low f.d ratios and low recruitment into the Uncompahgre mule deer population. �148 OBJECTIVES We propose to estimate pregnancy and fetal rates of adult mule deer does on the Uncompahgre Plateau during the 1998-1999 reproductive cycle. HYPOTHESES H"l: H,,2: Pregnancy rates of adult mule deer does on the Uncompahgre Plateau during the 1998-1999 reproductive cycle do not differ from historic pregnancy rates (93%) in Colorado. Fetal rates of adult mule deer does on the Uncompahgre Plateau during the 1998-1999 reproductive cycle do not differ from historic fetal rates (1.74 fawns per adult doe) in Colorado. EXPECTED RESULTS OR BENEFITS This research will help ascertain if pregnancy and fetal rates are outside the normal range expected for mule deer in Colorado and if they are contributing factors to the observed low December fawn:doe ratios. Knowledge of fetal rates is essential for determining what factors might be causing the low fawn:doe ratios observed in Colorado. Recruitment into a mule deer population is dependent on adequate pregnancy and fetal rates, and young survival through breeding age. This investigation will provide information on the early stages of the recruitment process, i.e. whether or not does are becoming pregnant and producing normal litters. These data are imperative before research and management experiments are conducted to estimate the effect of various experimental manipulations on fawn:doe ratios. APPROACH Capture and Handling of Does Forty mature does C:::. 2 years old) will be captured using a helicopter and netgun on the Uncompahgre Plateau during February 1999. The deer usually will be pursued for < 2 minutes. We expect < 2% of the captured does to sustain injuries or mortalities (R. M. Bartmann, Colorado Division of Wildlife and A. W. Alldredge, Colorado State University, personal communications). Captured does will be hobbled, blindfolded, and transported via helicopterto a base station, usually < 5 miles from the capture site to.minimize stress, where they will be processed. Does captured < 2 miles from the processing station will be released at the station, whereas does captured> 2 miles from the station will be released at the capture site. We anticipate that does will not be held for more than 30 minutes. Pregnancy and Fetal Detection The does will be manually restrained without the use of a tranquilizer to allow quick release at the processing site, unless we find sedation is necessary. If necessary, does will be sedated with ketamine and xylazine and reversed with yohimbine. The does will be placed in lateral or sternal recumbency, depending on location of the fetuses. We will use trans rectal ultrasonography, pioneered by Lindahl (1971) and applied in field situations by Barrett (1981), Smith and Lindzey (1982). White et al. (1989), and Stephenson et al. (1995), to estimate pregnancy and fetal rate. A portable ultrasound unit available from the Colorado State University Veterinary Teaching Hospital will be used; the �149 equipment will be transported in a mobile processing station to within 5 miles of capture sites to minimize transport time and stress on the does. Blood will be collected by carotid venipuncture and serum pregnancy-specific protein-B (PSPB) (Sasser et al, 1986) will be used to provide a second estimate of pregnancy rates (Stephenson et al. 1995). The use of ultrasonography to detect pregnancy is a well tested method and provides accurate estimates of pregnancy rates in elk (Cervus elephus) (Bingham et al. 1990, Revol and Wilson 1991, Willard et al, 1994). Transrectal ultrasonography was found to accurately detect pregnancy in domestic sheep (Schrick and Inskeep 1993) and fallow deer (Dama dama; Lenz et al. 1993). Because the method is not as well tested to detect the number of fetuses carried by pregnant female mule deer, we will have controls in our study to document the accuracy of the method and make adjustment in the observed fetal rate if necessary. Ten potentially pregnant does that are held in captivity in the Colorado Division of Wildlife Big Game Research Facility on the Foothills Campus, Fort Collins will be untrasounded for pregnancy and fetal rates as a control on the methodology and to test the proposed handling and restraint methods to be used on does during the field operation. These does will be under close surveillance throughout pregnancy and during fawning to verify the ultrasound results. In addition, we will have access to ~ 10 deer to be collected in January or February on the Rocky Mountain Arsenal. These does will be ultrasounded and then necropsied for immediate verification of the effectiveness of detecting the number of fetuses. Ninety-five percent of Colorado mule deer conception dates, based on back-dating from fetal forehead-rump measurements, over a 3 year span (n=135) were reported to be between November 18 and December 12 (Gill 1972). Ultrasounding for fetal rate is most successful on small ungulates during 30-60 days into pregnancy (LaRue Johnson, CSU Veterinary Teaching Hospital, personal communication). We will target the time period ofJanuary 20 through February 10 to do the. ultrasounding. Dr. LaRue Johnson will perform the ultrasounding procedures. Animal Care and Use All capture and handling of animals will comply with the standards of the Colorado State University and Colorado Division of Wildlife Animal Care and Use committees. We will secure trespass permission from study site owners prior to the study. Sample Sizes and Power of Tests The following power analysis (Table 1) was developed to estimate sample sizes necessary to detect various mean departures of fetal rates from historic levels for adult does. For the historic data, we used the weighted mean of fetal rates for adult does from the Poudre River drainage (n = 41, = 1.83, SD = 0.44; Medin and Anderson 1979), Middle park (n = 119, x = 1.86, SD = 0.57; Gill 1971), the Forbes-Trinchera Ranch (n = 18, x = 1.89, SD = 0.47; Freddy 1988), and the Piceance Basin ([1973: n = 61, x= 1.46, SD = 0.77][1978: n = 37, x = 1.65, SD = 0.72]; G. C. White, Colorado State University, personal communication) to estimate the historic mean (1.74) and standard deviation (0.64). We used an expected 1999 standard deviation equal to the historic standard deviation (0.64; G. C. White, personal communication). An alpha of 0.05 was used in our estimates of power. Power of 0.8 or higher is desired. x �150 Table 1. One-tailed power to estimate if 1999 fetal rates differ from historic rate of 1.74 fawns per mature doe (;:: 2 years old). Sample size (n) Fetal rate Power Sample size (n) Fetal rate Power 20 20 20 20 30 30 30 30 40 40 40 40 1.3 1.4 1.5 1.6 1.3 1.4 1.5 1.6 1.3 1.4 1.5 1.6 0.91 0.74 0.49 0.24 0.97 0.87 0.62 0.31 0.99 0.93 0.72 0.36 50 50 50 50 60 60 60 60 80 80 80 80 1.3 1.4 1.5 1.6 1.3 1.4 1.5 1.6 1.3 1.4 1.5 1.6 1.00 0.97 0.79 0.41 1.00 0.98 0.84 0.46 1.00 0.99 0.91 0.53 Regarding sample sizes, the desired differential in fetal rate that we wish to detect is somewhat problematic. We are primarily interested in ascertaining if pregnancy rates and fetal rates are a major contributor to the observed low December fawn:doe ratio (0.36 fawns/doe predicted by the regression equation in 1998) on the Uncompahgre Plateau. Considering historic fetal rates of L 74 fetuses/doe, there apparently is a major failure to breed, inadequate fetal production, a major loss of fetuses, high neonatal fawn mortality, or a combination of these factors which result in only 0.36 fawns/doe (predicted by regression) remaining in December 1998. On average, fawn:doe ratios in Colorado have declined by about 0.015 fawns/doe/year during the last 20 years which results in a decline of about 0.3 fawns/doe over the 20 years. Fawn:doe ratios on the Uncompahgre Plateau have declined by 0.018 fawns/doe/year during the last 17 years (1982-1998), resulting in a decline of 0.30 fawns/doe over the 17 years. Thus, to determine if fetal rates contribute to the observed changes, we will attempt to detect a change of 0.30 fawns/doe. A change of 0.30 fawns/doe represents 21 % of the 1.4 (1.74 - 0.36 [predicted by the regression equation for 1998] = 1.4) reduction in the fawn:doe ratio from fetal counts to December age-ratio counts. Thus, we believe that detecting 21% (±) of the overall loss of fetuses/fawns from the fetal stage of reproduction to December will provide strong evidence that fetal production is or is not a major contributing factor to low fawn:doe ratios. To detect a change of 0.30 fawns/doe, about 40 does will be necessary for our evaluation. Data Analyses Fetal rates obtained during 1999 on the Uncompahgre Plateau will be compared to historic rates with a Z test. Significance level to reject the null hypothesis will be at P < 0.05. LOCATION This study will be conducted on the Uncompahgre 62 (D-19). Plateau, Big Game Management Units 61 and �151 WORK SCHEDULE Jan-Feb 1999 Mar-Jun 1999 Capture and ultrasound does Data analysis and write report PERSONNEL Thomas M. Pojar F. Andelt LaRue W. Johnson David C. Bowden Kenneth P. Burnham Co-Principal Investigator, Colorado Division of WildlifeWilliam Co-Principal Investigator, Department of Fishery and Wildlife Biology, Colorado State University Co-Principal Investigator, College of Veterinary Medicine and Biomedical Sciences, Colorado State University Statistical Consultant, Department of Statistics, Colorado State University Colorado Cooperative Fish and Wildlife Research Unit, Department of Fishery and Wildlife Biology, Colorado State University ESTIMATED COST Item Helicopter capture Veterinary consulting Ultrasound rental Travel expenses Vehicle Mileage TOTAL 10% contingency GRAND TOTAL Number 40 does 1 1 15 days 1 month 2,000 miles CostlItem $ 400 $1,000 $ 600 $ 95 $ 400 $ 0.15 Total $16,000 $ 1,000 $ 600 $ 1,425 $ 400 $ 300 $19,725 1,972 $21,697 s LITERA TURE CITED Barrett, R H. 1981. Pregnancy diagnosis with doppler ultrasonic fetal pulse detectors. Wildlife Society Bulletin 9:60-62. Bartmann, R M. 1998. A prospectus on mule deer in Colorado. Colorado Division of Wildlife, Fort Collins, Colorado, Unpublished report. Bartmann, R M., and T. M. Pojar 1998a Experimental deer inventory. Colorado Division of Wildlife, FederalAid in Wildlife Restoration Job Progress Report, Project W-153-R-ll, July. _, and _. 1998b. Deer reproduction assessment. Colorado Division of Wildlife, Federal Aid in Wildlife Restoration Job Progress Report, Project W-153-R-11, July. Bingham, C. M., P. R. Wilson, and A S. Davies. 1990. Real-time ultrasonography for pregnancy diagnosis and estimation of fetal age in farmed red deer. Veterinary Record 126:102-106. Freddy, D. 1. 1988. Effect of elk harvest systems on elk breeding biology. Colorado Division of Wildlife, Federal Aid in Wildlife Restoration Job Progress Report, Project W-153-R-2, July. Gill, R B. 1971. Middle Park deer study - productivity and mortality. Colorado Division of Wildlife, Federal Aid in Wildlife Restoration Job Progress Report, Project W-38-R-25. July:189-207. _. 1972. Productivity studies of mule deer in Middle Park, Colorado: Second Annual Mule Deer Workshop, Elko, Nevada, January 12. Xerox. Lenz, M. F., A W. English, and A Dradjat. 1993. Real-time ultrasonography for pregnancy diagnosis and foetal ageing in fallow deer. Australian Veterinary Journal 70:373-375. �152 Lindahl, I. L. 1971. Pregnancy diagnosis in the ewe by intrarectal doppler. Journal of Animal Science 32:922-925. Medin, D. E., and A. E. Anderson. 1979. Modeling the dynamics of a Colorado mule deer population. Wildlife Monographs 68:1-77. Revel, B., and P. R Wilson. 1991. Ultrasonography of the reproductive tract and early pregnancy in red .deer. Veterinary Record 128:229-233. Salwasser, H., S. A. Holl, G. A. Ashcraft. 1978. Fawn production and survival in the North Kings River deer herd. California Fish and Game 64:38-52. Sasser, R G., C. A. Ruder, and K. A. Ivani. 1986. Pregnancy detection in farm animals by radioimmunoassay of a pregnancy-specific protein in serum. in 1. Hau, ed., Pregnancy proteins in animals. Walter de Gruyer & Co. Berlin. Schrick, F. N., and E. K. Inskeep. 1993. Determination of early pregnancy in ewes utilizing transrectal ultrasonography. Theriogenology 40:295-306. Smith, R B., and F. G. Lindzey. 1982. Use of ultrasound for detecting pregnancy in mule deer. Journal of Wildlife Management 46:1089-1092. Stephenson, T. R, 1. W. Testa. G. P. Adams, R G. Sasser, C. C. Schwartz, and K. 1. Hundertmark. 1995. Diagnosis of pregnancy and twinning in moose by ultrasonography and serum assay. Alces 31:167-172. Welker, H. 1. 1986. Fawn mortality in the Lake Hollow deer herd, Tehama County, California California Fish and Game 72:99-102. White, I. R, W. A. C. KcKelvy, S. Busby, A. Sneddon, and W. 1. Hamilton. 1989. Diagnosis of pregnancy and prediction of fetal age in red deer by real-time untrasonic scanning. The Veterinary Record 124:395-397. Willard, S. T., R G. Sasser, J. C. Gillespie, J. T. Jaques, T. H. Welsh, Jr., and RD. Randel. 1994. Methods for pregnancy determination and the effects of body condition on pregnancy status in Rocky Mountain elk (Cervus elephus nelsonii. Theriogenology 42:1095-1102. �153 DRAFT MANUSCRIPT ESTIMATION OF MULE DEER PREGNANCY AND FETAL RATES ON THE UNCOMPAHGRE PLATEAU, COLORADO USING TRANSRECTAL ULTRASOUND William F. Andelt, Thomas M. Pojar, and LaRue W. Johnson Abstract Mule deer (Odocoileus hemionus) populations apparently have declined in several western states during the 1990's. In Colorado, mule deer fawn:doe ratios in December have declined by 0.15 fawns/doe/year from 1972 through 1995. Significant declines in the fawn:doe ratio have especially occurred on the Uncompahgre Plateau near Montrose, Colorado. Thus, we estimated adult mule deer pregnancy and fetal rates on the Uncompahgre Plateau during the 1998-1999 reproductive cycle to determine if lower pregnancy or fetal production was the cause of the deer decline. Thirty-seven of 40 (93%) adult does on the Uncompahgre Plateau were detected pregnant with ultrasound, and 36 of the 40 does were detected pregnant with pregnancy-specific protein-B (PSPB). The proportion of does detected pregnant with ultrasound does not differ (Xl) = 0.053, P = 0.817) from the proportion of does pregnant (91.4% of303 does) in previous studies in Colorado. The average number of fetuses/doe for all adult does (pregnant and non-pregnant) on the Uncompahgre Plateau (n = 40, '! = 1.70, SE = 0.109) was not less (Z = 0.340, P = 0.367) than in previous studies (n = 276, '! = 1.74, SE = 0.038). The average number of fetuses/doe for pregnant adult does on the Uncompahgre Plateau (n = 37, '!= 1.84, SE = 0.082) also was not less (Z = 0.260, P = 0.397) than in previous studies (n=258, '! = 1.86, SE = 0.028). INTRODUCTION Mule deer (Odocoileus hemionus) populations apparently have declined in several western states of the United States during the 1990's (Unsworth et al. 1999). In Colorado, mule deer fawn:doe ratios in December have declined by 0.15 fawns/doe/year from 1972 through 1995 (G. C. White, Colorado State University, unpublished data). In particular, fawn:doe ratios on the Uncompahgre Plateau of Colorado have declined (I = -3.39, P = 0.004) by an average of 0.18 fawns/doe/year from 1982-1998, and have ranged from a high of 79 fawns/100 does in 1982 to a low of 32 fawns/l00 does in 1996 and 34 fawns/100 does in 1997 (Colorado Division of Wildlife, unpublished data). In addition to low December fawn:doe ratios, over-winter (December - May) fawn survival on the Uncompahgre Plateau was 49% during the winter of 1997-1998 (Bartmann and Pojar 1998a). The low December fawn:doe ratios are a concern to the public and game managers. Pregnancy rates of collected adult (2:2years old) female mule deer have ranged from 97 to 100% during the 1960's, 1970's, and 1980's in Colorado (Gill 1971, 1972; Medin and Anderson 1979; Freddy 1987, 1988). The average numbers offetuses per adult doe (fetal rate) have ranged from an average of 1.83 to 1.91 (Medin and Anderson 1979; Gill 1971, 1972; Freddy 1987, 1988) during the same period in various areas of Colorado. Several hypotheses exist to explain the decline in December mule deer fawn:doe ratios including a failure to breed, reduced fetal production, resorption/abortion of fetuses, or low summer and fall fawn survival (R. M. Bartmann, Colorado Division of Wildlife, personal communication) mediated by low �154 buck doe ratios and a protracted fawning period, poor habitat quality and reduced nutrition or hiding cover, more predators and increased predation, competition with elk (Cervus elaphus), disease, poisonous plants; drought, or perhaps a combination of these factors. Our objectives in this study were to determine adult mule deer pregnancy and fetal rates on the Uncompahgre Plateau, Colorado during the 1998-1999 reproductive cycle. We hypothesized that pregnancy and fetal rates did not differ from rates obtained during the 1960's, 1970's, and 1980's in Colorado. Transrectal ultrasonography has been used to accurately detect pregnancy in domestic sheep (Schrick and Inskeep 1993), fallow deer (Dama dama; Lenz et aI. 1993), red deer (Cervus elaphus; Bingham et aI. 1990, Wilson and Bingham 1990, Revol and Wilson 1991), and elk (Willard et aI. 1994). Pregnancy-specific protein-B has been used to reliably detect pregnancy in mule deer and white-tailed deer (Odocoi/eus virginianus; Wood et aI. 1986), muskoxen (Ovibos moschatus; Rowell et aI. 1989), fallow deer (Wilker et aI. 1993), moose (Alces alces; Haigh et aI. 1993, Stephenson et aI. 1995), and elk (Willard et aI. 1994, Noyes et aI. 1997). Because neither technique has been used extensively in mule deer, with the exception of Smith and Lindzey (1982) and Wood et aI. (1986), we also verified the accuracy of both techniques. STIJDY AREAS To assess accuracy in determining pregnancy and number of fetuses per mule deer doe, we investigated captive does at the Colorado Division of Wildlife's Foothills Research Facility in Northwest Fort Collins,and wild does that were culled on the USFWS Rocky Mountain Arsenal, Commerce City, Colorado. In our main study, we ultrasounded does on the Uncompahgre Plateau (Game Management Units 61 and 62, Data Analysis Unit (DAU) D-19) in southwestern Colorado. The plateau is about 100 km long and 29-40 km wide, and is bounded by the San Miguel and Dolores rivers on the southwest and west, the Uncompahgre and Gunnison rivers on the east, the Colorado river on the North, and an escarpment of Dakota sandstone above Pleasant Valley and Dallas creeks on the southeast (Anderson et aI. 1992). Topography ranges from gently rolling to steep canyons. Deer were captured on winter range consisting of Pinyon pine (Pinus edulis)-Utahjuniper (Juniperus osteospenna)-sagebrush interspersed with open sagebrush parks. The climate is semi-arid, with warm summers and cold winter. METHODS We evaluated 10__ captive and 15 wild mule deer does to ascertain the accuracy of ultrasound and PSPB for detecting pregnancy and fetal rates. We restrained 6 (5 or 6 years old) of the 10 captive does in a capture pen and immobilized them with an intramuscular injection of 420-500 mg ofketamine and 80-100 mg ofxylazine (a 5:1 mixture) on 25 January 1999. We also immobilized 4 yearling (about 1.6 years old) captive does with a dart gun using 250 mg oftelazol and 125 mg ofxylazine. After processing, tranquilization was reversed with 10 mg of yohimbine. We evaluated the 15 wild does (2 yearling and 13 adults), which were culled for herd management purposes by the U.S. Fish and Wildlife Service, on 31 January 1999. We captured 40 adult mule deer does on the Uncompahgre Plateau with a netgun fired from a Hughes 500C helicopter (Helicopter Capture Services, Marysvale, Utah, USA; Barrett et aI. 1982). The does were captured on deer winter range in proportion to their density within 8 strata located around the edge of the plateau. The deer were manually restrained, hobbled, blindfolded, and transported by helicopter to a base station, usually <6 km from the capture site, where they were processed on 5-8 February 1999. We aged each doe as ajuvenile, yearling, or adult by tooth replacement and wear, weighed each doe, and injected the first 32 does subcutaneously �155 with 5 ml of vitamins A (500,000 iu's), D3 (50,000 iu's), and E (1,500,000 iu's). We attached a 148 MHz telemetry collar, and, after processing, released all except 1 doe at the processing sites; the 1 doe was transported by helicopter to the capture site and released there because a steep canyon separated the capture and processing sites. In addition, 1 juvenile female, 2 yearling females, and 2 males were captured. The juvenile was immediately released at the processing site, the 2 yearlings were processed similar to the adult does and released at the processing site, and the 2 males were immediately released at the capture site. We processed and released does within about 30 minutes of capture. We placed the 25 does used for accuracy evaluation and the 40 study does in dorsal recumbency and used a portable ultrasound unit (Aloka 500V, Aloka, Wallingford, Connecticut, USA) with a 5 MHZ linear-array probe, powered by a portable generator in the field, to ascertain pregnancy and number of fetuses/doe via trans rectal ultrasonography (Smith and Lindzey 1982, Bingham et al. 1990, Stephenson et al. 1995). We also collected about 10 ml of blood viajugular venipuncture to assess levels ofPSPB and ascertain pregnancy status. The blood was allowed to clot, centrifuged, and, the serum was decanted and stored on ice and then frozen. The serum was submitted to BioTracking (Moscow, Idaho, USA) to assess levels ofPSPB (Stephenson et al. 1995, Noyes et al. 1997). We ascertained accuracy of the ultrasound and PSPB procedures by performing necropsies and counting the number of fetuses in 1 captive doe that died from chronic wasting disease and 3 other does that were euthanized because of chronic wasting disease (M. A. Wild, Colorado Division of Wildlife, personal communication). We also ascertained accuracy of the ultrasound and PSPB by counting the number of fawns born during May-June 1999 for the other 6 captive does, whereas we necropsied the does collected on the Rocky Mountain Arsenal to determine the number of fetuses. One of us (L WJ) had extensive experience using ultrasound techniques and conducted all ultrasound procedures. Apriori, we conducted a power analysis to estimated the sample size necessary to detect a departure in fetal rates from previous studies of fetal rates in Colorado. For the previous studies, we used the weighted mean of fetal rates for adult does from .the Poudre River drainage (n = 41, x = 1.83, SD = 0.44; Medin and Anderson 1979), Middle Park (n = 119, x= 1.86, SD = 0.57; Gill 1971), the Forbes-Trinchera Ranch (n = 33, x = 1.91, SD = 0.46; Freddy 1987, 1988), and the Piceance Basin ([1973: n = 61, x= 1.46, SD = 0.77][1978: n = 37, x= 1.65, SD = 0.72]; G. C. White, Colorado State University, personal communication) to estimate a historic mean (1.74) and standard deviation (0.64). We used an expected 1999 standard deviation equal to the historic standard deviation (0.64; G. C. White, personal communication) in our analysis. We were primarily interested in ascertaining if pregnancy rates and fetal rates were a major contributor to the observed low December fawn:doe ratio (0.36 fawns/doe predicted by the regression equation for 1998 using data on fawn:doe ratios from 1982-1998) on the Uncompahgre Plateau. Considering historic fetal rates of 1.74 fetuses/doe, and a predicted fawn:doe ratio of 0.36 remaining on the Uncompahgre Plateau in December 1998, there is considerable loss of fetuses or fawns. On average, fawn:doe ratios in Colorado have declined by about 0.015 fawns/doe/year from 1972 to 1995 which results in a decline of about 0.3 fawns/doe over the 23 years. Fawn:doe ratios on the Uncompahgre Plateau have declined by 0.018 fawns/doe/year during the last 17 years (1982-1998), resulting in a decline of 0.30 fawns/doe over the 17 years. Thus, we attempted to determine if fetal rates contributed to the observed change of 0.30 fawns/doe which represents only 21% of the observed reduction in the fawn:doe ratio of 1.4 (1.74 - 0.36 [predicted by the regression equation for 1998] = 1.4). Thus, our sample size of 40 does was aprior estimated to have power of 80% for detecting a change of 0.3 fawns/doe at an alpha of 0.05. The proportion of adult does pregnant during 1999 was compared to historic data with a chisquare test (proc Freq, SAS Institute 1988). Fetal rates obtained during 1999 were compared to historic rates with a Z-test. An alpha level 0[0.05 was considered significant. �156 RESULTS Via necropsy or observing newborn fawns, we ascertained that 1 of the 10 captive does did not have a fetus, 3 does produced 1 fetus or fawn, and 6 does produced twin fetuses or fawns. Via ultrasound, we estimated that 3 captive does had single fetuses, 6 does had twins, and 1 doe had triplets. Thus, for each of 3 does, we estimated 1 more fetus via ultrasound compared to necropsies or observation of newborn fawns. All does were detected pregnant via PSPB levels. We necropsied the 15 culled mule deer does (2 yearlings and 13 adults) and found 5 does (2 yearlings and 3 adults) were barren, 3 does each contained 1 fetus, 6 does contained twins, and 1 doe contained triplets. Using ultrasound, we accurately detected all 15 does as pregnant or not pregnant. We also detected the correct number of fetuses per doe for 12 of the 15 does. We incorrectly detected 3 fetuses via ultrasound in 1 doe but she had only 2 fetuses, and we detected only 1 of 2 fetuses in 2 other does. PSPB levels indicated that 12 of the 15 does were pregnant; two of the 12 does were not detected-pregnant via both ultrasound and necropsy .. Thirty-seven of the 40 (93%) adult does on the Uncompahgre Plateau were detected pregnant with ultrasound. This proportion does not differ (XZJ = 0.053, P = 0.817) from the proportion of does pregnant (91.4% of303 does) in previous studies in Colorado (Gill 1971, 1972; Medin and Anderson 1979; Freddy 1987, 1988; R M. Bartmann, Colorado Division of Wildlife, unpublished data). The average number of fetuses/doe for all adult does (pregnant and non-pregnant) on the Uncompahgre Plateau (n = 40, ~= 1.70, SE = 0.109) was not less (Z= 0.340, P = 0.367) than in previous studies (n = 276, ~= 1.74, SE = 0.038; Gill 1971, 1972; Medin and Anderson 1979; Freddy 1987, 1988; R M. Bartmann, Colorado Division of Wildlife, unpublished data). The average number of fetuses/doe for pregnant adult does on the Uncompahgre Plateau (n = 37, ~ = 1.84, SE = 0.082) also was not less (Z = 0.260, P = 0.397) than in previous studies (n = 258, ~ = 1.86, SE = 0.028). PSPB levels concurred with ultrasound in ascertaining pregnancy in 39 of the 40 does; PSPB levels indicated 1 doe was not pregnant whereas 2 fetuses were visually detected via ultrasound. DISCUSSION We failed to detect a difference in pregnancy and fetal rates of adult female mule deer between our and previous studies in Colorado. Bartmann and Pojar (1998b) also found a relative high pregnancy rate of93.0% for 29 does ~1 year old near Red Feather, Colorado during January 1998. Thus, a failure to breed or maintain pregnancy thru at least early February can be discounted as the cause of the low fawn:doe ratios observed on the Uncompahgre Plateau. We restricted our. analyses of pregnancy and fetal rates to adult (~2 years old) mule deer does. We did not include yearlings in our sample because a lower proportion of yearlings become pregnant and they have lower fetal rates which would have biased our comparisons to historic rates unless the proportions of yearlings in both samples were equal. �157 LITERATIJRECITED Anderson, A. E., D. C. Bowden, and D. M. Kattner. 1992. The puma on Uncompahgre Plateau, Colorado. Colorado Division of Wildlife Technical Publication No. 40. Fort Collins, Colorado, USA. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108-114. Bartmann, R M. 1998. A prospectus on mule deer in Colorado. Colorado Division of Wildlife, Fort Collins, Colorado, USA. Unpublished report. _, and T. M. Pojar 1998a. Experimental deer inventory. Colorado Division of Wildlife, Federal Aid in Wildlife Restoration, Project W-153-R-11, Progress Report . ._, and _. 1998b. Deer reproduction assessment. Colorado Division of Wildlife, Federal Aid in Wildlife Restoration, Project W-153-R-11, Progress Report. Bingham, C. M., P. R Wilson, and A. S. Davies. 1990. Real-time ultrasonography for pregnancy diagnosis and estimation of fetal age in farmed red deer. The Veterinary Record 126:102-106. Freddy, D. J. 1987. Effect of elk harvest systems on elk breeding biology. Colorado Division of Wildlife, Federal Aid in Wildlife Restoration, Project 01-03-047, Progress Report. Freddy, D. J. 1988. Effect of elk harvest systems on elk breeding biology. Colorado Division of Wildlife, Federal Aid in Wildlife Restoration, Project W-153-R-2, Progress Report. Gill, R B. 1971. Middle Park deer study - productivity and mortality. Colorado Division of Wildlife, Federal Aid in Wildlife Restoration, Project W-38-R-25, Progress Report. _. 1972. Productivity studies of mule deer in Middle Park, Colorado. Second Annual Mule Deer Workshop, Elko, Nevada Haigh, J. C., W. J. Dalton, C. A. Ruder, and R G. Sasser. 1993. Diagnosis of pregnancy in moose using a bovine assay for pregnancy-specific protein B. Theriogenology 40:905-911. Lenz, M. F., A. W. English, and A. Dradjat. 1993. Real-time ultrasonography for pregnancy diagnosis and foetal ageing in fallow deer. Australian Veterinary Journal 70:373-375. Medin, D. E., and A. E. Anderson. 1979. Modeling the dynamics of a Colorado mule deer population. Wildlife Monographs 68: 1-77. Noyes, J. H., R G. Sasser, B. K. Johnson, L. D. Bryant, and B. Alexander. 1997. Accuracy of pregnancy detection by serum protein (PSPB) in elk. Wildlife Society Bulletin 25:695-698. Revol, B., and P. R Wilson. 1991. Ultrasonography of the reproductive tract and early pregnancy in red deer. The Veterinary Record 128:229-233. Rowell, J. E., P. F. Flood, C. A. Ruder, and R G. Sasser. 1989. Pregnancy-specific protein in the plasma of captive muskoxen. Journal of Wildlife Management 53:899-901. . Salwasser, H., S. A. Hell, G. A. Ashcraft. 1978. Fawn production and survival in the North Kings River deer herd. California Fish and Game 64:38-52. SAS Institute Inc. 1988. SAS/STAT User's guide, release 6.03 edition, SAS Institute Inc., Cary, North Carolina, USA. Schrick, F. N., and E. K. Inskeep. 1993. Determination of early pregnancy in ewes utilizing transrectal ultrasonography. Theriogenology 40:295-306. Smith, R B., and F. G. Lindzey. 1982. Use of ultrasound for detecting pregnancy in mule deer. Journal of Wildlife Management 46:1089-1092. Stephenson, T. R, J. W. Testa G. P. Adams, R. G. Sasser, C. C. Schwartz, and K. J. Hundertmark. 1995. Diagnosis of pregnancy and twinning in moose by ultrasonography and serum assay. Alces 31:167-172. Unsworth, J. W., D. F. Pac, G. C. White, R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. �158 Wilker, c., B. Ball, T. Reimers, G. Sasser, M. Brunner, B. Alexander, and M. Giaquinto. 1993. Use of pregnancy-specific protein-B and estrone sulfate for determination of pregnancy on day 49 in fallow deer (Dama dama). Willard, S. T., R G. Sasser, J. C. Gillespie, J. T. Jaques, T. H. Welsh, Jr., and RD. Randel. 1994. Methods for pregnancy determination and the effects of body condition on pregnancy status in Rocky Mountain elk (Cervus elephus nelsoni). Theriogenology 42:1095-1102. Wilson, P. R, and C. M. Bingham. 1990. Accuracy of pregnancy diagnosis and prediction of calving date in red deer using real-time ultrasound scanning. The Veterinary Record 126:133-135. Wood, A. K., R E. Short, A. Darling, G. L. Dusek, R G. Sasser, and C. A. Ruder. 1986. Serum assays for detecting pregnancy in mule and white-tailed deer. Journal of Wildlife Management 50:684--687. �159 APPENDIX II Research Program Narrative IMPACT OF PREDATION AND VEGETATIVE COVER ON NEONATAL MULE DEER FAWN SURVIVAL 21 October 1998 Principal Investigators: THOMAS M. POJAR Colorado Division of Wildlife 317 W. Prospect Fort Collins, CO 80526 970-472-4308 WILLIAM F. ANDELT Department of Fishery and Wildlife Biology Colorado State University Fort Collins, CO 80523 970-491- 7093 �160 �161 PROGRAM NARRATIVE State of --"C=o=lo=r=ad=o~ _ Project No. W!..!.--.:..:15::..::3:....-R~ _ Work Package No. _;3:::..;0:<...:0'-"1 _ Task No. ----:.4 _ Cost Center 3430 Mammals Program Deer Conservation Impact of Predation and Vegetative Cover on Mule Deer Fawn Survival NEED There is evidence that the mule deer (Odocoileus hemionus) population in Colorado has declined during recent years due mostly to low fawn survival and subsequent low population recruitment (Bartmann 1997). Colorado Division of Wildlife quadrat surveys to estimate population size suggest that some mule deer herds have declined by 31 % since 1992 while some have not shown much change during this time (Bartmann 1997). On 2 areas where recent over-winter (1997-98) survival was estimated, rates of fawn and adult doe survival was 74% (n = 38) and 100% (n = 30) for the Red Feather herd (D-4) and 49% (n = 39) and 84% (n =31) for the Uncompahgre herd (D-19), respectively (Bartmann and Pojar 1998a). This supports the contention that some of the purported population declines may be influenced by over-winter fawn survival. However, some herds have low December fawn:doe ratios indicating possible deficiencies in 2 major population growth mechanisms - initial fawn production (pregnancy rates) and neonatal fawn survival (Bartmann 1998). Pregnancy rates during January 1998 were 93% for 29 does >1 year old in the Red Feather DAU (Bartmann and Pojar 1998b) which was similar to 92.0% ofl63 does (>2 years old) pregnant in the same area during 1961-1964 (Medin and Anderson 1979), a 94.8% pregnancy rate for adult does (n = 114) and a 94.0% pregnancy rate for adult and yearling does (n = 134) in Middle Park (Gill 1971), and rates of 89% (n = 47) during 1973 and 82% (n = 83) during 1978 in the Piceance Basin (Bartmann 1998). The 1998 data indicate that the proportion of does becoming pregnant probably was not contributing to the low fawn:doe ratios, however these data were not collected in the Uncompahgre DAU where ratios were lowest. Pregnancy rates will be obtained in the Uncompahgre DAU during January 1999 to ascertain if these rates are contributing to the low fawn:doe ratios. Fetal rates for does older than yearlings were 1.83 (n = 41) in the Cache la Poudre River drainage from 1961-1965 (Medin and Anderson 1979) and 1.82 (n = 114) in Middle Park (Gill 1971). If fetal rates are currently as high as they were in the 1960's, then high rates of fawn mortality are likely responsible for the low. and declining fawn:doe ratios. Thus, determining the causes and magnitude of fawn mortality from birth through December will be extremely important for understanding mule deer population trends in Colorado. Low neonatal fawn survival is probably the most sensitive indicator that the environmental needs of populations are not being met (Gaillard et al. 1998). Fawn:doe ratios obtained in December have declined by an average of 1.5 fawns per 100 does per year from 1972-1995 (Bartmann 1997, G. C. White, Dept. Fishery and Wildlife Biology, Colorado State University, pers. commun.). December 1997 fawn:doe ratios were 34.2 in the Uncompahgre DAU and 58.8 in the Red Feather DAU (Bartmann and Pojar 1998a). The December fawn:doe ratios and over-winter survival rates indicate that the status of the Uncompahgre deer herd is of greatest concern. Gruell (1986) hypothesizes that historic (1880-1930) domestic livestock grazing and fire suppression caused successional changes in rangeland communities favoring shrubby species resulting in optimal mule deer habitat during the period 1930-1960. These conditions resulted in the irruption and overpopulation of deer herds and may have reduced present day carrying capacity which may be �162 mediated by lower nutrition andlor greater susceptibility to predation (see also Salwasser et al. 1978). Fire suppression and removal of herbaceous species by grazing can foster conditions that favor woody species invasion (Skovlin 1991) and decadent/senescent stands of shrubby species (Clements and Young 1996) thereby reducing the canying capacity for deer. Once the stable state of a particular vegetation type has been disrupted, especially in arid or semi-arid rangeland types, the vegetative community may stabilize at a lower successional state that is highly resistant to change without drastic perturbations (Laycock 1991). It is possible this situation is in process on the Uncompahgre Plateau (and other areas of the West) with the constant expansion of the pinon-juniper vegetative community into sagebrush and grassland communities. A low fawn:doe ratio on the Uncompahgre DAU could result from several causes including malnutrition and predation. Competition with elk (Cervus elaphus) for food or space on wintering grounds, or for fawning sites on summer range may cause poor neonate survival through decreased nutrition or increased susceptibility to predation. Livestock grazing can reduce forbs and fawn hiding cover (Loft et al. 1987) which can influence nutrition and predation (Smith and LeCount 1979, Bowyer and Bleich 1984), but there likely is less livestock grazing now thanwhen deer were more abundant indicating recent grazing may not be the cause of the decline. A short-term drought may have reduced forage for lactating does, may have protracted the fawning period making fawns more susceptible to predators as in white-tailed deer (Odocoi/eus virginianus, Andelt et al. 1987), or reduced alternate prey for predators which may be exacerbating the low fawn:doe ratios. Smith and LeCount (1979) reported that survival of mule deer fawns was associated with winter forb yield and October-April rainfall of the winter-spring period preceding the fawning period. Coyotes (Canis latrans) and other predators kill and eat mule deer fawns and adults, however the population impact of this predation is variable. Connolly (1978) cited 31 studies which reported predation as a limiting or regulating influence on ungulate populations in North America Connolly (1978) also cited 27 studies which indicated that predators did not limit or control the size of ungulate populations. Filonov (1980) provides evidence that the total loss to natural mortality for ungulates is relatively stable and that it is maintained by a complex system of compensating mechanisms, one of which is predation. Because of functional substitution of mortality factors, the various reasons for natural mortality are independent of the total mortality and the range of mortality rates remains somewhat constant. An example of compensatory mortality was demonstrated in the Piceance Basin of Colorado where predator control reduced overwinter mule deer mortality to predators, but the effect was compensated for by additional mortality due to starvation (Bartrnann et al. 1992). We need to determine the causes of neonatal fawn mortality in areas with low fawn:doe ratios, such as Uncompahgre Plateau, so that we know where to focus experimental research and management. Predator density is one factor that may have a significant impact on fawn survival independent of vegetational or nutritional components of the habitat (Smith et al. 1986). Coyotes, black bears (Ursus americanus), mountain lions (Felis concolor), and bobcats (Felis rufus) are found on the Uncompahgre Plateau and in other areas of Colorado, and may have a limiting effect on mule deer. Coyotes have been responsible for high neonatal white-tailed deer fawn (Cook et al. 1971, Garner et al. 1976, Litvaitis and Bartush 1980, Stout 1982), high neonatal mule deer fawn (Steigers and Flinders 1980, Steigers 1981), and high overwinter mule deer fawn (White et al. 1987) mortality. Beck (1995) reported a density of O. 93 black bears/mf on the Uncompahgre Plateau, and black bears can prey on mule deer fawns (Smith 1983), white-tailed deer fawns (Ozoga and Verme 1982, 1986; Mathews and Porter 1988), and elk calves (Schlegel 1976). Anderson et al. (1992) estimated that a minimum population of34 mountain lions on 3,120 km2 of the Uncompahgre Plateau may have killed 1,885 to 2,060 mule deer or about 8 to 12% of the estimated wintering deer population during 1987. Welker (1986) reported that mountain lions killed 2 or 3 of 16 radioed mule deer fawns; Bleich and Taylor (1998) found that mountain lion predation on adult does was the primary source of mortality in northeastem California herds. Bobcats also kill deer (Litvaitis et al. 1986) and pronghorn fawns �163 (Antilocapra americana) (Autenrieth 1982, 1984). Although these predators kill deer, their impact on mule deer fawn:doe ratios in Colorado is unknown. If predators are responsible for the low fawn:doe ratios, we need to determine which predator is the major contributor, and if some factors, such as low hiding cover, less alternate prey (Horejsi 1982, Andelt et al. 1987), a protracted fawning season, predator abundance, and possibly a high predator:deer ratio are making fawns more vulnerable to predation. Height of vegetation apparently plays an important role in the survival of neonate ungulates. Riley and Dood (1984) and Boyer et al. (1998) reported that mule deer fawns and (Decker 1991) reported that white-tailed deer fawns selected habitat types with dense vegetative cover. Horejsi (1982) reported that in years of drought, mule deer fawn survival was lower in areas oflivestock grazing and attributed this to decreased ground cover. Benzon (1996) reported that white-tailed deer fawns in the Black Hills, South Dakota, selected an understory of grasses for diurnal bed sites; height and density of vegetation was higher at bed sites than what was randomly available. Tucker and Garner (1983) reported that bed sites of pronghorn <4 weeks old were found in cover taller than that in surrounding areas. Autenrieth (1982, 1984) suggested that taller cover was important for reducing pronghorn fawn mortality to predators. Rothchild (1993) reported that vegetation was taller at the bed sites of pronghorn fawns than at random sites and the vegetation also was taller at bed sites of surviving pronghorn fawns compared to fawns that did not survive. Rothchild et al. (1984) suggested that pronghorn fawns with larger home ranges may sustain higher coyote predation in tall grass prairie of east central Kansas. Fawns have an infinite number of potential bed site locations within their home ranges. Measurement of random sites that are possible bed sites should be within a typical home range of a fawn. Home ranges of mule deer fawns in northern Colorado were variable and estimated at 130 ha from 10 June through August (Geduldig 1981). Average summer home range size offawns was 185 ha in the Missouri River Breaks, Montana and home range size decreased with increased population size (Riley and Dood 1984). Home range size of fawns increased with age in south-central Washington and averaged 257 ha for fawns 60 days or older (Steigers and Flinders 1980). Mule deer fawns traveled variable distances each day, and these distances more than doubled when fawns were 61-90 days old (640 m) compared to when 1-30 days old (319 m) (Steigers and Flinders 1980). OBJECTIVES We propose to identify the agents of mortality and quantify the extent of depredations on neonatal mule deer fawns. We also propose to correlate height, density, and composition of vegetation at fawn birth and bed sites with fawn survival. HYPOTHESES Our hypotheses are divided under two research areas, predation and vegetative cover. 1. What is the contribution of coyotes and other predators to neonatal mule deer fawn mortality? Hoi: The proportions of fawns killed by coyotes, black bears, mountain lions, and bobcats do not differ from one another. H.,2: Predation on fawns is not related to sex or activity of fawns; activity as measured by the distance between successive bed sites. Ho3: Predation on fawns to 8 weeks of age is not related to the average distance between a fawn and its mother. H04: Survival offawns and causes of mortality do not differ among 3 vegetatively and ecologically different study areas. Hos: Fawn birth weights and measurements do not differ among study areas. Roo: Distances between fawn bed sites do not differ among study areas. �164 2. Do height and density, and/or composition of vegetation at fawn birth and bed sites correlate with the percent of fawns killed by predators? H"I: Predation rates on fawns is not related to height and density (visual obstruction readings), and/or composition of vegetation at birth and bed sites. Ho2: Height and density (visual obstruction readings, VOR), and/or composition of vegetation do not differ between fawn bed sites and random sites. Ho3: Height and density (VOR), and/or composition of vegetation at fawn bed sites and random sites do not vary among study areas. EXPECTED RESULTS OR BENEFITS This research will quantify the amount and causes of neonatal fawn mortality and contribution of various predators to this mortality. The influence of vegetative characteristics at birth and bed sites on subsequent fawn survival will be estimated. Knowledge of these factors will assist in evaluating the causes of low December fawn:doe ratios and will guide future research to determine the effect of various experimental manipulations on neonatal fawn survival. For political, social, and economic reasons, we need to have knowledge about the causes of low fawn:doe ratios before corrective action is taken. APPROACH Animal Care and Use: All capture and handling of animals will comply with the standards of the Colorado Division of Wildlife and Colorado State University Animal Care and Use Committees. We will obtain a collectors permit from the Colorado Division of Wildlife and secure trespass permission from study site owners prior to the study. Fawn Capture: In a related study, 40 radioed mature does will be monitored on each study area to determine adult doe survival (Federal Aid Project W-153-R); fawns will be located primarily by monitoring these does and capturing their fawns. Our objective will be to capture and radio 25-35 fawns on each of 3 study areas. Three study areas will be selected based on : a) contrasting vegetative/ecological characteristics, b) a range in fawn:doe ratios indicating contrasting fawn survival rates, and c) availability of radio collared does to minimize fawn capture costs and ensure randomness in fawn capture. Three candidate areas which meet all 3 characteristics are: The Uncompahgre, Middle Park, and Red Feathers DAUs. Fawns <10 days old will be captured and fitted with expandable, drop-off radio collars. Fawns younger than 24 hr will not be captured to allow dam-newborn fawn bonding (White et al. 1972). Precautions will be taken to minimize injury or abandonment by the dam from the capture operation based on recommendations of Beale and Smith (1973) and Livezey (1990). The radio collars and all handling equipment will be stored in containers with native shrub/tree (sagebrush or juniper) material to help mask "unnatural" odors. Plastic gloves will be worn by the handlers. Fawn Processing: Captured fawns will be blindfolded, weighed, and the right hind foot will be measured as an index of skeletal size. If age is not known from observed parturition, it will be estimated by hoof condition and length, umbilical condition, and general body size, activity, and hair coat condition will be subjective measures of age. Each fawn will be individually marked with a plastic ear tag in the right ear; the ear tag will have engraved numbers which contrast to the color of the tag. An expandable radio transmitter (with mortality sensor) collar will be fitted; the collar will be designed to drop off after about 6 months. Before release, the fawn will be examined for injuries and it's general health and condition will be noted and recorded. Each fawn will be carefully observed after release and, . if possible, observed until it is rejoined by it's mother. �165 Fawn and Doe Monitoring: Each fawn and doe will be radio located daily for the first 6 weeks of the fawn's life, every 3rd day for the next month, and then once weekly through the end of November. Precautions will be taken to avoid disturbance of the fawn or dam during radio relocation except when prescribed bed site measurements are taken (see below). Radios on mortality signal will be located and the carcass (if available) will be examined for cause of death using criteria outlined in White (1973), Wade and Bowns (1984), Acorn and Dorrance (1990), and Andelt et al. (1998). If the carcass is missing, the general area will be surveyed for signs of predation and our best judgement of cause of death will be made or the cause will be recorded as "unknown". Birth and Bed Site Measurements: Vegetative cover will be measured at birth sites when they can be positively identified. Measurements will be taken at bed sites every 5th day to minimize disturbance of individual fawns. If a bedded fawn is found and does not flush, the point of observation will be flagged and GPS coordinates will be recorded. Distance to the bed sitewill be measured with a laser range finding scope (e.g. Bushnell "yardage Pro") and the direction to the site will be measured with a hand held compass. Marked bed sites will be measured the next day to minimize any changes in vegetation from time of use. The objective will be to obtain lObed site measurements for each fawn during their first 8 weeks of life. Micro habitat vegetational measurements of birth and bed sites will be a modification of those taken by Benzon (1996). Cover of shrubs by species, grass, and forbs will be estimated using 4 0.1 m2 (20 x 50 em) quadrants (Daubenmire 1959) 'placed in the 4 cardinal directions length-wise from the bed center; the 20 em end of the frame will be centered on and touching the bed site center stake. Visual obstruction from the birthlbed site will be estimated from the 4 cardinal directions using a robel pole viewed from a distance of 4 m and height of 0.5 m from the site (Robel et al. 1970). No adjustments will be made for relief in the terrain. For every birth or bed site measured, we will take identical measurements on IO.random sites within aBO ha square area centered on the bed site. The random sites will be located using random coordinates. Sample Sizes and Power of Tests: Estimates of the means and variances of2 or more treatments are necessary before conducting power analyses. Estimates of these parameters are not known for the proportions of fawns killed by various predators, and the effect of vegetation height on susceptibility of fawns to predation. However, some educated guesses about the magnitude of these parameters will help determine sample sizes. We calculated power for our hypotheses of primary management interest including: 1) proportions of fawns killed by various predators, and 2) the effect of vegetation height on susceptibility of fawns to predation. We calculated for vegetation height vs. susceptibility to predation by using 2 height categories, whereas the actual analysis' will consist of regression using a continuum of vegetation heights. Thus, the power of the actual experiment likely will be somewhat less than presented below. We did not calculate power for our hypothesis of differences in vegetation height at fawn bed sites and random sites because sample sizes will be inherently larger that number 2 above. An alpha of 0.05 was used in all estimates of power. Power of 0.8 or higher is desired. Data Analyses: The proportions of fawns killed and the proportions killed by various causes of mortality (predation, disease, malnutrition) will be compared among study areas with chi-square (proc Freq, SAS Institute, Inc. 1988). The proportion of fawns killed by various predators and the proportions of male and female fawns killed by predators will be compared with chi-square. The effect of distance between bed sites and the distance between fawns and their mothers on fawn survival will be estimated with logistic regression (Proc Genmod, SAS Institute Inc. 1993) by blocking on study areas. Fawn birth weights and measurements, and distances between consecutive bed sites will be compared among study areas with analysis of variance. The effect of vegetation height and density (VOR) on survival of fawns will be estimated with logistic regression (proc Genmod, SAS Institute Inc. 1993) by blocking on study �166 Power for Evaluating if the Expected Relative Proportions of Fawns Killed by Coyotes, Black Bears, and Mountain Lions Varies from Equality (i.e. 0.33 killed by each predator): Fawns killed en) 10 20 30 50 75 100 10 20 30 50 75 100 Coyotes 0.5 0.5 0.5 0.5 0.5 0.5 0.7 0.7 0.7 0.7 0.7 0.7 Proportion killed by various predators Black Bears Mountain Lions 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 Power 0.16 0.27 0.39 0.60 0.79 0.90 0.59 0.89 0.98 1.00 1.00 1.00 I-Sided Power for the Effect of Vegetation Height on Survival of Fawns to 2 Months: Sample sizes Short 20 40 40 80 40 40 40 Tall 20 40 40 80 40 40 40 ProPQrtion surviving Short Tall 0.7 0.8 0.7 0.8 0.7 0.9 0.7 0.9 0.5 0.64 0.5. 0.7 0.5 0.8 Power 0.11 0.18 0.61 0.88 0.27 0.73 0.80 areas (Appendix 1). The height and density of vegetation at fawn bed sites and random sites will be compared with analysis of variance by blocking on study areas. The proportion of vegetation that is composed of forbs, grass, and shrubs will be compared between bed sites and random sites and among study areas with analysis of variance. Significance level to reject the null hypothesis will be at P < 0.05. LOCATION This study will be done on 3 areas, Uncompahgre Plateau (D-19), Middle Park (D-9), and Red Feathers (D-4). WORK SCHEDULE Oct-Dec 1998 Jan-May 1999 Jun-Dec 1999 Jan-Jun 2000 Complete Program Narrative and confirm study area selection Procure and test field equipment and hire summer crew Capture, radio collar, and track fawns Data analysis and report writing PERSONNEL Thomas M. Pojar Co-Principal Investigator, CDOW William F. Andelt Co-Principal Investigator, Department of Fishery and Wildlife Biology, Colorado State University �167 David C. Bowden Statistical Consultant, Department of Statistics, Colorado State University Gary C. White Analytical Consultant, Department of Fishery and Wildlife Biology, Colorado State University ESTIMATED COST Personal Services PFTE costs Andelt Pojar TFTE costs Operating Vehicle (4x4) Radios Field equipment Travel Expenses Capitol TOTAL 5% contingency GRAND TOTAL FY 98-99· FY 99-00b Totalc $ 56,250 $ 45,000 $ 18,216 $ 75,000 $ 60,000 $ 26,717 $131,250 $105,000 $ 44,933 $ 18,306 $ 22,000 $ 13,500 $ 1,800 $ 33,900 na na $ 2,250 0 $197,867 $ 9,893 $207,760 $ 52,206 $ 22,000 $ 13,500 $ 4,050 0 $372,939 $ 18,647 $391,586 o $175,072 $ 8,754 $183,826 "Cost for FY 1998-99, representing field operations during May-June, 1999. "Cost for completion of the 1999 field operation during FY 1999-00, July-December 1999. "Total cost for 1 complete field season spanning 2 fiscal years. LITERATURE CITED Acorn, R. C., and M. 1. Dorrance. 1990. Methods of investigating predation of livestock. Alberta Agric. Prot. Branch. Edmonton. 36pp. Andelt, W. F., M. Bruscino, and C. Niemeyer. 1998. Interpreting the physical evidence of predation on big game animals and domestic livestock. Unpublished. Andelt, W. F., 1. G. Kie, F. F. Knowlton, and K. Cardwell. 1987. Variation in coyote diets associated with season and successional changes in vegetation. Journal of Wildlife Management 51:273277. Anderson, A. E., D. C. Bowden, and D. M. Kattner. 1992. The puma on Uncompahgre Plateau, Colorado. Colorado Division of Wildlife Technical Publication No. 40. Fort Collins, Colorado, USA. Anderson, A. E., and D. E. Medin. 1965a. Reproductive studies. Colorado Game, Fish, and Parks Department, Federal Aid in Wildlife Restoration Job Progress Report, Project W-38-R-4. January:165-193. Anderson, A. E., and D. E. Medin. 1965b. Reproductive studies. Colorado Game, Fish, and Parks Department, Federal Aid in Wildlife Restoration Job Progress Report, Project W-38-R-4. January Part 4:522-552. �168 Anderson, A. E., and D. E. Medin. 1966. Reproductive studies. Colorado Game, Fish, and Parks Department, Federal Aid in Wildlife Restoration Job Progress Report, Project W-38-R-4. January Part 2:275-290. Autenrieth, R. E. 1982. Pronghorn fawn habitat use and vulnerability to predation. Pronghorn Antelope Workshop, Dickinson, ND. 10:1l2-13l. Autenrieth, R. E. 1984. Little lost pronghorn fawn study - condition, habitat use and mortality. Pronghorn Antelope Workshop, Corpus Christi, Texas. 11:49-70. Bartmann, R. M. 1997. Mule deer population status in Colorado - update. Colorado Division of Wildlife, Fort Collins, Colorado, Unpublished report. Bartmann, R. M. 1998. A prospectus on mule deer in Colorado. Colorado Division of Wildlife, Fort Collins, Colorado, Unpublished report. Bartmann, R. M., and T. M. Pojar 1998b. Deer reproduction assessment. Colorado Division of Wildlife, Federal Aid in Wildlife Restoration Job Progress Report, Project W-153-R-l1, July. Bartmann, R. M., and T. M. Pojar 1998a Experimental deer inventory. Colorado Division of Wildlife, Federal Aid in Wildlife Restoration Job Progress Report, Project W-153-R-ll, July. Bartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monographs 121:1-39. Beale, D. M., and A. D. Smith. 1973. Mortality of pronghorn antelope fawns in western Utah. Journal of Wildlife Management. 37:343-352. Beck, T. 1995. Development of black bear inventory techniques. Colorado Division of Wildlife, Federal Aid in Wildlife Restoration Job Progress Report, Project W-153-R-8, WP 5, Job 2, July. Benzon, T. A. 1996. Mortality and habitat use of white-tailed deer fawns in the Northern Black Hills, South Dakota, 1991-1994. Completion Report. Pittman-Robertson Project W-75-R-34. Bowyer, R. T., and V. C. Bleich. 1984. Effects of cattle grazing on selected habitats of southern mule deer. California Fish and Game 70:240-247. Boyer, R. T., 1. G. Kie, and V. Van Ballenberghe. 1998. Habitat selection by neonatal black-tailed deer: climate, forage, or risk of predation. Journal of Mammalogy 79:415-425. Clements, C. D., and 1. A. Young. 1997. A viewpoint: Rangeland health and mule deer habitat. Journal of Range Management 50:129-138. Connolly, G. E. 1978. Predators and predator control in big game management. Pages 369-394 in 1. L. Schmidt and D. L. Gilbert, eds. Big game of North America Stackpole Books, Harrisburg, Pa Daubenmire, R. 1959. A canopy-coverage method of vegetation analysis. Northwest Science. 33:4364. Filonov, C. 1980. Predator-prey problems in nature reserves of the European part of the RSFSR. Journal of Wildlife Management. 44:389-396. Gaillard, J. M., M. Festa-Bianchet, and N. G. Yoccoz. 1998. Population dynamics oflarge herbivores: variable recruitment with constant adult survival. Trends in Ecology and Evolution 13:58-63. Garner, G. W., J. A. Morrison, and J. C. Lewis. 1976. Mortality of white-tailed deer fawns in the Wichita Mountains, Oklahoma. Proceedings of the Annual Conference of Southeastern Fish and Wildlife Agencies 30:493-506. Geduldig. H. L. 1981. Summer home range of mule deer fawns. Journal of Wildlife Management 45:726-728. Gill, R. B. 1971. Middle Park deer study - productivity and mortality. Colorado Division of Wildlife, Federal Aid in Wildlife Restoration Job Progress Report, Project W-38~R-25. July:189-207. Gruell, G. E. 1986. Post-1900 mule deer irruptions in the Intermountain West. United States Department of agriculture, Forest Service. Intermountain Research Station, Ogden, Utah. General Technical Report INT-206. �169 Horejsi, R G. 1982. Mule deer fawn survival on cattle-grazed and ungrazed desert ranges. Arizona Game and Fish Department Federal Aid Project Report, Phoenix. 43pp. Laycock, W. A. 1991. Stable states and thresholds of range condition on North American rangelands: A viewpoint. Journal of Range Management 44:427-433. Litvaitis, J. A and W. S. Bartush. 1980. Coyote-deer interactions during the fawning season in Oklahoma Southwestern Naturalist 25: 117-118. Litvaitis, J. A, A G. Clark, and J. H. Hunt. 1986. Prey selection and fat deposits of bobcats (Felis rufus) during autunm and winter in Maine. Journal ofMammalogy 67:389-392. Livezey, K. B. 1990. Toward the reduction of marking-induced abandonment of newborn ungulates. Wildlife Society Bulletin 18:193-203. . Loft, E. R, J. W. Menke, J. G. Kie, and R. C. Bertram. 1987. Influence of cattle stocking rate on the structural profile of deer hiding cover. Journal of Wildlife Management 51:655-664. Mathews, N. E., and W. F. Porter. 1988. Black bear predation of white-tailed deer neonates in the central Adirondacks. Canadian Journal of Zoology 66: 1241-1242. Medin, D. E., and A. E. Anderson. 1979. Modeling the dynamics of a Colorado mule deer population. Wildlife Monographs 68: 1-77. Ozoga, J. J., and L. J. Verme. 1982. Predation by black bears on new-born white-tailed deer. Journal ofMammalogy 63:695-696. Ozoga, J. J., and L. J. Verme. 1986. Relation of maternal age to fawn-rearing success in white-tailed deer. Journal of Wildlife Management 50:480-486. Riley, S. J. and A. R Dood. 1984. Summer movements, home range, habitat use, and behavior of rriule deer fawns. Journal of Wildlife Management 48: 1302-1310. _ Robel, R J., J. N. Briggs, A. D. Dayton, and L. C. Hulbert. 1970. Relationships between visual obstruction measurements and weight of grassland vegetation. Journal of Range Management. 23:295-297. Rothchild, S. L. 1993. Mortality, home range, and habitat use of pronghorn fawns within tallgrass prairie of eastern Kansas .. M.S. thesis, Emporia State University. 85pp. Rothchild, S. L., E. J. Finck, and K. E. Sexson. 1994. Mortality and bedding site selection of pronghorn fawns in tall grass prairie. Proceedings of the Pronghorn Antelope Workshop 16:104-116. Salwasser, H., S. A Holl, G. A Ashcraft. 1978. Fawn production and survival in the North Kings River deer herd. California Fish and Game 64:38-52. SAS Institute Inc. 1988. SAS/STAT user's guide. Version 6.03. SAS Institute Inc., Cary, North Carolina. SAS Institute Inc. 1993. SASR Technical Report P-243, SAS/STATR Software: The GENMOD Procedure, Release 6.09, Cary, North Carolina Schlegel, M. 1976. Factors affecting calf elk survival in Northcentral Idaho. A progress report. Proceedings of the Western Association of State Game and Fish Commissioners 56:342-355. Skovlin, J. M. 1991. Fifty years of research progress: a historical document on the Starky Experimental Forest and Range. General Technical Report PNW-GTR-266. Portland, Oregon: U. S. Department of Agriculture, Forest Service, Pacific Northwest Research Station. Smith, R 1983. Mule deer reproduction and survival in the LaSal Mountains of Utah, 1983. M.S. thesis, Utah State University, Logan. Smith, R H., and A LeCount. 1979. Some factors affecting survival of desert mule deer fawns. Journal of Wildlife Management 43:657-665 Smith, R H., D. J. Neff, and N. G. Woolsey. 1986. Pronghorn response to coyote control - a benefit cost analysis. Wildlife Society Bulletin 14:226-231. . Steigers, W. D., Jr. 1981. Habitat use and mortality of mule deer fawns in western South Dakota Ph.D. Dissertation, Brigham Young University. 206pp. �170 Steigers, W. D., Jr., and 1. T. Flinders. 1980. Mortality arid movements of mule deer fawns in Washington. Journal of Wildlife Management 44:381-388. Stout, G. C. 1982. Effects of coyote reduction on white-tailed deer productivity on Fort Sill, Oklahoma. Wildlife Society Bulletin 10:329-332. Tucker, R. D. and G. W. Gamer. 1983. Habitat selection and vegetational characteristics of antelope fawn bedsites in West Texas. Wade, D. A, and J. E. Bowns. 1984. Procedures for evaluating predation on livestock and wildlife. Texas Agricultural Experiment Station Bulletin B-1429, College Station. Welker, H. J. 1986. Fawn mortality in the Lake Hollow deer herd, Tehama County, California California Fish and Game 72:99-102. White, G. C., R. A Garrott, R. M. Bartmann, L. H. Carpenter, and A W. Alldredge. 1987. Survival of mule deer in northwest Colorado. Journal of Wildlife Management 51:852-859. White, M. 1973. Description of remains of deer fawns killed by coyotes. Journal of Marnmalogy 54:291-293. . White, M., F. F. Knowlton, and W. C. Glazener. 1972. Effects of dam-newborn fawn behavior on capture and mortality. Journal of Wildlife Management 36:897-906. Prepared by _ Thomas M. Pojar Wildlife Researcher Colorado Division of Wildlife William F. Andelt Associate Professor Department of Fishery and Wildlife Biology Colorado State University Appendix 1. Logistic regression table for the effect of vegetation height on survival offawns. The analyses will be conducted with SAS using Proc Genmod. Numerator Source Height Areas DF 1 2 M.S. height area Denominator DF M.S. Fawns -5 Error (Model) Fawns -5 Error (Model) �127 Colorado Division of Wildlife Wildlife Research Report July 2000 JOB PROGRESS State of Project No. Colorado Cost Center 3430 W-153-R-13 Mammals Program Work Package No. _---'3:;...;:0'-"'0..:...1 Task No. 1 Period Covered: REPORT Deer Conservation _ Investigating Factors Contributing to Declining Mule Deer Numbers July 1 1999 June 30, 2000 Author: T. M. Pojar Personnel: _) .-- / T. Baker, T. Beck, C. Bishop, G. Bock, D. Coven, L. Dehart, B. Diamond, B. Dreher, J. Ellenberger, V. Graham, J. Griggs, D. Gustine, B. Hoffner, B. Lamont, M. King, K. Larsen, M. Mclain, H. McNally, K. Miller, M. W. Miller, E. Myers, J. Olterman, D . Schweitzer, T. Spraker, B. Watkins, S. Znamenacek. ABSTRACT Long-term data indicate that fawn:doe ratios have shown a significant and consistent decline over the past 20 years in most deer management units in Colorado as well as in most of the western states. This has resulted in a general decline in mule deer (Odocoileus hemionus) density and diminished recreational potential from this major resource. It is important for the Division of Wildlife to understand the factors contributing to the reduced fawn:doe ratios and the declining deer populations. Neonatal survival (birth to 6 months of age) is an important component of population recruitment. This is the second year of monitoring cause-specific neonatal fawn mortality on the Uncompahgre Plateau. In 1999, a sample of 50 fawns was radioed and during the 2000 fawning season 88 fawns were captured. Fawns were captured as . near birth as possible and are monitored to about 6 months of age. A new design radio collar that weighed 1109 was used on the 2000 season fawns. The collar was expandable and used latex tubing to facilitate drop-off in about 6 months. Neonatal fawn survival for 1999 was 38%. Of the 50 fawns collared, 30% died of sickness/starvation, 26% from predators (12% coyote, 8% feline, and 6% bear), 4% from unknown causes, and 2% from poaching. Thirty-one of 50 fawns died with 58% dying from causes other than predation. Monitoring is in progress for the 2000 season fawns. Early season 2000 mortality is less than it was in 1999. Deaths from sickness/starvation are half (10% vs. 20%) what they were in 1999 (as of . August 1) and predation is about three-fourths (15% vs. 20% that of 1999 (Figure 1). June 2000 was warm and dry, which may have affected early fawn survival. On the Uncompahgre, December fawn:doe ), ratios are negatively correlated (r = -0.8183) with the preceding June precipitation (R~.6697, P=0.0038). /i: d However, mean temperature for June had a weak positive correlation (r = 0.2756) and was not significant (R2 =.0.0760, P = 0.4409). The peak offawning was between June 14thand June 26th when 93% of the -, �128 fawns were caught (Figure 2). Year 2000 fawns were slightly larger than the 1999 fawns in terms of body weight and hindfoot length. Neither the difference in weight, 4.4692 (2000)vs. 4.3714 kg (1999), nor the difference in hing foot length, 10.474 (2000) vs. 10.297 (1999) inches, was statistically different, P = 0.5845 and P = 0.1309, respectively. �129 IMPACT OF PREDATION AND VEGETATIVE COVER ON MULE DEER FAWN SURVIVAL Thomas M. Pojar P. N. OBJECTIVES 1. Identify agents of neonatal mule deer fawn mortality from birth to 6 months of age. SEGMENT OBJECTIVES 1. Capture and radio collar 30-35 neonatal fawns each on the Uncompahgre Plateau (D-19), Middle Park (D-9), and Red Feathers (D-4). 2. Measure vegetative density and height at fawn bed sites. 3. Measure distance between successive bed sites for individual fawns. 4. Compare fawn survival and causes of mortality among 3 vegetational and ecologically different study areas. 5. Correlate height and density of vegetation at fawn birth sites and bed sites with percent of fawns killed by predators. 6. Compare height and density of vegetation at fawn bed sites and random sites. INTRODUCTION Major changes in the original Program Narrative include: 1. Limit the fawn collaring effort to the Uncompahgre Plateau (D-19). Because of the size of the area and diversity of vegetative communities on the Uncompahgre and because of intense political interest in this area, all resources were concentrated on the Uncompahgre during this segment. The target sampling intensity was increased to 80 fawns with emphasis placed on getting the sample widely distributed over the entire 2,300 square mile area. 2. Abandon the idea of measuring vegetation at bed sites because of evidence that persistent disturbance of bed sites reduces survival of young ungulates (Philips 1998). 3. The logistics of placing an adequate crew in remote fawning areas and the unknown of whether or not fawns could be captured in sufficient numbers given deer density, terrain, and vegetative cover encountered negated many of the objectives in the PN. In essence, this investigation was reduced to placing emphasis on capturing fawns on the Uncompahgre and monitoring their survival and causes of mortality. Summer fawn survival and causes of mortality are 2 important factors impacting deer population performance. STUDY AREA The study area is described in Pojar and Andelt (1999). METHODS Radio collared does were used to locate fawning areas although no special effort was made to capture the fawns of radioed does. The strategy for capturing fawns involved observing does for behavioral and physical signs, such as udder development, of having fawned. If it was determined that these indicators suggested that the doe had fawned, the area was searched for the fawn(s). Once located, �130 the fawn was approached from behind (out of direct eyesight) and restrained for weighing, measuring, and putting on the radio collar. Processing usually took less than 5 minutes. The collars used were made of 1.5-inch elastic with an expansion ratio of 1.5:1 and equipped with surgical tubing to allow the collar to drop off after about 6 months. Each collar weighed 110 g. The circumference of the collar was 10 inches, 6 of that was elastic. The 6 inches of elastic has the ability to expand to 9 inches. A l-inch loop was sewn in the elastic reducing the collar circumference to 9 inches. The loop was sewn with 4 stitches of cotton thread with the intent that these would rip out within 2-3 weeks after the collar was put on a fawn. When the collar was put on a fawn it was 9 inches in circumference and fit quite loosely. When the loop was released the circumference increased to 10 inches and as the elastic relaxed the collar could expand another 3 inches giving a totally expanded circumference of 13 inches. Collars were hung outside in the shade for several days before use to help dissipate any fabric and human scents associated with manufacture. When taken to the field in workers backpacks, the collars were sealed in zip loc bags. If a collar was to be recycled, i.e. removed from a dead fawn and reused, it was washed in tap water (no soap), hung out to dry, then placed in aplastic bag with native forage. RESULTS By recycling some of the collars, 88 fawns were radioed. The sample was roughly distributed according to deer density on the summer range. The sample of radioed fawns were distributed across the Uncompahgre Plateau as follows: North (Cold Springs) 29, central (25 Mesa) 16, south (Celesca) 33, and far south (10). Thus far, the collars are expanding as expected. From recovered collars, it appears the cotton stitching rips out in about 4 weeks or less. There was only one instance of suspected slippage of a collar and it is possible the fawn was killed or scavenged and the collar carried away from the carcass. Of the fawns recovered, there was no sign of the collar causing abrasions on the skin, however, there was slight hair loss in some cases from the collar being loose enough to rotate or move on the neck. The fawns captured during 2000 had a mean weight of 4.47 kg compared to 4.37 kg in 1999; this difference was not significant (P=0.5845). The hind foot length of 10.48 inches in 2000 was not different from that of 1999, 10.30 inches (P=0.1309). There was a noticeable difference in the weather, both precipitation and temperature, during the fawning period between 1999 and 2000 with the most recent year being warmer and dryer. Early sickness/starvation deaths (through August 1) in 2000 were half (10% vs. 20%) what they were in 1999 (Figure 1). It was theorized that the warmer, dryer June weather was a key factor in neonate fawn survival. Weather data for the Montrose area was tabulated for the past 10 years. I tested the hypothesis that June temperature and precipitation are related to the subsequent early winter (December) fawn:doe ratios collected during helicopter classification surveys done by management personnel. The correlation with precipitation was highly significant and negatively correlated (r = -0.8183) with subsequent fawn:doe ratios, R2 = 0.67, P = 0.0038; temperature had a slight positive correlation (r = 0.2756) with fawn:doe ratios but it was not significant, R2 = 0.0760, P = 0.4432. Although the search for newborn fawns began in early June, the first fawns were not caught until June 14thin 2000. In the 13-day period between June 14thand June 26th, 93% of the fawns were caught. The last fawn was caught on July 9th (Figure 2). In 1999, the 17-day period from June 13 to June 30,96% of the fawns were captured; the first fawns were captured on June 9th. From this data there is no evidence that a double fawning peak is occurring on the Uncompahgre Plateau. Data on captured fawns is in Table 1. Although the information for 2000 regarding mortalities only goes through August Pt, Figure 3 shows a comparison of total cause-specific mortality for 1999 (total) and 2000 (through August 1~. �131 LITERATURE CITED Philips, G. E. 1998. Effects of human-induced disturbance during calving season on reproductive success of elk in the upper Eagle River valley. Ph. D. Dissertation, Colorado State University, Ft. Collins. Pojar, T. M. and W. F. Andelt. 1999. Investigation offactors contributing to declining mule deer numbers. Wildlife Research Report, Mammals Research, Federal Aid Projects, Job Progress Report, Project W-153-R-12, Colorado Division of Wildlife, Fort Collins, Colorado, USA. Table l. Neonatal mule deer fawns captured and radio collared on the Uncompahgre Plateau, Colorado 2000. The mortality data is very preliminary and represents only deaths through August 1,2000. Probable Cause Radio I.D. Date of Capture location UTM Hindfoot Date Sex Wt Recovered of Death Capture Coordinates (kg) (inches) East North 150.023 6-15 703980 4268055 F 10.75 4.4 150.033 6-18 700954 4278742 F 10.25 4.7 150.044 6-16 702309 4280011 F 11.125 5.3 150.054 6-24 703446 4267697 F 11.25 5.1 150.064 6-18 700353 10.625 4274667 M 5.0 150.074 6-24 703050 4267256 M 11.00 4.4 150.084 6-24 703442 4267726 M 6.1 11.00 150.094 6-18 704622 4274176 M 11.00 5.0 150.104 6-22 700255 10.625 4282798 F 5.5 150.114 6-19 697815 5.3 11.375 4283738 M 150.134 6-20 6+ 11.875 703390 4279744 M Sick/starve 150.146 6-19 7-9 701207 4279027 M 9.75 3.9 150.155 6-20 702179 10.125 4275357 F 4.1 150.163 6-18 700357 10.50 4283325 M 4.2 7-19 Sick/starve 150.184 6-22 703468 10.00 4279642 M 4.0 150.192 6-20 702293 4.5 10.125 4281859 M 150.204 6-22 '4.7 700428 4283374 M 10.25 150.214 6-26 704463 10.50 4272215 F 4.7 Coyote 150.223 6-21 751358 6-23 4236752 M 3.6 10.50 150.234 6-20 727652 10.125 4252092 F 3.5 723136 150.244 6-26 4257872 F 10.625 5.7 150.254 6-22 730938 4264027 M 3.9 9.625 150.263 6-19 750827 4230756 M 4.0 10.25 150.272 6-25 700728 10.25 4284815 M 3.75 150.283 6-25 703475 10.50 4279355 F 5.0 10.00 150.294 6-26 241593 4235614 M 4.2 Sick/starve 150.303 6-21 760162 11.25 7-31 4237065 M 5.2 7-3 Bear 150.314 6-23 732743 4251998 M 11.125 5.0 150.324 6-26 241806 9.50 4236255 M 2.9 721362 10.00 8-1 Coyote 150.334 6-22 4264368 M 4.6 150.344 6-20 720095 4265861 F 5.2 11.00 Coyote 150.353 6-24 730048 9.75 7-27 4263042 F 3.4 150.362 6-20 748878 4231358 M 10.625 4.6 150.374 6-21 756526 10.75 4248692 M 5.1 150.383 6-26 728565 4254798 M 3.4 9.875 Coyote 150.393 6-21 10.125 7-5 743115 4255001 F 3.8 150.423 6-25 10.25 730831 4263910 M 4.0 150.434 6-18 757228 4232141 F 5.3 11.25 7-17 _~LcJcl~L ___ 6-20 4.~~!~2~_ __I4.~~I~ ___ J':____4.]___ J.9.:~Q_____ -----------__ l_?.9A4.4. -------- �132 Table 1 Continued Radio I.D. Date of Capture 150.455 150.464 150.473 150.484 150.495 150.504 150.514 150.524 150.545 150.553 150.564 150.573 150.584 150.593 150.604 151.114 151.123 151.144 151.153 151.164 151.173 151.195 151.203 151.233 151.245 151.254 151.263 151.274 151.284 151.294 151.314 151.324 151.333 150.013A 150.013B 150. 124A 150. 124B 150.172A 150.172B 150.403A 150.403B 150.415A 150.415B 150.533A 150.533B 151.134A 151.134B 151.183A 151.l83B 6-19 6-16 6-19 6-14 6-17 6-19 6-16 6-19 6-19 6-23 6-17 6-15 6-17 6-14 6-20 6-15 6-16 6-18 6-18 6-18 6-18 6-18 6-26 6-17 6-17 6-14 6-14 6-22 6-22 6-22 6-19 6-22 6-15 6-26 7-6 6-24 6-27 6-20 6-26 6-26 7-9 6-18 6-26 6-21 6-29 6-16 6-30 6-18 7-5 Capture location UTM Coordinates North East 762275 4233303 755233 4247651 760381 4232242 754488 4235165 756570 4233083 762275 4233303 4247828 754392 727428 4251901 749686 4229850 755706 4231637 4246702 757702 752002 4249457 754295 4247567 4249079 752525 754867 4231692 756942 4231730 754848 4331643 751910 4249670 751910 4249670 4249165 752547 752639 4248689 742905 4240385 731972 4251646 4251207 750332 754406 4248275 248077 4233576 248077 4233576 762261 4224531 761180 4220892 761688 4221672 761310 4220980 761688 4221672 761539 4223050 703376 4267562 704293 4269378 705226 4269426 4274463 704284 702297 4280730 730004 4261815 728978 4254395 702133 4280026 730056 4262079 730074 4261926 751358 4236752 707752 4270072 755167 4231731 707752 4270072 746380 4247269 701385 4279114 Sex Wt (kg) Hindfoot (inches) M M M F M M M F M M F M M M F M F F F F F M F M M F M F F F F F F F F F F M M M M F F M F F F F M 5.7 3.1 5.3 3.8 4.1 5.2 4.5 3.6 5.1 3.8 3.0 4.7 4.8 3.6 4.8 4.3 3.6 3.8 3.6 5.0 4.1 4.7 5.7 4.7 5.2 4.3 4.6 4.8 6.1 3.4 3.1 3.6 3.5 6.0 5.8 3.4 4.0 4.9 5.0' 4.3 6.7 3.5 3.0 3.7 6.1 4.4 5.0 4.0 6+ 11.50 9.625 10.50 10.50 10.25 10.75 10.625 10.25 10.25 10.50 9.625 11.00 11.125 10.125 10.25 10.50 9.875 10.25 9.75 10.50 10.625 10.75 11.125 10.25 11.375 10.83 10.25 10.43 12.20 10.16 8.86 10.00 9.76 11.00 11.00 8.00 10.25 11.00 10.875 10.625 11.50 9.875 8.875 10.50 10.625 10.75 11.50 10.125 11.75 Date Recovered Probable Cause of Death 7-8 Sick/starve 6-30 7-31 6-25 Coyote Bear?· Sick/starve 6-23 Sick/starve 7-3 Bear 6-24 . 7-5 6-23 7-31 6-21 Sick/starve Bear Coyote Sick/starve Coyote 6-26 7-6 Coyote Coyote �133 YEAR 2000 - Through Number of fawns radioed Number of fawn deaths Cause Predation Sick/starve Total Alive % of morts Number 13 9 66 August 1 88 % of Total Sample 22 Predation 'lb % of total 59% 41% 15% 10% 75% Sick/starve 10% Total Alive 75% YEAR 1999 - Through Number of fawns radioed Number of fawn deaths Cause Predation Sick/starve Total Alive Number 10 10 30 August 1 50 20 % ofTo~1 % of morts % of total 50% 20% 50% 20% 60% Sample Predation 20% Figure 1. Comparison of cause-specific mule deer fawn mortality, Uncompahgre Plateau, Colorado. represents mortality from collaring through August 1st. 14 "'C Cl) 12 •... .aa. 10 C'O () 8 - - 6 I-- o c ~ u. o o Z 4 f- f- 2 o lLU .IL_I_ Jul8 Jun 14 Jun 18 Jun 22 Jun 26 Jun 30 Jul4 Jun 16 Jun 20 Jun 24 Jun 28 Jut 2 Jut6 Figure 2. Capture dates for year 2000 fawns from the Uncompahgre Plateau, Colorado. Data �CAUSE-SPECIFIC FAWN MORTALITY - UNCOMPAHGRE 2000 (Through Aug 1) Number of fawns radloe 88 Num ber of fawn deaths 22 Cause Number % of mort % of total 8 36% 9% Coyote 4 18% 5% Bear Unk Pred 1 5% 1% Sick/starve 9 41% 10% 66 75% Total Alive •....• w .j::o. r----------------------------, % of Total Sample Bear 5% Unk Pred 1% Sick/starve 10% % or Mortalltl •• '" ,... CAUSE-SPECIFIC FAWN MORTALITY - UNCOMPAHGRE 1999 (Total for year) Number of fawns radloe 50 ...--'---------------------------, Number of fawn deaths ·31 Number % of mort % of total Cause Coyote 6 19% 12% 4 13% 8% Feline Bear 3 10% 6% Sick/starve 15 48% 30% Totalilive Unknown 2 6% 4% 38% Poached 1 3% 2% Total alive 19 38% ./. of Total Sample Coyote 12% Bear 6% % or Mortalltl.s •..". Unknown 4% Sick/starve 30% '''' Figure 3. Comparison of cause-specific mortality for 1999 and 2000, Uncompahgre Plateau, Colorado. The tabulation of 1999 includes the entire period from fawn capture to approximately 6 months of age. The data for 2000 only goes through August 1, 2000 (the date of this writing). � https://cpw.cvlcollections.org/files/original/8b8f55665099434ac957088b1ddf478a.pdf b4be7d2fb2190aa2718befefd68d3bea PDF Text Text 83 Colorado Division of Wildlife Wildlife Research Report July 1993 JOB PROGRESS REPORT State of ______..!C~o~l~o~r~a~d~o~---Project No. W-153-R-6 Mammals Research Work Plan No. -~3_ _ _ _ _ _ __ Elk Investigations Job No. ---~9_________ Estimating Survival Rates of Elk and Developing Techniques to Estimate Population Size Period Covered:; Author: July 1, 1992 - June 30, 1993 D. J. Freddy Personnel: R. Bartmann, and C. McCarty, CDOW; White, CSU. C. Vardeman, D. Bowden and G. Abstract A detailed study plan for this project was completed and approved through the required peer-review system. The study area selected was that portion of Game Management Unit 42 south of Newcastle and Rifle, Colorado . Two elk corral traps were refurbished for use in December 1993 and arrangements made to contract trapping of 100 elk using a helicopter . Radio-collars for adult cows and male and female calves were designed and purchased. �85 JOB PROGRESS REPORT ESTIMATING SURVIVAL RATES OF ELK AND DEVELOPING TECHNIQUES TO ESTIMATE POPULATION SIZE David J. Freddy P. N. OBJECTIVE Estimate survival rates of adult female and calf elk and develop techniques to estimate population size. SEGMENT OBJECTIVES 1. Complete detailed and approved study plan and select study area. 2. Refurbish elk traps and identify potential trap-site locations within the selected study area. 3. Design expandable radio-collars for calves and purchase and assemble collar materials transmitter units. A complete copy of the approved study plan presented. �86 STUDY PLAN FOR RESEARCH FOR 'FY 1992-93 - 1996-97 V State of ___c~o~l~o~r~a~d~o:.-_ _ Project No. 4400 Work Plan No. Job No. Hunt 3 9 0715 Mammals I Research Elk Investigations Estimating Survival Rates of Elk and Developing Techniques to Estimate Population Size ESTIMATING SURVIVAL RATES OF ELK AND DEVELOPING TECHNIQUES TO ESTIMATE POPULATION SIZE Principal Investigators David J. Freddy, Wildlife Researcher, Mammals Research R. Bruce Gill, Wildlife Research Leader, Mammals Research Cooperators John E. Ellenberger, Wildlife Biologist, Northwest Region James H. Olterman, Wildlife Biologist, Southwest Region David C. Bowden, Professor Statistics, Colo. St. Univ. Gary C. lJltite, Professor Wildlife Biology, Colo. St. Univ. STUDY PLAN APPROVAL Prepared by: _____________Date : _ _ _ __ Submitted by: ____________Date : _ _ _ __ Reviewed by: _____________Date : _________ _____________Date: _ _ _ __ _____________ Date: _ _ _ __ _____________ Date: _ _ _ __ Approved by: _____________ Date : _ _ _ __ Biometrician _____________Date: _ _ _ __ V �87 FREDDY STUDY PLAN PROGRAM NARRATIVE State of ---~C~o~l_o~r~a~d~o____ Project No. 4400 Work Plan No. Job No. I. \....,,._/ Hunt 3 9 0715 Mammals I Research Elk Investigations Estimating Survival Rates of Elk and Developing Techniques to Estimate Population Size NEED Elk (Cervus elaphus nelsoni) are a high-profile wildlife resource throughout much of Colorado because they provide recreation for persons who hunt, watch, and photograph wildlife (Freddy et al. 1993). The popularity of elk with the public is undoubtedly related to the expanding geographic distribution and increasing size of the State's elk population. This burgeoning elk resource has many benefits but frequent social, political, and economic conflicts suggest elk, to some degree, have reached a "social" carrying capacity. Balancing these uses and conflicts by maintaining acceptable numbers of elk is a challenge for future management of this species in Colorado. The positive and negative economic impacts of elk are often the most influential forces directing management of elk. In 1990 and 1991, 193-194,000 hunters harvested 46-51,000 elk in Colorado (CDOW 1990, 1991). In doing so, hunters generated at least $20 million in yearly license revenue to the Colorado Division of Wildlife (CDOW) and contributed about $250 million in direct and secondary expenditures to the yearly economy of Colorado (Freddy et al. 1993). Revenues derived from elk hunting provide the financial resources for many wildlife programs of the CDOW and for many private businesses in the State. However, rising numbers of elk and attendant harvests have also brought about conflicts with agricultural interests. Demands to reduce elk populations and expand payments for damage by elk to cultivated and native forage on private lands have resulted in the CDOW altering statewide objectives for elk from increasing to decreasing the total population and creating a Habitat Partnership Program with private land owners. This partnership to cooperatively manage elk that impact private lands may reach expenditures of $1 million annually. The core of the conflict in elk management, in part, stems from inadequate estimates of population size resulting in an inability to establish management objectives for specific populations that are agreeable to competing interests. Even when objectives are established, we can not readily quantify impacts of management decisions because current methodology for monitoring populations is neither precise nor accurate. �88 FREDDY STUDY PLAN The CDOW has relied on computer modeling, rather than direct measurements, to generate estimates of population size. This approach has been inadequate because model outputs have not been validated with independent estimates of population size. Estimates of population size and subsequent proposed harvest levels have, therefore, been subjected to endless debate. Computer models based on POP-II (Bartholow 1992) or POPMOD ('White 1992) function best when natural rates of survival or mortality, hunter harvests, and population size are reliably measured. In the current modeling process, survival of calves from June 1 until September 1 is implicitly estimated by measured post-hunting season calf:cow ratios, which along with bull:cow ratios, are used to estimate composition of the pre-hunting season population. However, the size of the preseason population is not measured. From this preseason population, hunter harvests, which are adequately measured, are subtracted resulting in a prewinter population. The prewinter population is multiplied by rates of survival, which are not measured, to estimate a postwinter population and subsequently the next preseason population. Obvious missing links in this modeling process are measured estimates of survival for calves and adult females during winter (Nelson and Peek 1982) and population size at some time during the annual cycle. Importantly, estimates of population size independent of the modeling process could be used to align models on measured values. The current modeling process is prone to many subjective estimates regarding population size and survival rates which has resulted in diminished confidence in using models to guide management of elk populations (Freddy 1987). The CDOW recognized the need to improve monitoring status of ungulate populations during 3 planning efforts. First, the statewide Long Range Plan (LRP), approved in 1988 and updated in 1991 (CDOW 1991b), highlighted the need for improved population data to accommodate increasing recreational demands and reduce conflicts among competing interests. The LRP targeted 25% of the revenues generated from increasing hunting license fees towards improving and expanding efforts to monitor ungulate populations. In 1991, increased fees associated with elk licenses generated approximately $4.5 millon in new revenue potentially resulting in $1.1 millon towards expanded population monitoring. Second, CDOW terrestrial biologists at a statewide meeting held in July 1991 stated that obtaining quantified estimates of survival rates and population size were their priority needs for developing more reliable models of elk populations. Third, the CDOW Deer and Elk Management Analysis Guide, approved in November 1991, identified as priority issues the need to obtain reliable estimates of population size and survival rates of calves and adult female elk (Freddy et al. 1993). Estimating both survival rates and population size are fundamental to developing systems for monitoring and predicting changes in elk populations. We believe estimating calf and adult female survival rates during winter and annual survival rates of adult females are a higher priority than estimating adult male survival rates primarily because most males are harvested when they reach legal age and contribute little to problems of either population growth or decline. Models having valid estimates of survival rates for calves and adult females along with currently obtained estimates of harvests and population composition will provide more defensible estimates of population . V �89 FREDDY STUDY PLAN size and trend than similar estimates derived from current models. However, an independent estimate of population size generated from a sampling system designed to enumerate elk would provide the opportunity to validate population estimates derived from models. Ideally, estimates of both survival and size would be obtained simultaneously. Changes in either calf or adult female survival have pronounced effects on population growth and greater effects than changes in fecundity (Nelson and Peek 1982). Sensitivity analyses indicated population growth is more affected by changes in survival rates of adult females than to equivalent changes in calf survival. A 15% increase in adult female survival resulted in a population increase 7x that of the increase from a 15% change in calf survival (Table 1). Although small changes in adult female survival can have major effects on population growth if compounded for 10-20 years, we expect calf survival to be more variable among years and our ability to detect changes in calf survival should be greater than detecting smaller, but important changes in adult female survival (White et al. 1987, Bear 1989, Bartmann et al. 1992). Because there are few estimates of either calf or adult female survival rates, we need to measure both simultaneously to document the relative differences in survival rates and patterns of mortality in order to develop more reliable population models. \.-,I' An example of how measurements of survival can alter our perception and modeling of populations involves the Piceance deer population in northwest Colorado. Measured overwinter survival of fawns (0.22) was much lower than perceived, and survival of adult females (0.82) was more stable among years than anticipated (White et al. 1987). Using measured rates of survival in existing models of this deer population resulted in radical changes in estimated population size, and served to focus wildlife managers on probable causes of low fawn recruitment and solutions to improve recruitment (Bartmann and White 1991, Bartmann et al. 1992). �90 FREDDY STUDY PLAN Table 1. Sensitivity analysis for 3, 10, and 15% increases in survival rates of calves and adult females for a beginning population of 1000 elk simulated for 20 years. Initial survival rates for calves (0.705) and adult cows (0.800) essentially stabilized population growth for 20 years. Growth rate increase equals the percent increase in yearly average growth rate compared to stabilized average growth rate, Change in Calf Survival Rate Calf Survival Ad. Female Survival 8 Ad. Male Survival Avg. Yrly. Growth Rate Growth Rate Increase End 20 Yr. Pop. Size 0% +3% +10% +15% 0.705 0.800 0.650 0.82% 0.00% 990 0.726 0.800 0.650 1.49% 82.5% 1109 0. 775 0.800 0.650 3.21% 293.0% 1441 0.811 0.800 0.650 4.61% 465.0% 1742 Change in Adult Cow Survival Rate +15% 0% +3% +10% Calf Survival Ad. Female Survival Ad. Male Survival Avg. Yrly. Growth Rate Growth Rate Increase End 20 Yr. Pop. Size a Ad.-adult ~ 0.705 0.800 0.650 0.82% 0.00% 990 0.705 0.824 0.650 3.35% 310% 1531 0.705 0.880 0.650 12.1% 1378% 4093 0.705 0.920 0.650 22.1% 2609% 8038 V 1 year old. Our lack of adequate population data on elk is not without just cause as known methodologies are limited and expensive. As resource.managers, we have likely reached a threshold where we must expand our investment in monitoring the processes that affect the growth or decline of the elk resource to accommodate the rising and often conflicting demands for managing elk. II . OBJECTIVES Our objectives are to provide reliable estimates of survival rates for calves and adult females overwinter and adult females throughout the year, and develop and test a system for estimating population size. Our efforts will focus on elk inhabiting winter ranges south of NewCastle and Rifle, Colorado. Our specific objectives are: 1) Estimate survival rates of calves from 1 December - 31 May within± 15% of the true survival rate at the 95% confidence interval and identify sources of mortality using adequate numbers of radio-collared calves. 2) Estimate winter (1 December - 31 May) and yearly (1 December - 30 November) survival rates of adult females within± 10% of the true V �91 FREDDY STUDY PLAN survival rate at the 95% confidence interval and identify sources of mortality using adequate numbers of radio-collared adults. 3) Monitor the fate of yearling bull elk radio-collared as calves in relation to antler-point restrictions designed to protect yearling bulls from harvest. 4) Develop and test a system to estimate population size during winter within± 20% of the true population size -at the 90% confidence interval using radio-collared elk to evaluate sighting bias and to provide population estimates based on random plot searches and mark-resight theory. 5) Provide data on general movements, distribution, and dispersal of radio-collared elk within the study area. Establish cooperative studies with universities and land management agencies to investigate specific aspects of habitats used by elk and to relate genetic characteristics of elk to their survival. III. EXPECTED RESULTS OR BENEFITS This investigation will provide estimates of survival rates for calf and adult female elk in a selected elk population during 4 consecutive years. These estimates will immediately assist CDOW biologists not only in refining population models for this specific population but also in providing estimates of survival that may be applicable to modeling other elk populations inhabiting similar habitats. Quantified estimates of survival will allow objective assessment of whether major restructuring of current population models is necessary. In the process of estimating survival rates, we will document sources of mortality. Our efforts to estimate population size are inherently more difficult than estimating survival rates. If we obtain relatively precise and unbiased estimates of true population size, the CDOW will have a method to obtain valid independent estimates of population size to compare with estimates derived from modeling, and therefore, provide another objective basis for assessing the current modeling process. If our efforts provide less than satisfactory results, we will have an objective basis for either continuing with refinements or discontinuing efforts to estimate population size and invest in measuring other population parameters. IV. APPROACH A. Survival General Approach We will estimate survival rates for calves (6 mos old) and adult females (~ 1 yr old) during winter, 1 December - 31 May, and for adult females annually, 1 December - 30 November, by using radio-telemetry collars that emit a mortality pulse code when animals remain motionless for 4-6 hours (White 1983, Bear 1989, White et al. 1987). Radios provide the ability to know the fate of individual animals (alive or dead) over discrete periods of time unlike marking animals with other types of tags that are seldom recovered upon the animal's death (White and Garrott 1990). Improved longevity of batteries �92 FREDDY STUDY PLAN for radio collars will allow us to monitor status of individual animals for up to 4+ years. Survival will be monitored from 1993-94 through 1996-97. V We assume that radio-collaring calf and adult female elk does not bias estimates of survival rates by jeopardizing or enhancing the welfare of individuals. This assumption can be violated in studies involving birds when transmitter packages exceed critical weight/body size ratios (Burger et al. 1991, Foster et al. 1992) but does not appear to be a detectable bias with ungulates when radio collars weigh about 0.8% of an animal's body weight (Garrott et al. 1985, White et al. 1987). Although some level of bias may occur, radio collars will likely provide much less positively biased estimates of survival compared to other indirect methods (Bartmann 1984, Zager and Leptich 1991). Radio collars used in this project will weigh about 1.1 kg and represent about 0.8% of calf and 0.5% of adult female body weights. Collars for adult females will be of fixed circumference and fitted for each individual. Collars for male and female calves will allow for expansion to adult size (Bear 1986, White et al. 1987, Appendix I). We assume that survival of those animals captured unbiasedly represents survival of individuals in the population. Individual behavior, social behavior, trapping method, and distribution of trapping effort all potentially bias those individuals trapped and marked (White el al. 1982, Garrott and White 1982). Recognizing these problems, we will capture elk with the objectives of marking animals throughout the distribution of the population and reducing influences of social hierarchies. The study area will be divided into several zones each having multiple capture sites to assure that marked elk represent the population. Capture quotas for calves and adult females will be established for each zone. A Hughes 500 helicopter and a net-gun will be used to capture individuals from groups of elk in more remote areas where portable corral traps would be inefficient to use. In those areas inhabited by people or having agricultural activities, we will use portable corral traps to capture groups of elk. Elk caught in corral traps are likely less affected by social dominance hierarchies than elk captured with single animal Clover traps (Garrott and White 1982, Bear et al. 1989). For groups of elk captured in traps, only a fraction of each group will be randomly selected and radio collared. We will attempt to collar equal numbers of adult females and calves within each capture method. Trapping will occur from approximately 29 November-20 December each year. While handling animals we will minimize capture stress and follow guidelines for animal welfare (Appendix II). V Each collared animal will be individually recognizable based on radio collar frequency, an external symbol/number on the top surface of the collar, and a numbered metal ear-tag placed in each ear. Radio collars will be white with a black symbol/number (Appendix III). Additionally, male and female calves or yearlings captured but not collared will be ear-tagged to provide additional data on dispersal when and if recovered as harvested animals during hunting seasons. Collared calves, and when possible, yearling females, will be weighed to nearest 0.5 kg. Additionally, hind foot lengths and total body lengths will be measured for calves. Blood samples will be obtained via venepuncture to assess pregnancy status of adult females using progesterone assays (Freddy u �93 FREDDY STUDY PLAN 1991) and from calves for potential genetic analyses related to measured survival rates (Pemberton et al. 1988). All collared elk will be aged as calves, yearlings, or young (2-3 yrs), prime·(4-10 yrs), and old(~ 10 yrs) adults based on visual inspection of tooth replacement and wear (Quimby and Gaab 1957). We will use periodic and systematic aerial and ground searches throughout areas inhabited by collared elk to determine.their life or death status based on pulse rates of collars (Gilmer et al. 1981). Animals will be monitored daily from the ground during winter with aerial searches done twice per month depending upon frequency of mortalities and relative ability to monitor all animals from the ground. During spring, summer, and fall we will conduct monthly or bi-monthly aerial searches. During fall hunting seasons, aerial searches may be conducted weekly in conjunction with intensified ground searches. Mortalities coarsely located during aerial surveys will be found from ground searches using hand-held antennas. Criteria for assigning probable cause of death (primarily during winter) will include body position, presence of bite or claw marks and subdermal hemorrhaging, tracks, drag marks, and when necessary rumen or tissue samples (Wade and Browns 1982). Potential causes of death include starvation, accidental trauma, plant poisoning, predation by black bears, mountain lions, coyotes, and domestic dogs, and legal and illegal hunter harvest (Bear 1989, Schlegel 1977). Analyses Number of animals collared depends on expected survival rates, desired precision, and degree of statistical power desired for comparative tests. The few studies that have estimated survival rates for elk used either age structure analyses or radio collars. Annual survival rates estimated from age structure analyses were: 64% for all calves (overwinter only) in Yellowstone National Park (Houston 1982); 61% for female and 26% for male calves in Jackson Hole, Wyoming (Boyce 1989); 90% for all calves in Colorado's White River population (Laake 1992): and for adult females, 97% (Houston 1982), 68% (Boyce 1989), and 79% (Laake 1992). Annual survival rates based on radio telemetry were: 93% for male and female calves pooled with adult females (nonhunted segments of the population) in northern New Mexico (White 1983); 75-94% for calves through their first 9 months in Rocky Mountain National Park, Colorado (Bear 1989); 100% for calves from shortly after birth through their first winter in central Colorado (de Vergie 1989); 88% for adult females in northern Idaho where all mortality was attributed to hunting (Leptich and Zager 1991); and nearly 100% for adult females in northern Colorado (overwinter only, G.D. Bear pers. comm.). These results suggest that survival may be> 70% for calves and> 85% for adult females·overwinter with annual survival rates for adult females varying due to intensity of hunting. We will radio-collar 75 calves (38 female, 37 male) yearly and initially collar 75 adult females and maintain at least 75 adults(~ 1 year old) collared in the population yearly to estimate rates of survival for these 2 age classes of animals. Assuming 75 collars remain functioning and survival is~ 0.7 or~ 0.9, 95% confidence intervals will be S ± 15% or S ± 8% of the �94 FREDDY STUDY PLAN mean survival rate, respectively. If survival for either calves or adults approaches 0.5 then considerably more collars must be deployed to maintain high precision (Appendix IV). These confidence intervals are calculated as 1.96 ± vVAR(S) where survival (S) is a binomial with a variance, VAR(S) - S(lS)/n, and n is number of collars deployed (White et al. 1982). Power (1-beta) to detect differences in survival rates~ 0.20 among years or age/sex classes will be~ 0.70 but power will be S 0.25 for differences in survival S 0.10 when survival rate is 0.7 and alpha - 0.05. If survival is 0.9, power will be ~ 0.80 for differences in survival~ 0.20 and 0.40 for differences S 0.10 (Appendix IV, White and Garrott 1990:32). Reliably detecting differences of 0.05 in survival rates would require deployment of about 1,000 collars yearly for each age class. Analyses of survival rates assumes that individual animals are known to be either alive or dead during specific time intervals of interest and that their individual life status is independent among animals. This binomial approach is most correct if all animals of interest enter and leave the population simultaneously and there are no animals of unknown status (censored) due to emigration or failure of radio collars (White and Garrott 1990). Typically, animals are trapped and marked over varying time periods and some animals emigrate or radio collars fail. To account for these problems, we will use a staggered entry Kaplan-Meier analysis, which is a modification of the binomial estimator, to estimate survival rates (Pollock et al. 1989, White and Garrott 1990, Bartmann et al. 1992). If no censoring occurs, the Kaplan-Meier approach will provide an estimate identical to the binomial estimator. We believe the binomial approach and attendant assumptions provides a more realistic basis for estimating survival rates than alternative approaches that assume constant survival rates for specified time intervals (Heisy and Fuller 1985). Staggered entry Kaplan-Meier analyses are readily available in programs written for SAS (SAS 1988, White and Garrott 1990, G. C. White, pers. comm.). The PC database program RADIOS (G. C. White pers. comm) will be used to store data on each collared animal. We will test the following generalized hypotheses: Ho: Survival Rate - S Scalves - Sadult females during winter Ho: Smale calves - Sfamale calves H0 : Sadult females H0 : Scalves during winter is equal among years is equal among years We anticipate the following comparative analyses of survival rates using data from animals of known fate. We will conduct pair-wise comparisons using log-rank tests (Pollock et al. 1989, Bartmann et al. 1992) to compare survival functions for: calves and adult females during winter and for male and female calves during winter. Program SURVIV (White and Garrott 1990) will be used to test for differences in survival rates among and between age classes and years (Table 2). For example, to compare survival rates of adult females among V V �95 FREDDY STUDY PLAN years, constraints for program SURVIV would be S1 - S2 , S1 - S3, S1 - S4, and to compare survival of adult females and calves within years, constraints would be S1 - S5, S2 - S61 S3 - S7 , S4 - S8 • Similar constraint pairings would occur for comparisons between male and female calves. These reduced models of survival would be tested against a generalized model that estimates all survival parameters individually, using log-likelihood ratio and goodness-offit tests in program SURVIV. Tests will be significant at PS 0.05. Table 2. SURVIV. Age/Sex Class ~ Generalized structure for analyzing survival rates with program Year Radios Deployed Proportions Lived Died Adult Females 93-94 94-95 95-96 96-97 75 75+ 75+ 75+ S1 S2 S3 S4 (l-S1) (l-S 2 ) (l-S3) (l-S4) Calves 93-94 94-95 95-96 94-95 75 75 75 75 Ss (1-Ss) (1-S6) (l-S7) (1-Sa) s6 S1 Sa -----------------------------------------------------------------------------93-94 37 (l-Sg) Male Calves Female Calves S9 94-95 95-96 94-95 37 37 37 S10 Su S12 (l-S10) (l-S 11 ) (l-S12) 93-94 94-95 95-96 94-95 38 38 38 38 S13 S14 S15 (l-S 13 ) (l-S14) (l-S15) (l-S 16 ) s16 We will assess whether calf survival can be predicted from body weight, hind foot length, or total body length at time of trapping (continuous variable) and sex, year, and trapping zone(s) (categorical variables) using logistic regression. We will also assess whether calf and/or adult survival can be predicted from weather variables such as mean monthly snow depth and temperature using logistic regression (PROC LOGISTIC, PROC CATMOD, SAS 1988; Bartmann et al. 1992). Tests will be significant at PS 0.05. We will use NOAA weather data from Rifle, Collbran, and Bonham Reservoir, Colorado, USDASCS snow depth coarses at Overland and Park Reservoirs on the Grand Mesa, and establish our own snow measurement coarse and weather station on the Garfield Creek State Wildlife Area deer/elk winter range. We will monitor fate of yearling bulls collared as calves immediately before and during hunting seasons to ascertain whether yearling bulls are illegally taken by hunters in areas where these bulls are protected from �96 FREDDY STUDY PLAN harvest by antler-point restrictions. Our selected study area is subject to antler-point restrictions during all hunting seasons but is adjacent to management areas where antler restrictions are not in effect during all seasons. Some yearling bulls may disperse into areas where they will be legal as yearlings, some will undoubtedly reside along the border where combinations of restrictions apply, and some will remain within the antler-point restriction area. We must assume hunters would select against yearling bulls having white radio collars compared to yearling bulls without collars resulting in negatively biased estimates of illegal harvest. However, if several collared bulls are taken illegally, we have evidence that a problem exists and the basis for designing a study specifically to test hypotheses regarding harvests of yearling bulls. If no bulls are illegally taken we cannot conclude that illegal harvest does not exist because the collars may effectively cause hunters to avoid taking marked bulls. u B. Population Size General Approach Counting all animals in a specific area is the intuitive approach to determining population size. Unfortunately, total counts over large areas are prohibitively costly in time and money and generally highly inaccurate because many animals present are not seen and counted (Caughley 1977). Efforts to improve population counts have followed 2 basic paths: designing sampling systems with statistical rigor to obtain precise estimates that often underestimate true population size (Siniff and Skoog 1964, Kufeld et. al. 1980), and developing procedures to correct for missed animals to improve accuracy of estimates (Bartmann et al. 1986, Samuel et al. 1987). Efforts to minimize numbers of animals missed during aerial counts often focused on variables that affect actual counting conditions such as experience of observers, height and speed of aircraft, type of aircraft, search time, weather conditions, and extent of snow background (Erickson and Siniff 1963, Caughley 1974, Gasaway et al. 1986). Although standardizing procedures may reduce variability between sampled counts, animals are stili missed, often in percentages exceeding 30% (Bartmann et al. 1986). Other efforts at improving accuracy attempted to model the probability of detecting animals. This probability function depends on group size, animal behavior, vegetation type and density, and age and sex class of individual (Floyd et al. 1979, Burnham et al. 1980, Samuel and Pollock 1981, Pollock and Kendall 1987, Samuel et al. 1987, Unsworth et al. 1990, Samuel et al. 1992). One solution to correcting for bias therefore lies in calculating probabilities of detection and correcting observed counts. We propose a multiple-step approach to developing a system to estimate elk densities: 1) estimate degree of sighting or counting bias on sample quadrats and develop an equation for predicting sightability using radiocollared elk as a known subpopulation, 2) apply predictive sighting bias correction factors to adjust counts of elk on sampled quadrats used to estimate density of elk over a large area and use mark-resight estimators based on radio-collared elk to provide an additional estimate of elk density from counts of marked and unmarked elk on the same quadrats, and 3) conduct multiple counts of elk on quadrats to test the repeatability of estimates based on the predictive sighting equation and mark-resight estimators. Elk radio-collared to estimate rates of survival will be used in developing methods to estimate population size. V �97 FREDDY STUDY PLAN Fundamentally, we believe a sampling system using quadrats instead of transects will be most appropriate for estimating densities of elk. Although line transect theory accounts for missing animals (negative sighting bias) (White et al. 1989), our experience with line transects to estimate densities of elk in portions of Middle Park, Colorado indicates too few groups of elk will likely be detected to provide reasonable precision, even with intensive repeated flights of transects (Freddy 1991). Additionally, quadrats are easier to fly than transects in rough terrain inhabited by elk. Initially, we will use a 2.59 km.2 quadrat. Developing a Sightability Model Given the premise of using a 2.59 km.2 quadrat as a sampling unit, we will estimate the probability, or degree of bias, of detecting elk on this size of quadrat using radio-collared elk (Samuel et al. 1987, Ackerman 1988). These authors used logistic regression to develop sighting models to predict the probability of detecting elk and mule deer (Odocoileus hemionus). Samuel et al. (1987) found group size and percent vegetation canopy coverage were the most useful variables to predict the probability of detecting elk in forested terrain of northern Idaho (n - 111 trials) while Ackerman (1988) found group size, animal activity, and vegetation type most useful for predicting detection of mule deer in brush-dominated habitats in southeastern Idaho (n 277 trials). Probability of sighting elk has been 0.35-0.95 in dense or .. open vegetation, respectively (Samuel et al. 1987) and 0.6-0.7 in mixed conifersagebrush (Artemisia tridentata) habitats (Bear et al. 1989). The logistic regression for predicting sightability is p.. exp(u) 1 + exp(u) where p - the sighting probability, (exp) - 2.718 1 and u - b 0 + b 1x 1 +b 2x 2 + ... bn'Xn, the multiple linear regression equation of (n) variables affecting sightability. Correction factors for each group of elk detected are obtained by inverting the estimated sighting probability (Samuel et al. 1987). The primary advantage of this sighting bias model approach is .that marked animals are not needed in the elk population once the model has been developed unlike traditional mark-resight models that require marking animals before every survey. The primary assumptions of this approach are: 1) Behavior of elk is not altered by the radio collar, 2) Visual detection of elk is not enhanced by the presence of the radio collar as concluded by Bartmann et al. 1987 1 Samuel et al. 1987, Ackerman 1988, and Bear et al. 1989 1 3) Selected radio-collared elk can be located to within a quadrat, 4) Collared elk can be individually identified either by radio frequency or collar symbol/numbers when detected, 5) Observers attempting to visually detect collared elk do not know the locations of marked elk, 6) The dependent variable, sighting probability, is asymptotic between values of O and 1, �98 FREDDY STUDY PLAN 7) The independent variables, whether numeric or categorical, are measured accurately, 8) Selected elk represent heterogenous habitats typically frequented by elk. V Major hypotheses to be tested are: 1) Logistic regression improves precision of sighting probability compared to a simple binomial sighting probability model, and 2) Predictive sighting models are repeatable among periods of time (years). The entire winter range within the study area will be divided into 2.59 lan2 units having boundaries based on topographic features instead of cadastral surveys. These irregularly shaped units will be approximately equal in size. Sightability tests will be conducted during 2 5-day periods in Januaryearly February 1994 and during 3 5-day periods in December, January, and February 1994-95. Our objective is to complete 80 test trials (16/day) on marked animals (seen or missed) during each 5-day period. We anticipate searching 16 quadrats/day at a rate of 15 min/quadrat resulting in 4-5 hours of helicopter use per day. During the 5 days, we will systematically cover the study area. During the first year, up to 150 different collared elk will be available which should reduce our need to repeat trials on individual elk during any 5-day period. We will conduct tests during the morning and afternoon and avoid mid-day to approximate traditional time periods when elk counts are conducted. Because sighting bias is largely associated with small groups of elk, we will attempt to fly when elk are widely distributed in small groups (Appendix VI). V A primary observer in a Cessna 185 will locate selected radio-collared elk (Gilmer et al. 1981) and assign each elk to a quadrat or quadrats if animals are near a quadrat boundary. Within 2 hours of locating an elk, 2 observers along with the pilot in a Bell-Soloy helicopter will search assigned quadrat(s) and enumerate all groups of elk and identify all radio-collared elk on the quadrat. Radio-collared elk will be identified either by the symbol/number on the collar and/or radio frequency using a receiver-scanner in the helicopter (Appendix I). When observers have completed their search of each quadrat they will check via radio with the primary observer to determine if they found the assigned elk. If the elk was found they will proceed to the next quadrat, but if the elk was missed the helicopter crew will immediately find the marked elk using telemetry and document its location as on or off the assigned quadrat. If more than 1 assigned elk occurs in a group only one animal will be used because observations may not be independent. If an assigned elk is missed and then found off the quadrat, the trial will be voided. To maintain vigilant observers, the helicopter crew will be assigned quadrats where there are no known radio-collared elk to serve as controls. For each group detected on a quadrat, observers will assign classification variables to describe conditions potentially affecting detection of each group whether or not the group contained a marked elk. These independent variables will be group size, activity of animal first detected in the group, vegetation type, percent vegetation canopy density, and V �99 FREDDY STUDY PLAN percent snow cover. These variables must also be noted for those groups containing assigned marked elk that are missed during the initial search of the quadrat. Additional independent variables assigned to all groups observed during a flight will be year, flight, pilot, observer, search time per quadrat, and age/sex of marked animals detected (Appendix V). During the first year we will attempt to use only one experienced observer, navigator, and pilot in the helicopter with the intent of controlling these variables. The helicopter crew will be trained prior to collecting data. We anticipate group size, activity, vegetation type, and percent vegetation canopy cover may significantly influence sighting probability (Samuel et al. 1987 and Ackerman 1988). We also anticipate that percent vegetation canopy cover may be difficult to judge. Samuel et al. (1987) worked primarily in habitats having tall conifers. They considered vertical canopy cover to be any vegetation obstructing the observer's view of an elk group for up to 30 m around the group (Unsworth et al. 1991). Ackerman (1988) correctly noted that animals are seldom viewed from vertical but rather from oblique angles. Confounding this perception of visual obstruction are • deciduous habitats, such as oakbrush (Quercus gambelii) and aspen (Populus tremuloides), that can be seen through during winter and only partially obscure one's view of elk even when canopy coverage may approach 100%. A similar problem involves sagebrush which can effectively hide a bedded animal with little distinguishable vertical canopy. We will follow suggestions by Unsworth et al. (1991) but anticipate this to be a difficult variable to estimate. Using stepwise logistic regression (SAS 1988), independent variables will be evaluated as to their significance in predicting the probability of sighting elk (groups seen or missed) to develop a predictive sightability model. Model fit will be assessed using AIC and log-likelihood values. We anticipate transforming some independent numeric variables .to improve conformity of data to model assumptions. We will compare simple binomial and complex sightability models using computer simulations to assess whether independent variables meaningfully improve precision of the sighting probability estimator and also compare the mean and variance of estimates of elk density derived from corrected and uncorrected counts of elk on quadrats. Sightability models among flight periods will be compared to assess the repeatability of the sighting model when pilot and observers remain constant. If we achieve a sample of 80 trials/flight period, power (1-beta) at alpha - 0.05 to detect a difference between sighting models will be about 0.57 (SE~0.016) (sighting models from Samuel et al. 1987 for sighting functions at 20 and 40% vegetation cover). Power improves to about 0.87 (SE=0.006) when n > 160 trials. These power calculations assume surveys will encounter groups of elk at frequencies of group sizes similar to those observed during preliminary sampling flights (Appendix VI, Table 2). If observed sizes of groups shift towards large groups then power of tests would be less because differences in sighting probabilities between models is most pronounced for small groups (Appendix VI). �100 FREDDY STUDY PLAN Comparing Population Estimators We will estimate total population size within the study area during winter using 3 approaches based on a random quadrat sampling system. This effort will begin in 1994-95 with one flight of quadrats to evaluate logistics, adequacy of sampling, and initial estimates of population size and then continue in 1995-96 and 1996-97 with 3 repeated flights each year if quality of results justifies the effort. We will randomly select about 15% of the potential quadrats and search these quadrats each flight following procedures suggested by Kufeld et al.(1980). Sampling may be stratified according to expected animal density based on distribution of radio-collared elk in 1993-94. For quadrats, we will use 3 estimators to calculate density of elk and all 3 estimators will use the same counts of elk generated from each flight: 1) uncorrected counts of elk on quadrats, 2) counts of elk on quadrats corrected for sighting bias based on the appropriate sighting bias model, and 3) counts of elk on quadrats corrected for sighting bias based on Lincoln-Petersen mark-resight estimator combined for repeated flights using the joint hypergeometric maximum likelihood estimator (JHE) (White and Garrott 1990). The Lincoln-Petersen estimator for each individual flight is: A N1 - -11!1 + 1} Cn 1 + 1) -1 V (m1 + 1) with an estimated variance of A A Var CN1) - ....!n1 + 1) Cn1 + 1) Cn~1lln1-=-m1l (m1 + 1) 2 (m1 + 2) A A A and a nominal 95% confidence interval of N1 ± 1.96 vVar(N1 ) where n 1 is number of radio-collared elk, n 1 is number of elk observed on each aerial survey, and m1 is number of marked elk observed on.each survey. We will compare means and variances of estimates derived from each method. Assumptions for each estimator are: Uncorrected Counts: a) All elk on the quadrats are counted and b) only elk on quadrats are counted (no boundary errors). Assumption a) is likely violated and is the reason for developing a sighting bias model to correct counts. Corrected Counts: c) assumptions are listed in the previous section. We will apply sighting bias corrections to each observed group of elk according to the sighting bias model. Lincoln-Petersen Mark-Resight: d) probability of sighting marked and unmarked elk is the same, e) marked and unmarked elk are correctly classified (marks are discernible), f) marked elk are randomly selected and distributed throughout the population or at least resighting effort is randomly distributed throughout the population, g) each animal has an V �101 FREDDY STUDY PLAN identical but independent probability of being resighted, h) number of marked elk (radio collars) in the population is known at the time of the survey which implies that marks are not lost or unaccounted for, and i) the population is geographically and demographically closed; i.e., there is no immigration or emigration, recruitment or death during the survey in a geographically defined area (Otis et al. 1978, Bear et al. 1989). Assumptions for Lincoln-Petersen mark-resight estimates may be the most difficult to achieve and evaluate. To achieve adequate numbers of marked animals in the population, elk marked each year must accumulate and be used during the years of population surveys. We implicitly assume that detection of white radio collars from year 8 is equal to detection of white collars from yearb when surveys are flown in yearc, otherwise there will be different sighting probabilities for each cohort of marked elk. We also must assume there will be some (1-5%) nonoperative radio collars on live elk which would create an error in knowing how many elk are marked within the area surveyed (fixed-winged flights will be conducted prior to population surveys to determine number of functioning radios in the sampled area using and observer independent of counters). This type of error would cause positive bias in population estimates. Bear et al. (1989) identified marker visibility, random sampling of the population, and heterogeneous sighting probabilities due to group size or vegetation type as problems associated with applying mark-resight estimators to elk populations. We have attempted to address these concerns by using high-contrast white collars not prone to color bias, helicopters to capture elk throughout the distribution of the population, use of random sampling to survey the population, and use of multiple independent aerial resight (recapture) surveys to reduce recapture bias associated with other recapture techniques and to improve precision of estimates (Otis et al. 1978, Bartmann et al. 1987, Bear et al. 1989, Neal et al. 1993). Effects of group size and vegetation type are sources of sighting heterogeneity which can bias the variance of the estimate and may negatively bias the estimate (Bear et al. 1989, Neal et al. 1993). Comparing variances of estimates derived from markresight and the sighting probability model may provide some insight into the severity of heterogenous sighting probabilities. Population estimates based on the JHE mark-resight approach will be calculated using the program NOREMARK (G. White, pers. comm.). We also used NOREMARK to simulate expected quality of population estimates for the JHE estimator given the following constraints. If the real population consists of 2,000 elk, 200 elk are marked with radio collars, about 10% of all elk are observed on each flight ([sighting probability - 0.60) x [proportion of area sampled - 0.15)), and 3 replicate flights are completed, we should achieve a 90% confidence interval about the true population of± 20% with a 1% positively biased estimate of population size. Confidence interval coverage would be 89% (n - 500 Monte Carlo simulations). During the second year of the project when mark-resight surveys will be initiated on quadrats, there could be about 200 marked elk in the population. Precision of JHE estimates would improve to± 15% if 300 marked elk were available or 5 replicate flights were completed. These would be potential design options if marked elk continued to accumulate in the third or fourth year of the project or helicopter hours were �102 FREDDY STUDY PLAN V reallocated among uses or increased. The JHE estimator does not require that marked elk are individually identified, only that marked and unmarked elk are discerned. If we can consistently identify individual elk by collar numbers or radio frequency, the Minta-Mangel mark-resight model can be used to estimate population size and associated variance (Minta and Mangel 1989) using program N0REMARX. Variance of this estimator can be calculated as described by Minta-Mangel (1989) or in a new approach developed by Bowden (1993, unpubl. ms.). Complete capture histories on each marked elk would also allow an assessment of bias associated with the estimate of N due to unequal individual capture probabilities. Unequal capture probabilities can be caused by individual animal behavior, time intervals between initial capture and resight surveys, and heterogeneity of capture probabilities due to factors of age, sex, social dominance, or habitat. Program CAPTURE will be usedAto evaluate the most appropriate capture probability model to estimate N (Rexstad and Burnham 1991). C. Distribution and Movements of Elk General Approach All visual sightings of live and dead collared elk will be assigned UTM coordinates as determined from USGS topographic maps. Visual sightings of collared elk during winter will occur during ground surveys to determine life/death status of elk and during aerial surveys to estimate population density. These locations will provide a general perspective of the distribution of elk on their winter range. V Seasonal patterns of movement will be obtained primarily from aerial surveys. Accurate seasonal locations on all collared elk would be cost prohibitive and not commensurate with the main objectives of the project. Therefore we plan to randomly select 20% of the calves (8 M, 8 F) and 20% of the adult females (15) and monitor their locations periodically to reveal patterns of seasonal movement. Yearling males marked as calves will be located prior to hunting seasons to monitor their fates in relation to antler-point restrictions. Distribution and movement data will be plotted following suggestions of White and Garrott (1990). V. SCHEDULE Activity Approximate Date Phase !--Complete detailed study plan, July 1992-June 1993 select study location, refurbish elk traps, purchase radio collars for calves and design expandable collars for calves. Phase !!--Complete initial purchase of July 1993-June 1994 radio collars; trap and radio-collar Trapping 12/93 elk to estimate survival; begin surveys Sighting bias to design sighting bias estimator. 1 & 2/94 V �103 FREDDY STUDY PLAN July 1994-June 1995 Phase III--Continue radio-collaring Trapping 12/94 elk for estimating survival, Sighting bias 12/94, continue testing sighting bias estimator, estimate population 1/95, 2/95; Pop. Est. 1 & 2/95 size using quadrats and mark-resight. Phase IV-- Estimate pop. size July 1995-June 1997 size using sighting bias correction, Trapping 12/95 & quadrats, and mark-resight. Continue 12/96; Pop. Est. trapping to estimate survival. 1/96 &1/97 VI. PERSONNEL PROGRAM RESPONSIBILITIES David J. Freddy: Principal Investigator responsible for final project design, organizing field personnel, obtaining and organizing data, financial control, coordinating publications. R. Bruce Gill: Provide administrative support, input for study design, and liaison with other administrative sections within the Division of Wildlife or other natural resource agencies. \,_,_,/ John H. Ellenberger and James H. Olterman: Provide input for study design and location, coordinate study with Regional activities, and logistical support. David C. Bowden and Gary C. White: Provide input for study design and statistical protocol, conduct data analyses, and provide software support. �104 FREDDY STUDY PLAN VII. V BUDGET-SUMMARY ACTIVITY FY92-93 Personal Operating Travel Capital Total 55,773 26,800 1,200 4,700 88,473 W/ 4% inflation BUDGET-ITEMIZED FY94-95 FY95-96 FY96-97 70,706 72,236 70,706 72,236 66,370 93,045 66,020 72,270 4,940 4,940 4,680 5,590 0 5,700 0 0 141,666 154,886 142,016 170.871 161,081 FY92-93 PERSONAL SERVICES $49,841 A. D.J. FREDDY 12 MOS $2,509 8. MAMMALS SECR. 1 MOS $0 C. UTILITY I 6 MOS OCT-MAR $2,723 D. UTILITY 1, TRAPPING E. WORK STUDY STUDENT, 1 YEAR $700 $0 F. VOLUNTEER FOR 20 DAYS $55,773 SUB-TOTAL OPERATING G. VEHICLES MILEAGE, 4X4, 3 $2,200 H. VEHICLES, SNOWMOBILES, ATV $0 I. NEW RADIO COLLARS $19,500 J. REFURBISHED RADIO COLLARS $0 $3,000 J. TRAPPING, BAIT, SUPPLIES $1,500 K. MISC. SUPPLIES, SERVICES L. FIXEDWINGED, MORT. SURV. $600 M. FIXEDWINGED, SIGHT. SURV. $0 N. HELICOPTER, SIGHT. SURVEYS $0 0. HELICOPTER, QUAD. SURVEYS $0 P. HELICOPTER, NONRANDOM SURV. $0 Q. HELICOPTER TRAPPING $0 SUB-TOTAL $26,800 TRAVEL R. DJF 10 DAYS, STANDARD $650 S. DJF DAYS TRAPPING $550 $0 T. VOLUNTEER, TRAPPING U. REGION PERSONS, TRAPPING $0 $0 V. REGION PERSONS, SURVEYS $1,200 SUB-TOTAL CAPITAL EXPENSES $3,200 W. RADIO TELEMETRY RECEIVER X. PACK-SETS/LAPTOP COMPUTER $1,500 $4,700 SUB-TOTAL PROJECT YEARLY TOTAL WITH 4X INFLATION FY93-94 $88,473 177,706 147,333 FY93-94 FY94-95 FY95-96 $56,796 $56,796 $2,500 $9,180 $3,060 $700 $0 $72,236 $56,796 $56,796 $2,500 $9,180 $3,060 $700 $0 $72,236 $2,500 $9,180 $1,530 $700 $0 $70,706 $2,500 $9,180 $1,530 $700 $0 $70,706 $5,720 $350 $8,600 $0 $2,500 $2,500 $3,150 $2,700 $21,750 $0 $0 $25,000 $72,270 $5,720 $350 $15,000 $4,050 $2,500 $2,500 $3,150 $4,050 $32,625 $4,350 $0 $18,750 $93,045 $5,720 $5,720 $350 $350 $6,000 $5,000 $6,750 $5,400 $2,500 $2,500 $2,500 $2,500 $3,150 $3,150 $1,800 $1,800 $0 $0 $21,750 $21,750 $4,350 $4,350 $12,500 $12,500 $66,020 ,$66,370 $650 $1,365 $0 $1,365 $1,300 $4,680 $650 $1,365 $0 $1,365 . $2,210 $5,590 $650 $1,365 $0 $1,365 $1,560 $4,940 $650 $1,365 $0 $1,3~5 $1,560 $4,940 $3,700 $2,000 $5,700 $0 $0 $0 $0 $0 $0 $0 $0 $0 147,697 FY96-97 $154,886 $170,871 $141,666 $142,016 $161,081 $177,706 $147,333 $147,697 """**************************************"""1Tt*"""1Tt*****"""************************************* 0 REGION TRAPPING MDAYS 65 REGION PILOT, HRS 20 REGION BIOL.,FLYING MDAYS REGION PICKUP(S)+TRAP TRAILER(S) TRAP A/N REGION SKIDOO ALPINE AND ATV 3 REGION PROP. TECH. MDAYS 21 65 20 TRAP A/N 3 21 80 34 TRAP A/N 3 21 55 24 TRAP A/N 3 21 55 24 TRAP A/N 3 -- V .. �105 FREDDY STUDY PLAN VIII. LOCATION We selected the eastern portion of GMU 42 within the Grand Mesa DAU (E14) for conducting this project (Appendix VII). This area encompasses about 1,028 km2 (397 mi 2 ) in the Divide Creek drainages located south of the towns of Newcastle and Rifle in west-central Colorado. We estimate that 1500-3000 elk inhabit about 642 km2 (248 mi 2 ) of winter range within this area. This area meets the following necessary criteria: 1) the elk population is relatively "closed" during the winter, that is, subject to minimal emigration or immigration, and assumes a somewhat static distribution during winter; 2) the winter range generally has reliable snow cover to facilitate detecting elk during population surveys and mixed vegetation types consisting of oakbrush, aspen, sagebrush, juniper-pinyon woodland (Juniperus osteosperma-Pinus edulis), mixed conifers (Picea sp., Abies sp.) and agricultural fields typical of many elk winter ranges in western Colorado; 3) there is dependable access for research purposes to both private and public lands and dependable public access to public lands for hunting; 4) there is minimal large-scale land-use conflict problems such as agricultural damage; 5) there is local, area, and regional CDOW support for conducting the project in this area; and 6) the airport at Rifle will provide readily accessible support services. IX. LITERATURE CITED Ackerman, B. B. 1988. Visability bias of mule deer aerial census procedures in southeast Idaho. Phd. Thesis. University of Idaho, Moscow. 106pp. Bartholow, J. 1992. Pop-II system documentation. Fossil Creek Software, Fort Collins, CO. 50pp. Bartmann, R. M. 1984. Estimating mule deer winter mortality in Colorado. J. Wildl. Manage. 48:262-267. _ _ , Carpenter, L. H., R. A. Garrott, and D. C. Bowden. 1986. Accuracy of helicopter counts of mule deer in pinyon-juniper woodland. Wildl. Soc. Bull. 14:356-363. ___ , and G. C. White. 1991. Compensatory effects of harvest in a mule deer population. Colo. Div. Wildl. Game Res. Rep. July:28-40. ___ , , L. H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildl. Monogr. 121. 39pp. ___ , , ___ , and R. A. Garrott. 1987. Aerial mark-recapture estimates of confined mule deer in pinyon-juniper woodland. J. Wildl. Manage. 51:41-46. Bear, G.D. 1986. Expanding telemetry collar for elk calves. Colo. Div. Wildl. Game Info. Leaflet 112. 2pp. 1989. Seasonal distribution and population characteristics of elk in Estes Valley, Colorado. Colo. Div. Wildl. Spec. Rep. 65. 14pp. ___ , G. C. White, L. H. Carpenter., R. B. Gill, and D. J. Essex. 1989. Evaluation of aerial mark-resighting estimates of elk populations. J. Wildl. Manage. 53:908-915. Bowden, D. C. A simple technique for estimating population size. Unpubl. ms. 17pp. Boyce, M. S. 1989. The Jackson elk herd, intensive wildlife management in North America. Cambridge Univ. Press, Cambridge. 306pp. �106 FREDDY STUDY PLAN Burger L. W., Jr., M. R. Ryan, D. P. Jones, and A. P Wywialowski. 1991. Radio transmitters bias estimation of movements and survival. J. Wildl. Manage. 55:693-697. Burnham, K. P., D.R. Anderson, and J. L. Laake. 1980. Estimation of density from line transect sampling of biological populations. Wildl. Mono. 72. 202pp. Caughley, G. 1974. Bias in aerial survey. J. Wildl. Manage. 38:921-933. 1977. Analysis of vertebrate populations. John Wiley & Sons, New York. 232pp. Colorado Division of Wildlife. 1990. Colorado big game harvest. Colo. Div. Wildl., Denver. 173pp. _ _ . 1991. Colorado big game harvest. Colo. Div. Wildl., Denver. 172pp. _ _ . 1991b. Long Range Plan (revised). Colo. Div. Wildl., Denver. 16pp. de Vergie, W. J. 1989. Elk movements, dispersal, and winter range carrying capacity in the upper Eagle River valley, Colorado. MSc. Thesis, Colorado St. Univ., Fort Collins. 124pp. Erickson, A. W., and D. B. Siniff. 1963. A statistical evaluation of factors influencing aerial survey results on brown bears. Trans. N. Am. Wildl. Conf. 28:391-409. Foster, C. C., E. D. Forsman, E. C. Meslow, G. S. Miller, J. A. Reid, F. F. Wanger, A. B. Carey, and J.B. Lint. 1992. Survival and reproduction of radio-marked adult spotted owls. J. Wildl. Manage. 56:91-95. Floyd, T. J., L. D. Mech, and M. E. Nelson. 1979. An improved method of censusing deer in deciduous-coniferous forests. J. Wildl. Manage. 43: 258-261. Freddy, D. J. 1987. The White River elk herd: a perspective, 1960-85. Colo. Div. Wildl. Tech. Publ. 37. 64pp. 1991. Elk census methodology. Colo. Div. Wildl. Game Res. Rep. July: 59-72. _ _ , D. L. Baker, R. M. Bartmann, and R. C. Kufeld 1993. Deer and elk management analysis guide, 1992-1994. Colo. Div. Wildl., Div. Rep. 17. 77pp. (In press). Garrott, R. A., and G. C. White. 1982. Age and sex selectivity in trapping mule deer. J. Wildl. Manage. 46:1083-1086. Garrott, R. A., R. M. Bartmann, and G. C. White. 1985. Comparison of radiotransmitter packages relative to deer fawn mortality. J. Wildl. Manage. 49:758-759. Gasaway, W. C., S. D. DuBois, D. J. Reed, and S. J. Harbo. 1986. Estimating moose population parameters from aerial surveys. Institute of Arctic Biol., Biol. Papers Univ. of Alaska No. 22. 108pp. . Gilmer, D. S., L. M. Cowardin, R. L. Duval, L. M. Mechlin, C. W. Shaiffer, and V. B. Kuechle. 1981. Procedures for the use of aircraft in wildlife biotelemetry studies. USDI Fish & Wildl. Serv. Resource Publ. 140. 19pp. Heisey, D. M., and T. K. Fuller. 1985. Evaluation of survival and causespecific mortality rates using telemetry data. J. Wildl. Manage. 49:668-674. Houston, D. B. 1982. The northern Yellowstone elk, ecology and management. Macmillan Publ. Co., Inc. New York. 474pp. Kufeld, R. C., J. H. 0lterman, and D. C. Bowden. 1980. A helicopter quadrat census for mule deer on Uncompahgre Plateau, Colorado. J. Wildl. Manage. 44:632-639. V ., �107 FREDDY STUDY PLAN Laake, J. L. 1992. Catch-per-unit-effort models: an application to an elk population in Colorado. Pages 44-55 in D.R. McCullough and R.H. Barrett, eds., Wildlife 2001: Populations. Elsevier Publ., London. Leptich, D. J., and P. Zager. 1991. Road access management effects on elk mortality and population dynamics. Pages 126-131 in A.G. Christensen, L. J. Lyon, and T. N. Lonner, comps., Proc. Elk Vulnerability Symp. Montana St. Univ. , Bozeman. Minta, S. and M. Mangel. 1989. A simple population estimate based on simulation for capture-recapture and capture resight data. Ecology 70:1738-1751. Neal, A. K., G. C. White, R. B. Gill, and D. F. Reed. 1993. Evaluation of mark-resight methods to estimate mountain sheep numbers. J. Wildl. Manage. (In press). Nelson, L. J., and J.M. Peek. 1982. Effect of survival and fecundity on rate of increase of elk. J. Wildl. Manage. 46:535-540. Pemberton, J.M., S. D. Albon, F. E. Guinness, T. H. Glutton-Brock, and R. J. Berry. 1988. Genetic variation and juvenile survival in red deer. Evolution 42:921-934. Pollock, K. H., and W. L. Kendall. 1987. Visibility bias in aerial surveys: A review of estimation procedures. J. Wildl. Manage. 51:502-510. ___ , S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. J. Wildl. Manage. 53:7-15. Quimby, D. C., and J. E. Gaab. 1957. Mandibular dentition as an age indicator in Rocky Mountain elk. J. Wildl. Manage. 21:134-153. Rexstad, E., and K. Burnham. 1991. User's guide for interactive program CAPTURE. Coop. Fish & Wildl. Res. Unit, Colorado St. Univ., Fort Collins. 29pp. Samuel, M. D. and K. H. Pollock. 1981. Correction of visibility bias in aerial surveys where animals occur in groups. J. Wildl. Manage. 45:993997. _ _ , E. O. Garton, M. W. Schlegel, and R. G. Carson. 1987. Visibility bias during aerial surveys of elk in northcentral Idaho. J. Wildl. Manage. 51:622-630. ___ , R. K. Steinhorst, E. 0. Garton, and J. W. Unsworth. 1992. Estimation of wildlife population ratios incorporating survey ~esign and visibility bias. J. Wildl. Manage. 56:718-725. SAS Institute, Inc. 1988. SAS Language guide for personal computers. SAS Institute Inc., Cary, NC. 1028pp. Sauer, J. R., and M. S. Boyce. 1983. Density dependence and survival of elk in northwestern Wyoming. J. Wildl. Manage. 47:31-37. Schlegel, M. W. 1977. Factors affecting calf elk survival on Coolwater Ridge in north central Idaho. Pages 35-37 in Proc. Western States Elk Workshop, Estes Park, Colo. Siniff, D. B., and R. 0. Skoog. 1964. Aerial censusing of caribou using stratified random sampling. J. Wildl. Manage. 28:391-401. Unsworth, J. W., L. Kuck, and E. 0. Garton. 1990. Elk sightability model validation at the National Bison Range, Montana. Wildl. Soc. Bull. 18:113-115. ___ , F. A. Leban, G. A. Sargeant, E. 0. Garton, M.A. Hurley, J. R. Pope, and D. J. Leptich. 1991. Aerial survey: user's manual with practical tips for designing and conducting aerial big game surveys. Idaho Dep. �108 FREDDY STUDY PLAN Game & Fish, Boise. 54pp. Wade, D. A., and J. E. Browns. 1982. Procedures for evaluating predation on livestock and wildlife. Texas Agric. Expt. Sta. Publ. B-1429. 42pp. u White, G. C. 1983. Numerical estimation of survival rates from band recovery and biotelemetry data. J. Wildl. Manage. 47:716-728. _ _ . 1992. DEAMAN database manager and population modeling procedures; Colorado Division of Wildlife User's manual and reference. Colorado St. Univ., Fort Collins. 109pp. ___ ,D.R. Anderson, K. P. Burnham, and D. L. Otis. 1982. Capturerecapture and removal methods for sampling closed populations. Los Alamos Natl. Lab. IA-8787-NERP. 235pp. ___ , R. M. Bartmann, L. H. Carpenter, and R. A. Garrott. 1989. Evaluation of aerial line transects for estimting mule deer densities. J. Wildl. Manage. 53:625-635. ___ , and R. A. Garrott. 1990. Analysis of wildlife radio-tracking data. Academic Press, Inc., San Diego. 383pp. ___ , R. M. Bartmann, L. H. Carpenter, and A. W. Alldrege. 1987. Survival of mule deer in northwest Colorado. J. Wildl. Manage. 51:852859. Zager, P., and D. J. Leptich. 1991. Comparing two methods of calculating elk survival rates. Pages 106-109 in A.G. Christensen, L. J. Lyon, and T. N. Lonner, comps., Proc. Elk Vulnerability Symp. Montana St. Univ., Bozeman. APPENDIX I Specifications for Radio Collars Pulse Rate Normal: 60-65 ppm Pulse Rate Mortality: 120-130 ppm Motion Sensor Delay: 4-6 hrs Batteries: 4+ year life, 2 lithium D-cells adults, 1 lithium D-cell calves Antenna: External whip, pvc coated Collar Material: 7.6 cm (3") wide white colored smooth surfaced conveyor belting, 2 layers sewn together, 0.64 cm (1/4") total thickness Additional Material: Bright white with black core Ritchie All-Flex rubber material for identifcation symbol/number placed as a sleeve over top portion of collar Collar Size: 61-81 cm (24-32") adult females, individually fitted 61-81 cm (24-32") expandable for female calves 61-97 cm (24-38") expandable for male calves Collar Weight: About 1.1 kg V �109 FREDDY STUDY PLAN APPENDIX II Animal Welfare Concerns There are 3 potential issues regarding animal welfare: use of ear-tags, use of radio collars, and capture techniques. Ear-tags: To permanently mark radio-collared animals and some non-collared animals, an aluminum, numbered and colored ear-tag will be placed in each ear. Tags are self-piercing using special pliers and have been used for many years on elk. To reduce potential problems of tag loss, tags will be placed on front edge of the ear about 1/3 to 1/2-way above the base of the ear and the edge of the tag will be flush with the edge of the ear. Radio collars: Fitted size of radio collars is based on measurements from hunter-killed elk and captive research elk and testing of actual collars on captive elk. The basic expandable design for calves has been extensively field tested on deer and moose. Capture Techniques: We will capture elk in traditional corral traps capable of holding several elk at one time and by net-gunning individual elk from a helicopter. We expect a 1-2% serious injury/mortality rate with either technique. Animals having a broken leg, neck, or pelvis will be euthanized with a gunshot to the head following euthanasia procedures in our Animal Welfare guide. All persons involved in capture will be trained to properly euthanize appropriate animals. Capture techniques will be constantly monitored and changed if necessary to insure that minor injuries to animals do not chronically occur. Animals captured with a helicopter will be hobbled and blindfolded and slung under the helicopter to a nearby field processing point where they will be collared, weighed, etc. and then released at that site. We aniticapte 1-5 minutes to capture the elk, 1-2 minutes to ferry the elk to .processing point, and 3-4 minutes to process the elk before release. This technique has been successfully used to capture and radio-collar elk in Arizona, Colorado, Idaho, Montana, Oregon, and Washington. Because of short handling times, mortalities due to capture have been rare. Animals captured in corral traps will be blindfolded as they are processed in the chutes. Calves may be hobbled or walked into a holding box for weighing. We prefer to trigger traps manually using a solenoid system so we can be selective about animals captured and reduce the time animals are in traps. However, traps may be triggered by animals entering and in such cases animals may be in the trap overnight. Food will be provided via alfalfa used as bait. We will use the minimum number of persons to process elk to reduce capture stress. �110 FREDDY STUDY PLAN . APPENDIX III Visual Identification System for Radio Collars Numbers, symbols, and letters will be used in ordered combinations to identify individual elk primarily during mark-resight aerial surveys. No more than 3 characters will be used to identify an individual and there will always be a 2-digit number. Numbers Used ( 8) : •0 , 1, 2 , 3 , 4 , 5 , 6 , and 7 Symbols Used (5): ., ■, ♦, ◄, + Letters Used (11): A, C, F, H, K, N, P, T, V, X, y 10, 20, 11, 21, 12, 22, 13, 23, 14, 24, 15, 25, 16, 26, 17, 27, Number Matrix 30, 40, 50, 60, 31, 41, 51, 61, 32, 42, 52, 62, 33, 43, 53, 63, 34, 44, 54, 64, 35, 45, 55, 65, 36, 46, 56, 66, 37, 47, 57, 67, 70 71 72 73 74 75 76 77 Each Symbol with each number represents 56 identification codes; all symbols with each number equals 280 codes or animals. Each letter with each number represents 56 codes; all letters with each number equals 616 codes or animals. Therefore, 896 different animals can be individually marked using this system. V �.L.L.L FREDDY STUDY PLAN APPENDIX IV ,. Precision and Power of Survival Estimates and Tests 30 ii 25 S•0.50 ! .. I ... 120 0 115 ~ !, 10 S •0.80 .. S •0.80 • S • 0.70 ■ • * 5 50 100 75 RADlO COLLARS DEPLOYED 1 =. 0 I I DETECT DIFFERENCES IN SURVIVAL OF: DIFF•O.25 DIFF•0.2 DIFF•0.1 DIFF•0.05 o.a ... --------······· 0.8 @ .. -.... ffi 0.4 ··········•••••••••••• ~ I ..----· --- ---------······ ................................... ...................... ···-·••" ..... 0.2 a. 0 20 40 60 75 80 100 RADIOCOUARS DEPLOYED 1.2 ! I I @ DETECT DIFFERENCES IN SURVIVAL OF: DIFF•O.25 DIFF•0.2 DIFF•0.1 DIFF•0.05 1 ---------~------------------------------ o.a ············••••••••••••••• 0.8 Ii ~ ~ ·~ I - .-··············· ................................................. ... 0.4 0.2 0 20 40 60 75 RADIO COLLARS DEPLOYED 80 100 �112 FREDDY STUDY PLAN V APPENDIX V Classification variables for elk groups and flight conditions for sightability trials, Group Size 1 2 3 Vegatation Type Activity Level Oakbrush-dense/continuous Oakbrush-scattered/clumped Aspen Sagebrush Pinyon-Juniper Tall Conifer Agri. Fields/Clearings Riparian Bedded Standing Moving 4 5 6 7 8 9 10-14 15-19 20-24 ~25 % Vegetation Canopy Vertical Cover 0-15 16-25 26-35 36-45 46-55 56-65 66-75 76-85 86-100 %Snow Cover Snow Type 0 Fresh New Snow Old snow with old tracks 1-25 26-50 51-75 76-99 100 Lighting Conditions Age/Sex Marked Elk Bright Sun/High Contrast Male Calf Female Calf Hazy Sun Adult Female Overcast, dull light Yearling Male Adult Male Search Rate Year/Period Pilot Observer 1993-94-I 1993-94-II 1994-95-I 1994-95-II 1994-95-III 1993 1993 1994 1994 km2 / min �113 FREDDY STUDY PLAN APPENDIX VI Power Considerations for Sightability Models Dr. David C. Bowden Consider the logistic model (1) below which describes the probability of observing elk groups for a given set of observational conditions. Assume that the same model also describes the probability of observing elk groups for a different set of conditions but with a different value for the parameter Po• Suppose data is obtained under both sets of conditions. This section describes the results of a simulation study which investigates the power of a test of the hypothesis that the parameter value for Po is the same for both sets of conditions. In the simulation study, observations were generated as if two sets of conditions applied. Then a test of equality of the 2 Po values was performed. This process was repeated 1000 times. The proportion of times that the null hypothesis was rejected estimates the power of the given test. Following Samuel et al. (1987), it is assumed that the probability of sighting elk groups in uniform habitat relative to percent vegetation cover can be adequately modeled as P = 1/ (1 + exp (- ((3 0 +P 1 ln (g)) ) (1) where pis the probability of sighting a group of size gin the specified habitat, Po and P1 are parameters and ln(g) is the natural logarithm of g. Those authors estimated the parameter values to be .22 and 1.55, respectively for habitat locations with 20 % vegetation cover and -0.78 and 1.55 with 40% vegetation cover. These sets of estimates were used to specify the parameter values for the simulation study. Table 1 gives values of p for the 2 sets of parameter values and a few choices of group sizes. Thus the problem of interest is, "What is the chance that a change in sighting probabilities as given in Table 1 will be detected if the conditions of the simulation study apply?" �114 FREDDY STUDY PLAN V Table 1. Probability of sighting given 2 sets of parameter values at different group sizes (g). Sample 1 uses Po = . 22 and P1 = 1, 55 in function (1) and Sample 2 uses Po m - • 78 and p1 = 1. 55. Group size(g) PROBABILITY SAMPLE 1 SAMPLE 2 1 .555 .314 2 .785 .573 3 .872 .716 4 .914 .797 5 .938 .847 6 .952 .881 10 .978 .942 15 .988 .968 20 .992 .979 30 .996 .989 Generation of observations for a single iteration in the simulation study is described next. A set of n elk groups were used to generate 2 samples of elk group observational data. Sample 1 was generated by taking an observation from a Bernoulli random variable for each group where the probability of success is given by probability function (1) with Po = .22 and p1 = 1. 55 , that is, a random uniform number on the interval [O, 1), say u, was selected, if u~p then the group was said to be sighted otherwise it was not sighted. The second sample of n observations was independently generated with the same group sizes using parameter values Po c - • 7 8 and p1 = 1. 55 and function (1) to obtain the needed probability of success. ·The 2 samples of sighting data were pooled and then analyzed using the SAS logistic regression procedure. The following model was fit to the pooled data set: p = 1 / ( 1 + exp ( - ( PO + P1 ln ( g > + P2 z ) ) ) (2) where z is an indicator variable which is equal to O if the observation came from sample 1 and is equal to 1 if the observation came from sample 2. The question of interest is, "What is the power of a «-.05 test of the hypothesis that p2 O versus an alternative hypothesis 11 2 .. O when actually p2 = -1 ?" This null hypothesis is equivalent to stating that the 2 samples were generated from the same logistic function. Of course the power of the test depends on the distribution of group sizes and the number of groups in each sample. It should also be clear from Table 1 that groups of sizes 1-10 are needed to obtain reasonable power unless very large numbers of groups are observed in both samples. Group sizes used in the study are those obtained from a sample survey i.: V �115 FREDDY STUDY PLAN of a winter elk population. This sample survey was a stratified random quadrat count in game management unit 23, 24 on January 18-19, 1989 by C. Reichert, J.Ellenberger, and D. Freddy. During the sample survey, 84 elk groups were observed and group size recorded (Table 2). Table 2. Frequency distribution of the 84 elk group sizes. Group size 1 2 3 4 5 6 7 8 9 Frequency 9 12 10 9 10 5 4 4 6 Group size 10 11 12 13 16 18 19 20 39 Frequency 1 3 2 3 1 1 1 2 1 Two independent simulation studies were performed using the frequency distribution ·of group sizes from Table 2 in the data generation step. Each study consisted in 1000 replications of pairs of samples generated as described above. In the first study, for each of the 84 listed group sizes, a determination of sighting or not was made, independently, using the 2 probability sighting functions. In the second study the frequency of groups of a given size·in Table 2 was doubled to give 2 samples of size 168, replicated 1000 times. For each replication the 2 samples were pooled and model (2) fitted. Then the null hypothesis that 11 2 = O was tested. The number of rejections of this null hypothesis is shown for the 1000 replications in Table 3 as 10 sets of 100 tests each. The standard error of the estimated power was calculated using variation among the 10 sets. Thus given 2 independent samples of observations each based on 84 groups with frequency distribution listed in Table 2, a 5% significance level test has an estimated power of . 57 ( se-.016) of rejecting H0 : 11 2 = O when P2 = -1. The estimated power when the number of groups is doubled (n-168 each) is estimated to be .873 ( se-.006). �116 FREDDY STUDY PLAN V Table 3. Proportion of rejections of the null hypothesis H0 : p2 = O vs HA: P2 ~ O when P2 = -1 , for 10 sets of 100 replications each, Cl = . 05. Set Sample I Sample II 1 .58 .89 2 .59 .89 3 .62 .88 4 .66 .86 5 .57 .91 6 .58 .88 7 .52 .85 8 .51 .88 9 .57 .86 10 .50 .90 Mean .57 .88 SE .016 .006 u Reference Samuel, M., E. Garton, M. Schlegel, and R. Carson. 1987. Visibility bias during aerial surveys of elk in northcentral Idaho. J. Wildl. Manage. 51: 622-630. V �.&..&., FREDDY STUDY PLAN APPENDIX VII Location of the study area in the eastern portion of Game Management Unit 42 within the Grand Mesa Data Analysis Unit E-14. 50 MILES NEWCASTLE GRAND MESA DAU E-14 UNIT B DARY GRAND JUNCTION �27 Colorado Division of Wildlife Wildlife Research Report July 1994 ~~ JOB PROGRESS REPORT State of Colorado Project No. W-153-R-7 Mammals Research Work Plan No. 3 Elk Investigations Job No. 9 Estimating Survival Rates of Elk and Developing Techniques to Estimate Population Size Period Covered: Author: July 1, 1993 - June 30, 1994 D. J. Freddy Personnel: F. Barnes, J. Broderick, G. Byrne, D. Crane, J. Ellenberger, J. Frothingham, V. Graham, J . Gray, R. Hays, c. Kolus, G. Loucks, J. Pelletier, S. Rivera, P. Will, CDOW; D. Bowden, C. Vardeman, G. White, CSU; M. Miller, C. McCarty, R. Witt, Volunteers; USFS Rifle, BLM Glenwood Springs, Cooperating. Abstract We radio-collared 73 calf elk (Cervus elaphus nelsoni) (6 mos old) and 68 adult female elk(~ 1 year old) in December 1993 to estimate survival rates during winter and used these same elk in 110 aerial sighting bias trials to develop models for estimating degree of negative sighting bias when counting elk with a helicopter on sample quadrats. Elk were captured using portable corral traps and helicopter net-gunning. Survival rates(± 95% CI) from December 1993 to June 1994 were 0 . 918 + 0.063 for calves and 0.956 + 0.049 for adult females. Suspected causes of death were primarily malnutrition and predation for calves and shooting and predation for adult females. Calves suspected of dying from malnutrition had marrow fat values< 13% at death and body weights< 102 kg at capture . Blood tests for brucellosis were negative in all 43 adult females sampled while progesterone assays indicated 74-80% of the adult females were pregnant. Observers did not count 15.4% of the elk during sighting trials resulting in a simple binomial correction factor of 0.846 ± 0.071 (± 95% CI). Univariate tests, however, indicated elk age, log(ln)initial group size, log(ln) total group size, elk behavior, vegetation type, percent occlusion cover, percent snow cover, and wind conditions affected sightability of elk (P ~ 0.05). Step-down regression analyses produced a complex 12 parameter sighting model inclusive of elk age, elk behavior, vegetation type, log total group size, and percent snow cover. Simpler models inclusive of log total group size or log total group size and percent occlusion cover may be more utilitarian and provide acceptably precise correction factors. The simple binomial correction model provided unacceptably inflated and imprecise correction factors which supported incorporating the variable total group size into sighting bias models. �29 JOB PROGRESS REPORT ESTIMATING SURVIVAL RATES OF ELK AND DEVELOPING TECHNIQUES TO ESTIMATE POPULATION SIZE David J. Freddy P.H. OBJECTIVE Estimate survival rates of adult female and calf elk and develop techniques to estimate population size. SEGMEw.l? OBJECTIVES 1. Radio-collar 75 adult female and 75 calf elk during December 1993 in Game Management Unit 42 south of Rifle, Colorado. 2. Estimate winter and annual survival rates of calf and adult female elk from known fates of radioed elk. 3. Estimate probability of sighting and counting elk during helicopter surveys using radioed elk as a known population. 4. Analyze survival and sightability data and summarize annually in Federal Aid Job Progress reports IRTRODUCTIOH Elk (Cervus elaphus nelsoni) are a high-profile and burgeoning wildlife resource throughout much of Colorado. This burgeoning resource has many benefits but frequent social, political, and economic conflicts suggest elk, to some degree, have reached a "social" carrying capacity. Many conflicts arise, in part, from an inadequate ability to predict the dynamics of elk populations. Key population parameters such as survival rates of calves and adult females and total population size are seldom measured resulting in computer models of elk population dynamics that are based on minimal data. Such models can be of little predictive value and, as such, diminish confidence in using models to guide management of elk populations. our objectives are to provide reliable estimates of survival rates for calves and adult females during winter and for adult females throughout the year for the period 1993-94 through 1997-98. Additionally, we will develop and test a system for estimating population size that will incorporate estimates of sighting bias in conjunction with a random sampling system using search quadrats as sample units. our winter study area encompasses about 839 Jan2 (324 mi2) in the eastern half of Game Management Unit 42 south and east of Rifle, Colorado~ Elk winter range vegetation types include juniper-pinyon woodland (Juniperus ost;eosperma-Pinus edulis), oakbrush-mountain shrub (Quercus gambelii-Amelanchier alnirolia), aspen (Populus t;remuloides), sagebrush µrt;emisia t;rident;at;a), and agricultural fields ·(Freddy 1993). �30 METHODS Marking V We placed radio collars (l72-176MHz) having mortality sensors on 73 calves (6 mos old), of which 36 were males and 37 were females, and 68 adult females(~ l year old). Of these, 54 calves and 46 adults were trapped from 2-6 December 1993 using helicopter net-gunning and 19 calves and 22 adults were trapped from 10-20 December using portable corral traps. Helicopter capture occurred at 17 remote sites located primarily on public lands while corral-trapping occurred at 3 sites primarily on private lands. Trapping effort was allocated among 8 geographic trap zones to assure that radioed elk were representative of most if not all segments of the population (Table 1). Number of elk to be captured in each zone was established prior to trapping based on distribution and relative numbers of elk observed during sex and age ratio classification flights completed annually in January from 1989 to 1993. Collars for adults were fixed in c·ircumference while those for both male and female calves were expandable to accommodate growth when animals became adults. Additionally, all radio collars had a 2 character visual identification code on the dorsal surface of the collar consisting of letters, numbers, or-symbols. Collars were white with black symbols. Calves captured by net-gunning were ferried by helicopter to processing points usually within 1.6 km of capture sites. At processing points, body weight, total body length, and hind foot length were measured and calves were then radio-collared and released. Similar measurements were also made on calves that were corral-trapped and then released at the trap site. Body measurements for calves were compared between sexes using Proc FREQ and Proc GLM (SAS 1988). Physiological Assays Serum samples were obtained from adult females captured by net-gunning for brucellosis tests (USDA Lab, Denver) and progesterone assays (Physiology Lab, Colo. St. Univ.). Fat content (percent dry matter) of femur marrow or lower jawbone marrow obtained from dead radioed elk was determined (Colo. Div. Wildl. Lab, Ft. Collins). Survival We monitored life or death .status of radioed elk during daily ground surveys and aerial surveys conducted at 2-4 week intervals from December 1993 through April 1994 and via monthly aerial surveys in May and June 1994. Survival rat~s (S) of radioed elk were calculated using the binomial estimator with a variance, VAR(S) = S(l-S)/n (White and Garrot 1990)(Proc FREQ, SAS 1988). We chose not to use the staggered entry approach and did not use a Kaplan-Meier approach because no animals were censored (White and Garrot 1990). Life or death status of all radioed calves was known for the period 2 December through 15 June. On 15 June, calves become yearlings for purposes of calculating rates of calf survival. Life·or death status was known for 65 of 68 adult females through 15 June with all females of known status through 28 April. Sighting Bias We conducted 110 aerial sighting bias trials that targeted individual radioed elk located on search quadrats (Samuel et al. 1987). Trials were conducted on 25, 27, 28, and 31 January, l, and 14-17 February, and on 9 March. Search quadrats were approximately 2.6 km (l mi2 ) in size with boundaries based on topographic features We used a Bell B-1-Soloy helicopter, the same pilot, and observation teams composed of 2 persons selected from 3 observers and 2 navigators for all V �31 trials. The Bell-Soley offers good forward and lateral visibility. Navigators were seated in the middle with the pilot to their left and the primary observer to their right. Navigators were highly experienced at flying sample quadrats and observers had high, moderate, low experience in flying sample quadrats. However, all observers were well experienced in conducting sex and age ratio helicopter survey flights for elk. The helicopter was flown at speeds of 65-95 Janph at 25-40 m above the vegetation canopy when flying the perimeter of search quadrats and weather permitting, at slower speeds and lower heights when searching the interior of quadrats. Searching time per quadrat ranged from 10-50 minutes. Sighting bias trails were usually conducted in 2 sessions each day. Op to 8 different radioed elk were located via telemetry with a Cessna 185 during each early morning and mid-day session. Locations of elk were assigned to a previously determined search quadrat. Assignment to a quadrat was confirmed by a technician on the ground who checked Loran-c locations obtained by plane with latitude-longitude locations of quadrats plotted on USGS 7.5 minute topographic maps. The navigator was given a list of 5-8 target elk and quadrats per session but no information as to where target elk were located within quadrats. Usually within 2 hours of locating target elk with the plane, the helicopter team searched the quadrat and counted all groups of elk until either·the target radioed elk was seen or until the entire quadrat had been searched and the target elk not seen. If a target elk was missed, the helicopter team located the elk via telemetry immediately after completing their search of the quadrat to determine whether the elk was on or off the quadrat at the time they flew the quadrat. The helicopter team recorded initial group size, total group size, activity behavior, vegetation type, percent occlusion cover, and percent snow cover for the portion of each group initially detected whether or not a group contained a radioed elk (Appendix I). Snow type, light conditions, wind conditions were assessed for each quadrat. When radioed elk were observed, they were usually individually identified by the number/symbol code, and if a target elk, the search was then terminated. Several radioed elk were often seen per quadrat. After completing the search of a quadrat, observers located target elk that were missed, and, if found on the quadrat, observers recorded all sighting variables for the missed group. Usually 10-12 trials were conducted each day. Successful trials occurred when the target elk was on the assigned quadrat whether or not that elk was seen. Unsuccessful tr-ials occurred when target elk were found off the assigned quadrat immediately subsequent to the time the quadrat was flown. Elk were off quadrats primarily due to their movements that occurred between the time elk were located with the plane and the time the helicopter team flew the quadrat. We wanted to avoid any positive effects on sightability that might be gained if an observer located a target elk during 1 trial and then was assigned that same target elk during another trial. Thus, elk were targets for an individual observer only once. However, the same navigator was involved in 2 trials with each of 2 elk, but the 2 trials were generally 2-4 weeks apart and we suspect the navigator had little influence on the ability of the ·-·.observation team to detect ·the target elk. We subsequently targeted 90 different radioed elk: 70 were targeted 1 time and 20 were targeted 2 times during the 110 trials and for 106 trials,· there was a unique combination of observer and navigator attempting to detect the targeted elk. Sighting bias models were developed using logistic regression (Pree GENMOD, SAS). The dichotomous classification of groups seen or missed was the dependent variable. Initial group size, total group size, activity behavior, vegetation type, percent occlusion cover, and percent snow cover, etc. were independent variables. Significance level for independent variables was PS 0.05 for univariate tests and PS 0.10 for stepwise regression in multivariate models. Relative efficiency of models was assessed using AIC values (Akaike Information Criterion= -2(-log likelihood value)+ 2[No. model parameters)). Observer, navigator, trap method, elk sex, elk age, activity, vegetation type, �32 snow type, and wind were treated as class variables and group size, percent occlusion cover, and percent snow cover were treated as continuous variables. Variances for estimates of total elk that were derived from sighting models followed a modified version of that presented by Steinhorst and Samuel (1989)(pers. comm. D. Bowden and G. White). V Movements We located 34 radioed elk (17 adult females, 17 calves [9 males, 8 females) at least once per month since capture via telemetry using a Cessna 185. These elk were selected at random from within trap zones and equalized by age class. These elk will be monitored seasonally to document general movements and areas of use with results reported next year. • RESULTS ARD DISCUSSION Survival From 2 December 1993 through 15 June 1994, 6 calves died (4 male, 2 female) and 3 adult females died (Table 2). Overwinter survial rate(± 951 CI) for calves was 0.918 ± 0.063 and for adult females was 0.956 ± 0.049. These survival rates were associated with a winter considered to be mild in both temperature and snow depth. Suspected· causes of death were primarily malnutrition and predation _for calves and shooting and predation for adult females (Table 2). Those calves dying from malnutrition had marrow fat values< 131 at death and body weights< 102 kg at capture. The 2 smallest male calves weighed (Table 3) died of suspected· malnutrition (Table 2). Male calves had larger body weights (P = 0.03), longer hind leg lengths (P = 0.07), and higher condition indexes (P = 0.03) than female calves. Total body lengths were not different between sexes (P = 0.36)(Table 4). Blood Assays Brucellosis tests were negative for all 43 adult females sampled. Progesterone assays indicated 25 (741) of 34 adult females were pregnant (1.37-4.69 ng/ml), 2 (61) were possibly pregnant (1.02-1.04 ng/ml), and 7 (201) were not pregnant(~ 0.37 ng/ml) (Freddy 1989). Of the females judged not pregnant, 5 were judged to be 2-4 years old and 2, 5-9 years old. Sighting Bias We completed 99 successful sighting trials involving 45 adult females and 54 calves of which 26 were males and 28 were females. Elk involved in 11 unsuccessful trials we1;e .6 adult females and 5 calves. Overall, observers missed 15.41 of the targeted elk (Table S). The degree of negative sighting bias was not different among observers (P = 0.53) ranging from 12.2 to 21.91. Therefore, the simple binomial correction factor(± 951 CI) for detecting elk was 0.846 ± 0.071. Of the 16 target elk missed during successful trials (representing 16 target groups), 14 (881) were calves (SM, 9F) and 2 (121) were adult females. Frequency of missing male and female calves was not different between sexes (P = 0.28). Calves missed were in groups of~ 4 elk, except for 1 group of 30 elk. Capturing calves with a helicopter or corral trap had no effect on their sightability (P = 0.66). Univariate tests indicated elk age (calf or adult), initial group size, log(ln) initial group size, total group size, log(ln) total group size, activity behavior, vegetation type, percent occlusion cover, percent snow cover, and wind conditions affected the probability of sighting elk (P ~ a.as, Tables 6, 7). The ln total group size was the most efficient of these single variable models (AIC = 74.678; intercept= 0.1215, group size coefficient= V �33 1.3073). Log transformations of either initial group size or total group size provided more efficient models than untransformed values (Table 7). The initial multivariate model for stepwise regression incorporated elk age, ln total group size, activity behavior, vegetation type, percent occlusion cover, percent snow cover, and wind conditions. Vegetation type, percent occlusion cover, and wind were not significant (Model A, Table 8). Proceeding in a step-down process, wind was dropped as the least significant variable from model A, then percent occlusion cover from model B, to derive model C in which elk age, activity behavior, vegetation type, ln total group size, and percent snow cover were all significant. However, model C necessitated estimates of coefficients for 12 parameters which makes this model unwieldy and likely subject to highly imprecise estimates of corrected numbers of elk although this model had a relatively low AIC value (Table 8). other models having fewer parameters were assessed. In model D having 3 primary variables, elk age was not significant, but ln total group size, percent occlusion cover, and the interaction of elk age and ln total group size were significant (Table 8). This significant interaction term reflected that although calves were the predominate class of elk missed, they were almost always in small groups when found after completely searching quadrats. We are not sure whether calves were missed because they were in separate small groups (assumed, true sighting bias) or because they represented a fraction of a larger group that was detected but the calves were not seen (count·ing bias). Missed groups involving targeted calves were seldom the only group of elk on the quadrat. Model E includes _ln total group size and percent occlusion cover which mimics variables used in sighting-models developed in Idaho for elk (Samuel et al. 1987). These models were surprisingly similar considering that the Idaho model was developed in tall conifer habitats and based only on sightability of adult elk (Samuel et al. 1987) (Table 9). Model F includes only ln total group size and had the highest AIC value but probability of detecting a·group increased rapidly to~ 0.90 for groups~ 6 elk (Table 8, Fig. 1). The degree to which selected models corrected number of elk counted and the variances of corrected counts were assessed. The simple binomial correction, model G, inflated counts the most (18%) and had nearly the highest variance (Table 10). Inflated counts resulted from applying the 0.846 correction factor to large groups of elk which had a sighting probability of nearly 1.0 (Fig. 1) and needed no adjustment in numbers. Models E and F having ln total group size or ln total group size and percent occlusion cover had similarly low variances and low corrections for counts of 3-4% (Table 10). These 2 models highlight the need to include group size as an adjustment variable for correcting counts. Adding percent snow cover as an additional variable, model H, markedly increased the variance and provided minimal additional correction to counts (Table 10). We caution, however, that small variances and potentially precise estimates about corrected counts per quadrat based on adjustments for group size may not result in highly precise estimates of population size because of variance associated with total counts of elk among sample quadrats when a sampling system is employed (otten et al. 1993). CORCLUSIORS We obtained acceptably precise estimates of calf and adult survival rates during winter and recommend continuing our current sampling effort of monitoring 75 radioed calves and 75 radioed adult females in 1994-95. Efforts to measure and develop criteria to adjust for negative bias in counts of elk were promising and we.recommend conducting an additional 100 sighting bias trials in 1994-95 to refine a sighting bias model. We are particularly interested in evaluating effects of elk age (calves) on sighting bias and assessing potential heterogeneity of sighting probabilities for individual elk among years by resighting elk in 1994-95 that were involved in trials in 199394. �34 LITERATURB CITED Freddy, D. J. 1989. Effect of elk harvest systems on elk breeding biology. Colo. Div. Wildl. Game Res. Rep. July(l): 35-60. Freddy, D. J. 1993. Estimating survival rates of elk and developing techniques to estimate population size. Colo. Div. Wildl. Game Res. Rep. July: 83-117. Samuel, M. D., E. O. Garton, M. W. Schlegel, and R. G. Carson. 1987. Visibility bias during aerial surveys of elk in northcentral Idaho. Wildl. Manage. 51:622-630. J. otten, M. R., J.B. Haufler, s. R. Winterstein, and L. c. Bender. 1993. An aerial censusing procedure for elk in Michigan. Wildl. Soc. Bull. 21: 73-80. SAS Institute Inc. 1988. SAS/STAT User's Guide, 6.03. SAS Institute, Inc., . Cary, NC. 1028pp. Steinhorst, R. K., and M. D. Samuel. 1989. Sightability adjustment methods for aerial surveys of wildlife populations. Biometrics 45:415-425. White, G. c., and R. A. Garrot. 1990. Analysis of wildlife radio-tracking data. Academic Press, Inc., San Diego. 383pp. V V �35 Table 1. Capture objectives and numbers of elk radioed in 8 trapzones, Game Management Unit 42, December, 1993. \,.._,I Trap Zone A B C D E F G H Blls Col11i;:ed• I.2tBJ. He1f.g ~2D:11 Capture O)zjectj.ve Bame Garfield Gibson Uncle Bob West Divide Hightower Middle Kamm West Kamm Dry Hollow 8 Adult Eel!H!lgs 4 12 12 10 10 8 8 10 10 25 29 21 17 0 19 29 21 17 0 0 8 6 6 0 44 35 0 35 7 9 1 19 150 141 100 41 36 37 68 20 24 26 All Calve! Hales [emale1 3 1 5 8 10 7 6 5 4 3 0 0 3 2 0 6 0 0 0 0 • Helio - Helic~pter net-gunning, Corral - corral-trap. Table 2. Causes of mortality for radioed elk in Game Management Unit 42, l December-ls June 1993-94. Radiocollar ~ l!:!!S!H!DSX Sex tgg• 172.090/93 172.542/93 172.690/93 172.899/93 173.000/93 173.262/93 173.289/93 173.461/93 173.469/93 F p 11 yrs 6 yrs 8 yrs 10 mos 11 mos 10 moo 10 mos 1 F F F M M M M 9'mos- 11 mos Date Dead Batimated cmase of Deatb 1/16/94 2/01/94 1/24/94 3/18/94 4/25/94 3/18/94 3/22/94 2/07/94 4/25/94 Wounding LOBS cougar kill Gunshot, poached Cougar kill Malnutrition Possible predation Undetermined Malnutrition Malnutrition Body wt:. Usg!b Harrow Fat E1;ce11Ji Da Hatte, not available 18.0° not available 2e.sd 12.6c 28.9d not available 2.1° 2. ,c 86 102 ·108 111 82 77 Approxlmata age at death. b Whole body weight at capture in December 1993. ° Fat content of either the right-or left femur bone marrow. d Fat content: of lower jaw bona marrow. Table 3. Frequency distribution of whole body weights for male and female elk ·calves trapped in Game Management Unit 42, December, 1993. Percentage per weight class shown in parentheses. Bodv Wgigb~ glll! (kg! 100-109 110-119 120-129 130-139 140-149 Total 3 (8.6) 6 (17.1) 12 (34.3) 10 (28.1) 1 (2.9) 1 (2.9) 35 (100) 8 (22.9) 12 (34.3) 4 (11.4) 6 (17.1) 0 (0.0) 0 (O.O) 35 (100) Sax 70-79 80-89 90-99 Mala 1 (2.9) 1 (2.9) Pamala 1 (2.9) 4 ( 11. 4) �36 Table 4. Body measurements for elk calves trapped in Game Management Unit 42, December, 1993. Mean Meaeurement; Body Weight (kg) 112. 7 191.2 Body Length (cm) Bindfoot Length (cm) 56.3 0.59 Condition Index (Wt./Body Length) MaJ.e CaJ.veg Min Max o Mean 141.0 35 210.0 36 61.0 36 0.67 35 SD 13.7 10.2 2.2 77.0 164.0 52.0 0.43 o.os Eemale ~iJ.ves SD Mi.n HM 103.8 12.2 188.9 54.8 9.8 76.0 167.0 51.0 123.0 35 207.0 35 0~55 2.1 o.os 0.46 59.0 0.64 u 34 34 Table S. Summary of elk sighting bias trials conducted in Game Management Unit 42, January-March 1994. Percentages shown in parentheses. 0bB8!::£8E Trials 6tt:eme51 Trials• Hot Usem2J.!} Trials Uae!lzie Trial Blk Detectgs! Trial Blk Hot ggtected JB 29 3 4 DK 36 45 26(90) 32(89) 22(84.6) JE 4 41(91) 36(87.8) 4(15.4) 7(21.9) 5(12.2) Totals 110 11 99(90) 83(84.6) 16(15.4) 25(78.1) V Table 6. Blk sightability survey.results by major independent variables from Game Management Unit 42, January-March, 1994. !IEialalSI IISh SIEmll!I ,,. 6 7 1 1 0 0 0 1 0 0.57 0.59 0.92 0.86 1.00 1.00 1.00 0.93 1.00 Iii.lad Ian 4 5 6 7-15 ~ 16-30 31+ v-... 'fype Cleariq/Ag Ripuian &&;Bruh O&Jcbnab PiDyaa-Junipm: 6 Aap8D 3 2 Tall CODJ.far ocal. COVU(~) 0-19 20-39 40-59 60-19 80-100 I 8 10 11 6 5 5 17 14 7 Baddad Standing Hoving Navigator GB VG 9 1 3 40 26 3 1 DH 1.00 1.00 1.00 O.89 O.81 Bllc,,Aga 0.33 calf sax o.so i'amala 22 25 23 12 1 O.96 0.89 O.79 o.ao 0.25 !!Ei.llalSI 11,111£1 lama ~ snow cover (I) 9 2 53 2 28 5 11 5 45 38 0.29 0.86 0.93 a.so 0.88 0-19 20-49 50-99 100 4 7 5 5 9 Trap (All elk) COffal Trap 4 BelJ.c:optar 12 Trap (calvaa) • corral Trap 3 Balicoptar 11 y-,£.Iblilty (group• 11aen / gmup■ seen plua groups iiilaaacl). 22 0.85 25 36 O.78 43 40 a.ea 0.96 0.74 21 19 O.81 0.68 27 56 O.87 O.82 11 0.79 0.73 29 3 2 0.40 3 4 29 48 0.97 0.84 1 9 0.57 Snow Type l're■h Old . Adult hmala 2 calf (H + I') 14 Hal.a 1 3 6 3 3 ~ Obaerver JB JB 0 0 0 5 11521 SIES!HDII lli.lllml IUD Behavior GJ:Oup S1:a 1 2 3 !1d1blo Wind Light ~ate St=ng 4 12 21 16 0 0 67 14 62 2 0.84 0.84 0.81 1.00 1.00 Ll.gbt Bright Dull Ba:y 12 52 0 11 4 20 0.81 1.00 0.83 V �37 Table 7. summary of variables ·tested in univariate tests using logistic regression for elk sightability trials, Game Management Unit 42, January-March, 1994. Likelihood Ratio Prob. >Chi-Square Variable Date of Sight Trial Sigh't; Trial Navigator Observer Individual Elk Sex of Elk Age of Elk Capture Type Initial Group Size Log(ln) Initial Group Size Total Group Size Log(ln) Total Group Size Elk Activity Vegetation Type Oc~lusion Cover I Snow Cover I Temperature Wind Conditions Light Conditions snow Type 8 AJ:C 0.7152 0.5936 0.2763 0.5399 0.5650 0.6257 0.0022 0.5453 0.0166 0.0044 0.0027 0.0001 . 0.0014 0.0405 0.0010 0.0275 0.3127 0.0449 0.1233 0.9797 Sign./ Nonsign. NSign NSign NSign NSign NSign • NSign Aic4 91.450 91.299 90.398 90.351 12.318 91.345 82.170 93.218 SIGN NSign SIGN SIGN SIGN SIGN SIGN SIGN SIGN SIGN 85.844 83.468 82.589 74.678 82.435 90.417 80.703 86.725 91.023 89.376 91.397 93.583 NSign SIGN NSign NSign = Akalke's Information Criterion Table 8. Multivariate analyses of factors affecting sighting probability based on stepwise logistic regression for elk in Game Management Unit 42, JanuaryHarch, 1994. Model A No. Parameters + Intercept 14 B 13 C D 12 6 E p 2 3 1 Ageaelk age class, Aic6 Parameters Used• Age , Act , Vega, Wind, LnGroup, Covert, sncw1• Age•, Act•, Vega•, LnGroup•, Covert, Snow%• Age•, Act•, Vega•, LnGroup•, Snow,• Age, LnGroup•, Covert• LnGroup•, Cover,• LnGroup• • • • Value 58.32 58.96 59.16 62.61 68.89 74.68 Act::aelk activity class, Vege~egetatlon type class, wind=wind class, LnGroup=log total group size, COvert=occlusion cover percent, Snorimsnow cover percent, * Denotes variable as significant . in model, Pm S 0.10). b Lower AIC value denotes better fitting model. �38 Table 9. Comparison of logistic regression models incorporating log(ln) group · size and percent occlusion cover or percent vegetation cover for sightability trials involving elk in Colorado and Idaho. Variable Coeffiecent Constant-Colorado 1.98 constant-Idaho 1.22 ln Group Size-Colorado 1.23 ln Group Size-Idaho 1.55 I 0ccl. Cover-Coloraod -0.04 I Vege. Cover-Idaho -o.os 1 SE 0.86 o. 674 0.40 0.37 4 0.02 0.01• Calculated from Table 2 in Samuel et al. 1987. Table 10. Estimated variance components and estimated corrected numbers of elk for selected sightability models for elk in Game Management Unit 42, 1994. Corrected Percent variance· 0 Totala gqg;aqt;ion Estimate0 COvarianca0 Model Elk 49,264 1,666 610 2,534 118 a-Binomial1 46,121 1,454 543 103 284 12 246 l'-LnGroup 1,462 104 873 264 111 498 B-LnGroup+ COvarl 284,549 1,489 -805,323 806,283 283,588 106 B-LnGroup+ Covarl+ Snpwt · • Variance from binomial sighting probability of groups. b Variance from predicted sighting probability_. 0 Variance frcm covariance of predicted sighting probability;; d Variance of estimate of total elk. • f Binomial model uses simple sighting bias correction of 0.846 applied to all observed groups of elk. 0 Total elk observed in groups u~ed to develop models was 1,405 elk in 99 groups. V V �39 z 1.0 o·o.s i== U 0.8 w ti 0.1 C 0.6 MEAN LL O 0.5 95 % Cl ~ 0.4 :l 0.3 m <C 0.2 m 0 0.1 £C Q. 0 1 2 3 4 5 6 7 7 20 60 150 400 1100 LOG (In) TOTAL GROUP SIZE 3 TOTAL GROUP SIZE RG. 1. Probability of detecting elk by group size, +/- 95% Cl, Game Management Unit 42, January - March, 1994. Group sizes observed during sighting ~ias trials ranged from 1 - 400. �40 15 Appendix I. Procedures for flying quadrats and definitions of variables assigned to each group of elk observed on guadrats. A. PROCEDURES FOR FLYING QUADRATS: 1. OBTAIN PROPER MAPS, ARRANGE IN ORDER OF NEED, DECIDE ON GENERAL ROUTE TO QUADRAT, HIGHLIGHT QUADRAT BORDERS lJITH FELT MARKER, AND DETERMINE LIKELY STARTING POINT. MAKE SURE NAVIGATOR HAS LISTING OF FREQUENCIES AND NECKBANDS TO DETECT (MINIMIZE ANY TALK ABOUT TARGET ANIMALS) AND THAT TELONICS TR-2 RECEIVER IS LOADED AND READY TO GO. CHECK TELEMETRY SYSTEM ON HELICOPTER WITH • DUMMY COLLAR. 2. FLY PERIMETER OF QUADRAT FIRST IN A CLOCKYISE MANNER SO THE INSIDE OF THE QUADRAT IS TO THE RIGHT OF THE PRIMARY OBSERVER. 3. DETERMINE STATUS OF GROUPS OF ELK ON PERIMETER. A. ELK MOVING OFF THE QUADRAT WHEN DETECTED ARE CONSIDERED ON THE QUADRAT. B. ELK MOVING ONTO THE QUADRAT WHEN DETECTED ARE CONSIDERED OFF THE QUADRAT. C. IF A GROUP IS STANDING ON THE PERIMETER, COUNT THOSE ELK INSIDE THE QUADRAT. D. RESIST THE TEMPTATION TO LEAVE THE PERIMETER TO COUNT A GROUP ON THE INSIDE OF THE QUADRAT. USE YOUR BEST JUDGEMENT AT THE TIME OF DETECTION. E. IF A MARKED ELK IS DETECTED NEAR THE BORDER BE SURE TO NOTE ALL IMPORTANT DATA VARIABLES AS THIS ELK COULD BE THE TARGET ELK. 4. FLY INTERIOR OF QUADRAT. A. FLY THE INTERIOR OF THE QUADRAT SYSTEMATICALLY IN STRIPS OR STRIP.CONTOURS. USE PROMINENT TERRAIN FEATURES TO DIVIDE THE QUADRAT INTO SMALLER COUNTING BLOCKS. IF DECIDE TO FLY STRIP-CONTOURS, WORKING FROM THE HIGHEST TO LOWEST ELEVATION USUALLY . WORKS BEST. B. BE PATIENT AND STRIVE FOR 100% COVERAGE. 5. WHEN MULTIPLE GROUPS OF ELK ARE DETECTED ·SIMULTANEOUSLY, FOLLOW THESES GUIDELINES. A. PRIMARY OBSERVER SHOULD BEGIN OBTAINING DATA ON GROUP NEAREST THE HELICOPTER AND USE HELICOPTER TO KEEP GROUPS SEPARATED. B. NAVIGATOR SHOULD FOCUS MOMENTARILY ON SECOND GROUP OBSERVED AND NOTE LOCATION, ACTIVITY AND NUMBER OF ANIMALS FIRST SEEN. C. AFTER COMPLETING DATA COLLECTION ON FIRST GROUP, PROCEED TO LOCATION OF SECOND GROUP, DETERMINE VEGETATION TYPE, OCCLUSION%, AND SNOW COVER% AT SITE OF DETECTION, THEN FIND AND COUNT SECOND GROUP. 6. RULES FOR MARKED ANIMALS A. FOR EVERY MARKED ELK SEEN (UNTIL THE TARGET ELK IS DETECTED) ·oBTAIN A POSITIVE IDENTIFICATION VIA THE NECKBAND NUMBER OR WITH TELEMETRY TO DETERMINE THAT THE MARKED ELK IS OR IS NOT THE TARGET ANIMAL. B. IF MARKED ELK ARE SEEN AFTER DETECTING THE TARGET ELK, OBTAIN AN IDENTIFICATION FROM THE.NECKBAND WITHOUT UNDULY HARASSING THE ELK OR USING TOO MUCH TIME. 7. RULES FOR DETECTING THE TARGET MARKED ELK A. YOU WILL ESTABLISH A FLIGHT PATTERN TO COVER THE ENTIRE QUADRAT. WHEN YOU HAVE COMPLETED THIS PATTERN, YOU ARE DONE FLYING THE QUADRAT. IF YOU HAVE NOT DETECTED THE TARGET ANIMAL DURING THIS PATTERN, DO NOT V �41 16 RE-SEARCH THE QUADRAT AS THIS WOULD BE DISHONEST; IE, DO NOT SEARCH FOREVER TO FIND THE TARGET ELK, IT MAY NOT BE THERE OR THERE MAY NOT BE ONE TO FIND! B. IF YOU DO NOT FIND THE TARGET ELK AFTER COMPLETING YOUR FLIGHT PATTERN, TURN ON THE RECEIVER AND START SEARCHING THE QUADRAT FOR A SIGNAL. C. IF THE TARGET ELK IS ON THE QUADRAT, COUNT THE GROUP AS THOUGH IT WAS ANY OTHER GROUP OF ELK, AND BE SURE TO OBTAIN ALL VARIABLES ASSOCIATED WITH THAT GROUP. THIS IS THE MOST IMPORTANT DATA WE CAN COLLECT IN REGARDS TO SIGHTING BIAS ADJUSTMENT FACTORS AS IT REPRESENTS AN ELK YOU DID NOT SEE. D. IF THE TARGET ELK IS DEFINITELY OFF THE QUADRAT SPEND SOME TIME LOCATING IT BECAUSE IT COULD BE NEAR THE BOUNDARY AND MAY HAVE RUN OFF THE QUADRAT (BACKTRACK TO DETERMINE IF LEFT QUADRAT). SPEND ENOUGH TIME SEARCHING TO DETERMINE TO YOUR SATISFACTION THAT THE ELK WAS LIKELY OFF THE OUADRAT DURING YOUR TIME ON THE OUADRAT. E. NOTE NUMBER OF MARKED ANIMALS SEEN IN EACH GROUP, AND ID'S OF THOSE ELK IF OBTAINED. LOCATIONS OF ELK SEEN ON A QUADRAT WILL BE ASSUMED TO BE THE MIDDLE OF THE QUADRAT FOR PURPOSES ~F DISTRIBUTION AND MOVEMENT INFORMATION UNLESS YOU OBTAIN A LORAN-C LOCATION FROM THE HELICOPTER. B. DEFINITIONS THAT APPLY TO EACH GROUP OF ELK DETECTED ON A QUADRAT. OUR DEFINITIONS ARE BASED ON THE HYPOTHESIS THAT DETECTING ELK IS DEPENDENT ON INITIALLY DETECTING SOME FRACTION OF EACH GROUP. 1. DETECTION GROUP SIZE: NUMBER OF ELK INITIALLY DETECTED WEN AN OBSERVER SEES A GROUP. THIS NUMBER WILL REPRESENT SOME PORTION OF THE TOTAL GROUP SIZE. 2. ANIMAL ACTIVITY: THE ACTIVITY OF THE ELK INITIALLY DETECTED OR THE ACTIVITY OF THE MAJORITY OF THE ELK INITIALLY DETECTED. ACTIVITY CLASSES WILL •• BE BEDDED-B, STANDING-S, MOVING-M. 3. VEGETATION TYPE: THE PREDOMINATE TYPE OF VEGETATION FOR 30 METERS AROUND THE INITIAL ELK SEEN. VEGETATION TYPES ARE: OK--OAKBRUSH-TALL MOUNTAIN SHRUB SG--SAGEBRUSH-(INCLUDING GREASEWOOD & RABBITBRUSH) AS--ASPEN PJ--PINYON-JUNIPER TC--TALL CONIFER (DOUGLAS FIR, ENGELMANN SPRUCE, LODGEPOLE PINE) • RP--RIPARIAN (COTTONWOOD/WILLOW BOTTOMS) CA--CLEARINGS/AGRICULTURAL FIELDS (CLEARINGS CAN BE NATURAL MEADOWS, CLEAR-CUTS COVERED WITH SNOW, DRILL-PADS, OR ANY SIZEABLE OPENING THAT APPEARS TO BE A CLEARING DUE TO SHORT VEGETATION AND/OR SNOW COVER) 4. % OCCLUSION COVER: AN ESTIMATE OF THE PERCENTAGE OF THE OBSBRVER'S VIEW BLOCKED BY VEGETATION INTERFERING WITH THE OBSERVERS ABILITY TO SEE THE INITIAL ELK DETECTED IN A SPACE OF 30 METERS AROUND THE ELK. OBSERVERS WILL ESTIMATE IN INCREMENTS OF 10%, IE; 0, 10, 20, 30, ... , 100. DATA WILL BE POOLED IATER. VEGETATION INCLUDES STEMS AND TRUNKS OF DECIDUOUS SHRUBS AND TREES, TRUNKS AND LEAVES OF EVERGREEN CONIFERS, LEAVES AND STEMS OF ANY OTHER EVERGREEN SPECIES. GRASS AND SHORT SAGEBRUSH SHOULD HAVE 0% OCCLUSION COVER �42 17 EXCEPT POSSIBLY WHEN ELK ARE BEDDED·;. 5. %SNOW COVER: THE PERCENTAGE OF BARE GROUND COVERED BY SNOW FOR 30 M AROUND _THE INITIAL ELK DETECTED. OBSERVERS WILL ESTIMATE IN INCREMENTS OF 10%, IE; 0, 10, 20, 30, .... 100. DATA WILL BE POOLED LATER. 6. TOTAL ELK IN GROUP: THE TOTAL NUMBER OF ELK COUNTED IN A GROUP. A GROUP IS DEFINED AS A CLUSTER OF ELK ACTING INDEPENDENTLY EITHER IN BEHAVIOR OR SPACE FROM ANOTHER CLUSTER OF ELK. GUIDELINES FOR DEFINING A GROUP ARE: CLUSTERS OF ELK SEPARATED BY >30 METERS OR CLUSTERS SHOWING DIFFERENT PATrERNS OF BEHAVIOR. DEFINING A GROUP CAN BE SUBJECTIVE IF ELK ARE AT HIGH DENSITIES ON THE QUADRAT. THESE DEFINITIONS APPLY TO AMBIENT CONDITIONS EXISTING ON EACH QUADRAT·AT THE TIME OF FLYING 1. WIND: WIND WAS : LIGHT-L, MODERATE-M, OR STRONG-S IN RELATION TO THE RELATIVE EFFECT OF THE WIND ON. THE ABILITY TO FLY THE QUADRAT IN A SLOW, LOW AND DELIBERATE MANNER. 2. LIGHT CONDITIONS: LIGHT CONDITIONS WERE EITHER: BRIGHT SUNSHINE WITH HIGH CONTRAST-B, HAZY SUNSHINE WITH HIGH THIN CLOUDS-H, OR DULL SUNSHINE DUE TO OVERCAST OF CLOUDS-D. 3. SNOW TYPE: SNOW TYPE IS EITHER FR.ESH SNOW THAT HAS FALLEN WITHIN THE PAST 48 HRS-F, OR OLD SNOW THAT IS OLDER THAN 48 HRS-0. ·sNOW CONDITIONS WILL LIKELY BE THE SAME FOR ALL GROUPS OF ELK DETECTED ON A QUADRAT. ~ 3. TEMPERATURE: TEMPERATURE (F) DURING THE MORNING AND AFTERNOON FLIGHTS AS MEASURED AT THE BASE OF OPERATIONS FOR EACH FLIGHT \ J ""--" �63 Colorado Division of Wildlife Wildlife Research Report July 1995 JOB PROGRESS REPORT State of Colorado Project No. W-153-R-8 Mammals Research Work Plan No. 3 Elk Investigations Job No. 9 Estimating Survival Rates of Elk and Developing Techniques to Estimate Population Size Period Covered: Author: July 1, 1994 - June 30, 1995 D. J. Freddy Personnel: F. Barnes, J. Broderick, G. Byrne, A. Coriell, D. Crane, J. Ellenberger, J . Frothingham, v. Graham, J. Gray, R. Hays, G. Loucks, P. Will, R. Witt CDOW; D. Bowden, C. Vardeman, G. White, CSU; K. Crane, C. McCarty, D. Ouren, volunteers; USFS Rifle, BLM Glenwood Springs, cooperating. ABSTRACT We radio-collared 69 calf elk (Cervus elaphus nelsoni) (6 months old) and an additional 14 adult female elk(~ 1 year old) in December 1994 to estimate survival rates during winter and used these same elk in 103 aerial sighting bias trials to develop models for estimating degree of negative sighting bias when counting elk with a helicopter on sample quadrats. Elk were captured using portable corral traps and helicopter net-gunning . Survival rates(± 95% CI) from December 1994 to June 1995 were 0 . 90 ± 0.07 for calves and 0 . 96 ± 0.04 for adult females. Suspected causes of death were malnutrition and predation for calves and shooting and predation for adult females. Calves suspected of dying from malnutrition had marrow fat values <13% at death and body weights <102 kg at capture while calves dying from predation had marrow fat values of 3.7-35.6% and body weights of 86-140kg. For adults ~1 year-old, hunting accounted for 88% of 24 deaths. Average sighting probability of elk groups was 82.3% which equates to a 17.7% negative sighting bias. There were no differences in sighting bias between years (P > 0.70) but there were differences in sighting bias among observers (P = 0 . 06) which ranged from 9 . 3% to 22.4%. Univariate tests indicated elk age (calf or adult), elk sex (calves only), initial group size, log(ln) initial group size, total group size, log(ln) total group size, activity behavior, vegetation type, percent vegetation occlusion cover, observer, navigator, pilot, and trapping method affected the probability of sighting elk (P ~ 06). Multivariate analyses incorporated elk age, ln total group size, activity behavior, vegetation type, percent vegetation occlusion cover, percent snow cover, navigator, observer, and year. The best sightability model out of 2,048 possible models was a complex 12 parameter model involving nearly all major variables. We are currently assessing the effects on sighting probability of using less complex models. �65 JOB PROGRESS REPORT BSTIMM?IHG SURVIVAL RADS OP BLK ARD DBVBLOPIHG BCIIHIQUBS m BSTIMAD POP~IOH SIZB David J. Freddy P. H. OBJECTIVE Estimate survival rates of adult female and calf elk and develop techniques to estimate population size. SEGMEtr.r OBJECTIVES 1. Radio-collar 75 calf and 15 adult female elk during December 1994 in Game Management Unit 42 south of Rifle, Colorado. 2. Estimate winter and annual survival rates of calf and adult female elk from known fates of radioed elk. 3. Estimate probability of sighting and counting elk during helicopter surveys using radioed elk as a known population. 4. Estimate density of elk in a portion of GMU-42 using a quadrat sampling system. 5. Analyze survival and sightability data and summarize annually in Federal Aid Job Progress reports INTRODUCTION our objectives are to provide reliable estimates of survival rates for calves and adult females during winter and for adult females throughout the year for the period 1993-94 through 1997-98. Additionally, we will develop and test a system for estimating population size that will incorporate estimates of sighting bias in conjunction with a random sampling system using search quadrats as sample units. our winter study area encompasses about 839 Jan2 (324 mi1 ) in the eastern half of Game Management Unit 42 south and east of Rifle, COlorado. Elk winter range vegetation types include juniper-pinyon woodland (Juniperus osteosperma-Pinus edulis), oakbrush-mountain shrub (Quercus gambelii-Amelanchier alnirolia), aspen (Populus tremuloides), sagebrush (Artemisia tridentata), and agricultural fields (Freddy 1993, 1994). METHODS Marking We placed radio collars (172-176MHz) having mortality sensors on 69 calves (6 months old), of which 33 were males and 36 were females, and 14 adult females (~ 1 year old). Of these, 65 calves and 2 adults were trapped from 7-11 December 1994 using helicopter net-gunning and 4 calves and 6 adult females were trapped from 13-21 December using portable corral traps. Six adult females were captured in corral traps 2 March 1995 to elucidate movements of elk associated with a specific area of private land. Helicopter capture occurred at 11 remote sites located primarily on public lands while corraltrapping occurred at 3 sites, 2 on private and 1 on public lands. Trapping effort was allocated among 8 geographic trap zones to assure that radioed elk were representative of most if not all segments of the population (Table 1). Radio collars were of the same type used in 1993 (Freddy 1994). �66 Calves captured by net-gunning were ferried by helicopter to processing points usually within 1.6 Jan of capture sites. At processing points, body weight, total body length, hind foot length, and rectal body temperature (F) were measured and calves were then radio-collared and released. Similar measurements were also made on calves that were corral-trapped and then released at the trap site. Body measurements for calves were compared between sexes using Proc FREQ, GLM, and REG (SAS 1988). V V survival We monitored life or death status of radioed elk during daily ground surveys and aerial surveys conducted at 2-4 week intervals from December 1994 through April 1995 and via monthly aerial surveys from May to November 1994 and May to June 1995. Survival rates (S) of radioed elk were calculated using the binomial estimator with a variance, VAR(S) = S(l-S)/n (White and Garrott 1990)(Proc FREQ, SAS 1988). We chose not to use the staggered entry approach and did not use a Kaplan-Meier approach because few animals were censored (White and Garrott 1990). Life or death status of all calves radioed in December 1994 (68 with functional collars) was known for the period 7 December 1994 through 15 June 1995 (1 male calf collar, 173.949/94, failed 2 weeks post-capture and although the calf was seen alive through 30 January 1995 it was excluded from estimates of survival rates). On 15 June, calves become yearlings for purposes of calculating rates of calf survival. Life or death status for adult females, yearling females, and yearling males collared in December 1993 or 1994 was known for 117 of 118 animals through 15 June 1995. All 6 adult females collared 2 March 1995 were alive as of 15 June 1995 but were not used in calculations of survival for time periods prior to 14 June 1995. Fat content (percent dry matter) of femur marrow and estimates of age based on dental cementum were obtained for dead radioed elk (Colo. Div. Wildl. Lab, Ft. Collins). Sighting Bias V V Procedures for conducting aerial sighting trials were identical in 1994 and 1995 (Freddy 1994). In 1995, we conducted 103 sighting bias trials that targeted radiocollared elk on 23, 24, 26, 28, and 30 January and 6-8 and 21 February. We again used a Bell-Soley helicopter and the same observers and navigators as in 1994. The pilot which flew all trials in 1994 flew 77 (75%) of the trials in 1995 with 2 additional pilots flying the remaining 26 trials. Observers indicated that all pilots provided comparable and acceptable service. Sighting bias models were developed using logistic regression (Proc GENMOD, SAS). The dichotomous classification of groups seen or missed was the dependent variable. Initial group size (ln), total group size (ln), activity behavior, vegetation type, percent occlusion cover, percent snow cover, snow type, year, observer, navigator, trap method, temperature, wind, lighting conditions, pilot, time interval, and time of day were independent variables. Observer, navigator, pilot, trap method, elk sex, elk age, activity, vegetation type, snow type, time interval, time of day, light conditions, and wind were treated as class variables and group size, percent occlusion cover, and percent snow cover were treated as continuous variables. Significance level for independent variables was P ~ 0.06 for univariate tests. Relative efficiency of models using all possible combinations of significant variables was assessed using AIC values (Akaike Information criterion= -2(-log likelihood value)+ 2[No. model parameters)). Population Estimates Elk population size in GMU 42 from East Alkali Creek to Beaver Creek was estimated during January and February 1995. Three independent estimates of size or density were obtained. On 19 January, a nonrandom elk sex/age V V �67 classification flight using a helicopter was flown during which numbers of marked (radiocollared) and unmarked elk were counted. From 22-24 February, 40 randomly selected quadrats approximately 1 mi2 (2.59 km2) in size were flown using a helicopter and during these flights numbers of marked and unmarked elk were counted. For these 2 flights, we used a simple Lincoln-Peterson estimator based on ratios of marked and unmarked elk (White and Garrott 1990) to estimate total numbers of elk. The third estimate of population size resulted from projecting the average number of elk counted per sample quadrat to the entire segment of winter range sampled by the quadrat system. The number of elk counted by direct observation at the time of the flights that were on private land areas not surveyed by the 3 flights were added to each of the 3 estimates of population size to arrive at an estimate of total elk in GMU 42 east of Beaver Creek. At this time, no adjustments have been make for sighting bias for counts of elk on quadrats. Movements We continued to locate 37 radioed elk at least once per month since capture to document seasonal movements via telemetry using a Cessna 185. These elk were selected at random from within trap zones and equalized by age class. As of 14 June 1995 these elk were classified as 15 adult females, 7 yearling females, 4 female calves, 7 yearling males, and 4 male calves. During 199394, 6 of 34 elk monitored for locations died. These 6 were replaced at random primarily with 6 month-old calves captured in December 1994 in the same trap zone(s) as those elk that died. During June 1995, we selected an additional 28 adult females at random to document locations during the calving period. As needed, we located other elk to document unusual movements. RESULTS AND DISCUSSION Survival ~ Between 1 December 1993 and 14 June 1995, 37 of 190 radiocollared elk died (Appendix 1). Of these 37, hunting was apparently involved in 571 of the deaths. For adults ~1 year-old, hunting accounted for as, of 24 deaths. There were 2 periods of mortality during the year. Calves died from February to May while adults died during fall and early winter when hunting seasons occurred (Fig. 1). Survival rates(± 951 CI) for calves during winter, 1 December - 14 June, were 0.92 ± 0.06 in 1993-94 and 0.90 ± 0.07 in 1994-95 (Table 3). Sex of dead calves was 4 male and 2 female in 1993-94 and 4 females and 3 males in 1994-95 (Table 2). These survival rates were associated with winters considered mild in temperature and having low or moderate snow depths. In 1994-95, considerable snow fell during March and April but usually melted rapidly at lower elevations. ~ Suspected causes of death for calves during winter were, mountain lion predation (311), malnutrition (231), and unknown (461) (Table 4). For the unknown category, bear predation was suspected once (3/1/95) as was lion predation (3/18/94) (Table 2). Those calves dying from malnutrition had marrow fat values <131 at death and body weights <102 kg at capture while calves likely dying from predation had marrow fat values 3.7-35.61 and body weights 86-140 kg (Table 2). There were no calf mortalities during capture and no evidence that capture procedures directly induced mortality in either year. In 1993-94 the first mortality involved an 82 kg female suspected of dying from malnutrition at 59 days post-capture while in 1994-95 the first mortality was an 86 kg female suspected of dying from mountain lion predation at 72 days post-capture. At capture, this last female had an abnormally articulating front shoulder which may have been injured during or prior to capture but the injury appeared to minimally affect her mobility upon release at capture. �68 Survival rates(± 951 CI) for adult females during winter, 1 December - 14 June, were 0.96 ± 0.05.in 1993-94 and 0.96 ± 0.04 in 1994-95 (Table 3). Of 7 winter deaths during both years, 5 (711) were due to legal (1) and illegal (l) hunting or wounding loss (3) during late rifle seasons (Table 4). Natural deaths (2) were attributed to mountain lion predation and unknown but possible lion predation. Of the 3 elk lost to wounding, 2 were suspicious kills. Both of these elk died from 1 rifle shot to the upper neck or head in relatively plain sight near or on an agricultural field where retrieval of the carcass was not difficult. These animals may have been accidentally shot, maliciously shot, or abandoned because of the radiocollar. Survival rates during winter 1994-95 were 1.00 for juvenile males and females age 18-23 months and radiocollared as calves (Table 3). V u Annual survival rate for adult females, 1 December 1993 - 30 November 1994, was 0.77 ± 0.10 (Table 3). Of the 15 deaths involving adult females marked in December 1993, 13 (87%) were associated with hunting. The 2 natural deaths were due to lion predation during winter and an unusual breech birth that occurred about 1 October 1994. Hunting deaths were attributed to legal hunting (6, fall seasons), illegal hunting (1, late season), wounding loss (5, 4 fall seasons, 1 late season), and presumed hunting (1, fall seasons). We examined all 5 carcasses attributed to wounding loss. Based on locations and physical position of the carcasses, we found no compelling evidence that hunters had likely found the animal and then left the carcass because it was radiocollared. Thus, we do not believe at this time that wounding loss during fall seasons is resulting from hunters being afraid to claim a radiocollared animal. Depending on the type of hunting season and GMU, yearling males with spike antlers are not legal quarry. In general, yearling males are not legal quarry in any part of the study area until the third combined rifle season in late October. In August 1994, there were 32 yearling males alive that were radiocollared as calves. Life/death status of all 32 was determined through 14 June 1995. Of these 32 yearlings, 4 (12.51) were known or presumed to be illegally taken during hunting seasons. Three carcasses were recovered in the following general condition: shot, eviscerated, and left in the field (1st rifle season); shot, not eviscerated, and left in the field (2nd rifle season); shot, eviscerated, head removed, partially packed-out, and radiocollar buried under debris (3rd rifle season). The telemetry signal of the fourth animal, presumed shot and removed, could not be heard after 4 October 1994. V Calf Body Size Male calves had larger body weights (P = 0.001), longer total body length (P = 0.06), longer hind leg lengths (P = 0.06), and higher condition indexes (P = 0.06) than female calves (Tables 5, 6). Body weights and measurements were not different between years for male or female calves (P > 0.30). Predicting body weight from body measurements does not look promising at this time, although all regressions were significant (P < 0.001). The multiple regression using body length and hindfoot length as independent variables provided the best correlation coefficients (.r) of 0.59-0.67. Sighting Bias In 1994 and 1995, we conducted 213 sighting trials of which 192 (90%) were successful, and each observer was involved in 70-72 sighting trials during both years (Table 7). Successful trials occurred when the targeted elk was on the assigned flight quadrat whether or not that elk was seen by observers. Unsuccessful trials occurred when the target elk was found off the assigned quadrat immediately subsequent to the time the quadrat was flown by the helicopter team. Elk were off quadrats primarily due to their movements between the time elk were located with the Cessna 185 and the time the helicopter team flew the quadrat. Usually only 1 elk was targeted per quadrat even though several radiocollared elk may have been present on a quadrat to insure that targeted elk were in separate and independent groups. V V �69 During the 2 years, 143 different elk were used in 192 successful trials (Table 8). Most elk were targeted only once during the 2 years. Targeted elk included adult females, yearling females, female calves, yearling males (spike-antlered), and male calves. Calves comprised 53-54% of the targeted elk within each year. Yearling males were only available as targets during 1995 (Table 9). We used as many different radiocollared elk as possible to avoid any positive effects on sightability that might have occurred if an observer or navigator had been involved in locating a target elk more than once. Observers were not assigned the same target- elk within the same year, except in 1995 an observer was given the same elk twice but the animal was on 2 different search quadrats. During both years, there were 11 cases where a target elk was given to the same observer for 2 trials. In 7 of these cases, the targeted elk was a calf the first year and a yearling the second year, while in the 4 remaining cases, the target elk was an adult in both years. In all 11 instances, elk were on different search quadrats in both years. During both years, there were 24 cases where a navigator was involved with the same target elk during 2 trials. Only 3 of these cases occurred in the same year and the targeted elk was on a different search quadrat on each occasion. Average sighting probability was 82.3% which equates to a 17.7% negative sighting bias (Table 7). There were no differences in sighting bias between years (P > 0.70) but there were differences in sighting bias among observers (P = 0.06) which ranged from 9.3% to 22.4% (Table 7). Univariate tests indicated elk age (calf or adult), elk sex (calves only) initial group size, log(ln) initial group size, total group size, log(ln) total group size, activity behavior, vegetation type, percent occlusion cover, observer, navigator, pilot, and trapping method affected the probability of sighting elk (P ~ 0.06, Tables 11, 12). Adult females were more sightable than calves (P = 0.03) as were male calves compared to female calves (P = 0.06). Elk caught in corral traps were more sightable than those caught with a helicopter and this effect was primarily associated with adult females (P = 0.05), not calves. Multivariate analyses incorporated elk age, ln total group size, activity behavior, vegetation type, percent vegetation occlusion cover, percent snow cover, navigator, observer, and year. According to AIC criteria, the best sightability model out of 2,048 possible models was a complex 12 parameter model involving nearly all major variables. Ln group size, percent vegetation cover, and observer persisted as significant variables as number of parameters was reduced (Table 13). Models incorporating elk age, observer, and navigator are likely not practical for field application. We are currently comparing these 12 best fitting models to assess what effects each model has on predicting sightability of our known population of elk groups. This process will hopefully produce a less complex sighting model. When the same variables are used to build sighting models, results from Idaho (Samuel et al. 1987) and Colorado strongly suggest there are common factors affecting sightability of elk in broadly different landscapes. For coniferous landscapes in Idaho the best elk sightability model was: Y = 1.22 + l.55(LnGroupSize) - 0.05(% vegetation cover). For oakbrush landscapes in Colorado, elk sightability was: Y = 1.23 + l.08(LnGroupSize) - 0.03(% vegetation cover). Effects on sightability of group size and vegetation cover are shown in Fig. 2. Unfortunately at this time, this 3 parameter model ranked 1,165 out of 2,048 Colorado models. The most efficient 3 parameter model incorporated elk activity instead of vegetation cover (Table 13). Group size is a variable that must be incorporated into any sighting model to correct for the propensity to miss groups containing <7 elk (Fig. 2). our data indicates that groups of elk observed but not targeted for sighting trials were smaller than our target groups (P =0.02). This indicates our radiocollared elk groups biasly represented all possible groups of elk which �70 V likely reflects a trapping bias. This data however, also strongly suggests that the need to correct for group size will be greater in actual surveys. Population Estimates Initial estimates of elk population size ranged from 2,049 to 4,339 and were generally imprecise (Table 10). For mark-resight estimators, the nonrandom flight provided the most precise estimate(± 271) because more unmarked and marked elk were observed compared to the quadrat flight(± 47% precision). However, both mark-resight flights provided estimates of similar magnitude: 3,810 and 4,339 elk. The lower quadrat estimate of 2,049 elk probably reflects a need to restratify and reallocate sample units. Precision of the quadrat sample estimate can likely be improved by altering strata, increasing sample size, and increasing the quadrats allocated to high density strata. The imprecision of both the quadrat and quadrat mark-resight estimates(+ 471) indicate more quadrats need to be flown along with refining strata boundaries. Count data indicates that major changes are needed in delineating the Garfield High and East Divide creek low strata. Of the 15 (38%) quadrats where no elk were counted, 8 occurred in these 2 strata. Consideration will be given to assigning each individual quadrat to high and low strata even if quadrats are adjacent to each other. Elk Movements As of 15 April 1995, 22 radioed elk had dispersed to a winter range outside the winter range in GMO 42 where they had originally been trapped in December 1993 (Table 14). Of 35 females collared as calves in December 1993 and recruited into the population as yearlings in June 1994, 8 (231) dispersed during summer-fall 1994 to another winter range. Likewise, of 32 male calves recruited as yearlings in June 1994, 6 (191) dispersed during summer-fall to another winter range. Eight (12%) of 65 adult females alive in June 1994 dispersed during summer-fall to another winter range. Seven (321) of the dispersing elk moved to areas outside of DAU E-14, with 6 of these elk moving into GMO 43 immediately east of DAU E-14 (Table 14). The longest dispersal movement involved an adult female elk (172.480/93) that moved to Beaver Creek west of Gunnison, a straight line distance of about 80 miles. This elk had been neckbanded as a young adult in Beaver Creek in 1991 (GMO 54), then radiocollared in GMO 42 in 1993, and subsequently returned to the drainage where it was originally marked in January 1994 (visual and telemetry location). As of June 1995, this elk had returned to GMO 42. We are currently creating a GIS land database for the area frequented by radiocollared elk. This is a volunteer cooperative effort through the National Biological Service. We believe plots of elk locations and movements using this database will be forthcoming in 1996. CONCLUSIONS We obtained acceptably precise estimates of calf and adult survival rates during winter and recommend continuing our current sampling effort of monitoring 75 radioed calves and >75 radioed adult females in 1995-96. Efforts to measure and develop criteria to adjust for negative bias in counts of elk were promising and we are currently assessing the relative efficiency of several sighting models. We recommend conducting the planned replicate surveys using sample quadrats to estimate elk density from mark-resight estimators and sighting bias correction models in 1995-96. V �71 LITERATURE CITED Freddy, D. J. 1993. Estimating survival rates of elk and developing techniques to estimate population size. Colo. Div. Wildl. Game Res. Rep. July: 83-117. Freddy, D. J. 1994. Estimating survival rates of elk and developing techniques to estimate population size. Colo. Div. Wildl. Game Res. Rep. July: 27-42. Samuel, M. D., E. o. Garton, M. w. Schlegel, and R. G. Carson. 1987. Visibility bias during aerial surveys of elk in northcentral Idaho. Wildl. Manage. 51:622-630. SAS Institute Inc. 1988. SAS/STAT User's Guide, 6.03. SAS Institute, Inc., Cary, NC. 1028pp. White, G. c., and R. A. Garrott. 1990. Analysis of wildlife radio-tracking data. Academic Press, Inc., San Diego. 383pp. Prepared by ___________________ David J. Freddy Life/Science Researcher J. �Table 1. Capture objectives and numbers of elk radioed in 8 trapzones, December 1993 and December 1994 1 GMU 42. Objectives and elk ca~tured for 1994 are in ~arentheses. Elk Collared8 Capture Calves Adult Trap Objective Total Helio Corral Males Females Females Zone Name 8 ( 6) 3 ( 3) Garfield 0( 0) 8( 6) 1( 3) 8( 5) 4( 0) A Gibson 25(17) 19 ( 7) 6.(10) 20( 8) 5( 5) 8( 6) 12 ( 6) B Uncle Bob 24(13) 29(13) 29(13) 0( 0) 10( 7) 7 ( 6) C 12( 0) 26(13) 0( 0) 5 ( 1) 21(11) 21(11) West Divide D 6( 8) 10( 2) 17(17) 17(17) Hightower 0( 0) 10(10) 4( 9) 3( 8) 10( 0) E 0(13) 0(13) Middle Mamm 10( 8) 0( 0) 0( 8) 0( 5) F 0( 0) 8( 8) 2 ( 0) 3 ( 0) West Mamm 6( 0) 6( 0) 0( 0) G 1( 0) 35 ( 6 )b 0( 0) 35( 6) Dry Hollow 44(21) 7 ( 0) 9( 0) 19( 6) H 141(83) 100(67) 41(16) 36(33) 37(36) 150(86) 68(14) All a Belio b All ~ = Helicopter net-gunning, corral= corral-trap. 6 adult females captured 2 March 1995 Table 2. Causes of mortality and body condition for radioed calf elk in GMU 42, 1 December 1993 - 14 June 1995. Date Estimated Body Marrow Fat Radiocollar Dead Cause of Death Sex Age8 Freggency Wt. 'kg}b Percent Drl!: Matter 172.899/93 173.000/93 173.262/93 173.289/93 173.461/93 173.469/93 173.870/94 174.119/94 174.140/94 173.789/94 173.640/94 174.170/94 173.589/95 F F M M M M F M M F F M F 9 mos 10 mos 9 mos 9 mos 8 mos 10 mos 8 mos 9 mos 9 mos 9 mos 9 mos 10 mos 12 mos 3/18/94 4/25/94 3/18/94 3/22/94 2/07/94 4/25/94 2/21/95 3/01/95 3/14/95 3/20/95 3/30/95 4/25/95 6/01/95 Mt.Lion kill Malnutrition Unknown, Predation? Unknown Malnutrition Malnutrition Mt.Lion kill Unknown, Predation? Mt.Lion kill Unknown Unknown Mt.Lion kill Unknown 86 102 108 111 82 28.54 12.6c 28.9 4 not available 2.lc 77 2. 7c 86 140 112 100 89 140 117 not available 3. 7c 35.6c 76.2c not available 17.4c not available a Approximate age at death, assume 15 June birthdate. body weight at capture in December 1993 or 1994. c Fat content of either the right or left femur bone marrow at death. b Whole d Fat C C content of lower jaw bone marrow at death. CC C C �( ( ( ( ( ( Table 3. Survival rates of radiocollared elk in GMU 42 for different age and sex classes during 4 time periods from 1 December 1993 through 14 June 1995. Survival rates calculated as a simple binomial of elk alive at end of time period divided by elk alive at beginning of time period with a variance of S(l-S)/n. Calves were 6-12 months old and collared at 6 months of age. Yearlings were 12 - 18 months old and juveniles were 18 - 23 months old and both were collared as 6 month old calves in December 1993. Adult females were ~l year-old when captured or recruited in December. Time Period Age/Sex Class 1 DEC 93 - 14 JUN 94 A8 0 6 Surv. Ratec Calves 73 6 15 JUN 94 - 30 NOV 94 A D Surv. Rate 1 DEC 93 - NOV 30 94 A D Surv. Rate 0.92 ± 0.06 1 Dec 94 - 14 JUN 95 A D Surv. Rate 68 7 0.90 ± 0.07 28 34f 0 O -1.00 1.00 94 4 0.96 ± 0.04 190 11 M + F Yearlings Male Female 32 35 4d le 0.88 ± 0.12 0.97 ± 0.06 Juveniles Male Female Adult Females 68 3 Totals 141 9 0.96 ± 0.05 648 12h 131 17 0.81 ± 0.15 67 15 67 15 0.77 ± 0.10 Number of radiocollared elk alive at beginning of time period. radiocollared elk dying during time period. c Survival rate± 95% confidence interval. d One elk (173.309/93) disappeared during October hunting seasons and presumed dead. e One elk (172.800/93) disappeared during November hunting seasons and presumed dead. f Yearling females are included in adult female survival rates during this time period. 8 One elk (172.011/93) censored from calculations because animal was missing. h One elk (172.649/93) disappeared during October hunting seasons and presumed dead. 8 b Number of �74 Table 4. Causes of deaths in radiocollared elk from GMO 42 between 1 December 1993 and 14 June 1995. Calves were 6-12 months old and collared at 6 months of age. Yearling males and females were 12-18 months old and collared as 6 month old calves. Juvenile males and females were 18-23 months old and collared as 6 month old calves. Adult females were ~l year-old when collared or recruited in December. Cause of Death Calves Malnutrition Predation-Lion Accident Legal Bunting Archery/Muzzle Rifle Rifle/Late Wounding Loss Archery/Muzzle Rifle Rifle/Late Illegal Bunting Presumed Bunting Unknown 3 4 0 0 Total Yearl. Males 0 0 0 0 0 0 0 0 0 0 0 6 13 4 0 0 0 1 0 0 0 0 0 Total 1 7 7 2 4 1 7 0 0 0 0 0 0 V 3 5 1 1 0 0 0 0 0 0 0 0 0 Adult Females 0 0 0 0 0 0 0 3 1a 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Elk Age/Sex Class Yearl. Juv. Juv. Females Males Females V 7 1 3 3 4 3 7 19 37 Elk (173.309/93) was a spike-antlered male that disappeared during October hunting seasons and presumed dead and illegally shot. b Elk (172.800/93) disappeared during November hunting season and is a presumed hunting mortality. c Elk (172.649/93) disappeared during October hunting seasons and is a presumed hunting mortality. a V �( ( ( ( ( Table 5. Frequency distribution of whole body weights for male and female elk calves trapped in Game · Management Unit 42. December, 1993 and 1994. Percentage per weight class shown in parentheses. Year Sex 70-79 80-89 90-99 Body Weight Class (kg) 100-109 120-129 110-119 130-139 140-149 Total 1993 1994 M M 1(2.9) 1(3.0) 1(2.9) 1(3.0) 3(8.6) 2(6.0) 6(17.1) 6(18.2) 12(34.3) 13(39.4) 10(28.6) 6(18.2) 1(2.9) 2(6.0) 1(2.9) 2(6.0) 35(100) 33(100) 1993 1994 F F 1(2.9) 2(5.7) 4(11.4) 4(11.4) 8(22.9) 7(20.0) 12(34.3) 10(28.6) 4(11.4) 10(28.6) 6(17.1) 2( 5.7) 0(0.0) 0(0.0) 0(0.0) 0(0.0) 35(100) 35(100) Table 6. Body measurements for elk calves trapped in Game Management Unit 42. December, 1993. 1994. SD 112.7 191.2 Body Length (cm) Hindfoot Length (cm) 56.3 Condition Index 0.59 (Wt./Body Length) 13.7 10.2 2.2 14.2 Measurement 1993 Body Weight (kg) 1994 Male Calves Min Max Mean Year Body Weight (kg) 113.5 190.3 Body Length (cm) Bindfoot Length (cm) 56.9 Condition Index 0.59 (Wt./Body Length) o.os 9.2 1.8 o.os Female Calves Min Max n Mean SD 77.0 164.0 52.0 0.43 141.0 35 210.0 36 61.0 36 0.67 35 103.8 188.9 54.8 12.2 9.8 2.1 76.0 167.0 51.0 0.46 123.0 35 207.0 35 59.0 34 0.64 34 71.0 168.0 140.0 33 204.0 32 59.0 32 0.69 32 103.0 186.3 55.1 13.3 70.0 166.0 128.0 35 207.0 36 59.0 36 0.65 35 so.a 0.42 o.ss o.os o.ss a.a 2.1 0.06 so.a 0.42 n ( �76 V Table 7. Summary of elk sighting bias trials conducted in GMU 42 during 1994 and 1995. Trials Target Target Trials Not Trials Elk Elk Not Observer Yr lttem;eted Useable Detected Useable Detected Broderick Ellenberger Masden Totals 94 95 94-95 94 95 94-95 94 95 94-95 94 95 94-95 29 42 71 36 36 72 45 25 70 110 103 213 3 7 10 4 1 5 4 2 6 11 10 21 26(90%) 35(90%) 61(90%) 32(89%) 35(97%) 67(93%) 41(91%) 23(92%) 64(91%) 99(90%) 93(90%) 192(90%) 22 27 49 25 27 49 36 22 58 83 76 159 ------ V 4(15.4%) 8(22.9%) 12(19.&%) 7(21.9%) 8(22.9%) 12(19.7%) 5(12.2%) 1( 4.3%) 6( 9.3%) 17(17.2%) 17(18.3%) 34(17.7%) Table 8. Numbers and frequency of individual elk targeted during successful sighting bias trials in GMU 42. 1994 and 1995. No. No. Elk No. Elk No. Elk Elk Used in Used in Used in Total Year Used 1 Trial 2 Trials 3 Trials Trials 1994 1995 1994-95 77 90 143 67 87 101 16 3 35 99 93 192 0 0 7 V Table 9. Age and sex of elk targeted during successful sighting bias trials in GMU 42. 1994 and 1995. Female Male Year Adult Yearling Calf Calf Total Yearling 1994 39(39%) 1995 20(21%) 1994-95 59(31%) 6( 6%) 28(28%) 10(11%) 26(28%) 16( 8%) 54(28%) 0( 0%) 15(16%) 15( 8%) 26(26%) 22(24%) 48(25%) 99(100%) 93(100%) 192(100%) Table 10. Summary of population estimates for elk in GMU 42, January and February. 1995. Estimate Private Total Densit! Land Elkb Elk/mi c Method Estimate Date + 95% CI8 Mark-Resight +399 3,810 3,411±27% 14 (267 mi2 ) 1/19/95 NonRandom Flight +565 4,339 Mark-Resight 3,744±47% 19 (234 mi2 ) 2/24/95 Random Quadrats Flight Random Quadrats +565 2,049 1,484±47% 9 (234 mi2 ) 2/24/95 Flight 8 Number of elk estimated In survey area actually flown. b Elk directly counted on private land area not included in survey flight. c Density for area represented by survey flight area plus private land area. V V �( ( ( ( ( Table 11. Elk sightability survey results by major independent variables from Game Management Unit 42, Januaa-March, 1994 and 1995. No. Grou:gs Missed Seen Variable ya Variable 0.52 0.63 Behavior Bedded standing Moving Group Size 12 9 2 5 0 1 4 1 2 3 4 5 6 7+ Vega. Type Clearing/Ag Riparian Sagebrush Oakbrush Pinyan-Juniper Aspen Tall Conifer 13 15 21 14 15 12 69 0.91 0.74 1.00 0.92 0.94 JE 0 0 1 14 9 7 2 13 1 6 89 42 7 1 1.00 1.00 0.86 0.86 0.82 o.so 0.33 Occl. Cover(\) 0-19 20-39 40-59 60-79 80-100 2 11 10 6 4 4 V=visibility • Navigator GB VG FB Observer JB 35 60 43 18 3 0.95 0.85 0.81 0.75 0.43 DM Elk Age Adult Female Calf (M + F) Yearling Calf Sex Male Female Trap (All elk) Corral Trap Helicopter Trap (calves) Corral Trap Helicopter No. Grou:12s Missed Seen Variable Missed Seen Snow Cover (%) 9 4 20 6 41 112 0.40 0.91 0.85 0-19 20-49 50-99 100 3 5 5 20 9 9 47 94 0.75 0.64 22 11 0 68 81 10 0.76 0.88 1.00 Snow Type Fresh Old 5 28 35 124 0.88 12 15 6 49 52 58 o.80 0.78 0.91 Wind Light Moderate Strong 32 1 0 137 20 2 0.81 0.95 1.00 4 21 8 55 81 23 0.93 0.79 0.74 Light Bright Dull Hazy 22 3 8 97 24 38 0.82 0.89 0.83 6 15 42 39 0.88 0.72 32 0 1 139 10 10 0.81 1.00 0.91 9 24 61 98 0.87 o.80 4 29 45 114 0.92 0.80 3 18 14 67 0.82 0.78 Pilot ET BB pp 0.90 0.82 0.82 Time (groups seen I groups seen plus gr~ups missed). AM PM ( �78 Table 12. Variables tested in univariate tests using logistic regression for elk sightability trials, Game Management Unit 42, January-March, 1994 and 1995. Significance level at P < 0.06. Variable Significant Date of Trial Sex of Elk Calves Age of Elk Initial Group Size Log(ln) Initial Group Size Total Group Size Log(ln) Total Group Size Elk Activity Vegetation Type Occlusion Cover % Observer. Navigator Pilot Snow Cover % Snow Type Temperature Wind Conditions Light Conditions Time Interval Time of Day Trap Type No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No No No No No No No Yes Sign. Year Effect Sign. Year Interaction No No No No No No No No No No No No Yes No No No No No No No No Yes No No No No No No No No No No Yes No No No No No No Yes Table 13. The best fitting sightability model at each level of 1 to 12 parameters according to Akaike Information Criterion (AIC). Rank is the relative rank of each model among 2,048 models which used variables in all possible combinations. "Yes" denotes variables used in models. Variables In Models No. Ln Elk Group Elk Vege. %Vege. %Snow AIC ParamObs. Cover Cover Nav. Year eters Age Size Act. Tv1>8 Rank 1 2 7 4 48 61 152 463 869 1019 1403 1890 2028 12 11 10 9 8 7 6 5 4 3 2 2 1a Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes V V V Yes Yes Yes Yes Yes Yes Yes Yes Yes a The one parameter model includes only the intercept which equates to the average sightability of elk which was 0.823. V �79 V ·~ Table 14. Radiocollared elk captured in GMU 42 during Deceri>er 1993 and thought to have dispersed out of GMU 42 to a new winter ranse as of 15 Aeril 1995. Locations ,GMUs2 determined~ aerial telemetrx. New Trap Location Descrietion Sex GMU Zone Age0 Freauencx 3+ Fourmile Ck (Loe.Elk) F 43 D 172.207/93 3+ 421,42 Beaver-Buzzard Cks, Alkali Ck. (Loe.Elk) F 172.219/93 D 5+ Ees t Muddy Ck F 521 C 172.249/93 2+ Middle Th0[1')Son Ck F 172.308/93 D 43 3+ Beaver Ck, Gumisonb F 54 172.480/93 H 5+ N.Fk.Gwv,ison, Near Paonia 172.459/93 F 521 E 2+ Roaring Fk R., Glenwood Sprgs. (Loe.Elk) 172.509/93 F 43 H 1+ F 421 Hawxhurst Ck 172.830/93 B 1+ Threemile Ck, Roaring Fk. River F 172.849/93 A 43 1+ East Muddy Ck, Paonia Reservoir F 172.921/93 C 521 1+ Redstone; Porcupine Ck F 172.929/93 C 43 1+ East Muddy Ck, Paonia Reservoir 172.961/93 F D 521 1+ Hawxhurst Ck, Brush Ck 173.010/93 F 421 G 1+ 173.020/93 F 421 Hawxhurst Ck, Brush CK G 2+ 173.070/93 F 42,West Parechute;Monunent Gulch E 1+ 173.090/93 F Roering Fk River, Crystal Ranch H 43 1+ 173.279/93 East Muddy Ck, Paonia Reservoir C M 521 1+ 173.300/93 C M 521 East Muddy Ck 1+ 173.332/93 Clover Gulch, Hawxhurst Ck D M 421 1+ 173.370/93 E M 421 Plateau Ck/Salt Ck 1+ 173.390/93 West Muddy Ck, Pilot Knob D M 521 1+ 173.450/93 Kimball Ck, Wallace Ck M 421 H 6 Age of elk in years as of April 1995 b This elk returned to GMU 42 during June 1995, a movement distance of about 80 airline miles. ~ \.._,,,I ~ \,,,..,/' Appendix 1. Mortalities of radiocollared elk from 1 Deceri>er 1993 - 14 June 1995. Age is approximate age in years of elk at death; C=calf 6·12 months old, Y=yearling 12-18 months old. Body weight measured in Deceni>er when eLk caetured as calves. Body Frequency ID/ Date Tree Cause of Death Year caetured Site Zone Sex Age Wt.,kg} Heard Dead 172.039/93 Undetermined mortality GR B F 2+ 06/14/95 172.080/93 GM A F 5+ 11/05/94 Legal harvest rifle season 172.090/93 F 5+ 01/16/94 Wounding loss late rifle season GR B 172.160/93 Legal harvest rifle season cc C F 2+ 10/23/94 172.181/93 F 2+ MC C 11/03/94 Wounding loss rifle season 172.201/93 GS C Wounding loss rifle season F 2+ 11/15/94 HY C F 2+ 10/04/94 Legal harvest archery/nuzzle season 172.258/93 172.277/93 F 2+ GS C 10/04/94 Wounding loss archery/nuzzle season F 5+ 172.290/93 SG 10/23/94 Legal harvest rifle season 172.369/93 AC F 2+ 09/18/94 Legal harvest archery/nuzzle season 172.409/93 AC F 5+ 12/29/94 WOWlding loss late rifle season 172.542/93 FS F 9+ 02/01/94 Mountain lion predation 172.570/93 F 9+ 11/03/94 WOWlding loss rifle season PG 172.581/93 F 2+ Legal harvest rifle sea·son PG 10/17/94 172.610/93 F 2+ PG 10/04/94 Accident, breech birth of calf 172.649/93 PG F Y+ 11/15/94 Disappear rifle season F 2+ 172.670/93 01/16/95 Legal harvest late rifle season PG 172.678/93 F 9+ 12/22/94 Wounding loss late rifle season PG FM F 5+ 172.690/93 01/24/94 Illegal harvest, poached 172.800/93 SR F Y+ 11/15/94 Disappear rifle season 172.899/93 BC F C 86.0 03/18/94 Mountain lion predation 173.000/93 WM G F C 102.0 04/25/94 Malnutrition M y 173.190/93 BC C Illegal harvest rifle season 10/29/94 111.0 BR A 173.232/93 M y 105.0 11/15/94 Illegal harvest rifle season 173.262/93 BC C M C 108.0 03/18/94 Undetermined mortality 173.289/93 MC C M C 111.0 03/22/94 Undetermined mortality 173.309/93 HY C M y 11/15/94 124.0 Disappear rifle season 173.320/93 HY C M y 118.0 10/20/94 Illegal harvest rifle season 173.461/93 M C 02/07/94 Malnutrition PG H 82.0 173.469/93 FS H M C 04/25/94 Malnutrition 77.0 173.589/94 0G B F C 117.0 06/01/95 Undetermined mortality 173.640/94 MC C F C 89.0 03/30/95 Undetermined mortality 173.789/94 F C MR E 100.0 03/20/95 Undetermined mortality 173.870/94 F C 86.0 02/21/95 SM D Mountain lion predation 174.119/94 F M C 03/01/95 MM 140.0 Undetermined mortality 174.140/94 M C F 112.0 03/14/95 MM Mountain lion predation 174.170/94 FM B M C 04/25/95 140.0 Mountain lion predation �87 Colorado Division of Wildlife Wildlife Research Report July 1996 JOB PROGRESS REPORT State of Colorado Project No. W-153-R-9 work Plan No. Mammals Research _..>L,.__________ Job No. _ _ _ _.....,.____________ Elk Investigations Estimating Survival Rates of Elk and Developing Techniques to Estimate Population size Period Covered: Author: July 1, 1995 - June 30, 1996 D. J. Freddy Personnel: J. Broderick, G. Byrne, A. Coriell, D. crane, J. Ellenberger, D. Fox, J . Frothingham, v. Graham, J. Gray, G. Loucks, D. Masden, C. Mehaffy, J. Ritchie, L. Stevens, P. Will, CDOW; D. Bowden, G. White, CSU; CSU Diagnostic Laboratory; K. Crane, c. McCarty, N. Miers, D. Ouren, A. Ryel, volunteers; BLM Glenwood Springs, NBS Ft. Collins, Rocky Mountain Elk Foundation, USFS Rifle, cooperating. ABSTRACT We captured and radio-collared 71 calf elk (Cervus elaphus nelsoni) 6-months old in December 1995 to estimate survival rates during winter 1995-96 and to increase numbers of radiocollared elk available for experiments utilizing mark-resight models to estimate population density. Elk were captured using helicopter net-gunning and portable corral traps. Survival rates(± 95% CI) for 6-11 month old calves during winter-spring were 0 . 92 ± 0.06, 0.90 ± 0.07, and 0.88 ± 0.08 in 1993-94, 1994-95, and 1995-96, respectively. Survival was similar among years (P > 0.50) and sexes (P > 0.70). Suspected primary causes of death for calves were predation (57%) and malnutrition (33%). Survival rates for adult females(~ 12 months old) during winter-spring were 0 . 96 ± 0 . 05, 0.96 ± 0.04, and 0.95 ± 0.04 also during 1993-94, 1994-95, and 1995-96 and survival was similar among years (P > 0.70). Hunting was the primary cause of death (57%) for adult females during winter-spring. Annual survival rates (l December-30 November) for adult females were 0.78 ± 0 . 10 and 0.84 ± 0.09 in 1993- 94 and 1994-95, respectively. Survival was similar between years (P > 0.50) and hunting annually accounted for 88% of the adult female deaths. Hunting removed 79 % of the adult 2-year old males. Survival of males from 6month old calf to 35-month old adult were 0.09 ± 0.10 compared to 0.86 ± 0.12 for females of the same cohort. Wounding loss on adult males was 9% and illegal loss of yearling males averaged 10%. Estimates of population size based on mark-resight estimators ranged from 3, 2 72 -3,618 elk or about 26 elk/mi2 of winter range. The IEJHE mark-resight model using data pooled from 1~~ 1ffil1~ii11 ~~1ffif BDOW01 □ 789 �4 aeria l flights provided the best estimate of population size which was 3,415 elk wit h a 95% CI of 3,129-3,807. Estimates based on counts of elk on randoml y selected 1 mi2-quadrats during 2 replicate flights were 2,165 and 3,175 e lk. Sighting bias models to adjust for negative bias of quadrat counts are cur rently being evaluated. We believe stratified sampling systems using quadrat s as sample plots and with counts corrected for sighting bias holds promise for providing reasonable estimates of population size. �:~JXf._ li10,)fh f ..:.1 :. 87 ... "--~ .:/{ \ .. ' / Colorado Division of Wildlife Wildlife Research Report July 1996 JOB PROGRESS REPOR~ state of Colorado Project No. W-153-R-9 Mammals Research Elk Investigations Work Plan No. Job No. ____ __________ __.._ : Estimating Survival Rates of Elk and Developing Techniques to Estimate Population size Period Covered: Author: ;~~~ :;;;tt·\: ( ~~ July 1, 1995 - June 30, 1996 D. J. Freddy Personnel: J. Broderick, G. Byrne, A. Coriell, D. Crane, J. Ellenberger, D. Fox, J. Frothingham, v. Graham, J. Gray, G. Loucks, D. Masden, C. Mehaffy, J. Ritchie, L. Stevens, P. Will, CDOW; D. Bowden, G. White, CSU; CSU Diagnostic Laboratory; K. Crane, c. McCarty, N. Miers, D. Ouren, A. Ryel, volunteers; BLM Glenwood Springs, NBS Ft. Collins, Rocky Mountain Elk Foundation, USFS Rifle, cooperating. ABSTRACT We captured and radio-collared 71 calf elk (Cervus elaphus nelsoni) 6-months. old in December 1995 to estimate survival rates during winter 1995-96 and to increase numbers of radiocollared elk available for experiments utilizing mark-resight models to estimate population density. Elk were captured using helicopter net-gunning and portable corral traps. Survival rates(± 95% CI) for 6-11 month old calves during winter-spring were 0.92 ± 0.06, 0.90 ± 0.07, and 0.88 ± 0.08 in 1993-94, 1994-95, and 1995-96, respectively. Survival was similar among years (P > 0.50) and sexes (P > 0.70). Suspected primary causes of death for calves were predation (57%) and malnutrition (33%). Survival rates for adult females(~ 12 months old) during winter-spring were 0.96 ± 0.05, 0.96 ± 0.04, and 0.95 ± 0.04 also during 1993-94, 1994-95, and 1995-96 and survival was similar among years (P > 0.70). Hunting was the primary cause of death (57%) for adult females during winter-spring. Annual survival rates (l December-JO November) for adult females were 0.7~ ± 0.10 and 0.84 ± 0.09 in 1993-94 and 1994-95, respectively. Survival was similar between years (P > 0.50) and hunting annually accounted for 88% of the adult female deaths. Hunting removed 79% of the adult 2-year old males. Survival of males from 6month old calf to 35-month old adult we~e 0.09 ± 0.10 compared to 0.86 ± 0.12 for females of the same cohort. Wounding loss on adult males was 9% and illegal loss of yearling males averaged 10%. Estimates of population size based on mark-resight estimators ranged from 3,272-3,618 elk or about 26 elk/mi2 of winter range. The IEJHE mark-res.ight model using data pooled f rem 1iiiHi1ili • BDOWD1 □ 789 �4 aerial flights provided the best estimate of population size which w~s elk with a 95% CI of 3,129-3,807. Estimates based on counts of- elk on " ,, .. :-. randomly selected 1 mi2 -quadrats during 2. replicate flights -were '2,'16sJka"/\.:. 3 I 175 elk. Sighting bias models to adjust for negative bias of quadr~t''9otii1€e :. are currently being evaluated. We believe stratified sampling ·systems ~:11~~?9 · • quadrats as sample plots and with counts corrected for sighting bias holds'· :> • ~. promise for providing reasonable estimates of population size. ,' �89 BS!rIMA!rIHG SURVIVAL RADS OF ELK AND DEVELOPING !rECBRIQlJBS m BS~IMAm POPULA~IOH SIZE David J. Freddy P, ff, OBJECTIVE Estimate survival rates of adult female a·nd calf elk and develop techniques to estimate population size. SEGMEHT OBJECTIVES 1. Radio-collar 75 calf and 15 adult female elk during December 1995 in Game Management Unit 42 south of Rifle, COlorado. 2. Estimate winter and annual survival r~tes of calf and adult elk from known fates of radioed elk. 3. Estimate density of elk in a portion of GMU-42 using 2 replicate flights of a quadrat sampling system. Apply sighting bias corrections to adjus~ numbers of elk counted during each replicate. Compare biase adjusted estimates with mark-resight estimates based on numbers of marked elk seen during the 2 replicate flights plus 2 nonrandom mark-resight flights. 4. Analyze data and summarize annually in Federal Aid Job Progress reports 5. Prepare_ draft of manuscript on sighting bias models developed from data collected in 1994 and 1995. 6. Continue to monitor locations and movements of selected radiocollared elk. INTRQDUCTXQH Our objectives-are to provide reliable estimates of survival rates for calves and -adult females during winter and for adult females throughout the year for the period 1993-94 through 1996-97. Additionally, we will develop and test a system for estimating population size that will incorporate estimates of sighting bias in conjunction with a random sampling system using search quadrats as sample units. our winter study area encompasses about 839 Jan2 (324 mi2 ) in the eastern half of Game Management Unit 42 south and east of Rifle, Colorado. Elk winter range vegetation types include juniper-pinyon woodland (Juniperus osteosperma-Pinus edulis), oakbrush-mountain sh~ub (Ouercus gambelii-Amelanchier alnifolia), aspen (Populus tremuloides), sagebrush (Artemisia tridentata), and agricultural fields (Freddy 1993, 1994). �90 METHODS • Marking We placed radio collars (172-176MBz) ·having mortality sensors on 71 calves (6 months old), of which 37 were males and 34 were females. Of these, 68 calves were trapped from. 8-11 December 1995 using helicopter net-gunning and 3 calves were trapped on 21 December using portable corral trap.a. Helicopter capture occurred at 10 remote sites located primarily. on public lands while corraltrapping occurred at 1 site on public lands. Trapping effort was all9cated among 8 geographic trap zones to assure that radioed elk were representative of most if n9t all segments of th~ population (Table 1). Radio collars were of the same type used in 1993 and 1994 (Freddy 1994). Calves captured by net-gunning were ferried by helicopter to processing points usually within 1. 6 Jan of capture sites. At processing points·, body weight, total body length, hind foot length, and rectal body temperature (F) were measured and calves were then radio-collared and released. Similar measurements were also made on calves that were_ corral-trapped and then released at the trap site. Body measurements for calves were compared between sexes and years using Pree FREQ, GLM, and REG (SAS 1988)·. survival We monitored life or death status of radioed elk during daily ground surveys and aerial surveys conducted at 2-4 week intervals from December 1995 through ·April 1996 and via monthly aerial surveys from M~y to November 1995 and May to. June 1996. Survival rates (S) of radioed elk were calculated using the binomial estimator with a variance, VAR(S) ~ S(l-S)/n (White and Garrott 1990) (Pree FREQ, .SAS 1988). Survival rates are expressed as the mean estimate ± the 95% confidence interval. We used x2 -contigency tests to compare survival rates. we· chose not to use the staggered entry approach and did not use a Kaplan-Meier approach because elk were captured in~ short time interval and few animals were censored (White and Garrott 1990). We defined 4 major time intervals for survival analyses: winter-spring was 1 December to 14 June, summer-fall was 15 June to 30 November, annual was 1 December to 30 November to coincide with timing of capture and radiqcallaring, and yearly for yearling elk aged 12-23 months was 15 June to subsequent.14 June. Life or death status of all calves radioed in.December 1995 (71) was known for the period 8 December 1995 through 15 June 1996 except for 1 male calf collar, 174.619/95 which apparently slipped off the elk in May 1996. on 15 June·,· calves become yearlings for purposes of calculating rates of calf surviYal. Life or death status for adult females, adult males, yearling females, and y~arling males collared iri December 1993 or 1994 was known for 198 of 205 elk through_ 15 June 1996. Causes of death were-estimated from multiple sources of evidence including: presence or absence of gunshot wounds, presence or absence of bite wounds on carcass and predator tracks or ~cat at carcass site, p:tiysical positioning of carcass remains whether buried, covered, scattered, or consolidated, re.lative amount of internal fa~ and marrow fat if pre~~nt with carcass, and results of histopathology an~ marrow fat analyses (Wade ~nd Browns 1982, Halfpenny and Biesiot 1986). Fat co_ntent (percent dry matter) of .bone· marrow and estimates of age based on d~ntal cementum were obtained for dead elk by the Colorado Division of Wildlife Laboratory while histopathology analyses were provided by ( / \ �91 the Colorado State University Veterinary Diagnostic Laborat.ory. Photographs were taken of nearly all mortalities so that physical evidence could be reviewed and judged by outside experts (pers. comm. A. Anderson, T. Beck, w. Andelt). Pgpulation Estimates Elk population size in GMU 42 from East Alkali Creek west to. Grass Mesa was estimated during mid-January and early March 1996. We counted elk using a· Bell-Soley helicopter on 4 flights. Two replicate flights involved counting elk on 53 quadrats about 1 mi 2 (2.59 km2 ) in size representing.a 40% strati£ ied random sample of t.he 132 mi2 winter range. • Each potential quadrat in 8 geograhic. blocks were ranked as a high or low density quadrat resulting in 14 strata. Intensity of sampling was 60% in high density and 32% in low density strata with a minimum of 3 quadrats in any one strata~ Two flights involved counting elk.encountered on a nonrandom route that traversed the same 132 mi 2 area. One flight of each type was conduct~d during the time periods 15-20 January and 29 February-3 March 1996. During helicopter surveys, fixedwinged flights were conducted to locate 219 radiocollared elk to determine whether these marked elk were within or outside the 132 mi2 sample area. During all flights, numbers of marked and unmarked elk were tallied by observers. We gener~ted estimates of population size using several markresight estimators for each individual flight, pairs of flights, and all 4 flights pooled using program NOREMARK (White 1996). We considered. estimates based on all 4 flights pooled to be our best estimate of population size- and the b~nchmark against which individual flights, especially·quadrats, would be compared. Estimates from quadrat sampling were computed using program DEAMAN (Colo. Div. Wildl. software). At this time, we have~ applied sighting bias correction factors to the counts of elk observed on quadrats •(Freddy 1995) .• Movements We continued to locate selected radioed.elk at least once per month since capture to document seasonal movements via telemetry using a Cessna 185. These elk were selected at random from within trap zones and equalized by age class in Janaury 1994. Elk from the original sample that died were replaced each subsequent January primarily with randomly selected 6 month-old calves of· the same sex from the same trap zone(s) as elk that died. As of 1 January 1996, these 4·4 elk were classified as 22 adult females, 4 yearling females, 6 • female ca·lves, 2 adult males, 4 yearling males, and 6 male calves. During June 1996, we agaln located 28 adult ·females that. were selected at random in 1995 to document locations during the calving period. As needed, we located other elk to document· unusual movements. RESULTS AND DISCUSSION c;apture Two elk calves died during helicopter net-gunning activities. These have been the only deaths during capture of 235 elk using the helicopter. one calf died_ of a bro~en neck during capture and one calf (174.800/95) di~d within 10 days of capture due to the effects of capture myopathy (histopathology, CSU Diagnostic Lab) and was subsequently censored from survival analyses. �92 survival ~ Between 1 December 1993 and 14 June 1996, 86 radiocollared elk died (Table 2, Appendix 1). Hunting was apparently· involved in 661 of the deaths. For adults ~1 year-old, hunting accounted for 87% of 65 deaths. There were 2 periods of mortality during the year. Calves died from February to May while adults died during fall and early winter when hunting seasons occurred (Fig. 1) . . Calves survival rates for 6-11 month old calves during winter-spring were 0.92 ± 0.06, 0.90 ± 0.01, and a.as± o.oa in 1993-94, 1994-95,· and 1995-96, respectively (Table 3). We failed to detect differences in survival of calves among years (X 22 = 0.40, P >0.50), between sexes· pooled among year·s (X21 = 0.58, P >0.70), and between sexes within each yearly cohort (X21 S 0.79, P > 0.70) (Tables 3, 5, 6). Sex of dead calves for all years was 12 male and 9 female (Table 2). These survival rates were associated with winters considered mild in temperature and having low or moderate snow depths. In 1994-95, considerable snow fell during March and April but usually melted rapidly at lower elevations. Causes of death for calves during winter-spring were suspected malnutrition (33%), mountain lion predation (33%), ·suspected predation (19%), bear predation (5%), and unknown cause (10%) (Tables 2, 4). Percent fat in bone marrow of calves suspected of dying from predation was 82 for males {(range 6495, n = 6) and 48 for females (range 29-63, n = 3) and for calves dying from suspected malnutrition, 14 for males (range 0.2-41, n = 3) and 11 for females (rang~ 8-15, n = 3) (Table 4). Yearlings -:i survival rates for elk 12-23.months of age were 0.86 ± 0.09·for males (n = 57) and 0.95 ± 0.05 for females (n = 66) when yearlings were pooled among years and hunting mortalities were included (Tables 5, 6). Excluding or censoring hunting mortalities increased survival rates to 0.98 ± 0.04 for males and 0.98 ± 0.03 for females. Males had lower survival rates when hunting mortalities were included (X2 1 = 3.38, P = 0.07) but not when hunting mortalities were • excluded (x\ = 0.03, P > 0.50). Of the 11 deaths involving elk 12-23 months old, 9 (82%) were hunting related (Table 3). Adult FemalQs ·Survival rates for adult females(~ 12 months of age) during winter-spring and inclusive· of hU:ntin_g mortalities were 0.96 ± 0.05, 0.96_ ± 0.04, and 0.-95 ± 0.04 in 1993-94, 1994-95, and 1995-96, respectively. Excluding hunting mortalities increased·survival to 0.98 ± 0.03, 0.99 ± 0.02, and 0.97 ± 0.03 in the same 3 years (Table 7). We failed to detect differences in survival among years inclusive (x22 = 0.37, P >= 0.50) or exclusive (X22 = 1.52, P > 0.70) of hunting mortalities. Of 14 winter-spring deaths, 8 (57_1) were related to late-season rifle hunting: 4 legal and 1 illegal harvest and 3 wounding losses (Table 2). Six natural deaths included 2 suspected predation, 3 of unknown cause,_ and 1 calvingrelated~ Of the 3 elk lost to wounding, 2 were suspici9us kills. Both of these elk died from 1 rifle shot to the upper neck or head in relatively plain ( ( �93 sight near or on an agricultural field where retrieval of the carcass was not difficult. These animals may_have been accidentally shot, maliciously shot, .or abandoned because of the radiocollar. Survival rates for adult females(~ 12 months of age) during summer-fall and -inclusive of hunting mortalities were 0.87 ± 0.07 and 0.94 ± 0.04 in ·1994 and 1995, respectively. Excludlng hunting mortalities increased survival to 0.99 ± 0.02 and 1.00 in the same 2 years (Table 7). We failed to detect differences in survival among years inclusive (X21 = 3.13, P = 0.09) or exclusive (x\ = 1.38, P > 0.70) of hunting mortalities. Hunting accounted for 20 (95%) of 21 summer-fall deaths. The one natural death was ~alvingrelated. During hunting seasons in Fall 1994 and 1995, 100 and 129 radiocollared adult females, respectively, were a~ailable to hunters (Table 7). Proportions of these marked elk lost to hunting-related mortalities were 12 1k in 1994 and 8% • in 1995. Annual survival rates for adult females(~ 12 months of age) marked as a cohort in December 1993, inclusive of hunting mortalities, were 0.78 ± 0.10 and 0.84 ± 0.09 in 1993-94 and 1994-95, respectively. Excluding hunting mortalities increased survival to 0.96 ± 0.05·and 0.98 ± 0.04, respectively (Table 8). We failed to detect differences in survival among years inclusive (X21 = 0.19, P > 0.50) or exclusive (x\ = 0.15, P > 0.50) of hunting mortalities. For this cohort of adult females, hunting accounted for 22 (88%) of 25 deaths during both years. Adult Males Survival of males from 6-month old calf t~ 35 months of age was 0.09 ± 0.10 compared to 0.86 ± 0.12 for females of.the same calf cohort (Table 5). Hunting was the overwhelming cause of mortality among male elk. From the 1993-94 calf cohort of 36 male calves, 28 (78%). lived to become 2-year old· legal branch-antlered bulls for ·the 1995 hunting seasons. Of these 28, 22 or 79% were harvested in the Fall of 1995 inclusive of 2 wounding losses (9% of 22). Rifle seasons accounted for 16 (73%) of the hunting mortalities (Table 2) . . Most of these 2.-year old bulls were harvest~d .in the Grand Mesa DAU or in adjacent GMO 43 .. One.exception was a bull taken during archery ·season in GMO 63 on Black Mesa about 100 miles south of its capture site in GMO 42. One surviving bull was located in March 1996 in GMO 4.7 east of Basalt, co about 35 miles southeast of its capture site in GMO 42. Yearling spike-antlered bulls were generally not legal quarry during hunting seasons. In 1994, 32 yearling bulls from the 1993-94 calf cohort entered the hunting seasons presumably as spike-antlered bulls and 4 (12.5%) were illegally taken with 3 taken during rifle seasons and 1 taken prior to rifle seasons. In 1995, 30 yearling bulls from the 1994-95 calf cohort entered the hunting season and 2 (7%) were illegally taken and 1 (3%) was fatally wounded during ar9hery season in an ar~a where the bull was legal q,.Jarry. cal:f Body ·size Body weights of male and females calves were largest in 1995 with the inerease most pronounced for males (Tables 9, 10).. Increases in weights were not unexpected as the locally moist summer of 1995 wa·s favorable to forage production on summer ranges. However, we failed to detect.differences in �94 body·weights among years (P = 0.206) and any interaction between year and sex (P = 0.380). overall, male calves had larger body weights (P = 0.0001), longer total body length (P = 0.0386), longer hind leg lengths (P = 0.0001), and higher condition indexes (P = 0.0001) than female calves (Table 10). Accurately predicting body weight from total body length, hind foot length, or a combination of·these 2 measurements does not look promising at this time, although all regressions were significant (P < 0.0001). The multiple regression using body length and hindfoot length as independent variables provided the best correlation coefficients (r2) of 0.63 for female calves, 0.69 for.male calves, and 0.69 for sexes combined. Calves suspected of dying from predation (Table 4) averaged 121 kg in weight for males (range 100-140 kg, n = 6) and 91 kg for females (range 79-117, n = 5). Predators thus appeared to take larger than average males and smaller than average females (Table 10). Calves suspected of dying from malnutrition (Table 4) averaged 95 kg in weight for males (~ange ~7-127, n = 3) and 107 kg for females (range 100~120, n = 3). Malnutrition thus appeared to affect smaller than average males .and average sized females (Table 10). Pgpulation Estimates Our elk population was not geographically or demographically closed during the time interval encompassing flights. During our first pair of flights, some elk were still moving down in elevation in response to heavy snow in early January resulting in.fewer than expected radiocollars on the sample area. By the March flights, all elk had reached elevations encompassed by the sample area but some elk moved lower in elevation onto private lands outside of but adjacent to the sample area and some elk moved to winter ranges adjacent to but also outside the sample area. These movements resulted in radiocollared elk moving into and out of a "pool" of elk that was different and larger than the elk population encompassed. by the. sample area. Between pairs of flights, 2 radiocollared elk died indicating some los~ of elk was occurring during our flight time interval. Fortunately, both quadrat and nonrandom flights reflected proportionately similar and positive changes in total elk and total marked elk counted between pairs of flights indicating that movements of marked and unmarked elk into the sample area were similar (Table 11). ( ( Population estimates from 4 mark-resight estimators ranged from 3,272-3,618 elk or 24.8-27.4 elk/mi2 of winter range within the sample area (Table 12). At· this time, the estimate of 3,415 elk generate~ by the Immigration/Emigration JHE mark-resight estimator is probably the best estimate of population size on the sample area because it accounts for movement of elk on and off the sample area. Adding the 165 elk directly counted on private lands in lower Divide Creek and Hunter.Mesa (57 mi 2 area) increased the population to 3,580 elk in the intensive study area. In January 1995, a single nonrandom flight produced a Lincoln-Peterson mark-resigh~ estimate of 3,411 elk and adding 399 elk counted on private lands in lower Divide creek and Hunter Mesa.to this estimate.increased the_ population to 3,810 elk in the intensive study area in 1995 (Freddy 1995). Population estimates for the larger "pool" of elk within which radiocollared elk were moving ranged from 3,969-4,004 elk as estimated by the IEJHE and BE mark-resight estimators (Table 12). This "pool" ~stimate wo~ld include el~ on the Divide creek and Hunter Mesa private lands and some addit!onal elk on areas ~djacent to the western boundary of the intensive study area. ( G �95 Population estimates from quadrat flights 1 (2,165 elk) and 2 (3,175 elk) were lower than the IEJHE mark-resight estimate (3,415) (Table 12). Although both quadrat flights were subject to sampling error, the lower . estimate from the first flight may reflect a less favorable distribution of elk ~uring that flight. Confidence intervals overlapped between quadrat flight 2 and IEJHE suggesting estimates were not different. Quadrat estimates will be adjusted upward for sighting bias and at that time more conclusive comparisons to markresight estimators will be made (Freddy 1995, Samuel et al . 1987). Elk densij:ies ~ 50/mi2 occurred within the Garfield Creek High, West Divide Creek 2 West High, and Grass Mesa Low strata which together encompass about 25 mi of winter range (Table 13). Elk Movements In December 1994, 33 male and 36 female calves and 8 adult females were radiocollared. Of these elk, 27 male and 31 female calves and all 8 adults survived to 30 November 1995. None of the adult females dispersed during winter 1995-96 to a winter range outside of the intensive study area in GMO 42. However, 25 (9M, 16F) or 43% .of the surviving yearling elk dispersed to winter ranges outside the intensive study area (Table 14). Rates of dispersal were 33% for males and 52% for females. Calves trapped in zones E and F, which encompass Alkali, Dry Hollow, and the Mamm creeks, comprised 68% (17) of the dispersing yearlings but only 45% of the surviving yearlings. Calves trapped in zones E and F dispersed to winter ranges near the towns of Collbran, Parachute-Rulison, and Paonia. Calves trapped in zones C and D which encompass West Divide Creek comprised 28% of the dispersing juveniles and these elk primarily moved south to winter ranges near Paonia and Paonia Reservoir. We are. currently creating a GIS land database for the area frequented by radiocollared elk. This is a cooperative effort through the National Biological Service and Rocky Mountain Elk Foundation . We believe plots of elk locations and movements using this database will be forthcoming in 1996. 12 lililllll CAL VE S 10 - ~ ADULT lfALES l<<<·>I ADULT FEtlALES 8 •••••• •••4••• •• •••• •••••• •••••• •••••••• •• •••••••••••• • •••••n•--• • ••• • •• •••u•••••••• • •••••• • ••••••••• • ••• • ••••• •••••• o o o o o o o o o o O o o U OO • • o • • O o o O oOOOO 00000000• • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • O oOo O O OOO O OOOOOO H Q P! r..i 6 •o• o o•••• ~ ~ 4 2 ···········-······· ···········~-,................................ . 0 Figure 1. Timing of deaths for calf and adult elk, 1993-1996. 12 months old and adults were >12 months old. Calves were 6- �96 CQNCLUSXQRS \._/ We obtained acceptably precise estimates .of calf and adult survival rates and recommend continuing our current sampling effort of capturing and radiocollaring 70 calves in 1996-97 and co~tinuing to monitor> 75 radioed adult females and> 25 radioed adult males in 1996-97. We recommend conducting replicate aerial surveys in 1996-97 to estimate elk density ~sing sample quadrats, mark-resight estimators, and sighting bias correction models to complete our evaluation evaluating techniques to estimate elk density. LITERATURE CITED Freddy~ D. J. 1993. Estimating survival rates of elk and developing techniques to estimate population size. Colo. Div. Wildl. Game Res. Rep.· July: 83-117. Freddy, D. J. 1994. Estimating survival rates of elk and developing techniques to estimate population size. Colo. Div. Wildl. Game Res. Rep. July: 27-42. Freddy, D. J. 1995. Estimating survival rates of elk and developing techniques to estimate population size. Colo. Div. Wildl. Game Res. Rep. J~ly. Halfpenny, J. c., and E. A. Biesiot. 1986. A field guide to mammal tracking in North America. Johnson Books, Boulder, co. 161pp. • Samuel, M. D., E. o. Garton, M. W. Schlegel, and R. G. Carson. 1987. Visibility bias during aerial surveys of elk in northcentral Idaho. Wildl. Manage. 51:622-630. J. SAS Institute Inc. 1988. SAS/STAT User's Guide, 6.03. SAS Institute, Inc., Cary, NC. 1028pp. Wade, D. A., and J. E. Browns·. 1982. Procedures for evaluating predation on livestock and wildlife. Texas Agric. Expt. Sta. Publ. B-1429. 42pp. White, G. c. 1996. NOREMARK: Population estimation from mark-resighting surveys. Wildl. Soc. Bull. 24:50-52. White, G. c., and R. A. Garrott. 1990. Analysis of wildlife radio-tracking data. Academic Press, Inc., San Diego. 383pp·. Prepared by ____________________ David J. Freddy Life/Science Researcher �( in 8 trapzones. December 1993.1994. and 1995 within Game Management Unit 42. Table 1. El.k capture objectives and numbers of elk radiocollared Elk Col lared1 Calves Collared ·Elk Captur~ Years Corral Objectives Males TotalLYeac Helicoeter Females Years Adult Females Collared Trap 93 94 95 93 94 95 Total 93 94 95 93 94 95 93 94 95 93 94 95 Total Zone Name 93 94 95 Total A B C D E F G H Garfield Gibson Uncle Bob West Divide Hightower Middle Manm West Manm Dry Hollow All 5 8 13 13 10 8 8 21 6 12 11 11 10 9 10 6 8 6 4 25 17 7 29 13 14 21 11 18 • 17 17 14 0 13 9 6 ·O 5 35 6b 0 8 6 4 19 7 4 29 13 14 21 11 18 17 17 14 0 13 9 6 0 5 0 0 0 0 0 0 6 10 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 35 6 0 18 49 56 50 48 22 11 41 150 86 75 141 83 71 100 67 68 41 295 8 20 24 26 10 10 8 44 16 3 3 4 0 2 7 3 1 5 4 7 10 1 8 9 ·8 8 5 0 1 0 0 36 ·33 37 5 10 5 1 8 7 6 3 0 3 9 3 3 6 3 6 4 8 10 8 6 5 4 O· 4 0 0 37 36 34 14 31 44 38 • 22 10 4 12 12 10 10 • 0 1 16 19 0 0 6 0 0 0 2 0 0 0 0 0 0 0 6 0 213 68 14 0 38 4 18 12 12 10 0 1 25 82 •Helicopter= Helicopter net-gunning, Corral= corral-trapping. 6 adult females cap~ured 2 March 1995 b All Table _3. Survival rates of calve~ with sexes pooled for 1 December - 14 June each year 1993-94, 1994-95, and 1995-96. Survival· rates (S) calculated as a mean estimate of (alive)/(alive + dead) and variance S(lS)/n collars. Elk Age and Time Period (dates) • Calves Cal~,es Calves Survival 'L 95\- CI U 9~%' CI n collars Censored1 Died Nonhunting Hunting 12/01/9306/14/94 12/01/9406/14/95 12/01/9506/14/96 0.92 0.85 0.98 73 .0. 90 0.83 0.97 0.88 0.81 0.96' 69 0 6 6 0 69b 0 7 7 0 2c 8 8 0 • Censored denotes collar failure and/or animal life/death status not known. b Includes (173.949/94) male whose collar'failed 12/94 but seen alive 1/95, 1/96. c Censored elk were (174.619/95) for slipped collar and (174.800/95) for trap-related mortality. �98 ~ Table 2. Causes of deaths in radiocollared elk between 1 December 1993 and 14 June 1996. Calves (M=male, F=female) were 6-12 months old and collared at 6 months of age. Yearling males and females were 12-18 months old and collared as 6 month old calves. Juvenile males and females were 18-23 months old and collared as 6 month old calves. Adult males and females were .2. 24-months old at time of death. Elk AgelSex Class at Death Total Adult Juvenile Calves Yearl±ng All Cause of Death M F F MF M F F M M Malnutrition Unknown-Suspect Malnutrition Predation-Lion Predation-Bear Unknown Predator Unknown-Suspect Predation Ac'cident-Birthing '....,_/ ') 2 2 0 0 0 0 0 1 2 5 2 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 2 0 0 0 0 0 1 0 0 0 0 0 Legal Hunting Archery Muzzleloading Archery/Muzzle Rifle-First Rifle-Second .·-· Rifle-Third Rifle-Late 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 5 1 Wounding Loss Archery/Muzzle Rifle-Regular Rifle-Late 0 0 0 0 0 0 1 Illegal Hunting Rifle-Regul'ar Out-of-Season 0 0 0 0 s• Presumed Hunting Disappear-Rifle Seasons 0 Unkl'lown Cause Totals 0 0 ·o 0 2 2 4 0 1 5 2 3 3 8 1 0 0 1 1 1 1 2 2 0 3 2 5 2 1 5 2 7 0 1 0 1 0 0 0 0 1 0 1 1 1 0 2 4 6 2 2· 4 6 0 0 0 () 0 6 2 4 0 2 4 0 2 0 0 0 0 2 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 lb lb 0 0 2 0 0 1 ·o 12 ·9 7 3 1 O· 0 0 0 0 0 0 0 0 0 0 0 0 0 o. .o 5 5 1 8 6 5 1 2 4 2 6 0 3 3 .5 0 0 5 1 1 1 2b 2b 3 3 6 0 0 2 2 3 5 0 22 32 42 44 86 4 3 0 ( ( • These elk were illegally shot as spike-antlered yearling males: (173.190/93), (173.232/93), (173.320/93), (113.919/94), (174.059/94). "Th~ elk disappeared during rifle hunting seasons and are.presumably dead and legally harvested: (17_2.207/93), (172.649/93), (172.800/93), (173.309/93), (173.390/93), (173.439/93). One exception may b-..; spike:..anttered yearling male (173.309/93) that disappeared during the first rifle season in 1994 when spike-antlered bulls were not legal. ( \ \J �( ( Table 4. Estimated causes of mortality and associated body condition for 21 radiocollared calf elk 6-11 14 June 1996. months old, 1 December 1993 Ag~• Body> Marrow. Trap Frequency ID/ t Fat Date Dead Cause of Death & Code No. Sex {months} Wt. {kg} zone Year Ca12tured - 172.899/93 173. 000/93 173. 262/93 173.289/93 173.461/93 173.469/93 173.589/94 173.640/94 173.789/94 173.870/94 174.119/94. 174.140/94 174.170/94 174.220/95 174.230/95 174.240/95 174.339/95 174.500/95 174 .·609/95" 174.689/95 .·174 ;789/95 C G C C H H B C E D F F B F F E E F F M M M M F F F F M M M M M M F C F B o· C F M M 9 10 ·9 9 8 10 12 9 9 86.0 102.0 108.0 111.0 82.0 77.0 117.0 89.0 100.-0 03/18/94 04/25/94 03/18/94 03/22/94 02/07/94 04/25/94 06/01/95 03/30/95 03/20/95 8 86.0 140.0 112.0 140.0 02/21/95 03/01/95 03/14/95 04/25/95 . 127. 0 120.0 115.0 79.0 78.0 120.0 100.0 04/08/96 . o~/24/96 04/17/96 03/27/96 04/16/96 04/15/96 02/05./96 05/22/96 9 9. 10 9 10 10 9 10 10 7 11 • Approximate age at death assuming rs June binhdate. • Whole body weight at capture in December of 1993, 1994, or 1995. c Fat content as percent dry matter of bone marrow from either femur of carcass. • Fat content as percent dry matter of bone marrow from either ~andible of carcass. • Fat content as percent dry matter of bone marrow from either humerus of carcass. 28. 50d 8. 03c 2a. sad 0 .21c 0. 27c 51. 71c 14 ."•71c 94. 97c 64 _44·c 76. osc 41. oac 86. 2ac 62. 97c 34. 56c 10. 2ac 91. 90c 75. 32e Lion Predation-3 Malnutrition-6 Unknown-11 Unknown-11 Malnutriton-6 Malnutrition-6 Unknown; suspect predation-30 Unknown; suspect p:t:edation-30 Unknown; suspect malnutrition-31 Lion Predation-3 Bear Predation-35 Lion·Predation-3 Lion Predation-3 Unknown; suspect malnutrition-31 Lion Predation-3 . Lion Predation-3 Unknown Predator-5 Unknown; suspect malnutrition-31 Malnutrition-6 Lion Predation-3 Unknown; suspect predat;.ion-30 �( ..... 8 Table 5. Survival rates.for the cohort of 6-month old elk calves radiocollared in December 1993 for 6 time periods from. 1 December 1993 through 14 June 1996·when elk were 35 months old. Survival rates for male and female calves pooled when 6-11 months old were. 0.92 (95% CI 0.86-0.98, n=73). Survival rates (S) calculated as a mean estimate of (alive)/{alive + dead) and variance S(l-S)/n collars. Elk Age (months) and Time Period (dates) 12 - 17 18 - 23 24 - 29 30 - 35 6 - 11 6 - 35 06/15/9412/01/9406/15/95.12/01/9512/0l/9312/01/9311/30/94 06/14/95 11/30/95 ~6/14/96 06/14/94 06/14/96 MALES Survival L 95% CI U 95% CI n collars Censored1 Died Nonhunting Hunting 0.89 0.78 0.99 36 0 4 4 0.88 ·o.76 0.99 32 0 4b 1.00 28 0 0 0 0 0 0 4 0.95 0.87 1.00 37 0 2 2 0.97 0.92 1.00 35 0 1' 0 1.00 0 1 0 0.21 0.06 0.37 28c 0 1.00 0.00 o. 00. 3 22d 0 0 0 0 22 3e 0.09 0.00 0.19 33 3 30 4 26 FEMALES Survival L 95% CI U 95% CI n collars Censored1 Died Nonhunting HUiiting 34 0 0 0 0.97 0.91 1.00 34 0 l 0 1 • Censored denotes collar failure and/or anima1 life/death status not known. 1> Collar (173~309/93) "disappeared" during Fall 1994 hunting seasons and assumed dead. ' Includes collar (173.241/93) that failed in August 1995 but bull seen alive Januacy 1996. d Three collars (173..390/93, 173.402/93, 173.439/93) "disappeared" during Fall 1995 hunting seasons and assumed dead. • Three collars censored; (173.241/93) failed, (173.340/93) not heard spring 1996, (173.381/93) slipped off 12/01/95. ' Collar (172.800/93) "disappeared" during Fall 1994 hunting seasons and assumed dead. 0.97 0.9~ 1.00 33 0 1 0 1 0.86 0.98 0.75 37 0 5 2 3 �( ( Table 6. Survival r.ates for the cohort· of 6-month old elk calves radiocollared in December 1994 for 4 time periods from 1 December 199~ through 14 June 1996 and survival rates for the cohort of 6-month elk calv.es radiocollared in December 1995 for one time period .. Survival rates for male and female caives pooleq when 6-·11 months old were 0.90 in 1994 (95% CI. 0.83-0.97, n=69) and 0.88 in 1995 (95% CI 0.81-0.96, n=69). Survival rates (S}calculated as a mean estimate of (alive)/(alive + dead) and variance S{l-S)/n collars. Elk Age (months) and Time Period (dates) • ____________1=9-9-4___C=a=l=f__C=o=h=o=r-t____· _ _ _ _ _ _ 1995 Calf-Cohort 12 - 17 18 - 23. 6 - 24 6 - 11 6 - 11 06/15/~512/01/95-· 12/01/9412/01/9512/01/9411/30/95 06/14/96 06/14/96. 06/14/96 06/14/95 MALES Survival L 95% CI U 95% CI n colla_rs Censored• Died Nonhunting Hunting FEMALES Survival L 95% CI U 95% CI n collars Censored• Died Nonhunting Hunting 0.91 0.81 1.00 0.90 0.79 1.00 33b 30b 0 3 3 0 0 3 0 3 0.89 0.78 0.99 36 0 4 4 0 0.97 0.91 1.00 32 0 1 0 1 0.95 0.87 1.00 22 5c 1 1 0. 0.97 0.90 1.00 30 1c 1 1 0 0.75 0.59 0.91 28 0.86 0.74 0.98 35 5 7 4 3 2d 0.83 Q.70 0.96 35 1 6 5 1 0.91 0.81 1.00 5 5 0 34 0 3 3· 0 • Censored denotes collar failure and/or animal life/death status not known. 11 Includes collar (173.949/94} that failed in December 1994 but male seen alive in January 199~ ' Censored elk were (173.949/94) failure, (173.981/94}~ (174.019/94), (174.090/94), (174.160/95) not heard spring 1996. • Censored elk were (174.619/95}slipped collar, (174.800/95)capture mortality. c Cenosred elk was (173.719/94) slipped collar. _., 0 _., �( ( ( .... 2 Table 7. Survival rates for all radiocollared adult female elk >12-months old during 5 periods from 1 December 1993 through 14 June 1996. Survival rates (S) calculated as a mean estimate of (alive)/(aliye + dead) and variance S (1-S) /n collars·. Elk Age and Time Period (dates) Adult Adult Adult Adult Adult 06/15/94.12/01/9406/15/9512/01/9512/01/9311/30/95 06/14/96 06/14/95 11/30/94 06/14/94 0.87 0.96 0.94 0.95 0.96 Survival 0.90 0.91 0.80 0.92 0.91 L 95% CI 1.00 0.98 0.99 0.94 1.00 U 95% CI 95bg 129bh 68bc 100bf 119 1 n collars 2j 0 0 0 0 Censored• 9d 13c 7 4 3 Died 0 4 1 1 1 Nonhunting 8 3 12 3 2 H~ting • Censored denotes collar failure and/or animal life/death stams unknown. b Includes collar (172.011 /93) that failed 4/ 1994 but seen alive in 1/ 1996 c Collars (172.800/93, 172.649/93) "disappeared" Fall 1994 hunting seasons.assumed dead. 4 Collar (172.207/93) "disappeared" Fall 1995 hunting·seasons, assumed dead. • Composition is 6 females 18+ and 62 females 30+ _months old. r Composed of 35-12+, 6-24+, and 59-36+ months old females. 'Composed of 34-18+, and 61-30+ montho old females. h Composed of 32-12+, 34-24+, and 63-36+-month old females. 1 Composed.of 30-18+ and 80-30+ monthos old females. J Censored (172.011/93) failure and (173.719/94) slipped collar. ·Table 8. Survival rates for the group of adult female elk >12-months old radiocollared in December 1993 for 8 time periods from 1 December 1993 through 14 June 1996. Survival rates (S) calculated as a mean estimate of (alive)/(alive + dead). and variance S(l-S)/n collars. Elk Age and.Time Period (dates) • Adult Adult Adult ·Adult Adult Adul-t Adult Adult 12/01/9406/15/9512/01/9512/01/9312/01/9412/01/9306/15/9412/0l/9311/30/94' 06/14/95 11/30/95 06/14/96' 11/30/95 06/14/96 11/30/94 06/i4/94 0.93 0.88 0.93 0.78 0.84 0.58 0.82 0.96 Survival 0.85 0.78 0.68 0.85 0. 75. 0.46 0.72 0.91 L 95% CI 0.88 0.70 0.97 1.00 0.93 0.99 0.91 1.00 U 95%-CI 53b f 68b r . 68bc 67f 65br 53br 49b r 42' n collars 1 11 . 0 .o 0 0 0 1 0 Censored• 29cd 15c 10d 4 · 6d 3 3 Died 6' l· 2 1 0 3 1 .Nonhunting 3 6 0 22 13 9 2 Hunting • Censored denotes collar failure and/or animal life/death stams unknown. 11 Includes collar (172.011/93) failed 4/1994 but seen alive 1/1996. c Collar (172.649/93) "disappeared" Fall 1994 hunting seasons, assumed dead. d Collar (172.207/93) "disappeared" Fall 1995 hunting seasons, assumed dead . .,---- I. \ • Composed of 6-18+ and 62-30+ month old females. r Composed of 100% females 24+ months old. 'Censored (172.011/93) failure of unknown status Spring 1996. �( ( ,, Table 9. Frequency distribution of whole body weights for male and female elk calves trapped in Game Management Unit 42. December. 1993. 1994, and 1995. Percentage per weight class shown in parentheses. Bod~ Weight Class {kg} Year Sex 60-69 70-79 80-89 90-99 100-109 110-119 120-129 130-139 140-149 Total 1993 1994 1995 M M M 0(0.0) 0(0.0) 0(0.0) 1 (2. 9) 1(3.0) 0(0.0) 1(2.9) 1(3.0) 0(0.0) 3(8.6) 2(6.0} 4(11.4) 6·(.17 .1) 6(18.2) 3(8.6) 12(34.3) 13(39.4) 8 (22. 9} 10(28.6) 6(18.2) 9(25.7) 1 (2. 9) . 2(6.0) 10(28.6) 1 (2. 9} 2(6.0) 1(~.9} 35(100) · 33 (100) 35(100) 1993 1994 1995 F F F 0(0.0) 0(0.0) 1(3.1) 1(2.9) 2(5.7) 2(6.3) 4(11.4) 4 (11. 4) 1(3.1) 8(22.9) 7(20.0) 3 (9.4) 12(34.3) 10(28.6) 14(43.8) 4 (11. 4) 10(28.6) 5(15.6) 6(17.1) 2( 5.7) 6(18.8) 0(0.0) 0(0.0) 0(0.0} 0(0.0) 0(0.0) o·(o. 0) 35 (100} 35(100)" 32 (100) Bod~ measurements for elk calves traooed in Game Management Unit 42. December. 1993, 1994·, 1995. Male Calves Female-Calves • Min Mean Max SD Min Max n SD n Mean Measurement Table 10. Year 1993 112.7 13.7 Body Weight (kg) 191.2 10.2 Body Length (cm} Hindfoot Length (cm} 56.3 • 2 .2 Condition Index8 0.59 o.~5 77.0 164.0 52.0 0.43 141.0 210.0 61. 0 0.67 35 36 36 35 103.8 188.9 54.8 0.55 12.2 9.8 2.1 0.05 76.0 167.0 51.0• 0.46 123.0 35 207.0 ·35 59.0 34 0.64 34 1994 113.5 14.2 Body Weight (kg} 190.3 9.2 Body Length (cm) 1.8 Hindfoot Length (cm} 56.9 0. 59. 0.05 Condition Index 71.0 168.0 50.0 o.42 140.0 204.0 59.0 0.69 33 32 32 32 103.0 186.3 55.1 0.55 13.3 8.8 2.1 0.06 70.0 166.0 50.0 0.42 128.0 207.0 59.0 0.65 35 36 36 35 1995 119.1 .13. 6 Body Weight (kg} 194.0 9.3 Body Length (cm} Hindfoot Length (cm} 57.4 2~2 0.61 0.05 Condition Index 92.0 166.0 53.0 0.50 141.0 206 .0. 61.0 0. 71 35 36 36 34· 105.7 189.4 54.9 0.56 -15.2 11.4 2.3 0.06 60.0 158.0 47.0 0.38 129.0 203.0 59.0 0.66 32 34 34 32 71.0 ·1~1.0 103 164.0 210.0 104 SO. O· 6LO 104 0.71 .101 0.42 104.l· 188.2 54.9 0.55 13.5 10.0 2.1 0.06 60.0 158.0 47.0 0.38 129.0 102 207.0 105 59.0 104 0.67 101 115.1 All .. Body Weight (kg) 191.9 Years Body Length (cm) 'Hindfoot Length (cm} 56.9 0.60 Condition Index 14.0 9.6 2.1 0.05 • Condition index= (Weight/body length). ... 8 �104 Table 11. Summary of flights to estimate elk population size and density on 132 mi 2 of winter range in GMU 42 south of Rifle and Newcastle, Colorado, January/March 1996. Total Flight Marked Marked Unmarked • Elk Flight Elk Elk Date Elk Time (hrs) Flight (M/D) Available Counted Counted Counted Quadrat 1, NonRandom 1 NonRandom 2 Quadrat 2 4 Flights Pooled Flight Quadrat 1 NoriRandom 1 NonRandom 2 Quadrat 2 • 1/15-18 1/19.-20 2/29 3/1-3 128 128 137· 137 32 54 80 49 853' 1300 1918 1324 885 1354 1998 1373 19.6 8.5 • 8.0 19.1 154• Marked/ Unmarked 0.0375 0.0415 0.0417 0.0370 Marked/ Marked+Unmarked 0.0362 0.0399 0.0400 0.0357 Proportion of Available Marks Seen {+/- 95% CI) 0.2500 (0.0749) 0.42.19 (0.0~57) 0.5839 (0.0825) 0.3577 (0.0803) --------- ---------------------------------------------------------------- Changes in Total' Elk Observed: Quadrat 1 vs. Quadrat 2: +488 elk or 55% increase over Quadrat 1 NRandom l vs. NRandom 2: +644 elk or·48% increase-over NRandom 1 Changes in Total Marked Elk Observed: Quadrat 1 vs. Quadrat 2: +17 marked elk or 53% increase over Quadrat 1 NRandom 1 vs . NRandom 2 : +2.6 marked elk or 4 8 %- increase over NRandom 1 Marked Elk Deaths: During the population estimate interval (1/15i96 - ·3/3/96), 2. radiocollared elk died: 174.689/95 male calf on 2/5/96 and. 172.070/93 ~dult female on 2/12/96 . . • Marked elk available with inunigration/emmigration·movements involving 41 elk; 25 elk moved onto and 16 moved off'. the intensive • sampling area during the time interval between 1/20/96 and 2/29/96. ") Table 12. Estimates of population size and density for elk on 132 mi 2 of winter range in GMU 42 south of Rifle ~nd Newcastle, Colorado, January/March 1996. Density Pvt. Model Population '. Flights ·95% Conf: Int. Land1 Estimator Estimate /mi2 JHEb 165 4 Flights Pooled 3,168-3,840 26.3 3,47~ IEJHEC 4 Flights Pooled 3, 129-3, 8_07 165 25.9 3,415 BEd Quad l+NonRandom 1 2,764-3,872 24.8 165 3,272 Quad 2+NonRandom-2 BE 3,169-4,130 27.4 165 3,618 4 Flights Pooled 3, 969c 3,621-4,446 IEJHE 4 Fligh~s Pooled 4, 004f BE 3,560-4,504 StRSg Quadrats-1;-'light 1 1,356-2,974 11 16.4 165 2,165 Q~adrats Flight 2 165 StRS 3,175 24.1 2' 205-4' 145 11 • Elk on p(ivate land enumerated by dir~ct counts must be added to p~pula.tion estimate for estimated total elk in the intensive study area. " Joint Hypergeometric Mark-Resight Estimator that"assumes a geographically closed population and homogeneous sighting probabilities of individual elk. • • Immigration/Emigration Joint Hypergeometric Mark-Resight Estimator that allows for mqvement of elk on/off intensive study· area but assumes homogeneous sighting probabilities of elk. d Bow.den Mark-Resight estimator allows for some violation of ,a geographically closed population but allows for heterogeneous sighting probabilities of elk. c Estimate of total pool of elk within which radiocollared elk are distributed on or just off the intensive study area; equates to N' estimate." r Estimate of total pool of elk; equates to fir. • 1 Stratified random sample of qua:drats with counts of elk not adjusted for sighting bias. h 90% confidence intervals. (' u I t �( ,,( ( :! Table 13. Counts of elk on quad.rats during 2 replicate flights in February and March, 1996-in GMU 42 south of Rifle and Newcastle, Colorado. Strata Qyadrats Elk ElklQyadrat Po:eulation Size Size {mi2> n Counted Mean N StdErr. Strata Total± 90% CI Quad.rat Flight 1 10 10 133 26.60 5 16.l07 26_5 Garfield High 266 83· 20.75 16 4 17.970 . 16 332 473 Garfield Low · 8 .80 16 16 5 44 4.023 106 141 E. Divide.Low 6 6 4 23.75 95 11. 800 143 116 w. Div.· East High 10 15. 0,0 10 3 45 6.535 150 107 w. Div. East Low ·3 5 28 5 9.33 5.283 47 43 W. Div. West High 9 3 13 4.33 2.762 9 41 w. Div. West Low 39 8 4 44 11.00 8 4.103 88 Hightower High 54 6 185 30.83 19 19 16.425 586 513 Hightower Low 3 3 15 5.00 0.000 3 15 0 D. Hollow High 3 0 0.00 0.000 9 9 0 0 -D. Hollow Low 5 161 53.. 67 27.282 268 5 3 224 w. Mamm High ,7 65' 3 13.00 3 5.674 7 91 w. Mamm·Low 0 0.00 0.000 0 9 9 3 0 Grass Mesa Low ·16.40 52 885 132 2165 809 Totals 90% Conf. Int ..on Population: 1356 to 2974 90% Conf. Int. as percen~ of estimte·: 37.38% -----------~-------------------------------------------------------------------- --------------------Quad.rat Flight· 2Garfield High Garfield Low E. Divide Low W.·Div. East High W. Div. East Low w. Div. West iiigh W. Div. West Low Hightower High Hightower Low D. Hollow High D. Hollow Low w. Mamm High w. Mamm Low Grass Mesa Low Totals 10 16 16 6 10 5 9 8 19 3 9 5 7 9 132 10 16 16 6 10 5 9 8 19 3 9 5 7 9 5 5 5 4 3 3 3 386 121 34 147 53 149 4 56 41 6 3 3 3 3 3 53 90%' Conf. Int. dn Population: 2205 tc;> 4145 11 6 77 87 16 189 1373 77 .20 24.20 6.80 36.75 17.67 4~L67 3.67 14.00 6.83 2.00 25.67 29.00 5.33 63.00 24.05 38.032 7.976 2.627 9.937 8.030 6.786 1.905 7 .9·21 3.608 0.000 20.957 10.973 1.764 39.466 772 387 10_9 221 177 248 33 112 130 6 231 145 37 567 3175 626 210 69 98 132 56 28 104 113 0 310 90 20 584 970 90% Conf. Int. as percent of estimate: 30.56% .... ~ �108 Table 14. Radiocollared elk captured on winter range in GMU 42 during December 1994 that dispersed out of GMU 42 to a new winter range as of 27 March 1996. Locations (GMUs) determined by aerial telemetry. Trap New Location.Description Age• Sex GMU Frequency Zone 173.621/94 LOWER WEST MUDDY CREEK C 1+ F 521 173.659/94 MCCLURE PASS HIGHWAY C 1+ F 521 DEADHORSE CREEK 173.66.9/94 D 1+ F 521 173.679/94 TERROR CREEK D 1+ F 521 HIGHTOWER MT. BUZZARD CREEK 173. 689/94 E 1+ 421 F 173.699/94 E SMALLEY GULCH 1+ F 421 173.709/94 D DEADHORSE CREEK 1+ F 521 173.719/94 D VEGA RESERVOIR 1+ 421 F 173.739/94 E MCCLURE PASS HIGHWAY 1+ F 521 173.750/94 1+ E LEROUX CREEK F .52 173.760/94 E COLLIER CREEK 1+ F 421 173.799/94 F 1+ MORRISANA MESA F 42 West 173.829/94 PARACHUTE, WALLACE CREEK F 1+ F 42 West 173.840/94LE MCCLURE PASS HIGHWAY 1+ F F 521 173.859/94 D TERROR CREEK 1+ F 521 173.990/94 C PAONIA RESERVOIR 1+ M 521 BUCK.MESA SOMMERSET 174.030/94LE E 1+ M 521 174.039/94 E HAWXHURST CREEK 1+ M 421 174.049/94 F VEGA RESERVOIR 1+ M 421 174.069/94 E HUBBARD CREEK 1+ M 521 174.080/94 E HAWXHURST CREEK 1+ 421 -~ .• M 174.109/94LE F 1+ GRASSEY GULCH 421 174.129/94 F 1+ LEROUX CREEK M 52 174.150/9:4 F 1+ PORCUPINE CREEK F 42 West 174.181/94 B 1+ LOWER EAST MUDDY CREEK M 521 1 Age of elk in years as of March 1996 b LE= Location elk rout~nely located. ( ( �107 ~ ( \..I ) ( \-.I Appendix I. Mortalities of 86 radiocollared elk from 1 ·oeceniJer 1.993 through 14 June 1996. Age is approximate age in years of elk at death; c=calf 6-12 months old, Y=yearling 12-18 months old. Body weight measured in December when elk caetured as calves. Frequency ID/ !;!ate Trae Bodi Site Zone Sex Ase \IT. ,ksl Heard Dead cause- of Death &Code Nunber Year caetured Legal harvest rifle season-28 BR A 172.030/93 5+ 10/27/95· F Unknown; suspect predation-30 GR B 2+ 06/14/95 172.039/93 F Unknown-11 A 02/12/96 172.070/93 BR F 11 Legal harvest rifle season-28 172.080/93 GM 11/05/94 A F 5+ Wounding loss late rifle season-26 172.090/93 01/16/94 GR B F 11 Wounding loss rifle season-25 172.139/93 BC C 10/19/95 F 17 Legal harvest rifle season-28 cc 2+ 10/23/94 172.160/93 C F Woun<Hng loss rifle season-25 • 172.181/93 11/03/94 MC C F 3 Wounding loss rifle season-25 172.201/93 11/15/94 GS C 3 F Disappear rifle season-22 172.207/93 SG 11/30/95 D F 2+ Legal -harvest archery/muzzle season-27 172.258/93 HY 10/04/94 C F 2+ Wounding loss archery/muzzle season-24 GS 10/04/94 112.2n193 C F 3 Legal harvest· rifle season-28 172.290/93 SG D 10/23/94 F 5+ Legal harvest archery season-33 172.369/93 AC E 09/18/94 F 2+ Legal- harvest late rifle season-29 172.369/94 FM 2+ 01/14/96 B F Wounding loss late rifle season-26 AC E 12/29/94 172.409/93 F 8 Legal ·harvest rifle season-28 FS 172.509/93 10/21/95 H F 3 Lion predation-3 FS 172.542/93 02/01/94 H F 6 Legal harvest late rifle season-29 5+ 172.549/93 PG H 11/28/95 F Wounding loss rifle season-25 PG 172.570/93 11/03/94 F 16 PG· Legal harvest rifle season-28 172.581/93 2+ 10/17/94 F Legal harvest late rifle season-29 FM 12/08/95 172.581/94 F 2+ Unkncwn-11 2+ 172.590/93 PG 04/24/96 F Accident; birthing/calving-32 172.610/93 10/04/94 PG F 3 Accident; birthing/calving-32 172.639/93 06/19/96 PG F 16 Disappear rifle season-22 172.649/93 PG 11/30/94 F 2 Legal harvest late rifle season-29 172.670/93 PG 01/16/95 F ·4 Wounding loss late rifle season-26 172.678/93 PG 12/22/94 F 9 Illegal kill-7 172.690/93 FM F 8 01/?4/94 Legal harvest rifle season-28 172.699/93 FM 5+ 11/05/95 F y Disappear rifle season-22 . SR 172.800/~ F 105.0 11/30/94 Lion predation-3 172.899/93 BC C C 86.0 03/18/94 F Legal harvest rifle season-28 GH 172.950/93 D 2 107.0 10/18/95 F Malnutrition-6 173.000/93 \lM G C 102.0 04/25/94 F Legal harvest rifle season-28 173.041/93 GM A M 2 107.0 10/19/95 Legal harvest -late rifle season-29 173.060/93 MH E 99.0 12/27/95 F 2 Legal harvest rifle season-28 173.120/93 GM A 99.0 10/14/95 M 2 Illegal kill rifle season-9 173.190/93 BC C 111.0 10/29/94 M y Wounding loss rifle season-25 GR 173.201/93 106.0 11/30/95 B M 2 Legal harvest rifle season-28 173.210/93 GC B M 2 108.0 10/19/95 Legal harvest rifle season-28 173.219/93 BC 111.0 11/06/95' C M 2 Illegal kill rifle season-9 173.232/93 BR A M y 101.0 11/15/94 BC Unknown-11 173.262/93 C ·108.0 03/18/94 M C Legal harvest rifle season-28 173~269/93 MC C M 2 121.0 11/06/95 • 173.289/93 MC C 111.0 03/22/94 .Unknown-11 M C 173.300/93 Legal harvest rifle season-28 GS C 119.0 10/24/95 M 2 Disappear rifle season-22 173.309/93 HY C M y -124.0 11/30/94Illegal kill rifle season-9 173.320/93 HY C 118.0 10/20/94 M y 1.egal harvest muzzleloading season-34 173.332/93 GH D M 2 125.0 09/12/95 SG Legal harvest rifle season-28 173.351/93 D .M 2 103.0 11/05/95 Legal harvest archery season-33 173.359/93 AC E 111.0 09/06/95 M 2 Legal harvest rifle season-28 173.370/93 MH E . 141.0 11/08/95 M 2 Disappear rifle season-22 173.390/93 MG D 113.0 11/30/95 M 2 ·Legal harvest rifle season-28 MG 173.402/93 D 2 M 118.0 10/18/95 Wounding loss rifle season-25 173.410/93 WM G M 2 90.0 11/02/95 Legal harvest rifle season-28 173.420/93 G WM 125.0 10/24/95 M 2 _Legal harvest archery season-33 173.429/93 MH .E M 2 126.0 09/14/95 173.439/93 PG H .M 2 o.o 11/30/95 Disappear rifle season-22 Legal harvest rifle season-28 173.450/93 PG H M 2 113.0 10/15/95 173.461/93 Malnutrition-6 PG H M C 82.0 02/07/94 173.461/94 CM A M y 123.0 09/19/95' Wounding loss archery/muzzle season-24 . 173.469/93 FS H M C n.o 04/25/94 Malnutrition-6 173.479/93 Legal harvest archery ~eason-33 FS H M 2 122.0 09/10/95 Legal harvest archery season-33 173.492/93 FS H M 2 110.0 09/03/95 Legal harvest archery season-33 FM 173.510/93 B M 2 91.Q 08/28/95 Legal harvest rifle season-28 173.521/93 FM B M 2 124.0 10/17/95 Legal harvest archery season-33 173.549/94 HM B F y 106.0 09/07/95 y Unknown-1l PR 173.580/94 A 114.0 12/21/95 F -----------------------------------------------------------. -------. -----------------·---------- �108 ~ Appendix I. Frequency ID/ Year Captured 173.589/94 173.640/94 173.789/94 173.870/94 173.919/94 174.059/94 174.101/94 174.119/94 174.140/94 174.170/94 174.220/95 174.230/95 174.240/95 174.339/95 174.500/95 174.609/95 174.689/95 174.789/95 Continued. Trae Date ~ sue Zone Sex Age \IT. Ckg> Heard Dead cause of Death OG -B MC MR C SM D BC MR MM MM MM C FM MS MS MO MO LB GB AP w E E F F f. B F F E E C B D C F F F F M MM M M M M M M F F F M M C C C C y y Y+ C C C C C C C C C C C 117.0 06/01/95 89.0 03/30/95 100.0 03/20/95 86.0 02/21/95 114.0 11/02/95 126.0 "10/17/95 117.0 05/24/96 .140.0 03/01/95 112.0 03/14/95 140.0 04/25/95 127.0. 04/08/96 120.0 . 04/24/96 115.0 04/17/96 79.0 03/27/96 78.0 04/16/96 120.0 04/15/96 100.0 02/05/96 0.0 05/22/96 Unknown; suspect predation-30 Unknown; suspect predatfon-30 Unknown; suspect malnutrition-31 Lion predation-3 Illegal kill rifle season-9 Illegal kill rifle season-9 Unknown; suspect predation-30 Beer predation-35 Lion predation-3 Lion predation .. 3 Unknown; suspect malnutrition-31 Lion predation-3 Lion predation-3 Unknown predator-5 Unknown; suspect malnutrition-31 Malnutrition-6 Lion predation-3 . Unknown; suspect predat i on-30 ( L -o-···_1ii_ 1 :' .. :i \;~:g)'., • �47 Colorado Division of Wildlife Wildlife Research Report July 1997 JOB PROGRESS REPORT State of Colorado Project No. W-153-R-10 Mammals Research work Plan No. - -><-- - -- -- - - - - Elk Investigations Job No. _ _ _ _.,___ _ _ _ _ _ _ _ _ __ Estimating Survival Rates of Elk Developing Techniques to Estimate Population Size Period Covered:; Author: July 1, 1996 - June 30, 1997 D. J. Freddy Personnel: J. Broderick, G. Byrne, J. Ellenberger, D. Fox, J. Frothingham, V. Graham, D. Masden, C. Mehaffy, M. Okata, J. Ritchie, P. Will, R. Winn, CDOW; D. Bowden, G. White, CSU; Diagnostic Laboratory, CSU; N. Miers, D. Ouren, volunteers; BLM Glenwood Springs, CO, NBS Ft. Collins, CO, USFS Rifle, CO, Rocky Mountain Elk Foundation, cooperating . Abstract We captured and radio-collared 70 calf elk (Cervus elaphus nelsoni) 6-months old in December 1996 to estimate survival rates during winter 1996-97 and to increase numbers of radiocollared elk available for experiments utilizing mark-resight models to estimate population density. Elk were captured using helicopter net-gunning and portable corral traps. Survival rates(± 95% CI) for 6-11 month old calves during winter-spring averaged 0.89 ± 0.04 with yearly rates ranging from 0.86 to 0.92 between 1993-94 and 1996-97. Survival was similar among years (P > 0.50) and sexes (P > 0 . 50). Primary causes of death for calves were suspected predation (58%) and suspe cted malnutritio n (26%) . Survival of adult females(~ 12 months old) during winter-spring averaged 0.96 ± 0.02 and was similar among years (P > 0 . 60) with yearly rates ranging from 0 . 94 to 0.98 between 1993-94 and 1996-97 . Hunting was the primary cause of death (59%) for adult females during winter-spring. Survival of elk 18-23 -months of age during winter-spring averaged 0.97 ± 0.03 for males and 0.97 ± 0.04 for females. Annual survival (1 December-30 November ) of adult females averaged 0.81 ± 0 . 06 from 1993-94 through 1995-96 and survival was simi lar among years (P > 0.50) with hunting accounting for 81% of the deaths . Survival during summer-fall for adult females averaged 0.90 ± 0.03 with hunting accounting for 95% of the deaths. Survival during summerfall for elk 12-17 -months of age averaged 0.87 ± 0.07 for males and 0.94 ± 0.05 for females with all deaths attributed to hunting . Surviv al during summer-fall for male elk 24-months old averaged 0.15 ± 0.10 with all deaths due to hunting. Elk population size in 1997 was 1,854 ± 194 (95% CI) based on random quadrat sampling and 3,610 ± 641 based on the JHE pooled mark-resight estimator. These estimates were different (P < 0.05) and continued the disturbing trend of quadrats estimating only 51-93% of the numbers of elk estimated by mark-resight estimators. Adjusting quadrat estimates with �48 sightin g bias formulas did not meaningfully narrow differences observed between quadrats and mark-resight estimators. Discrepancies in estimators reflect the possibility that we may be violating major assumptions(s) underlying either mark-resight or quadrat theory. We believe further field testing to evaluate potential sources of bias will be necessary to evaluate these techniques to estimate population size. ✓ �49 JOB PROGRESS REPORT ESTIMATING SURVIVAL RATES OF ELK AND DEVELOPING TECHNIQUES TO ESTIMATE POPULATION SIZE David J. Freddy P. B, OBJBCTXVB Estimate survival rates of adult female and calf elk and develop techniques to estimate population size. SEGMENT QBJICZXDI 1. Radio-collar 35 male and 35 female calf elk in December 1996 in GMU-42. 2. Estimate winter and annual survival rates of calf and previously radiocollared adult female and male elk from known fates of radiocollared animals. 3. Estimate density of elk in a portion of GMU-42 using 4 helicopter flights involving 2 nonrandom mark-resight flights and 2 flights using a random quadrat sampling system. Apply sighting bias corrections to adjust numbers of elk counted on quadrat flights. compare bias adjusted quadrat estimates with mark-resight estimates based on numbers of marked elk seen during all 4 flights. 4. Analyze data and summarize annually in Federal Aid Job Progress reports. 5. Prepare draft of manuscript on sighting bias models developed from data collected in 1994 and 1995. 6. Continue to monitor locations and movements of selected radiocollared elk. INTRQDUCTXOB our _objectives are to provide reliable estimates of survival rates for calves during winter and for adult females and males throughout the year for· the period 1993-94 through 1996-97. For adults, we are especially interested in survival rates inclusive ~f hunting mortalities which reflects human-induced rates of removal and exclusive of hunting mortalities which reflects natural rates of survival. Additionally, we will develop and test a system for estimating population size that will incorporate estimates of sighting bias in conjunction with a random sampling system using search quadrats as sample units. our winter study area encompasses about 839 lan2 (324 mi2 ) in the eastern half of Game Management Unit 42 south and east of Rifle, Colorado. Elk winter range vegetation types include juniper-pinyon woodland (Juniperus osteosperma-Pinus edulis), oakbrush-mountain shrub (Quercus gambeliiAmelanchier aJ.nirolia), aspen (Populus tremuloides), sagebrush (Artemisia tridentata), and agricultural fields (Freddy 1993, 1994). METHODS Cftpt;ura apd Markip,g we placed radio collars (172-176MHz) having mortality sensors on 70 6-month old calves, of which 35 were males and 35 were females. Calves were trapped �50 from 7-10 Deceml:>er 1996 using helicopter net-gunning. Helicopter capture ~ccurred at 10 remote sites located primarily on public lands. Trapping effort was allocated among 8 geographic trap zones to assure that radioed elk were representative of most if not all se~ents of the population (Table·l). Radio collars were of the same type used in previous years (Freddy 1994). V Calves captured by net-gunning were ferried by helicopter to processing points usually within 1.6 Jan of capture sites. At processing points, body weight, total body length, hind foot length, and rectal body temperature (F) were measured and calves were then radio-collared and released. Body measurements for calves were compared between sexes and years using Proo FREQ and GLM (SAS 1988). sury,iya1 We monitored life or death status of radioed elk during daily ground surveys and aerial surveys conducted at.2-4 week intervals from December 1996 through April 1997 and via aerial surveys at 2-4 week intervals from July to November 1996 and May to June 1997. Life or death status of all calves radioed in December 1996 (70) was known for the period 7 December 1996 through 14 June 1997 which is the time period over which survival rates for calves, age 6-11 months, are calculated. On 15 June, calves by definition become 12-mon~h old yearlings. Life or death status of 151 female and male elk previously collared in December 1993-1995 that had survived to 14 June 1996 was known for 148 of these elk through 15 June 1997. survival ·rates (S) of radioed elk were calculated using the binomial estimator with a variance, VAR(S) = S(l-S)/n (White and Garrott 1990). Survival rates are expressed as the mean estimate± the 95% confidence interval. We used x2 -contingency tests to compare survival rates. We defined 4 major time intervals for survival analyses: winter-spring was 1 December to 14 June, summer-fall was 15 June to 30 November, annual was 1 December to 30 November to coincide with timing of capture and radiocallaring, and yearly, only for yearling elk aged 12-23 months, was 15 June to subsequent 14 June. V Causes of death were estimated from multiple sources of evidence including: presence or absence of gunshot wounds, presence or absence of bite wounds on carcass and predator tracks or scat at carcass site, physical positioning of carcass remains whether buried, covered, scattered, or consolidated, relative amount of internal fat and marrow fat if present with carcass, and results of histopathology and marrow fat analyses (Wade and Browns 1982, Halfpenny and Biesiot 1986). Fat content (percent dry matter) of bone marrow and estimates of age based on dental cementum were obtained for dead elk by the COlorado Division of Wildlife Laboratory while histopathology analyses were provided by the Colorado State University Veterinary Diagnostic Laboratory. Photographs were taken of nearly all mortalities so that physical evidence could be reviewed and judged by outside experts (pers. comm. A. Anderson, T. Beck, w. Andelt). • • Pgpu1at,ion 11t,imates We continued to evaluate methods to estimate elk population size or density during winter in that portion of GMU 42 from East Alkali Creek west to Grass Mesa, south of Rifle, co. In previous years, 951 confidence intervals about population estimates based on quadrat sampling exceeded± 301 of the estimated population size. We addressed this problem by refining strata boundaries and \...,,) �51 by sampling 100\ of those quadrats in high density strata, effectively reducing sampling error to 0 in these high density strata. About 30% of the quadrats in low density strata were randomonly sampled which was similar to previous years. With this strategy, we selected 72 quadrats in 15 strata . representing a 53% sample of the 137 mi 2 (351 Jan2 ) winter range (Table 14). As in winter 1996, each potential quadrat was ·rated as high or low in expected density, individual quadrat boundaries remained unchanged from previous years and were based on topographic features, and quadrats were about 1 mi2 (2.6Jan2 ) in size. We felt this strategy to increase numbers of quadrats flown during 1 flight, instead of flying 2 replicates of a smaller sample of quadrats per flight as originally planned, would be more cost-effective in addressing problems of precision. We counted elk during 3 flights using a Bell-Soley helicopter having a pilot, observer, and navigator-observer, all of whom could potentially detect elk during surveys. One flight per day was flown on 10 and 13 February 1997 to count as many marked and unmarked elk as possible encountered along a nonrandom route within the 137 mi2 area with flying time limited to about 7.5 hours per flight. On 11 and 12 February, marked and unmarked elk were counted on quadrats. Two helicopters were used each day and assigned to different strata to shorten the total time interval over which quadrats were completed. Total flight time for both helicopters was 29 hours. Pilots were common to all 3 flights. The primary observer on nonrandom flights did not participate in quadrat flights. The navigator-observer for the nonrandom flights was 1 of 3 primary observers used on quadrat flights while navigator-observers for quadrat flights only flew quadrats. On February 11 and 12, fixed-wing flights were conducted to confirm locations of 223 radiocollared elk to determine whether these elk were within or outside of the 137 mi2 sample area. We generated estimates of population size for each individual flight and 3 flights pooled using several mark-resight estimators in program.NOREMARK (White 1996). We considered estimates based on a11· 3 flights pooled to be our best estimate of population size and the benchmark against which estunates based on quadrat sampling would be compared. Estimates of population size based.on stratified random quadrat sampling were computed using program DEAMAN (COlo. Div. Wildl. software). We then applied a sighting bias correction formula to each group of elk counted on each quadrat, resulting in adjus~ed counts per quadrat, and recalculated populations estimates based on quadrat sampling. To correct for negative sighting bias, we used a s~ple 2 parameter formula to correct for the probability of detecting an individual group: Ycc:orrectedCJrOup size> = l.1619(Log Group Size(countec1>) + 0.0806 (Freddy 1995). Movements We continued to locate selected radioed elk at least once per month since capture to document seasonal movements via telem~try using a Cessna 185. These elk were originally selected at random from within trap zones ·and equalized by age class in January 1994. Elk from the original sample that died were replaced each subsequent January primarily with randomly selected 6 month-old calves of the same sex from the same trap zone(s) as elk that died. As of 1 January 1997, these 44 elk were classified as 23 adult females, 4 yearling females, 5 female calves, 1 adult male, 5 yearling males, and 6 male calves. Durlng June 1997, we again located an additional 28 adult females, that were originally selected at random in 19.95, to document locations during the calving period. Females from the original sample that died were replaced each subsequent June with adult females randomly selected from the same trap �52 zones of elk that died. movements. As needed, we located other elk to document unusual RESULTS ANP pxscussJoR V capture There were no acute deaths of calves during capture. One male calf (174.619/96) died within 14 days of capture due to the effects of capture myopathy (histopathology, CSU Diagnostic Lab) and was subsequently censored from survival analyses ~Table 3). suryiyal Between 1 December 1993 and 14 June 1997, 146 radiocollared elk died (Table 2, Appendix 1). Hunting was apparently involved in 71% of the deaths. For adults ~12 months old, hunting accounted for 901 of 115 deaths. There were 2 periods of mortality during the year. Adults died primarily from September to January when hunting seasons occurred (Fig. 1) while calves died primarily from February to May (Fig. 2). calves survival for calves during winter-spring averaged 0.89 ± 0.04 (n = 280) with yearly rates ranging from 0.86 to 0.92 (Table 3). We failed to detect differences in survival of calves among years (X23 = 1.5, P >0.50), between sexes pooled among years (X21 = 0.43, P >0.50), and between sexes within each yearly cohort (x21 ~ 0.79, P > 0.40) (Tables 3, 5, 6, 7). Sex of dead calves for all years was 17 male and 14 female (Table 2). These survival rates were associated with winters considered mild in temperature. Snow was usually low or moderate in depth except in 1995 when considerable snow fell during March and April but usually melted rapidly at lower elevations, and in 1997 when a storm in January delivered at least 40 cm of snow on lower elevation winter ranges. Rain in January 1997 subsequently settled this snow to about 25 cm and caused severe crusting of snow. Elk in some segments of the lower winter range moved in excess of 24 Jan to·areas at even lower elevations along the Colorado River near Rulison and Parachute, co soon after this storm occurred. Survival rates of elk, however, remaineq high and apparently were not greatly affected by this weather event. Primary causes of death for all calves during winter-spring were malnutrition (131), suspected malnutrition (13%), mountain lion predation (35%), suspected predation (23%), and unknown cause (10%) (Tables 2, 4). Timing of deaths peaked in March and April for both males and females (Fig. 3). Deaths attributed to predation occurred from January into June and those attributed to malnutrition occurred from February to May (Fig. 2) but both presumed causes of mortality peaked in March and April suggesting a functional change in vulnerability of calves during these months. However, percent fat content in marrow of calves lost to predation and malnutrition suggested that malnutrition may not have predisposed calves to predation. Percent fat in bone marrow of calves suspected of dying from predation was 69 for.males (range 2-95, n = 10) and 36 f~r females (range 8-63, n = 8) and for calves dying from suspected malnutrition, 11 for males (range 0.2-41, n = 4) and 18 for females (range 8-31, n = 3) (Table 4). Possibly elk become more sedentary during March and April to conserve energy reserves and thus more predictable in behavior allowing predators, especially mountain lions, to hunt calves efficiently. V �53 Calf Body Size Averaged over all years, male calves had larger body weights, longer total body length, longer hind leg lengths, and higher condition indexes (PS 0.023) than female calves (Tables 11, 12). Differences in body size were greatest for weight, as males (115 kg) were about 81 larger in mass than females (106 kg). Differences between sexes in other body dimensions were <31. This advantage in mass, however, did not translate into higher rates of survival for male calves. Bddy weights were largest for males in 1995 (119 kg), for females in 1996 (110 kg), and for sexes combined in 1995 (113 kg) (Table 12), but we failed to detect differences in body weights among years (P = 0.15) and there was no interaction between year and sex (P = 0.34). Increases in weights in 1995 were not unexpected as the locally moist summer was favorable to forage production on summer ranges. Weights measured at capture in December for calves suspected of dying from predation (Table 4) averaged 116 kg for males (range 72-140 kg, n = 9) and 96 kg for females (range 78-120, n = 10). Predators thus appeared to take males of average and females of smaller than average size (Table 12). Weights of calves suspected of dying from malnutrition (Table 4) averaged 93 kg for males (range 77-127, n = 5) and 101 kg for females (range 100-102, n = 3). Malnutrition thu~ appeared to affect smaller than average sized males and females (Table 12). Weights of calves dying from all causes aver-aged 108 kg for males (range 72-140, n = 16) and 97 kg for females (range 78-120, n = 14) (Table 4), which was about 71 below average weights at capture for both sexes. Yearlings Yearly survival for elk 12-23 -months of age was 0.84 ± 0.08 for males (n = 90) and 0.91 ± 0.06 for females (n = 97) when averaged among years with hunting deaths included (Tables 5, 6, 7). Survival among years for males ranged from 0.79 to 0.88 and for females from 0.81 to 0.97. We failed to detect differences in yearly survival rates among years for males (X22 = 0.88, P > 0.50) but survival was lower (0.81) for females in 1996-97 (X22 = 5.75, P = 0.06). Yearling males were subjected to about the same rate of illegal hunting during all years which accounted for most of the male mortality, while in 1996-97, females had a lower survival rate because 5 of 6 deaths were hunting related (Tables 5, 6, 7). With hunting deaths included, we failed to detect differences in survival between sexes within (x\ s 2.25, P > 0.15) and among years (x\ = 1.71, P > 0.25). Of 23 deaths, 19 (831) were hunting related. Bunting deaths involved 12 males of which 9 (75%) were classified as illegal kills (Table 2). • Yearly survival for elk 12-23 -months of age was 0.97 ± 0.03 for males (n = 79) and 0.98 ± 0.02 for females (n = 91) when averaged among years with hunting deaths censored (Tables s, 6, 7). Survival among years for males ranged from 0.96 to 1.00 and for females from 0.97 to 1.00. We failed to detect differences· in yearly survival rates among years for males (X22 = 1.14, P > 0.50) and for females (x22 = 1.23, P > 0.50), and between sexes within and among years· (x\ s 0.007, P > 0.90). survival during summer-fall for yearling elk 12-17 -months of age was 0.87 ± 0.07 for males (n c 91) and 0.94 ± 0.05 for females (n = 98) when averaged among years ·and inclusive of hunting deaths, and 1.00 for both males (n = 79) and females (n = 92) with hunting deaths censored (Tables 5, 6, 7). survival among years for males ranged from 0.83 to 0.90 and for females from 0.87 to �54 0.97 inclusive of hunting mortalities. We failed to detect differences in survival rates among years for both males (X22 = 0.70, P > 0.50) and females (x22 = 3.63, P > 0.20) inclusive or exclusive of hunting deaths. Survival rates of females (0.94) were higher than males (0.87) when averaged among years and inclusive of hunting deaths (x2 1 = 2.73, P = 0.10). There were no non-hunting deaths during summer-fall for either males or females (Table 2). V survival during winter-spring for yearling elk 18-23 -months of age was 0.97 ± 0.03 for males (n = 78) and 0.97 ± 0.0.04 for females (n = 91) when averaged among years and inclusive of hunting deaths, and 0.97 ± 0.93 for males (n = 78) and 0.98 ± 0.03 for females (n = 90) wi~h hunting deaths·censored (Tables 5, 6, 7). We failed to detect differences in survival rates among years for both males (X22 = 1.16 P > 0.50) and females (X22 < 2.59, P > 0.30) inclusive or exclusive of hunting deaths. We also failed to detect differences in survival rates between sexes within and among years inclusive or exclusive of hunting deaths (X2 1 < 0.24 P > 0.50). During winter-spring, there were 2 nonhunti~g deaths each for males and females and 1 illegal hunting death for a female (Table 2). Yearling spike-antlered males were·generally not legal quarry during hunting seasons. In 1994, 32 yearling males from the 1993-94 calf cohort entered the hunting seasons presumably as spike-antlered males and 4 (131) were illegally taken during rifle seasons. In 1995, 30 yearling males from ~he 1994-95 calf cohort entered the hunting season and 2 (71) were illegally taken and 1 (31) was fatally wounded during archery season in an area where the elk was legal quarry. In 1996, 29 yearling males· from the 1995-96 calf cohort entered the hunting seasons and 4 (14%) were illegally taken during rifle seasons and 1 (3%) was an assumed wounding loss during rifle seasons because the animal had a 3 x 3 antler point configuration that may have resembled a legal branchantlered male (Tables 2, 5, 6, 7). • V Adult Females Survival during winter-spring for all adult females ~12-months of age was 0.96 ± 0.02 (n = 409) when averaged among years and inclusive of hunting deaths and 0.98 ± 0.01 (n = 399) with hunting deaths censored. Survival among years ranged from 0.94 to 0.98 and from o.·97 to 0.99, inclusive and exclusive of hunting mortalities, respectively. We failed to detect differences in survival among years inclusive (X23 = 1.93, P >= 0.50) or exclusive (X23 = 2.85, P > 0.60) of hunting mortalities (Table 8). There were 17 winter-spring deaths, of which 10 (591) involved hunting: 5 were killed and 3 were wounded during rifle late-seasons and 2 were illegally shot out of hunting seasons (Table 2)'. Seven natural deaths were attributed to mountain lion predation (1), suspected predation (2), complic~tions while calving (1), and unknown cause (3) (Table 2). survival during summer-fall for all adult females ~12-months of age was 0.90 ± 0.03 (n = 372) when averaged among years and inclusive of hunting deaths and 0.99 ± 0.01 (n = 337) with hunting deaths censored. Survival among years ranged from 0.87 to 0.94 and from 0.99 to 1.00, inclusive and exclusive of hunting mortalities, respectively. We failed to detect differences in survival among years inclusive (X22 = 3.31, P > 0.15) or exclusive (x22 = 1.24, P > 0.50) of hunting mortalities (Table 8). Hunting was involved in 35 (95%) of 37 summer-fall deaths: 24 were killed, 8 were wounded, and 3 disappeared during archery, muzzleloading, and rifle seasons. Natural deaths were attributed to suspected predation (1) and complications while calving (1) (Table 2). At the beginning of hunting seasons in fall 1994, 1995, and 1996 V · �55 there were 99, 129, and 142 radiocollared adult females (~12-months of age), respectively, available to hunters (Table 8, non-hunting deaths censored). Percent of marked elk removed by all types of hunting-related mortalities averaged 101 annually (range= 6-12). Annual survival for adult females ~12-months of age marked as a cohort in December 1993 was 0.81 ± 0.06 (n = 163) when averaged among years and inclusive of hunting deaths and 0.96 ± 0.04 (n = 138) with hunting deaths censored. Survival among years ranged from 0.78 to 0.88 and from 0.92 to 0.98, inclusive and exclusive of hunting mortalities, respectively. We failed to detect differences in survival among years inclusive (X21 = 0.19, P > 0.50) or exclusive (x21 = 0.15, P > 0.50) of hunting mortalities (Table 9). For this cohort of adult females, hunting was involved in 25 (81%) of 31 deaths during all years: 13 were killed, 9 were wounded, and 2 disappeared during archery, muzzleloading, and rifle seasons and 1 was illegally killed out of hunting seasons. Natural deaths were attributed to mountain lion predation (1), suspected predation (1), and complications while calving (2). Adult Males Survival during summer-fall for all adult males 24-months of age was 0.15 ± 0.10 (n = 53) when averaged among years and inclusive of hunting deaths and 1.00 (n = 8) with hunting deaths censored. Survival among years ranged from 0.08 to 0.21 inclusive of hunting mortalities. We failed to detect differences in survival among years (x2 1 = 1.86, P > 0.20) (Table 5, 6). All deaths were attributed to hunting. At 24-months of age, males are branchantlered and therefore become legal quarry for hunters. Two of 4 males that lived to be 36-months old survived their second hunting season as legal quarry. Survival during winter-spring for adult males >30-months of age was 1.00 (n = 8) during all years (Tables 5, 6). There were no non-hunting deaths of adult males. Hunting was the cause of mortality among male elk ~24-months of age. From the 1993-94 cohort of 36 male calves, 28 (78%) lived to become 24-month old ·branch-antlered males and enter the 1995 hunting seasons. Of these 28, 22 (791) were harvested in 1995 inclusive of 2 that disappeared during hunting seasons and 2 wounding losses (9% of 22). From the 1994-95 cohort of 33 male calves, 25 lived to 24-months and entered the 1996 hunting seasons. Of these 25, 23 (921) were harvested in 1996 inclusive of 2 that disappeared during hunting seasons, 1 illegally taken during rifle season, and 1 wounding loss (41 of 23). Regular rifle seasons accounted for 33 (701) of the hunting mortalities (Table 2). Most 24-month old males were harvested in the Grand Mesa DAU or in adjacent GMU 43. Exceptions were 3 males killed in GMUs 63 and 52 about 160 Jan south of where these males were captured as calves in GMU 42. The differential impact of hunting on survival and recruitment of males and femal~s to young adult age classes is demonstrated by net survival rates of calf cohorts. For 1993-94 and 1994-~5 calf cohorts, net survival from 6 to 36 months of age averaged 0.09 for males and 0.80 for females (Tables 5, 6). Survival Rates and Population Modeling We incorporated estimates of survival rates, population size, and composition into a simple spreadsheet model to assess potential for population growth or decline (Table 13). We used natural survival rates (hunting deaths censored) �58 for summer-fall and winter-spring time periods for adult males, yearling males, adult females, yearling females, and calves. Hunting deaths were incorporated as ha~vest removal rates for each sex/age class. We assumed survival of calves o-s months of age during summer was 1.0 because the model uses post-season calf:cow ratios ·(measured in January) as an estimate of net calf recruitment to the beginning of winter (December). We also assumed that the recruited sex ratio calves to December is 1:1. Modeling suggests that if harvest rates of antlered elk remained as measured and harvest rates of antlerless elk were reduced to 0, the population could grow at an annual rate of 17%. Current rates of ant~erless harvest allow annual growth rates of 71. To stabilize population growth, harvest rates on antlerless age classes must increase nearly two-fold. Obtaining this level of harvest could be a formidable management challenge. Pqpulation Estimates Elk population size in 1997 was 1,854 ± 194 (951 CI) based on quadrat sampling and 3,610 ± 641 based on the JHE pooled mark-resight estimator (Tables 14, 15, 16). The quadrat sample was lower than the JBE estimate (P s 0.05). We achieved our primary goal of improving precision of the quadrat estimate to facilitate detecting differences in quadrat and mark-resight estimators (Table 16). A disturbing trend has emerged in estimates of population size during the last 3 years. Quadrats provided estimates of size that represented only 51 to 931 of the numbers of elk estimated by pooled mark-resight estimators, excluding initial and preliminary quaclrat sampling in 1995 when the quadrat estimate was 39% of the mark-resight estimate. Mark-resight estimates have ranged from 3,177-3,924 elk based on pooled or individual flights while quadrat estimates have ranged from 1,854-3,387 elk (Table 16). The consistent magnitude in discrepancy among mark-resight and quadrat estimates was demonstrated well in 1997 as the 3, one-sample Lincoln-Peterson mark-resight estimates, whether based on marked and unmarked elk counted on nonrandom flights or on quadrats, were 3,289-3,924 elk while the estimate based on stratified quadrat sampling was 1,854 elk (Table 16). The 3 mark-resight estimates were based on 3 flights that were independent in methodology, strongly suggesting that the probability of detecting marked elk is consistent among different combinations of observers and types of flight patterns. A similar trend among estimates occurred in 1996 (Table 16). We adjusted quadrat estimates for sighting bias using a simple 2 parameter sighting bias model. Adjusting for sighting bias increased quadrat estimates only 7-14% (Table 16). Using more complex sighting models did not meaningfully narrow the differences observed between quadrat and mark-resight estimates. Discrepancies in estimators reflect the real possibility that we may be violating a major assumption(s) underlying either mark-resight estimators or quadrat sampling with sighting bias corrections. Three important potential sources of bias are: 1) detecting and accurately counting groups of elk but missing marked elk within detected groups would inflate mark-resight estimates, 2) overestimating numbers of marked elk within the sample area at the time of flights would inflate mark-resight estimates, and 3) detecting a lower than expected percentage of elk groups on quadrats would deflate quadrat estimates. If bias 1) is true, we suggest that behavior of marked elk when approached by a helicopter would be different than other elk in the same group V �57 with the possibility that marked elk either do not move with the group when the group is initially disturbed or they move away from the group reducing the chances of observers seeing marks. Another possibility is that some fraction of the white collars used as marks were not visible due to discoloration but we tend to discount this possibility because collars that were returned by hunters or were removed from other recovered mortalities were still white. If bias 2) is true, accuracy in determining elk locations during specific census location flights was much poorer than our measured accuracy of ~300 m. We cannot eliminate this source of bias, but each marked elk was located and the estimated GPS location verified on maps delineating the sample area. If bias 3) is true, our estimated sightability of elk groups on quadrats which was 82% and developed during 199 test trials (Freddy 1995) greatly 9verestimates the true detection rate achieved during manag~ent application of quadrat sampling, even though observers used during testing and application remained the same individuals. The magnitude of discrepancy between estimators would further suggest that we not only missed groups on quadrats, but groups relatively large in size, even though sightability of groups ~7 in size exceeded 0.90 (Freddy 1995). We are currently trying to evaluate the most likely source(s) of bias and then develop procedures to test the appropriate hypothesis during winter 1997-98. Elk Movements In December 19·95, 35 male and 34 female calves were radiocollared and of these elk, 24 males and 27 females survived to become 18-months old on 1 December 1996 (Table 7). By the end of winter 1996-97, 21 (7M, 14F) or 41% had dispersed to a winter range outside of the intensive winter range study area in GMU _42 where these elk were initially captured (Table 17). Rates of dispersal were 29% for males and 52% for females. Calves trapped in zones E, F, and G which encompass Alkali, Dry Hollow, and the Mamm creeks, comprised 43% (9)·of the dispersing yearlings and these elk moved priJnarily west to winter ranges near the towns of Collbran and Parachute-Rulison. Calves trapped in zones C and D which encompass West Divide Creek, also comprised 43% of the dispersing juveniles and these elk moved south and southwest to winter ranges near Paonia Reservoir and Hotchkiss. We completed initial draft maps (1:100,000 scale) of a GIS land database for the area frequented by radiocollared elk as part of a cooperative effort through the National Biological Service, Rocky Mountain Elk Foundation, USFS, and BLM. Layers of GIS information include topography, based on 1:24,000 scale USGS quadrangles, land ownership, hydrology, transportation, transportation management zones, winter trapzones for elk, all elk locations, locations of female elk in June, and locations of all elk mortalities. Maps will be reviewed for accuracy by cooperating agencies. we will acquire. vegetation-type coverage for the area and develop a relational database between elk locations and GIS layers to allow descriptive analyses of areas frequented by elk. sighting Bias Draft Manuscript Discrepancies in estimates of population size based on quadrats, sighting-bias adjusted quadrats, and mark-resight estimators prompted us to re-evaluate potential sources of bias in all of these estimators. We therefore, have delayed drafting and submitting for peer review a manuscript evaluating sighting bias adjusted estimates of population size. �58 CQRCLUSJQHS We obtained acceptably precise estimates of calf and adult survival rates and recommend monitoring survival of remaining radiocollared adult male and female elk during the next 2 years. We recommend conducting aerial surveys in 199798 to estimate bias in detecting and counting marked elk on sample quadrats and mark-resight surveys to complete our evaluation of techniques to estimate elk density. LXUMTQRB CJUP Freddy, D. J. 1993. Estimating survival rates of elk and developing techniques to estimate population size. Colo. Div. Wildl. Game Res. Rep. July: 83-117. • Freddy, D. J. 1994. Estimating survival rates of elk and developing techniques to estimate population size. COlo. Div. Wildl. Game Res. Rep. July: 27-42. Freddy, D. J. 1995. Estimating survival rates of elk and developing techniques to estimate population size. Colo. Div. Wildl. Game Res. Rep. July. Halfpenny, J. c., and E. A. Biesiot. 1986. A field guide to mammal tracking in North America. Johnson Books, Boulder, co. 161pp. SAS Institute Inc. 1988. SAS/STAT User's Guide, 6.03. SAS Institute, Inc., Cary, NC. 1028pp. Wade, D. A., and J. E~ Browns. 1982. Procedures for evaluating predation on livestock and wildlife. Texas Agric. Expt. Sta. Publ. B-1429. 42pp. White, G. c. 1996. NOREMARK: Population estimation from mark-resighting surveys. Wildl. Soc. Bull. 24:50-52. White, G. c., and R. A. Garrott. 1990. Analysis of wildlife radio-tracking data·. Academic Press, Inc., San Diego. 383pp. Prepared by - 1 . . U ~ ~ ~ ~ ~ ~ ~ ~ L - - - - - D L V �( ( ( Zone Name 93 24 95 96 umbers of elk rad oco lared i Decerrber Ca Elk Col lac Corral HelfcS!Qter Years Males Iota laea£ 93 94 5 6 93 94 95 96 Total 93 94 95 96 93-24 95 9 A B Garfield Gibson . ·Uncle Bob West Divide Hightower Middle Manm 8 5 6 8 20 8 12 12 24 13 11 12 26 13 11 10 10 10 10 .10 10 8 9 10 8 8 10 8 44 21 6 0 8 6 4 9 25 17 7 7 29 13 14 14 21 11 18 10 17 17 14 12 0 13 9 4 6 0 5 14 35 6b O 0 8 6 4 9 19 7 4 7 29 13 14 14 21 111810 17 17 14 12 0 13 9 4 6 0 5 14 0 0 0 0 .0 0 6 10 0 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 o •o· 35 6 0 0 27 56 70 60 60 26 25 41 3 3 1 3 5 5 4 4 10 7 10 7 5 1 8 3 4 9 8 7 0 8 5 2 2 0 1 9 7 0 0 0 3 7 7 5 2 5 0 38 150 86 75 70 141 83 71 70 100 67 68 70 41 16 3 0 365 36 33 37 35 37 36 34 35 Table Trap C o. E F G H ap ure West Manm Dry Hollow All Obfectjves and 1996 in G Years Total 1 8 7 6 3 3 6 3 6 4 8 10 3 ·s 6 0 5 4 3 0 4 9 0 0 6 Adult Females Collared 93 94 95 96 Total 48 50 26 24 16 4 12 12 10 10 0 1 19 0 6 0 2 0 0 0 6 0 0 0 0 283 68 14 0 23 58 0 0 0 0 0 0 0 0 0 0 0 0 0 4 18 12 12 10 0 1 25 82 •Helicopter= Helicopter net·gtmning, Corral= corral-~rapping. b All 6 adult females captured 2 March 1995 Table 3. Survival rates of 6-11 month old calves with sexes pooled for the winter-spring time period 1 December - 14 June each year i993-94 through 1996-97. Survival rates (S) calculated as a mean estimate of (alive)/(alive + dead) and variance S(l-S)/n collars. Elk Age and Time Period (dates) Calves Calves Calves Calves Calves 12/01/9312/01/9•12/01/95All Years 12/01/96•06/14/94 06/14/95 06/14/96 Pooled 06/14/97 Survival L 951 CI U 951 CI n collars Censored' Died Nonhunting Bunting 0.92 0.85 0.98 73 0 ,6 6 0 0.90 0.83 0.97 69b 0 7 7 0 0.88 0.81 0.96 69 2c 8 8 0 • Censored denotes collar failure and/or animal life/death status not known. ' Includes (173.949/94) male whose collar failed 12194 but seen alive 1195, 1196. c Censored elk were (174.619195) male for slipped collar and (174.800195) male for trap-related mortality. • Censored elk was (174.619196)male for trap-related mortality. 0.86 0.77 0.94 69 ld 10 10 0 0.89 0.85 0.93 280 3c,d 31 31 0 �60 Table 2. Causes of deaths in radiocollared elk between 1 December 1993 and 14 June 1997. Calves (M=male, F=female) were 6-11 months old and collared at 6 months of age. Yearling males and·females were 12-17 months old and juvenile males and females were 18-23 months old and all were collared as 6-month old calves. Adult-males and females were> 24-months old at time of death. Elk AgeLSex Class at Death cause of Death Calves Yearling Juvenile Adult Total and -Code M F F M F M F All M F M Natural Causes Malnutrit;i.on-6 2 2 0 2 2 4 0 0 0 0 0 Unknown-Suspect 3 Malnutrition-31 4 0 0 1 3 1 0 0 0 0 7 7 12 5 Predation-Lion-3 4 0 1 0 0 0 0 Predation-Bear-JS 1 0 1 0 0 0 1 0 0 0 0 0 Unknown Predator-5 0 0 1 0 0 1 1 0 0 0 Unknown-suspect Predation-30 2 11 2 5 3 8 0 1 1 0 0 Accident-Birthing-32 0 0 2 0 2 0 2 0 0 0 0 Accident-Fell-10 0 0 ·J 1 0 0 1 0 1 0 0 2 2 Unknown Cause-11 2 1 0 0 6 0 1 0 Subtotals 17 14 42 .Q. Q .Q. 1· ll 23 a a • Legal Bunting Archery-33 Muzzleloading-34 Archery/Muzzle-27 Rifle-First-28 Rifle-Second-28 Rifle-Third-28 Rifle-Late-29 Subtotals 0 0 0 0 0 0 .Q. 0 0 .Q. .Q. 0 .i Wounding Loss Archery/Muzzle-24 Rifle-Regular-25 Rifle-Late-26 Subtotals 0 0 0 Q 0 0 0 .Q. 1 1 1 0 :Illegal Bunting Rifle-Regular-9 OUt-of-Season-7 Subtotals 0 0 o· 0 Q .Q. • 0 0 .Q. Presumed Bunting Disappear-Rifle Seasons-22 Disappear-Archery/ Muzzle-21 Subtotals Totals 0 0 0 0 0 0 0 0 0 0 0 0 0 2 2 0 0 0 0 0 0 0 0 0 0 0 8 0 4 0 0 0 0 0 13 7 7 0 0 6 .Q. 39 25 0 0 0 0 Q 1 2 1 0 3 1. 9• .2 0 0 .Q. 0 0 .Q. 0 1 1. 0 lb lb 0 0 .Q. .o 0 1. 1. ·17' 14 12 6 0 7 4 0 a 0 4 .Q. 0 0 .Q. 0 2 1 1 6 8 4 4 3 1 0 12 7 1 17 14 11 6 13 7 7 0 39 29 il 2 3 2 6 4 4 7 4 6 9 3 16 .J. 10 0 3 .2 ll 1 0 10 0 -0 0 10 2 1. 1 1. a 0 3b 2b 4 3 0 .Q. 0 .Q. 0 a 1 .2 0 1 !. .J. §.. 2 3 47 45 78 68 146 1 10 2 12 7 • These elk were illegally shot as spike-andeml yearling males: (173.190/93), (173.232/93),(173.320/93), (173.919/94), (174.0S9/94) (174.140/95), (174.200/95), (174.679/9S), (174.719/95). • These elk clisappearcd during rifle hunting seasons and arc presumably dead and legally harvested: (171.207/93), (172.649193), (171.8()()/93), (173.309/93), (173.390/93), (173.439/93), (174.001/94), (174.181194). One exception may be yearling male (173.309/93) that disappeared during die first rifle season in 199:4 when spikc-antleml bplls were not legal. • V �( ( ( Table 4. Estimated causes of mortality and associated body condition for 31 radiocollared calf elk 6-11 months old 1 December 199 - 14 June 1 Marrow Frequency ID/ Trap Age• Date Dead t Fat Cause of Death & Code No. Year Captured Zone Sex (months) 28. soa 86.0 03/18/94 Lion Predation-3 172.899/93 C F 9 8. 03c 102.0 04/25/94 Malnutrition-6 173.000/93 G F 10 28. 88d 108.0 03/18/94 Unknown-11 173.262/93 C M 9 111.0 03/22/94 Unknown-11 173.289/93 C M • 9 N/A1 0.21c 82.0 02/07/94 Malnutrition-6 173.461/93 H M 8 0.27c 77 .-0 04/25/94 Malnutri'tion-6 173.469/93 H M 10 51.71c 1_17. 0 06/01/95 Unknown; suspect predation-30 173.S89/94 B F 12 89.0 17~.640/94 C. F 9 03/30/95 Unknown; suspect predation-30 N/A' 14. 71c 100.0 03/20/95 Unknown; suspect malnutrition-31 173.789/94 E F 9 86.0 02/21/95 Lion Predation-3 173.870/94 D F 8 N/A' 94. 97c 140.0 Bear Predation-35 03/01/95 174.119/94 F M 9 64. 44c .112.0 03/14/95 Lion Predation-3 174.140/94 F M 9 76. 05c: 140.0 04/25/95 Lion Predation-3 174.170/94 B M 10 41. osc 127.0 04/08/96 unknown; suspect malnutrition-31 174.220/95 F M 10 2 .47c 120.0 04/24/96 Lion Predation-3 174.230/95 F M 10 86. 2ac 115.0 04/17/96 Lion Predation-3 174.240/95 E M 10 62. 97c 79.0 03/27/96 unknown Predator-5 174.339/95 E F 9 34. 56c: 78.0 04/16/96 unknown; suspect predation-JO 174.500/95 C F 10 10. 29c 120.0 04/15/96 unknown; suspect predation-30 174.609/95 B F 10 91. goc 100.0 02/05/96 Lion. Predation-3 174.689/95 D M 8 05/22/96 unknown; suspect predation-30 75. 32° 174.789/95 C M 11 79. 99c 72.0 01/08/97 Lion Predation-3 173.370/96 E M 7 69. 36c 116.0 02/19/97 173.381/96 E M 8 unknown; suspect predation-30 30. soc 173.640/96. A F 9 102.0 03/16/97 Malnutrition-6 8 .age 89.5 174.520/96 A F 10 04/25/97 Lion Predation-3 103.0 174.689/96 G M 11 None1 05/14/97 Urµcnown; suspect malnutrition-31 51. 99c 04/09/97" 125.0 174.800/96 E M 10 Lion Predation-3 83. 94c 105.5 174.910/96 D F 8 02/27/97 unknown; suspect predation-JO 91.5 175.059/96 C F 8 02/13/97 N/A1 • Unknown-11 8.03c 111.0 175.130/96 A F 9 03/24/97 Lion·Predation-3 77. 5. l. 99c 175.170/96 D M 9 03/26/97 unknown; suspect malnutrition-31 • Approximate age at death assuming 15 June birthdate. • Whole body weight at capture in December of 1993, 1994, 1995, or 1996. c Pat content as percent dly matter· of bone manow from either femur of carcass. • Pat content as percent dly matter of bone marrow from either mandtl>le of carcass. • Pat content as peicent dly matter of bone marrow from either humerus of carcass. ' No suitable bones remaining with carcass. • No marrow observ~ in one long bone remaining with carcass. �Table 5. Survival rates for winter-spring and summer-fall time periods from 1 December 1993-14 June 1997· for the cohort of 6-morith old calves radiocollared in December 1993. Survival rates for male and female calves pooled when 6-11 months old were 0.92 (95% CI 0.86-0.98, n=73). Survival rates (S) calculated as a mean estimate of Calive)/Calive + dead) and variance SCl-S)/n collars. Elk Age (months) and Time Period (dates) 18 - 23 24 - 29 30 - 35 36 - 41 42 - 47 6 - 47 12 - 17 6 - 11 12/01/96._ 12/01/9406/15/9512/01/9506/15/9612/01/9306/15/9412/01/9306/14/95 11/30/95 06/14/96 11/30/96 06/14/97 11/30/94 06/14/97 06/14/94 MALES Survival L 951 CI U 951 CI n collars Censored• . Died Nonhunting Hunting 0.89 Q.78 0.99 36 0.88 0.76 0.99 32 1.00 0.21 0.06 0.37 1.00 0.00 0.00 1.00 28 4 4 0 4 4 0 0 29c 0 0.50 0.00 1.00 2c 0 2' 0 4b 0 0 2 0 32 0 4 0 22d 0 0 0 0 ·4 0 22 0 2 0 28 0.97 0.92 1.00 35 1.00 0.97 0.91 1.00 0.84 0.71 0.97 1.00 34 0~97 0.91 1.00 34 33 32 27 0.73 0.58 0.88 37 0 1' 0 0 0 0 0 0 0 1 l 0 l 5 0 ·10 0 0 0 2 8 0 FEMALES Survival L 951 CI U 951 CI n collars Censored• Died Nonhuriting ·Hunting 0.95 0.87 1.00 37 0 2 2 0 0 1 0 0 0 1 • Censored denotes collar failure and/or animal life/death status not known. 11 Collar (173.309/93) •disappeatecr during Fall 1994 hunting seasons and assumed dead. c Includes collar (173.241/93) that failed in August 1995 but bull seen alive January 1996. • Two collars (173.390/93), (173.439/93) "disappeared" during Fall 1995 hunting seasons and assumed dead. • Two collars censoml: (173.241/93) failed, (173.381/93) slipped off 12/01/95. ' Collar (172.800/93) •disappeared" during Fall 1994 hunting seasons and assumed dead. • Includes (173.340193) alive 5/17/'Tl. C C 5 0.06 0.00 0.14 34 2c �( ( ( Table 6. Survival·rates for winter-spring and summer-fall time periods from 1 December 1994-14 June .1997 for the cohort of 6-month old cal v.es radiocollared 'in December 1994. Survival rates for male and female calves·pooled when 6-11 months old were 0.90 (95% CI 0.83-0.97, na69). Survival rates (S) calculated as a mean estimate of {alive)/{alive + dead) and variance S(l-S)/n collars. . Elk Age (months) and Time Period (dates) 12 - 17 18 - 23 24 - 29 30 - 35 6 - 35 6 - 11 06/15/9512/01/9506/15/9612/01/9612/01/9412/01/9406/14/96 11/30/96 06/14/97 06/14/97 11/30/95 06/14/95 MALES Survival 0.91 L 951 CI 0.81 U 951 CI n collars Censored• Died Nonhunting Bunting 1.00 33b 0 3 3 0 0.90 0.79 1.00 30b 0 3 0 3 FEMALES Survival L 951 CI. U 95' CI n collars Censored• Died • Nonhunting Bunting 0.89 0.78 0.99 36 0 4 4 0 0.97 0.91 1.00 32 0 1 0 1 0.96 0.88 1.00 26 1c 1 1 0 0.97 0.90 1.00 30 ld 1 l· 0 . 0.08 0.00 0.19 25 0 23 0 23 ·1.00 0.93 0.83 1.00 29 0 2 0 2 0.96 • Censored denotes collar failure and/or animal life/death status not known. ~ Includes collar (173.949/94) that failed in December 1994 but male seen alive in January 1996 • Censored elk was (173.949/94)failure. which was killed in first rifle season 1996. • Censored elk was (173.719/94) slipped collar. • Includes (174.030/94)alive 6/4/97. 2e 0 0 0 0 0.89 1.00 27 0 1 0 1 0.06 0.00 0.15 32 1c 3Q 4 26 0.74 0.59 0.89 35 ld 9 5 4 �Table 7. Survival rates for winter-spring and summer-fall time periods from l December 1995-14 June 1997 for the cohort of 6-month old calves radiocollared in December 1995 and survival rates for winter-spring for 6-11 month old elk calves radiocollared in December 1996. Survival rates for male and female calves pooled when 6-11 months old were 0.88 in 1995 (95% CI 0.81-0.96, n=69) and 0.86 in 1996 (95% CI 0.77-0.94, n=69). Survival rates (Sl calculated as a mean estimate of (alive)/(alive+dead) and variance of S{l-S)/n collars. Elk Age {months) and Time Period (dates) • ·1995 Calf Cohort 1996 Calf Cohort 6·;_ 11 6 - 23 18 - 23 12 - 17 6 - 11 12/01/9612/0l/9506/15/9612/0l/9512/0l/9606/14/97 06/14/97 11/30/96 06/14/96 06/14/97 MALES Survival L 951 CI U 951 CI n collars Censored• Died Nonhunting Bunting 0.86 0.74 0.99· 35 2b. 5 5 0 FEMALES Survival L 951 CI U 951 CI n collars Censored• Died Nonhunting Hunting 0.91 0.81 1.00 34 0 3 3 0 0.83 ·o.68 0.97 29 1c 5 0 5 0.96 0.87 1.00 24 0 l 1 0.68 0.51 0.84 34 3b,c 11 6 0 5 5 5 0 0.87 0.75 0.99 31 0 0.93 0.82 1.00 27 0 0.86 0.98 0.74 35 0 4 0 2 1 4 l 0.74 0.58 0.89 34 0 9 4 5 0.85 0.73 0.98 34 ld 5 5 0 • Censored denotes collar failure and/or animal life/death stablS not known. • Censored elk were (174.619/95) slipped collar, (174.800/95) capture mortality. • Censored elk was (174.660195) slipped collar July 1997. • Censored elk was (174.619196) male trap-related mortality. C C C �( ( ( Table 8. Survival rates for winter-spring and sununer-fall. time periods from 1 December 1993 - 14 June 1997 for all radiocollared adult female elk ~12-months old. Survival rates .(S) calculated as a mean estimate of (alive)/(alive + dead} and variance S(l-S)/n collars. Elk Age and Time Period (dates) Adult Adult Adult Adult Adult Adult Adult 06/15/9412/01/9406/15/9512/01/9506/15/9612/01/9612/01/9311/30/94 11/30/95 06/14/95 11/30/96 06/14/97 06/14/9~ ~6/14/94 Survival L 951 CI U 951 CI n collars censored• Died Nqnhunting Hunting 0.96 0.91 1.00 68be 0 3 1 2 0.87 0.80 0.94 0.96 0.92 1.00 lOObf . 95b1 0 13c 1 12 0 4 1 3 0.94 0.90 0.98 129bh 0 9d 0 8 • Censored denotes collar fa11ure and/or animal life/death stams unknown. c Collars (172:800/93,172.649/93)"disappemc:r Fall 1994 hunting seasons,assumeddead. • Composition is 6 females 18+ and 62 females 30+ months old. • Composed of 34-18+, and 61-30+ months old females. 1 aomposed of 30-18+ ·and 89-30+ months old females. k Composed of 31-12+. 29-24+, and 83-36+ months old females. 0.94 0.90 0.98 1191 • 2J 7 4 3 0.89 .0.84 0.94 1431t 0 '16 1 15 0.98 0.95 1.00 1271 0 3 1 2 11 Includes collar (172.011/93) that failed 4/1994 but seen alive in 1/1996 • Collar (172.207/93) "disappeared" Fall 1995 hunting seasons, assumed dead. ' Composed of 35-12+, 6-24+, and 59-36+ months old females. h Composed of 32-12+, 34-24+, and 63-36+ months old females. J Censored (172.011/93) failwe and (173.719/94) slipped collar. 1 Composedof27-18+ and 100-30+ month old females. Table 9 . . survival rates for winter-spring and sumroer-fail time periods from 1 December 1993 - 14 June 1997 for the group of adult female elk that were ~12-months old when radiocollared in December 1993. survival rates (S) calculated as a mean estimate of Calive)/(alive + dead) and variance S(l-S)/n collars. Elk Age and Time Period (dates) Adult Adult Adult • Adult Adult Adult Adult 12/01/93- 06/15/94- 12/01/94- 06/15/95- 12/01/95- 06/15/96- 12/01/9606/14/94 11/30/94 06/14/95 11/30/95 06/14/96 11/30/96 06/14/97 0.82 0.96 0.93 0.88 0.93 0~92 Survival 1.00 0.72 0.91 0.85 0.78 0.85 0.84 L 951 CI 0.91 1.00 0.99 0.97 1.00 1.00 U 951 CI 68be 65bf 53b f 49b f 42f 39 1 36r n collars 0 0 1i 0 Censored• 0 0 0 4 6d 3 3 3 Died 0 Nonhunting 1 1 0 3 0 0 2 Bunting 3 6 0 3 0 • Censored denotes collar failure and/or animal life/death status unknown • Collar (172.649/93) "disappeared" Fall 1994 hunting seasons, assumed dead. • . Composed of 6-18+ and 62-30+ month old females. 1 . Censored (172.011/93) failure of unknown status spring 1996. 11 Includes collar (172.0ll/93)failed 411994 but seen alive 1/1996. • Collar (172.207/93) •disappeared" Fall 1995 hunting seasons, assumed dead. r Composed of 100% females 24+ months old. ffl �Table 10. Annual ·survival rates from 1 December 1993-30 November 1996 for the group of adult female elk that were ~12-months old when radiocollared in December 1993. Survival rates (S) calculated as a mean estimate of (al~ve)/(alive+dead) and variance S(l-S)/n collars. Adult Survival L 951 CI U 951 CI n collars Censored• Died· Nonhunting Bunting Elk Age and.Time Period (dates) Adult Adult Adult 12/01/93~11/30/94 0.78 0.68 0.88 12/0i/94-11/30/95 0.84 0.75 0.93 12/01/95-11/30/96 0.86. 0.75 0.97 12/01/93-06/14/97 0.54 0.42 0.66 . 68b,f 53b,f 42f 19· 67f 19 6 3 31 3 25 0 15c 0 ·2 1. 9 iOd 13 • Censored denotes collar failure and/or animal Jif~d~th etatua unknown. • Collar (172.649/9~) "disappeared" Fall 1994 bunting seasons, assumed dead. • Compoeed of 6-18+ and 62-30+ month old females. • 1 Censored (172.011/93) failure 4/1996. 6 b Includes collar (172.011/98) failed 4/1994 but seen alive 1/1996. Collar (172.207/93) "disappeared" Fall 1995 bunting seasons, assumed dead. ' Composed of 100% females 24+ months old. d Table 13. Data matrix for simple population model using spreadsheet software. We used a post-hunt (December) population composition of 50 calves:100 females (age 2 12-months) and 4 adult males:16 yearling: 100 females and an estimated population of 3,600 e~k. Annual population growth rate is about 7% when using this matrix Fall.Hunting Survival Rate Survival Rate Harvest Rate • Elk Sex/Age Class Winter-Spring summer-Fall Elk Sex/Age Class· Adult Males 0.82 1.00 1.00 Adult Males Age 230 months Age 24-29 months Yearling Males Age 12-17 months :t.00 0.13 Yearling Males Age 1·8-23 months 0.97 Adult Females Age 224 months 0.99 0.10 Adult Females Age ~30 months 0.98 Yearling Females Age 12-17 months 1.00 0.06 Yearling Females Age 18-2·3 months 0.97 Calves Age o-s months 1.00 o. 02 .Calves Age 6-11 months 0.89 1 ~stimated from harvest surveys. ( C ( �( ( ( Table 11. Frequency distribution of whole body weights for male and female elk calves trapped in Game Management unit 42. December. 1993~1996. Percentage per weight class shown in parentheses. Body Weight Class (kg) • Year sex 60-69 70-79 80-89 90-99 100-109 110-119 120-129 130-139 140-149 Total 1993 1994 1995 1996 ALL 1993 1994 1995 1996 ALL M M M M M F F F F ·F 0 (0. 0) 0 (0. 0) • 0(0.0) 0(0.0) 0(0.0) 0(0.0) 0(0.0) 1 (3 .1) 0(0.0) 1(0.7) 1 (2. 9) 1(3.0) 0(0.0) 3(8.6) 5(3.6) 1(2.9) 2(5.7) 2(6~3) 0(0.0) 5(3.7) 1( 2~9) 1( 3.0) 3( 8.6) 2( 6.0) 6(17.1) 6(18.2) 0 ( 0. 0) 4 ( 11. 4) 3 ( 8. 6) 0( 0.0) 3( 8.6) 6(17.1) 2( 1.4) 12( 8.7) 21(15.2) 12(34.3) 13(39.4) 8(22.9) 5(14.3) 38(27.50 4(11.4) 8(22.9) 4(11.4) 7120.0) 1( 3.1) 3( 9.4) l(' 2.9) 6(17.6) 10( 7.4) 24(17.6) 4(11.4) 10(28.6) 5(15.6) 10(29.4) 29(21.3) 12(34.3) 10(28.6) 14(43.8) 8(23.5) 44(32.4) 10(28.6) 6(18.2) 9(25.7) 13(37.1) 38 (27 -:5) 4 (2. 9) 35(100) 33(100) 35(100) 35(100) 138(100) 0(0.0) 0(0.0) 0(0.0) 0(0.0) 0(0.0) 35(100) 35(iOO) 32(100) 34 (100) 136(100) 1 ( 2. 9) 1(2.9) 2c 2 (6. 0) 1 (2. 9) 6·.o> 10(28.6) 5(14.3) 18 ('13. 0) 6(17~1) 0 ( 0. 0) 2 ( 5. 7) 6 (18 .·8) 0( 0.0) 8(23.5) 22(16.1) 1 ( 2. 9) 1 ( 0. 7) 0 ( 0. 0) 0(0.0) Body measurements for elk calves trapped in Game Management Unit 42. December. 1993-1996. Male Calves Female Calves Mean SD Min Max n Mean SD Min Max n Measurement 112.7 13.7 77.0 141.0 35 103.8 12.2 76.0 123.0 35 Body Weight (kg) ·191.2 10.2 164.0 210.0 36 188.9 9.8 167.0 207.0 35 Body Length. (cm) 2.2 52.0 61.0 36 54.8 2.1 51.0 59.0 34 Hindfoot Length (cm) 56.3 0.59 0.05 0.43 0.67 35 0.55 0.05 0.46 0.64· 34 Condition Ind~r Table 12. Year 1993 113.5 Body Weight (kg) Body Length (cm) 190.3 Hindfoot Length (cm) 56.9 0.59 Condition Index 14.2 9.2 1.8 0.05 71.0 168.0 50.0 0.42 140.0 204.0 59.0 0.69 33 32 32 32 103.0 186.3 55.1 0.55 13.3 2.1 0.06 70.0 128.0 166.0' 207.0 50.0 59.0 0.42 0.65 1995 Body Weight (kg) 119.l 194.0 Body Length (cm) Hindfoot Length (cm) 57.4 0.61 Condition Index 13.6 9.3 2.2 0.05 92.0 166.0 53.0 0.50 141.0 206.0 61.0 0~71 35 36 36 34 105.7 189.4 54 ..9 0.56 15.2 11.4 2.3 0.06 60.0 158.0 47.0 0.38 129.0 • 32 203.0 34 59.0 34 0.66 32 1996 114.3 Body Weight (kg) Body Length (cm) 187.6 Hindfoot Length (cm) 57.7 0.61 Condition Index 17.1 11.5 2.3 0.07 72.0 160.0 52.0 0.44 136.0 202.5 60.5 0.70 35 35 34 35 109.5 187.4 56.8 0.58 12.4 9.0 2.4 0.05 81.5 170.5 52.5 0.47 136.0 210.0 62.0 0.69 '114.9 Al.1- Body Weight {kg) 190.8 Years Body Length (cm) Hindfoot Length (cm) 57.1 0.60 Condition Index 14.8 10.3· 2.2 0.06 71.0 160.0 50.0 0.42- 141.0 1·38 210.0 139 61.0 138 0.71 136 105.5 188.0 55.4 0.56 13.4 9.8· 2.3 0.06 60.0 158.0 47.0 0.38 136.0-136 210.0 140 62.0 138 0.69 135 1994 • Condition index= (Weight/body length). 8.8 35 36 36 35 34 35 34 34 �20 Table 14. Colorado. Counts of elk on quadrats 11 and 12 February, 1997 in GMO 42 south of Rifle .and Newcastle, Strata Size (mi2> Strata Totals 10 17 16 6 11 10 17 16 6 11 Garfie;Ld High Garfield Low B. Divide Low w. Div. Bast High· w. Div. Bast Low W. Div. West High W. Div. West Low Hightower Low Upper Alkali Ck. High Upper Alkali Ck. Low Low.Alkali Ck. Low Upper DryHollow Low Lower Mamm Ck. Low W.Mamm-Grass Mesa Low W.Mamm-Grass Mesa High Qyadrats N n 5 5 9 9 4 5 10 10 13 9 9 4 5 10 10 13 10 8 3 6 4 4 3 3 4 3 5 Elk Counted 191 54 8 205 49 89 27 0 88 48 85 ElklQJaadrat Mean StdErr. Po:gulation Size Total+ 90% CI 19.10 6.75 2.67 34.17 12.25 22.25 9.00 0.00 22.00 16.00 17.00 191 115 43 205 135 111 81 0 4 . '58 86 14 17.20 4.67 244 34.86 1246 13.54 5 5 7 7 5 3 7 137 137 72 14.50 0.000 1.859 2.404 0.000 3.573 2.991 2.449 0.000 0.000 5.514 4 .• 738 3.309 2.423 1.520 0.000 0 52 63 0 65 25 36 0 88 0 80 170 45 78 54 52 145 224 23 244 13 0 1854 164 Table 15. Summary of helicopter flights used to estimate· elk population size and density on 137 mi2 of winter range in GMO 42 south of Rifle and Newcastle, Colorado, during February. 1997. Proportion ~arked Total Elk Marked Elk Marked Elk Elk Proportion Flight Available• Counted Counted Counted Hours Counted Dates Marked Type NonRandoml 2/10 Quadrats 2/11-12 NonRandom2 2/13 7.5 29.0 7.8 120 120 120 31 43 33 29 40 29 0.2583 0.3583 0.2750 838 1203 1069 869 1246 1102 0.0357 0.0345 0.0299 ~umber of marked elk within the 132 mi2 sample area. "Number of marked elk individually identified during flights based on number/symbol identifier seen on collar. ( C �( ( Table 1.6 .• Summary of flights to estimate .elk population size from 1994-1997 in GMU 42 south of Rifle and Newcastle 1 ColoradQ. Area (mi2) Minimum Elk 95% C. I. Population Approximate Method Flownllirs• %Precision CountedLFlight Estimate 95% Conf. Int. Estimator Flight Date Feb. 1997 Feb. 1997 Feb. 1997 Feb. 1997 Feb .. 1997 'All 1997 Quadrats-StRSb Quadrats~stRS~sBAc Quadrats-StRS-LPd NonRandom Flight,REPl-LP NonRandom Flight,REP2-LP 3 Flights Pooled-JHEC 137/29.0 137/29.0 137/29.0 137/ 7.5 137/ 7.8 137/44.3 1246 1246 1246 869 1102 1246 1.854 2105 3428 3289 3924 3610 1660-2048 1851-2359 2643-4213 2344-4233 2839-5010 3116-4251 11 12 23 -29 28 18 Jan. 1996 Jan. 1996 Jan. 1996 Mar. 1996 Mar. 1996 Mar. 1996 Jan. 1996 Mar. 1996 All 1996 All 1996 Quadrats,REPl-StRS Quadrats,REPl-StRS-SBA Quadrats,REPl.-StRS-LP Quadrats,REP2-StRS Quadrats,RBP2-StRS-SBA Quadrats,REP2-StRS-LP NonRandom Flight,REPl.-LP NonRandom Flight,REP2-LP 4 Flights Pooled-JHE 4 Flights Pooled-IEJHEe 132/19.6 132/19.6 132/19.6 132/19.l 132/1"9 .1 132/19.l 132/ 8.5 132/ 8.0 13·2/55. 2 132/55.2 885 885 885 1373 1373 1373 1354 1998 1998 1998 2165 2336 3463 3175 3387 3791 3177 3405 3472 3415 1265-3065 1294-3378 2477-4448 2098-4252 2189-4585 2975-4607 2560-3795 2941-3869 3168-3840 3129-3807. 42 45 28 34 35 Feb. 1995 Feb. 1995 Feb. _1995 Quadrats-StRS Quadrats-StRS-SBA Quadrats-StRS-LP 177/17.5 177/17.5 177 /11. 5 361 36.l 361 1484 169_8 3774 787-2181 928-2468 1993·-5555 8Hours of helicopter time per flight. 'Stratified random ·sampte of quadrats with counts of elk not adjusted for sighting bias. •Stratified random sample of quadrats with counts of elk adjusted for sighting bias. 'One sample Lincoln-Peterson mark-resight estimator. •Joint Hypergeometric mark-resight estimator for nonrandom and quadrat(unadjusted)flights pooled. . 'Immigration/Emigration Joint Hypergeometric mark-resight estimator that allows for movement of elk on/off intensive study area. 22 19 14 11 12 47 45 47 �70 Table 17. Radiocollared elk captured as calves on winter range in GMO 42 during December 1995. that dispersed out of GMO 42 to a new winter range as of Locations {GMOs} determined b~ aerial telemetry. March 1~97. Trap New Fregyency Age• Sex GMO Zone Location Descri~tion 173.899/95 F 21 F 42 West Battlement Creek F 173.909/95LE 21 F 421 Hawxhurst Creek 174.301/95 E 21 F Porucpine Creek 42 West 174-. 360/95 E 21 F Wallace Creek 42 West 174.370/95 E 21 F 521 Drift Creek F 174.391/95 D 21 Paonia Reservoir 521 174.431/95LE D 21 F Buck Mesa 521 174.451/95 D 21 F Thompson _Creek 43 174.491/95 C 21 F Holms Mesa 42 West B 174.532/95 21 F Thompson Creek 43 174.551/95 G 21 F Porcupine Creek 42 West 174.580/95 G 21 F .Porcupine Creek 42 West 174.589/95LE G 21 421 Hawxhurst Creek F A 174.600/95 21 F 42 West Holms Mesa 174.119/95 B 21 M 42 West Holms Mesa E 174.629/95LE 21 M 42 West Rulison D 174.719/95 21 Buck Mesa M 521 Grassey·Gulch C 174.770/95 21 M 421 C Anthracite Creek 174.780/95 21 M ·53 174.809/95 C 21 Sommerset M 521 174.851/95 C 21 M 52 Oak Mesa V • Age of elk in months as of March 1997. b LE-~ Location elk routinely located. ~ V �71 30 25 '--/ c::::n ADULTMALES 0 w 20 a:: w en 15 z 10 a :i!: ::, - ADULT FEMALES ~~ ~ n=61 n=54 5 0 , ~<I:- ., ~~ ~" #' 'b<8 MONTH Fig. 1. Timing of deaths for adut elk >=12-months of age, 1993-94-1996-97. 12 - 10 0 w 8 w m 6 z 4 0 a:: :E ::, ~ CALVES n =31 !!!!!!! PREDATION U;lmmm MALNUTRITION :I I I ______O_T_H_E_R_ _ _ _ _ __ mi.............. 2 0 JAN FEB MAR MAY APR JUN MONlH Fig. 2. liming and estimated causes of deaths for calf elk 1993-94 - 1996-97. Calves were of both sexes and 6-11 months of age. 7 6 0 5 - Uiliilii!i!I FEMALES n = 14 • MALES n=17 w c 0::: 4 w ~ => z ·111111 _______________ 3 UnH ·:::: 2 1 0 ~· JAN FEB APR MAR MAY JUN MONlH Fig. ·3. limi~ of deaths for male and female calf elk 1993-94 -1996-97. Calves were 6-11 montt:,s of age. �72 Appendix I. Mortalities of 146 radiocollared elk from 1 Oeced)er 1993 through 14 -June 1997. Age is eeeroxfmate e9e in iears of elk at death: C=calf 6-11 months old 1 Y:Xearlf!JS 12-23 months old. Frequency 10/ Death Tree Year caetured Site Zone Sex Age Date Cause of Death &Code Nu:nber 172.030/93 F BR A 5 27-0ct-95 Legal kill rifle season-28 172.039/93 GR B F Unknown-suspect predation-30 4 14-Jun-95 172.070/93 BR A F 11 12-Feb-96 Unknown-11 172.080/93 GM A F 6 05-Nov-94 Legal kill rifle season-28 GR B F 11 16-Jan-94 172.090/93 Wounding loss late rifle season-26 172.101/93 F SR B Legal kill rifle season-28 • 8 14-oct-96 GC B F 172.128/93 8 31-0Ct-96 Wounding loss rifle season-25 F 172.139/93 BC C 17 19-0Ct-95 Wounding loss rifle season-25 F 172.160/93 cc C 3 23-oct-94 Legal kill rifle season-28 172.181/93 MC C F 3 03-Nov-94 Wounding loss rifle season-25 172.201/93 GS C F 3 15-Nov-94 Wounding l_oss rifle season-25 172.207/93 SG D F 4 30-Nov-95 Disappear rifle season-22 . 172.258/93 HY C F 3 04-0ct-94 • Legal kill archery/n1.1zzle season-27 GS C 172.277/93 F 3 04-0ct-94 Wounding loss archery/n1.1zzl~-24 172.290/93 F SG D 6 23-0ct-94 Legal kill rifle season-28 172.290/94 SM D F 4 16-Sep-96 Legal kill nuzzleloading season-34 172.369/93 F AC E 18-Sep-94 3 Legal kill archery season-33 172.369/94 F FM B 4 Legal kill late rifle season-29 14-Jan-96 172.409/93 F AC E 29-Dec-94 8 Wounding loss late rifle season-26 MH E F 8 172.459/93 Legal kill rifle season-28 24-0Ct-96 FS ff F 3 21-0Ct-95 172.509/93 Legal kill rifle season-28 FS ff 172.542/93 F 6 01-Feb-94 Mountain lion predation-3 172.549/93 PG ff F 28-Nov-95 7 Legal kill late rifle season-29 172.570/93 PG ff F 16 03-Nov-94 Wounding loss rifle season-25 172.581/93 PG H 3 F Legal kill rifle season•28 17-oct-94 F 172.581/94 FM B 3 08-Dec-95 Legal kill late r.ffle season-29 F 172.590/93 PG ff 5 24-Apr-96 Unknown-11 F 172.610/93 • PG ff 3 04-0Ct-94 Accident, birthing/calving-32 F 172.639/93 PG ff 16 14-Jun-96 Accident, birthing/calving-32 172.649/93 F PG ff 2 30-Nov-94 Disappear rifle season-22 172.670/93 F PG H Legal kill late rifle season-29· 16-Jan-95 4 PG H F 9 22-Dec-94 Wounding loss late rifle season-26 172.678/93 F FM B 8 172.690/93 Illegal kill-7 24-Jan-94 172.699/93 FM B F 7 05-Nov-95 Legal kill rifle season-28 172.749/95 LS ff F 11 07-Aug-96 Unknown-suspect predation-30 y SR B 172.800/93 F 30-Nov-94 Disappear rifle season-22 172.821/93 EG B F 3 08-Sep-96 Legal kill archery season-33 F 172.890/93 BC C 3 06-Nov-96 Legal kill rifle season-28 172.899/93 BC C F C 18-Mar-94 Mountain lion predation-3 172.950/93 GH D F 2 18-0ct-95 ·Legal kill rifle season-28 F 173.000/93 WM G C 25-Apr-94 Malnutrition-6 173.041/93 GM A M 2 19-0Ct-95 Legal kill rifle season-28 MH E F 2 27-Dec-95 173.060/93 Legal kill late rifle season-29 173.081/93 FS H F Legal kill rifle season-28 3 22-0ct-96 GM A 173.120/93 M 2 14-0ct-95 Legal kill rifle season-28 173.140/93 PG H F 3 31-0Ct-96 Wounding loss rifle season-25 FS H 173.160/93 F 3 13-Nov-96 Legal kill rifle season-28 y 173.190/93 BC C M 29-oct-94 . Illegal kill rifle season-9 173.201/93 GR B M 2 30-Nov-95 Wounding loss rifle season-25 173.210/93 GC B M 2 19-0ct-95 Legal kill rifle season-28 BC C M 173.219/93 06-Nov-95 2 Legal kill rifle season-28 y 173.232/93 BR A M 15-Nov-94 Illegal kill rifle season-9 173.262/93 BC C M C 18-Mar-94 Unknown-11 173.262/94 HM B M 2 12-0ct-96 Legal ki U rifle ~eason-28 173.269/93 MC C M 2 06-Nov-95 Legal kill rifle season-28 3 MC C 173.279/93 M 12-0Ct-96 Legal kill rifle season-28 173.289/93 C 22.;.Mar-94 MC M C Unknown-11 173.289/94 PR A M 2 18-Sep-~6 Legal kill nuzzleloading season-34 GS C 173.300/93 M 2 Legal kill rifle season-28 24-0ct-95 y 173.309/93" HY C M 30-Nov-94 Disappear rifl_e season-22 y 173.320/93. HY C M 20-0ct-94 Illegal ldll rifle season-9 173.320/94 HM B M 2 Legal kill archery season-33 14·Sep·96 173~332/93 GH D M 2 12-Sep-95 Legal kill nuzzleloading season-34 SG · D 173.351/93 M 2 05-Nov-95 Legal kill rifle season-28 173.359/93 AC E M 2 06-sep-95 Legal kill archery season-33 173.370/93 MH E M 2 08-Nov-95 Legal kill rifle season-28 M· RG E 173.370/96 C 08-Jan-97 Mountain lion predation-3 RG E 173.381/96 M C 19-Feb-97 Unknown-suspect predation-30 MG D 173.390/93 M 2 • 30-Noy.:.95 Disappear rifle season-22 MG D M Legal kill rifle season-28 173.402/93 2 18-0ct-95 G 173.410/93 Wounding loss rifle .season-25 WM M 2 02-Nov-95 f7.f~420/93 M WM G. 2 24-oct-95 Legal ~~ Ll rff le season-28__ --------------------------------------------------- --------------------------------~--------------- V V' V �73 Appendix 1. <continued> Frequency ID/ Trap Year Captured Site Zone Sex 173.429/93 MH E M 173.439/93 PG H M 173.450/93 PG H M 173.461/93 PG H M 173.461/94 CM A M 173.469/93 FS H M PR A M 173.469/94 173.479/93 FS H M 173.492/93 FS H M 173.502/93 FS H M 173.510/93 FM B M 173.521/93 FM B M OG B M 173.540/94 173.549/94 HM B F 173.580/94 PR A F 173~589/94 OG B F 173.640/94 MC C F GM A F 173.640/96 BG E F 173.689/94 173.789/94 MR E F 173.829/94 MM F F SM D F 173.859/94 173.870/94 SM D F 173.919/94 BC C M 173.929/94 BC C M 173.939/94 BC C M .173.959/94 BC C M 173.970/94 MC C M 173.981/94 MP D M BC C M 173.990/94 174.001/94 BG E M 174.009/94 BG E M 174.019/94 BG E M 174.039/94 MR E M 174.049/94 MM F M 174.059/94 MR E M 174.069/94 BG E M 174.090/94 MR E M 174.101/94 MM F M 174.109/94 MM F M 174.119/94 MM F M 174.129/94 MM F M 174.140/94 MM F M GB B M 174.140/95 174.150/94 MM F M 174.160/94 MM F M 174.170/94 FM B M 174~181/94 FM B M 174.200/95 MS F M 174.220/95 MS F M 174.230/95 MS F M 174.240/95 MD E M 174.319/95 MS F F 174.329/95 MD E F 174.339/95 MD E F 174.360/95 MD E F 174.401/95 AP D F 174.500/95 LB C F 174.520/95 KR B F 174.520/96 GM A F BL A F 174.560/95 174.609/95 GB B F 174.679/95 HG E M 174.689/95 AP D M 174.689/96 EW G M. 174.729/95 VM D M 174.789/95 WD C M 174.800/96 BG E M 174.809/95 W C M 174.861/95 KR B M . ,174.910/96 LM D F 175.059/96 BC C F 175.130/96 GM A F 175.170/96 LM D M Death Date l 14-Sep-95 2 30-Nov-95 2 15-0ct-95 C 07-Feb-94 Y 19-Sep-95 Age C 25-Apr-94 2 20-0ct-96 2 2 3 2 2 2 Y Y C c C c 2 10·Sep·95 03·Sep·95 13-Nov-96 28-Aug-95 17-0ct-95 13-0ct.:.96 . 07-Sep-95 21-Dec-95 01-Jun-95 30-Mar-95 16-Mar-97 31-0ct-96 20-Mar-95 10-Jan-97 2 23-0ct-96 2 C • 21-Feb-95 Y 02-Nov-95 2 2 2 2 2 2 2 2 2 2 2 Y 2 2 Y 2 C 2 c Y 2 2 16-0ct-96 18-Sep-96 02-Nov-96 20-0ct-96 02-Nov-96 20-0ct-96 30-Nov-96 12-0ct-96 13-Nov-96 05-Sep-96 14-Sep-96 17-0ct-95 26-Sep-96 20-0ct-96 . 24-May-96 19-oct-96 01-Mar-95 14-oct-96 14-Mar-95 30-Nov-96 14-0ct-96 15-Sep-96 C 25-Apr-95 2 Y C C C Y Y c Y 2 30-Nov-96 18-0ct-96 08-Apr-96 24-Apr-96 17-Apr-96 02-0ct-96 C Y C Y C Y C C v 03-Sep-96 27-Mar-96 22-Apr-97 22-Sep-96 16-Apr-96 21·Sep·9~ 25-Apr-97 14-May-97 15-Apr-96 13-Nov-96 0S·Feb-96 14-May-97 17-oct-96 C C Y Y 09-Apr-97 22-Apr-97 31-0ct-96 C C - 13-Feb:.97 c C 22-May-96 27-Feb-97 24-Mar-97 26-Mar-97 Cause of Death & Code Nllli)er Legal kill archery season=33 Disappear rifle season-22 Legal kill rifle season-28 Malnutrition-6 Wounding loss archery/111.1zzle·24 Malnutrition-6 Legal kill rifle season-28 Legal kill archery season-33 Legal kill archery season-33 Legal kill rifle season-28 Legal kill archery season~33 Legal kill rifle season-28 Legal kill rifle season-28 Legal kill archery season-33 Unknown-11 Unknown-suspect predation-30 Unknown-suspect predatfon-30 Malnutrition-6 Legal kill rifle season-28 Unknown-suspect malnutrition-31 leg~l kill late rifle season-29 Legal kill rifle season-28 Mmntain lion predation-3 Illegal kill rifle season-9 Legal kill rifle season-28 Legal kill archery season-33 Legal kill rifle season-~8 Legal kill rifle season-28 Legal kill rifle season-28 Legal kill rifle season-28 Disappear archery/111.1zzle season-21 Legal kill rifle season-28. Illegal kill rifle season-9 Legal kill archery season-33 Legal kill 111.1zzleloading season-34 Illegal kill rifle season-9 Wounding loss archery/111.1zzle-24 Legal kill rifle season-28 Unknown-suspect predation-30 Legal kill rifle season-28 Bear predation-35 Legal kill rifle season-28 Mountain lion predation-3 Illegal kill rifle season-9 Legal kill rifle season-28 Legal Id l.l n.izzleloading season-34 Mountain lion predation-3 Disappear rifle season-22 Illegal kill rifle season-9 . Unknown-suspect malnutrition-31 Mountain lion predation-3 Mountain lion predation-3 Wounding loss archery/n.izzle-24 Legal kill archery season-33 Unknown predator-5 Unknown-suspect predation-30 Legal kill 111.1zzleloading season-34 · Unknown-suspect predation-30 Legal kill n.izzleloading season-34 Mountain lion predation-3 Illegal kill-7 Unknown-suspect predation-30 Illegal kill rifle season-9 Mountain lion predation-3 Unknown-suspect malnutrition-31 Illegal kill rifle season-9 Unknown-suspect predation-30 Mountain lion predation-3 Accident, fell-10 Wounding loss rifle season-25 Unknown-suspect predatfon-30 Unknown· 11 • Mountain· lion predation-3 Unknown-suspect malnutrition-3'1 �177 Colorado Division Wildlife Research July 1998 of Wildlife Report JOB PROGRESS state Work Colorado of Project No. Package W-153-R-11 Mammals 3002 Elk Inyestigations Task No. Period Author: REPORT Research Estimating Developing PQpulation Covered:i July Survival Rates of Elk and Techniques to Estimate Size 1, 1997 - June 30, 1998 D. J. Freddy Personnel: R. Adams, T. Beck, J. Broderick, G. Byrne, D. Crane, R. Del Piccolo, J. Ellenberger, J. Frothingham, V. Graham, D. Homan, R. Kahn, K. Madariaga, D. Masden, C. Mehaffy, P. Neil, J. olterman, J. Thompson, P. Will, K. Wright, S. Yamashita, D. Younkin, CDOW; W. Andelt, D. Bowden, G. White, CSU; Diagnostic Laboratory, CSU; D. Ouren, USGS-BRO, BLM Glenwood springs, CO, USFS Rifle, CO, Rocky Mountain Elk Foundation, cooperating. Abstract During 1997-98 we monitored survival of 210 adult elk radio-collared in previous years and we conducted an applied sighting bias trial to further test methods of estimating elk density. We summarized average survival rates (± 95% CI) for elk between 1993-94 and 1997-98. Survival during summer-fall for females age ~12 months averaged 0.90 ± 0.03 inclusive of hunting deaths (n=524 elk-years) and 0.99 ± 0.01 with hunting deaths censored (n=476 elk-years). Survival during winter-spring for females age ~18 months averaged 0.97 ± 0.01 inclusive of hunting deaths (n=547 elk-years) and 0.99 ± 0.01 with hunting deaths censored (n=536 elk-years). Survival during summer-fall for males age ~24 months was 0.20 ± 0.09 inclusive of hunting deaths (n=82 elk-years) and 1.00 with hunting deaths censored (n=16 elk-years). 'Survival during winterspring for males age ~30 months was 1.00 (n=13 elk-years). Hunting related deaths were the primary cause of mortality for adult elk and annually removed 80% of the adult males, 11% of the yearling males, and 9% of the adult females. For adult females, deaths due to wounding and illegal kills exceeded the absolute number of females dying from natural 'causes during 5 years. High adult and calf natural survival rates allow this population to grow an estimated 7% annually with current levels of hunting mortality. Further tests of sighting bias revealed that failure to count all, elk in groups, as opposed to failure to detect groups, is likely the primary cause of negatively biased estimates of elk density based on quadrat sampling. Counting error may approach 25%. Therefore, sighting bias models developed to account for errors in detecting elk on quadrats do not adequately increase estimates of elk numbers. We did not find evidence indicating mark-resight estimates of elk numberS were positively biased. �179 ESTIMATING SURVIVAL JOB PROGRESS REPORT RATES OF ELK AND DEVELOPING ESTIMATE POPULATION SIZE David TECHNIQUES TO J. Freddy P. H. OBJECTIVE Estimate survival rates of adult ,female, adult male, techniques to estimate population size. SEGMENT and calf elk and develop OBJECTIVES 1. Estimate winter, summer, and annual survival rates of adult female male elk from known fates of previously radio-collared elk. 2. Estimate density of elk in a portion of GMU-42 using a helicopter and employing mark-resight density estimators and random quadrat density estimators to evaluate potential errors in meeting assumptions of markresight theory and sighting bias correction theory. 3. Analyze data and summarize annually in Federal Aid Job Progress and Report. INTRODUCTION Our objectives are to provide reliable estimates of survival rates for calves during winter and for adult females and males throughout the year and to develop and test a system to estimate elk densities on winter ranges. Estimates of calf survival were obtained during winters 1993-94 through 199697 and summarized in Freddy (1997). We are continuing to obtain estimates of survival for adult males and females. For adults, we are interested in survival rates inclusive of hunting mortalities to document human-induced rates of survival and exclusive of hunting mortalities to estimate natural rates of survival. We are evaluating a system for estimating population size or density that incorporates estimates of sighting bias in conjunction with random sampling of search quadrats as sample units. OUr winter study area encompasses about 839 km2 (324 mi2) in the eastern half of Game Management Unit 42 south and east of Rifle, Colorado. This area is part of the Grand Mesa Elk Data Analysis Unit E-14. Elk winter habitats include juniper-pinyon woodland (Juniperus osteosperma-Pinus edulis), oakbrush-mountain shrub (Quercus gambelii-Amelanchier alnifolia), aspen (Populus tremuloides), sagebrush (Artemisia tridentata), and agricultural fields below elevations of 9,800 ft. In summer, elk use oakbrush-mountain shrub, aspen, and subalpine fir-Engelmann spruce (Abies lasiocarpa-Picea engelmannii) meadow systems throughout the Grand Mesa region up to elevations of 12,000 ft. (Freddy 1993). METHODS We placed radio collars (172-176MHz) having mortality sensors on elk calve.s age 6 months and adults age ~18 months during December 1993-1996 (Freddy 1997). Elk were captured using helicopter net-gunning and portable corral traps (Freddy 1997). During 1997-98, we monitored 210 adult elk having functioning radios as of July 1997. Collars were white and had black identification symbols on dorsal surfaces of collars to allow observers in helicopters to individually identify elk. �180 Estimating Elk Survival Rates We monitored life or death status of radioed elk with aerial surveys using a Cessna 185 at 2-4 week intervals from December 1993 to June 1998 and with daily ground surveys from December-April each year from 1993-94 to 1996-97. Survival rates (5) of radioed elk were calculated using the binomial estimator with a variance, VAR(S) = S(l-S)/n collars (White and Garrott 1990). Survival rates are expressed as the mean estimate ± the 95% confidence interval. We used x2-contingency tests to compare survival rates between sexes and among time intervals (PROC FREQ, SAS 1988). Sample sizes refer to numbers of individual radioed elk except when referenced as elk-years which denotes that individual radioed elk, because they survived for several years, were used repeatedly in successive yearly time intervals to calculate pooled average survival rates among years. Elk-years represents the numbers of times elk were at risk during time intervals. We defined 4 major time intervals for survival analyses: winter-spring was 1 December to 14 June, summer-fall was 15 June to 30 November, annual was 1 December to 30 November to coincide with timing of capture and collaring, and yearly was 15 June to subsequent 14 June used only for yearling elk aged 12-23 months. For adult elk during these time intervals, we calculated survival rates inclusive of natural and hunting-related mortalities, exclusive of hunting mortalities, and exclusive of natural mortalities. Excluding, or censoring hunting mortalities, provided estimates of natural survival rates. Excluding natural mortalities but including hunting mortalities provided estimates of hunting removal rates. Censoring elk associated with categories of mortalities reduced sample sizes used to estimate survival rates. Elk were also censored if radios failed or if their life/death status was unknown after an extended period of time. Archery, muzzleloading, and rifle hunting seasons occurred from about 1 September to 15 November each year during the summer-fall time interval. Laterifle seasons restricted hunters to taking only antlerless elk and occurred yearly from about 25 November-3l January which included portions of the summer-fall and winter-spring time intervals. Yearling spike-antlered males were generally not legal quarry in most areas frequented by radioed elk. Male elk become legal quarry when branch-antlered usually at age 2 years. Cause of death of radioed elk recovered in the field was estimated from presence or absence of gunshot wounds or bite wounds on carcass, predator tracks or scat at carcass site, physical positioning of carcass remains whether buried, covered, scattered, or consolidated (Wade and Browns 1982, Halfpenny and Biesiot 1986), relative amount of internal fat and marrow fat if present with carcass, and results of histopathology and marrow fat analyses. Fat content (percent dry matter) of bone marrow and estimates of age based on dental cementum were obtained for dead elk by the Colorado Division of Wildlife Laboratory while histopathology analyses were provided by the Colorado State University Veterinary Diagnostic Laboratory. Photographs were taken of nearly all mortalities so that physical evidence could be reviewed and judged by outside experts (pers. comm. T. Beck, W. Ahdelt). Estimating Elk Densities During 3 applied surveys in winters 199~ and 1997, we estimated numbers of elk on 132-137 mi2 of winter range with helicopter counts of marked (radioed) and unmarked elk on random quadrats and along nonrandom flight paths. Estimated numbers of elk for the 3 surveys based on actual counts of elk on quadrats were 51, 63, and 93% of the estimated numbers of elk obtained from markresight formulas using counts of marked and unmarked elk. Adjusting actual �181 quadrat counts with previously developed sighting bias models did not meaningfully increase quadrat estimates in comparison to mark-resight estimates (Freddy 1996, 1997). We had 3 important concerns: 1) sighting bias corrections developed during previous testing trials were biased high and provided inadequate corrections for applied surveys, 2) observers failed to see marks on elk in groups that were detected and counted accurately causing inflated mark-resight estimates of elk numbers, and 3) observers failed to count all elk within groups that were detected on quadrats with elk not counted inclusive of both marked and unmarked elk causing deflated counts of elk on quadrats. This last concern represented counting error that would not be accounted for by sighting bias corrections. In 1998, we conducted an applied sighting bias trial on quadrats along with mark-resight surveys to again compare estimates of elk numbers based on quadrat counts and mark-resight surveys. We selected 44 mi2 in West Divide and Alkali creeks that were a portion of the winter range sampled with quadrats and mark-resight surveys in 1996 and 1997 and a portion of the area used to conduct sighting bias trials in 1994 and 1995. This area was dominated by oakbrush and pinyon-juniper habitats. For this test, all marked elk were age ~18 months unlike previous sighting tests and winter range surveys that also included marked calves age 6 months. Locations of all radioed elk within this 44 mi2 area were obtained by telemetry using a Cessna 185 and ARNAV Star5000 GPS system on 26 February, 1998. Radioed elk represented a population of known size which we used to evaluate the magnitude of sighting detection bias. During 3 days from 27 February to 1 March, elk were counted on all 44, 1-mi2 quadrats using a Bell-Soloy helicopter flown at 40-50 mph. A flight crew was a pilot, navigator, and observer and crews changed each day among 3 observers, 2 navigators, and 1 pilot all of whom were involved in previous sighting bias tests and winter range surveys. Members of flight crews had no knowledge of the number or locations of marked elk within the survey area. Flight crews flew nearly 7 hours each day to complete quadrats which was similar to daily flying time and fatigue commonly associated with applied surveys. Quadrats were flown using the same helicopter and counting procedures that were used in previous sighting tests and surveys (Freddy 1994-1997). Observers recorded the standard sighting bias variables of group size, elk behavior, habitat type, percent screening cover, and percent snow cover for all groups seen. As flight crews completed quadrats, they maintained a list of marked elk seen and their GPS locations (Garmin Pilot III) and whether or not marked elk were or could have been individually identified. Those marked elk not seen or those seen but not individually identified during counts were then found by a second independent flight crew using a Bell-Soloy helicopter and telemetry. These missed elk were found on the same day an average 3.6 hours after quadrats were flown. For every missed' or unidentified marked elk group, the second flight crew recorded the GPS location (Garmin Pilot III) of the group and standard sighting bias variables. Those elk seen but not individually identified by quadrat crews were assigned as a specific individual elk based upon proximity of locations of where such elk were seen by quadrat crews compared to locations of these elk when found by the second flight crew. Locations and status of all marked elk were therefore known during the time quadrats were flown. A sighting trial occurred when marked elk were detected during quadrat counts and when marked elk were missed and later found by the second flight crew. To �182 maintain independent observations of marked elk, only 1 sighting trial occurred if there was more than 1 marked elk in a group. Otherwise, detecting or missing a marked elk constituted a sighting trial. Two mark-resight flights using the same helicopters were conducted on 1 and 2 March after quadrat counts were completed. These flights followed a nonrandom path through the same 44-mi2 with the intent of counting as many elk as possible within allotted flight time. The navigator for both flights had not participated in quadrat counts. The observer for the first flight was an observer during part of the quadrat counts while the observer for the second flight had not participated in quadrat counts. These 2 observers were the same persons who conducted mark-resight flights during winter range surveys in 1996 and 1997. At the conclusion of the second mark-resight flight, locations of collared elk not seen were confirmed to be within the sample area using the helicopter and telemetry. Weather conditions during flights were variable but similar to conditions experienced during previous winter range surveys. Quadrats were flown with generally acceptable visibility in hazy to bright sunshine and light to moderate wind speeds although brief snow squalls occurred on 27 and 28 February. Nonrandom flights were flown during hazy to bright sunshine and low wind speeds. Snow cover was nearly 100% in all areas with depths estimated to be 0.5 to 2.5 ft. Fresh snow was received on 24 and 25 February. Minimum night temperatures were 2-13 OF and were the coldest measured during the 14 days previous to flights. Maximum day temperatures were 27-35. OF. Number of elk on the search area was estimated using 3 approaches: from actual counts of elk on quadrats, from adjusting quadrat counts with sighting bias correction models, and from using the Bowden (BE) and lmmigration-EmigrationJHE (lMJHE) mark-resight estimators in program NOREMARK (White 1996). Elk counted on quadrats were potentially a total count because 100% of the search area was flown and because all 44 sample units were flown, there was no sampling variance associated with estimates based on quadrats. To adjust for elk groups likely missed on quadrats, we applied sighting bias corrections to elk groups counted on each quadrat which adjusted counts per quadrat and provided a sighting bias corrected total population estimate. We used 3 sighting correction models based on all sighting trials (n = 224) to .correct for probabilities of detecting elk groups, where probability of detection is [n = eU / 1 + eU] with u representing the linear regression of variables affecting the probability of detecting a group: 1. u = 1.342(LogGroupSize(countedd - 0.131 2. u = 1.230(LogGroupSize(co=ted» - 2.127(Bedded) + 0.291 3. u = 1.135(LogGroupSize(co=ted» - 1.822(Bedded) - 0.031(%Cover) + 1.517 probability of detecting an elk group was adjusted for group size, for bedded or not bedded behavior when detected, and for percent vegetation screening cover (Samuel et al. 1987, Freddy'1995). Relative AlC model performance values of 172, 160, and 154 for models 1-3, respectively, indicated model 3 was the most parsimonious model. We pooled counts of marked and unmarked elk seen during quadrat counts and mark-resight flights. Estimated numbers of elk based on all 3 flights pooled was considered to be the best estimate of population size and the benchmark against which quadrat counts were compared. Both the BE and lEJHE estimators account for movement of elk on and off the search area but the BE better accounts for heterogenous sighting probabilities among individual elk (Bowden 1995, White 1996). �183 Moyements We continued to precisely locate selected radioed elk at least once ~r month since their capture to document seasonal movements using a Cessna 185. These elk were originally selected at random from within trap zones and equalized by age class in January 1994. Elk from the original sample that died were replaced each subsequent January with randomly selected elk that were usually of the same sex, age class, and trap zone as elk that died. All replacement elk in January 1998 were yearling or older because additional calves were not marked in December 1997. As of 1 January 1998, these 44 elk were classified as 27 adult females, 5 yearling females, 4 adult males, and 8 yearling males. During June 1998, we again located an additional 28 adult females that were originally selected at random in 1995 to document their locations during the calving period. Females from the original 1995 sample that died were replaced each subsequent June with adult females randomly selected from the same trap zones of elk that died. As needed, we located other elk to document unusual movements. RESULTS AND DISCUSSIQH Survival Estimates Between 1 December 1993 and 14 June 1998, 181 radioed elk died of which 31 were calves 6-11 months old and 150 were adults ~12 months old (Table 1, Appendix I). Calves died of natural causes (Freddy 1997) while for adults ~12 months old, hunting accounted for 92% of 150 deaths. Yearlings Survival during summer-fall for yearling elk age 12-17 months was 0.89 ± 0.06 for males (n = 120) and 0.95 ± 0.04 for females (n = 128) when averaged among 4 years and inclusive of hunting deaths, and 1.00 for both males (n = 107) and females (n = 122) with hunting deaths censored. All deaths for males and females during summer-fall were due to hunting. Survival among years for males ranged from 0.83 to 0.97 and for females from 0.87 to 1.00 inclusive of hunting deaths. We failed to detect differences in survival between sexes within years (P ~ 0.13) and among years for both males (P ~ 0.39) and females (P ~ 0.09) inclusive of hunting deaths. Survival of females (0.95) tended to be higher than males (0.89) when averaged among years and inclusive of hunting deaths (P = 0.07) Cl'ables 2-5). Survival during winter-spring for yearling elk age 18-23 months was 0.98 ± 0.02 for males (n = 106) and 0.97 ± 0.0.03 for females (n = 121) when averaged among 4 years and inclusive of hunting deaths, and 0.98 ± 0.02 for males (n = 106) and 0.98 ± 0.02 for females (n = 120) with hunting deaths censored. Survival among years for males ranged from 0.96 to 1.00, inclusive or exclusive of hunting deaths. Survival for females ranged from 0.93 to 1.00 and 0.96 to 1.00 inclusive and exclusive of hunting deaths, respectively. We failed to detect differences in survival among ye~rs for both males (P ~ 0.51) and females (P ~ 0.22) inclusive or exclusive of hunting deaths. We also failed to detect differences in survival between sexes within years (P ~ 0.62) and among years (P ~ 0.76) inclusive or exclusive of hunting deaths (Tables 25). During winter-spring, there were 2 nonhunting deaths each for males and females and 1 illegal hunting death for a female (Table 1). Yearly survival for elk age 12-23 months was 0.87 ± 0.06 for males (n = 119) and 0.93 ± 0.05 for females (n = 127) when averaged among 4 years with hunting deaths included. Survival among years for males ranged from 0.79 to 0.97 and for females from 0.81 to 1.00. We failed to detect differences in yearly survival among years for males (P = 0.27) but survival was lower (0.81) for �184 females in 1996-97 (P = 0.02). Yearling males were subjected to about the same rate of illegal hunting during all years which accounted for most of the male mortality, while in 1996-97, females had a lower survival rate because 5 of 6 deaths were hunting related (Tables 2-5). With hunting deaths included, we failed to detect differences in survival between sexes within (P ~ 0.13) and among years (P = 0.15). Of 24 deaths involving yearling elk, 20 (83%) were hunting related (Table 1). With hunting deaths censored, yearly survival for elk age 12-23 months was 0.98 ± 0.02 for males (n = 106) and 0.98 ± 0.02 for females (n = 120) when averaged among 4 years. Survival among years for males and females ranged from 0.96 to 1.00. We failed to detect differences in yearly survival among years for males (P = 0.51) and for females (P = 0.50), and between sexes within and among years (P ~ 0.90) (Tables 2-5). Yearling spike-antlered males were generally not legal quarry during hunting seasons. Each year from 1994-1997, 29-32 yearling males radioed as calves entered the fall hunting seasons presumably as spike-antlered males. OVer 4 years, illegal harvest removed 10% of the yearling males with yearly rates ranging from 3 to 17%. Of 12 elk illegally killed, 11 were taken during rifle seasons and 1 during archery seasons. In addition to these 12 elk, 1 male was fatally wounded during archery season in an area where the elk was legal quarry. Overall, 11% of the yearling males were annually removed by hunting (Tables 1, 2-5). Adult Females Survival during summer-fall for all adult females age ~12 months was 0.90 ± 0.03 (n = 524 elk-years) when averaged among 4 years and inclusive of hunting deaths and 0.99 ± 0.01 (n = 476 elk-years) with hunting deaths censored. Survival among years ranged from 0.87 to 0.94 and from 0.99 to 1.00, inclusive and exclusive of hunting mortalities, respectively. We failed to detect differences in survival among years inclusive (P = 0.30) or exclusive (P = 0.46) of hunting mortalities (Table 6). Hunting was involved in 48 (96%) of 50 summer-fall deaths: 32 were killed, 12 were wounded, and 4 disappeared during hunting seasons and presumed legally killed. One natural death was attributed each to suspected predation and complications while calving. At the beginning of fall hunting seasons, there were 99, 129, 142, and 152 radioed adult females age ~12 months available to hunters in 1994, 1995, 1996 and 1997, respectively, representing 522 elk-years (Table 6, nonhunting deaths censored). During summer-fall, 9% of these radioeq elk were removed annually by all types of hunting mortality (range = 6-12%). Survival during winter-spring for all adult females age ~18 months was 0.97 ± 0.01 (n = 547 elk-years) when averaged among 5 years and inclusive of hunting deaths and 0.99 ± 0.01 (n = 536 elk-years) with hunting deaths censored. Survival among years ranged from 0.94 to 0.99 and from 0.97 to 0.99, inclusive and exclusive of hunting mortalities, respectively. We failed to detect differences in survival among years inclusive (P = 0.34) or exclusive (P = 0.40) of hunting mortalities (Table 6). There were 19 winter-spring deaths, of which 11 (58%) involved hunting with 5 killed and 3 wounded during rifle late-seasons and 3 illegally killed out of hunting seasons. Natural deaths were attributed to mountain lion predation (1), suspected predation (2), complications while calving (1), accidental fall (1), and unknown cause (3) (Table 1). During winter-spring, 2% of these radioed elk were removed annually by all types of hunting mortality. �185 Estimates of adult female survival using all females age ~12 months could be biased because of the constant yearly recruitment of yearling females into this radioed population of adults afforded by many radioed calves surviving to yearling age and no similar yearly recruitment of additional older adult females into the radioed population. We therefore estimated survival rates for adult females age ~24 months by excluding recruitment of yearlings age 1223 months and for 68 adult females initially collared as a cohort in December 1993 whose age at capture ranged from 18 months (n=6) to 10+ years. Survival during summer-fall for adult females age ~24 months was 0.89 ± 0.03 (n = 396 elk-years) when averaged among 4 years and inclusive of hunting deaths and 0.99 ± 0.01 (n = 354 elk-years) with hunting deaths censored. Survival among years ranged from 0.82 to 0.93 and from 0.99 to 1.00, inclusive and exclusive of hunting mortalities, respectively. We failed to detect differences in survival among years inclusive (P = 0.17) or exclusive (P = 0.38) of hunting mortalities (Table 7). Hunting was involved in 42 (95%) of 44 summer-fall deaths. During summer-fall, 11% of these radioed elk were removed annually by all types of hunting mortality (range 7-17%). Survival during winter-spring for adult females age ~30 months was 0.96 ± 0.02 (n = 420 elk-years) when averaged among 5 years and inclusive of hunting deaths and 0.98 ± 0.01 (n = 410 elk-years) with hunting deaths censored. Survival among years ranged from 0.93 to 0'.99 and from 0.97 to 0.99, inclusive and exclusive of hunting mortalities, respectively. We failed to detect differences in survival among years inclusive (P = 0.15) or exclusive (P = 0.39) of hunting deaths (Table 7). There were 16 winter-spring deaths with 10 (63%) attributed to hunting. During winter-spring, 2% of these radioed elk were removed annually by all types of hunting mortality (range 1-5%). Using the 1993 cohort of adult females, survival during summer-fall was 0.88 ± 0.05 (n = 189 elk-years) when averaged among 4 years and inclusive of hunting deaths and 0.99 ± 0.01 (n = 167 elk-years) with hunting deaths censored. Survival among years ranged from 0.82 to 0.94 and from 0.98 to 1.00, inclusive and exclusive of hunting deaths, respectively. We failed to detect differences in survival among years inclusive (P = 0.20) or exclusive (P = 0.55) of hunting mortalities (Table 8). Hunting was involved in 22 (96%) of 23 summer-fall deaths. Percent of these radioed elk removed during summerfall by all types of hunting-related mortalities averaged 12 annually (range = 6-17%). Interestingly, as this cohort collectively increased in age from 1994 to 1997, the annual percentage of elk removed by hunting during summer-fall steadily declined from 17 to 6%. For the 1993 cohort, survival during winter-spring age was 0.95 ± 0.03 (n = 233 elk-years) when averaged among 5.years and inclusive of hunting deaths and 0.98 ± 0.02 (n = 227 elk-years) with hunting deaths censored. Survival among years ranged from 0.92 to 1.00 and from 0.93 to 1.00, inclusive and exclusive of hunting mortalities, respectively. We failed detect differences in survival among years inclusive (P = 0.47) or exclusive (P = 0.17) of hunting deaths (Table 8). There were 11 winter-spring deaths, of which 6 (55%) involved hunting. During winter-spring, 3% of these radioed elk were removed annually by all types of hunting mortality. to Overall, survival rates for adult females during summer-fall and winter-spring were the same among age groups of adult females. Natural survival rates exceeded 0.98 for all time intervals and groupings (Table 9). �186 Annual survival rate using the 1993 cohort was 0.83 ± 0.07 (n = 199 elk-years) when averaged among 4 years and inclusive of hunting deaths and 0.97 ± 0.03 (n = 170 elk-years) with hunting deaths censored. Survival among years ranged from 0.78 to 0.94 and from 0.92 to 1.00, inclusive and exclusive of hunting mortalities, respectively. We failed to detect differences in survival among years inclusive (P = 0.17) or exclusive (P = 0.34) of hunting mortalities (Table 10). Annually, 14% of these radioed elk were removed by all types of hunting mortality (range 8-20%). Removal rates of females by hunting during summer-fall were compared between yearling females age 12-17 months and adult females age ~24 months. Removal rates were higher for adult females (0.11) than yearling females (0.05) when averaged among 4 years (P = 0.04). This difference was caused by extremes in removal rates in 1994 when 17% of adults and 3% of yearlings were removed (P =0.05) and in 1997 when 11% of adults and 0% yearlings were removed (P = 0.06). In 1995 and 1996, removal rates were not different (P > 0.41) but rates for yearlings (3%) tended to be lower than for adults (7-9%). These rates suggest that vulnerability of yearling and adult females to hunting is different. Adult Males Survival during summer-fall for adult males age 24-29 months was 0 •. 17 ± 0.09 (n = 75) when averaged among 3 years and inclusive of hunting deaths and 1.00 (n = 13) with hunting deaths censored. Survival among years ranged from 0.08 to 0.23 inclusive of hunting mortalities. We failed to detect differences in survival among years (P > 0.32) (Tables 2-4). All deaths were attributed to hunting resulting in removal rates averaging 83% collectively for all fall hunting seasons each year. Through fall 1997, 7 males had survived at least 1 year of hunting seasons and were available during at least 1 additional year of hunting seasons as 3 or 4 year-olds. Survival during summer-fall for adult males age ~24 months was 0.20 ± 0.09 (n = 82 elk-years) when averaged among 3 years and inclusive of hunting deaths and 1.00 (n = 16 elk-years) with hunting deaths censored. Survival among years ranged from 0.14 to 0.24 inclusive of hunting mortalities. We failed to detect differences in survival among years (P > 0.61) (Tables 2-4). All deaths were attributed to hunting resulting in removal rates averaging 80% collectively for all legal males during fall hunting seasons each year. Survival during winter-spring for adult males age ~30 months was 1.00 (n = 13 elk-years) during all years (Tables 2-5). There were no hunting or nonhunting deaths of adult males during winter-spring. Hunting was the cause of mortality among male elk age ~24 months. From the 1993-94 cohort of 36 male calves, 28 lived to age 24-months with 22 (79%) harvested in 1995 inclusive of 2 that disappeared 'and 2 wounding losses during rifle seasons. From the 1994-95 cohort of 33 male calves, 25 lived to 24months and 23 (92%) were harvested in 1996 including 2 that disappeared during seasons, 1 illegally taken during rifle seasons, and 1 wounding loss during archery/muzzleloading season. From the 1995-96 cohort of 33 male calves, 22 lived to 24-months and 17 (77%) were harvested in 1997 including 1 that disappeared, 1 illegally taken, and 4 wounding losses during rifle seasons. Comparative Survival of Adult Males and Females The differential impact of hunting on survival and recruitment of males and females to young adult age classes is demonstrated by net survival rates of �187 calf cohorts. For 1993-94, 1994-95, and 1995-96 calf cohorts, net survival from age 6 to 35 months averaged 0.10 for males and 0.73 for females (Tables 2-4). Only 3 (4%) of a potential 69 males survived from age 6 months to 48 months (Tables 2,3). Hunting Mortality Per Season Distribution of all types of hunting mortalities for male elk age >12 months during hunting seasons 1994-1997 was: Archery 15%, Muzzleloading 5%, RifleFirst 42%, Rifle-Second 25%, Rifle-Third 13%, Rifle-Late 0%, Illegal out-ofSeason 0%. Distribution of all hunting mortalities for female elk age >12 months was: Archery 14%, Muzzleloading 7%, Rifle-First 19%, Rifle-Second 31%, Rifle-Third 10%, Rifle-Late 15%, Illegal out-of-Season 4%. All female and nearly all male elk killed by hunters died within the Grand Mesa DAU E-14 or adjacent GMU 43. Exceptions were 6 adult males that dispersed long distances and were killed near Ruedi Reservoir, Black Mesa, Tomichi Dome, Gothic, and Anthracite Creek in GMUs 47, 63, 54, 55, 551, and 521 up to 100 mi south or east of their capture sites. These 6 males represented 9% of adult males age ~24 months that were taken by hunters. Adjusting Hunting Harvest Estimates Surveys used to estimate legal hunter harvest do not account for wounding losses or illegal kills. Estimates of the proportion of the legal kill that these additional deaths represent were made based on fates of radioed elk during hunting seasons 1994-97. We defined legal kill to be the number of radioed elk known to be harvested plus those whose radio signals disappeared during hunting seasons and were subsequently assumed to be legally killed. This legal kill would most likely represent the kill estimated by harvest surveys. Wounding losses and illegal kills represented deaths not accounted for by surveys. Total kill which was known for radioed elk could thus be expressed as (legal kill)*(X)= total kill where (X) represents a correction factor. correction factors by sex/age class were 1.16 for adult males, 1.11 for yearling males, and ~ 1.4 for all categories of adult females (Table 11). Rifle and late-rifle seasons involving adult female elk had high correction factors of 1.35-2.00. Losses of adult females not reported or estimated can be important in modeling populations. Total unreported losses of adult females during hunting seasons 1994-1997 was 18 which was 1.8x the 10 natural deaths of adult females in the comparative time interval of 1 December 1993-14 June 1998 (Tables 6, 11). An argument could be made that estimating unreported losses is as important as estimating natural mortality rates. Survival Rates and Population Modeling Survival and removal rates were incorporated into a simple spreadsheet model to assess trends in population growth (Table 12). Natural survival rates (hunting deaths censored) were used for summer-farl and winter-spring time intervals for adult males, yearling males, adult females, yearling females, and calves. Hunting deaths of all types were incorporated as total harvest removal rates for each sex/age class. Using total removal rates is a simpler method than using correction factors to account for unreported .harvest. This model assumed survival of calves 0-5 months of age during summer was 1.00 because the model uses post-season calf:cow ratios measured in January as an estimate of net calf recruitment to December. Recruited sex ratio of calves to December was assumed to be 1:1. �188 Modeling suggests that if harvest rates of antlered elk remained as measured and harvest rates of antler less elk were reduced to 0, the population could grow at an annual rate of 18%. Current rates of antlerless harvest allow annual growth rates of about 7.5%. To stabilize population growth, harvest rates on adult and yearling female and calf age classes must increase nearly 2X. population Estimates Movements of marked elk on and off the survey area during days when flights were conducted resulted in 34 marked elk (27 female, 7.male) and 38 marked elk (31 female, 7 male) within the area during quadrat and mark-resight flights, respectively. Elk movements were usually <600 yards and near sample area boundaries. Ratios of marked/unmarked elk counted were similar among quadrat and nonrandom flights ranging from 0.029 to 0.036 (Table 13). These ratios were generally lower than ratios of 0.031-0.042 observed during extensive survey flights in 1996 and 1997 when marked calves were available in addition to marked adults. Observers used 19.7±2.1 (95% CI) minutes to count 17.5 elk per quadrat (range=0-68). Observers 1B and 2E encountered similar densities of elk and had similar search times in oakbrush habitat while observer 3M had lower elk densities and higher search times in pinyon-juniper habitat (Table 14). During 3 extensive surveys in 1996 and 1997, these same observers used 17.6, 16.9, and 19.8 minutes to count 17.0, 25.9, and 17.3 elk, respectively, per quadrat (n=177). Searching effort during the 1998 sighting trial was therefore representative of searching effort during previous extensive surveys. During nonrandom flights, detection rates of marked elk were 0.32 and 0.45 with effective searching rates of 4.9 and 6.00 minutes per mi2 of search area during flights 1 and 2, respectively (Table 13). During nonrandom flights for extensive surveys in 1996 and 1997, detection rates of marked elk were 0.260.58 with effective searching rates of 3.3-3.9 minutes per mi2 of search area. Nonrandom flights in 1998 were therefore representative of flights during extensive surveys. Observers completed 32 sighting trials while counting elk on quadrats. Average sighting bias or detection rate for marked elk groups among observers was 0.78±0.15 (95% CI) (Table 14). This sighting bias rate was similar to the average sighting bias rate of 0.82+ 0.06 (95% CI) for these same 3 observers during sighting bias test trials (n=192) conducted in 1994 and 1995. Therefore, sighting bias rate estimated from test trials appears to be representative of sighting bias during applied surveys. Observers failed to detect 7 marked elk during sighting trials. When found after quadrats were completed, 6 of these elk were in groups of ~2 elk, 4 were bedded and appeared to have been bedded for several hours, 5 were in screening cover densities ~60%, and 5 were reluctant to move even when disturbed at close distances with the helicopter (Table 15). Importantly, the white collars were easily visible and thus, failure to detect marked elk was not likely due to failure to see marks. Of these 7 missed elk, 3 were males and 4 were females resulting in detection rates of 0.57 for adult males and 0.85 for adult females. Of the 7 missed elk, 3 were found in close proximity to other groups that were likely counted by quadrat crews. This lends credence to the possibility that some marked elk were not counted within detected groups because not all elk, �189 both marked and unmarked, were seen when groups were detected. This type of counting error is plausible as groups detected in dense pinyon-juniper or oakbrush often begin to move and observers concentrate on counting the moving elk and fail to see elk in the group that remain bedded or move in a different direction. Estimated number of elk on the search area was 1,179 based on the BE markresight estimator (Table 17). The 771 elk counted on quadrats represented 65% of the BE while sighting bias adjusted quadrat counts represented 73-75% of the BE. We propose that the 25% difference between sighting adjusted estimates and the BE represents counting error. This degree of counting error would equate to adding 2.2 elk per group to an average group size of 5.2 elk for the 148 groups detected (Table 18). Counting error, as opposed to detection error, appears to be the primary cause of negatively biased estimates of elk density based on quadrat sampling. Elk Moyements Yearling elk originally trapped as calves and surv1v1ng to age 18 months dispersed to winter ranges not affiliated with their natal capture winter range at rates of 43, 42, and 33% in winters 1995-96, 1996-97, 1997-98, respectively. Each year, 40-52% of the yearling females dispersed while only 25-33% of the yearling males dispersed. Yearling elk moved primarily west or south to winter ranges near the towns of Rulison, Parachute, Collbran, Hotchkiss, Paonia, and Paonia Reservoir. Specifically for the 1996 cohort of radioed calves, 28 males and 30 females survived to age 18 months in December 1997 (Table 5). By end of winter 199798, 19 (7M, 12F) or 33% had dispersed to other winter ranges (Table 19). Calves trapped in Alkali, Dry Hollow, and the Mammm creeks (zones E, F, G) comprised 79% of the dispersing yearlings and these elk moved primarily west to winter ranges near Collbran, Rulison, and Parachute. Calves trapped in West Divide Creek (zone D) comprised 16% of the dispersing yearlings and these elk moved to winter ranges near Collbran and also south towards Paonia Reservoir. We continue to cooperate with the USGS-BRO, USFS, and BLM with the support of the Rocky Mountain Elk Foundation to develop a GIS database allowing analyses relating elk locations, movements, and home ranges to habitat components. CONCLUSIQNS Hunting related mortalities were the primary cause of death for adult female and male elk. For adult males, all deaths to age 48 months were related to hunting. For adult females, deaths due to wounding and illegal kills exceeded the absolute number of females dying from natural causes during 5 years. Natural survival rates for adult females and males were ~0.98 during winterspring and summer-fall. High adult and calf survival rates allow this population to potentially increase 18% annually in the absence of hunting antlerless elk. Current hunting removal rates of 9% for adult females age ~12 months allows the modeled population to grow 7% annually. Failure to count all elk in groups, as opposed to failure to detect groups, appears to be the primary cause of negatively biased estimates of elk density based on quadrat sampling. Counting error may approach 25%. Sighting bias models developed to account for errors in detecting elk on quadrats inadequately increased estimates of elk numbers when compared to numbers estimated by mark-resight formulas. We did not find evidence indicating markresight estimates of elk numbers were positively biased. �190 LITERATURE CITED Bowden, D. C., and R. C. Kufeld. 1995. Generalized mark-sight population size estimation applied to Colorado moose. J. Wildl. Manage. 59: 840851. Freddy, D. J. 1993. Estimating survival rates of elk and developing techniques to estimate population size. Colo. Div. Wild1. Game Res. Rep. July: 83-117. Freddy, D. J. 1994. Estimating survival rates of elk and developing techniques to estimate population size. colo. Div. Wildl. Game Res. Rep. July: 27-42. Freddy, D. J. 1995. Estimating survival rates of elk and developing techniques to estimate population size. colo. Div. Wildl. Game Res. Rep. July: 63-79. Freddy, D. J. 1996. Estimating survival rates of elk and developing techniques to estimate population size. colo. Div. Wildl. Game Res. Rep. July: 87-108. Freddy, D. J. 1997. Estimating survival rates of elk and developing techniques to estimate population size. Colo. Div. Wildl. Game Res. Rep. July: 47-73. Hal.fpenny, J. C., and E. A. Biesiot. 1986. A field guide to mammal in North America. Johnson Books, Boulder, CO. 161pp. tracking Samuel, M. D., E. O. Garton, M. w. Schlegel, and R. G. Carson. 1987. Visibility bias during aerial surveys of elk in northcentral Idaho. Wildl. Manage. 51:622-630. SAS Institute Inc. 1988. Cary, NC. 1028pp. SAS/STAT User's Guide, 1982. Procedures Texas Agric. Expt. 6.03. SAS Institute, J. Inc., for evaluating predation on Sta. Publ. B-1429. 42pp. Wade, D. A., and J. E. Browns. livestock and wildlife. White, G. C. 1996. NOREMARK: Population estimation surveys. Wildl. Soc. Bull. 24:50-52. White, G. C., and R. A. Garrott. 1990. Analysis of wildlife data. Academic Press, Inc., San Diego. 383pp. from mark-resighting radio-tracking �191 Table 1. Causes of deaths in radio-collared elk between 1 December 1993 and 14 June 1998. Calves (M=male, F=female) were age 6-11 months and collared at age 6 months. Yearling males and females were age 12-17 months and juvenile males and females were age 18-23 months and all were collared at age 6 months. Adult males and females were ase >24 months at time of death. "Elk Ase!Sex Class at Death Yearling Juyenile Adult Total Cause of Death Calves All M F M F M F M F and -Code M F Natural causes o o 4 o o o 2 2 Malnutrition-6 2 2 o Unknown-Suspect o o o 3 1 4 Malnutrition-31 3 1 o o o o 1 5 13 Predation-Lion-3 o o o o 8 8 4 o o o o o Predation-Bear-35 o o o o o 0 o o o 1 1 Unknown Predator-5 o 1 o o o o Unknown-Suspect o 2 3 8 11 Predation-30 2 5 o o 1 1 2 o 0 o o 2 o 2 Accident-Birthing-32 o o o o 1 1 2 Accident-Fell-10 o 0 o o 1 o 1 1 o 2 2 4 6 Unknown Cause-11 2 1 o o o Subtotals .Q .Q .Q .a. .l.2. II sa 2. 111l 2. Legal Bunting 8 2 12 o 0 2 8 4 o Archery-33 o o 4 2 4 4 8 Muzzleloading-34 o 2 o o o 0 o 1 o 1 1 o Archery/Muzzle-27 o 0 o o o 29 22 7 22 7 Rifle-First-46 o 0 o o o o 20 o 0 10 10 10 10 o Rifle-Second-47 o o o 8 5 8 5 13 o 0 o Rifle-Third-48 o o o o 6 o 6 6 Rifle-Late-29 o 0 o o o o Subtotals .Q .Q .6..2. .Q .Q .Q sa aa sa II .i Wounding Loss 1 o 0 2 o 1 1 1 Archery/Muzzle-24 o o o 3 o 0 o o 1 1 2 Archery-52 1 1 o 3 2 o 0 o 1 1 2 Rifle-First-43 o o o 9 o 0 o 3 6 3 6 Rifle-Second-44 o o o 2 1 1 1 1 Rifle-Third-45 o 0 o o o o 1 1 o 1 o Rifle-Unk.Reg.-25 o 0 o o o o 3 o 3 o 3 Rifle-Late-26 o 0 o o o o Subtotals .Q .Q .Q .Q aa 1. II .a. .15. ~ ~ Illegal Bunting 5 o o 5 o Rifle-First-49 o 0 o o o 5 6 1 o 6 o Rifle-Second-50 o 0 5 o o o 1 o o o o 0 o o 1 Rifle-Third-51 1 o o 0 1 1 o 1 o Rifle-Unk.Reg.-9 o o o o 1 o o 1 o Archery Season-8 o 0 1 o o o 3 o 2 o 3 Out-of-Season-7 o 0 o o o 1 Subtotals .Q .Q .Q 11 2. 2. II .3. .u: .Q ~ Presumed Buntin~ 1 1 Missing-ArchMuzz-21 o 0 o o o o o 1 o o 0 5 o 2 3 3 2 Missing-Rifle 1st-40 o o o 3 1 Missing-Rifle 2nd-41 o 0 o 1 o 1 1 2 o o 0 o o o o o o Missing-Rifle 3rd-42 o o o Subtotals .Q .Q .Q .Q .Q .2. .3. .5. .5. .i ~ Totals 17 14 13 6 2 3 66 60 98 83 181 • These elk were illegally shot as spike-antlered yearling males: (173.190/93) (173.232/93) (173.309/93) (173.320/93) (173.919/94) (174.059/94) (174.140/95) (174.200/95) (174.679/95) (174.729/95) (174.861/95) (173.210/96). (173.309/93) disappeared in 1994 during first rifle season when spike-antlered yearling males were not legal and was assumed to have been taken illegally. All other illegal deaths were confirmed. • These elk disappeared during hunting seasons and remained missing for several months and are assumed dead and legally harvested: (172.207/93) (172.649/93) (172.800/93) (172.961/93) (173.390/93) (173.439/93) (174.001/94) (174.181/94) (174.770/95). �Table 2. Survival rates, inclusive of nonhunting and hunting mortalities, for winter-spring (WS, 1 December-14 June) and summer-fall (SF, 15 June-30 November) time periods from 1 December 1993-14 June 1998 for the cohort of calves age 6 months radio-collared in December 1993. Survival rate for male and female calves pooled when age 6-11 mohths was 0.92 (95% CI 0.86-0.98, n=73). Survival rates (S) calculated as a mean estimate of (alive)/(alive + dead) and variance S(1-S)/n collars. Elk Elk Age (months) and Time Period (WS or SF dates) 18-23 24-29 30-35 36-41 42-47 6-59 6-11 (mos) l2.:.ll ~ ~ .WS WS SF WS SF WS SF WS SF ALL 1994-95 1995 1995-96 1996 1996-97 1997 1997-98 1994 1993-98 1993-94 MALES Survival L 95% CI U 95% CI n collars censo.redDied Nonhunt Hunt 0.89 0.78 0.99 36 0.88 0.76 0.99 32 0 o b 1.00 28 o o o o 0.21 0.06 0.37 28° 1. 00 0.00 0.00 0.50 0.00 1. 00 1. 00 0.00 0.00 4 4 2f o 22d o o o 2 o o o o 1 19 33 o 2e 1 33 4 4 0 4 0.95 0.87 0.97 0.92 1. 00 1. 00 1. 00 0.97 0.91 1. 00 37 35 34 34 33 0.84 0.71 0.97 32 censo red- o o o o o Died Nonhunt. Hunt 2 2 1h 1 1 5 o o o o 1 o o o o 1 1 5 FEMALES Survival L 95% CI U 95% CI n collars o 4 o 1. 00 o 22 0.97 0.91 o 2 1. 00 27 o o o o 3e,9 o 4 1 29 0.89 0.76 1. 00 27 0.96 0.87 1. 00 24 0.62 0.46 0.78 37 o o o 31 o 1 1 3 3 o 11 • Censored denotes collar failure and/or animal life/death status not known. Collar (173.309/93) disappeared 1994 hunting seasons and assumed dead. • Includes (173.241/93) collar failure August 1995 but seen January 1996. d Two collars (173.390/93). (173.439/93) disappeared 1995 hunting seasons and assumed dead. • Two collars censored; (173.241/93). collar failed August 1995. (173.381/93) slipped off December Includes (173.340/93) alive May 1997. 9 Censored elk was (173.249/93) slipped collar September 1997. • Collar (172.961/93) disappeared 1997 hunting seasons and assumed dead. h Collar (172.800/93) disappeared 1994 hunting seasons and assumed dead. b 1995. 14 ,_. It) (IJ �Table 3. Survival rates, inclusive of nonhunting and hunting mortalities, for winter-spring (WS, 1 December-14 June) and summer-fall (SF, 15 June-30 November) time periods from 1 December 1994-14 June 1998 for the cohort of calves age 6 months radio-collared in December 1994. Survival rate for male and female calves pooled when age 6-11 months was 0.90 (95% CI 0.83-0.97, n=69). Survival rates (S) calculated as a mean estimate of (alive)/(alive + dead) and variance S(l-S)/n collars. Elk Age (months) and Time Period (WS or SF dates) 18-23 24-29 30-35 36-41 42-47 ~ 6-11 (mos) l2...:.l1 WS SF WS SF WS. ALL SF WS 1996-97 1997 1995-96 1996 1997-98 1994-98 1995 1994-~5 MALES Survival L 95% CI U 95% CI n collars censozedDied Nonhunt Hunt 0.91 0.81 1. 00 33b 0 3 3 0 0.90 0.79 1. 00 30b 0.96 0.88 1. 00 26 0.08 0.00 0.19 25 o o 23 o 1c 1 1 3 o 23 o o o o FEMALES Survival L 95% CI U 95% CI n collars 0.89 0.78 0.99 36 0.97 0.91 1. 00 32 0.97 0.90 1.00 30 0.93 0.83 1. 00 29 0.96 0.89 1. 00 27 0.85 0.70 0.99 26 censo.red= o o o o o Died Nonhunt Hunt 4 4 1 2 1 4 o 1e 1 1 o o o o 1 o 2 1 4 3 o 1. 00 2d 0.50 0.00 1. 00 2 1. 00 o o o o o 1 o 1 1 C 31 4 27 1.00 22 o o o o Censored denotes collar failure and/or animal life/death status not known. b Includes (173.949/94) collar failure December 1994 but seen January 1996 Censored elk was (173.949/94) collar failure, which was killed in first rifle season d Includes (174.030/94) alive June 1997 . • Censored elk was (173.719/94) slipped collar May 1996. 0.03 0.00 0.09 32 lC 0.63 0.46 0.79 35 1d 13 5 8 1996. I-' 10 W �Table 4. Survival rates, inclusive of nonhunting and hunting mortalities, for winter-spring (WS, 1 December-14 June) and summer-fall (SF, 15 June-30 November) time periods from 1 December 1995-14 June 1998 for the cohort of calves age 6 months radio-collared in December 1995. Survival rate for male and female calves pooled when age 6-11 months was 0.88 in 1995 (95% CI 0.81-0.96, n=69). Survival rates (S) calculated as a mean estimate of (alive)/(alive+dead) and variance of S(l-S)/n collars. Elk Age (months) and Time Period (WS or SF dates) 6-11(mos) 12-17 18-23 24-29 30-35 6-35 WS SF WS SF WS ALL 1997 1997-98 1995-98 1996-97 1996 1995-96 MALES Survival L 95% CI U 95% CI n collars 0.86 0.74 0.98 35 censored- 2h 0.83 0.68 0.97 2,9. F 5 0.96 0.87 1. 00 24 o 0.23 0.04 0.41 22 1d 17 o 1 1 5 o 17 0.93 0.82 1. 00 4 1e 0.13 0.01 0.24 32 sh,e,d,e o o o 28 1.00 Died Nonhunt' Hunt 5 5 0 FEMALES Survival L 95% CI U 95% CI n collars 0.91 0.81 1. 00 34 0.87 0.75 0.99 31 27 0.83 0.66 0.99 23 18 0.58 0.40 0.76 31 0 o o 2f .19 3f,9 3 3 0 4 2 4 o 1 1 o o o o 4 cenaoredDied Nonhunt Hunt • Censored b Censored Censored d Censored • Censored , Censored 9 Censored C 4 1. 00 o 4 6 22 13 9 denotes collar failure and/or animal life/death status not known. elk were (174.619/95) slipped collar May 1996, (174.800/95) capture mortality. elk was (174.660/95) slipped collar July 1996. elk was (174.671/95) likely collar failure July 1997 . elk was (174.190/95) slipped collar May 1998. elk· were (174.441/95) slipped collar August 1997. and (174.580/95) likely collar elk was (174.470/95) likely collar failure December 1997. failure July 1997. •.... 10 ,j:o �Table 5: Survival rates, inclusive of nonhunting and hunting mortalities, for winter-spring (WS, 1 December-14 June) and summer-fall (SF, 15 June-30 November) time periods from 1 December 1993-14 June 1998 for the cohort of calves age 6 months radio-collared in December 1996. Survival rate for male and female calves pooled when age 6-11 months was 0.86 (95% CI 0.81-0.96, n=69). Survival rates (S) calculated as a mean estimate of (alive)/(alive+dead) and variance of S(l-S)/n collars. Elk Age (months) and Time Period (WS or SF dates) 6-11(mos) 12-17 18-23 6-23 ALL WS SF WS 1996-98 1997-98 1997 1996-97 MALES Survival L 95% CI U 95% CI n collars cenaor-ed- 0.85 0.73 0.98 34 1b Died Nonhunt Hunt 5 5 0 FEMALES Survival L 95% CI U 95% CI n collars censo.redDied Nonhunt Hunt 0.86 0.74 0.98 35 0 5 5 0 • Censored b Censored 1. 00 0.97 0.90 1. 00 29 0 1 0 1 28 0 0 0 0 1. 00 1. 00 30 0 0 0 0 30 0 0 0 0 0.82 0.69 0.96 34 1b 6 5 1 0.86 0.74 0.98 35 0 5 5 0 denotes collar failure and/or animal life/death elk was (174.619/96) capture mortality. status not known. I-' 1.0 til �Table 6. Survival rates, inclusive of nonhunting and hunting mortalities, for winter-spring (WS 1 December14 June) and summer-fall (SF, 15 June-30 November) time periods from 1 December 1993 - 14 June 1998 for all Survival rates (S) calculated as a mean estimate of radio-collared adult female elk age ~12 months. (alive)/(alive + dead) and variance S(l-S)/n collars. Elk Age and Time Period (WS or SF dates) Adult Agylt Adult Adult Adult Adyl!;; Adylt AQylt AQJ.!.lt WS SF WS SF SF WS WS SF WS 1996 1995 1995-96 1996-97 1997 1997-98 1994-95 1994 1993-94 0.94 0.89 0.98 0.94 0.91 0.96 0 .. 99 0.'87 0.96 Survival 0.90 0.84 0.95 0.90 0.87 0.97 0.92 0.80 0.91 L 95% CI 0.94 0.98 1.00 0.98 0.96 1.00 1.00 0.94 1.00 U 95% CI 129b h 1271 100b f 68b e 95b 9 119i 143k 152m 138 n collars j n 1P 2 0 0 2 0 0 0 0 censoxed" 7 16 8d 3 13 2 4 13c 3 Died 4 1 1 0 0 1 1 1 1 Nonhunt 3 15 2 8 13 1 3 12 2 Hunt ---- 0 b Includes (172.011/93) collar failure April 1994, seen January 1996 • Censored denotes collar failure or life/death status unknown. d Collar (172.207/93) disappeared 1995 hunt seasons, assumed dead. • Collars (172.800/93,172.649/93) disappeared 1994 hunt seasons. f Composed of 35-12+, 6-24+, and 59-36+ months old females. • Composed of 6-18+ and 62-30+ months old females. h Composed of 32-12+, 34-24+, and 63-36+ months old females. 9 Composed of 34-18+ and ·61-30+ months old females. j Censored (172.011/93) failure and (173.719/94) slipped collar. 1 Composed of 30-18+ ~d 89-30+ months old females. I Composed of 27-18+ and 100-30+ month old females. k composed ot 31-12+, 29-24+, and 83-36+ months old females. m Composed of 30-12+, 23-24+, and 99-36+ months old females. n Censored (174.4-41/95) slipped collar August 1997 (174.580/95) likely failure June 1997 p Censored (174.470/95) likely collar failure August 1997. o Composed of 30-18+ and 108-30+ months old females. I-' 10 0'1 �Table 7. Survival rates, inclusive of nonhunting and hunting mortalities, for winter-spring (WS 1 December14 June) and summer-fall (SF, 15 June-30 November) time periods from 1 December 1993 - 14 June 1998 for radio-collared adult female elk age ~24 months. Survival rates (S) calculated as a mean estimate of (alive)/(alive + dead) and variance S(l-S)/n collars. Elk Age and Time Period (WS or SF dates) Adult Adyl!;; Adult 8,gult Agylt; 8,gylt; Agylt 8,gylt; Agyl!;; WS SF WS SF SF WS WS SF WS 1995 1995-96 1996· 1996-97 1997 1997-98 1994-95 1994 1993-94 0.93 0.89 0.93 0.99 0.89 0.98 0.82 0.93 0.95 Survival 0.84 0.88 0.88 0.97 0.84 0.96 0.87 0.72 0.90 L 95% CI 0.98 0.99 0.95 1. 00 0.95 1.00 0.91 0.99 1.00 U 95% CI 112 97 89 100 122 108 61 65 62 n collars 1b 1d 0 0 0 2c 0 0 cenaoxed0 7 6 12 1 13 2 4 12 3 Died 3 1 0 0 0 1 1 1 1 Nonhunt ·3 11 7 1 13 1 3 11 2 Hunt • Censored o Censored denotes collars collar failure or life/death (174.441/95) (174.580/95) status unknown. S d Censored Censored collar collar (172.011/93) 174.470/95) Table 8. Survival rates, inclusive of nonhuntng and hunting mortalities, for winter-spring (WS, 1 December14 June) and summer-fall (SF, 15 June-30 November) time periods from 1 December 1993 - 14 June 1998 for the group of adult female elk age ~18 months when radio-collared in December 1993. Survival rates (S) calculated as a mean estimate of (alive)/(alive + dead) and variance S(l-S)/n collars. Elk Age and Time Period (dates) Adult Adult 8,dylt Agylt; 8,dult Agyl!;; Adyl!;; 8,gl.l.lt My,lt WS SF SF WS WS SF WS SF WS 1995 1995-96 1996 1994-95 1996-97 1997 1994 1997-98 1993-94 0.93 0.92 0.88 1. 00 0.93 0.94 0.82 0.97 0.96 Survival 0.85 0.78 0.84 0.85 0.87 0.72 0.92 0.91 L 95% CI 0.97 1.00 1.00 0.99 1.00 0.91 1.00 1.00 U 95% CI 49b h 53b f 42h 65b f 68b e 391 361 36j 34j n collars 19 0 0 0 0 0 0 0 censor-ed- 0 d c 3 3 6 0 4 2 12 1 3 Died 0 3 0 0 1 0 0 1 1 Nonhunt 0 3 6 0 3 2 11 1 2 Hunt • Censored denotes collar failure or life/death status unknown o Collar (172.649/93) disappeared 1994 hunt seasons. • Composed of 6-·18+ and 62-30+ month old fe!llales. • Censored (172.011/93) collar failure April 1994. , Composed of 100% females 48+ months old. Includes (172.011/93) collar failure April 1994 but seen Jan. 1996. Collar (172.207/93) disappeared 1995 hunt seasons, assumed dead. , Composed of 100% females 24+ months old, h Composed of 100% females 36+ months old. l Composed of 100% females 60+ months old. b d I-' 10 -.J �Table 9. Summary of summer-fall (15 June-30 November) and winter-spring (1 December-14 June) survival rates inclusive and exclusive of hunting deaths among different age groupings of adult females averaged among years, 1993-94 - 1997-98. Survival rates (S) calculated as a mean estimate of (alive)/(alive + dead) and variance S(l-S)/n elk-years. Winter-Spring Average 5 Years Summer-Fall Average 4 Year's Age Grouping survival ±95% CI Elk-years Age Grouping Survival ±95% eI Elk-years Includes Hunting Age ~12 monthsb Age ~24 monthsC 1993 Cohortd Deathsa 0.90 0.89 0.88 0.03 0.03 0.05 524 396 189 Age ~18 months Age ~30 months 1993 Cohort 0.97 0.96 0.95 0.01 0.02 0.03 547 420 233 Excludes Hunting Age ~12 monthsb Age ~24 monthsC 1993 cohortd Deathse 0.99 0.99 0.99 0.01 0.01 0.01 476 354 167 Age ~18 months Age ~30 months 1993 Cohort 0.99 0.98 0.98 0.01 0.01 0.02 536 410 227 "Includes "Includes dIncludes "Includes hunting and natural deaths. bIncludes all females age ~12 months. only females age ~24 months by censoring yearling females recruited from calves. only females in 1993 trapped cohort having age 18 months to 10+ years. only natural deaths and represents natural survival rate. Table 10. Annual survival rates from 1 December-30 November each year 1993-94 - 1996-97 and overall survival through 14 June 1998 for the group of adult female elk age ~18 months when radio-collared in December 1993. Survival rates (S) calculated as a mean estimate of (alive)/(alive+dead) and variance S(1S)/n collars. Survival L 95% CI U 95% CI n collars censor-ed" Died Nonhunt Hunt Adult 19931994 0.78 0.68 0.88 68b•e 0 15C 2 13 Elk Age and Time Period Agult Adult Adult 19961995 1994 1997 1996 1995 0.94 0.86 0.84 0.87 0.75 0.75 1. 00 0.97 0.93 53b•f 36i 429 1h 0 0 d 2 6 10 0 3 1 2 3 9 " Censored denotes collar failure or life/death status unknown. " Collar (172.649/93) disappeared 1994 hunt seasons. • Composed of 6-18+ and 62-30+ month old females. 9 Composed of 36+ month old females. I Composed of 48+ month old females. (dates) AgyH 19931998 0.51 0.39 0.63 67 1h 33 6 27 Includes collar (i72.011/93) failed 4/1994 but seen alive 1/1996. Collar (172.207/93) disappeared 1995 hunt seasons, assumed dead . ' Composed of 100% females 24+ months old. h Censored (172.011/93) failure 4/1996. b d •.... 10 co �Table 11. Correction factors calculated to estimate total numbers of elk removed by all sources of hunting mortality from total numbers of elk 'legally removed' by hunting during 1994-1997 hunting seasons. Wounding Illegal Legal Total Correction Elk Sex/Age Class Or Loss Killb .xi u> xu iKilledc Factord Hunting Season fQ~ ~~~L8S~ ~lg~~~~ 57 7 2 66 0 1 12 13 1.16 1.11 e 41 15 3 59 1.44 35 12 3 50 1. 43f 37 14 2 53 1. 43 5 1 1 7 1.40 44 6 2 52 1.18 26 9 0 35 1.35 Rifle Late Season for Females Age ~ 12 months 6 3 3 12 2.00 Archery for Males Age ~ 99 0 0 9 1. 00 4 3h 0 7 1. 75 4 1 0 5 1.25 Males Age ~ 24 months Males Age 12-23 months Females Age ~ 12 months Females Age ~ 12 months Females Age ~ 24 months Females Age 12-23 months EQ~ Hynting ~easQn~ Rifle pt, 2nd, & 3rd for Males Age ~ 24 months Rifle pt, 2nd, & 3rd for Females Age ~ 12 months 24 Archery for Females Age ~ months 12 mbnths Muzzleloading for Males Age ~ 24 months Muzzleloading for Females Age ~ 12 months 0 5 0 5 1. 00 • Legal kill equals known-killed plus dl.sappeared during hunting seasons and assumed legally killed. These eLk represent kill that could be estimated by hunter harvest surveys. b Wounding loss and illegal kills would not be estimated by hunter harvest surveys. Total elk removed by all sources of hunting. d Multiply legal kill by correction factor to estimate total elk removed by hunting. For yearling males, mUltiply modeled numbers of yearlings by 1.11 if yearling males are not legal quarry . • Removal rate measured from known illegal hunting. ! Correction factor calculated without including mortalities associated with late rifle seasons. 9 Includes one legal kill that may have been from either archery or muzzleloading seasons. h Includes one wounding loss that may have been from either archery or muzzleloading seasons. the legal C estimates of I-' 10 10 �Table 12. Data matrix for simple population model using spreadsheet software. We used a post-hunt (December) population composition of 50 calves:100 females age L12 months and 4 adult males:16 yearling males:100 adult females and an estimated population of 3,600 elk. Fall Hunting Natural Survival Natural Survival Removal Rate Elk Sex/Age Class Rate Winter-Spring Rate Summer-Fall Elk Sex/Age Class 0.80 Adult Males 1. 00 1. 00 Adult Males Age L30 months Age 2.24 months Yearling Males Age 12-17 months 1. 00 0.11 Yearling Males Age 18-23 months 0.98 Adult Females Age L 24 months 0.99 0.11 Adult Females Age 2.30 months 0.98 Yearling Females Age 12-17 months 1. 00 0.05 Yearling Females Age 18-23 months 0.98 Calves Age 0-5 months 1. 00 0.02a Calves Age 6-11 months 0.89 • Estrmated from harvest surveys. IIJ o o �201 Table 13. Elk counted during quadrat and nonrandom flights on 44 mi2 in West Divide and Alkali creeks, GMU 42, 27 February-2 March,1998. Survey Marked Elk Unmarked Total Marked/ Flight Flight Available Seen Elk Seen Elk Seen Unmarked Hours Quadrats 34A 27 744 771 0.036 17.2 NonRandom-1 38 12 420 432 0.029 3.6 NonRandom-2 38 17 534 551 0.032 4.4 Totals 38 56 1698 1754 0.033 25.2 a Four elk off sample area during quadrat counts were on sample area during nonrandom flights. Table 14. Summary of quadrat flights for each observer, 27 February - 1 March, 1998. Total Minutes Marked Elk/ Sighting Trial Seen/ Flown/. Observer Quadrats Total Unmarked Quadrat Quadrat Marked Elk (Date) Flown (Range) (Range) Seen(%) Missed Elk Seen Seen 20.6 1B 18 371 0.039 18.5 14 1 (0-68) (2/28) (0.93) (10-27) 2E (2/27) 17 3M (3/01) 9 loa 5 314 0.040 18.5 (0-55) 17.8 (10-35) 1 86 0.012 9.6 (0-29) 25.4 (18-39) (0.67) 1 (0.50) 17.5 19.7 771 0.036 25 7 (0-68) (10-39) . (0.78) aTwo elk groups each contained 2 marked elk that were seen but each group represented only 1 sighting trial. Observer 2E saw a total of12 marked elk. Totals 44 Table 15. Status of marked elk missed on quadrats during sighting trials and subsequently found using telemetry. Marked Statul2When Marked Elk Foynd %Snow %Screening Group Elk Cover Cover Habitat Observer Missed Size Behavior Sex 100 Juniper 50 Standing 1 1 IB M 2E 5 F F M F M 1 sa 1 1A 1 Bedded Bedded Bedded Moving Bedded Oakbrush Oakbrush Oakbrush Oakbrush Oakbrush 70 70 70 70 60 100 100 100 100 100 2A 50 30 Juniper Standing 3M 1 F AThese 3 elk may represent groups or portions of groups of elk that were detected during quadrat counts but marked elk and other elk in the group were not seen due to elk behavior or vegetation. For all 3 groups, additional marked or unmarked elk were seen in close proximity. �202 Table 16. Frequencies at which individual marked elk were seen during quadrat and nonrandom flights, 27 February-2 March, 1998. Maximum frequency equals 3. Frequencies represent heterogeneity of sighting probabilities among elk Total Marks Frequency Seen During Flightsa Frequency Seenb Seen 0 1 2 3 1 56 5 20 8 5 5 a Of the 56 marked elk seen, 46 or 82% were individually identified by numbers/symbols on the collars or telemetry frequency. Elk not identified had lost the numbers/symbols on collars and were seen during nonrandom flights. b Includes 5 observations of elk seen but not individually identified. C Of the 5 elk not seen, 3 were females and 2 were males. Table 17. Estimated numbers of elk in survey area based on quadrats, sighting bias adjusted quadrats, and Bowden and Immigration-Emigration Joint Hypergeometric mark-resight estimators calculated using program NOREMARK. Both mark-resight estimators account for populations not closed to movement during time of surveys. Conf. Int. Width 25% Conf. Limits Total Elk Upper Percenta Lower Elk Estimator Estimate 771 Quadrats 944 14 124 882 820 Quadrats Adj. Model 1b c 952 814 16 138 Quadrats Adj. Model 2 883 14 116 915 857 799 Quadrats Adj. Model 3d 1,430 457 973 39 Bowden Estimator 1,179 1,337 370 33 967 Imm/EmmJHE Estimator l,115e a Width as percent of total estimate. b Quadrat counts adjusted; (LOG GROUP SIZE)+(INTERCEPT); AIC=172. C Quadrat counts adjusted; (LOG GROUP SIZE)+(BEDDED)+(INTERCEPT); AIC=160. d Quadrat counts adjusted; (LOG GROUP SIZE)+(BEDDED)+(%COVER)+(INTERCEPT); AIC=154. e Represents daily population estimate on sample area. Extant population estimate was 1,162 elk. Table 18. February-1 Total Groups 148 Number and size of elk groups seen during quadrat March 1998. FI:~gyen!;;;:l!: of Grou:Q Sizes Seen 25-30 10-24 7-9 4-6 1 2-3 2 18 21 34 35 38 counts, 27 Average Group Size 5.2 �203 Table 19. Radio-collared elk captured as calves on winter range in GMU 42 during December 1996 that dispersed out of GMU 42 to a new winter range as of March 1998. Locations (GMUs) determined by aerial telemetry. Trap New Frequency Zone Agea Sex GMU Location Description porcupine Creek-Holms Mesa 172.899/96 G 21 F 42 West 173.450/96 Wallace Creek-Samson Mesa G 21 M 42 West East Fork Wallace Creek 173.789/96 42 West G 21 F Low Beaver Creek 173.870/96 G 21 F 42 West Plateau Creek-Hayes Mesa 174.059/96 G 21 M 421 Taughenbauh-Grass Mesas 174.220/96 F 21 42 West M Porcupine Creek-Holms Mesa 174.789/96LE G 21 M 42 West 174.929/96 D 421 Pole Gulch-Collier Creek 21 F East Muddy-Little Henderson Creek 174.949/96LE D 21 521 F Hawxhurst Creek~Grassy Gulch 174.969/96 D 421 F 21 East Muddy-Little Henderson Creek 174.998/96 E 521 21 F Hawxhurst Creek 175.020/96 F 421 21 F Low Beaver Creek 175.039/96 G F 42 West 21 West Muddy-Ault Creek 175.050/96LE E 521 F 21 42 West Wallace Creek-Sugarloaf Mtn. 175.070/96 E F 21 Low Fourmile Creek 175.089/96 A F 43 21 Wallace Creek-Sugarloaf Mtn. 175.180/96 E M 42 West 21 Plateau Creek-Red Mountain 175.189/96LE E M 421 21 Low Wallace Creek-High Mesa 175.199/96 G M 42 West 21 a b Age of elk in months as of March 1998. LE = Location elk routinely located. �204 Appendix I. Mortalities of 181 radio-collared elk from 1 December 1993 through 14 June 1998. Age is aBeroximate age in years of elk at death: C=calf age 6-11 months, Y=yearling age 12-23 months. Frequency 101 Q~ath Ira!2 Age Year Ca~tured Site Zone Sex Date Cause of Death & Code Number 172.030/93 BR A F 5 27-0ct-95 Legal kill second rifle season-47 172.039/93 GR B F 4 14-Jun-95 Unknown-suspect predation-30 172.070/93 BR A F 11 12-Feb-96 Unknown-11 172.080/93 GM A F 6 05-Nov-94 Legal kill third rifle season-48 172.090/93 GR B F 11 16-Jan-94 Wounding loss late rifle season-26 172.101/93 SR B F 8 14-0ct-96 Legal kill first rifle season-46 172.128/93 GC B F 8 31-0ct-96 Wounding loss second rifle season-44 172.139/93 F BC C 17 19-0ct-95 Wounding loss first rifle season-43 172.160/93 CC C F 3 23-0ct-94 Legal kill second rifle season-47 172.181/93 MC C F 3 03-Nov-94 Wounding loss second rifle season-44 172.201/93 GS C F 3 15-Nov-94 Wounding loss second rifle season-44 172.207193 SG 0 F 4 30-Nov-95 Disappear first rifle season-40 172.258/93 HY F C 3 04-0ct-94 Legal kill archery/muzzle season-27 1n.277/93 GS C F 3 04-0ct-94 Wounding loss archery/muzzle-24 172.290/93 SG 0 F 6 23-0ct-94 Legal kill second rifle season-47 172.290/94 SM 0 F 4 16-Sep-96 Legal kill muzzleloading season-34 172.300/93 AC E F 8 30-0ct-97 Wounding loss second rifle season-44 172.369/93 AC E F 3 18-Sep-94 Legal kill archery season-33 172.369/94 FM B F 4 14-Jan-96 Legal kill late rifle season-29 172.409/93 AC F E 8 29-0ec-94 Wounding loss late rifle season-26 172.459/93 MH E F 8 24-0ct-96 Legal kill second rifle season-47 172.509/93 FS H F 3 21-0ct-95 Legal kill second rifle season-47 172.542/93 FS H F 01-Feb-94 Mountain lion predation-3 6 172.549/93 PG H F 7 28-Nov-95 Legal kill late rifle season-29 172.570/93 PG H F 16 03-Nov-94 Wounding loss second rifle season-44 172.581/93 PG H F 3 17-0ct-94 Legal kill first rifle season-46 172.581194 FM B F 3 08-0ec-95 Legal kill late rifle season-29 172.590/93 PG H F 24-Apr-96 Unknown-11 5 172.610/93 PG F 04-0ct-94 Accident, birthing/calving-32 H 3 172.639193 PG F 14-Jun-96 Accident, birthing/calving-32 H 16 172.649193 PG F 30-Nov-94 Disappear first rifle season-40 H 2 172.658/93 PG F Legal kill third rifle season-48 H 5 02-Nov-97 16~Jan-95 172.670/93 PG H F Legal kill late rifle season-29 4 172.678/93 PG H F 22-0ec-94 Wounding loss late rifle season-26 9 172.690/93 FM B F 24-Jan-94 Illegal kill-7 8 172.699/93 Legal kill third rifle season-48 FM B F 7 05-Nov-95 Unknown-suspect predation-30 172.749/95 LS H F 11 07-Aug-96 172.800/93 Disappear second rifle season-41 SR B F Y 30-Nov-94 172.821/93 EG Legal kill archery season-33 B F 3 08-Sep-96 Legal kill third rifle season-48 172.890/93 F 06-Nov-96 BC C 3 Mountain lion predation-3 172.899/93 BC C F 18-Mar-94 C Legal kill first rifle season-46 172.950/93 GH 0 F 18-0ct-95 2 172.961/93 F 30-0ct-97 Disappear second rifle season-41 MG 0 4 173.000/93 25-Apr-94 Malnutrition-6 \lM G F C 07-Jan-98 Accident-fell-10 173.010/93 \lM G F 4 Legal kill first rifle season-46 173.041/93 M 19-0ct-95 GM A 2 27_'Oec-95 Legal kill late rifle season-29 173.060/93 MH E F 2 Illegal kill-7 03-Mar-98 173.070/93 MH E F 5 Legal kill second rifle season-47 173.081193 FS H F 3 22-Oct-96 Legal kill second rifle season-47 24-0ct-97 173.090/93 FS H F 4 Legal kill second rifle season-47 173.100/93 F 30-0ct-97 FS H 4 Legal kill first rifle season-46 .14-Oct-95 173.120/93 GM A M 2 Wounding loss second rifle season-44 31-0ct-96 173. 140i93 PG H F 3 Legal kill third rifle season-48 173.160/93 FS F 13-Nov-96 H 3 Illegal kill second rifle season-50 BC 29-0ct-94 173.190/93 C M Y Wounding loss thtrd rifle season-45 173.201193 GR M 30-Nov-95 B 2 Legal ki II first rifle season-ee 19-0ct-95 173.210/93 GC B M 2 Illegal kill archery/muzz. season-8 30-0ct-97 173.210/96 LM 0 M 2 Legal kill third rifle season-48 06-Nov-95 173.219/93 BC M 2 C Illegal kill second rifle season-50 15-Nov-94 M Y 173.232193 BR A 18-Mar-94 173.262/93 BC M C Unknown-" C Legal kill first rifle season-46 M 12-0ct-96 173.262/94 HM B 2 Legal kill third rifle season-48 06-Nov-95 173.269/93 MC C M 2 Legal ki LL. first rifle season-46 12-0ct-96 MC 3 173.279/93 C M Unknown-11 22-Mar-94 MC C M C 173.289/93 Legal kill muzzleloading season-34 M 18-Sep-96 173.289/94 PR A 2 Legal kill second rifle season-47 24-Oct-95 GS C M 2 173.300/93 Illegal kill first rifle season-49 30-Nov-94 HY C M Y 173.309/93 Illegal kill first rifle season-49 20-0ct-94 HY C M Y 173.320/93 Legal kill archery season-33 14-Sep-96 HM M 2 173.320/94 B !~~}~L~~ _____ ~~__E_____ ~____ ~___ ~~~~~~:?] ______ ~~2!_~L~~~!~~~~2~l~[_s:2~~~:~~ __________________ . �205 A~ndix I. (continued) Frequency 10/ Ira~ Year CaE!tured Site Zone 173.340/93 D SG 173.351/93 SG D 173.359/93 AC E 173.370/93 MH E 173.370/96 RG E 173.381/96 RG E D 173.390/93 MG 173.402193 MG D 173.410/93 WM G WM 173.420/93 G 173.429/93 MH E 173.439/93 PG H 173.450/93 PG H 173.461/93 PG H 173.461/94 CM A 173.469/93 FS H 173.469/94 PR A 173.479/93 FS H 173.492/93 FS H 173.502193 FS H 173.510/93 FM B 173.521/93 FM B 173.540/94 OG B B 173.549/94 HM 173.569/94 PR A A 173.580/94 PR 173.589/94 OG B 173.610/94 BC C 173.640/94 MC C 173.640/96 GM A 173.649/94 BC C BG 173.689/94 E 173.789/94 E MR MM F 173.819/94 MM F 173.829/94 173.859/94 SM D 173.870/94 SM D 173.919/94 BC C 173.929/94 BC C 173.939/94 BC C BC 173.959/94 C C 173.970/94 MC 173.981/94 MP D 173.990/94 BC C 174.001/94 BG E E 174.009/94 BG BG E 174.019/94 MR E 174.039/94 174.049/94 MM F E 174.059/94 MR E 174.069/94 BG 174.080/94 MR E E 174.090/94 MR ,174.101/94 F MM 174.109/94 MM F F 174.119/94 MM 174.119/95 GB B F 174.129/94 MM 174.140/94 MM F B 174~140/95 GB F 174.150/94 MM F 174.160/94 MM FM B 174.170/94 FM B 174.181/94 MS F 174.200/95 MS F 174.210/95 MS F 174.220/95 MS F 174.230/95 MD E 174.240/95 MS F 174.319/95 MD E 174.329/95 Sex M M M M M M M M M M M M M M M M M M M M M M M F F F F F F F F F F F F F F M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M F F Q~ath Date 20-0ct-97 05-Nov-95 06-Sep-95 08-Nov-95 08-Jan-97 19-Feb-97 30-Nov-95 18-0ct-95 02-Nov-95 24-0ct-95 14-Sep-95 30-Nov-95 15-0ct-95 07-Feb-94 19-5ep-95 25-Apr-94 20-0ct-96 10-Sep-95 03-Sep-95 13-Nov-96 28-Aug-95 17-0ct-95 t3-0ct-96 07-Sep-95 16-0ct-97 21-Dec-95 01-Jun-95 14-Nov-97 30-Mar-95 16-Mar-97 11-0ct-97 31-0ct-96 20-Mar-95 30-0ct-97 10-Jan-97 23-0ct-96 21-Feb-95 02-Nov-95 16-0ct-96 18-Sep-96 02-Nov-96 20-0ct-96 02-Nov-96 20-0ct-96 30-Nov-96 12-Oct-96 13-Noy-96 05-Sep-96 14-Sep-96 17-0ct-95 26-Sep-96 11-0ct-97 20-Oct-96 24-May-96 19-0ct-96 01-Mar-95 30-Noy-97 14-0ct-96 14-Mar-95 30-Noy-96 14-0ct-96 15-Sep-96 25-Apr-95 30-Nov-96 18-0ct-96 11-Oct-97 08~Apr-96 24-Apr-96 17-Apr-96 C 02-Oct-96 Y 03-Sep-96 Y Age 4 2 2 2 C C 2 2 2 2 2 2 2 C Y C 2 2 2 3 2 2 2 Y 3 Y C 3 C C 3 2 C 3 2 2 C Y 2 2 2 2 2 2 2 2 2 2 2 Y 2 3 2 Y 2 C 2 2 C Y 2 2 C 2 Y 2 C C !~~~3)2L~~ _____ ~~ ___~ ____ ~ ____ ~___ Cause of Death & Code Number Legal kill second rifle season-47 Legal kill third rifle season-48 Legal kill archery season-33 Legal kill first rifle season-46 Mountain lion predation-3 Unknown-suspect predation-30 Disappear first rifLe season-40 Legal kill first rifle season-46 Wounding loss second rifle season-44 Legal kill second rifle season-s? 'Legal kill archery season-33 Disappear first rifle season-40 Legal kill first rifle season-46 Malnutrition-6 Wounding loss archery season-50 Malnutrition-6 Legal kilL second rifLe season-47 Legal kill archery season-33 Legal kiLL archery season-33 Legal kill third rifle season-48 Legal kiLL archery season-33 Legal kill first rifle season-46 Legal' kill first rifle season-46 Legal kiLL archery season-33 Wounding loss first rifle season-43 Unknown-11 Unknown-suspect predation-30 Wounding loss third rifLe season-45 Unknown-suspect predation-30 Malnutrition-6 Legal kill first rifle season-46 Legal kiLL first rifle season-46 Unknown-suspect maLnutrition-31 Legal kill second rifle season-47 LegaL kill late rifle season-29 Legal kill second rifle season-47 Mountain lion predation-3 Illegal kill second rifle season-50 Legal kill first rifle season-46 Legal kill archery season-33 Legal ki II third riHe season-48 Legal kill second rifle season-47 Legal kill third rifle season-48 Legal kill second rifle season-47 Disappear archery/muzzLe season-21 Legal kill first rifle season-47 ILlegal kiLL rifle season-9 Legal kill archery season-33 Legal kilL muzzleloading season-34 Illegal kill first rifLe season-49 Wounding loss archery/muzzle-24 Legal kill first rifle season-46 Legal kill second rifle season-47 Unknown-suspect predation-30 Legal kilL second rifle season-47 Mountain lion predation-3 Wounding loss rifle season-25 Legal kill first rifle season-46 Mountai'n lion predation-3 Illegal kill third rifle season-51 Legal kill first rifle season-46 Legal kill muzzleloading season-34 Mountain lion predation-3 Disappear second rifle season-41 Illegal kilL first rifle season-49 Legal,kill first rifle season-46 Unknown-suspect malnutrition-31 Mountain lion predation-3 Mountain lion predation-3 Wounding Loss archery season-52 Legal kill archery season-33 lI~~~~:?~ ______ u_n!~2~JP~~~!~~:? _________________________________ : �206 Appendix I. (continued) Frequency 101 Ie!!!:! Year Captured Site Zone 174.349/95 MD E 174.360/95 MD E 174.401/95 AP D 174.420/95 AP D \JI) 174.478/95 C 174.491195 lB C 174.500/95 lB C 174.520/95 KR B 174.520/96 GM A 174.560/95 Bl A 174.609/95 GB B 174.629/95 MD E 174.639/95 MD E 174.679/95 HG E 174.689/95 AP D 174.689/96 EW G 174.700/95 AP 0 174.710/95 AP 0 VM 0 174.729/95 VM 174.740/95 0 VM 174.750/95 0 174.770/95 WD C WD C 174.780/95 174.789/95 WD C 174.800/96 BG E 174.809/95 WD C 174.820/95 WD C 174.830/95 lB C 174.851/95 lB c 174.861/95 KR B lB c 174.870/95 KR B 174.880/95 174.889/95 we G Bl A 174.899/95 lM 0 174.910/96 Be c 175.059/96 175.130/96 GM A lM 0 175.170/96 [1g!!!1:! Sex F F F F F F F F F F F M M M M M M M M M M M M M M M M M M M M M M M F F Age 2 Y F c M 2 2 2 2 C Y C y' C 2 2 Y C C 2 2 Y 2 2 2 2 C C Y 2 2 2 Y 2 2 2 2 c c c Date 13-0ct-97 22-Apr-97 22-Sep-96 11-0ct-97 13-Sep-97 10-Sep-97 16-Apr-96 21-Sep-96 25-Apr-97 14-May-97 '15-Apr-96 11-0ct-97 21-0ct-97 13-Nov-96 05-Feb-96 14-May-97 16-0ct-97 15-0ct-97 17-0ct-96 08-Nov-97 05-Nov-97 16-0ct-97 11-0ct-97 22-May-96 09-Apr-97 22-Apr-97 14-0ct-97 13-0ct-97 22-0ct-97 31-0ct-96 30-0ct-97 11-0ct-97 20-0ct-97 30-0ct-97 27-Feb-97 13-Feb-97 24-Mar-97 26-Mar-97 Cause of Death & Code Number legal kill first rifle season·46 Unknown-suspect predation-30 legal kill muzzleloading season·34 legal kill first rifle season-46 legal kill muzzleloading season-34 Wounding loss archery season-52 Unknown-suspect predation-30 legal kill muzzleloading season·34 Mountain lion predation-3 Illegal kill-7 Unknown-suspect predation-30 legal kill first rifle season-46 legal kill second rifle season-47 Illegal kill second rifle season-50 Mountain lion predation-3 Unknown-suspect malnutrition-31 Wounding loss first rifle season-43 legal kill first rifle season-46 Illegal kill first rifle season-49 legal kill third rifle season-48 legal kill third rifle season-48 Disappear first rifle season-40 legal kill first rifle season-46 Unknown-suspect predation-30 Mountain lion predation-3 Accident, fell-10 legal kill first rifle season-46 legal kill first rifle season-46 Illegal kill second rifle season-50 Illegal kill second rifle seasonc50 Wounding loss second rifle season·44 legal kill first rifle season-46 legal kill second rifle season-47 Wounding loss second rifle season·44 Unknown-suspect predation·30 Unknown-11 Mountain lion predation·3 Unknown-suspect malnutrition-31 �172 �171 Colorado Division of Wildlife Wildlife Research Report July 1999 JOB PROGRESS REPORT Smteof ~C~o~lo~r~ad~o~ _ Mammals Research Project No. ---'W.:.,.--=1...:,5;::;..3-..:.R::....-=12=--_ Elk Investigations Work Package __ --=3::...:0:;,,::0:,:2'-_ Task No. ---=-1 _ Estimating Survival Rates of Elk and Developing Techniques to Estimate Population Size Period Covered: July 1, 1998 - June 30, 1999 Author: D. 1. Freddy Personnel: R. Adams; T. Beck J. Broderick, G. Byrne, D. Crane, R. Del Piccolo, J. Ellenberger, V. Graham, D. Homan, R. Kahn, K. Madariaga, D. Masden, C. Mehaffy, P. Neil, J. Olterman.J, Thompson, P. Will, K. Wright, S. Yamashita, D. Younkin, CD OW; D. Bowden, G. White.T, Gross, M. Kneeland, W. Andelt, CSU; D. Ouren, D. Schneider, T. Fancher USGS-BRD, BLM Glenwood Springs, CO, USFS Rifle, CO, Rocky Mountain Elk Foundation, cooperating. ABSTRACT We monitored survival and movements of 169 previously radio-collared elk composed of 136 adult females and 33 adult males from 1 July 1998 through 30 June 1999. Survival of adult females during summer-fall was 0.77 ± 0.07 (n = 136) inclusive of hunting deaths and 1.00 (n = 105) with hunting deaths censored while survival during winter-spring was 0.98 0.03 (n = 103). The 2 deaths of adult females during winter-spring were attributed to accident and malnutrition. Survival of adult males during summer-fall was 0.34 0.17 (n = 32) inclusive of hunting deaths and 1.00 (n = 11) with hunting deaths censored while survival during winter-spring was 1.00 (n = 11). A combination of unlimited either-sex elk hunting licenses and high numbers of antlerless elk hunting permits during rifle hunting seasons in 1998 contributed to removing 23% of the adult females which was sufficient to temper population growth. Progress continues towards developing habitat use models to predict areas important to managing elk. Draft manuscripts on elk survival rates and techniques to estimate population size are progressing. ± ± �173 ESTIMATING SURVIVAL RATES OF ELK AND DEVELOPING ESTIMATE POPULATION SIZE TECHNIQUES TO David 1. Freddy P. N. OBJECTIVE Estimate survival rates of adult female, adult male, and calf elk and develop techniques to estimate population size. SEGMENT OBJECTIVES 1. Estimate winter, summer, and annual survival rates of adult female and male elk from fates of previously radio-collared elk. 2. Develop an initial habitat use prediction model for elk inhabiting oakbrush dominated habitats based upon locations of radio-collared elk obtained from 1993 -98. 3. Analyze data, summarize annually in Federal Aid Job Progress Report, and prepare draft manuscripts regarding elk survival rates and techniques to estimate population size. INTRODUCTION Our objectives are to provide reliable estimates of survival rates for calves during winter and for adult females and males throughout the year and to develop and test a system to estimate elk densities on winter ranges. Estimates of calf survival were obtained during winters 1993-94 through 1996-97 and summarized in Freddy (1997). We are continuing to obtain estimates of survival for adult males and females. For adults, we are interested in survival rates inclusive of hunting mortalities to document human-induced rates of survival and exclusive of hunting mortalities to estimate natural rates of survival. We are evaluating a system for estimating population size or density that incorporates estimates of sighting bias in conjunction with random sampling of search quadrats as sample units. Our winter study area encompasses about 324 rnf (839 knr') in the eastern half of Game Management Unit 42 south and east of Rifle, Colorado. This area is part of the Grand Mesa Elk Data Analysis Unit E-14. Elk winter habitats include juniper-pinyon woodland (Juniperus osteosperma-Pinus edulis), oakbrush-mountain shrub (Quercus gambelii-Amelanchier alnifolia), aspen (Populus tremuloides), sagebrush (Artemisia tridentata), and agricultural fields below elevations of9,800 ft. In summer, elk use oakbrush-mountain shrub, aspen, and subalpine fir-Engelmann spruce (Abies lasiocarpa-Picea engelmannii) meadow systems throughout the Grand Mesa region up to elevations of 12,000 ft. (Freddy 1993). METHODS We placed radio collars (172-176MHz) having mortality sensors on elk calves age 6 months and adults age 2:18 months during December 1993-1996 (Freddy 1997). Elk were captured using helicopter net-gunning and portable corral traps (Freddy 1997). During 1998-99, we monitored 169 adult elk, 136 females and 33 males, having functioning radios as of July 1998. Collars were white and had black identification symbols on dorsal surfaces of collars to allow visual identification of individual elk. �174 Estimating Elk Survival Rates We monitored survival of radioed elk with aerial surveys using a Cessna 185 at 2-4 week intervals from December 1993 to June 1999 and with daily ground surveys from December-April each year from 1993-94 to 1996-97. Survival rates (S) of radioed elk were calculated using the binomial estimator with a variance, V AR(S) = S(I-S)/n collars (White and Garrott 1990). Survival rates are expressed as the mean estimate the 95% confidence interval. We used x2-contingency tests to compare survival rates between sexes and among time intervals (pROC FREQ, SAS 1988). Sample sizes refer to numbers of individual radioed elk. We defined 3 major time intervals for survival analyses: winter-spring was 1 December to 14 June, summer-fall was 15 June to 30 November, and annual was 1 December to 30 November to coincide with timing of capture and collaring. For adult elk during these time intervals, we calculated survival rates inclusive of natural and hunting-related mortalities, exclusive of hunting mortalities, and exclusive of natural mortalities. Excluding, or censoring hunting mortalities, provided estimates of natural survival rates. Excluding natural mortalities but including hunting mortalities provided estimates of hunting removal rates. Censoring elk associated with categories of mortalities reduced sample sizes used to estimate survival rates. Elk were also censored ifradios failed or if their life/death status was unknown after an extended period of time. Archery, muzzleloading, and rifle hunting seasons occurred from about 1 September to 15 November each year during the summer-fall time interval. Late-rifle seasons occurred in some years . and restricted hunters to taking only antlerless elk from about 25 November-31 January which included portions of the summer-fall and winter-spring time intervals. Yearling spike-antlered males were generally not legal quarry in most areas frequented by radioed elk. Male elk were legal quarry when branch-antlered usually at age 2 years. Cause of death of radioed elk recovered in the field was estimated from presence or absence of gunshot wounds or bite wounds on carcass, predator tracks or scat at carcass site, physical positioning of carcass remains whether buried, covered, scattered, or consolidated (Wade and Browns 1982, Halfpenny and Biesiot 1986), relative amount of internal fat and marrow fat if present with carcass, and results of histopathology and marrow fat analyses. Fat content (percent dry matter) of bone marrow and estimates of age based on dental cementum were obtained for dead elk by the Colorado Division of Wildlife Laboratory while histopathology analyses were provided by the Colorado State University Veterinary Diagnostic Laboratory. Photographs were taken of nearly all mortalities so that physical evidence could be reviewed and judged by outside experts (pers. comm. T. Beck, W. Andelt). ± Elk Movements and Habitat Use Models We continued to precisely locate selected radioed elk at least once per month since their capture to document seasonal movements using a Cessna 185. These elk were originally selected at random from within trap zones and equalized by age class in January 1994. Elk from the original sample that died were replaced each subsequent January with randomly selected elk that were usually of the same sex, age class, and trap zone as elk that died. All replacement elk in January 1999 were ;:::2-years old. As of 1 January 1999, these 43 elk were classified as 32 adult females and 11 adult males males. During June 1999, we again located an additional 28 adult females that were originally selected at random in 1995 to document their locations during the calving period. Females from the original 1995 sample that died were replaced each subsequent June with adult females randomly selected from the same trap zones of elk that died. As needed, we located other elk to document unusual movements. We continue to cooperate with the USGS-BRD, USFS, and BLM with the support of the Rocky Mountain Elk Foundation to create a GIS database allowing analyses relating elk locations, movements, and home ranges to habitat components with the intent of developing a predictive habitat use model. During 1998, initial efforts were undertaken to develop a model with Dr. 1. Gross at the Natural Resource Ecology Laboratory at Colorado State University. �175 Elk habitat models are being developed by using a staged process. Major stages included initial data analysis and reduction, development of predictive variables (i.e., GIS coverages), and . development and statistical analysis of alternative habitat models. Initial data analysis included an extensive process to locate and correct errors in the data set, create new variables (e.g., age, class, season), and to subset the data so that analyses included only the animals of interest. This first stage of this process was accomplished by developing a database application in MicroSoft Access© that validated data records and produced new tables with the data needed for further analysis. The entire data set consisted of 5,321 elk locations. We reduced the data set to include only data from elk that were randomly selected and systematically located during years 1993-1998. This reduced total elk locations to 2,888. Locations were subset into groups by sex, age (calf, yearling, adult), and season. Seasons were biologically defined as spring (1 April- 30 May), summer (1 June - 31 August), fall (1 September - 30 November), and winter (1 December - 31 March). Using ArcInfo© and Ar~VieW©, we linked each location to GIS coverages for elevation, slope, aspect (EW and NS), distance to a road, and to tasselcap indices for brightness, greenness, and wetness. To evaluate the ability of these variables to predict elk observations, we used ordinary least squares (Upton and Fingleton 1985) to identify significant relationships arid coefficients that were used for logistic regression. RESULTS AND DISCUSSION Between 1 December 1993 and 14 June 1999,235 radioed elk died of which 31 were calves 611 months old and 204 were adults ~12 months old (Table 1, Appendix I). Calves died of natural causes (Freddy 1997) while for adults hunting accounted for 93% of 204 deaths. Hunting accounted for 98% of 102 deaths for adult males and 88% of 102 deaths for adult females. . Survival Estimates Adult Females Summer-Fall Survival during summer-fall 1998 for adult females all age 2:.24 months was 0.77 ± 0.07 (n = 136) inclusive of hunting deaths and 1.00 (n = 105) with hunting deaths censored (Table 6). Net survival in 1998 was the lowest measured during 5 summer-fall periods due to increased hunting deaths that resulted in a hunting removal rate of 23%. Reduced survival due to hunting in 1998 was evident in all cohorts offemales (Tables 2-5,7,8). Natural survival rates remained >0.99 during all 5 years (Table 6). Increased hunting removal rates were caused by a combination of increased numbers of limitedentry rifle antlerless permits and first-time use of unlimited over-the-counter rifle either-sex permits. Unlimited either-sex permits were valid in Game Management Units (GMUs) 43,52, and 521 during the second and third rifle seasons while limited antIerless permits were available for GMUs 42 and 421 for the first, second, and third rifle seasons. These GMUs collectively encompassed the area inhabited by radioed female elk. Approximate locations of each radioed female were obtained prior to each season, each elk was assigned to a GMU, and life/death status of each elk was known prior to and following each season. During the second rifle season, adult female removal rates were greater in areas having unlimited either-sex permits (26%) than in areas having limited antIerless permits (6%) (P=O.OOI, X2 =10.7, dJ=I). During third rifle season, removal rates were not different between types of permits (P=0.53, X2 =0.39, dJ=I)(Table 10). Removal rates with antlerless permits were consistent between seasons while removal rates with either-sex permits declined drastically during the third season (Table 10). In this geographic area, unlimited either-sex permits provided a burst of female removal only during the second season. �178 Prepared by _ David J. Freddy Life/Science Researcher Table l. Causes of deaths in radio-collared elk between 1 December 1993 and 14 June 1999. Calves (M=male, F=female) were age 6-11 months and collared at age 6 months. Yearling males and females .were age 12-17 months and juvenile males and females were age 18-23 months and all were collared at age 6 months. Adult males and females were age >24 months at time of death. Elk Age/Sex Class at Death Adult Total Cause of Death Calves Yearling Juvenile M F M All and -Code M F M F M F F Natural Causes 0 3 Malnutrition-6 2 0 0 0 2 5 2 0 Unknown-Suspect Malnutrition-31 0 0 3 1 3 1 0 0 0 0 4 Predation-Lion-3 0 0 1 8 5 13 8 4 0 0 0 0 Predation-Bear- 35 0 0 0 0 0 0 0 0 0 0 0 0 0 Unknown Predator-5 0 0 0 0 0 1 1 1 Unknown-Suspect 0 Predation-30 2 1 1 2 3 8 11 5 0 0 Accident-Birthing-32 0 0 0 2 0 2 0 0 0 0 2 Accident-Fell-IO 0 1 0 0 2 2 0 0 0 1 3 Unknown Cause-l l 2 0 0 2 4 1 0 0 1 2 6 Subtotals 10 26 45 14 Q Q Q £ £ 12 11 Le~1 Hunting Arc ery-33 0 0 0 0 9 3 9 5 0 2 14 ·4 Muzzleloading-34 0 0 0 0 8 2 8 12 0 2 0 0 0 0 0 1 0 1 0 0 1 Arch~lMuzzle-27 28 28 Rifle- irst-46 0 0 0 0 0 0 11 11 39 15 15 17 Rifle-Second-47 0 0 0 0 0 17 32 0 0 0 9 12 9 12 Rifle- Third-48 0 0 0 0 21 0 0 0 0 0 6 6 6 Rifle-Late-29 0 0 0 69 52 56 69 125 Subtotals Q Q Q Q Q 1. Wound~Loss 0 2 2 2 Archery uzzle-24 0 0 0 0 0 2 4 0 0 0 1 1 2 3 Archery-52 0 0 1 1 0 0 1 3 1 Rifle-First-43 0 0 0 0 3 4 Rifle-Second-a-t 0 0 0 0 0 4 10 4 10 14 0 2 2 Rifle- Third-45 0 0 0 0 0 1 1 3 0 0 r 1 Rifle-Unk.Reg.-25 0 0 0 0 0 0 0 1 0 0 0 Rifle-Late-26 0 0 0 0 0 3 3 3 10 20 Subtotals 32 Q Q Q Q 11 £1 1 1 Illegal Hunting 0 5 Rifle-First-49 0 5 0 0 0 0 5 0 0 6 Rifle-Second-50 0 5 0 0 I 0 0 6 0 0 -. 0 0 Rifle-Third-51 1 0 0 0 0 1 0 1 0 0 0 1 0 0 Rifle-Unk.Reg.-9 0 0 0 0 1 1 0 I 0 0 0 0 1 0 1 Archery Season-8 0 0 Out-of-Season-7 0 0 0 1 0 2 0 3 3 0 0 0 12" Subtotals }. Q Q Q 11 1 £ £ 11 Presumed Hunting" Missing-ArchMuzz-21 0 0 1 0 0 0 0 0 0 1 1 3 3 Missing-Rifle Ist-40 0 0 0 0 0 0 3 3 6 Missing-Rifle 2nd-41 0 0 0 2 6 2 7 0 1 0 9 Missing-Rifle 3rd-42 0 0 0 0 0 0 0 0 0 0 0 Subtotals 10 16 Q Q Q Q .2 Q .2 .2 1 Totals 17 14 13 6 2 3 87 93 119 116 235 • These elk were illegally shot as spike-antlered yearling males: (173.190/93) (173.232/93) (173.309/93) (173.320/93) (173.919194) (174.059/94) (174.140/95) (174.200/95) (174.679/95) (174.729195) (174.861195) (173.210/96). (173.309/93) disappeared in 1994 during first rifle season when spike-antlered yearling males were not legal and was assumed to have been taken illegally. All other illegal deaths were confirmed. b These elk disappeared during hunting seasons and remained missing for several months and are assumed dead and legally harvested: (172.207193) (172.269193) (172.308193) (172.498193) (172.649/93) (172.800/93) (172.961/93) (172.991193) (173.390/93) (173.439193) (173.760/94) (174.001/94) (174.181/94) (174.770195) (174.929/96) (175.199/96). �Table 2. Survival rates, inclusive ofnonhunting and hunting mortalities, for winter-spring (WS, 1 December-14 June) and summer-fall (SF, 15 June-30 November) time periods from 1 December 1993-14 June 1999 for the cohort of calves age 6 months radio-collared in December 1993. Survival rate for male and female calves pooled when age 6-11 months was 0.92 (95% CI 0.86-0.98, n=73). Survival rates (S) calculated as a mean estimate of (alive)/(alive + dead) and variance S(1-S)/n collars. Elk Elk Age (months) and Time Period (WS or SF dates) 6-11 (mos) 12-17 18-23 24-29 30-35 36-41 42-47 48-53 54-59 60-65 66-71 6-71 WS SF WS SF WS SF WS SF WS SF WS ALL 1993-94 1994 .1994-95 1995 1995-96 1996 1996-97 1997 1997-98 1998 1998-99 1993-99 0.88 0.78 0.99 32 0 4b 0 4 1.00 0.76 0.99 28 0 0 0 0 0.21 1.00 0.06 0.37 4 0 0 0 0 0.50 0.00 0.00 4 22 0 2 1.00 0.00 1.00 2r 0 0 0 0 0.00 0.97 0.87 1.00 35 0 1.00 0.92 1.00 34 0 0 0 0 0.97 0.91 1.00 33 0 1 0 1 0.84 0.91 1.00 32 0 5 0 5 1.00 0.71 0.97 27 0 0 0 0 0.89 MALES Survival 0.89 L 95%CI U 95%CI n collars 36 Censored" Died 4 Nonhunt 4 Hoot 0 28< 0 224 0 22 1 0 1 0 1 0.00 33 1& 3-'& 33 4 29 FEMALES Survival 0.95 L95%CI U95%CI n collars 37 Censored" Died 2 Nonhunt 2 Hoot 0 Ih 0 1 0.97 34 0 1 0 1 27 0 31 0 3 0.96 0.76 1.00 24 0 1 1 0 0.87 0.87 1.00 23 0 31 0 3 1.00 0.72 1.00 18 0 0 0 0 0.51 0.34 0.69 35 21t 17 '3 14 2 • Censored denotes collar failure and/or animallifeldeath status not known. b Collar (173.309/93) disappeared 1994 hunting seasons and assumed dead. e Includes (173.241193) collar failure August 1995 but seen January 1996. d Collars (173.390/93),(173.439/93) disappeared 1995 hunting seasons and assumed dead. • Censored elk; (173.241193) failed August 1995, (173.381193) slipped collar December 1995. r Includes (173.340/93) alive May 1997. I Censored elk (173.249/93) slipped collar September 1997. k Collar (172.800/93) disappeared 1994 hunting seasons and assumed dead. i Collar (172.961193) disappeared 1997 hunting seasons and assumed dead. j Collar (172.991193) disappeared 1998 hunting seasons and assumed dead. k Censored elk (172.849/93) failed May 1999, (173.151193) failed February 1999. ..... .....) \0 �Table 3. Survival rates, inclusive ofnonhunting and hunting mortalities, for winter-spring (WS, 1 December-14 June) and summer-fall (SF, 15 June-30 November) time periods from 1 December 1994-14 June 1999 for the cohort of calves age 6 months radio-collared in December 1994. Survival rate for male and female calves pooled when age 6-11 months was 0.90 (95% CI 0.83-0.97, n=69). Survival rates (S) calculated as a mean estimate of (alive)/(alive + dead) and variance S(1-S)/n collars. Elk Age (months) and Time Period (WS or SF dates) 6-11 (mos) 12-17 WS SF 1994-95 1995 0.91 0.81 1.00 33b 0 3 3 0 0.90 0.79 1.00 30b 0 .0.97 0.78 0.99 32 0 1 0 1 0.97 0.91 1.00 30 0 1 1 0 18-23 ./ 24-29 30-35 36-41 42-47 48-53 54-59 6-59 WS SF WS SF WS SF WS ALL 1995-96 1996 1996-97 1997 1997-98 1998 1997-98 1994-98 0.96 0.88 1.00 26 1c 1 1 0 0.08 0.00 0.19 25 0 23 0 23 1.00 1.00 0.00 0.00 2d 0 0 0 0 0.50 0.00 1.00 2 0 1 0 1 1 0 0 0 0 0 18 0 0 0 31 2 31 4 27 0.93 0.90 1.00 29 0.96 0.83 1.00 27 0 1 0 1 0.85 0.89 1.00 26 0 4 0 ·4 1.00 0.70 0.99 22 0 0 0 0 0.64 1.00 0.42 0.85 14 0 0 0 0 MALES Survival L 95%CI U 95%CI n collars Censored" Died Nonhunt Hunt 3 0 3 FEMALES Survival 0.89 L 95%CI U 95%CI n collars 36 Censored" Died 4 Nonhunt 4 Hunt 0 I" 2 0 2 • Censored denotes collar failure and/or animallifeldeath status not known. Includes (173.949194) collar failure December 1994 but seen January 1996 • Censored elk:(173.949194) collar failure, which was killed in fJISt rifle season 1996. d Includes (174.030/94) alive June 1997. • Censored elk: (173.719194) slipped collar May 1996. f Collar (173.760/94) disappeared 1998 hunting seasons and assumed dead. • Censored elk: (174.030/94) slipped collar September 1998. b 22 0 gf 0 8 0.40 0.23 0.57 35 0 21 5 16 •...• 00 0 �Table 4. Survival rates, inclusive ofnonhunting and hunting mortalities, for winter-spring (WS, 1 December-14 June) and summer-fall (SF, 15 June-30 November) time periods from 1 December 1995-14 June 1999 for the cohort of calves age 6 months radio-collared in December 1995. Survival rate for male and female calves pooled when age 6-11 months was 0.88 in 1995 (95% CI 0.81-0.96, n=69). Survival rates (S) calculated as a mean estimate of (alive)/(alive+dead) and variance of S(I-S)/n collars. Elk Age (months) and Time Period (WS or SF dates) 24-29 36-41 18-23 30-35 SF WS SF WS 1996-97 1997 1997-98 1998 6-11(mos} WS 1995-96 12-17 SF 19.96 MALES Survival L 95%CI U 95%CI n collars Censored" Died Nonhunt Hunt 0.86 0.74 0.98 35 2b 5 5 0 0.83 0.68 0.97 29 1c 5 0 5 0.96 0.87 1.00 24 0 1 1 0 0.23 0.04 0.41 22 4 Id I' 17 0 17 0 0 0 FEMALES Survival L 95%CI U 95%CI n collars Censored' Died Nonhunt Hunt 0.91 0.81 1.00 34 0 3 3 0 0.87 0.75 0.99 31 0 4 0 0.93 0.82 1.00 27 0 2 1 1 0.83 0.66 0.99 23 2f 4 0 4 1.00 4 1.00 18 I' 0 0 0 0.25 0.00 0.86 4 0 3 0 3 0.89 0.73 1.00 18 0 2 0 ·2 42-47 WS 1998-99 1.00 1 0 0 0 0 1.00 16 0 0 0 0 6-47 ALL 1995-99 0.03 0.00 0.10 32 5 31 6 25 0.52 0.33 0.70 .31 3 15 4 11 • Censored denotes collar failure and/or animal life/death status not known. Censored elk (174.619/95) slipped collar May 1996, (174.800/95) capture mortality. • Censored elk (174.660/95) slipped collar July 1996. d Censored elk (174.671195) likely collar failure July 1997. • Censored elk (174.190/95) slipped collar May 1998. r Censored elk (174.441195) slipped collar August 1997 and (174.580/95) collar failure July 1997. I Censored elk (174.470195) collar failure December 1997. . b .... 00 .... �Table 5. Survival rates, inclusive ofnonhunting and hunting mortalities, for winter-spring (WS, 1 December-14 June) and summer-fall (SF, 15 June-30 November) time periods from 1 December 1993-14 June 1999 for the cohort of calves age 6 months radio-collared in December 1996. Survival rate for male and female calves pooled when age 6-11 months was 0.86 (95% CI 0.81-0.96, n=69). Survival rates (S) calculated as a mean estimate of (alive)/(alive+dead) and variance ofS(I-S)/n collars. Elk Age (months) and Time Period (WS or SF dates) 6-11 (mos) 12-17 18-23 24-29 30-35 6-35 WS SF WS SF WS ALL 1996-97 1997 1997-98 1998 1998-99 1996-98 0.85 0.73 0.98 34 0.97 0.90 1.00 29 0 1 0 1 1.00 1.00 28 0 0 0 0 0.36 0.17 0.54 28 0 184 0 18 0.29 0.14 0.45 34 1 24 1.00 0.80 MALES Survival L 95%CI U 95%CI n collars Censored' Died Nonhunt HWlt Ib 5 5 0 10 0 0 0 0 5 19 FEMALES Survival 0.86 L 95%CI U 95%CI n collars 35 Censored' Died 5 Nonhunt 5 Hunt 0 1.00 0.74 0.98 30 0 0 0 0 30 0 0 0 0 30 0 6< 0 6 • Censored denotes collar failure and/or animallifeldeath status not known. Censored elk. (174.619/96) capture mortality. C Collar (174.929/96) disappeared 1998 hunting seasons and assumed dead. d Collar (175.199/96) disappeared 1998 hunting seasons and assumed dead. b 1.00 0.65 0.95 24 0 0 0 0 0.69 0.53 0.85 35 0 11 5 6 o •... 00 N �Table 6. Survival rates, inclusive ofnonhunting and hunting mortalities, for winter-spring (WS 1 December-14 June) and summer-fall (SF, 15 June30 November) time periods from 1 December 1993 - 14 June 1999 for all radio-collared adult female elk age 2:,12months. Survival rates (S) calculated as a mean estimate of (alive)/(alive + dead) and variance S(1-S)/n collars. Adult WS Adult SF Adult WS Elk Age and Time Period (WS or SF dates) Adult Adult Adult Adult Adult SF WS SF WS SF 1993-94 1994 1994-95 1995 1995-96 1996 1996-97 0.94 0.90 0.98 129bh 0 8d 0 8 0.94 0.90 0.98 119i 2) 7 4 3 0.89 0.84 0.94 1431: 0 16 1 15 0.98 0.95 1.00 1271 0 3 1 2 .1 Survival L 95%CI U 9S%CI n collars Censored" Died Nonhunt Hunt 0.96 0.91 1.00 68b• 0 0.87 0.80 0.94 100bf 0 3 13< 1 2 I 12 0.96 0.92 1.00 9Sbl 0 4 1 3 • Censored denotes collar failure or life/death status unknown. b Includes (172.011/93) collar failure April 1994, seen January 1996 , Collars (172.800/93,172.649/93) disappeared 1994 hunt seasons. d Collar (172.207/93) disappeared 1995 hunt seasons, assumed dead. • Composed of 6-18+ and 62-30+ months old females. f Composed of35-12+, 6-24+, and 59-36+ months old females. • Composed of34-18+ and 61-30+ months old females. b Composed of32-12+, 34-24+, and 63-36+ months old females. I Composed of30-18+ and 89-30+ months old females. J Censored elk (172.011/93) failure and (173.719/94) slipped collar. k Composed of31-12+, 29-24+, and 83-36+ months old females. Adult WS Adult SF Adult WS 1997 1997-98 1998 1998-99 0.91· 0.87 0.96 152m 2D 13 0 13 0.99 0.97 1.00 138· IP 2 1 1 0.77 0.70 0.84 136<1 0 31' 0 31 0.98 0.95 1.00 103' 21 2 2 0 I Composed of27-18+ and 100-30+ month old females . ••Composed of30-12+, 23-24+, and 99-36+ months old females. • Censored elk (174.441/95) slipped collar August 1997 and (174.S80/95) likely failure June 1997. • Composed of30-18+ and 108-30+ months old females . P Censored elk (174.470/95) likely collar failure August 1997. q Composed of30-24+ and 106-36+ months old females. r Collars (172.269/93,172.308/93,172.498/93,172.991/93, 173,760/94, 174.929/96) disappeared 1998 hunt seasons and assumed dead. • Composed of24-30+ and 81-42+ months old females. I Censored elk (172.849/93, 173.151/93) failed collars. ,_. OQ W �186 APPENDIX I. Mortalities of235 radio-collared elk from 1 December 1993 through 14 June 1999. Age is aQQfoximate age in ~ears of elk at death: C=calf age 6-11 monthsz Y=yearling age 12-23 months. Frequency ID! Death TraQ Year caEtured Site Zone Sex Age Date Cause of death & code number 172.030/93 172.050/93 172.039/93 172.070/93 172.080/93 172.090/93 172.090/94 172.101193 172.128/93 172.139/93 172.160/93 172.181193 172.201193 172.207/93 172.229/93 172.258/93 172.269/93 172.277/93 172.290/93 172.290/94 172.300193 172.308/93 172.369/93 172.369/94 172.409/93 172.448/93 172.459/93 172.498/93 172.509/93 172.519/93 172.530/93 172.542/93 172.542/94 172.549/93 172.570/93 172.581193 172.581194 172.590/93 172.610/93 172.620/93 172.639/93 172.649/93 172.658/93 172.670/93 172.678/93 172.690/93 172.699/93 172.730/95 172.749/95 172.758/95 172.800/93 172.821/93 172.890/93 172.899/93 172.950/93 172.961193 172.991193 173.000/93 BR EG GR BR GM GR SM SR GC BC CC MC GS SG MC HY MC GS SG SM AC GH AC FM AC ·MH MH FS FS FS FS FS FM PG PG PG FM PG PG PG PG PG PG PG PG FM FM LS LS LS SR EG BC BC GH MG SG WM A B B A A B D B B C C C C D C C C C D D E D E B E E E H H H H H B H H H B H H H H H H H H B B H H H B B C C D D D G F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F F 5 12 4 11 6 11 11 8 8 17 3 3 3 4 8 3 12 3 6 4 8 8 3 4 8 12 8 12 3 6 8 6 15 7 16 3 3 5 3 5 16 2 5 4 9 8 7 19 11 15 Y 3 3 C 2 4 5 C 27-Oct-95 31-Oct-98 14-Jun-95 12-Feb-96 05-Nov-94 16-Jan-94 12-Oct-98 14-Oct-96 31-Oct-96 19-Oct-95 23-Oct-94 03-Nov-94 15-Nov-94 30-Nov-95 29-Oct-98 04-0ct-94 29-Oct-98 O4-Oct-94 23-Oct-94 16-Sep-96 30-0ct-97 29-Oct-98 18~94 14-Jan-96 29-Dec-94 . 01-Nov-98 24-Oct-96 15-Oct-98 21-Oct-95 12-Oct-98 29-Oct-98 01-Feb-94 06-Jan-99 28-Nov-95 03-Nov-94 17-Oct-94 08-Dec-95 i4-Apr-96 O4-Oct-94 04-Nov-98 14-Jun-96 30-Nov-94 02-Nov-97 16-Jan-95 22-Dec-94 24-Jan-94 05-Nov-95 05-May-99 07-Aug-96 28-Oct-98 30-Nov-94 08-Sep-96 06-Nov-96 18-Mar-94 18-Oct-95 30-Oct-97 29-Oct-98 25-Apr-94 Legal kill second rifle season-47 Legal kill third rifle season-48 Unknown-suspect predation-30 Unknown-l l Legal kill third rifle ~n-48 Wounding loss late rifle season-26 Legal kill first rifle season-46 Legal kill first rifle season-46 Wounding loss second rifle season-44 Wounding loss first rifle season-43 Legal kill second rifle season-47 Wounding loss second rifle season-44 Wounding loss second rifle season-44 Disappear first rifle season-40 Legal kill second rifle season-47 Legal kill archerylmuzzle season-27 Disappear second rifle season-41 Wounding loss archery/muzzle-24 Legal kill second rifle season-47 Legal kill muzzleloading season-34 Wounding loss second rifle season-44 Disappear second rifle season-41 Legal kill archery season-33 Legal kill late rifle season-29 Wounding loss late rifle season-26 Legal kill third rifle season-48 Legal kill second rifle season-47 Disappear [JISt rifle season-40 Legal kill second rifle season-47 Legal kill first rifle season-46 Wounding loss second rifle season-44 Mountain lion predation-3 Accident-irrigation ditch-If Legal kill late rifle season-29 Wounding loss second rifle season-44 Legal kill first rifle season-46 Legal kill late rifle season-29 Unknown-11 Accident. birthinglca1ving-32 Legal kill third rifle season-48 Accident. birthinglca1ving-32 Disappear first rifle season-40 Legal kill third rifle season-48 Legal kill late rifle season-29 Wounding loss late rifle season-26 megal kill-7 Legal kill third rifle season-48 Malnutrition-6 Unknown-suspect predation-30 Wounding loss second rifle season-44 Disappear second rifle season-41 Legal kill archery season-33 Legal kill third rifle season-48 Mountain lion predation-3 Legal kill first rifle season-46 Disappear second rifle season-41 Disappear second rifle season-41 . MaInutrition-6 --------------------------------------------------------------------------------------------- �187 Appendix I. (continued) Frequency IDI Trap Year captured Site Zone 173.000/94 173.010/93 173.041193 173.050/93 173.060/93 173.070/93 173.081193 173.090/93 173.100/93 173.120/93 173.140/93 173.160/93 173.171193 173.190/93 173.190/96 173.201193 173.210/93 173.210/96 173.219/93 173.219/96 . 173.232/93 173.232/96 173.262193 173.262/94 173.269/93 . .173.279/93 173.289/93 173.289/94 173.300/93 173.300/96 173.309/93 173.320/93 173.320/94 173.332/93 173.332/96 173.340/93 173.351193 173.351196 173.359/93 173.359/96 173.370/93 173.370/96 173.381196 173.390/93 173.402/93 173.410/93 173.410/96 173.420/93 173.429/93 173.439/93 173.450/93 173.461/93 173.461194 173.461196 173.469/93 173.469/94 173.479/93 173.479/96 173.492/93 HM WM GM MH MH MH FS FS FS GM PG FS FM BC BC GR GC LM BC CG BR BC BC HM MC MC MC PR GS MC HY HY HM GH BC SG SG BG AC MC MH RG RG MG MG WM EW WM MH PG PG PG CM GM FS PR FS RG FS B G A E E E H H H A H H B C C B B D C D AMY C C B C C C A C C C C B D C D D E E C E E E D D G G G E H H H AMY A H A H E H Death Sex· Age F F M F F F F F F M F F F M M M M M M M 4 4 2 5 2 5 3 4 4 2 3 3 5 Y 2 2 2 2 2 2 M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M 2 C 2 2 3 C 2 2 2 Y Y 2 2 2 4 2 2 2 2 2 C C 2 2 2 1 2 2 2 2 C M M M M M M 2 C 2 2 2 2 Date 31-Oct-98 07-Jan-98 19-Oct-95 31-Oct-98 27-Dec-95 03-Mar-98 22-Oct-96 24-Oct-97 30-0ct-97 14-Oct-95 31-Oct-96 13-Nov-96 10-Oct-98 29-Oct-94 1O-Oct-98 30-Nov-95 19-Oct-95 30-Oct-97 06-Nov-95 21-Oct-98 15-Nov-94 31-Oct-98 18-Mar-94 12-Oct-96 06-Nov-95 12-Oct-96 22-Mar-94 18-Sep-96 24-Oct-95 10-Oct-98 30-Nov-94 20-Oct-94 14-Sep-96 12-Sep-95 10-Oct-98 20-Oct-97 05-Nov-95 27-Oct-98 06-Sep-95 09-Nov-98 08-Nov-95 08-Jan-97 19-Feb-97 30-Nov-95 18-Oct-95 02-Nov-95 29-Sep-98 24-Oct-95 14-Sep-95 30-Nov-95 15-Oct-95 07-Feb-94 19-5ep-95 22-Oct-98 25-Apr-94 20-Oct-96 1O-Sep-95 l7-Oct-98 03-Sep-95 Cause of death & code number Legalkill third rifle season-48 Accident-fell-l0 Legalkill first rifle season-46 Legalkill third rifle season-48 Legalkill late rifle season-29 Illegalkill-7 Legalkill secondrifle season-47 Legalkill secondrifle season-47 Legalkill secondrifle season-47 Legalkill first rifle season-46 WOWldinglosssecondrifleseason-44 Legalkill third rifle season-48 Legalkill first rifle season-46 megal kill secondrifle season-50 Legalkill first rifle season-46 Woundingloss third rifle season-45 Legalkill first rifle season-46 Illegalkill archery/muzz.season-8 Legalkill third rifle season-48 Legalkill secondrifle season-47 illegal kill secondrifle season-50 Legalkill third rifle season-48 Unknown-ll Legalkill first rifle season-46 Legalkill third rifle season-48 Legalkill first rifle season-46 Unknown-ll Legalkill muzzleloadingseason-34 Legalkill secondrifle season-47 Legalkill first rifle season-46 Illegalkill first rifle season-49 megal kill first rifle season-49 Legalkill archeryseason-33 Legalkill muzzleloadingseason-34 Legalkill first rifle season-46 Legalkill secondrifle season-47 Legalkill third rifle season-48 Legalkill secondrifle season-47 Legalkill archeryseason-33 Woundingloss third rifle season-45 Legalkill first rifle season-46 Mountainlion predation-3 Unknown-suspectpredation-30 Disappearfirst rifle season-40 Legalkill first rifle season-46 Woundingloss secondrifle season-44 Woundingloss archery/muzzleseasons-24 Legalkill secondrifle season-47 Legalkill archeryseason-33 Disappearfirst rifle season-40 Legalkill first rifle season-46 Malnutrition-6 Woundingloss archeryseason-50 Legalkill secondrifle season-47 Malnutrition-6 Legalkill secondrifle season-47 Legalkill archeryseason-33 Legalkill secondrifle season-47 Legalkill archeryseason-33 !~~~~~~]--------~~-----~-------~----]----7--]]~~~!:?~-~g~!~~E:~~~~:_~~~~ _ �188 Appendix I. (continued) Frequency ID! Year Captured Trap. Site Zone Death Sex Age 173.510/93 FM FM FM M M F M F F F F F F F F 2 2 7 2 Y 3 Y C 3 C C 3 F F F F F F F F F F F F F M M M M M M M M M M M M M M M M M M 4 2 2 4 4 4 C 3 2 4 4 2 C Y 2 2 2 2 2 2 2 2 2 2 2 Y 2 3 2 Y 2 M M 2 2 M M M M M M M M Y 2 2 C 2 2 Y 2 173.521193 173.530/94 173.540/94 173.549/94 173.569/94 173.580/94 173.589/94 173.610/94 173.640/94 173.640/96 173.649/94 173.659/94 173.679/94 173.689/94 173.719/96 173.739/94 173.750/94 173.760194 173.789/94 173.819/94 173.829/94 173.840/94 173.849194 173.859/94 173.870/94 173.919/94 173.929/94 173.939/94 173.959/94 173.970/94 173.981/94 173.990/94 174.001194 174.009194 174.019/94 174.039/94 174.049194 174.059/94 174.069/94 174.080/94 174.090/94 174.101/94 174.109/94 174.119/94 174.119/95 174.129/94 174.140194 174.140/95 174.150/94 174.160/94 174.170/94 174.170/96 174.181194 174.200/95 174.210/95 174.220/95 174.230/95 174.230/96 00 HM PR PR 00 BC MC GM BC MC MP BG MC BG BG MR MR MM MM MM SM SM SM BC BC BC BC MC MP BC BG BG BG MR MM MR BG MR MR MM MM MM GB MM MM GB MM MM FM MC FM MS MS MS MS EW ~~~~~~2~ ~ B B B B B A A B C C A C CF4 D E C E E E E F F F D D D C C C C C D C E E E E F E E E E F F FMC B F FMC B F F B C B F F FMC FMC G M ~ ~ Date Cause of Death & Code Number 2 28-Aug-95 17-Oct-95 31-Oct-98 13-Oct-% 07-Sep-95 l6-Oct-97 21-Dec-95 01-Jun-95 14-Nov-97 30-Mar-95 16-Mar-97 1l-Oct-97 25-Oct-98 29-Oct-98 31-Oct-% 03-Nov-98 29-Oct-98 18-Oct-98 29-Oct-98 20-Mar-95 30-Oct-97 IO-Jan-97 29-Sep-98 20-Oct-98 23-Oct-96 21-Feb-95 02-Nov-95 l6-Oct-96 18-Sep-96 02-Nov-96 20-Oct-96 02-Nov-96 20-Oct-96 30-Nov-% 12-Oct-96 13-Nov-96 05-Sep-96 14-Sep-96 17-Oct-95 26-Sep-96 ll-Oct-97 20-Oct-96 24-May-96 19-Oct-96 01-Mar-95 30-Nov-97 14-Oct-96 14-Mar-95 30-Nov-96 14-Oct-96 15-Sep-96 25-Apr-95 10-Oct-98 30-Nov-96 18-Oct-96 ll-Oct-97 08-Apr-96 24-Apr-96 l3-Oct-98 Legal kill archeryseason-33 Legal kill first rifle season-46 Legal kill third rifle season-48 Legal kill first rifle season-46 Legal kill archeryseason-33 Woundingloss first rifle season-43 Unknown-ll Unknown-suspectpredation-30 Woundingloss third rifle season-45 Unknown-suspectpredation-30 Malnutrition-6 Legal kill first rifle season-46 Legal kill secondrifle season-47 Woundingloss first rifle season-43 Legalkillfrrstrifleseason-46 Legal kill third rifle season-48 Woundingloss secondrifle season-44 . Legal kill secondrifle season-47 Disappear secondrifle season-41 Unknown-suspectmalnutrition-31 Legal kill secondrifle season-47 Legal kill late rifle season-29 Woundingloss archery/muzzleseasons-24 Legal kill secondrifle season-47 Legal kill secondrifle season-47 Mountain lion predation-3 illegal kill secondrifle season-50 Legal kill first rifle season-46 Legal kill archeryseason-33 Legal kill third rifle season-48 Legal kill secondrifle season-47 Legal kill third rifle season-48 Legal kill secondrifle season-47 Disappeararchery/muzzleseason-21 Legal kill first rifle season-47 illegaI kill rifle season-9 Legal kill archeryseason-33 Legal kill muzzleloadingseason-34 illegal kill first rifle season-49 Woundingloss archery/muzzle-24 Legal kill first rifle season-46 Legal kill secondrifle season-47 Unknown-suspectpredation-30 Legal kill secondrifle season-47 Mountain lion predation-3 Woundingloss rifle season-25 Legal kill first rifle season-46 Mountainlionpredation-3 llIegalkillthirdrifleseason-51 Legal kill first rifle season-46 Legal kill muzzleloadingseason-34 Mountain lion predation-3 Legal kill first rifle season-46 Disappear secondrifle season-41 llIegaIkill first rifle season-49 Legal kill first rifle season-46 Unknown-suspectmalnutrition-31 Mountain lion predation-3 Legal kill first rifle season-46 ~ ~I:~~~~~ ~~~~~~~~P!~~~:~ _ �189 AEEendix I. {continued} Frequency ID! Tral! Site Zone Year caEtured 174.240/96 SR B MD 174.301195 E MS 174.319/95 F MD 174.329/95 E MD 174.339/95 E 174.349/95 MD E 174.360/95 MD E 174.401/95 AP D 174.420/95 AP D 174.478/95 WD e 174.491195 LB e 174.500/95 LB e 174.520/95 KR B 174.520/96 GM A 174.551/95 we G 174.560/95 BL A 174.609/95 GB B MD 174.629/95 E 174.639/95 MD E 174.650/95 HG E 174.679/95 HG E 174.689/95 AP D 174.689/96 EW G 174.700/95 AP D 174.710/95 AP D 174.729/95 VM D 174.740/95 VM D 174.750/95 VM D 174.760/95 VM D 174.770/95 WD e 174.780/95 WD e 174.789/95 WD e 174.789/96 EW G 174.800/96 BG E WD 174.809/95 e 174.820/95 WD e 174.830/95 LB e 174.841195 LB e 174.851195 LB e 174.861195 KR B 174.870/95 LB e 174.880/95 KR B 174.889/95 we G 174.899/95 BL A 174.910/96 LM D 174.929/96 LM D 174.959/96 LM D 174.998/96 BG E 175.059/96 Be e 175.130196 GM A 175.139/96 GM A 175.149/96 SR B 175.160/96 Be e 175.170196 LM D 175.189/96 RG E EW 175.199/96 G 175.209/96 MS F END Death Sex Age M 2 F 3 F Y F Y F e 2 F F Y F 2 F 2 2 F 2 F e F F Y F e 3 F F Y e F 2 M M 2 M 3 M Y e M e M M 2 M 2 M Y M 2 M 2 M 3 M 2 2 M e M 2 M M e y M 2 M M 2 3 M M 2 Y M M 2 2 M 2 M 2 M F e 2 F F 2 F 2 F e F e 2 F 2 F 2 M e M 2 M M 2 2 M Date 12-Sep-98 20-Sep-98 02-Oct-% 03-Sep-96 27-Mar-96 l3-Oct-97 22-Apr-97 22-Sep-% 1l-Oct-97 13-Sep-97 10-Sep-97 16-Apr-96 21-Sep-96 25-Apr-97 14-Oct-98 14-May-97 15-Apr-96 11-Oct-97 21-Oct-97 14-Sep-98 13-Nov-96 05-Feb-96 14-May-97 16-Oct-97 15-Oct-97 17-Oct-% 08-Nov-97 05-Nov-97 15-Sep-98 16-Oct-97 II-Oct-97 22-May-96 28-Oct-98 09-Apr-97 22-Apr-97 14-Oct-97 l3-Oct-97 12-Sep-98 22-Oct-97 31-Oct-96 30-Oct-97 11-Oct-97 20-Oct-97 30-Oct-97 27-Feb-97 29-Oct-98 29-Oct-98 21-Oct-98 13-Feb-97 24-Mar-97 28-Oct-98 19-Oct-98 17-Sep-98 26-Mar-97 l3-Oct-98 29-Oct-98 24-Oct-98 Cause of death & code number Legal kill muzzleloading season-34 Legal kill archery season-33 Wounding loss archery season-52 Legal kill archery season-33 Unknown predator-S Legal kill first rifle Season-46 Unknown-suspect predation-30 Legal kill muzzleloading season-34 Legal kill first rifle season-46 Legal kill muzzleloading season-34 Wounding loss archery season-52 Unknown-suspect predation-30 Legal kill muzzleloading season-34 Mountain lion predation-3 Legal kill first rifle season-46 Illegal kill-7 Unknown-suspect predation-30 Legal kill first rifle season-46 Legal kill second rifle season-47 Legal kill muzzleloading season-34 Illegal kill second rifle season-50 .Mountain lion predation-3 Unknown-suspect maInutrition-31 Wounding loss first rifle season-43 Legal kill first rifle season-46 Illegal kill first rifle season-49 Legal kill third rifle season-48 Legal kill third rifle season-48 Legal kill archery season-33 Disappear first rifle season-40 Legal kill first rifle season-46 Unknown-suspect predation-30 Wounding loss second rifle season-44 Mountain lion predation-S Accident, fell-lO Legal kill first rifle season-46 Legal kill first rifle season-46 Legal kill muzzleloading season-34 Illegal kill second rifle season-50 Illegal kill second rifle season-50 Wounding loss second rifle season-44 Legal kill first rifle season-46 Legal kill second rifle season-47 Wounding loss second rifle season-44 Unknown-suspect predation-30 Disappear second rifle season-41 Wounding loss second rifle season-44 Legal kill second rifle season-47 Unknown-l l Mountain lion predation-3 Legal kill second rifle season-47 Legal kill second rifle season-47 Legal kill muzzleloading season-34 Unknown-suspect maInutrition-31 Legal kill first rifle season-46 Disappear second rifle season-41 Legal kill second rifle season-47 �190 Appendix Il, Abstract from paper presented at Western States and Provinces Elk and Mule Deer Workshop, Salt Lake City, Utah, March 1999. ESTIMATING ELK DENSITIES IN COLORADO: PROGRESS WITH PERPLEXITIES DAVID J. FREDDY, Colorado Division of Wildlife, Research Center, 317 West Prospect Road, Fort Collins, CO, USA 80526 DAVID C. BOWDEN, Department of Statistics, Colorado State University, Fort Collins, CO 80523 USA GARY C. WIllTE, Department of Fishery and Wildlife Biology, Colorado State Univeristy, Fort Collins, CO 80523 USA Abstract: During winters 1994-1998, we evaluated helicopter survey methodologies to estimate elk densities injuniper-pinyon (Juniperus osteosperma-Pinus edulis) and oakbrush-mountain shrub (Quercus gambelii-Amelanchier alnifolia) habitats. We developed sighting bias correction models to correct for groups of elk not observed when counting elk on 2.59-km2 (l-mf) quadrats. Within a 350-km2 area, we compared elk densities obtained from a stratified random quadrat sampling system corrected for sighting bias with independent estimates of densities obtained from mark-resight models using known numbers of radio-collared elk (120-137 elk) within the quadrat survey area. Although observers detected about 80% of the elk groups on quadrats during sighting bias trials and resulting sighting bias models increased estimated elk densities on quadrats by 10-12%, quadrat densities were 2:,25%lower than mark-resight densities. Estimated densities of elk ranged from 6-10 elk/krrr', We propose that errors in counting numbers of elk in a group may contribute more to underestimating elk densities than errors in detecting groups of elk. �239 Colorado Division of Wildlife Wildlife Research Report July 2000 JOB FINAL REPORT Smreof ~C~m~o~rnrl~o~ _ Cost Center 3430 Project No. W~-1=5=3__=-R:..:.-"""'1=_3 _ Manunals Research Work Package -=3...:::;0=02=-- _ Task No. --=2'-- _ Elk Management Estimating Survival Rates of Elk and Developing Techniques to Estimate Population Size Period Covered: July 1, 1999 - June 30,2000 Author: D. J. Freddy \ ) Personnel: T. Beck, J. Broderick, G. Byrne, D. Crane, R Del Piccolo, J. Ellenberger, V. Graham, D. Homan, R. Kahn, K. Madariaga, D. Masden, C. Mehaffy, P. Neil, J. alterman, P. Will, K. Wright, S. Yamashita, D. Younkin, CDOW; D. Bowden, M. Conner, M. Kneeland, G. White, CSU; D. Ouren, D. Schneider, T. Fancher, R. Waltermeir, USGS-BRD, BLM Glenwood Springs, CO, USFS Rifle, CO, Rocky Mountain Elk Foundation, cooperating. ABSTRACT We monitored survival and movements of 112 previously radio-collared elk composed of 101 adult females and 11 adult males from 1 July 1999 through 30 June 2000. Survival of adult females during summer-fall was 0.89 ± 0.06 (n = 101) inclusive of hunting deaths and 1.00 (n = 90) with hunting deaths censored while survival during winter-spring was 0.98 ± 0.03 (n = 88) inclusive of hunting "deaths and 0.99 ± 0.02 with hunting deaths censored. The 1 non-hunting death of an adult female during winter-spring was attributed . to mountain lion predation. Survival of adult males during summer-fall was 0.27 ± 0.30 (n = 11) inclusive of hunting deaths and 1.00 (n = 3) with hunting deaths censored while survival during winter-spring was 1.00 (n = 3); With hunting deaths censored, annual survival rates during winter-spring were 2: 0.97 for adult females and 2: 0.96 for adult males during 7 consecutive winters while annual survival rates during summer-full were 2: 0.99 for both adult females and males during 6 consecutive summers. High adult survival and high over-winter calf survival (0.89 ± 0.04) allow this population to potentially increase> 15% annually in the absence of hunting antlerless elk. Hunting related mortalities were the primary cause of death for adult male and female elk. During 6 years, 98% of all adult male deaths and 89% of all adult female deaths were related to hunting. Hunting annually removed 76% of the adult branch-antlered males and 13% of the adult females. Annual wounding losses averaged 15% for adult males and 30% for adult females while 11% of yearling spike-antlered males were illegally harvested. Sighting bias corrections for counting elk on l-mf quadrats were developed and evaluated as a method to estimate elk densities. We found counts of elk on quadrats corrected for sighting bias associated with detection of elk groups .---.---ro rnvwLDL11i1llrilil~~llljilrnl BDOW016789 �240 continued to under-estimate true numbers of elk present. Evidence indicates under-counting numbers of elk in groups also contributes to underestimating true numbers of elk present. If a quadrat sampling system is employed to estimate densities of elk, counts of elk would need to be increased by 1.3x to approximate true numbers of elk present. Progress continues towards developing habitat use models to predict areas important to managing elk Several manuscripts on elk survival rates, techniques to estimate population size, and habitat use are progressing. �241 ESTIMATING SURVIVAL JOB FINAL REPORT RATES OF ELK AND DEVELOPING ESTIMATE POPULATION SIZE TECHNIQUES TO David J. Freddy P. N. OBJECTIVE Estimate survival rates of adult female, adult male, and calf elk and develop techniques to estimate population size. SEGMENT OBJECTIVES 1. Analyze data concerning estimates of elk survival rates and experiments to estimate elk numbers. 2. Prepare Job Final Report 3. Prepare a manuscript for submission to a scientific journal which describes and summarizes research results. INTRODUCTION Our objectives are to provide reliable estimates of survival rates for calves during winter and for adult females and males throughout the year and to develop and test a system to estimate elk densities on winter . ranges. Estimates of calf survival were obtained during winters 1993-94 through 1996-97 and summarized in Freddy (1997). We continue to obtain estimates of survival for adult males and females. For adults, we are interested in survival rates inclusive of hunting mortalities to document human-induced rates of survival and exclusive of hunting mortalities to estimate natural rates of survival. We are evaluating a system for estimating population size or density that incorporates estimates of sighting bias in conjunction with random sampling of search quadrats as sample units. Our winter study area encompasses about 324 mi2 (839 knr') in the eastern half of Game Management Unit 42 south and east of Rifle, Colorado. This area is part of the Grand Mesa Elk: Data Analysis Unit E-14. Elk winter habitats include juniper-pinyon woodland (Juniperus osteosperma-Pinus edulis), oakbrush-mountain shrub (Quercus gambelii-Amelanchier alnifolia), aspen (Populus tremuloides), sagebrush (Artemisia tridentata), and agricultural fields below elevations of 9,800 ft. In summer, elk:use oakbrush-mountain shrub, aspen, and subalpine fir-Engelmann spruce (Abies lasiocarpa-Picea engelmannii) meadow systems throughout the Grand Mesa region up to elevations of 12,000 ft. (Freddy 1993). METHODS We placed radio collars (l72-176MHz) having mortality sensors on elk calves age 6 months and adults age :::18 months during December 1993-1996 (Freddy 1997). Elk: were captured using helicopter net-gunning and portable corral traps (Freddy 1997). During 1999-00, we monitored 112 adult elk, 101 females and 11 males, having functioning radios as of July 1999. Collars were white and had black identification symbols on dorsal surfaces of collars to allow visual identification of individual elk. Estimating Elk Survival Rates During 1999-2000, we monitored survival of radioed elk with limited aerial surveys using a Cessna 185 and with opportunistic ground surveys. Extreme reductions in operating budgets during 1999-2000 �242 prevented following survey protocols used in previous years. Survival rates (S) of radioed elk were calculated using the binomial estimator with a variance, VAR(S) = S( I-S)/n collars (White and Garrott 1990). Survival rates are expressed as the mean estimate ± the 95% confidence interval. Sample sizes refer to numbers of individual radioed elk. Elk were censored if radios failed or if their life/death status was unknown after an extended period of time We defined 3 major time intervals for survival analyses: winter-spring was 1 December to 14 June, summer-fall was 15 June to 30 November, and annual was 1 December to 30 November to coincide with timing of capture and collaring. For adult elk during these time intervals, we calculated survival rates inclusive of natural and hunting-related mortalities, exclusive of hunting mortalities, and exclusive of natural mortalities. Excluding, or censoring hunting mortalities, provided estimates of natural survival rates. Excluding natural mortalities but including hunting mortalities provided estimates of hunting removal rates. Censoring elk associated with categories of mortalities reduced sample sizes used to estimate survival rates .. Archery, muzzleloading, and rifle hunting seasons occurred from about 1 September to 15 November each year during the summer-fall time interval. Late-rifle seasons occurred in some years and restricted hunters to taking only antlerless elk from about 25 November-31 January which included portions of the summerfall and winter-spring time intervals. Yearling spike-antlered males were generally not legal quarry in most areas frequented by radioed elk. Male elk were legal quarry when branch-antlered usually at age 2 years. Cause of death of radioed elk recovered in the field was estimated from presence or absence of gunshot wounds or bite wounds on carcass, predator tracks or scat at carcass site, physical positioning of carcass remains whether buried, covered, scattered, or consolidated (Wade and Browns 1982, Halfpenny and Biesiot 1986), relative amount of internal fat and marrow fat if present with carcass, and results of histopathology and marrow fat analyses. Fat content (percent dry matter) of bone marrow and estimates of age based on dental cementum were obtained for dead elk by the Colorado Division of Wildlife Laboratory while histopathology analyses were provided by the Colorado State University Veterinary Diagnostic Laboratory. Photographs were taken of nearly all mortalities so that physical evidence could be reviewed and judged by outside experts. Population Estimation Development of sighting bias corrections (Samuel et al. 1987) for aerial quadrat sampling systems to estimate elk densities terminated after winter 1998. We have continued to analyze existing data to evaluate and sources of negative bias associated with sighting.bias corrected counts of elk on quadrat sample units. Elk Movements and Habitat Use Models Aerial surveys to precisely locate radioed elk were terminated after June 1999 and development of habitat use predictive models were hampered by reduced budgets in 1999-2000. However, we continued an extensive process of database refinement and reduction. Using ArcInfo© and ArcVieW©, we linked all elk locations to GIS coverages for elevation, slope, aspect (EW and NS), land ownership, private land density, road density, hydrologic density, distance to private land, distance to roads, and distance to water. The entire data set consisted of 5,164 elk locations representing 365 individual elk. For developing predictive use models, we will use only data from elk that were randomly selected and systematically located during years 1993-1999 which represented 2,900 locations representing 140 individual elk. All locations were subset into groups by sex, age, (calf, yearling, adult) and season. Seasons were biologically defined as spring (1 April- 30 May), summer (1 June- 31 August), fall (1 September - 30 November) and winter (1 December - 31 March). �243 We also embarked on developing a supervised habitat type GIS coverage using ERDAS Imagine© and a thematic mapper image from June 1993. We had anticipated obtaining this data from another source during the last 3 years but such information did not become available. RESUL TS AND DISCUSSION Between 1 December 1993 and 14 June 2000, 256 radioed elk died of which 31 were calves 6-11 months old and 225 were adults ~12 months old (fable 1, Appendix I). Calves died of natural causes (Freddy 1997) while hunting accounted for 93% of 225 adult deaths. Hunting caused 98% of 110 adult male deaths and 89% of 115 adult female deaths. Survival Estimates Adult Females Summer-Fall Survival during summer-fall 1999 for adult females all age ~24 months was 0.89 ± 0.06 (n = 101) inclusive of hunting deaths and 1.00 (n = 90) with hunting deaths censored (fable 6). Survival rates for adult females during summer-fall with hunting deaths censored remained >0.99 during 6 consecutive years, 1994-1999 (fables 6-8) .. Net summer-fall survival increased to 0.89 in 1999 from a low of 0.77 in 1998 (fable 6). An increase in net survival was evident in nearly all cohorts of females (fables 2-8). Increased survival likely reflected the return to using traditional limited antlerless permits during the 1999 hunting seasons compared to using unlimited either-sex permits during portions of 1998 hunting seasons (Freddy 1999). Adult Females Winter-Spring Survival during winter-spring 1999-00 for adult females all age ~24 months was 0.98 ± 0.03 (n = 88) inclusive of hunting deaths and 0.99 ± 0.02 (n = 87) with hunting deaths censored (fable 6). The 1 natural death involved a female age 6 years that was killed by a mountain lion. Survival rates for adult females during winter-spring with hunting deaths censored remained ~0.97 for 7 consecutive winters, 1993-941999-00 (fables 6-8). Adult Females-Annual Annual survival from 1 December 1998 to November 30 1999 was 0.83 ± 0.16 (n = 24) inclusive of hunting deaths and 1.00 (n = 20) with hunting deaths censored for the 1993 cohort of radioed older adult females (fable 9). Annual survival among years for this cohort ranged from 0.71 to 0.94 and from 0.92 to 1.00, inclusive and exclusive of hunting mortalities, respectively. Adult Males Summer-Fall Survival during summer-fall 1999 for adult males all age ~36 months was 0.27 ± 0.30 (n = 11) inclusive of hunting deaths and 1.00 (n = 3) with hunting deaths censored (fable 11). With the unlimited branchantlered male hunting system used in the project area, the probability of a male living from age 6-months to 48-months was 5% (7 of 138 males, Tables 2-5). Adult Males Winter-Spring Survival during winter-spring 1999-00 for adult males all age ~42 months was 1.00 (n = 3)(fable 5). There have been only 2 non-hunting deaths of males age ~12-months during 6 years and both occurred during winter-spring and were attributed to an accidental fall and predation (fable 1). Wounding Loss Wounding loss during archery, muzzleloading, and rifle seasons in 1999 was 33% for adult males and 33% for adult females. This compares to 1994-99 averages of 15% for adult males and 30% for adult females. Wounding loss is defined as wounded / legal kill (Freddy 1997). �244 Removal Rates During the summer-fall period, hunting removal rates (1 - survival rate) averaged 0.76 for adult branchantlered males and 0.13 for adult females (Tables 10, 11). Peak removal rates of 0.92 for 24 month-old males and 0.86 for all adult males occurred in 1996 while minimum removal rates of 0.64 and 0.65 for these groups of males occurred in 1998. Peak removal rate of 0.23 occurred for adult females in 1998. Removal rates in 1998 were associated with unlimited numbers of either-sex hunting permits during portions of the 1998 seasons. These permits appeared to reduce hunting mortality on males and increase hunting mortality on females. During those years when cohorts of yearling (12-17 months old) and adult females (>24 months old) were available, removal rates for yearling females tended to be lower than adult females (Table 11). Radio Collar Status As of July 2000, 89 adult elk (3 males, 86 females) were alive with functioning radio collars and remained in the project area. Many of these collars should be subject to electronic failure soon. Quality radio collars (Lotek, Inc) with very low electronic failure rates allowed obtaining accurate survival rates on elk for 6.5 years. Collars having 1 D-celilithium battery remained functional for up to 78 consecutive months. Population Estimation Continuing analyses strongly indicate that under-counting numbers of elk in groups is an important source of error, or negative bias, and may be equivalent in magnitude to negative bias associated with missing entire groups of elk.. Thus, errors in detecting groups and in counting elk within groups contribute to underestimates of true numbers of elk present. The assessment of negative bias was possible because of independently obtained mark-resight estimates of elk numbers that provided robust estimates of true numbers of elk. Elk Movements and Habitat Models Spatial plots of elk distribution, movements, and seasonal ranges were completed. All outputs were prepared for color reproduction, and as such, are not suitable to reproduce in this report. Further statistical assessment of predictive habitat use variables is ongoing. We anticipate completion and evaluation of the habitat type coverage during Fall 2000. Once the vegetation coverage is completed, analyses of habitat use can proceed towards completion. Manuscripts The following manuscripts are in various stages of progress. ESTIMATING ERROR ASSOCIATED WITH AERIAL TELEMETRY LOCATIONS OF ELK ON GRAND MESA, COLORADO. Projected outlet is CDOW technical series publication. SPATIAL DISTRIBUTION OF ELK ON GRAND MESA, COLORADO. Projected outlet is CDOW technical series publication in cooperation with USGS-BRD with financial support from the Rocky Mountain Elk Foundation. SURVNAL OF ELK CALVES DURING WINTER ON GRAND MESA, COLORADO. outlet is Journal of Wildlife Management. SURVNAL OF ADULT ELK ON GRAND MESA, COLORADO. Management. Projected Projected outlet is Journal of Wildlife ESTIMATING ELK DENSITIES ON GRAND MESA, COLORADO. Wildlife Management. Projected outlet is Journal of �245 Additional publications regarding effects of helicopter net-gunning capture protocols on elk calves and refinements to portable elk corral traps are planned. CONCLUSIONS With hunting deaths censored, annual survival rates during winter-spring were 2: 0.97 for adult females and 2: 0.96 for adult males during 7 consecutive winters while annual survival rates during summer-fall were 2: 0.99 for both adult females and males during 6 consecutive summers. High adult survival in conjunction with high over-winter calf survival (0.89 ± 0.04) allows this population to potentially increase> 15% annually in the absence of hunting antlerless elk. Hunting related mortalities were the primary cause of death for adult male and female elk. During 6 years, 98% of all adult male deaths and 89% of all adult female deaths were related to hunting. Hunting annually removed 76% of the adult branch-antlered males and 13% of the adult females. Wounding losses averaged 15% for adult males and 30% for adult females while 11% of yearling spike-antlered males were illegally harvested. We found counts of elk on quadrats corrected for sighting bias associated with detection of elk groups continued to under-estimate true numbers of elk present. Evidence indicates under-counting numbers of elk in groups also contributes to underestimating true numbers of elk present. If a quadrat sampling system is employed to estimate densities of elk, counts of elk on quadrats would need to be increased by l.3x to approximate true numbers of elk present. LITERATURE CITED Freddy, D. J. 1993. Estimating survival rates of elk and developing techniques to estimate population size. Colorado Division of Wildlife Game Research Report. July: 83-117 .. Freddy, D. 1. 1997. Estimating survival rates of elk and developing techniques to estimate population size. Colorado Division of Wildlife Game Research Report. July: 47-73. Freddy, D. 1. 1999. Estimating survival rates of elk and developing techniques to estimate population size. Colorado Division of Wildlife Game Research Report. July: In press .. Halfpenny,1. C., and E. A Biesiot. 1986. A field guide to mammal tracking in North America. Johnson Books, Boulder, Colorado, USA. Samuel, M.D., RO. Garton, M.W. Schlegel, and RG. Carson. 1987. Visibility bias during aerial surveys of elk in northcentral Idaho. journal of Wildlife Management 51 :622-630. Wade, D. A, and J. E. Browns. 1982. Procedures for evaluating predation on livestock and wildlife. Texas Agricultural. Experiment Station Publication B-1429. White, G. C., and R A Garrott. 1990. Analysis of wildlife radio-tracking data. Academic Press, Inc., San Diego, California USA. Prepared by _ David 1. Freddy Life/Science Researcher �246 Table 1. Causes of death for radio-collared elk between I December 1993 and 14 June 2000. Calves (M=male, F=female) were age 6-11 months and collared at age 6 months. Yearling males and females were age 12-17 months and juvenile males and females were age 18-23 months and all were collared at age 6 months. Adult males and females were age >24 months at time of death. Elk Age/Sex Class at Death Cause of Death Calves Yearling Juvenile Adult Total and -Code All M F M F M F M F M F Natural Causes Malnutrition-6 2 2 0 0 1 2 3 5 0 0 0 Unknown-Suspect Malnutrition-31 0 0 3 4 3 1 0 0 0 1 0 O· 2 Predation-Lion- 3 8 4 0 0 8 6 14 0 0 Predation-Bear-35 0 0 0 0 0 0 0 0 0 0 0 Unknown Predator-5 0 1 0 1 0 0 0 0 1 0 0 Unknown-Suspect Predation-30 2 5 0 3 8 0 1 1 2 11 0 Accident -Birthing- 32 0 0 0 0 2 2 0 0 0 0 2 Accident-FeU-I0 0 0 0 0 2 1 2 0 0 1 3 Unknown Cause-l l 2 0 2 4 1 0 0 0 1 2 6 Subtotals 17 14 19 27 46 Q 11 Q Q ~ ~ Legal Hunting Archery-33 0 0 0 9 5 9 7 16 0 2 0 Muzzleloading-34 0 0 8 2 8 4 12 0 2 0 0 ArcherylMuzzle-27 0 0 0 0 0 1 0 1 1 0 0 Rifle-First-46 0 0 33 33 13 46 0 0 0 0 13 Rifle-Second-47 0 0 0 0 16 19 16 19 35 0 0 Rifle- Third-48 0 9 13 22 0 0 0 0 9 13 0 Rifle-Late-29 0 0 0 0 0 7 0 7 7 0 0 Rifle-Unk.Reg-28 0 0 0 0 0 0 0 1 1 1 0 Subtotals 75 75 65 61 140 Q Q Q Q Q 1 I Wound~Loss Archery uzzle-24 0 0 2 2 2 2 4 0 0 0 0 0 0 0 1 2 Archery-52 1 0 0 1 3 1 Rifle-Flrst-43 0 0 1 0 0 0 1 3 3 4 0 Rifle-Second-44 0 0 0 4 4 11 15 0 0 0 11 Riflc- Third-45 0 0 0 0 3 1 3 1 4 0 0 Rifle-Unk.Reg. -25 2 2 4 0 0 0 0 0 2 2 0 Rifle-Late-26 0 0 0 3 0 3 3 0 0 0 0 Subtotals 12 23 13 24 37 Q Q Q Q ! ! Illegal Hunting Rifle-First-49 0 0 0 5 5 0 0 0 0 5 0 Rifle-Second-50 6 0 0 5 0 0 1 0 0 6 0 Rifle- Third-51 0 0 0 0 0 0 1 0 1 1 0 Rifle-Unk.Reg, -9 0 0 0 0 1 0 1 0 1 0 0 Archery Season-S 0 0 0 0 0 0 1 0 1 0 1 Out -of-Season-? 0 0 0 0 2 0 3 3 0 0 1 Subtotals 12" 14 17 Q Q Q Q ! ~ ~ 2 Presumed Hunting" Missing-ArchMuzz-21 0 0 0 0 1 0 1 0 1 0 0 Missing-Rifle Ist-40 0 0 3 3 3 6 0 0 0 0 3 Missing-Rifle 2nd-41 0 0 2 0 0 0 6 2 7 9 1 Missing-Rifle 3rd-42 0 0 0 0 0 0 0 0 0 0 0 Subtotals 0 16 .Q .Q ...!.Q Q Q Q Q .2 ! Totals 17 14 13 6 2 3 95 106 127129 256 • These elk were illegally shot as spike-antlered yearling males: (173.190/93) (173.232/93) (173.309/93) (173.320/93) (173.919/94) (174.059/94) (174.140/95) (174.200/95) (174.679/95) (174.729/95) (174.861/95) (173.210/96). (173.309/93) disappeared in 1994 during first rifle season when spike-antlered yearling males were not legal and was assumed to have been taken illegally. All other illegal deaths were confirmed. b These elk disappeared during hunting seasons and remained missing for several months and are assumed dead and legally harvested: (172.207/93) (172.269/93) (172.308193) (172.498/93) (172.649193) (172.800/93) (172.961/93) (172.991193) (173.390/93) (173.439/93) (173.760/94) (174.001194) (174.181/94) (174.770/95) (174.929/96) (175.199/96). �247 Table 2. Survival rates, inclusive of non-hunting and hunting mortalities, for winter-spring (WS, 1 December-14 June) and summer-fall (SF, 15 June-30 November) time periods from 1 December 1993-14 June 2000 for the cohort of calves age 6 months radio-collared in December 1993. Survival rate for male and female calves pooled when age 6-11 months was 0.92 (95% CI 0.86-0.98, n~73). Survival rates (S) calculated as a mean estimate of (alive)/(alive + dead) and variance S(1-S)/n collars. Elk Age (months) and Time Period (WS or SF dates) 6-83 6-1l(mos) 12-17 18-23 24-29 30-35 36-41 42-47 48-53 54-59 60-65 66-71 72-77 78-83 SF WS SF WS SF WS SF WS SF WS SF WS ALL WS 1993-94 1994 1994-95 1995 1995-96 1996 1996-97 1997 1997-98 1998 1998-99 1999 1999-00 1993-00 MALES 0.00 0.88 1.00 Survival 0.89 0.21 1.00 0.50 1.00 0.00 L95%CI 0.78 0.76 0.06 0.00 0.00 U95%CI 0.99 0.99 0.37 0.00 1.00 4 2f 33 32 28 n collars 36 28c 4 1& 3"& o 0 Censored" 0 o 0 o 2" 22d 0 Died 4 33 4b 0 2 0 1 4 o 0 o 0 Nonhunt 4 o 0 o 29 4 Hunt 0 0 22 0 2 0 1 FEMALES Survival L 95%CI U 95% CI n collars Censored" Died Nonhunt Hunt 0.95 0.87 1.00 37 0 0.97 0.92 1.00 35 o 34 0 2 Ih 0 1 1 5 2 () o 0 0 o 0 o 1 If 5 1 1.00 0.97 0.91 1.00 34 0.97 0.91 1.00 33 0.84 0.71 0.97 32 o 0 o 1.00 27 0 0 0 0 0.87 0.72 1.00 23 o 0.96 0.87 1.00 24 0 1 1 3 0 3 0.89 0.76 1.00 27 o i 3 • Censored denotes collar failure and/or animal life/death status not known. b Collar (173.309/93) disappeared 1994 hunting seasons and assumed dead. c Includes (173.241193) collar failure August 1995 but seen January 1996. • Collars (173.390/93),(173.439/93) disappeared 1995 hunting seasons and assumed dead . • Censored elk; (173.241/93) failed August 1995, (173.381/93) slipped collar December 1995. f Includes (173.340/93) alive May 1997. , Censored elk: (173.249/93) slipped collar September 1997. k Collar (172.800/93) disappeared 1994 hunting seasons and assumed dead. i Collar (172.961193) disappeared 1997 hunting seasons and assumed dead. j Collar (172.991193) disappeared 1998 hunting seasons and assumed dead. k Censored elk: (172.849/93) failed May 1999, (173.151193) failed February 1999. 1.00 18 0.89 0.73 1.00 18 0.94 0.81 1.00 16 0.43 0.26 0.60 35 0 21;: 1 20 4 16 o 21;: o Ji 0 2 o 0 0 o 1 2 0 �248 Table 3. Survival rates, inclusive of non-hunting and hunting mortalities, for winter-spring (WS, 1 December-14 June) and summer-fall (SF, 15 June-30 November) time periods from 1 December 1994 -14 June 2000 for the cohort of calves age 6 months radio-collared in December 1994. Survival rate fot male and female calves pooled when age 6-11 months was 0.90 (95% CI 0.83-0.97, n=69). Survival rates (S) calculated as a mean estimate of ~alive2/~alive + dead2 and variance Sp-S2/n collars. Elk Ase {months~and Time Period {WS or SF dates} 6-11(mos} 12-17 18-23 24-29 30-35 54-59 60-65 66-71 36-41 42-47 48-53 6-71 . SF WS ALL SF WS WS WS WS WS SF SF SF 1994-95 1995 1995-96 1996 1996-97 1997 1997-98 1998 1998-99 1999 1999-00 1994-00 MALES 0.91 0.90 0.00 Survival 0.96 0.08 1.00 0.50 1.00 0.00 0.81 0.79 L 95%CI 0.88 0.00 0.00 1.00 U 95% CI 1.00 1.00 1.00 0.19 2d n collars 33b 30b 31 26 25 2 1 0 2C.~ 1c 1~ Censored' 0 0 0 0 0 0 3 3 31 Died 0 0 0 1 23 1 Nonhunt 3 0 0 4 1 0 0 0 0 0 3 27 HWlt 0 23 0 0 0 1 FEMALES Survival L 95% CI U95%CI n collars Censored' Died Nonhunt Hunt 0.89 0.78 0.99 36 0 4 4 0 0.97 0.91 1.00 32 0 1 0 1 0.97 0.90 1.00 30 1· 1 1 0 0.93 0.83 1.00 29 0 2 0 2 I 0.96 0.89 1.00 27 0 1 0 0.85 0.70 0.99 26 0 4 0 4 1.00 22 0 0 0 0 • Censored denotes collar failure and/or animal life/death status not known. Includes (173.949/94) collar failure December 1994 but seen January 1996 • Censored elk (173.949/94) collar failure, which was killed in first rifle season 1996. d Includes (174.030/94) alive June 1997. • Censored elk (173.719/94) slipped collar May 1996. f Collar (173.760/94) disappeared 1998 hunting seasons and assumed dead. • Censored elk (174.030/94) slipped collar September 1998. b Censored elk (173.799/94) collar failure February 2000. b 0.64 0.42 0.85 22 0 8f 0 8 1.00 14 0 0 ·0 0 0.86 0.66 1.00 14 0 2 0 2 0.91 0.72 1.00 11 Ih 1 0 1 0.29 0.14 0.45 34 2·,h 24 5 19 �249 Table 4. Survival rates, inclusive of non-hunting and hunting mortalities, for winter-spring (WS, 1 December-14 June) and summer-fall (SF, 15 June-30 November) time periods from 1 December 1995-14 June 2000 for the cohort of calves age 6 months radio-collared in December 1995. Survival rate for male and female calves pooled when age 6-11 months was 0.88 in 1995 (95% CI 0.81-0.96, n=69). Survival rates (S) calculated as a mean estimate of ~alivel/~alive+deadl and variance of Sp-S2/n collars. Elk Ag_e(months) and Time Period (WS or SF dates) 48-53 54-59 6-59 6-11 (mos) 12-17 18-23 30-35 36-41 42-47 24-29 ALL SF WS WS SF SF WS SF WS WS 1995-96 1996 1996-97 1999 1999-00 1995-00 1997 1997-98 1998 1998-99 MALES 0.00 Survival 0.86 0.83 0.96 0.23 1.00 0.25 1.00 0.00 0.74 L 95%CI 0.68 0.87 0.04 0.00 0.98 U95%CI 0.97 1.00 0.41 0.86 n collars 35 32 29 24 22 4 4 1 1 5b,C,d,. 2b Ie Id 1· 0 0 0 0 Censored" 5 5 1 32 Died 1 17 0 3 0 5 6 Nonhunt 0 0 0 0 0 0 1 0 5 0 17 0 1 26 Hunt 0 3 FEMALES Survival L95%CI U95%CI n collars Censored" Died Nonhunt Hunt • Censored Censored e Censored d Censored • Censored t Censored • Censored b 0.91 0.81 1.00 34 0 3 3 0 0.87 0.75 0.99 31 0 4 0 4 0.93 0.82 1.00 27 0 2 1 1 I 0.83 0.66 0.99 23 2f 1.00 Ig 4 0 4 0 0 0 18 0.89 0.73 1.00 18 0 2 0 2 1.00 16 0 0 0 0 denotes collar failure and/or animallifeldeath status not known. elk (174.619/95) slipped collar May 1996, (174.800/95) capture mortality. elk (174.660/95) slipped collar July 1996. elk (174.671/95) likely collar failure July 1997. elk (174.190/95) slipped collar May 1998. elk (174.441/95) slipped collar August 1997 and (174.580/95) collar failure July 1997. elk (174.470/95) collar failure December 1997. 0.88 0.70 1.00 16 0 2 0 2 1.00 14 0 0 0 0 0.36 0.18 0.53 31 3 17 4 13 �250 Table 5. Survival rates, inclusive of non-hunting and hunting mortalities, for winter-spring (WS, 1 December-14 June) and summer-fall (SF, 15 June-30 November) time periods from 1 December 1993-14 June 2000 for the cohort of calves age 6 months radio-collared in December 1996. Survival rate for male and female calves pooled when age 6-11 months was 0.86 (95% CI 0.81-0.96, n=69). Survival rates (S) calculated as a mean estimate of ~alive~/~alive+dead~ and variance of Sp-S~/n collars. Elk A~e (months2 and Time Period (WS or SF dates) 6-47 6-11 (mos) 36-41 42-47 12-17 24-29 30-35 18-23 ALL SF SF WS WS SF WS WS 1996-00 1999-00 1996-97 1997 1997-98 1998-99 1999 1998 MALES 1.00 0.09 0.85 0.97 1.00 0.30 Survival 1.00 0.36 0.00 L 95%CI 0.73 0.90 0.00 0.17 0.98 1.00 0.63 0.19 U 95% CI 0.54 n collars 34 10 3 34 29 28 28 10 1b 1b Censored" 0 0 0 0 0 0 0 Died Nonhunt Hunt 5 5 0 1 0 1 0 0 0 18d 0 18 0 0 0 7 0 7 0 0 FEMALES Survival L 95% CI U95%CI n collars Censored" Died Nonhunt Hunt 0.86 0.74 0.98 35 0 5 5 0 1.00 1.00 1.00 1.00 1.00 30 0 0 0 0 30 0 0 0 0 0.80 0.65 0.95 30 0 6C 0 6 24 0 0 0 0 24 0 0 0 0 24 0 0 0 0 • Censored denotes collar failure and/or animallifeldeath status not known. b Censored elk (174.619/96) capture mortality. c Collar (174.929/96) disappeared 1998 hunting seasons and assumed dead. d Collar (175.199196) disappeared 1998 hunting seasons and assumed dead. 31 5 26 0.69 0.53 0.85 35 0 11 5 6 �251 Table 6. Survival rates, inclusive of non-hunting and hunting mortalities, for winter-spring (WS 1 December-14 June) and summer-fall (SF, 15 June-30 November) time periods from 1 December 1993 - 14 June 2000 for all radio-collared adult female elk age :::12 months. Survival rates (S) calculated as a mean estimate of (alive)/(alive + dead2 and variance S{I-S2/n collars. Elk A8e and Time Period (WS or SF dates 2 Adult Adult Adult Adult Adult Adult Adult Adult Adult Adult Adult Adult Adult WS WS SF WS SF WS SF WS WS WS SF SF SF 1993-94 1994 1994-95 1995 1995-96 1996 1996-97 1997 1997-98 1998 1998-99 1999 1999-00 0.99 0.77 0.98 0.98 Survival 0.96 0.87 0.96 0.94 0.91 0.89 0.94 0.89 0.98 0.70 0.95 L95%CI 0.91 0.80 0.90 0.87 0.97 0.83 0.95 0.92 0.90 0.84 0.95 1.00 0.84 1.00 U95%CI 0.94 1.00 0.98 1.00 0.96 1.00 1.00 0.95 0.98 0.94 100b[ 95b& 68b• 1431< 1271 152m ncollars 129bh ll~ 138° 136q 103' 101ft 88v 2ft 2i Censored" IP 0 2' 0 2'" 0 0 0 0 0 0 c 4 Died 13 3 4 8 7 13 2 31r 2 11 2 16 3 Nonhunt 0 1 1 1 0 4 0 2 0 1 1 1 Hunt 13 2 12 8 31 0 11 1 3 3 15 2 • Censored denotes collar failure or life/death status unknown. e Collars (172.800/93,172.649/93) disappeared 1994 hunt seasons. • Composed of6-18+ and 62-30+ months old females. I Composed of34-18+ and 61-30+ months old females. ; Composed of 30-18+ and 89-30+ months old females. k Composed of31-12+, 29-24+, and 83-36+ months old females. • Composed of30-12+, 23-24+, and 99-36+ months old females. • Censored elk: (174.441/95) slipped collar August 1997 and (174.580/95) likely failure June 1997 P Censored elk: (174.470/95) likely collar failure August 1997. 'Collars (172.269/93, 172.308/93, 172.498/93, 172.991/93, 173,760/94, 174.929/96) disappeared 1998 hunt seIjSOns and assumed dead. • Composed of 24-36+ and 77-48+ months old females W Censored elk: (172.418/93 April 2000, 173.799/94 Feb.2000) Includes (172.011/93) collar failure April 1994, seen January 1996 Collar (172.207/93) disappeared 1995 hunt seasons, assumed dead. f Composed of35-12+, 6-24+, and 59-36+ months old females. k Composed of32-12+, 34-24+, and 63-36+ months old females. j Censored elk: (172.011/93) failure and (173.719/94) slipped collar. I Composed of27-18+ and 100-30+ month old females. b d o Composed of30-18+ and 108-30+ months old females. Composed of30-24+ and 106-36+ months old females. • Composed of 24-30+ and 81-42+ months old females. q I V Censored elk: (172.849/93, 173.151/93) failed collars. Composed of24-42+ and 66-54+ months old females �252 Table 7. Survival rates, inclusive of non-hunting and hunting mortalities, for winter-spring (WS 1 December-14 June) and summer-fall (SF, 15 June-30 November) time periods from 1 December 1993 - 14 June 2000 for radiocollared adult female elk age :::24 months. Survival rates (S) calculated as a mean estimate of (alive)/(alive + dead) and variance S{l-S~/n collars. Elk Ase and Time Period (WS or SF dates2 Adult Adult Adult Adult Adult Adult Adult Adult Adult Adult Adult Adult Adult WS WS SF WS SF WS WS SF WS SF SF WS SF 1993-94 1994 1994-95 1995 1995-96 1996 1996-97 1997 1997-98 1998 1998-99 1999 1999-00 Survival 0.77 0.89 0.98 0.95 0.82 0.93 0.93 0.89 0.89 0.98 0.98 0.93 0.99 L 95%CI 0.72 0.84 0.70 0.95 0.83 0.95 0.90 0.87 0.88 0.84 0.96 0.88 0.97 0.84 1.00 U 95% CI 1.00 1.00 1.00 1.00 0.95 0.91 0.99 0.98 0.99 0.95 0.95 n collars 136 103 101 62 65 97 100 122 108 88 61 89 112 Ib Id 2f 2< Censored" 2· 0 0 0 0 0 0 0 0 Died 3 12 7 31 11 4 6 12 13 2 2 2 1 Nonhunt 1 0 0 0 1 1 3 0 1 2 1 1 0 Hunt 2 11 7 13 31 3 3 11 1 0 11 1 1 • Censoreddenotescollarfailureor life/deathstatusunknown. Censoredelk (174.441/95)(174.580/95) e Censoredelk (172.849/93,173.151193) e Censoredelk (172.011/93) Censoredelk (174.470195) fCensored elk (172.418/93,173.799/94) B d Table 8. Survival rates, inclusive of non-hunting and hunting mortalities, for winter-spring (WS, 1 December-14 June) and summer-fall (SF, 15 June-30 November) time periods from 1 December 1993 - 14 June 2000 for the group of adult female elk age :::18 months when radio-collared in December 1993. Survival rates (S) calculated as a mean estimate of (alive)/(alive + dead) an? variance S(l-S)/n collars. Elk Ase and Time Period (WS or SF dates2 Adult Adult Adult Adult Adult Adult Adult Adult Adult Adult Adult Adult Adult SF WS WS SF WS SF WS SF WS SF WS SF WS 1993-94 1994 1994-95 1995 1995-96 1996 1996-97· 1997 1997-98 1998 1998-99 1999 1999-00 Survival 1.00 1.00 0.73 0.83 0.96 0.82 0.93 0.88 0.93 0.92 1.00 0.94 0.97 0.57 0.68 L95%CI 0.72 0.78 0.85 0.84 0.87 0.92 0.91 0.85 0.99 U95%CI 0.88 1.00 0.91 0.99 0.97 1.00 1.00 1.00 1.00 331: II)l 241: 241 68b• 34i 3~ 36i 36i n collars 65bf 53bf 49bh 42h o o Censored" 0 o 0 0 0 o 0 0 0 l' 12< Died 3 0 2 1 9 o 4 3 6d 4 3 o o Nonhunt o 0 0 0 o 1 1 0 3 1 Hunt 9 o 4 2 11 6 3 0 2 1 3 0 • Censored denotes collar failure or Iifcldcalh status unknown • CoDar (172.649/93) disappeared 1994 hunt seasons. • Composed of 6-18+ and 62-30+ month old females. • Censored (172.011/93) coDar failure Apri11994. i Composed of 100% females 48+ months old. • Composed of 100% females 72+ months old - Censored elk (172.418/93) collar failure Apri12000. • Includes (172.011/93) collar failure Apri11994 but seen Jan. 1996. • CoDar (172.207/93) disappeared 1995 hunt eeasons and assumed dead. r Composed of l000A. females 24+ months old. • Composed of 100% females 36+ months old. ; Composed of 100% females 60+ months old. , Composed of l000A. females 84+ months old �253 Table 9. Annual survival rates from 1 December-30 November each year 1993-94 - 1998-99 inclusive of nonhunting and hunting mortalities and overall survival through Novemeber 1999 for the group of adult female elk age ~18 months when radio-collared in December 1993. Survival rates (S) calculated as a mean estimate of (alive)/(alive+dead) and variance S(1-S)/n collars. Elk Age and Time Period (dates) Adult Adult Adult Adult Adult Adult Adult 19931994199519961997199819931994 1995 1996 1997 1998 1999 1999 Survival 0.78 0.84 0.86 0.94 0.71 0.83 0.30 L 95%CI 0.68 0.75 0.75 0.87 0.55 0.68 0.19 U 95% CI 0.88 0.93 0.97 1.00 0.86 0.99 0.41 68b,. 53b.f 34) 24k n collars 42g 36i 67 Ib Ib Censored" 0 0 0 0 0 d Died 15° 10 6 2 10 47 4 Nonhunt 2 1 3 0 0 6 0 Hunt 13 9 41 3 2 10 4 • Censored denotes collar failure or life/death status unknown b Includes collar (172.011193) failed 4/1994 seen alive 111996 d Collar (172.207/93) disappeared 1995 hunt seasons, assumed dead. t Composed of 100% females 24+ months old. ~ Censored (172.011193) failure 4/1996. j Composed of 60+ months old females. c Collar (172.649/93) disappeared 1994 hunt seasons • Composed of 6-18+ and 62-30+ months old females • Composed of36+ months old females I Composed of 48+ months old females k Composed of72+ months old females Table 10. Survival rates for adult male elk during summer-fall, 15 June-30 November, 1995-1999. All deaths due to hunting. Survival rates (S) calculated as 11 mean estimate of (alive)/(alive + dead) and variance S(I-S)/n collars. Removal rate = (1-S). All males radio-collared as 6-month old calves in December 1993-1996. Summer-Fall Time Period Year All Years 1995 1999 1996 1997 1998 MALES (age 24-29 months) Survival 0.21 0.22 0.08 0.23 0.35 L 95%CI 0.06 0.00 0.14 0.04 0.17 U95%CI 0.37 0.19 0.30 0.41 0.54 n collars 28 103 25 22 28 Died Hunting 80 22 23 17 18 Removal Rate 0.79 0.78 0.92 0.77 0.65 MALES (age 24-53 months) Survival 0.21 L 95% CI 0.06 U95%CI 0.37 n collars 28 Died Hunting 22 Removal Rate 0.79 0.14 0.01 0.27 29 25 0.86 0.24 0.06 0.42 25 19 0.76 0.33 0.17 0.50 33 21 0.64 0.27 0.00 0.57 11 8 0.73 0.24 0.17 0.32 126 95 0.76 �254 Table 11. Survival rates for adult female elk during summer-fall, 15 June-30 November, 1994-1999. For adult females, 2 natural deaths occurred during summer-fall in these 6 years and these 2 elk were censored from survival calculations. All deaths shown were due to hunting. Survival rates (S) calculated as a mean estimate of (alive) / (alive + dead) and variance of S(I-S) / n collars. Removal rate = (l -S). Females radio-collared as adults in 1993 or as 6-month calves in December 1993-1996. Summer-Fall Time Period Year 1994 1999 1995 1996 1997 1998 All Years FEMALES (age> 24 months) Survival 0.83 0.89 0.77 0.89 0.93 0.91 0.87 L 95% CI 0.73 0.70 0.83 0.88 0.85 0.84 0.84 U95%CI 0.92 0.98 0.95 0.84 0.95 0.89 0.96 n collars 64 110 122 136 99 628 97 Died Hunting 11 11 7 10 13 31 83 Removal Rate 0.17 0.11 0.07 0.09 0.11 0.23 0.13 FEMALES (age 12-17 months) 0.97 Survival L95% CI 0.92 U95%CI 1.00 n collars 35 Died Hunting 1 Removal Rate 0.03 0.97 0.91 1.00 32 1 0.03 0.87 0.75 0.99 31 4 0.13 1.00 30 0 0.00 0.95 0.92 0.99 128 6 0.05 �255 Appendi x I. MortaLities of 256 radio-coLLared eLk from 1 December 1993 through 14 June 2000. Age is aE2roximate a2e in ~ears of eLk at death: C=caLf a2e 6-11 months, Y=~earLin2 a2e 12-23 months. Death Frequency 101 Tra!;! Cause of Death & Code Number Date Age Zone Sex Year Ca~tured Site legaL kiLL second rifLe season-47 19-Oct-99 B 9 SR F 172.019/93 legaL kiLL second rifLe season-47 27-Oct-95 5 BR A F 172.030/93 legaL kiLL third rifLe season-48 31-Oct-98 F 12 EG B 172.050/93 Unknown-suspect predation-30 14-Jun-95 B F 4 GR 172.039/93 Unknown-11 12-Feb-96 F 11 BR A 172.070/93 legaL kiLL third rifLe season-48 05-Nov-94 6 GM A F 172.080/93 Wounding Loss late rifle season-26 16-Jan-94 F 11 GR B 172.090/93 legaL kiLL first rifLe season-46 12-Oct-98 0 F 11 SM 172.090/94 legaL kiLL first rifLe season-46 14-Oct-96 8 B F 172.101/93 SR Wounding Loss second rifLe season-44 31-Oct-96 8 B F GC 172.128/93 Wounding loss first rifLe season-43 19-0ct-95 17 BC C F 172.139/93 legaL kiLL second rifLe season-47 23-0ct-94 C F 3 CC 172.160/93 Wounding Loss second rifle season-44 03-Nov-94 3 MC C F 172.181/93 Wounding Loss second rifle season-44 15-Nov-94 3 GS C F 172.201/93 Disappear first rifle season-40 30-Nov-95 4 SG 0 F 172.207193 legaL kiLL rifle season-28 30-Nov-99 0 F 8 GH 172.219/93 legaL kiLL second rifle season-47 29-Oct-98 8 MC C F 172.229/93 legaL kiLL archery/muzzLe season-27 04-Oct-94 3 HY C F 172.258/93 Disappear second rifLe season-41 29-0ct-98 C F 12 MC 172.269/93 Wounding loss archerY/muzzLe-24 04-Oct-94 3 GS C F 172.277/93 legal kiLL second rifle season-47 23-0ct-94 D 6 SG F 172.290/93 legal kilL muzzleloading season-34 16-Sep-96 SM D F 4 172.290/94 Wounding loss second rifLe season-44 30-0ct-97 E 8 AC F 172.300/93 Disappear second rifle season-41 29-Oct-98 8 GH D F 172.308/93 legal kiLL first rifle season-46 10-Oct-99 D 11 SG F 172.330/93 legaL kiLL second rifle season-47 19-Oct-99 SG D F 11 172.358/93 legaL kiLL archery season-33 18-Sep-94 3 AC E F 172.369/93 legaL kiLL Late rifLe season-29 14-Jan-96 FM B F 4 172.369/94 Wounding Loss Late rifLe season-26 29-Dec-94 8 AC E F 172.409/93 legaL kiLL third rifle season-48 01-Nov-98 MH E F 172.448/93 I12 legaL kiLL second rifle season-47 24-0ct-96 E 8 MH F 172.459/93 Disappear first rifle season-40 15-0ct-98 12 FS H F 172.498/93 legaL kiLL second rifle season-47 21-0ct-95 3 FS H F 172.509/93 legaL kiLL first rifLe season-46 12-0ct-98 6 FS H F 172.519/93 Wounding Loss second rifLe season-44 29-Oct-98 8 FS H F 172.530/93 Mountain lion predation-3 01-Feb-94 6 FS H F 172.542193 Accident-irrigation ditch-10 06-Jan-99 15 FM B F 172.542/94 legaL kill late rifle season-29 28-Nov-95 7 PG H F 172.549/93 Wounding loss second rifle season-44 03-Nov-94 16 PG H F 172.570/93 legaL kiLL first rifLe season-46 17-Oct-94 PG H F 3 172.581/93 legal kill late rifle season-29 08-Dec-95 3 FM B F 172.581/94 Unknown-11 24-Apr-96 5 PG H F 172.590/93 Accident, birthing/calving-32 04-0ct-94 PG F 3 H 172.610/93 legaL kiLL third rifLe season-48 04-Nov-98 H F 5 PG 172.620/93 Accident, birthing/calving-32 14-Jun-96 16 PG H F 172.639/93 Disappear first rifle season-40 30-Nov-94 H F 2 172.649/93 PG legal kiLL third rifle season-48 02-Nov-97 H F 5 PG 172.658/93 legal kiLL late rifle season-29 16-Jan-95 4 PG H F 172.670/93 Wounding loss late rifle season-26 22-Dec-94 PG H F 9 1'72.678/93 Illegal kit l-7 24-Jan-94 B F 8 172.690/93 FM legaL kiLL third rifLe season-48 05-Nov-95 B 7 FM F 172.699/93 MaLnutrition-6 05-May-99 19 lS H F 172.730/95 Unknown-suspect predation-30 07-Aug-96 lS H F 11 172.749/95 Wounding Loss second rifLe season-44 28-Oct-98 15 lS H F 172.758/95 Wounding Loss second rifLe season-44 16-Oct-99 F 6 lS H 172.769/95 Mountain Lion predation-3 31-Dec-99 6 SR B F 172.790/93 Disappear second rifLe season-41 30-Nov-94 SR B F Y 172.800/93 legal kilL archery season-33 05-Sep-99 6 EG B F 172.810/93 legaL kilL archery season-33 08-S.ep-96 3 EG B F 172.821/93 legaL kilL third rifLe season-48 Q6-Nov-96 3 BC C F 172.890/93 Mountain Lion predation-3 18-Mar-94 BC C F C 172.899/93 legaL kiLL first rifLe season-46 18-0ct-95 GH D F 2 172.950/93 Disappear 'second rifLe season-41 30-Oct-97 MG D F 4 172.961/93 Disappear second rifLe season-41 29-0ct-98 D 5 SG F 172.991/93 Malnutrition-6 25-Apr-94 WM C 173.000/93 G F legaL kill third rifLe season-48 31-Oct-98 F 4 173.000/94 HM B Accident-feLL-10 07-Jan-98 WM 4 G F 173.010/93 legal kiLL first rifLe season-46 19-Oct-95 2 GM A M 173.041/93 �256 Appendix 1. (continued) Frequency ID/ Tral2 Year Captured Site Zone 173.050/93 MH E 173.060/93 MH E 173.070/93 E MH 173.081/93 FS H 173.090/93 FS H 173.100/93 FS H 173.120/93 GM A 173.140/93 PG H 173.160/93 FS H 173.180/93 PG H 173.171/93 FM B 173.190/93 BC C 173.190/96 BC C 173.201/93 GR B 173.210/93 GC B 173.210/96 LM D 173.219/93 BC C 173.219/96 CG D 173.232/93 BR A 173.232196 BC C 173.262/93 BC C 173.262/94 HM B 173.269/93 MC C 173.279/93 C MC 173.289/93 MC C 173.289/94 PR A 173.300/93 C GS 173.300/96 MC C 173.309/93 HY C 173.320/93 HY C 173.320/94 HM B 173.332/93 GH D 173.332/96 BC C 173.340/93 SG D 173.351/93 SG D 173.351/96 BG E 173.359/93 AC E 173.359/96 C MC 173.370/93 MH E 173.370/96 RG E RG' 173.381/96 E 173.390/93 MG D 173.402/93 MG D 173.410/93 WM G 173.410/96 EW G 173.420/93 WM G 173.429/93 MH E 173.429/96 G EW 173.439/93 PG H 173.450/93 PG H 173.450/96 EW G 173.461/93 PG H 173.461/94 CM A 173.461/96 GM A 173.469/93 FS H 173.469/94 PR A 173.479/93 FS H 173.479/96 RG E 173.492/93 FS H 173.492196 GM A 173.502/93 FS H 173.510/93 FM B 173.521/93 FM B 173.530/94 FM B 173.540/94 OG B 173.549/94 B HM 173.559/94 PR A 173.569/94 PR A 173.580/94 PR A Death Sex F F F F F F M F F F F M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M F M F F F F Age 5 2 5 3 4 4 2 3 3 6 5 Y 2 2 2 2 2 2 Y 2 C 2 2 3 C 2 2 2 Y Y ,2 2 2 4 2 2 2 2 2 C C 2 2 2 1 2 2 3 2 2 3 C Y 2 C 2 2 2 2 3 3 2 2 7 2 Y 5 3 Y Date 31-Oct·98 27-Dec-95 03-Mar-98 22-OCt-96 24-Oct-97 30-Oct-97 14-OCt-95 31-Oct-96 13-Noy-96 10-OCt-99 10-OCt-98 29-OCt-94 10-OCt-98 30-Noy-95 19-0ct-95 30-OCt-97 06-Noy-95 21-Oct-98 15-Noy-94 31-OCt-98 18-Mar-94 12-Oct-96 06-Noy-95 12-Oct-96 22-Mar-94 18-Sep-96 24-Oct-95 10-OCt-98 30-Noy-94 20-Oct-94 14-Sep-96 12-Sep-95 10-0ct-98 20-0ct-97 05-Noy-95 27-Oct-98 06-Sep-95 09-Noy-98 08-Noy-95 08-Jan-97 19-Feb-97 30-Noy-95 18-0ct-95 02-Noy-95 29-Sep-98 24-0ct-95 14-Sep-95 10-0ct-99 30-Noy-95 15-Oct-95 21-Oct-99 07-Feb-94 19-5ep-95 22-OCt-98 25-Apr-94 20-OCt-96 10-Sep-95 17-0ct-98 03-Sep-95 11-0ct-99 13-Noy-96 28-Aug-95 17-0ct-95 31-OCt-98 13-0ct-96 07-Sep-95 30-Oct-99 16-Oct-97 21-Dec-95 Cause of Death & Code Number Legal kill third rifle season-48 Legal kill late rifle season-29 Illegal kill-7 Legal kill second rifle season-47 Legal kill second rifle season-47 Legal kill second rifle season-47 Legal kill first rifle season-46 Wounding loss second rifle season-44 Legal kill third rifle season-48 Legal kill first rifle season-46 Legal kill first rifle season-46 Illegal kill second rifle season-50 Legal kill first rifle season-46 Wounding loss third rifle season-45 Legal kill first rifle season-46 Illegal kill archery/muzz. season-8 Legal kill third rifle season-48 Legal kill second rifle season-47 Illegal kill second rifle season-50 Legal kill third rifle season-48 Unknown-11 Legal ki II first rifle season-46 Legal k i II third rifLe season-48 LegaL kill first rifLe season-46 Unknown-11 LegaL kiLL muzzLeLoading season-34 LegaL kill second rifLe season-47 LegaL kiLL first rifle season-46 ILlegal kiLL first rifle season-49 Illegal kill first rifle season-49 LegaL kill archery season-33 LegaL kill muzzleloading season-34 Legal kiLL first rifle season-46 LegaL kill second rifLe season-47 Legal kill third rifle season-48 Legal kill second rifle season-47 Legal kill archery season-33 Wounding loss third rifle season-45 Legal kill first rifLe season-46 Mountain lion predation-3 Unknown-suspect predation-30 Disappear first rifle season-40 Legal kill first rifle season-46 Wounding (oss second rifle season-44 Wounding Loss archery/muzzle seasons-24 Legal kill second rifle season-47 Legal kill archery season-33 Legal kill first rifLe season-46 Disappear first rifLe season-40 Legal kill first rifle season-46 Legal kill second rifle season-47 Malnutrition-6 Wounding loss archery season-50 LegaL kill second rifLe season-47 MaLnutrition-6 Legal kill second rifle season-47 Legal kill archery season-33 LegaL kill second rifle season-47 Legal kiLL archery season-33 Legal kill first rifle season-46 LegaL kill third rifLe season-48 Legal kill archery season-33 LegaL kill first rifle season-46 Legal kill third rifle season-48 Legal kill first rifle season-46 Legal kill archery season-33 Legal kill third rifle season-48 Wounding loss first rifle season-43 Unknown-11 �257 A~ndix I. (continued) Frequency 10/ Tra!:! Zone Year Ca~tured Site 173.589/94 OG B 173.610/94 C BC 173.640/94 C MC 173.640/96 GM A 173.649/94 BC C 173.659/94 C MC 173.679/94 MP D 173.689/94 BG E 173.699/94 E BG 173.719/96 MC C 173.739/94 BG E 173.750/94 E BG 173.760/94 E MR 173.779/94 MR E 173.789/94 MR E 173.819/94 MM F 173.829/94 MM F 173.840/94 MM F 173.849/94 SM D 173.859/94 SM D 173.870/94 SM D 173.919/94 BC C C 173.929/94 BC 173.939/94 C BC 173.959/94 BC C 173.970/94 MC C 173.981/94 D MP 173.990/94 C BC 174.001/94 BG E 174.009/94 BG E 174.019/94 E BG 174.039/94 MR E 174.049/94 F MM 174.059/94 E MR 174.069/94 BG E 174.080/94 E MR 174.090/94 MR E 174.101/94 MM F 174.101/96 E~ G 174.109/94 MM F 174.119/94 MM F 174.119/95 GB B 174.129/94 MM F 174.140/94 F MM 174.140/95 GB B 174.150/94 MM F 174.160/94 MM F 174.170/94 FM B 174.170/96 C MC 174.181/94 FM B 174.200/95 MS F 174.210/95 MS F 174.220/95 MS F 174.220/96 F MS 174.230/95 MS F 174.230/96 G E~ 174.240/95 E MD 174.240/96 SR B 174.301/95 MD E 174.319/95 MS F 174.329/95 E MD 174.339/95 MD E 174.349/95 E MD 174.360/95 MD E 174.370/95 HG E 174.380/95 AP D 174.401/95 AP D 174.420/95 AP D 174.478/95 W C Death Sex F F F F F F F F F F F F F F F F F F F F F M M M M M M M M M M M M M M M M M M M M M M M M F F F F F F F F F F F Age C 3 C C 3 4 it 2 5 2 4 4 4 5 C 3 2 4 4 2 C Y 2 2 2 2 2 2 2 2 12 2 2 Y 2 3 2 Y 3 2 C 2 2 C Y 2 2 C 2 2 Y 2 C 3 C 2 C 2 3 Y Y C 2 Y 4 4 2 2 2 Date 01-Jun-95 14-Nov-97 30-Mar-95 16-Mar-97 11-0ct-97 25-0ct-98 29-Oct-98 31-Oct-96 30-Nov-99 03-Nov-98 29-Oct-98 18-Oct-98 29-Oct-98 31-Dec-99 20-Mar-95 30-Oct-97 10-Jan-97 29-Sep-98 20-Oct-98 23-0ct-96 21-Feb-95 02-Nov-95 16-Oct-96 18-Sep-96 02-Nov-96 20-Oct-96 02-Nov-96 20-Oct-96 30-Nov-96 12-0ct-96 13-Nov-96 05-Sep-96 14-Sep-96 17-0ct-95 26-Sep-96 11-0ct-97 20-Oct-96 24-May-96 10-0ct-99 19-Oct-96 01-Mar-95 30-Nov-97 14-0ct-96 14-Mar-95 30-Nov-96 14-0ct-96 15-Sep-96 25-Apr-95 10-0ct-98 30-Nov-96 18-0ct-96 11-0ct-97 08-Apr-96 11-0ct-99 24-Apr-96 13-0ct-98 17-Apr-96 12-Sep-98 20-Sep-98 02-Oct-96 03-Sep-96 27-Mar-96 13-Oct-97 22-Apr-97 30-Nov-99 30-Aug-99 22-Sep-96 11-Oct-97 13-Sep-97 Cause of Death & Code Number Unknown-suspect predation-30 ~ounding loss third rifle season-45 Unknown-suspect predation-30 Malnutrition-6 legal kill first rifle season-46 legal kill second rifle season-47 Wounding loss first rifle season-43 legal kill first rifle season-46 ~ounding loss rifle season-25 legal kill thi~d rifle season-48 ~ounding loss second rifle season-44 legal kill second rifle season-47 Disappear second rifle season-41 legal kill late rifle season-29 Unknown-suspect malnutrition-31 legal kill second rifle season-47 legal kill late rifle season-29 ~ounding loss archery/muzzle seasons-24 legal kill second rifle season-47 legal kill second rifle season-47 Mountain lion predation-3 Illegal kill second rifle season-50 legal kill first rifle season-46 legal kill archery season-33 legal kill third rifle season-48 legal kill second rifle season-47 legal kill third rifle season-48 legal kill second rifle season-47 Disappear archery/muzzle season-21 legal kill first rifle season-47 Illegal kill rifle season-9 legal kill archery season-33 legal kill muzzleloading season-34 Illegal kill first rifle season-49 ~ounding loss archery/muzzle-24 legal kill first rifle season-46 legal kill second rifle season-47 Unknown-suspect predation-30 legal kill first rifle season-46 legal kill second rifle season-47 Mountain lion predation-3 ~ounding loss rifle season-25 legal .ki ll first rifle season-46 Mountain lion predation-3 Illegal kill third rifle season-51 legal kill first rifle season-46 legal kill muzzleloading season-34 Mountain lion predation-3 legal kill first rifle season-46 Disappear second rifle season-41 Illegal kill first rifle season-49 legal kill first rifle season-46 Unknown-suspect malnutrition-31 legal kill first rifle season-46 Mountain lion predation-3 legal kill first rifle season-46 Mountain lion predation-3 legal kill muzzleloading season-34 legal kill archery season-33 ~ounding loss archery season-52 legal kill archery season-33 Unknown predator-5 legal kill first rifle season-46 Unknown-suspect predation-30 ~ounding loss rifle season-25 legal kill archery season-33 legal kill muzzleloading season-34 legal kill first rifle season-46 legal kill muzzleloading season-34 �258 Apeendix I. (continued) Frequency 10/ Tral;! Year Captured Site Zone 174.491/95 lB e 174.500/95 e lB 174.520/95 KR B 174.520/96 A GM 174.551/95 \Ie G 174.560/95 A Bl 174.609/95 GB B 174.629/95 E MD 174.639/95 MD E 174.650/95 HG E 174.660/96 B SR 174.679/95 E HG 174.689/95 D AP 174.689/96 G EW 174.700/95 0 AP 174.710/95 0 AP 174.719/95 VM 0 174.729/95 D VM 174.740/95 D VM 174.750/95 D VM 174.760/95 VM D 174.770/95 e W 174.780/95 e W 174.789/95 e W 174.789/96 G EW 174.800/96 E BG 174.809/95 W C 174.820/95 W C 174.830/95 lB e 174.841/95 e LB 174.851/95 e lB 174.861/95 KR B 174.870/95 e lB 174.880/95 B KR 174.889/95 G we 174.899/95 Bl A 174.910/96 LM D 174.929/96 D lM 174.959/96 lM 0 174.998/96 BG E 175.059/96 BC C 175.130/96 GM A 175.139/96 GM A 175.149/96 SR B 175.160/96 c Be 175.170/96 0 lM 175.180/96 E RG 175.189/96 RG E 175.199/96 EW G 175.209/96 MS F Oeath Sex F F F F F F F M M M M M M M M M M M M M M M M M M M M M M M M M M M M M F F F F F F F F M M M M M M Age 2 e Y e 3 Y e 2 2 3 3 Y e e 2 2 4 Y 2 2 3 2 2 e 2 e Y 2 2 3 12 Y 2 2 2 2 e 2 2 2 C C 2 2 2 C 3 2 2 2 Oate 10-Sep-97 16-Apr-96 21-Sep-96 25-Apr-97 14-Oct-98 14-May-97 15-Apr-96 11-Oct-97 21-0ct-97 14-Sep-98 07-Nov-99 13-Nov-96 05-Feb-96 14-May-97 16-Oct-97 15-0ct-97 10-0ct-99 17-0ct-96 08-Nov-97 05-Nov-97 15-Sep-98 16-Oct-97 11-0ct-97 22-May-96 28-0ct-98 09-Apr-97 22-Apr-97 14-Oct-97 13-0ct-97 12-Sep-98 22-0ct-97 31-0ct-96 30-0ct-97 11-0ct-97 20-0ct-97 30-0ct-97 27-Feb-97 29-Oct-98 29-0ct-98 21-0ct-98 13-Feb-97 24-Mar-97 28-0ct-98 19-0ct-98 17-Sep-98 26-Mar-97 30-Nov-99 13-0ct-98 29-0ct-98 24-0ct-98 Cause of Oeath & Code Number Wounding loss archery season-52 Unknown-suspect predation-30 legal kilL IlUzzleloading season-34 Mountain lion predation-3 legal kill first rifle season-46 Illegal kill-7 Unknown-suspect predation-30 legal kill first rifle season-46 legal kill second rifle season-47 legal kill IlUzzleloading season-34 Wounding loss third rifle season-45 Illegal kill second rifle season-50 Mountain lion predation-3 Unknown-suspect malnutrition-31 Wounding loss first rifle season-43 legal kill first rifle season-46 Legal kill first rifle season-46 Illegal kill first rifle season-49 Legal kill third rifle season-48 legal kill third rifle season-48 Legal kill archery season-33 Disapeear first rifle season-40 legal kill first rifle season-46 Unknown-suspect predation-30 Wounding loss second rifle season-44 Mountain lion predation-3 Accident, fell-10 legal kill first rifle season-46 legal kill first rifle season-46 Legal kill IlUzzleloading season-34 Illegal kill second rifle season-50 Illegal k ill second rifle season-50 Wounding loss second rifle season-44 legal kill first rifle season-46 legal kill second rifle season-47 Wounding loss second rifle season-44 Unknown-suspect predation-30 Disapeear second rifle season-41 Wounding loss second rifle season-44 legal kill second rifle season-47 Unknown-11 Mountain lion predation-3 legal kill second rifle season-47 legal kill second rifle season-47 legal kill IlUzzleloading season-34 Unknown-suspect malnutrition-31 Wounding loss rifle season-25 . legal kill first rifle season-46 Oisapeear second rifle season-41 Legal kill second rifle season-47 � https://cpw.cvlcollections.org/files/original/a2fae5d7a4eef8e55dda4123ea65bc20.pdf b5871955c8ba4bae18ec79e90191b315 PDF Text Text 207 Colorado Division of Wildlife Wildlife Research Report July 1998 Job Progress Report State of Colorado Cost center 3430 Project No. W-153-R-11 Mammals Program Project No. 3002 Elk conservation Work Program No. - - ~ L - - - - - - - Elk Movements in Response to Early season Hunting in the White River Area Period Covered: July 1, 1997 - June 30, 1998 Author: Mary Conner Personnel: Dave Freddy, Gary White, Rick Kahn, John Ellenberger, Jim Lipscomb, Bruce Gill, Jeff Madison ABSTRACT The effects of early hunting seasons on the movements and distribution of elk remains controversial. Those who hunt the regular rifle seasons claim that archery and muzzle loading rifle seasons drive elk off of public lands and onto private land refuges before the regular rifle season begins, making them unavailable to most who hunt in the regular rifle season.· Landowners complain that elk responses to early hunting seasons cause elk to spend longer periods on private property at the expense of the landowner. Beginning in 1996, the Colorado Division of Wildlife sponsored a management experiment designed to test the effects of early hunting on the movements and distribution of elk residing in and around the White River plateau. Four primary conclusions emerged from a preliminary analysis of 2 years of data: (1) In both 1996 and 1997, between 22-37% of the elk were on private land before mid-July. By mid-October, between 65-85% of the elk were on private land. The proportion of elk on private land was positively correlated with date; there was a 28-60% increase of elk on private land during the study period. (2-) The mean date of movement from public (non-refuge) to private land was not significantly different between treatments in the crossover analysis. Mean date of movement was different between treatments for 1997 but not for 1996. Mean date of movement was different between treatments for elk on the north half of the study area, but not found different for elk on the south half. The difference between late minus early date of movement was 6 days for elk receiving both treatments. �208 (3) Differences between elk on the north area and elk on the south area showed in responses of mean date of movement and proportion of elk on refuge areas. Mean date of movement between treatments was significantly different for elk on the north area but not for elk on the south area. The difference in proportion of elk on refuge areas between treatments was greater for elk on the north area than for elk on the south area. (4) Although all elk were captured on national forest land, some elk immediately moved to private land in 1996, and some elk were never located on public land during 1997 flights. "Manageable" elk include only those elk that are on public land within a month of early-season opening date. "Unmanageable• elk are those not on public land within a month of early-season opening date and are, therefore, not likely to be affected by changes in hunting patterns on public land. Between 39-64% of elk on public land one month before early-season opening date (unmanageable elk) moved to private land, and the mean date of movement including opening date in 3 of 4 cases. However, the elk that moved represent 31-41% of the entire herd (unmanageable and manageable elk). Therefore, any management decisions regarding hunter numbers is likely to affect at most 41%, and possibly less, of the White River elk. �209 ELK MOVEMENT IN RESPONSE TO EARLY SEASON IIUHTING IN THE WHITE RIVER AREA Mary Conner P.H. OBJECTIVE Test whether early-season hunting season causes movement of elk from areas of heavy hunting pressure to areas of lighter hunting pressure. SEGMENT OBJECTIVES 1. Estimate the mean date of movement of elk from non-refuge to refuge areas, and determine whether the timing is equivalent to the opening of archery season. 2. Evaluate alternative hypotheses about causes of elk movement such as livestock grazing, woodcutting, recreationalists, weather, or forage quality. 3. Publish analyses of elk movement in response to archery hunting in peerreviewed scientific journals. Preparation of manuscripts will begin during 1997-98. METHODS ARD MATERIALS Methods and materials used in this investigation are described in the detailed study plan reported in Conner 1996. RESULTS ARD DISCUSSIQH Preliminary analyses, results and conclusions of the 2-year study of the effects of early hunting seasons on the movements and distribution of elk in the White River area are summarized in Attachment 1 to this report. LITERATURE CITED Conner, M. 1996. Elk movements in response to early season hunting in the White River area. Colorado Division of Wildlife. Wildlife Research Report. July 1996 Part 1. Pp. 43-86. Prepared by Graduate Research Assistant ��211 Attachment 1 ELK MOVEMENTS IN RESPONSE TO EARLY-SEASON HUNTING IN 11IE WlilTE RIVER AREA: 1997 PRELIMINARY RESULTS Mary Conner March 31, 1998 EXECUTIVE SUMMARY This year was the second and final year of an experiment designed to determine if elk movements in the White River area were caused by early-season (archery and muzzleloading) hunting. The goal of this report is to present 1997 results in a timely manner. This interim report provides information for those involved with the White River elk to make relevant management decisions. Analyses that are more extensive will be included in my dissertation, which will be completed late spring or early summer 1998. Three main conclusions emerged from preliminary analyses of data from both years: (1) the proportion of elk located on private land (refuge) increased between July 15th - October 13th, (2) the timing of these movements weakly corresponded to the opening date of archery season, (3) elk on the north area had a stronger response to changes in archery-season opening date than elk on the south area, and (4) it is not clear how many elk in the White River area would be affected by management actions such as a reduction of archery hunting licenses. Conclusion (1): In both 1996 and 1997 between 22-37% of the elk were on private land before midJuly. By mid-October, between 65-85% of the elk were on private land. The proportion of elk on private land was positively correlated with date; there was a 28-60% increase of elk on private land during the study period. Conclusion (2): The mean date of movement from public (non-refuge) to private land was not significantly different between treatments in the crossover analysis. Mean date of movement was different between treatments for 1997 but not for 1996. Mean date of movement was different between treatments for elk on the north half of the study area but not found different for the south half elk. The difference between late minus early date of movement was 6 days for elk receiving both treatments. Conclusion (3): Differences between elk on the north area and elk on the south area showed in responses of mean date of movement and proportion of elk on refuge areas. Mean date of movement between treatments was significantly different for elk on the north area but not for elk on the south area. The difference in proportion of elk on refuge areas between treatments was greater for elk on the north area than for elk on the south area. Conclusion (4): Although all elk were captured on national forest land, some elk immediately moved to private land in 1996, and some elk were never located on public land during the 1997 flights. "Manageable" elk include only those elk that are on public land within a month of early-season opening date. "Unmanageable" elk are those elk not on public land within a month of early-season opening date and t rtherefore, not likely to be affected by changes in hunting patterns on public land. Between 39-64% of elk on public land a month before early-season opening date (manageable elk) moved to private land, and the mean date of movement included opening date in 3 of 4 cases. However, the elk that moved represent 3141 % of the entire herd (manageable and unmanageable elk). Therefore, any management decisions regarding hunter nwnbers are likely to affect at most 41 %, and possibly less, of the White River elk. INTRODUCTION As with the 1996 analyses, the treatment was difference in opening date of archery season and the primary response variables were mean date of movement and proportion of elk that move from public to private land I repeated all 1996 analyses for area x year to evaluate effects of archery hunting on elk that were on public land within a month of opening date and to compare to pilot study results (Appendix A: Methods). Because there are now 2 years of data, overall analyses including year and area effects were done to evaluate effects of changes in opening date of archery hunting. To evaluate treatment effects on mean date of movement, I did the crossover analysis described in my study plan (Appendix A: Methods). To evaluate �212 treatment effects on proportion of elk moving to refuge areas, I included area and year effects in the logistic regression model used in 1996 to predict the proportion of elk on refuge areas (Appendix A: Methods). Data from the entire 3-month study period were used in the crossover and logistic regression analyses. PRELIMINARY RESULTS Area x Year Analyses The timing of elk movements from non-refuge to refuge areas on the early- and late-opening treatments is presented for 1996 and 1997 (Fig. IA, B). Only elk moving within a month of opening date were used for within-treatment analyses to represent effects on manageable elk.Figure 1. Histograms of number of elk moving from public (non-refuge) to private (refuge) land with respect to archery-season opening date on the early-opening (A) and late opening (B) treatments; White River elk herd, 1996 and 1997. B A 1996 south side 1996 north side 4 4 Mun date of movemant8/26 CD Hlstortcal opening data .E 3 ~ ~3 E E .:.=: ='GI 2 ::, 0 !E 1 ::, 1 z z 0 0 IIIIIIIIIIIIIIIII Early opening: 8/24 § 1997 north side Mun Date of MovenwntB/18 1997 south side Hlstortcal opening data 4-r-----------------, 3 .:.=: 'ii 2 !E 1 0 1 ::, z Historical opening date CD C 1 ::, IIII~ IIIIIIIIIIII Late opening: 9/14 4 -r-----------------, !E Historical opening date ~2 0 !E Mean Date of Movement: 1115 CD C Mean Data of Movemant 916 z ~ ; IIt IIIIIIIIIIII Early opening: 8/23 I ; IIi IIIIIII IIIIIII Late opening 9/13 For elk moving from non-refuge to refuge areas, mean date of movement was tested under the null hypothesis that the mean date of movement was not different from opening date (Table 1). Between 39-64% of elk on non-refuge areas one month before opening date moved to refuge areas within a month of opening date (Table 2). The probability that classifications are independent of opening date (Table 2) evaluated the effect archery hunting has on elk location by testing whether movements between classifications occurred independently before and after opening of archery hunting. �213 Table 1. Mean dates of movement of elk moving from non-refuge to refuge areas. opening dates of archery season, and P-values from a t-test for differences between opening date and mean date of movement for earlyand late-opening treatments; White River study area. 1996 and 1997. Early opening Late opening Mean date of movement 95% Confidence interval 26 August 17 August- 18 August 11 August- 5 September 30 August- 6 September 26 August- Opening date of early-season 24 August 23 August 14 September 13 September P = 0.578 P = 0.178 P = 0.014 P = 0.131 - Ho: d _ = onenin2' date Table 2. Movements ofradio-collared elk between refuge and non-refuge areas with respect to opening date. early-and late opening treatments; White River study area. 1996 and 1997. Movement Combinations: Early opening Late opening Pre-opening -> post-opening YEAR 1 - South YEAR 2 - North YEAR 1 - North YEAR 2 - South Refuge -> refuge 11 8 4 11 Refuge -> non-refuge 5 0 2 2 Non-refuge-> refuge 12 18 15 14 Non-refuge -> non-refuge 19 10 20 15 Sample size 47 36 41 42 P= 0.047 P= 0.280 P=0.027 Probability movement combinations P=0.051 Overall Analyses on Mean Date of Movement The time frame for between-treatment analyses is the 3-month study period of July 15 th - October th 13 . Results from the crossover analysis on mean date of movement indicated no area (P = 0.138) or year effect (P = 0.644); however, there was a significant random effect of elk in area (P = 0.025). There was not a significant treatment effect (P = 0.159) on the mean date of elk movement from non-refuge to refuge areas (Table 3). Sequence effect refers to the fact that each elk received either a late-early or early-late sequence of treatments. Elk on the north half received the late-early treatment sequence. while elk on the south half received the early-late treatment sequence. Area effects and sequence effects were confounded. That is. it is impossible to determine if a sequence effect is really a difference due to the fact that the north side is different from the south side because of location of private-land refuges, topogQlphy, etc., or due to the fact that elk getting a late-opening treatment the first year followed by an early-opening treatment the second year will behave differently than elk getting the early-late sequence. I will refer to an area effect for clarity in the remainder of the paper. There appeared to be a treatment x year and treatment x area interaction in the crossover analysis that clouded the results. In a post hoc analysis, I found that there was a difference in mean date of movement between treatments for elk moving from non-refuge to refuge areas in 1997 but not in 1996 (Fig. 2 and Table 4). �214 Table 3. Least square mean differences in days between mean dates of movement for effects specified in the crossover analysis with their 95% confidence intervals; White River elk herd 1996 and 1997. No. of days difference No. of days difference No. of days difference between areas: ok between treatments: a; between years: Pi 7±9 2±7 6±8 Figure 2. Mean date of movement, 95% confidence interval, and archery opening date for elk on early-and late opening treatments, 1996 (A) and 1997 (B), White River elk herd. A 1996 Early open: South Mean date of movement: Late open: North Aug 26 Aug 31 Sept14 Aug 24 Opening Date: B 1997 Late open: South Early open: North Mean date of movement: Aug 18 Sept3 ~ IIII Aug 23 Opening Date: I Sept 13 Table 4. Differences between opening dates and mean dates of movement between early- and late-treatments, 95% confidence interval, and P-value for test of differences between mean opening dates 1996 and 1997; White River elk herd Year Days between .. 1996 1997 . dates 20 20 Days between mean p dates of movement 5± 11 16± 12 0.371 0.010 �215 I also did a post-hoc analysis to evaluate the treabnent effect by area, as it appeared that elk on the north area were moving to private land more readily than elk on the south area. There was a difference in mean date of movement between treabnents for elk on the north area but not for elk on the south area (Table 5). Table.5. Differences between opening dates and mean dates of movement between early- and late-treabnents for north- and south-side elk, 95% confidence interval, and P-value for test of differences between mean opening dates; White River elk herd 1996 and 1997. Area Days between Days between mean dates p North side ooenine: dates 20 of movement 13 ± 10 0.009 20 8 ± 14 0.267 (late-early treabnent) South side (earlv-late treabnent) I then analyzed date of movement only for individual elk that received both the early and late treabnent. Only 16 elk received both treabnents because some elk were not on public land for one year and some changed side of study area between years. For the elk getting both treabnents, I subtracted their early date of movement from their late date of movement to account for the fact that this was a repeated measurement on the elk. I tested the hull hypothesis that: Ho: Difference in date of movement for late treabnent • early treabnent = 0 Ha: Difference in date of movement for late treabnent - early treabnent > 0 . The difference in mean date of movement was 6 ± 7 days (p = 0.065). The distribution of differences in date of movement was not normal (Fig. 3); 12 of the 16 elk had a positive difference for date of movement. If movements were random, then there would be an equal number of positive and neg~tive differences. There was a greater count of positive differences than expected if date of movement was random with respect to treabnent (p = 0.038). Overall Analyses on Proportion of elk on Refuge Akaike's Information Criteria (AIC) was used to choose a model to predict the proportion of elk on refuge areas from a combination oftreabnent, area, year, and date effects (Fig. 4A, B). All models accounted for the repeated measurements taken on individual elk between treabnents. For the lowest AIC model, the logistic regression parameters for treabnent (P = 0.032), area (P < 0.001), year (P < 0.001), date (P < 0.001), and area x date (P < 0.00 I) were significantly different from z.ero. When these factors were controlled for there was a treabnent x date (P < 0.082) effect (Table 6). The borderline treabnent x date effect indicated that the slope, or rate of change in proportion of elk on refuge areas, was different for the 2 treabnent areas (Fig. 5). The significance of the area~ date parameter indicates that the slope, or rate of change in proportion of elk on refuge areas, was greater for elk on the north side (late-early sequence) compared to elk on the south side (early-late sequence; Fig. 6). Elk on the north area showed a greater response in proportion of elk on refuge area by treabnent compared to elk on the south area (Fig. 7A, B). The fitted proportion of elk on refuge for each area x year category and the corresponding predicted proportion of elk on refuge for the opposite treabnent was calculated to evaluate the effect of changing archery season opening date (Fig 8A, B, C,D). �216 Figure 3. Distribution of differences in date of movement between late and early treatments for elk experiencing both treatments; White River elk herd 1996 and 1997. ntlQUDiCT 7 6 4 3 2 1 -35 -25 -15 -5 5 15 Di t f•r•nce aovea•nt d.a.t• 25 35 (Late-early) Figure 4. Proportion of elk on refuge areas with respect to archery-season opening date, on early- and lateopening treatments for 1996 (A) and 1997 (B); White River elk herd. 1996 A Q) - 0.9 O> :::, 0.8 Q) L. C 0 ~ 0.7 0.6 Q) 0.5 0 0.4 ... , ,. ... C 0 :.::; 0L. 0.2 0L. 0.1 a. Cl.. Late treatment North side 0.0 (D 0) io ..r-- -- ~ N C:! r-- (D e? 0) C:! r-- ~ !Q a:) ~ (D ~ (D 0) (D ~ (D 0) (D 0) (D 0) -- t; ..... C:! ~ -- ..... ----' date: Early opening date: Late opening N ..... CD e? 0) (0 CD CD Aug24 O> e? 0) O> (0 O> Sept 14 C'5 C:! O> 0 (') O> 0 �217 Figure 4. (Continued) 1997 B Cl) 0.9 - . - - - - - - - - - - - - - - - - - - - - - - - - - , g> 0.8 e 0.1 ~ oC 0.6 ~ "a; 0.5 O C 0.4 j 0.3 [ 0.2 0 (l'. 0.1 ... ... ,, ,, 'Early treatment North side 0.0 +---r-,........,,.......,.---,---,---,--,--.-it-r--.,....-..--,........,--,--t,---,---,---,-~-,--....-....--1 i ~ .... ~ ~ ~.... Si....;::: co co ~ .... ~ ~ .... e.... 55 Early opening date: Aug23 .... I .... ~ .... a; .... ..,~ e.... ~ c3 .... I.... Late opening date: Sept 13 Table 6. Parameter estimates for logistic regression model used to predict the proportion of elk on refuge areas; White River elk herd 1996 and 1997. INTERCEPT SEQUENCE TREAT JULDAY YR JULDAY*TREAT -7.2244 4.2030 1.2382 0.0305 -0.2395 -0.0041 0.5082 0.5757 0.5684 0.0021 0.0606 0.0023 0.0001 0.0001 0.0323 0.0001 0.0001 0.0821 �218 Figure 5. Treatment r date effect for elk on early- and late-opening treatments 1996 and 1997; White River elk herd. Treatment Effect Q) C) .2 ! C 0 ~ Q) '5 C 0 :e0 a. e a. 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 •• • •• Early opening ~ • • • ••• • • I • • • • • • • • • . • • ~ • • • • • • • • • ••• Late opening • Early opening Late opening ■ 7/8 7/22 8/5 8/19 9/2 Date 9/16 9/30 10/14 Figure 6. Area x date effect for elk given the early-late or the late-early treatment sequence during 1996 and 1997; White River elk herd. Area Effect Q) C) :::, ! C 0 ~ Q) 0 C 0 :e0 a. e a. 0.9 0.8 early-late treatment 0.7 0.6 0.5 0.4 0.3 • •• • •• South side elk • • • • • • 0.2 0.1 0 ■. late-early treatment 7/8 7/22 8/5 8/19 9/2 Date ■ • South-side elk • North-side elk 9/16 9/30 10/14 �219 Figure 7. Fitted curves of proportion of elk on refuge areas by treatment from logistic regression analysis for elk on the north area (A) and elk on the south area (B); White River elk herd 1996 and 1997. (A) North Area: Elk movement to refuge by treatment 0.9 a, C> 0.8 .2 0.7 C 0.6 e 0 -a; ~ - 0.5 0 0.4 :e0 0.3 e 0.2 C 0 C. a. -North ear1y opening -fitted cutve 0.1 - North late opening - fitted cuive 0.0 7/8 7/28 9/6. 8/17 9/26 10/16 Date (B) South Area: Elk movement to refuge by treatment 0.9 a, C> 0.8 .2 0.7 C 0.6 ! ~- 0 ~ - 0.5 :e0 0.3 e 0.2 a, 0 C 0 C. a. ~ 0.4 -South ear1y opening- fitted cuive 0.1 - South late opening - fitted cutve 0.0 7/8 7/28 9/6 ,. 8/17 Date 9/26 10/16 �220 Figure 8. Fitted curves and predicted curves for reversed treatments, early treatment on south area 1996 (A), late treatment on north area 1996 (B), early treatment on north area 1997 (C), and late treatment on south area 1997 (D); White River elk herd. (A) (C) South Side - early treatmert 1996 North Side - early treatmert 1997 0.9 0.9 ., 0.8 a, 0.8 ,2 0.7 §, 0.7 C 0.6 l! 0.6 .>t! 0.5 .>t! I!! 0 .; 0 15 0.5 0.4 a; '5 0.4 .Q t: 0.3 -~ 0.3 e 0.2 C 8. n. J l===~I 0.1 0.0 718 7/28 8117 9/6 9/26 0.2 1-Filled early - Predicted late 0.1 0.0 718 10/16 7/28 8/17 Date C 0 .>t! ., 0.8 0.7 ,2 0.6 C 1-Filled late I a, I!! 0 0.7 0.6 .>t! 0.5 0 C 0.4 .Q t: 0.3 8. 0.9 0.8 .; e n. 10/16 South Side - late treatment 1997 0.9 I!! 9/26 (D) North Side - late treatment 1996 ,2 9/6 Date (B) .,a, I .; 0 C .Q t: 8. 0.2 e n. -Fitted late 0.1 - Predicted earty 0.0 0.5 0.4 0.3 0.2 0.1 - Predicted ea~y 0.0 718 7/28 8117 9/6 9/26 Date 10/16 718 7128 8/17 9/6 9/26 10/16 Date DISCUSSION AND FUTURE WORK Although all elk were captured on national forest land, some elk moved to private land early in the summer before the ±1 month time frame used for within-half analyses, and some elk were never located on public land during the 1997 flights. "Manageable" elk are the proportion of the herd on public land within a month of early-season opening date. "Unmanageable" elk are the proportion of the herd not on public land within a month of early-season opening date and are, therefore, not likely to be affected by changes in hunting patterns on public land. In determining effects of archery hunting activity on elk movement, two types of effects were examined; the effects on manageable elk, and the effects on total herd (manageable plus unmanageable elk). Between 39-64% of the manageable elk moved to private land within a month of archery-season opening date. For the manageable elk, confidence intervals on mean date of movement included opening date in 3 of 4 cases, and movements of elk between refuge and non-refuge areas was not independent of opening date in 3 of 4 cases. North-side elk experiencing the late-opening treatment were the exception in the tests, possibly due to several early movers that attenuated the treatment effect. In general, it appears that early season hunting activity has an effect on the manageable elk. �221 The effects of the manipulation of archery-season opening date on elk movements on the entire herd was evaluated by responses in mean date of movement and proportion of elk on refuge. The strongest test is the crossover analysis because it controls for external sources of variation when testing for treatment effects. The lack of a treatment effect in the crossover analysis is evidence that early-season hunting is not affecting the timing of elk movements from refuge to non-refuge areas. However, because there appeared to be interactions between treatment x area and treatment x year, post hoc analyses were done which reveled several pieces of evidence indicating that early-season hunting activity altered elk movements: (I) mean date of movement was different between treatments for elk on the north area, (2) mean date of movement was different between treatments for 1997, (3) for elk getting late and early treatments, there was a 6 day difference in late minus early date of movement, and (4) there was a significantly greater count of positive late minus early date of movement differences than expected if date of movement was random with respect to treatment. Also, from the logistic regression analysis of proportion of elk on refuge, the rate at which elk move from public land to private land refuge areas was borderline different for early- and late-opening treatments. Part of the ambiguity in evaluating the effects of archery hunting on elk movements may result from differences in movement patterns of elk on the north side versus elk on the south side of the study area. The steeper slope, or faster gain in proportion of elk on refuge areas over time for the north-side elk (Fig. 6) suggests that north-side elk move to private land more readily than south-side elk. This was surprising as south-side elk, which experienced the early-opening treatment the first year, were expected to move more readily the second year. Similarly, north-side elk, which experienced the late-opening treatment the first year would be taken by surprise by an early opening the second year, and were expected to move less readily. In addition to moving more readily, elk on the north area were more responsive to changes in opening of archery season. The timing of the movements of north-side elk was affected by archery season opening date, as indicated by the significant difference in mean date of movement betw~n treatments. South-side elk did not have as wide a spread in their mean date of movement between treatments and there was not a significant difference in mean date of movement between treatments. Also, the proportion of elk on refuge was greater during the early treatment compared to the late treatment for elk on the north area (Fig. 7A). For elk on the south area, the proportion of elk on refuges was similar for early and late treatments (Fig. 7B). It may be that the north-side elk move more readily due to the presence of a large private-land refuge adjacent to national forest land. Between 39-64% of the manageable elk moved to private land within a month of early-season opening date. However, the elk that moved represent 31-41 % of the entire herd. Hence, any management decisions regarding hunter numbers are likely to affect at most 41 %, and possibly less, of the White River elk. The question remains as to what percentage of the manageable elk would still move if hunting were reduced or eliminated. These results are not complete; I still need to analyre 1997 hunter survey data, test for elk avoidance of livestock, and calculate daily distances moved and elevational shifts with respect to early-season hunting activity. Additionally, I need to reclassify refuge areas; there needs to be a third land classification for habitat refuges. Habitat refuges will be defined as topographically steep areas where changes in elevation per linear distance are greater than some threshold (to be determined). Elk locations will be reclassified as public, private, or habitat refuge, and reanalyred in a multinomial model. Although it appears that north-side elk move more readily than south-side elk, results may change when habitat refuge areas are included in analyses. South-side elk may not need to move to private land because they move into habitat refuges. It may be that elk with available habitat refuges are of little management concern with respect to private-land problems. These results will be incorporated and discussed in my dissertation. ��223 APPENDIX A Methods METHODS The study area was split roughly east-west by the White River and North Fork of the White River into 2 halves for application of early- or late-opening treatments during 1996 and 1997. Game Management Unit (GMU) 12 and part of GMUs 23 and 24 comprised the north half, while GMU 33 and part of GMUs 23 and 24 comprised the south half. There was a 3-week difference in opening dates; archery hunting opened early, August 24th, on the south half, and late, September 14th, on the north half during 1996. Treatments were reversed in 1997; archery hunting opened late, September 13th on the north half, and early, August 23 rd , on the south half. Although data were collected on each elk from July 15 - October 13 for both years, withintreatment analyses were restricted to locations collected within a month of opening date (i.e. July 23 September 24 for early-opening treatment, and August 13 - October 13 for the late-opening treatment). The ±I-month interval allows evaluation of treatment effect on manageable elk and allowed for comparison to pilot study results. All other analyses were done over the 3-month study period for valid comparison of between-treatments effects on the entire herd (manageable plus unmanageable elk). The primary response variables were mean date of movement and proportion of elk that change classification (non-refuge ->refuge/ public ->private). All locations were labeled as either refuge or nonrefuge based on land ownership or management, and not on topographical considerations. That is, the Flat Tops Wilderness Area and all private land were labeled as refuge, while all National Forest, State Wildlife, and BLM lands were labeled as non-refuge. Area x Year Analyses This is a geographically nested design. with 2 levels of analysis. One level of analysis is within half, where the treatment is location on non-refuge or refuge areas. The time frame for within treatment analyses is ±1 month of opening date. The ±1 month interval allows evaluation of treatment effect on manageable elk. If hunting has no effect on elk movements, then elk movements from hunted non-refuge areas should not correlate with opening date. The primary hypothesis tested were: 1. Ho: Mean date of movement from non-refuge to refuge areas = opening date. Ha: Mean date of movement from non-refuge to refuge areas * opening date. 2. Ho: Location of elk on non-refuge and refuge areas was independent of the opening of archery hunting. Ha: Location of elk on non-refuge and refuge areas was not independent of the opening of archery hunting. These were the same hypotheses tested for the 1992,..1995 pilot study data; they were tested separately for both north and south halves of the study area. Differences in time of movement were tested with a one sample t-test. •Hypothesis 2 evaluated the effect archery hunting had on elk movements by testing whether changes in classifications (refuge->refuge, refuge->non-refuge, non-refuge->refuge, and non-refuge->nonrefuge) were independent of the presence of archery hunting. If archery hunting had no effect, then there should be no difference in elk locations before or after the opening of archery season. A chi-square test was used to evaluate if movement of elk between non-refuge and refuge areas was independent of being before or after opening date. �224 Overall Analyses on Mean Date of Movement The experimental unit is north or south half of the study area, the sampling unit is individual elk, and the treatment is early or late opening of archery season in this 2-period crossover design. The time frame for all overall analyses is the 3-month study period of July 15th -October 15th; the 3-month interval allows valid comparison between treatments for the entire herd (manageable and unmanageable elk). The crossover analysis used data from both years of the study to test the effects of early-season hunting on the mean date of elk movement from non-refuge to refuge areas (see methods: Elk movements in response to early-season hunting in the White River area: study plan). If hunting had no effect on elk movements, then there should be no difference between the mean date of movement to refuge areas between treatments. To test for a treatment effect on mean date of movement, the area (sequence) in which the treatments were applied,-the year, and the random effect of the individual sample elk must be accounted for. The corresponding analysis of variance model is: YiJ1d = µ + O.t + "1(1c) + (X; + A + &i11d, where: yij1d µ = date of movement for the fl' elk, in area k, treatment i, and year j = overall mean date of movement (for all elk, both years) bi = fixed effects due to area k (either north or south) = random effects due individual elk / in area k = fixed effects due to treatment i (early or late opening) = fixed effects due to year j (year I or year 2). tri(k) a; /3j Specific questions tested in the crossover ANOVA model were: 1. Mean date of movement for north-side elk (early-late treatment sequence)= mean date of movement for south-side elk (late-early treatment sequence), Ok= 0. 2. Elk randomly chosen from the south side (early-late sequence) were the same as elk randomly chosen from the north side (late-early sequence); 7rJ(k)-= 0. 3. Mean date of elk movement for early- opening treatment = mean date of elk movement for late opening treatment; a; = 0. 4. Mean date of elk movement of elk for year I of experiment= mean date of elk movement of elk for year 2; Pj = 0. Hypothesis 3 is the experimental hypothesis; hypotheses 1, 2, and 4 are needed to account for other sources of variation. Overall Analyses on Proportion of elk on Refuge The logistic regression model used in 1996 to test for treatment and date effects on the proportion of elk located on refuge areas (p): logit(p) =Po + P1(date)+ P2 (treatment)+ P3 (treatmentx.date) + e, was expanded to include area and year effects. Various combinations of treatment, area, year, and date were modeled. Akaike's Information Criterion (AIC) was used to select the best model given the precision and bias tradeoff. The DSCALE option was used in this analysis to account for the repeated measures on the individual elk and allow for inference to the entire herd. The model with the lowest AIC was: logit(p) = Po + P1 (area)+ P2 (year)+ p 3 (date)+ P4 (treatmemt) + Ps (treatment xdate) + P6 (areaxdate) + e. �225 From this model the following hypotheses were tested: 1: 2: 3: 4: 5: 6: Proportion of elk on refuges was independent of the area, given date, treatment, and year differences were controlled for; p, = 0. Proportion of elk on refuge areas was independent of year, given date, treatment, and area differences were controlled for; P2 = 0. Proportion of elk on refuge areas was independent of date, given area, year, and treatment differences were controlled for; P1 = 0. Proportion of elk on refuge areas was independent of treatment, given area, year, and date · differences were controlled for; p,;= 0. Early and late treatment had the same slope, that is, elk moved from non-refuge to refuge areas at the same rate on early- and late-treatment areas; PJ= 0. The south and north side had the same slope, that is, elk moved to refuge areas at the same rate whether they were on the south side (early-late treatment sequence) or on the north side (late-early treatment sequence); p6= 0. Hypotheses 5 is the experimental hypotheses. Hypothesis 5 was a test of proportion on refuge and timing of movement to refuge; that is, if archery hunting had no effect, then elk would be expected to move to refuge areas at the same rate for early- or late-opening treatments. A likelihood ratio test of the area x date effect (/J,) was an overall test of archery effects, given that elk would move the same on the 2 areas otherwise. Hypotheses 1-4 allow for accounting of additional sources of variation. Hypothesis 6 tested whether the rate of movement to refuge areas was different by area. This test was more biologically interesting than testing for a difference in the proportion of elk on refuge between areas. For example, a difference between areas may only indicate that more elk happened to begin on refuge areas on the north or south side, which is irrelevant to archery hunting activity. However, if elk on the south side receiving the early-treatment the first year were more wary the second year, then the south side elk should show a greater rate of movement than the north side (late-early sequence) elk. This effect was tested by the inclusion of the area x date interaction term in the logistic model. ��191 Colorado Division of Wildlife Wildlife Research Report July 1999 JOB FINAL REPORT State of _ _ _ _--=C...,o=lo=r=ad=o,.___ _ _ __ Project No. ------=-W_,__-..:;.1=53=--=R=--=-=12=-----Work Package _~3-0~02~-----Task No. 2 ------'------------- Cost Center 3430 Mammals Research Elk Conservation Elk Movements in Response to Early-season Hunting in the White River Area • Period Covered: July 1, 1998 - June 30, 1999 Author: M. M. Conner Personnel: G. White, D. Freddy, R Kahn, J. Maddison, J. Ellenberger ABSTRACT Understanding causes of elk movement to private land is important to wildlife managers who use hunting on public land as a population management tool. In the White River area of northwest Colorado, late-summer elk movements onto private lands made it difficult to harvest enough elk to maintain specified population levels. I conducted a 2-year field experiment to determine if early-season hunting (archery and muzzleloading) caused elk movement to private land during late summer. The study area was split into north and south treatment areas: each area received both an early- and lateopening treatment. Early-opening treatment was archery season that opened 1 week earlier than the historical opening, and late-opening treatment was an archery season that opened 2 weeks later, yielding a 23-week difference in opening dates. The 80-88 adult female elk used in this study were captured from random locations throughout the study area I relocated radiocollared elk approximately 2 times per week for a 3-month period surrounding early- and late-opening dates each year. Using a Geographical Information System (GIS) generated map, I classified each elk location as being on public or private land. Elk receiving the late-opening treatment moved 10 days later than elk receiving the early treatment (P = 0.013, one sided ANOVA). Because there appeared to be an interaction between area and treatment, I performed a post hoc analysis separately by area In the north area, elk receiving the late-opening treatment moved 14 days later than elk receiving the early treatment (P + 0.006), one sided t-test), compared to a 6-day difference for elk in the south area (P = 0.218). Approximately twice as many radiocollared elk moved to private land on the north area (44%) compared to the south area (23%). The experimental effect on the number of elk directly influenced by early-season hunting can be estimated by the abrupt increase in the proportion of elk moving to private land when hunting opened. Proportion of elk on private land increased 8-18% at the opening of early season. The greater response of elk to hunter activity on the north area may be due to topographical and migration differences between areas. The south area had densely forested canyons and cliffs, inaccessible to motor vehicles, which provided elk good refuge from hunters on public land. In �192 contrast, the north area offered less shelter on public land, but had refuge in large private-land tracts that bordered the area During the study period, elk moving to private land in the north were just beginning their fall migration, whereas elk in the south area were completing their migration by moving to private land. Elk on the north area that move onto private land do not risk depleting their winter forage, while elk on the south area may degrade their winter range by grazing an extra couple of months on their wintering grounds. Colorado Division of Wildlife.surveyed 34% and 37% of hunters using the study area during a post-hunt telephone survey in 1996 and 1997 to estimate the number of hunters afield per day with a maximum 95% CI of ±150 hunters. Daily use varied between 115-1,130 hunters per treatment area, with corresponding hunter densities of0.006-0.51 hunters/km2. The proportion of elk on private land was not strongly affected by hunter density. • Domestic sheep were accused of causing elk movements to private land during early-season hunting. I collected a location for at least one band of sheep per flight. From this data, I used nonparametric statistics and bootstrapping techniques to calculate the probability that elk avoided sheep at 3 spatial scales. Locations of elk were random with respect to sheep when all elk were used in analyses (P > 0.342), and when only elk within 1 km of a sheep band were used (P > 0.912). Although elk may avoid sheep at <l km, it is unlikely that they avoid sheep at large enou.gh distances to cause widespread elk movements to private land. Although early-season opening may not be the sole cause of elk movements to private land, the experimental results show that early-season hunting caused some elk to move to private land, especially on the north area The increase of elk on private land directly attributable_ to opening of hunting was 8-18%. Hence, the CDOW can reduce, at most, approximately 20% of elk movement to private land by manipulating early-season hunting. However, if elk movements are concentrated in problem areas, then manipulation of opening date may result in a proportionally higher reduction of elk movements in those areas. Managers concerned with reducing elk movement to private land should consider the public and private land refuges available to elk, as well as historical elk migration patterns. �193 ELK MOVEMENT IN RESPONSE TO EARLY-SEASON HUNTING IN THE WHITE RIVER AREA Mary M. Conner P. N. OBJECTIVES I. Test whether early-season hunting causes movement of elk from areas of heavy hunting pressure to areas of lower hunting pressure. SEGMENT NARRATIVES I. Estimate the mean date of movement of elk from non-refuge areas, and determine whether the timing is equivalent to the opening of archery season. 2. Evaluate alternative hypotheses about causes of elk movement such as livestock grazing, woodcutting, recreationalists, weather or forage quality. •• 3. Publish analyses of elk movement in response to archery hunting in peer-reviewed scientific journals. INTRODUCTION Following is the most recent available draft of the Ph.D. Thesis summarizing results of research activities directed toward accomplishment of Program Narrative and Segment Narrative Objectives. ��195 DISSERTATION ELK MOVEMENT IN RESPONSE TO EARLY-SEASON HUNTING IN THE WHITE RIVER AREA, COLORADO Submitted by Mary M. Conner Department of Fishery and Wildlife Biology In partial fulfillment of the requirements for the Degree of Doctor of Philosophy Colorado State University Fort Collins, Colorado Fall 1999 �196 COLORADO STAIB UNIVERSITY June 28, 1999 WE HEREBY RECOMMEND THAT THE DISSERTATION PREPARED UNDER OUR SUPERVISION BY MARY M. CONNER ENTITLED ELK MOVEMENTS IN RESPONSE TO EARLY-SEASON HUNTING IN THE WHITE RIVER AREA BE ACCEPIBD AS FULFILLING IN PART REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHIT.,OSOPHY. Committee on Graduate Work Advisor Department Head �197 ABSTRACT OF DISSERTATION ELK MOVEMENT IN RESPONSE TO EARLY-SEASON HUNTING IN THE WHITE RIVER AREA, COLORADO Understanding causes of elk movement to private land is important to wildlife managers who use hunting on public land as a population management tool. During the 1980s, late-summer elk movements onto private land in the White River area of northwest Colorado made it difficult to harvest enough elk to maintain specified population levels. I conducted a 2-year field experiment to determine if early-season hunting (archery and muzzleloading) caused elk movement to private land during late summer. The study area was split into north and south treatment areas; each area received both an early- and late-opening treatment. Early-opening treatment was an archery season that opened 1 week earlier than the historical opening, and late-opening treatment was an archery season that opened 2 weeks later, yielding a 3-week difference in opening dates. The 80-88 adult female elk used in this study were captured from random locations throughout the study area. I relocated radiocollared elk approximately 2 times per week for a 3-month period surrounding early- and late-opening dates each year. Using a Geographical Information System (GIS) generated map, I classified each elk location as being on public or private land. Elk receiving the late-opening treatment moved 10 days later than elk receiving the early treatment (P = 0.013, one-sided ANOVA). Because elk on the north area appeared to respond differently to treatment than elk on the south area, I performed a post hoc analysis separately by area. In the north area, elk receiving the lateopening treatment moved 14 days later than elk receiving the early treatment (P = 0.006, one-sided I-test), compared to a 6-day difference for elk in the south area (P = 0.218). Approximately twice as many radiocollared elk moved to private land on the north area (44%) compared to the south area (23%). The experimental effect on the number of elk directly influenced by early-season hunting can be estimated by the abrupt increase in the proportion of elk moving to private land when hunting opened. Proportion of elk on private land increased 8-18% at the opening of early season. The greater response of elk to hunter activity on the north area may be due to topographical and migration differences between areas. The south area had densely forested canyons and cliffs, inaccessible to motor vehicles, which provided elk good refuge from hunters on public land. In contrast, the north area offered less shelter on public land, but had refuge in large private-land tracts that bordered the area. During the study period, elk moving to private land in the north area were just beginning their fall migration, whereas elk in the south area were completing their migration by moving to private land. Elk on the north area that move onto private land do not risk depleting their winter forage, while elk on the south area may degrade their winter range by grazing an extra couple months on their wintering grounds. Colorado Division of Wildlife surveyed 34% and 37% of early-season hunters using the study area during a post-hunt telephone survey in 1996 and 1997. The goal was to survey enough hunters to estimate, during the study period, the number of hunters afield per day with a maximum width 95% CI of ±150 hunters. Daily use varied between 115-1, 130 hunters per treatment area, with corresponding hunter densities of 0.06-0.51 hunters/km2• The proportion of elk on private land was not affected by hunter density. Elk in the White River area responded more to the opening of hunting season than to hunter density. It may be that hunter density must surpass or drop below a threshold before elk begin to respond directly to hunters, and hunter densities did not cross any threshold during the study. If hunter density continues to be a suspected cause of increased elk movement to private land, then an experiment manipulating hunter density should be conducted. Domestic sheep also were considered to cause elk movements to pmrate land ,dwing early-season hunting. I collected a location for at least one band of sheep per flight. From these data, I used non-parametric statistics and randomization techniques to calculate the probability that elk avoided sheep at 3 spatial scales. Locations of elk were random with respect to sheep when all elk were used in analyses (P= 0.737 in 1996, P = 0.199 in 1997), when only elk within 5 km of a sheep band were used (P = 0.342 in 1996, P = 0.990 in 1997), and when only elk within I km ofa sheep band were used (P> 0.969 in 1996, P= 0.912 in 1997). Although elk may avoid sheep at �198 < I km, it is unlikely that they avoid sheep at large enough distances to cause widespread elk movements to private land. Although early-season opening may not be the sole cause of elk movements to private land, the experimental results show that early-season hunting caused some elk to move to private land, especially on the north area. The direct effect of hunting season is reflected in the 8-18% increase in proportion of elk on private land that occurred on opening day. Hence, in the White River area, the CDOW can reduce, at most, approximately 20% of elk movement to private land by manipulating early-season hunting. However, if elk movements are concentrated in problem areas, then manipulation of opening date may result in a proportionally higher reduction of elk movements in those areas. The elk not affected by opening of hunting season may be elk for which private land is enroute to their wintering area. Late summer movement onto private land may have become part of a learned migration behavior, changing hunt season opening and/or hunter density may have little effect on these elk. Future experiments and management actions should consider an early hunting season on private land, where problems occur, to retain elk on public land during late summer. Mazy M. Conner Department of Fishery and Wildlife Biology Colorado State University Fort Collins, CO 80523 Fall 1999 �199 TABLE OF CONTENTS ABSTRACT OF DISSERTATION ........................................................ . ACKN"OWLEDGMENTS ............................................................... . INTRODUCTION ..................................................................... . ELK MOVEMENTS IN RESPONSE TO DISTURBANCE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Elk Movement in Response to Human Activities , ........................................... . Elk Movement in Response to Hunting ................................................... . PROJECT OVERVIEW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . LITERATURE CITED . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . FIGURE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CHAPTER 1: ELK MOVEMENT IN RESPONSE TO EARLY-SEASON HUNTING IN NORTHWEST COLORADO ......................................................................... . INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . STUDY AREA ......................................................................... . METIIODS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental design ................................................................. . Data collection .................................................. ;· .................. . Data analysis ...................................................................... . REsULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . MANAGEMENT IMPLICATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . LITERATURE CITED . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CHAPTER 2: EFFECT OF HUNTER DENSITY ON ELK MOVEMENT TO PRIVATE LAND IN NORTHWEST COLORADO ............................................................ . INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . STUDY AREA ......................................................................... . METIIODS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Radio-telemetry Data Collection ........................................................ . Hunter Survey Data Collection ......................................................... . Estimating Number ofHunters ......................................................... . Estimating Number ofHunters Afield per Day . ............................................. . Estimating Mean Number ofDays per Hunter and Hunter-days ................................ . Elk Movement versus Hunter Density .................................................... . RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hunter Survey Data .................................................................. . Elk Movement versus Hunter Density .................................................... . DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . MANAGEMENT IMPLICATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . LITERATURE CITED . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ,. CHAPTER 3: ELK LOCATIONS IN RELATION TO DOMESTIC SHEEP BANDS AND A METHOD TO TESTFORAVOIDANCE OR ATTRACTION BETWEEN ANIMALS ........................... . INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . STUDY AREA ......................................................................... . METHODS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Data Collection ..................................................................... . Data Analysis ...................................................................... . �200 RESULTS DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental Results ................................................................. . Methodology for Determining Attraction and Avoidance ...................................... . MANAGEMENT IMPLICATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . LITERATURE CITED . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . APPENDIX 1: SAMPLE SIZE CALCULATION FOR HUNTER SURVEY ....................... . APPENDIX 2: STANDARD ERRORS FOR DAILY GMUESTIMATES OF HUNTERS AND ATVS AFIELD APPENDIX 3: MAPS OF ELK AND DOMESTIC SHEEP LOCATIONS ON SELECTED FLIGHTS ... ACKNOWLEDGMENTS ri .:1: Funding was provided by Colorado Division of Wildlife (CDOW). Colorado Habitat Partners Program in the Northwest Region provided funding for flights to collect elk locations. I greatly appreciate the financial support. From the CDOW, I would like to extend a special thanks to Dave Freddy for his insightful elk discussions, reviews, and helping to procure financial support. A second special thanks to Dick Bartmann; without his help, organization, funding procurement, and common sense I would have struggled with all the logistics of collaring and tracking over 80 elk. Also from the CDOW, thanks to Jeff Madison for all his help with obtaining funding from Colorado Habitat Partners Program. My major professor, Dr. Gary White, has been quite an inspiration to me. What can I say, he is one of the smartest people I know and I have been lucky to study with such an adept thinker. My committee member, Dr. Ken Burnham, has always been a great help with statistical and mathematical issues that were difficult for me, and never missed a detail in my dissertation work. His help greatly improved my dissertation. Dr. Bill Alldredge always had the common elk-sense aspect and practical questions. Without him, I might have forgotten the most important issues about elk in their natural environment Finally, Dr. Bruce Wunder provided me with the interesting big picture questions that I would not have considered with out him. Thank you to all my committee; I enjoyed and benefitted from your input and help tremendously. The proximate reasons that I was able to complete this academic endeavor were excellent academic advising and a good funding source, but the ultimate reason is the relationships that have nourished me since • childhood. Primarily, my parents are the reason I have been able to start and complete my doctorate. My mother, who importuned me to ''finish what you begin", and my dear father, with his natural Buddhist nature, both allowed me and encouraged me to do whatever it is to which I set out. Additionally, their financial support during graduate school kept me out of the poorhouse and kept my car out of the junkyard. I must thank my sister, who rolled her eyes and told me to "get a grip" many times during my graduate studies. I then thank my extended family, the cast of characters to whom I am related. They make me laugh and somehow think that I could do anything if that's the stock from which I originate. To my long time friends who work hard, but certainly expect to vacation well, and have little tolerance of any stupid, wimpy PhD excuses, thank you. My friends M. Kiuchi, P. Kurz, M. Zimmer, E. Soderstrom, A. Peet, D. Yonenaka, N. Larsen, M. Riker, T. Howe, A. Bruno, and T. Weller deserve special acknowledgements. Of the mentors from my academic career, I would like to thank and acknowledge Sister Rita Basta, Dr. Mike Jaeger, Dr. Dale McCullough, and Dr. Wayne Getz. Wayne, a special thanks to you, although you will never know how much you inspired me. I would also like to thank and acknowledge the iny:riad of graduate students who have made my life bright and graduate school stimulating. Most importantly, infinite thanks to my dear husband, John Shivik. He has been a stalwart support, fellow adventurer, paper reviewer, research partner, coffeemaker, cocktail alchemist, bad joke teller, and all round husband extrodinaire. Thank you, thank you, thank you all. I hope you like what you've been so instrumental in producing -this one's for you. �201 INTRODUCTION ELK MOVEMENTS IN RESPONSE TO DISTURBANCE Migration between high-elevation summer range and low-elevation winter range is common among Rocky Mountain elk (Cervus e/aphus ne/soni) in mountainous regions (Adams 1982). High-elevation ranges provide high levels of nutrition during the summer, but snow makes forage inaccessible during the winter. Fall migrations typically take place before the snow becomes too deep for animals to forage. In the White River area of Colorado, elk migrate between summer range, located in high-elevation national forest and wilderness areas, and lowelevation wintering grounds, located as near as the base of the summer range or as far as 80 km away. Historically, fall elk migration took place in December (Boyd 1970), but since the 1980s, fall movements have apparently advanced into August and September (Gray et al. 1994). During the development of the 1994 elk management plan for the White River herd, early elk movement from summer ranges onto public land was the most common complaint (Gray et al. 1994). In response to complaints. Colorado Division of Wildlife (CDOW) conducted a pilot study on 20 radiocollared cow elk from 1992-1995 to determine if and when elk were moving onto private land. Results from the pilot study indicated a correlation between early-season hunting activity and elk movement to private lands. However, it did not establish a causal relationship. Before instituting any management changes, COOW needed an experiment to determine whether early-season hunting activity was causing latesummer elk movements. I designed and conducted such an experiment in 1996-1997, and that experiment is the topic of my dissertation. In this introduction, I present background material on elk responses to an array of human activities in general and to hunting in particular. I intend this background material to provide a framework for understanding my experiment to determine the cause oflate-summer elk movements in the White River area. Elk Movement in Response to Human Activities Elk responses to recreational activities such as hiking, cross-counl:Iy skiing, and automobile driving depend on habituation to humans and on the presence of hunting. In Rocky Mountain National Park, where elk were accustomed to people and were not hunted, elk showed little response to approaches of people in automobiles or on foot, day or night (Schultz and Bailey 1978). In a study of another unhunted elk population in Yellowstone National Park, elk response to cross-counl:Iy skiers depended on how accustomed elk were to people. In areas of low human activity during the winter, elk responded with flight distances of 125-1,700 m, while in an area of high human activity, the flight distance was considerably shorter, ranging from Oto 300 m (Cassirer et al. 1992). Interestingly, elk in the area of high human activity showed a three-fold increase in flight distance when they were disturbed outside the area where people were present 24 hours a day (Cassirer et al. 1992). Thus, even habituated elk appear to flee human activity in areas where they do not expect such activity. Besides habituating to humans, elk may learn to differentiate between dangerous and benign situations with respect to road avoidance. In Rocky Mountain National Park, where elk were not hunted and traffic volume was high, Schultz and Bailey (1978) found that none of their 14 delineated elk behaviors changed with traffic volume, and there was little or no avoidance of the roads in winter. In contrast, in Roosevelt National Forest, which is adjacent to Rocky Mountain National Park, Rost (1975) found that a population of hunted elk avoided roads in winter. Wright (1983) found that mean distance ofradiocollared elk to jeep trails more than doubled, increasing from 800 m to 2, I 00 m when hunting season opened. In addition to danger level, the level of road activity also affected elk movements. After roads were closed in an experimental treatment, Cole et al. (1997) found a reduction in daily movement, home range size, and core area, all measures of elk movement Czech (1991) found a significant increase in mean elk distances to a road after it was opened to the general public with a corresponding increase in traffic. Elk also cross heavily traveled roads less often than lightly traveled roads (Hershey and Legee 1982). A third factor, cover, may also determine the degree,of elk response to disturbance. Logging disturbances changed normal elk movements (Edge and Marcun'i 1985): elk moved significantly longer distances away from logging areas than toward them. But both distance from disturbances, and areas of high elk use, were correlated to the amount of cover in the area (Edge and Marcum 1985). Similarly, Czech (1991) found that elk tolerance of logging operations was correlated positively with proximity to hiding cover. In one of the few manipulative experiments on elk movements. elk calves subjected to simulated surface-mining activities sought coniferous forest significantly more often than undisturbed calves (Kuck et al. 1985). Oil drilling has the same effect as mining: elk increased their use of forested habitats during drilling and post-drilling periods, compared with pre-drilling periods (Van Dyke and Klein 1996). Collectively, these disturbance experiments build a strong case for a relationship between elk movement and availability of cover. �202 All studies that evaluated elk movements after the disturbance ended found displacement to be a temporary condition in most situations. The Yellowstone elk displaced by cross-country skiers typically returned close to their original locations after people left the area (Cassirer et al. 1992). Road proximity resumed pre-hunting distances after hunting season closed (Wright 1983), and mean distance to roads decreased soon after the roads were closed for the season (Hershey and Legee 1982). During weekends, and immediately following the end oflogging activities, elk moved back into logged areas (Ward 1976, Edge and Marcum 1985). ;;a, Elk Movement in Response to Hunting Documented responses of elk to hunting include movement away from hunters or heavily hunted areas, increased movement, movement to thick cover, shifts in circadian patterns, and elevation shifts. Most responses seem to occur before winter migration, but some studies attribute changes in migration patterns to hunting pressures. For example, Knight (1970) and Morgantini and Hudson (1979) found a reversal of the normal downward autumn migration of elk, coinciding with the opening of the hunting season. Responses of elk to hunting activity may depend on the amount, location and quality of hiding cover, as well as the topography of the area. Altmann (1956) described an evasive "migration" of elk hunted adjacent to Yellowstone National Park. With the opening of hunting season, movements of these elk became relatively long (513 km), and the elk ceased long movements only when they reached the sanctuary of the Park. Similarly, Martinka ( 1969) found that 12 marked elk, which were in the National Elk Refuge just prior to hunting season, moved between 8 and 22 Ian, to areas closed to hunting in Grand Teton National Park. Other studies have found lessdramatic movement to refuge areas. With opening of hunting season, elk moved to densely forested (Irwin and Peek 1979) or shrubby (Morgantini and Hudson 1979) areas adjacent to their typical areas of activity, which provided refuge from hunting. Unpublished data for 1992-1995 from a pilot study on elk in the White River area (M. M. Conner, Colorado State University, unpublished data) found that distance moved around opening date(± 1 week from opening date) differed depending on habitat cover in the vicinity. Elk located in rugged and relatively inaccessible areas with access to thick timber and rough terrain (n = 11) moved an average of858 m/day (95% CI= 602, 1113). In contrast, elk located in road-accessible areas (n = 15), with little access to vegetative and topographic cover, moved an average of 1258 m/day (95% CI= 1011, 1505) to refuge on private land. Thus, elk without access to nearby habitat refuges moved significantly farther (401 m/day: 95% CI= 45, 756) than elk with access (P = 0.014, I-sided). Areas of rough topography and dense timber may serve as refuge; if such refuge is available, elk may not move far even when subject to high hunting pressure. Elk may respond to hunting method or density of hunters. Wright (1983) examined elk response to different hunting seasons: archery and muzzleloading, rifle-special elk, rifle-special deer, and deer and elk rifle. The greatest density of elk hunters was during the rifle-special elk season, and distances traveled by elk were also greatest during that season. Elk traveled three to four times farther during this season than during archery and muzzleloading season. Wright (1983) concluded that archery and muzzleloading hunters did not change the behavior of radiocollared cow elk. However, this result might be explained by low hunter densities (0.13 archery and muzzleloading hunters/km2, compared to 1.42 rifle hunterslkm2) rather than by the hunting method. In a 1985 study in the White River area, Consolidation Coal Company radiocollared 23 elk to evaluate their movement onto the company's protected land during early-season hunting. Eighty-seven percent of the elk were still found on National Forest lands midway through archery and muzzleloading seasons (Camp Dresser and McKee, Inc. 1986). But the lack of movement may be explained by a 63% reduction in archery hunters from 1984 to 1985, due to a change in hunting regulations (Gray et al. 1994). z.abn (1974), Lemke (1975), and Hershey and Legee (1982) noticed an increase in movements during the first 10-12 days of hunting season, followed by normal movement during the remainder of the season. These authors noted that hunting pressure was heavy during the first 10-12 days of hunting and relatively light for the remainder of the season. These studies of hunting suggest that elk behavior may change only when the density of hunters reaches a critical threshold. While some elk appear to have learned to move to refuge areas, elk already in refuges may learn to remain there in response to hunting pressure. In Jackson Hole, both resident and migratory elk use the National Elk Refuge for their winter range (Martinka 1969). Of the 183 marked elk in the study, resident elk were defined as elk found at any time from July through August within 15 km of the winter range, while migratory elk were defined as all other elk. In the fall, both resident and migratory elk had to cross hunting areas to move to their winter range in the National Elk Refuge. Resident elk were sensitive to hunter presence and tended to hang back in areas closed to hunting, while migratory elk were less hesitant to move through hunting areas to the winter range. These studies of elk moving to, or staying in, refuge areas support the theory that elk are responding, possibly with learned behavior, to hunting pressure. �203 Other studies found evidence to indicate that elk are not responding to hunting pressure, but are being differentially removed from heavily hunted areas. In Wyoming, hunting pressure focused on a resident population of elk just east of Yellowstone National Park, drastically reducing their numbers (Rudd et al. 1983 ). Like the Jackson Hole herd, migratory elk that summered in protected areas of Yellowstone National Park wintered with resident elk in unprotected range just outside of the park. Hunting in the unprotected range occurred before the migratory elk arrived from the park, drastically reducing the herd of resident elk (Rudd et al. 1983). The proportion of migratory to resident elk increased in the unprotected range, apparently due to hunting itself and not to huntinginduced shifts in elk behavior. Boyce (1991) noted that migration of elk wintering in the National Elk Preseive in Jackson Hole changed dramatically between the 1950s and 1980s. The primary migration routes shifted from openhunting national forest lands to routes through Grand Teton National Park, where hunting was more restricted. Between 1950 and 1954, 22% of the migrating elk were estimated to have moved through Grand Teton National Parle; by 1980-1984 the figure was 51%. Boyce (1991) hypothesized that this change maybe due to differential haIVest of the migrating animals and not due to shifts in elk behavior. However, there has been no experiment to test whether the cause of migration changes is differential removal or learned response. Similar to results from non-lethal disturbance studies, z.ahn (1974), Lemke (1975), and Hershey and Legee (1982) found that elk response to hunter disturbance was short-lived. These studies noted that elk movements return to normal after the initial 10-12 days of opening day, and that hunting is most intense during those initial days. Elk not only discontinued their erratic and increased movements, but also returned to their pre-season activity areas, indicating that hunting activity was a temporary disturbance. In conclusion, elk show highly plastic responses to human activity, habituating to people in areas where there is no hunting, but actively avoiding hunters. Elk response to human activity depends on the amount of, and the danger of, contact with humans. For example, avoidance of roads varies with the amount of human use and the danger level. The presence of dense cover for hiding seems to attenuate responses to human activity. Finally, elk responses to human activity do not seem to persist after the activity subsides. All studies of elk responses to hunting have been obseivational and any causal relationship between hunting and elk movements remains untested. PROJECT OVERVIEW Game Management Units (GMUs) 12, 23, 24, and 33 in northwestern Colorado compose the majority of the area of the White River elk herd. The White River herd has been one of the most heavily hunted, managed, and documented elk herds in Colorado (Boyd 1970, Freddy 1987, Gray et al. 1994). Like all elk herds in Colorado, the White River herd was decimated during the late 1800s, eventually being reduced to a few hundred elk (Bryant and Maser 1982). Since regulated hunting began in 1929, the population has been growing and was estimated to be 31,000 in 1993, the upper bound for COOW management objectives (Gray et al. 1994). The number of hunters using the area has grown along with the elk herd, with especially significant increases in the number of early-season hunters (archery and muzzleloading) between 1984 and 1992 (Fig. I.I). In the White River area, summer range is high-elevation public forest, and winter range is lower-elevation private land. Historically, the main White River elk migration from summer to winter ranges took place between late November and early January (Boyd 1970). However, Freddy (1987) noted that elk were beginning to be found in lower elevations, characteristically winter range, during the September and October hunting seasons, which was uncommon in the 1960s. Furthermore, there is now evidence that elk are moving in August and early September, during early-season archery hunting (Gray et al. 1994). Typically, early-season hunters did not hunt private lands; thus, private lands provided elk a refuge during early-season hunts. Late-summer elk movements to private land have led to complaints by local private landowners and resource managers (Gray et al. 1994). For private landowners, complaints focused on crop damage. For COOW, the primacy problem is herd size: because hunters have limited access to private lands, it is difficult for COOW to maintain or reduce herd size via haIVest (J. Madison and C. Reichert, COOW, personal communication). Hunter equity is a third issue: if a large proportion of elk are on private land, then the best hunting will be monopolized by those with enough money to buy trespass pennits (J. Madison and C. Reichert, COOW;' personal communication). Complaints about late-summer elk movement during the mid- to late-1980s prompted the CDOW to conduct a prelimiruuy study on late-summer elk movement to private land in the White River area. In 1992 the COOW trapped and radiocollared 20 adult female elk. A 4-year pilot study from 1992 to 1995 was conducted on 14 to 20 of these radiocollared elk from GMUs 12, 23, 24, and 33. During the study, 70-100% ofradiocollared elk located on public land areas moved to private or wilderness areas, and the mean day of movement was not different from opening day of early-season hunting. For elk moving from public to private land (n = 29) during the 4- year pilot study, the mean date of movement was 3 days (95% CI= -4, 11) from opening date (M. Conner, Colorado �204 State University, unpublished data). Although it did not establish a causal relationship, this study provided strong evidence of a correlation between elk movements and the opening of early-season hunting. To determine if archery and muzzleloading hunting caused elk movements in the White River area in the White River area, I conducted a 2-year experiment. In the White River area, other factors that may contribute to elk movement include historical movement patterns, long-term, learned movement responses, recreational activity, wood cutting, and livestock activity. Any or all of these factors may contribute to elk movement from July through October. However, because the pilot study identified archery hunting as the prime suspect for late-summer movements, and because the CDOW was interested in manageable rather than academic issues, this study focused on the effects of archery hunting on elk movement. In Chapter 1, I report on the field experiment designed to determine whether archery hunting causes elk movement: this experiment compares the timing of elk movements, and the proportion of elk on or moving to private land, with the opening date of archery season. In Chapter 2, I estimate daily hunter densities and compare elk movement to private lands with hunter density. In Chapter 3, I discuss the impact of domestic sheep grazing on elk movement. Each chapter is written to stand alone for publication and follows manuscript guidelines for The Journal of Wildlife Management. LITERATURE CITED Adams, A. W. 1982. Migration. Pages 301-321 in J. W. Thomas and D.E. Toweill, editors. Elk of North America: ecology and management. Stackpole Books, Harrisburg, Pennsylvania, USA. Altmann, M 1956. Patterns of herd behavior in free-ranging elk of Wyoming, Cervus_canadensis ne/soni. Zoologica 41:65-71. Boyce, M S. 1991. Migratory behavior and management of elk (Cervus e/aphus). Applied Animal Behaviour Science 29:239-250. Boyd, R J. 1970. Elk of the White River Plateau, Colorado. Colorado Division of Game, Fish and Parks Technical Publication 25, Fort Collins, Colorado, USA. Bryant, L. D., and C. Maser. 1982. Classification and Distribution. Pages 1-59 in J. W. Thomas and D.E. Toweill, editors. Elk of North America: ecology and management. Stackpole Books, Harrisburg, Pennsylvania, USA. Camp Dresser and McKee, Inc. 1986. Meeker PRLA elk migration study: monitoring report, volume 2. Prepared for Consolidation Coal Company. Camp Dresser and McKee, Inc., Denver, Colorado, USA. Cassirer, E. F., D. J. Freddy, and E. D. Ables. 1992. Elk responses to distwbances by cross-country skiers in Yellowstone National Park. Wildlife Society Bulletin 20:375-381. Cole, E. K., MD. Pope, and R G. Anthony. 1997. Effects of road management on movement and survival of Roosevelt elk. Journal of Wildlife Management 61: 1115-1126. Craighead, J. J., G. Atwell, and B. W. O'Gara. 1972. Elk migrations in and near Yellowstone National Park. Wildlife Monographs 29. Czech, B. 1991. Elk behavior in response to human distwbance at Mount St. Helens National Volcanic Monument. Applied Animal Behaviour Science 29:269-277. Edge, W. D., and C. L. Marcum. 1985. Movements of elk in reaction to logging distwbances. Journal of Wildlife Management 49:926-930. Freddy, D. J. 1987. The White River elk herd: A perspective, 1960-85. Colorado Division of Wildlife Technical Publication 37 Fort Collins, Colorado, USA. Gray, J. P., G. Byrne, and J. Madison. 1994. White River elk data analysis unit plan: game management units: 11,211,12,13,131,231,23,24,25,26,33. Colorado Division of Wildlife Internal Report, Grand Junction, Colorado, USA. Hershey, T. J., and T. A. Leege. 1982. Elk movements and habitat use on a managed forest in north-central Idaho. Idaho Department of Fish and Game Wildlife Bulletin 10, Boise, Idaho, USA. Irwin, L. L., and J. M. Peek. 1979. Relationships between road closures and elk behavior in northern Idaho. Pages 199-204 in M. S. Boyce and L. D. Hayden-Wing, editors. North American elk: ecology, behavior, and 'management University of Wyoming, Laramie, Wyoming, USA. it Knight, RR 1970. The Sun River elk herd Wildlife Monographs 23. Kuck, L., G. L. Hompland, and E. H. Merrill. 1985. Elk calf response to simulated distwbance in southeast Idaho. Journal of Wildlife Management 49:751-757. Lemke, T. 0. 1975. Movement and seasonal ranges of the Burdette Creek elk herd, and an investigation of sport hunting. Montana Fish and Game Department Job Final Report 32.01, Job No. BG-3.15, Missoula, Montana, USA. �205 Martinka. C. J. 1969. Population ecology of summer resident elk in Jackson Hole, Wyoming. Journal of Wildlife Management 33:465-481. Morgantini, L. E., and R J. Hudson. 1979. Human distribution and habitat selection by elk. Pages l32-l39 in M. S. Boyce and L. D. Hayden-Wing, editors. North American elk: ecology, behavior, and management. University of Wyoming, Laramie, Wyoming, USA Rost, G. R 1975. Responses of deer and elk to roads. Thesis, Colorado State University, Fort Collins, Colorado, USA. Rudd, W. J., AL. Ward, and L. L. Irwin. 1983. Do split hunting seasons influence elk migrations from Yellowstone National Park? Wildlife Society Bulletin 11:328-331. Schultz, R D., and J. A. Bailey. 1978. Responses of national park elk to human activity. Journal of Wildlife Management 42:91-100. Van Dyke, F., and W. C. Klein. 1996. Response of elk to installation of oil wells. Journal ofManunalogy 77:1028-1041. Ward, A L. 1976. Elk behavior in relation to timber harvest operations and traffic on Medicine Bow Range in south-central Wyoming. Pages 32-43 in S. Hieb, editor. Proceedings of the Elk-Logging-Roads Symposium. University ofldaho, Moscow, Idaho, USA Wright, K. L. 1983. Elk movements, habitat use, and the effects of hunting activity on elk behavior near Gunnison, Colorado. Thesis, Colorado State University, Fort Collins, Colorado, USA. z.abn, H. M. 1974. Seasonal movements of the Burdette Creek elk herd. Montana Fish and Game Department Job Final Report 32.01, Job BG-3.13, Missoula, Montana, USA 7000 - - - - - - - - - - - - - - - - ~ ~ ~ 6000 ::::J .c. C 0 95% Cl - .., . , ~-------., I ,. 5000 ,,. - -- _, ,. - , , - • •111, ' 111, ' " ... ' ' (/J m4000 ~ ai 3000 CD ~ 2000 CD .0 ~ 1000 z 0 -+----..---.------,------,.-----.----.----,------.---.----1 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 Year * Variance estimates not available 1984-1987. Figure I. 1. Number of archery and muzzleloading hunters using GMUs 12, 23, 24, and 33 in northwest Colorado, 1984-1994. Data from Deer, Elk, and Antelope Management (DEAMAN) database and software, Colorado Division of Wildlife, unpublished data. �206 �207 CHAPTER 1: ELK MOVEMENT IN RESPONSE TO EARLY-SEASON HUNTING IN NORTHWEST COLORADO INTRODUCTION Since the early 1900s, Rocky Mountain elk (Cervus elaphus nelsoni) populations have expanded their range from remote areas to much of the Rocky Mountain States (Bryant and Maser 1982). As elk range expanded, so too did elk contact with humans and human disturbances. Elk responses to hunting, road traffic, hikers, cross country skiers, snowmobiles, logging, mining, and oil drilling have been studied in order to reduce elk-human conflicts. Problems with elk responding to hunting center around elk moving onto hunting refuges, which are areas with no or low levels of hunting such as private land, national parks, or densely forested areas. Documented responses of elk to hunting include movement away from hunters or heavily hunted areas (Altmann 1956, Martinka 1969, Wright 1983), increased movement (Zahn 1974, Lemke 1975, Hershey and Legee 1982, Wright 1983), movement to dense vegetative cover (Irwin and Peek 1979, Morgantini and Hudson 1979), movement to private land (Wright 1983) or national parks (Altmann 1956, Martinka 1969), shifts in circadian patterns (Strohmeyer and Peek 1996), and elevation shifts in migration patterns (Knight 1970, Morgantini and Hudson 1979). Most responses to hunting occur before migration to winter range, but some studies have noted changes in migration patterns that were attributed to hunting pressures (Martinka 1969, Boyce 1991). All previous studies of elk responses to hunting have been observational, there have been·no experiments designed to test whether hunting activity causes elk movements. Elk inhabiting the White River area of northwest Colorado are no exception to the pattern of increased range expansion and increased conflict with humans. Prior to 1960, the fall migration of White River elk from high elevation public land, summer range, to lower elevation private land, winter range, (Fig 1.1) occurred during the month of December (Boyd 1970). However, Freddy (1987) noted that elk were beginning to be found in lower elevations during the September and October hunting seasons, which was an uncommon occurrence in the 1960s. Since the 1980s, fall movements have apparently advanced into August and September (Gray et al. 1994). Archery and muzzleloading hunting (early-season hunting) opens on the third Saturday of August in the White River area; around the time of the alleged elk movements onto private land. Since the 1960s, the number of earlyseason hunters using the White River area has grown, with especially significant increases between 1984 and 1992 (Fig. 1.2). Typically, archery hunters did not hunt private lands; thus, private lands provided elk a refuge during early-season hunts. As archery and muzzleloading hunter numbers increased, so too did complaints about early elk movements onto private land. For the private landowner, complaints about early elk movements focused on crop damage. For the Colorado Division of Wildlife (CDOW), the primary problem was herd size: it was difficult for CDOW to maintain or reduce herd size via harvest because hunters have limited access to private lands, (J. Madison and C. Reichert, Colorado Division of Wildlife, personal communication). Because of complaints during the mid- to late 1980s, CDOW conducted a preliminary study on late summer elk movement to private land in the White River area. From 1992 to 1995, 20 radiocollared elk were intensively monitored during August and September, approximately one month before and one month after opening day of archery hunting. Although the proportion and date of elk moving indicated a correlation between elk movement and opening of early-season hunting, a causal relationship was not established. Before instituting any management changes, the CDOW required. an experiment to determine if early-season hunting activity was causing late summer elk movements. My objective was to determine if archery hunting caused elk movement to private land during late summer. Specifically, I conducted a 2-year field experiment in :which_opening date of archery hunting was manipulated in a crossover design. The study area was split into 2 treatment areas; each area received an earlyand late-opening treatment. I tested the null hypotheses that; (1) the mean date of movement to private land was not different between early and late treatments, and (2) the proportion of elk on private land was not affected by the opening of archery hunting. �208 STUDY AREA The White River area of northwestern Colorado covered approximately 4,540 km2 and was composed of Game Management Units (GMUs) 12, 23, 24, and 33. In Colorado, GMUs were delineated to distribute hunters through allocation of hunting licenses. White River area land ownership was 34% private land and 66% public land (Fig 1.1), with 82% of public land being United States Forest Service (USFS). USFS land was mostly at high elevation and toward the center of the study area, with some Bureau of Land Management (BLM) areas at lower elevations in the southern sections of the study area. Private land was lower in elevation and comprised mainly ranches and coal mines. Topography, climate, and vegetation varied widely throughout the study area. Elevation ranged from 1,629 to 3,700 m. The central part of the study area was high elevation public land, split by the White River valley. The high elevation areas dropped off to lower elevations to the north, south, and western edges of the study area. Higher elevations had severe winters with heavy snowfall, while lower elevations had comparatively mild winters. Mean annual precipitation at 3,000 m in the National Forest areas was about 100 cm, compared with about 30 cm at lower elevations of <2,000 m. Vegetation types at elevations >2,600 m were in the montane/subalpine zone and included groves of aspen (Popu/us tremu/oides), Engelmann spruce (Picea enge/manni), and alpine fir (Abies /asiocarpa) interspersed with grassy meadows. Middle elevations (1,9802,600 m) were a transitional zone that consisted primarily of pinion pine (Pinus edu/is),jumper (Juniperus scopulorum), and big sagebrush (Artemisia tridentata). Lower elevations (<2,000 m) in the southern and northern parts of the study area were in the Great Basin zone with sage grasslands. Oakbrush patches were found at the middle and lower elevations with major shrubs that included Gambel's oak (Quercus gambe/ii), serviceberry (Amelanchier alnifo/ia), mountain mahogany (Cercocarpus montanus), chokecherry (Prunus virginiana), snowberry (Symphoricarpos utahensis), bitterbrush (Purshia tridentata), and rabbitbrush (Chrysothamnus nauseosus). Higher and middle elevation areas provided summer and fall forage for elk, while lower elevations were typically used by elk during winter. Boyd (1970) provides a detailed description of the study area. Hunting for elk was economically important to local residents in the study area (Gray et al. 1994). Guides, outfitters, local service sectors, and private landowners (who sell trespass permits) made a large proportion of their annual income during the elk and deer hunting seasons, with most hunting revenues accrued during rifle hunting seasons. MEmODS Experimental design The White River and North Fork of the White River divided the study area roughly east-west into 2 halves (Fig. 1.3). GMU 12 and parts of GMUs 23 and 24 composed the north treatment area, while GMU 33 and parts of GMUs 23 and 24 composed the south treatment area. Because only 2 elk moved from one side of the White River to the other within a year during the 4-year pilot study (M. Conner, Colorado State University, unpublished data), I felt there would be little interchange between treatment areas. Two treatments were applied: an archery season that opened 1 week earlier, and another that opened 2 weeks later than normal, yielding a 3week difference in opening dates. During the first year of the study, the early-opening treatment was randomly assigned to the south treatment area. Treatments were reversed the second year of the study. Thus, in 1996, archery hunting opened early, 24 August, in the south area, and late, 14 September, in the north area. When treatments were reversed in 1997, archery hunting opened early, 23 August, in the north area, and late, 13 September, in the south area The number of archery and muzzleloading licenses issued for elk and deer during the study,:'were based on the mean number per year calculated for 1990-1994 (4,985; Colorado Division of ,· •1 Wildlife unpublished data), with licenses allocated approximately 50% per treatment area. Restricting licenses kept hunter density consistent during the study and consistent with years of high elk movement complaints. Sample size for number of collared elk was based on the primary response variable, mean date of movement, for the following parameters; Cl= 0.05, ~ = 90%, effect siz.e = 7-day difference in mean date of movement between treatments, and estimated variance = 22. 7 days. The variance was based on pilot study data (M. Conner, Colorado State University, unpublished data). For these parameters, 36 cows per treatment area • �209 were required. To allow for some telemetry failures or erratic elk movements off the study area, extra cows were collared. Thus, 40 cows per treatment area were collared for the study. Elk capture locations were randomly picked from a uniform distribution by a computer. Random Universal Transverse Mercator (UTM) coordinates were generated until 80 points lay within the study area's national forest boundaries. Elk were captured by helicopter netgunning (Barrett et al. 1982) during mid-July in 1996 and 1997. Elk were captured on their summer range to assure a random sample for the July-October study period. The helicopter pilot flew to each randomly selected location and captured the first adult female elk found near that location. Although it was not practical to capture exactly as dictated by a random number algorithm, due to constraints on private land access and time, a reasonably representative sample was obtained with 2 elk being captured at some locations (Fig. 1.3). In addition to the 80 elk captured in 1996, 8 radiocollared elk from the pilot study were used in 1996 analyses. In 1997, 10 additional elk were captured to replace animals that died, bringing the total number of collared animals back to 80, with 40 per treatment area. Colorado State University Institutional Animal Care and Use Committee approved the capture and handling methods (protocol 95-180A-0l). Each elk was fitted with a 148-152 MHz radio collar with a mortality sensor. A high percentage of bulls are lost to hunting, which could reduce sample size below levels required to maintain the power of the hypothesis tests. Because this study was to take place during a hunting season, when bulls are desirable game, I wished to collar only adult females to minimize loss of sample size. During fall rutting season, August-October, there is a high degree of association (97%) between male and female elk (Franklin and Lieb 1979). I collared only adult female elk to represent the movement patterns of both sexes and maintain the power of the hypothesis tests. Data collection The relevant time frame for determining elk movement in response to archery hunting was defined as ±1 month of opening day. Because early treatment opening was 23-24 August and late treatment opening was 13-14 September, I collected locations 20 July- IO October. The study period ended slightly less than a month after late treatment opening date to avoid confounding elk movements due to archery hunting with effects due to rifle hunting, which opened 13-14 October. All analyses were done on data collected over the entire 3-month study period. I relocated radiocollared elk between 0700-1500 hr using a Cessna 182 or 185 fixed-wing aircraft with a 2-element Yagi antenna mounted to each strut of the airplane. For each elk relocation, UTM coordinates were recorded with a Global Positioning System that was not differentially corrected. I collected elk locations 2 times a week with 2-4 days between collections. A Geographical Information System (GIS) map was used to record elk locations, land ownership, and treatment boundaries. The GIS map was digitized from a United States Geological Survey map at a scale of 1:100,000. For each location of a telemetered cow, a 0 was assigned if the location was on public land (e.g. Flat Tops Wilderness Area, USFS, State Wildlife, and BLM lands) and a I was assigned if the location was on private land. Because I was interested in gross classifications of elk locations (public land versus private land), I did not do a formal check of the telemetry error. However, I did a cursory check of telemetry error when picking up collars from a different study in the area. The mean error on the collars I collected was 333 m (n = 24, 95% CI= 265, 40 I), with a range of 117-683 m. Data analysis The primary response variables were mean date of movement for elk that changed classification (occupying public and moving to private land) and proportion of elk on private land. I calculated th~ date of movement for each elk using logistic regression oflocation on public (y = 0) and private land (y = 1) versus date. The date of movement was defined as the date at which the logistic curve crossed 0.5 on the y-axis, indicating an equal probability of being on public or private land. Only dates of movement from public to private land during the study period were included in analyses of mean date of movement. If hunting had no effect on elk movements, then there should be no difference between the mean date of movement to private land between early- and late-opening date treatments. To test for a treatment effect on mean date of movement, I accounted for treatment, area, and year effects. The corresponding ANOVA model was: �210 where: yyu = date of movement for the ff' radiocollared elk, receiving treatment i, in year j and in area k, µ = overall mean date of movement (for all elk, both years), a., = fixed effects due to treatment i (early or late opening), 131 = fixed effects due to yearj (year 1 or year 2), ok = fixed effects due to area k (either north or south), and &yu = random error for the /'1' radiocollared elk. Specific hypotheses tested were: Ho 1: Mean date of elk movement for early-opening treatment = mean date of elk movement for late-opening treatment; a., ~ 0; HA1: Mean date of elk movement for early-opening treatment < mean date of elk movement for late-opening treatment; a., < 0; ffm: Mean date of elk movement for year 1 of experiment = mean date of elk movement for year 2; f31 = 0; HA2: Mean date of elk movement for year 1 of experiment :t:. mean date of elk movement for year 2; 131 -:#- 0; Hen: Mean date of movement for north-area elk = mean date of movement for south-area elk, ok= 0; HAJ: Mean date of movement for north-area elk mean date of movement for south-area elk, ok:t:. 0. * Hypothesis l was the experimental hypothesis; hypotheses 2 and 3 accounted for other sources of variation. I used PROC GLM (SAS Institute 1990) to estimate and test differences in mean date of movement by area, treatment, and year. Type m sum of squares were used for hypothesis tests. To determine whether elk receiving both early and late treatments on public land, regardless of area or year, were affected by opening date, I used a paired I-test blocked on individual elk. This individual blocking reduced confounding effects and better accounted for repeated measures taken on animals receiving both treatments; thus, the analysis was limited to those individuals that received both early- and late-opening date treatments. Fewer elk received both treatments than received a single treatment because some elk died after the first study period, some elk were on public land only one year (they stayed on private land the other year), and some changed treatment areas between years. For the elk getting both treatments, I subtracted their early date of movement from their late date of movement to account for the fact that this was a repeated measurement on the elk. I tested the null hypothesis: Ho: Difference in date of movement for late treatment - early treatment <= 0; HA: Difference in date of movement for late treatment - early treatment> 0. I used logistic regression models to predict the proportion of elk on private land on day i (j, i) from elk location data (public or private land). Models included date as a covariate, and hunting season, area and year effects as categorical variables. Julian date was used to model date effects so that year effects could be considered separately from study period date. Hunting season referred to the period before opening of archery season (hunting season= 0) and after opening (hunting season= 1). I began with a global model, which included date, hunting season, area, year, plus all 2-, 3- and 4-way interactions of date, hunting season, area, and year (fable 1.1, Model 1). I then developed 9 additional a priori hypothetical models for proportion of elk on private land (fable 1.1, Models 2-10). After analyzing the a priori models, I built 3 additional models to see if a better model could be built (fable 1.1, Models 11-13). �211 I followed the methodology of Burnham and Anderson ( 1998) to select an appropriate model. I used Akaike's Information Criterion (Akaike 1973), adjusted for overdispersion and corrected for small sample bias (QAICc), as the basis for objectively ranking models and selecting an appropriate "best approximating" model (Burnham et al. 1995). QAICc was defined as: QAICc =- 2{lni) +2K + 2K(K +I) n-K -l ' c where In .e is the natural logarithm of the likelihood function evaluated at the maximum likelihood estimates for a c given model, K is the number of estimable parameters from that model, n is sample size, and is the estimated overdispersion factor. QAICc was used instead of AICc because repeated measurements were taken on radiocollared elk. Lack of independence in the data, in this case repeated measures on elk, may lead to overdispersion or "extra-binomial variation" (Burnham and Anderson 1988). I estimated the overdispersion parameter ( c) from the global model, and then used this estimate to adjust (inflate) the variance estimates of model parameters and predicted values, and to calculate QAICc for all models (Burnham and Anderson 1998). Parameter estimates, c, and QAICc values were calculated using PROC GENMOD (SAS Institute 1993). The sample size, n, was the number of elk locations collected during the study period. The best approximating model was selected based on minimum QAICc. Models were ranked and compared using DQSAICc (Leberton et al. 1992, Burnham and Anderson 1998) and normalized QAICc weights (Buckland et al. 1997, Burnham and Anderson 1998). For the suite of models being compared, AQAICc was computed for each model as: AQAICc,,, = QAICc,,, - QAICc.,,n , where QAICc,,, was the QAICc value for the mth model and QAICc.,,n was the minimum QAICc among the suite models being compared. Essentially, AQAICc,., is an estimate of the distance between the best approximating model and model m. Normalized AICc weights (w.,) were estimated for each mth model: _ _!.~QAICc w m- e 2 ~> R "' l --~QAICc 2 , r r=l where R refers to the R models chosen for evaluation. AQAICc and normalized QAICc weights were used to address model selection uncertainty. Generally, models within 1-2 QAICc units of the selected model were considered competing models for explaining elk movements to private land. Another element of uncertainty in estimates of model precision is selection of the appropriate model (Buckland et al. 1997). If several sets of data, generated from the same underlying process, were evaluated using AICc, there would be some variation in the selected best model (Burnham and Anderson 1998). Estimates of precision from any one best model do not include model selection variance and may be biased low. Therefore, after model selection, I used model averaging to better estimate the precision of the estimates of daily proportion of elk on private land from all the models I developed. Average estimates of the daily proportion of elk on private land were calculated for each day i across all models considered. Because all models were logit models, modelaveraged daily proportion of elk on public land, P;, and var(p;) were calculated in the logit scale and backtransformed to calculate the appropriate confidence interval, which was slightly asymmetric. The transfomation used was: logit(J\) = 1n[ I - R !fa;]. logit(p;) = L Wm logit(P;m), and m=l �212 where wm is the normalized AICc weight. The variance of the model-averaged estimates, which included conditional sampling variance and model uncertainty was estimated: where M,,, is the mth model. The 95% confidence interval for the daily estimated proportion of elk on private land was calculated as: +95%CI(p;) e +9S%Cl~ogit(fa1)] - - - - - - - , and I +e +9S%CI~ogit(fa,)] -95%Cl(p;)= e -9S%CI[logit(fa1)] _ I +e -9S%CI~ogit(j,1)] To estimate the effect of hunting season on the proportion of elk on private land, I estimated the difference between the proportion of elk on private land immediately before and after opening date: e logit(fab) l +elogit(j,b) ' pa p where and b are the model-averaged proportion of elk on private land after and before opening date. The variance in estimated proportion of elk on private land before and after opening of archery hunting was calculated: �213 Procedures for estimating covariance of model-averaged values are in the development stage and untested (K. Burnham, Colorado Cooperative Fish and Wildlife Research Unit, personal communication). Thus, I assumed the covariance to be z.ero. This omission may result in a slightly inflated variance, and a possibility of not detecting a true effect of hunting season. The variance of the model-averaged estimated proportion of elk on private land was calculated using the delta method (Seber 1982): ~i, Let u =logit(pa), then var(pa) =v:J and "l1+e" A • ~ 2 du du du =-=viir(u )-=- =(-=-) viir(u) =[ A viir(p a) dfaa dfaa dfaa ]2 e" A A (l+e") 2 viir(u) • Substituting logit(p a) back in for u: 2 _ elogit(it> var(_p a) = _ _ [ 2 ( 1+ elogit(faa)) _ _ var[logit(p a)]= [fa a (l - pa) f var[logit(pa)]. - 1 The var(pb) was calculated similarly. The 95% CI was calculated: - - - - I\_ 95%Cl(p0 - Pb)~ (Pa - fib)± l.96se(p0 - Pb)· Because previous studies used changes in daily distances moved and elevation shifts to evaluate elk responses to disturbance, I calculated these metrics for comparison. Mean daily distances moved and mean daily elevations during the study period were calculated from the 80-radiocollared elk. Distances were calculated from consecutive UTM locations for each elk as straight-line distances. Because elk were not located at fixed time �214 intervals, distances between locations were expressed per day by dividing the distances moved by the number of days between locations. Daily distances were not absolute measures of distance moved, rather they were indices of movement. Elevations for each elk location came from a statewide I: l 00,000-scale elevation map. RESULTS Results from the ANOVA on mean date of elk movement indicated a treatment effect (P = 0.013, one sided), but no area or year effects (P > 0.100; Table 1.2). However, north- and south-area elk showed different responses to opening of archery hunting. To evaluate this apparent difference, I preformed a post hoc analysis separately by area. This analysis revealed a 14-day difference (P = 0.006, one sided) in mean date of movement between treatments for elk in the north area, versus a 6-day difference (P = 0.218, one sided) between treatments for elk in the south area (Table 1.3). Area and year were confounded within sequence of treatments this study. That is, elk in the north area received the late-early treatment sequence, while elk in the south area received the early-late treatment sequence. Thus, it was impossible to determine whether a sequence effect was really a difference resulting from (1) a difference between treatment areas because oflocation of private-land refuges, topography, etc., or (2) elk receiving a late-opening treatment the first year followed by an early-opening treatment the second year behave differently than elk receiving the early-late sequence. I reference this confounded effect as an area effect, which I will discuss later. Sixteen of the 80-radiocollared elk received both early and late treatments and moved from public to private land during the study period. That is, there were 32 paired movements (16 elk receiving 2 treatments and moving form public to private both years) compared to 60 unpaired movements (elk receiving treatments and moving from public to private land at least 1 year). Some elk moved to private land immediately after capture in 1996, and some elk never returned to public land in 1997, thus fewer than 80 elk were available for treatment each year. The difference between the paired and unpaired movements occurred because 4 elk moved the first year but died before the second year, 3 elk were collared in 1997 and moved only the second year, 11 elk only moved one year because they were on public land one year and private land the other year, 1 elk changed treatment areas and received late-late treatments (counting as 2 unpaired movements), and 8 elk moved from public to private land one year only. The difference in mean date of movement for the 16 elk receiving both early and late treatments was 5 days (95% CI= -3, 13; P = 0.091 one sided). Using model-averaged values, I generated predicted curves of proportion of elk on private land versus Julian day, hunt season, area, and year (Fig. 1.4 and 1.5). All logistic models to estimate the daily proportion of elk on private land had a significant positive date effect indicating that proportion of elk on private land increased from 19 July-10 October both years of the study (Fig. 1.4 and 1.5). The logistic regression model with the lowest AICc (Table 1. 1, model 13) was: logit(f,;) = --4.715-1.294(area) +4.987(hunt season)+ 0.015(Julian day)+ 0.009(area x Julian day) -0.018(hunt season xJulian day) , where f,; is the predicted proportion of elk on private land on Julian day i. The top 2 models, both post hoc models, (Table 1.1, models 13 and 12) were 1.85 AICc units from each other, while the remaining 11 models were over 3 AICc units greater than model 13 (Table 1.4). Hence, I considered models 13 and 12 as competing models. Model 12, the second model, was identical to model 13 except for an additional area x hunt season interaction. Thus, area, hunt season, Julian date,,;u-ea x Julian day, hunt season x Julian day, and area x hunt season may be important predictors of the proportion of elk on private land. The area effect arose from a higher proportion of elk on private land, in general, on the north area versus the south area, the hunt season effect arose from a higher proportion of elk on private land after hunting opened versus before, and the Julian date effect arose from the increase in elk on private land through time (the study period). The area x Julian day interaction arose from the steeper slope for the north area versus the south area, and the hunt season x Julian day interaction arose from the steeper slope before hunting season opened versus after it opened. �215 The experimental effect on the number of elk directly influenced by early-season hunting can be estimated by the abrupt increase in proportion of elk moving to private land when hunting opened. I used the difference in the model-averaged predicted proportion of elk on private land immediately before and immediately after opening day to estimate the direct effect of opening of hunting. Elk on private land increased 8-18% at the opening of early-season hunting (Table 1.5). During the study period, a significantly higher proportion of radiocollared elk moved to private land in the north treatment area compared to south area (P = 0.035; Table 1.6). On average, elk moved 613 m/day (95% CI= 594,632). The mean daily distance moved by elk before hunting season opened was 651 m/day (95% CI= 619, 683) and 575 m/day (95% CI= 540,610) after opening (Fig. 1.6 and 1. 7). During the study, the mean elevation of elk changed by -2.1 m/day (95% CI = -2.3, -1.9). Elk moved down in elevation an average of-2.4 m/day (95% CI= -2.9, -1.9) before hunting season opened and down -0.7 m/day (95% CI= -1.2, -0.2) after opening (Fig. 1.8 and 1.9). DISCUSSION Experimental ResulJs Although elk movements in the White River area may not be solely caused by archery hunting activity, the experimental results show that archery hunting activity has an effect on elk movements despite other factors, especially in the northern treatment area. The strength of a 2 x 2 crossover design is that it blocks or controls for two sources of external variation, allowing for much stronger inference of cause and effect than the same design without a strict crossover (Ratti and Garton 1994). The crossover is also more efficient than a randomiz.ed design (Ott 1993); that is, a randomiz.ed design requires a larger sample size. In this experiment, area and year were the external sources of variation controlled for by the crossover design, and opening date of archery hunting was the treatment Any factor that occurred on both halves of the study area, such as recreational activity, grazing, weather, forage phenology etc., that also caused elk to move, would lessen detection of an effect due to hunting activity. For example, if recreational activity caused some elk movement during the study period, then these recreationally induced movements would be unaffected by opening date of archery hunting and would serve to attenuate detection of archery hunting effects. If no elk were responding to archery hunting activity, either because they moved in response to another factor or they did not move at all, then I would fail to reject the null hypothesis and would conclude that archery hunting was not affecting elk movement. In the White River elk population, some elk may move regardless of what happens with archery hunting, because their movement was a response to some other factor; some elk may not move; and some elk may move due to changes in archery hunting activity. In the north treatment area, elk receiving the early treatment moved 14 days earlier than elk receiving the late treatment, indicating that some elk moved in response to archery season. Similarly, the effects of archery hunting on the daily proportion of elk on private land can also be distinguished from other effects. The positive slope for proportion of elk on private land resulted from steady movement of elk to private land during the study period and suggested that factors other than archery hunting may have affected elk movements. However, some of the movement to private land was directly attributable to hunting. That is, 23-44% of the radiocollared elk moved to private land during the study period, and 8-18% of this movement occurred at the opening of archery season. The significant jump in proportion of elk on private land at archery opening, beyond the steady increase, indicates that archery hunting activity did affect movement beyond that caused by other factors on both the north and south treatment areas. Sequence or carry-over effects can dilute detection of treatment effects in this type of study. A carryover effect occurs when the sequence of treatments affects responses to treatments (Ott 1993). In the case of White River elk, a carry-over effect would exist if south-area elk, receiving the early-opening treatment the first year, moved early the second year, before the late-opening date. A carry-over effect would be manifest on the north area if elk receiving the late-opening treatment the first year, moved late the second year, after the earlyopening date. In this experiment, elk on the north area moved slightly before opening date both years, while elk on the south area moved in accordance with opening date. Thus, I concluded that there was no carry-over effect due to the sequence of the treatments. I attributed differences in movements to area rather than sequence effects. A post hoc analyses on mean date of movement and daily proportion of elk on private land revealed that treatment effects were greater in the north area. When designing the study, I assumed that the north and south �216 areas were similar and would serve as controls for each other. Alleged elk movements to private land were perceived to be similar problem on both areas. The differences may have occurred because habitat, cover, topography, or any other area factors important to elk movement were different on the two treatment areas. Fortunately, each area received both treatments, so I could look at each area as its own control and evaluate each area separately. In the north treatment area there was a significant 14-day difference in mean date of movement between early and late treatments, while in the south treatment area there was a non-significant 6-day difference. In the north treatment area, about twice as many elk moved to private land ( 44%) compared to elk in the south area (23%). The difference in treatment response in the north and south areas may be due to cover and refuge configuration. In general, animals must balance the tradeoff between feeding and danger (Krebs and Davies 1993); elk may reduce use of open areas and use cover more frequently in times of perceived or real danger. Elk calves subjected to simulated surface-mining activities used coniferous forest significantly more often compared to undisturbed calves (Kuck et al. 1985). Similarly, elk subjected to oil drilling significantly increased use of forested habitats during drilling and post drilling periods, compared with the pre-drilling period (Van Dyke and Klein 1996). Where logging activities disturbed elk, areas of high elk use were negatively correlated to the amount of cover in the area (Edge and Marcum 1985, Edge et al. 1985, Czech 1991). In the White River area, the north area was rolling, with open parks interspersed with a few high peaks, while the higher elevation south area was essentially a plateau cut by steep narrow canyons and with steep, densely forested cliffs defining the plateau. These densely forested canyons and cliffs, inaccessible to motor vehicles, offered good refuge from hunters on public land. In contrast, the north area offered less shelter on public land, but had refuge in large private-land tracts that bordered the area. The south treatment area had less private land adjacent to public land, as BLM land was abundant at lower elevations. It may be that elk in the south area moved into forest refuges on public land after archery hunting opened, while elk in the north area moved to private-land refuges. More generally, elk movements during hunting seem to depend on location and quality of hiding cover. Elk decreased their daily movements after attaining sanctuary from hunting in national parks (Altmann 1956, Martinka 1969) or thick vegetation inaccessible to most hunters (Lemke 1975, Irwin and Peek 1979, Morgantini and Hudson 1979). I found a similar general pattern; elk on the north and south areas reduced their daily movements an average of 76 m/day after hunting season opened and they attained refuge from hunting on private land or in thick vegetation. a Scale Issues Three main issues became evident during the course of the study that fall under the rubric of large-scale field experiments: (1) a difference between movement responses for elk in the north and south areas, (2) a difference in the experimental results compared to pilot study results, and (3) failure of many elk to receive both treatments. As discussed, I think the difference in treatment responses between north and south elk were due to proximity of cover and private land hunting sanctuary. Additionally, I speculate that differences in wintering grounds and migration routes between the two treatment areas also affected movement responses. In the south treatment area, wintering grounds were at the south end of the study area. Southern elk were impaired from migrating beyond the study area by barriers created by Interstate 70 and the Colorado River. In the north treatment area, although private-land refuge was adjacent to the study area, wintering grounds were 30-80 km west of the study area During the study period, elk moving to private land in the north area were just beginning their fall migration, whereas elk in the south area were completing their migration by moving to private land. It may be that there was no energetic cost to northern elk to move to private land in late summer, while southern elk would risk depleting their winter forage by moving to private land in late summer. There was less movement during..the ex~ent than I expected based on pilot study data. During the experiment, 21-48% of elk moved to private land compared to 70-100% during the pilot study. I think this disparity arose because the pilot study lacked a random sample from the entire area. Pilot study animals were collared primarily in areas with perceived movement problems. When the spatial scope of sampling expanded such that the entire study area was randomly sampled, many elk were collared in areas with no previous record of movement Corresponding to the increased spatial variability in elk capture location was an increased variability in elk movement responses: elk in some areas did not move. The pilot study animals were not representatives of the entire study area, rather they represented areas with movement problems. �217 I designed this experiment with the intent that it be a crossover experiment, which used a crossover AN OVA model to test whether the mean date of elk movement from public land to private land was different between treatments. Exposure by each animal to both treatments was a requirement of the crossover analysis, and this requirement was impossible to implement in this large-scale field experiment. I did not have the control to ensure that each animal received both treatments; therefore, I analyzed the data as paired and unpaired data. Interpretation of mean date of movement results would have been more robust if all elk received both treatments. Mean date of movement between treatments was less dramatic for the paired analysis (5 days) compared to the unpaired analysis (10 days), though the difference was not significant. This difference may represent sampling variance; the small size of the paired sample makes this possibility likely. MANAGEMENT IMPLICATIONS Although there was a statistically significant effect of archery hunting on timing and number of elk moving to private land in the north treatment area, the question remains as to whether this effect was significant to elk management. The experimental effect on number of elk moving can be estimated by the abrupt increase in proportion of elk moving to private land when hunting opened. This jump was 8-18%; hence, the CDOW can reduce, at most, approximately 20% of the elk movement to private lands by manipulating archery hunting activity. However, 20% of a herd may represent a large enough number of animals to make management changes worthwhile. Additionally, if elk movements are concentrated in problem areas, then manipulation of opening date may result in a proportionally higher reduction of elk movements in those areas. Managers must decide whether the contribution of archery hunting activity to elk movement is large enough to warrant management action. Elk responded to archery hunting activity differently in north and south treatment areas; I concluded that archery hunting activity had little or no effect in the south area, but had a statistically significant effect on timing and proportion moving in the north area. Thus, managers concerned with elk movement need to be cautious if citing White River results as a reason for management actions; elk movement was dependent on the habitat and topography of an area, and these factors must be considered in any plan to reduce elk movements. Furthermore, because this study manipulated opening date of hunting activity but kept hunter numbers constant, the question remains of what percentage of the elk would still move if hunting were reduced or eliminated. Managers may also want to consider the time needed to make permanent changes in elk movement patterns. Elk responses to disturbance may be long- or short-lived depending on the type of disturbance and the type of elk response. For example, studies that evaluated elk movements after disturbances ceased found that displacement appears to be temporary in most situations. Elk displaced by cross-country skiers returned to their original locations after people left the area (Cassirer et al. 1992). During weekends, and immediately following the end oflogging activities, elk moved back into logged areas (Ward 1976, Edge and Marcum 1985). Even for more potentially lethal hunting disturbances, mean distances to roads returned to pre-hunting distances after hunting season closed (Hershey and Legee 1982, Wright 1983). Zahn (1974), Lemke (1975), and Hershey and Legee (1982) also found that elk response to hunter disturbance was short-lived. In the White River area, this short-lived response seems to characterize south-area elk. In 1996, hunting season opened early, 24 August, in the south area, and closed 14 September. A drop in the proportion of elk on private land occurred after closing date (Fig. 1.4a), indicating that some elk moved back to public land after hunting season closed. This was also seen in elevation shifts of south-area elk: after hunting season closed in 1996, mean daily elevations increased (Fig 1.9a), indicating that elk moved back up to public land after the disturbance ended. Elk in the south treatment area seem to exhibit a short-term response to hunting disturbances. On the other hand, if the elk response to disturbance was a migration shift, then the movement response may last longer. Elk that have learned to move to unhunted refuges and choose migration routes through lightly hunted areas (Altmann 1956, Martinka 1969) may continue these patterns ifno cost was incurred by the change. In the north area, the steady movement may be only partially a direct effect of hunting season. Some elk may learn to move to private land as a conditioned response to avoid hunting danger, and this movement may have become paired with some unconditioned stimuli, such as shortening day length. Because movement in the north area is the beginning of migration, and because timing of migration is thought to be stimulated by daylight cues (Krebs and Davies 1993), north-area elk may now move when the days begin to shorten in addition to the stimulus of hunter activity. As older cow elk have been found to assume leadership roles within a group (Franklin and Lieb 1979), �218 social learning may pass on this pattern to future generations. Thus, elk on the north area may move to private land in response to direct cues of hunter activity and indirect cues of day length. When considering elk movements in response to hunting activity, wildlife managers should consider the type of movements being exhibited by the elk as well as the refuge configuration in the area. Although reducing hunter numbers may reduce movement where elk responses to disturbance are short lived (probably short movements to habitat refuges), different management practices may be needed where migration patterns have become a fixed behavior. If elk have established a migration pattern onto private land, private land hunts, on lands to which elk move, may be a good solution to break the migration pattern and deter early elk movements onto private land. LITERATURE CITED Akaike, H. 1973. Information theory and an extension of the maximum likelihood principle. Pages 267-281 in B. N. Petran and F. Csak:i, editors. International symposium on information theory. Second edition. Akademiai Kiadi, Budapest, Hungary. Altmann, M. 1956. Patterns of herd behavior in free-ranging elk of Wyoming, Cervus canadensis ne/soni. Zoologica 41 :65-71. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net gun to capture large mammals. Wildlife Society Bulletin 10:108-114. .. Boyce, M S. 1991. Migratory behavior and management of elk (Cervus elaphus). Applied Animal Behavioral Science 29:239-250. Boyd, R J. 1970. Elk of the White River Plateau, Colorado. Colorado Division of Game, Fish and Parks Technical Publication 25, Fort Collins, Colorado, USA. Buckland, S. T., K.. P. Burnham, and N. H. Augustin. 1997. Model selection: an integral part of inference. Biometrics 53:603-618. Burnham, K.. P., D. R Anderson, and G. C. White. 1995. Model selection strategy in the analysis ofcapturerecapture data. Biometrics 51 :888-898. Burnham, K.. P., and D. R Anderson. 1998. Model selection and inference: a practical information-theoretic approach. Springer-Verlag, New York, New Yorlc, USA. Bryant, L. D., and C. Maser. 1982. Classification and Distribution. Pages 1-59 in J. W. Thomas and D. E. Toweill, editors. Elk of North America: ecology and management Stackpole Books, Harrisburg, Pennsylvania, USA. Cassirer, E. F., D. J. Freddy, and E. D. Ables. 1992. Elk responses to disturbances by cross-country skiers in Yellowstone National Park. Wildlife Society Bulletin 20:375-381. Czech, B. 1991. Elk behavior in response to human disturbance at Mount St Helens National Volcanic Monument. Applied Animal Behavioral Science 29:269-277. Edge, W. D., and C. L. Marcum. 1985. Movements of elk in reaction to logging disturbances. Journal of Wildlife Management 49:926-930. Edge, W. D., C. L. Marcum, and S. L. Olson. 1985. Effects oflogging activities on home-range fidelity of elk. Journal of Wildlife Management 49:741-744. Franklin, W. L., and J. W. Lieb. 1979. The social organization ofa sedentary population of North American elk: a model for understanding other populations. Pages 185-198 in M. S. Boyce and L. D. Hayden-Wing editors. North American elk: ecology, behavior, and management. University of Wyoming, Laramie, Wyoming, USA. Freddy, D. J. 1987. The White River elk herd: A perspective, 1960-85. Colorado Division of Wildlife Technical Publication 37, Fort Collins, Colorado, USA. Gray, J.P., G. Byrne, and J. Madison. 1994. White River elk data analysis unit plan: game management units: ll,211,12,13,i31,231,23,24,25,26,33. Colorado Division of Wildlife Internal Report, Grand Junction, Colorado, USA. .. Hershey, T. J., and T. A. Leege. 1982. Elk movements and habitat use on a managed forest in north-central Idaho. Idaho Department of Fish and Game Wildlife Bulletin 10, Boise, Idaho, USA. Irwin, L. L., and J.M. Peek. 1979. Relationships between road closures and elk behavior in northern Idaho. Pages 199-204 in M. S. Boyce and L. D. Hayden-Wing editors. North American elk: ecology, behavior, and management. University of Wyoming, Laramie, Wyoming, USA. Knight, RR 1970. The Sun River elk herd. Wildlife Monographs 23. Krebs, J. R, and N. B. Davies. 1993. An introduction to behavioural ecology. Third edition. Blackwell Science Ltd., Oxford, England. �219 Kuck, L., G. L. Hompland, and E. H. Merrill. 1985. Elk calf response to simulated disturbance in southeast Idaho. Journal of Wildlife Management 49:751-757. Leberton, J-D., K. P. Burnham, J. Clobert, and D. R Anderson. 1992. Modeling survival and testing biological hypotheses using marked animals: a unified approach with case studies. Ecological Monographs 62:67-118. Lemke, T. 0. 1975. Movement and seasonal ranges of the Burdette Creek elk herd, and an investigation of sport hunting. Montana Fish and Game Department Final Report 32.01, Job Number BG-3.15, Missoula, Montana, USA. Martinka, C. J. 1969. Population ecology of summer resident elk in Jackson Hole, Wyoming. Journal of Wildlife Management 33:465-481. Morgantini, L. E., and R J. Hudson. 1979. Human distribution and habitat selection by elk. Pages 132-139 in M. S. Boyce and L. D. Hayden-Wing editors. North American elk: ecology, behavior, and management. University of Wyoming, Laramie, Wyoming, USA. Ott, R L. 1993. An introduction to statistical methods and data analysis. Fourth edition. Wadsworth, Inc., Belmont, California, USA. Ratti, J. T., and E. 0. Garton. 1994. Research and experimental design. Pages 1-23 in S. Hieb, editor. Research and management techniques for wildlife and habitats. Fifth edition. The Wildlife Society, Bethesda, Maryland, USA. SAS Institute. 1990. SAS/STAT® user's guide, version 6.0. Fourth edition. SAS Institute, Inc., Cacy, North Carolina, USA. SAS Institute. 1993. SAS/STAT® software: The GENMOD procedure, release 6.09. SAS® Technical ReportP243, SAS Institute, Inc., Cacy, North Carolina, USA. .. Seber, G. A. F. 1982. The estimation of animal abundance and related parameters. Second edition. Edward Arnold, London, England. Strohmeyer, D. C., and J. M Peek. 1996. Wapiti home range and movement patterns in a sagebrush desert. Northwest Science 70:79-87. Van Dyke, F., and W. C. Klein. 1996. Response of elk to installation of oil wells. Journal ofMammalogy 77:1028-1041. Ward, A. L. 1976. Elk behavior in relation to timber harvest operations and traffic on Medicine Bow Range in south-central Wyoming. Pages 32-43 in S. Hieb, editor. Proceedings of the Elk-Logging-Roads Symposium. University of Idaho, Moscow, Idaho, USA. Wright, K. L. 1983. Elk movements, habitat use, and the effects of hunting activity on elk behavior near Gunnison, Colorado. Thesis, Colorado State University, Fort Collins, Colorado, USA. Zahn, ll M. 1974. Seasonal movements of the Burdette Creek elk herd. Montana Fish and Game Department Final Report 32.01, Job Number BG-3.13, Missoula, Montana, USA. �220 Table 1.1. Description and representation of a priori (models 1-12) and post hoc (models 11-13) models relating effects of year, area. opening of archery hunting season. and Julian date to daily proportion of radiocollared elk found on private land (p,) from 20 July to 10 October in the White River area. Colorado, 1996-1997. 1. HyPOthesis Global model: year, area. opening of archery hunting, Julian day, and all possible interactions 2. No year x date interactions 3. No year x main effects (A and S) interactions 4. No year effects 5. No area x date interactions 6. No area x main effects (Y and S) interactions 7. No area effects 8. No hunt season x date interactions 9. No hunt season x main effects (A and Y) interactions 10. No hunt season effects 11. Only hunt season effects 12. Only hunt season and area effects 13. Only hunt season and area effects with no S x A interaction Model structure • f3o+f31 (Y)+f32(A)+f3iS)+f3iD)+f3s(YxA)+f36(YxS) +f37CY xD )+f3s(Ax S)+f39(AxD)+f3 10(SxD) +f3 11 (YxAxS)+f3 12(YxAxD)+f3 13(YxSxD) +f3 14(AxSxD)+f3 15(YxAxSxD) f3o+f31 (Y)+f3iA)+f3 3(S)+f3iD)+f35(YxA)+f36(Y xS) +f3i{AxS)+f3s(AxD)+f39(SxD)+f3 10(YxAxS) +f3 11 (AxSxD) f3o+f31(Y)+f32(A)+f33(S)+f3iD)+f3 5(YxD) +f36(AxS)+f3i{AxD)+f3s(SxD)+f39(AxSxD) f3o+f3,(A)+f32(S)+f3iD)+f3 4(AxS)+f35(AxD) +f3iSxD)+f3i{AxSxD) f3o+f31 (Y)+f32(A)+f3iS)+f3iD)+f3s(YxA)+f36(YxS) +f37CY"xD)+f3s(AxS)+f3 9(SxD) +f3 10(YxAxS)+f3 11 (YxSxD) .. f3o+f31(Y)+f32(A)+f33(S)+f34(D)+f3 5(YxS)+f3~xD) +f3i{AxD)+f3s(SxD)+f39(YxSxD) f3o+f3,(Y)+f32(S)+f33(D)+f34(YxS)+f3 5(YxD) +f3iSxD)+f37CYxHxD) f3o+f3, (Y)+f32(A)+f3iS)+f34(D)+f3s(YxA)+f36(Y xS) +f37CY'xD)+f38(AxS)+f3 9(AxD)+f3 10(YxAxS) +f3 11 (YxAxD) f3o+f3,(Y)+f32(A)+f33(S)+f34(D)+f3s(YxA)+f36(YxD) +f3i{AxD )+f3s(SxD)+f39(YxAxD) f3o+f3,(A)+f32(Y)+f33(D)+f3iAxY)+f3 5(AxD) +f36(YxD)+f3i{Ax YxD) f3o+f3,(S)+f32(D)+f3iSxD) f3 0+f3 1(A)+f3iS)+f33(D)+f3iAxS)+f3 5(AxD)+f36(SxD) f30+f3 1(A)+f3 2(S)+f33(D)+f3 4(AxD)+f3 5(SxD) • Y represents year with 1996 = 0 and 1997 = 1, A represents treatment area with south area = 0 and north area = I, S represents archery hunting season with O= before opening and 1 = after opening, and D represents the covariate date, which is Julian day. The dependent variable (p,) was in logit scale. �221 Table 1.2. Days between opening dates, least-square mean differences with ±95% CI, and P-value for the null hypothesis of no difference between mean date of movement by treatment, year, and area for radiocollared elk in the White River study area, Colorado 1996-1997. Days between opening dates Treatment effect: (late treatment - early treatment) Year effect: (1997 - 1996) Area effect: (south area - north area) Days between mean dates of movement p 20 10±9 1 0 4±9 7±8 0.013• 0.346 0.100 • One-sided test Table 1.3. Differences between opening dates, least square mean differences with ±95% CI. and one-sided P-value for the null hypothesis of no difference between mean date of movement for radiocollared elk in the White River study area, Colorado 1996-1997. Treatment area North South Days between Opening dates 20 20 Days between mean dates of movement (late treatment - early treatment) p 14 ± 10 6 ± 15 0.006 0.218 �222 Table 1.4. Ranking of a priori (models 1-12) and post hoc (models 11-13) hypothesized models relating effects of year, area, opening of archery hunting season, and Julian day to daily proportion of radiocollared elk found on private land (p,) from 20 July to 10 October in the White River area, Colorado, 1996-1997. Models were ranked by QAICc values and normalized QAICc weights ( wm ). Model number b Kc QAICc AQAICc Wm 13o+13,(A)+l3i(S)+l3iD)+l3iAxD)+l3lSxD) 13 6 4718.68 0.00 0.495 13o+l31(A)+l32(S)+l3J(D)+l3iAxS)+l3£AxD) +l3 6(SxD) 12 7 4720.53 1.85 0.196 13o+l31(Y)+l3i(A)+l3J(S)+l3iD)+l3£YxA) +l3 6(YxD)+l3-;(AxD)+l3s(SxD)+l3g(YxAxD) 9 10 4721.92 3.24 0.098 13o+l3i(A)+l3i(S)+l33(D)+l3iAxS)+l3lAxD)+l3iSxD) +l3-;(AxSxD) 4 8 4722.42 3.74 0.076 13o+l31(Y)+l3i(A)+l3iS)+l34(D)+l3£YxS) +l3 6(YxD)+l3-;(AxD)+l3s(SxD)+l39 (YxSxD) 6 10 4723.37 4.69 0.048 13o+l3,(Y)+l3i(A)+l3iS)+l34(D)+l3£YxA) +l3 6(YxS)+l3-;(AxS)+l3s(AxD)+l3g(SxD) +l3 10(YxAxS)+l3 11 (AxSxD) 2 12 4723.57 4.90 0.043 13o+l3,(Y)+l3i(A)+l3iS)+l34(D)+l3iYxD) +l3iAxS)+l3-;(AxD)+l3 8(SxD)+l3lAxSxD) 3 10 4724.11 5.43 0.033 13o+l3,(Y)+l3i(A)+l3iS)+l34(D)+l35(YxA)+l3 6(YxS) +l3-;(YxD)+l3s(AxS)+l3g(SxD)+l3 10(Y xAxS) +l3 11 (YxSxD) 5 12 4726.6 7.92 0.009 1 13o+l31(Y)+l3i(A)+l3JCS)+l34(D)+l35(YxA)+l3 6(YxS) +l3-;(YxD)+l3s(AxS)+l3iAxD)+l3 10(SxD)+l3 11 (YxAxS) +l3 12(YxAxD)+l3 13(Y xSxD)+l3,◄<AxSxD) +l3 15(YxAxSxD) 8 130+131 (Y)+f3i(A)+l3J(S)+l34(D)+l3 5(YxA)+l3 6(Y x S) +l3-;(YxD)+l3s(AxS)+l3iAxD)+l3 10(Y xAx S) +l3 11 (YxAxD) 16 4730.41 11.74 0.001 12 4733.36 14.68 0.000 Model structure • 130+131 (A)+l32(Y)+f3 3(D)+13 i Ax Y)+l3£AxD) +l3 6(YxD)+l3-;(Ax Y xD) 10 8 4745.57 26.89 0.000 13o+l31(Y)+f3i(S)+l33(D)+l34(YxS)+l3 5(Y xD)+l3 6 (SxD) +l3-;(SxYxD) 7 8 4845.74 127.07 0.000 4855.07 136.39 11 4 0.000 13o+l3i(S)+l32(D)+l3J(SxD) • Y represents year with 1996 = 0 and 1997 = 1, A represents treatment area with south area= 0 and north area= 1, S represents archery hunting season with 0 = before opening and 1 = after opening, and D represents the covariate date, which is Julian day. The dependent variable (p,) was in logit scale. b Numbers correspond to those in Table 1.1 c Number of estimable parameters. �223 Table 1.5. Model-averaged estimates of hunt season effect represented by the difference of proportion of radiocollared elk on private land immediately before and after opening day and 95% confidence intervals; White River area, Colorado, 1996-1997. rrreatment Area 1997: North -early 1996: North - late 1996: South - early 1997: South - late Estimated oronortion of elk on orivate land Immediately Immediately Difference 95%CI after oneninv; of difference before oneninv; 0.095-0.257 0.408 0.584 0.176 -0.001-0.158 0.542 0.621 0.078 0.089-0.231 0.256 0.416 0.160 0.317 0.416 0.099 0.028-0.170 Table 1.6. Mean proportion ofradiocollared elk on private land during 1996-1997 for each treatment area in the White River area, Colorado, at the beginning (19 July) and end (10 October) of the stud.y period. Mean proportion of elk on private land 1996-1997 Treatment area North South July 19 0.200 0.161 October 10 0.636 0.393 Change 0.436 0.232 95% CI of change 0.296-0.577 0.105-0.360 �224 t N km 0 .~ ra........ . ... I 10 20 BLM Land Flat Tops Wilderness GMU Boundaries J$ State Wildlife Land D study Area Boundary II Forest Service Land White River .• Towns Colorado Figure 1.1. Alleged elk movements from high elevation public land to lower elevation private land, and land ownership in the White River study area, Colorado. 7000 - 6000 C 5000 ~ (I) C ,,. - . -.... :::, .c 0 ' u, cu (1) u, 4000 I >, "C cu 3000 0 2000 (1) L.. (1) .a § 1000 z 0 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 Year * Variance estimates not available 1984-1987. Figure 1.2. Number of archery and muzzleloading hunters using GMUs 12, 23, 24, and 33 in northwest Colorado, 1984-1994. Data from Deer, Elk, and Antelope Management (DEAMAN) database and software, Colorado Division of Wildlife, unpublished data. �225 t N .I! BLM Land fiffl Flat Tops WIiderness ~ State WIidiife Land D Study Area Boundary km 0 10 20 ■· Forest Service Land • Elk Capture Locations •· Towns Figure 1.3. North and south treabnent areas and July 1996 elk capture locations in the White River study area, Colorado. �226 Early Opening Late Opening a C 0.9 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 "'C C ca Q.) ca > ·c: 'J••--- •• ! -~- ·······-··· .'!.------------- ♦----- •.•. South 96 <X) ~ ;::: C. C !12 Opening date 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 Opening date ~ ~ cc ~ North 97 Cl) II) ~ ;:::: 0 ~ Q) 0 C 0 :e0 C. e a.. b d 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 Opening date North 96 Cl) Opening dale ------♦-- ......... -----·-•-·-•·· South 97 Cl) I() I() ;:::: ~ ~ i:: ~ ♦ Raw data Predicted --•95%CI Figure 1.4. Model averaged predicted values, 95% CI, and raw data of proportion ofradiocollared elk on private land versus date for (a) early treatment on the south area 1996, (b) early treatment on the north area 1997, (c) late treatment on the north area 1996, and (d) late treatment on the south area 1997, July-October in the White River study area, Colorado. �227 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 a Earl!' opening filled curve\ ■ •••• ■ ~ • ■ • I .I:. ■ •••• • ■ ■ •■ ~- -■•• • -■ ■ ■ ■ ii Late opening • Early opening Late opening fitted curv ----CX) IC) T"" CX) IC) T"" N CX) CX) T"" ...... ...... 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 CX) N --- 0) ~ ~ CX) CX) IC) N 0) T"" T"" 0) 0) co 0) - <-> 0 0 ~ 0) T"" 0 T"" T"" Date b ■ Late opening • Early opening Early opening fitted curve ■ • CX) IC) T"" ~ j:::: • • ••• ■ • ■ •• ■ .. - T"" CX) CX) a> IC) T"" CX) ■ N •• ■ - -- ■ ■ ■ • ■ ■ - -■ - •• ■ \ Late opening filled curve 0) !!? 0) ~ ~ Date -N 0) T"" T"" 0) 0) co ~ 0) <-> 0 T"" 0 T"" 0 T"" Figure 1.5. Model averaged predicted values and raw data of proportion of radiocollared elk on private land versus date for (a) north-treatment area and (b) south-treatment area during July-October 1996-1997 in the White River study area, Colorado. 1600 ~ - - - - - - - - - - - - - - - - - - - ~ -Soulh early 1996 Early opening Late opening -c. ._as 1400 --North early 1997 E ·-····· South late 1997 >- -c 1200 CD ~ 1000 (I) g 800 (ti .., -~ "'C ~ 'iii "'C C (ti (I) 600 400 200 :::E 0 co ..... ..... ,..._ I() ('.I ..... ,..._ ..... ..... co co ..... co I() ..... ..... co as ('.I ..... oi 0) ..... ..... 0) (0 ~ ~ 0 ..... 0 ..... 0 ..... Date Figure l.6. Mean distance moved per day by radiocollared elk during July-October 1996-1997 in the White River study area, Colorado. �230 �231 CHAPTER 2: EFFECT OF HUNTER DENSITY ON ELK MOVEMENT TO PRIVATE LAND IN NORTHWEST COLORADO INTRODUCTION Animals may modulate their response to stimuli depending on the quality, quantity, context, and intensity of the stimuli. Prey species, such as elk (Cervus elaphus), may show responses to varying densities of predators, such as hunters. Elk may increase their use of a refuge to avoid hunters; that is, as the density of predators increases, so too does the probability that elk are found in a refuge area. Or, elk may ignore hunters at low densities, and move to refuge areas only when hunter densities increase above some tolerance and danger threshold. To avoid hunters, elk have moved to refuges in national parks (Altmann 1956, Martinka 1969), private land (Wright 1983, Chapter 1), or densely vegetated areas (Irwin and Peek 1979, Morgantini and Hudson 1979) adjacent to their typical areas of activity. However, little research as been done on general elk responses to differing levels of hunter pressure, and no research has directly evaluated the effects of hunter density on elk use of refuge. Since the 1960s, the number of early-season (archery and muzzleloading) hunters using the White River area of northwest Colorado has grown, with especially significant increases between 1984 and 1992 (Fig. 2.1, DEAMAN, Colorado Division of Wildlife, unpublished data). Typically, archery hunters did not hunt private lands; thus, private lands provided elk a refuge during early-season hunts. As archery and muzzleloading hunter numbers increased, so too did complaints about early elk movements onto private land. For the Colorado Division of Wildlife (CDOW), the primary problem was herd size: it was difficult for CDOW to maintain or reduce herd size via harvest because hunters have limited access to private lands, (J. Madison and C. Reichert, Colorado Division of Wildlife, personal communication). Managers hypothesized that reduction of early-season hunter pressure, achievable through limiting licenses, would reduce movement onto private land. As part of an experiment about elk movement in response to early-season hunting activity, CDOW conducted an intensive hunter survey in the fall of 1996 and 1997. The study area was split into 2 treatment areas. A telephone survey of early-season hunters was conducted to estimate the daily number of hunters using each treatment area with a 95% confidence interval of ±150 hunters (total confidence interval of300 hunters). I used these data, along with elk location data, to determine how the density of early-season hunters affected elk movement to private land. My objectives were to: (1) summarize hunter survey data and provide estimates of hunter numbers, hunter-days, and mean days hunted per hunter by treatment area, Game Management Unit (GMU), hunt code, and day, (2) compute these estimates for hunters using ATVs, and (3) evaluate the relationship between hunter density and elk movement to private land. I used logistic regression and Akaike's Information Criteria (AIC) model selection (Burnham and Anderson 1998) to evaluate the relationship between hunter density and elk movement to private land. STUDY.AREA The White River area of northwestern Colorado covered approximately 4,540 km2 and was composed of Game Management Units (GMUs) 12, 23, 24, and 33. In Colorado, GMUs were delineated to distribute hunters through allocation of hunting licenses. White River area land ownership was 34% private land and 66% public land (Fig 2.2), with 82% of public land being United States Forest Service (USFS). USFS land was mostly at high elevation and toward the center of the study area, with some Bureau of Land Management (BLM) areas at lower elevations in the southern sections of the study area. Private land was lower in elevation and comprised mainly ranches and coal mines. Topography, climate, and vegetation varied widely throughout the study area. Elevation ranged from 1,629 to 3,700 m. The central part of the study area was high elevation public land, split by the White River valley. The high elevation areas dropped off to lower elevations to the north, south, and western edges of the study area. Higher elevations had severe winters with heavy snowfall, while lower elevations had comparatively mild winters. Mean annual precipitation at 3,000 min the National Forest areas was about 100 cm, compared �232 with about 30 cm at lower elevations of <2,000 m. Vegetation types at elevations >2,600 m were in the montane/subalpine zone and included groves of aspen (Populus tremuloides), Engelmann spruce (Picea engelmanni), and alpine fir (Abies lasiocarpa) interspersed with grassy meadows. Middle elevations (1,9802,600 m) were a transitional zone that consisted primarily of pinion pine (Pinus edu/is),juniper (Juniperos scopu/orum), and big sagebrush (Artemisia tridentata). Lower elevations (<2,000 m) in the southern and northern parts of the study area were in the Great Basin zone with sage grasslands. Oakbrush patches were found at the middle and lower elevations with major shrubs that included Gambel's oak (Quercus gambelii), servicebeny (Amelanchier alnifolia), mountain mahogany (Cercocarpus montanus), chokecherry (Prunus virginiana), snowberry (Symphoricarpos utahensis), bitterbrush (Purshia tridentata), and rabbitbrush (Chrysothamnus nauseosus). Higher and middle elevation areas provided summer and fall forage for elk, while lower elevations were typically used by elk during winter. Boyd (1970) provides a detailed description of the study area Hunting for large game was economically important to local residents in the study area (Gray et al. 1994). Guides, outfitters, local service sectors, and private landowners (who sell trespass perm.its) made a large proportion of their annual income during the elk and deer hunting seasons, with most hunting revenues accrued during rifle hunting season METHODS Radio-telemetry Data Collection As part of a study on effects of early-season opening date on elk movements to private land, the study area was divided roughly east-west into 2 halves along the White River and North Fork of the White River (Fig. 2.2). GMU 12 and parts of GMUs 23 and 24 composed the north treatment area, while GMU 33 and parts of GMUs 23 and 24 composed the south treatment area. Two treatments were applied: an archery season that opened 1 week earlier, and another that opened 2 weeks later than normal, yielding a 3-week difference in opening dates. Each area received both an early- and late-opening treatment During the first year of the study, the early-opening treatment was randomly assigned to the south treatment area. Treatments were reversed the second year of the study. Thus, in 1996, archery hunting opened early, 24 August, in the south area, and late, 14 September, in the north area. When treatments were reversed in 1997, archery hunting opened early, 23 August, in the north area, and late, 13 September, in the south area. Archery season was 4 weeks long during the earlyopening treatment and 3 weeks long during the late-opening treatment. Muzzleloading season was 1 week long and nested within archery season. Opening of muzzleloading season coincided with opening of archery season for the late-opening treatment, and occurred on the third weekend during the early-opening treatment. The 80 adult female elk used in this study were captured from random locations throughout the study area. Chapter 1 describes capture and collaring procedures in detail. Elk locations were collected 20 July-10 October 1996-1997. I relocated radiocollared elk between 0700-1500 hr using a Cessna 182 or 185 fixed-wing aircraft with a 2-element Yagi antenna mounted to each strut of the airplane. For each elk relocation, Universal Transverse Mercator (UTM) coordinates were recorded with a Global Positioning System that was not differentially corrected. During the early-season hunt, I collected elk locations 2 times a week with 2-4 days between collections. A Geographical Information System (GIS) map was used to record elk locations, land ownership, and treatment boundaries. The GIS map was digitized from a United States Geological Survey map at a scale of 1:100,000. For each location of a telemetered cow, a Owas assigned if the location was on public land (e.g. Flat Tops Wilderness Area, USFS, State Wildlife, and BLM lands) and a 1 was assigned if the location was on private land. Hunter Survey Data Collection Number of archery and muzzleloading hunting licenses for elk and deer in 1996 were based on the mean number calculated from 1990 - 1994 (4,985; Colorado Division of Wildlife unpublished data). Restricting licenses kept hunter density consistent during the study and consistent with years of high elk movement complaints. In the fall of 1996 and 1997, after archery and muzzleloading season ended, CDOW conducted a �233 telephone survey with a goal of sampling enough hunters to estimate the number of hunters afield on any day with a 95% confidence interval of ±150 hunters (total confidence interval width= 300 hunters; Appendix 1). Sample size was based on the binomial distribution and was conservative. That is, the maximum variance occurs when 50% of the license holders hunt, and enough license holders were surveyed to meet the ±150 hunters confidence interval under this condition. CDOW conducted the telephone survey following protocols typically used to collect hunter data. The sample was a simple random sample. The initial sample size was larger than required, with an expectation that 70-90% of hunters called will respond with a complete survey. If this requirement was not met, then additional hunters were again randomly chosen from those not sampled the first time. The survey began as soon as the hunting season closed, and continued for approximately 4 months. Telephone surveys have been found to provide valid estimates of harvest and days hunted (White 1993, Steinert et al. 1994). Hunt codes: D-E-012-01-A E-E-012-01-A E-F-012-01-M E-M-012-01-M D-M-012-01-M D-E-033-01-A E-E-033-01-A E-F-033-01-M E-M-033-01-M D-M-033-01-M - archery deer, either sex, on north area, - archery elk, either sex, on north area, - muzzleloading elk, cow, on north area, - muzzleloading elk, bull, on north area, - muzzleloading deer, buck, on north area, - archery deer, either sex, on south area, - archery elk, either sex, on south area, - muzzleloading elk, cow, on south area, - muzzleloading elk, bull, on south area, and - muzzleloading deer, buck, on south area, were surveyed. Hunters that purchased two licenses on the same treatment half (e.g. D-E-012-01-A and E-E012-01-A) were counted as one hunter and were proportionally assigned to hunt-code groups. The number of individuals purchasing licenses was the size of the statistical population sampled for estimates, and was used in all estimates of hunter number, hunter-days, and average days/hunter. Hunters were asked: 1. 2. 3. 4. 5. 6. 7. License identification number, Hunt code; If they hunted or did not hunt; In which GMUs they hunted; How many days they hunted per unit; If they hunted on day I, on day 2, on day 3, etc. of the 7season; and If they used an ATV. From these data, 3 categories of information were estimated for hunt-code group, treatment area, and GMU: Number of hunters that hunted and number of hunters that used ATVs; Number of hunters afield per day, and number of hunters with ATVs afield per day; and Number of hunter-days and average number of days in the field per hunter and per hunter using an ATV. The following notation is used in estimator formulas, with i, j, k, /, and m indexing the group for which the estimator is being used: i .. .., . j k I m - - day (day I is opening day, day 2 is 2nd day of the season etc.), "."' hunt code group (north treatment area contained hunt code groups: DE01201A, DM01201M, EE01201A, EF01201M, EM01201M; south treatment area contained hunt code groups: DE03301A, DM03301M, EE03301A, EF03301M, EM03301M), GMU (12, 23, 24, or 33), individual hunter, and treatment area (north or south half of study area). �234 For example, H;m is the estimated number of hunters afield on day i in treatment m, H;1cm is the estimated number of hunters afield on day i in GMU k, treatment area m, H1an is the estimated number of hunters that hunted in GMU k, treatment area m, etc. Additional notation is: N - N' - n I n - a - Number of individuals purchasing licenses, Number of licenses sold, Number of license holders surveyed who responded with at least partial infonnation, Number of license holders surveyed who reported hunting, fr - Number of license holders surveyed who reported hunting with an ATV, Estimated number of hunters, Estimated number of hunters using ATVs, Number of hunter-days reported for surveyed hunters, Number of hunter-days reported for surveyed hunters using ATVs, Number of hunter-days reported for surveyed hunters in a particular GMU, Estimated number of hunter-days, .. . Estimated number of hunter-days for hunters using A TVs, d f p - Mean number of days afield per hunter, Mean number of days afield per_ hunter using an A TV. Estimated probability that a license holder hunted, and r - Estimated probability that a license holder hunted with an A TV. H A d f u - D - Day i, hunter/, and A TV use were entere~ as~ or I. For example,. if a surveyed license holder hunted, then a I was entered, or, if they did not hunt, then a 0 was entered. Similarly, if a hunter hunted on.day i, then a I was entered, or, if the hunter did not hunt, then a 0 was entered for that day. If a hunter used an ATV . . .. then a I was entered, and the hunter was assumed to use the A TV every day hunting. The finite correction fact?r, (N~~ n) , was used for all estimates. N~te that N' was used in the finite correction because each - license, and not individual, was randomly sampled. Thus, a license holder purchasing 2 licenses could be interviewed for 0, I, or 2 of the licenses. The number of licenses sold varied between I 50-1,219 by hunt code and the finite correction factor varied between 0.45-0.72 (Table 2.4 and 2.5). The 95% confidence intervals were constructed similarly for all variables, e.g.: ii ±t.96.Jvar<.il> , or, where sample sizes were <100, the appropriate t-statistic was used in lieu of 1.96: Estimating Number of Hunters Estimates of number of hunters (if) and their associated variances were calculated for each hunt code group (J) separately, then summed over hunt~ode groups{/) for treatment (m) or GMU (k). The number of hunters from hunt code group (j) that hunted in treatment area (m) was estimated: �235 5 ii m = L ii J , and J=I 5 var(ilm) = L var(il1). J=I Similarly, the number of hunters using ATVs (A) from hunt code group (J) that hunted treatment area (m) was estimated: .~pf;·>= ""'' 1 (N'- -n-) r-(l-r-) J J J J N' ' ni J 5 Am= 2..,A1, and J=l 5 var(Am) = 2.., var(AJ). J=l The number of hunters (ii) from hunt code (J) that hunted in GMU (k) and treatment area (m) was estimated: " n'.k J PJk = - , ni :;...1· Va,.\Pjk ) = (N'-J -n J·) p" J·k(l-p"J-k) , N1 5 ii km = "I.., ii Jk , and J=I n1 �236 s L var(H jk). var(H km)= j=I Similarly, the number of hunters with AlVs (A) from hunt code (j) that hunted in GMU (k) and treatment area (m) was estimated: :;.~A Vcu\rjk ) = (N'-J -nJ·) r-k(l-r-k) J J I Nj nj s A1an = LAjk , and j=I var(Akm) = s L var(A jk). j=I Estimating Number of Hunters Afield per Day Estimates of total number of hunters (if) afield per day(,) and their associated daily variances were calculated for each hunt code group(/) separately, then summed over hunt code groups (j) for treatment (m) . . . . or GMU (k). The number of hunters afield on day (,) within treatment (m) was estimated: A s A H;m = LHij, and j=I var(H;m) = s L va.r(H ij). j=I Similarly, the number of hunters afield using AlVs ( A ) on day(,) within treatment area (m) was estimated: �237 a-I] r.--=1] , A nj var(r--) = (N'- -n -) r--(1-r--) J ~ J I] N'- IJ J I] nJ 5 ~ Aim= LAij ,and j=l 5 var(A;m) =}:: var(Aij). j=l The number of hunters(/) afield (ii) on day(,) in GMU (k), treatment area (m) was estimated: n.::.. i:.Pijkl l=l Pijk = - ' - - - - ' - - nj var(H ijk) = var(N j Pijk) = NJ var(Pijk), ~ 5 ~ Hil,m = L Hijk, and j=l 5 var(Hikm) = L var(H ijk). j=l Similarly, the number of hunters (l) using ATVs ( A ) afield on day (I) in GMU (k), treatment area (m) was estimated: �238 A A;1cm = 5 L Aiik , and A j=l s var(A;km) = L var(Aijk) j=l Estimating Mean Number ofDays per Hunter and Hunter-days - A Estimates of mean days afield per hunter ( d ) and total hunter-days ( D ), and their associated variances were calculated for each hunt code group (j) separately. For each hunt code group (J), mean days afjeld per hunter ( J ) was assumed to be independent of the proportion of license holders that hunted ( p). Estimates of mean days afield per hunter by treatment and GMU were calculated directly, and associated variances were summed over hunt code groups (j) for treatment (m) or GMU (k). The mean days afield per hunter and total number of hunter-days were estimated for hunters (l) from hunt code group (j) that hunted treatment area (m): nm Ldml -dm_-l=l --,-, nm - var(d m) = Ls var(d- j), J=l �239 5 A A Dm=L,Dj,and j=l 5 r, var(i> j). var(i>m) = j=l Similarly, for each hunt code group (j), mean days afield per hunter using an ATV ( j) was assumed to be independent of the proportion of license holders that hunted (; ). The mean days afield per hunter using an ATV ( j ) and total number of hunter-days with ATVs (ft) were estimated for hunters ({) from hunt code group (J) that hunted treatment area (m): F'_j =Ajfj =N/j/j, var(Fj) = var(N j1'j - var(/m) = 7j) = NJ f/ var{] j)+ 7/ var(rj )-(var(rj)var(/ ))], L5 var(/- j) , j=I A 5 A Fm= LFj ,and j=l 5 var(Fm) = L var(Fj) . j=l For each hunt code group (J), mean days afield per hunter ( d ) was assumed to be independent of the proportion of license holders that hunted ( p ). Mean days afield per hunter ( d) and total number of hunter-days ( D) were estimated for hunters ({) from hunt code group (J) that hunted GMU (k) and treatment area (m): �240 - 5 ~ var(d Jan)=}: var(djk), j=I 5 D1an = I, b jk , and j=I 5 var(D1an) = L var(D jk). j=I Similarly, for each hunt code group(/), mean days afield per hunter using an ATV(]) was assumed to be independent of the proportion of license holders that hunted ( r ). Mean days afield per hunter([) using an A TV ( J) and number of hunter-days using ATVs (fr) were estimated for hunters that hunted GMU (k) and treabnent area (m): ijk/ = fjl Pk/, �241 5 A Ljjkajk j=I ~ f1an=--s-"'iajk j=I s A var<}Jan)= A L var<}jk) · j=I s F1an = Iftjk , and j=I s var(F1cm)= Ivar<ftjk)j=I �242 Elk Movement versus Hunter Density I used logistic regression models to evaluate the effects of hunter density on elk movement to private land. The response variable representing elk movement to private land, proportion of elk on private land on day ; (v;), was calculated from elk location data (each elk location was classified as public or private land). Hunter density was the estimated density of hunters afield on day;, in hunters/km2 . Hunter density was not an exact measurement, rather it was an estimate with an error term. Typically, using a predictor variable with an error term results in the regression coefficient biased toward z.ero (Fuller 1987). That is, I would be less likely to detect an effect of hunter density than actually existed. However, if the variance of the predictor variable (hunter density) is large compared to the variance of the estimates, then the effect of the error in the predictor variable can be effectively ignored during regression analysis (Draper and Smith 1966). The variance of hunter density (0.0098) was large compared to the variance of the estimates (maximum variance= 0.0002); hence I ignored the error in hunter density. Because this experiment was not designed to manipulate hunter density, I used model selection to evaluate the effect of hunter density beyond other possible confounding factors. I previously built an extensive suite of models to evaluate the effects of hunt season opening on elk movement to private land (Chapter 1, Table 1.1 ). Instead of generating another long list of models, I began with the 2 best models from the existing set (Chapter 1, Table 1.4, Models 12 and 13); these 2 models were <2 QAICc units from each other and were competing models (Table 2.1, Models 1 and 2). My main goal was to evaluate the effect of hunter density on elk movement to private land, and compare this effect to hunt season opening. I replaced Julian day with hunter density to evaluate whether hunter density was a better predictor of elk on private land on day i than Julian day (Table 2.1, Models 3 and 4). To evaluate the importance of hunt season opening compared to hunter density, I first removed hunt season from Model 3, and then removed hunter density (Table 2.1, Models 5 and 6). Although treatment models did not work well predicting the proportion of elk on private land in the study manipulating opening date, I hypothesized that treatment may provide more information when hunter density was included in a model because hunter density patterns varied between earlyand late-opening treatments. To evaluate this idea, I added treatment to the best model that included hunter density (Table 2.1, Model 7), and replaced area with treatment (Table 2.1, Model 8). Finally, evaluate whether the cumulative interaction with hunters caused elk to move to private land, I replaced hunter density with cumulative hunter density in the best model that included hunter density (Table 2.1, Model 9). Note that none of these models were strictly a priori because I began with previously run best models and built from these best models. I followed the methodology of Burnham and Anderson (1998) to select an appropriate model. I used Akaike's Information Criterion (Akaike 1973), adjusted for overdispersion and corrected for small sample bias (QAICc), as the basis for objectively ranking models and selecting an appropria~l'best approximating" model (Burnham and Anderson 1998). QAICc was defined as: QAICc =- 2~l) + 2K + 2K(K +1), c n-K-1 where In l is the natural logarithm of the likelihood function evaluated at the maximum likelihood estimates for a given model, K is the number of estimable parameters from that model, n is sample siz.e, and cis the estimated overdispersion factor. QAICc was used instead of AICc because repeated measurements were taken on radiocollared elk. Lack of independence in the data, in this case repeated measures on elk, may lead to rr.,, overdispersion or "extra-binomial variation" (Burnham and Anderson 1988). I estimated the overdispel'$.ion ,, ,~ parameter ( c) from the global model, and then used this estimate to adjust (inflate) the variance estimates of model parameters and predicted values, and to calculate QAICc for all the models (Burnham and Anderson 1998). Parameter estimates, c, and QAICc values were calculated using PROC GENMOD (SAS Institute 1993). The sample siz.e, n, was the number of elk locations collected during the study period. �243 The best approximating model was selected based on minimum QAIO::. Models were ranked and compared using AQAIO:: (Leberton et al. 1992, Burnham and Anderson 1998) and normaliz.ed QAIO:: weights (Buckland et al. 1997, Burnham and Anderson 1998). For the suite of models being compared, AQAIO:: was computed for each model as: AQAICc., = QAICc., - QAICc.,,n . where QAIO::. was the QAIO:: value for the mth model and QAIO::""" was the minimum QAIO:: among the suite models being compared. Essentially, AQAIO::,.. is an estimate of the distance between the best approximating model and model m. Normaliz.ed QAIO:: weights (w.) were estimated for each mth model: __!_ An AICc A e 2"""'< - Wm = - - - 1- - - , R --AQAICc I;e 2 r r=l where R refers to the R models chosen for evaluation. AQAIO:: and normaliz.ed QAICc weights were used to· address model selection uncertainty. Generally, models within l-2 QAIO:: units of the selected model were considered competing models for explaining elk movements to private land. RESULTS Hunter Survey Data CDOW surveyed 35-36% of early-season hunters using the study area in 1996 and 1997. Early-season hunters were 77% archers and 23% muzzleloaders. Estimates of hunter use with and without ATVs are provided by treatment (fable 2.2) and GMU (fable 2.3). Estimates are provided by hunt code for 1996 (fable 2.4) and 1997 (fable 2.5). The estimated number of hunters afield per day and number of hunters using ATVs are shown with their 95% confidence intervals by treatment (Fig. 2.3) and GMU (Fig. 2.4). In 1996, daily information was collected for 23 days for each hunter, although hunting season extended 30 days during the early-opening treatment Therefore, 1996 graphs of the south area, where hunting opened early, show only the first 23 days of the season. In 1997, the survey was changed to collect information on hunter use for the entire 30-day season. Appendix 2 contains confidence intervals for daily GMU data. Elle Movement versus Hunter Density The best logistic regression model with the lowest QAIO:: value (fable 2.6, Model 1) was: This was the best model from the manipulative experiment and did not have a hunter density covariate. In fact, the first model with a hunter density covariate was 35.2 QAIO:: units from the top model (fable 2.6, Model .9), and the hunter density regression coefficient was not significantly different than zero in any of the models in which it was included. Model 9, with cumulative hunter density preformed similarly (AQAIO:: < 2) to its counterpart, Model 6 with straight hunter density (Table 2.6). Model 5, with hunter density, preformed poorly (AQAIO:: >> 2) compared to its counterpart, Model 6, with hunt season replacing hunter density (fable 2.6). �244 DISCUSSION By design, hunter use of the study area was constant from year to year and between treabnent areas. However, the daily pattern of use was different for early- and late-opening treabnents. When hunting opened early, hunter use was relatively constant. There were several small peaks coinciding with each weekend during the season, and one larger peak coinciding with the opening of muzzleloading season. When hunting opened late, there was a large peak on opening weekend, partially because muzzleloading season opened with archery season. The late treabnent peaked at higher densities of hunters (0.44-0.50 hunters/km2) compared to the early treabnent (0.28-0.36 hunters/km2). Several hunting studies indirectly provide insight about elk responses to different levels of hunting pressure. Wright (1983) examined elk response to different hunting seasons: archery and muzzleloading, versus 3 different types of rifle hunting seasons. Distances traveled by elk increased with hunter density, and were lowest during archery and muzzleloading season when hunter densities were the lowest. Although hunting method and hunter density were confounded, elk response to hunter density remains a possible explanation for increased elk movements. An earlier study in the White River area, designed to evaluate elk movement onto a coal company's protected land during early-season hunting, found that 13% of the radiocollared elk were on private lands midway through archery and muzzleloading seasons (Camp Dresser and McKee, Inc. 1986). During this study, 50% and 53% of the radiocollared elk were on private land on the same day during 1996 and 1997 respectively. The lack of movement onto private land in the earlier study may be explained by a 63% reduction in archery hunters for that year because of a change in hunting regulations (Gray et al. 1994). Thus, elk movement to private land may decrease when hunting pressure drops. Finally, Zahn (1974), Lemke (1975), and Hershey and Legee (1982) noticed an increase in distances moved by elk during the first 10-12 days of hunting season, followed by normal movement during the remainder of the season. These authors noted that hunting pressure was heavy during the initial 10-12 day of hunting, and relatively light for the remainder of the season. Although all these hunting studies are observational, taken together a pattern emerges; at high hunter densities elk increase their movements to refuge, and at low densities they decrease their movements to refuge. However, there is a confounding between the effects of hunter density and the opening of hunting season opening. That is, the increase in distances moved by elk during the initial 10-12 days of hunting season (Zahn 1974, Lemke 1975, Legee 1982) could be due to high densities of hunters or to the opening of hunting season itself. Elk in the White River area seemed to respond more to the opening of hunting season than to hunter density. That is, models that included a hunt season effect did significantly better than models without hunt season effect (Chapter 1). In contrast, models with a Julian day covariate (fable 2.6, Models 1 and 2) preformed far better than models with a hunter density covariate (fable 2.6, Models 3-5 and Models 7-9). In fact, hunter density can not be considered as a predictor of proportion of elk on private land because models with hunter density were a minimum of35.2 AQAICc units from the top model. Additionally, at the opening of earlyseason hunting, 8-18% of the study elk moved to private land (Chapter 1), while after opening, elk movement to private land was not related to hunter density (Fig. 2.5). Thus, hunter density had no effect on elk movements to private land in the White River area. It may be that hunter density must exceed a threshold before elk begin to respond directly to hunters, and hunter densities did not surpass this threshold during the study. MANAGEMENT IMPLICATIONS If hunter density continues to be a suspected cause of increased elk movement to private land, then an experiment manipulating hunter density should be conducted. Managers concerned with reducing elk movements to private land have an opportunity to evaluate effects of hunting pressure when setting license numbers for a particular area. Additionally, increasing hunter density on private land where problems are occurring may be the most direct strategy to retain elk on public land during late summer. Whatever the management strategy, manipulation of hunter density offers managers a good opportunity to conduct an adaptive management experiment to increase information regarding the management of elk movements to private land. �245 LITERATURE CITED Akaike, H. 1973. Information theory and an extension of the maximum likelihood principle. Pages 267-281 in B. N. Petran and F. Csaki, editors. International symposium on information theory. Second edition. Akademiai Kiadi, Budapest, Hungary. Altmann, M. 1956. Patterns of herd behavior in free-ranging elk of Wyoming, Cervus canadensis nelsoni. Zoologica 41 :65-71. Boyd, R J. 1970. Elk of the White River Plateau, Colorado. Colorado Division of Game, Fish and Parks Technical Publication 25, Fort Collins, Colorado, USA Buckland, S. T., K. P. Burnham, and N. H. Augustin. 1997. Model selection: an integral part of inference. Biometrics 53:603-618. Burnham, K. P., and D. R Anderson. 1998. Model selection and inference: a practical information-theoretic approach. Springer-Verlag, New York, New York, USA Camp Dresser and McKee, Inc. 1986. Meeker PRLA elk migration study: monitoring report, volume 2. Prepared for Consolidation Coal Company. Camp Dresser and McKee, Inc., Denver, Colorado, USA. Draper, N. Rand H. Smith Jr. 1966. Applied regression analysis. Second edition. John Wiley and Sons, New York, New York, USA Fuller, W. A 1987. Measurement error models. John Wiley and Sons, New York, New York, USA Gray, J.P.; G. Byrne, and J. Madison. 1994. White River elk data analysis unit plan: game management units: l l,21 l,12,13,13 l,231,23,24,25,26,33. Colorado Division of Wildlife Internal Report, Grand Junction, Colorado, USA Hershey, T. J., and T. A Leege. 1982. Elk movements and habitat use on a managed forest in north-central Idaho. Idaho Department of Fish and Game Wildlife Bulletin 10, Boise; Idaho, USA Irwin, L. L., and J.M. Peek. 1979. Relationships between road closures and elk behavior in northern Idaho. Pages 199-204 in M. S. Boyce and L. D. Hayden-Wing, editors. North American elk: ecology, behavior, and management. University of Wyoming, Laramie, Wyoming, USA Leberton, J-D., K. P. Burnham, J. Clobert, and D. R Anderson. 1992. Modeling survival and testing biological hypotheses using marked animals: a unified approach with case studies. Ecological Monographs 62:67118. Lemke, T. 0. 1975. Movement and seasonal ranges of the Burdette Creek elk herd, and an investigation of sport hunting. Montana Fish and Game Department Job Final Report 32.01, Job Number BG-3.15, Missoula, Montana, USA Martinka, C. J. 1969. Population ecology of summer resident elk in Jackson Hole, Wyoming. Journal of Wildlife Management 33:465-481. Morgantini, L. E., and R J. Hudson. 1979. Human distribution and habitat selection by elk. Pages 132-139 in M. S. Boyce and L. D. Hayden-Wing, editors. North American elk: ecology, behavior, and management. University of Wyoming, Laramie, Wyoming, USA SAS Institute. 1993. SAS/STAT® software: The GENMOD procedure, release 6.09. SAS® Technical Report P-243, SAS Institute, Inc., Caty, North Carolina, USA Steinert, S. F., H. D. Riffel, and G. C. White. 1994. Comparisons of big game harvest estimates from check stations and telephone surveys. Journal of Wildlife Management 58:335-340. White, G. C. 1993. Precision of harvest estimates obtained from incomplete responses. Journal of Wildlife Management 57:129-134. Wright, K. L. 1983. Elk movements, habitat use, and the effects of hunting activity on elk behavior near Gunnison, Colorado. Thesis, Colorado State University, Fort Collins, Colorado, USA. Zahn, H. M. 1974. Seasonal movements of the Burdette Creek elk herd. Montana Fish and Game Department Job Final Report 32.01, Job BG-3.13, Missoula, Montana, USA. �246 Table 2.1. Description and representation of models relating effects of area, treatment, hunt season opening, hunter density, and Julian date to daily proportion ofradiocollared elk found on private land (v,) from 20 July-10 October in the White River area, Colorado, 1996-1997. Hypothesis Model structure • I. Lowest AICc model from the experimental manipulation of early-season opening date b 2. Second lowest AICc model from the experimental f3 0+f3 1 (A)+f32(S)+f3iD)+f3◄(AxD)+f3 5 (SxD) manipulation of early-season opening date b +f3iAxS) 3. Hunter density is a better predictor than Julian day - Model f3o+f3 1(A)+f3 2(S)+f3J(H)+f3.(AxH) I with hunter density replacing the Julian day covariate • 4. Hunter density is a better predictor than Julian day - Model f3o+f3 1 (A)+f32(S)+f3 3(H)+f3◄(AxH)+f3s(AxS) 2 with hunter density replacing the Julian day covariate • 5. Hunter density is a better predictor than hunt season hunter density and area effects only 6. Hunt season is a better predictor than hunter density - hunt season and area effects only 7. Treatment is an important effect when hunter density is a covariate - add treatment to best model with hunter density 8. Treatment is an important effect when hunter density is a covariate - replace area with treatment 9. Cumulative hunter density better predicts elk movement to private land - use best model that includes hunter density and replace the hunter density covariate with a cumulative hunter density covariate • A represents treatment area with south area= 0 and north area = I, S represents archery hunting season with O = before opening and I = after opening, D represents the covariate date, which is Julian day, and H represents the covariate hunter density on day i in hunters/km2• The dependent variable (v,) was in logit scale. b See Chapter I, Table 1.4. • SxH cannot be estimated because hunter density= 0 before hunt season opening. �247 Table 2.2. Number of licenses sold (N), percent surveyed (s), estimates of hunter numbers ( ii ), hunter-days ( f> ), mean days hunted/hunter ( J ), hunters using ATVs (A), hunter-days with ATVs (ft), and mean number of days/hunter using an ATV ( ] ), by treatment area during archery and muzzleloading season in the White River area l 996-1997. The half-width of the 95% confidence interval is shown for all estimates. fl b Dc 37.2 36.9 1,362 ± 25 1,786 ± 20 8,999 ± 390 10, 716 ±426 6.4 ± I.I 6.0 ±0.8 35.1 1,715 ± 30 10,920 ±455 6.2 ± 1.0 Area Year N• s North 1996 1997 1,711 2,170 1996 2,253 South d A F 209 ± 34 1,555 ± 314 7.2 ±4.1 329 ± 45 2,134 ±392 6.4 ± 2.1 f 300 ±40 2,205 ± 380 7.1 ±2.8 1997 2,090 33.2 1,679 ± 26 11,128 ± 419 6.4 ± 0.9 258 ±40 2,083 ±408 7.6 ± 2.5 • Because some individuals purchased 2 licenses, the number of individuals purchasing licenses, which was used in all estimates, was slightly less than the number of licenses sold. b The estimated number of hunters is lower than the number oflicenses sold because some hunters purchasing licenses did not hunt and because some hunters purchased 2 licenses. c Estimated number of hunter-days and hunter-days with AlVs are sums of estimated hunter-days by hunt code (fables 2.4 and 2.5), and not the direct product of estimated number of hwiters x mean number of days/hunter. Table 2.3. Estimates of hunter numbers (fl), hunter-days (D ), mean days hunted/hunter (d), hunters using ATVs (A), hunter-days with ATVs (F), and mean number of days/hunter using an ATV(]), by GMU during archery and muzzleloading season in the White River area 1996-1997. The half-width of the 95% confidence interval is shown for all estimates. fl• Db F Year Area GMU J A Jc North 12 22 23 773 ±48 1,014 ± 58 437 ± 44 533 ± 52 291 ± 38 426 ±49 4,842 ± 441 6,295 ±484 2,653 ± 355 3,274 ± 483 1,504 ± 277 2,425 ± 401 6.0± 1.8 6.0± 1.2 5.8± 2.0 5.9± 2.2 5.0±2.5 5.3 ± 1.8 74 ±22 117 ± 30 104 ± 25 162 ± 33 75 ±21 102 ± 26 480 ± 181 571 ± 173 676 ±212 1,092 ±333 399 ± 154 643 ±296 6.3 ±7.1 5.0 ±3.1 6.3 ±4.7 6.3 ±3.5 5.3 ±7.5 5.8 ±3.9 1996 3,893 ±409 151 ± 30 890 ±235 5.7 ±3.6 715 ± 53 5.3± 1.8 1997 700 ±201 6.0 ±2.6 494 ± 51 2,770 ±358 5.5 ± 1.9 111 ±28 4.5 ±4.5 23 1996 342 ± 170 3,272 ± 375 5.4± 2.0 75 ±21 600 ± 51 1997 8.5 ± 7.4 577 ± 53 3,514 ±417 5.9±2.4 21 ± 12 195 ± 308 1996 5.7±2.0 132 ± 28 974 ±287 6.9 ±4.8 33 638 ± 52 3,735 ±456 1997 143 ± 31 1,188 ±302 8.0 ±3.1 746 ± 56 4,843 ±452 6.3± 1.4 • All GMU estimates are slightly greater than treatment or hunt code estimates because some hunters hunted in more than oneGMU. b Estimated number of hunter-days and hunter-days with AlVs are sums of estimated hunter-days by hunt code, and not the direct product of estimated number ofhunters x mean number of days/hunter. c Because of the small sample sizes, a log-based confidence interval should be used in some cases so that values for the 95% CI of ] are >O. South 22 1996 1997 1996 1997 1996 1997 �Table 2.4. Estimates of hunter numbers (H ), hunter-days (D ), mean days hunted/hunter (d ), hunters using ATVs (A), hunter-days with ATVs (fr), and mean number of day0iunter using an ATV (] ), by hunt code during archery and muzzleloading season in the White River area 1996. The half-width of the 95% confidence interval is shown for all estimates. Percent of Number of NORTH AREA fr b H d A f licenses sold • hunters surveyed 241 32.0 D-E-012-01-A 663 ± 83 94 ±7 7.1 ± 0.7 18±8 139 ±67 7.6 ±2.0 32.5 E-E-012-01-A 973 874 ± 22 6,368 ± 367 7.3 ± 0.4 149 ± 31 1,198 ± 302 8.1 ± 1.1 51.5 165 E-F-012-01-M 647 ± 69 5.1 ± 0.4 126 ± 8 12 ±6 66±43 5.3 ±2.7 178 48.9 E-M-012-01-M 808 ± 61 4.8 ± 0.4 167 ± 3 21 ±9 103 ± 48 4.8 ± 1.3 46.8 D-M-012-01-M 154 513 ± 59 101 ± 7 5.1±0.4 8±5 50 ±38 6.0 ± 3.2 SOUTH AREA 30.6 526 380 ± 128 D-E-033-01-A 292 ± 13 1,849 ± 187 6.3 ±0.6 59 ± 16 6.4 ± 1.3 1,219 32.0 7,184 ± 400 E-E-33-01-A 1,056 ± 23 6.8 ± 0.4 194 ± 35 1,579 ± 352 8.1 ± 1.1 49.3 150 581 ± 55 16 ±7 E-F-033-01-M 4.8 ± 0.4 86 ±41 5.3 ± 1.0 122 ± 6 47.5 160 5,8 ± 0.4 E-M-033-01-M, 140 ±6 817 ± 70 16±8 93 ±50 5.9 ± 1.6 45.5 198 490 ±77 4.7 ±0.5 14±7 4.9 ± 1.9 D-M-033-01-M 105 ± 12 68±41 • Because some individuals purchases 2 licenses, the number of individual purchasing licenses was slightly less the number of licenses sold. The number of individuals purchasing licenses was used for all estimates. Table 2.5. Estimates of hunter numbers (H ), hunter-days (D ), mean days hunted/hunter (d ), hunters using ATVs (A), hunter-days with ATVs (fr), and mean number of days/hunter using an ATV (] ), by hunt code during archery and muzzleloading season in the White River area 1997. The half-width of the 95% confidence interval is shown for all estimates. Percent of Number of NORTH AREA fr b H "J A ] licenses sold • hunters surveyed 402 54.7 D-E-012-01-A 1,243 ± 63 205 ±5 6.1 ± 0.3 41 ±8 253 ± 57 6.3 ±0.8 1,277 27.9 E-E-012-01-A 1,193±16 7,456 ± 407 6.3 ± 0.3 1,540 ± 377 233 ± 42 6.6 ± 1.1 166 43.4 E-F-012-01-M 134 ± 8 686 ± 70 5.1 ± 0.4 31 ± 11 193 ± 73 6.1±1.0 180 48.3 E-M-012-01-M 850 ± 63 5;3 ± 0.4 160 ± 5 21 ±9 131 ± 60 6.2 ± 1.4 145 45.5 D-M-012-01-M 480 ± 53 94 ±6 5.1 ± 0.5 16 ± 17 5.0 ±0.0 3±3 SOUTH AREA 30.0 D-E-033-01-A 416 223±7 1,503 ± 119 6.8±0.5 43 ± 13 330 ± 118 7.7 ± 1.4 28.4 1,185 E-E-33-01-A 1,058 ±23 7,640 ± 388 7.2±0.3 166 ± 34 1,516 ± 386 7.1±1.2 147 48.3 E-F-033-01-M 121 ±6 584±56 4.8 ±0.4 17 ±7 73 ±34 4.3 ±0.6 155 49.0 E-M-033-01-M 139±4 711 ±57 5.1 ±0.4 55 ± 34 11 ±6 4.8 ± 1.6 D-M-033-01-M 187 46.0 138 ±7 689 ± 63 5,0·± 0,4 21 ±8 109 ±4 9 5, 1 ± 1.2 • Because some individuals purchases 2 licenses, the number of individual purchasing licenses was slightly less the number of licenses sold. The number of individuals purchasing licenses was used for all estimates. N +'- 00 �249 Table 2.6. Ranking of models relating effects of area, treatment, hunt season opening, hunter density, cumulative hunter density, and Julian day to daily proportion ofradiocollared elk found on private land (v,) from 20 July-10 October in the White River area, Colorado, 1996-1997. Models were ranked by QAICc values and normalized QAICc weights ( ww ). Model number b Kc QAICc AQAICc wm f3 0+f3 1(A)+f3 2(S)+f33(D)+f3.(AxD)+f3s(SxD) +f3iAxS) 1 2 6 7 3760.28 3762.26 0.00 1.97 0.729 0.271 f3o+f31(A)+f32(S)+f3iC)+f3.(AxC) +f3s(AxS) 9 6 3795.45 35.17 f3o+f31(A)+f32(S)+f33(Ax S) f3o+f31(A)+f32(S)+f3iH)+f3.(AxH) +f3s(AxS) 6 4 3 7 4 6 3796.29 3798.59 36.01 38.31 40.13 0.000 0.000 0.000 Model structure • f3 0+f3 1(A)+f3 2(S)+f3 3(D)+f3.(AxD)+f3s(SxD) 5 3800.41 0.000 3802.34 8 42.06 0.000 4 5 3867.39 107.11 0.000 f3o+f31(A)+f32(D)+f3 3(Ax D) 3897.96 8 6 137.68 0.000 f3o+f31m+f32(S)+f33(H}+f34(fxH) +f3s(fxS) • A represents treabnent area with south area= 0 and north area= 1, S represents archeiy hunting season opening with before opening= 0 and after opening= 1, T represents treabnent with early opening= 0 and late opening= 1, D represents the covariate date, which is Julian day, H represents the covariate hunter density on day i in hunters/km2, and C represents the covariate cumulative hunter density on day i in hunters/km2. The dependent variable (v,) was in logit scale. b Numbers correspond to those in Table 2.1 c Number of estimable parameters. f3o+f3.(A)+f32(S)+f3iH)+f3.(AxH) f3o+f31(A)+f3 2(S)+f3iH)+f3 4(f)+f3s(AxH) +f3iAxS) +(3-ifxH) - 7000 f! Q) ______ , C 6000 . ~ .c. C 0 5000 U) co Q) C{) 4000 >, 1a 3000 Q) 0 2000 L- Q) .c ~ z 1000 0 1984 1985 1986 1987 1988 1989 1990 199119921993 1994 Year * Variance estimates not available 1984-1987. Figure 2.1. Number of archery and muzzleloading hwiters using GMUs 12, 23, 24, and 33 in Colorado, 19841994. Data from Deer, Elk, and Antelope Management (DEAMAN) database and software, Colorado Division of Wildlife wipublished data. �250 11· BLMLand llJ: Flat Tops Wilderness a. 0 10 State Wildlife Area □ Treatment Area Boundary Km 20 -II Forest Service Land Figure 2.2. North and south treatment areas and land ownership in the White River study area, Colorado. �251 a C Early Opening 1,200 Late Opening 1,200 Opening of muzzleloading End of muzzleloading 1,000 1,000 800 -- "'C ai ii: cu 600 800 South 1996 1 ~ 600 400 ~ 400 Cl) .c 200 :::s 0 E C "'C scu ...,. ~ m .... ~ ....a;...,. ~ 0 .... a1 b .. _ -- -- -- -- -- .... a1 ....a;...,. m a1 ....~ ,.._ a.... d 1,200 1,000 1,000 w -- - - . - 200 •- ---- ·-·- ·------- --••••·- ::::::::::-- -- = - = , , ~ E 1,200 .....u, ·- North 1996 800 .. 600 800 North 1997 600 400 400 200 200 ------ ---- ---· ··-- ---- ·- 0 C') N cc "'• •'"' •• •• ------ -- ·· -- ....a; C') ....a; C') 0 a1 0 a1 a1 Date LEGEND Hunters Hunters using A TVs •••••••••••• 95%CI Figure 2.3. Estimated daily number of hunters and hunters using A TVs in the (a) south treatment area 1996, (b) north treatment area 1997, ( c) north treatment area 1996, and (d} south treatment area 1997, during archery and muzzleloading season in the White River area, Colorado. The first date is opening date. Each date shown represents a Saturday. 1 There is a break in estimates for the south area in 1996 because survey data were only collected for 23 out of the 3'0 days of the early treatment. The survey was extended to 30 days in 1997. ,.": �252 a 1,000 1.000 "O B 1 A 800 co ii= 1997 C 1996 600 600 400 400 .0 200 200 ::l 0 ...cu (1) E C - ... - - ..... ~ ~ Cl) "O ..... I::: oi 0 - Cl) !!! ~ ~ CJI C B C ~ d 1,000 A B 1 C A 800 I I 600 400 400 200 200 0 B 0 a, E 600 .., ~ a Jg b cu 1,000 800 A 800 C ·- ... --.... :.-:, ··--•.• ···~-~-::,... -.r CJI 0 a, ~ "' 0 .., ~ ---- .. a t! ID Date LEGEND -GMU12 ......... GMU 23 --GMU24 -GMU33 A Opening date of archery season on early-treatment B Opening date of muzzleloading season and archery season on late treatment. C Closing date of muzzleloading season. Figure 2.4. Estimated daily number of (a) hunters 1996, (b) hunters using ATVs in 1996, (c) hunters 1997, and (d) hunters using A TVs 1997, during archery and muzzleloading season in the White River area, Colorado. Each date shown represents a ~aturday. 1 There is a break in estimates for GMU 23, 24, and 33 in 1996 because survey data were only collected for 23 out of the 30 days of the early treatment. The survey was extended to 30 days in 1997. �253 CHAPTER 3: ELK LOCATIONS IN RELATION TO DOMESTIC SHEEP BANDS AND A METHOD TO TEST FOR AVOIDANCE OR ATTRACTION BE1WEEN ANIMALS INTRODUCTION Hearsay, accusation, tall tales, and rumor dominate the discussion about Rocky Mountain elk (Cervus elaphus nelsoni) movements in response to domestic sheep (Ovis aries). Elk movement responses to sheep are a concern of hunters, who feel that livestock grazing on public land is a disturbance that elk avoid Hunters claim that domestic sheep force elk away from public hunting areas and onto private land, but there is little documentation of elk responses to domestic sheep. In one study of elk and livestock grazing, Clegg (1994) found that elk densities decreased rapidly after introduction of sheep, herders, and dogs. However, the distance of impact was not recorded making it difficult to assess whether sheep could move elk away from public hunting areas. Most studies of elk and sheep have originated from livestock concerns and concentrated on dietary overlaps and forage competition (Pickford and Reid 1943, Nichols 1957, Stevens 1966, Olsen and Hansen 1977, MacCracken and Hansen 1981, Beck et al. 1996). In the White River area, elk and domestic sheep graz.e high mountain meadows. Approximately 26 bands of domestic sheep used the study area between May and October. Early-season hunting (archery and muzzleloading), which occurs between late August and mid-October, overlaps with sheep grazing. During meetings about elk movements in response to early-season hunting, archery hunters expressed concern that domestic sheep cause elk to move off public land to private land during the early season. As part of the stakeholder process, CDOW agreed to collect locations of sheep bands in the area when collecting locations of 80 radiocollared elk for an archery hunting study. My objectives were to (1) determine if elk avoided sheep at several spatial scales and (2) develop a methodology applicable to radio-telemetry data to evaluate attraction or avoidance between animals. I used location data from radiocollared elk and visual locations of sheep bands from the summer and fall of 1996 and 1997. The null hypothesis was that elk locations were random with respect to sheep locations. I used nonparametric statistics and randomization techniques to calculate, under the null hypothesis, the probability that elk avoided sheep. STUDY .AREA The White River area of northwestern Colorado covered approximately 4,540 km2 and was composed of Game Management Units (GMUs) 12, 23, 24, and 33. In Colorado, GMUs were delineated to distribute hunters through allocation of hunting licenses. White River area land ownership was 34% private land and 66% public land (Fig. 3.1), with 82% of public land being United States Forest Service (USFS). USFS land was mostly at high elevation and toward the center of the study area, with some Bureau of Land Management (BLM) areas at lower elevations in the southern sections of the study area. Private land was lower in elevation and comprised mainly ranches and coal mines. Topography, climate, and vegetation varied widely throughout the study area. Elevation ranged from 1,629 to 3,700 m. The central part of the study area was high elevation public land, split by the White River valley. The high elevation areas dropped off to lower elevations to the north, south, and western edges of the study area. Higher elevations had severe winters with heavy snowfall, while lower elevations had comparatively mild winters. Mean annual precipitation at 3,000 min the National Forest areas was about 100 cm, compared with about 30 cm at lower elevations of <2,000 m. Vegetation types at elevations >2,600 m were in the montane/subalpine zone and included groves of aspen (Popu/us tremuloides), Engelmann spruce (Picea enge/manni), and alpine fir (Abies lasiocarpa) interspersed with grassy meadows. Middle elevations (l,9802,600 m) were a transitional zone that consisted primarily of pinion pine (Pinus edu/is),juniper (Juniperus scopulorom), and big sagebrush (Artemisia tridentata). Lower elevations (<2,000 m) in the southern and northern parts of the study area were in the Great Basin zone with sage grasslands. Oak.brush patches were found at the middle and lower elevations with major shrubs that included Gambel's oak (Quercus gambelii), serviceberry (Amelanchier alnifo/ia), mountain mahogany (Cercocarpus montanus), chokecherry (Prunus �254 virginiana), snowbcrry (Symphoricarpos utahensis), bittcrbrush (Purshia tridentata), and rabbitbrush (Chrysothamnus nauseosus). Higher and middle elevation areas provided summer and fall forage for elk, while lower elevations were typically used by elk during winter. Boyd (1970) provides a detailed description of the study area. USFS issued permits for sheep grazing on national forest lands on the study area. During the study period, approximately 26 bands of domestic sheep used the study area. Each band was assigned to a particular area called an allotment. Bands ranged in size from 750-1,350 sheep, and averaged approximately 1,043 sheep (95% CI= 981, 1,106; USFS unpublished data from 1992-1995). Band sizes are approximate because actual numbers were often less than permitted numbers (Mary Massey, United States Forest Service, Meeker Colorado, personal communication). Sheep arrived on the study area 16 June-I I July, and left 10 September-15 October. Herders and dogs stayed with the sheep bands and moved them through their allotment area. METHODS Data Collection The adult female elk used in this study were captured from randomly selected locations throughout the study area. Capture and collaring procedures are descried in detail in Chapter 1. Elk and sheep locations were collected 20 July-10 October 1996-1997. I relocated radiocollared elk between 0700-1~00 hr using a fixed-wing aircraft with a 2-element Yagi antenna mounted to each strut of the airplane. For each elk relocation, Universal Transverse Mercator (U1M) coordinates were recorded with a Global Positioning System. I collected elk locations 2 times a week with 2-4 days between collections. During each flight, the position of at least 1 sheep band was recorded. Because elk-sheep interactions were not the focus of the study, locations of sheep bands were opportunistically chosen. That is, I typically recorded the location of only one band of sheep seen and did not record any other bands, although up to 3 bands per day were sometimes seen. Thus, each band location was a sample of the population of 26 bands. Because I did not collect locations in a specific route each flight, the daily sample of sheep bands was fairly random (Fig. 3.1). Once a sheep band was seen, I flew over the band and recorded lITM coordinates at the approximate center of the band Data Analysis ·r r· All analyses were computed separately for each year. Distances between elk and sheep bands were calculated from UTM coordinates as straight-line distances. For each day a sheep band was located, I calculated the distance between the band and every radiocollared elk located that day. I called these paired distances because they were paired in time (same day). I then calculated the distance between the location of the sheep band, located that day, and the locations of each radiocollared elk located every other day. I called these unpaired distances because they were temporally unpaired During the study period, 20-23 locations were collected on each elk per year. Thus, for each elk, there was I paired distance and 19-22 unpaired distances for each day locations were collected. Note that there was no way to compute the variance on I measurement; hence paired and unpaired distances could not be compared using parametric methods. However, a rank representing the paired distance compared to the unpaired distances, between 1 elk and I sheep band, represents I sample. Thus, I used a ranking method to evaluate avoidance between elk and sheep bands. For each elk and each day locations were collected, I ordered distances, paired and unpaired, from smallest to largest, and assigned a ranking to each distance. That is, the smallest distance between a sheep band and an elk was assigned a 1, the next smallest distance a 2 and so on. Thus, for each elk and each day data were collected, there was a ranking of I to number of distances collected on that elk (Fig. 3.2). If elk were located randomly with respect to sheep bands, then paired distances·would be equal to unpaired distances, and the mean rank of paired distances would be equal to the mean rank unpaired rankings. If elk avoided sheep, then paired distances would be farther from sheep, on average, than unpaired distances, and the mean rank of paired distances would be significantly greater than the mean rank of unpaired rankings. Conversely, if elk were attracted to the sheep, then paired distances would be closer to sheep, on average, than unpaired distances, and the mean rank of paired distances would be significantly less than the mean rank of unpaired rankings (Fig. 3.3). This assumes that elk do not avoid an area where sheep were located after the sheep leave the area. �255 For each year, I extracted the rank of the paired distances for each elk and day that both the elk and sheep band were located, and calculated the mean ranking of the paired distances. The statistical expressions of the null and alternative hypothesis were: Ho: The mean rank of paired distances = the expected mean rank of unpaired ranks, and HA: The mean rank of paired distances> the expected mean rank of unpaired ranks, which tested the study hypothesis: Ho: Locations of elk were random with respect to locations of sheep; elk did not avoid sheep, and HA: Locations of elk were father to sheep than expected at random; elk avoided sheep. Note that it was possible that elk were attracted to sheep because sheep are graz.ed in areas preferred by elk for grazing, but this was not the question of interest. In order to evaluate the probability of the observed paired rank, I generated the distribution of paired ranks under the null hypothesis that elk and sheep were located randomly with respect to each other. To develop the null distribution, I randomiz.ed on sheep bands. That is, I randomly assigned a day to each sheep band. For example, a sheep band located on day 2, may be randomly assigned to day 6. After randomly assigning a day to each sheep band, paired and unpaired distances were calculated as described for the observed data. From each sample, I extracted the ranks of the randomly "paired" rank. From the collection of samples via the randomization procedure, I generated the distribution of mean ranks for paired elk and sheep locations that were random, and created a null distribution representing no avoidance (or attraction) between elk and sheep. For each randomiz.ed sample, the mean rank was saved and the randomization was re-run until there were 999 mean rankings. The mean of the 999 randomiz.ed mean ranks was the expected mean rank. The observed mean rank was merged with the randomiz.ed mean ranks, and the mean ranks were ordered from largest to smallest. The percentile of the observed sample was the P-value of the hypothesis test; that is, the probability of observing the rank of the sample or a larger rank under the null hypothesis that elk are located randomly with respect to sheep (Mooney and Duval 1993). Spatial scale may influence the ability to detect avoidance or attraction between animals (Doncaster 1990, Turchin 1998). That is, 'if elk were> 10 km from a particular sheep band, then it was not likely that there was any interaction between those elk and the sheep band. To include only elk close enough to interact with the sheep bands, I repeated this analysis at 2 other spatial scales. First, I re-analyz.ed the data using elk and sheep bands that were <5 km from each other on any day. Finally, I re-analyz.ed the data using elk and sheep bands that were < l km of each other. I did not examine closer spatial scales because sample sizes became too small. RESULTS The observed ranks were distributed without any apparent attraction or avoidance for 1996 (Fig. 3.4) and 1997 (Fig. 3.5). For all elk, elk <5 km from a sheep band, and elk <1 km ofa sheep band, the observed mean rank was not significantly greater than the expected mean rank under the null hypothesis that locations of elk were random with respect to sheep (Table 3.1). Thus, elk did not avoid sheep at the 3 spatial scales examined. For all spatial scales, the randomized mean rank remained constant (Table 3.1). In 1996, when approximately 22 locations were collected per elk, the randomized mean rank was 11.48-11.49 for the 3 spatial scales. In 1997, when approximately 23 locations were collected per elk, the randomized mean rank was 11.9311. 97 for the 3 spatial scales. Maps of elk and domestic sheep locations on several flights are presented in Appendix 3. DISUCSSION Experimental Results Elk did not appear to avoid sheep at any of the spatial scales I examined. Elk may avoid sheep at <l km, but I lacked sample sires needed for finer scale detection of avoidance. However, because elk did not avoid �256 sheep at > I km, it is extremely unlikely that domestic sheep caused any large scale movements of elk from public to private lands. Additionally, it may be that elk avoided the area used by sheep following sheep occupation. If this effect lasted for a long period of time, then the rank method used would be biased because would not detect this avoidance. If there is concern about elk avoidance of sheep at <l km, or elk avoidance of an area after sheep occupation, then a study designed to answer these questions is needed. Although there is little documentation of elk response to domestic sheep disturbances, studies have been conducted on other, similar, non-lethal disturbances. Elk responses to recreationalists, cross country skiers, cattle, and roads may illuminate possible elk responses to domestic sheep. In areas of continuous human presence, such as Rocky Mountain Park, elk showed little response to approaches of people in automobiles or on foot, day or night (Schultz and Bailey 1978). However, in the less intensely used Medicine Bow National Forest, Ward et al. (1973) found that elk avoided recreationalists (campers, fishermen, and picnickers) at <800 m. Elk Response to cross country skiers depended on the habituation of elk; in areas of high human activity elk moved short distances away from skiers, while in areas of low human activity, elk movement away from skiers was over 3 times greater than that of habituated elk (Cassirer et al. 1992). With respect to cattle, elk showed no movement responses to cattle grazing at> 100 m (Ward et al. 1973, Clegg 1994). In addition to habituating to non-lethal disturbances, elk may learn to distinguish between dangerous and harmless disturbances. In Rocky M01mtain National Parle, where elk were not hunted, Schultz and Bailey (1978) found no avoidance of roads in winter. In contrast, in Roosevelt National Forest, which is adjacent to Rocky Mo1D1tain National Parle, Rost (1975) found that a population ofh1D1ted elk avoided roads in winter. Wright (1983).found that mean distance to dirt roads more than doubled when hunting season opened. In the White River area, where sheep have been grazed for the past l 00 years, elk may have learned that sheep pose no threat, and like elk exposed to other nonlethal human activities, the White River elk may have habituated to sheep activities and ceased to avoid sheep. Methodology for Determining Attraction and Avoidance One of the difficulties in studying behavioral interactions between animals with radio telemetry data is lack of quantitative methods to determine if one animal's movements or locations are related to another animal's (White and Garrott 1990). Historically, radio-telemetry location data has been used to descn"be the size and shape of an animal's home range. Most studies collecting these type of data do not take into account the temporal information available in the telemetry data (White and Garrott 1990). This temporal information can be used to describe social interactions, such as avoidance or attraction, between animals. Two general approaches have been used to descn"be animal interaction. The first, static territorial overlap, describes the overlap of home ranges (Webster and Brooks 1981, Minta 1992) as an indication of attraction or avoidance. The second method, temporal or dynamic interactions, uses distance expectations based on expected home range use to test for attraction or avoidance (Dunn 1979, Macdonald et al. 1980, Doncaster 1990, Minta 1992). Both static and parametric temporal methods of testing for avoidance between animals require knowledge of home range siz.e or use. There are difficulties in estimating the mean area and variance of a home range, even if estimation procedures are accurate (Worton 1987). Additionally, home range statistics vary depending on the method of estimation (White and Garrott 1990). Parametric methods that test for temporal interaction assume that home range use follows a bivariate normal distribution (Jennrich and Turner 1969, Dunn 1979). It is unlikely that animals use space in a bivariate normal manner because resources, mates, and cover are all likely to be clumped and non-normal in distribution. To combat unlikely representations of home range siz.e of use, Doncaster (1990) suggests a nonparametric approach based on differences between observed and theoretical distributions of separation distances between 2 animals. All temporally paired distances and unpaired distances between 2 animals are calculated, and a critical distance is chosen for analysis of temporal interaction. Using this distance, data can be placed in a 2 x 2 contingency table showing frequen_cies of Pa.ii"~ and unpaired distances closer and farther than this critical distance. If distances are closer than expected, then the animals are attracted to each other; if distances are farther than expected, then the animals are avoiding each other. The rank method presented here is similar to Doncaster's (1990) nonparametric test for attractions and avoidance. Both methods have the advantage of not requiring assumptions about home range size or use compared to the static and parametric dynamic methods. However, there are 2 main advantages of the rank method over Doncaster's (1990) nonparametric approach. First, the rank method requires no arbitrary critical distance to evaluate attraction or avoidance. Although it may appear that the various spatial scales I used are similar to a critical distance, they are not. This study was not originally designed to test for attraction and �257 avoidance between elk and sheep, and the spatial scale over which I collected data (4,540 lan.2) was too large for accurate testing of attraction and avoidance between animals. Because of the large scale, I examined various spatial scales to subset out animals close enough to sense each other. Studies designed to test for avoidance or attraction interactions would not be at such a large scale and would not need to subset out animals close to each other. Additionally, at any scale examined, I did not set any critical distance at which attraction or avoidance was to occur. Second, Doncaster's (1990) method uses a chi-square test statistic, which works best when expected cell counts are ~5. The chi-square would not have worked well in this study because there was only 1 paired distance, which results in a count of 1 and O in cells containing the number of paired distances closer and farther than the critical distance. The chi-square test may not work well for any study with low sample sizes or unbalanced data (more locations on one animal than another). However, this problem is attenuated if a Fisher's exact test is used to evaluate the 2-way contingency table of paired and unpaired distances. MANAGEMENT IMPLICATIONS This study suggests that elk do not avoid domestic sheep at distances great enough to cause elk movements off public hunting grounds to private land. If concern about elk avoidance of sheep at distances of <1 km becomes serious, a more complete study, with radiocollared sheep from each band and a design tailored to the avoidance question, should be performed. The rank method used to test for elk avoidance of sheep could be used in many radio-telemetry data sets to evaluate attraction or avoidance between animals. The rank method, using a randomization procedure to generate the null distribution, is a flexible, powerful, and assumption reduced approach to test for attraction or avoidance behaviors between animals. The main requirement of the method is that simultaneous locations are taken on all animals in the study. LITERATURE CITED Beck, J. L., J. T. Flinders, D. R Nelson, and C. L. Clyde. 1996. Dietary overlap and preference of elk and domestic sheep in aspen-dominated habitats in north-central Utah. Pages 81-85 in K. E. Evans, compiler. Sharing common ground on western rangeland: proceedings of a livestock/big game symposium. U.S. Forest Service, Intermountain Research Station, Ogden, Utah, USA. Boyd, R J. 1970. Elk of the White River Plateau, Colorado. Colorado Division of Game, Fish and Parks Technical Publication 25, Fort Collins, Colorado, USA. Cassirer, E. F., D. J. Freddy, and E. D. Ables. 1992. Elk responses to disturbances by cross-country skiers in Yellowstone National Park. Wildlife Society Bulletin 20:375-381. Clegg, K. 1994. Density and feeding habits of elk and deer in relation to livestock disturbance. Thesis, Utah State University, Logan, Utah, USA. Doncaster, C. P. 1990. Nonparametric estimates of interaction from radio-tracking data. Journal of Theoretical Biology 143:431-443. Dunn, J.E. 1979. A complete test for dynamic territorial interaction. Pages 159-169 in F. M. Long, editor. Proceedings for the second international conference on wildlife biotelemetry. University of Wyoming, Laramie, Wyoming, USA. Jennrich, RI., and F. B. Turner. 1969. Measurement of non-circular home range. Journal of Theoretical Biology 22:227-237. MacCracken, J. G., and R M. Hansen. 1981. Diets of domestic sheep and other large herbivores in south central Colorado. Journal of Range Management 34:242-243. Macdonald, D. W., F. G. Ball, and N. G. Hough. J:980. The e¥aluation of home range size and configuration using radio tracking data. Pages 405-424 in C. J. Amlaner, Jr. and D. W. Macdonald, editors. A handbook on biotelemetry and radio tracking. Pergamon Press, Oxford, England. Minta, S. C. 1992. Tests of spatial and temporal interaction among animals. Ecological Applications 2:178204. Mooney, C. Z., and RD. Duval. 1993. Bootstrapping: a nonparametric approach to statistical inference. Sage Publishing Inc., Newbury Park, California, USA. �258 Nichols, L. Jr. 1957. Forage utilization by elk and domestic sheep in the White River National Forest. Thesis, Colorado State University, Fort Collins, Colorado, USA. Olsen, F. W., and R. M. Hansen. 1977. Food relationships of wild free-roaming horses to livestock and big game, Red Desert, Wyoming. Journal of Range Management 30: 17-20. Pickford, G. D. and E. H. Reid. 1943. Competition of elk and domestic livestock for summer range forage. Journal of Range Management 7:328-332. Rost, G. R. 1975. Responses of deer and elk to roads. Thesis, Colorado State University, Fort Collins, Colorado, USA. Schultz, R. D., and J. A. Bailey. 1978. Responses of national park elk to human activity. Journal of Wildlife Management 42:91-100. Stevens, D. R. 1966. Range relationships of elk and livestock, Crow Creek drainage, Montana. Journal of Wildlife Management 30:349-363. Turchin, P. 1998. Quantitative analysis of movement: measuring and modeling population redistribution in animals and plants. Sinauer, Sunderland, Massachusetts, USA. Ward, A. L., J. J. Cupel, A. L. Lea, C. Z. Oakeley, and R. W. Weeks. 1973. Elk behavior in relation to cattle grazing, forest recreation, and traffic. Transcripts from the 38th American Wildlife Natural Resource Conference 38:327-337. Webster, B. A, and R. J. Brooks. 1981. Social behavior ofMicrotus pennsylvanicus in relation to seasonal changes in demography. Journal ofMammalogy 64:738-751, White, G. C., and R. A. Garrott. 1990. Analysis of radio-tracking data. Academic Press, San Diego, California, USA Wright, K. L. 1983. Elk movements, habitat use, and the effects of hunting activity on elk behavior near Gunnison, Colorado. Thesis, Colorado State University, Fort Collins, Colorado, USA Worton, B. J. 1987. A review of models of home range for animal movement Ecological Modeling 38:277298. Table 3.1. Paired mean rank, randomiz.edmean rank, 95% confidence interval of the randomu:ed mean rank, and the probability of observing the sample rank or greater under the null hypothesis that elk locations are random with respect to sheep locations. Paired mean rank is the observed rank of temporally paired observations, and randomized mean rank is the expected rank of the temporally paired observations if elk are located randomly with respect to sheep. Num.berof observed rankings Paired mean rank Randomi7.ed meanraok Randomu:ed 95%CI 11.34 12.14 11.48 11.97 [11.47, 11.49] [11.96, 11.98] 0.737 0.199 11.71 10.59 11.48 11.93 [11.45, 11.52] [11.89, 11.96] 0.342 0.990.:. 9.09 9.84 11.49 12.00 [11.41, 11.57] [I 1.90, 12.10) 0.969 0.912 All elk 1,844 1996 1997 1,790 All e//c <5 km ofa sheep band on any day 1996 150 124' 1997 All elk <l km of a sheep band on any day 1996 34 1997 19 �259 t N Meeker Glenwood Springs • Rifle km 0 10 20 BLM Land Flat Tops Wilderness State Wildlife Land Study Area Boundary Forest Service Land Sheep Locations 1996 Sheep Locations 1997 Towns a ~ ..lo. * ■ Figure 3.1. Sheep bands located during flights 20 July-IO October 1996-1997, and land ownership in the White River Area, Colorado. Note that all white area (blank area) is private land. �260 Elk Day Sheep Band i 2 3 Temporally paired and unpaired distances PAIRED unpaired unpaired Rank 21 4 11 22 1 2 1 3 PAIRED unpaired 17 1 22 unpaired 3 Unpalied 2 . 2 unpaired 3 unpaired 21 15 7 PAIRED 20 I 1 22 2 i PAIRED I 2 2 2 21 3 unpaired oopalred 2 2 22 unpaired 2 2 9 15 7 2 3 2 . 22 uupalied .j 1 . PAIRED unpaired 21 22 • unpaired 2 2 2 2 unpaired unpaired 22 4 2 22 22 PAIRED 11 88 88 1 1 3 PAIRED unpaired unpaired 18 2 88 2 . 3 . 88 88 88 22 88 88 1 7 oopalred 2 Ot' l' e<l p .fl'RED 8 21 3 unpaired 1 22 unpaired 3 I 2 2 Figure 3.2. Representation of temporally paired and unpaired distances between elk and sheep. Only I sheep band was l~ted per day, therefore sheep band I was located on day 1, sheep band 2 was located on day 2 9tf. Ranks were randomly assigned as an example. Ranks of the temporally paired distances were extracted to • calculate the paired mean rank for a given year and spatial scale. �261 a -----·------------ Paired mean rank= 7.1 200 Expected mean rank= 5.5 >- 150 u C GI 6- 100 e u. 1 2 3 4 5 6 Rank 7 8 9 10 b 200 Paired mean rank= 3.9 Expected mean rank = 5.5 >- 150 u C GI 6- 100 e u. 50 I 0 ..µ,....._.....,......_,....u1•..,....11.,,l~'~r.....,L,-. 1 2 3 4 5 6 Rank 7 8 9. 10 9 10 C Paired mean rank= 5.5 200 Expected mean rank= 5.5 >-150 u . C a, ::, ec-100 u. 50 0 1 2 3 4 5 6 Rank 7 8 Figure 3.3. Theoretical distribution of paired ranks and paired mean rank under the expectation that (a) elk avoid domestic sheep, (b) elk are attracted to domestic sheep, or (c) elk are located rando!llly with respect to domestic sheep. Paired mean rank is the observed rank of temporally paired observations, and expected mean rank is the mean rank of all observations (temporally paired and unpaired) if elk are located randomly with respect to sheep. �262 a 120 - r - - - - - - - - - - - - : - - - - : - : - = - : - - - - - - - - , Paired mean rank= 11.34 100 Expected mean rank= 11.48 ~80 C CD 5-60 Kt40 20 0 ---..-.--..-.--..-. 1 3 5 7 9 11 13 15 17 19 21 23 Rank b 15 - r - - - - - - - - - - - - - - - - , Paired mean rank-= 11.71 Expected mean rank"" 11.48 >40 -CJ C CD ::s CJ" f5 IL 1 3 5 7 9 11 13 15 17 19 21 23 Rank C &~-------------------, 5 Paired mean rank= 9.09 Expected mean rank= 11.49 1 0 - . - . . - - -........-...---.......-.--.-...........,-,.-----1 1 3 5 7 9 11 13 15 17 19 21 23 Rank Figure 3.4. Observed distribution of paired ranks and paired mean rank for (a) all elk, (b) elk <5 km of a sheep band, and (c) elk <I km ofa sheep band, 20 July-IO October 1996 in the White River area, Colorado. Paired mean rank was the observed rank of temporally paired observations. Expected mean rank was the mean rank of all observations (temporally paired and unpaired) if elk were located randomly with respect to sheep, and was calculated from 999 samples that randomized on locations of sheep bands~ �263 a 120 ~ - - - -Paired - -mean - -rank - -c -----, 12.14 100 Expected mean rank = 11.97 ('; 80 C ~ 60 O" Cl) ~ 40 20 0 -flil,,...,....,....-,...,...,..,-,..,.....-,.,..,...,..,..,....,..,-.,.....,...,_,-i 1 3 5 7 9 11 13 15 17 19 21 23 Rank b 15 ~ - - - - - - - - - - - - - - - - - . Paired mean rank-= 10.69 Expected mean rank c 11.93 ~10 C G> ::s er e u. 5 1 3 5 7 9 11 13 15 17 19 21 23 Rank C 6~---------------~ Paired mean rank .. 9.84 5· Expected mean rank .. 12.00 1 0 -t-.,-,..-.-'r-T-.-...-,--.-,---+--,-..,-,--......,......-.........~ 1 3 5 7 9 11 13 15 17 19 21 23 Rank Figure 3.5. Observed distribution of paired ranks and paired mean rank for (a) all elk, (b) elk <5 km ofa sheep band, and (c) elk <l km ofa sheep band, 20 July-IO October 1997 in the White River area, Colorado. Paired mean rank was the observed rank of temporally paired observations. Expected mean rank was t;he mean rank of all observations (temporally paired and unpaired) if elk were located randomly with respect to sheep, and was calculated from 999 samples that randomized on locations of sheep bands. �264 �265 APPENDIX 1: SAMPLE SIZE CALCULATION FOR HUNTER SURVEY Note, G. White (Colorado State University, personal communication) derived this sample siz.e calculation. Hunt codes D-E-012-01-A. E-E-012-01-A. E-F-012-01-M, E-M-012-01-M, D-M-012-01-M, D-E-033-01-A. E-E033-01-A. E-F-033-01-M, E-M-033-01-M, and D-M-033-01-M used the study area during the study. To estimate total number of hunters on any day with a 95% confidence interval within ±150 hunters, the binomial distribution and proportion of licenses holders was used to derive the number of license holder to be surveyed, where: N Number of licenses holders (number of individuals purchasing licenses), p Estimated proportion of hunters hunting, n Number of licenses holders sampled/surveyed, H Estimated num.ber of hunters hunting, N-n N Finite population correction factor, H=pN,and N -n ~rt A)_ N -n N2 p(l- p) vaAr(HA)- - N2 v... \p - - -------'--. N N n The maximum variance will occur if 50% of the license holders hunt (i.e., p = 0.5), with the variance decreasing for larger or smaller percentages. For N license holders, the sample size (n) required to obtain a 95% confidence interval of ±150 with a= 0.05 and p = 0.5 is: On most days, the confidence interval will be less than ±150 because not exactly 50% of the hunters will be hunting. For example, if80% oflicense holders are hunting, then the variance of p is reduced from 0.25/n to 0.16/n (without the finite population correction factor). Thus, the confidence interval will be considerably smaller than ±150 hunters. �266 �267 APPENDIX 2: STANDARD ERRORS FOR DAILY GMU ESTIMATES OF HUNTERS AND ATVS AFIELD Estimated number of hunters afield per day by GMU and standard enors of the estimates for 1996. Stanoaro error of estimates Est1matoo fiunters af1elo DATE 8!:M/96 8/25/96 8/26/96 8/27/96 8/28/96 8/29/96 8/30/96 8/31/96 9/1/96 9/2/96 9/3/96 9/4/96 9/5196 9/6/96 9nl96 9/8/96 919196 9/10/96 9/11/96 9/12/96 9/13/96 9/14/96 9/15/96 9/16/96 9/17/96 9/18/96 9/19/96 9/20/96 9/21/96 9/22/96 9/23/96 9/24/96 9/25/96 9/26/96 9/27/96 9/28/96 9/29/96 9130196 10/1/96 10/2/96 10/3/96 10/4/96 10/5/96 10/6/96 Rl2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 568 551 498 460 410 362 314 282 209 155 124 111 91 80 80 74 73 70 64 61 63 66 51 H23 210 211 176 157 151 122 124 154 150 118 109 112 106 116 161 147 120 115 113 109 103 526 531 381 343 296 241 211 186 113 77 72 62 50 43 75 64 37 33 33 38 39 46 42 R2:il 127 135 129 119 106 85 84 121 132 126 105 105 94 88 119 118 116 114 113 112 113 421 400 265 243 197 163 133 117 47 30 32 25 22 28 36 33 26 29 3"1 31 25 19 21 R:33 218 226 185 176 152 129 130 177 175 151 133 119 112 118 163 146 74 69 65 71 74 211 213 94 78 52 43 39 34 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Rl2SE 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 23 23 23 22 21 21 20 19 18 16 14 14 12 12 12 11 11 11 11 10 11 11 10 1-12:JSE 17 17 16 15 ·15 14 14 15 15 14 13 13 13 14 16 15 14 14 H24SE 14 14 14 14 13 II 11 14 14 14 13 13 13 12 14 14 14 14 14 13 13 25 25 20 20 19 17 16 15 13 II II IO 9 9 13 II 11 8 8 8 8 8 9 9 14 14 22 22 16 16 15 14 13 12 9 7 7 6 6 7 8 7 7 7 7 7 7 6 6 H33SE 18 18 17 16 15 14 14 16 16 15 15 14 13 14 16 15 11 11 10 11 11 16 16 8 8 6 6 6 5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Note: H12 represents estimated number of hunter afield in GMU 12, H23 represents estimated number of hunter afield in GMU 23, etc. Hl2SE represents the standard error of the estimated number of hunters in GMUI2, Hl2SE represents the standard error of the estimated number of hunters in GMU23, etc. �268 Estimated number of hunters using ATVs afield per day by GMU and standard errors of the estimates for 1996. DATE 8724796 8/25/96 8/26/96 8/27/96 8/28/96 8/29/96 8/30/96 8/31/96 9/1/96 9/2/96 9/3/96 9/4/96 9/5/96 9/6/96 9nt96 9/8/96 9/9/96 9/10/96 9/11/96 9/12/96 9/13/96 9/14/96 9/15/96 9/16/96 9/17/96 9/18/96 9/19/96 9/20/96 9/21/96 9/22/96 9/23/96 9/24/96 9/25/96 9/26/96 9/27/96 9/28/96 9/29/96 9/30/96 10/1/96 10/2/96 10/3/96 10/4/96 10/5/96 10/6/96 Stanaaro error of estimates Esf1matea liunters usmg A I Qs af1elo HA12 HA23 HA24 HA'.33 HAl2SE HA23SE HA24SE HADSE 48 0 2o 0 9 6 51 9 6" 0 47 20 52 0 9 9 44 0 41 17 0 8 5 9 0 41 16 43 0 8 5 9 4 0 40 13 41 0 8 8 4 0 36 12 41 0 8 8 4 0 36 12 36 8 0 8 0 29 7 4 13 55 0 IO 4 7 0 26 13 51 0 9 17 5 0 11 50 0 4 IO 0 17 14 47 5 5 0 9 0 17 5 14 36 0 5 8 0 16 8 5 3 35 0 8 0 32 7 8 33 0 3 8 0 49 7 57 9 3 0 IO 0 46 6 48 0 9 3 9 0 40 8 6 26 0 3 7 0 8 39 7 23 3 0 7 0 8 39 7 23 0 3 7 0 33 8 2 5 18 0 6 28 7 0 11 4 15 0 5 14 50 139 77 45 9 IO 8 50 136 14 77 42 9 IO 8 47 90 II 59 13 9 9 3 46 85 11 8 12 9 55 3 37 77 46 12 8 IO 8 3 36 56 39 8 8 9 7 2 35 50 34 6 8 9 7 2 40 33 27 6 8 7 6 2 31 20 9 7 6 4 0 0 22 16 2 0 6 1 5 0 17 17 4 6 2 0 6 0 15 12 4 0 5 4 2 0 12 12 4 4 0 4 2 0 9 15 7 0 4 5 3 0 9 24 9 4 0 7 4 0 6 19 9 0 4 3 6 0 2 IO 8 0 I 4 4 0 2 4 8 0 0 3 4 2 4 8 0 0 3 4 4 2 8 0 0 4 3 0 7 6 0 0 0 3 3 3 10 0 2 0 0 4 0 3 7 0 0 0 2 3 0 Note: Hl2 represents estimated number of hunter afield in GMU 12, H23 represents estimated number of hunter afield in GMU 23, etc. Hl2SE represents the standard error of the estimated number of hunters in GMU12, Hl2SE represents the standard error of the estimated number of hunters in GMU23, etc. �269 Estimated number of hunters afield per day by GMU and standard errors of the estimates for 1997. Stanaara error of estimates H23SE H24SE H12SE DATE Eshmatoo liunfers af1ela H23 RI2 R24 H33 8723797 328 178 124 0 23 17 15 0 8/24/97 8/25/97 8/26/97 8/27/97 8/28/97 8/29/97 8/30/97 8/31/97 9/1/97 9/2/97 9/3/97 9/4/97 9/5/97 9/6/97 9nl97 9/8/97 9/9/97 9/10/97 9/11/97 9/12/97 9/13/97 9/14/97 9/15/97 9/16/97 9/17/97 9/18/97 9/19/97 9/20/97 9/21/97 9/22/97 9/23/97 9/24/97 9/25/97 9/26/97 9/27/97 9/28/97 9/29/97 9/30/97 10/1/97 10/2/97 10/3/97 10/4/97 10/5/97 314 282 252 226 206 191 220 211 191 151 139 115 107 130 168 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 23 22 21 20 19 19 20 20 19 17 17 15 15 16 17 ·11 17 15 14 14 13 13 14 14 14 13 13 13 12. 14 13 11 12 12 15 15 14 13 12 11 11 II 12 12 10 10 10 11 11 12 12 12 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 483 477 401 403 379. 317 249 275 240 17 23 22 21 21 20 161 148 0 0 -0 0 0 0 0 0 0 0 0 0 0 0 141 132 120 112 122 142 416 391 364 345 300 205 172 160 112 0 0 0 0 0 0 0 0 0 0 0 0 0 0 98 119 117 109 100 82 66 67 64 81 77 56 96 49 87 51 62 61 63 66 67 60 61 477 470 462 409 399 307 242 224 207 135 123 115 114 105 81 73 75 72 71 67 51 34 24 138 123 115 109 92 115 110 119 97 IOI 87 65 67 68 68 71 484 483 449 442 404 309 275. 223 149 57 50 53 52 50 46 50 44 50 39 37 37 40 45 .o 127 126 130 142 137 79 77 87 91 98 111 105 16 15 16 18 17 16 14 ii 11 11 26 26 25 25 24 22 21 19 16 10 10 10 10 10 9 IO IO IO 8 8 8 9 9 12 26 26 26 24 24 22 20 20 19 16 15 15 15 14 13 12 12 12 12 12 10 8 7 H33SE () 0 0 0 0 0 25 25 23 23 23 22 20 21 20 18 17 15 15 16 16 16 13 12 13 14 14 15 14 Note: Hl2 represents estimated number of hunter afield in GMU 12, H23 represents estimated number of hunter afield in GMU 23, etc. HI 2SE represents the standard error of the estimated number of hunters in GMU 12, Hl2SE represents the standard error of the estimated number of hunters in GMU23, etc. �270 Estimated number of hunters using ATVs afield per day by GMU and standard errors of the estimates for 1997. DATE 8723797 8/24/97 8(25/97 8(26/97 8f27/97 8(28/97 8/29/97 8/30/97 8/3 l/97 9/1/97 9f2/97 9/3/97 9/4/97 9/5/97 9/6/97 9nl91 9/8/97 9/9/97 9/10/97 9/l l/97 9/12/97 9/13/97 9/14/97 9/15/97 9/16/97 9/17/97 9/18/97 9/19/97 9f10l91 9fll/91 9(22/97 9(23/97 9f24/97 9f15l97 9f16l91 9(27/97 9(18/97 9/29/97 9/30/97 10/1/97 I0/2/97 10/3/97 I0/4/97 10/5/97 Estimatea nunters af1ela HA[2 HA23 HA24 43 70 34 39 36 70 32. 56 39 35 30 48 30 48 40 45 30 33 25 30 26 21 27 30 28 18 27 24 31 16 18 36 18 18 36 12 15 33 12 8 29 9 7 IO 34 5 30 8 12 30 8 30 8 5 2 8 30 28 3 5 7 21 5 110 29 50 32 110 50 112 51 30 20 113 48 22 106 48 81 ll 39 IO 77 33 52 7 30 7 19 35 14 0 7 0 14 7 0 8 7 8 0 7 0 8 7 0 15 3 0 19 3 0 19 3 0 19 3 0 19 3 0 19 3 0 19 3 0 12 3 0 12 3 HA33 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 99 IOI 94 85 78 63 56 64 58 54 40 37 35 42 48 39 29 22 24 22 34 34 34 Stanaara error of estimates HAl2SE HA23SE HA24SE HA33SE 12 8 9 0 12 9 8 0 IO 9 8 0 9 8 0 8 9 9 8 0 8 8 9 0 8 7 0 7 7 8 8 0 8 6 0 8 8 5 7 0 8 6 6 0 8 6 5 0 8 5 5 0 7 3 4 0 8 3 4 0 8 4 3 0 8 4 5 0 8 4 3 0 8 4 0 8 3 0 3 4 7 3 0 8 14 13 7 14 8 13 8 8 13 14 8 8 12 14 6 14 8 12 6 12 8 11 5 12 8 10 4 8 11 3 IO 6 11 3 8 4 11 6 0 0 6 4 9 4 0 4 9 4 0 4 9 4 4 9 0 6 3 IO 0 0 6 3 9 8 6 3 0 7 6 3 0 7 6 3 0 7 6 3 0 9 3 0 6 9 3 0 5 9 0 5 3 Note: Hl2 represents estimated number of hunter afield in GMU 12, H23 represents estimated number of hunter afield in GMU 23, etc. Hl2SE represents the standard error of the estimated number of hunters in GMUI2, H 12SE represents the standard error of the estimated number of hunters in GMU23, etc. �:.!__; l White River Elk Movement Study White River Elk Movement Study Elk and Shee~ Locations on July 19, 1996 Elk and Sheep Locations on July 24, 1996 Hamilton Km 20 z t:I ~ !-:I Hamilton 0 ~ ~ l'1!j ~ 40 en 0 ',ij l'1!j ~ ~ t:I t:I 0 ~ l'1!j en -i ~ n en = l'1!j l'1!j ~ I:"" 0 n > -i Rio Blanco ~ 0 z en 0 z en LEGEND BLM Land Flat Tops WIiderness State Wildlife Land Study Area Boundary White River National Forest Springs ■ * A ■ ..., White River Elk Location Sheep Location Towns l'1!j I:"" l'1!j n -i Springs ■ l'1!j t:I ',ij I:"" ~ (;') = -i en N -.J �White River Elk Movement Study White River Elk Movement Study Elk and Sheep Locations on July 31, 1996 Elk and Sheep Locations on August 4, 1996 Hamilton Hamilton Krn 0 20 40 LEGEND Springs • .BLM Land Flat Tops WIiderness State Wildlife Land Study Area Boundary White River National Forest - White River • Elk Location Sheep Location ■ Towns !* -------- Springs ■ �White River Elk Movement Study White River Elk Movement Study Elk and Sheep Locations on August 14, 1996 Elk and Sheep Locations on August 28, 1996 Hamilton Hamilton Km 0 20 40 Rio 81anco LEGEND Springs ■ * • ■ BLM Land Flat Tops Wilderness State Wildlife Land Study Area Boundary White River National Forest White River Elk Location Sheep Location Towns ts) ...J \.,.) �N -.I White River Elk Movement Study White River Elk Movement Study Elk and Sheep Locations on August 20, 1997 Elk and Sheep Locations on September 6, 1997 Hamilton Hamilton Km 20 0 LEGEND BLM Land Flat Tops WIiderness State Wildlife Land Study Area Boundary White River National Forest !* - White River • ■ Elk Location Sheep Location Towns 40 00 �White River Elk Movement Study White River Elk Movement Study Elk and Sheep Locations on September 17, 1997 Elk and Sheep Locations on Septermber 24, 1997 Hamilton Km 0 20 40 LEGEND BLM Land Flat Tops Wilderness State Wildlife Land Study Area Boundary White River National Forest - White River • Elk Location Sheep Location. ■ Towns !* N --.J '° �White River Elk Movement Study White River Elk Movement Study ~:~ c:-:c S:;ee:::: ~oc:::~:o~s 0:1 Cctober 1, ~ 997 Elk and '":,he::>~, Lr;r,rJ;:r:;n~: on October 5, 1997 Hamilton Hamilton Km 0 • 20 LEGEND BLM Land Flat Tops Wilderness Area State Wildlife Area Study Area Boundary White River National Forest - White River • Elk Location Sheep Location ■ Towns !* 40 � https://cpw.cvlcollections.org/files/original/0a33e58ed066698d08f7ebee6435722d.pdf 6733fc151d4a614da8db27b829c5bcab PDF Text Text 227 Colorado Division of Wildlife Wildlife Research Report July 1998 JOB PROGRESS REPORT State of Colorado cost center 3430 Project No. W-153-R-11 Mammals Program Work Package No. ___3_0_0-2______ Task No. _______.._______ Elk conservation : Monitoring and Managing-Chronic wasting Disease in Elk Period Covered: July 1, 1997 - June 30, 1998 w. Miller Authors: M. Personnel: s. Berry, K. Larsen, K. I. O'Rourke, T. R. Spraker, Wheeler, M.A. Wild, and E. S. Williams s. Tracy, E. ABSTRACT Elk from throughout Colorado were examined for occurrence of chronic wasting disease using a combination of targeted surveys and harvest or road-kill surveys. We continued to develop and modify a statewide targeted surveillance program for acquiring, examining, and reporting on CWD suspects submitted from Colorado. Between June 1997 and May 1998, 2 chronic wasting disease (CWD) cases were diagnosed among 6 •suspect• elk submitted from endemic game management units (GMUs) in northeastern Colorado; CWD was not diagnosed in any of 6 additional "suspect• elk submitted from elsewhere in Colorado. Harvest surveys were used to estimate CWD prevalence in enzootic management units. About 0.21 of elk harvested in Larimer County data analysis units (DAUs) (E4 or E9) tested positive for CWD via immunostaining; CWD was not detected in any of the harvested elk submitted from outside Larimer county. Requiring hunters to participate in harvest surveys continued to increase the number of samples submitted for CWD examination. Data from both targeted surveillance and surveys indicate that Larimer county remains the only focus of CWD in Colorado elk, although some undetected natural spread may be occurring. Compared to deer, however, CWD is a relatively rare disease in free-ranging elk in northcentral Colorado. Targeted surveillance of clinical suspects appears to be the most sensitive approach for initially detecting CWD in elk populations throughout Colorado, and will be continued statewide. Based on low prevalence detected in 1995-1997 surveys, future harvest and road-kill surveys will be conducted at about 5-yr intervals to estimate prevalence, monitor trends, and compare rates among endemic elk DAUS. ��229 MONITORING ARD MANAGING CHRONIC WASTING DISEASE IN ELK M.w. Miller P, H, OBJECTIVES (1) Design, conduct, and report results of: (a) targeted surveillance to estimate and detect changes in distribution of chronic wasting disease (CWD) in free-ranging elk populations; and (b) harvest or road-kill surveys to estimate and detect changes in prevalence of CWD in enzootic elk populations. (2) Design, conduct, and report results of experimental studies using captive elk naturally or experimentally infected with CWD. SEGMEHT OBJECTIVES (1) Conduct and report results of targeted surveillance to estimate and detect changes in distribution of CWD in free-ranging elk populations statewide. (2) Conduct and report results of harvest surveys to estimate prevalence of CWD in DAUS E4 and E9. (3) Observe epizootiological features of naturally-occurring CWD in captive elk. IHTRQDUCTIQH Chronic wasting disease (CWD) is a disease of native deer and elk characterized by behavioral changes and progressive loss of body condition that invariably lead to the death of affected animals (Williams and Young 1992). Neither the causative agent nor its mode of transmission have been identified. There are no tests currently available for diagnosing CWD in live animals, and postmortem tests require microscopic examination of brain tissue. There are no known treatments for CWD. Previous attempts to eradicate CWD from research facilities failed on at least 2 occasions (Williams and Young 1992). Although similar in some respects to other transmissible spongiform encephalopathies that affect domestic sheep (scrapie) and cattle (bovine spongiform encephalopathy; BSE; "mad cow disease"), there is no evidence suggesting CWD can be naturally transmitted to domestic livestock, or that scrapie or BSE can be transmitted to native cervids. Moreover, there is no evidence suggesting that CWD presents a threat toLhuman health. "Chronic wasting disease" was first recognized by biologists in the 1960's as a disease syndrome of captive deer held in wildlife research facilities in Ft. Collins, co, and was subsequently recognized in captive deer, and later in captive elk, in wildlife research facilities near Ft. Collins, Kremmling, and Meeker, co and Wheatland, WY (Williams and Young 1980, 1982). Since 1981, 74 cases of CWD have also been diagnosed in free-ranging mule deer, white-tailed deer, and elk from northeastern Colorado; most of these diagnoses have been made since 1990 (Spraker et al. 1997; M. w. Miller, unpubl. data). At present, the known world-wide distribution of CWD in wild cervids appears to �230 be limited to northeastern Colorado and southeastern Wyoming. Although CWD was first diagnosed in captive cervids, the original source of CWD in either captive cervids or free-ranging cervids is unknown; whether CWD in captive cervids really preceded CWD in wild cervids, or vice versa, is equally uncertain (Spraker et al. 1997). In Colorado, free-ranging CWD cases in elk have all originated from along the northern Front Range. Game Management Units (GMUs) yielding infected elk include 7, 9, 19, 191, and 20. All clinical elk cases have come from two Data Analysis Units (DAUs)(E4, E9). To date, examinations of elk from other parts of Colorado have not detected CWD. Preliminary data from hunter-killed deer indicate that even in enzootic areas CWD is relatively rare: it probably affects $l.5% of the elk in E4 and E9. No real trend in prevalence (increasing or decreasing) can be discerned from data available to date. In the absence of historical (20-30 years ago) prevalence data or reliable estimates of transmission rates, it is also unclear whether incidence of CWD in northcentral Colorado DAUs is stable or increasing, or whether short-term observations can accurately forecast long-term trends. The significance of CWD and its impacts on native elk populations are unclear. Preliminary results of simulation modeling to predict dynamics and impacts of CWD on affected deer populations suggest that, sustained at current prevalence (about 6.5% in deer), CWD could impact wild deer herds and lead to population declines (M. w. Miller and c. w. McCarty, unpubl. data). Although CWD prevalence is lower in elk than in deer in enzootic DAUs, it is conceivable that impacts could occur if prevalence were allowed to increase. Clearly, reliable estimates of CWD prevalence in wild elk populations are needed to guide policy decisions and monitor efficacy of management efforts. In the absence of data to the contrary, and considering the difficulties inherent in eliminating CWD from captive or wild cervid populations once established, it seems most prudent to assume CWD could adversely affect native deer or elk populations and manage to reduce its occurrence and prevent its further spread. Consequently, the Colorado Division of Wildlife needs to undertake a variety of actions to further understanding about CWD and its management through surveillance and experimental research in order to reduce the occurrence of CWD and minimize the risk of its spread to other native deer or elk populations in Colorado. MATERIALS AND METHODS Surveillance We monitored elk populations throughout Colorado for occurrence of CWD using a combination of targeted surveillance and harvest or road-kill surveys. These were organized and conducted as follows: Targeted(= clinical disease) surveillance: Elk showing clinical signs consist~nt with those seen in chronic wasting disease were collected by field personnel statewide and brain tissues examined for evidence of spongiform encephalopathy. The "suspect casen profile was defined as follows: • • • Species: Age: Signs: elk ~ 18 months emaciated and abnormal behavior &/or indifference to human activity &/or �231 increased salivation &/or tremor, stumbling, incoordination &/or difficulty or inefficiency in chewing/swallowing &/or increased drinking and urination Where possible, submissions were subjected to complete necropsy; in some situations, only heads were available for examination and sampling. In all cases, histopathology of brain tissue (Williams and Young 1993) was used to diagnose CWD; in some cases, immunohistochemistry or other ancillary tests were used to confirm or support diagnoses. Harvest surveys: In order to obtain reliable estimates of CWD prevalence that will serve as a basis for monitoring responses to management interventions, we continued conducting harvest surveys on select deer populations. During the 1997-1998 hunting seasons, fresh brain and select lymphatic tissues were collected from endemic and nonendemic GMUs. Brain tissues were examined at the Colorado State University Diagnostic Laboratory for histopathological lesions (Williams and Young, 1993)or anti-PrP immunostaining reactions (O'Rourke et al., 1998) consistent with CWD infection. Because sample sizes for most individual GMUs were too small to provide reliable prevalence estimates by GMO, we pooled data by DAU for comparisons within and among species. Small sample sizes also precluded meaningful analysis of data from deer or elk harvested outside known enzootic areas. Epizootiological Studies cwo Epidemiology of naturally-occurring in captive elk (Miller and Wild): we observed captive adult elk (n = 19) held at CDOW's Foothills Wildlife Research Facility for clinical signs of CWD and submitted all mortalities for complete necropsy and histopathological examination. A manuscript describing CWD epidemiology in captive elk (Miller et al., 1998) was revised and accepted for publication. RESULTS AND DISCUSSION Surveillance Targeted(= clinical disease) surveillance: Between June 1997 and May 1998, 2 chronic wasting disease (CWD) cases were diagnosed among 6 "suspect" elk submitted from endemic game management units (GMUs) in northeastern Colorado; CWD was not diagnosed in any of 6 additional "suspect" elk submitted from elsewhere in Colorado. Harvest surveys; During the 1997-1998 hunting seasons, fresh brain and select lymphatic tissues were collected from 653 elk harvested in enzootic GMUs • (Table l); 103 elk harvested or culled in other GMUs throughout Colorado were also sampled as negative controls. Estimated prevalence among harvested elk did not differ (P = 1.0) between E4 (0.2%) and E9 (0%). As in 1996, overall CWD prevalence among elk harvested in Larimer County was lower (P < 0.009) than in sympatric deer populations (see Miller, 1998). Even the foregoing prevalence estimates may be somewhat liberal because the definition of "positive" included subclinical cases where either histopathological lesions or anti-PrP immunostaining reactions in brain tissue were observed. The elk case identified in 1997, as well as all 4 elk cases in 1996, were classified as positive solely on the basis of immunostaining reactions. Although no known "false positives" were identified among the 103 elk examined from outside known enzootic DAUs in 1997, further evaluation of �232 both sensitivity and specificity of existing diagnostic techniques still appears warranted. During the 1997-1998 rifle seasons, harvest survey participation was required of successful elk hunters in both E4 and E9. In E4, 454 successful elk hunters submitted heads in 1997, compared to only 81 voluntary submissions in 1996. These observations reemphasize previous observations (Miller, 1997) that compelling participation in harvest surveys via regulation is the single most effective method for increasing sample sizes. Data from both targeted surveillance and surveys indicate that Larimer County remains the only focus of CWD in Colorado elk, although some undetected natural spread may be occurring. Compared to deer, however, CWD is a relatively rare disease in free-ranging elk in northcentral Colorado. Targeted surveillance of clinical suspects appears to be the most sensitive approach for initially detecting CWD in elk populations throughout Colorado, and will be continued statewide. Based on low prevalence detected in 1995-1997 surveys, future harvest and road-kill surveys will be conducted at about 5-yr intervals to estimate prevalence, monitor trends, and compare rates among endemic elk DAUs. Epizootiological Studies Epidemiology of naturally-occurring cwp in captive elk (Miller and Wild): None of the 19 captive adult elk held at CDOW's Foothills Wildlife Research Facility developed clinical CWD during June 1997-May 1998; one female (G86) that died showed no histological evidence of ewe. We will continue to monitor this naturally-infected captive herd and examine all mortalities for evidence of ewe. A manuscript describing CWD epidemiology in captive elk (Miller et al., 1998) was accepted for publication. Judging from requests for prepublication copies, this manuscript appears to be of considerable interest to those investigating and managing recently recognized foci of CWD in privately-owned captive elk throughout the us and Canada. ACKHQWLEDGMEHTS The statewide CWD monitoring and surveillance program described here relies heavily on efforts of dedicated field personnel throughout the Colorado Division of Wildlife, and truly represents a division-wide effort to improve our understanding and management of this important disease problems. In addition to those specifically listed, we collectively thank all of those regional and area biologists, district and area wildlife managers, volunteers, deer and elk hunters, and others who assisted by submitting suspect cases, harvested animals, or road-killed animals throughout the year. LITERATURE CITED Miller, M. w. 1997. Monitoring and managing wildlife chronic wasting disease in Colorado. in Wildlife Research Report, Mammals Research, Federal Aid Projects, Job Progress Report, Project W-153-R-10, WP2, Jl7. Colorado Division of Wildlife, Fort Collins, Colorado, USA, pp. 37-46. �233 1998. Monitoring and managing wildlife chronic wasting disease in deer. in Wildlife Research Report, Mammals Research, Federal Aid Projects, Job Progress Report, Project W-153-R-11, WP3001, T3. Colorado Division of Wildlife, Fort Collins, Colorado, USA, in press. ___ ,M.A. Wild, and E. s. Williams. 1998. Epizootiology of chronic wasting disease in captive Rocky Montain elk. J. Wildl. Dis. 34: 532538. O'Rourke, K. I., T. v. Baszler, J.M. Miller, T. R. Spraker, I. SadlerRiggleman, and D. P. Knowles. 1998. Monoclonal antibody F89/160.1.5 defines a conserved epitope on the ruminant prion protein. J. Clin. Microbiol. 36: 1750-1755. Spraker, T. R., M. W. Miller, E. S. Williams, D. M. Getzy, W. J. Adrian, G. G. Schoonveld, R. A. Spowart, K. I. O'Rourke, J.M. Miller, and P.A. Merz. 1997. Spongiform encephalopathy in free-ranging mule deer (Odocoileus hemionus), white-tailed deer (Odocoileus virginianus), and Rocky Mountain elk (Cervus ·elaphus nelsoni) in northcentral Colorado. J. Wildl. Dis. 33:1-6. Williams, E. s., ands. Young. 1980. Chronic Wasting disease of captive mule deer: A spongiform encephalopathy. Journal of Wildlife Diseases 16: 8998. ___ ,and___ 1982. Spongiform encephalopathy of Rocky Mountain elk. Journal of Wildlife Diseases 18: 465-471. ___ ,and___ 1992. Spongiform encephalopathies in Cervidae. Revue Scientifique et Technique Office International des Epizooties 11: 551567. ___ ,and___ 1993. Neuropathology of chronic wasting disease in mule deer (Odocoileus hemionus) and elk (Cervus elaphus nelsoni). Veterinary Pathology 30: 36-45. C Wildlife Research Veterinarian �234 Table 1. Results of 1997 CWD harvest surveys -- archery, muzzleloader, & rifle seasons. ELK DAU GMU # Examined # Positive E4 7 104 0 8 184 0 191 30 0 9 9 0 19 127 1 Total 454 1 20 199 0 Total 199 0 E9 Prevalence (95% Cl) 0.002 (0.0001-0.012) 0 (0-0.019) �281 Colorado Division of Wildlife Wildlife Research Report July 1999 JOB PROGRESS REPORT State of - - - - ~ " ' " " "Colorado -"'""""''-=-------Project No. _ _ _. . _W.:. . --=1=5;:;.. 3--=-R-=--=12;___ _ _ __ Work Package No. -~3....0___0=2_ _ _ _ _ __ Task No. _ _ _ _ _........;::3;,.____ _ _ _ __ Cost Center 3430 Mammals Program Elk Management Monitoring and Managing Chronic Wasting Disease in Elk Period Covered: July 1, 1998 -June 30, 1999 Authors: M. W. Miller and C. T. Larsen Personnel: K. I. O'Rourke, T. R Spraker, E. Wheeler, M.A. Wild, and E. S. Williams ABSTRACT Elk from throughout Colorado were examined for occurrence of chronic wasting disease using targeted surveillance. Between June 1998 and May 1999, 7 chronic wasting disease (CWD) cases were diagnosed . from among 12 "suspect" elk submitted from endemic game management units (GMUs) in northeastern Colorado. CWD was not diagnosed in any of 4 additional "suspect" elk submitted from elsewhere in Colorado. No cases ofCWD occurred among 23 adult elk held at CDOW's Foothills Wildlife Research Facility. �282 �283 MONITORING AND MANAGING CHRONIC WASTING DISEASE IN ELK M. W. Miller and C. T. Larsen P. N. OBJECTIVES (I) Design, conduct, and report results of: (a) targeted surveillance to estimate and detect changes in distribution of chronic wasting disease (CWD) in free-ranging elk populations; and (b) harvest or road-kill surveys to estimate and detect changes in prevalence of CWD in enzootic elk populations. (2) Design, conduct, and report results of experimental studies using captive elk naturally or experimentally infected with CWD. AGREEMENT OBJECTIVES (1) Conduct and report results of targeted surveillance to estimate and detect changes in distribution of CWD in free-ranging elk populations statewide. ( 2 ) Observe epizootiological features of naturally-occurring CWD in captive elk. MATERIALS AND METHODS Surveillance We monitored elk populations throughout Colorado for occurrence of CWD using a combination of targeted surveillance and harvest or road-kill surveys. These were organized and conducted as follows: Targeted(= clinical disease) surveillance: Elk showing clinical signs consistent with those seen in chronic wasting disease were collected by field personnel statewide and brain tissues examined for evidence of spongiform encephalopathy. Toe "suspect case" profile was defined as follows: • Species: elk • Age: 2: 18 months • Signs: emaciated and abnormal behavior &/or indifference to human activity &/or increased salivation &/or tremor, stumbling, incoordination &/or difficulty or inefficiency in chewing/swallowing &/or increased drinking and urination Where possible, submissions were subjected to complete necropsy; in some situations, only heads were available for examination and sampling. In all cases, histopathology of brain tissue (W"tlliams �284 and Young 1993) was used to diagnose CWD; in some cases, immunohistochemistry (O'Rourke et al., 1998) or other ancillary tests were used to confirm or support diagnoses. Harvest surveys: No elk harvest surveys were planned for this segment. Epizootiological Studies Epidemiology of naturally-occurring CWD in captive elk (Miller and Wild): We observed captive adult elk (n = 23) held at CDOW's Foothills Wildlife Research Facility for clinical signs ofCWD and submitted all mortalities for complete necropsy and histopathological examination. RESULTS AND DISCUSSION Surveillance Targeted(= clinical disease) surveillance: Between June 1998 and May 1999, 7 chronic wasting disease (CWD) cases were diagnosed among 12 "suspect" elk submitted from endemic game management units (GMUs) in northeastern Colorado; CWD was not diagnosed in any of 4 additional "suspect" elk submitted from elsewhere in Colorado. Since 1981, 23 clinical CWD cases have been diagnosed in elk from Larimer County GMUs (Fig. I). Harvest surveys: No elk harvest surveys were conducted in this segment, and none are planned for 1999-2000. Epizootiological Studies Epidemiology of naturally-occurring CWD in captive elk (Miller and Wild): None of the 23 captive adult elk held at CDOW's Foothills Wildlife Research Facility developed clinical CWD during June 1998-May 1999; 3 female elk that died during that period were not infected. We will continue to monitor this naturally-infected captive herd and examine all mortalities for evidence of CWD. A manuscript describing CWD epidemiology in captive elk during 1986-1997 was published in July (Miller et al., 1998). ACKNOWLEDGMENTS The statewide CWD monitoring and surveillance program described here relies heavily on efforts of dedicated field personnel throughout the Colorado Division of Wildlife, and truly represents a division-wide effort to improve our understanding and management of this important disease problems. In addition to those specifically listed, we collectively thank all of those regional and area biologists, district and area wildlife managers, volunteers, elk hlfflters, and .others 'Who assisted by submitting suspect cases, harvested animals, or road-killed animals throughout the year. �285 LITERATURE CITED Miller, M. W., M. A. Wild, and E. S. Williams. 1998. Epizootiology of chronic wasting disease in captive Rocky Montain elk. J. Wildl. Dis. 34: 532-538. O'Rourke, K. I., T. V. Baszler, J. M. Miller, T. R. Spraker, I. Sadler-Riggleman, and D. P. Knowles. 1998. Monoclonal antibody F89/160.1.5 defines a conserved epitope on the ruminant prion protein. J. Clin. Microbiol. 36: 1750-1755. Williams, E. S., and S. Young. 1993. Neuropathology of chronic wasting disease in mule deer (Odocoileus hemionus) and elk (Cervus elaphus nelsoni). Veterinary Pathology 30: 36-45. Prepared by _ _ _ _ _ _ _ _ _ _ __ Michael W. Miller Wildlife Research Veterinarian oCheyenne: South Platte River River 50 100 _ ___.__ _...____ km Figure 1. Between March 1981 and June 1999, 23 clinical CWD cases have been diagnosed in elk from GMUs in Larimer County, Colorado. �286 �233 L Colorado Division of Wildlife Wild)ife Research Report July 2000 JOB PROGRESS REPORT Stateof _ _ _ _ _ _C=o=l=orad==o_ _ _ __ Cost Center 3430 Project No. _ _ _ _ _W~-1=5___3___ -R___-=13=------ Mammals Program Work Package No. _____3---00=2.___ _ __ Elk Management Task No. 1 ---------------- Monitoring and Managing Chronic Wasting Disease in Elk Period Covered: July 1, 1999 - June 30, 2000 Authors: M. W. Miller and C. T. Larsen Personnel: K. I. O'Rourke, T. R Spraker, E. Wheeler, M.A. Wild, and E. S. Williams \ . ) ABSTRACT Elk from throughout Colorado were examined for occurrence of chronic wasting disease using targeted surveillance. Between June 1999 and May 2000, 5 chronic wasting disease (CWD) suspect elk were submitted from endemic game management units (GMUs) in northeastern Colorado; none of those suspects were infected :with CWD. In addition, CWD was not diagnosed in any of 7 other suspect elk submitted from elsewhere in Colorado. Locoism was diagnosed in 5 suspect elk (all from Jefferson County), and cause of emaciation could not be determined in 5 cases (including 4 from endemic GMUs). One case ofCWD occurred among 23 adult elk held at CDOW's Foothills Wildlife Research Facility. The last case in this herd occurred nearly 5 years ago. Exposure to run-off from pens holding CWD-infected deer was the most likely source of infection. BDOW016787 ��235 MONITORING AND MANAGING CHRONIC WASTING DISEASE IN ELK M. W. Miller and C. T. Larsen P. N. OBJECTIVES (I) Design, conduct, and report results of: (a) targeted surveillance to estimate and detect changes in distribution of chronic wasting disease (CWD) in free-ranging elk populations; and (b) harvest or road-kill surveys to estimate and detect changes in prevalence of CWD in enzootic elk populations. (2) Design, conduct, and report results of experimental studies using captive elk naturally or experimentally infected with CWD. AGREEMENT OBJECTIVES (I) Conduct and report results of targeted surveillance to estimate and detect changes in distribution of CWD in free-ranging elk populations statewide. f (2) Observe epizootiological features of naturally-occurring CWD in captive elk. MATERIALS AND METHODS Surveillance We monitored elk populations throughout Colorado for occurrence of CWD using a combination of targeted surveillance and harvest or road-kill surveys. These were organized and conducted as follows: Targeted(= clinical disease) surveillance: Elk showing clinical signs consistent with those seen in chronic wasting disease were collected by field personnel statewide and brain tissues examined for evidence of spongiform encephalopathy. The "suspect case" profile was defined as follows: • Species: elk • Age: ::::, 18 months • Signs: emaciated and abnormal behavior &/or indifference to human activity &/or increased salivation &/or tremor, stumbling, incoordination &/or difficulty or inefficiency in chewing/swallowing &/or increased drinking and urination �236 Where possible, submissions were subjected to complete necropsy; in some situations, only heads were available for examination and sampling. In all cases, histopathology of brain tissue (Williams and Young 1993) was used to diagnose CWD; in some cases, immunohistochemistry (O'Rourke et al., 1998) or other ancillary tests were used to confirm or support diagnoses. Harvest surveys: No elk harvest surveys were planned for this segment. Epizootiological Studies Epidemiology of naturally-occurring CWD in captive elk (Miller and Wild): We observed captive adult elk (n = 23) held at CDOW's Foothills Wildlife Research Facility for clinical signs ofCWD and submitted all mortalities for complete necropsy and histopathological examination. RESULTS AND DISCUSSION Surveillance Targeted(= clinical disease) surveillance: Between June 1999 and May 2000, no chronic wasting disease (CWD) cases were diagnosed among 5 "suspect" elk submitted from endemic game management units (GMUs) in northeastern Colorado; CWD was not diagnosed in any of 7 additional "suspect" elk submitted from elsewhere in Colorado. Since 1981, 23 clinical CWD cases have been diagnosed in elk from Larimer County GMUs (Fig. I). Harvest surveys: No elk harvest surveys were conducted in this segment, and none are planned for 20002001. Epizootiological Studies Epidemiology of naturally-occurring CWD in captive elk (Miller and Wild): One of the 23 captive adult elk held at CDOW's Foothills Wildlife Research Facility developed clinical CWD during June 1999-May 2000. The last case in this herd had occurred nearly 5 years earlier (February, 1995; Miller et al. 1998). Exposure to run-off from pens holding CWD-infected deer was the most likely source of infection. We will continue to monitor this naturally-infected captive herd and examine all mortalities for evidence of CWD. ACKNOWLEDGMENTS The statewide CWD monitoring and surveillance program described here relies heavily on efforts of dedicated field personnel throughout the Colorado Division of Wildlife, and truly represents a division-wide effort to improve our understanding and management of this important disease problems. In addition to those specifically listed, we collectively thank all of those regional and area biologists, district and area wildlife managers, volunteers, elk hunters, and others who assisted by submitting suspect cases, harvested animals, or road-killed animals throughout the year. �237 Number of cllnlcal cases per MU In elk I Lara RI River Cache la Pou River 50 100 - ~ ~ ~1~ - k m Figure 1. Between March 1981 and May 2000, 23 clinical CWD cases have been diagnosed in elk from GMUs in Larimer County, Colorado. Cases reported in Wyoming are also shown to provide context on overall geographic distribution of CWD in free-ranging elk. Figure modified from Miller et al. (2000). LITERATURE CITED Miller, M. W., M. A. Wild, and E. S. Williams. 1998. Epizootiology of chronic wasting disease in captive Rocky Montain elk. J. Wildl. Dis. 34: 532-538. Miller, M. W., E. S. Williams, C. W. McCarty, T. R. Spraker, T. J. Kreeger, C. T. Larsen, and E. T. Thome. 2000. Epizootiology of chronic wasting disease in free-ranging cervids in Colorado and Wyoming. Journal of Wildlife Diseases 38: 676-690. O'Rourke, K. I., T. V. Baszler, J.M. Miller, T. R. Spraker, I. Sadler-Riggleman, and D. P. Knowles. 1998. Monoclonal antibody F89/160.1.5 defines a conserved epitope on the ruminant prion protein. J. Clin. Microbiol. 36: 1750-1755. Williams, E. S., and S. Young. 1993. Neuropathology of chronic wasting disease in mule deer (Odocoileus hemionus) and elk (Cervus elaphus nelsoni). Veterinary Pathology 30: 36-45. Prepared by _ _ _ _ _ _ _ _ _ _ _ __ Michael W. Miller Wildlife Research Veterinarian � https://cpw.cvlcollections.org/files/original/c68d6536d44147a37876d9868aa2aac0.pdf 1b145e1e8eea81ec415403a31aaeb1ae PDF Text Text 273 Colorado Division of Wildlife Wildlife Research Report July2000 JOB PROGRESS REPORT State of Colorado Cost Center 3430 Project No. W-153-R-13 Mammals Research Program Work Package No. ----"'3--=-0.a.;03"--------- Carnivorous Mammals Conservation Study No. 1 Black bear muscle physiology Period Covered: July 1, 1999-June 30, 2000 Author: T.D.I. Beck Personnel: T. Beck, R. Firth, CDOW; H. Harlow, M. Hooker, T. Lohuis, L. Willmarth, U. of Wyoming \ ) ABSTRACT In 1999 10 black bears were captured in Grand County for studies on muscle disuse atrophy. Fat and muscle biopsies and blood samples were obtained from all ten bears. No direct in vivo measurements of hind limb muscle strength were obtained because of problems with electrical and computer equipment. Eight radio-collared bears were selected for study in December 1999. Direct muscle strength measurements, fat biopsies, muscle biopsies and blood samples were obtained from all. Six of these 8 bears were re-sampled successfully during the late-den period (March 2000). The other 2 bolted from their dens early and were not captured. Five of the bears handled in March 2000 were equipped with knockdown collars to facilitate recapture in June 2000. None of these collars worked as designed and could not be triggered to inject the immobilizing drug. Conventional trapping produced 6 bears for the comparative summer sample. Two of the ATS knock-down collars have been·retrieved and were submitted for examination to determine cause of failure. Data summary and analyses are still being conducted by U. of Wyoming personnel as the,major part of a Ph.D. dissertation. Final reports will be in the next segment. ) �275 BLACK BEAR MUSCLE PHYSIOLOGY Thomas D. I. Beck SEGMENT OBJECTIVES I. Obtain early-denning and late-denning comparative measurements of muscle strength, morphology, and protein metabolism in order to accurately describe the lack of muscle disuse atrophy. METHODS AND MATERIALS Black bears (Ursus americanus) were captured in cage traps throughout Grand County, Colorado during July 1999 and June 2000. All bears estimated to be adults were instrumented with standard VHF radio transmitter collars. Bears were located by aerial telemetry during the early-denning period ofmidNovember to December. Ground-based telemetry confirmed den entry. The early-season den work was begun as soon as all study bears had denned. Aerial tracking was used in early March to determine if bears had changed dens. All late-denning season work began in mid-March. The primary purpose of this project was to provide study bears for physiological research that is being funded through the Univ. of Wyoming under a National Science Foundation grant. This collaboration has been a joint undertaking since 1993. During den handlings, all bears were initially immobilized with Telazol administered by jab stick at the dosage rate of 7 mg/kg. Following imrilobiliz.ation, bears were removed from the den and placed on a foam pad which was on a tarp for insulation. When precipitation seemed likely, an overhead tarp was erected to keep rain/snow off bear. A sterile opthalmic ointment was placed in each eye and a stocking mask placed on the bear to shield the eyes. Each bear was then weighed. Whole body fat content was measured using bioelectrical impedence (Baumgartner et al. 1990). In vivo muscle strength of the hind limb was measured using a force transducer apparatus modified from a design used for human clinical research by one of the project collaborators (P. Iazzo, U. of Minn.). Electrical stimulation of the perinea! nerve caused the tibialis anterior to contract. This contraction was measured on a force plate. Using sterile surgical techniques, a 1-2 gm muscle biopsy was removed from the vastus Iateralis of the quadraceps. This sample was equally divided into 3 subsamples for later analyses for total protein content, total nitrogen content, RNA content, DNA content, muscle morphometry and fiber type, ubiquitin concentration, and in vitro muscle strength. A small fat sample was also removed from subcutaneous fat reserves for identification of fatty acid profiles. A 15-ml blood sample was obtained by venipuncture in the inguinal region. Blood samples were analyzed for key amino acid levels of activity, ubiquitin messenger RNA, and a general chemistry screen. All of the surgical procedures and laboratory -analyses were conducted by Dr. Henry Harlow and graduate student Tom Lohuis of the Univ. of Wyoming under a joint NSF grant. During the early-denning sample period, all biopsies were taken from the left hind limb while during the late-denning sample period the right hind limb was used. The in vivo muscle strength measurements were taken on the left hind limb in both periods. Following the work-up, all bears were placed back in their dens and the den entrances covered. In addition, 5 of the black bears handled in March 2000 were instrumented with knock-down collars manufactured by Advanced Telemetry Systems. These collars held syringes loaded with ketamine/xylazine �276 which could be activated by radio transmitter. The plan was to obtain the same data as collected in winter during the summer from the identical study bears. As a back-up to this system, traditional trapping protocols with cage traps would be implemented if needed to obtain a summer cohort of bears. Bears captured in June 2000 for this project were not radio instrumented and radio collars were removed from . any bears handled in June 2000. RESULTS AND DISCUSSION During July 1999 IO black bears were captured and collared. In vivo strength measurements were not obtained on any of these bears because of problems in both software and electrical hardware needed. The equipment had been modified to reduce the weight and volume of equipment. Bench tests of the equipment had been successful but we were unable to make the needed corrections once the field problems occurred during July 1999. The equipment was corrected before the den seasons. However, muscle and fat biopsies were obtained from all ten (9 males, 1 female) bears. During the early denning period (Nov-Dec 1999) 7 of these bears were located as well as 8 bears collared in previous years. From this group, a subset of 8 (7 females, 1 male) were selected for denning studies. Criteria for selection were primarily gender (females move less than males) and ease of den access based on topography, avalanche danger, and land ownership. Muscle and fat biopsies, blood samples, and in vivo muscle strength measurements were obtained for all 8 bears. Sampling was repeated in March 2000 on 6 of these 8 bears. All samples and measurements were successfully obtained on these six. One 1female black bear exited her den unusually early (early March) and did not return while another female exited her den upon the approach of the field crew. Neither of these bears re-denned so could not be captured. Five of these bears (4 females, 1 male) were fitted with the ATS knock-down collars while the remainder had a conventional radio collar. During June 2000, 4 of the 5 bears equipped with ATS knock-down collars were located via aerial and ground telemetry. Numerous attempts were made to activate the syringes on the collars for each of the 4 bears. None worked. Conventional trapping resulted in a total of 6 bears captured. One recapture had an ATS knock-down collar on while another recapture had a conventional collar. Additionally, a male equipped with the ATS knock-down collar was struck by a vehicle and killed. Thus we obtained 2 of the collars for analysis of the malfunctions. The 6 bears captured in June 2000 (5 males, 1 female) provided muscle biopsies, blood samples, and measurements of muscle strength. All laboratory analyses have been conducted by collaborators at University of Wyoming. However, data analysis has not been completed but should be in final form and will be the major portion of the Ph.D. dissertation of Tom Lohuis (expected completion Spring 2001). All surgical incisions were carefully examined during subsequent handlings. All healed without problems. No further data will be collected for these studies in the next segment. Analyses will be completed and an effort will be made in December 2000 to locate collared bears and remove as many collars as is feasible. LITERATURE CITED Baumgartner, R. N., W. C. Chumlea, and A. F. Roche. 1990. Bioelectrical impedance for body composition. Exercise and Sport Reviews 18:193-224. � https://cpw.cvlcollections.org/files/original/5e679ecf369ff3b260c49cfba6dcf0ab.pdf 96582220480cfba835ae39a0e0992563 PDF Text Text 235 Colorado Divison of Wildlife Wildlife Research Report July 1998 JOB FIHAL REPORT state of Colorado cost center 3430 Project No. W-153-R-11 Mammals Program Work Package No. _ _,,3~P~P~3_ _ _ __ Predatory Mammals Management Task No. Use of Sport Hunting to Reduce Puma Depredation on sheep Period Covered: July 1, 1997 - June 30, 1998 Author: Thomas D. I. Beck Personnel: T. Beck, J. Madison, J. Wallace; CDOW ABSTRACT A puma density sampling protocol was prepared based on probability sampling along line transects. A prototype sampling unit was designated. However, the appropriate snow conditions did not occur in Jan.-Mar., 1998 to allow for sampling. An evaluation of the logistical constraints on the technique, along with loss of management authority, resulted in the termination of the study. �237 Use of Sport Bunting to Reduce Puma Depredation to Sheep P.H. objectives 1. Evaluate the efficacy of liberal sport hunting quotas to reduce the density of puma. 2. Investigate the relationship of puma density to depredation levels on domestic sheep. segment Objectives 1. Compile historic data on sport kill of puma and puma depredation on sheep in DAU L-7. 2. Prepare detailed protocol on transect sampling and puma density estimation. 3. Conduct puma density estimation transects on 2 study areas. METHODS AND MATERIALS Kill of puma (.E.e.l..i.a. concolor) by hunters was tabulated from mandatory check files for the years 1988-1997 for DAU L-7. The frequency distribution of kill by time was calculated in one-week intervals in hopes of identifying periods where sampling would have minimal impact on hunting activities. Based on location of historic kills and terrain considerations, a pilot study site of approximately 400 Jan2 was delineated in the Piceance creek drainage. A detailed sampling protocol was prepared based on the techniques described by Becker (1991) and Van Sickle and Lindsey (1991). RESULTS AND DISCUSSION Hunter kill of puma increased from 12 in 1988 to 80 in 1997 in DAU L-7. The marked increase in kill began in 1993 and was the result of greater quotas allocated in hopes of reducing puma depredation on domestic sheep. During the period 1995-1997 a total of 209 puma were taken by hunters and an additional 30 were taken by either landowners or predator control agents in DAU L-7. Game damage payments for puma depredation statewide varied from $46,000 to $60,000 per year during 1988-1991. Since 1992 the annual payments have varied from $93,000-$132,000. The relative importance of possible causative factors is unknown. Approximately 50% of the statewide puma damage payments are made in DAU L-7. A detailed sampling protocol was prepared (Appendix A) but never implemented. During the January-March 1998 sampling period adequate snow cover did not exist in the sampling block. In trying to implement the sampling scheme several serious logistical impediments became apparent. Scheduling of helicopter time was problematic because of restrictions with state contracts and the higher priority status of ungulate census and survival work. The puma census would have to compete with other on-going work for both helicopter and personnel time. �238 The logistical impediments, along with the changed management status of puma in Colorado (shared authority with Colorado Dept. of Agriculture), make the successful completion of this job improbable. Thus it was decided by participants to cease the effort at density estimation for future years. �239 Appendix A PROTOCOL FOR PROBABILITY SAMPLING OF PUMA DEIISITY SAMPLING AREA: Area should be a rectangle of at least 390 km 2 , and must be at least 13 km wide on the shortest axis (baseline axis). The long axis (transect axis, 30 km) should be perpendicular to the majority of the major drainages in the area. You must delineate a specific rectangular area in order to make inclusion judgments. TRANSECTS: At least 3 transects should be flown in a straight line along the long axis for the entire 30 km length. Transect 1 should be randomly selected to start within the first 3 kms of the baseline axis. The remaining transects should be spaced 4 km distant (Example: Transect 1 begins at baseline point of 2 km, No. 2 at 6 km, No. 3 at 10 km). Flight time will be about 60 minutes to fly a 30 km transect plus about 10 minutes of search time per puma track set. MODIFICATIONS: If a 30 km transect cannot be fitted into the areas of concern, the long axis can be shortened. However, total area should remain at least 400 km2 and transects should be separated by at least 4 km. Thus a 20X20 km area would have 5 transects flown, of 20 km each. Long (30 km) transects are more efficient in terms of flight time. SURVEY PERIOD: The helicopter surveys should be flown 2-3 days following a snowfall which is sufficiently deep as to not melt out within 48 hours and completely cover old tracks (about 8 cm). Very deep snows (>20 cm) restrict puma movement and are to be avoided. Optimum snow depth would be 8-12 cm. There must be at least one complete night for movement following the snow. A 2-night period is better unless very high winds are occurring; however, more than 2 nights creates difficulty because of other animal tracks and wind-blown snow. PROCEDURE: Surveys should be conducted by a crew consisting of a trained observer and a navigator. The navigator should have detailed knowledge of the terrain. While biologists may be excellent observers, we should try to utilize experienced puma hunters for observers as they will more easily detect the track patterns of puma. All locations should be recorded by UTM. Fly a straight line along Transect 1 until a puma track is encountered. Record location. Fly the backtrack until you find where puma started moving following snowfall; record the location of the point where the track is farthest from the transect. Fly the front track until you locate puma (often you will not see animal but will see where tracks end in heavy cover); record location of the point where the track is farthest from the transect. Because of the non-linear nature of travel, the beginning and ending points may not be the farthest distance traveled parallel to the baseline axis. Record the extreme distances from the transect. Return to original transect line at the point where track was located. Continue along the line until the next puma track is encountered. Repeat the process to estimate travel distance for each set of tracks. The objective is to obtain a straight-line distance of travel parallel to the base-transect. At the end of transect 1, move 4 Jans along the baseline and return along transect 2, which parallels transect 1. Multiple puma tracks traveling together (female w/kittens or male-female pair) should only be counted as 1 track set. However, make a notation of the group size. There are separate analysis procedures which can be used if this is a common occurrence. �240 Do not count kitten tracks traveling separate from the mother. This is an unlikely occurrence and distinguishing from bobcat tracks while airborne is quite problematic. Kill sites will be characterized by lots of tracks in the near vicinity of the kill. Estimate the diameter of the circle which would include all the tracks. This diameter will be used as the travel distance. Most straight-line travel distances will be less than 3 km. However, in the event that a puma track crosses 2 transects, it should be separately recorded for each transect. Data to be recorded: location of each puma track along each transect; location of farthest points of travel of puma perpendicular to the transect; all in UTM's. Data to be calculated: straight-line distance parallel to baseline moved by each puma, in km's. Data for any puma where >50% of the travel distance was outside the designated rectangular survey area will not be used in the calculations. ANALYSIS: Population size is estimated from the following equations, from Becker (1991) and Van Sickle and Lindzey (1991): P;=x/(D/q) ½=L (lip;) where Pi= probability that the ith puma i sample; xi= distance, parallel to the baseltoetatnedeledtb~ Yhb aystpmaalcD = length of baseline; q = number of transects per systematic sample; and Tj = population estimate for the jth systmatic sample. since we will not be flying replicate samples for an area, variance of the estimate will be estimated using jackknife procedures and dividing the transects into segments (McDonald and Manly 1989). This will be done by a contract statistician. While replication is the preferred methodology, helicopter costs make this prohibitive. ASSUMPTIONS FOR PROBABILITY SAMPLING PROCEDURE: 1. 2. 3. 4. 5. 6. 7. 8. All animals move during the survey period. Puma tracks are readily recognizable. All animal tracks are continuous. Animal movements are independent of the sampling process. Post-snowstorm tracks can be distinguished. All puma tracks crossing a transect will be observed. Study area is rectangular in shape. All the transects are oriented perpendicular to the baseline axis. LITERATURE CITED Becker, E. F. 1991. A terrestrial furbearer estimator based on probability sampling. J. Wildl. Manage. 55(4):730-737. Van Sickle, w. D. and F. G. Lindzey. 1991. Evaluation of a cougar population estimator based on probability sampling. J. Wildl. Manage. 55(4):738743. � https://cpw.cvlcollections.org/files/original/06dc6b096d152e191d3c010703b1a9b0.pdf 95f42b9124861b075e9f8e6280eb7479 PDF Text Text 245 Colorado Division of Wildlife Wildlife Research Report July 1998 JOB PROGRESS REPORT State of Colorado cost center 3430 Project No. W-153-R-ll Mammals Program Work Plan No. 3004 Management of other ungulates Task No. Strategies for Managing Pasteurellosis in Mountain Sheep Populations Period Covered: July 1, 1997 - June 30, 1998 Authors: M. W. Miller and H.J. McNeil Personnel: s. Berry, J. George, K. Larsen, J. Vayhinger, T. Verry ABSTRACT We continued investigations of multivalent Pasteurella haemolytica supernatant vaccines in captive bighorn sheep (Ovis canadensis). A vaccine lot(PhSV lot #970520) that combined bighorn and domestic strains appears to contain antigenicity comparable to that of the original multivalent vaccine, and was selected for use in future laboratory and field studies. To evaluate vaccine delivery options, 30 captive bighorns were divided into 3 groups of 10 based on vaccine history and baseline P. haemolytica leukotoxin neutralizing antibody titers; treatment groups included hand injection, biobullet implantation and oral gavage of PhSV. Serum leukotoxin-neutralizing antibody titers differed among delivery treatments (P = 0.009). Neutralizing titers were highest among hand-injected bighorns; no serum antibody response was detected among bighorns vaccinated orally. Although neutralizing titers were lower among implanted bighorns than in the hand injected controls (P < 0.021), seroconversion rates did not differ (implantation= 6/10, hand injection= 9/10; P = 0.303). Although our data demonstrated that handinjection elicits higher absolute titers than either biobullet implantation or oral vaccination, biobullet implantation may also stimulate effective antibody responses to P. haemolytica supernatant vaccines. Further evaluation of biobullet vaccination against pneumonic pasteurellosis in free-ranging populations of wild bighorn sheep appears warranted. A total of 110 free-ranging bighorn sheep from 4 different herd units. in central Colorado were vaccinated with PhSV during January-March 1998. No adverse reactions or mortalities were detected among vaccinated bighorns through June 1998. Ongoing monitoring in the Tarryall-Kenosha herd complex will provide additional data on efficacy of vaccination to protect freeranging bighorns from pasteurellosis under field conditions. �247 EXPERIMENTS TO IDENTIFY AND MANAGE STRESS IN MOUNTAIN SHEEP POPULATIONS M. W. Miller and H.J. McNeil P, H, OBJECTIVES 1. Design, conduct, and report on experiments evaluating Pasteurella haemolytica vaccines and vaccine delivery systems and identify those with potential application in managing free-ranging bighorn populations. 2. Design, conduct, and report on field experiments evaluating management strategies for preventing pasteurellosis epizootics in bighorn populations SEGMENT OBJECTIVES 1. Conduct and report results of an experiment evaluating options for delivering a multivalent Pasteurella haemolytica supernatant vaccine to bighorn sheep. 2. Provide multivalent Pasteurella haemolytica supernatant vaccine for use in select bighorn sheep management activities statewide. STRATEGIES FOR MANAGING PASTEURELLOSIS IN MOUNTAIN SHEEP POPULATIONS Inability to control infectious disease outbreaks and subsequent mortality in mountain sheep populations represents a significant obstacle to long-term success in their management. Although the "bighorn pneumonia complex" has been studied intensively for over 3 decades, little is known about many aspects of its etiology and epizootiology. Moreover, management interventions recommended for preventing or controlling this problem remain untested. Although viral, bacterial, and parasitic agents have all been incriminated in these outbreaks, Pasteurella spp. are perhaps the most common pathogens associated with bronchopneumonia in bighorns. Two species, P. haemolytica and P. multocida, and several biotypes and/or serotypes within those species, have been isolated from bighorns during epizootics. Unfortunately, despite extensive diagnostic and experimental investigation, the epizootiology of pasteurellosis in wild bighorn populations is poorly understood. In the absence of knowledge about the epizootiology of pasteurellosis, effective strategies for managing pneumonia in bighorn populations have not emerged. Here, we report on a series of ongoing research studies designed to improve knowledge about various aspects of pasteurellosis epizootiology and management in bighorn sheep. METHODS ARD MATERIALS Refinement of experimental Pasteurella haemolytica supernatant vaccine (McNeil, Miller, and Shewen): A rabbit study was previously conducted with Pasteurella haemolytica supernatant vaccine (PhSV) lot #970520 (Miller and McNeil, 1997). Rabbits were exsanguinated after receiving three 0.5 ml doses �248 of vaccine at 2-wk intervals. All rabbits responded to vaccination by producing anti-leukotoxin antibodies. Protein profiles from Western blots (blotted with vaccinated rabbit serum) transferred from 12% precast PAGE gels containing the vaccine components of both PhSV lot #970520 and multivalent vaccine lot# 940902 (Miller et al., 1997) also showed strong vaccine-induced antibody reactivity. Based on these findings, a small pilot study in bighorn sheep was conducted. Six captive Rocky Mountain bighorn sheep were used. Sheep were paired on the basis of age and vaccination history. One bighorn from each pair received 2 ml PhSV lot #970520 intramuscularly; the other animals received 1 ml PhSV lot #970520 and 1 ml Presponsee mixed in the same syringe, delivered intramuscularly. (Presponsee is a commercially available supernatant vaccine of Pasteurella haemolytica Al.) We used this approach to establish whether the presence of P. haemolytica serotype Al was necessary to drive the marked increase in leukotoxin neutralizing titer seen in previous studies (Miller et al., 1997; Kraabel et al., 1998). Animals were bled prior to vaccination and again at day 7 and day 14. Serum leukotoxin neutralizing antibody titers stimulated by the two treatments were measured and compared using methods described by Miller et al. (1997). Delivery of Pasteurella haemolytica supernatant vaccine to bighorn sheep (McNeil and Miller): Thirty captive bighorn sheep were used in this study. Animals were divided into three groups of ten based on their vaccine history and baseline neutralizing titers. The three treatment groups included hand injection, biobullet implantation and oral gavage of vaccine. Four additional animals were intramuscularly injected with saline to monitor for changes in antibody titers unrelated to vaccination during the study period. PhSV lot #970520 was used for all treatments. Vaccine was lyophilized for use in biobullets and microspheres. The standard dose for intramuscular injection, 2 ml (Miller et al., 1997; Kraabel et al., 1998), yielded about 0.025 g lyophilized vaccine. Biobullets were packed with about 0.030 g lyophilized vaccine, then filled with methylcellulose filler for remote delivery. Lyophilized vaccine was also sent to Dr. Harm HogenEsch of Purdue University for encapsulation in alginate microspheres; 5 ml of alginate microsphere suspension was recommended as equivalent to a 2 ml dose of vaccine. The hand injection group received 2 ml PhSV injected intramuscularly into the left haunch. The biobullet group were vaccinated with the use of a specialized biobullet rifle from a distance of approximately 4 meters. Animals in this group had a small area on their left hindquarters shaved to facilitate confirming implantation. Animals receiving vaccine orally were dosed with the aid of a small stainless steel gavage tube. The tube was inserted into the animal's mouth and laid along the teeth; 5 ml of vaccine was then squirted into the sheep's mouth and subsequently swallowed. Animals were bled prior to vaccination and again at 1, 2, 4, 8 and 12 weeks postvaccination. Sera were collected and used to assess serological responses. Titers were measured using a leukotoxin neutralization assay, along with three separate direct agglutination assays that incorporated formanilized Pasteurella haemolytica serotypes Al, A2 or Tl0 as antigen (Miller et al., 1997; Kraabel et al., 1998). use of Pasteurella haemolytica supernatant vaccine in free-ranging bighorn .ab.e,ep (Miller, George, and Vayhinger): A total of 110 free-ranging bighorn sheep from 4 different herd units in central Colorado were vaccinated with �249 PhSV during January-March 1998 (Table 1). Bighorns from Georgetown herd unit were vaccinated just prior to translocation to the Arkansas River canyon for release; nearby resident bighorn populations appear to have recurrent problems with respiratory disease in this area, and translocated bighorns were considered potentially at risk. Bighorns in the Black Canyon and Twin Eagles herd units of the Tarryall-Kenosha complex were vaccinated in an attempt to prevent spread and/or minimize impacts of a pasteurellosis epidemic that started in the Sugarloaf herd unit in December 1997 (see WP7610-T4 for details); a few sheep in the Sugarloaf herd also were vaccinated in an attempt to enhance postepidemic survival in that subpopulation. Survival of vaccinated and unvaccinated individuals was compared opportunistically in conjunction with ongoing field studies. RESULTS AHD DISCUSSION Refinement of experimental Pasteurella haemolytica supernatant vaccine (McNeil, Miller, and Shewen): We observed no adverse effects in bighorns receiving PhSV lot #970520. All vaccinated bighorns responded with an increase in leukotoxin neutralizing titer. Based on these results, PhSV lot #970520 was judged an appropriate vaccine for use in the delivery experiment and in field applications. Delivery of Pasteurella haemolytica supernatant vaccine to bighorn sheep (McNeil and Miller): other than a mild transient lameness in animals vaccinated intramuscularly or remotely, no adverse reactions were observed following vaccination. Serum neutralizing antibody titers to P. haemolycica leukotoxin differed among delivery treatments (P=0.009) as well as among baseline titer/vaccination history groups (P=0.013). Neutralizing titers were highest among hand-injected bighorns; no serum antibody response was detected among bighorns vaccinated orally. Although neutralizing titers were lower among implanted bighorns than in the hand injected controls for ~2 weeks post vaccination (P < 0.021), seroconversion rates (defined as a ~2 log 2 increase in titers) in response to implantation (6/10) and hand injection (9/10) did not differ (P=0.303). Agglutinating antibody titers to Tl0 were high and did not differ over time or between delivery treatments. In addition, agglutinating antibody titers to Al did not vary over time or between groups. Agglutinating antibody titers to A2 were higher in the hand injected group than the orally vaccinated sheep during the time period between 2 and 4 weeks post vaccination (P < 0.026); A2 agglutinating titers in hand-injected and implanted sheep did not differ during the same time period (P = 0.07). Vaccination history/baseline titer categorization affected responses to serotype A2 surface antigens (P = 0.0001). our data support previous observations (Miller et al., 1997; Kraabel et al., 1998) that hand-injected P. haemolycica supernatant vaccines stimulate marked antibody responses in bighorn sheep; it follows that delivery via projectile syringe should stimulate similar antibody responses. Although our data demonstrate that hand-injection elicits higher absolute titers than either biobullet implantation or oral vaccination, biobullet implantation may also stimulate effective antibody responses to P. haemolycica supernatant vaccines. Whether lack of serum antibody response to oral vaccination truly reflects failure to stimulate immunity remains undetermined. Further evaluation of biobullet vaccination against pneumonic pasteurellosis in free-ranging populations of wild bighorn sheep appears warranted. �250 use of Pasteurella haemolytica supernatant vaccine in free-ranging bighorn Jlhe8l2 (Miller, George, and Vayhinger): As with previous captive animal studies (Miller et al., 1997; Kraabel et al., 1998; McNeil and Miller, above), no adverse effects were observed in free-ranging bighorns that received PhSV via hand-injection, projectile syringe, or biobullet implant. No mortality was been detected among vaccinated bighorns in the 4-6 mo after vaccination; however, little or no mortality occurred among unvaccinated bighorns in these populations during that time. We will continue to monitor survival rates among vaccinated and unvaccinated bighorns in the Tarryall and Kenosha subpopulations in conjunction with other ongoing studies. These data should be useful in evaluating efficacy of vaccination in protecting freeranging bighorns from pasteurellosis under field conditions. Kraabel, B. J., and M. W. Miller, J. A. Conlon, and H.J. McNeil. 1998. Evaluation of a multivalent Pasceurella haemolytica vaccine in bighorn sheep: Protection from experimental challenge. Journal of Wildlife Diseases 34: 325-333. Miller, M. W., J. A. COnlon, H.J. McNeil, J.M. Bulgin, and A. C. S. Ward. 1997. Evaluation of a multivalent Pasceurella haemolycica vaccine in bighorn sheep: Safety and serologic responses. Journal of Wildlife Diseases 33: 738-748 . ....._.L.-..lH.cer Wildlife Research Veterinarian Table 1. Free-ranging bighorn sheep treated with multivalent Pasceurella haemolycica supernatant vaccine during Jan-Mar 1998. Population (Herd Unit) Delivery Status Band Dart Biobullet healthy -translocated to Arkansas R. 25 0 0 healthy, future exposure likely 35 4 5 (Twin Eagles) healthy, future exposure likely 21 6 3 (Sugarloaf) pasteurellosis epidemic (12/97) 11 0 0 Georgetown (Georgetown) Tarryall-Kenosha (Black Canyon) �287 Colorado Division of Wildlife Wildlife Research Report July 1999 JOB PROGRESS REPORT -------~-~---- State of Colorado Project No. --------'W-'--.....1=5"-3--=-R"---=12=-----.........,. ___________ Work Plan No. 3004 Task No. 3 ____ ------------- Cost Center 3430 Mammals Program Management of Other Ungulates Strategies for Managing Pasteurellosis in Mountain Sheep Populations Period Covered: July I, 1998 -June 30, 1999 Authors: M. W. Miller, H.J. McNeil, and J. L. George Personnel: K. Larsen, S. Berry, J. Vayhinger ABSTRACT A pasteurellosis epidemic that began in the Sugarloaf Mountain subpopulation of the Tarryall-Kenosha bighorn herd complex extended to two adjacent subpopulations (Black Canyon, Twin Eagles) during winter 1998-1999. Sick and dead bighorns were found in the Twin Eagles subpopulation by December 1998, and in the Black Canyon subpopulation by January 1999. Based on preliminary data, vaccination with PhSV 911 mo earlier did noi appear to improve bighorn survival during subsequent epidemics. A group of bighorns translocated from Dome Rock to the Holy Cross Wilderness received PhSV during winter 1998-1999. No adverse effects have been observed in free-ranging bighorns that received PhSV last winter, or during 1997-1998. We began developing and evaluating a modified-live P. haemolytica vaccine. Using polymerase chain reaction (PCR) techniques, we screened candidate carrier strains and further characterized bacterial genomes linked to leukotoxin production. Leukotoxin A (//eta) gene of P. haemolytica TIO Eco. 100 was successfully amplified, and subsequent sequencing revealed that the I/eta gene ofthis bighorn-derived strain was approximately homologous to the published //eta gene of P. haemo/ytica TIO. We successfully ligated 123 and 1767 bp gene fragments. Initiaj. attempts to use PCR to increase yields for cloning failed, but such efforts continue. Plasmid transfers of I/eta are planned for next segment. �288 �289 EXPERIMENTS TO IDENTIFY AND MANAGE STRESS IN MOUNTAIN SHEEP POPULATIONS M. W. Miller, H.J. McNeil, and J. L. George P. N. OBJECTIVES 1. Design, conduct, and report on experiments evaluating Pasteurella haemolytica vaccines and vaccine delivery systems and identify those with potential application in managing free-ranging bighorn populations. 2. Design, conduct, and report on field experiments evaluating management strategies for preventing pasteurellosis epizootics in bighorn populations SEGMENT OBJECTIVES 1. Report results of an experiment evaluating options for delivering a multivalent Pasteurella haemolytica supernatant vaccine to bighorn sheep. 2. Provide multivalent Pasteurella haemolytica supernatant vaccine for use in select bighorn sheep management activities statewide. 3. Begin development and evaluation of a modified-live Pasteurella haemolytica vaccine for use in bighorn sheep. STRATEGIES FOR MANAGING PASTEURELLOSIS IN MOUNTAIN SHEEP POPULATIONS Inability to control infectious disease outbreaks and subsequent mortality in mountain sheep populations represents a significant obstacle to long-term success in their management. Although the "bighorn pneumonia complex" has been studied intensively for over 3 decades, little is known about many aspects of its etiology and epizootiology. Moreover, management interventions recommended for preventing or controlling this problem remain untested. Although viral, bacterial, and parasitic agents have all been incriminated in these outbreaks, Pasteurella spp. are perhaps the most common pathogens associated with bronchopneumonia in bighorns (Miller, 1999). Two species, P. haemolytica and P. multocida, and several biotypes and/or serotypes within those species, have been isolated from bighorns during epizootics. Unfortunately, despite extensive diagnostic and experimental investigation, the epizootiology of pasteurellosis in wild bighorn populations is poorly understood. In the absence of knowledge about the epizootiology ofpasteurellosis, effective strategies for managing pneumonia iif bighorn populations have not emerged. Here, we report on a series of ongoing research studies designed to improve knowledge about various aspects of pasteurellosis epizootiology and management in bighorn sheep. �290 METHODS AND MATERIALS Delivery of Pasteurella haemolytica supernatant vaccine to bighorn sheep (McNeil and Miller): Last segment we conducted an experiment evaluating options for delivering a multivalent Pasteurella haemolytica supernatant vaccine (PhSV) to bighorn sheep. A draft manuscript describing that research was accepted for publication in the Journal of Wildlife Diseases (see abstract, Appendix A). Use of Pasteurella haemolytica supernatant vaccine in free-ranging bighorn sheep (Miller, George, and Vayhinger): We continued evaluating survival of free-ranging bighorn sheep vaccinated with PhSV in the face of a pneumonia epidemic during winter 1997-1998 by comparing survival rates of vaccinated bighorns to those of unvaccinated individuals. Our evaluation was conducted in conjunction with other ongoing studies of the Tarryall-Kenosha bighorn herd complex. In addition, PhSV was used in select field operations during the winter of 1998-1999 (Table 2). We also continued exploring ways of adding Pasteurella multocida antigens to our vaccine formulation. Development of a modified-live Pasteurella haemolytica vaccine for bighorn sheep (McNeil, Lo, and Miller): We prepared a study plan and began developing and evaluating a modified-live Pasteurella haemolytica vaccine. Highlights of preliminary laboratory technique development and evaluations are reported. Bacteria: Pasteurella haemolytica TIO <Ea,_100) was obtained from a bighorn sheep that died during a pasteurellosis epidemic in southcentral Colorado in 1990 (Kraabel et al., 1997). Bacterial stocks were kept frozen at -70° C in Brain Heart Infusion Broth (BHIB) with 15 % glycerol. Thawed stock was used to inoculate blood agar plates incubated at 37° C overnight (0/N). A small volume (50 ml in a 175 ml Erlenmeyer flask) ofBHIB was inoculated with a single colony from the 0/N agar plate and incubated at 37° C, 120 rpm 0/N. The resulting culture was used to extract DNA for use in polymerase chain reactions (PCR). Preparation ofgenomic DNA: Two commercial kits were used to extract maximum amounts of genomic DNA fromP. haemolytica TIO <Eco.100). A QIAGEN Genomic-tip® (QIAGEN Inc., Mississauga, ON, Canada) was used and DNA was extracted according to the Bacterial DNA Isolation Protocol in the QIAGEN Genomic DNA Handbook. A PUREGENE® genomic DNA isolation kit (Gentra Systems, MN, USA) was also used to improve yields. Genomic DNA was used as template for subsequent PCR. PCR: PCR was utilized to amplify the leukotoxin A gene (lkta) of P. haemolytica TIO <Ea,_100). Many different primers and conditions were assessed. The gene was successfully amplified using primers based on the published P. haemolytica TIO Ieukotoxin A gene sequence (Genbank Accession Z26247). 5'-ATGGGAACTAAACTAACCCT-3' 5'-TAAGCTGCTCTAGCAAATTG-3' Conditions were optimized at 3.5 mM [Mg++] using the genomic DNA prepared by the PUREGENE® method as template. The thermal cycler (Perkin Elmer-,Cetus, NoFWalk, Connecticut, USA) was programmed for 94 C for 30 seconds, 51 C for 30 seconds, 72 C for 3 minutes and 30 seconds. This was repeated for 30 cycles. Two sections of the lkta gene also were amplified via PCR for use in subsequent ligation. Similarities between the sequence of the lkta gene for Eco. 100 and P. haemolytica Al (Genbank Accession M20730) were considered. Previous work in our laboratory (R. Lo, unpubl. data) used two existing Nael sites on the �291 lkta gene to delete a section of the gene and ligate the smaller fragments together to form a truncated gene. Although the Nae! sites on the lkta Eco.ioo were not a perfect match, using primer design, two sections were amplified. These sections incorporated exact Nae! restriction sites. The 123 base pair segment from the 5' end of the gene was amplified using the following primers: 5'-GGATCCCTIATGGGAACTAAAC-3' 5'-0TCTGGCCGGCTIGGGTIAA-3' The 1767 base pair segment from the 3' end of the gene was amplified using the following primers: 5'-TGCCGCCGGCTCTGTGGTI-3' 5'-CGAAACAATCTAGAGTIGCCAATC-3' Conditions were optimized at 3.5 mM [Mg++] using the genomic DNA prepared by the PUREGENE~ method as template. The thermal cycler was programmed for 94 C for 30 seconds, 59 C for 30 seconds, 72 C for 3 minutes and 30 seconds. This was repeated for 30 cycles. PCR was also used to try to increase the amount of ligated product. The following primers were used: 5'-GGATCCCTTATGGGAACTAAAC-3' 5'-CGAAACAATCTAGAGTTGCCAATC-3' Conditions were optimized at 3.5 mM [Mg++] using gel purified ligated product as the template. The thermal cycler was programmed for 94 C for 30 seconds, 64 C for 30 seconds, 72 C for 3 minutes and 30 seconds. This was repeated for 30 cycles. All PCR products were run on 0.75 % agarose gels (85 V for 23-45 minutes) and stained with ethidiurn bromide. Bands were visualized using ultra-violet (UV) light and the image captured by a GelDoc 1000 machine (Bio-Rad Laboratories, Mississauga, ON, Canada). PCR Product Purification: PCR products to be sequenced or used for further manipulation were purified using a QIAGEN QIAquick PCR purification kit~. Purified product was run on gels as before purification to ensure that the band in question was still visible. In instances where extraneous bands were visible, gel extraction was used to isolate the desired product. A low melting point 1.0% agarose gel was loaded with product and run at low voltage overnight (8-9 V, 1216 hours). The gel was then stained with ethidiurn bromide and visualized using UV light on a transillurninator. The desired band was cut out using a sterile scalpel blade and put into a tube. The tube was heated to melt the agar. This material was then purified in the same manner as the non-gel purified PCR products. Restriction Enzyme Digestion: Purified 123 bp and 1767 bp fragments were digested using Nae! (Promega 'Corporation, Madison, WI, USA ) with the recommended buffer for 1 hour at 37 C. The enzyme was then inactivated at 60 C for 15 minutes. A 1 µI aliquot was run on a 0.75% agarose gel to ensure product integrity. These products were to be ligated together. Restriction enzyme digestion was also used for mapping the ligated product_ Digestions were performed using Nae!, Apa! (Promega Corporation, Madison, WI, USA) and Sacl (New England Biolabs Ltd., �294 APPENDIX A EFFECTS OF DELIVERY METHOD ON SEROLOGICAL RESPONSES OF BIGHORN SHEEP TO A MULTIVALENT PASTEURELLA HAEMOLYTICA SUPERNATANT VACCINE Heather J. McNeil,1 Michael W. Miller,2 Jennifer A. Conlon,3 Ian K. Barker,1 and Patrica E. Shewen1 1 Department of Pathobiology, Ontario Veterinary College, University of Guelph, Guelph, Ontario, NlG 2Wl, Canada 2 Colorado Division of Wildlife, Wildlife Research Center, 317 West Prospect Road, Fort Collins, Colorado 80526-2097, USA 3 Merial Incorporated, 115 Transtech Drive, Athens, Georgia, 30601-1649, USA ABSTRACT: The efficacy and safety of a multivalent Pasteurella haemolytica supernatant vaccine (serotypes A2 and TIO) using different delivery systems were examined in captive Rocky Mountain bighorn sheep (Ovis canadensis canadensis). Twenty bighorn sheep were grouped according to baseline leukotoxin neutralizing antibody titers (~2 or >2 log/) and vaccination history (previously vaccinated or unvaccinated). Within these groups, animals were randomly assigned to one of two delivery treatments: hand injection (control) or biobullet implantation. All bighorns received a single dose from the same lot of vaccine (n = IO/treatment); four additional animals were injected intramuscularly with 0.9% saline as unvaccinated sentinels. Mild, transient lameness one day after hand injection or biobullet implantation was the only adverse effect. Serum neutralizing antibody titers to P. haemolytica leukotoxin differed among delivery treatments (P = 0.009) and among baseline titer/vaccination history groups (P = 0.013). Neutralizing titers were highest among hand-injected bighorns. Although neutralizing titers were lower among implanted bighorns than hand-injected controls at 1 wk (P = 0.002) and 2 wk (P = 0.021) after vaccination, seroconversion rates in response to implantation (6/10) and hand injection (9/10) did not differ (P = 0.303). Agglutinating antibody titers to TIO were high and did not vary over time or between delivery treatments. Agglutinating antibody titers to A2 in the hand-injected controls were not different (P £ 0.07) than those in bighorns vaccinated with biobullet implantation. These data demonstrate that although hand injection elicits higher absolute titers, biobullet implantation may also stimulate effective antibody responses to P. haemolytica supernatant vaccine. Further evaluation of biobullet vaccination against pneumonic pasteurellosis in free-ranging populations of wild bighorn sheep is warranted. Key words: Bighorn sheep, Ovis canadensis, Pasteurella haemolytica, pasteurellosis, vaccine delivery, vaccination. �277 \ Colorado Division of Wildlife Wildlife Research· Report July 2000 JOB PROGRESS REPORT State of _ _ _ _ _.... C=o=lo=rad=o'------- Cost Center 3430 Project No. -----'W;.;.....:-1=5=3--=R....--=-1=-3_ _ _ __ Mammals Program Work Plan No. ___................. 3004 ______ Task No. _ _ _ _ __.:.,1_ _ _ _ __ Management of Other Ungulates Strategies for Managing Pasteurellosis in Mountain Sheep Populations Period Covered: July 1, 1999 - June 30, 2000 Authors: M. W. Miller, R E. Briggs, and H. J. McNeil Personnel: K. Larsen, S. Berry, J. George, J. Vayhinger \ / I ABSTRACT We continued developing and evaluating modified-live Pasteurella spp. va~ines for use in bighorn sheep mangement. Based on preliminary successes in protecting both cattle and domestic sheep from experimental pasteurellosis, we began a pilot study to evaluate a genetically-modified live P. haemolytica vaccine (GMLPhV)in captive bighorns. Vaccine was supplied by USDA/ARS/NADC in Ames, IA. Of3 bighorns receiving about 7 x 108 colony forming units (CFUs) ofGMLPhV, 1 died about 3 days after vaccination and another showed signs of moderately severe respiratory disease from days 2-10 postinoculation (Pl); the GMLPhV strain was most likely responsible for morbidity and mortality in . vaccinated bighorns. Two previously unvaccinated bighorns housed with the 2 surviving vaccinates ·beginning 2 weeks PI seroconverted to P. haemolytica serotype 2 2-3 weeks later, suggesting colonization of these in-contact animals by GMLPhV strain despite failure to actually isolate GMLPhV strain from either in-contact bighorn. Both in-contact bighorns remanied healthy through 7 weeks postexposure. Ability of GMLPhV to protect bighorns from pasteurellosis will be evaluated via experimental challenge in early July. Further evaluation of this or a similar GMLPhV strain hinges on the vaccine's ability to prc!ect bighorns from expennental challenge. ) We also continued attempts to engineer an antigenic but nonpathogenic P. trehalosi strain for use as a modified-live vaccine in bighorns. Recent ligation attempts yielded a product of appropriate size, and restriction enzyme mapping indicated ligation was successful. As before, attempts to use polymerase chain reaction (PCR) to increase yields for cloning were disappointing, insofar as there were extraneous bands. However, we used gel purification to prepare concentrated ligated product to use in the subsequent insertplasmid ligation in E. coli. Several recipient colonies were recovered. A small number were screened and carried a plasmid of appropriate size (~ 7Kb). Restriction mapping results were encouraging. One cloned �278 construct (D22) showed favorable and consistent results. When used as a PCR template, plasmid from this clone produced a single, clear, 1889 bp band. Attempts to amplify the gfp gene using PCR were successful. A single band of appropriate size was visualized. Unfortunately, attempts to insert the gfp into the p2 l 76 containing the modified leukotoxin gene (referred to as D22) have been, as yet, unsuccessful. Short-term plans for continuing this work include PCR to amplify the green fluorescent pigment (gfp) gene with constructed blunt end restriction enzyme sites on each end. We hope to successfully clone this gene into the Nael site on the plasmid carried in D22. We then will electroporate the plasmid into P. trehalosi serotype 10 <Eco. 100) and replace the existing lktA with our LJ/ktA-gfp. We anticipate having a P. treha/osi vaccine strain available for testing in captive bighorn sheep sometime in the 00/01 segment. �279 EXPERIMENTS TO IDENTIFY AND MANAGE STRESS IN MOUNTAIN SHEEP POPULATIONS M. W. Miller, R. E. Briggs, and H. J. McNeil P. N. OBJECTIVES I. Design, conduct, and report on experiments evaluating Pasteurella haemolytica vaccines and vaccine delivery systems and identify those with potential application in managing free-ranging bighorn populations. 2. Design, conduct, and report on field experiments evaluating management strategies for preventing pasteurellosis epizootics in bighorn populations SEGMENT OBJECTIVES I. Continue development and evaluation of a modified-live Pasteurella haemolytica vaccine for use in bighorn sheep. STRATEGIES FOR MANAGING PASTEURELLOSIS IN MOUNTAIN SHEEP POPULATIONS Inability to control infectious disease outbreaks and subsequent mortality in mountain sheep populations represents a significant obstacle to Iong-tenn success in their management. Although the "bighorn pneumonia complex" has been studied intensively for over 3 decades, little is known about many aspects of its etiology and epizootiology. Moreover, management interventions recommended for preventing or controlling this problem remain untested. Although viral, bacterial, and parasitic agents have all been incriminated in these outbreaks, Pasteurella spp. are perhaps the most common pathogens associated with bronchopneumonia in bighorns (Miller 2000). Three species, P. haemolytica, P. trehalosi, and P. multocida, and several biotypes and/or serotypes within those species, have been isolated from bighorns during epidemics. Unfortunately, despite extensive diagnostic and experimental investigation, the epidemiology of pasteurellosis in wild bighorn populations is poorly understood. In the absence of knowledge about the epidemiology of pasteurellosis, effective strategies for managing pneumonia in bighorn populations have not emerged. Here, we report on a series of ongoing research studies designed to improve knowledge about various aspects of pasteurellosis epidemiology and management in bighorn sheep. METHODS AND MATERIALS Evaluation of a genetically-modified live Pasteurella haemolytica vaccine in bighorn sheep (Miller and Briggs): Based on preliminary successes in protecting both cattle and domestic sheep from experimental pasteurellosis (Briggs and Tatum, 1999), we began a pilot study to evaluate a genetically-modified live P. haemolytica vaccine (GMLPhV)in captive bighorns. The experimental GMLPhV strain ("Pasteurella haemolytica St2, Foreyt delta 3) was derived from a hemolytic, serotype 2 P. haemolytica strain (WSU-1) originally recovered from bighorn sheep cohoused �280 with domestic sheep (Foreyt 1989). The original strain was highly pathogenic in bighorns (Foreyt 1989, Foreyt et al. 1994, Foreyt and Silflow 1996), produced leukotoxin (Foreyt et al. 1994, Kraabel and Miller 1997) and carried a naturally-occurring leukotoxin gene (Green et al. 1999). Methods for genetic modifications used to produce vaccine strain are proprietary (R. Briggs, USDNARS). The candidate vaccine strain had not been tested previously in bighorn sheep. However, parallel pilot studies on safety and efficacy this vaccine strain via pareneteral delivery are underway in Pullman, WA, and other geneticallymodified P. haemolytica strains (Al from cattle and A5/6 from domestic sheep) already have been demonstrated as both safe and effective in respective domestic host species (Briggs and Tatum 1999, R. Briggs unpubl. data). Vaccinated bighorns received about 7 x 108 colony forming units (CFUs) of GMLPhV, sprayed intranasally, on 25 April 2000. Our pilot study focussed on safety, immunogenicity, transmissibility, and efficacy ofGMLPhV. To evaluate safety and immunogenicity, we used three pairs of previously unvaccinated captive bighorn sheep (4 males, 2 females). One bighorn from each pair was randomly assigned to one of two groups: intranasal vaccination with GMLPhV (n=3) or intranasal phosphate-buffered saline solution (PBS) in volumes equivalent to vaccine doses (control; n=3). Vaccinated and control groups were housed in separate I 00 m2 isolation pens. After a 2-wk acclimation period, respective vaccine and control treatments were administered as described(= wk 0). Blood for serology and pharyngeal swabs for bacteriology will be collected at wk-2, 0, 1, and 2. Safety and immunogenicity were evaluated by comparing morbidity, mortality, strain-specific P. haemolytica colonization rates, and humoral immune responses between groups over a 2-wk period. We used 4 captive bighorns from the safety study to evaluate vaccine transmissibility. Bighorns from the intranasal vaccination(= surviving prirrtiuy vaccinates; n=2) and control(= secondary vaccinates; n=2) groups were housed together in a single 200 m2 isolation pen immediately after sampling at wk 2 of the safety pilot. Additional blood for serology and nasal and pharyngeal swabs for bacteriology were collected at wk 3, 4, 5, 7, and I 0. Safety, immunogenicity, and transmissibility were evaluated by further monitoring of morbidity, mortality, strain-specific P. haemolytica colonization rates, and humoral immune responses between groups over an 8-wk period. We plan to use the four vaccinated bighorns (2 primary vaccinates, 2 secondary vaccinates) from the foregoing pilot studies and a fifth unvaccinated male bighorn to evaluate vaccine efficacy. After challenge, all five bighorns wiU be housed together in a single 200 m2 isolation pen. Subject bighorns will be challenged at 10 wk after vaccination of primary vaccinates. For challenge, we will use about 1 x 107 CFU of serotype 2 P. haemolytica strain (WSU-1) sprayed intranasally.Vaccine efficacy will be evaluated by comparing morbidity, mortality, strain-specific P. haemolytica colonization rates, and humoral immune responses of vaccinated bigho_rns to the control and to results from a previous study (Kraabel et al. 1998) . Development of a modified-live Pasteurella trehalosi vaccine for bighorn sheep (McNeil, Lo, and Miller): We continued developing and evaluating a modified-live Pasteurella trehalosi serotype 10 vaccine strain for use in bighorn sheep. Highlights of ongoing laboratory technique development and evaluations are reported. Polymerase chain reaction: We used polymerase chain reaction (PCR) techniques to amplify a 122 base pair segment from the 5' end of the leukotoxin gene, as previously described; using the following primers: 5'-GAAITCCCITATGGGAACTAAAC-3' 5'-GTCTGGCCGGCITGGGITAA-3' �281 Conditions were optimized at 3.5 rnM [Mg++] using genomic DNA prepared by the PUREGENE® method as template. The thermal cycler was programmed for 94 C for 30 seconds, 59 C for 30 seconds, 72 C for 3 minutes and 30 seconds. lbis was repeated for 30 cycles. PCR was also utilized to try to increase the amount of ligated product. The following primers were used: 5'-GAATICCCTIATGGGAACTAAAC-3' 5'-CGAAACAATCTAGAGTIGCCAATC-3' Conditions were optimized at 3.5 rnM [Mg++] using gel purified ligated product as the template. The thermal cycler was programmed for 94 C for 30 seconds, 64 C for 30 seconds, 72 C for 3 minutes and 30 seconds. This was repeated for 30 cycles. The above primers (used to amplify the ligated product) were also used to amplify the ligated product once it had been inserted into the plasmid p2176, as a means of mapping and confirmation. All PCR products were run on 0. 75 % agarose gels (85 V for 23-45 minutes) and stained with ethidiwn bromide. Bands were visualized using ultra-violet (UV) light and the image captured by a GelDoc I 000 machine (Bio-Rad Laboratories, Mississauga, ON, Canada). PCR Product Purification: We purified PCR products to be sequenced or used for further manipulation using a QIAGEN QIAquick PCR purification kit®. Purified product was run on gels as before purification to ensure that the band in question was still visible. I In instances where extraneous bands were visible, gel extraction was used to isolate the desired product. A low melting point 1.0 % agarose gel was loaded with product and run at low voltage overnight (8-9 V, 1216 hours). The gel was then stained with ethidiwn bromide and visualized using UV light on a transillwninator. The desired band was cut out using a sterile scalpel blade and put into a tube. The tube was heated to melt the agar. lbis material was then purified in the same manner as the non-gel purified PCR products. Restriction Enzyme Digestion: Purified 122 hp fragment was digested using Nae! (Promega Corporation, Madison, WI, USA) with the recommended buffer for I hour at 37 C. The enzyme was then inactivated at 60 C for 15 minutes. This product was ligated to the 1767 hp fragment described previously. Restriction enzyme digestion was also used for mapping the ligated product. Digestions were performed using Nae!, Apa! (Promega Corporation, Madison, WI, USA) and Sad (New England Biolabs Ltd., Mississauga, ON, Canada). Appropriate buffers were used for each reaction which was carried out at 37 C for I hour. Reactions were terminated by heat inactivation at 60 C for 15 minutes. A sample of product was run on a 0.75% agarose gel, stained with ethidiwn bromide and photographed. Restriction enzyme digestion was also used for mapping the plasmid construct (containing the 1889 hp insert). Digestions were performed using Nae!, Apa!, Sac!, EcoRI, Xbal, San and C/al. Enzymes are from Amersham Pharmacia Biotech, Baie d'Urfe, Quebec, Canada, unless otherwise indicated previously. The commercially available plasmid pBluescriptll (Stratagene, Aurora, ON, Canada) was used to troubleshoot digestion problems. Blunt End Ligation: Equal volwnes of the fragments to be ligated were mixed with T4 DNA ligase buffer (Gibco, Burlington, ON, Canada) according to manufacturer's instructions. Ligase (10% final volwne) �282 was then added and mixed thoroughly. The reaction was left at 16 C O/N. The ligated product was then run on a gel for visualization as well as used as template in further PCR. The product was then gel purified as previously described and used as template in further PCR. Blunt end ligation was also used in the preparation of the construct. Different ratios ofvector:insert were used to maximize the opportunities for ligation. Conditions were the same as above. Construct: Plasmid p2176 was digested with EcoRl andXbal in a double digest for I hour at 37 C. The ligated 1889 hp product was also double digested with the same two enzymes to prepare for ligation with the plasmid. After the digestion, enzymes were heat-inactivated for 15 minutes at 65 C. The plasmid digest mixture was then treated with calf intestinal alkaline phosphatase to minimiz.e the chances of recirculariz.ation. The reactions were then purified and concentrated using GENECLEAN® (BiolOI-Inc, from BioCan Scientific). These concentrated products were used in a blunt end ligation as reported above. Following ligation, the plasmid was transformed into competent E. coli cells (HB IO I) as described below. Preparing Competent E.coli Cells by CaC/2: A single colony of E.coli HBIOI was used to inoculate 3 ml of LT broth and incubated at 37 C O/N with shaking. This culture was then subcultured 1/40 in LT broth and incubated with the same conditions for I hour and 15 minutes. This culture was then put on ice for 40 minutes. After centrifugation (5000 rpm) for 5 minutes, the pellet was resuspended in ½ volume of cold, sterile 50mM CaC12 . This mixture was left on ice for I hour. After centrifugation, the pellet was resuspended in 1/10 volume of cold, sterile 50mM CaC12• These cells were used after 24 hours at 4 C, at which time maximum competency for foreign DNA uptake is expected. Transformation: In small glass tubes, 0.2 ml of competent cell suspension was mixed with IO µl of potentially ligated insert-plasmid and put on ice for I hour. This was followed by a 2 minute heat-shock at 42 C. Each tube was then supplemented with 0.3 ml of LT broth and left to "recover" at 37 C for 15 minutes. These cells were then plated onto LT agar plates containing ampicillin. The plasmid p2 l 76 carries an ampicillin resistance gene. Negative controls containing no insert DNA (digested plasmid only) as well as control cells containing no foreign DNA were used. Plates were incubated O/N at 37 C. Visible colonies were re-plated and screened by plasmid preparation and restriction enzyme mapping. Plasmid Preparations: Plasmid preparations were made by growing the colony to be screened O/N in 20 mis of LT broth containing ampicillin in a 125 ml erlenmyer flask. To incr~e plasmid yield (p2176 is a low copy plasmid), spectinomycin was added (50 µg/ml final concentration). Plasmid was then purified using Qiagen's QuickSpin® columns. Plasmid was then run on agarose gels to look for a product of expected size. Those which were of the expected size were further examined using restriction enzyme mapping. RESULTS AND DISCUSSION Evaluation of a genetically-modified live Pasteurella haemolytica vaccine in bighorn sheep (Miller and Briggs): Of3 bighorns receiving about 7 x 108 colony forming units (CFUs) ofGMLPhV, I died about 3 days after vaccination and another showed signs of moderately severe respiratory disease from days 2-10 postinoculation (PQ. Although bacteriology results were somewhat confounded by the reported isolation of a hemolytic P. haemolytica strain from lung tissue, the GMLPhV strain was most likely responsible for �283 morbidity and mortality in vaccinated bighorns. The sick bighorn recovered without medical intervention, and both surviving primary vaccinates showed strong elevations in both agglutinating and leukotoxin neutralizing antibody titers in response to vaccination (Fig. I). Two previously unvaccinated bighorns housed with the 2 surviving vaccinates beginning at wk 2 (2 weeks Pl) seroconverted to P. haemolytica serotype 2 2-3 weeks later (Fig. I), suggesting colonization of these incontact animals by GMLPhV strain despite failure to actually isolate GMLPhV strain from either incontact bighorn. Agglutinating antibody responses appeared to be less that those stimulated by primary vaccination (Fig. IA), and leukotoxin neutralizing antibody responses in secondary vaccinates were negligible (Fig. 1B). Both in-contact bighorns remanied healthy through 7 weeks postexposure (= wk 9). Ability of GMLPhV to protect bighorns from pasteurellosis will be evaluated via experimental challenge beginning in early July. Further evaluation of this or a similar GMLPhV strain hinges on the vaccine's ability to protect bighorns from expermental challenge, as well as the need for further modifications to improve vaccine strain safety in bighorn sheep. Development of a modified-live Pasteurella haemo/ytica vaccine for bighorn sheep (McNeil, Lo, and Miller):' Failed attempts to insert the ligated 123 bp-1767 bp fragment into p2176 led to the discovery of an error that was easily rectified. A different short fragment ( 122 bp) was prepared in the same way and ligated to the 1767 bp incorporating a different restriction enzyme on the 5'-end. Ligation of the 122 bp fragment to the 1767 bp fragment appears to have been successful. The product is the appropriate size, and restriction enzyme mapping supported our conclusion of successful ligation. As before, attempts to use PCR to increase yields for cloning were disappointing, insofar as there were extraneous bands. However, gel purification was used to prepare a high concentration of ligated product to use in the subsequent insert-plasmid ligation. Weeks postinoa.llation Figure 1. Serological responses of captive bighorn sheep to GMLPhV. A Agglutination titers rose in both primary (solid lines) and secondary (dashed lines) vaccinates after direct or indirect vaccination. B. Leukotoxin neutralizing (LN) titers showed marked elevation in primaryvaccinates; in contrast, LN titer responses were relatively weak in secondary vaccinates. An unvaccinated control showed no change in titers. Different symbols represent different individual bighorn sheep. Several recipient E. coli colonies were recovered. A small nwnber were screened and carried a plasmid of the appropriate size (~7Kb). Restriction mapping has been encouraging. One cloned construct (022) shows the most favourable and consistent results. When plasmid from this �284 clone is used as template in a PCR (using the 5'-primer used to make 122 bp fragment and the 3'-primer made to use the 1767 bp fragment) a single, very clear band of appropriate size ( 1889 bp) is visualized. We were concerned about the failure of Nae I digestion to create a single linearized band of 7 Kb, as expected. However, NaeI also failed to properly digest pBluescriptll, a commercially available plasmid vector with a single NaeI site. We concluded that the problem lies with either the enzyme or the digestion technique rather than with the insert in p2 l 76. It is important to sort out the problem with the Nae I because that is the proposed site for cloning the green fluorescent protein (gfp) gene. Other workers at the University of Guelph have shown that gfp can be expressed in P. haemolyti(!a. Short-term plans for continuing this work include PCR to amplify the gfp gene with constructed blunt end restriction enzyme sites on each end. We hope to successfully clone this gene into the NaeI site on the plasmid carried in D22. We then will electroporate the plasmid into P. haemolytica TIO CEco. 100) and replace the existing lktA with our .dlktA-gfp. We anticipate having a vaccine strain available for testing in captive bighorn sheep sometime in the 00/01 segment. LITERATURE CITED Briggs, R. E., and F. M. Tatum. 1999. New mucosal vaccine in beef cattle imparts rapid resistance to pneumonic pasteurellosis after mass-medicating on feed. USDA, REE, ARS, NADC, Ames IA. Unpublished report presented at USAHA, San Diego, CA, October 1999, 4 pp. Foreyt, W.J. 1989. Fatal Pasteurella haemolytica pneumonia in bighorn sheep after direct contact with clinically normal domestic sheep. Am. J. Vet. Res. 50:341-343. Foreyt, W.J. and R. M. Silflow. 1996. 1Attempted protection of bighorn sheep (Ovis canadensis) from pneumonia using a nonlethal cytotoxic strain of Pasteurella haemolytica, biotype A, serotype 11. J. Wildl. Dis. 32:315-321. Foreyt, W.J., K.P. Snipes, and R. W. Kasten. 1994. Fatal pneumonia following inoculation of healthy bighorn sheep withPasteurella haemolytica from healthy domestic sheep. J. Wildl. Dis. 30:137-145. Green, A. L., N. M. DuTeau, M. W. Miller, J. Triantis, and M. D. Salman. 1999. Polymerase chain reaction techniques for differentiating cytotoxic and noncytotoxic Pasteurella trehalosi from Rocky Mountain bighorn sheep. American Journal of Veterinary Research 60:583-588. Kraabel, B.J., and M.W. Miller. 1997. Effect of simulated stress on susceptibility of bighorn sheep neutrophils to Pasteurella haemolytica leukotoxin. J. Wildl. Dis. 33:558-566. Kraabel, B.J., M.W. Miller, J.A. Conlon, and H. McNeil. 1998. Evaluation ofa multivalent Pasteurella haemolytica vaccine in bighorn sheep: Protection from experimental challenge. J. Wildl. Dis. 34:325-333. Miller, M. W. 2000. Pasteurellosis. In Infectious Diseases of Wild Mammals, 3rd edition, E. S. Williams and I. K. Barker, eds. Iowa State University Press, Ames, Iowa, in press. Miller, M. W., J. A. Conlon, H.J. McNeil, J.M. Bulgin, and A. C. S. Ward. 1997. Evaluation of a multivalent Pasteurella haemolytica vaccine in bighorn sheep: safety and serologic responses. J. Wildl. Dis. 33: 738-748. Silflow, RM., Foreyt, W.J., and R.W. Leid. 1993. Pasteurella haemolytica cytotoxin dependant killing of neutrophils from bighorn and domestic sheep. J. Wildl. Dis. 29:30-35. Prepared by _ _ _ _ _ _ _ _ _ _ __ Michael W. Miller Wildlife Research Veterinarian � https://cpw.cvlcollections.org/files/original/a437c3334ab93087aef88d9389bef4d1.pdf 85cff8767672b5d1c03cbca0fbb0631b PDF Text Text 241 Colorado Division of Wildlife Wildlife Research Report July 1998 JOB PROGRESS REPORT state of Colorado cost center 3430 Project No. W-153-R-11 Mammals Program 3004 Work Package No. Task No. 1 Period covered: Author: &2 : Management of other ungulates Pronghorn Data Analysis_and Reporting July 1, 1997 - June 30, 1998 T.M. Pojar Abstract A herd structure estimate was made on the Middle Park pronghorn population during August, 1997 amd the total winter population estimate was made during January 1998. This information, along with like information since 1986 was incorporated into a spreadsheet population model. This model was provided to management personnel for.projecting effects of various management options. It was also used in public meetings for exploring public opinion of the proposed options. A manuscript was prepared summarizing the results of experiments in pronghorn inventory methods. It was submitted to the Journal of Wildlife Management and rejected on the basis that the topic addressed a relatively narrow issue. The manuscript will be modified and submitted either to the Wildlife Society Bulletin or the African Journal of Wildlife Management. The abstract is included in the results section. �243 PRONGHORN DATA ANALYSIS AND REPORTING Thomas M. Pojar P,H,QBJECTIYES Task 1 Test for density dependence in Middle Park population parameters and summarize the findings in manuscript format suitable for submission to a professional wildlife management or ecological journal. Task 2 Analyze and summarize the results of experiments in pronghorn inventory methods and report the findings in manuscript format suitable for submission to a professional wildlife management journal. SEGMENT OBJECTIVES Task 1 Test for density dependence in Middle Park population parameters and summarize the findings in manuscript format suitable for submission to a professional wildlife management or ecological journal. Task 2 Analyze and summarize the results of experiments in pronghorn inventory methods and report the findings in manuscript format suitable for submission RESULTS Task 1 Key population information in the form of herd structure estimates and a count of the wintering population was collected. This was included in the population data set that began in 1986 and used to build a spreadsheet population model. This model was used to project population responses to proposed management options and was provided to managment personnel. The overall objective of Task 1 was not fulfilled during this segment. Task 2 A manuscript was prepared that analyzed and summarized the results of experiment in pronghorn invenorty methods. It was rejected by the Journal of Wildlife Management as having too narrow a focus. The manuscript will be modified and sumbitted either to the Wildlife Society Bulletin or the African Journal of Wildlife. The following is the abstract for this article. PRONGBORH DENSITY ESTIMATES: HELICOPTER QUADRAT SURVEYS COMPARISON OF FIXED-WING LIRE TRANSECT ARD THOMAS M. POJAR1 , Colorado Division of Wildlife, 317 w. Prospect Road, Fort Collins, CO 80526, USA DAVID c. BOWDEN, Department of Statistics, Colorado State University, Fort Collins, CO 80523, USA JEFF D. MADISON, Colorado Division of Wildlife, P.O. Box 1181, Meeker, CO 81641, USA Estimates of pronghorn (Antilocapra americana) density in northwest Colorado sagebrush (Artemisia spp.) steppe habitat were compared using fixedwing line transect and helicopter quadrat surveys. Fixed-wing line transects offer wildlife managers a survey method that is both economical and statistically valid but this technique has not been evaluated with a standard or known density for pronghorn. We used the helicopter quadrat survey method as the standard and compare density estimates to fixed-wing line transect Abstract: �244 surveys. Fixed-wing line transect estimates were always less than quadrat estimates (t=0.682, P = 0.050) and were 83%, 69%, and 83% of the quadrat estimates respectively, for 3 years data. In addition, we analyzed the first and second line transect intervals as narrow strips of 50 m and 100 m, respectively. The narrow strips were not different from either the quadrats or lines (P > 0.070) and may offer managers a method that combines the economy of line transects with the analysis simplicity of quadrat data. We conclude that line transect density estimates should be adjusted upward by considering them approximately 0.78 (SE= 6.32) of quadrat estimates. At the density of pronghorn encountered in this study (approximately 5/km'), the cost of fixedwing line transect sampling for large scale management application is about 510% of the cost of helicopter quadrats for similar precision. Prepared Wildlife Researcher � https://cpw.cvlcollections.org/files/original/385b1830d75fd77330072f041e83d500.pdf 7bc75debf6db24d3492aef8e4a87d438 PDF Text Text 295 Colorado Division of Wildlife Wildlife Research Report July 1999 JOB FINAL REPORT State of _ _ _ _ _ _C=o=l=o=rad=o_ _ _ __ Project No. ----~W'-'---......1....5___ 3-___R___--=12"---_ __ Work Plan No. 3004 Job No. - - - - - - - - ' 4 ' - - - - - - - ----~~----- Mammals Research Mountain Goat Investigations Mountain goat numbers, distribution, and dispersal in the northern Collegiate range. Period Covered: July 1, 1998 - June 30, 1999 Author: D. F. Reed ABSTRACT A manuscript titled "Mark-resight population estimates of mountain goats in Colorado" (Reed and Vayhinger [in preparation]) was prepared. Additionally, a manuscript titled "A conceptual interference competition model for introduced mountain goats" was prepared and submitted to the Journal of Wildlife Management (Reed [in review]) and collaboration with Natural Resources Ecology Laboratory (NREL) occurred to extent and refine some of the Mt Evans data. �296 �297 MOUNTAIN GOAT NUMBERS, DISTRJBUTION, AND DISPERSAL IN THE NORTHERN COLLEGIATE RANGE Dale F. Reed P.N. OBJECTIVE To improve estimates of mountain goat populations by mark-resight methodology, to determine distribution, and to estimate dispersal rates in an increasing mountain goat population. SEGMENT OBJECTIVE I. Analyze data and prepare manuscripts. STUDY AREA The study area is described in the Program Narrative (Reed 1995). ME1HODS AND MATERIALS The methods are outlined in the Program Narrative and 1995 and 1996 progress reports (Appendix A in Reed 1995, Reed 1996). RESULTS Results have been reported in Reed ( 1996), Reed ( 1997), and in a manuscript titled "Mark-resight population estimates of mountain goats in Colorado" (Reed and Vayhinger [in preparation]. Additionally, a manuscript titled "A conceptual interference competition model for introduced mountain goats" was prepared and submitted to the Journal of Wildlife Management (Reed [in review]) and collaboration with NREL (ref Rocky Mountain National. Park contract) occurred to extent and refine some of the Mt Evans data by evaluating spatial distribution, habitat segregation, and implications of competitive displacement. LITERATURE CITED Reed, D. F. 1995. Mountain goat numbers, distribution, and dispersal in the northern Collegiate range. Colo. Div. Wildl. Res. Rep. July, 223-234pp. Reed, D. F. 1996. Mountain goat numbers, distribution, and dispersal in the northern Collegiate range. Colo. Div. Wildl. Res. Rep. July, 255-263pp. ; .· Reed, D. F. 1997. Mountain goat numbers, distribution, and dispersal in the northern Collegiate range. Colo. Div. Wildl. Res. Rep. July, 165-167pp. Prepared by _ _ _ _ _ __ DaleF. Reed � Dublin Core The Dublin Core metadata element set is common to all Omeka records, including items, files, and collections. For more information see, http://dublincore.org/documents/dces/. Title A name given to the resource Mammals research pdfs Description An account of the resource Pdfs of mammals research projects by title . Pdfs are linked from this page. Includes federal aid progress reports and non federal aid progress reports. If reports cover more than one year they are combined into one pdf. Dublin Core The Dublin Core metadata element set is common to all Omeka records, including items, files, and collections. For more information see, http://dublincore.org/documents/dces/. Title A name given to the resource Mammals Research Projects pdfs (1991-2000) Rights Information about rights held in and over the resource <a href="http://rightsstatements.org/vocab/NoC-NC/1.0/">No Copyright - Non-Commercial Use Only</a> https://cpw.cvlcollections.org/files/original/e42a9d55906409150beb289db16b230d.pdf f40f1b1bb64e3108228d7a501a3e6a1a PDF Text Text Colorado Division of Wildlife ._,, July 2006 - June 2007 WILDLIFE RESEARCH REPORT State of_ _ _ _ _ _=C=o..,,,lo=r=ad=o"---~---: ""'DC!c.iv!..a!i~s1!-"·o""n'--'o,. , f__,WC!..!.!il., ,_... dli£..,,e'--_ _ _ _ _ _ _ __ Cost Center 3430 : ""'M""a..,mm=""-a:::lsc.. R""e""s:.:::ce::ears.=c""h_ _ __ __ _ _ __ Work Package 3001 : ~D'-=e~er:.....C=o=ns,,__,e=rv..:...;a=t=io=n<----------Task No. 6 : Population Performance of Piceance Basin Mule : Deer in Response to Natural Gas Resource : Extraction and Mitigation Efforts to Address : Human Activity and Habitat Degradation Federal Aid Project: __W-'-'---~18=-=5'-'-R~----Period Covered: July 1, 2006 - June 30, 2007 Authors: C. R. Anderson, and D. J. Freddy Personnel: M. Alldredge, E. Bergman, C. Bishop, R. Kahn, P. Lukacs, T. Remington, M. Schuette; G. White, Colorado State University; H. Sawyer, Western Ecosystems Technology, Inc. ALI information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT A researcher FTE vacancy was filled with a newly hired person in December 2006 who became the project leader for this project. No preliminary field work could be completed in winter-spring 2007 as originally planned for FY2006-07 but assessments of potential study areas, resource inventory maps, and tentative study plan outlines were completed by June 2007. As such, field work for this project will begin winter 2007 and be centered in the Piceance Basin area of northwestern Colorado which is currently undergoing intensive natural gas development in one of the most extensive and important mule deer winter and transition range areas within the state. Our approach will be to experimentally evaluate habitat -treatments that may rehabilitate the landscape to benefit mule deer and to evaluate human-activity management alternatives to reduce the disturbance impacts on mule deer. This project will require a longterm commitment ofat least 10-years from private industry, the BLM, and the CDOW to assess if sustainable mule deer populations can persist within a highly disturbed landscape following implementation of beneficial habitat treatments and development practices. 103 �WILDLIFE RESEARCH REPORT POPULATION PERFORMANCE OF PICEANCE BASIN MULE DEER IN RESPONSE TO NATURAL GAS RESOURCE EXTRACTION AND MITIGATION EFFORTS TO ADDRESS HUMAN ACTIVITY AND HABITAT DEGREDATION CHARLES R. ANDERSON AND DAVID J. FREDDY P. N. OBJECTIVE To develop approaches to provide for energy extraction in a manner that maintains viable mule deer populations for future recreational and ecological purposes. SEGMENT OBJECTIVES I. Consult with regional personnel to select potential study sites for addressing habitat mitigation and energy development practices that benefit mule deer. 2. Plot historic and current energy development activities to assess potential treatment and control sites for experimental evaluation. 3. Develop draft study proposals for peer review and funiling solicitation. INTRODUCTION Extraction of natural gas from areas throughout western Colorado bas raised concerns among many public stakeholders and the Colorado Division of Wildlife that the cumulative impacts associated with this intense industrialization will dramatically and negatively affect the wildlife resources of the region. Concern is especially high for mule deer due to their recreational and economic importance as a principal game species and their ecological importance as one of the primary herbivores of the Colorado Plateau Ecoregion. Research evaluating the most effective strategies for minimizing and mitigating these activities will greatly enhance future management efforts to sustain mule deer populations for future recreational and ecological values. Our primary goal of this study is to develop approaches to provide for energy extraction in a manner that maintains viable mule deer populations for future recreational and ecological purposes. This may be accomplished by restoring or enhancing habitat conditions on or adjacent to disturbed sites and by modifying development practices. Due to the extensive energy development that is projected to occur over the next 20 years throughout much of the mule deer winter range in the northern Rocky Mountains of the western US, innovative approaches to energy development and mitigation methods are essential to sustain viable mule deer populations in the region. Impacts from development and conversely success of mitigation efforts are often assumed but rarely demonstrated, and these assumptions can only be confirmed by application of well designed research efforts conducted over sufficiently long time periods to measure responses. This project proposes to identify habitat mitigation and energy development approaches that sustain mule deer survival and recruitment during and after habitat disturbance from development activities. This effott will require coordination and cooperation between Colorado Division of Wildlife and the major energy companies. We anticipate this pa1tnersbip will benefit mule deer populations and foster the evolution of wildlife management and energy development practices that are compatible with other wildlife and human values associated with maintaining functional ecosystems over the long term. 104 �STUDY AREA The proposed study sites represent 6 segments of mule deer winter range in the Piceance Basin, southwest of Meeker, Colorado (Figure 1), and the primary energy companies developing these areas include Encana and Exxon-Mobile (Figure 2). Because of the varying levels of development and deer densities relative to differing winter population segments in the Piceance Basin, different experimental units (i.e., mule deer winter ranges) are uniquely suited for addressing different questions. Experimental designs monitoring mule deer responses to treatment (e.g., habitat mitigation) and control areas are necessary to differentiate cause-effect relationships from development versus environmental factors. Suitable control areas require that little or no previous development has occurred and that no development occurs during the experimental time frame. Ideally, both temporal and spatial control areas would be monitored to make valid comparisons to developed and subsequently mitigated sites; temporal controls provide measures of natural variability in mule deer population parameters over time and spatial controls provide measures of variability due to differences in geography. Once spatial and temporal variation is accounted for, inferences can be made relative to development disturbance or mitigation effects on mule deer. The North Ridge, Story/Willow Creek, and Yellow Creek deer population segment areas (Figure 1) currently exhibit little to no development, but it is currently unknown whether or not these areas will be developed in the future; there is potential for future oil shale development in the Story/Willow Creek and Yellow Creek deer areas. North Ridge appears least likely to be developed because it is outside of the current oil shale lease area and only a few wells have historically been drilled on or adjacent to the area, whereas the same cannot be said of the Story/Sprague or Yellow Creek areas. Thus, North ~dge would appear best suited as a temporal control site for comparison to other developed winter ranges within the Piceance Basin and may also serve as a geographic control for the Crooked Wash deer population segment located immediately north and adjacent to the Piceance Basin. The Story/Willow Creek and Yellow Creek deer may provide spatial controls for the Magnolia and Ryan Gulch deer population segments, respectively, but future development potential in these areas is unknown. If these areas become devel~ped in the future (either for oil shale or natural gas), they would provide BACI (Before-AfterControl-Impact) type comparisons strengthening our inference of development impacts on mule deer population performance. Magnolia, Crooked Wash, and Ryan Gulch deer areas have historically received relatively high development activity and currently exhibit moderate-high development, and appear likely to be developed extensively in the future based on the gas development layers currently available (Colorado Oil and Gas Conservation Commission). Pretreatment data in these areas will be represented by parameters associated with developed sites and the measured response will be in the form of habitat treatments and/or differing development practices, which will be measured in comparison to the control sites. We propose including 3 control sites (1 temporal/spatial control and 2 spatial controls) and 3 treatment sites to investigate mule deer response to habitat and/or development treatments (e.g., directional versus non-directional drilling, piping versus trucking condensate, etc.) across a range of deer densities (Table 1). We would strive to split high intensity extraction study sites into 2 halves with one half serving as the 'control' [standard development] and one half serving as the 'treatment' [improved development approach or improved habitat]. The above scenario addresses the potential for establishing control and treatment sites for evaluating mule deer population response to habitat treatments and/or development treatments, and may allow larger scale mule deer responses from energy development to be addressed by comparing control site parameters to developed site parameters; smaller scale inference would require collection of pretreatment data at developed sites (e.g., similar to mitigation treatments in the proposed design) and may not be possible unless the Yellow Creek or Story/Willow Creek areas are developed in the future. Modified versions of the proposed design could be implemented depending on 105 �the level offunding available and the degree to which industry willing to collaborate with this effort. We consider 3 study sites, likely North Ridge, Magnolia, and Crooked Wash, as the minimum number of study sites necessary to adequately address the objectives of this project; the additional proposed study areas will allow increased flexibility in the questions that are addressed and increase our inference relative to mule deer responses to habitat treatments and modifications of development practices. Furthermore, if we are not able to evaluate potential mitigating industrial operation and/or habitat improvements, this study would likely only have the potential to document negative impacts of intense energy extraction practices on mule deer. V RESPONSE VARIABLES To allow for competing hypotheses in regards to potential development and mitigation effects, 4 primary response variables will be measured including (1) overwinter fawn survival, (2) deer density, (3) habitat use patterns, and (4) adult female body condition. (1) To determine if mitigation and/or development treatments elicit a chronic survival response with a long-term population level effect, we will measure over-winter fawn survival in all experimental units. Based on past research (White and Bartmann 1998), treatment effects of 15% change in survival appear biologically significant. (2) To determine if habitat treatments or development practices elicit a brief survival response with a long-term population level effect, we will estimate deer density to determine if there is a difference in carrying capacity between treatment and control experimental units. Because mule deer may respond to development or mitigation at variable rates, we may not be able to detect differences in fawn survival, but estimating deer density will still allow us to determine if development or mitigation efforts have a population level effect. (3) To determine if habitat treatments or development practices elicit a shift in habitat use patterns, we will examine changes in Resource Selection Probability Functions (RSPF; Sawyer et al. 2006) pre- and post-habitat treatments, between areas exhibiting development practices, and compare RSPFs between developed and non-developed sites. We will infer population level impacts if fawn survival and/or deer densities differ relative to changes or differences in habitat use patterns. (4) To determine if adult female mule deer respond positively to habitat treatments and/or· changes in development practices, percent body fat and loin depth will be measured annually during late winter (Bishop et al. 2005, Bergman et al. 2005). We would expect a relatively rapid response in body condition following habitat or development treatments, indicating that treatments are having the intended positive effect. PROJECT EXPENSES Estimating fawn survival, deer density, deer behavioral responses, female body condition, and implementing small scale habitat improvements are costly endeavors involving the purchase of numerous standard VHF radio-collars, specialized GPS radio-collars, helicopter flight hours for deer capture/collaring and aerial surveys, machinery to physically alter the habitat, and personnel to adequately perform day-to-day data collection. If large scale habitat treatments are needed or desired, funding in addition to the estimates below will be required as habitat treatments cost $300 to $1,000/acre depending on the most appropriate treatment for a locale. Minimum cost estimates to design, implement, and evaluate responses of mule deer to habitat mitigation options range form $580,500 to $1,161,00 (most preferred design) per year depending on project design (Table 2). 106 u �LITERATURE CITED BISHOP, C. J., G. C. WHITE, D. J. FREDDY, and B. E. WATKINS. 2005. Effect of nutrition on mule deer recruitment and survival rates. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins. USA. BERGMAN, E. J., C. J. BISHOP, D. J. FREDDY, and G. C. WHITE. 2005. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Study Plan, Colorado Division of Wildlife, Fort Collins, USA. SAWYER, H., R. M. NIELSON, F. LINDZEY, and L. L. MCDONALD. 2006. Winter habitat selection of mule deer before and during development of a natural gas field. Journal of Wildlife Management 70:396-403. WHITE, G. C., and R. M. BARTMANN. 1998. Effect of density reduction on overwinter survival of freeranging mule deer fawns. Journal of Wildlife Management 62:214-225. Prepared by _ _ _ _ _ _ _ _ _ _ _ _ __ Charles R. Anderson, Wildlife Researcher Table 1. Relative density of natural gas wells and mule deer and experimental designation for potential study sites iri the Piceance Basin, Colorado, for addressing mule deer response to natural gas development practices and habitat mitigation. Relative density Experimental Study area Inactive wells Active wells Mule deer designation North Ridge Very low None High Temporal/spatial control Crooked Wash High High High Treatment Story/Willow Creek Low Low Moderate Spatial control Magnolia High High Moderate Treatment Yellow Creek Moderate Low Low Spatial control Ryan Gulch High Moderate Low Treatment ~ 107 �Table 2. Estimated costs for CDOW to conduct desired mule deer research in the Piceance Basin to assess impacts of natural gas extraction on mule deer and evaluate approaches to mitigate habitat impacts. Project should be conducted for 10 years to allow for adequate time to measure biological responses, 2008-2018. Cost Estimates Per Year Per Study Site (2008 dollars) Piceance Basin Mule Deer Research Telemetry collars & equipment $70,000 Helicopter Deer Capture & Surveys $68,500 Other Field Operations & Equipment $15,000 12 months TFTE Personnel {Tech I) $30,000 Vehicle Yearly Lease Plus Mileage (4x4 PU, & 45,000miles) Total cost Per Study Site/Yr Minimum Study One Control Site & Two Treatment Sites Acceptable Two Control Sites & Two Treatment Sites Best Three Control Sites & Three Treatment Sites Cost Per Year (2008 dollars) Cost Per Year (2008 dollars) Cost Per Year (2008 dollars) $580,500 $774,00 $1,161,000 $20,000 $193,500 108 V �Figure l. Proposed mule deer study sites relative to natural gas development in the Piceance Basin, Colorado, July 2007. Piceance Basin Research Areas + Inactive Wells + Drilling Permits • A.:ti,·c \\-d i~ ® ln.:acti,·t \\',:lb D Con1n1I Study - _ .-\n!:a Pmr,rnt<d Smdr .-\n::t.~ 0 ,1 & G,s FielJs C )Oil & Gas &sin, ---===""'""'"' 10 109 �Figure 2. Proposed mule deer study sites relative to the primary energy companies controlling natural gas leases in the Piccance Basin, Colorado, July 2007. Piceance Basin Research Areas 1111 C11,.•s:ipc:ikc Lc:isl.' 1111 Chc\Ton Lca.~e ~ Enc.inn L:nsc 1111 EXXClll 1..i:asc 1111 ~loh1h: lc:alic ~ \\'illi;uns Lc:asc Do,, & °'" Bosms 110 �-. ..J -- -- ...,,I --- ----- _., --- ..,,, . ..,I w -. _'J --_,_, -- Colorado Division of Wildlife July 2007 - June 2008 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Federal Aid Proj ect No. Division of Wildli fe Mammals Research Deer Conservation Population Pe1formance o f Piccance Basin Mule Deer in Response to N atural Gas Resource Extraction and M itigation Efforts to Address 1-luman Activity and Habitat Degradation - Stage 1, Objective 5; Patterns o f Mule Deer Distribution and Movements Colorado 3430 300 1 6 W-1 85-R Pe riod Covered: Jul y I , 2007 - June 30, 2008 Authors: C. R. Anderson and D. J. Freddy Personnel : J. Brodetick. B. de Vergie, D. Finley, L. Gcpfc.rt, C. Harty, K. Kaai, L. Kelly, S. Lockwood, R. Velarde, C DOW; R. Swisher, Quicksil ver Air, Inc. Project support received from Fede ra l Aid in Wildlife Restoration, Colorado M ule Deer Association, Colorado Oi l and Gas Conservation Commission, Williams Production LMT Co., EnCana Corp .. and Shell Petrole um. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR Q UOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. A BSTRACT We propose to experimenta lly eva luate habitat treatments that may rehabilitate the landscape to bene fi t mule deer (Odocoi!eus he111ion11s) and to eva luate human-acti vity management alternati ves to reduce the d isturbance of energy development impacts on mule deer. The Piccancc Basin o f northwestern Colorado was selected as the proj ect area due to ongoing natural gas development in one of the most extensive and important mule deer winter and transition range areas within the state. Assessments o f potential study areas, resource inventory maps, and tentati ve study plan outlines were presented to potent ial agency and industry cooperators. Suffic ient fund ing was secured to initiate a pilot study allowing refi nement of study area selection based on distribution of GPS collared deer, address logistics of deer captures and collaring efforts, and begin addressing one of the six proposed objecti ves by monjtoring deer movements from G PS localions in 5 study areas representing varying levels of e nergy development. We attached GPS collars collecting 5 fixes/day lo 75 ad ult female mule deer ( 15/study area) in January, 2008 to document deer movements and habitat use patterns among 5 deer wi nter ranges exposed to varyi ng levels of energy development. Over-winter survival of adult females was 90% (64 of 7 1) and typical for adult female mule deer in the western US. Data analyses of mule deer habitat use patterns will begin once GPS collars arc recovered in February, 2009. These data w ill provide deer beha vior information under ex,isting cond itions and serve as pre-treatment comparisons 10 future 63 �"""" i,,,,,.,I conditions fo llowing habitat treatments and/or improved development practices. Additional funding has become avai lable to initiate the full study proposal (see Appendix [) beginning November 2008, which will provide for evaluation of changes in body condition, fawn sw·vival, and deer densities relative to improved habitat treatments and energy development practices. This project will require additional funding commitments and cooperative agreements beyond spring 20 IOfrom private industry, the BLM, and the CDOW to assess if sustainable mule deer populations can persist within a highly disturbed landscape following implementation of beneficial habitat treatments and development practices. ...... """ \.iii """" "-iii ....., - ._./ \-iii i._..J \..ii 'l....il ....,,; ......, ~ ~ ..., """"" ""-' \=ii 64 ""'"' "--ii �WILDLIFE RESEARCH REPORT POPULATION PERFORMANCE OF PICEANCE BASIN MULE DEER IN RESPONSE TO NATURAL GAS RESOURCE EXTRACTION AND MITIGATION EFFORTS TO ADDRESS HUMAN ACTIVITY AND HABITAT DEGRADATION ...., STAGE I, OBJECTIVE 5: PATTERNS OF MULE DEER DISTRIBUTION & MOVEMENTS ..,.; CHARLES R. ANDERSON, JR. AND DAVID J. FREDDY ...., ..... -- P. N. OBJECTIVES l. To determine experimentally whether enhancing mule deer habitat conditions on winter and/or transition range elicits behavioral responses, improves body condition, increases overwinter fawn survival, or ultimately, population density on mule deer winter ranges exposed to extensive energy development. 2. To determine experimentally to what extent modification of energy development practices enhance habitat selection, body condition, over-winter fawn survival, and winter range mule deer densities. -.I SEGMENT OBJECTIVES 3. Assess the logistics capturing and collaring mule deer via helicopter net-gunning in 5 winter herd segments of the Piceance Basin, Colorado. 4. Improve delineation and identify degree of separation of winter range study sites based on deer distribution and movements from GPS collars collecting 5 fixes/day. 5. Monitor survival of adult female mule deer by daily ground tracking and bi-weekly aerial tracking. 6. Summarize data and present information in an annual Job Progress Report. INTRODUCTION .... .... .... Anderson and Freddy (2007) in their long-term research proposal identified 6 primary study objectives to assess measures to offset impacts of energy extraction on mule deer population performance . Much of the rationale for conducting the long-term research is presented in Appendix I. However, this progress report, beginning as of January 2008, focuses only on Objective 5 of the research proposal (Appendix I): monitoring distribution, movements, and habitat selection patterns of adult female mule deer on 5 potential segments of winter range in relation to varying levels of natural gas development, experimental modifications in energy developmental practices, and potential habitat improvement treatments. Long-term funding and support had not been secured to simultaneously address all 6 proposed study objectives on 5 potential winter range segments, but preliminary funding and support had been established to begin to address mule deer movement patterns relative to current natural gas development activities in the Piceance Basin. This initial effort during FY07-08 provided key information to I) document movement patterns and degree of spatial separation of deer among potential experimental control and treatment sites, 2) help refine study area boundaries, 3) begin documenting deer spatial use in proposed experimental control and treatment areas prior to implementing habitat or development improvements, and 4) provide an assessment of deer capture logistics and operational success of improved versions of GPS and VHF radio-telemetry collars. Monitoring spatial use patterns of 65 �deer is planned for at least 5 years as part of the forthcoming major study so that this first year of data acquisition establishes the foundation for long-term data acquisition process. Once longer term financial and administrative commitments have been established, we will incorporate the additional objectives into a revised study plan to achieve our overall goal of developing approaches to provide for energy extraction in a manner that maintains viable mule deer populations for future recreational and ecological purposes. We recently acquired the necessary funding to allow for the complete study proposal to be initiated by fall 2008 and continue through spring 20 I 0. STUDY AREA The Piceance Basin in northwest Colorado was selected as the project area due to its ecological importance as one of the largest migratory mule deer populations in North America and also exhibits one of the highest natural gas reserves in North America (Fig. l ). Historically, mule deer numbers on winter range were estimated between 15,000-22,000 (Bartmann 1975), and the current number of well pads (Appendix I: Fig. I) and projected number of gas wells in the Piceance Basin over the next 20 years is about 400 and 15,000, respectively. Mule deer winter range in the Piceance Basin is predominantly characterized as a topographically diverse pinion pine (Pinus edulis)~Utah juniper (Juniperus osteosperma; pinion-juniper) shrubland complex ranging from 1675 m to 2285 m in elevation (Bartmann and Steinert 1981 ). Pinion-juniper are the dominant overstory species and major shrub species include Utah serviceberry (Amelanchier utahensis), mountain mahogany (Cercocarpus montanus), bitterbrush (Purshia tridentata), big sagebrush (Artemisia tridentata), Gamble's oak (Quercus gambelii), mountain snowberry Symphoricarpos oreophilus), and rabbitbrush (Crysothamnus spp.; Bartmann et al. 1992). The Piceance Basin is segmented by numerous drainages characterized by stands of big sagebrush, saltbush (Atriplex spp.), and black greasewood (Sarcobatus vermiculatus), with the majority of the primary drainages having been converted to mixed-grass hay fields. Grasses and forbs common to the area consist ofwheatgrass (Agropyron spp.), blue grama (Boute/oua gracilis), needle and thread (Stipa comata), Indian rice grass (Oryzopsis hymenoides), arrowleafbalsamroot (Balsamorhiza sagittata), broom snakeweed (Gutierrezia sarothreae), pinnate tansymustard (Descurainia pinnata), milkvetch (Astragalus spp.), Lewis flax (Unum lewisii), evening primrose (Oenothera spp.), skyrocket gilia (Gilia aggregata), buckwheat (Erigonum spp.), Indian paintbrush (Castilleja spp.), and penstemon (Penstemon spp.; Gibbs 1978). The climate of the Piceance Basin is characterized by warm dry summers and cold winters with most of the annual moisture coming from spring snow melt. In our initial proposal, we outlined 6 potential study sites exhibiting varying winter deer densities and varying levels of energy development activity to provide control and treatment experimental units for evaluating improved habitat and development treatments (Appendix I: Table I, Fig. 2). Ultimately, I of the 6 proposed study sites was omitted partly due to funding limitations and ultimately because the omitted area (Crooked Wash) offered limited opportunity to examine habitat improvements due to dry moisture conditions inhibiting success of habitat treatments and future energy development in the area appeared unlikely due to extensive previous development precluding evaluations of improved development practices. The remaining 5 areas were maintained and North Ridge will serve as a temporal control area offering evaluations of annual variation in parameter estimates due to non-development factors from an undeveloped area, and Story/Sprague Gulch (formerly referred to as Story/Willow Creek) and Yellow Creek will serve as spatial control areas to the 2 treatment areas (Magnolia and Ryan Gulch, respectively), providing spatial comparisons from geographically and vegetatively similar areas exposed to minor levels of energy development compared to extensively developed areas receiving improved habitat and/or development treatments. Because the progression and extent of energy development in the future is currently unknown (to CDOW, at least), North Ridge may also serve as a spatial control area to Magnolia or possibly Ryan Gulch should the Story/Sprague Gulch or Yellow Creek study areas become developed in the future. 66 �METHODS -...I --..I -...., - -..I 1,-/ .... Tasks addressed this fiscal year included deer capture and collaring efforts, monitoring adult female mule deer survival, and downloading and plotting GPS location data monthly from a segment of the sample fitted with downloadable GPS collars (24 of 75 deer total). We employed helicopter netgunning techniques (Barrett et al. 1982, van Reenen 1982) to capture 15 adult female mule deer in each of 5 study areas (75 deer total). Once netted, deer were hobbled, blind folded, fitted with GPS collars, and released. Five deer in 4 of the 5 study areas and 4 deer in the Yellow Creek study area were fitted with remotely downloadable GPS collars (GPS-4400S; Lotek Wireless, Newmarket, Ontario, Canada) and the remaining deer in each area were fitted with store-on-board GPS collars (G2 I l OB; Advanced Telemetry Systems, Isanti, MN, USA). To insure GPS fixes for at least l year, both collar types were programmed to attempt a· fix every 5 hours and the fix schedule for store-on-board collars was reduced to attempt a fix every 23 hours July-October. Mule deer mortality monitoring consisted of ground tracking deer daily and aerial monitoring deer approximately every 2 weeks from fixed-wing aircraft. Once a mortality signal was detected, deer were located and necropsied to attempt determination of cause of death. We collected GPS locations from the 24 downloadable collars monthly via ground tracking, if possible, or using fixedwing aircraft. RES ULTS AND DISSCUSSION Deer Captures We captured and GPS collared 4 yearling and 71 adult female mule deer ( 15 deer/study area) from January 10-12, 2008 (Fig. 2). No significant injuries were noted during captures. In planning future capture efforts for adult female mule deer, we will anticipate about 25 captures/day/helicopter. Deer Mortalities We identified l yearling and 7 adult female mule deer mortalities from January-June, 2008 (Table l ). Although winter severity was relatively high this past winter, adult female survival (90%, n = 71) was typical of mule deer populations under normal winter conditions in the western US (Unsworth et al. 1999). Cause of mortality was determined for 4 of the 8 mortalities documented and varied between coyote predation, malnutrition, and vehicle collision (Table l ). Although the other 4 mortalities were undetermined due to timing of carcass inspection, winter severity was likely a factor given 3 of the 4 mortalities occurred during late May (Table I). GPS Data Collection and Deer Distribution GPS data downloads and collars retrieved from mortalities suggested collars were generally functioning as expected, but a few issues were noted that may warrant future attention. GPS location acquisition rates were high (>90%) for all collars except I where intermittent acquisition failures were common (Lotek GPS_4400S; 58% acquisition rate). The single collar exhibiting a low acquisition rate is acceptable relative to the 31 other collars exhibiting high acquisition rates, but the malfunctioning collar will be returned for evaluation once retrieved to potentially enhance collar performance in the future. We noted that false mortality signals (a mortality signal for an active deer) occurred for short durations ( I to a few days) on several occasions during winter monitoring, and we will increase the inactive time period to activate the mortality switch from 4 to 8 hours for future collar orders to try to address this problem. In addition, consultation with collar manufacturers will be conducted in an attempt to address the problem of inactive mortality signals occurring while deer are active. Another, more significant problem was noted when we unsuccessfully attempted to remotely detonate drop-off mechanisms on a few occasions (Lotek collars). The 20 Lotek collars currently in use will require remote detonation for retrieval in February, 2009, but the apparent unreliability of this device may require additional efforts to successfully retrieve the collars. We should consider the feasibility of using helicopter net-gunning to retrieve Lotck collars 67 �during capture e~forts scheduled for late February, 2009, assuming attempts to remotely detonate drop-off mechanisms fail. Monthly downloads and collars retrieved from mortalities yielded GPS movement and distribution data from 32 individuals during winter (Fig. 3), 28 during the spring transition period (Fig. 4), and 24 during early summer (Fig. 5). Observed winter deer distribution (Fig. 3) reasonably followed apriori expectations (Appendix I; Fig. 2) with minor differences in study area boundaries, as defined by deer use, except for the Story/Sprague study area, where wintering deer were distributed farther cast than expected (see Bartmann et al. 1992); this change in distribution may be due to changes in habitat conditions and/or potential increases in other ungulate populations (e.g., elk). Of the 32 deer monitored during winter, no interchange between winter herd segments was noted, but a few individuals traveled beyond areas of interest relative to control and treatment experimental units addressing energy development (Fig. 38). These movements can be addressed by either censoring those data or applying a covariate to the analyses. Based on the winter deer distribution data documented since January and the level of energy development activity present in April, 2008, we provide preliminary study area boundaries (Fig. 3) for future monitoring efforts to address experimental control (North Ridge, Yellow Creek, and Story/Sprague Gulch) and treatment (Ryan Gulch and Magnolia) areas addressing mule deer responses to beneficial habitat treatments and/or development activities. More specific boundaries will be assigned once data are analyzed from the remaining 43 collars scheduled for retrieval in February, 2009. During the spring transition period, deer from North Ridge and the northern half of Magnolia generally moved east, deer from southern Magnolia, Ycllow Creek, and Ryan Gulch moved south, and the Story/Sprague Gulch deer moved relatively short distances south and east (Fig. 4). As expected, summer deer distribution was more widely scattered than during winter with deer distributions radiating from the Piceance Basin to the northeast, east, southeast, and south generally following wintering deer from North Ridge, Magnolia-north, Story/Sprague Gulch, and Magnolia-south, Ryan Gulch, Yellow Creek. _,i FUTURE PLANS Funding has been recently secured to initiate the complete study proposal (Appendix I) beginning fall 2008 and continuing spring 2010. To address the other 5 study objectives outlined in Appendix I, we will attach VHF collars to 50 fawns/study area, increase our GPS sample to 20 GPS collared does/study area, measure body condition of 30 does/study area,- and add IO VHF collared does/study area to enhance mark-resight estimates. The period covered will represent existing development conditions or the pretreatment period and allow estimates of mule deer population parameters relative to current development practices and habitat conditions. Additional funding and cooperative agreements will be necessary to manipulate habitat conditions to benefit mule deer and modify development practices to enhance mule deer condition and survival on winter ranges exposed to energy development. We optimistically anticipate the opportunity to work cooperatively toward developing solutions for allowing the nation's energy reserves to be developed in a manner that benefits wildlife and the people who value both the wildlife and energy resources of Colorado. ...,, I._; LITERATORE CITED Anderson, C.R., Jr., and D. J. Freddy. 2007. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Draft Study Proposal, Colorado Division of Wildlife, Fort Collins, USA. Bartmann, R. M. 1975. Piccance deer study-population density and structure. Job Progress Report, Colorado Division of Wildlife, Fort Collins, Colorado, USA. Bartmann, R. B., and S. F. Steinert. 1981. Distribution and movements of mule deer in the White River Drainage, Colorado. Special Report No. 51. Colorado Division of Wildlife, Fort Collins, USA. 68 �...., -..,/ Bartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monograph No. 121. Barrett, M.W., J.W. Nolan, and L.D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10: 108-114. Gibbs, H. D. 1978. Nutritional quality of mule deer foods, Picearice Basin, Colorado. Thesis, Colorado State University, Fort Collins, USA. Unsworth, J. W., D. F. Pack, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. Van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA . ...,, val Prepared by _ _ _ _ _ _ _ _ _ __ Charles R. Anderson, Wildlife Researcher - ...,_ ..__I -.I Table I. Mortalities of GPS collared yearling and adult female mule deer in the Piceance Basin, Colorado, January-June, 2008. Deer ID Study area Mortality date Age class Apparent cause 150.194 150.235 219.159 150.094 219.149 150.275 216.706 217.615 Story/Sprague Gulch Magnolia Ryan Gulch North Ridge Story/Sprague Gulch Ryan Gulch North Ridge Magnolia 1/19/08 4/9/08 4/25/08 5/4/08 5/23/08 5/24/08 5/25/08 5/28/08 Young adult Young adult Yearling Young adult Old adult Young adult Old adult Young adult Undetermined Coyote predation Vehicle collision Coyote predation Malnutrition Undetermined Undetermined Undetermined 69 �- w \iii, Piceance Basin Gas Fields + Gas Basin + Figure I. Piceance Basin project area (dashed line) relative to mule deer winter range, oil and gas fields, and the oil and gas basin. ...., wii/1 Figure 2. Capture locations by study area (solid lines) ofG PS collared adul t fema le mule deer in the Piccance Basin, Colorado, January I 0-12, 2008. 70 - - �--- - -- - --- -- ---- -- -- Winter Mule Deer Locations and Natural Gas Development ,i, Active or Potentially Active Well Pads ' • Figure 3. Mule deer G PS locatio ns by preliminary study area boundary (solid lines) excl uding (lop) and including (bottom) active will pads and e nergy development tacili ties (as of April, 2008) in the Piceance Basin, Colorado. January-A pril , 2008. 71 ....., En ergy Development Facilities Mule cleer locations= circles. stars. triangles. pluses. ancl diamonds �\..,,; - Figure 4. GPS locations of Piccance Bas in mule deer during the spring transition period (April- May, 2008). Capture study site: circles= North Ridge, stars= Magno lia, triangles= Story/Sprague Gulch. diamonds= Ryan Gulch, pluses = Yellow Creek. GPS locations 613 - 7111 108 Deer ID (symbol s= capl u,o sl101 + 11 63_ 06J t 117!t_21 S * • 11 6.i_oH o 11; 0_ ,s~ o 11 n _:.i S; * • 11 65_085 a- 117 1_164 • 11 78_:.SE ,;') ~ 1167_ 10S 9 11 7:!_ 105 -,: 1179_165 • o 1169_1'5 - - "'""' ~ Figure 5. Summer range G PS locations of Piceanee Basin mule deer, June- Jul y, 2008. Capture study site: circles= North Ridge, stars = Magnolia, triangles = Story/Sprague Guk:h, diamonds= Rya n G ulch, pluses= Yellow Creek. 72 - �APPENDIX I PROGRAM NARRATIVE STUDY PLAN FOR MAMMALS RESEARCH FY 2007-08- FY 2012-13 POPULATION PERFORMANCE OF PICEANCE BASIN MULE DEER IN RESPONSE TO NATURAL GAS RESOURCE EXTRACTION AND MITIGATION EFFORTS TO ADDRESS HUMAN ACTIVITY AND HABITAT DEGRADATION ST AGE I, OBJECTIVE 5: PATTERNS OF MULE DEER DISTRIBUTION & MOVEMENTS A Research Study Plan submitted by: C.R. Anderson, Wildlife Researcher, Mammals Research, Colorado Division of Wildlife D.J. Freddy, Mammals Research Leader, Colorado Division of Wildlife 1,-1 >earl ...., ....,, -..,/ ...., A. Need Extraction of natural gas from areas throughout western Colora4o has raised concerns among many public stakeholders and the Colorado Division of Wildlife that the cumulative impacts associated with this intense industrialization will dramatically and negatively affect the wildlife resources of the region. Concern is especially high for mule deer due to their recreational and economic importance as a principal game species and their ecological importance as one of the primary herbivores of the Colorado Plateau Ecoregion. Extraction of natural gas will directly affect the potential suitability of the landscape used by mule deer by converting native habitat vegetation to drill pads, roads, or noxious weeds, by fragmenting habitat because of drill pads and roads, by increasing noise levels via compressor stations and vehicle traffic, and by increasing the year-round presence of human activities. Extraction will indirectly affect deer by increasing the human work-force populati<;m of the region and the subsequent need for developing additional landscape for human housing, supporting businesses, and upgraded road/transportation infrastructure. Additionally, incri~ed traffic on rural roads will raise the potential for vehicle-animal collisions and additive direct mortality to deer populations. Thus, research documenting these impacts and evaluating the most effective strategies for minimizing and mitigating these activities will greatly enhance future management efforts to sustain mule deer populations for future recreational and ecological values. The Piceance Basin in northwest Colorado supports one of the largest migratory mule deer populations in North America and also exhibits one of the highest natural gas reserves in North America. Projected energy development throughout northwest Colorado within the next 20 years is projected to be about 15,000 wells, many of which will occur in the Piceance Basin. The Piceance Basin (including the White River gas field immediately to the north) currently supports about 400 active gas well pads, 250 permits for development within the next year, and 200 energy development facilities (Colorado Oil and Gas Conservation Commission; Fig. 1). Wintering mule deer population segments in or immediately adjacent to the Piceance Basin include: Crooked Wash along the White River on the north edge of the Basin, North Ridge between Dry Fork of Piceance Cfeck and the White River in the northeastern portion of the Basin, Yellow Creek along Yellow Creek in the western portion of the Basin, Ryan Gulch between Ryan Gulch and Dry Gulch in the southwestern portion of the Basin, Magnolia north and east of Piceance Creek in the central portion of the Basin, and Story/Willow Creek between Willow Creek and Story Gulch in the southern portion of the Basin. Each of these wintering population segments has received varying levels of development, from little-no development in Story/Willow Creek and North Ridge, light development in Yellow Creek, and relatively high development in Ryan Gulch, Crooked Wash, and 73 �Magnolia segments (Fig. 2). Due to advances in resource extraction technology and the increased demand for natural gas, future development and extraction activities will likely focus on natural gas fields previously developed and expand into adjacent areas where previously identified oil shale reserves and natural gas basins provide additional resource extraction opportunities. Because of the variation in the geology relative to gas reserves in the area and the juxtaposition of differing mule deer winter herd segments, several opportunities are available to address different, but related, questions relative to natural gas extraction methods and mitigation efforts relative to mule deer habitat use patterns. Past Research The Piceance Basin has been the location of numerous research investigations conducted by the Colorado Division of Wildlife, Colorado State University, and others which addressed various aspects of mule deer ecology and management beginning in the 1970s and continuing through the mid 1990s. Previous investigations of Piceance Basin mule deer addressed food habits (Hansen and Dearden 1975, Hubbard and Hansen 1976, Gibbs 1978, Bartmann 1983), physiology (Bartmann 1986, Torbit ct al. 1988), development of management techniques (Freddy and Bowden 1983 b, Garrott and White 1984a, Lee et al. 1985, White and Bartmann 1994), efficacy of population sampling methods (Freddy and Bowden 1983a, Bartmann et al. 1986, 1987, White et al. 1989), and population dynamics (White and Bartmann 1983, 1998, Garrott and White 1984b, Lee 1984, Garrott et al. 1987, White ct al. 1987, Bartmann et at. 1992). Previous investigations of mule deer habitat use patterns in the Piceancc Basin (Garrott et al. 1987) suggested fall migration consistently occurred during November, but spring migration varied likely due to winter severity and body condition, where rapid migration was evident when deer were leaving winter range in good condition and delayed migration was indicative of deer transitioning from winter range in relatively poor condition. Garrott et al. ( 1987) also noted strong fidelity to seasonal ranges, that deer shifted from north to south slopes as winter severity increased, and that irrigated and fertilized hay meadows served as important transition areas during fall and spring migration periods. Bartmann et al. ( 1992) manipulated deer densities to demonstrate compensatory mortality in the Piceance Basin mule deer population, where overwinter fawn survival varied inversely with density and adult female survival remained relatively constant; fawn mortality rather than reproduction appeared to be the major process driving the density-dependent mechanism. White and Bartmann ( 1998) reduced deer densities by 75% in their treatment area and reported 16% higher overwinter fawn survival and fawn body mass averaging 0.8 kg higher than the control area, whereas adult female survival was comparable between areas supporting previous findings (Bartmann ct al. 1992). Empirical evidence of mule deer population response to habitat manipulations is currently limited, largely due to the logistical and financial difficulty in conducting long-term research sufficient to address this relationship. Density dependent relationships have been demonstrated (e.g., Bartmann et al. 1992) and habitat quality rather than proximate mortality factors (e.g., predation) appear to be the driving factor (Bartmann et al. 1992, Hurley and Zager 2004, Bishop et al. 2005). Bishop et al. (2005), however, demonstrated enhanced population performance in supplementally fed, free-ranging deer to simulate high quality habitat, and reported 18% higher fawn survival (fetus to yearling; fetus-neonate = 0.127, • overwinter= 0.240) and adult females averaged 5.5% more body fat, but reproduction and adult female survival were similar between treatment and control groups. Bergman et al. (2005, 2006) are currently investigating mule deer population response to habitat treatments in western Colorado, which will likely provide insight into our approach of addressing habitat treatments in response to energy development as this study progresses. Currently, research addressing mule deer activity in response to natural gas development is limited to one study from the Pinedale anticline in Wyoming (Sawyer et al. 2006). Sawyer et al. (2006) examined changes in distribution before and during development of a natural gas field, and observed shifts in mule deer habitat use away from well pads (2.7-3.7 km) within I year of development which continued throughout the study, suggesting indirect habitat loss may be substantially larger than direct 74 i..- �.,_, habitat loss and presumably results in deer using lower quality habitats that may ultimately lead to population decline. Mule deer habitat in Pinedale was much less topographically and vegetatively diverse than the Piceance Basin, however, and mule deer may respond differently where the habitat affords a higher degree of security cover. Mule Deer Response to Habitat Treatments and Changes in Development Practices Our primary goal of this study is to develop approaches to provide for energy extraction in a manner that maintains viable mule deer populations for future recreational and ecological purposes. This may be accomplished by restoring or enhancing habitat conditions on or adjacent to disturbed sites and by modifying development practices. Mitigating developed sites following disturbance requires reseeding or planting native vegetation, control of noxious weeds, and demonstrating success of mitigation efforts. Because mule deer are primarily browsers, shrub establishment will be essential, but shrub establishment is difficult and takes time for reemergence. Mule deer response to winter range mitigation efforts on disturbed sites will require relatively long-term monitoring to determine success of habitat treatments. More rapid habitat, and thus mule deer, responses can be expected from treating mule deer habitat adjacent to developed areas and by irrigating and fertilizing hay meadows adjacent to winter ranges (Garrott et al. 1987). Improving habitat conditions for or reverting succession of shrub communities using roller-chopping, hydro-axing, or fire can improve forage quality, and increasing forage quality and quantity by irrigating and fertilizing hay fields can improve mule deer body condition at critical times when transitioning to and from winter range. In addition to habitat treatments, mule deer may also benefit from modification in development practices that reduce human disturbance. Development practices that concentrate activities and/or minimize human disturbance will most likely minimize detrimental impacts to mule deer populations. Energy development practices that may be informative to investigate include directional versus non-directional drilling, piping versus trucking condensate from well pads, remotely versus directly monitoring gas wells, closing access roads following development, shifting from noisy diesel to quieter natural gas motors, and phased/clustered development where sections of deer winter range are developed while others remain undisturbed until development and mitigation are completed in developed sections. Determining the response of mule deer to specific development practices will require collaboration with the developer, and the specific conditions of the site being developed will dictate which development practices can feasibly be evaluated. Encana and Exxon-Mobile are the primary energy companies controlfing natural gas development in the Piceance Basin (Fig. 3 ). -..I Mule Deer Response to Energy Development Mule deer may negatively respond to energy development from direct reduction in forage availability from development activities, from indirect reduction of forage quality and quantity by shifting their distribution away from development activity to less preferred habitats, from negative physiological responses where deer maintain fidelity in areas exposed to development activities or from a combination of these factors. Depending on the extent and concentration of development, deer may also be able to adjust to development activities without population level impacts, and other factors (e.g., winter severity, drought, habitat succession, predation) also contribute to fluctuations in population performance/trajectory over time. Ultimately, reproduction and survival drive population perfonnance and, based on past research, focusing on fawn survival and recruitment appear to be the most influential parameters given the density dependent nature of these factors versus the apparent density independent nature of adult female survival and reproduction. Documenting proximate factors influencing fawn survival will also be useful and thus changes in distribution, deer density, body condition, and specific mortality factors should also be monitored. Comparing changes in mule deer population parameters relative to energy development will require that undeveloped control areas are monitored and predevelopment data are collected to determine whether or not and to what extent development versus environmental factors may be contributing. This will be challenging given development already in place and the unpredictability of future development that may occur. Large scale impacts from energy development may be detectable by comparing mule deer population parameters from undeveloped sites to ..,./ 75 -..I �developed sites, but natural variation due to geographic differences will be unaccounted for and add error to comparisons. Our ability to examine mule deer response to habitat mitigation and/or beneficial development practices will be better suited for demonstrating cause-effect relationships by allowing controlled experimental designs where habitat manipulation or modifying human behavior (i.e., development practices) provide the treatments for examining positive responses in mule deer population parameters. B. Objective The primary objectives for the long-term research proposal are as follows: I. Determine if winter range and riparian vegetation responds positively to habitat treatments; 2. Determine if fawn and yearling survival is positively influenced by winter range habitat treatments; 3. Determine if fawn and yearling survival is positively influenced by irrigating and fertilizing hay meadows adjacent to winter ranges; 4. Determine if modification of development practices positively influences mule deer population performance; 5. Determine if habitat treatments, changes in development practices, or natural gas development results in distributional shifts on mule deer winter range; 6. Determine if habitat treatments, changes in development practices, or natural gas development results in changing mule deer densities on winter range. The specific objective of this study plan is to address objective 5: Determine if habitat treatments, changes in development practices, or natural gas development result in distributional shifts on mule deer winter range in the Piceance Basin. The primary working hypotheses for the long-term research proposal are as follows: a. Landscape level habitat treatments do not influence forage quantity and quality; b. Fawn and yearling survival are not influenced by winter range habitat treatments; c. Fawn and yearling survival arc not influenced by modification of development practices; d. Mid-winter deer density does not fluctuate in response to habitat treatments, changes in development practices, or natural gas development; e. Mule deer habitat selection does not change in response to habitat treatments, changes in development practices, or natural gas development. The specific working hypothesis of this study plan is: Mule deer habitat selection docs not change in response to habitat treatments, changes in development practices, or natural gas development. C. Expected Results Due to the extensive energy development that is projected to occur over the next 20 years throughout much of the mule deer winter range in the northern Rocky Mountains of the western US, innovative approaches to energy development and mitigation methods are essential to sustain viable mule deer populations in the region. Impacts from development and conversely success of mitigation efforts are often assumed but rarely demonstrated, and these assumptions can only be confirmed by application of well designed research efforts conducted over sufficiently long time periods to measure responses. As a first step toward this effort, we propose to address mule deer habitat selection patterns relative to varying levels natural gas development and associated human activity and ultimately address mule distributional responses to habitat and development modifications anticipated to be beneficial to mule deer. This project will require coordination and cooperation between Colorado Division of Wildlife, land management agencies, and the major energy companies developing the Piceance Basin. We anticipate this partnership will benefit mule deer populations and foster the evolution of wildlife management and 76 -...,, �-energy development practices that are compatible with other wildlife and human values associated with maintaining functional ecosystems over the long term. .-I D. Approach 1. Experimental Approach a. Experimental Units Because of the varying levels of development and deer densities relative to differing winter population segments in the Piceance Basin, different experimental areas (i.e., mule deer winter ranges) arc uniquely suited for addressing mule deer habitat selection patterns relative to varying levels of energy development. Experimental designs monitoring mule deer responses to treatment (e.g., habitat mitigation, modified development practices) and control areas are necessary to differentiate cause-effect relationships from development versus environmental factors. Suitable control areas require that little or no previous development has occurred and that no development occurs during the experimental time frame. Ideally, both temporal and spatial control areas would be monitored to make valid comparisons to developed and subsequently mitigated sites; temporal controls provide measures of natural variability in mule deer population parameters over time and spatial controls provide measures of variability due to differences in geography. Once spatial and temporal variation is accounted for, inferences can be made relative to development disturbance or mitigation effects on mule deer. )'he North Ridge, Story/Willow Creek, and Yellow Creek deer population segment areas (Fig. 2) currently exhibit little to no development, but it is currently unknown whether or not these areas will be developed in the future; there is potential for future oil shale development in the Story/Willow Creek and Yellow Creek deer areas. North Ridge appears least likely to be developed because it is outside of the current oil shale lease area and. only a few natural gas wells have historically been drilled on or adjacent to the area, whereas some development is currently occurring and likely to increase in the Story/Willow Creek and Yell ow Creek areas. Thus, North Ridge would appear best suited as a temporal control site for comparison to other developed winter ranges within the Piceance Basin and may also serve as a geographic control for the Crooked Wash deer population segment located immediately north and adjacent to the Piceance Basin (as of Dec. 2007, the Crooked Was~ site ranks 61" in study priority and will not be sampled in the initial year due to limitedfunding). The Story/Willow Creek and Yellow Creek deer may provide spatial controls for the Magnolia and Ryan Gulch deer population segments, respectively, but future development potential in these areas is unknown. If these areas become developed in the future (either for oil shale or natural gas), they would provide BACI (Before-AfterControl-Impact) type comparisons strengthening our inference of development impacts on mule deer habitat selection patterns. Magnolia, Crooked Wash, and Ryan Gulch deer areas have historically received relatively high development activity and currently exhibit moderate-high development, and appear likely to be developed extensively in the future based on the gas development layers currently available (Colorado Oil and Gas Conservation Commission; Fig. 1). Pretreatment data in these areas will be represented by parameters associated with developed sites and the measured response will be in the form of habitat treatments and/or differing development practices, which will be measured in comparison to the control sites. We propose including 3 control sites ( 1 temporal/spatial control and 2 spatial controls) and 3 treatment sites to investigate mule deer response to habitat and/or development treatments (e.g., directional versus non-directional drilling, piping versus trucking condensate, etc.) across a range of deer densities (Table 1). We would strive to split high intensity extraction study sites into 2 halves with one half serving as the 'control' [standard development] and one half serving as the 'treatment' [improved development approach or improved habitat] (e.g., see Magnolia in Fig. 2). The above scenario addresses the potential for establishing control and treatment sites for evaluating shifts in mule deer habitat use patterns in response to habitat treatments and/or development treatments, and may allow larger scale 77 �comparisons in mule deer habitat use patterns relative to varying levels of energy development to be compared among experimental areas. Modified versions of the proposed design could be implemented depending on the level of funding available and the degree to which industry is willing to collaborate with this effort. We consider 3 study sites, likely North Ridge, Magnolia, and Ryan Gulch, as the minimum number of study sites necessary to adequately address the objectives of this project; the additional proposed study areas will allow increased flexibility in the questions that are addressed and increase our inference relative to mule deer responses to habitat treatments and modifications of development practices. Furthermore, ifwe are not able to evaluate potential for mitigating industrial operation and/or habitat improvements, this study would likely only have the potential to document negative impacts of intense energy extraction practices on mule deer. Table 1. Relative density of natural gas wells and mule deer and experimental designation for potential study sites in the Piceanace Basin, Colorado, for addressing mule deer response to natural gas development practices and habitat mitigation. Relative density Experimental Study area Inactive wells Active wells Mule deer designation North Ridge Very low None High Temporal/spatial control Crooked Washu High High High Treatment Story/Willow Creek Low Low Moderate Spatial control Magnolia High High Moderate Treatment Yellow Creek Moderate Low Low Spatial control Ryan Gulch High Moderate Low Treatment a As of Dec. 2007, for the initial research effort, the Crooked Wash study site ranks 6th in priority and will not be sampled due to limited funding. b. Response Variables To determine if habitat treatments or development practices elicit a shift in habitat use patterns, we will examine changes in Resource Selection Probability Functions (RSPF; Sawyer et al. 2006) preand post-habitat treatments, between areas exhibiting differing development practices, and compare RSPFs between developed and non-developed sites. Population level models for each study area will be compared to assess similarities and differences in habitat selection patterns relative to differing levels of energy development. We suggest relevant habitat attributes associated with mule deer response to habitat treatments and development practices include slope, aspect, elevation, habitat type, road density, distance to well pad, and development activity. Definition for development activity would vary depending on the development treatment investigated. For example, if the development treatment were applied to examine 78 ~ �fluid collection systems, the variable would be coded l or O depending on whether they were present or absent and the RSPF would be estimated relative to this effect. In another example, well pad visitation rate may be the variable of interest and the RSPF would be estimated for a continuous effect of increasing road traffic to well pads. 2. Sample Size I Power Calculations We anticipate 20 GPS collars per experimental area will be sufficient to provide population level inference based on similar studies with ungulates (Millspaugh and Marzluff 200 l) for addressing adult female mule deer habitat selection patterns for each study site. 3. Procedures a). Capture and Handling Methods A total of 120 adult female mule deer will be captured and OPS-collared (20/study area assuming 6 study sites, I00 deer during initial FY07-08 for 5 study sites). Helicopter net-gunning (Barrett et al. 1982, van Reenen 1982) will be used to complete the necessary sample in January 2008 and a combination of helicopter net-gunning and drop netting will be used during March of subsequent years. b). Monitoring Habitat Use Patterns Habitat use patterns on treatment and control sites will be evaluated applying the Resource Selection Probability Function (RSPF) approach of Sawyer et al. (2006), where resource selection is estimated using the relative frequency or absolute probability of use as a function of the predictor variables. This approach will consists of 5 basic steps including (I) estimate the relative frequency of use (an empirical estimate of probability of use) for a large number of sampling units for each GPS collared deer (20/study area), (2) use the relative frequency as the response variable in a multiple regression analysis to model the probability of use for each deer as a function of predictor variables, (3) develop a population level model from the individual deer models for each experimental area, (4) map predictions from each model annually to examine changes in habitat use patterns over time relative to treatment effects, and (5) compare population level model coefficients between treatment and control sites to examine differences in resource selection among non-developed, developed, and mitigated sites. Relative frequency of use for each deer will be estimated by counting the number of deer locations that occur within 100-m radii circular sampling units (representing habitat attributes) systematically sampled throughout each study area; 200-m-wide sample unit should be small enough to detect changes in deer movements and large enough to provide multiple locations for estimating use probability functions. c ). Habitat Manipulations The purpose of habitat manipulation would be 2-fold: 1) replace forage lost directly to surface destmction associated with gas pad/road/infrastructure development through rehabilitation of these areas, and 2) enhance suitable undisturbed vegetation. In both situations, the goal would be to provide habitats/vegetation having enhanced nutritional value to mule deer during fall (pre-winter) and spring (post-winter) migrations and during the critical winter period in order to improve body condition of deer and enhance their probability of survival. Placement of such habitat treatments would need to be evaluated and planned based on identification of priority areas within the Piceance Basin, in general, and specifically within experimental study sites. Opportunities within study sites would, in part, be dependent on cooperation of Energy Corporations, BLM, and private land owners, and site specific potentials that realistically can only be specifically determined after commitments are made in choosing experimental sites. We envision the potential to utilize a full-suite of habitat improvement options. These could include: enhancing existing sagebrush areas using combinations of herbicide, nitrogen fertilizer, chopping-mowing, reseeding with grasses-forbs, and in some cases reseeding with suitable sagebrush species; enhancing mountain brush habitats through burning, hydroaxing, and reseeding; enhancing 79 �~ pinyon-juniper habitats through hydroaxing, burning, and reseeding. Site specific situations could require using advanced mulching, seeding, and irrigation options to effectively rehabilitate sites. In all cases, we would attempt to layout experimental habitat improvements to facilitate evaluation of success both from the standpoint of vegetation rehabilitation and use by mule deer. Past research and monitoring of radio-collared mule deer in the Piceance Basin documented the high use and importance of cultivated hay fields along Piceance Creek. We envision considerable potential to improve management of hayfields to specifically address the needs of deer, especially during post-fall and pre-spring migrations of deer into and out of the Piceance Basin. The potential to manage hayfields for deer will be dependent on options to own or lease fee-title property and water rights. There may be nearly 10.000 acres of suitable hayfields located along Piceance, Ryan, Black Sulphur and Yell ow creeks. In general, we believe that hayfields using more efficient irrigation practices and planted with suitable varieties of alfalfa developed to be grazed more so than for traditional hay production and suitable to alkaline soils would offer high potential to enhance nutrition of deer at key periods of the year. We also could see potential to establish hayfields with appropriate varieties of cool-season grasses (bluegrass for example) that could be managed for high nutritional quality through annual burning, mowing, grazing, and irrigation practices. Such cool season grass fields could provide 'green' forage for deer both during spring "green-up' and fall •re-green' periods, especially if limited irrigation could be applied. The specific design and layout of reformed hayfield management would require considerable planning involving the expertise ofNRCS or University Extension programs and considerable cost (potentially millions of dollars) for fee title ownership of land and water rights, mechanical preparation of hayfields and irrigation systems, and annual management practices once fields were established. l_.i d) Evaluation ~f Development Practices We anticipate options for industry to alter extraction practices that would reduce and/or concentrate human activity and benefit deer by increasing the relative 'security' of existing or improved habitats for deer. Options could include: multi-well versus single-well drilling platforms to reduce well pad density; piping instead of trucking well-condensate; road closures that minimize where traffic occurs; time of day restrictions; remote well-monitoring, or other options that industry may be able to offer. The key to evaluating any of these industrial-human activity options would be to create experimental comparisons using "control' areas [current practices] versus 'treatment' areas [improved practices]. Which alternative practices are tested and in which potential study sites involved will depend upon cooperation from industry. Ideally, energy corporations would cooperate among themselves, the BLM, and with Division of Wildlife to help develop the best possible experimental design among extraction lease areas. l_.i e). Statistical Ana~vses Following Sawyer et al. (2006) for estimating Resource Selection Probability Functions, we will obtain population-level models for each experimental area by first estimating coefficients for each GPScollared deer. A negative binomial distribution will be used to fit the following general linear model (GLM): ln(E[r;]) = ln(total) +/Jo+ P1X1 + ... + /J,,Xp, where r; is the number oflocations for a OPS-collared deer within sampling unit i (i = /, 2, .... r), total is the total number of locations for that deer within each experimental unit, Po is the intercept tenn, P1, .. ./J,, are unknown coefficients for habitat variablesX,, ... ,.,~,. and E[.] denotes the expected value. We will estimate coefficients for the population-level model for each experimental unit following: " 1 11 P1. =- L/Jki' n j=I 80 '-' �-..I where pk.i is the estimate of coefficient k for individual j U = 1, ... ,n) and the variance wi 11 be estimated applying the variation among individual model coefficients. To compare habitat use patterns between areas and over treatment effects, we will map predicted probabilities of use for each study area by season. Differences (P < 0.05) between population level model coefficients will be compared between study areas using a I-test. 4. Project Schedule FY2007-08 FY2008-09 FY2009-I0 FY2010-l I FY201 l-12 FY2012-13 FY2013-14 FY2014-15 FY2015-16 FY2016-l 7 FY2017-18 FY2018-19 ~ -...,,I Pretreatment/Revised Program Narrative Study Plan Pretreatment/Progress Report (PR) Habitat and/or Development Treatments/PR Habitat-and/or Development Treatments/PR Monitor Deer Response/Progress Report Project Status Evaluation Monitor Deer Response/Progress Report Monitor Deer Response/Progress Report Monitor Deer Response/Progress Report Project Status Evaluation Monitor Deer Response/Progress Report Monitor Deer Response/Progress Report Monitor Deer Response/Completion Report Prepare and submit peer-reviewed publications 9/1/2007 8/1/2008 8/1/2009 8/1/2010 8/ 1/20 I I 8/1/2012 8/1/2013 8/1/2014 8/1/2015 8/1/2016 8/1/2017 8/1/2018 5. Annual Cost Estimates Estimating mule deer resource selection probability functions and implementing small scale habitat improvements are costly endeavors involving the purchase of specialized GPS radio-collars, helicopter flight hours for deer capture/collaring, machinery to physically alter the habitat, and personnel to adequately perform day-to-day data collection. If large scale habitat treatments arc needed or desired, funding in addition to the estimates below will be required as habitat treatments cost $300 to $1,000/acre depending on the most appropriate treatment for a locale. Key to evaluating mule deer responses to habitat and/or developmen't treatments will be sufficient and steady funding over a time horizon (minimum of 5-year commitments over the IO year study period) that allows for meaningful biological responses to occur and be measured. Cost estimates per year (2007 dollars for objective #5): GPS Equipment Costs: $200,000 Helicopter Capture Costs: $ 70,000 12 months TFTE: $ 30,000 Vehicle support: $ 20.000 Other field operations and equipment: $ 15,000 Total: $335,000 6. Personnel Charles R. Anderson, Jr., Wildlife Researcher, Project Leader, Colorado Division of Wildlife David J. Freddy, Mammals Research Leader, Colorado Division of Wildlife E. Location of Work The proposed research will take place in or adjacent to the Piceance Basin of northwest Colorado, primarily within Game Management Unit 22 of the White River mule deer DAU D-7, west and southwest of Meeker, Colorado (Fig. 2). 81 �F. Literature Cited Anderson, C.R., Jr., and D. J. Freddy. 2007. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Draft Study Proposal, Colorado Division of Wildlife, Fort Collins, USA. Bartmann, R. M. 1983. Composition and quality of mule deer diets pinyon-juniper winter range, Colorado. Journal of Range Management 36:534-541 Bartmann, R. M. 1986. Growth rates of mule deer fetuses under different winter conditions. Great Basin Naturalist 46:245-248. Bartmann, R. M., L. H. Carpenter, R. A. Garrott, and D. C. Bowden. 1986. Accuracy of helicopter counts of mule deer in pinyon-juniper woodland. Wildlife Society Bulletin 14:356-363. Bartmann, R. M., G. C. White, and L. H. Carpenter. 1987. Aerial mark-recapture estimates of confined mule deer in pinyon-juniper woodland. Journal of Wildlife Management 51 :41-46. Bartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monograph No. 121. Barrett, M.W., J.W. Nolan, and L.D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin I 0: I 08-114. Bergman, E. J., C. J. Bishop, D. J. Freddy, and G. C. White. 2005. Pilot evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer (study plan). Wildlife Research Report July: 23-35. Colorado Division of Wildlife, Fort Collins, USA. Bergman, E. J., C. J. Bishop, D. J. Freddy, and G. C. White. 2006. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer (study plan). Wildlife Research Report July: 67-89. Colorado Division of Wildlife, Fort Collins, USA. Bishop, C. J., G. C. White, D. J. Freddy, and B. E. Watkins. 2005. Effect of nutrition on mule deer recruitment and survival rates. Wildlife Research Report July: 37-65. Colorado Division of Wildlife, Fort Collins. USA. Coo~ R.C. 2000. Studies of body condition and reproductive physiology in Rocky Mountain Elk. Thesis, University ofldaho, Moscow, USA Freddy, D. J., and D. C. Bowden. 1983a. Sampling mule deer pellet-group densities injuniper-pinyon woodland. Journal of Wildlife Management 47:476-485. Freddy, D. J., and D. C. Bowden. 1983b. Efficacy of permanent and temporary pellet plots injunipcrpinyon woodland. Journal of Wildlife Management 47:512-516. Freddy, D.J., G.C. White, M.C. Kneeland, R.H. Kahn, J.W. Unsworth, W.J. DeVcrgie, V.K. Graham, J.H. Ellenberger, and C.H. Wagner. 2004. How many mule deer arc there? Challenges of credibility in Colorado. Wildlife Society Bulletin 32:916-927. Garrott, R. A., and G. C. White. 1984a. Evaluation of vaginal implants for mule deer. Journal of Wildlife Management 48:646-648. Garrott, R. A., and G. C. White. 1984b. Methodology for assessing the impacts of oil shale development on the Piceance Basin mule deer herd. Thome Ecological Institute Technical Publication 14:228. 231. Garrott, R. A., G. C. White, R. M. Bartmann, L. H. Carpenter, and A. W. Alldredge. 1987. Movements of female mule deer in northwest Colorado. Journal of Wildlife Management 51 :634-643. Gibbs, H. D. 1978. Nutritional quality of mule deer foods, Piceance Basin, Colorado. Thesis, Colorado State University, Fort Collins, USA. Gill, R.B. 1969. A quadrat count system for estimating game populations. Colorado Division of Game, Fish and Parks, Game Information Leaflet 76. Fort Collins, USA. Hansen, R. M., and B. L. Dearden. 1975. Winter foods of mule deer in the Piceance Basin, Colorado. Journal of Range Management 28:298-300. • Hubbard, R. E., and R. M. Hansen. 1976. Diets of horses, cattle, and mule deer in the Piceance Basin, Colorado. Journal of Range Management 29:389-392. Hurley, M., and P. Zager. 2004. Southeast mule deer ecology - Study I: Influence of predators on mule deer populations. Progress Report, Idaho Department of Fish and Game, Boise, USA. 82 __.i ...,, �..., Kufeld, R.C., J .H. Olterman, and D.C. Bowden. 1980. A helicopter quadrat census for mule deer on Uncompahgre Plateau, Colorado. Journal of Wildlife Management 44:632-639. Lee, J. E. 1984. Mule deer habitat use and movements on Piceance Basin winter range as estimated by radiotelemetry. Thesis, Colorado State University, Fort Collins, USA. Lee, J. E., G. C. White, R. A. Garrott, R. M. Bartmann, and A. W. Alldredge. 1985. Assessing accuracy of a radiotelemetry system for estimating animal locations. Journal of Wildlife Management 49:658-663. Millspuagh, J. J. and J.M. Marzluff. 2001. Radio tracking and animal populations. Academic Press, San Diego, USA. Ramsey, C.W. 1968. A drop-net deer trap. Journal of Wildlife Management 32: 187-190. Sawyer, H., R. M. Nielson, F. Lindzey, and L. L. McDonald. 2006. Winter habitat selection of mule deer before and during development of a natural gas field. Journal of Wildlife Management 70:396403 . Schmidt, R.L., W.H. Rutherford, and F.M. Bodenham. 1978. Colorado bighorn sheep-trapping techniques. Wildlife Society Bulletin 6: 159-163. Stephenshon, T.R., V.C. Bleich, B.M. Pierce, and G.P. Mulcahy. 2002. Validation of mule deer body composition using in vivo and post-mortem indices of nutritional condition. Wildlife Society Bulletin 30:557-564. • Stephenshon, T.R., T.R., K.J. Hundertmark, C.C. Schwartz, and V. Van Ballenberghe. 1998. Predicting body fat and body mass in moose with ultrasonography. Canadian Journal of Zoology 76:717722. Torbit, S. C., L. H. Carpenter, R. M. Bartmann, A. W. Alldredge, and G. C. White. 1988. Calibration of carcass fat indicies in wintering mule deer. Journal of Wildlife Management 52:582-588. Unsworth, J. W., D. F. Pack, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. Van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-421 in L. Nielsen, J. C. Ha.igh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. White, G. C. 1996. NOREMARK.: Population estimation from mark-resighting surveys. Wildlife Society Bulletin 24:50-52. White, G. C., and R. M. Bartmann. 1983. Estimation of survival rates from band recoveries of mule deer in Colorado. Journal of Wildlife Management 47:506-511. White, G. C., R. A. Garrott, R. M. Bartmann, L. H. Carpenter, and A. W. Alldredge. 1987. Survival of mule deer in northwest Colorado. Journal of Wildlife Management 51 :852-859. White, G. C., R. M. Bartmann, L. H. Carpenter, and R. A. Garrott. 1989. Evaluation of aerial line transects for estimating mule deer densities. Journal of Wildlife Management 53:625-635. White, G. C., and R. M. Bartmann. 1994. Drop nets versus helicopter net guns for capturing mule deer fawns. Wildlife Society Bulletin 22:248-252. White, G. C., and R. M. Bartmann. 1998. Effect of density reduction on overwinter survival of freeranging mule deer fawns. Journal of Wildlife Management 62:214-225. 83 �I ~ r, ' " "f ef; • , V:~-Rla° Bl.A ~~ . r;,. •. Piceance Basin . .:•. •• ~ -~ + Active Wells + Inactive Wells + Permitted Wells -'"""' ,._, ......,/ I.....; - ,L ,. !<hfl'o«&'1 Figure I. atural gas development in the Piceancc Basin, Colorado, July 2007. \,,_; "'-i'I 84 �_,J --...J , !,'' ' ~- ,:I,• Piceance Basin • --t~ell • Research Areas + Inactive Wells - + Drilling Permits -- - - == =· •--·_int4 ..1, ------- I~• Figure 2. Proposed mule deer study sites relative to natural gas development in the Piceance Basin, Colorado, Jul y 2007. 85 �\,...,, I...., Piceance Basin '-"I ~ Research Areas + Gas Lease Ownership 0 1. 1•1111.,J :--111,h .\n:a c::J l'11•j1•"\'1J Slmh \u:.11, ......., ~ 1111 l h\.'":,p,·ll.~• I ,•;i, c fllll'.t.~·, woll.';1,..Fu.::m:i I c.1~< EJ - F\.'\,11il ,·;",: m , 1,,1111( 1\•:i,;,• - \\1lh,111t, l .:.t,'(,: Figure 3. Proposed mule deer study sites relati ve to the primary energy companies controlling natural gas leases in the Piceance Basin, Colorado, July 2007. "- I,,..., 86 �- -- Colorado Division of Wildlife July 2008 - June 2009 WILDLIFE RESEARCH REPORT ...,,/ - _, State of- - - - - - ~ Colorado : Division of Wildlife ~ ~ ~- - - - - =-:...:...:..::-=-=-'-"-'-'-'c..:..:..=:..:..:..:.-=------ - - - - -- - C ost Center 3430 : ::..:.M=a=m'-'-'-'-m=a=ls~R=cs"'"'e""a'-'-rc=h- ' - - - - - - - - - - - - - - Work Package 300 I := D--=c=c"---r--=C--=o~n=se~1_·v~a~ti-=-o~n_ _ _ __ _ _ __ _ _ _ __ Task No.6 : Population Performance of Piccancc Basin Mule Deer in Response to Natural Gas Resource Extraction and Mitigation Efforts to Address Human Activity and Habitat Degradation Fcderal A id Proj ect:_--'-W-'------'1=8-=-5---'-= R ------Period Covered: July 1, 2008 -June 30, 2009 Author: C. R. Anderson Personnel: D. Alkire, E. Bergman, C. Bishop, J. Broderick, 8. dcVergic, D. Fi nley, C. Flickinger, D. Freddy, L. Gepfort, K. Kaai, L. Kelly, T. Knowles, P. Lendrum, P. Lukacs, B. Marsh, M. Reitz, T. Segal, K. Taylor, R. Velarde, CDOW; E. Hollowed, BLM; S. Monsen, Western Ecological Consulting, Inc.; G. White, Colorado State University; R. Swisher, Quicksilver Air, Jnc. Project support received from Federal Aid in Wildlife Restoration, Colorado Mule Deer Association, Colorado M ule Deer Foundation, Colorado Oil and Gas Conservation Commission, Colorado State Severance Tax Fun~, EnCana Corp., Shell Petroleum, and Williams Production LMT Co. ----...., - -.J All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT I propose to experimentally evaluate habitat treatments that may improve the landscape to benefit mule deer (Odocoileus hemionus) and evaluate human-activity management alternatives to reduce the disturbance of energy development impacts on mule deer. The Piccance Basin of northwestern Colorado was selected as the project area due to ongoing natural gas development in one of the most extensive and important mule deer winter and transition range areas within the state. The data presented here represent the first pretreatment year of a long-tenn study addressing habitat modifications and improved energy development practices intended to improve mule deer fitness in areas exposed to extensive energy development. I selected 5 winter ra nge study areas representing varying levels of development to serve as treatment (Ryan Guieb and Magnolia) and control (Yellow Creek, Story/Sprague, and North Ridge) sites and recorded habitat use and movement patterns using GPS collars (5 locations/day), estimated overwinter fawn and adult female sw-vival, estimated late winter body condition of adult females using ultrasonography, and estimated abundance using helicopter mark-rcsight surveys. 1 attached 250 VHF collars (SO/study area) to fawns in early December 2008 and 150 VHF ( I 0/study area) and GPS (20/study area) collars to adult female mule deer in late February-early March 2009. In comparing the data among study areas th is first year, Story/Sprague deer appear to be in better phys ical condition than deer from the other winter ranges examined. Migration patterns were similar among 4 of the 5 areas, but Story/Sprague deer traveled shorter distances and spent less time on winter range. Y cllow Creek fawns were lighter than 1I I �~ ~ other study areas and ex hibited the lowest survival o f the a reas investigated. North Ridge deer exhi bited the highest w inter range density and Magnolia and Ryan Gulc h deer exhibited the low.est densities. R easons fo r these differences arc cuITcntly unknown, but could be related to severa l factors including relative habitat conditions, duration on and distance to seasonal ranges, and extent of human activi ty throughout occupied habitats. Meaningful comparisons will be evident once treatments arc implemented a nd comparisons arc possible between areas that arc manipulated (treatment a reas; Ryan Gulch and Magnolia) and those that are not (control areas: Yellow Creek, Story/Sprague, a nd North Ridge). This project w ill require additiona l funding commitments and cooperative agreeme nts beyond spring 20 I0 from private industry, the BLM, and the CDOW to assess if sustai nable mule deer populations can persist w ithin a liighly disturbed landscape following implementation of beneficial habitat treatments a nd development practices. ~ \..,,I ......, '-I -..,..; ~ ..,,,,,,, \..,,I "'""' ;....; """ ......., ,._,. """" 'I;..;, 11 2 �...,/ WILDLIFE RESEARCH REPORT POPULATION PERFORMANCE OF PICEANCE BASIN MULE DEER IN RESPONSE TO NATURAL GAS RESOURCE EXTRACTION AND MITIGATION EFFORTS TO ADDRESS HUMAN ACTIVITY AND HABITAT DEGRADATION CHARLES R. ANDERSON, JR. P. N. OBJECTIVES 'allll I. 2. To determine experimentally whether enhancing mule deer habitat conditions on winter and/or transition range elicits behavioral responses, improves body condition, increases overwinter fawn survival, or ultimately, population density on mule deer winter ranges exposed to extensive energy development. To determine experimentally to what extent modification of energy development practices enhance habitat selection, body condition, over-winter fawn survival, and winter range mule deer densities. SEGMENT OBJECTIVES 1. Collect and reattach GPS collars (5 fixes/day) to maintain sample sizes for addressing mule deer habitat use and behavior patterns in 5 study areas experiencing varying levels of energy development of the Piceance Basin, Colorado. 2. Estimate late winter body condition of adult female mule deer in each of the 5 winter herd segments 3. Monitor over-winter survival of fawn and adult female mule deer by daily ground tracking and bi-weekly aerial tracking. 4. Conduct Mark-Resight helicopter surveys to estimate mule deer abundance in each study area. 5. Summarize data and present information in an annual Job Progress Report. INTRODUCTION Extraction of natural gas from areas throughout western Colorado has raised concerns among many public stakeholders and the Colorado Division of Wildlife that the cumulative impacts associated with this intense industrialization will dramatically and negatively affect the wildlife resources of the region. Concern is especially high for mule deer due to their recreational and economic importance as a principal game species and their ecological importance as one of the primary herbivores of the Colorado Plateau Ecoregion. Extraction of natural gas will directly affect the potential suitability of the landscape used by mule deer through conversion of native habitat vegetation with drill pads, roads, or noxious weeds, by fragmenting habitat because of drill pads and roads, by increasing noise levels via compressor stations and vehicle traffic, and by increasing the year-round presence of human activities. Extraction will indirectly affect deer by increasing the human work-force population of the region resulting in the need for additional landscape for human housing, supporting businesses, and upgraded road/transportation infrastructure. Additionally, increased traffic on rural roads will raise the potential for vehicle-animal collisions and additive direct mortality to deer populations. Thus, research documenting these impacts and evaluating the most effective strategics for minimizing and mitigating these activities will greatly enhance future management efforts to sustain mule deer populations for future recreational and ecological values. .._, 113 �The Piceance Basin in northwest Colorado contains one of the largest migratory mule deer populations in North America and also exhibits some of the largest natural gas reserves in North America. Projected energy development throughout northwest Colorado within the next 20 years is about 15,000 wells, many of which will occur in the Piceance Basin, which currently supports over 250 active gas well pads (http://cogcc.state.co.us). Anperson and Freddy (2008a) in their long-term research proposal • identified 6 primary study objectives to assess measures to offset impacts of energy extraction on mule deer population perfonnance. This progress report describes the first year of addressing mule deer population performance during the pretreatment phase, which includes monitoring habitat selection and behavior patterns of adult female mule deer, overwinter fawn and adult female survival, estimates of adult female body condition during late winter, and abundance estimates on 5 winter range herd segments in relation to varying levels of natural gas development in control and treatment experimental areas prior to proposed experimental modifications in energy developmental practices and potential habitat improvement treatments. STUDY AREAS The Piceance Basin between the cities of Rangely, Meeker, and Rifle in northwest Colorado was selected as the project area due to its ecological importance as one of the largest migratory mule deer populations in North America and because it exhibits one of the highest natural gas reserves in North America (Fig. I). Historically, mule deer numbers on winter range were estimated between 15,00022,000 (Bartmann 197 5), and the current number of well pads (Fig. I) and projected number of gas wells in the Piceance Basin over the next 20 years is about 250 and 15,000, respectively. Mule deer winter range in the Piceance Basin is predominantly characterized as a topographically diverse pinion pine (Pinus edulis)-Utah juniper (Juniperus osteosperma; pinion-juniper) shrubland complex ranging from 1675 m to 2285 m in elevation (Bartmann and Steinert 1981 ). Pinion-juniper are the dominant overstory species and major shrub species include Utah serviceberry (Amelanchier utahensis), mountain mahogany (Cercocarpus montanus), bitterbrush (Purshia tridentata), big sagebrush (Artemisia tridentata), Gamble's oak (Quercus gambelii), mountain snowberry Symphoricarpos oreophilus), and rabbitbrush ( Chrysothamnus spp.; Hartmann et al. 1992). The Piceance Basin is segmented by numerous drainages characterized by stands of big sagebrush, saltbush (Atrip/ex spp.), and black greasewood (Sarcobatus vermiculatus), with the majority of the primary drainages having been converted to mixed-grass hay fields. Grasses and forbs common to the area consist of wheatgrass (Agropyron spp. ), blue grama (Bouteloua gracilis), needle and thread (Stipa comata), Indian rice grass (Oryzopsis hymenoides), arrowleafbalsamroot (Ba/samorhiza sagittata), broom snakeweed (Gutierrezia sarothreae), pinnate tansymustard (Descurainia pinnata), milkvetch (Astragalus spp.), Lewis flax (Linum lewis ii), evening primrose (Oenothera spp.), skyrocket gilia (Gilia aggregata), buckwheat (Erigonum spp.), Indian paintbrush (Castilleja spp.), and penstemon (Penstemon spp.; Gibbs 1978). The climate of the Piceance Basin is characterized by warm dry summers and cold winters with most of the annual moisture coming from spring snow melt. Wintering mule deer population segments in the Piceance Basin include: North Ridge (57 km2) between Dry Fork of Piceance Creek and the White River in the northeastern portion of the Basin, Yellow Creek (70 km2) along Corral Gulch in the western portion of the Basin, Ryan Gulch (130 km 2) between Ryan Gulch and Dry Gulch in the southwestern portion of the Basin, Magnolia (130 km 2) north and east of Piceancc Creek in the central portion of the Basin, and Story/Sprague Gulch (90 km2) between Story Gulch and Sprague Gulch in the southern portion of the Basin (Fig. l ). Each of these wintering population segments has received varying levels of development, from no development in North Ridge, light development in Story/Sprague Gulch and Yellow Creek, and relatively high development in Ryan Gulch and Magnolia segments (Fig. I). Among the 5 study areas, Yellow Creek and Story/Sprague will serve as spatial controls to Ryan Gulch and Magnolia, respectively, and North Ridge will serve as a temporal control area. Because the progression and extent of energy development in the future is 114 �dynamic and currently unknown, North Ridge may also serve as a spatial control area to Magnolia or possibly Ryan Gulch should the Story/Sprague Gulch or Yellow Creek study areas become developed in the future. METHODS ...., ..,._ ~ Tasks addressed this fiscal year included mule deer capture and collaring efforts, monitoring overwinter fawn and adult female survival, estimating adult female body condition during late winter using ultrasonography, and estimating mule deer abundance applying helicopter mark-resight surveys. employed helicopter net-gunning techniques (Barrett et al. 1982, van Reenen 1982) to capture 50 fawns during early December and 30 adult females during late February-early March in each of the 5 study areas (250 fawns and 150 does total). Once netted, all deer were hobbled and blind folded. Fawns were weighed, radio-collared and released on site, and adult females were transported to a handling site for collection of body measurements and were fitted with GPS (20/area; 5 fixes/day; G2 l l OB, Advanced Telemetry Systems, Isanti, MN, USA) or VHF collars (IO/area) and released. Fawn collars were spliced and fitted with 2 lengths of rubber surgical tubing to facilitate collar drop during mid-summer-early autumn, adult VHF collars were attached static, and GPS collars were supplied with timed drop-off mechanisms scheduled to release early April, 20 I0. All radio-collars were equipped with mortality sensing options (i.e., increased pulse rate following 8 hrs of inactivity). Mule Deer Habitat Use and Movements I downloaded and organized data from GPS collars deployed during the pilot study (January 2008; see Anderson and Freddy 2008b) following collar drop and retrieval late February 2009. GPS collars redeployed late February-early March 2009 maintained the same fix schedule of attempting fixes every 5 hours. All well pads and roads present throughout the 5 study areas in spring 2009 were mapped using hand-held GPS units and data were incorporated into ArcGIS 9.2 for resource selection analyses. I plotted deer locations and recorded timing and distance of spring and fall 2008 migrations for each study area. Mule deer resource selection analyses for the first winter of research (January-May 2008) are pending acquisition of information on timing of road and well pad development and completion. Analyses of data from winter 2008-2009 will be conducted following retrieval of GPS collars in April 20 I0. Over-Winter Survival Mule deer mortality monitoring consisted of daily ground telemetry tracking and aerial monitoring deer approximately every 2 weeks from fixed-wing aircraft. Once a mortality signal was detected, deer were located and necropsied to assess cause of death. I estimated over-winter survival on a weekly basis using the staggered entry Kaplan-Meier procedure (Kaplan and Meier 1958, Pollock et al. 1989). Capture-related mortalities (any mortalities occurring within IO days of capture) and collar failures were censored from survival rate estimates. I estimated over-winter survival rates beginning 14 December, 2008-20 June, 2009 for adult females and 14 December, 2008-21 March, 2009 for fawns. Premature failure of surgical tubing integrity beginning late March inhibited my ability to reasonably estimate fawn survival beyond late March. Adult Female Body Measurements I applied ultrasonography techniques described by Stephenson et al. ( 1998, 2002) and Cook et al. (200 I) to measure maximum subcutaneous rump fat (mm) and loin depth (longissimus dorsi muscle, mm). I estimated a body condition score (BCS) for each deer by palpating the rump (Cook et al. 200 I). combined ultrasound rump fat measurements with BCS to develop an index (rLIVINDEX; Cook et al. 200 I, 2007) of the relative nutritional status of deer from each study area. I examined differences (P < 0.05) in nutritional status among study areas using a two-sample t-test. Other body measurements recorded included pregnancy status (pregnant, barren) via ultrasound, weight (kg), chest girth (cm), and ....,, 115 �hind-foot length (cm). Fetal counts were also recorded in 4 of the 5 study areas to assist a Vaginal Implant Transmitter (VIT) evaluation study (see Bishop 2009). Abundance Estimates . I conducted 4 (Ryan Gulch) or 5 (the remaining study areas) helicopter mark-resight surveys (2 observers and the pilot) during late March-early April, 2009 to estimate deer abundance in each of the 5 study areas. I delineated each study area from GPS locations during the same period the previous year and aerial telemetry locations of radio-collared deer within 2 weeks of the first survey. The survey boundary of each study area was then extended to the nearest section boundary and study areas were divided into 2.6 km2 sampling blocks. Aerial telemetry surveys were conducted during helicopter surveys to determine which marked deer were within each survey area. Initially, I randomly selected IO sampling blocks from each study area (total sampling blocks= 22-50/study area) for each survey and surveyed sampling blocks sequentially to minimize flight time. After the first 2-3 surveys, depending on the area, it became apparent that increasing the number of sampling blocks to improve precision could be accomplished without undue expense, and subsequent surveys included all sampling blocks for the smaller areas (North Ridge, Yellow Creek, Story/Sprague) or 40% of the sampling blocks for the larger areas (Ryan Gulch, Magnolia). I delineated flight paths in ArcGIS 9.2 prior to surveys following topographic contours (e.g., drainages, ridges) and approximating 500 m spacing throughout selected survey blocks; flight paths during surveys were followed using GPS navigation in the helicopter. All deer observed within and between sampling blocks within the study area were included in abundance estimates. Two approximately 12 x 12 cm pieces of Ritchey livestock banding material (Ritchey Manufacturing Co., Brighton, CO USA) were uniquely marked using number, symbol combinations and attached to each radio-collar to enhance mark-resight estimates. Each deer observed during surveys was recorded as mark ID#, unmarked, or unidentified mark. I used program MARK (White and Burnham 1999) applying the immigration-emigration mixed logit-normal model (McClintock et al. 2008) to estimate mule deer abundance and confidence intervals. For mark-resight model evaluations, I examined all parameter combinations of varying detection rates with survey occasion or effort (vary P with survey or effort), evaluating population size as equal or varied among surveys (a. = 0 or -=t= 0), and whether individual sighting probabilities (i.e., individual heterogeneity) were constant or varied (cl= 0 or -=t= 0). Model selection procedures followed the information-theoretic approach of Burnham and Anderson (2002). RESULTS AND DISSCUSSION -...I ~ '-' I._ ~ ~ Deer Captures and Survival The capture crew captured 253 fawns in early December 2008 and 150 does in late Februaryearly March 2009. Three fawn and Odoe mortalities occurred during capture and O fawn and 5 doe mortalities occurred during the myopathy period IO days post-capture. Fawn survival during mid-December 2008-late March 2009 varied from 0.688 (Yellow Creek) to 0.925 (Ryan Gulch; Fig 2., Table I). Fawn survival rates differed (P < 0.05) between the Ryan Gulch and Yellow Creek Study areas (Table I). Adult female survival mid-December 2008-late June 2009 varied from 0.762 (North Ridge) to 0.931 (Magnolia; Fig I), but were not different (P > 0.05) among study areas (Table I). Smaller sample sizes for adult females reduced my ability to detect differences relative to fawns, but the apparent lower survival of North Ridge females was partly due to 2 mortalities that occurred during early winter before the March capture effort when only 12-13 marked females were available in each study area. Overall, fawn survival was high during the period examined likely due to the mild winter conditions present through late March, and doe survival was consistent with other mule deer populations experiencing normal winter conditions in the western US (Unsworth et al. 1999). ~ 116 ~ �.__ .... Seasonal Movement Patterns Mule deer migration patterns during 2008 varied among study areas and within the Magnolia study area. North Ridge and north Magnolia deer migrated cast-west typically across US Highway 13; Yellow Creek, Ryan Gulch, and south Magnolia deer migrated south-north summering along the Roan Plateau; and Story/Sprague deer typically migrated relatively short distances south----"north (Fig 3.) . Although summer and winter ranges differed among study areas, distance and timing of migration was similar among 4 of the 5 study areas. Excluding the Story/Sprague study area, median date of spring migration occurred May 17, 2008 (all 4 study areas) and fall migration occurred from October 17-23, 2008; median straight-line migration distances ranged between 30.6 and 39.4 km among the 4 study areas. I noted unique migration patterns among Story/Sprague deer where median spring and fall migration occurred April 29 and December 17, 2008, respectively, and median migration distance was 9.6 km . Story/Sprague deer generally spent less time on winter range and required shorter migration distances to achieve their seasonal metabolic requirements. Mule Deer Body Measurements Body measurements of adult female mule deer recorded 27 February-6 March 2009 were typically highest from the Story/Sprague and North Ridge study areas and lowest from the Yellow Creek and Magnolia study areas (Table 2). Parameters most related to mule deer nutritional status (rLIVINDEX derived from rump fat and BCS; Cook et al. 200 l, 2007) suggested mule deer from the Story/Sprague study area were in the best condition and Yellow Creek deer were in the poorest condition. I observed significantly higher rLIVINDEX values (P < 0.05) among Story/Sprague females than females from the other 4 areas, but differences were not significant (P > 0.05) among the other 4 female groups. Early December fawn weights of males and females averaged 36.4 kg (n = 22, SD = 4.5) and 33.5 kg (n = 27, SD= 3.3) from Ryan Gulch, 33.9 kg (n = 22, SD= 3.6) and 30.5 (n = 28, SD= 4.9) from Yellow Creek, 37.0 kg (n = 24, SD= 3.5) and 33.5 kg (n = 26, SD= 4.0) from Magnolia, 35.8 kg (n = 26, SD= 4.8) and 33.2 kg (n = 24, SD= 3.0) from Story/Sprague, and 35.2 kg (n = 20, SD= 4.3) and 33.9 kg (n = 30, SD= 4.2) from North Ridge. Female fawns from Yellow Creek were significantly lighter (P < 0.05) than female fawns from the other 4 areas and Yellow Creek male fawns were significantly lighter than male fawns from Magnolia (P = 0.0 l 0). Mule Deer Population Estimates Mark-resight models that best predicted abundance estimates (lowest AICc; Burnham and Anderson 2002) exhibited constant population size across surveys (i.e., a= 0 suggesting population closure) and homogenous individual sightability (cr2 = 0) for all study areas, and variable sightability (P) across surveys in Ryan Gulch and Magnolia or with survey effort in North Ridge, Story/Sprague, and Ycllow Creek. North Ridge exhibited the highest deer density ( 18.1 /km2) and Ryan Gulch and Magnolia exhibited relatively low deer densities (5.6 and 6.6/km2; Table 3). ..., Abundance estimates were similarly precise from 4 of the 5 study areas (mean CV = 0.16-0.18), with Story/Sprague exhibiting the widest Cls (Table 3; mean CV = 0.29). A relatively small number of marked deer were sighted during surveys (Table 3) suggesting improved precision can be accomplished with increased sample sizes or increasing the number of surveys/study area. Increasing the number of marks/study area by 30 can easily be accomplished by extending GPS drop-off dates beyond the March capture period, which wasn't the case last winter. I also noted that complete coverage of each study area can reasonably be accomplished by increasing flight time by about 20 to 60 minutes/survey depending on the study area and should be more cost effective than increasing number of surveys/area. By increasing the number of marks and complete survey coverage/study area, CV s should improve likely providing detection of <30% change in population size. 117 �SUMMARY AND FUTURE PLANS The goal of this study is to investigate habitat treatments and energy development practices that enhance mule deer populations exposed to extensive energy development activity. The infonnation presented here provide data describing mule deer population parameters from the -first pre-treatment year of a long-term study intended to address how mule deer react to landscape scale habitat and human • activity modifications. The pretreatment period is intended to continue I to 2 more winters to provide baseline data to compare against intended improvements in habitat conditions and concentration/reduction in human development activities, which will be maintained for at least 5 years to provide sufficient time to measure how deer respond to these changes. Based on the data collected thus far, Story/Sprague deer appear to be in better physical condition than deer from the other winter ranges examined. Migration patterns were similar among 4 of the 5 areas, but Story/Sprague deer traveled shorter distances and spent less time on winter range. Yellow Creek fawns were lighter than other study areas and exhibited lowest survival of the areas investigated. North Ridge deer exhibited the highest winter range density and Magnolia and Ryan Gulch deer exhibited the lowest densities. Reasons for these differences are currently unknown, but could be related to several factors including relative habitat conditions, duration on and distance to seasonal ranges, and extent of human activity throughout occupied habitats. Meaningful comparisons will be evident once treatments are implemented and comparisons arc possible between areas that are manipulated (treatment areas) and those that are not (control areas). We are currently working towards a habitat improvement plan and identifying beneficial development practices that are both logistically and financially feasible to implement. Investigations of habitat treatment potential arc promising in the Magnolia and Ryan Gulch study areas and we expect positive native plant responses with potential acceleration of response through native seeding. Members of CDOW, BLM, and private consultants will be developing a habitat treatment plan for review and approval by the end of the year. Discussions with Williams Production LMT Co. have produced a clustered development plan to be implemented in the Ryan Gulch study area and new technologies will be implemented to reduce human activity through remote monitoring of well pads and fluid collection systems. I recently contracted with Dr. Terry Bowyer and Patrick Lendrum (MS candidate) of Idaho State University to begin a graduate project addressing mule deer migration and potential influences of human activity along migration routes. I collaborated with Chad Bishop this past winter/spring to test a new VIT design that improves VIT retention (see Bishop 2009) and will improve our ability to address neonate survival (in addition to overwinter survival) and identify fawning habitat on summer range; these factors arc not currently being addressed, but could strengthen our inference about mule deer and energy development if funding and cooperative agreements were developed for this purpose. We are beginning to work collaboratively with ExxonMobile Production Co. and Colorado State University to enhance funding and potentially provide graduate student assistance addressing additional components of mule deer/energy development interactions. Additional funding and cooperative agreements will be necessary to manipulate habitat conditions to benefit mule deer and our current funding sources will need to be maintained to continue monitoring mule deer population parameters at the current level. We optimistically anticipate the opportunity to work cooperatively toward developing solutions for allowing the nation's energy reserves to be developed in a manner that benefits wildlife and the people who value both the wildlife and energy resources of Colorado. LITERATURE CITED Anderson, C.R., Jr., and D. J. Freddy. 2008a. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Final Study Plan, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and D. J. Freddy. 2008b. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and 118 �...,,_ -...I ...t -...I habitat degradation-Stage I, Objective 5: Patterns o(.mule deer distribution & movements. Pilot Study, Colorado Division of Wildlife, Ft. Collins, CO, USA. Bartmann, R. M. 1975. Piccance deer study-population density and structure. Job Progress Report, Colorado Divison of Wildlife, Fort Collins, Colorado, USA. Bartmann, R. B., and S. F. Steinert. 1981. Distribution and movements ·of mule deer in the White River. Drainage, Colorado. Special Report No. 51, Colorado Division of Wildlife, Fort Collins, Colorado, USA. Bartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monograph No. 121 . Barrett, M.W., J.W. Nolan, and L.D. Roy. 1982. Evaluation ofa hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10: I 08-114. Bishop, C. J. 2009. Effectiveness of a modified vaginal implant transmitter for capturing mule deer neonates from targeted adult females. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA . Burnham, K. P., and D. R. Anderson. 2002. Model selection and multi-model inference: a practical information-theoretic approach. Second edition. Springer-Verlag, New York, New York, USA. Cook, R. C., J. G. Cook, D. L. Murray, P. Zager, B. K. Johnson, and M. W. Gratson. 2001. Development of predictive models of nutritional condition for rocky mountain elk. Journal of Wildlife Management 65:973-987. Cook, R. C., T. R. Stephenson, W. L. Meyers, J. G. Cook, and L.A. Shipley. 2007. Validating predictive models of nutritional condition for mule deer. Journal of Wildlife Management 71: 1934-1943. Gibbs, H. D. 1978. Nutritional quality of mule deer foods, Piceance Basin, Colorado. Thesis, Colorado State University, Fort Collins, Colorado, USA. Kaplan, E. L., and P. Meier. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 52:457-481. McClintock, B. T., G. C. White, K. P. Burnham, and M. A. Pride. 2008. A generalized mixed effects model of abundance for mark-resight data when sampling is without replacement. Pages 271289 in D. L. Thompson, E.G. Cooch, and M. J. Conroy, editors, Modeling demographic processes is marked populations. Springer, New York, New York, USA. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. C. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Stephenson, T. R., V. C. Bleich, B. M. Pierce, and G. P. Mulcahy. 2002. Validation of mule der body composition using in vivo and post-mortem indicies of nutritional condition. Wildlife Society Bulletin 30:557-564. Stephenson, T. R., K. J. Hundertmark, C. C. Swartz, and V. Van Ballenberghc. I 998. Predicting body fat and mass in moose with untrasonography. Canadian Journal of Zoology 76:717-722. Unsworth, J. W., D. F. Pack, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. Van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked individuals. Bird Study 46: 120-139. Prepared by_ _ _ _ _ _ _ _ _ _ _ _ _ _ __ Chuck R. Anderson, Wildlife Researcher ...I 119 �Table I. Survival rate estimates (S} of fawn ( 14 Dec. 2008-21 Mar. 2009) and adult female ( 14 Dcc.20 June, 2009) mule deer in 5 winter range study areas of the Piceance in northwest, Colorado. ',.'--'·~ ~ Cohort Study area ~ Initial sample size (n) March doe samplea (n) S(95% CI) ....., ~ ~ Fawns ~ Ryan Gulch 54 0.925 (0.853-1.000) -..., Yellow Creek 43 0.688 (0.546---0.839) ....., Magnolia 50 0.800 (0.688-0.911) Story/Sprague 47 0.823 (0.722-0.937) North Ridge 48 0.833 (0.728-0.939) ~ --,' __, ....., ~ ....., Adult females Ryan Gulch 12 28 0.893 (0. 778-1.000) Yellow Creek 13 28 0.890 (0.737-1.000) Magnolia 12 29 0.931 (0.839-1.000) Story/Sprague 13 29 0.862 (0.737-0.988) North Ridge 13 30 0. 762 (0.536-0.960) 3 Adult female sample size following capture and radio-collaring efforts late February-early March, 2009. 120 ~ �l l l l l f l l l l l l l l l l l l l l l l f l l l l l l l l l l l l l l l l f l l l l ( l t Table 2. Mean body measurements, Body Condition Score (BCS), and an index of relative nutritional status (rLIVINDEX) of adult female mule deer from 5 study areas in the Piceance Basin of northwest Colorado, late February-early March, 2009. Sample sizes= 30/study area and values in parentheses = SD. Study Area Weight (kg) Hind foot length (cm) Chest girth (cm) Loin depth (mm) Rump fat (mm) BCSa rLIVINDEXb Ryan Gulch 52.2 (5.7) 46.9 (1.8) 96.9 (4.1) 40.50 (3.03) 1.73 (1.78) 2.66 (0.55) 2.71 (0.68) Yellow Creek 52.9 (4.8) 47.2 (1.1) 97.4 (4.2) 40.17 (2.95) 1.47 (0.68) 2.50 (0.60) 2.51 (0.63) Story/Sprague 55.6 (5.7) 47.2 (1.2) 96.0 (4.1) 40.70 (3.72) 1.97 (1.00) 3.09 (0.72) 3.12 (0.77) Magnolia 55.3 (5.9) 47.7 (1.5) 87.5 (5.0) 40.53 (3.70) 1.30 (0.79) 2.56 (0.68) 2.57 (0.70) North Ridge 53.3 (5.6) 47.3 (3.3) 97.2 (4.9) 41.13 (2. 70) 1.57 (1.22) 2.60 (0.56) 2.62 (0.60) 11 Body condition score taken from palpations of the rump (Cook et al. 200 I) brLIVEINDEX = (cm rump fat - 0.2) + BCS if rump fat > 2 mm. Otherwise = BCS (Cook et al. 2001, 2007). 121 �Table 3. Mark-resight abunda nce (N) and density estimates of mule deer rrom 5 winter range he rd segments in the Piceance Basin. northwest Colorado, 25 March- 2 April. 2009. Data represent 4 s urveys r·rom Ryan Gulc h and 5 s urveys from the other 4 study areas. Study area Mean No. sighted Mean No. marked N (95°/ti Cl) Dens ity (deer/km") Rya n Gulc h Yellow Creek Magnol ia S tory/Sprague North Ridge 156 138 109 138 238 12 7 6 5 14 727 (626- 854) 720 (605- 870) 854 (7 16- 1,027) I, 125 (853-1,509) 1,028 (874- 1,230) 5.6 10 .3 6.6 12.4 18. 1 - - Well Pads & Study Area Boundaries CJ Ryan Gulch CJ Story_ Spra_gue Gulch Magnolia Nor1h Ridge Yellow Creek ,t Active well pads Figure I. Approxi mate study area boundaries relati ve to active natural gas well pads and ene rgy development fac ilities in the Piceance Bas in of northwest Colorado, spring 2009. 122 - �Fawn Survival, Dec. 2008-Mar. 2009 1.000 -· 1s= - - - . . 0.900 '------~---· .:~~-~~~-~~~~ ~,.,,;;;::- , 0.800 ! 0.700 ~ 0.600 ~ 0.500 ! - - ~ - - - - - - - - - - - - - - - ·; 0.400 ~ 0.300 0.200 0.100 0.000 0 ,-... M 0 r-,. v M r;- ~ M DO ,-... v rl I"') C: N .-t N N N I ~ ~ I 'f"'-1 ""'"4 NI ro ¢ """' 00 L/'I N o:::t co LI) .0 '- 00 u, I """' M '""' -, ""'"4 N .-t N ro '- 'f"'-1 N <l,.I ..0 00 C: u u N (,;J C """' !:: C: u. (l,I .0 .0 ~ ro '- .. ~ I (LI (LI 0 0 I u (LI -Ryan Gulch -story/Sprague Magnolia North Ridge I ~ I -Yellow Creek I I .... ....ro .... .... "O u. !ti (l) u. ~ (LI LL 0 C"J ~ Adult female survival, Dec. 2008-June 2009 a:g~~ f-,.._ _ -----·-=,~.......:·~~~'-====::::!llliilii---......·~·-··=•"··=·":-·. -Yellow c--eek 0.800 ! 0.700 ~ 0.600 ~ 0.500 ·:;; 0.400 ~ 0.300 0.200 0.100 0.000 -Ryan Gulch -story/SpragJe ... , Maenolia North Ridge 0 N M C ':-4 -J .... ro I u 00 a., N 0 u a., Cl r-.. ~ rl rt'l ,-t I LI') ~ N I .-, .-, '1 ~ I w ..0 ~ co -.J N I r'i N .0 M 00 N I rl LI) M N I OJ f'-1 I... N Lf) I.... r.J \ 0 ....re ....re LL CJ ~ ~ < ('J LL I O'I I N') m N I u: (l. (.il > r, C M r3 M '- ~ >· I Q. r3 'I"'" <t ~ ~ C I"- -.J 'I"'" " C: > -= (t ~ -·-~-·-•---·----·~-- Figure 2. Winter survival rates of fawn ( 14 December, 2008-21 March, 2009; top) and adult female ( 14 December-21 June, 2009; bottom) mule deer from 5 study areas in the Piceance Basin of northwest Colorado. Survival rates of Yellow Creek fawns were significantly lower (P < 0.05; Table 1) than survival of Ryan Gulch fawns. Survival rates among other fawn and doe groups were not significantly different (P > 0.05; Table 1). 1 123 �\,;.;;I ~....,,~" ' .. •• ....;..~... -· .,,,:... ~ -!.._t..f · i · .• •~. • l ~I ',·!;'~.('/~~... • ~ .... ~ ~ '. .. ;; -, ,( • - •.,, ,. • ·.:.,,;, . > , ~.--:... •• · '\ ,., - ... "'• <\ I ~ '\-1• :-~ .."'i..,... I' •. '· , ·, ... ...... ........ --., Plus= Yellow Creek Diamond= Ryan Gulch Circle= North Ridge Star= Magnolia Triangle= Stor1/Sprague Figure 3. Mule deer GPS locations from 5 winter range study areas (solid lines; 15 does/study area) in the Piceance Basin of northwest Colorado, January 2008- Fcbruary, 2009. -..... - - 124 �Colorado Division of Wildlife July 2009-June 2010 w .,,_,, - WlLDLIFE RESEARCH REPORT State of_ _ _ _ _ _..:::C=o=lo=r=ad =o"------- : ~D_i_vi_s_io_n_o_f_W_il_d_liii _e_ _ _ _ _ _ _ _ __ Cost Center 3430 : "'"'M""'a=m ==a= m ls~R=e=s=e=a= rc~h~ - - - - - - - - Work Package 3001 : =D~e~e~ r ~C~o~n~se=r_v=at~io~n_ _ _ _ _ _ _ _ __ Task No. 6 : Population Performance of Piceance Basin Mule Deer in Response to Natural Gas Resource Extraction and Mitigation Efforts to Address Human Activity and Habitat Degradation Federal Aid Proj ect: __W-'-'---=18=5~-R=-=-------Period Covered: July I, 2009 - June 30, 20 I 0 Authors: C.R. Anderson and C. J. Bishop 1w - Personnel: E. Bergman, J. Broderick, 8. deVergie, D. Finley, L. Gepfert, M. Grode, K. Kaai, T. Knowles, J. Lewis, P. Lukacs, K. Maysilles, M. Sirnchman, T. Swearingen, R. Velarde, S. Wilson, L. Wolfe, CDOW; E. Hollowed, BLM; S. Monsen, Western Ecological Consulting, Inc.; D. Freddy, Hoch Berg Enterprises; H. Sawyer, Western Ecosystems Technology; P. Lendrum, T Bowyer, Idaho State University; P. Doherty, G. Wittemyer, K. Wi lson, G. Wh ite, Colorado State University; M. Keech, L. Shelton, M. She lton, R. Swisher, Quicksilver Air, Inc.; D. Felix, Olathe Spray Service, lnc.; L. Coulter, Coulter Aviation. Project support received from Federal Aid in Wildlife Restoration, Colorado Mule Deer Association, Colorado Mule Deer Foundation, Colorado State Severance Tax Fund, EnCana Corp., Shell Petroleum, and Williams Production LMT Co. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the authors. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We propose to experimentally evaluate habitat treatments that may improve the landscape to benefit mule deer (Odocoileus hemionus) and evaluate human-activity management alternatives to reduce the disturbance of energy development impacts on mule deer. The Piceance Basin of northwestern Colorado was selected as the project area due to ongoing natural gas development in one of the most extensive and important mule deer winter and transition range areas within the state. The data presented here represent the first 2 pretreatment years of a long-term study addressing habitat modifications and improved energy development practices intended to improve mule deer fitness in areas exposed to extensive energy development. We modified the previous study design to monitor 4 winter range study areas representing varying levels of development to serve as treatment (Ryan Gu lch, North Magnolia, South Magnolia) and control (North Ridge) sites and recorded habitat use and movement patterns using GPS collars (5 locations/day), estimated overwinter fawn and annual adult female survival, estimated early and late winter body condition of adu lt females using ultrasonography, and estimated abundance using helicopter mark-resight surveys. We attached 250 VHF collars (50- 80/study area) to fawns and 80 VHF collars to does (20/study area) in early December 2009 and I 00 GPS collars (25/study area) to adult fema le mule deer in early March 20 l 0. Based on the data collected thus far, deer from all areas appear to 47 �be in reasonably good condition and are exhibiting high survival rates. Mild winter conditions the past 2 years certainly contributed to the observed mule deer popu lation parameters. lt will be informative to note how the different wintering mule deer herd segments react following a severe winter. Observed differences in winter concentration areas thus far may indicate behavioral modifications to areas of high development activity, but resource selection analyses will be necessary to confinn this supposition. We will continue to collect the various population and habitat use data across all study sites to evaluate the effectiveness of the habitat treatments scheduled to begin this fa ll. This approach will allow us to determine whether it is possible to effectively mitigate development impacts in highly developed areas, or whether it is better to allocate mitigation dollars toward less-impacted areas. We may also find that habitat mitigation efforts are not effective in developed areas at all, suggesting that habitat enhancement efforts may be only effective in areas that are not impacted by development. We are also evaluating deer behavioral responses to varying levels of development activity and habitat mitigation treatments. Thjs will allow us to assess the effectiveness of certain Best Management Practices (BMPs) and habitat manipulations for reducing disturbance to deer. The study is slated to run through at least 2015, and preferably 201 8, to adequately measure deer population responses to landscape level manipulations. 48 �WILDLIFE RESEARCH REPORT POPULATION PERFORMANCE OF PICEANCE BASIN MULE DEER IN RESPONSE TO NATURAL GAS RESOURCE EXTRACTION AND MITIGATION EFFORTS TO ADDRESS HUMAN ACTIVITY AND HABITAT DEGRADATION CHARLES R. ANDERSON, JR and CHAD J. BISHOP PROJECT NARRITIVE OBJECTIVES l. To determine experimentally whether enhancing mule deer habitat conditions on winter and/or transition range elicits behavioral responses, improves body condition, increases overwinter fawn survival, or ultimately, population density on mule deer winter ranges exposed to extensive energy development. 2. To determine experimentally to what extent modification of energy development practices enhance habitat selection, body condition, over-winter fawn survival, and winter range mule deer densities. ""1tsl SEGMENT OBJECTIVES 1. Collect and reattach GPS collars (5 location attempts/day) to maintain sample sizes for addressing mule deer habitat use and behavior patterns in 4 study areas experiencing varying levels of energy development of the Piceance Basin, northwest Colorado. 2. Estimate early and late winter body condition of adult female mule deer in each of the 4 winter herd segments 3. Monitor over-winter fawn and annual adult female mule deer survival by daily ground tracking and biweekly aerial tracking. 4. Conduct Mark-Resight helicopter surveys to estimate mule deer abundance in each study area. 5. Develop cooperative agreements to initiate habitat treatments for assessing efficacy of habitat improvement projects to mitigate energy development disturbances to mule deer. 6. Summarize data and present information in an annual Job Progress Report. INTRODUCTION Extraction of natural gas from areas throughout western Colorado has raised concerns among many public stakeholders and the Colorado Division of Wildlife that the cumulative impacts associated with this intense industrialization will dramatically and negatively affect the wildlife resources of the region. Concern is especially high for mule deer due to their recreational and economic importance as a principal game species and their ecological importance as one of the primary herbivores of the Colorado Plateau Ecoregion. Extraction of natural gas will directly affect the potential suitability of the landscape used by mule deer through conversion of native habitat vegetation with drill pads, roads, or noxious weeds, by fragmenting habitat because of drill pads and roads, by increasing noise levels via compressor stations and vehicle traffic, and by increasing the year-round presence of human activities. Extraction will indirectly affect deer by increasing the human work-force population of the region resulting in the need for additional landscape for human housing, supporting businesses, and upgraded 49 '-' �road/transportation infrastructure. Additionally, increased traffic on rural roads will raise the potential for vehicle-animal collisions and additive direct mortality to deer populations. Thus, research documenting these impacts and evaluating the most effective strategies for minimizing and mitigating these activities will greatly enhance future management efforts to sustain mule deer populations for future recreational and ecological values. The Piceance Basin in northwest Colorado contains one of the largest migratory mule deer populations in North America and also exhibits some of the largest natural gas reserves in North America. Projected energy development throughout northwest Colorado within the next 20 years is expected to reach about I 5,000 wells, many of which will occur in the Piceance Basin, which currently supports over 250 active gas well pads (http://cogcc.state.co.us). Anderson and Freddy (2008a) in their long-term research proposal identified 6 primary study objectives to assess measures to offset impacts of energy extraction on mule deer population performance. During the past 3 years, we have gathered baseline habitat utilization data from OPS-collared deer across the Piceance Basin to allow assessment of mitigation approaches that will be implemented over the next 2-3 years and evaluated for another 5-6 years. We initially selected 5 winter range study areas representing varying levels of development to serve as treatment and control sites. The past 2 years, we also estimated winter fawn survival and annual adult female survival, early and late winter body condition of adult females using ultrasonography, and deer abundance using helicopter mark-resight surveys. We started with 5 study sites to allow flexibility to respond to differences in deer behavior and changing energy development plans, which can directly affect experimental design. During the previous year, we refined our study design using our baseline deer data and current energy development plans of the major companies operating in Piceance Basin. We split I study area (Magnolia split into North and South Magnolia) based on differences in deer movement and behavior patterns from GPS data (Anderson 2009) and eliminated 2 other study sites (Story/Sprague and Yellow Creek) due to incompatible deer behavior patterns to adequately serve as control sites and to reduce the annual project budget to the minimum necessary to meet the original research objectives. This progress report describes the previous 2 years of addressing mule deer population performance during the pretreatment phase, which includes monitoring habitat selection and behavior patterns of adult female mule deer, overwinter fawn and adult female survival, estimates of adult female body condition during early and late winter, and abundance estimates on 4 winter range herd segments in relation to varying levels of natural gas development in control and treatment experimental areas prior to proposed experimental modifications in energy developmental practices and potential habitat improvement treatments. STUDY AREAS The Piceance Basin between the cities of Rangely, Meeker, and Rifle in northwest Colorado was selected as the project area due to its ecological importance as one of the largest migratory mule deer populations in North America and because it exhibits one of the highest natural gas reserves in North America (Fig. I). Historically, mule deer numbers on winter range were estimated between 15,00022,000 (Bartmann 1975), and the current number of well pads (Fig. I) and projected number of gas wells in the Piceance Basin over the next 20 years is about 250 and 15,000, respectively. Mule deer winter range in the Piceance Basin is predominantly characterized as a topographically diverse pinion pine (Pinus edulis)-Utah juniper (Juniperus osteosperma; pinion-juniper) shrubland complex ranging from 167 5 m to 2285 m in elevation (Hartmann and Steinert 1981 ). Pinion-juniper are the dominant overstory species and major shrub species include Utah serviceberry (Amelanchier utahensis), mountain mahogany (Cercocarpus montanus), bitterbrush (Purshia tridentata), big sagebrush (Artemisia tridentata), Gamble's oak (Quercus gambelii), mountain snowberry Symphoricarpos oreophilus), and rabbitbrush (Chrysothamnus spp.; Hartmann et al. 1992). The Piceance Basin is segmented by numerous drainages characterized by stands of big sagebrush, saltbush (Atriplex spp.), and black greasewood (Sarcobatus vermiculatus), with the majority of the primary drainages having been converted to mixed-grass hay ~l..,I '-" _, "'-' ~ '--' "w,/ \wl ~ '-" l_,i ~ w' "w,/ '-' lwJ 'w _, ~ I._.J w' ~ \._I ~ ~ w' ~ '_,J _,I '-" ~ \.._.,'- 1._.i 50 ~ l,_.i y �fields. Grasses and forbs common to the area consist of wheatgrass (Agropyron spp. ), blue grama (Bouteloua graci/is), needle and thread (Stipa comata), Indian rice grass (Oryzopsis hymenoides), arrowleafbalsamroot (Balsamorhiza sagittata), broom snakeweed (Gutierrezia sarothreae), pinnate tansymustard (Descurainia pinnata), milkvetch (Astragalus spp.), Lewis flax (Linum lewisii), evening primrose (Oenothera spp.), skyrocket gilia (Gilia aggregata), buckwheat (Erigonum spp.), Indian paintbrush (Castilleja spp.), and penstemon (Penstemon spp.; Gibbs 1978). The climate of the Piceance Basin is characterized by warm dry summers and cold winters with most of the annual moisture resulting from spring snow melt. 'cl Wintering mule deer population segments we investigated in the Piceance Basin include: North Ridge (57 km2)just north of the Dry Fork of Piceance Creek including the White River in the northeastern portion of the Basin, Ryan Gulch ( 130 km2) between Ryan Gulch and Dry Gulch in the southwestern portion of the Basin, North Magnolia (79 km2) between the Dry Fork of Piceance Creek and Lee Gulch in the north-central portion of the Basin, and South Magnolia (83 km2) between Lee Gulch and Piceance Creek in the south-central portion of the Basin (Fig. I). Each of these wintering population segments has received varying levels of natural gas development: no development in North Ridge, light development in North Magnolia (0.13 pads & facilities/km 2), and relatively high development in the Ryan Gulch (0.64 pads & facilities/km 2) and South Magnolia (0.81 pads & facilities/km 2) segments (Fig. 1). Among the 4 study areas, North Ridge will serve as an unmanipulated control site, Ryan Gulch will serve to address human-activity management alternatives (Best Management Practices; BMPs) that may benefit mule deer exposed to energy development, and North and South Magnolia will serve to address the utility of habitat treatments intended to enhance mule deer population performance in areas exposed to light (North Magnolia) and heavy (South Magnolia) energy development activities. METHODS Tasks addressed this fiscal year included mule deer capture and collaring efforts, monitoring overwinter fawn and annual adult female survival, estimating adult female body condition during early and late winter using ultrasonography, and estimating mule deer abundance applying helicopter markresight surveys. We employed helicopter net-gunning techniques (Barrett et al. 1982, van Reenen 1982) to capture 50-80 fawns and 20 adult females during early December and 25 adult females during early March in each of the 4 study areas (250 fawns and 180 does total). Once netted, all deer were hobbled and blind folded. Fawns were weighed, radio-collared and released on site, and adult females were transported to localized handling sites for collection of body measurements and were fitted with VHF (20/area during December) or OPS collars (25/area during March; 5 fixes/day; 0211 OB, Advanced Telemetry Systems, Isanti, MN, USA) and released. To provide direct measures of decline in overwinter body condition, we attempted to capture the same adult females during the March capture that were captured in December. Fawn collars were spliced and fitted with 2 lengths of rubber surgical tubing to facilitate collar drop during mid-summer-early autumn, adult VHF collars were attached static, and OPS collars were supplied with timed drop-off mechanisms scheduled to release early April, 2011. All radiocollars were equipped with mortality sensing options (i.e., increased pulse rate following 4 hrs of inactivity). Mule Deer Habitat Use and Movements We downloaded and organized data from OPS collars deployed March 2009 following collar drop and retrieval in early April 2010. OPS collars redeployed early March 2010 maintained the same fix schedule of attempting fixes every 5 hours. We plotted deer locations and recorded timing and distance of spring and fall 2009 migrations for each study area. Mule deer winter concentration areas were created using composite OPS data (winter locations since January 2008 from all deer) from each study area and mapped in ArcGIS (ver. 9.3) using Spatial Analyst (kernel probability density functions separated by 51 �~ '-' ~ vi quantiles). Mule deer resource selection analyses are pending completion of high resolution habitat data layers currently being developed by BLM (habitat data layers should be available by 2011 ). Mule Deer Survival Mule deer mortality monitoring consisted of daily ground telemetry tracking and aerial monitoring approximately every 2 weeks from fixed-wing aircraft. Once a mortality signal was detected, deer were located and necropsied to assess cause of death. We estimated over-winter survival on a weekly basis using the staggered entry Kaplan-Meier procedure (Kaplan and Meier 1958, Pollock et al. 1989). Capture-related mortalities (any mortalities occurring within IO days of capture) and collar failures were censored from survival rate estimates. We estimated survival rates 28 June 2009-26 June 2010 for adult females and 6 December 2009-27 March 2010 for fawns. Premature failure of surgical tubing integrity beginning late March inhibited our ability to reasonably estimate fawn survival beyond March 27, 2010. Adult Female Body Measurements We applied ultrasonography techniques described by Stephenson et al. ( 1998, 2002) and Cook et al. (2001) to measure maximum subcutaneous rump fat (mm) and loin depth (longissimus dorsi muscle, mm). We estimated a body condition score (BCS) for each deer by palpating the rump (Cook et al. 2001). We combined ultrasound rump fat measurements with BCS to develop an index (rLIVINDEX; Cook et al. 2001, 2007) of the relative nutritional status of deer from each study area. We examined differences (P < 0.05) in nutritional status among study areas using a two-sample I-test. We considered differences in body condition meaningful when either mean rump fat or rLIVINDEX differed statistically between comparisons. Other body measurements recorded included pregnancy status (pregnant, barren) via blood samples, weight (kg), chest girth (cm), and hind-foot length (cm). Abundance Estimates We conducted 5 (North Magnolia) or 4 (the remaining study areas) helicopter mark-resight surveys (2 observers and the pilot) during late March, 2010 to estimate deer abundance in each of the 4 study areas. We delineated each study area from GPS locations during the same period the previous year and aerial telemetry locations ofradio-collared deer within I week of the first mark-resight survey. Aerial fixed-wing telemetry surveys were conducted during helicopter surveys to determine which marked deer were within each survey area. We delineated flight paths in ArcGIS 9.3 prior to surveys following topographic contours (e.g., drainages, ridges) and approximating 500 m spacing throughout each study area; flight paths during surveys were followed using GPS navigation in the helicopter. Two approximately 12 x 12 cm pieces of Ritchey livestock banding material (Ritchey Livestock ID, Brighton, CO USA) were uniquely marked using number, symbol combinations and attached to each radio-collar to enhance mark-resight estimates. Each deer observed during surveys was recorded as mark ID#, unmarked, or unidentified mark. We used program MARK (White and Burnham 1999) applying the mixed logit-normal model (McClintock et al. 2008) to estimate mule deer abundance and confidence intervals. For mark-resight model evaluations, we examined parameter combinations of varying detection rates with survey occasion and whether individual sighting probabilities (i.e., individual heterogeneity) were constant or varied (cr2 = 0 or* 0). Model selection procedures followed the information-theoretic approach of Burnham and Anderson (2002). '~ "--'~ ~ i_,i _, '-' l_,.i \wl 'w '-' '-' \wl \wl 'w .._., '-i,I l._i .I '-"~ '-..,I ~ ',wl .._., vi '-i,I \_,I ~ '-' "--' vi vi '-" 1...,1 1...,1 ~ ~~ I._/ 52 ~ 'w ~ �RES UL TS AND DISSCUSSION -' Deer Captures and Survival The helicopter crew captured 253 fawns and 80 does in early December 2009 and 103 does in early March 2010. Seven fawn (ultimate cause= 4 cougar predation, 2 coyote predation, 1 drowning) and 2 doe mortalities (ultimate cause= tangled in fence and coyote predation) occurred within the IO day myopathy period following the December capture and 3 doe mortalities occurred during the March capture (all direct capture myopathy). Fawn survival during early-December 2009-late March 20 IO was similar among study areas (P > 0.05) ranging from 0.872 (Ryan Gulch) to 0.945 (North Magnolia; Table 1, Fig. 2). Although mean fawn survival was higher than last year among 3 of 4 study areas (with the exception of Ryan Gulch; see Anderson 2009), differences were statistically insignificant. Annual adult female survival was also similar among study areas (P > 0.05) ranging from 0.863 (North Ridge) to 0.943 (North Magnolia; Table I, Fig. I) and were comparable to last year (P > 0.05; Anderson 2009). The relatively high fawn survival observed the past 2 winters is likely due to the mild winter conditions present through late March, and doe survival was consistent with other mule deer populations experiencing normal winter conditions in the western US (Unsworth et al. 1999). 'el Seasonal Movement Patterns Migration patterns differed among areas with North Ridge and North Magnolia deer migrating east-west and South Magnolia and Ryan Gulch deer migrating south-north (Fig. 3). Median straight-line migration distances were similar ranging from 32.6 km (Ryan Gulch) to 40.1 km (North Ridge). Similar to seasonal ranges, most deer monitored exhibited strong fidelity to spring and fall migration routes (Fig. 3). Timing of mule deer migration during 2009 was similar among study areas with median spring migration dates occurring between 15 and 20 May and median fall migration dates occurring between 15 and 22 October. Migration dates were later compared to last year (Anderson 2009), occurring 8 to 16 days later in the spring and 11 to 14 days later in the fall. Length of migration was relatively short among areas averaging 5 to IO days in the spring and 4 to 7 days in the fall; these observations were comparable to last year. More detailed analyses of these migration data investigating the influence of human activity are currently being conducted by Patrick Lendrum and Terry Bowyer of Idaho State University. A final report including next year's migration data is scheduled to be completed by spring 2012. Winter concentration areas identified from January 2008-May 2010 (Fig. 4) reasonably followed study area boundaries delineated from deer locations applied the first winter of the project (Anderson and Freddy 2009b). We noted more continuous distributions from Ryan Gulch and North Ridge deer, with South Magnolia deer exhibiting the most fragmented and concentrated distributions, which may be related to relative development densities within each study area. Future resource selection analyses will address these differences relative to habitat attributes within each area. Minor modifications to study area boundaries will be applied in the future to better address winter deer use within each study area (Fig. 4). Mule Deer Body Condition Body condition measurements of adult female mule deer suggested that North and South Magnolia deer returned from summer range (December 2009) in better condition than North Ridge deer (P < 0.05) and condition of Ryan Gulch deer was intermediate and not significantly different (P > 0.05) from the other areas (Table 2). North and South Magnolia deer maintained relatively high body condition over winter, but only North Magnolia deer were in significantly better condition than deer from North Ridge and Ryan Gulch (P < 0.05; March 20 I 0, Table 2) by late winter. Paired comparisons of deer captured during December 2009 and March 20 IO indicted that mean rump fat and % body fat declined 8.3 mm and 6.9% in North Magnolia (n = 15), 8.1 mm and 6.9 % in South Magnolia (n = 16), 3.1 mm and 53 �4.0% in North Ridge (n = 16), and 6.3 mm and 6.6% in Ryan Gulch (n = 19). In comparing late winter body condition from 2009 to 2010, we noted significant improvement from North and South Magnolia deer and similar condition from North Ridge and Ryan Gulch deer. Pregnancy rates were expectedly high ranging from 84% in Ryan Gulch (n = 25) to 100% in South Magnolia (n = 25). Early December fawn weights of males and females averaged 39.5 kg (n = 30, SD= 4.3) and 36.5 kg (n = 30, SD= 3.2) from North Magnolia, 38.5 kg (n = 42, SD= 3.8) and 35. l (n = 18, SD= 4.0) from South Magnolia, 37.5 kg (n = 33, SD= 4.0) and 34.9 kg (n = 50, SD= 4.3) from North Ridge, and 37. l kg (n = 23, SD= 3.3) and 34.5 kg (n = 27, SD= 3.4) from Ryan Gulch. Fawn weights were similar among areas except that male and female fawns from North Ridge were larger than Ryan Gulch fawns (P < 0.05). Because North and South Magnolia study areas were not split until December 2009 and fawn locations were not sufficiently monitored prior to that time, comparisons to 2008 fawn weights were only possible by combining data from North and South Magnolia in 2009. Both males and females from the combined Magnolia area were larger during December 2009 than December 2008. Fawn weights from the other study areas were similar between years expect for males from North Ridge, which were also larger in 2009 (P = 0.047). \..._.,"-' i._.i ~ ~ ~ i_, ~ "-' ~ ~ ~ Mule Deer Population Estimates Mark-resight models that best predicted abundance estimates (lowest AICc; Burnham and Anderson 2002) exhibited homogenous individual sightability (cr2 = 0) for all study areas and variable sightability (P) across surveys in 3 of the 4 study areas; sightability was consistent across surveys in North Magnolia. North Ridge exhibited the highest deer density (20.1/knl) and comparably lower deer densities were observed in the other 3 areas (6.9-9.3/km2; Table 3). Abundance estimates were similar to last year (Anderson 2009) except in Ryan Gulch where deer numbers were significantly higher this year. It is unlikely deer abundance increased from 825 (95% CI= 672-1,016) to 1,442 (95% CI= 1112-1878) in 1 year, and we suspect this difference may be partially due differences in sampling approach between years. The abundance estimate from 2009 was derived from subsampling 20 to 40% of the Ryan Gulch study area (Anderson 2009), whereas the 2010 estimate was based on complete sampling of the entire study area. It is plausible that subsampling the study area resulted in a negative bias and we are more comfortable with the 2010 estimate derived from complete coverage of the study area. ~ 'w 'w' ~ I._,.; 41 ~l_,I ~ ~ '._.i Abundance estimates from 2010 were similarly precise from 3 of the 4 study areas (mean CV= 0.16-0.18), with Ryan Gulch exhibiting a relatively wide CI (Table 3; mean CV= 0.27). Number of marked deer was lowest from Ryan Gulch (n = 87) and increasing sample size would improve future estimates, as would increasing the number of mark-resight surveys. Additionally, winter concentration information from the past 3 winters (Fig. 4) can be used to more efficiently focus sampling effort potentially increasing mule deer sightability. Our goal is to achieve CVs of :S0.15 to allow detection of at least 30% population change. We will attempt to improve precision of future mark-resight abundance estimates by increasing sample size using VHF radiocollars and increasing the number of surveys when feasible; simulations suggest CVs can be improved by about 0.02 for each additional mark-resight survey (C. Anderson, unpublished data). ~ 'w vi ~ ~ ~ ~ ~ SUMMARY AND FUTURE PLANS The goal of this study is to investigate habitat treatments and energy development practices that enhance mule deer populations exposed to extensive energy development activity. The information presented here provide data describing mule deer population parameters from the first 2 years of the pretreatment period of a long-term study intended to address how mule deer react to landscape scale habitat and human activity modifications. The pretreatment period is intended to continue 1 to 2 more winters to provide baseline data to compare against intended improvements in habitat conditions and evaluation of concentration/reduction in human development activities, which will be maintained for at least 5 years to ~ w' _, - 'w ~~ '_,.I 54 '-" ~ '- �-.,I ..._, provide sufficient time to measure how deer respond to these changes. Based on the data collected thus far, deer from all areas appear to be in reasonably good condition and are exhibiting high survival rates. Mild winter conditions the past 2 years certainly contribute to the observed mule deer population parameters. It will be informative to note how the different wintering mule deer herd segments react following a severe winter. Observed differences in winter concentration areas (Fig. 4) may indicate behavioral modifications to areas of high development activity, but resource selection analyses will be necessary to confirm this supposition. We will continue to collect the various population and habitat use data across all study sites to evaluate the effectiveness of the habitat treatments. This approach will allow us to determine whether it is possible to effectively mitigate development impacts in highly developed areas, or whether it is better to allocate mitigation dollars toward less-impacted areas. We may also find that habitat mitigation efforts are not effective in developed areas at all, suggesting that habitat enhancement efforts may be only effective in areas that are not impacted by development. In a recent project conducted on the Uncomphahgre Plateau, Bergman et al. (2009) found that habitat treatments implemented in pinyon-juniper habitat in undeveloped areas were effective for deer. We are also evaluating deer behavioral responses to varying levels of development activity and habitat mitigation treatments. This will allow us to assess the effectiveness of certain BMPs and habitat manipulations for reducing disturbance to deer. We recently developed a habitat improvement plan and intend to begin implementation this fall with completion by fall 2011 if feasible or fall 2012 in the Magnolia study areas. In addition, hay field improvements have begun and will continue in the North Magnolia area and we plan to begin discussions addressing hay field improvements in the South Magnolia study area. Recent collaboration agreements with ExxonMobil Development Co. and Colorado State University will provide graduate research opportunities to enhance data collection and inference about mule deer/energy development interactions. Collaboration with Williams Production LMT Co. have produced a clustered development plan to be implemented in the Ryan Gulch study area and new technologies will be implemented to reduce human activity through remote monitoring of well pads and fluid collection systems. We are continuing to work with Dr. Terry Bowyer and Patrick Lendrum (MS candidate) of Idaho State University to address mule deer migration and potential influences of human activity along migration routes. Additional funding and cooperative agreements will be necessary to sustain this project through completion (through at least 2015 and preferably through 2018). We optimistically anticipate the opportunity to work cooperatively toward developing solutions for allowing the nation's energy reserves to be developed in a manner that benefits wildlife and the people who value both the wildlife and energy resources of Colorado. LITERATURE CITED \sl cl Anderson, C.R., Jr. 2009. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and D. J. Freddy. 2008a. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Final Study Plan, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and D. J. Freddy. 2008b. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation-Stage I, Objective 5: Patterns of_mule deer distribution & movements. Pilot Study, Colorado Division of Wildlife, Ft. Collins, CO, USA. Bartmann, R. M. 1975. Piceance deer study-population density and structure. Job Progress Report, Colorado Divison of Wildlife, Fort Collins, Colorado, USA. Bartmann, R. B., and S. F. Steinert. 1981. Distribution and movements of mule deer in the White River Drainage, Colorado. Special Report No. 51, Colorado Division of Wildlife, Fort Collins, Colorado, USA. 55 �'._,I i,_i '-" _.i Bartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monograph No. 121. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation ofa hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10: 108-114. Bergman, E. J., C. J. Bishop, D. J. Freddy, and G. C. White. 2009. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, USA. Burnham, K. P., and D.R. Anderson. 2002. Model selection and multi-model inference: a practical information-theoretic approach. Second edition. Springer-Verlag, New York, New York, USA. Cook, R. C., J. G. Cook, D. L. Murray, P. Zager, B. K. Johnson, and M. W. Gratson. 2001. Development of predictive models of nutritional condition for rocky mountain elk. Journal of Wildlife Management 65:973-987. Cook, R. C., T. R. Stephenson, W. L. Meyers, J. G. Cook, and L. A. Shipley. 2007. Validating predictive models of nutritional condition for mule deer. Journal of Wildlife Management 71 : 1934-1943. Gibbs, H. D. 1978. Nutritional quality of mule deer foods, Piceance Basin, Colorado. Thesis, Colorado State University, Fort Collins, Colorado, USA. Kaplan, E. L., and P. Meier. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 52:457-481. McClintock, B. T., G. C. White, K. P. Burnham, and M.A. Pride. 2008. A generalized mixed effects model of abundance for mark-resight data when sampling is without replacement. Pages 271289 in D. L. Thompson, E.G. Cooch, and M. J. Conroy, editors, Modeling demographic processes is marked populations. Springer, New York, New York, USA. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. C. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Stephenson, T. R., V. C. Bleich, B. M. Pierce, and G. P. Mulcahy. 2002. Validation of mule deer body composition using in vivo and post-mortem indices of nutritional condition. Wildlife Society Bulletin 30:557-564. Stephenson, T. R., K. J. Hundertmark, C. C. Swartz, and V. Van Ballenberghe. 1998. Predicting body fat and mass in moose with untrasonography. Canadian Journal of Zoology 76:717-722. Unsworth, J. W., D. F. Pack, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. Van Reenen, G. 1982. Field experience in the capture ofred deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-421 in L. Nielsen, J.C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked individuals. Bird Study 46: 120-139. \..v',.-1 ~ ~ \._I "-' \wl i_,i 'w ~ _, ~ ~ ~ _, ~ w ~ ~ -~ - '-' ._,, w '._,I ~ '..-' ~ _.I I-, ~ Prepared by _ _ _ _ _ _ _ _ __ Chuck Anderson, Wildlife Researcher ~ ~ ~ '..i ~ '-' '--"~ ~ 56 'w ~ "-' �Table I. Survival rate estimates (S) of fawn (6 Dec. 2009-27 Mar. 2010) and adult female (28 June 2009-26 June 20 l 0) mule deer from 4 winter range study areas of the Piceance Basin in northwest Colorado. Cohort Study area Initial sample size (n) March doe samplea (n) S(95% Cl) Fawns Ryan Gulch 47 0.872 (0. 777-0.968) South Magnolia 63 0.937 (0.876-0.997) North Magnolia 55 0.945 (0.884-1.000) North Ridge 80 0.912 (0.849-0.974) Adult females Ryan Gulch 25 47 0.868 (0.757-0.979) South Magnolia 12 38 0.873 (0. 757-0.989) North Magnolia 14 44 0.943 (0.866--1 .000) North Ridge 27 50 0.863 (0.748-0.978) ""-' ., aAdult female sample size following capture and radio-collaring efforts early March, 2010. vu \:sl 57 �Table 2. Mean rump fat (mm), Body Condition Score (BCS)3, and an index of relative nutritional status (rLIVINDEX/ of adult female mule deer from 4 study areas in the Piceance Basin of northwest Colorado, March and December 2009 and March 2010. Values in parentheses= SD. December 2009 March 2009 March 2010 rLIVINDEX Rump fat BCS rLIVINDEX Rump fat 1.73 (1.78) 2.66 (0.55) 2.71 (0.68) 8.35 (6.36) 4.06 (1.13) 4.71 (1.63) 2.31 (1.44) 2.35 (0.48) 2.41 (0.57) South Magnolia 1.47 (0.68) 2.50 (0.60) 2.51 (0.63) 10.05 (6.19) 4.07 (1.21) 4.87 (1.75) 3.12 (2.20) 2.64 (0.59) 2.78 (0.74) North Magnolia 1.30 (0.79) 2.56 (0.68) 2.57 (0.70) 10.20 (5.48) 4.25 (0.96) 5.07 (1.42) 3.15 (2.34) 2.85 (0.53) 2.99 (0.70) North Ridge 1.57 (1.22) 2.60 (0.56) 2.62 (0.60) 5.25 (5.65) 3.63 (1.11) 3.98 (1.59) 1.77 (1.11) 2.42 (0.49) 2.46 (0.54) Study Area Rump fat Ryan Gulch BCS BCS rLIVINDEX aBody condition score taken from palpations of the rump (Cook et al. 2001) brLIVEINDEX = (cm rump fat - 0.2) + BCS if rump fat> 2 mm. Otherwise= BCS (Cook et al. 2001, 2007). 58 ( ( ( ( { ( ( { l { ( ( ( ( { ( ( { ( ( { ( { ( { l ( { { ( ( { ( ( ( C{ ( ( { ( ( l C( { ( �Table 3. Mark-resight abunda nce (N) and density estimates of mule deer from 4 winter range herd segme nts in the Piceance Basin, northwest Colorado, 22- 3 1 March 2009. Data represent 5 resight surveys from North M agno lia and 4 resight surveys from the othe r 3 study areas. Study area Mean No. sighted Mean No. marked N (95% C l) Density (deer/km-) Ryan G ulch 125 II 1,442 (1,112- 1,878) 9.3 South Magnolia 103 18 575 (48 1- 692) 6.9 No1th Magnolia 102 14 595 (498- 715) 7.5 North Ridge 23 1 23 I, 145 (975- 1,348) 20. 1 i.,ftf ..., '-' ..., ...... ...., ..... ..., ! In developrrent J; lnJect1on well South Magnolia .t In production North Ridge G Development facility 2 4 ;,,liles Figure I. Mu le deer winter range study areas re lative lo active natural gas well pads a nd energy development faci lities in the Piceance Basin of north west Colorado, summer 20 l 0. 59 �~ -' ~ '_,i Adult Female Survival, 28 June 2009-26 June 2010 1.000 0.900 0.800 0.700 0.600 - - - - - - - - - - - - - - - - - - - - 0.500 0.400 0.300 - · · t - - - - - - - - - - - - - - - - - - - - 0.200 0.100 0.000 ! l I I I I I I I I I I I I I I Ii;. I I I I I I I I I ! l I I I I ...a. 6 " " N °'r:, °'r:, ~ ~ a. 00 en en <f: > °'c.o ... > .c "ro 2:- <C 6 ... a. u z > a. 2 ro 2 ~ L/".I L/".I N 2::::, :::, --, .-t N .-t I 1.11 N .-t .-t 0 N .-t rtl QJ 0 0 C: ro --, u 0 QJ z 0 1.11 c.o :::, :::, --, 00 I .-t QJ :::, 00 N C: ~ N I .-t 7 L/".I .-t 0 rtl rtl N .-t N C: .-t 'I.. --, LL ~ rtl 00 'I.. 'I.. <C ro '.._I ~ -North Ridge -North Magnolia -south Magnolia -Ryan Gulch :::, ~ ~ ~ ~ _, 6 C: N 7 QJ 0 \_,I C: rtl :::, --, > ro 2 <C _, N QJ ro I '-I ~ L/".I .-t N QJ L/".I .-t N 'I.. I I ~~ ~ 2 ~ \iwl Fawn Survival, 6 December 2009-27 March 2010 1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 ~ '-" -- ...-1 \.-,.'..; '-" I I N .-t I ~ u QJ 0 ~ °' r:, 6N .-t .-t u I I N u QJ QJ 0 0 N C: ro 7 "u N QJ 0 I I ~ I °' rtl 6 rtl N --, C: ro --, I C: C'0 .-t .-t C: ro --, r-!. .-t I I 0 rtl ~ N C: ro --, ~ .c QJ LL I I rtl .-t r-!. 0 I ~ "..!t LL LL N N .-t N .c .c QJ ..!t .c QJ QJ LL rtl C: C'0 I ~ 0 .-t N 2 " .-t ro I 00 N .c --, rtl 'I.. • 'I.. C'0 2 ~ -North Magnolia '_,I -south Magnolia ~ ""';J .-t N 'I.. 'I.. 2 2 ro \iwl -Ryan Gulch I I -North Ridge C'0 QJ LL Figure 2. Annual and winter survival rates of adult female (28 June 2009-26 June 201 0; top) and fawn (6 December, 2009-27 March, 201 0; bottom) mule deer from 4 winter range study areas in the Piceance Basin of northwest Colorado. Survival rates among fawn and doe groups were statistically similar (P > 0.05; Table 1). _, ~ _, -~ l._.i ~ ~ '-' ~ ~ 'w,I - l,_.J '....,~ ~ 60 \v '-" 1.,.1 �Ry-an Gulch - Diamonds South Magnolia - Stars c::J North Magnolia - Pluses North Ridge - Circles 5 10 20 .,-,____..-.,ill'!!::..::.,i ~!'iiiiiii~!iiiii~~~~~. M,les Figure 3. Mule deer migration routes from 4 winter range study areas in the Piceance Basin of no1ihwest Colorado, spring and fa ll 2009 . .,,.; ._., 6l �Mule deer winter concentration areas Study Area Kernel PDF L..J Ryan Gulch L J 95% D North Magnolia - 92% South Magnolia - 87% North Ridge -78% r ] 0 1.2525 5 - 61% Figure 4. Mule deer winter concentration areas (composite kernel Probability Density Functions; PDF) from 4 study areas in the Piceance Basin of northwest Colorado, December 2008- May 2010. Data from composite GPS locations of adult fema le mule deer by sn1dy area (5 GPS location attempts/day). 62 �- Colorado Division of Parks and Wildlife July I, 2010 - llllle 30, 2011 WILDLIFE RESEARCH REPORT State of_ _ _ _ _ __:C ,c_o""l""oc:.:ra::::d::::o<--_ _ _ _ _ : Division of Parks and Wildlife Cost Center 3430 : ""M""a:::..:mm ==a=ls=-= R=e=se=a=r=ch ' -'-------- - -- - -- Work Package 3001 : ,.D::. !e<>=e"'-r.. ., C<->o'n-'-'--"s"er """""'v'""a""ti' ""'o""n~---- - - - - - - - - Task No. 6 : Population Performance of Piceance Basin Mule Deer in Response to Natural Gas Resource Extraction and Miti gation Efforts to Address Human Activity and Habitat Degradation Federal Aid Project:._ ____:W . :. . - - -"1=8.:::..5-----"R~ - -- -Period Covered: July I, 2010 - June 30, 2011 Authors: C.R. Anderson and C. J. Bish op ....., ..... ....., - - Personnel: E. Bergman, J. Broderick, P. Damm, B. deVergie, D. Finley, L. Gepfert, M. Grode, C. Ha1ty, K. Kaai, T. Knowles, J. Lewis, P. Lukacs, T. Parks, B. Petch, M. Peterson, R. Velarde, L. Wolfe, CPW; E. Hollowed, L. Belmonte, BLM; S. Monsen, Western Ecological Consulting, Inc.; D. Freddy, Hoch Berg Enterprises; T. Graham, Ranch Advisory Partners; M. Wille, T & M Contractors.; H. Sawyer, Western Ecosystems Technology; P. Lendrum, T Bowyer, Idaho State University; P. Doherty, J. Northrup, G. Wittemyer, K. Wilson, G. White, Colorado State University; M. Keech, L. Shelton, M. She lton, R. Swisher, Quicksilver Air, Inc.; D. Felix, Olathe Spray Service, fnc.; L. Coulter, Coulter Aviation. Project support received from Federal Aid in Wildlife Restoration, Colorado Mule Deer Association, Colorado Mule Deer Foundation, Colorado State Severance Tax Fund, EnCana Corp., ExxonMobil Production Co., Marathon Oil Corp., Shell Petroleum, and Williams Production LMT Co. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the authors. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We propose to experimentally evaluate winter range habitat treatments and human-activity management alternatives intended to enhance mule deer (Odocoileus hemionus) populations exposed to energy development activities. The Piceance Basin of northwestern Colorado was selected as the project area due to ongoing natural gas development in one of the most extensive and important mule deer winter and transition range areas within the state. The data presented here represent the first 3 pretreatment years of a long-term shtdy addressing habitat modifications and improved energy development practices intended to improve mule deer fitness in areas exposed to extensive energy development. We monitored 4 winter range sh1dy areas representing varying levels of development to serve as treatment (Ryan Gulch, North Magnolia, South Magnolia) and control (North Ridge) sites and recorded habitat use and movement patterns using GPS collars (5 location attempts/day), estimated overwinter fawn and annual adult female survival, estimated early and Late winter body condition of adult females using ultrasonography, and estimated abundance using helicopter mark-resight surveys. We targeted 250 fawns (50- 80/study area) and 100 does (20-40/study area) in early December 201 0 for VHF and GPS radiocollar attachment, 51 �respectively, and 80 does in March 2011 (20/study area) for late winter body condition assessment and to increase our GPS radiocollar sample in 3 of the 4 areas ( IO of 20/area excluding Ryan Gulch). Based on the data collected since January 2008, deer from a ll areas appear to be in reasonably good condition and exhibited high survival rates the first 2 years, with lower winter fawn survival through mid-June this past winter in 3 of 4 study areas (excluding North Ridge), and winter range deer densities appear to be stable or increasing. Mild winter conditions the first 2 years fo llowed by more severe winter conditions this year Likely contributed to the observed survival rates and population trends. Observed differences in winter concentration areas thus far may indicate behavioral modifi cations to areas of high development activity, but resource selection analyses will be necessary to confirm this supposition. Pilot habi tat treatments ( 126 acres total) were completed January 20 I I and moist spring weather conditions have resulted in excellent vegetation response thus fa r. We will continue to collect the various population and habitat use data across all study sites to evaluate the effectiveness of additional habitat treatments (North and South Magnolia) scheduled for fall/winter 20I 2- 2013 ( 1,200 acres total). This evaluation will allow us to determine whether it is possible to effectively mitigate development impacts in highly developed areas, or whether it is better to allocate mitigation dollars toward less or non-impacted areas. In collaboration with Colorado State University, we are also evaluating deer behavioral responses to vary ing levels of development activ ity in the Ryan Gulch study area. This will allow us to assess the effectiveness of certain Best Management Practices (BMPs) for reducing disturbance to deer. The sh1dy is slated to run through at least 20 L7, and preferably 20 19, to adequately measure mule deer population responses to landscape leve l manipulations. - -- -- --- 52 -- �.._ WILDLIFE RESEARCH REPORT POPULATION PERFORMANCE OF PIC.EANCE BASIN MULE DEER IN RESPONSE TO NATURAL GAS RESOURCE EXTRACTION AND MITIGATION EFFORTS TO ADDRESS HUMAN ACTIVITY AND HABITAT DEGRADATION CHARLES R. ANDERSON, JR and CHAD J. BISHOP PROJECT NARRITIVE OBJECTIVES I. To determine experimentally whether enhancing mule deer habitat conditions on winter range elicits behavioral responses, improves body condition, increases overwinter fawn survival, or ultimately, population density on mule deer winter ranges exposed to extensive energy development. 2. To determine experimentally to what extent modification of energy development practices enhance habitat selection, body condition, over-winter fawn survival, and winter range mule deer densities. SEGMENT OBJECTIVES '-' I. Collect and reattach GPS collars to maintain sample sizes for addressing mule deer habitat use and behavior patterns in 4 study areas experiencing varying levels of energy development of the Piceance Basin, northwest Colorado. '-..I 2. Estimate early and late winter body condition of adult female mule deer in each of the 4 winter herd segments 3. Monitor over-winter fawn and annual adult female mule deer survival by daily ground tracking and biweekly aerial tracking. 4. Conduct Mark-Resight helicopter surveys to estimate mule deer abundance in each study area. 5. Initiate habitat treatments for assessing efficacy of habitat improvement projects to mitigate energy development disturbances to mule deer. INTRODUCTION '-I Extraction of natural gas from areas throughout western Colorado has raised concerns among many public stakeholders and Colorado Parks and Wildlife that the cumulative impacts associated with this intense industrialization will dramatically and negatively affect the wildlife resources of the region. Concern is especially high for mule deer due to their recreational and economic importance as a principal game species and their ecological importance as one of the primary herbivores of the Colorado Plateau Ecoregion. Extraction of natural gas will directly affect the potential suitability of the landscape used by mule deer through conversion of native habitat vegetation with drill pads, roads, or noxious weeds, by fragmenting habitat because of drill pads and roads, by increasing noise levels via compressor stations and vehicle traffic, and by increasing the year-round presence of human activities. Extraction will indirectly affect deer by increasing the human work-force population of the region resulting in the need for additional landscape for human housing, supporting businesses, and upgraded road/transportation infrastructure. Additionally, increased traffic on rural roads will raise the potential for vehicle-animal collisions and additive direct mortality to mule deer populations. Thus, research documenting these impacts and evaluating the most effective strategies for minimizing and mitigating these activities will 53 �greatly enhance future management efforts to sustain mule deer populations for future recreational and ecological values. The Piceance Basin in northwest Colorado contains one of the largest migratory mule deer populations in North America and also exhibits some of the largest natural gas reserves in North America. Projected energy development throughout northwest Colorado within the next 20 years is expected to reach about 15,000 wells, many of which will occur in the Piceance Basin, which currently supports over 250 active gas well pads (http://cogcc.state.co.us). Anderson and Freddy (2008a) in their long-term research proposal identified 6 primary study objectives to assess measures to offset impacts of energy extraction on mule deer population perfonnance. During the past 4 years, we have gathered baseline habitat utilization data from OPS-collared deer across the Piceance Basin to allow assessment of mitigation approaches that will be implemented over the next 1-2 years and evaluated for another 4-6 years. We are currently monitoring 1 control area without development (North Ridge), 2 areas with relatively high development activity (0.6-0.8 well pads & facilities/km 2 ; Ryan Gulch, South Magnolia), and another area with relatively minor development activity (0.1 well pads & facilities/km2; North Magnolia). In comparison to the un-manipulated control area (North Ridge), the North and South Magnolia areas will receive similar levels of mechanical habitat treatments to evaluate this mitigation technique in relation to differing development intensities, and deer behavior patterns relative to differing development activities in the Ryan Gulch area will be monitored to identify effective Best Management Practices (BMPs) for future application. This progress report describes the previous 3.5 years (Jan 2008-June 2011) of addressing mule deer population performance during the pretreatment phase on 4 winter range herd segments, which includes monitoring habitat selection and behavior patterns of adult female mule deer, overwinter fawn and adult female survival, estimates of adult female body condition during early and late winter, and abundance estimates. STUDY AREAS The Piceance Basin, located between the cities of Rangely, Meeker, and Rifle in northwest Colorado, was selected as the project area due to its ecological importance as one of the largest migratory mule deer populations in North America and because it exhibits one of the highest natural gas reserves in North America (Fig. 1). Historically, mule deer numbers on winter range were estimated between 20,000-30,000 (White and Lubow 2002), and the current number of well pads (Fig. I) and projected number of gas wells in the Piceance Basin over the next 20 years is about 250 and 15,000, respectively. Mule deer winter range in the Piceance Basin is predominantly characterized as a topographically diverse pinion pine (Pinus edulis)-Utahjuniper (Juniperus osteosperma; pinion-juniper) shrubland complex ranging from 1,675 m to 2,285 m in elevation (Bartmann and Steinert 1981 ). Pinion-juniper are the dominant overstory species and major shrub species include Utah serviceberry (Amelanchier utahensis), mountain mahogany (Cercocarpus montanus), bitterbrush (Purshia tridentata), big sagebrush (Artemisia tridentata), Gamble's oak (Quercus gambelii), mountain snowberry (Symphoricarpos oreophilus), and rabbitbrush ( Ch,ysothamnus spp.; Bartmann et al. 1992). The Piceance Basin is segmented by numerous drainages characterized by stands of big sagebrush, saltbush (Atriplex spp.), and black greasewood (Sarcobatus vermiculatus), with the majority of the primary drainages having been converted to mixedgrass hay fields. Grasses and forbs common to the area consist ofwheatgrass (Agropyron spp.), blue grama (Bouteloua gracilis), needle and thread (Stipa comata), Indian rice grass (Oryzopsis hymenoides), arrowleafbalsamroot (Balsamorhiza sagittata), broom snakeweed (Gutierrezia sarothreae), pinnate tansymustard (Descurainia pinna/a), milkvetch (Astragalus spp.), Lewis flax (Linum lewisii), evening primrose (Oenothera spp.), skyrocket gilia (Gilia aggregata), buckwheat (Erigonum spp.), Indian paintbrush (Castilleja spp.), and penstemon (Penstemon spp.; Gibbs 1978). The climate of the Piceance Basin is characterized by warm dry summers and cold winters with most of the annual moisture resulting from spring snow melt. 54 .,; .,; '-' ~ 'w \_.I ~ \_.I '-' I._.J 'w 1w 'w ~ '---'._ 'w .,; �'cl Wintering mule deer population segments we investigated in the Piceance Basin include: North Ridge (53 krn 2)just north of the Dry Fork of Piceance Creek including the White River in the northeastern portion of the Basin, Ryan Gulch ( 141 krn2) between Ryan Gulch and Dry Gulch in the southwestern portion of the Basin, North Magnolia (79 krn 2) between the Dry Fork of Piceance Creek and Lee Gulch in the north-central portion of the Basin, and South Magnolia (83 km2) between Lee Gulch and Piceance Creek in the south-central portion of the Basin (Fig. I). Each of these wintering population segments has received varying levels of natural gas development: no development in North Ridge, light development in North Magnolia (0.14 pads & facilities/km 2), and relatively high development in the Ryan Gulch (0.60 pads & facilities/km 2) and South Magnolia (0.86 pads & facilities/km2) segments (Fig. 1). Among the 4 study areas, North Ridge will serve as an unmanipulated control site, Ryan Gulch will serve to address human-activity management alternatives (BMPs) that benefit mule deer exposed to energy development, and North and South Magnolia will serve to address the utility of habitat treatments intended to enhance mule deer population performance in areas exposed to light (North Magnolia) and heavy (South Magnolia) energy development activities. METHODS Tasks addressed this period included mule deer capture and collaring efforts, monitoring overwinter fawn and annual adult female survival, estimating adult female body condition during early and late winter using ultrasonography, estimating mule deer abundance applying helicopter mark-resight surveys, and initiating winter range habitat treatments to benefit mule deer in areas experiencing disturbance from energy development activities. We employed helicopter net-gunning techniques (Barrett et al. 1982, van Reenen 1982) to capture 50-80 fawns and 20-40 adult females during early December 20 IO and 20 adult females during early March 2011 in each of the 4 study areas. Once netted, all deer were hobbled and blind folded. Fawns were weighed, radio-collared and released on site, and adult females were transported to localized handling sites for collection of body measurements and were fitted with GPS collars (20-40/area during December 20 I 0, 0-10/area during March 2011; 5 or 24 fixes/day; G2 l 10B, Advanced Telemetry Systems, Isanti, MN, USA) and released. To provide direct measures of decline in overwinter body condition, 20 does were recaptured in Ryan Gulch and 10 from the other 3 study areas that were captured the previous December; IO uncollared does were also captured in North Ridge, North Magnolia, and South Magnolia to increase sample sizes in those areas. Fawn collars were spliced and fitted with rubber surgical tubing to facilitate collar drop during mid-summerearly autumn and GPS collars were supplied with timed drop-off mechanisms scheduled to release early April of the year following deployment. All radio-collars were equipped with mortality sensing options (i.e., increased pulse rate following 4-8 hrs of inactivity). Mule Deer Habitat Use and Movements We downloaded and summarized data from GPS collars deployed March 2010 following collar drop and retrieval in early April 2011. GPS collars deployed early December 2010 maintained the same fix schedule of attempting fixes every 5 hours except in Ryan Gulch where fix rates were increased to I/hour to increase resolution of GPS data for evaluation of deer behavior patterns in relation to differing development activities. We plotted deer locations and recorded timing and distance of spring and fall 2010 migrations for each study area. Mule deer winter concentration areas were created using composite GPS data (winter locations March 2010-April 2011 from all deer; 5 location attempts/day) from each study area and mapped in ArcGIS (ver. 9.3) using Spatial Analyst (kernel probability density functions separated by quantiles). Mule deer resource selection analyses are pending completion of high resolution habitat data layers currently being developed by BLM. 55 �Mule Deer Survival Mule deer mortality monitoring consisted of daily ground telemetry tracking and aerial monitoring approximately every 2 weeks from fixed-wing aircraft on winter range and bi-weekly aerial monitoring on summer range. Once a mortality signal was detected, deer were located and necropsied to assess cause of death. We estimated weekly survival using the staggered entry Kaplan-Meier procedure (Kaplan and Meier 1958, Pollock et al. 1989). Capture-related mortalities (any mortalities occurring within IO days of capture) and collar failures were censored from survival rate estimates. We estimated survival rates 1 July 2010-30 June 2011 for adult females and early December 2010-mid June 2011 for fawns. Adult Female Body Measurements We applied ultrasonography techniques described by Stephenson et al. ( 1998, 2002) and Cook et al. (2001) to measure maximum subcutaneous rump fat (mm), loin depth (longissimus dorsi muscle, mm), and to estimate% body fat. We estimated a body condition score (BCS) for each deer by palpating the rump (Cook et al. 2001, 2007). We examined differences (P < 0.05) in nutritional status among study areas and between years using a two-sample /-test. We considered differences in body condition meaningful when mean rump fat or % body fat differed statistically between comparisons. Other body measurements recorded included pregnancy status (pregnant, barren) via blood samples, weight (kg), chest girth (cm), and hind-foot length (cm). ._, ._, ._, ~ ._, 'wl 1w Abundance Estimates ~ We conducted 4 (North Ridge) or 5 (the remaining study areas) helicopter mark-resight surveys (2 observers and the pilot) during late March, 2011 to estimate deer abundance in each of the 4 study areas. We delineated each study area from GPS locations collected during winter from previous years (since Jan 2008) and aerial telemetry locations of radio-collared deer within 1 week of the first markresight survey. Two aerial fixed-wing telemetry surveys/study area were conducted during helicopter mark-resight surveys to determine which marked deer were within each survey area. We delineated flight paths in ArcGIS 9.3 prior to surveys following topographic contours (e.g., drainages, ridges) and approximating 500 m spacing throughout each study area; flight paths during surveys were followed using GPS navigation in the helicopter. Two approximately 12 x 12 cm pieces of Ritchey livestock banding material (Ritchey Livestock ID, Brighton, CO USA) were uniquely marked using color, number, and symbol combinations and attached to each radio-collar to enhance mark-resight estimates. Each deer observed during surveys was recorded as mark ID#, unmarked, or unidentified mark. We used program MARK (White and Burnham 1999) applying the mixed logit-normal model (McClintock et al. 2008) to estimate mule deer abundance and confidence intervals. For mark-resight mode] evaluations, we examined parameter combinations of varying detection rates with survey occasion 2 and whether individual sighting probabilities (i.e., individual heterogeneity) were constant or varied (cr = Oor * 0). Model selection procedures followed the infonnation-theoretic approach of Burnham and Anderson (2002). RESULTS AND DISCUSSION Deer Captures and Survival The helicopter crew captured 264 fawns and I 07 does in December 2010 and 81 does during March 2011. Nine fawn mortalities (ultimate cause= 6 capture myopathy and 3 predation) occurred 56 '--".._ ._, ._, '-' '-' ._, �within the 10 day myopathy period following the December capture and 1 doe mortality each followed the December and March captures (ultimate cause= I capture myopathy and 1 predation). ...,_ Fawn survival from early-December 2010-mid June 2011 was similar (P > 0.05) among 3 of 4 study areas ranging from 0.48 to 0.51, with North Ridge fawns exhibiting marginally higher over-winter survival (0. 70; P < 0.10~ Table 1). In comparison to previous years, North Ridge fawn survival has been consistent since winter 2008/09, but survival in the other 3 areas was lower than last year and lower than the previous 2 years in Ryan Gulch (Fig. 2). Annual adult female survival was similar among study areas (P > 0.05) ranging from 0.77 (North Ridge) to 0.89 (Ryan Gulch; Table l) and was comparable to previous years (P > 0.05; Anderson 2009~ Anderson and Bishop 20 I 0). The relatively lower fawn survival observed this winter (3 of 4 study areas) was likely due to increased winter severity present through mid February, and doe survival was consistent with other mule deer populations experiencing normal winter conditions in the western US (Unsworth et al. 1999). Seasonal Movement Patterns ...., ....,; Migration patterns differed among areas with North Ridge and North Magnolia deer generally migrating east-west and South Magnolia and Ryan Gulch deer migrating south-north (Fig. 3). Median straight-line migration distances were similar ranging from 32.6 km (Ryan Gulch) to 41.3 km (North Magnolia). Similar to seasonal ranges, most deer monitored exhibited strong fidelity to spring and fall migration routes (Fig. 3). Timing of spring migration during 2010 was similar among study areas with median spring migration dates occurring between 8 and 16 May and median fall migration dates occurring between 15 and 23 October. Median migration duration was relatively short among areas ranging from 3 to 8 days in the spring and 2 to 6 days in the fall; these observations were comparable to previous years. More detailed analyses of these migration data investigating the influence of human activity are currently being conducted by Patrick Lendrum and Terry Bowyer ofldaho State University. A final report is scheduled to be completed by spring 2012. Winter concentration areas identified from March 2010-April 2011 (Fig. 4) reasonably followed study area boundaries delineated from previous OPS locations of adult female mule deer (Anderson and Bishop 2010). Winter concentration areas outside study area boundaries primarily resulted from atypical distribution shifts of some North Ridge deer. Within study areas, we noted more continuous distributions from North Magnolia and North Ridge deer, with Ryan Gulch and South Magnolia deer exhibiting more fragmented and concentrated distributions, which may be related to relative development densities and longevity within each study area. Future resource selection analyses will address these differences relative to habitat attributes within each area. Mule Deer Body Condition ~ Body condition measurements of adult female mule deer December 2010 were comparable to 1ast year (Anderson and Bishop 2010) with higher values evident from North and South Magnolia deer, intermediate from Ryan Gulch deer, and lower values from North Ridge deer (Table 2), but differences were only marginal (P < 0.01) between North Ridge and the 2 Magnolia populations (mm rump fat: P = 0.05-0.07). Unlike last year, deer coming into winter range with higher body condition did not maintain improved condition by late winter and all herd segments were similarly low when assessed in March 2011. The similarly low body condition among areas we observed during late winter can likely be attributed to increased winter severity this winter relative to last winter. Overwinter decline in mean% body fat ranged from 3.8% in Ryan Gulch to 4.7% in South Magnolia (Table 2). Pregnancy rates were expectedly high ranging from 95% to 100%/study area (n = 20/area). 57 �Similar to subtle trends in adult female body condition the past 3 years (Table 2), December fawn weights were slightly higher in 2009 than during 2008 and 2010 (Fig. 5). In 2009, male fawns from North and South Magnolia were heavier (P < 0.05) than during 2008 as were Ryan Gulch males when compared to 2010. Similarly, 2009 females were heavier from North Magnolia compared to 2008 and from North Magnolia and Ryan Gulch than during 20 IO (Fig. 5). In comparing fawn weights from December 2010, Ryan Gulch fawns were marginally (P = 0.055; South Magnolia females) or significantly lighter (P < 0.05; both sexes from the other 3 study areas and males from South Magnolia) than other fawns. Mule Deer Population Estimates Mark-resight models that best predicted abundance estimates (lowest AICc; Burnham and Anderson 2002) exhibited homogenous individual sightability (cr2 = 0) and constant sightability across surveys (P.) for South Magnolia and Ryan Gulch, homogenous individual sightability and variable sightability with survey period for North Ridge, and heterogeneous individual sightability with variable sightability across surveys for North Magnolia. North Ridge exhibited the highest deer density (22.9/km2), followed by North Magnolia ( l l .2/km2), with comparably lower deer densities in South Magnolia and Ryan Gulch (7 .6 and 8. 7/km 2 ; Table 3, Fig. 6). Abundance estimates were similar (P > 0.05) to last year except in North Magnolia where deer numbers increased from 595 to 884. Over the 3 year survey period so far the population trend in North Ridge appears to be increasing with a recent increase in North Magnolia and stability in the other 2 areas (Fig. 6). Abundance estimates from 2011 were similarly precise from all 4 study areas with the mean Confidence Interval Coefficient of Variation (CICV) ranging from 0.14-0.18. SUMMARY AND FUTURE PLANS The goal of this study is to investigate habitat treatments and energy development practices that enhance mule deer populations exposed to extensive energy development activity. The infonnation presented here provide data describing mule deer population parameters from the first 3.5 years of the pre-treatment period of a long-term study intended to address how mule deer react to landscape scale habitat and human activity modifications. The pretreatment period is intended to continue I to 2 more winters to provide baseline data to compare against intended improvements in habitat conditions and evaluation of concentration/reduction in human development activities, which will be maintained for 4--6 years to provide sufficient time to measure how deer respond to these changes. Based on the data collected thus far, deer from all areas appear to be in reasonably good condition and are exhibiting expected survival rates relative to changes in winter severity. Mild winter conditions the first 2 years and more severe winter conditions during the current year likely contributed to the observed mule deer population parameters. Observed differences in winter concentration areas (Fig. 4) may indicate behavioral modifications to areas of prolonged high development activity, but resource selection analyses will be necessary to confinn this supposition. We will continue to collect the various population and habitat use data across all study sites to evaluate the effectiveness of habitat improvements on winter range. This approach will allow us to determine whether it is possible to effectively mitigate development impacts in highly developed areas, or whether it is better to allocate mitigation dollars toward less or non-impacted areas. In a recent project conducted on the Uncomphahgre Plateau, Bergman et al. (2009) found that habitat treatments implemented in pinyon-juniper habitat in undeveloped areas were effective for deer. We are also evaluating deer behavioral responses to varying levels of development activity. This will allow us to assess the effectiveness of certain BMPs for reducing disturbance to wintering mule deer. We recently implemented a habitat improvement plan and completed our pilot habitat treatments January 201 l (126 acres total) and plan to complete the remaining treatments (~1,080 acres) in the 58 lwl �'-' -cl 'Cl 'Cl '-'~ '411 >.-I ,._,I '4fl -.-1 'Cl ,.,I .._ '-' Magnolia study areas by fall/winter 2012-2013; vegetation response thus far in the pilot treatment sites have been promising, likely due to the moist spring conditions this year. In addition, hay field improvements have been implemented in the North Magnolia area from a collaborative agreement with Williams Production LMT Co. Additional collaboration with Williams Production LMT Co. have produced a clustered development plan to be implemented in the Ryan Gulch study area and new technologies will be implemented to reduce human activity through remote monitoring of well pads and fluid collection systems. Recent collaboration agreements with ExxonMobil Development Co. and Colorado State University have provided graduate research opportunities to enhance data collection and inference about mule deer/energy development interactions. We are continuing to work with Dr. Terry Bowyer and Patrick Lendrum (MS candidate) of Idaho State University to address mule deer migration and potential influences of human activity along migration routes. Additional funding and cooperative agreements will be necessary to sustain this project through completion (through at least 2017 and preferably through 2019). We optimistically anticipate the opportunity to work cooperatively toward developing solutions for allowing the nation's energy reserves to be developed in a manner that benefits wildlife and the people who value both the wildlife and energy resources of Colorado. \1$1 LITERATORE CITED -.el 'Cl -.I -.e) vs/ 'el 'Cl'._...I -.I -...I -.I -.I -.I "-' 'el 'Ori 'Cl wl \t/lV 'Cl '-' 'Cl ..., ._, - ~ Anderson, C.R., Jr. 2009. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and D. J. Freddy. 2008a. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Final Study Plan, Colorado Division of Wildlife, Ft. Co11ins, CO, USA. Anderson, C.R., Jr., and D. J. Freddy. 2008b. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation-Stage I, Objective 5: Patterns of_mule deer distribution & movements. Pilot Study, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and C. J. Bishop. 2010. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA. Bartmann, R. M. 1975. Piceance deer study-population density and structure. Job Progress Report, Colorado Divison of Wildlife, Fort Collins, Colorado, USA. Bartmann, R. B., and S. F. Steinert. 1981. Distribution and movements of mule deer in the White River Drainage, Colorado. Special Report No. 51, Colorado Division of Wildlife, Fort Collins, Colorado, USA. Bartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monograph No. I 21. Barrett, M. W., J. W. Nolan, and L. D. Roy. I 982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin I 0: 108-114. Bergman, E. J., C. J. Bishop, D. J. Freddy, and G. C. White. 2009. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, USA. Burnham, K. P., and D.R. Anderson. 2002. Model selection and multi-model inference: a practical information-theoretic approach. Second edition. Springer-Verlag, New Yark, New York, USA. Cook, R. C., J. G. Cook, D. L. Murray, P. Zager, B. K. Johnson, and M. W. Gratson. 2001. Development of predictive models of nutritional condition for rocky mountain elk. Journal of Wildlife Management 65:973-987 . Cook, R. C., T. R. Stephenson, W. L. Meyers, J. G. Cook, and L.A. Shipley. 2007. Validating predictive models of nutritional condition for mule deer. Journal of Wildlife Management 71: 1934-1943. ._ '4,/ .., ._ 59 �Gibbs, H. D. 1978. Nutritional quality of mule deer foods, Piceance Basin, Colorado. Thesis, Colorado State University, Fort Collins, Colorado, USA. Kaplan, E. L., and P. Meier. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 52:457-481. McClintock, B. T., G. C. White, K. P. Burnham, and M.A. Pride. 2008. A generalized mixed effects model of abundance for mark-resight data when sampling is without replacement. Pages 271289 in D. L. Thompson, E.G. Cooch, and M. J. Conroy, editors, Modeling demographic processes is marked populations. Springer, New York, New York, USA. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. C. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Stephenson, T. R., V. C. Bleich, B. M. Pierce, and G. P. Mulcahy. 2002. Validation of mule deer body composition using in vivo and post-mortem indices of nutritional condition. Wildlife Society Bulletin 30:557-564. Stephenson, T. R., K. J. Hundertmark, C. C. Swartz, and V. Van Ballenberghe. 1998. Predicting body fat and mass in moose with untrasonography. Canadian Journal of Zoology 76:717-722. Unsworth, J. W., D. F. Pack, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63 :315-326. Van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked individuals. Bird Study 46:120-139. White, C. C., and B. C. Lubow. 2002. Fitting population models to multiple sources of observed data. Journal of Wildlife Management 66:300-309. Prepared by _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ Chuck R. Anderson, Wildlife Researcher 'w 60 �Table 1. Survival rate estimates (S) of fawn (1 Dec. 2010-18 June 2011) and adult female (1 July 2010-30 June 201 l) mule deer from 4 winter range study areas of the Piceance Basin in northwest Colorado. -.I Cohort ""51 Study area Initial sample size (n) March doe sample11 (n) S(95% CI) Fawns '411 Ryan Gulch 50 0.480 (0.342-0.618) South Magnolia 55 0.508 (0.375-0.640) North Magnolia 60 0.481 (0.351-0.610) North Ridge 77 0.697 (0.594--0.080) Adult females Ryan Gulch 31 51 0.892 (0.800-0.983) South Magnolia 28 53 0.832 (0.708-0.955) North Magnolia 32 54 0. 783 (0.654--0. 912) North Ridge 33 44 0.765 (0.622-0.908) 8 Adult female sample size following capture and radio-collaring efforts March, 2011. '-' 61 \/fSI �Table 2. Mean rump fat (mm), Body Condition Score (BCS3), and % body fat (% fat) of adult female mule deer from 4 study areas in the Piceance Basin of northwest Colorado, March and December, 2009-2011. Values in parentheses= SD. March 2009 December 2009 %fat Study Area Rump fat Ryan Gulch 1.73 (1.78) 2.66 (0.55) 7.54 (1.80) 8.35 (6.36) 4.06 (1.13) 12.96 (4.53) 2.31 (1.44) 2.35 (0.48) 6.69 (1.58) South Magnolia 1.47 (0.68) 2.50 (0.60) 7.26 (1.82) 10.05 ( 6.19) 4.07 ( 1.21) 13.46 (4.96) 3.12 (2.20) 2.64 (0.59) 7.70 (2.01) North Magnolia 1.30 (0.79) 2.56 (0.68) 6.96 (2.23) 10.67 (5.76) 4.25 (0.96) 13.92 (3.92) 3.15 (2.34) 2.85 (0.53) 8.28 (1.86) North Ridge 1.57 (1.22) 2.60 (0.56) 7.28 (1.66) 5.25 (5.65) 3.63 (1. 11) 11.02 (4.54) 1.77 (1.11) 2.42 (0.49) 6.83 (1.50) BCS Rump fat March 2010 BCS % fat Rump fat BCS %fat Table 2. Continued. December 2010 March 2011 Study Area Rump fat Ryan Gulch 7.75 (6.15) 3.34 (0.98) 10.82 (4.32) 1.55 (0.60) 2.53 (0.42) 7.05 (1.20) South Magnolia 9.85 (6.78) 3.30 (0.61) 11.21 (3.32) 1.65 (0.75) 2.35 (0.50) 6.56 (1.49) North Magnolia 9.55 (6.49) 2.56 (0.68) 11.65 (4.86) 1.65 (0.67) 2.53 (0.49) 7.06 (1.35) BCS %fat Rump fat BCS %fat North Ridge 6.14 (5.29) 3.32 (0.82) I 0.32 (3.39) 1.45 (0.76) 2.24 (0.49) 6.24 (1.45) 0 Body condition score taken from palpations of the rump following Cook et al. (200 I). 62 ( ( ( C( { C{ { { CC{ C( { CCCC{ CC( ( { C( ( { { C{ { ( ( CC( C( ( CCC{ �Table 3. Mark-resight abundance (N) and density estimates of mule deer from 4 winter range herd segments in the Piceance Basin, northwest Colorado, 29 March-4 April 2011. Data represent 4 resight surveys from North Ridge and 5 resight surveys from the other 3 study areas. Study area Mean No. sighted Mean No. marked N(95% CI) Density (deer/km2) Ryan Gulch 327 22 1,2 I 9 (1,040--1,431) 8.7 South Magnolia 156 21 630 (542-735) 7.6 North Magnolia 239 22 884 (739-1,060) 11.2 North Ridge 409 30 1,221 ( 1,067-1,399) 22.9 'cl 63 �- Mule Deer Winter Range Study Area Boundaries Study Area Well Pads & Facilities D Rfan Gulch c:J North Magnolia D South Magnolia North Ridge 0 3 J; J; In development or appl,cat,on for dnlhng Producing v.ell [, ll1Ject1on well El Development_fac1l1ties 6 12 Miles Fig ure I. Mule deer winter ra nge study areas relative to active natural gas well pads and energy development faci lities in the Piceance Basin of northwest Colorado, summer 20 11(Accessed http://cogcc.state.co.us/ Aug. 8, 20 I l ). 64 �Ryan Gulch fawn s-2008/09-2010/11 South Magnolia fawn S- 2008/09--2010/11 r===~~~:st~;-;~;;;;~;;;;~~g. ~:~~ 1.00 · 0.90 ····••.• •• 0.80 +--------~-""'°'<;:,----~.....-:-::--'-.....-:-:-=:-:7""""" 0.70 +-----------'-"r,---===--""-::,,---....a...• 0.60 + - - - - - - - - - - = 0.50 + - - - - - - - - - - - - - - - ~ - ~ - - 0.40 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __._ _ . . 3-~::. . . ·-~?=::·...............~.._;::z;-~~:.:~ 0 80 +------=,l.«•41#•1:n·•·•··· o:/U . ••••••••••••••••••• .,.,,...___ TITTTITn'Tff..,-- 0.40 + - - - - - - - - - - - - - - - - - - - - 0.30 · t - - - - - - - - - - - - - - - - - - - 0.20 + - - - - - - - - - - - - - - - - - - - 0.10 + - - - - - - - - - - - - - - - - - - - 0.00 +----r----.---.----.-..---.---.---,-----,.-,--,----,.---,---,----,---, 0.JO - 1 - - - - - - - - - - - - - - - - - - - 0.20 · · - - - - - - - - - - - - - - - - - - 0.10 · - - - - - - - - - - - - - - - - - - - o.oo -1----~--.-------,--,--.-,- . - - - - ~ North Magnolia fawn ~ - 2008/09-2010/11 ~:~~ +---~-;;::::. ::. ' •••. :•• ..--.. 0.60 •••••••••• 0.50 - j - - - - - - - - - - - - - - • • _ • • . . . c . ; • • • : : . £ . ' • ' - T T T I T O " r n n North Ridge fawn S• 2008/09-2010/11 1.UO • -····;•:.•• ~Y..::::·::~~~~~~ ~~ h••>.W!-....W!-.•,..•!-.•t>,- o.so 0.80 ,. • .0.•••..••nTl"o~~0,70 , ,........... ttltttt-0.GO +------------.........._. 0.50 ............ .. 0.40 + - - - - - - - - - - - - - - - - - - - 0.30 + - - - - - - - - - - - - - - - - - - - 0.20 + - - - - - - - - - - - - - - - - - - - 0.10 + - - - - - - - - - - - - - - - - - - - 0.00 -~-.,---------.---.---..---,---,---,---,---, ~.:#i/F_'i":§;~~~,.~.....zt - ...... ~.....-.~..... -•-•-......·nn!llIDI.l ·n·,. .~ 0.80 .................. ::::::au., .... . 0.70 - f - - - - - - - - - - - - - - - = - = = = = u u . . a 0.60 + - - - - - - - - - - - - - - - - - - - - 0.50 + - - - - - - - - - - - - - - - - - - - 0.40 + - - - - - - - - - - - - - - - - - - - 0.30 + - - - - - - - - - - - - - - - - - - - 0.20 + - - - - - - - - - - - - - - - - - - - O.lO + - - - - - - - - - - - - - - - - - - - 0.00 +---.----.---.---.--,--,----.----.---.----,.-,--,--~--.----,-----, 'el Figure 2. Over-winter (Dec-Mar) mule deer fawn survival (.5) from 4 study areas in the Piceance Basin, northwest Colorado, 2008/09 (red lines), 2009/10 (orange lines) and 20010/11 (blue lines). Solid lines= 5 and dashed lines = 95% CI. Comparable data among years December-March due to premature collar drop during 2008 and 2009. 'cl 'SI 65 �D 6'i,:i,;Rl'm,~-';!t~•r\l D North Ridge. Orcles North M agnolla • Pluses South Magnolia• Stars :::J Ryan Gulch. 01arronds 0 5 10 20 ,,~1les Figure 3. Mule deer migration routes from 4 winter range study areas in the Piceance Basin of northwest Colorado, spring and fall 2010. 66 �-- - - Mule Deer Winter Concentration Areas Study Area Kernel PDF O RyanGulch L J 95% North Magnolia -92% 5outh Magnolia - 87% North Ridge - 81% - 70% D D -- 125 25 ~l""""l~-~-~-~~~~IMiles Figure 4. Mule deer wi nter concentration areas (composite kernel Probability Density Functions; PDF) from 4 study areas in the Piceance Basin of no1thwest Colorado, March 20 I 0-April 20 11. Data from composite GPS localions (5 GPS location attempts/day) of adult female mule deer by study area. -...,, - 5 67 �- Male fawn weights 45.0 40.0 I- I 35 .0 tio I - f I I 30.0 ..:t: ... 25.0 .c: Ryan Gulch South Magnolia .!?:> 20.0 (II ~ ■ North Magnolia 15.0 North Ridge 10.0 5.0 0.0 Dec 2008 Dec2009 Dec 2010 Female fawn weights 40.0 35.0 I I I ·I 30.0 tlO 25.0 .... ..c: Ryan Gulch 20.0 South Magnolia 15.0 ■ North Magnolia 10.0 North Ridge .x OJ) 'iii $ 5.0 0.0 Dec 2008 Dec 2009 Dec 2010 Figure 5. Mean male and female fawn weights and 95% CI (error bars) from 4 mule deer study areas in the Piceance Basin, northwest Colorado, December 2008- 20 I 0. 68 �Late winter mule deer density 30 25 N E ~ 20 '.cu 15 - ... - - ... - --- - --- North Ridge ••••• • Ryan Gulch ~ - North Magnolia QJ C -)I@- 10 I South Magnolia 5 0 2009 2011 2010 Year Figure 6. Mule deer density estimates and 95% CI (error bars) from 4 winter range herd segments in the Piceance Basin, northwest Colorado, late winter 2009-2011. 'Cl 69 �--- Colorado Division of Parks and Wildlife July 1,201 l - June 30, 2012 WILDLIFE RESEARCH REPORT State of-------=-=-===~----Colorado : Division of Parks and Wildlife Cost Center 3430 : ===-=--===:.=c:'-------------Mammals Research Work Package 3001 : =D'-"e-=-er=--=C-=-o=n=se=rv-'-a=t=io=-=n~-----------Task No. 6 : Population Performance of Piceance Basin Mule Deer in Response to Natural Gas Resource Extraction and Mitigation Efforts to Address Human Activity and Habitat Degradation Federal Aid Project:_ ~ W~-~18=-=5~-~ R ----Period Covered: July 1, 201 1 - June 30, 2012 Authors: C.R. Anderson and C. J. Bishop """" Personnel: E. Bergman, T. Bryan, A. Burleson, B. deVergie, D. Finley, M. Fisher, L. Gepfert, C. Haity, D. Johnston, A. Jones, T. Knowles, J. Lewis, H. MacIntyre, J. Matijas, B. Panting, T. Parks, B. Petch, J. Rivale, J. Simpson, S. Singleton, M. Trump, B. Tycz, R. Velarde, L. Wolfe, CPW; E. Hollowed, L. Belmonte, BLM; S. Monsen, Western Ecological Consulting, Inc.; D. Freddy, Hoch Berg Enterprises; T. Graham, Ranch Advisory Partners; M. Wille, T & M Contractors.; P. Lendrum, T. Bowyer, ldaho State University; P. Doherty, J. Northrup, M. Peterson, G. Wittemyer, K. Wilson, G. White, Colorado State University; R. Swisher, S. Swisher, Quicksilver Air, Inc.; D. Felix, Olathe Spray Service, Inc.; L. Coulter, Coulter Aviation. Project support received from Federal Aid in Wildlife Restoration, Colorado Mu le Deer Association, Colorado Mule Deer Foundation, Colorado State Severance Tax Fund, EnCana Corp., ExxonMobil Production Co./XTO Energy, Marathon Oil Corp., Shell Petroleum, and Williams Production LMTCo. All information in this report is preliminary and subject to further evaluation. Information MAV NOT BE PUBLISHED OR QUOTED without permission of the authors. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT - - We propose to experimentally evaluate winter range habitat treatments and human-activity management alternatives intended to enhance mule deer (Odocoileus hemionus) populations exposed to energy-development activities. The Piceance Basin of northwestern Colorado was selected as the project area due to ongoing natural gas development in one of the most extensive and important mule deer winter and transition range areas in Colorado. The data presented here represent the first 4 pretreatment years of a long-term study addressing habitat improvements and evaluation of energy development practices intended to improve mule deer fitness in areas exposed to extensive energy development. We monitored 4 winter range study areas representing varying levels of development to serve as treatment (Ryan Gulch, North Magnolia, South Magnolia) and control (North Ridge) sites and recorded habitat use and movement patterns using GPS collars (2::5 location attempts/day), estimated overwinter fawn and annual adult female survival, estimated early and late winter body condition of adult fema les using ultrasonography, and estimated abundance using helicopter mark-resight surveys. We targeted 260 fawns (60-80/study area) and 140 does (30-40/study area) in early December 201 l for VHF and GPS radiocoUar attachment, respectively, and 140 does in March 201 2 (30-40/study area) for Late winter body condition assessment and to increase our GPS radiocollar sample in l of the 4 areas (24 in Ryan Gulch) to address neonate 48 �survival. Based on the data collected since January 2008, deer from all areas appear to be in reasonably good condition and have exhibited relatively high survival rates 3 of the 4 years (mean fawn S > 0.65) with lower winter fawn survival during 2010/11 in 3 of 4 study areas (mean S = 0.49 excluding North Ridge), and winter range deer densities appear to be stable. More extreme winter conditions during 2010/ 11 likely contributed to the observed decline in fawn survival rates. Pilot habitat treatments in North and South Magnolia (l 16 acres total) were completed January 2011 (Anderson and Bishop 20 11 ), another 54 acres were treated January 2012 to assess mechanical treatment methods (hydro-ax, rollerchop, chain), and all required NEPA surveys were completed this summer for the remaining sites (Fig. 6). The Biological Assessment should be completed during September 2012 allowing the remaining 1,030 acres to be treated using hydro-ax this winter. We will continue to collect the various population and habitat use data across all study sites to evaluate the effectiveness of habitat treatments (North and South Magnolia) scheduled for faLl/winter 2012- 2013 (1 ,200 acres total). This evaluation will allow us to determine whether it is possible to effectively mitigate development disturbance in highly developed areas, or whether it is better to allocate mitigation dollars toward less or non-impacted areas. In collaboration with Colorado State University, we are also evaluating deer behavioral responses to varying levels of development activity in the Ryan Gulch study area and neonate survival in relation to energy development from all study areas. This will allow us to assess the effectiveness of certain Best Management Practices (BMPs) for reducing disturbance to deer and include neonatal data to other demographic parameters for evaluation of mule deer/energy development interactions. The study is slated to run through at least 20 17, and preferably 2019, to adequately measure mule deer population responses to landscape level manipulations. -- - - 49 - �WILDLIFE RESEARCH REPORT POPULATION PERFORMANCE OF PICEANCE BASIN MULE DEER IN RESPONSE TO NATURAL GAS RESOURCE EXTRACTION AND MITIGATION EFFORTS TO ADDRESS HUMAN ACTIVITY AND HABITAT DEGRADATION CHARLES R. ANDERSON, JR and CHAD J. BISHOP PROJECT NARRITIVE OBJECTIVES >cl 1. To determine experimentally whether enhancing mule deer habitat conditions on winter range elicits behavioral responses, improves body condition, increases fawn survival, or ultimately, population density on mule deer winter ranges exposed to extensive energy development. 2. To determine experimentally to what extent modification of energy development practices enhance habitat selection, body condition, fawn survival, and winter range mule deer densities. SEGMENT OBJECTIVES 1. Collect and reattach GPS collars to maintain sample sizes for addressing mule deer habitat use and behavior patterns in 4 study areas experiencing varying levels of energy development of the Piceance Basin, northwest Colorado. 2. Estimate early and late winter body condition of adult female mule deer in each of the 4 winter herd segments using ultrasound techniques. 3. Monitor over-winter fawn and annual adult female mule deer survival by daily ground tracking and biweekly aerial tracking. 4. Conduct Mark-Resight helicopter surveys to estimate mule deer abundance in each study area. 5. Complete NEPA surveys to allow future habitat treatments for assessing efficacy of habitat improvement projects to mitigate energy development disturbances to mule deer. 6. Initiate neonate survival evaluations to complete demographic parameters for assessing mule deer/energy development interactions. INTRODUCTION Extraction of natural gas from areas throughout western Colorado has raised concerns among many public stakeholders and Colorado Parks and Wildlife (CPW) that the cumulative impacts associated with this intense industrialization will dramatically and negatively affect the wildlife resources of the region. Concern is especially high for mule deer due to their recreational and economic importance as a principal game species and their ecological importance as one of the primary herbivores of the Colorado . Plateau Ecoregion. Extraction of natural gas will directly affect the potential suitability of the landscape used by mule deer through conversion of native habitat vegetation with drill pads, roads, or noxious weeds, by fragmenting habitat because of drill pads and roads, by increasing noise levels via compressor stations and vehicle traffic, and by increasing the year-round presence of human activities. Extraction will indirectly affect deer by increasing the human work-force population of the region resulting in the need for additional landscape for human housing, supporting businesses, and upgraded road/transportation infrastructure. Additionally, increased traffic on rural roads will raise the potential for vehicle-animal collisions and additive direct mortality to mule deer populations. Thus, research documenting these relationships and evaluating the most effective strategies for minimizing and mitigating these activities will greatly enhance future management efforts to sustain mule deer populations for future recreational and ecological values. 50 �The Piceance Basin in northwest Colorado contains one of the largest migratory mule deer populations in North America and also exhibits some of the largest natural gas reserves in North America. Projected energy development throughout northwest Colorado within the next 20 years is expected to reach about 15,000 wells, many of which will occur in the Piceance Basin, which currently supports over 250 active gas well pads (http://cogcc.state.co.us; Fig. 1). Anderson and Freddy (2008a) in their longterm research proposal identified 6 primary study objectives to assess measures to offset impacts of energy extraction on mule deer population performance. During the past 4 years, we have gathered baseline habitat utilization data from OPS-collared deer across the Piceance Basin to allow assessment of mitigation approaches that will be implemented over the next 1-2 years and evaluated for another 4-6 years. We are currently monitoring 1 control area without development (North Ridge), 2 areas with relatively high development activity (0.6-0.9 well pads & facilities/km2; Ryan Gulch and South Magnolia), and another area with relatively minor development activity (0.1 well pads & facilities/km 2; North Magnolia). In comparison to the un-manipulated control area (North Ridge), the North and South Magnolia areas will receive similar levels of mechanical habitat treatments to evaluate this mitigation strategy relative to differing development intensities, and deer behavior patterns relative to differing development activities in the Ryan Gulch area will be monitored to identify effective Best Management Practices (BMPs) for future application. This progress report describes the previous 4.5 years (Jan 2008June 2011) of addressing mule deer population performance during the pretreatment phase on 4 winter range herd segments, which includes monitoring habitat selection and behavior patterns of adult female mule deer; spring/summer neonate, overwinter fawn and adult female survival; estimates of adult female body condition during early and late winter, and annual late-winter abundance estimates. STUDY AREAS The Piceance Basin, located between the cities of Rangely, Meeker, and Rifle in northwest Colorado, was selected as the project area due to its ecological importance as one of the largest migratory mule deer populations in North America and because it exhibits one of the highest natural gas reserves in North America (Fig. 1). Historically, mule deer numbers on winter range were estimated between 20,000-30,000 (White and Lubow 2002), and the current number of well pads (Fig.I) and projected number of gas wells in the Piceance Basin over the next 20 years is about 250 and 15,000, respectively. Mule deer winter range in the Piceance Basin is predominantly characterized as a topographically diverse pinion pine (Pinus edu/is)-Utahjuniper (Juniperus osteosperma; pinion-juniper) shrubland complex ranging from 1,675 m to 2,285 m in elevation (Bartmann and Steinert 1981 ). Pinion-juniper are the dominant overstory species and major shrub species include Utah serviceberry (Amelanchier utahensis), mountain mahogany (Cercocarpus montanus), bitterbrush (Purshia tridentata), big sagebrush (Artemisia tridentata), Gamble's oak (Quercus gambelii), mountain snowberry (Symphoricarpos oreophilus), and rabbitbrush (Chrysothamnus spp.; Bartmann et al. 1992). The Piceance Basin is segmented by numerous drainages characterized by stands of big sagebrush, saltbush (Atrip/ex spp.), and black greasewood (Sarcobatus vermiculatus), with the majority of the primary drainages having been converted to mixedgrass hay fields. Grasses and forbs common to the area consist ofwheatgrass (Agropyron spp.), blue grama (Bouteloua graci/is), needle and thread (Stipa comata), Indian rice grass (Oryzopsis hymenoides), arrowleafbalsamroot (Balsamorhiza sagittata), broom snakeweed (Gutierrezia sarothreae), pinnate tansymustard (Descurainia pinnata), milkvetch (Astragalus spp.), Lewis flax (Linum lewisii), evening primrose (Oenothera spp.), skyrocket gilia (Gilia aggregata), buckwheat (Erigonum spp.), Indian paintbrush (Castilleja spp.), and penstemon (Penstemon spp.; Gibbs 1978). The climate of the Piceance Basin is characterized by warm dry summers and cold winters with most of the annual moisture resulting from spring snow melt. Wintering mule deer population segments we investigated in the Piceance Basin include: North Ridge (53 km2) just north of the Dry Fork of Piceance Creek including the White River in the northeastern portion of the Basin, Ryan Gulch (141 km2) between Ryan Gulch and Dry Gulch in the 51 �southwestern portion of the Basin, North Magnolia (79 km2) between the Dry Fork of Piceance Creek and Lee Gulch in the north-central portion of the Basin, and South Magnolia (83 km2) between Lee Gulch and Piceance Creek in the south-central portion of the Basin (Fig. 1). Each of these wintering population segments has received varying levels of natural gas development: no development in North Ridge, light development in North Magnolia (0.14 pads & facilities/km2), and relatively high development in the Ryan Gulch (0.60 pads & facilities/km2) and South Magnolia (0.86 pads & facilities/km 2) segments (Fig. I). Among the 4 study areas, North Ridge will serve as an unmanipulated control site, Ryan Gulch will serve to address human-activity management alternatives (BMPs) that benefit mule deer exposed to energy development, and North and South Magnolia will serve to address the utility of habitat treatments intended to enhance mule deer population performance in areas exposed to light (North Magnolia) and heavy (South Magnolia) energy development activities. METHODS 'Cl Tasks addressed this period included mule deer capture and collaring efforts, monitoring overwinter fawn and annual adult female survival, estimating adult female body condition during early and late winter using ultrasonography, estimating mule deer abundance applying helicopter mark-resight surveys, working with BLM to complete NEPA surveys to proceed with mechanical habitat treatments fall/winter 2012, and initiation of evaluating neonate survival in developed and undeveloped landscapes. We employed helicopter net-gunning techniques (Barrett et al. 1982, van Reenen 1982) to capture 60--80 fawns and 30--40 adult females during early December 2011 and early March 2012 in each of the 4 study areas. Once netted, all deer were hobbled and blind folded. Fawns were weighed, radio-collared and released on site, and adult females were transported to localized handling sites for recording body measurements and fitted with GPS collars (30-40/area during December 2011, primarily recaptures during March 2012; 5 or 24 fixes/day; G2110D, Advanced Telemetry Systems, Isanti, MN, USA) and released. To provide direct measures of decline in overwinter body condition, 30 does were recaptured in each study area that were captured the previous December; 24 uncollared does were also captured in Ryan Gulch to achieve a desired sample size of 30/study area for monitoring neonate survival. Fawn collars were spliced and fitted with rubber surgical tubing to facilitate collar drop between mid-summer and early autumn, and GPS collars were supplied with timed drop-off mechanisms scheduled to release early in April of the year following deployment. All radio-collars were equipped with mortality sensing options (i.e., increased pulse rate following 4--8 hrs of inactivity). '151 Mule Deer Habitat Use and Movements We downloaded and summarized data from GPS collars deployed December 2010 following collar drop and retrieval in early April 2012. GPS collars deployed maintained the same fix schedule of attempting fixes every 5 hours except in Ryan Gulch where fix rates were programmed for I/hour to increase resolution of GPS data for evaluation of deer behavior patterns in relation to differing development activities. We plotted deer locations and recorded timing and distance of spring and fall 2011 migrations for each study area. Mule deer winter concentration areas were created using composite GPS data (March 2010 through April 2011 from all deer; 5 location attempts/day) from each study area and mapped in ArcGIS (ver. 9.3) using Spatial Analyst (kernel probability density functions separated by quantiles). Mule deer resource selection analyses are pending completion of high resolution habitat data layers currently being developed by BLM. 'el Mule Deer Survival Mule deer mortality monitoring consisted of daily ground telemetry tracking and aerial monitoring approximately every 2 weeks from fixed-wing aircraft on winter range and bi-weekly aerial monitoring on summer range. Once a mortality signal was detected, deer were located and necropsied to assess cause of death. We estimated weekly survival using the staggered entry Kaplan-Meier procedure (Kaplan and Meier 1958, Pollock et al. 1989). Capture-related mortalities (any mortalities occurring 52 �within 10 days of capture) and collar failures were censored from survival rate estimates. We estimated survival rates from 1 July 2011 through 30 June 2012 for adult females and from early December 2011mid June 2012 for fawns. Adult Female Body Measurements We applied ultrasonography techniques described by Stephenson et al. ( 1998, 2002) and Cook et al. (2001) to measure maximum subcutaneous rump fat (mm), loin depth (longissimus dorsi muscle, mm), and to estimate % body fat. We estimated a body condition score (BCS) for each deer by palpating the rump (Cook et al. 2001, 2007, 2009). We examined differences (P < 0.05) in nutritional status among study areas and between years using a two-sample I-test. We considered differences in body condition meaningful when mean rump fat or % body fat differed statistically between comparisons. Other body measurements recorded included pregnancy status (pregnant, barren) via blood samples, weight (kg), chest girth (cm), and hind-foot length (cm). Abundance Estimates We conducted 4 (North Ridge, North Magnolia) or 5 (Ryan Gulch, South Magnolia) helicopter mark-resight surveys (2 observers and the pilot) during late March/early April, 2012 to estimate deer abundance in each of the 4 study areas. We delineated each study area from GPS locations collected on winter range during the first 3 years of the study (Jan 2008 through April 2011 ). Two aerial fixed-wing telemetry surveys/study area were conducted during helicopter mark-resight surveys to determine which marked deer were within each survey area, and we confirmed adult female locations during surveys from GPS data acquired April 2012. We delineated flight paths in ArcGIS 9.3 prior to surveys following topographic contours (e.g., drainages, ridges) and approximating 500-600 m spacing throughout each study area; flight paths during surveys were followed using GPS navigation in the helicopter. Two approximately 12 x 12 cm pieces of Ritchey livestock banding material (Ritchey Livestock ID, Brighton, CO USA) were uniquely marked using color, number, and symbol combinations and attached to each radio-collar to enhance mark-resight estimates. Each deer observed during surveys. was recorded as mark ID#, unmarked, or unidentified mark. wl 'wl 'wl ~ We used program MARK (White and Burnham 1999), applying the immigration-emigration mixed logit-nonnal model (McClintock et al. 2008), to estimate mule deer abundance and confidence intervals. For mark-resight model evaluations, we examined parameter combinations of varying detection rates with survey occasion and whether individual sighting probabilities (i.e., individual heterogeneity) were constant or varied (cr2- = 0 or* 0). Model selection procedures followed the information-theoretic approach of Burnham and Anderson (2002). RESULTS AND DISCUSSION Deer Captures and Survival The helicopter crew captured 264 fawns and 138 does in Dec and Jan 2011 and 142 does during March 2012. Seventeen fawn mortalities (6.4%; ultimate cause= 6 capture myopathy, 10 predation, 1 vehicle collision) occurred within the 10 day myopathy period. Doe mortalities totaled 5 (3 .1 %; ultimate cause= 4 capture myopathy, 1 vehicle collision) and 7 (4.9%; all capture myopathy) within 10 days of the Dec and Jan and March capture periods, respectively. Mortality rates 10 days post capture have varied between 2-3% for fawns and 0-3% for does since Jan 2008, but were higher this year. Dry conditions and abnormally high dust from pipeline construction relative to previous years may be related. 'wl 'wl 'wl Fawn survival from early December 2010 through mid June 2011 was similar (P > 0.05) among study areas ranging from 0.60 to 0.75 (Table 1; all areas combined= 0.69, 95% CI= 0.63-0.74, n = 247). General comparisons to previous years suggest relatively high fawn survival during winter 2009-2010 and relatively low survival during winter 2010-2011 (Fig. 2), which correlates to some degree to winter 53 'wl I._/ �'SI 'cl 'cl -..I 'Cl severity. Exceptions include North Ridge, which has been stable throughout, and Ryan Gulch where relatively low precision of estimates do not allow statistical discrimination (Fig. 2). Annual adult female survival varied from 0.68 (North Magnolia) to 0.93 (Ryan Gulch; Table 1) this year and was comparable among study areas during 2011/12 and to previous years (P > 0.05) with the exception of North Magnolia deer exhibiting lower survival this year than during 2009/10 (Anderson and Bishop 20 I 0) and lower than Ryan Gulch this year. The relatively low adult female survival from North Magnolia may result in declining population trends if low survival persists. Spring Migration Patterns Collaboration with Idaho State University to direct a graduate student to address mule deer migration patterns in developed and undeveloped landscapes (funded from energy company contributions) has recently been completed. Two manuscripts have been prepared for publication; one is in review and the second has recently been accepted for publication (Lendrum et al. 2012). In addressing habitat selection during spring migration, Lendrum et al. (2012; Fig. 3) noted that mule deer migrating through the most developed landscapes exhibited longer step lengths (straight line distance between GPS locations) and selected habitats providing greater security cover versus more open areas with increased foraging opportunities through undeveloped landscapes. Migrating deer also selected areas closer to well pads, but avoided roads except in the highest developed areas where road densities may be too high for avoidance without significant deviations from traditional migration routes. These results suggest that deer may avoid disturbance where feasible or increase their rate of travel through highly developed landscapes where the energetic cost of avoidance may be too high. Mule Deer Body Condition Early-winter body condition measurements of adult female mule deer December 2011 were higher for deer from Ryan Gulch and North Ridge than previous years (P < 0.05), but were comparable for the North and South Magnolia deer (P > 0.05; Table 2). Comparisons among study areas in December suggested Ryan Gulch deer were in better condition that the other 3 areas. By late winter, however, body condition declined and deer from all study areas exhibited similar condition (Table 2). Improved condition of deer arriving on winter range was expected in December because of improved moisture conditions during spring and summer 2011. We were surprised that condition of North and South Magnolia deer did not mimic deer from the other 2 study areas, especially since there is summer ranges overlap with North Ridge and North Magnolia and Ryan Gulch and South Magnolia, respectively (Fig. 3). It was also surprising that deer from all study areas did not maintain higher condition by late winter given the mild winter conditions that were evident during 2011-2012, as was the case for North and South Magnolia deer during the mild winter of 2009-2010 (Table 2). Slightly higher late winter condition estimates were evident from all areas compared to 2009 and 2011, but these differences were not statistically significant (P > 0.05). December fawn weights were comparable to previous years and among study areas last year, with the exception of Ryan Gulch females which showed improvement over the previous year (Fig. 4). More detailed analyses will be conducted to identify factors potentially attributing to these observations. Neonate Survival To complete demographic parameters addressing mule deer-energy development interactions, CPW, Colorado State University, and ExxonMobil Production entered into a collaborative agreement to investigate neonate mule deer survival in developed and undeveloped landscapes (funded by ExxonMobil Production Co.). Mark Peterson (GRA) and Paul Doherty (CSU professor) will be assisting with this research, which began March 2012 and will continue for 3 years. To initiate this component of the study, we targeted 30 adult female mule deer/study area to receive Vaginal Implant Transmitters (VITs) during March 2012. Pregnancy rates during March were normal ranging from 96% to 98%/study area (n = 2846/area). March fetal counts ranged from 1.54 in South Magnolia to 1.92 in North Magnolia. We located 100 does with VITs and 97 neonates at parturition sites, with 85 neonates receiving radiocollars. Neonate 54 �survival will be monitored from June through December each year and compared among study areas relative to energy development activities. Mule Deer Population Estimates Mark-resight models that best predicted abundance estimates (lowest AICc; Burnham and Anderson 2002) exhibited variable sightability across surveys (P,) for all study areas and homogenous individual sightability (cr2 = 0) for North Ridge and South Magnolia deer and variable individual sightability (cr2 if:. 0) for North Magnolia and Ryan Gulch deer. North Ridge exhibited the highest deer density (18.3/km2), with comparably lower deer densities in the other 3 areas (7.4-9.2/krn2; Table 3, Fig. 5). Populations appear stable over the 4 year monitoring period exhibiting annual variation less than the error around point estimates, with the exception of North Magnolia which exhibited a positive increase in 2011 from the previous 2 years (Fig. 5). Abundance estimates from 2012 were similarly precise from all 4 study areas with the mean Confidence Interval Coefficient of Variation (CICV) ranging from 0.13-0.17. Magnolia Habitat Treatments In proceeding with mule deer habitat improvements in heavy (South Magnolia) and light developed areas (North Magnolia), we completed pilot habitat treatments in January 2011 (116 acres total; Anderson and Bishop 2011) and January 2012 (54 acres) to assess mechanical treatment methods (hydro-ax, roller-chop, chain). All required NEPA surveys were completed this summer for the remaining sites (Fig. 6). The Biological Assessment should be completed by September 2012, allowing the remaining 1,030 acres to be treated using hydro-ax during fall-winter 2012-2013. Vegetation response in the pilot treatment sites was promising by fall 2011 (Fig. 6), likely due to the moist conditions present during the previous spring and summer. Dryer conditions this spring inhibited a similar response, but treatments completed last January exhibited surprisingly good grass and forb growth; shrub response wasn't as vigorous as the previous year. All expenses addressing these habitat treatments will be covered through a Wildlife Management Plan agreement between CPW and ExxonMobil Production/XTO energy. SUMMARY AND COLLABORATIONS The goal of this study is to investigate habitat treatments and energy development practices that enhance mule deer populations exposed to extensive energy development activity. The information presented here provides data describing mule deer population parameters from the first 4.5 years of the pre-treatment period of a long-term study intended to address how mule deer react to landscape scale habitat and human activity modifications. The pretreatment period will continue through this fall to provide baseline data to compare against intended improvements in habitat conditions and evaluation of concentration and/or reduction in human development activities. Post-treatment monitoring will continue for 4-6 years to provide sufficient time to measure how deer respond to these changes. Based on the data collected thus far, deer from all areas appear to be in reasonably good condition and are exhibiting expected survival rates relative to changes in winter severity. We will continue to collect the various population and habitat use data across all study sites to evaluate the effectiveness of habitat improvements on winter range. This approach will allow us to determine whether it is possible to effectively mitigate development impacts in highly developed areas, or whether it is better to allocate mitigation dollars toward less or non-impacted areas. In a recent project conducted on the Uncomphahgre Plateau, Bergman et al. (2009) found that habitat treatments implemented in pinyon-juniper habitat in undeveloped areas were effective for deer. We are also evaluating deer behavioral responses to varying levels of development activity. This will allow us to assess the effectiveness of certain BMPs for reducing disturbance to wintering mule deer. Hay field improvements have been completed in the North Magnolia study area by Williams Production LMT Co. to fulfill a Wildlife Management Plan agreement with CPW; elk response has already been evident and mule deer response will continue to be monitored. Additional collaboration 55 'w �'cs/ \/ft!) with Williams Production LMT Co. has produced a clustered development plan recently implemented in the Ryan Gulch study area and new technologies will be implemented to reduce human activity through remote monitoring of well pads and fluid collection systems. Recent collaboration agreements with ExxonMobil Production Co. and Colorado State University have provided graduate research opportunities to enhance data collection and inference about mule deer-energy development interactions. Additional funding and cooperative agreements will be necessary to sustain this project through completion (at least 2017 and preferably through 2019). We pptimistically anticipate the opportunity to work cooperatively toward developing solutions for allowing the nation's energy reserves to be developed in a manner that benefits wildlife and the people who value both the wildlife and energy resources of Colorado. LITERATURE CITED Anderson, C.R., Jr. 2009. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and D. J. Freddy. 2008a. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Final Study Plan, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and D. J. Freddy. 2008b. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation-Stage I, Objective 5: Patterns o(mule deer distribution & movements. Pilot Study, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and C. J. Bishop. 2010. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and C. J. Bishop. 2011. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA. Bartmann, R. M. 1975. Piceance deer study-population density and structure. Job Progress Report, Colorado Divison of Wildlife, Fort Collins, Colorado, USA. Bartmann, R. B., and S. F. Steinert. 1981. Distribution and movements of mule deer in the White River Drainage, Colorado. Special Report No. 51, Colorado Division of Wildlife, Fort Collins, Colorado, USA. Bartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monograph No. 121. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin I 0: 108-114. Bergman, E. J., C. J. Bishop, D. J. Freddy, and G. C. White. 2009. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, USA. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multi-model inference: a practical information-theoretic approach. Second edition. Springer-Verlag, New York, New York, USA. Cook, R. C., J. G. Cook, D. L. Murray, P. Zager, B. K. Johnson, and M. W. Gratson. 2001. Development of predictive models of nutritional condition for rocky mountain elk. Journal of Wildlife Management 65 :973-987. Cook, R. C., T. R. Stephenson, W. L. Meyers, J. G. Cook, and L.A. Shipley. 2007. Validating predictive models of nutritional condition for mule deer. Journal of Wildlife Management 71:1934-1943. Cook, R. C., J. G. Cook, T. R. Stephenson, W. L. Meyers, S. M. McCorquodale, D. J. Vales, L. L. Irwin, P. Briggs Hall, R. D. Spencer, S. L. Murphie, K. A. Schoenecker, P. J. Miller. 2009. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife Management 74:880-896. 56 �Gibbs, H. D. 1978. Nutritional quality of mule deer foods, Piceance Basin, Colorado. Thesis, Colorado State University, Fort Collins, Colorado, USA. Kaplan, E. L., and P. Meier. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 52:457-481. Lendrum, P. E., C.R. Anderson, Jr., R. A. Long, J. K. Kie, and R. T. Bowyer. 2012. Habitat selection by mule deer during migration: effects of landscape structure and natural gas development. Ecosphere In press. Mcclintock, B. T., G. C. White, K. P. Burnham, and M.A. Pride. 2008. A generalized mixed effects model of abundance for mark-resight data when sampling is without replacement. Pages 271289 in D. L. Thompson, E.G. Cooch, and M. J. Conroy, editors, Modeling demographic processes is marked populations. Springer, New York, New York, USA. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. C. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Stephenson, T. R., V. C. Bleich, B. M. Pierce, and G. P. Mulcahy. 2002. Validation of mule deer body composition using in vivo and post-mortem indices of nutritional condition. Wildlife Society Bulletin 30:557-564. Stephenson, T. R., K. J. Hundertmark, C. C. Swartz, and V. Van Ballenberghe. 1998. Predicting body fat and mass in moose with untrasonography. Canadian Journal of Zoology 76:717-722. Unsworth, J. W., D. F. Pack, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63 :315-326. Van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked individuals. Bird Study 46:120-139. White, C. C., and B. C. Lubow. 2002. Fitting population models to multiple sources of observed data. Journal of Wildlife Management 66:300-309. Prepared by_ _ _ _ _ _ _ _ _ _ _ _ _ _ __ Charles R. Anderson, Jr., Wildlife Researcher 57 �Table I. Survival rate estimates (S) of fawn (3 Dec. 2011-18 June 2012) and adult female (I July 201130 June 2012) mule deer from 4 winter range study areas of the Piceance Basin in northwest Colorado. Cohort Study area Initial sample size (n) March doe sample8 (n) S(95% CI) Fawns Ryan Gulch 57 0.600 (0.466-0.734) South Magnolia 55 0.745 (0.630-0.861) North Magnolia 56 0.721 (0.601-0.842) North Ridge 73 0.681 (0.578-0.784) Adult females Ryan Gulch 44 67 0.927 (0.858-0.997) South Magnolia 30 45 0.903 (0.810-0.997) North Magnolia 31 49 0.683 (0.536-0.830) North Ridge 35 60 0.803 (0.698-0.908) aAdult female sample sizes following capture and radio-collaring efforts March, 2012. 58 �Table 2. Mean rump fat (mm), Body Condition Score (BCS8), and % body fat (% fat) of adult female mule deer from 4 study areas in the Piceance Basin of northwest Colorado, March and December, 2009-2012. Values in parentheses= SD. December 2009 March2009 % fat BCS Rump fat BCS March 2010 % fat Rump fat BCS %fat Study Area Rump fat Ryan Gulch 1.73 (1.78) 2.66 (0.55) 7.54 (1.80) 8.35 (6.36) 4.06 (1.13) 12.96 (4.53) 2.31 (1.44) 2.35 (0.48) 6.69 (1.58) South Magnolia 1.47 (0.68) 2.50 (0.60) 7.26 (1.82) 10.05 (6.19) 4.07 (1.21) 13.46 (4.96) 3.12 (2.20) 2.64 (0.59) 7.70 (2.01) North Magnolia 1.30 (0.79) 2.56 (0.68) 6.96 (2.23) 10.67 (5.76) 4.25 (0.96) 13.92 (3.92) 3.15 (2.34) 2.85 (0.53) 8.28 (1.86) North Ridge 1.57 (1.22) 2.60 (0.56) 7.28 (1.66) 5.25 (5.65) 3.63 (1.11) 11.02 (4.54) 1.77 (1.11) 2.42 (0.49) 6.83 (1.50) March 2011 December 2011 Table 2. Continued. December 2010 BCS % fat Rump fat BCS % fat Rump fat BCS %fat Study Area Rump fat Ryan Gulch 7.75 (6.15) 3.34 (0.98) 10.82 (4.32) 1.55 (0.60) 2.53 (0.42) 7.05 (1.20) 13.41 (6.93) 4.21 (1.17) 13.17 (3.64) South Magnolia 9.85 (6.78) 3.30 (0.61) 11.21 (3.32) 1.65 (0.75) 2.35 (0.50) 6.56 (1.49) 7.53 (4.66) 3.37 (0.76) North Magnolia 9.55 (6.49) 2.56 (0.68) 11.65 (4.86) 1.65 (0.67) 2.53 (0.49) 7.06 (1.35) 9.43 (6.41) 3.79 (0.93) 11.15 (3.57) North Ridge 6.14 (5.29) 3.32 (0.82) 10.32 (3.39) 1.45 (0.76) 2.24 (0.49) 6.24 (1.45) 9.81 (5.81) 3.62 (1.00) 11.22 (3.38) 9.95 (2.73) 59 ( CCC( ( { ( C{ ( C< ( ( ( CC( CCCC{ { { CC( C( CCC{ { ( CC( C{ C{ CC �Table 2. Continued. March 2012 Study Area Rump fat Ryan Gulch 2.15 (1.44) 2.74 (0.44) 7.22 (1.16) South Magnolia 1.71 (0.76) 2.58 (0.36) 6.97 (1.12) North Magnolia 1.87 (0.78) 2.85 (0.33) 7.65 (0.94) North Ridge 2.24 (1.58) 2.70 (0.35) 7.26 (1.05) 8 BCS % fat Body condition score taken from palpations of the rump following Cook et al. (2009). 60 �Table 3. Mark-resight abundance (N) and density estimates of mule deer from 4 winter range herd segments in the Piceance Basin, northwest Colorado, 27 March-4 April 2012. Data represent 4 helicopter resight surveys from North Ridge and North Magnolia and 5 resight surveys from Ryan Gulch and South Magnolia. Study area Mean No. sighted Mean No. marked N(95% CI) Density (deer/km2} Ryan Gulch 268 24 1,048 (897-1,243) 7.4 South Magnolia 161 25 630 (556-724) 7.6 North Magnolia 267 32 727 (648-840) 9.2 North Ridge 319 34 972 (862-1, 113) 18.3 ~ 61 �-- ...., Mule Deer VVinter Range Study Area Boundaries Well pads and facilities D Ryan Gulch 0 3 [iJ Development faci lities 6 12 Miles Figure l. Mule deer winter range study areas relative to acti ve natural gas well pads and energy development facilities in the Piceance Basin of northwest Colorado, summer 20 12 (Accessed http://cogcc.state.co.us/ Aug. 8, 20 12). - - Nor1h Magnolia ! In deve lopment or application for drilling ! Prod ucin g well [ l South Magnolia J; Injection well - -- Norlh Ridge 62 �Ryan Gulch fawn S- 2008/09-2010/11 South Magnolia fawns - 2008/09-2010/11 1.00 T -~~~~~~;;;;;;;==-=-------------- 1.00 -----· O.CJO i:E r 0.80 •·---·--·--·•··•----·-·---·0.70--0.60 ; 0.50 T:--------'-'-....-;;:-~......,:.,r::::='miimi,iiimiimir •••••••••••••••••••••••• . ::::::~.:.:;;;:::::~::·== 060 ~ ...... o:so : 1 - - - - - - - •••• •········ •• ••. 0.40 ~:_ _ _ _ _ _ ____;=---•• -•••- ••-•••- ••- •• -• ••- ••-•••- ••-.. -• ••- ..-•• -. . 1- - - - - - - - - 0.30 +------- 0.40 ~-_ _ _ _ _ _ _ _ _ _ _ _ _..:....,..,,..-. ..-..-..-. . 0.30 .;....;- - - - - - - - - - - - - - - - - 0.20 I 0,10 I 0.00 ! 0.20 ~:- - - - - - - - - - 0.10 + 0.00 : ! I ; i ; ~ ~ ~ ~ to ~ to ~ ~ ~ ~ ~ ~ II, ~ "'.., ~ ~'>j #li'Ji J> ~~ .....~ ,? w"' 'ti... ~-,; """"t ~"' ~ ~... _,I,, ,,,,~ \'Ii ~.... ~fc _:I)... ~~- i't ~~ " ,_"l) ..,,, -A... »~ ..e.""t ~ b ~ to-:. ~? ~,, , ~ to ~ .J'to ~ ~'J, ~ ~ ~~ tc~... ,_'ti ~"'... t/' l? ''Ii ,f.... ~,,,-.. '"' 'c~ ~ ~ ~ ~ ~ ~,, 'ti...!lo ~~ to.., ~'J,, ~!lo ~... ~l I .,il) ~~ ~,..., ~ol l•.., ~"' lfJ'lfJ' ~ '"'l ~ ~ '\)., ~"' 'tXr:/l t1-·~~,.,~ ~ ~~- ~ w' North Magnolia fawn s-2008/09-2010/11 North Ridge fawns - 2008/09-2010/11 1.00 . ,.... 0.90 •·-··------.!,..!ll.. 0.80 r--~·-··--.. --.......,._,.,......- !.!;:;~;_:~:~,:~'"• • ~:: r·······- ·····-·-·•··· .... 0 so .I I • ••••• •h ~:::•. ~-;.-;,~-;.-.-;....._ _ _ _ _ • - •••••••• ·• •••••• ·---·-···--· o:40 ; 0.30 ! 0.20 1 0.10 i 0.00 l r I ~to ~ :'> to ,.,:; to ~ ~ II, ,.'o ~ .., II, 'o 'd.., ~ ~~ #Ii~ i ';,:~ "',,, to... ~., 'd.., ~"" ,,,, ~... _,J.,~ ,--. ~.... ~<,, e... ~ ~... ~~ _A, _A,"l) ~ .,J,..,. -:,~· ,,... (1'" (1'" ,~ ~,,, '< ~ ~"' ~'I§ '1""" 'ti' ~ ~"O' ~ ~ ~ e ~., -°"' '"' <l ~'b-\ ~ Figure 2. Over-winter (Dec-Mar & June) mule deer fawn survival (5) from 4 study areas in the Piceance Basin, northwest Colorado, 2008/09 (red lines), 2009/10 (orange lines), 2010/11 (blue lines), and 2011/12 (black lines). Solid lines= Sand dashed lines= 95% CI. Comparable data among years DecemberMarch 2008-2009 and 2009-2010 due to premature collar drop and December-mid-June 2010-2011 and 2011-2012. 63 �- - North Ridge and North Magnolia Summer Range -- Colorado - 'W Figure 3. Mu le deer study areas in the Piceance Basin of northwestern Colorado, USA (Top), spring 2009 migration routes of adult female mule deer (n = 52; Lower left), and active natural-gas well pads (black dots) and roads (state, county, and natural-gas; white lines) from May 2009 (Lower right; from Lendrum et al. 20 12). - 64 �Male fawn weights 42.0 40.0 _ 38.0 C Ryan Gulch 0.0 ..::,:: ... ]> 36.0 C South Magnolia Q) II North Magnolia 5 34.0 C North Ridge 32.0 - 30.0 Dec 2008 Dec 2009 Dec 2010 Dec 2011 Female fawn weights _ 38 .0 - i - - - - - - - -- -- - - - -- - 0.0 ..::,:: ... ]> 36.0 C Ryan Gulch C South Magnolia cu ■ North Magnolia 5 34.0 CNorth Ridge 32.0 Dec 2008 Dec 2009 Dec 2010 Dec 2011 Figure 4. Mean male and female fawn weights and 95% CI (error bars) from 4 mule deer study areas in the Piceance Basin, northwest Colorado, December 2008- 201 1. 65 - �Late winter mule deer density 30.0 25.0 .... E 20.0 --- --- - t - - - - - + - - - - - - - , . - - · - - -- ~ '41111 "'cu;::- ts.o --- --- - North Ridge ••••••• Ryan Gulch cu C 10.0 - • North Magnolia - South Magnolia 5.0 0.0 2009 2010 2011 2012 Year Figure 5. Mule deer density estimates and 95% CI (error bars) from 4 winter range herd segments in the Piceance Basin, northwest Colorado, late winter 2009-2012. 66 �- North Magnolia treatement sites (587 acres) LJ BearSet_l 5_35b_ andG i BearSet_ l _8andA_E - LJ BearSet_36_54andJ •• '] GreasewoodSet_g16_g29 GreasewoodSet_g1_g 15 [ J GreasewoodSet_g30_g4 2 LeeOversights_a_fand 16_ 17 Mechanical treatment comparison (54 acres) North Hatch PIiot Treatments ( 116 acres) Mule Deer Study Areas North Magnolia Figure 6. Habitat treatment site delineations in 2 mule deer study areas (600 acres each) of the Piceance Basin, northwest Colorado (Top; cyan and yellow polygons have been completed and remaining sites are scheduled for treatment fall/winter 201 2/ I 3). January 2011 hydro-ax treatment-site photos from North Hatch Gulch during April (Lower left, aerial view) and October, 20 LI (Lower right, ground view). 67 - - �Colorado Parks and Wildlife July I, 2012 - June 30, 2013 WILDLIFE RESEARCH REPORT State of_ _ _ _ ______;:C::;..;o=lo.;;;.;r=a=d=-o_ _ _ _ _ : ""'"P=ar=k=-s-=an=d=---W____i=ld=h--·fe____________ Cost Center 3430 : =M=amm==al=s-=R=e=s=ear=c=h_ _ _ _ _ _ _ _ _ _ __ Work Package 3001 : =D__e__ er__C=on=s;;;..;;e.....rv.....a__t__ io__n_______________ Task No. 6 : Population Performance of Piceance Basin Mule Deer in Response to Natural Gas Resource Extraction and Mitigation Efforts to Address Human Activity and Habitat Degradation Federal Aid Project:__W.. . . . ,;-___ 1 8__5__ -R_ _ _ __ Period Covered: July 1, 2012 - June 30, 2013 \e,I 'el Author: C. R. Anderson, Jr. Personnel: N. Bellerose, E. Bergman, C. Bishop, E. Cato, A. Collier, D. Collins, B. deVergie, S. Eno, D. Finley, M. Fisher, B. Frankland, L. Gepfert, T. Gettelman, M. Grode, T. Jenkins, D. Johnston, T. Knowles, M. Melham, J. Matijas, S. Nagy, B. Petch, J. Rivale, R. Schilowsky, R. Velarde, L. Wolfe, CPW; E. Hollowed, L. Belmonte, BLM; D. Freddy, Hoch Berg Enterprises; T. Graham, Ranch Advisory Partners; M. Wille, T & M Contractors.; P. Lendrwn, T. Bowyer, Idaho State University; P. Doherty, J. Northrup, M. Peterson, G. Wittemyer, K. Wilson, Colorado State University; R. Swisher, S. Swisher, Quicksilver Air, Inc.; D. Felix, Olathe Spray Service, Inc.; L. Coulter, Coulter Aviation. Project support received from Federal Aid in Wildlife Restoration, Colorado Mule Deer Association, Colorado Mule Deer Foundation, Colorado State Severance Tax Fund, EnCana Corp., ExxonMobil Production Co./XTO Energy, Marathon Oil Corp., Shell Petroleum, and WPX Energy. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the authors. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT 'Cl .., We propose to experimentally evaluate winter range habitat treatments and human-activity management alternatives intended to enhance mule deer (Odocoileus hemionus) populations exposed to energy-development activities. The Piceance Basin of northwestern Colorado was selected as the project area due to ongoing natural gas development in one of the most extensive and important mule deer winter and transition range areas in Colorado. The data presented here represent the first 5 pretreatment years of a long-term study addressing habitat improvements and evaluation of energy development practices intended to improve mule deer fitness in areas exposed to extensive energy development. We monitored 4 winter range study areas representing varying levels of development to serve as treatment (North Magnolia, South Magnolia) and control (North Ridge, Ryan Gulch) sites and recorded habitat use and movement patterns using GPS collars (~5 location attempts/day), estimated overwinter fawn and annual adult female survival, estimated early and late winter body condition of adult females using ultrasonography, and estimated abundance using helicopter mark-resight surveys. During this research segment, we targeted 280 fawns (60-80/study area) and 170 does (30-70/study area) in early December 29 �2012 for VHF and GPS radiocollar attachment, respectively, and 140 does in March 2013 (30-40/study area) for late winter body condition assessment. Winter range habitat improvements resulting in 604 acres of mechanically treated pinion-juniper/mountain shrub habitats in each of the 2 treatment areas were completed April 2013. Post-treatment monitoring will continue for 4-6 years to provide sufficient time to measure how deer respond to these changes. Based on data collected during the pretreatment phase: ( 1) annual adult survival was consistent among areas averaging 80-84% annually, but overwinter fawn survival was more variable ranging from 48% to 85% within study areas, with annual and study area differences primarily due to annual weather conditions and in some cases density dependent influences; (2) migratory mule deer selected increased cover and increased their rate of travel through developed areas, but did not avoid development structures and avoided negative influences through behavioral shifts in timing and rate of migration; (3) mule deer body condition early and late winter was generally consistent within areas, with higher variability among study areas early winter, which likely relate to seasonal moisture within areas and relative forage capacity among areas; (4) mule deer densities appeared to increase in 3 of 4 areas, with a recent decline in North Ridge, but the most recent North Ridge density was comparable to the first 2 years of the study. Detailed habitat use analyses are still pending for the pretreatment period. We will continue to collect population and habitat use data across all study sites to evaluate the effectiveness of habitat improvements on winter range. This approach will allow us to determine whether it is possible to effectively mitigate development impacts in highly developed areas, or whether it is better to allocate mitigation dollars toward less or non-impacted areas. In collaboration with Colorado State University, we are also evaluating deer behavioral responses to varying levels of development activity in the Ryan Gulch study area and neonate survival in relation to energy development from all study areas. This will allow us to assess the effectiveness of certain Best Management Practices (BMPs) for reducing disturbance to deer and include neonatal data to other demographic parameters for evaluation of mule deer/energy development interactions. The study is slated to run through at least 2017, but extending the study through 2019 is preferable to adequately measure mule deer population responses to landscape level manipulations. w 1w w 30 �WILDLIFE RESEARCH REPORT POPULATION PERFORMANCE OF PICEANCE BASIN MULE DEER IN RESPONSE TO NATURAL GAS RESOURCE EXTRACTION AND MITIGATION EFFORTS TO ADDRESS HUMAN ACTMTY AND HABITAT DEGRADATION CHARLES R. ANDERSON, JR PROJECT NARRITIVE OBJECTIVES 1. To determine experimentally whether enhancing mule deer habitat conditions on winter range elicits behavioral responses, improves body condition, increases fawn survival, or ultimately, population density on mule deer winter ranges exposed to extensive energy development. 2. To determine experimentally to what extent modification of energy development practices enhance habitat selection, body condition, fawn survival, and winter range mule deer densities. SEGMENT OBJECTIVES 1. Collect and reattach GPS collars to maintain sample sizes for addressing mule deer habitat use and behavior patterns in 4 study areas experiencing varying levels of energy development of the Piceance Basin, northwest Colorado. 2. Estimate early and late winter body condition of adult female mule deer in each of the 4 winter herd segments using ultrasound techniques. 3. Monitor over-winter fawn and annual adult female mule deer survival by daily ground tracking and biweekly aerial tracking. 4. Conduct Mark-Resight helicopter surveys to estimate mule deer abundance in each study area. 5. Complete habitat treatments for assessing efficacy of habitat improvement projects to mitigate energy development disturbances to mule deer. 6. Continue neonate survival evaluations to complete demographic parameters for assessing mule deer/energy development interactions. INTRODUCTION Extraction of natural gas from areas throughout western Colorado has raised concerns among many public stakeholders and Colorado Parks and Wildlife (CPW) that the cumulative impacts associated with this intense industrialization will dramatically and negatively affect the wildlife resources of the region. Concern is especially high for mule deer due to their recreational and economic importance as a principal game species and their ecological importance as one of the primary herbivores of the Colorado Plateau Ecoregion. Extraction of natural gas will directly affect the potential suitability of the landscape used by mule deer through conversion of native habitat vegetation with drill pads, roads, or noxious weeds, by fragmenting habitat because of drill pads and roads, by increasing noise levels via compressor stations and vehicle traffic, and by increasing the year-round presence of human activities. Extraction will indirectly affect deer by increasing the human work-force population of the region resulting in the 31 �need for additional landscape for human housing, supporting businesses, and upgraded road/transportation infrastructure. Additionally, increased traffic on rural roads will raise the potential for vehicle-animal collisions and additive direct mortality to mule deer populations. Thus, research documenting these relationships and evaluating the most effective strategies for minimizing and mitigating these activities will greatly enhance future management efforts to sustain mule deer populations for future recreational and ecological values. The Piceance Basin in northwest Colorado contains one of the largest migratory mule deer populations in North America and also exhibits some of the largest natural gas reserves in North America. Projected energy development throughout northwest Colorado within the next 20 years is expected to reach about 15,000 wells, many of which will occur in the Piceance Basin, which currently supports over 250 active gas well pads (http://cogcc.state.co.us; Fig. 1). Anderson and Freddy (2008a) in their longterm research proposal identified 6 primary study objectives to assess measures to offset impacts of energy extraction on mule deer population performance. During the past 5 years, we gathered baseline habitat utilization and demographic data from radiocollared deer across the Piceance Basin to allow assessment of habitat mitigation approaches that were completed April 2013. We are currently monitoring 2 control areas: 1 with development (0.6 pads & facilities/km 2; Ryan Gulch) and 1 without (North Ridge). The control areas will be compared with 2 treatment areas experiencing similar development intensities (South Magnolia, 0.9 well pads & facilities/km 2 and North Magnolia, 0.1 well pads & facilities/km 2), that also recently received habitat improvements (604 acres each). Habitat and mule deer responses to mechanical habitat treatments will be evaluated over the next 4-6 years to assess the success of this habitat mitigation strategy to benefit mule deer exposed to energy development disturbance. In addition, mule deer behavior patterns in relation to energy development activities in the Ryan Gulch area are being monitored to identify effective Best Management Practices (BMPs) for future energy development planning. This progress report describes the previous 5.5 years (Jan 2008-June 2013) of mule deer population performance during the pretreatment phase on 4 winter range herd segments, which includes monitoring habitat selection and behavior patterns of adult female mule deer; spring/summer neonate, overwinter fawn and annual adult female survival; estimates of adult female body condition during early and late winter, and annual late-winter abundance/density estimates. ,._, 1w' STUDY AREAS The Piceance Basin, located between the cities of Rangely, Meeker, and Rifle in northwest Colorado, was selected as the project area due to its ecological importance as one of the largest migratory mule deer populations in North America and because it exhibits one of the highest natural gas reserves in North America (Fig. 1). Historically, mule deer numbers on winter range were estimated between 20,000-30,000 (White and Lubow 2002), and the current number of well pads (Fig.1) and projected number of gas wells in the Piceance Basin over the next 20 years is about 250 and 15,000, respectively. Mule deer winter range in the Piceance Basin is predominantly characterized as a topographically diverse pinion pine (Pinus edulis)-Utahjuniper (Juniperus osteosperma; pinion-juniper) shrubland complex ranging from 1,675 m to 2,285 min elevation (Bartmann and Steinert 1981). Pinion-juniper are the dominant overstory species and major shrub species include Utah serviceberry (Amelanchier utahensis), mountain mahogany (Cercocarpus montanus), bitterbrush (Purshia tridentata), big sagebrush (Artemisia tridentata), Gamble's oak (Quercus gambelii), mountain snowberry (Symphoricarpos oreophilus), and rabbitbrush ( Chrysothamnus spp.; Bartmann et al. 1992). The Piceance Basin is segmented by numerous drainages characterized by stands of big sagebrush, saltbush (Atriplex spp.), and black greasewood (Sarcobatus vermiculatus), with the majority of the primary drainages having been converted to mixedgrass hay fields. Grasses and forbs common to the area consist of wheatgrass (Agropyron spp. ), blue grama (Boute/oua gracilis), needle and thread (Stipa comata), Indian rice grass (Oryzopsis hymenoides), arrowleafbalsamroot (Ba/samorhiza sagittata), broom snakeweed (Gutierrezia sarothreae), pinnate tansymustard (Descurainia pinnata), milkvetch (Astraga/us spp.), Lewis flax (Linum /ewisii), evening 32 -' �".-I primrose (Oenothera spp.), skyrocket gilia (Gilia aggregata), buckwheat (Erigonum spp.}, Indian paintbrush (Castilleja spp.}, and penstemon (Penstemon spp.; Gibbs 1978). The climate of the Piceance Basin is characterized by warm dry summers and cold winters with most of the annual moisture resulting from spring snow melt. Wintering mule deer population segments we investigated include: North Ridge (53 km2) just north of the Dry Fork of Piceance Creek including the White River in the northeastern portion of the Basin, Ryan Gulch ( 141 km2) between Ryan Gulch and Dry Gulch in the southwestern portion of the Basin, North Magnolia (79 km2) between the Dry Fork of Piceance Creek and Lee Gulch in the northcentral portion of the Basin, and South Magnolia (83 km2) between Lee Gulch and Piceance Creek in the south-central portion of the Basin (Fig. I). Each of these wintering population segments has received varying levels of natural gas development: no development in North Ridge, light development in North Magnolia (0.14 pads & facilities/km 2), and relatively high development in the Ryan Gulch (0.60 pads & facilities/km 2) and South Magnolia (0.86 pads & facilities/km2) segments (Fig. I). Among the 4 study areas, North Ridge has served as an unmanipulated control site, Ryan Gulch will serve to address humanactivity management alternatives (BMPs) that benefit mule deer exposed to energy development and as a developed control area for comparison to the developed treatment area receiving habitat improvements (South Magnolia), and North and South Magnolia will allow us to assess the utility of habitat treatments intended to enhance mule deer population performance in areas exposed to light (North Magnolia) and heavy (South Magnolia) energy development activities. 'Cl METHODS Tasks addressed this period included mule deer capture and collaring efforts, monitoring neonate and overwinter fawn and annual adult female survival, estimating adult female body condition during early and late winter using ultrasonography, estimating mule deer abundance applying helicopter markresight surveys, and working with BLM and the contractor to complete mechanical habitat treatments by spring 2013. We employed helicopter net-gunning techniques (Barrett et al. 1982, van Reenen 1982) to target 280 fawns in December 2012/January 2013, 170 adult females during early December 2012, and 140 adult females (mostly recaptures) during early March 2013. Once netted, all deer were hobbled and blind folded. Fawns were weighed, radio-collared and released on site, and adult females were transported to localized handling sites for recording body measurements and fitted with GPS collars (5 or 48 fixes/day; G211 OD, Advanced Telemetry Systems, Isanti, MN, USA) and released. To provide direct measures of decline in overwinter body condition, we targeted 30adult females in each study area that were captured the previous December; Vaginal Implant Transmitters (VITs) were also inserted to assist with neonate capture and collaring efforts spring 2013. Fawn collars were spliced and fitted with rubber surgical tubing to facilitate collar drop between mid-summer and early autumn, and GPS collars were supplied with timed drop-off mechanisms scheduled to release early in April of the year following deployment. All radio-collars were equipped with mortality sensing options (i.e., increased pulse rate following 4-8 hrs of inactivity). Mule Deer Habitat Use and Movements We downloaded and summarized data from GPS collars deployed from December 2011 through April 2013. GPS collars maintained the same schedule of attempting to collect locations every 5 hours, except in Ryan Gulch where location rates were programmed for every 30 minutes to increase resolution of movement data for evaluation of deer behavior patterns in relation to differing development activities. We plotted deer locations and recorded timing and distance of spring and fall 2012 migrations for each study area. Mule deer winter concentration areas were created using composite GPS data (March 2010 through April 2011 from all deer; 5 location attempts/day) from each study area and mapped in ArcGIS (ver. 9.3) using Spatial Analyst (kernel probability density functions separated by quantiles). Mule deer 33 �resource selection analyses are pending completion of high resolution habitat data layers currently being developed by BLM. Mule Deer Survival Mule deer mortality monitoring consisted of daily ground telemetry tracking and aerial monitoring approximately every 2 weeks from fixed-wing aircraft on winter range and bi-weekly aerial monitoring on summer range. Once a mortality signal was detected, deer were located and necropsied to assess cause of death. We estimated weekly survival using the staggered entry Kaplan-Meier procedure (Kaplan and Meier 1958, Pollock et al. 1989). Capture-related mortalities (any doe/fawn mortalities occurring within 10 days of capture) and collar failures were censored from survival rate estimates. We estimated survival rates from 1 July 2012 through 30 June 2013 for adult females, from birth to mid December for neonates, and from early December 2012-mid June 2013 for fawns. Adult Female Body Measurements We applied ultrasonography techniques described by Stephenson et al. (1998, 2002) and Cook et al. (2001) to measure maximum subcutaneous rump fat (mm), loin depth (longissimus dorsi muscle, mm), and to estimate% body fat. We estimated a body condition score (BCS) for each deer by palpating the rump (Cook et al. 2001, 2007, 2009). We examined differences (P < 0.05) in nutritional status among study areas and between years using a two-sample t-test. We considered differences in body condition meaningful when mean rump fat or % body fat differed statistically between comparisons. Other body measurements recorded included pregnancy status (pregnant, barren) via blood samples, fetal counts using ultrasonagraphy, weight (kg), chest girth ( cm), and hind-foot length (cm). Abundance Estimates We conducted 4 (North Ridge, North Magnolia, South Magnolia) or 5 (Ryan Gulch) helicopter mark-resight surveys (2 observers and the pilot) during early April to estimate deer abundance in each of the 4 study areas. We delineated each study area from GPS locations collected on winter range during the first 3 years of the study (Jan 2008 through April 2011). Two aerial fixed-wing telemetry surveys/study area were conducted during helicopter mark-resight surveys to determine which marked deer were within each survey area, and we confirmed adult female locations during surveys from GPS data acquired April 2013. We delineated flight paths in ArcGIS 9.3 prior to surveys following topographic contours (e.g., drainages, ridges) and approximating 500-600 m spacing throughout each study area; flight paths during surveys were followed using GPS navigation in the helicopter. Two approximately 12 x 12 cm pieces of Ritchey livestock banding material (Ritchey Livestock ID, Brighton, CO USA) were uniquely marked using color, number, and symbol combinations and attached to each radio-collar to enhance mark-resight estimates. Each deer observed during surveys was recorded as mark ID#, unmarked, or unidentified mark. We used program MARK (White and Burnham 1999), applying the immigration-emigration mixed logit-normal model (McClintock et al. 2008), to estimate mule deer abundance and confidence intervals. For mark-resight model evaluations, we examined parameter combinations of varying detection rates with survey occasion and whether individual sighting probabilities (i.e., individual heterogeneity) were constant or varied (cr2 = 0 or -:t:- 0). Model selection procedures followed the information-theoretic approach of Burnham and Anderson (2002). RESULTS AND DISCUSSION Deer Captures and Survival The helicopter crew captured 277 fawns in Dec 2012-Jan 2013, 165 does in Dec 2012, and 138 does during March 2013. Eight fawn mortalities (2.9%; ultimate cause= 3 capture myopathy, 5 predation) occurred within the 10 day censorship period. Doe mortalities totaled 4 (2.5%; all capture 34 1w �.._ myopathy) and 4 (2.9%; 3 capture myopathy, 1 predation) within 10 days of the December and March capture periods, respectively. Mortality rates, 10 days post capture, have varied between 2-3% for fawns and 0--3% for does since Jan 2008, except during the 2011-2012 capture season where myopathy rates were higher (3-6%) due to dry, warm conditions (Anderson and Bishop 2012). Fawn survival from early December 2012 through mid June 2013 was similar (P > 0.05) among 3 study areas ranging from 0.75 to 0.85, but was lower in North Ridge (0.53; Table 1). General comparisons to previous years suggest relatively high fawn survival occurred during winters 2009-2010 and 2012-2013, and relatively low survival during winter 2010--2011 (Fig. 2), which correlates to some degree to winter severity. North Ridge exhibited lower survival during 2012-2013 (Fig. 2), which appeared to be driven by density dependent rather than climatic factors. Annual adult female survival varied from 0.73 (North Ridge) to 0.86 (North Magnolia; Table 1) during 2012-2013 and was comparable among study areas during 2012-2013 and to previous years (P > 0.05), with the exception oflower survival in North Magnolia during 2011-2012 (S = 0.68, Anderson and Bishop 2012). Sample sizes for adult female survival do not allow statistical discrimination among years unless large differences are evident (e.g., >15-20%). Spring Migration Patterns Collaboration with Idaho State University to address mule deer migration patterns in developed and undeveloped landscapes (funded from energy company contributions) has recently been completed. Two manuscripts have been accepted for publication (Lendrum et al. 2012, Lendrum et al. 2013; Appendix A). In addressing habitat selection during spring migration, Lendrum et al. (2012; Fig. 3) noted that mule deer migrating through the most developed landscapes exhibited longer step lengths (straight line distance between GPS locations) and selected habitats providing greater security cover than deer in undeveloped landscapes that migrated through more open areas that provided increased foraging opportunities. Migrating deer also selected areas closer to well pads, but avoided roads, except in the highest developed areas where road densities were likely too high for avoidance without significant deviations from traditional migration routes. In the second manuscript (Lendrum et al. 2013), we addressed biological and environmental factors influencing spring migration and assessed how energy development influenced migratory behavior. Overall, spring migration was influenced by snow depth, temperature, and green-up on winter and summer range; increasing temperatures, snow melt and emerging vegetation dictated timing of winter range departure and summer range arrival. Duration of Piceance Basin mule deer migration was short, averaging 4-8 days among the 4 areas (straight line distance between seasonal ranges averaged 33 - 45 km). Deer in poor condition migrated later than deer in good condition, but condition was similar among areas regardless of development status. Migrating deer from developed study areas did not avoid development structures, but departed later, arrived earlier and migrated more quickly than deer from undeveloped areas. While large changes in timing of migration could have nutritional consequences and negatively influence reproduction and neonate survival, the relatively minor shift we observed should not result in long-term fitness consequences. Migratory deer in the Piceance Basin appear to avoid negative effects of energy development through behavioral shifts in timing and rate of migration. Mule Deer Body Condition Early-winter body condition measurements of adult female mule deer from North Ridge and Ryan Gulch were lower than from deer from North Magnolia (P < 0.05), but were comparable otherwise (P > 0.05, Fig. 4, Table 2). By late winter, however, body condition declined and deer from all study areas exhibited similar condition (Fig. 4, Table 2). These observations have been generally consistent throughout the study, where early winter condition is variable between study areas and typically follows 35 �the pattern of better condition in North and South Magnolia deer, respectively poorer condition in Ryan Gulch and North Ridge, and poor condition in all areas by late winter. Exceptions occurred during late winter 2010 and early winter 2011, where North and South Magnolia and Ryan Gulch and North Ridge deer, respectively, exhibited improved condition than during other time periods (Fig. 4). December fawn weights by study area were higher in Ryan Gulch during 2012-2013, but were lower and have declined recently in the other 3 study areas (Fig. 5). In general, seasonal moisture conditions appear to be driving differences in annual body condition within study areas, but other factors appear related to differences among study areas. We suspect density dependent factors (forage capacity relative to deer density) are related to observed differences in early winter body condition among study areas. More detailed analyses will be conducted to identify factors attributing to these observations. Neonate Survival To complete demographic parameters addressing mule deer-energy development interactions, CPW, Colorado State University, and ExxonMobil Production entered into a collaborative agreement to investigate neonate survival in developed and undeveloped landscapes (funded by ExxonMobil Production Co.) beginning spring 2012. Mark Peterson (ORA) and Paul Doherty (CSU professor) are assisting with this research, which will continue through 2014. Neonate capture and collaring efforts totaled 85 during spring 2012 and 67 during spring 2013. Estimated neonate survival through midDecember 2012 was 0.39 (95% CI= 0.28-0.50). Factors influencing neonate mule deer survival from developed and undeveloped landscapes will be addressed by late 2014. Mule Deer Population Estimates Mark-resight models that best predicted abundance estimates (lowest AICc; Burnham and Anderson 2002) exhibited variable sightability across surveys (P,) for all study areas and homogenous individual sightability ( cr2 = 0) for South Magnolia deer and variable individual sightability ( cr2 -;p 0) for the other 3 areas. North Ridge exhibited the highest deer density (16.1/km2), with comparably lower deer densities in the other 3 areas (8.9-10.4/km2; Table 3, Fig. 6). Populations appeared to increase over the 5 year monitoring period in 3 of the study areas (from 6.S/km2 to 10.1/km2), with a recent decline in North Ridge since 2011 (from 22.8/km2 to 16.1/km2); the current North Ridge density is comparable to the first 2 years of the study (Fig. 6). The recent North Ridge decline was likely related to density dependent factors, which were also evident in lower early winter body condition (Fig. 4) and a recent increase in malnutrition mortalities of adult females (from O in 2010-2011 to 7 in 2012-13). Abundance estimates from 2013 were similarly precise from all 4 study areas with the mean Confidence Interval Coefficient of Variation (CICV) ranging from 0.15-0.16. Magnolia Habitat Treatments We proceeded with habitat improvements in high (South Magnolia) and low development areas (North Magnolia) during 2012-2013. We completed pilot habitat treatments in January 2011 (116 acres total; Anderson and Bishop 2011; Environmental Assessment: DOI-BLM-CO-110-2011-004-EA), mechanical treatment method comparison treatments (hydro-ax, roller-chop, chain) in January 2012 (54 acres), and hydro-axe habitat treatments in April 2013 (434 acres; Determination of NEPA Adequacy: DOI-BLM-CO-110-2012-0134-DNA), totaling 604 treated acres in each study area (Fig. 7). Vegetation response in the pilot treatment sites was visually evident by fall 2011 (Fig. 7), likely due to the moist conditions during the previous spring and summer. Early spring 2013 moisture has resulted in good vegetative responses from the most recently treated sites. Vegetation and mule deer responses will be documented for the next 4-6 years to assess the utility of this mitigation approach in benefiting mule deer exposed to energy development disturbance. All expenses addressing these habitat treatments will be covered through a Wildlife Management Plan agreement between CPW and ExxonMobil Production/XTO energy. 36 '._I �SUMMARY AND COLLABORATIONS The long-term goal of this study is to investigate habitat treatments and energy development practices that enhance mule deer populations exposed to extensive energy development activity. The information presented here summarizes mule deer population parameters from the first 5.5 years of the pre-treatment period. The pretreatment period was completed during spring 2013, providing baseline data for comparison with intended improvements in habitat conditions and reduction in human development activities. Winter range habitat improvements resulting in 604 acres of mechanically treated pinionjuniper/mountain shrub habitats in each of 2 study areas were completed April 2013. Post-treatment monitoring will continue for 4-6 years to provide sufficient time to measure how deer respond to these changes. Based on data collected prior to habitat improvements (i.e., pretreatment phase): (1) annual adult survival was consistent among areas averaging 80-84% annually, but overwinter fawn survival was variable, ranging from 48% to 85% within study areas, with annual and study area differences primarily due to annual weather conditions and in some cases density dependent influences; (2) migratory mule deer selected for areas with increased cover and increased their rate of travel through developed areas, and avoided negative influences through behavioral shifts in timing and rate of migration, but did not avoid development structures; (3) mule deer body condition early and late winter was consistent within areas, with higher variability among study areas early winter, which was likely related to seasonal moisture within areas and relative forage capacity among areas; (4) mule deer densities appear to be increasing in 3 of 4 areas, with a recent decline in North Ridge, but the current North Ridge density is comparable the first 2 years of the study. Detailed habitat use analyses are pending for the pretreatment period. We will continue to collect the various population and habitat use data across all study sites to evaluate the effectiveness of habitat improvements on winter range. This approach will allow us to determine whether it is possible to effectively mitigate development impacts in highly developed areas, or whether it is better to allocate mitigation dollars toward less or non-impacted areas. In a recent project conducted on the Uncomphahgre Plateau, Bergman et al. (2009) found that habitat treatments implemented in pinion-juniper habitat in undeveloped areas increased overwinter survival of fawns by a magnitude of 1.15. We are also evaluating deer behavioral responses to varying levels of development activity. This will allow us to assess the effectiveness of certain BMPs for reducing disturbance to wintering mule deer. ..._ -.I Hay field improvements have been completed in the North Magnolia study area by WPX Energy to fulfill a Wildlife Management Plan (WMP) agreement with CPW; elk (Cervus e/aphus) response has been evident but mule deer response has been minor. A similar WMP agreement between ExxonMobil/XTO Energy and CPW allowed completion and continued monitoring of mechanical habitat improvements in the Magnolia study areas. Additional collaboration with WPX Energy has resulted in a clustered development plan in the Ryan Gulch study area and new technologies will be implemented to further reduce human activity through remote monitoring of well pads and fluid collection systems. Collaborative research with Idaho State University and Colorado State University/ExxonMobil Production has produced 3 peer-reviewed publications addressing mule deer migration (Lendrum et al. 2012, 2013) and improved approaches to address animal habitat use patterns (Northrup et al. 2013 ); these publications are summarized in Appendix A. Additional funding and cooperative agreements will be necessary to sustain this project to completion (preferably through 2019). We anticipate the opportunity to work cooperatively toward developing solutions for allowing the nation's energy reserves to be developed in a manner that benefits wildlife and the people who value both the wildlife and energy resources of Colorado. 37 �LITERATORE CITED Anderson, C.R., Jr. 2009. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and D. J. Freddy. 2008a. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Final Study Plan, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C. R., Jr., and D. J. Freddy. 2008b. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation-Stage I, Objective 5: Patterns of_mule deer distribution & movements. Pilot Study, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and C. J. Bishop. 2010. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and C. J. Bishop. 2011. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and C. J. Bishop. 2012. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Parks and Wildlife, Ft. Collins, CO, USA. Bartrnann, R. M. 197 5. Piceance deer study-population density and structure. Job Progress Report, Colorado Divison of Wildlife, Fort Collins, Colorado, USA. Bartmann, R. B., and S. F. Steinert. 1981. Distribution and movements of mule deer in the White River Drainage, Colorado. Special Report No. 51, Colorado Division of Wildlife, Fort Collins, Colorado, USA. Bartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monograph No. 121. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108-114. Bergman, E. J., C. J. Bishop, D. J. Freddy, and G. C. White. 2009. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, USA. Burnham, K. P., and D.R. Anderson. 2002. Model selection and multi-model inference: a practical information-theoretic approach. Second edition. Springer-Verlag, New York, New York, USA. Cook, R. C., J. G. Cook, D. L. Murray, P. Zager, B. K. Johnson, and M. W. Gratson. 2001. Development of predictive models of nutritional condition for rocky mountain elk. Journal of Wildlife Management 65 :973-987. Cook, R. C., T. R. Stephenson, W. L. Meyers, J. G. Cook, and L.A. Shipley. 2007. Validating predictive models of nutritional condition for mule deer. Journal of Wildlife Management 71:1934-1943. Cook, R. C., J. G. Cook, T. R. Stephenson, W. L. Meyers, S. M. McCorquodale, D. J. Vales, L. L. Irwin, P. Briggs Hall, R. D. Spencer, S. L. Murphie, K. A. Schoenecker, P. J. Miller. 2009. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife Management 74:880-896. Gibbs, H. D. 1978. Nutritional quality of mule deer foods, Piceance Basin, Colorado. Thesis, Colorado State University, Fort Collins, Colorado, USA. Kaplan, E. L., and P. Meier. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 52:457-481. Lendrum, P. E., C.R. Anderson, Jr., R. A. Long, J. K. Kie, and R. T. Bowyer. 2012. Habitat selection by mule deer during migration: effects of landscape structure and natural gas development. Ecosphere 3(9):82. http://dx.doi.org/l 0. 38 �..., Lendrum, P. E., C.R. Anderson, Jr., K. L. Monteith, J. A. Jenks, R. T. Bowyer. 2013. Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes. PLoS ONE 8( 5): e64548. doi: 10.1371/journal.pone.0064548 McClintock, B. T., G. C. White, K. P. Burnham, and M.A. Pride. 2008. A generalized mixed effects model of abundance for mark-resight data when sampling is without replacement. Pages 271289 in D. L. Thompson, E.G. Cooch, and M. J. Conroy, editors, Modeling demographic processes is marked populations. Springer, New York, New York, USA. Northrup, J.M., M. B. Hooten, C.R. Anderson, Jr., and G. Wittemyer. 2013. Practical guidance on characterizing availability in resource selection functions under a use-availability design. Ecology 94(7):1456-1463. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. C. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Stephenson, T. R., V. C. Bleich, B. M. Pierce, and G. P. Mulcahy. 2002. Validation of mule deer body composition using in vivo and post-mortem indices of nutritional condition. Wildlife Society Bulletin 30:557-564. Stephenson, T. R., K. J. Hundertmark, C. C. Swartz, and V. Van Ballenberghe. 1998. Predicting body fat and mass in moose with untrasonography. Canadian Journal of Zoology 76:717-722. Unsworth, J. W., D. F. Pack, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. Van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-421 in L. Nielsen, J.C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked individuals. Bird Study 46:120-139. White, G. C., and B. C. Lubow. 2002. Fitting population models to multiple sources of observed data. Journal of Wildlife Management 66:300-309 . Prepared by_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ Charles R. Anderson, Jr., Mammals Research Leader 'Cl "el .,,,., 39 �Table 1. Survival rate estimates (S) of fawn (2 Dec. 2011-15 June 2013) and adult female (1 July 201230 June 2013) mule deer from 4 winter range study areas of the Piceance Basin in northwest Colorado. Cohort Study area Initial sample size (n) March doe sample8 (n) S(95% en Fawns Ryan Gulch 78 0.752 (0.655-0.849) South Magnolia 54 0. 778 (0.667-0.889) North Magnolia 61 0.852 (0.763-0.941) North Ridge 76 0.529 (0.416-0.643) Adult females Ryan Gulch 32 57 0.799 (0.671-0.926) South Magnolia 39 51 0.801 (0.675-0.927) North Magnolia 37 60 0.863 (0.752-0.975) North Ridge 42 58 0. 726 (0.595-0.857) 8 Adult female sample sizes following capture and radio-collaring efforts March, 2012. 40 �Table 2. Mean rump fat (mm), Body Condition Score (BCS8), and% body fat (% fat) of adult female mule deer from 4 study areas in the Piceance Basin of northwest Colorado, March and December, 2009-2013. Values in parentheses= SD. March 2009 December 2009 March 2010 Study Area Rump fat Ryan Gulch 1.73 (1.78) 2.66 (0.55) 7.54 (1.80) 8.35 (6.36) 4.06 (1.13) 12.96 (4.53) 2.31 (1.44) 2.35 (0.48) 6.69 (1.58) South Magnolia 1.47 (0.68) 2.50 (0.60) 7.26 (1.82) 10.05 (6.19) 4.07 (1.21) 13.46 (4.96) 3.12 (2.20) 2.64 (0.59) 7.70 (2.01) North Magnolia 1.30 (0.79) 2.56 (0.68) 6.96 (2.23) 10.67 (5. 76) 4.25 (0.96) 13.92 (3.92) 3.15 (2.34) 2.85 (0.53) 8.28 (1.86) North Ridge 1.57 (1.22) 2.60 (0.56) 7.28 (1.66) 5.25 (5.65) 3.63 (I.I 1) 11.02 (4.54) 1.77 (1.11) 2.42 (0.49) 6.83 (1.50) March 2011 December 2011 BCS % fat Rump fat BCS % fat Rump fat BCS % fat Table 2. Continued. December 2010 Study Area Rump fat Ryan Gulch 7.75 (6.15) 3.34 (0.98) 10.82 (4.32) 1.55 (0.60) 2.53 (0.42) 7.05 (1.20) 13.41 (6.93) 4.21 (1.17) 13.17 (3.64) South Magnolia 9.85 (6.78) 3.30 (0.61) 11.21 (3.32) 1.65 (0.75) 2.35 (0.50) 6.56 (1.49) 7.53 (4.66) 3.37 (0.76) North Magnolia 9.55 (6.49) 2.56 (0.68) 11.65 (4.86) 1.65 (0.67) 2.53 (0.49) 7.06 (1.35) 9.43 (6.41) 3.79 (0.93) 11.15 (3.57) North Ridge 6.14 (5.29) 3.32 (0.82) 10.32 (3.39) 1.45 (0.76) 2.24 (0.49) 6.24 (1.45) 9.81 (5.81) 3.62 (1.00) 11.22 (3.38) BCS % fat Rump fat 41 BCS %fat Rump fat BCS % fat 9.95 (2.73) �Table 2. Continued. March 2012 December 2012 Study Area Rump fat Ryan Gulch 2.15 (1.44) 2.74 (0.44) 7.22 (1.16) 6.34 (4.35) 3.30 (0.77) 9.34 (2.43) 1.87 (0.90) 2.65 (0.37) 6.90 (1.59) South Magnolia 1.71 (0.76) 2.58 (0.36) 6.97 (1.12) 8.13 (5.71) 3.41 (1.04) 10.22 (3.23) 2.03 (0.78) 2.62 (0.26) 7.17 (0.68) North Magnolia 1.87 (0.78) 2.85 (0.33) 7.65 (0.94) 9.80 (6.35) 3.89 (1.17) 11.25 (3.60) 1.81 (0.91) 2.16 (0.41) 6.91 (1.08) North Ridge 2.24 (1.58) 2.70 (0.35) 7.26 (1.05) 5.76 (4.10) 3.32 (0.82) 1.87 (0.73) 2.48 (0.34) 6.70 (1.12) 8 BCS %fat Rump fat BCS March 2013 % fat 9.06 (2.31) Rump fat BCS %fat Body condition score taken from palpations of the rump following Cook et al. (2009). ( ( ( C C C ( ( f C C C C { C C C C C C C ( C ( { ( C C C C C C C { C C C C C ( C l C { C { �'Cl Table 3. Mark-resight abundance (N) and density estimates of mule deer from 4 winter range herd segments in the Piceance Basin, northwest Colorado, 1--6 April 2013. Data represent 4 helicopter resight surveys from North Ridge, North Magnolia, and South Magnolia and 5 resight surveys from Ryan Gulch. Study area Mean No. sighted Mean No. marked N(95% Cl) Density (deer/km2) 'cl Ryan Gulch 245 27 1,309 (1,131-1,530) 9.3 South Magnolia 182 24 743 (644-875) 8.9 North Magnolia 261 29 950 (824--1,111) 10.4 North Ridge 314 31 858 (753-1,006) 16.1 43 �Mule Deer Winter Range Study Areas Mule deer study areas Well Pads & Facilities D North Magnolia J; 1 South Magnolia .!. b North Ridge Ryan Gulch 0 2.5 □ 5 In developmenl/ application for drilling Injection well Producing well Development facilities 10 ~ ~ ~ ~ ~ ; ; ~ ~ ~ ~ - - - - - •Miles Figure I. Mule deer winter range study areas relative to active natural gas well pads and energy development facilities in the Piceance Basin of northwest Colorado, swnmer 20 I 3 (Accessed http://cogcc.state.eo.us/ Aug. 19, 2013). 44 �Ryan Gulch fawn S South Magnolia fawn S 2008/09-2012/13 2008/09-2012/13 0.80 0.70 t===::~~~~~;;~;:~=:~~~~~~~~~~ 0.60 +----~~~-.:...:3111....._.............. 0.50 t-----__:~~...,l":":"'llliiiiiiiiiiiiiijiiii'" 0.40 - + - - - - - - - - Z - - •.- ... --..- ... -- ... --..-.-. 0.30 - + - - - - - - - - - - - - - 0.20 - - - - - - - - - - - - 0.10 - - - - - - - - - - - - 0.00 -+-r--.-.-..-.-..-.-.--.-.---.-.---.-.-..-.-..-.-.--.-.--------....-.-. 1.00 . - ~ T ' I ' " : . " ' ! ~ 0.90 0.80 0.70 +----&r--=~,......,.........-~~/!Nw.,.,.,,-0.60 -I-----~~~!!!!!!!!!!!!!~.....~-=-=-.:!,!_ 0.50 +----------=:.........,.......,.....,,...--~o.4o - + - - - - - - - - - - - - - - - ' ' " s - = - , - ~ 0.30 - + - - - - - - - - - - - - - 0.20 - + - - - - - - - - - - - - - 0.10 - - - - - - - - - - - - 0.00 -+-,-......,.........-,,....,....,...,....,....,....,......,.........,........,--.,-,....,....,...,....,........,..-.-,-, North Magnolia fawn S North Ridge fawn S 2008/09-2012/13 2008/09-2012/13 1.00 • 0.90 0.80 0.70 0.60 +-----__::_a.--...:::~~--......~............... 0.50 -t------~r;-.:: ..-..-.-.-=---llliiiiiiiiiiiiiiiiiiiii"" 0.40 - - - - - - - - ~ - - . •.-..-...-...-. 0.30 - - - - - - - - - - - - 0.20 - + - - - - - - - - - - - - - 0.10 - - - - - - - - - - - - 0.00 -+-,-.,......,-......-,,....,....,-r-r--.--r-"'T""T"".,......,-......-,,....,....,--.--,--r-,-....-.---, 1.00 0.90 0.80 0.70 0.60 +---------...-::-.::.a,-...c.:~:;:;===0.50 4 - - - - - - - - - - - 1 + , , . , - r - e , - - _ _ _ ; ; = . 0.40 - - - - - - - - - - - - - - - - " ~ 0.30 - - - - - - - - - - - - 0.20 - + - - - - - - - - - - - - - 0.10 - - - - - - - - - - - - 0.00 -+-,--.---,-......-,,....,....,-r-r-,-,-"'T""T"".,......,-......-,,....,....,-..--.--r-,-....-.-, ,-::~~~~~~=== 1.00 -1-0.90 'Cl Figure 2. Over-winter (Dec-Mar & June) mule deer fawn survival (5) from 4 study areas in the Piceance Basin, northwest Colorado, 2008/09 (red lines), 2009/10 (orange lines), 2010/11 (blue lines), 2011/12 (black lines), and 2012/13 (purple lines). Solid lines= 5 and dashed lines= 95% Cl. Comparable data among years December-March 2008-2009 and 2009-2010 due to premature collar drop and Decembermid-June 2010-2011, 2011-2012, and 2012-2013. 45 �lXJ North Ridge and North Magnolia Summer Range ..... Gulch ands Summer Figure 3. Mule deer study a reas in the Piceance Basin of northwestern Colorado, USA (Top), spring 2009 migration routes of adul t female mule deer (n = 52; Lower left), and active natural-gas well pads (black dots) and roads (state, county, and natural-gas; white lines) from May 2009 (Lower right; from Lendrum et al . 20 12). 46 - �Early winter rump fat 15 13 Ell .. J! 9 a. E ::, a:: 7 -North Ridge .§. -=-North Magnolia --Ryan Gulch -south Magnolia 5 "Cr/ 3 Dec 2009 Dec 2010 Dec 2011 Dec 2012 Late winter rump fat 4 3.5 e 3 -====- North Ridge .§. ,! 2.5 -North Magnolia a. E i ,==..-Ryan Gulch 2 -south Magnolia 1.5 1 Mar 2009 Mar 2010 Mar 2011 Mar 2012 Mar 2013 Figure 4. Mean early (early Dec., Top) and late winter (early Mar, Bottom) body condition (mm rump fat) of adult female mule deer from 4 winter range study areas in the Piceance Basin of northwest Colorodo, March 2009-March 2013. Error bars= 95% CI. 47 �Male fawn weights 42.0 T 40.0 38.0 .E 36.0 QI) 'iii ~ rl] ,_ n .~ tttt - ~~ 1T ]~ 1rr 34.0 t- ,- -n - 32.0 t- ,- - - 1, fl i l - DRyan Gulch Ii r~ 1 ' DSouth Magnolia D North Magnolia D North Ridge - - 30.0 Dec 2008 Dec 2009 Dec 2010 Dec 2011 Dec 2012 Female fawn weights 42.0 . . . . . - - - - - - - - - - - - - - - - - - - 40.0 + - - - - - - - - - - - - - - - - - - - - 38.0 DRyan Gulch tiii =.E 36.0 DSouth Magnolia QI) 'iii ~ D North Magnolia 34.0 D North Ridge 32.0 30.0 Dec 2008 Dec 2009 Dec 2010 Dec 2011 Dec 2012 Figure 5. Mean male and female fawn weights and 95% CI (error bars) from 4 mule deer study areas in the Piceance Basin, northwest Colorado, December 2008- 2012. 48 �Piceance Basin late winter mule deer density 30.0 25.0 20.0 -- -- -- N j ~ 15.0 cu cu - - North Ridge •••••• Ryan Gulch Q - 10.0 • North Magnolia - s o u t h Magnolia 5.0 0.0 2009 2010 2011 2012 2013 Year Figure 6. Mule deer density estimates and 95% CI (error bars) from 4 winter range herd segments in the Piceance Basin, northwest Colorado, late winter 2009-2013. 'Cl 'Cl 49 �North Magnolia 1reatement siles (587 acres) LJ BearSet_ l 5_35b_andG BearSet_ I _BandA_E LJ BearSe1_36_54andJ GreasewoodSet_g16_g29 GreasewoodSet_g I _g 15 D Greasewooc1Set_g30_g42 LeeOversigh1s_a_tand I B_ 17 Mechanical treatm,nt comparison (54 acres) North Hatch Pilot Treatments (1 I 6 acres) - Mule Deer Study Areas North Magnolia South Magnolia 2 4 8 Figure 7. Habitat treatment site delineations in 2 mule deer study areas (604 acres each) of the Piceance Basin, northwest Colorado (Top; cyan polygons completed Jan. 2011, yellow polygons completed Jan. 2012, and remaining polygons completed April 2013). January 2011 hydro-axe treatment-site photos from North Hatch Gulch during April (Lower left, aerial view) and October, 2011 (Lower right, ground view). 50 ...... �.._, -.cl ._ 'cl '-''-..I Appendix A. Abstracts of published manuscripts resulting from Piceance Basin mule deer/energy development interaction research collaborations. Abstract format specific to the respective journal requirements. 'cl Habitat selection by mule deer during migration: effects of landscape structure and natural-gas development 'WI 'Cl ._ ._ 1 1 PATRICK E. LENDRUM1. CHARLES R. ANDERSON, JR. 2, RYAN A. LoNG ,JoHN G. KIE .AND R. TERRY BoWYER1 1 Department of Biological Sciences, Idaho State University, Pocatello, Idaho 83209 USA 2Colorado Division of Parks and Wildlife, Grand Junction, Colorado 81505 USA 'Cl ..., Citation: Lendrum, P. E., C.R. Anderson, Jr., R. A. Long, J. G. Kie, and R. T. Bowyer. 2012. Habitat selection by mule deer during migration: effects of landscape structure and natural-gas development. Ecosphere 3(9):82. http://dx.doi.org/10. 1890/ES12-00165.I -.I Abstract. The disruption of traditional migratory routes by anthropogenic disturbances has shifted patterns of resource selection by many species, and in some instances has caused populations to decline. Moreover, in recent decades populations of mule deer (Odocoi/eus hemionus) have declined throughout much of their historic range in the western United States. We used resourceselection functions to determine if the presence of natural-gas development altered patterns of resource selection by migrating mule deer. We compared spring migration routes of adult female mule deer fitted with GPS collars (n = 167) among four study areas that had varying degrees of natural-gas development from 2008 to 20 IO in the Piceance Basin of northwest Colorado, USA. Mule deer migrating through the most developed area had longer step lengths (straight-line distance between successive GPS locations) compared with deer in less developed areas. Additionally, deer migrating through the most developed study areas tended to select for habitat types that provided greater amounts of concealment cover, whereas deer from the least developed areas tended to select habitats that increased access to forage and cover. Deer selected habitats closer to well pads and avoided roads in all instances except along the most highly developed migratory routes, where road densities may have been too high for deer to avoid roads without deviating substantially from established migration routes. These results indicate that behavioral tendencies toward avoidance of anthropogenic disturbance can be overridden during migration by the strong fidelity ungulates demonstrate towards migration routes. If avoidance is feasible, then deer may select areas further from development, whereas in highly developed areas, deer may simply increase their rate of travel along established migration routes. 'di -..I ,c;I -..I 'Cl ... 'Cl .._, '-'1'-s,I' Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes 'Cl .._ ......, Patrick E. Lendrum 1, Charles R. Anderson Jr. 2, Kevin L. Monteith 1•3, Jonathan A. Jenks4, R. Terry Bowyer1 1 Department of Biological Sciences, Idaho State University, Pocatello, Idaho, USA, 2 Colorado Division of Parks and Wildlife, Grand Junction, Colorado, USA, 3 Wyoming Cooperative Fish and Wildlife Research Unit, University of Wyoming, Laramie, Wyoming, USA,4 Department of Natural Resource Management, South Dakota State University, Brookings, South Dakota, USA -- Citation: Lendrum, P. E., C.R. Anderson, Jr., K. L. Monteith, J. A. Jenks, R. T. Bowyer. 2013. Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes. PLoS ONE 8(5): e64548. doi:10.1371/journal.pone.0064548 .,,_ Abstract 'Cl "Cl le/ -..I 'el .... ._ --._ ... - ~ Background: Migration is an adaptive strategy that enables animals to enhance resource availability and reduce risk of predation at a broad geographic scale. Ungulate migrations generally occur along traditional routes, many of which have been disrupted by anthropogenic disturbances. Spring migration in ungulates is of particular importance for conservation planning, because it is closely coupled with timing of parturition. The degree to which oil and gas development affects migratory patterns, and whether ungulate migration is sufficiently plastic to compensate for such changes, warrants additional study to better understand this critical conservation issue. Methodology/Principal Findings: We studied timing and synchrony of departure from winter range and arrival to summer range of female mule deer (Odocoileus hemionus) in northwestern Colorado, USA, which has one of the largest natural-gas reserves currently under development in North America. We hypothesized that in addition to local weather, plant phenology, and individual life-history characteristics, patterns of spring migration would be modified by disturbances associated with natural-gas extraction. We captured 205 adult female mule deer, equipped them with GPS collars, and observed patterns of spring migration during 2008-20 IO. Conclusions/Significance: Timing of spring migration was related to winter weather (particularly snow depth) and access to emerging vegetation, which varied among years, but was highly synchronous across study areas within years. Additionally, timing of migration was influenced by the collective effects of anthropogenic disturbance, rate of travel, distance traveled, and body condition of adult females. Rates of travel were more rapid over shorter migration distances in areas of high natural-gas development resulting in the delayed departure, but early arrival for females migrating in areas with high development compared with less-developed areas. Such shifts in behavior could have consequences for timing of arrival on birthing areas, especially where mule deer migrate over longer distances or for greater durations. , ._J 'ell ""'1:11 51 �Practical guidance on characterizing availability in resource selection functions under a use-availability design JOSEPH M. NORTI-IRUP 1, MEVIN 8. HOOTEN 1.u, CHARLES R. ANDERSON, JR."1, AND GEORGE WIITEMYER 1 1 Department of Fish, Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 2 U.S. Geological Survey, Colorado Cooperative Fish and Wildlife Research Unit, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 3 Colorado State University, Department of Statistics, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 4 Mammals Research Section Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction, Colorado 81505 USA Citation: Northrup, J.M., M. 8. Hooten, C.R. Anderson, Jr., and G. Wittemyer. 2013. Practical guidance on characterizing availability in resource selection functions under a use-availability design. Ecology 94(7): 1456-1463. Abstract. Habitat selection is a fundamental aspect of animal ecology, the understanding of which is critical to management and conservation. Global positioning system data from animals allow fine-scale assessments of habitat selection and typically are analyzed in a use-availability framewor~ whereby animal locations are contrasted with random locations (the availability sample). Although most use-availability methods are in fact spatial point process models, they often are fit using logistic regression. This framework offers numerous methodological challenges, for which the literature provides little guidance. Specifically, the size and spatial extent of the availability sample influences coefficient estimates potentially causing interpretational bias. We examined the influence of availability on statistical inference through simulations and analysis of serially correlated mule deer GPS data. Bias in estimates arose from incorrectly assessing and sampling the spatial extent of availability. Spatial autocorrelation in covariates, which is common for landscape characteristics, exacerbated the error in availability sampling leading to increased bias. These results have strong implications for habitat selection analyses using GPS data, which are increasingly prevalent in the literature. We recommend that researchers assess the sensitivity of their results to their availability sample and, where bias is likely, take care with interpretations and use cross validation to assess robustness. 52 1w �\._.,I Colorado Parks and Wildlife July 1, 2013 - June 30, 2014 WILDLIFE RESEARCH REPORT State of._ _ _ _ _ _C __o__l=o=ra=d=o_ _ _ _ _ : =-P=ar=k=s..;;;;an=d=-..;.W.;..:i=ld=l=if=◄ e_ _ _ _ _ _ _ _ __ Cost Center 3430 : =M=amm==a=l=-s=R=e=se=ar=c=h;:....__ _ _ _ _ _ _ _ _ __ Work Package 3001 : =D=e=er:....;C=o=n=s=erv:..:..:a=tio=n=--------------Task No. 6 : Population Performance of Piceance Basin Mule Deer in Response to Natural Gas Resource Extraction and Mitigation Efforts to Address Human Activity and Habitat Degradation Federal Aid Project:_ _ W_-__ 18;a...5_-R __________ Period Covered: July 1, 2013 - June 30, 2014 Author: C.R. Anderson, Jr. Personnel: N. Bellerose, E. Bergman, C. Bishop, E. Cato, A. Collier, D. Collins, B. deVergie, S. Eno, D. Finley, M. Fisher, B. Frankland, L. Gepfert, T. Gettelman, M. Grode, T. Jenkins, D. Johnston, T. Knowles, M. Melham, J. Matijas, S. Nagy, B. Petch, J. Rivale, R. Schilowsky, K. Stonehouse, R. Velarde, L. Wolfe, CPW; E. Hollowed, L. Belmonte, BLM; D. Freddy, Hoch Berg Enterprises; T. Graham, Ranch Advisory Partners; M. Wille, T & M Contractors.; P. Lendrum, T. Bowyer, Idaho State University; P. Doherty, J. Northrup, M. Peterson, G. Wittemyer, K. Wilson, Colorado State University; R. Swisher, S. Swisher, Quicksilver Air, Inc.; D. Felix, Olathe Spray Service, Inc.; L. Coulter, Coulter Aviation. Project support received from Federal Aid in Wildlife Restoration, Colorado Mule Deer Association, Colorado Mule Deer Foundation, Colorado State Severance Tax Fund, EnCana Corp., Exxoru\fobil Production Co./XTO Energy, Marathon Oil Corp., Shell Petroleum, and WPX Energy. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the authors. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We propose to experimentally evaluate winter range habitat treatments and human-activity management altemati ves intended to enhance mule deer (Odocoileus hemionus) populations exposed to energy-development activities. The Piceance Basin of northwestern Colorado was selected as the project area due to ongoing natural gas development in one of the most extensive and important mule deer winter and transition range areas in Colorado. The data presented here represent the first 5 pretreatment years and l year post treatment of a long-term study addressing habitat improvements and evaluation of energy development practices intended to improve mule deer fitness in areas exposed to extensive energy development. We monitored 4 winter range study areas representing varying levels of development to serve as treatment (North Magnolia, South Magnolia) and control (North Ridge, Ryan Gulch) sites and recorded habitat use and movement patterns using GPS collars (2::5 location attempts/day), estimated overwinter fawn and annual adult female survival, estimated early and late winter body condition of adult females using ultrasonography, and estimated abundance using helicopter mark-resight surveys. During this research segment, we targeted 240 fawns (60/study area) and 170 does (30-70/study area) in early I 1liiil~Iffll1rnm BDOW028744 �December 2013 for VHF and GPS rad.iocollar attachment, respectively, and 120 does in March 2013 (30/study area) for late winter body condition assessment. Winter range habitat improvements completed spring 2013 resulted in 604 acres of mechanically treated pinion-juniper/mountain shrub habitats in each of the 2 treatment areas with minor and extensive energy development, respectively. Post-treatment monitoring will continue for 4 years to provide sufficient time to measure how vegetation and deer respond to these changes. Based on data collected during the 5-year pretreatment phase and 1 year posttreatment: (1) annual adult survival was consistent among areas averaging 80-84% annually, but overwinter fawn survival was more variable ranging from 48% to 95% within study areas, with annual and study area differences primarily due to annual weather conditions and in some cases density dependent influences; (2) migratory mule deer selected increased cover and increased their rate of travel through developed areas, but did not avoid development structures and avoided negative influences through behavioral shifts in timing and rate of migration; (3) mule deer body condition early and late winter was generally consistent within areas, with higher variability among study areas early winter, which likely relate to seasonal moisture within areas and relative forage capacity among areas; (4) mule deer densities have increased in 3 of 4 areas, with fluctuating and recently increasing deer densities evident in the 4 t11 area; (5) post treatment vegetation responses have been promising with evidence of improved forage conditions, but longer tenn monitoring will be required to address the full potential of habitat mitigation efforts. Detailed habitat use analyses are still pending for the pretreatment period. We will continue to collect population and habitat use data across all study sites to evaluate the effectiveness of habitat improvements on winter range. This approach will allow us to detennine whether it is possible to effectively mitigate development impacts in highly developed areas, or whether it is better to allocate mitigation efforts toward less or non-impacted areas. In collaboration with Colorado State University, we are also evaluating deer behavioral responses to varying levels of development activity in the Ryan Gulch study area and neonate survival in relation to energy development from all study areas. This will allow us to assess the effectiveness of certain Best Management Practices (BMPs) for reducing disturbance to deer and include neonatal data to other demographic parameters for evaluation of mule deer/energy development interactions. The study is slated to run through 2018 to allow sufficient time for measuring mule deer population responses to landscape level manipulations. 2 -.., '-6,/ �WILDLIFE RESEARCH REPORT POPULATION PERFORMANCE OF PICEANCE BASIN MULE DEER IN RESPONSE TO NATURAL GAS RESOURCE EXTRACTION AND MITIGATION EFFORTS TO ADDRESS HUMAN ACTIVITY AND HABITAT DEGRADATION CHARLES R. ANDERSON, JR PROJECT NARRITIVE OBJECTIVES 1. To determine experimentally whether enhancing mule deer habitat conditions on winter range elicits behavioral responses, improves body condition, increases fa'Ml survival, or ultimately, population density on mule deer winter ranges exposed to extensive energy development. 2. To determine experimentally to what extent modification of energy development practices enhance habitat selection, body condition, fawn survival, and winter range mule deer densities. SEGMENT OBJECTIVES 1. Collect and reattach GPS collars to maintain sample sizes for addressing mule deer habitat use and behavior patterns in 4 study areas experiencing varying levels of energy development of the Piceance Basin, northwest Colorado. 2. Estimate early and late winter body condition of adult female mule deer in each of the 4 winter herd segments using ultrasound techniques. 3. Monitor over-winter fawn and annual adult female mule deer survival by daily ground tracking and biweekly aerial tracking. 4. Conduct Mark-Resight helicopter surveys to estimate mule deer abundance in each study area. 5. Monitor habitat treatment response for assessing efficacy of habitat improvement projects to mitigate energy development disturbances to mule deer. 6. Continue neonate survival evaluations to complete demographic parameters for assessing mule deer/energy development interactions. INTRODUCTION Extraction of natural gas from areas throughout western Colorado has raised concerns among many public stakeholders and Colorado Parks and Wildlife (CPW) that the cumulative impacts associated with this intense industrialization will dramatically and negatively affect the wildlife resources of the region. Concern is especially high for mule deer due to their recreational and economic importance as a principal game species and their ecological importance as one of the primary herbivores of the Colorado Plateau Ecoregion. Extraction of natural gas will directly affect the potential suitability of the landscape used by mule deer through conversion of native habitat vegetation with drill pads, roads, or noxious weeds, by fragmenting habitat because of drill pads and roads, by increasing noise levels via compressor stations and vehicle traffic, and by increasing the year-round presence of human activities. Extraction will indirectly affect deer by increasing the human work-force population of the region resulting in the 3 �need for additional landscape for human housing, supporting businesses, and upgraded road/transportation infrastructure. Additionally, increased traffic on rural roads will raise the potential for vehicle-animal collisions and additive direct mortality to mule deer populations. Thus, research documenting these relationships and evaluating the most effective strategies for minimizing and mitigating these activities will greatly enhance future management efforts to sustain mule deer populations for future recreational and ecological values. The Piceance Basin in northwest Colorado contains one of the largest migratory mule deer populations in North America and also exhibits some of the largest natural gas reserves in North America. Projected energy development throughout northwest Colorado within the next 20 years is expected to reach about 15,000 wells, many of which will occur in the Piceance Basin, which currently supports over 250 active gas well pads (http://cogcc.state.co.us; Fig. I). Anderson and Freddy (2008a) in their longterm research proposal identified 6 primary study objectives to assess measures to offset impacts of energy extraction on mule deer population performance. During the past 5 years, we gathered baseline habitat utilization and demographic data from radiocollared deer across the Piceance Basin to allow assessment of habitat mitigation approaches that were completed April 2013. We are currently monitoring 2 control areas: I with development (0.6 pads & facilities/km1 ; Ryan Gulch) and l without (North Ridge). The control areas will be compared with 2 treatment areas experiencing similar development intensities (South Magnolia, 0.9 well pads & facilities/km1 and North Magnolia, 0.1 well pads & facilities/km\ that also recently received habitat improvements ( 604 acres each). Habitat and mule deer responses to mechanical habitat treatments will be evaluated over the next 3-5 years to assess the success of this habitat mitigation strategy to benefit mule deer exposed to energy development disturbance. In addition, mule deer behavior patterns in relation to energy development activities in the Ryan Gulch area are being monitored to identify effective Best Management Practices (BMPs) for future energy development planning. This progress report describes the previous 6.5 years (Jan 2008-June 2014) of mule deer population performance during the pretreatment phase on 4 winter range herd segments, which includes monitoring habitat selection and behavior patterns of adult female mule deer; spring/summer neonate, oveJWinter fawn and annual adult female survival; estimates of adult female body condition during early and late winter; and annual late-winter abundance/density estimates. STUDY AREAS The Piceance Basin, located between the cities of Rangely, Meeker, and Rifle in northwest Colorado, was selected as the project area due to its ecological importance as one of the largest migratory mule deer populations in North America and because it exhibits one of the highest natural gas reserves in North America (Fig. I). Historically, mule deer numbers on winter range were estimated between 20,000-30,000 (White and Lubow 2002), and the current number of well pads (Fig. I) and projected number of gas wells in the Piceance Basin over the next 20 years is about 250 and 15,000, respectively. Mule deer winter range in the Piceance Basin is predominantly characterized as a topographically diverse pinion pine (Pinus edu/is)-Utahjuniper (Juniperus osteosperma; pinion-juniper) shrubland complex ranging from 1,675 m to 2,285 m in elevation (Bartmann and Steinert 1981 ). Pinion-juniper are the dominant overstory species and major shrub species include Utah serviceberry (Amelanclzier utahensis), mountain mahogany (Cercocarpus montanus), bitterbrush (Purshia tridentata), big sagebrush (Artemisia tridentata), Gamble's oak (Quercus gambelii), mountain snowberry (Symphoricarpos oreophilus), and rabbitbrush (Chrysothamnus spp.; Bartmann et al. 1992). The Piceance Basin is segmented by numerous drainages characterized by stands of big sagebrush, saltbush (Atriplex spp.), and black greasewood (Sarcobatus vermiculatus), with the majority of the primary drainages having been converted to mixedgrass hay fields. Grasses and forbs common to the area consist ofwheatgrass (Agropyron spp.), blue grama (Bouteloua gracilis), needle and thread (Stipa comara), Indian rice grass (O,yzopsis hymenoides), arrowleafbalsamroot (Balsamorhiza sagittata), broom snakeweed (Gutierrezia sarotlzreae), pinnate tansymustard (Descurainia pinnata ), milkvetch (Astragalus spp. ), Lewis tlax (Linum lewisii), evening 4 V -\ U �primrose (Oenothera spp.), skyrocket gilia (Gilia aggregata), buckwheat (Erigonum spp.), Indian paintbrush (Castilleja spp.), and penstemon (Penstemon spp.; Gibbs 1978). The climate of the Piceance Basin is characterized by warm dry summers and cold winters with most of the annual moisture resulting from spring snow melt and brief summer monsoonal rain storms. 2 Wintering mule deer population segments we investigated include: North Ridge (53 km ) just north of the Dry Fork of Piceance Creek including the White River in the northeastern portion of the Basin, Ryan Gulch ( 14 l km2) between Ryan Gulch and Dry Gulch in the southwestern portion of the Basin, North Magnolia (79 km 2) between the Dry Fork of Piceance Creek and Lee Gulch in the northcentral portion of the Basin. and South Magnolia (83 krn 2) between Lee Gulch and Piceance Creek in the south-central portion of the Basin (Fig. 1). Each of these wintering population segments has received varying levels of nanrral gas development: no development in North Ridge, light development in North Magnolia (0.1 l pads & facilities/km 2), and relatively high development in the Ryan Gulch (0.46 pads & facilities/km2) and South Magnolia (0.67 pads & facilities/knl) segments (Fig. l). Among the 4 study areas, North Ridge has served as an unmanipulated control site, Ryan Gulch will serve to address humanactivity management alternatives (BMPs) that benefit mule deer exposed to energy development and as a developed control area for comparison to the developed treatment area receiving habitat improvements (South Magnolia), and No11h and South Magnolia will allow us to assess the utility of habitat treatments intended to enhance mule deer population performance in areas exposed to light (North Magnolia) and heavy (South Magnolia) energy development activities. METHODS Tasks addressed this period included mule deer capnrre and collaring efforts, monitoring neonate and overwinter fawn and annual adult female survival, estimating adult female body condition during early and late winter using ultrasonography, estimating mule deer abundance applying helicopter markresight surveys, and monitoring vegetation responses to habitat treatments completed spring 2013. We employed helicopter net-gunning techniques (Barrett et al. 1982, van Reenen 1982) to target 240 fawns and 170 adult females during early December 2013, and 120 adult females (primarily recaptures) during early March 2014. Once netted, all deer were hobbled and blind folded. Fawns were weighed, radiocollared and released on site, and adult females were transported to localized handling sites for recording body measurements and fitted with GPS collars (5 or 48 fix attempts/day; G2110D, Advanced Telemetry Systems, Isanti, MN, USA) and released. To provide direct measures of decline in overwinter body condition, we targeted 30 adult females in each study area that were captured the previous December; Vaginal Implant Transmitters (VITs) were also inserted to assist with neonate capture and collaring efforts spring 2014. Fawn collars were spliced and fitted with rubber surgical tubing to facilitate collar drop between mid-summer and autumn, and GPS collars were supplied with timed drop-off mechanisms scheduled to release early in April of the year following deployment. All radio-collars were equipped with mortality sensing options (i.e., increased pulse rate following 8 hrs of inactivity). Mule Deer Habitat Use and Movements We downloaded and summarized data from GPS collars deployed and recovered since 2008. GPS collars maintained the same schedule of attempting to collect locations every 5 hours, except for 40 does in Ryan Gulch and l 0 control deer from North Ridge where location rates were programmed for every 30 minutes to increase resolution of movement data for evaluation of deer behavior patterns in relation to differing development activities. Joe Northrup (CSU PhD Candidate) is currently analyzing resource selection data relative to energy development activity and should have results finalized for inclusion in next year's annual report. Mule deer resource selection analyses to address success of habitat improvements are pending until vegetation responses are fully realized, which are anticipated by fall 2018. 5 �Mule Deer Survival V Mule deer mortality monitoring consisted of daily ground-telemetry tracking and aerial monitoring approximately every 2 weeks from fixed-wing aircraft on winter range and weekly aerial monitoring on summer range. Once a mortality signal was detected. deer were located and necropsied to assess cause of death. We estimated weekly survival using the staggered entry Kaplan-Meier procedure (Kaplan and Meier 1958, Pollock et al. 1989). Capture-related mortalities (any doe/fawn mortalities occurring within 10 days of capture; excluding neonates) and collar failures were censored from survival rate estimates. We estimated survival rates from I July 2013 through 30 June 2014 for adult females, from birth to mid December for neonates, and from early December 2013-mid June 2014 for fawns. Adult Female Body Measurements We applied ultrasonography techniques described by Stephenson et al. (1998, 2002) and Cook et al. (2001) to measure maximum subcutaneous rump fat (mm), loin depth (longissimus dorsi muscle, mm), and to estimate% body fat. We estimated a body condition score (BCS) for each deer by palpating the rump (Cook et al. 2001, 2007, 2009). We examined differences (P < 0.05) in nutritional status among study areas and between years using a two-sample t-test. We considered differences in body condition meaningful when mean rump fat or % body fat differed statistically between comparisons. Other body measurements recorded included pregnancy status (pregnant, barren) via blood samples, fetal counts using ultrasonagraphy, weight (kg), chest girth (cm), and hind-foot length (cm). Abundance Estimates We conducted 3 helicopter mark-resight surveys (2 observers and the pilot) during late March to estimate deer abundance in each of the 4 study areas; l to 2 additional surveys/study area were scheduled to achieve increased precision, but were not possible due to helicopter mechanical issues. We delineated each study area from GPS locations collected on winter range during the first 3 years of the study (Jan 2008 through April 2011 ). Two aerial fixed-wing telemetry surveys/study area were conducted during helicopter mark-resight surveys to determine which marked deer were within each survey area, and we confirmed adult female locations during surveys from GPS data acquired April 2014. We delineated flight paths in ArcGIS 9.3 prior to surveys following topographic contours (e.g., drainages, ridges) and approximating 500-600 m spacing throughout each study area; flight paths during surveys were followed using GPS navigation in the helicopter. Two approximately 12 x 12 cm pieces of Ritchey livestock banding material (Ritchey Livestock ID. Brighton, CO USA) were uniquely marked using color, number, and symbol combinations and attached to each radio-collar to enhance mark-resight estimates. Each deer observed during surveys was recorded as mark ID#, unmarked, or unidentified mark. We used program MARK (White and Burnham 1999), applying the immigration-emigration mixed logit-normal model (McClintock et al. 2008), to estimate mule deer abundance and confidence intervals. For mark-resight model evaluations, we examined parameter combinations of varying detection rates with survey occasion and whether individual sighting probabilities (i.e., individual heterogeneity) were constant or varied (a:!= 0 or ;t:. 0). Model selection procedures followed the information-theoretic approach of Burnham and Anderson (2002). RESULTS AND DISCUSSION Deer Captures and Survival The helicopter crew captured 242 fawns and 167 does during Dec 2013 and 118 does during March 2014. Four fawn monalities (I. 7%; ultimate cause= 2 capture myopathy, I predation, 1 vehicle 6 -.,, V �collision) occurred within the 10 day censorship period. Doe mortalities totaled 4 (2.4%; capture myopathy) and 2 (1.7%; capture myopathy) within 10 days of the December and March capture periods, respectively. Mortality rates, 10 days post capture, have varied between 2-3% for fawns and 0-3% for does since Jan 2008, except during the 2011-2012 capture season where myopathy rates were higher (36%) due to dry. warm conditions (Anderson and Bishop 2012). Fawn survival from early December 2013 through mid June 2014 was similar (P> 0.05) among study areas ranging from 0.84 to 0.95 (Table 1). General comparisons to previous years suggest relatively high fawn survival occurred during winters 2009-2010 and 2013-2014: and relatively low survival during winter 2010-2011 (Fig. 2), which correlates to summer forage condition evident from December fawn weights (Fig. 3) and winter severity. Annual adult female survival varied from 0.76 (North Magnolia) to 0.90 (South Magnolia; Table 1) during 2013-2014, but was comparable among study areas during 2013-2014 and to previous years (P > 0.05), with the exception of lower survival in North Magnolia during 2011-2012 (S= 0.68, Anderson and Bishop 2012). Sample sizes for adult female survival do not allow statistical discrimination among years unless large differences are evident (e.g., >15-20%). Spring Migration Patterns Collaboration with Idaho State University to address mule deer migration patterns in developed and undeveloped landscapes (funded from energy company contributions) has recently been completed. Three manuscripts have been accepted for publication (Lendrum et al. 2012, Lendrum et al. 2013; Lendrum et al. 2014; Appendix A). In addressing habitat selection during spring migration, Lendrum et al. (2012; Fig. 4) noted that mule deer migrating through the most developed landscapes exhibited longer step lengths (straight line distance between GPS locations) and selected habitats providing greater security cover than deer in undeveloped landscapes that migrated through more open areas that provided increased foraging opportunities. Migrating deer also selected areas closer to well pads, but avoided roads, except in the highest developed areas where road densities were likely too high for avoidance without significant deviations from traditional migration routes. In the second manuscript (Lendrum et al. 2013), we addressed biological and environmental factors influencing spring migration and assessed how energy development influenced migratory behavior. Overall, spring migration was influenced by snow depth, temperature, and green-up on winter and summer range; increasing temperatures. snow melt and emerging vegetation dictated timing of winter range departure and summer range arrival. Duration of Piceance Basin mule deer migration was short, with median migration durations of 3-8 days among the 4 areas (straight line distance between seasonal ranges averaged 32--40 km). Deer in poor condition migrated later than deer in good condition, but condition was similar among areas regardless of development status. Migrating deer from developed study areas did not avoid development structures, but departed later, arrived earlier and migrated more quickly than deer from undeveloped areas. While large changes in timing of migration could have nutritional consequences and negatively influence reproduction and neonate survival, the relatively minor shift we observed should not result in long-term fitness consequences. Migratory deer in the Piceance Basin appear to avoid negative effects of energy development through behavioral shifts in timing and rate of migration. In the third publication (Lendrum et al. 2014 ), we monitored migratory mule deer in the Piceance Basin to examined the relationship between the Normalized Difference Vegetation Index (NDVI). which is a course-scale measure of forage quality using a GIS assessment of vegetation greenness, and fecal nitrogen to assess the assumption that forage quality and deer diets can be reasonably linked to address 7 �deer habitat use patterns from remotely sensed data. We found that diet quality evident from fecal nitrogen and course measures of vegetation green-up were informative, and that Piceance Basin mule deer exhibited rapid migration (3 to 8 days depending on study area), left winter range following snow melt with lowest fecal N and NOVI values, and progressed to summer range as vegetation green-up and nitrogen levels increased, but ahead of peak vegetation green-up on summer range. I suspect this rapid migration strategy is evident for deer in relatively good condition and allows for early arrival on summer range to take advantage optimal forage conditions prior to parturition. Mule Deer Body Condition Early-winter body condition measurements of adult female mule deer were comparable among study areas during December 2013 (P > 0.05, Fig. 5, Table 2). Early winter condition this year was comparable to previous years with exception of relatively low condition expressed by North Ridge deer during 2009 and by North Ridge and Ryan Gulch deer last year; Ryan Gulch exhibited generally improved condition during December 2011 (Fig. 5). With the exception of Ryan Gulch does exhibiting relatively poor condition during 2014, late winter body condition was higher this year when compared to 2009 and 2011, but lower than North and South Magnolia does during 2010 (Fig. 5, Table 2). These observations appear more related to seasonal moisture conditions and winter severity than development intensity thus far. December 2013 fawn weights by study area were higher in all comparisons with the exception of Ryan Gulch females (Fig. 3). In general, seasonal moisture conditions appear to be driving differences in annual body condition within study areas, but other factors appear related to differences among study areas. We suspect density dependent factors (forage capacity relative to deer density) are related to observed differences in early winter body condition among study areas. More detailed analyses will be conducted to identify factors attributing to these observations. Neonate Survival To complete demographic parameters addressing mule deer-energy development interactions, CPW, Colorado State University, and ExxonMobil Production entered into a collaborative agreement to investigate neonate survival in developed and undeveloped landscapes (funded by ExxonMobil Production Co.) beginning spring 2012. Mark Peterson (Graduate Research Assistant) and Paul Doherty (CSU professor) are assisting with this research, which will continue through 2014. Neonate capture and collaring efforts totaled 85 during spring 2012, 67 during spring 2013, and 54 during spring 2014. Estimated neonate survival through mid-December 2012 was 0.39 (95% CI= 0.28-0.50) and 0.37 (95% CI= 0.25--0.48) from birth to mid-December 2013; neonate survival estimates for 2014 are pending. Factors influencing neonate mule deer survival from developed and undeveloped landscapes will be addressed by mid 2015. Mule Deer Population Estimates Mark-resight models that best predicted abundance estimates (lowest AICc; Burnham and Anderson 2002) exhibited variable sightability across surveys (P,) for all study areas and homogenous individual sightability (cr2 = 0) for South Magnolia deer and variable individual sightability (cr2 -:j:. 0) for the other 3 areas. North Ridge exhibited the highest deer density (22.2/km\ with comparably lower deer densities in the other 3 areas (9.7-10.9/km2; Table 3, Fig. 6). Populations have increased over the 6 year monitoring period in 3 of the study areas (from 6.5/km2 to l 0.4/km 2), with a fluctuating population in North Ridge (from 14.4/km2 to 22.8/km2); the current North Ridge density is comparable to the peak population estimate from 2011 (Fig. 6). The previous North Ridge decline was likely related to density dependent factors, evident in lower early winter body condition (Table 2) and an increase in malnutrition mortalities of radio-collared adult females during the decline (from O in 2010-2011 to 7 in 2012-13). 8 U �Abundance estimates from 2013 were similarly precise from all 4 study areas with the mean Confidence Interval Coefficient of Variation (CICV) ranging from 0.15--0.17. Magnolia Habitat Treatments We completed 116 acres of pilot habitat treatments in January 2011 (Anderson and Bishop 201 I; Environmental Assessment: DOI-BLM-CO-110-2011-004-EA}, 54 acres of mechanical treatment method comparison treatments (hydro-ax, roller-chop, chain) in January 2012 (Stephens 2014), and 1,038 acres of hydro-ax treatments in April 2013 (Determination of NEPA Adequacy: DOI-BLM-CO-I 10-2012-0134DNA), totaling 604 treated acres in each study area (Fig. 7). Vegetation response in the pilot treatment sites was visually evident by fall 20 l l (Fig. 7), and resulted in statistically significant (P < 0.05) increases in native grass cover and apparent (>50% increase) but not yet significant increases in native forb cover by spring 2013 (2014 results are still pending). Stephens (2014) reported that all 3 mechanical treatment methods compared resulted in roughly a 3 fold increase in grasses, forbs, and shrubs combined after 2 growing seasons (versus control sites), but cautioned that rollerchop treatments may be more vulnerable to invasive species response. Vegetative responses from 2013 hydro-ax treatments were visually evident following l growing season, but statistical comparisons are pre-mature. As anticipated, grass and forb responses should be evident 2 to 3 year post-treatment, with longer term response expected (3-5 years) from palatable shrubs. Of note, relatively high moisture conditions experienced during spring 2014 resulted in higher than normal prevalence of cheatgrass (Bromus tectorum); cheatgrass invasion has previously been minor to non-existent. Cheatgrass invasion, however, does not appear directly related to treatment sites because occurrence is evident in both treatment and control areas. We anticipate this outbreak will subside based on past competitive advantage of native species to dominate, but will continue to monitor species composition and address cheatgrass persistence in treatment and control sites. GPS data addressing deer use of treatment sites is just becoming available and will be analyzed as additional data are collected and vegetation responses progress. Vegetation and mule deer responses will be documented for the next 4 years to assess the utility of this mitigation approach in benefiting mule deer exposed to energy development disturbance. All expenses addressing these habitat treatments will be covered through a Wildlife Management Plan agreement between CPW and ExxonMobil Production/XTO energy. SUMMARY AND COLLABORATIONS The long-term goal of this study is to investigate habitat treatments and energy development practices that enhance mule deer populations exposed to extensive energy development activity. The information presented here summarizes mule deer population parameters from the 5-year pre-treatment period and I year post-treatment. The pretreatment period was completed during spring 2013, providing baseline data for comparison with intended improvements in habitat conditions and response to varying degrees in human development activity. Winter range habitat improvements resulting in 604 acres of mechanically treated pinion-juniper/mountain shrub habitats in each of2 study areas were completed April 2013, and preliminary vegetation responses appear promising. Post-treatment monitoring will continue for 4 years to provide sufficient time to measure how deer respond to these changes. Based on data collected prior to habitat improvements (i.e., pretreatment phase): (l) annual adult survival was consistent among areas averaging 80-84% annually, but overwinter fawn survival was variable, ranging from 48% to 85% within study areas, with annual and study area differences primarily due to annual weather conditions and in some cases density dependent influences; (2) migratory mule deer selected for areas with increased cover and increased their rate of travel through developed areas, and avoided negative influences through behavioral shifts in timing and rate of migration, but did not avoid 9 �development structures; (3) mule deer body condition early and late winter was consistent within areas, with higher variability among study areas early winter, which was likely related to seasonal moisture within areas and relative forage capacity among areas; (4) mule deer densities appear to be increasing in 3 of 4 areas, with midterm decline and recent increase in North Ridge. Detailed habitat use analyses are pending for the pretreatment period. We will continue to collect the various population and habitat use data across all study sites to evaluate the effectiveness of habitat improvements on winter range. This approach will allow us to detennine whether it is possible to effectively mitigate development impacts in highly developed areas, or whether it is better to allocate mitigation dollars toward less or non-impacted areas. In a recent project conducted on the Uncomphahgre Plateau, Colorado, Bergman et al. (2014) found that habitat treatments implemented in pinion-juniper habitat in undeveloped areas increased oveiwinter survival of fawns by a magnitude of I . 15. We are also evaluating deer behavioral responses to varying levels of development activity. This will allow us to assess the effectiveness of certain BMPs for reducing disturbance to wintering mule deer. Hay field improvements have been completed in the North Magnolia study area by WPX Energy to fulfill a Wildlife Management Plan (WMP) agreement with CPW; elk (Cervus elaphus) response has been evident but mule deer response has thus far been minor. A similar WMP agreement between ExxonMobil/XTO Energy and CPW allowed completion and continued monitoring of mechanical habitat improvements in the Magnolia study areas. Additional collaboration with WPX Energy has resulted in a clustered development plan in the Ryan Gulch study area and new technologies will be implemented to further reduce human activity through remote monitoring of well pads and fluid collection systems. Collaborative research with Idaho State University, Colorado State University/ExxonMobil Production, and Utah State University/Utah Division of Wildlife Resources has produced 6 peer-reviewed publications addressing mule deer migration (Lendrum et al. 2012, 2013, 2014), improved approaches to address animal habitat use patterns (Northrup et al. 2013 ), mule deer response to helicopter capture and handling (Northrup et al. 2014 ), and potential effects of male-biased harvest on mule deer productivity (Freeman et al. 2014); these publications are summarized in Appendix A. Additional funding and cooperative agreements will be necessary to sustain this project to completion (preferably through 2018). We anticipate the opportunity to work cooperatively toward developing solutions for allowing the nation's energy reserves to be developed in a manner that benefits wildlife and the people who value both the wildlife and energy resources of Colorado. LITERATURE CITED Anderson, C.R., Jr. 2009. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C. R., Jr.. and D. J. Freddy. 2008a. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Final Study Plan, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C. R., Jr., and D. J. Freddy. 2008b. Population perfonnance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation-Stage I, Objective 5: Patterns of_mule deer distribution & movements. Pilot Study, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and C. J. Bishop. 2010. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C. R., Jr., and C. J. Bishop. 2011. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA. V �Anderson, C.R., Jr., and C. J. Bishop. 2012. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Parks and Wildlife, Ft. Collins, CO, USA. Bartmann, R. M. 1975. Piceance deer study-population density and structure. Job Progress Report, Colorado Divison of Wildlife. Fort Collins, Colorado. USA. Bartmann, R. B., and S. F. Steinert. 1981. Distribution and movements of mule deer in the White River Drainage, Colorado. Special Report No. 51, Colorado Division of Wildlife, Fort Collins, Colorado, USA. Bartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monograph No. 12 l. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin I 0: 108-114. Bergman, E. J., C. J. Bishop, D. J. Freddy, G. C. White. and PF. Doherty, Jr. 2014. Habitat management influences over-winter survival of mule deer fawns in Colorado. Journal of Wildlife Management 78(3):448-455; DOI: 10.1002/jwmg.683 Burnham, K. P., and D.R. Anderson. 2002. Model selection and multi-model inference: a practical information-theoretic approach. Second edition. Springer-Verlag, New Yor~ New York, USA. Cook, R. C., J. G. Cook, D. L. Murray, P. Zager, B. K. Johnson, and M. W. Gratson. 2001. Development of predictive models of nutritional condition for rocky mountain elk. Journal of Wildlife Management 65 :973-987. Cook, R. C., T. R. Stephenson, W. L. Meyers, J. G. Cook, and L.A. Shipley. 2007. Validating predictive models of nutritional condition for mule deer. Journal of Wildlife Management 71: 1934-1943. Cook, R. C., J. G. Cook, T. R. Stephenson, W. L. Meyers, S. M. McCorquodale, D. J. Vales, L. L. Irwin, P. Briggs Hall, R. D. Spencer, S. L. Murphie, K. A. Schoenecker, P. J. Miller. 2009. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife Management 74:880-896. Freeman, E. D., R. T. Larsen, M. E. Peterson, C.R. Anderson, Jr., K. R. Hersey, and B. R. McMillan. 2014. Effects of male-biased harvest on mule deer: implications for rates of pregnancy, synchrony, and timing of parturition. Wildlife Society Bulletin; DOI: l 0.1002/wsb.450 Gibbs, H. D. 1978. Nutritional quality of mule deer foods, Piceance Basin, Colorado. Thesis, Colorado State University, Fort Collins. Colorado, USA. Kaplan, E. L., and P. Meier. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 52:457-481. Lendrum, P. E., C.R. Anderson, Jr., R. A. Long, J. K. Kie, and R. T. Bowyer. 2012. Habitat selection by mule deer during migration: effects of landscape structure and natural gas development. Ecosphere 3(9):82. http://dx.doi.org/l 0. Lendrum, P. E., C.R. Anderson, Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2013. Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes. PLoS ONE 8(5): e64548. doi: 10.13 7 l/joumal.pone.0064548 Lendrum, P. E., C.R. Anderson, Jr .. K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2014. Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns of forage quality. Mammalian Biology: http://dx.doi.org/10. l016/j.mambio.20l4.05.005 McClintock, B. T., G. C. White, K. P. Burnham, and M.A. Pride. 2008. A generalized mixed effects model of abundance for mark-resight data when sampling is without replacement. Pages 271289 in D. L. Thompson, E.G. Cooch, and M. J. Conroy, editors, Modeling demographic processes is marked populations. Springer, New York, New York, USA. Northrup, J.M.~ M. B. Hooten, C.R. Anderson, Jr., and G. Wittemyer. 2013. Practical guidance on characterizing availability in resource selection functions under a use-availability design. Ecology 94(7):1456-1463. 11 �Northrup, J.M., C.R. Anderson, Jr., and G. Wittemyer. 2014. Effects of helicopter capture and handling on movement behavior of mule deer. Journal of Wildlife Management 78(4):731-738; DOI: 10.1002/jwmg. 705 Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. C. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Stephens, G. J. 2014. Understory responses to mechanical removal ofpinyon-juniper overstory. MS Thesis, Colorado State University, Ft. Collins USA. Stephenson, T. R., V. C. Bleich, B. M. Pierce, and G. P. Mulcahy. 2002. Validation of mule deer body composition using in vivo and post-mortem indices of nutritional condition. Wildlife Society Bulletin 30:557-564. Stephenson, T. R., K. J. Hundertmark, C. C. Swartz, and V. Van Ballenberghe. 1998. Predicting body fat and mass in moose with untrasonography. Canadian Journal of Zoology 76:717-722. Unsworth, J. W., D. F. Pack, G. C. White, and R. M. Barnnann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. Van Reenen, G. 1982. Field experience in the capture ofred deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-42 l in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society. Milwaukee, USA. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked individuals. Bird Study 46: 120-139. White, G. C., and 8. C. Lubow. 2002. Fitting population models to multiple sources of observed data. Journal of Wildlife Management 66:300-309. u Prepared by_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ Charles R. Anderson, Jr., Mammals Research Leader --, 12 �Table 1. Survival rate estimates (S) of fawn (2 Dec. 2013-15 June 2014) and adult female (1 July 201330 June 2014) mule deer from 4 winter range study areas of the Piceance Basin in northwest Colorado. Cohort Study area Initial sample size (n) March doe samplea (n) S (95% Cl) Fawns Ryan Gulch 60 0.950 (0.895-1.000) South Magnolia 61 0.868 (0. 783--0.953) North Magnolia 60 0.883 (0.802--0.965) North Ridge 56 0.838 (0.741--0.935) Adult females Ryan Gulch 36 62 0.846 (0. 745--0.94 7) South Magnolia 31 55 0.901 (0.806--0.997) North Magnolia 33 55 0. 764 (0.637--0.890) North Ridge 29 51 0. 792 (0.666--0.918) aAdult female sample sizes following capture and radio-collaring efforts March, 2014. 13 �) ) () Table 2. Mean rump fat (mm). Body Condition Score (BCS'\ and% body fat(% fat) of adult female mule deer from 4 study areas in the Piceance Basin of northwest Colorado, March and December, 2009-2014. Values in parentheses= SD. March 2009 March 2010 December 2009 Study Arca Rump fat Ryan Gulch 1.73 ( 1.78) 2.66 (0.55) 7.08 ( 1.27) 8.35(6.36) 4.06(1.13) I0.54 (3.72) 2.31 ( 1.44) 2.35 (0.48) 6.37 ( 1.41) South Magnolia 1.29 (0.47) 2.51 (0.66) 6.74 (2.27) I 0.05 (6.19) 4.07 ( 1.21) 11.44 (3.50) 3.12 (2.20) 2.64 (0.59) 7.11 (l.69) North Magnolia 1.31 ( 1.01) 2.66 (0.68) 7.15 (1.63) 10.67 (5.76) 4.25 (0.96) 11.94 (3.39) 3.15 (2.34) 2.85 (0.53) 7.54 (1.53) North Ridge 1.57 ( 1.22) 2.60 (0.56) 6.81 ( 1.68) 5.25 (5.65) 3.63 (I.II) 9.37 (3.08) 1.77 ( 1.11) 2.42 (0.49) 6.39 ( I .45) March 2011 December 20 I I BCS % fat Rump fat BCS % fat Rump fat BCS % fat Table 2. Continued. December 20 l 0 % fat BCS % fat Ryan Gulch 7.26 (6.36) 3.24 (0.96) 9.69 (3.56) 1.55 (0.60) 2.53 (0.42) 6.72 (1.37) 13.41 (6.39) 4.21 ( 1.17) 13.17 (3.64) South Magnolia 9.85 (6.78) 3.30 (0.61) 11.27 (3.75) 1.65 (0.75) 2.35 (0.50) 6.15 (1.75) 8.19 (5.54) 3.41 (0.82) 10.34 (3.28) North Magnolia 9.55 (6.49) 3.46 ( 1.16) I0. 79 (4.26) 1.65 (0.67) 2.53 (0.49) 6.79 (1.47) 8.73 (5.77) 3.74 (0.91) I 0.73 (3.14) North Ridge 7.25 (5.41) 3.47 (0.86) 9.85 (3.02) 1.45 (0.76) 2.24 (0.49) 5.78 ( 1.79) 8.86 (5.37) 3.51 (0.99) 10.77 (3.33) 14 % fat BCS Rump fat Rump fat BCS Rump fat Study Area �Table 2. Continued. March 2012 December 2012 % fat BCS Rump fat BCS March 2013 BCS % fat Rump fat 9.34 (2.43) 1.87 (0.90) 2.65 (0.37) 6.90 ( 1.59) 7.00 (1.13) 8.30 (5.71) 3.46 (1.07) 10.32 (3.23) 2.06 (0.77) 2.65 (0.26) 7.19 (0.66) 1.90 (0.76) 2.84 (0.34) 7.62 (0.95) 9.66 (6.41) 3.84 (1.16) 11.18(3.64) 1.76 (0.91) 2.59 (0.41) 6.87 (1.11) 2.24 ( 1.58) 2.70 (0.35) 7.26 ( 1.05) 5.76 (4.10) 3.32 (0.82) 1.87 (0.73) 2.48 (0.34) 6.70(1.12) Study Arca Rump fat Ryan Gulch 2.15 ( 1.44) 2.74 (0.44) 7.22 ( 1.16) 6.34 (4.35) 3.30 (0.77) South Magnolia 1.68 (0.77) 2.59 (0.36) North Magnolia North Ridge 9.06 (2.31) % fat Table 2. Continued. December 2013 BCS March 2014 % fat Rump fat BCS % fat Study Arca Rump fat Ryan Gulch 9.27 (6.29) 3.47 (0.87) 10.61 (3.76) 1.69 (0.85) 2.68 (0.39) 7.03 (0.99) South Magnolia 11.27 (8.40) 3.99 (1.04) 11.40(4.16) 2.57 ( 1.61) 2.96 (0.30) 7.75 (0.68) North Magnolia 9.00 (6.15) 3.44 (0.78) 10.48 (3.25) 2.33 (2.12) North Ridge 11.17 (5.28) 3.85 (0.72) 1 1.66 (2.69) 2.38 ( 1.52) 2.68 (0.39) 2.80 (0.49) 7.31 (1.43) 7.16(1.14) aBody condition score taken from palpations of the rump following Cook ct al. (2009). 15 ) ) ) �Table 3. Mark-resight abundance (N) and density estimates of mule deer from 4 winter range herd segments in the Piceance Basin, northwest Colorado, 24-28 March 2014. Data represent 3 helicopter resight surveys from each study area. Study area Mean No. sighted Mean No. marked N(95% CI) Density (deer/km~) Ryan Gulch 328 _.,, - 1,498 (l,270-1.791) 10.6 South Magnolia 208 30 77 l (664-909) 9.7 North Magnolia 261 32 862(748-1,011) 10.9 North Ridge 413 33 1.183 (1.029-1,388) 22.~ u --....., 16 �Mule Deer Winter Range Study Areas Mul e deer study areas Zll!'-.:MP'3t:I D Nor1h Magnolia Soutn Magnolia l l0'1h Ridge ; yan Gulcn 25 ! In de\'elopme nt l Proaucing ,,,en Developmem rac •1nies 1 5 10 r-.111es Figure 1. Mule deer winter range study areas relative to active natural gas well pads and energy development facilities in the Piceance Basin of northwest Colorado, winter 20 13/ 14 (Accessed http://cogcc.state.eo.us/ Dec. 3 1.2013). 17 I Well Pads & Facilities �Ryan Gulch fawn S South Magnolia fawn S 2008/09-2013/14 2008/09-2013/14 1.00 0.90 0.80 t===:~:!~;i.;~~~~~~~~~f!~~~~ 0.70 -t0.60 +----......1.,~nllili~..;..:.;:-............... 0.50 0.40 0.30 •••••••••••••••• 0.20 - - - - - - - - - - - - - 0.10 0.00 North Magnolia fawn S North Ridge fawn S 2008/09-2012/13 2008/09-2013/14 1.00 T"ilir.-~ 0.90 ~•.....-1111,,ii;:: 0.80 0.70 0.60 4 -_ _ __.;;.......,_-311i._;:..u...-~............i~ 0.50 -t-1 -------S,,e;-:-:--==--a.iiiiiiiiiiiiiiiiiiiiiii" 0.40 - - - - - - - - - - - - - - - 0.30 - i . - - - - - - - - - - - - - 0.20 - + - - - - - - - - - - - - - - 0.10 - - - - - - - - - - - - - - - - 0.00 i I l 1 i l I i i I I I , I I i l i I I I I I I I . I t l 1.00 0.90 0.80 0.70 0.60 0.50 +--------___,,+ro....---,__;;;=0.40 0.30 0.20 . . . . . . - - - - - - - - - - - - - - 0.10 + - - - - - - - - - - - - - 0.00 I I I . I i I I I I I I i I I I I i I I I I I I I I i JI u Figure 2. Over-winter (Dec-Mar & June) mule deer fawn survival (5) from 4 study areas in the Piceance Basin, northwest Colorado, 2008/09 (red lines), 2009/10 (orange lines). 2010/11 (blue lines), 2011/12 (black lines), 2012/13 (purple lines), and 2013/14 (cyan lines). Solid lines= .Sand dashed lines= 95% Cl. Comparable data among years: December-March 2008-2010 due to premature collar drop and December-mid-June 2010-2014. ---. 18 �Male fawn weights 42.0 40.0 I 38.0 fa 36.0 ·a; tt I 0 s 34.0 rt ~ j Ii [ ~ I t f-- J ,_a ,- , 1 1, T ,_ - ,_ y 1:: 30.0 . C South Magnolia DNorth Magnolia I- C North Ridge 1: 32.0 Cl Ryan Gulch - - 1: - - Il l 11 - ' ~ =- Dec2008 Dec2009 Dec2010 Dec2011 Dec2012 Dec2013 Female fawn weights 42.0 40.0 bO 38.0 D Ryan Gulch -" -fa 36.0 s 34.0 T ~ i~f j .T It f n "ai l T j J 32.0 ' ,_: 30.0 !J ~ j IJ ,- I• 11 ,, ~ ,_ I t ~ f ttJ l a ~ DSouth Magnolia D North Magnolia DN orth Ri dge - Dec2008 Dec2009 Dec2010 Dec2011 Dec2012 Dec 2013 Figure 3. Mean male and female fawn weights and 95% CI (error bars) from 4 mule deer study areas in the Piceancc Basin. northwest Colorado, December :2008-20 13. 19 �r I- /*-- _ _ 1 / i \\. l.itah ;· . ·yomrng lf--I l Colorado r I .,__ r-- Figure 4. \fole deer study areas in the Piceance Basin of northwestern Colorado, USA (Top). spring 2009 migrati on routes of adu lt female mule deer (11 = 52: Lower left). and active natural-gas well pads (black dots) and roads (state. county, and natural-gas: white lines) from May 2009 (Lower 1ight; from Lendrum et al. 20 12). 20 �Early winter rump fat (mm) 16 14 12 1 i c==aNorth Ridge 10 -==-North Magnolia 8 a. ---m... E 6 = = Ryan Gulch ==-South Magnolia 4 2 0 Dec 2009 Dec 2010 Dec 2011 Dec 2012 Dec 2013 Late winter rump fat (mm) 4.00 3.50 3.00 e .§. 2.50 ==- North Ridge = ~ 2.00 North Magnolia =~Ryan Gulch C. E 1.50 == c==:aSouth Magnolia 1.00 0.50 0.00 Mar 2009 Mar 2010 Mar 2011 Mar 2012 Mar 2013 Mar 2014 Figure 5. Mean early (early Dec., Top) and late winter (early Mar., Bottom) body condition (mm rump fat) of adult female mule deer from 4 winter range study areas in the Piceance Basin of northwest Colorado, March 2009-March 2014. Error bars = 95% Cl. 21 �Piceance Basin late winter mule deer density 30.00 25.00 e 20.00 ~ North Ridge QI Ryan Gulch ":::15.00 QI Q North Magnolia 10.00 - s o u t h Magnolia 5.00 0.00 2009 2010 2011 2012 2013 2014 Year Figure 6. Mule deer density estimates and 95% CI (error bars) from 4 winter range herd segments in the Piceance Basin, northwest Colorado, late winter 2009-2014. u 22 V �\/orlh M ag~olla treatemem sites (58; ac,es) I = 3ear:':et_ • 5_35b_andC: 3ear5et_ • _aanctA _E 3ear:3:t_36_54andJ GreasewoodSet_ g I 6_g29 GreasewooctSet_QI.Jl 15 C Greasewooose,_g30_g4~ Lee0vers,gnts_a_land 16_ 17 Medlanrc31:re:=trret"\t companson (54 acres) Nor111 Hatch Pilot Treatments ( • • 6 acres; Mule Deer Study Areas Nor111 f,1agnoha Figure 7. Habitat treatment site delineations in 2 mule deer study areas (604 acres each) of the Piceance Bas in, northwest Colorado (Top; cyan polygons completed Jan. 2011 using hydro-axe; ye llow polygons completed Jan.2012 us ing hydro-axe. roller-chop. and chaining; and remaining polygons completed Ap1il 20 I 3 usi ng hydro-axe). January 2011 hydro-axe treatment-s ite phocos from North Hatch Gulch du1ing Apri l (Lower left, aeria l view) and October. 2011 (Lower right. grnund view). 23 �Appendix A. Abstracts of published manuscripts resulting from Piceance Basin mule deer/energy development interaction research collaborations. Abstract format specific to the respective journal requirements. Habitat selection by mule deer during migration: effects of landscape structure and natural-gas development 1 PATRICKE. LENDRUM 1, CHARLES R. ANDERSON. JR.~. RYAN A. LoNo • Jo11"1'\ G. KrE',AND R. TERRY BOWYER' 1 Department ofBiological Sciences, Idaho State University. Pocatello. Idaho 83209 USA Colorado Division of Parks and Wildlife, Grand Junction, Colorado 81505 USA 2 Citation: Lendrum, P. E.. C.R. Anderson. Jr.. R. A. Long, J. G. Kie. and R. T. Bowyer. 2012. Habitat selection by mule deer during migration: effects of landsc:ipe strucrure and natural-gas development. Ecosphere 3(9):82 http:.idx.doi.org/10.1890/ES 12-00165.1 Abstract. The disruption of traditional migratory routes by anthropogenic disturbances has shifted patterns of resource selection by many species. and in some instances has caused populations to decline. Moreover, in recent decades populations of mule deer (Odocoileus hemionus) have declined throughout much of their historic range in the western United States. We used resourceselection functions to determine if the presence of natural-gas development altered patterns ofresource selection by migrating mule deer. We compared spring migration routes of adult female mule deer fitted with GPS collars (11 = 167) among four study areas that had varying degrees of natural-gas development from 2008 to 20 IO in the Piceance Basin of northwest Colorado, USA. Mule deer migrating through the most developed area had longer step lengths (straight-line distance between successive GPS locations) compared with deer in less developed areas. Additionally. deer migrating through the most developed study areas tended to select for habitat types that provided greater amounts of concealment cover, whereas deer from the least developed areas tended to select habitats that increased access to forage and cover. Deer selected habitats closer to well pads and avoided roads in all instances except along the most highly developed migratory routes, where road densities may have been too high for deer to avoid roads without deviating substantially from established migration routes. These results indicate that behavioral tendencies toward avoidance of anthropogenic disturbance can be overridden during migration by the strong fidelity ungulates demonstrate towards migration routes. If avoidance is feasible, then deer may select areas further from development, whereas in highly developed areas. deer may simply increase their rate of travel along established migration routes. Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes Patrick E. Lendrum'. Charles R. Anderson Jr.!. Kevin L Monteith'·J. Jonathan A. Jenks\ R. Terry Bowyer' Department of Biological Sciences, Idaho State University, Pocatello, Idaho, USA,! Colorado Division of Parks and Wildlife, Grand Jwiction, Colorado, USA. 3 Wyoming Cooperative Fish and Wildlife Research Unit. University of Wyoming, Laramie. Wyoming, USA.4 Department of Natural Resource Management, South Dakota State University, Brookings, South Dakota, USA 1 Citation: Lendrum, P. E.. C.R. Anderson, Jr., K. L. Monteith, J. A. Jenks. R. T. Bowyer. :?013. Migrating Mule Deer: Effects of anthropogenically Altered Landscapes. PLoS ONE 8(5): e64548. DOI: I0.1371/joumal.pone.0064548 Abstract Background: Migration is an adaptive strategy that enables animals to enhance resource availability and reduce risk of predation at a broad geographic scale. Ungulate migrations generally occur along traditional routes, many of which have been disrupted by anthropogenic disturbances. Spring migration in ungulates is of particular importance for conservation planning, because it is closely coupled with timing of parturition. The degree to which oil and gas development affects migratory patterns, and whether ungulate migration is sutliciently plastic to compensate for such changes. warrants additional study to better understand this critical conservation issue. Methodology/Principal Findings: We studied timing and synchrony of departure from winter range and arrival to summer range of female mule deer (Odocoi/eus hemio11us) in northwestern Colorado, USA, which has one of the largest natural-gas reserves currently under development in North America. We hypothesized that in addition to local weather, plant phenology, and individual life-history characteristics, patterns of spring migration would be modified by disturbances associated with natural-gas extraction. We captured 205 adult female mule deer, equipped them with GPS collars, and observed patterns of spring migration during 2008-2010. Co11c/uslo11s/Signijica11ce: Timing of spring migration was related to winter weather (particularly snow depth) :111d access to emerging vegetation, which varied among years, but was highly synchronous across study areas within years. Additionally, timing of migration was intluenced by the collective effects of anthropogenic disturbance, rate of travel, distance traveled, and body condition of adult temales. Rates of travel were more rapid over shorter migration distances in areas of high natural-gas development resulting in the delayed departure. but early arrival for females migrating in areas with high development compared with less-developed areas. Such shifts in behavior could have consequences for timing of aiTival on birthing areas, especially where mule deer migrate over longer distances or for greater durations. 24 �~ Practical guidance on characterizing availability in resource selection functions under a use-availability design JOSEPH M.1'ORTIIRUP MEVIN B. HOOTEN ·! CHARLES R. ANDERSON. JR.". A.'ID GEORGE WITTEMYER' 'Department offish, Wildlite. and Conservation Biology. Colorado State University. 1474 Campus Delivery, Fort Collins, Colorado 80.523 USA =u.s. Geological Survey. Colorado Cooperative Fish and Wildlite Research Unit. 1474 Campus Delivery, Fon Collins, Colorado 80523 USA ;Colorado State University. Depanment of Statistics, Colorado State University, 1474 Campus Delivery. Fort Collins, Colorado 80523 USA J.'.\-tammals Research Section Colorado Parks and Wildlife. 71 l Independent Avenue. Grand Junction. Colorado 81505 USA 1 1 • 3 , Citation: 1':onhrup. J.M., M. B. Hooten. C.R. Anderson. Jr.. and G. Wittemyer. .:?013. Practical guidance on characterizing availability in resource selection functions under a use-availability design. Ecology 94(7): 1456-1463. http://dx.doi.org/ I0.1890/ I.:?- I688. I Abstract. Habitat selection is a fundamental aspect of animal ecology. the understanding of which is critical to management and conservation. Global positioning system data from animals allow fine-scale assessments of habitat selection and typically are analyzed in a use-availability framework. whereby animal locations are contrasted with random locations (the availability sample). Although most use-availability methods are in fact spatial point process models~ they often are fit using logistic regression. This framework otlers numerous methodological challenges, for which the literature provides little guidance. Specifically, the size and spatial extent of the availability sample influences coefficient estimates potentially causing interpretational bias. We examined the influence of availability on statistical inference through simulations and analysis of serially correlated mule deer GPS data. Bias in estimates arose from incorrectly assessing and sampling the spatial extent of availability. Spatial autocorrelation in covariates. which is common for landscape characteristics. exacerbated the error in availability sampling leading to increased bias. These results have strong implications for habitat selection analyses using GPS data, which are increasingly prevalent in the literature. We recommend that researchers assess the sensitivity of their results to their availability sample and, where bias is likely. take care with interpretations and use cross validation to assess robustness. Effects of Helicopter Capture and Handling on Movement Behavior of Mule Deer JOSEPH M. 1'ORTHRUP 1• CHARLES R. A:"l"DERSON. JR;_ A.'\ID GEORGE W1TTEMYER 1 1 Depanment offish, Wildlife, and Conservation Biology. Colorado State Uni\'ersity. 1474 Campus Delivery, Fort Collins. Colorado 805.23 USA ~Mammals Research Section Colorado Parks and Wildlife. 71 l lndependent Avenue. Grand Junction, Colorado 81505 USA Citation: Nonhrup, J. M.. C. R. Anderson. Jr.. and G. Wittemyer. 2014. Etlects of helicopter capture and handling on movement behavior of mule deer. Journal of Wildlife :Management 78(4):731-738: DOI: 10.1002/jwmg.705 ABSTRACT Research on wildlife movement, physiology. and reproductive biology often requires capture and handling of animals. Such invasive treatment can alter behavior, which may bias results or invalidate assumptions regarding representative behaviors. To assess the impacts of handling on mule deer (Odocoileus hemionus), a focal species for research in North America, we investigated pre- and post-recapture movements of collared individuals. and compared them to deer that were not recaptured (controls). We compared pre- and post-recapture movement rates (m/hr) and 24-hour straight-line displacement among recaptured and control deer. In addition. we examined the time it took recaptured deer to return to their pre-recapture home range. Both daily straight-line displacement and movement rate were marginally elevated relative to monthly averages for 24 hours following recapture, with non-significant elevation continuing for up to 7 days. Comparing movements averaged over 30 days before and after recapture, we found no differences in displacement. but movement rates demonstrated seasonal effects, with faster movements post- relative to pre-recapture in March and slower movements post- relative to pre-recapture in December. Relative to control deer movements, recaptured deer movement rates in March were higher immediately after recapture and lower in the second and third weeks following recapture. The median time to return to the pre-recapture home range was 13 hours. with 71 % of deer returning in the first day, and 91 % returning within 4 days. These results indicate a short period of elevated movements following recaptures, likely due co the deer returning to their home ranges, followed by weaker but non-significant depression of movements for up to 3 weeks. Censoring of the first day of data post capture from analyses is strongly supported. and removing additional days until the individual returns to its home range will control for the majority of impacts from capture. ,~: 2014 The Wildlite Society. 25 �Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns offorage quality V Patrick E. Lendrum\ Charles R. Anderson Jr.\ Kevin L. Monteith\ Jonathan A. knks'1. R. Terry Bowyer• • Depanment of Biological Sciences, Idaho State University, 921 South 8th Avenue, Stop 8007, Pocatello 83209, USA h Mammals Research Section Colorado Parks and Wildlite, 711 Independent Avenue. GrJnd Junction 81505, USA "Wyoming Cooperative Fish and Wildlife Research Unit. Depanment of Zoology and Physiology, University of Wyoming. 3166, IO00 East Vniversity Avenue. Laramie 82071, USA J Department of Natural Resource Management, South Dakota State t.:niversity. Box ~1408, Brookings 57007. USA Citation: Lendrum, P. E.. C.R. Anderson, Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2014. Relating the movement ofa rapidly migrating ungulate to spatiotemporal patterns of forage quality. Mammalian Biolo!:,ry: hnp: .. Jx.doi.,m.!il o. IO I6:j.mambio..:?O I4.05.005 ABSTRACT: Migratory ungulates exhibit recurring movements, otien along traditional routes between seasonal ranges each spring and autumn, which allow them to track resources as they become available on the landscape. We examined the relationship between spring migration of mule deer (Odocoileus hemionus) and forage quality, as indexed by spatiotemporal patterns of fecal nitrogen and remotely sensed greenness of vegetation (Normalized Difference Vegetation Index; NDVI) in spring 2010 in the Piceance Basin of northwestern Colorado, t;SA. NDVI increased throughout spring, and was affected primarily by snow depth when snow was present, and temperature when snow was absent. Fecal nitrogen was lowest when deer were on winter range before migration, increased rapidly to an asymptote during migration, and remained relatively high when deer reached summer range. Values of fecal nitrogen corresponded with increasing NDVI during migration. Spring migration for mule deer provided a way for these large mammals to increase access to a high-quality diet. which was evident in patterns of NOVI and fecal nitrogen. Moreover, these deer 'jumped" rather than ··surted" the green wave by arriving on summer range well before peak productivity of forage occurred. This rapid migration may aid in securing resources and seclusion from others on summer range in preparation for parturition, and to minimize detrimental factors such as predation. and malnutrition during migration. Effects of Male-Biased Harvest on Mule Deer: Implications for Rates of Pregnancy, Synchrony, and Timing of Parturition ERIC D. FREEMAN1, RAi'\IDY T. LARSEN1, :MARKE. PETERSON\ CHARLES R. A..'\IDERSON, JR.3, KENT R. HERSEY 4, AND BROCK R. McMILLAN 1 1 Department of Plant and Wildlife Sciences, Brigham Young t.:niversity, 27 5 WIDB, Provo, UT 84602, USA :: Department of Fish, Wildlite. and Conservation Biology. Colorado Scace University, 1474 Campus Delivery, Fort Collins, CO 80523. USA 3 Colorado Parks and Wildlife, 711 Independent Avenue. Grand Junction. CO 8I505, USA 4 Utah Division of Wildlife Resources, 1594 W North Temple, Salt Lake City, UT 8-1114. USA Citation: Freeman, E. D., R. T. Larsen, M. E. Peterson, C.R. Anderson. Jr.. K. R. Hersey. and B. R. McMillan. 2014. Effects of male-biased harvest on mule deer: implications for rates of pregnancy, synchrony, and timing ofpanurition. Wildlite Society Bulletin; DOI: 10.1002/wsb.450 ABSTRACT Evaluating how management practices intluence the population dynamics of ungulates may enhance future management of these species. For example, in mule deer ( Odocoileus hcmio11us), changes in male/female ratio due to malebiased harvest may alter rates of pregnancy, timing ofparrurition. and synchrony ofpamnition if inadequate numbers of males are present to fertilize females during their first estrous cycle. If rates of pregnancy or parturition are influenced by decreased male/female ratios, recruitment may be reduced (e.g., fewer births, later parturition resulting in lower survival of fa,vns, and a less synchronous panurition that potentially increases susceptibility of neonates to predation). Our objectives were to compare rates of pregnancy, synchrony of parturition, and timing of parturition between exploited mule deer populations with a relatively high (Piceance, CO, USA; 26 males/100 females) and a relatively low (Monroe, UT, USA; 14 males,100 females) male/female ratio. We determined rates of pregnancy via ultrasonography and timing of parturition via vaginal implant transmitters. We found no differences in rates of pregnancy (98.6% and 96.6%; z = 0.821; P = 0.794), timing of parturition (estimate= l.258; SE= 1.672; r= 0.752; P = 0.454), or synchrony of parturition (F = l .0i3; P = 0.859) between Monroe Mountain and Piceance Basin, respectively. The relatively low male/temale ratio on Monroe Mountain was not associated with a protracted period of parturition. This finding suggests that relatively low male/female ratios typical of heavily harvested populations do not influence population dynamics because recruitment remains unaffected. e.1014 The Wildlite Society. --- 26 �" \..,I Colorado Parks and Wildlife July 1, 2014-June 30. 2015 WILDLIFE RESEARCH REPORT State of________ C___ o__ lo.....ra___d.. .,_o_ _ _ _ _ : .;;...Pa=r=ks=-=an____d----a,.W__i__ ld__li.....fe____________ Cost Center 3430 : ta__m ____m . . . . a. . ___ls___R ___e_____s.....e__a ___ rc__h______________ Work Package 3001 : =D__e___e__ r ___ C___o_____n___ se____r_v____ at__io__n_______________ Task No. 6 : Population Performance of Piceance Basin Mule Deer in Response to Natural Gas Resource Extraction and Miti2ation Efforts to Address Human Activitv and Habitat Degradation Federal Aid Project:_ ___,;_W;._-.;;.. 18;;;..;;5;._-=R;.. .__ _ __ .;;.;rv__ Period Covered: July 1. :2014 - June 30. 2015 Author: C. R. Anderson, Jr. Personnel: N. Bellerose, E. Bergman, T. Blecha. D. Collins, J. Decoste, 8. deVergie. D. Finley. M. Fisher, L. Gepfert, D. Johnston, T. Knowles. D. Lewis, J. Matijas. N. Mesce, B. Petch, J. Rivale, G. Sanchez, R. Schilowsky, G. Smith, K. Stonehouse. R. Velarde. L. Wolfe. CPW; E. Hollowed, L. Belmonte, BLM; D. Freddy, Hoch Berg Enterprises; T. Graham, Ranch Advisory Partners; P. Doherty, J. Northrup, M. Peterson, G. Wittemyer, K. Wilson. Colorado State University; R. Swisher, S. Swisher, Quicksilver Air, Inc.; D. Felix, Olathe Spray Service, Inc.: L. Coulter, Coulter Aviation. Project support received from Federal Aid in Wildlife Restoration, Colorado Mule Deer Association, Colorado Mule Deer Foundation. Colorado State Severance Tax Fund, EnCana Corp., ExxonMobil Production Co./XTO Energy, Marathon Oil Corp., Shell Petroleum. and \\'PX Energy. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the authors. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT ~ We propose to experimentally evaluate winter range habitat treatments and human-activity management alternatives intended to enhance mule deer (Odocoileus hemionus) populations exposed to energy-development activities. The Piceance Basin of northwestern Colorado was selected as the project area due to ongoing natural gas development in one of the most extensive and important mule deer winter and transition range areas in Colorado. The data presented here represent the first 5 pretreatment years and 3 years post treatment of a long-term study addressing habitat improvements and evaluation of energy development practices intended to improve mule deer fitness in areas exposed to extensive energy development. We monitored 4 winter range study areas representing varying levels of development to serve as treatment (North Magnolia. South Magnolia) and control (North Ridge, Ryan Gulch) sites and recorded habitat use and movement patterns using GPS collars (?:5 location attempts/day), estimated neonatal, overwinter fawn and annual adult female survival. estimated early and late winter body condition of adult females using ultrasonography. and estimated abundance using helicopter mark-resight surveys. During this research segment, we targeted 240 fawns (60/study area) and 120 does (30/study area) in early December 2014 for VHF and GPS radiocollar attachment, respectively, and adult female ~COLO DIV WILDLIFE RESEARCH CTR LIB I~ Illffll ll~I IIIIII II ll~l lllll lllll ll~I IIII 11111111 BDOW029160 �body condition assessment. We attempted recapture of 120 does in March 2015 (30/study area) for late winter body condition assessment. Winter range habitat improvements completed spring 2013 resulted in 604 acres of mechanically treated pinion-juniper/mountain shrub habitats in each of the 2 treatment areas with minor and extensive energy development respectively. Post-treatment monitoring will continue for another 3 years to provide sufficient time to measure how vegetation and deer respond to these changes. Based on data collected during the 5-year pretreatment phase and 3 years post-treatment: (1) annual adult survival was consistent among areas averaging 79-87% annually, but overwinter fawn survival was variable. ranging from 48% to 95% within study areas~ with annual and study area differences primarily related to annual weather conditions; (2) migratory mule deer selected for areas with increased cover and increased their rate of travel through developed areas. and avoided negative influences through behavioral shifts in timing and rate of migration, but did not avoid development structures; (3) mule deer body condition early and late winter was consistent within areas~ with higher variability among study areas early winter, which was likely related to seasonal moisture within areas and relative forage capacity among areas; (4) mule deer exhibited behavioral plasticity in relation to energy development, where disturbance distance varied relative to diurnal extent and intensity of development activity, which may provide for several options in future development planning; (5) mule deer densities appear to be increasing in 3 of 4 areas, with a stable population in North Ridge; and (6) post treatment vegetation responses have been promising with evidence of improved forage conditions, but longer term monitoring will be required to address the full potential of habitat mitigation efforts. Detailed habitat use analyses are still pending for the pretreatment period. We will continue to collect population and habitat use data across all study sites to evaluate the effectiveness of habitat improvements on winter range. This approach will allow us to determine whether it is possible to effectively mitigate development impacts in highly developed areas, or whether it is better to allocate mitigation efforts toward less or non-impacted areas. In collaboration with Colorado State University. we are also monitoring neonate survival in relation to energy development from all study areas. This will allow us to include neonatal data to other demographic parameters for evaluation of mule deer/energy development interactions. The study is slated to run through 2018 to allow sufficient time for measuring mule deer population responses to landscape level manipulations. 2 V V �WILDLIFE RESEARCH REPORT POPULATION PERFORMANCE OF PICEANCE BASIN MULE DEER IN RESPONSE TO NATURAL GAS RESOURCE EXTRACTION AND MITIGATION EFFORTS TO ADDRESS HUMAN ACTIVITY AND HABITAT DEGRADATION CHARLES R. ANDERSON, JR PROJECT NARRITIVE OBJECTIVES 1. To determine experimentally whether enhancing mule deer habitat conditions on winter range elicits behavioral responses. improves body condition. increases fawn survival. or ultimately. population density on mule deer winter ranges exposed to extensive energy development. 2. To determine experimentally to what extent modification of energy development practices enhance habitat selection, body condition. fawn survival. and winter range mule deer densities. SEGMENT OBJECTIVES 1. Collect and reattach GPS collars to maintain sample sizes for addressing mule deer habitat use and behavior patterns in 4 study areas experiencing varying levels of energy development of the Piceance Basin, northwest Colorado. 2. Estimate early and late winter body condition of adult female mule deer in each of the 4 winter herd segments using ultrasound techniques. 3. Monitor over-winter fawn and annual adult female mule deer survival by daily ground tracking and biweekly aerial tracking. 4. Conduct Mark-Resight helicopter surveys to estimate mule deer abundance in each study area. 5. Monitor habitat treatment response for assessing efficacy of habitat improvement projects to mitigate energy development disturbances to mule deer. 6. Continue neonate survival evaluations to complete demographic parameters for assessing mule deer/energy development interactions. INTRODUCTION ~\._./ Extraction of natural gas from areas throughout western Colorado has raised concerns among many public stakeholders and Colorado Parks and Wildlife (CPW) that the cumulative impacts associated with this intense industrialization will dramatically and negatively affect the wildlife resources of the region. Concern is especially high for mule deer due to their recreational and economic importance as a principal game species and their ecological importance as one of the primary herbivores of the Colorado Plateau Ecoregion. Extraction of natural gas will directly affect the potential suitability of the landscape used by mule deer through conversion of native habitat vegetation with drill pads, roads. or noxious weeds, by fragmenting habitat because of drill pads and roads. by increasing noise levels via compressor stations and vehicle traffic. and by increasing the year-round presence of human activities. Extraction will indirectly affect deer by increasing the human work-force population of the region resulting in the .,.., �need for additional landscape for human housing. supporting businesses, and upgraded road/transportation infrastructure. Additional1y, increased traffic on rural roads will raise the potential for vehicle-animal collisions and additive direct mortality to mule deer populations. Thus, research documenting these relationships and evaluating the most effective strategies for minimizing an~ mitigating these activities will greatly enhance future management efforts to sustain mule deer populations for future recreational and ecological values. The Piceance Basin in northwest Colorado contains one of the largest migratory mule deer populations in North America and also exhibits some of the largest natural gas reserves in North America. Projected energy development throughout northwest Colorado within the next 20 years is expected to reach about 15,000 wells, many of which will occur in the Piceance Basin, which currently supports over 250 active gas well pads (http://cogcc.state.co.us; Fig. I). Anderson and Freddy (2008a) in their longterm research proposal identified 6 primary study objectives to assess measures to offset impacts of energy extraction on mule deer population performance. During the past 7 years. we gathered baseline habitat utilization and demographic data from radiocollared deer across the Piceance Basin to allow assessment of habitat mitigation approaches that were completed April 2013. We are currently monitoring 2 control areas: 1 with development (0.6 pads & facilities/km 2; Ryan Gulch) and I without (North Ridge). The control areas will be compared with 2 treatment areas experiencing similar development intensities (South Magnolia, 0.9 well pads & facilities/km 2 and North Magnolia, 0.1 well pads & facilities/km\ that also recently received habitat improvements (604 acres each). Habitat and mule deer responses to mechanical habitat treatments will be evaluated until spring 2018 to assess the success of this habitat mitigation strategy to benefit mule deer exposed to energy development disturbance. In addition, mule deer behavior patterns in relation to energy development activities in the Ryan Gulch area are being monitored to identify effective Best Management Practices (BrvtPs) for future energy development planning. This progress report describes the previous 7.5 years (Jan 2008-June 2015) of mule deer population performance during the pretreatment phase on 4 winter range herd segments, which includes monitoring habitat selection and behavior patterns of adult female mule deer; spring/summer neonate, overwinter fawn and annual adult female survival; estimates of adult female body condition during early and late winter; and annual late-winter abundance/density estimates. STUDY AREAS The Piceance Basin, located between the cities of Rangely. Meeker, and Rifle in northwest Colorado, was selected as the project area due to its ecological importance as one of the largest migratory mule deer populations in North America and because it exhibits one of the highest natural gas reserves in North America (Fig. I). Historically, mule deer numbers on winter range were estimated between 20,000-30,000 (White and Lubow 2002), and the current number of well pads (Fig. I) and projected number of gas wells in the Piceance Basin over the next 20 years is about 250 and 15,000, respectively. Mule deer winter range in the Piceance Basin is predominantly characterized as a topographically diverse pinion pine (Pinus edulis)-Utahjuniper (Juniperus osteosperma; pinion-juniper) shrubland complex ranging from I ~6i5 m to 2,285 m in elevation (Bartmann and Steinert 1981 ). Pinion-juniper are the dominant overstory species and major shrub species include Utah serviceberry (.A.melanchier utahensis). mountain mahogany (Cercocarpus montanus). bitterbrush (Purshia tridentata), big sagebrush (Artemisia tridentata), Gamble's oak (Quercus gambelii). mountain snowberry (Symphoricarpos oreophilus), and rabbitbrush (Chrysothamnus spp.; Bartmann et al. 1992). The Piceance Basin is segmented by numerous drainages characterized by stands of big sagebrush. salt bush (Atriplex spp.), and black greasewood (Sarcobatus vermiculatus), with the majority of the primary drainages having been converted to mixedgrass hay fields. Grasses and forbs common to the area consist of wheatgrass (Agropyron spp.), blue grama (Bouteloua gracilis). needle and thread (Stipa comata), Indian rice grass (Oryzopsis hymenoides). arrowleafbalsamroot (Balsamorhiza sagittata), broom snakeweed (Gutierrezia sarothreae), pinnate tansymustard (Descurainia pinnata). milkvetch (Astragalus spp.), Lewis flax (Limon lewisii), evening 4 V �~ primrose (Oenothera spp.), skyrocket gilia (Gilia aggregata). buckwheat (Erigonum spp.), Indian paintbrush (Castilleja spp.). and penstemon (Penstemon spp.; Gibbs 1978). The climate of the Piceance Basin is characterized by wann dry summers and cold winters with most of the annual moisture resulting from spring snow melt and brief summer monsoonal rain storms. Wintering mule deer population segments we investigated include: North Ridge (53 km 1) just north of the Dry Fork of Piceance Creek including the White River in the northeastern portion of the Basin, Ryan Gulch (141 kni) between Ryan Gulch and Dry Gulch in the southwestern portion of the Basin. North Magnolia (79 km 2) between the Dry Fork of Piceance Creek and Lee Gulch in the northcentral portion of the Basin. and South Magnolia (83 km 2) between Lee Gulch and Piceance Creek in the south-central portion of the Basin (Fig. l ). Each of these wintering population segments has received varying levels of natural gas development: no development in North Ridge. light development in North Magnolia (0.1 pads & facilities/km 1 ). and relatively high development in the Ryan Gulch (0.6 pads & facilities/krn 2) and South Magnolia (0.9 pads & facilities/km 1) segments (Fig. 1). Among the 4 study areas, North Ridge has served as an unmanipulated control site, Ryan Gulch will serve to address humanactivity management alternatives (BMPs) that benefit mule deer exposed to energy development and as a developed control area for comparison to the developed treatment area receiving habitat improvements (South Magnolia), and North and South Magnolia will allow us to assess the utility of habitat treatments intended to enhance mule deer population performance in areas exposed to light (North Magnolia) and heavy (South Magnolia) energy development activities. METHODS --.., '--" Tasks addressed this period included mule deer capture and collaring efforts. monitoring neonate. overwinter fawn and annual adult female survival. estimating adult female body condition during early and late winter using ultrasonography, estimating mule deer abundance applying helicopter mark-resight surveys, and monitoring vegetation responses to habitat treatments completed spring 2013. We employed helicopter net-gunning techniques (Barrett et al. 1982. van Reenen 1982) to target 240 fawns and 120 adult females during early December 2014. and 120 adult females (primarily recaptures) during early March 2015. Once netted, all deer were hobbled and blind folded. Fawns were weighed, radio-collared and released on site, and adult females were transported to localized handling sites for recording body measurements and fitted with GPS collars (5 fix attempts/day; G2 I 1OD. Advanced Telemetry Systems, Isanti, MN. USA) and released. To provide direct measures of decline in overwinter body condition. we targeted 30 adult females in each study area that were captured the previous December; Vaginal Implant Transmitters (VITs) were also inserted to assist with neonate capture and colJaring efforts spring 2015. Fawn collars were spliced and fitted with rubber surgical tubing to facilitate collar drop between midsummer and autumn, and GPS collars were supplied with timed drop-off mechanisms scheduled to release early in April of the year following deployment: All radio-collars were equipped with mortality sensing options (i.e., increased pulse rate following 8 hrs of inactivity). Mule Deer Habitat Use and Movements --., We downloaded and summarized data from GPS collars deployed and recovered since 2008. GPS collars maintained the same schedule of attempting to collect locations every 5 hours, except for 40 does in Ryan Gulch and 10 control deer from North Ridge where location rates were programmed for every 30-60 minutes to increase resolution of movement data for evaluation of deer behavior patterns in relation to differing development activities. Joe Northrup (CSU PhD Candidate) recently analyzed resource selection data relative to energy development and those results are addressed below. Mule deer resource selection analyses to address success of habitat improvements are pending until vegetation responses are fully realized, which are anticipated by fall 2018. 5 �Mule Deer Survival Mule deer mortality monitoring consisted of daily ground-telemetry tracking and aerial monitoring approximately every 2 weeks from fixed-wing aircraft on winter range and weekly aerial monitoring on summer range. Once a mortality signal was detected, deer were located and necropsied to assess cause of death. We estimated weekly survival using the staggered entry Kaplan-Meier procedure (Kaplan and Meier 1958. Pollock et al. 1989). Capture-related mortalities (any doe/fawn mortalities occurring within IO days of capture; excluding neonates) and collar failures were censored from survival rate estimates. We estimated survival rates from I July 2014 through 30 June 2015 for adult females, from birth to mid December for neonates. and from early December 2014-mid June 2015 for fawns. Adult Female Body Measurements We applied ultrasonography techniques described by Stephenson et al. ( 1998, 2002) and Cook et al. (2001) to measure maximum subcutaneous rump fat (mm), loin depth (longissimus dorsi muscle, mm), and to estimate% body fat. We estimated a body condition score (BCS) for each deer by palpating the rump (Cook et al. 2001~ 2007, 2009). We examined differences (P < 0.05) in nutritional status among study areas and between years using a two-sample t-test. We considered differences in body condition meaningful when mean rump fat or % body fat differed statistically between comparisons. Other body measurements recorded included pregnancy status (pregnant, barren) via blood samples, fetal counts using ultrasonagraphy, weight (kg). chest girth (cm), and hind-foot length (cm). Abundance Estimates We conducted 4 helicopter mark-resight surveys (2 observers and the pilot) during late March to estimate deer abundance in each of the 4 study areas. We delineated each study area from GPS locations collected on winter range during the first 3 years of the study (Jan 2008 through April 2011). Two aerial fixed-wing telemetry surveys/study area were conducted during helicopter mark-resight surveys to determine which marked deer were within each survey area, and we confirmed adult female locations during surveys from GPS data acquired April 2015. We delineated flight paths in ArcGIS 9.3 prior to surveys following topographic contours (e.g.. drainages, ridges) and approximating 500-600 m spacing throughout each study area; flight paths during surveys were followed using GPS navigation in the helicopter. Two approximately 12 x 12 cm pieces of Ritchey livestock banding material (Ritchey Livestock ID, Brighton, CO USA) were uniquely marked using color, number, and symbol combinations and attached to each radio-collar to enhance mark-resight estimates. Each deer observed during surveys was recorded as mark ID#. unmarked, or unidentified mark. We used program MARK (White and Burnham 1999), applying the immigration-emigration mixed logit-nonnal model (McClintock et al. 2008), to estimate mule deer abundance and confidence intervals. For mark-resight model evaluations, we examined parameter combinations of varying detection rates with survey occasion and whether individual sighting probabilities (i.e., individual heterogeneity) were constant or varied (cr~ = 0 or~ 0). Model selection procedures followed the information-theoretic approach of Burnham and Anderson (2002). RESULTS AND DISCUSSION Deer Captures and Survival The helicopter crew captured 242 fawns and 113 does during Dec 2014 and 120 does during March 2015. Eight fawn mortalities (3.3%: proximate cause= 4 capture myopathy, 4 predation) occurred within the 10 day censorship period. Doe monalities totaled I (0.9%; cougar predation) and 7 (5.8%; 6 6 �capture myopathy: 1 cougar predation) within 10 days of the December and March capture periods. respectively. Mortality rates. 10 days post capture. have varied between 2-3% for fawns and 0--3% for does since Jan 2008. except during the 2011-2012 capture season where myopathy rates were higher (36%) due to dry. warm conditions (Anderson and Bishop :W 12). and during the March 2015 doe captures. Nothing abnonnal occurred during March 20 I 5 capture efforts and reasons for the higher. than normal myopathy rate are unclear. Fawn survival from early December 2014 through mid June 20 I 5 was similar (P > 0.05) among study areas ranging from 0.86 to 0.95 (Table 1). General comparisons to previous years suggest relatively high fawn survival occurred during winters 2009-2010, 2013-2014. and 2014-2015 and relatively low survival during winter 20I0-2011 (Fig. 2). which correlates to summer forage condition evident from December fawn weights (Fig. 3) and winter severity. Annual adult female survival varied from 0.81 (Ryan Gulch) to 0.95 (North Magnolia; Table I) during 2014-2015~ but was comparable among study areas during 2014-2015 and to previous years (P > 0.05). with the exception of lower survival in North Magnolia during 2011-2012 (S= 0.68, Anderson and Bishop 2012). Sample sizes for adult female survival do not allow statistical discrimination among years unless large differences are evident (e.g.. >15-20%). Spring Migration Patterns Collaboration with Idaho State University to address mule deer migration patterns in developed and undeveloped landscapes (funded from energy company contributions) has recently been completed. Three manuscripts from this effort have been published (Lendrum et al. 20 I 2, Lendrum et al. :2013; Lendrum et al. 2014; Appendix A). ·'--' In addressing habitat selection during spring migration, Lendrum et al. (2012; Fig. 4) noted that mule deer migrating through the most developed landscapes exhibited longer step lengths (straight line distance between GPS locations) and selected habitats providing greater security cover than deer in undeveloped landscapes that migrated through more open areas that provided increased foraging opportunities. Migrating deer also selected areas closer to well pads, but avoided roads: except in the highest developed areas where road densities were likely too high for avoidance without significant deviations from traditional migration routes. In the second manuscript (Lendrum et al. 2013). we addressed biological and environmental factors influencing spring migration and assessed how energy development influenced migratory behavior. Overall. spring migration was influenced by snow depth, temperature. and green-up on winter and summer range; increasing temperatures. snow melt and emerging vegetation dictated timing of winter range departure and summer range arrival. Duration of Piceance Basin mule deer migration was short. with median migration durations of 3-8 days among the 4 areas (straight line distance between seasonal ranges averaged 32-40 km). Deer in poor condition migrated later than deer in good condition, but condition was similar among areas regardless of development status. Migrating deer from developed study areas did not avoid development structures~ but departed later. arrived earlier and migrated more quickly than deer from undeveloped areas. While large changes in timing of migration could have nutritional consequences and negatively influence reproduction and neonate survival~ the relatively minor shift we observed should not result in long-term fitness consequences. Migratory deer in the Piceance Basin appear to avoid negative effects of energy development through behavioral shifts in timing and rate of migration. In the third publication (Lendrum et al. 2014)~ we monitored migratory mule deer in the Piceance Basin to examined the relationship between the Normalized Difference Vegetation Index (NOVI), which is a course-scale measure of forage quality using a G IS assessment of vegetation greenness, and fecal 7 �nitrogen to assess the assumption that forage quality and deer diets can be reasonably linked to address deer habitat use patterns from remotely sensed data. We found that diet quality evident from fecal nitrogen and course measures of vegetation green-up were informative, and that Piceance Basin mule deer exhibited rapid migration (3 to 8 days depending on study area), left winter range following snow melt with lowest fecal N and NOVI values. and progressed to summer range as vegetation green-up and nitrogen levels increased~ but ahead of peak vegetation green-up on summer range. I suspect this rapid migration strategy is evident for deer in relatively good condition and allows for early arrival on summer range to take advantage optimal forage conditions prior to parturition. Mule Deer Body Condition Early-winter body condition measurements of adult female mule deer were comparable among study areas during December 2014 (P > 0.05) with the exception of South Magnolia does exhibiting greater rump fat than North Ridge does (P < 0.05, Fig. 5, Table 2). Early winter condition this year was comparable to previous years with exception of relatively low condition expressed by North Ridge deer during 2009 and 2012; Ryan Gulch exhibited generally improved condition during December 2011 (Fig. 5). With the exception of Ryan Gulch does exhibiting relatively poor condition during 2014, late winter body condition was higher this year when compared to 2009 and 2011, but lower than North and South Magnolia does during W 10 (Fig. 5~ Table 2), and may be trending upward due to relatively mild winters since 201 1. These observations appear more related to seasonal moisture conditions, relative deer densities (Fig. 6), and winter severity than development intensity thus far. December 2014 fawn weights by study area were comparable with the exception of heavier males from North Ridge in comparison to South Magnolia and heavier females in comparison to Ryan Gulch and North Magnolia (Fig. 3). OveralL fawn weights were relatively low in comparison to previous years (Fig. 3), and December fawn condition has been correlated with winter survival (Fig. 2), but high fawn survival persisted this year likely due to the mild winter and good spring/summer moisture improving forage conditions. More detailed analyses will be conducted to identify factors attributing to these observations. Mule Deer Behavioral Response to Energy Development We recently completed evaluations of deer behavior patterns in relation to energy development activities (Northrup et al. 2015). We found diurnal responses to development activity, where deer used timbered areas away from development activity while bedded during the day and moved into more open areas generally closer to developed areas while foraging at night. Avoidance of producing pads and roads declined from 400 m to 200 m and about 140 m to 60 m from daytime to nighttime, respectively, but increased from 600 m to 800 m for nighttime drilling pad activity. We suspect deer behaviorally respond to fluctuations in development activity, where road traffic and producing well pad activity decline at night. but drilling pad disturbance may increase from compressors and lights used to facilitate nighttime drilling activity. These evaluations were applied during an active drilling phase in the Piceance Basin and deer behavior was compromised in 25% (nighttime) to 50% (day time) of critical winter range during that period. However~ deer densities have increased (Fig. 6) and this suggests that deer can behaviorally mediate development disturbance at some level by taking advantage of fluctuations in development a~tivity to address their nutritional requirements. Given the plasticity in deer behavior, a number of potential options for future development planning exits including drilling schedule modifications (seasonal and/or diurnal). concentrated/staged development~ reducing road traffic~ and using light/noise barriers around drill rigs. It will be interesting to determine if habitat improvements will further reduce development disturbance and increase management options for future development planning. 8 V �'--1' Neonate Survival To complete demographic parameters addressing mule deer-energy development interactions, CPW, Colorado State University, and ExxonMobil Production entered into a collaborative agreement to investigate neonate survival in developed and undeveloped landscapes (funded by ExxonMobil Production Co.) beginning spring 2012. Mark Peterson (Graduate Research Assistant) and Paul Doherty (CSU professor) are assisting with this research. which was completed December 2014. Neonate capture and collaring efforts totaled 85 during spring 2012. 67 during spring 2013. and 54 during spring 2014. Estimated neonate survival through mid-December 2012 was 0.39 (95% Cl= 0.28-0.50). 0.37 (95% CI= 0.25-0.48) from birth to mid-December 20 l 3, and 0.57 (95% Cl= 0.44 - 0. 70) from binh to midDecember 2014. Data are currently being analyzed to address factors potentially influencing neonate mule deer survival from developed and undeveloped landscapes and should be finalized for next year's annual report. Mule Deer Population Estimates " '--1" Mark-resight models that best predicted abundance estimates (lowest AICi:; Burnham and Anderson 2002) exhibited variable sightability across surveys (P,) for all study areas and homogenous individual sightability (cr2 = 0) for South Magnolia deer and variable individual sightability (cr2 -:f; 0) for the other 3 areas. North Ridge exhibited the highest deer density ( 17.5/km\ with comparably lower deer densities in the other 3 areas (8.4-12.0/km 2; Table 3, Fig. 6). Populations have increased over the 7 year monitoring period in 3 of the study areas (from 6.5/km 1 to I0.31km2). with fluctuating population estimates from North Ridge (from 14.3/km 2 to 22.8/km 2); the current North Ridge density estimate is comparable to the mean population estimate of 18.8 deer/km 1 from the past 7 years and generally represents a stable population with the exception of 2013 when the lowest estimate occurred (Fig. 6). The fluctuating population estimates from North Ridge may be somewhat related to a permeable boundary in northwest portion of the study are~ where deer commonly cross the study area boundary. Assuming some of this variability is due to lack of closure. estimation methods will be investigated to attempt to reduce this variation. Abundance estimates from 20 I 4 were similarly precise from all 4 study areas with the mean Confidence Interval Coefficient of Variation (C ICY) ranging from 0. 13-0. 14. Magnolia Habitat Treatments We completed 116 acres of pilot habitat treatments in January 2011 (Anderson and Bishop 2011; Environmental Assessment: 001-BLM-CO- l 10-:20 I l-004-EA). 54 acres of mechanical treatment method comparison treatments (hydro-ax, roller-chop. chain) in January 2012 (Stephens 2014), and 1.038 acres of hydro-ax treatments in April 2013 (Determination of NEPA Adequacy: DOI-BLM-CO-110-2012-0134DNA)~ totaling 604 treated acres in each study area (Fig. 7). Vegetation response in the pilot treatment sites was visually evident by fall 2011 (Fig. 7). and resulted in statistically significant (P < 0.05) increases in native grass and forb cover by the 2014 growing season. 2015 results are pending, but shrub responses appear promising from visual inspection this spring. Stephens (2014) reported that all 3 mechanical treatment methods compared resulted in roughly a 3 fold increase in grasses. forbs, and shrubs combined after 2 growing seasons (versus control sites). but cautioned that rollerchop treatments may be more vulnerable to invasive species response. Vegetative responses from 2013 hydro-ax treatments were visually evident following 1 growing season. but statistical comparisons are pending. As anticipated, grass and forb responses should be evident 2 to 3 year post-treatment. with longer term response expected (3-5 years) from palatable shrubs. Of note, relatively high moisture conditions experienced during spring 2014 and 2015 resulted in higher than nonnal prevalence of cheatgrass (Bromus tectorum); cheatgrass invasion has previously been minor to non-existent. Cheatgrass invasion, however. does not appear directly related to treatment sites 9 �because occurrence is evident in both treatment and control areas. We anticipate this outbreak will subside based on past competitive advantage of native species to dominate, but will continue to monitor species composition and address cheatgrass persistence in treatment and control sites. u GPS data addressing deer use of treatment sites is just becoming available and will be analyzed as additional data are collected and vegetation responses progress. Vegetation and mule deer responses will be documented for the next 3 years to assess the utility of this mitigation approach in benefiting mule deer exposed to energy development disturbance. SUMMARY AND COLLABORATIONS The long-term goal of this study is to investigate habitat treatments and energy development practices that enhance mule deer populations exposed to extensive energy development activity. The information presented here summarizes mule deer population parameters from the 5-year pre-treatment period and 3 years post-treatment. The pretreatment period was completed by spring 2013, providing baseline data for comparison with intended improvements in habitat conditions and response to varying degrees in human development activity. Winter range habitat improvements resulting in 604 acres of mechanically treated pinion-juniper/mountain shrub habitats in each of 2 study areas were completed April 2013. and preliminary vegetation responses appear promising. Post-treatment monitoring will continue for 3 additional years to provide sufficient time to measure how deer respond to these changes. Based on data collected prior to habitat improvements (i.e., pretreatment phase): (1) annual adult survival was consistent among areas averaging 79-87% annually, but overwinter fawn survival was variable, ranging from 48% to 95% within study areas. with annual and study area differences primarily due to annual weather conditions; (2) migratory mule deer selected for areas with increased cover and increased their rate of travel through developed areas, and avoided negative influences through behavioral shifts in timing and rate of migration, but did not avoid development structures; (3) mule deer body condition early and late winter was generally consistent within areas, with higher variability among study areas early winter, which was likely related to seasonal moisture within areas and relative forage capacity among areas; (4) mule deer exhibited behavioral plasticity in relation to energy development, where disturbance distance varied relative to diurnal extent and magnitude of development activity, which may provide for several options in future development planning; and (5) mule deer densities appear to be increasing in 3 of 4 areas, with a stable population in North Ridge. Detailed habitat use analyses are pending for the pre and post-habitat treatment period. We will continue to collect the various population and habitat use data across study sites to evaluate the effectiveness of habitat improvements on winter range. This approach will allow us to determine whether it is possible to effectively mitigate development impacts in highly developed areas, or whether it is better to allocate mitigation dollars toward less or non-impacted areas. In a recent project conducted on the Uncomphahgre Plateau, Colorado, Bergman et al. (2014) found that habitat treatments implemented in pinion-juniper habitat in undeveloped areas increased overwinter survival of fawns by a magnitude of 1.15. Hay field improvements have been completed in the North Magnolia study area by WPX Energy to fulfill a Wildlife Management Plan (VlwlP) agreement with CPW; elk (Cervus elaphus) response has been evident, but mule deer response has thus far been minor. A similar WMP agreement between ExxonMobil/XTO Energy and CPW allowed completion and continued monitoring of mechanical habitat improvements in the Magnolia study areas. Collaborative research with agency biologists, graduate students. and university professors has produced 10 peer-reviewed publications addressing improved monitoring techniques for neonate mule deer captures (Bishop et al. 2011 ), mule deer migration (Lendrum et al. :2012, 2013~ 2014), improved approaches to address animal habitat use patterns (Northrup et al. 2013). mule deer response to helicopter capture and handling (Northrup et al. 2014a), potential effects of male-biased harvest on mule deer productivity (Freeman et al. 2014), mule deer genetics in relation to body condition and migration (Northrup et al. 2014b), spatial and temporal factors influencing 10 u u �~ auditory vigilance in mule deer (Lynch et al. 20 I 4). and the relationship of plant phenology with mule deer body condition (Seral et al. 2015); these publications are summarized in Appendix A. Additional funding and cooperative agreements will be necessary to sustain this project to completion (preferably through 2018). We anticipate the opportunity to work cooperatively toward developing solutions for allowing the nation's energy reserves to be developed in a manner that benefits wildlife and the people who value both the wildlife and energy resources of Colorado. LITERATURE CITED ~~ Anderson, C.R., Jr. 2009. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins. CO. USA. Anderson, C.R., Jr., and D. J. Freddy. 2008a. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Final Study Plan. Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and D. J. Freddy. 2008b. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation-Stage I. Objective 5: Patterns o(mule deer distribution & movements. Pilot Study, Colorado Division of Wildlife. Ft. Collins. CO, USA. Anderson, C.R., Jr., and C. J. Bishop. 2010. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report. Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C. R., Jr., and C. J. Bishop. 2011. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C. R., Jr., and C. J. Bishop. 2012. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Parks and Wildlife. Ft. Collins, CO, USA. Bartmann, R. M. 1975. Piceance deer study-population density and structure. Job Progress Report. Colorado Divison of Wildlife, Fort Collins, Colorado. USA. Bartmann, R. B., and S. F. Steinert. 1981. Distribution and movements of mule deer in the White River Drainage, Colorado. Special Report No. 51. Colorado Division of Wildlife. Fort Collins, Colorado, USA. Bartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatory mortality.in a Colorado mule deer population. Wildlife Monograph No. 121. .--., Barrett, M. W., J. W. Nolan. and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin I 0: I 08-114. Bergman, E. J., C. J. Bishop, D. J. Freddy, G. C. White. and PF. Doherty, Jr. 2014. Habitat management influences over-winter survival of mule deer fawns in Colorado. Journal of Wildlife Management 78(3):448-455; DOI: 10.1002/jwmg.683 Burnham, K. P., and D. R. Anderson. 2002. Model selection and multi-model inference: a practical information-theoretic approach. Second edition. Springer-Verlag, New York. New York. USA. Cook, R. C., J. G. Cook. D. L. Murray. P. Zager. B. K. Johnson, and M. W. Gratson. 2001. Development of predictive models of nutritional condition for rocky mountain elk. Journal of Wildlife Management 65 :973-987. Cook, R. C.. T. R. Stephenson, W. L. Meyers, J. G. Cook. and L.A. Shipley. 2007. Validating predictive models of nutritional condition for mule deer. Journal of Wildlife Management 71: 1934-1943. Cook. R. C., J. G. Cook. T. R. Stephenson, W. L. Meyers, S. M. McCorquodale. D. J. Vales. L. L. Irwin. P. Briggs Hall, R. D. Spencer, S. L. Murphie. K. A. Schoenecker, P. J. Miller. 2009. Revisions of rump fat and body scoring indices for deer. elk. and moose. Journal of Wildlife Management 74:880-896. II �Freeman. E. D.. R. T. Larsen, M. E. Peterson, C.R. Anderson, Jr., K. R. Hersey. and B. R. McMillan. 2014. Effects of male-biased harvest on mule deer: implications for rates of pregnancy, synchrony, and timing of parturition. Wildlife Society Bulletin; DOI: 10.1002/wsb.450 Gibbs, H. D. 1978. Nutritional quality of mule deer foods, Piceance Basin, Colorado. Thesis, Colorado State University. Fort Collins. Colorado, USA. Kaplan. E. L., and P. Meier. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 52:457-481. Lendrum, P. LC. R. Anderson, Jr., R. A. Long, J. K. Kie. and R. T. Bowyer. 2012. Habitat selection by mule deer during migration: effects of landscape structure and natural gas development. Ecosphere 3(9):82. http://dx.doi.orn/10. Lendrum, P. LC. R. Anderson, Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2013. Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes. PLoS ONE 8(5): e64548. doi: I 0.137 l/journal.pone.0064548 Lendrum, P. E., C.R. Anderson, Jr., K. L. Monteith. J. A. Jenks, and R. T. Bowyer. 2014. Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns of forage quality. Mammalian Biology: htto://dx.doi.onz/10.1016/j.mambio.2014.05.005 Lynch, E.. J.M. Northrup. M. F. McKenna, C.R. Anderson Jr., L. Angeloni, and G. Wittemyer. 2014. Landscape and anthropogenic features influence the use of auditory vigilance by mule deer. Behavioral Ecology; doi: 10.1093/beheco/aru 158. Mcclintock. B. T., G. C. White, K. P. Burnham, and M. A. Pride. 2008. A generalized mixed effects model of abundance for mark-resight data when sampling is without replacement. Pages 271289 in D. L. Thompson, E.G. Cooch, and M. J. Conroy, editors, Modeling demographic processes is marked populations. Springer, New York, New York. USA. Northrup, J.M., M. B. Hooten, C.R. Anderson, Jr., and G. Wittemyer. 2013. Practical guidance on characterizing availability in resource selection functions under a use-availability design. Ecology 94(7): 1456-1463. Northrup, J.M., C.R. Anderson, Jr., and G. Wittemyer. 2014a. Effects of helicopter capture and handling on movement behavior of mule deer. Journal of Wildlife Management 78( 4):731-738; DOI: 10.1002/jwmg. 705 Northrup, J.M., A. B. Shafer. C.R. Anderson Jr., D. W. Coltman, and G. Whittemyer. 2014b. Fine-scale genetic correlates to condition and migration in a wild cervid. Evolutionary Applications ISSN 1752-4571; doi: 10.1111/eva.12189 Northrup. J.M., C. R. Anderson, Jr.. and G. Winemyer. 2015. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer. Global Change Biology In press. Pollock. K. H.. S. R. Winterstein. C. M. Bunck, and P. C. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Searle, K. R., M. B. Rice. C.R. Anderson, C. Bishop and N. T. Hobbs. 2015. Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore. Oecologia ISSN 00298549: DOI 10.1007/s00442-015-3348-9 Stephens, G. J. 2014. Understory responses to mechanical removal of pinyon-juniper overstory. MS Thesis~ Colorado State University, Ft. Collins USA. Stephenson. T. R., V. C. Bleich, B. M. Pierce. and G. P. Mulcahy. 2002. Validation of mule deer body composition using in vivo and post-mortem indices of nutritional condition. Wildlife Society Bulletin 30:557-564. • Stephenson. T. R.. K. J. Hundertmark. C. C. Swartz, and V. Van Ballenberghe. 1998. Predicting body fat and mass in moose with untrasonography. Canadian Journal of Zoology 76:717-722. Unsworth. J. W.. D. F. Pack, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado. Idaho, and Montana. Journal of Wildlife Management 63 :315-326. 12 V �'-,/ Van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-4:! 1 in L. Nielsen. J. C. Haigh. and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society. Milwaukee, USA. White. G. C .. and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked individuals. Bird Study 46: 120-139. White, G. C., and B. C. Lubow. 2002. Fitting population models to multiple sources of observed data. Journal of Wildlife Management 66:300-309. Prepared by_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ Charles R. Anderson. Jr .. Mammals Research Leader 13 �Table 1. Survival rate estimates (S) of fawn (1 Dec. 2014-15 June 2015) and adult female (1 July 201430 June 2015) mule deer from 4 winter range study areas of the Piceance Basin in northwest Colorado. Cohort Study area Initial sample size (n) March doe sample3 (n) S (95% Cl) Fawns Ryan Gulch 59 0.932 (0.867-0.996) South Magnolia 58 0.931 (0.865-0.996) North Magnolia 58 0.948 (0.891-1.000) North Ridge 58 0.860 (0.770-0.950) Adult females Ryan Gulch 26 52 0.806 (0.671-0.941) South Magnolia 26 54 0.879 (0.764-0.994) North Magnolia 29 56 0.948 (0.875-1.000) North Ridge 27 49 0.851 (0. 737-0.966) 3 Adult female sample sizes following capture and radio-collaring efforts March~ 2015. 14 �( (/ ) Table 2. Mean rump fat (mm), Body Condition Score (BCS 0), and% body fat(% fat) of adult female mule deer from 4 study areas in the Piceance Basin of northwest Colorado, March and December, 2009-2015. Values in parentheses= SD. March 2010 December 2009 March 2009 Rump fat BCS % fat 8.35(6.36) 4.06(1.13) I 0.54 (3.72) 2.31 (1.44) 2.35 (0.48) 6.37 (1.41) 6.74 (2.27) 10.05 (6.19) 4.07 ( 1.21) 11.44 (3.50) 3.12 (2.20) 2.64 (0.59) 7.11 (1.69) North Magnolia 1.3 I ( 1.01) 2.66 (0.68) 7.15 ( 1.63) 10.67 (5.76) 4.25 (0.96) 11.94 (3.39) 3.15 (2.34) 2.85 (0.53) 7.54 ( 1.53) North Ridge 1.57 ( 1.22) 5.25 (5.65) 3.63 ( 1.11) 1.77 ( 1.1 I) 2.42 (0.4()) 6.39 ( I .45) Study Area Rump fat BCS % fat Ryan Gulch 1.73 (1.78) 2.66 (0.55) 7.08 ( 1.27) South Magnolia 1.29 (0.47) 2.51 (0.66) 2.60 (0.56) 6.81 ( 1.68) Rump fat BCS % fat 9.37 (3.08) Table 2. Continued. Decem her 20 I 1 March 201 I December 20 I 0 BCS % fat Study Area Rump fat RCS % fat Rump fat BCS % fat Ryan Gulch 7.26 (6.36) 3.24 (0.96) 9.69 (3.56) 1.55 (0.60) 2.53 (0.42) 6.72 ( I .37) 13.4 I (6.39) 4.21 ( 1.17) 13.17 (3.64) South Magnolia 9.85 (6.78) 3.30 (0.61) 11.27 (3.75) 1.65 (0.75) 2.35 (0.50) 6.15 (1.75) 8.18 (5.45) 3.41 (0.82) 10.34 (3.28) North Magnolia 9.55 (6.49) 3.46(1.16) 10. 79 ( 4.26) 1.65 (0.67) 2.53 (0.49) 6.79 (1.47) 8.76 (5.77) 3.74 (0.91) 10.73 (3.14) North Ridge 7.25 (5.4 I) 3.47 (0.86) 9.85 (3.02) 1.45 (0.76) 2.24 (0.49) 6.30 ( 1.65) 8.86 (5.37) 3.51 (0.99) I 0.77 (3.33) 15 Rump fat �Table 2. Continued. December 2012 March 2012 March '.2013 % fat Rump fat BCS % fat Rump fat BCS '½, fat 2.15 ( 1.44) 2.74 (0.44) 7.22 ( 1.16) 6.34 (4.35) 3.30 (0.77) 9.34 (2.43) 1.87 (0.90) 2.65 (0.37) 7.14 (0.89) South Magnolia 1.66 (0.77) 2.59 (0.36) 7.03 ( 1.13) 8.30 (5.71) 3.46 ( 1.07) I 0.32 (3.23) 2.06 (0.77) 2.65 (0.26) 7.19 (0.66) North Magnolia 1.90 (0.76) 2.84 (0.34) 7.61 (0.96) 9.66 (6.41) 3.84(1.16) 11.18 (3.64) 1.76 (0.91) 2.59 (0.4 I) 6.87 ( 1.11) North Riclge 2.24 ( 1.58) 2.70 (0.35) 7.26 ( 1.05) 5.76 (4.10) 3.32 (0.82) 9.06 (2.31) 1.87 (0.73) 2.48 (0.34) 6.70 ( 1.12) Study Area Rump fat Ryan Gulch BCS Table~- Continued. March 2014 December 2013 December 2014 % fat Rump fat 9.27 (6.29) 3.47 (0.87) I 0.61 (3.76) 1.69 (0.85) South Magnolia 11.27 (8.40) 3.99 ( 1.04) 11.40 (4.16) 2.57 ( 1.61) 2.96 (0.30) 7.75 (0.68) North Magnolia 9.00 (6.15) 3.44 (0.78) 10.48 (3.25) 2.33 (2.12) 2.80 (0.49) 7.31 (1.43) 9.52 (5.83) 3.83 ( 1.04) 11.18 (3.32) North Ridge 11.17 (5.28) 3.85 (0.72) I 1.66 (2.69) 2.38 ( 1.52) 2.68 (0.39) 7.16 (1.14) 7.93 (5.50) 3.74 (0.76) I 0.20 (3.0 I) Study Area Rump tat Ryan Gulch BCS BCS % fat 2.68 (0.39) 7.03 (0.99) Rump fat 8.50 (6.76) BCS % fat 3.69 ( 1.03) 10.56 (3.70) I 0.96 (6.82) 4.08 ( 1.06) 11.98 (3.81) 16 ( C ( �() Table 2. Continued. March 2015 Study Area Rump fat Ryan Gulch 2.62 (0.95) 2.89 (0.40) 7.44 (0.53) South Magno Iia 2.66 ( 1.36) 2.97 (0.55) 7.62 (0.74) North Magnolia 2.25 (0.97) :!.90 (0.42) 7.49 (0.90) North Ridge 2.28 ( 1.37) 2.92 (0.46) 7.43 ( 1.05) 11 BCS %fat Body condition score taken from palpations of the rump following Cook et al. (2009). 17 �Table 3. Mark-resight abundance (lv) and density estimates of mule deer from 4 winter range herd segments in the Piceance Basin, northwest Colorado. 23-27 March 2015. Data represent 4 helicopter resight surveys from each study area. Study area Mean No. sighted Mean No. marked N(95% CI) Density ( deer/km:) Ryan Gulch 309 31 1A11 (1,231-1.637) 10.0 South Magnolia 202 32 698 (618-800) 8.4 North Magnolia 288 35 950 (841-1,092) 12.0 North Ridge 285 29 931 (820-1.075) 17.5 u 18 �ule deer study areas ri or1n ~.1agno11a Soutn r.1agno1ta Well Pads & Facilities : 1n csevelopment :_ Prooucing \\ ell _ D eYt:IOpm t:nl f3t. lhttes rionh 'l.1age 2$ 10 ~~~~~iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii.lM1les Figure I. Mule deer winter range study areas relative to active natural gas well pads and energy development facilities in the Piceance Basin of northwest Colorado. winter 20 13/ 14 (Accessed http://cogcc.state.co.us/ Dec. 31, 20 13). 19 �Ryan Gulch fawn S South Magnolia fawn S 2008/09-2014/15 2008/09-2014/15 1.00 0.90 1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 ........... .,_,_. o.80 . _;.;~.;..~; ;f:.;.-.,;.~ . 0.70 t==~~.::_.~s~;;t:.~J:~:1~-,.;~ + 0.60 + - - - -- - - ·~ ~ __:__:_;~ ,;.:,a.1,1,,1,jl,l,l,j~ 0.50 +-- - - -~ r:--=:3--r:-:,,....--..0.40 + - -- - -- - - - - ''++-r- - - - -0.30 + - - - - - -- -- - - - - 0.20 - + - - - - - - - - - - - - - - 0.10 + - - - - - - -- -- - - - 0. 00 +-r--r--r-.,-,--r-r-r.-rr-rr,-,-y-,-,-..-.-r-r-r.,--,.-, North Magnolia fawn 5 North Ridge fawn 5 2008/09-2014/15 2008/09-2014/15 1.00 T + ~•~ ==! 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 I i I I I I I '" 1.00 ,-.,.~ !!II! 0.90 +-____e......u.._: 0.80 0.70 t====~~=~~;;;: 0.60 0.50 0.40 + - - -- - - -- - - - - ~~ 0.30 - + - - - - -- - - - -- - - - 0 . 20 + - - - - - - - - - - - - - 0.10 + - - - - - - - - - - - - - o. 00 +-,-~.,-,-.,-,-.,-,-.,-,-.,-,-.,-,-,-,-,-,-,-,-,..,....,--.-,r r , ........... I I I I I I I I I I I I I I I Figure 2. Over-winter (Dec-Mar & June) mule deer fawn survival (5) from 4 study areas in the Piceance Basin, northwest Colorado, 2008/09 (red lines), 2009/10 (orange lines), 20 10/ 11 (blue lines), 20 11 /12 (black lines), 2012/ 13 (purple lines), 2013/ 14 (cyan lines), and 2014/ 15 (brown lines). Solid lines = 5 and dashed lines = 95% Cl. Comparable data among years: December-March 2008-20 10 due to premature collar drop and December-mid-June 20I0-2015. 20 �Male fawn weights 42.0 40.0 32.0 30.0 .. DRyan Gulch I- I- I- ,_ I- I- ,_ ,_ I- I- ,_ I- 1- I- ,_ I- I- .... Dec 2008 ti: -.. 1-- .i.. _ I., l - - Dec 2010 Dec 2009 DSouth Magnolia ,.., 6. - Dec 2011 !::: _ Dec 2012 b, tm D North Ridge 6. l - - 1.. Dec 2013 ■ North Magnolia - I- Dec 2014 Female fawn weights 42.0 40.0 38.0 DRyan Gulch DSouth Magnolia 34.0 ~ 32.0 30.0 ~ I- .... - a~ I- .. - .. .. .. .. ,_ LI 0North Ridge T I- lei. ■ North Magnolia J I- ~ =. . ,. == .. .. = .u. ~ .. Dec 2008 Dec 2009 Dec 2010 Dec 2011 Dec 2012 Dec 2013 Dec 2014 Figure 3. Mean male and female fawn weights and 95% Cl (error bars) from 4 mule deer study areas in the Piceance Basin, northwest Colorado. December 2008-201 4. 21 �Gulch and South Mag Summer Range Figure 4. Mule deer study areas in the Piceance Basin of northwestern Colorado, USA (Top), spring 2009 migration routes of adu lt female mule deer (n = 52; Lower left), and active natural-gas well pads (black dots) and roads (state. county, and natural-gas; white lines) from May 2009 (Lower right; from Lendrum et al. 201 2). 22 �Early winter rump fat 16 14 l 12 ~ 10 ... .!!! 8 a. E ::J a: 6 - North Ridge - North Magnolia Ryan Gulch 4 - south Magnolia ,E 2.0 - North Ridge 0. - North Magnolia 2 0 <) :-,_'\- :-,,Cl R:,°> ~ ~<., '\.,C) ~<., <) <) '\.,C) '\.,C) ~<., ~<., <) ~ 0 :-,_'\.- '\.,C) ~<, <) '\.,C) ~<., <) Late winter rump fat 4.0 3.5 3.0 E .§. 2.5 E 1.5 ::J a: Ryan Gulch 1.0 - South Magnolia 0.5 0.0 Figure 5. Mean early (early Dec., Top) and late winter (early Mar., Bottom) body condition (mm rump fat) of adult female mule deer from 4 winter range study areas in the Piceance Basin of northwest Colorado, March 2009-March 20 15. Error bars = 95% Cl. 23 �Piceance Basin late winter mule deer density u 30.00 25.00 20.00 N E .:ii= ":::- 15.00 - Cl.I Cl.I - North Ridge • • •• • • Ryan Gulch C 10.00 - • North Magnolia - South Magnolia 5.00 0.00 2009 2010 2011 2013 2012 2014 2015 Year Figure 6. Mule deer density estimates and 95% Cl (error bars) from 4 winter range herd segments in the Piceance Basin, northwest Colorado, late winter 2009-2015. u 24 �Nonh Magn,Jl1a ueaten,ent sues (587 acres) Bea,Set_ l 5_~5b_andG Bear5et_ I _8ana;._E [ J 6eJr$et_3 6_5-landJ C,re asewoodSet_g I 6_g29 1?,n~a~ewoodSet_1J 1_g 15 Gre3~ewood5et_ g30_g4 2 LeeOvers,ghts_a_f and I 6_ 17 MethJntc Jl tr~atmer.t compa11$on 154 .1cres) - - ~Jorth HJtc.h PIiot Treatments ( 116 ac,es) - $oulh MJgnolia ue atment :ates (4-10 acres) 5011th MJgnolt7t :! A Figure 7. Habitat treatment site delineations in 2 mule deer study areas (604 acres each) of the Piceance Basin, northwest Colorado (Top; cyan polygons completed Jan. 20 I 1 using hydro-axe; yellow polygons completed Jan. 20 12 using hydro-axe, roller-chop, and chaining; and remaining polygons completed April 2013 using hydro-axe). January 20 11 hydro-axe treatment-site photos from North Hatch Gulch during April (Lower left, aerial view) and October, 2011 (Lower ri ght. ground view). 25 �Appendix A. Abstracts of published manuscripts resulting from Piceance Basin mule deer/energy development interaction research collaborations. Abstract format specific to the respective journal requirements. Effectiveness of a redesigned vaginal implant transmitter in mule deer CHAD J. BISHOP', CHARLES R. ANDERSON Jr. ', DANIEL P. WALSH', ERIC J. BERGMAN', PETER KUECHLE2, and JOHN ROTH 2 'Colorado Parks and Wildlife, Fort Collins, Colorado 80526 USA ' Advanced Telemetry Systems, Isanti, Minnesota 55040 USA Citation: Bishop. C. J .. C.R. Anderson Jr., D. P. Walsh, E. J. Bergman, P. Kuechle, and J. Roth. 2011 . Effectiveness ofa redesigned vaginal implant transmitter in mule deer. Journal of Wildlife Management 75(8): 1797-1806: DOI: I0. 1002/jwmg.229 ABSTRACT Our understanding of factors that limit mule deer ( Odocoileus hemionus) populations may be improved by evaluating neonatal survival as a function of dam characteristics under free-ranging conditions, which generally requires that both neonates and dams are radiocollared. The most viable te.chnique facilitating capture of neonates from radiocollared adult females is use of vaginal implant transmitters (VlTs). To date. VJTs have a llowed research opportunities that we re not previously possible: however. VITs are often expelled from adult females prepartum. which limits the ir effectiveness. We redesigned an ex isting VlT manufactured by Advanced Te lemetry Syste ms (A TS: Isanti, MN) by lengthening and widening wings used to retain the VIT in an adult female. O ur objective was to increase VlT retention rates and thereby increase the likelihood of locating birth sites and newborn fawns. We placed the newly des igned VlTs in 59 adult female mule deer and evaluated the probability of retention to parturition and the probability of detecting newborn fawn s. We also developed an equation for determining VlT sample size necessary to achieve a specified sample size of neonates. The probability of a VIT being retained until parturition was 0 .766 (SE 1/4 0.0605) and the probability ofa VIT being retained to within 3 days of parturition was 0.894 (SE ¼ 0.044 1). In a similar study using the original VlT wings (Bishop et al. 2007), the probability ofa VIT being retained until parturition was 0.447 (SE¼ 0.0468) and the probability of retention to within 3 days of parturition was 0.623 (SE¼ 0.0456). Thus. o ur design modification increased V IT retention to parturition by 0.3 19 (SE '/4 0.0765) and VIT retention to within 3 days of parturition by 0.27 1 (SE ¼ 0 .0634). Considering dams that retained V lTs to within 3 days of parturition, the probability of detecting at least I neonate was 0.952 (SE 1/4 0.0334) and the probability of detecting both fawns from twin litters was 0.588 (SE ¼ 0.0827). We expended approximately 12 person-hours per detected neonate. As a guide for researchers planning future studies. we found that VIT sample size should approximately equal the targeted neonate sample size. Our study expands opportunities for conducting research that links adult female attributes to productivity and offspring s urvival in mule deer. © 2014 The Wildlife Society. Habitat selection by mule deer during migration: effects of landscape structure and natural-gas development PATRICK E. LENDRUM' , CHARLES R. ANDERSON JR.1 , RYAN A. LONG', JOHN G. KIE', AND R. TERRY BOWYER' 'Department of Biological Sciences, Idaho State University. Pocatello, Idaho 83209 USA 'Colorado Parks and Wildliti:, Grand Junction, Colorado 8 1505 USA Citation: Lendrum, P. E., C. R. Anderson Jr., R. A. Long. J. G. Kie, and R. T. Bowyer. 2012. Habitat selection by mule deer during migration: efli:cts of landscape structure and natural-gas development. Ecosphere 3(9):82 hnp.//dx doo.org/10 t890/ESl2-00165. t Abstract. The disruption of traditional m igratory routes by anthropogenic disturbances has shifted patterns of resource selection by man)' species. and in some instances has caused populations to decline. Moreover. in recent decades populations of mule deer (Odocoileus hemionus) have declined throughout much of their historic range in the western United States. We used reso urceselection functions to determine if the presence of natural-gas development altered patterns of resource selection by migrating mule deer. We compared spring migration routes of adult female mule deer fitted with GPS collars (n = 167) among four study areas that had varying degrees of natural-gas development from 2008 to 2010 in the Piceance Basin of northwest Colorado. USA. Mule deer migrating through the most developed area had longer step lengths (straight-line distance between successive GPS locations) compared with deer in less developed areas. Additionally. deer migrating through the most developed study areas tended to select for habitat types that provided greater amounts of concealment cover. whereas deer from the least developed areas te nded to select habitats that increased access to forage and cover. Deer selected habitats closer to well pads and avoided roads in a ll instances except along the most highly deve loped migratory ro utes. where road densities may have been too high for deer to avo id roads without deviating substantially from established migration routes. These results indicate that behavioral tendencies toward avoidance of anthropogenic disturbance can be overridden during migration by the strong fide lity ungulates demonstrate towards migration routes. If avoidance is feasible. then deer may select areas fu rther from development. whereas in highly developed areas. deer may simply increase their rate of travel a long establ ished migration routes. 26 �Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes Patrick E. Lendrum\ Charles R. Anderson Jr.:, Kevin L ~lonteith 1.J, Jonathan A. Jenks\ R. Terrv Bon-ver 1 • Depanment of Biological Sciences. Idaho State University. Pocatello, Idaho. L'S..\. : Colorado D1visio~ of P:i;ks and Wildlife. Gr.ind Junction, Colorado, USA. 3 Wyoming Cooperative Fish and Wildlife Research Unit, L1niversity of Wyoming. Laramie. Wyoming, USA/ Depanment of Natural Resource Management. South Dakota State lJniversity. Brookings. South Dakota. CSA Citation: Lendrum, P. E., C.R. Anderson Jr.. K. L. :\-lomeith. J. A Jenks. R. T. Bowyer. 2013. Migrating :\-lule Deer: Effects of anthropogenically Altered Landscapes. PLoS ONE 8(5): e64548. DOI: I0.1371,joumal.pone.0064548 Abstract Background: Migration is an adaptive strategy that enables animals to enhance resource availability and reduce risk of predation at a broad geographic scale. Ungulate migrations generally occur along traditional routes. many of which have been disrupted by anthropogenic disturbances. Spring migration in ungulates is of panicular importance for conservation planning, because it is closely coupled with timing of parturition. The degree to which oil and gas development affects migratory patterns, and whether ungulate migration is sutliciently plastic to compensate for such changes. warrants additional study to better understand this critical conservation issue. Methodology/Principal Findings: We studied timing and synchrony of depanure from winter range and arrival to summer range of female mule deer (Odocoileus hemionus) in northwestern Colorado. USA. which has one of the largest natural-gas reserves currently under development in North America. We hypothesized that in addition to local weather, plant phenology, and individual life-history characteristics. patterns of spring migration would be modified by disturbances associated with natural-gas extraction. We captured 205 adult female mule deer. equipped them with GPS collars. and observed patterns of spring migration during 2008--2010. Conclusions/Significance: Timing of spring migration was related 10 winter weather (particularly snow depth) and access to emerging vegetation. which varied among years, but was highly synchronous across study areas within years. Additionally. timing of migration was influenced by the collective effects of anthropogenic disturbance. rate of travel. distance traveled. and body condition of adult females. Rates of travel were more rapid over shorter migration distances in areas of high natural-gas development resulting in the delayed departure. but early arrival tbr females migrating in areas with high development compared with less-developed areas. Such shifts in behavior could have consequences for timing of arrival on birthing areas. especially where mule deer migrate over longer distances or for greater durations. Practical guidance on characterizing availability in resource selection functions under a use-availability design JOSEPH M. NORTHRUP', MEVJN 8. HOOTEN 1.!..1, CHARLES R. ANDERSON JR.\ AND GEORGE WITTEMYER' 1 Department offish, Wildlife, and Conservation Biology, Colorado State University. 1474 Campus Delivery. Fort Collins, Colorado 80523 USA U.S. Gc:ological Survey, Colorado Cooperative Fish and Wildlife Research Unit, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 1 Colorado State University, Department of Statistics, Colorado State University. 1474 Campus Delivery. Fort Collins. Colorado 80523 USA ~Mammals Research Section Colorado Parks and Wildlire, 711 Independent Avenue. Gr.ind Junction. Colorado 81505 USA 2 Citation: Northrup. J. M., M. B. Hooten, C. R. Anderson Jr.. and G. Winemyer. 2013. Practical guidance on char.icterizing availability in resomce selection functions under a use-ava1lability design. Ecology <J4( 7): I 45o- l 4o3. hup:1/ux.do1.org/ I0.1890/12-1688. I Abstract. Habitat selection is a fundamental aspect of animal ecology. the understanding of which is critical to management and conservation. Global positioning system data from animals allow fine-scale assessments of habitat selection and typically are analyzed in a use-availability framework. whereby animal locations are contrasted with random locations {the availability sample). Although most use-availability methods are in fact spatial point process models. they often are fit using logistic regression. This framework offers numerous methodological challenges. for which the literature provides little guidance. Specifically. the size and spatial extent of the availability sample influences coetlicienc estimates potentially causing interpretational bias. We examined the influence of availability on statistical inference through simulations and analysis of serially correlated mule deer GPS data. Bias in estimates arose from incorrectly assessing and sampling the spatial extent of availability. Spatial autocorrelation in covariates. which is common for landscape characteristics. exacerbated the error in availability sampling leading to increased bias. These results have strong implications for habitat selection analyses using GPS data. which are increasingly prevalent in the literature. We recommend that researchers assess the sensitivity of their results to their availability sample and. where bias is likely. take care with interpretations and use cross validation to assess robustness. 27 �Effects of Helicopter Capture and Handling on Movement Behavior of Mule Deer JOSEPH M. NORTHRUP 1, CHARLES R. ANDERSON JR\ AND GEORGE WITTEMYER 1 1 Department of Fish. Wildlite. and Conservation Biology, Colorado State University. 1474 Campus Delivery, Fort Collins, Colorado 80523 USA :Mammals Research Section Colorado Parks and Wildlife. 711 Independent Avenue. Grand Junction, Colorado 81505 USA Citation: Northrup. J.M .. C.R. Anderson Jr., and G. Wittemyer. 2014. Effects of helicopter capture and handling on movement behavior of mule deer. Journal of Wildlife Management 78(4):731-738: DOI: 10.1002/j,,mg.705 ABSTRACT Research on wildlife movement. physiology, and reproductive biology often requires capture and handling of animals. Such invasive treatment can alter behavior, which may bias results or invalidate assumptions regarding representative behaviors. To assess the impacts of handling on mule deer (Odocoileus hemionus). a focal species for research in North America we investigated pre- and post-recapture movements of collared individuals. and compared them to deer that were not recaptured (controls). We compared pre- and post-recapture movement rates (m/hr) and 24-hour straight-line displacement among recaptured and control deer. In addition. we examined the time it took recaptured deer to return to their pre-recapture home range. Both daily straight-line displacement and movement rate were marginally elevated relative to monthly averages for 24 hours following recapture, with non-significant elevation continuing for up to 7 days. Comparing movements averaged over 30 days before and after recapture. we found no differences in displacement. but movement rates demonstrated seasonal effects. with faster movements post- relative to pre-recapture in March and slower movements post- relative to pre-recapture in December. Relative to control deer movements. recaptured deer movement rates in March were higher immediately after recapture and lower in the second and third weeks following recapture. The median time to return to the pre-recapture home range was 13 hours, with 71 % of deer returning in the first day. and 91 % returning within 4 days. These results indicate a short period of elevated movements following recaptures. likely due to the deer returning to their home ranges. followed by weaker but non-significant depression of movements for up to 3 weeks. Censoring of the first day of data post capture from analyses is strongly supponed, and removing additional days until the individual returns to its home range will control for the majority of impacts from capture. 2014 The Wildlife Society. e Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns of forage quality Patrick E. Lendrum■• Charles R. Anderson Jr.\ Kevin L. Monteitbc, Jonathan A. Jen~\ R. Terry Bowyer" • Department ofBiological Sciences. Idaho State University. 921 South 8th Avenue. Stop 8007, Pocatello 83209. USA h Mammals Research Section Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction 81505, USA • Wyoming Cooperative Fish and Wildlife Research Unit, Department of Zoology and Physiology, University of Wyoming, 3166, 1000 East University Avenue, Laramie 82071, USA .i Department of Natural Resource Management, South Dakota State University, Box 2140B. Brookings 57007, USA Citation: Lendrum. P. E.. C.R. Anderson Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2014. Relating the movement ofa rapidly migrating ungulate to spatiotcmporal patterns of forage quality. Mammalian Biology: http://dx.doi.or!?./I0.1016/j.mambio.:?014.05.005 ABSTRACT: Migratory ungulates exhibit recurring movements, often along traditional routes between seasonal ranges each spring and autumn. which allow them to track resources as they become available on the landscape. We examined the relationship between spring migration of mule deer (Odocoileus hemionus) and forage quality. as indexed by spatiotemporal patterns of fecal nitrogen and remotely sensed greenness of vegetation (Nonnalized Difference Vegetation Index: NDVI) in spring 2010 in the Piceance Basin of northwestern Colorado. USA. NOVI increased throughout spring, and was affected primarily by snow depth when snow was present. and temperature when snow was absent. Fecal nitrogen was lowest when deer were on winter range before migration. increased rapidly to an asymptote during migration. and remained relatively high when deer reached summer range. Values of fecal nitrogen corresponded with increasing NDVI during migration. Spring migration for mule deer provided a way for these large mammals to increase access to a high-quality diet. which was evident in patterns of NDVI and focal nitrogen. Moreover, these deer ..jumped" rather than '"surfed" the green wave by arriving on summer range well before peak productivity of forage occurred. This rapid migration may aid in securing resources and seclusion from others on summer range in preparation for parturition. and to minimize detrimental factors such as predation. and malnutrition during migration. 28 �Effects of Male-Biased Harvest on Mule Deer: Implications for Rates of Pregnancy, Synchrony, and Timing of Parturition ERIC D. FREEM..\.N 1, RANDY T. LARSEN 1, ~URK E. PETERSON!, CHARLES R. ANDERSOi'i JR.J, KENT R. HERSEY\ A~D BROCK R. :\'lc~OLLA.N' : Department of Plant and Wildlife Sciences. Brigham Young University. 275 WIDB. Provo. UT 84602, USA : Department of Fish. Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins. CO 80523. USA ; Colorado Parks and Wildlite. 71 I Independent Avenue. Grand Junction. CO 81505. USA 'Utah Division ofWildlite Resources. 1594 W North Temple. Salt Lake City. UT 84114. US.A Citation: Freeman. E. D., R. T. Larsen. M. E. Peterson, C.R. Anderson Jr.. K. R. Hersev. and B. R'.'McMillan. 2014. Effects of male-biased harvest on mule deer: implications for rates of pregnancy. synchrony. and timing of parturition. Wildlife Society Bulletin~ DOI: 10.1002/wsb.450 ABSTRACT Evaluating how management practices intluence the population dynamics of ungulates may enhance future management of these species. For example, in mule deer (Odocoi/eus hemionus). changes in male/female ratio due to malebiased harvest may alter rates of pregnancy, timing of parturition. and synchrony of parturition if inadequate numbers of mules are present to fertilize females during their first estrous cycle. If rates of pregnancy or parturition are intluenced by decreased male/female ratios, recruitment may be reduced (e.g.. tewer births. later parturition resulting in lower survival of fawns. and a less synchronous parturition that potentially increases susceptibility of neonates to predation). Our objectives were to compare rates of pregnancy. synchrony of parturition. and timing ofparrurition between exploited mule deer populations with a relatively high (Piceance. CO. USA: 26 males/100 temales) and a relatively low (Monroe, UT. USA: 14 males/100 females) male/female ratio. We detennined rates of pregnancy via ultrasonography and timing of parturition via vaginal implant transmitters. We found no differences in rates of pregnancy (98.6% and 96.6%: z = 0.821: P = 0.794), timing of parturition (estimate= 1.258: SE= 1.672: t = 0.752: P = 0.454), or synchrony of parturition (F= 1.073: P = 0.859) between Monroe Mountain and Piceance Basin. respectively. The relatively low male/female ratio on Monroe Mountain was not associated with a protracted period of parturition. This finding suggests that relatively low male/female ratios typical of heavily harvested populations do not influence population dynamics because recruinnent remains unaffected. Q 2014 The Wildlife Society. Fine-scale genetic correlates to condition and migration in a wild cervid Joseph .\I. Northrup, 1 Aaron B. A. Shafer/ Charles R. Anderson Jr/ Da,·id W. Coltman" and George Wittemyer 1 I Depanment of Fish. Wildlite, and Conservation Biology, Colorado State University. Fon Collins. CO, USA 2 Department of Evolutionary Biology. Evolutionary Biology Centre, Uppsala University, Uppsala. Sweden 3 Mammals Research s~ction, Colorado Parks and Wildlife. Grand Junction, CO. USA 4 Department of Biological Sciences, University of Alberta. Edmonton, AB. Canada. Citation: Northrup. J.M .. A. B. Shater, C.R. Anderson Jr., D. W. Coltman, and G. Whittemyer. :?OJ.i. Fine-scale genetic correlates to condition and migration in a wild cervid. Evolutionary Applications ISSN 1152-4571: doi: I 0. 1111 /eva. 12189 Abstract The relationship between genetic variation and phenotypic traits is fundamental to the study and management of natural populations. Such relationships often are investigated by assessing correlations between phenotypic traits and heterozygosity or genetic differentiation. Using an extensive data set compiled from free ranging mule deer (Odocoi/eus hemionus). we combined genetic and ecological data to (i) examine correlations between genetic differentiation and migration timing. (ii) screen for mitochondrial haplotypes associated with migration timing, and (iii) test whether nuclear heterozygosity was associated with condition. Migration was related to genetic ditlerentiation (more closely related individuals migrated closer in time) and mitochondrial haplogroup. Body fat was related to heterozygosity at two nuclear loci (with antagonistic patterns). one of which is situated near a known fat metabolism gene in mammals. Despite being focused on a widespread panmictic species. these findings revealed a link between genetic variation and imponant phenotypes at a tine scale. We hypothesize that these correlations are either the result of mixing refugial lineages or ditlerential mitochondrial hap lo types intluencing energetics. The maintenance of phenotypic diversity will be critical to enable the potential tracking of changing climatic conditions, and these correlates highlight the need to consider evolutionary mechanisms in management. even in widely distributed panmictic species. 29 �Landscape and anthropogenic features influence the use of auditory vigilance by mule deer Emma Lynch: Joseph M. Northrup,b Megan F. McKenna,c Charles R. Anderson Jr/ Lisa Angeloni,a.c and George Wittemyera." •Graduate Degree Program in Ecology. Colorado State University. 1474 Campus Delivery. Fort Collins. CO 80523. USA hDcpartment offish. Wildlife and Conservation Biology, Colorado State University, 1474 Campus Delivery. Fort Collins, CO 80523. USA ~atural Sounds and Night Skies Division. National Park Service, 1201 Oakridge Drive, Fort Collins, CO 80525, USA. dMammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road. Fort Collins. CO 80526, L:SA ~cpartment of Biology, Colorado State University, 1878 Campus Delivery. Fort Collins. CO 80523. USA Citation: Lynch. E., J.M. Northrup. M. F. McKenna. C. R Anderson Jr., L. Angeloni, and G. Wittemyer. 2014. Landscape and anthropogenic features influence the use of auditory vigilance by mule deer. Behavioral Ecology: doi: l0.10931beheco/arul58. While visual forms of vigilance behavior and their relationship with predation risk have been broadly examined, animals also employ other vigilance modalities such as auditory vigilance by listening for the acoustic cues of predators. Similar to the tradeoffs associated with visual vigilance. auditory behavior potentially structures the energy budgets and behavior of animals. The cryptic narure of auditory vigilance makes it difficult to study. but on-animal acoustical monitoring has rapidly advanced our ability to investigate behaviors and conditions related to sound. We utilized this technique to investigate the ways external stimuli in an active narural gas development field affect periodic pausing by mule deer (Odocoileus hemionus) within bouts of rumination-based mastication. To better understand the ecological properties that structure this behavior, we investigate spatial and temporal factors related to these pauses to determine if results are consistent with our hypothesis that pausing is used for auditory vigilance. We found that deer paused more when in forested cover and at night, where visual vigilance was likely to be less effective. Additionally, deer paused more in areas of moderate background sound levels, though responses to anthropogenic teatures were less clear. Our results suggest that pauses during rumination represent a fonn of auditory vigilance that is responsive to landscape variables. Further exploration of this behavior can tacilitate a more holistic understanding of risk perception and the costs associated with vigilance behavior. Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore Kate R. Searle 1 • }lindy 8. Rice 2 • Charles R. Anderson 2 • Chad Bishop= • N. T. Hobbs3 1 NERC Centre for Ecology and Hydrology, Bush Estate. Penicuik EH26 0QB, UK : Colorado Parks and Wildlife, 317 W. Prospect Road. Fort Collins, CO 80526, USA 3 Department of Ecosystem Science and Sustainability, Colorado State University. Fort Collins 80524, CO, USA Citation: Searle, K. R., M. B. Rice, C.R. Anderson, C. Bishop and N. T. Hobbs. :o 15. Asynchronous vegetation phenology enhances winter body condition ofa large mobile herbivore. Oecologia ISSN 0029-8549: DOI l0.1007/s00442-015-3348-9 Abstract Understanding how spatial and cemporal heterogeneity influence ecological processes tbrms a central challenge in ecology. Individual responses to heterogeneity shape population dynamics, therefore understanding these responses is central to sustainable population managemenL Emerging evidence has shown that herbivores track heterogeneity in nutritional quality of vegetation by responding to phenological differences in plants. We quantified the benefits mule deer (Odocoi/eus hemionus) accrue from accessing habitats with asynchronous plant phenology in northwest Colorado over 3 years. Our analysis examined both the direct physiological and indirect environmental effects of weather and vegetation phenology on mule deer winter body condition. We identified several important effects of annual weather patterns and topographical variables on vegetation phenology in the home ranges of mule deer. Crucially. temporal patterns of vegetation phenology were linked with differences in body condition. with deer tending to show poorer body condition in areas with less asynchronous vegetation green-up and later vegetation onset. The direct physiological effect of previous winter precipitation on mule deer body condition was much less important than the indirect effect mediated by vegetation phenology. 30 V �, Colorado Parks and Wildlife July 1, 2015 -June 30, 2016 ~ State of Cost Center Work Package Task No. Colorado 3430 3001 6 WILDLIFE RESEARCH REPORT : =-Pa=r.:::ks::a.. a=n=d;__W.:. .:. :.::il=dl=::aifi=-e_ _ _ _ _ _ _ _ _ _ __ :=M=a=mm=a=l=s~R=e=se=a=rc=h=--------------: =D-=e-=er~C-==o=ns=e=rv..;...a=t=io=n=--------------: Population Perfonnance of Piceance Basin Mule Deer in Response to Natural Gas Resource Extraction and Mitigation Efforts to Address Human Activity and Habitat Degradation Federal Aid Project: _____W___-.. .;;;1-=8=-5--=R___-___15_______ Period Covered: July 1, 2015 -June 30, 2016 Author: C.R. Anderson, Jr. --...,_ "-,,I Personnel: T. Asnicar, E. Bergman, K. Bond, E. Cardenas, D. Collins, 8. deVergie, D. Finley, M. Fisher, L. Gepfert, J. Hudson, D. Johnston, T. Knowles, D. Lewis, T. Mullins, J. Pelham, 8. Petch, J. Rivale, R. Schilowsky, G. Smith, D. Thibodeau, R. Velarde, T. Verzuh, S. Williams, L. Wolfe, CPW; E. Hollowed, L. Belmonte, BLM; T. Graham, Ranch Advisory Partners; P. Doherty, J. Northrup, M. Peterson, G. Wittemyer, K. Wilson, Colorado State University; R. Swisher, S. Swisher, Quicksilver Air, Inc.; D. Felix, Olathe Spray Service, Inc.; L. Coulter, Coulter Aviation. Project support received from Federal Aid in Wildlife Restoration, Colorado Mule Deer Association, Colorado Mule Deer Foundation, Muley Fanatic Foundation, Colorado State Severance Tax Fund, EnCana Corp., ExxonMobil Production Co./XTO Energy, Marathon Oil Corp., Shell Petroleum, and WPX Energy. All information in this report is preliminary and subject to further evaluation. Information MAV NOT BE PUBLISHED OR QUOTED without permission of the authors. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT ---.. ~ We propose to experimentally evaluate winter range habitat treatments and human-activity management alternatives intended to enhance mule deer (Odocoileus hemionus) populations exposed to energy-development activities. The Piceance Basin of northwestern Colorado was selected as the project area due to ongoing natural gas development in one of the most extensive and important mule deer winter and transition range areas in Colorado. The data presented here represent the first 9 years of data (5 years of pretreatment, 4 years post treatment) of a long-term study addressing habitat improvements and evaluation of energy development practices intended to improve mule deer fitness in areas exposed to extensive energy development. We monitored deer on 4 winter range study areas representing varying levels of development to serve as treatment (North Magnolia, South Magnolia) and control (North Ridge, Ryan Gulch) sites. We recorded habitat use and movement patterns, estimated neonatal, overwinter fawn and annual adult female survival, estimated early and late winter body condition of adult females, and estimated abundance. During this research segment, we targeted 240 fawns (60/study area) and 120 does (30/study area) in early December 2015 for VHF and GPS radiocollar attachment, respectively, and adult female body condition assessment. We attempted recapture of 120 does (30/study area) and 40 fawns (20 in 2 study areas) in March 2016 for late winter body condition assessment. Winter range habitat improvements completed spring 2013 resulted in 604 acres of mechanically treated pinionI 1iiililliii BDOW029663 �juniper/mountain shrub habitats in each of the 2 treatment areas with minor and extensive energy development, respectively. Post-treatment monitoring will continue for 2 additional years to provide sufficient time to measure how vegetation and deer respond to these changes. Based on data collected prior to habitat improvements (i.e., pretreatment phase): ( 1) annual adult survival was consistent among areas averaging 79-87% annually, but overwinter fawn survival was variable, ranging from 48% to 95% within study areas, with annual and study area differences primarily due to early winter fawn condition and annual weather conditions; (2) migratory mule deer selected for areas with increased cover and increased their rate of travel through developed areas, and avoided negative influences through behavioral shifts in timing and rate of migration, but did not avoid development structures; (3) mule deer body condition was generally consistent within areas, with higher variability among study areas early winter, primarily due to December lactation rates, and late winter condition appeared related to seasonal moisture and winter severity; (4) mule deer exhibited behavioral plasticity in relation to energy development, where disturbance distance varied relative to diurnal extent and magnitude of development activity, which may provide for several options in future development planning; (5) late winter mule deer densities have increased in all study areas, averaging up to 6% annual growth rates since 2008; and (6) post treatment vegetation responses have provided evidence of improved forage conditions, but longer term monitoring will be required to address the full potential of habitat mitigation efforts. Detailed habitat use analyses are still pending for the pretreatment period. We will continue to collect population and habitat use data across all study sites to evaluate the effectiveness of habitat improvements on winter range. This approach will allow us to determine whether it is possible to effectively mitigate development disturbances in highly developed areas, or whether it is better to allocate mitigation efforts toward less or non-impacted areas. In collaboration with Colorado State University, we monitored neonate survival in relation to energy development on all study areas. This wilJ allow us to include neonatal and parturition data with other demographic parameters to evaluate mule deer/energy development interactions. This study is slated to continue through 2018 to allow sufficient time for measuring mule deer population responses to landscape level manipulations. 2 '"'-,,,) �WILDLIFE RESEARCH REPORT POPULATION PERFORMANCE OF PICEANCE BASIN MULE DEER IN RESPONSE TO NATURAL GAS RESOURCE EXTRACTION AND MITIGATION EFFORTS TO ADDRESS HUMAN ACTIVITY AND HABIT AT DEGRADATION CHARLES R. ANDERSON, JR PROJECT NARRITIVE OBJECTIVES 1. To determine experimentally whether enhancing mule deer habitat conditions on winter range elicits behavioral responses, improves body condition, increases fawn survival, and ultimately, population density on mule deer winter ranges exposed to extensive energy development. 2. To determine experimentally to what extent modification of energy development practices enhance habitat selection, body condition, fawn survival, and winter range mule deer densities. SEGMENT OBJECTIVES ", 1. Collect and reattach GPS collars to maintain sample sizes for addressing mule deer habitat use and behavioral patterns in 4 study areas experiencing varying levels of energy development of the Piceance Basin, northwest Colorado. ~ 2. Estimate early and late winter body condition of adult female mule deer in each of the 4 winter herd segments using ultrasound techniques. Estimate early and late winter fawn weights in areas with and without habitat treatments to assess winter fawn condition relative to habitat improvements. 3. Monitor over-winter fawn and annual adult female mule deer survival by daily ground tracking and bi- weekly aerial tracking. 4. Conduct Mark-Resight helicopter surveys to estimate late winter mule deer abundance and density in each study area. 5. Monitor habitat treatment response for assessing efficacy of habitat improvement projects to mitigate energy development disturbances to mule deer. 6. Continue neonate survival and adult female parturition evaluations to complete demographic parameters for assessing mule deer/energy development interactions. INTRODUCTION ..-..._, '--" Extraction of natural gas from areas throughout western Colorado has raised concerns among many public stakeholders and Colorado Parks and Wildlife (CPW) that the cumulative impacts associated with this intense industrialization will dramatically and negatively affect the wildlife resources of the region. Concern is especially high for mule deer due to their recreational and economic importance as a principal game species and their ecological importance as one of the 3 �primary herbivores of the Colorado Plateau Ecoregion. Extraction of natural gas will directly affect the potential suitability of the landscape used by mule deer through conversion of native habitat vegetation with drill pads, roads, or introduction of noxious weeds, by fragmenting habitat with drill pads and roads, by increasing noise levels via compressor stations and vehicle traffic, and by increasing the year-round presence of human activities. Extraction will indirectly affect deer by increasing the human work-force population of the region resulting in the need for additional landscape conversion for human housing, supporting businesses, and upgraded road/transportation infrastructure. Additionally, increased traffic on rural roads will raise the potential for vehicle-animal collisions. Thus, research documenting these relationships and evaluating the most effective strategies for minimizing and mitigating these activities will greatly enhance future management U efforts to sustain mule deer populations for future recreational and ecological values. The Piceance Basin in northwest Colorado contains one of the largest migratory mule deer populations in North America and also covers some of the largest natural gas reserves in North America. Projected energy development throughout northwest Colorado within the next 20 years is expected to reach about 15,000 wells, many of which will occur in the Piceance Basin, which currently supports over 250 active gas well pads (http://cogcc.state.co.us; Fig. 1). Anderson and Freddy (2008a) in their long- term research proposal identified 6 primary study objectives to assess measures to offset impacts of energy extraction on mule deer population performance. During the first 5 years of this study, we gathered baseline habitat utilization and demographic data from radiocollared deer across the Piceance Basin to allow assessment of habitat mitigation approaches that were completed April 2013. We are currently monitoring 2 control areas: 1 with development (0.6 pads & facilities/km 2 ; Ryan Gulch) and I without (North Ridge). The control areas will be compared with 2 treatment areas experiencing similar development intensities (South Magnolia, 0.9 well pads & facilities/km 2 and North Magnolia, 0.1 well pads & facilities/km2), that also recently received habitat improvements {604 acres each). Habitat and mule deer responses to mechanical habitat treatments will be evaluated until spring 2018 to assess the success of this habitat mitigation strategy to benefit mule deer exposed to energy development disturbance. In addition, mule deer behavioral patterns in relation to energy development activities in the Ryan Gulch area are being monitored to identify effective Best Management Practices {BMPs) for future energy development planning. This progress report describes the previous 8.5 years (Jan 2008-June 2015) of mule deer population performance during the pretreatment phase on 4 winter range herd segments, which includes monitoring habitat selection and behavior patterns of adult female mule deer; spring/summer neonate, overwinter fawn and annual adult female survival; estimates of adult female body condition during early and late winter; and annual late-winter abundance/density estimates. STUDY AREAS The Piceance Basin, located between the cities of Rangely, Meeker, and Rifle in northwest Colorado, was selected as the project area due to its ecological importance as home to one of the largest migratory mule deer populations in North America and because it exhibits one of the highest natural gas reserves in North America (Fig. I). Historically, mule deer numbers on winter range were estimated between 20,000-30,000 (White and Lubow 2002), and the current number of well pads (Fig.I) and projected number of gas wells in the Piceance Basin over the next 20 years is about 250 and 15,000, respectively. Mule deer winter range in the Piceance Basin is predominantly characterized as a topographically diverse pinion pine (Pinus edulis)-Utah juniper (Jwziperus osteosperma; pinionjuniper) shrubland complex ranging from 1,675 m to 2,285 min elevation (Bartmann and Steinert 1981). Pinion-juniper are the dominant overstory species and major shrub species include Utah serviceberry (Amelanchier utahensis), mountain mahogany ( Cercocarpus montanus), bitterbrush (Purshia tridentata), big sagebrush (Artemisia tridentata), Gamble's oak (Quercus gambelii), mountain snowberry (Symphoricarpos oreophilus), and rabbitbrush (Chrysothamnus spp.; Bartmann et al. 1992). The Piceance Basin is segmented by numerous drainages characterized by stands ofbig 4 U �sagebrush, saltbush (Atriplex spp.), and black greasewood (Sarcobatus vermiculatus), with the majority of the primary drainages having been converted to mixed- grass hay fields. Grasses and forbs common to the area consist ofwheatgrass (Agropyron spp.), blue grama (Bouteloua gracilis), needle and thread (Stipa comata), Indian rice grass (Oryzopsis lzymenoides), arrowleafbalsamroot (Balsamorhi=a sagittata), broom snakeweed (Gutierrezia sarotlzreae), pinnate tansymustard (Descurainia pinnata), milkvetch (Astragalus spp.), Lewis flax (Linum lewisii), evening primrose (Oenothera spp.), skyrocket gilia (Gilia aggregata), buckwheat (Erigonum spp.), Indian paintbrush (Castilleja spp.), and penstemon (Penstemon spp.; Gibbs 1978). The climate of the Piceance Basin is characterized by wam1 dry summers and cold winters with most of the annual moisture resulting from spring snow melt and brief summer monsoonal rain storms. -..., \._.I Wintering mule deer population segments we are investigating include: North Ridge (53 km2) just north of the Dry Fork of Piceance Creek including the White River in the northeastern portion of the Basin, Ryan Gulch ( 141 km2) between Ryan Gulch and Dry Gulch in the southwestern portion of the Basin, North Magnolia (79 km2) between the Dry Fork of Piceance Creek and Lee Gulch in the north- central portion of the Basin, and South Magnolia (83 km2) between Lee Gulch and Piceance Creek in the south-central portion of the Basin ( Fig. 1). Each of these wintering population segments has received varying levels of natural gas development: no development in North Ridge, light development in North Magnolia (0.1 pads & facilities/km\ and relatively high development in the Ryan Gulch (0.6 pads & facilities/km2) and South Magnolia (0.9 pads & facilities/km 2) segments (Fig. 1). Development activity was high through 2011 and has declined substantially since natural gas prices began to decline in 2012. Among the 4 study areas, North Ridge has served as an unmanipulated control site, Ryan Gulch will serve to address human-activity management alternatives (BMPs) that benefit mule deer exposed to energy development and as a developed control area for comparison to the developed treatment area receiving habitat improvements (South Magnolia), and North and South Magnolia will allow us to assess the utility of habitat treatments intended to enhance mule deer population performance in areas exposed to light (North Magnolia) and relatively heavy (South Magnolia) energy development activities. METHODS Tasks addressed this period included mule deer capture and collaring, monitoring neonate, overwinter fawn and annual adult female survival, estimating adult female body condition during early and late winter using ultrasonography and winter fawn condition using early and late winter fawn weights, estimating mule deer abundance applying helicopter mark-resight surveys, and monitoring vegetation responses to habitat treatments completed spring 2013. We employed helicopter net- gunning techniques (Barrett et al. 1982, van Reen en 1982) to target 240 fawns and 120 adult females during early December 2015, and 120 adult females and 40 fawns (primarily recaptures) during early March 2016. Once netted, all deer were hobbled and blind folded. Fawns were weighed and radiocollared, and sex was recorded prior to release at the capture site. Adult females were transported to localized handling sites for recording body measurements and fitted with OPS collars (5 fix attempts/day; G2110D, Advanced Telemetry Systems, Isanti, MN, USA) prior to release. To provide direct measures of decline in overwinter body condition, we targeted 30 adult females in each study area that were captured the previous December; Vaginal Implant Transmitters (VITs) were also inserted in does on the Ryan Gulch and South Magnolia study areas to assist with neonate capture and collaring efforts spring 2016. During March, 20 fawns were recaptured, weighed and released in South Magnolia (in the habitat treatment areas) and Ryan Gulch (control area) to quantify overwinter declines in fawn body condition. Fawn collars were spliced and fitted with rubber surgical tubing to facilitate collar drop between mid-summer and autumn for winter fawns and during winter for neonates, and GPS collars were supplied with timed drop-off mechanisms scheduled to release early April of the year following deployment. All radio-collars were equipped with mortality sensing options (i.e., increased pulse rate following 8 hrs of inactivity). 5 �Mule Deer Habitat Use and Movements We downloaded and summarized data from GPS collars deployed and recovered since 2008. GPS collars maintained the same schedule of attempting to collect locations every 5 hours, except for 40 does in Ryan Gulch and IO control deer from North Ridge where location rates were programmed for every 30-60 minutes to increase resolution of movement data for evaluation of deer behavior patterns in relation to differing development activities. Joe Northrup (CSU PhD Candidate) recently analyzed resource selection data relative to energy development (Northrup 2015) and those results are addressed below. Mule deer resource selection analyses to address success of habitat improvements are pending until vegetation responses are fully realized, which are anticipated by fall 2018. U Mule Deer Survival Mule deer mortality monitoring consisted of daily ground-telemetry tracking and aerial monitoring approximately every 2 weeks from fixed-wing aircraft on winter range and weekly aerial monitoring on summer range. Once a mortality signal was detected, deer were located and necropsied to assess cause of death. We estimated weekly survival using the staggered entry Kaplan-Meier procedure (Kaplan and Meier 1958, Pollock et al. 1989). Capture-related mortalities (any doe/fawn mortalities occurring within 10 days of capture; excluding neonates) and collar failures were censored from survival rate estimates. We estimated survival rates from 1 July 2015 through 30 June 2016 for adult females, from birth to mid December for neonates, and from early December 2015-mid June 2016 for winter fawns. Adult Female Body Measurements We applied ultrasonography techniques described by Stephenson et al. (1998, 2002) and Cook et al. (2001) to measure maximum subcutaneous rump fat (mm), loin depth (longissimus dorsi muscle, mm), and to estimate% body fat. We estimated a body condition score (BCS) for each deer by palpating the rump (Cook et al. 2001, 2007, 2009). We examined differences (P < 0.05) in nutritional status among study areas and between years evident in non-overlapping 95% confidence intervals. We considered differences in body condition meaningful when mean rump fat or % body fat differed statistically between comparisons. Other body measurements recorded included pregnancy status (pregnant, barren) via blood samples, fetal counts using ultrasonagraphy, weight (kg), chest girth (cm), and hind-foot length (cm). Abundance Estimates We conducted 4 helicopter mark-resight surveys (2 observers and the pilot) during late March to estimate deer abundance in 3 of the 4 study areas and conducted a 5th survey in South Magnolia. We delineated each study area from GPS locations collected on winter range during the first 3 years of the study (Jan 2008 through April 2011 ). Two aerial fixed-wing telemetry surveys/study area were conducted during helicopter mark-resight surveys to determine which marked deer were within each survey area, and we confirmed adult female locations during surveys from GPS data acquired April 2016. We delineated flight paths in ArcGIS 10.0 prior to surveys following topographic contours (e.g., drainages, ridges) and approximating 500-600 m spacing throughout each study area; flight paths during surveys were followed using GPS navigation in the helicopter. Two 12 x 12 cm pieces of Ritchey livestock banding material (Ritchey Livestock ID, Brighton, CO USA) were uniquely marked using color, number, and symbol combinations and attached to each radio-collar to enhance mark-resight estimates. Each deer observed during surveys was recorded as mark ID#, unmarked, or unidentified mark. 6 U �We used program MARK (White and Burnham 1999), applying the immigration-emigration mixed logit-normal model (McClintock et al. 2008), to estimate mule deer abundance and confidence intervals. For mark-resight model evaluations, we examined parameter combinations of varying detection rates with survey occasion and whether individual sighting probabilities (i.e., individual heterogeneity) were constant or varied (cr2 = 0 or-:;; 0). Model selection procedures followed the infom1ation-theoretic approach of Burnham and Anderson (2002). RESULTS AND DISCUSSION Deer Captures and Survival The helicopter crew captured 242 fawns and 123 does during Dec 2015 and 115 does and 40 fawns during March 2016. Five fawn mortalities (2.1 %; proximate cause = 4 capture myopathy, I predation) occurred within the IO day censorship period during December and 4 fawn mortalities ( 10.0%; 3 capture myopathy, 1 predation) occurred during the March capture. Doe mortalities totaled 2 (1.6%; capture myopathy) and 4 (3.5%; capture myopathy) within 10 days of the December and March capture periods, respectively. Mortality rates 10 days post capture have typically varied between 2.5-3.5% for fawns and does since Jan 2008, except during the 2011-2012 capture season where myopathy rates were higher (3-6%) due to dry, warm conditions (Anderson and Bishop 2012). Excluding March fawn captures, myopathy rates were below expected levels during December and within normal levels during March. The relatively high myopathy rates for late winter fawn captures were likely linked to the severe winter conditions evident through January 2016. Fawn survival from early December 2015 through mid June 2016 was similar (P > 0.05) among study areas ranging from 0.74 to 0.84, with the exception of North Ridge where survival was lower (49%; Table I). A higher rate of malnutrition/disease (30%) was documented for North Ridge fawns during the 2015-16 winter than in previous years and other study areas (typically <5%). Premature collar drop during 2008-09 and 2009-l O did not allow for winter fawn survival estimates past late March, but survival rates among study areas were similar (P < 0.05) each year and comparable to 20·1 l-12 and 2012-13 (excluding North Ridge) during 2008-09 and to the higher survival rates from 2013-14 and 2014-15 during 2009-10 (Fig. 2). General comparisons to previous years suggest moderate to high fawn survival occurred during most winters and study areas with the exception of winter 2010-2011 for 3 of the 4 study areas and for North Ridge during winter 2012-13 and 2015-16 (Fig. 2). Low winter fawn survival (Fig. 2) appears to correlate with summer forage condition evident from lower December fawn weights (Fig. 3 ). Annual adult female survival varied from 0.72 (South Magnolia) to 0.89 (North Magnolia; Table I) during 2015-16, but was comparable among study areas during 2015-16 and to previous years (P > 0.05), with the exception oflower survival in North Magnolia during 2011-12 (S= 0.68, Anderson and Bishop 2012). Relatively low sample sizes per study area for adult female survival do not allow statistical discrimination among years unless large differences are evident (e.g., >15-20%). Estimates below 80% are biologically concerning if these values represent the respective population, but low statistical power precludes confirmation within study areas. When combined among study areas, annual survival estimates have varied from 79% in 2012-13 to 86% in 2014-15. Lower combined survival estimates are consistent with extreme environmental conditions consisting of dryer moisture conditions during late winter/spring (2012-13) or cold temperatures with heavy snow during early winter (2015-16). Spring Migration Patterns Collaboration with Idaho State University to address mule deer migration patterns in 7 �developed and undeveloped landscapes (funded from energy company contributions) has recently been completed. Four manuscripts from this effort have been published (Lendrum et al. 2012, Lendrum et al. 2013, Lendrum et al. 2014, Anderson and Bishop; Appendix A). u In addressing habitat selection during spring migration, Lendrum et al. (2012; Fig. 4) noted that mule deer migrating through the most developed landscapes exhibited longer step lengths (straight line distance between GPS locations) and selected habitats providing greater security cover than deer in undeveloped landscapes that migrated through more open areas that provided increased foraging opportunities. Migrating deer also selected areas closer to well pads, but avoided roads, except in the highest developed areas where road densities were likely too high for avoidance without significant deviations from traditional migration routes. In the second manuscript Lendrum et al. (2013) addressed biological and environmental factors influencing spring migration and assessed how energy development influenced migratory behavior. Overall, spring migration was influenced by snow depth, temperature, and green-up on winter and summer range; increasing temperatures, snow melt and emerging vegetation dictated timing of winter range departure and summer range arrival. Duration of Piceance Basin mule deer migration was short, with median migration durations of 3-8 days among the 4 areas (straight line distance between seasonal ranges averaged 32-40 km). Deer in poor condition migrated later than deer in good condition, but condition was similar among areas regardless of development status. Migrating deer from developed study areas did not avoid development structures, but departed later, arrived earlier and migrated more quickly than deer from undeveloped areas. While large changes in timing of migration could have nutritional consequences and negatively influence reproduction and neonate survival, the relatively minor shift we observed should not result in long-term fitness consequences. Migratory deer in the Piceance Basin appear to avoid negative effects of energy development through behavioral shifts in timing and rate of migration. In the third publication Lendrum et al. (2014), monitored migratory mule deer in the Piceance Basin to examined the relationship between the Normalized Difference Vegetation Index (NOVI), which is a course-scale measure of forage quality using a GIS assessment of vegetation greenness, and fecal nitrogen to assess the assumption that forage quality and deer diets can be reasonably iinked to address deer habitat use patterns from remotely sensed data. We found that diet quality evident from fecal nitrogen and course measures of vegetation green-up were informative, and that Piceance Basin mule deer exhibited rapid migration (3 to 8 days depending on study area), left winter range following snow melt with lowest fecal N and NOVI values, and progressed to summer range as vegetation green-up and nitrogen levels increased, but ahead of peak vegetation green-up on summer range. I suspect this rapid migration strategy is evident for deer in relatively good condition and allows for early arrival on summer range to take advantage optimal forage conditions prior to parturition. Anderson and Bishop (2014) summarized results from Lendrum et al. (2012, 2013) and Sawyer et al. (2012) addressing migratory mule deer and energy development in northwest Colorado and south-central Wyoming, respectively. The interactions between migratory mule deer and energy development identified by Lendrum et al. (2012, 2013) and Sawyer et al. (2012) suggest mule deer may benefit from energy development planning by considering thresholds of development that may alter migratory behavior. It appears that migration rate, migration routes, and stopover use, if present, may be altered at high development intensities. In addition, migratory mule deer may benefit by maintaining security cover along migration paths, and improved habitat conditions may facilitate more direct and rapid migration requiring less energy to complete migration. Enhancing permeability along migration routes by applying dispersed development plans (<2 well pads/km2 ) and minimizing disturbance to vegetation types by maintaining security cover should reduce impacts to migratory mule deer as well as other migratory ungulates. Where feasible, habitat improvement projects on 8 u �winter range and possibly stopover sites would also enhance migratory mule deer populations by enhancing energy reserves for long-distance movements and parturition shortly after summer range arrival. Where possible, directional drilling could be used to extract energy resources from underneath migration routes while maintaining no surface occupancy. Lastly, we emphasize that GPS studies now allow managers to accurately map migration routes for entire populations and identify relatively narrow corridors that are most heavily used thus allowing for the identification of the most important corridors for migrating ungulates. Where available, we encourage agencies to incorporate such migration corridors into land-use plans (e.g., resource management plans) and National Environmental Policy Act documents. Mule Deer Body Condition Early-winter body condition measurements of adult female mule deer during December 2015 were relatively high from Ryan Gulch does (P > 0.05) and relatively low from North Ridge does (P < 0.05, Fig. 5, Table 2). Early winter- condition this year was comparable to previous years with the exception ofrelatively low condition expressed by North Ridge deer during 2009, 2012 and 2015 and relatively high condition from Ryan Gulch does during December 2011 and 2015 (Fig. 5). Adult female body condition during early winter appears primarily related to the proportion of lactating does identified during December captures, where higher condition correlates with lower lactation rates. With the exception of Ryan Gulch and North Ridge does exhibiting relatively poor condition during 2014 and 2016, respectively, late winter body condition recently trended upward, but either stabilized or declined this past winter (North Ridge; Fig. 5, Table 2). The recent increase was likely related to relatively mild winters since 2011 and the downturn this past winter is likely related to the relatively severe winter conditions evident through January consisting oflow temperatures and deep snow on winter range; temperatures increased during February resulting in rapid snow-melt, which likely improved late winter condition above what might have resulted if severe winter conditions continued. Adult female body condition thus far appears more related to early winter lactation rates, seasonal moisture conditions, relative deer densities (Fig. 6), and winter severity than observed development intensity thus far. December 2015 fawn weights by study area were comparable with the exception of generally lighter fawns from North Ridge (Fig. 3). Overall, excluding North Ridge, fawn weights were moderate in comparison to previous years (Fig. 3 ), and December fawn condition has been correlated with winter fawn survival (Fig. 2), which was consistent with the low winter survival of North Ridge fawns this past winter. Because adult female body condition has been largely uninformative in regards to habitat treatment responses (pending further analyses), we began late winter fawn recaptures in South Magnolia (habitat treatment area) and Ryan Gulch (control area) to assess changes in over-winter condition. Weight loss was significantly less (P < 0.001) for fawns from the area receiving habitat treatments than for fawns from the untreated area. We will continue monitoring winter weight loss for fawns from the treatment and control area to evaluate over-winter fawn condition in areas with and without habitat improvements. Mule Deer Behavioral Response to Energy Development We recently completed evaluations of deer behavior patterns in relation to energy development activities (Northrup et al. 2015). We found diurnal responses to development activity, where deer used timbered areas away from development activity while bedded during the day and moved into more open areas generally closer to developed areas while foraging at night. Disturbance distances from producing pads and roads declined from 600 m to 200 m and about 140 m to 60 m from daytime to nighttime, respectively, but increased from 600 m to 800 m for nighttime drilling pad activity. We 9 �suspect deer behaviorally respond to fluctuations in development activity, where road traffic and producing well pad activity decline at night, but drilling pad disturbance may increase from compressors and lights used to facilitate nighttime drilling activity. These evaluations were applied during an active drilling phase in the Piceance Basin and deer use was influenced by development activity in 25% (nighttime) to 50% (day time) of critical winter range during that period. However, deer densities have comparably increased among developed and undeveloped study areas (Fig. 6) suggesting that deer can behaviorally mediate development disturbance under observed development and deer densities by taking advantage of fluctuations in development activity to address their nutritional requirements. Given the plasticity in deer behavior, a number of potential options for future development planning exits including drilling sc~edule modifications (seasonal and/or diurnal), concentrated/staged development, reducing road traffic, and using light/noise barriers around drill rigs. It will be interesting to determine if habitat improvements will further reduce development disturbance and increase management options for future development planning. Neonate Survival To complete demographic parameters addressing mule deer-energy development interactions, CPW, Colorado State University, and ExxonMobil Production entered into a collaborative agreement to investigate neonate survival and adult female parturition in developed and undeveloped landscapes (funded by ExxonMobil Production Co.) beginning spring 2012. Mark Peterson (Graduate Research Assistant) and Paul Doherty (CSU professor) assisted with this research, which was completed December 2014, and continued by CPW during 2015. Neonate capture and collaring efforts totaled 85 during spring 2012, 67 during spring 2013, 54 during spring 2014, and 59 during spring 2015. Estimated neonate survival through mid-December was 0.39 (95% CI= 0.28-0.50) during 2012, 0.37 (95% CI= 0.25-0.48) during 2013, 0.57 (95% CI= 0.44-0.70) during 2014, and 0.36 (95% CI= 0.23-0.49) during 2015. Through December 2015, predation was the highest mortality factor (averaging about 50% annually), with relatively low incidents of starvation/disease directly influencing spring/summer fawn survival (mean= 3.2%). Manuscripts addressing neonate fawn survival and adult female parturition in relation to energy development are currently in preparation and review for publication. Mule Deer Population Estimates Mark-resight models that best predicted abundance estimates (lowest AICc; Burnham and Anderson 2002) exhibited variable sightability across surveys (P,) for all study areas and variable individual sightability (cr2 = 0) for North Magnolia deer and homogenous sightability (cr2 -;/:. 0) for the other 3 areas. North Ridge exhibited the highest deer density (26.0/km2), with comparable but lower deer densities in the other 3 areas (10.7-13.2/km2; Table 3, Fig. 6). Densities increased over the 8 year monitoring period in all study areas ranging from an estimated 50% increase in North Ridge to a 103% increase in North Magnolia (mean estimated increase across study areas= 78%); North Ridge deer appeared to decline during 2012 and 2013, but have increased recently, while the other 3 areas have exhibited consistent and similar rates of increase (Fig. 6). Abundance estimates from 2016 were similarly precise from all 4 study areas with the mean Confidence Interval Coefficient of V ariati on (CICV) ranging from 0.14-0.15. Magnolia Habitat Treatments We completed 116 acres of pilot habitat treatments in January 2011 (Anderson and Bishop 2011; Environmental Assessment: DOI-BLM-CO-110-2011-004-EA), 54 acres of mechanical treatment method comparison treatments (hydro-ax, roller-chop, chain) in January 2012 (Stephens 2014), and 1,038 acres of hydro-ax treatments in April 2013 (Determination of NEPA Adequacy: DOI-BLM-CO-110-2012-0134- DNA), totaling 604 treated acres in each study area (Fig. 7). IO u �·,'. -....,_; Vegetation response in the pilot treatment sites was visually evident by fall 2011 (Fig. 7), and resulted in statistically significant (P < 0.05) increases in native grass and forb cover by the 2014 growing season. 2016 results are pending, but shrub responses appear promising from data collected this spring. Stephens (2014) reported that all 3 mechanical treatment methods compared resulted in roughly a 3 fold increase in grasses, forbs, and shrubs combined after 2 growing seasons (versus control sites), but cautioned that rollerchop treatments may be more vulnerable to invasive species response. Vegetative responses from 2013 hydro-ax treatments were visually evident following I growing season and shrub responses have been notable during the 4 th growing season, but statistical comparisons are still pending. As anticipated, grass and forb responses were evident 2 to 3 years post-treatment, with longer term response expected (3-5 years) for palatable shrubs. Of note, relatively high moisture conditions experienced during spring 2014 and 2015 resulted in higher than normal prevalence of cheatgrass (Bromus tectontm); cheatgrass invasion has previously been minor to non-existent. Cheatgrass invasion, however, does not appear directly related to treatment sites because occurrence is evident in both treatment and control areas. We anticipate this outbreak will subside based on past competitive advantage of native species to dominate, but will continue to monitor species composition and address cheatgrass persistence in treatment and control sites. GPS data addressing deer use of treatment sites is becoming available and will be analyzed as additional data are collected and vegetation responses progress. We observed improved fawn condition (P < 0.00 I) in South Magnolia following the 4 th growing season of habitat treatments when compared to fawn condition in the Ryan Gulch control area. Vegetation and mule deer responses will continue to be documented for the next 2 years to assess the utility of this mitigation approach in benefiting mule deer exposed to energy development disturbance. SUMMARY AND COLLABORATIONS The long-term goal of this study is to investigate habitat treatments and energy development practices that enhance mule deer populations exposed to extensive energy development activity. The information presented here summarizes mule deer population parameters from the 5-year pretreatment period and 4 years post-treatment. The pretreatment period was completed by spring 2013, providing baseline data for comparison with intended improvements in habitat conditions and response to varying degrees in human development activity. Winter range habitat improvements resulting in 604 acres of mechanically treated pinion-juniper/mountain shrub habitats in each of 2 study areas were completed by April of 2013, and subsequent vegetation responses have met or exceeded expectations. Post-treatment monitoring will continue for 2 additional years to provide sufficient time to measure how deer respond to these changes. Based on data collected through year 8 of this IO-year project: (1) annual adult female survival was consistent among areas averaging 7987% annually, but overwinter fawn survival was variable, ranging from 48% to 95% within study areas, with annual and study area differences primarily due to early winter fawn condition and annual weather conditions; (2) migratory mule deer selected for areas with increased cover and increased their rate of travel through developed areas, and avoided negative influences through behavioral shifts in timing and rate of migration, but did not avoid development structures; (3) mule deer body condition early and late winter was generally consistent within areas, with higher variability among study areas early winter, primarily due to December lactation rates, and late winter condition related to seasonal moisture and winter severity; (4) mule deer exhibited behavioral plasticity in relation to energy development, where disturbance distance varied relative to diurnal extent and magnitude of development activity, which may provide for several options in future development planning; and (5) late winter mule deer densities have increased in all study areas, ranging from 50% in North Ridge to 103% in North Magnolia. Detailed habitat use analyses are pending for the pre and post-habitat treatment period. We will continue to collect the various population and habitat use data across II �study sites to evaluate the effectiveness of habitat improvements on winter range. This approach will allow us to determine whether it is possible to effectively mitigate development impacts in highly developed areas, or whether it is better to allocate mitigation dollars toward less or non-impacted areas. In a recent project conducted on the Uncomphahgre Plateau, Colorado, Bergman et al. (2014) found that habitat treatments implemented in pinion-juniper habitat in undeveloped areas increased overwinter survival of fawns by a magnitude of 1.15. Hay field improvements have been completed in the North Magnolia study area by WPX Energy to fulfill a Wildlife Management Plan (WMP) agreement with CPW; elk (Cervus elaphus) response has been evident, but mule deer response has thus far been minor. A similar WMP agreement between ExxonMobil/XTO Energy and CPW a11owed completion and continued monitoring of mechanical habitat improvements in the Magnolia study areas. Collaborative research with agency biologists, graduate students, and university professors has produced 12 scientific publications addressing improved monitoring techniques for neonate mule deer captures (Bishop et al. 2011), mule deer migration (Lendrum et al. 2012, 2013, 2014; Anderson and Bishop 2014), improved approaches to address animal habitat use patterns (Northrup et al. 2013), mule deer response to helicopter capture and handling (Northrup et al. 2014a), potential effects of male-biased harvest on mule deer productivity (Freeman et al. 2014), mule deer genetics in relation to body condition and migration (Northrup et al. 2014b ), spatial and temporal factors influencing auditory vigilance in mule deer (Lynch et al. 2014), the relationship of plant phenology with mule deer body condition (Seral et al. 2015), and mule deer responses to differing energy development activities to inform future development planning (Northrup et al. 2015); these publications are summarized in Appendix A. Additional funding and cooperative agreements will be necessary to sustain this project to completion (preferably through 2018). We anticipate the opportunity to work cooperatively toward developing solutions for allowing the nation's energy reserves to be developed in a manner that benefits wildlife and the people who value both the wildlife and energy resources of Colorado. 12 �LITERATURE CITED Anderson, C.R., Jr. 2009. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and D. J. Freddy. 2008a. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Final Study Plan, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and D. J. Freddy. 2008b. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation-Stage I, Objective 5: Patte111s of mule deer distribution & movements. Pilot Study, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and C. J. Bishop. 2010. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and C. J. Bishop. 2011. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C. R., Jr., and C. J. Bishop. 2012. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Parks and Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and C. J. Bishop. 2014. Migration patterns of adult female mule deer in response to energy development. Pages 47-50 in Transactions of the 79 th North American Wildlife & Natural Resources Conference (R. A. Coon & M. C. Dunfee, eds.). Wildlife Management Institute, Gardners, PA, USA. ISSN 0078-1355. Bartmann, R. M. 1975. Piceance deer study-population density and structure. Job Progress Report, Colorado Division of Wildlife, Fort Collins, Colorado, USA. Bartmann, R. B., and S. F. Steinert. 1981. Distribution and movements of mule deer in the White River Drainage, Colorado. Special Report No. 51, Colorado Division of Wildlife, Fort Collins, Colorado, USA. Bartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monograph No. 121. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin I 0: 108-114. Bergman, E. J., C. J. Bishop, D. J. Freddy, G. C. White, and P F. Doherty, Jr. 2014. Habitat management influences over-winter survival of mule deer fawns in Colorado. Journal of Wildlife Management 78(3):448-455; DOI: 10.1002/jwmg.683 Burnham, K. P ., and D. R. Anderson. 2002. Model selection and multi-model inference: a practical information-theoretic approach. Second edition. Springer-Verlag, New York, New York, USA. Cook, R. C., J. G. Cook, D. L. Murray, P. Zager, B. K. Johnson, and M. W. Gratson. 2001. Development of predictive models of nutritional condition for rocky mountain elk. Journal of Wildlife Management 65:973-987. Cook, R. C., T. R. Stephenson, W. L. Meyers, J. G. Cook, and L.A. Shipley. 2007. Validating predictive models of nutritional condition for mule deer. Journal of Wildlife Management 71: 1934-1943. 13 �Cook, R. C., J. G. Cook, T. R. Stephenson, W. L. Meyers, S. M. McCorquodale, D. J. Vales, L. L. Irwin, P. Briggs Hall, R. D. Spencer, S. L. Murphie, K. A. Schoenecker, P. J. Miller. 2009. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife Management 74:880-896. Freeman, E. D., R. T. Larsen, M. E. Peterson, C. R. Anderson, Jr., K. R. Hersey, and B. R. McMillan. 2014. Effects of male-biased harvest on mule deer: implications for rates of pregnancy, synchrony, and timing of parturition. Wildlife Society Bulletin; DOI: 10.1002/wsb.450 Gibbs, H. D. 1978. Nutritional quality of mule deer foods, Piceance Basin, Colorado. Thesis, Colorado State University, Fort Collins, Colorado, USA. Kaplan, E. L., and P. Meier. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 52:457-481. Lendrum, P. E., C.R. Anderson, Jr., R. A. Long, J. K. Kie, and R. T. Bowyer. 2012. Habitat selection by mule deer during migration: effects of landscape structure and natural gas development. Ecosphere 3(9):82. http://dx.doi.org/10. Lendrum, P. E., C.R. Anderson, Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2013. Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes. PLoS ONE 8(5): e64548. doi:I0.1371/journal.pone.0064548 Lendrum, P. E., C.R. Anderson, Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2014. Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns of forage quality. Mammalian Biology: http://dx.doi.org/10.1016/j.mambio.2014.05.005 Lynch, E., J.M. Northrup, M. F. McKenna, C.R. Anderson Jr., L. Angeloni, and G. Wittemyer. 2014. Landscape and anthropogenic features influence the use of auditory vigilance by mule deer. Behavioral Ecology; doi:10.1093/beheco/arul58. McClintock, B. T., G. C. White, K. P. Burnham, and M.A. Pride. 2008. A generalized mixed effects model of abundance for mark-resight data when sampling is without replacement. Pages 271- 289 in D. L. Thompson, E.G. Cooch, and M. J. Conroy, editors, Modeling demographic processes is marked populations. Springer, New York, New York, USA. Northrup, J.M. 2015. Behavioral response of mule deer to natural gas development in the Piceance Basin. Dissertation, Colorado State University, Fort Collins, USA. Northrup, J.M., M. 8. Hooten, C.R. Anderson, Jr., and G. Wittemyer. 2013. Practical guidance on characterizing availability in resource selection functions under a use-availability design. Ecology 94(7): 1456-1463. Northrup, J.M., C.R. Anderson, Jr.. and G. Wittemyer. 2014a. Effects of helicopter capture and handling on movement behavior of mule deer. Journal of Wildlife Management 78(4):731738; DOI: I 0.1002/jwmg. 705 Northrup, J.M., A. 8. Shafer, C.R. Anderson Jr., D. W. Coltman, and G. Whittemyer. 2014b. Finescale genetic correlates to condition and migration in a wild cervid. Evolutionary Applications ISSN 1752-4571; doi: 10.1111/eva.12189 Northrup, J. M., C. R. Anderson, Jr., and G. Wittemyer. 2015. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer. Global Change Biology, doi: 10.1111/gcb.13037. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. C. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Sawyer, H., M. J. Kauffman, A. D. Middleton, T. A. Morrison, R. M. Nielson, and T. B. Wycoff. 2012. A framework for understanding semi-permeable barrier effects on migratory ungulates. Journal of Applied Ecology 50:68-78. Searle, K. R., M. B. Rice, C.R. Anderson, C. Bishop and N. T. Hobbs. 2015. Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore. Oecologia ISSN 0029- 8549; DOI I 0.1007/s00442-015-3348-9 Stephens, G. J. 2014. Understory responses to mechanical removal ofpinyon-juniper overstory. MS Thesis, Colorado State University, Ft. Collins USA. 14 ~ �~ Stephenson, T. R., V. C. Bleich, B. M. Pierce, and G. P. Mulcahy. 2002. Validation of mule deer body composition using in vivo and post-mortem indices of nutritional condition. Wildlife Society Bulletin 30:557-564. Stephenson, T. R., K. J. Hundertmark, C. C. Swartz, and V. Van Ballenberghe. 1998. Predicting body fat and mass in moose with untrasonography. Canadian Journal of Zoology 76:717-722. Unsworth, J. W., D. F. Pack, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. Van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-421 in L. Nielsen, J.C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked individuals. Bird Study 46: 120-139. White, G. C., and B. C. Lubow. 2002. Fitting population models to multiple sources of observed data. Journal of Wildlife Management 66:300-309. Prepared by_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ Charles R. Anderson, Jr., Mammals Research Leader ............ 15 �Table 1. Survival rate estimates (S) of fawn (1 Dec. 2015-15 June 2016) and adult female (1 July 201530 June 2016) mule deer from 4 winter range study areas of the Piceance Basin in northwest Colorado. Cohort Study area Initial sample size (n) March doe sample8 (n) S(95% CI) Fawns Ryan Gulch 60 0. 744 (0.632-0.856) South Magnolia 59 0.844 (0.751-0.938) North Magnolia 59 0.811 (0.711-0.912) North Ridge 59 0.492 (0.364-0.619) Adult females Ryan Gulch 30 56 0.837 (0.714-0.960) South Magnolia 33 53 0. 723 (0.580-0.867) North Magnolia 28 56 0.886 (0.778-0.994) North Ridge 23 50 0.752 (0.602-0.902) 8 Adult female sample sizes following capture and radio-collaring efforts March, 2016. 16 u �( ) () ( Table 2. Mean rump fat (mm), Body Condition Score (BCS8), and % body fat (% fat) of adult female mule deer from 4 study areas in the Piceance Basin of northwest Colorado, March and December, 2009-2016. Values in parentheses= SD. March 2009 March 2010 December 2009 %fat Rump fat Ryan Gulch 1.73 ( 1.78) 2.66 (0.55) 7.08 ( 1.27) 8.35 (6.36) 4.06 (1.13) 10.54 (3.72) 2.31 ( 1.44) 2.35 (0.48) 6.37 (1.41) South Magnolia 1.29 (0.47) 2.51 (0.66) 6.74 (2.27) 10.05 (6.19) 4.07 ( 1.21) 11.44 (3.50) 3.12 (2.20) 2.64 (0.59) 7.11 (1.69) North Magnolia 1.3 I ( 1.01) 2.66 (0.68) 7.15 (l.63) 10.67 (5.76) 4.25 (0.96) 11.94 (3.39) 3.15 (2.34) 2.85 (0.53) 7.54 (1.53) North Ridge 1.57 (1.22) 2.60 (0.56) 6.81 ( 1.68) 5.25 (5.65) 3.63 ( 1.11) 9.37 (3.08) 1. 77 ( 1.11) 2.42 (0.49) 6.39 (1.45) March 2011 December 2011 BCS Rump fat BCS %fat Rump fat BCS % fat Study Area Table 2. Continued. December 2010 %fat Rump fat BCS %fat Rump fat BCS %fat Ryan Gulch 7.26 (6.36) 3.24 (0.96) 9.69 (3.56) 1.55 (0.60) 2.53 (0.42) 6.72 (1.37) 13.41 (6.39) 4.21 (1.17) 13.17 (3.64) South Magnolia 9.85 (6.78) 3.30 (0.61) 11.27 (3.75) 1.65 (0.75) 2.35 (0.50) 6.15 ( I. 75) 8.18 (5.45) 3.41 (0.82) 10.34 (3.28) North Magnolia 9.55 (6.49) 3.46 ( 1.16) 10.79 (4.26) 1.65 (0.67) 2.53 (0.49) 6.79 (1.47) 8.76 (5.77) 3.74 (0.91) 10.73 (3.14) North Ridge 7.25 (5.41) 3.47 (0.86) 9.85 (3.02) 1.45 (0.76) 6.30 ( 1.65) 8.86 (5.37) 3.51 (0.99) 10.77 (3.33) 15 2.24 (0.49) Rump fat BCS Study Area �Table 2. Continued. December 2012 March 2012 March 2013 % fat Rump fat BCS % fat Rump fat 2.15 (1.44) 2.74 (0.44) 7.22 (1.16) 6.34 (4.35) 3.30 (0.77) 9.34 (2.43) 1.87 (0.90) South Magnolia 1.66 (0.77) 2.59 (0.36) 7.03 (1. 13) 8.30 (5.71) 3.46 ( 1.07) 10.32 (3.23) 2.06 (0.77) 2.65 (0.26) 7.19 (0.66) North Magno Iia 1.90 (0.76) 2.84 (0.34) 7.61 (0.96) 9.66 (6.41) 3.84 ( 1.16) 11.18 (3.64) 1.76 (0.91) 2.59 (0.41) 6.87 ( 1.11) North Ridge 2.24 (1.58) 2.70 (0.35) 7.26 (1.05) 5.76 (4.10) 3.32 (0.82) 1.87 (0.73) 2.48 (0.34) Study Area Rump fat Ryan Gulch BCS 9.06 (2.31) BCS % fat 2.65 (0.37) 7.14 (0.89) 6.70 ( 1.12) Table 2. Continued. March 2014 December 2013 BCS % fat Rump fat BCS December 2014 % fat Rump fat Study Area Rump fat Ryan Gulch 9.27 (6.29) 3.47 (0.87) 10.61 (3.76) 1.69 (0.85) 2.68 (0.39) 7.03 (0.99) 8.50 (6.76) 3.69 (1.03) 10.56 (3.70) South Magnolia 11.27 (8.40) 3.99 ( 1.04) 11.40 (4.16) 2.57 (l.61) 2.96 (0.30) 7.75 (0.68) 10.96 (6.82) 4.08 (1.06) 11.98 (3.81) North Magnolia 9.00 (6.15) 3.44 (0.78) 10.48 (3.25) 2.33 (2.12) 2.80 (0.49) 7.31 (1.43) 9.52 (5.83) 3.83 (1.04) 11.18 (3.32) North Ridge 11.17 (5.28) 3.85 (0.72) 11.66 (2.69) 2.38 ( 1.52) 2.68 (0.39) 7.16 (1.14) 7.93 (5.50) 3.74 (0.76) 10.20 (3.01) BCS %fat 16 C ( C �( ( ) ( Table 2. Continued. March 2015 December 2015 Rump fat BCS % fat 12.80 (6.83) 4.24 (0.88) 12.89 (3.72) 2.29 (0.64) 2.81 (0.37) 7.29 (0.52) 7.62 (0.74) 6.93 (4.83) 3.83 (0.89) 9.83 (2.69) 2.07 (1.39) 2.78 (0.39) 7.46 (0.93) 2.90 (0.42) 7.49 (0.90) 8.79 (6.01) 3.79 (0.85) 10.81 (3.54) 2.43(1.01) 2.71 (0.39) 7.17 (0.87) 2.92 (0.46) 7.43 (1.05) 5.47 (5.49) 3.25 (0.66) 1.58 (0.70) 2.51 (0.41) 6.73 (1.26) Study Area Rump fat BCS % fat Ryan Gulch 2.62 (0.95) 2.89 (0.40) 7.44 (0.53) South Magnolia 2.66 (1.36) 2.97 (0.55) North Magnolia 2.25 (0.97) North Ridge 2.28 (1.37) 3 March 2016 Rump fat BCS 8ody condition score taken from palpations of the rump following Cook et al. (2009). 17 % fat 9.35 (2.75) �Table 3. Mark-resight abundance (N) and density estimates of mule deer from 4 winter range herd segments in the Piceance Basin, northwest Colorado, 21 March-2 April 2016. Data represent 4 helicopter resight surveys from 3 study areas and 5 resight surveys from South Magnolia. Study area Mean No. sighted Mean No. marked N(95% en Density (deer/kni2) Ryan Gulch 441 32 1,754 (1,534-2,025) 12.5 South Magnolia 199 23 888 (780-I,027) 10.7 North Magnolia 310 29 1,045 (910-1,226) 13.2 North Ridge 489 26 1,381 ( 1,204-1,609) 26.0 18 �Well P1ad.s 6 Faci lities Sovci \I ilgnolla ~yan GLHh 2, - O t:\••d O JJffl t::N f .!l, 10 •,fll~~ Figure I. Mule deer winter range study areas relative to active natural gas well pads and energy development facilities in the Piceance Basin of northwest Colorado, winter 20 I 3/14 (Accessed http://cogcc.state.eo.us/ Dec. 3 1, 2013). Development activity has subsided with no additional drilling since 2013. 19 w,e~ �Winter fawn survival 2010-11 - 2015-16 1.00 frl 0.90 0.80 -- 0.60 - 0.50 - 0.40 0.30 0.20 0.10 0.00 - --- - ~-- 1--- 0.70 - 1--- - - - - - - 2011-12 2012-13 ,- - 1--- ,- ,- - - D Ryan Gulch - D South Magnolia - D North Ridge ■ North Magnolia - --~ 2010-11 1--- ,- - - ~ - - -- 2013-14 2014-15 2015-16 Figure 2. Over-winter (Dec-June) mule deer fawn survival (5) from 4 study areas in the Piceance Basin, northwest Colorado. Error bars= 95% Cl. 20 �Male fawn weights 42.0 40.0 iii 38.0 =-fa 36.0 0Ryan Gulch DSouth Magnolia 'cii 3: 34.0 ~ ~ II North Magnolia ' ', l d 32.0 30.0 ii j l H D North Ridge p Dec Dec Dec Dec Dec Dec Dec Dec 2008 2009 2010 2011 2012 2013 2014 2015 Female fawn weights 42.0 40.0 -;:a 38.0 = fD URyan Gulch 36.0 DSouth Magnolia ~ 34.0 ■ North Magnolia aNorth Ridge 32.0 30.0 Dec Dec Dec Dec Dec Dec Dec Dec 2008 2009 2010 2011 2012 2013 2014 2015 Figure 3. Mean male and female fawn weights and 95% CI ( error bars} from 4 mule deer study areas in the Piceance Basin, northwest Colorado, December 2008-2015. 21 �Colorado Figure 4. Mule deer study areas in the Piceance Basin of northwestern Colorado, USA (Top), spring 2009 migration routes of adult female mule deer (n = 52; Lower left), and active natural-gas well pads (black dots) and roads (state, county, and natural-gas; white lines) from May 2009 (Lower right; from Lendrum et al. 2012). 22 �Early winter rump fat 16 14 _ 12 E .§. 10 -North Ridge J! -North Magnolia .. 8 ~ E 6 a: 4 ::::s ----~ Ryan Gulch -south Magnolia 2 0 Dec Dec Dec Dec Dec Dec Dec 2009 2010 2011 2012 2013 2014 2015 Late winter rump fat 4.00 - , - - - - - - - - - - - - - - - - - - - - - - - - - - 3.50 - t - - - - - - - - - - - - - - - - - - - - - - - - 3.00 - t - - - - - - - - - - - - - - - - - - - - - - - - - - t--7:;t-~--~~-=-:;~~;;z~~~1.50 t=l~~;~t;~~i~~;;E~~:!~~~~t~~=~~~~St= 2.50 2 00 • 1.00 - - - - - - - - - - - - - - - - ~ - - - - - - - - - - -North Ridge -North Magnolia -Ryan Gulch -south Magnolia 0.50 - - - - - - - - - - - - - - - - - - - - - - 0.00 Mar Mar Mar Mar Mar Mar Mar Mar 2009 2010 2011 2012 2013 2014 2015 2016 Figure 5. Mean early {early Dec., Top) and late winter (early Mar., Bottom) body condition (mm rump fat) of adult female mule deer from 4 winter range study areas in the Piceance Basin of northwest Colorado, March 2009-March 2016. Error bars = 95% CI. ........._ 23 �Piceance Basin late winter mule deer density 35.00 30.00 25.00 N j 20.00 - ~ t 15.00 - North Ridge • • • • • • Ryan Gulch Q - 10.00 • North Magnolia - s o u t h Magnolia 5.00 0.00 2009 2010 2011 2012 2013 2014 2015 2016 Year Figure 6. Mule deer density estimates and 95% CI (error bars) from 4 winter range herd segments in the Piceance Basin, northwest Colorado, late winter 2009-2016. Estimates for North Ridge 2014 and 2015 and for North Magnolia 2015 were adjusted upward (using GPS migration data) to account for early migration from winter range prior to and during surveys. V 24 �M :iilnrl• 1 t -~.Jt?n".,, :""t ':I~('$ t[•67 ,:'\(T'f." >I t:t:.:r::~t_ 1 : _ Ji;L_ .inJ 1., - -: : ~•. t- l'\1JI f17:,-t_f1_y1 ~ :;r:: .t::..1: •"11,;U': r"l_~ Jrl_g-1 :i - ~'!C'\E·r;1gt.:.. a_ J f"trll [ 4 1: Mule Deer Study Areas Figure 7. Habitat treatment site delineations in 2 mule deer study areas (604 acres each) of the Piceance Basin, northwest Colorado (Top; cyan polygons completed Jan. 2011 using hydro-axe; yellow polygons completed Jan. 20 I 2 using hydro-axe, roller-chop, and chaining; and remaining polygons completed April 20 I 3 using hydro-axe). January 20 11 hydro-axe treatment-site photos from North Hatch Gulch during April (Lower left, aerial view) and October, 2011 (Lower right, ground view). 25 �Appendix A. Abstracts of published manuscripts resulting from Piceance Basin mule deer/energy development interaction research collaborations. Abstract format specific to the respective journal requirements. U Effectiveness of a redesigned vaginal implant transmitter in mule deer CHAD J. BISHOP 1, CHARLES R. ANDERSON Jr. 1, DAi"llEL P. WALSH 1, ERIC J. BERGMAN 1, PETER KUECHLE 2, and JOHN ROTH 2 'Colorado Parks and Wildlife, Fort Collins. Colorado 80526 USA 2 Advanced Telemetry Systems. Isanti, Minnesota 55040 USA Citation: Bishop, C. J., C.R. Anderson Jr., D. P. Walsh. E. J. Bergman, P. Kuechle, and J. Roth. 2011. Effectiveness ofa redesigned vaginal implant transmitter in mule deer. Journal of Wildlife Management 75(8):1797-1806; DOI: 10.1002/jwmg.229 ABSTRACT Our understanding of factors that limit mule deer (Odocoileus hemionus) populations may be improved by evaluating neonatal survival as a function of dam characteristics under free-ranging conditions, which generally requires that both neonates and dams are radiocollared. The most viable technique facilitating capture of neonates from radiocollared adult females is use of vaginal implant transmitters (VITs). To date, VITs have allowed research opportunities that were not previously possible; however, VITs are often expeJled from adult females prepartum, which limits their effectiveness. We redesigned an existing VIT manufactured by Advanced Telemetry Systems (ATS; Isanti, MN) by lengthening and widening wings used to retain the VIT in an adult female. Our objective was to increase VIT retention rates and thereby increase the likelihood of locating birth sites and newborn fawns. We placed the newly designed VITs in 59 adult female mule deer and evaluated the probability of retention to parturition and the probability of detecting newborn fawns. We also developed an equation for determining VIT sample size necessary to achieve a specified sample size of neonates. The probability of a VIT being retained until parturition was 0. 766 (SE =0.0605) and the probability of a VIT being retained to within 3 days of parturition was 0.894 (SE = 0.0441 ). In a similar study using the original VIT wings (Bishop et al. 2007), the probability of a VIT being retained until parturition was 0.447 (SE= 0.0468) and the probability of retention to within 3 days of parturition was 0.623 (SE= 0.0456). Thus, our design modification increased VIT retention to parturition by 0.319 (SE= 0.0765) and VIT retention to within 3 days of parturition by 0.271 (SE= 0.0634). Considering dams that retained VITs to within 3 days of parturition, the probability of detecting at least I neonate was 0.952 (SE = 0.0334) and the probability of detecting both fawns from twin litters was 0.588 (SE = 0.0827). We expended approximately 12 person-hours per detected neonate. As a guide for researchers plaMing future studies, we found that VIT sample size should approximately equal the targeted neonate sample size. Our study expands opportunities for conducting research that links adult female attributes to productivity and offspring survival in mule deer.© 2014 The Wildlife Society. Habitat selection by mule deer during migration: effects of landscape structure and natural-gas development PATRICKE. LENDRUM 1• CHARLES R. ANDERSON JR. 2, RYAi"l A. LONG1.JOHN G. KIE 1, A."lD R. TERRY BOWYER1 1 Department of Biological Sciences, Idaho State University, Pocatello, Idaho 83209 USA :?Colorado Parks and Wildlife, Grand Junction, Colorado 81505 USA Citation: Lendrum, P. E., C.R. Anderson Jr., R. A. Long, J. G. Kie, and R. T. Bowyer. 2012. Habitat selection by mule deer during migration: effects of landscape structure and natural-gas development. Ecosphere 3(9):82 http://dx.doi.org/l 0.1890/ES 12-00165. l Abstract. The disruption of traditional migratory routes by anthropogenic disturbances has shifted patterns of resource selection by many species, and in some instances has caused populations to decline. Moreover, in recent decades populations of mule deer (Odocoi/eus hemionus) have declined throughout much of their historic range in the western United States. We used resourceselection functions to determine if the presence of natural-gas development altered patterns ofresource selection by migrating mule deer. We compared spring migration routes of adult female mule deer fitted with GPS collars (n = 167) among four study areas that had varying degrees ofnatural-gas development from 2008 to 2010 in the Piceance Basin of northwest Colorado, USA. Mule deer migrating through the most developed area had longer step lengths (straight-line distance between successive GPS locations) compared with deer in less developed areas. Additionally, deer migrating through the most developed study areas tended to select for habitat types that provided greater amounts of concealment cover, whereas deer from the least developed areas tended to select habitats that increased access to forage and cover. Deer selected habitats closer to well pads and avoided roads in all instances except along the most highly developed migratory routes, where road densities may have been too high for deer to avoid roads without deviating substantially from established migration routes. These results indicate that behavioral tendencies toward avoidance of anthropogenic disturbance can be overridden during migration by the strong fidelity ungulates demonstrate towards migration routes. If avoidance is feasible, then deer may select areas further from development, whereas in highly developed areas. deer may simply increase their rate of travel along established migration routes. 26 ·v �Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes Patrick E. Lendrum 1, Charles R. Anderson Jr.2, Kevin L. Monteith 1"', Jonathan A. Jenks'. R. Terry Bowyer' 1 Department of Biological Sciences, Idaho State University, Pocatello, Idaho, USA, 2 Colorado Division of Parks and Wildlife, Grand Junction, Colorado, USA, 3 Wyoming Cooperative Fish and Wildlife Research Unit, University of Wyoming, Laramie, Wyoming, USA,4 Department of Natural Resource Management, South Dakota State University, Brookings. South Dakota, USA Citation: Lendrum, P. E., C.R. Anderson Jr., K. L. Monteith. J. A. Jenks. R. T. Bowyer. 2013. Migrating Mule Deer: Effects of anthropogcnically Altered Landscapes. PloS ONE 8(5): c64548. DOI: 1O. l 371/joumal.pone.0064548 Abstract Background: Migration is an adaptive strategy that enables animals to enhance resource availability and reduce risk of predation at a broad geographic scale. Ungulate migrations generally occur along traditional routes, many of which have been disrupted by anthropogenic disturbances. Spring migration in ungulates is of particular importance for conservation planning, because it is closely coupled with timing of parturition. The degree to which oil and gas development affects migratory patterns, and whether ungulate migration is sufficiently plastic to compensate for such changes, warrants additional study to better understand this critical conservation issue. Methodolog)'IPrincipal Findi11gs: We studied timing and synchrony of departure from winter range and arrival to summer range offemale mule deer (Odocoileus hemionus) in northwestern Colorado, USA, which has one of the largest natural-gas reserves currently under development in North America. We hypothesized that in addition to local weather, plant phenology, and individual life-history characteristics, patterns of spring migration would be modified by disturbances associated with natural-gas extraction. We captured 205 adult female mule deer, equipped them with GPS collars, and observed patterns of spring migration during 2008-20 l 0. Condusions/Signijicance: Timing of spring migration was related to winter weather (particularly snow depth) and access to emerging vegetation, which varied among years, but was highly synchronous across study areas within years. Additionally, timing of migration was influenced by the collective effects of anthropogenic disturbance, rate of travel, distance traveled, and body condition of adult females. Rates of travel were more rapid over shorter migration distances in areas of high natural-gas development resulting in the delayed departure, but early arrival for females migrating in areas with high development compared with less-developed areas. Such shifts in behavior could have consequences for timing of arrival on birthing areas, especially where mule deer migrate over longer distances or for greater durations. \.,,_I Practical guidance on characterizing availability in resource selection functions under a use-availability design JOSEPH M. NORTHRUP 1, !\ilEVIN B. HOOTEN 1.l.3, CHARLES R. ANDERSON JR.'. AND GEORGE WITTEMYER1 'Department of Fish, Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 2 U.S. Geological Survey, Colorado Cooperative Fish and Wildlife Research Unit, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 3 Colorado State University, Department of Statistics, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 4 Mammals Research Section Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction, Colorado 81505 USA Citation: Northrup, J.M., M. B. Hooten, C.R. Anderson Jr., and G. Wittemyer. 2013. Practical guidance on characterizing availability in resource selection functions under a use-availability design. Ecology 94(7):1456-1463. http://dx.doi.org/10.1890/12-1688.I Abstract. Habitat selection is a fundamental aspect of animal ecology, the understanding of which is critical to management and conservation. Global positioning system data from animals allow fine-scale assessments of habitat selection and typically are analyzed in a use-availability framework, whereby animal locations are contrasted with random locations (the availability sample). Although most use-availability methods are in fact spatial point process models, they often are fit using logistic regression. This framework offers numerous methodological challenges, for which the literature provides little guidance. Specifically, the size and spatial extent of the availability sample influences coefficient estimates potentially causing inteipretational bias. We examined the influence of availability on statistical inference through simulations and analysis of serially correlated mule deer GPS data. Bias in estimates arose from incorrectly assessing and sampling the spatial extent of availability. Spatial autocorrelation in covariates, which is common for landscape characteristics, exacerbated the error in availability sampling leading to increased bias. These results have strong implications for habitat selection analyses using GPS data, which are increasingly prevalent in the literature. We recommend that researchers assess the sensitivity of their results to their availability sample and, where bias is likely, take care with interpretations and use cross validation to assess robustness . ..-....., 27 �Effects of Helicopter Capture and Handling on Movement Behavior of Mule Deer JOSEPH M. NORTHRUP', CHARLES R. ANDERSON JR2, AND GEORGE WITIEMYER1 1 Dcpanment of Fish, Wildlife, and Conservation Biology, Colorado State University. 1474 Campus Delivery, Fort Collins, Colorado 80523 USA Mammals Research Section Colorado Parks and Wildlife. 711 Independent Avenue, Grand Junction, Colorado 81505 USA 2 Citation: Nonhrup, J.M .. C.R. Anderson Jr., and G. Wittcmyer. 2014. Effects of helicopter capture and handling on movement behavior of mule deer. Journal of Wildlife Management 78(4):731-738; DOI: 10.1002/jwmg.705 ABSTRACT Research on wildlife movement, physiology, and reproductive biology often requires capture and handling of animals. Such invasive treatment can alter behavior, which may bias results or invalidate assumptions regarding representative behaviors. To assess the impacts of handling on mule deer (Odocoileus hemionus), a focal species for research in North America, we investigated pre- and post-recapture movements of collared individuals, and compared them to deer that were not recaptured (controls). We compared pre- and post-recapture movement rates (mihr) and 24-hour straight-line displacement among recaptured and control deer. In addition, we examined the time it took recaptured deer to return to their pre-recapture home range. Both daily straight-line displacement and movement rate were marginally elevated relative to monthly averages for 24 hours following recapture, with non-significant elevation continuing for up to 7 days. Comparing movements averaged over 30 days before and after recapture, we found no differences in displacement, but movement rates demonstrated seasonal effects, with faster movements post- relative to pre-recapture in March and slower movements post- relative to pre-recapture in December. Relative to control deer movements, recaptured deer movement rates in March were higher immediately after recapture and lower in the second and third weeks following recapture. The median time to return to the pre-recapture home range was 13 hours, with 71 % of deer returning in the first day, and 91 % returning within 4 days. These results indicate a short period of elevated movements following recaptures, likely due to the deer returning to their home ranges, followed by weaker hut non-significant depression of movements for up to 3 weeks. Censoring of the first day of data post capture from analyses is strongly supported, and removing additional days until the individual returns to its home range will control for the majority of impacts from capture. © 2014 The Wildlife Society. Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns of forage quality Patrick E. Lendrum•. Charles R. Anderson Jr.\ Kevin L. Monteith\ Jonathan A. Jenksd, R. Terry Bowyer• ~ Department ofBiological Sciences, Idaho State University, 921 South 8th Avenue, Stop 8007, Pocatello 83209, USA h Mammals Research Section Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction 81505, USA c Wyoming Cooperative Fish and Wildlife Research Unit, Department of Zoology and Physiology, University of Wyoming. 3166. 1000 East University Avenue, Laramie 82071, USA d Department of Natural Resource Management, South Dakota State University, Box 2140B, Brookings 57007, USA Citation: Lendrum, P. E., C.R. Andenmn Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2014. Relating the movement ofa rapidly migrating ungulate to spatiotemporal patterns of forage quality. Mammalian Biology: http:·'/dx.doi.org/10. l 0 16/j.mambio.20 I4.05.005 ABSTRACT: Migratory ungulates exhibit recurring movements, often along traditional routes between seasonal ranges each spring and autumn, which allow them to track resources as they become available on the landscape. We examined the relationship between spring migration of mule deer (Odocoi/eus hemionus) and forage quality, as indexed by spatiotemporal patterns of fecal nitrogen and remotely sensed greenness of vegetation (Normalized Difference Vegetation Index; NDVI) in spring 2010 in the Piceance Basin of northwestern Colorado, USA. NDVI increased throughout spring, and was affected primarily by snow depth when snow was present, and temperature when snow was absent. Fecal nitrogen was lowest when deer were on winter range before migration, increased rapidly to an asymptote during migration, and remained relatively high when deer reached summer range. Values of fecal nitrogen corresponded with increasing NOVI during migration. Spring migration for mule deer provided a way for these large mammals to increase access to a high-quality diet, which was evident in patterns of NOVI and fecal nitrogen. Moreover. these deer "jumped" rather than "surfed" the green wave by arriving on summer range well before peak productivity of forage occurred. This rapid migration may aid in securing resources and seclusion from others on summer range in preparation for parturition. and to minimize detrimental factors such as predation. and malnutrition during migration. 28 �Effects of Male-Biased Harvest on Mule Deer: Implications for Rates of Pregnancy, Synchrony, and Timing of Parturition ERIC D. FREEMAN 1, RANDY T. LARSEN 1, MARK E. PETERSON2, CHARLES R. ANDERSON JR.3, KENT R. HERSEY', AND BROCK R. McMILLAN 1 1 Department of Plant and Wildlife Sciences, Brigham Young University, 275 WIDB, Provo, UT 84602, USA Department of Fish, Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fon Collins, CO 80523, USA Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction, CO 81505, USA 4 Utah Division of Wildlife Resources, 1594 W Nonh Temple, Salt Lake City, UT 84114, USA 2 3 Citation: Freeman, E. D., R. T. Larsen. M. E. Peterson. C.R. Anderson Jr.. K. R. Hersey, and B. R. McMillan. 2014. Effects of male-biased harvest on mule deer: implications for rates of pregnancy, synchrony, and timing of parturition. Wildlife Society Bulletin; DOI: 10.1002/wsb.450 ABSTRACT Evaluating how management practices influence the population dynamics of ungulates may enhance future management of these species. For example, in mule deer (Odocoileus hemionus), changes in male/female ratio due to malebiased harvest may alter rates of pregnancy, timing of parturition, and synchrony of parturition if inadequate numbers of males are present to fertilize females during their first estrous cycle. If rates of pregnancy or parturition are influenced by decreased male/female ratios, recruitment may be reduced ( e.g., fewer births, later parturition resulting in lower survival of fawns, and a less synchronous parturition that potentially increases susceptibility of neonates to predation). Our objectives were to compare rates of pregnancy, synchrony of parturition, and timing of parturition between exploited mule deer populations with a relatively high (Piceance, CO, USA; 26 males/100 females) and a relatively low (Monroe, UT, USA; 14 males/100 females) male/female ratio. We determined rates of pregnancy via ultrasonography and timing of parturition via vaginal implant transmitters. We found no differences in rates of pregnancy (98.6% and 96.6%; z = 0.821; P = 0.794), timing of parturition (estimate= 1.258; SE= l.672; t = 0.752; P = 0.454), or synchrony of parturition (F = l.073; P = 0.859) between Monroe Mountain and Piceance Basin, respectively. The relatively low male/female ratio on Monroe Mountain was not associated with a protracted period of parturition. This finding suggests that relatively low male/female ratios typical of heavily harvested populations do not influence population dynamics because recruitment remains unaffected. @ 2014 The Wildlife Society. Fine-scale genetic correlates to condition and migration in a wild cervid Joseph M. Nortbrup, 1 Aaron B. A. Shafer,2 Charles R. Anderson Jr,3 David W. Coltman" and George Wittemyer 1 I Dcpanment of Fish, Wildlife, and Conservation Biology, Colorado State University, Fort Collins, CO, USA 2 Dcpanment of Evolutionary Biology, Evolutionary Biology Centre. Uppsala University, Uppsala, Sweden 3 Mammals Research Section, Colorado Parks and Wildlife, Grand Junction, CO, USA 4 Department of Biological Sciences, University of Alhena, Edmonton, AB, Canada. Citation: Northrup, J.M., AB. Shafer, C.R. Anderson Jr., D. W. Coltman, and G. Whittemyer. 2014. Fine-scale genetic correlates to condition and migration in a wild cervid. Evolutionary Applications ISSN 1752-4571; doi: 10.1111/eva.12189 Abstract The relationship between genetic variation and phenotypic traits is fundamental to the study and management of natural populations. Such relationships often are investigated by assessing correlations between phenotypic traits and heterozygosity or genetic differentiation. Using an extensive data set compiled from free ranging mule deer ( Odocoilcus hemionus), we combined genetic and ecological data to (i) examine correlations between genetic differentiation and migration timing, (ii) screen for mitochondrial haplotypes associated with migration timing, and (iii) test whether nuclear heterozygosity was associated with condition. Migration was related to genetic differentiation (more closely related individuals migrated closer in time) and mitochondrial haplogroup. Body fat was related to heterozygosity at two nuclear loci (with antagonistic patterns), one of which is situated near a known fat metabolism gene in mammals. Despite being focused on a widespread panmictic species, these findings revealed a link between genetic variation and important phenotypes at a fine scale. We hypothesize that these correlations are either the result of mixing refugial lineages or differential mitochondrial haplotypes influencing energetics. The maintenance of phenotypic diversity will be critical to enable the potential tracking of changing climatic conditions, and these correlates highlight the need to consider evolutionary mechanisms in management, even in widely distributed panmictic species. 29 �Landscape and anthropogenic features influence the use of auditory vigilance by mule deer Emma Lynch,■ Joseph M. Northrup,b Megan F. McKenna,C Charles R. Anderson Jr,d Lisa Angeloni,... and George Wittemyera.b aGraduatc Degree Program in Ecology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA bDcpartmcnt offish. Wildlife and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA ~atural Sounds and Night Skies Division, National Park Service, 1201 Oakridge Drive, Fort Collins, CO 80525, USA, dMammals Research Section, Colorado Parks and Wildlife. 317 W. Prospect Road, Fort Collins, CO 80526. USA COepartmcnt of Biology, Colorado State University. 1878 Campus Delivery, Fort Collins, CO 80523, USA Citation: Lynch, E.. J.M. Northrup, M. F. McKenna. C.R. Anderson Jr., L. Angeloni, and G. Wittemyer. 2014. Landscape and anthropogenic features influence the use of auditory vigilance by mule deer. Behavioral Ecology: doi: 10.1093/beheco/aru 158. While visual forms of vigilance behavior and their relationship with predation risk have been broadly examined, animals also employ other vigilance modalities such as auditory vigilance by listening for the acoustic cues of predators. Similar to the tradeoffs associated with visual vigilance, auditory behavior potentially structures the energy budgets and behavior of animals. The cryptic nature of auditory vigilance makes it difficult to study, but on-animal acoustical monitoring has rapidly advanced our ability to investigate behaviors and conditions related to sound. We utilized this technique to investigate the ways external stimuli in an active natural gas development field affect periodic pausing by mule deer (Odocoileus hemionus) within bouts of rumination-based mastication. To better understand the ecological properties that structure this behavior, we investigate spatial and temporal factors related to these pauses to determine if results are consistent with our hypothesis that pausing is used for auditory vigilance. We found that deer paused more when in forested cover and at night, where visual vigilance was likely to be less effective. Additionally, deer paused more in areas of moderate background sound levels, though responses to anthropogenic features were less clear. Our results suggest that pauses during rumination represent a form of auditory vigilance that is responsive to landscape variables. Further exploration of this behavior can facilitate a more holistic understanding of risk perception and the costs associated with vigilance behavior. Migration Patterns of Adult Female Mule Deer in Response to Energy Development Charles R. Anderson Jr. and Chad J. Bishop Mammals Research Section. Colorado Parks and Wildlife. 317 W. Prospect Road, Fort Collins, CO 80526. USA Citation: Anderson. C.R., Jr., and C. J. Bishop. 2014. Migration patterns of adult female mule deer in response to energy development. Pages 47-50 in Transactions of the 79lh North American Wildlife & Natural Resources Conference (R. A. Coon & M. C. Dunfee. eds.). Wildlife Management Institute, Gardners, PA, USA. ISSN 0078-1355. Migration is an adaptive strategy that enables animals to enhance resource availability and reduce risk of predation at a broad geographic scale. Ungulate migrations generally occur along traditional routes, many of which have been disrupted by anthropogenic disturbances. Spring migration in ungulates is of particular imponance for conservation planning because it is closely coupled with timing of parturition. The degree to which oil and gas development affects migratory patterns, and whether ungulate migration is sufficiently prepared to compensate for such changes, has recently been investigated in Colorado and Wyoming (Lendrum et al. 2012, 2013; Sawyer et al. 2012). Lendrum et al. (2012, 2013) and Sawyer et al. (2012) address mule de~r (Odocoileus hemionus) migration patterns in relation to energy development from northwest Colorado and south-central Wyoming, respectively. We address results from the Colorado and Wyoming studies and then compare similarities and differences. The interactions between migratory mule deer and energy development identified by Lendrum et al. (2012, 2013) and Sawyer et al. (2012) suggest mule deer may benefit from energy development planning by considering thresholds of development that may alter migratory behavior. It appears that migration rate, migration routes, and stopover use, if present, may be altered at high development intensities. In addition, migratory mule deer may benefit by maintaining security cover along migration paths, and improved habitat conditions may facilitate more direct and rapid migration requiring less energy to complete migration. Enhancing permeability along migration routes by applying dispersed development plans (<2 well pads/km2) and minimizing disturbance to vegetation types by maintaining security cover should reduce impacts to migratory mule deer as well as other migratory ungulates. Where feasible. habitat improvement projects on winter range and possibly stopover sites would also enhance migratory mule deer populations by enhancing energy reserves for long-distance movements and parturition shortly after summer range arrival. Where possible, directional drilling could be used to extract energy resources from underneath migration routes while maintaining no surface occupancy. Lastly, we emphasize that GPS studies now allow managers to accurately map migration routes for entire populations and identify relatively narrow corridors that are most heavily used thus allowing for the identification of the most important corridors for migrating ungulates. Where available. we encourage agencies to incorporate such migration corridors into land-use plans (e.g .. resource management plans) and National Environmental Policy Act documents. 30 �Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore Kate R. Searle1 • Mindy 8. Rice 1 • Charles R. Anderson1 • Chad Bishop 1 • N. T. Hobbs3 1 NERC Centre for Ecology and Hydrology, Bush Estate, Pcnicuik EH26 0QB, UK 2 Colorado Parks and Wildlife, 317 W. Prospect Road, Fon Collins, CO 80526, USA 3 Department of Ecosystem Science and Sustainability, Colorado State University. Fon Collins 80524, CO, USA Citation: Searle, K. R., M. B. Rice, C.R. Anderson, C. Bishop and N. T. Hobbs. 2015. Asynchronous vegetation phcnology enhances winter body condition of a large mobile herbivore. Occologia ISSN 00:?9-8549; DOI 10.1007ls00442-0 l 5-3348-9 Abstract Understanding how spatial and temporal heterogeneity influence ecological processes forms a central challenge in ecology. Individual responses to heterogeneity shape population dynamics, therefore understanding these responses is central to sustainable population management. Emerging evidence has shown that herbivores track heterogeneity in nutritional quality of vegetation by responding to phenological differences in plants. We quantified the benefits mule deer (Odocoileus /Jemionus) accrue from accessing habitats with asynchronous plant phenology in northwest Colorado over 3 years. Our analysis examined both the direct physiological and indirect environmental effects of weather and vegetation phenology on mule deer winter body condition. We identified several important effects of annual weather patterns and topographical variables on vegetation phenology in the home ranges of mule deer. Crucially, temporal patterns of vegetation phenology were linked with differences in body condition, with deer tending to show poorer body condition in areas with less asynchronous vegetation green-up and later vegetation onset. The direct physiological effect of previous winter precipitation on mule deer body condition was much less important than the indirect effect mediated by vegetation phenology. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer JOSEPH M. NORTHRUP', CHARLES R. ANDERSON JR .2 and GEORGE WITTEMYER 1• 3 --...., '-.,,/ 'Department of Fish, Wildlife and Conservation Biology, Colorado State University. Fort Collins, CO, USA ?Mammals Research Section. Colorado Parks and Wildlife, Fort Collins, CO. USA 3 Graduate Degree Program in Ecology, Colorado State University, Fort Collins, CO, USA Citation: Northrup, J.M., C.R. Anderson. Jr., and G. Winemyer. 2015. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer. Global Change Biology, doi: IO. l l I llgcb.13037 Abstract Extraction of oil and natural gas (hydrocarbons) from shale is increasing rapidly in North America, with documented impacts to native species and ecosystems. With shale oil and gas resources on nearly every continent, this development is set to become a major driver of global land-use change. It is increasingly critical to quantify spatial habitat loss driven by this development to implement effective mitigation strategies and develop habitat offsets. Habitat selection is a fundamental ecological process, influencing both individual fitness and population-level distribution on the landscape. Examinations of habitat selection provide a natural means for understanding spatial impacts. We examined the impact of natural gas development on habitat selection patterns of mule deer on their winter range in Colorado. We fit resource selection functions in a Bayesian hierarchical framework, with habitat availability defined using a movement-based modeling approach. Energy development drove considerable alterations to deer habitat selection patterns, with the most substantial impacts manifested as avoidance of well pads with active drilling to a distance of at least 800 m. Deer displayed more nuanced responses to other infrastructure, avoiding pads with active production and roads to a greater degree during the day than night. In aggregate, these responses equate to alteration of behavior by human development in over SO% of the critical winter range in our study area during the day and over 25% at night. Compared to other regions, the topographic and vegetative diversity in the study area appear to provide refugia that allow deer to behaviorally mediate some of the impacts of development. This study, and the methods we employed, provides a template for quantifying spatial take by industrial activities in natural areas and the results offer guidance for policy makers. mangers, and industry when attempting to mitigate habitat loss due to energy development. 31 �Colorado Parks and Wildlife July 1, 2016 June 30, 2017 WILDLIFE RESEARCH REPORT : Parks and Wildlife : Mammals Research : Deer Conservation : Population Performance of Piceance Basin Mule Deer in Response to Natural Gas Resource Extraction and Mitigation Efforts to Address Human Activity and Habitat Degradation State of Cost Center Work Package Task No. Colorado 3430 3001 6 Federal Aid Project: W-243-R2 : Period Covered: July 1, 2016 June 30, 2017 Author: C. R. Anderson, Jr. Personnel: E. Bergman, E. Cardenas, D. Collins, B. deVergie, D. Finley, M. Fisher, L. Gepfert, J. Hudson, D. Johnston, T. Knowles, D. Lewis, T. Mullins, J. Pelham, B. Petch, J. Rivale, R. Schilowsky, G. Smith, R. Velarde, T. Verzuh, S. Williams, L. Wolfe, CPW; E. Hollowed, L. Belmonte, BLM; T. Graham, Ranch Advisory Partners; P. Doherty, J. Northrup, M. Peterson, G. Wittemyer, K. Wilson, Colorado State University; R. Swisher, S. Swisher, Quicksilver Air, Inc.; B. Mallow, Kiwi Air, Inc.; L. Coulter, Coulter Aviation. Project support received from Federal Aid in Wildlife Restoration, Colorado Mule Deer Association, Colorado Mule Deer Foundation, Muley Fanatic Foundation, Colorado State Severance Tax Fund, EnCana Corp., ExxonMobil Production Co./XTO Energy, Marathon Oil Corp., Shell Petroleum, and WPX Energy. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the authors. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We propose to experimentally evaluate winter range habitat treatments and human-activity management alternatives intended to enhance mule deer (Odocoileus hemionus) populations exposed to energy-development activities. The Piceance Basin of northwestern Colorado was selected as the project area due to ongoing natural gas development in one of the most extensive and important mule deer winter and transition range areas in Colorado. The data presented here represent the first 9 years of data (5 years of pretreatment, 4 years post treatment) of a long-term study addressing habitat improvements and evaluation of energy development practices intended to improve mule deer fitness in areas exposed to extensive energy development. We monitored deer on 4 winter range study areas representing varying levels of development to serve as treatment (North Magnolia, South Magnolia) and control (North Ridge, Ryan Gulch) sites. We recorded habitat use and movement patterns, estimated neonatal, overwinter fawn and annual adult female survival, estimated early and late winter body condition of adult females, and estimated late winter annual abundance/density. During this research segment, we targeted 240 fawns (60/study area) and 120 does (30/study area) in early December 2016 for VHF and GPS radiocollar attachment, respectively, and adult female body condition assessment. We attempted recapture of 120 does (30/study area) and 40 fawns (20 in 2 study areas) in March 2017 for late winter body condition assessment. Winter range habitat improvements completed spring 2013 resulted in 604 acres of 1 �mechanically treated pinion-juniper/mountain shrub habitats in each of the 2 treatment areas with relatively minor and extensive energy development, respectively. Post-treatment monitoring will continue for another year to provide sufficient time to measure how vegetation and deer respond to these changes. Based on data collected through year 9 of this 10-year project: (1) annual adult female survival was consistent among areas averaging 79-87% annually, but overwinter fawn survival was variable, ranging from 31% to 95% within study areas, with annual and study area differences primarily due to early winter fawn condition, annual weather conditions, and winter conditions potentially enhancing predation success; (2) migratory mule deer selected for areas with increased cover and increased their rate of travel through developed areas, and avoided negative influences through behavioral shifts in timing and rate of migration, but did not avoid development structures; (3) mule deer body condition early and late winter was generally consistent within areas, with higher variability among study areas early winter, primarily due to December lactation rates, and late winter condition related to seasonal moisture and winter severity; (4) mule deer exhibited behavioral plasticity in relation to energy development, where disturbance distance varied relative to diurnal extent and magnitude of development activity, which may provide for several options in future development planning; (5) late winter mule deer densities have consistently increased in 3 of 4 study areas, averaging about +6% annually, with the North Ridge study area exhibiting erratic population changes that may be an artifact of periodic migration behavior prior to survey timing; and (6) post treatment vegetation responses have provided evidence of improved forage conditions, but longer term monitoring will be required to address the full potential of habitat mitigation efforts. Detailed habitat use analyses are still pending for the pre and post-treatment periods. We will continue to collect population and habitat use data across all study sites to evaluate the effectiveness of habitat improvements on winter range. This approach will allow us to determine whether it is possible to effectively mitigate development disturbances in highly developed areas, or whether it is better to allocate mitigation efforts toward less or non-impacted areas. In collaboration with Colorado State University, we monitored neonate survival in relation to energy development on all study areas since 2012. This will allow us to include neonatal and parturition data with other demographic parameters to evaluate mule deer/energy development interactions. This study is slated to continue through 2018 to allow sufficient time for measuring mule deer population responses to landscape level manipulations. 2 �WILDLIFE RESEARCH REPORT POPULATION PERFORMANCE OF PICEANCE BASIN MULE DEER IN RESPONSE TO NATURAL GAS RESOURCE EXTRACTION AND MITIGATION EFFORTS TO ADDRESS HUMAN ACTIVITY AND HABITAT DEGRADATION CHARLES R. ANDERSON, JR PROJECT NARRITIVE OBJECTIVES 1. To determine experimentally whether enhancing mule deer habitat conditions on winter range elicits behavioral responses, improves body condition, increases fawn survival, and ultimately, population density on mule deer winter ranges exposed to extensive energy development. 2. To determine experimentally to what extent modification of energy development practices enhance habitat selection, body condition, fawn survival, and winter range mule deer densities. SEGMENT OBJECTIVES 1. Collect and reattach GPS collars to maintain sample sizes for addressing mule deer habitat use and behavioral patterns in 4 study areas experiencing varying levels of energy development of the Piceance Basin, northwest Colorado. 2. Estimate early and late winter body condition of adult female mule deer in each of the 4 winter herd segments using ultrasound techniques. Estimate early and late winter fawn weights in areas with and without habitat treatments to assess winter fawn condition relative to habitat improvements. 3. Monitor over-winter fawn and annual adult female mule deer survival by daily ground tracking and bi-weekly aerial tracking. 4. Conduct Mark-Resight helicopter surveys to estimate late winter mule deer abundance and density in each study area. 5. Monitor habitat treatment response for assessing efficacy of habitat improvement projects to mitigate energy development disturbances to mule deer. 6. Continue neonate survival and adult female parturition evaluations to complete demographic parameters for assessing mule deer/energy development interactions. INTRODUCTION Extraction of natural gas from areas throughout western Colorado has raised concerns among many public stakeholders and Colorado Parks and Wildlife (CPW) that the cumulative impacts associated with this intense industrialization will dramatically and negatively affect the wildlife resources of the region. Concern is especially high for mule deer due to their recreational and economic importance as a principal game species and their ecological importance as one of the 3 �primary herbivores of the Colorado Plateau Ecoregion. Extraction of natural gas will directly affect the potential suitability of the landscape used by mule deer through conversion of native habitat vegetation with drill pads, roads, or introduction of noxious weeds, by fragmenting habitat with drill pads and roads, by increasing noise levels via compressor stations and vehicle traffic, and by increasing the year-round presence of human activities. Extraction will indirectly affect deer by increasing the human work-force population of the region resulting in the need for additional landscape conversion for human housing, supporting businesses, and upgraded road/transportation infrastructure. Additionally, increased traffic on rural roads will raise the potential for vehicle-animal collisions. Thus, research documenting these relationships and evaluating the most effective strategies for minimizing and mitigating these activities will greatly enhance future management efforts to sustain mule deer populations for future recreational and ecological values. The Piceance Basin in northwest Colorado contains one of the largest migratory mule deer populations in North America and also covers some of the largest natural gas reserves in North America. Projected energy development throughout northwest Colorado within the next 20 years is expected to reach about 15,000 wells, many of which will occur in the Piceance Basin, which currently supports over 250 active gas well pads (http://cogcc.state.co.us; Fig. 1). Anderson and Freddy (2008) in their long-term research proposal identified 6 primary study objectives to assess measures to offset impacts of energy extraction on mule deer population performance. During the first 5 years of this study, we gathered baseline habitat utilization and demographic data from radiocollared deer across the Piceance Basin to allow assessment of habitat mitigation approaches that were completed April 2013. We are currently monitoring 2 control areas: 1 with development (0.6 pads & facilities/km2; Ryan Gulch) and 1 without (North Ridge). The control areas will be compared with 2 treatment areas experiencing similar development intensities (South Magnolia, 0.9 well pads & facilities/km2 and North Magnolia, 0.1 well pads & facilities/km2), that also received habitat improvements from 2011– 2013 (604 acres each). Habitat and mule deer responses to mechanical habitat treatments will be evaluated until 2018 to assess the success of this habitat mitigation strategy to benefit mule deer exposed to energy development disturbance. In addition, mule deer behavioral patterns in relation to energy development activities in the area are being monitored to identify effective Best Management Practices (BMPs) for future energy development planning. This progress report describes the previous 9.5 years (Jan 2008–June 2017) of mule deer population performance during the pretreatment phase on 4 winter range herd segments, which includes monitoring habitat selection and behavior patterns of adult female mule deer; spring/summer neonate, overwinter fawn and annual adult female survival; estimates of adult female body condition and fawn weights during early and late winter; and annual late-winter abundance/density estimates. STUDY AREAS The Piceance Basin, located between the cities of Rangely, Meeker, and Rifle in northwest Colorado, was selected as the project area due to its ecological importance as home to one of the largest migratory mule deer populations in North America and because it exhibits one of the highest natural gas reserves in North America (Fig. 1). Historically, mule deer numbers on winter range were estimated between 20,000–30,000 (White and Lubow 2002), and the current number of well pads (Fig.1) and projected number of gas wells in the Piceance Basin over the next 20 years is about 250 and 15,000, respectively. Mule deer winter range in the Piceance Basin is predominantly characterized as a topographically diverse pinion pine (Pinus edulis)-Utah juniper (Juniperus osteosperma; pinionjuniper) shrubland complex ranging from 1,675 m to 2,285 m in elevation (Bartmann and Steinert 1981). Pinion-juniper are the dominant overstory species and major shrub species include Utah serviceberry (Amelanchier utahensis), mountain mahogany (Cercocarpus montanus), bitterbrush (Purshia tridentata), big sagebrush (Artemisia tridentata), Gamble’s oak (Quercus gambelii), mountain snowberry (Symphoricarpos oreophilus), and rabbitbrush (Chrysothamnus spp.; Bartmann et al. 1992). The Piceance Basin is segmented by numerous drainages characterized by stands of big 4 �sagebrush, saltbush (Atriplex spp.), and black greasewood (Sarcobatus vermiculatus), with the majority of the primary drainages having been converted to mixed-grass hay fields. Grasses and forbs common to the area consist of wheatgrass (Agropyron spp.), blue grama (Bouteloua gracilis), needle and thread (Stipa comata), Indian rice grass (Oryzopsis hymenoides), arrowleaf balsamroot (Balsamorhiza sagittata), broom snakeweed (Gutierrezia sarothreae), pinnate tansymustard (Descurainia pinnata), milkvetch (Astragalus spp.), Lewis flax (Linum lewisii), evening primrose (Oenothera spp.), skyrocket gilia (Gilia aggregata), buckwheat (Erigonum spp.), Indian paintbrush (Castilleja spp.), and penstemon (Penstemon spp.; Gibbs 1978). The climate of the Piceance Basin is characterized by warm dry summers and cold winters with most of the annual moisture resulting from spring snow melt and brief summer monsoonal rain storms. Wintering mule deer population segments we are investigating include: North Ridge (53 km2) just north of the Dry Fork of Piceance Creek including the White River in the northeastern portion of the Basin, Ryan Gulch (141 km2) between Ryan Gulch and Dry Gulch in the southwestern portion of the Basin, North Magnolia (79 km2) between the Dry Fork of Piceance Creek and Lee Gulch in the north-central portion of the Basin, and South Magnolia (83 km2) between Lee Gulch and Piceance Creek in the south-central portion of the Basin (Fig. 1). Each of these wintering population segments has received varying levels of natural gas development: no development in North Ridge, light development in North Magnolia (0.1 pads & facilities/km2), and relatively high development in the Ryan Gulch (0.6 pads & facilities/km2) and South Magnolia (0.9 pads & facilities/km2) segments (Fig. 1). Development activity was high through 2011 and has declined substantially since natural gas prices began to decline in 2012. Among the 4 study areas, North Ridge has served as an unmanipulated control site, Ryan Gulch will serve to address human-activity management alternatives (BMPs) that benefit mule deer exposed to energy development and as a developed control area for comparison to the developed treatment area receiving habitat improvements (South Magnolia), and North and South Magnolia will allow us to assess the utility of habitat treatments intended to enhance mule deer population performance in areas exposed to light (North Magnolia) and relatively heavy (South Magnolia) energy development activities. METHODS Tasks addressed this period included mule deer capture and collaring, monitoring neonate, overwinter fawn and annual adult female survival, estimating adult female body condition during early and late winter using ultrasonography and winter fawn condition measuring early and late winter fawn weights, estimating mule deer abundance applying helicopter mark-resight surveys, and monitoring vegetation responses to habitat treatments completed spring 2013. We employed helicopter netgunning techniques (Barrett et al. 1982, van Reenen 1982) to target 240 fawns and 120 adult females during early December 2016, and 120 adult females and 40 fawns (primarily recaptures) during early March 2017. Once netted, all deer were hobbled and blind folded. Fawns were weighed and radiocollared, and sex was recorded prior to release at the capture site. Adult females were transported to localized handling sites for recording body measurements and fitted with GPS collars (5 fix attempts/day; G2110D, Advanced Telemetry Systems, Isanti, MN, USA) prior to release. To provide direct measures of decline in overwinter body condition, we targeted 30 adult females in each study area that were captured the previous December. During March, 20 fawns were recaptured, weighed and released in South Magnolia (within the habitat treatment areas) and Ryan Gulch (control area) to quantify overwinter declines in fawn body condition. Fawn collars were spliced and fitted with rubber surgical tubing to facilitate collar drop between mid-summer and autumn for winter fawns and during winter for neonates, and GPS collars were supplied with timed drop-off mechanisms scheduled to release early April of the year following deployment. All radio-collars were equipped with mortality sensing options (i.e., increased pulse rate following 8 hrs of inactivity). 5 �Mule Deer Habitat Use and Movements We downloaded and summarized data from GPS collars deployed and recovered since 2008. GPS collars maintained the same schedule of attempting to collect locations every 5 hours, except for 40 does in Ryan Gulch and 10 control deer from North Ridge where location rates were programmed for every 30-60 minutes to increase resolution of movement data for evaluation of deer behavior patterns in relation to differing development activities. Joe Northrup (CSU PhD Candidate) recently analyzed resource selection data relative to energy development (Northrup 2015) and those results are addressed below. Mule deer resource selection analyses to address success of habitat improvements are pending until vegetation responses are fully realized, which will begin by fall 2018. Mule Deer Survival Mule deer mortality monitoring consisted of daily ground-telemetry tracking and aerial monitoring approximately every 2 weeks from fixed-wing aircraft on winter range and weekly aerial monitoring on summer range. Once a mortality signal was detected, deer were located and necropsied to assess cause of death (Stonehouse et al., 2016). We estimated weekly survival using the staggered entry Kaplan-Meier procedure (Kaplan and Meier 1958, Pollock et al. 1989). Capture-related mortalities (any doe/fawn mortalities occurring within 10 days of capture; excluding neonates) and collar failures were censored from survival rate estimates. We estimated survival rates from 1 July 2016 through 30 June 2017 for adult females, from birth to mid December for neonates, and from early December 2016–mid June 2017 for winter fawns. Adult Female Body Measurements We applied ultrasonography techniques described by Stephenson et al. (1998, 2002) and Cook et al. (2001) to measure maximum subcutaneous rump fat (mm), loin depth (longissimus dorsi muscle, mm), and to estimate % ingesta-free body fat. We estimated a body condition score (BCS) for each deer by palpating the rump (Cook et al. 2001, 2007, 2009). We examined differences (P < 0.05) in nutritional status among study areas and between years evident in non-overlapping 95% confidence intervals. We considered differences in body condition meaningful when mean rump fat or % body fat differed statistically between comparisons. Other body measurements recorded included pregnancy status (pregnant, barren) via blood samples, fetal counts using ultrasonagraphy, weight (kg), chest girth (cm), and hind-foot length (cm). Abundance Estimates We conducted 4 helicopter mark-resight surveys (2 observers and the pilot) during late March/early April to estimate deer abundance in all 4 study areas. We delineated each study area from GPS locations collected on winter range during the first 3 years of the study (Jan 2008 through April 2011). Two aerial fixed-wing telemetry surveys/study area were conducted during helicopter mark-resight surveys to determine which marked deer were within each survey area, and we confirmed adult female locations during surveys from GPS data acquired April 2017. We delineated flight paths in ArcGIS 10.0 prior to surveys following topographic contours (e.g., drainages, ridges) and approximating 500600 m spacing throughout each study area; flight paths during surveys were followed using GPS navigation in the helicopter. Two 12 x 12 cm pieces of Ritchey livestock banding material (Ritchey Livestock ID, Brighton, CO USA) were uniquely marked using color, number, and symbol combinations and attached to each radio-collar to enhance mark-resight estimates. Each deer observed during surveys was recorded as mark ID#, unmarked, or unidentified mark. 6 �We used program MARK (White and Burnham 1999), applying the immigration-emigration mixed logit-normal model (McClintock et al. 2008), to estimate mule deer abundance and confidence intervals. For mark-resight model evaluations, we examined parameter combinations of varying detection rates with survey occasion and whether individual sighting probabilities (i.e., individual heterogeneity) were constant or varied (2 = 0 or 0). Model selection procedures followed the information-theoretic approach of Burnham and Anderson (2002). RESULTS AND DISCUSSION Deer Captures and Survival The helicopter crew captured 242 fawns and 121 does during Dec 2016 and 122 does and 41 fawns during March 2017. Sixteen fawn mortalities (6.6%; proximate cause = 5 capture myopathy, 11 predation) occurred within the 10 day censorship period during December and 2 fawn mortalities (4.9%; 1 capture myopathy, 1 predation) occurred during the March capture. Doe mortalities totaled 2 (1.7%; capture myopathy) and 4 (3.3%; capture myopathy) within 10 days of the December and March capture periods, respectively. Mortality rates 10 days post capture have typically varied between 2.5– 3.5% for fawns and does since Jan 2008, except during the 2011–2012 capture season where myopathy rates were higher (3–6%) due to dry, warm conditions (Anderson and Bishop 2012). Excluding December fawn captures, myopathy rates were comparable to expected levels when compared to previous years. The relatively high myopathy rates (including ongoing predation that occurred within the 10-day censorship period) for early winter fawn captures were likely linked to the relatively severe winter conditions evident through January 2017, resulting in crusted snow conditions, and unusually high levels of coyote (Canis latrans) predation; crusted snow conditions may have enhanced predation by coyotes by enabling them to stay above the snow while fleeing deer were inhibited by breaking through the crust. Fawn survival estimates from early December 2016 through mid June 2017 were similar (overlapping 95% CIs) among 3 study areas ranging from 0.51 to 0.67, with a lower survival estimate from South Magnolia fawns (0.31; Table 1, Fig. 1). In comparison to previous years, coyote predation was much more common in all areas this winter and the dominant proximate cause of mortality in South Magnolia. This increase in coyote predation may be related to multiple factors including increasing coyote numbers in response to high rabbit densities the past few years following by a recent drop in the rabbit population in combination with crusted snow conditions enhancing coyote predation success on mule deer fawns. General comparisons to previous years suggest relatively low fawn survival this winter, which was comparable to the lower survival rates observed during 2010-11 (Fig. 2). Both low survival winters coincided with harsher weather conditions with an added influence of predation coupled with crusted snow this past winter. Winter fawn survival (Fig. 2) also appears to correlate with summer forage conditions as suggested from relative December fawn weights (Fig. 3). Annual adult female survival varied from 0.73 (Ryan Gulch) to 0.91 (North Magnolia; Table 1) during 2016–17, but was comparable among study areas during 2016–17 and to previous years (P > 0.05), with the exception of lower survival in North Magnolia during 2011–12 (Ŝ = 0.68, Anderson and Bishop 2012). Relatively low sample sizes per study area for adult female survival do not allow statistical discrimination among years unless large differences are evident (e.g., >1520%). Estimates below 80% are biologically concerning if these values represent the respective population, but low statistical power precludes confirmation within study areas. When combined among study areas, annual survival estimates have varied from 79% in 2012-13 to 86% in 2014-15 and was 83% this year, which is comparable to the long term average. Lower combined survival estimates are consistent with extreme environmental conditions consisting of dryer moisture conditions during late winter/spring (2012-13) and/or cold temperatures with heavy snow during early winter (2015-16). 7 �Spring Migration Patterns Collaboration with Idaho State University to address mule deer migration patterns in developed and undeveloped landscapes (funded from energy company contributions) has been completed. Four manuscripts from this effort have been published (Lendrum et al. 2012, Lendrum et al. 2013, Lendrum et al. 2014, Anderson and Bishop 2014; Appendix A). In addressing habitat selection during spring migration, Lendrum et al. (2012; Fig. 4) noted that mule deer migrating through the most developed landscapes exhibited longer step lengths (straight line distance between GPS locations) and selected habitats providing greater security cover than deer in undeveloped landscapes that migrated through more open areas that provided increased foraging opportunities. Migrating deer also selected areas closer to well pads, but avoided roads, except in the highest developed areas where road densities were likely too high for avoidance without significant deviations from traditional migration routes. In the second manuscript Lendrum et al. (2013) addressed biological and environmental factors influencing spring migration and assessed how energy development influenced migratory behavior. Overall, spring migration was influenced by snow depth, temperature, and green-up on winter and summer range; increasing temperatures, snow melt and emerging vegetation dictated timing of winter range departure and summer range arrival. Duration of Piceance Basin mule deer migration was short, with median migration durations of 3–8 days among the 4 areas (straight line distance between seasonal ranges averaged 32–40 km). Deer in poor condition migrated later than deer in good condition, but condition was similar among areas regardless of development status (Table 2). Migrating deer from developed study areas did not avoid development structures, but departed later, arrived earlier and migrated more quickly than deer from undeveloped areas. While large changes in timing of migration could have nutritional consequences and negatively influence reproduction and neonate survival, the relatively minor shift we observed should not result in longterm fitness consequences. Migratory deer in the Piceance Basin appear to avoid negative effects of energy development through behavioral shifts in timing and rate of migration. In the third publication Lendrum et al. (2014), monitored migratory mule deer in the Piceance Basin to examined the relationship between the Normalized Difference Vegetation Index (NDVI), which is a course-scale measure of forage quality using a GIS assessment of vegetation greenness, and fecal nitrogen to assess the assumption that forage quality and deer diets can be reasonably linked to address deer habitat use patterns from remotely sensed data. We found that diet quality evident from fecal nitrogen and course measures of vegetation green-up were informative, and that Piceance Basin mule deer exhibited rapid migration (3 to 8 days depending on study area), left winter range following snow melt with lowest fecal N and NDVI values, and progressed to summer range as vegetation green-up and nitrogen levels increased, but ahead of peak vegetation green-up on summer range. It is plausible that this rapid migration strategy is evident for deer in relatively good condition and allows for early arrival on summer range to take advantage of optimal forage conditions prior to parturition. Anderson and Bishop (2014) summarized results from Lendrum et al. (2012, 2013) and Sawyer et al. (2012) addressing migratory mule deer and energy development in northwest Colorado and south-central Wyoming, respectively. The interactions between migratory mule deer and energy development identified by Lendrum et al. (2012, 2013) and Sawyer et al. (2012) suggest mule deer may benefit from energy development planning by considering thresholds of development that may alter migratory behavior. It appears that migration rate, migration routes, and stopover use, if present, may be altered at high development intensities. In addition, migratory mule deer may benefit by maintaining security cover along migration paths, and improved habitat conditions may facilitate more direct and rapid migration requiring less energy to complete migration. Enhancing permeability along migration routes by applying dispersed development plans (<2 well pads/km2) and minimizing 8 �disturbance to vegetation types by maintaining security cover should reduce impacts to migratory mule deer as well as other migratory ungulates. Where feasible, habitat improvement projects on winter range and possibly stopover sites would also enhance migratory mule deer populations by enhancing energy reserves for long-distance movements and parturition shortly after summer range arrival. Where possible, directional drilling could be used to extract energy resources from underneath migration routes while maintaining no or minimal surface occupancy. Lastly, we emphasize that GPS studies now allow managers to accurately map migration routes for entire populations and identify relatively narrow corridors that are most heavily used thus allowing for the identification of the most important corridors for migrating ungulates. Where available, we encourage agencies to incorporate such migration corridors into land-use plans (e.g., resource management plans) and National Environmental Policy Act documents. Mule Deer Body Condition Early-winter body condition measurements of adult female mule deer during December 2016 were similar among study areas (P < 0.05, Fig. 5, Table 2). Early winter condition this year was moderate to low compared to previous years with notably high condition exhibited from Ryan Gulch does during December 2011 and 2015 (Fig. 5). Adult female body condition during early winter appears primarily related to the proportion of lactating does identified during December captures, where higher condition correlates with lower lactation rates. Since 2011, excluding Ryan Gulch during 2014, North Ridge during 2016 and North Magnolia this year, late winter body condition initially trended upward and appears to be stabilizing recently (Fig. 5, Table 2). The observed increase was likely related to relatively mild winters from 2012 - 2015 and the generally stabilizing condition may be related to more severe conditions during early winter the past 2 years; temperatures increased during mid-winter resulting in snow-melt, which likely improved late winter condition above what might have resulted if severe winter conditions continued. Adult female body condition thus far appears more related to early winter lactation rates, seasonal moisture conditions, relative deer densities (Fig. 6), and winter severity than observed development intensity thus far. December 2016 fawn weights were generally lower than observed during previous years (about 3.6 kg below average; Fig. 3). December fawn condition has been correlated with winter fawn survival (Fig. 2), which was consistent with the relatively low winter fawn survival observed this past winter. Because adult female body condition has been largely uninformative in regards to habitat treatment responses (pending further analyses), we began late winter fawn recaptures in South Magnolia (habitat treatment area) and Ryan Gulch (control area) to assess changes in over-winter body condition the past 2 years. Fawns from both areas exhibited significant weight loss (P < 0.05) during 2015-16, with fawns from the treatment area exhibiting significantly less weight loss (P < 0.001; -3.1 kg) than fawns from the untreated area (-5.6 kg). Results from winter 2016-17 differed in that fawns from both areas exhibited similarly reduced body condition early winter (about 4 kg lighter on average) and maintained their weights into late winter (negligible weight differences from Dec to Mar). These results are conflicting with respect to habitat treatment effects, but will require more detailed analyses to address other factors that may influence nutritional benefits of habitat improvements on winter range. We will continue monitoring over-winter condition of fawns from the treatment and control areas to evaluate over-winter fawn condition in areas with and without habitat improvements. Mule Deer Behavioral Response to Energy Development We recently completed evaluations of deer behavior patterns in relation to energy development activities (Northrup et al. 2015, 2016a, 2016b). We found diurnal responses to development activity, where deer used timbered areas away from development activity while bedded during the day and 9 �moved into more open areas generally closer to developed areas while foraging at night. Disturbance distances from producing pads and roads declined from 600 m to 200 m and about 140 m to 60 m from daytime to nighttime, respectively, but increased from 600 m to 800 m for nighttime drilling pad activity. We suspect deer behaviorally respond to fluctuations in development activity, where road traffic and producing well pad activity decline at night, but drilling pad disturbance may increase from compressors and lights used to facilitate nighttime drilling activity. These evaluations were applied during an active drilling phase in the Piceance Basin and deer use was influenced by development activity in 25% (nighttime) to 50% (day time) of critical winter range during that period. However, deer densities have comparably increased among developed and undeveloped study areas (Fig. 6) suggesting that deer can behaviorally mediate development disturbance under observed development and deer densities by taking advantage of fluctuations in development activity to address their nutritional requirements. Given the plasticity in deer behavior, a number of potential options for future development planning exits including drilling schedule modifications (seasonal and/or diurnal), concentrated/staged development, reducing road traffic, and using light/noise barriers around drill rigs. It will be informative to determine if habitat improvements will further reduce development disturbance and increase management options for future development planning. Neonate Survival To complete demographic parameters addressing mule deer–energy development interactions, CPW, Colorado State University, and ExxonMobil Production entered into a collaborative agreement to investigate neonate survival and adult female parturition in developed and undeveloped landscapes (funded by ExxonMobil Production Co.) beginning spring 2012. Mark Peterson (Graduate Research Assistant) and Paul Doherty (CSU professor) assisted with this research, which was completed December 2014, and continued by CPW during 2015 and 2016. Neonate capture and collaring efforts totaled 85 during spring 2012, 67 during spring 2013, 54 during spring 2014, 59 during spring 2015, and 58 during spring 2016. Overall, estimated neonate survival through mid-December was 0.39 (95% CI = 0.28–0.50) during 2012, 0.37 (95% CI = 0.25–0.48) during 2013, 0.57 (95% CI = 0.44 – 0.70) during 2014, 0.36 (95% CI = 0.23–0.49) during 2015, and 0.32 (95% CI = 0.19–0.44) during 2016. Through December 2016, predation was the highest proximate mortality factor (averaging about 50% annually), with relatively low incidents of starvation/disease directly influencing spring/summer fawn survival (mean < 4%). Manuscripts addressing neonate fawn survival and adult female parturition in relation to energy development are currently in preparation and review for publication. Mule Deer Population Estimates Mark-resight models that best predicted abundance estimates (lowest AICc; Burnham and Anderson 2002) exhibited variable sightability across surveys (Pt) for all study areas, and variable individual sightability (2 = 0) for North Magnolia deer and homogenous sightability (2 ≠ 0) for the other 3 areas. North Ridge exhibited the highest deer density (15.8/km2), with comparable but lower deer densities in the other 3 areas (10.3–11.9/km2; Table 3, Fig. 6). Densities similarly increased over the 9 year monitoring period in 3 of the 4 study areas averaging about 6% annual increases from North Magnolia, South Magnolia and Ryan Gulch. Mule deer density estimates from North Ridge have been erratic with a significant decline this past year (Fig. 6). Biological support for the recent decline is lacking based on similar demographic parameter estimates when compared to the other study areas showing population growth (Tables 1 and 2, Figs. 2, 3 and 5). Reasons for the recent decline in North Ridge are unclear, but lack of closure partially due to early migration has been an issue in the past and may have artificially reduced the population estimate this year. Active GPS collars addressing last year’s movement patterns will be collected April 2018 and movement data will be assessed to address the potential closure issue. 10 �Abundance estimates from 2017 were similarly precise from all 4 study areas, but error of estimates was greater (mean CICVs ranged from 0.19 – 0.22) than past surveys (typically ~0.15). Increased error was likely associated with reduced sightability from a different helicopter (Hughes MD 500) used than during previous years (Bell 47). We shifted to the Hughes MD 500 helicopter to increase power and consistently allow for 3 people in the helicopter to assist with deer surveys. The Bell 47 is less powerful and, depending on weather conditions, may only provide for 2 people in the helicopter, but provides for increased visibility. Based on estimate comparisons using the 2 different helicopters, the Bell 47 provides improved population estimates even with the potential limitation of survey personnel and will be used for future surveys. Magnolia Habitat Treatments We completed 116 acres of pilot habitat treatments in January 2011 (Anderson and Bishop 2011; Environmental Assessment: DOI-BLM-CO-110-2011-004-EA), 54 acres of mechanical treatment method comparison treatments (hydro-ax, roller-chop, chaining) in January 2012 (Stephens 2014), and 1,038 acres of hydro-ax treatments in April 2013 (Determination of NEPA Adequacy: DOIBLM-CO-110-2012-0134- DNA), totaling 604 treated acres in each study area (Fig. 7). Vegetation response in the pilot treatment sites was visually evident by fall 2011 (Fig. 7), and resulted in statistically significant (P < 0.05) increases in native grass and forb cover by the 2014 growing season. 2016 results are still pending, but shrub responses appear promising from data collected last fall. Stephens (2014) reported that all 3 mechanical treatment methods compared resulted in roughly a 3 fold increase in grasses, forbs, and shrubs combined after 2 growing seasons (versus control sites), but cautioned that rollerchop treatments may be more vulnerable to invasive species response. Vegetative responses from 2013 hydro-ax treatments were visually evident following 1 growing season and shrub responses have been notable since the 4th growing season, but statistical comparisons are still pending. As anticipated, grass and forb responses were evident 2 to 3 years post-treatment, with longer term response expected (3-5 years) for palatable shrubs. Of note, relatively high moisture conditions experienced during spring 2014 and 2015 resulted in higher than normal prevalence of cheatgrass (Bromus tectorum); cheatgrass invasion has previously been minor to non-existent. Cheatgrass invasion, however, does not appear directly related to treatment sites because occurrence is evident in both treatment and control areas. We anticipate this outbreak will subside based on past competitive advantage of native species to dominate, but will continue to monitor species composition and address cheatgrass persistence in treatment and control sites. GPS data addressing deer use of treatment sites is becoming available and will be analyzed as additional data are collected and vegetation responses progress. We observed improved fawn condition (P < 0.001) in South Magnolia following the 4 th growing season of habitat treatments when compared to fawn condition in the Ryan Gulch control area. Vegetation and mule deer responses will continue to be documented for the next year to assess the utility of this mitigation approach in benefiting mule deer exposed to energy development disturbance. SUMMARY AND COLLABORATIONS The long-term goal of this study is to investigate habitat treatments and energy development practices that enhance mule deer populations exposed to extensive energy development activity. The information presented here summarizes mule deer population parameters from the 4-year pretreatment period and 5 years post-treatment. The pretreatment period was completed by spring 2013, providing baseline data for comparison with intended improvements in habitat conditions and response to varying degrees in human development activity. Winter range habitat improvements resulting in 604 acres of mechanically treated pinion-juniper/mountain shrub habitats in each of 2 11 �study areas were completed by April of 2013, and subsequent vegetation responses have met or exceeded expectations. Post-treatment monitoring will continue for another year to provide sufficient time to measure how deer respond to these changes. Based on data collected through year 9 of this 10-year project: (1) annual adult female survival was consistent among areas averaging 79-87% annually, but overwinter fawn survival was variable, ranging from 31% to 95% within study areas, with annual and study area differences primarily due to early winter fawn condition, annual weather conditions, and winter conditions potentially enhancing predation success; (2) migratory mule deer selected for areas with increased cover and increased their rate of travel through developed areas, and avoided negative influences through behavioral shifts in timing and rate of migration, but did not avoid development structures; (3) mule deer body condition early and late winter was generally consistent within areas, with higher variability among study areas early winter, primarily due to December lactation rates, and late winter condition related to seasonal moisture and winter severity; (4) mule deer exhibited behavioral plasticity in relation to energy development, where disturbance distance varied relative to diurnal extent and magnitude of development activity, which may provide for several options in future development planning; and (5) late winter mule deer densities have consistently increased in 3 of 4 study areas, averaging about +6% annually, with the North Ridge study area exhibiting erratic population changes that may be an artifact of periodic migration behavior prior to survey timing. Detailed habitat use analyses are pending for the pre and post-habitat treatment periods. We will continue to collect the various population and habitat use data across study sites to evaluate the effectiveness of habitat improvements on winter range. This approach will allow us to determine whether it is possible to effectively mitigate development impacts in highly developed areas, or whether it is better to allocate mitigation dollars toward less or non-impacted areas. In a previous project conducted on the Uncomphahgre Plateau, Colorado, Bergman et al. (2014) found that habitat treatments implemented in pinion-juniper habitat in undeveloped areas increased overwinter survival of fawns by a magnitude of 1.15. Hay field improvements have been completed in the North Magnolia study area by WPX Energy to fulfill a Wildlife Management Plan (WMP) agreement with CPW; elk (Cervus elaphus) response has been evident, but mule deer response has thus far been minor. A similar WMP agreement between ExxonMobil/XTO Energy and CPW allowed completion and continued monitoring of mechanical habitat improvements in the Magnolia study areas. Collaborative research with agency biologists, graduate students, and university professors has produced 15 scientific publications addressing improved monitoring techniques for neonate mule deer captures (Bishop et al. 2011), approaches to address proximate mortality factors from field necropsies of mule deer (Stonehouse et al. 2016), mule deer migration (Lendrum et al. 2012, 2013, 2014; Anderson and Bishop 2014), improved approaches to address animal habitat use patterns (Northrup et al. 2013), mule deer response to helicopter capture and handling (Northrup et al. 2014a), potential effects of male-biased harvest on mule deer productivity (Freeman et al. 2014), mule deer genetics in relation to body condition and migration (Northrup et al. 2014b), spatial and temporal factors influencing auditory vigilance in mule deer (Lynch et al. 2014), the relationship of plant phenology with mule deer body condition (Seral et al. 2015), and mule deer responses to differing energy development activities to inform future development planning (Northrup et al. 2015, Northrup et al. 2016a, Northrup et al. 2016b); these publications are summarized in Appendix A. Ongoing cooperative agreements will be necessary to sustain this project to completion (through 2018). We anticipate the opportunity to work cooperatively toward developing solutions for allowing the nation’s energy reserves to be developed in a manner that benefits wildlife and the people who value both the wildlife and energy resources of Colorado. 12 �LITERATURE CITED Anderson, C. R., Jr., and D. J. Freddy. 2008. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Final Study Plan, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C. R., Jr., and C. J. Bishop. 2011. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C. R., Jr., and C. J. Bishop. 2012. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Parks and Wildlife, Ft. Collins, CO, USA. Anderson, C. R., Jr., and C. J. Bishop. 2014. Migration patterns of adult female mule deer in response to energy development. Pages 47-50 in Transactions of the 79th North American Wildlife & Natural Resources Conference (R. A. Coon & M. C. Dunfee, eds.). Wildlife Management Institute, Gardners, PA, USA. ISSN 0078-1355. Bartmann, R. M. 1975. Piceance deer study—population density and structure. Job Progress Report, Colorado Division of Wildlife, Fort Collins, Colorado, USA. Bartmann, R. B., and S. F. Steinert. 1981. Distribution and movements of mule deer in the White River Drainage, Colorado. Special Report No. 51, Colorado Division of Wildlife, Fort Collins, Colorado, USA. Bartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monograph No. 121. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108-114. Bergman, E. J., C. J. Bishop, D. J. Freddy, G. C. White, and P F. Doherty, Jr. 2014. Habitat management influences over-winter survival of mule deer fawns in Colorado. Journal of Wildlife Management 78(3):448–455; DOI: 10.1002/jwmg.683 Burnham, K. P., and D. R. Anderson. 2002. Model selection and multi-model inference: a practical information-theoretic approach. Second edition. Springer-Verlag, New York, New York, USA. Cook, R. C., J. G. Cook, D. L. Murray, P. Zager, B. K. Johnson, and M. W. Gratson. 2001. Development of predictive models of nutritional condition for rocky mountain elk. Journal of Wildlife Management 65:973-987. Cook, R. C., T. R. Stephenson, W. L. Meyers, J. G. Cook, and L. A. Shipley. 2007. Validating predictive models of nutritional condition for mule deer. Journal of Wildlife Management 71:1934-1943. Cook, R. C., J. G. Cook, T. R. Stephenson, W. L. Meyers, S. M. McCorquodale, D. J. Vales, L. L. Irwin, P. Briggs Hall, R. D. Spencer, S. L. Murphie, K. A. Schoenecker, P. J. Miller. 2009. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife Management 74:880-896. Freeman, E. D., R. T. Larsen, M. E. Peterson, C. R. Anderson, Jr., K. R. Hersey, and B. R. McMillan. 2014. Effects of male-biased harvest on mule deer: implications for rates of pregnancy, synchrony, and timing of parturition. Wildlife Society Bulletin; DOI: 10.1002/wsb.450 Gibbs, H. D. 1978. Nutritional quality of mule deer foods, Piceance Basin, Colorado. Thesis, Colorado State University, Fort Collins, Colorado, USA. Kaplan, E. L., and P. Meier. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 52:457-481. 13 �Lendrum, P. E., C. R. Anderson, Jr., R. A. Long, J. K. Kie, and R. T. Bowyer. 2012. Habitat selection by mule deer during migration: effects of landscape structure and natural gas development. Ecosphere 3(9):82. http://dx.doi.org/10. Lendrum, P. E., C. R. Anderson, Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2013. Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes. PLoS ONE 8(5): e64548. doi:10.1371/journal.pone.0064548 Lendrum, P. E., C. R. Anderson, Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2014. Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns of forage quality. Mammalian Biology: http://dx.doi.org/10.1016/j.mambio.2014.05.005 Lynch, E., J. M. Northrup, M. F. McKenna, C. R. Anderson Jr., L. Angeloni, and G. Wittemyer. 2014. Landscape and anthropogenic features influence the use of auditory vigilance by mule deer. Behavioral Ecology; doi:10.1093/beheco/aru158. McClintock, B. T., G. C. White, K. P. Burnham, and M. A. Pride. 2008. A generalized mixed effects model of abundance for mark—resight data when sampling is without replacement. Pages 271- 289 in D. L. Thompson, E. G. Cooch, and M. J. Conroy, editors, Modeling demographic processes is marked populations. Springer, New York, New York, USA. Northrup, J. M. 2015. Behavioral response of mule deer to natural gas development in the Piceance Basin. Dissertation, Colorado State University, Fort Collins, USA. Northrup, J. M., M. B. Hooten, C. R. Anderson, Jr., and G. Wittemyer. 2013. Practical guidance on characterizing availability in resource selection functions under a useavailability design. Ecology 94(7):1456-1463. Northrup, J. M., C. R. Anderson, Jr., and G. Wittemyer. 2014a. Effects of helicopter capture and handling on movement behavior of mule deer. Journal of Wildlife Management 78(4):731738; DOI: 10.1002/jwmg.705 Northrup, J. M., A. B. Shafer, C. R. Anderson Jr., D. W. Coltman, and G. Whittemyer. 2014b. Finescale genetic correlates to condition and migration in a wild cervid. Evolutionary Applications ISSN 1752-4571; doi: 10.1111/eva.12189 Northrup, J. M., C. R. Anderson, Jr., and G. Wittemyer. 2015. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer. Global Change Biology, doi: 10.1111/gcb.13037. Northrup, J. M., C. R. Anderson, Jr., M. B. Hooten, and G. Wittemyer. 2016a. Movement reveals scale dependence in habitat selection of a large ungulate. Ecological Applications 26:27462757 Northrup, J. M., C. R. Anderson, Jr., and G. Wittemyer. 2016b. Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid. Diversity and Distributions 22: 547-557, DOI: 10.1111/ddi.12417 Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. C. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Sawyer, H., M. J. Kauffman, A. D. Middleton, T. A. Morrison, R. M. Nielson, and T. B. Wycoff. 2012. A framework for understanding semi-permeable barrier effects on migratory ungulates. Journal of Applied Ecology 50:68-78. Searle, K. R., M. B. Rice, C. R. Anderson, C. Bishop and N. T. Hobbs. 2015. Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore. Oecologia ISSN 0029- 8549; DOI 10.1007/s00442-015-3348-9 Stephens, G. J. 2014. Understory responses to mechanical removal of pinyon-juniper overstory. MS Thesis, Colorado State University, Ft. Collins USA. Stephenson, T. R., V. C. Bleich, B. M. Pierce, and G. P. Mulcahy. 2002. Validation of mule deer body composition using in vivo and post-mortem indices of nutritional condition. Wildlife Society Bulletin 30:557-564. Stephenson, T. R., K. J. Hundertmark, C. C. Swartz, and V. Van Ballenberghe. 1998. Predicting body fat and mass in moose with untrasonography. Canadian Journal of Zoology 76:717-722. 14 �Stonehouse, K. F., C. R. Anderson Jr., M. E. Peterson, and D. R. Collins. 2016. Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus). Colorado Parks and Wildlife Technical Report No. 48, First Edition, 317 W. Prospect Rd., Ft. Collins, CO USA. DOW-R-T-48-16, ISSN 0084-8883. Unsworth, J. W., D. F. Pack, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. Van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked individuals. Bird Study 46:120-139. White, G. C., and B. C. Lubow. 2002. Fitting population models to multiple sources of observed data. Journal of Wildlife Management 66:300-309. Prepared by Charles R. Anderson, Jr., Mammals Research Leader 15 �Table 1. Survival rate estimates (Ŝ) of fawn (1 Dec. 2016–15 June 2017) and adult female (1 July 2016– 30 June 2017) mule deer from 4 winter range study areas of the Piceance Basin in northwest Colorado. Cohort Study area Initial sample size (n) March doe samplea (n) Ŝ (95% CI) Fawns Ryan Gulch 61 0.506 (0.381–0.632) South Magnolia 53 0.312 (0.183–0.441) North Magnolia 54 0.655 (0.525–0.785) North Ridge 60 0.665 (0.546–0.785) Adult females Ryan Gulch 28 50 0.726 (0.570–0.882) South Magnolia 23 46 0.849 (0.720–0.997) North Magnolia 29 57 0.914 (0.829–0.999) North Ridge 19 47 0.845 (0.721–0.968) a Adult female sample sizes following capture and radio-collaring efforts March, 2017. 16 �Table 2. Mean rump fat (mm), Body Condition Score (BCSa), and % ingesta-free body fat (% fat) of adult female mule deer from 4 study areas in the Piceance Basin of northwest Colorado, March and December, 2009–2017. Values in parentheses = SD. March 2009 December 2009 March 2010 Study Area Rump fat Ryan Gulch 1.73 (1.78) 2.66 (0.55) 7.08 (1.27) 8.35 (6.36) 4.06 (1.13) 10.54 (3.72) 2.31 (1.44) 2.35 (0.48) 6.37 (1.41) South Magnolia 1.29 (0.47) 2.51 (0.66) 6.74 (2.27) 10.05 (6.19) 4.07 (1.21) 11.44 (3.50) 3.12 (2.20) 2.64 (0.59) 7.11 (1.69) North Magnolia 1.31 (1.01) 2.66 (0.68) 7.15 (1.63) 10.67 (5.76) 4.25 (0.96) 11.94 (3.39) 3.15 (2.34) 2.85 (0.53) 7.54 (1.53) North Ridge 1.57 (1.22) 2.60 (0.56) 6.81 (1.68) 5.25 (5.65) 3.63 (1.11) 9.37 (3.08) 1.77 (1.11) 2.42 (0.49) 6.39 (1.45) March 2011 December 2011 BCS % fat Rump fat BCS % fat Rump fat BCS % fat Table 2. Continued. December 2010 Study Area Rump fat BCS % fat Rump fat Ryan Gulch 7.26 (6.36) 3.24 (0.96) 9.69 (3.56) 1.55 (0.60) 2.53 (0.42) 6.72 (1.37) 13.41 (6.39) 4.21 (1.17) 13.17 (3.64) South Magnolia 9.85 (6.78) 3.30 (0.61) 11.27 (3.75) 1.65 (0.75) 2.35 (0.50) 6.15 (1.75) 8.18 (5.45) 3.41 (0.82) 10.34 (3.28) North Magnolia 9.55 (6.49) 3.46 (1.16) 10.79 (4.26) 1.65 (0.67) 2.53 (0.49) 6.79 (1.47) 8.76 (5.77) 3.74 (0.91) 10.73 (3.14) North Ridge 7.25 (5.41) 3.47 (0.86) 9.85 (3.02) 1.45 (0.76) 2.24 (0.49) 6.30 (1.65) 8.86 (5.37) 3.51 (0.99) 10.77 (3.33) 15 BCS % fat Rump fat BCS % fat �Table 2. Continued. March 2012 December 2012 BCS % fat Rump fat BCS March 2013 Study Area Rump fat % fat Ryan Gulch 2.15 (1.44) 2.74 (0.44) 7.22 (1.16) 6.34 (4.35) 3.30 (0.77) 9.34 (2.43) 1.87 (0.90) 2.65 (0.37) 7.14 (0.89) South Magnolia 1.66 (0.77) 2.59 (0.36) 7.03 (1.13) 8.30 (5.71) 3.46 (1.07) 10.32 (3.23) 2.06 (0.77) 2.65 (0.26) 7.19 (0.66) North Magnolia 1.90 (0.76) 2.84 (0.34) 7.61 (0.96) 9.66 (6.41) 3.84 (1.16) 11.18 (3.64) 1.76 (0.91) 2.59 (0.41) 6.87 (1.11) North Ridge 2.24 (1.58) 2.70 (0.35) 7.26 (1.05) 5.76 (4.10) 3.32 (0.82) 1.87 (0.73) 2.48 (0.34) 6.70 (1.12) 9.06 (2.31) Rump fat BCS % fat Table 2. Continued. December 2013 BCS March 2014 % fat Rump fat Study Area Rump fat Ryan Gulch 9.27 (6.29) 3.47 (0.87) 10.61 (3.76) 1.69 (0.85) 2.68 (0.39) 7.03 (0.99) 8.50 (6.76) 3.69 (1.03) 10.56 (3.70) South Magnolia 11.27 (8.40) 3.99 (1.04) 11.40 (4.16) 2.57 (1.61) 2.96 (0.30) 7.75 (0.68) 10.96 (6.82) 4.08 (1.06) 11.98 (3.81) North Magnolia 9.00 (6.15) 3.44 (0.78) 10.48 (3.25) 2.33 (2.12) 2.80 (0.49) 7.31 (1.43) 9.52 (5.83) 3.83 (1.04) 11.18 (3.32) North Ridge 11.17 (5.28) 3.85 (0.72) 11.66 (2.69) 2.38 (1.52) 2.68 (0.39) 7.16 (1.14) 7.93 (5.50) 3.74 (0.76) 10.20 (3.01) 16 BCS December 2014 % fat Rump fat BCS % fat �Table 2. Continued. March 2015 December 2015 BCS % fat Rump fat BCS March 2016 Study Area Rump fat % fat Ryan Gulch 2.62 (0.95) 2.89 (0.40) 7.44 (0.53) 12.80 (6.83) 4.24 (0.88) 12.89 (3.72) 2.29 (0.64) 2.81 (0.37) 7.29 (0.52) South Magnolia 2.66 (1.36) 2.97 (0.55) 7.62 (0.74) 6.93 (4.83) 3.83 (0.89) 9.83 (2.69) 2.07 (1.39) 2.78 (0.39) 7.46 (0.93) North Magnolia 2.25 (0.97) 2.90 (0.42) 7.49 (0.90) 8.79 (6.01) 3.79 (0.85) 10.81 (3.54) 2.43 (1.01) 2.71 (0.39) 7.17 (0.87) North Ridge 2.28 (1.37) 2.92 (0.46) 7.43 (1.05) 5.47 (5.49) 3.25 (0.66) 1.58 (0.70) 2.51 (0.41) 6.73 (1.26) 9.35 (2.75) Table 2. Continued. December 2016 Study Area Rump fat Ryan Gulch 8.20 (4.90) 3.94 (0.97) 10.46 (2.70) 2.39 (0.74) 2.49 (0.38) 6.78 (0.97) South Magnolia 6.27 (4.62) 3.54 (0.88) 9.37 (2.53) 2.48 (0.77) 2.57 (0.35) 7.09 (0.63) North Magnolia 7.90 (5.52) 3.86 (1.01) 10.34 (3.14) 1.82 (0.72) 2.53 (0.27) 7.05 (0.58) North Ridge 7.74 (5.48) 3.85 (0.95) 10.01 (3.09) 2.30 (1.61) 2.74 (0.45) 7.23 (1.21) a BCS March 2017 % fat Rump fat BCS Body condition score taken from palpations of the rump following Cook et al. (2009). 17 % fat Rump fat BCS % fat �Table 3. Mark-resight abundance (N) and density estimates of mule deer from 4 winter range herd segments in the Piceance Basin, northwest Colorado, 27 March–4 April 2017. Data represent 4 helicopter resight surveys from all 4 study areas. Study area Mean No. sighted Mean No. marked N (95% CI) Density (deer/km2) Ryan Gulch 284 16 1,446 (1,184–1,792) 10.3 South Magnolia 171 14 991 (803–1,246) 11.9 North Magnolia 188 18 904 (757–1,098) 11.4 North Ridge 182 15 840 (695–1,040) 15.8 18 �deer study reas Well Pad & F•cllllles f 1n oeve10pmem 7 oe.ie10pm en1 · 1es 10 IIH Figure 1. Mule deer winter range study areas relative to active natural gas well pads and energy development facilities in the Piceance Basin of northwest Colorado, winter 2013/14 (Accessed http://cogcc.state.co.us/ Dec. 31, 2013). Development activity has subsided with no additional drilling since 2013. 19 �Winter fawn survival 2010-11 – 2016-17 1.00 fir 0.90 0.80 -- 0.70 0.60 0.50 .._ 0.40 0.30 0.20 0.10 --- --- I- I- 1 ,- - f 1 T T -- 1- i --I- 0.00 - f- - f- f- - □ Ryan Gulch □ South Magnolia • North Magnolia □ North Ridge f- -f- 2010-11 2011-12 2012-13 2013-14 2014-15 2015-16 2016-17 Figure 2. Over-winter (Dec–June) mule deer fawn survival (Ŝ) from 4 study areas in the Piceance Basin, northwest Colorado. Error bars = 95% CI. Winter survival rates from 2009-10 and 2010-11 are unavailable due to pre-mature collar drop, but 2009-10 and 2010-11 survival mirrored rates observed during 2011-12 and 2012-13 (excluding North Ridge), respectively, until collars began dropping during mid – late March of those years. 20 �Male fawn weights 42.0 Weight (kg) 40.0 38.0 ~ - 34.0 32.0 30.0 a Ryan Gulch - 36.0 28.0 - - - - = - == Dec 2008 Dec 2009 - - - - - - - = - == Dec 2010 Dec 2011 Dec 2012 a South Magnolia • North Magnolia a North Ridge =- Dec 2013 Dec 2014 Dec 2015 Dec 2016 Female fawn weights 42.0 Weight (kg) 40.0 38.0 a Ryan Gulch a South Magnolia 36.0 34.0 30.0 28.0 - - 32.0 == - - - - = , == Dec 2008 Dec 2009 Dec 2010 Dec 2011 - = , == = , == Dec 2012 Dec 2013 • - - North Magnolia - = Dec 2014 , == rtP Dec 2015 Dec 2016 I r a North Ridge Figure 3. Mean male and female fawn weights and 95% CI (error bars) from 4 mule deer study areas in the Piceance Basin, northwest Colorado, December 2008–2016. 21 �North Rief!,~ and ~Orth 1'fo(!!'Olia Summet f1"1g~ Ryan Gulc!i imd Sou\b M Swl1ther Ran_ge Figure 4. Mule deer study areas in the Piceance Basin of northwestern Colorado, USA (Top), spring 2009 migration routes of adult female mule deer (n = 52; Lower left), and active natural-gas well pads (black dots) and roads (state, county, and natural-gas; white lines) from May 2009 (Lower right; from Lendrum et al. 2012). 22 �Early winter rump fat 16 Rump fat (mm) 14 12 10 North Ridge 8 North Magnolia 6 Ryan Gulch 4 South Magnolia 2 0 Dec 2009 Dec 2010 Dec 2011 Dec 2012 Dec 2013 Dec 2014 Dec 2015 Dec 2016 Late winter rump fat 4.0 Rump fat (mm) 3.5 3.0 2.5 North Ridge 2.0 North Magnolia 1.5 Ryan Gulch 1.0 South Magnolia 0.5 0.0 Mar 2009 Mar 2010 Mar 2011 Mar 2012 Mar 2013 Mar 2014 Mar 2015 Mar 2016 Mar 2017 Figure 5. Mean early (early Dec., Top) and late winter (early Mar., Bottom) body condition (mm rump fat) of adult female mule deer from 4 winter range study areas in the Piceance Basin of northwest Colorado, March 2009–March 2017. Error bars = 95% CI. 23 �Piceance Basin late winter mule deer density 35.00 30.00 Deer/km2 25.00 20.00 North Ridge 15.00 Ryan Gulch North Magnolia 10.00 South Magnolia 5.00 0.00 2009 2010 2011 2012 2013 2014 2015 2016 2017 Year Figure 6. Mule deer density estimates and 95% CI (error bars) from 4 winter range herd segments in the Piceance Basin, northwest Colorado, late winter 2009–2017. Estimates for North Ridge 2014 and 2015 and for North Magnolia 2015 were adjusted upward (using GPS migration data) to account for early migration from winter range prior to and during surveys. 24 �eear5et_15_J b_Jndl.l aA_E C.... llearSet_36_ ., dSP't_.1!11l_gi9 er •l H • h Pilot T~ tmerm; 11 H 11Cresl ae sI Study Ar s Figure 7. Habitat treatment site delineations in 2 mule deer study areas (604 acres each) of the Piceance Basin, northwest Colorado (Top; cyan polygons completed Jan. 2011 using hydro-axe; yellow polygons completed Jan. 2012 using hydro-axe, roller-chop, and chaining; and remaining polygons completed April 2013 using hydro-axe). January 2011 hydro-axe treatment-site photos from North Hatch Gulch during April (Lower left, aerial view) and October, 2011 (Lower right, ground view). 25 �Appendix A. Abstracts of published manuscripts resulting from Piceance Basin mule deer/energy development interaction research collaborations. Abstract format specific to the respective journal requirements. Effectiveness of a redesigned vaginal implant transmitter in mule deer CHAD J. BISHOP1, CHARLES R. ANDERSON Jr. 1, DANIEL P. WALSH1, ERIC J. BERGMAN1, PETER KUECHLE2, and JOHN ROTH2 1 Colorado Parks and Wildlife, Fort Collins, Colorado 80526 USA 2 Advanced Telemetry Systems, Isanti, Minnesota 55040 USA Citation: Bishop, C. J., C. R. Anderson Jr., D. P. Walsh, E. J. Bergman, P. Kuechle, and J. Roth. 2011. Effectiveness of a redesigned vaginal implant transmitter in mule deer. Journal of Wildlife Management 75(8):1797-1806; DOI: 10.1002/jwmg.229 ABSTRACT Our understanding of factors that limit mule deer (Odocoileus hemionus) populations may be improved by evaluating neonatal survival as a function of dam characteristics under free-ranging conditions, which generally requires that both neonates and dams are radiocollared. The most viable technique facilitating capture of neonates from radiocollared adult females is use of vaginal implant transmitters (VITs). To date, VITs have allowed research opportunities that were not previously possible; however, VITs are often expelled from adult females prepartum, which limits their effectiveness. We redesigned an existing VIT manufactured by Advanced Telemetry Systems (ATS; Isanti, MN) by lengthening and widening wings used to retain the VIT in an adult female. Our objective was to increase VIT retention rates and thereby increase the likelihood of locating birth sites and newborn fawns. We placed the newly designed VITs in 59 adult female mule deer and evaluated the probability of retention to parturition and the probability of detecting newborn fawns. We also developed an equation for determining VIT sample size necessary to achieve a specified sample size of neonates. The probability of a VIT being retained until parturition was 0.766 (SE = 0.0605) and the probability of a VIT being retained to within 3 days of parturition was 0.894 (SE = 0.0441). In a similar study using the original VIT wings (Bishop et al. 2007), the probability of a VIT being retained until parturition was 0.447 (SE = 0.0468) and the probability of retention to within 3 days of parturition was 0.623 (SE = 0.0456). Thus, our design modification increased VIT retention to parturition by 0.319 (SE = 0.0765) and VIT retention to within 3 days of parturition by 0.271 (SE = 0.0634). Considering dams that retained VITs to within 3 days of parturition, the probability of detecting at least 1 neonate was 0.952 (SE = 0.0334) and the probability of detecting both fawns from twin litters was 0.588 (SE = 0.0827). We expended approximately 12 person-hours per detected neonate. As a guide for researchers planning future studies, we found that VIT sample size should approximately equal the targeted neonate sample size. Our study expands opportunities for conducting research that links adult female attributes to productivity and offspring survival in mule deer. © 2014 The Wildlife Society. Habitat selection by mule deer during migration: effects of landscape structure and natural-gas development PATRICK E. LENDRUM1, CHARLES R. ANDERSON JR.2, RYAN A. LONG1, JOHN G. KIE1, AND R. TERRY BOWYER1 1 Department of Biological Sciences, Idaho State University, Pocatello, Idaho 83209 USA 2 Colorado Parks and Wildlife, Grand Junction, Colorado 81505 USA Citation: Lendrum, P. E., C. R. Anderson Jr., R. A. Long, J. G. Kie, and R. T. Bowyer. 2012. Habitat selection by mule deer during migration: effects of landscape structure and natural-gas development. Ecosphere 3(9):82 http://dx.doi.org/10.1890/ES12-00165.1 Abstract. The disruption of traditional migratory routes by anthropogenic disturbances has shifted patterns of resource selection by many species, and in some instances has caused populations to decline. Moreover, in recent decades populations of mule deer (Odocoileus hemionus) have declined throughout much of their historic range in the western United States. We used resourceselection functions to determine if the presence of natural-gas development altered patterns of resource selection by migrating mule deer. We compared spring migration routes of adult female mule deer fitted with GPS collars (n = 167) among four study areas that had varying degrees of natural-gas development from 2008 to 2010 in the Piceance Basin of northwest Colorado, USA. Mule deer migrating through the most developed area had longer step lengths (straight-line distance between successive GPS locations) compared with deer in less developed areas. Additionally, deer migrating through the most developed study areas tended to select for habitat types that provided greater amounts of concealment cover, whereas deer from the least developed areas tended to select habitats that increased access to forage and cover. Deer selected habitats closer to well pads and avoided roads in all instances except along the most highly developed migratory routes, where road densities may have been too high for deer to avoid roads without deviating substantially from established migration routes. These results indicate that behavioral tendencies toward avoidance of anthropogenic disturbance can be overridden during migration by the strong fidelity ungulates demonstrate towards migration routes. If avoidance is feasible, then deer may select areas further from development, whereas in highly developed areas, deer may simply increase their rate of travel along established migration routes. 26 �Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes Patrick E. Lendrum1, Charles R. Anderson Jr.2, Kevin L. Monteith1,3, Jonathan A. Jenks4, R. Terry Bowyer1 1 Department of Biological Sciences, Idaho State University, Pocatello, Idaho, USA, 2 Colorado Division of Parks and Wildlife, Grand Junction, Colorado, USA, 3 Wyoming Cooperative Fish and Wildlife Research Unit, University of Wyoming, Laramie, Wyoming, USA,4 Department of Natural Resource Management, South Dakota State University, Brookings, South Dakota, USA Citation: Lendrum, P. E., C. R. Anderson Jr., K. L. Monteith, J. A. Jenks, R. T. Bowyer. 2013. Migrating Mule Deer: Effects of anthropogenically Altered Landscapes. PLoS ONE 8(5): e64548. DOI: 10.1371/journal.pone.0064548 Abstract Background: Migration is an adaptive strategy that enables animals to enhance resource availability and reduce risk of predation at a broad geographic scale. Ungulate migrations generally occur along traditional routes, many of which have been disrupted by anthropogenic disturbances. Spring migration in ungulates is of particular importance for conservation planning, because it is closely coupled with timing of parturition. The degree to which oil and gas development affects migratory patterns, and whether ungulate migration is sufficiently plastic to compensate for such changes, warrants additional study to better understand this critical conservation issue. Methodology/Principal Findings: We studied timing and synchrony of departure from winter range and arrival to summer range of female mule deer (Odocoileus hemionus) in northwestern Colorado, USA, which has one of the largest natural-gas reserves currently under development in North America. We hypothesized that in addition to local weather, plant phenology, and individual life-history characteristics, patterns of spring migration would be modified by disturbances associated with natural-gas extraction. We captured 205 adult female mule deer, equipped them with GPS collars, and observed patterns of spring migration during 2008–2010. Conclusions/Significance: Timing of spring migration was related to winter weather (particularly snow depth) and access to emerging vegetation, which varied among years, but was highly synchronous across study areas within years. Additionally, timing of migration was influenced by the collective effects of anthropogenic disturbance, rate of travel, distance traveled, and body condition of adult females. Rates of travel were more rapid over shorter migration distances in areas of high natural-gas development resulting in the delayed departure, but early arrival for females migrating in areas with high development compared with less-developed areas. Such shifts in behavior could have consequences for timing of arrival on birthing areas, especially where mule deer migrate over longer distances or for greater durations. Practical guidance on characterizing availability in resource selection functions under a use–availability design JOSEPH M. NORTHRUP1, MEVIN B. HOOTEN1,2,3, CHARLES R. ANDERSON JR.4, AND GEORGE WITTEMYER1 1 Department of Fish, Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 2 U.S. Geological Survey, Colorado Cooperative Fish and Wildlife Research Unit, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 3 Colorado State University, Department of Statistics, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 4 Mammals Research Section Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction, Colorado 81505 USA Citation: Northrup, J. M., M. B. Hooten, C. R. Anderson Jr., and G. Wittemyer. 2013. Practical guidance on characterizing availability in resource selection functions under a useavailability design. Ecology 94(7):1456-1463. http://dx.doi.org/10.1890/12-1688.1 Abstract. Habitat selection is a fundamental aspect of animal ecology, the understanding of which is critical to management and conservation. Global positioning system data from animals allow fine-scale assessments of habitat selection and typically are analyzed in a use–availability framework, whereby animal locations are contrasted with random locations (the availability sample). Although most use–availability methods are in fact spatial point process models, they often are fit using logistic regression. This framework offers numerous methodological challenges, for which the literature provides little guidance. Specifically, the size and spatial extent of the availability sample influences coefficient estimates potentially causing interpretational bias. We examined the influence of availability on statistical inference through simulations and analysis of serially correlated mule deer GPS data. Bias in estimates arose from incorrectly assessing and sampling the spatial extent of availability. Spatial autocorrelation in covariates, which is common for landscape characteristics, exacerbated the error in availability sampling leading to increased bias. These results have strong implications for habitat selection analyses using GPS data, which are increasingly prevalent in the literature. We recommend that researchers assess the sensitivity of their results to their availability sample and, where bias is likely, take care with interpretations and use cross validation to assess robustness. 27 �Effects of Helicopter Capture and Handling on Movement Behavior of Mule Deer JOSEPH M. NORTHRUP1, CHARLES R. ANDERSON JR2, AND GEORGE WITTEMYER1 1 Department of Fish, Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 2 Mammals Research Section Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction, Colorado 81505 USA Citation: Northrup, J. M., C. R. Anderson Jr., and G. Wittemyer. 2014. Effects of helicopter capture and handling on movement behavior of mule deer. Journal of Wildlife Management 78(4):731-738; DOI: 10.1002/jwmg.705 ABSTRACT Research on wildlife movement, physiology, and reproductive biology often requires capture and handling of animals. Such invasive treatment can alter behavior, which may bias results or invalidate assumptions regarding representative behaviors. To assess the impacts of handling on mule deer (Odocoileus hemionus), a focal species for research in North America, we investigated pre- and post-recapture movements of collared individuals, and compared them to deer that were not recaptured (controls). We compared pre- and post-recapture movement rates (m/hr) and 24-hour straight-line displacement among recaptured and control deer. In addition, we examined the time it took recaptured deer to return to their pre-recapture home range. Both daily straight-line displacement and movement rate were marginally elevated relative to monthly averages for 24 hours following recapture, with non-significant elevation continuing for up to 7 days. Comparing movements averaged over 30 days before and after recapture, we found no differences in displacement, but movement rates demonstrated seasonal effects, with faster movements post- relative to pre-recapture in March and slower movements post- relative to pre-recapture in December. Relative to control deer movements, recaptured deer movement rates in March were higher immediately after recapture and lower in the second and third weeks following recapture. The median time to return to the pre-recapture home range was 13 hours, with 71% of deer returning in the first day, and 91% returning within 4 days. These results indicate a short period of elevated movements following recaptures, likely due to the deer returning to their home ranges, followed by weaker but non-significant depression of movements for up to 3 weeks. Censoring of the first day of data post capture from analyses is strongly supported, and removing additional days until the individual returns to its home range will control for the majority of impacts from capture. © 2014 The Wildlife Society. Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns of forage quality Patrick E. Lendruma, Charles R. Anderson Jr.b, Kevin L. Monteithc, Jonathan A. Jenksd, R. Terry Bowyera a Department of Biological Sciences, Idaho State University, 921 South 8th Avenue, Stop 8007, Pocatello 83209, USA b Mammals Research Section Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction 81505, USA c Wyoming Cooperative Fish and Wildlife Research Unit, Department of Zoology and Physiology, University of Wyoming, 3166, 1000 East University Avenue, Laramie 82071, USA d Department of Natural Resource Management, South Dakota State University, Box 2140B, Brookings 57007, USA Citation: Lendrum, P. E., C. R. Anderson Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2014. Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns of forage quality. Mammalian Biology: http://dx.doi.org/10.1016/j.mambio.2014.05.005 ABSTRACT: Migratory ungulates exhibit recurring movements, often along traditional routes between seasonal ranges each spring and autumn, which allow them to track resources as they become available on the landscape. We examined the relationship between spring migration of mule deer (Odocoileus hemionus) and forage quality, as indexed by spatiotemporal patterns of fecal nitrogen and remotely sensed greenness of vegetation (Normalized Difference Vegetation Index; NDVI) in spring 2010 in the Piceance Basin of northwestern Colorado, USA. NDVI increased throughout spring, and was affected primarily by snow depth when snow was present, and temperature when snow was absent. Fecal nitrogen was lowest when deer were on winter range before migration, increased rapidly to an asymptote during migration, and remained relatively high when deer reached summer range. Values of fecal nitrogen corresponded with increasing NDVI during migration. Spring migration for mule deer provided a way for these large mammals to increase access to a high-quality diet, which was evident in patterns of NDVI and fecal nitrogen. Moreover, these deer “jumped” rather than “surfed” the green wave by arriving on summer range well before peak productivity of forage occurred. This rapid migration may aid in securing resources and seclusion from others on summer range in preparation for parturition, and to minimize detrimental factors such as predation, and malnutrition during migration. 28 �Effects of Male-Biased Harvest on Mule Deer: Implications for Rates of Pregnancy, Synchrony, and Timing of Parturition ERIC D. FREEMAN1, RANDY T. LARSEN1, MARK E. PETERSON2, CHARLES R. ANDERSON JR.3, KENT R. HERSEY4, AND BROCK R. McMILLAN1 1 Department of Plant and Wildlife Sciences, Brigham Young University, 275 WIDB, Provo, UT 84602, USA 2 Department of Fish, Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA 3 Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction, CO 81505, USA 4 Utah Division of Wildlife Resources, 1594 W North Temple, Salt Lake City, UT 84114, USA Citation: Freeman, E. D., R. T. Larsen, M. E. Peterson, C. R. Anderson Jr., K. R. Hersey, and B. R. McMillan. 2014. Effects of male-biased harvest on mule deer: implications for rates of pregnancy, synchrony, and timing of parturition. Wildlife Society Bulletin; DOI: 10.1002/wsb.450 ABSTRACT Evaluating how management practices influence the population dynamics of ungulates may enhance future management of these species. For example, in mule deer (Odocoileus hemionus), changes in male/female ratio due to malebiased harvest may alter rates of pregnancy, timing of parturition, and synchrony of parturition if inadequate numbers of males are present to fertilize females during their first estrous cycle. If rates of pregnancy or parturition are influenced by decreased male/female ratios, recruitment may be reduced (e.g., fewer births, later parturition resulting in lower survival of fawns, and a less synchronous parturition that potentially increases susceptibility of neonates to predation). Our objectives were to compare rates of pregnancy, synchrony of parturition, and timing of parturition between exploited mule deer populations with a relatively high (Piceance, CO, USA; 26 males/100 females) and a relatively low (Monroe, UT, USA; 14 males/100 females) male/female ratio. We determined rates of pregnancy via ultrasonography and timing of parturition via vaginal implant transmitters. We found no differences in rates of pregnancy (98.6% and 96.6%; z = 0.821; P = 0.794), timing of parturition (estimate = 1.258; SE = 1.672; t = 0.752; P = 0.454), or synchrony of parturition (F = 1.073; P = 0.859) between Monroe Mountain and Piceance Basin, respectively. The relatively low male/female ratio on Monroe Mountain was not associated with a protracted period of parturition. This finding suggests that relatively low male/female ratios typical of heavily harvested populations do not influence population dynamics because recruitment remains unaffected. © 2014 The Wildlife Society. Fine-scale genetic correlates to condition and migration in a wild cervid Joseph M. Northrup,1 Aaron B. A. Shafer,2 Charles R. Anderson Jr,3 David W. Coltman4 and George Wittemyer1 1 Department of Fish, Wildlife, and Conservation Biology, Colorado State University, Fort Collins, CO, USA 2 Department of Evolutionary Biology, Evolutionary Biology Centre, Uppsala University, Uppsala, Sweden 3 Mammals Research Section, Colorado Parks and Wildlife, Grand Junction, CO, USA 4 Department of Biological Sciences, University of Alberta, Edmonton, AB, Canada. Citation: Northrup, J. M., A. B. Shafer, C. R. Anderson Jr., D. W. Coltman, and G. Whittemyer. 2014. Fine-scale genetic correlates to condition and migration in a wild cervid. Evolutionary Applications ISSN 1752-4571; doi: 10.1111/eva.12189 Abstract The relationship between genetic variation and phenotypic traits is fundamental to the study and management of natural populations. Such relationships often are investigated by assessing correlations between phenotypic traits and heterozygosity or genetic differentiation. Using an extensive data set compiled from free ranging mule deer (Odocoileus hemionus), we combined genetic and ecological data to (i) examine correlations between genetic differentiation and migration timing, (ii) screen for mitochondrial haplotypes associated with migration timing, and (iii) test whether nuclear heterozygosity was associated with condition. Migration was related to genetic differentiation (more closely related individuals migrated closer in time) and mitochondrial haplogroup. Body fat was related to heterozygosity at two nuclear loci (with antagonistic patterns), one of which is situated near a known fat metabolism gene in mammals. Despite being focused on a widespread panmictic species, these findings revealed a link between genetic variation and important phenotypes at a fine scale. We hypothesize that these correlations are either the result of mixing refugial lineages or differential mitochondrial haplotypes influencing energetics. The maintenance of phenotypic diversity will be critical to enable the potential tracking of changing climatic conditions, and these correlates highlight the need to consider evolutionary mechanisms in management, even in widely distributed panmictic species. 29 �Landscape and anthropogenic features influence the use of auditory vigilance by mule deer Emma Lynch,a Joseph M. Northrup,b Megan F. McKenna,c Charles R. Anderson Jr,d Lisa Angeloni,a,e and George Wittemyera,b Graduate Degree Program in Ecology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA b Department of Fish, Wildlife and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA c Natural Sounds and Night Skies Division, National Park Service, 1201 Oakridge Drive, Fort Collins, CO 80525, USA, d Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526, USA e Department of Biology, Colorado State University, 1878 Campus Delivery, Fort Collins, CO 80523, USA a Citation: Lynch, E., J. M. Northrup, M. F. McKenna, C. R. Anderson Jr., L. Angeloni, and G. Wittemyer. 2014. Landscape and anthropogenic features influence the use of auditory vigilance by mule deer. Behavioral Ecology; doi:10.1093/beheco/aru158. While visual forms of vigilance behavior and their relationship with predation risk have been broadly examined, animals also employ other vigilance modalities such as auditory vigilance by listening for the acoustic cues of predators. Similar to the tradeoffs associated with visual vigilance, auditory behavior potentially structures the energy budgets and behavior of animals. The cryptic nature of auditory vigilance makes it difficult to study, but on-animal acoustical monitoring has rapidly advanced our ability to investigate behaviors and conditions related to sound. We utilized this technique to investigate the ways external stimuli in an active natural gas development field affect periodic pausing by mule deer (Odocoileus hemionus) within bouts of rumination-based mastication. To better understand the ecological properties that structure this behavior, we investigate spatial and temporal factors related to these pauses to determine if results are consistent with our hypothesis that pausing is used for auditory vigilance. We found that deer paused more when in forested cover and at night, where visual vigilance was likely to be less effective. Additionally, deer paused more in areas of moderate background sound levels, though responses to anthropogenic features were less clear. Our results suggest that pauses during rumination represent a form of auditory vigilance that is responsive to landscape variables. Further exploration of this behavior can facilitate a more holistic understanding of risk perception and the costs associated with vigilance behavior. Migration Patterns of Adult Female Mule Deer in Response to Energy Development Charles R. Anderson Jr. and Chad J. Bishop Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526, USA Citation: Anderson, C. R., Jr., and C. J. Bishop. 2014. Migration patterns of adult female mule deer in response to energy development. Pages 47-50 in Transactions of the 79th North American Wildlife & Natural Resources Conference (R. A. Coon & M. C. Dunfee, eds.). Wildlife Management Institute, Gardners, PA, USA. ISSN 0078-1355. Migration is an adaptive strategy that enables animals to enhance resource availability and reduce risk of predation at a broad geographic scale. Ungulate migrations generally occur along traditional routes, many of which have been disrupted by anthropogenic disturbances. Spring migration in ungulates is of particular importance for conservation planning because it is closely coupled with timing of parturition. The degree to which oil and gas development affects migratory patterns, and whether ungulate migration is sufficiently prepared to compensate for such changes, has recently been investigated in Colorado and Wyoming (Lendrum et al. 2012, 2013; Sawyer et al. 2012). Lendrum et al. (2012, 2013) and Sawyer et al. (2012) address mule deer (Odocoileus hemionus) migration patterns in relation to energy development from northwest Colorado and south-central Wyoming, respectively. We address results from the Colorado and Wyoming studies and then compare similarities and differences. The interactions between migratory mule deer and energy development identified by Lendrum et al. (2012, 2013) and Sawyer et al. (2012) suggest mule deer may benefit from energy development planning by considering thresholds of development that may alter migratory behavior. It appears that migration rate, migration routes, and stopover use, if present, may be altered at high development intensities. In addition, migratory mule deer may benefit by maintaining security cover along migration paths, and improved habitat conditions may facilitate more direct and rapid migration requiring less energy to complete migration. Enhancing permeability along migration routes by applying dispersed development plans (<2 well pads/km2) and minimizing disturbance to vegetation types by maintaining security cover should reduce impacts to migratory mule deer as well as other migratory ungulates. Where feasible, habitat improvement projects on winter range and possibly stopover sites would also enhance migratory mule deer populations by enhancing energy reserves for long-distance movements and parturition shortly after summer range arrival. Where possible, directional drilling could be used to extract energy resources from underneath migration routes while maintaining no surface occupancy. Lastly, we emphasize that GPS studies now allow managers to accurately map migration routes for entire populations and identify relatively narrow corridors that are most heavily used thus allowing for the identification of the most important corridors for migrating ungulates. Where available, we encourage agencies to incorporate such migration corridors into land-use plans (e.g., resource management plans) and National Environmental Policy Act documents. 30 �Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore Kate R. Searle1 · Mindy B. Rice2 · Charles R. Anderson2 · Chad Bishop2 · N. T. Hobbs3 NERC Centre for Ecology and Hydrology, Bush Estate, Penicuik EH26 0QB, UK 2 Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526, USA 3 Department of Ecosystem Science and Sustainability, Colorado State University, Fort Collins 80524, CO, USA 1 Citation: Searle, K. R., M. B. Rice, C. R. Anderson, C. Bishop and N. T. Hobbs. 2015. Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore. Oecologia ISSN 0029-8549; DOI 10.1007/s00442-015-3348-9 Abstract Understanding how spatial and temporal heterogeneity influence ecological processes forms a central challenge in ecology. Individual responses to heterogeneity shape population dynamics, therefore understanding these responses is central to sustainable population management. Emerging evidence has shown that herbivores track heterogeneity in nutritional quality of vegetation by responding to phenological differences in plants. We quantified the benefits mule deer (Odocoileus hemionus) accrue from accessing habitats with asynchronous plant phenology in northwest Colorado over 3 years. Our analysis examined both the direct physiological and indirect environmental effects of weather and vegetation phenology on mule deer winter body condition. We identified several important effects of annual weather patterns and topographical variables on vegetation phenology in the home ranges of mule deer. Crucially, temporal patterns of vegetation phenology were linked with differences in body condition, with deer tending to show poorer body condition in areas with less asynchronous vegetation green-up and later vegetation onset. The direct physiological effect of previous winter precipitation on mule deer body condition was much less important than the indirect effect mediated by vegetation phenology. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer JOSEPH M. NORTHRUP1 , CHARLES R. ANDERSON JR .2 and GEORGE WITTEMYER1 , 3 1 Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, CO, USA 2 Mammals Research Section, Colorado Parks and Wildlife, Fort Collins, CO, USA 3 Graduate Degree Program in Ecology, Colorado State University, Fort Collins, CO, USA Citation: Northrup, J. M., C. R. Anderson, Jr., and G. Wittemyer. 2015. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer. Global Change Biology, doi: 10.1111/gcb.13037 Abstract Extraction of oil and natural gas (hydrocarbons) from shale is increasing rapidly in North America, with documented impacts to native species and ecosystems. With shale oil and gas resources on nearly every continent, this development is set to become a major driver of global land-use change. It is increasingly critical to quantify spatial habitat loss driven by this development to implement effective mitigation strategies and develop habitat offsets. Habitat selection is a fundamental ecological process, influencing both individual fitness and population-level distribution on the landscape. Examinations of habitat selection provide a natural means for understanding spatial impacts. We examined the impact of natural gas development on habitat selection patterns of mule deer on their winter range in Colorado. We fit resource selection functions in a Bayesian hierarchical framework, with habitat availability defined using a movement-based modeling approach. Energy development drove considerable alterations to deer habitat selection patterns, with the most substantial impacts manifested as avoidance of well pads with active drilling to a distance of at least 800 m. Deer displayed more nuanced responses to other infrastructure, avoiding pads with active production and roads to a greater degree during the day than night. In aggregate, these responses equate to alteration of behavior by human development in over 50% of the critical winter range in our study area during the day and over 25% at night. Compared to other regions, the topographic and vegetative diversity in the study area appear to provide refugia that allow deer to behaviorally mediate some of the impacts of development. This study, and the methods we employed, provides a template for quantifying spatial take by industrial activities in natural areas and the results offer guidance for policy makers, mangers, and industry when attempting to mitigate habitat loss due to energy development. 31 �Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid Joseph M. Northrup1, Charles R. Anderson Jr2 and George Wittemyer1,3 1 Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, CO, USA 2 Mammals Research Section, Colorado Parks and Wildlife, Fort Collins, CO, USA 3 Graduate Degree Program in Ecology, Colorado State University, Fort Collins, CO, USA Citation: Northrup, J. M., C. R. Anderson, Jr., and G. Wittemyer. 2016. Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid. Diversity and Distributions 22: 547-557, DOI: 10.1111/ddi.12417 ABSTRACT Aim The space an animal uses over a given time period must provide the resources required for meeting energetic needs, reproducing and avoiding predation. Anthropogenic landscape change in concert with environmental dynamics can strongly structure space-use. Investigating these dynamics can provide critical insight into animal ecology, conservation and management. Location The Piceance Basin, Colorado, USA. Methods We applied a novel utilization distribution estimation technique based on a continuous-time correlated random walk model to characterize range dynamics of mule deer during winter and summer seasons across multiple years. This approach leverages secondorder properties of movement to provide a probabilistic estimate of space-use. We assessed the influence of environmental (cover and forage), individual and anthropogenic factors on interannual variation in range use of individual deer using a hierarchical Bayesian regression framework. Results Mule deer demonstrated remarkable spatial philopatry, with a median of 50% overlap (range: 8–78%) in year-to-year utilization distributions. Environmental conditions were the primary driver of both philopatry and range size, with anthropogenic disturbance playing a secondary role. Main conclusions Philopatry in mule deer is suspected to reflect the importance of spatial familiarity (memory) to this species and, therefore, factors driving spatial displacement are of conservation concern. The interaction between range behaviour and dynamics in development disturbance and environmental conditions highlights mechanisms by which anthropogenic environmental change may displace deer from familiar areas and alter their foraging and survival strategies. Movement reveals scale dependence in habitat selection of a large ungulate Joseph M. Northrup,1 Charles R. Anderson Jr.,2 Mevin B. Hooten,3 and George Wittemyer4 1 Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA 2 Mammals Research Section, Colorado Parks and Wildlife, Fort Collins, Colorado 80523 USA 3 U.S. Geological Survey, Colorado Cooperative Fish and Wildlife Research Unit, Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA 4 Department of Fish, Wildlife and Conservation Biology and Graduate Degree Program in Ecology, Colorado State University, Fort Collins, Colorado 80523 USA Citation: Northrup, J. M., C. R. Anderson, Jr., M. B. Hooten, and G. Wittemyer. 2016. Movement reveals scale dependence in habitat selection of a large ungulate. Ecological Applications 26:2746-2757 Abstract. Ecological processes operate across temporal and spatial scales. Anthropogenic disturbances impact these processes, but examinations of scale dependence in impacts are infrequent. Such examinations can provide important insight to wildlife–human interactions and guide management efforts to reduce impacts. We assessed spatiotemporal scale dependence in habitat selection of mule deer (Odocoileus hemionus) in the Piceance Basin of Colorado, USA, an area of ongoing natural gas development. We employed a newly developed animal movement method to assess habitat selection across scales defined using animal-centric spatiotemporal definitions ranging from the local (defined from five hour movements) to the broad (defined from weekly movements). We extended our analysis to examine variation in scale dependence between night and day and assess functional responses in habitat selection patterns relative to the density of anthropogenic features. Mule deer displayed scale invariance in the direction of their response to energy development features, avoiding well pads and the areas closest to roads at all scales, though with increasing strength of avoidance at coarser scales. Deer displayed scale-dependent responses to most other habitat features, including land cover type and habitat edges. Selection differed between night and day at the finest scales, but homogenized as scale increased. Deer displayed functional responses to development, with deer inhabiting the least developed ranges more strongly avoiding development relative to those with more development in their ranges. Energy development was a primary driver of habitat selection patterns in mule deer, structuring their behaviors across all scales examined. Stronger avoidance at coarser scales suggests that deer behaviorally mediated their interaction with development, but only to a degree. At higher development densities than seen in this area, such mediation may not be possible and thus maintenance of sufficient habitat with lower development densities will be a critical best management practice as development expands globally. 32 �Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus) Kourtney F. Stonehouse, ,1,2 Charles R. Anderson Jr.,1 Mark E. Peterson,1,2 and David R. Collins1 1 Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 90526 USA 2 Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA Citation: Stonehouse, K. F., C. R. Anderson Jr., M. E. Peterson, and D. R. Collins. 2016. Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus). Colorado Parks and Wildlife Technical Report No. 48, First Edition, 317 W. Prospect Rd., Ft. Collins, CO USA. DOW-R-T-48-16, ISSN 0084-8883. This technical report provides general guidelines for conducting mortality site investigations to help investigators distinguish predation from scavenging and other causes of death. General health indices are also provided to assess whether or not deer may have died from malnutrition or disease or if these factors may have predisposed deer to predation. Lastly, these guidelines will assist investigators in identifying predatory species or scavengers involved through the examination of physical evidence at deer mortality sites. The information presented here is based primarily on field experience gained from a long term research effort in northwest Colorado investigating mule deer mortality sites over several years (http://cpw.state.co.us/learn/Pages/ResearchMammalsRP-04.aspx) and literature review where referenced. We acknowledge that proximate and ultimate cause of death can be difficult or impossible to detect from field necropsy alone and examples presented here largely represent proximate causes of mortality; efforts discerning ultimate cause will require specific tissue sample collections, where possible, submitted to a veterinary diagnostic laboratory. Within this technical report are numerous photographs documenting characteristics of predator attacks on mule deer and signs left by predatory and scavenging species. Additional pictures illustrate differences between healthy and unhealthy tissues and organs. While reading this document, be aware that each mortality investigation is unique and observations in the field may differ from illustrations provided here. Appendix I provides a sample necropsy form to assist in conducting mortality investigations. 33 �Colorado Parks and Wildlife July 1, 2017 - June 30, 2018 WILDLIFE RESEARCH REPORT State of __________ C__o=Io__ra=d=o________ : .....P=ar:.:.:ks:-=a=nd=-a..Wa...::i=ld=li=fe;a....__ _ _ _ _ _ _ _ _ _ __ Cost Center 3430 : __M=a__m__m=a=I__s __R=e__ se=ar__c__h______________ Work Package 3001 : =D.....ee.....r__C __o__n=s__erva.. . .,.; .;a=ti__ on_ _ _ _ _ _ _ _ _ _ _ __ Task No. 6 : Population Performance of Piceance Basin Mule Deer in Response to Natural Gas Resource Extraction and Mitigation Efforts to Address Human Activity and Habitat Degradation Federal Aid Project: ________ W___-2___4___3___ -R3 __________ Period Covered: July 1, 2017 - June 30, 2018 Author: C. R. Anderson, Jr. Personnel: E. Bergman, D. Collins, B. deVergie, D. Finley, M. Fisher, L. Gepfert, D. Johnston, T. Knowles, B. Petch, J. Rivale, G. Samsill, E. Sawa, G. Smith, R. Velarde, S. Williams, L. Wolfe, CPW; L. Belmonte, BLM; T. Graham, Ranch Advisory Partners; P. Doherty, J. Northrup, M. Peterson, G. Wittemyer, K. Wilson, Colorado State University; R. Swisher, S. Swisher, Quicksilver Air, Inc.; D. Felix, Olathe Spray Service, Inc.; L. Coulter, Coulter Aviation. Project support received from Federal Aid in Wildlife Restoration, Colorado Mule Deer Association, Colorado Mule Deer Foundation, Muley Fanatic Foundation, Colorado State Severance Tax Fund, EnCana Corp., ExxonMobil Production Co./XTO Energy, Marathon Oil Corp., Shell Petroleum, and WPX Energy. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the authors. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We propose to experimentally evaluate winter range habitat treatments and human-activity management alternatives intended to enhance mule deer (Odocoileus hemionus) populations exposed to energy-development activities. The Piceance Basin of northwestern Colorado was selected as the project area due to ongoing natural gas development in one of the most extensive and important mule deer winter and transition range areas in Colorado. The data presented here represent preliminary and final results of a 10-year research project addressing habitat improvements and evaluation of energy development practices intended to improve mule deer fitness in areas exposed to extensive energy development. We monitored deer on 4 winter range study areas representing varying levels of development to serve as treatment (North Magnolia, South Magnolia) and control (North Ridge, Ryan Gulch) sites. We recorded habitat use and movement patterns, estimated neonatal, overwinter fawn and annual adult female survival, estimated early and late winter body condition of adult females, and estimated annual abundance among study areas. During this research segment, we targeted 240 fawns (60/study area) and 120 does (30/study area) in early December 2017 for VHF and GPS radiocollar attachment, respectively, and adult female body condition assessment. We attempted recapture of 120 does (30/study area) and 40 fawns (20 in 2 study areas) in March 2018 for late winter body condition assessment. Winter range habitat improvements completed spring 2013 resulted in 604 acres of mechanically treated pinionjuniper/mountain shrub habitats in each of the 2 treatment areas with minor and extensive energy I �development, respectively. Based on final (migration, mule deer behavioral responses, reproductive success and neonate survival) and preliminary data analyses for this 10-year project: ( 1) annual adult female survival was consistent among areas averaging 79-87% annually, but overwinter fawn survival was variable, ranging from 31 % to 95% within study areas, with annual and study area differences primarily due to early winter fawn condition (Dec fawn mass), annual weather conditions, and factors associated with predation on winter range; (2) mule deer body condition early and late winter was generally consistent within areas, with higher variability among study areas early winter, primarily due to December lactation rates, and late winter condition related to seasonal moisture and winter severity; (3) late winter mule deer densities increased through 2016 in all study areas, ranging from a 50% increase in North Ridge to a 103% increase in North Magnolia, however densities have stabilized recently in 3 of the 4 study areas and a recent decline in density was evident in North Ridge; (4) migratory mule deer selected for areas with increased cover and increased their rate of travel through developed areas, and avoided negative influences through behavioral shifts in timing and rate of migration, but did not avoid development structures; (5) mule deer exhibited behavioral plasticity in relation to energy development, where disturbance distance varied relative to diurnal extent and magnitude of development activity, which may provide for several options in future development planning; and (6) energy development activity under existing conditions did not influence pregnancy rates, fetal rates or early fawn survival (0-6 months), but may have reduced neonatal survival (March until birth) when drought conditions persisted during the third trimester of doe parturition. Final results are pending to address vegetation and mule deer responses to assess habitat treatment mitigation options for energy development planning, and final results addressing the interaction of mule deer behavioral and demographic factors associated with energy development activity have recently been submitted for scientific review and publication. Completion of this project, including final data collection, analyses and interpretation of results, is anticipated by fall/winter 2020. u 2 �WILDLIFE RESEARCH REPORT POPULATION PERFORMANCE OF PICEANCE BASIN MULE DEER IN RESPONSE TO NATURAL GAS RESOURCE EXTRACTION AND MITIGATION EFFORTS TO ADDRESS HUMAN ACTMTY AND HABITAT DEGRADATION CHARLES R. ANDERSON, JR PROJECT NARRITIVE OBJECTIVES I. To determine experimentally whether enhancing mule deer habitat conditions on winter range elicits behavioral responses, improves body condition, increases fawn survival, and ultimately, population density on mule deer winter ranges exposed to extensive energy development. 2. To determine experimentally to what extent modification of energy development practices enhance habitat selection, body condition, fawn survival, and winter range mule deer densities. SEGMENT OBJECTIVES I. Collect and reattach OPS collars to maintain sample sizes for addressing mule deer habitat use and behavioral patterns in 4 study areas experiencing varying levels of energy development in the Piceance Basin, northwest Colorado. 2. Estimate early and late winter body condition of adult female mule deer in each of the 4 winter herd segments using ultrasound techniques. Estimate early and late winter fawn weights in areas with and without habitat treatments to assess winter fawn condition relative to habitat improvements. 3. Monitor over-winter fawn and annual adult female mule deer survival by daily ground tracking and bi-weekly aerial tracking. 4. Conduct Mark-Resight helicopter surveys to estimate late winter mule deer abundance and density in each study area. 5. Monitor habitat treatment response for assessing efficacy of habitat improvement projects to mitigate energy development disturbances to mule deer. 6. Complete investigations of neonate fawn survival and adult female parturition relationships relative to mule deer/energy development interactions. INTRODUCTION Extraction of natural gas from areas throughout western Colorado has raised concerns among many public stakeholders and Colorado Parks and Wildlife (CPW) that the cumulative impacts associated with this intense industrialization will dramatically and negatively affect the wildlife resources of the region. Concern is especially high for mule deer due to their recreational and economic importance as a principal game species and their ecological importance as one of the 3 �primary herbivores of the Colorado Plateau Ecoregion. Extraction of natural gas will directly affect the potential suitability of the landscape used by mule deer through conversion of native habitat vegetation with drill pads, roads, or introduction of noxious weeds, by fragmenting habitat with drill pads and roads, by increasing noise levels via compressor stations and vehicle traffic, and by increasing the year-round presence of human activities. Extraction will indirectly affect deer by increasing the human work-force population of the region resulting in the need for additional landscape conversion for human housing, supporting businesses, and upgraded road/transportation infrastructure. Additionally, increased traffic on rural roads will raise the potential for vehicle-animal collisions. Thus, research documenting these relationships and evaluating the most effective strategies for minimizing and mitigating these activities will greatly enhance future management efforts to sustain mule deer populations for future recreational and ecological values. The Piceance Basin in northwest Colorado contains one of the largest migratory mule deer populations in North America and also covers some of the largest natural gas reserves in North America Projected energy development throughout northwest Colorado within the next 20 years is expected to reach about 15,000 wells, many of which will occur in the Piceance Basin, which currently supports over 250 active gas well pads (http://cogcc.state.co.us; Fig. I). Anderson and Freddy (2008) in their long-term research proposal identified 6 primary study objectives to assess measures to offset impacts of energy extraction on mule deer population performance. During the first 5 years of this study, we gathered baseline habitat utilization and demographic data from radiocollared deer across the Piceance Basin to allow assessment of habitat mitigation approaches that were completed April 2013. We recently completed monitoring 2 control areas: I with development (0.6 pads & facilities/km 2; Ryan Gulch) and I without (North Ridge). The control areas will be compared with 2 treatment areas experiencing similar development intensities (South Magnolia, 0.9 well pads & facilities/km 2 and North Magnolia, 0.1 well pads & facilities/km 2), that also received habitat improvements (604 acres each). Habitat and mule deer responses to mechanical habitat treatments will be evaluated through spring 2018 to assess the success of this habitat mitigation strategy to benefit mule deer exposed to energy development disturbance. In addition, mule deer behavioral patterns in relation to energy development activities in the area are being monitored to identify effective Best Management Practices (BMPs) for future energy development planning. This progress report describes the previous IO years (Jan 2008-June 2018) of mule deer population performance during the pre and post-treatment phases on 4 winter range herd segments, which includes monitoring habitat selection, migration and behavior patterns of adult female mule deer; parturition success; spring/summer neonate, overwinter fawn and annual adult female survival; estimates of adult female body condition during early and late winter; and annual late-winter abundance/density estimates. STUDY AREAS The Piceance Basin, located between the cities of Rangely, Meeker, and Rifle in northwest Colorado, was selected as the project area due to its ecological importance as home to one of the largest migratory mule deer populations in North America and because it exhibits one of the highest natural gas reserves in North America (Fig. I). Historically, mule deer numbers on winter range were estimated between 20,000-30,000 (White and Lubow 2002), and the current number of well pads (Fig. I) and projected number of gas wells in the Piceance Basin over the next 20 years is about 250 and 15,000, respectively. Mule deer winter range in the Piceance Basin is predominantly characterized as a topographically diverse pinion pine (Pinus edulis)-Utahjuniper (Juniperus osteosperma; pinionjuniper) shrubland complex ranging from 1,675 m to 2,285 m in elevation (Bartmann and Steinert 1981). Pinion-juniper are the dominant overstory species and major shrub species include Utah serviceberry (Amelanchier utahensis), mountain mahogany (Cercocarpus montanus), bitterbrush (Purshia tridentata), big sagebrush (Artemisia tridentata), Gamble's oak (Quercus gambelii), mountain snowberry (Symphoricarpos oreophilus), and rabbitbrush (Chrysothamnus spp.; Bartmann et al. 1992). The Piceance Basin is segmented by numerous drainages characterized by stands of big 4 u �sagebrush, saltbush (Atriplex spp.), and black greasewood (Sarcobatus vermiculatus), with the majority of the primary drainages having been converted to mixed-grass hay fields. Grasses and forbs common to the area consist ofwheatgrass (Agropyron spp.), blue grama (Bouteloua gracilis), needle and thread (Stipa comata), Indian rice grass (Oryzopsis hymenoides), arrowleafbalsamroot (Balsamorhiza sagittata), broom snakeweed (Gutierrezia sarothreae), pinnate tansymustard (Descurainia pinnata), milkvetch (Astragalus spp.), Lewis flax (Linum /ewisii), evening primrose (Oenothera spp.), skyrocket gilia (Gilia aggregata), buckwheat (Erigonum spp.), Indian paintbrush (Castilleja spp.), and penstemon (Penstemon spp.; Gibbs 1978). The climate of the Piceance Basin is characterized by warm dry summers and cold winters with most of the annual moisture resulting from spring snow melt and brief summer monsoonal rain storms. Wintering mule deer population segments we are investigating include: North Ridge (53 km 2) just north of the Dry Fork of Piceance Creek including the White River in the northeastern portion of the Basin, Ryan Gulch ( 141 km2) between Ryan Gulch and Dry Gulch in the southwestern portion of the Basin, North Magnolia (79 km2) between the Dry Fork of Piceance Creek and Lee Gulch in the north-central portion of the Basin, and South Magnolia (83 km2) between Lee Gulch and Piceance Creek in the south-central portion of the Basin (Fig. 1). Each of these wintering population segments has received varying levels of natural gas development: no development in North Ridge, light development in North Magnolia (0.1 pads & facilities/knl), and relatively high development in the Ryan Gulch (0.6 pads & facilities/km 2) and South Magnolia (0.9 pads & facilities/km 2) segments (Fig. I). Development activity was high through 2011 and has declined substantially since natural gas prices began to decline in 2012. Among the 4 study areas, North Ridge has served as an unmanipulated control site, Ryan Gulch will serve to address human-activity management alternatives (BMPs) that benefit mule deer exposed to energy development and as a developed control area for comparison to the developed treatment area receiving habitat improvements (South Magnolia), and North and South Magnolia will allow us to assess the utility of habitat treatments intended to enhance mule deer population performance in areas exposed to light (North Magnolia) and relatively heavy (South Magnolia) energy development activities. METHODS Tasks addressed this period included mule deer capture and collaring, monitoring neonate, overwinter fawn and annual adult female survival, estimating adult female body condition during early and late winter using ultrasonography and winter fawn condition measuring early and late winter fawn weights, estimating mule deer abundance applying helicopter mark-resight surveys, and monitoring vegetation responses to habitat treatments completed spring 2013. We employed helicopter netgunning techniques (Barrett et al. 1982, van Reenen 1982) to target 240 fawns and 120 adult females during early December 2016, and 120 adult females and 40 fawns (primarily recaptures) during early March 2017. Once netted, all deer were hobbled and blind folded. Fawns were weighed and radiocollared, and sex was recorded prior to release at the capture site. Adult females were transported to localized handling sites for recording body measurements and fitted with GPS collars (5 fix attempts/day; 0211 OD, Advanced Telemetry Systems, Isanti, MN, USA) prior to release. To provide direct measures of decline in overwinter body condition, we targeted 30 adult females in each study area that were captured the previous December. During March, 20 fawns were recaptured, weighed and released in South Magnolia (in the habitat treatment areas) and Ryan Gulch (control area) to quantify overwinter declines in fawn body condition. Fawn collars were spliced and fitted with rubber surgical tubing to facilitate collar drop between mid-summer and autumn for winter fawns and during winter for neonates, and GPS collars were supplied with timed drop-off mechanisms scheduled to release early April of the year following deployment. All radio-collars were equipped with mortality sensing options (i.e., increased pulse rate following 8 hrs of inactivity). 5 �Mule Deer Habitat Use and Movements We downloaded and summarized data from GPS collars deployed and recovered since 2008. GPS collars maintained the same schedule of attempting to collect locations every 5 hours, except for 40 does in Ryan Gulch and IO control deer from North Ridge where location rates were programmed for every 30-60 minutes to increase resolution of movement data for evaluation of deer behavior patterns in relation to differing development activities. Joe Northrup (CSU PhD Candidate) recently analyzed resource selection data relative to energy development (Northrup 2015) and those results are addressed below. Mule deer resource selection analyses to address success of habitat improvements are pending until vegetation responses are fully realized, which will begin by fall 2018. Mule Deer Survival Mule deer mortality monitoring consisted of daily ground-telemetry tracking and aerial monitoring approximately every 2 weeks from fixed-wing aircraft on winter range and weekly aerial monitoring on summer range. Once a mortality signal was detected, deer were located and necropsied to assess cause of death (Stonehouse et al. 2016). We estimated weekly survival using the staggered entry Kaplan-Meier procedure (Kaplan and Meier 1958, Pollock et al. 1989). Capture-related mortalities (any doe/fawn mortalities occurring within 10 days of capture; excluding neonates) and collar failures were censored from survival rate estimates. We estimated survival rates from 1 July 2016 through 30 June 2017 for adult females, from birth to mid-December for neonates, and from early December 2016-mid June 2017 for winter fawns. Adult Female Body Measurements We applied ultrasonography techniques described by Stephenson et al. (1998, 2002) and Cook et al. (2001) to measure maximum subcutaneous rump fat (mm), loin depth (longissimus dorsi muscle, mm), and to estimate% ingesta-free body fat. We estimated a body condition score (BCS) for each deer by palpating the rump (Cook et al. 2001, 2007, 2009). We examined differences (P < 0.05) in nutritional status among study areas and between years evident in non-overlapping 95% confidence intervals. We considered differences in body condition meaningful when mean rump fat or % body fat differed statistically between comparisons. Other body measurements recorded included pregnancy status (pregnant, barren) via blood samples, fetal counts using ultrasonagraphy, weight (kg), chest girth (cm), and hind-foot length (cm). Abundance Estimates We conducted 4 helicopter mark-resight surveys (2 observers and the pilot) during late March/early April to estimate deer abundance in all 4 study areas. We delineated each study area from GPS locations collected on winter range during the first 3 years of the study (Jan 2008 through April 2011 ). Two aerial fixed-wing telemetry surveys/study area were conducted during helicopter mark-resight surveys to determine which marked deer were within each survey area, and we confirmed adult female locations during surveys from GPS data acquired April 2017. We delineated flight paths in ArcGIS 10.0 prior to surveys following topographic contours (e.g., drainages, ridges) and approximating 500-600 m spacing throughout each study area; flight paths during surveys were followed using GPS navigation in the helicopter. Two 12 x 12 cm pieces of Ritchey livestock banding material (Ritchey Livestock ID, Brighton, CO USA) were uniquely marked using color, number, and symbol combinations and attached to each radio-collar to enhance mark-resight estimates. Each deer observed during surveys was recorded as mark ID#, unmarked, or unidentified mark. 6 �We used program MARK (White and Burnham 1999), applying the immigration-emigration mixed logit-normal model (McClintock et al. 2008), to estimate mule deer abundance and confidence intervals. For mark-resight model evaluations, we examined parameter combinations of varying detection rates with survey occasion and whether individual sighting probabilities (i.e., individual heterogeneity) were constant or varied (cr2 = 0 or :;c 0). Model selection procedures followed the information-theoretic approach of Burnham and Anderson (2002). RESULTS AND DISCUSSION Deer Captures and Survival The helicopter crew captured 237 fawns and 121 does during Dec 2017 and 103 does and 40 fawns during March 2018. Fifteen fawn mortalities (6.3%; proximate cause= 4 capture myopathy, 10 predation, 1 vehicle collision) occurred within the 10-day censorship period during December and 3 fawn mortalities (7.5%; 1 capture myopathy, I predation, 1 fence entanglement) occurred during the March capture. Doe mortalities totaled 2 (1.7%; capture myopathy) and 4 (3.9%; 3 capture myopathy, 1 predation) within IO days of the December and March capture periods, respectively. Mortality rates within 10 days post capture have typically varied between 2.5-3.5% for fawns and does since Jan 2008, except during the 2011-2012 capture season where myopathy rates were higher (3--6%) due to dry, warm conditions (Anderson and Bishop 2012) and 2016-17 (6.6% for Dec fawns) when reduced fawn condition may have enhanced coyote predation success. The relatively high myopathy rates (including ongoing predation that occurred within the 10-day censorship period) for early winter fawn captures during 2017-2018 were similar to last year with lighter December fawn weights and high coyote predation. Relatively higher fawn myopathy rates were expected during late winter captures and have ranged from 4.9 to 10.0% the past 3 years. Fawn survival from early December 2017 through mid-June 2018 was more variable among study areas than in past years ranging from 0.34 to 0. 76 (Table l, Fig. 2). Based on CI overlap, North Ridge and North Magnolia fawns exhibited higher survival than Ryan Gulch fawns, and North Ridge fawn survival was also higher than in South Magnolia (Table 1). The range in winter fawn survival was unusual in comparison to previous years (Fig. 2), but correlated closely with the variability observed in December fawn weights (Fig. 3); early winter fawn condition (as reflected in Dec fawn mass) likely contributes to over-winter survival potential, and reduced weights the past 2 years is likely related to the lower survival rates observed recently from Ryan Gulch and South Magnolia (Fig. 2, Fig. 3). Premature collar drop during 2008-09 and 2009- IO did not allow for winter fawn survival estimates past March, but survival rates among study areas were similar (P < 0.05) each year and comparable to 2011-12 and 2012-13 (excluding North Ridge) during 2008-09 and to the higher survival rates from 2013-14 and 2014-15 during 2009-10 (Fig. 2). General comparisons to previous years suggest moderate to high fawn survival occurred during most winters and study areas with the exception of winter 2010-2011 for 3 of the 4 study areas, North Ridge during winter 2012-13 and 2015-16, and Ryan Gulch and South Magnolia the past 2 years (Fig. 2). Low winter fawn survival (Fig. 2) may to correlate with summer forage condition assuming lower December fawn weights (Fig. 3) represent an index of summer fall forage conditions; severe winter conditions can also strongly influence winter fawn survival, but winter conditions during this study have been mild to moderate with the exception of winter 2010-11, which may have more strongly influenced winter fawn survival that year. Annual adult female survival varied from 0.73 (North Ridge) to 0.87 (South Magnolia; Table 1) during 2017-18, but was comparable among study areas and to previous years (P > 0.05), with the exception oflower survival in North Magnolia during 2011-12 (S = 0.68, Anderson and Bishop 2012). Relatively low sample sizes per study area for adult female survival do not allow statistical 7 �discrimination among years unless large differences are evident (e.g.,> 15-20%). Estimates below 80% are biologically concerning if these values represent the respective population, but low statistical power precludes confirmation within study areas. When combined among study areas, annual survival estimates have varied from 79% in 2012-13 to 86% in 2014-15, but consistent CI overlap including large sample sizes (exceeding 100 in July and 200 in Mar annually) supports consistent annual doe survival during this study. Adult female mule deer exhibit consistently high survival rates unless extreme weather events and/or habitat degradation persists, which has not been evident since 2008. Mule Deer Body Condition Early-winter body condition measurements of adult female mule deer during December 2017 were relatively low among study areas compared to previous years and Ryan Gulch does exhibited the lowest condition estimates among study areas (P < 0.05; Fig. 4, Table 2). Although fall body condition is likely related to spring/summer forage conditions, doe condition is also influenced by energy expended for fawn rearing, and appears to be strongly influenced by lactation status; I observed a strong correlation between December lactation rate and body condition (mm rump fat; P = 0.004, r = 0.62, n = 20). Thus, while fall body condition represents an index of nutritional status entering winter, it also appears to be a useful metric to assess fall reproductive status, where low fat levels represent high fall fawning rates; the low December fat levels observed from Ryan Gulch does during 2017 was associated with highest fall lactation rate (0.63) recorded during the study. In contrast, late winter condition appears more strongly related to winter severity and winter range forage conditions and low fat levels observed during December do not necessarily manifest into poor late winter condition (Fig. 4, Table 2). Late winter doe condition among study areas this past winter was comparable to the long term average (Fig. 4), and was reflective of mild to moderate winter conditions consisting of infrequent snow storms and minimal snow pack on winter range. December 2017 fawn weights by study area represented a gradient in fawn condition ranging from low to high for Ryan Gulch and North Ridge, respectively (Fig. 3), which corresponded to winter fawn survival (Fig. 2). Overall fawn condition for 3 of 4 study areas (excluding North Ridge) has declined the past 2 years (Fig. 3) and may be related to changes in summer forage conditions (further analyses pending). Because adult female body condition has been largely uninformative in regards to habitat treatment responses (pending further analyses), we began late winter fawn recaptures in South Magnolia (habitat treatment area) and Ryan Gulch (reference area) to assess changes in over-winter condition. Weight loss during winter 2015-16 was significantly less (P < 0.001) for fawns from the area receiving habitat treatments than for fawns from the untreated area, but no net weight loss was detected during the following 2 winters for either study area (P ~ 0.396), suggesting strong annual effects. Vegetation measurements from treatment and control sites indicate recent summer/fall use of shrubs potentially negating forage benefits on winter range. Additional investigations to address this issue will be conducted summer/fall 2018 to confirm summer/fall use of treatment sites and whether or not intended forage benefits on habitat treatment sites persist on winter range. Mule Deer Population Estimates Mark-resight models that best predicted abundance estimates (lowest AICc; Burnham and Anderson 2002) exhibited variable sightability across surveys (P,) for all study areas and variable individual sightability (cr2 = 0) for North Magnolia deer and homogenous sightability (cr2 f. 0) for the other 3 areas. During 2018 North Ridge exhibited the highest deer density ( l 5.8/km2), with comparable but lower deer densities in the other 3 areas (9.2-l l .3/k.m2; Table 3, Fig. 5). Abundance estimates from 2018 were similarly precise from all 4 study areas with the mean Confidence Interval Coefficient of Variation (CICV) ranging from 0.12-0.17 (Table 3). Densities increased over the first 8- 8 u �year monitoring period in all study areas ranging from an estimated 50% increase in North Ridge to a 103% increase in North Magnolia (mean estimated increase across study areas= 78%); North Ridge deer appeared to decline during 2012 and 2013, but subsequently increased, while the other 3 areas exhibited consistent and similar rates ofincrease from 2009-2016 (mean annual increase= 0.064; Fig. 5). Excluding the North Ridge study area, late winter mule deer densities have apparently stabilized since 2016 (Fig. 5). The reason for decline since 2016 for the North Ridge deer population is unclear and not completely explained by demographic parameters monitored during the study. Erratic population estimates observed from North Ridge may be partially attributed to lack of geographic closure more commonly associated with this study area (primarily from earlier spring migration timing). Population vital rates will be analyzed and compared to abundance estimates to assess factors contributing to population change by study area. Spring Migration Patterns Collaboration with Idaho State University to address mule deer migration patterns in developed and undeveloped landscapes (funded from energy company contributions) has been completed. Four manuscripts from this effort have been published (Lendrum et al. 2012, Lendrum et al. 2013, Lendrum et al. 2014, Anderson and Bishop 2014; Appendix A). In addressing habitat selection during spring migration, Lendrum et al. (2012; Fig. 6) noted that mule deer migrating through the most developed landscapes exhibited longer step lengths (straight line distance between GPS locations) and selected habitats providing greater security cover than deer in undeveloped landscapes that migrated through more open areas that provided increased foraging opportunities. Migrating deer also selected areas closer to well pads, but avoided roads, except in the highest developed areas where road densities were likely too high for avoidance without significant deviations from traditional migration routes. In the second manuscript Lendrum et al. (2013) addressed biological and environmental factors influencing spring migration and assessed how energy development influenced migratory behavior. Overall, spring migration was influenced by snow depth, temperature, and green-up on winter and summer range; increasing temperatures, snow melt and emerging vegetation dictated timing of winter range departure and summer range arrival. Duration of Piceance Basin mule deer migration was short, with median migration durations of 3-8 days among the 4 areas (straight line distance between seasonal ranges averaged 32-40 km). Deer in poor condition migrated later than deer in good condition, but condition was similar among areas regardless of development status. Migrating deer from developed study areas did not avoid development structures, but departed later, arrived earlier and migrated more quickly than deer from undeveloped areas. While large changes in timing of migration could have nutritional consequences and negatively influence reproduction and neonate survival, the relatively minor shift we observed should not result in long-term fitness consequences. Migratory deer in the Piceance Basin appear to avoid negative effects of energy development through behavioral shifts in timing and rate of migration. In the third publication Lendrum et al. (2014), monitored migratory mule deer in the Piceance Basin to examined the relationship between the Normalized Difference Vegetation Index (NOVI), which is a course-scale measure of forage quality using a GIS assessment of vegetation greenness, and fecal nitrogen to assess the assumption that forage quality and deer diets can be reasonably linked to address deer habitat use patterns from remotely sensed data. We found that diet quality evident from fecal nitrogen and course measures of vegetation green-up were informative, and that Piceance Basin mule deer exhibited rapid migration (3 to 8 days depending on study area), left winter range following snow melt with lowest fecal N and NDVI values, and progressed to summer range as vegetation green-up and nitrogen levels increased, but ahead of peak vegetation green-up on summer range. I suspect this rapid migration strategy is evident for deer in relatively good condition and 9 �allows for early arrival on summer range to take advantage optimal forage conditions prior to parturition. Anderson and Bishop (2014) summarized results from Lendrum et al. (2012, 2013) and Sawyer et al. (2012) addressing migratory mule deer and energy development in northwest Colorado and south-central Wyoming, respectively. The interactions between migratory mule deer and energy development identified by Lendrum et al. (2012, 2013) and Sawyer et al. (2012) suggest mule deer may benefit from energy development planning by considering thresholds of development that may alter migratory behavior. It appears that migration rate, migration routes, and stopover use, if present, may be altered at high development intensities. In addition, migratory mule deer may benefit by maintaining security cover along migration paths, and improved habitat conditions may facilitate more direct and rapid migration requiring less energy to complete migration. Enhancing permeability along migration routes by applying dispersed development plans (>2 well pads/km2) and minimizing disturbance to vegetation types by maintaining security cover should reduce impacts to migratory mule deer as well as other migratory ungulates. Where feasible, habitat improvement projects on winter range and possibly stopover sites would also enhance migratory mule deer populations by increasing energy reserves for long-distance movements and parturition shortly after summer range arrival. Where possible, directional drilling could be used to extract energy resources from underneath migration routes while maintaining no surface occupancy. Lastly, we emphasize that GPS studies now allow managers to accurately map migration routes for entire populations and identify relatively narrow corridors that are most heavily used thus allowing for the identification of the most important corridors for migrating ungulates. Where available, we encourage agencies to incorporate such migration corridors into land-use plans (e.g., resource management plans) and National Environmental Policy Act documents. Mule Deer Behavioral Response to Energy Development u We completed evaluations of deer behavior patterns in relation to energy development activities (Northrup et al. 2015). We found diurnal responses to development activity, where deer used timbered areas away from development activity while bedded during the day and moved into more open areas generally closer to developed areas while foraging at night. Disturbance distances from producing pads and roads declined from 600 m to 200 m and about 140 m to 60 m from daytime to nighttime, respectively, but increased from 600 m to 800 m for nighttime drilling pad activity (pad response depicted in Fig. 7). We suspect deer behaviorally respond to fluctuations in development activity, where road traffic and producing well pad activity decline at night, but drilling pad disturbance may increase from compressors and lights used to facilitate nighttime drilling activity. These evaluations were applied during an active drilling phase in the Piceance Basin and deer use was influenced by development activity in 25% (nighttime) to 50% (day time) of critical winter range during that period. However, deer densities have comparably increased among developed and undeveloped study areas (excluding North Ridge; Fig. 5) suggesting that deer can behaviorally mediate development disturbance under observed development and deer densities by taking advantage of fluctuations in development activity to address their nutritional requirements. Given the plasticity in deer behavior, a number of potential options for future development planning exits including drilling schedule modifications (seasonal and/or diurnal), concentrated/staged development, reducing road traffic, and using light/noise barriers around drill rigs. It will be interesting to determine if habitat improvements will further reduce development disturbance and increase management options for future development planning. Reproductive Success and Neonate Survival To complete demographic parameters addressing mule deer-energy development interactions, CPW, Colorado State University, and ExxonMobil Production entered into a collaborative agreement 10 V �to investigate reproductive success (Peterson et al. 2017), including pregnancy rates (early Mar) and fetal survival (Mar until birth), and early fawn survival (0- 6 months; Peterson et al. 2018) in developed and relatively undeveloped landscapes beginning spring 2012 and continuing through Dec 2014. We applied statistical models to address reproductive success under contrasting energy development scenarios and noted that pregnancy and in utero fetal rates (early Mar; n = 346) were high (0.948, SE= 0.012 and 1.877, SE = 0.029, respectively) and statistically indistinguishable between study areas. Fetal survival (n = 383), however, was lower (P < 0.05) in the developed study area during 1 of 3 years (2012; Fig. 8) when drought conditions were present, suggesting the combination of severe weather conditions and development activity under observed conditions may influence fetal survival. There was no apparent influence from energy development in 0-6 month fawn survival (n = 184) based on similar mortality rates between study areas; mean daily mortality probabilities from predation, malnutrition and unknown causes were nearly identical (Fig. 9). These results suggest that natural gas development did not exert measureable influence on mule deer pregnancy rates, fetal rates or early fawn survival, but may have negatively influenced fetal survival during 2012 when does were exposed to drought conditions during the third trimester. These findings are consistent with developed areas in a production phase (little to no drilling activity) exhibiting moderate pad densities (0.4--0.9 pads/km2), and relationships may differ in areas of higher pad densities and/or drilling activity. Magnolia Habitat Treatments We completed 116 acres of pilot habitat treatments in January 2011 (Anderson and Bishop 2011; Environmental Assessment: DOI-BLM-CO-110-2011-004-EA), 54 acres of mechanical treatment method comparison treatments (hydro-ax, roller-chop, chain) in January 2012 (Stephens 2014), and 1,038 acres of hydro-ax treatments in April 2013 (Determination ofNEPA Adequacy: DOI-BLM-CO-110-2012-0134- DNA), totaling 604 treated acres in each study area (Fig. 10). Vegetation response in the pilot treatment sites was visually evident by fall 201 I (Fig. 10), and resulted in statistically significant (P < 0.05) increases in native grass and forb cover by the 2014 growing season. Final results are pending, but shrub responses appear promising from data collected through spring 2018. Stephens (2014) reported that all 3 mechanical treatment methods compared resulted in roughly a 3-fold increase in grasses, forbs, and shrubs combined after 2 growing seasons (versus control sites), but cautioned that rollerchop treatments may be more vulnerable to invasive species response. Vegetative responses from 2013 hydro-ax treatments were visually evident following I growing season and shrub responses have been notable during the 4th growing season, but statistical comparisons are still pending. As anticipated, grass and forb responses were evident 2 to 3 years posttreatment, with longer tenn response expected (3-5 years) for palatable shrubs. Ofnote, relatively high moisture conditions experienced during spring 2014 and 2015 resulted in higher than normal prevalence of cheatgrass (Bromus tectorum); cheatgrass invasion has previously been minor to non-existent in this area. Cheatgrass invasion, however, does not appear directly related to treatment sites because occurrence is evident in both treatment and control areas. We anticipate this outbreak will subside based on past competitive advantage of native species to dominate, but will continue to monitor species composition and address cheatgrass persistence in treatment and control sites. GPS data addressing deer use of treatment sites has been collected through April 2018, with collars from the Dec 2107 - May 2018 sample (n = I 06) still on deer in the field. The remaining collars will be collected during the final capture effort in March 2019. The final spring vegetation response measurements for habitat treatment and control areas were collected the past spring and final shrub response data will be collected Sep. 20 I 8. Final data analyses will be initiated once OPS collars are collected in March. Thus far, we observed improved fawn condition (P < 0.00 I) in South Magnolia following the 4th growing season of habitat treatments when compared to fawn 11 �condition in the Ryan Gulch control area, but we did not detect a response the following 2 winters. Ongoing data analyses suggests that fall shrub condition appears to have declined recently indicating that summer/fall shrub use may be increasing and potentially inhibiting the intended benefit of habitat treatments on winter range. We deployed remote cameras on treatment and control areas this summer and will continue monitoring into the fall to further address summer/fall use and identify species utilizing treatment sites on mule deer winter range. Although results are preliminary, vegetation responses through the first 4 years post treatment provided the intended forage benefit and there is some evidence that fawn condition improved as a result. Recent changes in habitat use by multiple species (potentially including wild horses and livestock) may have reduced winter forage benefits recently, but additional data collection and analyses will be necessary for confirmation. Analyses of doe use of treatment sites throughout the study are still pending and will provide information addressing the utility of habitat improvement projects as a mitigation technique to offset energy development disturbance on mule deer winter range. V SUMMARY AND COLLABORATIONS The long-term goal of this study is to investigate habitat treatments and energy development practices that enhance mule deer populations exposed to extensive energy development activity. The information presented here summarizes mule deer population parameters from the 10-year study period, with the final year of data collection for some parameters (i.e., habitat treatment response and adult female habitat use) remaining. The pretreatment period was completed by spring 2013, providing baseline data for comparison with intended improvements in habitat conditions and response to varying degrees in human development activity. Winter range habitat improvements resulting in 604 acres of mechanically treated pinion-juniper/mountain shrub habitats in each of2 study areas were completed by April of 2013, and subsequent vegetation responses have met or exceeded expectations through 2016. The post-treatment monitoring period was completed June 2018, with the final year of habitat use and habitat treatment response data collection still pending. Based on final (migration, mule deer behavioral responses, reproductive success and neonate survival) and preliminary data analyses for this 10-year project: ( 1) annual adult female survival was consistent among areas averaging 79-87% annually, but overwinter fawn survival was variable, ranging from 31 % to 95% within study areas, with annual and study area differences primarily due to early winter fawn condition, annual weather conditions, and factors associated with predation on winter range; (2) mule deer body condition early and late winter was generally consistent within areas, with higher variability among study areas early winter, primarily due to December lactation rates, and late winter condition related to seasonal moisture and winter severity; (3) late winter mule deer densities increased through 2016 in all study areas, ranging from 50% in North Ridge to 103% in North Magnolia, but have stabilized recently in 3 of the 4 study areas with recent decline evident in North Ridge; (4) migratory mule deer selected for areas with increased cover and increased their rate of travel through developed areas, and avoided negative influences through behavioral shifts in timing and rate of migration, but did not avoid development structures; (5) mule deer exhibited behavioral plasticity in relation to energy development, where disturbance distance varied relative to diurnal extent and magnitude of development activity, which may provide for several options in future development planning; and (6) energy development activity under existing conditions did not influence pregnancy rates, fetal rates or early fawn survival (0-6 months), but may have reduced neonatal survival (March until birth) when drought conditions persisted during the third trimester of doe parturition. Final results are pending to address vegetation and mule deer responses to assess habitat treatment mitigation options for energy development planning, and final results addressing the interaction of mule deer behavioral and demographic factors associated with energy development activity have recently been submitted for scientific review and publication. Completion of this project, including final data collection, analyses and interpretation of results, is anticipated by fall/winter 2020. 12 u �Hay field improvements were completed during 2012 in the North Magnolia study area by WPX Energy to fulfill a Wildlife Management Plan (WMP) agreement with CPW; rapid and continued elk (Cervus elaphus) use of these areas was evident, but mule deer response has been minor. A similar WMP agreement between ExxonMobil/XTO Energy and CPW allowed completion and continued monitoring of mechanical habitat improvements in the Magnolia study areas. Collaborative research with agency biologists, graduate students, and university professors has produced I 8 scientific publications addressing improved monitoring techniques for neonate mule deer captures (Bishop et al. 2011), mule deer migration (Lendrum et al. 2012, 2013, 2014; Anderson and Bishop 2014), improved approaches to address animal habitat use patterns (Northrup et al. 2013), mule deer response to helicopter capture and handling (Northrup et al. 2014a), potential effects of male-biased harvest on mule deer productivity (Freeman et al. 2014), mule deer genetics in relation to body condition and migration (Northrup et al. 2014b), spatial and temporal factors influencing auditory vigilance in mule deer (Lynch et al. 2014), the relationship of plant phenology with mule deer body condition (Seral et al. 2015), the influence ofindividual and temporal factors affecting late winter body condition estimates of adult female mule deer (Bergman et at. 20 I 8), and mule deer behavioral and demographic responses to energy development activities to inform future development planning (Northrup et al. 20 I 5, 2016a, 2016b, Peterson et al. 2017, 2018); these publications are summarized in Appendix A. We anticipate the opportunity to work cooperatively toward developing solutions for allowing the nation's energy reserves to be developed in a manner that benefits wildlife and the people who value both the wildlife and energy resources of Colorado. 13 �LITERATURE CITED Anderson, C.R., Jr., and D. J. Freddy. 2008. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Final Study Plan, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and C. J. Bishop. 2011. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and C. J. Bishop. 2012. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Parks and Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and C. J. Bishop. 2014. Migration patterns of adult female mule deer in response to energy development. Pages 47-50 in Transactions of the 79th North American Wildlife & Natural Resources Conference (R. A. Coon & M. C. Dunfee, eds.). Wildlife Management Institute, Gardners, PA, USA. ISSN 0078-1355. Bartmann, R. M. 1975. Piceance deer study-population density and structure. Job Progress Report, Colorado Division of Wildlife, Fort Collins, Colorado, USA. Bartmann, R. B., and S. F. Steinert. 1981. Distribution and movements of mule deer in the White River Drainage, Colorado. Special Report No. 51, Colorado Division of Wildlife, Fort Collins, Colorado, USA. Bartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monograph No. 121. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation ofa hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10: 108-114. Bergman, E. J., C.R. Anderson Jr., C. J. Bishop, A. A. Holland, and J.M. Northrup. 2018. Variation in ungulate body fat: individual versus temporal effects. Journal of Wildlife Management 82:130-137, DOI: 10:1002/jwmg.21334 Bishop, C. J., C. R. Anderson Jr., D. P. Walsh, E. J. Bergman, P. Kuechle, and J. Roth. 2011. Effectiveness of a redesigned vaginal implant transmitter in mule deer. Journal of Wildlife Management 75(8): 1797-1806; DOI: 10.1002/jwmg.229 Burnham, K. P., and D.R. Anderson. 2002. Model selection and multi-model inference: a practical information-theoretic approach. Second edition. Springer-Verlag, New York, New York, USA. Cook, R. C., J. G. Cook, D. L. Murray, P. Zager, B. K. Johnson, and M. W. Gratson. 2001. Development of predictive models of nutritional condition for rocky mountain elk. Journal of Wildlife Management 65:973-987. Cook, R. C., T. R. Stephenson, W. L. Meyers, J. G. Cook, and L.A. Shipley. 2007. Validating predictive models of nutritional condition for mule deer. Journal of Wildlife Management 71: 1934-1943. Cook, R. C., J. G. Cook, T. R. Stephenson, W. L. Meyers, S. M. McCorquodale, D. J. Vales, L. L. Irwin, P. Briggs Hall, R. D. Spencer, S. L. Murphie, K. A. Schoenecker, P. J. Miller. 2009. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife Management 74:880-896. Freeman, E. D., R. T. Larsen, M. E. Peterson, C.R. Anderson, Jr., K. R. Hersey, and B. R. McMillan. 2014. Effects of male-biased harvest on mule deer: implications for rates of pregnancy, synchrony, and timing of parturition. Wildlife Society Bulletin; DOI: 10. 1002/wsb.450 Gibbs, H. D. 1978. Nutritional quality of mule deer foods, Piceance Basin, Colorado. Thesis, Colorado State University, Fort Collins, Colorado, USA. 14 �Kaplan, E. L., and P. Meier. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 52:457-481. Lendrum, P. E., C.R. Anderson, Jr., R. A. Long, J. K. Kie, and R. T. Bowyer. 2012. Habitat selection by mule deer during migration: effects of landscape structure and natural gas development. Ecosphere 3(9):82. http://dx.doi.org/ I0. Lendrum, P. E., C.R. Anderson, Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2013. Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes. PLoS ONE 8(5): e64548. doi: IO. l 371/journal.pone.0064548 Lendrum, P. E., C.R. Anderson, Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2014. Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns of forage quality. Mammalian Biology: http://dx.doi.org/10.1016/j.mambio.2014.05.005 Lynch, E., J.M. Northrup, M. F. McKenna, C.R. Anderson Jr., L. Angelor:ii, and G. Wittemyer. 2014. Landscape and anthropogenic features influence the use of auditory vigilance by mule deer. Behavioral Ecology; doi: 10.1093/beheco/aru 158. McClintock, B. T., G. C. White, K. P. Burnham, and M. A. Pride. 2008. A generalized mixed effects model of abundance for mark-resight data when sampling is without replacement. Pages 271- 289 in D. L. Thompson, E.G. Cooch, and M. J. Conroy, editors, Modeling demographic processes is marked populations. Springer, New York, New York, USA. Northrup, J. M. 2015. Behavioral response of mule deer to natural gas development in the Piceance Basin. Dissertation, Colorado State University, Fort Collins, USA. Northrup, J.M., M. 8. Hooten, C.R. Anderson, Jr., and G. Wittemyer. 2013. Practical guidance on characterizing availability in resource selection functions under a use-availability design. Ecology 94(7):1456-1463. Northrup, J. M., C. R. Anderson, Jr., and G. Wittemyer. 2014a. Effects of helicopter capture and handling on movement behavior of mule deer. Journal of Wildlife Management 78(4):731738; DOI: IO. I 002/jwmg.705 Northrup, J.M., A. B. Shafer, C.R. Anderson Jr., D. W. Coltman, and G. Whittemyer. 2014b. Finescale genetic correlates to condition and migration in a wild cervid. Evolutionary Applications ISSN 1752-4571; doi: 10.1111/eva.12189 Northrup, J. M., C. R. Anderson, Jr., and G. Wittemyer.2015. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer. Global Change Biology, doi: 10.1111/gcb.13037. Northrup, J.M., C.R. Anderson, Jr., M. B. Hooten, and G. Wittemyer. 2016. Movement reveals scale dependence in habitat selection of a large ungulate. Ecological Applications 26:2746-2757 Northrup, J.M., C.R. Anderson, Jr., and G. Wittemyer. 2016. Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid. Diversity and Distributions 22: 547-557, DOI: IO.Ill I/ddi.12417 Peterson, M. E., C.R. Anderson Jr., J.M. Northrup, and P. F. Doherty Jr. 2017. Reproductive success of mule deer in a natural gas development area. Wildlife Biology doi: I0.1111/wlb.00341 Peterson, M. E., C.R. Anderson Jr., J.M. Northrup, and P. F. Doherty Jr. 2018. Mortality of mule deer fawns in a natural gas development area. Journal of Wildlife Management 82: 1135-1148, DOI: 10.1002/jwmg.21476 Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. C. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Sawyer, H., M. J. Kauffman, A. D. Middleton, T. A. Morrison, R. M. Nielson, and T. B. Wycoff. 2012. A framework for understanding semi-permeable barrier effects on migratory ungulates. Journal of Applied Ecology 50:68-78. Searle, K. R., M. B. Rice, C.R. Anderson, C. Bishop and N. T. Hobbs. 2015. Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore. Oecologia ISSN 0029- 8549; DOI 10.1007/s00442-015-3348-9 Stephens, G. J. 2014. Understory responses to mechanical removal of pinyon-juniper overstory. MS Thesis, Colorado State University, Ft. Collins USA. 15 �Stephenson, T. R., V. C. Bleich, B. M. Pierce, and G. P. Mulcahy. 2002. Validation of mule deer body composition using in vivo and post-mortem indices of nutritional condition. Wildlife Society Bulletin 30:557-564. Stephenson, T. R., K. J. Hundertmark, C. C. Swartz, and V. Van Ballenberghe. 1998. Predicting body fat and mass in moose with untrasonography. Canadian Journal of Zoology 76:717-722. Stonehouse, K. F., C.R. Anderson Jr., M. E. Peterson, and D.R. Collins. 2016. Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus). Colorado Parks and Wildlife Technical Report No. 48, First Edition, 317 W. Prospect Rd., Ft. Collins, CO USA. DOW-R-T-48-16, ISSN 0084-8883. Unsworth, J. W., D. F. Pack, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. Van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-421 in L. Nielsen, J.C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked individuals. Bird Study 46: 120-139. White, G. C., and B. C. Lubow. 2002. Fitting population models to multiple sources of observed data. Journal of Wildlife Management 66:300-309. Prepared by_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ Charles R. Anderson, Jr., Mammals Research Leader 16 �Table 1. Survival rate estimates (S) of fawn (1 Dec. 2017-15 June 2018) and adult female (1 July 201730 June 2018) mule deer from 4 winter range study areas of the Piceance Basin in northwest Colorado. Cohort Study area Initial sample size (n) March doe sample8 (n) S (95% Cl) Fawns Ryan Gulch 53 0.335 (0.207-0.464) South Magnolia 53 0.479 (0.343-0.616) North Magnolia 57 0.665 (0.543-0.788) North Ridge 58 0.757 (0.645-0.868) Adult females Ryan Gulch 27 52 0.832 (0.722-0.942) South Magnolia 32 56 0.869 (0. 757-0.980) North Magnolia 27 49 0.781 (0.650---0.913) North Ridge 24 45 0. 731 (0.583-0.878) aAdult female sample sizes following capture and radio-collaring efforts March, 2017. 17 �Table 2. Mean rump fat (mm) and % ingesta-free body fat' (% fat) of adult female mule deer from 4 study areas in the Piceance Basin of northwest Colorado, March and December, 2009-2018. Values in parentheses= SD. March 2009 Study Area Rump fat Ryan Gulch %fat December 2009 March 2010 December 20 I0 Rump fat %fat Rump fat % fat Rump fat %fat 1.73 (1.78) 7.08 (1.27) 8.35 (6.36) 10.54 (3.72) 2.31 (1.44) 6.37 (1.41) 7.26 (6.36) 9.69 (3.56) South Magnolia 1.29 (0.47) 6.74 (2.27) IO.OS (6.19) 11.44 (3.50) 3.12 (2.20) 7.11 (1.69) 9.85 (6.78) 11.27 (3.75) North Magnolia 1.31 (1.01) 7.15 (1.63) 10.67 (5.76) 11.94 (3.39) 3.15 (2.34) 7.54 (1.53) 9.55 (6.49) 10. 79 (4.26) North Ridge 1.57 (1.22) 6.81 (1.68) 5.25 (5.65) 1.77 (1.11) 6.39 (1.45) 7.25 (5.41) 9.85 (3.02) 9.37 (3.08) Table 2. Continued. March 2011 Study Area Rump fat %fat Ryan Gulch 1.55 (0.60) 6.72 (1.37) South Magnolia December 2011 Rump fat %fat March 2012 December 2012 Rump fat %fat Rump fat %fat 13.41 (6.39) 13.17 (3.64) 2.15 (1.44) 7.22 (1.16) 6.34 (4.35) 9.34 (2.43) 1.65 (0.75) 6.15 (1.75) 8.18 (5.45) 10.34 (3.28) 1.66 (0.77) 7.03 (1.13) 8.30 (5.71) I 0.32 (3.23) North Magnolia 1.65 (0.67) 6.79 (1.47) 8.76 (5.76) 10. 73 (3. I4) 1.90 (0.76) 7.61 (0.96) 9.66 (6.41) I 1.18 (3.64) North Ridge 1.45 (0.76) 6.30 (1.65) 8.86 (5.65) 10. 77 (3.33) 2.24 (1.58) 7.26 (1.05) 5.76 (4.10) 9.06 (2.31) 15 C ( C �( ( (_ Table 2. Continued. March 2013 December 2013 March 2014 December 2014 %fat 9.27 (6.29) 10.61 (3.76) 1.69 (0.85) 7.03 (0.99) 8.50 (6.76) 10.56 (3.70) 2.06 (0.77) 7.19 (0.66) 11.27 (8.40) 11.40 (4.16) 2.57 (1.61) 7.75 (0.68) 10.96 (6.82) 11.98 (3.8 I) North Magnolia 1.76 (0.91) 6.87 (1.11) 9.00 (6.15) 10.48 (3.25) 2.33 (2.12) 7.31 (1.43) 9.52 (5.83) 11.18 (3.32) North Ridge 1.87 (0.73) 6.70 (1.12) 11.17 (5.28) 11.66 (2.69) 2.38 (1.52) 7. 16 (1.14) 7.93 (5.50) Rump fat %fat Ryan Gulch 1.87 (0.90) 7.14 (0.89) South Magnolia Rump fat % fat Rump fat %fat Rump fat Study Area 10.20 (3.01) Table 2. Continued. March 2015 December 2015 March 2016 December 2016 Rump fat %fat Rump fat %fat 12.80 (6.83) 12.89 (3.72) 2.29 (0.64) 7.29 (0.52) 8.20 (4.90) 10.46 (2. 70) 9.83 (2.69) 2.07 (1.39) 7.46 (0.93) 6.27 (4.62) 9.37 (2.53) 2.25 (0.97) 7.49 (0.90) 8.79 (6.01) 10.81 (3.54) 2.43 (1.01) 7.17 (0.87) 7.90 (5.52) 10.34 (3.14) 2.28 (1.37) 7.43 (1.05) 5.47 (5.49) 1.58 (0.70) 6.73 (1.26) 7.74 (5.48) 10.01 (3.09) Study Area Rump fat %fat Rump fat Ryan Gulch 2.62 (0.95) 7.44 (0.53) South Magnolia 2.66 (1.36) 7.62 (0.74) 6.93 (4.83) North Magnolia North Ridge %fat 9.35 (2.75) 16 �Table 2. Continued. March 2017 December 2017 March 2018 Study Area Rump fat % fat Rump fat %fat Rump fat % fat Ryan Gulch 2.39 (0.74) 6.78 (0.97) 4.47 (3.57) 8.62 (1.80) 2.13 (0.76) 7.40 (0.50) South Magnolia 2.48 (0.77) 7.09 (0.63) 6.67 (5.23) 9.56 (2.73) 2.19 (1.18) 7.40 (0.72) North Magnolia 1.82 (0.72) 7.05 (0.58) 6.16 (4.32) 9.23 (2.47) 1.87 (0.63) 7.15(1.11) North Ridge 2.30 (1.37) 7.23 (1.21) 6.60 (4.29) 9.38 (2.35) 2.35 (0.80) 7.73 (1.03) 8 lngesta-free body fat calculated following Cook et al. (2009). 17 ( ( C �Table 3. Mark-resight abundance (N) and density estimates of mule deer from 4 winter range herd segments in the Piceance Basin, northwest Colorado. 26-31 March 2018. Data represent 4 helicopter resight surveys from 3 of 4 study areas. with South Magnolia receiving 5 surveys. Study area Mean No. sighted Mean No. marked N(95% Cl) Density (deer/km 2 ) Ryan Gulch 418 19 1.397 ( 1.186-1.674) 9.9 South Magnolia 239 26 764 (678-874) 8.1 North Magnolia 319 27 899 (774-1.070) 9.8 North Ridge 305 32 838 (748-954) 15.8 18 �,,. Mule Deer Winter Range Study Areas Mule deer study areas ' D r1onh l.!agnol1a Well Pads & Facilities .: Jn development ; PrOductng well _ Deve1opment Jae111J,es Sou1n Magnolia 25 10 ,.111.i:s Figure I. Mule deer winter range study areas relative to active natural gas wel l pads and energy development facilities in the Piceance Basin of northwest Colorado. winter 2013/ 14 (Accessed http://cogcc.state.eo.us/ Dec.31.2013). Development activity has subsided with minimal drilling activity since 2013. 19 �Winter fawn survival 2010-11 - 2017-18 1.00 tr 0.90 1 0.80 - 1-- - 0.40 - 0.30 - - 0.20 - - 0.10 - - - 0.70 0.60 0.50 - - - a T 1 T T i-- t-- i - - -- - - - - ,_ - - - - - - - - - - □ Ryan Gulch □ South Magnolia El North Magnoha 0.00 2010-11 2011-12 2012-13 2013-14 2014-15 2015-16 2016-17 2017-18 Figure 2. Over-winter (Dec-June) mule deer fawn survival (5) from 4 study areas in the Piceance Basin. no11hwest Colorado. 20 I0-11 - 20 17-18. E1rnr bars = 95% Cl. Fa,vn survival estimates past late March unavailable for winter 2008-09 and 2009-10 due to premature collar drop. but survival estimates were comparable to 20 11-1 2 and 20 12- 13 (excluding 2012- 13 North Ridge) during 2008-09 and to the higher survival rates from 20 13-1 4 and 2014- 15 during 2009-10. 20 �Male faw n w eights 42.0 40.0 r - 30.0 - - - '- - - - '- - - - - '- - - '- - -, DRvan Gulch D South Magno!ii) - c......: - ■North MagnoliJ '- - D Nort h Ridge '- 28.0 Dec Dec Dec Dec Dec Dec Dec Dec Dec Dec 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Female fawn weights 42.0 40.0 co 38.0 D Rvan Gulch ... :::. 36.0 ..c: D South M agnolia -~ 34.0 3: ■North Magnolia 32.0 D NcrthRidge 30.0 28.0 Dec 2008 Dec 2009 Dec 2010 Dec Dec Dec Dec Dec Dec Dec 2011 2012 2013 2014 2015 2016 2017 Figure 3. Mean male and female fawn weights and 95% C I (error bars) from 4 mule deer study areas in the Piceance Basin. 11011hwest Colorado. December 2008- 20 I 7. 21 �Early winter rump fat (mm) 16 I 14 12 10 8 6 I 4 2 0 Dec 2009 Dec 2010 Dec 2011 Dec 2012 Dec 2013 Dec 2014 Dec 2015 Dec 2016 Dec 2017 ■ North Ridge ■ North Magnolia Ryan Gu lch ■ South Magnolia Late winter rump fat (mm) 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 Mar 2009 Mar 2010 Mar 2011 Mar 2012 Mar 2013 Mar 2014 Mar 2015 Mar 2016 Mar 2017 Mar 2018 ■ North Ridge ■ North M agnolia Ryan Gu lch ■ South Magnolia Figure 4. M ean early (early Dec .. Top) and late winter (early Mar.. Bollom) body conditi on (mm rump fat) of adult female mule deer from 4 w inter range study areas in the Piceance Basin of northwest Colorado. March 2009- March 20 18. E1Tor bars = 95% CI. �Piceance Basin late winter mule deer density 35.00 30.00 25.00 l":::- 20.00 - t 15.00 C - North Ridge • • • ••• Ryan Gulch - • North Magnolia - South Magnolia 0.00 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Year ---• - ---•···-- -------------------------' Figure 5. Mule deer density estimates and 95% CI (error bars) from 4 winter range herd segments in the Piceance Basin, northwest Colorado, late winter 2009-2018. Estimates for were adjusted upward (using GPS migration data) to account for early migration from winter range prior to and during surveys during years when early migration biased estimates low (North Ridge 2014-2017, North Magnolia 2015 and 2017, Ryan Gulch 2017). u 24 ------------ �Colorado Figure 6. Mule deer study areas in the Piceance Basin of northwestern Colorado, USA (Top), spring 2009 migration routes of adul t female mule deer (n = 52; Lower left), and acti ve natural-gas well pads (black dots) and roads (state, county, and natural-gas; white lines) from May 2009 ( Lower right; from Lendrum et al. 20 12). 25 �I ::·+----~ ---+----+ --- -f - - - r---+----+ 8 'a' C") I I I I I I I I I Prod 400 Prod 600 Prod 800 Prod 1000 Drlll 400 Drlll 600 Drill 800 Drlll 1000 Covariates j +---+----+----+---- ------ -----t----+ O i, 1J e 8 'a' C") I Prod 400 Prod 600 Prod 800 Drlll 400 Covariates Prod 1000 Drill 600 Drill 800 Drill 1000 Figure 7. Posterior distributions of population-level coefficients related to natural gas development for RSF models during the (a) day and (b) night for 53 adult female mule deer in the Piceance Basin, Northwest Colorado. Dashed line indicates 0 selection or avoidance of the habitat features. 'Drill' and 'Prod' refer to well pads where there was active drilling or producing pads, respectively. The numbers following 'Drill' or 'Prod' represent the concentric buffer over which the number of well pads was calculated (e.g., 'Drill 600' is the number of well pads with active drilling between 400-600 m from the deer location; from Northup et al. 20 I 5). 26 �,......_ 1.00 Q) - - - '-- 0.80 rt iii .... ro> 0.60 -~ :l Cl) roQ) 0.40 LL 0.20 0.00 2012 2013 2014 Year o High development o Low development I Figure 8. Model-averaged estimates of fetal survival (± 95% Cl) of mule deer fetuses from early March until birth (late May- June) in high and low energy development study areas of the Piceance Basin. northwest Colorado. 2012-20 14 ( from Peterson et al. 2017). ?;' 0.016 iii .:: 0 E 0 0.012 &' 1i -gro 0.008 ls. .z:. ·a; 0 0.004 High development Low development development Low development development Low development 2012 2012 2013 2013 2014 2014 High High Study area □ Predation □ Malnutrition ■ Unknown mortality Figure 9. Mean daily probability of death by predation. malnutrition. or unknown mortality(± 95% Cl) of mule deer fawns (0 to 6 months old) in high and lovv energy development study areas of the Piceance Basin Colorado. 2012-20 14 (from Peterson et al. 2018). 27 �North M agnoha treal~ment sites (587 ac,esl Bear~et_ 15_35b_JndG Bea,~ t_ I _Band~ _E Bear5et_36_54ana., G 1 ~ J $~\l't00dSet_ g 16_g19 •~1easewoocSe1_g I _g :s G,ea~ewoodSe:_g30_g4 2 LaeOvers,ryhts_a_rand I~- 17 Mecriamc31tre 1trrent compari son t5J acres.) Figure I 0. Habitat treatment site delineations in 2 mule deer study areas (604 acres each) of the Piceance Basin. northwest Colorado (Top: cyan polygons completed Jan. 20 I I using hydro-axe: yellow pol ygons completed Jan. 20 l 2 using hydro-axe. roller-chop. and chaining: and remaining polygons completed April 2013 using hydro-axe). January 20 11 hydro-axe treatment-site photos from N011h Hatch Gulch during April (Lower left. aerial view) and October. 201 1 (Lower ri ght. ground view). 25 �Appendix A. Abstracts of published manuscripts resulting from Piceance Basin mule deer/energy development interaction research collaborations. Abstract format specific to the respective journal requirements. Effectiveness of a redesigned vaginal implant transmitter in mule deer CHAD J. BISHOP CHARLES R. ANDERSON Jr. 1, DANIEL P. WALSH ERIC J. BERGMAN 1, PETER KUECHLE and JOHN ROTH2 1 Colorado Parks and Wildlife, Fort Collins, Colorado 80526 USA 2 Advanced Telemetry Systems, Isanti, Minnesota 55040 USA 2 1 , 1, , Citation: Bishop, C. J., C. R. Anderson Jr., D. P. Walsh, E. J. Bergman, P. Kuechle, and J. Roth. 2011. Effectiveness of a redesigned vaginal implant transmitter in mule deer. Journal of Wildlife Management 75(8):1797-1806; DOI: 10.1002/jwmg.229 ABSTRACT Our understanding of factors that limit mule deer (Odocoileus hemionus) populations may be improved by evaluating neonatal survival as a function of dam characteristics under free-ranging conditions, which generally requires that both neonates and dams are radiocollared. The most viable technique facilitating capture of neonates from radiocollared adult females is use of vaginal implant transmitters (VITs). To date, VITs have allowed research opportunities that were not previously possible; however, VITs are often expelled from adult females prepartum, which limits their effectiveness. We redesigned an existing VIT manufactured by Advanced Telemetry Systems (ATS; Isanti, MN) by lengthening and widening wings used to retain the VIT in an adult female. Our objective was to increase VIT retention rates and thereby increase the likelihood of locating birth sites and newborn fawns. We placed the newly designed VITs in 59 adult female mule deer and evaluated the probability of retention to parturition and the probability of detecting newborn fawns. We also developed an equation for determining VIT sample size necessary to achieve a specified sample size of neonates. The probability ofa VIT being retained until parturition was 0.766 (SE= 0.0605) and the probability of a VIT being retained to within 3 days of parturition was 0.894 (SE = 0.0441 ). In a similar study using the original VIT wings (Bishop et al. 2007), the probability of a VIT being retained until parturition was 0.447 (SE= 0.0468) and the probability of retention to within 3 days of parturition was 0.623 (SE= 0.0456). Thus, our design modification increased VIT retention to parturition by 0.319 (SE== 0.0765) and VlT retention to within 3 days of parturition by 0.271 (SE== 0.0634). Considering dams that retained VITs to within 3 days of parturition, the probability of detecting at least l neonate was 0.952 (SE= 0.0334) and the probability of detecting both fawns from twin litters was 0.588 (SE = 0.0827). We expended approximately 12 person-hours per detected neonate. As a guide for researchers planning future studies, we found that VIT sample size should approximately equal the targeted neonate sample size. Our study expands opportunities for conducting research that links adult female attributes to productivity and offspring survival in mule deer.© 2014 The Wildlife Society. Habitat selection by mule deer during migration: effects of landscape structure and natural-gas development PATRICK E. LENDRUM 1, CHARLES R. ANDERSON JR. 2, RYAN A. LONG 1, JOHN G. KIE 1, AND R. TERRY BOWYER 1 1 Department ofBiological Sciences, Idaho State University, Pocatello, Idaho 83209 USA 2colorado Parks and Wildlife, Grand Junction, Colorado 81505 USA Citation: Lendrum, P. E., C.R. Anderson Jr., R. A. Long, J. G. Kie, and R. T. Bowyer. 2012. Habitat selection by mule deer during migration: effects of landscape structure and natural-gas development. Ecosphere 3(9):82 http://dx.doi.org/10.1890/ES 12-00165.1 Abstract. The disruption of traditional migratory routes by anthropogenic disturbances has shifted patterns of resource selection by many species, and in some instances has caused populations to decline. Moreover, in recent decades populations of mule deer (Odocoi/eus hemionus) have declined throughout much of their historic range in the western United States. We used resourceselection functions to detennine if the presence of natural-gas development altered patterns of resource selection by migrating mule deer. We compared spring migration routes of adult female mule deer fitted with GPS collars (n = 167) among four study areas that had varying degrees of natural-gas development from 2008 to 2010 in the Piceance Basin of northwest Colorado, USA. Mule deer migrating through the most developed area had longer step lengths (straight-line distance between successive GPS locations) compared with deer in less developed areas. Additionally, deer migrating through the most developed study areas tended to select for habitat types that provided greater amounts of concealment cover, whereas deer from the least developed areas tended to select habitats that increased access to forage and cover. Deer selected habitats closer to well pads and avoided roads in all instances except along the most highly developed migratory routes, where road densities may have been too high for deer to avoid roads without deviating substantially from established migration routes. These results indicate that behavioral tendencies toward avoidance of anthropogenic disturbance can be overridden during migration by the strong fidelity ungulates demonstrate towards migration routes. If avoidance is feasible, then deer may select areas further from development, whereas in highly developed areas, deer may simply increase their rate of travel along established migration routes. 26 �Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes Patrick E. Lendrum1, Charles R. Anderson Jr. 2, Kevin L Monteith 1.l, Jonathan A. Jenks.., R. Terry Bowyer 1 1 Department of Biological Sciences, Idaho State University, Pocatello, Idaho, USA, 2 Colorado Division of Parks and Wildlife, Grand Junction, Colorado, USA, 3 Wyoming Cooperative Fish and Wildlife Research Unit, University of Wyoming, Laramie, Wyoming, USA,4 Department of Natural Resource Management, South Dakota State University, Brookings, South Dakota, USA Citation: Lendrum, P. E., C.R. Anderson Jr., K. L. Monteith, J. A. Jenks, R. T. Bowyer. 2013. Migrating Mule Deer: Effects of anthropogenically Altered Landscapes. PLoS ONE 8(5): e64548. DOI: 10.l371~oumal.pone.0064548 Abstract Background: Migration is an adaptive strategy that enables animals to enhance resource availability and reduce risk of predation at a broad geographic scale. Ungulate migrations generally occur along traditional routes, many of which have been disrupted by anthropogenic disturbances. Spring migration in ungulates is of particular importance for conservation planning, because it is closely coupled with timing of parturition. The degree to which oil and gas development affects migratory patterns, and whether ungulate migration is sufficiently plastic to compensate for such changes, warrants additional study to better understand this critical conservation issue. Methodology/Principal Findings: We studied timing and synchrony of departure from winter range and arrival to summer range of female mule deer (Odocoi/eus hemionus) in northwestern Colorado, USA, which has one of the largest natural-gas reserves currently under development in North America. We hypothesized that in addition to local weather, plant phenology, and individual life-history characteristics, patterns of spring migration would be modified by disturbances associated with natural-gas extraction. We captured 205 adult female mule deer, equipped them with GPS collars, and observed patterns of spring migration during 2008-2010. Condusions/Signijicance: Timing of spring migration was related to winter weather (particularly snow depth) and access to emerging vegetation, which varied among years, but was highly synchronous across study areas within years. Additionally, timing of migration was influenced by the collective effects of anthropogenic disturbance, rate of travel, distance traveled, and body condition of adult females. Rates of travel were more rapid over shorter migration distances in areas of high natural-gas development resulting in the delayed departure, but early arrival for females migrating in areas with high development compared with less-developed areas. Such shifts in behavior could have consequences for timing of arrival on birthing areas, especially where mule deer migrate over longer distances or for greater durations. Practical guidance on characterizing availability in resource selection functions under a use-availability design JOSEPH M. NORTHRUP 1, MEVIN B. HOOTEN 1.J.J, CHARLES R. ANDERSON JR..., AND GEORGE WITIEMYER 1 1 Department offish, Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 2 U.S. Geological Survey, Colorado Cooperative Fish and Wildlife Research Unit, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 3 Colorado State University, Department of Statistics, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 4 Mammals Research Section Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction, Colorado 81505 USA Citation: Northrup, J. M., M. B. Hooten, C. R. Anderson Jr., and G. Wittemyer. 2013. Practical guidance on characterizing availability in resource selection functions under a use-availability design. Ecology 94(7):1456-1463. http:l/dx.doi.org/l0.1890/12-1688.1 Abstract. Habitat selection is a fundamental aspect of animal ecology, the understanding of which is critical to management and conservation. Global positioning system data from animals allow fine-scale assessments of habitat selection and typically are analyzed in a us~availability framework, whereby animal locations are contrasted with random locations (the availability sample). Although most us~availability methods are in fact spatial point process models, they often are fit using logistic regression. This framework offers numerous methodological challenges, for which the literature provides little guidance. Specifically, the size and spatial extent of the availability sample influences coefficient estimates potentially causing interpretational bias. We examined the influence of availability on statistical inference through simulations and analysis of serially correlated mule deer GPS data. Bias in estimates arose from incorrectly assessing and sampling the spatial extent of availability. Spatial autocorrelation in covariates, which is common for landscape characteristics, exacerbated the error in availability sampling leading to increased bias. These results have strong implications for habitat selection analyses using GPS data, which are increasingly prevalent in the literature. We recommend that researchers assess the sensitivity of their results to their availability sample and, where bias is likely, take care with interpretations and use cross validation to assess robustness. 27 u �Effects of Helicopter Capture and Handling on Movement Behavior of Mule Deer JOSEPH M. NORTHRUP', CHARLES R. ANDERSON JR2, AND GEORGE WITTEMYER 1 1Department of Fish, Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 2Mammals Research Section Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction, Colorado 81505 USA Citation: Northrup, J.M., C.R. Anderson Jr., and G. Wittemyer. 2014. Effects of helicopter capture and handling on movement behavior of mule deer. Journal ofWildlife Management 78(4):731-738~ DOI: 10.1002/jwmg.705 ABSTRACT Research on wildlife movement, physiology, and reproductive biology often requires capture and handling of animals. Such invasive treatment can alter behavior, which may bias results or invalidate assumptions regarding representative behaviors. To assess the impacts of handling on mule deer (Odocoileus hemionus), a focal species for research in North America, we investigated pre- and post-recapture movements of collared individuals, and compared them to deer that were not recaptured (controls). We compared pre- and post-recapture movement rates (m/hr) and 24-hour straight-line displacement among recaptured and control deer. In addition, we examined the time it took recaptured deer to return to their pre-recapture home range. Both daily straight-line displacement and movement rate were marginally elevated relative to monthly averages for 24 hours following recapture, with non-significant elevation continuing for up to 7 days. Comparing movements averaged over 30 days before and after recapture, we found no differences in displacement, but movement rates demonstrated seasonal effects, with faster movements post- relative to pre-recapture in March and slower movements post- relative to pre-recapture in December. Relative to control deer movements, recaptured deer movement rates in March were higher immediately after recapture and lower in the second and third weeks following recapture. The median time to return to the pre-recapture home range was 13 hours, with 71 % of deer returning in the first day, and 91 % returning within 4 days. These results indicate a short period of elevated movements following recaptures, likely due to the deer returning to their home ranges, followed by weaker but non-significant depression of movements for up to 3 weeks. Censoring of the first day of data post capture from analyses is strongly supported, and removing additional days until the individual returns to its home range will control for the majority of impacts from capture. © 2014 The Wildlife Society. Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns offorage quality Patrick E. Lendrum•, Charles R. Anderson Jr. h, Kevin L Monteithr, Jonathan A. Jenbd, R. Terry Bowyer• • Department ofBiological Sciences, Idaho State University, 921 South 8th Avenue, Stop 8007, Pocatello 83209, USA b Mammals Research Section Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction 81 SOS, USA c Wyoming Cooperative Fish and Wildlife Research Unit, Department of Zoology and Physiology, University of Wyoming, 3166, 1000 East University Avenue, Laramie 82071, USA d Department of Natural Resource Management, South Dakota State University, Box 2140B, Brookings 57007, USA Citation: Lendrum, P. E., C.R. Anderson Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2014. Relating the movement ofa rapidly migrating ungulate to spatiotemporal patterns of forage quality. Mammalian Biology: http://dx.doi.org/lO. l016/j.mambio.2014.0S.00S ABSTRACT: Migratory ungulates exhibit recurring movements, often along traditional routes between seasonal ranges each spring and autumn, which allow them to track resources as they become available on the landscape. We examined the relationship between spring migration of mule deer (Odocoileus hemionus) and forage quality, as indexed by spatiotemporal patterns of fecal nitrogen and remotely sensed greenness of vegetation (Nonnalized Difference Vegetation Index; NOVI) in spring 2010 in the Piceance Basin of northwestern Colorado, USA. NOVI increased throughout spring, and was affected primarily by snow depth when snow was present, and temperature when snow was absent. Fecal nitrogen was lowest when deer were on winter range before migration, increased rapidly to an asymptote during migration, and remained relatively high when deer reached summer range. Values of fecal nitrogen corresponded with increasing NDVI during migration. Spring migration for mule deer provided a way for these large mammals to increase access to a high-quality diet, which was evident in patterns of NOVI and fecal nitrogen. Moreover, these deer "jumped" rather than "surfed" the green wave by arriving on summer range well before peak productivity of forage occurred. This rapid migration may aid in securing resources and seclusion from others on summer range in preparation for parturition, and to minimize detrimental factors such as predation, and malnutrition during migration. 28 �Effects of Male-Biased Harvest on Mule Deer: Implications for Rates of Pregnancy, Synchrony, and Timing of Parturition V ERIC D. FREEMAN', RANDY T. LARSEN', MARKE. PETERSON1, CHARLES R. ANDERSON JR.3, KENT R. HERSEY', AND BROCK R. McMILLAN' 1 Depanment of Plant and Wildlife Sciences, Brigham Young University, 275 WIDB, Provo, UT 84602, USA 2 Department of Fish, Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA 3 Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction, CO 81505, USA 4 Utah Division of Wildlife Resources, 1594 W North Temple, Salt Lake City, UT 84114, USA Citation: Freeman, E. D., R. T. Larsen, M. E. Peterson, C.R. Anderson Jr., K. R. Hersey, and B. R. McMillan. 2014. Effects of male-biased harvest on mule deer: implications for rates of pregnancy, synchrony, and timing ofpanurition. Wildlife Society Bulletin~ DOI: 10.1002/wsb.450 ABSTRACT Evaluating how management practices influence the population dynamics of ungulates may enhance future management of these species. For example, in mule deer (Odocoileus hemionus), changes in male/female ratio due to malebiased harvest may alter rates of pregnancy, timing of parturition, and synchrony of parturition if inadequate numbers of males are present to fertilize females during their first estrous cycle. If rates of pregnancy or parturition are influenced by decreased male/female ratios, recruitment may be reduced (e.g., fewer births, later parturition resulting in lower survival of fawns, and a less synchronous parturition that potentially increases susceptibility of neonates to predation). Our objectives were to compare rates of pregnancy, synchrony of parturition, and timing of parturition between exploited mule deer populations with a relatively high (Piceance, CO, USA; 26 males/100 females) and a relatively low (Monroe, UT, USA; 14 males/100 females) male/female ratio. We determined rates of pregnancy via ultrasonography and timing of parturition via vaginal implant transmitters. We found no differences in rates of pregnancy (98.6%and 96.6%; z = 0.821; P = 0.794), timing of parturition (estimate= 1.258; SE= 1.672; 1 = 0.752; P = 0.454), or synchrony of parturition (F = 1.073; P = 0.859) between Monroe Mountain and Piceance Basin, respectively. The relatively low male/female ratio on Monroe Mountain was not associated with a protracted period of parturition. This finding suggests that relatively low male/female ratios typical of heavily harvested populations do not influence population dynamics because recruitment remains unaffected. © 2014 The Wildlife Society. Fine-scale genetic correlates to condition and migration in a wild cervid Joseph M. Northrup,' Aaron B. A. Shafer,1 Charles R. Anderson Jr,3 David W. Coltman" and George Wittemyer 1 I Department offish, Wildlife, and Conservation Biology, Colorado State University, Fort Collins, CO, USA 2 Department of Evolutionary Biology, Evolutionary Biology Centre, Uppsala University, Uppsala, Sweden 3 Mammals Research Section, Colorado Parks and Wildlife, Grand Junction, CO, USA 4 Depanment of Biological Sciences, University of Alberta, Edmonton, AB, Canada. Citation: Northrup, J.M., A. B. Shafer, C.R. Anderson Jr., D. W. Coltman, and G. Whittemyer. 2014. Fine-scale genetic correlates to condition and migration in a wild cervid. Evolutionary Applications ISSN 1752-4571 ~ doi: 10.1111/eva.12 l 89 Abstract The relationship between genetic variation and phenotypic traits is fundamental to the study and management of natural populations. Such relationships often are investigated by assessing correlations between phenotypic traits and heterozygosity or genetic differentiation. Using an extensive data set compiled from free ranging mule deer (Odocoileus hemionus), we combined genetic and ecological data to (i) examine correlations between genetic differentiation and migration timing, (ii) screen for mitochondrial haplotypes associated with migration timing, and (iii) test whether nuclear heterozygosity was associated with condition. Migration was related to genetic differentiation (more closely related individuals migrated closer in time) and mitochondrial haplogroup. Body fat was related to heterozygosity at two nuclear loci (with antagonistic patterns), one of which is situated near a known fat metabolism gene in mammals. Despite being focused on a widespread panmictic species, these findings revealed a link between genetic variation and important phenotypes at a fine scale. We hypothesize that these correlations are either the result of mixing refugial lineages or differential mitochondrial haplotypes influencing energetics. The maintenance of phenotypic diversity will be critical to enable the potential tracking of changing climatic conditions, and these correlates highlight the need to consider evolutionary mechanisms in management, even in widely distributed panmictic species. 29 V �,~ Landscape and anthropogenic features influence the use of auditory vigilance by mule deer Emma Lynch: Joseph M. Northrup,b Megan F. McKenna,r Charles R. Anderson Jr,d Lisa Angeloni,~ and George Wittemye.-a.b 'Graduate Degree Program in Ecology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA ti>epartment of Fish, Wildlife and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA ~atural Sounds and Night Skies Division, National Park Service, 1201 Oakridge Drive, Fort Collins, CO 80525, USA, dMammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526, USA COepartment of Biology, Colorado State University. 1878 Campus Delivery, Fort Collins, CO 80523, USA Citation: Lynch, E., J.M. Northrup, M. F. McKenna, C.R. Anderson Jr., L. Angeloni, and G. Wittemyer. 2014. Landscape and anthropogenic features influence the use of auditory vigilance by mule deer. Behavioral Ecology~ doi: I0.1093/beheco/aru 158. While visual forms of vigilance behavior and their relationship with predation risk have been broadly examined, animals also employ other vigilance modalities such as auditory vigilance by listening for the acoustic cues of predators. Similar to the tradeoffs associated with visual vigilance, auditory behavior potentially structures the energy budgets and behavior of animals. The cryptic nature of auditory vigilance makes it difficult to study, but on-animal acoustical monitoring has rapidly advanced our ability to investigate behaviors and conditions related to sound. We utilized this technique to investigate the ways external stimuli in an active natural gas development field affect periodic pausing by mule deer (Odocoileus hemionus) within bouts of rumination-based mastication. To better understand the ecological properties that structure this behavior, we investigate spatial and temporal factors related to these pauses to determine if results are consistent with our hypothesis that pausing is used for auditory vigilance. We found that deer paused more when in forested cover and at night, where visual vigilance was likely to be less effective. Additionally, deer paused more in areas of moderate background sound levels, though responses to anthropogenic features were less clear. Our results suggest that pauses during rumination represent a form of auditory vigilance that is responsive to landscape variables. Further exploration of this behavior can facilitate a more holistic understanding of risk perception and the costs associated with vigilance behavior. Migration Patterns of Adult Female Mule Deer in Response to Energy Development Charles R. Anderson Jr. and Chad J. Bishop Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526, USA Citation: Anderson, C.R., Jr., and C. J. Bishop. 2014. Migration patterns of adult female mule deer in response to energy development. Pages 47-50 in Transactions of the 79lh North American Wildlife & Natural Resources Conference (R. A. Coon & M. C. Dunfee, eds.). Wildlife Management Institute, Gardners, PA, USA. ISSN 0078-1355. Migration is an adaptive strategy that enables animals to enhance resource availability and reduce risk of predation at a broad geographic scale. Ungulate migrations generally occur along traditional routes, many of which have been disrupted by anthropogenic disturbances. Spring migration in ungulates is of particular importWlce for conservation planning because it is closely coupled with timing of parturition. The degree to which oil and gas development affects migratory patterns, and whether ungulate migration is sufficiently prepared to compensate for such changes, has recently been investigated in Colorado and Wyoming (Lendrum et al. 2012, 2013; Sawyer et al. 2012). Lendrum et al. (2012, 2013) and Sawyer et al. (2012) address mule deer (Odocoileus hemionus) migration patterns in relation to energy development from northwest Colorado and south-central Wyoming, respectively. We address results from the Colorado and Wyoming studies and then compare similarities and differences. The interactions between migratory mule deer and energy development identified by Lendrum et al. (2012, 2013) and Sawyer et al. (2012) suggest mule deer may benefit from energy development planning by considering thresholds of development that may alter migratory behavior. It appears that migration rate, migration routes, and stopover use, if present, may be altered at high development intensities. In addition, migratory mule deer may benefit by maintaining security cover along migration paths, and improved habitat conditions may facilitate more direct and rapid migration requiring less energy to complete migration. Enhancing permeability along migration routes by applying dispersed development plans (>2 well pads/km 2) and minimizing disturbance to vegetation types by maintaining security cover should reduce impacts to migratory mule deer as well as other migratory ungulates. Where feasible, habitat improvement projects on winter range and possibly stopover sites would also enhance migratory mule deer populations by enhancing energy reserves for long-distance movements and parturition shortly after summer range arrival. Where possible, directional drilling could be used to extract energy resources from underneath migration routes while maintaining no surface occupancy. Lastly, we emphasize that GPS studies now allow managers to accurately map migration routes for entire populations and identify relatively narrow corridors that are most heavily used thus allowing for the identification of the most important corridors for migrating ungulates. Where available, we encourage agencies to incorporate such migration corridors into land-use plans (e.g., resource management plans) and National Environmental Policy Act documents. 30 �Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore Kate R. Searle 1 • Mindy B. Rice2 • Charles R. Anderson 2 • Chad Bishop2 • N. T. Hobbs3 1 NERC Centre for Ecology and Hydrology. Bush Estate, Penicuik EH26 0QB. UK Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins. CO 80526, USA 3 Department of Ecosystem Science and Sustainability, Colorado State University. Fort Collins 80524, CO, USA 2 Citation: Searle. K. R.. M. B. Rice, C.R. Anderson, C. Bishop and N. T. Hobbs. 2015. Asynchronous vegetation phenology enhances winter body condition ofa large mobile herbivore. Oecologia ISSN 0029-8549; OOl 10.1007/s00442-015-3348-9 Abstract Understanding how spatial and temporal heterogeneity influence ecological processes forms a central challenge in ecology. Individual responses to heterogeneity shape population dynamics, therefore understanding these responses is central to sustainable population management. Emerging evidence has shown that herbivores track heterogeneity in nutritional quality of vegetation by responding to phenological differences in plants. We quantified the benefits mule deer (Odocoileus hemionus) accrue from accessing habitats with asynchronous plant phenology in northwest Colorado over 3 years. Our analysis examined both the direct physiological and indirect environmental effects of weather and vegetation phenology on mule deer winter body condition. We identified several important effects of annual weather patterns and topographical variables on vegetation phenology in the home ranges of mule deer. Crucially, temporal patterns of vegetation phenology were linked with differences in body condition, with deer tending to show poorer body condition in areas with less asynchronous vegetation green-up and later vegetation onset. The direct physiological effect of previous winter precipitation on mule deer body condition was much less important than the indirect effect mediated by vegetation phenology. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer JOSEPH M. NORTHRUP', CHARLES R. ANDERSON JR.,2 and GEORGE WITTEMYER 1 • 3 1Department offish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, CO, USA 2Mammals Research Section. Colorado Parks and Wildlife, Fort Collins, CO, USA 3Graduate Degree Program in Ecology, Colorado State University, Fort Collins, CO, USA Citation: Northrup. J.M., C.R. Anderson, Jr., and G. Wittemyer. 2015. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer. Global Change Biology, doi: 10.1111/gcb.13037 Abstract Extraction of oil and natural gas (hydrocarbons) from shale is increasing rapidly in North America, with documented impacts to native species and ecosystems. With shale oil and gas resources on nearly every continent, this development is set to become a major driver of global land-use change. It is increasingly critical to quantify spatial habitat loss driven by this development to implement effective mitigation strategies and develop habitat offsets. Habitat selection is a fundamental ecological process, influencing both individual fitness and population-level distribution on the landscape. Examinations of habitat selection provide a natural means for understanding spatial impacts. We examined the impact of natural gas development on habitat selection patterns of mule deer on their winter range in Colorado. We fit resource selection functions in a Bayesian hierarchical framework, with habitat availability defined using a movement-based modeling approach. Energy development drove considerable alterations to deer habitat selection patterns, with the most substantial impacts manifested as avoidance of well pads with active drilling to a distance of at least 800 m. Deer displayed more nuanced responses to other infrastructure, avoiding pads with active production and roads to a greater degree during the day than night. In aggregate, these responses equate to alteration of behavior by human development in over 50% of the critical winter range in our study area during the day and over 25% at night. Compared to other regions, the topographic and vegetative diversity in the study area appear to provide refugia that allow deer to behaviorally mediate some of the impacts of development. This study, and the methods we employed, provides a template for quantifying spatial take by industrial activities in natural areas and the results offer guidance for policy makers, mangers, and industry when attempting to mitigate habitat loss due to energy development. 31 �Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid Joseph M. Northrup•, Charles R. Anderson Jr and George Wittemyer1.l 1Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, CO, USA 2 Mammals Research Section, Colorado Parks and Wildlife, Fort Collins, CO, USA 3 Graduate Degree Program in Ecology, Colorado State University, Fort Collins, CO, USA Citation: Northrup, J.M., C. R Anderson, Jr., and G. Wittemyer. 2016. Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid. Diversity and Distributions 22: 547-557, DOI: I0. I I I l/ddi.12417 ABSTRACT Aim The space an animal uses over a given time period must provide the resources required for meeting energetic needs, reproducing and avoiding predation. Anthropogenic landscape change in concert with environmental dynamics can strongly structure space-use. Investigating these dynamics can provide critical insight into animal ecology, conservation and management. Location The Piceance Basin, Colorado, USA. Methods We applied a novel utilization distribution estimation technique based on a continuous-time correlated random walk model to characterize range dynamics of mule deer during winter and summer seasons across multiple years. This approach leverages secondorder properties of movement to provide a probabilistic estimate of space-use. We assessed the influence of environmental (cover and forage), individual and anthropogenic factors on interannual variation in range use of individual deer using a hierarchical Bayesian regression framework. Results Mule deer demonstrated remarkable spatial philopatry, with a median of 50% overlap (range: 8-78%) in year-to-year utilization distributions. Environmental conditions were the primary driver of both philopatry and range size, with anthropogenic disturbance playing a secondary role. Main conclusions Philopatry in mule deer is suspected to reflect the importance of spatial familiarity (memory) to this species and, therefore, factors driving spatial displacement are of conservation concern. The interaction between range behaviour and dynamics in development disturbance and environmental conditions highlights mechanisms by which anthropogenic environmental change may displace deer from familiar areas and alter their foraging and survival strategies. Movement reveals scale dependence in habitat selection of a large ungulate Joseph M. Northrup, 1 Charles R. Anderson Jr.,2 Mevin 8. Hooten,3 and George Wittemyel" 'Department ofFish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA 2 Mammals Research Section, Colorado Parks and Wildlife, Fort Collins, Colorado 80523 USA 3 U.S. Geological Survey, Colorado Cooperative Fish and Wildlife Research Unit, Department ofFish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA 4 0epartment ofFish, Wildlife and Conservation Biology and Graduate Degree Program in Ecology, Colorado State University, Fort Collins, Colorado 80523 USA Citation: Northrup, J.M., C.R. Anderson, Jr., M. 8. Hooten, and G. Wittemyer. 2016. Movement reveals scale dependence in habitat selection ofa large ungulate. Ecological Applications 26:2746-2757 Abstract. Ecological processes operate across temporal and spatial scales. Anthropogenic disturbances impact these processes, but examinations of scale dependence in impacts are infrequent. Such examinations can provide important insight to wildlife-human interactions and guide management efforts to reduce impacts. We assessed spatiotemporal scale dependence in habitat selection of mule deer (Odocoileus hemionus) in the Piceance Basin of Colorado, USA, an area of ongoing natural gas development. We employed a newly developed animal movement method to assess habitat selection across scales defined using animal-centric spatiotemporal definitions ranging from the local (defined from five hour movements) to the broad (defined from weekly movements). We extended our analysis to examine variation in scale dependence between night and day and assess functional responses in habitat selection patterns relative to the density of anthropogenic features. Mule deer displayed scale invariance in the direction of their response to energy development features, avoiding well pads and the areas closest to roads at all scales, though with increasing strength of avoidance at coarser scales. Deer displayed scale-dependent responses to most other habitat features, including land cover type and habitat edges. Selection differed between night and day at the finest scales, but homogenized as scale increased. Deer displayed functional responses to development, with deer inhabiting the least developed ranges more strongly avoiding development relative to those with more development in their ranges. Energy development was a primary driver of habitat selection patterns in mule deer, structuring their behaviors across all scales examined. Stronger avoidance at coarser scales suggests that deer behaviorally mediated their interaction with development, but only to a degree. At higher development densities than seen in this area, such mediation may not be possible and thus maintenance of sufficient habitat with lower development densities will be a critical best management practice as development expands globally. 32 �Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus) Kourtney F. Stonehouse,•.z Charles R. Anderson Jr., 1 Mark E. Peterson, 1.2 and David R. Collins1 'Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 90526 USA 2Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA Citation: Stonehouse, K. F., C.R. Anderson Jr., M. E. Peterson, and D.R. Collins. 2016. Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus). Colorado Parks and Wildlife Technical Report No. 48, First Edition, 317 W. Prospect Rd., Ft. Collins, CO USA. DOW-R-T-48-16, ISSN 0084-8883. This technical report provides general guidelines for conducting mortality site investigations to help investigators distinguish predation from scavenging and other causes of death. General health indices are also provided to assess whether or not deer may have died from malnutrition or disease or if these factors may have predisposed deer to predation. Lastly, these guidelines will assist investigators in identifying predatory species or scavengers involved through the examination of physical evidence at deer mortality sites. The information presented here is based primarily on field experience gained from a long term research effort in northwest Colorado investigating mule deer mortality sites over several years (http://cpw.state.co.us/leam/Pages/ResearchMammalsRP-04.aspx) and literature review where referenced. We acknowledge that proximate and ultimate cause of death can be difficult or impossible to detect from field necropsy alone and examples presented here largely represent proximate causes of mortality; efforts discerning ultimate cause will require specific tissue sample collections, where possible, submitted to a veterinary diagnostic laboratory. Within this technical report are numerous photographs documenting characteristics of predator attacks on mule deer and signs left by predatory and scavenging species. Additional pictures illustrate differences between healthy and unhealthy tissues and organs. While reading this document, be aware that each mortality investigation is unique and observations in the field may differ from illustrations provided here. Appendix I provides a sample necropsy form to assist in conducting mortality investigations. Reproductive success of mule deer in a natural gas development area Mark E. Peterson,' Charles R. Anderson Jr.,2 Joseph M. Northrup•, and Paul F. Doherty Jr. 1 1Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA 2Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 90526 USA Citation: Peterson, M. E., C.R. Anderson Jr., J.M. Northrup, and P. F. Doherty Jr. 2017. Reproductive success of mule deer in a natural gas development area. Wildlife Biology doi: IO. Ill l/wlb.00341 Abstract: Natural gas development is increasing across North America and causing concern over the potential impacts on wildlife populations and their habitat, particularly for ungulate species. Understanding how this development impacts reproductive success metrics that are influential for ungulate population dynamics is important to guide management of ungulates. However, the influences of natural gas development on reproductive success metrics of mule deer Odocoileus hemionus have not been studied. We used statistical models to examine the influence of natural gas development and temporal factors on reproductive success metrics of mule deer in the Piceance Basin, northwest Colorado during 2012-2014. We focused on study areas with relatively high or low levels of natural gas development. Pregnancy and in utero fetal rates were high and statistically indistinguishable between study areas. Fetal survival rates increased over time and survival was lower in the high versus low development study areas in 2012 possibly influenced by drought coupled with habitat loss and fragmentation associated with development. Our novel results suggest managers should be concerned with the influences of development on fetal survival, particularly during extreme environmental conditions (e.g. drought) and our results can be used to guide development planning and/or mitigation. Developers and wildlife managers should continue to collaborate on development planning, such as implementing habitat treatments to improve forage availability and quality, minimizing disturbance to hiding and foraging habitat particularly during parturition, and implementing directional drilling to minimize pad disturbance density to increase fetal survival in developed areas. 33 �Colorado Parks and Wildlife July I, 2018-June 30, 2019 WILDLIFE RESEARCH REPORT State of ______C=o=l=o=ra=d=o_ _ _ _ _: !..P.!!!ar~ks~an~d~W~il~d~li~~e:..,___ _ _ _ _ _ _ _ _ __ Cost Center 3430 : .uM.!!am~m.:.!:a~l~s..!.:R~e~se~ar:::..:c=:.:.h:.. .__ _ _ _ _ _ _ _ _ __ Work Package 3001 : .::::D:..:::ee:::r:..-C~o=n~s~e::..;rv~a=ti=on=-=--------------Task No. 6 : Population Performance of Piceance Basin Mule Deer in Response to Natural Gas Resource Extraction and Mitigation Efforts to Address Human Activity and Habitat Degradation Federal Aid Project:. ____W__-=-24"""""3'---"--"R3=-----Period Covered: July I, 2018-June 30, 2019 Author: C. R. Anderson, Jr. Personnel: E. Bergman, D. Bilyeu-Johnston, G. Cahal, D. Collins, B. deVergie, D. Finley, M. Fisher, L. Gepfert, T. Knowles, B. Petch, J. Rivale, E. Sawa, Z. Swennes, M. Way, L. Wolfe, CPW; L. Belmonte, BLM; T. Graham, Ranch Advisory Partners; P. Doherty Jr., J. Northrup, M. Peterson, G. Wittemyer, K. Wilson, Colorado State University; R. Swisher, S. Swisher, Quicksilver Air, Inc.; D. Felix, Olathe Spray Service, Inc.; L. Coulter, Coulter Aviation; H. Sawyer, Western Ecosystems Technology, Inc. Project support received from Federal Aid in Wildlife Restoration, Colorado Mule Deer Association, Colorado Mule Deer Foundation, Muley Fanatic Foundation, Colorado State Severance Tax Fund, EnCana Corp., ExxonMobil Production Co./XTO Energy, Marathon Oil Corp., Shell Petroleum, and WPX Energy. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the authors. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT ~- We propose to experimentally evaluate winter range habitat treatments and human-activity management alternatives intended to enhance mule deer (Odocoileus hemionus) populations exposed to energy-development activities. The Piceance Basin of northwestern Colorado was selected as the project area due to ongoing natural gas development in one of the most extensive and important mule deer winter and transition range areas in Colorado. The data presented here represent preliminary and final results of a I0-year research project addressing habitat improvements and evaluation of energy development practices intended to improve mule deer fitness in areas exposed to extensive energy development. We monitored deer on 4 winter range study areas representing relatively high (Ryan Gulch, South Magnolia) and low (North Magnolia, North Ridge) levels of development activity to address factors influencing deer behavior and demographics and to evaluate success of habitat treatments as a mitigation option. We recorded habitat use and movement patterns, estimated annual neonatal, overwinter fawn and annual adult female survival, estimated annual early and late winter body condition of adult females, and estimated annual abundance among study areas. Winter range habitat improvements completed spring 2013 resulted in 604 acres of mechanically treated pinion-juniper/mountain shrub habitats in each of 2 treatment areas with minor (North Magnolia) and extensive (South Magnolia) energy development, respectively. During this research segment, we removed store-on-board GPS collars from adult female mule deer, addressed �mule deer winter concentration areas during a post-drilling production phase, measured vegetation response of habitat treatment sites and established camera grids to address summer/fall use of habitat treatments. Based on final (migration, mule deer behavioral responses, reproductive success and neonate survival) and preliminary data analyses for this IO-year project: (I) annual adult female survival was consistent among areas averaging 79-87% annually, but overwinter fawn survival was variable, ranging from 31 % to 95% within study areas, with annual and study area differences primarily due to early winter fawn condition, annual weather conditions, and factors associated with predation on winter range; (2) mule deer body condition early and late winter was generally consistent within areas, with higher variability among study areas early winter, primarily due to December lactation rates, and late winter condition related to seasonal moisture and winter severity; (3) late winter mule deer densities increased through 2016 in all study areas, ranging from 50% in North Ridge to I03% in North Magnolia, but have stabilized recently in 3 of the 4 study areas with recent decline evident in North Ridge; (4) migratory mule deer selected for areas with increased cover and increased their rate of travel through developed areas, and avoided negative influences through behavioral shifts in timing and rate of migration, but did not avoid development structures; (5) mule deer exhibited behavioral plasticity in relation to energy development, where disturbance distance varied relative to diurnal extent and magnitude of development activity, which may provide for several options in future development planning; and (6) energy development activity under existing conditions did not influence pregnancy rates, fetal rates or early fawn survival (0-6 months), but may have reduced neonatal survival (March until birth) when drought conditions persisted during the third trimester of doe parturition. Final results are pending to address vegetation and mule deer responses to assess habitat treatment mitigation options for energy development planning, and final results addressing the interaction of mule deer behavioral and demographic factors associated with energy development activity have recently been submitted for scientific peer-review and publication. Final data collection addressing GPS collar recovery and summer/fall use of habitat treatment sites will be completed by December 2019. Completion of this project, including data analyses and interpretation of results, is anticipated by fall/winter 2020-21. 2 �WILDLIFE RESEARCH REPORT POPULATION PERFORMANCE OF PICEANCE BASIN MULE DEER IN RESPONSE TO NATURAL GAS RESOURCE EXTRACTION AND MITIGATION EFFORTS TO ADDRESS HUMAN ACTMTY AND HABITAT DEGRADATION CHARLESR.ANDERSON,JR PROJECT NARRITIVE OBJECTIVES 1. To determine experimentally whether enhancing mule deer habitat conditions on winter range elicits behavioral responses, improves body condition, increases fawn survival, and ultimately, population density on mule deer winter ranges exposed to extensive energy development. 2. To determine experimentally to what extent modification of energy development practices enhance habitat selection, body condition, fawn survival, and winter range mule deer densities. SEGMENT OBJECTIVES I. Recover remaining GPS collars to address the final year of adult female mule deer habitat use and behavioral patterns in 4 study areas experiencing varying levels of energy development and response to habitat treatments as mitigation in the Piceance Basin, northwest Colorado. 2. Develop winter utilization distributions using mule deer GPS data to inform future development planning for the Piceance Basin mule deer winter range. 3. Monitor vegetation responses from habitat treatments for assessing efficacy of habitat improvement projects to mitigate energy development disturbances to mule deer. 4. Continue to evaluate large herbivore use of habitat treatments during summer/fall using remote camera sampling. INTRODUCTION Extraction of natural gas from areas throughout western Colorado has raised concerns among many public stakeholders and Colorado Parks and Wildlife (CPW) that the cumulative impacts associated with this intense industrialization will dramatically and negatively affect the wildlife resources of the region. Concern is especially high for mule deer due to their recreational and economic importance as a principal game species and their ecological importance as one of the primary herbivores of the Colorado Plateau Ecoregion. Extraction of natural gas will directly affect the potential suitability of the landscape used by mule deer through conversion of native habitat vegetation with drill pads, roads, or introduction of noxious weeds, by fragmenting habitat with drill pads and roads, by increasing noise levels via compressor stations and vehicle traffic, and by increasing the year-round presence of human activities. Extraction will indirectly affect deer by increasing the human work-force population of the region resulting in the need for additional landscape conversion for human housing, supporting businesses, and upgraded road/transportation infrastructure. Additionally, increased traffic on rural roads will raise the potential for vehicle-animal collisions. Thus, research documenting these relationships and evaluating the most effective strategies for minimizing and mitigating these activities will greatly enhance future management efforts to sustain mule deer populations for future recreational and ecological values. 3 �The Piceance Basin in northwest Colorado contains one of the largest migratory mule deer populations in North America and also covers some of the largest natural gas reserves in North America. Projected energy development throughout northwest Colorado within the next 20 years is expected to reach about 15,000 wells, many of which will occur in the Piceance Basin, which currently supports over 250 active gas well pads (http://cogcc.state.co.us; Fig. I). Anderson and Freddy (2008) in their long-term research proposal identified 6 primary study objectives to assess measures to offset impacts of energy extraction on mule deer population performance. During the first 5 years of this study, we gathered baseline habitat utilization and demographic data from radiocollared deer across the Piceance Basin to allow assessment of habitat mitigation approaches that were completed April 2013. We recently completed monitoring 2 control areas: I with development (0.6 pads & facilities/km 2; Ryan Gulch) and I without (North Ridge). The control areas will be compared with 2 treatment areas experiencing similarly contrasting development intensities (South Magnolia, 0.9 well pads & facilities/km 2 and North Magnolia, 0.1 well pads & facilities/km 2), that also received habitat improvements (604 acres each). Habitat and mule deer responses to mechanical habitat treatments will be evaluated through spring 2019 to assess the success of this habitat mitigation strategy to benefit mule deer exposed to energy development disturbance. In addition, mule deer behavioral patterns in relation to energy development activities in the area have been monitored to identify effective Best Management Practices (BMPs) for future energy development planning. This progress report describes the previous IO years (Jan 2008-June 2018) of mule deer population performance during the pre and post-treatment phases on 4 winter range herd segments. This includes monitoring habitat selection, migration and behavior patterns of adult female mule deer; parturition success; spring/summer neonate, overwinter fawn and annual adult female survival; estimates of adult female body condition during early and late winter; and annual late-winter abundance/density estimates. The final year of GPS data collection from existing mule deer collars, vegetation response to habitat treatments and summer/fall use of treatment sites is ongoing and anticipated for completion by December 2019. STUDY AREAS The Piceance Basin, located between the cities of Rangely, Meeker, and Rifle in northwest Colorado, was selected as the project area due to its ecological importance as home to one of the largest migratory mule deer populations in North America and because it exhibits one of the highest natural gas reserves in North America (Fig. I). Historically, mule deer numbers on winter range were estimated between 20,000-30,000 (White and Lubow 2002), and the current number of well pads (Fig. I) and projected number of gas wells in the Piceance Basin over the next 20 years is about 250 and 15,000, respectively. Mule deer winter range in the Piceance Basin is predominantly characterized as a topographically diverse pinion pine (Pinus edu/is)-Utahjuniper (Juniperus osteosperma; pinionjuniper) shrubland complex ranging from 1,675 m to 2,285 min elevation (Bartmann and Steinert 1981 ). Pinion-juniper are the dominant overstory species and major shrub species include Utah serviceberry (Amelanchier utahensis), mountain mahogany (Cercocarpus montanus), bitterbrush (Purshia tridentata), big sagebrush (Artemisia tridentata), Gamble's oak (Quercus gambelii), mountain snowberry (Symphoricarpos oreophilus), and rabbitbrush (Chrysothamnus spp.; Bartmann et al. 1992). The Piceance Basin is segmented by numerous drainages characterized by stands of big sagebrush, saltbush (Atrip/ex spp.), and black greasewood (Sarcobatus vermiculatus), with the majority of the primary drainages having been converted to mixed-grass hay fields. Grasses and forbs common to the area consist ofwheatgrass (Agropyron spp.), blue grama (Bouteloua gracilis), needle and thread (Stipa comata), Indian rice grass (Oryzopsis hymenoides), arrowleafbalsamroot (Balsamorhiza sagittata), broom snakeweed (Gutierrezia sarothreae), pinnate tansymustard (Descurainia pinnata), milkvetch (Astragalus spp.), Lewis flax (Linum lewisii), evening primrose (Oenothera spp.), skyrocket gilia (Gi/ia aggregata), buckwheat (Erigonum spp.), Indian paintbrush (Castilleja spp.), and penstemon (Penstemon spp.; Gibbs 1978). The climate of the Piceance Basin is characterized by warm 4 �dry summers and cold winters with most of the annual moisture resulting from spring snow melt and brief summer monsoonal rainstorms. 2 Wintering mule deer population segments we are investigating include: North Ridge (53 km ) just north of the Dry Fork of Piceance Creek including the White River in the northeastern portion of the Basin, Ryan Gulch ( 141 km2) between Ryan Gulch and Dry Gulch in the southwestern portion of the Basin, North Magnolia (79 km 2) between the Dry Fork of Piceance Creek and Lee Gulch in the north-central portion of the Basin, and South Magnolia (83 km2) between Lee Gulch and Piceance Creek in the south-central portion of the Basin (Fig. 1). Each of these wintering population segments has received varying levels of natural gas development: no development in North Ridge, light development in North Magnolia (0.1 pads & facilities/km 2), and relatively high development in the Ryan Gulch (0.6 pads & facilities/km 2) and South Magnolia (0.9 pads & facilities/km 2) segments (Fig. 1). Development activity was high through 2011 and into 2012, but has declined substantially since natural gas prices began to decline by fall 2012. Among the 4 study areas, North Ridge has served as an unmanipulated control site, Ryan Gulch will serve to address human-activity management alternatives (BMPs) that benefit mule deer exposed to energy development and as a developed control area for comparison to the developed treatment area receiving habitat improvements (South Magnolia), and North and South Magnolia will allow us to assess the utility of habitat treatments intended to enhance mule deer population performance in areas exposed to light (North Magnolia) and relatively high (South Magnolia) energy development activities. METHODS Tasks addressed this period include adult female mule deer captures to remove store-on-board GPS collars, delineate winter mule deer concentration areas using GPS locations during a non-drilling, production phase (2012 - 2018), continue spring and fall vegetation measurements to address vegetation response to habitat treatments completed spring 2013, and document summer/fall use of habitat treatment sites using remote camera sampling. We employed helicopter net-gunning techniques (Barrett et al. 1982, van Reenen 1982) to capture and remove GPS collars from adult females during early March 2018. Once netted, all deer were hobbled and blind folded. Adult females were transported to localized handling sites for recording body measurements and remove GPS collars (5 fix attempts/day; 021 lOD, Advanced Telemetry Systems, Isanti, MN, USA) prior to release. GPS collars were supplied with timed drop-off mechanisms scheduled to release July 2019 for recovery from adult females that were not captured during March capture efforts. All radio-collars were equipped with mortality sensing options (i.e., increased pulse rate following 8 hrs of inactivity). Mule Deer Habitat Use and Movements We downloaded and summarized data from GPS collars deployed and recovered since 2008. OPS collars maintained the same schedule of attempting to collect locations every 5 hours, except for 40 does in Ryan Gulch and 10 control deer from North Ridge where location rates were programmed for every 30-60 minutes to increase resolution of movement data for evaluation of deer behavior patterns in relation to differing development activities. Joe Northrup (CSU PhD Candidate) analyzed resource selection data relative to energy development (Northrup 2015) and those results are addressed below. We contracted with Western Ecosystems Technology, Inc. (Laramie, WY USA) to develop winter utilization distributions from adult female mule deer GPS locations collected during a post-drilling production phase (2012-2018) in the Piceance Basin, using the Brownian bridge movement modeling approach described by Home et al. (2007). 5 �Mule Deer Survival Mule deer mortality monitoring consisted of daily ground-telemetry tracking and aerial monitoring approximately every 2 weeks from fixed-wing aircraft on winter range and weekly aerial monitoring on summer range. Once a mortality signal was detected, deer were located and necropsied to assess cause of death (Stonehouse et al. 2016). We estimated weekly survival using the staggered entry Kaplan-Meier procedure (Kaplan and Meier I 958, Pollock et al. 1989). Capture-related mortalities (any doe/fawn mortalities occurring within 10 days of capture; excluding neonates) and collar failures were censored from survival rate estimates. We estimated annual survival rates from 1 July through 30 June for adult females, from birth to mid-December for neonates, and from early December-mid June for winter fawns. V Adult Female Body Measurements We applied ultrasonography techniques described by Stephenson et al. (1998, 2002) and Cook et al. (200 I) to measure maximum subcutaneous rump fat (mm), loin depth (longissimus dorsi muscle, mm), and to estimate % ingesta-free body fat. We estimated a body condition score (BCS) for each deer by palpating the rump (Cook et al. 2001, 2007, 2009). We examined differences (P < 0.05) in nutritional status among study areas and between years evident in non-overlapping 95% confidence intervals. We considered differences in body condition meaningful when mean rump fat or% body fat differed statistically between comparisons. Other body measurements recorded included pregnancy status (pregnant, barren) via ultrasonography and blood samples, fetal counts using ultrasonography, weight (kg), chest girth (cm), and hind-foot length (cm). Abundance Estimates We conducted 3-5 helicopter mark-resight surveys (2 observers and the pilot) during late March/early April to annually estimate deer abundance in all 4 study areas. We delineated study area boundaries from GPS locations collected on winter range during the first 3 years of the study (Jan 2008 through April 20 I I). Two aerial fixed-wing telemetry surveys/study area were conducted during helicopter mark-resight surveys to determine which marked deer were within each survey area, and we confinned adult female locations during surveys from GPS data acquired the following April each year. We delineated flight paths in ArcGIS 10.0 prior to surveys following topographic contours (e.g., drainages, ridges) and approximating 500-600 m spacing throughout each study area; flight paths during surveys were followed using GPS navigation in the helicopter. Two 12 x 12 cm pieces of Ritchey livestock banding material (Ritchey Livestock ID, Brighton, CO USA) were uniquely marked using color, number, and symbol combinations and attached to each radio-collar to enhance markresight estimates. Each deer observed during surveys was recorded as mark ID#, unmarked, or unidentified mark. We used program MARK (White and Burnham 1999), applying the immigration-emigration mixed logit-nonnal model (Mcclintock et al. 2008), to estimate mule deer abundance and confidence intervals. For mark-resight model evaluations, we examined parameter combinations of varying detection rates with survey occasion and whether individual sighting probabilities (i.e., individual heterogeneity) were constant or varied (cr2 = 0 or* 0). Model selection procedures followed the infonnation-theoretic approach of Burnham and Anderson (2002). 6 V �RESULTS AND DISCUSSION Deer Captures and Survival The helicopter crew captured 56 does during March 2019 to remove store-on-board GPS collars. No recapture myopathies occurred during March capture efforts. Forty-four does remain collared with GPS collars programmed to drop during July 2019. The remaining collars will be located and recovered during summer/fat I 2019. Data collection for estimating deer survival rates ended July 2018. Survival results for the duration of the study are reported in the next 2 paragraphs. Fawn survival from early December 2017 through mid-June 2018 was more variable than in past years ranging from 0.34 to 0. 76 (Table 1, Fig. 2). Based on CI overlap, North Ridge and North Magnolia fawns exhibited higher survival than Ryan Gulch fawns and North Ridge fawn survival was also higher than in South Magnolia {Table 1). The range in winter fawn survival was unusual in comparison to previous years (Fig. 2), but correlated with similar variability observed in December fawn weights (Fig. 3); early winter fawn condition likely contributes to over-winter survival potential, and reduced condition the past 2 years is likely related to the lower survival rates observed recently from Ryan Gulch and South Magnolia (Fig. 2, Fig. 3). Premature collar drop during 2008-09 and 2009-10 did not allow for winter fawn survival estimates past late March, but survival rates among study areas were similar (P < 0.05) each year and comparable to 20 I 1-12 and 2012-13 (excluding North Ridge) during 2008-09 and to the higher survival rates from 2013-14 and 2014-15 during 200910 (Fig. 2). General comparisons to previous years suggest moderate to high fawn survival occurred during most winters and study areas with the exception of winter 2010-2011 for 3 of the 4 study areas, North Ridge during winter 2012-13 and 2015-16, and Ryan Gulch and South Magnolia the past 2 years (Fig. 2). Low winter fawn survival (Fig. 2) appears to correlate with summer forage condition evident from lower December fawn weights (Fig. 3); severe winter conditions can also strongly influence winter fawn survival, but winter conditions during this study have been mild to moderate with the exception of winter 2010-11, which may have more strongly influenced winter fawn survival that year. Annual adult female survival varied from 0.73 (North Ridge) to 0.87 (South Magnolia; Table 1) during 2017-18, but was comparable among study areas and to previous years (P > 0.05), with the exception of lower survival in North Magnolia during 2011-12 (S = 0.68, Anderson and Bishop 2012). Relatively low sample sizes per study area for adult female survival do not allow statistical discrimination among years unless large differences are evident (e.g.,> 15-20%). Estimates below 80% are biologically concerning if these values represent the respective population, but low statistical power precludes confinnation within study areas. When combined among study areas, annual survival estimates have varied from 79% in 2012-13 to 86% in 2014-15, but consistent CI overlap including large sample sizes (exceeding 100 in July and 200 in Mar annually) supports consistent annual doe survival during this study. Adult female mule deer exhibit consistently high survival rates unless extreme weather events and/or habitat degradation persists, which has not been evident since 2008. Mule Deer Body Condition Data collection to address mule deer body condition ended March 2018. Body condition results for the duration of the study are reported in the next 3 paragraphs. Early-winter body condition measurements of adult female mule deer during December 2017 were collectively relatively low among study areas compared to previous years and Ryan Gulch does 7 �exhibited the lowest condition estimates among study areas (P < 0.05; Fig. 4, Table 2). Although fall body condition is likely related to spring/summer forage conditions, doe condition is also influenced by energy expended for fawn rearing, and appears to be strongly influenced by lactation status; I observed a strong correlation between December lactation rate and body condition (mm rump fat; P = 0.004, r = 0.62, n = 20). Thus, while fall body condition represents an index of nutritional status entering winter, it also appears to be a useful metric to assess fall reproductive status, where low fat levels represent high fall fawning rates; the low December fat levels observed from Ryan Gulch does during 2017 was associated with highest fall lactation rate (0.63) recorded during the study. In contrast, late winter condition appears more strongly related to winter severity and winter-range forage conditions and low fat levels observed during December do not necessarily manifest into poor late winter condition (Fig. 4, Table 2). Late winter doe condition among study areas this past winter was comparable to the long-term average (Fig. 4), and was reflective of mild to moderate winter conditions consisting of infrequent snowstorms and minimal snow pack on winter range. December 2017 fawn weights by study area represented a gradient in fawn condition ranging from low to high for Ryan Gulch and North Ridge, respectively (Fig. 3), which corresponded to winter fawn survival (Fig. 2). Overall fawn condition for 3 of 4 study areas (excluding North Ridge) has declined the past 2 years (Fig. 3) and may be related to changes in summer forage conditions (further analyses pending). Because adult female body condition has been largely uninformative in regards to habitat treatment responses (pending further analyses), we began late winter fawn recaptures in South Magnolia (habitat treatment area) and Ryan Gulch (reference area) to assess changes in over-winter condition. Weight loss during winter 2015-16 was significantly less (P < 0.001) for fawns from the area receiving habitat treatments than for fawns from the untreated area, but no net weight loss was detected during the following 2 winters for either study area (P ~ 0.396). Vegetation measurements from treatment and control sites indicate recent summer/fall use of shrubs potentially negating forage benefits on winter range. Additional investigations to address this issue will be conducted summer/fall 2018 and 2019 to confirm summer/fall use of treatment sites and whether or not intended forage benefits on habitat treatment sites persist on winter range. V Mule Deer Population Estimates Data collection to address annual mule deer abundance/density estimates ended March 2018. Population size/density results for the duration of the study are reported in the paragraph below. Mark-resight models that best predicted abundance estimates (lowest AICc; Burnham and Anderson 2002) exhibited variable sightability across surveys (P,) for all study areas and variable individual sightability (cr2 = 0) for North Magnolia deer and homogenous sightability (cr2 ;/; 0) for the other 3 areas. During 2018, North Ridge exhibited the highest deer density (15.8/km2), with comparable but lower deer densities in the other 3 areas (9.2-l l .3/km2 ; Table 3, Fig. 5). Abundance estimates from 2018 were similarly precise from all 4 study areas with the mean Confidence Interval Coefficient of Variation (CICV) ranging from 0.12--0.17 (Table 3). Densities increased over the first 8year monitoring period in all study areas ranging from an estimated 50% increase in North Ridge to a 103% increase in North Magnolia (mean estimated increase across study areas= 78%); North Ridge deer appeared to decline during 2012 and 2013, but subsequently increased, while the other 3 areas exhibited consistent and similar rates ofincrease from 2009-2016 (mean annual increase= 0.064; Fig. 5). Excluding the North Ridge study area, late winter mule deer densities have apparently stabilized since 2016 (Fig. 5). The reason for decline since 2016 for the North Ridge deer population is unclear and not completely explained by demographic parameters monitored during the study. Erratic population estimates observed from North Ridge may be partially attributed to lack of geographic closure more commonly associated with this study area (primarily from earlier spring migration 8 V �timing). Population vital rates will be analyzed and compared to abundance estimates to assess factors contributing to population change by study area. Spring Migration Patterns Collaboration with Idaho State University to address mule deer migration patterns in developed and undeveloped landscapes (funded from energy company contributions) has been completed. Four manuscripts from this effort have been published (Lendrum et al. 2012, Lendrum et al. 2013, Lendrum et al. 2014, Anderson and Bishop 2014; Appendix A). In addressing habitat selection during spring migration, Lendrum et al. (2012; Fig. 6) noted that mule deer migrating through the most developed landscapes exhibited longer step lengths (straight line distance between GPS locations) and selected habitats providing greater security cover than deer in undeveloped landscapes that migrated through more open areas that provided increased foraging opportunities. Migrating deer also selected areas closer to well pads, but avoided roads, except in the highest developed areas where road densities were likely too high for avoidance without significant deviations from traditional migration routes. In the second manuscript, Lendrum et al. (2013) addressed biological and environmental factors influencing spring migration and assessed how energy development influenced migratory behavior. Overall, spring migration was influenced by snow depth, temperature, and green-up on winter and summer range; increasing temperatures, snow melt and emerging vegetation dictated timing of winter range departure and summer range arrival. Duration of Piceance Basin mule deer migration was short, with median migration durations of 3-8 days among the 4 areas (straight-line distance between seasonal ranges averaged 32-40 km). Deer in poor condition migrated later than deer in good condition, but condition was similar among areas regardless of development status. Migrating deer from developed study areas did not avoid development structures, but departed later, arrived earlier and migrated more quickly than deer from undeveloped areas. While large changes in timing of migration could have nutritional consequences and negatively influence reproduction and neonate survival, the relatively minor shift we observed should not result in long-tenn fitness consequences, which was supported by Peterson et al. (2018; see Reproductive Success and Neonate Survival below). Migratory deer in the Piceance Basin appear to avoid negative effects of energy development through behavioral shifts in timing and rate of migration. In the third publication Lendrum et al. (2014) monitored migratory mule deer in the Piceance Basin to examined the relationship between the Nonnalized Difference Vegetation Index (NOVI), which is a course-scale measure of forage quality using a G IS assessment of vegetation greenness, and fecal nitrogen to assess the assumption that forage quality and deer diets can be reasonably linked to address deer habitat use patterns from remotely sensed data. We found that diet quality evident from fecal nitrogen and course measures of vegetation green-up were infonnative, and that Piceance Basin mule deer exhibited rapid migration (3 to 8 days depending on study area), left winter range following snow melt with lowest fecal N and NOVI values, and progressed to summer range as vegetation green-up and nitrogen levels increased, but ahead of peak vegetation green-up on summer range. I suspect this rapid migration strategy is evident for deer in relatively good condition and allows for early arrival on summer range to take advantage optimal forage conditions prior to parturition. Anderson and Bishop (2014) summarized results from Lendrum et al. (2012, 2013) and Sawyer et al. (2012) addressing migratory mule deer and energy development in northwest Colorado and south-central Wyoming, respectively. The interactions between migratory mule deer and energy development identified by Lendrum et al. (2012, 2013) and Sawyer et al. (2012) suggest mule deer may benefit from energy development planning by considering thresholds of development that may 9 �alter migratory behavior. It appears that migration rate, migration routes, and stopover use, if present, may be altered at high development intensities. In addition, migratory mule deer may benefit by maintaining security cover along migration paths, and improved habitat conditions may facilitate more direct and rapid migration requiring less energy to complete migration. Enhancing permeability along migration routes by applying dispersed development plans (<2 well pads/km 2) and minimizing disturbance to vegetation types by maintaining security cover should reduce impacts to migratory mule deer as well as other migratory ungulates. Where feasible, habitat improvement projects on winter range and possibly stopover sites would also enhance migratory mule deer populations by increasing energy reserves for long-distance movements and parturition shortly after summer range arrival. Where possible, directional drilling could be used to extract energy resources from underneath migration routes while maintaining no surface occupancy. Lastly, we emphasize that GPS studies now allow managers to accurately map migration routes for entire populations and identify relatively narrow corridors that are most heavily used thus allowing for the identification of the most important corridors for migrating ungulates. Where available, we encourage agencies to incorporate such migration corridors into land-use plans (e.g., resource management plans) and National Environmental Policy Act documents. V Mule Deer Behavioral Response to Energy Development We completed evaluations of deer behavior patterns in relation to energy development activities (Northrup et al. 2015). We found diurnal responses to development activity, where deer used timbered areas away from development activity while bedded during the day and moved into more open areas generally closer to developed areas while foraging at night. Disturbance distances from producing pads and roads declined from 600 m to 200 m and about 140 m to 60 m from daytime to nighttime, respectively, but increased from 600 m to 800 m for nighttime drilling pad activity (pad response depicted in Fig. 7). We suspect deer behaviorally respond to fluctuations in development activity, where road traffic and producing well pad activity decline at night, but drilling pad disturbance may increase from compressors and lights used to facilitate nighttime drilling activity. These evaluations were applied during an active drilling phase in the Piceance Basin and deer use was influenced by development activity in 25% (nighttime) to 50% (daytime) of critical winter range during that period. However, deer densities have comparably increased among developed and undeveloped study areas (excluding North Ridge; Fig. 5) suggesting that deer can behaviorally mediate development disturbance under observed development and deer densities by taking advantage of fluctuations in development activity to address their nutritional requirements. Given the plasticity in deer behavior, a number of potential options for future development planning ex.its including drilling schedule modifications (seasonal and/or diurnal), concentrated/staged development, reducing road traffic, and using light/noise barriers around drill rigs. It will be interesting to determine if habitat improvements will further reduce development disturbance and increase management options for future development planning. Reproductive Success and Neonate Survival To complete demographic parameters addressing mule deer-energy development interactions, CPW, Colorado State University, and ExxonMobil Production entered into a collaborative agreement to investigate reproductive success (Peterson et al. 2017), including pregnancy rates (early Mar) and fetal survival (Mar until birth), and early fawn survival (0 - 6 months; Peterson et al. 2018) in developed and relatively undeveloped landscapes beginning spring 2012 and continuing through Dec 2014. We applied statistical models to address reproductive success under contrasting energy development scenarios and noted that pregnancy and in utero fetal rates (early Mar; n = 346) were high (0.948, SE= 0.012 and 1.877, SE= 0.029, respectively) and statistically indistinguishable between study areas. Fetal survival (n = 383), however, was lower (P < 0.05) in the developed study IO V �area during 1 of 3 years (2012; Fig. 8) when drought conditions were present, suggesting the combination of severe weather conditions and development activity under observed conditions may influence fetal survival. There was no apparent influence from energy development in 0--6 month fawn survival (n = 184) based on similar mortality rates between study areas; mean daily mortality probabilities from predation, malnutrition and unknown causes were nearly identical (Fig. 9). These results suggest that natural gas development did not exert measureable influence on mule deer pregnancy rates, fetal rates or early fawn survival, but may have negatively influenced fetal survival during 2012 when does were exposed to drought conditions during the third trimester. These findings are consistent with developed areas in a production phase (little to no drilling activity) exhibiting moderate pad densities (0.4-0.9 pads/km2), and relationships may differ in areas of higher pad densities and/or drilling activity. Winter Range Habitat Treatments and Habitat Utilization Distributions We completed 116 acres of pilot habitat treatments in January 2011 (Anderson and Bishop 2011; Environmental Assessment: DOI-BLM-CO-110-2011-004-EA), 54 acres of mechanical treatment method comparison treatments (hydro-ax, roller-chop, chain) in January 2012 (Stephens 2014), and 1,038 acres of hydro-ax treatments in April 2013 (Determination ofNEPA Adequacy: DO1-BLM-CO-I I 0-2012-0134-DNA), totaling 604 treated acres in each study area (Fig. 10). Vegetation response in the pilot treatment sites was visually evident by fall 2011 (Fig. 10), and resulted in statistically significant (P < 0.05) increases in native grass and forb cover by the 2014 growing season. Final results are pending, but shrub responses appear promising from data collected through spring 2019. Stephens (2014) reported that all 3 mechanical treatment methods compared resulted in roughly a 3-fold increase in grasses, forbs, and shrubs combined after 2 growing seasons (versus control sites), but cautioned that rollerchop treatments may be more vulnerable to invasive species response. Vegetative responses from 2013 hydro-ax treatments were visually evident following 1 growing season and shrub responses have been notable during the 4 th growing season, but statistical comparisons are still pending. As anticipated, grass and forb responses were evident 2 to 3 years posttreatment, with longer term response expected (3-5 years) for palatable shrubs. Of note, relatively high moisture conditions experienced during spring 2014 and 2015 resulted in higher than normal prevalence of cheatgrass (Bromus tectorum); cheatgrass invasion has previously been minor to non-existent in this area. Cheatgrass invasion, however, does not appear directly related to treatment sites because occurrence is evident in both treatment and control areas. We anticipate this outbreak will subside based on past competitive advantage of native species to dominate, but we will continue to monitor species composition and address cheatgrass persistence in treatment and control sites. OPS data addressing deer use of treatment sites has been collected through March 2019, with remaining collars from the Dec 2017 sample (n = 44) still on deer in the field. The remaining collars will be collected during summer/fall 2019 (collars were programmed to drop July 2019). The final spring vegetation response measurements for habitat treatment and control areas were collected th is past spring (final data analyses pending) and final shrub response data will be collected Sep. 2019. Final data analyses will be initiated once OPS collars are collected this summer/fall. Thus far, we observed improved fawn condition (P < 0.001) in South Magnolia following the 4 th growing season of habitat treatments when compared to fawn condition in the Ryan Gulch control area, but we did not detect a response the following 2 winters. Ongoing data analyses suggests that fall shrub condition appears to have declined recently indicating that summer/fall shrub use may be increasing and potentially inhibiting the intended benefit of habitat treatments on winter range. We deployed remote cameras on treatment and control areas July 2018 and 2019 to further address summer/fall use and identify species utilizing treatment sites on mule deer winter range. Although results are 11 �preliminary, vegetation responses through the first 4 years post treatment provided the intended forage benefit and there is some evidence that fawn condition improved as a result. Recent changes in habitat use by multiple species (potentially including wild horses and livestock) may have reduced winter forage benefits recently, but additional data collection and analyses will be necessary for confirmation. Analyses of doe use of treatment sites throughout the study are still pending and will provide information addressing the utility of habitat improvement projects as a mitigation technique to offset energy development disturbance on mule deer winter range. We delineated mule deer winter concentration areas for each study area from 2012 - 2018, which represents a non-drilling (drilling activity had ended by fall 2012) production phase (active well pads producing natural gas) in the Piceance Basin. Results exhibiting high use areas during winter are reported in Appendix B. This information is intended to provide guidance for site selection in future development planning (i.e., placement of well pads, facilities and roads). Directional drilling technology should provide options for development activity to occur in areas of relatively low use on mule deer winter range. Ultimately, we will expand our analyses to address mule deer winter concentration areas throughout the entire Piceance Basin (and possibly other mule deer pinion-juniper winter ranges), but final results will not be available until 2021. SUMMARY AND COLLABORATIONS The long-term goal of this study is to investigate habitat treatments and energy development practices that enhance mule deer populations exposed to extensive energy development activity. The information presented here summarizes mule deer population parameters from the 10-year study period, with the final year of data collection for some parameters (i.e., habitat treatment response and adult female habitat use) remaining. The pretreatment period was completed by spring 2013, providing baseline data for comparison with intended improvements in habitat conditions and response to varying degrees in human development activity. Winter range habitat improvements resulting in 604 acres of mechanically treated pinion-juniper/mountain shrub habitats in each of2 study areas were completed by April of 2013, and subsequent vegetation responses have met or exceeded expectations through 2016. The post-treatment monitoring period was completed June 2018, with the final year of habitat use and habitat treatment response data collection still pending. Based on final (migration, mule deer behavioral responses, reproductive success and neonate survival) and preliminary data analyses for this 10-year project: (I) annual adult female survival was consistent among areas averaging 79-87% annually, but overwinter fawn survival was variable, ranging from 31 % to 95% within study areas, with annual and study area differences primarily due to early winter fawn condition, annual weather conditions, and factors associated with predation on winter range; (2) mule deer body condition early and late winter was generally consistent within areas, with higher variability among study areas early winter, primarily due to December lactation rates, and late winter condition related to seasonal moisture and winter severity; (3) late winter mule deer densities increased through 2016 in all study areas, ranging from 50% in North Ridge to 103% in North Magnolia, but have stabilized recently in 3 of the 4 study areas with recent decline evident in North Ridge; (4) migratory mule deer selected for areas with increased cover and increased their rate of travel through developed areas, and avoided negative influences through behavioral shifts in timing and rate of migration, but did not avoid development structures; (5) mule deer exhibited behavioral plasticity in relation to energy development, where disturbance distance varied relative to diurnal extent and magnitude of development activity, which may provide for several options in future development planning; and (6) energy development activity under existing conditions did not influence pregnancy rates, fetal rates or early fawn survival (0-6 months}, but may have reduced neonatal survival (March until birth) when drought conditions persisted during the third trimester of doe parturition. Final results are pending to address vegetation and mule deer responses to assess habitat treatment mitigation options for energy development planning, and final results addressing the 12 V �interaction of mule deer behavioral and demographic factors associated with energy development activity have been submitted for scientific review and publication. Completion of this project, including final data collection, analyses and interpretation ofresults, is anticipated by fall/winter 2020-21. Hay field improvements were completed during 2012 in the North Magnolia study area by WPX Energy to fulfill a Wildlife Management Plan (WMP) agreement with CPW; rapid and continued elk (Cervus e/aphus) use of these areas was evident, but mule deer response has been minor. A similar WMP agreement between ExxonMobil/XTO Energy and CPW allowed completion and continued monitoring of mechanical habitat improvements in the Magnolia study areas. Collaborative research with agency biologists, graduate students, and university professors has produced 19 scientific publications addressing improved monitoring techniques for neonate mule deer captures (Bishop et al. 2011, Peterson et al. 2018b); mule deer migration (Lendrum et al. 2012, 2013, 2014; Anderson and Bishop 2014), improved approaches to address animal habitat use patterns (Northrup et al. 2013); mule deer response to helicopter capture and handling (Northrup et al. 2014a); potential effects of male-biased harvest on mule deer productivity (Freeman et al. 2014 ); mule deer genetics in relation to body condition and migration (Northrup et al. 2014b); spatial and temporal factors influencing auditory vigilance in mule deer (Lynch et al. 2014); the relationship of plant phenology with mule deer body condition (Seral et al. 2015); approaches to identify cause-specific mortality in mule deer from field necropsies (Stonehouse et al. 2016); the influence of individual and temporal factors affecting late winter body condition estimates of adult female mule deer (Bergman et al. 2018); and mule deer behavioral and demographic responses to energy development activities to inform future development planning (Northrup et al. 2015, 2016a, 2016b, Peterson et al. 2017, 2018a). These publications are summarized in Appendix A and results describing mule deer concentration areas among study areas are reported in Appendix B. We anticipate the opportunity to work cooperatively toward developing solutions for allowing the nation's energy reserves to be developed in a manner that benefits wildlife and the people who value both the wildlife and energy resources of Colorado. 13 �LITERATURE CITED Anderson, C.R., Jr., and D. J. Freddy. 2008. Population perfonnance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Final Study Plan, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and C. J. Bishop. 201 I. Population perfonnance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Division of Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and C. J. Bishop. 2012. Population perfonnance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Parks and Wildlife, Ft. Collins, CO, USA. Anderson, C.R., Jr., and C. J. Bishop. 2014. Migration patterns ofadult female mule deer in response to energy development. Pages 47-50 in Transactions of the 79 th North American Wildlife & Natural Resources Conference (R. A. Coon & M. C. Dunfee, eds.). Wildlife Management Institute, Gardners, PA, USA. ISSN 0078-1355. Bartmann, R. M. 1975. Piceance deer study-population density and structure. Job Progress Report, Colorado Division of Wildlife, Fort Collins, Colorado, USA. Bartmann, R. 8., and S. F. Steinert. 1981. Distribution and movements of mule deer in the White River Drainage, Colorado. Special Report No. 51, Colorado Division of Wildlife, Fort Collins, Colorado, USA. Bartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monograph No. 121. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108-114. Bergman, E. J., C.R. Anderson Jr., C. J. Bishop, A. A. Holland, and J.M. Northrup. 2018. Variation in ungulate body fat: individual versus temporal effects. Journal of Wildlife Management 82:130-137, DOI: 10:1002/jwmg.21334 Bishop, C. J., C.R. Anderson Jr., D. P. Walsh, E. J. Bergman, P. Kuechle, and J. Roth. 2011. Effectiveness of a redesigned vaginal implant transmitter in mule deer. Journal of Wildlife Management 75(8):1797-1806; DOI: 10.1002/jwmg.229 Burnham, K. P., and D.R. Anderson. 2002. Model selection and multi-model inference: a practical information-theoretic approach. Second edition. Springer-Verlag, New York, New York, USA. Cook, R. C., J. G. Cook, D. L. Murray, P. Zager, B. K. Johnson, and M. W. Gratson. 2001. Development of predictive models of nutritional condition for rocky mountain elk. Journal of Wildlife Management 65:973-987. Cook, R. C., T. R. Stephenson, W. L. Meyers, J. G. Cook, and L.A. Shipley. 2007. Validating predictive models of nutritional condition for mule deer. Journal of Wildlife Management 71: 1934-1943. Cook, R. C., J. G. Cook, T. R. Stephenson, W. L. Meyers, S. M. McCorquodale, D. J. Vales, L. L. Irwin, P. Briggs Hall, R. D. Spencer, S. L. Murphie, K. A. Schoenecker, P. J. Miller. 2009. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife Management 74:880-896. Freeman, E. D., R. T. Larsen, M. E. Peterson, C.R. Anderson, Jr., K. R. Hersey, and B. R. McMillan. 2014. Effects of male-biased harvest on mule deer: implications for rates of pregnancy, synchrony, and timing of parturition. Wildlife Society Bulletin; DOI: 10.1002/wsb.450 Gibbs, H. D. 1978. Nutritional quality of mule deer foods, Piceance Basin, Colorado. Thesis, Colorado State University, Fort Collins, Colorado, USA. 14 V V �Home, J. S., E. 0. Garton, S. M Krone, and J. S. Lewis. 2007. Analyzing animal movements using Brownian bridges. Ecology 88:2354-2363. Kaplan, E. L., and P. Meier. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 52:457-481. Lendrum, P. E., C.R. Anderson, Jr., R. A. Long, J. K. Kie, and R. T. Bowyer. 2012. Habitat selection by mule deer during migration: effects of landscape structure and natural gas development. Ecosphere 3(9):82. http://dx.doi.org/10. Lendrum, P. E., C.R. Anderson, Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2013. Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes. PLoS ONE 8(5): e64548. doi: 10.1371/joumal.pone.0064548 Lendrum, P. E., C.R. Anderson, Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2014. Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns of forage quality. Mammalian Biology: http://dx.doi.org/10.10 I6/i .mambio.2014.05.005 Lynch, E., J.M. Northrup, M. F. McKenna, C.R. Anderson Jr., L. Angeloni, and G. Wittemyer. 2014. Landscape and anthropogenic features influence the use of auditory vigilance by mule deer. Behavioral Ecology; doi: 10.1093/beheco/aru 158. McCiintock, B. T., G. C. White, K. P~ Burnham, and M. A. Pride. 2008. A generalized mixed effects model of abundance for mark-resight data when sampling is without replacement. Pages 271- 289 in D. L. Thompson, E.G. Cooch, and M. J. Conroy, editors, Modeling demographic processes is marked populations. Springer, New York, New York, USA. Northrup, J.M. 2015. Behavioral response of mule deer to natural gas development in the Piceance Basin. Dissertation, Colorado State University, Fort Collins, USA. Northrup, J.M., M. B. Hooten, C.R. Anderson, Jr., and G. Wittemyer. 2013. Practical guidance on characterizing availability in resource selection functions under a use-availability design. Ecology 94(7): 1456-1463. Northrup, J.M., C.R. Anderson, Jr., and G. Wittemyer. 2014a. Effects of helicopter capture and handling on movement behavior of mule deer. Journal of Wildlife Management 78(4):731738; DOI: 10.1002/jwmg.705 Northrup, J.M., A. B. Shafer, C.R. Anderson Jr., D. W. Coltman, and G. Whittemyer. 2014b. Finescale genetic correlates to condition and migration in a wild cervid. Evolutionary Applications ISSN 1752-4571; doi: 10.1111/eva.12189 Northrup, J.M., C. R. Anderson, Jr., and G. Wittemyer. 2015. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer. Global Change Biology, doi: 10.1111/gcb.13037. Northrup, J. M., C. R. Anderson, Jr., M. B. Hooten, and G. Wittemyer. 2016. Movement reveals scale dependence in habitat selection of a large ungulate. Ecological Applications 26:2746-2757 Northrup, J.M., C.R. Anderson, Jr., and G. Wittemyer. 2016. Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid. Diversity and Distributions 22: 547-557, DOI: 10.1111/ddi.12417 Peterson, M. E., C.R. Anderson Jr., J.M. Northrup, and P. F. Doherty Jr. 2017. Reproductive success of mule deer in a natural gas development area. Wildlife Biology doi: 10.1111/wlb.00341 Peterson, M. E., C.R. Anderson Jr., J.M. Northrup, and P. F. Doherty Jr. 2018. Mortality of mule deer fawns in a natural gas development area. Journal of Wildlife Management 82:1135-1148, DOI: 10.1002/jwmg.21476 Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. C. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Sawyer, H., M. J. Kauffman, A. D. Middleton, T. A. Morrison, R. M. Nielson, and T. B. Wycoff. 2012. A framework for understanding semi-permeable barrier effects on migratory ungulates. Journal of Applied Ecology 50:68-78. 15 �Searle, K. R., M. B. Rice, C.R. Anderson, C. Bishop and N. T. Hobbs. 2015. Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore. Oecologia ISSN 0029- 8549; DOI IO. 1007/s00442-015-3348-9 Stephens, G. J. 2014. Understory responses to mechanical removal of pinyon-juniper overstory. MS Thesis, Colorado State University, Ft. Collins USA. Stephenson, T. R., V. C. Bleich, 8. M. Pierce, and G. P. Mulcahy. 2002. Validation of mule deer body composition using in vivo and post-mortem indices of nutritional condition. Wildlife Society Bulletin 30:557-564. Stephenson, T. R., K. J. Hundertmark, C. C. Swartz, and V. Van Ballenberghe. 1998. Predicting body fat and mass in moose with untrasonography. Canadian Journal of Zoology 76:717-722. Stonehouse, K. F., C. R. Anderson Jr., M. E. Peterson, and D. R. Collins. 2016. Approaches to field investigations of cause-specific mortality in mule deer ( Odocoileus hemionus). Colorado Parks and Wildlife Technical Report No. 48, First Edition, 317 W. Prospect Rd., Ft. Collins, CO USA. DOW-R-T-48-16, ISSN 0084-8883. Unsworth, J. W., D. F. Pack, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. Van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-421 in L. Nielsen, J.C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked individuals. Bird Study 46: 120-139. White, G. C., and B. C. Lubow. 2002. Fitting population models to multiple sources of observed data. Journal of Wildlife Management 66:300-309. Prepared by_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ Charles R. Anderson, Jr., Mammals Research Leader V 16 �Table I. Survival rate estimates (S) of fawn (I Dec. 2017-15 June 2018) and adult female (1 July 201730 June 2018) mule deer from 4 winter range study areas of the Piceance Basin in northwest Colorado. Cohort Study area Initial sample size (n) March doe samplea (n) 8(95% CI) Fawns Ryan Gulch 53 0.335 (0.207-0.464) South Magnolia 53 0.479 (0.343-0.616) North Magnolia 57 0.665 (0.543-0.788) North Ridge 58 0. 757 (0.645-0.868) Adult females '..,,,_/ Ryan Gulch 27 52 0.832 (0.722-0.942) South Magnolia 32 56 0.869 (0. 757-0.980) North Magnolia 27 49 0.781 (0.65~.913) North Ridge 24 45 0. 731 (0.583--0.878) aAdult female sample sizes following capture and radio-collaring efforts March, 2018. 17 �Table 2. Mean rump fat (mm) and % ingesta-free body fat8 (% fat) of adult female mule deer from 4 study areas in the Piceance Basin of northwest Colorado, March and December, 2009-2018. Values in parentheses= SD. March 2009 Study Area Rump fat Ryan Gulch 1.73(1.78) 7.08 (1.27) South Magnolia %fat December 2009 Rump fat %fat March 2010 December 20 I 0 Rump fat %fat Rump fat %fat 8.35 (6.36) 10.54 (3.72) 2.31 (1.44) 6.37 (1.41) 7.26 (6.36) 9.69 (3.56) 1.29 (0.47) 6.74 (2.27) 10.05 (6.19) 11.44 (3.50) 3.12 (2.20) 7.11 ( 1.69) 9.85 (6.78) 11.27 (3.75) North Magnolia 1.31 (1.01) 7.15 (1.63) 10.67 (5.76) 11.94 (3.39) 3.15 (2.34) 7.54 (1.53) 9.55 (6.49) 10. 79 (4.26) North Ridge 1.57 (1.22) 6.81 (1.68) 5.25 (5.65) 1.77 (I. I I) 6.39 (1.45) 7.25 (5.41) 9.85 (3.02) 9.37 (3.08) Table 2. Continued. March 2011 Study Area Rump fat Ryan Gulch 1.55 (0.60) 6.72 (1.37) South Magnolia %fat December 2011 Rump fat %fat March 2012 December 2012 Rump fat %fat Rump fat %fat 13.41 (6.39) 13.17 (3.64) 2.15 (1.44) 7.22 (1.16) 6.34 (4.35) 9.34 (2.43) 1.65 (0.75) 6.15 (1.75) 8.18 (5.45) 10.34 (3.28) 1.66 (0.77) 7.03(1.13) 8.30 (5.71) 10.32 (3.23) North Magnolia 1.65 (0.67) 6.79 (1.47) 8.76 (5.76) 10.73 (3.14) 1.90 (0.76) 7.61 (0.96) 9.66 (6.41) I 1.18 (3.64) North Ridge 1.45 (0.76) 6.30 (1.65) 8.86 (5.65) 10.77 (3.33) 2.24 (1.58) 7.26 (1.05) 5.76 (4.10) 9.06 (2.31) 15 C ( ( �( ( Table 2. Continued. March 2013 Study Area Rump fat Ryan Gulch 1.87 (0.90) 7.14 (0.89) South Magnolia %fat December 2013 Rump fat %fat March 2014 December 2014 Rump fat %fat Rump fat % fat 9.27 (6.29) 10.61 (3.76) 1.69 (0.85) 7.03 (0.99) 8.50 (6.76) 10.56 (3.70) 2.06 (0.77) 7.19 (0.66) 11.27 (8.40) 11.40 (4.16) 2.57 (1.61) 7.75 (0.68) 10.96 (6.82) 11.98 (3.81) North Magnolia 1.76 (0.91) 6.87 (1.11) 9.00 (6.15) 10.48 (3.25) 2.33 (2.12) 7.31 (1.43) 9.52 (5.83) 11.18 (3.32) North Ridge 1.87 (0.73) 6.70 (1.12) 11.17 (5.28) 11.66 (2.69) 2.38 (1.52) 7. 16 (1.14) 7.93 (5.50) 10.20 (3.01) Table 2. Continued. March 2015 Study Area Rump fat % fat Ryan Gulch 2.62 (0.95) 7.44 (0.53) South Magnolia December 2015 Rump fat %fat March 2016 December 2016 Rump fat %fat Rump fat %fat 12.80 (6.83) 12.89 (3.72) 2.29 (0.64) 7.29 (0.52) 8.20 (4.90) 10.46 (2.70) 2.66 (1.36) 7.62 (0.74) 6.93 (4.83) 9.83 (2.69) 2.07 (1.39) 7.46 (0.93) 6.27 (4.62) North Magnolia 2.25 (0.97) 7.49 (0.90) 8.79 (6.01) 10.81 (3.54) 2.43 (1.01) 7.17 (0.87) 7.90 (5.52) 10.34 (3.14) North Ridge 2.28 (1.37) 7.43 (1.05) 5.47 (5.49) 1.58 (0.70) 6.73 (1.26) 7.74 (5.48) 10.01 (3.09) 9.35 (2.75) 16 9.37 (2.53) �Table 2. Continued. March 2017 December 2017 March 2018 Study Area Rump fat %fat Rump fat %fat Rump fat %fat Ryan Gulch 2.39 (0.74) 6.78 (0.97) 4.47 (3.57) 8.62 (1.80) 2.13 (0.76) 7.40 (0.50) South Magnolia 2.48 (0.77) 7.09 (0.63) 6.67 (5.23) 9.56 (2.73) 2.19 (1.18) 7.40 (0.72) North Magnolia 1.82 (0.72) 7.05 (0.58) 6.16 (4.32) 9.23 (2.47) 1.87 (0.63) 7.15(1.11) North Ridge 2.30 (1.37) 7.23 (1.21) 6.60 (4.29) 9.38 (2.35) 2.35 (0.80) 7.73 (1.03) 8 lngesta-free body fat calculated following Cook et al. (2009). 17 C ( �Table 3. Mark-resight abundance (N) and density estimates of mule deer from 4 winter range herd segments in the Piceance Basin, northwest Colorado, 26-31 March 2018. Data represent 4 helicopter resight surveys from 3 of 4 study areas, with South Magnolia receiving 5 surveys. Study area Mean No. sighted Mean No. marked N(95% Cl) Density (deer/km 2) Ryan Gulch 418 19 1,397 ( 1, 186-1 ,674) 9.9 South Magnolia 239 26 764 (678-874) 8.1 North Magnolia 319 27 899 (774-1,070) 9.8 North Ridge 305 32 838 (748-954) 15.8 18 �Mule Deer Winter Range Study Areas Mule deer study areas Well Pads & Facilities D Nonn Magnolia Soutn Magnoha ! 1n development ► Pr0<1uc1ng well _ Oevelopm ent lac11,1,es 10 M ilts Figure I. Mule deer winter range study areas relative to active natural gas well pads and energy development facilities in the Piceance Basin of northwest Colorado. winter 20 13/14 (Accessed http://cogcc.state.co.us/. Dec.31.201 3). Development activity has subsided with minimal drilling activity since fall 20 12. 19 �Winter fawn survival 2010-11- 2017-18 1.00 ~T , j j,__ Ji j~ j 0.90 0.80 - I- I- I- I- I-- I-- 1-- 1-- 0.70 0.60 0.50 ....... 0.40 0.30 0.20 0.10 0.00 T ' T I- '- I- I- '- I-- 1-- I- I- I- I- I- I-- '- □ South Mag noli3 '- I I- I-- 1-- I-- I- 1-- I- I- '- 1-- ._ I•' I- 1-- I- I- '-- '- I-- I- - I- - I- ,, ~- -~ ' - f.l a Ryan Gulch ■ North Magnolia - □ North Ridge Ii 2010-11 2011-12 2012-13 2013-14 2014-15 2015-16 2016-17 2017-18 Figure 2. Over-winter (Dec- June) mule deer fawn survival (5) from 4 study areas in the Piceance Basin. northwest Colorado. 20 I 0-11 - 2017- 18. ErTor bars = 95% C l. Fawn survival estimates past late March unavai lable for winter 2008-09 and 2009-1 0 due to premature collar drop. but survival estimates were comparable to 20 11-1 2 and 20 12- 13 (excl uding 2012-1 3 North Ridge) during 2008-09 and to the hi gher survival rates from 20 I3- 14 and 20 I 4- 15 during 2009- I 0. 20 �Male fawn weights 42.0 40.0 ~ .JII 38.0 Q0 - ... ='=- 36.0 .t:: ._ ·a; 34.0 3: 32.0 _11 I 30.0 ~ ,. lrll! ~ - - - - -.... I- I- I- '- I- I- I- '- .... .... L- L- ~ ,__ .,_ ,- - ..... 28.0 Dec 2008 Dec 2009 Dec 2010 Dec 2011 -- I• Dec 2012 Dec 2013 Dec 2014 Dec 2015 Dec 2016 D Ryan Gulch ... D South Magnolia ■ North M agnolia D North Ridge Dec 2017 Female fawn weights 42.0 40.0 Q0 38.0 ... D Rvan Gulch ='=- 36.0 .t:: D south Magnolia -~ 34.0 3: ■North Magnolia 32.0 D Nonh Ridgo 30.0 28.0 Dec 2008 Dec 2009 Dec 2010 Dec 2011 Dec 2012 Dec 2013 Dec 2014 Dec 2015 Dec 2016 Dec 2017 Figure 3. Mean male and female fawn weights and 95% CI (error bars) from 4 mule deer study areas in the Piceance Basin. northwest Colorado. December 2008- 20 17. 21 �Early winter rump fat {mm) 16 14 12 10 8 6 4 2 0 Dec 2009 Dec 2010 Dec 2011 Dec 2012 Dec 2013 Dec 2014 Dec 2015 Dec 2016 Dec 2017 ■ North Ridge ■ North Magnolia ■ Ryan Gulch ■ South Magnolia Late winter rump fat {mm) 4.0 3.5 3.0 2.5 2.0 II 1.5 1.0 0.5 0.0 Mar 2009 Mar 2010 Mar 2011 Mar 2012 Mar 2013 Mar 2014 Mar 2015 Mar 2016 Mar 2017 Mar 2018 ■ North Ridge ■ North Magnolia ■ Ryan Gulch ■ South Magnolia Figure 4. Mean early (early Dec., Top) and late winter (early Mar .. Bottom) body condition (mm rump fat) of adult female mule deer from 4 winter range study areas in the Piceance Basin of northwest Colorado. March 2009-March 10 18. E1Tor bars = 95% C l. �Piceance Basin late winter mule deer density 35.00 30.00 25.00 } 20.00 ";::- - t 15.00 C - North Ridge • • • • • • Ryan Gulch 10.00 5.00 - • North Magnolia - South Magnolia 0.00 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Year Figure 5. Mule deer density estimates and 95% CI (error bars) from 4 winter range herd segments in the Piceance Basin, northwest Colorado, late winter 2009-2018. Estimates for were adjusted upward ( using GPS migration data) to account for early migration from winter range prior to and during surveys during years when early migration biased estimates low (North Ridge 2014-2017, North Magnolia 2015 and 2017, Ryan Gulch 2017). 24 -------------- ------- �North Ridge and North Magnolia Summer Range Figure 6. Mule deer study areas in the Piceance Basin of northwestern Colorado. USA (Top). spring 2009 migration routes of adult female mule deer (11 = 52; Lower left). and active natural-gas well pads (black dots) and roads (state, county. and natural-gas; white lines) from May 2009 (Lower right: from Lendrum et al. 2012). 25 �j O _.,t·- - - - · - - - -+- - - - ... - - - - -.. - - - - - , ,- - - t" ~ 'E .... - I~ 13 --+----+ I ' e 8 7M I I I T I I J Prod 400 Prod 600 Prod 800 Prod 1000 Drill 400 Covariates I Drill 600 OriO 800 Drill 1000 Prod 400 Prod 600 Prod 800 Prod 1000 Drill 600 Drill 800 Drlll 1000 C") I Drill 400 Covariates Figure 7. Posterior distributions of population-level coefficients related to natural gas development for RSF models during the (a) day and (b) night for 53 adult female mule deer in the Piceance Basin, Northwest Colorado. Dashed line indicates 0 selection or avoidance of the habitat features. 'Drill' and 'Prod' refer to well pads where there was active drilling or producing pads, respectively. The numbers following 'Drill' or ·Prod' represent the concentric buffer over which the number of well pads was calculated (e.g., 'Drill 600' is the number of well pads with active drilling between 400-600 m from the deer location; from Northup et al. 2015). 26 �,......,_ 1.00 - '-- 0.80 Q) --~ ·, I, ~ ro 0.60 > -~ :::, (/) cu 0.40 Q) u. 0.20 0.00 2013 2012 2014 Year □ High development □ Low development I Figure 8. Model-averaged estimates of fetal s urvi val (± 95% Cl) of mule deer fetuses from early March until birth (late May- June) in high and low energy development study areas of the Piceance Basin, northwest Colorado, 20 12-20 14 (from Peterson et al. 2017). f;' 0.016 iii ,::: 0 E 0 0.012 ~ :a m .g 0.008 5. ~ 7il Cl 0.004 development Low development development Low development development Low development 2012 2012 2013 2013 2014 2014 High High High Study area □ Predation c Malnutrition ■ Unknown mortality Figure 9. Mean daily probability of death by predation. malnutrition, or unknown mortality(± 95% Cl) of mule deer fawns (0 to 6 months old) in high and low energy development study areas of the Piceance Basin Colorado, 2012-2014 (from Peterson et al. 2018). 27 �Norlh M agnoba tr<alement sates (587 aoes) C Bea,Sel_ I 5_35b_andG Bea,Sel _ I _BandA_E c::__, 8 earSe1_36_54andJ GreasewoodSe1_g I 6_929 GreasewoodSe1_g1 _915 I GreasewooaSe1_g30_!)'l2 LeeOvers,ghts_a_tand 16_ 17 Med'lanrc,1 treatment compansoo 15-1 acres) - - Norlh Hatch Pilot Treatments 111 6 acres ) South Magnolia Figure I 0. Habitat treatment site delineations in 2 mule deer study areas (604 acres each) of the Piceance Basin. northwest Colorado (Top: cyan polygons completed Jan. 201 I using hydro-axe; yellow polygons completed Jan. 20 I 2 using hydro-axe, roller-chop. and chaining: and remaining polygons completed April 2013 using hydro-axe). January 2011 hydro-axe treatment-site photos from North Hatch Gulch during Apri l (Lower left. aerial view) and October. 20 I I (Lower ri ght. ground view). 25 �Appendix A. Abstracts of published manuscripts resulting from Piceance Basin mule deer/energy development interaction research collaborations. Abstract format specific to the respective journal requirements. Effectiveness of a redesigned vaginal implant transmitter in mule deer CHAD J. BISHOP', CHARLES R. ANDERSON Jr. 1, DANIEL P. WALSH 1, ERIC J. BERGMAN 1, PETER KUECHLE 2, and JOHN ROTH2 'Colorado Parks and Wildlife, Fort Collins, Colorado 80526 USA 2 Advanced Telemetry Systems, Isanti, Minnesota 55040 USA Citation: Bishop, C. J., C. R. Anderson Jr., D. P. Walsh, E. J. Bergman, P. Kuechle, and J. Roth. 2011. Effectiveness of a redesigned vaginal implant transmitter in mule deer. Journal of Wildlife Management 75(8): 1797-1806; DOI: I0.1002~wmg.229 '.._I ABSTRACT Our understanding of factors that limit mule deer (Odocoileus hemionus) populations may be improved by evaluating neonatal survival as a function of dam characteristics under free-ranging conditions, which generally requires that both neonates and dams are radiocollared. The most viable technique facilitating capture of neonates from radiocollared adult females is use of vaginal implant transmitters (VITs). To date, VITs have allowed research opportunities that were not previously possible; however, VITs are often expelled from adult females prepartum, which limits their effectiveness. We redesigned an existing VIT manufactured by Advanced Telemetry Systems (ATS; Isanti, MN) by lengthening and widening wings used to retain the VIT in an adult female. Our objective was to increase VIT retention rates and thereby increase the likelihood of locating birth sites and newborn fawns. We placed the newly designed VITs in 59 adult female mule deer and evaluated the probability of retention to parturition and the probability of detecting newborn fawns. We also developed an equation for determining VIT sample size necessary to achieve a specified sample size of neonates. The probability of a VIT being retained until parturition was 0.766 (SE= 0.0605) and the probability ofa VIT being retained to within 3 days of parturition was 0.894 (SE = 0.0441 ). In a similar study using the original VIT wings (Bishop et al. 2007), the probability of a VIT being retained until parturition was 0.447 (SE= 0.0468) and the probability of retention to within 3 days of parturition was 0.623 (SE= 0.0456). Thus, our design modification increased VIT retention to parturition by 0.319 (SE= 0.0765) and VIT retention to within 3 days of parturition by 0.271 (SE= 0.0634). Considering dams that retained VITs to within 3 days of parturition, the probability of detecting at least I neonate was 0.952 (SE= 0.0334) and the probability of detecting both fawns from twin litters was 0.588 (SE = 0.0827). We expended approximately 12 person-hours per detected neonate. As a guide for researchers planning future studies, we found that VIT sample size should approximately equal the targeted neonate sample size. Our study expands opportunities for conducting research that links adult female attributes to productivity and offspring survival in mule deer.© 2014 The Wildlife Society. Habitat selection by mule deer during migration: effects of landscape structure and natural-gas development PATRICK E. LENDRUM', CHARLES R. ANDERSON JR. 2, RY AN A. LONG 1, JOHN G. KIE 1, AND R. TERRY BOWYER' 'Department of Biological Sciences, Idaho State University, Pocatello, Idaho 83209 USA 2cotorado Parks and Wildlife, Grand Junction, Colorado 81505 USA Citation: Lendrum, P. E., C.R. Anderson Jr., R. A. Long, J. G. Kie, and R. T. Bowyer. 2012. Habitat selection by mule deer during migration: effects of landscape structure and natural-gas development. Ecosphere 3(9):82 http·//dx.doi.org/10.1890/ES 12-00165.1 Abstract. The disruption of traditional migratory routes by anthropogenic disturbances has shifted patterns of resource selection by many species, and in some instances has caused populations to decline. Moreover, in recent decades populations of mule deer (Odocoileus hemionus) have declined throughout much of their historic range in the western United States. We used resourceselection functions to determine if the presence of natural-gas development altered patterns of resource selection by migrating mule deer. We compared spring migration routes of adult female mule deer fitted with GPS collars (n = 167) among four study areas that had varying degrees of natural-gas development from 2008 to 2010 in the Piceance Basin of northwest Colorado, USA. Mule deer migrating through the most developed area had longer step lengths (straight-line distance between successive GPS locations) compared with deer in less developed areas. Additionally, deer migrating through the most developed study areas tended to select for habitat types that provided greater amounts of concealment cover, whereas deer from the least developed areas tended to select habitats that increased access to forage and cover. Deer selected habitats closer to well pads and avoided roads in all instances except along the most highly developed migratory routes, where road densities may have been too high for deer to avoid roads without deviating substantially from established migration routes. These results indicate that behavioral tendencies toward avoidance of anthropogenic disturbance can be overridden during migration by the strong fidelity ungulates demonstrate towards migration routes. If avoidance is feasible, then deer may select areas further from development, whereas in highly developed areas, deer may simply increase their rate of travel along established migration routes. 26 �Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes Patrick E. Lendrum', Charles R. Anderson Jr.2, Kevin L Monteith 1,J, Jonathan A. Jenks", R. Terry Bowyer' 1 Department ofBiological Sciences, Idaho State University, Pocatello, Idaho, USA, 2 Colorado Division of Parks and Wildlife, Grand Junction, Colorado, USA, 3 Wyoming Cooperative Fish and Wildlife Research Unit, University of Wyoming, Laramie, Wyoming, USA.4 Department of Natural Resource Management, South Dakota State University, Brookings, South Dakota, USA Citation: Lendrum, P. E., C.R. Anderson Jr., K. L. Monteith, J. A. Jenks, R. T. Bowyer. 2013. Migrating Mule Deer: Effects of anthropogenically Altered Landscapes. PLoS ONE 8(5): e64548. DOI: I0.13711oumal.pone.0064548 Abstract Background: Migration is an adaptive strategy that enables animals to enhance resource availability and reduce risk of predation at a broad geographic scale. Ungulate migrations generally occur along traditional routes, many of which have been disrupted by anthropogenic disturbances. Spring migration in ungulates is of particular importance for conservation planning, because it is closely coupled with timing of parturition. The degree to which oil and gas development affects migratory patterns, and whether ungulate migration is sufficiently plastic to compensate for such changes, warrants additional study to better understand this critical conservation issue. Met/1odology/Prindpal Findings: We studied timing and synchrony of departure from winter range and arrival to summer range offemale mule deer (Odocoileus hemionus) in northwestern Colorado, USA, which has one of the largest natural-gas reserves currently under development in North America. We hypothesized that in addition to local weather, plant phenology, and individual life-history characteristics, patterns of spring migration would be modified by disturbances associated with natural-gas extraction. We captured 205 adult female mule deer, equipped them with GPS collars, and observed patterns of spring migration during 2008-20 I0. Conclusions/Signijicance: Timing of spring migration was related to winter weather (particularly snow depth) and access to emerging vegetation, which varied among years, but was highly synchronous across study areas within years. Additionally, timing of migration was influenced by the collective effects of anthropogenic disturbance, rate of travel, distance traveled, and body condition of adult females. Rates of travel were more rapid over shorter migration distances in areas of high natural-gas development resulting in the delayed departure, but early arrival for females migrating in areas with high development compared with less-developed areas. Such shifts in behavior could have consequences for timing of arrival on birthing areas, especially where mule deer migrate over longer distances or for greater durations. Practical guidance on characterizing availability in resource selection functions under a use-availability design JOSEPH M. NORTHRUP', MEVIN B. HOOTEN 1,2,J, CHARLES R. ANDERSON JR.", AND GEORGE WITTEMYER1 1 Department offish, Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 2 U.S. Geological Survey, Colorado Cooperative Fish and Wildlife Research Unit, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 3 Colorado State University, Department of Statistics, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 4 Mammals Research Section Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction, Colorado 81505 USA Citation: Northrup, J. M., M. B. Hooten, C. R. Anderson Jr., and G. Wittemyer. 2013. Practical guidance on characterizing availability in resource selection functions under a use-availability design. Ecology 94(7): 1456-1463. http:l/dx.doi.org/l0.1890/12-1688. I Abstract. Habitat selection is a fundamental aspect of animal ecology, the understanding of which is critical to management and conservation. Global positioning system data from animals allow fine-scale assessments of habitat selection and typically are analyzed in a useravailability framework, whereby animal locations are contrasted with random locations (the availability sample). Although most u~availability methods are in fact spatial point process models, they often are fit using logistic regression. This framework offers numerous methodological challenges, for which the literature provides little guidance. Specifically, the size and spatial extent of the availability sample influences coefficient estimates potentially causing interpretational bias. We examined the influence of availability on statistical inference through simulations and analysis of serially correlated mule deer GPS data. Bias in estimates arose from incorrectly assessing and sampling the spatial extent of availability. Spatial autocorrelation in covariates, which is common for landscape characteristics, exacerbated the error in availability sampling leading to increased bias. These results have strong implications for habitat selection analyses using OPS data, which are increasingly prevalent in the literature. We recommend that researchers assess the sensitivity of their results to their availability sample and, where bias is likely, take care with interpretations and use cross validation to assess robustness. 27 �Effects of Helicopter Capture and Handling on Movement Behavior of Mule Deer JOSEPH M. NORTHRUP 1, CHARLES R. ANDERSON JR2, AND GEORGE WITTEMYER 1 1Department of Fish, Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 2 Mammals Research Section Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction, Colorado 81505 USA Citation: Northrup, J. M., C. R. Anderson Jr., and G. Wittemyer. 2014. Effects of helicopter capture and handling on movement behavior of mule deer. Journal of Wildlife Management 78(4):731-738; DOI: 10.1002/jwmg.705 ABSTRACT Research on wildlife movement, physiology, and reproductive biology often requires capture and handling of animals. Such invasive treatment can alter behavior, which may bias results or invalidate assumptions regarding representative behaviors. To assess the impacts of handling on mule deer (Odocoileus hemionus), a focal species for research in North America, we investigated pre- and post-recapture movements of collared individuals, and compared them to deer that were not recaptured (controls). We compared pre- and post-recapture movement rates (m/hr) and 24-hour straight-line displacement among recaptured and control deer. In addition, we examined the time it took recaptured deer to return to their pre-recapture home range. Both daily straight-line displacement and movement rate were marginally elevated relative to monthly averages for 24 hours following recapture, with non-significant elevation continuing for up to 7 days. Comparing movements averaged over 30 days before and after recapture, we found no differences in displacement, but movement rates demonstrated seasonal effects, with faster movements post- relative to pre-recapture in March and slower movements post- relative to pre-recapture in December. Relative to control deer movements, recaptured deer movement rates in March were higher immediately after recapture and lower in the second and third weeks following recapture. The median time to return to the pre-recapture home range was 13 hours, with 71% of deer returning in the first day, and 91% returning within 4 days. These results indicate a short period of elevated movements following recaptures, likely due to the deer returning to their home ranges, followed by weaker but non-significant depression of movements for up to 3 weeks. Censoring of the first day of data post capture from analyses is strongly supported, and removing additional days until the individual returns to its home range will control for the majority of impacts from capture. © 2014 The Wildlife Society. ~ Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns offorage quality Patrick E. Lendrum•, Charles R. Anderson Jr.\ Kevin L Monteithc, Jonathan A. Jenksd, R. Terry Bowyer8 • Department of Biological Sciences, Idaho State University, 921 South 8th Avenue, Stop 8007, Pocatello 83209, USA b Mammals Research Section Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction 81505, USA "Wyoming Cooperative Fish and Wildlife Research Unit, Department of Zoology and Physiology, University of Wyoming, 3166, 1000 East University Avenue, Laramie 82071, USA d Department of Natural Resource Management, South Dakota State University, Box 2140B, Brookings 57007, USA Citation: Lendrum, P. E., C.R. Anderson Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2014. Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns of forage quality. Mammalian Biology: http://dx.doi.org/l0.1016/j.mambio.2014.05.005 ABSTRACT: Migratory ungulates exhibit recurring movements, often along traditional routes between seasonal ranges each spring and autumn, which allow them to track resources as they become available on the landscape. We examined the relationship between spring migration of mule deer (Odocoileus hemionus) and forage quality, as indexed by spatiotemporal patterns offecal nitrogen and remotely sensed greenness of vegetation (Normalized Difference Vegetation Index; NOVI) in spring 2010 in the Piceance Basin of northwestern Colorado, USA. NDVI increased throughout spring, and was affected primarily by snow depth when snow was present, and temperature when snow was absent. Fecal nitrogen was lowest when deer were on winter range before migration, increased rapidly to an asymptote during migration, and remained relatively high when deer reached summer range. Values offecal nitrogen corresponded with increasing NOVI during migration. Spring migration for mule deer provided a way for these large mammals to increase access to a high-quality diet, which was evident in patterns of NOVI and fecal nitrogen. Moreover, these deer 'jumped,. rather than "surfed" the green wave by arriving on summer range well before peak productivity of forage occurred. This rapid migration may aid in securing resources and seclusion from others on summer range in preparation for parturition, and to minimize detrimental factors such as predation and malnutrition during migration. 28 �Effects of Male-Biased Harvest on Mule Deer: Implications for Rates of Pregnancy, Synchrony, and Timing of Parturition ERIC D. FREEMAN', RANDY T. LARSEN', MARKE. PETERSON2, CHARLES R. ANDERSON JR.3, KENT R. HERSEY', AND BROCK R. McMILLAN' 1 Department of Plant and Wildlife Sciences, Brigham Young University, 275 WIDB, Provo, lIT 84602, USA 2 Department offish, Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA 3 Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction, CO 81 SOS, USA 4 Utah Division of Wildlife Resources, 1594 W North Temple, Salt Lake City, lIT 84114, USA Citation: Freeman, E. D., R. T. Larsen, M. E. Peterson, C.R. Anderson Jr., K. R. Hersey, and B. R. McMillan. 2014. Effects of male-biased harvest on mule deer: implications for rates of pregnancy, synchrony, and timing of parturition. Wildlife Society Bulletin; DOI: IO. I002/wsb.450 ABSTRACT Evaluating how management practices influence the population dynamics of ungulates may enhance future management of these species. For example, in mule deer (Odocoileus hemionus), changes in male/female ratio due to malebiased harvest may alter rates of pregnancy, timing of parturition, and synchrony of parturition if inadequate numbers of males are present to fertilize females during their first estrous cycle. If rates of pregnancy or parturition are influenced by decreased male/female ratios, recruitment may be reduced (e.g., fewer births, later parturition resulting in lower survival of fawns, and a less synchronous parturition that potentially increases susceptibility of neonates to predation). Our objectives were to compare rates of pregnancy, synchrony of parturition, and timing of parturition between exploited mule deer populations with a relatively high (Piceance, CO, USA; 26 males/JOO females) and a relatively low (Monroe, UT, USA; 14 males/lO0 females) male/female ratio. We detennined rates of pregnancy via ultrasonography and timing of parturition via vaginal implant transmitters. We found no differences in rates of pregnancy (98.6% and 96.6%; z = 0.821; P = 0.794), timing of parturition (estimate= 1.258; SE= 1.672; t = 0.752; P = 0.454), or synchrony of parturition (F = 1.073; P = 0.859) between Monroe Mountain and Piceance Basin, respectively. The relatively low male/female ratio on Monroe Mountain was not associated with a protracted period of parturition. This finding suggests that relatively low male/female ratios typical of heavily harvested populations do not influence population dynamics because recruitment remains unaffected.© 2014 The Wildlife Society. Fine-scale genetic correlates to condition and migration in a wild cervid Joseph M. Norlhrup, 1 Aaron B. A. Shafer,2 Charles R. Anderson Jr/ David W. Coltman4 and George Wittemyer 1 I Department offish, Wildlife, and Conservation Biology, Colorado State University, Fort Collins, CO, USA 2 Department of Evolutionary Biology, Evolutionary Biology Centre, Uppsala University, Uppsala, Sweden 3 Mammals Research Section, Colorado Parks and Wildlife, Grand Junction, CO, USA 4 Department ofBiological Sciences, University of Alberta, Edmonton, AB, Canada. Citation: Northrup, J.M., A. B. Shafer, C.R. Anderson Jr., D. W. Colbnan, and G. Whittemyer. 2014. Fine-scale genetic correlates to condition and migration in a wild cervid. Evolutionary Applications ISSN 1752-4571; doi: 10.1111/eva.12189 Abstract The relationship between genetic variation and phenotypic traits is fundamental to the study and management of natural populations. Such relationships often are investigated by assessing correlations between phenotypic traits and heterozygosity or genetic differentiation. Using an extensive data set compiled from free ranging mule deer (Odocoileus hemionus), we combined genetic and ecological data to (i) examine correlations between genetic differentiation and migration timing, (ii) screen for mitochondrial haplotypes associated with migration timing, and (iii) test whether nuclear heterozygosity was associated with condition. Migration was related to genetic differentiation (more closely related individuals migrated closer in time) and mitochondrial haplogroup. Body fat was related to heterozygosity at two nuclear loci (with antagonistic patterns), one of which is situated near a known fat metabolism gene in mammals. Despite being focused on a widespread panmictic species, these findings revealed a link between genetic variation and important phenotypes at a fine scale. We hypothesize that these correlations are either the result of mixing refugial lineages or differential mitochondrial haplotypes influencing energetics. The maintenance of phenotypic diversity will be critical to enable the potential tracking of changing climatic conditions, and these correlates highlight the need to consider evolutionary mechanisms in management, even in widely distributed panmictic species. 30 �Landscape and anthropogenic features influence the use of auditory vigilance by mule deer Emma Lynch,• Joseph M. Northrup,b Megan F. McKenna,c Charles R. Anderson Jr,d Lisa Angeloni,a.e and George Wittemye.,..i, Graduate Degree Program in Ecology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA ~partment of Fish, Wildlife and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA ~atural Sounds and Night Skies Division, National Parle Service, 1201 Oakridge Drive, Fort Collins, CO 80525, USA, dMammals Research Section, Colorado Pruks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526, USA ceepartment of Biology, Colorado State University, 1878 Campus Delivery, Fort Collins, CO 80523, USA 1 Citation: Lynch, E., J.M. Northrup, M. F. McKenna, C.R. Anderson Jr., L. Angeloni, and G. Wittemyer. 2014. Landscape and anthropogenic features influence the use of auditory vigilance by mule deer. Behavioral Ecology~ doi: I0.1093/beheco/aru 158. While visual fonns of vigilance behavior and their relationship with predation risk have been broadly examined, animals also employ other vigilance modalities such as auditory vigilance by listening for the acoustic cues of predators. Similar to the tradeoffs associated with visual vigilance, auditory behavior potentially structures the energy budgets and behavior of animals. The cryptic nature of auditory vigilance makes it difficult to study, but on-animal acoustical monitoring has rapidly advanced our ability to investigate behaviors and conditions related to sound. We utilized this technique to investigate the ways external stimuli in an active natural gas development field affect periodic pausing by mule deer (Odocoileus hemionus) within bouts of rumination-based mastication. To better understand the ecological properties that structure this behavior, we investigate spatial and temporal factors related to these pauses to detennine if results are consistent with our hypothesis that pausing is used for auditory vigilance. We found that deer paused more when in forested cover and at night, where visual vigilance was likely to be less effective. Additionally, deer paused more in areas of moderate background sound levels, though responses to anthropogenic features were less clear. Our results suggest that pauses during rumination represent a form of auditory vigilance that is responsive to landscape variables. Further exploration of this behavior can facilitate a more holistic understanding of risk perception and the costs associated with vigilance behavior. Migration Patterns of Adult Female Mule Deer in Response to Energy Development Charles R. Anderson Jr. and Chad J. Bishop Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526, USA Citation: Anderson, C. R., Jr., and C. J. Bishop. 2014. Migration patterns of adult female mule deer in response to energy development. Pages 47-50 in Transactions of the 7'11' North American Wildlife & Natural Resources Conference (R. A. Coon & M. C. Dunfee, eds.). Wildlife Management Institute, Gardners, PA, USA. ISSN 0078-1355. Migration is an adaptive strategy that enables animals to enhance resource availability and reduce risk of predation at a broad geographic scale. Ungulate migrations generally occur along traditional routes, many of which have been disrupted by anthropogenic disturbances. Spring migration in ungulates is of particular importance for conservation planning because it is closely coupled with timing of parturition. The degree to which oil and gas development affects migratory patterns, and whether ungulate migration is sufficiently prepared to compensate for such changes, has recently been investigated in Colorado and Wyoming (Lendrum et al. 2012, 2013; Sawyer et al. 2012). Lendrum et al. (2012, 2013) and Sawyer et al. (2012) address mule deer (Odocoileus hemionus) migration patterns in relation to energy development from northwest Colorado and south-central Wyoming, respectively. We address results from the Colorado and Wyoming studies and then compare similarities and differences. The interactions between migratory mule deer and energy development identified by Lendrum et al. (2012, 2013) and Sawyer et al. (2012) suggest mule deer may benefit from energy development planning by considering thresholds of development that may alter migratory behavior. It appears that migration rate, migration routes, and stopover use, if present, may be altered at high development intensities. In addition, migratory mule deer may benefit by maintaining security cover along migration paths, and improved habitat conditions may facilitate more direct and rapid migration requiring less energy to complete migration. Enhancing penneability along migration routes by applying dispersed development plans (>2 well pads/km 2) and minimizing disturbance to vegetation types by maintaining security cover should reduce impacts to migratory mule deer as well as other migratory ungulates. Where feasible, habitat improvement projects on winter range and possibly stopover sites would also enhance migratory mule deer populations by enhancing energy reserves for long-distance movements and parturition shortly after summer range arrival. Where possible, directional drilling could be used to extract energy resources from underneath migration routes while maintaining no surface occupancy. Lastly, we emphasize that GPS studies now allow managers to accurately map migration routes for entire populations and identify relatively narrow corridors that are most heavily used thus allowing for the identification of the most important corridors for migrating ungulates. Where available, we encourage agencies to incorporate such migration corridors into land-use plans (e.g., resource management plans) and National Environmental Policy Act documents. 31 �Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore Kate R. Searle' • Mindy B. Rice2 • Charles R. Anderson 2 • Chad Bishop2 • N. T. Hobbs3 NERC Centre for Ecology and Hydrology, Bush Estate, Penicuik EH26 OQB. UK 2 Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80S26, USA 3 Department of Ecosystem Science and Sustainability. Colorado State University, Fort Collins 80524, CO, USA 1 Citation: Searle, K. R., M. B. Rice, C.R. Anderson, C. Bishop and N. T. Hobbs. 201S. Asynchronous vegetation phenology enhances winter body condition ofa large mobile herbivore. Oecologia ISSN 0029-8S49~ 00110.1007/s00442-015-3348-9 Abstract Understanding how spatial and temporal heterogeneity influence ecological processes fonns a central challenge in ecology. Individual responses to heterogeneity shape population dynamics, therefore understanding these responses is central to sustainable population management. Emerging evidence has shown that herbivores track heterogeneity in nutritional quality of vegetation by responding to phenological differences in plants. We quantified the benefits mule deer (Odocoileus hemionus) accrue from accessing habitats with asynchronous plant phenology in northwest Colorado over 3 years. Our analysis examined both the direct physiological and indirect environmental effects of weather and vegetation phenology on mule deer winter body condition. We identified several important effects of annual weather patterns and topographical variables on vegetation phenology in the home ranges of mule deer. Crucially, temporal patterns of vegetation phenology were linked with differences in body condition, with deer tending to show poorer body condition in areas with less asynchronous vegetation green-up and later vegetation onset. The direct physiological effect of previous winter precipitation on mule deer body condition was much less important than the indirect effect mediated by vegetation phenology. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer JOSEPH M. NORTHRUP 1, CHARLES R. ANDERSON JR.,2 and GEORGE WITTEMYER 1 • 3 1Department offish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, CO, USA 2Mammals Research Section, Colorado Parks and Wildlife, Fort Collins, CO, USA 3Graduate Degree Program in Ecology, Colorado State University, Fort Collins, CO, USA Citation: Northrup, J. M., C. R. Anderson, Jr., and G. Wittemyer. 20 IS. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer. Global Change Biology, doi: IO.I I l l/gcb.13037 Abstract Extraction of oil and natural gas (hydrocarbons) from shale is increasing rapidly in North America, with documented impacts to native species and ecosystems. With shale oil and gas resources on nearly every continent, this development is set to become a major driver of global land-use change. It is increasingly critical to quantify spatial habitat loss driven by this development to implement effective mitigation strategies and develop habitat offsets. Habitat selection is a fundamental ecological process, influencing both individual fitness and population-level distribution on the landscape. Examinations of habitat selection provide a natural means for understanding spatial impacts. We examined the impact of natural gas development on habitat selection patterns of mule deer on their winter range in Colorado. We fit resource selection functions in a Bayesian hierarchical framework, with habitat availability defined using a movement-based modeling approach. Energy development drove considerable alterations to deer habitat selection patterns, with the most substantial impacts manifested as avoidance of well pads with active drilling to a distance of at least 800 m. Deer displayed more nuanced responses to other infrastructure, avoiding pads with active production and roads to a greater degree during the day than night. In aggregate, these responses equate to alteration of behavior by human development in over 500/4 of the critical winter range in our study area during the day and over 25% at night. Compared to other regions, the topographic and vegetative diversity in the study area appear to provide refugia that allow deer to behaviorally mediate some of the impacts of development. This study, and the methods we employed, provides a template for quantifying spatial take by industrial activities in natural areas and the results offer guidance for policy makers, mangers, and industry when attempting to mitigate habitat loss due to energy development. 32 �Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid Joseph M. Northrup•, Charles R. Anderson Jr and George Wittemyer•.l 1 Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, CO, USA 2Mammals Research Section, Colorado Parks and Wildlife, Fort Collins, CO, USA Graduate Degree Program in Ecology, Colorado State University, Fort Collins, CO, USA 3 Citation: Northrup, J.M., C.R. Anderson, Jr., and G. Wittemyer. 2016. Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid. Diversity and Distributions 22: 547-557, 001: 10.111 J/ddi.12417 ABSTRACT Aim The space an animal uses over a given time period must provide the resources required for meeting energetic needs, reproducing and avoiding predation. Anthropogenic landscape change in concert with environmental dynamics can strongly structure space-use. Investigating these dynamics can provide critical insight into animal ecology, conservation and management. Location The Piceance Basin, Colorado, USA. Methods We applied a novel utilization distribution estimation technique based on a continuous-time correlated random walk model to characterize range dynamics of mule deer during winter and summer seasons across multiple years. This approach leverages secondorder properties of movement to provide a probabilistic estimate of space-use. We assessed the influence of environmental (cover and forage), individual and anthropogenic factors on interannual variation in range use of individual deer using a hierarchical Bayesian regression framework. Results Mule deer demonstrated remarkable spatial philopatry, with a median of 50% overlap (range: 8-78%) in year-to-year utilization distributions. Environmental conditions were the primary driver of both philopatry and range size, with anthropogenic disturbance playing a secondary role. Main conclusions Philopatry in mule deer is suspected to reflect the importance of spatial familiarity (memory) to this species and, therefore, factors driving spatial displacement are of conservation concern. The interaction between range behaviour and dynamics in development disturbance and environmental conditions highlights mechanisms by which anthropogenic environmental change may displace deer from familiar areas and alter their foraging and survival strategies. Movement reveals scale dependence in habitat selection of a large ungulate Joseph M. Northrup,• Charles R. Anderson Jr.,2 Mevin B. Hooten,3 and George Wittemye.-4 'Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA Mammals Research Section, Colorado Parks and Wildlife, Fort Collins, Colorado 80523 USA 3 U.S. Geological Survey, Colorado Cooperative Fish and Wildlife Research Unit, Department offish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA 4 Department offish, Wildlife and Conservation Biology and Graduate Degree Program in Ecology, Colorado State University, Fort Collins, Colorado 80523 USA 2 Citation: Northrup, J.M., C. R. Anderson, Jr., M. B. Hooten, and G. Wittemyer. 2016. Movement reveals scale dependence in habitat selection of a large ungulate. Ecological Applications 26:2746-2757 Abstract. Ecological processes operate across temporal and spatial scales. Anthropogenic disturbances impact these processes, but examinations of scale dependence in impacts are infrequent. Such examinations can provide important insight to wildlife-human interactions and guide management efforts to reduce impacts. We assessed spatiotemporal scale dependence in habitat selection of mule deer (Odocoi/eus hemionus) in the Piceance Basin of Colorado, USA, an area ofongoing natural gas development. We employed a newly developed animal movement method to assess habitat selection across scales defined using animal-centric spatiotemporal definitions ranging from the local (defined from five hour movements) to the broad (defined from weekly movements). We extended our analysis to examine variation in scale dependence between night and day and assess functional responses in habitat selection patterns relative to the density of anthropogenic features. Mule deer displayed scale invariance in the direction of their response to energy development features, avoiding well pads and the areas closest to roads at all scales, though with increasing strength of avoidance at coarser scales. Deer displayed scale-dependent responses to most other habitat features, including land cover type and habitat edges. Selection differed between night and day at the finest scales, but homogenized as scale increased. Deer displayed functional responses to development, with deer inhabiting the least developed ranges more strongly avoiding development relative to those with more development in their ranges. Energy development was a primary driver of habitat selection patterns in mule deer, structuring their behaviors across all scales examined. Stronger avoidance at coarser scales suggests that deer behaviorally mediated their interaction with development, but only to a degree. At higher development densities than seen in this area, such mediation may not be possible and thus maintenance of sufficient habitat with lower development densities will be a critical best management practice as development expands globally. 33 �Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus) Kourtney F. Stonehouse, 1.J Charles R. Anderson Jr., 1 Mark E. Peterson,•.2 and David R. Collins• 1 Mammals Research Section. Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 90S26 USA 2Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80S23 USA Citation: Stonehouse, K. F., C.R. Anderson Jr.• M. E. Peterson, and D.R. Collins. 2016. Approaches to field investigations ofcaus~specific mortality in mule deer {Odocoileus hemionus). Colorado Parks and Wildlife Technical Report No. 48, First Edition, 317 W. Prospect Rd., Ft. Collins, CO USA. DOW-R-T-48-16, ISSN 0084-8883. This technical report provides general guidelines for conducting mortality site investigations to help investigators distinguish predation from scavenging and other causes of death. General health indices are also provided to assess whether or not deer may have died from malnutrition or disease or if these factors may have predisposed deer to predation. Lastly, these guidelines will assist investigators in identifying predatory species or scavengers involved through the examination of physical evidence at deer mortality sites. The infonnation presented here is based primarily on field experience gained from a long tenn research effort in northwest Colorado investigating mule deer mortality sites over several years (http://cpw.state.co.us/leam/Pages/ResearchMammalsRP-04.aspx) and literature review where referenced. We acknowledge that proximate and ultimate cause of death can be difficult or impossible to detect from field necropsy alone and examples presented here largely represent proximate causes ofmortaJity; efforts discerning ultimate cause will require specific tissue sample collections, where possible, submitted to a veterinary diagnostic laboratory. Within this technical report are numerous photographs documenting characteristics of predator attacks on mule deer and signs left by predatory and scavenging species. Additional pictures illustrate differences between healthy and unhealthy tissues and organs. While reading this document, be aware that each mortality investigation is unique and observations in the field may differ from illustrations provided here. Appendix I provides a sample necropsy fonn to assist in conducting mortality investigations. Reproductive success of mule deer in a natural gas development area Mark E. Peterson,1 Charles R. Anderson Jr., 2 Joseph M. Northrup•, and Paul F. Doherty Jr. 1 1Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80S23 USA 2Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 90S26 USA Citation: Peterson, M. E., C. R. Anderson Jr., J.M. Northrup, and P. F. Doherty Jr. 2017. Reproductive success of mule deer in a natural gas development area. Wildlife Biology doi: IO. I l I l/wlb.00341 Abstract: Natural gas development is increasing across North America and causing concern over the potential impacts on wildlife populations and their habitat, particularly for ungulate species. Understanding how this development impacts reproductive success metrics that are influential for ungulate population dynamics is important to guide management of ungulates. However, the influences of natural gas development on reproductive success metrics of mule deer Odocoi/eus hemionus have not been studied. We used statistical models to examine the influence of natural gas development and temporal factors on reproductive success metrics of mule deer in the Piceance Basin, northwest Colorado during 2012-2014. We focused on study areas with relatively high or low levels of natural gas development. Pregnancy and in utero fetal rates were high and statistically indistinguishable between study areas. Fetal survival rates increased over time and survival was lower in the high versus low development study areas in 2012 possibly influenced by drought coupled with habitat loss and fragmentation associated with development. Our novel results suggest managers should be concerned with the influences of development on fetal survival, particularly during extreme environmental conditions (e.g. drought) and our results can be used to guide development planning and/or mitigation. Developers and wildlife managers should continue to collaborate on development planning, such as implementing habitat treatments to improve forage availability and quality, minimizing disturbance to hiding and foraging habitat particularly during parturition, and implementing directional drilling to minimize pad disturbance density to increase fetal survival in developed areas. 34 �Variation in ungulate body fat: individual versus temporal effects Eric J. Bergman,1 Charles R. Anderson Jr., 1 Chad J. Bishop, 1 A. Andrew Holland, 1 and Joseph M. Northrup1 1Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 90526 USA 2Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA Citation: Bergman, E. J., C. R. Anderson Jr., C. J. Bishop, A. A. Holland, and J.M. Northrup. 2018. Variation in ungulate body fat: individual versus temporal effects. Journal of Wildlife Management 82: 130-137, 001: I0: I002/jwmg.21334 ABSTRACT The use ofultrasonograhic measurements of muscle and body fat represent a relatively new data stream that can be used to address questions regarding ungulate condition. We have learned that measurements of body fat and presumably overall body condition among individual animals, even those taken from the same herd at that same time, are highly variable. Relatively little consideration has been given to the sources of variation in body fat and other physiological parameters in wildlife populations. We evaluated the components of variation in late-winter mule deer (Odocoileus hemionus) body fat estimates: sampling variation (i.e., variation induced by the particular set of individuals that were sampled) and process variation (i.e., variation stemming from biological processes) with a long-tenn data set (2002-2015) from Colorado, USA. We collected our data from across Colorado as part of historicaJ research, ongoing research, and periodic population monitoring programs. Mean percent ingesta-free body fat (%IFBF) for sampled mule deer was 7.20 :1: 1.20% (SD). Covariates related to individuaJ deer explained approximately 4% of the total variation in %IFBF and annual effects explained an additional 13% of the variation. Substantial residua] variation in %1FBF (83%) remained unexplained. The source of the 83% of unexplained variation is partially linked to fine-scale spatial dynamics but also additional individual metrics we were unable to capture, primarily the presence or absence of dependent young. We speculate that the primary factors influencing late-winter mule deer body fat and overall condition are individual in nature. These results present a cautionary check on herd level inference that can be made from individuaJ late-winter body fat estimates and we postulate that for mule deer, alternative and additional body condition metrics may offer added utility in management scenarios. However, an important next step to better understand wildlife population health is to evaJuate the sources and magnitude of variation within other body condition metrics, with the goal of further refining data that can better allow biologists to incorporate herd health into population management recommendations. Mortality of mule deer fawns in a natural gas development area Mark E. Peterson,' Charles R. Anderson Jr.,1 Joseph M. Northrup 1,and Paul F. Doherty Jr.' 1 Department ofFish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA 2Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 90526 USA Citation: Peterson, M. E., C.R. Anderson Jr., J.M. Northrup, and P. F. Doherty Jr. 2018. Mortality of mule deer fawns in a natural gas development area. Journal ofWildlifeManagement 82:1135-1148, 001: 10.1002/jwmg.2l476 ABSTRACT Recent natural gas development has caused concern among wildlife managers, researchers, and stakeholders over the potential effects on wildlife and their habitats. Specifically, understanding how this development and other factors influence mule deer (Odocoileus hemionus) fawn (i.e., 0--6 months old) mortality rates, recruitment, and subsequently population dynamics have been identified as knowledge gaps. Thus, we tested predictions concerning the relationship between natural gas development, adult female, fawn birth, and temporal (weather) characteristics on fawn mortality in the Piceance Basin of northwestern Colorado, USA, from 2012-2014.We captured and radio-collared 184 fawns and estimated apparent cause-specific mortality in areas with relatively high or low levels of natural gas development using a multi-state model. Mean daily predation probability was similar in the high versus low development areas. Predation was the leading cause of fawn mortality in both areas and decreased from 0-14 days old. Black bear (Ursus americanus; 22% of all mortalities, n = 17) and cougar (Fe/is concolor; 36% of all mortalities, n =6) predation was the leading cause of mortality in the high and low development areas, respectively. Predation of fawns was negatively correlated with the distance from a female's core area to a producing well pad on winter or summer range. Contrary to expectations, predation of fawns was positively correlated with rump fat thickness of adult females. Well pad densities and development activity were relatively low during our study, indicating that the observed intensity of development did not appear to influence daily predation probability. Our results suggest maintaining development activity thresholds at levels we observed to potentially minimize the effects of development on fawn mortality. However, we caution that higher development intensity and drilling activity in flatter, less rugged areas with less concealment cover could influence fawn mortality. Managers should maintain low development densities in areas where topography and vegetation offer less concealment. Overall, region-specific data (e.g., development intensity, topography, predator assemblages, and associated predation risk) are needed to better understand the effects of natural gas development on fawn mortality. 35 �Using maternal mule deer movements to estimate timing of parturition and assist fawn captures Mark E. Peterson, 1 Charles R. Anderson Jr.,2 Mathew W. Alldredge 2,and Paul F. Doherty Jr. 1 1 Depanment offish, Wildlife and Conservation Biology, Colorado State University, Fon Collins, Colorado 80523 USA 2 Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fon Collins, CO 90526 USA Citation: Peterson, M. E., C. R. Anderson Jr., M. W. Alldredge, and P. F. Doherty Jr. 2018. Using maternal mule deer movements to estimate timing of parturition and assist fawn captures. Wildlife Society Bulletin; DOI: l0.1002/wsb.935 ABSTRACT Movement patterns of maternal ungulates have been used to determine parturition dates and aid in locating fawns, which may be important for understanding reproductive rates (e.g., pregnancy and fetal), but such methods have not been validated for mule deer (Odocoileus hemionus). We first determined timing of parturition using vaginal implant transmitters (VITs) and then predicted timing of parturition using VITs in conjunction with Global Positioning System collar data in the Piceance Basin of northwestern Colorado, USA, during 2012-2014. We examined daily movement rate to determine differences in movement rate among days (7 days pre- and postpartum) and for movement patterns indicative of parturition. Mean daily movement rate (m/day) of I 02 maternal deer decreased by 46% from I day preparturition (mean = 1,253, SD = 1,091) to parturition date (mean =682, S = 574), and remained at this low rate 1-7 days postpartum. We applied an independent data set to validate predicted parturition dates based on daily movement rate. We estimated day of parturition correctly (i.e., day 0), within 1-3 days postparturition, and_4 days postparturition offield-reported dates for 10 (29%), 21 (60%), and 4 (II%) maternal females, respectively. For novel data sets, we predict that a mule deer female whose daily movement rate decreases by_46% and remains low _3 days postparturition particularly when preceded by a sudden increase in movement-has given birth. However, we caution that disturbance of deer by field crews should be minimized, and if birth sites are not found, neonatal mortality will be underestimated. Our results can help determine timing and general location of parturition as an aid in capturing fawns when the use ofVITs is not feasible, with the ultimate objective of estimating pregnancy, fetal, and fawn survival rates if birth sites are found. 36 V �u Appendix B. Winter utilization distributions of adult female mule deer during a post-drilling, production phase (2012 -2018) in the Piceance Basin, Colorado USA. u 37 �Winter Range Analysis for Piceance Mule Deer Summary Prepared for: Colorado Parks and Wildlife 317 W. Prospect Rd ., Fort Collins, CO 80526 Prepared by: Hall Sawyer and Andrew Telander Western Ecosystems Technology, Inc. 200 South 2nd St. , Laramie, Wyoming November 2018 ~ NATURAL RESOURCES • SCIENTIFIC SOLUTIONS WESli 38 �~ OVERVIEW The purpose of this analysis was to delineate winter distribution patterns within four designated study areas (North Magnolia, South Magnolia, Ryan Gulch, and North Ridge) using GPS data collected from mule deer during six winters (2012-13 through 2017-18). The GPS data were used to estimate overall utilization distributions for each study area, so that managers could map and visualize intensity of use within the four winter ranges. Recognizing which winter range parcels are used more intensively than others can help inform management and land-use decisions associated with this important mule deer herd. APPROACH Data Collection Colorado Parks and Wildlife (CPW) provided us with GPS data collected during the winters of, 2012-13, 201314, 2014-15, 2015-16, 2016-17, and 2017-18. These data were combined into one data frame so that all fields could be organized, cleaned and formatted for analysis. Winter Distribution To ensure data reflected winter use rather than late-autumn or early-spring migration, we restricted our analysis to data collected between December 01 and March 15 each winter. We further restricted analysis to individuals that collected 30 or more days of data. Table 1 shows sample sizes listed by year and study area. Once winter sequences were extracted for each animal, we used the Brownian bridge movement model (BBMM: Horne et al. 2007) and the "BBMM" package in R to estimate a winter utilization distribution (UD) for each animal. The BBMM was preferred over traditional kernel-based estimators because it incorporates movement of the animal by using sequential locations common with GPS studies. Winter distribution patterns were estimated using 444,822 locations collected from 925 GPS-collared deer during the winters 2012-2017 (Fig. 1, Table 1). Within each study area (n = 4), we then averaged the individual animal UDs to create a population-level UD for each winter, including 2012-13, 2013-14, 2014-15, 2015-16, 2016-17, and 2017-18. As a final step, we then averaged all 6 winter UDs together in each study area to produce a final UD intended to represent the overall winter distribution (Fig. 2). As an extra step, we created a shapefile of UD contours ranging from 0.1 to 0.9, to assist with visualizing and differentiating high-use from low-use areas (Figs.3 - 8). Such contours (e.g., 0.50) are often used to help agencies identify or modify existing crucial winter ranges boundaries and target areas for habitat improvements. 39 �WY Clly UT D North Magnolia me * Soltla'<> a co "o.w North Ridge w+ Ryan Gulch " • 0 f 3 °"1a Source H•to-"lof G«,Q,~ m, Cmr~AIW'Sy!-litffl UA0 198]UTM Zonc 13:.1 0 C.. 1118/20 1& Aur:'lor R~MM"I ~ WESli Figure 1. GPS locations (n = 444,822) locations collected from 925 GPS-collared mule deer during the winters of 2012-13, 2013-14, 2014-15, 2015-16, 2016-17, and 2017-18 in the Piceance Basin of northwest Colorado. 40 �Intensity of Use WY • ,. C'O = D High North Ridge Low Ryan Gulch l m, .,,, • Dal.a Sourte Na.tl0n3J Geogar,rvc Coou>rnaie System N;.o 1983 UT M Zent tlN 0.1e i 14'2'018 ,\Utnol R And~uon ~ WESli Figure 2. Winter utilization distributions (UD) of mule deer, averaged from winters 2011-12, 2012-13, 201314, 2014-15, 2015-16, 2016-17, and 2017-18. 41 �North Magnolia North Ridge Ryan Gulch South Magnolia contour contour contour contour 0.1 0.1 0.1 0.1 0.2 0.2 0.2 0.2 0.3 0.3 0.3 0.3 0.4 0.4 0.4 0.4 0.5 0.5 0.5 0.5 Contem may noc rer>e(;1 Natl0a3 Geogtapr11c·s e-urr•.: 1map p(>JIC) Sourcep ~a;.ona:1 ,Geoo1aph1t Es,, OeLo1me 1-fERE .UNEP \\'Cl.IC 'USG1:t rl ASA, ESA ◄'-IR:Afl GEBC'1 1,t)r,.p. mcr~men1PCo<p •.,en. Figure 3. Selected contou rs (0. 1, 0.2., 0.3, 0.4, 0.5) from winter utilization distributions (UD) of mule deer, averaged from winters 2012-13, 2013-14, 2014-15, 2015-16, 2016-17, and 2017-18. 42 �contour contour 0.1 0.1 contour 0.2 0.2 0.2 0.3 0.3 0.3 0.4 0.4 0.4 0.5 0.5 0.5 . 0.1 Figure 4. Selected contours (0.1, 0.2., 0.3, 0.4, 0.5) from w inter utilization distributions (UD) of mule deer, averaged from winters 2012-13, 2013-14, 2014-15, 2015-16, 2016-17, and 2017-18. 43 �Figure 5. Selected contours (0.1, 0.2., 0.3, 0.4, 0.5) from winter utilization distributions (UD) of mule deer in North Ridge study area, averaged from winters 2012-13, 2013-14, 2014-15, 2015-16, 2016-17, and 2017-18. 44 �Figure 6. Selected contours (0.1, 0.2., 0.3, 0.4, 0.5) from winter utilization distributions (UD) of mule deer in North Magnolia study area, averaged from winters 2012-13, 2013-14, 2014-15, 2015-16, 2016-17, and 2017- 18. 45 �Figure 7. Selected contours (0.1, 0.2., 0.3, 0.4, 0.5) from winter utilization distributions (UD} of mule deer in South Magnolia study area, averaged from w inters 2012-13, 2013-14, 2014-15, 2015-16, 2016-17, and 201718. 46 �Figure 8. Selected contours (0.1, 0.2., 0.3, 0.4, 0.5) from winter utilization distributions (UD) of mule deer in Ryan Gulch study area, averaged from winters 2012-13, 2013-14, 2014-15, 2015-16, 2016-17, and 2017-18. Table 1. Sample sizes of GPS-collared mule deer by winter and study area. 47 �Study Area North Magnolia North Magnolia North Magnolia North Magnolia North Magnolia North Magnolia North Magnolia North Ridge North Ridge North Ridge North Ridge North Ridge North Ridge North Ridge Ryan Gulch Ryan Gulch Ryan Gulch Ryan Gulch Ryan Gulch Ryan Gulch Ryan Gulch South Magnolia South Magnolia South Magnolia South Magnolia South Magnolia South Magnolia South Magnolia All Study Areas Winter Period 2012-13 2013-14 2014-15 2015-16 2016-17 2017-18 Sub-total 2012-13 2013-14 2014-15 2015-16 2016-17 2017-18 Sub-total 2012-13 2013-14 2014-15 2015-16 2016-17 2017-18 Sub-total 2012-13 2013-14 2014-15 2015-16 2016-17 2017-18 Sub-total Total n 28 43 44 54 52 16 237 20 39 43 43 46 18 209 27 42 42 54 48 24 237 27 41 45 56 46 27 242 925 REFERENCES Horne, J. S., E. 0. Garton, S. M. Krone, and J. S. Lewis. 2007. Analyzing animal movements using Brownian Bridges. Ecology 88:2354-2363. Sawyer, H., M. J. Kauffman, R. M. Nielson, and J. S. Horne. 2009. Identifying and prioritizing ungulate migration routes for landscape-level conservation. Ecological Applications 19:2016-2025. 48 --•---- ----------- �Colorado Parks and Wildlife July I, 2019- June 30, 2020 WILDLIFE RESEARCH REPORT State of _______C.. ;ao__,;lo:a. ; .ra=-d__oa.,.__ _ _ _ : __P=ar__ksaa.. .,; . ,. a...... n d__W.. . . . .,il___ dl__iTI___ e _ _ _ _ _ _ _ _ _ _ __ Cost Center 3430 : . __M__a=m=m . . . a_____l___s ___R__e__ se__a___rc__h_______________ Work Package 3001 : =D=e=er,.._C=-o=n=s=erv---=a=ti=on=-=-------------Task No. 6 : Population Performance of Piceance Basin Mule Deer in Response to Natural Gas Resource Extraction and Mitigation Efforts to Address Human Activity and Habitat Degradation Federal Aid Project:_--'-'W---'-2;;:..;4~3-=--R.;::..;4,_____ _ __ Period Covered: July 1, 2019 - June 30, 2020 Author: C. R. Anderson, Jr. Personnel: D. Bilyeu-Johnston, D. Collins, B. deVergie, D. Finley, L. Gepfert, T. Knowles, B. Petch, J. Rivale, Z. Swennes, M. Way, CPW; L. Belmonte, BLM; J. Northrup, B. Gerber, G. Wittemyer, Colorado State University; L. Coulter, Coulter Aviation. Project support received from Federal Aid in Wildlife Restoration, Colorado Mule Deer Association, Colorado Mule Deer Foundation, Muley Fanatic Foundation, Colorado State Severance Tax Fund, Caerus Oil and Gas LLC, EnCana Corp., ExxonMobil Production Co./XTO Energy, Marathon Oil Corp., Shell Petroleum, and WPX Energy. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW's right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. �WILDLIFE RESEARCH REPORT POPULATION PERFORMANCE OF PICEANCE BASIN MULE DEER IN RESPONSE TO NATURAL GAS RESOURCE EXTRACTION AND MITIGATION EFFORTS TO ADDRESS HUMAN ACTMTY AND HABITAT DEGRADATION CHARLES R. ANDERSON, JR PROJECT NARRITIVE OBJECTIVES 1. To determine experimentally whether enhancing mule deer habitat conditions on winter range elicits behavioral responses, improves body condition, increases fawn survival, and ultimately, population density on mule deer winter ranges exposed to extensive energy development. 2. To determine experimentally to what extent modification of energy development practices enhance habitat selection, body condition, fawn survival, and winter range mule deer densities. SEGMENT OBJECTIVES 1. Recover remaining GPS collars to address the final year of adult female mule deer habitat use and behavioral patterns in 4 study areas experiencing varying levels of energy development in the Piceance Basin, northwest Colorado. 2. Finalizing monitoring of vegetation responses to habitat treatments for assessing efficacy of habitat improvement projects to mitigate energy development disturbances to mule deer. 3. Complete evaluation of large herbivore use of habitat treatments during summer/fall using remote camera sampling. PROJECT OVERVIEW AND RESEARCH SUMMARY We propose to experimentally evaluate winter range habitat treatments and human-activity management alternatives intended to enhance mule deer (Odocoileus hemionus) populations exposed to energy-development activities. The Piceance Basin of northwestern Colorado was selected as the project area due to ongoing natural gas development in one of the most extensive and important mule deer winter and transition range areas in Colorado. The data presented here represent preliminary and final results of a 10-year research project addressing habitat improvements as mitigation and evaluation of deer responses to energy development activities to inform future development planning options on important seasonal ranges. From2008-2019, we monitored deer on 4 winter range study areas representing relatively high (Ryan Gulch, South Magnolia) and low (North Magnolia, North Ridge) levels of development activity (Fig. 1) to address factors influencing deer behavior and demographics and to evaluate success of habitat treatments as a mitigation option. We recorded adult female habitat use and movement patterns; estimated neonatal, overwinter fawn and annual adult female survival; estimated annual early and late winter body condition, pregnancy and fetal rates of adult females; and estimated annual mule deer abundance among study areas. Winter range habitat improvements completed spring 2013 resulted in 604 acres of mechanically treated pinion-juniper/mountain shrub habitats in each of2 treatment areas (Fig. 2) with minor (North Magnolia) and extensive (South Magno1ia) energy development, respectively. During this research segment, we recovered the remaining store-on-board GPS collars from adult female mule deer during spring/summer 2019, completed the final year of measuring vegetation 2 �responses of habitat treatments completed spring 2013 and collected camera grid detections of summer/fall herbivore use of habitat treatment and control sites (preliminary results reported in Appendix B). Based on final (migration, mule deer behavioral responses. reproductive success and neonate survival; see Anderson 2019 for detailed methods and results and Appendix A for publication abstracts) and preliminary data analyses (vegetation and herbivore response to habitat treatments, Appendix B) for this I0-year project: ( 1) annual adult female survival was consistent among areas averaging 79-87% annually, but overwinter fawn survival was variable, ranging from 31 % to 95% within study areas, with annual and study area differences primarily due to early winter fawn condition. annual weather conditions, and factors associated with predation on winter range; (2) mule deer body condition early and late winter was generally consistent within areas. with higher variability among study areas early winter, primarily due to December lactation rates. and late winter condition related to seasonal moisture and winter severity; (3) late winter mule deer densities increased through 2016 in all study areas, ranging from 50% in North Ridge to 103% in North Magnolia. but have stabilized recently in 3 of the 4 study areas with recent decline evident in North Ridge (Fig. 3); (4) migratory mule deer selected for areas with increased cover and increased their rate of travel through developed areas. and avoided negative influences through behavioral shifts in timing and rate of migration, but did not avoid development structures (Fig. 4); (5) mule deer exhibited behavioral plasticity in relation to energy development. without evidence of demographic effects. where disturbance distance varied relative to diurnal extent and magnitude of development activity (Fig 5), which provide for useful mitigation options in future development planning; and (6) energy development activity under existing conditions did not influence pregnancy rates, fetal rates or early fawn survival (0-6 months), but may have reduced neonatal survival (March until birth) during 2012 when drought conditions persisted during the third trimester of doe parturition (Fig. 6). Final results are pending to address vegetation and mule deer responses to assess habitat treatment mitigation options for energy development planning. Final data collection efforts for this project were completed by spring 2020. Collaborative research with agency biologists. graduate students, and university professors has produced 22 scientific publications addressing improved monitoring techniques for neonate mule deer captures (Bishop et al. 2011. Peterson et al. :!O 18b); development and evaluation of a remote mule deer collaring device (Bishop et al. 2019); mule deer migration relative to energy development (Lendrum et al. 2012, 20 I3. 20 I4; Anderson and Bishop 20 I4). improved approaches to address animal habitat use pattems (Northrup et al. 2013 ); mule deer response to helicopter capture and handling (Northrup et al. 2014a); potential effects of male-biased harvest on mule deer productivity (Freeman et al. 2014); mule deer genetics in relation to body condition and migration (Northrup et al. 2014b); acoustic monitoring to investigate spatial and temporal factors influencing mule deer vigilance (Lynch et al. 2014) and foraging behavior (Northrup et al. 2019); the relationship of plant phenology with mule deer body condition (Seral et al. 2015); approaches to identify cause-specific mortality in mule deer from field necropsies (Stonehouse et al. 2016); the influence of individual and temporal factors affecting late winter body condition estimates of adult female mule deer (Bergman et al. 2018); and mule deer behavioral and demographic responses to energy development activities to inform future development planning (Northrup et al. 20 I5, 2016a. 20 I6b, in press. Peterson et al. 2017. 2018a). These publications are summarized in Appendix A and preliminary results describing vegetation and herbivore responses to habitat treatments are reported in Appendix B. We anticipate the opportunity to work cooperatively toward developing solutions for allowing the nation's energy reserves to be developed in a manner that benefits wildlife and the people who value both the wildlife and energy resources of Colorado and elsewhere. 3 �~, ' Mule Deer Winter Range Study Areas Mule deer study areas Well Pad• & Facilities Q ! 1nc:~lop:nen1 - De-w"!k)f)mtnt f.JCll'IIIE-5 uew1nMagnooa SOuin "1apn0ha 10 Mdo Figure I. Mule deer winter range study areas relative to active natural gas well pads and energy development fac ilities in the Piceance Basin of n011hwest Colorado. winter 2013/ 14 (Accessed http://cogcc.state.co.us/ December 31.20 13: energy development drilling activity has been minor since 2013). 4 �Nollh Magnolia treatement sites (587 acres) Bea,Set_ 15_35b_andG 8e,rSel_ 1_BancA_E I 8e3rSft_36_54andJ GreasewoooSet_g16_g~9 Greasev-.ood5 et_g1 _g 1$ Gre a; awoodSet_ g30_g4 2 l eeOvers,ghts_J_fand 16_ 17 Mechantcal treatment comparison {54 acres) - - No11h Hatch Pilot Treatmen ts! 116 1cres1 Soutn Magnolia J Figure 2. Habitat treatment site delineations in 2 mule deer study areas (604 acres each) of the Piceance Basin. northwest Colorado (Top; cyan polygons completed Jan 20 I I using hydro-axe: yellow pol ygons completed Jan 20 12 using hydro-axe. ro ller-chop. and chaining; and remaining polygons completed Apr 2013 using hydro-axe). January 201 1 hydro-axe treatment-site photos from North Hatch Gulch during April (Lower left. aerial view) and October. 20 I I (Lower right. ground view). 5 �Piceance Basin late w inter mule deer density 35.00 30.00 25.00 NE -"" -:::(IJ (IJ / 20.00 15.00 .I. 0 4 .. ....... 10.00 :I: 5.00 - - North Ridge • • • • • • Ryan Gulch - • North Magnolia - South Magnolia 0.00 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Year Figure 3. Mule deer density estimates and 95% Cl (enor bars) from 4 winter range herd segments in the Piceance Basin, northwest Colorado. late w inter 1009- 10 18. Figure 4. M ule deer study areas in the Piceance Basin of northwestern Colorado. USA (Top). spring 2009 migration routes of adult fema le mule deer (n = 51: Lower left). and active natural-gas well pads (black dots) and roads (state. county. and natural-gas: white lines) from May 1009 (Lower right: from Lendrum et al. 20 12: http://d:-:.doi.om/ I 0. 1890 ES 11-00165. 1). 6 �'i' t _ _ _ _ _ r Ptod 400 Prod 600 P,ud 800 Prc,d 1000 Oriti ,IOt, 01111 600 Oriti 600 Oull 1000 On~ 500 Onll BOC 01111 100~ Covarut1os - r -- 'i'' ·r r Prod 400 Proo 600 P1od 800 Proa '1000 Outt 1\ t)~ n I <t Figure 5. Posterior distributions of populati on-level coeffi cients related to natural gas development for RSF models during the day (top) and ni ght (bottom) for 53 adu lt fema le mul e deer in the Piceance Basi n. northwest Colorado. Dashed line indicates 0 selection or avoidance (below the line) of the habitat features. ' Drill' and 'Prod' represent d rilling and producing well pads, respectively. T he numbers fo llow ing ' Dri ll' or ' Prod' represent the distance from respect ive well pads evaluated (e.g., 'Drill 600' is the number of well pads with active dri lling between 400-600 m from the deer location; from No1thrup et al. 20 15; http://o nli nelibrarv.wilev.com/do i/ I0.1 I 1l /gcb.13037/abstract). Road disturbance was relatively minor (- 60-120 m, not illustrated above). 1.00 (1) 0.80 r-t- - ..._ ~ .._ rt co.... ro> 0 .60 -~ ~ Cl) ro 0.40 Q) LL I I 0 .20 0.00 2012 20 13 2014 Year o High development □ Low development I Figure 6. Model averaged estimates of mu le deer fetal s urvival from early Marc h unti l bi 1th (late May-June) in high and low energy develo pment study areas of the Piceance Basin, northwest Colorado, 2012- 20 14 (from Peterson et a l. 20 17; http://www. bioone.org/doi/ pdf/ I 0.298 1/wlb.00341 ). 7 �LITERATURE CITED Anderson, C.R., Jr. 2019. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Federal Aid in Wildlife Restoration Annual Report W-243-R3, Ft. Collins, CO USA. Anderson, C.R., Jr., and C. J. Bishop. 2014. Migration patterns of adult female mule deer in response to energy development. Pages 47-50 in Transactions of the 79th North American Wildlife & Natural Resources Conference (R. A. Coon & M. C. Dunfee, eds.). Wildlife Management Institute, Gardners, PA, USA. ISSN 0078-1355. Bergman, E. J., C.R. Anderson Jr., C. J. Bishop, A. A. Holland, and J.M. Northrup. 2018. Variation in ungulate body fat: individual versus temporal effects. Journal of Wildlife Management 82:130-137, DOI: 10:1002/jwmg.21334 Bishop, C. J., C.R. Anderson Jr., D. P. Walsh, E. J. Bergman, P. Kuechle, and J. Roth. 2011. Effectiveness of a redesigned vaginal implant transmitter in mule deer. Journal of Wildlife Management 75(8): 1797-1806; DOI: 10.1002/jwmg.229 Bishop, C. J., M. W. Alldredge, D. P. Walsh, E. J. Bergman, C. R. Anderson Jr., D. Kilpatrick, J. Bakel, and C. Fabvre. 2019. A noninvasive automated device for remotely collaring and weighing mule deer. Wildlife Society Bulletin 43:717-725; doi.org/10.1002/wsb.1034 Freeman, E. D., R. T. Larsen, M. E. Peterson, C.R. Anderson, Jr., K. R. Hersey, and B. R. McMillan. 2014. Effects of male-biased harvest on mule deer: implications for rates of pregnancy, synchrony, and timing of parturition. Wildlife Society Bulletin; DOI: 10.1002/wsb.450 Lendrum, P. E., C.R. Anderson, Jr., R. A. Long, J. K. Kie, and R. T. Bowyer. 2012. Habitat selection by mule deer during migration: effects of landscape structure and natural gas development. Ecosphere 3(9):82. http://dx.doi.org/10. Lendrum, P. E., C.R. Anderson, Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2013. Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes. PLoS ONE 8(5): e64548. doi: 10.1371/journal.pone.0064548 Lendrum, P. E., C.R. Anderson, Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2014. Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns of forage quality. Mammalian Biology: http://dx.doi.org/10.1016/j .mambio.2014.05.005 Lynch, E., J.M. Northrup, M. F. McKenna, C.R. Anderson Jr., L. Angeloni, and G. Wittemyer. 2014. Landscape and anthropogenic features influence the use of auditory vigilance by mule deer. Behavioral Ecology; doi: 10.1093/beheco/aru 158. Northrup, J. M., M. 8. Hooten, C. R. Anderson, Jr., and G. Wittemyer. 2013. Practical guidance on characterizing availability in resource selection functions under a use-availability design. Ecology 94(7): 1456-1463. Northrup, J.M., C.R. Anderson, Jr., and G. Wittemyer. 2014a. Effects of helicopter capture and handling on movement behavior of mule deer. Journal of Wildlife Management 78(4):731738; DOI: 10.1002/jwmg.705 Northrup, J.M., A. B. Shafer, C.R. Anderson Jr., D. W. Coltman, and G. Whittemyer. 2014b. Finescale genetic correlates to condition and migration in a wild cervid. Evolutionary Applications ISSN 1752-4571; doi: 10.1111/eva.12 l 89 Northrup, J. M., C. R. Anderson, Jr., and G. Wittemyer. 2015. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer. Global Change Biology, doi: 10.1111/gcb.13037. Northrup, J.M., C.R. Anderson, Jr., M. B. Hooten, and G. Wittemyer. 2016a. Movement reveals scale dependence in habitat selection of a large ungulate. Ecological Applications 26:27462757 8 �Northrup, J.M., C.R. Anderson, Jr., and G. Wittemyer. 2016b. Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid. Diversity and Distributions 22: 547-557, DOI: 10.11 11 /ddi. I2417 Northrup, J.M., A. Avrin, C.R. Anderson Jr., E. Brown, and G. Wittemyer. 2019. On-animal acoustic monitoring provides insight to ungulate foraging behavior. Journal ofMammalogy 100:14791489; https://doi.org/10.1093/imammal/gyzl 24 Northrup, J.M., C.R. Anderson Jr., 8. D. Gerber, and G. Wittemyer. In Press. Behavioral and demographic responses of mule deer to energy development on winter range. Wildlife Monographs. Peterson, M. E., C.R. Anderson Jr., J.M. Northrup, and P. F. Doherty Jr. 2017. Reproductive success of mule deer in a natural gas development area. Wildlife Biology doi: 10.1111/wlb.00341 Peterson, M. E., C.R. Anderson Jr., J.M. Northrup, and P. F. Doherty Jr. 2018a. Mortality of mule deer fawns in a natural gas development area. Journal of Wildlife Management 82:1135-1148, DOI: 10.1002/jwmg.21476 Peterson, M. E., C. R. Anderson Jr., M. W. Alldredge, and P. F. Doherty Jr. 2018b. Using maternal mule deer movements to estimate timing of parturition and assist fawn captures. Wildlife Society Bulletin 42:616-621; DOI: 10.1002/wsb.935 Searle, K. R., M. 8. Rice, C. R. Anderson, C. Bishop and N. T. Hobbs. 2015. Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore. Oecologia ISSN 0029- 8549; DOI 10.1007/s00442-015-3348-9 Stonehouse, K. F., C.R. Anderson Jr., M. E. Peterson, and D.R. Collins. 2016. Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus). Colorado Parks and Wildlife Technical Report No. 48, First Edition, 317 W. Prospect Rd., Ft. Collins, CO USA. DOW-R-T-48-16, ISSN 0084-8883 . ...,_; Prepared by_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ Charles R. Anderson, Jr., Mammals Research Leader 9 �Appendix A. Abstracts of published manuscripts resulting from Piceance Basin mule deer/energy development interaction research collaborations. Abstract format specific to the respective journal requirements. Effectiveness of a redesigned vaginal implant transmitter in mule deer CHAD J. BISHOP', CHARLES R ANDERSON Jr. 1, DANIEL P. WALSH', ERIC J. BERGMAN', PETER KUECHLE2, and JOHN ROTH 2 1 Colorado Parks and Wildlife, Fort Collins, Colorado 80526 USA Advanced Telemetry Systems, Isanti, Minnesota 55040 USA 2 Citation: Bishop, C. J., C. R. Anderson Jr., D. P. Walsh, E. J. Bergman, P. Kuechle, and J. Roth. 2011. Effectiveness of a redesigned vaginal implant transmitter in mule deer. Journal of Wildlife Management 75(8):1797-1806; DOI: 10.1002/jwmg.229 ABSTRACT Our understanding of factors that limit mule deer (Odocoi/eus hemionus) populations may be improved by evaluating neonatal survival as a function of dam characteristics under free-ranging conditions, which generally requires that both neonates and dams are radiocollared. The most viable technique facilitating capture of neonates from radiocollared adult females is use of vaginal implant transmitters (VITs). To date, VITs have allowed research opportunities that were not previously possible; however, VITs are often expelled from adult females prepartum, which limits their effectiveness. We redesigned an existing VIT manufactured by Advanced Telemetry Systems (ATS; Isanti, MN) by lengthening and widening wings used to retain the VIT in an adult female. Our objective was to increase VIT retention rates and thereby increase the likelihood oflocating birth sites and newborn fawns. We placed the newly designed VITs in 59 adult female mule deer and evaluated the probability of retention to parturition and the probability of detecting newborn fawns. We also developed an equation for determining VIT sample size necessary to achieve a specified sample size of neonates. The probability ofa VIT being retained until parturition was 0.766 (SE= 0.0605) and the probability of a VIT being retained to within 3 days of parturition was 0.894 (SE= 0.0441). In a similar study using the original VIT wings (Bishop et al. 2007), the probability ofa VIT being retained until parturition was 0.447 (SE= 0.0468) and the probability of retention to within 3 days of parturition was 0.623 (SE= 0.0456). Thus, our design modification increased VIT retention to parturition by 0.319 (SE= 0.0765) and VIT retention to within 3 days of parturition by 0.271 (SE= 0.0634). Considering dams that retained VITs to within 3 days of parturition, the probability of detecting at least 1 neonate was 0.952 (SE= 0.0334) and the probability of detecting both fawns from twin litters was 0.588 (SE = 0.0827). We expended approximately I 2 person-hours per detected neonate. As a guide for researchers planning future studies, we found that VIT sample size should approximately equal the targeted neonate sample size. Our study expands opportunities for conducting research that links adult female attributes to productivity and offspring survival in mule deer.© 2014 The Wildlife Society. Habitat selection by mule deer during migration: effects of landscape structure and natural-gas development PATRICK E. LENDRUM 1, CHARLES R ANDERSON JR 2, RYAN A. LONG 1, JOHN G. KIE', AND R TERRY BOWYER' 1 0epartment of Biological Sciences, Idaho State University, Pocatello, Idaho 83209 USA 2colorado Parks and Wildlife. Grand Junction. Colorado 81505 USA Citation: Lendrum, P. E., C. R. Anderson Jr., R. A. Long. J. G. Kie, and R. T. Bowyer. 2012. Habitat selection by mule deer during migration: effects of landscape structure and natural-gas development. Ecosphere 3(9):82 http://dx.doi.org/ I0.1890/ES 12-00165. I Abstract. The disruption of traditional migratory routes by anthropogenic disturbances has shifted patterns of resource selection by many species, and in some instances has caused populations to decline. Moreover, in recent decades populations of mule deer (Odocoi/eus hemionus) have declined throughout much of their historic range in the western United States. We used resourceselection functions to determine if the presence of natural-gas development altered patterns of resource selection by migrating mule deer. We compared spring migration routes of adult female mule deer fitted with GPS collars (n = I 67) among four study areas that had varying degrees ofnatural-gas development from 2008 to 2010 in the Piceance Basin of northwest Colorado, USA. Mule deer migrating through the most developed area had longer step lengths (straight-line distance between successive GPS locations) compared with deer in less developed areas. Additionally, deer migrating through the most developed study areas tended to select for habitat types that provided greater amounts of concealment cover, whereas deer from the least developed areas tended to select habitats that increased access to forage and cover. Deer selected habitats closer to well pads and avoided roads in all instances except along the most highly developed migratory routes. where road densities may have been too high for deer to avoid roads without deviating substantially from established migration routes. These results indicate that behavioral tendencies toward avoidance of anthropogenic disturbance can be overridden during migration by the strong fidelity ungulates demonstrate to,vards migration routes. If avoidance is feasible, then deer may select areas further from development. whereas in highly developed areas, deer may simply increase their rate of travel along established migration routes. 10 u �Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes Patrick E. Lendrum', Charles R. Anderson Jr.\ Kevin L. Monteith'·\ Jonathan A. Jenks\ R. Terry Bowyer' 1 Department of Biological Sciences, Idaho State University, Pocatello, Idaho, USA, 2 Colorado Division of Parks and Wildlife, Grand Junction, Colorado, USA, 3 Wyoming Cooperative Fish and Wildlife Research Unit, University of Wyoming, Laramie, Wyoming, USA,~ Department of Natural Resource Management, South Dakota State University, Brookings, South Dakota, USA Citation: Lendrum, P. E., C.R. Anderson Jr., K. L. Monteith, J. A. Jenks, R. T. Bowyer. 2013. Migrating Mule Deer: Effects of anthropogenically Altered Landscapes. PLoS ONE 8(5): e64548. DOI: 10.137l~oumal.pone.0064548 Abstract Background: Migration is an adaptive strategy that enables animals to enhance resource availability and reduce risk of predation at a broad geographic scale. Ungulate migrations generally occur along traditional routes, many of which have been disrupted by anthropogenic disturbances. Spring migration in ungulates is of particular importance for conservation planning, because it is closely coupled with timing of parturition. The degree to which oil and gas development affects migratory patterns, and whether ungulate migration is sufficiently plastic to compensate for such changes, warrants additional study to better understand this critical conservation issue. Methodology/Principal Findings: We studied timing and synchrony of departure from winter range and arrival to summer range of female mule deer (Odocoileus hemionus) in northwestern Colorado, USA, which has one of the largest natural-gas reserves currently under development in North America We hypothesized that in addition to local weather, plant phenology, and individual life-history characteristics, patterns of spring migration would be modified by disturbances associated with natural-gas extraction. We captured 205 adult female mule deer, equipped them with GPS collars, and observed patterns of spring migration during 2008-20 l 0. Condusions/Signijicance: Timing of spring migration was related to winter weather (particularly snow depth) and access to emerging vegetation, which varied among years, but was highly synchronous across study areas within years. Additionally, timing of migration was influenced by the collective effects of anthropogenic disturbance, rate of travel, distance traveled, and body condition of adult females. Rates of travel were more rapid over shoner migration distances in areas of high natural-gas development resulting in the delayed departure, but early arrival for females migrating in areas with high development compared with less-developed areas. Such shifts in behavior could have consequences for timing of arrival on birthing areas, especially where mule deer migrate over longer distances or for greater durations. Practical guidance on characterizing availability in resource selection functions under a use-availability design JOSEPH M. NORTHRUP 1, MEVIN 8. HOOTEN 1,2,3, CHARLES R. ANDERSON JR.\ AND GEORGE \VITTEMYER 1 1 Department offish, Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA U.S. Geological Survey, Colorado Cooperative Fish and Wildlife Research Unit, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 3 Colorado State University, Department of Statistics, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA ~Mammals Research Section Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction, Colorado 81505 USA 2 Citation: Northrup, J. M., M. 8. Hooten, C. R. Anderson Jr., and G. Wittemyer. 2013. Practical guidance on characterizing availability in resource selection functions under a use-availability design. Ecology 94(7): I 4S6-1463. http://dx.doi.org/10. I 890/12-1688.1 Abstract. Habitat selection is a fundamental aspect of animal ecology, the understanding of which is critical to management and conservation. Global positioning system data from animals allow fine-scale assessments of habitat selection and typically are analyzed in a us~availability framework, whereby animal locations are contrasted with random locations (the availability sample). Although most use-availability methods are in fact spatial point process models, they often are fit using logistic regression. This framework offers numerous methodological challenges, for which the literature provides little guidance. Specifically, the size and spatial extent of the availability sample influences coefficient estimates potentially causing interpretational bias. We examined the influence of availability on statistical inference through simulations and analysis of serially correlated mule deer OPS data. Bias in estimates arose from incorrectly assessing and sampling the spatial extent of availability. Spatial autocorrelation in covariates, which is common for landscape characteristics. exacerbated the error in availability sampling leading to increased bias. These results have strong implications for habitat selection analyses using GPS data, which are increasingly prevalent in the literature. We recommend that researchers assess the sensitivity of their results to their availability sample and, where bias is likely, take care with interpretations and use cross validation to assess robustness. 11 �Effects of Helicopter Capture and Handling on Movement Behavior of Mule Deer JOSEPH M. NORTHRUP', CHARLES R. ANDERSON JR2, AND GEORGE WITIEMYER 1 'Department of Fish. Wildlife. and Conservation Biology, Colorado State University. 1474 Campus Delivery. Fort Collins. Colorado 80523 USA Mammals Research Section Colorado Parks and Wildlife. 711 Independent Avenue. Grand Junction, Colorado 81505 USA 2 Citation: Northrup. J. M.. C. R. Anderson Jr., and G. Wittemyer. 2014. Effects of helicopter capture and handling on movement behavior of mule deer. Journal of Wildlife Management 78(4):731-738; DOI: 10.1002/jwmg.705 ABSTRACT Research on wildlife movement, physiology, and reproductive biology often requires capture and handling of animals. Such invasive treatment can alter behavior, which may bias results or invalidate assumptions regarding representative behaviors. To assess the impacts of handling on mule deer (Odocoileus hemionus), a focal species for research in North America, we investigated pre- and post-recapture movements of collared individuals, and compared them to deer that were not recaptured {controls). We compared pre- and post-recapture movement rates (m/hr) and 24-hour straight-line displacement among recaptured and control deer. In addition, we examined the time it took recaptured deer to return to their pre-recapture home range. Both daily straight-line displacement and movement rate were marginally elevated relative to monthly averages for 24 hours following recapture, with non-significant elevation continuing for up to 7 days. Comparing movements averaged over 30 days before and after recapture, we found no differences in displacement, but movement rates demonstrated seasonal effects, with faster movements post- relative to pre-recapture in March and slower movements post- relative to pre-recapture in December. Relative to control deer movements, recaptured deer movement rates in March were higher immediately after recapture and lower in the second and third weeks following recapture. The median time to return to the pre-recapture home range was 13 hours, with 71 % of deer returning in the first day, and 91 % returning within 4 days. These results indicate a short period of elevated movements following recaptures, likely due to the deer returning to their home ranges, followed by weaker but non-significant depression of movements for up to 3 weeks. Censoring of the first day of data post capture from analyses is strongly supported, and removing additional days until the individual returns to its home range will control for the majority of impacts from capture. © 2014 The Wildlife Society. Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns offorage quality Patrick E. Lendrum•, Charles R. Anderson Jr.b, Kevin L. Monteith~, Jonathan A. Jenksd, R. Terry Bowyer• a Department ofBiological Sciences, Idaho State University, 921 South 8th Avenue. Stop 8007. Pocatello 83209, USA " Mammals Researth Section Colorado Parks and Wildlife, 71 I Independent Avenue. Grand Junction 81505, USA .: Wyoming Cooperative Fish and Wildlife Research Unit. Department of Zoology and Physiology. University of Wyoming.3166. 1000 East University Avenue, Laramie 82071. USA "Department of Natural Resource Management. South Dakota State University, Box 21408. Brookings 57007. USA Citation: Lendrum. P. E .• C.R. Anderson Jr.• K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2014. Relating the movement ofa rapidly migrating ungulate to spatiotemporal patterns of forage quality. Mammalian Biology: http·//dx.doi.org/J0.1016/j mambio.20)4.05 005 ABSTRACT: Migratory ungulates exhibit recurring movements, often along traditional routes between seasonal ranges each spring and autumn. which allow them to track resources as they become available on the landscape. We examined the relationship between spring migration of mule deer (Odocoileus hemionus) and forage quality, as indexed by spatiotemporal patterns of fecal nitrogen and remotely sensed greeMess of vegetation (Nonnalized Difference Vegetation Index; NOVI) in spring 2010 in the Piceance Basin of northwestern Colorado, USA. NOVI increased throughout spring, and was affected primarily by snow depth when snow was present, and temperature when snow was absent. Fecal nitrogen was lowest when deer were on winter range before migration, increased rapidly to an asymptote during migration, and remained relatively high when deer reached summer range. Values of fecal nitrogen corresponded with increasing NOVI during migration. Spring migration for mule deer provided a way for these large mammals to increase access to a high-quality diet, which was evident in patterns of NOVI and fecal nitrogen. Moreover, these deer "jumpect•· rather than ·•surfed.. the green wave by arriving on summer range well before peak productivity of forage occurred. This rapid migration may aid in securing resources and seclusion from others on summer range in preparation for parturition, and to minimize detrimental factors such as predation and malnutrition during migration. 12 �Effects of Male-Biased Harvest on Mule Deer: Implications for Rates of Pregnancy, Synchrony, and Timing of Parturition ERIC D. FREEMAN', RANDY T. LARSEN 1, MARKE. PETERSON2, CHARLES R. ANDERSON JR. 3, KENT R. HERSEY', AND BROCK R. McMILLAN' 1 Department of Plant and Wildlife Sciences, Brigham Young University, 275 WIDB, Provo, UT 84602, USA 2 Department of Fish, Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA 3 Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction, CO 81505, USA 4 Utah Division of Wildlife Resources, 1594 W North Temple, Salt Lake City, UT 84114, USA Citation: Freeman, E. D., R. T. Larsen, M. E. Peterson, C.R. Anderson Jr., K. R. Hersey, and B. R. McMillan. 2014. Effects of male-biased harvest on mule deer: implications for rates of pregnancy, synchrony, and timing of parturition. Wildlife Society Bulletin; DOI: 10.1002/wsb.450 ABSTRACT Evaluating how management practices influence the population dynamics of ungulates may enhance future management of these species. For example, in mule deer (Odocoileus hemionus), changes in male/female ratio due to male• biased harvest may alter rates of pregnancy, timing of parturition, and synchrony of parturition if inadequate numbers of males are present to fertilize females during their first estrous cycle. If rates of pregnancy or parturition are influenced by decreased male/female ratios, recruitment may be reduced (e.g., fewer births, later parturition resulting in lower survival of fawns, and a less synchronous parturition that potentially increases susceptibility of neonates to predation). Our objectives were to compare rates of pregnancy, synchrony of parturition, and timing of parturition between exploited mule deer populations with a relatively high (Piceance, CO, USA; 26 males/100 females) and a relatively low (Monroe, UT, USA; 14 males/100 females) male/female ratio. We determined rates of pregnancy via ultrasonography and timing of parturition via vaginal implant transmitters. We found no differences in rates of pregnancy (98.6% and 96.6%; z = 0.821; P = 0.794), timing of parturition (estimate= 1.258; SE= 1.672; t = 0.152; P = 0.454), or synchrony of parturition (F= 1.073; P = 0.859) between Monroe Mountain and Piceance Basin, respectively. The relatively low male/female ratio on Monroe Mountain was not associated with a protracted period of parturition. This finding suggests that relatively low male/female ratios typical of heavily harvested populations do not influence population dynamics because recruitment remains unaffected.© 2014 The Wildlife Society. Fine-scale genetic correlates to condition and migration in a wild cervid Joseph M. Northrup', Aaron B. A. Sharer2, Charles R. Anderson Jr.3, David W. Coltman", and George Wittemyer 1 1 Department of Fish, Wildlife, and Conservation Biology, Colorado State University, Fort Collins, CO. USA 2 Department of Evolutionary Biology. Evolutionary Biology Centre. Uppsala University. Uppsala, Sweden 3 Mammals Research Section, Colorado Parks and Wildlife, Grand Junction, CO, USA 4 Department of Biological Sciences, University of Alberta, Edmonton, AB, Canada. Citation: Northrup, J. M., A. B. Shafer, C. R. Anderson Jr., D. W. Coltman, and G. Whittemyer. 2014. Fine-scale genetic correlates to condition and migration in a wild cervid. Evolutionary Applications ISSN 1752-4571; doi: 10.1111/eva.12189 Abstract The relationship between genetic variation and phenotypic traits is fundamental to the study and management of natural populations. Such relationships often are investigated by assessing correlations between phenotypic traits and heterozygosity or genetic differentiation. Using an extensive data set compiled from free ranging mule deer (Odocoileus hemionus), we combined genetic and ecological data to (i) examine correlations between genetic differentiation and migration timing, (ii) screen for mitochondrial haplotypes associated with migration timing, and (iii) test whether nuclear heterozygosity was associated with condition. Migration was related to genetic differentiation (more closely related individuals migrated closer in time) and mitochondrial haplogroup. Body fat was related to heterozygosity at two nuclear loci (with antagonistic patterns), one of which is situated near a known fat metabolism gene in mammals. Despite being focused on a widespread panmictic species, these findings revealed a link between genetic variation and important phenotypes at a fine scale. We hypothesize that these correlations are either the result of mixing refugial lineages or differential mitochondrial haplotypes influencing energetics. The maintenance of phenotypic diversity will be critical to enable the potential tracking of changing climatic conditions, and these correlates highlight the need to consider evolutionary mechanisms in management. even in widely distributed panmictic species. 13 �Landscape and anthropogenic features influence the use of auditory vigilance by mule deer Emma Lynch•, Joseph M. Northrupti, Megan F. Mc Kenna~. Charles R. Anderson Jr. d, Lisa Angeloniu, and George Wittemye..-.t, Graduate Degree Program in Ecology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA "Department offish, Wildlife and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA 'Natural Sounds and Night Skies Division, National Park Service, 1201 Oakridge Drive, Fort Collins, CO 80525, USA, dMammals Research Section. Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526, USA COepanment ofBiology, Colorado State University, 1878 Campus Delivery, Fort Collins, CO 80523, USA 0 Citation: Lynch, E., J.M. Northrup, M. F. McKenna, C.R. Anderson Jr., L. Angeloni, and G. Wittemyer. 2014. Landscape and anthropogenic features influence the use of auditory vigilance by mule deer. Behavioral Ecology; doi: 10.1093/beheco/aru I58. While visual fonns of vigilance behavior and their relationship with predation risk have been broadly examined, animals also employ other vigilance modalities such as auditory vigilance by listening for the acoustic cues of predators. Similar to the tradeoffs associated with visual vigilance, auditory behavior potentially structures the energy budgets and behavior of animals. The cryptic nature of auditory vigilance makes it difficult to study, but on-animal acoustical monitoring has rapidly advanced our ability to investigate behaviors and conditions related to sound. We utilized this technique to investigate the ways extemaJ stimuli in an active natural gas development field affect periodic pausing by mule deer (Odocoileus hemionus) within bouts of rumination-based mastication. To better understand the ecological properties that structure this behavior, we investigate spatial and temporal factors related to these pauses to detennine if results are consistent with our hypothesis that pausing is used for auditory vigilance. We found that deer paused more when in forested cover and at night, where visual vigilance was likely to be less effective. Additionally, deer paused more in areas of moderate background sound levels, though responses to anthropogenic features were less clear. Our results suggest that pauses during rumination represent a fonn of auditory vigilance that is responsive to landscape variables. Further exploration of this behavior can facilitate a more holistic understanding of risk perception and the costs associated with vigilance behavior. Migration Patterns of Adult Female Mule Deer in Response to Energy Development Charles R. Anderson Jr. and Chad J. Bishop Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526, USA Citation: Anderson, C. R., Jr., and C. J. Bishop. 2014. Migration patterns of adult female mule deer in response to energy development. Pages 47-S0 in Transactions of the 7f1h North American Wildlife & Natural Resources Conference (R. A. Coon & M. C. Dunfee, eds.). Wildlife Management Institute, Gardners. PA, USA. ISSN 0078-1355. Migration is an adaptive strategy that enables animals to enhance resource availability and reduce risk of predation at a broad geographic scale. Ungulate migrations generally occur along traditional routes, many of which have been disrupted by anthropogenic disturbances. Spring migration in ungulates is of particular importance for conservation planning because it is closely coupled with timing of parturition. The degree to which oil and gas development affects migratory patterns, and whether ungulate migration is sufficiently prepared to compensate for such changes, has recently been investigated in Colorado and Wyoming (Lendrum et al. 2012, 2013: Sawyer et al. 2012). Lendrum et al. (2012, 2013) and Sawyer et al. (2012) address mule deer ( Odocoi/eus hemionus) migration patterns in relation to energy development from northwest Colorado and south-central Wyoming, respectively. We address results from the Colorado and Wyoming studies and then compare similarities and differences. The interactions between migratory mule deer and energy development identified by Lendrum et al. (2012, 2013) and Sawyer et al. (2012) suggest mule deer may benefit from energy development planning by considering thresholds of development that may alter migratory behavior. It appears that migration rate, migration routes, and stopover use, if present, may be altered at high development intensities. In addition, migratory mule deer may benefit by maintaining security cover along migration paths, and improved habitat conditions may facilitate more direct and rapid migration requiring less energy to complete migration. Enhancing penneability along migration routes by applying dispersed development plans (<2 well pads/km2) and minimizing disturbance to vegetation types by maintaining security cover should reduce impacts to migratory mule deer as well as other migratory ungulates. Where feasible, habitat improvement projects on winter range and possibly stopover sites would also enhance migratory mule deer populations by enhancing energy reserves for long-distance movements and parturition shortly after summer range arrival. Where possible, directional drilling could be used to extract energy resources from underneath migration routes while maintaining no surface occupancy. Lastly. we emphasize that GPS studies now allow managers to accurately map migration routes for entire populations and identify relatively narrow corridors that are most heavily used thus allowing for the identification of the most important corridors for migrating ungulates. Where available, we encourage agencies to incorporate such migration corridors into land-use plans (e.g., resource management plans) and National Environmental Policy Act documents. 14 �Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore Kate R. Searle1 • Mindy 8. Rice2 • Charles R. Anderson 2 • Chad Bishop2 • N. T. Hobbs3 1 NERC Centre for Ecology and Hydrology, Bush Estate, Penicuik EH26 OQB, UK 2 Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80S26, USA 3 Department of Ecosystem Science and Sustainability, Colorado State University, Fort Collins 80524, CO, USA Citation: Searle, K. R., M. B. Rice, C. R. Anderson, C. Bishop and N. T. Hobbs. 2015. Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore. Oecologia ISSN 0029-8549; DOI I0.1007/s00442-015-3348-9 Abstract Understanding how spatial and temporal heterogeneity influence ecological processes forms a central challenge in ecology. Individual responses to heterogeneity shape population dynamics, therefore understanding these responses is central to sustainable population management Emerging evidence has shown that herbivores track heterogeneity in nutritional quality of vegetation by responding to phenological differences in plants. We quantified the benefits mule deer (Odocoileus hemionus) accrue from accessing habitats with asynchronous plant phenology in northwest Colorado over 3 years. Our analysis examined both the direct physiological and indirect environmental effects of weather and vegetation phenology on mule deer winter body condition. We identified several important effects of annual weather patterns and topographical variables on vegetation phenology in the home ranges of mule deer. Crucially, temporal patterns of vegetation phenology were linked with differences in body condition, with deer tending to show poorer body condition in areas with less asynchronous vegetation green-up and later vegetation onset. The direct physiological effect of previous winter precipitation on mule deer body condition was much less important than the indirect effect mediated by vegetation phenology. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer JOSEPH M. NORTHRUP 1, CHARLES R. ANDERSON JR. 2, and GEORGE WITTEMYER 1 • 3 1Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, CO, USA 2Manunals Research Section, Colorado Parks and Wildlife, Fort Collins, CO, USA 3 Graduate Degree Program in Ecology, Colorado State University, Fort Collins, CO. USA Citation: Northrup, J.M., C.R. Anderson, Jr.. and G. Wittemyer. 2015. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer. Global Change Biology, doi: 10.1111/gcb.13037 Abstract Extraction of oil and natural gas (hydrocarbons) from shale is increasing rapidly in North America, with documented impacts to native species and ecosystems. With shale oil and gas resources on nearly every continent, this development is set to become a major driver of global land-use change. It is increasingly critical to quantify spatial habitat loss driven by this development to implement effective mitigation strategies and develop habitat offsets. Habitat selection is a fundamental ecological process, influencing both individual fitness and population-level distribution on the landscape. Examinations of habitat selection provide a natural means for understanding spatial impacts. We examined the impact of natural gas development on habitat selection patterns of mule deer on their winter range in Colorado. We fit resource selection functions in a Bayesian hierarchical framework, with habitat availability defined using a movement-based modeling approach. Energy development drove considerable alterations to deer habitat selection patterns, with the most substantial impacts manifested as avoidance of well pads with active drilling to a distance of at least 800 m. Deer displayed more nuanced responses to other infrastructure, avoiding pads with active production and roads to a greater degree during the day than night. In aggregate, these responses equate to alteration of behavior by human development in over 50% of the critical winter range in our study area during the day and over 25% at night. Compared to other regions, the topographic and vegetative diversity in the study area appear to provide refugia that allow deer to behaviorally mediate some of the impacts of development. This study, and the methods we employed, provides a template for quantifying spatial take by industrial activities in natural areas and the results offer guidance for policy makers, mangers, and industry when attempting to mitigate habitat loss due to energy development 15 �Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid Joseph M. Northrup•, Charles R. Anderson Jr, and George Wittemyer 1.J 'Department offish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, CO, USA 2Mammals Research Section, Colorado Parks and Wildlife, Fort Collins, CO, USA 3Graduate Degree Program in Ecology, Colorado State University, Fort Collins, CO, USA Citation: Northrup, J.M., C.R. Anderson, Jr., and G. Wittemyer. 2016. Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid. Diversity and Distributions 22: 547-557, DOI: 10.1111/ddi.12417 ABSTRACT Aim The space an animal uses over a given time period must provide the resources required for meeting energetic needs, reproducing and avoiding predation. Anthropogenic landscape change in concert with environmental dynamics can strongly structure space-use. Investigating these dynamics can provide critical insight into animal ecology, conservation and management. Location The Piceance Basin, Colorado. USA. Methods We applied a novel utilization distribution estimation technique based on a continuous-time correlated random walk model to characterize range dynamics of mule deer during winter and summer seasons across multiple years. This approach leverages secondorder properties of movement to provide a probabilistic estimate of space-use. We assessed the influence of environmental (cover and forage), individual and anthropogenic factors on interannual variation in range use of individuaJ deer using a hierarchical Bayesian regression framework. Results Mule deer demonstrated remarkable spatial philopatry, with a median of 50% overlap (range: 8-78%) in year-to-year utilization distributions. Environmental conditions were the primary driver of both philopatry and range size, with anthropogenic disturbance playing a secondary role. Main conclusions Philopatry in mule deer is suspected to reflect the importance of spatial familiarity (memory) to this species and, therefore, factors driving spatial displacement are of conservation concern. The interaction between range behaviour and dynamics in development disturbance and environmental conditions highlights mechanisms by which anthropogenic environmental change may displace deer from familiar areas and alter their foraging and survival strategies. Movement reveals scale dependence in habitat selection of a large ungulate Joseph M. Northrup•, Charles R. Anderson Jr. 2, Mevin B. Hooten 3, and George Wittemyer' 'Department offish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA 2Mammals Research Section, Colorado Parks and Wildlife, Fort Collins, Colorado 80523 USA 3U.S. Geological Survey, Colorado Cooperative Fish and Wildlife Research Unit, Department offish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA 4 Department of Fish, Wildlife and Conservation Biology and Graduate Degree Program in Ecology, Colorado State University, Fort Collins, Colorado 80523 USA Citation: Northrup, J.M., C.R. Anderson. Jr., M. B. Hooten, and G. Wittemyer. 2016. Movement reveals scale dependence in habitat selection ofa large ungulate. Ecological Applications 26:2746-2757 Abstract. Ecological processes operate across temporal and spatial scales. Anthropogenic disturbances impact these processes, but examinations of scale dependence in impacts are infrequent. Such examinations can provide important insight to wildlife-human interactions and guide management efforts to reduce impacts. We assessed spatiotemporal scale dependence in habitat selection of mule deer (Odocoileus hemionus) in the Piceance Basin of Colorado, USA, an area ofongoing natural gas development. We employed a newly developed animal movement method to assess habitat selection across scales defined using animal-centric spatiotemporal definitions ranging from the local (defined from five hour movements) to the broad (defined from weekly movements). We extended our analysis to examine variation in scale dependence between night and day and assess functional responses in habitat selection patterns relative to the density of anthropogenic features. Mu le deer displayed scale invariance in the direction of their response to energy development features, avoiding well pads and the areas closest to roads at all scales, though with increasing strength of avoidance at coarser scales. Deer displayed scale-dependent responses to most other habitat features, including land cover type and habitat edges. Selection differed between night and day at the finest scales, but homogenized as scale increased. Deer displayed functional responses to development. with deer inhabiting the least developed ranges more strongly avoiding development relative to those with more development in their ranges. Energy development was a primary driver of habitat selection patterns in mule deer, structuring their behaviors across all scales examined. Stronger avoidance at coarser scales suggests that deer behaviorally mediated their interaction with development. but only to a degree. At higher development densities than seen in this area. such mediation may not be possible and thus maintenance of sufficient habitat with lower development densities will be a critical best management practice as development expands globally. 16 V �Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus) Kourtney F. Stonehouse•.2, Charles R. Anderson Jr. 1, Mark E. Peterson•.2, and Da,·id R. Collins' 1Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fon Collins, CO 90526 USA 2 Department offish, Wildlife and Conservation Biology, Colorado State University, Fon Collins, Colorado 80523 USA Citation: Stonehouse, K. F., C.R. Anderson Jr., M. E. Peterson, and D.R. Collins. 2016. Approaches to field investigations of cause-specific mortality in mule deer (Odocoi/eus hemionus). Colorado Parks and Wildlife Technical Repon No. 48, First Edition, 317 W. Prospect Rd., Ft. Collins, CO USA. DOW-R-T-48-16, ISSN 0084-8883. This technical report provides general guidelines for conducting mortality site investigations to help investigators distinguish predation from scavenging and other causes of death. General health indices are also provided to assess whether or not deer may have died from malnutrition or disease or if these factors may have predisposed deer to predation. Lastly, these guidelines will assist investigators in identifying predatory species or scavengers involved through the examination of physical evidence at deer mortality sites. The information presented here is based primarily on field experience gained from a long term research effort in northwest Colorado investigating mule deer mortality sites over several years (http://cpw.state.co.us/leam/Pages/ResearchMammalsRP-04.aspx) and literature review where referenced. We acknowledge that proximate and ultimate cause of death can be difficult or impossible to detect from field necropsy alone and examples presented here largely represent proximate causes of mortality; efforts discerning ultimate cause will require specific tissue sample collections, where possible, submitted to a veterinary diagnostic laboratory. Within this technical report are numerous photographs documenting characteristics of predator attacks on mule deer and signs left by predatory and scavenging species. Additional pictures illustrate differences between healthy and unhealthy tissues and organs. While reading this document, be aware that each mortality investigation is unique and observations in the field may differ from illustrations provided here. Appendix I provides a sample necropsy form to assist in conducting mortality investigations. Reproductive success of mule deer in a natural gas development area Mark E. Peterson', Charles R. Anderson Jr. 2,Joseph M. Northrup 1,and Paul F. Doherty Jr. 1 1 Department offish, Wildlife and Conservation Biology, Colorado State University, Fon Collins, Colorado 80523 USA Mammals Researeh Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fon Collins, CO 90526 USA 2 Citation: Peterson, M. E., C. R. Anderson Jr., J.M. Nonhrup, and P. F. Doheny Jr. 2017. Reproductive success of mule deer in a natural gas development area. Wildlife Biology doi: IO. I I I I/wlb.00341 Abstract: Natural gas development is increasing across North America and causing concern over the potential impacts on wildlife populations and their habitat, particularly for ungulate species. Understanding how this development impacts reproductive success metrics that are influential for ungulate population dynamics is important to guide management of ungulates. However, the influences of natural gas development on reproductive success metrics of mule deer Odocoi/eus hemionus have not been studied. We used statistical models to examine the influence of natural gas development and temporal factors on reproductive success metrics of mule deer in the Piceance Basin, northv,est Colorado during 2012-2014. We focused on study areas with relatively high or low levels of natural gas development. Pregnancy and in utero fetal rates were high and statistically indistinguishable between study areas. Fetal survival rates increased over time and survival was lower in the high versus low development study areas in 2012 possibly influenced by drought coupled with habitat loss and fragmentation associated with development. Our novel results suggest managers should be concerned with the influences of development on fetal survival, particularly during extreme environmental conditions (e.g. drought) and our results can be used to guide development planning and/or mitigation. Developers and wildlife managers should continue to collaborate on development planning, such as implementing habitat treatments to improve forage availability and quality, minimizing disturbance to hiding and foraging habitat particularly during parturition, and implementing directional drilling to minimize pad disturbance density to increase fetal survival in developed areas. 17 �Variation in ungulate body fat: individual versus temporal effects EricJ. Bergman•. Charles R. Anderson Jr:, Chad J. Bishop'. A. Andrew Holland 1• and Joseph M. Northrup2 1 Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 90526 USA 2Oepartment of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA Citation: Bergman, E. J., C.R. Anderson Jr., C. J. Bishop, A. A. Holland, and J.M. Northrup. 2018. Variation in ungulate body fat: individual versus temporal effects. Journal of Wildlife Management 82:130-137, 001: 10:1002/jwmg.21334 ABSTRACT The use of ultrasonograhic measurements of muscle and body fat represent a relatively new data stream that can be used to address questions regarding ungulate condition. We have learned that measurements of body fat and presumably overall body condition among individual animals, even those taken from the same herd at that same time, are highly variable. Relatively little consideration has been given to the sources of variation in body fat and other physiological parameters in wildlife populations. We evaluated the components of variation in late-winter mule deer (Odocoileus hemionus) body fat estimates: sampling variation (i.e., variation induced by the particular set of individuals that were sampled) and process variation (i.e., variation stemming from biological processes) with a long-term data set (2002-2015) from Colorado, USA. We collected our data from across Colorado as part of historical research, ongoing research, and periodic population monitoring programs. Mean percent ingesta-free body fat (%1FBF) for sampled mule deer was 7.20 :I: 1.20% (SD). Covariates related to individual deer explained approximately 4% of the total variation in %1FBF and annual effects explained an additional 13%ofthe variation. Substantial residual variation in %IFBF (83%) remained unexplained. The source of the 83% of unexplained variation is partially linked to fine-scale spatial dynamics but also additional individual metrics we were unable to capture, primarily the presence or absence of dependent young. We speculate that the primary factors influencing late-winter mule deer body fat and overall condition are individual in nature. These results present a cautionary check on herd level inference that can be made from individual late-winter body fat estimates and we postulate that for mule deer, alternative and additional body condition metrics may offer added utility in management scenarios. However, an important next step to better understand wildlife population health is to evaluate the sources and magnitude of variation within other body condition metrics, with the goal of further refining data that can better allow biologists to incorporate herd heaJth into population management recommendations. Mortality of mule deer fawns in a natural gas development area Mark E. Peterson•, Charles R. Anderson Jr. 2• Joseph M. Northrup 1,and Paul F. Doherty Jr. 1 1 Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 90526 USA 2 Citation: Peterson, M. E., C.R. Anderson Jr., J.M. Northrup, and P. F. Doherty Jr. 2018. Mortality of mule deer fawns in a natural gas development area. Journal of Wildlife Management 82:1135-1148, DOI: 10.1002/jwmg.21476 ABSTRACT Recent natural gas development has caused concern among wildlife managers, researchers, and stakeholders over the potential effects on wildlife and their habitats. Specifically, understanding how this development and other factors influence mule deer (Odocoileus hemionus) fawn (i.e., 0-6 months old) mortality rates, recruitment, and subsequently population dynamics have been identified as knowledge gaps. Thus, we tested predictions concerning the relationship between natural gas development. adult female, fawn birth, and temporal (weather) characteristics on fawn mortality in the Piceance Basin of northwestern Colorado, USA, from 2012-2014.We captured and radio-collared 184 fawns and estimated apparent cause-specific mortality in areas with relatively high or low levels of natural gas development using a multi-state model. Mean daily predation probability was similar in the high versus low development areas. Predation was the leading cause of fawn mortality in both areas and decreased from 0-14 days old. Black bear ( Ursus americanus; 22% of all mortalities, n = 17) and cougar (Fe/is concolor; 36% of all mortalities, n = 6) predation was the leading cause of mortality in the high and low development areas, respectively. Predation of fawns was negatively correlated with the distance from a female's core area to a producing well pad on winter or summer range. Contrary to expectations, predation of fawns was positively correlated with rump fat thickness of adult females. Well pad densities and development activity were relatively low during our study, indicating that the observed intensity of development did not appear to influence daily predation probability. Our results suggest maintaining development activity thresholds at levels we observed to potentially minimize the effects of development on fawn mortality. However, we caution that higher development intensity and drilling activity in flatter, less rugged areas with less concealment cover could influence fawn mortality. Managers should maintain low development densities in areas where topography and vegetation offer less concealment. Overall, region-specific data (e.g., development intensity, topography, predator assemblages, and associated predation risk) are needed to better understand the effects of natural gas development on fawn mortality. 18 �Using maternal mule deer movements to estimate timing of parturition and assist fawn captures Mark E. Peterson•. Charles R. Anderson Jr. 2, Mathew W. Alldredge2.and Paul F. Doherty Jr.' 'Department of Fish, Wildlife and Conservation Biology, Colorado State University. Fon Collins, Colorado 80523 USA 2Mammals Research Section. Colorado Parks and Wildlife, 317 W. Prospect Road, Fon Collins, CO 90526 USA Citation: Peterson. M. E., C. R. Anderson Jr., M. W. Alldredge, and P. F. Doherty Jr. 2018. Using maternal mule deer movements to estimate timing of parturition and assist fawn captures. Wildlife Society Bulletin 42:616-621; DOI: 10.1002/wsb.935 ABSTRACT Movement patterns of maternal ungulates have been used to detennine parturition dates and aid in locating fawns, which may be important for understanding reproductive rates (e.g., pregnancy and fetal), but such methods have not been validated for mule deer (Odocoileus hemionus). We first detennined timing of parturition using vaginal implant transmitters (VITs) and then predicted timing of parturition using VITs in conjunction with Global Positioning System collar data in the Piceance Basin of northwestern Colorado, USA, during 2012-2014. We examined daily movement rate to detennine differences in movement rate among days (7 days pre- and postpartum) and for movement patterns indicative of parturition. Mean daily movement rate (m/day) of 102 maternal deer decreased by 46% from 1 day preparturition (mean = 1,253, SD= 1,091) to parturition date (mean = 682, S = 574), and remained at this low rate 1-7 days postpartum. We applied an independent data set to validate predicted parturition dates based on daily movement rate. We estimated day of parturition correctly (i.e., day 0), within 1-3 days postparturition, and_4 days postparturition of field-reported dates for 10 (29%), 21 (60%), and 4 ( 11 %) maternal females, respectively. For novel data sets, we predict that a mule deer female whose daily movement rate decreases by _46% and remains low _3 days postparturition particularly when preceded by a sudden increase in movement-has given birth. However, we caution that disturbance of deer by field crews should be minimized, and if birth sites are not found, neonatal monality will be underestimated. Our results can help detennine timing and general location of parturition as an aid in capturing fawns when the use of VITs is not feasible, with the ultimate objective of estimating pregnancy, fetal, and fawn survival rates if birth sites are found. On-animal acoustic monitoring provides insight to ungulate foraging behavior Joseph M. Northrup•, Alexandra Avrin 1• Charles R. Anderson, Jr. 2, Emma Brownl, and George Wittemyer 1 'Department of Fish, Wildlife and Conservation Biology, Colorado State University. Fort Collins, Colorado 80523 USA ?Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fon Collins, CO 90526 USA 3 National Parle Service Natural Sounds and Night Skies Division, Fon Collins, CO 80525 USA Citation: Nonhrup, J.M., A. Avrin, C.R. Anderson Jr., E. Brown, and G. Wittemyer. 2019. On-animal acoustic monitoring provides insight to ungulate foraging behavior. Journal of Mammalogy I 00: 1479-1489: https·//doi org/ I0.1093/jmammal/gyz 124 Abstract Foraging behavior underpins many ecological processes; however. robust assessments of this behavior for free-ranging animals are rare due to limitations to direct observations. We leveraged acoustic monitoring and GPS tracking to assess the factors influencing foraging behavior of mule deer (Odocoileus hemionus). We deployed custom-built acoustic collars with GPS radiocollars on mule deer to measure location-specific foraging. We quantified individual bites and steps taken by deer, and quantified two metrics of foraging behavior: the number of bites taken per step and the number of bites taken per unit time. which relate to foraging intensity and efficiency. We fit statistical models to these metrics to examine the individual, environmental. and anthropogenic factors influencing foraging. Deer in poorer body condition took more bites per step and per minute and foraged for longer irrespective of landscape properties. Other patterns varied seasonally with major changes in deer condition. In December, when deer were in better condition, they took fewer bites per step and more bites per minute. Deer also foraged more intensely and efficiently in areas of greater forage availability and greater movement costs. During March, when deer were in poorer condition, foraging was not influenced by landscape features. Anthropogenic factors weakly structured foraging behavior in December with no relationship in March. Most research on animal foraging is interpreted under the framework of optimal foraging theory. Departures from predictions developed under this framework provide insight to unrecognized factors influencing the evolution of foraging. Our results only confonned to our predictions when deer were in better condition and ecological conditions were declining, suggesting foraging strategies were state-dependent. These results advance our understanding of foraging patterns in wild animals and highlight novel observational approaches for studying animal behavior. 19 �A noninvasive automated device for remotely collaring and weighing mule deer Chad J. Bishop', Mathew W. Alldredge 1, Daniel P. Walsh 1, Eric J. Bergman', Charles R. Anderson Jr. 1, Darlene Kilpatrick, Joe Bakel 2,and Christophe Fabvre2 1Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526 USA 20ynamic Group Circuit Design, Inc., 2629 Redwing Road, Fort Collins, CO 80525 USA Citation: Bishop, C. J., M. W. Alldredge, D. P. Walsh, E. J. Bergman, C.R. Anderson Jr., D. Kilpatrick, J. Bakel, and C. Fabvre. 2019. A noninvasive automated device for remotely collaring and weighing mule deer. Wildlife Society Bulletin 43:717-725; doi.org/10.1002/wsb.1034 ABSTRACT Wildlife biologists capture deer (Odocoileus spp.) annually to attach transmitters and collect basic information (e.g., animal mass and sex) as part of ongoing research and monitoring activities. Traditional capture techniques induce stress in animals and can be expensive, inefficient, and dangerous. They are also impractical for some urbanized settings. We designed and evaluated a device for mule deer (0. hemionus) that automatically attached an expandable radiocollar to a ~-month-old fawn and recorded the fawn's mass and sex, without physically restraining the animal. The device did not require on-site human presence to operate. Students and faculty in the Mechanical Engineering Department at Colorado State University produced a conceptual model and early prototype. Professional engineers at Dynamic Group Circuit Design, Inc. in Fort Collins, Colorado, USA, produced a fully functional prototype of the device. Using the device. we remotely collared, weighed, and identified sex of8 free-ranging mule deer fawns during winters 2010-2011 and 2011-2012. Collars were modified to shed from deer approximately I month after the collaring event. Two fawns were successfully recollared after they shed the first collars they received. Thus, we observed 10 successful collaring events involving 8 unique fawns. Fawns demonstrated minimal response to collaring events, either remaining in the device or calmly exiting. A fawn typically required :::I weeks of daily exposure before fully entering the device and extending its head through the outstretched collar, which was necessary for a collaring event to occur. This slow acclimation period limited utility of the device when compared with traditional capture techniques. Future work should focus on device modifications and altered baiting strategies that decrease fawn acclimation period, and in turn. increase collaring rates, providing a noninvasive and perhaps cost-effective alternative for monitoring mid- to large-sized mammal species. © 2019 The Wildlife Society. Behavioral and demographic responses of mule deer to energy development on winter range Joseph M. Northrup', J. M., Charles R Anderson Jr. 2, Brian D. Gerber, and George Wittemyer 1 1Department offish, Wildlife and Conservation Biology, Colorado State University. 1474 Campus Deliveiy, Fort Collins, CO 80523 USA 2 Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fon Collins, CO 80S26 USA Department of Natural Resources Science, University of Rhode Island. I Greenhouse Road, Kingston, RI 02881 USA 3 Citation: Northrup, J.M., C.R. Anderson Jr., B. D. Gerber, and G. Wittemyer. /n Press. Behavioral and demographic responses of mule deer to energy development on winter range. Wildlife Monographs. ABSTRACT Anthropogenic habitat modification is a major driver of global biodiversity loss. In North America, one of the primary sources of habitat modification over the last two decades has been exploration for and production of oil and natural gas (hydrocarbon development), which has led to demographic and behavioral impacts to numerous wildlife species. Developing effective measures to mitigate these impacts has become a critical task for wildlife managers and conservation practitioners. However, this task has been hindered by the difficulties involved in identifying and isolating factors driving population responses. Current research on responses of wildlife to development predominantly quantifies behavior, but it is not always clear how these responses scale to demography and population dynamics. Concomitant assessments of behavior and population-level processes are needed to gain the mechanistic understanding required to develop effective mitigation approaches. We simultaneously assessed the demographic and behavioral responses of a mule deer population to natural gas development on winter range in the Piceance Basin of Colorado, USA from 2008 to 2015. Notably, this was the period when development declined from high levels of active drilling to production phase activity (i.e., no drilling). We focused our data collection on two contiguous mule deer winter range study areas that experienced starkly different levels of hydrocarbon development within the Piceance Basin. We assessed mule deer behavioral responses to a range of development features with varying levels of associated human activity by examining habitat selection patterns of nearly 400 individual adult female mule deer. Concurrently, we assessed the demographic and physiological effects of natural gas development by comparing annual adult female and over-winter fawn (6-monthold animals) survival, December fawn mass, adult female late and early winter body fat, age, pregnancy rates, fetal counts and lactation rates in December across the two study areas. Strong differences in habitat selection between the two study areas were apparent. Deer in the less developed study area avoided development during the day and night, while selecting habitat presumed to be used for foraging. Deer in the heavily developed study area selected habitat presumed to be used for thermal and security cover to a greater degree. Deer faced with higher densities of development avoided areas with more well pads during the day and responded neutrally or selected for these areas at night. Deer in both study areas showed a strong reduction in use of areas around well pads that were being drilled, which is the phase of energy development associated with the greatest amount of human presence, vehicle traffic, noise and artificial light. Despite divergent habitat selection patterns, we found no effects of development on individual condition or reproduction 20 V �and found no differences in any of the physiological or vital rate parameters measured at the population level. However, deer density and annual increases in density were higher in the low development area. Our results indicated that deer in the less developed area avoided development, whereas those in the heavily developed area altered their behavioral patterns to use habitat with more cover and areas near development during times when there was reduced human activity. The recorded behavioral alterations did not appear to be associated with demographic or physiological costs, possibly because populations are below winter range carrying capacity. We discuss potential drivers of the difference in population density between the two areas, suggesting development caused a population decline prior to our study (when development was initiated) or that there were area specific differences in habitat quality, juvenile dispersal, neonatal or juvenile survival; however, we lack the required data to contrast evidence for these mechanisms. Given our results, it appears that deer can adjust to relatively high densities of well pads in the production phase (the period with markedly lower human activity on the landscape), provided there is sufficient vegetative and topographic cover afforded to them and populations are below carrying capacity. The strong reaction to wells in the drilling phase of development suggests mitigation efforts should focus on this activity and stage of development. Many of the wells in this area were directionally drilled from multiple well pads, leading to a reduced footprint of disturbance, but still drove strong behavioral responses. Our results also indicate the likely value of mitigation efforts focusing on reducing human activity (i.e., vehicle traffic, light and noise). In combination, these findings indicate that attention should be paid to the spatial configuration of the final development footprint to ensure adequate cover. In our study system, minimizing the road network through landscape-level development planning would be valuable (i.e., exploring a maximum road density criteria). Lastly, our study highlights the importance of concomitant assessments of behavior and demography to provide a comprehensive understanding of how wildlife respond to habitat modification. 21 �Appendix B. Preliminary results of habitat treatment responses and herbivore use of treated sites. Vegetation and camera data to accompany the study' Population performance ofPiceance Basin mule deer in response to natural gas resource selection and mitigation efforts to address human activity and habuatdegradation' Principal Investigators: Danielle Johnston (Danielle.bilyeu@state.co.us). Chuck Anderson (chuck.anderson@state.co.us) Collaborators: Colorado Parks and Wildlife, BLM-White River Field Office, Idaho State University, Colorado State University, Federal Aid in Wildlife Restoration, EnCana Corp., ExxonMobil Prod. Co./XTO Energy, Marathon Oil Corp., Shell Petroleum, WPX Energy, Colorado Mule Deer Assn., Muley Fanatic Found., Colorado Mule Deer Found., Colorado State Severance Tax Fund, Boone & Crocket Club, and Safari Club Int. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW's right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. In 2011 and 2013, about 1,200 acres of pinyon and juniper (PJ) mastication treatments were completed in the Magnolia region of the Piceance Basin. Treated parcels averaged 7 acres in size, and were intended to increase winter range quality for deer. The treatments were part of a study to evaluate the effectiveness of PJ removal as mitigation for impacts of natural gas development on deer, with outcomes assessed in terms of deer population and demographic parameters. This summary addresses some side questions relevant to the main study, with outcomes assessed in terms of vegetation response and animal use of vegetation treatments. We were interested in quantifying the understory forage produced by the mastication treatments. We used paired masticated/control point-intercept transects on a subset of parcels (Graham 2013) to quantify cover of plant groups relevant to deer nutrition. We used belt transects and trained ocular estimation, with benchmarks (Johnston 2018), to estimate summer utilization on individual shrubs, then scaled these to the plot level (Bilyeu. Cooper et al. 2007). We used belt transects of shrub canopy measurements, coupled with biomass equations developed for the study area (Johnston 2018) to quantify winter forage production of key browse species. Winter forage production was defined as current-year stems, not including leaves, not including biomass removed by summer browsing, and not including very small stems which would likely be shed prior to winter (Johnston 2018). We were interested in how summer use of treatments, and use of treatments by non-target animals, impacted winter forage availability. Ten cattle exclosures, distributed broadly throughout the study area (Figure 1), were built within mastication treatments in 2011 and 2013. We assessed plant cover and summer shrub utilization within these using techniques described above. On paired masticated/control transects, we deployed Reconyx Hyperfire cameras July-November 2018-2019. These were programmed to facilitate creating an index of use: 5 pictures per motion trigger, 3 second interval between pictures, a 5 minute wait time between triggers, and a sensitivity setting of High (Rhodes, Larsen et al. 2018). An animal observed with their head down or other indication of foraging in one or more of the photos in a 5 photo set was counted as one foraging event, and non-foraging occurrences were counted similarly. Sampling efforts by year are given in Table I. Because the plant cover data contained many zeros, we modeled presence/absence of each plant group separately from its cover where present (Fletcher, Mackenzie et al. 2005), using the lme4 package in R (Bates 2005). For both analyses, treatment, year, and their interaction were considered fixed effects, year was included as a categorical variable, and pair ID and plot ID were included as random effects. We used a similar approach for camera data for cattle and elk, which also contained many zeros. 22 �1 ~ ~ In general, grasses responded positively to treatment (Figure 2a). Wheatgrass presence, wheatgrass cover, and needlegrass presence were higher in treated than untreated plots. Poa grass presence was higher in treated plots by 2018, although poa grass presence and cover initially had a negative response to treatment. Cheatgrass presence also responded positively to treatment (Figure 2a). Wheatgrasses, poas, and cheatgrass all had significant year•treatment interactions for either presence or cover. Interannual variation in cover was greater in masticated plots than in control plots for these species groups (Figure 2a). Forbs responded positively to treatment. Annual forb and perennial forb presence were higher in treated than untreated plots (Figure 2b ). Some shrubs responded positively to treatment, while others did not. Snowberry cover was lower in treated plots in 2013, but in 2016 and 2018, cover was higher in treated plots (Figure 2c). Variation in snowberry cover was greater in masticated than in control plots (Figure 2c). Bitterbrush did not display any significant effects until 2018, when cover was higher in treated plots (Figure 2c). Serviceberry cover was lower in treated plots over all years (Figure 2d). Sagebrush cover was initially lower in treated plots, but by 2018 this difference was no longer significant (Figure 2d). Summer utilization of serviceberry and mountain mahogany in 2018 was significantly higher in masticated than in control plots, but no differences were detected in bitterbrush or sagebrush. Winter forage production, which was summed over serviceberry, mountain mahogany, and bitterbrush, was significantly higher in masticated plots than in unmasticated plots in all years except 2016, when the pattern was reversed (Figure 3). There was no significant effect of exclosures on any plant cover group or on summer utilization in 2018. Deer, horse, elk, and cattle all foraged more often in masticated plots than in controls in 2018 (Figure 4). Cattle were only observed foraging at 6 of20 locations, horse were observed at 9, deer at 19, and elk at 6. Mastication treatments had many positive effects on forage availabilty, including higher cover of desirable grass groups such as poa grasses and wheatgrasses, higher cover of perennial forbs, and usually higher productivity of winter-available shrub forage. There were some negative effects and some differences in effects among years, however. Cheatgrass was higher in masticated plots than in controls, and snowberry cover was higher in masticated plots in 2016 and 2018. 2016 was an unusual year compared to other years of this study, with very high productivity of grasses (including cheatgrass, especially in masticated plots), and unusually high productivity of winter-available forage of desirable shrubs in control but not masticated plots. Summer shrub utilization in 2018 was higher in masticated plots than in controls. We lack any data on utilization from 2016, which might have helped explain if the lower production of winter-available forage in masticated plots was due to higher summer utilization in those plots that year. Another explanation for the 2016 results is that good conditions for grass, cheatgrass, and/or snowberry productivity in masticated plots led to increased competition which lessened productivity of desirable forage shrubs. All four of the large herbivores of interest foraged more frequently in summer and fall in masticated plots than in control plots in 2018. The impact of cattle was concentrated in only a few plots, but they did forage frequently in plots where they occurred. Cattle use ended in September, prior to the period of heavy use by deer in October. The data from the cattle exclosures does not indicate that cattle are having any measurable negative effect on forage resources. In summary the impact of cattle on the forage resources available to deer in mastication treatments seems minimal. However, the effect of the sum of cattle, horse, and elk foraging may have some impact. In 2019, we collected vegetation data and camera data. 2019 is the last year of data collection for this study, and final analyses will be incorporated into publications in 2020-21. LITERATURE CITED Bates, D. (2005). "Fitting linear mixed models in R." R news 5( 1). Bilyeu, D. M., D. J. Cooper and N. T. Hobbs (2007). "Assessing impacts of large herbivores on shrubs: tests of scaling factors for utilization rates from shoot-level measurements." Journal of Applied Ecology 44(1 ): 168-175. 23 �Fletcher, D., D. D. Mackenzie and E. Villouta (2005). "Modelling skewed data with many zeros: a simple approach combining ordinary and logistic regression." Environmental and ecological statistics 12: 45-54. Graham, T. (2013). Magnolia habitat manipulation project vegetative monitoring: June 2013 notes on data collection and methods used, Ranch Advisory Partners, LLC: 7. Johnston, D. B. (2018). Wildlife Research Report: Examining the effectiveness of mechanical treatments as a restoration technique for mule deer habitat. Fort Collins, CO, Colorado Parks and Wildlife. Rhodes, A. C., R. T. Larsen and S. B. S. Clair (2018). "Differential effects of cattle, mule deer, and elk herbivory on aspen forest regeneration and recruitment." Forest Ecology and Management 422: 273280. u 24 �Table I. Number of transects sampled for a given data type each year. 2011 2012 2013 2014 2015 2016 2018 Variables quantified 69 90* 90* 159 145 107t Percent cover of plant functional groups 70t 27t 63 Winter-available forage of bitterbrush, serviceberry, mountain mahogany (ShrubMassPerArea) Summer utilization of bitterbrush, serviceberry, mountain mahogany, and sagebrush Index of deer, elk, horse, and cattle use in summer and fall, as determined by trail camera (EventsPerDay) * Pretreatment data collected 2011-2012 will be added to a later report. tlncludes 24-30 locations taken at exclosure sites. 25 ------------------------ -··-·- ··-------- 2019 40 (camera sites) 75t 75t 75t 75t 40 (2 40 (2 cameras each) cameras each) �Figurl.! I. Sampling locations ll'ithin the .\lagnolia region o(the Piceance Basin. 16 �b 10 o- a .... Q) 15 - .... Q) > > 0 0 (.) 'E Q) 7 5- (.) 10 - ....Q) (.) wht;.,Sss + . PC 'E Q) ' a. # # a. 5· 5.0 - ~ Q) o- 2500- I 2013 ' 2016 2014 2018 2013 year C , -.... ,r-' \ (.) C \ 6- 2016 2018 d 0- sno~berr,,+, C ,, > 0 2018 year T 9.... Q) I ' 2016 ' 2014 ' Q) (.) Q) a. 3- '\ .... Q) 6 > s 0 (.) 'E Q) 4 - ....Q) (.) 'T 0. 2- 2013 2014 2016 20 13 2018 ' 2014 year year Figure]. Cover ofsome pla111f11nctional groups and spl!Cies i111porta111.fhr eval11a1i11g hahilal <Jllaliz,•. Dashed lines indicale maslica!ed plots and solid lines are co111m/.1·. .·/ .. _ .. or .. _.. sign indica/es significant posit in: or ncgatin: 111uin effect of 111asticatiu11 C1c:ro.u _1·c w ·.,· (<l 0 . 051. .. l' .. indic:at,•.,· that !he sign[ficant ej /ect was obsen•ed in the prese11ce 1ahse11ce ww~i·s is. and "( ' .. i11dicales a si~11i/icant effect in !he cover-where -prese111 analysis. 27 �- 1....... : i: - . ~ Treatment "'~ "' .D • masbcatec 2 9C - control -t 1 .r:: V, .., C: ::, a., a. <l> b.O r.n - ~ .2 .!!! .D * * ~ "' "'> 30 • "' .:.. 1 1 * T T 1 * + T -'- 2:! C: 3 . o· :"\ •J 20 1 : .' 016 . o,e year Figure 3. Mass <?f'11·i11ter-a1•ailahle.fi>rnge rcurrent-year stem muss measured in Septe111her. 1101 i11cludi11g /eal'es or mass re11101•ed hy s11111111er hrmrsingJ p<!r unit shrub area. Data are .1·11111111ed urer .1·erl'icehen:1·. mountain 111ahoga11y. and hillerhrush. ,\'= 8.fi>r 1013 and 1015 and 15-31.fhr other years. No 1ra11sects i11side.fe11ce.1· ll'ere i11cluded Error har.1· = SE. Stars indicate sign(fica111 d(fferences at alpha = (J.()5 28 �Not foraging, near camera Foraging, near camera b 75 · >. 75 · (l! 0 iii c.. (l! I Treatment 50 · iii 50 · c.. f/) control Q) ■ masticated c > w en .S 25· en 25 · ~ 0 u.. O· O· ' callle ' elk ' horse canle mule deer ' elk horse mule' deer Herbivore Herbivore Fig ure -I. a) Average number a/foraging events per hectare per day between mid-July and midNovember, 2018 in control versus masticated plots. Stars indicate signijlcant differences at a. = 0. 05. t indicates a sign(ficant d(fference in presence offoraging events. b) Average number of non.foraging observations per hectare per day. 29 �"--'. Colorado Parks and Wildlife July I, 2020 - June 30, 2021 WILDLIFE RESEARCH REPORT Colorado : Parks and Wildlife ---------==--------3430 : =M=amm==a=l~s.; ;. ;R;. a:; es=--e--ar__c__h______________ State of Cost Center Work Package Task No. .:..=.=.::.::...=;=-~===------------- 3001 6 : =D-"'"e___ er_C__o__n__s___ erv~at___io__n_______________ : Population Performance of Piceance Basin Mule Deer in Response to Natural Gas Resource Extraction and Mitigation Efforts to Address Human Activity and Habitat Degradation Federal Aid Project: ______ W__-___ 24.....3.....-.aa..;RS _______ Period Covered: July l, 2020 - June 30, 2021 Author: C.R. Anderson, Jr. Personnel: D. Bilyeu-Johnston, K. Aagaard, CPW; J. Northrup, Ontario Ministry ofNatural Resources and Forestry; B. Gerber, University of Rhode Island; G. Wittemyer, Colorado State University. Project support received from Federal Aid in Wildlife Restoration, Colorado Mule Deer Association, Colorado Mule Deer Foundation, Muley Fanatic Foundation, Colorado State Severance Tax Fund, Caerus Oil and Gas LLC, EnCana Corp., ExxonMobil Production Co./XTO Energy, Marathon Oil Corp., Shell Petroleum, Williams and WPX Energy. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW's right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. Colo Para Wlklllfa RaNnll'I Ub 11~111mm1mm11~mm~1~1~111111 3 2333 00000 1059 I �WILDLIFE RESEARCH REPORT POPULATION PERFORMANCE OF PICEANCE BASIN MULE DEER IN RESPONSE TO NATURAL GAS RESOURCE EXTRACTION AND MITIGATION EFFORTS TO ADDRESS HUMAN ACTIVITY AND HABITAT DEGRADATION CHARLES R. ANDERSON, JR PROJECT NARRITIVE OBJECTIVES 1. To determine experimentally whether enhancing mule deer habitat conditions on winter range elicits behavioral responses, improves body condition, increases fawn survival, and ultimately, population density on mule deer winter ranges exposed to extensive energy development. 2. To determine experimentally to what extent modification of energy development practices enhance habitat selection, body condition, fawn survival, and winter range mule deer densities. SEGMENT OBJECTIVES 1. Complete publication of mule deer behavioral and demographic responses to energy development activity. 2. Continue analyses and begin manuscript preparation to address effectiveness of habitat treatments as a mitigation option for mule deer management and energy development planning. PROJECT OVERVIEW AND RESEARCH SUMMARY We propose to experimentally evaluate winter range habitat treatments and human-activity management alternatives intended to enhance mule deer ( Odocoileus hemionus) populations exposed to energy-development activities. The Piceance Basin of northwestern Colorado was selected as the project area due to ongoing natural gas development in one of the most extensive and important mule deer winter and transition range areas in Colorado. The data presented here represent preliminary and final results of a 10-year research project addressing habitat improvements as mitigation and evaluation of deer responses to energy development activities to inform future development planning options on important seasonal ranges. From2008-2019, we monitored deer on 4 winter range study areas representing relatively high (Ryan Gulch, South Magnolia) and low (North Magnolia, North Ridge) levels of development activity (Fig. 1) to address factors influencing deer behavior and demographics and to evaluate success of habitat treatments as a mitigation option. We recorded adult female habitat use and movement patterns; estimated neonatal, overwinter fawn and annual adult female survival; estimated annual early and late winter body condition, pregnancy and fetal rates of adult females; and estimated annual mule deer abundance among study areas. Winter range habitat improvements completed spring 2013 resulted in 604 acres of mechanically treated pinion-juniper/mountain shrub habitats in each of 2 treatment areas (Fig. 2) with minor (North Magnolia) and extensive (South Magnolia) energy development, respectively. During this research segment, we finalized publication of mule deer behavioral and demographic responses to energy development activity (Northrup et al. 2021; Appendix A) and continued ~nalyses of vegetation and mule deer responses to habitat treatments intended as a mitigation option to offset energy development disturbance (preliminary results reported in Appendix B). Based on final (migration, mule deer behavioral and demographic responses, reproductive success and neonate survival; see Anderson 2019 for detailed methods and results and Appendix A for publication abstracts) and preliminary data 2 V �analyses (vegetation and herbivore response to habitat treatments, Appendix B) for this 10-year project: (1) annual adult female survival was consistent among areas averaging 79-87% annual1y, but overwinter fawn survival was variable, ranging from 31 % to 95% within study areas, with annual and study area differences primarily due to early winter fawn condition, annual weather conditions, and factors associated with predation on winter range; (2) mule deer body condition early and late winter was generally consistent within areas. with higher variability among study areas early winter, primarily due to December lactation rates. and late winter condition related to seasonal moisture and winter severity; (3) late winter mule deer densities increased through 2016 in all study areas, ranging from 50% in North Ridge to I03% in North Magnolia, but have stabilized recently in 3 of the 4 study areas with recent decline evident in North Ridge (Fig. 3); (4) migratory mule deer selected for areas with increased cover and increased their rate of travel through developed areas, and avoided negative influences through behavioral shifts in timing and rate of migration, but did not avoid development structures (Fig. 4); (5) mule deer exhibited behavioral plasticity in relation to energy development. without evidence of demographic effects, where disturbance distance varied relative to diurnal extent and magnitude of development activity (Fig 5), which provide for useful mitigation options in future development planning; and (6) energy development activity under existing conditions did not influence pregnancy rates, fetal rates or early fawn survival (0-6 months), but may have reduced neonatal survival (March until birth) during 2012 when drought conditions persisted during the third trimester of doe parturition (Fig. 6). Final results are pending to address vegetation and mule deer responses to assess habitat treatment mitigation options for energy development planning. Final data collection efforts for this project were completed by spring 2020 (final GPS collar recovery). Collaborative research with agency biologists, graduate students. and university professors has produced 22 scientific publications addressing improved monitoring techniques for neonate mule deer captures (Bishop et al. 2011, Peterson et al. 2018b); development and evaluation of a remote mule deer collaring device (Bishop et al. 2019); mule deer migration relative to energy development (Lendrum et al. 2012, 2013, 2014; Anderson and Bishop 2014), improved approaches to address animal habitat use patterns (Northrup et al. 2013); mule deer response to helicopter capture and handling (Northrup et al. 2014a); potential effects of male-biased harvest on mule deer productivity (Freeman et al. 2014 ); mule deer genetics in relation to body condition and migration (Northrup et al. 2014b); acoustic monitoring to investigate spatial and temporal factors influencing mule deer vigilance (Lynch et al. 2014) and foraging behavior (Northrup et al. 2019); the relationship of plant phenology with mule deer body condition (Sera) et al. 2015); approaches to identify cause-specific mortality in mule deer from field necropsies (Stonehouse et al. 2016); the influence of individual and temporal factors affecting late winter body condition estimates of adult female mule deer (Bergman et al. 2018); and mule deer behavioral and demographic responses to energy development activities to inform future development planning (Northrup et al. 2015, 2016a. 2016b, 2021. Peterson et al. 2017, 2018a). These publications are summarized in Appendix A and preliminary results describing vegetation and herbivore responses to habitat treatments are reported in Appendix B. We anticipate the opportunity to work cooperatively toward developing solutions for allowing the nation·s energy reserves to be developed in a manner that benefits wildlife and the people who value both the wildlife and energy resources of Colorado and elsewhere. 3 �Mule Deer Winter Range Study Areas Mule deer 9tucly areas Well Pl dl & FacllltiOI CJ llOM uaono,.a ! 1n .leY~k>pm~ nl soorn Magno.aa 11"'111 ~""1• 10 Figure 1. Mule deer winter range study areas relati ve to active natural gas well pads and energy development fac ilities in the Piceance Basin of northwest Colorado. wi nter 20 13/14 (Accessed http://cogcc.state.co.us/ December 31. 20 13: energy development dri ll ing activity has been minor since 20 13). 4 �North M agnolia trea tement sites (587 acre s) LJ BearSet_ 15_ 35b_andG SearSet_ 1_8anc1A_E c::::J BearSet_36_54 andJ GreasewoodSet_g 16_g29 GreasewoodSet_g1 _g 15 c::::J Greasewood5et_g30_g42 LeeOvers,ghts_a_fandl 6_ 17 Mechanical treatrrent companson {54 acres) - - North Hatch Pilot Treatments ( 116 acres) South Magnol•a Figure 2. Habitat treatment site de! ineations in 2 mule deer study areas (604 acres each) of the Piceance Basin. northwest Colorado (Top; cyan polygons completed Jan 2011 using hydro-axe; yellow polygons completed Jan 20 12 using hydro-axe. roller-chop. and chaining: and remaining polygons completed Apr 2013 using hydro-axe). January 20 11 hydro-axe treatment-site photos from North Hatch Gulch during April (Lower left. aerial view) and October. 20 I I (Lower right. ground view). 5 �Piceance Basin late winter mule deer density 35.00 30.00 25.00 e 20.00 ~ - ~ cu 15.00 cu - North Ridge • • • • • • Ryan Gulch 0 10.00 5.00 - • North Magnolia - South Magnolia 0.00 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Year Figure 3. Mule deer density estimates and 95% Cl (e1rnr bars) from 4 wi nter range herd segments in the Piceance Basin, northwest Colorado, late winter 2009-2018. Figure 4. Mule deer study areas in the Piceance Basin of northwestern Colorado, USA (Top), spring 2009 migration routes of adult female mule deer (11 = 52: Lower left). and active natural-gas well pads (black dots) and roads (state. county, and natural-gas: white lines) from May 2009 (Lower right: from Lendrum et al. 20 12; http://dx.doi.org/ I 0. 1890/ES 12-00 165. I). 6 �I:~<f •• "' ..... •··r -r ·+ --+ g 7 ... ' -' - - - - - - , - - - , - - - M Prod 400 Prod 600 Prod 600 Prod 1000 Onn i:oo Onll 600 Onll 800 Crill 1000 Onfl 600 Ortn 800 Qt-NI H)OO Ca1anates '- M P,oa 400 Ptod 600 Prod 800 Proa 1000 Ot\11 400 ' I Figure 5. Posterior distributions of population-level coefficients related to natural gas development for RSF models during the day (top) and night (bonom) for 53 adult female mule deer in the Piceance Basin. northwest Colorado. Dashed line indicates 0 selection or avoidance (below the line) of the habitat features. ' Drill' and ·Prod ' represent drilling and producing well pads, respectively. The numbers following 'Drill' or 'Prod· represent the distance from respective well pads evaluated (e.g., 'Drill 600' is the number of well pads with active drilling between 400-600 m from the deer location: from Northrup et al. 20 15; http://onlinelibrary.wilev.com/doi/ I 0. 111I /gcb. 13037/abstract). Road disturbance was relatively minor (- 60-120 m. not illustrated above). 1.00 QJ 0.80 r---a-- - ._ -- + "§ ro> 0.60 ·2: :::; en cii 0.40 Q) u.. 0.20 0.00 2012 2013 Year □ High development 2014 □ Low development I Figure 6. Model averaged estimates of mule deer fetal survival from early March until birth (late May-June) in high and low energy development study areas of the Piceance Basin. no11hwest Colorado. 2012- 2014 (from Peterson et al. 2017: http://www.bioone.org/doi/pdf/l 0.298 I/wlb.00341 ). 7 �LITERATURE CITED Anderson, C. R., Jr.20 19. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation effo11s to address human activity and habitat degradation. Federal Aid in Wildlife Restoration Annual Repo11 W-243-R3. Ft. Collins. CO USA. Anderson. C.R.. Jr., and C. J. Bishop. 2014. Migration patterns of adult female mule deer in response to energy development. Pages 47-50 in Transactions of the 79•h North American Wildlife & Natural Resources Conference (R. A. Coon & M. C. Dunfee. eds.). Wildlife Management Institute, Gardners, PA, USA. ISSN 0078-1355. Bergman, E. J., C.R. Anderson Jr .. C. J. Bishop. A. A. Holland. and J.M. Northrup. 1018. Variation in ungulate body fat: individual versus temporal effects. Journal of Wildlife Management 82:130-137. DOI : 10:1002/jwmg.21334 Bishop. C. J.. C. R. Anderson Jr.. D. P. Walsh. E. J. Bergman. P. Kuechle. and J. Roth.20 11. Effectiveness of a redesigned vaginal implant transmitter in mule deer. Journal of Wildlife Management 75(8): 1797-1806; DOI : I0.1002/jwmg.229 Bishop. C. J., M. W. Alldredge, D. P. Walsh, E. J. Bergman. C.R. Anderson Jr.. D. Kilpatrick. J. Bake!, and C. Fabvre.2019. A noninvasive automated device for remotely collaring and weighing mule deer. Wildlife Society Bulletin 43:7 17-725; doi.org/ I0.1002/wsb. l 034 Freeman. E. D., R. T. Larsen, M. E. Peterson. C. R. Anderson. Jr.. K. R. Hersey. and B. R. McMillan. 20 14. Effects of male-biased harvest on mule deer: implications for rates of pregnancy, synchrony, and timing ofpa11urition. Wildlife Society Bulletin; DOI: 10.1002/wsb.450 Lendrum. P. E.. C. R. Anderson, Jr.. R. A. Long, J. K. Kie. and R. T. Bowyer. 20 12. Habitat selection by mule deer during migration: effects of landscape structure and natural gas development. Ecosphere 3(9):82. http://dx.doi.org/ I0. Lendrum, P. E.. C.R. Anderson. Jr .. K. L. Monteith. J. A. Jenks. and R. T. Bowyer. 2013. Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes. PLoS ONE 8(5): e64548. doi: I 0.1371 /joumal.pone.0064548 Lendrum. P. E.. C.R. Anderson. Jr.. K. L. Monteith. J. A. Jenks. and R. T. Bowyer. 20 14. Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns of forage quality. Mammalian Biology: http:, dx.doi.om ' J0. 10 16 j.mambio.20 1-tOS.005 Lynch. E., J. M. Northrup. M. F. McKenna. C. R. Anderson Jr.. L. Angeloni, and G. Wittemyer. 20 14. Landscape and anthropogenic features influence the use of auditory vigilance by mule deer. Behavioral Ecology; doi: I 0.1093/beheco/aru 158. Northrup. J. M., M. B. Hooten, C. R. Anderson. Jr.. and G. Wittemyer. 2013 . Practical guidance on characterizing availability in resource selection functions under a use-availability design. Ecology 94(7): 1456- 1463. Northrup, J.M .. C.R. Anderson, Jr.. and G. Wittemyer. 2014a. Effects of helicopter capture and handling on movement behavior of mule deer. Journal of Wildlife Management 78(4):73 1738; DOI: 10.1002/jwmg.705 Northrup, J.M .. A. B. Shafer, C.R. Anderson Jr.. D. W. Coltman. and G. Whittemyer. 20 14b. Finescale genetic correlates to condition and migration in a wild cervid. Evolutionary Applications ISSN 1752-457 1; doi: I0.1 I 11 /eva. 11 I89 Northrup, J.M .. C.R. Anderson. Jr.. and G. Wittemyer.2015. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer. Global Change Biology. doi: IO.I I I I/gcb. 13037. Northrup. J.M .. C.R. Anderson. Jr .. M. B. Hooten. and G. Wittemyer. 2016a. Movement reveals scale dependence in habitat selection ofa large ungulate. Ecological Applications 26:27462757 8 �Northrup. J.M .. C. R. Anderson. Jr.. and G. Winemyer. 2016b. Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid. Diversity and Distributions 22: 547-557. DOI : I0.11 11 /ddi. I24 17 Northrup. J.M., A. Avrin. C. R. Anderson Jr.. E. Brown. and G. Winemyer. 201 9. On-animal acoustic monitoring provides insight to ungulate foraging behavior. Journal of Mammalogy I00: 14791489: https://doi .ore./ 10. 1093/jmamma]/gyz 12-1 Northrup. J. M.. C. R. Anderson Jr.. B. D. Gerber. and G. Wittemyer. 202 1. Behavioral and demographic responses of mule deer to energy development on winter range. Wildlife Monographs 208: 1-37; 202 1: DOI : 10.1002/wmon.1060 Peterson. M. E.. C.R. Anderson Jr.. J.M. Northrup. and P. F. Doherty Jr. 2017 . Reproductive success of mule deer in a natural gas development area. Wildli fe Biology doi : 10.111 l/wlb.0034 1 Peterson. M. E.. C. R. Anderson Jr.. J. M. Northrup. and P. F. Doherty Jr. 20 18a. Mortality of mule deer fawns in a natural gas development area. Journal of Wildlife Management 82: 11 35-11 48. DOI: 10.1002/jwmg.2 1476 Peterson. M. E.. C. R. Anderson Jr.. M. W. Alldredge. and P. F. Doherty Jr. 20 18b. Using maternal mule deer movements to estimate timing of pa1turition and assist fawn captures. Wildlife Society Bulletin 42:6 16-62 1: DOI: I0.1002/wsb.935 Searle. K. R., M. B. Rice. C.R. Anderson. C. Bishop and N. T. Hobbs. 20 15. Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore. Oecologia ISSN 0029- 8549: DOI I0.1007/s00442-0 15-3348-9 Stonehouse. K. F.. C. R. Anderson Jr.. M. E. Peterson. and D. R.Collins. 20 16. Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus he111io1111s). Colorado Parks and Wildlife Technical Repo11 No. 48. First Edition. 317 W. Prospect Rd.. Ft. Collins. CO USA. DOW-R-T-48-16. ISSN 0084-8883. Prepared by_ _ _ _ __ __ _ __ _ __ _ _ _ __ Charles R. Anderson. Jr.. Mammals Research Leader 9 �Appendix A. Abstracts of published manuscripts resulting from Piceance Basin mule deer/energy development interaction research colla borations. Abstract form at specific to the respective journal requirements. Effectiveness of a redesigned vaginal implant transmitter in mule deer CHADJ. BISII OP1, CIIARLES R. ANDERSON .Jr. ', DANIEL P. \\'.-\LSll 1, ERIC ,I. II ERGI\IAN 1, PETER h: l iECIII.E', and .JOH N ROTH' 'Colorado Parks and Wildlili:. Fon Collins. Colorado 80526 USA 'Advanced Tdcmwy Systems. Isanti. Minnesota 550~0 US/\ Citation: Bishop. C. J . C R Anderson Jr.. D. P. Walsh. E J. Bergman. P. Kuechle. and J. Roth. 20 1 I. Effecti,·cness ofa redesigned vaginal implant transmincr in mule deer. Journal of Wildlifi: Managenwn 75(8 ): 1797-1806. DOI. IO I00'.:!/j11mg.229 ABSTRACT Our understanding of factors that limit mule deer (Odocoilrms he111io1111s) populations may be improved by evaluating neonatal survival as a function of dam characteristics under free-ranging conditions. which generally requires that both neonates and dams are radiocollared. The most viable technique facilitating capture of neonates from radiocoll arcd adult females is use of vaginal implant transmitters (VITs). To date. VITs have allowed research opponunitics that were not previously possible: however. vrrs arc ollcn expelled from adult lcmalcs prcpanum. which limits their cffcctivc.;ncss. We rcdcsigned an existing VIT manufactured by Advanced Telemetry Systems (ATS: Isanti. MN) by leng1hcning and widening wings used Lo retain the VIT in an adult female. Our objective was to increase VIT retention rates and lhercby increase the likelihood of locating birth sites and newborn fawns. We placed the newly designed VITs in 59 adult female mule deer and evaluated the probabi lity of re1ention to parturition and the probability of detecting ncwbom fawns. We also developed an equation for detern1ining VIT sample size necessary to achieve a specified sample size of neonates. The probabil ity ofa VIT being retained until parturition was 0.766 (SE = 0.0605) and the probability ofa VIT being retained to within 3 days of parturition was 0.894 (SE = 0.044 1). In a similar study using the original VIT wings (Bishop ct al. 2007). the probabi lity of a VIT being retained unti l parturition was 0.447 (SE = 0.0468) and the probabil ity of retention to within 3 days or parturition was 0.623 (SI: = 0.0456). Thus. our design modification increased VIT retention to pai1urition by 0.3 19 (SE = 0.0765) and VIT retention to within 3 days of parturition by 0.27 1 (SE = 0.0634). Considering dams that retained VITs to within 3 days of parturition. the probabil ity of detecting at least I neonate was 0.952 (SE= 0.033-l) and the probability of detecting both fawns from twin lillcrs was 0.588 (SE = 0.0827). We expended approximately 12 person-hours per detected neonate. As a guide for researchers planning future stud ies. we found that VIT sample size should approximately equal the targeted neonate sample size. Our study expands oppo1tunities for conducting research that links adult female attributes to producth·ity and olTspring survival in mule deer. t 20 1-t The Wildlife Society. Habitat selection by mule deer during migration: effects of landscape structure and natural-gas development PATRICh: E. LENDRlli\1 1, CHA RLES R. ANDERSON .m.'. RYAN A. LONG'. ,JOII N G. h:IE1, .\ND R. TERRY BOWYER' Depanment of Biological Sciences. Idaho State University. l'ocoldlo. Idaho 83209 tJS/\ 'Colorado Parks and Wildlife. Grand Junction. Colorado 81505 USA 1 Citation: Lendrum. I' E.. C R. Anderson Jr.. R. /I Long. J G Kie. and R. T Bo11)cr 2012 1labilal sd cction by mule deer during m1gra11on. ellccts of landscape slnicturc and natural-gas development. Ecosphcrc 3( 9 ):82 hnr 1J, Jn, ore, Ill I N9ll'ES Ic-lllll r,5 I A bstract. The disruption of trad itional migratory routes by anthropogenic disturbances has shifted pallcms of resource selection by many species. and in some instances has caused populalions to decline. Moreover. in recent decades popu lations of mule deer (Odocoileus he111io1111s) have declined throughout much of their historic range in the western United Slates. We used resourceselection functions to detem1ine if the presence of natural-gas development altered pauems of resource selection by migrating mule deer. We compared spring migration routes of adult female mule deer fitted with GPS collars (11 = 167) among four study areas that had varying degrees of natural-gas development from 2008 to 20 IO in the Piceance Basin of n011hwcs1 Colorado. USA. Mule deer migrating through the most developed area had longer step lengths (straight-line distance between successive GPS locations) compared with deer in less developed areas. Additionally. deer migrating through the most developed study areas tended to select for habitat types that provided greater amounts of concealment cover. whereas deer from the least developed areas tended to select habitats that increased access to forage and cover. Deer selected habitats closer to \\•ell pad, and avoided roads in all instances except along the most highly developed migratory routes. where road densities may have been too high for deer to avoid roads without deviating substantial ly from established migration routes. These results indicate that behavioral 1endencies toward avoidance of anthropogenic disturbance can be ovc1,-iddcn during migration by the strong lidelity ungulates demonstrate towards migration routes. If avoidance is feas ible. then d1:cr may select areas further from development. whereas in highly developed areas. deer may simply increase their rate of travel along establ ished migration routes. 10 �Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes Patrick E. Lendrum•, Charles R. Anderson Jr. 2, Ke,·in L. Monteith 1.J, Jonathan A. Jenb~, R. Terry Bowyer' 1 Department of Biological Sciences. Idaho State University. Pocatello. Idaho. USA.; Colorado Division of Parks and Wildlife, Grand Junction. Colorado. USA.~ Wyoming Cooperative Fish and Wildlife Research Unit University of Wyoming. Laramie. Wyoming. USA.~ Deparunent of Natural Resource Management. South Dakota State University. Brookings. South Dakota. USA Citation: Lendrum. P. E.. C.R. Anderson Jr.. K. L. Monteith. J. A. Jenks. R. T. Bowyer. 2013. Migrating Mule Deer: Effects of anthropogenically Altered Landscapes. PLoS ONE 8(5): c64548. DOI: I0.1371/joumal.pone.0<164548 Abstract Background: Migration is an adaptive strategy that enables animals to enhance resource availability and reduce risk of predation at a broad geographic scale. Ungulate migrations generally occur along traditional routes. many of which have been disrupted by anthropogenic disturbances. Spring migration in ungulates is of particular importance for conservation planning. because it is closely coupled with timing of parturition. The degree to which oil and gas development affects migratory patterns. and whether ungulate migration is sufficiently plastic to compensate for such changes. warrants additional study to better understand this critical conservation issue. Met/1odolog)'/Principal Fi11dings: We studied timing and S)11Chrony of departure from winter range and arrival to summer range offemale mule deer (Odocoileus /remiom,s) in northwestern Colorado. USA. which has one of the largest natural-gas reserves currently under development in North America. We hypothesized that in addition to local weather. plant phcnology. and individual life-history characteristics. patterns of spring migration would be modified by disturbances associated with natural-gas extraction. We captured 205 adult female mule deer. equipped them with GPS collars. and observed patterns of spring migration during 2008--20 I0. Condusions/Sig11ijicance: Timing of spring migration was related to winter weather (particularly snow depth) and access to emerging vegetation. which varied among years. but was highly synchronous across study areas within years. Additionally. timing of migration was influenced by the collective effects of anthropogenic disturbance. rate of travel. distance traveled. and body condition of adult females. Rates of travel were more rapid over shorter migration distances in areas of high natural-gas development resulting in the delayed departure. but early arrival for females migrating in areas with high development compared with less-developed areas. Such shifts in behavior could have consequences for timing of arrival on birthing areas. especially where mule deer migrate over longer distances or for greater durations. Practical guidance on characterizing availability in resource selection functions under a use-availability design 1 JOSEPH M. NORTHRllP , ME\'IN 8. HOOTEN'"• CHARI.ES R. ANDERSON JR.~. AND GEORGE WITTEMYER 1 1 Depanment of Fish. Wildlife. and Conservation Biology. Colorado State University. 1474 Campus Delivery. Fort Collins. Colorado 80523 USA :u.s. Geological Survey. Colorado Cooperative Fish and Wildlife Research Unit 1474 Campus Delivery. Fon Collins. Colorado 80523 USA ·'Colorado State University. Department of Statistics. Colorado State University. 1474 Campus Delivery. Fon Collins. Colorado 80523 USA ~Mammals Research Section Colorado Parks and Wildlife. 711 Independent Avenue. Grand Junction. Colorado 81505 USA Citation: Northrup. J.M .. M. B. Hooten. C.R. Anderson Jr.. and G. Wittemyer. :?013. Practical guidance on characterizing availability in resource selection functions under a use-availability design. Ecology 94(7 ): 1456-1463. hnp://dx.doi.org/10.1890/12-1688.1 Abstract. Habitat selection is a fundamental aspect of animal ecology. the understanding of which is critical to management and conservation. Global positioning system data from animals allow fine-scale assessments of habitat selection and typically are analyzed in a use-availability framework. whereby animal locations are contrasted with random locations (the availability sample). Although most use-availability methods are in foct spatial point process models. they often are fit using logistic regression. This framework offers numerous methodological challenges. for which the literature provides little guidance. Specifically. the size and spatial extent of the availability sample influences coefficient estimates potentially causing interpretational bias. We examined the influence of availability on statistical inference through simulations and analysis of serially correlated mule deer GPS data. Bias in estimates arose from incorrectly assessing and sampling the spatial extent of availability. Spatial autocorrelation in covariates. which is common for landscape characteristics. exacerbated the error in availability sampling leading to increased bias. These results have strong implications for habitat selection analyses using GPS data. which are increasingly prevalent in the literature. We recommend that researchers assess the sensitivity of their results to their availability sample and. where bias is likely. take care with interpretations and use cross validation to assess robustness. 11 �Effects of Helicopter Capture and Handling on Movement Behavior of Mule Deer JOSEPH l\l. NORTHRUP', CHARLES R. A;-.OERSON JR'. .-\~O GEORGE \\'ITTEl\lYER1 'Department of Fish. Wildlife. and Conservation Biology. Colorado State Universir:,. 1-17-l Campus Delivery. Fort Collins. Colorado 80523 USA 'Mammals Research S<'Clion Colorado Parks and Wildlife. 711 Independent Avenue. Grand Junction. Colorado 81505 USA Citation: Northrup. J. M.. C. R. Anderson Jr.. and G. Wittemyer. 201-l Effects ofhdicopter capture and handling on mowment behavior of mule deer. Journal of Wildlife Management 78(4):731 -738: DOI: 10. IOO:!/J\\1ng.705 ABSTRACT Research on wildlife movement. physiology. and reproductive biology often requires capture and handling of animals. Such invasive treatment can alter behavior. which may bias results or invalidate assumptions regarding representative behaviors. To assess the impacts of handling on mule deer (Odocoile11s he111io1111s). a focal species for research in North America. we investigated pre- and post-recapture movements of collared individuals. and compared them to deer that were not recaptured (controls). We compared pre- and post-recapture movement rates (1n/hr) and 24-hour straight-line displacement among recaptured and control deer. In addition. we examined the time it took recaptured deer to return to their pre-recapture home range. Both daily straight-line displacement and movement rate were marginally elevated relative to monthly averages for 24 hours following. recapture. with non-significant elevation continuing for up to 7 days. Comparing movements averaged over 30 days before and after recapture. we found no differences in displacement. but movement rates demonstrated seasonal effects. with faster movements post- relative to pre-recapture in March and slower movements post- relative to pre-recapture in December. Relative to control deer movements. recaptured deer movement rates in March were higher immediately after recapture and lower in the second and third weeks fo llowing recapture. The median time lo return to the pre-recapture home range was 13 hours. with 71 % of deer returning in the first day. and 91 % returning " ~thin 4 days. These results indicate a shon period of elevated movements following recaptures. likely due to the deer returning to their home ranges. followed by weaker but non-significant depression of movements for up to 3 weeks. Censoring of the first day o r data post capture from analyses is strongly supponed. and removing additional days until the individual returns to its home range will control for the majority of impacts from capture. © 20 14 The Wildlife Society. Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns of forage quality Patrick E. Lendrum', Charles R. Anderson Jr. 1' . h:cvin L. :\lonteith', Jonallrnn .-\. ,Jenks'', R. Terry Bowyer-" ' Department ofBiolog,cal Sciences. Idaho State University. 9::? I South 8th Avenue. Stop 8007. Pocotdlo 83209. USA 1' Mammals Research Section Colorado Parks and Wildlik 711 Independent Avenue. Grand Junction 81505. USA ' Wyoming Cooperative Fish and Wildlife Research Unit Department of Zoology and Physiology. Universny of Wyoming. 3166. I000 East University Avenue. Laramie 82071. USA J Department of Natural Resource Management. South Dakota State University. Box ::?1-108. Brookings 57007. USA Citation: Lendrum. I'. E.. C. R. Anderson Jr.. KL. Monteith. J. A. Jenks. and R. T. Bo\\yer. 201-1. Relating the movement ofa rapidly migrating ungulate 10 spatiotemporal patterns of forage quality. Mammalian Biology: hnn 1/Jx dn, unµ'I 11 IOI6 11nrnmhsn 101-l 05 OIJ:\ ABSTRACT: Migratory ungulates exhibit recurring movements. o ften along trad itional routes between seasonal ranges each spring and autumn. which allow them lo track resources as they become avaiIable on the landscape. We examined the relationship between spring migration of mule deer (Odocoileus he111io1111s) and forage quality. as indexed by spatiotemporal pauerns of fecal nitrogen and remotely sensed greenness of vegetation (Nom1al izcd Difference Vegetation Index: NOV I) in spring 2010 in the Piceance Basin ofnorthwestem Colorado. USA. NDVI increased throughout spring. and was affected primarily by snow depth when snow was present. and temperature when snow was absent. Fecal nitrogen was lowest when deer were on winter range before migration. increased rapidly to an asymptote during migration. and remained relatively high when deer reached summer range. Values of fecal nitrogen corresponded with increasing NDV I during migration. Spring migration for mule deer provided a way for these large mammals to increase access lo a high-qual ity diet. which was evident in pauerns of NOVI and fecal nitrogen. Moreover. these deer ·jumped.. rather than ..surfed.. the green wave by arriving on summer range well before peak productivity of forage occurred. This rapid migration may aid in securing resources and seclusion from others on summer range in preparation for panurition. and to minimize detrimental factors such as predation and malnutrition during migration. 12 �Effects of Male-Biased Harvest on Mule Deer: Implications for Rates of Pregnancy, Synchrony, and Timing of Parturition ERIC D. FREEMAN'. RAND\' T. LARSEN 1, MARKE. PETF.RSON2, CHARLES R. ANDERSON JR.J. KENT R. HERSEY\ AND BROCK R. Mcl\llLLAN 1 1 Depanment of Plant and Wildlife Sciences. Brigham Young University. 275 WIDB. Provo. UT 84602. USA : Depanment of Fish. Wildlife. and Conservation Biology. Colorado State University. 1474 Campus Delivery. Fon Collins. CO 80523, USA ·' Colorado Parks and Wildlife. 711 Independent Avenue. Grand Junction. CO 81 SOS. USA ~ Utah Division of Wildlife Resources. 1594 W Nonh Temple. Salt Lake City. UT 84114. USA Citation: Freeman. E. D.. R. T. Larsen. M. E. Peterson. C.R. Anderson Jr.. K. R. Hersey. and B. R. McMillan. 2014. Effects of male-biased harvest on mule deer: implications for rat~-s of pregnancy. synchrony. and timing of panurition. Wildlife Society Bulletin: DOI: 10.1002/wsb.450 ABSTRACT Evaluating how management practices influence the population dynamics of ungulates may enhance future management of these species. For example. in mule deer (Odocoileus hemi01ms). changes in male/female ratio due to malebiased harvest may alter rates of pregnancy. timing of parturition. and synchrony of parturition if inadequate numbers of males are present to fertilize females during their first estrous cycle. If rates of pregnancy or parturition are influenced by decreased male/female ratios. recruitment may be reduced (e.g .. fewer births. later parturition resulting in lower survival of favms. and a less S)nchronous parturition that potentially increases susceptibility of neonates to predation). Our objectives were to compare rates of pregnancy. synchrony of parturition. and timing of parturition between exploited mule deer populations with a relatively high (Piceance. CO. USA: 26 males/JOO females) and a relatively low (Monroe. UT. USA: 14 males/JOO females) male/female ratio. We determined rates of pregnancy via ultrasonography and timing of parturition via vaginal implant transmitters. We found no differences in rates of pregnancy (98.6% and 96.6%: z = 0.821: P = 0.794). timing of parturition (estimate= 1.258: SE= 1.672: t = 0. 752: P = 0.454 ). or S)11chrony of parturition (F = 1.073: P = 0.859) between Monroe Mountain and Piceance Basin. respectively. The relatively low male/female ratio on Monroe Mountain was not associated with a protracted period of parturition. This finding suggests that relatively low male/female ratios typical of heavily harvested populations do not influence population d)11amics because recruitment remains unaffected. C 2014 The Wildlife Society. Fine-scale genetic correlates to condition and migration in a wild cervid Joseph M. Northrup', Aaron B. A. Shafer, Charles R. Anderson Jr.J, Da,·id W. Coltman~. and George \\'ittemyer1 I Department offish. Wildlife. and Conservation Biology. Colorado State University. Fon Collins. CO. USA 2 Department of Evolutionary Biology. Evolutionary Biology Centre. Uppsala University. Uppsala. Sweden 3 Mammals Research Section. Colorado Parks and Wildlife. Grand Jun~1ion. CO. USA 4 Depanment of Biological Sciences. University of Alberta. Edmonton. AB. Canada. Citation: Nonhrup. J.M .. A. B. Shafer. C.R. Anderson Jr.. D. W. Coltman. and G. Whittemyer. 2014. Fine-scale genetic correlates to condition and migration in a \\ild cervid. Evolutionary Applications ISSN 1752-4571: doi: IO. I I I l/eva.12189 Abstract The relationship between genetic variation and phcnotypic traits is fundamental to the study and management of natural populations. Such relationships often arc investigated by assessing correlations between phenotypic traits and heterozygosity or genetic differentiation. Using an extensive data set compiled from free ranging mule deer (Odocoileus hemionus). we combined genetic and ecological data to (i) examine correlations between genetic differentiation and migration timing. (ii) screen for mitochondrial haplotypes associated ,,ith migration timing. and (iii) test whether nuclear heterozygosity was associated with condition. Migration was related to genetic differentiation (more closely related individuals migrated closer in time) and mitochondrial haplogroup. Body fat was related to heterozygosity at two nuclear loci (with antagonistic patterns). one of which is situated near a knO\m fat metabolism gene in mammals. Despite being focused on a widespread panmictic species. these findings revealed a link between genetic variation and important phenotypes at a fine scale. We hypothesize that these correlations are either the result of mixing refugial lineages or differential mitochondrial haplotypes influencing energetics. The maintenance of phenotypic diversity will be critical to enable the potential tracking of changing climatic conditions. and these correlates highlight the need to consider evolutionary mechanisms in management. even in widely distributed panmictic species. 13 �Landscape and anthropogenic features influence the use of auditory vigilance by mule deer Emma Lynch•, Joseph M. Northrupb, Megan F. McKennaC. Charles R. Anderson Jr.", Lisa Angeloni~, and George Wittemye~• "Graduate Degree Program in Ecology. Colorado State University. 1474 Campus Delivery. Fon Collins. CO 80513. USA 1lepanment offish, Wildlife and Conservation Biology. Colorado State University. 1474 Campus Delivery. Fon Collins. CO 80523. USA "Natural Sounds and Night Skies Division, National Park Service. 1201 Oakridge Drive. Fon Collins. CO 80525. USA. JMammals Research Section, Colorado Parks and Wildlife. 317 W. Prospect Road. Fon Coll ins. CO 80526, USA cDepartment of Biology, Colorado State University, 1878 Campus Delivery. Fon Collins. CO 80513. USA Citation: Lynch, E., J.M. Nonhrup, M. F. McKenna. C.R. Anderson Jr.. L. Angeloni. and G. Winemyer. 2014. Landscape and anthropogenic features influence the use of auditory vigilance by mule deer. Behavioral Ecology: doi: I0.1093/beheco/aru 158. While visual fonns of vigilance behavior and their relationship with predation risk have been broadly examined. animals also employ other vigilance modalities such as auditory vigilance by listening for the acoustic cues of predators. Similar to the tradeoffs associated with visual vigilance. auditory behavior potentially structures the energy budgets and behavior of animals. The cryptic nature of auditory vigilance makes it difficult to study. but on-animal acoustical monitoring has rapidly advanced our ability to investigate behaviors and conditions related to sound. We utilized this technique to investigate the ways external stimuli in an active natural gas development field affect periodic pausing by mule deer (Odocoi/eus hemionus) within bouts of rumination-based maslication. To better understand the ecological properties that structure this behavior. we investigate spatial and temporal factors related to these pauses to detennine ifresults are consistent with our hypothesis that pausing is used for auditory vigilance. We found that deer paused more when in forested cover and at night where visual vigilance was likely to be less effective. Additionally. deer paused more in areas of moderate background sound levels. though responses to anthropogenic features were less clear. Our results suggest that pauses during rumination represent a fonn of auditory vigilance that is responsive to landscape variables. Further e~-ploration of this behavior can facilitate a more holistic understanding of risk perception and the costs associated with vigilance behavior. Migration Patterns of Adult Female Mule Deer in Response to Energy Development V Charles R. Anderson Jr. and Chad J. Bishop Mammals Research Section, Colorado Parks and Wildlife. 317 W. Prospect Road. Fon Collins. CO 80526. USA Citation: Anderson, C.R .. Jr.. and C. J. Bishop. 2014. Migration panems of adult female mule deer in response to energy development. Pages 47.50 in Transactions of the 79lh North American Wildlife & Natural Resources Conference (R. A. Coon & M. C. Dunfee. eds.). Wildlife Manae.ement Institute. Gardners. PA. USA. ISSN 0078•1355. - Migration is an adaptive strategy that enables animals to enhance resource availability and reduce risk of predation at a broad geographic scale. Ungulate migrations generally occur along traditional routes. many of which have been disrupted by anthropogenic disturbances. Spring migration in ungulates is of particular importance for conservation planning because it is closely coupled with timing of parturition. The degree to which oil and gas development affects migratory patterns. and whether ungulate migration is sufficiently prepared to compensate for such changes. has recently been investigated in Colorado and Wyoming (Lendrum et al. 2012. 2013: Sawyer et al. 2012). Lendrum et al. (2012. 2013) and Sawyer et al. (2012) address mule deer (Odoc:oi/eus hemionus) migration patterns in relation to energy development from northwest Colorado and south-central Wyoming. respectively. We address results from the Colorado and Wyoming studies and then compare similarities and differences. The interactions between migratory mule deer and energy development identified by Lendrum et al. (2012. 2013) and Sawyer et al. (2012) suggest mule deer may benefit from energy development planning by considering thresholds of development that may alter migratory behavior. It appears that migration rate. migration routes. and stopover use. if present may be altered at high development intensities. In addition. migratory mule deer may benefit by maintaining security cover along migration paths. and improved habitat conditions may facilitate more direct and rapid migration requiring less energy to complete migration. Enhancing penneability along migration routes by applying dispersed development plans (<2 well pads/km:?) and minimizing disturbance to vegetation types by maintaining security cover should reduce impacts to migratory mule deer as well as other migratory ungulates. Where feasible. habitat improvement projects on winter range and possibly stopover sites would also enhance migratory mule deer populations by enhancing energy reserves for long-distance movements and parturition shortly after summer range arrival. Where possible. directional drilling could be used to extract energy resources from underneath migration routes while maintaining no surface occupancy. Lastly. we emphasize that GPS studies now allow managers to accurately map migration routes for entire populations and identify relatively narrow con·idors that arc most heavily used thus allowing for the identification of the most important corridors for migrating ungulates. Where available. we encourage agencies to incorporate such migration corridors into land-use plans (e.g .. resource management plans) and National Environmental Policy Act documents. 14 �Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore Kate R. Searle' • Mindy B. Rice 2 • Charles R. Anderson 2 • Chad Rishop 2 • N. T. HobbsJ 1 NERC Centre for Ecology and Hydrology. Bush Estate. Penicuik El 126 0QB. UK ~ Colorado Parks and Wildlife. 317 W. Prospect Road. Fort Collins. CO 80526. USA ~ Department of Ecosystem Science and Sustainability. Colorado State University. Fon Collins 80524. CO. USA Citation: Searle. K. R.. M. B. Rice. C. R. Anderson. C. Bishop and N. T. Hobbs. 2015. Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore. Occologia ISSN 0029-8549: DOI I 0.1007/s00442-015-3348-9 Abstract Understanding how spatial and temporal heterogeneity influence ecological processes fonns a central challenge in ecology. Individual responses to heterogeneity shape population dynamics. therefore understanding these responses is central to sustainable population management. Emerging evidence has sho\\11 that herbivores track heterogeneity in nutritional quality of vegetation by responding to phenological differences in plants. We quantified the benefits mule deer (Odocoileus hemionus) accrue from accessing habitats with asynchronous plant phcnology in northwest Colorado over 3 years. Our analysis examined both the direct physiological and indirect environmental effects of weather and vegetation phenology on mule deer winter body condition. We identified several important effects of annual weather patterns m1d topographical variables on vegetation phenology in the home ranges of mule deer. Crucially. temporal patterns of vegetation phenology were linked with differences in body condition. ,,ith deer tending to show poorer body condition in areas ,-.ith less asynchronous vegetation green-up and later vegetation onset The direct physiological effect of previous winter precipitation on mule deer body condition was much less important than the indirect effect mediated by vegetation phenology. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer JOSEPH 1\1. NORTHRUP'. CHARLES R. ANDERSON JR. 2• and GEORGE WIITE!\IYER1.J 'Department of Fish. Wildlife and Conservation Biology. Colorado State University. Fon Collins. CO. USA ~Mammals Research Section. Colorado Parks and Wildlife. Fort Collins. CO. USA )Graduate Degree Program in Ecology. Colorado State University. Fon Collins. CO. USA Citation: Nonhrup. J.M .. C. R. Anderson. Jr.. and G. Wittcmyer. 2015. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer. Global Change Biology. doi: IO. I I I l/gcb.13037 Abstract Extraction ofoil and natural gas (hydrocarbons) from shale is increasing rapidly in North America. with documented impacts to native species and ecosystems. With shale oil and gas resources on nearly every continent this development is set to become a major driver of global land-use change. It is increasingly critical to quantil)' spatial habitat loss driven by this development to implement effective mitigation strategies and develop habitat oflscts. Habitat selection is a fundamental ecological process. influencing both individual fitness and population-level distribution on the landscape. Examinations of habitat selection provide a natural means for understanding spatial impacts. We examined the impact of natural gas development on habitat selection patterns of mule deer on their ,\inter range in Colorado. We fit resource selection functions in a Bayesian hierarchical framework. with habitat availability defined using a movement-based modeling approach. Energy development drove considerable alterations to deer habitat selection patterns. with the most substantial impacts manifested as avoidance of well pads with active drilling to a distance of at least 800 m. Deer displayed more nuanced responses to other infrastructure. avoiding pads with active production and roads to a greater degree during the day than night. In aggregate. these responses equate to alteration of behavior by human development in over 50% of the critical winter range in our study area during the day and over 25% at night. Compared to other regions. the topographic and vegetative diversity in the study area appear to provide refugia that allow deer to behaviorally mediate some of the impacts of development. This study. and the methods we employed. provides a template for quantifying spatial take by industrial activities in natural areas and the results offer guidance for policy makers. mangers. and industry when attempting to mitigate habitat loss due to energy development. 15 �Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid Joseph M. Northrup•, Charles R. Anderson Jr. and George Wittemyer 1.., 1Department of Fish. Wildlife and Conservation Biology, Colorado State University. Fon Collins. CO. USA ~Mammals Research Section. Colorado Parks and Wildlife, Fon Collins. CO, USA JGraduate Degree Program in Ecology. Colorado State University. Fon Collins. CO. USA Citation: Nonhrup, J. M.. C. R. Anderson. Jr., and G. Winemycr. .::?0 16. Environmental dynamics and anthropogenic development alter philopatl)' and space-use in a Nonh American cervid. Diversity and Db1ributions .:?.:?: 547-557. DOI: I0. I l I l/ddi. l.::?417 ABSTRACT Aim The space an animal uses over a given time period must provide the resources required for meeting energetic needs. reproducing and avoiding predation. Anthropogenic landscape change in concert with environmental dynamics can strongly structure space-use. Investigating these dynamics can provide critical insight into animal ecology. conservation and management. Location The Piceance Basin. Colorado. USA. Methods We applied a novel utilization distribution estimation technique based on a continuous-time correlated random walk model to characterize range dynamics of mule deer during winter and summer seasons across multiple years. This approach leverages secondorder properties of movement to provide a probabilistic estimate of space-use. We assessed the influence of environmental (cover and forage). individual and anthropogenic factors on interannual variation in range use of individual deer using a hierarchical Bayesian regression framework. Results Mule deer demonstrated remarkable spatial philopatry. with a median of 50% overlap (range: 8-78%) in year-to-year utilization disuibutions. Environmental conditions were the primary driver of both philopatry and range size. with anthropogenic disturbance playing a secondary role. Main conclusions Philopatry in mule deer is suspected to reflect the importance of spatial familiarity (memory) to this species and. therefore. factors driving spatial displacement are of conservation concern. The interaction between range behaviour and dynamics in development disturbance and environmental conditions highlights mechanisms by which anthropogenic environmental change may displace deer from familiar areas and alter their foraging and survival strategics. Movement reveals scale dependence in habitat selection of a large ungulate Joseph M. Northrup1, Charles R. Anderson Jr.:. Mnin B. Hooten". and George Wittemye~ 1 Depanment of Fish. Wildlife and Conservation Biology, Colorado State University. Fon Collins. Colorado 80523 USA :Mammals Research Section. Colorado Parks and Wildlife. Fon Collins. Colorado 805:!3 USA 3U.S. Geological Survey. Colorado Cooperative Fish and Wildlife Research Unit. Depanment of Fish. Wildlife and Conservation Biology. Colorado State University. Fon Collins. Colorado 80523 USA ~Depanment of Fish. Wildlife and Conservation Biology and Graduate Degree Program in Ecology. Colorado State University. Fon Collins. Colorado 80523 USA Citation: Nonhrup, J.M .. C.R. Anderson. Jr.. M. B. Hooten. and G. Wincmyer. 2016. Movement reveals scale dependence in habitat selection ofa large ungulate. Ecological Applications 26:2746-2757 Abstract. Ecological processes operate across temporal and spatial scales. Anthropogenic disturbances impact these processes. but examinations of scale dependence in impacts are infrequent. Such examinations can provide important insight to wildlife-human interactions and guide management efforts to reduce impacts. We assessed spatiotemporal scale dependence in habitat selection of mule deer (Odocoileus hemionus) in the Piceance Basin of Colorado. USA. an area of ongoing natural gas development We employed a newly developed animal movement method to assess habitat selection across scales defined using animal-centric spatiotemporal definitions ranging from the local (defined from five hour movements) to the broad (defined from weekly movementc;). We extended our analysis to examine variation in scale dependence between night and day and assess functional responses in habitat selection patterns relative to the density of anthropogenic features. Mule deer displayed scale invariance in the direction of their response to energy development features. avoiding well pads and the areas closest to roads at all scales. though with increasing strength of avoidance at coarser scales. Deer displayed scale-dependent responses to most other habitat features. including land cover type and habitat edges. Selection differed between night and day at the finest scales. but homogenized as scale increased. Deer displayed functional responses to development with deer inhabiting the least developed ranges more strongly avoiding development relative to those \\'1th more development in their ranges. Energy development was a primary driver of habitat selection patterns in mule deer. sb1lcturing their behaviors across all scales examined. Stronger avoidance at coarser scales suggests that deer behaviorally mediated their interaction with development. but only to a degree. At higher development densities than seen in this area. such mediation may not be possible and thus maintenance of sufficient habitat with lower development densities "ill be a critical best management practice as development expands globally. 16 �Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus) Kourtney F. Stonehouse•.2. Charles R. Anderson Jr. 1, Mark E. Peterson•.2, and David R. Collins' 1Mammals Research Section. Colorado Parks and Wildlife. 317 W. Prospect Road. Fort Collins. CO 90526 USA :Department offish. Wildlife and Conservation Biology. Colorado Stute University. Fort Collins. Colorado 80523 USA Citation: Stonehouse. K. F.. C.R. Anderson Jr.. M. E. Peterson. and D.R. Collins . .2016. Approaches to field investigations of cause-specific monality in mule deer (Odocoi/eus hemionus). Colorado Parks and Wildlife Technical Report No. 48. First Edition. 317 W. Prospect Rd .. Ft. Collins. CO USA. DOW-R-T-48-16. ISSN 0084-8883. This technical report provides general guidelines for conducting mortality site investigations to help investigators distinguish predation from scavenging and other causes of death. General health indices are also provided to assess whether or not deer may have died from malnutrition or disease or if these factors may have predisposed deer to predation. Lastly. these guidelines will assist investigators in identifying predatory species or scavengers involved through the examination of physical evidence at deer mortality sites. The infonnation presented here is based primarily on lield experience gained from a long term research effort in northwest Colorado investigating mule deer mortality sites over several years (http://cpw.state.co.us/leam/Pages/ResearchMammalsRP-04.aspx) and literature review where referenced. We acknowledge that proximate and ultimate cause of death can be difficult or impossible to detect from field necropsy alone and examples presented here largely represent proximate causes of mortality: efforts discerning ultimate cause will require specific tissue sample collections. where possible. submitted to a veterinary diagnostic laboratory. Within this technical report are numerous photographs documenting characteristics of predator attacks on mule deer and signs left by predatory and scavenging species. Additional pictures illustrate differences between healthy and unhealthy tissues and organs. While reading this documenL be aware that each mortality investigation is unique and observations in the field may differ from illustrations provided here. Appendix I provides a sample necropsy form to assist in conducting mortality investigations. Reproductive success of mule deer in a natural gas development area Mark E. Peterson 1, Charles R. Anderson Jr. 2, Joseph M. Northrup 1,and Paul F. Doherty Jr. 1 'Department of Fish. Wildlife and Conservation Biology. Colorado State University. Fort Collins. Colorado 80523 USA :Mammals Research Section. Colorado Parks and Wildlife. 317 W. Prospect Road. Fort Collins. CO 90526 USA Citation: Peterson. M. E.. C. R. Anderson Jr.. J.M. Northrup. and P. F. Doherty Jr. 2017. Reproductive success of mule deer in a natural gas development area. Wildlife Biology doi: IO. I I I I/wlb.00341 Abstract: Natural gas development is increasing across North America and causing concern over the potential impacts on wildlife populations and their habitat. particularly for ungulate species. Understanding how this development impacts reproductive success metrics that are influential for ungulate population dynamics is important to guide management of ungulates. However. the influences of natural gas development on reproductive success metrics of mule deer Odocoi/eus hemio11us have not been studied. We used statistical models to examine the influence of natural gas development and temporal factors on reproductive success metrics of mule deer in the Piceance Basin. northwest Colorado during 2012-2014. We focused on study areas with relatively high or low levels of natural gas development. Pregnancy and in utero fetal rates were high and statistically indistinguishable between study areas. Fetal sun•ival rates increased over time and survival was lower in the high versus low development study areas in 2012 possibly influenced by drought coupled with habitat loss and fragmentation associated with development. Our novel results suggest managers should be concerned with the influences of development on fetal survival. particularly during extreme environmental conditions (e.g. drought) and our results can be used to guide development planning and/or mitigation. Developers and wildlife managers should continue to collaborate on development planning. such as implementing habitat treatments to improve forage availability and quality. minimizing disturbance to hiding and foraging habitat particularly during parturition. and implementing directional drilling to minimize pad disturbance density to increase fetal survival in developed areas. 17 �Variation in ungulate body fat: individual versus temporal effects Eric J. Bergman•, Charles R. Anderson Jr. 1• Chad J. Bishop•, A, Andrew Holland 1, and Joseph 1\1, Northrup? 'Colorado Parks and Wildlife. 317 W. Prospect Road. Fon Collins. CO 90526 USA :Depanment of Fish. Wildlife and Conservation Biology. Colorado State University. Fon Collins. Colorado 805::?3 USA Citation: Bergman, E. J.. C. R. Anderson Jr.. C. J. Bishop. A. A. Holland. and J. M. Nonhrup. .'.!0 18. Variation in ungulate body fat: individual versus temporal effects. Journal of Wildlife Management 8.'.!: I30-137. DOI: IO: I00.'.!/j\\mg.::? 133.J ABSTRACT The use ofultrasonograhic measurements of muscle and body fat represent a relatively new data stream that can be used to address questions regarding ungulate condition. We have learned that measurements of body fat and presumably overall body condition among individual animals. even those taken from the same herd at that same time. are highly variable. Relatively little consideration has been given to the sources of variation in body fat and other physiological parameters in "•;Jdlife populations. We evaluated the components of variation in late-winter mule deer (Odocoileus hemiomts) body fat estimates: sampling variation (i.e .. variation induced by the particular set of individuals that were sampled) and process variation (i.e .. variation stemming from biological processes) with a long-tenn data set (2002-2015) from Colorado. USA. We collected our data from across Colorado as part of historical research. ongoing research. and periodic population monitoring programs. Mean percent ingesta-frce body fat (%1FBF) for sampled mule deer was 7.20 ± 1.20% (SD). Covariates related to individual deer explained approximately-'% of the total variation in %1FBF and annual effects explained an additional 13% of the variation. Substantial residual variation in %1FBF (83%) remained unex'J)lained. The source of the 83% of unexplained variation is partially linked to line-scale spatial dynamics but also additional individual metrics we were unable to capture, primarily the presence or absence of dependent young. We speculate that the primary factors influencing late-winter mule deer body fat and overall condition are individual in nature. These results present a cautionary check on herd level inference that can be made from individual late-winter body fat estimates and we postulate that for mule deer, alternative and additional body condition metrics may offer added utility in management scenarios. 1lowever. an important next step to better understand wildlife population health is to evaluate the sources and magnitude of variation within other body condition metrics. with the goal of further refining data that can better allow biologists to incorporate herd health into population management recommendations. Mortality of mule deer fawns in a natural gas development area Mark E. Peterson•, Charles R. Anderson Jr. 2, Joseph l\l. Northrup•, and Paul F. Doherty Jr. 1 'Department offish. Wildlife and Conservation Biology. Colorado State Uniwrsity. Fon Collins, Colorado 80523 USA ?Mammals Research Section. Colorado Parks and Wildlife. 317 W. Prospect Road. Fon Collins. CO 905.'.!6 USA Citation: Peterson. M. E.. C.R. Anderson Jr .. J.M. Nonhrup. and P. F. Doheny Jr. ::?018. Monality of mule deer fa\\TIS in a natural gas development area. Journal of Wildlife Management 82:1135-1148. DOI: 10. I00::?/j,,mg.21476 ABSTRACT Recent natural gas development has caused concern among wildlife managers. researchers. and stakeholders over the potential effects on wildlife and their habitats. Specifically. understanding how this development and other factors influence mule deer (Odocoileus hemi01ms) fawn (i.e .. 0-6 months old) mortality rates. recruitment and subsequently population dynamics have been identified as knowledge gaps. Thus. we tested predictions concerning the relationship between natural gas development adult female. fa,..11 birth. and temporal (weather) characteristics on fa, ..n mortality in the Piccance Basin of northwestern Colorado. USA. from 2012-2014.We captured and radio-collared 184 fawns and estimated apparent cause-specific mortality in areas with relatively high or low levels of natural gas development using a multi-state model. Mean daily predation probability was similar in the high versus low development areas. Predation was the leading cause offawn mortality in both areas and decreased from 0-14 days old. Black bear ( Ursus americam1s: 22% of all mortalities. 11 = 17) and cougar (Fe/is co11color: 36% of all mortalities. 11 = 6) predation was the leading cause of mortality in the high and low development areas. respectively. Predation of fawns was negatively correlated with the distance from a female"s core area to a producing well pad on winter or summer range. Contrary to expectations. predation of fawns was positively correlated with rump fat thickness of adult females. Well pad densities and development activity were relatively low during our study. indicating that the observed intensity of development did not appear to influence daily predation probability. Our results suggest maintaining development activity thresholds at levels we observed to potentially minimize the effects of development on fawn mortality. However. we caution that higher development intensity and drilling activity in flatter. less rugged areas with less concealment cover could influence fawn mortality. Managers should maintain low development densities in areas where topography and vegetation offer less concealment. Overall. region-specific data (e.g .. development intensity. topography. predator assemblages. and associated predation risk) arc needed to better understand the effects of natural gas development on fawn mortality. 18 �Using maternal mule deer movements to estimate timing of parturition and assist fawn captures Mark E. Peterson', Charles R. Anderson ,Jr.', Mathew W. Alldredge', and Paul F. Doherty ,I r.' 1Depamnent of Fish. Wildlife and Conservation Biology. Colorado State Uni\'crsily. Fon Collins. Colorado 80523 USA 'Mammals Research Section. Colorado Parks and Wildlife. 317 W. Prospect Road. Fon Coll ms. CO 90526 USA Citation: Peterson. M. E.. C R. Anderson Jr.. M. W. Alldredge. and P. F Doheny Jr 2018. Using maternal mule deer movements lo estimate liming of panurilion and assist fa\\T1 captures Wildlife Society Bullelin ➔2 : 616-6~1 : DOI: 10 1002/wsb,935 ABSTRACT Movement patterns of maternal ungulates have been used Lo determine parturition dates and aid in locating fawns. which may be important for understanding reproductive rates (e.g .. pregnancy and fetal). but such methods have not been validated for mule deer ( Odocoi/eus he111ia1111s). We first detem1ined timing of parturition using vaginal implant transmitters (VITs) and then predicted timing of parturition using VlTs in conjunction with Global Positioning System collar data in the Piceance Basin of northwestern Colorado. USA. during 20 12-20 1--1. We examined daily movement rate to determine differences in movement rate among days (7 days pre• and postpartum) and for movement patterns indicative of parturition, Mean daily movement rate (m/day) of I 02 maternal deer decreased by --16% from I day preparturition (mean = l.253. SD = 1.091) to parturition date (mean = 682. S = 574). and remained at this low rate 1-7 days postpartum. We applied an independent data set to validate predicted parturition dates based on daily movement rate. We estimated day of parturition cotTcctly (i.e .. day 0). with in 1-3 days postparturition. and _ 4 days postparrurition of field-n:ported dates for IO (29%). 2 1 (60%). and 4 ( 11%) maternal lcmalcs. respecti vely. For novel data s.:ts. w.: predict that a mule deer female whose dai ly movement rate decreases by _ 46% and remains low _3 days postparturition particularly when preceded by a sudden increase in movement-has given birth. However. we caution that disturbance of deer by field crews should be minimized. and if birth sites are not found. neonatal mortal ity will be underestimated. Our results can help determine timing and general location of parturition as an aid in capturing fawns when the use of VITs is not feasible. with the ultimate objective of estimating pregnancy. fetal. and fawn survival rates ifbirth sites arc found. On-animal acoustic monitoring provides insight to ungulate foraging behavior Joseph M. Northrup'. Alexandra Anin 1, Charles R. Anderson. Jr.'. Emma Browns, and Gcori:c Wittcmycr 1 1Depanmenl of Fish. Wildlife and Conservation Biolog,. Colorado Stale Unhersit~. Fon Collins. Colorado 80523 USA 1Mammals Research Section. Colorado Parks and Wildlife. 317 W. Prospect Road. Fon Collins. CO 905'.!6 USA ' National Park Service Natural Sounds and Night Skies Division. Fon Collins. CO 805~5 USA Citalion: Northrup. J. M.. A i\vnn. C. R. Anderson Jr.. E. Bro\\n. and G. Wiuem~er. '.!019. On-animal acoustic monitoring provides insight to ungulate foraging behavior. Journal of Mammalogy I00: I-l 79-1 ➔89: hnps //do, nrt!/IIJ.10'/3h111urnm:1II;;\/ 11-l Abstract Foraging behavior underpins many ecological processes: ho"·cvcr. rohust assessments of this behavior for free-ranging animals arc rare due to limitations to direct obse1vations. We leveraged acoustic mon itoring and GPS tracking to assess the factors influencing foragi ng behavior of mule deer (Odocoile us he111io1111s). We deployed custom-built acoustic collars with GPS radiocollars on mule deer to measure location-specific foraging. We quantified individual bites and steps taken by deer. and quantified two metrics of forag ing behavior: the number of bites taken p~r step and the number of bites taken per unit time. which relate to foragi ng intensity and efficiency. We fi t statistical models to these metrics to examine the individual. environmental. and anthropogenic factors influencing foraging. Deer in poorer body condition took more bites per step and per minute and foraged for longer irrespective of landscape properties. Other patterns varied seasonally with major changes in deer condition. In December. when deer were in better condition. they took fewer bites per step and more bites per minute. Deer also fo raged more intensely and efliciently in areas of greater forage availability and greater movement costs. During March. when deer were in poorer condition. foraging was not influenced by landscape features. Anthropogenic factors weakly su·uctured foraging behavior in December "ith no relationship in March. Most research on animal foraging is interpreled under the framework of optimal foraging theory. Departures from predictions developed under this framework provide insight to unrecogn ized factors influenc.ing the evolution of foraging. Our results on ly conformed to our pred ictions when deer were in better condition and ecological conditions were declini ng. suggesting foraging strategies were state-dependent. These results advance our understanding of foragi ng patterns in wild animals and highlight novel observational approaches for studying animal behavior. 19 �A noninvasive automated device for remotely collaring and weighing mule deer Chad J. Bishop•, Mathew W, Alldredge•, Daniel P. Walsh 1. Eric: J, Bergman 1. Charles R. Anderson Jr.'. Darlene Kilpatrick. Joe Bakel 1, and Christophe Fabvre1 1 Colorado Parks and Wildlife. 317 W. Prospect Road. Fon Collins. CO 80526 USA ~Dynamic Group Circuit Design. Inc .. 2629 Redwing Road, Fort Collins. CO 80525 USA Citation: Bishop, C. J., M. W. Alldredge, D. P. Walsh, E. J. Bergman, C.R. Anderson Jr .. D. Kilpatrick. J. Bakel. and C. Fabvrc. 2019. A noninvasive automated device for remotely collaring and weighing mule deer. Wildlife Society Bulletin 43:717-725: doi.org/10.1002/wsb. I034 ABSTRACT Wildlife biologists capture deer (Odocoi/eus spp.) annually to attach transmitters and collect basic information (e.g .. animal mass and sex) as part of ongoing research and monitoring activities. Traditional capture techniques induce stress in animals and can be expensive. inefficient and dangerous. They are also impra"-tical for some urbanized settings. We designed and evaluated a device for mule deer (0. hemionus) that automatically attached an expandable radiocollar to a 2::6-month-old fawn and recorded the fawn's mass and sex. without physically restraining the animal. The device did not require on-site human presence to operate. Students and faculty in the Mechanical Engineering Department at Colorado State University produced a conceptual model and early prototype. Professional engineers at D)11amic Group Circuit Design. Inc. in Fort Collins. Colorado. USA. produced a fully functional prototype of the device. Using the device. we remotely collared, weighed. and identified sex of 8 free-ranging mule deer fa\\ns during winters 2010-2011 and 2011-2012. Collars were modified to shed from deer approximately I month after the collaring event. Two fawns were successfully recollared after they shed the first collars they received. Thus. we observed IO successful collaring events involving 8 unique fawns. Fawns demonstrated minimal response to collaring events. either remaining in the de,,ice or calmly exiting. A fawn typically required 2::1 weeks of daily e~-posure before fully entering the device and extending its head through the outstretched collar. which was necessary for a collaring event to occur. This slow acclimation period limited utility of the device when compared with traditional capture techniques. Future work should focus on device modifications and altered baiting strategies that decrease fawn acclimation period, and in tum, increase collaring rates, providing a noninvasive and perhaps cost-effective alternative for monitoring mid- to large-sized mammal species. © 2019 The Wild Ii fe Society. Behavioral and demographic responses of mule deer to energy development on winter range Joseph M. Northrup', J.M., Charles R. Anderson Jr. 1, Brian D. Gerber. and George Wittemycr 1 1Department of Fish, Wildlife and Conservation Biology, Colorado State University. 14 74 Campus Delivery, Fort Collins. CO 80523 USA 2Mammals Research Section. Colorado Parks and Wildlife. 317 W. Prospect Road. Fort Collins. CO 80526 USA 'Department of Natural Resources Science, University of Rhode Island. I Gr~nhousc Road. Kingston. RI 02881 USA Citation: Nonhrup, J. M.. C. R. Anderson Jr.. B. D. Gerber. and G. Winemyer. 2021. Bcha\'ioral and demographic responses of mule deer to energy development on winter range. Wildlife Monographs 208: 1-37: 2021: DOI: 10.1002/\\mon.1060. ABSTRACT Anthropogenic habitat modification is a major driver of global biodiversity loss. In North America. one of the primary sources of habitat modification over the last 2 decades has been exploration for and production of oil and natural gas (hydrocarbon development). which has led to demographic and behavioral impacts to numerous wildlife species. Developing effective measures to mitigate these impacts has become a critical task for wildlife managers and conservation practitioners. However. this task has been hindered by the difficulties involved in identifying and isolating factors driving population responses. Current research on responses of v.ildlife to development predominantly quantifies behavior. but it is not always clear how these responses scale to demography and population dynamics. Concomitant assessments of behavior and population-level processes are needed to gain the mechanistic understanding required to develop effective mitigation approaches. We simultaneously assessed the demographic and behavioral responses of a mule deer population to natural gas development on winter range in the Piceance Basin of Colorado. USA. from 2008 to 2015. Notably. this was the period when development declined from high levels of active drilling to only production phase activity (i.e .. no drilling). We focused our data collection on 2 contiguous mule deer winter range study areas that experienced starkly different levels of hydrocarbon development within the Piceance Basin. We assessed mule deer behavioral responses to a range of development features with varying levels of associated human activity by examining habitat selection patterns of nearly 400 individual adult female mule deer. Concurrently. we assessed the demographic and physiological effects of natural gas development by comparing annual adult female and overwinter fawn (6-monthold animals) survival. December fa,m mass. adult female late and early winter body fat. age. pregnancy rates. fetal counts. and lactation rates in December between the 2 study areas. Strong differences in habitat selection between the 2 study areas were apparent. Deer in the less-developed study area avoided development during the day and night and selected habitat presumed to be used for foraging. Deer in the heavily developed study area selected habitat presumed to be used for thermal and security cover to a greater degree. Deer faced with higher densities of development avoided areas with more well pads during the day and responded neutrally or 20 u �\__,I selected for these areas at night. Deer in both study areas showed a strong reduction in use of areas around well pads that were being drilled. which is the phase of energy development associated with the greatest amount of human presence. vehicle traffic. noise. and artificial light. Despite divergent habitat selection patterns. we found no effects of development on individual condition or reproduction and found no differences in any of the physiological or vital rate parameters measured at the population level. However. deer density and annual increases in density were higher in the low-development area. Thus. the recorded behavioral alterations did not appear to be associated with demographic or physiological costs measured at the individual level. possibly because populations are below winter range carrying capacity. Differences in population density between the 2 areas may be a result of a population decline prior to our study (when development was initiated) or area-specific differences in habitat quality, juvenile dispersal. or neonatal or juvenile survival: however. we lack the required data to contrast evidence for these mechanisms. Given our results. it appears that deer can adjust to relatively high densities of well pads in the production phase (the period with markedly lower human activity on the landscape). provided there is sufficient vegetative and topographic cover afforded to them and populations are below carrying capacity. The strong reaction to wells in the drilling phase of development suggests mitigation efforts should focus on this activity and stage of development Many of the wells in this area were directionally drilled from multiplewell pads. leading to a reduced footprint of disturbance. but were still related to strong behavioral responses. Our results also indicate the likely value of mitigation efforts focusing on reducing human activity (i.e .. vehicle tratlic. lighL and noise). In combination. these findings indicate that attention should be paid to the spatial configuration of the final development footprint to ensure adequate cover. In our study system. minimizing the road network through landscape-level development planning would be valuable (i.e.~ exploring a maximum road density criteria). Lastly. our study highlights the imponancc of concomitant assessments of behavior and demography to provide a comprehensive understanding of how wildlife respond to habitat modification. 0 2021 The Wildlife Society. 21 �Appendix B. Preliminary results of habitat treatment responses and herbivore use of treated sites. Vegetation and camera data to accompany the study ' Population performance of Piceance Basin mule deer in response to natural gas resource selection and mitigation efforts to address human activi~1• and habitat degradation ' Principal Investigators: Danielle Johnston (Danielle.bil\'eu ll state.co.us). Chuck Anderson (chuck.anderson@state.co. us) Collaborators: Colorado Parks and Wildlife. BLM-White River Field Office. Idaho State University, Colorado State University, Federal Aid in Wildlife Restoration, EnCana Corp.. ExxonMobil Prod. Co./XTO Energy, Marathon Oil Corp., Shell Petroleum WPX Energy. Colorado Mule Deer Assn .. Muley Fanatic Found. , Colorado Mule Deer Found .. Colorado State Severance Tax Fund. Boone & Crocket Club, and Safari Club Int. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW's right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. In 2011 and 2013, about I ,200 acres of pinyon and juniper (PJ) mastication treatments were completed in the Magnolia region of the Piceance Basin. Treated parcels averaged 7 acres in size. and were intended to increase winter range quality for deer. The treatments were pa11 of a study to evaluate the effectiveness of PJ removal as mitigation for impacts of natural gas development on deer. with outcomes assessed in terms of deer population and demographic parameters. This summary addresses some side questions relevant to the main study, with outcomes assessed in terms of vegetation response and animal use of vegetation treatments. We were interested in quantifying the understory forage produced by the mastication treatments. We used paired masticated/control point-intercept transects on a subset of parcels (Graham 20 I 3) to quantify cover of plant groups relevant to deer nutrition. We used belt transects and trained ocular estimation, with benchmarks (Johnston 2018), to estimate summer utilization on individual shrubs. then scaled these to the plot level (Bilyeu. Cooper et al. 2007). We used belt transects of shrub canopy measurements. coupled with biomass equations developed for the study area (Johnston 2018) to quantify winter forage production of key browse species. Winter forage production was defined as current-year stems. not including leaves. not including biomass removed by summer browsing. and not including very small stems which wou ld likely be shed prior to winter (Johnston 20 18). We were interested in how summer use of treatments. and use of treatments by non-target animals. impacted winter forage availability. Ten cattle exclosures. distributed broadly throughout the study area (Figure 1), were built within mastication treatments in 20 11 and 20 13. We assessed plant cover and summer shrub uti lization within these using techniques described above. On paired masticated/control transects, we deployed Reconyx Hypertire cameras July-November 20I8-2019. These were programmed to facilitate creating an index of use: 5 pictures per motion trigger. 3 second interval between pictures. a 5 minute wait time between triggers, and a sensitivity sening of High (Rhodes. Larsen et al. 20 18). An animal observed with their head down or other indication of foraging in one or more of the photos in a 5 photo set was counted as one foraging event, and non-foraging occurrences were counted similarl y. Sampl ing efforts by year are given in Table I. Because the plant cover data contained many zeros. we modeled presence/absence of each plant group separately from its cover where present (Fletcher. Mackenzie et al. 2005). using the lme4 package in R (Bates 2005). For both analyses, treatment. year. and their interaction were considered ft:xed effects. year was included as a categorical variable. and pair ID and plot ID were included as random effects. We used a similar approach for camera data for cattle and elk. which also contained many zeros. 22 �--- ---- ----- In general, grasses responded positively to treatment (Figure 2a). Wheatgrass presence, wheatgrass cover, and needlegrass presence were higher in treated than untreated plots. Poa grass presence was higher in treated plots by 2018, although poa grass presence and cover initially had a negative response to treatment. Cheatgrass presence also responded positively to treatment (Figure 2a). Wheatgrasses, poas, and cheatgrass all had significant year*treatment interactions for either presence or cover. Interannual variation in cover was greater in masticated plots than in control plots for these species groups (Figure 2a). Forbs responded positively to treatment. Annual forb and perennial forb presence were higher in treated than untreated plots (Figure 2b). Some shrubs responded positively to treatment, while others did not. Snowberry cover was lower in treated plots in 2013, but in 2016 and 2018, cover was higher in treated plots (Figure 2c). Variation in snowberry cover was greater in masticated than in control plots (Figure 2c). Bitterbrush did not display any significant effects until 2018, when cover was higher in treated plots (Figure 2c). Serviceberry cover was lower in treated plots over all years (Figure 2d). Sagebrush cover was initially lower in treated plots, but by 2018 this difference was no longer significant (Figure 2d). Summer utilization of serviceberry and mountain mahogany in 2018 was significantly higher in masticated than in control plots, but no differences were detected in bitterbrush or sagebrush. Winter forage production, which was summed over serviceberry, mountain mahogany, and bitterbrush, was significantly higher in masticated plots than in unmasticated plots in all years except 2016, when the pattern was reversed (Figure 3). There was no significant effect of exclosures on any plant cover group or on summer utilization in 2018. Deer, horse, elk, and cattle all foraged more often in masticated plots than in controls in 2018 (Figure 4). Cattle were only observed foraging at 6 of 20 locations, horse were observed at 9, deer at 19, and elk at 6. Mastication treatments had many positive effects on forage availabilty, including higher cover of desirable grass groups such as poa grasses and wheatgrasses. higher cover of perennial forbs, and usually higher productivity of winter-available shrub forage. There were some negative effects and some differences in effects among years, however. Cheatgrass was higher in masticated plots than in controls, and snowberry cover was higher in masticated plots in 2016 and 2018. 2016 was an unusual year compared to other years of this study, with very high productivity of grasses (including cheatgrass, especially in masticated plots), and unusua11y high productivity of winter-available forage of desirable shrubs in control but not masticated plots. Summer shrub utilization in 2018 was higher in masticated plots than in controls. We lack any data on utilization from 2016, which might have helped explain if the lower production of winter-available forage in masticated plots was due to higher summer utilization in those plots that year. Another explanation for the 2016 results is that good conditions for grass, cheatgrass, and/or snowberry productivity in masticated plots led to increased competition which lessened productivity of desirable forage shrubs. All four of the large herbivores of interest foraged more frequently in summer and fa11 in masticated plots than in control plots in 2018. The impact of cattle was concentrated in only a few plots, but they did forage frequently in plots where they occurred. Cattle use ended in September, prior to the period of heavy use by deer in October. The data from the cattle exclosures does not indicate that cattle are having any measurable negative effect on forage resources. In summary the impact of cattle on the forage resources available to deer in mastication treatments seems minimal. However, the effect of the sum of cattle, horse, and elk foraging may have some impact. In 2019, we collected vegetation data and camera data. 2019 is the last year of data collection for this study, and final analyses will be incorporated into publications in 2020-21. LITERATURE CITED Bates, D. (2005). "Fitting linear mixed models in R." R news 5(1). Bilyeu, D. M., D. J. Cooper and N. T. Hobbs (2007). "Assessing impacts of large herbivores on shrubs: tests of scaling factors for utilization rates from shoot-level measurements." Journal of Applied Ecology 44( I): I68-175. 23 �Fletcher, D., D. D. Mackenzie and E. Villouta (2005). "Modelling skewed data with many zeros: a simple approach combining ordinary and logistic regression." Environmental and ecological statistics 12: 45-54. Graham, T. (2013). Magnolia habitat manipulation project vegetative monitoring: June 2013 notes on data collection and methods used, Ranch Advisory Partners, LLC: 7. Johnston, D. 8. (2018). Wildlife Research Report: Examining the effectiveness of mechanical treatments as a restoration technique for mule deer habitat. Fort Collins, CO. Colorado Parks and Wildlife. Rhodes, A. C., R. T. Larsen and S. B. S. Clair (2018). "Differential effects of cattle, mule deer, and elk herbivory on aspen forest regeneration and recruitment." Forest Ecology and Management 422: 273280. 24 �Table 1. Number of transects sampled for a given data t:ype each year. Variables quantified 2011 2012 2013 2014 2015 2016 2018 90* 90* 145 69 Percent cover of plant l07t 159 functional groups Winter-available forage of bitterbrush, serviceberry, mountain mahogany (ShrubMassPerArea) Summer utilization of bitterbrush, service berry, mountain mahogany, and sagebrush Index of deer, elk, horse, and 70t 27t 63 2019 75t 40 (camera sites) 75t 75t 75t 40 40 cattle use in summer and fall, (2 (2 as determined by trail camera (EventsPerDay) * Pretreatment data collected 20 I 1-20 I2 will be added to a later report. tlncludes 24-30 locations taken at exclosure sites. cameras each) cameras each) 25 �Figure 1. Sampling locations within the Magnolia region r~f"the Pic:eance Basin. 26 �b 10.0 - a .... Q) e 15 · ra s I > ._ a, > 7.5 - 0 0 _!: 10 · ca, 5.o - (.) C: Q) ~ ~ Q) Q) a. a. 5 · o- 2.5 · 0.0 • ' 2013 ' 2014 ' 2018 ' 2016 ' 2013 .... Q) snowb~ +, C '' ,, 8 c 6· , ' Q) ~ 2016 2018 d 8- I 9- ' 2018 year year C ' 2016 ' 2014 ' ai 6 • > 8 ~ 4 _ sa ....Q) (.) Q) a. a. 3· 2· 20'13 ' 2016 2014 2013 2018 ' 2014 year year Figure 2. Cover ofsome p!antfimcriona! groups and species imporranr.for evaluaring habitar quality. Dashed lines indicare masricared plots and solid lines are confrols. A "+ " or .. _.. sign indicates significant posifive or negative main effect of masticafion across years (a = 0. 05). "P .. indicates that the sign[ficant ejject was observed in the prese11celabse11ce ana(rsis. and "C •• indicates a significant effect in the cover-where-present ana(vsis. 27 �_ 20 · NE --- T ~ Treatment "'~ __.. control ..0 "' 2 90 - • masbcated .c l "' * ..1.. ...1.. I I 2013 20 14 2015 2016 2018 year Figure 3. Mass of winter-available forage (curre111-year stem mass measured in September, not including leaves or mass removed by summer browsing) per unit shrub area. Data are summed over serviceberry, mountain mahogany. and bitterbrush. N=8.fhr 2013 and 2015 and ]5-31.for other years. No transects inside fences were included. Error bars = SE. Stars indicate siKn(ficant differences at alpha = 0. 05 28 �Not foraging, near camera Foraging, near camera a 75 · .,,>- 75 • 0 ai Cl. .,, :I: Treatment 50 · ai 50 • Cl. control 1/l ■ masticated C: Ql > w Cl .!: 25 • 25 • * Cl !!! 0 u.. O· O· cattle elk horse callle mule deer elk horse mule deer Herbivore Herbivore Figure -I. a) Average number offoraging eve111s per hectare per day between mid-July and midNovember. 2018 in control versus masticated plots. Stars indicate sign!ficant differences at a = 0.05. t indicates a sig11(ficcmt d(ffere11ce in presence <~fjoraging events. h) Average number of non.foraging observations per hectare per day. 29 �> Colorado Parks and Wildlife July 1, 2021 - June 30, 2022 WILDLIFE RESEARCH REPORT State of _ _ _ _ _____.:::C=o=lo=r=ad=o=--_ _ _ _ _ : ~P=ar:..!.:ks=an:.:.:d::....:..W:..:i=ld:::.:li=fe=--------------Cost Center 3430 : .:.:.M=a=mm=a=l=s=R=e=se=a=rc=h=-------------Work Package 3001 : .::::D~e.:.:er~C::::.o=n=s=erv:...:..::a=ti=on~-----------Task No. 6 : Population Performance of Piceance Basin Mule Deer in Response to Natural Gas Resource Extraction and Mitigation Efforts to Address Human Activity and Habitat Degradation Federal Aid Project:. ______W__-"""""24__3a.. .-""""'R'"""'-6a.. .__ _ __ Period Covered: July 1, 2021 - June 30, 2022 Author: C. R. Anderson, Jr. Personnel: D. Bilyeu-Johnston, CPW; J. Northrup, Ontario Ministry ofNatural Resources and Forestry; R. Marrotte and Helena Rheault, Environmental and Life Sciences Graduate Program, Trent University; B. Gerber, University of Rhode Island. Project support received from Federal Aid in Wildlife Restoration, Colorado Mule Deer Association, Colorado Mule Deer Foundation, Muley Fanatic Foundation, Colorado State Severance Tax Fund, Caerus Oil and Gas LLC, EnCana Corp., ExxonMobil Production Co./XTO Energy, Marathon Oil Corp., Shell Petroleum, Williams and WPX Energy. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW's right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. 1 �WILDLIFE RESEARCH REPORT POPULATION PERFORMANCE OF PICEANCE BASIN MULE DEER IN RESPONSE TO NATURAL GAS RESOURCE EXTRACTION AND MITIGATION EFFORTS TO ADDRESS HUMAN ACTIVITY AND HABITAT DEGRADATION CHARLES R. ANDERSON, JR PROJECT NARRITIVE OBJECTIVES l. To determine experimentally whether enhancing mule deer habitat conditions on winter range elicits behavioral responses, improves body condition, increases fawn survival, and ultimately, population density on mule deer winter ranges exposed to extensive energy development. 2. To determine experimentally to what extent modification of energy development practices enhance habitat selection, body condition, fawn survival, and winter range mule deer densities. SEGMENT OBJECTIVES I. Finalize publication of results addressing the role of memory in seasonal habitat selection patterns of migratory mule deer during different time scales. 2. Continue analyses of vegetation and mule deer responses to habitat treatments intended as a mitigation option to offset energy development disturbance. ' 3. Submit final results addressing vegetation and mule deer responses to 3 mechanical treatment methods on pinyon-juniper winter range for publication. 4. Construct a web-based energy development planning tool to guide future energy development to minimize and/or mitigate mule deer disturbance on winter range. PROJECT OVERVIEW AND RESEARCH SUMMARY We propose to experimentally evaluate winter range habitat treatments and human-activity management alternatives intended to enhance mule deer (Odocoileus hemionus) populations exposed to energy-development activities. The Piceance Basin of northwestern Colorado was selected as the project area due to ongoing natural gas development in one of the most extensive and important mule deer winter and transition range areas in Colorado. The data presented here represent preliminary and final results of a 10-year research project addressing habitat improvements as mitigation and evaluation of deer responses to energy development activities to inform future development planning options on important seasonal ranges. From2008-2019, we monitored deer on 4 winter range study areas representing relatively high (Ryan Gulch, South Magnolia) and low (North Magnolia, North Ridge) levels of development activity (Fig. 1) to address factors influencing deer behavior and demographics and to evaluate success of habitat treatments as a mitigation option. We recorded adult female habitat use and movement patterns; estimated neonatal, overwinter fawn and annual adult female survival; estimated annual early and late winter body condition, pregnancy and fetal rates of adult females; and estimated annual mule deer abundance among study areas. Winter range habitat improvements completed spring 2013 resulted in 604 acres of mechanically treated pinion-juniper/mountain shrub habitats in each of 2 treatment areas (Fig. 2) with minor (North Magnolia) and extensive (South Magnolia) energy development, respectively. 2 �During this research segment, we finalized publication of results investigating the role of memory in seasonal habitat selection of migratory mule deer over different time scales (Rheault et al. 2021; Appendix A), continued analyses of vegetation and mule deer responses to habitat treatments intended as a mitigation option to offset energy development disturbance (preliminary results reported in Appendix B), submitted final results addressing vegetation and mule deer responses to 3 mechanical treatment methods on pinyon-juniper winter range for publication (Johnston and Anderson /11 review), and constructed a web-based energy development planning tool to guide future energy development to minimize and/or mitigate mule deer disturbance on winter range (Marrotte et al. In prep). Based on final (migration, mule deer behavioral and demographic responses, reproductive success and neonate survival; see Anderson 2019 for detailed methods and results and Appendix A for publication abstracts) and preliminary data analyses (vegetation and herbivore response to habitat treatments, Appendix B) for this 10-year project: (I) annual adult female survival was consistent among areas averaging 79-87% annually, but overwinter fawn survival was variable, ranging from 31 % to 95% within study areas, with annual and study area differences primarily due to early winter fawn condition, annual weather conditions, and factors associated with predation on winter range; (2) mule deer body condition early and late winter was generally consistent within areas, with higher variability among study areas early winter, primarily due to December lactation rates, and late winter condition related to seasonal moisture and winter severity; (3) late winter mule deer densities increased through 2016 in all study areas, ranging from 50% in North Ridge to I03% in North Magnolia, but have stabilized recently in 3 of the 4 study areas with recent decline evident in North Ridge (Fig. 3); (4) migratory mule deer selected for areas with increased cover and increased their rate of travel through developed areas, and avoided negative influences through behavioral shifts in timing and rate of migration, but did not avoid development structures (Fig. 4); (5) mule deer exhibited behavioral plasticity in relation to energy development, without evidence of demographic effects, where disturbance distance varied relative to diurnal extent and magnitude of development activity (Fig 5), which provide for useful mitigation options in future development planning; and (6) energy development activity under existing conditions did not influence pregnancy rates, fetal rates or early fawn survival (0-6 months), but may have reduced neonatal survival (March until birth) during 2012 when drought conditions persisted during the third trimester of doe parturition ( Fig. 6). Final results are pending to address vegetation and mule deer responses to assess habitat treatment mitigation options for energy development planning and spatial planning tool development is in progress. Final data collection efforts for this project was completed by spring 2020 (final GPS collar recovery). Collaborative research with agency biologists. graduate students. and university professors has produced 23 scientific publications addressing improved monitoring techniques for neonate mule deer captures (Bishop et al. 2011, Peterson et al. 2018b); development and evaluation of a remote mule deer collaring device (Bishop et al. 2019); mule deer migration relative to energy development (Lendrum et al. 2012. 2013, 2014; Anderson and Bishop 2014), improved approaches to address animal habitat use patterns (Northrup et al. 2013; Rheault et al. 2021 ); mule deer response to helicopter capture and handling (Northrup et al. 2014a ); potential effects of male-biased harvest on mule deer productivity (Freeman et al. 2014); mule deer genetics in relation to body condition and migration (Northrup et al. 2014b); acoustic monitoring to investigate spatial and temporal factors influencing mule deer vigilance (Lynch et al. 2014) and foraging behavior (Northrup et al. 2019); the relationship of plant phenology with mule deer body condition (Searle et al. 2015); approaches to identify cause-specific mortality in mule deer from field necropsies (Stonehouse et al. 2016); the influence of individual and temporal factors affecting late winter body condition estimates of adult female mule deer (Bergman ct al. 2018 ); and mule deer behavioral and demographic responses to energy development activities to inform future development planning (Northrup et al. 2015, 2016a, 2016b, 2021, Peterson ct al. 2017, 2018a). These publications arc summarized in Appendix A and preliminary results describing vegetation and herbivore responses to habitat treatments are reported in Appendix B. Wc anticipate the opportunity to work cooperatively toward developing solutions for allowing the nation's energy reserves to be developed in a manner that benefits wildlife and the people who value both the wildlife and energy resources of Colorado and elsewhere. 3 �Wl!II Pads & Faclll~es South MagnOf•a ,i:,rtn R idge .C In oevelopme nt l Producing 'n'CII _ Development tac1:1t1 es 10 f,t.le-s Figure I. Mule deer winter range study areas relative to active natural gas well pads and energy development facil ities in the Piceance Basin of northwest Colorado, winter 2013/14 (Accessed http://cogcc.state.co. us/ December 3 1, 2013; energy development drilling activity has been minor since 20 12). 4 �North Magnoha treatement sites (587 acres) D Be;; rSet_15_35b_andC, r--• BearSet_ 1_8andA_E D BearSet_36_54andJ GreasewoodSet_g I 6_g29 GreasewoodSet_g I _g 15 D GreasewoodSet_g30_g-l 2 LeeOversrghts_a_fand 16_ 17 t,lechanrcar treatment companson (54 acres) - - North Hatch Prlo t Trea tments ( 116 acres) le Deer Study Areas North Magnolra South Magnolia 4 Figure 2. Habitat treatment site delineations in 2 mule deer study areas (604 acres each) of the Piceance Basin, northwest Colorado (Top; cyan polygons completed Jan 2011 using hydro-axe; yellow polygons completed Jan 201 2 using hydro-axe, roller-chop, and chaining; and remaining polygons completed Apr 201 3 using hydro-axe). January 20 I I hydro-axe treatment-site photos from North Hatch Gulch during April (Lower left, aerial view) and October, 20 11 (Lower right, ground view). 5 �Piceance Basin late winter mule deer density 35.00 30.00 25.00 1-:::- 20.00 - t 15.00 0 - North Ridge • • • • • • Ryan Gulch 10.00 - • North Magnolia 5.00 - South Magnolia 0.00 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Year Figure 3. Mule deer density estimates and 95% C I (error bars) from 4 winter range herd segments in the Piceance Basin. northwest Colorado, late winter 2009-2018. Figure 4. Mule deer study areas in the Piceancc Basin of northwestern Colorado, USA (Top), spring 2009 migration routes of adult female mule deer (11 = 52: Lower left), and acti ve natural-gas well pads (black dots) and roads (state, county, and natural-gas: white lines) from May 2009 (Lower right; from Lendrum et al. 20 12: http://dx.doi.org/l 0. I 890/ES I "'-00 l 65 . 1). 6 �M I Prod 400 Prod 600 Prod 800 Prod 1000 Drill 400 Dnll 6(10 Drill 800 Dnll 1000 Dr<ll 600 Drill SQQ Drill 1000 Covanaces M I Prod 400 Prod 600 Prod 800 Prod 1000 Dr<II 400 Figure 5. Posterior distributions of population-level coefficients related to natural gas development for RSF models during the day (top) and night (bottom) for 53 adult female mule deer in the Piceance Bas in, northwest Colorado. Dashed line indicates 0 selection or avoidance (below the line) of the habitat features. 'Drill ' and ' Prod ' represent drilling and producing well pads, respectively. The numbers fo llowing ' Drill ' or ' Prod' represent the distance from respective well pads evaluated (e.g.. ' Drill 600' is the number of well pads with active drilling between 400- 600 m from the deer location; from Northrup et al. 2015; http://onlinelibrary.wiley.com/doi/ I 0. 1111 /gcb.13037/abstract). Road disturbance was relatively minor (~60- 120 m, not illustrated above). 1.00 0.80 Q) ro I... ro> 0.60 -~ ::, Cf) ro 0.40 a> lL 0.20 0.00 2013 2012 2014 Year □ High development □ Low development I Figure 6. Model averaged estimates of mule deer fetal survival from early March until bi11h (late MayJune) in high and low energy development study areas of the Piccancc Basin, northwest Colorado, 20 122014 (from Peterson et al. 2017; http://www.bioone.org/doi/ pdf/l 0.298 I/w lb.0034 1). 7 �LITERATURE CITED Anderson, C. R., Jr. 2019. Population performance of Piceance Basin mule deer 111 response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Federal Aid in Wildlife Restoration Annual Report W-243-R3, Ft. Collins. CO USA. Anderson, C.R., Jr., and C. J. Bishop. 2014. Migration patterns of adult female mule deer in response to energy development. Pages 47-50 in Transactions of the 79th North American Wildlife & Natural Resources Conference (R. A. Coon & M. C. Dunfee, eds.). Wildlife Management Institute, Gardners. PA, USA. ISSN 0078- I 355. Bergman. E. J.. C. R. Anderson Jr., C. J. Bishop. A. A. Holland, and J. M. Northrup. 20 18. Variation in ungulate body fat: individual versus temporal effects. Journal of Wildlife Management 82: 130-137, DOI: 10: 1002/jwmg.21334 Bishop, C. J .. C. R. Anderson Jr.. D. P. Walsh. E. J. Bergman. P. Kuechle. and J. Roth. 20 11 . Effectiveness of a redesigned vaginal implant transmitter in mule deer. Journal of Wildlife Management 75(8): 1797-1806; DOI : 10.1002/j wmg.229 Bishop, C. J., M. W. Alldredge, D. P. Walsh, E. J. Bergman. C.R. Anderson Jr., D. Kilpatrick, J. Bake!, and C. Fabvre. 2019. A noninvasive automated device for remotely collaring and weighing mule deer. Wildlife Society Bulletin 43:717-725; doi.org/10.1002/wsb. l 034 Freeman, E. D., R. T . Larsen, M. E. Peterson, C. R. Anderson, Jr., K. R. Hersey, and B. R. Mc Millan. 2014. Effects of male-biased harvest on mule deer: implications for rates of pregnancy, synchrony, and timing of parturition. Wildlife Society Bulletin; DOI: I 0.1002/wsb.450 Johnston, D. B., and C. R. Anderson Jr. /11 Re view. Plant and mule deer responses to pinyon-juniper removal by three mechanical methods. Wildlife Society Bulletin. Lendrum, P. E. , C. R. Anderson, Jr., R. A. Long, J. K. Kie, and R. T. Bowyer. 2012. Habitat selection by mule deer during migration: effects of landscape structure and natural gas development. Ecosphere 3(9):82. http ://dx.doi.org/10. Lendrum, P. E. , C.R. Anderson, Jr.. K. L. Monteith. J. A. Jenks, and R. T. Bowyer. 2013. Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes. PLoS ONE 8(5): e64548. doi: l 0.13 7 l/joumal.pone.0064548 Lendmm, P. E., C.R. Anderson, Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 20 14. Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns of forage quality. Mammalian Biology: http://dx.doi.orn/ l 0. 10 16/ j.mambio.20 14.05.005 Lynch, E., J . M. Northrup, M. F. McKenna, C.R. Anderson Jr., L. Angeloni, and G. Wittemyer. 2014. Landscape and anthropogenic featu res influence th e u se of auditory vig ilance by mule deer . Behavioral Ecology; doi: 10.1093/beheco/aru 158. Marrotte, R. R., C. R. Anderson Jr., and J. M. Northrup. /11 Prep . Developing a spatial planning tool for natural gas development on mule deer winter range. Northrup, J. M., M. B. Hooten, C.R. Anderson, Jr., and G. Wittemyer. 2013 . Practical guidance on characterizing availability in resource selection functions under a use-availability design. Ecology 94(7): 1456-1463 . Northrnp, J. M., C.R. Anderson, Jr., and G . Wittemycr. 2014a. Effects of helicopter capture and handling on movement behavior of mule deer. Journal of Wildlife Management 78(4 ):73 1738; DOI: 10.1002/jwmg.705 Northrup, J. M., A. B. Shafer, C. R. Anderson Jr., D. W. Coltman, and G. Whittcmyer. 2014b. Finescale genetic correlates to condition and migration in a wild cervid. Evolutionary Applications ISSN 1752-4571 ; doi : 1O. l l l l/eva. 121 89 Northrup, J. M., C.R. Anderson, Jr., and G. Wittcmyer.201 5. Quantifying spatial habitat loss from h ydrocarbon development through assessing habitat selection patterns of mule deer. Global Change Biology, doi : I 0.1111/gcb. 13037. 8 �Northrup, J.M., C.R. Anderson, Jr., M. B. Hooten, and G. Wittemyer. 2016a. Movement reveals scale dependence in habitat selection of a large ungulate. Ecological Applications 26:27462757 Northrup, J.M., C.R. Anderson, Jr., and G. Wittemyer. 2016b. Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid. Diversity and Distributions 22: 547-557, DOI: 10.1111/ddi.12417 Northrup, J.M., A. Avrin, C.R. Anderson Jr., E. Brown, and G. Wittemyer. 2019. On-animal acoustic monitoring provides insight to ungulate foraging behavior. Journal of Mammalogy 100: 14791489; https://doi.org/lO. I093/jmammal/gyz124 Northrup, J.M., C. R. Anderson Jr., B. D. Gerber, and G. Wittemyer. 2021. Behavioral and demographic responses of mule deer to energy development on winter range. Wildlife Monographs 208: 1-37; 2021; DOI: 10.1002/wmon.1060 Peterson, M. E., C.R. Anderson Jr., J.M. Northrup, and P. F. Doherty Jr. 2017. Reproductive success of mule deer in a natural gas development area. Wildlife Biology doi: IO. l l l l/wlb.00341 Peterson, M. E., C. R. Anderson Jr., J. M. Northrup, and P. F. Doherty Jr. 2018a. Mortality of mule deer fawns in a natural gas development area. Journal of Wildlife Management 82:1135-1148, DOI: l0.1002/jwmg.21476 Peterson, M. E., C. R. Anderson Jr., M. W. Alldredge, and P. F. Doherty Jr. 2018b. Using maternal mule deer movements to estimate timing of parturition and assist fawn captures. Wildlife Society Bulletin 42:616-621; DOI: 10.l002/wsb.935 Rheault, H., C. R. Anderson Jr., M. Bonar, R. R. Marrotte, T. R. Ross, G. Wittemyer, and J. M. Northrup. 2021. Some memories never fade: inferring multi-scale memory effects on habitat selection of a migratory ungulate using step-selection functions. Frontiers in Ecology and Evolution 9:702818; doi: I 0.3389/fevo.2021. 702810 Searle, K. R., M. B. Rice, C.R. Anderson, C. Bishop and N. T. Hobbs. 2015. Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore. Oecologia ISSN 0029- 8549; DOI 10.1007/s00442-015-3348-9 Stonehouse, K. F., C. R. Anderson Jr., M. E. Peterson, and D. R. Collins.2016. Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus). Colorado Parks and Wildlife Technical Report No. 48, First Edition, 317 W. Prospect Rd., Ft. Collins, CO USA. DOW-R-T-48-16, ISSN 0084-8883. Prepared by Ch UC k And erson Digitally signed by Chuck Anderson Date:2022.10,0605:49:08-06'00' Charles R. Anderson, Jr., Mammals Research Leader 9 �Appendix A. Abstracts of published manuscripts resulting from Piceance Basin mule deer/energy development interaction research collaborations. Abstract format specific to the respective journal requirements. Effectiveness of a redesigned vaginal implant transmitter in mule deer CHAD J. BISHOP 1, CHARLES R. ANDERSON Jr. 1, DANIEL P. WALSH 1, ERIC J. BERGMAN', PETER KUECHLE 2, and JOHN ROTH2 1 Colorado Parks and Wildlife, Fort Collins, Colorado 80526 USA 2 Advanccd Telemetry Systems, Isanti, Minnesota 55040 USA Citation: Bishop, C. J., C.R. Anderson Jr., D. P. Walsh, E. J. Bergman, P. Kuechle, and J. Roth. 2011. Effectiveness ofa redesigned vaginal implant transmitter in mule deer. Journal ofWildlife Management 75(8):1797-1806; DOI: 10.1002/jwmg.229 ABSTRACT Our understanding of factors that limit mule deer (Odocoileus hemionus) populations may be improved by evaluating neonatal survival as a function of dam characteristics under free-ranging conditions, which generally requires that both neonates and dams are radiocollared. The most viable technique facilitating capture of neonates from radiocollared adult females is use of vaginal implant transmitters (VITs). To date, VITs have allowed research opportunities that were not previously possible; however, VITs are often expelled from adult females prepartum, which limits their effectiveness. We redesigned an existing VIT manufactured by Advanced Telemetry Systems (ATS; Isanti, MN) by lengthening and widening wings used to retain the VIT in an adult female. Our objective was to increase VIT retention rates and thereby increase the likelihood of locating birth sites and newborn fawns. We placed the newly designed VITs in 59 adult female mule deer and evaluated the probability of retention to parturition and the probability of detecting newborn fawns. We also developed an equation for determining VIT sample size necessary to achieve a specified sample size of neonates. The probability of a VIT being retained until parturition was 0.766 (SE= 0.0605) and the probability of a VIT being retained to within 3 days of parturition was 0.894 (SE= 0.0441 ). In a similar study using the original VIT wings (Bishop et al. 2007), the probability of a VIT being retained until parturition was 0.447 (SE= 0.0468) and the probability of retention to within 3 days of parturition was 0.623 (SE= 0.0456). Thus, our design modification increased VIT retention to parturition by 0.319 (SE= 0.0765) and VIT retention to within 3 days of parturition by 0.271 (SE = 0.0634 ). Considering dams that retained VITs to within 3 days of parturition, the probability of detecting at least I neonate was 0.952 (SE= 0.0334) and the probability of detecting both fawns from twin litters was 0.588 (SE = 0.0827). We expended approximately 12 person-hours per detected neonate. As a guide for researchers planning future studies, we found that VIT sample size should approximately equal the targeted neonate sample size. Our study expands opportunities for conducting research that links adult female attributes to productivity and offspring survival in mule deer.© 2014 The Wildlife Society. Habitat selection by mule deer during migration: effects of landscape structure and natural-gas development PATRICK E. LENDRUM 1, CHARLES R. ANDERSON JR.2, RYAN A. LONG 1, JOHN G. KIE 1, AND R. TERRY BOWYER1 1Department ofBiological Sciences, Idaho State University, Pocatello, Idaho 83209 USA 2 Colorado Parks and Wildlife, Grand Junction, Colorado 81 SOS USA Citation: Lendrum, P. E., C.R. Anderson Jr., R. A Long, J. G. Kie, and R. T. Bowyer. 2012. Habit:it selection by mule deer during migration: effects of landscape structure and natural-gas development. Ecosphere 3(9):82 http://dx.doi.org/lO. l 890/ES 12-00165. l Abstract. The disruption of traditional migratory routes by anthropogenic disturbances has shifted patterns of resource selection by many species, and in some instances has caused populations to decline. Moreover, in recent decades populations of mule deer (Odocoileus hemionus) have declined throughout much of their historic range in the western United States. We used resourceselection functions to determine if the presence ofnatural-gas development altered patterns of resource selection by migrating mule deer. We compared spring migration routes of adult female mule deer fitted with GPS collars (n = 167) among four study areas that had varying degrees of natural-gas development from 2008 to 2010 in the Piceance Basin of northwest Colorado, USA. Mule deer migrating through the most developed area had longer step lengths (straight-line distance between successive GPS locations) compared with deer in less developed areas. Additionally, deer migrating through the most developed study areas tended to select for habitat types that provided greater amounts of concealment cover, whereas deer from the least developed areas tended to select habitats that increased access to forage and cover. Deer selected habitats closer to well pads and avoided roads in all instances except along the most highly developed migratory routes, where road densities may have been too high for deer to avoid roads without deviating substantially from established migration routes. These results indicate that behavioral tendencies toward avoidance of anthropogenic disturbance can be overridden during migration by the strong fidelity ungulates demonstrate towards migration routes. If avoidance is feasible, then deer may select areas further from development, whereas in highly developed areas, deer may simply increase their rate of travel along established migration routes. 10 �Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes Patrick E. Lendrum•, Charles R. Anderson Jr.2, Kevin L. Monteith 1.3, Jonathan A. Jenks", R. Terry Bowyer' 1 Department of Biological Sciences, Idaho State University, Pocatello, Idaho, USA, 2 Colorado Division of Parks and Wildlife, Grand Junction, Colorado, USA, 3 Wyoming Cooperative Fish and Wildlife Research Unit, University of Wyoming, Laramie, Wyoming, USA,4 Department of Natural Resource Management, South Dakota State University, Brookings, South Dakota, USA Citation: Lendrum, P. E., C.R. Anderson Jr., K. L. Monteith, J. A Jenks, R. T. Bowyer. 2013. Migrating Mule Deer: Effects of anthropogcnically Altered Landscapes. PloS ONE 8(5): e64548. DOI: I0.1371/journal.pone.0064548 Abstract Background: Migration is an adaptive strategy that enables animals to enhance resource availability and reduce risk of predation at a broad geographic scale. Ungulate migrations generally occur along traditional routes, many of which have been disrupted by anthropogenic disturbances. Spring migration in ungulates is of particular importance for conservation planning, because it is closely coupled with timing of parturition. The degree to which oil and gas development affects migratory patterns, and whether ungulate migration is sufficiently plastic to compensate for such changes, warrants additional study to better understand this critical conservation issue. Methodology/Principal Findings: We studied timing and synchrony of departure from winter range and arrival to summer range offemale mule deer (Odocoileus hemionus) in northwestern Colorado, USA, which has one of the largest natural-gas reserves currently under development in North America. We hypothesized that in addition to local weather, plant phenology, and individual life-history characteristics, patterns of spring migration would be modified by disturbances associated with natural-gas extraction. We captured 205 adult female mule deer, equipped them with GP$ collars, and observed patterns of spring migration during 2008-20 I0. Concl11sions/Signijicance: Timing of spring migration was related to winter weather (particularly snow depth) and access to emerging vegetation, which varied among years, but was highly synchronous across study areas within years. Additionally, timing of migration was influenced by the collective effects of anthropogenic disturbance, rate of travel, distance traveled, and body condition of adult females. Rates of travel were more rapid over shorter migration distances in areas of high natural-gas development resulting in the delayed departure, but early arrival for females migrating in areas with high development compared with less-developed areas. Such shifts in behavior could have consequences for timing of arrival on birthing areas, especially where mule deer migrate over longer distances or for greater durations. Practical guidance on characterizing availability in resource selection functions under a use-availability design JOSEPH M. NORTHRUP 1, MEVIN B. HOOTEN 1,2,3, CHARLES R. ANDERSON JR.4, AND GEORGE WITTEMYER1 1 Department of Fish, Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 2 U.S. Geological Survey, Colorado Cooperative Fish and Wildlife Research Unit, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 3 Colorado State University, Department of Statistics, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 4 Mammals Research Section Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction, Colorado 81505 USA Citation: Northrup, J.M., M. B. Hooten, C.R. Anderson Jr., and G. Wittemyer. 2013. Practical guidance on characterizing availability in resource selection functions under a use-availability design. Ecology 94(7): 1456-1463. http://dx.doi.org/ 10.1890/12-1688.l Abstract. Habitat selection is a fundamental aspect of animal ecology, the understanding of which is critical to management and conservation. Global positioning system data from animals allow fine-scale assessments of habitat selection and typically are analyzed in a use-availability framework, whereby animal locations are contrasted with random locations (the availability sample). Although most use-availability methods are in fact spatial point process models, they often are fit using logistic regression. This framework offers numerous methodological challenges, for which the literature provides little guidance. Specifically, the size and spatial extent of the availability sample influences coefficient estimates potentially causing interpretational bias. We examined the influence of availability on statistical inference through simulations and analysis of serially correlated mule deer GPS data. Bias in estimates arose from incorrectly assessing and sampling the spatial extent of availability. Spatial autocorrelation in covariates, which is common for landscape characteristics, exacerbated the error in availability sampling leading to increased bias. These results have strong implications for habitat selection analyses using GPS data, which are increasingly prevalent in the literature. We recommend that researchers assess the sensitivity of their results to their availability sample and, where bias is likely, take care with interpretations and use cross validation to assess robustness. 11 �Effects of Helicopter Capture and Handling on Movement Behavior of Mule Deer JOSEPH M. NORTHRUP', CHARLES R. ANDERSON JRZ, AND GEORGE WITTEMYER 1 'Department offish, Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 2 Mammals Research Section Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction, Colorado 81505 USA Citation: Northrup, J.M., C. R. Anderson Jr., and G. Wittemyer. 2014. Effects of helicopter capture and handling on movement behavior of mule deer. Journal of Wildlife Management 78(4):731-738; DOI: 10.1002/jwmg.705 ABSTRACT Research on wildlife movement, physiology, and reproductive biology often requires capture and handling of animals. Such invasive treatment can alter behavior, which may bias results or invalidate assumptions regarding representative behaviors. To assess the impacts of handling on mule deer (Odocoileus hemionus), a focal species for research in North America, we investigated pre- and post-recapture movements of collared individuals, and compared them to deer that were not recaptured (controls). We compared pre- and post-recapture movement rates (m/hr) and 24-hour straight-line displacement among recaptured and control deer. In addition, we examined the time it took recaptured deer to return to their pre-recapture home range. Both daily straight-line displacement and movement rate were marginally elevated relative to monthly averages for 24 hours following recapture, with non-significant elevation continuing for up to 7 days. Comparing movements averaged over 30 days before and after recapture, we found no differences in displacement, but movement rates demonstrated seasonal effects, with faster movements post- relative to pre-recapture in March and slower movements post- relative to pre-recapture in December. Relative to control deer movements, recaptured deer movement rates in March were higher immediately after recapture and lower in the second and third weeks following recapture. The median time to return to the pre-recapture home range was 13 hours, with 71 % of deer returning in the first day, and 91 % returning within 4 days. These results indicate a short period of elevated movements following recaptures, likely due to the deer returning to their home ranges, followed by weaker but non-significant depression of movements for up to 3 weeks. Censoring of the first day of data post capture from analyses is strongly supported, and removing additional days until the individual returns to its home range will control for the majority ofimpacts from capture. © 2014 The Wildlife Society. Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns of forage quality Patrick E. Lendrum■, Charles R. Anderson Jr. b' Kevin L. MontelthC, Jonathan A. Jenksd, R. Terry Bowyer• Department of Biological Sciences, Idaho State University, 921 South 8th Avenue, Stop 8007, Pocatello 83209, USA b Mammals Research Section Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction 81505, USA c Wyoming Cooperative Fish and Wildlife Research Unit, Department of Zoology and Physiology, University of Wyoming, 3166, 1000 East University Avenue, Laramie 82071, USA .s Department of Natural Resource Management, South Dakota State University, Box 2140B, Brookings 57007, USA u Citation: Lendrum, P. E., C.R. Anderson Jr., K. L Monteith, J. A Jenks, and R. T. Bowyer. 2014. Relating the movement of a rapidly migrating ungulate to spatiotemporal patterns of forage quality. Mammalian Biology: http://dx.doi.ony'l0.1016/j.mambio.2014.05.005 ABSTRACT: Migratory ungulates exhibit recurring movements, often along traditional routes between seasonal ranges each spring and autumn, which allow them to track resources as they become available on the landscape. We examined the relationship between spring migration of mule deer (Odocoileus hemionus) and forage quality, as indexed by spatiotemporal patterns of fecal nitrogen and remotely sensed greenness of vegetation (Normalized Difference Vegetation Index; NOVI) in spring 20 l Oin the Piceance Basin of northwestern Colorado, USA. NOVI increased throughout spring, and was affected primarily by snow depth when snow was present, and temperature when snow was absent. Fecal nitrogen was lowest when deer were on winter range before migration, increased rapidly to an asymptote during migration, and remained relatively high when deer reached summer range. Values offecal nitrogen corresponded with increasing NOVI during migration. Spring migration for mule deer provided a way for these large mammals to increase access to a high-quality diet, which was evident in patterns of NOVI and fecal nitrogen. Moreover, these deer 'jumped" rather than "surfed" the green wave by arriving on summer range well before peak productivity of forage occurred. This rapid migration may aid in securing resources and seclusion from others on summer range in preparation for parturition, and to minimize detrimental factors such as predation and malnutrition during migration. 12 �\..,_I Effects of Male-Biased Harvest on Mule Deer: Implications for Rates of Pregnancy, Synchrony, and Timing of Parturition ERIC D. FREEMAN 1, RANDY T. LARSEN 1, MARKE. PETERSON2, CHARLES R. ANDERSON JR.3, KENT R. HERSEY\ AND BROCK R. McMILLAN 1 1 Department of Plant and Wildlife Sciences, Brigham Young University, 275 WIDB, Provo, UT 84602, USA 2 Department offish, Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA 3 Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction, CO 81505, USA 4 Utah Division of Wildlife Resources, 1594 WNorth Temple, Salt Lake City, UT 84114, USA Citation: Freeman, E. D., R. T. Larsen, M. E. Peterson, C.R. Anderson Jr., K. R. Hersey, and B. R. McMillan. 2014. Effects of male-biased harvest on mule deer: implications for rates of pregnancy, synchrony, and timing of parturition. Wildlifo Society Bulletin; DOI: I0.1002/wsb.450 ABSTRACT Evaluating how management practices influence the population dynamics of ungulates may enhance future management of these species. For example, in mule deer (Odocoileus hemionus), changes in male/female ratio due to male- biased harvest may alter rates of pregnancy, timing of parturition, and synchrony of parturition if inadequate numbers of males are present to fertilize females during their first estrous cycle. If rates of pregnancy or parturition are influenced by decreased male/female ratios, recruitment may be reduced (e.g., fewer births, later parturition resulting in lower survival of fawns, and a less synchronous parturition that potentially increases susceptibility of neonates to predation). Our objectives were to compare rates of pregnancy, synchrony of parturition, and timing of parturition between exploited mule deer populations with a relatively high (Piceance, CO, USA; 26 males/100 females) and a relatively low (Monroe, UT, USA; 14 males/100 females) male/female ratio. We determined rates of pregnancy via ultrasonography and timing of parturition via vaginal implant transmitters. We found no differences in rates of pregnancy (98.6% and 96.6%; z = 0.821; P = 0.794), timing of parturition (estimate= 1.258; SE= 1.672; I = 0. 752; P = 0.454), or synchrony of parturition (F = 1.073; P = 0.859) between Monroe Mountain and Piceance Basin, respectively. The relatively low male/female ratio on Monroe Mountain was not associated with a protracted period of parturition. This finding suggests that relatively low male/female ratios typical of heavily harvested populations do not influence population dynamics because recruitment remains unaffected.© 2014 The Wildlife Society. Fine-scale genetic correlates to condition and migration in a wild cervid Joseph M. Northrup•, Aaron B. A. Shafe.-%, Charles R. Anderson Jr.3, David W. Coltman', and George Wlttemyer 1 I Department offish, Wildlife, and Conservation Biology, Colorado State University, Fort Collins, CO, USA 2 Department of Evolutionary Biology, Evolutionary Biology Centre, Uppsala University, Uppsala, Sweden 3 Mammals Research Section, Colorado Palks and Wildlife, Grand Junction, CO, USA 4 Department of Biological Sciences, University of Alberta, Edmonton, AB, Canada. Citation: Northrup, J.M., AB. Shafer, C.R. Anderson Jr., D. W. Coltman, and G. Whittemyer. 2014. Fine-scale genetic correlates to condition and migration in a wild cervid. Evolutionary Applications ISSN 1752-457 l; doi: I0.1111/eva 12189 Abstract The relationship between genetic variation and phenotypic traits is fundamental to the study and management of natural populations. Such relationships often are investigated by assessing correlations between phenotypic traits and heterozygosity or genetic differentiation. Using an extensive data set compiled from free ranging mule deer (Odocoileus hemionus), we combined genetic and ecological data to (i) examine correlations between genetic differentiation and migration timing, (ii) screen for mitochondrial haplotypes associated with migration timing, and (iii) test whether nuclear beterozygosity was associated with condition. Migration was related to genetic differentiation (more closely related individuals migrated closer in time) and mitochondrial baplogroup. Body fat was related to heterozygosity at two nuclear loci (with antagonistic patterns), one of which is situated near a known fat metabolism gene in mammals. Despite being focused on a widespread panmictic species, these findings revealed a link between genetic variation and important phenotypes at a fine scale. We hypothesize that these correlations are either the result of mixing refugial lineages or differential mitochondrial haplotypes influencing energetics. The maintenance of phenotypic diversity will be critical to enable the potential tracking of changing climatic conditions, and these correlates highlight the need to consider evolutionary mechanisms in management, even in widely distributed panmictic species. 13 �Landscape and anthropogenic features influence the use of auditory vigilance by mule deer Emma Lynch•, Joseph M. Northrupb, Megan F. MeKennaC, Charles R. Anderson Jr.d, Lisa Angeloni¥, and George Wlttemyer-,b •Graduate Degree Program in Ecology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA bDepartmcnt offish, Wildlife and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA •Natural Sounds and Night Skies Division, National Park Service, 1201 Oakridge Drive, Fort Collins, CO 80525, USA, dMammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526, USA 9)cpartmcnt ofBiology, Colorado State University, 1878 Campus Delivery, Fort Collins, CO 80523, USA Citation: Lynch, E., J.M. Northrup, M. F. McKenna, C.R. Anderson Jr., L. Angeloni, and G. Wittcmycr. 2014. Landscape and anthropogenic features influence the use of auditory vigilance by mule deer. Behavioral Ecology; doi: I0.1093/beheco/aru I58. While visual forms of vigilance behavior and their relationship with predation risk have been broadly examined, animals also employ other vigilance modalities such as auditory vigilance by listening for the acoustic cues of predators. Similar to the tradeoffs associated with visual vigilance, auditory behavior potentially structures the energy budgets and behavior of animals. The cryptic nature of auditory vigilance makes it difficult to study, but on-animal acoustical monitoring has rapidly advanced our ability to investigate behaviors and conditions related to sound. We utilized this technique to investigate the ways external stimuli in an active natural gas development field affect periodic pausing by mule deer (Odocoileus hemionus) within bouts of rumination-based mastication. To better understand the ecological properties that structure this behavior, we investigate spatial and temporal factors related to these pauses to determine if results are consistent with our hypothesis that pausing is used for auditory vigilance. We found that deer paused more when in forested cover and at night, where visual vigilance was likely to be less effective. Additionally, deer paused more in areas of moderate background sound levels, though responses to anthropogenic features were less clear. Our results suggest that pauses during rumination represent a fonn of auditory vigilance that is responsive to landscape variables. Further exploration of this behavior can facilitate a more holistic understanding of risk perception and the costs associated with vigilance behavior. Migration Patterns of Adult Female Mule Deer in Response to Energy Development Charles R. Anderson Jr. and Chad J. Bishop Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526, USA Citation: Anderson, C.R., Jr., and C. J. Bishop. 2014. Migration patterns of adult female mule deer in response to energy development. Pages 47-50 in Transactions of the 79rJt North American Wildlife & Natural Resources Conference (R. A. Coon & M. C. Dunfee, eds.). Wildlife Management Institute, Gardners, PA, USA. ISSN 0078-1355. Migration is an adaptive strategy that enables animals to enhance resource availability and reduce risk of predation at a broad geographic scale. Ungulate migrations generally occur along traditional routes, many of which have been disrupted by anthropogenic disturbances. Spring migration in ungulates is of particular importance for conservation planning because it is closely coupled with timing of parturition. The degree to which oil and gas development affects migratory patterns, and whether ungulate migration is sufficiently prepared to compensate for such changes, has recently been investigated in Colorado and Wyoming (Lendrum et al. 2012, 2013; Sawyer et al. 2012). Lendrum et al. (2012, 2013) and Sawyer et al. (2012) address mule deer (Odocoileus hemionus) migration patterns in relation to energy development from northwest Colorado and south-central Wyoming, respectively. We address results from the Colorado and Wyoming studies and then compare similarities and differences. The interactions between migratory mule deer and energy development identified by Lendrum et al. (2012, 2013) and Sawyer et al. (2012) suggest mule deer may benefit from energy development planning by considering thresholds of development that may alter migratory behavior. It appears that migration rate, migration routes, and stopover use, if present, may be altered at high development intensities. In addition, migratory mule deer may benefit by maintaining security cover along migration paths, and improved habitat conditions may facilitate more direct and rapid migration requiring less energy to complete migration. Enhancing permeability along migration routes by applying dispersed development plans (<2 well pads/krn2) and minimizing disturbance to vegetation types by maintaining security cover should reduce impacts to migratory mule deer as well as other migratory ungulates. Where feasible, habitat improvement projects on winter range and possibly stopover sites would also enhance migratory mule deer populations by enhancing energy reserves for long-distance movements and parturition shortly after summer range arrival. Where possible, directional drilling could be used to extract energy resources from underneath migration routes while maintaining no surface occupancy. Lastly, we emphasize that OPS studies now allow managers to accurately map migration routes for entire populations and identify relatively narrow corridors that are most heavily used thus allowing for the identification of the most important corridors for migrating ungulates. Where available, we encourage agencies to incorporate such migration corridors into land-use plans (e.g., resource management plans) and National Environmental Policy Act documents. 14 �Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore Kate R. Searle1 • Mindy B. Rlce2 • Charles R. Anderson2 • Chad Bishop2 • N. T. Hobbs3 1 NERC Centre for Ecology and Hydrology, Bush Estate, Penicuik EH26 OQB, UK 2 Colorado Parlcs and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526, USA 3 Department of Ecosystem Science and Sustainability, Colorado State University, Fort Collins 80524, CO, USA Citation: Searle, K. R., M. B. Rice, C.R. Anderson, C. Bishop and N. T. Hobbs. 2015. Asynchronous vegetation phcnology enhances winter body condition of a large mobile herbivore. Occologia ISSN 0029-8549; DOI I0.1007/s00442-015-3348-9 Abstract Understanding how spatial and temporal heterogeneity influence ecological processes forms a central challenge in ecology. Individual responses to heterogeneity shape population dynamics, therefore understanding these responses is central to sustainable population management. Emerging evidence has shown that herbivores track heterogeneity in nutritional quality of vegetation by responding to phenological differences in plants. We quantified the benefits mule deer (Odocoi/eus hemionus) accrue from accessing habitats with asynchronous plant phenology in northwest Colorado over 3 years. Our analysis examined both the direct physiological and indirect environmental effects of weather and vegetation phenology on mule deer winter body condition. We identified several important effects of annual weather patterns and topographical variables on vegetation phenology in the home ranges of mule deer. Cmcially, temporal patterns of vegetation phenology were linked with differences in body condition, with deer tending to show poorer body condition in areas with less asynchronous vegetation green-up and later vegetation onset. The direct physiological effect of previous winter precipitation on mule deer body condition was much less important than the indirect effect mediated by vegetation phenology. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer JOSEPH M. NORTHRUP 1, CHARLES R. ANDERSON JR. 2, and GEORGE WITTEMYER 1• 3 'Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, CO, USA Mammals Research Section, Colorado Parks and Wildlife, Fort Collins, CO, USA 3 Graduate Degree Program in Ecology, Colorado State University, Fort Collins, CO, USA 2 Citation: Northrup, J. M., C. R. Anderson, Jr., and G. Wittemyer. 2015. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer. Global Change Biology, doi: IO.I I I l/gcb.13037 Abstract Extraction of oil and natural gas (hydrocarbons) from shale is increasing rapidly in North America, with documented impacts to native species and ecosystems. With shale oil and gas resources on nearly every continent, this development is set to become a major driver of global land-use change. It is increasingly critical to quantify spatial habitat loss driven by this development to implement effective mitigation strategies and develop habitat offsets. Habitat selection is a fundamental ecological process, influencing both individual fitness and population-level distribution on the landscape. Examinations of habitat selection provide a natural means for understanding spatial impacts. We examined the impact of natural gas development on habitat selection patterns of mule deer on their winter range in Colorado. We fit resource selection functions in a Bayesian hierarchical framework, with habitat availability defined using a movement-based modeling approach. Energy development drove considerable alterations to deer habitat selection patterns, with the most substantial impacts manifested as avoidance of well pads with active drilling to a distance of at least 800 m. Deer displayed more nuanced responses to other infrastructure, avoiding pads with active production and roads to a greater degree during the day than night. In aggregate, these responses equate to alteration of behavior by human development in over 50% of the critical winter range in our study area during the day and over 25% at night. Compared to other regions, the topographic and vegetative diversity in the study area appear to provide refugia that allow deer to behaviorally mediate some of the impacts of development. This study, and the methods we employed, provides a template for quantifying spatial take by industrial activities in natural areas and the results offer guidance for policy makers, mangers, and industry when attempting to mitigate habitat loss due to energy development. 15 �Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid Joseph M. Northrup', Charles R. Anderson Jr, and George Wittemyer•.J 1Department offish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, CO, USA 2 Mammals Research Section, Colorado Parks and Wildlife, Fort Collins, CO, USA 3Graduate Degree Program in Ecology, Colorado State University, Fort Collins, CO, USA Citation: Northrup, J.M., C.R. Anderson, Jr., and G. Wittemyer. 2016. Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid Diversity and Distributions 22: 547-557, DOI: 10.1111/ddi.12417 ABSTRACT Aim The space an animal uses over a given time period must provide the resources required for meeting energetic needs, reproducing and avoiding predation. Anthropogenic landscape change in concert with environmental dynamics can strongly structure space-use. Investigating these dynamics can provide critical insight into animal ecology, conservation and management. Location The Piceance Basin, Colorado, USA. Methods We applied a novel utilization distribution estimation technique based on a continuous-time correlated random walk model to characterize range dynamics of mule deer during winter and summer seasons across multiple years. This approach leverages secondorder properties of movement to provide a probabilistic estimate of space-use. We assessed the influence of environmental (cover and forage), individual and anthropogenic factors on interannual variation in range use of individual deer using a hierarchical Bayesian regression framework. Results Mule deer demonstrated remarkable spatial philopatry, with a median of 50% overlap (range: 8-78%) in year-to-year utilization distributions. Environmental conditions were the primary driver of both philopatry and range size, with anthropogenic disturbance playing a secondary role. Main conclusions Philopatry in mule deer is suspected to reflect the importance of spatial familiarity (memory) to this species and, therefore, factors driving spatial displacement are of conservation concern. The interaction between range behaviour and dynamics in development disturbance and environmental conditions highlights mechanisms by which anthropogenic environmental change may displace deer from familiar areas and alter their foraging and survival strategies. Movement reveals scale dependence in habitat selection of a large ungulate Joseph M. Northrup 1, Charles R. Anderson Jr. 2, Mevin 8. Hooten 3, and George Wittemyer4 'Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA Mammals Research Section, Colorado Parks and Wildlife, Fort Collins, Colorado 80523 USA 3 U.S. Geological Survey, Colorado Cooperative Fish and Wildlife Research Unit, Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA 4 Department offish, Wildlife and Conservation Biology and Graduate Degree Program in Ecology, Colorado State University, Fort Collins, Colorado 80523 USA 2 Citation: Northrup, J.M., C.R. Anderson, Jr., M. B. Hooten, and G. Wittemyer. 2016. Movement reveals scale dependence in habitat selection ofa large ungulate. Ecological Applications 26:2746-2757 Abstract. Ecological processes operate across temporal and spatial scales. Anthropogenic disturbances impact these processes, but examinations of scale dependence in impacts are infrequent. Such examinations can provide important insight to wildlife-human interactions and guide management efforts to reduce impacts. We assessed spatiotemporal scale dependence in habitat selection of mule deer (Odocoileus hemionus) in the Piceance Basin of Colorado, USA, an area of ongoing natural gas development. We employed a newly developed animal movement method to assess habitat selection across scales defined using animal-centric spatiotemporal definitions ranging from the local (defined from five hour movements) to the broad (defined from weekly movements). We extended our analysis to examine variation in scale dependence between night and day and assess functional responses in habitat selection patterns relative to the density of anthropogenic features. Mule deer displayed scale invariance in the direction of their response to energy development features, avoiding well pads and the areas closest to roads at all scales, though with increasing strength of avoidance at coarser scales. Deer displayed scale-dependent responses to most other habitat features, including land cover type and habitat edges. Selection differed between night and day at the finest scales, but homogenized as scale increased. Deer displayed functional responses to development, with deer inhabiting the least developed ranges more strongly avoiding development relative to those with more development in their ranges. Energy development was a primary driver of habitat selection patterns in mule deer, structuring their behaviors across all scales examined. Stronger avoidance at coarser scales suggests that deer behaviorally mediated their interaction with development, but only to a degree. At higher development densities than seen in this area, such mediation may not be possible and thus maintenance of sufficient habitat with lower development densities will be a critical best management practice as development expands globally. 16 �~ Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus) Kourtney F. Stonehouse•.i, Charles R. Anderson Jr. 1, Mark E. Peterson•.i, and David R. Collins• 1Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 90526 USA 2 Department offish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA Citation: Stonehouse, K. F., C.R. Anderson Jr., M. E. Peterson, and D.R. Collins. 2016. Approaches to field investigations ofcausc-spccifie mortality in mule deer (Odocoi/eus hemionus). Colorado Parks and Wildlife Technical Report No. 48, First Edition, 317 W. Prospect Rd., Ft. Collins, CO USA. DOW-R-T-48-16, ISSN 0084-8883. This technical report provides general guidelines for conducting mortality site investigations to help investigators distinguish predation from scavenging and other causes of death. General health indices are also provided to assess whether or not deer may have died from malnutrition or disease or if these factors may have predisposed deer to predation. Lastly. these guidelines will assist investigators in identifying predatory species or scavengers involved through the examination of physical evidence at deer mortality sites. The information presented here is based primarily on field experience gained from a long term research effort in northwest Colorado investigating mule deer mortality sites over several years (http://cpw.state.co.us/learn/Pages/ResearchMammalsRP-04.aspx) and literature review where referenced. We acknowledge that proximate and ultimate cause of death can be difficult or impossible to detect from field necropsy alone and examples presented here largely represent proximate causes of mortality; efforts discerning ultimate cause will require specific tissue sample collections, where possible, submitted to a veterinary diagnostic laboratory. Within this technical report are numerous photographs documenting characteristics of predator attacks on mule deer and signs left by predatory and scavenging species. Additional pictures illustrate differences between healthy and unhealthy tissues and organs. While reading this document, be aware that each mortality investigation is unique and observations in the field may differ from illustrations provided here. Appendix I provides a sample necropsy form to assist in conducting mortality investigations. Reproductive success of mule deer in a natural gas development area Mark E. Peterson•, Charles R. Anderson Jr.2, Joseph M. Northrup 1,and Paul F. Doherty Jr.' 'Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA 2 Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospi:ct Road, Fort Collins, CO 90526 USA \..-I Citation: Peterson, M. E., C.R. Anderson Jr., J.M. Northrup, and P. F. Doherty Jr. 2017. Reproductive success of mule deer in a natural gas development area. Wildlife Biology doi: I0.1 I I l/wlb.00341 Abstract: Natural gas development is increasing across North America and causing concern over the potential impacts on wildlife populations and their habitat, particularly for ungulate species. Understanding how this development impacts reproductive success metrics that are influential for ungulate population dynamics is important to guide management of ungulates. However, the influences of natural gas development on reproductive success metrics of mule deer Odocoileus hemionus have not been studied. We used statistical models to examine the influence of natural gas development and temporal factors on reproductive success metrics of mule deer in the Piceance Basin, northwest Colorado during 2012-2014. We focused on study areas with relatively high or low levels of natural gas development. Pregnancy and in utero fetal rates were high and statistically indistinguishable between study areas. Fetal survival rates increased over time and survival was lower in the high versus low development study areas in 2012 possibly influenced by drought coupled with habitat loss and fragmentation associated with development. Our novel results suggest managers should be concerned with the influences of development on fetal survival, particularly during extreme environmental conditions (e.g. drought) and our results can be used to guide development planning and/or mitigation. Developers and wildlife managers should continue to collaborate on development planning, such as implementing habitat treatments to improve forage availability and quality, minimizing disturbance to hiding and foraging habitat particularly during parturition, and implementing directional drilling to minimize pad disturbance density to increase fetal survival in developed areas. 17 �Variation in ungulate body fat: individual versus temporal effects Eric J. Bergman 1, Charles R. Anderson Jr. 1, Chad J. Bishop', A. Andrew Holland', and Joseph M. Northrup 2 'Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 90526 USA 2 Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA Citation: Bergman, E. J., C.R. Anderson Jr., C. J. Bishop, A. A. Holland, and J.M. Northrup. 2018. Variation in ungulate body fat: individual versus temporal effects. Journal of Wildlife Management 82:130-137, DOI: 10:1002/jwmg.21334 ABSTRACT The use of ultrasonograhic measurements of muscle and body fat represent a relatively new data stream that can be used to address questions regarding tmgulate condition. We have learned that measurements of body fat and presumably overall body condition among individual animals, even those taken from the same herd at that same time, are highly variable. Relatively little consideration has been given to the sources of variation in body fat and other physiological parameters in wildlife populations. We evaluated the components of variation in late-winter mule deer (Odocoileus hemionus) body fat estimates: sampling variation (i.e., variation induced by the particular set of individuals that were sampled) and process variation (i.e., variation stemming from biological processes) with a long-term data set (2002-20 IS) from Colorado, USA. We collected our data from across Colorado as part of historical research, ongoing research, and periodic population monitoring programs. Mean percent ingesta-free body fat (%IFBF) for sampled mule deer was 7.20 ± 1.20% (SO). Covariates related to individual deer explained approximately 4% of the total variation in %IFBF and annual effects explained an additional 13% of the variation. Substantial residual variation in %IFBF (83%) remained unexplained. The source of the 83% of unexplained variation is partially linked to fine-scale spatial dynamics but also additional individual metrics we were unable to capture, primarily the presence or absence of dependent young. We speculate that the primary factors influencing late-winter mule deer body fat and overall condition are individual in nature. These results present a cautionary check on herd level inference that can be made from individual late-winter body fat estimates and we postulate that for mule deer, alternative and additional body condition metrics may offer added utility in management scenarios. However, an important next step to better understand wildlife population health is to evaluate the sources and magnitude of variation within other body condition metrics, with the goal of further refining data that can better allow biologists to incorporate herd health into population management recommendations. Mortality of mule deer fawns in a natural gas development area Mark E. Peterson•, Charles R. Anderson Jr.2,Joseph M. Northrup 1,and Paul F. Doherty Jr.' 'Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA 2Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 90526 USA Citation: Peterson, M. E., C. R. Anderson Jr., J. M. Northrup, and P. F. Doherty Jr. 20 I8. Mortality of mule deer fawns in a natural gas development area. Journal of Wildlife Management 82:1135-1148, DOI: 10.1002/jwmg.21476 ABSTRACT Recent natural gas development has caused concern among wildlife managers, researchers, and stakeholders over the potential effects on wildlife and their habitats. Specifically, understanding how this development and other factors influence mule deer (Odocoileus hemionus) fawn (i.e., 0-6 months old) mortality rates, recruitment, and subsequently population dynamics have been identified as knowledge gaps. Thus, we tested predictions concerning the relationship between natural gas development, adult female, fawn birth, and temporal (weather) characteristics on fawn mortality in the Piceance Basin of northwestern Colorado, USA, from 2012-2014.We captured and radio-collared 184 fawns and estimated apparent cause-specific mortality in areas with relatively high or low levels of natural gas development using a multi-state model. Mean daily predation probability was similar in the high versus low development areas. Predation was the leading cause of fawn mortality in both areas and decreased from 0-14 days old. Black bear (Ursus americanus; 22% of all mortalities, n = 17) and cougar (Felis conco/or; 36% of all mortalities, n =6) predation was the leading cause of mortality in the high and low development areas, respectively. Predation of fawns was negatively correlated with the distance from a female's core area to a producing well pad on winter or summer range. Contrary to expectations, predation of fawns was positively correlated with rump fat thickness of adult females. Well pad densities and development activity were relatively low during our study, indicating that the observed intensity of development did not appear to influence daily predation probability. Our results suggest maintaining development activity thresholds at levels we observed to potentially minimize the effects of development on fawn mortality. However, we caution that higher development intensity and drilling activity in flatter, less rugged areas with less concealment cover could influence fawn mortality. Managers should maintain low development densities in areas where topography and vegetation offer less concealment. Overall, region-specific data (e.g., development intensity, topography, predator assemblages, and associated predation risk) are needed to better understand the effects of natural gas development on fawn mortality. 18 �,__., Using maternal mule deer movements to estimate timing of parturition and assist fawn captures Mark E. Peterson•, Charles R. Anderson Jr.2, Mathew W. Alldredge2,and Paul F. Doherty Jr.' 1Departmcnt of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA 2Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 90526 USA Citation: Peterson, M. E., C.R. Anderson Jr., M. W. Alldredge, and P. F. Doherty Jr. 2018. Using maternal mule deer movements to estimate timing of parturition and assist fawn captures. Wildlife Society Bulletin 42:616-621; DOI: I0.1002/wsb.935 ABSTRACT Movement patterns of maternal ungulates have been used to determine parturition dates and aid in locating fawns, which may be important for understanding reproductive rates (e.g., pregnancy and fetal), but such methods have not been validated for mule deer (Odocoileus hemionus). We first determined timing of parturition using vaginal implant transmitters (VITs) and then predicted timing of parturition using VITs in conjunction with Global Positioning System collar data in the Piceance Basin of northwestern Colorado, USA, during 2012-2014. We examined daily movement rate to determine differences in movement rate among days (7 days pre- and postpartum) and for movement patterns indicative of parturition. Mean daily movement rate (m/day) of 102 maternal deer decreased by 46% from 1 day preparturition (mean = 1,253, SD= 1,091) to parturition date (mean = 682, S = 574), and remained at this low rate 1-7 days postpartum. We applied an independent data set to validate predicted parturition dates based on daily movement rate. We estimated day of parturition correctly (i.e., day 0), within 1-3 days postparturition, and_4 days postparturition offield-reported dates for IO (29%), 21 (60%), and 4 (11%) maternal females, respectively. For novel data sets, we predict that a mule deer female whose daily movement rate decreases by_46% and remains low _3 days postparturition particularly when preceded by a sudden increase in movement-has given birth. However, we caution that disturbance of deer by field crews should be minimized, and if birth sites are not found, neonatal mortality will be underestimated. Our results can help determine timing and general location of parturition as an aid in capturing fawns when the use of VITs is not feasible, with the ultimate objective of estimating pregnancy, fetal, and fawn survival rates if birth sites are found. On-animal acoustic monitoring provides insight to ungulate foraging behavior Joseph M. Northrup•, Alexandra Avrin 1, Charles R. Anderson, Jr. 2, Emma Brown3, and George Wittemyer 1 'Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA 2 Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 90526 USA 3National Parle Service Natural Sounds and Night Skies Division, Fort Collins, CO 80525 USA Citation: Northrup, J.M., A. Avrin, C.R. Anderson Jr., E. Brown, and G. Wittemyer. 2019. On-animal acoustic monitoring provides insight to ungulate foraging behavior. Journal ofMammalogy 100:1479-1489; https://doi.org/10.1093/jmammaVgyzl24 Abstract Foraging behavior underpins many ecological processes; however, robust assessments of this behavior for free-ranging animals are rare due to limitations to direct observations. We leveraged acoustic monitoring and GPS tracking to assess the factors influencing foraging behavior of mule deer (Odocoileus hemionus). We deployed custom-built acoustic collars with GPS radiocollars on mule deer to measure location-specific foraging. We quantified individual bites and steps taken by deer, and quantified two metrics of foraging behavior: the number of bites taken per step and the number of bites taken per unit time, which relate to foraging intensity and efficiency. We fit statistical models to these metrics to examine the individual, environmental, and anthropogenic factors influencing foraging. Deer in poorer body condition took more bites per step and per minute and foraged for longer irrespective of landscape properties. Other patterns varied seasonally with major changes in deer condition. In December, when deer were in better condition, they took fewer bites per step and more bites per minute. Deer also foraged more intensely and efficiently in areas of greater forage availability and greater movement costs. During March, when deer were in poorer condition, foraging was not influenced by landscape features. Anthropogenic factors weakly structured foraging behavior in December with no relationship in March. Most research on animal foraging is interpreted under the framework of optimal foraging theory. Departures from predictions developed under this framework provide insight to unrecognized factors influencing the evolution of foraging. Our results only conformed to our predictions when deer were in better condition and ecological conditions were declining, suggesting foraging strategies were state-dependent. These results advance our understanding of foraging patterns in wild animals and highlight novel observational approaches for studying animal behavior. 19 �A noninvasive automated device for remotely collaring and weighing mule deer Chad J. Bishop 1, Mathew W. Alldredge 1, Daniel P. Walsh 1, Eric J. Bergman 1, Charles R. Anderson Jr. 1, Darlene Kilpatrick, Joe Bakel2, and Christophe Fabvre1 1 Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526 USA 2 Dynamic Group Circuit Design, Inc., 2629 Redwing Road, Fort Collins, CO 80525 USA Citation: Bishop, C. J., M. W. Alldredge, D. P. Walsh, E. J. Bergman, C.R. Anderson Jr., D. Kilpatrick, J. Bake!, and C. Fabvre. 2019. A noninvasive automated device for remotely collaring and weighing mule deer. Wildlife Society Bulletin 43:717-725; doi.org/ 10.1002/wsb. I034 ABSTRACT Wildlife biologists capture deer (Odocoileus spp.) annually to attach transmitters and collect basic information (e.g., animal mass and sex) as part of ongoing research and monitoring activities. Traditional capture techniques induce stress in animals and can be expensive, inefficient, and dangerous. They are also impractical for some urbanized settings. We designed and evaluated a device for mule deer (0. hemionus) that automatically attached an expandable radiocollar to a ~6-month-old fawn and recorded the fawn's mass and sex, without physically restraining the animal. The device did not require on-site human presence to operate. Students and faculty in the Mechanical Engineering Department at Colorado State University produced a conceptual model and early prototype. Professional engineers at Dynamic Group Circuit Design, Inc. in Fort Collins, Colorado, USA, produced a fully functional prototype of the device. Using the device, we remotely collared, weighed, and identified sex of 8 free-ranging mule deer fawns during winters 2010-2011 and 2011-2012. Collars were modified to shed from deer approximately I month after the collaring event. Two fawns were successfully recollared after they shed the first collars they received. Thus, we observed IO successful collaring events involving 8 unique fawns. Fawns demonstrated minimal response to collaring events, either remaining in the device or calmly exiting. A fawn typically required ~l weeks of daily exposure before fully entering the device and extending its head through the outstretched collar, which was necessary for a collaring event to occur. This slow acclimation period limited utility of the device when compared with traditional capture techniques. Future work should focus on device modifications and altered baiting strategies that decrease fawn acclimation period, and in tum, increase collaring rates, providing a noninvasive and perhaps cost-effective alternative for monitoring mid- to large-sized mammal species. © 2019 The Wildlife Society. Behavioral and demographic responses of mule deer to energy development on winter range Joseph M. Northrup 1,J. M., Charles R. Anderson Jr. 2, Brian D. Gerber, and George Wittemyer 1 1Department of Fish, Wildlife and Conservation Biology, Colorado State University. 1474 Campus Delivery, Fort Collins, CO 80523 USA 2Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526 USA 3Departrnent of Natural Resources Science, University of Rhode Island. 1 Greenhouse Road, Kingston, RI 02881 USA Citation: Northrup, J.M., C.R. Anderson Jr., B. D. Gerber, and G. Wittemyer. 2021. Behavioral and demographic responses of mule deer to energy development on winter range. Wildlife Monographs 208:1-37; 2021; DOI: 10.1002/wmon.1060. ABSTRACT Anthropogenic habitat modification is a major driver of global biodiversity loss. In North America, one of the primary sources of habitat modification over the last 2 decades has been exploration for and production of oil and natural gas (hydrocaroon development), which has led to demographic and behavioral impacts to numerous wildlife species. Developing effective measures to mitigate these impacts has become a critical task for wildlife managers and conservation practitioners. However, this task has been hindered by the difficulties involved in identifying and isolating factors driving population responses. Current research on responses of wildlife to development predominantly quantifies behavior, but it is not always clear how these responses scale to demography and population dynamics. Concomitant assessments of behavior and population-level processes are needed to gain the mechanistic understanding required to develop effective mitigation approaches. We simultaneously assessed the demographic and behavioral responses of a mule deer population to natural gas development on winter range in the Piceance Basin of Colorado, USA, from 2008 to 2015. Notably, this was the period when development declined from high levels of active drilling to only production phase activity (i.e., no drilling). We focused our data collection on 2 contiguous mule deer winter range study areas that experienced starkly different levels of hydrocarbon development within the Piceance Basin. We assessed mule deer behavioral responses to a range of development features with varying levels of associated human activity by examining habitat selection patterns of nearly 400 individual adult female mule deer. Concurrently, we assessed the demographic and physiological effects of natural gas development by comparing annual adult female and overwinter fawn (6-monthold animals) survival, December fawn mass. adult female late and early winter body fat, age. pregnancy rates. fetal counts, and lactation rates in December between the 2 study areas. Strong differences in habitat selection between the 2 study areas were apparent. Deer in the less-developed study area avoided development during the day and night, and selected habitat presumed to be used for foraging. Deer in the heavily developed study area selected habitat presumed to be used for thermal and security cover to a greater degree. Deer faced with higher densities of development avoided areas with more well pads during the day and responded neutrally or 20 V �~ selected for these areas at night. Deer in both study areas showed a strong reduction in use of areas around well pads that were being drilled, which is the phase of energy development associated with the greatest amount of human presence, vehicle traffic, noise, and artificial light. Despite divergent habitat selection patterns, we found no effects of development on individual condition or reproduction and found no differences in any of the physiological or vital rate parameters measured at the population level. However, deer density and annual increases in density were higher in the low-development area. Thus, the recorded behavioral alterations did not appear to be associated with demographic or physiological costs measured at the individual level, possibly because populations are below winter range carrying capacity. Differences in population density between the 2 areas may be a result of a population decline prior to our study (when development was initiated) or area-specific differences in habitat quality, juvenile dispersal, or neonatal or juvenile survival; however, we lack the required data to contrast evidence for these mechanisms. Given our results, it appears that deer can adjust to relatively high densities of well pads in the production phase (the period with markedly lower human activity on the landscape), provided there is sufficient vegetative and topographic cover afforded to them and populations are below carrying capacity. The strong reaction to wells in the drilling phase of development suggests mitigation efforts should focus on this activity and stage of development. Many of the wells in this area were directionally drilled from multiplewell pads, leading to a reduced footprint of disturbance, but were still related to strong behavioral responses. Our results also indicate the likely value of mitigation efforts focusing on reducing human activity (i.e., vehicle traffic, light, and noise). In combination, these findings indicate that attention should be paid to the spatial configuration of the final development footprint to ensure adequate cover. In our study system, minimizing the road network through landscape-level development planning would be valuable (i.e., exploring a maximum road density criteria). Lastly, our study highlights the importance of concomitant assessments of behavior and demography to provide a comprehensive understanding of how wildlife respond to habitat modification. © 2021 The Wildlife Society. Some memories never fade: inferring multi-scale memory effects on habitat selection of a migratory ungulate using step-selection functions Helena Rheault 1, Charles R. Anderson Jr.2, Meagwin Bonar', Robby R. Marrotte•, Tyler R. Rossl, Geroge Wittemyer4, and Joseph M. Northrup •.s 'Environmental and Life Sciences Graduate Program, Trent University, Peterborough, ON, Canada 2 Mammals Research Section, Colorado Parks and Wildlife, Fort Collins, CO, United States 3 Department of Biology, York University, Toronto, ON, Canada 4 Department offish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, CO, United States, 5 Ontario Ministry of Natural Resources and Forestry, Peterborough, ON, Canada Citation: Rheault, H., C.R. Anderson Jr., M. Bonar, R.R. Marrotte, T. R. Ross, G. Wittemyer, and J.M. Northrup. 2021. Some memories never fade: inferring multi-scale memory effects on habitat selection of a migratory wigulate using step-selection functions. Frontiers in Ecology and Evolution 9:702818; doi: I0.3389/fevo.202 l.702810 ABSTRACT: Understanding how animals use infonnation about their environment to make movement decisions underpins our ability to explain drivers of and predict animal movement. Memory is the cognitive process that allows species to store infonnation about experienced landscapes, however, remains an understudied topic in movement ecology. By studying how species select for familiar locations, visited recently and in the past, we can gain insight to how they store and use local infonnation in multiple memory types. In this study, we analyzed the movements of a migratory mule deer (Odocoileus hemionus) population in the Piceance Basin of Colorado, United States to investigate the influence of spatial experience over different time scales on seasonal range habitat selection. We inferred the influence of short and long-tenn memory from the contribution to habitat selection of previous space use within the same season and during the prior year, respectively. We fit step-selection functions to GPS collar data from 32 female deer and tested the predictive ability of covariates representing current environmental conditions and both metrics of previous space use on habitat selection, inferring the latter as the influence of memory within and between seasons (summer vs. winter). Across individuals, models incorporating covariates representing both recent and past experience and environmental covariates perfonned best. In the top model, locations that had been previously visited within the same season and locations from previous seasons were more strongly selected relative to environmental covariates, which we interpret as evidence for the strong influence of both short- and long-tenn memory in driving seasonal range habitat selection. Further, the influence of previous space uses was stronger in the summer relative to winter, which is when deer in this population demonstrated strongest philopatry to their range. Our results suggest that mule deer update their seasonal range cognitive map in real time and retain long-tenn infonnation about seasonal ranges, which supports the existing theory that memory is a mechanism leading to emergent space-use patterns such as site fidelity. Lastly, these findings provide novel insight into how species store and use infonnation over different time scales. 21 �Appendix B. Preliminary results of habitat treatment responses and herbivore use of treated sites. Vegetation and camera data to accompany the study 'Population performance of Piceance Basin mule deer in response to natural gas resource selection and mitigation efforts to address human activity and habuatdegradation' Principal Investigators: Danielle Johnston (Danielle.bilyeu@state.co.us), Chuck Anderson (chuck.anderson@state.co.us) Collaborators: Colorado Parks and Wildlife, BLM-White River Field Office, Idaho State University, Colorado State University, Federal Aid in Wildlife Restoration, EnCana Corp., ExxonMobil Prod. Co./XTO Energy, Marathon Oil Corp., Shell Petroleum, WPX Energy, Colorado Mule Deer Assn., Muley Fanatic Found., Colorado Mule Deer Found., Colorado State Severance Tax Fund, Boone & Crocket Club, and Safari Club Int. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW's right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. In 2011 and 2013, about 1,200 acres of pinyon and juniper (PJ) mastication treatments were completed in the Magnolia region of the Piceance Basin. Treated parcels averaged 7 acres in size, and were intended to increase winter range quality for deer. The treatments were part of a study to evaluate the effectiveness of PJ removal as mitigation for impacts of natural gas development on deer, with outcomes assessed in terms of deer population and demographic parameters. This summary addresses some side questions relevant to the main study, with outcomes assessed in terms of vegetation response and animal use ofvegetation treatments. We were interested in quantifying the understory forage produced by the mastication treatments. We used paired masticated/control point-intercept transects on a subset of parcels (Graham 2013) to quantify cover of plant groups relevant to deer nutrition. We used belt transects and trained ocular estimation, with benchmarks (Johnston 2018), to estimate summer utilization on individual shrubs, then scaled these to the plot level (Bilyeu, Cooper et al. 2007). We used belt transects of shrub canopy measurements, coupled with biomass equations developed for the study area (Johnston 2018) to quantify winter forage production of key browse species. Winter forage production was defined as current-year stems, not including leaves, not including biomass removed by summer browsing, and not including very small stems which would likely be shed prior to winter (Johnston 2018). We were interested in how summer use of treatments, and use of treatments by non-target animals, impacted winter forage availability. Ten cattle exclosures, distributed broadly throughout the study area (Figure 1), were built within mastication treatments in 2011 and 2013. We assessed plant cover and summer shrub utilization within these using techniques described above. On paired masticated/control transects, we deployed Reconyx Hyperfire cameras July-November 2018-2019. These were programmed to facilitate creating an index of use: 5 pictures per motion trigger, 3 second interval between pictures, a 5 minute wait time between triggers, and a sensitivity setting of High (Rhodes, Larsen et al. 2018). An animal observed with their head down or other indication of foraging in one or more of the photos in a 5 photo set was counted as one foraging event, and non-foraging occurrences were counted similarly. Sampling efforts by year are given in Table I. Because the plant cover data contained many zeros, we modeled presence/absence of each plant group separately from its cover where present (Fletcher, Mackenzie et al. 2005), using the lme4 package in R (Bates 2005). For both analyses, treatment, year, and their interaction were considered fixed effects, year was included as a categorical variable, and pair ID and plot ID were included as random effects. We used a similar approach for camera data for cattle and elk, which also contained many zeros. 22 �In general, grasses responded positively to treatment (Figure 2a). Wheatgrass presence, wheatgrass cover, and needlegrass presence were higher in treated than untreated plots. Poa grass presence was higher in treated plots by 2018, although poa grass presence and cover initially had a negative response to treatment. Cheatgrass presence also responded positively to treatment (Figure 2a). Wheatgrasses, poas, and cheatgrass all had significant year*treatment interactions for either presence or cover. Interannual variation in cover was greater in masticated plots than in control plots for these species groups (Figure 2a). Forbs responded positively to treatment. Annual forb and perennial forb presence were higher in treated than untreated plots (Figure 2b). Some shrubs responded positively to treatment, while others did not. Snowberry cover was lower in treated plots in 2013, but in 2016 and 2018, cover was higher in treated plots (Figure 2c). Variation in snowberry cover was greater in masticated than in control plots (Figure 2c). Bitterbrush did not display any significant effects until 2018, when cover was higher in treated plots (Figure 2c). Serviceberry cover was lower in treated plots over all years (Figure 2d). Sagebrush cover was initially lower in treated plots, but by 2018 this difference was no longer significant (Figure 2d). Summer utilization of serviceberry and mountain mahogany in 2018 was significantly higher in masticated than in control plots, but no differences were detected in bitterbrush or sagebrush. Winter forage production, which was summed over serviceberry, mountain mahogany, and bitterbrush, was significantly higher in masticated plots than in unmasticated plots in all years except 2016, when the pattern was reversed (Figure 3). There was no significant effect of exclosures on any plant cover group or on summer utilization in 2018. Deer, horse, elk, and cattle all foraged more often in masticated plots than in controls in 2018 (Figure 4). Cattle were only observed foraging at 6 of20 locations, horse were observed at 9, deer at 19, and elk at 6. Mastication treatments had many positive effects on forage availabilty, including higher cover of desirable grass groups such as poa grasses and wheatgrasses, higher cover of perennial forbs, and usually higher productivity of winter-available shrub forage. There were some negative effects and some differences in effects among years, however. Cheatgrass was higher in masticated plots than in controls, and snowberry cover was higher in masticated plots in 2016 and 2018. 2016 was an unusual year compared to other years of this study, with very high productivity of grasses (including cheatgrass, especially in masticated plots), and unusually high productivity of winter-available forage of desirable shrubs in control but not masticated plots. Summer shrub utilization in 2018 was higher in masticated plots than in controls. We lack any data on utilization from 2016, which might have helped explain if the lower production of winter-available forage in masticated plots was due to higher summer utilization in those plots that year. Another explanation for the 2016 results is that good conditions for grass, cheatgrass, and/or snowberry productivity in masticated plots led to increased competition which lessened productivity of desirable forage shrubs. All four of the large herbivores of interest foraged more frequently in summer and fall in masticated plots than in control plots in 2018. The impact of cattle was concentrated in only a few plots, but they did forage frequently in plots where they occurred. Cattle use ended in September, prior to the period of heavy use by deer in October. The data from the cattle exclosures does not indicate that cattle are having any measurable negative effect on forage resources. In swnmary the impact of cattle on the forage resources available to deer in mastication treatments seems minimal. However, the effect of the sum of cattle, horse, and elk foraging may have some impact. In 2019, we collected vegetation data and camera data. 2019 is the last year of data collection for this study, and final analyses will be incorporated into publications in 2020-21. LITERATURE CITED Bates, D. (2005). "Fitting linear mixed models in R." R news 5(1). Bilyeu, D. M., D. J. Cooper and N. T. Hobbs (2007). "Assessing impacts oflarge herbivores on shrubs: tests of scaling factors for utilization rates from shoot-level measurements." Journal of Applied Ecology 44(1): 168-175. 23 �Fletcher, D., D. D. Mackenzie and E. Villouta (2005). "Modelling skewed data with many zeros: a simple approach combining ordinary and logistic regression." Environmental and ecological statistics 12: 45-54. Graham, T. (2013). Magnolia habitat manipulation project vegetative monitoring: June 2013 notes on data collection and methods used, Ranch Advisory Partners, LLC: 7. Johnston, D. B. (2018). Wildlife Research Report: Examining the effectiveness of mechanical treatments as a restoration technique for mule deer habitat. Fort Collins, CO, Colorado Parks and Wildlife. Rhodes, A. C., R. T. Larsen and S. B. S. Clair (2018). "Differential effects of cattle, mule deer, and elk herbivory on aspen forest regeneration and recruitment." Forest Ecology and Management 422: 273280. \..,! V 24 �Table 1. Number of transects sampled for a given data type each year. 2011 2012 2013 2014 2015 2016 2018 Variables quantified 145 90* 90* 159 69 Percent cover of plant 107t functional groups Winter-available forage of bitterbrush, serviceberry, mountain mahogany 70t 27t 63 75t 2019 40 (camera sites) 75t 75t 75t 40 (2 cameras each) 40 (2 cameras each) (ShrubMassPerArea) Summer utilization of bitterbrush, serviceberry, mountain mahogany, and sagebrush Index of deer, elk, horse, and cattle use in summer and fall, as determined by trail camera (EventsPerDay) * Pretreatment data collected 2011-2012 will be added to a later report. tincludes 24-30 locations taken at exclosure sites. 25 �Figure 1. Sampling locatio11s within the Magnolia region o/'the Piceance Basin. 26 �b 10.0 - a 9 ras 15 '- ._ Q) > Q) > 7.5 - 0 0 (.) ~ 10 - c<ll s.o- c: Q) e (.) ,_ Q) Q) a. a. 5 - 0- 2.5 - 0.0 I I I I I I I I 2013 2014 2016 2018 2013 201 4 2016 201 8 year I d 8- snowb~ +, C , \ , ai 5 - C 9'Q) > 0 (.) year \ , 'E 6 Q) ~ Q) a. 3- ' '' > S 0 (.) ~ 4 _ sa '1 ~ Q) a. 2- I I I I I I I I 2013 2014 2016 2018 2013 2014 2016 2018 year year Figure 2. Cover ofsome p/antfunctional groups and species important.for evaluating habitat quality. Dashed lines indicate masticated plots and solid lines are controls. A " +" or "-" sign indicates sign~ficant positive or negative main effect of mastication across years (o. = 0. 05). "P" indicates that the signjficant effect was obsen 1ed in the presence/absence a11a~vsis, and ··c" indicates a sig11!fica11t effect in the cover-where-present a11azvsis. 27 �_ 20 · NE ....... .:!'! re ~ re Treatment ..... control .D .... . masticated .c T.. 2 90 · ...., "' /1 / C :, .... QJ I * * a. I ~ 60 · "'.... .E * ~ .D re re .f \ l\ T- > * ..L.. '1' 30 • ~ ~C 3 T 2013 20,~ 2015 ,016 2016 year Figure 3. Mass o_f-winrer-availableforage (curre11t-.1·ear stem mass measured in September. not including leaves or mass removed by s11111111er browsing) per unit shrub area. Data are summed over serviceberry. mountain mahogany. and bitterbrush. N=lifor 2013 and 2015 and 25-3/for other years. No transects inside.fences " ·ere included. Error bars= SE. Srars indicate sign(ficant d(fferences at alpha= 0.05 28 �Foraging, near camera Not foraging, near camera b a 75 · >. 75· n, Cl ai Cl. n, ::c Treatment 50 • ai 50 • Cl. control rJ> i: Q) ■ masticated > w Cl 25 · .!: 25 · Cl * ~ i 0 u. O· cattle ' elk O· ' horse mule' deer Ill cattle Herbivore I elk horse mule deer Herbivore Figure 4. a) Average number o.fforag ing events per hectare per day between 111id-J11(1' and midNovember, 2018 in control versus masticated plots. Stars indicate sig11[!ica11t d(lfere11ces vt a. = 0.05. indicates a sign[ficant d[fference in presC:'11ce o.fforaging e1·ents. b) ,-l1•erage 1111mher <?f"nonforaging observations per hectare per day. t 29 �Colorado Parks and Wildlife July 2022 -June 2023 WILDLIFE RESEARCH FINAL REPORT State of __________C.. ;;;o__Io__r__ad=o~_ _ _ _ _: __P=ar=k__s __a__ nd___W ___i__ld__I___iti__ e _ _ _ _ _ _ _ _ _ _ _ __ Cost Center 3430 : .;;..;.M=a=m.....m .......... al__s__R.....e__se.....a__r..... ch_______________ Work Package 3001 : .....D.....e__ er__C_o_n.....s.....e_rv__a___ti_o_n_ _ _ _ _ _ _ _ _ _ _ __ Task No. 6 : Population Performance of Piceance Basin Mule Deer in Response to Natural Gas Resource Extraction and Mitigation Efforts to Address Human Activity and Habitat Degradation Federal Aid Project:_ ___,;,W.....-,_2__ 43__-__R__-7_______ Period Covered: July 1, 2022 - June 30, 2023 Author: C. R. Anderson, Jr. Personnel: D. Bilyeu-Johnston, CPW; J. Northrup, Ontario Ministry of Natural Resources and Forestry; R. Marrotte and M. Bonar, Environmental and Life Sciences Graduate Program, Trent University; B. Gerber, University of Rhode Island. Project support received from Federal Aid in Wildlife Restoration, Colorado Mule Deer Association, Colorado Mule Deer Foundation, Muley Fanatic Foundation, Colorado State Severance Tax Fund, Caerus Oil and Gas LLC, EnCana Corp., ExxonMobil Production Co./XTO Energy, Marathon Oil Corp., Shell Petroleum, Williams and WPX Energy. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW's right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. �WILDLIFE RESEARCH FINAL REPORT POPULATION PERFORMANCE OF PICEANCE BASIN MULE DEER IN RESPONSE TO NATURAL GAS RESOURCE EXTRACTION AND MITIGATION EFFORTS TO ADDRESS HUMAN ACTIVITY AND HABITAT DEGRADATION CHARLES R. ANDERSON, JR PROJECT NARRITIVE OBJECTIVES 1. To determine experimentally whether enhancing mule deer habitat conditions on winter range elicits behavioral responses, improves body condition, increases fawn survival, and ultimately, population density on mule deer winter ranges exposed to extensive energy development. 2. To determine experimentally to what extent modification of energy development practices enhance habitat selection, body condition, fawn survival, and winter range mule deer densities. SEGMENT OBJECTIVES 1. Finalize publication of results addressing vegetation and mule deer responses to 3 mechanical treatment methods on pinyon-juniper winter range. 2. Finalize development of a web-based energy development planning tool to guide future energy development to minimize and/or mitigate mule deer disturbance on winter range. 3. Submit final Federal Aid in Wildlife Restoration report to complete this research project. PROJECT OVERVIEW AND RESEARCH SUMMARY We experimentally evaluated mule deer (Odocoileus hemionus) response to energy-development activities and habitat treatments to address energy development planning and mitigation options. The Piceance Basin of northwestern Colorado was selected as the project area due to ongoing natural gas development in one of the most extensive and important mule deer winter and transition range areas in Colorado. The data presented here represent final results of an I I-year research project addressing habitat improvements as mitigation and evaluation of deer responses to energy development activities to inform future development planning options on important seasonal ranges. From 2008-2019, we monitored deer on 4 winter range study areas representing relatively high (Ryan Gulch, South Magnolia) and low (North Magnolia, North Ridge) levels of development activity (Fig. 1) to address factors influencing deer behavior and demographics and to evaluate success of habitat treatments as a mitigation option. We recorded adult female habitat use and movement patterns; estimated neonatal, overwinter fawn and annual adult female survival; estimated annual early and late winter body condition, pregnancy and fetal rates of adult females; and estimated annual mule deer abundance among study areas. Winter range habitat improvements completed spring 2013 resulted in 604 acres of mechanically treated pinyon-juniper/mountain shrub habitats in each of 2 treatment areas (Fig. 2) with minor (North Magnolia) and extensive (South Magnolia) energy development, respectively. During this final research segment, we finalized publication of results addressing vegetation and mule deer responses to 3 mechanical treatment methods on pinyon-juniper winter range (Johnston and Anderson 2023), and finalized development of a web-based energy development planning tool to guide future energy development to minimize and/or mitigate mule deer disturbance on winter range (Marrotte et al. 2022; Appendix B). Based on final data analyses (see Anderson 2019 for detailed methods and results and Appendix A for publication abstracts) for this I I-year project: (1) annual adult female survival 2 V �was consistent among areas averaging 79-87% annually, but overwinter fawn survival was variable, ranging from 31% to 95% within study areas, with annual and study area differences primarily due to early winter fawn weights, annual weather conditions, and factors associated with predation on winter range (e.g., crusted snow); (2) mule deer body condition early and late winter was generally consistent within areas, with higher variability among study areas early winter, primarily due to December lactation rates, and late winter condition related to seasonal moisture and winter severity; (3) late winter mule deer densities increased through 2016 in all study areas, ranging from a 50% increase in North Ridge to a 103% increase in North Magnolia, but stabilized through 2018 in 3 of the 4 study areas with a recent decline evident in North Ridge (Fig. 3); (4) migratory mule deer selected for areas with increased cover and increased their rate of travel through developed areas, and avoided negative influences through behavioral shifts in timing and rate of migration, but did not avoid development structures (Fig. 4); (5) mule deer exhibited behavioral plasticity in relation to energy development, without evidence of demographic effects, where disturbance distance varied relative to diurnal extent and magnitude of development activity (Fig 5), which provide for useful mitigation options in future development planning; (6) energy development activity under existing conditions did not influence pregnancy rates, fetal rates or early fawn survival (0-6 months), but may have reduced neonatal survival (March until birth) during 2012 when drought conditions persisted during the third trimester of doe parturition (Fig. 6); (7) rollerchopped plots provided the best combination of hiding cover and winter forage relative to mule deer use on winter range (Fig. 7), but mastication or chaining, applied leaving dispersed security cover, may be better options at large scales or when invasive species concerns exist; and (8) these results informed development of a spatial planning tool to guide future energy development on mule deer winter range (Appendix B). Final data collection efforts for this project were completed by spring 2020 (final GPS collar recovery). Collaborative research with agency biologists, graduate students, and university professors produced 26 scientific publications addressing improved monitoring techniques for neonate mule deer captures (Bishop et al. 2011, Peterson et al. 2018b); development and evaluation of a remote deer collaring device (Bishop et al. 20 I 9); mule deer migration relative to energy development (Lendrum et al. 2012, 2013, 2014; Anderson and Bishop 2014), improved approaches to address animal habitat use patterns (Northrup et al. 2013; Rheault et al. 2021 ); mule deer response to helicopter capture and handling (Northrup et al. 20 I4a); potential effects of male-biased harvest on mule deer productivity (Freeman et al. 2014); mule deer genetics in relation to body condition and migration (Northrup et al. 2014b, Bonar et al. 2022); acoustic monitoring to investigate spatial and temporal factors influencing mule deer vigilance (Lynch et al. 2014) and foraging behavior (Northrup et al. 2019); the relationship of plant phenology with mule deer body condition (Searle et al. 20 I 5); approaches to identify cause-specific mortality in mule deer from field necropsies (Stonehouse et al. 2016); the influence of individual and temporal factors affecting late winter body condition estimates of adult female mule deer (Bergman et al. 20 I 8); mule deer behavioral and demographic responses to energy development activities to inform future development planning (Northrup et al. 2015, 2016a, 2016b, 2021, Peterson et al. 2017, 2018a); plant and mule deer responses to 3 mechanical treatment methods on winter range (Johnston and Anderson 2023), and application of web-based planning tool to guide future energy development (Marrotte et al. 2022; Appendix B). These publications with management implications where appropriate are summarized in Appendix A and Appendix B. We anticipate the opportunity to work cooperatively toward developing solutions for allowing the nation's energy reserves to be developed in a manner that benefits wildlife and the people who value both the wildlife and energy resources of Colorado and elsewhere. 3 �Well Pads & Facilities SOutn MagnOka l In deve!Opment ! Proaucing w~I _ OevelOpment tacllmes 10 Milts Figure I. Mule deer winter range study areas relati ve to active natural gas well pads and energy development fac ilities in the Piceance Basin o f northwest Colorado. winter 20 I 3/ 14 (A ccessed http://cogcc.state.co. us/ December 31.2013: energy development drilling acti vi ty has been minor since 2012). 4 �North Magnolia treatemem sites (587 acres) LJ BearSet_l5_35b_andG BearSet_ l _BandA._E LJ BearSet_36_54anclJ GreasewoodSet_gl 6_g29 GreasewoodSet_gl _g 15 LJ GreasewoodSet_g30_g42 LeeOvers,ghts_a_fandl 6_ 17 Mechanical treatment companson (54 acres) - - NOr1h Hatch Pilot Treatments ( 116 acres) South Magnolia 2 a Figure 2. Habitat treatment site delineations in 2 mule deer study areas (604 acres each) of the Piceance Basin, northwest Colorado (Top; cyan polygons completed Jan 20 I I using hydro-axe; yell ow polygons completed Jan 20 12 using hydro-axe, roll er-chop, and chaining; and remaining polygons completed Apr 20 13 using hydro-axe). January 20 11 hydro-axe treatment-site photos from North Hatch Gulch during April (Lower left, aerial view) and October, 20 11 (Lower ri ght, ground vi ew). 5 �Piceance Basin late winter mule deer density 35.00 30.00 25.00 1 20.00 - ";::- g: 15.00 C 10.00 5.00 - North Ridge - - - Ryan Gulch - • North Magnolia - South Magnolia 0.00 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Year Figure 3. Mule deer density estimates and 95% Cl (error bars) from 4 winter range herd segments in the Piceance Basin, northwest Colorado, late winter 2009-20 18. Figure 4. Mule deer study areas in the Piceance Basin of northwestern Colorado, USA (Top), spring 2009 migration routes of adult fe male mule deer (n = 52; Lower left), and active natural-gas well pads (black dots) and roads (state, county, and natu ral-gas; white lines) from May 2009 (Lower right; from Lendrum et al. 20 12; http://dx.do i.org/ 10.1890/ES 12-00 165. 1). 6 �i~1 f - ,;, - - - + - - - - .... - - ~- - - - - - 7 - - - -t- - - - + J J7 ,., I Prod 400 Proo 600 PrOd 800 PrOd 1000 Or1U 400 Onll 600 o ,m soo Orlll 1000 Covariates - - - 4----4-- ~- --- ?--- A · · l · · f - - + j o ~ ~ I ~ J c;- ,., I Prod • OO PrOd 600 Prod 800 PrOd 1000 Drill 400 ~----·-·-· Orill 600 Onll 800 o nu 1000 Figu re 5. Posterior distri butions of populat ion-level coefficients related to natural gas development for RSF models d uring the day (top) and night (bottom) for 53 adult female mule deer in the Piceance Basin, northwest Colorado. Das hed li ne indicates 0 selection or avo idance (below the li ne) of the habitat features. ' Drill' and ' Prod' represent drilling and produc ing well pads, respectively. The numbers following ' Drill' or ' Prod' represent the distance from respective well pads evaluated (e.g., ' Drill 600' is the number of well pads with active drilling between 400-600 m from the deer location; from Northrup et al. 20 15; http://onli nelibrary.wiley.com/doi/10. 1 I 11/gcb. 13037/abstract). Road disturbance was re lat ively minor (- 60-1 20 m, not illustrated above). 1 .00 a, 0 .80 r---9- - - - - rl- I§ cii 0 .60 > -~ ::, 1/l ro 0 .40 w LL 0.20 0.00 2012 2013 2014 Year □ High development □ Low development I Figure 6. Model averaged esti mates of mu le deer fetal survival from early Ma rch until bi11h (late MayJune) in high and low energy deve lopment study areas of the Piceance Basin, northwest Colorad o. 20 1220 14 ( from Peterson et a l. 20 17; http://www.bioone.org/do i/pdf/ l 0.298 1/wlb.00341 ). 7 �C C 4 co ..c: .._ be ~ 3 ro ab -0 Cl) (/) ::, "Cl) 2 Q) I -0 !,,... (1) ......, C s1 0 CON MAST CHAIN ROLLER Figure 7. Winter mule deer use days from GPS locations (days/ha) over a 5-year period in control plots and plots treated to remove pi nyon and juniper trees by 3 different methods: CON (control). MAST (masticated). CHAIN (chained), and ROLLER (roller-chopped). Bars not sharing letters are significantly different at a = 0.05. Error bars = SE (from Johnston and Anderson 2023; https://doi.org/ I0. 1002/wsb. I42 I). 8 �LITERATURE CITED Anderson. C. R .. Jr. 20 19. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human acti vity and habitat degradati on. Federal Aid in Wi ldlife Restorati on A nnual Report W-243-R3. Ft. Coll ins. CO USA. A nderson, C. R.. Jr.. and C. J. Bishop.2014. M igrati on patterns of adult female mule deer in response to energy development. Pages 47-50 in Transactions of the 79'" North Ameri can Wi ldlife & Natural Resources Conference (R. A. Coon & M. C. Dunfee, eds.). Wildlife M anagement Institute. Gardners, PA, USA. ISSN 0078-1355. Bergman. E. .I.. C. R. Anderson Jr.. C. J. Bishop. A. A. Holl and. and J.M. Northrup. 20 18. Variation in ungulate body fat: individual versus temporal effects. Journal of Wi ldlife Management 82:130- 137. DO I: 10: 1002/j w mg.2 l 334 Bishop, C. J., C. R. Anderson .Ir.. D. P. Walsh. E. .I. Bergman, P. Kuechle. and J. Roth. 20 I I . Effecti veness ora redesigned vaginal implant transmitter in mule deer. Journal or Wild life Management 75(8): 1797-1806. DOI: I 0.1002/jwmg.229 Bishop, C. J., M . W. A lldredge, D. P. Walsh, E. J. Bergman, C. R. Anderson Jr.. D. K ilpatri ck, .I. Bake I, and C. Fabvre. 20 19. A noninvasive automated device for remotely collaring and weighing mule deer. Wildli fe Society Bulletin 43:717-725. doi.org/ 10. 1002/wsb. I 034 Bonar. M. S. J. Anderson, C. R Anderson Jr. G. Wittemyer, J. M. Northrup. and A. B. A. Shafer. 2022 Genomic correlates ror migratory direction in a free-ranging cervid. Proceedings of the Royal Society B 289: 2022 1969. hllps://doi.org/ l O.l098/rspb.2022.1969 Freeman. E. D .. R. T. Larsen. M. E. Peterson. C. R. Anderson. Jr.. K. R. Hersey. and B. R. McM illan. 20 14. Effects of male-biased harvest on mule deer: implications for rates of pregnancy, synchrony, and timing of parturition. Wildlife Society Bulletin 38(4):806-8 11. DOI: I 0. I 002/wsb.450 Johnston. 0 . B .. and C.R. A nderson Jr. 2023. Plant and mule deer responses to pinyon-juniper removal by three mechanical methods. Wild Ii fe Society Bulletin 47:e 142 1. https://doi.org/10. 1002/wsb. I 42 I Lendrum. P. E., C.R. An derson, .Ir., R. A. Long, J. K. Kie, and R. T. Bowyer. 20 12. Habitat selection by mule deer during migrati on: effects of landscape structure and natural gas development. Ecosphere 3(9):82. http://dx.doi.om/ I 0. Lendrum. P. E.. C. R. Anderson, .Ir .. K. L. Monteith, J. A . Jenks. and R. T. Bowyer. 20 13. Migrating Mule Deer: Effects o f A nthropogenically Altered Landscapes. PLoS ONE 8(5): e64548. doi: 10. I 371 /journal.pone.0064548 Lendrum, P. E., C. R. A nderson, .Ir., K . L. Monteith, .I. A. Jenks. and R. T. Bowyer.20 14. Relating the movement of a rapid ly migrating ungulate to spatiotemporal patterns o f forage quality. Mammal ian Biology 79(6):369-3 75. http://dx.doi.om/ I 0. 10 16/j .mambio.20 14.05.005 Lynch, E.. J.M. Northrup, M . F. McKenna, C. R. A nderson Jr., L. Angeloni, and G. Wittemyer. 2014. Landscape and anthropogenic features influence the use of auditory vigilance by mule deer. Behavioral Ecology 26( I ) :75-82. doi: I 0.1 093/beheco/aru 158. Marrolle, R. R., C. R. A nderson Jr., and .I. M. Northrup. 2022. Developing a spatial planning tool for natural gas development on mule deer winter range. Final Report to Bureau o f' Land Management : Grant Agreement LI 8AC00068. I 4pp. Northrup. J.M .. M. B. Hooten. C.R. Anderson. Jr.. and G. Wittemyer. 20I3. Practical guidance on characteri zing availability in resource selection functi ons under a use-avai lability design. Ecology 94(7): 1456- 1463 . Northrup, J. M .. C. R. Anderson, Jr., and G. Winemy er.2014a. Effects of helicopter capture and handling on movement behavior of mule deer. Journal of Wildlife Management 78(4):731738. DOI: 10. 1002/j wmg.705 9 �Northrup. J.M .. A. B. A. Shafer. C.R. A nderson Jr. . D. W. Coltman. and G. Whi ttemyer. 20 I 4b. Finescale geneti c correlates to condition and migration in a wild cervid. Evolutionary Applications 7(8):937-948: doi: IO. I I l I /eva.12 I89 No11hrup. J. M .. C. R. A nderson. Jr.. and G. Willemyer.20 I 5. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selecti on patterns of mule deer. Global Change Biology 2 1( 11 ):386 1-3970. doi: I 0. I I I l /gcb. I 3037. Northrup. J.M .. C.R. Anderson, Jr.. M. B. Hooten. and G. W itternyer. 2016a. Movement reveals scale dependence in habitat selection or a large ungulate. Ecological Applications 26:27462757. North rup, J.M .. C. R. Anderson. .Ir.. and G. Wirtemyer. 2016b. Environmental dynam ics and anthropogenic development alter philopatry and space-use in a North American cervid . Diversity and Di stributions 22 :547-557 . DOI: I 0. 1I I I /ddi. I24I7 Northrup, J.M. , A . Avrin . C.R. Anderson Jr., E. Brown, and G. Wittemyer. 20 19. On-animal acoustic monitoring provides insight to ungulate foraging behavior. Journal or Marnmalogy I 00: I 4791489. https://doi.org/ I 0. 1093/j mam mal/gyzl 24 Northrup, J.M .. C.R. A nderson .Ir.. B. D. Gerber, and G. Wittemyer. 202 I . Behavioral and demographic responses of mule deer to energy development on winter range. Wildlife Monographs208: l -37;202 I : DOI : I 0. I002/wmon.I060 Peterson, M . E.. C. R. A nderson .Ir., .l. M. Northrup, and P. F. Doherty Jr. 20 17. Reproductive success or mule deer in a natural gas development area. Wildlife Bio logy. doi: I 0. I1II / w lb.0034I Peterson. M. E.. C.R. A nderson .Jr.. J.M. Northrup. and P. F. Dohe11y Jr. 20 18a. Mortality of mule deer fawns in a natural gas development area. Journal of Wildli fe Management 82: I I 35-11 48. DOI: 10.1002/j w mg.2 1476 Peterson. M. E.. C. R. A nderson .Ir.. M. W. A lldredge. and P. F. Doherty Jr. 20 18b. Using maternal mule deer movements lo estimate timing of parturition and assist faw n captures. Wildlife Society Bulletin 42:6 16-62 1. DOI: I 0. 1002/wsb.935 Rheault. H .. C. R. A nderson Jr.. M. Bonar. R. R. Marrotte. T. R. Ross. G. W ittemyer, and J. M. Northrup. 202 1. Some memories never fade: inferring multi-scale memory effects on habitat selection of' a migratory ungulate using step-selection functi ons. Frontiers in Ecology and Evolution 9:7028 18. doi : I 0.3389/fevo.202 I. 7028 10 Searl e. K. R., M. B. Rice. C. R. A nderson. C. Bishop and N . T. Hobbs. 20 15. Asynchronous vegetati on phenology enhances w inter body condition of a large mobile herbivore. Oecologia 179:377-39 1. DOI I 0.1 007/s00442-015-3348-9 Stonehouse. K. F.. C. R. Anderson .Ir.. M. E. Peterson. and D. R. Collins.20 16. Approaches to fi eld investi gations or cause-speci fie mortality in mule deer ( Odocoi/eus hemionus). Colorado Parks and Wi ldlife Technical Report No. 48, First Edition, 317 W. Prospect Rd., Ft. Collins. CO USA. DOW-R-T-48-I 6. ISSN 0084-8883. Prepared by Chuck Anderson °"Jllally dgl\4'd DYC"hutl. , N... w• °'~ '°"''" ,,,...,.,... Charles R. A nderson. .Jr.. Mammals Research Leader 10 �Appendix A. Abstracts of published manuscripts resulting from Piceance Basin mule deer/energy development interaction research collaborations. Abstract format specific to the respective journal requirements. Effectiveness of a redesigned vaginal implant transmitter in mule deer CIIAD J. BISIIOI'', CIIARLES R i\N DERSON .Jr.', D,\:\IEJ. P. \\'ALSII ', ERIC .I. BERG~IA~ 1• l'ETEn h'. lltCIII.E'. :i nti ,IOII N ROTIJ' 1Colorado Parks and Wildlilc. Fort Collins. Colorado 80526 lJSA 'Advanced Telemetry Systems, lsn1111, Minnesota 55040 USA Citation ' Bishop, C J.. C R. Anderson Jr.. D P. Walsh. I: J. Bergman. r Kuechle, and J l~oth. 2011 Effecu vcn~ss ot'a redesign~d vaginal implant transmillcr 111 mule deer Joumal or Wildlilc Management 75(8 ): I797-1806. DOI I() I002/jwmg 229 ABSTRACT Our understanding of factors that limit mule deer (Odocoileus '1e111io1111s) populations may be improved by evaluating neonata l survival as a funct ion of dam chuructcristics under free-ranging conditions. which generally requires that both neonates and dams are radiocollarcd. l'hc most viable technique lacil i1a1ing capture of neonates lrom radiocollarcd adu lt females is use of vaginal implant transmillcrs (VITs). To date. VITs have allowed research opportunities that were not previously possible: however. VITs arc oflen expelled from adult females prepartum. which limits their effectiveness. We redesigned an existing VIT manufactured by /\dvam:cd Telemetry Systems (t\TS: Isanti. MN) by lengthening and widen ing wings used 10 retain the VIT in an adult lemale. Our objct:tive was to increase VIT retentio n rates und thereby increase 1hc likelihood of locating birth sites and ncwbom lawns. We placed the newly designed VITs in 59 adult female mule deer and evaluated the probability of rc1e111ion to parturition and the probability of detecting newborn fawns. We also developed an equation for determining VIT samplc size necessary 10 ach ieve a specified sample size of neonates. The probability of a VIT being n:tnincd until parturition was 0.766 (SE = 0.0605) and the probability ofa VIT being retained lo within 3 days of parturit ion was 0.894 (SE = 0.044 1). In a similar study using the original VIT wings (Bishop ct al. 2007). the probability ofa VIT being retained unti l parturition was 0.-147 (SE= 0.0468) and the probability of retention 10 within 3 days of parturition was 0.623 (SI~= (J.0456). Thus. our design modification increased VIT retention 10 pal'lurition by 0.3 19 (SE = 0.0765) and VIT retention lo with in 3 day of parturition by 0.271 (SE = 0.063-1 ). Considering dams that retained V l'l s to within 3 days of parturition. the probability of detecting at least I neonate was 0.952 (SE = 0.0334) and the probability ofdc1cc1ing both fa\1ns from twin li11ers was 0.588 ( E = 0.0827). We expended appro.xima1cly 12 person-hours per detected neonate. /\s a guide for researchers planning future studies. we fou nd that VIT sample size: should approximately equal the targeted neonate sample size. Our study expands oppor1uni1ics fo r conducting research that links adult female att ributes 10 prod uctivity and offspring survival in mule deer. © 20 14 The Wildlife Society. Habitat selection by mule deer during migration: effects of landscape structure and natural-gas development l'ATRICI< E. LtNDRUM', CIIARLES It A:\'DERSOI'\ .m . 1• n YAN ,,. LONG', ,IOIIN G. h:IE 1, AN D R. TERRY BOWYER' 'Department of Uiological Sciences. Idaho State University, l'ocatello, Idaho 83209 USA ~Colorndo Parks and Wildlilc. Grund Junction. Colorado 81505 lJSA Ci1a11on. Lendrum. P. E., C. R Anderson Jr., R. A Long, J G Kie. and R. r. 1301\ycr 2012 llabitat sclccuon hy mule deer during rnigrnuM c1kc1s of landscape structure and natural-gas development l:cosphcrc 3(9):82 hnp /Id~ do1 ore/ I() 1890/1:S12-001(,5 I Abstract. The disruption of tradit ional migratory routes by anthropogenic disturbances has shil'led pallerns of resource se lection by many species. and in some instances has caused populatio ns 10 decline. Moreover. in recent decades populations of'111ulc deer (Odocoileus he111ion11s) have clcclincd throughout much of their historic range in the western Uni 11.:d States. We ust.:d rcsoure1.:sclec1ion functions 10 determine ii'thc presence ofnaturnl-gas dcvelopmenl altered pallerns of resource selection by migrating mule deer. \Ve compared spring migration routes of adult fema le mule deer lilled with GPS collars (11 = 167) among four study areas 1ha1 had varying degrees of natural-gas development from 2008 to 20 IO in the Piccance Basin of nonlmcst Colorado. US/\. Mule deer migrating through the most developed area had longer step lengths (straight-line distuncc between sueccssivl'. GPS lorntions) compared with deer in less developed areas. Add itionally. deer migrat ing through the most developed study ar1.:as tended 10 select for habitnt types that provided grcntcr amounts or concealment cover. whereas deer li·om the least clcvclopcd areas tended 10 select habitats that increased access 10 forage and cover. Deer sclc1.:1ed habitats closer 10 we ll pads and avoided roads in all instances except along the most highly dl'.vcloped migratory routes. where road densities may have been 100 high for deer 10 avoid roads without deviati ng substantially li-0111 cslllblishcd migration routes. These results indicate that behavioral tendencies toward avoidance o r anthropogenic disturbance can be overridden during migration b) the strong fidelit) ungulates demonstrate towards migration routes. If avoidance is leasiblc. then deer may select areas further from development. whereas in highly dcvelopcd areas. deer may simply increase their rate ol'travcl along eswblishcd migration routes. 10 �Migrating Mule Deer: Effects of Anthropogenically Altered Landscapes Patrick E. Lendrum', Charles R. Anderson Jr.2, Kevin L. Monteith 1.J, Jonathan A. Jenks4, R. Terry Bowyer' 1 Department ofBiological Sciences, Idaho State University, Pocatello, Idaho, USA, 2 Colorado Division of Parks and Wildlife, Grand Junction, Colorado, USA, J Wyoming Cooperative Fish and Wildlife Research Unit, University of Wyoming, Laramie, Wyoming, USA,4 Department of Natural Resource Management, South Dakota State University, Brookings, South Dakota, USA Citation: Lendrum, P. E., C.R. Anderson Jr., K. L. Monteith, J. A. Jenks, R. T. Bowyer. 2013. Migrating Mule Deer: Effects of anthropogenically Altered Landscapes. PLoS ONE 8(5): e64548. DOI: I0.1371/joumal.pone.0064548 Abstract Background: Migration is an adaptive strategy that enables animals to enhance resource availability and reduce risk of predation at a broad geographic scale. Ungulate migrations generally occur along traditional routes, many of which have been disrupted by anthropogenic disturbances. Spring migration in ungulates is of particular importance for conservation planning, because it is closely coupled with timing of parturition. The degree to which oil and gas development affects migratory patterns. and whether ungulate migration is sufficiently plastic to compensate for such changes. warrants additional study to better understand this critical conservation issue. Met/,odo/ogy/Principa/ Fim/i11gs: We studied timing and synchrony of departure from winter range and arrival to summer range of female mule deer (Odocoileus hemionus) in northwestern Colorado. USA. which has one of the largest natural-gas reserves currently under development in North America. We hypothesized that in addition to local weather. plant phenology. and individual life-history characteristics. patterns of spring migration would be modified by disturbances associated with natural-gas extraction. We captured 205 adult female mule deer. equipped them with GPS collars. and observed patterns of spring migration during 2008-20 I 0. Conclusi011s/Signijica11ce: Timing of spring migration was related to winter weather (particularly snow depth) and access to emerging vegetation. which varied among years. but was highly synchronous across study areas within years. Additionally. timing of migration was influenced by the collective effects of anthropogenic disturbance, rate of travel, distance traveled. and body condition of adult females. Rates of travel were more rapid over shorter migration distances in areas of high natural-gas development resulting in the delayed departure. but early arrival for females migrating in areas with high development compared with less-developed areas. Such shifts in behavior could have consequences for timing of arrival on birthing areas. especially where mule deer migrate over longer distances or for greater durations. Practical guidance on characterizing availability in resource selection functions under a use-availability design JOSEPH M. NORTHRUP', ME\'IN B. HOOTEN 1·u, CHARLES R. ANDERSON JR. 4, AND GEORGE WITTEMYER1 1 Department of Fish, Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 2 U.S. Geological Survey, Colorado Cooperative Fish and Wildlife Research Unit, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA ~Colorado State University, Department of Statistics, Colorado State University, 1474 Campus Delivery, Fort Collins, Colorado 80523 USA 4 Mammals Research Section Colorado Parks and Wildlife, 711 Independent Avenue, Grand Junction, Colorado 81505 USA Citation: Northrup, J. M., M. B. Hooten, C. R. Anderson Jr., and G. Wittemyer. 2013. Practical guidance on characterizing availability in resource selection functions under a USt.'--availability design. Ecology 94(7): 1456-1463. http://dx.doi.org/10.1890/12-1688. I Abstract. Habitat selection is a fundamental aspect of animal ecology. the understanding of which is critical to management and conservation. Global positioning system data from animals allow fine-scale assessments of habitat selection and typically arc analyzed in a use-availability framework. whereby animal locations arc contrasted with random locations (the availability sample). Although most use-availability methods are in fact spatial point process models. they often are fit using logistic regression. This framework offers numerous methodological challenges, for which the literature provides little guidance. Specifically. the size and spatial extent of the availability sample influences coefficient estimates potentially causing interprctational bias. We examined the influence of availability on statistical inference through simulations and analysis of serially correlated mule deer GPS data. Bias in estimates arose from incorrectly assessing and sampling the spatial extent of availability. Spatial autocorrelation in covariates. which is common for landscape characteristics. exacerbated the error in availability sampling leading to increased bias. These results have strong implications for habitat selection analyses using GPS data. which are increasingly prevalent in the literature. We recommend that researchers assess the sensitivity of their results to their availability sample and. where bias is likely, take care with interpretations and use cross validation to assess robustness. u 1I �Effects of Helicopter Capture and Handling on Movement Behav ior of Mule Deer ,JOSEPII ~I. NORTIIRUP', Clli\Rl.ES H. ,\ 'DERSO .m'. AN D GEORG E W ITTLJ\IYEH' I Dcpanmcnt oi' Fish, Wildl1lc. and Conscrva1ion 13iology, Colorado S1;11c U111vcrsi1y, 1474 Campus Dd 1vcry, Fon Collins. Colomclo 80523 USA cMa111muls Research Sccl ion Colorado Parks and Wildlilc. 7 11 lndcpcndcm /\venue, Grand Junc11on. Colorado 81505 US/\ Ci1a11on : Northrup. J. M .. C R. i\nclcrson Jr .. and G. Wiucmycr 20 14 Effcc1s ol'hclicoplcr caplurc uml handling on movcmcnl behavior oi'mulc deer Journal ofWildlilc Managcmml 78(4).731-738. DOI : 10 1002/jwmg 705 .\BSTRA CT Research on wildl ilc movement physiology. and reproducl ive biology oncn requires caplurc and handling oi' ani mals. Such in vasive lreauncnl cun alter behavior. whid1 may hias results or invalidaie assumptions regarding representat ive behaviors. T o assess the impacts of handling on mule deer (Odocoileus he111io1111s). a focal species for research in North America. we investigated pre- and post-recapture movements of co ll ared indiv iduals. and compared them to deer t hat wen: not rccnpturcd (control s). We compared pre- and post-recapture movement rutcs (111/hr) and 24-hour straight-line displacement among recaptured and contro l deer. In addition. we e:--amined the time it took recaptured deer to return to their pre-recapture home range. Both daily straight-line displacement and movement rate wen: ,narginally elevated relative to monthly averages for 24 hours lo Ilowing recapture. with non- signi Ii cant elevation continuing for up to 7 days. Comparing movements avcrag1:d over 30 da) s before and afler recapture. we found no d i fferences in displac1:111ent. but movement rates demonstrated seasonal effects. with fos ter movements post- relat ive to pre-recapture in March and slower movements post- relative to pre-recapture in December. Relati e to control deer movements. recaptured deer mo, cment rates in March were higher im mediately a lier recapture and lower i n the second and third weeks followi ng recapture. The median time to return to the pre-recapture home range was 13 hours. with 71 % of deer returning in the first day. and 9 1% returning withi n ~ days. r hese results ind icate a short period 0 1· elcvuled movements fo llowing recaptures. likely due to the deer returning to their home ranges. followed by weaker but non-signili cant depression of movements for up to 3 weeks. Censoring ol' thc lirst day of data post capture from analyses is strongly supported. and remov ing additional days unt il the individual returns to ils home range will control for the 111:i_jority o r impacts from capture. V 20 1-l 'Ille W ildl i fe Society. Relating the movement of a rapid ly migrating ungulate to spatiotemporal patterns of forage quality Patrick E. Lendrum•, Charles R. Anderson Jr.", Kevin I.. Monteith' , .lonntlrnn A. ,Jcnki , R. Terry IIOll')'Cr• ·' Department of Biological Sciences. Idaho State University, 921 South 8th /\venue, Stop 8007. l'o<:atcllo 83209. US/\ '' Mammals Research Section Col()rauo Parks and Wildlifo. 711 Independent /\venue. Grand Junction 81505. US/\ ' Wyoming Cooperative Fish and Wilclhlc Research Unit. Dcpartmclll of Zoology and Physiology. University of Wyoming, 3 166, 1000 East Universlly /\venue. Laramie 82071. US/\ " Dcpanmcm of Natural Resource Management. South Dakolu Swtc Univcr.;1ty. Box 2140l3. l3rookmg, 57007, USA C1tut1011 I.end rum. P E., C R. Anderson Jr. K I.. Monteith. J i\ Jenks. and R. T Bm,~•cr. 20I4 , Rclaung the movcmclll or a rnp1clly migrating ungulate 10 spmiotcmporal pancms or forage quality. M amnmhan Bmlogy· http./tdx do1 org/10 10 16/J mamb10 2014 05 005 ABSTRACT: Migratory ungu lates exhibil recurring movements. 011cn along tradilional routes between seasonal ranges each spring and autumn. which allow them 10 track resources as they become available on the landscape. We examined the relat ionship between spri ng migration o r mule deer (Odornileus he111ion11s) and forage qual ity. as indexed by spatiotcmporal putlerns ol'fi::cal nitrogen and remotely sensed greenness of'vegetation (Normalized Diilcrence Vegetat ion Index: NDVI) in spring 20 10 in the Piceancc Basin of'nortbwcstem Colorado. US/\. N D V I increased throughout spring. and was artcctcd primar ily by snow depth when snow was present. and tempcruture when snow was absent. Fecal nitrogen \\US IO\\CSt when deer were on w inter range before m igration. increased rapidly to an asymptote during migration. and remained relatively high when deer reached summer range. Values of foca l nitrogen corresponded with increasing NDVI duri ng migration. Spring migration for mule deer provided a way for these large mam mals to increase access to a high-qua lity diet. which was evident in patterns of N D V I and fecal nitrogen. Moreover. these deer "jumped" rather than "surfed" the green wave by ar riving on summer range well before peak producti vity or forage occurred. This rapid migration may aid in securing resources and seclusion fro m others on summer range in preparation for parturition. and to min imize detrimental factors such ns predat ion and malnutrit ion during m igrat ion. 12 �Effects of Male-Biased Harvest on Mule Deer: Implications for Rates of Pregnancy, Synchrony, and Timing of Parturition ERIC D. FREEMAN', RANDY T. l.ARSEN 1, MARKE. PETERSON 2, CHARLES R. ANDERSON JR.3, KENT R. HERSEY\ AND BROCK R. McMILLAN' 1 Department of Plant and Wildlife Sciences, Brigham Young University, 275 WIDB, Provo, UT 84602, USA 2 Department of Fish, Wildlife, and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA 3 Colorado Parks and Wildlite, 711 Independent Avenue, Grand Junction, CO 81505, USA 1 Utah Division of Wildlife Resources. I594 W North Temple, Salt Lake City, UT 84114, USA Citation: Freeman, E. D., R. T. Larsen, M. E. Peterson, C.R. Anderson Jr., K. R. Hersey, and B. R. McMillan. 2014. Effects of male-biased harvest on mule deer: implications for rates of pregnancy, synchrony, and timing of parturition. Wildlife Society Bulletin; DOI: 10.1002/wsb.450 ABSTRACT Evaluating how management practices influence the population dynamics of ungulates may enhance future management of these species. For example. in mule deer (Odocoileus hemionus), changes in male/female ratio due to male- biased harvest may alter rates of pregnancy. timing of parturition. and synchrony of parturition if inadequate numbers of males are present to fertilize females during their first estrous cycle. If rates of pregnancy or parturition are influenced by decreased male/female ratios, recruitment may be reduced (e.g .. fewer births. later parturition resulting in lower survival of fawns. and a less synchronous parturition that potentially increases susceptibility of neonates to predation). Our objectives were to compare rates of pregnancy. synchrony of parturition. and timing of parturition between exploited mule deer populations with a relatively high (Piceance, CO. USA: 26 males/I 00 females) and a relatively low (Monroe. UT. USA: 14 males/I 00 females) male/female ratio. We determined rates of pregnancy via ultrasonography and timing of parturition via vaginal implant transmitters. We found no differences in rates of pregnancy (98.6% and 96.6%: z = 0.821: P = 0.794). timing of parturition (estimate= 1.258: SE= 1.672: I= 0.752: P = 0.454), or synchrony of parturition (F = 1.073: P = 0.859) between Monroe Mountain and Piceance Basin, respectively. The relatively low male/female ratio on Monroe Mountain was not associated with a protracted period of parturition. This finding suggests that relatively low male/female ratios typical of heavily harvested populations do not influence population dynamics because recruitment remains unaffected.© 2014 The Wildlife Society. Fine-scale genetic correlates to condition and migration in a wild cervid Joseph M. Northrup', Aaron B. A. Shafer2, Charles R. Anderson Jr.J, David W. Coltman~, and George Wittemyer 1 I Department offish, Wildlite, and Conservation Biology, Colorado State University, Fort Collins, CO, USA 2 Department of Evolutionary Biology, Evolutionary Biology Centre, lJppsala University, Uppsala, Sweden 3 Mammals Research Section, Colorado Parks and Wildlife, Grand Junction, CO, USA 4 Department of Biological Sciences, University of Alberta, Edmonton, AB, Canada. Citation: Northrup, J. M., A. B. Shafer, C. R. Anderson Jr., D. W. Coltman, and G. Whittemyer. 2014. Fine-scale genetic correlates to condition and migration in a wild cervid. Evolutionary Applications ISSN 1752-4571; doi: IO. l I I l/eva.12189 Abstract The relationship between genetic variation and phenotypic traits is fundamental to the study and management of natural populations. Such relationships often are investigated by assessing correlations between phenotypic traits and heterozygosity or genetic differentiation. Using an extensive data set compiled from free ranging mule deer (Odocoi/eus hemionus). we combined genetic and ecological data to (i) examine correlations between genetic differentiation and migration timing. (ii) screen for mitochondrial haplotypes associated with migration timing. and (iii) test whether nuclear heterozygosity was associated with condition. Migration was related to genetic differentiation (more closely related individuals migrated closer in time) and mitochondrial haplogroup. Body fat was related to heterozygosity at two nuclear loci (with antagonistic patterns). one of which is situated near a known fat metabolism gene in mammals. Despite being focused on a widespread panmictic species. these findings revealed a link between genetic variation and important phenotypes at a fine scale. We hypothesize that these correlations are either the result of mixing refugial lineages or differential mitochondrial haplotypes influencing energetics. The maintenance of phenotypic diversity will be critical to enable the potential tracking of changing climatic conditions, and these correlates highlight the need to consider evolutionary mechanisms in management. even in widely distributed panmictic species. 13 u �Landscape and anthropogenic features influence the use of auditory vigilance by mule deer Emma Lynth•, Joseph M. Northrupi., Megan F. MtKenna", Charles R. Anderson Jr.d, Lisa Angeloniu, and George Wittemyer.,i. •Graduate Degree Program in Ecology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA "Department offish, Wildlife and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523. USA "Natural Sounds and Night Skies Division, National Park Service, 1201 Oakridge Drive, Fort Collins, CO 80525, USA, JMammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526, USA "Department of Biology, Colorado State University, 1878 Campus Delivery, Fort Collins, CO 80523, USA Citation: Lynch, E., J.M. Northrup, M. F. McKenna, C.R. Anderson Jr., L. Angeloni, and G. Wittemyer. 2014. Landscape and anthropogenic features influence the use of auditory vigilance by mule deer. Behavioral Ecology; doi: I0.1093/beheco/aru 158. While visual fonns of vigilance behavior and their relationship with predation risk have been broadly examined, animals also employ other vigilance modalities such as auditory vigilance by listening for the acoustic cues of predators. Similar to the tradeofTs associated with visual vigilance. auditory behavior potentially structures the energy budgets and behavior of animals. The cryptic nature of auditory vigilance makes it difficult to study, but on-animal acoustical monitoring has rapidly advanced our ability to investigate behaviors and conditions related to sound. We utilized this technique to investigate the ways external stimuli in an active natural gas development field aITect periodic pausing by mule deer (Odocoileus hemionus) within bouts of rumination-based mastication. To better understand the ecological properties that structure this behavior. we investigate spatial and temporal factors related to these pauses to determine if results are consistent with our hypothesis that pausing is used for auditory vigilance. We found that deer paused more when in forested cover and at night. where visual vigilance was likely to be less effective. Additionally. deer paused more in areas of moderate background sound levels. though responses to anthropogenic features were less clear. Our results suggest that pauses during rumination represent a fonn of auditory vigilance that is responsive to landscape variables. Further exploration of this behavior can facilitate a more holistic understanding of risk perception and the costs associated with vigilance behavior. Migration Patterns of Adult Female Mule Deer in Response to Energy Development Charles R. Anderson Jr. and Chad J. Bishop Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, fort Collins, CO 80526, USA Citation: Anderson, C.R., Jr., and C. J. Bishop. 2014. Migration patterns of adult female mule deer in response to energy development. Pages 47-50 in Transactions of the 791h North American Wildlife & Natural Resources Conference (R. A Coon & M. C. Dunfee, eds.). Wildlifo Management Institute, Gardners, PA, USA. ISSN 0078-1355. Migration is an adaptive strategy that enables animals to enhance resource availability and reduce risk of predation at a broad geographic scale. Ungulate migrations generally occur along traditional routes. many of which have been disrupted by anthropogenic disturbances. Spring migration in ungulates is of particular importance for conservation planning because it is closely coupled with timing of parturition. The degree to which oil and gas development aITects migratory patterns. and whether ungulate migration is sufficiently prepared to compensate for such changes. has recently been investigated in Colorado and Wyoming (Lendrum et al. 2012, 2013; Sawyer et al. 2012). Lendrum et al. (2012, 2013) and Sawyer et al. (2012) address mule deer (Odocoileus hemionus) migration patterns in relation to energy development from northwest Colorado and south-central Wyoming. respectively. We address results from the Colorado and Wyoming studies and then compare similarities and differences. The interactions between migratory mule deer and energy development identi fled by Lendrum et al. (2012. 2013) and Sawyer et al. (2012) suggest mule deer may benefit from energy development planning by considering thresholds of development that may alter migratory behavior. It appears that migration rate. migration routes. and stopover use, if present. may be altered at high development intensities. In addition. migratory mule deer may benefit by maintaining security cover along migration paths. and improved habitat conditions may facilitate more direct and rapid migration requiring less energy to complete migration. Enhancing penneability along migration routes by applying dispersed development plans (<2 well pads/km2) and minimizing disturbance to vegetation types by maintaining security cover should reduce impacts to migratory mule deer as well as other migratory ungulates. Where feasible. habitat improvement projects on winter range and possibly stopover sites would also enhance migratory mule deer populations by enhancing energy reserves for long-distance movements and parturition shortly after summer range arrival. Where possible. directional drilling could be used to extract energy resources from underneath migration routes while maintaining no surface occupancy. Lastly. we emphasize that GPS studies now allow managers to accurately map migration routes for entire populations and identify relatively narrow corridors that are most heavily used thus allowing for the identification of the most important corridors for migrating ungulates. Where available. we encourage agencies to incorporate such migration corridors into land-use plans (e.g., resource management plans) and National Environmental Policy Act documents. 14 �Asynchronous vegetation phenology enhances winter body condition of a large mobile herbivore Kate R. Searle 1 • Mindy B. Rice2 • Charles R. Anderson 2 • Chad Bishop2 • N. T. HobbsJ 1 NERC Centre for Ecology and Hydrology, Bush Estate, Penicuik EI-126 0QB, UK i Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526, USA ) Department of Ecosystem Science and Sustainability, Colorado State University, Fort Collins 80524, CO, USA Citation: Searle, K. R., M. B. Rice, C.R. Anderson, C. Bishop and N. T. Hobbs. 2015. Asynchronous vegetation phenology enhances winter body condition ofa large mobile herbivore. Oecologia ISSN 0029-8549: DOI I0.1007/s00442-015-3348-9 Abstract Understanding how spatial and temporal heterogeneity influence ecological processes forms a central challenge in ecology. Individual responses to heterogeneity shape population dynamics. therefore understanding these responses is central to sustainable population management. Emerging evidence has shown that herbivores track heterogeneity in nutritional quality of vegetation by responding to phenological differences in plants. We quantified the benefits mule deer (Odocoileus hemionus) accrue from accessing habitats with asynchronous plant phenology in northwest Colorado over 3 years. Our analysis examined both the direct physiological and indirect environmental effects of weather and vegetation phenology on mule deer winter body condition. We identified several important effects of annual weather patterns and topographical variables on vegetation phenology in the home ranges of mule deer. Crucially. temporal patterns of vegetation phenology were linked with differences in body condition. with deer tending to show poorer body condition in areas with less asynchronous vegetation green-up and later vegetation onset. The direct physiological effect of previous winter precipitation on mule deer body condition was much less important than the indirect effect mediated by vegetation phenology. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer ,JOSEPH M. NORTHRUP', CHARLES R. ANDERSON JR. 2, and GEORGE WITTEMYER 1·J 'Department offish, Wildlilc and Conservation Biology, Colorado State University, Fort Collins, CO, USA 2 Mammals Research Section, Colorado Parks and Wildlife, Fort Collins, CO, USA ~Graduate Degree Program in Ecology, Colorado State University, Fort Collins, CO, USA Citation: Northrup, J.M., C.R. Anderson, Jr., and G. Wittemyer. 2015. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer. Global Change Biology, doi: IO. I I I l/gcb.13037 Abstract Extraction of oil and natural gas (hydrocarbons) from shale is increasing rapidly in North America. with documented impacts to native species and ecosystems. With shale oil and gas resources on nearly every continent, this development is set to become a major driver of global land-use change. It is increasingly critical to quantify spatial habitat loss driven by this development to implement effective mitigation strategies and develop habitat offsets. Habitat selection is a fundamental ecological process, influencing both individual fitness and population-level distribution on the landscape. Examinations of habitat selection provide a natural means for understanding spatial impacts. We examined the impact of natural gas development on habitat selection patterns of mule deer on their winter range in Colorado. We fit resource selection functions in a Bayesian hierarchical framework. with habitat availability defined using a movement-based modeling approach. Energy development drove considerable alterations to deer habitat selection patterns. with the most substantial impacts manifested as avoidance of well pads with active drilling to a distance of at least 800 m. Deer displayed more nuanced responses to other infrastructure, avoiding pads with active production and roads to a greater degree during the day than night. In aggregate. these responses equate to alteration of behavior by human development in over 50% of the critical winter range in our study area during the day and over 25% at night. Compared to other regions. the topographic and vegetative diversity in the study area appear to provide refugia that allow deer to behaviorally mediate some of the impacts of development. This study. and the methods we employed. provides a template for quantifying spatial take by industrial activities in natural areas and the results offer guidance for policy makers. mangers. and industry when attempting to mitigate habitat loss due to energy development. 15 �Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid Joseph M. Northrup', Charles R. Anderson Jr 2, and George WiUemyer 1·J Department offish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, CO, USA Mammals Research Section, Colorado Parks and Wildlife, Fort Collins, CO, USA JGraduate Degree Program in Ecology, Colorado State University, Fort Collins, CO, USA 1 2 Citation: Northrup, J.M., C.R. Anderson, Jr., and G. Wittemyer. 2016. Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid. Diversity and Distributions 22:547-557, DOI: I0.111 l/ddi.12417 ABSTRACT Aim The space an animal uses over a given time period must provide the resources required for meeting energetic needs. reproducing and avoiding predation. Anthropogenic landscape change in concert with environmental dynamics can strongly structure space-use. Investigating these dynamics can provide critical insight into animal ecology. conservation and management. Location The Piceancc Basin, Colorado, USA. Methods We applied a novel utilization distribution estimation technique based on a continuous-time correlated random walk model to characterize range dynamics of mule deer during winter and summer seasons across multiple years. This approach leverages secondorder properties of movement to provide a probabilistic estimate of space-use. We assessed the influence of environmental (cover and forage), individual and anthropogenic factors on intcrannual variation in range use of individual deer using a hierarchical Bayesian regression framework. Results Mule deer demonstrated remarkable spatial philopatry. with a median of50% overlap (range: 8-78%) in year-to-year utilization distributions. Environmental conditions were the primary driver of both philopatry and range size. with anthropogenic disturbance playing a secondary role. Main conclusions Philopatry in mule deer is suspected to reflect the importance of spatial familiarity (memory) to this species and, therefore, factors driving spatial displacement arc of conservation concern. The interaction between range behaviour and dynamics in development disturbance and environmental conditions highlights mechanisms by which anthropogenic environmental change may displace deer from familiar areas and alter their foraging and survival strategies. Movement reveals scale dependence in habitat selection of a large ungulate Joseph M. Northrup•, Charles R. Anderson Jr. 2, Mevin B. Hootenl, and George WiUemyer-4 Department offish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA Mammals Research Section, Colorado Parks and Wildlife, Fort Collins, Colorado 80523 USA JU.S. Geological Survey, Colorado Cooperative Fish and Wildlife Research Unit, Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA 4 Department offish, Wildlife and Conservation Biology and Graduate Degree Program in Ecology, Colorado State University, Fort Collins, Colorado 80523 USA 1 2 Citation: Northrup, J.M., C.R. Anderson, Jr., M. B. Hooten, and G. Wittemyer. 2016. Movement reveals scale dependence in habitat selection ofa large ungulate. Ecological Applications 26:2746-2757 Abstract. Ecological processes operate across temporal and spatial scales. Anthropogenic disturbances impact these processes. but examinations of scale dependence in impacts arc infrequent. Such examinations can provide important insight to wildlife-human interactions and guide management efforts to reduce impacts. We assessed spatiotemporal scale dependence in habitat selection of mule deer (Odocoi/eus hemionus) in the Piceance Basin of Colorado, USA. an area of ongoing natural gas development. We employed a newly developed animal movement method to assess habitat selection across scales defined using animal-centric spatiotemporal definitions ranging from the local (defined from five hour movements) to the broad (defined from weekly movements). We extended our analysis to examine variation in scale dependence between night and day and assess functional responses in habitat selection patterns relative to the density of anthropogenic features. Mule deer displayed scale invariance in the direction of their response to energy development features, avoiding well pads and the areas closest to roads at all scales, though with increasing strength of avoidance at coarser scales. Deer displayed scale-dependent responses to most other habitat features. including land cover type and habitat edges. Selection differed between night and day at the finest scales. but homogenized as scale increased. Deer displayed functional responses to development. with deer inhabiting the least developed ranges more strongly avoiding development relative to those with more development in their ranges. Energy development was a primary driver of habitat selection patterns in mule deer. structuring their behaviors across all scales examined. Stronger avoidance at coarser scales suggests that deer behaviorally mediated their interaction with development. but only to a degree. At higher development densities than seen in this area. such mediation may not be possible and thus maintenance of sufficient habitat with lower development densities will be a critical best management practice as development expands globally. 16 �Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus) u Kourtney F. Stonehouse 1•2, Charles R. Anderson Jr. 1, Mark E. Peterson•.z, and David R. Collins' 'Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 90526 USA ~Department ofFish, Wildlife and Conservation Biology, Colorado State University. Fort Collins, Colorado 80523 USA Citation: Stonehouse, K. F., C.R. Anderson Jr., M. E. Peterson, and D.R. Collins. 2016. Approaches to field investigations of cause-specific mortality in mule deer (Odoco;/eus hemionus). Colorado Parks and Wildlife Technical Report No. 48, First Edition, 317 W. Prospect Rd., Ft. Collins, CO USA. DOW-R-T-48-16, ISSN 0084-8883. This technical report provides general guidelines for conducting mortality site investigations to help investigators distinguish predation from scavenging and other causes of death. General health indices arc also provided to assess whether or not deer may have died from malnutrition or disease or if these factors may have predisposed deer to predation. Lastly. these guidelines will assist investigators in identifying predatory species or scavengers involved through the examination of physical evidence at deer mortality sites. The information presented here is based primarily on field experience gained from a long term research effort in northwest Colorado investigating mule deer mortality sites over several years (http://cpw.statc.co.us/learn/Pages/ResearchMammalsRP-04.aspx) and literature review where referenced. We acknowledge that proximate and ultimate cause of death can be difficult or impossible to detect from field necropsy alone and examples presented here largely represent proximate causes of mortality; efforts discerning ultimate cause will require specific tissue sample collections. where possible. submitted to a veterinary diagnostic laboratory. Within this technical report are numerous photographs documenting characteristics of predator attacks on mule deer and signs left by predatory and scavenging species. Additional pictures illustrate differences between healthy and unhealthy tissues and organs. While reading this document. be aware that each mortality investigation is unique and observations in the field may differ from illustrations provided here. Appendix I provides a sample necropsy form to assist in conducting mortality investigations. Reproductive success of mule deer in a natural gas development area Mark E. Peterson', Charles R. Anderson Jr.2, Joseph M. Northrup1,and Paul F. Doherty Jr. 1 'Department ofFish. Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA ~Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 90526 USA Citation: Peterson, M. E., C. R. Anderson Jr., J. M. Northrup, and P. F. Doherty Jr. 2017. Reproductive success of mule deer in a natural gas development area. Wildlife Biology, doi: 10.1111/wlb.00341 Abstract: Natural gas development is increasing across North America and causing concern over the potential impacts on wildlife populations and their habitat. particularly for ungulate species. Understanding how this development impacts reproductive success metrics that arc influential for ungulate population dynamics is important to guide management of ungulates. However. the influences of natural gas development on reproductive success metrics of mule deer Odocoileus hemionus have not been studied. We used statistical models to examine the influence of natural gas development and temporal factors on reproductive success metrics of mule deer in the Piceancc Basin, northwest Colorado during 2012-2014. We focused on study areas with relatively high or low levels of natural gas development. Pregnancy and in utcro fetal rates were high and statistically indistinguishable between study areas. Fetal survival rates increased over time and survival was lower in the high versus low development study areas in 2012 possibly influenced by drought coupled with habitat loss and fragmentation associated with development. Our novel results suggest managers should be concerned with the influences of development on fetal survival. particularly during extreme environmental conditions (e.g. drought) and our results can be used to guide development planning and/or mitigation. Developers and wildlife managers should continue to collaborate on development planning. such as implementing habitat treatments to improve forage availability and quality. minimizing disturbance to hiding and foraging habitat particularly during parturition. and implementing directional drilling to minimize pad disturbance density to increase fetal survival in developed areas. 17 u �V ariati on in ungulate body fat: individual versus temporal effects Eric J. Bergman•, Charles R. Anderson Jr. 1, Chad J. Bishop1, A. Andrew Holland', and Joseph M. Northrup 2 'Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 90526 USA 2Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA Citation: Bergman, E. J., C. R. Anderson Jr., C. J. Bishop, A. A. Holland, and J. M. Northrup.2018. Variation in ungulate body fat: individual versus temporal effects. Journal of Wildlife Management 82: 130-137, DOI: 10: 1002/jwmg.21334 ABSTRACT The use ofultrasonograhic measurements of muscle and body fat represent a relatively new data stream that can be used to address questions regarding ungulate condition. We have learned that measurements of body fat and presumably overall body condition among individual animals, even those taken from the same herd at that same time, are highly variable. Relatively little consideration has been given to the sources of variation in body fat and other physiological parameters in wildlife populations. We evaluated the components of variation in late-winter mule deer (Odocoileus hemionus) body fat estimates: sampling variation (i.e .. variation induced by the particular set of individuals that were sampled) and process variation (i.e .. variation stemming from biological processes) with a long-term data set (2002-2015) from Colorado. USA. We collected our data from across Colorado as part of historical research, ongoing research, and periodic population monitoring programs. Mean percent ingesta-free body fat (%1FBF) for sampled mule deer was 7.20 ± 1.20% (SD). Covariates related to individual deer explained approximately 4% of the total variation in %IFBF and annual effects explained an additional 13% of the variation. Substantial residual variation in %IFBF (83%) remained unexplained. The source of the 83% of unexplained variation is partially linked to line-scale spatial dynamics but also additional individual metrics we were unable to capture, primarily the presence or absence of dependent young. We speculate that the primary factors influencing late-winter mule deer body fat and overall condition are individual in nature. These results present a cautionary check on herd level inference that can be made from individual late-winter body fat estimates and we postulate that for mule deer, alternative and additional body condition metrics may offer added utility in management scenarios. However. an important next step to better understand wildlife population health is to evaluate the sources and magnitude of variation within other body condition metrics, with the goal of further refining data that can better allow biologists to incorporate herd health into population management recommendations. Mortality of mule deer fawns in a natural gas development area Mark E. Peterson•, Charles R. Anderson Jr. 2, Joseph M. Northrup', and Paul F. Doherty Jr. 1 'Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523 USA 2 Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 90526 USA Citation: Peterson, M. E., C.R. Anderson Jr., J.M. Northrup, and P. F. Doherty Jr. 2018. Mortality of mule deer fawns in a natural gas development area. Journal of Wildlife Management 82: 1135-1148, DOI: 10.1002/jwmg.21476 ABSTRACT Recent natural gas development has caused concern among wildlife managers, researchers, and stakeholders over the potential effects on wildlife and their habitats. Specifically, understanding how this development and other factors influence mule deer (Odocoileus hemionus) fawn (i.e., 0-6 months old) mortality rates, recruitment, and subsequently population dynamics have been identified as knowledge gaps. Thus, we tested predictions concerning the relationship between natural gas development, adult female, fawn birth, and temporal (weather) characteristics on fawn mortality in the Piceance Basin of northwestern Colorado, USA, from 2012-2014. We captured and radio-collared 184 fawns and estimated apparent cause-specific mortality in areas with relatively high or low levels of natural gas development using a multi-state model. Mean daily predation probability was similar in the high versus low development areas. Predation was the leading cause of fawn mortality in both areas and decreased from 0-14 days old. Black bear ( Ursus americanus; 22% of all mortalities, 11 = 17) and cougar (Fe/is concolor: 36% of all mortalities. 11 = 6) predation was the leading cause of mortality in the high and low development areas. respectively. Predation of fawns was negatively correlated with the distance from a female's core area to a producing well pad on winter or summer range. Contrary to expectations, predation of fawns was positively correlated with rump fat thickness of adult females. Well pad densities and development activity were relatively low during our study, indicating that the observed intensity of development did not appear to influence daily predation probability. Our results suggest maintaining development activity thresholds at levels we observed to potentially minimize the effects of development on fawn mortality. However, we caution that higher development intensity and drilling activity in flatter, less rugged areas with less concealment cover could influence fawn mortality. Managers should maintain low development densities in areas where topography and vegetation offer less concealment. Overall, region-specific data (e.g., development intensity, topography, predator assemblages, and associated predation risk) are needed to better understand the effects of natural gas development on fawn mortality. 18 �Using matern al mule deer movements to estimate timing of parturition and assist fawn captures .\ lark E. l'rterson' , Charil's R. Ande rson .fr.'. 1\ lathrw \\'. i\lldredgr'.and Paul F. Ooht·rt)' .Ir.' 'Dcpanmcm of Fish. Wild hie and Conscrvatmn Biology. Colorndo State U111vcrs1ty. Fort Collins. Colorado 80523 US!\ :Mammals Research Sct·t1on. Colorndo Parks and Wlldl1li:. 317 W Prospect Road. Fort Coll ms. CO 90526 US!\ Citation: Peterson, M E. C R. Anderson Jr.. M W. Alldredge. and I'. F. Doheny Jr 2018. Using maternal mule deer movements tu cstnnatc Liming of parturition uml assist fown enplurcs. Wild lilc Society Bulletin 42:616-621: DOI· IO 1002/wsb.935 ABSTRACT Movement pallerns of maternal ungulates have been used 10 determine parturition daies and aid in locating fawns. which may be im portant for unders1and ing reproduc1 ive rates (e.g.. pregnancy and fe tal). but such methods have not been validated for mule deer (Odocoileus hemionus). We first de1ennined timing of parturition using vaginal implant 1rnnsmi11crs (VfTs) and then predicted timing of parturition using VITs in conjunction \\ith Global Positioning System co llar data in the Piccancc Basin of north\\'estern Colorado. USA. during 20 12-201-1. We examined daily movcmcn1 rate to determine diflcrcnccs in movement rate among days (7 days pre- and postpartum) and li>r movement patlerns indicative or parturition. Mean daily movement rate (m/day) o r I02 maternal deer decreased by 46% from I day prcparlurition (mean = 1.253. SD = 1.09 1) 10 panurition date (mean = 682. SD = 574). and remained at this low rate 1-7 days postpanum. We applied an independent data set 10 va lidate predicted parturition dates based on daily movement rate. We estimated day of' parturition correctly (i.e .. day 0). within 1-3 days pos1parluri1 ion. and 2:4 days postparturition of lie Id-reported dates for IO (29%). 2 1 (60%). and 4 ( 11 %) materna l females. respectively. For novel data sets. we predict thm a mule deer female whose daily movcmcn l rate decreases by 2:46<Vo and remains low 2:3 days postparturition particularly when preceded by a sudden increase in 111ove111c111-has given birth. However. we caut ion that disturbance or deer hy field crews should be minimized. and il' bi11h sites arc 1101 found. neonatal mortality wil l be unclcrcstimaled. Our results can help determine timing and ~cncrnl location of parturition as an aid in capturing fawns when the use of'VITs is not fcasihlc. with the ultimate objective of estimating pregnancy. fetal. and fa" n survival rates if birth sites arc found. £, 20 18 The Wildlife Society. On-animal aco ustic monitoring provides insig ht to ungulate forag ing behavior .Joseph M. North r up', ;\lcxn ndrn i\vrin ', C hurlcs n. A nd erson, Jr.', Emma Brown', :111d George Willcmycr' 'Dcpanmcnt olTish. Wildlil'e and Conservation Biology. Colorado State University, Fon Collins, Colorado 80523 US/\ :Mammals Research Secuon. Colorado Parks and Wildlilc. 317 W Prospect Roud , Fon Collms, CO 90526 US/\ ' National Park Scrv1tc Natural Sounds and Night Skies Division. Fort Collins. CO 80525 USA Citation Nonhni[J. J M . i\ i\vrm. C.R. Anderson Jr . 1,. Bro\\n. and G Wittcmy~r 201 9 On-an11nal acoustic mo111tonng provides msight to ungulate roraging bchavwr Journal or Mammalogy I00: I479-1489. https·//doi orn/10 I093/unammalleH 12~ Abstract foraging behavior underpins many ecological processes: however. robust assessments or1h is behavior tor free-ranging animals are rare due 10 limitations 10 direct observat ions. We leveraged acoustic mon itoring and GPS tracking 10 assess the factors in0ucncing foraging behavior of' mulc deer (Odocoileus he111io1111s). We deployed custom-built acoustic collars with GPS radiornllars on mu le deer to measure lotalion-sp~cific foraging. We quantified individual bites and steps taken by deer. and quantified two metrics o r foraging behavior: lhe number of'bitcs taken per step and the number o r bites taken per uni t time. which re late to foraging intensity and ellicicncy. We fit statistical models 10 these metrics to examine the individual. environmental. and anthropogenic factors influencing foraging. Deer in poorer body condition took more bites per step and per minute and foraged for longer irrespecti ve or landscape properties. Other pallcrns varied sca,onally with major changes in deer condition. In December. when deer were in belier condition. they took fc w..:r bites per step and more bites per minute. Deer also foraged more intensely and efficiently in areas orgreater forage availabil ity and greater movement costs. During March. when deer were in poorer condition. foraging was not influenced by landscape features. Anthropogenic lilctors weakly sLructurcd fora ging behavior in December with no relationship in Murch. Most research on animal foraging is i111Crprc1cd under the framework of'optimal foraging theory. Depa11urcs from predictions deve loped under this framework providc insight 10 unrecognized factors influencing the evolut ion o r forug ing. Our results on ly con lim11ed 10 our predictions when deer were in lx:llcr condition and ecological conditions were declining. suggesting forag ing strategics were stale-dependent. These results advm1cc o ur understanding of forag ing pallcrns in wild an imals and highlight novel observational approaches for studying animal behavior. 19 �A noninvasive automated device for remotely collaring and weighing mule deer Chad J. Bishop1, Mathew W. Alldredge•, Daniel P. Walsh', Eric J. Bergman•, Charles R. Anderson Jr. 1, Darlene Kilpatrick1, Joe Bakel 2, and Christophe Fabvre 2 1 Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526 USA 2 Dynamic Group Circuit Design, Inc., 2629 Redwing Road, Fort Collins, CO 80525 USA Citation: Bishop, C. J., M. W. Alldredge, D. P. Walsh, E. J. Bergman, C. R. Anderson Jr., D. Kilpatrick, J. Bakel, and C. Fabvre.2019. A noninvasive automated device for remotely collaring and weighing mule deer. Wildlife Society Bulletin 43:717-725; doi.org/10.1002/wsb.1034 ABSTRACT Wildlife biologists capture deer (Odocoileus spp.) annually to attach transmitters and collect basic information (e.g .. animal mass and sex) as part of ongoing research and monitoring activities. Traditional capture techniques induce stress in animals and can be expensive, inefficient, and dangerous. They arc also impractical for some urbanized settings. We designed and evaluated a device for mule deer (0. hemionus) that automatically attached an expandable radiocollar to a ~-month-old fawn and recorded the fawn's mass and sex, without physically restraining the animal. The device did not require on-site human presence to operate. Students and faculty in the Mechanical Engineering Department at Colorado State University produced a conceptual model and early prototype. Professional engineers at Dynamic Group Circuit Design. Inc. in Fort Collins, Colorado, USA, produced a fully functional prototype of the device. Using the device, we remotely collared, weighed, and identified sex of 8 free-ranging mule deer fawns during winters 20I0-2011 and 2011-2012. Collars were modified to shed from deer approximately I month after the collaring event. Two fawns were successfully recollared after they shed the first collars they received. Thus. we observed IO successful collaring events involving 8 unique fawns. Fawns demonstrated minimal response to collaring events. either remaining in the device or calmly exiting. A fawn typically required :::1 weeks of daily exposure before fully entering the device and extending its head through the outstretched collar, which was necessary for a collaring event to occur. This slow acclimation period limited utility of the device when compared with traditional capture techniques. Future work should focus on device modifications and altered baiting strategies that decrease fawn acclimation period. and in tum. increase collaring rates, providing a noninvasive and perhaps cost-effective alternative for monitoring mid-to large-sized mammal species.© 2019 The Wildlife Society. Behavioral and demographic responses of mule deer to energy development on winter range Joseph M. Northrup', J.M., Charles R. Anderson Jr. 2, Brian D. Gerber\ and George Wittemyer1 1 Department offish, Wildlife and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523 USA 2 Mammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526 USA 3 Department of Natural Resources Science, University of Rhode Island, I Greenhouse Road, Kingston, RI 02881 USA Citation: Northrup, J.M., C. R. Anderson Jr., B. D. Gerber, and G. Wittemyer. 2021. Behavioral and demographic responses of mule deer to energy development on winter range. Wildlife Monographs 208: 1-37; 2021; DOI: 10.1002/wmon.1060. ABSTRACT Anthropogenic habitat modification is a major driver of global biodiversity loss. In North America. one of the primary sources of habitat modification over the last 2 decades has been exploration for and production of oil and natural gas (hydrocarbon development), which has led to demographic and behavioral impacts to numerous wildlife species. Developing effective measures to mitigate these impacts has become a critical task for wildlife managers and conservation practitioners. However. this task has been hindered by the difficulties involved in identifying and isolating factors driving population responses. Current research on responses of wildlife to development predominantly quantifies behavior, but it is not always clear how these responses scale to demography and population dynamics. Concomitant assessments of behavior and population-level processes are needed to gain the mechanistic understanding required to develop effective mitigation approaches. We simultaneously assessed the demographic and behavioral responses of a mule deer population to natural gas development on winter range in the Piceance Basin of Colorado. USA. from 2008 to 2015. Notably. this was the period when development declined from high levels of active drilling to only production phase activity (i.e., no drilling). We focused our data collection on 2 contiguous mule deer winter range study areas that experienced starkly different levels of hydrocarbon development within the Piceance Basin. We assessed mule deer behavioral responses to a range of development features with varying levels of associated human activity by examining habitat selection patterns of nearly 400 individual adult female mule deer. Concurrently. we assessed the demographic and physiological effects of natural gas development by comparing annual adult female and overwinter fawn (6-monthold animals) survival. December fawn mass. adult female late and early winter body fat. age. pregnancy rates. fetal counts. and lactation rates in December between the 2 study areas. Strong differences in habitat selection between the 2 study areas were apparent. Deer in the less-developed study area avoided development during the day and night. and selected habitat presumed to be used for foraging. Deer in the heavily developed study area selected habitat presumed to be used for thermal and security cover to a greater degree. Deer faced with higher densities of development avoided areas with more well pads during the day and responded neutrally or 20 �selected for these areas at night. Deer in both study areas showed a strong reduction in use of areas around well pads that were being drilled. which is the phase of energy development associated with the greatest amount of human presence. vehicle traffic, noise. and artificial light. Despite divergent habitat selection patterns. we found no effects of development on individual condition or reproduction and found no differences in any of the physiological or vital rate parameters measured at the population level. However, deer density and annual increases in density were higher in the low-development area. Thus, the recorded behavioral alterations did not appear to be associated with demographic or physiological costs measured at the individual level, possibly because populations are below winter range carrying capacity. Differences in population density between the 2 areas may be a result of a population decline prior to our study (when development was initiated) or area-specific differences in habitat quality,juvenile dispersal, or neonatal or juvenile survival; however, we lack the required data to contrast evidence lor these mechanisms. Given our results. it appears that deer can adjust to relatively high densities of well pads in the production phase (the period with markedly lower human activity on the landscape). provided there is sufficient vegetative and topographic cover afforded to them and populations are below carrying capacity. The strong reaction to wells in the drilling phase of development suggests mitigation efforts should focus on this activity and stage of development. Many of the wells in this area were directionally drilled from multiplewell pads. leading to a reduced footprint of disturbance. but were still related to strong behavioral responses. Our results also indicate the likely value of mitigation efforts focusing on reducing human activity (i.e .. vehicle traffic. light. and noise). In combination. these findings indicate that attention should be paid to the spatial configuration of the final development footprint to ensure adequate cover. In our study system. minimizing the road network through landscape-level development planning would be valuable (i.e .. exploring a maximum road density criteria). Lastly. our study highlights the importance of concomitant assessments of behavior and demography to provide a comprehensive understanding of how wildlife respond to habitat modification.© 2021 The Wildlife Society. Some memories never fade: inferring multi-scale memory effects on habitat selection of a migratory ungulate using step-selection functions Helena Rheault', Charles R. Anderson Jr. 2, Meagwin Bonar•, Robby R. Marrotte•, Tyler R. RossJ, Geroge Wittemyer\ and Joseph M. Northrup•.s 'Environmental and Life Sciences Graduate Program, Trent University, Peterborough, ON, Canada ~Mammals Research Section, Colorado Parks and Wildlife, Fon Collins, CO, United States )Depanment of Biology, York University, Toronto, ON, Canada ~Depanment of Fish, Wildlife and Conservation Biology, Colorado State University, Fon Collins, CO, United States, 5Ontario Ministry of Natural Resources and Forestry, Peterborough, ON, Canada u Citation: Rheault, II., C. R. Anderson Jr .. M. Bonar, R. R. Marrotte, T. R. Ross, G. Wittemyer, and J. M. Nonhrup. 2021. Some memories never fade: inferring multi-scale memory effects on habitat selection of a migratory ungulate using step-selection functions. Frontiers in Ecology and Evolution 9:702818; doi: I0.3389/fevo.2021.702810 ABSTRACT: Understanding how animals use information about their environment to make movement decisions underpins our ability to explain drivers of and predict animal movement. Memory is the cognitive process that allows species to store information about experienced landscapes. however. remains an understudied topic in movement ecology. By studying how species select for familiar locations. visited recently and in the past. we can gain insight to how they store and use local information in multiple memory types. In this study. we analyzed the movements of a migratory mule deer (Odocoileus hemionus) population in the Piceance Basin of Colorado. United States to investigate the innucncc of spatial experience over different time scales on seasonal range habitat selection. We inferred the innuence of short and long-term memory from the contribution to habitat selection of previous space use within the same season and during the prior year. respectively. We fit step-selection functions to OPS collar data from 32 female deer and tested the predictive ability of covariates representing current environmental conditions and both metrics of previous space use on habitat selection. inferring the latter as the influence of memory within and between seasons (summer vs. winter). Across individuals. models incorporating covariates representing both recent and past experience and environmental covariates performed best. In the top model. locations that had been previously visited within the same season and locations from previous seasons were more strongly selected relative to environmental covariates, which we interpret as evidence for the strong influence of both short- and long-term memory in driving seasonal range habitat selection. Further. the influence of previous space uses was stronger in the summer relative to winter. which is when deer in this population demonstrated strongest philopatry to their range. Our results suggest that mule deer update their seasonal range cognitive map in real time and retain long-term information about seasonal ranges. which supports the existing theory that memory is a mechanism leading to emergent space-use patterns such as site fidelity. Lastly. these findings provide novel insight into how species store and use information over different time scales. 21 u �Genomic correlates for migratory direction in a free-ranging cervid iVtaci:win Uonar1, Spcntcr,J. ;\1ukrson', C harles R. Anderson .Jr', George Witlcmycr 3, Joseph ~I. Northrup,. ·' and Aaron B. A. Shafer' 11':nv,ronmcntal & I.ire Sc iences Graduate Program. Trent University. Peterborough. Ontario. Canada K9L OG2 ' Mammals Research Sccuon, Colorado Parks and Wildlife, Fort Coll111s. CO 8()523. US/\ ' Department or Fish. Wildl1f'c and Conservation !1,ology. Colorado State lJ111vers1ty. Fon Collin,. CO 80513. USA 'W,ldhlc Research and Monitoring Section. Ontario Ministry or Natural Resources & Forestry. Peterborough. Ontario, Canada K9.I 3C7 Citation Bonar, M, S. J Anderson, C. R Anderson Jr. G. Wiltemyer. J. M. Northrup. and i\ H A. Shafer. 2022 Genomic correlates for migrutory direction in a frcc-rangmg ccrv,d Proceed ings 01'1hc Royal Society 13 289· 20221969 hltps//do, or0 l11 1098/rsph 2022 1969 ABSTRACT: Animal migrations arc some of the most ubiquitous and one of the most threatened ecological processes globally./\ wide range of migratory behu viours occur in nature. and this behaviour is not uniform among and within spl:cics. where even individuals in the same population can exhibit differences. While the environment largely drives migratory behaviour. it is necessary to understand the genetic mechanisms influcndng migration to elucidate the potential of migratory species to cope with novel conditions aml adapt to environmental change. In this study. we identified genes associated with a migratory trait by undertaking pooled genome-wide scuns on a natural population of migrating mule deer. We identified genomic regions associated with variation in migratory direction. including flTM I. a gene linked to the fonnation of lipids. and DPPA3. a gl:ne linked to epigenetic modifications of the maternal line. Such a genetic basis for a migratory trait co111 ribu1cs to the adaptive pmcntial of the species and might afTcl:I the flexibility of individuals to change their behaviour in the lace of changes in their environment. Plant and mule deer responses to pinyon-juniper removal by three mechanical methods Dnniclle Bilyeu J ohnston' and C harles R. Anderson .Jr.' 'Colorado Parks and Wildlife. 711 lndcpcndenl Avenue. Grand Junction. CO 81505, USA 1Colorndo Parks and Wildlitc. 317 Prospect Avenue. Fort Collins. CO 80526. lJSA C1ta\1on Johnston. D. 13, and C. R_Anderson Jr 2023 Plam and mule deer responses to Jllll} on-Juniper removal b) 1hrcl' mrcha111cal rncthOtb_ W1ldhfc Soc1c1y Oullct,n ~7:cl412 h11ps//c.lo1 org/10.1002/wsb 1~21 A bstn1ct Land managers in western North America often reverse succession by removing pin yon (/'inus spp.) and juniper (J1111ipern.1· spp.) trees to reduce fi re risk and increase forage Jor wildlife und livestock. Because prescribed fire carries inherent risks. mechanical methods such as chaining, roller-chopping, and mastication arc olicn used. Ml:chanical methods differ in cost and the size or wood) debris produced. and may diflc rcntially impact plum and animal rl:sponses. We i111plcmc111cd a rando111 i1.ccl. complete block. split-plot experiment in December 20 11 in the l'iceancc Basin. northwcslcrn Colorado. USA. to compare mcchnnirnl methods and to explore seeding (subplot) interactions. We assessed vegetation 1-, 2-, 5-, and 6-years post-treatment, and mule deer (Odocoileus he111io11us) response via GPS locations 3-8 years post-treatment. By 2016, treated plots had 3- 5 times higher perennial grass cover and - 10 times higher cheatgrass (Bro11111s recrorum) cover than untreated control plots. Rollcrchopped plots had both the highest non-native annual forb cover. and when seeded. the highest density ofbitterbrush (l'11rshia 1ride111ara). a nutritious shrub used by mule deer. Masticated plots had higher binerbrush use during summer and fa ll. leaving less lorngc avai lable for winter. Days of winter mule deer use from GPS po int locations in chained and rollerchopped plots was - 70% highl:r than in control plots. while wimcr use in masticated plots was similar to control plots. Mule deer use appears related to a combination of hiding cover. resulting from residual woody debris. and winter forage availability. Roller-chopped plots provide the best combination of hiding cover and winter forage. but mastication or chaining. applied leaving dispersed security cover. may be better options at large scah.:s or when invasive species concerns ex ist. 22 �Appendix B. Final Report to U. S. Department of Interior Bureau of Land Management: Developing a spatial planning tool for natural gas development on mule deer winter range. Developing a spatial planning tool for natural gas development on mule deer winter range Robby M. Marrotte, Department of Biological Sciences, Trent University Charles R. Anderson Jr., Mammals Research Section, Colorado Parks and Wildlife Joseph M. Northrup, Environmental and Life Sciences Graduate Program, Trent University Purpose Developing a spatial planning tool for natural gas development on mule deer winter range. Objectives Using existing data collected on mule deer in the Piceance Basin, Colorado, we developed a tool that allows land managers to assess the potential impacts of future hydrocarbon development on mule deer behaviour and populations. This project had two phases: I. Statistical modelling of movement data to optimize predictions of deer habitat selection. 2. Development of a user-friendly, web-based platform to assist in the development planning process by optimizing placement of infrastructure that minimizes disturbance to mule deer utilizing winter range. Yearly Summaries • 2021 o Data and covariate gathering and development ■ We cleaned mule deer location data and filtered transition and summer range data (Table I). ■ We defined the area of interest as the 4 study areas used by the mule deer in the Piceance Basin winter range addressed by Northrup et al. (2021 ): North Ridge, North Magnolia, South Magnolia, and Ryan Gulch (Figure 1). ■ We retrieved all necessary spatial and spatiotemporal data. Locations of roads, pipelines and facilities were digitized from National Agricultural Imagery Program (NAIP) imagery and ground truthed between 2010 and 2015. • Mule deer winter range study areas (Retrieved from Northrup et al. 2021 ). • Digital Elevation Model (DEM; Retrieved from https://earthexplorer.usgs.gov/). • Daily snow depth between 2009-2015 (Retrieved from Northrup et al. 2021). • Road network (Retrieved from Northrup et al. 2021 ). • Facility locations (Retrieved from Northrup et al. 2021 ). 23 �• • o Pipeline network (Retrieved from Northrup et al. 2021 ). Landsat 8 (LS8) imagery between 2012-2019 (Retrieved from Google Earth Engine). • Modis imagery between 2009-2019 (Retrieved from Google Earth Engine). • Location and daily status of wells between 2009-2019 (Retrieved from cogcc.state.co.us). Wells were grouped onto pads and pad boundaries were digitized using NAIP imagery. Status of the pad was assigned as the status of the well with the most active developmente.g., if there were two wells on a pad and one was producing gas and the other was being drilled, the status of the pad was set as "drilling." ■ We built static and spatiotemporal layers (Table 2). • Static layers o Digital Elevation Model (DEM) derived: Elevation, Terrain Ruggedness Index (TRI), Slope, Solar-radiation Aspect Index. o Climate: Long-term average and standard deviation of snow depth. o Roads: Density of roads within I 00-meter distance bands between 0.1-1 km. o Facilities: Density of facilities within I 00-meter distance bands between 0.1-1 km. o Pipelines: Density of pipelines within I 00-meter distance bands between 0.1-1 km. o LS8 derived: summer bands 1-7, summer NOVI, summer NOVI slope, winter bands 1-7, winter NOVI, winter NOVI slope. • Spatiotemporal layers o Modis derived: Biweekly NOVI, red, near-infrared, blue and middle-infrared. o Producing well pads derived: Daily density of producing well pads within I 00-meter distance bands between 0.1-1 km. o Drilling well pads derived: Daily density of producing well pad within I 00-meter distance bands between 0.1-1 km. Model development ■ We created the background (available) and habitat use (i.e., deer GPS collar) locations. We set a ratio of 30 background locations for each habitat use location within each range. We based the probability of each background location on the frequency of use locations during each day between 20092019. Consequently, if 5% of use locations were on January 2nd, 2008, the same proportion of background locations were assigned to this day. • We trained machine learning resource selection functions using Extreme Gradient Boosting (XGBoost; Chen et al. 2015) using 80% of the entire dataset. We used 20% of the training data (i.e., 16% of the entire dataset) for 24 �model validation to help guide and tune the hyperparameters. We then used the remaining 20% of the data for testing the accuracy out-of-sample. o Dashboard • We developed a dashboard using the R Shiny package and Leaflet interactive maps that can be used to predict the impact of the placement of well pads (Figure 2). • 2022 o o o o o o o We deployed a prototype of the application on a University of Rhode Island server (shiny.celsrs.uri.edu/bgerber/). We fixed bugs relating to version differences between the server and the previous infrastructure on which we built the application. We added the ability to predict the impact of roads in addition to well pads (Figure 3). We fine-tuned the artificial intelligence model to increase the model's predictive accuracy on validation data. We used the remaining data (testing data) to determine the weakness of the models across ranges and drilling periods (Table 3). We found that the model was generally capable of accurately predicting the drilling (2008-2012) and producing (2012-2019) periods. Comparatively, the model accuracy was lowest for North Magnolia during the producing period. We invited resource managers to test-run the application and used their feedback to make it more user-friendly. We added the ability to visualize the percent change of habitat use during the dri Hing and producing period (Figure 4 ). We finalized the model and application and uploaded a stable release of the application on the University of Rhode Island server. We wrote the first draft of a manuscript detailing the steps to create the application. We plan on submitting this manuscript to the Journal of Wildlife Management in December 2023. Background Oil and natural gas development has seen significant increases across North America since the turn of the century (USEIA 2015), bringing substantial environmental impacts to developed areas. In western North America, much of this development has overlapped with the ranges of wildlife species (Northrup & Wittemyer 2013). One species for which this development has generated significant concern is the mule deer. Mule deer are an important recreational and economic resource across the intermountain west but have seen large-scale population fluctuations over the last several decades (Unsworth et al. 1999), along with recent declines (Bergman et al. 2015). Hydrocarbon development in mule deer winter range elicits behavioural responses from deer, including relative reductions in the use of large areas in their winter range (Sawyer et al. 2006, 2009; Northrup et al. 2015, 2016a, 2016b ). During winter, deer have a negative energy balance, leading to declining conditions (Monteith et al. 2013) and occasional large-scale mortality events (White & Bartmann 1998). Thus, displacement from preferred areas or increased movements due to human activity could exacerbate these issues. Hydrocarbon extraction is projected to continue to increase for the next two decades (USEIA 2022), modifying substantial areas of new land, much of which will be in the 25 U �intermountain West (McDona ld et al. 2009). Considering this ongoi ng and impend ing development in the mule deer winter range. managers need a more in-depth understandi ng of the impacts of hydrocarbon development. A major need for land and wildli fe managers is spatial decision support too ls that incorporate currentl y ex isting knowledge on how hydrocarbon development impacts mule deer to allow for science-based development and mitigation planni ng. Specifical ly. managers need tools that can be used to determine how much development to allow in an area, where and when to allow development to proceed (e.g .. how to spatia lly configure development infrastructu re on the landscape to red uce impacts to critica l habitat), and the types of mitigation measures to implement to reduce the impacts o f deve lopment on mule deer. We leveraged ex isting large tempora l and spatial scale datasets on mu le deer habitat selection and demography fro m the Piceance Basin of Colorado to develop a spatial planning tool that can be used by managers in an adaptive management framework to plan development infrastructure and guide mitigation planning. We applied IO years or combined movement, survival. and populat ion abundance in fo rmation. Much of these data have been previously analyzed and been used to quantify behav ioral and demographic responses to energy development. We used this ex isti ng in fo rmation in conjunction with new analyses that focused on opti mizing our ability to predict the spatial responses of deer to energy development to produce a plan ning too l. The spatial planning tool that we developed wi ll allow land managers lo assess how the spatial pattern of proposed development would impact mule deer behavior on pinyon juniper winter range. Further, it prov ides estimates of the uncertainty in expected impacts to deer and the opportunity to expl ore impacts under varying winter and moisture conditions. We envision a user- friendly platform that would ultimately allow managers and developers the ability to optimize the development footprint such that impacts to deer populations and habitat can be mini mized. Applicability of Planning Tool and Next Steps The model underlying the shi nyapp developed for this proj ect was tit and tested using winter range data fro m the Piceance Basin. As such, the app is most valid for applicati on to the winter ranges in the Piceance Bas in from which the data originated. However, the mode l has potential utility outside of the Piceance Basin provided sufficient caution is taken in interpretation of the outputs. Several factors will influence how accurate the model is outside of the Piceance Bas in: I) the similarity of the habitat. including both vegetation and topography (e.g., topographica lly diverse dominated by pinion-juni per overstory), 2) the sim ilari ty of the development infrastructu re, and 3) deer density, which is directly linked to their use of habitat. In the coming months. we will directl y test the applicabil ity of the developed models to mule deer habitat use outside of the specific winter ranges within the Picea nce Basi n and using data completely outside of the Piceance Basin. Th is wi ll provide some guidance on the utility of the model for development planning elsewhere. Further. we plan to deve lop a companion tool that wiII allow users to apply the model elsewhere in Colorado. This too l will require user inputs for existing in frastructure of roads. well pads, pipelines. and fac ilities (e.g., compressor stations, gas plants). Prior to development of this com pan ion too l. resource ma nagers can contact Dr. Joseph Northru p at joe.northrup@gmaiI.com to discuss use outside of the Piceance Basin and coordinate application. Such application will again requi re user inputs fo r we ll pads, roads, pi peli nes and fac ilities. Further. caution in interpretation wi ll be needed. 26 �Tables Table 1. The number of individuals and GPS fixes for adult female mule deer monitored on winter range in the Piceance Basin, Colorado USA between December 2008 to March 2019. Number of Does Number of Fixes Winter North Magnolia North Ridge Ryan Gulch South Magnolia 2007-2008 2008-2009 2009-2010 2010-2011 2011-2012 2012-2013 2013-2014 2014-2015 2015-2016 2016-2017 2017-2018 2018-2019 Total 4 15 34 52 10 11 27 31 23 36 56 55 48 44 67 56 56 47 43 48 22 43 39 43 27 26 36 19 39 35 43 49 44 48 14 5 12 34 51 60 71 51 49 50 59 67 29 499 387 442 538 38 Total North Magnolia North Ridge Ryan Gulch South Magnolia Total 30 77 142 190 207 220 181 191 173 172 199 84 1,866 1,116 2,515 7,296 18,486 19,337 25,091 19,617 21,782 20,016 17,814 15,597 4,802 173,469 3,338 4,573 8,371 7,091 12,752 15,287 13,998 17,825 11,504 10,791 12,126 6,179 123,835 2,967 3,484 8,232 18,129 2,061 15,681 13,269 19,245 24,145 19,195 20,451 6,549 153,408 1,801 2,081 4,985 17,697 22,162 25,382 19,894 15,372 17,955 19,906 29,424 10,100 186,759 9,222 12,653 28,884 61,403 56,312 81,441 66,778 74,224 73,620 67,706 77,598 27,630 637,471 27 C C C �( ( ( Table 2. Habitat use prediction categories for mule deer does on their winter range in the Piceance Basin, Colorado, USA. Derived predictors Category Variation Elevation (m) Terrain ruggedness index (TRI) Cover Static Forage Static Description Sources Topographic based predictors Obtained from the United States Geological Survey https://earthexplorer .usgs.gov/ Daily modelled snow depth from 2008-2015 Obtained and derived by Liston and Elder (2006), Northrup et al. (2016b), Northrup et al. (2021) For each, median value for December-March 2013-2019 and June-September 2013-2019 USGS Landsat 8 Level 2, Collection 2, Tier 1 https://developers.google.com/earthengine/datasets/catalog/LAN DSAT_LC08_CO2_Tl_L2 Nearest value in time between 2009 and 2019 Obtained from the Google Earth Engine MOD13Ql.006 Terra Vegetation Indices 16-Day Global 250m https://developers.google.com/earthengine/datasets/catalog/MODIS_006_MOD13Q1 The density of roads for several distance bands Obtained from the United States Geological Survey Digitized from aerial imagery obtained from the National Agricultural Imagery Program https://earthexplorer.usgs.gov/ The density of pipelines for several distance bands Obtained from the White River Bureau of Land Management office and supplemented from aerial imagery obtained from the National Agricultural Imagery Program https://earthexplorer.usgs.gov/ Radiation Slope Mean Snow Depth Sd Snow Depth LB Bl Ultra Blue (0.435-0.451 µm) LB B2 Blue (0.452-0.512 µm) LB B3 Green (0.533-0.590 µm) LB B4 Red (0.636-0.673 µm) LB BS NIR (0.851-0.879 µm) Coverand forage Static LS 86 SIR 1 (1.566-1.651 µm) LS B7 SIR 2 (2.107-2.294 µm) NOVI NOVI Slope Modis Red (645nm) Modis NIR (858nm) Modis Blue (469nm) Modis MIR (2130nm/2105 2155nm) Coverand forage Spatiotemporal Modis NOVI Road density within 0-200, 200400, 400-600, 600-800, and 8001000 meters Pipeline density within 0-200, 200400, 400-600, 600-800, and 8001000 meters Anthropogenic Anthropogenic Static Static Static 28 �Facility density within 0-200, 200400, 400-600, 600-800, and 8001000 meters Pad density within 0-200, 200-400, 400-600, 600-800, and 800-1000 meters The density of natural gas facilities for several distance bands Anthropogenic Nearest value in time between 2009 and 2019 for density of drilling and producing well pads for several distance bands Anthropogenic Digitized from aerial imagery obtained from the National Agricultural Imagery Program https://earthexplorer.usgs.gov/ and validated on the ground Obtained from the Colorado Oil & Gas Conservation Commission cogcc.state.co.us Spatiotemporal Pad density within 0-200, 200-400, 400-600, 600-800, and 800-1000 meters 29 ( C ( �Table 3. Model accuracy(%) for adult female mule deer habitat use in 4 winter range study areas in the Piceance Basin, Colorado between 2008-2019. Sensitivity is the accuracy of the locations where mule deer were located from their GPS collars and specificity was the accuracy of the background availability data. Range Development All Ranges North Magnolia North Ridge Ryan Gulch South Magnolia All Ramies North Magnolia North Ridge Ryan Gulch South Magnolia All Ranges North Ma2nolia North Ridge Ryan Gulch South Magnolia Low/High Low None High High Period Trainine: 94.44 93.04 Sensitivity Validation 74.78 73.91 Testin~ 74.76 73.96 95.24 94.44 95.21 73.82 73.94 76.90 74.46 74.13 76.22 75.43 74.00 71.25 93.35 91.00 73.89 72.11 73.62 71.13 72.11 69.40 94.96 94.40 93.75 73.59 73.33 76.29 74.91 73.89 74.96 74.74 73.86 71.63 94.83 93.84 75.10 74.61 75.17 75.06 72.30 69.56 95.36 94.45 95.73 73.91 74.12 77.12 74.27 74.20 76.65 75.72 74.03 71.11 Drilling (20082012) Low/High Low None Hi,i;h High Producing (20122019) Low/High Low None High High Both (20082019) 30 Specificity 72.25 69.52 �Figures J O 10'11 J Q 05'1l - - - " '' ~ JO 00'11 -' / / ~ 39 95'tl 2 :, "' ...J 39 90'tl 39 85'11 I ~ I > I ' ' l \\ Nonh Magnolia I ~ I I I I l \ I I •,, Ryan Gulch South Magnolia ,--.. \; \ '' 39 80'11 108 JO'I\ I ' '- I ' .....I 108 35'\V 108 30'\V 108 25'\V 108 2o•w 108 15'\'V 108 ,o,w Longnude Figure I. Mu le deer winter range study areas in the Piceance Basin, Colorado, USA. The study areas are described in Northrup et al. (2021 ). Study areas contoured in red represent high development areas with numerous active natural gas wells and study areas contoured in black represent low development areas with few (North Magnolia) or no active natural gas wells (North Ridge). 31 �0 D-J Mule Ow / scnpW app • Shiny D X r-:tp:. '127.0.0.1-6291 Hydrocarbon Impact on Mule Deer PrtdlCI Impact R15fl Page (002) (0 2 0 ~, (OJ 06) (06 08) (0 81) Topography SUHI Mop ., Winier Ranges ~ WellPodS "' R~ds Figure 2. Mule deer hydrocarbon impact dashboard for predicting the impact of plac ing natural gas wells and roads within the Magnol ia, Ryan Gu lch, and North Ridge winter range study areas. The model was developed from mule deer GPS co llar data acquired between 2009- 2019. 32 �u Hydrocarbon Impact on Mule Deer • 0 - ~ · -- - Hydrocarbon lmpaC1 on Mule Deer Figu re 3. Example of placing natural gas well pads within the South Magnol ia winter range study area. A) Area of interest for well pad development. B) Placement of new well pads and service roads. 33 �.----Hydrocarbon Impact on Mule Deer ...... _ ..._ ..... .-----.. Hydrocarbon Impact on Mule Deer .----Hydrocarbon Impact on Mule Deer • ..__ ,._,,.. '"""""._.,.....,,,_""""' a ·--·· 0 • -;_.- • - Hydrocarbon Impact on Mule Deer Figure 4. Predicted habitat use by mule deer during the winter months during the drilling phase (A) and during the producing phase (B). Percent change in habitat use during the dri ll ing phase relative to pred icted map with no well pads; green shading indicates an increase in pred icted probabi lity of use, while purple shad ing indicates a decrease in predicted probability of use and (C) producing phase relative to predicted map with no well pads; green shading indicates an increase in predicted probability of use, while purple shading ind icates a decrease in predicted probability of use (D). Pred icted habitat use and change in habitat use are re lative to available habitat and scaled by category. Complete avoidance on ly occurred directly on well pads. The large apparent avoidance and change in habitat use apparent around well pads in the fi gures represent change in habitat use re lat ive to baseline, but are not complete avoidance of the areas. Note that because the percent change is relative, apparentl y large percent changes can occur with sma ll absolute change; e.g., a change from 0.0 I to 0.02 is small in abso lute terms but would be a I00% increase in probabi Iity of use. 34 �References Bergman. E.J ., Doherty. P.F., White, G.C. and Holland, A.A., 20 15. Density dependence in mule deer: a rev iew of evidence. Wild Ii fe Biology, 21 ( I), pp. 18-29. Chen, T., He, T .. Benesiy. M., Khotil ov ich. V.. Tang, Y.. Cho. H. and Chen, K. , 20 15. Xgboost: extreme gradient boosting. R package version 0.4-2. I(4), pp.1 -4. Northrup, J.M. and Wittemyer. G .. 20 13. Characterising the impacts of emerging energy development on wild Ii fe. with an eye towards mi tigation. Ecology letters, 16( I), pp. I I2125. Northrup, J.M .. Anderson Jr, C. R. and Wittemyer, G., 20 15. Quantifying spatial habitat loss from hydrocarbon development through assessi ng habitat selection patterns of mu le deer. Global change biology, 2 1( 11 ), pp.3961 -3970. Northrup. J.M.. Anderson Jr, C. R.. Hooten, M.B. and Wittemyer, G., 20 16a. Movement reveals scale dependence in habitat selection of a large ungulate. Ecological Applications, 26(8), pp.2746-2757. Northrup. J.M., Anderson Jr, C. R. and Wittemyer, G., 20 16b. Environmenta l dynamics and anthropogenic development alter philopatry and space-use in a North American cervid. Diversity and Distributions. 22(5). pp.547-557. Northrup. J.M., Anderson Jr. C. R. , Gerber, B.D. and Wiltemyer. G., 202 1. Behavioral and demograph ic responses of mule deer to energy development on wi nter range. Wild life Monographs. 208( I), pp. 1-3 7. McDonald. R.I., Fargione, J., Kiesecker, J., Miller. W.M. and Powell , .I ., 2009. Energy sprawl or energy effi ciency: climate policy impacts on natural habitat for the United States of America. PloS one. 4(8). p.e6802. Monteith. K.L.. Stephenson. T.R.. Bleich. V.C.. Conner, M.M., Pierce, B.M. and Bowyer, R.T.. 2013. Risk-sensitive allocation in seasonal dynamics of fat and protein reserves in a longli ved mammal. Journal of Animal Ecology. 82(2). pp.377-388. Sawyer. H., Kauffman. M..J . and Nielson. R.M .. 2009. Influence of well pad activity on winter habitat se lection patterns of mule deer. The Journal of Wildlife Management, 73(7), pp. I052- 106 1. Sawyer. H.. Nielson. R.M .. Lindzey. F. and McDonald. L.L.. 2006. Winter habitat se lection of mule deer before and during development of a natural gas field. The Journal of Wildlife Management. 70(2). pp.396-403. United States Energy In formation Administration (USE IA).20 15. Crude Oil and Natural Gas Exploratory and Deve lopment Wells. http://www.eia.gov/dnav/ng/NG ENR WELLEND SI A.htm United States Energy In formation Adm inistration (USE IA). 2022. Annual Energy Outlook 2022. https ://www .eia.gov/outlooks/arch ive/aeo2 I/pd f/ A EO Narrative 202 1.pd f Unsworth. .I . W., Pac. D.F .. White. G.C. and Bartmann , R.M .. 1999. Mule deer surv ival in Colorado, Idaho. and Montana. The Journal of Wi ldli fe Management, pp.3 15-326. White. G.C. and Bartmann, R.M .. 1998. Effect of density reduction on overwinter survival of freeranging mule deer fa wns. The Journal of wild life management, pp.21 4-225. 35 � https://cpw.cvlcollections.org/files/original/eb70939c35cfe6de79639420bdf5e1e7.pdf cafa5849028cd3d2bec66ccdf165940c PDF Text Text Colorado Parks and Wildlife July 2018 – June 2019 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Colorado 3430 3002 Task No. Federal Aid Project No. 1 W-242-R-3 : Parks and Wildlife : Mammals Research : Evaluating factors influencing elk recruitment in Colorado : : Period Covered: July 1, 2018 – June 30, 2019 Authors: N.D. Rayl, M.W. Alldredge, and C.R. Anderson Personnel: V. Ashby, B. Banulis, N. Bealer, T. Bonacquista, K. Bond, M. Caddy, K. Crane, D. Collins, R. Delpiccolo, J. Dewhirst, K. Duckett, R. de Vergie, W. de Vergie, R. Ebel-Childs, J. Ennis, D. Finley, P. Firmin, M. Fisher, A. Fowler, K. Fox, W. Hiler, B. Holder, E. Jones, A. Kirby, J. Lambert, S. Lambert, A. Larson, D. Lewis, K. Logan, K. Mahaffie, K. Middledorf, M. Miller, C. Murray, B. Nance, W. O’Malley, A. Orlando, M. Ortega, J. Pollock, N. Renneker, M. Richman, M. Shaw, I. Smith, J. Stanton, M. Swaro, L. Sweanor, J. Taylor, L. Temple, M. Trujillo, G. Tuck, A. Vitt, S. Wagner, C. Wallace, N. Waring, K. Yeager, J. Yost, E. Sawa, and L. Wolfe, CPW; J. Clark, J. Kelley, R. Swisher, S. Swisher, A. Orlando, Quicksilver Air, Inc.. Project support received from Federal Aid in Wildlife Restoration, Rocky Mountain Elk Foundation, CPW Big Game Auction and Raffle, and Bar NI/Cabot Foundation. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Over the last two decades, wildlife managers in Colorado have become increasingly concerned about declining winter elk calf recruitment (estimated using juvenile:adult female ratios) in the southern portion of the state. Although juvenile:adult female ratios are often highly correlated with juvenile elk survival, they are an imperfect estimate of recruitment because they are affected by harvest, pregnancy rates, juvenile survival, and adult female survival. Thus, there is a need for elk research in Colorado based upon monitoring of marked individuals to evaluate factors affecting each stage of production and survival. In 2016, Colorado Parks and Wildlife (CPW) began a 2-year pilot study to investigate factors influencing elk recruitment in 2 study areas in the state. In FY2018-19, CPW expanded this pilot study work into a 3rd study area and for 6 additional years to better determine how predators, habitat, and weather conditions are impacting elk recruitment in Colorado. This progress report covers the 1st half of the 1st field season of this new project (March 2019-June 2019). During this past fiscal year, we focused on working with stakeholders and collaborators on research logistics, and capturing and collaring elk. Field efforts were 1 �centered on 2 objectives: 1) capturing adult female elk, and collaring and outfitting pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior, and 2) capturing and collaring newborn elk to collect data on calf survival and cause-specific mortality. We captured 71 adult female elk and radio-collared 62 pregnant elk and outfitted them with VITs. Estimates of average pregnancy rates ranged from 86-100% across herds, and estimates of mean ingesta-free body fat ranged from 5.8-7.1% across herds. During the 2019 calving season, we captured and collared 146 elk calves, including 87% (54 of 62 calves) of the calves born to collared females. Averaged across herds, the average date of calving was June 2nd. 2 �WILDLIFE RESEARCH REPORT EVALUATING FACTORS INFLUENCING ELK RECRUITMENT IN COLORADO NATHANIEL D. RAYL, MAT W. ALLDREDGE, AND CHUCK R. ANDERSON JR. PROJECT NARRATIVE OBJECTIVES The objectives of this project are to 1) estimate elk calf survival and cause-specific mortality rates from birth to age 1 to evaluate the importance of mortality sources for elk calf survival, 2) evaluate the influence of biotic (birth date, birth mass, gender, maternal body condition, habitat conditions) and abiotic factors (previous and current weather conditions) on seasonal mortality risk of elk calves from birth to age 1, 3) assess the health of elk herds by quantifying pregnancy rates and percent ingesta-free body fat (IFBF) of adult female elk, and 4) evaluate the influence of biotic (age, body condition, reproductive status, habitat conditions and selection) and abiotic factors (previous and current weather conditions) on pregnancy rates and IFBF. SEGMENT OBJECTIVES 1. Work with personnel from CPW Areas 6, 10, 11, and 18, and private landowners on field research logistics. 2. Capture adult female elk, and collar and outfit pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior. 3. Capture and collar newborn elk to collect data on calf survival and cause-specific sources of mortality. INTRODUCTION In Colorado, elk (Cervus canadensis) are an important natural resource that are valued for ecological, consumptive, aesthetic, and economic reasons. In 1910, fewer than 1,000 elk remained in Colorado (Swift 1945), but today the state population is estimated to be the largest in the country, with approximately 282,000 elk. Over the last two decades, however, there has been increasing concern among wildlife managers in Colorado about declining winter calf recruitment (estimated using juvenile:adult female ratios) in the southern portion of the state (Fig. 1; Colorado Parks and Wildlife, unpublished data). In many elk populations, similar declining trends in juvenile recruitment have been observed. A recent synthesis examining elk recruitment in the western United States from 19892010 found evidence of a long-term reduction of 0.48 juveniles/100 adult females/year (Lukacs et al. 2018). Lukacs et al. (2018) discovered associations between recruitment and forage productivity that suggested nutritional conditions on either summer or winter ranges had the most influence on elk recruitment. These associations varied by geographic region: winter range conditions appeared to be more influential in southern areas (Colorado, Utah, and parts of Wyoming), whereas summer range conditions appeared to be more influential in northern areas 3 �(Idaho, Montana, Oregon, Washington, parts of Wyoming). Lukacs et al. (2018) also found that lower total winter precipitation in the previous winter was associated with lower recruitment the next year. The presence of wolves (Canis lupus) and grizzly bears (Ursus arctos) were also associated with lower recruitment. Although juvenile:adult female ratios are often highly correlated with juvenile elk survival (Raithel et al. 2007, Harris et al. 2008), they are an imperfect estimate of recruitment because they are affected by harvest, pregnancy rates, juvenile survival, and adult female survival (Caughley 1974, Gaillard et al. 2000, Harris et al. 2008, DeCesare et al. 2012, Lukacs et al. 2018). This makes it difficult to identify ultimate factors influencing population dynamics using age ratio data alone, as multiple scenarios can produce equivalent ratios (Caughley 1974). Thus, long-term demographic studies based on monitoring of marked individuals are necessary to reliably test biological hypotheses and evaluate factors affecting each stage of production and survival (Gaillard et al. 2000, Clutton-Brock and Sheldon 2010, Proffitt et al. 2014). In the absence of harvest, the population dynamics of ungulates are normally characterized by high and stable adult female survival and variable juvenile survival (Gaillard et al. 1998, 2000). Among vital rates, changes in adult female survival have the potential to exert the most influence on population growth rates, but usually do not because of low year-to-year variability (Gaillard et al. 1998, 2000). Instead, adult female survival is typically buffered against moderate environmental variation, and thus not strongly influenced by climatic or densitydependent factors (Gaillard et al. 2000). Indeed, in a large-scale meta-analysis, Brodie et al. (2013) demonstrated that adult female elk survival was principally related to human harvest, and found that this harvest was an additive source of mortality. In contrast to adult female survival, and despite its lower elasticity (de Kroon et al. 1986), juvenile survival frequently has a greater effect on population growth rates of ungulates because of its high variability (Gaillard et al. 1998, 2000, Raithel et al. 2007). Juvenile survival of ungulates may be influenced by abiotic or biotic factors such as environmental conditions, forage quality or quantity, population density, maternal body condition, and predation (Barber-Meyer et al. 2008, Griffin et al. 2011, Monteith et al. 2014, Bastille-Rousseau et al. 2016). Complex interactions among these factors frequently make it difficult to identify the relative role of top-down and bottom-up factors affecting calf survival (Linnell et al. 1995). Further complexity may be introduced if top-down and bottom-up forces are simultaneously and variably influencing survival (Bowyer et al. 2005, Monteith et al. 2014). Juvenile survival has been found to vary substantially among and within elk populations on an annual basis (Garrott et al. 2003, Raithel et al. 2007, Griffin et al. 2011). Griffin et al. (2011) synthesized elk neonatal survival across 12 populations in the northwestern United States, and found that survival declined after warmer previous summers and with more predator species, and increased with higher May precipitation. In resource-limited populations, weather conditions may heavily influence juvenile survival. For example, in an unharvested elk population, prior to the establishment of wolves, Garrott et al. (2003) found that the dominant source of calf mortality for an unharvested elk population in Yellowstone National Park was starvation, with more severe winters leading to lower calf survival. Elsewhere, mortality due to predation has been identified as the most significant cause of death for elk calves (e.g., Barber-Meyer et al. 2008). In some systems black bears (Ursus americanus) are the dominant predator (White et al. 2010, Tatman et al. 2018), whereas in others mountain lions (Puma concolor) kill the most calves (Eacker et al. 2016). Although non-predation deaths (i.e., disease, starvation) may be low where high levels of juvenile predation occur, it remains difficult to determine whether neonate 4 �predation represents an additive or compensatory source of mortality (Linnell et al. 1995, Barber-Meyer et al. 2008, Griffin et al. 2011). Individual calf characteristics may also influence risk of mortality. To date, evidence for sex-biased mortality in elk calves is equivocal. Some studies have documented increased vulnerability of male calves to predation (Smith and Anderson 1996, Eacker et al. 2016), whereas others have demonstrated increased vulnerability of female calves (White et al. 2010). It has been suggested that differences may be due to the hunting behavior of the dominant predator, with stalking predators, such as mountain lions, disproportionately killing male calves, which may engage in riskier exploratory behavior (Eacker et al. 2016). Conclusions about the effect of birth mass on elk calf survival are similarly ambiguous. Some studies have reported that birth mass influenced survival probability (Singer et al. 1997, White et al. 2010), but others have not detected an influence of birth mass on neonate mortality (Smith and Anderson 1996, Eacker et al. 2016). The ability for nutritional resources on the landscape to support ungulate populations is reflected in the nutritional condition of individuals within those populations (Parker et al. 2009, Cook et al. 2013, Monteith et al. 2014). As nutritional resources decline there is typically a predictable sequence of changes in the vital rates of large herbivore populations: first, juvenile survival decreases, then age of first reproduction increases, followed by decreased fertility of adult females, and finally increased mortality rates of adults (Eberhardt 1977a, 1977b, 2002, Gaillard et al. 1998, 2000). This general sequence has been confirmed for elk, with forage quality and associated nutritional body condition demonstrated to affect intra-uterine survival, yearling and adult pregnancy rates, birth dates, birth weight, calf growth, and winter survival of juveniles and adults (Cook et al. 2004a, 2004b, 2013, Proffitt et al. 2016). When nutritional resources are limited, elk calves may be lighter at birth, be born later, and have slower growth rates during summer, which may predispose them to predation mortality. Cook et al. (2004a, 2013) demonstrated that summer-autumn nutrition can play a central role in determining elk productivity, as, during this time, animals must meet the demands of lactation and accrue sufficient fat to get pregnant and survive the winter. For elk, nutritional resources on the landscape may be affected by wild and domestic herbivory (Vavra and Sheehy 1996, Vavra et al. 2007), timber management (Visscher and Merrill 2009), climatic conditions (Middleton et al. 2013), and fire history (Proffitt et al. 2016). To properly determine factors affecting juvenile ungulate survival bottom-up nutritional effects and top-down predation effects should be evaluated together (Monteith et al. 2014). Understanding how these effects are influencing elk population dynamics in Colorado is critical for guiding management actions. In 2016, Colorado Parks and Wildlife (CPW) initiated a 2-year pilot study to investigate factors influencing elk recruitment in 2 study areas in the state. During the pilot study, researchers collected data on annual elk calf survival, pregnancy rates, and latewinter body condition of adult female elk. In July 2018 – July 2019, we expanded this research into a 3rd study area to better determine how predators, habitat, and weather conditions are impacting elk recruitment in Colorado, and to provide management recommendations for increasing juvenile recruitment. STUDY AREAS For management purposes, the elk population in Colorado is divided into 43 Data Analysis Units (DAUs), each of which encompass the year-round range of an elk herd. This 5 �project is being conducted in 3 DAUs, 2 with low juvenile:adult female ratios (E-20, E-33), and 1 with high juvenile:adult female ratios (E-2), which will serve as a reference area (Fig. 1). The Uncompahgre Plateau elk herd (DAU E-20; 5,858 km2) is on the Uncompahgre Plateau in southwest Colorado, USA. In 2017, the post-hunt population of the Uncompahgre Plateau herd was estimated to be ~8,500 elk. From 2013-2017 juvenile:adult female ratios in E20 averaged 30 calves per 100 adult females. Landownership is a mixture of BLM (38%), USFS (37%), private (24%), and state (1%) lands. Elevations range from 1,390 to 3,150 m. The plateau is characterized by a mixture of pinyon-juniper (Pinus edulis, Juniperus osteosperma) woodlands and sage-grassland communities (Artemisia spp., Cercocarpus montanus, Achnatherum hymenoides) at lower elevations. At mid elevations, ponderosa pine (Pinus ponderosa) and mountain shrub communities (Amelanchier alnifolia, Arctostaphylos, Artemisia spp., Quercus gambelii, Symphoricarpos spp.) predominate. Spruce-fir (Picea engelmannii, Abies lasiocarpa, Pseudotsuga menziesii) and aspen (Populus tremuloides) forests dominate at higher elevations. The Trinchera elk herd (DAU E-33; 8,601 km2) is in southeast Colorado, USA. In 2017, the post-hunt population of the Trinchera herd was estimated to be ~16,600 elk. From 2013-2017 juvenile:adult female ratios in E-33 averaged 26 calves per 100 adult females. Landownership is a mixture of private (89%), USFS (3%), state (3%), BLM (2%), U.S. Fish and Wildlife Service (USFWS; 1%), and other (2%) lands. Elevations range from 1,640 to 4,370 m. Lower elevations are characterized by agriculture and sage-grassland communities. At mid elevations, pinyonjuniper woodlands, ponderosa pine and mountain shrub forests, and spruce-fir and aspen forests predominate. Alpine tundra communities dominate at higher elevations. The Bear’s Ears elk herd (DAU E-2; 7,293 km2) is in northwest Colorado, USA. In 2017, the post-hunt population of the Bear’s Ears herd was estimated to be 20,000-24,000 elk. From 2013-2017 juvenile:adult female ratios in E-2 averaged 56 calves per 100 adult females. Landownership is a mixture of private (50%), USFS (25%), BLM (19%), state (5%) and other (1%) lands. Elevations range from 1,730 to 3,710 m. The DAU is characterized by sagegrassland communities at lower elevations. At mid elevations, mountain shrub communities predominate. Spruce-fir and aspen forests dominate at higher elevations. METHODS Capture and handling — We captured adult female elk ≥2 years of age from each study herd by helicopter net-gunning during late winter (March). During capture, we marked individuals with ear tags, collected a blood sample, and measured hind foot length, chest girth, and age based on tooth eruption and wear patterns. We used a portable ultrasound machine to assess whether or not captured elk were pregnant, and estimated the percent of ingesta-free body fat (IFBF) following methods detailed in Cook et al. (2010). We verified non-pregnancies using pregnancy-specific protein B (PSPB) analysis of sampled blood. From each study herd, we outfitted pregnant elk with vaginal implant transmitters (VITs) and Global Positioning System (GPS) radio-collars that attempt to acquire a location every 2 h. We deployed VITs that use the satellite communication capabilities of the collar on the adult female to send a notification when the VIT is expelled, signifying a birth. After receiving a birth notification from a VIT, we went to the birth site to capture and collar the newborn elk calf. We blindfolded calves, and wore latex gloves to minimize the transfer of human scent. We measured body mass, hind foot length, chest girth, and determined the gender of captured calves. We outfitted elk neonates with expandable GPS radio-collars or 6 �VHF proximity collars that communicated with the collar on the adult female, and are designed to drop off after 12 months. We also opportunistically located, captured, and collared additional neonates to increase sample sizes. We collected additional measurements (hair moisture, incisor and upper canine eruption, hoof, dew claw, and navel condition) from opportunistically captured calves to estimate age at capture following Johnson (1951) and Eacker (2015). We attempted to handle calves for <5 minutes to minimize stress. During all captures, we followed CPW’s animal care and use guidelines for capturing and handling elk (CPW ACUC #09-2008). Cause-specific mortality — Within 24 hours of detecting a mortality signal from an elk collar, we attempted to conduct a systematic field investigation to determine the cause of death. We searched the area surrounding kill sites for evidence of predator presence, including predator scats, tracks, and hair, or signs of a struggle (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). We examined elk calf carcasses for evidence of canine puncture wounds, subcutaneous hemorrhaging and bruising, aspirated blood in the mouth, nose, or trachea, claw or bite marks on the hide, cracked or chewed bones, and characteristic consumption patterns (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). We also collected calf carcasses when they were available to verify field assessments with laboratory necropsies performed by a CPW veterinarian. Nutritional condition of adult female elk — The body fat of lactating and non-lactating adult female elk can vary substantially, as lactating females are more sensitive than non-lactating females to their nutritional environment (Cook et al. 2004a, 2013). Therefore, it is difficult to interpret the body condition of adult female elk in late winter without knowing whether or not they experienced the energetic demands of lactation throughout the previous growing season (Cook et al. 2004a, 2013). We will use the late-winter body condition of prime-aged adult female elk that successfully raised a calf the previous year to assess whether or not our study herds may be experiencing nutritional limitations. RESULTS AND DISCUSSION During March 2019, we captured a total of 71 adult female elk by helicopter net-gunning, 3 from the Bear’s Ears herd, 37 from the Trinchera herd, and 31 from the Uncompahgre Plateau herd. We radio-collared 62 pregnant elk and outfitted them with VITs, 30 each from the Trinchera and Uncompahgre Plateau herds, and 2 from the Bear’s Ears herd. We exceeded the mortality threshold established by our ACUC protocol due to acute mortalities that occurred during the capture process (4 from the Trinchera herd, 1 from the Bear’s Ears herd). Therefore, we ceased capture operations prior to reaching our target sample of 30 collared pregnant female elk in the Bear’s Ears herd. In 2019, we estimated that pregnancy rates of adult female elk were 100% in the Bear’s Ears herd (95% CI = 44-100%; n = 3), 91% in the Trinchera herd (95% CI = 76-97%; n = 33), and 97% in the Uncompahgre Plateau herd (95% CI = 84-100%; n = 31; Fig. 2). Elk populations experiencing good to excellent summer-autumn nutrition typically have pregnancy rates ≥90% (Cook et al. 2013). We estimated the mean IFBF of adult female elk to be 5.8% from the Bear’s Ears herd, 6.4% from the Trinchera herd, and 7.1% from the Uncompahgre Plateau herd (Fig. 3). When late-winter IFBF values are <8-9% for adult female elk that have lactated through the previous 7 �growing season, this suggests that there may be nutritional limitations, but it does not identify whether limitations are a result of summer-autumn or winter nutrition (R. Cook, personal communication). During May and June 2019, we captured and collared 146 elk calves, 51 from the Bear’s Ears herd, 46 from the Trinchera herd, and 49 from the Uncompahgre Plateau herd. From the Bear’s Ears herd, we successfully captured and collared 100% (2/2) of the calves of collared adult female elk outfitted with VITs. From the Trinchera herd, we successfully captured and collared 90% (27/30) of the calves of collared adult female elk outfitted with VITs. From the Uncompahgre Plateau herd, we successfully captured and collared 83% (25/30) of the calves of collared adult female elk outfitted with VITs. The estimated mean date of calving was June 11 in the Bear’s Ears herd, June 1 in the Trinchera herd, and June 3 in the Uncompahgre Plateau herd (Fig. 4). SUMMARY During FY18-19 we successfully worked with private landowners and personnel from CPW to coordinate field research logistics and initiate the first year of this study. We collected data on body condition and reproduction by capturing adult female elk, and we outfitted 62 pregnant females with GPS collars and VITs. We did not reach our target sample size of 30 collared pregnant females from the Bear’s Ears herd because we halted capture operations due to acute mortalities that occurred during helicopter net-gunning. As a result, we had to adjust our sampling strategy for elk calves in this area to capture a greater number of opportunistically encountered calves due to the low number of calves available to capture from collared adult female elk. We successfully captured and collared >45 newborn elk from each study area, meeting our sample size objective, and allowing us to collect data on calf survival and causespecific sources of mortality. We will continue to collect data on elk survival and cause-specific sources of mortality throughout the year. LITERATURE CITED Barber-Meyer, S. M., L. D. Mech, and P. J. White. 2008. Elk calf survival and mortality following wolf restoration to Yellowstone National Park. Wildlife Monographs 169:1–30. Bastille-Rousseau, G., J. A. Schaefer, K. P. Lewis, M. A. Mumma, E. H. Ellington, N. D. Rayl, S. P. Mahoney, D. Pouliot, and D. L. Murray. 2016. Phase-dependent climate-predator interactions explain three decades of variation in neonatal caribou survival. Journal of Animal Ecology 85:445–456. Bowyer, T., D. K. Person, and M. P. Becky. 2005. Detecting top-down versus bottom-up regulation of ungulates by large carnivores: implications for conservation of biodiversity. Pages 342–361 in J. C. Ray, K. H. Redford, R. S. Steneck, and J. Berger, editors. Large Carnivores and the Conservation of Biodiversity. Island Press, Washinton, D.C., USA. Brodie, J., H. Johnson, M. Mitchell, P. Zager, K. Proffitt, M. Hebblewhite, M. Kauffman, B. Johnson, J. Bissonette, C. Bishop, J. Gude, J. Herbert, K. Hersey, M. Hurley, P. M. Lukacs, S. Mccorquodale, E. Mcintire, J. Nowak, H. Sawyer, D. Smith, and P. J. White. 2013. Relative influence of human harvest, carnivores, and weather on adult female elk survival across western North America. Journal of Applied Ecology 50:295–305. Caughley, G. 1974. Interpretation of age ratios. Journal of Wildlife Management 38:557–562. 8 �Clutton-Brock, T., and B. C. Sheldon. 2010. Individuals and populations: the role of long-term, individual-based studies of animals in ecology and evolutionary biology. Trends in Ecology and Evolution 25:562–573. Cook, J. G., B. K. Johnson, R. C. Cook, R. A. Riggs, T. Delcurto, L. D. Bryant, and L. L. Irwin. 2004a. Effects of summer-autumn nutrition and parturition date on reproduction and survival of elk. Wildlife Monographs 155:1–61. Cook, R. C., J. G. Cook, and L. D. Mech. 2004b. Nutritional condition of northern Yellowstone elk. Journal of Mammalogy 85:714–722. Cook, R. C., J. G. Cook, T. R. Stephenson, W. L. Myers, S. M. Mccorquodale, D. J. Vales, L. L. Irwin, P. B. Hall, R. D. Spencer, S. L. Murphie, K. A. Schoenecker, and P. J. Miller. 2010. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife Management 74:880–896. Cook, R. C., J. G. Cook, D. J. Vales, B. K. Johnson, S. M. McCorquodale, L. A. Shipley, R. A. Riggs, L. L. Irwin, S. L. Murphie, B. L. Murphie, K. A. Schoenecker, F. Geyer, P. B. Hall, R. D. Spencer, D. A. Immell, D. H. Jackson, B. L. Tiller, P. J. Miller, and L. Schmitz. 2013. Regional and seasonal patterns of nutritional condition and reproduction in elk. Wildlife Monographs 184:1–44. Decesare, N. J., M. Hebblewhite, M. Bradley, K. G. Smith, D. Hervieux, and L. Neufeld. 2012. Estimating ungulate recruitment and growth rates using age ratios. Journal of Wildlife Management 76:144–153. Eacker, D. R. 2015. Linking the effects of risk factors on annual calf survival to elk population dynamics in the Bitterroot Valley, Montana. M.S. thesis, University of Montana, Missoula, MT, USA. Eacker, D. R., M. Hebblewhite, K. M. Proffitt, B. S. Jimenez, M. S. Mitchell, and H. S. Robinson. 2016. Annual elk calf survival in a multiple carnivore system. Journal of Wildlife Management 80:1345–1359. Eberhardt, L. L. 1977a. “Optimal” management policies for marine mammals. Wildlife Society Bulletin 5:162–169. Eberhardt, L. L. 1977b. Optimal policies for conservation of large mammals, with special references to marine ecosystems. Environmental Conservation 4:205–212. Eberhardt, L. L. 2002. A paradigm for population analysis of long-lived vertebrates. Ecology 83:2841–2854. Gaillard, J.-M., M. Festa-Bianchet, and N. G. Yoccoz. 1998. Population dynamics of large herbivores: variable recruitment with constant adult survival. Trends in Ecology and Evolution 13:58–63. Gaillard, J.-M., M. Festa-Bianchet, N. G. Yoccoz, A. Loison, and C. Toigo. 2000. Temporal variation in fitness components and population dynamics of large herbivores. Annual Review of Ecology, Evolution, and Systematics 31:367–393. Garrott, R. A., L. L. Eberhardt, P. J. White, and J. Rotella. 2003. Climate-induced variation in vital rates of an unharvested large-herbivore population. Canadian Journal of Zoology 81:33–45. Griffin, K. A., M. Hebblewhite, H. S. Robinson, P. Zager, S. M. Barber-Meyer, D. Christianson, S. Creel, N. C. Harris, M. A. Hurley, D. H. Jackson, B. K. Johnson, W. L. Myers, J. D. Raithel, M. Schlegel, B. L. Smith, C. White, and P. J. White. 2011. Neonatal mortality of elk driven by climate, predator phenology and predator community composition. Journal of Animal Ecology 80:1246–1257. 9 �Harris, N. C., M. J. Kauffman, and L. S. Mills. 2008. Inferences about ungulate population dynamics derived from age ratios. Journal of Wildlife Management 72:1143–1151. Johnson, D. E. 1951. Biology of the elk calf, Cervus canadensis nelsoni. Journal of Wildlife Management 15:396–410. de Kroon, H., A. Plaisier, J. van Groenendael, and H. Caswell. 1986. Elasticity: the relative contribution of demographic parameters to population growth rate. Ecology 67:1427–1431. Linnell, J. D. C., R. Aanes, and R. Andersen. 1995. Who killed Bambi? The role of predation in the neonatal mortality of temperate ungulates. Wildlife Biology 1:209–223. Lukacs, P. M., M. S. Mitchell, M. Hebblewhite, B. K. Johnson, H. Johnson, M. Kauffman, K. M. Proffitt, P. Zager, J. Brodie, K. Hersey, A. A. Holland, M. Hurley, S. McCorquodale, A. Middleton, M. Nordhagen, J. J. Nowak, D. P. Walsh, and P. J. White. 2018. Factors influencing elk recruitment across ecotypes in the western United States. Journal of Wildlife Management 82:698–710. Middleton, A. D., M. J. Kauffman, D. E. Mcwhirter, J. G. Cook, R. C. Cook, A. A. Nelson, M. D. Jimenez, and R. W. Klaver. 2013. Animal migration amid shifting patterns of phenology and predation: lessons from a Yellowstone elk herd. Ecology 94:1245–1256. Monteith, K. L., V. C. Bleich, T. R. Stephenson, B. M. Pierce, M. M. Conner, J. G. Kie, and R. T. Bowyer. 2014. Life-history characteristics of mule deer: effects of nutrition in a variable environment. Wildlife Monographs 186:1–62. Parker, K. L., P. S. Barboza, and M. P. Gillingham. 2009. Nutrition integrates environmental responses of ungulates. Functional Ecology 23:57–69. Proffitt, K. M., J. A. Cunningham, K. L. Hamlin, and R. A. Garrott. 2014. Bottom-up and topdown influences on pregnancy rates and recruitment of northern Yellowstone elk. Journal of Wildlife Management 78:1383–1393. Proffitt, K. M., M. Hebblewhite, W. Peters, N. Hupp, and J. Shamhart. 2016. Linking landscapescale differences in forage to ungulate nutritional ecology. Ecological Applications 26:2156– 2174. Raithel, J. D., M. J. Kauffman, and D. H. Pletscher. 2007. Impact of spatial and temporal variation in calf survival on the growth of elk population. Journal of Wildlife Management 71:795– 803. Singer, F. J., A. Harting, K. K. Symonds, and M. B. Coughenour. 1997. Density dependence, compensation, and environmental effects on elk calf mortality in Yellowstone National Park. Journal of Wildlife Management 61:12–25. Smith, B. L., and S. H. Anderson. 1996. Patterns of neonatal mortality of elk in northwest Wyoming. Canadian Journal of Zoology 74:1229–1237. Stonehouse, K. F., C. R. J. Anderson, M. E. Peterson, and D. R. Collins. 2016. Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus). Technical Publication Number 48, Colorado Parks and Wildlife, Fort Collins, CO, USA. Swift, L. W. 1945. A partial history of the elk herds of Colorado. Journal of Mammalogy 26:114– 119. Tatman, N. M., S. G. Liley, J. W. Cain, and J. W. Pitman. 2018. Effects of calf predation and nutrition on elk vital rates. Journal of Wildlife Management 82:1417–1428. Vavra, M., C. G. Parks, and M. J. Wisdom. 2007. Biodiversity, exotic plant species, and herbivory: the good, the bad, and the ungulate. Forest Ecology and Management 246:66–72. Vavra, M., and D. P. Sheehy. 1996. Improving elk habitat characteristics with livestock grazing. Rangelands 18:182–185. 10 �Visscher, D. R., and E. H. Merrill. 2009. Temporal dynamics of forage succession for elk at two scales: implications of forest management. Forest Ecology and Management 257:96–106. White, C. G., P. Zager, and M. W. Gratson. 2010. Influence of predator harvest, biological factors, and landscape on elk calf survival in Idaho. Journal of Wildlife Management 74:355–369. Prepared by Nathaniel D. Rayl, Wildlife Researcher 11 �Calves:100 adult females (2013-2017) Insufficient data 0 25-300 30-350 35-400 40-450 45-500 50-55 E-4 55-60 E-51 E-99 j E-24 E-33 0 30 60 Miles Figure 1. Number of elk calves per 100 adult females observed during December-February aerial surveys (5-year average from 2013-2017) within elk Data Analysis Units (DAUs; labeled with black text) in Colorado, USA. 12 �1.0 3 33 31 0.9 0.8 +-' C cu 0.7 C 0) ~ 0.6 a. C o 0.5 t: 0 g-0.4 ~ o.. 0.3 0.2 0.1 0.0 Bear's Ears Trinchera Uncompahgre Plateau Herd Figure 2. Estimated average pregnancy rates of adult female elk from the Bear’s Ears, Trinchera, and Uncompahgre Plateau herds sampled during late winter 2019 in Colorado, USA. The sample size is given at the top of the 95% binomial confidence intervals (black lines). 13 �12.5- ~ 10.0- cf:. --- +-' ~ >, "'O 0 .o 7.5a, Q) I... I ro +-' CJ) Q) C) C 5.Q - 2.5Bear's Ears Trinchera Uncompahgre Plateau Herd Figure 3. The estimated ingesta-free body fat (%) of adult female elk from the Bear’s Ears (n = 3), Trinchera (n = 33), and Uncompahgre Plateau (n = 31) herds during late-winter 2019 in Colorado, USA. 14 �Jun 17 - Jun 10 - .....roQ) "'O cn Jun 03 c "> ro () May 27 - May 20 Bear's Ears Trinchera Herd Uncompahgre Plateau Figure 4. The distribution of calving dates of adult female elk estimated from vaginal implant transmitters (VITs) from the Bear’s Ears (n = 2), Trinchera (n = 30), and Uncompahgre Plateau (n = 30) herds during 2019 in Colorado, USA. 15 �Colorado Parks and Wildlife July 2019 – June 2020 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Colorado 3430 3002 1 : : : : Federal Aid Project No. W-242-R4 : Parks and Wildlife Mammals Research Elk Conservation Evaluating factors influencing elk recruitment in Colorado Period Covered: July 1, 2019 – June 30, 2020 Authors: N.D. Rayl, M.W. Alldredge, and C.R. Anderson Jr. Personnel: J. Anderson, J. Ashe, B. Banulis, E. Babbitt, T. Bonacquista, K. Bond, G. Bullington, M. Caddy, K. Crane, D. Collins, R. Delpiccolo, K. Deschenes, J. Dewhirst, K. Duckett, W. de Vergie, R. Ebel-Childs, J. Ennis, D. Finley, P. Firmin, M. Fisher, K. Fox, L. Gephert, K. Hayes, W. Hiler, B. Holder, E. Jones, T. Kishimoto, D. Lewis, K. Logan, K. Mahaffie, M. Melton, K. Middledorf, M. Miller, C. Murray, A. Orlando, M. Ortega, J. Pollock, B. Reimann, N. Renneker, M. Richman, T. Robinson, K. Russo, E. Sawa, I. Smith, T. Smith, J. Stanton, M. Swaro, L. Sweanor, J. Taylor, L. Temple, M. Trujillo, G. Tuck, A. Vitt, N. Waring, J. Yost, S. Waters, and L. Wolfe, CPW; J. Clark, H. Cushman, J. Larrivee, A. Orlando, S. Strike, R. Swisher, S. Swisher, T. Triple, Quicksilver Air, Inc.. Project support received from Federal Aid in Wildlife Restoration, Rocky Mountain Elk Foundation, CPW Big Game Auction and Raffle, and Virginia Wellington Cabot Foundation. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. ABSTRACT Over the last two decades, wildlife managers in Colorado have become increasingly concerned about declining winter elk calf recruitment (estimated using juvenile:adult female ratios) in the southern portion of the state. Although juvenile:adult female ratios are often highly correlated with juvenile elk survival, they are an imperfect estimate of recruitment because they are affected by harvest, pregnancy rates, juvenile survival, and adult female survival. Thus, there is a need for elk research in Colorado based upon monitoring of marked individuals to evaluate factors affecting each stage of production and survival. In 2016, Colorado Parks and Wildlife (CPW) began a 2-year pilot study to investigate factors influencing elk recruitment in 2 study areas in the state. In FY2018-19, CPW expanded this pilot study work into a 3rd study area and for 6 additional years to better determine how predators, habitat, and weather conditions are impacting elk recruitment in Colorado. During the past fiscal year we focused on working with stakeholders and collaborators 1 �on research logistics, and capturing and collaring elk. Field efforts were centered on 2 objectives: 1) capturing adult female elk, and collaring and outfitting pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior, and 2) capturing and collaring newborn and 6-month old elk to collect data on calf survival and cause-specific mortality. We radio-collared 98 pregnant elk and outfitted them with VITs. Estimates of average pregnancy rates ranged from 78-93% across herds, and estimates of mean ingesta-free body fat ranged from 6.5-7.5% across herds. We radio-collared 50 6-month old elk calves in December 2019. During the 2020 calving season, we radio-collared127 elk calves, including 91% of the calves born to collared females (85 of 93 calves). Mean calving date across herds was 31 May. 2 �WILDLIFE RESEARCH REPORT EVALUATING FACTORS INFLUENCING ELK RECRUITMENT IN COLORADO NATHANIEL D. RAYL, MAT W. ALLDREDGE, AND CHUCK R. ANDERSON JR. PROJECT NARRATIVE OBJECTIVES The objectives of this project are to 1) estimate elk calf survival and cause-specific mortality rates from birth to age 1 to evaluate the importance of mortality sources for elk calf survival, 2) evaluate the influence of biotic (birth date, birth mass, gender, maternal body condition, habitat conditions) and abiotic factors (previous and current weather conditions) on seasonal mortality risk of elk calves from birth to age 1, 3) assess the health of elk herds by quantifying pregnancy rates and percent ingesta-free body fat (IFBF) of adult female elk, and 4) evaluate the influence of biotic (age, body condition, reproductive status, habitat conditions and selection) and abiotic factors (previous and current weather conditions) on pregnancy rates and IFBF. SEGMENT OBJECTIVES 1. Work with personnel from CPW Areas 6, 10, 11, and 18, and private landowners on field research logistics. 2. Capture adult female elk, and collar and outfit pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior. 3. Capture and collar newborn and 6-month old elk to collect data on calf survival and cause-specific sources of mortality. INTRODUCTION In Colorado, elk (Cervus canadensis) are an important natural resource that are valued for ecological, consumptive, aesthetic, and economic reasons. In 1910, fewer than 1,000 elk remained in Colorado (Swift 1945), but today the state population is estimated to be the largest in the country, with approximately 282,000 elk. Over the last two decades, however, there has been increasing concern among wildlife managers in Colorado about declining winter calf recruitment (estimated using juvenile:adult female ratios) in the southern portion of the state (Fig. 1; Colorado Parks and Wildlife, unpublished data). In many elk populations, similar declining trends in juvenile recruitment have been observed. A recent synthesis examining elk recruitment in the western United States from 19892010 found evidence of a long-term reduction of 0.48 juveniles/100 adult females/year (Lukacs et al. 2018). Lukacs et al. (2018) discovered associations between recruitment and forage productivity that suggested nutritional conditions on either summer or winter ranges had the most influence on elk recruitment. These associations varied by geographic region: winter range conditions appeared to be more influential in southern areas (Colorado, Utah, and parts of Wyoming), whereas summer range conditions appeared to be more influential in northern areas 3 �(Idaho, Montana, Oregon, Washington, parts of Wyoming). Lukacs et al. (2018) also found that lower total winter precipitation in the previous winter was associated with lower recruitment the next year. The presence of wolves (Canis lupus) and grizzly bears (Ursus arctos) were also associated with lower recruitment. Although juvenile:adult female ratios are often highly correlated with juvenile elk survival (Raithel et al. 2007, Harris et al. 2008), they are an imperfect estimate of recruitment because they are affected by harvest, pregnancy rates, juvenile survival, and adult female survival (Caughley 1974, Gaillard et al. 2000, Harris et al. 2008, Decesare et al. 2012, Lukacs et al. 2018). This makes it difficult to identify ultimate factors influencing population dynamics using age ratio data alone, as multiple scenarios can produce equivalent ratios (Caughley 1974). Thus, long-term demographic studies based on monitoring of marked individuals are necessary to reliably test biological hypotheses and evaluate factors affecting each stage of production and survival (Gaillard et al. 2000, Clutton-Brock and Sheldon 2010, Proffitt et al. 2014). In the absence of harvest, the population dynamics of ungulates are normally characterized by high and stable adult female survival and variable juvenile survival (Gaillard et al. 1998, 2000). Among vital rates, changes in adult female survival have the potential to exert the most influence on population growth rates, but usually do not because of low year-to-year variability (Gaillard et al. 1998, 2000). Instead, adult female survival is typically buffered against moderate environmental variation, and thus not strongly influenced by climatic or densitydependent factors (Gaillard et al. 2000). Indeed, in a large-scale meta-analysis, Brodie et al. (2013) demonstrated that adult female elk survival was principally related to human harvest, and found that this harvest was an additive source of mortality. In contrast to adult female survival, and despite its lower elasticity (de Kroon et al. 1986), juvenile survival frequently has a greater effect on population growth rates of ungulates because of its high variability (Gaillard et al. 1998, 2000, Raithel et al. 2007). Juvenile survival of ungulates may be influenced by abiotic or biotic factors such as environmental conditions, forage quality or quantity, population density, maternal body condition, and predation (Barber-Meyer et al. 2008, Griffin et al. 2011, Monteith et al. 2014, Bastille-Rousseau et al. 2016). Complex interactions among these factors frequently make it difficult to identify the relative role of top-down and bottom-up factors affecting calf survival (Linnell et al. 1995). Further complexity may be introduced if top-down and bottom-up forces are simultaneously and variably influencing survival (Bowyer et al. 2005, Monteith et al. 2014). Juvenile survival has been found to vary substantially among and within elk populations on an annual basis (Garrott et al. 2003, Raithel et al. 2007, Griffin et al. 2011). Griffin et al. (2011) synthesized elk neonatal survival across 12 populations in the northwestern United States, and found that survival declined after warmer previous summers and with more predator species, and increased with higher May precipitation. In resource-limited populations, weather conditions may heavily influence juvenile survival. For example, in an unharvested elk population, prior to the establishment of wolves, Garrott et al. (2003) found that the dominant source of calf mortality in Yellowstone National Park was starvation, with more severe winters leading to lower calf survival. Elsewhere, mortality due to predation has been identified as the most significant cause of death for elk calves (e.g., Barber-Meyer et al. 2008). In some systems black bears (Ursus americanus) are the dominant predator (White et al. 2010, Tatman et al. 2018), whereas in others mountain lions (Puma concolor) kill the most calves (Eacker et al. 2016). Although non-predation deaths (i.e., disease, starvation) may be low where high levels of juvenile predation occur, it remains difficult to determine whether neonate predation represents 4 �an additive or compensatory source of mortality (Linnell et al. 1995, Barber-Meyer et al. 2008, Griffin et al. 2011). Individual calf characteristics may also influence risk of mortality. To date, evidence for sex-biased mortality in elk calves is equivocal. Some studies have documented increased vulnerability of male calves to predation (Smith and Anderson 1996, Eacker et al. 2016), whereas others have demonstrated increased vulnerability of female calves (White et al. 2010). It has been suggested that differences may be due to the hunting behavior of the dominant predator, with stalking predators, such as mountain lions, disproportionately killing male calves, which may engage in riskier exploratory behavior (Eacker et al. 2016). Conclusions about the effect of birth mass on elk calf survival are similarly ambiguous. Some studies have reported that birth mass influenced survival probability (Singer et al. 1997, White et al. 2010), but others have not detected an influence of birth mass on neonate mortality (Smith and Anderson 1996, Eacker et al. 2016). The ability for nutritional resources on the landscape to support ungulate populations is reflected in the nutritional condition of individuals within those populations (Parker et al. 2009, Cook et al. 2013, Monteith et al. 2014). As nutritional resources decline there is typically a predictable sequence of changes in the vital rates of large herbivore populations: first, juvenile survival decreases, then age of first reproduction increases, followed by decreased fertility of adult females, and finally increased mortality rates of adults (Eberhardt 1977a, 1977b, 2002, Gaillard et al. 1998, 2000). This general sequence has been confirmed for elk, with forage quality and associated nutritional body condition demonstrated to affect intra-uterine survival, yearling and adult pregnancy rates, birth dates, birth weight, calf growth, and winter survival of juveniles and adults (Cook et al. 2004a, 2004b, 2013, Proffitt et al. 2016). When nutritional resources are limited, elk calves may be lighter at birth, be born later, and have slower growth rates during summer, which may predispose them to predation mortality. Cook et al. (2004a, 2013) demonstrated that summer-autumn nutrition can play a central role in determining elk productivity, as, during this time, animals must meet the demands of lactation and accrue sufficient fat to get pregnant and survive the winter. For elk, nutritional resources on the landscape may be affected by wild and domestic herbivory (Vavra and Sheehy 1996, Vavra et al. 2007), timber management (Visscher and Merrill 2009), climatic conditions (Middleton et al. 2013), and fire history (Proffitt et al. 2016). To properly determine factors affecting juvenile ungulate survival bottom-up nutritional effects and top-down predation effects should be evaluated together (Monteith et al. 2014). Understanding how these effects are influencing elk population dynamics in Colorado is critical for guiding management actions. In 2016, Colorado Parks and Wildlife (CPW) initiated a 2-year pilot study to investigate factors influencing elk recruitment in 2 study areas in the state. During the pilot study, researchers collected data on annual elk calf survival, pregnancy rates, and latewinter body condition of adult female elk. In July 2018 – July 2019, we expanded this research into a 3rd study area to better determine how predators, habitat, and weather conditions are impacting elk recruitment in Colorado, and to provide management recommendations for increasing juvenile recruitment. STUDY AREAS For management purposes, the elk population in Colorado is divided into 43 Data Analysis Units (DAUs), each of which encompass the year-round range of an elk herd. This 5 �project is being conducted in 3 DAUs, 2 with low juvenile:adult female ratios (E-20, E-33), and 1 with high juvenile:adult female ratios (E-2), which will serve as a reference area (Fig. 1). The Uncompahgre Plateau elk herd (DAU E-20; 5,858 km2) is on the Uncompahgre Plateau in southwest Colorado, USA. In 2017, the post-hunt population of the Uncompahgre Plateau herd was estimated to be ~8,500 elk. From 2013-2017 juvenile:adult female ratios in E20 averaged 30 calves per 100 adult females. Landownership is a mixture of BLM (38%), USFS (37%), private (24%), and state (1%) lands. Elevations range from 1,390 to 3,150 m. The plateau is characterized by a mixture of pinyon-juniper (Pinus edulis, Juniperus osteosperma) woodlands and sage-grassland communities (Artemisia spp., Cercocarpus montanus, Achnatherum hymenoides) at lower elevations. At mid elevations, ponderosa pine (Pinus ponderosa) and mountain shrub communities (Amelanchier alnifolia, Arctostaphylos, Artemisia spp., Quercus gambelii, Symphoricarpos spp.) predominate. Spruce-fir (Picea engelmannii, Abies lasiocarpa, Pseudotsuga menziesii) and aspen (Populus tremuloides) forests dominate at higher elevations. The Trinchera elk herd (DAU E-33; 8,601 km2) is in southeast Colorado, USA. In 2017, the post-hunt population of the Trinchera herd was estimated to be ~16,600 elk. From 2013-2017 juvenile:adult female ratios in E-33 averaged 26 calves per 100 adult females. Landownership is a mixture of private (89%), USFS (3%), state (3%), BLM (2%), U.S. Fish and Wildlife Service (USFWS; 1%), and other (2%) lands. Elevations range from 1,640 to 4,370 m. Lower elevations are characterized by agriculture and sage-grassland communities. At mid elevations, pinyonjuniper woodlands, ponderosa pine and mountain shrub forests, and spruce-fir and aspen forests predominate. Alpine tundra communities dominate at higher elevations. The Bear’s Ears elk herd (DAU E-2; 7,293 km2) is in northwest Colorado, USA. In 2017, the post-hunt population of the Bear’s Ears herd was estimated to be 20,000-24,000 elk. From 2013-2017 juvenile:adult female ratios in E-2 averaged 56 calves per 100 adult females. Landownership is a mixture of private (50%), USFS (25%), BLM (19%), state (5%) and other (1%) lands. Elevations range from 1,730 to 3,710 m. The DAU is characterized by sagegrassland communities at lower elevations. At mid elevations, mountain shrub communities predominate. Spruce-fir and aspen forests dominate at higher elevations. METHODS Capture and handling — We captured adult female elk ≥2 years of age from each study herd by helicopter net-gunning during late winter (March). During capture, we marked individuals with ear tags, collected a blood sample, and measured hind foot length, chest girth, and age based on tooth eruption and wear patterns. We used a portable ultrasound machine to assess whether or not captured elk were pregnant, and estimated the percent of ingesta-free body fat (IFBF) following methods detailed in Cook et al. (2010). We verified non-pregnancies using pregnancy-specific protein B (PSPB) analysis of sampled blood. From each study herd, we outfitted pregnant elk with vaginal implant transmitters (VITs) and Global Positioning System (GPS) radio-collars that attempt to acquire a location every 2 h. We deployed VITs that use the satellite communication capabilities of the collar on the adult female to send a notification when the VIT is expelled, signifying a birth. After receiving a birth notification from a VIT, we went to the birth site to capture and collar the newborn elk calf. We blindfolded calves, and wore latex gloves to minimize the transfer of human scent. We measured body mass, hind foot length, chest girth, and determined the gender of captured calves. We outfitted elk neonates with expandable GPS radio-collars or 6 �VHF proximity collars that communicated with the collar on the adult female, and are designed to drop off after 12 months. We also opportunistically located, captured, and collared additional neonates to increase sample sizes. We collected additional measurements (hair moisture, incisor and upper canine eruption, hoof, dew claw, and navel condition) from opportunistically captured calves to estimate age at capture following Johnson (1951) and Eacker (2015). In December, we captured 6-month old elk calves from each study herd by helicopter netgunning. During capture, we measured body mass, hind foot length, chest girth, and determined the gender of captured calves. We outfitted calves with expandable GPS radio-collars that were scheduled to drop off after 6 months. During all captures, we followed CPW’s animal care and use guidelines for capturing and handling elk (CPW ACUC #09-2008). Cause-specific mortality — Within 24 hours of detecting a mortality signal from an elk collar, we attempted to conduct a systematic field investigation to determine the cause of death. We searched the area surrounding kill sites for evidence of predator presence, including predator scats, tracks, and hair, or signs of a struggle (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). We examined elk calf carcasses for evidence of canine puncture wounds, subcutaneous hemorrhaging and bruising, aspirated blood in the mouth, nose, or trachea, claw or bite marks on the hide, cracked or chewed bones, and characteristic consumption patterns (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). We also collected calf carcasses when they were available to verify field assessments with laboratory necropsies performed by a CPW veterinarian. Nutritional condition of adult female elk — The body fat of lactating and non-lactating adult female elk can vary substantially, as lactating females are more sensitive than non-lactating females to their nutritional environment (Cook et al. 2004a, 2013). Therefore, it is difficult to interpret the body condition of adult female elk in late winter without knowing whether or not they experienced the energetic demands of lactation throughout the previous growing season (Cook et al. 2004a, 2013). We will use the late-winter body condition of prime-aged adult female elk that successfully raised a calf the previous year to assess whether or not our study herds may be experiencing nutritional limitations. RESULTS AND DISCUSSION During March 2020, we captured 113 adult female elk by helicopter net-gunning, 43 from the Bear’s Ears herd, 27 from the Trinchera herd, and 43 from the Uncompahgre Plateau herd. We radio-collared 98 pregnant elk and outfitted them with VITs, 40 each from the Bear’s Ears and Uncompahgre Plateau herds, and 18 from the Trinchera herd. Additionally, we collared 1 non-pregnant elk from the Trinchera herd. In 2020, we estimated that pregnancy rates of adult female elk were 93% in the Bear’s Ears and Uncompahgre Plateau herds (both 95% CI = 81-98%; n = 43), and 78% in the Trinchera herd (95% CI = 59-89%; n = 27; Fig. 2). Elk populations experiencing good to excellent summer-autumn nutrition typically have pregnancy rates ≥90% (Cook et al. 2013). We estimated the mean IFBF of adult female elk to be 6.51% from the Bear’s Ears herd, 7.51% from the Trinchera herd, and 7.03% from the Uncompahgre Plateau herd (Fig. 3). When late-winter IFBF values are <8-9% for adult female elk that have lactated through the previous growing season, this suggests that there may be nutritional limitations, but it does not identify 7 �whether limitations are a result of summer-autumn or winter nutrition (R. Cook, personal communication). In December 2019, we collared 50 6-month old elk calves, 25 each from the Bear’s Ears and Uncompahgre Plateau elk herds. The mean weight of calves from the Bear’s Ears herd was 101.8 kg (95% CI = 96.5-107.2 kg) from the Bear’s Ears herd and 113.9 kg (95% CI = 108.4119.4 kg) from the Uncompahgre Plateau elk herd. During May-July 2020, we captured and collared 127 elk calves, 54 from the Bear’s Ears herd, 21 from the Trinchera herd, and 52 from the Uncompahgre Plateau herd. From the Bear’s Ears and Uncompahgre Plateau herds, we successfully captured and collared 90% (35/39) of the calves of adult female elk outfitted with VITs. From the Trinchera herd, we successfully captured and collared 100% (15/15) of the calves of adult female elk outfitted with VITs. The estimated mean date of calving was 31 May in the Bear’s Ears and Uncompahgre Plateau herds, and 3 June in the Trinchera herd (Fig. 4). SUMMARY During FY2019-20 we successfully worked with private landowners and personnel from CPW to coordinate field research logistics and initiate the second year of this study. We collected data on body condition and reproduction by capturing adult female elk, and we outfitted 99 pregnant females with GPS collars and VITs. We successfully captured and collared 127 newborn elk and 50 6-month old elk calves, meeting our sample size objectives, and allowing us to collect data on calf survival and cause-specific sources of mortality. We will continue to collect data on elk survival and cause-specific sources of mortality throughout the year. LITERATURE CITED Barber-Meyer, S. M., L. D. Mech, and P. J. White. 2008. Elk calf survival and mortality following wolf restoration to Yellowstone National Park. Wildlife Monographs 169:1–30. Bastille-Rousseau, G., J. A. Schaefer, K. P. Lewis, M. A. Mumma, E. H. Ellington, N. D. Rayl, S. P. Mahoney, D. Pouliot, and D. L. Murray. 2016. Phase-dependent climate-predator interactions explain three decades of variation in neonatal caribou survival. Journal of Animal Ecology 85:445–456. Bowyer, T., D. K. Person, and M. P. Becky. 2005. Detecting top-down versus bottom-up regulation of ungulates by large carnivores: implications for conservation of biodiversity. Pages 342–361 in J. C. Ray, K. H. Redford, R. S. Steneck, and J. Berger, editors. Large Carnivores and the Conservation of Biodiversity. Island Press, Washinton, D.C., USA. Brodie, J., H. Johnson, M. Mitchell, P. Zager, K. Proffitt, M. Hebblewhite, M. Kauffman, B. Johnson, J. Bissonette, C. Bishop, J. Gude, J. Herbert, K. Hersey, M. Hurley, P. M. Lukacs, S. Mccorquodale, E. Mcintire, J. Nowak, H. Sawyer, D. Smith, and P. J. White. 2013. Relative influence of human harvest, carnivores, and weather on adult female elk survival across western North America. Journal of Applied Ecology 50:295–305. Caughley, G. 1974. Interpretation of age ratios. Journal of Wildlife Management 38:557–562. Clutton-Brock, T., and B. C. Sheldon. 2010. Individuals and populations: the role of long-term, individual-based studies of animals in ecology and evolutionary biology. Trends in Ecology and Evolution 25:562–573. 8 �Cook, J. G., B. K. Johnson, R. C. Cook, R. A. Riggs, T. Delcurto, L. D. Bryant, and L. L. Irwin. 2004a. Effects of summer-autumn nutrition and parturition date on reproduction and survival of elk. Wildlife Monographs 155:1–61. Cook, R. C., J. G. Cook, and L. D. Mech. 2004b. Nutritional condition of northern Yellowstone elk. Journal of Mammalogy 85:714–722. Cook, R. C., J. G. Cook, T. R. Stephenson, W. L. Myers, S. M. Mccorquodale, D. J. Vales, L. L. Irwin, P. B. Hall, R. D. Spencer, S. L. Murphie, K. A. Schoenecker, and P. J. Miller. 2010. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife Management 74:880–896. Cook, R. C., J. G. Cook, D. J. Vales, B. K. Johnson, S. M. McCorquodale, L. A. Shipley, R. A. Riggs, L. L. Irwin, S. L. Murphie, B. L. Murphie, K. A. Schoenecker, F. Geyer, P. B. Hall, R. D. Spencer, D. A. Immell, D. H. Jackson, B. L. Tiller, P. J. Miller, and L. Schmitz. 2013. Regional and seasonal patterns of nutritional condition and reproduction in elk. Wildlife Monographs 184:1–44. Decesare, N. J., M. Hebblewhite, M. Bradley, K. G. Smith, D. Hervieux, and L. Neufeld. 2012. Estimating ungulate recruitment and growth rates using age ratios. Journal of Wildlife Management 76:144–153. Eacker, D. R. 2015. Linking the effects of risk factors on annual calf survival to elk population dynamics in the Bitterroot Valley, Montana. M.S. thesis, University of Montana, Missoula, MT, USA. Eacker, D. R., M. Hebblewhite, K. M. Proffitt, B. S. Jimenez, M. S. Mitchell, and H. S. Robinson. 2016. Annual elk calf survival in a multiple carnivore system. Journal of Wildlife Management 80:1345–1359. Eberhardt, L. L. 1977a. “Optimal” management policies for marine mammals. Wildlife Society Bulletin 5:162–169. Eberhardt, L. L. 1977b. Optimal policies for conservation of large mammals, with special references to marine ecosystems. Environmental Conservation 4:205–212. Eberhardt, L. L. 2002. A paradigm for population analysis of long-lived vertebrates. Ecology 83:2841–2854. Gaillard, J.-M., M. Festa-Bianchet, and N. G. Yoccoz. 1998. Population dynamics of large herbivores: variable recruitment with constant adult survival. Trends in Ecology and Evolution 13:58–63. Gaillard, J.-M., M. Festa-Bianchet, N. G. Yoccoz, A. Loison, and C. Toigo. 2000. Temporal variation in fitness components and population dynamics of large herbivores. Annual Review of Ecology, Evolution, and Systematics 31:367–393. Garrott, R. A., L. L. Eberhardt, P. J. White, and J. Rotella. 2003. Climate-induced variation in vital rates of an unharvested large-herbivore population. Canadian Journal of Zoology 81:33–45. Griffin, K. A., M. Hebblewhite, H. S. Robinson, P. Zager, S. M. Barber-Meyer, D. Christianson, S. Creel, N. C. Harris, M. A. Hurley, D. H. Jackson, B. K. Johnson, W. L. Myers, J. D. Raithel, M. Schlegel, B. L. Smith, C. White, and P. J. White. 2011. Neonatal mortality of elk driven by climate, predator phenology and predator community composition. Journal of Animal Ecology 80:1246–1257. Harris, N. C., M. J. Kauffman, and L. S. Mills. 2008. Inferences about ungulate population dynamics derived from age ratios. Journal of Wildlife Management 72:1143–1151. Johnson, D. E. 1951. Biology of the elk calf, Cervus canadensis nelsoni. Journal of Wildlife Management 15:396–410. 9 �de Kroon, H., A. Plaisier, J. van Groenendael, and H. Caswell. 1986. Elasticity: the relative contribution of demographic parameters to population growth rate. Ecology 67:1427–1431. Linnell, J. D. C., R. Aanes, and R. Andersen. 1995. Who killed Bambi? The role of predation in the neonatal mortality of temperate ungulates. Wildlife Biology 1:209–223. Lukacs, P. M., M. S. Mitchell, M. Hebblewhite, B. K. Johnson, H. Johnson, M. Kauffman, K. M. Proffitt, P. Zager, J. Brodie, K. Hersey, A. A. Holland, M. Hurley, S. McCorquodale, A. Middleton, M. Nordhagen, J. J. Nowak, D. P. Walsh, and P. J. White. 2018. Factors influencing elk recruitment across ecotypes in the western United States. Journal of Wildlife Management 82:698–710. Middleton, A. D., M. J. Kauffman, D. E. Mcwhirter, J. G. Cook, R. C. Cook, A. A. Nelson, M. D. Jimenez, and R. W. Klaver. 2013. Animal migration amid shifting patterns of phenology and predation: lessons from a Yellowstone elk herd. Ecology 94:1245–1256. Monteith, K. L., V. C. Bleich, T. R. Stephenson, B. M. Pierce, M. M. Conner, J. G. Kie, and R. T. Bowyer. 2014. Life-history characteristics of mule deer: effects of nutrition in a variable environment. Wildlife Monographs 186:1–62. Parker, K. L., P. S. Barboza, and M. P. Gillingham. 2009. Nutrition integrates environmental responses of ungulates. Functional Ecology 23:57–69. Proffitt, K. M., J. A. Cunningham, K. L. Hamlin, and R. A. Garrott. 2014. Bottom-up and topdown influences on pregnancy rates and recruitment of northern Yellowstone elk. Journal of Wildlife Management 78:1383–1393. Proffitt, K. M., M. Hebblewhite, W. Peters, N. Hupp, and J. Shamhart. 2016. Linking landscapescale differences in forage to ungulate nutritional ecology. Ecological Applications 26:2156– 2174. Raithel, J. D., M. J. Kauffman, and D. H. Pletscher. 2007. Impact of spatial and temporal variation in calf survival on the growth of elk population. Journal of Wildlife Management 71:795– 803. Singer, F. J., A. Harting, K. K. Symonds, and M. B. Coughenour. 1997. Density dependence, compensation, and environmental effects on elk calf mortality in Yellowstone National Park. Journal of Wildlife Management 61:12–25. Smith, B. L., and S. H. Anderson. 1996. Patterns of neonatal mortality of elk in northwest Wyoming. Canadian Journal of Zoology 74:1229–1237. Stonehouse, K. F., C. R. J. Anderson, M. E. Peterson, and D. R. Collins. 2016. Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus). Technical Publication Number 48, Colorado Parks and Wildlife, Fort Collins, CO, USA. Swift, L. W. 1945. A partial history of the elk herds of Colorado. Journal of Mammalogy 26:114– 119. Tatman, N. M., S. G. Liley, J. W. Cain, and J. W. Pitman. 2018. Effects of calf predation and nutrition on elk vital rates. Journal of Wildlife Management 82:1417–1428. Vavra, M., C. G. Parks, and M. J. Wisdom. 2007. Biodiversity, exotic plant species, and herbivory: the good, the bad, and the ungulate. Forest Ecology and Management 246:66–72. Vavra, M., and D. P. Sheehy. 1996. Improving elk habitat characteristics with livestock grazing. Rangelands 18:182–185. Visscher, D. R., and E. H. Merrill. 2009. Temporal dynamics of forage succession for elk at two scales: implications of forest management. Forest Ecology and Management 257:96–106. White, C. G., P. Zager, and M. W. Gratson. 2010. Influence of predator harvest, biological factors, and landscape on elk calf survival in Idaho. Journal of Wildlife Management 74:355–369. 10 �Prepared by Nathaniel D. Rayl, Wildlife Researcher 11 �Calves:100 adult females (2013-2017) Insufficient data 0 25-300 30-350 35-400 40-450 45-500 50-55 E-4 55-60 E-51 E-99 j E-24 E-33 0 30 60 Miles Figure 1. Number of elk calves per 100 adult females observed during December-February aerial surveys (5-year average from 2013-2017) within elk Data Analysis Units (DAUs; labeled with black text) in Colorado, USA. 12 �■ Bears Ear's ■ Trinchera 1.0 0.9 31 27 II Uncompahgre Plateau 43 31 43 31 0.8 ..... C ~ 0.7 O> c.. 0.6 (]) C o 0.5 t:'. 8_0.4 0 l- o... 0.3 0.2 0.1 o.o~- 2017 2018 2019 2020 Year Figure 2. Estimated average pregnancy rates of adult female elk from the Bear’s Ears, Trinchera, and Uncompahgre Plateau herds sampled during late winter 2017-2020 in Colorado, USA. The sample size is given at the top of the 95% binomial confidence intervals (black lines). 13 �11 - • 1o- •• --- 9- ~ 0 ._.... ..... ~ >, ""O 8- 0 .0 (I) (I) 7- L.. '+I .....roen 6(I) 0) C 5- 4- • 3' Bear's Ears Trinchera Herd ' Uncompahgre Plateau Figure 3. The estimated ingesta-free body fat (%) of adult female elk from the Bear’s Ears (n = 43), Trinchera (n = 25), and Uncompahgre Plateau (n = 42) herds during late-winter 2020 in Colorado, USA. 14 �Jul 01 - • •• Jun 15(]) ...... ro ""O 0) C > ro Jun 01 U I May 15- • ' Bear's Ears ' Trinchera Uncompahgre Plateau Herd Figure 4. The distribution of calving dates of adult female elk estimated from vaginal implant transmitters (VITs) from the Bear’s Ears (n = 39), Trinchera (n = 15), and Uncompahgre Plateau (n = 39) herds during 2020 in Colorado, USA. 15 �Colorado Parks and Wildlife July 2020 – June 2021 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Colorado 3430 3002 1 : : : : Federal Aid Project No. W-242-R5 : Parks and Wildlife Mammals Research Elk Conservation Evaluating factors influencing elk recruitment in Colorado Period Covered: July 1, 2020 – June 30, 2021 Authors: N.D. Rayl, M.W. Alldredge, and C.R. Anderson Jr. Personnel: J. Anderson, B. Banulis, T. Bonacquista, K. Bond, M. Caddy, A. Cole, K. Crane, S. Crews, R. Delpiccolo, K. Deschenes, J. Dewhirst, K. Duckett, W. de Vergie, D. Finley, P. Firmin, M. Fisher, K. Fox, L. Gephert, K. Hayes, C. Hernandez, W. Hiler, B. Holder, J. Irvin, E. Jones, A. Kircher, T. Kishimoto, D. Lewis, K. Logan, K. Mahaffie, K. Middledorf, M. Miller, H. Mondin, E. Monfort, C. Murray, R. Nielson, P. Nol, A. Orlando, M. Ortega, J. Pollock, N. Renneker, J. Richards, M. Richman, T. Robinson, K. Russo, E. Sawa, I. Smith, T. Smith, M. Swaro, L. Sweanor, J. Taylor, L. Temple, M. Trujillo, G. Tuck, E. VanNatta, A. Vitt, N. Waring, S. Waters, and M. Wood, CPW; J. Clark, H. Cushman, B. Dooling, T. Herby, A. Orlando, R. Swisher, S. Swisher, and T. Triple, Quicksilver Air, Inc.. Project support received from Federal Aid in Wildlife Restoration, Rocky Mountain Elk Foundation, and CPW Big Game Auction and Raffle. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. ABSTRACT Over the last two decades, wildlife managers in Colorado have become increasingly concerned about declining winter elk calf recruitment (estimated using juvenile:adult female ratios) in the southern portion of the state. Although juvenile:adult female ratios are often highly correlated with juvenile elk survival, they are an imperfect estimate of recruitment because they are affected by harvest, pregnancy rates, juvenile survival, and adult female survival. Thus, there is a need for elk research in Colorado based upon monitoring of marked individuals to evaluate factors affecting each stage of production and survival. Colorado Parks and Wildlife (CPW) is conducting research in 3 study areas to better determine how predators, habitat, and weather conditions are impacting elk recruitment in Colorado. From July 1, 2020 – June 30, 2021, we focused on working with stakeholders and collaborators on research logistics, and capturing and collaring elk. Field efforts were centered on 2 objectives: 1) capturing adult female elk, and collaring and 1 �outfitting pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior, and 2) capturing and collaring newborn and 6-month old elk to collect data on calf survival and cause-specific mortality. We radio-collared 50 6-month old elk calves in December 2020. In March 2021, we radio-collared 100 pregnant elk and outfitted them with VITs. Estimates of average pregnancy rates ranged from 91-95% across herds, and estimates of mean ingesta-free body fat ranged from 7.01-7.78 across herds. During the 2021 calving season, we radio-collared 126 elk calves. Mean calving date across herds was 2 June. 2 �WILDLIFE RESEARCH REPORT EVALUATING FACTORS INFLUENCING ELK RECRUITMENT IN COLORADO NATHANIEL D. RAYL, MAT W. ALLDREDGE, AND CHUCK R. ANDERSON JR. PROJECT NARRATIVE OBJECTIVES The objectives of this project are to 1) estimate elk calf survival and cause-specific mortality rates from birth to age 1 to evaluate the importance of mortality sources for elk calf survival, 2) evaluate the influence of biotic (birth date, birth mass, gender, maternal body condition, habitat conditions) and abiotic factors (previous and current weather conditions) on seasonal mortality risk of elk calves from birth to age 1, 3) assess the health of elk herds by quantifying pregnancy rates and percent ingesta-free body fat (IFBF) of adult female elk, and 4) evaluate the influence of biotic (age, body condition, reproductive status, habitat conditions and selection) and abiotic factors (previous and current weather conditions) on pregnancy rates and IFBF. SEGMENT OBJECTIVES 1. Work with personnel from CPW Areas 6, 10, 11, and 18, and private landowners on field research logistics. 2. Capture adult female elk, and collar and outfit pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior. 3. Capture and collar neonate and 6-month old elk to collect data on calf survival and causespecific sources of mortality. INTRODUCTION In Colorado, elk (Cervus canadensis) are an important natural resource that are valued for ecological, consumptive, aesthetic, and economic reasons. In 1910, fewer than 1,000 elk remained in Colorado (Swift 1945), but today the state population is estimated to be the largest in the country, with over 292,000 elk. Over the last two decades, however, there has been increasing concern among wildlife managers in Colorado about declining winter calf recruitment (estimated using juvenile:adult female ratios) in the southern portion of the state (Fig. 1; Colorado Parks and Wildlife, unpublished data). In many elk populations, similar declining trends in juvenile recruitment have been observed. A recent synthesis examining elk recruitment in the western United States from 19892010 found evidence of a long-term reduction of 0.48 juveniles/100 adult females/year (Lukacs et al. 2018). Lukacs et al. (2018) discovered associations between recruitment and forage productivity that suggested nutritional conditions on either summer or winter ranges had the most influence on elk recruitment. These associations varied by geographic region: winter range conditions appeared to be more influential in southern areas (Colorado, Utah, and parts of Wyoming), whereas summer range conditions appeared to be more influential in northern areas 3 �(Idaho, Montana, Oregon, Washington, parts of Wyoming). Lukacs et al. (2018) also found that lower total winter precipitation in the previous winter was associated with lower recruitment the next year. The presence of wolves (Canis lupus) and grizzly bears (Ursus arctos) were also associated with lower recruitment. Although juvenile:adult female ratios are often highly correlated with juvenile elk survival (Raithel et al. 2007, Harris et al. 2008), they are an imperfect estimate of recruitment because they are affected by harvest, pregnancy rates, juvenile survival, and adult female survival (Caughley 1974, Gaillard et al. 2000, Harris et al. 2008, Decesare et al. 2012, Lukacs et al. 2018). This makes it difficult to identify ultimate factors influencing population dynamics using age ratio data alone, as multiple scenarios can produce equivalent ratios (Caughley 1974). Thus, long-term demographic studies based on monitoring of marked individuals are necessary to reliably test biological hypotheses and evaluate factors affecting each stage of production and survival (Gaillard et al. 2000, Clutton-Brock and Sheldon 2010, Proffitt et al. 2014). In the absence of harvest, the population dynamics of ungulates are normally characterized by high and stable adult female survival and variable juvenile survival (Gaillard et al. 1998, 2000). Among vital rates, changes in adult female survival have the potential to exert the most influence on population growth rates, but usually do not because of low year-to-year variability (Gaillard et al. 1998, 2000). Instead, adult female survival is typically buffered against moderate environmental variation, and thus not strongly influenced by climatic or densitydependent factors (Gaillard et al. 2000). Indeed, in a large-scale meta-analysis, Brodie et al. (2013) demonstrated that adult female elk survival was principally related to human harvest, and found that this harvest was an additive source of mortality. In contrast to adult female survival, and despite its lower elasticity (de Kroon et al. 1986), juvenile survival frequently has a greater effect on population growth rates of ungulates because of its high variability (Gaillard et al. 1998, 2000, Raithel et al. 2007). Juvenile survival of ungulates may be influenced by abiotic or biotic factors such as environmental conditions, forage quality or quantity, population density, maternal body condition, and predation (Barber-Meyer et al. 2008, Griffin et al. 2011, Monteith et al. 2014, Bastille-Rousseau et al. 2016). Complex interactions among these factors frequently make it difficult to identify the relative role of top-down and bottom-up factors affecting calf survival (Linnell et al. 1995). Further complexity may be introduced if top-down and bottom-up forces are simultaneously and variably influencing survival (Bowyer et al. 2005, Monteith et al. 2014). Juvenile survival has been found to vary substantially among and within elk populations on an annual basis (Garrott et al. 2003, Raithel et al. 2007, Griffin et al. 2011). Griffin et al. (2011) synthesized elk neonatal survival across 12 populations in the northwestern United States, and found that survival declined after warmer previous summers and with more predator species, and increased with higher May precipitation. In resource-limited populations, weather conditions may heavily influence juvenile survival. For example, in an unharvested elk population, prior to the establishment of wolves, Garrott et al. (2003) found that the dominant source of calf mortality in Yellowstone National Park was starvation, with more severe winters leading to lower calf survival. Elsewhere, mortality due to predation has been identified as the most significant cause of death for elk calves (e.g., Barber-Meyer et al. 2008). In some systems black bears (Ursus americanus) are the dominant predator (White et al. 2010, Tatman et al. 2018), whereas in others mountain lions (Puma concolor) kill the most calves (Eacker et al. 2016). Although non-predation deaths (i.e., disease, starvation) may be low where high levels of juvenile predation occur, it remains difficult to determine whether neonate predation represents 4 �an additive or compensatory source of mortality (Linnell et al. 1995, Barber-Meyer et al. 2008, Griffin et al. 2011). Individual calf characteristics may also influence risk of mortality. To date, evidence for sex-biased mortality in elk calves is equivocal. Some studies have documented increased vulnerability of male calves to predation (Smith and Anderson 1996, Eacker et al. 2016), whereas others have demonstrated increased vulnerability of female calves (White et al. 2010). It has been suggested that differences may be due to the hunting behavior of the dominant predator, with stalking predators, such as mountain lions, disproportionately killing male calves, which may engage in riskier exploratory behavior (Eacker et al. 2016). Conclusions about the effect of birth mass on elk calf survival are similarly ambiguous. Some studies have reported that birth mass influenced survival probability (Singer et al. 1997, White et al. 2010), but others have not detected an influence of birth mass on neonate mortality (Smith and Anderson 1996, Eacker et al. 2016). The ability for nutritional resources on the landscape to support ungulate populations is reflected in the nutritional condition of individuals within those populations (Parker et al. 2009, Cook et al. 2013, Monteith et al. 2014). As nutritional resources decline there is typically a predictable sequence of changes in the vital rates of large herbivore populations: first, juvenile survival decreases, then age of first reproduction increases, followed by decreased fertility of adult females, and finally increased mortality rates of adults (Eberhardt 1977a, 1977b, 2002, Gaillard et al. 1998, 2000). This general sequence has been confirmed for elk, with forage quality and associated nutritional body condition demonstrated to affect intra-uterine survival, yearling and adult pregnancy rates, birth dates, birth weight, calf growth, and winter survival of juveniles and adults (Cook et al. 2004a, 2004b, 2013, Proffitt et al. 2016). When nutritional resources are limited, elk calves may be lighter at birth, be born later, and have slower growth rates during summer, which may predispose them to predation mortality. Cook et al. (2004a, 2013) demonstrated that summer-autumn nutrition can play a central role in determining elk productivity, as, during this time, animals must meet the demands of lactation and accrue sufficient fat to get pregnant and survive the winter. For elk, nutritional resources on the landscape may be affected by wild and domestic herbivory (Vavra and Sheehy 1996, Vavra et al. 2007), timber management (Visscher and Merrill 2009), climatic conditions (Middleton et al. 2013), and fire history (Proffitt et al. 2016). To properly determine factors affecting juvenile ungulate survival bottom-up nutritional effects and top-down predation effects should be evaluated together (Monteith et al. 2014). Understanding how these effects are influencing elk population dynamics in Colorado is critical for guiding management actions. In 2016, Colorado Parks and Wildlife (CPW) initiated a 2-year pilot study to investigate factors influencing elk recruitment in 2 study areas in the state. During the pilot study, researchers collected data on annual elk calf survival, pregnancy rates, and latewinter body condition of adult female elk. In July 2018 – July 2019, we expanded this research into a 3rd study area to better determine how predators, habitat, and weather conditions are impacting elk recruitment in Colorado, and to provide management recommendations for increasing juvenile recruitment. STUDY AREAS For management purposes, the elk population in Colorado is divided into 43 Data Analysis Units (DAUs), each of which encompass the year-round range of an elk herd. This 5 �project is being conducted in 3 DAUs, 2 with low juvenile:adult female ratios (E-20, E-33), and 1 with high juvenile:adult female ratios (E-2), which will serve as a reference area (Fig. 1). The Uncompahgre Plateau elk herd (DAU E-20; 5,858 km2) is on the Uncompahgre Plateau in southwest Colorado, USA. In 2017, the post-hunt population of the Uncompahgre Plateau herd was estimated to be ~8,500 elk. From 2013-2017 juvenile:adult female ratios in E20 averaged 30 calves per 100 adult females. Landownership is a mixture of BLM (38%), USFS (37%), private (24%), and state (1%) lands. Elevations range from 1,390 to 3,150 m. The plateau is characterized by a mixture of pinyon-juniper (Pinus edulis, Juniperus osteosperma) woodlands and sage-grassland communities (Artemisia spp., Cercocarpus montanus, Achnatherum hymenoides) at lower elevations. At mid elevations, ponderosa pine (Pinus ponderosa) and mountain shrub communities (Amelanchier alnifolia, Arctostaphylos, Artemisia spp., Quercus gambelii, Symphoricarpos spp.) predominate. Spruce-fir (Picea engelmannii, Abies lasiocarpa, Pseudotsuga menziesii) and aspen (Populus tremuloides) forests dominate at higher elevations. The Trinchera elk herd (DAU E-33; 8,601 km2) is in southeast Colorado, USA. In 2017, the post-hunt population of the Trinchera herd was estimated to be ~16,600 elk. From 2013-2017 juvenile:adult female ratios in E-33 averaged 26 calves per 100 adult females. Landownership is a mixture of private (89%), USFS (3%), state (3%), BLM (2%), U.S. Fish and Wildlife Service (USFWS; 1%), and other (2%) lands. Elevations range from 1,640 to 4,370 m. Lower elevations are characterized by agriculture and sage-grassland communities. At mid elevations, pinyonjuniper woodlands, ponderosa pine and mountain shrub forests, and spruce-fir and aspen forests predominate. Alpine tundra communities dominate at higher elevations. The Bear’s Ears elk herd (DAU E-2; 7,293 km2) is in northwest Colorado, USA. In 2017, the post-hunt population of the Bear’s Ears herd was estimated to be 20,000-24,000 elk. From 2013-2017 juvenile:adult female ratios in E-2 averaged 56 calves per 100 adult females. Landownership is a mixture of private (50%), USFS (25%), BLM (19%), state (5%) and other (1%) lands. Elevations range from 1,730 to 3,710 m. The DAU is characterized by sagegrassland communities at lower elevations. At mid elevations, mountain shrub communities predominate. Spruce-fir and aspen forests dominate at higher elevations. METHODS Capture and handling — We captured adult female elk ≥2 years of age from each study herd by helicopter net-gunning during late winter (March). During capture, we marked individuals with ear tags, collected a blood sample, and measured hind foot length, chest girth, and age based on tooth eruption and wear patterns. We used a portable ultrasound machine to assess whether or not captured elk were pregnant, and estimated the percent of ingesta-free body fat (IFBF) following methods detailed in Cook et al. (2010). We verified non-pregnancies using pregnancy-specific protein B (PSPB) analysis of sampled blood. From each study herd, we outfitted pregnant elk with vaginal implant transmitters (VITs) and Global Positioning System (GPS) radio-collars that attempt to acquire a location every 2 h. We deployed VITs that use the satellite communication capabilities of the collar on the adult female to send a notification when the VIT is expelled, signifying a birth. After receiving a birth notification from a VIT, we went to the birth site to capture and collar the newborn elk calf. We blindfolded calves, and wore latex gloves to minimize the transfer of human scent. We measured body mass, hind foot length, chest girth, and determined the gender of captured calves. We outfitted elk neonates with expandable GPS radio-collars or 6 �VHF proximity collars that communicated with the collar on the adult female, and are designed to drop off after 12 months. We also opportunistically located, captured, and collared additional neonates to increase sample sizes. We collected additional measurements (hair moisture, incisor and upper canine eruption, hoof, dew claw, and navel condition) from opportunistically captured calves to estimate age at capture following Johnson (1951) and Eacker (2015). In December, we captured 6-month old elk calves from each study herd by helicopter netgunning. During capture, we measured body mass, hind foot length, chest girth, and determined the gender of captured calves. We outfitted calves with expandable GPS radio-collars that were scheduled to drop off after 6 months. During all captures, we followed CPW’s animal care and use guidelines for capturing and handling elk (CPW ACUC #09-2008). Cause-specific mortality — Within 24 hours of detecting a mortality signal from an elk collar, we attempted to conduct a systematic field investigation to determine the cause of death. We searched the area surrounding kill sites for evidence of predator presence, including predator scats, tracks, and hair, or signs of a struggle (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). We examined elk carcasses for evidence of canine puncture wounds, subcutaneous hemorrhaging and bruising, aspirated blood in the mouth, nose, or trachea, claw or bite marks on the hide, cracked or chewed bones, and characteristic consumption patterns (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). We also collected calf carcasses when they were available to verify field assessments with laboratory necropsies performed by a CPW veterinarian. Nutritional condition of adult female elk — The body fat of lactating and non-lactating adult female elk can vary substantially, as lactating females are more sensitive than non-lactating females to their nutritional environment (Cook et al. 2004a, 2013). Therefore, it is difficult to interpret the body condition of adult female elk in late winter without knowing whether or not they experienced the energetic demands of lactation throughout the previous growing season (Cook et al. 2004a, 2013). RESULTS AND DISCUSSION In December 2020, we collared 50 6-month old elk calves, 25 each from the Bear’s Ears and Uncompahgre Plateau elk herds. The mean weight of calves from the Bear’s Ears herd was 97.9 kg (95% CI = 90.9-104.9 kg) and 102.8 kg (95% CI = 96.6-108.9 kg) from the Uncompahgre Plateau elk herd. During March 2021, we radio-collared 100 pregnant elk and outfitted them with VITs, 40 each from the Bear’s Ears and Uncompahgre Plateau herds, and 20 from the Trinchera herd. We estimated that pregnancy rates of adult female elk were 93% (95% CI = 82-98%) in the Bear’s Ears herd, 95% (95% CI = 78-100%) in the Trinchera herd, and 91% (95% CI = 80-97%) in the Uncompahgre Plateau herd (Fig. 2). Elk populations experiencing good to excellent summerautumn nutrition typically have pregnancy rates ≥90% (Cook et al. 2013). We estimated the mean IFBF of adult female elk to be 7.01% from the Bear’s Ears herd, 7.78% from the Trinchera herd, and 7.22% from the Uncompahgre Plateau herd (Fig. 3). When late-winter IFBF values are <8-9% for adult female elk that have lactated through the previous growing season, this suggests that there may be nutritional limitations, but it does not identify whether limitations are a result of summer-autumn or winter nutrition (R. Cook, personal communication). 7 �During May-July 2021, we captured and collared 126 elk calves, 53 from the Bear’s Ears herd, 21 from the Trinchera herd, and 52 from the Uncompahgre Plateau herd. The estimated mean date of calving was 1 June in the Bear’s Ears herd, 3 June in the Trinchera herd, and 4 June in the Uncompahgre Plateau herd. SUMMARY From July 1, 2020 – June 30, 2021 we successfully worked with private landowners and personnel from CPW to coordinate field research logistics and initiate the third year of this study. During FY2019-20 we successfully worked with private landowners and personnel from CPW to coordinate field research logistics and initiate the second year of this study. We collected data on body condition and reproduction by capturing adult female elk, and we outfitted 100 pregnant females with GPS collars and VITs. We successfully captured and collared 126 newborn elk and 50 6-month old elk calves, meeting our sample size objectives, and allowing us to collect data on calf survival and cause-specific sources of mortality. We will continue to collect data on elk survival and cause-specific sources of mortality throughout the year. LITERATURE CITED Barber-Meyer, S. M., L. D. Mech, and P. J. White. 2008. Elk calf survival and mortality following wolf restoration to Yellowstone National Park. Wildlife Monographs 169:1–30. Bastille-Rousseau, G., J. A. Schaefer, K. P. Lewis, M. A. Mumma, E. H. Ellington, N. D. Rayl, S. P. Mahoney, D. Pouliot, and D. L. Murray. 2016. Phase-dependent climate-predator interactions explain three decades of variation in neonatal caribou survival. Journal of Animal Ecology 85:445–456. Bowyer, T., D. K. Person, and M. P. Becky. 2005. Detecting top-down versus bottom-up regulation of ungulates by large carnivores: implications for conservation of biodiversity. Pages 342–361 in J. C. Ray, K. H. Redford, R. S. Steneck, and J. Berger, editors. Large Carnivores and the Conservation of Biodiversity. Island Press, Washinton, D.C., USA. Brodie, J., H. Johnson, M. Mitchell, P. Zager, K. Proffitt, M. Hebblewhite, M. Kauffman, B. Johnson, J. Bissonette, C. Bishop, J. Gude, J. Herbert, K. Hersey, M. Hurley, P. M. Lukacs, S. Mccorquodale, E. Mcintire, J. Nowak, H. Sawyer, D. Smith, and P. J. White. 2013. Relative influence of human harvest, carnivores, and weather on adult female elk survival across western North America. Journal of Applied Ecology 50:295–305. Caughley, G. 1974. Interpretation of age ratios. Journal of Wildlife Management 38:557–562. Clutton-Brock, T., and B. C. Sheldon. 2010. Individuals and populations: the role of long-term, individual-based studies of animals in ecology and evolutionary biology. Trends in Ecology and Evolution 25:562–573. Cook, J. G., B. K. Johnson, R. C. Cook, R. A. Riggs, T. Delcurto, L. D. Bryant, and L. L. Irwin. 2004a. Effects of summer-autumn nutrition and parturition date on reproduction and survival of elk. Wildlife Monographs 155:1–61. Cook, R. C., J. G. Cook, and L. D. Mech. 2004b. Nutritional condition of northern Yellowstone elk. Journal of Mammalogy 85:714–722. Cook, R. C., J. G. Cook, T. R. Stephenson, W. L. Myers, S. M. Mccorquodale, D. J. Vales, L. L. Irwin, P. B. Hall, R. D. Spencer, S. L. Murphie, K. A. Schoenecker, and P. J. Miller. 2010. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife 8 �Management 74:880–896. Cook, R. C., J. G. Cook, D. J. Vales, B. K. Johnson, S. M. McCorquodale, L. A. Shipley, R. A. Riggs, L. L. Irwin, S. L. Murphie, B. L. Murphie, K. A. Schoenecker, F. Geyer, P. B. Hall, R. D. Spencer, D. A. Immell, D. H. Jackson, B. L. Tiller, P. J. Miller, and L. Schmitz. 2013. Regional and seasonal patterns of nutritional condition and reproduction in elk. Wildlife Monographs 184:1–44. Decesare, N. J., M. Hebblewhite, M. Bradley, K. G. Smith, D. Hervieux, and L. Neufeld. 2012. Estimating ungulate recruitment and growth rates using age ratios. Journal of Wildlife Management 76:144–153. Eacker, D. R. 2015. Linking the effects of risk factors on annual calf survival to elk population dynamics in the Bitterroot Valley, Montana. M.S. thesis, University of Montana, Missoula, MT, USA. Eacker, D. R., M. Hebblewhite, K. M. Proffitt, B. S. Jimenez, M. S. Mitchell, and H. S. Robinson. 2016. Annual elk calf survival in a multiple carnivore system. Journal of Wildlife Management 80:1345–1359. Eberhardt, L. L. 1977a. “Optimal” management policies for marine mammals. Wildlife Society Bulletin 5:162–169. Eberhardt, L. L. 1977b. Optimal policies for conservation of large mammals, with special references to marine ecosystems. Environmental Conservation 4:205–212. Eberhardt, L. L. 2002. A paradigm for population analysis of long-lived vertebrates. Ecology 83:2841–2854. Gaillard, J.-M., M. Festa-Bianchet, and N. G. Yoccoz. 1998. Population dynamics of large herbivores: variable recruitment with constant adult survival. Trends in Ecology and Evolution 13:58–63. Gaillard, J.-M., M. Festa-Bianchet, N. G. Yoccoz, A. Loison, and C. Toigo. 2000. Temporal variation in fitness components and population dynamics of large herbivores. Annual Review of Ecology, Evolution, and Systematics 31:367–393. Garrott, R. A., L. L. Eberhardt, P. J. White, and J. Rotella. 2003. Climate-induced variation in vital rates of an unharvested large-herbivore population. Canadian Journal of Zoology 81:33–45. Griffin, K. A., M. Hebblewhite, H. S. Robinson, P. Zager, S. M. Barber-Meyer, D. Christianson, S. Creel, N. C. Harris, M. A. Hurley, D. H. Jackson, B. K. Johnson, W. L. Myers, J. D. Raithel, M. Schlegel, B. L. Smith, C. White, and P. J. White. 2011. Neonatal mortality of elk driven by climate, predator phenology and predator community composition. Journal of Animal Ecology 80:1246–1257. Harris, N. C., M. J. Kauffman, and L. S. Mills. 2008. Inferences about ungulate population dynamics derived from age ratios. Journal of Wildlife Management 72:1143–1151. Johnson, D. E. 1951. Biology of the elk calf, Cervus canadensis nelsoni. Journal of Wildlife Management 15:396–410. de Kroon, H., A. Plaisier, J. van Groenendael, and H. Caswell. 1986. Elasticity: the relative contribution of demographic parameters to population growth rate. Ecology 67:1427–1431. Linnell, J. D. C., R. Aanes, and R. Andersen. 1995. Who killed Bambi? The role of predation in the neonatal mortality of temperate ungulates. Wildlife Biology 1:209–223. Lukacs, P. M., M. S. Mitchell, M. Hebblewhite, B. K. Johnson, H. Johnson, M. Kauffman, K. M. Proffitt, P. Zager, J. Brodie, K. Hersey, A. A. Holland, M. Hurley, S. McCorquodale, A. Middleton, M. Nordhagen, J. J. Nowak, D. P. Walsh, and P. J. White. 2018. Factors influencing elk recruitment across ecotypes in the western United States. Journal of Wildlife 9 �Management 82:698–710. Middleton, A. D., M. J. Kauffman, D. E. Mcwhirter, J. G. Cook, R. C. Cook, A. A. Nelson, M. D. Jimenez, and R. W. Klaver. 2013. Animal migration amid shifting patterns of phenology and predation: lessons from a Yellowstone elk herd. Ecology 94:1245–1256. Monteith, K. L., V. C. Bleich, T. R. Stephenson, B. M. Pierce, M. M. Conner, J. G. Kie, and R. T. Bowyer. 2014. Life-history characteristics of mule deer: effects of nutrition in a variable environment. Wildlife Monographs 186:1–62. Parker, K. L., P. S. Barboza, and M. P. Gillingham. 2009. Nutrition integrates environmental responses of ungulates. Functional Ecology 23:57–69. Proffitt, K. M., J. A. Cunningham, K. L. Hamlin, and R. A. Garrott. 2014. Bottom-up and topdown influences on pregnancy rates and recruitment of northern Yellowstone elk. Journal of Wildlife Management 78:1383–1393. Proffitt, K. M., M. Hebblewhite, W. Peters, N. Hupp, and J. Shamhart. 2016. Linking landscapescale differences in forage to ungulate nutritional ecology. Ecological Applications 26:2156– 2174. Raithel, J. D., M. J. Kauffman, and D. H. Pletscher. 2007. Impact of spatial and temporal variation in calf survival on the growth of elk population. Journal of Wildlife Management 71:795– 803. Singer, F. J., A. Harting, K. K. Symonds, and M. B. Coughenour. 1997. Density dependence, compensation, and environmental effects on elk calf mortality in Yellowstone National Park. Journal of Wildlife Management 61:12–25. Smith, B. L., and S. H. Anderson. 1996. Patterns of neonatal mortality of elk in northwest Wyoming. Canadian Journal of Zoology 74:1229–1237. Stonehouse, K. F., C. R. J. Anderson, M. E. Peterson, and D. R. Collins. 2016. Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus). Technical Publication Number 48, Colorado Parks and Wildlife, Fort Collins, CO, USA. Swift, L. W. 1945. A partial history of the elk herds of Colorado. Journal of Mammalogy 26:114– 119. Tatman, N. M., S. G. Liley, J. W. Cain, and J. W. Pitman. 2018. Effects of calf predation and nutrition on elk vital rates. Journal of Wildlife Management 82:1417–1428. Vavra, M., C. G. Parks, and M. J. Wisdom. 2007. Biodiversity, exotic plant species, and herbivory: the good, the bad, and the ungulate. Forest Ecology and Management 246:66–72. Vavra, M., and D. P. Sheehy. 1996. Improving elk habitat characteristics with livestock grazing. Rangelands 18:182–185. Visscher, D. R., and E. H. Merrill. 2009. Temporal dynamics of forage succession for elk at two scales: implications of forest management. Forest Ecology and Management 257:96–106. White, C. G., P. Zager, and M. W. Gratson. 2010. Influence of predator harvest, biological factors, and landscape on elk calf survival in Idaho. Journal of Wildlife Management 74:355–369. Prepared by Nathaniel D. Rayl, Wildlife Researcher 10 �Calves:100 adult females (2013-2017) Insufficient data 0 25-300 30-350 35-400 40-450 45-500 50-55 E-4 55-60 E-51 E-99 j E-24 E-33 0 30 60 Miles Figure 1. Number of elk calves per 100 adult females observed during December-February aerial surveys (5-year average from 2013-2017) within elk Data Analysis Units (DAUs; labeled with black text) in Colorado, USA. 11 �■ Bears Ear's ■ Trinchera 1.0 27 31 2017 2018 II Uncompahgre Plateau 31 0.9 0.8 ..... C ~ 0.7 O> a. 0.6 Q) C o 0.5 t 8_0.4 e CL 0.3 0.2 0.1 0.0 2019 2020 2021 Year Figure 2. Estimated average pregnancy rates of adult female elk from the Bear’s Ears, Trinchera, and Uncompahgre Plateau herds sampled during late winter 2017-2021 in Colorado, USA. The sample size is given at the top of the 95% binomial confidence intervals (black lines). 12 �13- • 12• .?f:. 11 - ---.....cu .__ 10~ "'O 0 ..c Q) Q) 9- '.__ .....roen aQ) g> 7 6- 5' Bear's Ears ' Trinchera Herd ' Uncompahgre Plateau Figure 3. The estimated ingesta-free body fat (%) of adult female elk from the Bear’s Ears (n = 44), Trinchera (n = 22), and Uncompahgre Plateau (n = 47) herds during late-winter 2021 in Colorado, USA. 13 �Colorado Parks and Wildlife July 2021 – June 2022 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Colorado 3430 3002 1 : : : : Federal Aid Project No. W-242-R6 : Parks and Wildlife Mammals Research Elk Conservation Evaluating factors influencing elk recruitment in Colorado Period Covered: July 1, 2021 – June 30, 2022 Authors: N.D. Rayl, M.W. Alldredge, and C.R. Anderson Jr. Personnel: R. Aberle, J. Alley, T. Bonacquista, K. Bond, M. Brown, M. Caddy, M. Calahan, Z. Chrisman, A. Cole, D. Corcoran, K. Crane, S. Crew, A. Davis, B. de Vergie, K. Duckett, L. Emerick, D. Finley, P. Firmin, K. Fischer, M. Fisher, K. Fox, A. Friedel, M. Gallagher, M. George, L. Gephert, J. Goncalves, K. Hayes, A. Kircher, J. Lambert, D. Leer, E. Los, K. Middledorf, L. Miller, M. Miller, S. Mollett, H. Mondin, E. Monfort, E. Newkirk, P. Nol, K. Oldham, S. Olson, M. Ortega, J. Ortiz Calo, J. Pollock, J. Potter, N. Renneker, B. Rubalcaba, E. Sawa, S. Sinclair, G. Smith, B. Smith, R. Sralla, M. Trujillo, E. VanNatta, A. Vitt, S. Waters, H. Westacott, M. Wood, CPW; J. Clark, H. Cushman, B. Dooling, A. Orlando, R. Swisher, S. Swisher, Quicksilver Air, Inc.. Project support received from Federal Aid in Wildlife Restoration, Rocky Mountain Elk Foundation, and CPW Big Game Auction and Raffle. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. ABSTRACT Over the last two decades, wildlife managers in Colorado have become increasingly concerned about declining winter elk calf recruitment (estimated using juvenile:adult female ratios) in the southern portion of the state. Although juvenile:adult female ratios are often highly correlated with juvenile elk survival, they are an imperfect estimate of recruitment because they are affected by harvest, pregnancy rates, juvenile survival, and adult female survival. Thus, there is a need for elk research in Colorado based upon monitoring of marked individuals to evaluate factors affecting each stage of production and survival. Colorado Parks and Wildlife (CPW) is conducting research in 3 study areas to better determine how predators, habitat, and weather conditions are impacting elk recruitment in Colorado. From July 1, 2021 – June 30, 2022, we focused on working with stakeholders and collaborators on research logistics, and capturing and collaring elk. Field efforts were centered on 2 objectives: 1) capturing adult female elk, and collaring and outfitting pregnant females with vaginal implant transmitters (VITs) to collect data on elk 1 �demography, body condition, reproduction, and behavior, and 2) capturing and collaring newborn and 6-month old elk to collect data on calf survival and cause-specific mortality. We radio-collared 50 6-month old elk calves in December 2021. In March 2022, we radio-collared 80 pregnant elk and outfitted them with VITs. Estimates of average pregnancy rates ranged from 87-91% across herds, and estimates of mean ingesta-free body fat ranged from 7.22-8.02 across herds. During the 2022 calving season, we radio-collared 107 elk calves. Mean calving date across herds was 2 June. 2 �WILDLIFE RESEARCH REPORT EVALUATING FACTORS INFLUENCING ELK RECRUITMENT IN COLORADO NATHANIEL D. RAYL, MAT W. ALLDREDGE, AND CHUCK R. ANDERSON JR. PROJECT NARRATIVE OBJECTIVES The objectives of this project are to 1) estimate elk calf survival and cause-specific mortality rates from birth to age 1 to evaluate the importance of mortality sources for elk calf survival, 2) evaluate the influence of biotic (birth date, birth mass, gender, maternal body condition, habitat conditions) and abiotic factors (previous and current weather conditions) on seasonal mortality risk of elk calves from birth to age 1, 3) assess the health of elk herds by quantifying pregnancy rates and percent ingesta-free body fat (IFBF) of adult female elk, and 4) evaluate the influence of biotic (age, body condition, reproductive status, habitat conditions and selection) and abiotic factors (previous and current weather conditions) on pregnancy rates and IFBF. SEGMENT OBJECTIVES 1. Work with personnel from CPW Areas 6, 10, 11, and 18, and private landowners on field research logistics. 2. Capture adult female elk, and collar and outfit pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior. 3. Capture and collar neonate and 6-month old elk to collect data on calf survival and causespecific sources of mortality. INTRODUCTION In Colorado, elk (Cervus canadensis) are an important natural resource that are valued for ecological, consumptive, aesthetic, and economic reasons. In 1910, fewer than 1,000 elk remained in Colorado (Swift 1945), but today the state population is estimated to be the largest in the country, with over 292,000 elk. Over the last two decades, however, there has been increasing concern among wildlife managers in Colorado about declining winter calf recruitment (estimated using juvenile:adult female ratios) in the southern portion of the state (Fig. 1; Colorado Parks and Wildlife, unpublished data). In many elk populations, similar declining trends in juvenile recruitment have been observed. A recent synthesis examining elk recruitment in the western United States from 19892010 found evidence of a long-term reduction of 0.48 juveniles/100 adult females/year (Lukacs et al. 2018). Lukacs et al. (2018) discovered associations between recruitment and forage productivity that suggested nutritional conditions on either summer or winter ranges had the most influence on elk recruitment. These associations varied by geographic region: winter range conditions appeared to be more influential in southern areas (Colorado, Utah, and parts of Wyoming), whereas summer range conditions appeared to be more influential in northern areas 3 �(Idaho, Montana, Oregon, Washington, parts of Wyoming). Lukacs et al. (2018) also found that lower total winter precipitation in the previous winter was associated with lower recruitment the next year. The presence of wolves (Canis lupus) and grizzly bears (Ursus arctos) were also associated with lower recruitment. Although juvenile:adult female ratios are often highly correlated with juvenile elk survival (Raithel et al. 2007, Harris et al. 2008), they are an imperfect estimate of recruitment because they are affected by harvest, pregnancy rates, juvenile survival, and adult female survival (Caughley 1974, Gaillard et al. 2000, Harris et al. 2008, Decesare et al. 2012, Lukacs et al. 2018). This makes it difficult to identify ultimate factors influencing population dynamics using age ratio data alone, as multiple scenarios can produce equivalent ratios (Caughley 1974). Thus, long-term demographic studies based on monitoring of marked individuals are necessary to reliably test biological hypotheses and evaluate factors affecting each stage of production and survival (Gaillard et al. 2000, Clutton-Brock and Sheldon 2010, Proffitt et al. 2014). In the absence of harvest, the population dynamics of ungulates are normally characterized by high and stable adult female survival and variable juvenile survival (Gaillard et al. 1998, 2000). Among vital rates, changes in adult female survival have the potential to exert the most influence on population growth rates, but usually do not because of low year-to-year variability (Gaillard et al. 1998, 2000). Instead, adult female survival is typically buffered against moderate environmental variation, and thus not strongly influenced by climatic or densitydependent factors (Gaillard et al. 2000). Indeed, in a large-scale meta-analysis, Brodie et al. (2013) demonstrated that adult female elk survival was principally related to human harvest, and found that this harvest was an additive source of mortality. In contrast to adult female survival, and despite its lower elasticity (de Kroon et al. 1986), juvenile survival frequently has a greater effect on population growth rates of ungulates because of its high variability (Gaillard et al. 1998, 2000, Raithel et al. 2007). Juvenile survival of ungulates may be influenced by abiotic or biotic factors such as environmental conditions, forage quality or quantity, population density, maternal body condition, and predation (Barber-Meyer et al. 2008, Griffin et al. 2011, Monteith et al. 2014, Bastille-Rousseau et al. 2016). Complex interactions among these factors frequently make it difficult to identify the relative role of top-down and bottom-up factors affecting calf survival (Linnell et al. 1995). Further complexity may be introduced if top-down and bottom-up forces are simultaneously and variably influencing survival (Bowyer et al. 2005, Monteith et al. 2014). Juvenile survival has been found to vary substantially among and within elk populations on an annual basis (Garrott et al. 2003, Raithel et al. 2007, Griffin et al. 2011). Griffin et al. (2011) synthesized elk neonatal survival across 12 populations in the northwestern United States, and found that survival declined after warmer previous summers and with more predator species, and increased with higher May precipitation. In resource-limited populations, weather conditions may heavily influence juvenile survival. For example, in an unharvested elk population, prior to the establishment of wolves, Garrott et al. (2003) found that the dominant source of calf mortality in Yellowstone National Park was starvation, with more severe winters leading to lower calf survival. Elsewhere, mortality due to predation has been identified as the most significant cause of death for elk calves (e.g., Barber-Meyer et al. 2008). In some systems black bears (Ursus americanus) are the dominant predator (White et al. 2010, Tatman et al. 2018), whereas in others mountain lions (Puma concolor) kill the most calves (Eacker et al. 2016). Although non-predation deaths (i.e., disease, starvation) may be low where high levels of juvenile predation occur, it remains difficult to determine whether neonate predation represents 4 �an additive or compensatory source of mortality (Linnell et al. 1995, Barber-Meyer et al. 2008, Griffin et al. 2011). Individual calf characteristics may also influence risk of mortality. To date, evidence for sex-biased mortality in elk calves is equivocal. Some studies have documented increased vulnerability of male calves to predation (Smith and Anderson 1996, Eacker et al. 2016), whereas others have demonstrated increased vulnerability of female calves (White et al. 2010). It has been suggested that differences may be due to the hunting behavior of the dominant predator, with stalking predators, such as mountain lions, disproportionately killing male calves, which may engage in riskier exploratory behavior (Eacker et al. 2016). Conclusions about the effect of birth mass on elk calf survival are similarly ambiguous. Some studies have reported that birth mass influenced survival probability (Singer et al. 1997, White et al. 2010), but others have not detected an influence of birth mass on neonate mortality (Smith and Anderson 1996, Eacker et al. 2016). The ability for nutritional resources on the landscape to support ungulate populations is reflected in the nutritional condition of individuals within those populations (Parker et al. 2009, Cook et al. 2013, Monteith et al. 2014). As nutritional resources decline there is typically a predictable sequence of changes in the vital rates of large herbivore populations: first, juvenile survival decreases, then age of first reproduction increases, followed by decreased fertility of adult females, and finally increased mortality rates of adults (Eberhardt 1977a, 1977b, 2002, Gaillard et al. 1998, 2000). This general sequence has been confirmed for elk, with forage quality and associated nutritional body condition demonstrated to affect intra-uterine survival, yearling and adult pregnancy rates, birth dates, birth weight, calf growth, and winter survival of juveniles and adults (Cook et al. 2004a, 2004b, 2013, Proffitt et al. 2016). When nutritional resources are limited, elk calves may be lighter at birth, be born later, and have slower growth rates during summer, which may predispose them to predation mortality. Cook et al. (2004a, 2013) demonstrated that summer-autumn nutrition can play a central role in determining elk productivity, as, during this time, animals must meet the demands of lactation and accrue sufficient fat to get pregnant and survive the winter. For elk, nutritional resources on the landscape may be affected by wild and domestic herbivory (Vavra and Sheehy 1996, Vavra et al. 2007), timber management (Visscher and Merrill 2009), climatic conditions (Middleton et al. 2013), and fire history (Proffitt et al. 2016). To properly determine factors affecting juvenile ungulate survival bottom-up nutritional effects and top-down predation effects should be evaluated together (Monteith et al. 2014). Understanding how these effects are influencing elk population dynamics in Colorado is critical for guiding management actions. In 2016, Colorado Parks and Wildlife (CPW) initiated a 2-year pilot study to investigate factors influencing elk recruitment in 2 study areas in the state. During the pilot study, researchers collected data on annual elk calf survival, pregnancy rates, and latewinter body condition of adult female elk. In July 2018 – July 2019, we expanded this research into a 3rd study area to better determine how predators, habitat, and weather conditions are impacting elk recruitment in Colorado, and to provide management recommendations for increasing juvenile recruitment. STUDY AREAS For management purposes, the elk population in Colorado is divided into 43 Data Analysis Units (DAUs), each of which encompass the year-round range of an elk herd. This 5 �project is being conducted in 3 DAUs, 2 with low juvenile:adult female ratios (E-20, E-33), and 1 with high juvenile:adult female ratios (E-2), which will serve as a reference area (Fig. 1). The Uncompahgre Plateau elk herd (DAU E-20; 5,858 km2) is on the Uncompahgre Plateau in southwest Colorado, USA. In 2017, the post-hunt population of the Uncompahgre Plateau herd was estimated to be ~8,500 elk. From 2013-2017 juvenile:adult female ratios in E20 averaged 30 calves per 100 adult females. Landownership is a mixture of BLM (38%), USFS (37%), private (24%), and state (1%) lands. Elevations range from 1,390 to 3,150 m. The plateau is characterized by a mixture of pinyon-juniper (Pinus edulis, Juniperus osteosperma) woodlands and sage-grassland communities (Artemisia spp., Cercocarpus montanus, Achnatherum hymenoides) at lower elevations. At mid elevations, ponderosa pine (Pinus ponderosa) and mountain shrub communities (Amelanchier alnifolia, Arctostaphylos, Artemisia spp., Quercus gambelii, Symphoricarpos spp.) predominate. Spruce-fir (Picea engelmannii, Abies lasiocarpa, Pseudotsuga menziesii) and aspen (Populus tremuloides) forests dominate at higher elevations. The Trinchera elk herd (DAU E-33; 8,601 km2) is in southeast Colorado, USA. In 2017, the post-hunt population of the Trinchera herd was estimated to be ~16,600 elk. From 2013-2017 juvenile:adult female ratios in E-33 averaged 26 calves per 100 adult females. Landownership is a mixture of private (89%), USFS (3%), state (3%), BLM (2%), U.S. Fish and Wildlife Service (USFWS; 1%), and other (2%) lands. Elevations range from 1,640 to 4,370 m. Lower elevations are characterized by agriculture and sage-grassland communities. At mid elevations, pinyonjuniper woodlands, ponderosa pine and mountain shrub forests, and spruce-fir and aspen forests predominate. Alpine tundra communities dominate at higher elevations. The Bear’s Ears elk herd (DAU E-2; 7,293 km2) is in northwest Colorado, USA. In 2017, the post-hunt population of the Bear’s Ears herd was estimated to be 20,000-24,000 elk. From 2013-2017 juvenile:adult female ratios in E-2 averaged 56 calves per 100 adult females. Landownership is a mixture of private (50%), USFS (25%), BLM (19%), state (5%) and other (1%) lands. Elevations range from 1,730 to 3,710 m. The DAU is characterized by sagegrassland communities at lower elevations. At mid elevations, mountain shrub communities predominate. Spruce-fir and aspen forests dominate at higher elevations. METHODS Capture and handling — We captured adult female elk ≥2 years of age from each study herd by helicopter net-gunning during late winter (March). During capture, we marked individuals with ear tags, collected a blood sample, and measured hind foot length, chest girth, and age based on tooth eruption and wear patterns. We used a portable ultrasound machine to assess whether or not captured elk were pregnant, and estimated the percent of ingesta-free body fat (IFBF) following methods detailed in Cook et al. (2010). We verified non-pregnancies using pregnancy-specific protein B (PSPB) analysis of sampled blood. From each study herd, we outfitted pregnant elk with vaginal implant transmitters (VITs) and Global Positioning System (GPS) radio-collars that attempt to acquire a location every 2 h. We deployed VITs that use the satellite communication capabilities of the collar on the adult female to send a notification when the VIT is expelled, signifying a birth. After receiving a birth notification from a VIT, we went to the birth site to capture and collar the newborn elk calf. We blindfolded calves, and wore latex gloves to minimize the transfer of human scent. We measured body mass, hind foot length, chest girth, and determined the gender of captured calves. We outfitted elk neonates with expandable GPS radio-collars or 6 �VHF proximity collars that communicated with the collar on the adult female, and are designed to drop off after 12 months. We also opportunistically located, captured, and collared additional neonates to increase sample sizes. We collected additional measurements (hair moisture, incisor and upper canine eruption, hoof, dew claw, and navel condition) from opportunistically captured calves to estimate age at capture following Johnson (1951) and Eacker (2015). In December, we captured 6-month old elk calves from each study herd by helicopter netgunning. During capture, we measured body mass, hind foot length, chest girth, and determined the gender of captured calves. We outfitted calves with expandable GPS radio-collars that were scheduled to drop off after 6 months. During all captures, we followed CPW’s animal care and use guidelines for capturing and handling elk (CPW ACUC #09-2008). Cause-specific mortality — Within 24 hours of detecting a mortality signal from an elk collar, we attempted to conduct a systematic field investigation to determine the cause of death. We searched the area surrounding kill sites for evidence of predator presence, including predator scats, tracks, and hair, or signs of a struggle (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). We examined elk carcasses for evidence of canine puncture wounds, subcutaneous hemorrhaging and bruising, aspirated blood in the mouth, nose, or trachea, claw or bite marks on the hide, cracked or chewed bones, and characteristic consumption patterns (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). We also collected calf carcasses when they were available to verify field assessments with laboratory necropsies performed by a CPW veterinarian. Nutritional condition of adult female elk — The body fat of lactating and non-lactating adult female elk can vary substantially, as lactating females are more sensitive than non-lactating females to their nutritional environment (Cook et al. 2004a, 2013). Therefore, it is difficult to interpret the body condition of adult female elk in late winter without knowing whether or not they experienced the energetic demands of lactation throughout the previous growing season (Cook et al. 2004a, 2013). RESULTS AND DISCUSSION In December 2021, we collared 50 6-month old elk calves, 25 each from the Bear’s Ears and Uncompahgre Plateau elk herds. The mean weight of calves from the Bear’s Ears herd was 100.2 kg (95% CI = 94.7-105.7 kg) and 103.7 kg (95% CI = 99-108.3 kg) from the Uncompahgre Plateau elk herd. During March 2022, we radio-collared 100 pregnant elk and outfitted them with VITs, 40 each from the Bear’s Ears and Uncompahgre Plateau herds. We estimated that pregnancy rates of adult female elk were 91% (95% CI = 79-96%) in the Bear’s Ears herd, and 87% (95% CI = 7494%) in the Uncompahgre Plateau herd (Fig. 2). Elk populations experiencing good to excellent summer-autumn nutrition typically have pregnancy rates ≥90% (Cook et al. 2013). We estimated the mean IFBF of adult female elk to be 7.22% from the Bear’s Ears herd and 8.02% from the Uncompahgre Plateau herd (Fig. 3). When late-winter IFBF values are <8-9% for adult female elk that have lactated through the previous growing season, this suggests that there may be nutritional limitations, but it does not identify whether limitations are a result of summer-autumn or winter nutrition (R. Cook, personal communication). 7 �During May-July 2021, we captured and collared 107 elk calves, 54 from the Bear’s Ears herd, and 53 from the Uncompahgre Plateau herd. The estimated mean date of calving was 1 June in the Bear’s Ears herd, and 2 June in the Uncompahgre Plateau herd. SUMMARY From July 1, 2021 – June 30, 2022 we successfully worked with private landowners and personnel from CPW to coordinate field research logistics and initiate the fourth year of this study. We collected data on body condition and reproduction by capturing adult female elk, and we outfitted 80 pregnant females with GPS collars and VITs. We successfully captured and collared 107 newborn elk and 50 6-month old elk calves, meeting our sample size objectives, and allowing us to collect data on calf survival and cause-specific sources of mortality. LITERATURE CITED Barber-Meyer, S. M., L. D. Mech, and P. J. White. 2008. Elk calf survival and mortality following wolf restoration to Yellowstone National Park. Wildlife Monographs 169:1–30. Bastille-Rousseau, G., J. A. Schaefer, K. P. Lewis, M. A. Mumma, E. H. Ellington, N. D. Rayl, S. P. Mahoney, D. Pouliot, and D. L. Murray. 2016. Phase-dependent climate-predator interactions explain three decades of variation in neonatal caribou survival. Journal of Animal Ecology 85:445–456. Bowyer, T., D. K. Person, and M. P. Becky. 2005. Detecting top-down versus bottom-up regulation of ungulates by large carnivores: implications for conservation of biodiversity. Pages 342–361 in J. C. Ray, K. H. Redford, R. S. Steneck, and J. Berger, editors. Large Carnivores and the Conservation of Biodiversity. Island Press, Washinton, D.C., USA. Brodie, J., H. Johnson, M. Mitchell, P. Zager, K. Proffitt, M. Hebblewhite, M. Kauffman, B. Johnson, J. Bissonette, C. Bishop, J. Gude, J. Herbert, K. Hersey, M. Hurley, P. M. Lukacs, S. Mccorquodale, E. Mcintire, J. Nowak, H. Sawyer, D. Smith, and P. J. White. 2013. Relative influence of human harvest, carnivores, and weather on adult female elk survival across western North America. Journal of Applied Ecology 50:295–305. Caughley, G. 1974. Interpretation of age ratios. Journal of Wildlife Management 38:557–562. Clutton-Brock, T., and B. C. Sheldon. 2010. Individuals and populations: the role of long-term, individual-based studies of animals in ecology and evolutionary biology. Trends in Ecology and Evolution 25:562–573. Cook, J. G., B. K. Johnson, R. C. Cook, R. A. Riggs, T. Delcurto, L. D. Bryant, and L. L. Irwin. 2004a. Effects of summer-autumn nutrition and parturition date on reproduction and survival of elk. Wildlife Monographs 155:1–61. Cook, R. C., J. G. Cook, and L. D. Mech. 2004b. Nutritional condition of northern Yellowstone elk. Journal of Mammalogy 85:714–722. Cook, R. C., J. G. Cook, T. R. Stephenson, W. L. Myers, S. M. Mccorquodale, D. J. Vales, L. L. Irwin, P. B. Hall, R. D. Spencer, S. L. Murphie, K. A. Schoenecker, and P. J. Miller. 2010. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife Management 74:880–896. Cook, R. C., J. G. Cook, D. J. Vales, B. K. Johnson, S. M. McCorquodale, L. A. Shipley, R. A. Riggs, L. L. Irwin, S. L. Murphie, B. L. Murphie, K. A. Schoenecker, F. Geyer, P. B. Hall, R. D. Spencer, D. A. Immell, D. H. Jackson, B. L. Tiller, P. J. Miller, and L. Schmitz. 2013. 8 �Regional and seasonal patterns of nutritional condition and reproduction in elk. Wildlife Monographs 184:1–44. Decesare, N. J., M. Hebblewhite, M. Bradley, K. G. Smith, D. Hervieux, and L. Neufeld. 2012. Estimating ungulate recruitment and growth rates using age ratios. Journal of Wildlife Management 76:144–153. Eacker, D. R. 2015. Linking the effects of risk factors on annual calf survival to elk population dynamics in the Bitterroot Valley, Montana. M.S. thesis, University of Montana, Missoula, MT, USA. Eacker, D. R., M. Hebblewhite, K. M. Proffitt, B. S. Jimenez, M. S. Mitchell, and H. S. Robinson. 2016. Annual elk calf survival in a multiple carnivore system. Journal of Wildlife Management 80:1345–1359. Eberhardt, L. L. 1977a. “Optimal” management policies for marine mammals. Wildlife Society Bulletin 5:162–169. Eberhardt, L. L. 1977b. Optimal policies for conservation of large mammals, with special references to marine ecosystems. Environmental Conservation 4:205–212. Eberhardt, L. L. 2002. A paradigm for population analysis of long-lived vertebrates. Ecology 83:2841–2854. Gaillard, J.-M., M. Festa-Bianchet, and N. G. Yoccoz. 1998. Population dynamics of large herbivores: variable recruitment with constant adult survival. Trends in Ecology and Evolution 13:58–63. Gaillard, J.-M., M. Festa-Bianchet, N. G. Yoccoz, A. Loison, and C. Toigo. 2000. Temporal variation in fitness components and population dynamics of large herbivores. Annual Review of Ecology, Evolution, and Systematics 31:367–393. Garrott, R. A., L. L. Eberhardt, P. J. White, and J. Rotella. 2003. Climate-induced variation in vital rates of an unharvested large-herbivore population. Canadian Journal of Zoology 81:33–45. Griffin, K. A., M. Hebblewhite, H. S. Robinson, P. Zager, S. M. Barber-Meyer, D. Christianson, S. Creel, N. C. Harris, M. A. Hurley, D. H. Jackson, B. K. Johnson, W. L. Myers, J. D. Raithel, M. Schlegel, B. L. Smith, C. White, and P. J. White. 2011. Neonatal mortality of elk driven by climate, predator phenology and predator community composition. Journal of Animal Ecology 80:1246–1257. Harris, N. C., M. J. Kauffman, and L. S. Mills. 2008. Inferences about ungulate population dynamics derived from age ratios. Journal of Wildlife Management 72:1143–1151. Johnson, D. E. 1951. Biology of the elk calf, Cervus canadensis nelsoni. Journal of Wildlife Management 15:396–410. de Kroon, H., A. Plaisier, J. van Groenendael, and H. Caswell. 1986. Elasticity: the relative contribution of demographic parameters to population growth rate. Ecology 67:1427–1431. Linnell, J. D. C., R. Aanes, and R. Andersen. 1995. Who killed Bambi? The role of predation in the neonatal mortality of temperate ungulates. Wildlife Biology 1:209–223. Lukacs, P. M., M. S. Mitchell, M. Hebblewhite, B. K. Johnson, H. Johnson, M. Kauffman, K. M. Proffitt, P. Zager, J. Brodie, K. Hersey, A. A. Holland, M. Hurley, S. McCorquodale, A. Middleton, M. Nordhagen, J. J. Nowak, D. P. Walsh, and P. J. White. 2018. Factors influencing elk recruitment across ecotypes in the western United States. Journal of Wildlife Management 82:698–710. Middleton, A. D., M. J. Kauffman, D. E. Mcwhirter, J. G. Cook, R. C. Cook, A. A. Nelson, M. D. Jimenez, and R. W. Klaver. 2013. Animal migration amid shifting patterns of phenology and predation: lessons from a Yellowstone elk herd. Ecology 94:1245–1256. 9 �Monteith, K. L., V. C. Bleich, T. R. Stephenson, B. M. Pierce, M. M. Conner, J. G. Kie, and R. T. Bowyer. 2014. Life-history characteristics of mule deer: effects of nutrition in a variable environment. Wildlife Monographs 186:1–62. Parker, K. L., P. S. Barboza, and M. P. Gillingham. 2009. Nutrition integrates environmental responses of ungulates. Functional Ecology 23:57–69. Proffitt, K. M., J. A. Cunningham, K. L. Hamlin, and R. A. Garrott. 2014. Bottom-up and topdown influences on pregnancy rates and recruitment of northern Yellowstone elk. Journal of Wildlife Management 78:1383–1393. Proffitt, K. M., M. Hebblewhite, W. Peters, N. Hupp, and J. Shamhart. 2016. Linking landscapescale differences in forage to ungulate nutritional ecology. Ecological Applications 26:2156– 2174. Raithel, J. D., M. J. Kauffman, and D. H. Pletscher. 2007. Impact of spatial and temporal variation in calf survival on the growth of elk population. Journal of Wildlife Management 71:795– 803. Singer, F. J., A. Harting, K. K. Symonds, and M. B. Coughenour. 1997. Density dependence, compensation, and environmental effects on elk calf mortality in Yellowstone National Park. Journal of Wildlife Management 61:12–25. Smith, B. L., and S. H. Anderson. 1996. Patterns of neonatal mortality of elk in northwest Wyoming. Canadian Journal of Zoology 74:1229–1237. Stonehouse, K. F., C. R. J. Anderson, M. E. Peterson, and D. R. Collins. 2016. Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus). Technical Publication Number 48, Colorado Parks and Wildlife, Fort Collins, CO, USA. Swift, L. W. 1945. A partial history of the elk herds of Colorado. Journal of Mammalogy 26:114– 119. Tatman, N. M., S. G. Liley, J. W. Cain, and J. W. Pitman. 2018. Effects of calf predation and nutrition on elk vital rates. Journal of Wildlife Management 82:1417–1428. Vavra, M., C. G. Parks, and M. J. Wisdom. 2007. Biodiversity, exotic plant species, and herbivory: the good, the bad, and the ungulate. Forest Ecology and Management 246:66–72. Vavra, M., and D. P. Sheehy. 1996. Improving elk habitat characteristics with livestock grazing. Rangelands 18:182–185. Visscher, D. R., and E. H. Merrill. 2009. Temporal dynamics of forage succession for elk at two scales: implications of forest management. Forest Ecology and Management 257:96–106. White, C. G., P. Zager, and M. W. Gratson. 2010. Influence of predator harvest, biological factors, and landscape on elk calf survival in Idaho. Journal of Wildlife Management 74:355–369. Prepared by Nathaniel D. Rayl, Wildlife Researcher 10 �Calves:100 adult females (2013-2017) Insufficient data 0 25-300 30-350 35-400 40-450 45-500 50-55 E-4 55-60 E-51 E-99 j E-24 E-33 0 30 60 Miles Figure 1. Number of elk calves per 100 adult females observed during December-February aerial surveys (5-year average from 2013-2017) within elk Data Analysis Units (DAUs; labeled with black text) in Colorado, USA. 11 �■ E-2 Bear's Ears 1.0 31 E-20 Uncompat1gre Plateau ■ E-33 Trinchera 31 32 43 42 4 5 46 30 0.9 0.8 _. C cu 0.7 C O> ~ 0.6 a. C .Q 0.5 t 0 g-0.4 ,_ 0... 0.3 0.2 0.1 0.0 J....__ 2_0,,_1-7..___ _ 20~1-8...____ _ 2_0...., 19 - -- 2020 2021 2022 Year Figure 2. Estimated average pregnancy rates of adult female elk from the Bear’s Ears, Trinchera, and Uncompahgre Plateau herds sampled during late winter 2017-2022 in Colorado, USA. The sample size is given at the top of the 95% binomial confidence intervals (black lines). 12 �13• • •• 12.?f2. 11 - ---.....cu .._ 10>"'O 0 ..c Q) Q) 9- • • L.. .._ .....coen 8Q) g> 76- 5- I • E-2 Bear's Ears E-20 Uncompahgre Plateau Herd Figure 3. The estimated ingesta-free body fat (%) of adult female elk from the Bear’s Ears (n = 46) and Uncompahgre Plateau (n = 46) herds during late-winter 2022 in Colorado, USA. 13 �Colorado Parks and Wildlife July 2022 – June 2023 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Colorado 3430 3002 1 : : : : Federal Aid Project No. W-242-R-7 : Parks and Wildlife Mammals Research Elk Conservation Evaluating factors influencing elk recruitment in Colorado Period Covered: July 1, 2022 – June 30, 2023 Authors: N.D. Rayl, M.W. Alldredge, and C.R. Anderson Jr. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. ABSTRACT Over the last two decades, wildlife managers in Colorado have become increasingly concerned about declining winter elk calf recruitment (estimated using juvenile:adult female ratios) in the southern portion of the state. Although juvenile:adult female ratios are often highly correlated with juvenile elk survival, they are an imperfect estimate of recruitment because they are affected by harvest, pregnancy rates, juvenile survival, and adult female survival. Thus, there is a need for elk research in Colorado based upon monitoring of marked individuals to evaluate factors affecting each stage of production and survival. Colorado Parks and Wildlife (CPW) is conducting research in 3 study areas to better determine how predators, habitat, and weather conditions are impacting elk recruitment in Colorado. From July 1, 2022 – June 30, 2023, we focused on working with stakeholders and collaborators on research logistics, and capturing and collaring elk. Field efforts were centered on 2 objectives: 1) capturing adult female elk, and collaring and outfitting pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior, and 2) capturing and collaring newborn and 6-month old elk to collect data on calf survival and cause-specific mortality. We radio-collared 46 6-month old elk calves in December 2022. In March 2023, we radio-collared 80 pregnant elk and outfitted them with VITs. Estimates of average pregnancy rates ranged from 95-98% across herds, and estimates of mean ingesta-free body fat ranged from 6.19-6.87% across herds. During the 2023 calving season, we radio-collared 97 elk calves. 1 �WILDLIFE RESEARCH REPORT EVALUATING FACTORS INFLUENCING ELK RECRUITMENT IN COLORADO NATHANIEL D. RAYL, MAT W. ALLDREDGE, AND CHUCK R. ANDERSON JR. PROJECT NARRATIVE OBJECTIVES The objectives of this project are to 1) estimate elk calf survival and cause-specific mortality rates from birth to age 1 to evaluate the importance of mortality sources for elk calf survival, 2) evaluate the influence of biotic (birth date, birth mass, gender, maternal body condition, habitat conditions) and abiotic factors (previous and current weather conditions) on seasonal mortality risk of elk calves from birth to age 1, 3) assess the health of elk herds by quantifying pregnancy rates and percent ingesta-free body fat (IFBF) of adult female elk, and 4) evaluate the influence of biotic (age, body condition, reproductive status, habitat conditions and selection) and abiotic factors (previous and current weather conditions) on pregnancy rates and IFBF. SEGMENT OBJECTIVES 1. Work with personnel from CPW Areas 6, 10, 11, and 18, and private landowners on field research logistics. 2. Capture adult female elk, and collar and outfit pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior. 3. Capture and collar neonate and 6-month old elk to collect data on calf survival and causespecific sources of mortality. INTRODUCTION In Colorado, elk (Cervus canadensis) are an important natural resource that are valued for ecological, consumptive, aesthetic, and economic reasons. In 1910, fewer than 1,000 elk remained in Colorado (Swift 1945), but today the state population is estimated to be the largest in the country, with over 292,000 elk. Over the last two decades, however, there has been increasing concern among wildlife managers in Colorado about declining winter calf recruitment (estimated using juvenile:adult female ratios) in the southern portion of the state (Fig. 1; Colorado Parks and Wildlife, unpublished data). In many elk populations, similar declining trends in juvenile recruitment have been observed. A recent synthesis examining elk recruitment in the western United States from 19892010 found evidence of a long-term reduction of 0.48 juveniles/100 adult females/year (Lukacs et al. 2018). Lukacs et al. (2018) discovered associations between recruitment and forage productivity that suggested nutritional conditions on either summer or winter ranges had the most influence on elk recruitment. These associations varied by geographic region: winter range conditions appeared to be more influential in southern areas (Colorado, Utah, and parts of Wyoming), whereas summer range conditions appeared to be more influential in northern areas 2 �(Idaho, Montana, Oregon, Washington, parts of Wyoming). Lukacs et al. (2018) also found that lower total winter precipitation in the previous winter was associated with lower recruitment the next year. The presence of wolves (Canis lupus) and grizzly bears (Ursus arctos) were also associated with lower recruitment. Although juvenile:adult female ratios are often highly correlated with juvenile elk survival (Raithel et al. 2007, Harris et al. 2008), they are an imperfect estimate of recruitment because they are affected by harvest, pregnancy rates, juvenile survival, and adult female survival (Caughley 1974, Gaillard et al. 2000, Harris et al. 2008, Decesare et al. 2012, Lukacs et al. 2018). This makes it difficult to identify ultimate factors influencing population dynamics using age ratio data alone, as multiple scenarios can produce equivalent ratios (Caughley 1974). Thus, long-term demographic studies based on monitoring of marked individuals are necessary to reliably test biological hypotheses and evaluate factors affecting each stage of production and survival (Gaillard et al. 2000, Clutton-Brock and Sheldon 2010, Proffitt et al. 2014). In the absence of harvest, the population dynamics of ungulates are normally characterized by high and stable adult female survival and variable juvenile survival (Gaillard et al. 1998, 2000). Among vital rates, changes in adult female survival have the potential to exert the most influence on population growth rates, but usually do not because of low year-to-year variability (Gaillard et al. 1998, 2000). Instead, adult female survival is typically buffered against moderate environmental variation, and thus not strongly influenced by climatic or densitydependent factors (Gaillard et al. 2000). Indeed, in a large-scale meta-analysis, Brodie et al. (2013) demonstrated that adult female elk survival was principally related to human harvest, and found that this harvest was an additive source of mortality. In contrast to adult female survival, and despite its lower elasticity (de Kroon et al. 1986), juvenile survival frequently has a greater effect on population growth rates of ungulates because of its high variability (Gaillard et al. 1998, 2000, Raithel et al. 2007). Juvenile survival of ungulates may be influenced by abiotic or biotic factors such as environmental conditions, forage quality or quantity, population density, maternal body condition, and predation (Barber-Meyer et al. 2008, Griffin et al. 2011, Monteith et al. 2014, Bastille-Rousseau et al. 2016). Complex interactions among these factors frequently make it difficult to identify the relative role of top-down and bottom-up factors affecting calf survival (Linnell et al. 1995). Further complexity may be introduced if top-down and bottom-up forces are simultaneously and variably influencing survival (Bowyer et al. 2005, Monteith et al. 2014). Juvenile survival has been found to vary substantially among and within elk populations on an annual basis (Garrott et al. 2003, Raithel et al. 2007, Griffin et al. 2011). Griffin et al. (2011) synthesized elk neonatal survival across 12 populations in the northwestern United States, and found that survival declined after warmer previous summers and with more predator species, and increased with higher May precipitation. In resource-limited populations, weather conditions may heavily influence juvenile survival. For example, in an unhunted elk population, prior to the establishment of wolves, Garrott et al. (2003) found that the dominant source of calf mortality in Yellowstone National Park was starvation, with more severe winters leading to lower calf survival. Elsewhere, mortality due to predation has been identified as the most significant cause of death for elk calves (e.g., Barber-Meyer et al. 2008). In some systems black bears (Ursus americanus) are the dominant predator (White et al. 2010, Tatman et al. 2018), whereas in others mountain lions (Puma concolor) kill the most calves (Eacker et al. 2016). Although nonpredation deaths (i.e., disease, starvation) may be low where high levels of juvenile predation occur, it remains difficult to determine whether neonate predation represents an additive or 3 �compensatory source of mortality (Linnell et al. 1995, Barber-Meyer et al. 2008, Griffin et al. 2011). Individual calf characteristics may also influence risk of mortality. To date, evidence for sex-biased mortality in elk calves is equivocal. Some studies have documented increased vulnerability of male calves to predation (Smith and Anderson 1996, Eacker et al. 2016), whereas others have demonstrated increased vulnerability of female calves (White et al. 2010). It has been suggested that differences may be due to the hunting behavior of the dominant predator, with stalking predators, such as mountain lions, disproportionately killing male calves, which may engage in riskier exploratory behavior (Eacker et al. 2016). Conclusions about the effect of birth mass on elk calf survival are similarly ambiguous. Some studies have reported that birth mass influenced survival probability (Singer et al. 1997, White et al. 2010), but others have not detected an influence of birth mass on neonate mortality (Smith and Anderson 1996, Eacker et al. 2016). The ability for nutritional resources on the landscape to support ungulate populations is reflected in the nutritional condition of individuals within those populations (Parker et al. 2009, Cook et al. 2013, Monteith et al. 2014). As nutritional resources decline there is typically a predictable sequence of changes in the vital rates of large herbivore populations: first, juvenile survival decreases, then age of first reproduction increases, followed by decreased fertility of adult females, and finally increased mortality rates of adults (Eberhardt 1977a, 1977b, 2002, Gaillard et al. 1998, 2000). This general sequence has been confirmed for elk, with forage quality and associated nutritional body condition demonstrated to affect reproductive failures (i.e., abortions or still births), yearling and adult pregnancy rates, birth dates, birth weight, calf growth, and winter survival of juveniles and adults (Cook et al. 2004a, 2004b, 2013, Proffitt et al. 2016). When nutritional resources are limited, elk calves may be lighter at birth, born later, and have slower growth rates during summer, which may predispose them to predation mortality. Cook et al. (2004a, 2013) demonstrated that summer-autumn nutrition can play a central role in determining elk productivity, as, during this time, animals must meet the demands of lactation and accrue sufficient fat reserves to get pregnant and survive the winter. For elk, nutritional resources on the landscape may be affected by wild and domestic herbivory (Vavra and Sheehy 1996, Vavra et al. 2007), timber management (Visscher and Merrill 2009), climatic conditions (Middleton et al. 2013), and fire history (Proffitt et al. 2016). To properly determine factors affecting juvenile ungulate survival bottom-up nutritional effects and top-down predation effects should be evaluated together (Monteith et al. 2014). Understanding how these effects are influencing elk population dynamics in Colorado is critical for guiding management actions. In 2016, Colorado Parks and Wildlife (CPW) initiated a 2-year pilot study to investigate factors influencing elk recruitment in 2 study areas in the state. During the pilot study, researchers collected data on annual elk calf survival, pregnancy rates, and latewinter body condition of adult female elk. In July 2018 – July 2019, we expanded this research into a 3rd study area to better determine how predators, habitat, and weather conditions are impacting elk recruitment in Colorado, and to provide management recommendations for increasing juvenile recruitment. STUDY AREAS For management purposes, the elk population in Colorado is divided into 43 Data Analysis Units (DAUs), each of which encompass the year-round range of an elk herd. This 4 �project is being conducted in 3 DAUs, 2 with low juvenile:adult female ratios (E-20, E-33), and 1 with high juvenile:adult female ratios (E-2), which will serve as a reference area (Fig. 1). The Uncompahgre Plateau elk herd (DAU E-20; 5,858 km2) is on the Uncompahgre Plateau in southwest Colorado, USA. From 2013-2017 juvenile:adult female ratios in E-20 averaged 30 calves per 100 adult females. Landownership is a mixture of BLM (38%), USFS (37%), private (24%), and state (1%) lands. Elevations range from 1,390 to 3,150 m. The plateau is characterized by a mixture of pinyon-juniper (Pinus edulis, Juniperus osteosperma) woodlands and sage-grassland communities (Artemisia spp., Cercocarpus montanus, Achnatherum hymenoides) at lower elevations. At mid elevations, ponderosa pine (Pinus ponderosa) and mountain shrub communities (Amelanchier alnifolia, Arctostaphylos, Artemisia spp., Quercus gambelii, Symphoricarpos spp.) predominate. Spruce-fir (Picea engelmannii, Abies lasiocarpa, Pseudotsuga menziesii) and aspen (Populus tremuloides) forests dominate at higher elevations. The Trinchera elk herd (DAU E-33; 8,601 km2) is in southeast Colorado, USA. From 2013-2017 juvenile:adult female ratios in E-33 averaged 26 calves per 100 adult females. Landownership is a mixture of private (89%), USFS (3%), state (3%), BLM (2%), U.S. Fish and Wildlife Service (USFWS; 1%), and other (2%) lands. Elevations range from 1,640 to 4,370 m. Lower elevations are characterized by agriculture and sage-grassland communities. At mid elevations, pinyon-juniper woodlands, ponderosa pine and mountain shrub forests, and spruce-fir and aspen forests predominate. Alpine tundra communities dominate at higher elevations. The Bear’s Ears elk herd (DAU E-2; 7,293 km2) is in northwest Colorado, USA. From 2013-2017 juvenile:adult female ratios in E-2 averaged 56 calves per 100 adult females. Landownership is a mixture of private (50%), USFS (25%), BLM (19%), state (5%) and other (1%) lands. Elevations range from 1,730 to 3,710 m. The DAU is characterized by sagegrassland communities at lower elevations. At mid elevations, mountain shrub communities predominate. Spruce-fir and aspen forests dominate at higher elevations. METHODS Capture and handling — We captured adult female elk ≥2 years of age from each study herd by helicopter net-gunning during late winter (March). During capture, we marked individuals with ear tags, collected a blood sample, and measured hind foot length, chest girth, and age based on tooth eruption and wear patterns (Keiss 1969). We used a portable ultrasound machine to assess whether or not captured elk were pregnant, and estimated the percent of ingesta-free body fat (IFBF) following methods detailed in Cook et al. (2010). We verified nonpregnancies using pregnancy-specific protein B (PSPB) analysis of sampled blood. From each study herd, we outfitted pregnant elk with vaginal implant transmitters (VITs) and Global Positioning System (GPS) radio-collars that attempt to acquire a location every 2 h. We deployed VITs that use the satellite communication capabilities of the collar on the adult female to send a notification when the VIT is expelled, signifying a birth. After receiving a birth notification from a VIT, we went to the birth site to capture and collar the newborn elk calf. We blindfolded calves, and wore latex gloves to minimize the transfer of human scent. We measured body mass, hind foot length, chest girth, and determined the gender of captured calves. We outfitted elk neonates with expandable GPS radio-collars or VHF proximity collars that communicated with the collar on the adult female, and are designed to drop off after 12 months. We also opportunistically located, captured, and collared additional neonates to increase sample sizes. We collected additional measurements (hair moisture, incisor 5 �and upper canine eruption, hoof, dew claw, and navel condition) from opportunistically captured calves to estimate age at capture following Johnson (1951) and Eacker (2015). In December, we captured 6-month old elk calves from each study herd by helicopter netgunning. During capture, we measured body mass, hind foot length, chest girth, and determined the gender of captured calves. We outfitted calves with expandable GPS radio-collars that were scheduled to drop off after 6 months. During all captures, we followed CPW’s animal care and use guidelines for capturing and handling elk (CPW ACUC #09-2008). Cause-specific mortality — Within 24 hours of detecting a mortality signal from an elk collar, we attempted to conduct a systematic field investigation to determine the cause of death. We searched the area surrounding kill sites for evidence of predator presence, including predator scats, tracks, and hair, or signs of a struggle (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). We examined elk carcasses for evidence of canine puncture wounds, subcutaneous hemorrhaging and bruising, aspirated blood in the mouth, nose, or trachea, claw or bite marks on the hide, cracked or chewed bones, and characteristic consumption patterns (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). We also collected calf carcasses when they were available to verify field assessments with laboratory necropsies performed by a CPW veterinarian. Nutritional condition of adult female elk — The body fat of lactating and non-lactating adult female elk can vary substantially, as lactating females are more sensitive than non-lactating females to their nutritional environment (Cook et al. 2004a, 2013). Therefore, it is difficult to interpret the body condition of adult female elk in late winter without knowing whether or not they experienced the energetic demands of lactation throughout the previous growing season (Cook et al. 2004a, 2013). RESULTS AND DISCUSSION In December 2022, we collared 46 6-month old elk calves, 21 from the Bear’s Ears and 25 from the Uncompahgre Plateau elk herd. The mean weight of calves from the Bear’s Ears herd was 105.1 kg (95% CI = 99.1-111.0 kg) and 111.3 kg (95% CI = 104.5-118.1 kg) from the Uncompahgre Plateau elk herd. During March 2023, we radio-collared 80 pregnant elk and outfitted them with VITs, 40 each from the Bear’s Ears and Uncompahgre Plateau herds. We estimated that pregnancy rates of adult female elk were 98% (95% CI = 87-100%) in the Bear’s Ears herd, and 95% (95% CI = 8599%) in the Uncompahgre Plateau herd (Fig. 2). Elk populations experiencing good to excellent summer-autumn nutrition typically have pregnancy rates ≥90% (Cook et al. 2013). We estimated the mean IFBF of adult female elk to be 6.19% from the Bear’s Ears herd and 6.87% from the Uncompahgre Plateau herd. When late-winter IFBF values are <8-9% for adult female elk that have lactated through the previous growing season, this suggests that there may be nutritional limitations, but it does not identify whether limitations are a result of summer-autumn or winter nutrition (R. Cook, personal communication). During May-July 2023, we captured and collared 97 elk calves, 43 from the Bear’s Ears herd, and 54 from the Uncompahgre Plateau herd. 6 �SUMMARY From July 1, 2022 – June 30, 2023 we successfully worked with private landowners and personnel from CPW to coordinate field research logistics and initiate the fifth year of this study. We collected data on body condition and reproduction by capturing adult female elk, and we outfitted 80 pregnant females with GPS collars and VITs. We successfully captured and collared 97 newborn elk and 46 6-month old elk calves, meeting our sample size objectives, and allowing us to collect data on calf survival and cause-specific sources of mortality. LITERATURE CITED Barber-Meyer, S. M., L. D. Mech, and P. J. White. 2008. Elk calf survival and mortality following wolf restoration to Yellowstone National Park. Wildlife Monographs 169:1–30. Bastille-Rousseau, G., J. A. Schaefer, K. P. Lewis, M. A. Mumma, E. H. Ellington, N. D. Rayl, S. P. Mahoney, D. Pouliot, and D. L. Murray. 2016. Phase-dependent climate-predator interactions explain three decades of variation in neonatal caribou survival. Journal of Animal Ecology 85:445–456. Bowyer, T., D. K. Person, and M. P. Becky. 2005. Detecting top-down versus bottom-up regulation of ungulates by large carnivores: implications for conservation of biodiversity. Pages 342–361 in J. C. Ray, K. H. Redford, R. S. Steneck, and J. Berger, editors. Large Carnivores and the Conservation of Biodiversity. Island Press, Washinton, D.C., USA. Brodie, J., H. Johnson, M. Mitchell, P. Zager, K. Proffitt, M. Hebblewhite, M. Kauffman, B. Johnson, J. Bissonette, C. Bishop, J. Gude, J. Herbert, K. Hersey, M. Hurley, P. M. Lukacs, S. Mccorquodale, E. Mcintire, J. Nowak, H. Sawyer, D. Smith, and P. J. White. 2013. Relative influence of human harvest, carnivores, and weather on adult female elk survival across western North America. Journal of Applied Ecology 50:295–305. Caughley, G. 1974. Interpretation of age ratios. Journal of Wildlife Management 38:557–562. Clutton-Brock, T., and B. C. Sheldon. 2010. Individuals and populations: the role of long-term, individual-based studies of animals in ecology and evolutionary biology. Trends in Ecology and Evolution 25:562–573. Cook, J. G., B. K. Johnson, R. C. Cook, R. A. Riggs, T. Delcurto, L. D. Bryant, and L. L. Irwin. 2004a. Effects of summer-autumn nutrition and parturition date on reproduction and survival of elk. Wildlife Monographs 155:1–61. Cook, R. C., J. G. Cook, and L. D. Mech. 2004b. Nutritional condition of northern Yellowstone elk. Journal of Mammalogy 85:714–722. Cook, R. C., J. G. Cook, T. R. Stephenson, W. L. Myers, S. M. Mccorquodale, D. J. Vales, L. L. Irwin, P. B. Hall, R. D. Spencer, S. L. Murphie, K. A. Schoenecker, and P. J. Miller. 2010. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife Management 74:880–896. Cook, R. C., J. G. Cook, D. J. Vales, B. K. Johnson, S. M. McCorquodale, L. A. Shipley, R. A. Riggs, L. L. Irwin, S. L. Murphie, B. L. Murphie, K. A. Schoenecker, F. Geyer, P. B. Hall, R. D. Spencer, D. A. Immell, D. H. Jackson, B. L. Tiller, P. J. Miller, and L. Schmitz. 2013. Regional and seasonal patterns of nutritional condition and reproduction in elk. Wildlife Monographs 184:1–44. 7 �Decesare, N. J., M. Hebblewhite, M. Bradley, K. G. Smith, D. Hervieux, and L. Neufeld. 2012. Estimating ungulate recruitment and growth rates using age ratios. Journal of Wildlife Management 76:144–153. de Kroon, H., A. Plaisier, J. van Groenendael, and H. Caswell. 1986. Elasticity: the relative contribution of demographic parameters to population growth rate. Ecology 67:1427–1431. Eacker, D. R. 2015. Linking the effects of risk factors on annual calf survival to elk population dynamics in the Bitterroot Valley, Montana. M.S. thesis, University of Montana, Missoula, MT, USA. Eacker, D. R., M. Hebblewhite, K. M. Proffitt, B. S. Jimenez, M. S. Mitchell, and H. S. Robinson. 2016. Annual elk calf survival in a multiple carnivore system. Journal of Wildlife Management 80:1345–1359. Eberhardt, L. L. 1977a. “Optimal” management policies for marine mammals. Wildlife Society Bulletin 5:162–169. Eberhardt, L. L. 1977b. Optimal policies for conservation of large mammals, with special references to marine ecosystems. Environmental Conservation 4:205–212. Eberhardt, L. L. 2002. A paradigm for population analysis of long-lived vertebrates. Ecology 83:2841–2854. Gaillard, J.-M., M. Festa-Bianchet, and N. G. Yoccoz. 1998. Population dynamics of large herbivores: variable recruitment with constant adult survival. Trends in Ecology and Evolution 13:58–63. Gaillard, J.-M., M. Festa-Bianchet, N. G. Yoccoz, A. Loison, and C. Toigo. 2000. Temporal variation in fitness components and population dynamics of large herbivores. Annual Review of Ecology, Evolution, and Systematics 31:367–393. Garrott, R. A., L. L. Eberhardt, P. J. White, and J. Rotella. 2003. Climate-induced variation in vital rates of an unharvested large-herbivore population. Canadian Journal of Zoology 81:33–45. Griffin, K. A., M. Hebblewhite, H. S. Robinson, P. Zager, S. M. Barber-Meyer, D. Christianson, S. Creel, N. C. Harris, M. A. Hurley, D. H. Jackson, B. K. Johnson, W. L. Myers, J. D. Raithel, M. Schlegel, B. L. Smith, C. White, and P. J. White. 2011. Neonatal mortality of elk driven by climate, predator phenology and predator community composition. Journal of Animal Ecology 80:1246–1257. Harris, N. C., M. J. Kauffman, and L. S. Mills. 2008. Inferences about ungulate population dynamics derived from age ratios. Journal of Wildlife Management 72:1143–1151. Johnson, D. E. 1951. Biology of the elk calf, Cervus canadensis nelsoni. Journal of Wildlife Management 15:396–410. Keiss, R. E. 1969. Comparison of eruption-wear patterns and cementum annuli as age criteria in elk. Journal of Wildlife Management 33:175-180. Linnell, J. D. C., R. Aanes, and R. Andersen. 1995. Who killed Bambi? The role of predation in the neonatal mortality of temperate ungulates. Wildlife Biology 1:209–223. Lukacs, P. M., M. S. Mitchell, M. Hebblewhite, B. K. Johnson, H. Johnson, M. Kauffman, K. M. Proffitt, P. Zager, J. Brodie, K. Hersey, A. A. Holland, M. Hurley, S. McCorquodale, A. Middleton, M. Nordhagen, J. J. Nowak, D. P. Walsh, and P. J. White. 2018. Factors influencing elk recruitment across ecotypes in the western United States. Journal of Wildlife Management 82:698–710. Middleton, A. D., M. J. Kauffman, D. E. Mcwhirter, J. G. Cook, R. C. Cook, A. A. Nelson, M. D. Jimenez, and R. W. Klaver. 2013. Animal migration amid shifting patterns of phenology and predation: lessons from a Yellowstone elk herd. Ecology 94:1245–1256. 8 �Monteith, K. L., V. C. Bleich, T. R. Stephenson, B. M. Pierce, M. M. Conner, J. G. Kie, and R. T. Bowyer. 2014. Life-history characteristics of mule deer: effects of nutrition in a variable environment. Wildlife Monographs 186:1–62. Parker, K. L., P. S. Barboza, and M. P. Gillingham. 2009. Nutrition integrates environmental responses of ungulates. Functional Ecology 23:57–69. Proffitt, K. M., J. A. Cunningham, K. L. Hamlin, and R. A. Garrott. 2014. Bottom-up and topdown influences on pregnancy rates and recruitment of northern Yellowstone elk. Journal of Wildlife Management 78:1383–1393. Proffitt, K. M., M. Hebblewhite, W. Peters, N. Hupp, and J. Shamhart. 2016. Linking landscapescale differences in forage to ungulate nutritional ecology. Ecological Applications 26:2156– 2174. Raithel, J. D., M. J. Kauffman, and D. H. Pletscher. 2007. Impact of spatial and temporal variation in calf survival on the growth of elk population. Journal of Wildlife Management 71:795– 803. Singer, F. J., A. Harting, K. K. Symonds, and M. B. Coughenour. 1997. Density dependence, compensation, and environmental effects on elk calf mortality in Yellowstone National Park. Journal of Wildlife Management 61:12–25. Smith, B. L., and S. H. Anderson. 1996. Patterns of neonatal mortality of elk in northwest Wyoming. Canadian Journal of Zoology 74:1229–1237. Stonehouse, K. F., C. R. J. Anderson, M. E. Peterson, and D. R. Collins. 2016. Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus). Technical Publication Number 48, Colorado Parks and Wildlife, Fort Collins, CO, USA. Swift, L. W. 1945. A partial history of the elk herds of Colorado. Journal of Mammalogy 26:114– 119. Tatman, N. M., S. G. Liley, J. W. Cain, and J. W. Pitman. 2018. Effects of calf predation and nutrition on elk vital rates. Journal of Wildlife Management 82:1417–1428. Vavra, M., C. G. Parks, and M. J. Wisdom. 2007. Biodiversity, exotic plant species, and herbivory: the good, the bad, and the ungulate. Forest Ecology and Management 246:66–72. Vavra, M., and D. P. Sheehy. 1996. Improving elk habitat characteristics with livestock grazing. Rangelands 18:182–185. Visscher, D. R., and E. H. Merrill. 2009. Temporal dynamics of forage succession for elk at two scales: implications of forest management. Forest Ecology and Management 257:96–106. White, C. G., P. Zager, and M. W. Gratson. 2010. Influence of predator harvest, biological factors, and landscape on elk calf survival in Idaho. Journal of Wildlife Management 74:355–369. Prepared by Nathaniel D. Rayl, Wildlife Researcher 9 �Calves:100 adult females (2013-2017) Insufficient data 0 25-300 30-350 35-400 40-450 45-500 50-55 E-4 55-60 E-51 E-99 j E-24 E-33 0 30 60 Miles Figure 1. Number of elk calves per 100 adult females observed during December-February aerial surveys (5-year average from 2013-2017) within elk Data Analysis Units (DAUs; labeled with black text) in Colorado, USA. 10 �■ E-2 Bear's Ears 27 1.0 0.9 31 31 32 30 31 E-20 Uncompat1gre Plateau ■ E-33 Trinchera 43 42 j 0.8 _. C cu 0.7 C 0) ~ 0.6 a. C .Q 0.5 t 0 g-0.4 ,_ 0.. 0.3 0.2 0.1 0.0.,____2_0.,,._1..... 7'----- 2--'0,18 - --2~0-19..._____ 2020 2021 2022 2023 Year Figure 2. Estimated average pregnancy rates of adult female elk from the Bear’s Ears, Trinchera, and Uncompahgre Plateau herds sampled during late winter 2017-2023 in Colorado, USA. The sample size is given at the top of the 95% binomial confidence intervals (black lines). 11 �Colorado Parks and Wildlife July 2023 – June 2024 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Colorado 3430 3002 1 : : : : Federal Aid Project No. W-242-R-8 : Parks and Wildlife Mammals Research Elk Conservation Evaluating factors influencing elk recruitment in Colorado Period Covered: July 1, 2023 – June 30, 2024 Authors: N.D. Rayl, M.W. Alldredge, and C.R. Anderson Jr. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. ABSTRACT Over the last two decades, wildlife managers in Colorado have become increasingly concerned about declining winter elk calf recruitment (estimated using juvenile:adult female ratios) in the southern portion of the state. Although juvenile:adult female ratios are often highly correlated with juvenile elk survival, they are an imperfect estimate of recruitment because they are affected by harvest, pregnancy rates, juvenile survival, and adult female survival. Thus, there is a need for elk research in Colorado based upon monitoring of marked individuals to evaluate factors affecting each stage of production and survival. Colorado Parks and Wildlife (CPW) is conducting research in 3 study areas to better determine how predators, habitat, and weather conditions are impacting elk recruitment in Colorado. From July 1, 2023 – June 30, 2024, we focused on working with stakeholders and collaborators on research logistics, and capturing and collaring elk. Field efforts were centered on 2 objectives: 1) capturing adult female elk, and collaring and outfitting pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior, and 2) capturing and collaring newborn and 6-month old elk to collect data on calf survival and cause-specific mortality. We radio-collared 50 6-month old elk calves in December 2023. In March 2024, we radio-collared 80 pregnant elk and outfitted them with VITs. Estimates of average pregnancy rates ranged from 87-98% across herds, and estimates of mean ingesta-free body fat ranged from 7.32-8.03% across herds. During the 2024 calving season, we radio-collared 102 elk calves. 1 �WILDLIFE RESEARCH REPORT EVALUATING FACTORS INFLUENCING ELK RECRUITMENT IN COLORADO NATHANIEL D. RAYL, MAT W. ALLDREDGE, AND CHUCK R. ANDERSON JR. PROJECT NARRATIVE OBJECTIVES The objectives of this project are to 1) estimate elk calf survival and cause-specific mortality rates from birth to age 1 to evaluate the importance of mortality sources for elk calf survival, 2) evaluate the influence of biotic (birth date, birth mass, gender, maternal body condition, habitat conditions) and abiotic factors (previous and current weather conditions) on seasonal mortality risk of elk calves from birth to age 1, 3) assess the health of elk herds by quantifying pregnancy rates and percent ingesta-free body fat (IFBF) of adult female elk, and 4) evaluate the influence of biotic and abiotic factors on pregnancy rates and IFBF. SEGMENT OBJECTIVES 1. Work with personnel from CPW Areas 6, 10, 11, and 18, and private landowners on field research logistics. 2. Capture adult female elk, and collar and outfit pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior. 3. Capture and collar neonate and 6-month old elk to collect data on calf survival and causespecific sources of mortality. INTRODUCTION In Colorado, elk (Cervus canadensis) are an important natural resource that are valued for ecological, consumptive, aesthetic, and economic reasons. In 1910, fewer than 1,000 elk remained in Colorado (Swift 1945), but today the state population is estimated to be the largest in the country, with over 292,000 elk. Over the last two decades, however, there has been increasing concern among wildlife managers in Colorado about declining winter calf recruitment (estimated using juvenile:adult female ratios) in the southern portion of the state (Fig. 1; Colorado Parks and Wildlife, unpublished data). In many elk populations, similar declining trends in juvenile recruitment have been observed. A recent synthesis examining elk recruitment in the western United States from 19892010 found evidence of a long-term reduction of 0.48 juveniles/100 adult females/year (Lukacs et al. 2018). Lukacs et al. (2018) discovered associations between recruitment and forage productivity that suggested nutritional conditions on either summer or winter ranges had the most influence on elk recruitment. These associations varied by geographic region: winter range conditions appeared to be more influential in southern areas (Colorado, Utah, and parts of Wyoming), whereas summer range conditions appeared to be more influential in northern areas (Idaho, Montana, Oregon, Washington, parts of Wyoming). Lukacs et al. (2018) also found that 2 �lower total winter precipitation in the previous winter was associated with lower recruitment the next year. The presence of wolves (Canis lupus) and grizzly bears (Ursus arctos) were also associated with lower recruitment in northern areas. Although juvenile:adult female ratios are often highly correlated with juvenile elk survival (Raithel et al. 2007, Harris et al. 2008), they are an imperfect estimate of recruitment because they are affected by harvest, pregnancy rates, juvenile survival, and adult female survival (Caughley 1974, Gaillard et al. 2000, Harris et al. 2008, Decesare et al. 2012, Lukacs et al. 2018). This makes it difficult to identify ultimate factors influencing population dynamics using age ratio data alone, as multiple scenarios can produce equivalent ratios (Caughley 1974). Thus, long-term demographic studies based on monitoring of marked individuals are necessary to reliably test biological hypotheses and evaluate factors affecting each stage of production and survival (Gaillard et al. 2000, Clutton-Brock and Sheldon 2010, Proffitt et al. 2014). In the absence of harvest, the population dynamics of ungulates are normally characterized by high and stable adult female survival and variable juvenile survival (Gaillard et al. 1998, 2000). Among vital rates, changes in adult female survival have the potential to exert the most influence on population growth rates, but usually do not because of low year-to-year variability (Gaillard et al. 1998, 2000). Instead, adult female survival is typically buffered against moderate environmental variation, and thus not strongly influenced by climatic or densitydependent factors (Gaillard et al. 2000). Indeed, in a large-scale meta-analysis, Brodie et al. (2013) demonstrated that adult female elk survival was principally related to human harvest, and found that this harvest was an additive source of mortality. In contrast to adult female survival, and despite its lower elasticity (de Kroon et al. 1986), juvenile survival frequently has a greater effect on population growth rates of ungulates because of its high variability (Gaillard et al. 1998, 2000, Raithel et al. 2007). Juvenile survival of ungulates may be influenced by abiotic or biotic factors such as environmental conditions, forage quality or quantity, population density, maternal body condition, and predation (Barber-Meyer et al. 2008, Griffin et al. 2011, Monteith et al. 2014, Bastille-Rousseau et al. 2016). Complex interactions among these factors frequently make it difficult to identify the relative role of top-down and bottom-up factors affecting calf survival (Linnell et al. 1995). Further complexity may be introduced if top-down and bottom-up forces are simultaneously and variably influencing survival (Bowyer et al. 2005, Monteith et al. 2014). Juvenile survival has been found to vary substantially among and within elk populations on an annual basis (Garrott et al. 2003, Raithel et al. 2007, Griffin et al. 2011). Griffin et al. (2011) synthesized elk neonatal survival across 12 populations in the northwestern United States, and found that survival declined after warmer previous summers and with more predator species, and increased with higher May precipitation. In resource-limited populations, weather conditions may heavily influence juvenile survival. For example, in an unhunted elk population, prior to the establishment of wolves, Garrott et al. (2003) found that the dominant source of calf mortality in Yellowstone National Park was starvation, with more severe winters leading to lower calf survival. Elsewhere, mortality due to predation has been identified as the most significant cause of death for elk calves (e.g., Barber-Meyer et al. 2008). In some systems black bears (Ursus americanus) are the dominant predator (White et al. 2010, Tatman et al. 2018), whereas in others mountain lions (Puma concolor) kill the most calves (Eacker et al. 2016). Although nonpredation deaths (i.e., disease, starvation) may be low where high levels of juvenile predation occur, it remains difficult to determine whether neonate predation represents an additive or 3 �compensatory source of mortality (Linnell et al. 1995, Barber-Meyer et al. 2008, Griffin et al. 2011, Monteith et al. 2014). Individual calf characteristics may also influence risk of mortality. To date, evidence for sex-biased mortality in elk calves is equivocal. Some studies have documented increased vulnerability of male calves to predation (Smith and Anderson 1996, Eacker et al. 2016), whereas others have demonstrated increased vulnerability of female calves (White et al. 2010). It has been suggested that differences may be due to the hunting behavior of the dominant predator, with stalking predators, such as mountain lions, disproportionately killing male calves, which may engage in riskier exploratory behavior (Eacker et al. 2016). Conclusions about the effect of birth mass on elk calf survival are similarly ambiguous. Some studies have reported that birth mass influenced survival probability (Singer et al. 1997, White et al. 2010), but others have not detected an influence of birth mass on neonate mortality (Smith and Anderson 1996, Eacker et al. 2016). The ability for nutritional resources on the landscape to support ungulate populations is reflected in the nutritional condition of individuals within those populations (Parker et al. 2009, Cook et al. 2013, Monteith et al. 2014). As nutritional resources decline there is typically a predictable sequence of changes in the vital rates of large herbivore populations: first, juvenile survival decreases, then age of first reproduction increases, followed by decreased fertility of adult females, and finally increased mortality rates of adults (Eberhardt 1977a, 1977b, 2002, Gaillard et al. 1998, 2000). This general sequence has been confirmed for elk, with forage quality and associated nutritional body condition demonstrated to affect intra-uterine survival, yearling and adult pregnancy rates, birth dates, birth weight, calf growth, and winter survival of juveniles and adults (Cook et al. 2004a, 2004b, 2013, Proffitt et al. 2016). When nutritional resources are limited, elk calves may be lighter at birth, born later, and have slower growth rates during summer, which may predispose them to predation mortality. Cook et al. (2004a, 2013) demonstrated that summer-autumn nutrition can play a central role in determining elk productivity, as, during this time, animals must meet the demands of lactation and accrue sufficient fat reserves to get pregnant and survive the winter. For elk, nutritional resources on the landscape may be affected by wild and domestic herbivory (Vavra and Sheehy 1996, Vavra et al. 2007), timber management (Visscher and Merrill 2009), climatic conditions (Middleton et al. 2013), and fire history (Proffitt et al. 2016). To properly determine factors affecting juvenile ungulate survival bottom-up nutritional effects and top-down predation effects should be evaluated together (Monteith et al. 2014). Understanding how these effects are influencing elk population dynamics in Colorado is critical for guiding management actions. In 2016, Colorado Parks and Wildlife (CPW) initiated a 2-year pilot study to investigate factors influencing elk recruitment in 2 study areas in the state. During the pilot study, researchers collected data on annual elk calf survival, pregnancy rates, and latewinter body condition of adult female elk. In July 2018 – July 2019, we expanded this research into a 3rd study area to better determine how predators, habitat, and weather conditions are impacting elk recruitment in Colorado, and to provide management recommendations for increasing juvenile recruitment. STUDY AREAS For management purposes, the elk population in Colorado is divided into 43 Data Analysis Units (DAUs), each of which encompass the year-round range of an elk herd. This 4 �project is being conducted in 3 DAUs, 2 with low juvenile:adult female ratios (E-20, E-33), and 1 with high juvenile:adult female ratios (E-2), which will serve as a reference area (Fig. 1). The Uncompahgre Plateau elk herd (DAU E-20; 5,858 km2) is on the Uncompahgre Plateau in southwest Colorado, USA. From 2013-2017 juvenile:adult female ratios in E-20 averaged 30 calves per 100 adult females. Landownership is a mixture of BLM (38%), USFS (37%), private (24%), and state (1%) lands. Elevations range from 1,390 to 3,150 m. The plateau is characterized by a mixture of pinyon-juniper (Pinus edulis, Juniperus osteosperma) woodlands and sage-grassland communities (Artemisia spp., Cercocarpus montanus, Achnatherum hymenoides) at lower elevations. At mid elevations, ponderosa pine (Pinus ponderosa) and mountain shrub communities (Amelanchier alnifolia, Arctostaphylos, Artemisia spp., Quercus gambelii, Symphoricarpos spp.) predominate. Spruce-fir (Picea engelmannii, Abies lasiocarpa, Pseudotsuga menziesii) and aspen (Populus tremuloides) forests dominate at higher elevations. The Trinchera elk herd (DAU E-33; 8,601 km2) is in southeast Colorado, USA. From 2013-2017 juvenile:adult female ratios in E-33 averaged 26 calves per 100 adult females. Landownership is a mixture of private (89%), USFS (3%), state (3%), BLM (2%), U.S. Fish and Wildlife Service (USFWS; 1%), and other (2%) lands. Elevations range from 1,640 to 4,370 m. Lower elevations are characterized by agriculture and sage-grassland communities. At mid elevations, pinyon-juniper woodlands, ponderosa pine and mountain shrub forests, and spruce-fir and aspen forests predominate. Alpine tundra communities dominate at higher elevations. The Bear’s Ears elk herd (DAU E-2; 7,293 km2) is in northwest Colorado, USA. From 2013-2017 juvenile:adult female ratios in E-2 averaged 56 calves per 100 adult females. Landownership is a mixture of private (50%), USFS (25%), BLM (19%), state (5%) and other (1%) lands. Elevations range from 1,730 to 3,710 m. The DAU is characterized by sagegrassland communities at lower elevations. At mid elevations, mountain shrub communities predominate. Spruce-fir and aspen forests dominate at higher elevations. METHODS Capture and handling — We captured adult female elk ≥2 years of age from each study herd by helicopter net-gunning during late winter (March). During capture, we marked individuals with ear tags, collected a blood sample, and measured hind foot length, chest girth, and age based on tooth eruption and wear patterns. We used a portable ultrasound machine to assess whether or not captured elk were pregnant, and estimated the percent of ingesta-free body fat (IFBF) following methods detailed in Cook et al. (2010). We verified non-pregnancies using pregnancy-specific protein B (PSPB) analysis of sampled blood. From each study herd, we outfitted pregnant elk with vaginal implant transmitters (VITs) and Global Positioning System (GPS) radio-collars that attempt to acquire a location every 2 h. We deployed VITs that use the satellite communication capabilities of the collar on the adult female to send a notification when the VIT is expelled, signifying a birth. After receiving a birth notification from a VIT, we went to the birth site to capture and collar the newborn elk calf. We blindfolded calves, and wore latex gloves to minimize the transfer of human scent. We measured body mass, hind foot length, chest girth, and determined the gender of captured calves. We outfitted elk neonates with expandable GPS radio-collars or VHF proximity collars that communicated with the collar on the adult female, and are designed to drop off after 12 months. We also opportunistically located, captured, and collared additional neonates to increase sample sizes. We collected additional measurements (hair moisture, incisor 5 �and upper canine eruption, hoof, dew claw, and navel condition) from opportunistically captured calves to estimate age at capture following Johnson (1951) and Eacker (2015). In December, we captured 6-month old elk calves from each study herd by helicopter netgunning. During capture, we measured body mass, hind foot length, chest girth, and determined the gender of captured calves. We outfitted calves with expandable GPS radio-collars that were scheduled to drop off after 6 months. During all captures, we followed CPW’s animal care and use guidelines for capturing and handling elk (CPW ACUC #09-2008). Cause-specific mortality — Within 24 hours of detecting a mortality signal from an elk collar, we attempted to conduct a systematic field investigation to determine the cause of death. We searched the area surrounding kill sites for evidence of predator presence, including predator scats, tracks, and hair, or signs of a struggle (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). We examined elk carcasses for evidence of canine puncture wounds, subcutaneous hemorrhaging and bruising, aspirated blood in the mouth, nose, or trachea, claw or bite marks on the hide, cracked or chewed bones, and characteristic consumption patterns (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). We also collected calf carcasses when they were available to verify field assessments with laboratory necropsies performed by a CPW veterinarian. Nutritional condition of adult female elk — The body fat of lactating and non-lactating adult female elk can vary substantially, as lactating females are more sensitive than non-lactating females to their nutritional environment (Cook et al. 2004a, 2013). Therefore, it is difficult to interpret the body condition of adult female elk in late winter without knowing whether or not they experienced the energetic demands of lactation throughout the previous growing season (Cook et al. 2004a, 2013). RESULTS AND DISCUSSION In December 2023, we collared 50 6-month old elk calves, 25 each from the Bear’s Ears and Uncompahgre Plateau elk herds. The mean weight of calves from the Bear’s Ears herd was 100.4 kg (95% CI = 94.3-106.5 kg) and 103.2 kg (95% CI = 96.8-109.5 kg) from the Uncompahgre Plateau elk herd. During March 2023, we radio-collared 80 pregnant elk and outfitted them with VITs, 40 each from the Bear’s Ears and Uncompahgre Plateau herds. We estimated that pregnancy rates of adult female elk were 87% (95% CI = 74-94%) in the Bear’s Ears herd, and 98% (95% CI = 88100%) in the Uncompahgre Plateau herd (Fig. 2). Elk populations experiencing good to excellent summer-autumn nutrition typically have pregnancy rates ≥90% (Cook et al. 2013). We estimated the mean IFBF of adult female elk to be 7.32% from the Bear’s Ears herd and 8.03% from the Uncompahgre Plateau herd. When late-winter IFBF values are <8-9% for adult female elk that have lactated through the previous growing season, this suggests that there may be nutritional limitations, but it does not identify whether limitations are a result of summer-autumn or winter nutrition (R. Cook, personal communication). During May-July 2024, we captured and collared 102 elk calves, 50 from the Bear’s Ears herd, and 52 from the Uncompahgre Plateau herd. 6 �SUMMARY From July 1, 2023 – June 30, 2024 we successfully worked with private landowners and personnel from CPW to coordinate field research logistics and initiate the sixth year of this study. We collected data on body condition and reproduction by capturing adult female elk, and we outfitted 80 pregnant females with GPS collars and VITs. We successfully captured and collared 102 newborn elk and 50 6-month old elk calves, meeting our sample size objectives, and allowing us to collect data on calf survival and cause-specific sources of mortality. LITERATURE CITED Barber-Meyer, S. M., L. D. Mech, and P. J. White. 2008. Elk calf survival and mortality following wolf restoration to Yellowstone National Park. Wildlife Monographs 169:1–30. Bastille-Rousseau, G., J. A. Schaefer, K. P. Lewis, M. A. Mumma, E. H. Ellington, N. D. Rayl, S. P. Mahoney, D. Pouliot, and D. L. Murray. 2016. Phase-dependent climate-predator interactions explain three decades of variation in neonatal caribou survival. Journal of Animal Ecology 85:445–456. Bowyer, T., D. K. Person, and M. P. Becky. 2005. Detecting top-down versus bottom-up regulation of ungulates by large carnivores: implications for conservation of biodiversity. Pages 342–361 in J. C. Ray, K. H. Redford, R. S. Steneck, and J. Berger, editors. Large Carnivores and the Conservation of Biodiversity. Island Press, Washinton, D.C., USA. Brodie, J., H. Johnson, M. Mitchell, P. Zager, K. Proffitt, M. Hebblewhite, M. Kauffman, B. Johnson, J. Bissonette, C. Bishop, J. Gude, J. Herbert, K. Hersey, M. Hurley, P. M. Lukacs, S. Mccorquodale, E. Mcintire, J. Nowak, H. Sawyer, D. Smith, and P. J. White. 2013. Relative influence of human harvest, carnivores, and weather on adult female elk survival across western North America. Journal of Applied Ecology 50:295–305. Caughley, G. 1974. Interpretation of age ratios. Journal of Wildlife Management 38:557–562. Clutton-Brock, T., and B. C. Sheldon. 2010. Individuals and populations: the role of long-term, individual-based studies of animals in ecology and evolutionary biology. Trends in Ecology and Evolution 25:562–573. Cook, J. G., B. K. Johnson, R. C. Cook, R. A. Riggs, T. Delcurto, L. D. Bryant, and L. L. Irwin. 2004a. Effects of summer-autumn nutrition and parturition date on reproduction and survival of elk. Wildlife Monographs 155:1–61. Cook, R. C., J. G. Cook, and L. D. Mech. 2004b. Nutritional condition of northern Yellowstone elk. Journal of Mammalogy 85:714–722. Cook, R. C., J. G. Cook, T. R. Stephenson, W. L. Myers, S. M. Mccorquodale, D. J. Vales, L. L. Irwin, P. B. Hall, R. D. Spencer, S. L. Murphie, K. A. Schoenecker, and P. J. Miller. 2010. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife Management 74:880–896. Cook, R. C., J. G. Cook, D. J. Vales, B. K. Johnson, S. M. McCorquodale, L. A. Shipley, R. A. Riggs, L. L. Irwin, S. L. Murphie, B. L. Murphie, K. A. Schoenecker, F. Geyer, P. B. Hall, R. D. Spencer, D. A. Immell, D. H. Jackson, B. L. Tiller, P. J. Miller, and L. Schmitz. 2013. Regional and seasonal patterns of nutritional condition and reproduction in elk. Wildlife Monographs 184:1–44. 7 �Decesare, N. J., M. Hebblewhite, M. Bradley, K. G. Smith, D. Hervieux, and L. Neufeld. 2012. Estimating ungulate recruitment and growth rates using age ratios. Journal of Wildlife Management 76:144–153. Eacker, D. R. 2015. Linking the effects of risk factors on annual calf survival to elk population dynamics in the Bitterroot Valley, Montana. M.S. thesis, University of Montana, Missoula, MT, USA. Eacker, D. R., M. Hebblewhite, K. M. Proffitt, B. S. Jimenez, M. S. Mitchell, and H. S. Robinson. 2016. Annual elk calf survival in a multiple carnivore system. Journal of Wildlife Management 80:1345–1359. Eberhardt, L. L. 1977a. “Optimal” management policies for marine mammals. Wildlife Society Bulletin 5:162–169. Eberhardt, L. L. 1977b. Optimal policies for conservation of large mammals, with special references to marine ecosystems. Environmental Conservation 4:205–212. Eberhardt, L. L. 2002. A paradigm for population analysis of long-lived vertebrates. Ecology 83:2841–2854. Gaillard, J.-M., M. Festa-Bianchet, and N. G. Yoccoz. 1998. Population dynamics of large herbivores: variable recruitment with constant adult survival. Trends in Ecology and Evolution 13:58–63. Gaillard, J.-M., M. Festa-Bianchet, N. G. Yoccoz, A. Loison, and C. Toigo. 2000. Temporal variation in fitness components and population dynamics of large herbivores. Annual Review of Ecology, Evolution, and Systematics 31:367–393. Garrott, R. A., L. L. Eberhardt, P. J. White, and J. Rotella. 2003. Climate-induced variation in vital rates of an unharvested large-herbivore population. Canadian Journal of Zoology 81:33–45. Griffin, K. A., M. Hebblewhite, H. S. Robinson, P. Zager, S. M. Barber-Meyer, D. Christianson, S. Creel, N. C. Harris, M. A. Hurley, D. H. Jackson, B. K. Johnson, W. L. Myers, J. D. Raithel, M. Schlegel, B. L. Smith, C. White, and P. J. White. 2011. Neonatal mortality of elk driven by climate, predator phenology and predator community composition. Journal of Animal Ecology 80:1246–1257. Harris, N. C., M. J. Kauffman, and L. S. Mills. 2008. Inferences about ungulate population dynamics derived from age ratios. Journal of Wildlife Management 72:1143–1151. Johnson, D. E. 1951. Biology of the elk calf, Cervus canadensis nelsoni. Journal of Wildlife Management 15:396–410. de Kroon, H., A. Plaisier, J. van Groenendael, and H. Caswell. 1986. Elasticity: the relative contribution of demographic parameters to population growth rate. Ecology 67:1427–1431. Linnell, J. D. C., R. Aanes, and R. Andersen. 1995. Who killed Bambi? The role of predation in the neonatal mortality of temperate ungulates. Wildlife Biology 1:209–223. Lukacs, P. M., M. S. Mitchell, M. Hebblewhite, B. K. Johnson, H. Johnson, M. Kauffman, K. M. Proffitt, P. Zager, J. Brodie, K. Hersey, A. A. Holland, M. Hurley, S. McCorquodale, A. Middleton, M. Nordhagen, J. J. Nowak, D. P. Walsh, and P. J. White. 2018. Factors influencing elk recruitment across ecotypes in the western United States. Journal of Wildlife Management 82:698–710. Middleton, A. D., M. J. Kauffman, D. E. Mcwhirter, J. G. Cook, R. C. Cook, A. A. Nelson, M. D. Jimenez, and R. W. Klaver. 2013. Animal migration amid shifting patterns of phenology and predation: lessons from a Yellowstone elk herd. Ecology 94:1245–1256. 8 �Monteith, K. L., V. C. Bleich, T. R. Stephenson, B. M. Pierce, M. M. Conner, J. G. Kie, and R. T. Bowyer. 2014. Life-history characteristics of mule deer: effects of nutrition in a variable environment. Wildlife Monographs 186:1–62. Parker, K. L., P. S. Barboza, and M. P. Gillingham. 2009. Nutrition integrates environmental responses of ungulates. Functional Ecology 23:57–69. Proffitt, K. M., J. A. Cunningham, K. L. Hamlin, and R. A. Garrott. 2014. Bottom-up and topdown influences on pregnancy rates and recruitment of northern Yellowstone elk. Journal of Wildlife Management 78:1383–1393. Proffitt, K. M., M. Hebblewhite, W. Peters, N. Hupp, and J. Shamhart. 2016. Linking landscapescale differences in forage to ungulate nutritional ecology. Ecological Applications 26:2156– 2174. Raithel, J. D., M. J. Kauffman, and D. H. Pletscher. 2007. Impact of spatial and temporal variation in calf survival on the growth of elk population. Journal of Wildlife Management 71:795– 803. Singer, F. J., A. Harting, K. K. Symonds, and M. B. Coughenour. 1997. Density dependence, compensation, and environmental effects on elk calf mortality in Yellowstone National Park. Journal of Wildlife Management 61:12–25. Smith, B. L., and S. H. Anderson. 1996. Patterns of neonatal mortality of elk in northwest Wyoming. Canadian Journal of Zoology 74:1229–1237. Stonehouse, K. F., C. R. J. Anderson, M. E. Peterson, and D. R. Collins. 2016. Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus). Technical Publication Number 48, Colorado Parks and Wildlife, Fort Collins, CO, USA. Swift, L. W. 1945. A partial history of the elk herds of Colorado. Journal of Mammalogy 26:114– 119. Tatman, N. M., S. G. Liley, J. W. Cain, and J. W. Pitman. 2018. Effects of calf predation and nutrition on elk vital rates. Journal of Wildlife Management 82:1417–1428. Vavra, M., C. G. Parks, and M. J. Wisdom. 2007. Biodiversity, exotic plant species, and herbivory: the good, the bad, and the ungulate. Forest Ecology and Management 246:66–72. Vavra, M., and D. P. Sheehy. 1996. Improving elk habitat characteristics with livestock grazing. Rangelands 18:182–185. Visscher, D. R., and E. H. Merrill. 2009. Temporal dynamics of forage succession for elk at two scales: implications of forest management. Forest Ecology and Management 257:96–106. White, C. G., P. Zager, and M. W. Gratson. 2010. Influence of predator harvest, biological factors, and landscape on elk calf survival in Idaho. Journal of Wildlife Management 74:355–369. Prepared by Nathaniel D. Rayl, Wildlife Researcher 9 �Figure 1. Number of elk calves per 100 adult females observed during December-February aerial surveys (5-year average from 2013-2017) within elk Data Analysis Units (DAUs; labeled with black text) in Colorado, USA. 10 �Figure 2. Estimated average pregnancy rates of adult female elk from the Bear’s Ears, Trinchera, and Uncompahgre Plateau herds sampled during late winter 2017-2024 in Colorado, USA. The sample size is given at the top of the 95% binomial confidence intervals (black lines). 11 � https://cpw.cvlcollections.org/files/original/99b97322203e8f53f30b744bcbef4e39.pdf 0366640cc05defa506b6f588d22e209e PDF Text Text Colorado Parks and Wildlife July 2019 – June 2020 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Colorado 3430 3002 2 : : : : Federal Aid Project No. W-242-R-4 : Parks and Wildlife Mammals Research Elk Conservation Response of Elk to Human Recreation at Multiple Scales: demographic shifts and behaviorally mediated fluctuations in local abundance Period Covered: July 1, 2019 - June 30, 2020 Authors: E.J. Bergman and N.D. Rayl Personnel: R. Baker, T. Brtis, M. Fisher, L. Gepfert, J. Groves, A. Hart, K. Hayes, W. Hiler, T. Kishimoto, D. Lewis, J. Mao, A. McLaine, K. Middledorf, A. Orlando, K. Russo, K. Tesch, L. Wolfe, and M. Yamashita, CPW; J. Clark, H. Cushman, J. Larrivee, A. Orlando, S. Strike, R. Swisher, S. Swisher, and T. Triple, Quicksilver Air, Inc.. Project support received from Federal Aid in Wildlife Restoration, Great Outdoors Colorado, Pitkin County Open Space and Trails, and Rocky Mountain Elk Foundation. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. ABSTRACT During the reporting period we focused on working with stakeholders and collaborators on research logistics, deploying field cameras, and capturing and collaring elk. Field efforts were centered on 3 objectives: 1) deploying/retrieving field cameras (i.e., cameras capable of timelapse and motion activation to estimate elk abundance), 2) capturing adult female elk, and collaring and outfitting pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior, and 3) capturing and collaring newborn and 6-month old elk calves to collect data on calf survival and cause-specific mortality. We retrieved 119 cameras (deployed during fall 2019) and we redeployed 238 cameras (spring 2020) across 8 study units, and radio-collared 40 pregnant elk and outfitted them with VITs. Estimated the pregnancy rate was 95%, and the mean ingesta-free body fat of adult females was 8.1%. We radio-collared 25 6-month old elk calves in December 2019. During the 2020 calving 1 �season, we radio-collared 54 elk calves, including 93% of the calves born to collared females. We estimated that the average date of calving was 3 June. Cameras deployed during spring 2020 will be retrieved from the field during the fall of 2020 and photo processing will occur during the winter of 2020–2021. 2 �WILDLIFE RESEARCH REPORT RESPONSE OF ELK TO HUMAN RECREATION AT MULTIPOLE SCALES: DEMOGRAPHIC SHIFTS AND BEHAVIORALLY MEDIATED FLUCTUATIONS IN LOCAL ABUNDANCE ERIC J. BERGMAN AND NATHANIEL D. RAYL PROJECT NARRATIVE OBJECTIVES This project has objectives on 2 scales. At the broad, elk herd-level scale, we will estimate pregnancy rates, calf survival rates, and cause-specific mortality rates to evaluate the importance of mortality sources for elk calf survival. More specifically, we will evaluate the influence of biotic (birth date, birth mass, gender, maternal body condition, habitat conditions), abiotic (previous and current weather conditions), and human-induced factors (i.e., relative exposure to recreational activities) on seasonal mortality risk of elk calves from birth to age 1 and on pregnancy rates of mature female elk. At the narrower geographic and temporal scale, we will use short-term (~3-4 weeks) changes in elk abundance within small study units (<65 km2) as a tool to evaluate the influence of human recreation on elk distribution. At this narrower scale, the primary objective is to evaluate the role that human recreation (e.g., hiking, mountain biking, horseback riding, trail running, hunting, etc.) has on the behavioral distribution of elk on spring calving, summer, and fall transition ranges. Coupled to the objective of detecting behaviorally influenced changes in abundance and density, we will evaluate the effectiveness of current recreational closures maintained by ski areas, counties, and federal land management agencies. SEGMENT OBJECTIVES 1. Work with personnel from CPW Areas 8 and 10, and private landowners on field research logistics. 2. Deploy field cameras (i.e., cameras capable of time-lapse) to estimate elk abundance. 3. Capture adult female elk, and collar and outfit pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior. 4. Capture and collar newborn and 6-month old elk to collect data on calf survival and cause-specific sources of mortality. INTRODUCTION The role of outdoor recreation within the state of Colorado is difficult to overstate. According to Colorado's Statewide Comprehensive Outdoor Recreation Plan (SCORP), outdoor recreation contributes 511,000 jobs, $62.5 billion in economic output, and $9.4 billion in local, state, and federal tax revenue (State of Colorado 2019). Outdoor recreation includes multiple activities such as biking, camping, climbing, fishing, hiking, horseback riding, hunting, shooting, skiing and wildlife watching. The Outdoor Foundation estimates that in the United States, there 3 �are nearly 30 million hikers, just under 7 million mountain bikers, over 14 million hunters, and nearly 23 million wildlife watchers (Outdoor Foundation 2018). While difficult to quantify, it is a reasonable assumption that many individual outdoor enthusiasts actively participate in more than one of these activities. Thus, the economies of Colorado, its counties, and its communities, rely on managing the landscape for a multitude of outdoor recreational opportunities. However, there is also evidence that human activities have an impact on wildlife. While trail-based recreation has the potential to impact many species, recent concerns in Colorado have focused on elk (Cervus canadensis; Durango Herald 2018, Steamboat Pilot and Today 2018, Vail Daily 2018). The sensitivity of elk to human presence and human activity has been a topic of interest for many decades. Preliminary research, focused on the effects of logging and vehicle use along road networks, provided consistent and clear evidence that elk-use declined in areas with high road densities, and as road use increased (Lyon and Christensen 2002). Similarly, research in Colorado evaluating the repeated displacement and disturbance of elk by people on foot provided evidence of suppressed recruitment rates following human disturbance (Phillips and Alldredge 2000, Shively et al. 2005). Experimental evaluation of the impact of hunter presence on elk movements and elk distribution has also occurred in Colorado (Conner et al. 2001, Vieira et al. 2003). This research demonstrated that the presence of hunters shifted elk off public lands and onto neighboring private lands. More recently, recreational trail use (all-terrain vehicle (ATV) riding, hiking, biking, and horseback riding) impacts on elk use of areas with trails was experimentally evaluated in Oregon. Wisdom et al. (2018) found that elk avoided areas with trails when recreationists of any type were present. Thus, regardless of human activity, behavioral displacement of elk by humans is well documented. In Colorado, increasing public concerns over human recreational use have coincided with declines in elk productivity, but a direct relationship to this activity in Colorado remains unaddressed. During FY 2016–2017, Colorado Parks and Wildlife (CPW) initiated a large-scale pilot study designed to evaluate pregnancy rates, elk calf survival, and causes of elk calf mortality (Alldredge 2016). At the onset, it was recognized that many factors contribute to suppression of pregnancy rates and calf survival. In addition to hunting, deteriorating habitat quality, habitat loss, and predation are key factors that may influence Colorado’s ungulate herds. Likewise, factors such as disease and competition may also play a role. Less clear, however, are the effects that human recreation may exert on the population dynamics of elk and other large ungulates. Past research has also reported individual behavioral responses of elk exposed to recreational stimuli. However, an alternate approach to studying behavioral displacement would shift the focus away from individual animals and link elk distribution to specific geographic areas. One limitation to studying individual animals is that the presence or absence of unmarked animals within the study area is largely ignored. However, access management and land management planning decisions are intrinsically tied to geographic areas. Thus, knowledge about the presence, absence, and abundance of a species of interest is of great value to managers. 4 �STUDY AREAS This study is occurring in two study areas. The northern study area focuses on the Bear’s Ears elk herd between Craig and Steamboat Springs. Within the Bear’s Ears herd, the fine scale camera-based behavioral portion of this study is centered on the Routt County segment of the herd that uses the Elk River drainage near the community of Steamboat Springs. The Bear’s Ears study area will be sampled using 4 study units: Mad Creek, Buffalo Pass, Walton Rim, and Hwy 40/Ferndale. The northernmost Mad Creek study unit has few existing trails but has been identified as a potential site for future trail development. Immediately south of the Mad Creek study unit is the Buffalo Pass study unit. Extensive trail development in this study area occurred during the past 5-10 years and it is currently an important and key area for many trail-based recreational activities. Further south is the Walton Rim study unit. Bounded to the north by the Steamboat Springs ski area, Walton Rim currently has little or no recreational trail use and plans for future trail development in this unit currently do not exist. Finally, immediately south of Walton Rim falls the Hwy 40/Ferndale study unit. Currently the Hwy 40/Ferndale study unit has nominal trail development and use, but plans for future trail construction in this unit are being considered. With the exception of Walton Rim, all of the camera-based study units in the Bear’s Ears study area have the potential to experience extensive trail development and use. The southern study area is focused on the Avalanche Creek elk herd along the Roaring Fork River between Glenwood Springs and Aspen. Four camera-based study units in the Avalanche Creek study area have also been identified. The southernmost of these units is the Snowmass unit. This unit, managed by Pitkin County and White River National Forest, has existing trails but is managed with seasonal closures (low elevation trail closures in place until May 16th, and high elevation trail closures in place until June 21st) to protect elk wintering and calving areas. Near the Snowmass study unit (and also in the southern portion of this study area) is the Wildcat study unit. The Wildcat study unit is centered on private property and has nominal recreational trail use, allowing it to serve as a reference area. Further north and nearer the community of Carbondale are 2 additional study units. The eastern most of these additional 2 units is The Crown, which is managed by the Bureau of Land Management and has extensive recreational use, but also has winter closures for mechanized and motorized recreation. Immediately to the south and west of The Crown is the fourth study unit. This unit, Two Shoes Ranch, is privately managed and has little recreational use and minimal trail development. METHODS Camera Sampling — Recent development of non-invasive abundance estimation techniques provide opportunities to quantify species in finite areas over relatively short periods. Camera based Space-To-Event (STE) and Instantaneous Sampling (IS) methods provide tools to estimate abundance without expensive flight time (Moeller et al. 2018). An inherent property of these new techniques is that the scope of inference applies to geographic areas and not individual animals. We deployed remote field cameras (HP2X, Reconyx, Holmen, Wisconsin, USA) to estimate elk abundance and density within small geographic areas (<65 km2) and during short time frames (~3–4 weeks). We deployed cameras across 8 study units (4 within the Bear’s Ears study area and 4 within the Avalanche Creek study area). We designed grids composed of 1.6 km cells to overlay each study unit. Within each cell, we used generalized random-tessellation stratification (GRTS) sampling (Stevens and Olsen 2004, Kincaid and Olsen 2017) to select 2 5 �coarse camera locations. We selected final camera site locations (<250 m of the randomly selected coarse locations) in the field with the specific objective of maximizing detection probability of elk. We deployed cameras during the spring and early summer seasons. We programmed the cameras to take pictures at 10-minute intervals throughout the day. Elk capture and handling — We captured adult female elk ≥2 years of age by helicopter netgunning during late winter (March). During capture, we marked individuals with ear tags, collected a blood sample, and measured hind foot length, chest girth, and age based on tooth eruption and wear patterns. We used a portable ultrasound machine to assess whether or not captured elk were pregnant, and estimated the percent of ingesta-free body fat (IFBF) following methods detailed in Cook et al. (2010). We verified non-pregnancies using pregnancy-specific protein B (PSPB) analysis of sampled blood. We outfitted pregnant elk with vaginal implant transmitters (VITs) and Global Positioning System (GPS) radio-collars that attempt to acquire a location every 2 h. We deployed VITs that use the satellite communication capabilities of the collar on the adult female to send a notification when the VIT is expelled, signifying a birth. After receiving a birth notification from a VIT, we went to the birth site to capture and collar the newborn elk calf. We blindfolded calves, and wore latex gloves to minimize the transfer of human scent. We measured body mass, hind foot length, chest girth, and determined the gender of captured calves. We outfitted elk neonates with expandable GPS radio-collars or VHF proximity collars that communicated with the collar on the adult female, and are designed to drop off after 12 months. We also opportunistically located, captured, and collared additional neonates to increase sample sizes. We collected additional measurements (hair moisture, incisor and upper canine eruption, hoof, dew claw, and navel condition) from opportunistically captured calves to estimate age at capture following Johnson (1951) and Eacker (2015). We attempted to handle calves for <5 minutes to minimize stress. In December, we captured 6-month old elk calves by helicopter net-gunning. During capture, we measured body mass, hind foot length, chest girth, and determined the gender of captured calves. We outfitted calves with expandable GPS radio-collars that were scheduled to drop off after 6 months. During all captures, we followed CPW’s animal care and use guidelines for capturing and handling elk (CPW ACUC #09-2008). Cause-specific mortality — Within 24 hours of detecting a mortality signal from an elk collar, we attempted to conduct a systematic field investigation to determine the cause of death. We searched the area surrounding kill sites for evidence of predator presence, including predator scats, tracks, and hair, or signs of a struggle (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). We examined elk calf carcasses for evidence of canine puncture wounds, subcutaneous hemorrhaging and bruising, aspirated blood in the mouth, nose, or trachea, claw or bite marks on the hide, cracked or chewed bones, and characteristic consumption patterns (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). We also collected calf carcasses when they were available to verify field assessments with laboratory necropsies performed by a CPW veterinarian. Nutritional condition of adult female elk — The body fat of lactating and non-lactating adult female elk can vary substantially, as lactating females are more sensitive than non-lactating females to their nutritional environment (Cook et al. 2004, 2013). Therefore, it is difficult to interpret the body condition of adult female elk in late winter without knowing whether or not 6 �they experienced the energetic demands of lactation throughout the previous growing season (Cook et al. 2004, 2013). We will use the late-winter body condition of prime-aged adult female elk that successfully raised a calf the previous year to assess whether or not our study herds may be experiencing nutritional limitations. RESULTS AND DISCUSSION During March 2020, we radio-collared 40 pregnant elk from the Avalanche Creek elk herd and outfitted them with VITs. Estimated the pregnancy rate was 95% (95% CI = 85-99%; n = 43). Elk populations experiencing good to excellent summer-autumn nutrition typically have pregnancy rates ≥90% (Cook et al. 2013). We estimated the mean IFBF of adult female elk to be 8.1%. When late-winter IFBF values are <8-9% for adult female elk that have lactated through the previous growing season, this suggests that there may be nutritional limitations, but it does not identify whether limitations are a result of summer-autumn or winter nutrition (R. Cook, personal communication). In December 2019, we collared 25 6-month old elk calves from the Avalanche Creek elk herd. The mean weight of 6-month old calves was 115.8 kg (95% CI = 110.8-120.8 kg). During May-July 2020, we captured and collared 54 elk calves from the Avalanche Creek herd. We successfully captured and collared 93% (37/40) of the calves of adult female elk outfitted with VITs. The estimated mean date of calving was 3 June for the Avalanche Creek herd. During the summer of 2019, a total of 384,455 photos were taken by the 118 cameras deployed across 8 study units. Automated photo recognition software is being developed and applied to these photos to expedite future analyses. SUMMARY During FY2019-20 we successfully worked with private landowners and personnel from CPW to coordinate field research logistics and initiate the second year of this study. We collected data on body condition and reproduction by capturing adult female elk, and we outfitted 40 pregnant females with GPS collars and VITs. We successfully captured and collared 54 newborn elk and 25 6-month old elk calves, meeting our sample size objectives, and allowing us to collect data on calf survival and cause-specific sources of mortality. We will continue to collect data on elk survival and cause-specific sources of mortality throughout the year. Field cameras were also successfully deployed. LITERATURE CITED Alldredge, M. 2016. Pilot study – elk recruitment and habitat use in Colorado. Program Narrative Study Plan, Colorado Parks and Wildlife, Fort Collins, CO, USA. Barber-Meyer, S. M., L. D. Mech, and P. J. White. 2008. Elk calf survival and mortality following wolf restoration to Yellowstone National Park. Wildlife Monographs 169:1–30. Conner, M.M., G.C. White, and D.J. Freddy. 2001. Elk movement in response to early-season hunting in northwest Colorado. Journal of Wildlife Management 65:926–940. Cook, J. G., B. K. Johnson, R. C. Cook, R. A. Riggs, T. Delcurto, L. D. Bryant, and L. L. Irwin. 2004. Effects of summer-autumn nutrition and parturition date on reproduction and survival of elk. Wildlife Monographs 155:1–61. 7 �Cook, R. C., J. G. Cook, T. R. Stephenson, W. L. Myers, S. M. Mccorquodale, D. J. Vales, L. L. Irwin, P. B. Hall, R. D. Spencer, S. L. Murphie, K. A. Schoenecker, and P. J. Miller. 2010. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife Management 74:880–896. Cook, R. C., J. G. Cook, D. J. Vales, B. K. Johnson, S. M. McCorquodale, L. A. Shipley, R. A. Riggs, L. L. Irwin, S. L. Murphie, B. L. Murphie, K. A. Schoenecker, F. Geyer, P. B. Hall, R. D. Spencer, D. A. Immell, D. H. Jackson, B. L. Tiller, P. J. Miller, and L. Schmitz. 2013. Regional and seasonal patterns of nutritional condition and reproduction in elk. Wildlife Monographs 184:1–44. Durango Herald. 2018. Where have all the elk gone? Published 15 November 2018, accessed 16 November 2018 (https://durangoherald.com/articles/250613-where-have-all-the-elkgone). Eacker, D. R. 2015. Linking the effects of risk factors on annual calf survival to elk population dynamics in the Bitterroot Valley, Montana. M.S. thesis, University of Montana, Missoula, MT, USA. Eacker, D. R., M. Hebblewhite, K. M. Proffitt, B. S. Jimenez, M. S. Mitchell, and H. S. Robinson. 2016. Annual elk calf survival in a multiple carnivore system. Journal of Wildlife Management 80:1345–1359. Johnson, D. E. 1951. Biology of the elk calf, Cervus canadensis nelsoni. Journal of Wildlife Management 15:396–410. Kincaid, T.M., and A.R. Olsen. 2017. Spsurvey: spatial survey design and analysis. R package version 3.4. https://CRAN.R-roject.org/package=spsurvey. Lyon, L.J., and A.G. Christensen. 2002. Elk and land management in D.E. Toweill and J.W. Thomas, eds., North American Elk: ecology and management. Smithsonian Institute Press, Washington D.C., USA. Moeller, A.K., P.M. Lukacs, and J.S. Horne. 2018. Three novel methods to estimate abundance of unmarked animals using remote cameras. Ecosphere 9:e02331. Phillips, G.E. and A.W. Alldredge. 2000. Reproductive success of elk following disturbance by humans during calving season. Journal of Wildlife Management 64:521–530. Shively, K.J., A.W. Alldredge, and G.E. Phillips. 2005. Elk reproductive response to removal of calving season disturbance by humans. Journal of Wildlife Management 69:1073–1080. State of Colorado, Colorado Statewide Comprehensive Outdoor Recreation Plan. 2019. https://cpw.state.co.us/Documents/Trails/SCORP/Final-Plan/2019-SCORP-Report.pdf Accessed 23 January 2019. Steamboat Pilot and Today. 2018. Newly formed group advocates to slow trail building in Routt National Forest to protect wildlife. Published 21 October 2018, accessed 16 November 2018 (https://www.steamboatpilot.com/news/newly-formed-group-advocates-toslow-trail-building-in-routt-national-forest-to-protect-wildlife/). Stevens, D.L., and A.R. Olsen. 2004. Spatially balanced sampling of natural resources. Journal of the American Statistical Association 99:262–278. Stonehouse, K. F., C. R. J. Anderson, M. E. Peterson, and D. R. Collins. 2016. Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus). Technical Publication Number 48, Colorado Parks and Wildlife, Fort Collins, CO, USA. Outdoor Foundation. 2018. Outdoor Participation Report. https://outdoorindustry.org/resource/2018-outdoor-participation-report/. Accessed 26 September 2018. 8 �Vail Daily. 2018. Eagle County officials concerned by wildlife population declines. Published 14 September 2018, accessed 16 November 2018 (https://www.vaildaily.com/news/eaglecounty-officials-concerned-by-wildlife-population-declines/). Vieira, M.E.P., M.M. Conner, G.C. White, and D.J. Freddy. 2003. Effects of archery hunter numbers and opening dates on elk movement. Journal of Wildlife Management 67:717– 728. Wisdom, M.J., H.K. Preisler, L.M. Naylor, R.G. Anthony, B.K. Johnson, and M.M. Rowland, 2018. Elk responses to trail-based recreation on public forests. Forest Ecology and Management 411:223–233. Prepared by Eric J. Bergman, Wildlife Researcher 9 �Colorado Parks and Wildlife July 2020 – June 2021 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Colorado 3430 3002 2 : : : : Federal Aid Project No. W-242-R-5 : Parks and Wildlife Mammals Research Elk Conservation Response of Elk to Human Recreation at Multiple Scales: demographic shifts and behaviorally mediated fluctuations in local abundance Period Covered: July 1, 2020 – June 30, 2021 Authors: E.J. Bergman and N.D. Rayl Personnel: R. Baker, Z. Durbin, R. Ebel-Childs, M. Fisher, L. Gepfert, J. Groves, W. Hiler, D. Lewis, J. Mao, K. Middledorf, S. Mollett, E. Monfort, P. Nol, A. Orlando, S. Sandefur, E. Sawa, N. Starling, K. Tesch, and M. Yamashita, CPW; J. Clark, H. Cushman, B. Dooling, T. Herby, A. Orlando, R. Swisher, S. Swisher, and T. Triple, Quicksilver Air, Inc.. Project support received from Federal Aid in Wildlife Restoration, Rocky Mountain Elk Foundation, and CPW Big Game Auction and Raffle. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. ABSTRACT During the reporting period we focused on working with stakeholders and collaborators on research logistics, deploying field cameras, and capturing and collaring elk. Field efforts were centered on 3 objectives: 1) deploying/retrieving field cameras (i.e., cameras capable of timelapse and motion activation to estimate elk abundance), 2) capturing adult female elk, and collaring and outfitting pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior, and 3) capturing and collaring newborn and 6-month old elk calves to collect data on calf survival and cause-specific mortality. During spring (FY 19-20) and early summer (FY 20-21) of 2020 we deployed 238 cameras across 8 study units. During fall, 154 of these cameras were retrieved. Retrieval of 84 cameras was prevented by closures due to the Middle Fork Fire near Steamboat Springs, and the 1 �subsequent onset of winter. The remaining cameras were retrieved during spring of 2021. Downloading and cataloging of photos collected during the summer of 2020 occurred during summer 2021, with approximately 4.6 million photos collected. We radio-collared 25 6-month old elk calves in December 2020. In March 2021, we radio-collared 40 pregnant elk and outfitted them with VITs. We estimated the pregnancy rate was 85% and the mean ingesta-free body fat of adult females was 8.2%. During the 2021 calving season, we radio-collared 51 elk calves. We estimated that the average date of calving was 2 June. WILDLIFE RESEARCH REPORT RESPONSE OF ELK TO HUMAN RECREATION AT MULTIPOLE SCALES: DEMOGRAPHIC SHIFTS AND BEHAVIORALLY MEDIATED FLUCTUATIONS IN LOCAL ABUNDANCE ERIC J. BERGMAN AND NATHANIEL D. RAYL PROJECT NARRATIVE OBJECTIVES This project has objectives on 2 scales. At the broad, elk herd-level scale, we will estimate pregnancy rates, calf survival rates, and cause-specific mortality rates to evaluate the importance of mortality sources for elk calf survival. More specifically, we will evaluate the influence of biotic (birth date, birth mass, gender, maternal body condition, habitat conditions), abiotic (previous and current weather conditions), and human-induced factors (i.e., relative exposure to recreational activities) on seasonal mortality risk of elk calves from birth to age 1 and on pregnancy rates of mature female elk. At the narrower geographic and temporal scale, we will use short-term (~3-4 weeks) changes in elk abundance within small study units (<65 km2) as a tool to evaluate the influence of human recreation on elk distribution. At this narrower scale, the primary objective is to evaluate the role that human recreation (e.g., hiking, mountain biking, horseback riding, trail running, hunting, etc.) has on the behavioral distribution of elk on spring calving, summer, and fall transition ranges. Coupled to the objective of detecting behaviorally influenced changes in abundance and density, we will evaluate the effectiveness of current recreational closures maintained by ski areas, counties, and federal land management agencies. SEGMENT OBJECTIVES 1. Work with personnel from CPW Areas 8 and 10, and private landowners on field research logistics. 2. Deploy field cameras (i.e., cameras capable of time-lapse) to estimate elk abundance. 3. Capture adult female elk, and collar and outfit pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior. 4. Capture and collar newborn and 6-month old elk to collect data on calf survival and cause-specific sources of mortality. 2 �INTRODUCTION The role of outdoor recreation within the state of Colorado is difficult to overstate. According to Colorado's Statewide Comprehensive Outdoor Recreation Plan (SCORP), outdoor recreation contributes 511,000 jobs, $62.5 billion in economic output, and $9.4 billion in local, state, and federal tax revenue (State of Colorado 2019). Outdoor recreation includes multiple activities such as biking, camping, climbing, fishing, hiking, horseback riding, hunting, shooting, skiing and wildlife watching. The Outdoor Foundation estimates that in the United States, there are nearly 30 million hikers, just under 7 million mountain bikers, over 14 million hunters, and nearly 23 million wildlife watchers (Outdoor Foundation 2018). While difficult to quantify, it is a reasonable assumption that many individual outdoor enthusiasts actively participate in more than one of these activities. Thus, the economies of Colorado, its counties, and its communities, rely on managing the landscape for a multitude of outdoor recreational opportunities. However, there is also evidence that human activities have an impact on wildlife. While trail-based recreation has the potential to impact many species, recent concerns in Colorado have focused on elk (Cervus canadensis; Durango Herald 2018, Steamboat Pilot and Today 2018, Vail Daily 2018). The sensitivity of elk to human presence and human activity has been a topic of interest for many decades. Preliminary research, focused on the effects of logging and vehicle use along road networks, provided consistent and clear evidence that elk-use declined in areas with high road densities, and as road use increased (Lyon and Christensen 2002). Similarly, research in Colorado evaluating the repeated displacement and disturbance of elk by people on foot provided evidence of suppressed recruitment rates following human disturbance (Phillips and Alldredge 2000, Shively et al. 2005). Experimental evaluation of the impact of hunter presence on elk movements and elk distribution has also occurred in Colorado (Conner et al. 2001, Vieira et al. 2003). This research demonstrated that the presence of hunters shifted elk off public lands and onto neighboring private lands. More recently, recreational trail use (all-terrain vehicle (ATV) riding, hiking, biking, and horseback riding) impacts on elk use of areas with trails was experimentally evaluated in Oregon. Wisdom et al. (2018) found that elk avoided areas with trails when recreationists of any type were present. Thus, regardless of human activity, behavioral displacement of elk by humans is well documented. In Colorado, increasing public concerns over human recreational use have coincided with declines in elk productivity, but a direct relationship to this activity in Colorado remains unaddressed. During FY 2016–2017, Colorado Parks and Wildlife (CPW) initiated a large-scale pilot study designed to evaluate pregnancy rates, elk calf survival, and causes of elk calf mortality (Alldredge 2016). At the onset, it was recognized that many factors contribute to suppression of pregnancy rates and calf survival. In addition to hunting, deteriorating habitat quality, habitat loss, and predation are key factors that may influence Colorado’s ungulate herds. Likewise, factors such as disease and competition may also play a role. Less clear, however, are the effects that human recreation may exert on the population dynamics of elk and other large ungulates. Past research has also reported individual behavioral responses of elk exposed to recreational stimuli. However, an alternate approach to studying behavioral displacement would shift the focus away from individual animals and link elk distribution to specific geographic areas. One limitation to studying individual animals is that the presence or absence of unmarked animals within the study area is largely ignored. However, access management and land 3 �management planning decisions are intrinsically tied to geographic areas. Thus, knowledge about the presence, absence, and abundance of a species of interest is of great value to managers. STUDY AREAS This study is occurring in two study areas. The northern study area focuses on the Bear’s Ears elk herd between Craig and Steamboat Springs. Within the Bear’s Ears herd, the fine scale camera-based behavioral portion of this study is centered on the Routt County segment of the herd that uses the Elk River drainage near the community of Steamboat Springs. The Bear’s Ears study area will be sampled using 4 study units: Mad Creek, Buffalo Pass, Walton Rim, and Hwy 40/Ferndale. The northernmost Mad Creek study unit has few existing trails but has been identified as a potential site for future trail development. Immediately south of the Mad Creek study unit is the Buffalo Pass study unit. Extensive trail development in this study area occurred during the past 5-10 years and it is currently an important and key area for many trail-based recreational activities. Further south is the Walton Rim study unit. Bounded to the north by the Steamboat Springs ski area, Walton Rim currently has little or no recreational trail use and plans for future trail development in this unit currently do not exist. Finally, immediately south of Walton Rim falls the Hwy 40/Ferndale study unit. Currently the Hwy 40/Ferndale study unit has nominal trail development and use, but plans for future trail construction in this unit are being considered. With the exception of Walton Rim, all of the camera-based study units in the Bear’s Ears study area have the potential to experience extensive trail development and use. The southern study area is focused on the Avalanche Creek elk herd along the Roaring Fork River between Glenwood Springs and Aspen. Four camera-based study units in the Avalanche Creek study area have also been identified. The southernmost of these units is the Snowmass unit. This unit, managed by Pitkin County and White River National Forest, has existing trails but is managed with seasonal closures (low elevation trail closures in place until May 16th, and high elevation trail closures in place until June 21st) to protect elk wintering and calving areas. Near the Snowmass study unit (and also in the southern portion of this study area) is the Wildcat study unit. The Wildcat study unit is centered on private property and has nominal recreational trail use, allowing it to serve as a reference area. Further north and nearer the community of Carbondale are 2 additional study units. The eastern most of these additional 2 units is The Crown, which is managed by the Bureau of Land Management and has extensive recreational use, but also has winter closures for mechanized and motorized recreation. Immediately to the south and west of The Crown is the fourth study unit. This unit, Two Shoes Ranch, is privately managed and has little recreational use and minimal trail development. METHODS Camera Sampling — Recent development of non-invasive abundance estimation techniques provide opportunities to quantify species in finite areas over relatively short periods. Camera based Space-To-Event (STE) and Instantaneous Sampling (IS) methods provide tools to estimate abundance without expensive flight time (Moeller et al. 2018). An inherent property of these new techniques is that the scope of inference applies to geographic areas and not individual animals. We deployed remote field cameras (HP2X, Reconyx, Holmen, Wisconsin, USA) to estimate elk abundance and density within small geographic areas (<65 km2) and during short 4 �time frames (~3–4 weeks). We deployed cameras across 8 study units (4 within the Bear’s Ears study area and 4 within the Avalanche Creek study area). We designed grids composed of 1.6 km cells to overlay each study unit. Within each cell, we used generalized random-tessellation stratification (GRTS) sampling (Stevens and Olsen 2004, Kincaid and Olsen 2017) to select 2 coarse camera locations. We selected final camera site locations (<250 m of the randomly selected coarse locations) in the field with the specific objective of maximizing detection probability of elk. We deployed cameras during the spring and early summer seasons. We programmed the cameras to take pictures at 10-minute intervals throughout the day. Elk capture and handling — We captured adult female elk ≥2 years of age by helicopter netgunning during late winter (March). During capture, we marked individuals with ear tags, collected a blood sample, and measured hind foot length, chest girth, and age based on tooth eruption and wear patterns. We used a portable ultrasound machine to assess whether or not captured elk were pregnant, and estimated the percent of ingesta-free body fat (IFBF) following methods detailed in Cook et al. (2010). We verified non-pregnancies using pregnancy-specific protein B (PSPB) analysis of sampled blood. We outfitted pregnant elk with vaginal implant transmitters (VITs) and Global Positioning System (GPS) radio-collars that attempt to acquire a location every 2 h. We deployed VITs that use the satellite communication capabilities of the collar on the adult female to send a notification when the VIT is expelled, signifying a birth. In December, we captured 6-month old elk calves by helicopter net-gunning. During capture, we measured body mass, hind foot length, chest girth, and determined the gender of captured calves. We outfitted calves with expandable GPS radio-collars that were scheduled to drop off after 6 months. During all captures, we followed CPW’s animal care and use guidelines for capturing and handling elk (CPW ACUC #09-2008). Cause-specific mortality — Within 24 hours of detecting a mortality signal from an elk collar, we attempted to conduct a systematic field investigation to determine the cause of death. We searched the area surrounding kill sites for evidence of predator presence, including predator scats, tracks, and hair, or signs of a struggle (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). We examined elk carcasses for evidence of canine puncture wounds, subcutaneous hemorrhaging and bruising, aspirated blood in the mouth, nose, or trachea, claw or bite marks on the hide, cracked or chewed bones, and characteristic consumption patterns (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). Nutritional condition of adult female elk — The body fat of lactating and non-lactating adult female elk can vary substantially, as lactating females are more sensitive than non-lactating females to their nutritional environment (Cook et al. 2004a, 2013). Therefore, it is difficult to interpret the body condition of adult female elk in late winter without knowing whether or not they experienced the energetic demands of lactation throughout the previous growing season (Cook et al. 2004a, 2013). RESULTS AND DISCUSSION In December 2020, we collared 25 6-month old elk calves from the Avalanche Creek herd. The mean weight of calves was 109.0 kg (95% CI: 103.3-114.8 kg). During March 2021, we radio-collared 40 pregnant elk and outfitted them with VITs. We estimated the pregnancy 5 �rate of adult female elk was 85% (72-93%). Elk populations experiencing good to excellent summer-autumn nutrition typically have pregnancy rates ≥90% (Cook et al. 2013). We estimated the mean IFBF of adult female elk to be 8.2%. When late-winter IFBF values are <8-9% for adult female elk that have lactated through the previous growing season, this suggests that there may be nutritional limitations, but it does not identify whether limitations are a result of summerautumn or winter nutrition (R. Cook, personal communication). During the summer of 2020, approximately 4.6 million photos were taken by the 238 cameras deployed across 8 study units. These photos are actively being archived. Automated photo recognition software continues to be developed and will be applied to these photos to expedite future analyses. However, subsampling of photos (and longer time intervals) will occur prior to analysis. SUMMARY During July 2020 – June 2021 we successfully worked with private landowners and personnel from CPW to coordinate field research logistics and initiate the third year of this study. We collected data on body condition and reproduction by capturing adult female elk, and we outfitted 40 pregnant females with GPS collars and VITs. We successfully captured and collared 51 newborn elk and 25 6-month old elk calves, meeting our sample size objectives, and allowing us to collect data on calf survival and cause-specific sources of mortality. We will continue to collect data on elk survival and cause-specific sources of mortality throughout the year. Field cameras were also successfully deployed. LITERATURE CITED Alldredge, M. 2016. Pilot study – elk recruitment and habitat use in Colorado. Program Narrative Study Plan, Colorado Parks and Wildlife, Fort Collins, CO, USA. Barber-Meyer, S. M., L. D. Mech, and P. J. White. 2008. Elk calf survival and mortality following wolf restoration to Yellowstone National Park. Wildlife Monographs 169:1–30. Conner, M.M., G.C. White, and D.J. Freddy. 2001. Elk movement in response to early-season hunting in northwest Colorado. Journal of Wildlife Management 65:926–940. Cook, J. G., B. K. Johnson, R. C. Cook, R. A. Riggs, T. Delcurto, L. D. Bryant, and L. L. Irwin. 2004. Effects of summer-autumn nutrition and parturition date on reproduction and survival of elk. Wildlife Monographs 155:1–61. Cook, R. C., J. G. Cook, T. R. Stephenson, W. L. Myers, S. M. Mccorquodale, D. J. Vales, L. L. Irwin, P. B. Hall, R. D. Spencer, S. L. Murphie, K. A. Schoenecker, and P. J. Miller. 2010. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife Management 74:880–896. Cook, R. C., J. G. Cook, D. J. Vales, B. K. Johnson, S. M. McCorquodale, L. A. Shipley, R. A. Riggs, L. L. Irwin, S. L. Murphie, B. L. Murphie, K. A. Schoenecker, F. Geyer, P. B. Hall, R. D. Spencer, D. A. Immell, D. H. Jackson, B. L. Tiller, P. J. Miller, and L. Schmitz. 2013. Regional and seasonal patterns of nutritional condition and reproduction in elk. Wildlife Monographs 184:1–44. Durango Herald. 2018. Where have all the elk gone? Published 15 November 2018, accessed 16 November 2018 (https://durangoherald.com/articles/250613-where-have-all-the-elkgone). 6 �Eacker, D. R. 2015. Linking the effects of risk factors on annual calf survival to elk population dynamics in the Bitterroot Valley, Montana. M.S. thesis, University of Montana, Missoula, MT, USA. Eacker, D. R., M. Hebblewhite, K. M. Proffitt, B. S. Jimenez, M. S. Mitchell, and H. S. Robinson. 2016. Annual elk calf survival in a multiple carnivore system. Journal of Wildlife Management 80:1345–1359. Johnson, D. E. 1951. Biology of the elk calf, Cervus canadensis nelsoni. Journal of Wildlife Management 15:396–410. Kincaid, T.M., and A.R. Olsen. 2017. Spsurvey: spatial survey design and analysis. R package version 3.4. https://CRAN.R-roject.org/package=spsurvey. Lyon, L.J., and A.G. Christensen. 2002. Elk and land management in D.E. Toweill and J.W. Thomas, eds., North American Elk: ecology and management. Smithsonian Institute Press, Washington D.C., USA. Moeller, A.K., P.M. Lukacs, and J.S. Horne. 2018. Three novel methods to estimate abundance of unmarked animals using remote cameras. Ecosphere 9:e02331. Phillips, G.E. and A.W. Alldredge. 2000. Reproductive success of elk following disturbance by humans during calving season. Journal of Wildlife Management 64:521–530. Shively, K.J., A.W. Alldredge, and G.E. Phillips. 2005. Elk reproductive response to removal of calving season disturbance by humans. Journal of Wildlife Management 69:1073–1080. State of Colorado, Colorado Statewide Comprehensive Outdoor Recreation Plan. 2019. https://cpw.state.co.us/Documents/Trails/SCORP/Final-Plan/2019-SCORP-Report.pdf Accessed 23 January 2019. Steamboat Pilot and Today. 2018. Newly formed group advocates to slow trail building in Routt National Forest to protect wildlife. Published 21 October 2018, accessed 16 November 2018 (https://www.steamboatpilot.com/news/newly-formed-group-advocates-toslow-trail-building-in-routt-national-forest-to-protect-wildlife/). Stevens, D.L., and A.R. Olsen. 2004. Spatially balanced sampling of natural resources. Journal of the American Statistical Association 99:262–278. Stonehouse, K. F., C. R. J. Anderson, M. E. Peterson, and D. R. Collins. 2016. Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus). Technical Publication Number 48, Colorado Parks and Wildlife, Fort Collins, CO, USA. Outdoor Foundation. 2018. Outdoor Participation Report. https://outdoorindustry.org/resource/2018-outdoor-participation-report/. Accessed 26 September 2018. Vail Daily. 2018. Eagle County officials concerned by wildlife population declines. Published 14 September 2018, accessed 16 November 2018 (https://www.vaildaily.com/news/eaglecounty-officials-concerned-by-wildlife-population-declines/). Vieira, M.E.P., M.M. Conner, G.C. White, and D.J. Freddy. 2003. Effects of archery hunter numbers and opening dates on elk movement. Journal of Wildlife Management 67:717– 728. Wisdom, M.J., H.K. Preisler, L.M. Naylor, R.G. Anthony, B.K. Johnson, and M.M. Rowland, 2018. Elk responses to trail-based recreation on public forests. Forest Ecology and Management 411:223–233. Prepared by 7 �Eric J. Bergman, Wildlife Researcher 8 �Colorado Parks and Wildlife July 2021 – June 2022 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Colorado 3430 3002 2 : : : : Federal Aid Project No. W-242-R-6 : Parks and Wildlife Mammals Research Elk Conservation Response of Elk to Human Recreation at Multiple Scales: demographic shifts and behaviorally mediated fluctuations in local abundance Period Covered: July 1, 2021 – June 30, 2022 Authors: E.J. Bergman and N.D. Rayl Personnel: R. Aberle, J. Arntson, R. Baker, R. Black, K. Bond, D. Corcoran, S. Crews, Z. Durbin, P. Firmin, K. Fischer, M. Fisher, K. Fox, M. Gallagher, L. Gephert, J. Groves, K. Hatch, W. Hiler, Julie Mao, E. Los, M. McDaniel, K. Middledorf, L. Miller, E. Monfort, E. Newkirk, P. Nol, A. Orlando, J. Pollock, J. Potter, E. Sawa, A. Schneller, B. Smith, K. Tesch, E. VanNatta, S. Waters, M. Wood, M. Yamashita, CPW; J. Clark, H. Cushman, B. Dooling, A. Orlando, R. Swisher, S. Swisher, Quicksilver Air, Inc.. Project support received from Federal Aid in Wildlife Restoration, Rocky Mountain Elk Foundation, and CPW Big Game Auction and Raffle. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. ABSTRACT During the reporting period we focused on working with stakeholders and collaborators on research logistics, deploying field cameras, and capturing and collaring elk. Field efforts were centered on 3 objectives: 1) deploying/retrieving field cameras (i.e., cameras capable of timelapse and motion activation to estimate elk abundance), 2) capturing adult female elk, and collaring and outfitting pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior, and 3) capturing and collaring newborn and 6-month old elk calves to collect data on calf survival and cause-specific mortality. During spring (FY 20-21) and early summer (FY 21-22) of 2021 we deployed 238 cameras across 8 study units. During spring (FY21-22) and summer (FY2022-23) of 2022 data cards were 1 �retrieved. Downloading and cataloging of photos collected during the summer of 2021 occurred during summer 2022, with approximately 4.1 million photos collected. We radio-collared 25 6month old elk calves in December 2021. In March 2022, we radio-collared 40 pregnant elk and outfitted them with VITs. We estimated the pregnancy rate was 95% and the mean ingesta-free body fat of adult females was 7.9%. During the 2022 calving season, we radio-collared 53 elk calves. We estimated that the average date of calving was 3 June. WILDLIFE RESEARCH REPORT RESPONSE OF ELK TO HUMAN RECREATION AT MULTIPOLE SCALES: DEMOGRAPHIC SHIFTS AND BEHAVIORALLY MEDIATED FLUCTUATIONS IN LOCAL ABUNDANCE ERIC J. BERGMAN AND NATHANIEL D. RAYL PROJECT NARRATIVE OBJECTIVES This project has objectives on 2 scales. At the broad, elk herd-level scale, we will estimate pregnancy rates, calf survival rates, and cause-specific mortality rates to evaluate the importance of mortality sources for elk calf survival. More specifically, we will evaluate the influence of biotic (birth date, birth mass, gender, maternal body condition, habitat conditions), abiotic (previous and current weather conditions), and human-induced factors (i.e., relative exposure to recreational activities) on seasonal mortality risk of elk calves from birth to age 1 and on pregnancy rates of mature female elk. At the narrower geographic and temporal scale, we will use short-term (~3-4 weeks) changes in elk abundance within small study units (<65 km2) as a tool to evaluate the influence of human recreation on elk distribution. At this narrower scale, the primary objective is to evaluate the role that human recreation (e.g., hiking, mountain biking, horseback riding, trail running, hunting, etc.) has on the behavioral distribution of elk on spring calving, summer, and fall transition ranges. Coupled to the objective of detecting behaviorally influenced changes in abundance and density, we will evaluate the effectiveness of current recreational closures maintained by ski areas, counties, and federal land management agencies. SEGMENT OBJECTIVES 1. Work with personnel from CPW Areas 8 and 10, and private landowners on field research logistics. 2. Deploy field cameras (i.e., cameras capable of time-lapse) to estimate elk abundance. 3. Capture adult female elk, and collar and outfit pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior. 4. Capture and collar newborn and 6-month old elk to collect data on calf survival and cause-specific sources of mortality. 2 �INTRODUCTION The role of outdoor recreation within the state of Colorado is difficult to overstate. According to Colorado's Statewide Comprehensive Outdoor Recreation Plan (SCORP), outdoor recreation contributes 511,000 jobs, $62.5 billion in economic output, and $9.4 billion in local, state, and federal tax revenue (State of Colorado 2019). Outdoor recreation includes multiple activities such as biking, camping, climbing, fishing, hiking, horseback riding, hunting, shooting, skiing and wildlife watching. The Outdoor Foundation estimates that in the United States, there are nearly 30 million hikers, just under 7 million mountain bikers, over 14 million hunters, and nearly 23 million wildlife watchers (Outdoor Foundation 2018). While difficult to quantify, it is a reasonable assumption that many individual outdoor enthusiasts actively participate in more than one of these activities. Thus, the economies of Colorado, its counties, and its communities, rely on managing the landscape for a multitude of outdoor recreational opportunities. However, there is also evidence that human activities have an impact on wildlife. While trail-based recreation has the potential to impact many species, recent concerns in Colorado have focused on elk (Cervus canadensis; Durango Herald 2018, Steamboat Pilot and Today 2018, Vail Daily 2018). The sensitivity of elk to human presence and human activity has been a topic of interest for many decades. Preliminary research, focused on the effects of logging and vehicle use along road networks, provided consistent and clear evidence that elk-use declined in areas with high road densities, and as road use increased (Lyon and Christensen 2002). Similarly, research in Colorado evaluating the repeated displacement and disturbance of elk by people on foot provided evidence of suppressed recruitment rates following human disturbance (Phillips and Alldredge 2000, Shively et al. 2005). Experimental evaluation of the impact of hunter presence on elk movements and elk distribution has also occurred in Colorado (Conner et al. 2001, Vieira et al. 2003). This research demonstrated that the presence of hunters shifted elk off public lands and onto neighboring private lands. More recently, recreational trail use (all-terrain vehicle (ATV) riding, hiking, biking, and horseback riding) impacts on elk use of areas with trails was experimentally evaluated in Oregon. Wisdom et al. (2018) found that elk avoided areas with trails when recreationists of any type were present. Thus, regardless of human activity, behavioral displacement of elk by humans is well documented. In Colorado, increasing public concerns over human recreational use have coincided with declines in elk productivity, but a direct relationship to this activity in Colorado remains unaddressed. During FY 2016–2017, Colorado Parks and Wildlife (CPW) initiated a large-scale pilot study designed to evaluate pregnancy rates, elk calf survival, and causes of elk calf mortality (Alldredge 2016). At the onset, it was recognized that many factors contribute to suppression of pregnancy rates and calf survival. In addition to hunting, deteriorating habitat quality, habitat loss, and predation are key factors that may influence Colorado’s ungulate herds. Likewise, factors such as disease and competition may also play a role. Less clear, however, are the effects that human recreation may exert on the population dynamics of elk and other large ungulates. Past research has also reported individual behavioral responses of elk exposed to recreational stimuli. However, an alternate approach to studying behavioral displacement would shift the focus away from individual animals and link elk distribution to specific geographic areas. One limitation to studying individual animals is that the presence or absence of unmarked animals within the study area is largely ignored. However, access management and land 3 �management planning decisions are intrinsically tied to geographic areas. Thus, knowledge about the presence, absence, and abundance of a species of interest is of great value to managers. STUDY AREAS This study is occurring in two study areas. The northern study area focuses on the Bear’s Ears elk herd between Craig and Steamboat Springs. Within the Bear’s Ears herd, the fine scale camera-based behavioral portion of this study is centered on the Routt County segment of the herd that uses the Elk River drainage near the community of Steamboat Springs. The Bear’s Ears study area will be sampled using 4 study units: Mad Creek, Buffalo Pass, Walton Rim, and Hwy 40/Ferndale. The northernmost Mad Creek study unit has few existing trails but has been identified as a potential site for future trail development. Immediately south of the Mad Creek study unit is the Buffalo Pass study unit. Extensive trail development in this study area occurred during the past 5-10 years and it is currently an important and key area for many trail-based recreational activities. Further south is the Walton Rim study unit. Bounded to the north by the Steamboat Springs ski area, Walton Rim currently has little or no recreational trail use and plans for future trail development in this unit currently do not exist. Finally, immediately south of Walton Rim falls the Hwy 40/Ferndale study unit. Currently the Hwy 40/Ferndale study unit has nominal trail development and use, but plans for future trail construction in this unit are being considered. With the exception of Walton Rim, all of the camera-based study units in the Bear’s Ears study area have the potential to experience extensive trail development and use. The southern study area is focused on the Avalanche Creek elk herd along the Roaring Fork River between Glenwood Springs and Aspen. Four camera-based study units in the Avalanche Creek study area have also been identified. The southernmost of these units is the Snowmass unit. This unit, managed by Pitkin County and White River National Forest, has existing trails but is managed with seasonal closures (low elevation trail closures in place until May 16th, and high elevation trail closures in place until June 21st) to protect elk wintering and calving areas. Near the Snowmass study unit (and also in the southern portion of this study area) is the Wildcat study unit. The Wildcat study unit is centered on private property and has nominal recreational trail use, allowing it to serve as a reference area. Further north and nearer the community of Carbondale are 2 additional study units. The eastern most of these additional 2 units is The Crown, which is managed by the Bureau of Land Management and has extensive recreational use, but also has winter closures for mechanized and motorized recreation. Immediately to the south and west of The Crown is the fourth study unit. This unit, Two Shoes Ranch, is privately managed and has little recreational use and minimal trail development. METHODS Camera Sampling — Recent development of non-invasive abundance estimation techniques provide opportunities to quantify species in finite areas over relatively short periods. Camera based Space-To-Event (STE) and Instantaneous Sampling (IS) methods provide tools to estimate abundance without expensive flight time (Moeller et al. 2018). An inherent property of these new techniques is that the scope of inference applies to geographic areas and not individual animals. We deployed remote field cameras (HP2X, Reconyx, Holmen, Wisconsin, USA) to estimate elk abundance and density within small geographic areas (<65 km2) and during short time frames (~3–4 weeks). We deployed cameras across 8 study units (4 within the Bear’s Ears 4 �study area and 4 within the Avalanche Creek study area). We designed grids composed of 1.6 km cells to overlay each study unit. Within each cell, we used generalized random-tessellation stratification (GRTS) sampling (Stevens and Olsen 2004, Kincaid and Olsen 2017) to select 2 coarse camera locations. We selected final camera site locations (<250 m of the randomly selected coarse locations) in the field with the specific objective of maximizing detection probability of elk. We deployed cameras during the spring and early summer seasons. We programmed the cameras to take pictures at 10-minute intervals throughout the day. Elk capture and handling — We captured adult female elk ≥2 years of age by helicopter netgunning during late winter (March). During capture, we marked individuals with ear tags, collected a blood sample, and measured hind foot length, chest girth, and age based on tooth eruption and wear patterns. We used a portable ultrasound machine to assess whether or not captured elk were pregnant, and estimated the percent of ingesta-free body fat (IFBF) following methods detailed in Cook et al. (2010). We verified non-pregnancies using pregnancy-specific protein B (PSPB) analysis of sampled blood. We outfitted pregnant elk with vaginal implant transmitters (VITs) and Global Positioning System (GPS) radio-collars that attempt to acquire a location every 2 h. We deployed VITs that use the satellite communication capabilities of the collar on the adult female to send a notification when the VIT is expelled, signifying a birth. In December, we captured 6-month old elk calves by helicopter net-gunning. During capture, we measured body mass, hind foot length, chest girth, and determined the gender of captured calves. We outfitted calves with expandable GPS radio-collars that were scheduled to drop off after 6 months. During all captures, we followed CPW’s animal care and use guidelines for capturing and handling elk (CPW ACUC #09-2008). Cause-specific mortality — Within 24 hours of detecting a mortality signal from an elk collar, we attempted to conduct a systematic field investigation to determine the cause of death. We searched the area surrounding kill sites for evidence of predator presence, including predator scats, tracks, and hair, or signs of a struggle (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). We examined elk carcasses for evidence of canine puncture wounds, subcutaneous hemorrhaging and bruising, aspirated blood in the mouth, nose, or trachea, claw or bite marks on the hide, cracked or chewed bones, and characteristic consumption patterns (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). Nutritional condition of adult female elk — The body fat of lactating and non-lactating adult female elk can vary substantially, as lactating females are more sensitive than non-lactating females to their nutritional environment (Cook et al. 2004a, 2013). Therefore, it is difficult to interpret the body condition of adult female elk in late winter without knowing whether or not they experienced the energetic demands of lactation throughout the previous growing season (Cook et al. 2004a, 2013). RESULTS AND DISCUSSION In December 2021, we collared 25 6-month old elk calves from the Avalanche Creek herd. The mean weight of calves was 116.5 kg (95% CI: 111.3-121.7 kg). During March 2022, we radio-collared 40 pregnant elk and outfitted them with VITs. We estimated the pregnancy rate of adult female elk was 95% (85-99%). Elk populations experiencing good to excellent 5 �summer-autumn nutrition typically have pregnancy rates ≥90% (Cook et al. 2013). We estimated the mean IFBF of adult female elk to be 7.9%. When late-winter IFBF values are <8-9% for adult female elk that have lactated through the previous growing season, this suggests that there may be nutritional limitations, but it does not identify whether limitations are a result of summerautumn or winter nutrition (R. Cook, personal communication). During the summer of 2021, approximately 4.1 million photos were taken by the 238 cameras deployed across 8 study units. These photos are actively being archived. Automated photo recognition software continues to be developed and will be applied to these photos to expedite future analyses. However, subsampling of photos (and longer time intervals) will occur prior to analysis. SUMMARY During July 2021 – June 2022 we successfully worked with private landowners and personnel from CPW to coordinate field research logistics and initiate the third year of this study. We collected data on body condition and reproduction by capturing adult female elk, and we outfitted 40 pregnant females with GPS collars and VITs. We successfully captured and collared 53 newborn elk and 25 6-month old elk calves, meeting our sample size objectives, and allowing us to collect data on calf survival and cause-specific sources of mortality. We will continue to collect data on elk survival and cause-specific sources of mortality throughout the year. Field cameras were also successfully deployed. LITERATURE CITED Alldredge, M. 2016. Pilot study – elk recruitment and habitat use in Colorado. Program Narrative Study Plan, Colorado Parks and Wildlife, Fort Collins, CO, USA. Barber-Meyer, S. M., L. D. Mech, and P. J. White. 2008. Elk calf survival and mortality following wolf restoration to Yellowstone National Park. Wildlife Monographs 169:1–30. Conner, M.M., G.C. White, and D.J. Freddy. 2001. Elk movement in response to early-season hunting in northwest Colorado. Journal of Wildlife Management 65:926–940. Cook, J. G., B. K. Johnson, R. C. Cook, R. A. Riggs, T. Delcurto, L. D. Bryant, and L. L. Irwin. 2004. Effects of summer-autumn nutrition and parturition date on reproduction and survival of elk. Wildlife Monographs 155:1–61. Cook, R. C., J. G. Cook, T. R. Stephenson, W. L. Myers, S. M. Mccorquodale, D. J. Vales, L. L. Irwin, P. B. Hall, R. D. Spencer, S. L. Murphie, K. A. Schoenecker, and P. J. Miller. 2010. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife Management 74:880–896. Cook, R. C., J. G. Cook, D. J. Vales, B. K. Johnson, S. M. McCorquodale, L. A. Shipley, R. A. Riggs, L. L. Irwin, S. L. Murphie, B. L. Murphie, K. A. Schoenecker, F. Geyer, P. B. Hall, R. D. Spencer, D. A. Immell, D. H. Jackson, B. L. Tiller, P. J. Miller, and L. Schmitz. 2013. Regional and seasonal patterns of nutritional condition and reproduction in elk. Wildlife Monographs 184:1–44. Durango Herald. 2018. Where have all the elk gone? Published 15 November 2018, accessed 16 November 2018 (https://durangoherald.com/articles/250613-where-have-all-the-elkgone). Eacker, D. R. 2015. Linking the effects of risk factors on annual calf survival to elk population 6 �dynamics in the Bitterroot Valley, Montana. M.S. thesis, University of Montana, Missoula, MT, USA. Eacker, D. R., M. Hebblewhite, K. M. Proffitt, B. S. Jimenez, M. S. Mitchell, and H. S. Robinson. 2016. Annual elk calf survival in a multiple carnivore system. Journal of Wildlife Management 80:1345–1359. Johnson, D. E. 1951. Biology of the elk calf, Cervus canadensis nelsoni. Journal of Wildlife Management 15:396–410. Kincaid, T.M., and A.R. Olsen. 2017. Spsurvey: spatial survey design and analysis. R package version 3.4. https://CRAN.R-roject.org/package=spsurvey. Lyon, L.J., and A.G. Christensen. 2002. Elk and land management in D.E. Toweill and J.W. Thomas, eds., North American Elk: ecology and management. Smithsonian Institute Press, Washington D.C., USA. Moeller, A.K., P.M. Lukacs, and J.S. Horne. 2018. Three novel methods to estimate abundance of unmarked animals using remote cameras. Ecosphere 9:e02331. Phillips, G.E. and A.W. Alldredge. 2000. Reproductive success of elk following disturbance by humans during calving season. Journal of Wildlife Management 64:521–530. Shively, K.J., A.W. Alldredge, and G.E. Phillips. 2005. Elk reproductive response to removal of calving season disturbance by humans. Journal of Wildlife Management 69:1073–1080. State of Colorado, Colorado Statewide Comprehensive Outdoor Recreation Plan. 2019. https://cpw.state.co.us/Documents/Trails/SCORP/Final-Plan/2019-SCORP-Report.pdf Accessed 23 January 2019. Steamboat Pilot and Today. 2018. Newly formed group advocates to slow trail building in Routt National Forest to protect wildlife. Published 21 October 2018, accessed 16 November 2018 (https://www.steamboatpilot.com/news/newly-formed-group-advocates-toslow-trail-building-in-routt-national-forest-to-protect-wildlife/). Stevens, D.L., and A.R. Olsen. 2004. Spatially balanced sampling of natural resources. Journal of the American Statistical Association 99:262–278. Stonehouse, K. F., C. R. J. Anderson, M. E. Peterson, and D. R. Collins. 2016. Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus). Technical Publication Number 48, Colorado Parks and Wildlife, Fort Collins, CO, USA. Outdoor Foundation. 2018. Outdoor Participation Report. https://outdoorindustry.org/resource/2018-outdoor-participation-report/. Accessed 26 September 2018. Vail Daily. 2018. Eagle County officials concerned by wildlife population declines. Published 14 September 2018, accessed 16 November 2018 (https://www.vaildaily.com/news/eaglecounty-officials-concerned-by-wildlife-population-declines/). Vieira, M.E.P., M.M. Conner, G.C. White, and D.J. Freddy. 2003. Effects of archery hunter numbers and opening dates on elk movement. Journal of Wildlife Management 67:717– 728. Wisdom, M.J., H.K. Preisler, L.M. Naylor, R.G. Anthony, B.K. Johnson, and M.M. Rowland, 2018. Elk responses to trail-based recreation on public forests. Forest Ecology and Management 411:223–233. Prepared by Eric J. Bergman, Wildlife Researcher 7 �8 �Colorado Parks and Wildlife July 2022 – June 2023 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Colorado 3430 3002 2 : : : : Federal Aid Project No. W-242-R-7 : Parks and Wildlife Mammals Research Elk Conservation Response of Elk to Human Recreation at Multiple Scales: demographic shifts and behaviorally mediated fluctuations in local abundance Period Covered: July 1, 2022 – June 30, 2023 Authors: E.J. Bergman and N.D. Rayl All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. ABSTRACT During the reporting period we focused on working with stakeholders and collaborators on research logistics, deploying field cameras, and capturing and collaring elk. Field efforts were centered on 3 objectives: 1) deploying/retrieving field cameras (i.e., cameras capable of timelapse and motion activation to estimate elk abundance), 2) capturing adult female elk, and collaring and outfitting pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior, and 3) capturing and collaring newborn and 6-month old elk calves to collect data on calf survival and cause-specific mortality. During spring (FY 21-22) and early summer (FY 22-23) of 2021 we deployed 238 cameras across 8 study units. Concurrently, data from the summer of 2022 data cards were retrieved. Downloading and cataloging of photos collected during the summer of 2022 is ongoing, with an expected total of 3.5–4.0 million photos collected. During FY21-22, a research contract with Colorado State University was developed, with the intent of having a postdoctoral researcher develop an artificial intelligence processing and work flow system for analyzing all unclassified photos. Similarly, a collaborative relationship with the USFS Rocky Mountain Research Station was developed, for the purpose of using cell phone data to estimate human activity in study areas associated with this project. We radio-collared 25 6-month-old elk calves in December 2022. In March 2023, we radio-collared 40 pregnant elk and outfitted them with VITs. We estimated the 1 �pregnancy rate was 93% and the mean ingesta-free body fat of adult females was 8.1%. During the 2023 calving season, we radio-collared 60 elk calves. 2 �WILDLIFE RESEARCH REPORT RESPONSE OF ELK TO HUMAN RECREATION AT MULTIPOLE SCALES: DEMOGRAPHIC SHIFTS AND BEHAVIORALLY MEDIATED FLUCTUATIONS IN LOCAL ABUNDANCE ERIC J. BERGMAN AND NATHANIEL D. RAYL PROJECT NARRATIVE OBJECTIVES This project has objectives on 2 scales. At the broad, elk herd-level scale, we will estimate pregnancy rates, calf survival rates, and cause-specific mortality rates to evaluate the importance of mortality sources for elk calf survival. More specifically, we will evaluate the influence of biotic (birth date, birth mass, gender, maternal body condition, habitat conditions), abiotic (previous and current weather conditions), and human-induced factors (i.e., relative exposure to recreational activities) on seasonal mortality risk of elk calves from birth to age 1 and on pregnancy rates of mature female elk. At the narrower geographic and temporal scale, we will use short-term (~3-4 weeks) changes in elk abundance within small study units (<65 km2) as a tool to evaluate the influence of human recreation on elk distribution. At this narrower scale, the primary objective is to evaluate the role that human recreation (e.g., hiking, mountain biking, horseback riding, trail running, hunting, etc.) has on the behavioral distribution of elk on spring calving, summer, and fall transition ranges. Coupled to the objective of detecting behaviorally influenced changes in abundance and density, we will evaluate the effectiveness of current recreational closures maintained by ski areas, counties, and federal land management agencies. SEGMENT OBJECTIVES 1. Work with personnel from CPW Areas 8 and 10, and private landowners on field research logistics. 2. Deploy field cameras (i.e., cameras capable of time-lapse) to estimate elk abundance. 3. Capture adult female elk, and collar and outfit pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior. 4. Capture and collar newborn and 6-month old elk to collect data on calf survival and cause-specific sources of mortality. INTRODUCTION The role of outdoor recreation within the state of Colorado is difficult to overstate. According to Colorado's Statewide Comprehensive Outdoor Recreation Plan (SCORP), outdoor recreation contributes 511,000 jobs, $62.5 billion in economic output, and $9.4 billion in local, state, and federal tax revenue (State of Colorado 2019). Outdoor recreation includes multiple activities such as biking, camping, climbing, fishing, hiking, horseback riding, hunting, shooting, 3 �skiing and wildlife watching. The Outdoor Foundation (2018) estimates that in the United States, there are nearly 30 million hikers, just under 7 million mountain bikers, over 14 million hunters, and nearly 23 million wildlife watchers. While difficult to quantify, it is a reasonable assumption that many individual outdoor enthusiasts actively participate in more than one of these activities. Thus, the economies of Colorado, its counties, and its communities, rely on managing the landscape for a multitude of outdoor recreational opportunities. However, there is also evidence that human activities have an impact on wildlife. While trail-based recreation has the potential to impact many species, recent concerns in Colorado have focused on elk (Cervus canadensis; Durango Herald 2018, Steamboat Pilot and Today 2018, Vail Daily 2018). The sensitivity of elk to human presence and human activity has been a topic of interest for many decades. Preliminary research, focused on the effects of logging and vehicle use along road networks, provided consistent and clear evidence that elk-use declined in areas with high road densities, and as road use increased (Lyon and Christensen 2002). Similarly, research in Colorado evaluating the repeated displacement and disturbance of elk by people on foot provided evidence of suppressed recruitment rates following human disturbance (Phillips and Alldredge 2000, Shively et al. 2005). Experimental evaluation of the impact of hunter presence on elk movements and elk distribution has also occurred in Colorado (Conner et al. 2001, Vieira et al. 2003). This research demonstrated that the presence of hunters shifted elk off public lands and onto neighboring private lands. More recently, recreational trail use (all-terrain vehicle (ATV) riding, hiking, biking, and horseback riding) impacts on elk use of areas with trails was experimentally evaluated in Oregon. Wisdom et al. (2018) found that elk avoided areas with trails when recreationists of any type were present. Thus, regardless of human activity, behavioral displacement of elk by humans is well documented. In Colorado, increasing public concerns over human recreational use have coincided with declines in elk productivity, but a direct relationship to this activity in Colorado remains unaddressed. During FY 2016–2017, Colorado Parks and Wildlife (CPW) initiated a large-scale pilot study designed to evaluate pregnancy rates, elk calf survival, and causes of elk calf mortality (Alldredge 2016). At the onset, it was recognized that many factors contribute to suppression of pregnancy rates and calf survival. In addition to hunting, deteriorating habitat quality, habitat loss, and predation are key factors that may influence Colorado’s ungulate herds. Likewise, factors such as disease and competition may also play a role. Less clear, however, are the effects that human recreation may exert on the population dynamics of elk and other large ungulates. Past research has also reported individual behavioral responses of elk exposed to recreational stimuli. However, an alternate approach to studying behavioral displacement would shift the focus away from individual animals and link elk distribution to specific geographic areas. One limitation to studying individual animals is that the presence or absence of unmarked animals within the study area is largely ignored. However, access management and land management planning decisions are intrinsically tied to geographic areas. Thus, knowledge about the presence, absence, and abundance of a species of interest is of great value to managers. STUDY AREAS This study is occurring in two study areas. The northern study area focuses on the Bear’s Ears elk herd between Craig and Steamboat Springs. Within the Bear’s Ears herd, the fine scale camera-based behavioral portion of this study is centered on the Routt County segment of the herd that uses the Elk River drainage near the community of Steamboat Springs. The Bear’s Ears 4 �study area will be sampled using 4 study units: Mad Creek, Buffalo Pass, Walton Rim, and Hwy 40/Ferndale. The northernmost Mad Creek study unit has few existing trails but has been identified as a potential site for future trail development. Immediately south of the Mad Creek study unit is the Buffalo Pass study unit. Extensive trail development in this study area occurred during the past 5-10 years and it is currently an important and key area for many trail-based recreational activities. Further south is the Walton Rim study unit. Bounded to the north by the Steamboat Springs ski area, Walton Rim currently has little or no recreational trail use and plans for future trail development in this unit currently do not exist. Finally, immediately south of Walton Rim falls the Hwy 40/Ferndale study unit. Currently the Hwy 40/Ferndale study unit has nominal trail development and use, but plans for future trail construction in this unit are being considered. With the exception of Walton Rim, all of the camera-based study units in the Bear’s Ears study area have the potential to experience extensive trail development and use. The southern study area is focused on the Avalanche Creek elk herd along the Roaring Fork River between Glenwood Springs and Aspen. Four camera-based study units in the Avalanche Creek study area have also been identified. The southernmost of these units is the Snowmass unit. This unit, managed by Pitkin County and White River National Forest, has existing trails but is managed with seasonal closures (low elevation trail closures in place until May 16th, and high elevation trail closures in place until June 21st) to protect elk wintering and calving areas. Near the Snowmass study unit (and also in the southern portion of this study area) is the Wildcat study unit. The Wildcat study unit is centered on private property and has nominal recreational trail use, allowing it to serve as a reference area. Further north and nearer the community of Carbondale are 2 additional study units. The eastern most of these additional 2 units is The Crown, which is managed by the Bureau of Land Management and has extensive recreational use, but also has winter closures for mechanized and motorized recreation. Immediately to the south and west of The Crown is the fourth study unit. This unit, Two Shoes Ranch, is privately managed and has little recreational use and minimal trail development. METHODS Camera Sampling — Recent development of non-invasive abundance estimation techniques provide opportunities to quantify species in finite areas over relatively short periods. Camera based Space-To-Event (STE) and Instantaneous Sampling (IS) methods provide tools to estimate abundance without expensive flight time (Moeller et al. 2018). An inherent property of these new techniques is that the scope of inference applies to geographic areas and not individual animals. We deployed remote field cameras (HP2X, Reconyx, Holmen, Wisconsin, USA) to estimate elk abundance and density within small geographic areas (<65 km2) and during short time frames (~3–4 weeks). We deployed cameras across 8 study units (4 within the Bear’s Ears study area and 4 within the Avalanche Creek study area). We designed grids composed of 1.6 km cells to overlay each study unit. Within each cell, we used generalized random-tessellation stratification (GRTS) sampling (Stevens and Olsen 2004, Kincaid and Olsen 2017) to select 2 coarse camera locations. We selected final camera site locations (<250 m of the randomly selected coarse locations) in the field with the specific objective of maximizing detection probability of elk. We deployed cameras during the spring and early summer seasons. We programmed the cameras to take pictures at 10-minute intervals throughout the day. 5 �Elk capture and handling — We captured adult female elk ≥2 years of age by helicopter netgunning during late winter (March). During capture, we marked individuals with ear tags, collected a blood sample, and measured hind foot length, chest girth, and age based on tooth eruption and wear patterns. We used a portable ultrasound machine to assess whether or not captured elk were pregnant, and estimated the percent of ingesta-free body fat (IFBF) following methods detailed in Cook et al. (2010). We verified non-pregnancies using pregnancy-specific protein B (PSPB) analysis of sampled blood. We outfitted pregnant elk with vaginal implant transmitters (VITs) and Global Positioning System (GPS) radio-collars that attempt to acquire a location every 2 hrs. We deployed VITs that use the satellite communication capabilities of the collar on the adult female to send a notification when the VIT is expelled, signifying a birth. In December, we captured 6-month old elk calves by helicopter net-gunning. During capture, we measured body mass, hind foot length, chest girth, and determined the gender of captured calves. We outfitted calves with expandable GPS radio-collars that were scheduled to drop off after 6 months. During all captures, we followed CPW’s animal care and use guidelines for capturing and handling elk (CPW ACUC #09-2008). Cause-specific mortality — Within 24 hours of detecting a mortality signal from an elk collar, we attempted to conduct a systematic field investigation to determine the cause of death. We searched the area surrounding kill sites for evidence of predator presence, including predator scats, tracks, and hair, or signs of a struggle (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). We examined elk carcasses for evidence of canine puncture wounds, subcutaneous hemorrhaging and bruising, aspirated blood in the mouth, nose, or trachea, claw or bite marks on the hide, cracked or chewed bones, and characteristic consumption patterns (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). Nutritional condition of adult female elk — The body fat of lactating and non-lactating adult female elk can vary substantially, as lactating females are more sensitive than non-lactating females to their nutritional environment (Cook et al. 2004a, 2013). Therefore, it is difficult to interpret the body condition of adult female elk in late winter without knowing whether or not they experienced the energetic demands of lactation throughout the previous growing season (Cook et al. 2004a, 2013). RESULTS AND DISCUSSION In December 2022, we collared 25 6-month old elk calves from the Avalanche Creek herd. The mean weight of calves was 109.9 kg (95% CI: 102.9-117.0 kg). During March 2023, we radio-collared 40 pregnant elk and outfitted them with VITs. We estimated the pregnancy rate of adult female elk was 93% (82-98%). Elk populations experiencing good to excellent summer-autumn nutrition typically have pregnancy rates ≥90% (Cook et al. 2013). We estimated the mean IFBF of adult female elk to be 7.9%. When late-winter IFBF values are <8-9% for adult female elk that have lactated through the previous growing season, this suggests that there may be nutritional limitations, but it does not identify whether limitations are a result of summerautumn or winter nutrition (R. Cook, personal communication). During May-July 2023, we captured and collared 60 elk calves from the Avalanche Creek herd. During the summer of 2022, between 3.5–4.0 million photos were taken by the 238 cameras deployed across 8 study units. These photos are actively being archived. Automated 6 �photo recognition software is being developed, in collaboration with Colorado State University. Once this process is developed, it will be applied to these photos to expedite future analyses. Late during FY21–22, a collaborative effort with researchers from the USFS Rocky Mountain Research Station was initiated, to quantify human activity in our study areas. This activity, inferred from cell phone location data, will provide an index of human recreation. As part of this collaboration, efforts to disentangle different types of human activity from repeated location data will be made. SUMMARY From July 1, 2022 – June 30, 2023 we successfully worked with private landowners and personnel from CPW to coordinate field research logistics and initiate the fifth year of this study. We collected data on body condition and reproduction by capturing adult female elk, and we outfitted 40 pregnant females with GPS collars and VITs. We successfully captured and collared 60 newborn elk and 25 6-month old elk calves, meeting our sample size objectives, and allowing us to collect data on calf survival and cause-specific sources of mortality. Field cameras were also successfully deployed. LITERATURE CITED Alldredge, M. 2016. Pilot study – elk recruitment and habitat use in Colorado. Program Narrative Study Plan, Colorado Parks and Wildlife, Fort Collins, CO, USA. Barber-Meyer, S. M., L. D. Mech, and P. J. White. 2008. Elk calf survival and mortality following wolf restoration to Yellowstone National Park. Wildlife Monographs 169:1–30. Conner, M.M., G.C. White, and D.J. Freddy. 2001. Elk movement in response to early-season hunting in northwest Colorado. Journal of Wildlife Management 65:926–940. Cook, J. G., B. K. Johnson, R. C. Cook, R. A. Riggs, T. Delcurto, L. D. Bryant, and L. L. Irwin. 2004. Effects of summer-autumn nutrition and parturition date on reproduction and survival of elk. Wildlife Monographs 155:1–61. Cook, R. C., J. G. Cook, T. R. Stephenson, W. L. Myers, S. M. Mccorquodale, D. J. Vales, L. L. Irwin, P. B. Hall, R. D. Spencer, S. L. Murphie, K. A. Schoenecker, and P. J. Miller. 2010. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife Management 74:880–896. Cook, R. C., J. G. Cook, D. J. Vales, B. K. Johnson, S. M. McCorquodale, L. A. Shipley, R. A. Riggs, L. L. Irwin, S. L. Murphie, B. L. Murphie, K. A. Schoenecker, F. Geyer, P. B. Hall, R. D. Spencer, D. A. Immell, D. H. Jackson, B. L. Tiller, P. J. Miller, and L. Schmitz. 2013. Regional and seasonal patterns of nutritional condition and reproduction in elk. Wildlife Monographs 184:1–44. Durango Herald. 2018. Where have all the elk gone? Published 15 November 2018, accessed 16 November 2018 (https://durangoherald.com/articles/250613-where-have-all-the-elkgone). Eacker, D. R. 2015. Linking the effects of risk factors on annual calf survival to elk population dynamics in the Bitterroot Valley, Montana. M.S. thesis, University of Montana, Missoula, MT, USA. Eacker, D. R., M. Hebblewhite, K. M. Proffitt, B. S. Jimenez, M. S. Mitchell, and H. S. Robinson. 2016. Annual elk calf survival in a multiple carnivore system. Journal of Wildlife 7 �Management 80:1345–1359. Johnson, D. E. 1951. Biology of the elk calf, Cervus canadensis nelsoni. Journal of Wildlife Management 15:396–410. Kincaid, T.M., and A.R. Olsen. 2017. Spsurvey: spatial survey design and analysis. R package version 3.4. https://CRAN.R-roject.org/package=spsurvey. Lyon, L.J., and A.G. Christensen. 2002. Elk and land management in D.E. Toweill and J.W. Thomas, eds., North American Elk: ecology and management. Smithsonian Institute Press, Washington D.C., USA. Moeller, A.K., P.M. Lukacs, and J.S. Horne. 2018. Three novel methods to estimate abundance of unmarked animals using remote cameras. Ecosphere 9:e02331. Phillips, G.E. and A.W. Alldredge. 2000. Reproductive success of elk following disturbance by humans during calving season. Journal of Wildlife Management 64:521–530. Shively, K.J., A.W. Alldredge, and G.E. Phillips. 2005. Elk reproductive response to removal of calving season disturbance by humans. Journal of Wildlife Management 69:1073–1080. State of Colorado, Colorado Statewide Comprehensive Outdoor Recreation Plan. 2019. https://cpw.state.co.us/Documents/Trails/SCORP/Final-Plan/2019-SCORP-Report.pdf Accessed 23 January 2019. Steamboat Pilot and Today. 2018. Newly formed group advocates to slow trail building in Routt National Forest to protect wildlife. Published 21 October 2018, accessed 16 November 2018 (https://www.steamboatpilot.com/news/newly-formed-group-advocates-toslow-trail-building-in-routt-national-forest-to-protect-wildlife/). Stevens, D.L., and A.R. Olsen. 2004. Spatially balanced sampling of natural resources. Journal of the American Statistical Association 99:262–278. Stonehouse, K. F., C. R. J. Anderson, M. E. Peterson, and D. R. Collins. 2016. Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus). Technical Publication Number 48, Colorado Parks and Wildlife, Fort Collins, CO, USA. Outdoor Foundation. 2018. Outdoor Participation Report. https://outdoorindustry.org/resource/2018-outdoor-participation-report/. Accessed 26 September 2018. Vail Daily. 2018. Eagle County officials concerned by wildlife population declines. Published 14 September 2018, accessed 16 November 2018 (https://www.vaildaily.com/news/eaglecounty-officials-concerned-by-wildlife-population-declines/). Vieira, M.E.P., M.M. Conner, G.C. White, and D.J. Freddy. 2003. Effects of archery hunter numbers and opening dates on elk movement. Journal of Wildlife Management 67:717– 728. Wisdom, M.J., H.K. Preisler, L.M. Naylor, R.G. Anthony, B.K. Johnson, and M.M. Rowland, 2018. Elk responses to trail-based recreation on public forests. Forest Ecology and Management 411:223–233. Prepared by Eric J. Bergman, Wildlife Researcher 8 �Colorado Parks and Wildlife July 2023 – June 2024 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Colorado 3430 3002 2 : : : : Federal Aid Project No. W-242-R-8 : Parks and Wildlife Mammals Research Elk Conservation Response of Elk to Human Recreation at Multiple Scales: demographic shifts and behaviorally mediated fluctuations in local abundance Period Covered: July 1, 2023 – June 30, 2024 Authors: E.J. Bergman and N.D. Rayl All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. ABSTRACT During the reporting period we focused on working with stakeholders and collaborators on research logistics, deploying field cameras, and capturing and collaring elk. Field efforts were centered on 3 objectives: 1) deploying/retrieving field cameras (i.e., cameras capable of timelapse and motion activation to estimate elk abundance), 2) capturing adult female elk, and collaring and outfitting pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior, and 3) capturing and collaring newborn and 6-month old elk calves to collect data on calf survival and cause-specific mortality. During spring (FY 23-24) and early summer (FY 24-25) of 2024 we deployed 238 cameras across 8 study units. Concurrently, data from the summer of 2023 data cards were retrieved. Downloading and cataloging of photos collected during the summer of 2023 occurred during late summer 2024. During FY 23-24, all photos collected during this study were archived on a DNR server. A CPW funded postdoctoral researcher at Colorado State University identified a work flow system for analyzing all unclassified photos. All photos from 2019 and 2020 were processed accordingly. We radio-collared 25 6-month-old elk calves in December 2023. In March 2024, we radio-collared 40 pregnant elk and outfitted them with VITs. We estimated the pregnancy rate was 90% and the mean ingesta-free body fat of adult females was 8.3%. During the 2024 calving season, we radio-collared 55 elk calves. 1 �WILDLIFE RESEARCH REPORT RESPONSE OF ELK TO HUMAN RECREATION AT MULTIPOLE SCALES: DEMOGRAPHIC SHIFTS AND BEHAVIORALLY MEDIATED FLUCTUATIONS IN LOCAL ABUNDANCE ERIC J. BERGMAN AND NATHANIEL D. RAYL PROJECT NARRATIVE OBJECTIVES This project has objectives on 2 scales. At the broad, elk herd-level scale, we will estimate pregnancy rates, calf survival rates, and cause-specific mortality rates to evaluate the importance of mortality sources for elk calf survival. More specifically, we will evaluate the influence of biotic (birth date, birth mass, gender, maternal body condition, habitat conditions), abiotic (previous and current weather conditions), and human-induced factors (i.e., relative exposure to recreational activities) on seasonal mortality risk of elk calves from birth to age 1 and on pregnancy rates of mature female elk. At the narrower geographic and temporal scale, we will use short-term (~3-4 weeks) changes in elk abundance within small study units (<65 km2) as a tool to evaluate the influence of human recreation on elk distribution. At this narrower scale, the primary objective is to evaluate the role that human recreation (e.g., hiking, mountain biking, horseback riding, trail running, hunting, etc.) has on the behavioral distribution of elk on spring calving, summer, and fall transition ranges. Coupled to the objective of detecting behaviorally influenced changes in abundance and density, we will evaluate the effectiveness of current recreational closures maintained by ski areas, counties, and federal land management agencies. SEGMENT OBJECTIVES 1. Work with personnel from CPW Areas 8 and 10, and private landowners on field research logistics. 2. Deploy field cameras (i.e., cameras capable of time-lapse) to estimate elk abundance. 3. Capture adult female elk, and collar and outfit pregnant females with vaginal implant transmitters (VITs) to collect data on elk demography, body condition, reproduction, and behavior. 4. Capture and collar newborn and 6-month old elk to collect data on calf survival and cause-specific sources of mortality. INTRODUCTION The role of outdoor recreation within the state of Colorado is difficult to overstate. According to Colorado's Statewide Comprehensive Outdoor Recreation Plan (SCORP), outdoor recreation contributes 511,000 jobs, $62.5 billion in economic output, and $9.4 billion in local, state, and federal tax revenue (State of Colorado 2019). Outdoor recreation includes multiple activities such as biking, camping, climbing, fishing, hiking, horseback riding, hunting, shooting, 2 �skiing and wildlife watching. The Outdoor Foundation (2018) estimates that in the United States, there are nearly 30 million hikers, just under 7 million mountain bikers, over 14 million hunters, and nearly 23 million wildlife watchers. While difficult to quantify, it is a reasonable assumption that many individual outdoor enthusiasts actively participate in more than one of these activities. Thus, the economies of Colorado, its counties, and its communities, rely on managing the landscape for a multitude of outdoor recreational opportunities. However, there is also evidence that human activities have an impact on wildlife. While trail-based recreation has the potential to impact many species, recent concerns in Colorado have focused on elk (Cervus canadensis; Durango Herald 2018, Steamboat Pilot and Today 2018, Vail Daily 2018). The sensitivity of elk to human presence and human activity has been a topic of interest for many decades. Preliminary research, focused on the effects of logging and vehicle use along road networks, provided consistent and clear evidence that elk-use declined in areas with high road densities, and as road use increased (Lyon and Christensen 2002). Similarly, research in Colorado evaluating the repeated displacement and disturbance of elk by people on foot provided evidence of suppressed recruitment rates following human disturbance (Phillips and Alldredge 2000, Shively et al. 2005). Experimental evaluation of the impact of hunter presence on elk movements and elk distribution has also occurred in Colorado (Conner et al. 2001, Vieira et al. 2003). This research demonstrated that the presence of hunters shifted elk off public lands and onto neighboring private lands. More recently, recreational trail use (all-terrain vehicle (ATV) riding, hiking, biking, and horseback riding) impacts on elk use of areas with trails was experimentally evaluated in Oregon (Wisdom et al. 2018), and reported that elk avoided areas with trails when recreationists of any type were present. Thus, regardless of human activity, behavioral displacement of elk by humans is well documented. In Colorado, increasing public concerns over human recreational use have coincided with declines in elk productivity, but a direct relationship to this activity in Colorado remains unaddressed. During FY 2016–2017, Colorado Parks and Wildlife (CPW) initiated a large-scale pilot study designed to evaluate pregnancy rates, elk calf survival, and causes of elk calf mortality (Alldredge 2016). At the onset, it was recognized that many factors contribute to suppression of pregnancy rates and calf survival. In addition to hunting, deteriorating habitat quality, habitat loss, and predation are key factors that may influence Colorado’s ungulate herds. Likewise, factors such as disease and competition may also play a role. Less clear, however, are the effects that human recreation may exert on the population dynamics of elk and other large ungulates. Past research has also reported individual behavioral responses of elk exposed to recreational stimuli (Phillips and Alldredge 2000, Shively et al. 2005, Wisdom et al. 2018). However, an alternate approach to studying behavioral displacement would shift the focus away from individual animals and link elk distribution to specific geographic areas. One limitation to studying individual animals is that the presence or absence of unmarked animals within the study area is largely ignored. However, access management and land management planning decisions are intrinsically tied to geographic areas. Thus, knowledge about the presence, absence, and abundance of a species of interest is of great value to managers. STUDY AREAS This study is occurring in two study areas. The northern study area focuses on the Bear’s Ears elk herd between Craig and Steamboat Springs. Within the Bear’s Ears herd, the fine scale camera-based behavioral portion of this study is centered on the Routt County segment of the 3 �herd that uses the Elk River drainage near the community of Steamboat Springs. The Bear’s Ears study area will be sampled using 4 study units: Mad Creek, Buffalo Pass, Walton Rim, and Hwy 40/Ferndale. The northernmost Mad Creek study unit has few existing trails but has been identified as a potential site for future trail development. Immediately south of the Mad Creek study unit is the Buffalo Pass study unit. Extensive trail development in this study area occurred during the past 5-10 years and it is currently an important and key area for many trail-based recreational activities. Further south is the Walton Rim study unit. Bounded to the north by the Steamboat Springs ski area, Walton Rim currently has little or no recreational trail use and plans for future trail development in this unit currently do not exist. Finally, immediately south of Walton Rim falls the Hwy 40/Ferndale study unit. Currently the Hwy 40/Ferndale study unit has nominal trail development and use, but plans for future trail construction in this unit are being considered. With the exception of Walton Rim, all of the camera-based study units in the Bear’s Ears study area have the potential to experience extensive trail development and use. The southern study area is focused on the Avalanche Creek elk herd along the Roaring Fork River between Glenwood Springs and Aspen. Four camera-based study units in the Avalanche Creek study area have also been identified. The southernmost of these units is the Snowmass unit. This unit, managed by Pitkin County and White River National Forest, has existing trails but is managed with seasonal closures (low elevation trail closures in place until May 16th, and high elevation trail closures in place until June 21st) to protect elk wintering and calving areas. Near the Snowmass study unit (and also in the southern portion of this study area) is the Wildcat study unit. The Wildcat study unit is centered on private property and has nominal recreational trail use, allowing it to serve as a reference area. Further north and nearer the community of Carbondale are 2 additional study units. The eastern most unit is The Crown, which is managed by the Bureau of Land Management and has extensive recreational use, but also has winter closures for mechanized and motorized recreation. Immediately to the south and west of The Crown is the Two Shoes Ranch, which is privately managed and has little recreational use and minimal trail development. METHODS Camera Sampling — Recent development of non-invasive abundance estimation techniques provide opportunities to quantify species in finite areas over relatively short periods. Camera based Space-To-Event (STE) and Instantaneous Sampling (IS) methods provide tools to estimate abundance without expensive flight time (Moeller et al. 2018). An inherent property of these new techniques is that the scope of inference applies to geographic areas and not individual animals. We deployed remote field cameras (HP2X, Reconyx, Holmen, Wisconsin, USA) to estimate elk abundance and density within small geographic areas (<65 km2) and during short time frames (~3–4 weeks). We deployed cameras across 8 study units (4 within the Bear’s Ears study area and 4 within the Avalanche Creek study area). We designed grids composed of 1.6 km cells to overlay each study unit. Within each cell, we used generalized random-tessellation stratification (GRTS) sampling (Stevens and Olsen 2004, Kincaid and Olsen 2017) to select 2 coarse camera locations. We selected final camera site locations (<250 m of the randomly selected coarse locations) in the field with the specific objective of maximizing detection probability of elk. We deployed cameras during the spring and early summer seasons. We programmed the cameras to take pictures at 10-minute intervals throughout the day. 4 �Elk capture and handling — We captured adult female elk ≥2 years of age by helicopter netgunning during late winter (March). During capture, we marked individuals with ear tags, collected a blood sample, and measured hind foot length, chest girth, and age based on tooth eruption and wear patterns (Quimby and Gaab 1957). We used a portable ultrasound machine to assess whether or not captured elk were pregnant, and estimated the percent of ingesta-free body fat (IFBF) following methods detailed in Cook et al. (2010). We verified non-pregnancies using pregnancy-specific protein B (PSPB) analysis of sampled blood. We outfitted pregnant elk with vaginal implant transmitters (VITs) and Global Positioning System (GPS) radio-collars that attempt to acquire a location every 2 hrs. We deployed VITs that use the satellite communication capabilities of the collar on the adult female to send a notification when the VIT is expelled, signifying a birth. In December, we captured 6-month old elk calves by helicopter net-gunning. During capture, we measured body mass, hind foot length, chest girth, and determined the gender of captured calves. We outfitted calves with expandable GPS radio-collars that were scheduled to drop off after 6 months. During all captures, we followed CPW’s animal care and use guidelines for capturing and handling elk (CPW ACUC #09-2008). Cause-specific mortality — Within 24 hours of detecting a mortality signal from an elk collar, we attempted to conduct a systematic field investigation to determine the cause of death. We searched the area surrounding kill sites for evidence of predator presence, including predator scats, tracks, and hair, or signs of a struggle (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). We examined elk carcasses for evidence of canine puncture wounds, subcutaneous hemorrhaging and bruising, aspirated blood in the mouth, nose, or trachea, claw or bite marks on the hide, cracked or chewed bones, and characteristic consumption patterns (Barber-Meyer et al. 2008, Eacker et al. 2016, Stonehouse et al. 2016). Nutritional condition of adult female elk — The body fat of lactating and non-lactating adult female elk can vary substantially, as lactating females are more sensitive than non-lactating females to their nutritional environment (Cook et al. 2004a, 2013). Therefore, it is difficult to interpret the body condition of adult female elk in late winter without knowing whether or not they experienced the energetic demands of lactation throughout the previous growing season (Cook et al. 2004a, 2013). RESULTS AND DISCUSSION In December 2023, we collared 25 6-month old elk calves from the Avalanche Creek herd. The mean weight of calves was 108.8 kg (95% CI: 101.0-116.6 kg). During March 2024, we radio-collared 40 pregnant elk and outfitted them with VITs. We estimated the pregnancy rate of adult female elk was 90% (95% CI: 78-96%). Elk populations experiencing good to excellent summer-autumn nutrition typically have pregnancy rates ≥90% (Cook et al. 2013). We estimated the mean IFBF of adult female elk to be 8.3%. When late-winter IFBF values are <89% for adult female elk that have lactated through the previous growing season, this suggests that there may be nutritional limitations, but it does not identify whether limitations are a result of summer-autumn or winter nutrition (R. Cook, personal communication). During May-July 2024, we captured and collared 55 elk calves from the Avalanche Creek herd. 5 �During the summer of 2023, approximately 4.0 million photos were taken by the 238 cameras deployed across 8 study units. These photos were uploaded into appropriate databases and are actively being archived on a DNR server. Automated photo recognition processes were applied to photos collected during 2019 and 2020. These processes reduce the number of photos requiring human evaluation by ~90%. It is expected that during FY 24-25 human evaluation of photos collected from 2019-2023 will be complete. SUMMARY From July 1, 2023 – June 30, 2024 we successfully worked with private landowners and personnel from CPW to coordinate field research logistics and initiate the fifth year of this study. We collected data on body condition and reproduction by capturing adult female elk, and we outfitted 40 pregnant females with GPS collars and VITs. We successfully captured and collared 55 newborn elk and 25 6-month old elk calves, meeting our sample size objectives, and allowing us to collect data on calf survival and cause-specific sources of mortality. Field cameras were also successfully maintained for the final summer of data collection. LITERATURE CITED Alldredge, M. 2016. Pilot study – elk recruitment and habitat use in Colorado. Program Narrative Study Plan, Colorado Parks and Wildlife, Fort Collins, CO, USA. Barber-Meyer, S. M., L. D. Mech, and P. J. White. 2008. Elk calf survival and mortality following wolf restoration to Yellowstone National Park. Wildlife Monographs 169:1–30. Conner, M.M., G.C. White, and D.J. Freddy. 2001. Elk movement in response to early-season hunting in northwest Colorado. Journal of Wildlife Management 65:926–940. Cook, J. G., B. K. Johnson, R. C. Cook, R. A. Riggs, T. Delcurto, L. D. Bryant, and L. L. Irwin. 2004. Effects of summer-autumn nutrition and parturition date on reproduction and survival of elk. Wildlife Monographs 155:1–61. Cook, R. C., J. G. Cook, T. R. Stephenson, W. L. Myers, S. M. Mccorquodale, D. J. Vales, L. L. Irwin, P. B. Hall, R. D. Spencer, S. L. Murphie, K. A. Schoenecker, and P. J. Miller. 2010. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife Management 74:880–896. Cook, R. C., J. G. Cook, D. J. Vales, B. K. Johnson, S. M. McCorquodale, L. A. Shipley, R. A. Riggs, L. L. Irwin, S. L. Murphie, B. L. Murphie, K. A. Schoenecker, F. Geyer, P. B. Hall, R. D. Spencer, D. A. Immell, D. H. Jackson, B. L. Tiller, P. J. Miller, and L. Schmitz. 2013. Regional and seasonal patterns of nutritional condition and reproduction in elk. Wildlife Monographs 184:1–44. Durango Herald. 2018. Where have all the elk gone? Published 15 November 2018, accessed 16 November 2018 (https://durangoherald.com/articles/250613-where-have-all-the-elkgone). Eacker, D. R. 2015. Linking the effects of risk factors on annual calf survival to elk population dynamics in the Bitterroot Valley, Montana. M.S. thesis, University of Montana, Missoula, MT, USA. Eacker, D. R., M. Hebblewhite, K. M. Proffitt, B. S. Jimenez, M. S. Mitchell, and H. S. Robinson. 2016. Annual elk calf survival in a multiple carnivore system. Journal of Wildlife Management 80:1345–1359. 6 �Quimby, D.C., and J.E. Gaab. 1957. Mandibular dentition as an age indicator in Rocky Mountain elk. Journal of Wildlife Management 21:435-451. Johnson, D. E. 1951. Biology of the elk calf, Cervus canadensis nelsoni. Journal of Wildlife Management 15:396–410. Kincaid, T.M., and A.R. Olsen. 2017. Spsurvey: spatial survey design and analysis. R package version 3.4. https://CRAN.R-roject.org/package=spsurvey. Lyon, L.J., and A.G. Christensen. 2002. Elk and land management in D.E. Toweill and J.W. Thomas, eds., North American Elk: ecology and management. Smithsonian Institute Press, Washington D.C., USA. Moeller, A.K., P.M. Lukacs, and J.S. Horne. 2018. Three novel methods to estimate abundance of unmarked animals using remote cameras. Ecosphere 9:e02331. Phillips, G.E. and A.W. Alldredge. 2000. Reproductive success of elk following disturbance by humans during calving season. Journal of Wildlife Management 64:521–530. Shively, K.J., A.W. Alldredge, and G.E. Phillips. 2005. Elk reproductive response to removal of calving season disturbance by humans. Journal of Wildlife Management 69:1073–1080. State of Colorado, Colorado Statewide Comprehensive Outdoor Recreation Plan. 2019. https://cpw.state.co.us/Documents/Trails/SCORP/Final-Plan/2019-SCORP-Report.pdf Accessed 23 January 2019. Steamboat Pilot and Today. 2018. Newly formed group advocates to slow trail building in Routt National Forest to protect wildlife. Published 21 October 2018, accessed 16 November 2018 (https://www.steamboatpilot.com/news/newly-formed-group-advocates-toslow-trail-building-in-routt-national-forest-to-protect-wildlife/). Stevens, D.L., and A.R. Olsen. 2004. Spatially balanced sampling of natural resources. Journal of the American Statistical Association 99:262–278. Stonehouse, K. F., C. R. J. Anderson, M. E. Peterson, and D. R. Collins. 2016. Approaches to field investigations of cause-specific mortality in mule deer (Odocoileus hemionus). Technical Publication Number 48, Colorado Parks and Wildlife, Fort Collins, CO, USA. Outdoor Foundation. 2018. Outdoor Participation Report. https://outdoorindustry.org/resource/2018-outdoor-participation-report/. Accessed 26 September 2018. Vail Daily. 2018. Eagle County officials concerned by wildlife population declines. Published 14 September 2018, accessed 16 November 2018 (https://www.vaildaily.com/news/eaglecounty-officials-concerned-by-wildlife-population-declines/). Vieira, M.E.P., M.M. Conner, G.C. White, and D.J. Freddy. 2003. Effects of archery hunter numbers and opening dates on elk movement. Journal of Wildlife Management 67:717– 728. Wisdom, M.J., H.K. Preisler, L.M. Naylor, R.G. Anthony, B.K. Johnson, and M.M. Rowland, 2018. Elk responses to trail-based recreation on public forests. Forest Ecology and Management 411:223–233. Prepared by Eric J. Bergman, Wildlife Researcher 7 � https://cpw.cvlcollections.org/files/original/9815eb0d52c429687679a2ba40a7f183.pdf cb86f6d1af74c21682ee29960f6f0c4d PDF Text Text Colorado Parks and Wildlife July 2022 – June 2023 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Colorado 3430 3003 Task No. 1 Federal Aid Project No. W-244-R-7 : Parks and Wildlife : Mammals Research : Predatory Mammals Conservation : Bobcat population density estimation: A pilot study : Period Covered: July 1, 2022 – June 30, 2023 Authors: S. Frank, J. Ivan, M. Vieira, and J. Runge All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. ABSTRACT Two bobcat study areas, one with low (Skull Creek) and another with high historical harvest (Piceance), were established in 2022. Camera trap grids were deployed in each 400 km2 study area, but only Piceance was complete and where the majority of capture efforts occurred, due to a severe winter season. Thirty-nine bobcat captures yielded 26 unique individual bobcats, 23 of which were collared (n = 19 in Piceance; n = 4 in Skull Creek). Population estimation using a mark-resight model was only possible in Piceance, where data were sufficient, and resulted in 34.62 bobcats (CV = 0.38, SE = 12.80) for the 400 km2 study area. Resights were regarded as low, potentially due to severe winter/reduced movement, suboptimal camera locations, and constrained trapping efforts (for marked individuals). Capture effort will be distributed and prioritized in hard-to-access winter areas prior to winter, more toward the center of the study area, and camera locations are being relocated to the best sites for resighting bobcats, as opposed to those areas best representing the habitat of a given grid cell. Altogether, these modifications will increase resights, provide more robust population estimates, and have a higher chance of estimating sex-specific densities that can be related to harvest statistics. �FINAL PERFORMANCE REPORT State: Colorado Project #: F22AF01995 Project Title: Predatory Mammals Conservation Period Covered: July 1, 2022 - June 30, 2023 (W-244-R-7) OBJECTIVES By June 30, 2023 1. Conduct mandatory checks on all bears and lions harvested in Colorado and maintain a database of all mandatory check data. Based on predicted license sales, historic participation and success rates, and on historic non-hunt mortality averages, we predict that about 1,900 black bear and about 600 mountain lion mortality records will be entered into our mortality database. 2. As part of the mandatory check process, collect tooth samples from harvested bears and lions to determine age and sex. This information will be used to provide insight into the age and sex structure of the population. Based on historic tooth collection rates, we predict that about 1,800 black bear teeth and about 550 mountain lion teeth will be processed for cementum age, and for black bear, female reproductive intervals determined. 3. Continue refinement of a human-bear / human-lion / human-wildlife conflict application that is accessible through multiple web based platforms (pc, tablet, smartphone). This is a multi-year project with initial roll-out of the low-code AppSheet version having occurred in April 2019 for bear/lion incidents and the fall of 2019 for bear/lion mortalities. The continuation of this project will involve both ongoing refinement and evaluation of the AppSheet application and continued scoping of the functional needs that agency staff will have for the database, structural planning, interaction with other agency platforms, and programming of a custom-designed application. Both the scoping of the custom application and ongoing evaluation of the low-code application will include assessing the flexibility of the platform to eventually include collection of other mandatory mortality data for other species (bobcats) and incident and conflict data beyond just bears and lions. 4. Evaluate relationships between bobcat density and sex/age composition with relative bobcat harvest (high versus low harvest). Results will be applied inform future bobcat harvest management. APPROACH Bear and Mountain Lion Mandatory Checks & Database Management Mandatory checks are performed by CPW personnel at all offices and in the field. Mandatory check information includes hunter and license information, as well as information about the �animal harvested (species, estimated age, sex, breeding status, harvest location, prior marks). As part of the mandatory check process, all bear and lion mortalities are marked with a numbered identification seal. The mandatory check process requires seals to be distributed to CPW personnel and information from the mandatory check forms to be entered into a central database, largely via the mortality application. This database is used to compile annual mortality estimates and to observe trends over time. A comprehensive training program has been initiated for CPW personnel that perform mandatory checks. Bear and Mountain Lion Tooth Analysis Tooth analysis can provide valuable information on the age and gender structure of bear and lion mortalities. Aging bears and lions based only on visual carcass inspection can have a high degree of error. Beginning in 2006, CPW began collecting premolar tooth samples, with the permission of hunters, as part of the bear and lion mandatory check process. However, a substantial percentage of hunters did not voluntarily allow a tooth to be extracted from their animal. In 2008, Wildlife Commission regulations required hunters to allow CPW to obtain a tooth sample. As a result, useable tooth samples are now collected from about 97% of hunterkilled bears and from 87% of hunter-killed lions (2018-2020 average). In addition to hunterharvested animals, tooth samples are collected from all bear and lion mortalities handled by CPW (e.g., vehicle mortalities, conflict mortalities) whenever possible. Tooth samples are submitted to a commercial laboratory for aging based on cementum annuli. In female bears it is also possible to use cementum deposition to identify years that females reared cubs. Because of the value of this information for management, CPW anticipates continuing bear and lion tooth collection and analysis as a routine part of the mandatory check process. Human-Wildlife Conflict Database and Recording Application Development CPW developed and deployed a low-code mobile application (AppSheets) to record information about the number, type, location, and date of human-bear and human-lion conflicts beginning in April 2019. This initial phase of use consisted of an evaluation to assess if this low-code application met the current needs for recording and tracking conflict incident information and if its use could be expanded to include bear/lion mandatory mortality check aspects described in Objective 1. During the fall of 2019, AppSheet was used statewide for bear/lion mandatory mortality data collection. Evaluation and integration of this data from this new source is currently underway. While AppSheet shows promise for most of the agency needs surrounding record-keeping of these data, an initial Request for Information and Request for Proposal scoping for development of a custom-designed application will continue. This application and database need to be secure because individual citizen’s and visitor’s identities will be contained within. The database will need geospatial and temporal update capabilities. The database will need to be accessible for data entry in real time or near-real time and from multiple platforms (PC, tablet, or smartphone devices). The database must also allow secure access for data analysis, plotting, and charting capabilities. Lastly the application and database will need to interact with other existing and future CPW software, applications and programs. Density and Sex/Age Composition Relative to Bobcat Harvest �We propose 2 400 km2 study areas of contiguous Colorado bobcat habitat that have different levels of bobcat harvest density (1 high harvest, 1 low harvest). Candidate study areas of high and low harvest will be identified through CPW harvest records. Within each of the 2 study areas, ~30 bobcats will be captured during September-November. Each animal will be uniquely marked with eartags and a GPS radiocollar. From early November (pre-harvest)–April (post-harvest), 80 remote trail cameras will be deployed per study area to provide pre- and post-hunt density estimates using marks (ear tags, GPS collars) to uniquely identify individual animals. Harvest rates on each study area will be estimated from the marked bobcats that are harvested and presented for mandatory inspection. During March, we will again capture bobcats to increase sample size and improve post-hunt density estimates. Spatially-explicit capture-recapture (SECR) models and spatial mark-resight estimators will be employed to estimate density, likely using both techniques in a “hybrid” fashion to improve precision. Given the cyclical nature of bobcat prey populations and the expected relationship with changes in bobcat density, we propose continuing this work over 5 years on each study area. The baseline sex/age composition of the pre-hunt population will be developed as bobcats are captured, and compared to sex/age composition data from harvested bobcats presented for mandatory inspection. We will evaluate how these changes in composition align with changes in density and how those compare to currently used predictors of change in population trend. GPS collars will provide habitat use, home range and survival estimates. This information will be useful to develop future population modeling approaches for bobcat management. Improving our understanding of bobcat populations in Colorado will provide us with a strong basis in supporting future management decisions and regulations. RESULTS 1. Bear and Mountain Lion Mandatory Checks & Database Management Mandatory checks and database management have been conducted as described in the foregoing section. The following table lists the number of checks conducted since this project began. The mortality database is available for managers to analyze gender, visual estimation of age, spatial and temporal mortality distribution, season, method of take for hunter harvest, and other non-hunt types of mortality. Year 2018-2019 2019-2020 2020-2021 2021-2022 2022-2023 Black Bear Mortality Mandatory Checks 1,620 1,792 2,210 2,003 2,171 Mountain Lion Mortality – Mandatory Checks 653 624 668 621 646 2. Bear and Mountain Lion Tooth Cementum Analysis Age and gender composition of hunter harvest and total mortality are used along with other �biological and social metrics to monitor the affect of hunting on population trajectory and human social parameters. Data is primarily analyzed at the DAU level to inform the management strategies in the 17 black bear DAUs and 8 mountain lion DAUs. In 2020, a new management plan for the West Slope of Colorado was developed which reduced the number of statewide lion DAUs down to 8. The number, type of samples, and basic age and gender data averages are displayed in the following tables. Year 2018-2019 2019-2020 2020-2021 2021-2022 2022-2023 Black Bear Tooth Sample Count 1,472 1,659 1,967 1,773 1,837 Mountain Lion Tooth Sample Count 531 594 585 587 581 Coarse scale analysis of black bear teeth show: the long-term age range of black bear in our mortality database is age class 0-30. Generally, black bears in non-hunt forms of mortality are equal, on average, to those in hunter harvest, but this is not annually consistent. Black Bear 2018-2019 Hunter Harvest Average Age Non-Hunt Mortality Average Age Age Range Female Average Age Primipatry Proportion in Harvest Proportion in Non-Hunt Mortality Male 4.6 4.6 0-27 62% 66% Female 6.4 5.4 0-30 4.6 38% 34% Black Bear 2019-2020 Hunter Harvest Average Age Non-Hunt Mortality Average Age Age Range Female Average Age Primipatry Proportion in Harvest Proportion in Non-Hunt Mortality Male 5.0 4.6 0-27 63% 74% Female 6.5 5.3 0-30 4.7 37% 26% Black Bear 2020-2021 Hunter Harvest Average Age Non-Hunt Mortality Average Age Age Range Female Average Age Primipatry Male 4.9 4.5 0-21 Female 6.4 6.1 0-26 4.5 �Proportion in Harvest Proportion in Non-Hunt Mortality 58% 70% 42% 30% Male 4.8 4.1 0-22 Female 6.1 5.5 0-29 4.8 36% 34% Black Bear 2021-2022 Hunter Harvest Average Age Non-Hunt Mortality Average Age Age Range Female Average Age Primipatry Proportion in Harvest Proportion in Non-Hunt Mortality Black Bear 2022-2023 Hunter Harvest Average Age Non-Hunt Mortality Average Age Age Range Female Average Age Primipatry Proportion in Harvest Proportion in Non-Hunt Mortality 64% 66% Male 4.6 4.4 0-24 58% 66% Female 6.2 6.6 0-26 4.7 42% 34% Coarse scale analysis of mountain lion teeth show: the long-term age range of lions in hunter harvest is age class 0-16 years. In non-hunt mortality the age range is age class 0-14 years. In contrast with bears, the age of lions in non-hunt mortality is generally younger than those in hunter harvest. Mountain Lion 2018-2019 Hunter Harvest Average Age Non-Hunt Mortality Average Age Age Range Proportion in Harvest Proportion in Non-Hunt Mortality Male 2.9 2.2 0-10 60% 51% Female 3.0 3.1 0-14 40% 49% Male 2.8 2.1 0-9 61% 43% Female 2.7 2.2 0-7 39% 57% Male 2.7 1.4 0-9 Female 2.7 2.1 0-10 Mountain Lion 2019-2020 Hunter Harvest Average Age Non-Hunt Mortality Average Age Age Range Proportion in Harvest Proportion in Non-Hunt Mortality Mountain Lion 2020-2021 Hunter Harvest Average Age Non-Hunt Mortality Average Age Age Range �Proportion in Harvest Proportion in Non-Hunt Mortality 61% 48% 39% 52% Male 2.1 1.7 0-7 59% 47% Female 2.4 2.7 0-9 41% 53% Male 2.4 1.9 0-7 59% 45% Female 2.6 1.9 0-10 41% 55% Mountain Lion 2021-2022 Hunter Harvest Average Age Non-Hunt Mortality Average Age Age Range Proportion in Harvest Proportion in Non-Hunt Mortality Mountain Lion 2022-2023 Hunter Harvest Average Age Non-Hunt Mortality Average Age Age Range Proportion in Harvest Proportion in Non-Hunt Mortality 3. Human-Wildlife Conflict Database Development A low-code application platform was selected, purchased and developed by CPW staff for recording lion and bear incidents (sighting and conflicts). This mobile application was distributed to staff cell phones and computers and was put into statewide use for recording all reported incidents beginning April 1, 2019. Staff developed supporting reference materials for using the application and created training tools and webinars to inform staff on how to use this information. Total incidents reported annually for bears and lions are reported below. Year 2019 (partial year starting Apr) 2020 2021 2022 2023 (as of Sept 15, 2023) Total Black Bear Incidents Recorded 5,392 4,971 3,706 4,292 2,894 Total Mountain Lion Incidents Recorded 723 868 767 788 564 The internal CPW workgroup responsible for the bear/lion incident application has also developed and distributed an application for recording drug use and handling information for all immobilized wildlife. This workgroup has also developed and distributed an application for the entering mortality data from bears and lions. This application replaces the previous paper mandatory mortality form which was used to record data from all mortalities of bears and lions. This workgroup continues to meet and work on future steps in development of a more robust application platform that could integrate the three existing apps into multiple streams of currently-used agency data and software. 4. Density and Sex/Age Composition Relative to Bobcat Harvest A bobcat pilot study was conducted during the 2022-2023 field season for the purposes of honing capture strategies/methods in relation to data collected on cameras and from GPS-collared bobcats for �the purposes of population density estimation. Two study areas were selected, each 400 km2, with one depicting ‘high’ historical legal harvest (>2.55 bobcats/100km2; hereafter “Piceance”) and the other ‘low’ historical harvest (near null; hereafter “Skull Creek”). Due to a rare, severe winter season, snow depth precluded access to some of Piceance and most of Skull Creek. A camera trap grid of 100 cameras was successfully deployed in Piceance and only 37 of 100 cameras were deployed in Skull Creek. Bobcat captures were conducted from November 2022 through April 2023 in Piceance and from February through March 2023 in Skull Creek. Each study area below contains 100 2x2 km cells with camera locations shown with gray dots, live cage trap sites in yellow, successful live trap sites in red. Capture effort (trap nights) and success (captures), and the demographic break-down of bobcat �capture/collar success is shown below. Twenty-six unique bobcats were captured and 23 of those were collared. Overall, the sex ratio of captured and collared animals was male-biased, i.e. 27:12 and 17:9, respectively. Study area Piceance Skull Creek Total Study Area Trap Nights Number of Captures 1376 295 1671 34 5 39 Sex Female Piceance Male Skull Creek Female Male Total Number of Individuals Age Adult Sub-adult Adult Sub-adult Adult Subadult Adult Trap Nights / New Individual Capture 66 59 -- Trap Nights / All Captures 21 5 26 40 59 -- Captured New Individuals 5 4 21 4 2 1 2 39 4 2 12 3 2 1 2 26 Collared 4 1 12 2 2 0 2 23 There were 947 bobcat images, yielding 119 ‘hits’ or independent detections. Most detections were unmarked (n = 93), 8 were marked, and 17 were unknown whether marked or unmarked bobcats. Six of the 17 (35%) of the marked, collared bobcats were detected. We estimated bobcat population density via a mark-resight immigration-emigration mixed logit-normal model for Piceance, due to a lack of geographic closure. For Piceance, bobcat density was estimated for the beginning of the period from November 19, 2022 through April 23, 2023, as that corresponds to the last camera trap set-up at the beginning of the season and first SD card pick-up at the end of the season; demographic closure was violated during this period, due to legal harvest (n = 8), so the derived estimate corresponds to density at the beginning of the sampling period. Preliminary results yielded a bobcat density of 34.62 bobcats (CV = 0.38, SE = 12.80) for the 400 km2 Piceance study area, i.e., 8.41 bobcats/100 km2. Bobcat density estimation was not possible for Skull Creek, due to too few collared bobcats (n = 4) and detectors on the camera grid. We suspect that the Piceance density estimate is biased low, due to low number of bobcat resights, due to many marked animals occurring off-grid (spatially biased trapping effort), reduced movement during a severe winter, and mostly due to suboptimal camera locations. Camera locations were initially chosen based on representing the majority of habitat type and a good travel route/location for bobcat detection. In several cases, however, lower represented habitats likely provided better locations for detecting bobcats in a camera grid cell and/or bobcat behavior on the landscape was better understood from both experience and acquired GPS data. Animals captured closer to the study area edges were less likely to be resighted on camera sets (black bars below): �Although individuals with more GPS locations on-grid generally had higher resightability (red bars below), several marked bobcats were not resighted (black bars below), indicating that camera locations were potentially suboptimal: Findings from this pilot year are driving changes and modifications to capture strategy and camera trap locations. Capturing animals more toward the center of the study area and relocating camera locations toward ‘better’ bobcat microsites and possibly increasing scent lure refresh rates will yield higher resight rates and improve population density estimates for the two study areas. Moreover, increased sample size of marked individuals and resight rates are required to estimate sex ratio and to make harvestrelated associations with density within the study areas. PERSONNEL Mark Vieira Chuck Anderson Shane Frank Megan Sims CPW, Carnivore and Furbearer Program Coordinator CPW, Mammals Research Leader CPW, Wildlife Research Scientist CPW, Federal Aid Coordinator 970-472-4368 970-472-4335 970-646-2961 303-291-7622 � https://cpw.cvlcollections.org/files/original/1999fe70189d36896db00a004b3c7c06.pdf f1f28f515bd1396a565784d5645b54f1 PDF Text Text Colorado Parks and Wildlife July 2023 – June 2024 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Federal Aid Project No. Colorado 3430 3003 : Parks and Wildlife (CPW) : Mammals Research : Bobcat Population Density and Demographics 1 : Evaluation of bobcat population density and demographics across habitat types and harvest levels in Colorado W-244-R, Segment 8 : Period Covered: July 1, 2023 – June 30, 2024 Authors: S.C. Frank, J. Ivan, M. Vieira, J. Runge All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. ABSTRACT Bobcat (Lynx rufus) populations in Colorado are available for legal harvest each year. Data on bobcat population densities in Colorado are sparse, and Colorado Parks and Wildlife has assumed a bobcat density of 15 bobcats/100 km2 within bobcat core habitat. Colorado constituents have indicated that they desire a more robust understanding of the population-level impacts of harvest on bobcat populations. This study will address that demand by quantifying bobcat density and bobcat prey availability, particularly as it relates to human harvest. From July 1, 2023 – June 30, 2024, we focused on working with landowners and stakeholders to ensure continuation of the project and to enhance our abilities to maintain camera grids and capture bobcats. Field efforts were centered on several objectives: 1) set up the remaining 70% of cameras in the low harvest legacy study area 2) continue capturing bobcats, 3) estimate bobcat population density for both study areas, 4) sample prey species tissue (roadkill) for bobcat dietary analyses, and 5) conduct line transect of observing leporids and their tracks for distance sampling. We set up the remaining cameras, pulled/replaced SD cards from all cameras, and maintained the camera grid throughout the season. We captured 16 bobcats and (re)collared 14 and collected diet samples and morphometric data. A preliminary mark-resight population estimate was possible for the high harvest legacy study area from previous data collection (20222023), but photo image processing is in progress for 2023-2024 population density estimation for both study areas from data collected through April 2024. About 25% of the 800,000 images 14 �collected have been photo identified. Preliminary results on dietary analysis using stable isotopes are promising and more samples are required. We conducted ~50% (~100 km) of the target transect length (200 km) for distance sampling of leporids, but there were not enough live observations (N = 6) to fit a model. Pellet plots will be used as an index for leporid abundance in the future. EVALUATION OF BOBCAT POPULATION DENSITY AND DEMOGRAPHICS ACROSS HABITAT TYPES AND HARVEST LEVELS IN COLORADO SHANE C. FRANK, JAKE IVAN, MARK VIEIRA, JON RUNGE PROJECT NARRATIVE OBJECTIVES A background objective of this project is to set up two bobcat study areas similar in habitat type and topography, but differing in historical bobcat harvest legacy, i.e., one with low and another with high historical bobcat harvest. Within each of these study areas, objectives are to (1) estimate bobcat population density for before and after legal harvest, (2) estimate bobcat survival and cause-specific mortality for the study population(s), (3) determine bobcat sex-age composition of the study population(s), (4) determine bobcat diet for the study population(s), (5) estimate leporid (relative) prey abundance, (6) explore associations between both harvest demographics and leporid abundance with bobcat density to identify potential indicators of population change, and (6) extrapolate density from study areas to other comparable habitat/vegetation cover in Colorado. SEGMENT OBJECTIVES (1) Work with Area 6 personnel and maintain positive relationships with landowners, particularly in the Skull Creek study area where 25%+ of the land is private to ensure adequate access for bobcat captures and maintaining camera grids. (2) Set up the remaining 70 camera sets in the Skull Creek study area, pull SD cards from all of the previously deployed cameras (N = ~130) and move locations as necessary to access bobcat travel routes. (3) Capture, collar/mark, sample, measure bobcats following the ACUC capture/handling protocol and monitor/investigate cause-specific mortalities of the marked population. (4) Estimate population density for both Skull Creek/low harvest legacy and Piceance/high harvest legacy study areas. (5) Sample roadkill or potential bobcat prey for muscle tissue that will be applied to stable isotope analyses and inference of bobcat diet composition. (6) Perform line transects and document both live and track observations. 14 �INTRODUCTION Bobcat Management in Colorado Bobcat (Lynx rufus) populations in Colorado are available for legal harvest each year from December 1 until the end of February. As with all furbearer species that Colorado Parks and Wildlife (CPW) manages for sustainable harvest, management decisions are informed using the best available data. Currently, bobcats are harvested in Colorado without bag limits. However, data on bobcat population densities in Colorado are sparse and limited to point estimates derived during winter from 2 small (80 km2) study areas near the Uncompahgre Plateau and along the Front Range (Boulder) in 2009 and 2010, respectively (Lewis et al. 2015). Density estimates from those study areas ranged from 16-24 bobcats/100 km2. Beyond those temporally and geographically isolated estimates, population density estimates are not available for Colorado. In lieu of that, Colorado bobcat density is inferred from estimates in other States (California: 115153/100 km2, Lembeck 1978, Lembeck and Gould 1979; Oregon: 77/100 km2, Witmer and DeCalesta 1986; Arizona: 24-28/100 km2, Jones and Smith 1979, Lawhead 1984; Utah: 6.2/100 km2, Karpowitz 1981; Idaho: 5.4 adults/100 km2, Bailey 1974; Nevada: 20/100 km2, Golden 1982). Compared to these other study locations, bobcat habitat quality in Colorado is considered moderate. Colorado Parks and Wildlife has assumed a bobcat density of 15 bobcats/100 km2 within bobcat core habitat, i.e. constrained to <9,500 feet elevation and within 7 km distance to woodland and shrubland vegetation types (Figure 1a). This assumed density, when applied to the bobcat management area level (Fig. 1b) is most likely a conservative range, as it falls slightly below Lewis et al.’s (2015) estimates and is well under estimates from most other western States. Bobcat harvest management in Colorado is partially based on a 7-year study conducted in Idaho that examined how bobcat populations responded to exploitation (Knick 1987, 1990). Knick (1987, 1990) found that annual harvest had little impact on bobcat populations until it reached 20% of population size. Colorado Parks and Wildlife has not set a bag limit on bobcats, but considers a derived harvest rate acceptable statewide if it does not exceed 17% of assumed bobcat density (15/100 km2), i.e., 2.55 bobcats/100 km2 at the bobcat management area scale (Figure 1b). This threshold has not been exceeded in any of the bobcat management areas and ranged from 0.71 – 1.73 bobcat /100 km2 in 2016-2017 (Colorado Parks and Wildlife Furbearer Management Report). Furbearer harvest, including the legal regulated take of bobcat, has recently fallen under more intense scrutiny by some of CPW’s constituents. For example, continuation of harvest was challenged in recent citizen-proposed petitions to the Colorado Parks and Wildlife Commission (PWC), by Colorado Senate (Bill SB22-031), and more recently by a proposed ballot initiative that would ban harvest of bobcats and other wild felids. The basis for the past petitions and for the current ballot initiative include social, ethical and biological factors. Two criticisms of current bobcat management have focused on a lack of a robust population and density estimates for the State of Colorado and reliance on an inferred scientific basis for regulating harvest. More succinctly, CPW has not evaluated the relationships between bobcat density, demographics, and harvest specific to Colorado. These challenges to CPW’s bobcat management practices clearly indicate that constituents desire a more robust understanding of the population-level impacts of harvest on bobcat populations. This study will address that demand by quantifying bobcat density and bobcat prey availability, particularly as it relates to human harvest. 14 �STUDY AREAS Our selection of study areas was guided by several factors, including reported bobcat harvest from 2017-2019, key bobcat habitats of Colorado (State Wildlife Action Plan - SWAP 2015), topography (DEM), and logistical considerations. We identified study areas of 20 x 20 km2 with one of high (2.55-5 bobcats harvested annually/100 km2) and the other low historical (2017-2019) bobcat harvest (0-1 bobcats harvested annually/100 km2). Aside from historical bobcat harvest rates, the two study areas were chosen to be as similar as possible across other environmental factors. Bobcats select rocky outcrops, areas with access to water, and prefer cover and edges (Armstrong et al. 2011), all of which the interface between pinyon-juniper and sagebrush communities provide. Pinyon-juniper (Pinus edulis and Juniperus spp.) and sagebrush (Artemisia spp.) vegetation communities comprise approximately 90% of each study area. Pinyon-juniper (Pinus edulis-Juniperus spp.) communities are regarded as preferred bobcat habitat (Armstrong et al. 2011). In Colorado, pinyon-juniper is often mixed with sagebrush communities; the chosen study areas had a predominant mixture of both pinyon-juniper (4470%) and sagebrush (23-42%) habitat types. The high (south; ‘Piceance’) and low harvest (north; ‘Skull Creek’) study areas (Figure 2) fall within CPW game management units 22 and 10, respectively. Elevation ranges from ~5500 to 8600 ft in Skull Creek and from ~5900 to 7800 ft in the Piceance study area. Nearly 25% of Skull Creek is private land; ~5% private land exists in Piceance. The majority of land in each study area is managed by the Bureau of Land Management, with Skull Creek containing approximately 40 km2 of designated wilderness. Piceance has better road and trail access than Skull Creek. This project requires cooperation and coordination with CPW Biologists, Area Wildlife Managers, and District Wildlife Managers primarily in the northwest part of the state, i.e., Area 6, but also statewide for bobcat teeth and carcass collection from donated bobcat carcasses. Additionally, work occurs on lands managed by BLM, and coordination is ongoing with appropriate Wildlife Program Leads. METHODS Bobcat Live Captures Bobcats will be captured using box traps, and trapping will occur from September-April. The use of dogs and leg-hold traps may be utilized if live cage trapping success is insufficient. Capture methods including hound pursuit and leg-hold traps will follow CPW’s ACUC approved capture guidelines for felids (ACUC protocol #HG-001, Mountain Lion/Canada Lynx/Bobcat Capture and Handling Guidelines, 2015 update). We will set out bait and traps near travel routes and landscape features where bobcats often occur. We will bait, chemical, and visual attractants to lure bobcats to trapping sites and noted on camera, where traps will then be armed. Box traps will be our principal method of capture. We perform daily trap checks (~within 10a-12p) a maximum time-in-trap of 26 hours. We set traps to provide trapped animals with adequate protection from the elements and potential food until they can be released. Trapping is halted and cages locked closed in the case of inclement weather. Trapped bobcats will be chemically immobilized, weighed, and then marked with eartags, pit tags, and fitted with GPS collar (see Marking on capture form in Appendix II), not to exceed 5% of the animals body weight. GPS collars are programmed to record a position and cycle throughout the day. Collars are scheduled to drop off the animal after 104 weeks through an electronic release on the collar. Bobcats will be sexed, aged, measured (head, body, teeth, paws), and their reproductive status will be assessed 14 �(developed nipples, testicles present); samples will be taken (tissue biopsy, blood, hair, whisker – cut not pulled); and vitals (respiration, heart rate, temperature) and reflexes will be monitored at regular intervals throughout capture and handling (Appendix II) prior to receiving a reversal drug and being released. Bobcat Camera Deployment for Population Density Estimation via Traditional Mark-Resight In each study area (see “Location” below), an array of 100 Reconyx HP2X camera traps were deployed and maintained active year-round. Each 400 km2 study area was divided into 100 cells (2 x 2 km size), with each cell receiving one camera placed >200 m from cell boundaries and within favorable bobcat microhabitat (game trail, ridges, washes, rocky outcrops) to detect bobcats. Resightings of individually-identified bobcats and sightings of unmarked bobcats will be jointly used to estimate population density. Bobcats detected from cameras will be individually identified through a combination of eartag ID, collar ID, and potentially individual pelage markings (Heilbrun et al. 2003). Camera sets will have scent lure and visual attractants (e.g. flagging, feathers) to attract bobcats toward camera sets/viewsheds for image capture. Bobcat Cause-Specific Mortality and Survival Bobcat mortality will be monitored via GPS-collared animals as part of this study. A ‘mortality signal’ that is automatically sent via sms and email when collar-based accelerometer data suggest no movement has occurred for at least 8 consecutive hours. If a mortality signal is triggered, the VHF beacon will be used to locate the collar and the animal (carcass) if present. If the animal is dead, then a basic, in-field necropsy will be performed to ascertain cause-of-death and compile environmental information. Information on location, animal body condition, ectoparasite load, trauma (e.g. gunshot, bludgeon/road strike, predation), and other information specific to predation will be collected for assignment of mortality cause. If disease is suspected, e.g., no obvious trauma, then the carcass will be collected and sent to the CPW Health Lab for necropsy. Survival and cause-specific mortality will be estimated using a hierarchical multi-state model (Kéry and Schaub 2012). Most bobcats (Nexpected = ~120 over the course of the 5 year study) will be telemetered and monitored for approximately two years, but some dead, marked individuals can be recovered after telemetry equipment has been removed or fails (via eartags and/or pit tags). Due to mandatory seal checks, bobcat mortalities from legal harvest offer ‘perfect detection’ with or without GPS collars on the animal, whereas other sources of mortality (e.g. predation and disease) are not likely be detected unless occurring on telemetered animals. Recovery rate of individual fates is therefore different, depending on mortality source and telemetry performance, and if ignored will produce biased estimates of survival. We plan to use a combination of capture-recapture data (live trapping), telemetry (GPS positional ‘recaptures’), and dead recovery information from animals to create capture histories and fit a joint capturerecapture mark-recovery model, as depicted in Kéry and Schaub (2012). Bobcat Sex and Age Bobcat sex and age information are collected by CPW staff for both marked (via capture) and unmarked bobcats. Sex and age for unmarked bobcats are collected during mandatory seal checks at CPW field offices during and shortly following the end of the bobcat harvest season in February. Whether marked or unmarked, bobcat ages fall into either juvenile or adult classes. For 14 �the unmarked bobcats, harvest information data requests will be submitted to the Terrestrial Section at the end of each harvest season, following ample time to receive and process bobcat seal check information from each Area. Sex ratio of the study population can be determined in several ways, but challenging factors for bobcats include (1) the lack of sexually distinguishing characters that are observable on camera and (2) the high likelihood of sex-dependent probability of capture and resight rates. The lack of camera-based sexual dimorphism precludes the possibility of estimating sex ratio from unmarked animals alone. In this project, sex-specific population density estimates that account for sexspecific probabilities of detection/resight/recapture are the preferred method to derive sex ratios, but this is contingent upon having a representative amount of marked resights for both sexes and for all study areas. Alternatively, sex-specific survival rates can be used to derive sex ratios (Ancona et al. 2017), which will be a secondary approach, using the multistate model above. This approach is limited, however, as it relies on assumptions that are rarely met in wild populations: harvest probabilities of different age classes are equal, age structure remains constant over time, and recruitment rates do not differ between males and females (Ancona et al. 2017). Raw counts of harvest are generally avoided, but dead, recovered ‘recaptures’ of marked bobcats can be used to derive sex ratios, assuming that harvest is not sex-biased. This is not likely, particularly for trapping, where larger male pelts are generally preferred and selected, and females are sometimes released to enhance/sustain reproduction and population persistence. In other harvested bobcat populations, however, hunting is more selective of larger males compared to trapping (Allen et al. 2018). A review of Colorado’s harvest data suggests that hunting-based legal harvest methods for bobcats and not trapping per se is less biased and could be used as a third approach, but the magnitude of sex-bias by harvest type has not verified. Given the data, all possible approaches will be compared for corroboration of derived sex ratios, but we realize that none may be a rigorous estimate of the actual sex ratio in the study population. We may consider a non-invasive sample of collected bobcat scat using scent dogs and subsequent genetic analysis to identify sex, but this would require additional and/or reprioritized project funds. Bobcat Diet Bobcat diet will be assessed via stable isotope ratio (δ13C and δ15 N) analysis, i.e., information derived from GPS-marked bobcat samples, such as blood, hair, and whiskers, in combination with tissue samples taken from taxa that are potential prey. Tissue samples from prey taxa will be collected primarily via roadkill and by other means (e.g. grouse wings already collected by Area 6). Individual bobcat GPS locations will be analyzed respective to the same individual’s stable isotopic signatures (consumed prey), which will give insight into how bobcats use the landscape for prey acquisition. The availability of prey helps provide context for bobcat resource acquisition and selection, in addition to potentially helping explain spatial and temporal variation in population density (see Leporid Line Transects). The bobcat dietary work is a focus of a Masters project in cooperation with the University of Wyoming (UWyo). A graduate student will be co-supervised by the bobcat project leader, Shane Frank, and Assistant Professor Joe Holbrook (UWyo), and she will further develop the scope and methods addressing bobcat diet nearer to the start of her project in spring 2024. Leporid Line Transects and Distance Sampling We attempted to estimate leporid abundance in the study areas using distance sampling. Distance sampling allows for the estimation of animal abundance using the probability of 14 �detection from line transects, which decays as distance increases from the line. Crucial assumptions are that (1) there is 100% detection on the line (0m from the line) and (2) objects do not move (or are detected multiple times from the line). Leporids can flush, but they typically do so within close proximity, meaning the identification of original location is likely known, especially in snow. The number of detections per kilometer determines the sample size required to achieve desired precision for an abundance estimate (Buckland et al. 2015) . Leporids can range from as low as 0 animals to as high as 17 animals per kilometer transect (Fernandez-deSimon et al. 2011, van Strien et al. 2011, Robinson et al. 2014). We calculated optimal CVs by varying the number of animals detected/km and transect length and fixing variation (b) of animals detected per km (b = 3; Burnham et al. 1980) (Figure 6). RESULTS AND DISCUSSION Pilot Study Results Analyzed (November 2022 – April 2023) Between November 2022 and April 2023, CPW conducted a pilot study in two study areas. Due to stochastic and severe winter conditions in both study areas, particularly deep and persistent snow, the camera trap array in Piceance was fully deployed but only one third of the cameras were deployed in Skull Creek (Fig. 2, gray dots). Due to limited and remote access, capture efforts were similarly limited in Skull Creek. These conditions led to a higher level of capture effort and success in Piceance, (Fig. 1, more orange and blue dots in Piceance). Thirtynine bobcats were captured, consisting of 26 newly caught individuals and 13 recaptures. Seventeen bobcats were collared and ‘available’ for subsequent resighting in Piceance, whereas four were available in Skull Creek. The sex ratio (male:female) of captured and collared bobcats was male-skewed about 3:1 (n = 21) in Piceance, but female-skewed with 2:3 (n = 5) in Skull Creek. Most captured individuals were adults (n = 20 of 26), and most captured sub-adults (n = 5 of 6) were too small to collar safely, i.e., ≤15 lbs. (Table 1). There were no recorded mortalities of GPS-collared bobcats outside of legal harvest during this period. Bobcat density estimation was not possible for Skull Creek, due to small sample size (n = 4) and incomplete camera deployment (n = 37), and SD cards were not collected in April following two months of data collection; field efforts were prioritized toward Piceance. There were 947 bobcat images from Piceance camera traps, yielding 119 independent detections. Most detections were of unmarked individuals (n = 93), 8 were of marked individuals, and 17 were unknown whether marked or unmarked bobcats. Six of the 17 (35%) of the marked, collared bobcats were detected. We estimated the Piceance bobcat population density via a mark-resight immigration-emigration mixed logit-normal model (addressing lack of geographic closure). Bobcat density was estimated for the period from November 19, 2022 through April 23, 2023, corresponding to trap set and retrieval dates. Demographic closure was violated during this period, due to legal harvest (n = 8 with 2/8 marked), so the derived estimate corresponds to density at the beginning of the sampling period (McClintock 2014). A preliminary bobcat density of 34.62 bobcats (CV = 0.38, SE = 12.80) was estimated for the 400 km2 Piceance study area, i.e., 8.41 bobcats/100 km2. We suspect that the Piceance density estimate is biased low, due to many marked animals occurring off-grid (spatially biased trapping effort near study area borders), reduced movement during a severe winter, and due to suboptimal camera locations (low number of both marked/unmarked bobcat resights). Camera locations were primarily chosen to represent the dominant habitat type in a given cell and secondarily for anticipated preferred travel routes/locations to enhance bobcat detection. In several cases, however, lower 14 �represented habitats or edge habitats likely provided better locations for detecting bobcats in a camera grid cell and our understanding of bobcat spatial behavior improved from both experience and acquired GPS data. Bobcat captures and camera images (October 2023 – April 2024) We captured 16 bobcats (9 new and 7 recaptures) with 22 currently collared bobcats following harvest and previous captures/collaring. Trap effort was higher this season (~2700 trap nights vs ~1700 trap nights), but success was much less. We suspect that the mild onset of the winter, i.e., very little snow, allowed bobcats to take advantage of prey that is usually unavailable under snow, e.g., small rodents. We also suspect that bobcat movement was reduced during this period as well, as they were not as present at bait sites nor were interested in our roadkill baits like the previous year. For this reason, we implemented the use of hounds in a limited fashion. We released hounds on four occasions and had a successful (re)capture as a result in one instance. Given the proof of concept and in practice, we aim to utilize this capture method in the future to help increase capture effort and success. We successfully set up the rest of the camera grid in Skull Creek and collected SD cards from cameras that recorded images from October 2023 until April 2024 for both study areas. We have collected ~800,000 camera images for this period and have photo identified ~25% of them. A population density cannot be estimated from (April 2023-April 2024) until all bobcat images have been identified. We have collected ~24,000 GPS locations. Bobcat mortality rates are acquired from both GPS-collared and ear-tagged bobcats (N = 31; N = 5 ear-tag only) and cause-specific mortality will be estimated at the end of the 5-year study to maximize sample size or when study areas change and sample sizes will no longer increase. Altogether, we have newly captured and marked 26 males and 11 females, but only 31 have had collars deployed, due to size constraints. We have not estimated the sexspecific probabilities of detection on camera for each of these cohorts yet, as estimation depends on individual bobcat ID in camera images (in progress) and mark-resight density estimation. Leporid Line Transects (October 2023 – April 2024) We conducted ~50% of our targeted 200 km total transect length for leporid live observations and track counts. For each low and high harvest legacy study areas, we walked ~50 km length of transects. We only observed two jackrabbits for low harvest legacy and four cottontails for the high harvest legacy study area. Similarly, we observed more leporid tracks crossing the transect line(s) within the high harvest legacy study area (N = 471) compared to the low harvest legacy area (N = 186). Live observations of leporids was too low to fit a distance sampling model for abundance estimation and the field effort was substantial. We have therefore opted to use pellet plots around camera sets in the future to corroborate the use of leporid detection rates on cameras as an index of relative leporid abundance within and across study areas. Bobcat Diet In addition, blood, hair, and whisker samples have been taken from bobcats during capture, and roadkill tissue samples of potential bobcat prey have been collected for stable isotopic analysis. Preliminary analysis shows that bobcat inert hair and whisker tissues do differentiate from prey and specific prey taxa stable isotopes tend to cluster (Figure 3; e.g. ground squirrel/marmotini and mule deer/Odocoileus hemionus), but we need more prey samples to properly run mixture models and ‘assign’ dietary preference to individual bobcats. 14 �SUMMARY From July 1, 2023 – June 30, 2024 we successfully worked with private landowners and personnel from CPW to coordinate field research logistics and initiate the 2nd year of this study. We captured 16 bobcats, of which we newly collared 9 and replaced 4 collars; bobcats were otherwise too small for collars. We collected bobcat GPS locations, morphometrics, and took biological samples, including information on bobcat diet. Capture sample size objectives were likely not met this year, due to weather effects on capture success, but population density estimation relies on camera resight or detection rate as well, which is currently being estimated. Previous capture efforts and success might render a population density estimate possible. We have ~10 collared bobcats in each study area, which is ~the minimum number required. Despite the winter weather challenge in captures, it provides an opportunity to contrast bobcat diet between two capture seasons. We discovered that the transect method was not effective for population estimation of leporids using distance sampling, because we did not register enough live observations. Pellet plots will be used in the future to correlate it as a leporid abundance index with camera detection rates. We successfully set up the remaining camera sets in the Skull Creek study area and improved camera locations and successfully maintained both study area camera grids, enhancing our ability to estimate population size in subsequent periods. 14 �LITERATURE CITED Allen, M. L., N. M. Roberts, and T. R. Van Deelen. 2018. Hunter selection for larger and older male bobcats affects annual harvest demography. R Soc Open Sci 5:180668. Ancona, S., F. V. Denes, O. Kruger, T. Szekely, and S. R. Beissinger. 2017. Estimating adult sex ratios in nature. Philos Trans R Soc Lond B Biol Sci 372. Armstrong, D. M., J. P. Fitzgerald, and C. A. Meaney. 2011. Mammals of Colorado, Second Edition. University Press of Colorado. Bailey, T. N. 1974. Social Organization in a Bobcat Population. The Journal of Wildlife Management 38. Buckland, S. T., T. A. Marques, C. S. Oedekoven, and E. A. Rexstad. 2015. Distance Sampling: Methods and Applications. Pages 1 online resource (XV, 277 pages 280 illustrations, 224 illustrations in color in Methods in Statistical Ecology,. Springer International Publishing : Imprint: Springer,, Cham. Burnham, K. P., D. R. Anderson, and J. L. Laake. 1980. Estimation of density from line transect sampling of biological populations. Biometrical Journal 24. Fernandez-de-Simon, J., F. Díaz-Ruiz, F. Cirilli, F. S. Tortosa, R. Villafuerte, M. DelibesMateos, and P. Ferreras. 2011. Towards a standardized index of European rabbit abundance in Iberian Mediterranean habitats. European Journal of Wildlife Research 57:1091-1100. Golden, H. 1982. Bobcat populations and environmental relationships in northwestern Nevada. University of Nevada, Reno. Heilbrun, R. D., N. J. Silvy, M. E. Tewes, and M. J. Peterson. 2003. Using automatically triggered cameras to individually identify bobcats. Wildlife Society Bulletin 31:748-755. Jones, J. H., and N. S. Smith. 1979. Bobcat Density and Prey Selection in Central Arizona. The Journal of Wildlife Management 43. Karpowitz, J. T. 1981. Home ranges and movements of Utah bobcats with reference to habitat selection and prey base., Brigham Young University. Kéry, M., and M. Schaub. 2012. Bayesian population analysis using WinBUGS : a hierarchical perspective. 1st edition. Academic Press, Boston. Knick, S. T. 1987. Ecology of bobcats in southeastern Idaho. University of Montana. Knick, S. T. 1990. Ecology of Bobcats Relative to Exploitation and a Prey Decline in Southeastern Idaho. Wildlife Monographs 108:1-42. Lawhead, D. N. 1984. Bobcat Lynx rufus Home Range, Density and Habitat Preference in South-Central Arizona. The Southwestern Naturalist 29. Lembeck, M. 1978. Bobcat Study, San Diego County, California. State of California, The Resources Agencey, Department Fish and Game. Lembeck, M., and J. Gould, G.I. Dynamics of harvested and unharvested bobcat populations in California. National Wildlife Federation Science and Technology, 1979. Lewis, J. S., K. A. Logan, M. W. Alldredge, L. L. Bailey, S. VandeWoude, and K. R. Crooks. 2015. The effects of urbanization on population density, occupancy, and detection probability of wild felids. Ecol Appl 25:1880-1895. McClintock, B. T. 2014. Mark-resight models. in E. Cooch, and G. White, editors. Program MARK: a gentle introduction. Robinson, Q. H., D. Bustos, and G. W. Roemer. 2014. The application of occupancy modeling to evaluate intraguild predation in a model carnivore system. Ecology 95:3112-3123. 14 �SWAP, S. W. A. P.-. 2015. Colorado Parks and Wildlife. van Strien, A. J., J. J. A. Dekker, M. Straver, T. van der Meij, L. L. Soldaat, A. Ehrenburg, and E. van Loon. 2011. Occupancy dynamics of wild rabbits (Oryctolagus cuniculus) in the coastal dunes of the Netherlands with imperfect detection. Wildlife Research 38. Witmer, G. W., and D. S. deCalesta. 1986. Resource use by unexploited sympatric bobcats and coyotes in Oregon. Canadian Journal of Zoology 64:2333-2338. Prepared by Shane C. Frank, Wildlife Researcher FIGURES Figure 1. Bobcat Core Habitat (1a) and Bobcat Management Areas (1b). Colorado Parks and Wildlife’s core habitat model and map (1a) was developed to represent bobcat habitat (lavender) within the state. While bobcats may occur anywhere in the state, the core habitat model was developed to conservatively represent essential bobcat habitat. Core habitat was constrained to less than 9,500 feet elevation and to woodland and shrubland vegetation types identified in CPW’s Basinwide vegetation layer. Vegetation classifications were buffered to approximately 7 km distance in order to smooth boundaries. Colorado Parks and Wildlife’s geographical bobcat management structure is reflected by figure 1b. Bobcat management areas are depicted by uniformly colored polygons, CPW’s Management Regions are fall within the yellow borders, and CPW’s Game Management Units are reflected by black borders and numbered. 14 �Figure 2. Map depicting the Skull Creek or low harvest legacy study area (GMU 10) and Piceance or high harvest legacy study area (GMU 22) in northwest Colorado. Each study area is a 10 x 10 grid of 100, 2 x 2 km2 cells. Each 400 km2 grid has one camera per cell (not shown), which are used as detectors of collared/bobcats for mark-resight population estimation. Orange and blue dots signify trap sites for the 2022-2023 and 2023-2024 capture seasons, respectively. Red dots are successful capture sites, some of which repeated across seasons. Figure 3. Delta 15 Nitrogen (y-axis) and 13 Carbon stable isotope (x-axis) amounts for bobcat 14 �whisker, guard hair, and potential prey muscle tissues. Circles of the same color close together indicate clustering of prey items and implies differentiability from other prey items if there is separation between clusters. More prey tissue samples and stable isotope values will help create precise “envelopes” for prey differentiation. Bobcat guard hair (green open circles) and whiskers (lavender open circles) depict variation among individual bobcat diet and season (not shown here), indicating that bobcats might show variation in prey selection and use. 14 � https://cpw.cvlcollections.org/files/original/b73314202416fc330b7a222a396bb8b8.pdf 4f16bc31d3b82205b3c3f9be64241c7f https://cpw.cvlcollections.org/files/original/d6480d79932cc2e2ecc973a001523f12.pdf 7c15ccb17fd05d852e124e10f748145e PDF Text Text ÿ ÿ ÿ ÿ 23435673ÿ965 ÿ67ÿ �474���ÿ ��������ÿ��������ÿ������ ÿ�!""��#ÿ ÿ �$%$&$ÿ�'%(ÿ")%*+),*%-ÿ*%ÿ�).),$&)ÿ ÿ /01234ÿ63701048ÿÿÿ9:;<ÿ=>ÿ?@=AB9:C0ÿD@>ÿ?@=Eÿ ÿ /12CF2GH;ÿIC70JK2LHK31J8ÿÿÿM12FÿN40;;>ÿM12FON40;;PJKHK0OF3O:JQÿ9HR0ÿI7HC>ÿ9HR0OI7HCPJKHK0OF3O:JQÿSF3KKÿ TH2K>ÿSF3KKOTH2KPJKHK0OF3O:Jÿÿÿ ÿ U44ÿ��35V6W�3ÿ�ÿWX�ÿY53Z�[Wÿ\VV65]ÿ�ÿY5�4�V�65]ÿ67ÿ\^Z�[WÿW3ÿ�\5WX�5ÿ�_64\6W�3 ÿ̀ÿ a�35V6W�3ÿbUcÿdefÿghÿ9igjaklhmÿenÿoiefhmÿp�WX3\WÿY�5V��3ÿ3�ÿWX�ÿY5�[�Y64ÿ �_�W�q6W35ÿ̀ÿb6�Y\46W�3ÿ3�ÿWX��ÿ76W6ÿ^�]37ÿWX6Wÿ[3W6��7ÿ�ÿWX�ÿ\VV65]ÿ�ÿ7�[3\56q�7ÿ̀ÿÿ ÿ rsÿtsÿuvvwxyÿywÿxuzywxuÿtÿ{|t}~uÿw€~ty|wsÿwvÿtst‚tÿ~ƒs„ÿ†‡ˆ‰ÿ‹ŒˆŒŽˆ‘ÿywÿy’uÿzw€y’uxsÿ wxy|wsÿwvÿy’u|xÿvwx“uxÿxts”u•ÿ–—˜ÿ|s‚|{|‚€t~zÿ™uxuÿxu|syxw‚€šu‚ÿ|sywÿw~wxt‚wÿvxw“ÿ—›››B–œœžÿÿrsÿ–œ—œ•ÿ y’uÿw~wxt‚wÿŸ|{|z|wsÿwvÿ |~‚~|vuÿsw™ÿw~wxt‚wÿ¡tx¢zÿts‚ÿ |~‚~|vuÿ£¡ ¤‘ÿ‚uyux“|su‚ÿy’tyÿy’uÿ xu|syxw‚€šy|wsÿuvvwxyÿ“uyÿt~~ÿ}usš’“tx¢zÿwvÿz€ššuzz•ÿts‚ÿy’tyÿy’uÿw€~ty|wsÿwvÿtst‚tÿ~ƒs„ÿ|sÿy’uÿzytyuÿ ™tzÿttxusy~ƒÿ{|t}~uÿts‚ÿzu~v¥z€zyt|s|s”žÿÿrsÿwx‚uxÿywÿyxtš¢ÿy’uÿuxz|zyusšuÿwvÿy’|zÿsu™ÿw€~ty|wsÿts‚ÿ y’€zÿ‚uyux“|suÿy’uÿ~ws”¥yux“ÿz€ššuzzÿwvÿy’uÿxu|syxw‚€šy|ws•ÿtÿ“|s|“t~~ƒ¥|s{tz|{u•ÿzytyu™|‚uÿ“ws|ywx|s”ÿ xw”xt“ÿ|zÿxu¦€|xu‚žÿÿŸ€x|s”ÿ–œ—§B–œ—˜ÿ¡ ÿ|s|y|tyu‚ÿtÿwxy|wsÿwvÿy’uÿzytyu™|‚uÿ“ws|ywx|s”ÿzš’u“uÿ ‚uzšx|}u‚ÿ|sÿr{tsÿ–œ—¨‘ÿ}ƒÿšw“~uy|s”ÿz€x{uƒzÿ|sÿtÿxts‚w“ÿzt“~uÿwvÿ“ws|ywx|s”ÿ€s|yzÿˆÿ©ÿªœ‘ÿÿvxw“ÿ y’uÿ«tsÿ¬€tsÿ­w€syt|szÿ|sÿzw€y’™uzyÿw~wxt‚wÿˆÿ©ÿ—1›ÿywyt~ÿ€s|yz®ÿ̄|”€xuÿ—‘žÿÿ Ÿ€x|s”ÿ–œ—1B–œ—˜•ÿuxzwssu~ÿvxw“ÿ¡ ÿts‚ÿ°ž«žÿ̄wxuzyÿ«ux{|šuÿ°«¯«‘ÿšw“~uyu‚ÿy’uÿvw€xy’ÿ ƒutxÿwvÿ“ws|ywx|s”ÿ™wx¢ÿwsÿy’|zÿzt“uÿzt“~užÿÿ«uš|v|št~~ƒ•ÿ—§ÿ€s|yzÿ™uxuÿzt“~u‚ÿ{|tÿzsw™ÿyxtš¢|s”ÿ z€x{uƒzÿšws‚€šyu‚ÿ}uy™uusÿŸušu“}uxÿ—ÿts‚ÿ­txš’ÿ̈—žÿÿ±sÿutš’ÿwvÿ—B¨ÿ|s‚uus‚usyÿwšštz|wsz•ÿz€x{uƒÿ šxu™zÿzutxš’u‚ÿxwt‚™tƒzÿt{u‚ÿxwt‚zÿts‚ÿ~w””|s”ÿxwt‚z‘ÿts‚ÿyxt|~zÿvwxÿ~ƒs„ÿyxtš¢zžÿÿxu™zÿzutxš’u‚ÿy’uÿ “t„|“€“ÿ~|sutxÿ‚|zytsšuÿwvÿxwt‚zÿwzz|}~uÿ™|y’|sÿutš’ÿz€x{uƒÿ€s|yÿ”|{usÿztvuyƒÿts‚ÿ~w”|zy|št~ÿšwszyxt|syzžÿÿ ²tš’ÿz€x{uƒÿšw{uxu‚ÿtÿ“|s|“€“ÿwvÿ—œÿ~|sutxÿ¢|~w“uyuxzÿž–ÿ“|~uz‘ÿ‚|zyx|}€yu‚ÿtšxwzzÿtyÿ~utzyÿ–ÿ¦€t‚xtsyzÿ wvÿy’uÿ€s|yžÿÿ³sÿt‚‚|y|wst~ÿ€s|yÿ™tzÿzš’u‚€~u‚ÿvwxÿzsw™ÿyxtš¢ÿz€x{uƒzÿ}€yÿz€x{uƒzÿ™uxuÿswyÿšw“~uyu‚žÿ́’uÿ xu“t|s|s”ÿ̈ªÿ€s|yzÿšw€~‚ÿswyÿ}uÿz€x{uƒu‚ÿ{|tÿzsw™ÿyxtš¢|s”žÿÿrszyut‚•ÿz€x{uƒÿšxu™zÿ‚u~wƒu‚ÿ§ÿtzz|{uÿ |svxtxu‚ÿ“wy|wsÿšt“uxtzÿ|sÿutš’ÿwvÿy’uzuÿ€s|yzÿ‚€x|s”ÿvt~~ÿ–œ—1žÿÿt“uxtzÿ™uxuÿ}t|yu‚ÿ™|y’ÿ{|z€t~ÿ tyyxtšytsyzÿts‚ÿzšusyÿ~€xuÿywÿus’tsšuÿ‚uyušy|wsÿwvÿ~ƒs„ÿ~|{|s”ÿ|sÿy’uÿtxutžÿÿt“uxtzÿ™uxuÿxuyx|u{u‚ÿ‚€x|s”ÿ z€““uxÿ–œ—˜ÿts‚ÿt~~ÿ’wywzÿ™uxuÿtxš’|{u‚ÿts‚ÿ{|u™u‚ÿ}ƒÿtyÿ~utzyÿ–ÿw}zux{uxzÿywÿ‚uyux“|suÿzuš|uzÿ xuzusyÿ|sÿutš’žÿÿt“uxtÿ‚tytÿ™uxuÿy’usÿ}|ssu‚ÿz€š’ÿy’tyÿutš’ÿwvÿ—œÿ—ª¥‚tƒÿux|w‚zÿvxw“ÿŸušu“}uxÿ—ÿ Kµ13:Lµÿ¶G12;ÿD@ÿ·HJÿF3CJ240104ÿHCÿ̧3FFHJ23C>¹ÿHC4ÿHC<ÿGµ3K3ÿ3ºÿHÿ;<C»ÿ3¼KH2su‚ÿ‚€x|s”ÿtÿ—ª¥‚tƒÿux|w‚ÿ ·HJÿF3CJ240104ÿHÿ̧40K0FK23C¹ÿ4:12CLÿKµHKÿ3FFHJ23COÿÿÿÿÿÿ «€x{uƒwxzÿšw{uxu‚ÿª¨˜ÿ¢“ÿ‚€x|s”ÿzsw™ÿyxtš¢|s”ÿz€x{uƒzÿts‚ÿ‚uyušyu‚ÿ~ƒs„ÿtyÿ1ÿ€s|yzÿ´t}~uÿ—‘žÿÿ ´’|zÿxuxuzusyzÿtÿ§¥ƒutxÿ~w™ÿ|sÿzsw™ÿyxtš¢|s”ÿuvvwxyÿts‚ÿ|zÿtxy|t~~ƒÿ‚€uÿywÿz™|yš’|s”ÿ§ÿyxtš¢|s”ÿ€s|yzÿywÿ št“uxtÿ€s|yzÿz|sšuÿy’uÿ}u”|ss|s”ÿwvÿy’uÿxw½ušy•ÿts‚ÿtxy|t~~ƒÿ‚€uÿywÿ~tš¢ÿwvÿzsw™ÿ‚€x|s”ÿy’uÿ–œ—1B–œ—˜ÿ ™|syuxžÿÿ«€x{uƒwxzÿ™uxuÿws~ƒÿt}~uÿywÿšw“~uyuÿ̈ÿ{|z|yzÿywÿ̈¾ÿwvÿzsw™ÿyxtš¢|s”ÿ€s|yzÿ|sÿ–œ—1B–œ—˜•ÿ ™’uxutzÿ|sÿxu{|w€zÿƒutxz•ÿªœB1—¾ÿ™uxuÿz€x{uƒu‚ÿ̈ÿy|“uzžÿÿ́’uÿ“utsÿ‚|zytsšuÿz€x{uƒu‚ÿtzÿ™u~~ÿtzÿy’uÿ s€“}uxÿwvÿ€s|yzÿ™|y’ÿ~ƒs„ÿxu“t|su‚ÿz|“|~txÿywÿxu{|w€zÿƒutxzžÿÿ«€x{uƒwxzÿšw~~ušyu‚ÿ“wxuÿ’wywzÿ‚€x|s”ÿ –œ—1B–œ—˜ÿy’tsÿ|sÿtsƒÿwy’uxÿƒutx•ÿt”t|sÿ‚€uÿywÿz~w™~ƒÿxu~tš|s”ÿzsw™ÿyxtš¢|s”ÿ€s|yzÿ™|y’ÿšt“uxtÿ€s|yzÿ w{uxÿy’uÿƒutxzžÿÿ¿w™u{ux•ÿ™uÿšw~~ušyu‚ÿÀªœ¾ÿwvÿy’uÿ€z€t~ÿs€“}uxÿwvÿ~ƒs„ÿ’wywzÿts‚ÿy’uÿs€“}uxÿwvÿ€s|yzÿ ™|y’ÿ~ƒs„ÿšwsy|s€u‚ÿywÿ‚uš~|suÿ´t}~uÿ–‘žÿÿ¡ux’tzÿy’|zÿxuv~ušyzÿtÿxut~ÿ‚uš~|suÿ|sÿ~ƒs„ÿt}€s‚tsšuÿwxÿ ‚|zyx|}€y|ws•ÿwxÿux’tzÿy’uÿ~tš¢ÿwvÿzsw™ÿš’ts”u‚ÿ~ƒs„ÿ}u’t{|wxÿts‚ÿ’t}|ytyÿ€zu•ÿxuz€~y|s”ÿ|sÿvu™uxÿ’wywzžÿÿ Átš¢ÿwvÿzsw™ÿšw€~‚ÿ’t{uÿt~yuxu‚ÿw€xÿ‚uyušy|wsÿxw}t}|~|yƒÿtzÿ™u~~ÿšt“uxtzÿtxuÿwz|y|wsu‚ÿywÿtššw€syÿvwxÿ —B–ÿ“ÿwvÿzsw™•ÿ™’|š’ÿsu{uxÿtšš€“€~tyu‚‘žÿÿ³~yuxsty|{u~ƒ•ÿ~tš¢ÿwvÿ‚uyušy|wszÿšw€~‚ÿ’t{uÿ}uusÿ‚€uÿywÿy’uÿ su™ÿ~€xuÿÂ6H70C¹JÿÃ23;HK31ÿAQÿÄ2CC0J3KHÿÅ1HG;2C0ÿ/134:FKJ>ÿµKKGJ8ÆÆ···OÇ2CCK1HGG134OF3ÇÆÈ3¼FHK¥ts‚¥ 1ÿ ÿ �ÿ ÿ ÿ ÿ 23456789 �61��6�ÿ��ÿ��ÿ�4ÿ���������ÿ��!"ÿ#!ÿ$%"!ÿ&!ÿ%'!(ÿ)"!*+,%'$-ÿ./+0�%1�2ÿ3%!""!'ÿ4,"5!6'ÿ 2 8��7ÿ87966���:;�88��<98=�:9>6789 :787?�@789 A�B�ÿC�D>�ÿ4DED�;DC;�:ÿÿF4<98 4D�;37ÿ 8�ÿ8D4=��ÿ�4ÿ�49�ÿD4ÿ;8�ÿD8�ÿ94<94�7ÿ8�ÿ>DG�4=ÿ�ÿ�<<� ;ÿ9ÿ��8>�4�ÿ�8�8ÿ<D 98ÿ8��;�ÿ �4ÿ<���8ÿ�� �94�:ÿÿH�ÿ��;;ÿ��ÿ8�ÿIDE�46�ÿJ�9;D98ÿ�ÿ;8�ÿ�4ÿ���1�����7ÿ�8�87ÿ�<ÿD 9>7D4��ÿC3ÿDÿ 498>D;ÿ�49�<D;;7ÿ>D3ÿD;;9�ÿ�ÿ9ÿ8�89�7� �E�;3ÿD�����ÿ8�ÿD��ÿ9<ÿ8�ÿ;DGÿ9<ÿ�� �94�:ÿÿI9>7D8�ÿ9ÿ 78�E�9�ÿ3�D8�7ÿ��ÿ9CD�4�ÿ4��ÿ;345ÿ�� �94�ÿDÿDÿD>�8Dÿ4�ÿ4�D8ÿH9;<ÿI8��GÿKD��7ÿC ÿ<D�;�ÿ9ÿ �� ÿ;345ÿ�4ÿ8�ÿ77�8ÿH�;;�D>�ÿI8��Gÿ98ÿ�4ÿ8�ÿ2DÿLD8�DÿM94D�4�ÿ4988ÿ9<ÿI8���ÿNO�=8�ÿ��:ÿÿP4ÿ D ;ÿ<�>D;�ÿ��8ÿG��4�ÿ�D�ÿ�� �ÿDÿD>�8D�ÿ�4ÿDÿ4�ÿ4�D8ÿK;D989ÿQ���8E9�87ÿ8�ÿ9 >�4�4=ÿ8Dÿ Dÿ;�D�ÿ�9>�ÿ8�789 �94ÿ9 88�ÿ�4ÿ8�ÿ� 3ÿD8�D:ÿ H�ÿ��ÿ8�ÿQÿNQÿR�E�;97>�4ÿI98�ÿS�D>ÿ���1�ÿ)�T0�5!ÿUVW�"06ÿN2DDG�ÿ���1�ÿ9ÿ<�ÿ�D4D8ÿ 9 7D43ÿ>9�;�ÿNMDX�4Y��ÿ[\ÿ]^_ÿ���`�ÿ9ÿ98ÿ�8E�3ÿDDÿ��4=ÿ789=8D>ÿMPQXÿNH8��ÿD4ÿ a 848D>ÿ�����:ÿÿS8�7ÿ��ÿ���>D�ÿ8�ÿ789CDC�;�3ÿ9<ÿDÿ4�ÿC��4=ÿ9 7��ÿN�:�:7ÿ���ÿC3ÿ;345ÿ9E�8ÿ 8�ÿ98��ÿ9<ÿ8�ÿ��4�87ÿD;94=ÿ��8ÿ8�ÿ789CDC�;�3ÿ9<ÿ�� �4=ÿDÿ;345ÿNb�ÿ=�E�4ÿ8Dÿ8�ÿ4�ÿ�D�ÿ ,TT%)+!(cÿÿUd%"*!-ÿe!#,(6ÿ�f(ÿU-!�"6ÿ&!"!ÿ"!�!(ÿ�'ÿ5",%)ÿ*�"+�1$!'ÿ'9ÿ8Dÿ��ÿ9;7ÿCD��ÿ94ÿ 78�E�9�ÿ�98G7ÿ��ÿD;;9�ÿ�� �94ÿ789CDC�;�3ÿNb�ÿ9ÿED83ÿC3ÿ�8E�3ÿ>�897ÿ��ÿD;;9�ÿ<98ÿ�� �94ÿ 789CDC�;�3ÿ<98ÿ������1ÿ9ÿ�<<�8ÿ<89>ÿ98�8ÿ3�D8�ÿ �ÿ9ÿ8�ÿ;DGÿ9<ÿ�49�ÿ98ÿ4��ÿ;8�7ÿD4ÿg�ÿ�4; �ÿDÿ C8���4=ÿ��D�94ÿ�<<� ÿ<98ÿ�� �94ÿDÿD>�8D�ÿN;345ÿ�4ÿ9ÿ>9E�ÿ>98�ÿ�4ÿ;D�ÿ��4�8ÿ�8�4ÿ8�3ÿC�=�4ÿ9ÿ C8��7ÿD4ÿ8�ÿ�89;ÿ�494�8ÿD>�8D�ÿ>98�ÿ9<�4�:ÿÿH�ÿD;�9ÿ94���8�ÿDÿ���ÿ9<ÿ9ED8�D��ÿ8Dÿ 9;ÿ79�4�D;;3ÿ�57;D�4ÿED8�D�94ÿ�4ÿ9 7D43ÿNh�ÿ�4; �4=ÿ789798�94ÿ9<ÿ8�ÿ4�ÿ8Dÿ�D�ÿ9E�8�ÿC3ÿ �78 �6<�8ÿ<98��7ÿDE�8D=�ÿ3�D8�ÿ��4�ÿCD8GÿC��;�ÿ�4<��D�947ÿED8�DC�;�3ÿN�D4D8ÿ�E�D�94�ÿ�4ÿ3�D8�ÿ��4�ÿ CD8GÿC��;�ÿ�4<��D�947ÿ789798�94ÿ9<ÿ8�ÿ4�ÿ�>7D �ÿC3ÿCD8GÿC��;��7ÿ789798�94ÿ9<ÿ8�ÿ4�ÿ8Dÿ�D�ÿ C84�ÿ 8�4=ÿi>>�8ÿ���g7ÿD4ÿ8�ÿ4>C�8ÿ9<ÿ7899�ÿ9<ÿ98�8ÿ�7����ÿ8Dÿ9;ÿ79�4�D;;3ÿ�>7D ÿ 78���4�ÿ9<ÿ;345ÿN�:=:7ÿ�49��89�ÿ8D8��ÿD�ÿDÿ<99ÿ�98�7ÿ939��ÿD�ÿ79�4�D;ÿ9>7��98��:ÿÿH�ÿ;�>��ÿ 98ÿ>9�;ÿ��ÿC3ÿ<�8�ÿ���4=ÿDÿ=�4�8D;ÿ�8 8�ÿ<98ÿhÿ�8�;�ÿD������4=ÿ<�ÿ9<ÿED8�9�ÿ9>C�4D�94�ÿ9<ÿ ED8�DC;��ÿ�57� �ÿ9ÿD<<� ÿb:ÿÿH�ÿ8�4ÿ<�5�ÿ8�ÿC��j<��4=ÿ�8 8�ÿ<98ÿb7ÿD4ÿD������ÿ9>C�4D�94�ÿ9<ÿ 8�ÿ9ED8�D��ÿ�57� �ÿ9ÿ�4<;�4�ÿh7ÿD;;9��4=ÿ7ÿ9ÿ�ÿ9<ÿ8���ÿ9ED8�D��ÿDÿDÿ�>�7ÿ�4ÿD ��94ÿ9ÿ8�ÿ 9ED8�D��ÿ94ÿ�� �94:ÿÿH�ÿ�4; �ÿDDÿ<89>ÿ8�ÿ7�;9ÿ� 3ÿN����������ÿD�ÿ��;;ÿD�ÿ8�ÿ<�8�ÿkÿ3�D8�ÿ9<ÿ >94�98�4=ÿN���k����1�ÿ9ÿ>D5�>�Y�ÿ�8D8�4=ÿ9<ÿ�4<98>D�94ÿD89��ÿ�8E�3�:ÿÿÿ P�ÿ�4ÿ7D�ÿ3�D8�7ÿ8�ÿC��j<��4=ÿ>9�;ÿ8D8D �8�Y�ÿ9 7D43ÿD�ÿDÿ<4 �94ÿ9<ÿ�ÿ9ED8�D��9ÿ8�ÿ 789798�94ÿ9<ÿ8�ÿ�D>7;�ÿ4�ÿ9E�8�ÿC3ÿ�78 �j<�8ÿ<98��ÿD4ÿ8�ÿ4>C�8ÿ9<ÿ7899�ÿ9<ÿ8D8��ÿ8�98�ÿDÿ D>�8Dÿ�D�94�ÿNP77�4�5ÿ��:ÿÿl4ÿC98ÿD���7ÿ8�ÿD��9�D�94ÿ�D�ÿ79���E�7ÿ�4�D�4=ÿ8Dÿ8�ÿ789CDC�;�3ÿ 9<ÿ;345ÿ��ÿ�48�D��ÿ��8ÿ>98�ÿ�78 �j<�8ÿD4ÿ>98�ÿ8D8��:ÿÿm8�8ÿ9ED8�D��ÿD77�D8�ÿ�4ÿ97ÿ>9�;�ÿ��8ÿ �78 �j<�87ÿC ÿD ��94ÿ9<ÿ8���ÿ9ED8�D��ÿ�ÿ49ÿ�>789E�ÿPlIÿ�98��ÿC�394ÿ8�ÿ>9�;ÿ��8ÿ�78 �j<�8ÿ 94;3ÿNP77�4�5ÿ��:ÿÿS8��ÿ78�49>�494ÿ�4�D��ÿ8Dÿ8���ÿ98�8ÿED8�DC;��ÿ��8�ÿ49ÿ�4<98>D�E�:ÿÿ R�� �94ÿ789CDC�;�3ÿ�D�ÿ8�;D�E�;3ÿ8�=8ÿ<98ÿ�49�ÿ8DG�4=ÿ�8E�3�ÿNbÿAÿ�:̀�7ÿinÿAÿ�:�B�7ÿD4ÿ8�;D�E�;3ÿ ;9�ÿ<98ÿD>�8Dÿ�8E�3�ÿNbÿAÿ�:��7ÿinÿAÿ�:�g�ÿ 8�4=ÿR��>C�8oO�C8D83ÿD4ÿP78�;7ÿD;89=8ÿ�� �94ÿDÿ D>�8D�ÿ�48�D��ÿ9ÿ�:k�ÿNinÿAÿ�:���ÿ 8�4=ÿC8���4=ÿ��D�94ÿNMD88�ÿD�ÿ�57� �:ÿÿH�ÿ<94ÿDÿ ��=4�<�D47ÿ4�=D�E�ÿ�<<� ÿ9<ÿ8�ÿ��4�8ÿ9<ÿ��������1ÿ94ÿ�� �94ÿ789CDC�;�3ÿ<98ÿD>�8D�ÿNbÿAÿ�:�B7ÿinÿ Aÿ�:�gÿ<98ÿR��>C�8oO�C8D83ÿD4ÿP78�;pÿbÿAÿ�:�g7ÿinÿAÿ�:�kÿ<98ÿC8���4=ÿ��D�94�:ÿÿlÿ��ÿ4;�D8ÿ�8�8�8ÿ 8��ÿ897ÿ�4ÿ�� �94ÿ789CDC�;�3ÿ�D�ÿ �ÿ9ÿ8�ÿ;DGÿ9<ÿ�49�ÿ98ÿ8�ÿD;�84D�ÿ��4ÿ;8��:ÿÿH�ÿ���>D�ÿ #�ÿq�rÿ,�ÿ#!ÿ'�e)$!ÿ%f+'ÿ+fÿ#!ÿd�fÿs%�f6'ÿ&!"!ÿ,TT%)+!(ÿ1-ÿ$-ftÿ.uvrÿT,f�+(!fT!ÿ+f!"*�$wÿ�q� ��x�:ÿÿm 7D43ÿ���>D��ÿ<89>ÿ8�ÿ��������1ÿ>94�98�4=ÿ�<<98ÿ��8�ÿ8�ÿ8�=8��ÿ9C��8E�ÿ9ÿD�:ÿÿ y9��E�87ÿ94<��4�ÿ�4�8ED;�ÿ��8�ÿz��ÿ;D8=�7ÿ9��4=ÿ9ÿ8�ÿ�58Dÿ�<<� ÿ4���ÿ9ÿ>9�;ÿ�� �94ÿ�4ÿ ��������1ÿD4ÿ9ÿ8�ÿ;9�7ÿ7998;3ÿ���>D�ÿ�� �94ÿ789CDC�;�3ÿ8Dÿ8��;�ÿNO�=8�ÿ��:ÿÿS8�ÿ�7D�D;ÿ (+'"+1%+,fÿ,�ÿ$-ftÿ+fÿ#!ÿd�fÿs%�f6'ÿ"!e�+f!(ÿ$�"5!$-ÿ%fT#�f5!(ÿ.4+5%"!ÿ�{cÿ ÿ 1ÿ ÿ �ÿ ÿ ÿ ÿ 2345674865ÿ345 ÿ ÿ ���ÿ��ÿ��ÿ�����ÿ���������ÿ���������ÿ�!ÿ"�����ÿ#$�%ÿ��ÿ"�#�����&ÿ' �#(�����ÿ�!ÿ)*����+�ÿ ,��+ÿ�-.�/ÿ01ÿÿ2�#�#�!�ÿ3�+���45ÿ3�*���ÿ6ÿ��77�#+�ÿ"�#�����ÿ,��8+ÿ���ÿ2�#�#�!���ÿ 9���ÿ"�##��+�ÿ")�ÿ:�;�ÿ5��*&<<4*��+�����4��(+<#����<,��+<3�+���45��77�#+,(=+��+*%ÿ >��8��ÿ��ÿ>�ÿ���?�ÿ,�48��ÿ@3���8@&ÿ3ÿ"���ÿ!��ÿ���8ÿ;��#$+�+�ÿÿA��+���ÿ����-�ÿ5��*+&<<4�����6 *��B�4����<��=<*�48��+<3���8<3���8�*�!�ÿ ��4C��D���ÿE�ÿ��ÿ��ÿE�ÿF�45�#+�ÿ��ÿ;�ÿ3�$#��ÿC�ÿG�ÿ,�##�48�ÿ>�ÿ>�ÿH��#�$�ÿ���ÿ��ÿ'�ÿG���+�ÿ���I�ÿ )44(*��4$ÿ�+��7�����ÿ���ÿ7���#��&ÿ��!�����ÿ*������+ÿ���ÿ�$��7�4+ÿ�!ÿ+*�4��+ÿ �44(����4��ÿ;4���7�4ÿ,��++�ÿ)%!����ÿ:�����ÿC����7�ÿ 3ÿE��#�*7���ÿ"���ÿJ��7�ÿ���?�ÿ3&ÿ�ÿ#��(��ÿ���ÿ������7���ÿ!��ÿ+����+��4�#ÿ4�7*(����ÿ3ÿ 9�(�������ÿ!��ÿ�����+��4�#ÿ"�7*(����ÿA������ÿ;(+�����ÿ5��*&<<����36*��B�4�����ÿ 25����ÿK�ÿ"��ÿ���ÿC�ÿ,�ÿH(��5�7�ÿ�111�ÿ,����7ÿ�;3C&ÿ+(���#ÿ�+��7�����ÿ!��7ÿ*�*(#�����+ÿ �!ÿ7��8��ÿ���7�#+�ÿH���ÿ��(�$ÿLIÿ�(**#�7���&���.��?�ÿ ÿ ÿ J�=#�ÿ��ÿÿ�(77��$ÿ+����+��4+ÿ!��7ÿ+���ÿ���48��ÿ�!!����ÿ ÿ M:���+ÿ C7ÿ ����ÿC7ÿ M:���+ÿ ���5ÿ M>$�%ÿ MK�����4�ÿ �(��$��ÿ �(��$��ÿ M",2ÿ M:�9�ÿ ���+��ÿ �(��$��ÿ >$�%ÿ J��48+ÿ ��7*#�+ÿ NJ���#Oÿ *��ÿA�+��ÿ ,��+����#ÿ ,��+����#ÿ /ÿ ��ÿ ?=ÿ ??Lÿ ����ÿ ��ÿ ��ÿ ���L.���-ÿ �?ÿ 4 /ÿ �Lÿ 1ÿ 1?/ÿ ���1ÿ ��ÿ Iÿ ���-.���Iÿ �/ÿ � ?ÿ ��ÿ /ÿ /��ÿ �?��ÿ ��ÿ ?ÿ ���I.���/ÿ �Iÿ � � L ÿ / ÿ 1 ÿ � ÿ � ? ÿ � ? � ÿ � L ÿ -ÿ ���/.���?ÿ �ÿF(7=�� ÿ�!ÿ�����4ÿ+�7*#�+ÿN+4��ÿ��ÿ5���Oÿ4�##�4���ÿ��ÿ=�48���48��ÿ*(�����ÿ#$�%ÿ���48+ÿ =EF;ÿ � � � #$+�+ÿ4��!��7+ÿ�5��ÿ�##ÿ+�7*#�+ÿ4�##�4���ÿ!��7ÿ*(�����ÿ#$�%ÿ���48+ÿ����ÿ#$�%ÿ 4EF;ÿ ���#$+�+ÿ4��!��7+ÿ�5��ÿIÿ�!ÿ1ÿ+�7*#�+ÿ����ÿ#$�%ÿN�ÿ4�$����ÿ�ÿ���5��ÿ7(#�ÿ����ÿ��ÿ5(7���ÿ ��(������7����Oÿ EF;ÿ���#$+�+ÿ4��!��7��ÿ�5��ÿ-ÿ�!ÿ/ÿ+�7*#�+ÿ����ÿ#$�%ÿN�ÿ4�$����ÿ�ÿ+���+5��ÿ5���Oÿ �EF;ÿ 4��!��7�����ÿ*�����ÿ ÿ ÿ ÿ J�=#�ÿ��ÿÿ�(77��$ÿ+����+��4+ÿ!��7ÿ4�7���ÿ�!!����ÿ ÿ M:���+ÿ M:���+ÿ2��5ÿ M,5���+ÿ M,5���+ÿ M"�7���+ÿ M",2ÿ M:�9�ÿ ���+��ÿ �(��$��ÿ >$�%�ÿ NJ���#Oÿ N>$�%Oÿ 2��5ÿ>$�%ÿ ,��+����#ÿ ,��+����#ÿ ?ÿN/Oÿ ��L�I1Lÿ ���ÿ �Lÿ LIÿ ��ÿ ���L.���-ÿ ��ÿ /ÿNIOÿ ����-�Lÿ L--ÿ ��ÿ ��ÿ 1ÿ ���-.���Iÿ ��ÿ IÿN-Oÿ �I?�/�-ÿ �-�ÿ ��ÿ �1ÿ 1ÿ ���I.���/ÿ ��ÿ -ÿNLOÿ �/���/1ÿ 1�ÿ ?ÿ �-ÿ ?ÿ ���/.���?ÿ �-ÿ �ÿF(7=�� ÿ��ÿ*�����5�+�+ÿ����4���+ÿ(���+ÿ���5ÿ#$�%ÿ�(���ÿ�5�ÿ�!!�4��#ÿ+(��$ÿ*�����ÿNE�4ÿ�.;*�ÿ��Oÿ ÿ 1ÿ ÿ �ÿ ÿ ��ÿ ��ÿ ÿ ÿ ÿ ÿ ÿ ÿ 345678ÿ19ÿÿ ÿ��4��745ÿ78�6���ÿ��7ÿ��ÿ��8ÿ�6778�ÿ�����45ÿ�8���ÿ��21���21��ÿ��ÿ��ÿ��8ÿ�6�6���48ÿ ��4��745ÿ8���7�ÿ��21!��21��"ÿ#�ÿ$6�ÿ%�6��4�"ÿ��6��&8��ÿ'���7���9ÿÿ'���78�ÿ64��ÿ�(ÿ)ÿ*2�ÿ4�4���8ÿ ����8ÿ�8�8��8�ÿ��ÿ7����ÿ�7��ÿ��8ÿ���6���4�ÿ��ÿ64��ÿ�(ÿ)ÿ1�+�ÿ8�������45ÿ� ÿ���4���ÿ4ÿ��8ÿ#�ÿ $6�ÿ%�6��4�9ÿÿ ÿ&878ÿ�8�8��8�ÿ4ÿ1�ÿ64��ÿ4ÿ�21���21�ÿ��ÿ1+ÿ64��ÿ�6�6���48�ÿ�4�8ÿ ��4��745ÿ�85�ÿ4ÿ�21!��21*9ÿ 12ÿ ÿ �ÿ ÿ ÿ ÿ 1.0 0.9 0.8 0.7 (I) V, :::::l ....0 0.6 ' >, ~ 0.5 .c 0.4 :.cn::, 0 ,._ a.. -------------- ------- ------------ -------- ---------- 0.3 • 0.2 ------- ---- ---------- • . ------1/ --------- ------ ---- -------- 0.1 0.0 2010 2011 2012 2013 2014 Year 2015 2016 2017 234567ÿ89ÿÿ 7���76�47ÿ��5�����ÿ7��3���7�ÿ��ÿ���ÿ���37��7ÿ3��76���ÿ�6ÿ�����ÿ���ÿ3�ÿ��7ÿ���ÿÿ !"#ÿ%&!#'"(#)*ÿ)&!'+,-)'ÿ.&/&0"1&2ÿÿ34-"05ÿ(#1(6"'-)ÿ,+-#ÿ'+-ÿ-77&0')ÿ,-0-ÿ(#('("'-1ÿ8-292*ÿ:;<;=11>ÿ 8?1@=1AB9ÿÿ ÿ C��7�3�ÿ19ÿÿ 7ÿ�77��3�ÿ67�5��ÿ�6ÿ���ÿ� �3�63�4ÿ���ÿ� 7��7ÿ3�ÿ��7ÿ���ÿD5��ÿ 5���3��>ÿ � 6� >ÿ8?1?=:;<E2ÿÿF"#G(#9)ÿ"0-ÿH")-1ÿ&#ÿIG"(G-5)ÿJ#7&0K"'(&#ÿ.0('-0(&#ÿ"1L!)'-1ÿ7&0ÿ)K"//ÿ)"KM/-ÿ �3N7ÿOCP��B9ÿÿQ7�ÿ��63�R7�ÿS767ÿ���3767ÿ��ÿ���63��7�ÿ�ÿ3��6�ÿ7��3���3�ÿ�ÿ��5�����ÿOTB9ÿÿQ�7ÿ ���7�7ÿ� 7ÿ�7�ÿO�ÿUÿ�VBÿ3��57ÿ�ÿ��R3���3��ÿ�ÿ�S >ÿ3�ÿ� 3�3�ÿ�ÿ� 73�4ÿ7�7��3�ÿOWBÿ��ÿ�ÿ �5���3�ÿ�ÿ�56�7�ÿ�7�� >ÿR6773�4ÿ�7���>ÿ��ÿ��7ÿS���SÿS3��76ÿ�ÿ8?1@ÿ�6ÿ���76��9ÿÿX��ÿ��7ÿR7��ÿ 1?ÿ� 7�ÿ�67ÿ��S�9ÿ 7ÿ CP��ÿ YCP��ÿ CP��ÿZ��ÿ [ 9ÿ\�69ÿ � WO]7��^_8`"0-ÿaÿbM0!6-c(0ÿaÿ4-"0^ÿ @dV9eÿ ?9?ÿ ?9ddÿ 11ÿ W8f-)'^_8c&gÿaÿbM0!6-c(0ÿaÿ4-"0^ÿ @d�9?ÿ 89@ÿ ?911ÿ 11ÿ W8f-)'^ÿ_ÿ8bM0!6-c(0ÿaÿ4-"0^ÿ @d�9eÿ e9?ÿ ?91?ÿ 1?ÿ W8f-)'^ÿ_ÿ8h0&Mf!0#ÿaÿbM0!6-c(0ÿaÿ4-"0^ÿ @�?9eÿ d9?ÿ ?9?Vÿ 11ÿ W8f-)'^ÿ_ÿ8.&i&'-ÿaÿbM0!6-c(0ÿaÿ4-"0^ÿ @�?9Vÿ d9eÿ ?9?�ÿ 11ÿ W8f-)'^ÿ_ÿ8f&H6"'ÿaÿbM0!6-c(0ÿjÿk7�6Bÿ @�?9Aÿ d9�ÿ ?9?�ÿ 11ÿ W8f-)'^ÿ_ÿ840)b(#6-f--'/-ÿaÿbM0!6-c(0ÿaÿ4-"0^ÿ @�?9Aÿ d9�ÿ ?9?�ÿ 11ÿ W8f-)'^ÿ_ÿ8.&!9"0ÿaÿbM0!6-c(0ÿaÿ4-"0^ÿ @�?9�ÿ d9Vÿ ?9?�ÿ 11ÿ W8f-)'^ÿ_ÿ8blf--'/-m(//ÿaÿbM0!6-c(0ÿaÿ4-"0^ÿ @�19dÿ �91ÿ ?9?eÿ 11ÿ W 8 f) ' ^ ÿ _ÿ 8 h 0 & M f' / ÿ a ÿ b M 0 ! 6 c ( 0 ÿ a ÿ 4" 0 ^ ÿ @�19�ÿ �98ÿ ?9?eÿ 11ÿ ÿ �]7�� ��3��3�4ÿ��65��567ÿ�6ÿ7�7��3�ÿ�6R�R33��ÿ3��57ÿ7��7���ÿ�6ÿ�56�7�ÿ�7�� >ÿR6773�4ÿ�7���>ÿ ��ÿ��ÿ7��7��ÿ�ÿ��7ÿS���SÿS3��76ÿ�ÿ8?1@ÿ�ÿ���76��9ÿ ÿ ÿ ÿ ÿ ÿ 11ÿ ÿ �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Canada Lynx Monitoring in Colorado Period Covered: July 1, 2018 − June 30, 2019 Principal Investigators: Eric Odell, Eric.Odell@state.co.us; Jake Ivan, Jake.Ivan@state.co.us; Scott Wait, Scott.Wait@state.co.us All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. In an effort to restore a viable population of Canada lynx (Lynx canadensis) to the southern portion of their former range, 218 individuals were reintroduced into Colorado from 1999−2006. In 2010, the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) determined that the reintroduction effort met all benchmarks of success, and that the population of Canada lynx in the state was apparently viable and self-sustaining. In order to track the persistence of this new population and thus determine the long-term success of the reintroduction, a minimally-invasive, statewide monitoring program is required. During 2014−2019 CPW initiated a portion of the statewide monitoring scheme described in Ivan (2013) by completing surveys in a random sample of monitoring units (n = 50) from the San Juan Mountains in southwest Colorado (n = 179 total units; Figure 1). During 2018−2019 personnel from CPW and USFS completed the fifth year of monitoring work on this same sample. Specifically, 14 units were sampled via snow tracking surveys conducted between December 1 and March 31. On each of 1–3 independent occasions, survey crews searched roadways (paved roads and logging roads) and trails for lynx tracks. Crews searched the maximum linear distance of roads possible within each survey unit given safety and logistical constraints. Each survey covered a minimum of 10 linear kilometers (6.2 miles) distributed across at least 2 quadrants of the unit. The remaining 36 units could not be surveyed via snow tracking. Instead, survey crews deployed 4 passive infrared motion cameras in each of these units during fall 2018. Cameras were baited with visual attractants and scent lure to enhance detection of lynx living in the area. Cameras were retrieved during summer or fall 2019 and all photos were archived and viewed by at least 2 observers to determine species present in each. Camera data were then binned such that each of 10 15-day periods from December 1 through April 30 was considered an ‘occasion,’ and any photo of a lynx obtained during a 15-day period was considered a ‘detection’ during that occasion. Surveyors covered 510 km (317 mi) during snow tracking surveys and detected lynx at 6 units (Table 1). This represents a 5-year low in snow tracking effort and is due mostly to the record-setting snows experienced during the 2018–2019 winter. However, the mean distance surveyed per visit as well as the number of units with lynx remained similar to previous years. Surveyors collected more photos during 2018–2019 than in any other year. This was due in part to replacing snow tracking units with camera units in recent years, but mostly because many cameras were not retrieved until late summer or fall 2019 due to access issues related to the heavy snow pack. For the second year in a row we collected <50% of the number of lynx photos collected during the initial years of the monitoring effort, although the number of units with lynx returned to ‘normal’ after last year’s low (Table 2). Perhaps the abnormal snow patterns during the past few years (lack of snow in 2017–18, record snow in 2018–19) impacted our detection probability. Alternatively, lack of detections could have been due to the new lure (Caven’s 2 �Violator 7; Minnesota Trapline Products, https://www.minntrapprod.com/Bobcat-andLynx/products/829/) we used in 2017–2018 and 2018–19 after the lure we used previously (Pikauba; Luerres Forget’s Lures, http://www.leurresforget.com/product.php?id_product=15) became unavailable. Unfortunately, the changes in snow and lure are confounded, thus making it difficult to determine which factor resulted in fewer detections. We will use the same new lure in 2019–2020, which if accompanied by a normal snowfall, may allow us to retrospectively assess the lack of detections. Compared to previous years, we obtained new lynx detections at a camera unit near Table Mountain northwest of Creede and one north of Lemon Reservior. Also, we detected lynx again for only the second season at a unit west of Trujillo Meadows, near the New Mexico border. However, we failed to detect lynx in two units near Silverton that have had detections each winter since the inception of monitoring (Figure 1). Potential tracks were observed in each of these, but conditions were such that they could not be confirmed. An adult female with kittens was detected at cameras in a unit near Platoro Reservoir, thus documenting that at least some reproduction occurred in the study area. We used the R (R Development Core Team 2018) package ‘RMark’ (Laake 2018) to fit standard occupancy models (MacKenzie et al. 2006) to our survey data using program MARK (White and Burnham 1999). Thus, we estimated the probability of a unit being occupied (i.e., used) by lynx over the course of the winter (ψ), along with the probability of detecting a lynx (p) given that the unit was occupied. ‘Survey method’ and ‘year’ were treated as group variables so that we could, based on previous work, 1) allow detection probability to vary by survey method, 2) allow for detection probability for 2017–18 and/or 2018–19 to differ from other years due to abnormal snow or new lure, and 3) include a breeding season effect for detection at cameras (lynx tend to move more in late winter when they begin to breed, and thus should encounter cameras more often). We also considered a suite of covariates that could potentially explain variation in occupancy including proportion of the unit that was covered by spruce/fir forest, average years since bark beetle infestation, variability (standard deviation) in years since bark beetle infestation, proportion of the unit impacted by bark beetles, proportion of the unit that was burned during Summer 2013, and the number of photos of other species that could potentially impact presence of lynx (e.g., snowshoe hares as a food source, coyotes as potential competitors). We limited our model set by first setting a general structure for ψ while assessing fit of various combinations of variables expected to affect p. We then fixed the best-fitting structure for p, and assessed combinations of the covariates expected to influence ψ, allowing up to 2 of these covariates at a time, in addition to the covariates on detection. We included data from the pilot study (2010–11) as well as the first five years of monitoring (2014−2019) to maximize sharing of information across surveys. Since the inception of our monitoring program, the best-fitting model characterized occupancy as a function of 2 covariates: the proportion of the sample unit covered by spruce-fir forest and the number of photos of hares recorded at camera stations (Appendix 1). However, for the 2018–19 sampling year, the best fitting model characterized occupancy as a function of proportion of the sample unit covered by spruce-fir and by the number of cougar photos recorded at camera sites. The association with spruce-fir was positive, indicating that the probability of lynx use increased with more spruce-fir; the association with cougars was negative, indicating that probability of lynx use decreased with more photos of cougars. The second best model included bobcat photos in addition to spruce-fir; again lynx use was negatively associated with increased bobcat photos. Other covariates appeared in top models with spruce-fir, but addition of these covariates did not improve AICc scores beyond the model with spruce-fir only (Appendix 1). This phenomenon indicates that these other variables were not informative. Detection probability was relatively high for snow tracking surveys (p = 0.59, SE=0.05), and relatively low for camera surveys (p = 0.22, SE = 0.03) during December−February and April, although detection at cameras increased to 0.39 (SE = 0.07) during breeding season (March) as expected. We found a significant, negative effect on p during winters when Violator 7 was used as lure (p = 0.03, SE = 0.01 for December−February and April; p = 0.06, SE = 0.03 for breeding season), although it is unclear whether this drop in detection probability was due to abnormal snowpack or the alternate scent lure. We estimated that 31% of the sample units in the San Juan’s were occupied by lynx (95% confidence interval: 12–60%) 3 �during 2018–19. Confidence intervals were quite large for the second year in a row, owing to the extra parameter needed to model the “Violator 7 effect and to the low, poorly estimated detection probability that resulted (Figure 2). The spatial distribution of lynx in the San Juans remained largely unchanged (Figure 1). LITERATURE CITED Ivan, J. S. 2013. Statewide Monitoring of Canada lynx in Colorado: Evaluation of Options. Pages 15-27 in Wildlife Research Report - Mammals. Colorado Parks and Wildlife., Fort Collins, CO, USA. http://cpw.state.co.us/learn/Pages/ResearchMammalsPubs.aspx Laake, J. L. 2018. Package 'RMark': R Code for Mark Analysis. Version 2.2.5. https://cran.rproject.org/web/packages/RMark/RMark.pdf. MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. Academic Press, Oxford, United Kingdom. R Development Core Team. 2018. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 Supplement:120-138. Table 1. Summary statistics from snow tracking effort. Season #Units Surveyed #Units with Lynx #Lynx Tracks #Genetic Samplesa Km Surveyed (Total) Mean Km Surveyed per Visit #CPW Personnel #USFS Personnel 2014–2015 24 8 13 10b 1,088 20.1 30 13 2015–2016 17 7 14 9c 987 21.9 23 6 2016–2017 16 8 13 7 d 703 18.0 20 8 2017–2018 14 7 9 3e 578 19.3 14 5 2018–2019 14 6 7 2 510 19.6 16 5 e Number of genetic samples (scat or hair) collected via backtracking putative lynx tracks b DNA analysis confirms that all samples collected from putative lynx tracks were lynx c DNA analysis confirms that 6 of 9 samples were lynx (1 coyote, 1 either mule deer or human, 1undetermined) d DNA analyses confirmed that 5 of 7 samples were lynx (1 coyote, 1 snowshoe hare) e DNA confirmation pending a Table 2. Summary statistics from camera effort. Season 2014–2015 #Units Surveyed 32 2015–2016 31 #Units With Lynx #Photos (Total) #Photos (Lynx) #Cameras With Lynx #CPW Personnel 46 #USFS Personnel 12 134,694 301 14 33 9 101,534 455 10 2016–2017 33 29 9 168,705 251 10 6 (5) 2017–2018 35 35 8 173,279 90 8 5 (4) 2018–2019 36 31 7 204,243 60 10 7 (5) a Number in parenthesis indicates units with lynx during the official survey period (Dec 1–Apr 30) 8 (7) 7 (6) 4 �a) b) Figure 1. Lynx monitoring results for a) the current sampling season (2018–2019) and b) the cumulative monitoring effort (2014–2019), San Juan Mountains, southwest Colorado. Colored units (n = 50) indicate those selected at random from the population of units (n = 179) encompassing lynx habitat in the San Juan Mountains. Lynx were detected in 12 units in 2018−2019 and 23 units cumulatively since monitoring began in 2014−2015. 5 �1.0 0.9 -- 0.8 0.7 >, u C: m a. :::i u u 0 ·-~ 0.6 ti 0.5 0.40.3 0.2 0.1 - l I II lt -- -~ 0.0 I I I I I I I I I 2010 2011 2012 2013 2014 2015 2016 2017 2018 Year Figure 2. Model-averaged occupancy estimates and 95% confidence intervals for occupancy of Canada lynx in the San Juan Mountains, southwest Colorado. ‘Year’ indicates when the efforts were initiated (e.g., 2010−11, 2018−19). Appendix 1. Model selection results for lynx monitoring data collected in the San Juan Mountains, Colorado, 2010–2019. Rankings are based on Akaike’s Information Criterion adjusted for small sample size (AICc). Ten variables were considered as covariates to inform estimation of occupancy (ψ). The complete model set (n = 56) included all combinations of two, in addition to modeling detection (p) as a function of survey method, breeding season, and alternate lure used during the 2017–18 and 2018–19 seasons. Only the best 10 models are shown. Model AICc ∆AICc AICc Wts No. Par. a p(Best ) ψ (Cougar + Prop Spruce/Fir) 817.89 0 0.64 12 p(Best) ψ (Bobcat + Prop Spruce/Fir) 820.87 2.98 0.15 12 p(Best) ψ (Prop Spruce/Fir) 822.92 5.03 0.05 11 p(Best) ψ (Prop Burned + Prop Spruce/Fir) 824.14 6.26 0.03 12 p(Best) ψ (Coyote + Prop Spruce/Fir) 824.26 6.38 0.03 12 p(Best) ψ (Years Since Beetles + Prop Spruce/Fir) 824.46 6.57 0.02 12 p(Best) ψ (Fox + Proportion Spruce/Fir) 824.61 6.72 0.02 12 p(Best) ψ (Hare + Proportion Spruce/Fir) 825.03 7.14 0.02 12 p(Best) ψ (Prop Beetle + Prop Spruce/Fir) 825.06 7.17 0.02 12 p(Best) ψ (Variability Beetles + Prop Spruce/Fir) 825.08 7.19 0.02 12 a Best-fitting structure for detection probability included effects for survey method, breeding season, and an effect for the 2017–18 and 2018–19 survey seasons when Violator 7 was used for lure rather than Pikauba. 6 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Canada lynx monitoring in Colorado Period Covered: July 1, 2019 − June 30, 2020 Principal Investigators: Eric Odell, Eric.Odell@state.co.us; Jake Ivan, Jake.Ivan@state.co.us; Scott Wait, Scott.Wait@state.co.us; Morgan Hertel, Morgan.Hertel@state.co.us Personnel: Brad Weinmeister, Evan Phillips, Nate Seward, Brent Frankland All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. In an effort to restore a viable population of Canada lynx (Lynx canadensis) to the southern portion of their former range, 218 individuals were reintroduced into Colorado from 1999−2006. In 2010, the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) determined that the reintroduction effort met all benchmarks of success, and that the population of Canada lynx in the state was apparently viable and self-sustaining. To track the persistence of this new population and thus determine the long-term success of the reintroduction, a minimally-invasive, statewide monitoring program is required. During 2014−2020 CPW initiated a portion of the statewide monitoring scheme described in Ivan (2013) by completing surveys in a random sample of monitoring units (n = 50) from the San Juan Mountains in southwest Colorado (n = 179 total units; Figure 1). During 2019−2020 personnel from CPW and USFS completed the sixth year of monitoring work on this same sample. Specifically, 14 units were sampled via snow tracking surveys conducted between December 1 and March 31. On each of 1–3 independent occasions, survey crews searched roadways (paved roads and logging roads) and trails for lynx tracks. Crews searched the maximum linear distance of roads possible within each survey unit given safety and logistical constraints. Each survey covered a minimum of 10 linear kilometers (6.2 miles) distributed across at least 2 quadrants of the unit. The remaining 36 units could not be surveyed via snow tracking. Instead, survey crews deployed 4 passive infrared motion cameras in each of these units during fall 2019. Cameras were baited with visual attractants and scent lure to enhance detection of lynx living in the area. Cameras were retrieved during summer or fall 2020 and all photos were archived and viewed by at least 2 observers to determine species present in each. Camera data were then binned such that each of 10 15-day periods from December 1 through April 30 was considered an ‘occasion,’ and any photo of a lynx obtained during a 15-day period was considered a ‘detection’ during that occasion. Surveyors covered 650 km during snow tracking surveys and detected lynx at 6 units (Table 1). These results are among the lowest recorded for the project, but mirror those recorded during the past 3 years (Table 1). Surveyors collected more than 3 times the photos during 2019–2020 than have been collected in any other year. This can be mostly attributed to the use of new, more sensitive cameras along with new, high capacity memory cards. However, for the third year in a row we collected <50% of the number of lynx photos taken during the initial years of the monitoring effort (Table 2). In fact, the 36 lynx photos collected during the 2019−20 season was the fewest recorded since the inception of the project. We initially considered at least 3 possible explanations for the lack of photos collected in recent years. First, we hypothesized that abnormal snow patterns (lack of snow in 2017–18, record snow in 2 �2018–19) could have impacted detection probability. Second, lack of detections could have been due to the new lure (Caven’s Violator 7; Minnesota Trapline Products, https://www.minntrapprod.com/Bobcatand-Lynx/products/829/) we used in 2017–18, 2018–19, and 2019−20 after the lure we used previously (Pikauba; Luerres Forget’s Lures, http://www.leurresforget.com/product.php?id_product=15) became unavailable. Finally, it could be that lynx have disappeared from a number of camera units. Unfortunately, the changes in snow and lure were confounded for a few years, thus making it difficult to determine which factor resulted in fewer detections. However, 2019−20 was a normal snow year, yet the number of lynx photos was still low. This indicates that abnormal snow was not the cause of the pattern we observed. Also, the number of snow tracking units with lynx has remained fairly steady throughout the project; we can think of no reason why snow track units would remain occupied while lynx blinked out of camera units, unless just by chance. Thus, we suggest that the new lure is less effective than the original. Fortunately the original formulation is again available and will be deployed for the 2020−21 survey. We plan to utilize this lure for the remainder of the survey efforts, provided it remains available. We obtained lynx detections for only the second time at a camera unit near Wolf Creek Pass. Lynx were again detected at Lizard Head Pass after no detections last year, and in all four snow tracking units along the Hwy 550 corridor after two of the four went without detections in 2018−19. However, we failed to detect lynx in at the Table Mountain Unit northwest of Creede, at Lemon Reservoir, at Little Squaw Creek west of Creede, and at Trujillo Meadows near the New Mexico border, where they had been detected the previous two seasons (Figure 1). We used the R (R Development Core Team 2018) package ‘RMark’ (Laake 2018) to fit multipleseason (i.e., “dynamic”) occupancy models (MacKenzie et al. 2006) to our survey data using program MARK (White and Burnham 1999). Thus, we estimated the derived probability of a unit being occupied (i.e., used) by lynx over the course of the winter (ψ), along with the probability of detecting a lynx (p) given that the unit was occupied, the probability a unit that was unused in one year was used the next (i.e., “local colonization”, γ), and the probability a used unit became unused from one year to the next (i.e., “local extinction”, ε). Based on previous work, we treated ‘survey method’ as a group variable so that we could allow p to vary by method. Additionally, we allowed p for 2017–18, 2018–19, and 2019–20 to differ from other years due to the new lure, and we included a breeding season effect for detection at cameras (lynx tend to move more in late winter when they begin to breed, and thus should encounter cameras more often). Also based on previous work, we specified initial ψ in the time series to be a function of the proportion of the unit that was covered by spruce/fir forest. We then allowed annual estimates of ε to be constant or a function of average years since bark beetle infestation, proportion of the unit impacted by bark beetles, proportion of the unit that was burned during Summer 2013, and the number of photos of other species that could potentially impact presence of lynx (e.g., snowshoe hares as a food source; coyotes, bobcats, foxes, and cougars as potential competitors). We allowed annual estimates of γ to be constant or a function of snowshoe hares. We limited our model set by first setting a general structure for ψ while assessing fit of various combinations of variables expected to affect p. We then fixed the best-fitting structure for p, and assessed combinations of the covariates expected to influence ε or γ, allowing up to 2 of these covariates at a time, in addition to the covariates on detection. We made inference from the best-fitting model as selected via Akaikie’s Information Criterion (AIC), adjusted for small sample size (Burnham and Anderson 2002). As has been the case since the inception of our monitoring program, the proportion of the sample unit covered by spruce-fir forest was positively associated with the initial occupancy estimate in the time series. Local colonization probability was estimated to be low (γ = 0.03, SE = 0.01 ) and constant; local extinction was also low, but in some years twice that of colonization (ε = 0.03 to 0.06, SE = 0.03 to 0.05). Furthermore, in all of the top models, ε was negatively (but weakly) associated with the number of coyote photos collected on the year indicating that the probability of extinction of a unit in any given year goes up as the index of coyote abundance goes down (Appendix 1). Local extinction was also significantly, positively associated with the number of fox photos in the top model, suggesting that extinction is more likely in units in which we detected fox more often. Other models for ε that performed better than a 3 �constant structure included a negative relationship with number of snowshoe hare photos (less likely to go extinct as hare index increases), a positive relationship with the number of bobcat photos (more likely to go extinct as bobcat index increases), and a positive association with proportion of a unit impacted by beetles. However, the hare, bobcat, and beetle models were not as well supported as those including coyotes and foxes. The five occupancy growth rates (λ) estimated between surveys were all near 1.0, indicating a stable distribution with little to no growth (Figure 2). Similar to previous years, detection probability was relatively high for snow tracking surveys (p = 0.59, SE=0.05), and relatively low for camera surveys (p = 0.23, SE = 0.04) during December−February and April, although detection at cameras increased to 0.34 (SE = 0.07) during breeding season (March) as expected. We found a significant, negative effect on p during winters when Violator 7 was used as lure (p = 0.08, SE = 0.02 for December−February and April; p = 0.13, SE = 0.05 for breeding season). We estimated that 29% of the sample units in the San Juan’s were occupied by lynx (95% confidence interval: 15–43%) during 2019– 20 (Figure 2). The spatial distribution of lynx in the San Juans remained largely unchanged (Figure 1). Table 1. Summary statistics from snow tracking effort. Season 2014-2015 #Units Surveyed 24 #Units with Lynx 8 #Lynx Tracks 13 2015-2016 17 7 14 #Genetic Samplesa 10b 9c 13 7d 2016-2017 16 8 Km Surveyed (Total) 1,088 Mean Km Surveyed per Visit 20.1 #CPW Personnel 30 #USFS Personnel 13 987 21.9 23 6 703 18.0 20 8 2017-2018 14 7 9 3e 578 19.3 14 5 2018-2019 14 6 7 2e 510 19.6 16 5 10 2b 650 19.7 15 3 2019-2020 15 6 Number of genetic samples (scat or hair) collected via backtracking putative lynx tracks b DNA analysis confirms that all samples collected from putative lynx tracks were lynx c DNA analysis confirms that 6 of 9 samples were lynx (1 coyote, 1 either mule deer or human, 1undetermined) d DNA analyses confirmed that 5 of 7 samples were lynx (1 coyote, 1 snowshoe hare) e DNA analysis confirms 1 sample was lynx; remaining samples were not analyzed a Table 2. Summary statistics from camera effort. Season 2014-2015 #Units Surveyed 32 2015-2016 2016-2017 31 33 2017-2018 35 2018-2019 36 2019-2020 36 #Units With Lynx #Photos (Total) #Photos (Lynx) #Cameras With Lynx 8 7 6 5 6 4 134,694 101,534 168,705 173,279 204,243 701,724 301 455 251 90 59 36 14 10 10 8 9 4 4 #CPW Personnel 46 #USFS Personnel 12 33 29 9 9 35 8 31 7 29 6 �a) b) Figure 1. Lynx monitoring results for a) the current sampling season (2019–2020) and b) the cumulative monitoring effort (2014–2020), San Juan Mountains, southwest Colorado. Colored units (n = 50) depicted here are those selected at random from the population of units (n = 179) encompassing lynx habitat in the San Juan Mountains. Lynx were detected in 11 units in 2019−2020 and 23 units cumulatively since monitoring began in 2014−2015. 5 �- 1.2 I I 1.0 0.9 0.8 - I I - 1.1 I - 1.0 - 0.9 - 0.8 >, u 0.7 - - 0. 7 0. ::l 0.6 - - 0.6 C rel u u 0 0.5 0.4 0. 3 0.2 0. 1 - - 0. 5 I I I I I I - 0.4 Cl '"' 0 ~ ,..... ~ :::0 QI ,..... (1) - 0.3 - 0.2 - 0. 1 0.0 - - 0.0 I I I 2014 2015 2016 I I I 201 7 2018 2019 Year Figure 2. Occupancy estimates (Ψ, filled circles, left axis) and annual growth rate (λ) in occupancy between surveys (open circles, right axis) for Canada lynx in the San Juan Mountains, southwest Colorado. ‘Year’ indicates when the efforts were initiated (e.g., winter 2014−15, winter 2019−20). Growth rates less than 1.0 indicate a decline in occupancy; those >1.0 indicate an increase. Literature Cited Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd edition. Springer, New York, New York, USA. Ivan, J. S. 2013. Statewide monitoring of Canada lynx in Colorado: evaluation of options. Pages 15–27 in Wildlife research report: Mammals. Colorado Parks and Wildlife., Fort Collins, USA. http://cpw.state.co.us/learn/Pages/ResearchMammalsPubs.aspx. Laake, J. L. 2018. Package “RMark”: R Code for Mark Analysis. Version 2.2.5. https://cran.rproject.org/web/packages/RMark/RMark.pdf. MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. Academic Press, Oxford, United Kingdom. R Development Core Team. 2018. No Title. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 Supplem:120–138. 6 �Appendix 1. Model selection results for lynx monitoring data collected in the San Juan Mountains, Colorado, 2014–2020. Rankings are based on Akaike’s Information Criterion adjusted for small sample size (AICc). Eight variables were considered as covariates to inform estimation of local extinction (ε); one was considered for local colonization (γ). The complete model set (n = 46) included all combinations of two of these covariates, in addition to modeling detection (p) as a function of survey method, breeding season, and alternate lure used during the 2017–18, 2018–19, and 2019–2020 seasons. Only the best 10 models are shown. Model ψ (Prop Spruce/Fir) ε (Coyote + Fox) γ (.) p (Best) ψ (Prop Spruce/Fir) ε (Coyote) γ (.) p (Best) ψ (Prop Spruce/Fir) ε (Coyote + PropBeetle) γ (.) p (Best) ψ (Prop Spruce/Fir) ε (Coyote + Hare) γ (.) p (Best) ψ (Prop Spruce/Fir) ε (Bobcat + Coyote) γ (.) p (Best) ψ (Prop Spruce/Fir) ε (.) γ (.) p (Best) ψ (Prop Spruce/Fir) ε (Coyote + PropBurn) γ (.) p (Best) ψ (Prop Spruce/Fir) ε (BKAvg + Coyote) γ (.) p (Best) ψ (Prop Spruce/Fir) ε (Cougar + Coyote) γ (.) p (Best) ψ (Prop Spruce/Fir) ε (Bobcat) γ (.) p (Best) AICc 574.54 576.43 576.50 576.61 577.17 578.01 578.12 578.21 578.30 578.50 ∆AICc 0.00 1.89 1.96 2.07 2.63 3.47 3.58 3.67 3.76 3.96 AICc Wts 0.19 0.08 0.07 0.07 0.05 0.03 0.03 0.03 0.03 0.03 No. Par. 10 9 10 10 10 8 10 10 10 9 Best-fitting structure for detection probability included effects for survey method, breeding season, and an effect for the 2017–18, 2018–19, and 2019–20 survey seasons when Violator 7 was used for lure rather than Pikauba. a 7 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Canada lynx monitoring in Colorado 2020 – 2021 Period Covered: July 1, 2020 − June 30, 2021 Principal Investigators: Eric Odell, Eric.Odell@state.co.us; Morgan Hertel, Morgan.Hertel@state.co.us; Jake Ivan, Jake.Ivan@state.co.us All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. In an effort to restore a viable population of Canada lynx (Lynx canadensis) to the southern portion of their former range, 218 individuals were reintroduced into Colorado from 1999−2006. In 2010, the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) determined that the reintroduction effort met all benchmarks of success and that the population of Canada lynx in the state was apparently viable and self-sustaining. In order to track the persistence of this new population and thus determine the long-term success of the reintroduction, a minimally-invasive, statewide monitoring program is required. From 2014−2021 CPW initiated a portion of the statewide monitoring scheme described in Ivan (2013) by completing surveys in a random sample of monitoring units (n = 50) from the San Juan Mountains in southwest Colorado (n = 179 total units; Figure 1). During the 2020−2021 winter, personnel from CPW and USFS completed the seventh year of monitoring work on this same sample. Fourteen units were sampled via snow-tracking surveys conducted between December 1 and March 31. On each of 1–3 independent occasions, survey crews searched roadways (snow-covered paved roads and logging roads) and trails for lynx tracks. Crews searched the maximum linear distance of roads possible within each survey unit given safety and logistical constraints. Each survey covered a minimum of 10 linear kilometers (6.2 miles) distributed across at least 2 quadrants of the unit. The remaining 36 units could not be surveyed via snow tracking. Instead, survey crews deployed 4 passive infrared motion cameras in each of these units during fall 2020. Cameras were lured with visual attractants and scent lure to enhance detection of lynx in the area. Cameras were retrieved during summer or fall 2021 and all photos were archived and viewed by at least 2 observers to determine species present in each. Camera data were then binned such that each of 10 15-day periods from December 1 through April 30 was considered an ‘occasion,’ and any photo of a lynx obtained during a 15-day period was considered a ‘detection’ during that occasion. Surveyors covered 744 km during snow tracking surveys and detected lynx at 7 units (Table 1). In 2020-21 surveyors collected more DNA samples than in previous years, likely because new environmental DNA (eDNA) sampling is more efficient to collect than the previous scat or hare sampling. As in 2019-20, significantly more photos were collected in 2020-21 than in the first 5 seasons of sampling. This can be mostly attributed to the use of new, more sensitive cameras along with new, highcapacity memory cards. However, for the fourth year in a row, we collected <50% of the number of lynx photos taken during the initial years of the monitoring effort (Table 2). In fact, the 36 lynx photos collected during the 2019-20 and 2020-21 seasons are the fewest recorded since the inception of the project. We initially considered at least 3 possible explanations for the lack of photos collected in recent years. First, we hypothesized that abnormal snow patterns (lack of snow in 2017–18, record snow in 2018–19) could have impacted detection probability. Second, lack of detections could have been due to 2 �the new lure (Caven’s Violator 7; Minnesota Trapline Products, https://www.minntrapprod.com/Bobcatand-Lynx/products/829/) we used in 2017–18, 2018–19, 2019-20, and 2020-21 after the lure we used previously (Pikauba; Luerres Forget’s Lures, http://www.leurresforget.com/product.php?id_product=15) became unavailable. Finally, it could be that lynx have disappeared from a number of camera units. Unfortunately, the changes in snow and lure were confounded for a few years, thus making it difficult to determine which factor resulted in fewer detections. However, 2019-20 and 2020-21 were normal snow years, yet the number of lynx photos was still low. This suggests that abnormal snow was not the cause of the pattern we observed. Also, the number of snow tracking units with lynx has remained fairly steady throughout the project; we can think of no reason why snow track units would remain occupied while lynx blinked out of camera units, unless just by chance. Thus, we suggest that the new lure is less effective than the original. Fortunately the original formulation, Pikauba, is again available and will be deployed for the 2021-22 survey. We plan to utilize this lure for the remainder of the survey efforts, provided it remains available. We obtained lynx detections for the first time in a unit near Mesa Mountain in the La Garitas. This detection represents the northernmost detection of lynx since surveys began. We also detected lynx for the first time in the unit that encompasses Fern Creek and lower Trout Creek west of Creede. This unit, however, is surrounded by other units where lynx have been detected several times previously. After a 1-year absence, lynx were again detected in the Barlow Creek Unit near Rico and the Pass Creek Unit near Wolf Creek Pass; lynx were not detected at the two units adjacent to Pass Creek, or at the southern Conejos Peak Unit after having been detected in all 3 last year (Figure 1). We used the R (R Development Core Team 2018) package ‘RMark’ (Laake 2018) to fit multipleseason (i.e., “dynamic”) occupancy models (MacKenzie et al. 2006) to our survey data using program MARK (White and Burnham 1999). Thus, we estimated the derived probability of a unit being occupied (i.e., used) by lynx over the course of the winter (ψ), along with the probability of detecting a lynx (p) given that the unit was occupied, the probability a unit that was unused in one year was used the next (i.e., “local colonization,” γ), and the probability a used unit became unused from one year to the next (i.e., “local extinction,” ε). For each model we fit for the analysis, we specified that the initial ψ in the time series should be a function of the proportion of the unit that is covered by spruce/fir forest – the single most important and consistent predictor of ψ in past analyses. For sake of comparison we fit a base model in which p was specified to be constant for the duration of the survey. Based on previous work, however, we considered several other structures for p we anticipated would fit better. We fit models that specified 1) p could vary by survey method (i.e., detection could be different for cameras compared to snowtracking), 2) p could be higher during breeding season when lynx tend to move more and are therefore more likely to be detected by track or at a camera, and 3) p for cameras deployed from 2017–21 could be different than p for other years due to the lure substitution. Additionally we fit a model in which the effect of breeding season was only allowed to act on cameras, not snowtracking. We allowed annual estimates of ε and γ to be different each year (i.e., assuming occupancy dynamics were not random but instead dependent on the year previous and the population is not at equilibrium), which allowed derived ψ to vary as freely as possible given the data. We used Akaike’s Information Criterion (AIC), adjusted for small sample size (Burnham and Anderson 2002) to identify the best-fitting model from this small set. Ultimately, we fit a linear model through the time series of ψ estimates to estimate the slope of the trend in occupancy through time. Ideally we would test other predictions of lynx occupancy to see, for instance, if colonization or extinction were influenced by bark beetles, fire, or the presence of competitors or prey species. However, we do not currently have enough data to test these predictions in addition to assessing trend, which is the highest priority. As has been the case since the inception of our monitoring program, the proportion of the sample unit covered by spruce-fir forest was significantly and positively associated with the initial occupancy estimate in the time series. Even though local colonization and extinction were allowed to vary freely from year to year, annual estimates were near zero and varied little (ε = 0.00–0.08; γ = 0.00–0.10). Accordingly, derived occupancy was relatively stable across years (ψ = 0.26–0.38). The slope of the trend 3 �in occupancy through time was slightly positive but not significantly different from zero (β = 0.017, SE = 0.01; Figure 2). These results suggests that future analyses may benefit from fitting models that hypothesize occupancy is at or near equilibrium and extinction/colonization are either Markovian (as modeled here) or possibly zero. Similar to previous years, detection probability was relatively high for snow tracking surveys (p = 0.69, SE = 0.06), lower for camera surveys (p = 0.23, SE = 0.03) using Pikauba, and lowest for camera surveys utilizing Violator 7 (p = 0.06, SE = 0.02). We estimated that 38% of the sample units in the San Juan’s were occupied by lynx (95% confidence interval: 20–55%) during 2020–21 (Figure 2). The spatial distribution of lynx in the San Juan mountains remained largely unchanged (Figure 1). Table 1. Summary statistics from snow tracking effort. Lynx DNAb 8 6 Km Surveyed (Total) 884 987 Mean Km Surveyed per Visit 20.1 21.9 #CPW Personnelc 30 23 #USFS Personnelc 13 6 Season 2014–2015 2015–2016 #Units Surveyed 18 17 #Units with Lynx 7 7 2016–2017 16 8 13 7 5 703 18.0 20 8 2017–2018 14 7 9 3 1 578 19.3 14 5 2018–2019 14 6 8 2 1 510 19.6 16 5 2019–2020 14 7 11 3 2 640 19.4 15 3 2020–2021 15 9 14 12 7 790 18.8 17 3 #Lynx Tracks 12 14 #Genetic Samplesa 8 9 Number of genetic samples (scat, hair, or eDNA) collected via backtracking putative lynx tracks b Number of genetic samples that came back positive for Lynx c Number of staff that participate in the annual sampling effort a Table 2. Summary statistics from camera effort. Season 2014–2015 2015–2016 #Units Surveyed 31 31 2016–2017 33 2017–2018 35 2018–2019 35 2019–2020 36 2020–2021 35 #Units With Lynx #Photos (Total) #Photos (Lynx) #Cameras With Lynx 7 7 6 5 6 4 3 133,483 101,534 168,705 173,279 201,782 706,074 347,868 184 455 251 90 59 36 36 11 10 10 8 9 4 3 4 #CPW Personnel 46 33 #USFS Personnel 12 9 29 9 35 8 31 7 29 6 23 5 �a) b) Years With Lynx Detections 0 010 • 2') □ , 10 . .J - Zlo • l) - llo • 1) - • 10 • 1) - 510 • <) - •10 • 6) - 71o -7) Figure 1. Lynx monitoring results for a) the current sampling season (2020–2021) and b) the cumulative monitoring effort (2014–2021), San Juan Mountains, southwest Colorado. Colored units (n = 50) depicted here are those selected at random from the population of units (n = 179) encompassing lynx habitat in the San Juan Mountains. Lynx were detected in 12 units in 2020−2021 and 24 units cumulatively since monitoring began in 2014−2015. 5 �1.0 0.9 0.8 0.7 >, 0.6- C. :::J 0.5- u C ro u u 0 0.4 0.3 0.2 0.1 0.0 I I I I I I I 2014 2015 2016 2017 2018 2019 2020 Figure 2. Occupancy estimates (Ψ) and trend (including 95%CI) for Canada lynx in the San Juan Mountains, southwest Colorado. ERRATA: We note here that some data in Tables 1 and 2, and Figure 1 are incongruent with reports issued for the previous two seasons. This was due to inadvertent removal of filters in our database that were originally set to exclude pilot data from report tables, figures, and input files. These filters have been restored. The cumulative tables and figures presented here are accurate and supersede discrepancies with previous reports. Literature Cited Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd edition. Springer, New York, New York, USA. Ivan, J. S. 2013. Statewide monitoring of Canada lynx in Colorado: evaluation of options. Pages 15–27 in Wildlife research report: Mammals. Colorado Parks and Wildlife., Fort Collins, USA. http://cpw.state.co.us/learn/Pages/ResearchMammalsPubs.aspx. Laake, J. L. 2018. Package “RMark”: R Code for Mark Analysis. Version 2.2.5. https://cran.rproject.org/web/packages/RMark/RMark.pdf. MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. Academic Press, Oxford, United Kingdom. R Development Core Team. 2018. No Title. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 Supplem:120–138. 6 �Appendix 1. Model selection results for lynx monitoring data collected in the San Juan Mountains, Colorado, 2014–2021. Rankings are based on Akaike’s Information Criterion adjusted for small sample size (AICc). We mostly sought to tease out best fitting models for detection, allowing constant detection (.), along with effects for survey type (ST), breeding season (B), substituting Violator 7 lure for Pikauba (V), and interactions to allow lure and breeding to act only on cameras. For these models we fixed the initial ψ to be a function of spruce-fir forest while local extinction (ε) and colonization (γ) were estimated annually to allow for non-equilibrium estimates in ψ that depended on previous year’s occupancy state. Post-hoc, we added tested for equilibrium conditions (ε (.) γ (.) ) or that occupancy from year to year was random ({ε = 1- γ}). Model ψ (Prop Spruce/Fir) ε (t) γ (t) p (ST+V+ST*V) ψ (Prop Spruce/Fir) ε (t) γ (t) p (ST+B+V+ST*V) ψ (Prop Spruce/Fir) ε (t) γ (t) p (ST+B+V+ST*B+ST*V) ψ (Prop Spruce/Fir) ε (t) γ (t) p (ST) ψ (Prop Spruce/Fir) ε (t) γ (t) p (ST+B) ψ (Prop Spruce/Fir) ε (.) γ (.) p (.) ψ (Prop Spruce/Fir) ε (t) γ (t) p (.) ψ (Prop Spruce/Fir) {ε = 1- γ}p (1) 7 AICc 674.04 675.88 676.77 697.55 699.41 749.98 768.42 914.99 ∆AICc 0.00 1.85 2.74 23.52 25.38 75.95 94.38 240.95 AICc Wts 0.61 0.24 0.15 0.00 0.00 0.00 0.00 0.00 No. Par. 17 18 19 15 16 4 14 8 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Canada lynx monitoring in Colorado 2021 – 2022 Period Covered: July 1, 2021 − June 30, 2022 Principal Investigators: Eric Odell, Eric.Odell@state.co.us; Morgan Hertel, Morgan.Hertel@state.co.us; Jake Ivan, Jake.Ivan@state.co.us All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. In an effort to restore a viable population of Canada lynx (Lynx canadensis) to the southern portion of their former range, 218 individuals were reintroduced into Colorado from 1999−2006. In 2010, the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) determined that the reintroduction effort met all benchmarks of success and that the population of Canada lynx in the state was apparently viable and self-sustaining. In order to track the persistence of this new population and thus determine the long-term success of the reintroduction, a minimally-invasive, statewide monitoring program is required. From 2014−2022 CPW initiated a portion of the statewide monitoring scheme described in Ivan (2013) by completing surveys in a random sample of monitoring units (n = 50) from the San Juan Mountains in southwest Colorado (n = 179 total units; Figure 1). During the 2021−2022 winter, personnel from CPW and USFS completed the eighth year of monitoring work on this same sample. Fourteen units were sampled via snow-tracking surveys conducted between December 1 and March 31. On each of 1–3 independent occasions, survey crews searched roadways (snow-covered paved roads and logging roads) and trails for lynx tracks. Crews searched the maximum linear distance of roads possible within each survey unit given safety and logistical constraints. Each survey covered a minimum of 10 linear kilometers (6.2 miles) distributed across at least 2 quadrants of the unit. The remaining 36 units could not be surveyed via snow tracking. Instead, survey crews deployed 4 passive infrared motion cameras in each of these units during fall 2021. Cameras were lured with visual attractants and scent lure to enhance detection of lynx in the area. Cameras were retrieved during summer or fall 2022 and all photos were archived and viewed by at least 2 observers to determine species present in each. Camera data were then binned such that each of 10 15-day periods from December 1 through April 30 was considered an ‘occasion,’ and any photo of a lynx obtained during a 15-day period was considered a ‘detection’ during that occasion. Surveyors covered 692 km during snow tracking surveys and detected only 6 lynx tracks at 4 units, both all-time low for the program (Table 1). Significantly, more photos were collected in the past three seasons than in the first 5 seasons of sampling. This can be mostly attributed to the use of new, more sensitive cameras along with new, high-capacity memory cards. After four seasons (2017-2020) in which we collected the fewest lynx photos of any set of years on the project (<50% of the number of lynx photos taken during the initial years of the monitoring effort), the number of lynx photos collected this year rebounded substantially (Table 2). This substantiates our previous conclusions that the Violator7 lure (in use during those 4 season) was less effective than the Pikauba lure used this year and during the first 3 years of sampling. Pikauba will be utilized for the remainder of the survey efforts, provided it remains available. 8 �We obtained lynx detections in the La Garita Mountains north of Creede for first time in 5 years. Lynx were detected in the two units near Conejos Peak after having not been detected last year. Snowtracking surveys did not provide lynx detections in either the Mineral Creek or Molas Pass units near Silverton, nor at the Lime Creek unit south of Creede. This lack of detections is notable because these 3 units are among the most reliable for detecting lynx in the entire study area; each has provided lynx detections for 6–7 of the 8 years these areas have been surveyed (Figure 1). We used the R package (R Development Core Team 2018) ‘RMark’ (Laake 2018) to fit multipleseason (i.e., “dynamic”) occupancy models (MacKenzie et al. 2006) to our survey data using program MARK (White and Burnham 1999). Thus, we estimated the derived probability of a unit being occupied (i.e., used) by lynx over the course of the winter (ψ), along with the probability of detecting a lynx (p) given that the unit was occupied, the probability a unit that was unused in one year was used the next (i.e., “local colonization,” γ), and the probability a used unit became unused from one year to the next (i.e., “local extinction,” ε). For each model we fit for the analysis, we specified that the initial ψ in the time series should be a function of the proportion of the unit that is covered by spruce/fir forest – the single most important and consistent predictor of ψ in past analyses. For sake of comparison we fit a base model in which p was specified to be constant for the duration of the survey. However, based on previous work, we considered several other structures for p we anticipated would fit better. We fit models that specified 1) p could vary by survey method (i.e., detection could be different for cameras compared to snowtracking), 2) p could be higher during breeding season when lynx tend to move more and are therefore more likely to be detected by track or at a camera, and 3) p for cameras deployed from 2017–21 could be different than p for other years due to the lure substitution. Additionally we fit a model in which the effect of breeding season was only allowed to act on cameras, not snowtracking. We allowed annual estimates of ε and γ to be different each year (i.e., assuming occupancy dynamics were not random but instead dependent on the year previous and the population is not at equilibrium), which allowed derived ψ to vary as freely as possible given the data. We used Akaike’s Information Criterion (AIC), adjusted for small sample size (Burnham and Anderson 2002) to identify the best-fitting model from this small set. Ultimately, we fit a linear model through the time series of ψ estimates to estimate the slope of the trend in occupancy through time. Ideally we would test other predictions of lynx occupancy to see, for instance, if colonization or extinction were influenced by bark beetles, fire, or the presence of competitors or prey species. However, we do not currently have enough data to test these predictions in addition to assessing trend, which is the highest priority. As has been the case since the inception of our monitoring program, the proportion of the sample unit covered by spruce-fir forest was significantly and positively associated with the initial occupancy estimate in the time series. Even though local colonization and extinction were allowed to vary freely from year to year, annual estimates were near zero and varied little (ε = 0.00–0.08; γ = 0.00–0.10) up until the most recent season when extinction probability was high (ε = 0.40, SE = 0.15). Accordingly, derived occupancy was relatively stable across years (ψ = 0.26–0.35), but dropped to the lowest level observed to date this past season (ψ = 0.23, SE = 0.07). The slope of the trend in occupancy through time was zero (β = 0.001, SE = 0.01; Figure 2), indicating stability. Similar to previous years, detection probability was relatively high for snow tracking surveys (p = 0.65, SE = 0.06), lower for camera surveys (p = 0.22, SE = 0.03) using Pikauba, and lowest for camera surveys utilizing Violator 7 (p = 0.06, SE = 0.02). We estimated that 24% of the sample units in the San Juan’s were occupied by lynx (95% confidence interval: 11–37%) during 2021–22 (Figure 2). The broad spatial distribution of lynx in the San Juan’s remained largely unchanged with the exception of no detection in 3 core snow tracking units where lynx are usually detected (Figure 1). 9 �Table 1. Summary statistics from snow tracking effort. Lynx DNAb 8 6 Km Surveyed (Total) 884 987 Mean Km Surveyed per Visit 20.1 21.9 #CPW Personnelc 30 23 #USFS Personnelc 13 6 18.0 20 8 Season 2014-2015 2015-2016 #Units Surveyed 18 17 #Units with Lynx 7 7 2016-2017 16 8 13 7 5 703 2017-2018 14 7 9 3 1 578 19.3 14 5 2018-2019 14 6 8 2 1 510 19.6 16 5 2019-2020 14 7 11 3 2 640 19.4 15 3 2020-2021 15 9 14 12 7 790 18.8 17 3 2021-2022 13 4 6 5 4 692 18.7 11 3 #Lynx Tracks 12 14 #Genetic Samplesa 8 9 Number of genetic samples (scat, hair, or eDNA) collected via backtracking putative lynx tracks b Number of genetic samples that came back positive for Lynx c Number of staff that participate in the annual effort a Table 2. Summary statistics from camera effort. Season 2014-2015 #Units Surveyed 31 2015-2016 31 2016-2017 33 2017-2018 35 2018-2019 35 2019-2020 36 2020-2021 2021-2022 35 35 #Units With Lynx #Photos (Total) #Photos (Lynx) #Cameras With Lynx 7 7 6 5 6 4 3 5 133,483 101,534 168,705 173,279 201,782 706,074 347,868 576,288 184 455 251 90 59 36 36 116 11 10 10 8 9 4 3 7 10 #CPW Personnel 46 #USFS Personnel 12 33 9 29 9 35 8 31 7 29 6 23 23 5 4 �a) b) Figure 1. Lynx monitoring results for a) the current sampling season (2021–2022) and b) the cumulative monitoring effort (2014–2022), San Juan Mountains, southwest Colorado. Colored units (n = 50) depicted here are those selected at random from the population of units (n = 179) encompassing lynx habitat in the San Juan Mountains. Lynx were detected in 9 units in 2021−2022 and 25 units cumulatively since monitoring began in 2014−2015. 11 �1.00.90.80.7>, u ro 0.6- C C. :::J u u 0 0.50.40.30.20.1 0.0I I I I I I I I 2014 2015 2016 2017 2018 2019 2020 2021 Figure 2. Occupancy estimates (Ψ) and trend (including 95%CI) for Canada lynx in the San Juan Mountains, southwest Colorado. Literature Cited Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd edition. Springer, New York, New York, USA. Ivan, J. S. 2013. Statewide monitoring of Canada lynx in Colorado: evaluation of options. Pages 15–27 in Wildlife Research Report: Mammals. Colorado Parks and Wildlife, Fort Collins, USA. http://cpw.state.co.us/learn/Pages/ResearchMammalsPubs.aspx. Laake, J. L. 2018. Package “RMark”: R Code for Mark Analysis. Version 2.2.5. https://cran.rproject.org/web/packages/RMark/RMark.pdf. MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. Academic Press, Oxford, United Kingdom. R Development Core Team. 2018. No Title. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 Supplem:120–138. 12 �Appendix 1. Model selection results for lynx monitoring data collected in the San Juan Mountains, Colorado, 2014–2022. Rankings are based on Akaike’s Information Criterion adjusted for small sample size (AICc). We mostly sought to tease out best fitting models for detection, allowing constant detection (.), along with effects for survey type (ST), breeding season (B), substituting Violator 7 lure for Pikauba (V), and interactions to allow lure and breeding to act only on cameras. For these models we fixed the initial ψ to be a function of spruce-fir forest while local extinction (ε) and colonization (γ) were estimated annually to allow for non-equilibrium estimates in ψ that depended on previous year’s occupancy state. Post-hoc, we added tested for equilibrium conditions (ε (.) γ (.) ) or that occupancy from year to year was random ({ε = 1- γ}). Model ψ (Prop Spruce/Fir) ε (t) γ (t) p (ST+V+ST*V) ψ (Prop Spruce/Fir) ε (t) γ (t) p (ST+B+V+ST*B+ST*V) ψ (Prop Spruce/Fir) ε (t) γ (t) p (ST+B+V+ ST*V) ψ (Prop Spruce/Fir) ε (t) γ (t) p (ST) ψ (Prop Spruce/Fir) ε (t) γ (t) p (ST+B) ψ (Prop Spruce/Fir) ε (.) γ (.) p (.) ψ (Prop Spruce/Fir) ε (t) γ (t) p (.) ψ (Prop Spruce/Fir) {ε = 1- γ}p (.) 13 AICc 784.65 786.47 786.86 804.81 807.00 859.30 880.01 1038.81 ∆AICc 0.00 1.81 2.21 20.16 22.34 74.64 95.36 254.16 AICc Wts 0.58 0.23 0.19 0.00 0.00 0.00 0.00 0.00 No. Par. 19 21 20 17 18 4 16 9 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Canada lynx monitoring in Colorado 2022 – 2023 Period Covered: December 1, 2022 − April 30, 2023 Principal Investigators: Jake Ivan, Jake.Ivan@state.co.us; Tim Brtis; Lori McCurdy All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. In an effort to restore a viable population of Canada lynx (Lynx canadensis) to the southern portion of their former range, 218 individuals were reintroduced into Colorado from 1999−2006. In 2010, the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) determined that the reintroduction effort met all benchmarks of success and that the population of Canada lynx in the state was apparently viable and self-sustaining. In order to track the persistence of this new population and thus determine the long-term success of the reintroduction, a minimally-invasive, statewide monitoring program is required. From 2014−2023 CPW initiated a portion of the statewide monitoring scheme described in Ivan (2013) by completing surveys in a random sample of monitoring units (n = 50) from the San Juan Mountains in southwest Colorado (n = 179 total units; Figure 1). During the 2022−2023 winter, personnel from CPW and USFS completed the ninth year of monitoring work on this same sample. Thirteen units were sampled via snow-tracking surveys conducted between December 1 and March 31. On each of 1–3 independent occasions, survey crews searched roadways (snow-covered paved roads and logging roads) and trails for lynx tracks. Crews searched the maximum linear distance of roads possible within each survey unit given safety and logistical constraints. Each survey covered a minimum of 10 linear kilometers (6.2 miles) distributed across at least 2 quadrants of the unit. Thirty-five units could not be surveyed via snow tracking. Instead, survey crews deployed 4 passive infrared motion cameras in each of these units during fall 2022. Cameras were lured with visual attractants and scent lure to enhance detection of lynx in the area. Cameras were retrieved during summer or fall 2023 and all photos were archived and viewed by at least 2 observers to determine species present in each. Camera data were then binned such that each of 10 15-day periods from December 1 through April 30 was considered an ‘occasion,’ and any photo of a lynx obtained during a 15-day period was considered a ‘detection’ during that occasion. Surveyors covered 730 km during snow tracking surveys and detected 10 lynx tracks at 5 units (Table 1). This is a slight increase over the program-low of 6 tracks in 4 units observed in 2021–22. Lynx were detected via camera sampling in only one unit during the 2023–23 survey season, which is two fewer units than the previous program low for cameras, which was observed in 2020–21. Snow depths during the 2022–23 season were among the highest ever recorded and a number of cameras were buried for days to weeks, which could have resulted in fewer lynx detections. Also, after 9 seasons of sampling, perhaps resident individuals are developing fatigue to the lures used on the project. In response to the potential for lure fatigue, 117 cameras were passively (i.e., no lure) deployed along roads, trails, and other potential travel routes during fall 2023 in 16 camera units that have had lynx detections in the past. Deployments followed protocols established by (King et al. 2020) and (Anderson et al. 2023). These cameras will be retrieved in summer 2024. Detections at these deployments, and not at traditional camera stations in the same unit, would support the notion that lynx are exhibiting lure fatigue, and future 2 �sampling could switch to passive sampling to capture lynx moving along natural travel routes rather than luring them to a predetermined camera set. Given the program-low in snowtracking detections in 2021– 22, and program-low in camera detections this season (2022–23), it is also possible that lynx distribution declined sharply over the past two survey seasons, which would indicate a decline in the population as well. Lynx were once again detected during snowtrack surveys at Molas Pass and South Mineral, after having gone undetected there in 2021–22. Cameras picked up lynx near Wolf Creek Pass for only the 3rd time in 9 years of sampling, but failed to detect lynx at Rio Grand Reservoir, Lizard Head Pass, and Conejos Peak for only the 2nd or 3rd time since the monitoring program began (Figure 1). We used the R package (R Development Core Team 2018) ‘RMark’ (Laake 2018) to fit multipleseason (i.e., “dynamic”) occupancy models (MacKenzie et al. 2006) to our survey data using program MARK (White and Burnham 1999). Thus, we estimated the derived probability of a unit being occupied (ψ), or used, by lynx over the course of the winter, along with the probability of detecting a lynx (p) given that the unit was occupied, the probability a unit that was unused in one year was used the next (i.e., “local colonization,” γ), and the probability a used unit became unused from one year to the next (i.e., “local extinction,” ε). For each model we fit for the analysis, we specified that the initial ψ in the time series should be a function of the proportion of the unit that is covered by spruce/fir forest – the single most important and consistent predictor of ψ in past analyses. For sake of comparison we fit a base model in which p was specified to be constant for the duration of the survey. However, based on previous work, we considered several other structures for p we anticipated would fit better. We fit models that specified 1) p could vary by survey method (i.e., detection could be different for cameras compared to snowtracking), 2) p could be higher during breeding season when lynx tend to move more and are therefore more likely to be detected by track or at a camera, and 3) p for cameras deployed from 2017–21 could be different than p for other years due to the lure substitution. Additionally we fit a model in which the effect of breeding season was only allowed to act on cameras, not snowtracking. We allowed annual estimates of ε and γ to be different each year (i.e., assuming occupancy dynamics were not random but instead dependent on the year previous and the population is not at equilibrium), which allowed derived ψ to vary as freely as possible given the data. We used Akaike’s Information Criterion (AIC), adjusted for small sample size (Burnham and Anderson 2002) to identify the best-fitting model from this small set. Ultimately, we fit a linear model through the time series of ψ estimates to estimate the slope of the trend in occupancy through time. Ideally we would test other predictions of lynx occupancy to see, for instance, if colonization or extinction were influenced by bark beetles, fire, or the presence of competitors or prey species. However, we do not currently have enough data to test these predictions in addition to assessing trend, which is the highest priority. As has been the case since the inception of our monitoring program, the proportion of the sample unit covered by spruce-fir forest was positively associated with the initial occupancy estimate in the time series. Even though local colonization and extinction were allowed to vary freely from year to year, annual estimates were near zero and varied little (ε = 0.00–0.11; γ = 0.00–0.10) up until the most recent 2 seasons when extinction probability was high (ε21–22 = 0.36, SE = 0.18; ε22–23 = 0.73, SE = 0.17). Accordingly, derived occupancy was relatively stable across years (ψ = 0.25–0.34), but dropped to the lowest level observed to date this past season (ψ = 0.11, SE = 0.05). The slope of the trend in occupancy through time was slightly negative but not statistically different from zero (β = -0.007, SE = 0.01; Figure 2). Similar to previous years, detection probability was relatively high for snow tracking surveys (p = 0.65, SE = 0.06), lower for camera surveys (p = 0.22, SE = 0.03) using Pikauba, and lowest for camera surveys utilizing Violator 7 (p = 0.06, SE = 0.02). We estimated that 11% of the sample units in the San Juan’s were occupied by lynx (95% confidence interval: 2–20%) during 2022–23 (Figure 2). 3 �Table 1. Summary statistics from snow tracking effort. Lynx DNAb 8 Km Surveyed (Total) 884 Mean Km Surveyed per Visit 20.1 #CPW Personnelc 30 #USFS Personnelc 13 Season 2014-2015 #Units Surveyed 18 #Units with Lynx 7 2015-2016 17 7 14 9 6 987 21.9 23 6 2016-2017 16 8 13 7 5 703 18.0 20 8 2017-2018 14 7 9 3 1 578 19.3 14 5 2018-2019 14 6 8 2 1 510 19.6 16 5 2019-2020 14 7 11 3 2 640 19.4 15 3 2020-2021 15 9 14 12 7 790 18.8 17 3 2021-2022 13 4 6 5 4 692 18.7 11 3 2022-2023 15 5 10 9 7 730 18.3 15 2 #Lynx Tracks 12 #Genetic Samplesa 8 Number of genetic samples (scat, hair, or eDNA) collected via backtracking putative lynx tracks b Number of genetic samples that came back positive for Lynx c Number of staff that participate in the annual effort a Table 2. Summary statistics from camera effort. Season 2014-2015 #Units Surveyed 31 2015-2016 31 2016-2017 33 2017-2018 35 2018-2019 35 2019-2020 36 2020-2021 35 2021-2022 35 2022-2023 35 #Units With Lynx #Photos (Total) #Photos (Lynx) #Cameras With Lynx 7 7 6 5 6 4 3 5 1 133,483 101,534 168,705 173,279 201,782 706,074 347,868 576,288 531,083 184 455 251 90 59 36 36 116 4 11 10 10 8 9 4 3 7 1 4 #CPW Personnel 46 #USFS Personnel 12 33 9 29 9 35 8 31 7 29 6 23 5 23 4 31 3 �a) b) Figure 1. Lynx monitoring results for a) the current sampling season (2022–2023) and b) the cumulative monitoring effort (2014–2023), San Juan Mountains, southwest Colorado. Colored units (n = 50) depicted here are those selected at random from the population of units (n = 179) encompassing lynx habitat in the San Juan Mountains. Lynx were detected in 6 units in 2022−2023 and 25 units cumulatively since monitoring began in 2014−2015. 5 �1.00.9 0.8 0.7- > u 0.6 - <ti 0.. ::::, 0.5 - 0 0.4- C: u u 0.3 0.2 0.1 0.0I I I I I I I I I 2014 2015 2016 2017 2018 2019 2020 2021 202;1 Figure 2. Occupancy estimates (Ψ) and trend (including 95% CI) for Canada lynx in the San Juan Mountains, southwest Colorado. LITERATURE CITED Anderson, A. K., J. S. Waller, and D. H. Thornton. 2023. Canada lynx occupancy and density in Glacier National Park. Journal of Wildlife Management e22383:1–24. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd ed. Springer, New York, New York, USA. Ivan, J. S. 2013. Statewide monitoring of Canada lynx in Colorado: evaluation of options. Pages 15–27 in Wildlife research report - mammals. Colorado Parks and Wildlife., Fort Collins, Colorado, USA. http://cpw.state.co.us/learn/Pages/ResearchMammalsPubs.aspx. King, T. W., C. Vynne, D. Miller, S. Fisher, S. Fitkin, J. Rohrer, J. I. Ransom, and D. Thornton. 2020. Will lynx lose their edge? Canada lynx occupancy in Washington. Journal of Wildlife Management 84:705–725. Laake, J. L. 2018. Package “RMark”: R Code for Mark Analysis. Version 2.2.5. https://cran.rproject.org/web/packages/RMark/RMark.pdf. MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. Academic Press, Oxford, United Kingdom. R Development Core Team. 2018. No Title. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 Supplem:S120–S138. 6 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Canada lynx monitoring in Colorado 2023 – 2024 Period Covered: December 1, 2023  December 30, 2024 Principal Investigators: Jake Ivan, Jake.Ivan@state.co.us; Tim Brtis All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. In an effort to restore a viable population of Canada lynx (Lynx canadensis) to the southern portion of their former range, 218 individuals were reintroduced into Colorado from 19992006. In 2010, the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) determined that the reintroduction effort met all benchmarks of success and that the population of Canada lynx in the state was apparently viable and self-sustaining. In order to track the persistence of this new population and thus determine the long-term success of the reintroduction, a minimally-invasive, statewide monitoring program is required. From 2014−2024 CPW initiated a portion of the statewide monitoring scheme described in Ivan (2013) by completing surveys in a random sample of monitoring units (n = 50) from the San Juan Mountains in southwest Colorado (n = 179 total units; Figure 1). During the 2023−2024 winter, personnel from CPW and USFS completed the tenth year of monitoring work on this same sample. Fifteen units were sampled via snow-tracking surveys conducted between December 1 and March 31. On each of 1–3 independent occasions, survey crews searched roadways (snow-covered paved roads and logging roads) and trails for lynx tracks. Crews searched the maximum linear distance of roads possible within each survey unit given safety and logistical constraints. Each survey covered a minimum of 10 linear kilometers (6.2 miles) distributed across at least 2 quadrants of the unit. Thirty-five units could not be surveyed via snow tracking. Instead, survey crews deployed 4 passive infrared motion cameras in each of these units during fall 2023. Cameras were lured with visual attractants and scent lure to enhance detection of lynx in the area. Cameras were retrieved during summer or fall 2024 and all photos were archived and viewed by at least 2 observers to determine species present in each. Camera data were then binned such that each of 10 15-day periods from December 1 through April 30 was considered an ‘occasion,’ and any photo of a lynx obtained during a 15-day period was considered a ‘detection’ during that occasion. Surveyors covered 826 km during snow tracking surveys and detected 11 lynx tracks at 6 units (Table 1). This is considered a rebound from the program-low of 6 tracks in 4 units observed in 2021–22. Lynx were detected via camera sampling in 3 units during the 2023–24 survey season, which also represents a rebound from the previous program low (1 unit) for cameras, which was observed in 2022– 23. In response to program low in camera detections during the 2022-23 winter, and our thinking that a potential explanation could be fatigue to the lured camera sets in use for nearly a decade, 117 cameras were passively (i.e., no lure) deployed along roads, trails, and other potential travel routes during fall 2023 in 16 camera units that have had lynx detections in the past. Deployments followed protocols established by (King et al. 2020) and (Anderson et al. 2023). These cameras were retrieved in summer 2024. During the usual analysis period we recorded 384 lynx detections at 11 units, 35 cameras. That is, the passive sampling scheme produced lynx detection in more units and cameras than at any point during the decade-long sampling using traditional lured sets. It also produced more lynx photos than all but one 2 �year during that same timeframe. Coupled with the increased efficiency and reduced cost of deploying passive sets, the monitoring program will transition to this methodology in coming years. Lynx were once again detected in the upper Rio Grande Reservoir and Conejos Peak areas, after having gone undetected there in 2022–23. However, no lynx were detected near Lizard Head Pass or west of Lake City for the second year in a row (Figure 1). We used the R package (R Development Core Team 2018) ‘RMark’ (Laake 2018) to fit multipleseason (i.e., “dynamic”) occupancy models (MacKenzie et al. 2006) to our survey data using program MARK (White and Burnham 1999). Thus, we estimated the derived probability of a unit being occupied (), or used, by lynx over the course of the winter, along with the probability of detecting a lynx (p) given that the unit was occupied, the probability a unit that was unused in one year was used the next (i.e., “local colonization,” ), and the probability a used unit became unused from one year to the next (i.e., “local extinction,” ). For each model we fit for the analysis, we specified that the initial  in the time series should be a function of the proportion of the unit that is covered by spruce/fir forest – the single most important and consistent predictor of  in past analyses. For sake of comparison we fit a base model in which p was specified to be constant for the duration of the survey. However, based on previous work, we considered several other structures for p we anticipated would fit better. We fit models that specified 1) p could vary by survey method (i.e., detection could be different for cameras compared to snowtracking), 2) p could be higher during breeding season when lynx tend to move more and are therefore more likely to be detected by track or at a camera, and 3) p for cameras deployed from 2017–21 could be different than p for other years due to a lure substitution. Additionally we fit a model in which the effect of breeding season was only allowed to act on cameras, not snowtracking. We allowed annual estimates of  and  to be different each year (i.e., assuming occupancy dynamics were not random but instead dependent on the year previous and the population is not at equilibrium), which allowed derived  to vary as freely as possible given the data. We used Akaike’s Information Criterion (AIC), adjusted for small sample size (Burnham and Anderson 2002) to identify the best-fitting model from this small set. Ultimately, we fit a linear model through the time series of  estimates to estimate the slope of the trend in occupancy through time. Ideally we would test other predictions of lynx occupancy to see, for instance, if colonization or extinction were influenced by bark beetles, fire, or the presence of competitors or prey species. However, we do not currently have enough data to test these predictions in addition to assessing trend, which is the highest priority. As has been the case since the inception of our monitoring program, the proportion of the sample unit covered by spruce-fir forest was positively associated with the initial occupancy estimate in the time series. Even though local colonization and extinction were allowed to vary freely from year to year, annual estimates were near zero and varied little ( = 0.00–0.11;  = 0.00–0.8) except for the the interval between the 2021–22 and 2022–23 seasons when extinction probability was high (21–22 = 0.35, SE = 0.14). Accordingly, derived occupancy was relatively stable across years ( = 0.25–0.31), but dropped to a program low the past two winters ( ≈ 0.17, SE = 0.05). The slope of the trend in occupancy through time was slightly negative but not statistically different from zero ( = -0.008, SE = 0.01; Figure 2). Similar to previous years, detection probability was relatively high for snow tracking surveys (p = 0.60, SE = 0.05), lower for camera surveys (p = 0.21, SE = 0.02) using Pikauba, and lowest for camera surveys utilizing Violator 7 (p = 0.07, SE = 0.02). We estimated that 17% of the sample units in the San Juan’s were occupied by lynx (95% confidence interval: 6–27%) during 2023–24 (Figure 2). 3 �Table 1. Summary statistics from snow tracking effort. Lynx DNAb 8 Km Surveyed (Total) 884 Mean Km Surveyed per Visit 20.1 #CPW Personnelc 30 #USFS Personnelc 13 Season 2014-2015 #Units Surveyed 18 #Units with Lynx 7 2015-2016 17 7 14 9 6 987 21.9 23 6 2016-2017 16 8 13 7 5 703 18.0 20 8 2017-2018 14 7 9 3 1 578 19.3 14 5 2018-2019 14 6 8 2 1 510 19.6 16 5 2019-2020 14 7 11 3 2 640 19.4 15 3 2020-2021 15 9 14 12 7 790 18.8 17 3 2021-2022 13 4 6 5 4 692 18.7 11 3 2022-2023 15 5 10 9 7 730 18.3 15 2 2023-2024 15 6 11 10 6 826 19.7 14 3 #Lynx Tracks 12 #Genetic Samplesa 8 Number of genetic samples (scat, hair, or eDNA) collected via backtracking putative lynx tracks b Number of genetic samples that came back positive for lynx c Number of staff that participate in the annual effort a Table 2. Summary statistics from camera effort. Season 2014-2015 #Units Surveyed 31 2015-2016 31 2016-2017 33 2017-2018 35 2018-2019 35 2019-2020 36 2020-2021 35 2021-2022 35 2022-2023 35 2023-2024 35 #Units With Lynx #Photos (Total) #Photos (Lynx) #Cameras With Lynx 7 7 6 5 6 4 3 5 1 3 133,483 101,534 168,705 173,279 201,782 706,074 347,868 576,288 531,083 601,371 184 455 251 90 59 36 36 116 4 336 11 10 10 8 9 4 3 7 1 4 4 #CPW Personnel 46 #USFS Personnel 12 33 9 29 9 35 8 31 7 29 6 23 5 23 4 31 3 24 3 �a) b) Figure 1. Lynx monitoring results for a) the current sampling season (2023–2024) and b) the cumulative monitoring effort (2014–2024), San Juan Mountains, southwest Colorado. Colored units (n = 50) depicted here are those selected at random from the population of units (n = 179) encompassing lynx habitat in the San Juan Mountains. Lynx were detected in 8 units in 2023−2024 and 25 units cumulatively since monitoring began in 2014−2015. 5 �1.00.90.80.7- > u 0.6- RJ 0.. ::::, 0.5- 0 0.4- C u u 0.30.20.10.0I I I I I I I I I I 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 Figure 2. Occupancy estimates () and trend (including 95%CI for each) for Canada lynx in the San Juan Mountains, southwest Colorado. Literature Cited Anderson, A. K., J. S. Waller, and D. H. Thornton. 2023. Canada lynx occupancy and density in Glacier National Park. Journal of Wildlife Management e22383:1–24. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd edition. Springer, New York, New York, USA. Ivan, J. S. 2013. Statewide monitoring of Canada lynx in Colorado: evaluation of options. Pages 15–27 in Wildlife Research Report - Mammals. Colorado Parks and Wildlife., Fort Collins, Colorado, USA.https://spl.cde.state.co.us/artemis/nrserials/nr616internet/nr616201213internet.pdf. King, T. W., C. Vynne, D. Miller, S. Fisher, S. Fitkin, J. Rohrer, J. I. Ransom, and D. Thornton. 2020. Will lynx lose their edge? Canada lynx occupancy in Washington. Journal of Wildlife Management 84:705–725. Laake, J. L. 2018. Package “RMark”: R Code for Mark Analysis. Version 2.2.5. https://cran.rproject.org/web/packages/RMark/RMark.pdf. MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. Academic Press, Oxford, United Kingdom. R Development Core Team. 2018. No Title. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 Supplem:120–138. 6 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Canada lynx monitoring in Colorado, Winter 2024–2025 Period Covered: January 1, 2025 − December 31, 2025 Principal Investigators: Jake Ivan, Jake.Ivan@state.co.us; Tim Brtis; Bob Inman; All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. In an effort to restore a viable population of Canada lynx (Lynx canadensis) to the southern portion of their former range, 218 individuals were reintroduced into Colorado from 1999−2006. In 2010, the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) determined that the reintroduction effort met all benchmarks of success and that the population of Canada lynx in the state was apparently viable and self-sustaining. In order to track the persistence of this new population and thus determine the long-term success of the reintroduction, a minimally-invasive statewide monitoring program was required. Beginning in 2014 CPW initiated a portion of the statewide monitoring scheme described in Ivan (2013) by completing surveys in a random sample of monitoring units (n = 50) from the San Juan Mountains in southwest Colorado (n = 179 total units; Figure 1). During the 2024−2025 winter, personnel from CPW and USFS completed the eleventh year of monitoring work on this same sample. Fourteen units were scheduled for snow-tracking surveys between December 1 and April 30. On each of 1–3 independent occasions, survey crews searched roadways (snow-covered paved roads and logging roads) and trails for lynx tracks. Crews searched the maximum linear distance of roads possible within each survey unit given safety and logistical constraints. Each survey covered a minimum of 10 linear km (6.2 miles) distributed across at least 2 quadrants of the unit. Thirty-six units could not be surveyed via snow tracking. Instead, survey crews deployed 4 passive infrared motion cameras in each of these units during fall 2024. Cameras were lured with visual attractants and scent lure to enhance detection of lynx in the area. Cameras were retrieved during summer or fall 2025. All photos were archived, filtered to “animals” using the AI platform Pytorch-Wildlife to run MegaDetector v5a (Hernandez et al. 2024) and viewed by at least 2 observers to determine species present in each. Camera data were then binned such that each of 10 15-day periods from December 1 through April 30 was considered an ‘occasion,’ and any photo of a lynx obtained during a 15-day period was considered a ‘detection’ during that occasion. Surveyors covered 495 km during snow tracking surveys and detected 9 lynx tracks at 4 units (Table 1). This represents the lowest snow-tracking effort (km surveyed) and matches the fewest number of snow-tracking units with lynx, since the inception of the monitoring program. However, 2 units in Area 17 and 1 unit in Area 16 were not surveyed due to a lack of snow and poor tracking conditions. Lynx were detected via camera sampling in 4 units during the 2024–2025 survey season, which represents a continued rebound from the previous program low (1 unit) for cameras, which was observed in 2022– 2023 (Table 2). Lynx were detected in the La Garita Mountains northeast of Creede for the first time since the 2021–2022 survey season and were again detected near Lizard Head Pass after having gone undetected there for two winters (Figure 1). In response to a program low in camera detections during the 2022–2023 winter, and a potential explanation that fatigue to the lured camera sets in use for nearly a decade could be the cause, we initiated 2 �a research project to assess an alternate approach to camera sampling for lynx that has been successful elsewhere within their range. Accordingly, 117 cameras were passively (i.e., no lure) deployed along roads, trails, and other potential travel routes during fall 2023 in 16 camera units that have had lynx detections in the past. These “research deployments” were in addition to our concurrent camera or snowtracking efforts and followed protocols established by King et al. (2020) and Anderson et al. (2023), including twice the cameras per unit as our traditional approach. Camera data were retrieved in summer 2024, but the cameras were left in place with original batteries to sample through winter 2024–25. During the usual analysis period we recorded 384 lynx detections at 35 cameras in 11 units during 2023–2024 and 691 lynx detections at 32 cameras in 8 units during 2024–2025. Thus, the passive sampling scheme performed as well or better than our traditional lured sets but incurred less effort to deploy and was free from concerns regarding lure fatigue. Therefore, the monitoring program transitioned to this approach for the 2025–2026 deployments. Only 9% of cameras were dead when they were retrieved during summer 2025 after 2 years of deployment as passive sets. Another 25% registered “0 battery” but were still operating. We used the R package (R Development Core Team 2018) ‘RMark’ (Laake 2018) to fit multipleseason (i.e., “dynamic”) occupancy models (MacKenzie et al. 2006) to our survey data using program MARK (White and Burnham 1999). Thus, we estimated the derived probability of a unit being occupied (ψ), or used, by lynx over the course of the winter, along with the probability of detecting a lynx (p) given that the unit was occupied, the probability a unit that was unused in one year was used the next (i.e., “local colonization,” γ), and the probability a used unit became unused from one year to the next (i.e., “local extinction,” ε). For each model we fit for the analysis, we specified that the initial ψ in the time series should be a function of the proportion of the unit that is covered by spruce/fir forest – the single most important and consistent predictor of ψ in past analyses. For sake of comparison, we fit a base model in which p was specified to be constant for the duration of the survey. However, based on previous work, we considered several other structures for p we anticipated would fit better. We fit models that specified 1) p could vary by survey method (i.e., detection could be different for cameras compared to snowtracking), 2) p could be higher during breeding season when lynx tend to move more and are therefore more likely to be detected by track or at a camera, and 3) p for cameras deployed from 2017–21 could be different than p for other years due to a lure substitution. Additionally, we fit a model in which the effect of breeding season was only allowed to act on cameras, not snowtracking. We allowed annual estimates of ε and γ to be different each year (i.e., assuming occupancy dynamics were not random but instead dependent on the year previous and the population is not at equilibrium), which allowed derived ψ to vary as freely as possible given the data. We used Akaike’s Information Criterion (AIC), adjusted for small sample size (Burnham and Anderson 2002) to identify the best-fitting model from this small set. Ultimately, we fit a linear model through the time series of ψ estimates to estimate the slope of the trend in occupancy through time. Alternative models to further test predictions relating lynx occupancy to bark beetles, fire, or the presence of competitors or prey species are ongoing but here we focus on status and trend. As has been the case since the inception of our monitoring program, the proportion of the sample unit covered by spruce-fir forest was positively associated with the initial occupancy estimate in the time series. Even though local colonization and extinction were allowed to vary freely from year to year, annual estimates were near zero and varied little (ε = 0.00–0.9; γ = 0.00–0.8) except for the intervals between the 2021–22 and 2022–23 seasons when extinction probability was high (ε21–22 = 0.30, SE = 0.14; ε22–23 = 0.20, SE = 0.13). Derived occupancy oscillated between ψ = 0.17–0.32) and appears to be rebounding after hitting a low of 0.17 during winter 2023–2024. The slope of the trend in occupancy through time was slightly negative but not statistically different from zero (β = -0.005, SE = 0.01; Figure 2). Similar to previous years, detection probability was relatively high for snow tracking surveys (p = 0.62, SE = 0.05), lower for camera surveys (p = 0.19, SE = 0.02) using Pikauba lure, and lowest for camera surveys utilizing Violator 7 lure (p = 0.06, SE = 0.02). We estimated that 20% of the sample units in the San Juan’s were occupied by lynx (95% confidence interval: 8–32%) during 2024–25 (Figure 2). 3 �Literature Cited Anderson, A. K., J. S. Waller, and D. H. Thornton. 2023. Canada lynx occupancy and density in Glacier National Park. Journal of Wildlife Management e22383:1–24. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd edition. Springer, New York, New York, USA. Ivan, J. S. 2013. Statewide monitoring of Canada lynx in Colorado: evaluation of options. Pages 15–27 Wildlife Research Report - Mammals. Colorado Parks and Wildlife., Fort Collins, Colorado, USA. https://cpw.cvlcollections.org/files/original/e43c3c5a908d16f25ff25c5a42ff1180.pdf. Hernandez, A., Z. Miao, L. Vargas, R. Dodhia, P. Arbelaez, and J. M. L. Ferres. 2024. Pytorch-Wildlife: A collaborative deep learning framework for conservation. arXiv King, T. W., C. Vynne, D. Miller, S. Fisher, S. Fitkin, J. Rohrer, J. I. Ransom, and D. Thornton. 2020. Will lynx lose their edge? Canada lynx occupancy in Washington. Journal of Wildlife Management 84:705–725. Laake, J. L. 2018. Package “RMark”: R Code for Mark Analysis. Version 2.2.5. https://cran.rproject.org/web/packages/RMark/RMark.pdf. MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. Academic Press, Oxford, United Kingdom. R Development Core Team. 2018. No Title. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 Supplem:120–138. 4 �Table 1. Summary statistics from snow tracking effort. Lynx DNAb 8 Km Surveyed (Total) 884 Mean Km Surveyed per Visit 20.1 #CPW Personnelc 30 #USFS Personnelc 13 9 6 987 21.9 23 6 13 7 5 703 18.0 20 8 9 3 1 578 19.3 14 5 6 8 2 1 510 19.6 16 5 7 11 3 2 640 19.4 15 3 Season 2014-2015 #Units Surveyed 18 #Units with Lynx 7 #Lynx Tracks 12 #Genetic Samplesa 8 2015-2016 17 7 14 2016-2017 16 8 2017-2018 14 7 2018-2019 14 2019-2020 14 2020-2021 15 9 14 12 7 790 18.8 17 3 2021-2022 13 4 6 5 4 692 18.7 11 3 2022-2023 15 5 10 9 7 730 18.3 15 2 2023-2024 15 6 11 10 6 826 19.7 14 3 2024-2025 11 4 9 9 5 495 18.3 10 2 a Number of genetic samples (scat, hair, or eDNA) collected via backtracking putative lynx tracks Number of genetic samples that came back positive for Lynx c Number of staff that participate in the annual effort b Table 2. Summary statistics from camera effort. a Season 2014-2015 #Units Surveyed 31 2015-2016 31 2016-2017 33 2017-2018 35 2018-2019 35 2019-2020 36 2020-2021 35 2021-2022 35 2022-2023 35 2023-2024 35 2024-2025 36 #Units With Lynx #Photos (Total)a #Photos (Lynx) a #Cameras With Lynx 7 7 6 5 6 4 3 5 1 3 4 29,767 15,357 32,525 65,993 42,796 109,587 69,848 121,076 90,140 85,901 17,924b 262 423 227 53 33 21 36 118 4 336 121 11 10 10 8 9 4 3 7 1 4 5 #CPW Personnel 46 #USFS Personnel 12 33 9 29 9 35 8 31 7 29 6 23 5 23 4 31 3 24 3 24 3 Number photos collected during the survey period (December 1 – April 30). Note that the numbers in these columns are different than those reported from the same columns in previous years. Previously we reported the total photos collected from deployment to retrieval, summed across all cameras. This depiction focuses only on the period of interest and is a fairer comparison between years. b The 2024–2025 field season was the first during which the photos collected during the survey period (December 1 – April 30) were initially run through MegaDetector v5a to cull photos of people and vehicles as well as empty photos. We retained and tagged photos for this project with a confidence level for “animal” >0.20. Previous research using training data from this project suggested that this confidence level was useful for improving efficiency, but also conservative. That is, we would miss very few animals at this level of confidence but would remove most nonanimal photos from the dataset. 5 �a) b) Figure 1. Lynx monitoring results for a) the current sampling season (2024–2025) and b) the cumulative monitoring effort (2014–2025), San Juan Mountains, southwest Colorado. Colored units (n = 50) depicted here are those selected at random from the population of units (n = 179) encompassing lynx habitat in the San Juan Mountains. Lynx were detected in 8 units in 2024−2025 and 25 units cumulatively since monitoring began in 2014−2015. 6 �Figure 2. Occupancy estimates (Ψ) and trend (including 95% CI for each) for Canada lynx in the San Juan Mountains, southwest Colorado. 7 � https://cpw.cvlcollections.org/files/original/5ec8df0f0e754e947d575c8931367848.pdf 527ce74442b45924ff49391ec9954940 PDF Text Text Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Evaluation of accelerometer collars and methods development for domestic cattle Period Covered: January 1, 2023-December 31, 2023 Principal Investigators: Ellen Brandell, ellen.brandell@state.co.us All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. Livestock production is an important component of Colorado’s economy (University of Arkansas accessed 2023, Bureau of Land Management accessed 2023), as well as ingrained in the state’s culture and heritage – cattle production in particular. Colorado citizens are concerned about the effects of reestablishing gray wolves (Canis lupus) on livestock (Niemiec et al. 2022), and given the geographic constraints of CRS 33-2-105.8 (Colorado General Assembly 2020, CPW 2023) and suitable wolf habitat in Colorado (Ditmer et al. 2022), wolves and livestock will spatially overlap in western Colorado. Wolves may affect livestock both directly and indirectly; direct effects include depredation, which has already occurred in the state. Indirect effects, such as increased stress or vigilance behavior, are much more difficult to observe and quantify. Indirect effects of wolves on cattle have been documented in other western states or laboratory experiments, such as decreased weight gain (Ramler et al. 2014) and increased stress (Cooke et al. 2013). However, these negative effects are not ubiquitous across studies, and the majority of published literature on this topic lacks a mechanistic understanding. For example, cattle movement rates (Laporte et al. 2010, Bailey et al. 2018) and physiology (Cooke et al. 2013) in response to wolf presence have been studied, but unless changes in movement rates or physiology have direct implications for weight gain, pregnancy rates, or animal health, it might not be important to a producer or impact the operation’s economics. In a future research project, we aim to link cattle behavior and movement in response to wolf presence to cattle stress levels, weight gain, and pregnancy rates. Quantifying the mechanisms of changes in cattle stress, weight gain, and pregnancy rates is critical for identifying whether a causal relationship exists between wolf exposure and cattle responses, the magnitude of this effect, and subsequent consequences for producers’ bottom line. However, before we can launch a research project, we need to test the field equipment and develop data collection methods. In spring 2023, we began a methods testing project to evaluate GPS and accelerometer collars on beef cattle. We had three goals of this methods testing project: (1) assess proper fit of GPS/accelerometer collars on both adult female cows and calves throughout the grazing season; (2) develop methods to calibrate accelerometer data to common cattle behaviors; (3) test field equipment, and improve equipment as needed. We outfitted 20 cows with collars in May and June 2023. More specifically, we collared and monitored 10 cow-calf pairs from two cattle operations (one in Northeast Colorado, one in Northwest Colorado). Cow-calf pairs are of interest as calves are the most vulnerable to predation. Data collection ranged from approximately 1-5 months while cattle were grazing on allotments (e.g., USFS, BLM). We obtained a high-quality visual observation of all collared animals at least twice per month, and often multiple times a week. Visual observations were obtained by CPW staff, the livestock owner, or ranch personnel. Animal condition and collar fit was assessed visually, and with associated photos and video 36 �where possible. We used this information to determine if collars needed to be periodically adjusted. Calf collars had a section of elastic to allow for growth in between adjustments. Accelerometers collect triaxial data (x, y, and z axes) 8 times per second (8 Hz). Accelerometers have been used on cattle and other grazing species to identify behaviors and quantify time budgets (Riaboff et al. 2020, Riaboff et al. 2022). We will create time budgets by specifying cattle behaviors such as feeding, resting, ruminating, moving, acting vigilant, and grooming. We will calibrate cattle behavior by performing focal follows, where an individual cow or calf is observed for a predetermined amount of time (20 minutes), and the timing of different behaviors is recorded (Riaboff et al. 2022). One adult female cow per operation was outfitted with a camera collar as well to provide constant behavioral validation data. The observation data is compared with the triaxial data patterns, and unique data patterns are labeled as specific behaviors using machine learning algorithms (Riaboff et al. 2020, Riaboff et al. 2022). Collars will also collect geospatial data at short, regular intervals to calculate distance moved and movement rates (Bailey et al. 2018). We are currently organizing and analyzing these data. Experiences from this methods testing project will help guide equipment decisions, data collection methods, and fieldwork as we develop a larger-scale research project focusing on indirect effects of predators on livestock. Literature Cited Bailey, D. W., M. G. Trotter, C. W. Knight, and M. G. Thomas. 2018. Use of GPS tracking collars and accelerometers for rangeland livestock production research. Translational Animal Science 2:81-88. Bureau of Land Management. Colorado rangeland management and grazing. <https://www.blm.gov/programs/natural-resources/rangeland-and-grazing/rangeland-health/colorado>. Accessed 2023. Clark, P. E., D. E. Johnson, L. L. Larson, M. Louhaichi, T. Roland, and J. Williams. 2017. Effects of wolf presence on daily travel distance of range cattle. Rangeland ecology & management 70:657-665. Colorado General Assembly. 2020 Colorado ballot analysis, proposition 114, reintroduction and management of gray wolves. <https://leg.colorado.gov/ballots/reintroduction-and-management-graywolves>. Accessed 2023. Colorado Parks and Wildlife. 2023. Colorado wolf restoration and management plan. Denver, USA. Cook, R. C., J. G. Cook, D. J. Vales, B. K. Johnson, S. M. Mccorquodale, L. A. Shipley, R. A. Riggs, L. L. Irwin, S. L. Murphie, and B. L. Murphie. 2013. Regional and seasonal patterns of nutritional condition and reproduction in elk. Wildlife Monographs 184:1-45. Ditmer, M. A., G. Wittemyer, S. W. Breck, and K. R. Crooks. 2022. Defining ecological and socially suitable habitat for the reintroduction of an apex predator. Global Ecology and Conservation 38:e02192. Laporte, I., T. B. Muhly, J. A. Pitt, M. Alexander, and M. Musiani. 2010. Effects of wolves on elk and cattle behaviors: implications for livestock production and wolf conservation. PLoS One 5:e11954. Niemiec, R., R. E. Berl, M. Gonzalez, T. Teel, J. Salerno, S. Breck, C. Camara, M. Collins, C. Schultz, and D. Hoag. 2022. Rapid changes in public perception toward a conservation initiative. Conservation Science and Practice 4:e12632. Ramler, J. P., M. Hebblewhite, D. Kellenberg, and C. Sime. 2014. Crying wolf? A spatial analysis of wolf location and depredations on calf weight. American Journal of Agricultural Economics 96:631-656. Riaboff, L., S. Couvreur, A. Madouasse, M. Roig-Pons, S. Aubin, P. Massabie, A. Chauvin, N. Bédère, and G. Plantier. 2020. Use of predicted behavior from accelerometer data combined with GPS data to explore the relationship between dairy cow behavior and pasture characteristics. Sensors 20:4741. Riaboff, L., L. Shalloo, A. F. Smeaton, S. Couvreur, A. Madouasse, and M. T. Keane. 2022. Predicting livestock behaviour using accelerometers: A systematic review of processing techniques for ruminant behaviour prediction from raw accelerometer data. Computers and Electronics in Agriculture 192:106610. 37 �University of Arkansas Division of Agriculture Research & Extension. Economic impact of agriculture. <https://economic-impact-of-ag.uada.edu/colorado/>. Accessed 2023. 38 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Evaluation of accelerometer collars and methods development for domestic cattle Period Covered: January 1, 2024-December 31, 2024 Principal Investigator: Ellen Brandell, ellen.brandell@state.co.us All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. Livestock production is an important component of Colorado’s economy (University of Arkansas, Bureau of Land Management), as well as ingrained in the state’s culture and heritage– cattle production in particular. Colorado citizens are concerned about the effects of re-establishing gray wolves (Canis lupus) on livestock (Niemiec et al. 2022), and given the geographic constraints of CRS 33-2-105.8 (Colorado General Assembly 2020, CPW 2023) and suitable wolf habitat in Colorado (Ditmer et al. 2022), wolves and livestock will spatially overlap in western Colorado. Wolves may affect livestock both directly and indirectly; direct effects include depredation, which has already occurred in the state. Indirect effects, such as increased stress or vigilance behavior, are much more difficult to observe and quantify. Indirect effects of wolves on cattle have been documented in other western states or laboratory experiments, such as decreased weight gain (Ramler et al. 2014) and increased stress (Cooke et al. 2013). However, these negative effects are not ubiquitous across studies, and the majority of published literature on this topic lacks a mechanistic understanding. For example, cattle movement rates (Laporte et al. 2010, Bailey et al. 2018) and physiology (Cooke et al. 2013) in response to wolf presence have been studied, but unless changes in movement rates or physiology have direct implications for weight gain, pregnancy rates, or animal health, it might not be important to a producer or impact the operation’s economics. In a future research project, we aim to link cattle behavior and movement in response to wolf presence to cattle stress levels, weight gain, and pregnancy rates. Quantifying the mechanisms of changes in cattle stress, weight gain, and pregnancy rates is critical for identifying whether a causal relationship exists between wolf exposure and cattle responses, the magnitude of this effect, and subsequent consequences for producers’ bottom line. However, before we can launch a research project, we need to test the field equipment and develop data collection methods. In spring 2023, we began a methods testing project is to evaluate GPS and accelerometer collars on beef cattle. We had three goals of this methods testing project: (1) Assess proper fit of GPS/accelerometer collars on both adult female cows and calves throughout the grazing season; (2) develop methods to calibrate accelerometer data to common cattle behaviors; and (3) test field equipment, and improve equipment as needed. We applied what we learned in 2023 to a 2024 field season where we refined our data collection protocols. We outfitted 20 cows with collars in May and June 2023. More specifically, we collared and monitored 10 cow-calf pairs from two cattle operations (one in Northeast Colorado, one in Northwest Colorado). Cow-calf pairs are of interest as calves are the most vulnerable to predation. In May 2024, we outfitted collars on 8 cows from one operation in Northwest Colorado. Data collection ranged from approximately 1-5 months while cattle were grazing on allotments (e.g., USFS, BLM) or privately owned pastures. We obtained a high-quality visual observation of all collared animals at least twice per month, and often multiple times a week. Visual observations were obtained by CPW staff, the livestock owner, or ranch personnel. Animal condition and collar fit was assessed visually, and with associated photos and 33 �video where possible. We used this information to determine if collars needed to be periodically adjusted. Calf collars had a section of elastic to allow for growth in between adjustments. Accelerometers collected triaxial data (x, y, and z axes) 8 times per second (8 Hz). Accelerometers have been used on cattle and other grazing species to identify behaviors and quantify time budgets (Riaboff et al. 2020, Riaboff et al. 2022). We will create time budgets by specifying cattle behaviors such as feeding, resting, ruminating, moving, acting vigilant, and grooming. We will calibrate cattle behavior by correlating observer behavior with accelerometer data. We performed focal follows, where an individual cow or calf is observed for a predetermined amount of time (20 minutes), and recorded the timing of different behaviors (Riaboff et al. 2022). In 2023, one adult female cow per operation was outfitted with a camera collar as well to provide constant behavioral validation data. Currently, we are assessing the observation data and labeling as specific behaviors; the next step is to use machine learning algorithms to correspond these behaviors with triaxial data patterns (Riaboff et al. 2020, Riaboff et al. 2022). Collars also collected geospatial data at short, regular intervals (5 minutes), which will be used to calculate distance moved and movement rates (Bailey et al. 2018). We are currently organizing and cleaning these locational data. Experiences from this methods testing project will help guide equipment decisions, data collection methods, and fieldwork as we develop a larger-scale research project focusing on indirect effects of predators on livestock. Data collected in 2023 and 2024 should be adequate to develop time budgets and behavioral models. We plan to analyze these data before moving forward with a full research project. The timeline for this research is also dependent on wolf activity in the state and partnerships with livestock producers, and therefore we will be adaptable moving forward. Literature Cited: Bailey, D. W., M. G. Trotter, C. W. Knight, and M. G. Thomas. 2018. Use of GPS tracking collars and accelerometers for rangeland livestock production research. Translational Animal Science 2:81-88. Bureau of Land Management. Colorado rangeland management and grazing. <https://www.blm.gov/programs/natural-resources/rangeland-and-grazing/rangelandhealth/colorado>. Accessed 2023. Clark, P. E., D. E. Johnson, L. L. Larson, M. Louhaichi, T. Roland, and J. Williams. 2017. Effects of wolf presence on daily travel distance of range cattle. Rangeland ecology & management 70:657-665. Colorado General Assembly. 2020 Colorado ballot analysis, proposition 114, reintroduction and management of gray wolves. <https://leg.colorado.gov/ballots/reintroduction-and-management-graywolves>. Accessed 2023. Colorado Parks and Wildlife. 2023. Colorado wolf restoration and management plan. Denver, USA. Cook, R. C., J. G. Cook, D. J. Vales, B. K. Johnson, S. M. Mccorquodale, L. A. Shipley, R. A. Riggs, L. L. Irwin, S. L. Murphie, and B. L. Murphie. 2013. Regional and seasonal patterns of nutritional condition and reproduction in elk. Wildlife Monographs 184:1-45. Ditmer, M. A., G. Wittemyer, S. W. Breck, and K. R. Crooks. 2022. Defining ecological and socially suitable habitat for the reintroduction of an apex predator. Global Ecology and Conservation 38:e02192. Laporte, I., T. B. Muhly, J. A. Pitt, M. Alexander, and M. Musiani. 2010. Effects of wolves on elk and cattle behaviors: implications for livestock production and wolf conservation. PLoS One 5:e11954. Niemiec, R., R. E. Berl, M. Gonzalez, T. Teel, J. Salerno, S. Breck, C. Camara, M. Collins, C. Schultz, and D. Hoag. 2022. Rapid changes in public perception toward a conservation initiative. Conservation Science and Practice 4:e12632. Ramler, J. P., M. Hebblewhite, D. Kellenberg, and C. Sime. 2014. Crying wolf? A spatial analysis of wolf location and depredations on calf weight. American Journal of Agricultural Economics 96:631-656. 34 �Riaboff, L., S. Couvreur, A. Madouasse, M. Roig-Pons, S. Aubin, P. Massabie, A. Chauvin, N. Bédère, and G. Plantier. 2020. Use of predicted behavior from accelerometer data combined with GPS data to explore the relationship between dairy cow behavior and pasture characteristics. Sensors 20:4741. Riaboff, L., L. Shalloo, A. F. Smeaton, S. Couvreur, A. Madouasse, and M. T. Keane. 2022. Predicting livestock behaviour using accelerometers: A systematic review of processing techniques for ruminant behaviour prediction from raw accelerometer data. Computers and Electronics in Agriculture 192:106610. University of Arkansas Division of Agriculture Research & Extension. Economic impact of agriculture. <https://economic-impact-of-ag.uada.edu/colorado/>. Accessed 2023. 35 � https://cpw.cvlcollections.org/files/original/bf3b4b130d161a206874eb5c95d50f37.pdf f0574ddc95855e2000f2b6d80813a3d5 PDF Text Text Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Elk sightability for abundance estimation in Colorado Period Covered: January 01, 2024 – December 31, 2024 Principal Investigators: Rachel Smiley, rachel.smiley@state.co.us; Mat Alldredge, mat.alldredge@state.co.us; Chuck Anderson, chuck.anderson@state.co.us; Andy Holland, andy.holland@state.co.us; Jon Runge, jon.runge@state.co.us All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. Rocky Mountain Elk (Cervus canadensis) is an iconic species throughout western North America and especially in Colorado, with high recreational value to hunters, photographers, artists and wildlife enthusiasts in general. Elk populations can fluctuate greatly and are responsive to ecological conditions such as habitat succession, wildfires, climatic conditions, harvest regimes, and predation pressure (Wang et al. 2002, Brodie et al. 2013, Paterson et al. 2022). To maintain sustainable populations, managers must understand the various factors that influence population dynamics and use the best available information to set herd objectives, harvest strategies, and monitoring programs. Estimates of animal abundance are critical for understanding changes in population size and proximity to objectives. In Colorado, growing interest in elk population size and concern over the factors that may impact populations emphasizes the need for estimates of elk abundance. The myriad of factors that can influence population dynamics of elk highlight the need for estimates of abundance to understand the overall population status and when management intervention is warranted. Understanding wildlife population dynamics allows managers to estimate population size from one year to the next. If population size is reliably estimated for one year and the future number of losses and gains are monitored, then the population size and trend can be estimated and modeled into the future. Conducting yearly population counts via aerial surveys is expensive and not practical across the entire state on an annual basis. Reliable population models, however, require estimates of population size as a starting point and periodic re-estimation of population size can help anchor models over long time periods (Lukacs and Nowak 2023). Sightability surveys are a relatively time- and cost-effective approach with demonstrated utility for estimating abundance (Anderson et al. 1998, Wal et al. 2011, Fieberg 2012, Phillips et al. 2019). To parameterize sightability models, observers conduct trial flights to determine whether marked animals are seen or not seen in a variety of sighting conditions. Using data collected during the trial flights we will model how factors influence sightability, such as group size, vertical cover, and snow cover. Once the model is parameterized, biologists can conduct sightability surveys, in which they count and classify groups of elk and record the variables that influence sightability. When applied to sightability surveys, the sightability model uses detection probabilities for groups of animals seen and missed during development to correct for sighting conditions influencing detection and provides corrected abundance estimates. Development of an elk sightability model will give the state a tool to obtain periodic elk estimates at the DAU level. For this study, we are using females collared in the Elk Monitoring areas. Additionally, we are including males in the study to account for differences in habitat or group characteristics between sexes. In winter 2024-2025 we collared 15 male elk in the South Park study area (DAU E-18, E- 22, and E-23) and 15 male elk in the Gunnison study area (DAU E-5, E-25, and E-43). In February 2025, we completed 37 sightability trials on collared elk between the two study areas (Figures 1, 2). The trials 27 �included groups that ranged from 1 to 150 elk in a variety of vegetation and snow cover classes, which will allow for evaluation of how these variables influence the detection of elk. We will deploy 30 more collars on male elk in winter 2025-2026 in the Piney River (DAU E-12) and Middle Park (DAU E-13, E8) and continue trials to get a large enough sample size for a robust sightability model. Literature Cited Anderson, C. R., D. S. Moody, B. L. Smith, F. G. Lindzey, and R. P. Lanka. 1998. Development and evaluation of sightability models for summer elk surveys. The Journal of Wildlife Management 62:1055. Brodie, J., H. Johnson, M. Mitchell, P. Zager, K. Proffitt, M. Hebblewhite, M. Kauffman, B. Johnson, J. Bissonette, C. Bishop, J. Gude, J. Herbert, K. Hersey, M. Hurley, P. M. Lukacs, S. McCorquodale, E. McIntire, J. Nowak, H. Sawyer, D. Smith, and P. J. White. 2013. Relative influence of human harvest, carnivores, and weather on adult female elk survival across western North America. Journal of Applied Ecology 50:295–305. Fieberg, J. R. 2012. Estimating population abundance using sightability models: R SightabilityModel package. Journal of Statistical Software 51:1-20. Lukacs, P. M., and J. J. Nowak. 2023. Modeling population dynamics of black-tailed and mule deer. Pages 95–102 in. Ecology and management of black-tailed and mule deer of North America. First edition. CRC Press, Boca Raton, Florida, USA. Paterson, J. T., K. M. Proffitt, and J. J. Rotella. 2022. Incorporating vital rates and harvest into stochastic population models to forecast elk population dynamics. The Journal of Wildlife Management 86:e22189. Phillips, E. C., C. P. Lehman, R. W. Klaver, A. R. Jarding, S. P. Rupp, J. A. Jenks, and C. N. Jacques. 2019. Evaluation of an elk detection probability model in the Black Hills, South Dakota. Western North American Naturalist 79:551. Wal, E. V., P. D. McLoughlin, and R. K. Brook. 2011. Spatial and temporal factors influencing sightability of elk. The Journal of Wildlife Management 75:1521–1526. Wang, G., N. Thompson Hobbs, F. J. Singer, D. S. Ojima, and B. C. Lubow. 2002. Impacts of climate changes on elk population dynamics in Rocky Mountain National Park, Colorado, U.S.A. Climatic Change 54:205–223. 28 �Figure 1. Number of trials where target elk were seen and not seen in Gunnison and South Park study areas in 2025. Figure 2. Percent of trials where male and female elk were seen and not seen in Gunnison and South Park study areas in 2025. 29 � https://cpw.cvlcollections.org/files/original/a71464e50b9c86698d96ef32fdb45e6d.pdf 21da1afeb34af35644c6f14b5d6432e0 PDF Text Text ÿ ÿ ÿ ÿ 34546784ÿ76 ÿ78ÿ��585���ÿ ��������ÿ��������ÿ��� ��!ÿ�"##��$ÿ ÿ �%&'()%*)ÿ+&ÿ&+,)-.ÿ/0%01)/)%.ÿ+%ÿ-%+2-3+)ÿ30,)ÿ4)%-5.6ÿ5%ÿ'+41)7+')ÿ0%4ÿ-7,(*)8&5,ÿ -6-.)/-ÿ5%ÿ�+'+,04+ÿ ÿ 9:;<=>ÿ@=A:;:>BÿÿÿCDEFÿGHÿIJGKLCDM:ÿNJHÿIJGOÿ ÿ 9;<MP<QREÿSMA:TU<VRU=;TBÿÿÿCRW:ÿSARMHÿCRW:XSARMYTURU:XP=XDTZÿ[;<Pÿ\:]W<;WHÿ[;<PX\:]W<;WYTURU:XP=XDTÿ ÿ ^55ÿ��46_7`�4ÿ�ÿ̀a�ÿb64c�d`ÿe__76fÿ�ÿb6�5�_�76fÿ78ÿegc�d`ÿ̀4ÿ�e6`a�6ÿ�h75e7`�4iÿÿ j�46_7`�4ÿk^lÿmnoÿpqÿrpsjtuqvÿnwÿxrnoqvÿy�`a4e`ÿb�6_��4ÿ4�ÿ̀a�ÿb6�d�b75ÿ �h�`�z7`46iÿÿk7�be57`�4ÿ4�ÿ̀a��ÿ87`7ÿg�f48ÿ̀a7`ÿd4`7��8ÿ�ÿ̀a�ÿe__76fÿ�ÿ8�d4e67z�8iÿÿÿ ÿ {|}~€‚|}ƒ|„ÿ‚|}ÿ†|ƒ†ƒ|„ÿ€|†‡€ˆ†~ÿˆ‚~ÿ‰Š‹ŒŽÿ‘‹’“”•Ž–ÿ}~|€ƒ—ÿƒ|ÿ˜†™†‚}†ÿƒ€ÿ ƒš†‚|ÿ›~œ‚€~ÿˆ‚~€ÿœ† šƒ€~ÿžŸ ÿ†¡ÿˆ~ÿ}ƒ~ÿ†¡ÿˆ~ÿ€‚~¢~|}‚|„~~}£ÿ¡~}~‚™™—ÿˆ~‚~|~}ÿ˜‚|‚}‚ÿ ™—|¤ÿ‰Š¥•¦ÿ”•§‹•Ž“Žÿ̈{©ª©ÿ«ƒ€ˆÿ‚|}ÿ¬ƒ™}™ƒ¡~ÿª~­ƒœ~ÿ2ŸŸŸ£ÿ®­‚|ÿ‚|}ÿªˆ~|¯ÿ2Ÿ1°–©ÿÿ«†~€ÿ‚|‚„~ ~|ÿ ƒ€ÿ‚|ÿƒš†‚|ÿ}ƒ­~ÿ†¡ÿ€|†‡€ˆ†~ÿˆ‚~ÿ}~|€ƒ—£ÿ‚|}ÿ‚™™ÿ±‚ƒ†|‚™ÿ«†~€€ÿƒ|ÿ˜†™†‚}†ÿ‚~ÿ~²ƒ~}ÿ†ÿ ƒ|œ™}~ÿ‚|‚„~ ~|ÿ}ƒ~œƒ†|ÿ‚ƒ~}ÿ‚ÿœ†|€~­‚ƒ†|ÿ†¡ÿ˜‚|‚}‚ÿ™—|¤ÿ‚|}ÿ€|†‡€ˆ†~ÿˆ‚~ÿ‚€ÿš~ÿˆ~ÿ ª†ˆ~|ÿ³†œ¯ƒ~€ÿ́—|¤ÿµ ~|} ~|ÿ‰ª³´µÿ̈ˆš€¶··‡‡‡©¡€©€}‚©„†­·}~‚ƒ™·2·™‚|} ‚|‚„~ ~|·ÿ š™‚||ƒ|„·¸œƒ}¹ÿ€~™š}›º»º°¼°º–©ÿÿµÿˆ~ÿ€‚ ~ÿƒ~£ÿ«†~€€ÿƒ|ÿˆ~ÿ³~„ƒ†|ÿ‚~ÿœ† š~™™~}ÿ†ÿ~~ÿƒ›~ÿ š†}œƒ†|ÿ‚|}ÿ‚|‚„~ ~|ÿ~€š†|€~ÿ†›™ƒ„‚ƒ†|€©ÿÿªœˆÿ‚œƒ­ƒƒ~€ÿ‚—ÿ}~š~€€ÿ€|†‡€ˆ†~ÿˆ‚~ÿ}~|€ƒ—£ÿ ƒš†­~ÿƒ£ÿ†ÿˆ‚­~ÿƒ¤~}ÿ~¡¡~œ€ÿ}~š~|}~|ÿ†|ÿƒ~ÿ~™‚š€~}ÿ€ƒ|œ~ÿˆ~ÿ‚œƒ­ƒ—©ÿÿ½~~ÿ‡~ÿ}~€œƒ›~ÿ‚ÿ €‚ š™ƒ|„ÿ€œˆ~ ~ÿ†ÿ‚€€~€€ÿƒš‚œ€ÿ†¡ÿœ† †|ÿ¡†~€ÿ‚|‚„~ ~|ÿ~œˆ|ƒ²~€ÿ†|ÿ€|†‡€ˆ†~ÿˆ‚~ÿ}~|€ƒ—ÿ ƒ|ÿ›†ˆÿ™†}„~š†™~ÿšƒ|~ÿ‚|}ÿ€šœ~¢¡ƒÿ€—€~ €ÿ€‚~‡ƒ}~©ÿÿÿÿÿ ¾†ÿ€~™~œÿ¡†~€ÿ€‚|}€ÿ¡†ÿ€‚ š™ƒ|„£ÿ‡~ÿ¡ƒ€ÿ€~}ÿ{©ÿª©ÿ«†~€ÿª~­ƒœ~ÿ‰{ª«ª–ÿ€š‚ƒ‚™ÿ}‚‚ÿ†ÿ }~™ƒ|~‚~ÿ‚™™ÿ€šœ~¢¡ƒÿ‚|}ÿ™†}„~š†™~ÿšƒ|~ÿ€‚|}€ÿ‰€‚ ÿ1–ÿ†|ÿ{ª«ªÿ™‚|}ÿƒ|ÿ˜†™†‚}†£ÿ‚|}ÿƒ}~|ƒ¡ƒ~}ÿ ‚™™ÿ†¡ÿˆ~ÿ‚|‚„~ ~|ÿ‚œƒ­ƒƒ~€ÿˆ‚ÿˆ‚­~ÿ†œœ~}ÿƒ|ÿ~‚œˆÿ€‚|}ÿ†­~ÿƒ~©ÿÿ¬ƒˆÿœ†|€™‚ƒ†|ÿ¡† ÿˆ~ÿ {ª«ªÿ³~„ƒ†|ÿ2ÿ́—|¤¢ªƒ™­ƒœ™~ÿ¾~‚ £ÿ‡~ÿˆ~|ÿ„†š~}ÿ~™~­‚|ÿ¡†~€ÿ‚|‚„~ ~|ÿ‚œƒ­ƒƒ~€ÿ ‰€‚ ÿ2–ÿƒ|†ÿ¿ÿ›†‚}ÿœ‚~„†ƒ~€¶ÿ~­~|¢‚„~}ÿ‚|‚„~ ~|£ÿ|~­~|¢‚„~}ÿ‚|‚„~ ~|£ÿˆƒ||ƒ|„£ÿ‚|}ÿ | ‚|‚„~}ÿ‰œ†|†™€–©ÿÿÿ¬~ÿˆ~|ÿ}†œ ~|~}ÿˆ~ÿ2¢}~œ‚}~ÿ‡ƒ|}†‡ÿ‰€‚ ÿ»–ÿ}ƒ|„ÿ‡ˆƒœˆÿˆ~ÿ†€ÿ ~œ~|ÿ‚œƒ†|ÿ‡‚€ÿœ† š™~~}©ÿÿ«ƒ|‚™™—£ÿ‡~ÿ€~™~œ~}ÿ‚ÿ€š‚ƒ‚™™—ÿ›‚™‚|œ~}ÿ‚|}† ÿ€‚ š™~ÿ†¡ÿºÿ€‚|}€ÿ ‡ƒˆƒ|ÿ~‚œˆÿœ† ›ƒ|‚ƒ†|ÿ†¡ÿ¡†~€ÿ—š~ÿÀÿ‚|‚„~ ~|ÿ‚œƒ­ƒ—ÿÀÿœ† š™~ƒ†|ÿ}‚~ÿœ† ›ƒ|‚ƒ†|ÿ†ÿ~|€~ÿ ‡~ÿ€‚ š™~}ÿ‚ÿœ† š™~~ÿ„‚}ƒ~|ÿ†¡ÿƒ~ÿ€ƒ|œ~ÿƒš™~ ~|‚ƒ†|ÿ¡†ÿ~‚œˆÿ‚|‚„~ ~|ÿ‚œƒ­ƒ—ÿ†¡ÿƒ|~~€ÿ ƒ|ÿ~‚œˆÿ¡†~€ÿ—š~ÿ†¡ÿƒ|~~€ÿ‰«ƒ„~ÿ1£ÿ2–©ÿÿ¾ˆ~~ÿƒ€ÿ|†ÿÁP=ÂQE:U<=Mÿ>RU:Ãÿ¡†ÿ| ‚|‚„~}ÿœ†|†™€£ÿ€†ÿ ‡~ÿ€ƒš™—ÿ€‚ š™~}ÿ1Ÿÿ‚|}† ™—ÿ€~™~œ~}ÿ€‚|}€ÿ¡† ÿˆƒ€ÿœ† ›ƒ|‚ƒ†|©ÿÿµ™€†£ÿ|~­~|¢‚„~}ÿ™†}„~š†™~ÿ šƒ|~ÿ~‚~|€ÿ‚~ÿ‚~£ÿ€†ÿ‡~ÿ}ƒ}ÿ|†ÿ€‚ š™~ÿˆ‚ÿœ† ›ƒ|‚ƒ†|£ÿ™~‚­ƒ|„ÿ‚ÿ†‚™ÿ†¡ÿ•ÿ¹ÿ1Ÿºÿ€‚|}€ÿ €‚ š™~}©ÿÿ ¬~ÿ€‚ š™~}ÿ•ÿ¹ÿºŸÿ1¢ 2ÿœƒœ™‚ÿš™†€ÿ‡ƒˆƒ|ÿ~‚œˆÿ†¡ÿˆ~ÿ•ÿ¹ÿ1Ÿºÿ€‚|}€ÿ€~™~œ~}ÿ¡†ÿ€‚ š™ƒ|„©ÿÿ ꆁ€ÿ‡~~ÿ‚™€†ÿ€~™~œ~}ÿƒ|ÿ‚ÿ€š‚ƒ‚™™—ÿ›‚™‚|œ~}£ÿ‚|}† ÿ¡‚€ˆƒ†|©ÿÿ®|ÿ2Ÿ1¼£ÿ~œˆ|ƒœƒ‚|€ÿœ™~‚~}ÿ‚|}ÿœ†|~}ÿ €|†‡€ˆ†~ÿˆ‚~ÿš~™™~€ÿƒ|ÿ~‚œˆÿš™†©ÿÿÄ~™™~ÿƒ|¡† ‚ƒ†|ÿ¡† ÿ|œ™~‚~}ÿš™†€ÿƒ€ÿ™~€€ÿ‚œœ‚~ÿˆ‚|ÿˆ‚ÿ¡† ÿ š~­ƒ†€™—ÿœ™~‚~}ÿš™†€ÿ›~œ‚€~ÿ|œ™~‚~}ÿš™†€ÿƒ|­‚ƒ‚›™—ÿƒ|œ™}~ÿš~™™~ÿ‚œœ ™‚ƒ†|ÿ‚œ†€€ÿ€~­~‚™ÿ —~‚€ÿ‰½†}„~€ÿ‚|}ÿљ™€ÿ2ŸŸ¼–©ÿÿ¾ˆ~ÿ}~„~~ÿ†ÿ‡ˆƒœˆÿš~­ƒ†€ÿ—~‚€ÿ‚~ÿ~š~€~|~}ÿœ‚|ÿ}~š~|}ÿ†|ÿ™†œ‚™ÿ ‡~‚ˆ~ÿœ†|}ƒƒ†|€£ÿ€ƒ~ÿœ†|}ƒƒ†|€ÿ‚ÿˆ~ÿš™†£ÿ‚|}ÿ­‚ƒ‚›ƒ™ƒ—ÿƒ|ÿ‚œ‚™ÿ€|†‡€ˆ†~ÿˆ‚~ÿ}~|€ƒ—ÿ†­~ÿ š~­ƒ†€ÿ‡ƒ|~€©ÿÿ±~­~ˆ~™~€€£ÿ}‚‚ÿ¡† ÿ|œ™~‚~}ÿš™†€ÿœ‚|ÿ›~ÿ‚ÿ~‚€†|‚›™~ÿƒ|}ƒœ‚†ÿ†¡ÿ~™‚ƒ­~ÿ €|†‡€ˆ†~ÿˆ‚~ÿ}~|€ƒ—ÿ‰½†}„~€ÿ‚|}ÿљ™€ÿ2ŸŸ¼–©ÿÿ¾ˆ~~¡†~£ÿ‡~ÿœ† š~}ÿˆ~ÿ~‚|ÿ| ›~ÿ†¡ÿ š~™™~€·š™†ÿ¡†ÿ~‚œˆÿ¡†~€ÿ—š~ÿÀÿ‚|‚„~ ~|ÿ‚œƒ­ƒ—ÿÀÿœ† š™~ƒ†|ÿ}‚~ÿœ† ›ƒ|‚ƒ†|ÿ†ÿ‚¯~ÿ‚|ÿƒ|ƒƒ‚™ÿ ‚€€~€€ ~|ÿ†¡ÿˆ~ÿƒš‚œ€ÿ†¡ÿ¡†~€ÿ‚|‚„~ ~|ÿ†|ÿ€|†‡€ˆ†~ÿˆ‚~ÿ}~|€ƒ—ÿ‰«ƒ„~ÿ2–©ÿÿ±†~ÿˆ‚ÿˆ~€~ÿ 12ÿ ÿ �ÿ ÿ ÿ ÿ 3453678ÿ9ÿ7 �3�7ÿ� 7ÿ57483��ÿ7ÿ69 87ÿ�ÿ�78��ÿ3453678ÿ�875ÿ94ÿ5�ÿ6976�75ÿ9�ÿ67 75ÿ�9�8ÿ34ÿ ��1�ÿ8�9�5ÿ�97ÿ66� �7�ÿ57�36�ÿ849�8�97ÿ� 7ÿ57483���ÿÿÿ �73�34 �ÿ78��8ÿ8���78�ÿ���ÿ��7ÿ�3��78�ÿ849�8�97ÿ� 7ÿ57483�378ÿ966�ÿ34ÿ�4� 4�75ÿ8��67� 3ÿ978�8�ÿÿ� 4�7�74�ÿ9ÿ8�6�ÿ8�458ÿ�747 �ÿ758ÿ�9ÿ9�7ÿ849�8�97ÿ� 7ÿ57483�378ÿ��9���ÿ 57483�378ÿ69�� �7ÿ�9ÿ�4� 4�75ÿ8�458ÿ64ÿ�7ÿ6�37�75ÿ��!"�ÿ�7 8ÿ�7ÿ�47�74��75ÿ�7��74�8ÿ9ÿ �!��ÿ�7 8ÿ�7ÿ��34434��ÿÿ#�74��75ÿ�7��74�8ÿ��7 ÿ�9ÿ57�788ÿ� 7ÿ57483�378ÿ�!$�ÿ�7 8ÿ�7ÿ �7��74��ÿÿ%�7ÿ9��983�7ÿ��7 8ÿ�9ÿ�7ÿ��7ÿ34ÿ95�7�97ÿ�347ÿ8�8�7�8�ÿÿ%��ÿ38ÿ��34434�ÿÿ�7�39�8�ÿ �4� 4�75ÿ8�45ÿ64ÿ75ÿ�9ÿ�3��7ÿ� 7ÿ57483�378ÿ�!$�ÿ�7 8ÿ�98���7��74�ÿ69�� 75ÿ�9ÿ�4� 4�75ÿ 8�458�ÿÿ&3�3 �ÿ� 7ÿ57483�378ÿ34ÿ7�74��75ÿ�7��74�8ÿ��7 ÿ�3��7ÿ��4ÿ34ÿ�4� 4�75ÿ8�458ÿ��!$�ÿ �7 8ÿ�98���7��74�ÿ'(3��7ÿ�)�ÿÿÿ ÿ *+,-./,0.-ÿ2+,-34ÿ ÿ 56789:;ÿ=>ÿ?>;ÿ@A7ÿB>ÿC>ÿDEFF:>ÿGHHI>ÿJ9:E8AEA8ÿK9L@FÿM9FF9Nÿ:OPQ9R:ÿK6Pÿ:A6S:T69ÿT@P9:>ÿU6P9:Nÿ ?L6F68Rÿ@A7ÿD@A@89V9ANÿGWXYZ[ZI!Z[GX>ÿ \Q@A;ÿ]>ÿC>;ÿ@A7ÿ^>ÿD>ÿCT9A_>ÿGHZX>ÿ̀EAN9Pÿ7E9Nÿ@A7ÿTOANEA8ÿ:OLL9::ÿ6Kÿa@A@7@ÿFRAbÿEAÿa6F6P@76>ÿ ^T9ÿ]6OPA@Fÿ6Kÿ̀EF7FEK9ÿD@A@89V9ANÿIHYZHc[!ZHWI>ÿ d>C>ÿUE:Tÿ@A7ÿ̀EF7FEK9ÿC9PQEL9>ÿGHHH>ÿ?A7@A89P97ÿ@A7ÿNTP9@N9A97ÿSEF7FEK9ÿ@A7ÿMF@AN:Yÿ 79N9PVEA@NE6Aÿ6KÿNTP9@N9A97ÿ:N@NO:ÿK6PÿNT9ÿL6ANE8O6O:ÿd>ÿC>ÿ7E:NEALNÿM6MOF@NE6Aÿ:98V9ANÿ6Kÿ NT9ÿa@A@7@ÿFRAbÿ@A7ÿP9F@N97ÿPOF9;ÿKEA@FÿPOF9>ÿU979P@Fÿe98E:N9PÿXWYZXHWG!ZXHIX>ÿ ÿ ÿ -----1 * Treatment ÿ ÿ UE8OP9ÿZ>ÿÿB6L@NE6Aÿ6Kÿ@GFFÿ:N@A7:ÿfgÿhÿZHWiÿ:9F9LN97ÿK6PÿM9FF9NÿMF6Nÿ:@VMFEA8;ÿ]OA9!C9MN9Vj9PÿGHZI>ÿÿkNÿ 9@LTÿ:N@A7;ÿgÿhÿWHÿZlV ÿM9PV@A9ANÿM9FF9NÿMF6N:ÿS9P9ÿEA:N@FF97ÿ@NÿP@A76VÿF6L@NE6A:;ÿ@A7ÿ@FFÿ:A6S:T69ÿT@P9ÿ M9FF9N:ÿSENTEAÿNT9:9ÿS9P9ÿL6OAN97ÿ@A7ÿLF9@P97>ÿÿ 12ÿ ÿ �ÿ ÿ ÿ Even-aged Unmanaged 16 16 ·c. 12 14 a.> a.> C +-' 0 Q_ en +-' 0 10 a. ÿ Uneven-aged 16 14 14 14 12 12 12 10 10 10 8 8 8 a.> CJ) "O 0 .....I 0 (1) c1959 (1) Q_ C ro (1) ~ Thinned 16 L. i;::: Q) (.) :::, L. a. 1959- 1978 1979-1998 1959-1978 1999-2018 1979-1998 1999· 2018 16 16 16 16 14 14 14 14 12 12 12 12 10 10 10 10 8 8 8 1959-1978 1979-1998 1999·2018 1959-1978 1979-1998 1999-2018 6 (f) <1959 1959-1978 1979-1998 1999-2018 1959-1978 1979-1998 Treatment Year 1999-2018 345678ÿÿÿ 8�ÿ����ÿ����ÿ�8��8�������ÿ����8�ÿ�ÿ8��ÿ!"#$%ÿ%&'#ÿ(ÿ)*+*,#)#+%ÿ*-%./.%&ÿ(ÿ-!)'0#%.!+ÿ ÿ 1*%#ÿ-!)2.+*%.!+ÿÿ3���8�ÿ�4�8�ÿ4��4��8ÿ��8ÿ�8�ÿ�8��8�������ÿ4�7ÿ��8ÿ6�� �58�ÿ4�78��ÿ�5�8ÿ4�7ÿ ����74���ÿ��ÿ� �58�8��ÿ��4���ÿÿ6��58���8ÿ�4�8ÿ�����ÿ78ÿ778�5ÿ� �58�ÿ4�ÿ6�878�ÿ4��4��8ÿ ��8784�78ÿ��ÿ�6��ÿ�����ÿ9878ÿ����8�ÿÿ:��8ÿ��8ÿ8;�78�8ÿ�6��487ÿ�8�4��8�ÿ<5ÿÿ�4�5�8ÿ���ÿ4�ÿ��8ÿ���58���8ÿ �4�88ÿ878�=58�ÿ��8�ÿÿ>�4�ÿ4�ÿ��8ÿ��4��ÿ8��4� �8ÿ�87478�ÿ47��ÿ��ÿ�4ÿ��8ÿ��ÿ4�ÿ���ÿ�����8�4��ÿ��8ÿ<4�8ÿ 4���6�4�5ÿ���ÿ47��ÿÿ�4�5�8ÿ����ÿ4�ÿ9�4��ÿ��8ÿ78758ÿ�8��8�ÿ��6��ÿ9 �ÿ�ÿ�7�87ÿ�4ÿ� 5�4�6�8ÿ�7587ÿ���ÿ �59�878ÿ8��8ÿ98ÿ����8�ÿÿ>�8�8ÿ��ÿ�88�ÿ�6��8��ÿ�ÿ�78�ÿ78ÿ���ÿ���65��ÿ��ÿ78��ÿ�45�ÿ�8��4�48�ÿ4�ÿ 7875ÿ���ÿ��87�6��ÿÿ>�8ÿ��ÿ��4��ÿ94��ÿ����4�8�ÿ����7�ÿ877�7ÿ4�ÿ��8ÿ8��4� �8ÿ44ÿ��4�ÿ���8��4�ÿ�6��487ÿ4�ÿ ��4��8�ÿ ÿ ÿ ÿ 12ÿ ÿ �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Influence of forest management on snowshoe hare density in lodgepole and spruce-fir systems in Colorado Period Covered: July 1, 2018  June 30, 2019 Principal Investigators: Jake Ivan, Jake.Ivan@state.co.us; Eric Newkirk, Eric.Newkirk@state.co.us All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. Understanding and monitoring snowshoe hare (Lepus americanus) density in Colorado is important because hares comprise 70% of the diet of the state-endangered, federally threatened Canada lynx (Lynx canadensis; U.S. Fish and Wildlife Service 2000, Ivan and Shenk 2016). Forest management is an important driver of snowshoe hare density, and all National Forests in Colorado are required to include management direction aimed at conservation of Canada lynx and snowshoe hare as per the Southern Rockies Lynx Amendment (SRLA; https://www.fs.usda.gov/detail/r2/landmanagement/ planning/?cid= stelprdb5356865). At the same time, Forests in the Region are compelled to meet timber production and management response obligations. Such activities may depress snowshoe hare density, improve it, or have mixed effects dependent on the specific activity and the time elapsed since that activity was initiated. Here we describe a sampling scheme to assess impacts of common forest management techniques on snowshoe hare density in both lodgepole pine and spruce-fir systems in Colorado. To select forest stands for sampling, we first used U. S. Forest Service (USFS) spatial data to delineate all spruce-fir and lodgepole pine stands (stratum 1) on USFS land in Colorado, and identified all of the management activities that have occurred in each stand over time. With consultation from the USFS Region 2 Lynx-Silviculture Team, we then grouped relevant forest management activities (stratum 2) into 4 broad categories: even-aged management, uneven-aged management, thinning, and unmanaged controls. We wanted to assess both the immediate and long-term impacts of management on hare densities. Therefore, when selecting stands for sampling, we took the additional step of binning the date of the most recent management activity into 2-decade intervals (i.e., 0-20, 20-40, and 40-60 years before 2018). We then selected a spatially balanced random sample of 5 stands within each combination of forest type × management activity × time interval. This design ensured that we sampled the complete gradient of time since implementation for each management activity of interest in each forest type of interest. There is no notion of “completion date” for unmanaged controls, so we simply sampled 10 randomly selected stands from this combination. Also, uneven-aged lodgepole pine treatments are rare, so we did not sample that combination, leaving a total of n = 105 stands sampled (Figure 1). During summer 2018, we established n = 50 1-m2 permanent circular plots within each of the n = 105 stands selected for sampling. Plot locations within each stand were selected in a spatially balanced, random fashion. Technicians cleared and counted snowshoe hare pellets in each plot as they were established. These same plots were re-visited and re-counted during summer 2019. In addition to sampling the previously cleared plots from 2018, technicians were able to install plots at 2 more replicate sites for each combination of forest type × management activity × time interval, meaning that inference 2 �from future years will be based on 7 stands within each combination, or n = 128 total stands (note that this total also reflects a handful of stands that were re-classified based field observations, along with new stands that were brought into the sample in 2019 to replace those that were reclassed). Pellet information from cleared plots is more accurate than that from uncleared plots because uncleared plots usually include pellet accumulation across several years (Hodges and Mills 2008). The degree to which previous years are represented can depend on local weather conditions, site conditions at the plot, and variability in actual snowshoe hare density over previous winters. Data from cleared plots necessarily reflects hare activity from the previous 12 months, and tracks true density more closely. Therefore, we focused the current analysis on the 2019 data from previously cleared plots. For each forest type × management activity combination, we plotted mean pellet counts against “year since activity,” then fit a curve (e.g., quadratic function) through the data (Figure 2). Results from this preliminary analysis suggest that on average the highest snowshoe hare densities typically occur in unmanaged spruce-fir forests, and that unmanaged spruce-fir forests are estimated to have twice the relative hare density of unmanaged lodgepole pine forests. For both forest types, the fitted line suggests that even-aged management (e.g., clearcutting), immediately depresses relative hare density to near zero, but density rebounds and peaks 20-40 years after management before declining again 40-60 years after. Estimated peak hare densities after even-aged management in lodgepole systems tend to be higher than the control condition, but in spruce-fir systems estimated peak densities approach, but never match, the control condition. In both forest types, thinning (which often occurs 20-40 years after stands undergo even-aged management, especially in lodgepole), immediately depresses hare densities, but densities are estimated to slowly recover through time in nearly linear fashion, reaching their maximum 45-55 years after the treatment. As with the even-aged treatment, maximum hare density after thinning in lodgpole systems is estimated to be higher than the control condition, whereas in spruce-fir systems, the maximum hare density matches that of the control sites. Uneven-aged management of spruce-fir forests results in a similar snowshoe hare trajectory as that observed in thinned spruce-fir forests. Note the two outliers on the right side of the even-aged lodgepole panel. These “high density” sites are represent even-aged lodgepole stands that happen to be surrounded by high quality spruce-fir forest on at least two sides. Thus, the high relative hare density observed at these sites may be due to the quality habitat in adjacent stands rather than by the quality of the sampled stands themselves. While we left them on the figure for transparency, we excluded them when fitting the curve as they appear to be true outliers. Also note that in some cases, 95% CIs are relatively large and overlap the control reference line in some panels. Thus, even though the fitted lines indicate the relationships discussed above, evidence for some of these patterns is moderate or weak. In future years, each panel will include cleared plot data from 6 additional sites, and each site will have data from multiple years (i.e., repeated measures). Both phenomena will greatly improve sample sizes, diminish the role of a few outlying data points, and tighten up our estimate, and corresponding inference, regarding the response of snowshoe hare density to forest management through time. Literature Cited: Hodges, K. E., and L. S. Mills. 2008. Designing fecal pellet surveys for snowshoe hares. Forest Ecology and Management 256:1918-1926. Ivan, J. S., and T. M. Shenk. 2016. Winter diet and hunting success of Canada lynx in Colorado. The Journal of Wildlife Management 80:1049-1058. U.S. Fish and Wildlife Service. 2000. Endangered and threatened wildlife and plants: determination of threatened status for the contiguous U. S. distinct population segment of the Canada lynx and related rule, final rule. Federal Register 65:16052–16086. 3 �Figure 1. Location of all stands (n = 105) resampled for snowshoe hare pellets, June-September 2019. Unmanaged Q) 0 0.. Cl) .._. Uneven-aged Thinned 12 12 12 12 ·5. 10 10 10 10 Q) 8 C .._. Even-aged 0 Cl. Q) C) "O 0 ....J Q) 6 6 4 0 0 , 60 Q) 0 0 10 20 30 40 50 60 0 0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60 0.. C ro Q) ~ .... .;: 12 12 12 12 10 10 10 10 I Q) (.) ....::JCl. 8 6 Cl) 0 0 , 60 0 0 10 20 30 40 50 60 0 10 20 30 40 50 60 Years Since Treatment Figure 2. Fitted quadratic function (white line) and 95% CI (shaded polygon) relating pellet counts (i.e., relative snowshoe hare density) to time elapsed since treatment for each forest type × management activity combination. Dotted lines indicate the mean pellets/plot for the unmanaged controls for each forest type. 4 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Influence of forest management on snowshoe hare density in lodgepole and spruce-fir systems in Colorado Period Covered: July 1, 2019 − June 30, 2020 Principal Investigators: Jake Ivan, Jake.Ivan@state.co.us; Eric Newkirk, Eric.Newkirk@state.co.us All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. Understanding and monitoring snowshoe hare (Lepus americanus) density in Colorado is important because hares comprise 70% of the diet of the state-endangered, federally threatened Canada lynx (Lynx canadensis; U.S. Fish and Wildlife Service 2000, Ivan and Shenk 2016). Forest management is an important driver of snowshoe hare density, and all National Forests in Colorado are required to include management direction aimed at conservation of Canada lynx and snowshoe hare as per the Southern Rockies Lynx Amendment (SRLA; https://www.fs.usda.gov/detail/r2/landmanagement/ planning/?cid= stelprdb5356865). At the same time, Forests in the Region are compelled to meet timber production obligations. Such activities may depress snowshoe hare density, improve it, or have mixed effects dependent on the specific activity and the time elapsed since that activity was initiated. Here we describe a sampling scheme to assess impacts of common forest management techniques on snowshoe hare density in both lodgepole pine (Pinus contorta) and spruce-fir (Picea engelmannii – Abies lasiocarpa) systems in Colorado. To select forest stands for sampling, we first used U. S. Forest Service (USFS) spatial data to delineate all spruce-fir and lodgepole pine stands (stratum 1) on USFS land in Colorado, and identified all of the management activities that have occurred in each stand over time. With consultation from the USFS Region 2 Lynx-Silviculture Team, we then grouped relevant forest management activities (stratum 2) into 4 broad catetories: even-aged management, uneven-aged management, thinning, and unmanaged controls. We wanted to assess both the immediate and long-term impacts of management on hare densities. Therefore, when selecting stands for sampling, we took the additional step of binning the date of the most recent management activity into 2-decade intervals (i.e., 0-20, 20-40, and 40-60 years before 2018). We then selected a spatially balanced random sample of 5 stands within each combination of forest type × management activity × time interval. This design ensured that we sampled the complete gradient of time since implementation for each management activity of interest in each forest type of interest. There is no notion of “completion date” for unmanaged controls, so we simply sampled 10 randomly selected stands from this combination. Also, uneven-aged lodgepole pine treatments are rare, so we did not sample that combination (Figure 1). During summer 2018, we established n = 50 1-m2 permanent circular plots within each of the stands selected for sampling. Plot locations within each stand were selected in a spatially balanced, random fashion. Technicians cleared and counted snowshoe hare pellets in each plot as they established them. These same plots were re-visited and re-counted during summers 2019 and 2020. In addition to sampling the previously cleared plots from 2018, technicians were able to install plots at 2 more replicate sites for each combination of forest type × management activity × time interval during 2019. Additionally, a handful of stands visited in 2019 and 2020 were re-classified or tossed based on field 7 �observations and new stands were sampled in their place by pulling the next one from the spatially balanced list. Currently, then inference is based on n = 130 total stands. Pellet information from cleared plots is more accurate than that from uncleared plots because uncleared plots usually include pellet accumulation across several years (Hodges and Mills 2008). The degree to which previous years are represented can depend on local weather conditions, site conditions at the plot, and variability in actual snowshoe hare density over previous winters. Data from cleared plots necessarily reflects hare activity from the previous 12 months, and tracks true density more closely. Therefore, we focused the current analysis on the 2019 and 2020 data from previously cleared plots. For each forest type × management activity combination, we plotted mean pellet counts against “year since activity”, then fit a curve (e.g., quadratic function) through the data (Figure 2). Results from this preliminary analysis suggest that on average the highest snowshoe hare densities typically occur in unmanaged spruce-fir forests, and that unmanaged spruce-fir forests are estimated to have twice the relative hare density of unmanaged lodgepole pine forests (Figure 2). For both forest types, the fitted line suggests that even-aged management (e.g., clearcutting), immediately depresses relative hare density to near zero, but density rebounds and peaks 20-40 years after management before declining again 40-60 years after. Estimated peak hare densities after even-aged management in lodgepole systems tend to be higher than the control condition. However, in spruce-fir systems the estimated fitted line is flatter and peak densities fell well short of the control condition. In both forest types, thinning (which often occurs 20-40 years after stands undergo even-aged management, especially in lodgepole), immediately depresses hare densities. In spruce-fir stands, densities were estimated to slowly recover through time in nearly linear fashion. However, they follow a peaked response in lodgepole pine, similar to the response to even-aged management. Uneven-aged management of spruce-fir forests results in immediate depression of relative hare density, which then recovers back to pre-treatment levels approximately 30 years after the treatment. Note the outlier on the right side of the even-aged lodgepole panel. This “high density” site is an even-aged lodgepole stand that happens to be surrounded by high quality spruce-fir forest on at least two sides. Thus, the high relative hare density observed at this site may be due to the quality habitat in adjacent stands rather than by the quality of the sampled stand itself. While we left the point on the figure for transparency, we excluded it when fitting the curve as it appears to be a true outlier (including it “flattens” the curve somewhat such that it crosses the control line at about 55 years). Literature Cited: Hodges, K. E., and L. S. Mills. 2008. Designing fecal pellet surveys for snowshoe hares. Forest Ecology and Management 256:1918-1926. Ivan, J. S., and T. M. Shenk. 2016. Winter diet and hunting success of Canada lynx in Colorado. The Journal of Wildlife Management 80:1049-1058. U.S. Fish and Wildlife Service. 2000. Endangered and threatened wildlife and plants: determination of threatened status for the contiguous U. S. distinct population segment of the Canada lynx and related rule, final rule. Federal Register 65:16052–16086. 8 �* Control Treatment Figure 1. Location of all stands (n = 130) resampled for snowshoe hare pellets, June-September 2020. Unmanaged Even-aged Uneven-aged Thinned (1) C ·a. (1) +-' 0 - a.. Cl) +-' 0 4 0.. (1) Cl "C 2 0 _J 0 Q) 0 >b0 Q) 0 10 20 30 40 50 bO 0 10 20 30 40 50 bO 0 10 20 30 40 50 b0 0 10 20 30 40 50 b0 a.. C .... co Q) ..:::I ~ :::, ....0.. (1) (.) 4 4 (/) 0 0 >b0 0 10 20 30 40 50 0 b0 10 20 30 40 50 b0 Years Since Treatment Figure 2. Fitted quadratic function (white line) and 95% CI (shaded polygon) relating pellet counts (i.e., relative snowshoe hare density) to time elapsed since treatment for each forest type × management activity combination. Dotted lines indicate the mean pellets/plot for the unmanaged controls for each forest type. 9 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Influence of forest management on snowshoe hare density in lodgepole and spruce-fir systems in Colorado Period Covered: July 1, 2019 − June 30, 2020 Principal Investigators: Jake Ivan, Jake.Ivan@state.co.us; Eric Newkirk, Eric.Newkirk@state.co.us All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. Understanding and monitoring snowshoe hare (Lepus americanus) density in Colorado is important because hares comprise 70% of the diet of the state-endangered, federally threatened Canada lynx (Lynx canadensis; U.S. Fish and Wildlife Service 2000, Ivan and Shenk 2016). Forest management is an important driver of snowshoe hare density, and all National Forests in Colorado are required to include management direction aimed at conservation of Canada lynx and snowshoe hare as per the Southern Rockies Lynx Amendment (SRLA; https://www.fs.usda.gov/detail/r2/landmanagement/ planning/?cid= stelprdb5356865). At the same time, Forests in the Region are compelled to meet timber production obligations. Such activities may depress snowshoe hare density, improve it, or have mixed effects dependent on the specific activity and the time elapsed since that activity was initiated. Here we describe a sampling scheme to assess impacts of common forest management techniques on snowshoe hare density in both lodgepole pine (Pinus contorta) and spruce-fir (Picea engelmannii – Abies lasiocarpa) systems in Colorado. To select forest stands for sampling, we first used U. S. Forest Service (USFS) spatial data to delineate all spruce-fir and lodgepole pine stands (stratum 1) on USFS land in Colorado, and identified all of the management activities that have occurred in each stand over time. With consultation from the USFS Region 2 Lynx-Silviculture Team, we then grouped relevant forest management activities (stratum 2) into 4 broad categories: even-aged management, uneven-aged management, thinning, and unmanaged controls. We wanted to assess both the immediate and long-term impacts of management on hare densities. Therefore, when selecting stands for sampling, we took the additional step of binning the date of the most recent management activity into 2-decade intervals (i.e., 0-20, 20-40, and 40-60 years before 2018). We then selected a spatially balanced random sample of 5 stands within each combination of forest type × management activity × time interval. This design ensured that we sampled the complete gradient of time since implementation for each management activity of interest in each forest type of interest. There is no notion of “completion date” for unmanaged controls, so we simply sampled 10 randomly selected stands from this combination. Also, uneven-aged lodgepole pine treatments are rare, so we did not sample that combination (Figure 1). During summer 2018, we established n = 50 1-m2 permanent circular plots within each of the stands selected for sampling. Plot locations within each stand were selected in a spatially balanced, random fashion. Technicians cleared and counted snowshoe hare pellets in each plot as they established them. These same plots were re-visited and re-counted during summers 2019 and 2020. In addition to sampling the previously cleared plots from 2018, technicians were able to install plots at 2 more replicate sites for each combination of forest type × management activity × time interval during 2019. Also, a handful of stands visited in 2019 and 2020 were re-classified or tossed because ground-truthing revealed 8 �they did not actually fit in the stratum for which they were selected. New stands were sampled in their place by pulling the next one from the spatially balanced list. Similarly, a handful more stands were replaced during the 2021 field season, and 12 new stands were selected to replace those that burned during the 2020 fire season. Currently, inference is based on n = 130 total stands. Finally, in 2021, we sampled vegetation metrics in each stand that will hopefully account for the considerable noise we have observed (highly variable results for some strata) and allow us to better assess the effects of the treatments themselves. This vegetation sampling will be completed during the 2022 field season. Pellet information from cleared plots is more accurate than that from uncleared plots because uncleared plots usually include pellet accumulation across several years (Hodges and Mills 2008). The degree to which previous years are represented can depend on local weather conditions, site conditions at the plot, and variability in actual snowshoe hare density over previous winters. Data from cleared plots necessarily reflects hare activity from the previous 12 months, and tracks true density more closely. Therefore, we focused the current analysis on the 2019-21 data from previously cleared plots. For each forest type × management activity combination, we plotted mean pellet counts against “year since activity”, then fit a curve (e.g., quadratic function) through the data (Figure 2). Results from this preliminary analysis suggest that on average the highest snowshoe hare densities typically occur in unmanaged spruce-fir forests, and that unmanaged spruce-fir forests are estimated to have twice the relative hare density of unmanaged lodgepole pine forests (Figure 2). For both forest types, the fitted line suggests that even-aged management (e.g., clearcutting), immediately depresses relative hare density to near zero, but density rebounds and peaks 20-40 years after management before declining again 40-60 years after. Estimated peak hare densities after even-aged management in lodgepole systems tend to be higher than the control condition. However, in spruce-fir systems the estimated fitted line is flatter and peak densities fell well short of the control condition. In both forest types, thinning (which often occurs 20-40 years after stands undergo even-aged management, especially in lodgepole), immediately depresses hare densities. In spruce-fir stands, densities were estimated to slowly recover through time in nearly linear fashion. However, they follow a peaked response in lodgepole pine, similar to the response to even-aged management. Uneven-aged management of spruce-fir forests results in immediate depression of relative hare density, which then recovers back to pre-treatment levels approximately 30 years after the treatment. Note the outlier on the right side of the even-aged lodgepole panel (Figure 2). This “high density” site is an even-aged lodgepole stand that happens to be surrounded by high quality spruce-fir forest on at least two sides. Thus, the high relative hare density observed at this site may be due to the quality habitat in adjacent stands rather than by the quality of the sampled stand itself. While we left the point on the figure for transparency, we excluded it when fitting the curve as it appears to be a true outlier (including it “flattens” the curve somewhat such that it crosses the control line at about 55 years). Literature Cited: Hodges, K. E., and L. S. Mills. 2008. Designing fecal pellet surveys for snowshoe hares. Forest Ecology and Management 256:1918-1926. Ivan, J. S., and T. M. Shenk. 2016. Winter diet and hunting success of Canada lynx in Colorado. The Journal of Wildlife Management 80:1049-1058. U.S. Fish and Wildlife Service. 2000. Endangered and threatened wildlife and plants: determination of threatened status for the contiguous U. S. distinct population segment of the Canada lynx and related rule, final rule. Federal Register 65:16052–16086. 9 �Figure 1. Location of all stands (n = 130) resampled for snowshoe hare pellets, June-September 2020. Unmanaged Q) 10 Uneven-aged Even-aged 10 C ·a. ....... 0 0.... en ....... 8 Q) 0 C. Q) 4 4 0 ...J 0 >60 0.... ~ 6 Cl Q) co Q) 6 "tJ Q) C Thinned 10 10 10 0 0 10 20 JO 40 50 60 0 10 10 .... .;:::: 8 8 Q) (.) 6 6 4 4 10 20 JO 40 50 60 0 10 20 JO 40 50 60 10 I ....::::J C. . Cf) - _,,,__ - .... ~ - - •• - _ "'J i:.. - - - -~~~ -a1.• ~ •t· ~~t -' ~t=:.}: 4 0 , 60 0 0 10 20 JO 40 50 60 0 10 20 JO 40 50 60 0 10 20 JO 40 50 Years Since Treatment Figure 2. Fitted quadratic function (white line) and 95% CI (shaded polygon) relating pellet counts (i.e., relative snowshoe hare density) to time elapsed since treatment for each forest type × management activity combination. Dotted lines indicate the mean pellets/plot for the unmanaged controls for each forest type. 10 60 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Influence of forest management on snowshoe hare density in lodgepole and spruce-fir systems in Colorado Period Covered: July 1, 2021 − June 30, 2022 Principal Investigators: Jake Ivan, Jake.Ivan@state.co.us; Eric Newkirk, Eric.Newkirk@state.co.us All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. Understanding and monitoring snowshoe hare (Lepus americanus) density in Colorado is imperative because hares comprise 70% of the diet of the state-endangered, federally threatened Canada lynx (Lynx canadensis; U.S. Fish and Wildlife Service 2000, Ivan and Shenk 2016). Forest management is an important driver of snowshoe hare density, and all National Forests in Colorado are required to include management direction aimed at conservation of Canada lynx and snowshoe hare as per the Southern Rockies Lynx Amendment (SRLA; https://www.fs.usda.gov/detail/r2/landmanagement/ planning/?cid= stelprdb5356865). At the same time, Forests in the Region are compelled to meet timber production obligations. Such activities may depress snowshoe hare density, improve it, or have mixed effects dependent on the specific activity and the time elapsed since that activity was initiated. Here we describe a sampling scheme to assess impacts of common forest management techniques on snowshoe hare density in both lodgepole pine (Pinus contorta) and spruce-fir (Picea engelmannii – Abies lasiocarpa) systems in Colorado. To select forest stands for sampling, we first used U. S. Forest Service (USFS) spatial data to delineate all spruce-fir and lodgepole pine stands (stratum 1) on USFS land in Colorado, and identified all of the management activities that have occurred in each stand over time. With consultation from the USFS Region 2 Lynx-Silviculture Team and USFS Rocky Mountain Research Station, we then grouped relevant forest management activities (stratum 2) into 4 broad categories: even-aged management, uneven-aged management, thinning, and unmanaged controls. We wanted to assess both the immediate and long-term impacts of management on hare densities. Therefore, when selecting stands for sampling, we took the additional step of binning the date of the most recent management activity into 2decade intervals (i.e., 0-20, 20-40, and 40-60 years before 2018). We then selected a spatially balanced random sample of 5 stands within each combination of forest type × management activity × time interval. This design ensured that we sampled the complete gradient of time since implementation for each management activity of interest in each forest type of interest. There is no notion of “completion date” for unmanaged controls, so we simply sampled 10 randomly selected stands from this combination. Also, uneven-aged lodgepole pine treatments are rare, so we did not sample that combination (Figure 1). During summer 2018, we established n = 50 1-m2 permanent circular plots within each of the stands selected for sampling. Plot locations within each stand were selected in a spatially balanced, random fashion. Technicians cleared and counted snowshoe hare pellets in each plot as they established them. These same plots were re-visited and re-counted during summers 2019 and 2020. In addition to sampling the previously cleared plots from 2018, technicians were able to install plots at 2 more replicate sites for each combination of forest type × management activity × time interval during 2019. Also, a 14 �handful of stands visited in 2019 and 2020 were re-classified or excluded because ground-truthing revealed they did not actually fit in the stratum for which they were selected. New stands were sampled in their place by pulling the next one from the spatially balanced list. Similarly, a handful more stands were replaced during the 2021 field season, and 12 new stands were selected to replace those that burned during the 2020 fire season. Currently, inference is based on n = 130 total stands. Finally, in 2021 and 2022, we sampled vegetation metrics in each stand that will hopefully account for the considerable noise we have observed (highly variable results for some strata) and allow us to better assess the effects of the treatments themselves. Pellet information from cleared plots is more accurate than that from uncleared plots because uncleared plots usually include pellet accumulation across several years (Hodges and Mills 2008). The degree to which previous years are represented can depend on local weather conditions, site conditions at the plot, and variability in actual snowshoe hare density over previous winters. Data from cleared plots necessarily reflects hare activity from the previous 12 months, and tracks true density more closely. Therefore, we focused the current analysis on the 2019–22 data from previously cleared plots. For each forest type × management activity combination, we plotted mean pellet counts against “year since activity,” then fit a curve (e.g., quadratic function) through the data (Figure 2). Results from this preliminary analysis suggest that on average the highest snowshoe hare densities typically occur in unmanaged spruce-fir forests, and that unmanaged spruce-fir forests are estimated to have twice the relative hare density of unmanaged lodgepole pine forests (Figure 2). For both forest types, the fitted line suggests that even-aged management (e.g., clearcutting), immediately depresses relative hare density to near zero, but density rebounds and peaks 20-40 years after management before declining again 40-60 years after. Estimated peak hare densities after even-aged management in lodgepole systems tend to be higher than the control condition. However, in spruce-fir systems the estimated fitted line is flatter and peak densities fell short of the control condition. In both forest types, thinning (which often occurs 20-40 years after stands undergo even-aged management, especially in lodgepole) immediately depresses hare densities. In spruce-fir stands, densities were estimated to slowly recover through time in nearly linear fashion. However, they follow a peaked response in lodgepole pine, similar to the response to even-aged management. Uneven-aged management of spruce-fir forests results in immediate depression of relative hare density, which then recovers back to pre-treatment levels approximately 30 years after the treatment. Literature Cited: Hodges, K. E., and L. S. Mills. 2008. Designing fecal pellet surveys for snowshoe hares. Forest Ecology and Management 256:1918-1926. Ivan, J. S., and T. M. Shenk. 2016. Winter diet and hunting success of Canada lynx in Colorado. The Journal of Wildlife Management 80:1049-1058. U.S. Fish and Wildlife Service. 2000. Endangered and threatened wildlife and plants: determination of threatened status for the contiguous U. S. distinct population segment of the Canada lynx and related rule, final rule. Federal Register 65:16052–16086. 15 �I '-·---··-- * Control Treatment Figure 1. Location of all stands (n = 130) resampled for snowshoe hare pellets, June-September 2022. Unmanaged Even-aged Uneven-aged Thinned Q) ... -... 0 Q_ (/) C ·o.. Q) 0 c.. Q) 0) 2 -0 0 .J 0 (].) (].) >60 0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60 >60 0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60 Q_ C cu (].) ~ .... .;:::: (I) (.) ::::, a. Cl) Years Since Treatment Figure 2. Fitted quadratic function (white line) and 95% CI (shaded polygon) relating pellet counts (i.e., relative snowshoe hare density) to time elapsed since treatment for each forest type × management activity combination. Dotted lines indicate the mean pellets/plot for the unmanaged controls for each forest type. 16 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Influence of forest management on snowshoe hare density in lodgepole and spruce-fir systems in Colorado Period Covered: January 1, 2022 − December 31, 2023 Principal Investigators: Jake Ivan, Jake.Ivan@state.co.us; All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. Understanding and monitoring snowshoe hare (Lepus americanus) density in Colorado is imperative because hares comprise 70% of the diet of the state-endangered, federally threatened Canada lynx (Lynx canadensis; U.S. Fish and Wildlife Service 2000, Ivan and Shenk 2016). Forest management is an important driver of snowshoe hare density, and all National Forests in Colorado are required to include management direction aimed at conservation of Canada lynx and snowshoe hare as per the Southern Rockies Lynx Amendment (SRLA; https://www.fs.usda.gov/detail/r2/landmanagement/ planning/?cid= stelprdb5356865). At the same time, Forests in the Region are compelled to meet timber production obligations. Such activities may depress snowshoe hare density, improve it, or have mixed effects dependent on the specific activity and the time elapsed since that activity was initiated. Here I describe a sampling scheme to assess impacts of common forest management techniques on snowshoe hare density in both lodgepole pine (Pinus contorta) and spruce-fir (Picea engelmannii – Abies lasiocarpa) systems in Colorado. To select forest stands for sampling, I first used U. S. Forest Service (USFS) spatial data to delineate all spruce-fir and lodgepole pine stands (stratum 1) on USFS land in Colorado, and identified all of the management activities that have occurred in each stand over time. With consultation from the USFS Region 2 Lynx-Silviculture Team and USFS Rocky Mountain Research Station, I then grouped relevant forest management activities (stratum 2) into 4 broad categories: even-aged management, uneven-aged management, thinning, and unmanaged controls. I wanted to assess both the immediate and long-term impacts of management on hare densities. Therefore, when selecting stands for sampling, I took the additional step of binning the date of the most recent management activity into 2decade intervals (i.e., 0-20, 20-40, and 40-60 years before 2018). I then selected a spatially balanced random sample of 5 stands within each combination of forest type × management activity × time interval. This design ensured that I sampled the complete gradient of time since implementation for each management activity of interest in each forest type of interest. There is no notion of “completion date” for unmanaged controls, so I simply sampled 10 randomly selected stands from this combination. Also, uneven-aged lodgepole pine treatments are rare, so I did not sample that combination (Figure 1). During summer 2018, I established n = 50 1-m2 permanent circular plots within each of the stands selected for sampling. Plot locations within each stand were selected in a spatially balanced, random fashion. Technicians cleared and counted snowshoe hare pellets in each plot as they established them. These same plots were re-visited and re-counted during summers 2019 and 2023. In addition to sampling the previously cleared plots from 2018, technicians were able to install plots at 2 more replicate sites for each combination of forest type × management activity × time interval during 2019. In 2021 and 2022, we sampled vegetation metrics in each stand to help account for extraneous noise in the data and allow us 7 �to better assess the effects of the treatments themselves. A handful of initially selected stands were reclassified or excluded during 2019–2022 because ground-truthing and/or vegetation metrics revealed they did not actually fit in the stratum for which they were selected. New stands were sampled in their place by pulling the next one from the spatially balanced list. Similarly, 12 new stands were selected to replace those that burned during the 2020 fire season. Currently, inference is based on n = 130 total stands. Finally, prior to the 2023 field season, I computed the sampling variance of the pellet count for each time interval within each treatment. We sampled additional stands in the 3 most variable bins in an effort to reduce variability and improve our understanding of snowshoe hare response to these treatments. Pellet information from cleared plots is more accurate than that from uncleared plots because uncleared plots usually include pellet accumulation across several years (Hodges and Mills 2008). The degree to which previous years are represented can depend on local weather conditions, site conditions at the plot, and variability in actual snowshoe hare density over previous winters. Data from cleared plots necessarily reflects hare activity from the previous 12 months, and tracks true density more closely. Therefore, I focused the current analysis on the 2019–23 data from previously cleared plots. For each forest type × management activity combination, I plotted mean pellet counts against “year since activity,” then fit a curve (e.g., quadratic function) through the data (Figure 2). Results from this preliminary analysis suggest that on average the highest snowshoe hare densities typically occur in unmanaged spruce-fir forests, and that unmanaged spruce-fir forests are estimated to have more than twice the relative hare density of unmanaged lodgepole pine forests (Figure 2). For both forest types, the fitted line suggests that even-aged management (e.g., clearcutting), immediately depresses relative hare density to near zero, but density rebounds and peaks 20-40 years after management before declining again (lodgepole systems) or leveling off (sprue-fir systems) 40-60 years after. Estimated peak hare densities after even-aged management in lodgepole systems tend to be higher than the control condition. However, in spruce-fir systems the estimated fitted line is flatter and peak densities fell short of the control condition. In both forest types, thinning (which often occurs 20-40 years after stands undergo even-aged management, especially in lodgepole) immediately depresses hare densities. In spruce-fir stands, densities were estimated to slowly recover through time in nearly linear fashion. However, they follow a peaked response in lodgepole pine, similar to the response to even-aged management. Uneven-aged management of spruce-fir forests results in immediate depression of relative hare density, which then recovers back to pre-treatment levels approximately 40 years after the treatment. Literature Cited Hodges, K. E., and L. S. Mills. 2008. Designing fecal pellet surveys for snowshoe hares. Forest Ecology and Management 256:1918-1926. Ivan, J. S., and T. M. Shenk. 2016. Winter diet and hunting success of Canada lynx in Colorado. The Journal of Wildlife Management 80:1049-1058. U.S. Fish and Wildlife Service. 2000. Endangered and threatened wildlife and plants: determination of threatened status for the contiguous U.S. distinct population segment of the Canada lynx and related rule, final rule. Federal Register 65:16052–16086. 8 �Figure 1. Location of all stands (n = 130) resampled for snowshoe hare pellets, June-September 2023. Unmanaged Even-aged Uneven-aged Thinned Q) C ·a. .....0 0a. >60 -.D 0 10 20 JO 40 50 60 0 10 20 JO 40 50 60 0 10 20 JO 40 50 60 >60 0 10 20 JO 40 50 60 0 10 2') JO 40 50 60 0 10 20 JO 40 50 60 Q) a.. .....Q) Cl) Q) 2 Ol "O 0 ...J Q) a.. C cu Q) ~ '-- ~ Q) <..) ::, a. (f) Years Since Treatment Figure 2. Fitted quadratic function (white line) and 95% CI (shaded polygon) relating pellet counts (i.e., relative snowshoe hare density) to time elapsed since treatment for each forest type × management activity combination. Dotted lines indicate the mean pellets/plot for the unmanaged controls for each forest type. 9 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Influence of forest management on snowshoe hare density in lodgepole and spruce-fir systems in Colorado Period Covered: January 1, 2024  December 31, 2024 Principal Investigator: Jake Ivan, Jake.Ivan@state.co.us; All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. Understanding and monitoring snowshoe hare (Lepus americanus) density in Colorado is imperative because hares comprise 70% of the diet of the state-endangered, federally threatened Canada lynx (Lynx canadensis; U.S. Fish and Wildlife Service 2000, Ivan and Shenk 2016). Forest management is an important driver of snowshoe hare density, and all National Forests in Colorado are required to include management direction aimed at conservation of Canada lynx and snowshoe hare as per the Southern Rockies Lynx Amendment (SRLA; https://www.fs.usda.gov/detail/r2/landmanagement/ planning/?cid= stelprdb5356865). At the same time, Forests in the Region are compelled to meet timber production obligations. Such activities may depress snowshoe hare density, improve it, or have mixed effects dependent on the specific activity and the time elapsed since that activity was initiated. Here we describe a sampling scheme to assess impacts of common forest management techniques on snowshoe hare density in both lodgepole pine (Pinus contorta) and spruce-fir (Picea engelmannii – Abies lasiocarpa) systems in Colorado. To select forest stands for sampling, we first used U. S. Forest Service (USFS) spatial data to delineate all spruce-fir and lodgepole pine stands (stratum 1) on USFS land in Colorado, and identified all of the management activities that have occurred in each stand over time. With consultation from the USFS Region 2 Lynx-Silviculture Team and USFS Rocky Mountain Research Station, we then grouped relevant forest management activities (stratum 2) into 4 broad categories: even-aged management, uneven-aged management, thinning, and unmanaged controls. We wanted to assess both the immediate and long-term impacts of management on hare densities. Therefore, when selecting stands for sampling, we took the additional step of binning the date of the most recent management activity into 2-decade intervals (i.e., 0-20, 20-40, and 40-60 years before 2018). We then selected a spatially balanced random sample of 5 stands within each combination of forest type × management activity × time interval. This design ensured that we sampled the complete gradient of time since implementation for each management activity of interest in each forest type of interest. There is no notion of “completion date” for unmanaged controls, so we simply sampled 10 randomly selected stands from this combination. Also, uneven-aged lodgepole pine treatments are rare, so we did not sample that combination (Figure 1). During summer 2018, we established n = 50 1-m2 permanent circular plots within each of the stands selected for sampling. Plot locations within each stand were selected in a spatially balanced, random fashion. Technicians cleared and counted snowshoe hare pellets in each plot as they established them. These same plots were re-visited and re-counted during summers 2019 through 2024. In addition to sampling the previously cleared plots from 2018, technicians were able to install plots at 2 more replicate sites for each combination of forest type × management activity × time interval during 2019. In 2021 and 2022, we sampled vegetation metrics in each stand to help account for extraneous noise in the data and 7 �allow us to better assess the effects of the treatments themselves. A handful of initially selected stands were re-classified or excluded during 2019–2023 because ground-truthing and/or vegetation metrics revealed they did not actually fit in the stratum for which they were selected. New stands were sampled in their place by pulling the next one from the spatially balanced list. Similarly, 12 new stands were selected to replace those that burned during the 2020 fire season. Currently, inference is based on n = 137 total stands. Finally, prior to the 2023 field season, we computed the sampling variance of the pellet count for each time interval within each treatment. We sampled additional stands in the 3 most variable bins in an effort to reduce variability and improve our understanding of snowshoe hare response to these treatments. Pellet information from cleared plots is more accurate than that from uncleared plots because uncleared plots usually include pellet accumulation across several years (Hodges and Mills 2008). The degree to which previous years are represented can depend on local weather conditions, site conditions at the plot, and variability in actual snowshoe hare density over previous winters. Data from cleared plots necessarily reflects hare activity from the previous 12 months, and tracks true density more closely. Therefore, we focused the current analysis on the 2019–24 data from previously cleared plots. For each forest type × management activity combination, we plotted mean pellet counts against “year since activity,” then fit a curve (e.g., quadratic function) through the data (Figure 2). Results from this preliminary analysis suggest that on average the highest snowshoe hare densities typically occur in unmanaged spruce-fir forests, and that unmanaged spruce-fir forests are estimated to have more than twice the relative hare density of unmanaged lodgepole pine forests (Figure 2). For both forest types, the fitted line suggests that even-aged management (e.g., clearcutting), immediately depresses relative hare density to near zero, but density rebounds and peaks 20-40 years after management before declining again (lodgepole systems) or leveling off (sprue-fir systems) 40-60 years after. Estimated peak hare densities after even-aged management in lodgepole systems tend to be higher than the control condition. However, in spruce-fir systems the estimated fitted line is flatter and peak densities fell short of the control condition. In both forest types, thinning (which often occurs 20-40 years after stands undergo even-aged management, especially in lodgepole) immediately depresses hare densities. In spruce-fir stands, densities were estimated to slowly recover through time in nearly linear fashion. However, they follow a peaked response in lodgepole pine, similar to the response to even-aged management. Uneven-aged management of spruce-fir forests results in immediate depression of relative hare density, which then recovers back to pre-treatment levels and beyond approximately 40 years after the treatment. The final season of field sampling for this project was summer 2024. Literature Cited: Hodges, K. E., and L. S. Mills. 2008. Designing fecal pellet surveys for snowshoe hares. Forest Ecology and Management 256:1918-1926. Ivan, J. S., and T. M. Shenk. 2016. Winter diet and hunting success of Canada lynx in Colorado. The Journal of Wildlife Management 80:1049-1058. U.S. Fish and Wildlife Service. 2000. Endangered and threatened wildlife and plants: determination of threatened status for the contiguous U. S. distinct population segment of the Canada lynx and related rule, final rule. Federal Register 65:16052–16086. 8 �0 D,n<>U<Jr t.•toon,I 0 Cheyenne Fort Collins Monum<1nt 0 • .- • OBoulder 0enver 0 0 CD 0 GrandJ~ Greeley Co l or;,do • 0 Colorado Springs Cu, t uri(, N•(,on,I SiteType ~,ot,o,,Ar~, 0 • Control 0 Treatment 0 6) (J 0 F¥mi 0 on Figure 1. Location of all stands (n = 137) resampled for snowshoe hare pellets, June-August 2024. Unmanaged Even-aged Uneven-aged Thinned Q) -0 0... Cl) C ·a. Q) 0 c.. Q) O> --0 0 ....J 0 Q) Q) >60 0 10 20 30 40 50 60 , 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60 0 10 20 30 40 50 60 0... C ,_ ro Q) ..-:::: ~ ::::, (I) (.) Q_ Cf) 0 0 Years Since Treatment Figure 2. Fitted quadratic function (white line) and 95% CI (shaded polygon) relating pellet counts (i.e., relative snowshoe hare density) to time elapsed since treatment for each forest type × management activity combination. Dotted lines indicate the mean pellets/plot for the unmanaged controls for each forest type. 9 � https://cpw.cvlcollections.org/files/original/97737a5539e8a29a1f4eda3c8ee4bdb8.pdf 875f25e54186206e7a558b2d4515969f PDF Text Text Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Mule deer population response to cougar population manipulation Period Covered: January 1, 2023 – December 31, 2023 Principal Investigators: Mat Alldredge, mat.alldredge@state.co.us; Allen Vitt, allen.vitt@state.co.us; Bryan Lamont, bryan.lamont@state.co.us; Ty Woodward, tyrel.woodward@state.co.us; Jamin Grigg, jamin.grigg@state.co.us; Chuck Anderson, chuck.anderson@state.co.us All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. The adopted Colorado mule deer (Odocoileus hemionus) strategy identified predation as one of the potential factors limiting Colorado mule deer populations. Since the adoption of the mule deer strategy by the Colorado Parks and Wildlife (CPW) Commission, members of the CPW Leadership Team developed a plan to implement the strategy. To inform predator harvest and management decisions, staff examined existing data sets related to predator and deer relationships. In June 2015, CPW personnel from the SE Region, Terrestrial, and Research branches met to explore the concept for a project that examines how deer demographic parameters may change following cougar population suppression. Deer Data Analysis Unit (DAU) D-16 had experienced significant deer mortality from cougars. This study initiated in 2017 in D-16 and the adjacent D-34 as a manipulative study to examine the effects of cougar predation on mule deer and simultaneously examine the effects of cougar harvest on the cougar population. To assess the effect of management manipulations, it was necessary to develop an experimental framework including a control and treatment study area. Otherwise, the magnitude of the effect would be unknown as other limiting factors fluctuate. D-34 is an adjacent mule deer DAU to the south of D-16, which has a similar mule deer population size and habitat. Using D-16 and D-34 in a crossover design allowed for the manipulation of a potential limiting factor for mule deer population growth or survival and examine similarities in the response as the control and treatment are switched between the areas. The study's first objective was to assess the impact of cougar predation on mule deer survival and determine if this impact could be manipulated by altering cougar densities. The second objective was to assess how this manipulation would affect the cougar population in terms of intraspecific mortality and human conflict. The manipulation involved increasing cougar harvest in D-16 for the first 3 years of the study and then reducing harvest to a low level for the following 6 years and doing the reverse in D-34 with a reduced harvest for the first 6 years and increased harvest in the last 3 years. During this time we would monitor deer mortality from cougars, measure cougar density, and assess intraspecific cougar mortality and cougar/human conflict in both study areas. To date, deer survival has been relatively high (86% average doe survival D-16 and D-34; 64% average winter fawn survival D-16; 84% average winter fawn survival D-34) in both study areas across years and deer mortality associated with cougars has been low (5.6% does D-16; 7.2% does D-34; 4.2% fawns D-16; 2.1% fawns D-34). Because deer survival was relatively high in the area and mortality associated with cougars was relatively low during the first 6 years of the study, we stopped investigating the impact of cougar predation on deer survival. The remaining treatment was to increase cougar harvest in D34, which presumably would increase deer survival. However, it was decided that it would not be 33 �possible to measure an effect if it did occur with relatively high deer survival evident during the period of low cougar harvest/relatively high cougar density. Graduate student, Annie Hart, at Colorado State University is continuing her Master’s project examining the deer data. The first part of her project examines how variation in natural forage abundance influences mule deer selection of agricultural resources. The other part of her project will model adult and juvenile survival to help understand the costs and benefits of migration. This is using a state uncertainty modeling approach to estimate survival of migrant and resident fawns, which incorporates the survival of individuals that die before their movement strategy is classified. The cougar population component of the study is continuing with assessing impacts of cougar harvest in D-16 and D-34. We continue to estimate cougar density in both study areas and are monitoring intraspecific effects and cougar/human conflict. As this continues, we will maintain a low cougar harvest (quota of 12) in D-16 but need to increase the cougar quota in D-34. The quota in D-34 had been reduced to 15 since the study started, but we proposed an increase in the quota to 35 cougars to start in the 20232024 hunting season, which was approved by the CPW Wildlife Commission in 2023. During the study we have captured and collared 108 cougars in D-16 and 120 in D-34. Last year we captured 11 in D-16 and 20 in D-34. The higher captures in D-34 were related to increased sample size requirements for the cougar survey in D-34 that year. Over the last couple of years collars have been failing sooner than expected, presumably because collar batteries are not lasting as long as they used to. To date, we have completed 3 density estimates in each D-16 and D-34 with preliminary estimates ranging from 2.7 to 3.1 independent cougars per 100 km2. This does not account for any cougars that may have been harvested prior to the initiation of the survey each year. We have not detected a significant change in density relative to changes in harvest quotas or achieved harvest. In 2023 the density estimate was conducted in D-34. Cougar mortality has been relatively low throughout the study, with the majority of this attributable to hunting mortality. Other sources of mortality include disease, intraspecific killing, human conflict removal and unknown. Intraspecific mortality has ranged from 1 to 2 incidences yearly in D-16 and 1 to 3 in D-34 for collared cougars. Cougar/human conflict is variable between years and study areas. This conflict may include livestock depredation, pet depredation, being in unacceptable locations, or aggressive behaviors toward humans. We show conflict rates from 2000-2023 (Figure 1) which shows the variability across time. There may also be variability in these data from how it was reported and recorded, most notably the switch to an electronic/online approach of the conflict app in 2019. D-34 had some of the highest conflict, especially in 2021 and 2023, but historical conflict rates also had occasional high years as well. 34 �Figure 1: Number of human/cougar conflicts in DAUs D-16 and D-34 by year. This does not include sightings. 35 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Mule deer population response to cougar population manipulation Period Covered: January 1, 2024 – December 31, 2024 Principal Investigators: Mat Alldredge, mat.alldredge@state.co.us; Allen Vitt, allen.vitt@state.co.us; Bryan Lamont, bryan.lamont@state.co.us; Ty Woodward, tyrel.woodward@state.co.us; Jamin Grigg, jamin.grigg@state.co.us; Chuck Anderson, chuck.anderson@state.co.us All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. The adopted Colorado mule deer (Odocoileus hemionus) strategy identified predation as one of the potential factors limiting Colorado mule deer populations. Since the adoption of the mule deer strategy by the Colorado Parks and Wildlife (CPW) Commission, members of the CPW Leadership Team developed a plan to implement the strategy. To inform predator harvest and management decisions, staff examined existing data sets related to predator and deer relationships. In June 2015, CPW personnel from the SE Region, Terrestrial, and Research branches met to explore the concept for a project that examines how deer demographic parameters may change following cougar population suppression. Deer Data Analysis Unit (DAU) D-16 had experienced significant deer mortality from cougars. This study initiated in 2017 in D-16 and the adjacent D-34 as a manipulative study to examine the effects of cougar predation on mule deer and simultaneously examine the effects of cougar harvest on the cougar population. To assess the effect of management manipulations, it was necessary to develop an experimental framework including a control and treatment study area. Otherwise, the magnitude of the effect would be unknown as other limiting factors fluctuate. D-34 is an adjacent mule deer DAU to the south of D-16, which has a similar mule deer population size and habitat. Using D-16 and D-34 in a crossover design allowed for the manipulation of a potential limiting factor for mule deer population growth or survival and examine similarities in the response as the control and treatment are switched between the areas. The study's first objective was to assess the impact of cougar predation on mule deer survival and determine if this impact could be manipulated by altering cougar densities. The second objective was to assess how this manipulation would affect the cougar population in terms of intraspecific mortality and human conflict. The manipulation involved increasing cougar harvest in D-16 for the first 3 years of the study and then reducing harvest to a low level for the following 6 years and doing the reverse in D-34 with reduced harvest for the first 6 years and increased harvest in the last 3 years. During this time we would monitor deer mortality from cougars, measure cougar density, and assess intraspecific cougar mortality and cougar/human conflict in both study areas. To date, deer survival has been relatively high (86% average doe survival D-16 and D-34; 64% average winter fawn survival D-16; 84% average winter fawn survival D-34) in both study areas across years and deer mortality associated with cougars has been low (5.6% does D-16; 7.2% does D-34; 4.2% fawns D-16; 2.1% fawns D-34). Because deer survival was relatively high in the area and mortality associated with cougars was relatively low during the first 6 years of the study, we stopped investigating the impact of cougar predation on deer survival. The remaining treatment was to increase cougar harvest in D34, which presumably would increase deer survival. However, it was decided that it would not be 30 �possible to measure an effect if it did occur with relatively high deer survival evident during the period of low cougar harvest/relatively high cougar density. Graduate student, Annie Hart, at Colorado State University finished her Master’s project examining the deer data. The first part of her project examined how variation in natural forage abundance influenced mule deer selection of agricultural resources. The other part of her project modeled adult and juvenile survival to help understand the costs and benefits of migration. This used a state uncertainty modeling approach to estimate survival of migrant and resident fawns, which incorporates the survival of individuals that die before their movement strategy is classified. The cougar population component of the study is continuing with assessing impacts of cougar harvest in D-16 and D-34. We continue to estimate cougar density in both study areas and are monitoring intraspecific effects and cougar/human conflict. As this continues, we will maintain a low cougar harvest (quota of 12) in D-16 but need to increase the cougar quota in D-34. The quota in D-34 had been reduced to 15 since the study started, but we increased the quota to 35 cougars to start in the 2023-2024 hunting season, which was approved by the CPW Wildlife Commission in 2023 and will continue through the 2025-2026 hunting season. This total harvest of 35 was achieved in the 2023-2024 hunting season. During the study we have captured and collared 124 cougars in D-16 and 129 in D-34. Last year we captured 11 in D-16 and 20 in D-34. The higher captures in D-34 were related to increased sample size requirements for the cougar survey in D-34. Over the last two years collars have been failing sooner than expected, presumably because collar batteries are not lasting as long as they used to. To date, we have completed 4 density estimates in each D16 and D34 with preliminary estimates ranging from 2.7 to 3.1 independent cougars per 100 km2. This does not account for any cougars that may have been harvested prior to the initiation of the survey each year. We have not detected a significant change in density relative to changes in harvest quotas or achieved harvest. In 2024 the density estimate was conducted in D16, which is the final estimate for this area. Cougar mortality has been relatively low throughout the study, with the majority of this attributable to hunting mortality. Other sources of mortality include disease, intraspecific killing, human conflict removal and unknown. Intraspecific mortality has ranged from 1 to 2 incidences yearly in D16 and 1 to 3 in D34 for collared cougars. Cougar/human conflict is variable between years and study areas. This conflict may include livestock depredation, pet depredation, being in unacceptable locations, or aggressive behaviors toward humans. We show conflict rates from 2000-2023 (Figure 1) which shows the variability across time. There may also be variability in these data from how it was reported and recorded, most notably the switch to an electronic/online approach of the conflict app in 2019. D34 had some of the highest conflict, especially in 2021 and 2023, but historical conflict rates also had occasional high years as well. 31 �Figure 1: Number of human/cougar conflicts in DAUs D-16 and D-34 by year. This does not include sightings. 32 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Mule deer population response to cougar population manipulation Period Covered: January 1, 2025-December 31, 2025 Principal Investigators: Mat Alldredge, mat.alldredge@state.co.us; Allen Vitt, allen.vitt@state.co.us; Bryan Lamont, bryan.lamont@state.co.us; Ty Woodward, tyrel.woodward@state.co.us; Jamin Grigg, jamin.grigg@state.co.us; Chuck Anderson chuck.anderson@state.co.us All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. The adopted Colorado mule deer (Odocoileus hemionus) strategy identified predation as one of the potential factors limiting Colorado mule deer populations. Since the adoption of the mule deer strategy by the Colorado Parks and Wildlife (CPW) Commission, members of the CPW Leadership Team developed a plan to implement the strategy. To inform predator harvest and management decisions, staff examined existing data sets related to predator and deer relationships. In June 2015, CPW personnel from the SE Region, Terrestrial, and Research branches met to explore the concept for a project that examines how deer demographic parameters may change following cougar population suppression. Deer Data Analysis Unit (DAU) D-16 had experienced significant deer mortality from cougars. This study initiated in 2017 in D-16 and the adjacent D-34 as a manipulative study to examine the effects of cougar predation on mule deer and simultaneously examine the effects of cougar harvest on the cougar population. To assess the effect of management manipulations, it was necessary to develop an experimental framework including a control and treatment study area. Otherwise, the magnitude of the effect would be unknown as other limiting factors fluctuate. D-34 is an adjacent mule deer DAU to the south of D-16, which has a similar mule deer population size and habitat. Using D-16 and D-34 in a crossover design allowed for the manipulation of a potential limiting factor for mule deer population growth or survival and examine similarities in the response as the control and treatment are switched between the areas. The study's first objective was to assess the impact of cougar predation on mule deer survival and determine if this impact could be manipulated by altering cougar densities. The second objective was to assess how this manipulation would affect the cougar population in terms of intraspecific mortality and human conflict. The manipulation involved increasing cougar harvest in D-16 for the first 3 years of the study and then reducing harvest to a low level for the following 6 years and doing the reverse in D-34 with reduced harvest for the first 6 years and increased harvest in the last 3 years. During this time we would monitor deer mortality from cougars, measure cougar density, and assess intraspecific cougar mortality and cougar/human conflict in both study areas. To date, deer survival has been relatively high (86% average doe survival D-16 and D-34; 64% average winter fawn survival D-16; 84% average winter fawn survival D-34) in both study areas across years and deer mortality associated with cougars has been low (5.6% does D-16; 7.2% does D-34; 4.2% fawns D-16; 2.1% fawns D-34). Because deer survival was relatively high in the area and mortality associated with cougars was relatively low during the first 6 years of the study, we stopped investigating the impact of cougar predation on deer survival. The remaining treatment was to increase cougar harvest in D34, which presumably would increase deer survival. However, it was decided that it would not be 35 �possible to measure an effect, if it did occur, with relatively high deer survival evident during the period of low cougar harvest/relatively high cougar density. Graduate student, Annie Hart, at Colorado State University, finished her Master’s project examining the deer data and is currently working on publishing these results. The first part of her project examined how variation in natural forage abundance influenced mule deer selection of agricultural resources. The other part of her project modeled adult and juvenile survival to help understand the costs and benefits of migration. This used a state uncertainty modeling approach to estimate survival of migrant and resident fawns, which incorporates the survival of individuals that die before their movement strategy is classified. The cougar population component of the study is continuing with assessing impacts of cougar harvest in D-16 and D-34. We continue to estimate cougar density and are monitoring intraspecific effects and cougar/human conflict. As this continues, we will maintain a low cougar harvest (quota of 12) in D16 but need to increase the cougar quota in D-34. The quota in D-34 had been reduced to 15 since the study started, but we proposed an increase in the quota to 35 cougars to start in the 2023-2024 hunting season, which was approved by the CPW Wildlife Commission in 2023. This total harvest was achieved in the 2023-2024 and 2024-2025 hunting seasons. To date, we have completed 4 density estimates in each D-16 and D-34 with preliminary estimates ranging from 2.7 to 3.1 independent cougars per 100 km2. This does not account for any cougars that may have been harvested prior to the initiation of the survey each year. We have not detected a significant change in density relative to changes in harvest quotas or achieved harvest. In 2024 the density estimate was conducted in D-16, which is the final estimate for this area. No population estimate was done during 2025 with the final population estimate being scheduled for D-34 in winter of 2026. During the study we captured and collared 124 cougars in D-16 and 133 in D-34. Last year we did not capture in D-16 as population estimates are completed there. Five cougars were captured in D-34 last year to maintain sample size. Over the last couple of years collars have been failing sooner than expected, presumably because collar batteries are not lasting as long as they used to. Cougar mortality has been relatively low throughout the study, with the majority of this attributable to hunting mortality. Other sources of mortality include disease, intraspecific killing, human conflict removal and unknown. Intraspecific mortality has ranged from 1 to 2 incidents yearly in D16 and 1 to 3 in D34 for collared cougars. Cougar/human conflict is variable between years and study areas. This conflict may include livestock depredation, pet depredation, being in unacceptable locations, or aggressive behaviors toward humans. We show conflict rates from 2000-2023 (Figure 1) which shows the variability across time. There may also be variability in these data from how it was reported and recorded, most notably the switch to an electronic/online approach of the conflict app in 2019. D-34 had some of the highest conflict, especially in 2021 and 2023, but historic conflict rates also had occasional high years as well. 36 �Figure 1: Number of human/cougar conflicts in DAUs D-16 and D-34 by year. This does not include sightings. 37 � https://cpw.cvlcollections.org/files/original/ebc16546fe7d12442a8647cef852df81.pdf 6fc87bc6beb69db59af275b35c4b51e4 PDF Text Text Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Pilot evaluation of prey distribution and moose recruitment following exposure to wolf predation risk in North Park, Colorado Period Covered: January 1, 2022 – December 31, 2022 Principal Investigators: Eric Bergman, eric.bergman@state.co.us; Ellen Brandell, ellen.brandell@state.co.us All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. During November 2020, Colorado voters passed Proposition 114 (subsequently codified as Colorado Revised Statue 33-2-105.8), which directed Colorado Parks and Wildlife (CPW) and the CPW Wildlife Commission to develop a gray wolf (Canis lupus) reintroduction and management plan for Colorado by the end of 2023. Wolves are a native species to Colorado and prior to westward European expansion they occurred throughout the Rocky Mountains and into Colorado’s eastern plains (Feldhamer et al. 2003). Since the 1940s, wolf presence in Colorado has been sporadic (Warren 1942, Lechleitner 1969, Armstrong et al. 2011). Beginning in the early 2000s, CPW documented occasional wolf presence in Colorado (Colorado Parks and Wildlife 2021), primarily in North Park. During the summer of 2021, a pack comprised of 2 adults and 6 pups was observed. Between dispersal and reproduction of wolves from neighboring states and reintroductions mandated by Colorado Revised Statute 33-2-105.8, wolves will become a consistent feature on Colorado’s landscape, and specifically in North Park. The return of wolves to Colorado’s landscape has already generated interest in future research projects. Between the 1940s and present day, and largely in the absence of wolves, Colorado’s ungulate prey populations (i.e., elk (Cervus americanus), mule deer (Odocoileus hemionus), and moose (Alces alces)) adapted to many changes. These changes included successional change in vegetation, increases and reductions in competition with other native herbivores and livestock, novel diseases, predation from mountain lions (Felis concolor), black bears (Ursus americanus), and coyotes (Canis latrans), but also increased human activity, human disturbance, and large increases in human infrastructure. Moose experienced deliberate management transplants between the late 1970s (Denney 1976) and mid-2000s. By 2022, Colorado’s moose population was estimated to be 3,000–3,500 animals (Colorado Parks and Wildlife, unpublished data). Similarly, during the 1940s it was believed there were 45,000 elk in Colorado (Swift 1945) and population growth during the next 6–7 decades led to a peak of ~300,000 animals during the late 1990s and early 2000s (CPW, unpublished data). This research is generally focused on predator-prey dynamics and how wolves will influence wild prey. Specifically, this research will measure prey survival, productivity, and distribution. To supplement survival and spatial data collected from moose during 2013–2019 (Bergman 2022), we initiated capture and collaring efforts of cow and calf moose during the winter of 2021–2022. These efforts demonstrated that moose calf abundance and subsequent moose calf density in North Park were insufficient to accommodate the necessary sample size for the initial study design of this project. Historically modeled estimates for the North Park moose herd suggest it is comprised of 600–800 animals. Sex and age distribution data from this herd simultaneously indicate there are ~70 bulls/100 cows and ~52 calves/100 cows, thereby lending evidence that there are ~140–190 calves in North Park. However, it is likely that >50% of these calves reside on private lands during winter, making their access for capture purposes 23 �logistically difficult. Accordingly, there are likely only ~70–95 calves available on public land, of which CPW would need to capture 65%-85% to meet sample size requirements. Capturing such a large proportions of this calf population is both logistically and financially difficult and preliminary efforts in North Park provided evidence that it would be infeasible to capture 60 moose calves each winter. However, capture efforts of cow moose between 2013–2019 (Bergman 2022) and again during the winter of 2021–2022 provided evidence of adequate densities to accommodate robust capturing and collaring efforts, thereby presenting alternative opportunities to estimate calf survival. Advancements in satellite collar technology make it feasible for researchers to attain location data from moose that were collected only a few hours earlier. When coupled with VHF capabilities, researchers have the ability to quickly relocate and observe animals. For the purposes of this study, this technology will allow researchers to observe cow moose, but also observe if cow moose are accompanied by a calf (<12 months old). Repeated observations of cows and calves in this manner, and gathered at key points in time, will allow researchers to approximate calf survival by quantifying the decay in calf/cow ratios from birth to the yearling age class (Lukacs et al. 2004). While these data will not provide causespecific calf mortality estimates, they will improve population models that inform moose ecology and harvest management decision making for the North Park moose herd. To implement this alternative approach to estimating calf survival, a total of 80 cow moose will be collared in North Park. Approximately 65 additional collars will be deployed during winter of 2022– 2023. Collars will be deployed in a spatially balanced manner, with 40 collars on both the northern and southern halves of North Park. To expand this research to include additional prey species, 40 cow elk will be captured and collared during the winter of 2022–2023. Once available for observation, these elk will serve as sentinel animals that will allow researchers to quantify group size behavior, spatial distribution, and habitat use, relative to any known wolf activity. Data collected from cow moose during 2022 did not deviate from data collected during 2013– 2019. Between 2012–2022 survival of cow moose ranged from 91.2%–94.8%. During the same period, pregnancy rates of moose ranged from 54.8%–88.0. Literature Cited: Armstrong, D. M., J. P. Fitzgerald, and C. A. Meaney. 2011. Mammals of Colorado (2nd Edition). University Press of Colorado, Boulder, USA. Bergman, E. J. 2022. Incorporation of moose life history traits, nutritional status, and browse characteristics in Shiras moose management in Colorado. Federal Aid in Wildlife Restoration Annual Report W-245-R4, Ft. Collins, CO USA. Denney, R. N. 1976. A Proposal for the Reintroduction of Moose into Colorado. Colorado Parks and Wildlife, Ft. Collins, USA. Feldhamer, G.A., B.C. Thompson, and J.A. Chapman. 2003. Wild mammals of North America: biology, management, and conservation. Johns Hopkins University Press, Baltimore, MD, USA. Lechleitner, R. R. 1969. Wild mammals of Colorado: their appearance, habits, distribution, and abundance. Pruett Publishing, Boulder, CO, USA. Lukacs, P. M., V. J. Dreitz, F. L. Knopf, and K. P. Burnham. 2004. Estimating survival probabilities of unmarked dependent young when detection is imperfect. Condor 106:926–931. Swift, L. W. 1945. A partial history of the elk herds of Colorado. Journal of Mammalogy 26:114–119. Warren, E. R. 1942. The mammals of Colorado: their habits and distribution (2nd Edition). University of Oklahoma Press, Norma, USA. 24 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Pilot evaluation of prey distribution and moose recruitment following exposure to wolf predation risk in North Park, Colorado Period Covered: January 1, 2023 – December 31, 2023 Principal Investigators: Eric Bergman, eric.bergman@state.co.us; Ellen Brandell, ellen.brandell@state.co.us All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. During November 2020, Colorado voters passed Proposition 114 (subsequently codified as Colorado Revised Statue 33-2-105.8), which directed Colorado Parks and Wildlife (CPW) and the CPW Wildlife Commission to develop a gray wolf (Canis lupus) reintroduction and management plan for Colorado by the end of 2023 (CPW 2023). Wolves are a native species to Colorado and prior to westward European expansion they occurred throughout the Rocky Mountains and into Colorado’s eastern plains (Feldhamer et al. 2003). Since the 1940s, wolf presence in Colorado has been sporadic (Warren 1942, Lechleitner 1969, Armstrong et al. 2011, CPW 2023). Beginning in the early 2000s, CPW documented occasional wolf presence in Colorado (Colorado Parks and Wildlife 2021), primarily in North Park. During the summer of 2021, a pack comprised of 2 adults and 6 pups was observed in North Park. In December 2023, CPW introduced 10 wolves into the state from Oregon, fulfilling the December 31, 2023 deadline set in CRS 33-2-105.8. Between immigration, reintroduction, and reproduction, wolves will become a consistent feature on Colorado’s landscape, and specifically in North Park. The return of wolves to Colorado’s landscape has already generated interest in future research projects. Between the 1940s and present day, and largely in the absence of wolves, Colorado’s ungulate prey populations (i.e., elk (Cervus americanus), mule deer (Odocoileus hemionus), and moose (Alces alces) adapted to many changes. These changes included successional change in vegetation, increases and reductions in competition with other native herbivores and livestock, novel diseases, predation from mountain lions (Puma concolor), black bears (Ursus americanus), bobcats (Lynx rufus) and coyotes (Canis latrans), but also increased human activity, human disturbance, and large increases in human infrastructure. Moose experienced deliberate management transplants between the late 1970s (Denney 1976) and mid-2000s. By 2022, Colorado’s moose population was estimated to be 3,000–3,500 animals (CPW, unpublished data). Similarly, during the 1940s it was believed there were 45,000 elk in Colorado (Swift 1945) and population growth during the next 6–7 decades led to a peak of ~300,000 animals during the late 1990s and early 2000s (CPW, unpublished data). This research is generally focused on predator-prey dynamics and how wolves will influence wild prey. Specifically, this research will measure prey survival, productivity, and distribution. To supplement survival and spatial data collected from moose during 2013–2019 (Bergman 2022), we initiated capture and collaring efforts of cow and calf moose during the winter of 2021–2022. These efforts demonstrated that moose calf abundance and subsequent moose calf density in North Park were insufficient to accommodate the necessary sample size for the initial study design of this project. Historically modeled estimates for the North Park moose herd suggest it is comprised of 600–800 animals. Sex and age distribution data from this herd simultaneously indicate there are ~70 bulls/100 cows and ~52 calves/100 cows, thereby lending evidence that there are ~140–190 moose calves in North Park. However, it is likely 11 �that >50% of these calves reside on private lands during winter, making their access for capture purposes logistically difficult. Accordingly, there are likely only ~70–95 calves available on public land, of which CPW would need to capture 65%-85% to meet sample size requirements. Capturing such a large proportions of this calf population is both logistically and financially difficult and preliminary efforts in North Park provided evidence that it would be infeasible to capture 60 moose calves each winter. However, capture efforts of cow moose between 2013–2019 (Bergman 2022) and again during the winter of 2021–2022 provided evidence of adequate densities to accommodate robust capturing and collaring efforts, thereby presenting alternative opportunities to estimate calf survival. Advancements in satellite collar technology make it feasible for researchers to attain location data from moose that were collected only a few hours earlier. When coupled with VHF capabilities, researchers have the ability to quickly relocate and observe animals. For the purposes of this study, this technology will allow researchers to observe cow moose, but also observe if cow moose are accompanied by a calf (<12 months old). Repeated observations of cows and calves in this manner, and gathered at key points in time, will allow researchers to approximate calf survival by quantifying the decay in calf/cow ratios from birth to the yearling age class (Lukacs et al. 2004). While these data will not provide causespecific calf mortality estimates, they will improve population models that inform moose ecology and harvest management decision making for the North Park moose herd. To implement this alternative approach to estimating calf survival, a total of 80 cow moose will be collared in North Park. In addition to the previously collared moose, 65 moose were collared for the first time in February 2023. Collars were be deployed in a spatially balanced manner, with approximately 40 collars on both the northern and southern halves of North Park. Calf-at-heel surveys were conducted in June and December 2023. 92% and 71% of moose with active collars were observed in the June and December surveys, respectively. Preliminary calf-at-heel ratios were 0.63 and 0.43 calves/cow during the first two surveys. Further analysis and estimation of monthly and annual calf survival rates will be done in the future when all data have been collected. There was some collar failure over the year, which effectively reduces sample size due to inability to locate collared moose during surveys. We plan to collar an additional 5–10 moose in the winter 2023–2024 to meet our desired sample size for calf-at-heel surveys. Data collected from cow moose during 2022 did not deviate from data collected during 2013–2019. Between 2012–2022 survival of cow moose ranged from 91.2%–94.8%. During the same period, pregnancy rates of moose ranged from 54.8%–88.0%. To expand this research to include additional prey species, 40 cow elk were collared in February 2023. These elk will serve as sentinel animals that will allow researchers to quantify group size behavior, spatial distribution, and habitat use, relative to any known wolf activity. To collect these data, we aimed to obtain aerial visual observations of all collared elk on a monthly basis and record the habitat type they occurred in and the size of the elk group they resided in. In addition to estimating group size from the air, we took photographs, allowing us to count elk in groups. We conducted seven aerial surveys from March to December, 2023, and located 50% of collared elk per flight on average. This resulted in 9–19 unique elk groups observed per survey. We will continue approximately monthly elk surveys in addition to the continual locational data collection on GPS collars. Six collared elk died over the year, therefore we plan to collar elk in the winter 2023–2024 to retain our desired sample size of 40 elk. Literature Cited Armstrong, D. M., J. P. Fitzgerald, and C. A. Meaney. 2011. Mammals of Colorado, 2nd ed. University Press of Colorado, Boulder, USA. Colorado Parks and Wildlife. 2023. Colorado wolf restoration and management plan. Denver, USA. Denney, R. N. 1976. A proposal for the reintroduction of moose into Colorado. Colorado Division of Wildlife planning document. 12 �Feldhamer, G. A., B. C. Thompson, and J. A. Chapman. 2003. Wild mammals of North America: biology, management, and conservation. JHU Press, Baltimore, Maryland, USA. Lechleitner, R. R. 1969. Wild mammals of Colorado: their appearance, habits, distribution, and abundance. Pruett Publ. Co., Boulder, Colorado, USA. Swift, L. W. 1945. A partial history of the elk herds of Colorado. Journal of Mammalogy 26:114–119. Warren, E. R. 1942. The mammals of Colorado: their habits and distribution, 2nd (revised) ed. Univ. Oklahoma Press, Norman, USA. 13 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Pilot evaluation of prey distribution and moose recruitment following exposure to wolf predation risk in North Park, Colorado Period Covered: January 1, 2024 – December 31, 2024 Principal Investigator: Ellen Brandell, ellen.brandell@state.co.us All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. During November 2020, Colorado voters passed Proposition 114 (subsequently codified as Colorado Revised Statue 33-2-105.8), which directed Colorado Parks and Wildlife (CPW) and the CPW Wildlife Commission to develop a gray wolf (Canis lupus) reintroduction and management plan for Colorado by the end of 2023 (CPW 2023). Wolves are a native species to Colorado and prior to westward European expansion they occurred throughout the Rocky Mountains and into Colorado’s eastern plains (Feldhamer et al. 2003). Since the 1940s, wolf presence in Colorado has been sporadic (Warren 1942, Lechleitner 1969, Armstrong et al. 2011, CPW 2023). Beginning in the early 2000s, CPW documented occasional wolf presence in Colorado (Colorado Parks and Wildlife 2021), primarily in North Park. During the summer of 2021, a pack comprised of 2 adults and 6 pups was observed in North Park, demonstrating the first wolf reproduction in Colorado in nearly 80 years. In December 2023, CPW introduced 10 wolves into the state from Oregon, fulfilling the December 31, 2023 deadline set in CRS 33-2-105.8. CPW continued their efforts by introducing 15 wolves from British Columbia into Colorado in January 2025. Between immigration, reintroduction, and reproduction, wolves will become a consistent feature on Colorado’s landscape, and specifically in North Park. The return of wolves to Colorado’s landscape has generated interest in future research projects. Between the 1940s and present day, and largely in the absence of wolves, Colorado’s ungulate prey populations (i.e., elk (Cervus americanus), mule deer (Odocoileus hemionus), and moose (Alces alces)) adapted to many changes. These changes included successional change in vegetation, increases and reductions in competition with other native herbivores and livestock, novel diseases, predation from mountain lions (Puma concolor), black bears (Ursus americanus), and coyotes (Canis latrans), but also increased human activity, human disturbance, and large increases in human infrastructure. Moose experienced deliberate management transplants between the late 1970s (Denney 1976) and mid-2000s. By 2022, Colorado’s moose population was estimated to be 3,000–3,500 animals (CPW, unpublished data). Similarly, during the 1940s it was believed there were 45,000 elk in Colorado (Swift 1945) and population growth during the next 6–7 decades led to a peak of ~300,000 animals during the late 1990s and early 2000s (CPW, unpublished data). This research is generally focused on predator-prey dynamics and how wolves will influence wild prey. Specifically, this research will measure prey survival, productivity, and behavior. To supplement survival and spatial data collected from moose during 2013–2019 (Bergman 2022), we initiated capture and collaring efforts of cow and calf moose during the winter of 2021–2022. These efforts demonstrated that moose calf abundance and subsequent moose calf density in North Park were insufficient to accommodate the necessary sample size for the initial study design of this project. Historically modeled estimates for the North Park moose herd suggest it is comprised of 600–800 animals. Sex and age distribution data from this herd simultaneously indicate there are ~70 bulls/100 cows and ~52 calves/100 11 �cows, thereby suggesting that there are ~140–190 calves in North Park. However, it is likely that >50% of these calves reside on private lands during winter, making their access for capture purposes logistically difficult. Accordingly, there are likely only ~70–95 calves available on public land, of which CPW would need to capture 65%-85% to meet sample size requirements. Capturing such a large proportion of this calf population is both logistically and financially difficult, and preliminary efforts in North Park provided evidence that it would be infeasible to capture 60 moose calves each winter. However, capture efforts of cow moose from 2013–2019 (Bergman 2022), and again during the winter of 2021–2022, provided evidence of adequate densities to accommodate robust capturing and collaring efforts, thereby presenting alternative opportunities to estimate calf survival. Advancements in satellite collar technology make it feasible for researchers to attain location data from moose that were collected only a few hours earlier. When coupled with VHF capabilities, researchers have the ability to quickly relocate and observe animals. For the purposes of this study, this technology will allow researchers to observe cow moose, but also observe if cow moose are accompanied by a calf (<12 months old). Repeated observations of cows and calves in this manner, and gathered at key points in time, will allow researchers to approximate calf survival by quantifying the decay in calf/cow ratios from birth to the yearling age class (Lukacs et al. 2004). While these data will not provide causespecific calf mortality estimates, they will improve population models that inform moose ecology and harvest management decision making for the North Park moose herd. To implement this alternative approach to estimating calf survival, we planned to capture and collar a total of 80 cow moose in North Park. In addition to the previously collared moose, 65 moose were collared for the first time in February 2023. Collars were be deployed in a spatially balanced manner, with approximately 40 collars on both the northern and southern halves of North Park. Three calf-at-heel surveys are to be conducted per biological year during June, December, and April; this allows for calculation of survival post-parturition, prior to their first winter, and at nearly one-year old. Calf-at-heel surveys were conducted for the 2023 biological year in June, December, and May, as well as for the 2024 biological year in June and December so far (Table 1). In each survey, cows may not have been located due to dense cover, animal movement from last known GPS location, inaccessible terrain, or collar malfunction. Over time, sample size decreased due to collar failures and harvest. We collared six additional moose in March 2024 to bolster sample size. Further analysis and estimation of monthly and annual calf survival rates will be done in the future when data collection is complete. Thus far, data collected from cow moose during 2023 and 2024 did not deviate from data collected during 2013–2019. From 2012–2022, survival of cow moose ranged from 91.2%–94.8%. During the same period, pregnancy rates of moose ranged from 54.8%–88.0%. Table 1. Preliminary summary of calf-at-heel surveys. Cows observed is reported as a proportion and number. Calf:cow ratios are unadjusted and should not be interpreted as survival. Biological Year 2023 2023 2023 2024 2024 Month June December May June December Cows Observed 0.93 (n = 53) 0.71 (n = 37) 0.86 (n = 51) 0.82 (n = 46) 0.90 (n = 44) Calf:Cow Ratio 0.60 0.43 0.30 0.54 0.48 To expand this research to include additional prey species, 40 cow elk were collared in February 2023. These elk will serve as sentinel animals that will allow researchers to quantify group size behavior, spatial distribution, and habitat use, relative to any known wolf activity. Collars were be deployed in a spatially balanced manner, with approximately 20 collars on both the northern and southern halves of North Park. Six additional elk were collared in March 2024 to maintain our sample size following harvest. 12 �To collect these data, we aimed to obtain aerial visual observations of all collared elk on a monthly basis and record the habitat type and the elk group size. In addition to estimating group size from the air, we took photographs, allowing us to count elk in groups. We conducted sixteen aerial surveys from March 2023 to December 2024 (7 in 2023, 9 in 2024), and located 58% of collared elk per flight on average. This resulted in an average of 13.13 unique elk groups observed per survey. We will continue approximately monthly elk surveys in addition to the continual locational data collection on GPS collars. Literature Cited: Armstrong, D. M., J. P. Fitzgerald, and C. A. Meaney. 2011. Mammals of Colorado (2nd Edition). University Press of Colorado, Boulder, USA. Colorado Parks and Wildlife. 2023. Colorado wolf restoration and management plan. Denver,USA. Denney, R. N. 1976. A proposal for the reintroduction of moose into Colorado. Colorado Parks and Wildlife, Ft. Collins, USA. Feldhamer, G. A., B. C. Thompson, and J. A. Chapman. 2003. Wild mammals of North America: biology, management, and conservation. Johns Hopkins University Press, Baltimore, MD, USA. Lechleitner, R. R. 1969. Wild mammals of Colorado: their appearance, habits, distribution, and abundance. Pruett Publishing Company, Boulder, Colorado, USA. Swift, L. W. 1945. A partial history of the elk herds of Colorado. Journal of Mammalogy 26:114–119. Warren, E. R. 1942. The Mammals of Colorado: their habits and distribution (2nd Edition). University of Oklahoma Press, Norma, USA. 13 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Pilot evaluation of prey distribution and moose recruitment following exposure to wolf predation risk in North Park, Colorado Period Covered: January 1, 2025-December 31, 2025 Principal Investigators: Ellen Brandell, ellen.brandell@state.co.us All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. During November 2020, Colorado voters passed Proposition 114 (subsequently codified as Colorado Revised Statue 33-2-105.8), which directed Colorado Parks and Wildlife (CPW) and the CPW Wildlife Commission to develop a gray wolf (Canis lupus) reintroduction and management plan for Colorado by the end of 2023 (CPW 2023). Wolves are a native species to Colorado and prior to westward European expansion they occurred throughout the Rocky Mountains and into Colorado’s eastern plains (Feldhamer et al. 2003). Since the 1940s, wolf presence in Colorado has been sporadic (Warren 1942, Lechleitner 1969, Armstrong et al. 2011, CPW 2023). Beginning in the early 2000s, CPW documented occasional wolf presence in Colorado (Colorado Parks and Wildlife 2021), primarily in North Park. During the summer of 2021, a pack comprised of 2 adults and 6 pups was observed in North Park, demonstrating the first wolf reproduction in Colorado in nearly 80 years. In December 2023, CPW introduced 10 wolves into the state from Oregon, fulfilling the December 31, 2023 deadline set in CRS 33-2-105.8. CPW continued their efforts by introducing 15 wolves from British Columbia into Colorado in January 2025. Between immigration, reintroduction, and reproduction, wolves will become a consistent feature on Colorado’s landscape, and specifically in North Park. The return of wolves to Colorado’s landscape has generated interest in future research projects. Between the 1940s and present day, and largely in the absence of wolves, Colorado’s ungulate prey populations (i.e., elk (Cervus americanus), mule deer (Odocoileus hemionus), and moose (Alces alces)) adapted to many changes. These changes included successional change in vegetation, increases and reductions in competition with other native herbivores and livestock, novel diseases, predation from mountain lions (Puma concolor), black bears (Ursus americanus), and coyotes (Canis latrans), but also increased human activity, human disturbance, and large increases in human infrastructure. Moose experienced deliberate management transplants between the late 1970s (Denney 1976) and mid-2000s. By 2022, Colorado’s moose population was estimated to be 3,000–3,500 animals (CPW, unpublished data). Similarly, during the 1940s it was believed there were 45,000 elk in Colorado (Swift 1945) and population growth during the next 6–7 decades led to a peak of ~300,000 animals during the late 1990s and early 2000s (CPW, unpublished data). This research is generally focused on predator-prey dynamics and how wolves will influence wild prey. Specifically, this research will measure prey survival, productivity, and behavior. To supplement survival and spatial data collected from moose during 2013–2019 (Bergman 2022), we initiated capture and collaring efforts of cow and calf moose during the winter of 2021–2022. These efforts demonstrated that moose calf abundance and subsequent moose calf density in North Park were insufficient to accommodate the necessary sample size for the initial study design of this project. Historically modeled estimates for the North Park moose herd suggest it is comprised of 600–800 animals. Sex and age distribution data from this herd simultaneously indicate there are ~70 bulls/100 cows and ~52 calves/100 13 �cows, thereby lending evidence that there are ~140–190 calves in North Park. However, it is likely that >50% of these calves reside on private lands during winter, making their access for capture purposes logistically difficult. Accordingly, there are likely only ~70–95 calves available on public land, of which CPW would need to capture 65%-85% to meet sample size requirements. Capturing such a large proportion of this calf population is both logistically and financially difficult, and preliminary efforts in North Park provided evidence that it would be infeasible to capture 60 moose calves each winter. However, capture efforts of cow moose from 2013–2019 (Bergman 2022), and again during the winter of 2021–2022, provided evidence of adequate densities to accommodate robust capturing and collaring efforts, thereby presenting alternative opportunities to estimate calf survival. Advancements in satellite collar technology make it feasible for researchers to attain location data from moose that were collected only a few hours earlier. When coupled with VHF capabilities, researchers have the ability to quickly relocate and observe animals. For the purposes of this study, this technology will allow researchers to observe cow moose but also observe if cow moose are accompanied by a calf (<12 months old). Repeated observations of cows and calves in this manner, and gathered at key points in time, will allow researchers to approximate calf survival by quantifying the decay in calf/cow ratios from birth to the yearling age class (Lukacs et al. 2004). While these data will not provide causespecific calf mortality estimates, they will improve population models that inform moose ecology and harvest management decision making for the North Park moose herd. To implement this alternative approach to estimating calf survival, we planned to capture and collar a total of 80 cow moose in North Park. In addition to the previously collared moose, 65 moose were collared for the first time in February 2023. Collars were deployed in a spatially balanced manner, with approximately 40 collars on both the northern and southern halves of North Park. Three calf-at-heel surveys will be conducted per biological year in approximately June, December, and April; this allows for calculation of survival post-parturition, prior to their first winter, and at nearly one-year old. Calf-at-heel surveys were conducted for the 2023 biological year in June, December, and May, the 2024 biological year in June, December, and April, and the 2025 biological year in June so far (Table 1). Wind and unfavorable weather conditions pushed the December 2025 survey to January 2026. In each survey, cows may not have been located due to dense cover, animal movement from last known GPS location, inaccessible terrain, or collar malfunction. Over time, sample size decreased due to collar failures and mortality, which was primary due to harvest. We collared six additional moose in March 2024 to bolster sample size. Attrition can be viewed in Table 1. Further analysis and estimation of monthly and annual calf survival rates will be done in the future when data collection is complete. From 2012–2022, survival of cow moose ranged from 91.2%– 94.8%. During the same period, pregnancy rates of moose ranged from 54.8%–88.0%. Thus far, data collected from cow moose during 2023 and 2024 did not deviate from data collected during 2013–2019. However, preliminarily, mortality rates appeared to be higher in 2025 due to harvest, vehicle collisions, and natural causes. Table 1. Preliminary summary of calf-at-heel surveys. Cows observed is reported as a proportion and number. Calf:cow ratios are unadjusted and should not be interpreted as survival. Biological Year 2023 2023 2023 2024 2024 2024 2025 Month June December May June December April June Cows Observed 0.93 (n=53) 0.71 (n=37) 0.86 (n=51) 0.82 (n=46) 0.90 (n=44) 0.96 (n=44) 0.93 (n=42) 14 Calf:Cow Ratio 0.60 0.43 0.30 0.54 0.48 0.43 0.67 �To expand this research to include additional prey species, 40 cow elk were collared in February 2023. These elk will serve as sentinel animals that will allow researchers to quantify group size behavior, spatial distribution, and habitat use, relative to any known wolf activity. Collars were deployed in a spatially balanced manner, with approximately 20 collars on both the northern and southern halves of North Park. Six additional elk were collared in March 2024 to maintain our sample size following harvest. To collect these data, we aimed to obtain aerial visual observations of all collared elk on a monthly basis and record the habitat type they occurred in and the size of the elk group they resided in. In addition to estimating group size from the air, we took photographs, allowing us to count elk in groups. We conducted 26 aerial surveys from March 2023 to December 2025 (7 in 2023, 10 in 2024, 9 in 2025) and located 62.4% of collared elk per flight on average. This resulted in an average of 13.0 unique elk groups observed per survey. Elk collars are scheduled to drop off in January 2026. We will collect all dropped collars in 2026 and begin analysis. Literature Cited Armstrong, D. M., J. P. Fitzgerald, and C. A. Meaney. 2011. Mammals of Colorado (2nd Edition). University Press of Colorado, Boulder, USA. Colorado Parks and Wildlife. 2023. Colorado wolf restoration and management plan. Denver, Colorado. 261 pages. Denney, R. N. 1976. A proposal for the reintroduction of moose into Colorado. Colorado Parks and Wildlife, Ft. Collins, USA. Feldhamer, G. A., B. C. Thompson, and J. A. Chapman. 2003. Wild mammals of North America: biology, management, and conservation. Johns Hopkins University Press, Baltimore, Maryland, USA. Lechleitner, R. R. 1969. Wild mammals of Colorado: their appearance, habits, distribution, and abundance. Pruett Publishing Company, Boulder, Colorado, USA. Swift, L. W. 1945. A partial history of the elk herds of Colorado. Journal of Mammalogy 26:114–119. Warren, E. R. 1942. The mammals of Colorado: their habits and distribution (2nd Edition). University of Oklahoma Press, Norma, USA. 15 � https://cpw.cvlcollections.org/files/original/3288fd6b782bb14ed47c8124c74f9a93.pdf 6bcf6ec0bfc2d05255c6d5d48d990e45 PDF Text Text Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Plant and mule deer responses to pinyon-juniper removal by three mechanical methods (follow-up to: Examining the effectiveness of mechanical treatments as a restoration technique for mule deer habitat) Period Covered: July 1, 2020 – June 30, 2021 Principal Investigators: Danielle Johnston (Danielle.bilyeu@state.co.us), Chuck Anderson (chuck.anderson@state.co.us) Personnel: C. Bishop, D. Collins, K. Kain, S. VanNortwick, B. deVergie, D. Finley, L. Gepfert, T. Knowles, B. Petch, J. Rivale, Z. Swennes, M. Way, CPW; L. Belmonte, E. Hollowed, BLM; M. Paschke, G. Stephens, B. Wolk, J. Northrup, B. Gerber, G. Wittemyer, Colorado State University; L. Coulter, Coulter Aviation. Project support received from Federal Aid in Wildlife Restoration, Colorado Mule Deer Association, Colorado Mule Deer Foundation, Muley Fanatic Foundation, Colorado State Severance Tax Fund, Caerus Oil and Gas LLC, EnCana Corp., ExxonMobil Production Co./XTO Energy, Marathon Oil Corp., Shell Petroleum, and WPX Energy. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. Land managers in western North America often reverse succession by removing pinyon (Pinus spp.) and juniper (Juniperus spp.) trees to reduce fire risk and/or increase forage for wildlife or livestock (Monaco and Gunnell 2020). Because prescribed fire is risky, mechanical methods such as chaining, rollerchopping, and mastication are often used (Figure 1). Mechanical methods differ in cost and in the size of woody debris produced, and may also differ in plant and animal responses. We implemented a randomized, complete-block, split-plot experiment in December 2011 in the Piceance Basin, northwestern Colorado, USA, to compare chaining, rollerchopping, mastication and control (whole plots, n = 7) and to explore seeding (subplot) interactions (Figure 2). We assessed plants 1, 2, 5, and 6 years post-treatment, and mule deer (Odocoileus hemionus) response via GPS locations 3-8 years post-treatment. Early results were published previously (Stephens et al. 2016); this effort combines follow-up vegetation data with mule deer responses. By 2016, treated plots had 3-5 times higher perennial grass cover and ~10 times higher cheatgrass (Bromus tectorum) cover than controls (Figure 3). Rollerchopped plots had both the highest annual species cover, and when seeded, also the highest density of bitterbrush (Purshia tridentata), a nutritious shrub for mule deer (Figure 4). Winter deer GPS point detections in chained and rollerchopped plots were almost twice as high as control (P < 0.001), while detections in masticated plots were about 20% higher than control (P ≤ 0.042; Figure 5). Deer detections appear related to a combination of relative hiding cover, resulting from residual woody debris, and winter forage availability. Masticated plots received higher bitterbrush use during summer/fall than chained or rollerchopped plots (P < 0.05; Figure 6). This may have made masticated plots less attractive the following winter, as ungulates tend to browse the most palatable plants and plant parts first (Armstrong and Macdonald 1992). Rollerchopped and chained plots appeared to provide the best combination of mule deer cover and winter forage, but mastication, applied leaving dispersed security cover, may be a viable option where invasive species concerns exist. 17 �Literature Cited: Armstrong, H. M., and A. J. Macdonald. 1992. Tests of different methods for measuring and estimating utilization rate of heather (Calluna-Vulgaris) by vertebrate herbivores. Journal of Applied Ecology 29:285-294. Monaco, T. A., and K. L. Gunnell. 2020. Understory vegetation change following woodland reduction varies by plant community type and seeding status: a region-wide assessment of ecological benefits and risks. Plants 9:1113. Stephens, G. J., D. B. Johnston, J. L. Jonas, and M. W. Paschke. 2016. Understory responses to mechanical treatment of pinyon-juniper in northwestern Colorado. Rangeland Ecology & Management 69:351-359. Figure 1. Equipment, residual structure, and vegetation response 9 years post-treatment for a) chaining, b) rollerchopping, and c) mastication. 18 �Figure 2. Location of tree removal and control plots within north and south Magnolia winter range study areas in the Piceance Basin, Rio Blanco County, Colorado, USA. 19 �Figure 3. Percent cover of A) snowberry, B) perennial grasses, C) exotic annual forbs, and D) cheatgrass 1-6 years following implementation of 3 pinyon and juniper removal methods, unseeded subplots only. Points not sharing letters are significantly different at α = 0.05 for within-year contrasts between treatments. Error bars = 95% CIs. 20 �Figure 4. 2017 bitterbrush density within seeded (solid outline) and unseeded (dashed outline) subplots 6 years after implementation of 3 pinyon and juniper removal methods: CON (control), MAST (masticated), CHAIN (chained, and ROLLER (rollerchopped). Star indicates a significant contrast between seeded and unseeded subplots at α = 0.05. Error bars = 95% CIs. Figure 5. Mule deer GPS locations (points/ha) in winter over a 5-year period in control plots and plots treated to remove pinyon and juniper trees by 3 different methods: CON (control), MAST (masticated), CHAIN (chained), and ROLLER (rollerchopped). Bars not sharing letters are significantly different at α = 0.056. Error bars = 95% CIs. A B Figure 6. Percent of current year growth removed by herbivory during the growing season for A) bitterbrush and B) serviceberry 6 years following implementation of 3 pinyon and juniper removal methods: CON (control), MAST (masticated), CHAIN (chained), and ROLLER (rollerchopped), unseeded subplots only. Bars not sharing letters are significantly different at α = 0.05. Error bars = 95% CIs. 21 � https://cpw.cvlcollections.org/files/original/a9c3a8e3f306c3f5788b2b0aeb8f3c7b.pdf 710515316eedcaa81e674255860be4db PDF Text Text Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Predator community effects pilot study: browsing of upland shrubs in Middle Park, Colorado Period Covered: January 1, 2025 – December 31, 2025 Authors: Danielle B. Johnston, Michael Peyton, and Lauren Brandt Principal Investigators: Danielle B. Johnston (Habitat Researcher, CPW), Ellen Brandell (Wildlife Researcher, CPW), Brian Gerber (Research Ecologist, USGS and Assistant Unit Leader (Colorado Cooperative Fish and Wildlife Research Unit) Project Collaborators: Michael Peyton (Postdoctoral Researcher, CPW); Terrestrial and Field Operations Staff, Area 8; Terrestrial and Field Operations Staff, Area 9; Anthony Vorster, Research Scientist, Colorado State University; Nicholas Young, Research Scientist, Colorado State University All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Effects of predator addition or removal on ecosystem processes are potentially numerous but sometimes subtle. Possible impacts of wolf reintroduction to Colorado ecosystems include changes in prey abundance and/or behavior and indirect effects on plant communities mediated by shifts in browsing pressure. Existing CPW efforts to quantify deer and elk space use and abundance in monitoring areas will offer an opportunity to tie deer and elk responses to wolf space use. In addition, we are proposing to examine the impact of browsing pressure changes on upland shrub communities in Middle Park and the upper Colorado River valley (GMUs 15, 18, 181, 27, 28, 35, 36, 361, 37, 371). Changes in browsing pressure may occur in the near term (1-10 years) due to changes in wolf space use, may occur over a longer time span (10-20 years) as ungulate populations respond numerically to wolf presence, both may occur, or neither. Upland shrub communities include deciduous shrubs such as serviceberry, bitterbrush, and mountain mahogany as well as sagebrush, all of which provide critical winter forage for deer and elk. Shrub seed production and recruitment, which are important for long-term shrub community sustainability, may be sensitive to browsing pressure changes. A study design capable of detecting browsing pressure changes, and the impact of those changes on shrub growth and reproduction, must consider landscape variability and the timescale over which we can reasonably detect changes. We seek to design a cost-efficient study with robust statistical power. To do so, we plan to address three components over a two year period. These components include: 1) improving shrub community maps; 2) determining sample size and design via data simulation; and 3) developing efficient field methods and protocols to �measure shrub productivity, browsing pressure, seed production, and recruitment. We made progress on each of these goals in 2025. Leveraging year end FY25 funds and in collaboration with the NASA DEVELOP internship program, we created new maps of vegetation in deer and elk winter concentration areas in the Middle Park and Upper Colorado River Valley areas. This involved ground-truthing vegetation at 481 locations, 120 with a full protocol with complete plant cover data, and 361 with a quick protocol consisting of cover estimates. The NASA DEVELOP team used these points to train a model based on imagery from Landsat 8-9, Sentinel-2, radar, and lidar. This resulted in a map with overall accuracy of 83% which included detailed classifications such as serviceberry, mountain mahogany, and bitterbrush. To determine sample size and design, we employed a postdoctoral researcher, Michael Peyton, to simulate how shrub cover may respond to changes in browsing pressure. Michael’s simulation incorporates many of the important factors that influence shrub growth, such as soil moisture, fire, and grazing. The simulation represents shrub communities in the landscape of the Middle Park and Upper Colorado River Valley study site in silico (i.e. by computer simulations), parameterized by available environmental data from remote sensing and field sources. The model is initialized using aboveground biomass estimates from across the study site and computes annual biomass change based on major system drivers. Specifically, shrub biomass accumulation is computed from the contribution of soil moisture, soil nitrogen, and their interaction to annual productivity, while biomass is removed through fire, cattle impacts, browsing, and maintenance respiration. These gain and loss terms are informed by established relationships in the published literature for shrubs in this system. Uncertainty and process noise are incorporated in simulation outputs by iteratively testing model behavior across plausible parameter ranges for these relationships. Finally, the simulation tests differences in sample size, stratification approaches, and statistical methods to provide estimates of detection probability under different design choices. These outputs will help guide our sampling approach and provide a defensible basis for determining where and how to prioritize sampling effort. To develop efficient field methods, we completed tasks related to shrub aging, seed production method development, plot setup, and browsing pressure estimation. The ability to accurately age young shrubs is critical for our ability to assess recruitment. In 2025 we used root crown ring counts to age 75 shrubs. 40 of these were of known ages, having come from prior experiments, and this allowed us to assess our ability to age shrubs by various methods. We were able to determine actual age from ring counts with good accuracy (R2 = 0.88), and root crown diameter approximated ring-estimated age with reasonable accuracy (R2 = 0.66). For seed production, we counted seeds on 19 shrubs, documented shrub density classes with photographs, and assessed our ability to visually estimate density class. We completed 6 practice plots combining estimates of shrub density, productivity, seed production, and recruitment, which allowed us to streamline our protocols. Finally, we constructed grazing cages around 26 shrubs in early fall 2025 to prepare them for a browsing pressure method development study. In early 2026, we will select locations for permanent shrub sampling plots based on our maps of shrub communities and the results of the simulation analysis. In spring of 2026, we will continue our browsing pressure methods study and refine methods for seed production, productivity, and recruitment through the summer and fall. We will establish as many of the permanent plots as possible and begin taking data to be used in long-term assessments. We also plan to submit a manuscript concerning the sample size estimation process. �BACKGROUND Gray wolves (Canis lupus) are re-establishing in Colorado, joining mountain lions (Puma concolor) and black bears (Ursus americanus) as key predators of ungulates. This re-establishment is expected to cause community effects, defined as direct or indirect changes to vegetation, prey, or other predators. The complexity of Colorado's ecosystems—characterized by wide ranges in elevation, plant communities, hunting practices, and land use—makes conclusively demonstrating a classical trophic cascade unlikely. Trophic cascades fundamentally involve changes in one trophic level causing changes in two or more lower trophic levels. While these are dramatic in simple, discrete systems such as tide pools (Paine 1980), complex communities are often buffered by numerous species interactions (Brice et al. 2022). We focus instead on wolf community effects, which encompass a wide range of potential interactions, including competition with other predators and alterations to the movement, habitat selection, diet, and/or density of elk (Cervus canadensis) and mule deer (Odocoileus hemionus) as well as impacts to vegetation. A key community effect we aim to evaluate is the impact of wolves on elk and mule deer space use. We will capitalize on existing Colorado Parks and Wildlife (CPW) monitoring efforts, which maintain Global Positioning System (GPS) collars on ungulates within five existing Mule Deer Intensive Monitoring Areas and five new, spatially overlapping Elk Intensive Monitoring Areas. This provides extensive data on movement, space use patterns, and survival, and will be supplemented by new efforts to estimate elk abundance. This coordination of efforts is intended to require minimal additional budget over the next few years. The main focus of this pilot phase is the impact of wolves on upland shrub communities, a critical forage resource for ungulates. We specifically select upland deciduous shrubs, such as serviceberry (Amelanchier alnifolia), mountain mahogany (Cercocarpus montanus), and bitterbrush (Purshia tridentata), along with sagebrush (Artemisia tridentata). We have strategically chosen to focus on these upland shrubs instead of more commonly studied species like willow or aspen. This decision is based on several ecological and management objectives. First, unlike willow and aspen, the impact of browsing on upland shrubs has received very little scientific attention in the context of top-down predator effects. By focusing here, we aim to fill a significant gap in Rocky Mountain ecology. Secondly, serviceberry, mountain mahogany, and bitterbrush are critical, highvalue forage species that receive heavy ungulate browsing on winter ranges in Colorado. They constitute a large fraction of winter diet, particularly for mule deer (Bartmann 1983), and increasing their productivity is a common CPW management objective. Although sagebrush is less preferred, it remains an important part of mule deer winter diet (Richens 1967, Carpenter et al. 1979), and CPW is actively involved in its restoration (Tarbox et al. 2025). Lastly, deciduous upland shrubs may be vulnerable to climate changes such as increased drought or fire. As upland plants, they are dependent on precipitation for water and therefore may be more susceptible to drought than willows, which rely on stream or ground water. Furthermore, while aspen often benefits from fire, upland shrubs often decline following fire �(Updike et al. 1992). Investigating how browsing pressure changes interact with these climate-driven vulnerabilities is crucial for future management. Wolves can influence browsing pressure on upland deciduous shrubs through two types of indirect effects: ● ● Behaviorally-Mediated Indirect Effects (BMIE): Occur when wolves change prey behavior (e.g., space use, increased vigilance), which then drives reduced browsing in certain areas. ○ Timeline: Changes could occur relatively quickly (1–10 years) once wolves are establish and introduce spatially heterogeneous predation risk. ○ Pattern: Responses may be finer-scale and uncertain, as they are influenced by both topdown (predator) and bottom-up (forage preference, site fidelity) processes. Ungulates may adopt vigilance, altered diel patterns, and/or group size changes instead of abandoning preferred areas (Mao et al. 2005, Liley and Creel 2008, Kohl et al. 2018). Density-Mediated Indirect Effects (DMIE): Occur when wolves reduce prey density (population size), which drives reduced browsing across the landscape. ○ Timeline: Requires a longer timescale (10–20 years) for wolf populations to reach a size that significantly reduces elk and/or deer density. ○ Pattern: Expected to be straightforward: lower prey density leads to decreased landscapescale browsing pressure. Given the complexities of BMIEs and the wide-ranging nature of wolves and ungulates, DMIEs are likely the dominant driver of long-term plant community response (Schmitz et al. 2004). Our proposed long-term study will be designed to detect changes mediated by both mechanisms, requiring a significant time commitment and resources. To ensure feasibility, statistical power, and cost-effectiveness for a long-term study, we are conducting a two-year pilot study focused on elk and mule deer winter ranges in Middle Park and the upper Colorado River valley (GMUs 15, 18, 181, 27, 28, 35, 36, 361, 37, 371), areas with current wolf activity and co-occurring elk and deer monitoring efforts. The pilot study focuses on three key goals: 1. Mapping: Develop an accurate spatial map of the extent of focal plant communities. 2. Sample Size Estimation: Perform a simulation exercise to determine the feasibility of detecting effects on shrubs from browsing pressure changes of varying magnitude and extent and to identify critical covariates to measure (e.g. soil moisture). 3. Field Methods Development: Identify the most efficient field methodology for measuring browsing pressure, recruitment, and growth of shrubs. Determine efficient ways to measure critical covariates. The overall goal is to inform a robust, low-annual-cost observational study that can be sustained by internal CPW funding to monitor wolf community effects over the necessary extended timeframe. �Ultimately, we hope to relate vegetation responses to predator and prey data to address broader hypotheses, such as those outlined below. Note that these hypotheses may change based on feasibility as revealed in our pilot study 1: ● Are wolves causing changes in upland deciduous shrub and sagebrush communities? ○ ○ Behaviorally-mediated indirect effects ■ H1a: (Bottom-up processes dominate). Browsing pressure and wolf space use metrics will be either uncorrelated or positively correlated. Height gain, cover, seed production, and recruitment of shrubs (or temporal changes in those metrics) will be uncorrelated or negatively correlated with wolf space use metrics. ■ H1b: (Top-down processes dominate). Browsing pressure and wolf space use metrics will be negatively correlated. Height gain, cover, seed production, and recruitment of shrubs (or temporal changes in those metrics) will be positively correlated with wolf space use metrics. Density-mediated indirect effects ■ ● H2: Browsing pressure will lessen with increasing wolf density. Height gain, cover, seed production, and recruitment of shrubs will increase with increasing wolf density. Is ungulate use limiting to upland deciduous shrubs and sagebrush? ○ ○ Behavioral direct effects ■ H3: Browsing pressure will be positively correlated with elk and mule deer space use. ■ H4a: (Bottom-up processes dominate) Height gain, cover, seed production, and recruitment of shrubs will be positively correlated with elk and mule deer space use. ■ H4b: (Top-down processes dominate) Height gain, cover, seed production, and recruitment of shrubs will be negatively correlated with elk and mule deer space use. Density direct effects ■ H5: Browsing pressure will be positively correlated with elk and mule deer density. Height gain, cover, seed production, and recruitment of shrubs will be negatively correlated with elk and mule deer density. Here, we separate hypotheses related to space use versus density. We define space use as a measure of time spent formulated from GPS data, such as a probability density of wolf GPS locations. Space use has a spatial resolution relevant to individual vegetation sampling locations, and a temporal resolution at the season level or finer. We define density as population at the level of our entire study area with a temporal resolution of one year. 1 �Importantly, we are not proposing to examine wolf-prey relationships in this pilot study, but nonetheless, these hypotheses are important for guiding our work. 2025 ACTIVITIES In 2025, we made progress on all three of our goals for the pilot study: shrub mapping, sample size determination via data simulation, and field methods development. The first step to support these processes was precisely defining our study area. Study area definition We chose to work in ungulate winter range because this is where we expect a reduction in browsing pressure to be most likely to benefit plants. We defined winter range in Middle Park and the Upper Colorado River Valley based on species activity mapping (SAM) layers. We consulted with CPW personnel knowledgeable about the layers (Chuck Anderson, Dani Neumann, Jon Runge) and considered the spatial extent of Winter Range, Winter Concentration Area, and Severe Winter Range for both elk and mule deer. We concluded that Winter Concentration Areas were the most likely to receive winter browsing pressure from both elk and mule deer in most years. We then explored the outputs from intersecting versus merging the Winter Concentration Area layers for species (elk and mule deer) and for SAM mapping years (2014, 2018, and 2023), preferring to intersect the layers to more easily design a study that captures both elk and mule deer responses. To ensure that we are capturing the majority of upland shrub communities experiencing winter browsing pressure, we merged SAM layers across the 3 years for each species before intersecting them, and then buffered the result by 1 km in order to connect polygons that were near each other and produce a more cohesive final layer. Shrub mapping CPW biologists consistently report that nationwide remotely-sensed products, such as LANDFIRE, do a poor job of mapping deciduous upland shrub communities in Colorado. Accurate maps of the shrub communities we wish to sample are required for assessing landscape scale responses to changes in browsing pressure. In 2025, we addressed this problem by leveraging year-end FY 25 funds and a program free to us, the NASA DEVELOP remote sensing internship program. In June 2025, we captured plant community data at 481 points within our study area, as defined above. Locations for sampling were selected on public land, and within 1.6 km of roads. We collected data at 125 randomly selected points and 356 opportunistic points. To increase our likelihood of sufficiently sampling within upland deciduous shrub communities, which are rare on the landscape, we selected half of the random points from the Rocky Mountain Lower Montane-Foothill Shrubland classification from SWReGAP, the most accurate pre-existing classification we had identified based on local field knowledge. The remainder were selected without regard to prior remote sensing classification. We verified with the NASA DEVELOP team manager, Tony Vorster, that the spatial extent of our sampling locations were sufficient to support modeling to the entire study area, as they encompassed the range of elevations and shrub communities of interest. At 120 randomly selected points, we collected cover data for a 5m radius plot in two ways: 1) quick ocular assessment of cover of the three most dominant plant groups visible from above, categorized by groupings of interest (serviceberry, bitterbrush, mountain mahogany, perennial grass, perennial forb, �pinyon, juniper, aspen, rabbitbrush, sagebrush, snowberry, riparian shrub, riparian tree, other shrubs); and 2) complete line-point-intercept data taken for all canopy layers, to species, in a circle-and-crosshairs pattern. At the remainder of the points, we collected cover data only by quick ocular assessments. At all points, we collected categorical, qualitative estimates of shrub browsing pressure, which was used to initialize our sample size estimation data simulation (see following section). All data was taken with Survey 123 and sent weekly to the NASA DEVELOP team, who consisted of Ashley Bañuelos, Erin Burke, Scott Mohan, and Sheyla Rios Galeano. The NASA DEVELOP team found that our quick assessments related to the LPI data by R2 of 0.64 to 0.87 for our shrub species of interest. They used the quick assessments in a cluster analysis to define plant community associations, keeping in mind our interest in distinguishing between upland shrub community types. Next, they modeled these vegetation classifications across our study area using imagery from Landsat 8-9, Sentinel-2, Radar, and Lidar (Figure 1). Figure 1. Procedure used by the NASA DEVELOP interns to map shrub communities in ungulate winter range in Middle Park and the Upper Colorado River Valley. Image credit: Ashley Bañuelos, Erin Burke, Scott Mohan, and Sheyla Rios Galeano. We are still awaiting the finished report from NASA DEVELOP, but we have received a map and accuracy assessments of each classification. The map based on LANDSAT had slightly higher overall accuracy than the SENTINEL map, but the SENTINEL map has a finer resolution. Accuracy for serviceberry was satisfactory, with 83% user accuracy (83% of pixels labeled as serviceberry are actually dominated by serviceberry) and 85% producer accuracy (pixels labeled as serviceberry capture 85% of pixels actually dominated by serviceberry). However, accuracy for bitterbrush was less satisfactory, with 100% user accuracy but only 25% producer accuracy, indicating that many points dominated by bitterbrush were not captured by the model (Table 1). �Table 1. Classifications modeled by the NASA DEVELOP intern team for ungulate winter ranges in Middle Park and the Upper Colorado River Valley, with accuracy for LANDSAT-based and SENTINEL-based models. 43 PREDICTORS NO TERRAIN VARIABLES #ID Class LANDSAT Producer Accuracy SENTINEL User Accuracy Area (sq miles) Percent cover Producer Accuracy User Accuracy Area (sq miles) Percent over 1 Rabbitbrush and Snowberry 0.63 0.83 42.53 6.1% 0.75 0.83 20.73 3.0% 2 Bitterbrush 0.42 1 0.15 0.0% 0.25 1 0.12 0.0% 3 Mountain Mahogany 0.63 0.95 4.84 0.7% 0.56 0.86 9.67 1.4% 4 Perennial Grass /Sagebrush 0.99 0.77 419.68 60.0% 0.97 0.79 404.04 58.4% 5 Aspen /Gambel Oak 0.47 0.90 5.66 0.8% 0.42 0.62 44.15 6.4% 6 Other Conifer 0.86 0.83 67.41 9.6% 0.84 0.88 88.72 12.8% 7 Serviceberry 0.67 0.94 12.29 1.8% 0.85 0.83 24.54 3.5% 8 Pinyon Juniper 0.9 0.83 42.85 6.1% 0.81 0.64 34.64 5.0% 9 Mixed Riparian Shrub and Tree 0.33 1 1.91 0.3% 0.33 1 4.68 0.7% 10 Water 0.93 1 5.91 0.8% 0.90 1 5.16 0.7% 11 Agriculture 0.97 0.9 67.15 9.6% 0.94 0.88 32.90 4.8% 12 Developed/ Bare Ground 0.77 0.9 22.91 3.3% TOTAL OF OBS 614 699.87 0.63 0.88 614 18.13 692.06 OVERALL ACCURACY (GT DATA) 0.83 0.82 OUT OF BAG ERROR 0.39 0.37 2.6% �This may be because bitterbrush is often intermixed with many other species, including sagebrush, serviceberry, mountain mahogany, and pinyon. In practice, it was very difficult to find locations that had bitterbrush cover exceeding 25%. Even so, the model will be a useful starting point to guide long-term study site selection, as it provides a finer-scale and more shrub-focused classification than previously available (Figure 2). Figure 2. Modeled vegetation classification map for ungulate winter ranges in Middle Park and the Upper Colorado River Valley, based on 10m2 pixel resolution Sentinel imagery. Sample size estimation: simulation Approach Determining sample sizes and study duration is dependent on the number of indirect effects we are interested in measuring (i.e., response variables), the number of variables we are interested in quantifying or controlling for (i.e., explanatory variables), and the strength of their interactions. We started to address this by conducting data simulations, informed by pilot field data, remotely sensed data products, and peer-reviewed literature, which explore what magnitude of effects are feasible to detect under different design decisions. Brian Gerber (USGS Colorado Cooperative Fish and Wildlife Research Unit) led this component with the help of a post-doctorate researcher, Michael Peyton, whom we hired through Colorado State University with allocated FY25 and FY26 Mammals Research funds. Our study design must be capable of accounting for (via study design or modeling) the factors that can impact shrub productivity and biomass change. To determine how to parameterize the simulation, we first identified the primary drivers influencing shrub dynamics in Middle Park. We selected soil moisture, soil nitrogen, fire, cattle use, and ungulate browsing as drivers to be explicitly modelled in the simulation, while other factors – which may have minor contributions to biomass change – were modelled as process noise. �We simulated the study site as a collection of 20 m x 20 m cells encompassing the extent of the shrub community in Middle Park. We leveraged the NASA DEVELOP shrub map, masked by vegetation communities identified as upland shrublands, to determine the operational study area. Using this sampling frame, we extracted data from remotely sensed and publicly available sources capturing aboveground biomass (NASA GEDI), soil moisture (ERA5-LAND), soil nitrogen (SoilGrids), burn probability (Colorado Wildfire Risk Assessment), and cattle allotment jurisdictions (BLM and USFS grazing allotment boundaries). Raw data sources or probability distributions capturing these data were used to initialize conditions and project the annual variability of major drivers across cells, when applicable. This simulation is not spatially explicit; we determined that modelling this variation without spatial structure was sufficient for our primary objective of evaluating sample design and detection power. Methods The simulation proceeds in annual timesteps, with biomass gains and losses computed as a function of measured relationships among major drivers from published, peer-reviewed literature. Estimates of these relationships vary among studies, so each relationship is designed to be modelled across multiple plausible levels to capture the range of variation estimated in the literature. All processes outlined below are modelled with process noise informed by the literature, ensuring the contribution of other drivers not explicitly incorporated into the model are indirectly captured in the spatial and temporal variability across cells. First, annual biomass gain is computed through the effect of soil moisture, soil nitrogen, and their interaction on annual primary productivity. These gains are added to the existing crop of standing biomass on a per-cell basis. Next, fire is simulated by using per-cell burn probability to project and identify burned cells during this timestep. Cells identified as burned incur biomass losses based on probability distributions derived from the literature on measured losses due to fire events in this ecosystem. Post-fire trajectories of biomass development differ from the dynamics observed in unburned areas, but estimates differ widely based on a variety of factors, including pre-burn site conditions, existing vegetation, and founder effects. Moreover, fire can alter abiotic conditions such as soil moisture and nitrogen, directly influencing factors important for productivity. Due to the uncertainty involved in the effects of fire on these factors collectively, we modelled postfire trajectories directly, using a stepwiselinear form with no growth immediately following fire initialization followed by recovery. Both the recovery trajectory and the years of recovery (before returning to pre-fire conditions) were allowed to vary across the range of estimates found in the literature. Next, the effect of cattle on shrub biomass is computed by a two step process: first, the probability of cattle use is applied within allotment cells, then the grazing effect is applied on cells selected that year. Estimates of actual use are not available by public record nor are collected by government agencies, so estimates of usage probability come from personal communication with the Bureau of Land Management and the US Forest Service, with appropriate noise to encapsulate uncertainty. Estimates of the effect of cattle on shrub productivity range widely, with most sources reporting some losses due to herbivory but some sources reporting minor gains due to a reduction in grass competition. We therefore parameterized the effects of cattle to range from small, positive change in growth to progressively larger losses, informed by the literature. Effects were scaled to the current year’s productivity but applied to standing biomass, allowing effects that exceeded annual productivity projections to impact existing biomass. �Finally, the model applies winter browsing pressure from ungulates, which come from estimates of losses in the field. Similarly to the effects of cattle, these effects are scaled to the current year’s productivity, but are applied to standing biomass and can exceed annual productivity. Mule deer and elk show some degree of site fidelity across years, which we captured by assigning cells to none, low, medium, or high browsing pressure categories but allowed significant variation in year-to-year proportional losses within each category (except ‘none’) to reflect temporal variability in site use. Browsing intensity at a site depends heavily on the composition of shrubs within a vegetation community and their palatability. To determine the distribution of browsing pressure categories across vegetation communities, we used field estimates of browsing pressure (categorized similarly as none, low, medium, or high) to understand differences in browsing pressure across the vegetation communities identified by NASA DEVELOP. We detected two distinct distributions of browsing intensity categories across vegetation communities using PERMANOVA, which we collapsed into two vegetation ecotypes broadly consisting of shared composition. We identified a generally less-extensively browsed, Sagebrushdominated system (hereafter, Sagebrush community), and a more heavily-browsed system abundant with palatable shrubs (e.g. bitterbrush, mountain mahogany, serviceberry; hereafter, Mtn. Shrub community). Using these community distinctions, we identified the proportion of the landscape at the study site consisting of the Sagebrush community (94.2%) and the Mtn. Shrub community (5.8%) to apply our field-derived measurements of browsing pressure, and randomly assigned each cell a vegetation community based on these proportions. Then, for each cell, we applied a browsing intensity category (none, low, medium, or high) based on probability distributions of these categories specific to the cell’s assigned vegetation community. Thus, vegetation communities here represent differences in initial browsing conditions, with the Sagebrush community representing sites dominated by less palatable vegetation, and Mtn. Shrub communities representing those with more palatable species. Respiration costs are an important constraint on growth and maximum biomass. We modelled this dynamic as a loss of standing biomass based on the initial biomass at the current timestep. This loss function follows a well-established power law, where respiration losses are proportional to the ¾ power of standing biomass. While the exponent of this power law is well-studied, the constant used to scale standing biomass to biomass loss from respiration is not. Therefore, we tested a range of values for this constant to determine which value allows us to best approximate the observed values of standing biomass in Middle Park. We proceeded using 0.3, with the option of adjusting as new information becomes available. Respiration losses are calculated from initial biomass after annual productivity is added to the standing crop, but applied after losses from fire, cattle, and ungulate browsing. Impacts of predator space-use and density on ungulate density and behavior are extremely variable and contingent, and current data do not support parameterizing wolf - ungulate interactions. Rather than explicitly modelling predator-prey dynamics, we instead impose a change in browsing pressure directly as a shift in annual herbivory losses. This allows for the testing of sample design across a range of plausible scenarios while avoiding the additional uncertainty from poorly constrained parameters. We first allowed the simulation to run for 30 annual timesteps (i.e. 30 years) to allow initial conditions to equilibrate, then applied a shift in browsing pressure which effectively subtracted the proportion of annual productivity removed through ungulate browsing by 0.1 (low), 0.3 (medium), or 0.5 (high). �We used shrub cover as the primary metric for detecting vegetation change to align with commonly used monitoring protocols. We converted standing biomass to cover using a saturating Michaelis-Menton curve from the literature, with process noise to reflect variation in canopy architecture. Sample plots were randomly assigned within vegetation communities at the beginning of each simulation run and sampled each year to represent permanent plots. Each plot is equivalent to a single 20 m x 20 m cell. Random noise was incorporated to reflect sampling error for sampled cover and sampled browsing pressure. The spatial extent of wolf indirect effects on vegetation is unknown and difficult to predict. To test how the spatial extent of predator impacts influences detection, and to capture across the range of uncertainty, we modelled four scenarios corresponding to differing breadths of browsing pressure change: local (10% of the landscape), regional (30% of the landscape), landscape (60% of the landscape), and all (100% of the landscape). Within each of these scenarios, we applied low, medium, and high browsing pressure shifts to run a total of 12 scenarios corresponding to differences in the spatial extent and intensity of predator impacts on browsing. Across these scenarios, we tested four sample sizes reflecting the sampling effort plausible for each field season: 30, 60, 90, and 130 sample plots within each vegetation community for a total of 60, 120, 189, and 260 samples across both Sagebrush and Mt. Shrub communities. Thus, we tested across 48 scenarios to understand detection with varying sampling effort. To evaluate how stratified sampling influences detection probability, we ran each set of 48 scenarios under five sampling schema: stratification by soil moisture, soil nitrogen, initial browsing pressure, and cattle allotments, plus an unstratified design within each vegetation community. We used a simple allocation structure with an equal number of plots within each stratum. For continuous variables (soil moisture and soil nitrogen), we defined strata as low, medium, and high by splitting the data into three groups of equal size. For categorical variables (initial browsing pressure category and cattle allotments), strata corresponded to the existing categories. Comparing detection performance across these sampling schemes allowed quantification of the effect of stratification on our ability to detect changes. Detection probability depends not only on sample design but also on the statistical methods used to evaluate vegetation change. To compare alternative statistical methods, we implemented four statistical tests representing different approaches to test for the impact of wolves on shrub dynamics. Because each test was applied to 1000 simulation iterations for each of the 48 scenarios, we used streamlined implementations of each statistical approach that retain their core structure while keeping computation tractable. For the first three tests, we compared measurements from a five-year window immediately before wolf establishment to a five-year window following a 10-year equilibration period after establishment. We also fit a fourth model that compared temporal trends in cover across the pre- and postestablishment periods. First, we used a simple paired t-test on plot-level mean cover in the pre- and post-establishment windows, representing a naive test for a step change in vegetation. Second, we fit a mixed-effects model comparing cover across pre- and post-establishment periods, with period as a fixed effect and plot as a random effect. For scenarios with stratified sampling, we also included a fixed effect for stratum. This test represented a modelling approach that explicitly accounts for repeated measurements and stratification, when applicable. Third, we used a piecewise Structural Equation Model (SEM) which required two conditions for detection: (1) a mixed-effects model testing the effect of pre- vs post-establishment period �on browsing pressure, and (2) a mixed-effects model testing the effect of browsing pressure on shrub cover while also including establishment period as a covariate to capture other changes within each period. Both models included plot as a random effect. Together, these models provide an explicit test of a tri-trophic cascade by establishing both the link between browsing pressure (representing ungulate population density) and wolf establishment, and the link between browsing pressure and shrub cover. Finally, we fit a mixed effect model to test differences in the temporal trend between the five-year window immediately before establishment and a five year window immediately after establishment. Cover was modelled as a function of time, establishment period, and their interaction, with a plot-level random effect. Detection occurred when the time x period interaction was significant, indicating a temporal slope change between the pre- and post-establishment period. Results Detection probability was strongly influenced by both spatial scale and the magnitude of the browsing pressure shift (Figure 3). �Figure 3. Detection probability for the Mtn. Shrub community across four alternative statistical approaches. Rows show four different statistical tests used to detect an indirect effect of wolf establishment on shrub cover: (a) simple paired t-test on plot-level mean cover pre- vs. postestablishment, (b) mixed-effects model comparing pre- and post-establishment periods, (c) piecewise Structural Equation Model (SEM) testing both (i) the effect of establishment on browsing pressure and (ii) the effect of browsing on cover, and (d) mixed-effects model testing a change in temporal slope in the pre- and post-establishment periods. Columns show differences in the spatial extent of establishment, with local (10% of the landscape), regional (30% of the landscape), landscape (60% of the landscape), and all (100% of the landscape) from left to right. Lines indicate the magnitude of the shift in browsing pressure, with blue lines showing a low shift (0.1), yellow lines showing a medium shift (0.3), and green lights showing a high shift (0.5). The y axis shows detection probability, computed as the percentage of simulation iterations (n = 1000) where a shift was detected using the given statistical approach. The x axis shows variation in sample size. Each point represents a single simulation run for a given combination of spatial scale, sample size, browsing shift magnitude, and statistical test. At the local scale (10%), all four statistical approaches had low power, and differences in detection among browsing shifts were small. As spatial scale increased to regional (30%) and landscape (60%) extents, detection improved and the magnitude of the browsing-pressure shift became more important for detection probability. While moderate (0.3) and high (0.5) shifts in browsing pressure met our target detection (~ 80% power) at landscape scales, low (0.1) shifts remained difficult to detect across methods. Detection improved with sample size, but gains generally diminished beyond ~ 60 plots for most combinations of scale, shift magnitude, and statistical test. This pattern suggests that a design with �no fewer than 60 plots can achieve sufficient power given the trade-off between power and sampling effort. We also tracked the proportion of plots that burn over a 35-year period, which indicated that ~ 7% of plots will likely burn over the course of the monitoring effort. Factoring in this expected loss, we tentatively suggest targeting ~ 70 - 80 plots to maintain effective sample sizes using our current models. This estimate may be refined as the simulation work continues. Of the first three tests, the simple t-test provided the lowest rate of detection, followed by the design-aware mixed-effects modelling approach. These differences were most pronounced at the regional scale with moderate to high shifts in browsing pressure. Piecewise SEM showed the greatest promise, providing detection that approached or exceeded our target power at the regional scale (but see sensitivity analysis results below). The final test comparing temporal trends in cover demonstrated the lowest detection rate across all methods, suggesting detecting slope changes may be difficult under a simple regression framework. Detecting temporal trends may require more sophisticated time-series approaches in practice, which were not implemented here due to computational constraints. Comparisons between stratified and unstratified sampling under our equal-allocation design showed only modest differences in detection across methods. We plan to continue testing the potential benefits of stratification using a stratified sampling estimator that explicitly accounts for heterogeneity in variance to determine if they will provide additional gains in detection. These tests of detection were conducted under a common set of baseline simulation settings for all scenarios. However, key processes in the real system are uncertain, and published estimates for many relationships span a broad range. To evaluate how this uncertainty may influence our conclusions about detection probability, we conducted a sensitivity analysis on major model drivers and relationships. We varied parameters controlling the strength of soil moisture, soil nitrogen, and their interaction on productivity, fire-related biomass loss and recovery time, the frequency of allotment use and the magnitude of cattle impacts, respiration costs (i.e. the respiration constant), and sampling error for cover and browsing pressure. For each parameter, we varied across each categorical distinction (or three plausible values for continuous variables) while holding all other parameters at their baseline settings under an unstratified sample design. Sample size, the magnitude of the shift in browsing pressure following establishment, and the spatial extent of establishment were held constant across simulation runs. For each parameter setting, we recomputed detection probabilities and compared them to the baseline, allowing us to identify which assumptions most strongly influence detection and which conclusions are robust to uncertainty in model structure. The sensitivity analysis showed detection probability was most strongly influenced by sampling error in cover and browsing pressure (Figure 4). �Figure 4. Sensitivity of detection probability to simulation parameters for the design-aware mixedeffects modelling approach (a) and the piecewise SEM (b). The y-axis shows the absolute change in detection probability when simulation parameters are varied across their low, medium, and high settings relative to a common baseline configuration where all categorical parameters are set at their medium values. Variables include the strength of the relationships between productivity and soil moisture (SM_level), the strength of the relationship between productivity and soil N (N_level), the strength of the soil moisture and N interaction on productivity (SM_N_level), the amount of variance in productivity attributed to soil moisture, N, and process noise (var_level), the amount of variance attributed to the soil moisture x N interaction (int_level), the slope of the post-fire biomass accumulation trajectories (pf_level), the number of post-fire recovery years (rec_years), the effect of cattle on standing biomass (cattle_level), cattle allotment use probability (cattle_use), sample noise for shrub cover (cover_sd), sample noise for browsing intensity (sample_browse_sd), and the maintenance respiration cost constant (resp_constant). Colors indicate sampling between both communities with equal sample size (All; red), sampling only within the Mtn Shrub community (Mt_Shrub; green), and sampling only within the Sagebrush community (Sagebrush; blue). Bars show the range of |Δ % detection| values across all scenarios compared to baseline conditions, with larger values indicating greater influence of simulation parameters on detection probability. Note differences between panels in the y-axis scale. �Increases in observation error for cover substantially reduced detection across all four statistical approaches, emphasizing that accurate, precise, and repeatable field estimates of cover are critical for detecting indirect effects of wolves on vegetation. Piecewise SEM performance was particularly sensitive to sampling error in browsing pressure and the ability to accurately track changes over time; performance sharply declined when estimates of browsing pressure included substantial error. In practice, measurements of indirect effects are often evaluated using both browsing metrics and independent estimates of ungulate density and space-use. This finding underscores the importance of investing sampling effort in these measurements, and suggest that combining estimates of browsing pressure and ungulate data may help to constrain the effect of noise on detection, especially for mechanistic tests of indirect effects. Field methods development Shrub recruitment Heavy ungulate browsing can create a ‘regeneration debt’ in woody plants, whereby recruitment of new individuals is insufficient to replace mature individuals, resulting in shifts in plant community composition (Miller et al. 2023). Seedlings of bitterbrush, mountain mahogany, and serviceberry can be killed by ungulate browsing, even though mature plants are quite resilient to browsing (Shepard 1971, Paschke et al. 2003). Thus, regeneration is an important and sensitive metric with respect to browsing pressure changes. Our proposal included quantifying juvenile shrub density as a metric of regeneration. We have decided that juvenile shrub density, while interesting, is not sufficient to understand regeneration processes. This is because juvenile shrub density could decrease for either good reasons (juveniles grow to become adults) or for bad reasons (juveniles die). Probabilities for juvenile emergence, survival, and transition to the mature class would be much more informative. This requires tagged individuals. Thus, we developed methods for tagging and relocating juveniles in 2025. We placed permanently marked center points for each plot and recorded polar coordinates for each tagged individual (Figure 5). �Figure 5. 5 gallon bucket lids permanently marking plot centers allow us to precisely align a device for measuring and relocating juvenile shrubs via polar coordinates. Our proposal included ancillary work to ensure that we could consistently and reliably classify juveniles, including testing methods for aging young shrubs. We assessed our ability to age shrubs via bud scar counts, ring counts at the root crown, and root crown diameter by starting with 40 shrubs of known age harvested from decommissioned prior experiments or purchased from local nurseries. Shrubs can be aged using ring counts in a similar manner to trees, if the counts are performed at the point of germination, which is identified as the lowest part of the plant containing pith (Telewski 1993). We sectioned the shrubs in 1cm increments, sanded the sections with progressively finer sandpaper, and used dissecting scopes to identify the correct section (Cooper et al. 2003). Observers without knowledge of the shrub’s actual age then counted the rings. A regression between actual age and ring count had an R2 of 0.88. For ages over 2, ring counts tended to underestimate age, likely due to missing rings occurring during drought years. Therefore our ring count estimates should be regarded as minimum ages (Figure 6). �Figure 6. Relationship between root crown diameter and age (as estimated by ring counts at the root crown) for a lumped sample of bitterbrush (circles), mountain mahogany (triangles), sagebrush (squares), and serviceberry (crosses). Next, we harvested 34 bitterbrush, mountain mahogany, serviceberry, and sagebrush shrubs of unknown ages, counted bud scars, measured root crown diameter, and then sectioned them and counted rings. Bud scar counts proved a very unreliable estimate of age, with an R2 of only 0.13 when regressed versus ring-estimated age. Plants of ages 5, 7, 10, and 24 had only 2 bud scars, implying that shrubs often lose entire leaders. Diameter at root crown performed better, with an R2 of 0.66 when regressed versus ring-estimated age. Species were lumped for this analysis (Figure 6). We decided to use root crown diameter as our basis for classifying juveniles, and we considered that shrubs having survived at least 4 years would have passed the most vulnerable stage of higher seedling mortality. Our model estimates that shrubs with root crown diameter of 3mm have a ringestimated age of 4.2 (3.5, 4.8) years, therefore we plan to define juveniles as having a diameter of 3mm or less. As serviceberry plants produce suckers, we excavated several small serviceberry plants to develop criteria for distinguishing suckers from juveniles based on the tapering of the tap root (Figure 7). �Figure 7. Serviceberry suckers (a) can be distinguished from serviceberry juveniles (b) by whether or not the tap root tapers in diameter with depth. We ultimately created a scheme of 5 different age and size related classifications that are relevant to shrub origin, survival, and contribution to forage resources: baby, sucker, juvenile, dwarf, and mature. Table 2 describes these classifications, why they matter, and what type of monitoring we plan to perform. Some criteria may undergo further refinement. Table 2. Shrub recruitment classes to be monitored. Name Approx Age Criteria Relevance 2025 and Planned monitoring baby Less than 1 year Stem diameter less than 1mm. Easily identifiable. Starting point for recruitment Counted, but not tagged; tagging not practical due to small size and high mortality. juvenile Less than about 4 years Stem diameter less than 3mm (Cutoff of 5mm was used in 2025), tapering by at least a third within 7cm of the ground surface. If serviceberry, more than 1 m from canopy of mature serviceberry. Canopy area less than 200 cm2. Stage which may contribute to future recruitment, but vulnerable to mortality. Survival may be sensitive to browsing pressure. Counted and tagged so individual fate can be determined. sucker Less than about 4 Stem diameter less than 3mm (Cutoff of 5mm was Stage which may contribute to future recruitment. Likely to be Counted in 2025. Plan to begin tagging in 2026 so �Name 2025 and Planned monitoring Approx Age Criteria Relevance years used in 2025), NOT tapering by at least a third within 7cm of the ground surface. If serviceberry, more than 1 m from canopy of mature serviceberry. Canopy area less than 200 cm2. less vulnerable to mortality, with individual fate can be survival less sensitive to determined. browsing pressure, than juveniles. dwarf More than about 4 Stem diameter more than 3mm (Cutoff of 5mm was years used in 2025). If serviceberry, more than 1 m from canopy of mature serviceberry. Canopy area less than 200 cm2. Stage which is less vulnerable to mortality than juveniles or suckers, but which does not yet provide meaningful forage resources. Transition to mature plants may be sensitive to browsing pressure. Counted in 2025. Plan to begin tagging in 2026 so individual fate can be determined. Mature More than about 4 Canopy area more than 200 cm2. Stem diameter years unlikely to be less than 3mm. Stage which is less vulnerable to mortality than juveniles or suckers and provides meaningful contribution to forage resources and seed production. 8 randomly selected individuals per species per plot to be tagged in 2026 for mortality estimates, seed production, productivity, canopy area, and browsing pressure Plot-scale sampling strategy In spring 2025, we discussed temporal aspects of data collection and plot spacing. We considered the merits of different temporal strategies for collecting plot-scale data, such as measuring all plots every year, half the plots every other year, or a third of the plots every three years. After considering our future desire to relate plot-scale vegetation data to animal space use and density, we decided that taking data at every plot in every year is necessary. We also considered how far plots should be spaced in order for each plot to be considered independent with regards to ungulate movement, and decided that vegetation plots should be spaced by a minimum of 400m. In late summer 2025, we completed practice plots in 6 locations to try out different sampling strategies for tagging immature shrubs as well as for gathering data on mature shrubs. Our initial strategy was to measure all shrubs within a defined area, as this would allow estimates to be calculated on a perarea basis from our measurements. We assessed several plot areas/shapes (80m2 circle, 40m2 hourglass or pinwheel, 20m2 thin pinwheel). For immature shrubs, allowing a flexible plot size/shape for each species in each location was successful, allowing us to complete our surveys within about 30 minutes per plot. In some cases, �however, we tagged fewer than 10 plants per species. In 2026, we may survey for immature shrubs over a plot radius of 7m rather than 5m. For mature shrubs, flexible plot area allowed us to gather data more efficiently, but even so, each survey took 1-3 hours to complete. Choosing the most appropriate plot area/shape for each species at each location proved difficult, and we ended up with excessive numbers of measurements for certain species, particularly sagebrush. We are interested in height, seed production, productivity, canopy area, and density for mature shrubs. In 2025 we measured all parameters for all shrubs within our defined area. In 2026, we propose to randomly select 8 mature shrubs per species per plot for seed production, productivity, canopy area, and height measurements. We will tag these shrubs and collect polar coordinates to facilitate relocation over time, similar to the immature shrub survey. In this way, we will obtain data on mature plant mortality probability. This approach does not directly provide a measure of shrub density or total cover, nor does it provide productivity or seed production on a per-area basis. We will be able to get estimates of these parameters by coupling our measurements on marked plants with line-point intercept (LPI) measurements of percent cover taken at the plot scale in late summer. We will collect LPI in a crosshairs pattern centered on a permanent plot marker, allowing us to remeasure exactly the same transects each year, and providing a complete picture of the plant community, including herbaceous plants. The marked plants will allow us to calculate productivity and seed production per unit of shrub canopy area, and the LPI data will provide a measure of shrub canopy area at the plot level. We can also obtain a rough estimate of shrub density by dividing plot-level shrub area estimated by LPI by average shrub canopy area estimated from marked plants. Shrub seed production Seed production is the starting point for regeneration and essential to plant community persistence. Serviceberry, mountain mahogany, and bitterbrush produce flowers and seeds on wood that grew the prior growing season. If most or all of the current year’s growth is browsed, very little seed will be produced (Hormay 1943, Clements and Young 2001). Browsing can also reduce sagebrush seed production due to energetic constraints (Wambolt and Sherwood 1999). Thus, seed production is a metric of interest that may be sensitive to browsing pressure changes. There is no established, reproducible method for efficiently estimating bitterbrush, serviceberry, and mountain mahogany seed production. Seeds can be collected in traps or grates, but this method requires infrastructure to be left in the field and multiple site visits per year (Clements and Young 2001). For big sagebrush, inflorescence length correlates well with seed production per inflorescence (Landeen et al. 2017), but methods to estimate inflorescence density are still needed. In 2025, we began to develop seed production methods for bitterbrush, serviceberry, mountain mahogany, and sagebrush by correlating ocular estimates of seed density classes with actual numbers of seeds (for bitterbrush, serviceberry, and mountain mahogany) or numbers of inflorescences (for sagebrush). To do this, we photographed shrubs, estimated their seed density class on a scale of 0-10, and measured their canopy area. For smaller or more sparsely seeded shrubs, we then picked and counted all seeds on the shrub. For larger or more densely seeded shrubs, we randomly selected three 0.03m2 vertical columns of the shrub canopy. We then photographed each of these three subsamples, estimated the seed density class of each, and then picked and counted the seeds within them. Our naive estimates of seed �density class correlated with actual seed density by an R2 of 0.56 for sagebrush, 0.57 for mountain mahogany, and 0.70 for bitterbrush. We were unable to test our accuracy for serviceberry, due to very poor seed production in 2025. We arrayed our photographs in order of actual seed density to create training materials to improve our future ocular estimates of seed density classes. In 2026, we will continue this work by including serviceberry plants and estimating and counting samples of higher seed density for the other species. We will also determine if seed density estimates are more accurate when made at the scale of the whole plant or when a few subsamples are selected and then averaged. To do this, we will complete photographs, estimates, and counts at the scale of whole plants and at the scale of subsamples for the same shrubs. Shrub productivity Shrub productivity is similar to shrub seed production and browsing pressure in that standardized, efficient methods have not been developed. Shrub productivity is important to measure as it is known to be predictive of herbivore use. In addition, shrub productivity is a function of many site-level variables, including precipitation, soil quality, aspect, and competition. Our modeling exercises may reveal that it will be more efficient to measure shrub productivity directly, rather than model or collect data on each of the site-level factors that contribute to it (Kauffman et al. 2010). For trees, site productivity can be measured by measuring changes in trunk diameter from one year to the next. Shrubs do not offer such a handy, discrete way of measuring annual growth, as they consist of multiple stems of differing ages with the palatable current-year shoots scattered in a complex three-dimensional distribution(Rutherford 1979). The gold standard is to quantify all current-year growth by either clipping and weighing all current-year shoots, or by measuring many shoots and then applying allometric equations (Johnston et al. 2007). Both of these options are too time-consuming for our needs. In 2025 we attempted a method relating the diameters of the largest two shoot diameters per shrub to current year growth biomass. We utilized data from prior experiments to assess the method and found that it failed to predict biomass adequately (R2 of 0.18 - 0.32). In 2026 we will compile detailed shrub allometry data from additional prior experiments and use machine learning to determine the minimum number of measurements needed to assess biomass with accuracy of at least R2 0.65. In late summer, we will test the resulting method by sampling off plot and clipping all current-year biomass from vertical columns of shrubs with defined volume. Shrub browsing pressure We plan to assess browsing pressure by trained ocular estimates of the percentage of current-year biomass removed (Johnston and Anderson 2023). This method has proven informative for landscape patterns, but requires well-trained observers, and has yet to be calibrated against weighed biomass to ensure accuracy for upland deciduous shrubs. If proven reliable, this method would be preferable to alternative methods which suffer from various limitations. For instance, the percentage of browsed shoots can be an informative metric for low to moderate browsing pressures, but above ~40% biomass removal it becomes inaccurate because a single bite in a larger diameter stem might represent more biomass removed than multiple bites taken from higher on the plant (Armstrong and Macdonald 1992). Browsing history by growth form is informative, �but only reliable for species that grow taller than herbivores can reach; smaller statured plants, and those growing in poor conditions, can look stunted or clubby even in the absence of herbivory (Keigley 1998). Comparing shoot-level biomass estimates pre- versus post-browsing is accurate, but is extremely time consuming and requires marked plants(Bilyeu et al. 2007, 2008). In fall 2025, we constructed grazing cages around 25 shrubs which we will use for a clipping study in early spring 2026 to test our ability to make accurate ocular estimates. We will use two observers. Observer 1 will select a target percentage of prior (2025) growth to remove, e.g. 25%, and then attempt to clip that percentage in a pattern similar to that of an ungulate (i.e. larger and more accessible shoots clipped preferentially), retaining the clipped portion. Photographs will be taken both before and after clipping. Observer 2, kept blind from the percentage chosen by Observer 1, will then estimate the percentage removed. Next, we will clip the remaining current annual growth on the protected plant, retaining the clipped material. After drying and weighing the two clipped components, we will calculate true percentages of biomass removed and then compare these to the estimates made by Observer 2. We will likely move the grazing cages to nearby plants in order to repeat this process in fall 2026/spring 2027. If this method seems viable, we will use the photographs to create training materials to ensure accuracy over time. Browsing pressure estimates will be made in spring in order to capture prior winter browsing activity. In May 2026, we will also begin making browsing pressure estimates on tagged individuals at each of our permanent locations. Because we will be simultaneously developing our methods and tagging individuals, we likely will not be able to complete measurements at all sites. Covariate data In 2025, we made progress on how to quantify three potentially important covariates at plot scale: soil moisture, soil nutrients, and cattle use. We identified soil moisture probes that include on-board dataloggers and will be suitable to quantify plot-level soil moisture at an economical price. We purchased 60 of these with FY25 year-end funds. We also identified cost-effective soil nutrient probes and purchased these as well. Soil nutrient and moisture probes will be kept in storage until we have selected permanent sites for our long-term study. We noted cattle use by presence/absence of cattle fecal piles and tracks within study plots, and this data was used to help parameterize the sample size simulation. 2026 PLANNED ACTIVITIES In early 2026, we plan to use the results of the sample size simulation and estimated fieldwork times to make an informed decision about sample size. We will decide how to distribute plots within our study area, making use of the shrub community maps provided by the NASA DEVELOP interns. If necessary we will reduce our study scope to stay within a feasible budget, which we judge to be the amount of fieldwork 2 technicians could complete in six to seven weeks. We will begin our clipping study for browsing pressure method development and begin taking browsing pressure measurements on permanent plots in April to May. We will continue our efforts to develop seed production and productivity metrics in July. In August, we will continue setting up permanent plots and take data on shrub recruitment, seed production, productivity, and cover. �We will produce a manuscript describing our process for the sample size data simulation and including a metanalysis of how prior community effects studies have selected sample size. LITERATURE CITED Armstrong, H. M., and A. J. Macdonald. 1992. Tests of Different Methods for Measuring and Estimating Utilization Rate of Heather (Calluna-Vulgaris) by Vertebrate Herbivores. Journal of Applied Ecology 29:285–294. Bartmann, R. M. 1983. Composition and quality of mule deer diets on pinyon-juniper winter range, Colorado. Journal of Range Management 36:534–541. Bilyeu, D. M., D. J. Cooper, and N. T. Hobbs. 2007. Assessing impacts of large herbivores on shrubs: tests of scaling factors for utilization rates from shoot-level measurements. Journal of Applied Ecology 44:168–175. Bilyeu, D. M., D. J. Cooper, and N. T. Hobbs. 2008. Water tables constrain height recovery of willow on Yellowstone’s Northern Range. Ecological Applications 18:80–92. Brice, E. M., E. J. Larsen, and D. R. MacNulty. 2022. Sampling bias exaggerates a textbook example of a trophic cascade. Ecology Letters 25:177–188. Carpenter, L. H., O. C. Wallmo, and R. B. Gill. 1979. Forage diversity and dietary selection by wintering mule deer. Journal of Range Management 32:226–229. Clements, C. D., and J. A. Young. 2001. Antelope bitterbrush seed production and stand age. Journal of Range Management 54:269–273. Cooper, D. J., D. C. Andersen, and R. A. Chimner. 2003. Multiple pathways for woody plant establishment on floodplains at local to regional scales. Journal of Ecology 91:182–196. Hormay, A. L. 1943. Bitterbrush in California. Research note, U.S. Department of Agriculture, Forest Service, California Forest and Range Experiment Station, Berkeley, CA. Johnston, D. B., and C. R. Anderson. 2023. Plant and mule deer responses to pinyon‐juniper removal by three mechanical methods. Wildlife Society Bulletin 47:e1421. Johnston, D. B., D. J. Cooper, and N. T. Hobbs. 2007. Elk browsing increases aboveground growth of water-stressed willows by modifying plant architecture. Oecologia 154:467–478. Kauffman, M. J., J. F. Brodie, and E. S. Jules. 2010. Are wolves saving Yellowstone’s aspen? A landscape-level test of a behaviorally mediated trophic cascade. Ecology 91:2742–2755. Keigley, R. B. 1998. Browse evaluation By Analysis of Growth Form: Volume I Methods For Evaluating Condition and Trend. Montana Fish Wildlife and Parks, Bozeman, MT. Kohl, M. T., D. R. Stahler, M. C. Metz, J. D. Forester, M. J. Kauffman, N. Varley, P. J. White, D. W. Smith, and D. R. MacNulty. 2018. Diel predator activity drives a dynamic landscape of fear. Ecological Monographs 88:638–652. �Landeen, M. L., L. Allphin, S. G. Kitchen, and S. L. Petersen. 2017. Seed Production Estimation for Mountain Big Sagebrush (Artemisia tridentata ssp vaseyana). Rangeland Ecology & Management 70:633–637. Liley, S., and S. Creel. 2008. What best explains vigilance in elk: characteristics of prey, predators, or the environment? Behavioral Ecology 19:245–254. Mao, J. S., M. S. Boyce, D. W. Smith, F. J. Singer, D. J. Vales, J. M. Vore, and E. H. Merrill. 2005. Habitat selection by elk before and after wolf reintroduction in Yellowstone National Park. Journal of Wildlife Management 69:1691–1707. Miller, K. M., S. J. Perles, J. P. Schmit, E. R. Matthews, A. S. Weed, J. A. Comiskey, M. R. Marshall, P. Nelson, and N. A. Fisichelli. 2023. Overabundant deer and invasive plants drive widespread regeneration debt in eastern United States national parks. Ecological Applications 33:e2837. Paine, R. T. 1980. Food webs- linkage, interaction strength and community infrastructure- the 3rd Tansley lecture. Journal of Animal Ecology 49:667–685. Paschke, M. W., E. F. Redente, and S. L. Brown. 2003. Biology and establishment of mountain shrubs on mining disturbances in the Rocky Mountains, USA. Land Degradation & Development 14:459– 480. Richens, V. B. 1967. Characteristics of mule deer herds and their range in northeastern Utah. Journal of Wildlife Management 31:651-+. Rutherford, M. C. 1979. Plant-Based Techniques for Determining Available Browse and Browse Utilization: A Review. Botanical Review 45:203–228. Schmitz, O. J., V. Krivan, and O. Ovadia. 2004. Trophic cascades: the primacy of trait-mediated indirect interactions. Ecology Letters 7:153–163. Shepard, H. R. 1971. Effects of clipping on key browse species in Southwestern Colorado, Technical Publication Number 28. L. E. Yeager, editor. Game Range Investigations, Colorado Division of Game, Fish, and Parks, Denver, Co. Tarbox, B. C., N. J. Van Lanen, A. P. Monroe, and C. L. Aldridge. 2025. Tiered spatial conservation prioritizations for sagebrush ecosystems in northwest Colorado. U.S. Geological Survey data release. Telewski, F. W. 1993. Determining the germination date of woody plants: a proposed method for locating the root/shoot interface. Tree-ring bulletin 53. Updike, D. R., E. R. Loft, and F. A. Hall. 1992. Wildfires on big sagebrush/antelope bitterbrush range in northeastern California: implications for deer populations. Biological Conservation 59:277. Wambolt, C. L., and H. W. Sherwood. 1999. Sagebrush response to ungulate browsing in Yellowstone. Journal of Range Management 52:363–369. � https://cpw.cvlcollections.org/files/original/3b0bfe015cc442396464ded8e7f65d3d.pdf c32596c76286462f065a41239d4e5831 https://cpw.cvlcollections.org/files/original/9a8a89746e0db4c8ab77eb1d3567b389.pdf 19a4b451b914cb6c0d70b70e7c0e8107 PDF Text Text Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Spatiotemporal effects of human recreation on elk behavior: an assessment within critical time stages Period Covered: July 1, 2019-June 30, 2020 Principal Investigators: Nathaniel Rayl, nathaniel.rayl@state.co.us; Eric Bergman, eric.bergman@state.co.us; Joe Holbrook, Joe.Holbrook@uwyo.edu All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. The influence of recreational disturbance on ungulate populations is of particular interest to wildlife managers in Colorado, as there is growing concern about its potential impacts within the state. Currently, the western United States is experiencing some of the highest rates of human population growth in the country, with growth in rural and exurban areas frequently outpacing growth in urban areas. Additionally, participation in outdoor recreation is also increasing. In Colorado, the number of individuals participating in recreational activities, and the associated demand for recreational opportunities, appear to be increasing. Understanding potential impacts of recreational activity on elk spatial ecology in Colorado is critical for guiding management actions, as altered movements may result in reduced foraging time and higher energetic costs, which may decrease fitness. We are studying elk from the resident portion of the Bear’s Ears elk herd (DAU E-2) in Colorado to determine potential impacts of recreational activities on this population (Fig. 1). This research project is a collaboration between Colorado Parks and Wildlife (CPW) and the Haub School of Environment and Natural Resources at the University of Wyoming, and will form the basis of an M.S. thesis for a graduate student enrolled at the Haub School. In January 2020, we collared 30 adult female elk from the resident portion of the Bear's Ears elk herd on U.S. Forest Service (USFS) land near Steamboat Springs. The estimated pregnancy rate was 93% (95% CI: 79-98%). This spring, summer, and fall we will be deploying trail counters and cameras at trailheads in the study area, and handing out GPS units to recreationists to quantify human recreation on the landscape and evaluate how elk respond to recreationists. 25 �DENVER • Colorado 5 10 Figure 1. Routt National Forest study area located in northwest Colorado, USA. 26 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Spatiotemporal effects of human recreation on elk behavior: an assessment within critical time stages Period Covered: July 1, 2020 − June 30, 2021 Principal Investigators: Nathaniel Rayl, nathaniel.rayl@state.co.us; Eric Bergman, eric.bergman@state.co.us; Joe Holbrook, Joe.Holbrook@uwyo.edu All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. The influence of recreational disturbance on ungulate populations is of particular interest to wildlife managers in Colorado, as there is growing concern about its potential impacts within the state. Currently, the western United States is experiencing some of the highest rates of human population growth in the country, with growth in rural and exurban areas frequently outpacing growth in urban areas. Additionally, participation in outdoor recreation is also increasing. In Colorado, the number of individuals participating in recreational activities, and the associated demand for recreational opportunities, appear to be increasing. Understanding potential impacts of recreational activity on elk spatial ecology in Colorado is critical for guiding management actions, as altered movements may result in reduced foraging time and higher energetic costs, which may decrease fitness. We are studying elk from the resident portion of the Bear’s Ears elk herd (DAU E-2) in Colorado to determine potential impacts of recreational activities on this population. This research project is a collaboration between Colorado Parks and Wildlife (CPW) and the Haub School of Environment and Natural Resources at the University of Wyoming, and forms the basis of an M.S. thesis for a graduate student enrolled at the Haub School. In January 2020 and January 2021, we collared 30 and 26 adult female elk, respectively, from the resident portion of the Bear's Ears elk herd on U.S. Forest Service (USFS) land near Steamboat Springs. In both years, the estimated pregnancy rate was 93% (95% CI: 79-98%). From May-October 2020 we deployed trail counters at 22 trailheads in the Routt National Forest (Figure 1). We recorded roughly 100,000 people departing and returning from these trailheads. Among individual trailheads, we documented average daily traffic counts ranging from 2-325 people (Figure 2). Most traffic was recorded on weekends with noticeable lulls in traffic frequency observed during weekdays. During the 2021 field season, we again deployed trail counters at the 22 trailheads, and also added additional trail counters at 1-km intervals along each trail for up to 5-km from the trailhead. These additional trail counters are being deployed on a rotating basis to sample each trail. Data collected from these additional trail counters will provide an estimate of the decay of traffic along trails. During the 2020 and 2021 field season, we distributed handheld GPSs to recreationists (hikers, bikers, hunters) to record detailed tracks of human use within this trail system (Figure 3). In 2020, we collected over 100 GPS tracks. GPS tracks from recreationists and hunters will allow us to better quantify human recreation on the landscape and evaluate how elk respond to recreationists. 32 �Figure 1. Routt National Forest study area located in northwest Colorado, USA. 33 �Figure 2. Daily trends in trailhead traffic documented with trail counters from June through October 2020, excluding Fish Creek Falls, Mad Creek, and Red Dirt trailheads, which received average daily counts >200. Figure 3. GPS track (blue) recorded from recreational mountain biker on trail system (white) in August 2020. Note the off-trail use near Long Lake. 34 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Spatiotemporal effects of human recreation on elk behavior: an assessment within critical time stages Period Covered: January 1, 2022 – December 31, 2022 Principal Investigators: Nathaniel Rayl, nathaniel.rayl@state.co.us; Eric Bergman, eric.bergman@state.co.us; Joe Holbrook, Joe.Holbrook@uwyo.edu All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. The influence of recreational disturbance on ungulate populations is of particular interest to wildlife managers in Colorado, as there is growing concern about its potential impacts within the state. Currently, the western United States is experiencing some of the highest rates of human population growth in the country, with growth in rural and exurban areas frequently outpacing growth in urban areas. Additionally, participation in outdoor recreation is also increasing. In Colorado, the number of individuals participating in recreational activities, and the associated demand for recreational opportunities, appear to be increasing. Understanding potential impacts of recreational activity on elk spatial ecology in Colorado is critical for guiding management actions, as altered movements may result in reduced foraging time and higher energetic costs, which may decrease fitness. We are studying elk from the resident portion (i.e., non-migratory) of the Bear’s Ears elk herd (DAU E-2) in Colorado to determine potential impacts of recreational activities on this population. This research project is a collaboration between Colorado Parks and Wildlife (CPW) and the Haub School of Environment and Natural Resources at the University of Wyoming, and forms the basis of an M.S. thesis for a graduate student enrolled at the Haub School. In January 2020 and January 2021, we collared 30 and 26 adult female elk, respectively, from the resident portion of the Bear's Ears elk herd on U.S. Forest Service (USFS) land near Steamboat Springs. We estimated pregnancy rates of 93% (95% CI: 79-98%) in 2020 and 96% (95% CI: 81-100%) in 2021. From May-October 2020 we deployed trail counters at 22 trailheads in the Routt National Forest (Figure 1). We recorded roughly 100,000 people departing and returning from these trailheads. Among individual trailheads, we documented average daily traffic counts ranging from 2-325 people (Figure 2). Most traffic was recorded on weekends with noticeable lulls in traffic frequency observed during weekdays. During the 2021 field season, we again deployed trail counters at the 22 trailheads, and also added additional trail counters at 1-km intervals along each trail for up to 5-km from the trailhead. These additional trail counters are being deployed on a rotating basis to sample each trail. Data collected from these additional trail counters will provide an estimate of the decay of traffic along trails. During the 2020 and 2021 field season, we distributed handheld GPSs to recreationists (hikers, bikers, hunters) to record detailed tracks of human use within this trail system (Figure 3). In 2020, we collected over 100 GPS tracks. These tracks from recreationists and hunters will allow us to better quantify human recreation on the landscape and evaluate how elk respond to recreationists. 33 �Figure 1. Routt National Forest study area located in northwest Colorado, USA. 34 �Figure 2. Daily trends in trailhead traffic documented with trail counters from June through October 2020, excluding Fish Creek Falls, Mad Creek, and Red Dirt trailheads, which received average daily counts >200. Figure 3. GPS track (blue) recorded from recreational mountain biker on trail system (white) in August 2020. Note the off-trail use near Long Lake. 35 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Spatiotemporal effects of human recreation on elk behavior: an assessment within critical time stages Period Covered: January 1, 2023-December 31, 2023 Principal Investigators: Nathaniel Rayl, nathaniel.rayl@state.co.us; Eric Bergman, eric.bergman@state.co.us; Joe Holbrook, Joe.Holbrook@uwyo.edu All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. The influence of recreational disturbance on ungulate populations is of particular interest to wildlife managers in Colorado, as there is growing concern about its potential impacts within the state. Currently, the western United States is experiencing some of the highest rates of human population growth in the country, with growth in rural and exurban areas frequently outpacing growth in urban areas. Additionally, participation in outdoor recreation is also increasing. In Colorado, the number of individuals participating in recreational activities, and the associated demand for recreational opportunities, appear to be increasing. Understanding potential impacts of recreational activity on elk spatial ecology in Colorado is critical for guiding management actions, as altered movements may result in reduced foraging time and higher energetic costs, which may decrease fitness. We are studying elk from the resident portion of the Bear’s Ears elk herd (DAU E-2) in Colorado to determine potential impacts of recreational activities on this population. This research project is a collaboration between Colorado Parks and Wildlife (CPW) and the Haub School of Environment and Natural Resources at the University of Wyoming, and forms the basis of an M.S. thesis for a graduate student (Eric VanNatta, also CPW Area 10 Terrestrial Biologist) enrolled at the Haub School. In January 2020 and January 2021, we collared 30 and 26 adult female elk, respectively, from the resident portion of the Bear's Ears elk herd on U.S. Forest Service (USFS) land near Steamboat Springs. We estimated pregnancy rates of 93% (95% CI: 79-98%) in 2020 and 96% (95% CI: 81-100%) in 2021. From May-October 2020 we deployed trail counters at 22 trailheads in the Routt National Forest (Fig. 1). We recorded roughly 100,000 people departing and returning from these trailheads. Among individual trailheads, we documented average daily traffic counts ranging from 2-325 people (Fig. 2). Most traffic was recorded on weekends with noticeable lulls in traffic frequency observed during weekdays. During the 2021 field season, we again deployed trail counters at the 22 trailheads, and also added additional trail counters at 1-km intervals along each trail for up to 5-km from the trailhead. These additional trail counters are being deployed on a rotating basis to sample each trail. Data collected from these additional trail counters will provide an estimate of the decay of traffic along trails. During the 2020 and 2021 field season, we distributed handheld GPSs to recreationists (hikers, bikers, hunters) to record detailed tracks of human use within this trail system (Fig. 3). In 2020, we collected over 100 GPS tracks. These tracks from recreationists and hunters will allow us to better quantify human recreation on the landscape and evaluate how elk respond to recreationists. In fall 2023, Eric VanNatta successfully completed and defended his M.S. proposal at the University of Wyoming and finished processing and cleaning the trail counter dataset. 22 �Figure 1. Routt National Forest study area located in northwest Colorado, USA. 23 �Figure 2. Daily trends in trailhead traffic documented with trail counters from June through October 2020, excluding Fish Creek Falls, Mad Creek, and Red Dirt trailheads, which received average daily counts >200. Figure 3. GPS track (blue) recorded from recreational mountain biker on trail system (white) in August 2020. Note the off-trail use near Long Lake. 24 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Spatiotemporal effects of human recreation on elk behavior: an assessment within critical time stages Period Covered: January 1, 2024-December 31, 2024 Principal Investigators: Nathaniel Rayl, nathaniel.rayl@state.co.us; Eric Bergman, eric.bergman@state.co.us; Joe Holbrook, Joe.Holbrook@uwyo.edu All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. The influence of recreational disturbance on ungulate populations is of particular interest to wildlife managers in Colorado, as there is growing concern about its potential impacts within the state. Currently, the western United States is experiencing some of the highest rates of human population growth in the country, with growth in rural and exurban areas frequently outpacing growth in urban areas. Additionally, participation in outdoor recreation is also increasing. In Colorado, the number of individuals participating in recreational activities, and the associated demand for recreational opportunities, appear to be increasing. Understanding potential impacts of recreational activity on elk spatial ecology in Colorado is critical for guiding management actions, as altered movements may result in reduced foraging time and higher energetic costs, which may decrease fitness. We are studying elk from the resident portion of the Bear’s Ears elk herd (DAU E-2) in Colorado to determine potential impacts of recreational activities on this population. This research project is a collaboration between Colorado Parks and Wildlife (CPW) and the Haub School of Environment and Natural Resources at the University of Wyoming, and forms the basis of an M.S. thesis for a graduate student (Eric VanNatta, also CPW Area 10 Terrestrial Biologist) enrolled at the Haub School. In January 2020 and January 2021, we collared 30 and 26 adult female elk, respectively, from the resident portion of the Bear's Ears elk herd on U.S. Forest Service (USFS) land near Steamboat Springs. We estimated pregnancy rates of 93% (95% CI: 79-98%) in 2020 and 96% (95% CI: 81-100%) in 2021. From May-October 2020 we deployed trail counters at 22 trailheads in the Routt National Forest (Figure 1). We recorded roughly 100,000 people departing and returning from these trailheads. Among individual trailheads, we documented average daily traffic counts ranging from 2-325 people (Figure 2). Most traffic was recorded on weekends with noticeable lulls in traffic frequency observed during weekdays. During the 2021 field season, we again deployed trail counters at the 22 trailheads, and also added additional trail counters at 1-km intervals along each trail for up to 5-km from the trailhead. These additional trail counters are being deployed on a rotating basis to sample each trail. Data collected from these additional trail counters will provide an estimate of the decay of traffic along trails. During the 2020 and 2021 field season, we distributed handheld GPSs to recreationists (hikers, bikers, hunters) to record detailed tracks of human use within this trail system (Figure 3). In 2020, we collected over 100 GPS tracks. These tracks from recreationists and hunters will allow us to better quantify human recreation on the landscape and evaluate how elk respond to recreationists. In fall 2023, Eric VanNatta successfully completed and defended his M.S. proposal at the University of Wyoming and finished processing and cleaning the trail counter dataset. In 2024, Eric worked on analyses for the first chapter of his thesis, which should be completed in early 2025. 22 �Figure 1. Routt National Forest study area located in northwest Colorado, USA. 23 �Figure 2. Daily trends in trailhead traffic documented with trail counters from June through October 2020, excluding Fish Creek Falls, Mad Creek, and Red Dirt trailheads, which received average daily counts >200. Figure 3. GPS track (blue) recorded from recreational mountain biker on trail system (white) in August 2020. Note the off-trail use near Long Lake. 24 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Spatiotemporal effects of human recreation on elk behavior: an assessment within critical time stages Period Covered: January 1, 2025-December 31, 2025 Principal Investigators: Nathaniel Rayl, nathaniel.rayl@state.co.us; Eric Bergman, eric.bergman@state.co.us; Joe Holbrook, Joe.Holbrook@uwyo.edu All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. By providing this summary, CPW does not intend to waive its rights under the Colorado Open Records Act, including CPW’s right to maintain the confidentiality of ongoing research projects. CRS § 24-72-204. The influence of recreational disturbance on ungulate populations is of particular interest to wildlife managers in Colorado, as there is growing concern about its potential impact within the state. Currently, the western United States is experiencing some of the highest rates of human population growth in the country, with growth in rural and exurban areas frequently outpacing growth in urban areas. Additionally, participation in outdoor recreation is also increasing. In Colorado, the number of individuals participating in recreational activities, and the associated demand for recreational opportunities, appear to be increasing. Understanding potential impacts of recreation on elk spatial ecology in Colorado is critical for guiding management actions, as altered movements may result in reduced foraging time and higher energetic costs, which may decrease fitness. We studied elk from the resident portion of the Bear’s Ears elk herd (DAU E-2) in Colorado to determine potential impacts of recreational activities on this population. This research project was a collaboration between Colorado Parks and Wildlife (CPW) and the Haub School of Environment and Natural Resources at the University of Wyoming and formed the basis of an M.S. thesis for a graduate student (Eric Van Natta, also CPW Area 10 Terrestrial Biologist) enrolled at the Haub School. In January 2020 and January 2021, we collared 30 and 26 adult female elk, respectively, from the resident portion of the Bear's Ears elk herd on U.S. Forest Service (USFS) land near Steamboat Springs. We estimated pregnancy rates of 93% (95% CI: 79-98%) in 2020 and 96% (95% CI: 81-100%) in 2021. From May-October 2020 we deployed trail counters at 22 trailheads in the Routt National Forest (Figure 1). We recorded roughly 100,000 people departing and returning from these trailheads. Among individual trailheads, we documented average daily traffic counts ranging from 2-325 people (Figure 2). Most traffic was recorded on weekends with noticeable lulls in traffic frequency observed during weekdays. During the 2021 field season, we again deployed trail counters at the 22 trailheads and also added additional trail counters at 1-km intervals along each trail for up to 5-km from the trailhead. These additional trail counters were deployed on a rotating basis to sample each trail and provide an estimate of the decay of traffic along trails. During the 2020 and 2021 field season, we distributed handheld GPSs to recreationists (hikers, bikers, hunters) to record detailed tracks of human use within this trail system (Figure 3). In 2020, we collected over 100 GPS tracks. These tracks from recreationists and hunters will allow us to better quantify human recreation on the landscape and evaluate how elk respond to recreationists. In fall 2025, Eric Van Natta successfully defended his M.S. thesis at the University of Wyoming. We are currently preparing his two thesis chapters for publication. 24 �Figure 1. Routt National Forest study area located in northwest Colorado, USA. 25 �Figure 2. Daily trends in trailhead traffic documented with trail counters from June through October 2020, excluding Fish Creek Falls, Mad Creek, and Red Dirt trailheads, which received average daily counts >200. Figure 3. GPS track (blue) recorded from recreational mountain biker on trail system (white) in August 2020. Note the off-trail use near Long Lake. 26 � Dublin Core The Dublin Core metadata element set is common to all Omeka records, including items, files, and collections. For more information see, http://dublincore.org/documents/dces/. Title A name given to the resource Mammals research pdfs Description An account of the resource Pdfs of mammals research projects by title . Pdfs are linked from this page. Includes federal aid progress reports and non federal aid progress reports. If reports cover more than one year they are combined into one pdf. Dublin Core The Dublin Core metadata element set is common to all Omeka records, including items, files, and collections. For more information see, http://dublincore.org/documents/dces/. Title A name given to the resource Mammals Research Projects pdfs (2021-current) Rights Information about rights held in and over the resource <a href="http://rightsstatements.org/vocab/NoC-NC/1.0/">No Copyright - Non-Commercial Use Only</a> https://cpw.cvlcollections.org/files/original/e2e32a7cea8fa0563e4b41436c52e943.pdf 267eda9c65bdfa9cc30b3a360b88df1b PDF Text Text PROGRAi~ PROGRESS REPORT COLORADO Project No. Project Type:~__R_e_s_e_a_r_c_h________ SE - 3 - 2 Job No. 2 Work Plan No. __I..._I=I~-------Job Title: Endangered Wildlife Investigations: Lynx and Wolverine Verifcation Period Covered: Personnel: October 1, 1978 to April 30, 1979 Steven J. Bissel, John R. Torres, James C. Halfpenny, David Nead. ABSTRACT Over 3,000 posters soliciting information were distributed to District Wildlife Hanagers, taxidermists, trappers, outfitters, libraries, universities, federal and state agencies. Museums were visited and individuals were interviewed to obtain historical background. Distributional maps were produced to evaluate the best possible areas for locating both species. Field work was conducted in four areas fo~ · wolverine and two areas for lynx. Over 400 miles of trail were searched for tracks and signs. Trained dogs were used to verify "cat" tracks. Baited hair snags were set to obtain hair samples. One wolverine shot near the Colorado-Utah border was processed during the project. J· ~ Two lynx were tracked in the central portion of Colorado. Compilation of reports and sightings indicate that both the wolverine and the lynx probably still have breeding populations in Colorado. further work is needed to verify wolverine existence. However, The known population of Coloradoan lynx should be monitored to determine their range and evaluate habitat. i 1iliilOOiWiiOO~ BDOW021592 �TABLE OF CONTENTS Page Program Segment Narrative • 1 Methods and Materials 1 Results 2 Discussion 3 .... . . Literature Cited ·~ ii 7 �-~3 r LYNX AND WOLVERINE VERIFICATION \~ PROGRAM SEGMENT NARRATIVE Due to the lack of data, it is impossible to determine the current status of either the lynx (Lynx canadensis)or the wolverine (Gulo gulo) in Colorado. Only two confirmed wolverine reports have been made in recent years while confirmed lynx reports number but ten (Bissell, 1978a and 1978 b; Halfpenny et. al., 1979 and in press; T.errill, 1970). Little is known about their natural history in Colorado and essential habitat has not~been determined (Bissel, 1978a). Armstrong (1972), Lechleitner (1969), and Warren (1942) list historical records of both species. No published ecological studies exist for either mammal in the continental United States. European literature indicates that the diet of the wolverine consists mainly of cervid species with hare (Lepus) and ptarmigan (Alopex) often of minor importance. of the lynx consists largely of hares. The diet The wolverine primarily subsists as a scavenger while the lynx is an active predator. Both animals, under the right conditions, can kill large ungulates. Both mammals are mainly boreal in distribution, being found at or above timberline in dense conifer forests and rocky areas. Denning areas have not been investigated in the States. The purpose of this study was to verify wolverine and lynx populations and determine their range within Colorado. METHODS AND MATERIALS Reports of wolverine and lynx were solicited from the public by distributing over 3,000 posters to District Wildlife Manager, taxidermists, trappers, outfitters, libraries, universities, federal and state agencies and using radio, television, and newspaper stories. Based on the descriptions given, reports were identified as to species of mammal observed and the quality of the report was then determined. Reliable reports were used to determine main geographic areas of interest. Museums were visited and interviews conducted to obtain �2 historical information to compare to the reports that were received. Based on the quality and quantity of reports, four main areas possibly supporting wolverine populations were selected for field work. In these areas, baited hair snags (cylinders made out of hardware cloth and barbed wire; Hummel, 1978) were placed. Trails were searched by snowmobiles and crosscountry skis to locate mammal tracks. Two areas were selected for lynx verifications work. and skis were again used to search trails for tracks. and houndsman was retained. Snowmobiles A local trapper His trained dogs were used to verify "cat" tracks. RESULTS From February 1, 1979 to April 30, 1979, project personnel, District Wildlife Managers, and volunteers traveled over 2500 miles on 420 miles of trail. During this time two lynx were trailed on the Frying Pan River. These tracks represent the first reports of lynx since 1974. Six recent lynx reports and three older reports were obtained from interviews. Two previously unreported mounted lynx specimens were observed. One of these animals, trapped on the Vail ski slopes, is now in the law enforcement office of the Division. The other, which was trapped near Guanella Pass, resides in the Jefferson County Outdoor School collection. A third lynx (reported by Terrill, 1971) is in the private collection of Pete Sherwood. The three lynx were harvested in 1974, 1972, and 1969 respectively. On October 25, 1978, project personnel monitored the release by private individuals of a male and a female wolverine at a site about 15 miles due south of Aspen. These animals, trapped in Canada about three months before the release, were part of a movie being filmed by Stouffer Enterprises and John Denver. No reports have been received by the Division office indicating that these animals have been observed. Early in the project, pieces of a wolverine skull found in the San Juan Mountains of Colorado, were sent to the Division. It was not possible to determine how long the animal had been dead but a reasonable guess maybe between 10 and 25 years. �J Both tracking and hair snags failed to provide evidence of wolverine. Although hair was obtained from 30% of the hair snags, no wolverine hair was found. Aproximately 60 reports may have actually observed wolverine. Of the 60 reports, 25 were rated as probable sightings and 35 as possible sightings, a lower quality category. On March 15, 1979, a male wolverine in reproductive condition was shot near the Colorado-Utah border. Due.to legal hassles revolving around the exact location of the death of the animal, the individual has retained the animal and will not allow it to be used for scientific investigations. The animal however must have had at least part of its range in Colorado, if not all of its range. Data from reports, museums and interviews were used to construct distributional maps for Colorado (fig-. 1 and 2). Results of this study were presented at the Colorado - Wyoming Academy of Science. Lynx distributional information has been incorporated into a paper to be submitted to the Journal of Marnrnalogy. Bibliographic material obtained through the study will be published as part of the Bibliography~ Mustelids series. Finally, Mammalian Species has agreed to publish our Wolverine Mammalian Species which will be completed during the coming year. DISCUSSION Wolverine verification: The wolverine killed in northwestern Colorado and several very credible reports indicate that a small population exists in Colorado. Evaluation of sighting reports indicate populations may be highest south and southwest of the Flat Tops in Garfield County and on the west, south, and east sides of North Park in Jackson County. These two areas should be intensively worked during another field season. lines should be set.up and checked weekly. Hair snag Frequent runs should be made on snow covered trails to check for tracks. �4 Lynx verification: Tracks and records indicate a small population of lynx in Pitkin, Eagle, and Lake Counties. Further work will undoubtably expand the known range of this population. It is highly probable that work in other remote areas of similar habitat will disclose additional populations. The Colorado lynx population represents the existance of lynx at their extreme limits.Knowledge of this population important. is therefore very Our field work indicates that· lynx are ..."rare" in Colorado, and more knowl~dge is needed before their status can be evaluated. �Colorado Wolverine Reports State map shown is overlayed with a 9,000 foot contour.. * 41 - r Skull of a Wolverine received from Mr. Al Williams Delta, Colorado. e --- Wolverine reports received prior to verification program October 1, 1978. • ---Wolverine reports received • October 1, 1978 to April 30, 1979. Figur~ ( 1. Colorado Wolverine Reports. ( ( �• *-- C0 L 0 RAD0 L Y NX REC0 RDS Recent specimens .._ -- Tr~k records obtained by the Lynx Verification Program 1978-79 ft -- Colorado Trapper's Survey 1975 • -- Literature records from Allen (1874), Armstrong (1972), Cary (1911), Seton (1929), Warren (1906, 1942), and Young (1958). The 9,000 foot contour is shown. ------~---------------------------SOUTHERN LIMITS OF LYNX DISTRIBUTION Extralirnital records indicated by triangles. Figure 2. ( Colorado Lynx Records and the Southern Limits of Lynx Distribution. c ( �7 Literature Cited Armstrong, D. M. 1972. Distribution of Mammals in Colorado. Univ. Mus. Nat. Hist., 3:x + 1 - 415. Monogr., Bissell, s. J. 1978:1.. Pp. 76,77,80, and 81 in Torres, J. et al. 1978. Essential Habitat for Threatened or Endangered Wildlife in Colorado. Bissell, S. J. 1978b. Bissell, s. J. (ed.). 1978. Colorado Mammal Distribution Latilong Study. Colo. Div. Wild!. DOW-R-D-10 ii + 20 pp. Halfpenny, J. C., D. Nead, and S. J. Bissell. 1979. Colorado Wolverine-Lynx Verification Program • . J. Colo.-Wyo. Acad. Sci., 11(1) :89. Abst • .. Halfpenny, J. c., S. J. Bissell, and D. Nead. 1979. The southern limits of lynx (Lynx canadensis) distribution in North America with special reference to Colorado. To be submitted to J. Wildl. Man. in the fall of 1979. Hummel, J. E. 1978. Furbearer study, Sierra County, California. 1977-78. The Resource Agency, State of Calif. 21 pp. Fed. Project Report W-54-R-10. Lechleitner, R. R. Wild mammals of Colorado: their appearance, habits, distirbution, and abundance. Pruett Publishing Co., Boulder, xiv + 254 pp. Terrell, B. 1971. Lynx. Colo. Outdoors, 20(5) :19. Warren, E. R. 1942. The Mammals of Colorado. Colo. College Puhl., Gen. Ser. 19 (Sci. Ser., 46): 225-274. � Dublin Core The Dublin Core metadata element set is common to all Omeka records, including items, files, and collections. For more information see, http://dublincore.org/documents/dces/. Title A name given to the resource Mammals research pdfs Description An account of the resource Pdfs of mammals research projects by title . Pdfs are linked from this page. Includes federal aid progress reports and non federal aid progress reports. If reports cover more than one year they are combined into one pdf. Dublin Core The Dublin Core metadata element set is common to all Omeka records, including items, files, and collections. For more information see, http://dublincore.org/documents/dces/. Title A name given to the resource Mammals Research Projects pdfs (1971-1980) Rights Information about rights held in and over the resource <a href="http://rightsstatements.org/vocab/NoC-NC/1.0/">No Copyright - Non-Commercial Use Only</a> Dublin Core The Dublin Core metadata element set is common to all Omeka records, including items, files, and collections. For more information see, http://dublincore.org/documents/dces/. Title A name given to the resource Mammals research pdfs Description An account of the resource Pdfs of mammals research projects by title . Pdfs are linked from this page. Includes federal aid progress reports and non federal aid progress reports. If reports cover more than one year they are combined into one pdf. Dublin Core The Dublin Core metadata element set is common to all Omeka records, including items, files, and collections. For more information see, http://dublincore.org/documents/dces/. Title A name given to the resource Mammals Research Projects pdfs (1939-1950) Rights Information about rights held in and over the resource <a href="http://rightsstatements.org/vocab/NoC-NC/1.0/">No Copyright - Non-Commercial Use Only</a> https://cpw.cvlcollections.org/files/original/4d39ef53e50c14bab61789a8af09d358.pdf a73978f7d9a8fa37f23d1fbf5c972867 PDF Text Text Colorado Division of Wildlife Wildlife Research Report •July 2001 and July 2002 JOB PROGRESS REPORT State of Colorado Division of Wildlife-Mammals Research Work Package No. -~0_6~6_2_ _ _ _ _ __ Preble's Meadow Jumping Mouse Conservation Task No. Effects of Resource Addition on Preble's Meadow Jumping Mouse (Zapus hudsonius preblei) Movement Patterns 2 Period Covered: July I, 2000- June 30, 2002 Author: Anne Trainor Personnel: T. M. Shenk, G. C. White, K. Wilson Interim Report - Preliminary Results This work continues, and precise analysis of data has yet to be accomplished. Manipulation or interpretation of these data beyond that contained in this report should be labeled as such and is discouraged. ABSTRACT Preble's meadow jumping mouse (Zapus hudsonius preblei; PMJM) is federally listed as threatened under the Endangered Species Act (ESA). Habitat conservation plans (HCPs) as defined in Section IO of the ESA, allow for 'take' of species and their habitat on private property. HCPs attempt to minimize take and provide for mitigation. Collection of reliable information and an increased understanding of PMJM habitat requirements are essential for the development of effective mitigation strategies for this species. Thus, our objectives are to (I) determine how the presence ofresource additions influences the distribution of individual PMJM within a population, and (2) to quantify and compare microhabitat characteristics among areas PMJM used heavily to areas of no use. A manipulation experiment will be conducted in sections of riparian habitat and adjacent grasslands in Douglas County, Colorado in 2002 and 2003. ��3 Research Prospectus Effects of Resource Addition on Preble's Meadow Jumping Mouse (Zapus hudsonius preble1) Movement Patterns Anne Trainor, Tanya Shenk, and Kenneth Wilson Problem: The U.S. Fish and Wildlife Service (USFWS) listed the Preble's meadow jumping mouse (Zapus hudsonius preb/ei; PMJM) as a threatened species in 1998 under the Endangered Species Act (USFWS 1999). Upon listing, little was known about the biology and habitat requirements of this subspecies within its range along the Front Range of Colorado and southeastern Wyoming. Since listing, a number of projects (e.g., long-term monitoring, surveying, and movement studies) have collected valuable information throughout Colorado (Schorr 2001, Meaney 2000, Shenk and Sivert 1999). However, information on specific habitat requirements and their relationship to the distribution, density, survival and reproduction of PMJM is still lacking. The threatened status of PMJM requires management decisions be made despite our limited knowledge. In particular, the species and its habitat are subject to habitat conservation plans (HCPs). HCPs are written for endangered and threatened species to compensate for authorized "take" with mitigation practices (Bingham and Noon 1998). HCPs require the use of the "best available" science to determine the biological needs of target species (Harding et al. 2001). Collection of reliable information for the species will improve the mitigation practices developed for HCPs. Well-designed habitat manipulation experiments provide the strongest inference to determine cause and effect relationships. Understanding of the species habitat requirements will enable the development of effective mitigation strategies. A manipulation experiment will be conducted in Douglas County, Colorado (Columbine Open Space) during 2002 and 2003 to advance our understanding of PMJM habitat requirements. We will manipulate sections of the riparian habitat and adjacent grassland within the 100-year flood plain. The site will be manipulated by adding patches (3 m x 2.43 m) of artificial resources (food and cover). Time limitations (2 field seasons are inadequate for vegetation to establish) and funding (cost of planting and sustaining vegetation) will restrict this manipulation experiment to simulating habitat with temporary structures and food supplementation. The treatments will be placed in areas of low use based on past monitoring studies conducted by the Colorado Division of Wildlife (CDOW) during 1998-2000 within 60 m of East Plum Creek. PMJM will be radio tracked before and after the manipulation to determine if PMJM locations can be altered through the addition of resources. Research Objectives: We propose two primary objectives: 1) determine how the presence ofresource additions influences the distribution of individual PMJM within a population, and 2) quantify habitat characteristics of PMJM on a micro habitat scale. Desired outcome: We want to examine if the distribution of individual PMJM can be altered in response to the addition ofresources (food and cover) and to quantify relevant microhabitat characteristics where PMJM have been detected. Approach: A field experiment will be conducted during 2002-2003 (June-August) to test if PMJM can be attracted to areas where they haye not previously been detected within the 100-year flood plain. Study Site- Riparian habitat within the Columbine Open Space, owned by Douglas County Open Space managed by the CDOW and the adjacent grassland. Columbine Open Space was selected �4 because PMJM were monitored for 3 years by the CDOW ( 1998-2000), providing site-specific information on PMJM locations before this manipulation experiment. Methods- PMJM will be trapped using non-folding Sherman live traps (7.6 cm x 8.9 cm x 22.9 cm) placed 5 m apart along approximately 0.5 km transects adjacent to both sides of East Plum Creek for a minimum of 5 consecutive nights. Trapping procedures will be in accordance with the guidelines published by the USFWS ( 1999). Species other than PMJM will be recorded with trap location and immediately released. The following information will be recorded for captured PMJM: unique identification, trap location, weight, sex, age, and reproductive condition. PMJM will be scanned for a passive integrated transponder (PIT) tag. Newly captured individuals will have a unique PIT-tag injected and individuals ~18 grams will be anesthetized with isoflurane to fit a 1-g radio transmitter (Holohil Systems Ltd Ontario, Canada). All methods were approved by the Animal Care and Use Committee of Colorado State University (Authorization Number A3572-0I). Radio telemetry will be used to monitor locations of individuals for a 21-day period, the battery life of the radio transmitters. Observers will attempt to stay approximately 3 m from the radio-tagged individual to avoid influencing PMJM movement. Observations taken 3 m or greater from PMJM did not influence movement (T. Shenk, CDOW personal comm.). The following information will be recorded at each relocation: individual identification, time, weather, and surrounding vegetation. All data will be combined into a geographical information system (GIS) database using ArcView®3.2 (Environmental Systems Research Institute, Redlands, California, U.S.A.). The manipulation experiment will consist of 5 phases: I) select areas of little or no previous use by PMJM based on CDOW location data (1998-2000) collected at Columbine Open Space, 2) record pre-treatment location data ofradio-tagged individuals for 6 nights, 3) select placement of treatment plots based on pre-treatment and CDOW location data, 4) add resources to treatment plots, and 5) record post-treatment location data ofradio-tagged individuals. Two sessions (June and July) of the manipulation experiment will be conducted each year. A digital map with a grid cell size of 9 m x 9 m has been constructed for the entire study site with ArcView®3.2 (Environmental Systems Research Institute, Redlands, California, U.S.A.) software. CDOW location data was pooled into a single coverage over the grid to establish areas~ 1,000 m2 containing only low use cells (<2 locations/cell based on CDOW location data) within the 100-year flood plain. Location of treatments will be selected with a stratified random design from a set of candidate cells meeting a criteria developed based on PMJM biology (sparse vegetation and little food source) within 60 m of East Plum Creek, and low historical use. The artificial cover, simulating vertical complexity, will be constructed with wheat straw and tree branches distributed in a patch (3 m x 2.43 m). Burlap cloth will be suspended 30 cm over the tree branches and straw. Food supplements composed of an equal mixture of whole wheat, dehydrated alfalfa pellets and sweet feed will be placed on cardboard trays (0.16 m x 0.3 m) within the straw and branches as an attractant and a source of high protein. The dimensions of the treatments were selected to balance the manageability of construction and decrease the chance of inter and intraspecies domination within a treatment. Quantification of microhabitat variables in areas of high use will be examined by comparing a random sample of cells (9 m x 9 m) containing~ 99 % of PMJM locations for each session and a random sample of cells with no locations detected. Two line transects will be randomly placed in each selected cell with 6 quadrat frames (50 cm x 20 cm) evenly distributed per line transect (Daubenmire 1959). The variables measured in each cell will include percent bare ground, shrub, grass, and forb cover and vegetation composition . .J �5 Analysis- The location data will be analyzed with linear regression. The response variable will be the number of locations detected in a cell. A suite of candidate models will be developed as predictors of the response variable. Akaike's information criterion (AIC) will be applied to select the best "approximating" model (Burnham and Anderson 2002). The independent habitat variables of interest for the models include distance from the center of the cell to the nearest water, area and juxtaposition of nearest shrub, and presence of wetland grasses in the cell. Additional variables to be included in the models are period (pre- or post-treatment), sex, session, and year. The microhabitat data collected from the Daubenmire plots will be analyzed with Proc GLM (SAS 2002) to test for differences in means among areas of high use and no use by PMJM. Schedule: Fall 2001.. ............... Formation of committee and write study plan development Spring 2002 .............. Completion of study plan and preparation for field season Summer 2002 ........... Begin data collection Fall 2002 ................. Begin data analysis Spring 2003 .............. Continue data analysis, begin thesis and complete comprehensive oral examination Summer 2003 ........... Complete data collection Fall 2003 ................. Complete data analysis Winter 2004 .............. Complete thesis Budget: Fiscal Year 2001-02 Refurbished Holohil radio collars Technicians Housing Vehicles Supplies Computer Tuition GRA Faculty Support FY 2001-02 Total $2,000 1@$1,250/month for 3 months $833/month for 3 months 1@$200/month for 3 months (including mileage) $700 $2,000 $2,880 $1,300/month for 12 months $3,830.00 $33,779 Fiscal Year 2002-03 Refurbished Holohil radio collars PIT tags Technicians Technicians Supplies Housing Vehicles Stipend Tuition Faculty Support FY 2002-03 Total $500 $1,000 1@$1,250/month for 3 months 1@$1,250/month for 2 months $500 $833/month for 3 months 1@$200/month for 3 months (including mileage) $1,300/month for 12 months $679.00 . $3830.00 $31,458 �6 Fiscal Year 2003-04 Stipend FY 2003-04 Total Project Total $1,300/month for 8 months $10,400 $75,637 Potential cooperators: Funding is provided by the CDOW; Douglas County has given permission to use the Open Space; Colorado State University has provided office space, equipment, computers, and adviser. Alternative and obstacles: Alternatives considered include modeling habitat utilization at a microhabitat and site specific scales. Potential obstacles include 1) low number of radio-tagged PMJM resulting in low power for the manipulation experiment, 2) other species deterring PMJM from using the additional resources, and 3) radio-tagged mice do not detect the additional resources. PMJM have demonstrated general site fidelity to daytime nesting sites and nighttime feeding sites (Shenk and Sivert 1999). It is possible PMJM have established use areas and do not easily alter their use patterns. Literature Cited: Bingham, B. B., and B. R. Noon. 1998. The use of core areas in comprehensive mitigation strategies. Conservation Biology 12:241-243. Burnham K. P., and D.R. Anderson. 2002. Model selection and multimodel inference. Second edition. Springer, New York, New York, USA. Daubenmire, R. 1959. A canopy-coverage method ofvegetational analysis. Northwest Science 33:4364. Harding, E., E. Crone, B. D. Elderd, J.M. Hoekstra, A. J. McKerrow, J. D. Perrine, J. Regetz, L. J. Rissler, A.G. Stanley, E. L. Walters and NCEAS Habitat Conservation Plan Working Group. 2001. The scientific foundations of habitat conservation plans: a quantitative assessment. Conservation Biology 15:488-500. Meaney, C. A. 2000. Monitoring for Preble's meadow jumping mice along South Boulder Creek and Four Ditches. Boulder, Colorado, USA. Report prepared for the Colorado Division of Wildlife. SAS Institute. 2002. SAS Version 8.2. SAS Institute, Cary, North Carolina, USA. Schorr, R. 2001. Meadow jumping mice (Zapus hudsonius preblei) on the U.S. Air Force Academy, El Paso County, Colorado, USA. Shenk, T. M., and M. Sivert. 1999. Movement patterns of Preble's meadow jumping mouse (Zapus hudsonius preblei) as they very across time and space. Annual Report to the Colorado Division of Wildlife. Fort Collins, Colorado, USA. U. S. Fish and Wildlife Service. 1999. Interim Survey Guidelines for Preble's meadow jumping mouse. U.S. Fish and Wildlife Service. Denver, Colorado, USA. �13 ketamine (3 mg/kg) administered intramuscularly (IM) with either an extendible pole-syringe or a pressurized syringe-dart fired from a Dan-Inject air pistol. Immobilized lynx were monitored continuously for decreased respiration or hypothermia. If lynx exhibited decreased respiration 2mg/kg ofDopram was administered under the tongue. lfrespiration was severely decreased, the animal was ventilated with a resuscitation bag. If medetomidine/ketamine were the immobilization drug, the antagonist Antisedan was administered. Hypothermic (body temperature < 95° F) animals were warmed with hand warmers and blankets. While immobilized, the lynx were fitted with a replacement VHF/satellite collar and blood and hair samples were collected. Once the animal was processed recovery was expedited by injecting the antagonist Antisedan IM if medetomodine/ketemine was used for immobilization. The lynx was monitored until it was sufficiently recovered to move safely on its own. No antagonist is available for Te)ezol so lynx anesthetized with this drug were monitored until the animal recovered on its own. If captured and in poor body condition the lynx was anesthetized with Telezol (2 mg/kg) and returned to the Colorado holding facility for rehabilitation. Reproduction Reproductive status of all female lynx was determined prior to release through radiographs. Pregnancy was confirmed through radiographs if the bones of the fetuses had begun to ossify. All females known to be pregnant or thought to possibly be pregnant on release were monitored closely from their release through the following August to determine reproductive success. Females remaining within a limited area immediately after release through August were located and observed to look for accompanying kittens or a den site. Females that had been released in 1999 and were alive in spring 2000 were monitored for proximity to males during breeding season and for site fidelity to a given area during the denning period of May and June 2000. Each female lynx from the 1999 releases was directly observed in summer 2000 over 3-5 different visits to look for accompanying kittens or evidence of denning. Each female alive in May 2001 that exhibited stationary movement patterns in June 2001 was observed in summer or fall 2001 to look for accompanying kittens. Females were also snow-tracked in winter months to look for accompanying kitten tracks. Hunting Behavior Snow-tracking of released lynx provided preliminary information on hunting behavior by documenting location of kills, food caches, chases, and diet composition estimated through scat analysis. Snow-tracking was conducted during February-May 1999 (Year 1), November 1999-May 2000 (Year 2), and November 2000 -April 2001 (Year 3). Prey from failed and successful hunting attempts were identified by either tracks or remains. Scat analysis also provided information on foods consumed. Scat samples were collected wherever found and labeled with location and individual lynx identification. Only part of the scat was collected, the remainder was left where found so as not to interfere with the possibility that the scat was being used by the animal as a territory mark. Habitat Use Gross habitat use was documented by recording canopy vegetation at aerial locations. More refined descriptions of habitat use by reintroduced lynx were obtained through snow-tracking and sitescale habitat data collection. Specific objectives for the site-scale habitat data collection included: 1. Describe and quantify site-scale habitat use by lynx reintroduced to Colorado. 2. Compare site-scale habitat use among types of sites ( e.g., kills vs. long-duration beds). 3. Compare site-scale habitat use between sexes. 4. Compare habitat use over years. 5. Develop methodology that will result in data that will be comparable to data collected in studies investigating the ecology of snowshoe hare in Colorado. �14 Snow-tracking Locations from aerial- and satellite-tracking were used to help ground-trackers locate lynx tracks in snow. Snowmobiles, where permitted, were used to gain the closest possible access to the lynx tracks without disturbing the animal. From that point, the tracking team used snowshoes to access tracks. Once tracks were found, the ground crew back- or forward-tracked the animal if it was far enough away not to be disturbed. Back-tracking generally avoided the possibility of disturbing the lynx by moving away from the animal rather than towards the animal. However, monitoring of the lynx through radiotelemetry was used to assure that the ground crew was staying a sufficient distance away from the lynx in the event the lynx might double back on its tracks. Radio-telemetry was also used in forward-tracking to make sure the team did not disturb the animal. If it appeared the lynx began to move in response to the observers, the observers stopped following the tracks. If the lynx began to move and the movement did not appear to be a response to the observers, the ground crew continued following the track. An attempt was made in Year I and Year 2 to track each lynx. In Year 3 we attempted to track all lynx within the Core Release Area. Ground crews were instructed to track lynx only where it was safe to travel. Restrictions to safe travel included avalanche danger and extremely rugged terrain. Ground crews worked in pairs and were fully equipped for winter back-country survival. Data Collection For each day of tracking the date, lynx being tracked, slope, aspect, UTM coordinates, elevation, general habitat description, and summary of the days tracking were recorded. Aspect was defined as the direction of 'downhill' or 'fall line' on a slope. This is the direction along the ground in a dihedral angle between the horizontal and the plane of the ground surface. Units were compass direction that most closely defined the cardinal points (e.g., N, NW, etc.). Slope was defined as the dihedral angle between the horizontal and the plane of the ground surface (e.g., 45° ). There were 4 levels of intensity of human activity recorded. They included: 1. None: track was not found off an existing snowmobile, ski, or snow shoe track. Distance to nearest human track is greater than 1.0 km 2. Low: track was found near low human activity (e.g., existing snowmobile or ski track) 3. Medium: track found near medium human activity (detected the presence of other people in the area during tracking effort). 4. High: track found near high human activity (e.g., detected presence of many people nearby, near major road, near housing). There were 2 categories for recording detection of tracks of other species. They included "M" for tracks from multiple animals of the same species and "T" for detection of tracks of only a single animal of the species. Once a track was located there were 2 types of'sites' that were encountered. Site I areas needed documentation but either did not reflect areas lynx selected for specific habitat features, or sites that occurred too frequently to measure each in detail. Site II areas were places where lynx may have selected habitat features. At each of the 2 types of sites the date, lynx tracked, slope, aspect, forest structure class, UTM coordinates, and elevation was recorded. Forest structure classes included grass/forb, shrub/seedling, sapling/pole, mature, and old growth as defined in Table 3. For Site I areas, the only additional data that was collected was identification of what the site was used for (e.g., short-duration bed), and a brief description of the site. These sites included the start and end of the track being followed, the location of scat, and short-duration beds defined as being small in size (approximating an area a lynx would crouch), and with little ice formed in the bed indicating little time spent there. The Site II areas included areas that might reflect specific habitat features lynx selected for. These sites required habitat sampling (see below) and included locations where the following were found: kills, start of chases, territory marks (e.g., spray sites, buried scat, scat placed on prominent locations), long-duration beds (encompasses an area where a lynx would have lain for an extended period, iced bottom), travel (if no other sites sampled in last hour), and road crossing (both sides of road). �15 Table 3. Definitions of forest structure classes used to describe habitat sites (Thomas 1979). Forest Structure Class Definition Grass/forb The grass/forb stage is created naturally by a catastrophic event, such as wildfire, and is typified by the near complete absence of snags, litter or down material in the aspen and ponderosa pine types, or vice versa in the lodgepole or subalpine forest types. Shrub/seedling The shrub/seedling stage occurs when tree seedlings or shrubs grow up to 2.5 cm at diameter breast height (DBH), either naturally or artificially through planting. Sapling/pole The sapling/pole stage is a young stage where tree DBH's range from 2.5-17 .5 cm with tree heights ranging 1.8-13 .5 m. These trees are 5-100 years of age, depending on species and site condition. Mature The mature stage occurs when tree diameters reach a relatively large size (25-50 cm) and the trees are usually 90 or more years old. Forest stands begin to experience accelerated mortality from disease and insects. Old-growth The old-growth stage occurs when a mature stand is at advanced age (100 years for aspen or 200 years for spruce), is very slow growing, and has advanced degrees of disease, insects, snags, and down, dead material. An exception to this occurs in ponderosa pine and aspen types where these old-growth stands typically experience low densities of down dead material or snags. Description of the Habitat Plot A habitat sampling plot was completed wherever a Site II was encountered. The habitat plot consisted of a 12 m x 12 m square defined by a series of 25 points placed in 5 rows of 5 with the center point being on the object that defined the site (e.g., a kill) (Figure 1). Each point was 3 m apart. The 12 m x 12 m sampling square exceeded the minimum requirement of0.01 ha. Recommended by Curtis (1959) for sampling trees. Measurements taken at each of the 25 points included: 1. Snow depth - measured vertically by an avalanche probe marked in cm. , 6 11 16 2i.... 2. Understory - measured from top of snow to 150 cm above snow in a column of 3-cm radius i + t + t !2 • • • • • around the avalanche probe. Because understory i + t t + t measurements were influenced by vegetation outside !3 • • • • • the perimeter of the 25 sampling points (12 m x 12 i + t + t m) the area used for esfimating undersofy cover was 4. •-•-• • 15 m by 15 m. At each point, crews recorded all + t shrubs, trees and coarse woody debris (CWD) that -~-~-~•-•-•-•_2F fell within this column and was visible above the 12 m snow. Crews also recorded number of branches of 15 m each species that fell within the column at 3 different height categories (0-0.5 m, 0.51-1.0 m, 1.01-1.5 m). 3. Overstory: measured at 150 cm above snow with a sighting tube. The tube was made of PVC pipe, with a curved viewing end and a crosshair made of wire on the opposite end. The sighting tube Figure 1. Design of site-scale habitat sampwas attached to the avalanche probe used to measure ling plot. Each point was 3 m apart. The object snow depth. Species that hit the crosshair were that triggered the habitat sampling (e.g., a kill) recorded at each of the 25 points in the vegetation was located at the center point. plot. Ganey and Block (1994) found this method of h•-•-• •-• �16 measuring canopy cover (with ~ 20 sample points per plot; Laymon 1988) provided greater precision among observers. 4. Species composition: all the different species of tree or shrub that hit. the crosshair of the sighting tube at each of the 25 points were recorded. • Tree composition of the vegetation plot was recorded by species and diameter at breast height (DBH). Snow depth was used in conjunction with this recorded DBH to estimate true DBH. Within the 12 m x 12 m square all conifers and deciduous trees were recorded by DBH size class (A= 0-15 cm, B = 15.1-30 cm, C = 30.1-45 cm, D = 45.1-60 cm, E = »60 cm). Area for the tree composition analysis was 12 m x 12 m. Understory was estimated as: (1) percent occurrence within the vegetation plot (number of points with understory/total number of points surveyed) and (2) mean percent occurrence and variance by species and height category over the total points sampled within the vegetation plot. Overstory was estimated as percent occurrence over the vegetation plot (number of points with overstory/total number of points surveyed). Results Assessment of Release Protocols A total of 41 lynx were released in Colorado in 1999 under 5 different release protocols (Table 2). Release protocols were modified as new information became available from monitoring the released lynx through radio-telemetry and snow-tracking. Each modification of the release protocols decreased the percent of animals dying from starvation (Table 4). Three of the 4 animals released under Protocol 1 died of starvation within 6 weeks of their release and the fourth was recaptured and returned to the holding facility where she recovered and was later re-released. Reevaluation of the condition of animals released under the Protocol I suggested that these animals might not have been in optimal physical condition when released. Therefore, Protocol 2 was initiated. Most lynx gained considerable body weight while in captivity (Wild 1999). Nine lynx were released under this second protocol. Of these, 1 juvenile female died of starvation 7 weeks after release. After the starvation death under Protocol 2, Protocol 3 was developed (3-week minimum holding time, high quality diet, no release prior to May 1). Twenty lynx were released under Protocol 3 with no starvation deaths of these animals occurring within 6 months post-release. Six females were released under Protocol 3P (known to be pregnant) and 2 under Protocol 3P? (possibly pregnant). Two of the 6 pregnant lynx released died of starvation within 6 months post-release. An assessment of the fates of each lynx under all 5 release protocols used in 1999 led to release protocols for lynx released in 2000. Release Protocols 2 and 3 resulted in the fewest starvation mortalities up to 8 months after release date. The common element in both protocols 2 and 3 was increased captivity time in the Colorado holding facility. The single starvation mortality for lynx released under Protocol 2 in 1999 was also the only juvenile released under that protocol and the only animal released in February (the other 8 Protocol 2 lynx were released in March 1999). Thus, all lynx released in 2000 were released under either Protocol 2 or 3 but not before April 1. Because of the high percentage of starvation mortalities in females pregnant on release, we also attempted to avoid reintroducing lynx that were known to be pregnant. This was best accomplished by trying to have animals captured for the reintroduction effort in Canada prior to their breeding season. A series of 11 models (Table 5) were developed using various combinations of the hypothesized factors that may have affected survival up to 8 months post-release: (1) whether the release was in winter or spring (Rel), (2) whether the released lynx was an adult or kitten (age), (3) sex of lynx released (sex), ( 4) whether or not females were released while pregnant (preg) and the interaction of pregnancy and age of the female (adult vs. kitten), and (5) the duration of holding time in the Colorado facility (DCF). Survival time and DCF were modeled with and without a log transformation (Ln) because of possible threshold effects overtime. We used AICc as the model selection criterion to select the model that best explains the data (Table 5). The model that best fit the data was {S(age+preg+Rel+LnT+LnDCF}, �17 which suggested pregnancy had a deleterious effect on survival of females, with the effect being stronger on kittens than adults (Figure 2). This model also indicated that winter releases led to higher mortality than spring releases for both non-pregnant kittens (Figure 3) and non-pregnant adults (Figure 4), with no sex effects on either age class. Lastly, long stays in the Colorado holding facility increased survival if the duration was at least 21 days with no significant decrease or increase in survival for stays longer than 21 days (Figures 2, 3, 4). Table 4. Starvation mortalities and recaptures of poor body condition lynx reintroduced to Colorado under the 5 release Erotocols over 2 }:'.ears. Release Protocol Year Total Number Released I 1999 4 2 2 3 3 3P? 3P? 3P 3P 1999 2000 1999 2000 1999 2000 1999 2000 9 41 20 10 2 3 6 I Number of. Starvation Mortalities a 3b 1c 1c % Mortality Number of Recaptures in Poor Body Condition a 75 11 0 0 0 0 2d 0 100 11 2 0 0 0 0 33 0 I 0 0 0 0 0 0 2 0 0 0 0 33 0 % Failure of Release Protocol 0 0 within 8 months of release. b I juvenile, 2 adults. c juvenile. d adults. 8 Table 5. Model selection results of the a priori models concerning the effects of age, sex, pregnancy, season ofrelease, and amount of time spent in the Colorado holding facility on survival oflynx 8 months Eost-release. Ranking based on AICc values. AICc #Model AICc ~Cc Weight Deviance Pars. {S(age+preg+Rel+LnT+LnDCF} {S(age*preg+Rel+LnT+LnDCF} {S(age+preg+Rel+T+LnDCF} {S(age+preg+Rel+T+T2+LnDCF} {S(age+Rel+preg+LnDCF} {S(age+preg+Rel+T'+T2'+LnDCF} {S(age+preg+Rel+T"+T2"+LnDCF} {S(age*preg +Rel+LnDCF} {S(age*Rel+preg+LnDCF} {S(age+Rel+preg+DCF} {S(age+ReI+ereg+DCF+DCF2} 200.120 201.027 202.702 203.225 203.266 204.069 204.936 205.265 205.289 205.609 205.760 0 1.91 2.58 3.10 3.15 3.95 4.82 5.14 5.17 5.49 5.64 0.28305 0.10908 0.07784 0.05993 0.05871 0.03930 0.02547 0.02161 0.02135 0.01819 0.01687 6 7 6 7 5 7 7 6 6 5 6 188.036 187.914 190.618 189.113 193.206 189.957 190.824 193.181 193.205 195.549 193.676 �18 Kittens -; > .E Adults 1 = ~ 0.8 > i 0.6 =E= u 50 Two-week Interval Figure 2. Effects of pregnancy and tiine spent in the Colorado holding facility on survival of pregnant kittens and adult females. Wmter Release Spring Release Two-week lnlemll Figure 3. Effects ofrelease season and time spent in the Colorado holding facility on survival of non-pregnant kittens. Spring Rd~ T~weeklnterval Figure 4. Effects of release season and time spent in the Colorado holding facility on survival of non-pregnant adults. �19 Movement Patterns A total of 2,158 aerial VHF locations for all 96 reintroduced lynx have been collected to date (Figure 5, Figure 6). An additional 4,020 satellite locations (1,375 satellite locations if multiple locations for a single night were averaged and counted as only I location) for 49 of the 51 lynx fitted with dual collars have been collected. Two satellite collars never worked after the lynx were released. The majority of movements in 1999 away from the an area encompassed by alO0-km radius area centered on the release sites (Core Release Area) were to the north (Figure 5), although some movements occurred to the south into New Mexico and west into Utah as well. A single male from the 1999 releases traveled to Nebraska where he was shot in violation of Nebraska regulations. Initial dispersal habitats used by lynx released in 1999 were highly variable, from high elevation Engelmann spruce/subalpine fir to Nebraska agricultural lands. Dispersal movement directions for lynx released in 2000 were similar to those of lynx released in 1999 (Figure 6). Most movements away from the Core Release Area were to the north. However, more animals remained within the Core Release Area. Numerous travel corridors have been used repeatedly by more than one lynx, possibly suggesting route selection based on olfactory cues. These travel corridors include the Cochetopa Hills area for northerly movements, the Rio Grande Reservoir-SilvertonLizardhead Pass for movements to the west, and southerly movements down the east side of Wolf Creek Pass to the southeast through the Conejos River Valley. Lynx appear to remain faithful to an area during winter months, and exhibit more extensive movements away from these areas in the summer. Such movement patterns have also been documented by native lynx in Wyoming and Montana (Squires and Laurion 1999). Most lynx currently being tracked are within the Core Release Area (Figure 7). Mortalities occurred throughout the areas through which lynx moved. However, mortalities occurred in New Mexico in higher proportion to all lynx locations in that area than elsewhere (Figure 8). Survival and Mortality Factors Of the 96 lynx released, 39 mortalities have been recorded to date. From the 1999 releases (41 animals) we have had 24 known mortalities (Table 6). From the 2000 releases (55 animals) we have 15 known mortalities (Table 6). Of the total 9 confirmed starvation deaths, 3 were associated with animals released in less than ideal body condition (released under Protocol I) and 2 were lynx less than I-year old (Table 4). Fourteen of the mortalities died of unknown causes. In 4 of these cases starvation could be ruled out as cause of death by evidence of good body condition through examination of bone marrow. Pneumonic plague could be ruled out in all 14 cases. Delayed retrieval of carcasses resulted in advanced deterioration of the body, making determination of cause of death impossible. Necropsy results for 3 female lynx released in 2000, indicate they died from pneumonic plague. Two of these lynx were in good condition, with abdominal fat, no muscle wasting, and fat in the bone marrow. The only gross lesions were an acute fibrinous pneumonia (i.e., lung infection of short duration). These lynx had probably only been sick a few days before they died. A third female was in poore~ body condition when found. Plague was diagnosed by flourescent antibody tests and isolation of Yersinia pestis from lung and spleen samples. A fourth lynx was also diagnosed with plague after she was hit by a car. A male lynx, recaptured near Laramie, Wyoming, tested positive for plague titers but did not have active plague. Thus, he had been exposed to plague but either did not contract the disease or recovered from the disease. Recaptures Seven lynx have been recaptured and 6 subsequently re-released since their initial release. Lynx BC99F6 was released in 1999 under Protocol 1. Her behavior and incidental sightings by the public suggested the lynx was in poor condition. We trapped her using a Tomahawk™ live trap baited with rabbit. She was recaptured the first night (March 25, 1999) we set the trap. On capture, we found she was severely emaciated. We anesthetized her with Telezol (2 mg/kg) and returned her to the Colorado holding facility. She was rehabilitated through diet. The lynx gained weight steadily and was re-released on May �20 . . --------i l J __) I ----M• ~-- •• • • Locations of lynx Released in 1999 ('JHigh-ys • •• 'c:::J Colorado Counties [-=-j New Mexico Counties ···'·'··'··· Wyoming Cowities NNe!nslm Counties s 300 - 0 300 Figure 5. Locations oflynx released in 1999, obtained through aerial telemetry. 600 Kilometers �21 • VHF·locations of lynxrele~ed in 200:0 • • PTJ locations of lynx released. In 2000 Highways. ·Colontdo Counties N_ew Mexico C ouities r---; ~oming Counties • A/ t:J c:J 75/Nebras1ca Counti~ N ./\./ utah C~unties W*E 300 - s 0 600 .Kilometers Figure 6. Locations of lynx released in 2000. Gray circles indicate locations obtained from satellite collars. Black circles are locations obtained through aerial telemetry. �22 Last Known Locations • alive ■ collar off .~ missing /'/Highways Colorado Counties .-.-, New Mexico Counties t:::=J s -- 300 0 300 600 Kilometers Figure 7. Last known locations oflynx. Circles depict locations oflynx currently being tracked. Triangles are last known locations of missing lynx. �23 • .4· ·Mortality looation·s of lynx releasecl·in 2000 • Mort.lity looations of lynx released iii 1999 / \ / Highways t::J Colorado Counties CJ NeW. Mexico· Counties N . W * E s -- 300 0 300 600 Kilometers Figure 8. Locations of lynx mortalities. Circles depict mortalities of lynx released in 1999, triangles depict mortalities from lynx released in 2000. I �24 Table 6. Causes of death for lynx released into southwestern Colorado in 1999 and 2000. 1999 1999 2000 2000 2000 Male Female Male Female Unknown Cause Starvation 1 6 1 1 Road-kill 2 2 2 Shot 1 Human-caused a Trauma - unknown cause Possible predation Plague 3 Unknown 2 3 2 2 Unknown - not starvation b 1 2 Total Mortalities 7 17 4 10 1 Total 9 5 5 2 1 3 9 4 39 a Cut collar found, no carcass. b Starvation ruled out by condition of bone marrow. 28, 1999. She was hit by a car on Interstate 70 on July 19, 1999. Necropsy results indicated she was in excellent body condition at her time of death. Lynx AK99M9 was released on May 12, 1999 and recaptured on March 24, 2000. Field observations by the lynx monitoring crew suggested that the lynx was severely emaciated. Live-trapping the lynx failed, so the lynx was darted with Telazol (3 mg/kg) using a Dan-Inject CO 2 pistol. Physical examination revealed severe emaciation (6 kg). The lynx was returned to the Colorado holding facility and rehabilitated through diet. The lynx gained weight steadily and was re-released on May 3, 2000 but has not been located since and is listed as missing. Lynx AK99F2 was released on May 7, 1999 and recaptured on April 18, 2000. Field observations by the lynx monitoring crew suggested that the lynx was emaciated. She was live-trapped with a Tomahawk™ live trap with one night's effort. On capture, we found she was emaciated. We anesthetized her with Telezol (2 mg/kg) and returned her to the Colorado holding facility. She was rehabilitated through diet. The lynx gained weight steadily and was re-released on May 22, 2000. This lynx is currently in the Core Release Area. Lynx BC00F7 was released on April 2, 2000 and recaptured on February 11, 2001. She was severely emaciated and was captured by anesthetizing her with Telazol delivered IM by a jab-pole. She was returned to the Frisco Creek Wildlife Rehabilitation Center but died that night from emaciation and hypothermia. Lynx BC00M13 was released on April 2, 2000 and recaptured on March 21, 2001 near Laramie, Wyoming. He had been observed by a homeowner on his porch. We recaptured the lynx because this type of behavior was not considered normal. On examination he was in good body condition. After a period of observation this lynx was re-released at the Rio Grande Reservoir on April 24, 2001. This lynx had previously been listed as one of our 15 missing lynx as he had not been located since Sept 2000. This lynx is currently in the Core Release Area Lynx YK99F5 was recaptured on Aprill 9, 2001 to have her radio collar changed. She was captured in a live trap baited with one of her own kills. She was in very good body condition. We anesthetized her with Telazol (3mg/k.g), processed and released her on the same site where she was captured. Only her cut collar was found on October 17, 2001, cause of death is assumed human-caused. Lynx AK99F5 was treed by hounds and anesthetized with Telazol (3 mg/kg) on September 2, 2001. Her collar was exchanged, hair and blood samples were collected. She was in very good body condition and showed no evidence of lactation. She was re-released on the site she was recaptured once she recovered from the Telazol. This lynx remains in the same area as her recapture, within the Core Release Area. �25 Reproduction Six lynx released in 1999 were known to be pregnant (Table 2, Release Protocol 3P), and 2 other females released may have been pregnant (Table 2, Release Protocol 3P?). Three of the 6 lynx known to have been pregnant on release in 1999 died within 2 months after release: 2 starved and 1 was killed on the road. Long-distance movements and lack of stationary movement patterns of the other 3 lynx known to have been pregnant on release in 1999 suggests these females did not have young with them by July 1999. Of the 2 females that might have been pregnant, movement patterns were not suggestive of a female rearing young. It is not known if any other females bred and/or had young once released, however no females snow-tracked in Year 2 had young with them. Beginning in March 2000 both male and female lynx began to exhibit extensive movements (> 100 km) away from areas they had used throughout the winter. For example, 1 male moved from the area near Frisco he used in the winter to the area west of Lizardhead Pass, a straight line distance of approximately 270 km (Figure 9). Such movements by both females and males put them in close (< 5 km) proximity to a lynx of the opposite sex. These extreme movements may have been related to breeding behavior. All 7 females alive in spring 2000 were documented in close(< 5 km) proximity to a male during the breeding season and could have bred. Two isolated males did not move during March or April and thus were not in close proximity to a known _female during the breeding season. This was a male that had used the area in and adjacent to the northwest comer of Rocky Mountain National Park and a male that used the area around Cuchara, Colorado throughout the winter. The 7 females in the wild during breeding season 2000 were monitored for site fidelity to a given area during the denning period of May and June. Each of these 7 females was directly observed in summer 2000 over 3-5 different visits to look for accompanying kittens. No kittens were found. The question of whether they successfully bred or had kittens at some point in 2000 is unknown. However, no kittens were found during the following winter through snow-tracking. From radiographs taken of the 35 females released in 2000, after breeding season,1 female was known to be pregnant and 3 were possibly pregnant. Movement patterns suggested none of these 4 females had kittens with them by July 2000. Of the 49 lynx being tracked on a regular basis during the March 2001 breeding season, there were 29 females and 20 males. We documented movements that may have been related to breeding. The largest movement observed was a male that moved to Laramie, Wyoming and was subsequently recaptured, rehabilitated and re-released in the Core Release Area in Colorado after the breeding season. Other movements were of a much smaller scale, 10-30 km. These movements were primarily movements of males towards a female. We documented 10 potential 'pairs' where a pair was defined as a male and female within 5 km of each other and in the same drainage. More pairs could have occurred which we did not document from aerial- or ground-tracking because of the time delays between lynx locations. To date, no reproduction has been documented in 2001 from direct observations of females. Snow-tracking efforts this winter will focus initially on females in an attempt to document possible kittens through tracks. Current Status Of the total 96 lynx released we have 39 known mortalities (Table 7). We currently are listing 16 lynx as missing - 11 males, 5 females. We have not heard signals on 13 (11 males, 2 females) of these lynx since at least December 2000. The remaining 3 missing lynx are females that have been lost for less than 1 year. Possible reasons for not locating these missing lynx include (1) long distance dispersal, beyond the areas currently being searched, (2) radio failure, or (3) destruction of the radio (e.g., run over by car). We continue to search for all missing lynx during both aerial and ground searches. There have been 4 incidents whe;:re lynx missing for over a year have returned to the Core Release Area and are now once again being monitored on a regular basis. Thus, it is premature to consider missing lynx as lost to the Colorado lynx program. However, of the 16 missing lynx, 3 have collars whose battery life expired spring 2001 and will probably never be located through telemetry. At least 1 of the missing lynx is a mortality where we know a collar was found on a road kill bot the collar was not returned to the �26 NHighways -□ Colorado Counties Lynx YK99M3 Movements 19990513 • 19990521 • 19990521 • 19990715 • 19990715 • 19990923 • 19990923 • 19991027 • 19991027 • 19991229 • 19991229 • 20000229 • 20000229 • 20000329 • 20000329 • 20000512 N W*E s -- 300 0 300 600 Kiloinete Figure 9. Movements of a male lynx in breeding season 2000. Straight-line distance from winter use area to the area used during breeding season in approximately 270 km. The larger the circle the more recent the date, uptoMay2000 �27 CDOW for identification. One female is known to have slipped her collar. Thus, we are currently tracking 41 lynx. Table 7. Current status of lynx reintroduced to Colorado. Females Males Unknown Released 57 39 Known Dead 27 11 1 Missing 5 11 Slipped Collar 1 Tracking 24 17 TOTAL 96 39 16a 1 41 a l is unknown mortality. Hunting Behavior Snow-tracking of released lynx provided preliminary information on hunting behavior by documenting location of kills, food caches, chases, and through scat analysis. Prey from failed and successful hunting attempts were identified by either tracks or remains. Scat analysis also provided infonnation on foods consumed. During Year 1 a total of 10 kills were located. All the snow-tracking effort was conducted on 9 lynx released under Protocols 1 and 2. Any lynx released under Protocol 3 were released too late to track. In Year 2, ground crews tracked 13 of the lynx released in 1999. Two other lynx were being located during this time but were not in areas covered by snow. We found 64 kills and collected 109 scat samples that will be analyzed for content. Lynx released in 2000 were released too late to snow track in Year 2. In Year 3, field crews snow-tracked 48 lynx, documented 86 kills and collected 189 scat samples. Data collected on kills (Figure 10) suggests the reintroduced lynx are feeding on their preferred prey species, snowshoe hare (Lepus americanus) and pine (red) squirrel (Tamiasciurus hudsonicus) in similar proportions as those reported for northern lynx during lows in the snowshoe hare cycle (Aubry et al., 1999). Caution must be used in interpreting the proportion of identified kills. Such a proportion ignores other food items that ~e consumed in their entirety. For example, through snow-tracking we have some evidence that lynx are mousing and several of the fresh 70 carcasses have yielded small mammals in 60 the gut on necropsy. 1131999 ■ 1999-00 ■ 2000-01 However, the extent of small 50 ~ mammals in the diet are not accurately 'o 40 portrayed by information collected based a.. _8 30 on prey remains in snow. Nearly all the e::S 20 scat samples collected have been found z through snow-tracking efforts and thus will 10 be representative of winter diet only. The summer diet of lynx elsewhere has been 0 documented to include less snowshoe hare Snowshoe Cottontail Other Red and more alternative prey than in winter hare squirrel (Mowat et al. 1999). Species = Habitat Use Gross habitat use was documented from 2441 aerial locations oflynx I Figure 10. Winter diet of reintroduced lynx estimated from snow-tracking data. �28 collected from February 1999 through December 2001. Throughout the year 80 Engelmann spruce (Picea engelmannii) 70 / subalpine fir (Abies lasiocarpa) (S/F) "'= 60 was the dominant cover used by lynx -~ (Figure 11 ). A mix of Engelmann cc 50 CJ spruce, subalpine fir and aspen 0 40 ~ (Populus tremuloides) (S/F/A) was the = ll> 30 second most common cover type used CJ i.. ll> throughout the year. Various riparian p.., 20 and riparian mix areas was the third 10 most common cover type where lynx were found during the daytime flights. 0 Use of S/F and S/F/A was similar J J A S O N D J F M A M throughout the year. There was a trend Months in increased use of riparian areas beginning in July, peaking in Noveml ■ s/F DS/F/A ■ Riparian ber, and dropping off December through June. Figure 11. Percent aerial locations in Engelmann spruce A total of 4 73 site-scale habitat subalpine fir forests (S/F), Engelmann spruce- subalpine firplots were completed in Year 3. The aspen forests (S/F/A), and riparian areas by month. majority understory species at all 3 heights was Engelmann spruce, followed by subalpine fir, willow (Salix spp.) and aspen (Figure 12). Various other species such as Ponderosa pine (Pinus ponderosa), lodgepole pine (Pinus contorta), cottonwood (Populus sargentii), birch 100 (Betula spp.) and others were also found in less than 5% of the habitat 9 0 ,□ LOW aMEDIUM •HIGH plots. Coarse woody debris was also 8 0 present in 10-35% of plots. If present, 70 willow provided the greatest percent cover within a plot (Figure 13) 6 0 = ll> followed by Engelmann spruce, CJ 50 ~ subalpine fir, aspen and coarse woody i:i... 40 debris. 30 Engelmann spruce provided a mean of35.87 % overstory within 20 86.68% of the plots (Figure 14). 10 Subalpine fir and aspen provided 0 overstory for< 50% of the plots, but SF CWD WI AS LO ES when present provided approximately Species the same mean percent cover as Engelmann spruce (Figure 14). Willow and lodgepole pine provided fewer than l 0% of the plots with Figure 12. Mean percent cover of habitat plot by understory cover, but when present provided tree/shrub species Engelmann spruce (ES), subalpine fir (SF), nearly the same percent cover as the willow (WI), aspen (AS), lodgepole pine (LO), and coarse other tree species (Figure 14). woody debris (CWD) if species is present. Mean percent The most common tree species cover is estimated for 3 height levels above the snow (low = in the habitat plots was Engelmann 0-0.5 m, medium= 0.51-1.0 m, high= 1.1-1.5 m). - I - �29 50 45 ,□ LOW ■ MEDIUM ■ HIGH I 40 35 30 = . 25 .,u ., I>. 20 15 10 5 0 ES SF CWD WI Species AS LO Figure 13. Mean percent cover of habitat plots by understory tree or shrub species Engelmann spruce (ES), subalpine fir (SF), willow (WI), aspen (AS), lodgepole pine (LO), and coarse woody debris (CWD) if species is present. 100 90 80 70 EIES 60 c...u ...u "- ■ ES Snag ■ SF 50 □ SF Snag ■ AS 40 ■ AS Snag ■ WI 30 EILO 20 10 0 Percent Plots With These Species Mean Percent Cover if Species Present Figure 14. Percent plots with overstory tree species Engelmann • spruce (ES), subalpine fir (SF), willow (WI), aspen (AS), lodgepole pine (LO), and coarse woody debris (CWD). Mean percent overstory cover if tree species present. spruce (Figure 15). Subalpine fir and aspen were also present in> 35% of the plots. Most habitat plots were vegetated with trees of DBH < 6" (Figure 15). As DBH increased, percent occurrence decreased within the plot. The larger DBH trees (> 18") within the plots were generally Engelmann spruce with fewer subalpine firs of that DBH class present in the habitat plots. No willow or aspen of DBH > 18" were present in any of the plots. Of the 5 most common tree species in the habitat plots, mean number of trees for each DBH size class ranged from 0.18 to 10.82 except for willow which averaged 74.83 plants per plot (Figure 16). Areas of willow used by lynx are typically dense willow thickets. Discussion Of the 96 lynx released in Colorado in 1999 and 2000 we are currently monitoring 41 lynx on a regular basis and an additional 16 lynx may still be alive, although not being monitored. We have 39 confirmed mortalities. Survival oflynx released in the second year has been higher than lynx released in the first year. Human-caused mortalities due to vehicle collision, gunshot, and the mortalities where only a cut collar was found comprise the greatest known cause of mortality for the reintroduced lynx (31 % ). Mortalities due to starvation (23%) were minimized with improved release protocols. Only 2 of the 55 lynx released in 2000 died of starvation and 1 of those �30 died 8.5 months post-release. Three lynx died of plague, 1 road kill tested positive for plague, and 1 lynx had plague positive titers while healthy. Carnivores are most often exposed to plague by eating infected rodents or by being bitten by rodent fleas (Biggens and Kosoy 200] ). Although it is known that felids are highly susceptible to plague (Aiello 1998), the 5 cases of plague in lynx reintroduced to Colorado are the first documented for this species. Dispersal movement patterns for lynx released in 2000 were similar to those of lynx released in 1999. However, more animals remained within the Core Release Area. This increased site fidelity may be due to the presence of con-specifics in the area on release. Numerous travel corridors have been used repeatedly by more than ] lynx, possibly suggesting route selection based on olfactory cues. These travel corridors include the Cochetopa Hills area for northerly movements, the Rio Grande Reservoir-SilvertonLizardhead Pass for movements to the west, and southerly movements down the east side of Wolf Creek Pass to the southeast to the Conejos River Valley. Lynx appear to remain faithful to an . area during winter months, and exhibit more extensive movements away from these areas in the summer. Most lynx currently being tracked are within the Core Release Area. During the summer of 2000 and 2001, several lynx that had been faithful to a 100 D0-6"DBH 90 .6.1-12" DBH 80 •12.1-18" DBH D 18.1-24" DBH 70 • > 24" DBH 60 = "' :::'. 50 "' 0.. 40 30 20 I0 0 ES SF CWD WI AS LO Species Figure 15. Percent of habitat plots with tree species Engelmann spruce (ES), subalpine fire (SF), willow (WI), aspen (AS), lodgepole pine (LO), and coarse woody debris (CWD) by diameter at breast height (DBH) size class. 100 90 □ 0-6"DBH .6.1-12" DBH 80 .12.1-18" DBH 70 = ." "' D18.1-24" DBH . > 24" DBH 60 50 "' 0.. 40 30 20 10 0 ES SF CWD WI AS LO Species Figure 16. Mean number of trees or shrubs in habitat plots with tree species Engelmann spruce (ES), subalpine fire (SF), willow (WI), aspen (AS), lodgepole pine (LO), and coarse woody debris (CWD) by diameter at breast height (DBH) size class. �31 given area during the winter months made large movements away from their winter-use areas. Extensive summer movements away from areas used throughout the rest of the year have been documented in native lynx in Wyoming and Montana (Squires and Laurion 1999). In winter, lynx reintroduced to Colorado appear to be feeding on their preferred prey species, snowshoe hare and red squirrel in similar proportions as those reported for northern lynx during lows in the snowshoe hare cycle (Aubry et al., 1999). Caution must be used in interpreting the proportion of identified kills. Such a proportion ignores other food items that are consumed in their entirety and thus are biased towards larger prey and may not accurately represent the proportion of smaller prey items, such as microtines, in lynx winter diet. Through snow-tracking we have evidence that lynx are mousing and several of the fresh carcasses have yielded small mammals in the gut on necropsy. Nearly all the scat samples collected have been found through snow-tracking efforts and thus are representative of winter diet only. However, the summer diet of lynx has been documented to include less snowshoe hare and more alternative prey than in winter (Mowat et al., 1999). Reproduction is critical to achieving a self-sustaining viable population of lynx in Colorado. Although females have been monitored and observed during each denning season, no kittens have been found to date. Snow-tracking has also not provided evidence that any of the females tracked had kittens with them. However, the question of whether they successfully bred or had kittens at some point is unknown. With only 7 females from the 1999 releases in the wild in spring 2000 it was expected that there _might not be successful reproduction in 2000. However, the extreme movements observed by both females and males in March and April 2000 may have been related to breeding behavior. March and April are the natural breeding periods for northern lynx (Tumlison 1987). From observations of the 29 females alive in summer 2001, we have not yet documented kittens. We may still find evidence of kittens through snow-tracking efforts in winter 2001-02. Mowat et al. (1999) suggest lynx and snowshoe hare select similar habitats except that hares select more dense stands than lynx. Very dense understory limits hunting success of the lynx and provides refugia for hares. Given the high proportion of snowshoe hare in the lynx diet in Colorado, we might then assume the habitats used by reintroduced lynx also depict areas where snowshoes hare are abundant and available for capture by lynx in Colorado. From both aerial locations taken throughout the year and from the site-scale habitat data collected in winter, the most common areas used by lynx are in stands of Engelmann spruce and subalpine fir. This is in contrast to adjacent areas of Ponderosa pine, pinyonjuniper, aspen and oakbrush. The lack oflodgepole pine in the areas used by the lynx may be more reflective of the limited amount oflodgepole pine in southwestern Colorado, the Core Research Area, rather than avoidance of this tree species. Hodges (1999) summarized habitats used by snowshoe hare from 15 studies as areas of dense understory cover from shrubs, stands that are densely stocked, and stands at ages where branches have more lateral cover. Species composition and stand age appears to be less correlated with hare habitat use than is understory structure (Hodges 1999). The stands need to be old enough to provide dense cover and browse for the hares and cover for the lynx. In winter, the cover/browse needs to be tall enough to still provide browse and cover in average snow depths. Hares also use riparian areas and mature forests with understory. Site-scale habitat use documented for lynx in Colorado indicate lynx are most commonly using areas with Engelmann spruce understory present from the snow line to at least 1.5 m above the snow. The mean percent understory cover within the habitat plots is typically less than 15% regardless ofunderstory species. However, if the understory species is willow, percent understory cover is typically double that,.with mean number of shrubs per plot approximately 80, far greater than for any other understory species. • In winter, hares browse on small diameter woody stems (<0.25"), bark and needles. In summer hares shifts their diet to include forbs, grasses, and other succulents as well as continuing to browse on woody stems. This shift in diet may express itself in seasonal shifts in habitat use, using more or denser coniferous cover in winter than in summer. The increased use ofriparian areas by lynx in Colorado from July to November may reflect a seasonal shift in hare habitat in Colorado. Major (1989) suggested lynx hunted the edge of dense riparian willow stands. The use of these edge habitats may allow lynx to hunt �32 hares that live in habitats nonnally too dense to hunt effectively. The use of riparian areas and riparianEngelmann spruce-subalpine fir and riparian- aspen mixes documented in Colorado may stem from a similar hunting strategy. However, too little is known about habitat use by hares in Colorado to test this hypothesis at this time. Lynx also require sufficient denning habitat. Denning habitat has been described by Koehler ( 1990) and Mowat et al. ( 1999) as areas having dense downed trees, roots, or dense live vegetation. No den sites have been located as yet in Colorado for comparison. Through extensive monitoring of released animals we were able to continuously evaluate and modify release protocols to improve survival of released lynx. The primary element in later, more successful release protocols was increased time in captivity at the Colorado holding facility. Increasing the amount of time lynx were held in the Colorado holding facility provided each lynx with an opportunity to increase body weight and acclimate to the climate, elevation, and local conditions of the environment they would be released into. Although most lynx were housed in individual pens, with a few sharing a pen with one other lynx, the holding facility also allowed the lynx to hear and smell each other throughout this acclimation period. Such contact may have provided time for social interactions to occur. Such social interactions may improve the likelihood these animals could fonn a breeding population. If additional lynx are released in Colorado the following guidelines are recommended in establishing release protocols. Translocated animals should be adults and females should not be pregnant on release. Once lynx are moved from their place of origin they should be held a minimum of 3 weeks in a local holding facility to provide a high quality diet for gaining optimal body condition prior to release in the new area, acclimation time to adjust to local conditions, and possible social interactions. Animals should be released in the spring to ensure the highest prey abundance in the release area. These release protocol guidelines may also prove useful if other states attempt lynx reintroductions or augmentations. Future Research Future research will include the continued monitoring of lynx released in Colorado that have remained in the Core Release Area. Such monitoring will include continued data collection and analysis on survival and mortality factors, reproduction, habitat use, winter and summer diet, and movement patterns. If additional funding becomes available, reintroduced lynx that have moved beyond the Core Release Area should also be monitored, particularly those lynx using areas near the Interstate Highway 70 corridor. We will continue to attempt to recapture lynx to replace radio collars that are either malfunctioning or scheduled to stop functioning. Any Colorado born lynx will be radio collared once they reach a minimum of 10 months of age. Studies have been initiated to refine mark-recapture techniques to estimate abundance oflynx from hair-snag data. Such an approach would provide a non-invasive technique for estimating abundance. A snowshoe hare ecology study was initiated in 2001 to describe density of hares in various forest stands and which habitats and topographic features are most important to hare density and survival. From this research, management prescriptions may be designed to better manage forests for optimal hare populations. Maintaining abundant and widespread snowshoe hare populations is essential to establishing lynx in Colorado. Through funding provided by Colorado Department of Transportation (COOT) a detailed analysis of lynx movement patterns as they relate to highways has been initiated. The feasibility of augmenting this reintroduction effort by releasing additional animals from Canada and Alaska is being considered by CDOW to improve the likelihood of establishing a viable population oflynx in Colorado. Funding is being sought to develop protocols for collecting data on lynx summer diet by using dogs trained to locate lynx scat. �33 If viable, self-sustaining populations oflynx are established in Colorado, habitat manipulation studies will be needed to more fully understand how lynx respond to their habitat and how best to alter habitats to maintain and enhance lynx populations. Acknowledgments The lynx reintroduction program involved the efforts of literally hundreds of people across North America, in Canada and the U.S. Any attempt to properly acknowledge all the people who played a role in this effort is at risk of missing many people. The following list should be considered to be incomplete. CDOWCLAWS Team (1998-2001): Bill Andree, Tom Beck, Gene Byrne, Bruce Gill, Mike Grode, Rick Kahn (Program Leader), Dave Kenvin, Todd Malmsbury, Jim Olterman, Dale Reed, John Seidel, Scott Wait, Margaret Wild CDOW: John Mumma (Director 1996-2000), Conrad Albert, Jerry Apker, Cary Carron, Don Crane, Larry DeClaire, Phil Ehrlich, Lee Flores, Delana Friedrich, Dave Gallegos, Juanita Garcia, Drayton Harrison, Jon Kindler, Ann Mangusso, Jerrie McKee, Melody Miller, Mike Miller, Kirk Navo, Robin Olterman, Jerry Pacheo, Mike Reid, Ellen Salem, Eric Schaller, Mike Sherman, Jennie Slater, Steve Steinert, Kip Stransky, Suzanne Tracey, Anne Trainor, Brad Weinmeister, Nancy Wild, Perry Will, Brent Woodward, Kelly Woods, Kevin Wright. Lynx Advisory Team (19982001): Steve Buskirk, Jeff Copeland, Dave Kenny, John Krebs, Brian Miller (Co-leader), Mike Phillips, Kim Poole, Rich Reading (Co-leader), Rob Ramey, John Weaver. U.S. Forest Sen,ice: Kit Buell, Joan Friedlander, Jerry Mastel, John Squires, Fred Wahl. U.S. Fish And Wildlife Sen,ice: Lee Carlson, Gary Patton (1998-2000), Kurt Broderdorp. State Agencies: Gary Koehler (Washington). National Park Sen,ice: Steve King. Colorado State University: Alan B. Franklin, Gary C. White. Colorado Natural Heritage Program: Rob Schorr, Mike Wunder. Alaska: ADF&G: Cathie Harms, Mark Mcnay, Dan Reed (Regional Manager), Wayne Reglin (Director), Ken Taylor (Assist. Director), Ken Whitten, Randy Zarnke, Other:Ron Perkins (trapper), Dr. Cort Zachel (veterinarian). British Columbia: Dr. Gary Armstrong (veterinarian), Mike Badry (government), Paul Blackwell (trapper coordinator), Trappers: Dennis Brown, Ken Graham, Tom Sbo, Terry Stocks, Ron Teppema, Matt Ounpuu. Yukon: Government: Arthur Hoole (Director), Harvey Jessup, Brian Pelchat, Helen Slama, Trappers:Roger Alfred, Ron Chamber, Raymond Craft, Lance Goodwin, Jerry Kruse, Elizabeth Hofer, Jurg Hofer, Guenther Mueller (YK Trapper's Association), Ken Reeder, Rene Rivard (Trapper coordinator), Russ Rose, Gilbert Tulk, Dave Young. Alberta: Al Cook. Northwest Territories: Albert Bourque, Robert Mulders (Forbearer Biologist), Doug Steward (Director NWT Renewable Res.), Fort Providence Native People. Colorado Holding Facility: Herman and Susan Dieterich. Pilots: Dell Dhabolt, Larry Gepfert, Al Keith, Jim Olterman, Matt Secor, Whitey Wannamaker, Dave Younkin. Field Crews: Bryce Bateman, Bob Dickman, Denny Morris, Gene Orth, Chris Parmater, Jake Powell, Jeremy Rockweit, Jennifer Zahratka. Photographs: Tom Beck, Bruce Gill, Mary Lloyd, Rich Reading, Rick Thompson. Funding: CDOW, GOCO, Turner Foundation, U.S. Forest Service, Vail Associates. Literature Cited Aiello, S. E., editor 1998. The Merck Veterinary Manual. Eighth Edition. Merck & Co., Inc. Whitehorse Station, New Jersey. Aubry, K. B., G. M. Koehler, J. R. Squires. 1999. Ecology of Canada lynx in southern boreal forests. Pages 373-396 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U.S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. Biggins, D. E. and M. Y. Kosoy. 2001. Influences of introduced plague in North American mammals: implications from ecology of plague in Asia. Journal of Mammalogy 82: 906-916. Burnham, K. P. and D.R. Anderson. 1998. Model Selection and Inference: A Practical InformationTheoretic Approach. Springer-Verlag, New York, New York. �34 Byrne, G. 1998. Core area release site selection and considerations for a Canada lynx reintroduction in Colorado. Report for the Colorado Division of Wildlife. Curtis, J. T. 1959. The vegetation of Wisconsin. University of Wisconsin Pres, Madison. Ganey, J. L. and W. M. Block. A comparison of two techniques for measuring canopy closure. Western Journal of Applied Forestry 9:1: 21-23. Hodges, K. E. 1999. Ecology of snowshoe hares in southern boreal and montane forests. Pages 163-206 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. Koehler, G. M. 1990. Population and habitat characteristics oflynx and snowshoe hares in north central Washington. Canadian Journal of Zoology 68:845-851. Laymon, S. A. 1988. The ecology of the spotted owl in the central Sierra Nevada, California. PhD Dissertation University of California, Berkeley, California. Major, A. R. 1989. Lynx, Lynx canadensis canadensis (Kerr) predation patterns and habitat use in the Yukon Territory, Canada. M. S. Thesis, State University of New York, Syracuse. Mowat, G., K. G. Poole, and M. O'Donoghue. 1999. Ecology of lynx in northern Canada and Alaska. Pages 265-306 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U.S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53: 7-15. Poole, K. G., G. Mowat, and B. G. Slough. 1993. Chemical immobilization oflynx. Wildlife Society Bulletin 21:136-140. Seidel, J., B. Andree, S. Berlinger, K. Buell, G. Byrne, B. Gill, D. Kenvin, and D. Reed. 1998. Draft strategy for the conservation and reestablishment of lynx and wolverine in the southern Rocky Mountains. Report for the Colorado Division of Wildlife. Shenk, T. M. 1999. Program narrative: Post-release monitoring of reintroduced lynx (Lynx canadensis) to Colorado. Report for the Colorado Division of Wildlife. Squires, J. R. and T. Laurion. 1999. Lynx home range and movements in Montana and Wyoming: preliminary results. Pages 337-349 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University Press of Colorado, Boulder, Colorado. Thomas, J. W. Ed. 1979. Wildlife habitats in managed forests - the Blue Mountains of Oregon and Washington. USDA Agricultural Handbook No. 553. U.S. Government Printing Office. Washington, D. C. Tumlison, R. 1987. Mammalian Species: Fe/is lynx. American Society ofMamrnalogists. U. S. Fish and Wildlife Service. 2000. Endangered and threatened wildlife and plants: final rule to list the contiguous United States distinct population segment of the Canada lynx as a threatened species. Federal Register 65, Number 58. White, G. C. and K. P. Burnham. 1999. Program MARK: Survival estimation from populations of marked animals. Bird Study 46 (suppl):120-138. Wild, M.A. 1999. Lynx veterinary services and diagnostics. Job Progress Report for the Colorado Division of Wildlife. Fort Collins, Colorado. �1 JOB PROGRESS REPORT Stateof_ _ _ _ _ _~C=o=l=or=a=d=o_ _ _ __ Division of Wildlife - Mammals Research Work Package No. --~0~6~62~------- Preble's Meadow Jumping Mouse Conservation Task No. - - - - - ~ 2 ~ - - - - - - - Effects of Resource Addition on Preble's Meadow Jumping Mouse (Zapus hudsonius preblei) Movement Patterns Period Covered: July 1, 2002 - June 30, 2003 Author: Anne M. Trainor. Personnel: T. M. Shenk, K. Wilson, G. C. White Interim Report - Preliminary Results This work continues, and precise analysis ofdata has yet to be accomplished. Manipulation or interpretation of these data beyond that contained in this report should be labeled as such and is discouraged. ABSTRACT The Preble's meadow jumping mouse (Zapus hudsonius preblei; PMJM) is a federally threatened species. Improving our understanding of PMJM habitat is essential for the development of effective management strategies for conservation of the species. Thus, the objectives of our research were to compare microhabitat characteristics among low and high use areas within PMJM habitat and to determine how the addition of artificial resources influence the movement patterns of PMJM. A comparison of microhabitat characteristics from a random sample of "high-use" and "no-use" areas indicated a greater (P < 0.0001) shrub canopy cover in "high-use" areas verses "no-use" areas (47.7% ± 29.8%, 12.6% ± 14.11 %, respectively). Further, "high-use" areas had greater basal cover (P = 0.013) and bare ground (P = 0.0459) and "no-use" areas contained a greater (P = 0.0331) abundance of forb canopy cover. We conducted a manipulation experiment where we constructed patches of artificial resources (food and cover) in areas without previous PMJM activity. PMJM were radio collared and located hourly before and after the addition of food and cover. The majority of PMJM movements were not influenced by the addition of resources in 2002. These results may be due to site fidelity or lack of exploratory movement to locate the additional resources �2 Effects of Resource Addition on Preble's Meadow Jumping Mouse (Zapus hudsonius preb/ei) Movement Patterns Anne M. Trainor Colorado State University INTRODUCTION The U.S. Fish and Wildlife Service (USFWS) listed the Preble's meadow jumping mouse (Zapus hudsonius preblei; PMJM) as a threatened species in 1998 under the Endangered Species Act (USFWS 1999). Upon listing, little was known about the biology and habitat requirements of this subspecies within its range along the Front Range of Colorado and southeastern Wyoming. Since listing, a number of projects (e.g., long-term monitoring, surveying, and movement studies) have collected valuable information throughout Colorado (Schorr 2001, Meaney 2000, Shenk and Sivert 1999). However, information on specific habitat requirements and their relationship to the distribution, density, survival and reproduction of PMJM is still lacking. The threatened status of PMJM requires management decisions be made despite our limited knowledge. In particular, the species and its habitat are subject to habitat conservation plans (HCPs). HCPs are written for endangered and threatened species to compensate for authorized "take" through mitigation practices (Bingham and Noon 1998). HCPs require the use of the "best available" science to determine the biological needs of target species (Harding et al. 2001). Collection ofreliable information for the species will improve the mitigation practices developed for HCPs. Well-designed habitat manipulation experiments provide the strongest inference to determine cause and effect relationships. Understanding of the species habitat requirements will enable the development of effective mitigation strategies. A manipulation experiment was conducted in Douglas County, Colorado (Columbine Open Space) during 2002 and 2003 to advance our understanding of PMJM habitat requirements. We manipulated sections of the riparian habitat and adjacent grassland within the I 00-year flood plain. The site was manipulated by adding patches (3 m x 2.43 m) of artificial resources (food and cover). Time limitations of only a 2-year study were inadequate for vegetation to establish and limited funding (cost of planting and sustaining vegetation) restricted this manipulation experiment to simulating habitat with temporary structures and food supplementation. The treatments were placed in areas of low use based on past monitoring studies conducted by the Colorado Division of Wildlife (CDOW) during 1998-2000. PMJM were radio tracked before and after the manipulation to determine if PMJM movements were altered through the addition of resources. We propose two primary objectives: 1) determine how the presence ofresource additions influences the distribution of individual PMJM within a population, and 2) to quantify habitat characteristics of PMJM on a microhabitat scale. We want to examine if the distribution of individual PMJM can be altered in response to the addition of resources (food and cover) and to quantify relevant microhabitat characteristics where PMJM have been detected. �3 STUDY AREA The study was conducted within the riparian habitat within Columbine Open Space, owned by Douglas County Open Space managed by the CDOW and the adjacent grassland. Columbine Open Space was selected because PMJM were monitored for 3 years by the CDOW ( 1998-2000), providing sitespecific information on PMJM locations before this manipulation experiment.· METHODS PMJM were trapped using non-folding Sherman live traps (7.6 cm x 8.9 cm x 22.9 cm) placed 5m apart along approximately 0.5 km transects adjacent to both sides of East Plum Creek for a minimum of 5 consecutive nights. Trapping procedures were in accordance with the guidelines published by the USFWS (1999). Species other than PMJM were recorded by trap location and immediately released. The following information was recorded for captured PMJM: unique identification, trap location, weight, sex, age, and reproductive condition. PMJM were scanned for a passive integrated transponder (PIT) tag. Newly captured individuals were marked by inserting a unique PIT-tag. Individuals ;::18 grams were anesthetized with isoflurane and fitted with a 1-g radio transmitter (Holohil Systems Ltd Ontario, Canada). All methods were approved by the Animal Care and Use Committee of Colorado State University (Authorization Number A3572-0l). Radio telemetry was used to monitor locations of individuals for a 21-day period, the battery life of the radio transmitters. Observers attempted to stay approximately 3 m from the radio-tagged individual to avoid influencing PMJM movement. Observations taken 3 m or greater from PMJM did not influence movement (T. Shenk, CDOW personal. comm.). The following information was recorded at each relocation: individual identification, time, weather, and surrounding vegetation. All data were combined into a geographical information system (GIS) database using ArcView®3 .2 (Environmental Systems Research Institute, Redlands, California, U.S.A.). The manipulation experiment consisted of 5 phases: 1) selection of areas oflittle or no previous use by PMJM based on CDOW location data (1998-2000) collected at Columbine Open Space, 2) recording of pre-treatment location data of radio-tagged individuals for 6 nights, 3) selection of treatment plot location based on pre-treatment and CDOW location data, 4) addition ofresources to treatment plots, and 5) recording of post-treatment location ofradio-tagged individuals. Two sessions (June and July) of the manipulation experiment were conducted each year. A digital map with a grid cell size of 9 m x 9 m was constructed for the entire study site with ArcView®3.2 (Environmental Systems Research Institute, Redlands, California, U.S.A.) software. CDOW location data was pooled into a single coverage over the grid to establish areas::: 1,000 m2 containing only low use cells (<2 locations/cell based on CDOW location data) within the 100-year flood plain. Location of treatment plots was selected with a stratified random design from a set of candidate cells meeting criteria developed to describe poor PMJM habitat (sparse vegetation and little food) within 60 m of East Plum Creek, and low historical use. The artificial cover, simulating vertical complexity, was constructed with wheat straw and tree branches distributed in a patch (3 m x 2.43 m). Burlap cloth was suspended 30 cm over the tree branches and straw. Food supplements composed of an equal mixture of whole wheat, dehydrated alfalfa pellets and sweet feed were placed on cardboard trays (0.16 m x 0.3 m) within the straw and branches as an attractant and a source of high protein. The dimensions of the treatments were selected to balance the manageability of construction and decrease the chance of inter and intra-species domination within a treatment. Quantification of microhabitat variables in areas of high use were examined by comparing a random sample of cells (9 m x 9 m) containing ::: 99 % of PMJM locations for each session to a random �4 sample of cells where no PMJM locations detected. Two line transects were randomly placed in each selected cell with 6 quadrat frames (SO cm x 20 cm) evenly distributed per line transect (Daubenmire 1959). The variables measured in each cell included percent bare ground, shrub, grass, and forb cover and vegetation composition. The location data were analyzed using linear regression. The response variable was the number of locations detected in a cell. A suite of candidate models was developed as predictors of the response variable. Akaike's information criterion (AIC) was applied to select the best "approximating" model (Burnham and Anderson 2002). The independent habitat variables of interest for the models included distance from the center of the cell to the nearest water, area and juxtaposition of nearest shrub, and presence of wetland grasses in the cell. Additional variables -included in the models were period (pre- or post-treatment), sex, session, and year. The microhabitat data collected from the Daubenmire plots were analyzed with Proc GLM (SAS 2002) to test for differences in means among areas of high use and no use by PMJM. PRELIMINARY RESULTS A comparison of microhabitat characteristics from a random sample of "high-use" and "no-use" areas indicated a greater (P < 0.0001) shrub canopy cover in "high-use" areas verses "no-use" areas (47.7% ± 29.8%, 12.6% ± 14.11 %, respectively). Further, "high-use" areas had greater basal cover (P = 0.013) and bare ground (P = 0.0459) and "no-use" areas contained a greater (P = 0.0331) abundance of forb canopy cover. We conducted a manipulation experiment where we constructed patches of artificial resources (food and cover) in areas without previous PMJM activity. PMJM were radio collared and located hourly before and after the addition of food and cover. The majority of PMJM movements were not influenced by the addition of resources in 2002. These results may be due to site fidelity or lack of exploratory movement to locate the additional resources LITERATURE CITED Bingham, B. B. and B. R. Noon. 1998. The use of core areas in comprehensive mitigation strategies. Conservation Biology 12:241-243. Burnham K. P. and D.R. Anderson. 2002. Model selection and multimodel inference. Second edition. Springer, New York, New York, USA. Daubenmire, R. 1959. A canopy-coverage method ofvegetational analysis. Northwest Science 33:4364. Harding, E., E. Crone, B. D. Elderd, J.M. Hoekstra, A. J. McKerrow, J. D. Perrine, J. Regetz, L. J. Rissler, A.G. Stanley, E. L. Walters and NCEAS Habitat Conservation Plan Working Group. 2001. The scientific foundations of habitat conservation plans: a quantitative assessment. Conservation Biology 15:488-500. Meaney, C. A. 2000. Monitoring for Preble's meadow jumping mice along South Boulder Creek and Four Ditches. Boulder, Colorado, USA. Report prepared for the Colorado Division of Wildlife. SAS Institute. 2002. SAS Version 8.2. SAS Institute, Cary, North Carolina, USA. Schorr, R. 2001 Meadow jumping mice (Zapus hudsonius preblei) on the U.S. Air Force Academy, El Paso County, Colorado, USA. Shenk, T. M. and M. Sivert. 1999. Movement patterns of Preble's meadow jumping mouse (Zapus hudsonius preblei) as they very across time and space. Annual Report to the Colorado Division of Wildlife. Fort Collins, Colorado, USA. U.S. Fish and Wildlife Service. 1999. Interim Survey Guidelines for Preble's meadow jumping mouse. U.S. Fish and Wildlife Service. Denver, Colorado, USA. �Colorado Division of Wildlife Wildlife Research Report July 2003 – June 2004 JOB PROGRESS REPORT State of Project No. Work Package No. Colorado Task No. 2 : Cost Center 3430 : Mammals Research : Preble’s Meadow Jumping Mouse Conservation Effects of Resource Addition on Preble’s Meadow Jumping Mouse (Zapus hudsonius : preblei) Movement Patterns 0662 Federal Aid Project: N/A : Period Covered: July 1, 2003 - June 30, 2004 Author: Anne M. Trainor. Personnel: T. M. Shenk, K. R. Wilson, G. C. White All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT A thesis, entitled ‘Influence of resource supplementation on movements of Preble’s meadow jumping mouse (Zapus hudsonius preblei) and habitat use characteristics,’ was completed and submitted to Colorado State University in partial fulfillment of a Master of Science degree. The thesis is available from The Colorado Division of Wildlife Library or the Colorado State University Library. Included in this report is an abstract of the thesis. 1 �JOB PROGRESS REPORT INFLUENCE OF RESOURCE SUPPLEMENTATION ON MOVEMENTS OF PREBLE’S MEADOW JUMPING MOUSE (Zapus hudsonius preblei) AND HABITAT USE CHARACTERISTICS Anne M. Trainor ABSTRACT Riparian wetlands are complex ecosystems containing great species diversity that may easily be affected by anthropogenic disturbances. Preble’s meadow jumping mouse (Zapus hudsonius preblei) is a federally threatened species dependent upon riparian wetlands. It has been the subject of habitat management and conservation efforts involving restoration and mitigation projects along the eastern Front Range of Colorado and southeastern Wyoming. Although habitat improvements for Z. h. preblei are designed for multiple spatial scales, most knowledge about the species’ habitat requirements has been described at a broad landscape scale. In addition, few projects have directly evaluated the mouse’s response to restoration and mitigation projects. The first objective of this study was to determine how supplementation using artificial resources influences the spatial movement patterns of a Z. h. preblei population. Previous studies described Z. h. preblei use areas through live trapping. This study more precisely evaluated Z. h. preblei spatial use by applying radio telemetry within a riparian ecosystem. I conducted an experiment by constructing treatment plots of artificial resources (food and cover) in areas with no previous detections of Z. h. preblei during 3 prior years (1998-2000) of intensive monitoring. Z. h. preblei were radio collared and then located hourly during nightly activity periods before and after the addition of food and cover. The second objective of this study was to improve understanding about micro-habitat characteristics that Z. h. preblei use. During the resource supplementation experiment, Z. h. preblei response to treatment plots varied by year with only 1 of 13 radio-tagged individuals using supplemental resources during 2002 and 6 of 8 in 2003. The lower use in 2002 may have been due to drought conditions, which decreased available herbaceous cover and thus protection from predators. While treatment plot use increased in 2003, the overall use was relatively low when compared to natural, high-use areas. The mean proportion of treatment plot use in 2003 was = 5.9% (SE =1.4%, range = 0% to 12%). Limited use of treatment plots may have been due to site fidelity and minimal exploratory movements by Z. h. preblei or to elevated predation risk. A comparison of micro-habitat characteristics from random samples of high-use and no-use areas indicated that areas used intensely by Z. h. preblei were closer to the center of the creek bed and positively associated with shrub, grass, and woody debris cover. Distance to center of the creek bed, percent shrub cover, and grass cover had the greatest relative importance of the habitat variables modeled in describing high-use areas. High-use areas contained three times the percent of grass cover as forb cover. There was a greater proportion of wetland shrub and grass cover in high-use versus no-use cells. However, proportion of cover type (shrub or grass) did not vary greatly between high-use and no use cells. Within riparian wetlands, the identification of key micro-habitat components that are intensively used by Z. h. preblei could improve conservation and management programs. In addition, results from the resource supplementation experiment suggest that TP pˆ mitigation and restoration may not ensure use of areas by threatened and endangered species. Therefore, understanding how species respond to changes in 2 �areas where they currently live will require development of more efficient and effective mitigation projects, and monitoring by conservation biologists and wildlife managers will be essential. Prepared by _______________________ Anne M. Trainor, Colorado State University 3 � https://cpw.cvlcollections.org/files/original/b97e5be10a7fc3f0ebee9c561b49ff4a.pdf c30aa4a6d7abfdf4fcee933246c8d0ee PDF Text Text 29 Colorado Division of Wildlife Wild.life Research Annual Report July 2000 JOB PROGRESS REPORT State of Colorado Cost Center 3430 Project No. W-153-R-13 Mammals Research Program Lynx Reintroduction Work Package No. ____0-=-67"""'0'-------Task No. l Post-Release Monitoring of Reintroduced Lynx Period Covered: July l, 1999 - June 30, 2000 Author: Tanya M. Shenk Personnel: Gene Byrne, Rick Kahn, Dave Kenvin, Jim Olterman, Scott Wait, Whitey Wannamaker, Margaret Wild, Dave Younkin ABSTRACT In an effort to reestablish a viable population of lynx (Lynx canadensis) in Colorado, 41 lynx were reintroduced into southwestern Colorado in 1999 and an additional 55 lynx released in Spring 2000. Release protocols were evaluated by closely monitoring each lynx released in 1999 through radiotelemetry. Number of mortalities and causes of each mortality were documented. With this new infonnation, release protocols were modified in an effort to release each lynx with the highest probability of survival. Three different release protocols were used in 1999. Differences in release protocol included the length of time animals were kept in the Colorado holding facility and timing of the release. Percent mortality due to starvation within six months of release date decreased with each modification of release protocols (75% under Protocol 1, 11 % under Protocol 2, 0% under Protocol 3 except for female lynx released pregnant). Of the 55 lynx released in 2000, 41 were released in April (Protocol 2) and 14 were released in May (Protocol 3) following a minimum of three weeks in the Colorado holding facility. Of the total 96 lynx released, 63 are being followed on a regular basis within Colorado, six male lynx have not been located since October 1999, one lynx possibly slipped a collar, and 26 lynx are known to have died. Known mortality factors included starvation (7), gunshot (3), vehicle collision (3), trauma (2), predation (1), and disease (l). Cause of death could not be determined for nine mortalities. Initial dispersal movement patterns of the lynx released in 1999 were extremely variable. Dispersal habitat used by the lynx released in 1999 were also highly variable, from high elevation Engelmann spruce/subalpine fir to Nebraska agricultural lands. Lynx released in 2000 have remained closer to their release site and fewer have been observed using atypical lynx habitats. Through snow-tracking efforts (221 snow-tracking days) in 1999 and 2000 we have located 147 kills, 371 beds, and 132 scats. Of the 147 kills, 75% were of snowshoe hare (Lepus americanus), 23% were pine (red) squirrel (Tamiasciurus hudsonicus), and the remaining 2% were made up of other mammals and birds. We collected 132 scat samples that will be analyzed later for content. No reproduction has been documented to date, however whether or not lynx have bred and had litters at some point remains unknown. �30 �31 POST-RELEASE MONITORING OF REINTRODUCED LYNX Tanya M. Shenk P. N. OBJECTIVE I. Post release monitoring of lynx released in Colorado. SEGMENT OBJECTIVES FY99-00 2. 3. 4. 5. Estimate first year survival rates oflynx reintroduced to Colorado. Identify first year mortality factors of lynx reintroduced to Colorado. Describe first year movement patterns of lynx reintroduced to Colorado. Refine and prioritize needed research components to develop sound management strategies for lynx in Colorado. 6. Prepare a Federal Aid Job Progress Report. PERFORMANCE INDICATORS FY99-00 I. 2. 3. 4. 5. 6. 7. 8. Survival estimates of lynx reintroduced to Colorado. Summary of mortality factors oflynx reintroduced to Colorado. Description and analysis of habitats used by lynx reintroduced to Colorado. Description of movement patterns for lynx reintroduced to Colorado. Reproduction estimates of lynx reintroduced to Colorado. Evaluation and modification of release protocols for reintroducing lynx. Sites selected for second year release of lynx Refinement and prioritization of needed research components to develop sound management strategies for lynx in Colorado. 9. Report on first year release of lynx in Colorado: movement patterns, habitat use, survival and reproduction INTRODUCTION In an effort to reestablish a viable population of lynx (Lynx canadensis) to Colorado, 41 lynx were reintroduced into southwestern Colorado in the spring of 1999 and an additional 55 lynx were released in Spring 2000. Monitoring of these lynx is crucial to evaluating the progress of this lynx reintroduction effort. The monitoring program will also provide information and data critical to improving release techniques to ensure the highest probability of survival for each individual lynx released in future years of the Colorado effort, and perhaps in other reintroduction efforts. The post-release monitoring program for the reintroduced lynx has two primary goals. The first goal is to obtain regular locations of released lynx. From these locations we will be able to determine how many lynx remain in Colorado and their locations relative to each other. Given this information and knowing the sex of each individual we will be able to assess the feasability of these lynx to form a breeding core from which a viable population might be established. Also from these data we can describe general movement patterns and habitats used. The second primary goal of the monitoring program is to estimate survival of the reintroduced lynx and, where possible, determine cause of mortality of reintroduced lynx. �32 Additional goals of the post-release monitoring program for lynx reintroduced to the southern Rocky Mountains include refining descriptions of habitat use and movement patterns, determining food habits, and obtaining information on reproduction. When the lynx establish home ranges that encompass their preferred habitat, more emphasis will be placed on refining descriptions of movement patterns and habitat use. Lynx is currently a species listed as threatened under the Endangered Species Act (ESA) of 1973, as amended (16 U.S. C. 1531 et. seq.)(U. S. Fish and Wildlife Service 2000). As a listed species, information specific to the ecology of the lynx in its southern range such as habitats used, movement patterns, mortality factors, survival, and reproduction in Colorado will be needed to develop recovery goals and conservation strategies for this species specific to its southern range. Thus, an additional objective of the post-release monitoring program is to develop conservation strategies relevant to lynx in Colorado. OBJECTIVES The initial post-release monitoring of reintroduced lynx will emphasize five primary objectives: 1. Assess and modify release protocols to enure the highest probability of survival. 2. To obtain regular locations of released lynx to describe general movement patterns and habitats used by lynx. 3. Determine causes of mortality occurring in reintroduced lynx. 4. Estimate survival of lynx reintroduced to Colorado. 5. Estimate reproduction of reintroduced lynx. Three additional objectives will run concurrently or become active after lynx become established in an area that encompasses their movements. These objectives include: 6. Better refine descriptions of habitats used by reintroduced lynx. 7. Better refine descriptions of daily and overall movement patterns of reintroduced lynx. 8. Describe food habits and prey of reintroduced lynx. The data collected during the post-release monitoring will be analyzed to evaluate habitat use, movement patterns, reproduction and survival. These data will be used to further the knowledge about habitat requirements for this species in the southern Rocky Mountains. Thus, the final objective for the postrelease monitoring plan is to: 9. Refine habitat protection recommendations and conservation strategies based on information collected from released lynx. STUDY AREA Five areas throughout Colorado were evaluated as potential lynx habitat (Byrne 1998). Criteria investigated in these five areas for comparison were (1) relative snowshoe hare densities (Reed at al., unpublished data), (2) road density, (3) size of area, (4) juxtaposition of habitats within the area, (5) historical records oflynx observations, and (6) public issues. Based on results from this analysis, the San Juan Mountains of southwestern Colorado were selected as the.release area for reintroducing lynx. Ten release sites within the San Juan Mountains were selected based on land ownership and accessability during time ofrelease for the 41 animals released in 1999. Of the 55 lynx released in Spring 2000, 45 were released at Rio Grande Reservoir and ten lynx were released at three sites west of the Continental Divide. Based on current locations of the majority of the released lynx, the core research area remains in the southern San Juan Mountains, however lynx may need to be captured from areas of non-suitable habitat in Colorado and adjacent states. �33 METHODS Assessment ofRelease Protocols A total of 41 lynx were released in 1999 at selected areas in the San Juan Mountains of southwestern Colorado. Prior to release each lynx was examined and age, sex, and body condition determined. Each lynx was fitted with a TelonicsTM VHF radio-collar for post-release monitoring. The collars were also equipped with a mortality switch that activates if the collar remains motionless for a period of four hours or more. Specific release sites were selected based on land ownership and accessability during times of release. Lynx were transported from the holding facility to the release site in individual cages. Release site location was recorded in Universal Trans Mercator (UTM) coordinates and identification of all other lynx released at the same location, on the same day, was recorded. Behavior of the lynx on release and movement away from the release site was documented. Monitoring of the survival and mortality factors (see below) of each lynx was used to modify release protocols in 1999 in an attempt to release each lynx with the highest probability of survival. Release protocols for the 55 lynx released in 2000 were developed from survival, mortality factor, and movement pattern data obtained from lynx released in 1999. Documenting Movement Patterns To obtain regular locations of released lynx to determine general movement patterns and habitats used by reintroduced lynx a combination of satellite, aerial and ground radio-tracking were conducted. Locations and general habitat descriptions of each location were recorded and mapped for all locations. All 41 of the lynx released in 1999 were monitored from the air through radio-tracking. Frequent flights (three times a week) were critical during the initial post-release periods because of the greater likelihood of dispersal and mortality in reintroduced carnivores. Every effort was made to locate every lynx during each flight during this period. Sixty days from the date of the last re~ease, aerial locations of the radio-collared lynx were to be determined two times per week for the remainder of the life of the transmitters. Flights were also conducted three times per week, weather permitting, to locate lynx during the snow-tracking field season (December through April) to aid in the snow-tracking efforts. When possible at least one observer flew with the pilot to become familiar with the terrain, to operate the radio telemetry receiver, and to record the global positioning system (OPS) locations of the lynx. Generally,- the pilot circled a strong telemetry signal and then bisected the circle activating the OPS unit when approaching directly overhead. The date and time of the beginning and ending of the flight, the time each collar was located, the UTM coordinates for each animal located, general weather conditions, primary overstory vegetation type, and name of the personnel were recorded. All locations were entered into a database for mapping and data analysis. Fifty-one of the 55 lynx released in spring 2000 were fitted with SirtrackTM dual VHF/satellite transmitter collars. The remaining four lynx were fitted with TelonicsTM VHF collars identical to those used on lynx released in 1999. Each dual collar weighed 137-156 grams. The satellite component of each collar is programmed to be active for 12 hours per week. The 12-hour active periods are staggered throughout the week, with approximately seven collars being active each day of the week. Signals from the collars allow for locations of the animals to be made via Argos, NASA, and NOAA satellites. The location information was processed by ServiceArgos and distributed daily to the Colorado Division of Wildlife through e-mail messages. Both the VHF and satellite transmitter in the dual collar has a mortality switch which is triggered by four or more hours of stationarity. Determining Causes of Mortality To determine causes of mortality occurring in reintroduced lynx every effort was made to locate and retrieve carcasses of dead lynx as soon as possible. When a mortality signal (75 ppm vs 50 ppm for the TelonicsTM VHF transmitters, 20bpm vs 40bpm for the SirtrackTM VHF transmitters, 0 activity for �34 Sirtrack™ PIT) was heard during either satellite, aerial or ground surveys, the location (UTM coordinates) was recorded. Ground crews located and retrieved the carcasses. The immediate area was searched for evidence of other predators and the carcass photographed in place before removal. Additionally, the mortality site was described, habitat associations, and exact location were recorded. Any scat found near the dead lynx that appeared to be from the lynx was collected. All carcasses were transported immediately to the Colorado State University Veterinary Hospital for a post mortem exam. Lynx carcasses were not frozen but kept cool. If carcasses were already frozen due to field conditions, this was noted on the field form. The objectives of the post-mortem examination were to 1) determine the cause of death and document with evidence, 2) collect samples for a variety of research projects, and 3) archive samples for future reference (research or forensic). The gross necropsy and histology were performed by, or under the lead and direct supervision of a board certified veterinary pathologist. At least one research personnel from the Colorado Division of Wildlife involved with the lynx program was also present. In general, the protocol followed standard procedures used for thorough post-mortem examination and sample collection for histopathology and diagnostic testing. Some additional data/samples were routinely collected for research, forensics, and archiving. Other data/samples were collected based on the circumstances of the death (e.g., photographs, video, radiographs, bullet recovery, samples for toxicology or other diagnostic tests, etc.,). The CDOW retained all samples and carcass remains with the exception of tissues in formalin for histopathology, brain for rabies exam, feces for parasitology, external parasites for ID, and other diagnostic samples. Estimating Survival Survival rates of lynx reintroduced to Colorado will be estimated using the Kaplan-Meier method with staggered entries (Pollock et al. 1989). Documenting Habitat Use and Hunting Behavior More refined descriptions of habitats used by reintroduced lynx were obtained through snow-tracking of animals. Data were collected on habitats used, daybed and hunting bed locations, and travel corridors. Hunting and feeding behavior information was also collected by documenting prey taken, prey chases, relative abundance of prey (tracks and sightings), and use of carrion. Snow-tracking was conducted during February-May, 1999 and beginning again in November, 1999 through April 2000. Locations from the aerial-tracking were used to help ground-trackers "locate lynx tracks in the snow. One or more persons working together conducted the snow-tracking surveys. Snowmobiles, where permitted, were used to gain the closest possible access to the lynx tracks without disturbing the animal. From that point, snowshoes were used by the tracking team to reach the tracks. Once tracks are found, the ground crew back-tracked the animal. Back-tracking avoided the possibility of disturbing the lynx by moving away from the animal rather than toward the animal. However, monitoring of the lynx through radiotelemetry was also used to assure that ground crews stayed a sufficient distance away from the lynx in the event the lynx might double back on its tracks. If the lynx began to move in response to the observers, the observers retreated. If the lynx began to move and the movement did not appear to be a response to the observers, the crew continued to follow and record locations, habitats used, and behavioral information for as long as possible. Locations oflynx tracks were recorded using a Garmin XL12 GPS and 7.5° topographic map. Habitat descriptions included overstory and understory vegetation and seral stage. Locations and behavioral observations that could be interpreted from the tracks (e.g., chases, scent marking) were recorded. These data will be used for mapping and spatial analyses and analyzed to make inferences on how different habitats are used, frequency of use, daily movement patterns, hunting areas, daybed locations, den sites, and travel corridors. Data will also be used to document any changes in habitat use as animals begin to settle into a home range. �35 An attempt was made to locate tracks from all lynx. However, first priority was given to locating any animal that appeared to be consistently in the same location from aerial surveys. Such stationarity may indicate an injured, starving, or otherwise traumatized animal. Data on hunting behavior was collected by location of kills, food caches, chases, and through scat analysis. Prey from attempted and successful hunting attempts were identified by either tracks or prey remains. Information from scat analysis will also provide information on foods consumed. Scat samples were collected wherever found, recording location and individual lynx identification. Only part of the scat was collected, the remainder was left where found so as not to interfere with the possibility the scat was being used by the animal as a territory mark. Comparisons of food composition and percent occurrence will be made within and among individuals. Analyses of temporal, spatial, and individual differences will be conducted to provide information on feeding ecology of reintroduced lynx in the southern Rocky Mountains. Estimating Reproduction Reproductive status of all female lynx was determined prior to release through radiographs. All females known to be pregnant or thought to possibly be pregnant on release were monitored closely from their release through the following August to determine reproductive success. Females remaining within a limited area immediately after release through August were located and observed to look for accompanying kittens or a den site. Females that had been released in 1999 and were alive in spring 2000 were monitored for proximity to males during breeding season and for site fidelity to a given area during the denning period of May and June. Each female lynx from the 1999 releases were directly observed in summer 2000 over 3-5 different visits to look for accompanying kittens or evidence of denning. Locations of both males and females released in 1999 were evaluated during March and April 2000 to document proximity of males to females in an attempt to determine if breeding could have occurred. RESULTS Assessment ofRelease Protocols A total of 41 lynx were released in Colorado in 1999 under five different release protocols (Table I). The initial release protocol called for the immediate release of females once they passed veterinary inspection in Colorado. Males were to be held for a period of weeks until females established a territory, and then males were to be released near female territories. Four animals were released in early February, however, three of these died of starvation within six weeks of their release and the fourth was recaptured and returned to the holding facility where she recovered and was later re-released (Table 2). Reevaluation on the condition of animals released under the first protocol suggested that these animals may not have been in optimal physical shape when released. Therefore, a second release protocol was initiated whereby lynx were held at the Colorado holding facility for a minimum of three weeks and fed high quality diets to encourage weight gain. Most lynx gained considerable body weight while in captivity (Wild 1999). Nine lynx were released under this second protocol (Table 2). Of these nine lynx, one juvenile female died of starvation seven weeks after release. After the starvation death of the first lynx under the second protocol, a third release protocol was developed that called for releasing all subsequent lynx in the spring after a minimum stay in the holding facility of at least three weeks (Table I). A spring release would assure the lynx were released when prey was most abundant (i.e., young of the year would be most abundant and hibernating and migratory prey would be available). Twenty lynx were released under this protocol (Table 2). Additionally, six females were released under this third protocol that were known to be pregnant (Protocol 3P) and two that were possibly pregnant (3P?). No lynx reintroduced under Protocol 3 died of starvation within six months postrelease (Table 2). However, two of the six lynx released when pregnant died of starvation within six months post-release. �36 An assessment of the fates of each lynx under all five release protocols used in 1999 led to release protocols for lynx released in 2000. Release protocols 2 and 3 resulted in the fewest post-release (up to six months after release date) starvation mortalities (Table 2). The common element in both protocols was increased captivity time in the Colorado holding facility. The single starvation mortality for lynx released under Protocol 2 in 1999 was also the only juvenile released under that protocol and the only animal released in February (the other eight Protocol 2 lynx were released in March 1999). Thus, all lynx released in 2000 were released under either Protocol 2 or 3 but not before April I. Because of the high percentage of starvation mortalities in females pregnant on release (Table 2), we also attempted to avoid reintroducing lynx that were known to be pregnant. This was best accomplished by trying to have animals captured for the reintroduction effort in Canada and Alaska prior to their breeding season. Movement Patterns Through extensive aerial and satellite tracking, we continue to search and locate 63 of the 70 lynx with collars on and assumed to be alive (one lynx has presumably slipped her collar). We have 1206 satellite locations for 49 of the 51 lynx fitted with dual collars (2 satellite collars never worked after the lynx were released) and 1122 aerial VHF locations for all 96 reintroduced lynx. Six males from the 1999 releases have not been found since at least 1 October 1999. Possible reasons for not locating these six males include (I) long distance dispersal, beyond the areas currently being searched, (2) radio failure, or (3) destruction of the radio (e.g., run over by car). We continue to search for all missing lynx during both aerial and ground searches. Last known locations for each of the 70 lynx assumed to be alive are presented in Figure I. Initial dispersal movement patterns of the lynx released in 1999 were extremely variable. Dispersal habitat used by lynx released in 1999 has been highly variable, from high elevation Engelmann spruce/Subalpine fir to Nebraska agricultural lands. However, numerous travel corridors have been used repeatedly by more than one lynx, possibly suggesting route selection based on olfactory cues. Dispersal movement patterns of lynx released in 2000 were much less than those observed by lynx released in 1999. Most of 2000 releases have remained within an area encompassed by 100 km radius circle from the release locations. Most movement away from this core area has been to the north (Figure I). We currently have six lynx using areas near Interstate 70. Survival and Mortality Factors - Of the 96 lynx released, 26 mortalities have been recorded to date (Table 3). From the 1999 releases (41 animals) we have had 22 known mortalities (6 from starvation, 8 unknown, 3 gunshot, 2 hit by car, 2 trauma, and 1 predation). We have six missing males. We are following 13 of the lynx from the 1999 releases on a regular basis. From the 2000 releases (55 animals) we have four known mortalities (1 hit by car, 1 disease, I starvation, 1 unknown) and one animal that possibly slipped her collar. We are following the remaining 50 animals on a regular basis. Of the total seven confirmed starvation deaths, three were associated with animals released in less than ideal body condition and two were lynx less than one year old. Percent mortality due to starvation decreased with each modification of release protocols (75% under Protocol 1, 11 % under Protocol 2, 0% under Protocol 3). Necropsy results for lynx BC00F3, a female released on April 2, 2000 near Creede, Colorado indicated she died from pneumonic plague. The lynx was in fairly good condition, there was some abdominal fat, no muscle wasting, and the bone marrow had fat in it. The only gross lesion was an acute fibrinous pneumonia (i.e., lung infection of short duration). The lynx had probably only been sick a few days before it died. The carcass was recovered near her release site. Plague was diagnosed by flourescent antibody test and isolation of Yersinia pestis from lung and spleen samples. �37 Recaptures Three lynx have been recaptured and subsequently re-released since their initial release. Lynx BC99F6 was released in 1999 under Protocol l. Her behavior and incidental sightings by the public suggested the lynx was in poor condition. We trapped her using a TomahawkTM live trap baited with rabbit. She was recaptured the first night (March 25, 1999) we set the trap. On capture, we found she was severely emaciated. We anesthetized her with Telezol (2 mg/kg) and returned her to the Colorado holding facility. She was rehabilitated through diet. The lynx gained weight steadily and was re-released on May 28, 1999. She was hit by a car on Interstate 70 on July 19, 1999. Necropsy results indicated she was in excellent body condition at her time of death. Lynx AK.99M9 was released on May 12, 1999 and recaptured on March 24, 2000. Field observations by the lynx monitoring crew suggested that the lynx was severely emaciated. Live-trapping the lynx failed, so the lynx was darted with Telazol (3 mg/kg) using a Dan-Inject CO2 pistol. Physical examination revealed severe emaciation (6 kg). The lynx was returned to the Colorado holding facility and rehabilitated through diet. The lynx gained weight steadily and was re-released on May 3, 2000. Lynx AK.99F2 was released on May 7, 1999 and recaptured on April 18, 2000. Field observations by the lynx monitoring crew suggested that the lynx was emaciated. She was live-trapped with a TomohawkTM live trap with one nights effort. On capture, we found she was emaciated. We anesthetized her with Telezol (2 mg/kg) and returned her to the Colorado holding facility. She was rehabilitated through diet. The lynx gained weight steadily and was re-released on May 22, 2000. Habitat Use and Hunting Behavior February 1999-May 1999 Through snow-tracking, we were able to document habitat use, daily movement patterns, and hunting behavior of the earlier released lynx. Snow-tracking of lynx began shortly after the first release, Feb. 6, and continued until May 15, 1999. Although we tried to continue beyond May 15, efforts beyond this date did not yield any information because of either the lack of snow in the areas where the lynx were, or the snow conditions were too difficult to track in (hard, crusty, patchy). Because the majority (28) of the lynx were first released under Protocol 3, after May 6, the snow-tracking effort focused on the 13 lynx released prior to this date, under Release Protocols 1 and 2. Approximately 114 km of lynx tracks were followed. These tracks were from 11 different lynx, with kilometers tracked for any individual varying from 1 to 31 kilometers (Table 4). Two lynx (one female and one male) from Release Protocols 1 and 2 were never snow-tracked because we were either not able to locate the animals or because when we did locate them we could not readily access where they were. Daybeds and hunting beds were each located for eight of the lynx. Prey chases or kills were found for four lynx, scat samples were collected from five lynx, and possibly from a sixth. From the kills found and from initial examination of the scat samples, the lynx fed on snowshoe hare (Lepus americanus), pine (red) squirrel (Tamiasciurus hudsonicus), and waterfowl. All the snow-tracking effort was conducted on nine lynx released under Protocols 1 and 2. Any lynx released under Protocol 3 were released too late to track. November 1999 -April 2000 Ground crews tracked 13 of the lynx released in 1999 during this period (Table 4). Two other lynx were being located during this time but were not in snow. A total of 139 kills or chases were located, 75% were snowshoe hare, 23% were pine (red) squirrel, and the remaining 2% were made up of other mammals and birds. We collected 115 scat samples that will be analyzed for content. Lynx released in 2000 were released too late to snow track. �38 Reproduction Six lynx released under Protocol 3 in 1999 were known to be pregnant (Table 1, Release Protocol 3P). Two other females may have been pregnant, the radiographs were suggestive but inconclusive (Table 1, Release Protocol 3P?). Three of the six lynx known to have been pregnant on release in 1999 died within two months after release. Two starved and one was killed on the road (Table 2). Long distance movements and lack of stationarity in the movement patterns of the other three lynx known to have been pregnant on release in 1999 suggests these females did not have young with them by July 1999. Of the two females that might have been pregnant, movement patterns were not suggestive of a female rearing young. It is not known if any other females bred and/or had young once released, however no females snow-tracked November 1999 through April 2000 had young with them. From radiographs taken of the 35 females released in 2000, one female was known to be pregnant and three were possibly pregnant. Movement patterns suggest that none of these females have kittens with them as of July 2000. There were seven females released in 1999 that were alive during the Spring 2000 breeding season. All seven females were in close (< 5 km) proximity to a male during the breeding season and could have bred. The seven females were monitored closely for stationary movement patterns, indicative of denning, from May-July 2000. Ground trackers also walked in on all seven females for visual observations on a minimum of three occasions and two females were visited on five occasions. No kittens were observed. However, the question of whether they successfully bred or had kittens at some point in 2000 is unknown. One of these females has since died and three others have made movements of over 100 km. Although we are confident none of the six live females have kittens at this time, for further confirmation we will snow-track each of these females as soon as they are in areas with fresh snow to check for kitten tracks. Beginning in March 2000 both male and female lynx began to exhibit extensive movements (> 100 km) away from areas they had used throughout the winter. For example, female (AK.99F3) moved from the area near Grizzly Gulch she used throughout the winter to the Wolf Creek Pass area, a straight line distance of approximately 255km (Figure 3). Male YK99M3 moved from the area near the Climax mine which he had used throughout the winter to Taylor Mesa, a straight line distance of approximately 270km (Figure 4). Such movements by both females and males put them in close(< 5 km) proximity to a lynx of the opposite sex. Two isolated males did not move during March or April and thus were not in close proximity to a known female during reeding season. This was a male that had used the area in and adjacent to the northwest comer of Rocky Mountain National Park and a male that used the area around Cuchara, Colorado throughout the winter. DISCUSSION Monitoring of lynx reintroduced to southwestern Colorado is crucial to evaluating the progress of the lynx reintroduction. Monitoring of these released lynx provides information and data necessary for improving release techniques to ensure the highest probability of survival for each individual lynx released in future years, and perhaps in other areas. Lynx is currently a species listed as threatened under the ESA. Information collected on the progress of the lynx reintroduction program, including habitats used, movement patterns, mortality factors, survival, and reproduction, could also be used to help develop recovery goals and conservation strategies for this species specific to its southern range. Three release protocols were used in the reintroduction of lynx to Colorado in 1999. Release protocols were modified as new information became available from monitoring the released lynx through radio-telemetry and snow-tracking. Each modification of the release protocols decreased the percent of animals dying from starvation. The primary element in later, more successful release protocols was an increased time in captivity at the Colorado holding facility. Increasing the amount of time lynx were held in �39 the Colorado holding facility provided each lynx with an opportunity to increase body weight and acclimate to the climate, elevation, and local conditions of the environment they would be released into. Although most lynx were housed in individual pens, with a few sharing a pen with one other lynx, the holding facility also allowed the lynx to hear and smell each other throughout this acclimation period. Such contact may have provided time for social interactions to occur. Such social interactions. may improve the likelihood these animals could form a breeding population. Post-release monitoring provided preliminary information on habitat use specific to Colorado that might later be used to refine habitat protection and management recommendations specific to Colorado. However, caution must be used in interpreting the information collected to date on habitats used by the introduced lynx. The aerial locations and snow-tracking results do provide some information but may also reflect behavior of displaced animals. General observations to note may be repeated use by multiple lynx of certain travel corridors and lack of use of tundra areas for any length of time. Both these habitat use characteristics have been noted for naturally occurring lynx populations. Preliminary data collected on kills suggests the reintroduced lynx are feeding on their preferred prey species, snowshoe hare and pine (red) squirrel in similar proportions as those reported for northen lynx during lows in the snowshoe hare cycle (Aubry et al., 1999). Caution must be used in interpreting the proportion of identified kills. Such a proportion ignores other food items that are consumed in their entirety. Through snow-tracking we have evidence that lynx are mousing and several of the fresh carcasses have yielded small mammals in the gut on necropsy. Nearly all the scat samples collected have been found through snow-tracking efforts and thus are representative of winter diet only. However, the summer diet of lynx has been documented to include less snowshoe hare and more alternative prey than in winter (Mowat et al., 1999). The extreme movements observed by both females and males in March and April 2000 may have been related to breeding behavior. March and April are the natural breeding periods for northern lynx (Tumlison 1987). We do not know if any of the females bred or had kittens but we are fairly sure that no female has kittens at this time. With only seven females from the 1999 releases in the wild in spring 2000 it was not unexpected that there might not be successful reproduction in 2000. During the summer of 2000, some lynx that were released in 1999 and had been faithful to a given area have made large movements away from these areas. Extensive summer movements away from areas used throughout the rest of the year have been documented by native lynx in Wyoming and Montana (Squires and Laurion 1999). Proposed monitoring and research include continued aerial radiotelemetry to document current locations and movement patterns, documentation of mortalities and causes of death, use of snow-tracking to document habitat use and hunting behavior, and further assessment of snowshoe hare densities in the state. The habitats used by the lynx will continue to be identified, mapped, and analyzed. These data will be used to further the knowledge about habitat requirements and preferences for this species in the southern Rocky Mountains. This information will be used to identify other blocks of potential habitat located throughout the Southern Rocky Mountains and evaluate conflicts that might jeopardize the recovery of lynx in Colorado. If conflicts are identified, such information can be used to develop conservation strategies and recommend land management strategies to mitigate them. ACKNOWLEDGMENTS The Colorado lynx reintroduction program and post-release monitoring is a large project involving many people. John Mumma, former director of the Colorado Division of Wildlife was instrumental in the implementation of the program. Rick Kahn of the CDOW is the program leader. Many CDOW biologists, researchers, wildlife managers and other personnel are involved in the program and or have advised us in the development of the monitoring protion of the program including Bill Andree, Tom Beck, Gene Byrne, Bruce Gill, Dave Kenvin, Todd Malmsbury, Jim Olterman, Dale Reed, John Seidel, Scott Wait, Margaret �40 Wild. The Lynx Advisory Team members from outside the CDOW include Steve Buskirk, Jeff Copeland, Dave Kenny, Steve King, John Krebs, Brian Miller, Gary Patton, Jerry Mastel, Kim Poole, Rob Ramey, Rich Reading, John Weaver, and Mike Wunder. We thank Susan and Herman Dieterich of the Frisco Creek Wildlife Rehabilitation Center for the care and maintenance of the lynx while being held in Colorado. The aerial post-release monitoring has been conducted by state pilots including Dell Dhabolt, Jim Olterman, Matt Secor, Whitey Wannamaker, and Dave Younkin. Ground field crew members include Bob Dickman, Chris Parmater, Jake Powell, and Jennifer Zahratka. Jon Kindler and Anne Trainor of CDOW were most helpful in preparation of the maps. Funding has been provided by Vail Associates, Turner Foundation, Great Outdoors Colorado (GOCO), and the Colorado Division of Wildlife. LITERATURE CITED Aubry, K. B., G. M. Koehler, and J. R. Squires. 1999. Ecology of Canada lynx in southern boreal forests. in Ecology and Conservation of Lynx in the United States. General Technical Report for U.S. D. A. Rocky Mountain Research Station. University Press of Colorado. Byrne, G. I 998. Core area release site selection and considerations for a Canada lynx reintroduction in Colorado. Report for the Colorado Division of Wildlife. Mowat, G., B. G. Slough, and S. Boutin. 1996. Lynx recruitment during a snowshoe hare population peak and decline in southwest Yukon. Journal of Wildlife Management 60:441-452. Mowat, G. and B. G. Slough. 1998. Some observations on the natural history and behaviour of the Canada lynx, Lynx canadensis. Canadian Field Naturalist 112: 32-36. Mowat, G., K. G. Poole, and M. O'Donoghue. 1999. Ecology oflynx in northern Canada and Alaska. in Ecology and Conservation of Lynx in the United States. General Technical Report for U.S. D. A. Rocky Mountain Research Station. University Press of Colorado. Nava, J. 1970. The reproductive biology of the Alaska lynx. M.S. Thesis University of Alaska, Fairbanks. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of wildlife management 53: 7-15. Poole, K. G., G. Mowat, and B. G. Slough. 1993. Chemical immobilization oflynx. Wildlife Society Bulletin 21:136-140. Seidel, J., B. Andree, S. Berlinger, K. Buell, G. Byrne, B. Gill, D. Kenvin, and D. Reed. 1998. Draft strategy for the conservation and reestablishment of lynx and wolverine in the southern Rocky Mountains. Report for the Colorado Division of Wildlife. Slough, B. G. 1999. Characteristics of Canada lynx, Lynx canadensis, maternal dens and denning habitat. Canadian Field Naturalist 113:605-608. Squires, J. R. and T. Laurion. 1999. Lynx home range and movements in Montana and Wyoming: preliminary results. in Ecology and Conservation of Lynx in the United States. General Technical Report for U.S. D. A. Rocky Mountain Research Station. University Press of Colorado. Tumlison, R. 1987. Mammalian Species: Fe/is lynx. American Society of Mammalogists. U.S. Fish and Wildlife Service. 2000. Endangered and threatened wildlife and plants: final rule to list the contiguous United States distinct population segment of the Canada lynx as a threatened species. Federal Register 63, Number 58. Wild, M.A. 1999. Lynx veterinary services and diagnostics. Job Progress Report for the Colorado Division of Wildlife. Fort Collins, Colorado. �41 Table 1. Release protocols for l)'TIX released in southwestern Colorado in 1999. Protocol 1 Description Release females as soon as they pass veterinary inspection in Colorado. Release males once females appear to have settled into an area. 2 Release males or females after they have been held in Colorado holding facility for a minimum of 3 weeks. During this holding period, the lynx were fed high quality diets to encourage weight gain, assuring each l)'TIX would be released in optimal physical condition. Such a minimal holding period also provided an opportunity for the l)'TIX to acclimate to the climate, elevation, and local conditions of the environment they would be released into. Although most l)'TIX were housed in individual pens, with a few sharing a pen with one other l)'TIX, the holding facility allowed the l)'TIX to hear and smell each other throughout this acclimation period. Such contact may also have provided time for social interactions to occur. 3 All l)'TIX to be kept in the holding facility for not only the minimal three week period but until spring. A spring release would assure the l)'TIX were released when prey was most abundant (i.e., young of the year would be most abundant and hibernating prey would be available). Coupled with the minimum holding period of three weeks, these lynx would also be released when in optimal physical condition and after a period of acclimation to their new surroundings. 3P Pregnant females released under Protocol 3. 3P? Possibly pregnant females released under Protocol 3. Table 2. Summary of number of l)'TIX released under each release protocol and numbers of lynx mortalities six months post-release for lynx released into southwestern Colorado in 1999 and four months post-release for l)'TIX released in 2000. 1999 2000 Protocol 1 2 3 3P 3P? Total Number released Mortalities 6 months post-release (n, %) Number released Mortalities 4 months post-release (n, %) Female Male Starvation Other Female Male Starvation Other 3 3 8 6 2 22 1 6 12 3, 75% 1, 11% 0, 00/4 2,33% 0, 00/4 6, 14% 0,0% 0,0% 3, 15% 2,33% 0, 00/4 5, 12% 0 25 6 l 3 35 0 16 4 1,2% l, 100/4 0, 00/4 0,0% l, 1% 3, 7% 0,0% 0,0% 0,0% 3, 5% 19 20 �42 Table 3. Release and mortality information for lynx released into southwestern Colorado in 1999 and 2000. Mortality Infonnation Release Infonnation Animal ID Sex Age Date Site Protocol Date Cause of death BC99Ml M 8mo 2/4/99 Goose Creek 1 2/24/99 staivation BC99F9 BC99F7 BC99F8 AK99F4 AK99M23 BC99F6 AK99Fl7 AK99F8 AK99Fl8 AK99Fl0 BC99M2 AK99F27 AK99M6 AK99Fl5 YK99F4 AK99Mll BC00F3 YKOOM5 YK99F3 YK99M6 AKOOF4 AK99Fl3 YKOOF17 BC99Ml0 AK99F25 YKOOF6 F F F F M F F F F F M F M F F M F M F M F F F M F F 2+ 3+ 9mo 1-2 1-2 2+ 2-3 5+ 1-2 l0mo 4+ lOmo 5 2-3 4-5 2-3 1 lOmos 2 3 lOmos lOmo 1 3-4 lOmo 2 2/3/99 2/3/99 2/20/99 5/7/99 5/14/99 2/4/99 5/10/99 5/10/99 5/14/99 5/12/99 3/19/99 5/14/99 5/13/99 5/14/99 5/13/99 5/12/99 4/2/00 4/2/00 5/10/99 5/13/99 5/22/00 5/12/99 4/17/00 3/19/99 5/7/99 4/2/00 Goose Creek Goose Creek Red Mtn Creek Sand Bench Love Lake Goose Creek First Fork First Fork Love Lake Lemon Res Red Mtn Creek Love Lake Vallecito Res Love Lake Vallecito Res Lemon Res Goose Creek Beaver Meadows First Fork Vallecito Res Rio Grande Res Lemon Res Rio Grande Res Red Mtn Creek Sand Bench Rio Grande Res 1 1 2 3p 3 1 3p 3p 3 3p 2 3 3 3 3 3 2 2 3 3 3 3 2 2 3 2 2/26/99 3/16/99 4/10/99 6/13/99 6/18/99 7/19/99 7/22/99 7/30/99 8/25/99 9/13/99 10/20/99 10/31/99 11/16/99 11/24/99 1/25/00 1/29/00 5/24/00 5/25/00 6/7/00 6/19/00 6/19/00 6/22/00 7/29/00 8/2/00 8/10/00 8/17/00 staivation staivation staivation staivation shot hit by car hit by car staivation trawna, emaciation wucnown. not staivation wucnown, not staivation shot shot blunt trauma predation, emaciation wucnown pneumonic plague staivation wucnown. not staivation wucnown slipped collar? wucnown wucnown. not staivation wucnown wucnown. not staivation hit br car Table 4. Habitat use and hunting behavior as described by summarizing kills, mousing activity, territory marks, hunting beds, day beds, and chases for each lynx tracked. Total number of days tracked to date and number of scat samples collected are also summarized. Data presented here are for the 1999-2000 snowtracking field season. Tracking Period No. of lynx tracked Kills Beds Scats Tracking Days Feb 99 - May 99 11 8 71 17 84 Nov 99 - Apr 00 13 139 300 115 137 �. -- _J --- ---- . - I --- -- -- - - _/ .,l_ .. ,--, - -, WYOMING • I, ---r '- 7 I i-r-~';'l---.....L i I --··t~ - 7, ., __ !_ ,J ' -- ·•·-. ,--.,_I • /t£WMEXICO • i I I /-------/ j -- ! s Figure 1. Most recent locations, as of August 17, 2000, of the 70 lynx known or assumed to be alive from the 96 lynx reintroduced to southwestern Colorado in 1999 and 2000. Black triangles indicate last known VHF location, black circles indicate last known satellite location. Black lines are Colorado highways, grey lines are county boundaries. Each location is identified with the animal code. �! __ L _ _ ___ : j -1 l ! ! l- - -- J - . ------ ~ i NEBRA~KA WfOMING i ---~r· '' I ,I :~. 't! -- -- -~ --- I I -- -- - -1 -------- - - --/~EW::o I i ? i • ; I 'I l ? <._ \ i Figure 2. Locations of all 26 known lynx mortalities from the 96 lynx reintroduced to southwestern Colorado in 1999 and 2000. Different symbols indicate different causes of death: starvation (e), hit by car(+), predation(*), disease (A), gunshot(*), and unknown(?). Dark lines within the Colorado border are highways, grey lines are county lines. �NEBRASKA WYOMING I~ NEW MEXICO ------- ·-·-;------ .,.r---.L~ s Figure 3. Movements of lynx female AK.99F3 from her release to August 2000. Smallest circles are oldest locations with circles increasing in size as date becomes more recent. Largest circle is most recent location. Black lines are Colorado highways, grey lines are county boundaries. �NEBRASKA WYOMING ·1-----(~~~- -~\ f- NEW MEXICO Figure 4. Movements of lynx male YK99M3 from his release to August 2000. Smallest circles are oldest locations with circles increasing in size as date becomes more recent. Largest circle is most recent location. Black lines are Colorado highways, grey lines are county boundaries. �7 Colorado Division of Wildlife Wildlife Research Report July 2001 and July 2002 PROGRESS REPORT State of_._ _ _ _ _----=Ca.ao=lo=r=a=do.:.,__ _ _ __ Division of Wildlife - Mammals Research Work Package No._ _---"-0-=-67'-'0.....__ _ _ __ Lynx Conservation Task No. ---------=-1_ _ _ _ _ __ Post-Release Monitoring of Lynx Reintroduced to Colorado Period Covered: January 1, 2001-December 31, 2001 Author: Tanya M. Shenk, Ph. D. Personnel: A. B. Franklin, L. Gephert, R. Kahn, A. Keith, D. Kenvin, G. Miller, J. Olterman, M. Secor, C. Wagner, S. Wait, G. C. White, D. Younkin Interim Report - Preliminary Results This work continues, and precise analysis ofdata has yet to be accomplished. Manipulation or interpretation of these data beyond that contained in this report should be labeled as such and is discouraged. ABSTRACT In an effort to establish a viable population of lynx (Lynx canadensis) in Colorado 96 lynx were reintroduced into southwestern Colorado in 1999 and 2000. Release protocols were evaluated by monitoring released individuals through radiotelemetry. Numbers of mortalities and causes of death were documented and this information used to modify subsequent release protocols in an effort to attain the highest probability of survival for released lynx. In general, release protocols were modified by increasing length of time lynx were kept at the Colorado holding facility, delaying time of release to spring, and releasing non-pregnant females. Mortality due to starvation decreased as earlier protocols were modified. A suite of hypotheses was developed to model early survival and factors that may have influenced survival, including sex, age on capture, pregnancy, time spent in the Colorado holding facility, and release time. Models were evaluated using AICc model selection and model averaging used to estimate survival rates. There have been 39 confirmed deaths. Human-caused mortality factors such as gunshot and vehicle collision are the highest cause of death for lynx> 8 months post-release. Locations of each lynx were collected through aerial- or satellite-tracking to document movement patterns. Initial dispersal movement patterns and distances traveled by lynx released in 1999 were highly variable and • more extreme than movements oflynx released in 2000. Movement patterns suggest lynx are pairing in March, but successful reproduction has not been do~umented to date. Snow-tracking results indicate the primary winter prey are snowshoe hare (Lepus americanus) and red squirrel (Tamiasciurus hudsonicus), with waterfowl and other mammals and birds forming a minor part of the winter diet. Site-scale habitat data collected from snow-tracking efforts indicate Engelmann spruce (Picea engelmannii) and subalpine fir (Abies lasiocarpa) are the most common forest stands used by lynx in southwestern Colorado. There is a seasonal trend in use of willows (Salix spp.) with use peaking in November and being at its lowest in May and June. ��9 Post-Release Monitoring of Lynx Reintroduced to Colorado Annual Progress Report for the U.S. Fish and Wildlife Service December 2001 Interim Report - Preliminary Results This work continues, and precise analysis ofdata has yet to be accomplished. Manipulation or interpretation of these data beyond that contained in this report should be labeled as such, and is discouraged. Tanya M. Shenk Mammals Research Colorado Division of Wildlife Abstract In an effort to establish a viable population oflynx (Lynx canadensis) in Colorado 96 lynx were reintroduced into southwestern Colorado in 1999 and 2000. Release protocols were evaluated by monitoring released individuals through radiotelemetry. Numbers of mortalities and causes of death were documented and this information used to modify subsequent release protocols in an effort to attain the highest probability of survival for released lynx. In general, release protocols were modified by increasing length of time lynx were kept at the Colorado holding facility, delaying time ofrelease to spring, and releasing non-pregnant females. Mortality due to starvation decreased as earlier protocols were modified. A suite of hypotheses was developed to model early survival and factors that may have influenced survival, including sex, age on capture, pregnancy, time spent in the Colorado holding facility, and release time. Models were evaluated using AICc model selection and model averaging used to estimate survival rates. There have been 39 confirmed deaths. Human-caused mortality factors such as gunshot and vehicle collision are the highest cause of death for lynx >8 months post-release. Locations of each lynx were collected through aerial- or satellite-tracking to document movement patterns. Initial dispersal movement patterns and distances traveled by lynx released in 1999 were highly variable and more extreme than movements of lynx released in 2000. Movement patterns suggest lynx are pairing in March, but successful reproduction has not been documented to date. Snow-tracking results indicate the primary winter prey are snowshoe hare (Lepus americanus) and red squirrel (Tamiasciurus hudsonicus), with waterfowl and other mammals and birds forming a minor part of the winter diet. Site-scale habitat data collected from snow-tracking efforts indicate Engelmann spruce (Picea engelmannii) and subalpine fir (Abies lasiocarpa) are the most common forest stands used by lynx in southwestern Colorado. There is a seasonal trend in use of willows (Salix spp.) with use peaking in November and being at its lowest in May and June. Introduction In an effort to establish a viable population of lynx (Lynx canadensis) in Colorado (Seidel et al. 1998), 41 lynx were reintroduced into southwestern Colorado in the winter and spring of 1999 and an additional 55 lynx were released in April and May of 2000. Post-release monitoring of these lynx is crucial to evaluating the progress of this reintroduction effort. The monitoring program also provides information and data critical for improving release techniques to ensure the highest probability of survival for each individual lynx released in the Colorado effort, and perhaps in other reintroduction efforts. The post-release monitoring program for the reintroduced lynx has 2 primary goals. The first goal is to determine how many lynx remain in Colorado and their locations relative to each other. Given this information and knowing the sex of each individual we can assess whether these lynx can form a breeding core from which a viable population might be established. From these data we can also describe general movement patterns and habitats used. The second primary goal of the monitoring �program is to estimate survival of the reintroduced lynx and, where possible, determine cause of mortality of reintroduced lynx. Such information will help in assessing and modifying release protocols and management of lynx once they have been released. . Additional goals of the post-release monitoring program for lynx reintroduced to th~ southern Rocky Mountains include refining descriptions of habitat use and movement patterns, determining hunting habits, and obtaining information on reproduction. When the lynx establish home ranges that encompass their preferred habitat, more emphasis will be placed on refining descriptions of movement patterns and habitat use. Lynx is listed as threatened under the Endangered Species Act (ESA) of 1973, as amended ( 16 U.S. C. 1531 et. seq.) (U.S. Fish and Wildlife Service 2000). As a listed species, information specific to the ecology of the lynx in its southern range such as habitats used, movement patterns, mortality factors, survival, and reproduction in Colorado will be needed to develop recovery goals and conservation strategies for this species specific to its southern Rocky Mountain range. Thus, an additional objective of the post-release monitoring program is to develop conservation strategies relevant to lynx in Colorado Objectives The initial post-release monitoring of reintroduced lynx will emphasize five primary objectives: 1. Assess and modify release protocols to enure the highest probability of survival for each lynx released. 2. Obtain regular locations ofreleased lynx to describe general movement patterns and habitats used by lynx. 3. Determine causes of mortality in reintroduced lynx. 4. Estimate survival of lynx reintroduced to Colorado. 5. Estimate reproduction oflynx reintroduced to Colorado. Three additional objectives will be emphasized after lynx display site fidelity to an area: 6. Refine descriptions of habitats used by reintroduced lynx. 7. Refine descriptions of daily and overall movement patterns of reintroduced lynx. 8. Describe hunting habits and prey of reintroduced lynx. Information gained to achieve these objectives will form a basis for the development of lynx conservation strategies in the southern Rocky Mountains. Study Area Five areas throughout Colorado were evaluated as potential lynx habitat (Byrne 1998). Criteria investigated in these 5 areas for comparison were (1) relative snowshoe hare densities (Reed at al., unpublished data), (2) road density, (3) size of area, (4)juxtaposition ofhapitats within the area, (5) historical records of fynx observations, and (6) puolic issues: Based on results from this analysis, the San Juan Mountains of southwestern Colorado were selected as the release area for reintroducing lynx. Ten release sites within the San Juan Mountains were selected based on land ownership and accessability during time of release for the 41 animals released in 1999. Of the 55 lynx released in spring 2000, 45 were released at Rio Grande Reservoir and 10 lynx were released at 3 sites west of the Continental Divide. Based on current locations of the majority of the released lynx, the core research area remains in the southern San Juan Mountains. Methods Reintroduction Effort A total of 96 lynx were released at selected areas in the San Juan Mountains of southwestern Colorado (Table 1). Estimated age, sex and body condition were ascertained and recorded for each lynx prior to release (see Wild 1999). Specific release sites were selected based on land ownership and accessibility during times of release. Lynx were transported from the holding facility to the release site in cages (usually!, occasionally 2 lynx per cage). Release site location was recorded in Universal �11 Transverse Mercator (UTM) coordinates and identification of all other lynx released at the same location, on the same day, was recorded. Behavior of the lynx on release and movement away from the release site were documented. Table 1. Colorado lynx reintroduction effort. Assessment of Release Protocols Year Females Males TOTAL In 1999, lynx were released under 5 41 1999 22 19 different release protocols (Table 2). Protocol 1 55 2000 35 20 called for the immediate release of females once TOTAL 57 39 96 they passed veterinary inspection in Colorado. Males were to be held for a period of weeks u_ntil females established a territory, and then males were to be released near female territories. Four lynx were released under this protocol with poor survival. Protocol 2 was developed whereby lynx were held at the Colorado holding facility for a minimum of 3 weeks and fed high quality diets to encourage weight gain. Nine lynx were released under Protocol 2. After a starvation death under Protocol 2, Protocol 3 was developed, requiring the 3-week minimum holding time and high-quality feeding of Protocol 2 plus a release date no earlier than May 1. A spring release would assure that lynx were released when prey was most abundant (i.e., young of the year would be most abundant and hibernating and migratory prey would be available). Twenty lynx were released under Protocol 3. Additionally, 6 females were released under Protocol 3 that were known to be pregnant (Protocol 3P) and 2 that were possibly pregnant (Protocol 3P?). An assessment of the fates of each lynx under all 5 release protocols used in 1999 led to release protocols for lynx released in 2000. Release protocols 2 and 3 resulted in the fewest post-release (up to 8 months after release date) starvation mortalities. The common element in both protocols was increased captivity time in the Colorado hold~ng facility. The single starvation mortality for lynx released under Protocol 2 in 1999 was also the only juvenile released under that protocol and the only animal released in February (the other 8 Protocol 2 lynx were released in March 1999). Thus, all lynx released in 2000 were released under either Protocol 2 or 3 but not before April 1. Because of the high percentage of starvation mortalities in females pregnant on release, we also attempted to avoid reintroducing lynx that were known to be pregnant. This was best accomplished by trying to have animals captured for the reintroduction effort in Canada prior to their breeding season. Table 2. Release protocols for lynx released in southwestern Colorado in 1999 and 2000. Protocol Description 1 Release females as soon as they pass veterinary inspection in Colorado. Release males once females appear to have settled into an area. 2 Release males or females after they have been held in Colorado holding facility for a minimum of 3 weeks and fed a high quality diet. 3 Release males or females after they have been held in Colorado holding facility for a minimum of 3 weeks, fed a high quality diet, and released no earlier than May 1. 3P Pregnant females released under Protocol 3. 3P? Possibly pregnant females released under Protocol 3. To evaluate the efficacy of the changes in release protocols we developed a series of a priori hypotheses concerning factors that affected lynx survival up to 8 months post-release. These factors included (1) the timing ofrelease (winter vs spring), (2) age oflynx released ( adults vs. kittens), (3) sex oflynx released, (4) whether or not females were released while pregnant and the interaction of pregnancy and age of the female (adult vs. kitten), and (5) the duration of holding time in the Colorado facility. A series of 11 models were developed using various combinations of these factors. We used �12 AICc (Burnham and Anderson 1998) as the model selection criterion to select the model that best explained the data. Movement Patterns To determine general movement patterns and habitats used by reintroduced lynx, regular locations of released lynx were collected through a combination of aerial, satellite and ground radiotracking. Locations and general habitat descriptions at each location were recorded and mapped. Frequent flights (at least 2 times per week) were critical during the initial post-release periods because of the greater likelihood of dispersal and mortality in reintroduced carnivores during this period. Every effort was made to locate every lynx each flight during this period. All 41 of the lynx released in the winter and spring of 1999 were fitted with Telonics™ VHF radio-collars, equipped with a mortality switch that activates if the collar remains motionless for 4 hours or more. Fifty-one of the 55 lynx released in the spring 2000 were fitted with Sirtrack™ dual satelliteNHF radio-collars (the other 4 lynx were fitted with Telonics™ VHF collars). These collars also had a mortality indicator switch that operated on both the satellite and VHF mode. The satellite component of each collar was programmed to be active for 12 hours per week. The 12-hour active periods were staggered throughout the week, with approximately 7 collars being active each day of the week. Signals from the collars allowed for locations of the animals to be made via Argos, NASA, and NOAA satellites. The location information was processed by ServiceArgos and distributed to the CDOW through e-mail messages. Survival and Mortality Factors When a mortality signal (75 ppm vs. 50 ppm for the Telonics™ VHF transmitters, 20 bpm vs. 40 bpm for the Sirtrack™ VHF transmitters, 0 activity for Sirtrack™ PTT) was heard during either satellite, aerial or ground surveys, the location (UTM coordinates) was recorded. Ground crews then located and retrieved the carcass as soon as possible. The immediate area was searched for evidence of other predators and the carcass photographed in place before removal. Additionally, the mortality site was described, habitat associations, and exact location were recorded. Any scat found near the dead lynx that appeared to be from the lynx was collected. All carcasses were transported immediately to the Colorado State University Veterinary Hospital for a post mortem exam to 1) determine the cause of death and document with evidence, 2) collect samples for a variety ofresearch projects, and 3) archive samples for future reference (research or forensic). The gross necropsy and histology were performed by, or under the lead and direct supervision of a board certified veterinary pathologist. At least one research personnel from the Colorado Division of Wildlife involved with the lynx program was also present. The protocol followed standard procedures used for thorough post-mortem examination and sample collection for histopathology and diagnostic testing (see Shenk 1999 for details). Some additional data/samples were routinely collected for research, forensics, and archiving. Other data/samples were collected based on the circumstances of the death (e.g., photographs, video, radiographs, bullet recovery, samples for toxicology or other diagnostic tests, etc.). The CDOW retained all samples and carcass remains with the exception of tissues in formalin for histopathology, brain for rabies exam, feces for parasitology, external parasites for ID, and other diagnostic samples. Survival rates oflynx reintroduced to Colorado were estimated using the Kaplan-Meier method with staggered entries (Pollock et al. 1989) in Program MARK (White and Burnham 1999). Recaptures Recaptures were attempted on lynx that were either in poor body condition or need to have their radio collars replaced. Methods ofrecapture included trapping using a Tomahawk™ live trap baited with a rabbit, and darting lynx with Telazol (3 mg/kg) using a Dan-Inject CO2 pistol (modified from Poole et al. 1993 as recommended by M.Wild, DVM). Hounds trained to pursue felids were also used to tree lynx for capture. Treed lynx were immobilized with Telazol or medetomidine (0.09mg/kg) and �21 JOB PROGRESS REPORT Stateof _ _ _ _ _____,C=o=l=or=a=d=o_ _ _ __ Division of Wildlife - Mammals Research Work Package No. _06~7_0_ _ _ _ _ _ __ Lynx Conservation Task No. _ _ _ _ _~1_ _ _ _ _ _ __ Post-Release Monitoring of Lynx Reintroduced to Colorado Period Covered: July 1, 2002 - June 30, 2003 Author: Tanya M. Shenk Personnel: R. Dickman, R. Kahn, A. Keith, G. Miller, C. Wagner, S. Wait, D. Younkin Interim Report - Preliminary Results This work continues, and precise analysis ofdata has yet to be accomplished. Manipulation or interpretation of these data beyond that contained in this report should be labeled as such and is discouraged ABSTRACT Reproduction is critical to the success of any reintroduction effort if a self-sustaining, viable population is the ultimate goal of the conservation effort. As of winter 2002-2003, no reproduction had been documented from lynx reintroduced to Colorado beginning in winter 1999. However, the low density of lynx present in Colorado by winter 2002-2003 limited the ability to answer the question of whether Colorado is suitable to sustain a viable lynx population because either insufficient habitat or lynx at too low a density to achieve reproductive success could have resulted in the lack of reproduction. Following an analysis of possible management options, it was decided that an augmentation of this reintroduction effort was necessary to eliminate an ambiguous result if successful reproduction had not occurred under densities such as exist in winter 2002-03. The reintroduction effort was augmented with 33 additional animals, released within the Core Area in April 2003, to increase lynx density so that the question of whether lynx can sustain viable populations in Colorado could be more definitively addressed. Based on dispersal patterns of lynx released in 2000, the second cohort, it was hypothesized that lynx released in the Core Area would show the necessary site fidelity to increase lynx densities to enhance the probability of successful reproduction. The first lynx kittens documented to be born to lynx reintroduced to Colorado were found on May 21, 2003. A total of 6 dens and 16 kittens were found in 2003. From results to date it can be concluded that CDOW has developed release protocols that ensure high initial post-release survival, and on an individual level lynx have demonstrated they can survive long-term in areas of Colorado. It had also been documented that reintroduced lynx could exhibit site fidelity, engage in breeding behavior and produce kittens. What is yet to be demonstrated is whether Colorado conditions can support the recruitment necessary to offset annual mortality for a population to sustain itself. Monitoring ofreintroduced lynx will continue in an effort to document such viability. �22 Post-Release Monitoring of Lynx (Lynx canadensis) Reintroduced to Colorado Tanya M. Shenk Mammals Research Colorado Division of Wildlife INTRODUCTION The Canada lynx (Lynx canadensis) occurs throughout the boreal forests of northern North America. Colorado represents the southern-most historical distribution of lynx, where the species occupied the higher elevation, montane forests in the state. Little was known about the population dynamics or habitat use of this species in their southern distribution. Lynx were extirpated or reduced to a few animals in the state by the late l 970's. Given the isolation of Colorado to the nearest northern populations, the Colorado Division of Wildlife (CDOW) considered reintroduction as the only option to attempt to reestablish the species in the state. A key question to be asked when considering the re-establishment of any species is, "What is different now from when they disappeared?" For lynx, the causative factor(s) of their extirpation may never be known. Many of the hypothesized factors, however, have changed substantially since the early and mid-l 900's. For example, widespread predator poisoning no longer occurs; conservation of wildlife habitat is now given much stronger consideration in public land management decisions; trapping and hunting are more strictly regulated and regulations enforced; and in some areas, at least, the passage of time has allowed the landscape to recover from abuses of the past, perhaps to a state that is more conducive to lynx survival. It must be acknowledged, however, that there may be other detrimental factors operating now that did not exist previously. In particular, increased human density and development have occurred in some areas and exotic diseases such as plague have been introduced in Colorado. The uncertainty surrounding the cause of the extirpation of lynx and the effects of current conditions in Colorado on lynx makes it impossible to predict with confidence whether Colorado has sufficient habitat to sustain viable population(s) oflynx. In order to perform the best test of this question the CDOW led a cooperative effort to reintroduce wild-trapped lynx from Canada and Alaska into southwestern Colorado beginning in 1999. It was hoped the effort would clarify whether or not Colorado is or is not suitable for sustaining viable lynx populations, provided the fate of the released animals could be determined. The goal of the Colorado lynx reintroduction program is to establish a viable population of lynx in this state. Evaluation of incremental achievements necessary for establishing viable populations is an interim method of assessing if the reintroduction effort is progressing towards success. There are seven critical criteria for achieving a viable population: ( 1) development of release protocols that lead to a high initial post-release survival of reintroduced animals, (2) long-term survival of lynx in Colorado, (3) development of site fidelity by the lynx to areas supporting good habitat in densities sufficient to breed, (4) reintroduced lynx must breed, (5) breeding must lead to reproduction of surviving kittens (6) lynx born in Colorado must reach breeding age and reproduce successfully, and (7) recruitment must be equal to or greater than mortality. Prior to the reintroduction, it was hypothesized that a minimum of 100 animals would need to be released for a fair evaluation of the suitable/unsuitable question. In 1999 and 2000, 96 lynx (57 females, 39 males) were released into the San Juan Mountains of southwestern Colorado. The 1999 cohort of 41 individuals scattered widely, and suffered a first year mortality of 17 (41 %) lynx (Shenk 2001). The 2000 cohort of 55 animals, being released into areas already occupied (although sparsely) by the previous year's animals, were more sedentary, and experienced a first year mortality of 10 (18%) lynx. Humancaused mortalities due to vehicle collision, gunshot, and the mortalities where only a cut collar was found �23 comprise the greatest known cause of mortality for all the reintroduced lynx (31 %). Mortalities due to starvation (23%) were minimized with the improved release protocols. To date, only 2 of the 55 lynx released in 2000 died of starvation. However, the improved survival of reintroduced lynx provided only partial evidence that Colorado could sustain a viable population of the species. As of winter 2003, no successful reproduction had been documented. This lack of reproduction resulted in an increased emphasis on the question of whether or not Colorado could provide sufficient habitat to sustain a selfsustaining population of lynx. Two options existed to address the problem of answering the suitable/unsuitable question. The first was to continue to monitor the existing animals for recruitment, with the understanding that the probability of detection would decrease rapidly as radio-collars failed, and the probability of successful pairing might further decrease with lowered densities due to natural mortality. Possible outcomes include 1) the animals currently out there would eventually reproduce with sufficient success to establish a viable population of lynx, 2) the animals currently out there would reproduce although not in sufficient numbers to offset mortality or 3) the animals currently out there would fail to reproduce. The primary reasons for outcomes 2 and 3 are either that Colorado does not have sufficient habitat to support viable populations of lynx or there were too few lynx released to achieve sufficient successful reproduction. Thus, the question of whether or not Colorado can support viable population(s) oflynx would remain arguable. A second option would be to supplement the existing lynx by re-introducing additional lynx over multiple years into the Core Area to attain a density approaching that of established populations of lynx. The possible outcomes could be any of those listed for the first option. The difference, however, would be that the low-density explanation for failure to establish a viable population would be difficult to support. Thus, CDOW could more definitively address the question of the suitability of Colorado for lynx populations. An analysis of these two options was conducted to determine the best management strategy to pursue to enhance the ability to assess the outcome. An update of the post-release monitoring program was also conducted. OBJECTIVES The initial post-release monitoring ofreintroduced lynx will emphasize 5 primary objectives: 1. Assess and modify release protocols to enure the highest probability of survival for each lynx released. 2. Obtain regular locations ofreleased lynx to describe general movement patterns and habitats used by lynx. 3. Determine causes of mortality in reintroduced lynx. 4. Estimate survival oflynx reintroduced to Colorado. 5. Estimate reproduction oflynx reintroduced to Colorado. Three additional objectives will be emphasized after lynx display site fidelity to an area: 6. Refine descriptions of habitats used by reintroduced lynx. 7. Refine descriptions of daily and overall movement patterns of reintroduced lynx. 8. Describe hunting habits and prey ofreintroduced lynx. Information gained to achieve these objectives will form a basis for the development of l y ~ oervation "'-. strategies in the southern Rocky Mountains. Lastly, an analysis was conducted to evaluat tow --------7/ management options for assessing Colorado's suitability for sustaining a viable lynx populat10 . �24 METHODS Augmentation An analysis of two management options was conducted to determine the best management strategy to pursue to enhance the ability to assess whether Colorado provided suitable habiat for a viable, self-sustaining population of lynx. In response to the completed analysis, the current reintroduction effort was augmented with additional animals, released within the Core Area, to increase lynx density so that the question of whether lynx can sustain viable populations in Colorado could be more definitively addressed. These new releases were conducted under the protocols found to maximize survival (see Shenk 1999). Based on dispersal patterns of lynx released in 2000, the second cohort, it was hypothesized that lynx released in the Core Area would show the necessary site fidelity to increase lynx densities to enhance the probability of successful reproduction. Movement Patterns To determine general movement patterns and habitat used by reintroduced lynx, regular locations of released lynx were collected through a combination of aerial, satellite and ground radio-tracking. Locations and general habitat descriptions at each location were recorded and mapped. Frequent flights (at least 2 times per week) were critical during the initial post-release periods because of the greater likelihood of dispersal and mortality in reintroduced carnivores during this period. Every effort was made to locate all lynx each flight during this period. All lynx released in the winter and spring of 1999 were fitted with Telonics™ VHF radio-collars, equipped with a mortality switch that activates if the collar remains motionless for 4 hours or more. Fiftyone of the 55 lynx released in the spring 2000 were fitted with Sirtrack™ dual satellite/VHF radio-collars (the other 4 lynx were fitted with Telonics™ VHF collars). All 33 lynx released in 2003 were fitted with Sirtrack™ dual satellite/VHF radio-collars. These collars also had a mortality indicator switch that operated on both the satellite and VHF mode. The satellite component of each collar was programmed to be active for 12 hours per week. The 12-hour active periods were staggered throughout the week, with approximately 7 collars being active each day of the week. Signals from the collars allowed for locations of the animals to be made via Argos, NASA, and NOAA satellites. The location information was processed by ServiceArgos and distributed to the CDOW through e-mail messages. Survival and Mortality Factors When a mortality signal (75 ppm vs. 50 ppm for the Telonics™ VHF transmitters, 20 bpm vs. 40 bpm for the Sirtrack™ VHF transmitters, 0 activity for Sirtrack™ PTT) was heard during either satellite, aerial or ground surveys, the location (UTM coordinates) was recorded. Ground crews then located and retrieved the carcass as soon as possible. The immediate area was searched for evidence of other predators and the carcass photographed in place before removal. Additionally, the mortality site was described, habitat associations, and exact location were recorded. Any scat found near the dead lynx that appeared to be from the lynx was collected. All carcasses were transported immediately to the Colorado State University Veterinary Hospital for a post mortem exam to 1) determine the cause of death and document with evidence, 2) collect samples for a variety of research projects, and 3) archive samples for future reference (research or forensic). The gross necropsy and histology were performed by, or under the lead and direct supervision of a board certified veterinary pathologist. At least one research personnel from the CDOW involved with the lynx program was also present. The protocol followed standard procedures used for thorough post-mortem examination and sample collection for histopathology and diagnostic testing (see Shenk �25 1999 for details). Some additional data/samples were routinely collected for research, forensics, and archiving. Other data/samples were collected based on the circumstances of the death (e.g., photographs, video, radiographs, bullet recovery, samples for toxicology or other diagnostic tests, etc.). The CDOW retained all samples and carcass remains with the exception of tissues in formalin for histopathology, brain for rabies exam, feces for parasitology, external parasites for ID, and other diagnostic samples. Reproduction Females were monitored for proximity to males during breeding season and for site fidelity to a given area during the denning period of May and June. Each female that exhibited stationary movement patterns in May or June 2003 was observed to look for accompanying kittens. If kittens were found at a den site they were weighed, sexed and photographed. Each kitten was uniquely marked by inserting a sterile passive integrated transponder (PIT, Biomark, Inc., Boise, Idaho, USA) tag subcutaneously between the shoulder blades. Time spent at the den was minimized to ensure the least amount of disturbance to the female and the kittens. Weight, PIT-tag number, sex and any distinguishing characteristics of each kitten was recorded. Den site location was recorded as Universal Transmercator (UTM) Coordinates. Other data to be recorded include general vegetation characteristics, elevation, weather, field personnel, time at the den, and behavioral responses of the kittens and female. RESULTS Rationale for Augmentation Thirty-six reintroduced lynx were known to be in the Core Area in winter 2002-2003, which is approximately 10,000 mi 2 . Thus, the lowest possible density oflynx in the Core Area was approximately 2 lynx/ 500 mi2, an area slightly larger than Rocky Mountain National Park. The highest density of reintroduced lynx in the Core Area was approximately 3 lynx/ 500 mi2, if all the missing lynx at that time were currently there but not being detected due to faulty radio collars. If additional naturally occurring lynx were in the area these densities could have been even higher. Lynx densities reported for natural populations occurring in northern habitats range from< 13 lynx/ 500 mi2 during snowshoe hare lows to 104-259 lynx/ 500 mi2 in years of peak hare densities in mature forests. The densities of lynx reported for populations in the north during the low in the hare cycle may not represent the lowest densities at which lynx could exist and maintain viable populations. At these lows, northern lynx still reproduce although at a much lower rate then when the hare density is higher. This low reproductive rate could be related to poor body condition, low lynx densities, or a combination of both. What can be assumed is that lynx occurring at these low densities are able to rebound and achieve higher densities. Given that reintroduced lynx in Colorado are in good body condition, CDOW may only need to increase densities to achieve reproductive rates that would sustain a viable population of lynx. Densities of lynx reported for their northern range reflect densities where lynx habitat is more uniform and consistent than in Colorado. In Colorado, although the Core Area is described as 10,000 mi2, lynx are not using the Core Area uniformly but rather are dispersed in patches throughout the Core Area. Therefore the densities calculated for lynx in Colorado are not directly comparable to those estimated from the north. It is difficult, however, to estimate an appropriate correction factor for Colorado densities to make them comparable to those reported for northern populations. Therefore, the number of lynx needed to augment the current population to achieve a density of 13 lynx/ 500 mi2 under several combinations of current density and percent of the Core Area that has suitable habitat was estimated (Table 1). �26 Although this analysis required numerous assumptions, an augmentation effort of at least 150 animals (no more than 50 per year) is a minimum target for achieving densities oflynx conducive to successful reproduction and recruitment. Once minimum densities have been achieved, additional releases should continue over four to five years to maintain the minimum densities considered necessary for successful reproduction. Monitoring of the lynx population throughout the augmentation will be critical and should be conducted rigorously. The target density of 13 lynx I 500 mi2 is based on the lowest densities documented for northern populations. However, lynx may be able to rebound from lower densities. Thus, through monitoring CDOW should estimate at what densities reproduction occurs, and at what densities successful recruitment of animals occurs. This may happen at densities lower than low lynx densities estimated for the north. Table 1. Estimates of current densities of reintroduced lynx in the Core Area under various combinations of number of lynx and percent suitable habitat. Calculations of how many lynx would be needed under these conditions to achieve densities similar to the lowest densities reported for northern populations are presented and the number of additional lynx needed to achieve this density. Minimum Density Lynx needed to achieve additional lynx lynx/ 500 mi2 13 lynx/ 500 mi2 Density Assumptions needed 1 No.of lynx % Area suitable mimmum 100% 1.8 260 224 maximum 100% 2.6 260 208 mimmum 75% 2.4 195 159 maximum 75% 3.5 195 143 mmimum 50% 3.6 130 94 maximum 50% 5.2 130 78 Assumes no mortality. Augmentation Based on the adoption of the augmentation management strategy, 33 lynx were released in April 2003, bringing the total number of lynx reintroduced to Colorado to 129 (Table 2). The 33 lynx reintroduced in 2003 had been captured in Quebec, Manitoba and British Columbia. These new releases were conducted under the protocols found to maximize survival (see Shenk 2001). All 33 lynx were released in the Core Area of southwestern Colorado. Each lynx was released with a dual VHF /satellite radio collar so that the lynx can be monitor for movement and mortality. Estimated age, sex and body condition were ascertained and recorded for each lynx prior to release (see Wild 1999). Specific release sites were selected based on land ownership and accessibility during times of release. Lynx were transported from the holding facility to the release site in cages (usually 1, occasionally 2 lynx per cage). Release site location was recorded in Universal Transverse Mercator (UTM) coordinates and identification of all other lynx released at the same location, on the same day, was recorded. Behavior of the lynx on release and movement away from the-release site were documented. �27 Table 2. Colorado lynx reintroduction effort as of June 30, 2004. Females Males TOTAL Year 22 19 41 1999 35 20 55 2000 17 16 33 2003 74 55 129 TOTAL Reproduction Nine pairs of lynx were documented during the 2003 breeding season (March and April). In May and June 2003, 6 dens and a total of 16 kittens were found in the lynx core research area in southwestern Colorado (Table 3). At all dens the females appeared in excellent condition, as did the kittens. The kittens weighed from 270-500 grams. Lynx kittens weigh approximately 200 grams at birth and do not open their eyes until they are 10-17 days old. Dens were found when field crews walked in on females that exhibited virtually no movement for at least 10 days from both aerial and ground telemetry. Table 3. Reproduction information for summer 2003. Kittens Date Den Female Release Year Found Females 2000 BCO0F8 5/21/03 ? 2000 BC00F19 5/26/03 l 2000 YK00F16 6/19/03 l YK99Fl 1999 6/10/03 2 YK00F19 2000 6/11/03 YK00Fl0 2000 5/31/03 2 TOTAL 7 Males ? l 1 2 2 7 Total 2 2 2 3 3 4 16 The dens were scattered throughout the Core Area, with no dens found outside the Core Area. All the dens were in Engelmann spruce/subalpine fir forests in areas of extensive downfall. Elevations ranged from 3240-3557 m (10,630 - 11,670 feet). Field crews weighed, photographed, and PIT-tagged the kittens. Field crews also took hair samples from the kittens for genetic work in an attempt to confirm paternity. Kittens were processed as quickly as possible (11-32 minutes) to minimize the time the kittens were without their mother. While working with the kittens the females remained nearby, often making themselves visible to us. The females generally continued a low growling vocalization the entire time personnel were at the den. In all cases, the female returned to the den site once field crews left the area. Locations •The 2003 releases have remained in the Core Area with the exception of 2 lynx that went briefly to New Mexico but subsequently returned to Colorado. Most lynx continue to use terrain within the Core Area: New Mexico north to Gunnison, west as far as Taylor Mesa and east to Monarch Pass. There are some lynx north of Gunnison up to the 170 corridor and in the :Caylor Park area. No lynx are known to be north of 170 at this time. Mortalities Of the total 129 lynx released in 1999, 2000 and 2003 there are 46 known mortalities. Of these 46 mortalities, 25 are from the 1999 releases, 20 are from the 2000 releases, and 1 is from the 2003 releases. Causes of death are listed in Table 4. �28 Table 4. Causes of death for lynx released into southwestern Colorado in 1999, 2000 and 2003. 2000 2003 2003 2000 2000 1999 1999 Male Female Male Female Unk Male Female Cause of Death 1 6 1 Starvation 1 3 Hit by Vehicle 2 1 1 3 1 1 Shot Probable Predation 1 3 Plague Unknown: Human Caused 1 2 1 Probable Shot Probable Hit by Vehicle 2 Unknown: Not Starvation 1 2 1 2 4 3 1 Unknown 1 12 Total Mortalities 8 17 7 1 1 Total 9 6 6 1 3 4 2 4 11 46 Current Status At this time, CDOW is tracking 61 of the 83 lynx still possibly alive. A lynx is listed as missing if a signal has not been heard from the animal for at least 1 year. There are 21 lynx that CDOW has not heard signals on since at least June 30, 2002 (fable 5). One of these missing lynx cannot be identified but was hit by a truck in New Mexico, thus only 20 are truly missing. Possible reasons for not locating these missing lynx include (1) long distance dispersal, beyond the areas currently being searched, (2) radio failure, or (3) destruction of the radio (e.g., run over by car). CDOW continues to search for all missing lynx during both aerial and ground searches. Expanded flights outside the research area during the summer and fall months may yield locating these missing lynx. Two of the lynx released in 2000 have probably slipped their collars. Thus, CDOW has tracked 61 individual lynx since at least June 30, 2002. Table 5. Status oflynx reintroduced to Colorado as of June 30, 2003. Females Males TOTALS Unk.nown 74 55 129 Released 29 46 Known Dead 16 1 Possible Alive 45 39 83 7 14 Missing 21 (1 is unknown mortality) 1-2 Slipped Collar 1 1? 37 24 Tracking 61 �29 DISCUSSION The low density oflynx present in Colorado in winter 2002-2003 limited the ability to answer to the question of whether Colorado has sufficient suitable habitat to sustain a viable lynx population. At that time, the lack of successful reproduction may have reflected either insufficient habitat or lynx at too low a density to achieve reproductive success. It was decided that an augmentation of this reintroduction effort was necessary to eliminate an ambiguous result if successful reproduction had not occurred under densities existing in winter 2002-03. In order to maintain densities equal to those in areas that have maintained breeding populations the CDOW would need to reintroduce 50 lynx per year for the next three years and augment the population with an additional I 0-12 lynx for years 4 through 6. The reintroduction effort was augmented with 33 additional animals, released within the Core Area in April 2003, to increase lynx density so that the question of whether lynx can sustain viable populations in Colorado could be more definitively addressed. Based on dispersal patterns of lynx released in 2000, the second cohort, it was assumed lynx released in the Core Area would show the necessary site fidelity to increase lynx densities to enhance the probability of successful reproduction. The first lynx kittens documented to be born to lynx reintroduced to Colorado were found on May 21, 2003. A total of 6 dens and 16 kittens were found in 2003. While this is a milestone CDOW has been hoping to achieve, live births are the first step towards recruitment. Recruitment into a population would require these kittens to survive through their first year oflife and produce offspring of their own. To achieve a viable population of lynx, enough kittens need to be recruited into the population to offset the mortality that occurs in that year and hopefully even add more so that the population can grow. Although den sites will not be visited again until fall 2003, so as not to disturb the female and kittens further, the female's movement patterns will be monitored through aerial telemetry. During fall 2003, demales with kittens will be observed through walk-ins to try to count the number of kittens still with her. Alternatively, the females will be snow-tracked once there is sufficient snowfall on the ground to document the presence and number of kittens. Kittens typically stay with their mothers until they are I 0 months old. The Colorado lynx reintroduction effort has overcome most obstacles encmm.tered so far. From results to date it can be concluded that the CDOW has developed release protocols that ensure high initial post-release survival (Shenk 2001), and on an individual level lynx have demonstrated they can survive long-term in areas of Colorado. It had also been documented that reintroduced lynx could exhibit site fidelity, engage in breeding behavior and produce kittens. What is yet to be demonstrated is whether Colorado conditions can support the recruitment necessary to more-than-offset annual mortality for a population to sustain itself. Monitoring of reintroduced lynx will continue in an effort to document such viability. LITERATURE CITED Shenk, T. M. 1999. Program narrative: Post-release monitoring ofreintroduced lynx (Lynx canadensis) to Colorado. Report for the Colorado Division of Wildlife. Shenk, T. M. 200 I. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report. Colorado Division of Wildlife. Wild, M.A. 1999. Lynx veterinary services and diagnostics. Job Progress Report for the Colorado Division of Wildlife. Fort Collins, Colorado. �Colorado Division of Wildlife Wildlife Research Report July 2003- June 2004 JOB PROGRESS REPORT State of Colorado Project No. Work Package No. 0670 Task No. 1 : : : : Federal Aid Project: : N/A Cost Center 3430 Mammals Research Lynx Conservation Post-Release Monitoring of Lynx Reintroduced to Colorado Period Covered: July 1, 2003 - June 30, 2004 Author: Tanya M. Shenk Personnel: R. Dickman, L. Gepfert, R. Kahn, A. Keith, G. Merrill, G. Miller, C. Wagner, S. Wait, S. Waters, L. Wolfe, D. Younkin All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT In an effort to establish a viable population of lynx (Lynx canadensis) in Colorado, a reintroduction effort was initiated in 1999. The reintroduction effort was augmented with the release of 37 additional animals in April 2004, bringing the total to 166 lynx reintroduced to southwestern Colorado. Each lynx is released with dual satellite and VHF radio transmitters to allow intensive monitoring of animals after release. Through documentation of lynx mortalities and causes of death, human-caused mortality factors such as gunshot and vehicle collision are currently the highest source of mortality for reintroduced lynx. Locations of each lynx were collected through aerial- or satellite-tracking to document movement patterns. Most lynx remain in the southwestern quarter of Colorado. Reproduction was first documented during the 2003 reproduction season. A second successful breeding season was documented in 2004 with 11 dens and 30 kittens found as of June 30, 2004. Snow-tracking results indicate the primary winter prey species are snowshoe hare (Lepus americanus) and red squirrel (Tamiasciurus hudsonicus), with other mammals and birds forming a minor part of the winter diet. Site-scale habitat data collected from snowtracking efforts indicate Engelmann spruce (Picea engelmannii) and subalpine fir (Abies lasiocarpa) are the most common forest stands used by lynx in southwestern Colorado. From results to date it can be concluded that CDOW has developed release protocols that ensure high initial post-release survival, and on an individual level lynx have demonstrated they can survive long-term in areas of Colorado. It has also been documented that reintroduced lynx could exhibit site fidelity, engage in breeding behavior and produce kittens. What is yet to be demonstrated is whether Colorado conditions can support the recruitment necessary to offset annual mortality for a population to sustain itself. Monitoring of reintroduced lynx will continue in an effort to document such viability. 5 �JOB PROGRESS REPORT POST-RELEASE MONITORING OF LYNX (Lynx canadensis) REINTRODUCED TO COLORADO Tanya M. Shenk SEGMENT OBJECTIVES The initial post-release monitoring of reintroduced lynx will emphasize 5 primary objectives: 1. Assess and modify release protocols to ensure the highest probability of survival for each lynx released. 2. Obtain regular locations of released lynx to describe general movement patterns and habitats used by lynx. 3. Determine causes of mortality in reintroduced lynx. 4. Estimate survival of lynx reintroduced to Colorado. 5. Estimate reproduction of lynx reintroduced to Colorado. Three additional objectives will be emphasized after lynx display site fidelity to an area: 6. Refine descriptions of habitats used by reintroduced lynx. 7. Refine descriptions of daily and overall movement patterns of reintroduced lynx. 8. Describe hunting habits and prey of reintroduced lynx. Information gained to achieve these objectives will form a basis for the development of lynx conservation strategies in the southern Rocky Mountains. INTRODUCTION The Canada lynx occurs throughout the boreal forests of northern North America. Colorado represents the southern-most historical distribution of lynx, where the species occupied the higher elevation, montane forests in the state. Little was known about the population dynamics or habitat use of this species in their southern distribution. Lynx were extirpated or reduced to a few animals in the state by the late 1970’s. Given the isolation of Colorado to the nearest northern populations, the Colorado Division of Wildlife (CDOW) considered reintroduction as the only option to attempt to reestablish the species in the state. A reintroduction effort was begun in 1999. To date, 166 wild-caught lynx from Alaska and Canada have been released in southwestern Colorado. The goal of the Colorado lynx reintroduction program is to establish a self-sustaining, viable population of lynx in this state. Evaluation of incremental achievements necessary for establishing viable populations is an interim method of assessing if the reintroduction effort is progressing towards success. There are seven critical criteria for achieving a viable population: (1) development of release protocols that lead to a high initial post-release survival of reintroduced animals, (2) long-term survival of lynx in Colorado, (3) development of site fidelity by the lynx to areas supporting good habitat in densities sufficient to breed, (4) reintroduced lynx must breed, (5) breeding must lead to reproduction of surviving kittens (6) lynx born in Colorado must reach breeding age and reproduce successfully, and (7) recruitment must be equal to or greater than mortality. The post-release monitoring program for the reintroduced lynx has 2 primary goals. The first goal is to determine how many lynx remain in Colorado and their locations relative to each other. Given this information and knowing the sex of each individual, we can assess whether these lynx can form a breeding core from which a viable population might be established. From these data we can also describe general movement patterns and habitats used. The second primary goal of the monitoring program is to 6 �estimate survival of the reintroduced lynx and, where possible, determine cause of mortality of reintroduced lynx. Such information will help in assessing and modifying release protocols and management of lynx once they have been released. Additional goals of the post-release monitoring program for lynx reintroduced to the southern Rocky Mountains include refining descriptions of habitat use and movement patterns, determining hunting habits, and obtaining information on reproduction. When the lynx establish home ranges that encompass their preferred habitat, more emphasis will be placed on refining descriptions of movement patterns and habitat use. Lynx is listed as threatened under the Endangered Species Act (ESA) of 1973, as amended (16 U. S. C. 1531 et. seq.)(U. S. Fish and Wildlife Service 2000). As a listed species, an additional objective of the post-release monitoring program is to develop conservation strategies relevant to lynx in Colorado. Therefore, information specific to the ecology of the lynx in its southern Rocky Mountain range such as habitats used, movement patterns, mortality factors, survival, and reproduction in Colorado is needed. METHODS Reintroduction Effort All 2004 lynx releases were conducted under the protocols found to maximize survival (see Shenk 2001). Estimated age, sex and body condition were ascertained and recorded for each lynx prior to release (see Wild 1999). Specific release sites were selected based on land ownership and accessibility during times of release. Lynx were transported from the holding facility to the release site in individual cages. Release site location was recorded in Universal Transverse Mercator (UTM) coordinates and identification of all lynx released at the same location, on the same day, was recorded. Behavior of the lynx on release and movement away from the release site were documented. Movement Patterns All lynx released in spring 2004 were fitted with SirtrackTM dual satellite/VHF radio-collars. These collars have a mortality indicator switch that operated on both the satellite and VHF mode. The satellite component of each collar was programmed to be active for 12 hours per week. The 12-hour active periods were staggered throughout the week. Signals from the collars allowed for locations of the animals to be made via Argos, NASA, and NOAA satellites. The location information was processed by ServiceArgos and distributed to the CDOW through e-mail messages. To determine general movement patterns of reintroduced lynx, regular locations of released lynx were collected through a combination of aerial, satellite and ground radio-tracking. Locations and general habitat descriptions at each location were recorded and mapped. Survival and Mortality Factors When a mortality signal (75 ppm vs. 50 ppm for the Telonics™ VHF transmitters, 20 bpm vs. 40 bpm for the Sirtrack™ VHF transmitters, 0 activity for Sirtrack™ PTT) was heard during either satellite, aerial or ground surveys, the location (UTM coordinates) was recorded. Ground crews then located and retrieved the carcass as soon as possible. The immediate area was searched for evidence of other predators and the carcass photographed in place before removal. Additionally, the mortality site was described and habitat associations and exact location were recorded. Any scat found near the dead lynx that appeared to be from the lynx was collected. All carcasses were transported immediately to the Colorado State University Veterinary Hospital for a post mortem exam to 1) determine the cause of death and document with evidence, 2) collect samples for a variety of research projects, and 3) archive samples for future reference (research or 7 �forensic). The gross necropsy and histology were performed by, or under the lead and direct supervision of a board certified veterinary pathologist. At least one research personnel from the CDOW involved with the lynx program was also present. The protocol followed standard procedures used for thorough post-mortem examination and sample collection for histopathology and diagnostic testing (see Shenk 1999 for details). Some additional data/samples were routinely collected for research, forensics, and archiving. Other data/samples were collected based on the circumstances of the death (e.g., photographs, video, radiographs, bullet recovery, samples for toxicology or other diagnostic tests, etc.). The CDOW retained all samples and carcass remains with the exception of tissues in formalin for histopathology, brain for rabies exam, feces for parasitology, external parasites for ID, and other diagnostic samples. Reproduction Females were monitored for proximity to males during breeding season and for site fidelity to a given area during the denning period of May and June. Each female that exhibited stationary movement patterns in May or June 2004 was observed to look for accompanying kittens. Kittens found at den sites were weighed, sexed and photographed. Each kitten was uniquely marked by inserting a sterile passive integrated transponder (PIT, Biomark, Inc., Boise, Idaho, USA) tag subcutaneously between the shoulder blades. Time spent at the den was minimized to ensure the least amount of disturbance to the female and the kittens. Weight, PIT-tag number, sex and any distinguishing characteristics of each kitten was also recorded. Den site location was recorded as UTM coordinates. General vegetation characteristics, elevation, weather, field personnel, time at the den, and behavioral responses of the kittens and female were also recorded. Diet Winter diet of reintroduced lynx was estimated by documenting successful kills through snowtracking. Prey species from failed and successful hunting attempts were identified by either tracks or remains. Scat analysis also provided information on foods consumed. Scat samples were collected wherever found and labeled with location and individual lynx identification. Only part of the scat was collected; the remainder was left in place in the event that the scat was being used by the animal as a territory mark. RESULTS Reintroduction Effort Based on the adoption of the approved augmentation management strategy (Shenk 2003), 37 lynx (17 females and 20 males) were released in April 2004, bringing the total number of lynx reintroduced to Colorado to 166 (Table 1). Lynx for the 2004 augmentation were captured in Quebec and British Columbia. All 37 lynx were released at previously used release sites in southwestern Colorado. Table 1. Colorado lynx reintroduction effort as of June 30, 2004. Year Females Males TOTAL 1999 22 19 41 2000 35 20 55 2003 17 16 33 2004 17 20 37 TOTAL 91 75 166 8 �Movement Patterns Most of the lynx released in 2004 have remained in the southwestern quarter of Colorado, with the exception of 2 lynx that went briefly to New Mexico but subsequently returned to Colorado. The majority of surviving lynx from the entire reintroduction effort continue to use areas from New Mexico north to Gunnison, west as far as Taylor Mesa and east to Monarch Pass. There are some lynx north of Gunnison up to the I70 corridor and in the Taylor Park area. No lynx are known to be north of I70 at this time. Mortalities Of the total 166 adult lynx released in 1999, 2000, 2003 and 2004 we have 56 known mortalities. Of these 56 mortalities, 25 are from the 1999 releases, 23 are from the 2000 releases, 4 are from the 2003 releases, and 4 are from the 2004 releases. Causes of death are listed in Table 2. Of the 16 kittens known to have been born in Colorado in 2003, we have 7 confirmed mortalities and 3 possible mortalities. Table 2. Causes of death for adult lynx released into southwestern Colorado in 1999, 2000, 2003 and 2004. 1999 Total Cause of death 2000 2003 2004 M F M F U M F M F Starvation 1 Hit by Vehicle Shot 6 1 2 3 Probable Predation 1 1 3 1 9 1 1 1 1 7 7 1 1 Plague 4 4 Unknown: Human Caused 1 1 Probable Shot 1 Probable Hit by Vehicle 2 1 4 2 Unknown: Not Starvation 1 2 Unknown 2 1 2 1 4 1 4 2 1 Human Caused Total 1 14 1 8 17 7 15 1 3 1 6 2 1 2 56 As of June 30, 2004, CDOW was actively tracking 85 of the 110 lynx still possibly alive (Table 3). Of the remaining 25 remaining lynx possibly alive, 24 were ‘missing’ as of June 30, 2004 (Table 3). A lynx was listed as missing if a signal has not been heard from the animal for at least 1 year. One of these missing lynx is the unknown mortality, thus only 23 are truly missing. Possible reasons for not locating these missing lynx include (1) long distance dispersal, beyond the areas currently being searched, (2) radio failure, or (3) destruction of the radio (e.g., run over by car). CDOW continues to search for all missing lynx during both aerial and ground searches. Two of the lynx released in 2000 are thought to have slipped their collars. Thus, CDOW tracked 85 individual lynx since at least June 30, 2003. 9 �Table 3. Status of adult lynx reintroduced to Colorado as of June 30, 2004. Females Males Unknown TOTALS Released 91 75 Known Dead 35 20 Possible Alive 56 55 110 Missing 9 15 23 (1 is unknown mortality) Slipped Collar 1 1? 1-2 Tracking 46 39 85 166 1 56 Reproduction Of the 6 females that had kittens in 2003, 1 died and 2 had collars that shut off prior to denning season in 2004. Of the 3 that could be monitored during the 2004 denning season, 1 had a litter of 2 kittens (YK00F10), 1 did not have kittens (BC00F08) and it is highly probable the third female (YK99F1) has kittens with her based on her movement patterns. We are still trying to document her status. The 2004 dens were scattered throughout Colorado and 1 den was found in southern Wyoming. Most of the dens were in Engelmann spruce/subalpine fir forests in areas of extensive downfall. Elevations at the den sites ranged from 2652-3560 m (8701 - 11,680 feet). We weighed, photographed, and PIT-tagged the kittens and recorded sex. We also took hair samples from the kittens for genetic work in an attempt to confirm paternity. We processed the kittens as quickly as possible (15-35 minutes) to minimize the time the kittens were without their mother. Four of the females would not leave the den until we reached out to pick up a kitten. While we were working with the kittens the females remained nearby, often remaining visible to us. The females generally continued a low growling vocalization the entire time we were at the den. In all cases, the female returned to the den site once we left the area. Four of the 6 females that we know had kittens in summer 2003 were still with kittens at the end of April 2004. Two of those females still had 2 kittens with them at that time. Visual observations in February 2004 of one female with 2 kittens indicated all 3 were in good body condition. Snow-trackers documented at least 1 snowshoe hare kill by a kitten in winter 2003-04. The mortality of the female YK00F16 and her 1 kitten from plague was not due to poor habitat or prey conditions, and thus we might assume she would have raised the 1 kitten to this stage as well. Three probable kitten deaths from female YK00F19 were from 1 litter that most likely failed very early. Through snow-tracking an unknown female (no radio frequency heard in the area of the tracks) we also documented 1-2 additional kittens born spring 2003 and still alive in winter 2004. Of the 16 kittens we found in summer 2003, we documented the following by April 2004: 6 confirmed alive, 7 confirmed dead, and 3 some evidence dead (Table 4). Although we tried, we were not able to capture any of the 6 surviving kittens to fit them with radio collars. Unless we capture or find any of these kittens from other methods we will not know their fate beyond this 10 months of survival. 10 �Table 4. Known reproduction for summer 2003 and subsequent kitten fates by April 2004. Female Release Year Date Den Found Females Males Total BC00F8 2000 5/21/2003 ? ? 2 Kittens Known Alive in April 2004 1 BC00F19 2000 5/26/2003 1 1 2 1 1 YK00F16 2000 6/19/2003 1 1 2 0 2 YK99F1 1999 6/10/2003 2 1 3 2 1 YK00F19 2000 6/11/2003 1 2 3 ? 3? YK00F10 2000 5/31/2003 2 2 4 2 2 16 6 7, 3? Kittens Born TOTAL Kittens Known Dead in April 2004 1 In spring 2004 we had 26 females from the releases in 1999, 2000 and 2003 that had active radio collars. We documented 18 possible mating pairs of lynx during breeding season. We defined a possible mating pair as any male and female documented within at least 1 km of each other in breeding season through either flight data or snow-tracking data. All 4 of the females that had kittens with them through winter 2003-04 bred again this spring, 2 with the same male they successfully bred with last spring. During May-June 2004 we found 11 dens and a total of 30 kittens (Table 5). Dens were found when we walked in on females that exhibited virtually no movement for at least 10 days from both aerial and ground telemetry. At all dens the females appeared in excellent condition, as did the kittens. The kittens weighed from 250-770 grams. Lynx kittens weigh approximately 200 grams at birth and do not open their eyes until they are 10-17 days old. Three of the 11 females with kittens were from the 2003 releases (Table 5). Table 5. Lynx reproduction documented in 2004. Kittens Born Female Release Year Date Den Found Females Males Total YK00F2 2000 5/28/2004 3 1 4 AK00F2 2000 5/31/2004 2 1 3 YK00F1 2000 6/1/2004 3 YK00F15 2000 6/4/2004 1 2 3 BC00F14 2000 6/7/2004 1 2 3 BC00F18 2000 6/10/2004 4 YK00F10 2000 6/17/2004 1 BC03F02 2003 6/25/2004 BC03F10 2003 6/26/2004 BC03F09 2003 YK00F7 2000 TOTAL 3 4 1 2 2 2 1 1 2 6/29/2004 1 1 2 6/30/2004 1 1 2 18 12 30 11 �Diet Winter diet of lynx was documented through detection of kills found through snow-tracking. In each winter, the most common prey item was snowshoe hare (Lepus americanus), followed by red squirrel (Tamiasciurus hudsonicus) (Table 6). Table 6. Number of kills found each winter field season through snow-tracking of lynx and percent composition of kills of the three primary prey species. Field Season 1999 1999-2000 2000-2001 2001-2002 2002-2003 2003-2004 n 9 81 88 54 65 37 Snowshoe Hare 55.56 67.46 67.41 90.74 90.77 67.56 Prey (%) Red Squirrel Cottontail 22.22 0 19.27 1.20 19.10 8.99 5.56 0 6.15 0 27.03 2.70 Other 22.22 12.05 4.49 3.70 3.08 2.70 DISCUSSION In an effort to establish a viable population of lynx in Colorado, a reintroduction effort was initiated in 1999. The reintroduction effort was augmented with the release of 37 additional animals in April 2004, bringing the total to 166 lynx reintroduced to southwestern Colorado. Locations of each lynx were collected through aerial- or satellite-tracking to document movement patterns and to detect mortalities. Most lynx remain in the southwestern quarter of Colorado. Humancaused mortality factors such as gunshot and vehicle collision are currently the highest causes of death. Reproduction was first documented from the 2003 reproduction season. A second successful breeding season was documented in 2004 with 11 dens and 30 kittens found as of June 30, 2004. Live births are the first step towards recruitment. Recruitment into a population would require these kittens to survive through their first year of life and produce offspring of their own. To achieve a viable population of lynx, enough kittens need to be recruited into the population to offset the mortality that occurs in that year and hopefully even add more so that the population can grow. Snow-tracking of released lynx provided preliminary information on hunting behavior by documenting location of kills, food caches, chases, and diet composition estimated through scat analysis. Snow-tracking results indicate the primary winter prey species are snowshoe hare and red squirrel, with other mammals and birds forming a minor part of the winter diet. Site-scale habitat data collected from snow-tracking efforts indicate Engelmann spruce and subalpine fir are the most common forest stands used by lynx in southwestern Colorado. From results to date it can be concluded that CDOW has developed release protocols that ensure high initial post-release survival, and on an individual level lynx have demonstrated they can survive long-term in areas of Colorado. It has also been documented that reintroduced lynx could exhibit site fidelity, engage in breeding behavior and produce kittens. What is yet to be demonstrated is whether Colorado conditions can support the recruitment necessary to offset annual mortality for a population to sustain itself. Monitoring of reintroduced lynx will continue in an effort to document such viability. 12 �LITURATURE CITED Shenk, T. M. 1999. Program narrative: Post-release monitoring of reintroduced lynx (Lynx canadensis) to Colorado. Job Progress Report for the Colorado Division of Wildlife. Fort Collins, Colorado. __________ 2001. Post-release monitoring of lynx reintroduced to Colorado. Job Progress Report for the Colorado Division of Wildlife. Fort Collins, Colorado. __________ 2003. Post-release monitoring of lynx reintroduced to Colorado. Job Progress Report for the Colorado Division of Wildlife. Fort Collins, Colorado. U. S. Fish and Wildlife Service. 2000. Endangered and threatened wildlife and plants: final rule to list the contiguous United States distinct population segment of the Canada lynx as a threatened species. Federal Register 65, Number 58. Wild, M. A. 1999. Lynx veterinary services and diagnostics. Job Progress Report for the Colorado Division of Wildlife. Fort Collins, Colorado. Prepared by _______________________________ Tanya M. Shenk, Wildlife Researcher 13 �Colorado Division of Wildlife July 2004 – June 2005 WILDLIFE RESEARCH REPORT State of Cost Center Work Package Task No. Federal Aid Project: Colorado 3430 0670 1 : : : : N/A : Division of Wildlife Mammals Research Lynx Conservation Post-Release Monitoring of Lynx Reintroduced to Colorado Period Covered: July 1, 2004- June 30, 2005 Author: T. M. Shenk Personnel: L. Baeten, B. Diamond, R. Dickman, S. Dieterich, D. Freddy, L. Gepfert, R. Kahn, A. Keith, G. Merrill, G. Miller, J. Stewart, C. Wagner, S. Wait, S. Waters, L. Wolfe, D. Younkin All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT In an effort to establish a viable population of lynx (Lynx canadensis) in Colorado, the Colorado Division of Wildlife (CDOW) initiated a reintroduction effort in 1997 with the first lynx released in February 1999. A total of 166 lynx were released from 1999-2004 and an augmentation of 38 additional animals (20 males:18 females) was completed in 2005 resulting in a total of 204 lynx reintroduced to southwestern Colorado. Each lynx was released with dual satellite and VHF radio transmitters to allow intensive monitoring of animals after release. Locations of each lynx were collected through aerial- or satellite-tracking to document movement patterns. Most lynx remain in the southwestern quarter of Colorado. Through documentation of lynx mortalities and causes of death, human-caused mortality factors, such as gunshot and vehicle collision, are currently the highest source of mortality for reintroduced lynx. Reproduction was first documented during the 2003 reproduction season with 6 dens and 16 kittens found. A second successful breeding season was documented in 2004 with 30 kittens found at 11 dens and an addition 9 kittens found after denning season. In 2005, 46 kittens were found at 16 dens with an additional den located but not visited for safety reasons. Data collected from snowtracking indicate the primary winter prey species are snowshoe hare (Lepus americanus) and red squirrel (Tamiasciurus hudsonicus), with other mammals and birds forming a minor part of the winter diet. Sitescale habitat data collected from snow-tracking efforts indicate Engelmann spruce (Picea engelmannii) and subalpine fir (Abies lasiocarpa) are the most common forest stands used by lynx in southwestern Colorado. Results to date have demonstrated that CDOW has developed release protocols that ensure high initial post-release survival, and on an individual level, lynx have demonstrated an ability to survive long-term in areas of Colorado. Reintroduced lynx have also exhibited site fidelity, engaged in breeding behavior and produced kittens. What is yet to be demonstrated is whether conditions in Colorado can support the recruitment necessary to offset annual mortality for a population to remain viable for several generations of lynx. Monitoring of reintroduced lynx will continue in an effort to document such viability. 1 �WILDLIFE RESEARCH REPORT POST RELEASE MONITORING OF LYNX (LYNX CANADENSIS) REINTRODUCED TO COLORADO TANYA M. SHENK P. N. OBJECTIVE The initial post-release monitoring of lynx reintroduced into Colorado will emphasize 5 primary objectives: 1. Assess and modify release protocols to ensure the highest probability of survival for each lynx released. 2. Obtain regular locations of released lynx to describe general movement patterns and habitats used by lynx. 3. Determine causes of mortality in reintroduced lynx. 4. Estimate survival of lynx reintroduced to Colorado. 5. Estimate reproduction of lynx reintroduced to Colorado. Three additional objectives will be emphasized after lynx display site fidelity to an area: 6. Refine descriptions of habitats used by reintroduced lynx. 7. Refine descriptions of daily and overall movement patterns of reintroduced lynx. 8. Describe hunting habits and prey of reintroduced lynx. Information gained to achieve these objectives will form a basis for the development of lynx conservation strategies in the southern Rocky Mountains. SEGMENT OBJECTIVES 1. Release additional adult lynx captured in Canada in southwestern Colorado during spring 2005. 2. Complete winter 2004-05 field data collection on lynx habitat use, hunting behavior, diet, mortalities, and movement patterns. 3. Complete winter 2004-05 lynx trapping field season to collar Colorado born lynx and re-collar adult lynx. 4. Complete spring 2005 field data on lynx reproduction. 5. Summarize and analyze data and publish information as Progress Reports, peer-reviewed manuscripts for appropriate scientific journals, or CDOW technical publications. INTRODUCTION The Canada lynx occurs throughout the boreal forests of northern North America. Colorado represents the southern-most historical distribution of lynx, where the species occupied the higher elevation, montane forests in the state. Little was known about the population dynamics or habitat use of this species in their southern distribution. Lynx were extirpated or reduced to a few animals in the state by the late 1970’s due, most likely, to predator control efforts such as poisoning and trapping. Given the isolation of Colorado to the nearest northern populations, the CDOW considered reintroduction as the only option to attempt to reestablish the species in the state. A reintroduction effort was begun in 1997, with the first lynx released in Colorado in 1999. To date, 204 wild-caught lynx from Alaska and Canada have been released in southwestern Colorado. The goal of the Colorado lynx reintroduction program is to establish a self-sustaining, viable population of 2 �lynx in this state. Evaluation of incremental achievements necessary for establishing viable populations is an interim method of assessing if the reintroduction effort is progressing towards success. There are 7 critical criteria for achieving a viable population: 1) development of release protocols that lead to a high initial post-release survival of reintroduced animals, 2) long-term survival of lynx in Colorado, 3) development of site fidelity by the lynx to areas supporting good habitat in densities sufficient to breed, 4) reintroduced lynx must breed, 5) breeding must lead to reproduction of surviving kittens 6) lynx born in Colorado must reach breeding age and reproduce successfully, and 7) recruitment must be equal to or greater than mortality. The post-release monitoring program for the reintroduced lynx has 2 primary goals. The first goal is to determine how many lynx remain in Colorado and their locations relative to each other. Given this information and knowing the sex of each individual, we can assess whether these lynx can form a breeding core from which a viable population might be established. From these data we can also describe general movement patterns and habitat use. The second primary goal of the monitoring program is to estimate survival of the reintroduced lynx and, where possible, determine causes of mortality for reintroduced lynx. Such information will help in assessing and modifying release protocols and management of lynx once they have been released to ensure their highest probability of survival. Additional goals of the post-release monitoring program for lynx reintroduced to the southern Rocky Mountains included refining descriptions of habitat use and movement patterns and describing successful hunting habitat once lynx established home ranges that encompassed their preferred habitat. Specific objectives for the site-scale habitat data collection include: 1) describe and quantify site-scale habitat use by lynx reintroduced to Colorado, 2) compare site-scale habitat use among types of sites (e.g., kills vs. long-duration beds), and 3) compare habitat features at successful and unsuccessful snowshoe hare chases. The program will also investigate the ecology of snowshoe hare in Colorado. Documenting reproduction is critical to the success of the program and lynx are monitored intensively to document breeding, births, survival and recruitment of lynx born in Colorado. Site-scale habitat descriptions of den sites are also collected and compared to other sites used by lynx. Lynx is listed as threatened under the Endangered Species Act (ESA) of 1973, as amended (16 U. S. C. 1531 et. seq.)(U. S. Fish and Wildlife Service 2000). Colorado is included in the federal listing as lynx habitat. Thus, an additional objective of the post-release monitoring program is to develop conservation strategies relevant to lynx in Colorado. To develop these conservation strategies, information specific to the ecology of the lynx in its southern Rocky Mountain range, such as habitat use, movement patterns, mortality factors, survival, and reproduction in Colorado is needed. STUDY AREA Southwestern Colorado is characterized by wide plateaus, river valleys, and rugged mountains that reach elevations over 4200 m. Engelmann spruce-subalpine fir is the most widely distributed coniferous forest type at elevations most typically used by lynx. The Core Research Area is defined as areas bounded by the New Mexico state line to the south, Taylor Mesa to the west and Monarch Pass on the north and east and > 2900 meters in elevation. METHODS REINTRODUCTION Effort All 2005 lynx releases were conducted under the protocols found to maximize survival (see Shenk 2001). Estimated age, sex and body condition were ascertained and recorded for each lynx prior to 3 �release (see Wild 1999). Specific release sites were those used in earlier years of the project and were selected based on land ownership and accessibility during times of release (Byrne 1998). Lynx were transported from the Frisco Creek Wildlife Rehabilitation Center, where they were held from their time of arrival in Colorado, to their release site in individual cages. Release site location was recorded in Universal Transverse Mercator (UTM) coordinates and identification of all lynx released at the same location, on the same day, was recorded. Behavior of the lynx on release and movement away from the release site were documented. Distribution and Movement Patterns All lynx released in 1999 were fitted with TelonicsTM radio-collars. All lynx released since 1999, with the exception of 5 males released in spring 2000, were fitted with SirtrackTM dual satellite/VHF radio-collars. These collars have a mortality indicator switch that operated on both the satellite and VHF mode. The satellite component of each collar was programmed to be active for 12 hours per week. The 12-hour active periods for individual collars were staggered throughout the week. Signals from the collars allowed for locations of the animals to be made via Argos, NASA, and NOAA satellites. The location information was processed by ServiceArgos and distributed to the CDOW through e-mail messages. To determine general movement patterns of reintroduced lynx, regular locations of released lynx were collected through a combination of aerial, satellite and ground radio-tracking. Locations were recorded in UTM coordinates and general habitat descriptions for each ground and aerial location were recorded. Survival and Mortality Factors When a mortality signal (75 beats per minute [bpm] vs. 50 bpm for the Telonics™ VHF transmitters, 20 bpm vs. 40 bpm for the Sirtrack™ VHF transmitters, 0 activity for Sirtrack™ PTT) was heard during either satellite, aerial or ground surveys, the location (UTM coordinates) was recorded. Ground crews then located and retrieved the carcass as soon as possible. The immediate area was searched for evidence of other predators and the carcass photographed in place before removal. Additionally, the mortality site was described and habitat associations and exact location were recorded. Any scat found near the dead lynx that appeared to be from the lynx was collected. All carcasses were transported to the Colorado State University Veterinary Teaching Hospital (CSUVTH) for a post mortem exam to 1) determine the cause of death and document with evidence, 2) collect samples for a variety of research projects, and 3) archive samples for future reference (research or forensic). The gross necropsy and histology were performed by, or under the lead and direct supervision of a board certified veterinary pathologist. At least one research personnel from the CDOW involved with the lynx program was also present. The protocol followed standard procedures used for thorough post-mortem examination and sample collection for histopathology and diagnostic testing (see Shenk 1999 for details). Some additional data/samples were routinely collected for research, forensics, and archiving. Other data/samples were collected based on the circumstances of the death (e.g., photographs, video, radiographs, bullet recovery, samples for toxicology or other diagnostic tests, etc.). From 1999–2004 the CDOW retained all samples and carcass remains with the exception of tissues in formalin for histopathology, brain for rabies exam, feces for parasitology, external parasites for ID, and other diagnostic samples. Since 2005 carcasses are disposed of at the CSUVTH with the exception of the lower canine, fecal samples, stomach content samples and tissue or bone marrow samples to be delivered by CDOW to the Center for Disease control for plague testing. The lower canine is sent to Matson Labs (Missoula, Montana) for aging and the fecal and stomach content samples are evaluated for diet. 4 �Reproduction Females were monitored for proximity to males during each breeding season. We defined a possible mating pair as any male and female documented within at least 1 km of each other in breeding season through either flight data or snow-tracking data. Females were then monitored for site fidelity to a given area during each denning period of May and June. Each female that exhibited stationary movement patterns in May or June were closely monitored to locate possible dens. Dens were found when field crews walked in on females that exhibited virtually no movement for at least 10 days from both aerial and ground telemetry. Kittens found at den sites were weighed, sexed and photographed. Each kitten was uniquely marked by inserting a sterile passive integrated transponder (PIT, Biomark, Inc., Boise, Idaho, USA) tag subcutaneously between the shoulder blades. Time spent at the den was minimized to ensure the least amount of disturbance to the female and the kittens. Weight, PIT-tag number, sex and any distinguishing characteristics of each kitten was also recorded. Beginning in 2005, blood and saliva samples were collected for genetic identification. During the den site visits, den site location was recorded as UTM coordinates. General vegetation characteristics, elevation, weather, field personnel, time at the den, and behavioral responses of the kittens and female were also recorded. Once the females moved the kittens from the natal den area, den sites were visited again and site-specific habitat data were collected (see Habitat Use section below). Recaptures Recaptures were attempted for either lynx that were in poor body condition or lynx that needed to have their radio-collars replaced due to failed or failing batteries or to radio-collar kittens born in Colorado once they reached at least 10-months of age when they were nearly adult size. Methods of recapture included 1) trapping using a Tomahawk™ live trap baited with a rabbit and visual and scent lures, 2) calling in and darting lynx using a Dan-Inject CO2 rifle, 3) custom box-traps modified from those designed by other lynx researchers (Kolbe et al. 2003) and 4) hounds trained to pursue felids were also used to tree lynx and then the lynx was darted while treed. Lynx were immobilized either with Telazol (3 mg/kg; modified from Poole et al. 1993 as recommended by M. Wild, DVM) or medetomidine (0.09mg/kg) and ketamine (3 mg/kg; as recommended by L. Wolfe, DVM)) administered intramuscularly (IM) with either an extendible pole-syringe or a pressurized syringe-dart fired from a Dan-Inject air rifle. Immobilized lynx were monitored continuously for decreased respiration or hypothermia. If a lynx exhibited decreased respiration 2mg/kg of Dopram was administered under the tongue; if respiration was severely decreased, the animal was ventilated with a resuscitation bag. If medetomidine/ketamine were the immobilization drugs, the antagonist Atipamezole hydrochloride (Antisedan) was administered. Hypothermic (body temperature < 95o F) animals were warmed with hand warmers and blankets. While immobilized, lynx were fitted with replacement SirtrackTM VHF/satellite collar and blood and hair samples were collected. Once an animal was processed, recovery was expedited by injecting the equivalent amount of the antagonist Antisedan IM as the amount of medetomidine given, if medetomodine/ketemine was used for immobilization. Lynx were then monitored while confined in the box-trap until they were sufficiently recovered to move safely on their own. No antagonist is available for Telezol so lynx anesthetized with this drug were monitored until the animal recovered on its own in the box-trap and then released. If captured and in poor body condition lynx were anesthetized with Telezol (2 mg/kg) and returned to the Frisco Creek Wildlife Rehabilitation Center for treatment. HABITAT USE Gross habitat use was documented by recording canopy vegetation at aerial locations. More refined descriptions of habitat use by reintroduced lynx were obtained through following lynx tracks in 5 �the snow (i.e., snow-tracking) and site-scale habitat data collection conducted at sites found through this method to be used by lynx. Snow-tracking Locations from aerial- and satellite-tracking were used to help ground-trackers locate lynx tracks in snow. Snowmobiles, where permitted, were used to gain the closest possible access to the lynx tracks without disturbing the animal. From that point, the tracking team used snowshoes to access tracks. Once tracks were found, the ground crew back- or forward-tracked the animal if it was far enough away not to be disturbed. Back-tracking generally avoided the possibility of disturbing the lynx by moving away from the animal rather than towards the animal. However, monitoring of the lynx through radio-telemetry was used to assure that the ground crew was staying a sufficient distance away from the lynx in the event the lynx might double back on its tracks. Radio-telemetry was also used in forward-tracking to make sure the team did not disturb the animal. If it appeared the lynx began to move in response to the observers, the observers stopped following the tracks. If the lynx began to move and the movement did not appear to be a response to the observers, the ground crew continued following the track. An attempt was made in Season 1 (February-May 1999) and Season 2 (December 1999-April 2000) to snow-track each lynx. In Season 3 (December 2000-April 2001), we attempted to snow-track all lynx within the Core Research Area. In tracking Season 4 (December 2001-April 2002), Season 5 (December 2002-April 2003), Season 6 (December 2003-April 2004) and Season 7 (December 2004April 2005) we attempted to track all accessible lynx in the Core Research Area and some lynx north of the Core Research Area. Ground crews were instructed to track lynx only where it was safe to travel. Restrictions to safe travel included avalanche danger and extremely rugged terrain. Ground crews worked in pairs and were fully equipped for winter back-country survival. Data Collection For each day of tracking the date, lynx being tracked, slope, aspect, UTM coordinates, elevation, general habitat description, and summary of the days tracking were recorded. Aspect was defined as the direction of 'downhill' or 'fall line' on a slope. This is the direction along the ground in a dihedral angle between the horizontal and the plane of the ground surface. Units were compass degrees. Slope was defined as the dihedral angle between the horizontal and the plane of the ground surface (e.g., 45"). Once a track was located there were 2 types of 'sites' that were encountered. Site I areas needed documentation but either did not reflect areas lynx selected for specific habitat features, or were sites that occurred too frequently to measure each in detail. These sites included the start and end of the track being followed, the location of scat, and short-duration beds defined as being small in size (approximating an area a lynx would crouch), and with little ice formed in the bed indicating little time spent there. Site II areas included areas that might reflect specific habitat features lynx selected for and included locations where the following were found: kills, start of chases, territory marks (e.g., spray sites, buried scat, scat placed on prominent locations), long-duration beds (encompasses an area where a lynx would have lain for an extended period, iced bottom), and road crossing (both sides of road). In addition, habitat plots were conducted along lynx travel routes if no other sites sampled in last hour. At each of the 2 types of sites the date, lynx tracked, slope, aspect, forest structure class, UTM coordinates, and elevation were recorded. Forest structure classes included grass/forb, shrub/seedling, sapling/pole, mature, and old growth as defined in Table 1. For Site I areas, the only additional data that was collected was identification of what the site was used for (e.g., short-duration bed), and a brief description of the site. Habitat plots (see below) were conducted at Site II areas. 6 �Description of the Habitat Plot The habitat plot consisted of a 12 m x 12 m square defined by a series of 25 points placed in 5 rows of 5 with the center point being on the object that defined the site (e.g., a kill)(Figure 1). Each point was 3 m apart. The 12 m x 12 m sampling square exceeded the minimum requirement of 0.01 ha. recommended by Curtis (1959) for sampling trees. Measurements taken at each of the 25 points included: 1. Snow depth - measured vertically by an avalanche probe marked in cm. 2. Understory - measured from top of snow to 150 cm above snow in a column of 3-cm radius around the avalanche probe. Because understory measurements were influenced by vegetation outside the perimeter of the 25 sampling points (12 m x 12 m) the area used for estimating undersory cover was 15 m by 15 m. At each point, crews recorded all shrubs, trees and coarse woody debris (CWD) that fell within this column and was visible above the snow. Crews also recorded number of branches of each species that fell within the column at 3 different height categories (0-0.5 m, 0.51-1.0 m, 1.01-1.5 m). 3. Overstory: measured at 150 cm above snow with a sighting tube. The tube was made of PVC pipe, with a curved viewing end and a crosshair made of wire on the opposite end. The sighting tube was attached to the avalanche probe used to measure snow depth. Species that hit the crosshair were recorded at each of the 25 points in the vegetation plot. Ganey and Block (1994) found this method of measuring canopy cover (with 20 sample points per plot; Laymon 1988) provided greater precision among observers. 4. Species composition: all the different species of tree or shrub that hit the crosshair of the sighting tube at each of the 25 points were recorded. 5. Tree composition of the vegetation plot was recorded by species and diameter at breast height (DBH). Snow depth was used in conjunction with this recorded DBH to estimate true DBH. Within the 12 m x 12 m square all conifers and deciduous trees were recorded by DBH size class (A = 0-6 in, B = 6.1-12 in, C = 12.1 -18 in, D = 18.1-24 in, E = > 24 in). Area for the tree composition analysis was 12 m x 12 m. Understory was estimated as: 1) percent occurrence within the vegetation plot (number of points with understory/total number of points surveyed) and 2) mean percent occurrence and variance by species and height category over the total points sampled within the vegetation plot. Overstory was estimated as percent occurrence over the vegetation plot (number of points with overstory/total number of points surveyed). DIET AND HUNTING BEHAVIOR Winter diet of reintroduced lynx was estimated by documenting successful kills through snowtracking. Prey species from failed and successful hunting attempts were identified by either tracks or remains. Scat analysis also provided information on foods consumed. Scat samples were collected wherever found and labeled with location and individual lynx identification. Only part of the scat was collected (approximately 75%); the remainder was left in place in the event that the scat was being used by the animal as a territory mark. Site-scale habitat data collected for successful and unsuccessful snowshoe hare kills were compared. RESULTS REINTRODUCTION Effort From 1999 through 2004 166 lynx were reintroduced into southwestern Colorado. An additional 37 lynx were released in April 2005 (17 females and 20 males), one female was released in June 2005. This brings the total number of lynx released in Colorado to 204 (Table 2). These lynx released in 2005 were captured in Quebec, British Columbia and Manitoba. All lynx were released in the Core Research 7 �Area of southwestern Colorado at or near previously used release sites in southwestern Colorado. Lynx were released with dual VHF/satellite radio collars so they can be monitored for movement and mortality. The CDOW plans to release up to 15 lynx annually from 2006-2008. Distribution and Movement Patterns A total of 7421 aerial VHF locations for all 204 reintroduced lynx have been collected to date. An additional 14,788 satellite locations have been collected. Most lynx released remained in the southwestern quarter of Colorado. The majority of surviving lynx from the entire reintroduction effort continue to use areas from New Mexico north to Gunnison, west as far as Taylor Mesa and east to Monarch Pass. Most movements away from the Core Research Area were to the north. Numerous travel corridors have been used repeatedly by more than one lynx. These travel corridors include the Cochetopa Hills area for northerly movements, the Rio Grande Reservoir-SilvertonLizardhead Pass for movements to the west, and southerly movements down the east side of Wolf Creek Pass to the southeast through the Conejos River Valley. Lynx appear to remain faithful to an area during winter months, and exhibit more extensive movements away from these areas in the summer. Such movement patterns have also been documented by native lynx in Wyoming and Montana (Squires and Laurion 1999). Survival and Mortality Factors Of the total 204 adult lynx released from 1999-2005 there are 66 known mortalities. Of these 66 mortalities, 26 are from the 1999 releases, 24 are from the 2000 releases, 5 are from the 2003 releases, 8 are from the 2004 releases, and 3 are from the 2005 releases. Causes of death are listed in Table 3. Starvation was a significant cause of mortality in the first year of releases only. Mortalities occurred throughout the areas through which lynx moved. As of June 30, 2005, CDOW was actively tracking 110 of the 138 lynx still possibly alive. There are 29 lynx that we have not heard signals on since at least June 30, 2004 and these animals are classified as ‘missing’ (Table 4). One of these missing lynx is a mortality of unknown identity, thus only 28 are truly missing. Possible reasons for not locating these missing lynx include 1) long distance dispersal, beyond the areas currently being searched, 2) radio failure, or 3) destruction of the radio (e.g., run over by car). CDOW continues to search for all missing lynx during both aerial and ground searches. Two of the missing lynx released in 2000 are thought to have slipped their collars. Reproduction 2003.-- Nine pairs of lynx were documented during the 2003 breeding season (March and April). In May and June, 6 dens and a total of 16 kittens were found in the lynx Core Research Area in southwestern Colorado (Table 5). At all dens the females appeared in excellent condition, as did the kittens. The kittens weighed from 270-500 grams. Lynx kittens weigh approximately 200 grams at birth and do not open their eyes until they are 10-17 days old. The dens were scattered throughout the Core Research Area, with no dens found outside the core area. All the dens were in Engelmann spruce/subalpine fir forests in areas of extensive downfall. Elevations ranged from 3240-3557 m. Field crews weighed, photographed, PIT-tagged the kittens and . took hair samples from the kittens for genetic work in an attempt to confirm paternity. Kittens were processed as quickly as possible (11-32 minutes) to minimize the time the kittens were without their mother. While working with the kittens the females remained nearby, often making themselves visible to the field crews. The females generally continued a low growling vocalization the entire time personnel were at the den. In all cases, the female returned to the den site once field crews left the area. 8 �Four of the 6 females that we know had kittens in summer 2003 were still with kittens at the end of April 2004. Two of those females still had 2 kittens with them at that time. Visual observations in February 2004 of one female with 2 kittens indicated all 3 were in good body condition. The mortality of female YK00F16 and her 1 kitten in October 2003 from plague was not due to poor habitat or prey conditions, and thus we might assume she would have raised the 1 kitten to this stage as well. Three probable kitten deaths from female YK00F19 were from 1 litter that most likely failed very early. Through snow-tracking in winter 2003-04 an unknown female (no radio frequency heard in the area of the tracks) we also documented 1-2 additional kittens born spring 2003 and still alive in winter 2004. Of the 16 kittens we found in summer 2003, we documented the following by April 2004: 6 confirmed alive, 7 confirmed dead, and 3 some evidence dead. Although we tried, we were not able to capture any of the 6 surviving kittens to fit them with radio-collars. 2004.-- In Spring 2004, 26 females from the releases in 1999, 2000 and 2003 had active radiocollars. Of these, we documented 18 possible mating pairs of lynx during breeding season. All 4 of the females that had kittens with them through winter 2003-04 bred again spring 2004, 2 with the same male they successfully bred with spring 2003. During May-June 2004 we found 11 dens and a total of 30 kittens (Table 6). At all dens the females appeared in excellent condition, as did the kittens. The kittens weighed from 250-770 grams. Three of the 11 females with kittens were from the 2003 releases (Table 6). Three additional litters were documented after denning season through either observation of a female lynx with kittens or snow-tracking females with kittens that were not one of the 11 females found on dens. From the size of the kittens they would have been born during the normal denning season in May or June. Nine additional kittens were observed from these litters for a total of 39 known kittens born in 2004. Two of these additional litters were documented from direct follow-ups to sighting made by the public and reported to CDOW. Two females that had kittens in 2003 and reared at least part of their litters through March 2004, bred and had kittens again in 2004. Two of the litters documented by direct observation or snow-tracking are from females whose collars no longer work. Seven kittens born in 2004 were captured at 10-months of age and fitted with dual satellite/VHF collars. All 7 are alive and currently being monitored. 2005.-- In spring 2005 we had 34 females from the releases in 1999, 2000, 2003 and 2004 that had active radio-collars. We documented 23 possible mating pairs of lynx during breeding season. During May-June 2005 we visited 16 dens and found a total of 46 kittens (Table 7). At all dens the females appeared in excellent condition, as did the kittens. An additional female had a den we were not able to get to during May or June due to high water. Female BC03F03 was hit and killed on I70 on 5/19/2005. She had 2 fetuses in her uterus, so would have contributed to reproduction this year had she lived. We weighed, photographed, PIT-tagged the kittens and recorded sex. We also took blood samples from the kittens for genetic work in an attempt to confirm paternity. While we were working with the kittens the females remained nearby, often remaining visible to us. The females generally continued a low growling vocalization the entire time we were at the den. In all cases, the female returned to the den site once we left the area. All of the 2005 dens were scattered throughout the high elevation areas of Colorado, south of Interstate 70. Most of the dens were in Engelmann spruce/subalpine fir forests in areas of extensive downfall. Elevations ranged from 3117-3586 m. We weighed, photographed, and PIT-tagged the kittens, recorded sex and took hair samples from the kittens for genetic work in an attempt to confirm paternity. Four of the females would not leave the den until we reached out to pick up a kitten. While we were working with the kittens the females remained nearby, often remaining visible to us. The females 9 �generally continued a low growling vocalization the entire time we were at the den. In all cases, the female returned to the den site once we left the area. One female, YK00F10 has had litters 3 years in a row. In 2003 she had 4 kittens and raised 2 through the winter. In 2004 she had 2 kittens and raised both through the winter, this year she had 4 kittens again. She has had all 3 litters in the same general area and has had the same mate for 3 years. Eight additional females had a second litter in Colorado this year. Three females from the 2004 releases had litters in 2005. This is the second year in a row we had females released the prior spring, find a territory and a mate within a year and produced live young. In reproduction season 2004 we had 3 females released in spring 2003 that did the same thing. Of those 3, 2 successfully raised at least part of their litters through winter 2005. Den Sites.--A total of 33 dens have been found. All of the dens except one have been scattered throughout the high elevation areas of Colorado, south of I-70. One den was found in southeastern Wyoming, near the Colorado border. Dens were located on steep ( x slope = 29o), north-facing, high elevation ( x = 3347 m) slopes (Figure 2). The dens were typically in Engelmann spruce/subalpine fir forests in areas of extensive downfall (Figures 3, 4, 5). Recaptures Two adult lynx were captured in 2001 for collar replacement. One lynx was captured in a tomahawk live-trap, the other was treed by hounds and then anesthetized using a jab pole. Five adult lynx were captured in 2002; 3 were treed by hounds and 2 were captured in padded leghold traps. In 2004, 1 lynx was captured with a Belisle snare and 6 other adult lynx were captured in box-traps. Trapping effort was substantially increased in winter and spring 2005 and 12 adult lynx were captured and re-collared. In addition, 7 kittens born in Colorado in 2004 were also captured and collared. All lynx captured in 2005 were caught in box-traps. All captured lynx were fitted with new Sirtrack TM dual VHF/satellite collars. Six adult lynx were captured from March 1999-June 30, 2005 because they were in poor body condition. Five of these lynx were successfully treated at the Frisco Creek Rehabilitation Center and rereleased in the Core Research Area. One lynx, BC00F7, died from starvation and hypothermia. Two lynx were captured because they were in atypical habitat outside the state of Colorado. They were held at Frisco Creek Rehabilitation Center for a minimum of 3 weeks and re-released in the Core Research Area in Colorado. Prior to release these lynx were fitted with new Sirtrack TM dual VHF/satellite collars. HABITAT USE Landscape-scale daytime habitat use was documented from 7421 aerial locations of lynx collected from February 1999-June 30, 2005. Throughout the year Engelmann spruce / subalpine fir was the dominant cover used by lynx. A mix of Engelmann spruce, subalpine fir and aspen (Populus tremuloides) was the second most common cover type used throughout the year. Various riparian and riparian-mix areas were the third most common cover type where lynx were found during the daytime flights. Use of Engelmann spruce-subalpine fir forests and Engelmann spruce-subalpine fir-aspen forests was similar throughout the year. There was a trend in increased use of riparian areas beginning in July, peaking in November, and dropping off December through June. Site-scale habitat data collected from snow-tracking efforts indicate Engelmann spruce and subalpine fir were also the most common forest stands used by lynx for all activities during winter in southwestern Colorado. Comparisons were made among sites used for long beds, dens, travel and where they made kills. Little difference in aspect, mean slope and mean elevation were detected for 3 of the 4 site types including long beds, travel and kills where lynx typically use gentler slopes ( x = 15.7o ) at an mean elevation of 3173 m, and varying aspects with a slight preference for north-facing slopes (Figure 2). 10 �Den sites however, were located at higher elevations ( x = 3347 m), steeper slopes ( x = 29o) and more commonly on north-facing slopes (Figure 2). Mean percent total overstory was higher for long bed and kill sites than travel or den sites (Figure 3). Engelmann spruce provided a mean of 35.87% overstory for kills and long beds, with travel sites averaging 28% and den sites having the lowest mean percent overstory of 23% (Figure 3). Mean percent subalpine fir or aspen overstory did not vary across use sites (Figure 3). Willow overstory was highly variable and no dens were located in willow overstory. A total of 1841 site-scale habitat plots were completed in winter from December 2002 through April 2005. The most common understory species at all 3 height categories above the snow (low = 00.5m, medium = 0.51 - 1.0 m, high = 1.1 - 1.5 m) was Engelmann spruce, subalpine fir, willow (Salix spp.) and aspen (Figure 4). Various other species such as Ponderosa pine (Pinus ponderosa), lodgepole pine (Pinus contorta), cottonwood (Populus sargentii), birch (Betula spp.) and others were also found in less than 5% of the habitat plots. If present, willow provided the greatest percent cover within a plot followed by Engelmann spruce, subalpine fir, aspen and coarse woody debris for long beds, kills and travel sites. Areas documented in willow used by lynx are typically on the edge of willow thickets as tracks are quickly lost within the thicket. Den sites had significantly higher percent understory cover for all three height categories. Understory at den sites was primarily made up of coarse woody debris (Figure 3). The most common tree species documented in the site-scale habitat plots was Engelmann spruce Figure 5). Subalpine fir and aspen were also present in >35% of the plots. Most habitat plots were vegetated with trees of DBH < 6" (Figure 5). As DBH increased, percent occurrence decreased within the plot. Although decreasing in abundance as size increased, most lynx use sites had trees in each of the DBH categories, indicating mature forest stands except for dens. Den sites had a broad spectrum of Engelmann spruce tree sizes, including > 18” but no large subalpine fir or aspen trees. While Engelmann spruce and subalpine fir occurred in similar densities for kills, long beds and travel sites, den sites had twice the density of subalpine firs found at all other sites (Figure 5). DIET AND HUNTING BEHAVIOR Winter diet of lynx was documented through detection of kills found through snow-tracking. Prey species from failed and successful hunting attempts were identified by either tracks or remains. Scat analysis also provided information on foods consumed. A total of 400 kills were located from February 1999-April 2005. We collected 671 scat samples from February 1999-April 2004 that will be analyzed for content. In each winter, the most common prey item was snowshoe hare, followed by red squirrel (Table 8). A comparison of percent overstory for successful and unsuccessful snowshoe hare chases indicated lynx were more successful at sites with slightly higher percent overstory, if the overstory species were Englemann spruce, subalpine fir or willow. Lynx were slightly less successful in areas of greater aspen overstory (Figure 6). This trend was repeated for percent understory at all 3 height categories (Figure 7) except that higher aspen understory improved hunting success. Higher density of Engelmann spruce and subalpine fir increased hunting success while increased aspen density decreased hunting success (Figure 8). DISCUSSION In an effort to establish a viable population of lynx in Colorado, CDOW initiated a reintroduction effort in 1997 with the first lynx released in winter 1999. From 1999 through spring 2004, 166 lynx were released in the Core Research Area. The reintroduction effort was augmented with the release of 37 11 �additional animals in April 2005 and 1 in June 2005, bringing the total to 204 lynx reintroduced to southwestern Colorado. Locations of each lynx were collected through aerial- or satellite-tracking to document movement patterns and to detect mortalities. Most lynx remain in the southwestern quarter of Colorado. Dispersal movement patterns for lynx released in 2000 and subsequent years were similar to those of lynx released in 1999. However, more animals released in 2000 and subsequent years remained within the Core Research Area than those released in 1999. This increased site fidelity may have been due to the presence of con-specifics in the area on release. Numerous travel corridors have been used repeatedly by more than 1 lynx. These travel corridors include the Cochetopa Hills area for northerly movements, the Rio Grande Reservoir-Silverton-Lizardhead Pass for movements to the west, and southerly movements down the east side of Wolf Creek Pass to the southeast to the Conejos River Valley. Lynx appear to remain faithful to an area during winter months, and exhibit more extensive movements away from these areas in the summer. Most lynx currently being tracked are within the Core Research Area. During the summer months, lynx were documented to make extensive movements away from their winter use areas. Extensive summer movements away from areas used throughout the rest of the year have been documented in native lynx in Wyoming and Montana (Squires and Laurion 1999). Human-caused mortality factors such as gunshot and vehicle collision are currently the highest causes of death. Reproduction is critical to achieving a self-sustaining viable population of lynx in Colorado. Reproduction was first documented from the 2003 reproduction season and again in 2004 and 2005. Additional reproduction is likely to have occurred in females we are no longer tracking, and from Colorado born lynx that have not been collared. The dens we find are more representative of the minimum number of litters and kittens in a reproduction season. Live-births and over-winter survival of kittens are the first steps towards recruitment into the breeding population defined as when these Colorado-born lynx will produce offspring of their own. To achieve a viable population of lynx, enough kittens need to be recruited into the population to offset the mortality that occurs in that year and hopefully even exceed the mortality rate for an increasing population. Mowat et al. (1999) suggest lynx and snowshoe hare select similar habitats except that hares select more dense stands than lynx. Very dense understory limits hunting success of the lynx and provides refugia for hares. Given the high proportion of snowshoe hare in the lynx diet in Colorado, we might then assume the habitats used by reintroduced lynx also depict areas where snowshoes hare are abundant and available for capture by lynx in Colorado. From both aerial locations taken throughout the year and from the site-scale habitat data collected in winter, the most common areas used by lynx are in stands of Engelmann spruce and subalpine fir. This is in contrast to adjacent areas of Ponderosa pine, pinyon juniper, aspen and oakbrush. The lack of lodgepole pine in the areas used by the lynx may be more reflective of the limited amount of lodgepole pine in southwestern Colorado, the Core Research Area, rather than avoidance of this tree species. Hodges (1999) summarized habitats used by snowshoe hare from 15 studies as areas of dense understory cover from shrubs, stands that are densely stocked, and stands at ages where branches have more lateral cover. Species composition and stand age appears to be less correlated with hare habitat use than is understory structure (Hodges 1999). The stands need to be old enough to provide dense cover and browse for the hares and cover for the lynx. In winter, the cover/browse needs to be tall enough to still provide browse and cover in average snow depths. Hares also use riparian areas and mature forests with understory. Site-scale habitat use documented for lynx in Colorado indicate lynx are most commonly using areas with Engelmann spruce understory present from the snow line to at least 1.5 m above the snow. The mean percent understory cover within the habitat plots is typically less than 15% regardless of understory species. However, if the understory species is willow, percent understory cover is typically 12 �double that, with mean number of shrubs per plot approximately 80, far greater than for any other understory species. In winter, hares browse on small diameter woody stems (<0.25"), bark and needles. In summer, hares shift their diet to include forbs, grasses, and other succulents as well as continuing to browse on woody stems. This shift in diet may express itself in seasonal shifts in habitat use, using more or denser coniferous cover in winter than in summer. The increased use of riparian areas by lynx in Colorado from July to November may reflect a seasonal shift in hare habitat use in Colorado. Major (1989) suggested lynx hunted the edge of dense riparian willow stands. The use of these edge habitats may allow lynx to hunt hares that live in habitats normally too dense to hunt effectively. The use of riparian areas and riparian-Engelmann spruce-subalpine fir and riparian-aspen mixes documented in Colorado may stem from a similar hunting strategy. However, too little is known about habitat use by hares in Colorado to test this hypothesis at this time. Lynx also require sufficient denning habitat. Denning habitat has been described by Koehler (1990) and Mowat et al. (1999) as areas having dense downed trees, roots, or dense live vegetation. We found this to be in true in Colorado as well. In addition, the dens used by reintroduced lynx were at high elevation, steep north-facing slopes. All females that were documented with kittens denned in areas within their winter-use area. Snow-tracking of released lynx provided information on hunting behavior and diet through documentation of kills, food caches, chases, and diet composition estimated through prey remains. Snowtracking results indicate the primary winter prey species are snowshoe hare and red squirrel, with other mammals and birds forming a minor part of the winter diet. In winter, lynx reintroduced to Colorado appear to be feeding on their preferred prey species, snowshoe hare and red squirrel in similar proportions as those reported for northern lynx during lows in the snowshoe hare cycle (Aubry et al., 1999). Caution must be used in interpreting the proportion of identified kills. Such a proportion ignores other food items that are consumed in their entirety and thus are biased towards larger prey and may not accurately represent the proportion of smaller prey items, such as microtines, in lynx winter diet. Through snowtracking we have evidence that lynx are mousing and several of the fresh carcasses have yielded small mammals in the gut on necropsy. The summer diet of lynx has been documented to include less snowshoe hare and more alternative prey than in winter (Mowat et al., 1999). All evidence suggests reintroduced lynx are finding adequate food resources. SUMMARY From results to date it can be concluded that CDOW has developed release protocols that ensure high initial post-release survival, and on an individual level lynx have demonstrated they can survive long-term in areas of Colorado. It has also been documented that reintroduced lynx could exhibit site fidelity, engage in breeding behavior and produce kittens. What is yet to be demonstrated is whether current conditions in Colorado can support the recruitment necessary to offset annual mortality for a population to sustain itself. Monitoring of reintroduced lynx will continue in an effort to document such viability. 13 �ACKNOWLEDGEMENTS The lynx reintroduction program involves the efforts of literally hundreds of people across North America, in Canada and the U. S. Any attempt to properly acknowledge all the people who played a role in this effort is at risk of missing many people. The following list should be considered to be incomplete. CDOW CLAWS Team (1998-2001): Bill Andree, Tom Beck, Gene Byrne, Bruce Gill, Mike Grode, Rick Kahn (Program Leader), Dave Kenvin, Todd Malmsbury, Jim Olterman, Dale Reed, John Seidel, Scott Wait, Margaret Wild. CDOW: John Mumma (Director 1996-2000), Conrad Albert, Jerry Apker, Laurie Baeten, Cary Carron, Don Crane, Larry DeClaire, Phil Ehrlich, Lee Flores, Delana Friedrich, Dave Gallegos, Juanita Garcia, Drayton Harrison, Jon Kindler, Ann Mangusso, Jerrie McKee, Melody Miller, Mike Miller, Kirk Navo, Robin Olterman, Jerry Pacheo, Mike Reid, Ellen Salem, Eric Schaller, Mike Sherman, Jennie Slater, Steve Steinert, Kip Stransky, Suzanne Tracey, Anne Trainor, Brad Weinmeister, Nancy Wild, Perry Will, Lisa Wolfe, Brent Woodward, Kelly Woods, Kevin Wright. Lynx Advisory Team (1998-2001): Steve Buskirk, Jeff Copeland, Dave Kenny, John Krebs, Brian Miller (Co-leader), Mike Phillips, Kim Poole, Rich Reading (Co-leader), Rob Ramey, John Weaver. U. S. Forest Service: Kit Buell, Joan Friedlander, Dale Gomez, Jerry Mastel, John Squires, Fred Wahl, Nancy Warren. U. S. Fish and Wildlife Service: Lee Carlson, Gary Patton (1998-2000), Kurt Broderdorp. State Agencies: Gary Koehler (Washington). National Park Service: Steve King. Colorado State University: Alan B. Franklin, Gary C. White. Colorado Natural Heritage Program: Rob Schorr, Mike Wunder. Alaska: ADF&G: Cathie Harms, Mark Mcnay, Dan Reed (Regional Manager), Wayne Reglin (Director), Ken Taylor (Assist. Director), Ken Whitten, Randy Zarnke, Other:Ron Perkins (trapper), Dr. Cort Zachel (veterinarian). British Columbia: Dr. Gary Armstrong (veterinarian), Mike Badry (government), Paul Blackwell (trapper coordinator), Trappers: Dennis Brown, Ken Graham, Tom Sbo, Terry Stocks, Ron Teppema, Matt Ounpuu. Yukon: Government: Arthur Hoole (Director), Harvey Jessup, Brian Pelchat, Helen Slama, Trappers: Roger Alfred, Ron Chamber, Raymond Craft, Lance Goodwin, Jerry Kruse, Elizabeth Hofer, Jurg Hofer, Guenther Mueller (YK Trapper’s Association), Ken Reeder, Rene Rivard (Trapper coordinator), Russ Rose, Gilbert Tulk, Dave Young. Alberta: Al Cook. Northwest Territories: Albert Bourque, Robert Mulders (Furbearer Biologist), Doug Steward (Director NWT Renewable Res.), Fort Providence Native People. Quebec: Luc Farrell, Pierre Fornier. Colorado Holding Facility: Herman and Susan Dieterich, Loree Harvey, Rachel Riling. Pilots: Dell Dhabolt, Larry Gepfert, Al Keith, Jim Olterman, Matt Secor, Whitey Wannamaker, Steve Waters, Dave Younkin. Field Crews (1999-2005): Steve Abele, Brandon Barr, Bryce Bateman, Todd Bayless, Ryan Besser, Mandi Brandt, Brad Buckley. Patrick Burke, Paula Capece, Stacey Ciancone, Doug Clark, John DePue, Shana Dunkley, Tim Hanks, Matt Holmes, Andy Jennings, Susan Johnson, Paul Keenlance, Tony Lavictoire, Clay Miller, Denny Morris, Kieran O’Donovan, Gene Orth, Chris Parmater, Jake Powell, Jeremy Rockweit, Josh Smith, Adam Strong, Dave Unger, David Waltz, Andy Wastell, Lyle Willmarth, Leslie Witter, Kei Yasuda, Jennifer Zahratka. Research Associates: Bob Dickman, Grant Merrill. Data Analysts: Karin Eichhoff, Joanne Stewart, Anne Trainor. Data Entry: Charlie Blackburn, Patrick Burke, Rebecca Grote, Angela Hill, Mindy Paulek. Photographs: Tom Beck, Bruce Gill, Mary Lloyd, Rich Reading, Rick Thompson. Funding: CDOW, Great Outdoors Colorado (GOCO), Turner Foundation, U.S.D.A. Forest Service, Vail Associates. 14 �LITERATURE CITED AUBRY, K. B., G. M. KOEHLER, J. R. SQUIRES. 1999. Ecology of Canada lynx in southern boreal forests. Pages 373-396 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. BYRNE, G. 1998. Core area release site selection and considerations for a Canada lynx reintroduction in Colorado. Report for the Colorado Division of Wildlife. CURTIS, J. T. 1959. The vegetation of Wisconsin. University of Wisconsin Pres, Madison. GANEY, J. L. AND W. M. BLOCK. 1994. A comparison of two techniques for measuring canopy closure. Western Journal of Applied Forestry 9:1: 21-23. HODGES, K. E. 1999. Ecology of snowshoe hares in southern boreal and montane forests. Pages 163206 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. KOEHLER, G. M. 1990. Population and habitat characteristics of lynx and snowshoe hares in north central Washington. Canadian Journal of Zoology 68:845-851. KOLBE, J. A,, J. R. SQUIRES, T. W. PARKER. 2003. An effective box trap for capturing lynx. Journal of Wildlife Management 31:980-985. LAYMON, S. A. 1988. The ecology of the spotted owl in the central Sierra Nevada, California. PhD Dissertation University of California, Berkeley, California. MAJOR, A. R. 1989. Lynx, Lynx canadensis canadensis (Kerr) predation patterns and habitat use in the Yukon Territory, Canada. M. S. Thesis, State University of New York, Syracuse. MOWAT, G., K. G. POOLE, AND M. O’DONOGHUE. 1999. Ecology of lynx in northern Canada and Alaska. Pages 265-306 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. POOLE, K. G., G. MOWAT, AND B. G. SLOUGH. 1993. Chemical immobilization of lynx. Wildlife Society Bulletin 21:136-140. SHENK, T. M. 1999. Program narrative: Post-release monitoring of reintroduced lynx (Lynx canadensis) to Colorado. Report for the Colorado Division of Wildlife. __________. 2001. Post-release monitoring of lynx reintroduced to Colorado. Job Progress Report for the Colorado Division of Wildlife. Fort Collins, Colorado. SQUIRES, J. R. AND T. LAURION. 1999. Lynx home range and movements in Montana and Wyoming: preliminary results. Pages 337-349 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University Press of Colorado, Boulder, Colorado. U. S. FISH AND WILDLIFE SERVICE. 2000. Endangered and threatened wildlife and plants: final rule to list the contiguous United States distinct population segment of the Canada lynx as a threatened species. Federal Register 65, Number 58. WILD, M. A. 1999. Lynx veterinary services and diagnostics. Job Progress Report for the Colorado Division of Wildlife. Fort Collins, Colorado. Prepared by _______________________________ Tanya M. Shenk, Wildlife Researcher 15 �Table 1. Definitions of forest structure classes used to describe habitat sites (Thomas 1979). Forest Structure Class Definition Grass/forb The grass/forb stage is created naturally by a catastrophic event, such as wildfire, and is typified by the near complete absence of snags, litter or down material in the aspen and ponderosa pine types, or vice versa in the lodgepole or subalpine forest types. Shrub/seedling The shrub/seedling stage occurs when tree seedlings or shrubs grow up to 2.5 cm at diameter breast height (DBH), either naturally or artificially through planting. Sapling/pole The sapling/pole stage is a young stage where tree DBH's range from 2.5-17.5 cm with tree heights ranging 1.8-13.5 m. These trees are 5-100 years of age, depending on species and site condition. Mature The mature stage occurs when tree diameters reach a relatively large size (25-50 cm) and the trees are usually 90 or more years old. Forest stands begin to experience accelerated mortality from disease and insects. Old-growth The old-growth stage occurs when a mature stand is at advanced age (100 years for aspen or 200 years for spruce), is very slow growing, and has advanced degrees of disease, insects, snags, and down, dead material. An exception to this occurs in ponderosa pine and aspen types where these old-growth stands typically experience low densities of down dead material or snags. Table 2. Lynx released in Colorado from February 1999 through June 30, 2005. Year Females Males TOTAL 1999 22 19 41 2000 35 20 55 2003 17 16 33 2004 17 20 37 2005 18 20 38 TOTAL 109 95 204 16 �Table 3. Causes of death for adult lynx released into southwestern Colorado in 1999-2005 as of June30, 2005. Number of Mortalities Cause of Death Unknown 22 9 Starvation Hit by Vehicle 9 Shot 8 Probable Shot 6 Plague 4 Probable Predation 2 2 Probable Hit by Vehicle Other Human Caused 2 Illness 1 Territorial Dispute 1 Total Mortalities 66 Table 4. Status of adult lynx reintroduced to Colorado as of June 30, 2005. Females Males Unknown Released 109 95 Known Dead 40 25 1 Possible Alive 69 70 Missing 16 13 Tracking 53 57 a 1 is unknown mortality TOTALS 204 66 138 28a 110 Table 5. Lynx reproduction documented in 2003. Female BC00F8 BC00F19 YK00F16 YK99F1 YK00F19 YK00F10 Release Year 2000 2000 2000 1999 2000 2000 Date Den Found 5/21/03 5/26/03 6/19/03 6/10/03 6/11/03 5/31/03 TOTAL Females ? 1 1 2 1 2 7 17 Number of Kittens Males ? 2 1 2 1 2 1 3 2 3 2 4 7 16 Total �Table 6. Lynx reproduction documented in 2004. Female ID YK00F2 AK00F2 YK00F1 YK00F15 BC00F14 BC00F18 YK00F10 BC03F02 BC03F10 BC03F09 YK00F7 YK99F1 Unknown Unknown TOTAL Release Year 2000 2000 2000 2000 2000 2000 2000 2003 2003 2003 2000 1999 Previous Litter Date Den Found 5/28/2004 5/31/2004 6/1/2004 6/4/2004 6/7/2004 6/10/2004 6/17/2004 6/25/2004 6/26/2004 6/29/2004 6/30/2004 Date Kittens Found Dec 2004 Sept 2004 Feb 2005 Number of Kittens Females Males Total 3 1 4 2 1 3 3 3 1 2 3 1 2 3 4 4 1 1 2 2 2 2 2 1 1 2 1 1 2 2 4 3 19 11 39 Table 7. Lynx reproduction documented in 2005. Female ID AK00F02 YK00F15 YK00F10 YK00F11 YK00F01 YK00F07 BC00F18 BC03F02 BC03F01 QU03F06 QU03F04 QU03F07 BC03F09 BC03F10 BC04F01 BC04F03 BC04F05 TOTAL Release year 2000 2000 2000 2000 2000 2000 2000 2003 2003 2003 2003 2003 2003 2003 2004 2004 2004 Previous Litters 2004 2004 2003, 2004 2004 2004 2004 2004 2004 2004 Date Den Found 5/21/2005 5/28/2005 6/1/2005 6/9/2005 6/10/2005 6/14/2005 6/24/2005 5/25/2005 5/27/2005 6/5/2005 6/14/2005 6/16/2005 6/27/2005 6/11/2005 6/19/2005 6/23/2005 Total 3 2 4 2 3 3 2 2 4 3 3 4 2 ? 3 3 3 46 Number of Kittens Males Females 2 1 1 1 2 2 2 2 1 1 2 1 1 1 1 2 2 3 1 2 3 1 1 1 2 3 25 1 3 21 Table 8. Number of kills found each winter field season through snow-tracking of lynx and percent composition of kills of the three primary prey species. Field Season 1999 1999-2000 2000-2001 2001-2002 2002-2003 2003-2004 2004-2005 n 9 83 89 54 65 37 78 Snowshoe Hare 55.56 67.47 67.42 90.74 90.77 67.57 83.33 Prey (%) Red Squirrel Cottontail 22.22 0 19.28 1.20 19.10 8.99 5.56 0 6.15 0 27.03 2.70 10.26 0 18 Other 22.22 12.05 4.49 3.70 3.08 2.70 6.41 �·•-•--• •-• • • e •• •--•--•. 6 1 r 1 l l i 1 l 2• • • 1 l EB 3• 1 l l 4. l l se--•--•--•--• 12meters 15meters Figure 1. Design of site-scale habitat plot sampling plot. Each of the 25 points are 3 meters apart (the first 6 points are labeled 1-6). The object that triggered a habitat plot (e.g., kill ) is the center point, depicted by the hollow circle. The actual pints encompass a 12 m x 12 m square but the understory and overstory data collected are influenced by vegetation occurring within a 15 m x 15 m square. DEN SITES LONG BED SITES KILL SITES n = 33 n = 458 n = 216 x Elevation = 3347 m x Elevation = 3179 m x SE = 33 m Slope = 29° SE = 2° x SE= 9 m Slope = 14° SE = 0.5° TRAVEL SITES n = 441 x Elevatio n= 3145 m x Elevation = 3194 m SE = 14 m SE = 9 m x Slope = 16° x Slope = 17° SE = 0.8° SE = 0.6° Figure 2. Frequency of aspect, mean elevation and SE and mean slope and SE for 4 lynx use sites; dens, long beds, kills and travel. 19 �80 70 60 ...., 50 IC: 8 40 V P.. 30 20 10 0 ES SF AS \XII Total Cover Tree Species I ■ Long Beds ■ Kills □ Travel □ Den Sitesl Figure 3. Mean percent overstory by tree species Engelmann spruce (ES), subalpine fir (SF), aspen (AS), willow (WI) and total cover for 4 different lynx use sites: long beds, kill sites, travel and den sites. Confidence intervals (95%) are depicted by error bars. 10-r----------------~ ~ ES SF C\JI/D AS WI Total 60 Co~ lO ~ ~-r-----------------, V nu1I i ~1~B·~ p.... 1".S ES SF Den Sites 40 ITT nl Cll'ID lo ll'/1 WI 20 10 T,tl.C ..., 'I' otu "' " Understory Species CWD 30 Most Common Understory Species ES= Engelmann spruce SF = Subalpine fir \l7I= Willow AS= Aspen LO = Lodgepole pine C\YD = Coarse woody debris C ov~r Figure 4. Mean percent understory by tree species Engelmann spruce (ES), subalpine fir (SF), coarse woody debris (CWD), aspen (AS), willow (WI) and total cover for 4 different lynx use sites: long beds, kill sites, travel and den sites. 20 �1000 ~ - - ~ - - - ~ 800 Kills 600 ES SF SF ES AS AS Tree Species 1 ■ o-6 □ 6-12 12-18 □ 18-24 □ >24 I Figure 5. Mean tree density by species Engelmann spruce (ES), subalpine fir (SF) and aspen (AS) and dbh size class for 4 different lynx use sites. 80 70 ...u 60 > 0 50 _µ 40 lJ t:: u ... 30 u u p_, 20 10 0 SF AS WI Total Cover Tree Species I■ SuocessfuJ. Chases ~ Unsuocessful Chases I Figure 6. Mean percent overstory by tree species Engelmann spruce (ES), subalpine fir (SF), aspen (AS), willow (WI) and total cover for successful and unsuccessful snowshoe hare chases. Confidence intervals (95%) are depicted by error bars. 21 �50 45 H 40 G) i> 35 0 u.w 30 25 ~ G) u 20 H G) 15 p., 10 5 0 ES SF C\XJD T o tal Cover \XII T ree Species ■ 0 - 0. 5 m S IE 0 - 0.5 m U ■ 0.5 - 1.0 m S IE 0.5 - 1. 0 m U □ 1.0 - 1.5 m S ll1! 10 - 1.5 m U Figure 7. Mean percent understory by tree species Engelmann spruce (ES), subalpine fir (SF), aspen (AS), willow (WI) and total cover for 3 different understory height categories for successful and unsuccessful snowshoe hare chases. Confidence intervals (95%) are depicted by error bars. 1200 I 0-6 SC ir:il 0-6 UC 1000 .... V ~ u - ■ 6-12 SC 800 ir:il 6-12 UC V ::r:: □ 12-18 SC 600 ~ 12-18 UC r.n V V .... 400 ■ 18-24 SC E--< ir:il 18-24 UC 200 I >24 SC ir:il >24 UC 0 ES SF AS Tree Species Figure 8. Mean tree density by species Engelmann spruce (ES), subalpine fir (SF) and aspen (AS) and 5 DBH size classes for successful chases (SC) and unsuccessful chases (UC) of snowshoe hare. 22 �Colorado Division of Wildlife July 2005 - June 2006 WILDLIFE RESEARCH REPORT State of Cost Center Work Package Task No. Colorado 3430 0670 1 Federal Aid Project: N/A : Division of Wildlife : Mammals Research : Lynx Conservation : Post-Release Monitoring of Lynx Reintroduced to Colorado : Period Covered: July 1, 2005 - June 30, 2006 Author: T. M. Shenk Personnel: L. Baeten, B. Diamond, R. Dickman, D. Freddy, L. Gepfert, J. Ivan, R. Kahn, A. Keith, G. Merrill, T. Spraker, S. Wait, S. Waters, L. Wolfe, D. Younkin All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT In an effort to establish a viable population of lynx (Lynx canadensis) in Colorado, the Colorado Division of Wildlife (CDOW) initiated a reintroduction effort in 1997 with the first lynx released in February 1999. From 1999-2005, 204 lynx were released in Colorado. Fourteen additional animals (8 males: 6 females) were released in spring 2006 resulting in a total of 218 lynx reintroduced to southwestern Colorado. We documented survival, movement patterns, reproduction, and habitat-use through aerial (n = 8680) and satellite (n = 18, 963) tracking. Most lynx remained near the core release area in southwestern Colorado. From 1999-2006, there were 80 mortalities of released adult lynx. Approximately 31.3% were human-induced which were attributed to collisions with vehicles or gunshot. Malnutrition and disease/illness accounted for 21.3% of the deaths while 32.5% of the deaths were from unknown causes. Reproductive females had the smallest 90% utilization distribution home ranges ( x = 75.2 km2, SE = 15.9 km2 ), followed by attending males ( x = 102.5 km2, SE = 39.7 km2) and nonreproductive animals ( x = 653.8 km2, SE = 145.4 km2). Reproduction was first documented in 2003 with subsequent successful reproduction in 2004 and 2005. Four dens with 11 kittens were found in 2006. Lynx CO04F07, a female lynx born in Colorado in 2004 was the mother of one of these litters which documented the first recruitment of Colorado-born lynx into the Colorado breeding population. From snow-tracking, the primary winter prey species (n = 426) were snowshoe hare (Lepus americanus, annual x = 75.1%, SE = 5.17) and red squirrel (Tamiasciurus hudsonicus, annual x = 15.3%, SE = 3.09); other mammals and birds formed a minor part of the winter diet. Mature Engelmann spruce (Picea engelmannii)-subalpine fir (Abies lasiocarpa) forest stands with 42-65% canopy cover and 15-20% conifer understory cover were the most commonly used areas in southwestern Colorado. Little difference in aspect (slight preference for north-facing slopes), slope ( x = 15.7°) or elevation ( x = 3173 m) were detected for long beds, travel and kill sites (n = 1841). Den sites (n = 37) however, were located at higher 1 �elevations ( x = 3354 m, SE = 31 m) on steeper ( x = 30°, SE = 2°) and more commonly north-facing slopes with a dense understory of coarse woody debris. A study to evaluate snowshoe hare densities, demography and seasonal movement patterns among small and medium tree-sized lodgepole pine stands and mature spruce/fir stands was initiated in 2005 and will continue through 2009. Results to date have demonstrated that CDOW has developed release protocols that ensure high initial post-release survival followed by high long-term survival, site fidelity, reproduction and recruitment of Colorado-born lynx into the Colorado breeding population. What is yet to be demonstrated is whether Colorado can support sufficient recruitment to offset annual mortality for a viable lynx population over time. Monitoring continues in an effort to document such viability. 2 �WILDLIFE RESEARCH REPORT POST RELEASE MONITORING OF LYNX (LYNX CANADENSIS) REINTRODUCED TO COLORADO TANYA M. SHENK P. N. OBJECTIVE The initial post-release monitoring of lynx reintroduced into Colorado will emphasize 5 primary objectives: 1. Assess and modify release protocols to ensure the highest probability of survival for each lynx released. 2. Obtain regular locations of released lynx to describe general movement patterns and habitats used by lynx. 3. Determine causes of mortality in reintroduced lynx. 4. Estimate survival of lynx reintroduced to Colorado. 5. Estimate reproduction of lynx reintroduced to Colorado. Three additional objectives will be emphasized after lynx display site fidelity to an area: 6. Refine descriptions of habitats used by reintroduced lynx. 7. Refine descriptions of daily and overall movement patterns of reintroduced lynx. 8. Describe hunting habits and prey of reintroduced lynx. Information gained to achieve these objectives will form a basis for the development of lynx conservation strategies in the southern Rocky Mountains. SEGMENT OBJECTIVES 1. Release additional adult lynx captured in Canada in southwestern Colorado during spring 2006. 2. Complete winter 2005-06 field data collection on lynx habitat use, hunting behavior, diet, mortalities, and movement patterns. 3. Complete winter 2005-06 lynx trapping field season to collar Colorado born lynx and re-collar adult lynx. 4. Complete spring 2006 field data on lynx reproduction. 5. Summarize and analyze data and publish information as Progress Reports, peer-reviewed manuscripts for appropriate scientific journals, or CDOW technical publications. 6. Complete a study plan to evaluate snowshoe hare densities, demography and seasonal movement patterns among small and medium tree-sized lodgepole pine stands and mature spruce/fir stands. INTRODUCTION The Canada lynx occurs throughout the boreal forests of northern North America. Colorado represents the southern-most historical distribution of lynx, where the species occupied the higher elevation, montane forests in the state. Little was known about the population dynamics or habitat use of this species in their southern distribution. Lynx were extirpated or reduced to a few animals in the state by the late 1970’s due, most likely, to predator control efforts such as poisoning and trapping. Given the isolation of Colorado to the nearest northern populations, the CDOW considered reintroduction as the only option to attempt to reestablish the species in the state. 3 �A reintroduction effort was begun in 1997, with the first lynx released in Colorado in 1999. To date, 218 wild-caught lynx from Alaska and Canada have been released in southwestern Colorado. The goal of the Colorado lynx reintroduction program is to establish a self-sustaining, viable population of lynx in this state. Evaluation of incremental achievements necessary for establishing viable populations is an interim method of assessing if the reintroduction effort is progressing towards success. There are 7 critical criteria for achieving a viable population: 1) development of release protocols that lead to a high initial post-release survival of reintroduced animals, 2) long-term survival of lynx in Colorado, 3) development of site fidelity by the lynx to areas supporting good habitat in densities sufficient to breed, 4) reintroduced lynx must breed, 5) breeding must lead to reproduction of surviving kittens 6) lynx born in Colorado must reach breeding age and reproduce successfully, and 7) recruitment must equal or be greater than mortality over an extended period of time. The post-release monitoring program for the reintroduced lynx has 2 primary goals. The first goal is to determine how many lynx remain in Colorado and their locations relative to each other. Given this information and knowing the sex of each individual, we can assess whether these lynx can form a breeding core from which a viable population might be established. From these data we can also describe general movement patterns and habitat use. The second primary goal of the monitoring program is to estimate survival of the reintroduced lynx and, where possible, determine causes of mortality for reintroduced lynx. Such information will help in assessing and modifying release protocols and management of lynx once they have been released to ensure their highest probability of survival. Additional goals of the post-release monitoring program for lynx reintroduced to the southern Rocky Mountains included refining descriptions of habitat use and movement patterns and describing successful hunting habitat once lynx established home ranges that encompassed their preferred habitat. Specific objectives for the site-scale habitat data collection include: 1) describe and quantify site-scale habitat use by lynx reintroduced to Colorado, 2) compare site-scale habitat use among types of sites (e.g., kills vs. long-duration beds), and 3) compare habitat features at successful and unsuccessful snowshoe hare chases. Documenting reproduction is critical to the success of the program and lynx are monitored intensively to document breeding, births, survival and recruitment of lynx born in Colorado. Site-scale habitat descriptions of den sites are also collected and compared to other sites used by lynx. The program will also investigate the ecology of snowshoe hare in Colorado. A study comparing snowshoe hare densities among mature stands of Engelmann spruce (Picea engelmannii)/subalpine fir (Abies lasiocarpa), lodgepole pine (Pinus contorta) and Ponderosa pine (Pinus ponderosa) was completed in 2004 with highest hare densities found in Engelmann spruce/subalpine fir stands and no hares found in Ponderosa pine stands. A study to evaluate the importance of young, regenerating lodgepole pine and mature Engelmann spruce/subalpine fir stands in Colorado by examining density and demography of snowshoe hares that reside in each was initiated in 2005 and will continue through 2009. Lynx is listed as threatened under the Endangered Species Act (ESA) of 1973, as amended (16 U. S. C. 1531 et. seq.)(U. S. Fish and Wildlife Service 2000). Colorado is included in the federal listing as lynx habitat. Thus, an additional objective of the post-release monitoring program is to develop conservation strategies relevant to lynx in Colorado. To develop these conservation strategies, information specific to the ecology of the lynx in its southern Rocky Mountain range, such as habitat use, movement patterns, mortality factors, survival, and reproduction in Colorado is needed. 4 �STUDY AREA Southwestern Colorado is characterized by wide plateaus, river valleys, and rugged mountains that reach elevations over 4200 m. Engelmann spruce-subalpine fir is the most widely distributed coniferous forest type at elevations most typically used by lynx. The Core Release Area is defined as areas bounded by the New Mexico state line to the south, Taylor Mesa to the west and Monarch Pass on the north and east and > 2900 m in elevation (Figure 1). The lynx-established core area is roughly bounded by areas used by lynx in the Taylor Park/ Collegiate Peak areas in central Colorado and includes areas of continuous use by lynx, including areas used during breeding and denning (Figure 1). METHODS REINTRODUCTION Effort All 2006 lynx releases were conducted under the protocols found to maximize survival (see Shenk 2001). Estimated age, sex and body condition were ascertained and recorded for each lynx prior to release (see Wild 1999). Specific release sites were those used in earlier years of the project and were selected based on land ownership and accessibility during times of release (Byrne 1998). Lynx were transported from the Frisco Creek Wildlife Rehabilitation Center, where they were held from their time of arrival in Colorado, to their release site in individual cages. Release site location was recorded in Universal Transverse Mercator (UTM) coordinates and identification of all lynx released at the same location, on the same day, was recorded. Behavior of the lynx on release and movement away from the release site were documented. Distribution and Movement Patterns All lynx released in 1999 were fitted with TelonicsTM radio-collars. All lynx released since 1999, with the exception of 5 males released in spring 2000, were fitted with SirtrackTM dual satellite/VHF radio-collars. These collars have a mortality indicator switch that operated on both the satellite and VHF mode. The satellite component of each collar was programmed to be active for 12 hours per week. The 12-hour active periods for individual collars were staggered throughout the week. Signals from the collars allowed for locations of the animals to be made via Argos, NASA, and NOAA satellites. The location information was processed by ServiceArgos and distributed to the CDOW through e-mail messages. To determine general movement patterns of reintroduced lynx, regular locations of released lynx were collected through a combination of aerial, satellite and ground radio-tracking. Locations were recorded in UTM coordinates and general habitat descriptions for each ground and aerial location were recorded. Home Range Annual home ranges were calculated as a 95% utilization distribution using a kernel home-range estimator for each lynx we had at least 30 locations for within a year. A year was defined as March 15 – March 14 of the following year. Locations used in the analyses were collected from September 1999 – January 2006 and all locations obtained for an individual during the first six months after its release were eliminated from any home range analyses as it was assumed movements of lynx initially post-release may not be representative of normal habitat use. Locations were obtained either through aerial VHF surveys or locations or the midpoint (ArcView Movement Extension) of all high quality (accuracy rating of 01km) satellite locations obtained within a single 24-hour period. All locations used within a single home range analysis were taken a minimum of 24 hours apart. 5 �Home range estimates were classified as being for a reproductive or non-reproductive animal. A reproductive female was defined as one that had kittens with her; a reproductive male was defined as a male whose movement patterns overlapped that of a reproductive female. If a litter was lost within the defined year a home range described for a reproductive animal were estimated using only locations obtained while the kittens were still with the female. Survival Survival was estimated as ragged telemetry data using the nest survival models in Program MARK (White and Burnham 1999). Mortality Factors When a mortality signal (75 beats per minute [bpm] vs. 50 bpm for the Telonics™ VHF transmitters, 20 bpm vs. 40 bpm for the Sirtrack™ VHF transmitters, 0 activity for Sirtrack™ PTT) was heard during either satellite, aerial or ground surveys, the location (UTM coordinates) was recorded. Ground crews then located and retrieved the carcass as soon as possible. The immediate area was searched for evidence of other predators and the carcass photographed in place before removal. Additionally, the mortality site was described and habitat associations and exact location were recorded. Any scat found near the dead lynx that appeared to be from the lynx was collected. All carcasses were transported to the Colorado State University Veterinary Teaching Hospital (CSUVTH) for a post mortem exam to 1) determine the cause of death and document with evidence, 2) collect samples for a variety of research projects, and 3) archive samples for future reference (research or forensic). The gross necropsy and histology were performed by, or under the lead and direct supervision of a board certified veterinary pathologist. At least one research personnel from the CDOW involved with the lynx program was also present. The protocol followed standard procedures used for thorough post-mortem examination and sample collection for histopathology and diagnostic testing (see Shenk 1999 for details). Some additional data/samples were routinely collected for research, forensics, and archiving. Other data/samples were collected based on the circumstances of the death (e.g., photographs, video, radiographs, bullet recovery, samples for toxicology or other diagnostic tests, etc.). From 1999–2004 the CDOW retained all samples and carcass remains with the exception of tissues in formalin for histopathology, brain for rabies exam, feces for parasitology, external parasites for ID, and other diagnostic samples. Since 2005 carcasses are disposed of at the CSUVTH with the exception of the lower canine, fecal samples, stomach content samples and tissue or bone marrow samples to be delivered by CDOW to the Center for Disease control for plague testing. The lower canine, from all carcasses, is sent to Matson Labs (Missoula, Montana) for aging and the fecal and stomach content samples are evaluated for diet. Reproduction Females were monitored for proximity to males during each breeding season. We defined a possible mating pair as any male and female documented within at least 1 km of each other in breeding season through either flight data or snow-tracking data. Females were then monitored for site fidelity to a given area during each denning period of May and June. Each female that exhibited stationary movement patterns in May or June were closely monitored to locate possible dens. Dens were found when field crews walked in on females that exhibited virtually no movement for at least 10 days from both aerial and ground telemetry. Kittens found at den sites were weighed, sexed and photographed. Each kitten was uniquely marked by inserting a sterile passive integrated transponder (PIT, Biomark, Inc., Boise, Idaho, USA) tag subcutaneously between the shoulder blades. Time spent at the den was minimized to ensure the least amount of disturbance to the female and the kittens. Weight, PIT-tag number, sex and any distinguishing 6 �characteristics of each kitten was also recorded. Beginning in 2005, blood and saliva samples were collected and archived for genetic identification. During the den site visits, den site location was recorded as UTM coordinates. General vegetation characteristics, elevation, weather, field personnel, time at the den, and behavioral responses of the kittens and female were also recorded. Once the females moved the kittens from the natal den area, den sites were visited again and site-specific habitat data were collected (see Habitat Use section below). Captures Captures were attempted for either lynx that were in poor body condition or lynx that needed to have their radio-collars replaced due to failed or failing batteries or to radio-collar kittens born in Colorado once they reached at least 10-months of age when they were nearly adult size. Methods of recapture included 1) trapping using a Tomahawk™ live trap baited with a rabbit and visual and scent lures, 2) calling in and darting lynx using a Dan-Inject CO2 rifle, 3) custom box-traps modified from those designed by other lynx researchers (Kolbe et al. 2003) and 4) hounds trained to pursue felids were also used to tree lynx and then the lynx was darted while treed. Lynx were immobilized either with Telazol (3 mg/kg; modified from Poole et al. 1993 as recommended by M. Wild, DVM) or medetomidine (0.09mg/kg) and ketamine (3 mg/kg; as recommended by L. Wolfe, DVM)) administered intramuscularly (IM) with either an extendible pole-syringe or a pressurized syringe-dart fired from a Dan-Inject air rifle. Immobilized lynx were monitored continuously for decreased respiration or hypothermia. If a lynx exhibited decreased respiration 2mg/kg of Dopram was administered under the tongue; if respiration was severely decreased, the animal was ventilated with a resuscitation bag. If medetomidine/ketamine were the immobilization drugs, the antagonist Atipamezole hydrochloride (Antisedan) was administered. Hypothermic (body temperature < 95o F) animals were warmed with hand warmers and blankets. While immobilized, lynx were fitted with replacement SirtrackTM VHF/satellite collar and blood and hair samples were collected. Once an animal was processed, recovery was expedited by injecting the equivalent amount of the antagonist Antisedan IM as the amount of medetomidine given, if medetomodine/ketemine was used for immobilization. Lynx were then monitored while confined in the box-trap until they were sufficiently recovered to move safely on their own. No antagonist is available for Telezol so lynx anesthetized with this drug were monitored until the animal recovered on its own in the box-trap and then released. If captured and in poor body condition, lynx were anesthetized with either Telezol (2 mg/kg) or medetomodine/ketemine and returned to the Frisco Creek Wildlife Rehabilitation Center for treatment. HABITAT USE Gross habitat use was documented by recording canopy vegetation at aerial locations. More refined descriptions of habitat use by reintroduced lynx were obtained through following lynx tracks in the snow (i.e., snow-tracking) and site-scale habitat data collection conducted at sites found through this method to be used by lynx. Snow-tracking Locations from aerial- and satellite-tracking were used to help ground-trackers locate lynx tracks in snow. Snowmobiles, where permitted, were used to gain the closest possible access to the lynx tracks without disturbing the animal. From that point, the tracking team used snowshoes to access tracks. Once tracks were found, the ground crew back- or forward-tracked the animal if it was far enough away not to be disturbed. Back-tracking generally avoided the possibility of disturbing the lynx by moving away from the animal rather than towards the animal. However, monitoring of the lynx through radio-telemetry was used to assure that the ground crew was staying a sufficient distance away from the lynx in the event the lynx might double back on its tracks. Radio-telemetry was also used in forward-tracking to make sure 7 �the team did not disturb the animal. If it appeared the lynx began to move in response to the observers, the observers stopped following the tracks. If the lynx began to move and the movement did not appear to be a response to the observers, the ground crew continued following the track. An attempt was made in Season 1 (February-May 1999) and Season 2 (December 1999-April 2000) to snow-track each lynx. In Season 3 (December 2000-April 2001), we attempted to snow-track all lynx within the Core Release Area. In tracking Season 4 (December 2001-April 2002), Season 5 (December 2002-April 2003), Season 6 (December 2003-April 2004), Season 7 (December 2004-April 2005) and Season 8 (December 2005-March 2006) we attempted to track all accessible lynx in the Core Release Area and some lynx north of the Core Release Area. Ground crews were instructed to track lynx only where it was safe to travel. Restrictions to safe travel included avalanche danger and extremely rugged terrain. Ground crews worked in pairs and were fully equipped for winter back-country survival. Data Collection For each day of tracking the date, lynx being tracked, slope, aspect, UTM coordinates, elevation, general habitat description, and summary of the days tracking were recorded. Aspect was defined as the direction of 'downhill' or 'fall line' on a slope. This is the direction along the ground in a dihedral angle between the horizontal and the plane of the ground surface. Units were compass degrees. Slope was defined as the dihedral angle between the horizontal and the plane of the ground surface (e.g., 45°). Once a track was located there were 2 types of 'sites' that were encountered. Site I areas needed documentation but either did not reflect areas lynx selected for specific habitat features, or were sites that occurred too frequently to measure each in detail. These sites included the start and end of the track being followed, the location of scat, and short-duration beds defined as being small in size (approximating an area a lynx would crouch), and with little ice formed in the bed indicating little time spent there. Site II areas included areas that might reflect specific habitat features lynx selected for and included locations where the following were found: kills, start of chases, territory marks (e.g., spray sites, buried scat, scat placed on prominent locations), long-duration beds (encompasses an area where a lynx would have lain for an extended period, iced bottom), and road crossing (both sides of road). In addition, habitat plots were conducted along lynx travel routes if no other sites were sampled in the last hour. At each of the 2 types of sites the date, lynx tracked, slope, aspect, forest structure class, UTM coordinates, and elevation were recorded. Forest structure classes included grass/forb, shrub/seedling, sapling/pole, mature, and old growth as defined in Table 1. For Site I areas, the only additional data that was collected was identification of what the site was used for (e.g., short-duration bed), and a brief description of the site. Habitat plots (see below) were conducted at Site II areas. Description of the Habitat Plot The habitat plot consisted of a 12 m x 12 m square defined by a series of 25 points placed in 5 rows of 5 with the center point being on the object that defined the site (e.g., a kill)(Figure 2). Each point was 3 m apart. The 12 m x 12 m sampling square exceeded the minimum requirement of 0.01 ha. recommended by Curtis (1959) for sampling trees. Measurements taken at each of the 25 points included: 1. Snow depth - measured vertically by an avalanche probe marked in cm. 2. Understory - measured from top of snow to 150 cm above snow in a column of 3-cm radius around the avalanche probe. Because understory measurements were influenced by vegetation outside the perimeter of the 25 sampling points (12 m x 12 m) the area used for estimating undersory cover was 15 m by 15 m. At each point, crews recorded all shrubs, trees and coarse woody debris (CWD) that fell within this column and was visible above the snow. Crews also recorded number of branches of each species that fell within the column at 3 different height categories (0-0.5 m, 0.51-1.0 m, 1.01-1.5 m). 8 �3. 4. 5. Overstory: measured at 150 cm above snow with a sighting tube. The tube was made of PVC pipe, with a curved viewing end and a crosshair made of wire on the opposite end. The sighting tube was attached to the avalanche probe used to measure snow depth. Species that hit the crosshair were recorded at each of the 25 points in the vegetation plot. Ganey and Block (1994) found this method of measuring canopy cover (with 20 sample points per plot; Laymon 1988) provided greater precision among observers. Species composition: all the different species of tree or shrub that hit the crosshair of the sighting tube at each of the 25 points were recorded. Tree composition of the vegetation plot was recorded by species and diameter at breast height (DBH). Snow depth was used in conjunction with this recorded DBH to estimate true DBH. Within the 12 m x 12 m square all conifers and deciduous trees were recorded by DBH size class (A = 0-6 in, B = 6.1-12 in, C = 12.1 -18 in, D = 18.1-24 in, E = > 24 in). Area for the tree composition analysis was 12 m x 12 m. Understory was estimated as: 1) percent occurrence within the vegetation plot (number of points with understory/total number of points surveyed) and 2) mean percent occurrence and variance by species and height category over the total points sampled within the vegetation plot. Overstory was estimated as percent occurrence over the vegetation plot (number of points with overstory/total number of points surveyed). DIET AND HUNTING BEHAVIOR Winter diet of reintroduced lynx was estimated by documenting successful kills through snowtracking. Prey species from failed and successful hunting attempts were identified by either tracks or remains. Scat analysis also provided information on foods consumed. Scat samples were collected wherever found and labeled with location and individual lynx identification. Only part of the scat was collected (approximately 75%); the remainder was left in place in the event that the scat was being used by the animal as a territory mark. Site-scale habitat data collected for successful and unsuccessful snowshoe hare kills were compared. SNOWSHOE HARE ECOLOGY A study plan was designed to evaluate the importance of young, regenerating lodgepole pine (Pinus contorta) and mature Engelmann spruce / subalpine fir stands in Colorado by examining density and demography of snowshoe hares that reside in each. Specifically, the study was designed to evaluate small and medium lodgepole pine stands and large spruce/fir stands where the classes “small”, “medium”, and “large” refer to the diameter at breast height (dbh) of overstory trees as defined in the United States Forest Service R2VEG Database (small = 2.54−12.69 cm dbh, medium = 12.70−22.85 cm, and large = 22.86−40.64 cm dbh; J. Varner, United States Forest Service, personal communication). The study design was also developed to identify which of the numerous density-estimation procedures available perform accurately and consistently using an innovative, telemetry augmentation approach as a baseline. Movement patterns and seasonal use of deciduous cover types such as riparian willow will be assessed. Finally, the study was designed to further expound on the relationship between density, demography, and stand type by examining how snowshoe hare density and demographic rates vary with specific vegetation, physical, and landscape characteristics of a stand. 9 �RESULTS REINTRODUCTION Effort From 1999 through 2005 204 lynx were reintroduced into southwestern Colorado. An additional 14 lynx were released in April 2006 (6 females: 8 males), bringing the total number of lynx released in Colorado to 218 (Table 2). Lynx released in 2006 were captured in British Columbia and Yukon. These 14 lynx were released in the Core Release Area of southwestern Colorado at or near previously used release sites in southwestern Colorado. Lynx were released with dual VHF/satellite radio collars so they could be monitored for movement, reproduction and survival. The CDOW does not plan to release any additional lynx in 2007. Distribution and Movement Patterns A total of 8680 aerial VHF locations for all 218 reintroduced lynx have been collected to date (June 30, 2006). An additional 18,963 satellite locations have been collected. Most lynx released in 2006 remained in southwestern Colorado. The majority of surviving lynx from the entire reintroduction effort continue to use high elevation (> 2900 m), forested areas from New Mexico north to Gunnison, west as far as Taylor Mesa and east to Monarch Pass. Most movements away from the Core Release Area were to the north. Numerous travel corridors have been used repeatedly by more than one lynx. These travel corridors include the Cochetopa Hills area for northerly movements, the Rio Grande Reservoir-SilvertonLizardhead Pass for movements to the west, and southerly movements down the east side of Wolf Creek Pass to the southeast through the Conejos River Valley. Lynx appear to remain faithful to an area during winter months, and exhibit more extensive movements away from these areas in the summer. Such movement patterns have also been documented by native lynx in Wyoming and Montana (Squires and Laurion 1999). Home Range Reproductive females had the smallest 90% utilization distribution annual home ranges ( x = 75.2 km2, SE = 15.9 km2, n = 19), followed by attending males ( x = 102.5 km2, SE = 39.7 km2, n = 4). Nonreproductive females had the largest annual home ranges ( x = 703.9 km2, SE = 29.8 km2, n = 32) followed by non-reproductive males ( x = 387.0 km2, SE = 73.5 km2, n = 6). Combining all nonreproductive animals yielded a mean annual home range of 653.8 km2 (SE = 145.4 km2, n = 38). Survival Initial survival rate estimates for reintroduced lynx were completed, however, further analyses need to be conducted before estimates will be presented. As of June 30, 2006, CDOW was actively tracking 95 of the 138 lynx still possibly alive. There are 43 lynx that we have not heard signals on since at least June 30, 2005 and these animals are classified as ‘missing’ (Table 3). One of these missing lynx is a mortality of unknown identity, thus only 42 are truly missing. Possible reasons for not locating these missing lynx include 1) long distance dispersal, beyond the areas currently being searched, 2) radio failure, or 3) destruction of the radio (e.g., run over by car). CDOW continues to search for all missing lynx during both aerial and ground searches. Two of the missing lynx released in 2000 are thought to have slipped their collars. Mortality Factors Of the total 218 adult lynx released from 1999-2006 there are 80 known mortalities as of June 30, 2006. Causes of death are listed in Table 4. Starvation was a significant cause of mortality in the first 10 �year of releases only. Mortalities occurred throughout the areas through which lynx moved. Approximately 31.3% were human-induced which were attributed to collisions with vehicles or gunshot. Malnutrition and disease/illness accounted for 21.3% of the deaths while 32.5% of the deaths were from unknown causes (Table 4). Reproduction 2003.-- Nine pairs of lynx were documented during the 2003 breeding season (March and April) from the 17 females we were monitoring. In May and June, 6 dens and a total of 16 kittens were found in the lynx Core Release Area in southwestern Colorado (Table 5, Figure 1). At all dens the females appeared in excellent condition, as did the kittens. The kittens weighed from 270-500 grams. Lynx kittens weigh approximately 200 grams at birth and do not open their eyes until they are 10-17 days old. The dens were scattered throughout the Core Release Area, with no dens found outside the core area. All the dens were in Engelmann spruce/subalpine fir forests in areas of extensive downfall. Elevations ranged from 3240-3557 m. Field crews weighed, photographed, PIT-tagged the kittens and took hair samples from the kittens for genetic work in an attempt to confirm paternity. Kittens were processed as quickly as possible (11-32 minutes) to minimize the time the kittens were without their mother. While working with the kittens the females remained nearby, often making themselves visible to the field crews. The females generally continued a low growling vocalization the entire time personnel were at the den. In all cases, the female returned to the den site once field crews left the area. Four of the 6 females that we know had kittens in summer 2003 were still with kittens at the end of April 2004. Two of those females still had 2 kittens with them at that time. Visual observations in February 2004 of one female with 2 kittens indicated all 3 were in good body condition. The mortality of female YK00F16 and her 1 kitten in October 2003 from plague was not due to poor habitat or prey conditions, and thus we might assume she would have raised the 1 kitten to this stage as well. Three probable kitten deaths from female YK00F19 were from 1 litter that most likely failed very early. Through snow-tracking in winter 2003-04 an unknown female (no radio frequency heard in the area of the tracks) we also documented 1-2 additional kittens born spring 2003 and still alive in winter 2004. Of the 16 kittens we found in summer 2003, we documented the following by April 2004: 6 confirmed alive, 7 confirmed dead, and 3 some evidence dead. Although we tried, we were not able to capture any of the 6 surviving kittens to fit them with radio-collars for subsequent monitoring. 2004.-- In Spring 2004, 26 females from the releases in 1999, 2000 and 2003 had active radiocollars. Of these, we documented 18 possible mating pairs of lynx during breeding season. All 4 of the females that had kittens with them through winter 2003-04 bred again spring 2004; 2 with the same male they successfully bred with spring 2003. During May-June 2004 we found 11 dens and a total of 30 kittens (Table 6). At all dens the females appeared in excellent condition, as did the kittens. The kittens weighed from 250-770 grams. Three of the 11 females with kittens were from the 2003 releases (Table 6). Three additional litters were documented after denning season through either observation of a female lynx with kittens or snow-tracking females with kittens that were not one of the 11 females found on dens. From the size of the kittens they would have been born during the normal denning season in May or June. Nine additional kittens were observed from these litters for a total of 39 known kittens born in 2004. Two of these additional litters were documented from direct follow-ups to sighting made by the public and reported to CDOW. Two females that had kittens in 2003 and reared at least part of their litters through March 2004, bred and had kittens again in 2004. Two of the litters documented by direct observation or snow-tracking are from females whose collars were no longer functioning. Seven kittens born in 2004 were captured at approximately 10-months of age and fitted with dual satellite/VHF collars. Six of the 7 were still alive 11 �and being monitored as of June 30, 2006. The cut collar of one kitten CO04M15 was left at the Silverton Post Office on October 25, 2005. We assume this lynx is dead. 2005.-- In spring 2005 we had 40 females from the releases in 1999, 2000, 2003 and 2004 that had active radio-collars. We documented 23 possible mating pairs of lynx during breeding season. During May-June 2005 we visited 16 dens and found a total of 46 kittens (Table 7). At all dens the females appeared in excellent condition, as did the kittens. An additional female (BC03F10) had a den we were not able to get to during May or June due to high water during spring run-off. Female BC03F03 was hit and killed on I-70 on 5/19/2005. She had 2 fetuses in her uterus, so would have contributed to reproduction this year had she lived. We weighed, photographed, PIT-tagged the kittens and recorded sex. We also took blood samples from the kittens for genetic work in an attempt to confirm paternity. While we were working with the kittens the females remained nearby, often remaining visible to us. The females generally continued a low growling vocalization the entire time we were at the den. In all cases, the female returned to the den site once we left the area. All of the 2005 dens were scattered throughout the high elevation areas of Colorado, south of I70. Most of the dens were in Engelmann spruce/subalpine fir forests in areas of extensive downfall. Elevations ranged from 3117-3586 m. We weighed, photographed, and PIT-tagged the kittens, recorded sex and took hair samples from the kittens for genetic work in an attempt to confirm paternity. Four of the females would not leave the den until we reached out to pick up a kitten. While we were working with the kittens the females remained nearby, often remaining visible to us. The females generally continued a low growling vocalization the entire time we were at the den. In all cases, the female returned to the den site once we left the area. One female, YK00F10 has had litters 3 years in a row. In 2003 she had 4 kittens and raised 2 through the winter. In 2004 she had 2 kittens and raised both through the winter, in 2005 she had 4 kittens again. She has had all 3 litters in the same general area and has had the same mate for 3 years. Eight additional females had their second litter in Colorado in 2005. Three females from the 2004 releases had litters in 2005. Year 2005 was the second consecutive year that we had females released the prior spring, find a territory and a mate within a year and produced live young. In reproduction season 2004 we had 3 females released in spring 2003 that also produced live young the next year. Of those 3, 2 successfully raised at least part of their litters through winter 2005. Seven kittens born in 2005 were captured at approximately 10-months of age and fitted with dual satellite/VHF collars. One of the 7 was still alive and being monitored as of June 30, 2006. 2006.--In spring 2006, 42 females were being monitored. We found 4 dens in May and June 2006 with 11 kittens total (Table 8). Lynx CO04F07, a female lynx born in Colorado in 2004, was the mother of one of these litters which documented the first recruitment of Colorado-born lynx into the Colorado breeding population. The percent of tracked females found with litters in 2006 was lower (0.095) than in the 3 previous years (0.413, SE = 0.032, Table 9). However, all demographic and habitat characteristics measured at the 4 dens that were found in 2006 were comparable to all other dens found (Table 9). Mean number of kittens per litter from 2003-2006 was 2.78 (SE = 0.05) and sex ratio of females to males was equal ( x = 1.14, SE = 0.14). 12 �Den Sites.--A total of 37 dens have been found from 2003-2006. All of the dens except one have been scattered throughout the high elevation areas of Colorado, south of I-70. In 2004, 1 den was found in southeastern Wyoming, near the Colorado border. Dens were located on steep ( x slope = 30o , SE=2o), north-facing, high elevation ( x = 3354 m, SE = 31 m) slopes (Figure 3). The dens were typically in Engelmann spruce/subalpine fir forests in areas of extensive downfall of coarse woody debris (Figures 4, 5, 6). All dens were located within the winter use areas used by the females. Captures Two adult lynx were captured in 2001 for collar replacement. One lynx was captured in a tomahawk live-trap, the other was treed by hounds and then anesthetized using a jab pole. Five adult lynx were captured in 2002; 3 were treed by hounds and 2 were captured in padded leghold traps. In 2004, 1 lynx was captured with a Belisle snare and 6 other adult lynx were captured in box-traps. Trapping effort was substantially increased in winter and spring 2005 and 12 adult lynx were captured and re-collared. Eight reintroduced lynx were captured in winter and spring 2006. All lynx captured in 2005 and 2006 were caught in box-traps. All captured lynx were fitted with new Sirtrack TM dual VHF/satellite collars. Seven adult lynx were captured from March 1999-June 30, 2006 because they were in poor body condition. Five of these lynx were successfully treated at the Frisco Creek Rehabilitation Center and rereleased in the Core Release Area. One lynx, BC00F7, died from starvation and hypothermia. Lynx QU04M07 died on February 5, 2006 at the rehabilitation center. Necropsy results documented starvation as the cause of death that was precipitated by hydrocephalus and bronchopneumonia (unpublished data T. Spraker, CSUVTH). Two lynx were captured because they were in atypical habitat outside the state of Colorado. They were held at Frisco Creek Rehabilitation Center for a minimum of 3 weeks and rereleased in the Core Release Area in Colorado. Prior to release these lynx were fitted with new Sirtrack TM dual VHF/satellite collars. In addition, 14 Colorado-born kittens were captured and collared at approximately 10-months of age. Seven 2004-born kittens were collared in spring 2005, and 7 2005-born kittens collared in spring 2006. HABITAT USE Landscape-scale daytime habitat use was documented from 7421 aerial locations of lynx collected from February 1999-June 30, 2005. Throughout the year Engelmann spruce / subalpine fir was the dominant cover used by lynx. A mix of Engelmann spruce, subalpine fir and aspen (Populus tremuloides) was the second most common cover type used throughout the year. Various riparian and riparian-mix areas were the third most common cover type where lynx were found during the daytime flights. Use of Engelmann spruce-subalpine fir forests and Engelmann spruce-subalpine fir-aspen forests was similar throughout the year. There was a trend in increased use of riparian areas beginning in July, peaking in November, and dropping off December through June. Site-scale habitat data collected from snow-tracking efforts indicate Engelmann spruce and subalpine fir were also the most common forest stands used by lynx for all activities during winter in southwestern Colorado. Comparisons were made among sites used for long beds, dens, travel and where they made kills. Little difference in aspect, mean slope and mean elevation were detected for 3 of the 4 site types including long beds, travel and kills where lynx typically use gentler slopes ( x = 15.7o ) at an mean elevation of 3173 m, and varying aspects with a slight preference for north-facing slopes (Figure 3). Mean percent total overstory was higher for long bed and kill sites than travel or den sites (Figure 4). Engelmann spruce provided a mean of 35.87% overstory for kills and long beds, with travel sites averaging 28% and den sites having the lowest mean percent overstory of 23% (Figure 4). Mean percent 13 �subalpine fir or aspen overstory did not vary across use sites (Figure 4). Willow overstory was highly variable and no dens were located in willow overstory. A total of 1841 site-scale habitat plots were completed in winter from December 2002 through April 2005. The most common understory species at all 3 height categories above the snow (low = 00.5m, medium = 0.51 - 1.0 m, high = 1.1 - 1.5 m) was Engelmann spruce, subalpine fir, willow (Salix spp.) and aspen (Figure 5). Various other species such as Ponderosa pine (Pinus ponderosa), lodgepole pine (Pinus contorta), cottonwood (Populus sargentii), birch (Betula spp.) and others were also found in less than 5% of the habitat plots. If present, willow provided the greatest percent cover within a plot followed by Engelmann spruce, subalpine fir, aspen and coarse woody debris for long beds, kills and travel sites. Areas documented in willow used by lynx are typically on the edge of willow thickets as tracks are quickly lost within the thicket. Den sites had significantly higher percent understory cover for all three height categories. Understory at den sites was primarily made up of coarse woody debris (Figure 5). The most common tree species documented in the site-scale habitat plots was Engelmann spruce Figure 6). Subalpine fir and aspen were also present in >35% of the plots. Most habitat plots were vegetated with trees of DBH < 6" (Figure 6). As DBH increased, percent occurrence decreased within the plot. Although decreasing in abundance as size increased, most lynx use sites had trees in each of the DBH categories, indicating mature forest stands except for dens. Den sites had a broad spectrum of Engelmann spruce tree sizes, including > 18” but no large subalpine fir or aspen trees. While Engelmann spruce and subalpine fir occurred in similar densities for kills, long beds and travel sites, den sites had twice the density of subalpine firs found at all other sites (Figure 6). DIET AND HUNTING BEHAVIOR Winter diet of lynx was documented through detection of kills found through snow-tracking. Prey species from failed and successful hunting attempts were identified by either tracks or remains. Scat analysis also provided information on foods consumed. A total of 400 kills were located from February 1999-April 2005. We collected 671 scat samples from February 1999-April 2004 that will be analyzed for content. In each winter, the most common prey item was snowshoe hare, followed by red squirrel (Table 10). A comparison of percent overstory for successful and unsuccessful snowshoe hare chases indicated lynx were more successful at sites with slightly higher percent overstory, if the overstory species were Englemann spruce, subalpine fir or willow. Lynx were slightly less successful in areas of greater aspen overstory (Figure 7). This trend was repeated for percent understory at all 3 height categories (Figure 8) except that higher aspen understory improved hunting success. Higher density of Engelmann spruce and subalpine fir increased hunting success while increased aspen density decreased hunting success (Figure 9). SNOWSHOE HARE ECOLOGY A study plan was completed to evaluate snowshoe hare densities, demography and seasonal movement patterns among small and medium tree-sized lodgepole pine stands and mature spruce/fir stands (Appendix I). DISCUSSION In an effort to establish a viable population of lynx in Colorado, CDOW initiated a reintroduction effort in 1997 with the first lynx released in winter 1999. From 1999 through spring 2005, 204 lynx were 14 �released in the Core Release Area. The reintroduction effort was augmented with the release of 14 additional animals in April 2006, bringing the total to 218 lynx reintroduced to southwestern Colorado. Locations of each lynx were collected through aerial- or satellite-tracking to document movement patterns and to detect mortalities. Most lynx remain in the high elevation, forested areas in southwestern Colorado. Dispersal movement patterns for lynx released in 2000 and subsequent years were similar to those of lynx released in 1999. However, more animals released in 2000 and subsequent years remained within the Core Release Area than those released in 1999. This increased site fidelity may have been due to the presence of con-specifics in the area on release. Numerous travel corridors have been used repeatedly by more than 1 lynx. These travel corridors include the Cochetopa Hills area for northerly movements, the Rio Grande Reservoir-Silverton-Lizardhead Pass for movements to the west, and southerly movements down the east side of Wolf Creek Pass to the southeast to the Conejos River Valley. Lynx appear to remain faithful to an area during winter months, and exhibit more extensive movements away from these areas in the summer. Most lynx currently being tracked are within the Core Release Area. During the summer months, lynx were documented to make extensive movements away from their winter use areas. Extensive summer movements away from areas used throughout the rest of the year have been documented in native lynx in Wyoming and Montana (Squires and Laurion 1999). Humancaused mortality factors such as gunshot and vehicle collision are currently the highest causes of death. Reproduction is critical to achieving a self-sustaining viable population of lynx in Colorado. Reproduction was first documented from the 2003 reproduction season and again in 2004, 2005 and 2006. Reproduction in 2006 included a Colorado-born female giving birth to 2 kittens, documenting the first recruitment of Colorado-born lynx into the Colorado breeding population. Additional reproduction is likely to have occurred in all years from females we are no longer tracking, and from Colorado-born lynx that have not been collared. The dens we find are more representative of the minimum number of litters and kittens in a reproduction season. To achieve a viable population of lynx, enough kittens need to be recruited into the population to offset the mortality that occurs in that year and hopefully even exceed the mortality rate for an increasing population. Mowat et al. (1999) suggest lynx and snowshoe hare select similar habitats except that hares select more dense stands than lynx. Very dense understory limits hunting success of the lynx and provides refugia for hares. Given the high proportion of snowshoe hare in the lynx diet in Colorado, we might then assume the habitats used by reintroduced lynx also depict areas where snowshoes hare are abundant and available for capture by lynx in Colorado. From both aerial locations taken throughout the year and from the site-scale habitat data collected in winter, the most common areas used by lynx are in stands of Engelmann spruce and subalpine fir. This is in contrast to adjacent areas of Ponderosa pine, pinyon juniper, aspen and oakbrush. The lack of lodgepole pine in the areas used by the lynx may be more reflective of the limited amount of lodgepole pine in southwestern Colorado, the Core Release Area, rather than avoidance of this tree species. Hodges (1999) summarized habitats used by snowshoe hare from 15 studies as areas of dense understory cover from shrubs, stands that are densely stocked, and stands at ages where branches have more lateral cover. Species composition and stand age appears to be less correlated with hare habitat use than is understory structure (Hodges 1999). The stands need to be old enough to provide dense cover and browse for the hares and cover for the lynx. In winter, the cover/browse needs to be tall enough to still provide browse and cover in average snow depths. Hares also use riparian areas and mature forests with understory. Site-scale habitat use documented for lynx in Colorado indicate lynx are most commonly using areas with Engelmann spruce understory present from the snow line to at least 1.5 m above the snow. The mean percent understory cover within the habitat plots is typically less than 15% regardless of understory species. However, if the understory species is willow, percent understory cover is typically 15 �double that, with mean number of shrubs per plot approximately 80, far greater than for any other understory species. In winter, hares browse on small diameter woody stems (<0.25"), bark and needles. In summer, hares shift their diet to include forbs, grasses, and other succulents as well as continuing to browse on woody stems. This shift in diet may express itself in seasonal shifts in habitat use, using more or denser coniferous cover in winter than in summer. The increased use of riparian areas by lynx in Colorado from July to November may reflect a seasonal shift in hare habitat use in Colorado. Major (1989) suggested lynx hunted the edge of dense riparian willow stands. The use of these edge habitats may allow lynx to hunt hares that live in habitats normally too dense to hunt effectively. The use of riparian areas and riparian-Engelmann spruce-subalpine fir and riparian-aspen mixes documented in Colorado may stem from a similar hunting strategy. However, too little is known about habitat use by hares in Colorado to test this hypothesis at this time. Lynx also require sufficient denning habitat. Denning habitat has been described by Koehler (1990) and Mowat et al. (1999) as areas having dense downed trees, roots, or dense live vegetation. We found this to be in true in Colorado as well. In addition, the dens used by reintroduced lynx were at high elevations and on steep north-facing slopes. All females that were documented with kittens denned in areas within their winter-use area. Snow-tracking of released lynx provided information on hunting behavior and diet through documentation of kills, food caches, chases, and diet composition estimated through prey remains. Snowtracking results indicate the primary winter prey species are snowshoe hare and red squirrel, with other mammals and birds forming a minor part of the winter diet. In winter, lynx reintroduced to Colorado appear to be feeding on their preferred prey species, snowshoe hare and red squirrel in similar proportions as those reported for northern lynx during lows in the snowshoe hare cycle (Aubry et al., 1999). Caution must be used in interpreting the proportion of identified kills. Such a proportion ignores other food items that are consumed in their entirety and thus are biased towards larger prey and may not accurately represent the proportion of smaller prey items, such as microtines, in lynx winter diet. Through snowtracking we have evidence that lynx are mousing and several of the fresh carcasses have yielded small mammals in the gut on necropsy. The summer diet of lynx has been documented to include less snowshoe hare and more alternative prey than in winter (Mowat et al., 1999). All evidence suggests reintroduced lynx are finding adequate food resources. SUMMARY From results to date it can be concluded that CDOW has developed release protocols that ensure high initial post-release survival, and on an individual level lynx have demonstrated they can survive long-term in areas of Colorado. It has also been documented that reintroduced lynx could exhibit site fidelity, engage in breeding behavior and produce kittens that are recruited into the Colorado breeding population. What is yet to be demonstrated is whether current conditions in Colorado can support the recruitment necessary to offset annual mortality for a population to sustain itself. Monitoring of reintroduced lynx will continue in an effort to document such viability. ACKNOWLEDGEMENTS The lynx reintroduction program involves the efforts of literally hundreds of people across North America, in Canada and the U. S. Any attempt to properly acknowledge all the people who played a role in this effort is at risk of missing many people. The following list should be considered to be incomplete. 16 �CDOW CLAWS Team (1998-2001): Bill Andree, Tom Beck, Gene Byrne, Bruce Gill, Mike Grode, Rick Kahn (Program Leader), Dave Kenvin, Todd Malmsbury, Jim Olterman, Dale Reed, John Seidel, Scott Wait, Margaret Wild. CDOW: John Mumma (Director 1996-2000), Bruce McCloskey (Director 2001-present), Conrad Albert, Jerry Apker, Laurie Baeten, Cary Carron, Don Crane, Larry DeClaire, Phil Ehrlich, Lee Flores, Delana Friedrich, Dave Gallegos, Juanita Garcia, Drayton Harrison, Jon Kindler, Ann Mangusso, Jerrie McKee, Melody Miller, Mike Miller, Kirk Navo, Robin Olterman, Jerry Pacheo, Mike Reid, Ellen Salem, Eric Schaller, Mike Sherman, Jennie Slater, Steve Steinert, Kip Stransky, Suzanne Tracey, Anne Trainor, Brad Weinmeister, Nancy Wild, Perry Will, Lisa Wolfe, Brent Woodward, Kelly Woods, Kevin Wright. Lynx Advisory Team (1998-2001): Steve Buskirk, Jeff Copeland, Dave Kenny, John Krebs, Brian Miller (Co-leader), Mike Phillips, Kim Poole, Rich Reading (Co-leader), Rob Ramey, John Weaver. U. S. Forest Service: Kit Buell, Joan Friedlander, Dale Gomez, Jerry Mastel, John Squires, Fred Wahl, Nancy Warren. U. S. Fish and Wildlife Service: Lee Carlson, Gary Patton (1998-2000), Kurt Broderdorp. State Agencies: Gary Koehler (Washington). National Park Service: Steve King. Colorado State University: Alan B. Franklin, Gary C. White. Colorado Natural Heritage Program: Rob Schorr, Mike Wunder. Alaska: ADF&G: Cathie Harms, Mark Mcnay, Dan Reed (Regional Manager), Wayne Reglin (Director), Ken Taylor (Assist. Director), Ken Whitten, Randy Zarnke, Other:Ron Perkins (trapper), Dr. Cort Zachel (veterinarian). British Columbia: Dr. Gary Armstrong (veterinarian), Mike Badry (government), Paul Blackwell (trapper coordinator), Trappers: Dennis Brown, Ken Graham, Tom Sbo, Terry Stocks, Ron Teppema, Matt Ounpuu. Yukon: Government: Arthur Hoole (Director), Harvey Jessup, Brian Pelchat, Helen Slama, Trappers: Roger Alfred, Ron Chamber, Raymond Craft, Lance Goodwin, Jerry Kruse, Elizabeth Hofer, Jurg Hofer, Guenther Mueller (YK Trapper’s Association), Ken Reeder, Rene Rivard (Trapper coordinator), Russ Rose, Gilbert Tulk, Dave Young. Alberta: Al Cook. Northwest Territories: Albert Bourque, Robert Mulders (Furbearer Biologist), Doug Steward (Director NWT Renewable Res.), Fort Providence Native People. Quebec: Luc Farrell, Pierre Fornier. Colorado Holding Facility: Herman and Susan Dieterich, Loree Harvey, Rachel Riling. Pilots: Dell Dhabolt, Larry Gepfert, Al Keith, Jim Olterman, Matt Secor, Whitey Wannamaker, Steve Waters, Dave Younkin. Field Crews (1999-2006): Steve Abele, Brandon Barr, Bryce Bateman, Todd Bayless, Nathan Berg, Ryan Besser, Mandi Brandt, Brad Buckley. Patrick Burke, Braden Burkholder, Paula Capece, Stacey Ciancone, Doug Clark, John DePue, Shana Dunkley, Tim Hanks, Dan Haskell, Matt Holmes, Andy Jennings, Susan Johnson, Paul Keenlance, Patrick Kolar, Tony Lavictoire, Clay Miller, Denny Morris, Kieran O’Donovan, Gene Orth, Chris Parmater, Jake Powell, Jeremy Rockweit, Jenny Shrum, Josh Smith, Heather Stricker, Adam Strong, Dave Unger, David Waltz, Andy Wastell, Lyle Willmarth, Leslie Witter, Kei Yasuda, Jennifer Zahratka. Research Associates: Bob Dickman, Grant Merrill. Data Analysts: Karin Eichhoff, Joanne Stewart, Anne Trainor. Data Entry: Charlie Blackburn, Patrick Burke, Rebecca Grote, Angela Hill, Mindy Paulek. Photographs: Tom Beck, Bruce Gill, Mary Lloyd, Rich Reading, Rick Thompson. Funding: CDOW, Great Outdoors Colorado (GOCO), Turner Foundation, U.S.D.A. Forest Service, Vail Associates. LITERATURE CITED AUBRY, K. B., G. M. KOEHLER, J. R. SQUIRES. 1999. Ecology of Canada lynx in southern boreal forests. Pages 373-396 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. BYRNE, G. 1998. Core area release site selection and considerations for a Canada lynx reintroduction in Colorado. Report for the Colorado Division of Wildlife. CURTIS, J. T. 1959. The vegetation of Wisconsin. University of Wisconsin Pres, Madison. GANEY, J. L. AND W. M. BLOCK. 1994. A comparison of two techniques for measuring canopy closure. Western Journal of Applied Forestry 9:1: 21-23. 17 �HODGES, K. E. 1999. Ecology of snowshoe hares in southern boreal and montane forests. Pages 163206 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. KOEHLER, G. M. 1990. Population and habitat characteristics of lynx and snowshoe hares in north central Washington. Canadian Journal of Zoology 68:845-851. KOLBE, J. A., J. R. SQUIRES, T. W. PARKER. 2003. An effective box trap for capturing lynx. Journal of Wildlife Management 31:980-985. LAYMON, S. A. 1988. The ecology of the spotted owl in the central Sierra Nevada, California. PhD Dissertation University of California, Berkeley, California. MAJOR, A. R. 1989. Lynx, Lynx canadensis canadensis (Kerr) predation patterns and habitat use in the Yukon Territory, Canada. M. S. Thesis, State University of New York, Syracuse. MOWAT, G., K. G. POOLE, AND M. O’DONOGHUE. 1999. Ecology of lynx in northern Canada and Alaska. Pages 265-306 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. POOLE, K. G., G. MOWAT, AND B. G. SLOUGH. 1993. Chemical immobilization of lynx. Wildlife Society Bulletin 21:136-140. SHENK, T. M. 1999. Program narrative: Post-release monitoring of reintroduced lynx (Lynx canadensis) to Colorado. Report for the Colorado Division of Wildlife. __________. 2001. Post-release monitoring of lynx reintroduced to Colorado. Job Progress Report for the Colorado Division of Wildlife. Fort Collins, Colorado. SQUIRES, J. R. AND T. LAURION. 1999. Lynx home range and movements in Montana and Wyoming: preliminary results. Pages 337-349 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University Press of Colorado, Boulder, Colorado. U. S. FISH AND WILDLIFE SERVICE. 2000. Endangered and threatened wildlife and plants: final rule to list the contiguous United States distinct population segment of the Canada lynx as a threatened species. Federal Register 65, Number 58. WHITE, G.C. AND K. P. BURNHAM. 1999. Program MARK: Survival estimation from populations of marked animals. Bird Study 46 Supplement, 120-138. WILD, M. A. 1999. Lynx veterinary services and diagnostics. Job Progress Report for the Colorado Division of Wildlife. Fort Collins, Colorado. Prepared by _______________________________ Tanya M. Shenk, Wildlife Researcher 18 �Table 1. Definitions of forest structure classes used to describe habitat sites (Thomas 1979). Forest Structure Class Definition Grass/forb The grass/forb stage is created naturally by a catastrophic event, such as wildfire, and is typified by the near complete absence of snags, litter or down material in the aspen and ponderosa pine types, or vice versa in the lodgepole or subalpine forest types. Shrub/seedling The shrub/seedling stage occurs when tree seedlings or shrubs grow up to 2.5 cm at diameter breast height (DBH), either naturally or artificially through planting. Sapling/pole The sapling/pole stage is a young stage where tree DBH's range from 2.517.5 cm with tree heights ranging 1.8-13.5 m. These trees are 5-100 years of age, depending on species and site condition. Mature The mature stage occurs when tree diameters reach a relatively large size (25-50 cm) and the trees are usually 90 or more years old. Forest stands begin to experience accelerated mortality from disease and insects. Old-growth The old-growth stage occurs when a mature stand is at advanced age (100 years for aspen or 200 years for spruce), is very slow growing, and has advanced degrees of disease, insects, snags, and down, dead material. An exception to this occurs in ponderosa pine and aspen types where these old-growth stands typically experience low densities of down dead material or snags. Table 2. Lynx released in Colorado from February 1999 through June 30, 2006. Year Females Males TOTAL 1999 22 19 41 2000 35 20 55 2003 17 16 33 2004 17 20 37 2005 18 20 38 2006 6 8 14 TOTAL 115 103 218 Table 3. Status of adult lynx reintroduced to Colorado as of June 30, 2006. Females Released Known Dead Possible Alive Missing Tracking a 1 is unknown mortality 115 46 69 20 49 Males 103 33 70 24 46 19 Unknown 1 TOTALS 218 80 138 43a 95 �Table 4. Causes of death for lynx released into southwestern Colorado from 1999-2006 as of June30, 2006. Cause of Death Unknown Hit by Vehicle Starvation Shot Other Trauma Probable Shot Plague Predation Probable Predation Illness Total Mortalities Number of Mortalities 26 11 10 9 7 5 5 3 2 2 80 Table 5. Lynx reproduction documented in 2003. Female BC00F8 BC00F19 YK00F16 YK99F1 YK00F19 YK00F10 Release Year 2000 2000 2000 1999 2000 2000 Date Den Found 5/21/03 5/26/03 6/19/03 6/10/03 6/11/03 5/31/03 TOTAL Females ? 1 1 2 1 2 7 Number of Kittens Males ? 1 1 1 2 2 7 Total 2 2 2 3 3 4 16 Table 6. Lynx reproduction documented in 2004. Female ID YK00F2 AK00F2 YK00F1 YK00F15 BC00F14 BC00F18 YK00F10 BC03F02 BC03F10 BC03F09 YK00F7 YK99F1 Unknown Unknown TOTAL Release Year 2000 2000 2000 2000 2000 2000 2000 2003 2003 2003 2000 1999 Previous Litters Date Den Found 5/28/2004 5/31/2004 6/1/2004 6/4/2004 6/7/2004 6/10/2004 6/17/2004 6/25/2004 6/26/2004 6/29/2004 6/30/2004 6/2004 Date Kittens Found Dec 2004 Sept 2004 Feb 2005 20 Number of Kittens Females Males Total 3 1 4 2 1 3 3 3 1 2 3 1 2 3 4 4 1 1 2 2 2 2 2 1 1 2 1 1 2 2 4 3 19 11 39 �Table 7. Lynx reproduction documented in 2005. Female ID AK00F02 YK00F15 YK00F10 YK00F11 YK00F01 YK00F07 BC00F18 BC03F02 BC03F01 QU03F06 QU03F04 QU03F07 BC03F09 BC03F10 BC04F01 BC04F03 BC04F05 BC04F04 TOTAL Release Year 2000 2000 2000 2000 2000 2000 2000 2003 2003 2003 2003 2003 2003 2003 2004 2004 2004 2004 Previous Litters 2004 2004 2003, 2004 2004 2004 2004 2004 2004 2004 Date Den Found 5/21/2005 5/28/2005 6/1/2005 6/9/2005 6/10/2005 6/14/2005 6/24/2005 5/25/2005 5/27/2005 6/5/2005 6/14/2005 6/16/2005 6/27/2005 6/2005 6/11/2005 6/19/2005 6/23/2005 Date Kittens Found 12/20/2005 12/10/2005 Number of Kittens Males Females Total 2 1 3 1 1 2 2 2 4 2 2 2 1 3 1 2 3 1 1 2 1 1 2 2 2 4 3 3 1 2 3 3 1 4 1 1 2 2 2 1 3 1 3 4 3 3 1 1 26 22 50 Table 8. Lynx reproduction in 2006. Female ID AK00F15 AK00F05 BC03F10 CO04F07 TOTAL Release Year 2000 2000 2003 Year Born in Colorado Previous Litters 2004, 2005 2004 2004, 2005 Date Den Found 5/21/2006 6/7/2006 6/9/2006 6/17/2006 2004 Number of Kittens Males Females Total 1 3 4 1 2 3 1 1 2 2 2 5 6 11 Table 9. Lynx reproduction summary statistics for 2003-2006. Additional Litters Found in Winter Mean # Kittens/Litter 16 0.462 2 0.425 1 2.67 (SE = 0.33 2.83 (SE = 0.24) 2.88 (SE = 0.18) 2.75 (SE = 0.47) 2.78 (SE = 0.05) 11 Year # Females Tracked # Dens Found in May/June % Tracked Females with Kittens 2003 17 6 0.353 2004 26 11 2005 40 17 Mean 2003-05 2006 Mean 2003-06 Total Kittens Found 39 50 Sex Ratio M/F 1.0 1.5 0.8 0.413 (SE =0.032) 42 4 0.095 0.334 (SE = 0.083) 21 TOTAL 116 1.2 1.14 (SE = 0.14) �Table 10. Number of kills found each winter field season through snow-tracking of lynx and percent composition of kills of the three primary prey species. Field Season 1999 1999-2000 2000-2001 2001-2002 2002-2003 2003-2004 2004-2005 n 9 83 89 54 65 37 78 Snowshoe Hare 55.56 67.47 67.42 90.74 90.77 67.57 83.33 Prey (%) Red Squirrel Cottontail 22.22 0 19.28 1.20 19.10 8.99 5.56 0 6.15 0 27.03 2.70 10.26 0 Other 22.22 12.05 4.49 3.70 3.08 2.70 6.41 Figure 1. Lynx are monitored throughout Colorado and by satellite throughout the western United States. The lynx core release area, where all lynx were released, is located in southwestern Colorado. A lynxestablished core use area has developed in the Taylor Park and Collegiate Peak area in central Colorado. 22 �: 6 : !1 • - • - • • - • l 1 l f , 1 12• • • • • i i f i l i , e EB e e 1• 1 1 l 1 l f : !4. •-•-• • 1 l 1 , !5•-•-•-•-• I - - - - - --- - --- -- - -- ---- - - - - -- ---- - -- - - - - - - - - - - l 12meters 15 meters Figure 2. Design of site-scale habitat plot sampling plot. Each of the 25 points are 3 meters apart (the first 6 points are labeled 1-6). The object that triggered a habitat plot (e.g., kill ) is the center point, depicted by the hollow circle. The actual pints encompass a 12 m x 12 m square but the understory and overstory data collected are influenced by vegetation occurring within a 15 m x 15 m square. TRAVEL n = 441 LONG BED n = 458 x Elevation= 3194 m SE = 9 m x Slope = 17° SE = 0.6° x Elevation= 3179 m SE = 9 m x Slope = 14° SE = 0 5° SSH KILL n = 238 DEN n = 37 x Elevation= 3145 m SE = 14 m x Slope = 16° SE = 0 8° x Elevation = 3354 SE= 31 m x Slope= 30° SE~. . . ' Figure 3. Frequency of aspect with mean vector and 95%confidence interval depicted as grey bars on graphs for 4 lynx use sites; dens, long beds, kills and travel as well as mean elevation and SE and mean slope and SE . 23 �80 70 60 'i:: 50 8 40 V p_. 30 20 10 0 ES SF AS \XII Total Cover Tree Species I ■ L ong Beds ■ Kills □ T ravel □ Den Sites I Figure 4. Mean percent overstory by tree species Engelmann spruce (ES), subalpine fir (SF), aspen (AS), willow (WI) and total cover for 4 different lynx use sites: long beds, kill sites, travel and den sites. Confidence intervals (95%) are depicted by error bars. 40 ~ - - - - - - - - - - - - - - ~ ~J; 01 ITT,CTl ~.rr1I 70 - r - - - - - - - - - - - - - - - - - - - , 60 Den Sites lO 40 ~o 20 10 Most Common Understory Species ES= Engelmann spruce SF = Subalpine fir 'w'.I = Willow AS = Aspen LO = Lodgepole pine C\YD = Coarse woody debri s Understory Species Figure 5. Mean percent understory by tree species Engelmann spruce (ES), subalpine fir (SF), coarse woody debris (CWD), aspen (AS), willow (WI), and total cover for 4 different lynx use sites: long beds, kill sites, travel, and den sites. 24 �Kills 800 60J (].) 40J ~ .su 20J (].) ~ SF ES SF ES AS AS Tree Species I■ o-6 0 6-12 12-18 018-24 0 > 241 Figure 6. Mean tree density by species Engelmann spruce (ES), subalpine fir (SF), and aspen (AS) and dbh size class for 4 different lynx use sites. 80 70 ...u 60 :,. 0 50 ~ 40 u C: ...uuu 30 p... 20 10 0 FS SF AS WI Total Cover: Tree Species I ■ Successful Chases i:r,!] Unsuccessful Chases I Figure 7. Mean percent overstory by tree species Engelmann spruce (ES), subalpine fir (SF), aspen (AS), willow (WI) and total cover for successful and unsuccessful snowshoe hare chases. Confidence intervals (95%) are depicted by error bars. 25 �50 45 H 40 (!) ~ 35 0 u.J-> 30 25 ~ (!) u 20 H (!) 15 p_, 10 5 0 ES SF C\X/D \XII AS Tree Species Tot:al Cover ■ 0 - 0. 5 m S rlol 0 - 0.5 m U ■ 0.5 - 1.0 m S rlol 0.5 - 1.0 m U □ 1.0 - 1.5 m S ~ 1.0 - 1.5 m U Figure 8. Mean percent understory by tree species Engelmann spruce (ES), subalpine fir (SF), apsen (AS), willow (WI), and total cover for 3 different understory height categories for successful and unsuccessful snowshoe hare chases. Confidence intervals (95%) are depicted by error bars. 1200 I 0-6 SC l!I 0-6 UC 1000 <l) 1 6-12 SC w Zlu 800 ri!I 6-12 UC <l) X 600 ,.,, ll 12-18 SC -....._ 1!112-18 UC <l) <l) w 400 118-24 SC E--< l!I 18-24 UC 200 I >24 SC ri!1 >24 UC 0 ES SF AS Tree Species Figure 9. Mean tree density by species Engelmann spruce (ES), subalpine fir (SF), and aspen (AS) and 5 dbh size classes for successful chases (SC) and unsuccessful chases (UC) of snowshoe hares. 26 �APPENDIX I PROGRAM NARRATIVE STUDY PLAN FOR MAMMALS RESEARCH FY 2005-06 – FY 2009-10 State of: Cost Center: Work Package: Task No.: Colorado 3430 0670 2 Federal Aid Project No.: : Division of Wildlife : Mammals Research : Lynx Conservation : Density, Demography, and Seasonal Movements of Snowshoe Hare in Colorado N/A : DENSITY, DEMOGRAPHY, AND SEASONAL MOVEMENTS OF SNOWSHOE HARES IN COLORADO Principal Investigators Jacob S. Ivan, Ph. D. Candidate, Colorado State University Tanya M. Shenk, Wildlife Researcher, Mammals Research, Colorado Division of Wildlife Cooperators Gary C. White, Professor, Fishery and Wildlife Biology, Colorado State University STUDY PLAN APPROVAL Prepared by: _____________________________ Date: __________________ Submitted by: ____________________________ Date: ___________________ Reviewed by: ____________________________ Date: ___________________ ____________________________ Date: ___________________ ____________________________ Date: ___________________ Reviewed by: ____________________________ Biometrician Date: ___________________ Approved by: ____________________________ Mammals Research Leader Date: ___________________ 27 �DENSITY, DEMOGRAPHY, AND SEASONAL MOVEMENTS OF SNOWSHOE HARES IN COLORADO NEED A program to reintroduce the threatened Canada lynx (Lynx canadensis) into Colorado was initiated in 1997. Since that time, 204 lynx have been released in the state, and an extensive effort to determine their movements, habitat use, reproductive success, and food habits has ensued (Shenk 2005). Analysis of scat collected from winter snow tracking indicates that snowshoe hares (Lepus americanus) comprise 65–90% of the winter diet of reintroduced lynx (T. Shenk, Colorado Division of Wildlife, unpublished data). Thus, as in the far north where the intimate relationship between lynx and snowshoe hares has captured the attention of ecologists for decades, it appears that the existence of lynx in Colorado and the success of the reintroduction effort may hinge on maintaining adequate and widespread populations of hares. Colorado represents the extreme southern range limit for both lynx and snowshoe hares (Hodges 2000). At this latitude, habitat for each species is less widespread and more fragmented compared to the continuous expanse of boreal forest at the heart of lynx and hare ranges. Neither exhibits dramatic cycles as occur farther north, and typical lynx (≤2−3 lynx/100km2; Aubry et al. 2000) and hare (≤1−2 hares/ha; Hodges 2000) densities in the southern part of their range correspond to cyclic lows form northern populations (2-30 lynx/100 km2, 1−16 hares/ha; Aubry et al. 2000, Hodges 2000, Hodges et al. 2001). Whereas extensive research on lynx-hare ecology has occurred in the boreal forests of Canada, literature regarding the ecology of these species in the southern portion of their range is relatively sparse. This scientific uncertainty is acknowledged in the “Canada Lynx Conservation Assessment and Strategy,” a formal agreement between federal agencies intended to provide a consistent approach to lynx conservation on public lands in the lower 48 states (Ruediger et al. 2000). In fact, one of the explicit guiding principles of this document is to “retain future options…until more conclusive information concerning lynx management is developed.” Thus, management recommendations in this agreement are decidedly conservative, especially with respect to timber management, and are applied broadly to cover all habitats thought to be of possible value to lynx and hare. This has caused controversy where recommendations conflict with competing resource management goals. Accurate identification and detailed description of lynx-hare habitat in the southern Rocky Mountains would permit more informed and refined management recommendations. A commonality throughout the snowshoe hare literature, regardless of geographic location, is that hares are associated with dense understory vegetation that provides both browse and protection from elements and predators (Wolfe et al. 1982, Litvaitis et al. 1985, Hodges 2000, Homyack et al. 2003, Miller 2005). In western mountains, this understory can be provided by relatively young conifer stands regenerating after stand-replacing fires or timber harvest (Sullivan and Sullivan 1988, Koehler 1990, Koehler 1990, Bull et al. 2005) as well as mature, uneven-aged stands (Beauvais 1997, Griffin 2004). Hares may also take advantage of seasonally abundant browse and cover provided by deciduous, open habitats (e.g., riparian willow [Salix spp.], aspen [Populus tremuloides]; Wolff 1980, Miller 2005). In drier portions of hare range, such as Colorado, regenerating stands can be relatively sparse, and hares may be more associated with mesic, late-seral forest and/or riparian areas than with young stands (Ruggiero et al. 2000). Numerous investigators have sought to determine the relative importance of these distinctly different habitat types with regards to snowshoe hare ecology. Most previous evaluations were based on hare density or abundance (Bull et al. 2005), indices to hare density and abundance (Wolfe et al. 1982, Koehler 1990, Beauvais 1997, Miller 2005), survival (Bull et al. 2005), and/or habitat use (Dolbeer and 28 �Clark 1975). Each of these approaches provides insight into hare ecology, but taken singly, none provide a complete picture and may even be misleading. For example, extensive use of a particular habitat type may not accurately reflect the fitness it imparts on individuals, and density can be high even in “sink” habitats (Van Horne 1983). A more informative approach would be to measure density, survival, and habitat use simultaneously in addition to recruitment and population growth rate through time. Griffin (2004) employed such an approach and found that summer hare densities were consistently highest in young, dense stands. However, he also noted that only dense mature stands held as many hares in winter as in summer. Furthermore hare survival seemed to be higher in dense mature stands, and only dense mature stands were predicted (by matrix projection) to impart a mean positive population growth rate on hares. Griffin’s (2004) study occurred in the relatively moist forests of Montana, which share many similarities but also many notable differences with Colorado forests including levels of fragmentation, species composition, elevation, and annual precipitation. Density estimation is a key component in assessing the value of a particular stand type and is the common currency by which hare populations are compared across time and space. However, it can be a difficult metric to estimate accurately. Density estimation based on capture-recapture methods is a welldeveloped field (Otis et al. 1978, White et al. 1982), but is often too costly and labor intensive to be implemented on scales necessary to effectively monitor density over a biologically meaningful area. Also, density can be difficult to assess from grid-trapping efforts because it is often unclear how much area was effectively sampled by the grid (Williams et al. 2002:314). Different approaches can produce density estimates that differ by an order of magnitude even when calculated from the same data (Zahratka 2004). Indices such as pellet plot counts and distance sampling of pellet groups can be used to estimate density, but each of these has limitations as well (Krebs et al. 1987, Eriksson 2006). Pellet plot counts are typically conducted by laying out numerous rectangular or circular plots along transect lines randomly placed within a study site. All pellets occurring within the plot are counted and removed on an annual basis. The mean number of pellets per plot is then inserted into a regression equation that gives an estimate of hare density (Krebs et al. 1987). Estimates from this technique correlate well with density estimates derived from simultaneous mark-recapture studies occurring in the same area (Krebs et al. 2001, Murray et al. 2002, Mills et al. 2005, Homyack et al. 2006). However, because fecal deposition rates can vary by season and diet, and because pellet decomposition rates can vary with altitude, climate, aspect, precipitation, and cover type, region-specific, stand-specific, and/or season-specific equations should be developed before this technique is employed for a given area and season (Krebs et al. 2001, Prugh and Krebs 2004, Murray et al. 2005). Density estimates vary with plot size and shape, requiring equations specific to these geometric considerations as well (McKelvey et al. 2002). Pellet counts tend to yield more precise and unbiased density estimates when plots are visited and cleared more than once per year (e.g., plots cleared in the fall and then counted in the spring to estimate winter density) because variability in deposition and decomposition rates is reduced (Homyack et al. 2006). However, this requires considerably more work and expense than an annual survey. Some studies have conducted pellet plot counts without first clearing plots (e.g., Bartmann and Byrne 2001). This saves time and money, but requires the ability to discern fresh (this year) pellets from old pellets, which can be difficult and is generally not a recommended approach (Prugh and Krebs 2004, Murray et al. 2005). Distance sampling is a well-developed method for estimating the density of objects in a given area (Buckland et al. 2001). In general, observers walk a pre-defined sampling transect and record each object of interest along with the perpendicular distance of that object from the transect line. This information is then used to develop a detection function which is in turn used to estimate density (Buckland et al. 2001). The method assumes all objects on the line are seen with certainty, objects are not double-counted, distance measures are accurate, and transect lines are located randomly within a study area (Buckland et al. 2001). Recently, distance sampling has been used to indirectly estimate hare density 29 �by first estimating the pellet group density of hares, then using fecal deposition and decomposition rates as a link back to hare density (Eriksson 2006). In general, distance sampling is more efficient than pellet plot counts as it does not require the tedious layout of hundreds of plots or counting individual pellets. This advantage is most recognizable in situations where pellet groups occur at low densities. Conversely, at extremely high densities, it may become difficult to distinguish pellet groups, and plots may be preferable (Marques et al. 2001). Regardless, distance sampling of pellet groups to estimate animal density also requires habitat and season specific decomposition and defecation rates, which can be difficult to obtain (Marques et al. 2001). For this project, I have chosen to provide land managers with information relating demographic rates, as well as density, to stand characteristics. Thus, I will use mark-recapture techniques as data from such an approach can provide information on both density and demography. I will address the “effective trapping area” issue using a new approach that augments mark-recapture data with telemetry locations of animals using the grid. The study outlined below is designed principally to evaluate the importance of young, regenerating lodgepole pine (Pinus contorta) and mature Engelmann spruce (Picea engelmannii)/ subalpine fir (Abies lasiocarpa) stands in Colorado by examining density and demography of snowshoe hares that reside in each (Figure 1). My hope is that information gathered from this research will be drawn upon as managers make routine decisions, leading to landscapes that include stands capable of supporting abundant populations of hares. I assume that if management agencies focus on providing habitat, hares will persist. Specifically, I will evaluate small and medium lodgepole pine stands and large spruce/fir stands where the classes “small”, “medium”, and “large” refer to the diameter at breast height (dbh) of overstory trees as defined in the United States Forest Service R2VEG Database (small = 2.54−12.69 cm dbh, medium = 12.70−22.85 cm, and large = 22.86−40.64 cm dbh; J. Varner, United States Forest Service, personal communication). To maximize comparability, I will choose lodgepole stands so that all are generating from harvest or all are regenerating following fire. I also intend to identify which of the numerous density-estimation procedures available perform accurately and consistently using an innovative, telemetry augmentation approach as a baseline. I will assess movement patterns and seasonal use of deciduous cover types such as riparian willow. Finally, I will further expound on the relationship between density, demography, and stand type by examining how snowshoe hare density and demographic rates vary with specific vegetation, physical, and landscape characteristics of a stand. Figure 1. Purported high quality snowshoe hare habitat in Colorado. From left to right: small lodgepole pine, medium lodgepole pine, and large Engelmann spruce/subalpine fir. 30 �OBJECTIVES 1) Compare telemetry-corrected estimates of density to those that would have been obtained from other commonly employed techniques used to convert population size estimated from a trapping grid to density (i.e., mean maximum distance moved, ½ mean maximum distance moved, ½ trap interval, nested grids, Program DENSITY). The purpose is to determine which common technique requiring less effort most consistently matches estimates from the intensive, telemetry-corrected approach. 2) Assess the relative value of the 3 stand types that purportedly provide high quality hare habitat by estimating and comparing survival, recruitment, finite population growth rate, and maximum (late summer) and minimum (late winter) snowshoe hare densities for each type. 3) Describe the timing, duration, and extent of broad-scale, seasonal movement patterns of snowshoe hares. 4) Relate specific vegetation, physical, and landscape characteristics of the 3 stand types to snowshoe hare density and demographics. APPROACH Hypotheses 1) In general, snowshoe hare density in Colorado will be relatively low (≤0.5 hares/ha) compared to densities reported in northern boreal forests, even immediately post-breeding when an influx of juveniles will bolster hare numbers. 2) Snowshoe hare density will be consistently highest in small lodgepole pine stands, followed by large spruce/fir and medium lodgepole pine, respectively. 3) Survival will generally be highest in mature (large) spruce/fir stands followed by small and medium lodgepole pine, respectively. 4) Finite population growth rate will be consistently at or above 1.0 in mature spruce/fir stands with survival contributing most significantly to the growth rate. Finite growth rates for the lodgepole pine stands will be more variable. 5) Snowshoe hares will significantly shift their home ranges to make use of abundant food and cover provided by riparian willow (and/or aspen) habitats in summer. 6) Snowshoe hare density, survival, and recruitment will be highly correlated with understory cover and stem density. Experimental Design/Procedures Variables.--The response variables of interest for this project include stand-specific snowshoe hare density (D), apparent survival (φ), recruitment (f), finite population growth rate (λ), and a metric of seasonal movement. Density is the number of hares per unit area and will be estimated using a variety of conventional techniques as well as a rigorous method that incorporates radio telemetry. The standspecific demographic parameters will be estimated primarily from capture-mark-recapture methods. As such, apparent survival is defined as the probability that a marked animal alive and in the population at time i survives and is in the population at time i + 1. Apparent survival encompasses losses due to both death and emigration. Recruitment is the number of new animals in the population at time i + 1 per animal in the population at time i. New recruits can arise from on-site reproduction as well as immigration. The finite population growth rate is the number of animals in a given age class at time i + 1 divided by the number present at time i. Shifts in home range will be assessed by comparing the seasonal proportion of telemetry locations in deciduous habitats using multi-response permutation procedures (MRPP; Zimmerman et al. 1985, White and Garrott 1990). Potential explanatory variables for snowshoe hare density, demographics, and movement include general species composition and structural stage of each stand in which response variables are measured. Additionally, stem density, horizontal cover, and canopy cover (to a lesser extent) are highly correlated 31 �with snowshoe hare abundance and habitat use (Wolfe et al. 1982, Litvaitis et al. 1985, Hodges 2000, Zahratka 2004, Miller 2005). Thus, I will further characterize vegetation in each stand by measuring stem density by size class (1-7 cm, 7.1-10 cm, and >10 cm), percent canopy cover, percent horizontal cover of understory and basal area. Basal area is an easily obtainable metric that may be correlated with the other variables and is recorded routinely during timber cruises, whereas the others are not. Thus, it might prove a useful link for biologists designing management strategies for snowshoe hare. Additionally, I will record physical covariates such as ambient temperature, precipitation, and snow depth at each stand during sampling periods as well as precipitation 1-3 years prior to sampling. Finally, I will calculate potentially important landscape metrics such as patch size and level of fragmentation. Location--.Identification of a suitable study area for this project and others that may follow is ongoing. The general study area must consist of an interspersion of young lodgepole pine and mature spruce/fir forest juxtaposed closely with open, seasonal habitats such as riparian willow. Within this general area, 3 sites will be selected such that 1) the 3 stand types of interest (small and medium lodgepole, large spruce/fir) occur within each site, 2) sites are close enough geographically to minimize differences due to climate, weather, and topography, but are far enough apart to be considered independent (e.g., 3 sites might occur in 3 different, but adjacent drainages), 3) each stand type within a site is adjacent to a riparian area, and 4) stand types of interest occur within 1 km of an access road (for logistical purposes). Such an arrangement often occurs in east-west drainages where spruce/fir grows on the north-facing slope, lodgepole pine covers the south-facing slope, and a riparian/willow area with road access separates the two (Figure 2). Additionally, sites must 1) include stands of suitable size and shape to admit a 16.5-ha trapping grid, 2) be consistent in their management history (i.e., all lodgepole pine stands in all sites must be either thinned or un-thinned, all regenerating after fire or all regenerating after harvest), and 3) be consistent in their intensity of use by lynx (core areas or not). I recently obtained the U.S. Forest Service R2VEG GIS database (newest, most detailed stand inventory information available statewide) and am currently working to objectively select a suite of potential study sites that satisfy the above-stated conditions. Depending on the number of potential sites within this suite, I will choose a small set of provisional study areas to ground-truth based on logistical considerations (e.g, housing, access). I will randomly select the final study sites from among those that appeared qualitatively suitable during ground-truthing. Prior to data collection I will more intensively sample the vegetation characteristics of the various stand types within the selected study sites to ensure that they represent intended conditions. Sampling.--All trapping and handling procedures will be approved by the Colorado State University Animal Care and Use Committee and filed with the Colorado Division of Wildlife. Snowshoe hares breed synchronously and generally exhibit 2 birth pulses in Colorado (although in some years, some individuals may have 3 litters), with the first pulse terminating approximately June 5−20 and the second approximately July 15–25 (Dolbeer 1972). To obtain a maximum density estimate, I will begin data collection on site 1 immediately following the second birth pulse in late July. Along with a crew of 5 technicians, I will deploy one 7 × 12 trapping grid (50-m spacing between traps; grid covers 16.5 ha) in each of the 3 stand types of interest following Griffin (2004) and Zahratka (2004). Grid locations and orientation will be chosen randomly within each stand subject to the logistical constraint that they must be within 1 km of a road. Traps will be deployed in all 3 stands in a single day. As traps are deployed, they will be locked open and “pre-baited” with apple slices and commercial rabbit chow. On days 2-4, the crew will continue pre-baiting, replacing apples and rabbit chow as necessary. The purpose of this extended pre-baiting is to maximize capture rates when trapping begins. This will minimize the number of trap-nights needed to capture the desired number of animals which in turn will minimize trappingrelated stress as well as the likelihood that American marten (Martes americana) will key into trap lines and prey on entrapped hares, as has occurred in previous studies (J. Zahratka, personal communication). During pilot work in winter 2005, I observed low but increasing capture rates (<0.20) during the first 3 32 �nights of trapping, with higher, more stable capture probabilities after 3 days (approximately 0.35–0.45). Thus 3 days of pre-baiting seems reasonable. Study Area Site 1 D Site 2 Site 3 _____A.._____ \ ( Summer Winter FY06-07 FY08-09 FY07-08 Figure 2. Experimental design for study of snowshoe hare density, demography, and movement. Within the study area, 3 sites, each consisting of 3 forest stand types (light to dark gray shades) and a riparian area (medium gray shade), will be sampled (dotted trapping grids) during late summer and late winter for 3 years. 33 �Traps will be set on the afternoon of the 4th day and checked early each morning and again in the evening on days 5–9. By checking traps in both morning and evening I prevent hares from being entrapped >13 hours, which should minimize capture stress. Based on Zahratka (2004) and personal experience, I anticipate capturing up to 10–15 individual hares per grid. A crew of 2 people will work together on each grid to check traps and process captures as quickly as possible. All captured hares will be coaxed out of the trap and into a dark handling bag by blowing quick shots of air on them from behind. Hares will remain in the handling bag, physically restrained with their eyes covered, for the entire handling process. Each individual will be aged, sexed, marked with a passive integrated transponder (PIT) tag and temporary ear mark (to track PIT tag retention), then released. Aging will consist of assigning each individual as either juvenile (<1 year old, <1000 g) or adult (≥1 year old, ≥1000 g) based on weight. This criterion is accurate through the end of September at which point juveniles are difficult to distinguish from adults (K. Hodges, University of British Columbia; P. Griffin, University of Montana, personal communication). After the first day of trapping, all captured hares will be scanned for a PIT tag prior to any handling and those already marked will be recorded and immediately released. Traps and bait will be completely removed from the grid on day 10. In addition to PIT tags and ear marks, I will radio collar up to 10 hares captured on each grid with a 28-g mortality-sensing transmitter (BioTrack, LTD) to facilitate unbiased density estimation as well as assessment of seasonal movements. I expect heterogeneity in snowshoe hare movements and use of the grid area, with potential bias surfacing due to location at which a hare is captured (e.g., hares captured on the edge of a grid may use the grid area differently than those captured at the center), and differential behavioral responses to trapping (e.g., young individuals may have lower capture probabilities and thus may be more likely to be captured on later occasions). To guard against the first potential bias, I will randomly select a starting trap location each morning and run the grid systematically from that point. Thus, the first several hares encountered (and collared) will be as likely to be from the inner part of the grid as from the edge. To protect against the second potential source of bias, I will refrain from deploying the final 3 collars until days 4 and 5 of the trapping session. Immediately following the removal of traps, the field crew will begin work locating each radiocollared hare 1–2 times per day for 10 days. Locations will be obtained by “homing” on a signal (Samuel and Fuller 1996, Griffin 2004) taking care not to push hares while approaching them. Technicians will record their location with hand-held GPS units (Garmin model 12XL) as soon as a visual of the collared hare is obtained or if the signal can be picked up by the receiver without an antenna. Using the same make and model collars, Griffin (2004) found that hares are usually within ~15m when the signal came be received without an antenna (Griffin 2004). I will test this assumption with my telemetry equipment over a variety of transmitter locations and orientations. Because hares are largely nocturnal (Keith 1964, Mech et al. 1966, Foresman and Pearson 1999), an effort will be made to conduct telemetry work at various times of the night (safety and logistics permitting) and day to gather a representative sample of locations for each hare. The crew will gather telemetry locations for radio-collared hares on site 1 for 8−9 days. Then the 10−day trapping procedure and 8 to 9−day telemetry work will be repeated on the 3 grids comprising site 2 (Figure 3). The cycle will be repeated once more for grids on site 3 (Figure 3). The entire process will be repeated during the following winter when densities should be at a minimum. In summary, for any given 9-week sampling period, I will collect data from 9 total grids, 1 grid in each of 3 habitat types (stand types) across 3 sites. Sampling will occur during 2 such 9-week periods each year − once in late summer and once in late winter – and will continue for 3 years (Figure 2). During the interim between intensive trapping and telemetry work, a single technician and myself will attempt to gather 1–2 telemetry locations/hare/month in order to keep closer tabs on these individuals, 34 �determine more precisely when mortality occurs, and retrieve collars from dead hares. Telemetry work will also occur during “pre-baiting” days to determine which hares are still alive and immediately available to be sampled by the grid during the ensuing trapping period. Jul Aug Sep 1 Jan 1 2 3 4 2 3 4 5 6 7 8 5 6 7 8 9 10 11 9 10 11 12 13 14 15 12 13 14 15 16 17 18 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 1 2 3 4 6 7 8 9 10 13 14 15 16 19 20 21 22 23 24 25 26 27 28 29 30 31 1 5 2 3 4 5 6 7 8 11 12 9 10 11 12 13 14 15 17 18 19 16 17 18 19 20 21 22 Feb 20 21 22 23 24 25 26 23 24 25 26 27 28 1 27 28 29 30 31 1 2 Mar 2 3 4 5 6 7 8 3 4 5 6 7 8 9 9 10 11 12 13 14 15 10 11 12 13 14 15 16 16 17 18 19 20 21 22 17 18 19 20 21 22 23 23 24 25 26 27 28 29 24 25 26 27 28 29 30 30 31 Figure 3. Approximate annual data collection schedule for trapping (�) and telemetry (�). Dates and weeks will change depending on calendar year and pay schedule. During telemetry work, the 6-person crew will be divided into 2 teams, only one of which will be working at any given time. Monthly locations on radio-collared hares will also be collected in the interim between the intensive sampling periods indicated here. Vegetation sampling at each stand will follow protocols established through previous snowshoe hare and lynx work in Colorado (Zahratka 2004, T. Shenk, Colorado Division of Wildlife, personal communication). Specifically, on each of the 9 live-trapping grids, I will lay out 5 × 5 grids (3-m spacing) of vegetation sampling points centered on 15 of the 84 trap locations (Figure 4). At each of the 25 vegetation sampling points, I will record: 1) distance to the nearest woody stem 1.0−7.0 cm, 7.1−10.0 cm, and >10.0 cm in diameter at heights of 0.1 m and 1.0 m above the ground (to capture both summer [0.1 m] and winter [1.0 m] stem density; Barbour et al. 1999), 2) horizontal cover in 0.5-m increments above the ground up to 2 m (Nudds 1977), and 3) canopy cover [present or absent] using a densitometer. Additionally, at the center of all 15 vegetation sampling grid points (i.e., at the trap location), I will measure basal area using an angle gauge. These measurements will be gathered once at the start of the project, unless conditions change due to disturbance such as fire. Temperature will be monitored hourly at each grid during the 6-week intensive sampling periods using data loggers. During winter sampling periods, snow depth measurements will be recorded daily at the same 15 trap locations used to quantify the vegetative attributes of that stand. 35 �0 0 0 0 0 0 0 0 • 0 • 0 • 0 0 0 0 0 0 0 0 0 • 0 • 0 • 0 0 0 0 0 0 0 l som □ □ □ □ □ □ □ □ □ □ □ □ ■ □ □ □ □ □ □ □ □ □ □ □ □ l 3m 0 • 0 • 0 0 0 0 0 0 0 0 0 • 0 • 0 • 0 0 0 0 0 0 0 0 0 • 0 • 0 • 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Figure 4. 15 trap locations (•) on 7 × 12 trapping grid where vegetation will be sampled by measuring stem density horizontal cover, and canopy cover at the 25 points on each 5 × 5 subgrid (inset). In addition, basal area will be measured at the trap location (�) on which each of the 15 subgrids are centered. Data Analysis Density.--I will assume that hare populations are demographically and geographically closed during the short 5-day mark-recapture sampling periods. To obtain a density estimate for each grid, I will use the Huggins closed capture model (Huggins 1989, 1991) in Program MARK (White and Burnham 1999) with some modifications. The basic Huggins estimator (no individual covariates) is based on the fact that if pj is the probability that a hare in the population will be captured (and marked) for the first time on trapping occasion j, then p * = 1 − (1 − p1 )...(1 − p5 ) is the probability that an individual is captured at least once during a 5-day trapping period (i.e., j = 1,…,5). Accordingly, the basic Huggins estimator for population size, N̂ , is Nˆ = M t +1 / p* where M t +1 is the total number of hares captured. The estimator can be re-written to allow each of the M t +1 individuals captured to have their own p*. In that case, Nˆ = M t +1 ∑1 / p . Presumably hares that reside near the edge of a grid encounter fewer traps and * i i =1 are less likely to be captured than hares residing near the center of a grid. To account for this, I will take advantage of the Huggins model with individual covariates to model p* by using the logit link function of program MARK to model pi* as a function of di, where di is distance from the edge of the grid for hare i based on mean capture coordinates. A naïve density estimate for each grid would then be Dˆ = Nˆ / A where A is the area of the grid. However, this gives full credit to all hares, even those whose home range only partially overlaps the grid, which results in a density estimate that is biased high. To correct for this bias, I will determine the proportion, ( ~ pk ), of telemetry locations for each of the k = 1,…,10 radiocollared hares that fall within the “naïve grid area.” By incorporating data from multiple grids, a logistic regression model will be developed to estimate p% i for all M t +1 animals captured on a grid based on 36 �distance from the edge of the grid for hare i (di). Replacing the numerator (i.e., 1) in the Huggins ⎛ M t +1 ⎞ ~ p / p ⎟ A. ⎜∑ ⎟ estimator with ( p% i ), gives a density estimate, Dˆ = ⎜ ⎝ i =1 i * i ⎠ The above-stated approach assumes that radio-collared hares neither gravitate toward nor avoid the former grid area after the 5 days of trapping, 10–20 locations per hare is enough to provide a reasonable representation of the proportion of time they spend on the grid, and their use of the grid area is representative of other hares that were captured but not collared (i.e., that the logistic regression model of p% i is a useful model). I contend that this type of estimate from grid-based trapping can be construed as a relatively unbiased estimate of density. Using these point estimates and their associated confidence intervals, I will compare hare density among seasons, years, and stand types. I will also compare these “true” density estimates to those that would have been obtained using other available methods such as ½ mean maximum distance moved (Wilson and Anderson 1985, Williams et al. 2002:314-315), full mean maximum distance moved (Parmenter et al. 2003), ½ trap interval (Parmenter et al. 2003), “nested grids” (White et al. 1982:120-131), and Program DENSITY (Efford et al. 2004). I Demography.--I will analyze mark-recapture data using Pradel temporal symmetry models (Pradel 1996, Nichols and Hines 2002) in a robust design framework (Williams et al. 2002:523-554), which will be available in Program MARK by summer 2006. Thus, I will treat summer and winter sampling occasions as primary periods, and the 5-day trapping sessions within each as secondary periods. The Pradel temporal symmetry models employ both forward and reverse-time evaluation of capture histories to provide estimates of apparent survival ( φ̂ ) and seniority ( γ̂ ). Apparent survival, φi, is the probability that a marked animal alive and in the population at time i survives and is in the population at time i + 1. The seniority parameter, γi , is the reverse-time analogue of survival. Reading backward through a capture history, it is the probability that a marked animal alive and in the population at time i was alive and in the sampled population at time i − 1. If N is the number of animals present in the population, N i φi ≈ N i +1γ i +1 and N i +1 / N i = φi / γ i +1 = λ i . Also, if fi is recruitment rate, or the number of recruits at time i + 1 per animal present at time i, then N i +1 = N i φi + N i f i . Rearranging and substituting into the previous equation gives f i = φi (1/ γ i − 1) . Thus, using Pradel models, one can estimate recruitment and finite population growth rate in addition to survival (Pradel 1996, Nichols and Hines 2002). I will use Akaike’s Information Criterion corrected for small sample size (AICc; Burnham and Anderson 1998) to determine whether models with time-dependent parameters or constant parameters are best supported by the data. I will derive estimates of the above-mentioned parameters from the best model or from model averaging. I anticipate pooling capture data across sites to obtain φˆ i , λˆ i , and fˆi for each stand type for each interval between primary sampling periods (5 estimates of each). I also anticipate simply estimating these parameters for “generic hares”, treating both juveniles and adults as a single group or age class. Given that juveniles are morphometrically indistinguishable from adults by their first fall of life (K. Hodges, University of British Columbia; P. Griffin, University of Montana, personal communication), adult and juvenile survival rates are similar (Griffin 2004), and there is little evidence for age-specific differences in pregnancy rates or litter size (Dolbeer 1972), this approach seems justified. However, if I happen to capture sufficient numbers of juveniles and adults, the design I have laid out here allows for treating the age classes separately. This, in turn, may permit me to decompose the contribution that fi makes to λi into the portion of that contribution due to on-site reproduction and that due to immigration (Nichols et al. 2000). Similarly, it may also be possible using my telemetry data to decompose apparent survival, φi , into emigration and mortality. Such fortuitous situations would facilitate the identification of source and sink habitats if they exist. 37 �Seasonal Movements.--I will assess whether snowshoe hares seasonally shift their home ranges using the multi-response permutation procedure (MRPP; Zimmerman et al. 1985, White and Garrott 1990:134-135). Under this approach, telemetry locations are grouped by season (summer and winter), and an MRPP statistic is calculated as the weighted average of the distance between all possible pairs of locations within groups compared to the average distance between all possible pairs ignoring groups. The null hypothesis is that the distribution of locations is the same for both groups (seasons). Sufficiently small values of the test statistic suggest that within group distances are smaller than distances measured ignoring groups, which is evidence against the null in favor of a group (seasonal) effect. P-values are obtained by calculating the percentile of the observed test statistic relative to all possible test statistics that could be computed by re-arranging the data into all possible groups of 2. The MRPP procedure is sensitive and can detect even small changes in use of an area (White and Garrott 1990:136). I propose a priori that changes in proportional use of deciduous habitats <0.10 in magnitude are unlikely to be biologically significant. Vegetation.--I will calculate mean stem density, horizontal cover, canopy cover, and basal area for each season−stand type as well as temperature, precipitation, snow depth information, and landscape metrics. These will be entered into the MARK design matrix as covariates to population size (~density) and survival in a random effects analysis. As such, I will be able to quantify the amount of variation in population size or survival that is due to differences in vegetation, landscape, or weather relative to the amount left to other causes. Sample size.--I conducted power analyses to determine the probability of discerning meaningful differences in density and survival for hares occupying different stand types. For density, I postulated that foraging lynx likely do not discriminate among stands that differ by only a few hares. However, it seems probable that if hare density in one stand is twice that of another, a lynx would choose the former given the opportunity. Thus, I conducted power calculations to determine the probability of distinguishing differences in densities between 2 stand types in which one had twice the density of hares as the second. Specifically, using the Huggins closed capture model (Huggins 1989, Huggins 1991) in Program MARK, I specified the number of hares (N) present in each of 2 groups (i.e., 2 stand types), allowed capture (p) and recapture (c) probabilities to vary with time but constrained them to be equal and the same for each group, then simulated this scenario 1000 times for a range of realistic capture probabilities. For each simulation I calculated a 95% confidence interval for the mean difference in N̂ between the 2 groups and determined the proportion of all simulations in which this confidence interval did not include zero. This proportion is the power, or probability of discerning a difference between the 2 groups when one actually exists. I compared 2-fold differences in density at the low (5 vs. 10 hares/grid) and high (15 vs. 30 hares/grid) end of the range of hare numbers and I expect to observe (Zahratka 2004). I also simulated the power to detect differences between 17 and 39 hares/grid, corresponding to recently published cut-points for low and high hare densities in the context of lynx conservation (Mills et al. 2005). Given capture/recapture probabilities I observed during winter 2005 (approximately 0.35–0.45), I expect to have reasonable power to detect 2-fold differences in density even if I encounter relatively few hares per grid (Figure 5). 38 �% Non-overlapping 95% CIs Density Power Analysis 100 90 80 70 60 50 40 30 20 10 0 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.50 0.55 0.60 0.45 capture/recapture probability ---+--- N=5 vs. N=10 N=15 vs. N=30 ___,,_ ---I N=17 vs. N=39 Figure 5. Power for distinguishing differences in snowshoe hare density between 2 habitat types when a difference actually exists. Gray area indicates the capture probability realized by the 3rd day of trapping during a pilot study in winter 2005. N indicates number of hares per grid, a range of roughly 0.1 (N = 5) to 0.7 hares/ha (N = 39). I conducted power analyses for survival in a similar manner using the Huggins estimator (Huggins 1989, Huggins 1991) in a robust design framework (Williams et al. 2002:524-556). For this analysis, I specified 3 primary periods (e.g., 3 years) with 5 secondary occasions for each. I established either 30 or 45 hares in each of 2 groups (i.e., pooled an expected 10-15 hares/grid across the 3 grids in a given habitat type), specified a different survival rate for each, and allowed p and c to vary with time but constrained them to be equal and the same for each group as before. I then specified a general model that assumed survival rates varied among groups and a second, reduced model that assumed survival rates were the same for each group. After 1000 simulations under a given scenario of hare numbers, capture probabilities, and survival rates, I conducted a likelihood ratio test between each pair of general and reduced models. As before, I used the proportion of significant tests as an estimate of power to detect differences in survival. I compared survival rates of 0.4 vs. 0.5, 0.3 vs. 0.5, and 0.2 vs. 0.5. These rates span the range of annual hare survival rates reported in the literature (Dolbeer 1972, Dolbeer and Clark 1975, Griffin 2004). Also, because each comparison is anchored at 0.5, these calculations provide a conservative estimate of power due to the nature of binomial probabilities. That is, I would be more likely to distinguish the difference between 0.1 and 0.2 than between 0.4 and 0.5 even though the difference in both cases is 0.1 because the sampling variance of the estimate for the same sample size is maximal at 0.5 and declines to 0 for survival rates of 0 or 1. Results indicate that I have ≥80% chance of discerning real differences in survival of ≥0.3 (Figure 6), but only 40-65% chance (depending on number of hares captured) of detecting a difference of 0.2, and very little chance of detecting differences smaller than 0.2. However, I plan to combine my telemetry data with my trapping data in the MARK Robust design model using separate groups for each data type. This should enhance my precision and power, thus making the prospect of detecting differences as small as 0.2 a possibility. 39 �Survival Power Analysis (N = 45) 100 Capture/Recapture Probability I-+- 0.2 vs. 0.5 --- 0.3 vs. 0.5 -------- _._..,,.- 0.60 0.10 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0 V / 0.55 10 0 ✓ 0.50 - 20 / - 0.45 ------------ / 30 ---- / 0.40 / 40 / / 0.35 20 L---"" 10 .---------- 50 / - 0.30 / - _,....... 70 60 / 0.25 V /' / / 90 80 0.20 70 60 50 40 30 ---- ____.,- 80 0.15 100 90 % Significant LR Tests % Significant LR Tests Survival Power Analysis (N = 30) Capture/Recapture Probability I-+- 0.2 vs. 0.5 --- 0.3 vs. 0.5 0.4 vs. 0.5 0.4 vs. 0.5 Figure 6. Power, or probability of distinguishing differences in snowshoe hare survival between 2 habitat types when differences actually exist. N = 30 (left) and N = 45 (right) correspond to reasonable estimates of the number of hares I expect to capture in each habitat type. Gray area indicates the capture probability realized by the 3rd day of trapping during a pilot study in winter 2005. To complete a power analysis for λ̂ requires running simulations of Pradel models in a robust design framework. This capability is not yet available in Program MARK, so such an analysis has not been completed. Sampling 15 vegetation plots per trapping grid provided reasonably precise characterizations of similar stands in similar locations during a previous study (Zahratka 2004). I trust this level of sampling will be adequate for the present study as well. If not, more plots can be established at a later date given that vegetative characteristics are unlikely to change appreciably over a few years. Project Schedule I will begin the first 9-week data collection period in mid July 2006. The first winter sampling period will begin in February 2007. Intensive sampling will occur across a total of 3 summer and 3 winter periods, with monthly telemetry work interspersed between the main sampling periods. All fieldwork will terminate with the winter 2009 sampling period. Analysis, write-up, and submission to journal outlets will occur during summer and Fall 2009. I plan to graduate during spring semester 2010. Personnel Jacob S. Ivan, Ph. D. student, Colorado State University will be the primary investigator responsible for the design and conduct of the study. Tanya M. Shenk, Mammals Research, Colorado Division of Wildlife, and Gary C. White, Professor, Colorado State University will serve as primary advisors. Also, as most lynx/hare habitat occurs on United States Forest Service (USFS) land, this project will require cooperation and coordination with USFS biologists and district rangers for permission and possibly logistical support (housing, campsites, trucks). As presented here, this project will require an estimated minimum of 3,600 person-hours/year (5 technicians, 720 hours) in technician labor to complete the intensive 9-week sampling periods as well as 360 person-hours/year of technician labor to run the monthly telemetry operation. Thus, completion of the 3-year project will require an estimated minimum of 11,880 person-hours in addition to time spent by the primary investigator, advisors, and cooperators. 40 �Estimated Annual Cost FY06-07 FY07-08 FY08-09 TFTE (5 techs, 360, $11.13/hr, 11.16% overhead)* $ 22,270 $ 22,830 $ 23,410 TFTE (1 tech, 360 hours, $11.13/hr, 11.16% overhead)** $ $ 4,565 $ 4,679 Personnel 4,454 Operating PURCHSERV (Ph.D. Stipend, tuition, minimal supplies)*** $ 27,500 $ 27,500 $ 27,500 SUPPLIES (bait, snowmobile repairs, handling supplies, etc.) $ $ 4,000 $ 4,000 EQUIPMENT (radio collars) $ 11,500 $ 11,500 $ 11,500 INSTTRAV $ 1,500 $ 1,500 $ 1,500 VEHICLE LEASE/MILEAGE (3 vehicles, 5 months/year)** $ 5,328 $ 5,328 $ 5,328 4,000 Travel TOTAL COST $76,552 $77,223 $77,917 TOTAL COST TO SSH BUDGET $43,724 $44,395 $45,089 *Assumes 2.5% cost-of-living wage increase/year **Telemetry work during interim between sampling periods ***Will be charged to budget centers other than lynx/snowshoe hare EXPECTED RESULTS/BENEFITS 1) Seasonal density estimates and associated variability will help establish where Colorado lies on the continuum of hare densities reported in the literature. Whether densities are relatively high or low, stable or highly variable, or drastically different or roughly equal among cover types could influence future land management decisions as well as decisions regarding the lynx reintroduction process. 2) Combined with Zahratka (2004) and future research, density estimates from this project may elucidate the degree to which hare populations fluctuate or cycle in Colorado, a phenomenon of interest to wildlife ecologists and managers. 3) Comparison of “known” densities to those obtained from other commonly used methods will inform future research and monitoring programs which techniques are likely to produce results that are most consistently in agreement with the intensively derived estimates reported from this project. This knowledge will also enhance interpretation of previously reported hare densities in Colorado and elsewhere. 4) Assessment of density, demographic parameters, and their variability among habitat types will help identify which type(s) consistently support(s) high hare numbers and productivity. The current, conservative approach to lynx/hare conservation is to treat all potential habitat as equally and highly valuable, although this has not been substantiated scientifically, especially in Colorado. This project should determine if the current approach is justified or if there is a disparity in the value of different habitat types relative to lynx-hare conservation. If the latter is true, those charged with managing forests may be allowed more flexibility to accommodate competing resource uses while maintaining lynx/hare habitat. 41 �5) Assessment of density and demographic parameters should help identify the general time period over which succession carries young, regenerating lodgepole pine stands into and then out of service as snowshoe hare habitat. It is apparent that stands in fresh clear cuts and mature lodgepole stands do not provide quality hare habitat (Zahratka 2004). The value of small and medium lodgepole stands to hares has not been quantified in Colorado and is of interest to resource managers. 6) Knowledge regarding the presence or absence of large-scale seasonal movements, and the extent to which this occurs will inform managers about the value of peripheral vegetation (other than conifer forest, such as riparian willow or aspen), will identify when and for how long peripheral vegetation is likely to be used, and will potentially identify other snowshoe hare management issues that have not received prior consideration. 7) A description and comparison of vegetation and landscape characteristics among the 3 stand types and their relationship to snowshoe hare demography and movement patterns should further aid land managers in creating and maintaining lynx/hare habitat. RELATED FEDERAL PROJECTS Given that the majority of lynx/hare habitat occurs on United States Forest Service land, this project will require cooperation with local ranger districts, regional biologists, and researchers within that agency. As soon as I have completed provisional study site selection, I will contact the appropriate collaborators to obtain permission, appropriate permits, etc. LITERATURE CITED AUBRY, K. B., G. M. KOEHLER, AND J. R. SQUIRES. 2000. Ecology of Canada lynx in southern boreal forests. Pages 373-396 in Ruggiero, L. F., K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, and J. R. Squires, editors. Ecology and conservation of lynx in the United States. Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Collins, Colorado, USA. BARBOUR, M. G., J. H. BURK, W. E. PITTS, F. S. GILLIAM, AND M. W. SCHWARTZ. 1999. Terrestrial plant ecology, Addison Wesley Longman, Inc., Menlo Park, California, USA. BARTMANN, R. M. AND G. BYRNE. 2001. Analysis and critique of the 1998 snowshoe hare pellet survey. Colorado Division of Wildlife 20. BEAUVAIS, G. P. 1997. Mammals in fragmented forests in the Rocky Mountains: community structure, habitat selection, and individual fitness. Dissertation, University of Wyoming, Laramie, Wyoming, USA. BUCKLAND, S. T., D. R. ANDERSON, K. P. BURNHAM, J. L. LAAKE, D. L. BORCHERS, AND L. THOMAS. 2001. Introduction to distance sampling: estimating abundance of biological populations, Oxford University Press, Inc., New York, New York, USA. BULL, E. L., T. W. HEATER, A. A. CLARK, J. F. SHEPHERD, AND A. K. BLUMTON. 2005. Influence of precommercial thinning on snowshoe hares. U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station, General Technical Report PNW-RP-562. DOLBEER, R. A. AND W. R. CLARK. 1975. Population ecology of snowshoe hares in the central Rocky Mountains. Journal of Wildlife Management 39:535-549. DOLBEER, R. A. 1972. Population dynamics of snowshoe hare in Colorado. Dissertation, Colorado State University, Fort Collins, Colorado, USA. EFFORD, M. G., D. K. DAWSON, AND C. S. ROBBINS. 2004. DENSITY: software for analysing capturerecapture data from passive detector arrays. Animal Biodiversity and Conservation 27:1-12. ERIKSSON, H. M. 2006. Snowshoe hare densities in post-fire vegetation. Thesis, University of Alaska Fairbanks, Fairbanks, Alaska, USA. FORESMAN, K. R. AND D. E. PEARSON. 1999. Activity patterns of American martens, Martes americana, snowshoe hares, Lepus americanus, and red squirrels, Tamiasciurus hudsonicus, in westcentral 42 �Montana. The Canadian Field-Naturalist 113:386-389. GRIFFIN, P. C. 2004. Landscape ecology of snowshoe hares in Montana. Dissertation, University of Montana, Missoula, Montana, USA. HODGES, K. E. 2000. Ecology of snowshoe hares in southern boreal and montane forests. Pages 163-206 in Ruggiero, L. F., K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, and J. R. Squires, editors. Ecology and conservation of lynx in the United States. University Press of Colorado, Boulder, Colorado, USA. HODGES, K. E., C. J. KREBS, D. S. HIK, C. I. STEFAN, E. A. GILLIS, AND C. E. DOYLE. 2001. Snowshoe hare demography. Pages 141-178 in Krebs, C. J., S. Boutin, and R. Boonstra, editors. Ecosystem dynamics of the boreal forest: the Kluane project. Oxford University Press, New York, New York, USA. HOMYACK, J. A., D. J. HARRISON, AND W. B. KROHN. 2003. Effects of precommercial thinning on select wildlife species in northern Maine, with special emphasis on snowshoe hare. Maine Cooperative Fish and Wildlife Research Unit, Orono, Maine, USA. HOMYACK, J. A., D. J. HARRISON, J. A. LITVAITIS, AND W. B. KROHN. 2006. Quantifying densities of snowshoe hares in Maine using pellet plots. Wildlife Society Bulletin 34:74-80. HUGGINS, R. M. 1989. On the statistical analysis of capture-recapture experiments. Biometrika 76:133140. HUGGINS, R. M. 1991. Some practical aspects of a conditional likelihood approach to capture experiments. Biometrics 47:725-732. KEITH, L. B. 1964. Daily activity pattern of snowshoe hares. Journal of Mammalogy 45:626-627. KOEHLER, G. M. 1990. Snowshoe hare, Lepus americanus, use of forest successional stages and population changes during 1985-1989 in North-central Washington. Canadian Field-Naturalist 105:291-293. KOEHLER, G. M. 1990. Population and habitat characteristics of lynx and snowshoe hares in north central Washington. Canadian Journal of Zoology 68:845-851. KREBS, C. J., B. SCOTT GILBERT, S. BOUTIN, AND E. AL. R. BOONSTRA. 1987. Estimation of snowshoe hare population density from turd transects. Canadian Journal of Zoology 65:565-567. KREBS, C. J., R. BOONSTRA, V. NAMS, M. O'DONOGHUE, K. E. HODGES, AND S. BOUTIN. 2001. Estimating snowshoe hare population density from pellet plots: a further evaluation. Canadian Journal of Zoology 79:1-4. LITVAITIS, J. A., J. A. SHERBURNE, AND J. A. BISSONETTE. 1985. Influence of understory characteristics on snowshoe hare habitat use and density. Journal of Wildlife Management 49:866-873. MARQUES, F. F. C., S. T. BUCKLAND, D. GOFFIN, C. E. DIXON, D. L. BORCHERS, B. A. MAYLE, AND A. J. PEACE. 2001. Estimating deer abundance from line transect surveys of dung: sika deer in southern Scotland. Journal of Applied Ecology 38:349-363. MCKELVEY, K. S., G. W. MCDANIEL, L. S. MILLS, AND P. C. GRIFFIN. 2002. Effects of plot size and shape on pellet density estimates for snowshoe hares. Wildlife Society Bulletin 30:751-755. MECH, L. D., K. L. HEEZEN, AND D. B. SINIFF. 1966. Onset and cessation of activity in cottontail rabbit and snowshoe hares in relation to sunset and sunrise. Animal Behaviour 14:410-413. MILLER, M. A. 2005. Snowshoe hare habitat relationships in northwest Colorado. Thesis, Colorado State University, Fort Collins, Colorado, USA. MILLS, L. S., P. C. GRIFFIN, K. E. HODGES, K. MCKELVEY, L. RUGGIERO, AND T. ULIZIO. 2005. Pellet count indices compared to mark-recapture estimates for evaluating snowshoe hare density. Journal of Wildlife Management 69:1053-1062. MURRAY, D., E. ELLSWORTH, AND A. ZACK. 2005. Assessment of potential bias with snowshoe hare fecal pellet-plot counts. Journal of Wildlife Management 69:385-395. MURRAY, D. L., J. D. ROTH, E. ELLSWORTH, A. J. WIRSING, AND T. D. STEURY. 2002. Estimating lowdensity snowshoe hare populations using fecal pellet counts. Canadian Journal of Zoology 80:771-781. NICHOLS, J. D. AND J. E. HINES. 2002. Approaches for the direct estimation of lambda, and demographic 43 �contributions to lambda, using capture-recapture data. Journal of Applied Statistics 29:539-568. NICHOLS, J. D., J. E. HINES, J. D. LEBRETON, AND R. PRADEL. 2000. Estimation of contributions to population growth: a reverse-time capture-recapture approach. Ecology 81:3362-3376. NUDDS, T. D. 1977. Quantifying the vegetative structure of wildlife cover. Wildlife Society Bulletin 5:113-117. OTIS, D. L., K. P. BURNHAM, G. C. WHITE, AND D. R. ANDERSON. 1978. Statisical inference from capture data on closed animal populations. Wildlife Monographs 62: PARMENTER, R. R., T. L. YATES, D. R. ANDERSON, K. P. BURNHAM, J. L. DUNNUM, A. B. FRANKLIN, M. T. FRIGGENS, B. C. LUBOW, M. MILLER, G. S. OLSON, C. A. PARMENTER, J. POLLARD, E. REXSTAD, T. SHENK, T. R. STANLEY, AND G. C. WHITE. 2003. Small-mammal density estimation: a field comparison of grid-based vs. web-based density estimators. Ecological Monographs 73:126. PRADEL, R. 1996. Utilization of capture-mark-recapture for the study of recruitment and population growth rate. Biometrics 52:703-709. PRUGH, L. R. AND C. J. KREBS. 2004. Snowshoe hare pellet-decay rates and aging in different habitats. Wildlife Society Bulletin 32:386-393. RUEDIGER, B., J. CLAAR, S. GNIADEK, B. HOLT, LEWIS LYLE, S. MIGHTON, B. NANEY, G. PATTON , T. RINALDI, J. TRICK, A. VENDEHEY, F. WAHL, N. WARREN, D. WENGER, AND A. WILLIAMSON. 2000. Canada lynx conservation assessment and strategy. U.S. Department of Agriculture, Forest Service, U.S. Department of Interior, Fish and Wildlife Service, Bureau of Land Management, National Park Service R1-00-53 U.S. Department of Agriculture, Forest Service, U.S. Department of Interior, Fish and Wildlife Service, Bureau of Land Management, National Park Service, Missoula, Montana, USA. RUGGIERO, L. F., K. B. AUBRY, S. W. BUSKIRK, G. M. KOEHLER, C. J. KREBS, K. S. MCKELVEY, AND J. R. SQUIRES. 2000. The scientific basis for lynx conservation: qualified insights. Pages 443-454 in Ruggiero, L. F., K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, and J. R. Squires, editors. Ecology and conservation of lynx in the United States. Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Collins, Colorado, USA. SAMUEL, M. D. AND M. R. FULLER. 1996. Wildlife radiotelemetry. Pages 370-418 in Bookhout, T. A., editors. Research and Management Techniques for Wildlife and Habitats. Allen Press, Inc., Lawrence, Kansas, USA. SHENK, T. M. 2005. General locations of lynx (Lynx canadensis) reintroduced to southwestern Colorado from February 4, 1999 through February 1, 2005. Colorado Division of Wildlife Colorado Division of Wildlife, Fort Collins, Colorado, USA. SULLIVAN, T. P. AND D. S. SULLIVAN. 1988. Influence of stand thinning on snowshoe hare population dynamics and feeding damage in lodgepole pine forest. Journal of Applied Ecology 25:791-805. VAN HORNE, B. 1983. Density as a misleading indicator of habitat quality. Journal of Wildlife Management 47:893-901. WHITE, G. C., D. R. ANDERSON, K. P. BURNHAM, AND D. L. OTIS. 1982. Capture-recapture and removal methods for sampling closed populations. Los Alamos National Laboratory, Los Alamos, New Mexico, USA. WHITE, G. C. AND R. A. GARROTT. 1990. Analysis of wildlife radio-tracking data, Academic Press, San Diego, California, USA. WILLIAMS, B. K., J. D. NICHOLS, AND M. J. CONROY. 2002. Analysis and management of animal populations, Academic Press, San Diego, California, USA. WILSON, K. R. AND D. R. ANDERSON. 1985. Evaluation of two density estimators of small mammal population size. Journal of Mammalogy 66:13-21. WOLFE, M. L., N. V. DEBYLE, C. S. WINCHELL, AND T. R. MCCABE. 1982. Snowshoe hare cover relationships in northern Utah. Journal of Wildlife Management 46:662-670. WOLFF, J. O. 1980. The role of habitat patchiness in the population dynamics of snowshoe hares. Ecological Monographs 50:111-130. 44 �ZAHRATKA, J. L. 2004. The population and habitat ecology of snowshoe hares (Lepus americanus) in the southern Rocky Mountains. Thesis, The University of Wyoming, Laramie, Wyoming, USA. ZIMMERMAN, G. M., H. GOETZ, AND P. W. JR. MIELKE. 1985. Use of an improved statistical method for group comparisons to study effects of prairie fire. Ecology 66:606-611. 45 � https://cpw.cvlcollections.org/files/original/939c9c7f86d7cd43cc99d606aa5c6865.pdf a6b5b0747f7dd2d2165a1723826bb852 PDF Text Text Colorado Division of Wildlife July 2006 - June 2007 WILDLIFE RESEARCH REPORT State of Cost Center Work Package Task No. Colorado 3430 0670 1 Federal Aid Project: N/A : Division of Wildlife : Mammals Research : Lynx Conservation : Post-Release Monitoring of Lynx : Reintroduced to Colorado : Period Covered: July 1, 2006 - June 30, 2007 Author: T. M. Shenk Personnel: L. Baeten, B. Diamond, R. Dickman, D. Freddy, L. Gepfert, J. Ivan, R. Kahn, A. Keith, G. Merrill, M. Schuette, B. Smith, T. Spraker, S. Wait, S. Waters, L. Wolfe, D. Younkin All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT In an effort to establish a viable population of lynx (Lynx canadensis) in Colorado, the Colorado Division of Wildlife (CDOW) initiated a reintroduction effort in 1997 with the first lynx released in February 1999. From 1999-2007, 218 lynx were released in Colorado. We documented survival, movement patterns, reproduction, and landscape habitat-use through aerial (n = 9496) and satellite (n = 23,791) tracking. Most lynx remained near the core release area in southwestern Colorado. From 1999June 2007, there were 98 mortalities of released adult lynx. Approximately 30.6% were human-induced which were attributed to collisions with vehicles or gunshot. Starvation and disease/illness accounted for 19.4% of the deaths while 35.7% of the deaths were from unknown causes. Reproductive females had the smallest 90% utilization distribution home ranges ( x = 75.2 km2, SE = 15.9 km2 ), followed by attending males ( x = 102.5 km2, SE = 39.7 km2) and non-reproductive animals ( x = 653.8 km2, SE = 145.4 km2). Reproduction was first documented in 2003 with subsequent successful reproduction in 2004, 2005 and 2006. No dens were documented in 2007. From snow-tracking, the primary winter prey species (n = 506 kills) were snowshoe hare (Lepus americanus, annual x = 74.9%, SE = 4.6, n = 9) and red squirrel (Tamiasciurus hudsonicus, annual x = 16.5%, SE = 4.1, n = 9); other mammals and birds formed a minor part of the winter diet. Lynx use-density surfaces were generated to illustrate relative use of areas throughout Colorado and areas of use in New Mexico, Utah and Wyoming. Within the areas of high use in southwestern Colorado, site-scale habitat use, documented through snow-tracking, supports mature Engelmann spruce (Picea engelmannii)-subalpine fir (Abies lasiocarpa) forest stands with 42-65% canopy cover and 15-20% conifer understory cover as the most commonly used areas in southwestern Colorado. Little difference in aspect (slight preference for north-facing slopes), slope ( x = 15.7°) or elevation ( x = 3173 m) were detected for long beds, travel and kill sites (n = 1841). Den sites (n = 37) 1 �however, were located at higher elevations ( x = 3354 m, SE = 31 m) on steeper ( x = 30°, SE = 2°) and more commonly north-facing slopes with a dense understory of coarse woody debris. The first year of a study to evaluate snowshoe hare densities, demography and seasonal movement patterns among small and medium tree-sized lodgepole pine stands and mature spruce/fir stands was completed in 2006-2007 and will continue through 2009 (see Appendix I of this report). Results to date have demonstrated that CDOW has developed lynx release protocols that ensure high initial post-release survival followed by high long-term survival, site fidelity, reproduction and recruitment of Colorado-born lynx into the Colorado breeding population. What is yet to be demonstrated is whether Colorado can support sufficient recruitment to offset annual mortality for a viable lynx population over time. Monitoring continues in an effort to document such viability. 2 �WILDLIFE RESEARCH REPORT POST RELEASE MONITORING OF LYNX (LYNX CANADENSIS) REINTRODUCED TO COLORADO TANYA M. SHENK P. N. OBJECTIVE The initial post-release monitoring of Canada lynx (Lynx canadensis) reintroduced into Colorado will emphasize 5 primary objectives: 1. Assess and modify release protocols to ensure the highest probability of survival for each lynx released. 2. Obtain regular locations of released lynx to describe general movement patterns and habitats used by lynx. 3. Determine causes of mortality in reintroduced lynx. 4. Estimate survival of lynx reintroduced to Colorado. 5. Estimate reproduction of lynx reintroduced to Colorado. Three additional objectives will be emphasized after lynx display site fidelity to an area: 6. Refine descriptions of habitats used by reintroduced lynx. 7. Refine descriptions of daily and overall movement patterns of reintroduced lynx. 8. Describe hunting habits and prey of reintroduced lynx. Information gained to achieve these objectives will form a basis for the development of lynx conservation strategies in the southern Rocky Mountains. SEGMENT OBJECTIVES 1. Complete winter 2006-07 field data collection on lynx habitat use at the landscape scale, hunting behavior, diet, mortalities, and movement patterns. 2. Complete winter 2006-07 lynx trapping field season to collar Colorado born lynx and re-collar adult lynx. 3. Complete spring 2007 field data on lynx reproduction. 4. Summarize and analyze data and publish information as Progress Reports, peer-reviewed manuscripts for appropriate scientific journals, or CDOW technical publications. 5. Complete the first year of field work to evaluate snowshoe hare (Lepus americanus) densities, demography and seasonal movement patterns among small and medium tree-sized lodgepole pine stands and mature spruce/fir stands (see Appendix I). INTRODUCTION The Canada lynx occurs throughout the boreal forests of northern North America. Colorado represents the southern-most historical distribution of lynx, where the species occupied the higher elevation, montane forests in the state. Little was known about the population dynamics or habitat use of this species in their southern distribution. Lynx were extirpated or reduced to a few animals in the state by the late 1970’s due, most likely, to predator control efforts such as poisoning and trapping. Given the isolation of Colorado to the nearest northern populations, the CDOW considered reintroduction as the only option to attempt to reestablish the species in the state. 3 �A reintroduction effort was begun in 1997, with the first lynx released in Colorado in 1999. To date, 218 wild-caught lynx from Alaska and Canada have been released in southwestern Colorado. The goal of the Colorado lynx reintroduction program is to establish a self-sustaining, viable population of lynx in this state. Evaluation of incremental achievements necessary for establishing viable populations is an interim method of assessing if the reintroduction effort is progressing towards success. There are 7 critical criteria for achieving a viable population: 1) development of release protocols that lead to a high initial post-release survival of reintroduced animals, 2) long-term survival of lynx in Colorado, 3) development of site fidelity by the lynx to areas supporting good habitat in densities sufficient to breed, 4) reintroduced lynx must breed, 5) breeding must lead to reproduction of surviving kittens 6) lynx born in Colorado must reach breeding age and reproduce successfully, and 7) recruitment must equal or be greater than mortality over an extended period of time. The post-release monitoring program for the reintroduced lynx has 2 primary goals. The first goal is to determine how many lynx remain in Colorado and their locations relative to each other. Given this information and knowing the sex of each individual, we can assess whether these lynx can form a breeding core from which a viable population might be established. From these data we can also describe general movement patterns and habitat use. The second primary goal of the monitoring program is to estimate survival of the reintroduced lynx and, where possible, determine causes of mortality for reintroduced lynx. Such information will help in assessing and modifying release protocols and management of lynx once they have been released to ensure their highest probability of survival. Additional goals of the post-release monitoring program for lynx reintroduced to the southern Rocky Mountains included refining descriptions of habitat use and movement patterns and describing successful hunting habitat once lynx established home ranges that encompassed their preferred habitat. Specific objectives for the site-scale habitat data collection include: 1) describe and quantify site-scale habitat use by lynx reintroduced to Colorado, 2) compare site-scale habitat use among types of sites (e.g., kills vs. long-duration beds), and 3) compare habitat features at successful and unsuccessful snowshoe hare chases. Documenting reproduction is critical to the success of the program and lynx are monitored intensively to document breeding, births, survival and recruitment of lynx born in Colorado. Site-scale habitat descriptions of den sites are also collected and compared to other sites used by lynx. The program will also investigate the ecology of snowshoe hare in Colorado. A study comparing snowshoe hare densities among mature stands of Engelmann spruce (Picea engelmannii)/subalpine fir (Abies lasiocarpa), lodgepole pine (Pinus contorta) and Ponderosa pine (Pinus ponderosa) was completed in 2004 with highest hare densities found in Engelmann spruce/subalpine fir stands and no hares found in Ponderosa pine stands. A study to evaluate the importance of young, regenerating lodgepole pine and mature Engelmann spruce/subalpine fir stands in Colorado by examining density and demography of snowshoe hares that reside in each was initiated in 2005 and will continue through 2009 (see Appendix I). Lynx is listed as threatened under the Endangered Species Act (ESA) of 1973, as amended (16 U. S. C. 1531 et. seq.)(U. S. Fish and Wildlife Service 2000). Colorado is included in the federal listing as lynx habitat. Thus, an additional objective of the post-release monitoring program is to develop conservation strategies relevant to lynx in Colorado. To develop these conservation strategies, information specific to the ecology of the lynx in its southern Rocky Mountain range, such as habitat use, movement patterns, mortality factors, survival, and reproduction in Colorado is needed. 4 �STUDY AREA Southwestern Colorado is characterized by wide plateaus, river valleys, and rugged mountains that reach elevations over 4200 m. Engelmann spruce-subalpine fir is the most widely distributed coniferous forest type at elevations most typically used by lynx. The Core Release Area is defined as areas bounded by the New Mexico state line to the south, Taylor Mesa to the west and Monarch Pass on the north and east and > 2900 m in elevation (Figure 1). The lynx-established core area is roughly bounded by areas used by lynx in the Taylor Park/Collegiate Peak areas in central Colorado and includes areas of continuous use by lynx, including areas used during breeding and denning (Figure 1). METHODS REINTRODUCTION Effort All lynx releases were conducted under the protocols found to maximize survival (see Shenk 2001). Estimated age, sex and body condition were ascertained and recorded for each lynx prior to release (see Wild 1999). Specific release sites were those used in earlier years of the project and were selected based on land ownership and accessibility during times of release (Byrne 1998). Lynx were transported from the Frisco Creek Wildlife Rehabilitation Center, where they were held from their time of arrival in Colorado, to their release site in individual cages. Release site location was recorded in Universal Transverse Mercator (UTM) coordinates and identification of all lynx released at the same location, on the same day, was recorded. Behavior of the lynx on release and movement away from the release site were documented. Movement, Distribution and Relative Use of Areas by Lynx To monitor lynx movements and thus determine distribution and relative use of areas all released lynx were fitted with radio collars. All lynx released in 1999 were fitted with TelonicsTM radio-collars. All lynx released since 1999, with the exception of 5 males released in spring 2000, were fitted with SirtrackTM dual satellite/VHF radio-collars. These collars have a mortality indicator switch that operated on both the satellite and VHF mode. The satellite component of each collar was programmed to be active for 12 hours per week. The 12-hour active periods for individual collars were staggered throughout the week. Signals from the collars allowed for locations of the animals to be made via Argos, NASA, and NOAA satellites. The location information was processed by ServiceArgos and distributed to the CDOW through e-mail messages. Datasets.-- To determine recent (post-reintroduction) movement and distribution of lynx reintroduced, born or initially trapped in Colorado and relative use of areas by these lynx, regular locations of lynx were collected through a combination of aerial and satellite tracking. Locations were recorded and general habitat descriptions for each aerial location was recorded. The first dataset of lynx locations included all locations obtained from daytime flights conducted with a Cessna 185 or similar aircraft to locate lynx by their VHF collar transmitters (hereafter aerial locations). VHF transmitters have been used on lynx since the first lynx were released in February 1999. The second type of lynx location data was collected via satellite from the satellite collar transmitters placed on the lynx (hereafter satellite locations). Satellite transmitter collars were first used for lynx in April 2000. These satellite collars also contained a VHF transmitter which also allowed locating lynx from the air or ground. All locations were recorded in Universal Transverse Mercator (UTM) coordinates using the CONUS NAD27 datum. Flights to obtain lynx aerial locations were typically conducted on a weekly basis throughout most summer and winter months and twice a week during the den search field season (May 15 – June 30), depending on weather and availability of planes and pilots. Flights were typically concentrated in the high elevation (> 2700 m) southwest quadrant of Colorado which encompasses the core lynx release and 5 �research area (Figure 1). Flights during the den seasons were conducted to obtain locations on all female lynx within the state wearing an active VHF transmitter. VHF transmitters were outfitted with sufficient batteries to last 60 months. The satellite transmitters were designed to provide locations on a weekly basis with sufficient batteries to last for 18 months. Lynx may not be exhibiting typical behavior or habitat use within the first few months after their release in Colorado. Therefore, a subset of each of the aerial and satellite datasets was created that eliminated the first 180 days (approximately 6 months) of locations obtained for each lynx immediately after their initial release. As a result, the truncated aerial location dataset contained lynx locations from September 1999 through March 2007 while the truncated satellite location dataset began October 2000 and extended through March 2007. Accuracy of both aerial and satellite locations varied with the environmental conditions at the time the location was obtained. Accuracy of aerial locations was influenced by weather with accuracy ranging from 50 - 500 meters. Satellite location accuracy was also influenced by atmospheric conditions and position of the satellites. Satellite location accuracy ranged from 150 meters -10 km. Movement and Distribution.-- To document all known lynx locations maps were generated with all aerial and satellite locations displayed. Due to lynx movements outside of Colorado, particularly into the states of New Mexico, Utah and Wyoming we further evaluated lynx use throughout those three states, as well as the data would allow. All individual lynx located at least once in these 3 states (nontruncated datasets) were identified and tallied for each year. To document consistency and known use of these states after the initial effect of being reintroduced was minimized (i.e., 180 days post-release), each individual lynx located at least once in these states from the truncated datasets were identified and tallied. Relative Use.-- To document relative use of areas by lynx, 90% kernel use-density surfaces were calculated for truncated satellite and aerial lynx locations using the ArcGIS Spatial Analyst Kernel Density Tool. Due to differences in data collection frequency and accuracy between datasets, the truncated satellite and truncated aerial data were analyzed separately for generating the lynx use-density surfaces. These use-density surfaces fit a smoothly curved surface over each lynx location. The surface value was highest at the location of the point and diminished with increasing distance from the point. A fixed kernel was used with a smoothing parameter of 5 km, reaching 0 at the search radius distance from the point. Only a circular neighborhood was possible. The volume under the surface equaled the total value for the point. The use-density at each output GIS raster cell was calculated by adding the values of all the kernel surfaces from all the lynx point locations that overlaid each raster cell center. The kernel function was based on the quadratic kernel function described in Silverman (1986, p. 76, equation 4.5). The use-density surfaces were calculated at 100 m resolution. To enhance graphic displays of higher usedensity areas, density values representing single locations were not displayed. Home Range Annual home ranges were calculated as a 95% utilization distribution using a kernel home-range estimator for each lynx we had at least 30 locations for within a year. A year was defined as March 15 – March 14 of the following year. Locations used in the analyses were collected from September 1999 – January 2006 and all locations obtained for an individual during the first six months after its release were eliminated from any home range analyses as it was assumed movements of lynx initially post-release may not be representative of normal habitat use. Locations were obtained either through aerial VHF surveys or locations or the midpoint (ArcView Movement Extension) of all high quality (accuracy rating of 01km) satellite locations obtained within a single 24-hour period. All locations used within a single home range analysis were taken a minimum of 24 hours apart. 6 �Home range estimates were classified as being for a reproductive or non-reproductive animal. A reproductive female was defined as one that had kittens with her; a reproductive male was defined as a male whose movement patterns overlapped that of a reproductive female. If a litter was lost within the defined year a home range described for a reproductive animal were estimated using only locations obtained while the kittens were still with the female. Survival Survival was estimated as ragged telemetry data using the nest survival models in Program MARK (White and Burnham 1999). Mortality Factors When a mortality signal (75 beats per minute [bpm] vs. 50 bpm for the Telonics™ VHF transmitters, 20 bpm vs. 40 bpm for the Sirtrack™ VHF transmitters, 0 activity for Sirtrack™ PTT) was heard during either satellite, aerial or ground surveys, the location (UTM coordinates) was recorded. Ground crews then located and retrieved the carcass as soon as possible. The immediate area was searched for evidence of other predators and the carcass photographed in place before removal. Additionally, the mortality site was described and habitat associations and exact location were recorded. Any scat found near the dead lynx that appeared to be from the lynx was collected. All carcasses were transported to the Colorado State University Veterinary Teaching Hospital (CSUVTH) for a post mortem exam to 1) determine the cause of death and document with evidence, 2) collect samples for a variety of research projects, and 3) archive samples for future reference (research or forensic). The gross necropsy and histology were performed by, or under the lead and direct supervision of a board certified veterinary pathologist. At least one research personnel from the CDOW involved with the lynx program was also present. The protocol followed standard procedures used for thorough post-mortem examination and sample collection for histopathology and diagnostic testing (see Shenk 1999 for details). Some additional data/samples were routinely collected for research, forensics, and archiving. Other data/samples were collected based on the circumstances of the death (e.g., photographs, video, radiographs, bullet recovery, samples for toxicology or other diagnostic tests, etc.). From 1999–2004 the CDOW retained all samples and carcass remains with the exception of tissues in formalin for histopathology, brain for rabies exam, feces for parasitology, external parasites for ID, and other diagnostic samples. Since 2005 carcasses are disposed of at the CSUVTH with the exception of the lower canine, fecal samples, stomach content samples and tissue or bone marrow samples to be delivered by CDOW to the Center for Disease control for plague testing. The lower canine, from all carcasses, is sent to Matson Labs (Missoula, Montana) for aging and the fecal and stomach content samples are evaluated for diet. Reproduction Females were monitored for proximity to males during each breeding season. We defined a possible mating pair as any male and female documented within at least 1 km of each other in breeding season through either flight data or snow-tracking data. Females were then monitored for site fidelity to a given area during each denning period of May and June. Each female that exhibited stationary movement patterns in May or June were closely monitored to locate possible dens. Dens were found when field crews walked in on females that exhibited virtually no movement for at least 10 days from both aerial and ground telemetry. Kittens found at den sites were weighed, sexed and photographed. Each kitten was uniquely marked by inserting a sterile passive integrated transponder (PIT, Biomark, Inc., Boise, Idaho, USA) tag subcutaneously between the shoulder blades. Time spent at the den was minimized to ensure the least 7 �amount of disturbance to the female and the kittens. Weight, PIT-tag number, sex and any distinguishing characteristics of each kitten was also recorded. Beginning in 2005, blood and saliva samples were collected and archived for genetic identification. During the den site visits, den site location was recorded as UTM coordinates. General vegetation characteristics, elevation, weather, field personnel, time at the den, and behavioral responses of the kittens and female were also recorded. Once the females moved the kittens from the natal den area, den sites were visited again and site-specific habitat data were collected (see Habitat Use section below). Captures Captures were attempted for either lynx that were in poor body condition or lynx that needed to have their radio-collars replaced due to failed or failing batteries or to radio-collar kittens born in Colorado once they reached at least 10-months of age when they were nearly adult size. Methods of recapture included 1) trapping using a Tomahawk™ live trap baited with a rabbit and visual and scent lures, 2) calling in and darting lynx using a Dan-Inject CO2 rifle, 3) custom box-traps modified from those designed by other lynx researchers (Kolbe et al. 2003) and 4) hounds trained to pursue felids were also used to tree lynx and then the lynx was darted while treed. Lynx were immobilized either with Telazol (3 mg/kg; modified from Poole et al. 1993 as recommended by M. Wild, DVM) or medetomidine (0.09mg/kg) and ketamine (3 mg/kg; as recommended by L. Wolfe, DVM)) administered intramuscularly (IM) with either an extendible pole-syringe or a pressurized syringe-dart fired from a Dan-Inject air rifle. Immobilized lynx were monitored continuously for decreased respiration or hypothermia. If a lynx exhibited decreased respiration 2mg/kg of Dopram was administered under the tongue; if respiration was severely decreased, the animal was ventilated with a resuscitation bag. If medetomidine/ketamine were the immobilization drugs, the antagonist Atipamezole hydrochloride (Antisedan) was administered. Hypothermic (body temperature < 95o F) animals were warmed with hand warmers and blankets. While immobilized, lynx were fitted with replacement SirtrackTM VHF/satellite collar and blood and hair samples were collected. Once an animal was processed, recovery was expedited by injecting the equivalent amount of the antagonist Antisedan IM as the amount of medetomidine given, if medetomodine/ketemine was used for immobilization. Lynx were then monitored while confined in the box-trap until they were sufficiently recovered to move safely on their own. No antagonist is available for Telezol so lynx anesthetized with this drug were monitored until the animal recovered on its own in the box-trap and then released. If captured and in poor body condition, lynx were anesthetized with either Telezol (2 mg/kg) or medetomodine/ketemine and returned to the Frisco Creek Wildlife Rehabilitation Center for treatment. HABITAT USE Gross habitat use was documented by recording canopy vegetation at aerial locations. More refined descriptions of habitat use by reintroduced lynx were obtained through following lynx tracks in the snow (i.e., snow-tracking) and site-scale habitat data collection conducted at sites found through this method to be used by lynx. See Shenk (2006) for detailed methodologies. DIET AND HUNTING BEHAVIOR Winter diet of reintroduced lynx was estimated by documenting successful kills through snowtracking. Prey species from failed and successful hunting attempts were identified by either tracks or remains. Scat analysis also provided information on foods consumed. Scat samples were collected wherever found and labeled with location and individual lynx identification. Only part of the scat was collected (approximately 75%); the remainder was left in place in the event that the scat was being used by the animal as a territory mark. Site-scale habitat data collected for successful and unsuccessful snowshoe hare kills were compared. 8 �SNOWSHOE HARE ECOLOGY A study designed to evaluate the importance of young, regenerating lodgepole pine and mature Engelmann spruce / subalpine fir stands in Colorado by examining density and demography of snowshoe hares that reside in each was initiated in 2005. Specifically, the study was designed to evaluate small and medium lodgepole pine stands and large spruce/fir stands where the classes “small”, “medium”, and “large” refer to the diameter at breast height (dbh) of overstory trees as defined in the United States Forest Service R2VEG Database (small = 2.54−12.69 cm dbh, medium = 12.70−22.85 cm, and large = 22.86−40.64 cm dbh; J. Varner, United States Forest Service, personal communication). The study design was also developed to identify which of the numerous density-estimation procedures available perform accurately and consistently using an innovative, telemetry augmentation approach as a baseline. Movement patterns and seasonal use of deciduous cover types such as riparian willow were assessed. Finally, the study was designed to further expound on the relationship between density, demography, and stand-type by examining how snowshoe hare density and demographic rates vary with specific vegetation, physical, and landscape characteristics of a stand. RESULTS REINTRODUCTION Effort From 1999 through 2006, 218 lynx were reintroduced into southwestern Colorado (Table 1). No lynx were released in 2007. All lynx were released with either VHF or dual VHF/satellite radio collars so they could be monitored for movement, reproduction and survival. The CDOW does not plan to release any additional lynx in 2008. Movement Patterns and Distribution Numerous travel corridors were used repeatedly by more than one lynx. These travel corridors include the Cochetopa Hills area for northerly movements, the Rio Grande Reservoir-SilvertonLizardhead Pass for movements to the west, and southerly movements down the east side of Wolf Creek Pass to the southeast through the Conejos River Valley. Lynx appear to remain faithful to an area during winter months, and exhibit more extensive movements away from these areas in the summer. A total of 9496 aerial and 23791 satellite locations were obtained from the 218 reintroduced lynx, radio-collared Colorado kittens (n = 14) and unmarked lynx captured in Colorado (n = 2) as of June 30, 2007. The majority of these locations were in Colorado (Figure 2). Some reintroduced lynx dispersed outside of Colorado into Arizona, Idaho, Iowa, Kansas, Montana, Nebraska, Nevada, New Mexico, South Dakota, Utah and Wyoming (Figure 2). The majority of surviving lynx from the reintroduction effort currently continue to use high elevation (> 2900 m), forested terrain in an area bounded on the south by New Mexico north to Independence Pass, west as far as Taylor Mesa and east to Monarch Pass. Most movements away from the Core Release Area were to the north. Relative Use The lynx use-density surfaces resulting from the fixed kernel analyses provided relative probabilities of finding lynx in areas throughout their distribution. A single use-density surface was calculated separately for both the aerial (n = 8058) and satellite truncated datasets (n = 16240). Relative Use in Colorado.-- All 218 lynx released in Colorado, all radio-collared kittens and 2 captured unmarked adults were located at least once in Colorado. The majority of these lynx remained in Colorado. The use-density surfaces within Colorado were displayed separately for both the aerial (Figure 3) and satellite truncated datasets (Figure 4). Of the total locations available in the 9 �truncated datasets used to generate the use-density surfaces, 7953 of the aerial locations and 13,241 of the satellite locations were in Colorado. Aerial and satellite use-density surfaces indicated similar high usedensity areas. Satellite locations indicated broader spatial use by lynx because satellite collars provided more locations than flights. The use-density surface for lynx use in Colorado indicates two primary areas of use. The first is the Core Research Area (see Figure 1) and a secondary core centered in the Collegiate Peaks Wilderness (Figures 3 and 4). High use is also documented for 1) the area east of Dillon, on both the north and south sides of I70 and 2) the area north of Hwy 50 centered around Gunnison and then north to Crested Butte. These last 2 high use areas are smaller in extent than the 2 core areas. Relative Use in New Mexico.-- Combining the non-truncated aerial (n = 81) and satellite lynx location (n = 928) datasets, lynx used New Mexico consistently and with an increasing number of individuals from 1999 through 2006 (Table 2). Data for 2007 represents only a partial year and thus trend in numbers of individuals using New Mexico for 2007 cannot be made, however continued use of New Mexico into 2007 was documented Sixty lynx (37 females: 23 males) were found within New Mexico from February 1999 through March 2007 (Table 2). Excluding all aerial and satellite lynx locations collected in the first 180 days after release (truncated datasets; n = 61 aerial locations, n = 569 satellite locations), a total of 35 individual lynx (22 females: 13 males) were found within New Mexico from September 1999 through March 2007 (Table 3). The decrease in number of lynx frequenting New Mexico in 2001 through 2003 (Tables 2 and 3) was more likely due to fewer satellite collars functioning in those years rather than indicating less use of the area by lynx. The satellite transmitters placed on lynx in 2000 were failing and no new lynx were released or re-collared in 2001 and 2002. This decrease in satellite locations is present throughout the lynx distribution and is also reflected in the numbers presented below for Utah and Wyoming The use-density surface for lynx use in New Mexico indicates the primary area of use being located either immediately south of the Colorado border and south of the Conejos River Valley (an area of high use in Colorado) or east of Taos (Figure 5). The use-density surfaces throughout both Colorado and New Mexico are displayed so that lynx use within New Mexico can be directly compared to lynx use throughout Colorado (Figure 6). Relative Use in Utah.-- Combining the non-truncated aerial (n = 10) and satellite lynx location (n = 574) datasets, lynx used the analysis area consistently and with an increasing number of individuals from 1999 through 2006 (Table 4). Data for 2007 represents only a partial year and thus trend in numbers of individuals using the state for 2007 cannot be made, however continued use of Utah into 2007 was documented. Twenty-two lynx (7 females: 15 males) were found within Utah from February 1999 through March 2007 (Table 4). Excluding all aerial and satellite lynx locations collected in the first 180 days after release (truncated datasets; n = 7 aerial locations, n = 399 satellite locations), 17 individual lynx (6 females: 11 males) were found within Utah from September 1999 through March 2007 (Table 5). The use-density surface for lynx use in Utah indicates the primary area of use being located in the Uinta Mountains (Figure 7). The use-density surfaces throughout both Colorado and Utah are displayed so that lynx use within Utah can be directly compared to lynx use throughout Colorado (Figure 8). Relative Use in Wyoming.-- Combining the non-truncated aerial (n = 34) and satellite lynx location (n = 1780) datasets, lynx used the analysis area consistently and with an increasing number of individuals from 1999 through 2006 (Table 6). Data for 2007 represents only a partial year and thus trend in numbers of individuals using the state for 2007 cannot be made, however continued use of the Wyoming into 2007 was documented. Thirty-three lynx (14 females: 19 males) were found within 10 �Wyoming from February 1999 through March 2007 (Table 6). Excluding all aerial and satellite lynx locations collected in the first 180 days after release (truncated datasets; n = 28 aerial locations, n = 1533 satellite locations), 27 individual lynx (13 females: 14 males) were found within Wyoming from September 1999 through March 2007 (Table 7). The use-density surface for lynx use in Wyoming indicates the primary area of use being located either immediately north of the Colorado border in the Medicine Bow National Forest or in the northwest quadrant of the state including areas in Yellowstone and Teton National Parks and the Laramie Range (Figure 9). The use-density surfaces throughout both Colorado and Wyoming are displayed so that lynx use within Wyoming can be directly compared to lynx use throughout Colorado (Figure 10). Home Range Reproductive females had the smallest 90% utilization distribution annual home ranges ( x = 75.2 km2, SE = 15.9 km2, n = 19), followed by attending males ( x = 102.5 km2, SE = 39.7 km2, n = 4). Nonreproductive females had the largest annual home ranges ( x = 703.9 km2, SE = 29.8 km2, n = 32) followed by non-reproductive males ( x = 387.0 km2, SE = 73.5 km2, n = 6). Combining all nonreproductive animals yielded a mean annual home range of 653.8 km2 (SE = 145.4 km2, n = 38). Survival Initial survival rate estimates for reintroduced lynx were completed, however, further analyses need to be conducted before estimates will be presented. As of June 30, 2007, CDOW was actively monitoring/tracking 71 of the 120 lynx still possibly alive (Table 8). There are 50 lynx that we have not heard signals on since at least June 30, 2006 and these animals are classified as ‘missing’ (Table 8). One of these missing lynx is a mortality of unknown identity, thus only 49 are truly missing. Possible reasons for not locating these missing lynx include 1) long distance dispersal, beyond the areas currently being searched, 2) radio failure, or 3) destruction of the radio (e.g., run over by car). CDOW continues to search for all missing lynx during both aerial and ground searches. Two of the missing lynx released in 2000 are thought to have slipped their collars. Mortality Factors Of the total 218 adult lynx released, we have 98 known mortalities as of June 30, 2007 (Table 9). The primary known causes of death included 30.6% human-induced deaths which were confirmed or probably caused by collisions with vehicles or gunshot. Starvation and disease/illness accounted for 19.4% of the deaths; starvation was a significant cause of mortality in the first year of releases only. An additional 35.7% of known mortalities were from unknown causes. Mortalities occurred throughout the areas where lynx moved, including 13 in New Mexico, 4 in Wyoming and Nebraska, 3 in Utah and 1 each in Arizona, Kansas and Montana (Figure 2, Table 10). Reproduction Field crews weighed, photographed, PIT-tagged the kittens and took hair, blood and saliva samples from the kittens for genetic work in an attempt to confirm paternity. Lynx kittens weigh approximately 200 grams at birth and do not open their eyes until they are 10-17 days old. Kittens were processed as quickly as possible (11-32 minutes) to minimize the time the kittens were without their mother. While working with the kittens the females remained nearby, often making themselves visible to the field crews. The females generally continued a low growling vocalization the entire time personnel were at the den. In all cases, the female returned to the den site once field crews left the area. At all dens the females appeared in excellent condition, as did the kittens. 11 �2003.-- Nine pairs of lynx were documented during the 2003 breeding season (March and April) from the 17 females we were monitoring. In May and June, 6 dens and a total of 16 kittens were found in the lynx Core Release Area in southwestern Colorado (Table 11, Figure 1). The kittens weighed from 270-500 grams. The dens were scattered throughout the Core Release Area, with no dens found outside the core area. All the dens were in Engelmann spruce/subalpine fir forests in areas of extensive downfall. Elevations ranged from 3240-3557 m. Four of the 6 females that we know had kittens in summer 2003 were still with kittens at the end of April 2004. Two of those females still had 2 kittens with them at that time. Visual observations in February 2004 of one female with 2 kittens indicated all 3 were in good body condition. The mortality of female YK00F16 and her 1 kitten in October 2003 from plague was not due to poor habitat or prey conditions, and thus we might assume she would have raised the 1 kitten to this stage as well. Three probable kitten deaths from female YK00F19 were from 1 litter that most likely failed very early. Through snow-tracking in winter 2003-04 an unknown female (no radio frequency heard in the area of the tracks) we also documented 1-2 additional kittens born spring 2003 and still alive in winter 2004. Of the 16 kittens we found in summer 2003, we documented the following by April 2004: 6 confirmed alive, 7 confirmed dead, and 3 some evidence dead. Although we tried, we were not able to capture any of the 6 surviving kittens to fit them with radio-collars for subsequent monitoring. 2004.-- In Spring 2004, 26 females from the releases in 1999, 2000 and 2003 had active radiocollars. Of these, we documented 18 possible mating pairs of lynx during breeding season. All 4 of the females that had kittens with them through winter 2003-04 bred again spring 2004; 2 with the same male they successfully bred with spring 2003. During May-June 2004 we found 11 dens and a total of 30 kittens (Table 11). The kittens weighed from 250-770 grams. Three of the 11 females with kittens were from the 2003 releases. Three additional litters were documented after denning season through either observation of a female lynx with kittens or snow-tracking females with kittens that were not one of the 11 females found on dens. From the size of the kittens they would have been born during the normal denning season in May or June. Nine additional kittens were observed from these litters for a total of 39 known kittens born in 2004. Two of these additional litters were documented from direct follow-ups to sighting made by the public and reported to CDOW. Two females that had kittens in 2003 and reared at least part of their litters through March 2004, bred and had kittens again in 2004. Two of the litters documented by direct observation or snow-tracking are from females whose collars were no longer functioning. Seven kittens born in 2004 were captured at approximately 10-months of age and fitted with dual satellite/VHF collars. Six of the 7 were still alive and being monitored as of June 30, 2006. The cut collar of one kitten CO04M15 was left at the Silverton Post Office on October 25, 2005. We assume this lynx is dead. 2005.-- In spring 2005 we had 40 females from the releases in 1999, 2000, 2003 and 2004 that had active radio-collars. We documented 23 possible mating pairs of lynx during breeding season. During May-June 2005 we visited 16 dens and found a total of 46 kittens (Table 11). An additional female (BC03F10) had a den we were not able to get to during May or June due to high water during spring run-off. Female BC03F03 was hit and killed on I-70 on 5/19/2005. She had 2 fetuses in her uterus, so would have contributed to reproduction this year had she lived. All of the 2005 dens were scattered throughout the high elevation areas of Colorado, south of I70. Most of the dens were in Engelmann spruce/subalpine fir forests in areas of extensive downfall. Elevations ranged from 3117-3586 m. Four of the females would not leave the den until we reached out to pick up a kitten. 12 �One female, YK00F10 has had litters 3 years in a row. In 2003 she had 4 kittens and raised 2 through the winter. In 2004 she had 2 kittens and raised both through the winter, in 2005 she had 4 kittens again. She has had all 3 litters in the same general area and has had the same mate for 3 years. Eight additional females had their second litter in Colorado in 2005. Three females from the 2004 releases had litters in 2005. Year 2005 was the second consecutive year that we had females released the prior spring find a territory and a mate within a year and produced live young. In reproduction season 2004 we had 3 females released in spring 2003 that also produced live young the next year. Of those 3, 2 successfully raised at least part of their litters through winter 2005. Seven kittens born in 2005 were captured at approximately 10-months of age and fitted with dual satellite/VHF collars. One of the 7 was still alive and being monitored as of June 30, 2007. 2006.-- In spring 2006, 42 females were being monitored. We found 4 dens in May and June 2006 with 11 kittens total (Table 11). Lynx CO04F07, a female lynx born in Colorado in 2004, was the mother of one of these litters which documented the first recruitment of Colorado-born lynx into the Colorado breeding population. There were at least 2 surviving kittens as of spring 2007. We were unsuccessful in capturing these kittens for collar placement. The percent of tracked females found with litters in 2006 was lower (0.095) than in the 3 previous years (0.413, SE = 0.032, Table 11). However, all demographic and habitat characteristics measured at the 4 dens that were found in 2006 were comparable to all other dens found (Table 11). Mean number of kittens per litter from 2003-2006 was 2.78 (SE = 0.05) and sex ratio of females to males was equal ( x = 1.14, SE = 0.14). 2007.-- During May and June 2007 we monitored 34 females for reproduction (Table 11). No dens were found. Den Sites.-- A total of 37 dens have been found from 2003-2006. All of the dens except one have been scattered throughout the high elevation areas of Colorado, south of I-70. In 2004, 1 den was found in southeastern Wyoming, near the Colorado border. Dens were located on steep ( x slope = 30o , SE=2o), north-facing, high elevation ( x = 3354 m, SE = 31 m) slopes. The dens were typically in Engelmann spruce/subalpine fir forests in areas of extensive downfall of coarse woody debris (Shenk 2006). All dens were located within the winter use areas used by the females. Captures Two adult lynx were captured in 2001 for collar replacement. One lynx was captured in a tomahawk live-trap, the other was treed by hounds and then anesthetized using a jab pole. Five adult lynx were captured in 2002; 3 were treed by hounds and 2 were captured in padded leghold traps. In 2004, 1 lynx was captured with a Belisle snare and 6 adult lynx were captured in box-traps. Trapping effort was substantially increased in winter and spring 2005 and 12 adult lynx were captured and re-collared. Eight reintroduced lynx were captured in winter and spring 2006. In 2007, 11 reintroduced adult lynx were captured and re-collared. All lynx captured in Colorado from 2005-2007 were caught in box-traps. In addition, as part of the collaring trapping effort, 14 Colorado-born kittens were captured and collared at approximately 10-months of age. Seven 2004-born kittens were collared in spring 2005, and 7, 2005-born kittens were collared in spring 2006. We were not successful at capturing and collaring any kittens born in 2006 in winter 2006-07. We did however, capture 2 adults (approximate age 2 years old) in winter 2006-07 that had no PIT-tags or radio collars. We assume these 2 lynx were from litters born in Colorado that were never found at dens (i.e., why there were no PIT-tags). All lynx captured for collaring 13 �or re-collaring were fitted with new Sirtrack TM dual VHF/satellite collars and re-released at their capture locations. Seven adult lynx were captured from March 1999-June 30, 2007 because they were in poor body condition (Table 12). Five of these lynx were successfully treated at the Frisco Creek Rehabilitation Center and re-released in the Core Release Area. One lynx, BC00F7, died from starvation and hypothermia within 1 day of capture at the rehabilitation center. Lynx QU04M07 died 3 days after capture at the rehabilitation center. Necropsy results documented starvation as the cause of death that was precipitated by hydrocephalus and bronchopneumonia (unpublished data T. Spraker, CSUVTH). Seven lynx were captured (either by CDOW personnel or conservation personnel in other states) because they were in atypical habitat outside the state of Colorado (Table 12). They were held at Frisco Creek Rehabilitation Center for a minimum of 3 weeks, fitted with new Sirtrack TM dual VHF/satellite collars and re-released in the Core Release Area in Colorado. Five of these 7 lynx were still alive 6 months post-re-release but 3 had already dispersed out of Colorado and 2 stayed in Colorado through June 30, 2007. Two lynx died within 6 months of re-release: 1 died of starvation in Colorado and the other died of unknown causes in Nebraska. Two lynx captured out of state and re-released currently remain in Colorado. HABITAT USE Landscape-scale daytime habitat use was documented from 9496 aerial locations of lynx collected from February 1999-June 30, 2007. Throughout the year Engelmann spruce - subalpine fir was the dominant cover used by lynx. A mix of Engelmann spruce, subalpine fir and aspen (Populus tremuloides) was the second most common cover type used throughout the year. Various riparian and riparian-mix areas were the third most common cover type where lynx were found during the daytime flights. Use of Engelmann spruce-subalpine fir forests and Engelmann spruce-subalpine fir-aspen forests was similar throughout the year. There was a trend in increased use of riparian areas beginning in July, peaking in November, and dropping off December through June. Site-scale habitat data collected from snow-tracking efforts indicate Engelmann spruce and subalpine fir were also the most common forest stands used by lynx for all activities during winter in southwestern Colorado. Comparisons were made among sites used for long beds, dens, travel and where they made kills. Little difference in aspect, mean slope and mean elevation were detected for 3 of the 4 site types including long beds, travel and kills where lynx typically use gentler slopes ( x = 15.7o ) at an mean elevation of 3173 m, and varying aspects with a slight preference for north-facing slopes. See Shenk (2006) for more detailed analyses of habitat use. DIET AND HUNTING BEHAVIOR Winter diet of lynx was documented through detection of kills found through snow-tracking. Prey species from failed and successful hunting attempts were identified by either tracks or remains. Scat analysis also provided information on foods consumed. A total of 506 kills were located from February 1999-April 2007. We collected over 900 scat samples from February 1999-April 2007 that will be analyzed for content. In each winter, the most common prey item was snowshoe hare, followed by red squirrel (Tamiusciurus hudsonicus; Table 13). The percent of snowshoe hare kills found however, varied annually from a low of 55.56% in 1999 to a high of 90.77% in winter 2002-2003. SNOWSHOE HARE ECOLOGY The first year of a study to evaluate snowshoe hare densities, demography and seasonal movement patterns among small and medium tree-sized lodgepole pine stands and mature spruce/fir stands was completed and preliminary results presented (Appendix I). 14 �DISCUSSION In an effort to establish a viable population of lynx in Colorado, CDOW initiated a reintroduction effort in 1997 with the first lynx released in winter 1999. From 1999 through spring 2007, 218 lynx were released in the Core Release Area. Locations of each lynx were collected through aerial- or satellite-tracking to document movement patterns and to detect mortalities. Most lynx remain in the high elevation, forested areas in southwestern Colorado. The use-density surfaces for lynx use in Colorado indicate two primary areas of use. The first is the Core Research Area (see Figure 1) and a secondary core centered in the Collegiate Peaks Wilderness (Figures 3, 4). High use is also documented for 1) the area east of Dillon, on both the north and south sides of I70 and 2) the area north of Hwy 50 centered around Gunnison and then north to Crested Butte. These last 2 high use areas are smaller in extent than the 2 core areas. Dispersal movement patterns for lynx released in 2000 and subsequent years were similar to those of lynx released in 1999 (Shenk 2000). However, more animals released in 2000 and subsequent years remained within the Core Release Area than those released in 1999. This increased site fidelity may have been due to the presence of con-specifics in the area on release. Numerous travel corridors within Colorado have been used repeatedly by more than 1 lynx. These travel corridors include the Cochetopa Hills area for northerly movements, the Rio Grande Reservoir-Silverton-Lizardhead Pass for movements to the west, and southerly movements down the east side of Wolf Creek Pass to the southeast to the Conejos River Valley. Lynx appear to remain faithful to an area during winter months, and exhibit more extensive movements away from these areas in the summer. Reproductive females had the smallest 90% utilization distribution home ranges ( x = 75.2 km2, SE = 15.9 km2), followed by attending males ( x = 102.5 km2, SE = 39.7 km2) and non-reproductive animals ( x = 653.8 km2, SE = 145.4 km2). Most lynx currently being tracked are within the Core Release Area. During the summer months, lynx were documented to make extensive movements away from their winter use areas. Extensive summer movements away from areas used throughout the rest of the year have been documented in native lynx in Wyoming and Montana (Squires and Laurion 1999). Current data collection methods used for the Colorado lynx reintroduction program were not specifically designed to address the reintroduced lynx movements or use of areas in other states. In particular, the core research and release area were in Colorado. Therefore, the number of aerial locations obtained would be far fewer in other states than in Colorado which would bias low the number of lynx and intensity of lynx use documented outside the state. In contrast, obtaining satellite locations is not biased by the location of the lynx. Satellite locations are, however, biased by the shorter time the satellite transmitters function, approximately 18 months versus 60 months for the VHF transmitters used to obtain the aerial locations. However, data collected to meet objectives of the lynx reintroduction program were used to provide information to help address the question of lynx use outside of Colorado. Due to the rarity of flights conducted outside Colorado, only use-density surfaces generated from satellite locations were used to document relative lynx use of areas in New Mexico, Utah and Wyoming. New Mexico and Wyoming have been used continuously by lynx since the first year lynx were released in Colorado (1999) to the present (Tables 2, 6). Lynx reintroduced in Colorado were first documented in Utah in 2000 (Table 4) and are still being documented there to date. In addition, all levels of lynx use-density documented throughout Colorado are also represented in New Mexico, Utah and Wyoming from none to the highest level of use (Figures 5, 7, 9). One den was found in Wyoming. Although no reproduction has been documented in New Mexico or Utah to date, documenting areas of the 15 �highest intensity of use and the continuous presence of lynx within these states for over six years does suggest the potential for year-round residency of lynx and reproduction in those states. The use-density surface for lynx use in New Mexico indicates the primary areas of use being located immediately south of the Colorado border and south of the Conejos River Valley (an area of high use in Colorado) or east of Taos (Figure 5). In Utah, the primary area of use is located in the Uinta Mountains (Figure 7). Lynx use in Wyoming is focused in 2 primary areas, the Medicine Bow National Forest in south-central Wyoming and in the northwest quadrant of the state including areas in Yellowstone and Teton National Parks and the Laramie Range (Figure 9). From 1999-June 2007, there were 98 mortalities of released adult lynx. Human-caused mortality factors are currently the highest causes of death with approximately 30.6% attributed to collisions with vehicles or gunshot. Starvation and disease/illness accounted for 19.4% of the deaths while 35.7% of the deaths were from unknown causes. Lynx mortalities were documented throughout all areas lynx used, including 28 (28.6%) occurring in other states (Figure 2, Table10). Half of the out-of-state mortalities were documented in New Mexico. Reproduction is critical to achieving a self-sustaining viable population of lynx in Colorado. Reproduction was first documented from the 2003 reproduction season and again in 2004, 2005 and 2006. Lower reproduction occurred in 2006 (Table 11) but did include a Colorado-born female giving birth to 2 kittens, documenting the first recruitment of Colorado-born lynx into the Colorado breeding population. No reproduction was documented in 2007. The cause of the decreased reproduction in 2006 and 2007 is unknown. One possible explanation would be a decrease in prey abundance. Additional reproduction is likely to have occurred in all years from females we were no longer tracking, and from Colorado-born lynx that have not been collared. The dens we find are more representative of the minimum number of litters and kittens in a reproduction season. To achieve a viable population of lynx, enough kittens need to be recruited into the population to offset the mortality that occurs in that year and hopefully even exceed the mortality rate to achieve an increasing population. The use-density surfaces depict intensity of use by location. Why certain areas would be used more intensively than others should be explained by the quality of the habitat in those areas. Characteristics of areas used by lynx, as documented through aerial locations and snow-tracking of lynx in the Colorado core research area, include mature Engelmann spruce-subalpine fir forest stands with 4265% canopy cover and 15-20% conifer understory cover (Shenk 2006). Within these forest stand types, lynx appear to have a slight preference for north-facing, moderate slopes ( x = 15.7°) at high elevations ( x = 3173 m; Shenk 2006). Snow-tracking of released lynx also provided information on hunting behavior and diet through documentation of kills, food caches, chases, and diet composition estimated through prey remains. The primary winter prey species (n = 506) were snowshoe hare (Table 12) with an annual x = 74.9% (SE = 4.6, n = 9) and red squirrel (annual x = 16.5%, SE = 4.1, n = 9). Thus, areas of good habitat must also support populations of snowshoe hare and red squirrel. In winter, lynx reintroduced to Colorado appear to be feeding on their preferred prey species, snowshoe hare and red squirrel in similar proportions as those reported for northern lynx during lows in the snowshoe hare cycle (Aubry et al. 1999). Environmental conditions in the springs and summers of 2003 and 2006 resulted in high cone crops during their following winters based on field observations, resulting in increased red squirrel abundance. This may partially explain the higher percent of red squirrel kills, and thus a lower percent of snowshoe hare kills, found in winters 2003-04 and 2006-07 (Table 12). 16 �Caution must be used in interpreting the proportion of identified kills. Such a proportion ignores other food items that are consumed in their entirety and thus are biased towards larger prey and may not accurately represent the proportion of smaller prey items, such as microtines, in lynx winter diet. Through snow-tracking we have evidence that lynx are mousing and several of the fresh carcasses have yielded small mammals in the gut on necropsy. The summer diet of lynx has been documented to include less snowshoe hare and more alternative prey than in winter (Mowat et al., 1999). All evidence suggests reintroduced lynx are finding adequate food resources to survive. Mowat et al. (1999) suggest lynx and snowshoe hare select similar habitats except that hares select more dense stands than lynx. Very dense understory limits hunting success of the lynx and provides refugia for hares. Given the high proportion of snowshoe hare in the lynx diet in Colorado, we might then assume the habitats used by reintroduced lynx also depict areas where snowshoes hare are abundant and available for capture by lynx in Colorado. From both aerial locations taken throughout the year and from the site-scale habitat data collected in winter, the most common areas used by lynx are in stands of Engelmann spruce and subalpine fir. This is in contrast to adjacent areas of Ponderosa pine, pinyon juniper, aspen and oakbrush. The lack of lodgepole pine in the areas used by the lynx may be more reflective of the limited amount of lodgepole pine in southwestern Colorado, the Core Release Area, rather than avoidance of this tree species. Hodges (1999) summarized habitats used by snowshoe hare from 15 studies as areas of dense understory cover from shrubs, stands that are densely stocked, and stands at ages where branches have more lateral cover. Species composition and stand age appears to be less correlated with hare habitat use than is understory structure (Hodges 1999). The stands need to be old enough to provide dense cover and browse for the hares and cover for the lynx. In winter, the cover/browse needs to be tall enough to still provide browse and cover in average snow depths. Hares also use riparian areas and mature forests with understory. Site-scale habitat use documented for lynx in Colorado indicate lynx are most commonly using areas with Engelmann spruce understory present from the snow line to at least 1.5 m above the snow. The mean percent understory cover within the habitat plots is typically less than 15% regardless of understory species. However, if the understory species is willow, percent understory cover is typically double that, with mean number of shrubs per plot approximately 80, far greater than for any other understory species. In winter, hares browse on small diameter woody stems (<0.25"), bark and needles. In summer, hares shift their diet to include forbs, grasses, and other succulents as well as continuing to browse on woody stems. This shift in diet may express itself in seasonal shifts in habitat use, using more or denser coniferous cover in winter than in summer. The increased use of riparian areas by lynx in Colorado from July to November may reflect a seasonal shift in hare habitat use in Colorado. Major (1989) suggested lynx hunted the edge of dense riparian willow stands. The use of these edge habitats may allow lynx to hunt hares that live in habitats normally too dense to hunt effectively. The use of riparian areas and riparian-Engelmann spruce-subalpine fir and riparian-aspen mixes documented in Colorado may stem from a similar hunting strategy. However, too little is known about habitat use by hares in Colorado to test this hypothesis at this time. Lynx also require sufficient denning habitat. Denning habitat has been described by Koehler (1990) and Mowat et al. (1999) as areas having dense downed trees, roots, or dense live vegetation. We found this to be in true in Colorado as well (Shenk 2006). In addition, the dens used by reintroduced lynx were at high elevations and on steep north-facing slopes. All females that were documented with kittens denned in areas within their winter-use area. 17 �SUMMARY From results to date it can be concluded that CDOW developed release protocols that ensure high initial post-release survival of lynx, and on an individual level, lynx demonstrated they can survive longterm in areas of Colorado. We also documented that reintroduced lynx exhibited site fidelity, engaged in breeding behavior and produced kittens that were recruited into the Colorado breeding population. What is yet to be demonstrated is whether current conditions in Colorado can support the recruitment necessary to offset annual mortality in order to sustain the population. Monitoring of reintroduced lynx will continue in an effort to document such viability. ACKNOWLEDGEMENTS The lynx reintroduction program involves the efforts of literally hundreds of people across North America, in Canada and USA. Any attempt to properly acknowledge all the people who played a role in this effort is at risk of missing many people. The following list should be considered to be incomplete. CDOW CLAWS Team (1998-2001): Bill Andree, Tom Beck, Gene Byrne, Bruce Gill, Mike Grode, Rick Kahn (Program Leader), Dave Kenvin, Todd Malmsbury, Jim Olterman, Dale Reed, John Seidel, Scott Wait, Margaret Wild. CDOW: John Mumma (Director 1996-2000), Russell George (Director 2001-2003), Bruce McCloskey (Director 2004-2007), Conrad Albert, Jerry Apker, Laurie Baeten, Cary Carron, Don Crane, Larry DeClaire, Phil Ehrlich, Lee Flores, Delana Friedrich, Dave Gallegos, Juanita Garcia, Drayton Harrison, Jon Kindler, Ann Mangusso, Jerrie McKee, Gary Miller, Melody Miller, Mike Miller, Kirk Navo, Robin Olterman, Jerry Pacheo, Mike Reid, Tom Remington, Ellen Salem, Eric Schaller, Mike Sherman, Jennie Slater, Steve Steinert, Kip Stransky, Suzanne Tracey, Anne Trainor, Scott Wait, Brad Weinmeister, Nancy Wild, Perry Will, Lisa Wolfe, Brent Woodward, Kelly Woods, Kevin Wright. Lynx Advisory Team (1998-2001): Steve Buskirk, Jeff Copeland, Dave Kenny, John Krebs, Brian Miller (Co-Leader), Mike Phillips, Kim Poole, Rich Reading (Co-Leader), Rob Ramey, John Weaver. U. S. Forest Service: Kit Buell, Joan Friedlander, Dale Gomez, Jerry Mastel, John Squires, Fred Wahl, Nancy Warren. U. S. Fish and Wildlife Service: Lee Carlson, Gary Patton (1998-2000), Kurt Broderdorp. State Agencies: Alaska: ADF&G: Cathie Harms, Mark Mcnay, Dan Reed (Regional Manager), Wayne Reglin (Director), Ken Taylor (Assist. Director), Ken Whitten, Randy Zarnke, Other:Ron Perkins (trapper), Dr. Cort Zachel (veterinarian). Washington: Gary Koehler. National Park Service: Steve King. Colorado State University: Alan Franklin, Gary White. Colorado Natural Heritage Program: Rob Schorr, Mike Wunder. Canada: British Columbia: Dr. Gary Armstrong (veterinarian), Mike Badry (government), Paul Blackwell (trapper coordinator), Trappers: Dennis Brown, Ken Graham, Tom Sbo, Terry Stocks, Ron Teppema, Matt Ounpuu. Yukon: Government: Arthur Hoole (Director), Harvey Jessup, Brian Pelchat, Helen Slama, Trappers: Roger Alfred, Ron Chamber, Raymond Craft, Lance Goodwin, Jerry Kruse, Elizabeth Hofer, Jurg Hofer, Guenther Mueller (YK Trapper’s Association), Ken Reeder, Rene Rivard (Trapper coordinator), Russ Rose, Gilbert Tulk, Dave Young. Alberta: Al Cook. Northwest Territories: Albert Bourque, Robert Mulders (Furbearer Biologist), Doug Steward (Director NWT Renewable Res.), Fort Providence Native People. Quebec: Luc Farrell, Pierre Fornier. Colorado Holding Facility: Herman and Susan Dieterich, Kate Goshorn, Loree Harvey, Rachel Riling. Pilots: Dell Dhabolt, Larry Gepfert, Al Keith, Jim Olterman, Matt Secor, Brian Smith, Whitey Wannamaker, Steve Waters, Dave Younkin. 18 �Field Crews (1999-2007): Steve Abele, Brandon Barr, Bryce Bateman, Todd Bayless, Nathan Berg, Ryan Besser, Jessica Bolis, Mandi Brandt, Brad Buckley. Patrick Burke, Braden Burkholder, Paula Capece, Stacey Ciancone, Doug Clark, John DePue, Shana Dunkley, Tim Hanks, Carla Hanson, Dan Haskell, Nick Hatch, Matt Holmes, Andy Jennings, Susan Johnson, Paul Keenlance, Patrick Kolar, Tony Lavictoire, Jenny Lord, Clay Miller, Denny Morris, Kieran O’Donovan, Gene Orth, Chris Parmater, Jake Powell, Jeremy Rockweit, Jenny Shrum, Josh Smith, Heather Stricker, Adam Strong, Dave Unger, David Waltz, Andy Wastell, Mike Watrobka, Lyle Willmarth, Leslie Witter, Kei Yasuda, Jennifer Zahratka. Research Associates: Bob Dickman, Grant Merrill. Data Analysts: Karin Eichhoff, Joanne Stewart, Anne Trainor. Data Entry: Charlie Blackburn, Patrick Burke, Rebecca Grote, Angela Hill, Mindy Paulek. Mary Schuette and Dave Theobald provided assistance with the GIS analysis and M. Schuette generated the maps used in this report Photographs: Tom Beck, Bruce Gill, Mary Lloyd, Rich Reading, Rick Thompson. Funding: CDOW, Great Outdoors Colorado (GOCO), Turner Foundation, U.S.D.A. Forest Service, Vail Associates, Colorado Wildlife Heritage Foundation. LITERATURE CITED AUBRY, K. B., G. M. KOEHLER, J. R. SQUIRES. 1999. Ecology of Canada lynx in southern boreal forests. Pages 373-396 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. BYRNE, G. 1998. Core area release site selection and considerations for a Canada lynx reintroduction in Colorado. Report for the Colorado Division of Wildlife. CURTIS, J. T. 1959. The vegetation of Wisconsin. University of Wisconsin Pres, Madison. HODGES, K. E. 1999. Ecology of snowshoe hares in southern boreal and montane forests. Pages 163206 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. KOEHLER, G. M. 1990. Population and habitat characteristics of lynx and snowshoe hares in north central Washington. Canadian Journal of Zoology 68:845-851. KOLBE, J. A., J. R. SQUIRES, T. W. PARKER. 2003. An effective box trap for capturing lynx. Journal of Wildlife Management 31:980-985. LAYMON, S. A. 1988. The ecology of the spotted owl in the central Sierra Nevada, California. PhD Dissertation University of California, Berkeley, California. MAJOR, A. R. 1989. Lynx, Lynx canadensis canadensis (Kerr) predation patterns and habitat use in the Yukon Territory, Canada. M. S. Thesis, State University of New York, Syracuse. MOWAT, G., K. G. POOLE, AND M. O’DONOGHUE. 1999. Ecology of lynx in northern Canada and Alaska. Pages 265-306 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. POOLE, K. G., G. MOWAT, AND B. G. SLOUGH. 1993. Chemical immobilization of lynx. Wildlife Society Bulletin 21:136-140. SHENK, T. M. 1999. Program narrative Study Plan: Post-release monitoring of reintroduced lynx (Lynx canadensis) to Colorado. Colorado Division of Wildlife, Fort Collins, Colorado. __________. 2001. Post-release monitoring of lynx reintroduced to Colorado. Job Progress Report, Colorado Division of Wildlife, Fort Collins, Colorado. __________. 2006. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report July: 1-45. Colorado Division of Wildlife, Fort Collins, Colorado 19 �SILVERMAN, B.W. 1986. Density Estimation for Statistics and Data Analysis. Chapman and Hall. New York, New York, USA. SQUIRES, J. R. AND T. LAURION. 1999. Lynx home range and movements in Montana and Wyoming: preliminary results. Pages 337-349 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University Press of Colorado, Boulder, Colorado. U. S. FISH AND WILDLIFE SERVICE. 2000. Endangered and threatened wildlife and plants: final rule to list the contiguous United States distinct population segment of the Canada lynx as a threatened species. Federal Register 65, Number 58. WHITE, G.C. AND K. P. BURNHAM. 1999. Program MARK: Survival estimation from populations of marked animals. Bird Study 46 Supplement, 120-138. WILD, M. A. 1999. Lynx veterinary services and diagnostics. Job Progress Report for the Colorado Division of Wildlife. Fort Collins, Colorado. Prepared by ___________________________________ Tanya M. Shenk, Wildlife Researcher Table 1. Lynx released in Colorado from February 1999 through June 30, 2007. No lynx were released in 2001, 2002 or 2007. Year Females Males TOTAL 1999 22 19 41 2000 35 20 55 2003 17 16 33 2004 17 20 37 2005 18 20 38 2006 6 8 14 TOTAL 115 103 218 20 �Table 2. All individual lynx (n = 60) documented through either aerial or satellite locations (nontruncated datasets) by year in New Mexico from February 1999 – March 2007. Lynx ID AK99F10 AK99F13 AK99F17 AK99F3 AK99F5 AK99F8 AK99M11 AK99M26 AK99M9 BC99M4 YK99F3 YK99M3 YK99M6 YK99M7 AK00F2 AK00F5 BC00F10 BC00F14 BC00F6 BC00F8 BC00M04 BC00M11 BC00M4 YK00F11 YK00F2 YK00F4 YK00F7 BC03F03 BC03F04 BC03F06 BC03F08 BC03M02 BC03M05 BC03M08 QU03F01 QU03F04 QU03F07 BC04F02 BC04F03 BC04F05 BC04M02 BC04M13 QU04F05 QU04F08 QU04F09 QU04M02 QU04M04 BC05F04 BC05M02 QU05F03 QU05M01 QU05M05 YK05F01 YK05M01 BC06F05 BC06F07 BC06F09 BC06M12 YK06F01 YK06M01 Total Lynx 1999 X 2000 2001 2002 Year 2003 2004 2005 2006 2007 X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X 8 11 2 3 9 21 17 17 X X X X X X X X 14 X 2 �Table 3. All individual lynx (n = 35) documented at least 180 days after their initial release (truncated datasets) through either aerial or satellite locations, by year in New Mexico from September 1999 – March 2007. Lynx ID AK99F13 AK99F3 AK99F5 AK99M11 AK99M9 BC99M4 YK99F3 YK99M6 AK00F2 AK00F5 BC00F14 BC00F8 BC00M04 BC00M11 BC00M4 YK00F11 YK00F2 YK00F4 YK00F7 BC03F03 BC03F06 BC03M02 BC03M08 QU03F04 QU03F07 BC04F02 BC04M13 QU04F05 QU04F08 QU04F09 QU05M05 YK05M01 BC06F07 BC06M12 YK06F01 Total Lynx 2000 X X 2001 2002 2003 Year 2004 X 2005 2006 2007 X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X 6 2 3 2 12 22 X X X X X X 10 X X X X X X 11 X 2 �Table 4. All individual lynx (n = 22) documented through either aerial or satellite locations (nontruncated datasets) by year in Utah from February 1999 – March 2007. Lynx ID AK99F5 AK00F5 AK00M3 BC00M09 BC00M13 YK00F7 BC03F03 BC03M06 BC03M08 BC03M10 QU03F03 BC04M01 QU04M04 QU04M05 BC05M01 BC05M03 CO05F20 QU05F05 QU05M03 QU05M08 YK05M01 YK06M01 Total Lynx 2000 2001 2002 2003 Year 2004 2005 X 2006 2007 X X X X X X X X X X X X X X X X X X X X X 1 1 0 2 4 X X X X X 7 7 5 Table 5. All individual lynx (n = 17) documented at least 180 days after their initial release (truncated datasets) through either aerial or satellite locations, by year in Utah from September 1999 – March 2007. Year Lynx ID AK99F5 AK00F5 AK00M3 BC00M09 YK00F7 BC03M06 BC03M10 QU03F03 BC04M01 QU04M04 BC05M01 BC05M03 CO05F20 QU05F05 QU05M03 YK05M01 YK06M01 Total Lynx 2000 2001 2002 2003 2004 2005 X 2006 2007 X X X X X X X X X X X X X X X X 0 1 0 0 23 3 6 X X X X X 7 5 �Table 6. All individual lynx (n = 33) documented through either aerial or satellite locations (nontruncated datasets) by year in Wyoming from February 1999 – March 2007. Lynx ID AK99M6 BC00F14 BC00M13 YK00F11 BC03F03 BC03M02 BC03M06 BC03M09 QU03M01 BC04F02 BC04M01 BC04M08 BC04M13 CO04F10 CO04M05 CO04M06 QU04F01 QU04F02 QU04F07 QU04M04 QU04M05 BC05M03 BC05M08 MB05F01 MB05F02 MB05F03 QU05F04 QU05F05 QU05F08 QU05M08 YK05M03 BC06M10 BC06M13 Total Lynx 1999 X 2001 2003 X X Year 2004 2005 2006 2007 X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X 1 1 2 16 24 14 X X X X X X X X X 16 X X X X X 5 �Table 7. All individual lynx (n = 27) documented at least 180 days after their initial release (truncated datasets) through aerial or satellite locations, by year in Wyoming from September 1999 – March 2007. Year Lynx ID BC00F14 BC00M13 YK00F11 BC03F03 BC03M02 BC03M06 BC03M09 QU03M01 BC04F02 BC04M08 BC04M13 CO04F10 CO04M05 CO04M06 QU04F01 QU04F02 QU04M04 QU04M05 BC05M03 MB05F01 MB05F02 MB05F03 QU05F04 QU05F05 QU05F08 QU05M08 BC06M13 Total Lynx 2001 2003 2004 X X X X 2005 2006 2007 X X X X X X X X X X X X X X X X X X X X X X X X X 1 2 14 11 X X X X X X X X X X X X X X 15 X X X X X 5 Table 8. Status of adult lynx reintroduced to Colorado as of June 30, 2007. Lynx Released Known Dead Possible Alive Missing Monitoring/tracking a 1 is unknown mortality Females 115 54 61 23 38 Males 103 43 60 27 33 Unknown 1 TOTALS 218 98 120 49a 71 Table 9. Causes of death for all lynx released into southwestern Colorado 1999-2006 as of June30, 2007. Cause of Death Unknown Gunshot Hit by Vehicle Starvation Other Trauma Plague Probable Gunshot Predation Probable Predation Illness Total Mortalities Mortalities In Colorado (%) 20 (57.1) 7 (53.8) 8 (66.7) 9 (90.0) 7 (87.5) 7 (100) 4 (80) 3 (100) 3 (100) 2 (100) 70 (71.4) Total (%) 35 (35.7) 13 (13.3) 12 (12.2) 10 (10.2) 8 (8.1) 7 (7.1) 5 (5.1) 3 (3.1) 3 (3.1) 2 (2.0) 98 25 Outside Colorado (%) 15 (42.9) 6 (46.2) 4 (33.3) 1 (10.0) 1 (12.5) 0 (0) 1 (20) 0 (0) 0 (0) 0 (0) 28 (28.6) �Table 10. Known lynx mortalities (n = 28) and causes of death documented by state outside of Colorado from February 1999 – June 30, 2007. Lynx ID AK99F8 Unknown AK99M11 YK99M06 AK99F13 YK00F04 BC99M04 QU05M01 QU04F05 QU03F07 BC00M04 YK06F01 BC03M08 BC06F07 AK99M06 AK99M01 QU05M08 MB05F02 BC00F14 QU04F07 BC06M10 QU04F02 AK00M03 QU05M03 YK06M01 YK99F01 YK00M03 YK05M03 State Date Mortality Recorded Cause of Death New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico Nebraska Nebraska Nebraska Nebraska Wyoming Wyoming Wyoming Wyoming Utah Utah Utah Arizona Kansas Montana 7/30/1999 2000 1/27/2000 6/19/2000 6/22/2000 4/20/2001 6/7/2002 8/22/2005 8/26/2005 9/15/2005 7/19/2006 10/19/2006 10/19/2006 1/8/2007 11/16/1999 1/11/2005 10/1/2006 2/13/2007 7/28/2004 9/21/2004 8/15/2006 3/14/2007 7/2/2001 10/26/2005 12/4/2006 9/15/2005 9/30/2005 11/8/2005 Starvation Hit by Vehicle Unknown Probable Gunshot Unknown Gunshot Gunshot Unknown Hit by Vehicle Unknown Unknown Unknown Unknown Gunshot Gunshot Snared (Other Trauma) Unknown Gunshot Unknown Unknown Vehicle Collision Unknown Unknown Unknown Unknown Gunshot Vehicle Collision Unknown Table 11. Lynx reproduction summary statistics for 2003-2007. No reproduction was documented from 1999-2002 or in 2007. Year Females Tracked Dens Found in May/June 2003 2004 2005 2006 2007 Total 17 26 40 42 34 6 11 17 4 0 Percent Tracked Females with Kittens 0.353 0.462 0.425 0.095 0.0 Additional Litters Found in Winter 2 1 26 Mean Kittens Per Litter (SE) 2.67 (0.33) 2.83 (0.24) 2.88 (0.18) 2.75 (0.47) Total Kittens Found 16 39 50 11 0 116 Sex Ratio M/F (SE) 1.0 1.5 0.8 1.2 1.14 (0.14) �Table 12. Lynx captured because they were in poor body condition or were in atypical habitat and their fates 6 months post re-release and as of June 30, 2007. Lynx ID BC99F6 Date of Capture 3/25/1999 State Where Captured Colorado Reason For Capture Poor body condition Date of Re-release 5/28/1999 Status 6 Months Post Re-release Dead AK99M9 3/24/2000 Colorado 5/3/2000 Missing AK99F2 4/18/2000 Colorado 5/22/2000 BC00F7 2/11/2001 Colorado Alive in Colorado Dead BC00M13 3/21/2001 Wyoming BC03M08 9/5/2003 Colorado QU04M07 2/2/2006 Colorado Poor body condition Poor body condition Poor body condition Poor body condition Poor body condition Poor body condition BC04M01 11/5/2004 Utah QU04F02 4/10/2005 Nebraska QU05M08 11/25/2005 Wyoming QU04M04 12/5/2006 Utah YK00F7 12/12/2006 Utah YK05M02 1/1/2007 Kansas BC04M08 1/22/2007 Wyoming N/A 4/24/2001 1/1/2004 N/A Alive in Colorado Alive in Colorado Dead Died 7/19/1999 in Colorado from vehicle collision Last located 5/3/2000, collar failure Last located 7/30/2003 in Colorado Died at Rehab Center on 2/12/2001 Last located 10/26/2004 in Colorado Died in New Mexico of unknown causes 10/19/06 Died at Rehab Center on 2/5/2006 from hydrocephalous and pneumonia Atypical habitat Atypical habitat 12/5/2004 Atypical habitat Atypical habitat Atypical habitat Atypical habitat Atypical habitat 4/18/2006 Dead 1/20/2007 1/20/2007 Dead in Colorado Alive in Utah Died 3/14/2007 in Wyoming (good habitat) of unknown causes Died of unknown causes in Nebraska 10/1/2006 Died of starvation in Colorado, found 3/19/07 In Utah as of 6/30/2007 2/2/2007 Alive in Iowa In Iowa as of 6/30/2007 2/15/2007 Alive in Colorado In Colorado as of 6/30/2007 5/7/2005 Alive in Colorado Alive in Wyoming Current Status In Colorado as of 6/30/2007 Table 13. Number of kills found each winter field season through snow-tracking of lynx and percent composition of kills of the three primary prey species. Field Season 1999 1999-2000 2000-2001 2001-2002 2002-2003 2003-2004 2004-2005 2005-2006 2006-2007 n 9 83 89 54 65 37 78 50 41 Snowshoe Hare 55.56 67.47 67.42 90.74 90.77 67.57 83.33 90.00 61.00 Prey (%) Cottontail Red Squirrel 22.22 0 19.28 1.20 19.10 8.99 5.56 0 6.15 0 27.03 2.70 10.26 0 0.08 0 39.0 0 27 Other 22.22 12.05 4.49 3.70 3.08 2.70 6.41 0.02 0 �Figure 1. Lynx are monitored throughout Colorado and by satellite throughout the western United States. The lynx core release area, where all lynx were released, is located in southwestern Colorado. A lynx-established core use area has developed in the Taylor Park and Collegiate Peak area in central Colorado. 28 �Figure 2. All documented lynx locations (non-truncated datasets) obtained from either aerial (yellow circles) or satellite (red circles) tracking from February 1999 through June 30, 2007. All known lynx mortality locations (n = 97) are displayed as stars. 29 �Figure 3. Use-density surface for lynx aerial locations (truncated dataset) in Colorado from September 1999-March 2007 30 �Figure 4. Use-density surface for lynx satellite locations (truncated dataset) in Colorado from September 1999-March 2007. 31 �Figure 5. Use-density surface for lynx satellite locations (truncated dataset) in New Mexico from September 1999-March 2007 32 �Figure 6. Use-density surface for lynx satellite locations (truncated dataset) in Colorado and New Mexico from September 1999-March 2007. 33 �Figure 7. Use-density surface for lynx satellite locations (truncated dataset) for Utah from September 1999-March 2007. 34 �Figure 8. Use-density surface for lynx satellite locations (truncated dataset) in Colorado and Utah from September 1999-March 2007. 35 �Figure 9. Use-density surface for lynx satellite locations (truncated dataset) in Wyoming from September 1999-March 2007. 36 �Figure 10. Use-density surface for lynx satellite locations (truncated dataset) in Colorado and Wyoming from September 1999-March 2007. 37 �APPENDIX I Colorado Division of Wildlife July 2006 - June 2007 WILDLIFE RESEARCH REPORT State of Cost Center Work Package Task No. Colorado 3430 0670 2 Federal Aid Project: N/A : Division of Wildlife : Mammals Research : Lynx Conservation : Density, Demography, and Seasonal Movements of Snowshoe Hare in Colorado : Period Covered: July 1, 2006- June 30, 2007 Author: J. S. Ivan, Ph. D. Candidate, Colorado State University Personnel: T. M. Shenk, CDOW and G. C. White of Colorado State University. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT A program to reintroduce the threatened Canada lynx (Lynx canadensis) into Colorado was initiated in 1997 with the first lynx release in 1999. Analysis of scat collected from winter snow tracking indicated that snowshoe hares (Lepus americanus) comprised 65–90% of the winter diet of reintroduced lynx. Thus, existence of lynx in Colorado and success of the reintroduction effort hinge at least in part on maintaining adequate and widespread populations of hares. Beginning in July 2006, I initiated a study to assess the relative value of 3 forest stand types (mature [“large”] spruce/fir, sapling [“small”] lodgepole pine, pole-sized [“medium”] lodgepole pine) that purportedly provide high quality hare habitat in Colorado. Estimates and comparisons of survival, recruitment, finite population growth rate, and maximum (late summer) and minimum (late winter) snowshoe hare densities for each stand will provide the metrics for assessing value. Number of individuals captured, number of captures, and number of locations obtained per hare during the first year of the project appear adequate for attaining the objectives of this study. Some hare deaths due to capture myopathy (most likely cause) occurred during initial trapping periods in both the summer and winter sampling seasons. However, changes to the trapping protocol, trapping schedule, and bait provided seem to have alleviated the problem. Densities during summer were highest in small lodgepole stands (0.47 hares/ha, 95% CI: 0.41-0.54), followed by large spruce/fir (0.18 hares/ha, 95% CI: 0.12-0.25) and medium lodgepole (0.02 hares/ha, 95% CI: 0.01-0.03). During winter, densities in small lodgepole stands dropped and became more variable across replicates (0.18 hares/ha, 95% CI: 0.01-0.35). Medium lodgepole stands gained hares (0.07 hares/ha, 95% CI: 0.050.10). Spruce/fir stands remained at the same density as during summer (0.17 hares/ha, 95% CI: 0.110.23). 38 �WILDIFE RESEARCH REPORT DENSITY, DEMOGRAPHY, AND SEASONAL MOVEMENTS OF SNOWSHOE HARE IN COLORADO JACOB S. IVAN P. N. OBJECTIVE Assess the relative value of 3 forest stand types (old spruce/fir, sapling lodgepole, pole-sized lodgepole) that purportedly provide high quality snowshoe hare (Lepus americanus) habitat by estimating survival, recruitment, finite population growth rate, and maximum (late summer) and minimum (late winter) snowshoe hare densities for each type. SEGMENT OBJECTIVES 1. Complete mark-recapture work across all replicate stands during late summer (mid-July through midSeptember) and winter (mid-January through March). 2. Obtain daily telemetry locations on radio-tagged hares for 10 days immediately after capture periods, as well as monthly between primary trapping sessions. 3. Locate, retrieve, and refurbish radio tags as mortalities occur. 4. Summarize initial sampling efforts and provide initial density estimates for Wildlife Research Reports for Colorado Division of Wildlife (CDOW). INTRODUCTION A program to reintroduce the threatened Canada lynx (Lynx canadensis) into Colorado was initiated in 1997 with the first lynx released in 1999. Since that time, 204 lynx have been released in the state, and an extensive effort to determine their movements, habitat use, reproductive success, and food habits has ensued (Shenk 2006). Analysis of scat collected from winter snow tracking indicates that snowshoe hares (Lepus americanus) comprise 65–90% of the winter diet of reintroduced lynx (T. Shenk, Colorado Division of Wildlife, unpublished data). Thus, as in the far north where the intimate relationship between lynx and snowshoe hares has captured the attention of ecologists for decades, the existence of lynx in Colorado and success of the reintroduction effort may also hinge on maintaining adequate and widespread populations of hares. Colorado represents the extreme southern range limit for both lynx and snowshoe hares (Hodges 2000). At this latitude, habitat for each species is less widespread and more fragmented compared to the continuous expanse of boreal forest at the heart of lynx and hare ranges. Neither species exhibits dramatic cycles as occur farther north, and typical lynx (≤2−3 lynx/100km2; Aubry et al. 2000) and hare (≤1−2 hares/ha; Hodges 2000) densities in the southern part of their range correspond to cyclic lows form northern populations (2-30 lynx/100 km2, 1−16 hares/ha; Aubry et al. 2000, Hodges 2000, Hodges et al. 2001). Whereas extensive research on lynx-hare ecology has occurred in the boreal forests of Canada, literature regarding the ecology of these species in the southern portion of their range is relatively sparse. This scientific uncertainty is acknowledged in the “Canada Lynx Conservation Assessment and Strategy,” a formal agreement between federal agencies intended to provide a consistent approach to lynx conservation on public lands in the lower 48 states (Ruediger et al. 2000). In fact, one of the explicit guiding principles of this document is to “retain future options…until more conclusive information concerning lynx management is developed.” Thus, management recommendations in this agreement are 39 �decidedly conservative, especially with respect to timber management, and are applied broadly to cover all habitats thought to be of possible value to lynx and hare. This has caused controversy where recommendations conflict with competing resource management goals. Accurate identification and detailed description of lynx-hare habitat in the southern Rocky Mountains would permit more informed and refined management recommendations. A commonality throughout the snowshoe hare literature, regardless of geographic location, is that hares are associated with dense understory vegetation that provides both browse and protection from elements and predators (Wolfe et al. 1982, Litvaitis et al. 1985, Hodges 2000, Homyack et al. 2003, Miller 2005). In western mountains, this understory can be provided by relatively young conifer stands regenerating after stand-replacing fires or timber harvest (Sullivan and Sullivan 1988, Koehler 1990, Koehler 1990, Bull et al. 2005) as well as mature, uneven-aged stands (Beauvais 1997, Griffin 2004). Hares may also take advantage of seasonally abundant browse and cover provided by deciduous, open habitats (e.g., riparian willow [Salix spp.], aspen [Populus tremuloides]; Wolff 1980, Miller 2005). In drier portions of hare range, such as Colorado, regenerating stands can be relatively sparse, and hares may be more associated with mesic, late-seral forest and/or riparian areas than with young stands (Ruggiero et al. 2000). Numerous investigators have sought to determine the relative importance of these distinctly different habitat types with regards to snowshoe hare ecology. Most previous evaluations were based on hare density or abundance (Bull et al. 2005), indices to hare density and abundance (Wolfe et al. 1982, Koehler 1990, Beauvais 1997, Miller 2005), survival (Bull et al. 2005), and/or habitat use (Dolbeer and Clark 1975). Each of these approaches provides insight into hare ecology, but taken singly, none provide a complete picture and may even be misleading. For example, extensive use of a particular habitat type may not accurately reflect the fitness it imparts on individuals, and density can be high even in “sink” habitats (Van Horne 1983). A more informative approach would be to measure density, survival, and habitat use simultaneously in addition to recruitment and population growth rate through time. Griffin (2004) employed such an approach and found that summer hare densities were consistently highest in young, dense stands. However, he also noted that only dense mature stands held as many hares in winter as in summer. Furthermore hare survival seemed to be higher in dense mature stands, and only dense mature stands were predicted (by matrix projection) to impart a mean positive population growth rate on hares. Griffin’s (2004) study occurred in the relatively moist forests of Montana, which share many similarities but also many notable differences with Colorado forests including levels of fragmentation, species composition, elevation, and annual precipitation. Density estimation is a key component in assessing the value of a particular stand type and is the common currency by which hare populations are compared across time and space. However, density can be a difficult metric to estimate accurately. Density estimation based on capture-recapture methods is a well-developed field (Otis et al. 1978, White et al. 1982), but is often too costly and labor intensive to be implemented on scales necessary to effectively monitor density over a biologically meaningful area. Also, density can be difficult to assess from grid-trapping efforts because it is often unclear how much area was effectively sampled by the grid (Williams et al. 2002:314). Different approaches can produce density estimates that differ by an order of magnitude even when calculated from the same data (Zahratka 2004). Indices such as pellet plot counts and distance sampling of pellet groups can be used to estimate density, but each of these has limitations as well (Krebs et al. 1987, Eriksson 2006). Pellet plot counts are typically conducted by laying out numerous rectangular or circular plots along transect lines randomly placed within a study site. All pellets occurring within the plot are counted and removed on an annual basis. The mean number of pellets per plot is then inserted into a regression equation that gives an estimate of hare density (Krebs et al. 1987). Estimates from this technique correlate well with density estimates derived from simultaneous mark-recapture studies occurring in the 40 �same area (Krebs et al. 2001, Murray et al. 2002, Mills et al. 2005, Homyack et al. 2006). However, because fecal deposition rates can vary by season and diet, and because pellet decomposition rates can vary with altitude, climate, aspect, precipitation, and cover type, region-specific, stand-specific, and/or season-specific equations should be developed before this technique is employed for a given area and season (Krebs et al. 2001, Prugh and Krebs 2004, Murray et al. 2005). Density estimates vary with plot size and shape, requiring equations specific to these geometric considerations as well (McKelvey et al. 2002). Pellet counts tend to yield more precise and unbiased density estimates when plots are visited and cleared more than once per year (e.g., plots cleared in the fall and then counted in the spring to estimate winter density) because variability in deposition and decomposition rates is reduced (Homyack et al. 2006). However, this requires considerably more work and expense than an annual survey. Some studies have conducted pellet plot counts without first clearing plots (e.g., Bartmann and Byrne 2001). This saves time and money, but requires the ability to discern fresh (this year) pellets from old pellets, which can be difficult and is generally not a recommended approach (Prugh and Krebs 2004, Murray et al. 2005). Distance sampling is a well-developed method for estimating the density of objects in a given area (Buckland et al. 2001). In general, observers walk a pre-defined sampling transect and record each object of interest along with the perpendicular distance of that object from the transect line. This information is then used to develop a detection function which is in turn used to estimate density (Buckland et al. 2001). The method assumes all objects on the line are seen with certainty, objects are not double-counted, distance measures are accurate, and transect lines are located randomly within a study area (Buckland et al. 2001). Recently, distance sampling has been used to indirectly estimate hare density by first estimating the pellet group density of hares, then using fecal deposition and decomposition rates as a link back to hare density (Eriksson 2006). In general, distance sampling is more efficient than pellet plot counts as it does not require the tedious layout of hundreds of plots or counting individual pellets. This advantage is most recognizable in situations where pellet groups occur at low densities. Conversely, at extremely high densities, it may become difficult to distinguish pellet groups, and plots may be preferable (Marques et al. 2001). Regardless, distance sampling of pellet groups to estimate animal density also requires habitat and season specific decomposition and defecation rates, which can be difficult to obtain (Marques et al. 2001). For this project, I have chosen to provide land managers with information relating demographic rates, as well as density, to forest stand characteristics. Thus, I will use mark-recapture techniques as data from such an approach can provide information on both density and demography. I will address the “effective trapping area” issue using a new approach that augments mark-recapture data with telemetry locations of animals using the grid. The study outlined below is designed principally to evaluate the importance of young, regenerating lodgepole pine (Pinus contorta) and mature Engelmann spruce (Picea engelmannii)/ subalpine fir (Abies lasiocarpa) stands in Colorado by examining density and demography of snowshoe hares that reside in each (Figure 1). My hope is that information gathered from this research will be drawn upon as managers make routine decisions, leading to landscapes that include stands capable of supporting abundant populations of hares. I assume that if management agencies focus on providing habitat, hares will persist. Specifically, I will evaluate small and medium lodgepole pine stands and large spruce/fir stands where the classes “small”, “medium”, and “large” refer to the diameter at breast height (dbh) of overstory trees as defined in the United States Forest Service R2VEG Database (small = 2.54−12.69 cm dbh, medium = 12.70−22.85 cm, and large = 22.86−40.64 cm dbh; J. Varner, United States Forest Service, personal communication). I also intend to identify which of the numerous density-estimation procedures available perform accurately and consistently using an innovative, telemetry augmentation approach as a 41 �baseline. I will assess movement patterns and seasonal use of deciduous cover types such as riparian willow. Finally, I will further expound on the relationship between density, demography, and stand type by examining how snowshoe hare density and demographic rates vary with specific vegetation, physical, and landscape characteristics of a stand. Hypotheses 1) In general, snowshoe hare density in Colorado will be relatively low (≤0.5 hares/ha) compared to densities reported in northern boreal forests, even immediately post-breeding when an influx of juveniles will bolster hare numbers. 2) Snowshoe hare density will be consistently highest in small lodgepole pine stands, followed by large spruce/fir and medium lodgepole pine, respectively. 3) Survival will generally be highest in mature (large) spruce/fir stands followed by small and medium lodgepole pine, respectively. 4) Finite population growth rate will be consistently at or above 1.0 in mature spruce/fir stands with survival contributing most significantly to the growth rate. Finite growth rates for the lodgepole pine stands will be more variable. 5) Snowshoe hares will significantly shift their home ranges to make use of abundant food and cover provided by riparian willow (and/or aspen) habitats in summer. 6) Snowshoe hare density, survival, and recruitment will be highly correlated with understory cover and stem density. STUDY AREA The study area stretches from Taylor Park to Pitkin in central Colorado (Figure 2). Elevation ranges from 2700 m to 4000 m. Sagebrush (Artemisia spp.) dominates broad, low-lying valleys. Most montane areas are covered by even-aged, large-diameter lodgepole pine forests with sparse understory. Moist, north-facing slopes and areas near tree line are dominated by large-diameter Engelmann spruce/subalpine fir. Interspersed along streams and rivers are corridors of willow. Patches of aspen occur sporadically on southern exposures. This area was chosen over other potential study areas in the state because 1) it contained numerous examples of the 3 stand types of interest (more southern regions lack naturally occurring stands of lodgepole pine), 2) it was not subject to confounding effects of largescale mountain pine beetle outbreak as were more northern stands, and 3) an adequate number of radio frequencies were available to support a large study with hundreds of radio-tagged individuals. Within the study area I selected sample stands based on the following: Potential replicate stands were required to be 1) close enough geographically to minimize differences due to climate, weather, and topography, but are far enough apart to be considered independent, 2) adjacent to one or more riparian willow corridors, 3) within 1 km of an access road for logistical purposes, 4) of suitable size and shape to admit a 16.5-ha trapping grid, and 5) consistent in their management history (i.e., replicate lodgepole pine stands were clear-cut and/or thinned within 1-2 years of each other). I queried the U.S. Forest Service R2VEG GIS database using the criteria listed above to initially develop a suite of potential sample stands. I further narrowed this suite after obtaining updated standlevel information from local USFS personnel (Art Haines, Silviculturalist, USFS Gunnison Ranger District, personal communication). Finally, I ground-truthed potential stands and qualitatively assessed their representativeness and similarity to other potential replicates. Given the numerous constraints imposed, very few stands met all criteria. Thus, I was unable to randomly select sample stands from a population of suitable stands. Rather, I subjectively chose the “best” stands from among the handful that met my criteria. Small lodgepole stands rarely occur on the landscape in patches large enough to fit a full 42 �7 x 12 trapping grid. To accommodate this, I sampled 6 replicate small lodgepole stands (rather than 3) using 6 x 7 trapping grids (1/2 size). METHODS Experimental Design/Procedures Variables.--The response variables of interest for this project include stand-specific snowshoe hare density (D), apparent survival (φ), recruitment (f), finite population growth rate (λ), and a metric of seasonal movement. Density is the number of hares per unit area and will be estimated using a variety of conventional techniques as well as a rigorous method that incorporates radio telemetry. The standspecific demographic parameters will be estimated primarily from capture-mark-recapture methods. As such, apparent survival is defined as the probability that a marked animal alive and in the population at time i survives and is in the population at time i + 1. Apparent survival encompasses losses due to both death and emigration. Recruitment is the number of new animals in the population at time i + 1 per animal in the population at time i. New recruits can arise from on-site reproduction as well as immigration. The finite population growth rate is the number of animals in a given age class at time i + 1 divided by the number present at time i. Shifts in home range will be assessed by comparing the seasonal proportion of telemetry locations in deciduous habitats using multi-response permutation procedures (MRPP; Zimmerman et al. 1985, White and Garrott 1990). Potential explanatory variables for snowshoe hare density, demographics, and movement include general species composition and structural stage of each stand in which response variables are measured. Additionally, stem density, horizontal cover, and canopy cover (to a lesser extent) are highly correlated with snowshoe hare abundance and habitat use (Wolfe et al. 1982, Litvaitis et al. 1985, Hodges 2000, Zahratka 2004, Miller 2005). Thus, I will further characterize vegetation in each stand by measuring stem density by size class (1-7 cm, 7.1-10 cm, and >10 cm), percent canopy cover, percent horizontal cover of understory and basal area. Basal area is an easily obtainable metric that may be correlated with the other variables and is recorded routinely during timber cruises, whereas the others are not. Thus, it might prove a useful link for biologists designing management strategies for snowshoe hare. Additionally, I will record physical covariates such as ambient temperature, precipitation, and snow depth at each stand during sampling periods as well as precipitation 1-3 years prior to sampling. Finally, I will calculate potentially important landscape metrics such as patch size and level of fragmentation. Sampling.--All trapping and handling procedures have been approved by the Colorado State University Animal Care and Use Committee and filed with the Colorado Division of Wildlife. Snowshoe hares breed synchronously and generally exhibit 2 birth pulses in Colorado (although in some years, some individuals may have 3 litters), with the first pulse terminating approximately June 5−20 and the second approximately July 15–25 (Dolbeer 1972). To obtain a maximum density estimate, I began data collection on the first suite of sites immediately following the second birth pulse in late July. Along with a crew of 5 technicians, I deployed one 7 × 12 trapping grid (50-m spacing between traps; grid covers 16.5 ha) in the large spruce/fir and medium lodgepole stands within the first suite, along with 2 6 × 7 grids in 2 small lodgepole stands. Grid set up and trap deployment followed Griffin (2004) and Zahratka (2004). Grid locations and orientation within each stand were chosen subjectively to accommodate logistical constraints and to ensure that hares using the grid had ample opportunity to use adjacent riparian willow zones. Traps were deployed in all 4 stands in a single day. As traps are deployed, they were locked open and “pre-baited” with apple slices and commercial rabbit chow. During winter, hay cubes were added to traps as well (see Discussion). On days 2-4, the crew continued pre-baiting, replacing apples and rabbit chow as necessary. The purpose of this extended pre-baiting was to maximize capture rates when trapping began. This minimized the number of trap-nights needed to capture the desired number of animals which in turn minimized trapping-related stress as well as the likelihood that 43 �American marten (Martes americana) keyed into trap lines and preyed on entrapped hares, as has occurred in previous studies (J. Zahratka, personal communication). During pilot work in winter 2005, I observed low but increasing capture rates (<0.20) during the first 3 nights of trapping, with higher, more stable capture probabilities after 3 days (approximately 0.35–0.45). Thus 3 days of pre-baiting seems reasonable. Traps were set on the afternoon of the 4th day and checked early each morning and again in the evening on days 5–9. By checking traps in both morning and evening I prevent hares from being entrapped >13 hours, which should minimize capture stress. A crew of 2 people worked together on each grid to check traps and process captures as quickly as possible. All captured hares were coaxed out of the trap and into a dark handling bag by blowing quick shots of air on them from behind. Hares remained in the handling bag, physically restrained with their eyes covered, for the entire handling process. Each individual was aged, sexed, marked with a passive integrated transponder (PIT) tag and temporary ear mark (to track PIT tag retention), then released. Aging consisted of assigning each individual as either juvenile (<1 year old, <1000 g) or adult (≥1 year old, ≥1000 g) based on weight. This criterion is accurate through the end of September at which point juveniles are difficult to distinguish from adults (K. Hodges, University of British Columbia; P. Griffin, University of Montana, personal communication). After the first day of trapping, all captured hares were scanned for a PIT tag prior to any handling and those already marked were recorded and immediately released. Traps and bait were completely removed from the grid on day 10. In addition to PIT tags and ear marks, I radio collared up to 10 hares captured on each grid with a 28-g mortality-sensing transmitter (BioTrack, LTD) to facilitate unbiased density estimation as well as assessment of seasonal movements. I expected heterogeneity in snowshoe hare movements and use of the grid area, with potential bias surfacing due to location at which a hare is captured (e.g., hares captured on the edge of a grid may use the grid area differently than those captured at the center), and differential behavioral responses to trapping (e.g., young individuals may have lower capture probabilities and thus may be more likely to be captured on later occasions). To guard against the first potential bias, I randomly selected a starting trap location each morning and ran the grid systematically from that point. Thus, the first several hares encountered (and collared) were as likely to be from the inner part of the grid as from the edge. To protect against the second potential source of bias, I refrained from deploying the final 3 collars until days 4 and 5 of the trapping session. Immediately following the removal of traps, the field crew began work locating each radiocollared hare 1–2 times per day for 10 days. Most locations were obtained by triangulation from relatively close proximity, but some were obtained by “homing” on a signal (Samuel and Fuller 1996, Griffin 2004) taking care not to push hares while approaching them. Because hares are largely nocturnal (Keith 1964, Mech et al. 1966, Foresman and Pearson 1999), I made an effort to conduct telemetry work at various times of the night (safety and logistics permitting) and day to gather a representative sample of locations for each hare. The crew gathered telemetry locations for radio-collared hares on the initial sites for 8 to 10 days. Then the 10−day trapping procedure and 8 to 10−day telemetry work were repeated on the 3 grids comprising suite 2 (Figure 3). The cycle was repeated once more for grids in suite 3 (Figure 3). The entire process was repeated during the winter when densities should have been at a minimum. In summary, for any given 9-week sampling period, I collected data from 12 total grids, 1 spruce/fir, 1 medium lodgepole, and 2 small lodgepole across 3 replicates. Sampling will occur during 2 such 9-week periods each year − once in late summer and once in late winter – and will continue for 3 years. During the interim between intensive trapping and telemetry work, monthly telemetry checks were conducted from the air to track mortalities and facilitate retrieval of collars from dead hares. Telemetry 44 �work was also occur during “pre-baiting” days after the initial summer sampling session to determine which hares were still alive and immediately available to be sampled by the grid during the ensuing trapping period. Vegetation sampling at each stand will follow protocols established through previous snowshoe hare and lynx work in Colorado (Zahratka 2004, T. Shenk, Colorado Division of Wildlife, personal communication). Specifically, on each of the 12 live-trapping grids, I will lay out 5 × 5 grids (3-m spacing) of vegetation sampling points centered on 15 of the 84 trap locations (Figure 4; 9 points will be sampled on each of the ½-sized small lodgepole stands). At each of the 25 vegetation sampling points, I will record: 1) distance to the nearest woody stem 1.0−7.0 cm, 7.1−10.0 cm, and >10.0 cm in diameter at heights of 0.1 m and 1.0 m above the ground (to capture both summer [0.1 m] and winter [1.0 m] stem density; Barbour et al. 1999), 2) horizontal cover in 0.5-m increments above the ground up to 2 m (Nudds 1977), and 3) canopy cover [present or absent] using a densitometer. Additionally, at the center of all 15 vegetation sampling grid points (i.e., at the trap location), I will measure basal area using an angle gauge. These measurements will be gathered once at the start of the project, unless conditions change due to disturbance such as fire. Temperature will be monitored hourly at each grid during the 6-week intensive sampling periods using data loggers. During winter sampling periods, snow depth measurements will be recorded daily at the same 15 trap locations used to quantify the vegetative attributes of that stand. Data Analysis Density.--I assumed that hare populations were demographically and geographically closed during the short 5-day mark-recapture sampling periods. To obtain a density estimate for each grid, I used the Huggins closed capture model (Huggins 1989, 1991) in Program MARK (White and Burnham 1999) with some modifications. The basic Huggins estimator (no individual covariates) is based on the fact that if pj is the probability that a hare in the population is captured (and marked) for the first time on trapping occasion j, then p * = 1 − (1 − p1 )...(1 − p5 ) is the probability that an individual is captured at least once during a 5-day trapping period (i.e., j = 1,…,5). Accordingly, the basic Huggins estimator for population size, N̂ , is Nˆ = M t +1 / p* where M t +1 is the total number of hares captured. The estimator can be re-written to allow each of the M t +1 individuals captured to have their own p*. In that case, M t +1 Nˆ = ∑1 / pi* . Presumably hares that reside near the edge of a grid encounter fewer traps and are less i =1 likely to be captured than hares residing near the center of a grid. To account for this, I took advantage of the Huggins model with individual covariates to model p* by using the logit link function of program MARK to model pi* as a function of di, where di is distance from the edge of the grid for hare i based on mean capture coordinates. A naïve density estimate for each grid would then be Dˆ = Nˆ / A where A is the area of the grid. However, this gives full credit to all hares, even those whose home range only partially overlaps the grid, which results in a density estimate that is biased high. To correct for this bias, I determined the proportion, ( ~ pk ), of telemetry locations for each of the k = 1,…,10 radio-collared hares that fell within the “naïve grid area.” By incorporating data from multiple grids, a logistic regression model was developed to estimate p% i for all M t +1 animals captured on a grid based on distance from the edge of the grid for hare i (di). Replacing the numerator (i.e., 1) in the Huggins estimator with ( p% i ), gives ⎛ M t +1 ⎞ ~ p / p ⎟ A. ⎟ ⎜∑ a density estimate, Dˆ = ⎜ ⎝ i =1 i * i ⎠ The above-stated approach assumes that radio-collared hares neither gravitate toward nor avoid the former grid area after the 5 days of trapping, 10–20 locations per hare is enough to provide a 45 �reasonable representation of the proportion of time they spend on the grid, and their use of the grid area is representative of other hares that were captured but not collared (i.e., that the logistic regression model of p% i is a useful model). I contend that this type of estimate from grid-based trapping can be construed as a relatively unbiased estimate of density. Using these point estimates and their associated confidence intervals, I compared hare density among seasons and stand types. I will also compare these “true” density estimates to those that would have been obtained using other available methods such as ½ mean maximum distance moved (Wilson and Anderson 1985, Williams et al. 2002:314-315), full mean maximum distance moved (Parmenter et al. 2003), ½ trap interval (Parmenter et al. 2003), “nested grids” (White et al. 1982:120-131), and Program DENSITY (Efford et al. 2004). Demography.--I will analyze mark-recapture data using Pradel temporal symmetry models (Pradel 1996, Nichols and Hines 2002) in a robust design framework (Williams et al. 2002:523-554), which will be available in Program MARK by summer 2006. Thus, I will treat summer and winter sampling occasions as primary periods, and the 5-day trapping sessions within each as secondary periods. The Pradel temporal symmetry models employ both forward and reverse-time evaluation of capture histories to provide estimates of apparent survival ( φ̂ ) and seniority ( γ̂ ). Apparent survival, φi, is the probability that a marked animal alive and in the population at time i survives and is in the population at time i + 1. The seniority parameter, γi , is the reverse-time analogue of survival. Reading backward through a capture history, it is the probability that a marked animal alive and in the population at time i was alive and in the sampled population at time i − 1. If N is the number of animals present in the population, N i φi ≈ N i +1γ i +1 and N i +1 / N i = φi / γ i +1 = λ i . Also, if fi is recruitment rate, or the number of recruits at time i + 1 per animal present at time i, then N i +1 = N i φi + N i f i . Rearranging and substituting into the previous equation gives f i = φi (1/ γ i − 1) . Thus, using Pradel models, one can estimate recruitment and finite population growth rate in addition to survival (Pradel 1996, Nichols and Hines 2002). I will use Akaike’s Information Criterion corrected for small sample size (AICc; Burnham and Anderson 1998) to determine whether models with time-dependent parameters or constant parameters are best supported by the data. I will derive estimates of the above-mentioned parameters from the best model or from model averaging. I anticipate pooling capture data across sites to obtain φˆ i , λˆ i , and fˆi for each stand type for each interval between primary sampling periods (5 estimates of each). I also anticipate simply estimating these parameters for “generic hares”, treating both juveniles and adults as a single group or age class. Given that juveniles are morphometrically indistinguishable from adults by their first fall of life (K. Hodges, University of British Columbia; P. Griffin, University of Montana, personal communication), adult and juvenile survival rates are similar (Griffin 2004), and there is little evidence for age-specific differences in pregnancy rates or litter size (Dolbeer 1972), this approach seems justified. However, if I happen to capture sufficient numbers of juveniles and adults, the design I have laid out here allows for treating the age classes separately. This, in turn, may permit me to decompose the contribution that fi makes to λi into the portion of that contribution due to on-site reproduction and that due to immigration (Nichols et al. 2000). Similarly, it may also be possible using my telemetry data to decompose apparent survival, φi , into emigration and mortality. Such fortuitous situations would facilitate the identification of source and sink habitats if they exist. Seasonal Movements.--I will assess whether snowshoe hares seasonally shift their home ranges using the multi-response permutation procedure (MRPP; Zimmerman et al. 1985, White and Garrott 1990:134-135). Under this approach, telemetry locations are grouped by season (summer and winter), and an MRPP statistic is calculated as the weighted average of the distance between all possible pairs of locations within groups compared to the average distance between all possible pairs ignoring groups. The 46 �null hypothesis is that the distribution of locations is the same for both groups (seasons). Sufficiently small values of the test statistic suggest that within group distances are smaller than distances measured ignoring groups, which is evidence against the null in favor of a group (seasonal) effect. P-values are obtained by calculating the percentile of the observed test statistic relative to all possible test statistics that could be computed by re-arranging the data into all possible groups of 2. The MRPP procedure is sensitive and can detect even small changes in use of an area (White and Garrott 1990:136). I propose a priori that changes in proportional use of deciduous habitats <0.10 in magnitude are unlikely to be biologically significant. Vegetation.--I will calculate mean stem density, horizontal cover, canopy cover, and basal area for each season−stand type as well as temperature, precipitation, snow depth information, and landscape metrics. These will be entered into the MARK design matrix as covariates to population size (~density) and survival in a random effects analysis. As such, I will be able to quantify the amount of variation in population size or survival that is due to differences in vegetation, landscape, or weather relative to the amount left to other causes. Sample size.--I conducted power analyses to determine the probability of discerning meaningful differences in density and survival for hares occupying different stand types. For density, I postulated that foraging lynx likely do not discriminate among stands that differ by only a few hares. However, it seems probable that if hare density in one stand is twice that of another, a lynx would choose the former given the opportunity. Thus, I conducted power calculations to determine the probability of distinguishing differences in densities between 2 stand types in which one had twice the density of hares as the second. Specifically, using the Huggins closed capture model (Huggins 1989, Huggins 1991) in Program MARK, I specified the number of hares (N) present in each of 2 groups (i.e., 2 stand types), allowed capture (p) and recapture (c) probabilities to vary with time but constrained them to be equal and the same for each group, then simulated this scenario 1000 times for a range of realistic capture probabilities. For each simulation I calculated a 95% confidence interval for the mean difference in N̂ between the 2 groups and determined the proportion of all simulations in which this confidence interval did not include zero. This proportion is the power, or probability of discerning a difference between the 2 groups when one actually exists. I compared 2-fold differences in density at the low (5 vs. 10 hares/grid) and high (15 vs. 30 hares/grid) end of the range of hare numbers and I expect to observe (Zahratka 2004). I also simulated the power to detect differences between 17 and 39 hares/grid, corresponding to recently published cut-points for low and high hare densities in the context of lynx conservation (Mills et al. 2005). Given capture/recapture probabilities I observed during winter 2005 (approximately 0.35–0.45), I expect to have reasonable power to detect 2-fold differences in density even if I encounter relatively few hares per grid (Figure 5). I conducted power analyses for survival in a similar manner using the Huggins estimator (Huggins 1989, Huggins 1991) in a robust design framework (Williams et al. 2002:524-556). For this analysis, I specified 3 primary periods (e.g., 3 years) with 5 secondary occasions for each. I established either 30 or 45 hares in each of 2 groups (i.e., pooled an expected 10-15 hares/grid across the 3 grids in a given habitat type), specified a different survival rate for each, and allowed p and c to vary with time but constrained them to be equal and the same for each group as before. I then specified a general model that assumed survival rates varied among groups and a second, reduced model that assumed survival rates were the same for each group. After 1000 simulations under a given scenario of hare numbers, capture probabilities, and survival rates, I conducted a likelihood ratio test between each pair of general and reduced models. As before, I used the proportion of significant tests as an estimate of power to detect differences in survival. 47 �I compared survival rates of 0.4 vs. 0.5, 0.3 vs. 0.5, and 0.2 vs. 0.5. These rates span the range of annual hare survival rates reported in the literature (Dolbeer 1972, Dolbeer and Clark 1975, Griffin 2004). Also, because each comparison is anchored at 0.5, these calculations provide a conservative estimate of power due to the nature of binomial probabilities. That is, I would be more likely to distinguish the difference between 0.1 and 0.2 than between 0.4 and 0.5 even though the difference in both cases is 0.1 because the sampling variance of the estimate for the same sample size is maximal at 0.5 and declines to 0 for survival rates of 0 or 1. Results indicate that I have ≥80% chance of discerning real differences in survival of ≥0.3 (Figure 6), but only 40-65% chance (depending on number of hares captured) of detecting a difference of 0.2, and very little chance of detecting differences smaller than 0.2. However, I plan to combine my telemetry data with my trapping data in the MARK Robust design model using separate groups for each data type. This should enhance my precision and power, thus making the prospect of detecting differences as small as 0.2 a possibility. To complete a power analysis for λ̂ requires running simulations of Pradel models in a robust design framework. This capability is not yet available in Program MARK, so such an analysis has not been completed. Sampling 15 vegetation plots per trapping grid provided reasonably precise characterizations of similar stands in similar locations during a previous study (Zahratka 2004). I trust this level of sampling will be adequate for the present study as well. If not, more plots can be established at a later date given that vegetative characteristics are unlikely to change appreciably over a few years. RESULTS AND DISCUSSION Much of the analysis presented above is not possible or meaningful without several seasons of data, especially the survival, recruitment, and growth rate models. Below, I present a basic summary, relevant observations, and initial density estimates from the inaugural year of this project. I captured 75 hares 166 times during July-September 2006. I captured 99 hares 243 times during January-March 2007 (Table 1). Fourteen of these individuals were captured during both the summer and winter sampling sessions. During summer, I captured over twice as many individuals in small lodgepole stands as in spruce/fir. I captured only a few individuals in medium lodgepole stands. During winter, captures were more evenly distributed among the stands (Table 1). During the initial trapping session of the summer trapping period, 6 hares were captured, handled, and released (seemingly without harm) but were found dead in traps 1-3 days later. I collected the carcasses and submitted them for necropsy. Cause of death was attributed to capture myopathy, which is relatively common in lagomorphs (Laurie Baeten DVM, and Lisa Wolfe DVM, Colorado Division of Wildlife, personal communication). I subsequently altered my trapping protocol to further minimize both the amount of time a hare could be entrapped as well as the handling time at each capture. No trap deaths occurred during the remainder of the sampling season aside from 4 hares that succumbed to predation while inside traps. During the initial 2 trapping sessions of the winter trapping period, 6 more hares were captured, handled, and released multiple times, again with seemingly little adverse reaction, only to be found dead on a subsequent trapping occasion. Several more hares died during the 10-day telemetry session immediately following trapping. These “telemetry deaths” could have been due to natural causes, effects of capture, or a combination of both. Again, carcasses were submitted for necropsy, and again capture myopathy was cited as a potential cause of death. Further examination of the data indicated that hares trapped ≥3 days in a row were much more likely to die in a trap or during telemetry than other hares. Thus, I further modified the trapping protocol by locking traps open on day 3 of the 5-day trapping period so that hares could not be trapped more than 2 days in a row. Additionally, I began providing hay cubes 48 �in the traps as roughage to complement the high quality alfalfa pellets and apples. After implementing these changes, I did not observe any further trap-deaths or telemetry-deaths for the rest of the season. I averaged 9.9 and 6.3 locations per radio-tagged hare during the summer and winter sampling sessions, respectively (Table 2). Thus, “proportion of time on grid,” which is critical to my density estimation procedure, was based on relatively few points per individual for the first 2 sampling periods, and I was unable to attain my goal of 10-20 locations per individual. Following the winter field season, I conducted a series of simulations to examine the effects of sample size on precision of density estimates. I found that 1) the variability between hares (“proportion on grid” ranges from 0.00-1.00) overwhelms the variability within hares (i.e., the binomial variance for proportion of time on grid for any single individual, which decreases as number of locations increases), and 2) given a fixed effort, the variance of the density estimate is minimized by increasing the number of individuals collared as opposed to increasing the number of locations per individual. Thus, it is better to radio-tag more hares and get fewer locations than to tag fewer hares and get more locations. I will continue to deploy as many collars as possible, and will strive for 10-20 locations per individual, but the level of sampling achieved during the first 2 field seasons appears sufficient to detect the large differences in density that occur on the landscape. During summer, density estimates followed hypotheses 1) and 2) above. Specifically, hare density in small lodgepole stands was twice that observed in spruce/fir, which was more than twice that observed in medium lodgepole stands (Figure 7). However, even the relatively high density found in the small lodgepole stands was relatively low compared to densities that have been reported in other parts of hare range (Griffin 2004, Hodges 2000). However, different methods for computing density make this type of comparison difficult. During winter, hare densities remained the same in spruce/fir stands. Hare density in medium lodgpole stands more than doubled, although still remained relatively low compared to other stand types. Density in the small lodgepole stands dropped significantly compared to summer levels and was more variable among replicates. Hare density is likely driven by availability of food and cover. I submit that the interplay between food, cover, and snow depth provides a plausible explanation for the density patterns observed during the first year of this study. Spruce/fir stands probably provide adequate access to both necessities during both summer and winter due to their uneven-aged, multi-layered structure. Medium lodgepole stands, on the other hand, apparently provide very little forage/cover for hares during summer as the canopy in these stands is generally ≥1 meter off the ground. However, in winter, accumulated snow may bring that canopy back into reach for hares. Conversely, small lodgepole stands provided abundant food and cover during summer, but accumulated snow during winter brings hares closer to the crowns of the young trees, which then provide less cover. SUMMARY • The number of snowshoe hares captured, the number of captures, and the number of locations obtained per hare during the first year appeared adequate for attaining the objectives of this study. • Some deaths due to capture myopathy (most likely cause) occurred during initial trapping periods in each sampling season. Changes to the trapping protocol, trapping schedule, and bait provided seem to have alleviated the problem. • Snowshoe hare densities during summer were highest in small lodgepole stands, followed by large spruce/fir and medium lodgepole. During winter, densities in small lodgepole stands dropped and became more variable across replicates. Medium lodgepole stands gained hares. Spruce/fir stands remained at the same density as during summer. 49 �ACKNOWLEDGEMENTS Ken Wilson (CSU), Bill Romme (CSU), Paul Doherty (CSU), Dave Freddy (CDOW), Chad Bishop (CDOW), and Paul Lukacs (CDOW) provided helpful insight on the design of this study. We appreciate the invaluable logistical support provided by Mike Jackson (USFS), Jake Spritzer (USFS), Margie Michaels (CDOW), Gabriele Engler (CSU), Brandon Diamond (CDOW), Chris Parmeter (CDOW), Kathaleen Crane (CDOW), Lisa Wolfe (CDOW), and Laurie Baeten (CDOW). The following hardy individuals collected the hard-won data presented in this report: Braden Burkholder, Matt Cuzzocreo, Brian Gerber, Belita Marine, Adam Behney, Pete Lundberg, Katie Yale, Britta Schielke, Cory VanStratt, Mike Watrobka, Meredith Goss, Sidra Blake, Keith Rutz, Rob Saltmarsh, Jennie Sinclair, and Evan Wilson. Funding was provided by the Colorado Division of Wildlife. LITERATURE CITED AUBRY, K. B., G. M. KOEHLER, AND J. R. SQUIRES. 2000. Ecology of Canada lynx in southern boreal forests. Pages 373-396 in Ruggiero, L. F., K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, and J. R. Squires, editors. Ecology and conservation of lynx in the United States. Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Collins, Colorado, USA. BARBOUR, M. G., J. H. BURK, W. E. PITTS, F. S. GILLIAM, AND M. W. SCHWARTZ. 1999. Terrestrial plant ecology, Addison Wesley Longman, Inc., Menlo Park, California, USA. BARTMANN, R. M. AND G. BYRNE. 2001. Analysis and critique of the 1998 snowshoe hare pellet survey. Colorado Division of Wildlife 20. BEAUVAIS, G. P. 1997. Mammals in fragmented forests in the Rocky Mountains: community structure, habitat selection, and individual fitness. Dissertation, University of Wyoming, Laramie, Wyoming, USA. BUCKLAND, S. T., D. R. ANDERSON, K. P. BURNHAM, J. L. LAAKE, D. L. BORCHERS, AND L. THOMAS. 2001. Introduction to distance sampling: estimating abundance of biological populations, Oxford University Press, Inc., New York, New York, USA. BULL, E. L., T. W. HEATER, A. A. CLARK, J. F. SHEPHERD, AND A. K. BLUMTON. 2005. Influence of precommercial thinning on snowshoe hares. U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station, General Technical Report PNW-RP-562. DOLBEER, R. A. AND W. R. CLARK. 1975. Population ecology of snowshoe hares in the central Rocky Mountains. Journal of Wildlife Management 39:535-549. DOLBEER, R. A. 1972. Population dynamics of snowshoe hare in Colorado. Dissertation, Colorado State University, Fort Collins, Colorado, USA. EFFORD, M. G., D. K. DAWSON, AND C. S. ROBBINS. 2004. DENSITY: software for analysing capturerecapture data from passive detector arrays. Animal Biodiversity and Conservation 27:1-12. ERIKSSON, H. M. 2006. Snowshoe hare densities in post-fire vegetation. Thesis, University of Alaska Fairbanks, Fairbanks, Alaska, USA. FORESMAN, K. R. AND D. E. PEARSON. 1999. Activity patterns of American martens, Martes americana, snowshoe hares, Lepus americanus, and red squirrels, Tamiasciurus hudsonicus, in westcentral Montana. The Canadian Field-Naturalist 113:386-389. GRIFFIN, P. C. 2004. Landscape ecology of snowshoe hares in Montana. Dissertation, University of Montana, Missoula, Montana, USA. HODGES, K. E. 2000. Ecology of snowshoe hares in southern boreal and montane forests. Pages 163-206 in Ruggiero, L. F., K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, and J. R. Squires, editors. Ecology and conservation of lynx in the United States. University Press of Colorado, Boulder, Colorado, USA. HODGES, K. E., C. J. KREBS, D. S. HIK, C. I. STEFAN, E. A. GILLIS, AND C. E. DOYLE. 2001. Snowshoe hare demography. Pages 141-178 in Krebs, C. J., S. Boutin, and R. Boonstra, editors. Ecosystem 50 �dynamics of the boreal forest: the Kluane project. Oxford University Press, New York, New York, USA. HOMYACK, J. A., D. J. HARRISON, AND W. B. KROHN. 2003. Effects of precommercial thinning on select wildlife species in northern Maine, with special emphasis on snowshoe hare. Maine Cooperative Fish and Wildlife Research Unit, Orono, Maine, USA. HOMYACK, J. A., D. J. HARRISON, J. A. LITVAITIS, AND W. B. KROHN. 2006. Quantifying densities of snowshoe hares in Maine using pellet plots. Wildlife Society Bulletin 34:74-80. HUGGINS, R. M. 1989. On the statistical analysis of capture-recapture experiments. Biometrika 76:133140. HUGGINS, R. M. 1991. Some practical aspects of a conditional likelihood approach to capture experiments. Biometrics 47:725-732. KEITH, L. B. 1964. Daily activity pattern of snowshoe hares. Journal of Mammalogy 45:626-627. KOEHLER, G. M. 1990. Snowshoe hare, Lepus americanus, use of forest successional stages and population changes during 1985-1989 in North-central Washington. Canadian Field-Naturalist 105:291-293. KOEHLER, G. M. 1990. Population and habitat characteristics of lynx and snowshoe hares in north central Washington. Canadian Journal of Zoology 68:845-851. KREBS, C. J., B. SCOTT GILBERT, S. BOUTIN, AND E. AL. R. BOONSTRA. 1987. Estimation of snowshoe hare population density from turd transects. Canadian Journal of Zoology 65:565-567. KREBS, C. J., R. BOONSTRA, V. NAMS, M. O'DONOGHUE, K. E. HODGES, AND S. BOUTIN. 2001. Estimating snowshoe hare population density from pellet plots: a further evaluation. Canadian Journal of Zoology 79:1-4. LITVAITIS, J. A., J. A. SHERBURNE, AND J. A. BISSONETTE. 1985. Influence of understory characteristics on snowshoe hare habitat use and density. Journal of Wildlife Management 49:866-873. MARQUES, F. F. C., S. T. BUCKLAND, D. GOFFIN, C. E. DIXON, D. L. BORCHERS, B. A. MAYLE, AND A. J. PEACE. 2001. Estimating deer abundance from line transect surveys of dung: sika deer in southern Scotland. Journal of Applied Ecology 38:349-363. MCKELVEY, K. S., G. W. MCDANIEL, L. S. MILLS, AND P. C. GRIFFIN. 2002. Effects of plot size and shape on pellet density estimates for snowshoe hares. Wildlife Society Bulletin 30:751-755. MECH, L. D., K. L. HEEZEN, AND D. B. SINIFF. 1966. Onset and cessation of activity in cottontail rabbit and snowshoe hares in relation to sunset and sunrise. Animal Behaviour 14:410-413. MILLER, M. A. 2005. Snowshoe hare habitat relationships in northwest Colorado. Thesis, Colorado State University, Fort Collins, Colorado, USA. MILLS, L. S., P. C. GRIFFIN, K. E. HODGES, K. MCKELVEY, L. RUGGIERO, AND T. ULIZIO. 2005. Pellet count indices compared to mark-recapture estimates for evaluating snowshoe hare density. Journal of Wildlife Management 69:1053-1062. MURRAY, D., E. ELLSWORTH, AND A. ZACK. 2005. Assessment of potential bias with snowshoe hare fecal pellet-plot counts. Journal of Wildlife Management 69:385-395. MURRAY, D. L., J. D. ROTH, E. ELLSWORTH, A. J. WIRSING, AND T. D. STEURY. 2002. Estimating lowdensity snowshoe hare populations using fecal pellet counts. Canadian Journal of Zoology 80:771-781. NICHOLS, J. D. AND J. E. HINES. 2002. Approaches for the direct estimation of lambda, and demographic contributions to lambda, using capture-recapture data. Journal of Applied Statistics 29:539-568. NICHOLS, J. D., J. E. HINES, J. D. LEBRETON, AND R. PRADEL. 2000. Estimation of contributions to population growth: a reverse-time capture-recapture approach. Ecology 81:3362-3376. NUDDS, T. D. 1977. Quantifying the vegetative structure of wildlife cover. Wildlife Society Bulletin 5:113-117. OTIS, D. L., K. P. BURNHAM, G. C. WHITE, AND D. R. ANDERSON. 1978. Statisical inference from capture data on closed animal populations. Wildlife Monographs 62: PARMENTER, R. R., T. L. YATES, D. R. ANDERSON, K. P. BURNHAM, J. L. DUNNUM, A. B. FRANKLIN, M. T. FRIGGENS, B. C. LUBOW, M. MILLER, G. S. OLSON, C. A. PARMENTER, J. POLLARD, E. 51 �REXSTAD, T. SHENK, T. R. STANLEY, AND G. C. WHITE. 2003. Small-mammal density estimation: a field comparison of grid-based vs. web-based density estimators. Ecological Monographs 73:126. PRADEL, R. 1996. Utilization of capture-mark-recapture for the study of recruitment and population growth rate. Biometrics 52:703-709. PRUGH, L. R. AND C. J. KREBS. 2004. Snowshoe hare pellet-decay rates and aging in different habitats. Wildlife Society Bulletin 32:386-393. RUEDIGER, B., J. CLAAR, S. GNIADEK, B. HOLT, LEWIS LYLE, S. MIGHTON, B. NANEY, G. PATTON , T. RINALDI, J. TRICK, A. VENDEHEY, F. WAHL, N. WARREN, D. WENGER, AND A. WILLIAMSON. 2000. Canada lynx conservation assessment and strategy. U.S. Department of Agriculture, Forest Service, U.S. Department of Interior, Fish and Wildlife Service, Bureau of Land Management, National Park Service R1-00-53 U.S. Department of Agriculture, Forest Service, U.S. Department of Interior, Fish and Wildlife Service, Bureau of Land Management, National Park Service, Missoula, Montana, USA. RUGGIERO, L. F., K. B. AUBRY, S. W. BUSKIRK, G. M. KOEHLER, C. J. KREBS, K. S. MCKELVEY, AND J. R. SQUIRES. 2000. The scientific basis for lynx conservation: qualified insights. Pages 443-454 in Ruggiero, L. F., K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, and J. R. Squires, editors. Ecology and conservation of lynx in the United States. Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Collins, Colorado, USA. SAMUEL, M. D. AND M. R. FULLER. 1996. Wildlife radiotelemetry. Pages 370-418 in Bookhout, T. A., editors. Research and Management Techniques for Wildlife and Habitats. Allen Press, Inc., Lawrence, Kansas, USA. SHENK, T. M. 2005. General locations of lynx (Lynx canadensis) reintroduced to southwestern Colorado from February 4, 1999 through February 1, 2005. Colorado Division of Wildlife Colorado Division of Wildlife, Fort Collins, Colorado, USA. __________. 2006. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report July: 1-45. Colorado Division of Wildlife, Fort Collins, Colorado SULLIVAN, T. P. AND D. S. SULLIVAN. 1988. Influence of stand thinning on snowshoe hare population dynamics and feeding damage in lodgepole pine forest. Journal of Applied Ecology 25:791-805. VAN HORNE, B. 1983. Density as a misleading indicator of habitat quality. Journal of Wildlife Management 47:893-901. WHITE, G. C., D. R. ANDERSON, K. P. BURNHAM, AND D. L. OTIS. 1982. Capture-recapture and removal methods for sampling closed populations. Los Alamos National Laboratory, Los Alamos, New Mexico, USA. WHITE, G. C. AND R. A. GARROTT. 1990. Analysis of wildlife radio-tracking data, Academic Press, San Diego, California, USA. WILLIAMS, B. K., J. D. NICHOLS, AND M. J. CONROY. 2002. Analysis and management of animal populations, Academic Press, San Diego, California, USA. WILSON, K. R. AND D. R. ANDERSON. 1985. Evaluation of two density estimators of small mammal population size. Journal of Mammalogy 66:13-21. WOLFE, M. L., N. V. DEBYLE, C. S. WINCHELL, AND T. R. MCCABE. 1982. Snowshoe hare cover relationships in northern Utah. Journal of Wildlife Management 46:662-670. WOLFF, J. O. 1980. The role of habitat patchiness in the population dynamics of snowshoe hares. Ecological Monographs 50:111-130. ZAHRATKA, J. L. 2004. The population and habitat ecology of snowshoe hares (Lepus americanus) in the southern Rocky Mountains. Thesis, The University of Wyoming, Laramie, Wyoming, USA. ZIMMERMAN, G. M., H. GOETZ, AND P. W. JR. MIELKE. 1985. Use of an improved statistical method for group comparisons to study effects of prairie fire. Ecology 66:606-611. Prepared by _________________________________________________ Jacob S. Ivan, Graduate Student, Colorado State University 52 �Table 1. Number of snowshoe hares (Lepus americanus) captured during 5-day trapping sessions conducted during July-September 2006 and January-March 2007 on 3 medium lodgepole, 3 spruce/fir, and 6 small lodgepole stands on the Gunnison National Forest, Taylor Park and Pitkin, Colorado. ____________________________________________________________________________________ Number of Hares Captured (Total Captures) ___________________________________________________________ Summer 2006 Winter 2007 Both Summer and Winter ____________________________________________________________________________________ Medium Lodgepole 3 24 2 Small Lodgepole 50 40 10 Spruce/Fir 22 35 2 ____________________________________________________________________________________ Table 2. Number of snowshoe hares (Lepus americanus) radio-collared and tracked during 10-day sessions immediately following 5-day trapping periods July-September 2006 and January-March 2007 on the Gunnison National Forest, Taylor Park and Pitkin, Colorado. ____________________________________________________________________________________ Summer 2006 Winter 2007 ____________________________________________________________________________________ Number of Hares Collared 41 79 Number of Locations 407 510 Number of Locations/Hare 9.9 6.5 ____________________________________________________________________________________ Figure 1. Purported high quality snowshoe hare habitat in Colorado. From left to right: small lodgepole pine, medium lodgepole pine, and large Engelmann spruce/subalpine fir. 53 �Figure 2. Study area near Taylor Park and Pitkin, Colorado including medium lodgepole (squares), small lodgepole (circles), and spruce/fir (triangles) stands selected for mark-recapture sampling. 54 �Jul Aug Sep 1 Jan 1 2 3 4 2 3 4 5 6 7 8 5 6 7 8 9 10 11 9 10 11 12 13 14 15 12 13 14 15 16 17 18 16 17 18 19 20 21 22 19 20 21 22 23 24 25 23 24 25 26 27 28 29 26 27 28 29 30 31 1 30 31 1 2 3 4 5 2 3 4 5 6 7 8 Feb 6 7 8 9 10 11 12 9 10 11 12 13 14 15 13 14 15 16 17 18 19 16 17 18 19 20 21 22 20 21 22 23 24 25 26 23 24 25 26 27 28 1 27 28 29 30 31 1 2 2 3 4 5 6 7 8 3 4 5 6 7 8 9 9 10 11 12 13 14 15 10 11 12 13 14 15 16 16 17 18 19 20 21 22 17 18 19 20 21 22 23 23 24 25 26 27 28 29 24 25 26 27 28 29 30 30 31 Mar Figure 3. Approximate annual data collection schedule for trapping (�) and telemetry (�). Dates and weeks will change depending on calendar year and pay schedule. During telemetry work, the 6-person crew will be divided into 2 teams, only one of which will be working at any given time. Monthly locations on radio-collared hares will also be collected in the interim between the intensive sampling periods indicated here. Figure 4. 15 trap locations (•) on 7 × 12 trapping grid where vegetation will be sampled by measuring stem density horizontal cover, and canopy cover at the 25 points on each 5 × 5 subgrid (inset). In addition, basal area will be measured at the trap location (�) on which each of the 15 subgrids are centered. 55 �Density Power Analysis % Non-overlapping 95% CIs 100 90 80 70 60 50 40 30 20 10 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0 capture/recapture probability N=5 vs. N=10 N=15 vs. N=30 N=17 vs. N=39 Figure 5. Power for distinguishing differences in snowshoe hare density between 2 habitat types when a difference actually exists. Gray area indicates the capture probability realized by the 3rd day of trapping during a pilot study in winter 2005. N indicates number of hares per grid, a range of roughly 0.1 (N = 5) to 0.7 hares/ha (N = 39). Survival Power Analysis (N = 45) 100 100 90 80 90 80 % Significant LR Tests 70 60 50 40 30 20 10 0 70 60 50 40 30 20 Capture/Recapture Probability 0.2 vs. 0.5 0.3 vs. 0.5 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.10 0.60 0.55 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 10 0 0.15 % Significant LR Tests Survival Power Analysis (N = 30) Capture/Recapture Probability 0.4 vs. 0.5 0.2 vs. 0.5 0.3 vs. 0.5 0.4 vs. 0.5 Figure 6. Power, or probability of distinguishing differences in snowshoe hare survival between 2 habitat types when differences actually exist. N = 30 (left) and N = 45 (right) correspond to reasonable estimates of the number of hares I expect to capture in each habitat type. Gray area indicates the capture probability realized by the 3rd day of trapping during a pilot study in winter 2005. 56 �Figure 7. Snowshoe hare density and 95% confidence intervals in 3 types of stands in central Colorado as determined by mark-recapture with telemetry augmentation, July-September 2006 and January-March 2007. 57 �Colorado Division of Wildlife July 2007- June 2008 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 0670 1 Federal Aid Project No. N/A : : : : : Division of Wildlife Mammals Research Lynx Conservation Post-Release Monitoring of Lynx Reintroduced to Colorado Period Covered: July 1, 2007 - June 30, 2008 Author: T. M. Shenk Personnel: L. Baeten, B. Diamond, R. Dickman, D. Freddy, L. Gepfert, J. Ivan, R. Kahn, A. Keith, G. Merrill, B. Smith, T. Spraker, S. Wait, S. Waters, L. Wolfe All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT In an effort to establish a viable population of Canada lynx (Lynx canadensis) in Colorado, the Colorado Division of Wildlife (CDOW) initiated a reintroduction effort in 1997 with the first lynx released in February 1999. From 1999-2007, 218 wild-caught lynx from Canada and Alaska were released in Colorado. We documented survival, movement patterns, reproduction, and landscape habitatuse through aerial (n = 10,935) and satellite (n = 26,082) tracking. Most lynx remained near the core release area in southwestern Colorado. From 1999-August 2008, there were 112 mortalities of released adult lynx. Approximately 30.4% were either human-induced or likely human-induced through either collisions with vehicles or gunshot. Starvation and disease/illness accounted for 18.8% of the deaths while 36.6% of the deaths were from unknown causes. Of these mortalities, 26.8% occurred outside of Colorado. Monthly mortality rate was lower inside the study area than outside, and slightly higher for male than for female lynx, although 95% confidence intervals for sexes overlapped. Mortality was higher immediately after release (first month = 0.0368 [SE = 0.0140] inside the study area, and 0.1012 [SE = 0.0359] outside the study area), and then decreased according to a quadratic trend over time. Reproductive females had the smallest 90% utilization distribution home ranges ( x = 75.2 km2, SE = 15.9 km2), followed by attending males ( x = 102.5 km2, SE = 39.7 km2) and non-reproductive animals ( x = 653.8 km2, SE = 145.4 km2). Reproduction was first documented in 2003 with subsequent successful reproduction in 2004, 2005 and 2006. No dens were documented in 2007 or 2008. From snow-tracking, the primary winter prey species (n = 548 kills) were snowshoe hare (Lepus americanus, annual x = 73.3%, SE = 4.7, n = 10) and red squirrel (Tamiasciurus hudsonicus, annual x = 18.2%, SE = 4.2, n = 10); other mammals and birds formed a minor part of the winter diet. Lynx use-density surfaces were generated to illustrate relative use of areas throughout Colorado. Within the areas of high use in southwestern Colorado, site-scale habitat use, documented through snow-tracking, supports mature 1 �Engelmann spruce (Picea engelmannii)-subalpine fir (Abies lasiocarpa) forest stands with 42-65% canopy cover and 15-20% conifer understory cover as the most commonly used areas in southwestern Colorado. Little difference in aspect (slight preference for north-facing slopes), slope ( x = 15.7°) or elevation ( x = 3173 m) were detected for long beds, travel and kill sites (n = 1841). Den sites (n = 37) however, were located at higher elevations ( x = 3354 m, SE = 31 m) on steeper ( x = 30°, SE = 2°) and more commonly north-facing slopes with a dense understory of coarse woody debris. Two years of a study to evaluate snowshoe hare densities, demography and seasonal movement patterns among small and medium tree-sized lodgepole pine stands and mature spruce/fir stands have been completed in 2006-2008 and will continue through 2009 (see Appendix I of this report). Results to date have demonstrated that CDOW has developed lynx release protocols that ensure high initial post-release survival followed by high long-term survival, site fidelity, reproduction and recruitment of Colorado-born lynx into the Colorado breeding population. What is yet to be demonstrated is whether Colorado can support sufficient recruitment to offset annual mortality for a viable lynx population over time. Monitoring continues in an effort to document such viability. 2 �WILDLIFE RESEARCH REPORT POST RELEASE MONITORING OF LYNX (LYNX CANADENSIS) REINTRODUCED TO COLORADO TANYA M. SHENK P. N. OBJECTIVE The initial post-release monitoring of Canada lynx (Lynx canadensis) reintroduced into Colorado will emphasize 5 primary objectives: 1. Assess and modify release protocols to ensure the highest probability of survival for each lynx released. 2. Obtain regular locations of released lynx to describe general movement patterns and habitats used by lynx. 3. Determine causes of mortality in reintroduced lynx. 4. Estimate survival of lynx reintroduced to Colorado. 5. Estimate reproduction of lynx reintroduced to Colorado. Three additional objectives will be emphasized after lynx display site fidelity to an area: 6. Refine descriptions of habitats used by reintroduced lynx. 7. Refine descriptions of daily and overall movement patterns of reintroduced lynx. 8. Describe hunting habits and prey of reintroduced lynx. Information gained to achieve these objectives will form a basis for the development of lynx conservation strategies in the southern Rocky Mountains. SEGMENT OBJECTIVES 1. Complete winter 2007-08 field data collection on lynx habitat use at the landscape scale, hunting behavior, diet, mortalities, and movement patterns. 2. Complete winter 2007-08 lynx trapping field season to collar Colorado born lynx and re-collar adult lynx. 3. Complete spring 2008 field data on lynx reproduction. 4. Summarize and analyze data and publish information as Progress Reports, peer-reviewed manuscripts for appropriate scientific journals, or CDOW technical publications. 5. Complete the second year of field work to evaluate snowshoe hare (Lepus americanus) densities, demography and seasonal movement patterns among small and medium tree-sized lodgepole pine stands and mature spruce/fir stands (see Appendix I). INTRODUCTION The Canada lynx occurs throughout the boreal forests of northern North America. Colorado represents the southern-most historical distribution of lynx, where the species occupied the higher elevation, montane forests in the state. Little was known about the population dynamics or habitat use of this species in their southern distribution. Lynx were extirpated or reduced to a few animals in the state by the late 1970‘s due, most likely, to predator control efforts such as poisoning and trapping. Given the isolation of Colorado to the nearest northern populations, the CDOW considered reintroduction as the only option to attempt to reestablish the species in the state. 3 �A reintroduction effort was begun in 1997, with the first lynx released in Colorado in 1999. To date, 218 wild-caught lynx from Alaska and Canada have been released in southwestern Colorado. The goal of the Colorado lynx reintroduction program is to establish a self-sustaining, viable population of lynx in this state. Evaluation of incremental achievements necessary for establishing viable populations is an interim method of assessing if the reintroduction effort is progressing towards success. There are 7 critical criteria for achieving a viable population: 1) development of release protocols that lead to a high initial post-release survival of reintroduced animals, 2) long-term survival of lynx in Colorado, 3) development of site fidelity by the lynx to areas supporting good habitat in densities sufficient to breed, 4) reintroduced lynx must breed, 5) breeding must lead to reproduction of surviving kittens 6) lynx born in Colorado must reach breeding age and reproduce successfully, and 7) recruitment must equal or be greater than mortality over an extended period of time. The post-release monitoring program for the reintroduced lynx has 2 primary goals. The first goal is to determine how many lynx remain in Colorado and their locations relative to each other. Given this information and knowing the sex of each individual, we can assess whether these lynx can form a breeding core from which a viable population might be established. From these data we can also describe general movement patterns and habitat use. The second primary goal of the monitoring program is to estimate survival of the reintroduced lynx and, where possible, determine causes of mortality for reintroduced lynx. Such information will help in assessing and modifying release protocols and management of lynx once they have been released to ensure their highest probability of survival. Additional goals of the post-release monitoring program for lynx reintroduced to the southern Rocky Mountains included refining descriptions of habitat use and movement patterns and describing successful hunting habitat once lynx established home ranges that encompassed their preferred habitat. Specific objectives for the site-scale habitat data collection include: 1) describe and quantify site-scale habitat use by lynx reintroduced to Colorado, 2) compare site-scale habitat use among types of sites (e.g., kills vs. long-duration beds), and 3) compare habitat features at successful and unsuccessful snowshoe hare chases. Documenting reproduction is critical to the success of the program and lynx are monitored intensively to document breeding, births, survival and recruitment of lynx born in Colorado. Site-scale habitat descriptions of den sites are also collected and compared to other sites used by lynx. The program will also investigate the ecology of snowshoe hare in Colorado. A study comparing snowshoe hare densities among mature stands of Engelmann spruce (Picea engelmannii)/subalpine fir (Abies lasiocarpa), lodgepole pine (Pinus contorta) and Ponderosa pine (Pinus ponderosa) was completed in 2004 with highest hare densities found in Engelmann spruce/subalpine fir stands and no hares found in Ponderosa pine stands. A study to evaluate the importance of young, regenerating lodgepole pine and mature Engelmann spruce/subalpine fir stands in Colorado by examining density and demography of snowshoe hares that reside in each was initiated in 2005 and will continue through 2009 (see Appendix I). Lynx is listed as threatened under the Endangered Species Act (ESA) of 1973, as amended (16 U. S. C. 1531 et. seq.)(U. S. Fish and Wildlife Service 2000). Colorado is included in the federal listing as lynx habitat. Thus, an additional objective of the post-release monitoring program is to develop conservation strategies relevant to lynx in Colorado. To develop these conservation strategies, information specific to the ecology of the lynx in its southern Rocky Mountain range, such as habitat use, movement patterns, mortality factors, survival, and reproduction in Colorado is needed. 4 �STUDY AREA Byrne (1998) evaluated five areas within Colorado as potential lynx habitat based on (1) relative snowshoe hare densities (Bartmann and Byrne 2001), (2) road density, (3) size of area, (4) juxtaposition of habitats within the area, (5) historical records of lynx observations, and (6) public issues. Based on results from this analysis, the San Juan Mountains of southwestern Colorado were selected as the core reintroduction area, and where all lynx were reintroduced. Wild Canada lynx captured in Alaska, British Columbia, Manitoba, Quebec and Yukon were transported to Colorado and held at The Frisco Creek Wildlife Rehabilitation Center located within the reintroduction area prior to release. Post-release monitoring efforts were focused in a 20,684 km2 study area which included the core reintroduction area, release sites and surrounding high elevation sites (> 2,591 m). The area encompassed the southwest quadrant of Colorado and was bounded on the south by New Mexico, on the west by Utah, on the north by interstate highway 70, and on the east by the Sangre de Cristo Mountains (Figure 1). Southwestern Colorado is characterized by wide plateaus, river valleys, and rugged mountains that reach elevations over 4,200 m. Engelmann spruce/subalpine fir is the most widely distributed coniferous forest type within the study area. The lynx-established core area is roughly bounded by areas used by lynx in the Taylor Park/Collegiate Peak areas in central Colorado and includes areas of continuous use by lynx, including areas used during breeding and denning (Figure 1). METHODS REINTRODUCTION Effort Wild Canada lynx were captured in Alaska, British Columbia, Manitoba, Quebec and Yukon and transported to Colorado where they were held at the Frisco Creek Wildlife Rehabilitation Center prior to release. All lynx releases were conducted under the protocols found to maximize survival (see Shenk 2001). Estimated age, sex and body condition were ascertained and recorded for each lynx prior to release (see Wild 1999). Lynx were transported from the rehabilitation facility to their release site in individual cages. Specific release site locations were recorded in Universal Transverse Mercator (UTM) coordinates and identification of all lynx released at the same location, on the same day, was recorded. Behavior of the lynx on release and movement away from the release site were documented. Movement, Distribution and Relative Use of Areas by Lynx To monitor lynx movements and thus determine distribution and relative use of areas all released lynx were fitted with radio collars. All lynx released in 1999 were fitted with TelonicsTM radio-collars. All lynx released since 1999, with the exception of 5 males released in spring 2000, were fitted with SirtrackTM dual satellite/VHF radio-collars. These collars have a mortality indicator switch that operated on both the satellite and VHF mode. The satellite component of each collar was programmed to be active for 12 hours per week. The 12-hour active periods for individual collars were staggered throughout the week. Signals from the collars allowed for locations of the animals to be made via Argos, NASA, and NOAA satellites. The location information was processed by ServiceArgos and distributed to the CDOW through e-mail messages. Datasets.-- To determine recent (post-reintroduction) movement and distribution of lynx reintroduced, born or initially trapped in Colorado and relative use of areas by these lynx, regular locations of lynx were collected through a combination of aerial and satellite tracking. Locations were recorded and general habitat descriptions for each aerial location was recorded. The first dataset of lynx locations included all locations obtained from daytime flights conducted with a Cessna 185 or similar aircraft to locate lynx by their VHF collar transmitters (hereafter aerial locations). VHF transmitters have been used on lynx since the first lynx were released in February 1999. The second type of lynx location 5 �data was collected via satellite from the satellite collar transmitters placed on the lynx (hereafter satellite locations). Satellite transmitter collars were first used for lynx in April 2000. These satellite collars also contained a VHF transmitter which also allowed locating lynx from the air or ground. All locations were recorded in Universal Transverse Mercator (UTM) coordinates using the CONUS NAD27 datum. Flights to obtain lynx aerial locations were typically conducted on a weekly basis throughout most summer and winter months and twice a week during the den search field season (May 15 – June 30), depending on weather and availability of planes and pilots. Flights were typically concentrated in the high elevation (> 2700 m) southwest quadrant of Colorado which encompasses the core lynx release and research area (Figure 1). Flights during the den seasons were conducted to obtain locations on all female lynx within the state wearing an active VHF transmitter. VHF transmitters were outfitted with sufficient batteries to last 60 months. The satellite transmitters were designed to provide locations on a weekly basis with sufficient batteries to last for 18 months. Lynx may not be exhibiting typical behavior or habitat use within the first few months after their release in Colorado. Therefore, a subset of each of the aerial and satellite datasets was created that eliminated the first 180 days (approximately 6 months) of locations obtained for each lynx immediately after their initial release. As a result, the truncated aerial location dataset contained lynx locations from September 1999 through March 2007 while the truncated satellite location dataset began October 2000 and extended through March 2007. Accuracy of both aerial and satellite locations varied with the environmental conditions at the time the location was obtained. Accuracy of aerial locations was influenced by weather with accuracy ranging from 50 - 500 meters. Satellite location accuracy was also influenced by atmospheric conditions and position of the satellites. Satellite location accuracy ranged from 150 meters -10 km. Movement and Distribution.-- To document all known lynx locations maps were generated with all aerial and satellite locations displayed. Due to lynx movements outside of Colorado, particularly into the states of New Mexico, Utah and Wyoming we further evaluated lynx use throughout those three states, as well as the data would allow. All individual lynx located at least once in these 3 states (nontruncated datasets) were identified and tallied for each year. To document consistency and known use of these states after the initial effect of being reintroduced was minimized (i.e., 180 days post-release), each individual lynx located at least once in these states from the truncated datasets were identified and tallied. Relative Use.-- To document relative use of areas by lynx, 90% kernel use-density surfaces were calculated for truncated satellite and aerial lynx locations using the ArcGIS Spatial Analyst Kernel Density Tool. Due to differences in data collection frequency and accuracy between datasets, the truncated satellite and truncated aerial data were analyzed separately for generating the lynx use-density surfaces. These use-density surfaces fit a smoothly curved surface over each lynx location. The surface value was highest at the location of the point and diminished with increasing distance from the point. A fixed kernel was used with a smoothing parameter of 5 km, reaching 0 at the search radius distance from the point. Only a circular neighborhood was possible. The volume under the surface equaled the total value for the point. The use-density at each output GIS raster cell was calculated by adding the values of all the kernel surfaces from all the lynx point locations that overlaid each raster cell center. The kernel function was based on the quadratic kernel function described in Silverman (1986, p. 76, equation 4.5). The use-density surfaces were calculated at 100 m resolution. To enhance graphic displays of higher usedensity areas, density values representing single locations were not displayed. 6 �Home Range Annual home ranges were calculated as a 95% utilization distribution using a kernel home-range estimator for each lynx we had at least 30 locations for within a year. A year was defined as March 15 – March 14 of the following year. Locations used in the analyses were collected from September 1999 – January 2006 and all locations obtained for an individual during the first six months after its release were eliminated from any home range analyses as it was assumed movements of lynx initially post-release may not be representative of normal habitat use. Locations were obtained either through aerial VHF surveys or locations or the midpoint (ArcView Movement Extension) of all high quality (accuracy rating of 01km) satellite locations obtained within a single 24-hour period. All locations used within a single home range analysis were taken a minimum of 24 hours apart. Home range estimates were classified as being for a reproductive or non-reproductive animal. A reproductive female was defined as one that had kittens with her; a reproductive male was defined as a male whose movement patterns overlapped that of a reproductive female. If a litter was lost within the defined year a home range described for a reproductive animal were estimated using only locations obtained while the kittens were still with the female. Survival Multi-state mark-recapture models were used to estimate monthly mortality rates and described in detail in Devineau et al. 2008 (in review). This approach accommodated missing data and allowed exploration of factors possibly affecting lynx survival such as sex, time spent in pre-release captivity, movement patterns, and origin. Mortality Factors When a mortality signal (75 beats per minute [bpm] vs. 50 bpm for the Telonics™ VHF transmitters, 20 bpm vs. 40 bpm for the Sirtrack™ VHF transmitters, 0 activity for Sirtrack™ PTT) was heard during either satellite, aerial or ground surveys, the location (UTM coordinates) was recorded. Ground crews then located and retrieved the carcass as soon as possible. The immediate area was searched for evidence of other predators and the carcass photographed in place before removal. Additionally, the mortality site was described and habitat associations and exact location were recorded. Any scat found near the dead lynx that appeared to be from the lynx was collected. All carcasses were transported to the Colorado State University Veterinary Teaching Hospital (CSUVTH) for a post mortem exam to 1) determine the cause of death and document with evidence, 2) collect samples for a variety of research projects, and 3) archive samples for future reference (research or forensic). The gross necropsy and histology were performed by, or under the lead and direct supervision of a board certified veterinary pathologist. At least one research personnel from the CDOW involved with the lynx program was also present. The protocol followed standard procedures used for thorough post-mortem examination and sample collection for histopathology and diagnostic testing (see Shenk 1999 for details). Some additional data/samples were routinely collected for research, forensics, and archiving. Other data/samples were collected based on the circumstances of the death (e.g., photographs, video, radiographs, bullet recovery, samples for toxicology or other diagnostic tests, etc.). From 1999–2004 the CDOW retained all samples and carcass remains with the exception of tissues in formalin for histopathology, brain for rabies exam, feces for parasitology, external parasites for ID, and other diagnostic samples. Since 2005 carcasses are disposed of at the CSUVTH with the exception of the lower canine, fecal samples, stomach content samples and tissue or bone marrow samples to be delivered by CDOW to the Center for Disease control for plague testing. The lower canine, from all carcasses, is sent to Matson Labs (Missoula, Montana) for aging and the fecal and stomach content samples are evaluated for diet. 7 �Reproduction Females were monitored for proximity to males during each breeding season. We defined a possible mating pair as any male and female documented within at least 1 km of each other in breeding season through either flight data or snow-tracking data. Females were then monitored for site fidelity to a given area during each denning period of May and June. Each female that exhibited stationary movement patterns in May or June were closely monitored to locate possible dens. Dens were found when field crews walked in on females that exhibited virtually no movement for at least 10 days from both aerial and ground telemetry. Kittens found at den sites were weighed, sexed and photographed. Each kitten was uniquely marked by inserting a sterile passive integrated transponder (PIT, Biomark, Inc., Boise, Idaho, USA) tag subcutaneously between the shoulder blades. Time spent at the den was minimized to ensure the least amount of disturbance to the female and the kittens. Weight, PIT-tag number, sex and any distinguishing characteristics of each kitten was also recorded. Beginning in 2005, blood and saliva samples were collected and archived for genetic identification. During the den site visits, den site location was recorded as UTM coordinates. General vegetation characteristics, elevation, weather, field personnel, time at the den, and behavioral responses of the kittens and female were also recorded. Once the females moved the kittens from the natal den area, den sites were visited again and site-specific habitat data were collected (see Habitat Use section below). Captures Captures were attempted for either lynx that were in poor body condition or lynx that needed to have their radio-collars replaced due to failed or failing batteries or to radio-collar kittens born in Colorado once they reached at least 10-months of age when they were nearly adult size. Methods of recapture included 1) trapping using a Tomahawk™ live trap baited with a rabbit and visual and scent lures, 2) calling in and darting lynx using a Dan-Inject CO2 rifle, 3) custom box-traps modified from those designed by other lynx researchers (Kolbe et al. 2003) and 4) hounds trained to pursue felids were also used to tree lynx and then the lynx was darted while treed. Lynx were immobilized either with Telazol (3 mg/kg; modified from Poole et al. 1993 as recommended by M. Wild, DVM) or medetomidine (0.09mg/kg) and ketamine (3 mg/kg; as recommended by L. Wolfe, DVM)) administered intramuscularly (IM) with either an extendible pole-syringe or a pressurized syringe-dart fired from a Dan-Inject air rifle. Immobilized lynx were monitored continuously for decreased respiration or hypothermia. If a lynx exhibited decreased respiration 2mg/kg of Dopram was administered under the tongue; if respiration was severely decreased, the animal was ventilated with a resuscitation bag. If medetomidine/ketamine were the immobilization drugs, the antagonist Atipamezole hydrochloride (Antisedan) was administered. Hypothermic (body temperature < 95o F) animals were warmed with hand warmers and blankets. While immobilized, lynx were fitted with replacement SirtrackTM VHF/satellite collar and blood and hair samples were collected. Once an animal was processed, recovery was expedited by injecting the equivalent amount of the antagonist Antisedan IM as the amount of medetomidine given, if medetomodine/ketemine was used for immobilization. Lynx were then monitored while confined in the box-trap until they were sufficiently recovered to move safely on their own. No antagonist is available for Telezol so lynx anesthetized with this drug were monitored until the animal recovered on its own in the box-trap and then released. If captured and in poor body condition, lynx were anesthetized with either Telezol (2 mg/kg) or medetomodine/ketemine and returned to the Frisco Creek Wildlife Rehabilitation Center for treatment. HABITAT USE Gross habitat use was documented by recording canopy vegetation at aerial locations. More 8 �refined descriptions of habitat use by reintroduced lynx were obtained through following lynx tracks in the snow (i.e., snow-tracking) and site-scale habitat data collection conducted at sites found through this method to be used by lynx. See Shenk (2006) for detailed methodologies. DIET AND HUNTING BEHAVIOR Winter diet of reintroduced lynx was estimated by documenting successful kills through snowtracking. Prey species from failed and successful hunting attempts were identified by either tracks or remains. Scat analysis also provided information on foods consumed. Scat samples were collected wherever found and labeled with location and individual lynx identification. Only part of the scat was collected (approximately 75%); the remainder was left in place in the event that the scat was being used by the animal as a territory mark. Site-scale habitat data collected for successful and unsuccessful snowshoe hare kills were compared. SNOWSHOE HARE ECOLOGY To further our understanding of snowshoe hare ecology in Colorado, a study was conducted comparing snowshoe hare densities among mature stands of Engelmann spruce/subalpine fir, lodgepole pine (Pinus contorta) and Ponderosa pine (Pinus ponderosa). The highest hare densities were found in Engelmann spruce/subalpine fir stands and no hares found in Ponderosa pine stands (Zahratka and Shenk 2008). A second study was initiated in 2005 to evaluate the importance of young, regenerating lodgepole pine and mature Engelmann spruce / subalpine fir stands in Colorado by examining density and demography of snowshoe hares that reside in each (Ivan 2005). Specifically, this study was designed to evaluate small and medium lodgepole pine stands and large spruce/fir stands where the classes ― small‖, ―medium‖, and ―l arge‖ refer to the diameter at breast height (dbh) of overstory trees as defined in the United States Forest Service R2VEG Database (small = 2.54 12.69 cm dbh, medium = 12.70 22.85 cm, and large = 22.86 40.64 cm dbh; J. Varner, United States Forest Service, personal communication). The study design was also developed to identify which of the numerous hare density-estimation procedures available perform accurately and consistently using an innovative, telemetry augmentation approach as a baseline. In addition, movement patterns and seasonal use of deciduous cover types such as riparian willow were assessed. Finally, the study was designed to further expound on the relationship between density, demography, and stand-type by examining how snowshoe hare density and demographic rates vary with specific vegetation, physical, and landscape characteristics of a stand. RESULTS REINTRODUCTION Effort From 1999 through 2006, 218 wild-caught lynx were reintroduced into southwestern Colorado (Table 1). No lynx were released in 2007 or 2008. All lynx were released with either VHF or dual VHF/satellite radio collars so they could be monitored for movement, reproduction and survival. The CDOW does not plan to release any additional lynx in 2009. Movement Patterns and Distribution Numerous travel corridors were used repeatedly by more than one lynx. These travel corridors include the Cochetopa Hills area for northerly movements, the Rio Grande Reservoir-SilvertonLizardhead Pass for movements to the west, and southerly movements down the east side of Wolf Creek Pass to the southeast through the Conejos River Valley. Lynx appear to remain faithful to an area during winter months, and exhibit more extensive movements away from these areas in the summer. 9 �A total of 10,935 aerial and 26,082 satellite locations were obtained from the 218 reintroduced lynx, radio-collared Colorado kittens (n = 14) and unmarked lynx captured in Colorado (n = 2) as of August 27, 2008. The majority of these locations were in Colorado (Figure 2). Some reintroduced lynx dispersed outside of Colorado into Arizona, Idaho, Iowa, Kansas, Montana, Nebraska, Nevada, New Mexico, South Dakota, Utah and Wyoming (Figure 2). The majority of surviving lynx from the reintroduction effort currently continue to use high elevation (> 2900 m), forested terrain in an area bounded on the south by New Mexico north to Independence Pass, west as far as Taylor Mesa and east to Monarch Pass. Most movements away from the Core Release Area were to the north. Relative Use The lynx use-density surfaces resulting from the fixed kernel analyses provided relative probabilities of finding lynx in areas throughout their distribution. A single use-density surface was calculated separately for both the aerial (n = 8058) and satellite truncated datasets (n = 16240). All 218 lynx released in Colorado, all radio-collared kittens and 2 captured unmarked adults were located at least once in Colorado. The majority of these lynx remained in Colorado. The use-density surfaces within Colorado were displayed separately for both the aerial (Figure 3) and satellite truncated datasets (Figure 4). Of the total locations available in the truncated datasets used to generate the usedensity surfaces, 7953 of the aerial locations and 13,241 of the satellite locations were in Colorado. Aerial and satellite use-density surfaces indicated similar high use-density areas. Satellite locations indicated broader spatial use by lynx because satellite collars provided more locations than flights. The use-density surface for lynx use in Colorado indicates two primary areas of use. The first is the Core Research Area (see Figure 1) and a secondary core centered in the Collegiate Peaks Wilderness (Figures 1, 3 and 4). High use is also documented for 1) the area east of Dillon, on both the north and south sides of I70 and 2) the area north of Hwy 50 centered around Gunnison and then north to Crested Butte. These last 2 high use areas are smaller in extent than the 2 core areas. Relative use-density surfaces were also generated for New Mexico, Wyoming and Utah and presented in detail in Shenk (2007). Home Range Reproductive females had the smallest 90% utilization distribution annual home ranges ( x = 75.2 km2, SE = 15.9 km2, n = 19), followed by attending males ( x = 102.5 km2, SE = 39.7 km2, n = 4). Nonreproductive females had the largest annual home ranges ( x = 703.9 km2, SE = 29.8 km2, n = 32) followed by non-reproductive males ( x = 387.0 km2, SE = 73.5 km2, n = 6). Combining all nonreproductive animals yielded a mean annual home range of 653.8 km2 (SE = 145.4 km2, n = 38). Survival Detailed analyses of lynx mortality was completed and described in Devineau et al. 2008 (in review). Monthly mortality rate was lower inside the study area than outside, and slightly higher for male than for female lynx, although 95% confidence intervals for sexes overlapped. Mortality was higher immediately after release (first month = 0.0368 [SE = 0.0140] inside the study area, and 0.1012 [SE = 0.0359] outside the study area), and then decreased according to a quadratic trend over time. As of August 27, 2008, CDOW was actively monitoring/tracking 45 of the 106 lynx still possibly alive (Table 2). There are 62 lynx that we have not heard signals on since at least August 27, 2007 and these animals are classified as ‗missing‘ (Table 2). One of these missing lynx is a mortality of unknown identity, thus only 61 are truly missing. Possible reasons for not locating these missing lynx include 1) 10 �long distance dispersal, beyond the areas currently being searched, 2) radio failure, or 3) destruction of the radio (e.g., run over by car). CDOW continues to search for all missing lynx during both aerial and ground searches. Two of the missing lynx released in 2000 are thought to have slipped their collars. Mortality Factors Of the total 218 adult lynx released, we have 112 known mortalities as of August 27, 2008 (Table 2). Starvation was a significant cause of mortality in the first year of releases only. The primary known causes of death included 30.4% human-induced deaths which were confirmed or probably caused by collisions with vehicles or gunshot (Table 3). Malnutrition and disease/illness accounted for 18.8% of the deaths. An additional 36.6% of known mortalities were from unknown causes. Mortalities occurred throughout the areas through which lynx moved, with 26.8% occurring outside of Colorado. The out of state mortalities included 14 in New Mexico, 4 in Wyoming, Utah and Nebraska, and 1 each in Arizona, Kansas, Iowa and Montana (Figure 2, Table 4). Reproduction Reproduction was first documented in 2003 when 6 dens and a total of 16 kittens were found in the lynx Core Release Area in southwestern Colorado. Reproduction was also documented in 2004, 2005 and 2006. No dens were found in 2007 or 2008 (Table 5). Field crews weighed, photographed, PIT-tagged the kittens and checked body condition. Beginning in 2005, we also collected blood samples from the kittens for genetic work in an attempt to confirm paternity Kittens were processed as quickly as possible (11-32 minutes) to minimize the time the kittens were without their mother. While working with the kittens the females remained nearby, often making themselves visible to the field crews. The females generally continued a low growling vocalization the entire time personnel were at the den. In all cases, the female returned to the den site once field crews left the area. At all dens the females appeared in excellent condition, as did the kittens. The kittens weighed from 270-500 grams. Lynx kittens weigh approximately 200 grams at birth and do not open their eyes until they are 10-17 days old. The percent of tracked females found with litters in 2006 was lower (0.095) than in the 3 previous years (0.413, SE = 0.032, Table 5). However, all demographic and habitat characteristics measured at the 4 dens that were found in 2006 were comparable to all other dens found. Mean number of kittens per litter from 2003-2006 was 2.78 (SE = 0.05) and sex ratio of females to males was equal ( x = 1.14, SE = 0.14). More details of reproduction in 2003-06 were presented in Shenk (2007). Den Sites.-- A total of 37 dens were found from 2003-2006. All of the dens except one have been scattered throughout the high elevation areas of Colorado, south of I-70. In 2004, 1 den was found in southeastern Wyoming, near the Colorado border. Dens were located on steep ( x slope = 30o , SE=2o), north-facing, high elevation ( x = 3354 m, SE = 31 m) slopes. The dens were typically in Engelmann spruce/subalpine fir forests in areas of extensive downfall of coarse woody debris (Shenk 2006). All dens were located within the winter use areas used by the females. No dens were found in either 2007 or 2008 even though up to 34 adult females were monitored intensively during the denning period (Table 5). Captures Two adult lynx were captured in 2001 for collar replacement. One lynx was captured in a tomahawk live-trap, the other was treed by hounds and then anesthetized using a jab pole. Five adult lynx were captured in 2002; 3 were treed by hounds and 2 were captured in padded leghold traps. In 2004, 1 lynx was captured with a Belisle snare and 6 adult lynx were captured in box-traps. Trapping effort was substantially increased in winter and spring 2005 and 12 adult lynx were captured and re-collared. Eight 11 �reintroduced lynx were captured in winter and spring 2006. In 2007, 11 reintroduced adult lynx were captured and re-collared and an additional 10 in 2008. All lynx captured in Colorado from 2005-2008 were caught in box-traps. In addition, as part of the collaring trapping effort, 14 Colorado-born kittens were captured and collared at approximately 10-months of age. Seven 2004-born kittens were collared in spring 2005, and 7, 2005-born kittens were collared in spring 2006. We were not successful at capturing and collaring any kittens born in 2006 in winter 2006-07. We did however, capture 2 adults (approximate age 2 years old) in winter 2006-07 that had no PIT-tags or radio collars. We assume these 2 lynx were from litters born in Colorado that were never found at dens (i.e., why there were no PIT-tags). All lynx captured for collaring or re-collaring were fitted with new Sirtrack TM dual VHF/satellite collars and re-released at their capture locations. Seven adult lynx were captured from March 1999-August 27, 2008 because they were in poor body condition (Table 6). Five of these lynx were successfully treated at the Frisco Creek Rehabilitation Center and re-released in the Core Release Area. One lynx, BC00F07, died from starvation and hypothermia within 1 day of capture at the rehabilitation center. Lynx QU04M07 died 3 days after capture at the rehabilitation center. Necropsy results documented starvation as the cause of death for this lynx that was precipitated by hydrocephalus and bronchopneumonia (unpublished data T. Spraker, CSUVTH). Seven lynx were captured (either by CDOW personnel or conservation personnel in other states) because they were in atypical habitat outside the state of Colorado (Table 6). They were held at Frisco Creek Rehabilitation Center for a minimum of 3 weeks, fitted with new Sirtrack TM dual VHF/satellite collars and re-released in the Core Release Area in Colorado. Five of these 7 lynx were still alive 6 months post-re-release but 3 had already dispersed out of Colorado and 1 stayed in Colorado through August 27, 2008. Two of these lynx died within 6 months of re-release: 1 died of starvation in Colorado and the other died of unknown causes in Nebraska. One lynx captured out of state and re-released currently remains in Colorado. HABITAT USE Landscape-scale daytime habitat use was documented from 9496 aerial locations of lynx collected from February 1999-June 30, 2007. Throughout the year Engelmann spruce - subalpine fir was the dominant cover used by lynx. A mix of Engelmann spruce, subalpine fir and aspen (Populus tremuloides) was the second most common cover type used throughout the year. Various riparian and riparian-mix areas were the third most common cover type where lynx were found during the daytime flights. Use of Engelmann spruce-subalpine fir forests and Engelmann spruce-subalpine fir-aspen forests was similar throughout the year. There was a trend in increased use of riparian areas beginning in July, peaking in November, and dropping off December through June. Site-scale habitat data collected from snow-tracking efforts indicate Engelmann spruce and subalpine fir were also the most common forest stands used by lynx for all activities during winter in southwestern Colorado. Comparisons were made among sites used for long beds, dens, travel and where they made kills. Little difference in aspect, mean slope and mean elevation were detected for 3 of the 4 site types including long beds, travel and kills where lynx typically use gentler slopes ( x = 15.7o ) at an mean elevation of 3173 m, and varying aspects with a slight preference for north-facing slopes. See Shenk (2006) for more detailed analyses of habitat use. 12 �DIET AND HUNTING BEHAVIOR Winter diet of lynx was documented through detection of kills found through snow-tracking. Prey species from failed and successful hunting attempts were identified by either tracks or remains. Scat analysis also provided information on foods consumed. A total of 548 kills were located from February 1999-April 2008. We collected over 950 scat samples from February 1999-April 2008 that will be analyzed for content. In each winter, the most common prey item was snowshoe hare, followed by red squirrel (Tamiusciurus hudsonicus; Table 7). The percent of snowshoe hare kills found however, varied annually from a low of 55.56% in 1999 to a high of 90.77% in winter 2002-2003. An annual mean of 73.29% (SE = 4.67) snowshoe hare kills in the diet has been documented. A comparison of percent overstory for successful and unsuccessful snowshoe hare chases indicated lynx were more successful at sites with slightly higher percent overstory, if the overstory species were Englemann spruce, subalpine fir or willow. Lynx were slightly less successful in areas of greater aspen overstory. This trend was repeated for percent understory at all 3 height categories except that higher aspen understory improved hunting success. Higher density of Engelmann spruce and subalpine fir increased hunting success while increased aspen density decreased hunting success. SNOWSHOE HARE ECOLOGY Two years of a 3-year study to evaluate snowshoe hare densities, demography and seasonal movement patterns among small and medium tree-sized lodgepole pine stands and mature spruce/fir stands have been completed and preliminary results presented (see Appendix I). DISCUSSION In an effort to establish a viable population of lynx in Colorado, CDOW initiated a reintroduction effort in 1997 with the first lynx released in winter 1999. From 1999 through spring 2007, 218 lynx were released in the Core Release Area. Locations of each lynx were collected through aerial- or satellite-tracking to document movement patterns and to detect mortalities. Most lynx remain in the high elevation, forested areas in southwestern Colorado. The use-density surfaces for lynx use in Colorado indicate two primary areas of use. The first is the Core Research Area (see Figure 1) and a secondary core centered in the Collegiate Peaks Wilderness (Figures 1, 3, 4). High use is also documented for 1) the area east of Dillon, on both the north and south sides of I70 and 2) the area north of Hwy 50 centered around Gunnison and then north to Crested Butte. These last 2 high use areas are smaller in extent than the 2 core areas. Dispersal movement patterns for lynx released in 2000 and subsequent years were similar to those of lynx released in 1999 (Shenk 2000). However, more animals released in 2000 and subsequent years remained within the Core Release Area than those released in 1999. This increased site fidelity may have been due to the presence of con-specifics in the area on release. Numerous travel corridors within Colorado have been used repeatedly by more than 1 lynx. These travel corridors include the Cochetopa Hills area for northerly movements, the Rio Grande Reservoir-Silverton-Lizardhead Pass for movements to the west, and southerly movements down the east side of Wolf Creek Pass to the southeast to the Conejos River Valley. Lynx appear to remain faithful to an area during winter months, and exhibit more extensive movements away from these areas in the summer. Reproductive females had the smallest 90% utilization distribution home ranges ( x = 75.2 km2, SE = 15.9 km2), followed by attending males ( x = 102.5 km2, SE = 39.7 km2) and non-reproductive animals ( x = 653.8 km2, SE = 145.4 km2). Most lynx currently being tracked are within the Core Release Area. During the summer months, lynx were documented to 13 �make extensive movements away from their winter use areas. Extensive summer movements away from areas used throughout the rest of the year have been documented in native lynx in Wyoming and Montana (Squires and Laurion 1999). Current data collection methods used for the Colorado lynx reintroduction program were not specifically designed to address the reintroduced lynx movements or use of areas in other states. In particular, the core research and release area were in Colorado. Therefore, the number of aerial locations obtained would be far fewer in other states than in Colorado which would bias low the number of lynx and intensity of lynx use documented outside the state. In contrast, obtaining satellite locations is not biased by the location of the lynx. Satellite locations are, however, biased by the shorter time the satellite transmitters function, approximately 18 months versus 60 months for the VHF transmitters used to obtain the aerial locations. However, data collected to meet objectives of the lynx reintroduction program were used to provide information to help address the question of lynx use outside of Colorado. Due to the rarity of flights conducted outside Colorado, only use-density surfaces generated from satellite locations were used to document relative lynx use of areas in New Mexico, Utah and Wyoming. New Mexico and Wyoming have been used continuously by lynx since the first year lynx were released in Colorado (1999) to the present. Lynx reintroduced in Colorado were first documented in Utah in 2000 and are still being documented there to date. In addition, all levels of lynx use-density documented throughout Colorado are also represented in New Mexico, Utah and Wyoming from none to the highest level of use (Shenk 2007). One den was found in Wyoming. Although no reproduction has been documented in New Mexico or Utah to date, documenting areas of the highest intensity of use and the continuous presence of lynx within these states for over six years does suggest the potential for yearround residency of lynx and reproduction in those states. From 1999-August 2008, there were 112 mortalities of released adult lynx. Human-caused mortality factors are currently the highest causes of death with approximately 30.4% attributed to collisions with vehicles or gunshot. Starvation and disease/illness accounted for 18.8% of the deaths while 36.6% of the deaths were from unknown causes. Lynx mortalities were documented throughout all areas lynx used, including 30 (26.8%) occurring in other states (Figure 2, Table 3). Nearly half (14 of 30) of the out-of-state mortalities were documented in New Mexico. Monthly mortality rate was lower inside the study area than outside, and slightly higher for male than for female lynx, although 95% confidence intervals for sexes overlapped. Mortality was higher immediately after release (first month = 0.0368 [SE = 0.0140] inside the study area, and 0.1012 [SE = 0.0359] outside the study area), and then decreased according to a quadratic trend over time. Reproduction is critical to achieving a self-sustaining viable population of lynx in Colorado. Reproduction was first documented from the 2003 reproduction season and again in 2004, 2005 and 2006. Lower reproduction occurred in 2006 (Table 5) but did include a Colorado-born female giving birth to 2 kittens, documenting the first recruitment of Colorado-born lynx into the Colorado breeding population. No reproduction was documented in 2007 or 2008. The cause of the decreased reproduction from 2006 08 is unknown. One possible explanation would be a decrease in prey abundance. Additional reproduction is likely to have occurred in all years from females we were no longer tracking, and from Colorado-born lynx that have not been collared. The dens we find are more representative of the minimum number of litters and kittens in a reproduction season. To achieve a viable population of lynx, enough kittens need to be recruited into the population to offset the mortality that occurs in that year and hopefully even exceed the mortality rate to achieve an increasing population. The use-density surfaces depict intensity of use by location. Why certain areas would be used more intensively than others should be explained by the quality of the habitat in those areas. 14 �Characteristics of areas used by lynx, as documented through aerial locations and snow-tracking of lynx in the Colorado core research area, include mature Engelmann spruce-subalpine fir forest stands with 4265% canopy cover and 15-20% conifer understory cover (Shenk 2006). Within these forest stand types, lynx appear to have a slight preference for north-facing, moderate slopes ( x = 15.7°) at high elevations ( x = 3173 m; Shenk 2006). Snow-tracking of released lynx also provided information on hunting behavior and diet through documentation of kills, food caches, chases, and diet composition estimated through prey remains. The primary winter prey species (n = 548) were snowshoe hare (Table 7) with an annual x = 73.3% (SE = 4.7, n = 10) and red squirrel (annual x = 18.2%, SE = 4.2, n = 10). Thus, areas of good habitat must also support populations of snowshoe hare and red squirrel. In winter, lynx reintroduced to Colorado appear to be feeding on their preferred prey species, snowshoe hare and red squirrel in similar proportions as those reported for northern lynx during lows in the snowshoe hare cycle (Aubry et al. 1999). Environmental conditions in the springs and summers of 2003 and 2006 resulted in high cone crops during their following winters based on field observations, resulting in increased red squirrel abundance. This may partially explain the higher percent of red squirrel kills, and thus a lower percent of snowshoe hare kills, found in winters 2003-04 and 2006-07 (Table 7). Caution must be used in interpreting the proportion of identified kills. Such a proportion ignores other food items that are consumed in their entirety and thus are biased towards larger prey and may not accurately represent the proportion of smaller prey items, such as microtines, in lynx winter diet. Through snow-tracking we have evidence that lynx are mousing and several of the fresh carcasses have yielded small mammals in the gut on necropsy. The summer diet of lynx has been documented to include less snowshoe hare and more alternative prey than in winter (Mowat et al., 1999). All evidence suggests reintroduced lynx are finding adequate food resources to survive. Mowat et al. (1999) suggest lynx and snowshoe hare select similar habitats except that hares select more dense stands than lynx. Very dense understory limits hunting success of the lynx and provides refugia for hares. Given the high proportion of snowshoe hare in the lynx diet in Colorado, we might then assume the habitats used by reintroduced lynx also depict areas where snowshoes hare are abundant and available for capture by lynx in Colorado. From both aerial locations taken throughout the year and from the site-scale habitat data collected in winter, the most common areas used by lynx are in stands of Engelmann spruce and subalpine fir. This is in contrast to adjacent areas of Ponderosa pine, pinyon juniper, aspen and oakbrush. The lack of lodgepole pine in the areas used by the lynx may be more reflective of the limited amount of lodgepole pine in southwestern Colorado, the Core Release Area, rather than avoidance of this tree species. Hodges (1999) summarized habitats used by snowshoe hare from 15 studies as areas of dense understory cover from shrubs, stands that are densely stocked, and stands at ages where branches have more lateral cover. Species composition and stand age appears to be less correlated with hare habitat use than is understory structure (Hodges 1999). The stands need to be old enough to provide dense cover and browse for the hares and cover for the lynx. In winter, the cover/browse needs to be tall enough to still provide browse and cover in average snow depths. Hares also use riparian areas and mature forests with understory. Site-scale habitat use documented for lynx in Colorado indicate lynx are most commonly using areas with Engelmann spruce understory present from the snow line to at least 1.5 m above the snow. The mean percent understory cover within the habitat plots is typically less than 15% regardless of understory species. However, if the understory species is willow, percent understory cover is typically double that, with mean number of shrubs per plot approximately 80, far greater than for any other understory species. 15 �In winter, hares browse on small diameter woody stems (<0.25"), bark and needles. In summer, hares shift their diet to include forbs, grasses, and other succulents as well as continuing to browse on woody stems. This shift in diet may express itself in seasonal shifts in habitat use, using more or denser coniferous cover in winter than in summer. The increased use of riparian areas by lynx in Colorado from July to November may reflect a seasonal shift in hare habitat use in Colorado. Major (1989) suggested lynx hunted the edge of dense riparian willow stands. The use of these edge habitats may allow lynx to hunt hares that live in habitats normally too dense to hunt effectively. The use of riparian areas and riparian-Engelmann spruce-subalpine fir and riparian-aspen mixes documented in Colorado may stem from a similar hunting strategy. However, too little is known about habitat use by hares in Colorado to test this hypothesis at this time. Lynx also require sufficient denning habitat. Denning habitat has been described by Koehler (1990) and Mowat et al. (1999) as areas having dense downed trees, roots, or dense live vegetation. We found this to be in true in Colorado as well (Shenk 2006). In addition, the dens used by reintroduced lynx were at high elevations and on steep north-facing slopes. All females that were documented with kittens denned in areas within their winter-use area. SUMMARY From results to date it can be concluded that CDOW developed release protocols that ensure high initial post-release survival of lynx, and on an individual level, lynx demonstrated they can survive longterm in areas of Colorado. We also documented that reintroduced lynx exhibited site fidelity, engaged in breeding behavior and produced kittens that were recruited into the Colorado breeding population. What is yet to be demonstrated is whether current conditions in Colorado can support the recruitment necessary to offset annual mortality in order to sustain the population. Monitoring of reintroduced lynx will continue in an effort to document such viability. ACKNOWLEDGMENTS The lynx reintroduction program involves the efforts of literally hundreds of people across North America, in Canada and USA. Any attempt to properly acknowledge all the people who played a role in this effort is at risk of missing many people. The following list should be considered to be incomplete. CDOW CLAWS Team (1998-2001): Bill Andree, Tom Beck, Gene Byrne, Bruce Gill, Mike Grode, Rick Kahn (Program Leader), Dave Kenvin, Todd Malmsbury, Jim Olterman, Dale Reed, John Seidel, Scott Wait, Margaret Wild. CDOW: John Mumma (Director 1996-2000), Russell George (Director 2001-2003), Bruce McCloskey (Director 2004-2007), Conrad Albert, Jerry Apker, Laurie Baeten, Cary Carron, Don Crane, Larry DeClaire, Phil Ehrlich, Lee Flores, Delana Friedrich, Dave Gallegos, Juanita Garcia, Drayton Harrison, Jon Kindler, Ann Mangusso, Jerrie McKee, Gary Miller, Melody Miller, Mike Miller, Kirk Navo, Robin Olterman, Jerry Pacheo, Mike Reid, Tom Remington, Ellen Salem, Eric Schaller, Mike Sherman, Jennie Slater, Steve Steinert, Kip Stransky, Suzanne Tracey, Anne Trainor, Scott Wait, Brad Weinmeister, Nancy Wild, Perry Will, Lisa Wolfe, Brent Woodward, Kelly Woods, Kevin Wright. Lynx Advisory Team (1998-2001): Steve Buskirk, Jeff Copeland, Dave Kenny, John Krebs, Brian Miller (Co-Leader), Mike Phillips, Kim Poole, Rich Reading (Co-Leader), Rob Ramey, John Weaver. U. S. Forest Service: Kit Buell, Joan Friedlander, Dale Gomez, Jerry Mastel, John Squires, Fred Wahl, Nancy Warren. U. S. Fish and Wildlife Service: Lee Carlson, Gary Patton (1998-2000), Kurt Broderdorp. 16 �State Agencies: Alaska: ADF&G: Cathie Harms, Mark Mcnay, Dan Reed (Regional Manager), Wayne Reglin (Director), Ken Taylor (Assist. Director), Ken Whitten, Randy Zarnke, Other:Ron Perkins (trapper), Dr. Cort Zachel (veterinarian). Washington: Gary Koehler. National Park Service: Steve King. Colorado State University: Alan Franklin, Gary White. Colorado Natural Heritage Program: Rob Schorr, Mike Wunder. Canada: British Columbia: Dr. Gary Armstrong (veterinarian), Mike Badry (government), Paul Blackwell (trapper coordinator), Trappers: Dennis Brown, Ken Graham, Tom Sbo, Terry Stocks, Ron Teppema, Matt Ounpuu. Yukon: Government: Arthur Hoole (Director), Harvey Jessup, Brian Pelchat, Helen Slama, Trappers: Roger Alfred, Ron Chamber, Raymond Craft, Lance Goodwin, Jerry Kruse, Elizabeth Hofer, Jurg Hofer, Guenther Mueller (YK Trapper‘s Association), Ken Reeder, Rene Rivard (Trapper coordinator), Russ Rose, Gilbert Tulk, Dave Young. Alberta: Al Cook. Northwest Territories: Albert Bourque, Robert Mulders (Furbearer Biologist), Doug Steward (Director NWT Renewable Res.), Fort Providence Native People. Quebec: Luc Farrell, Pierre Fornier. Colorado Holding Facility: Herman and Susan Dieterich, Kate Goshorn, Loree Harvey, Rachel Riling. Pilots: Dell Dhabolt, Larry Gepfert, Al Keith, Jim Olterman, Matt Secor, Brian Smith, Whitey Wannamaker, Steve Waters, Dave Younkin. Field Crews (1999-2007): Steve Abele, Brandon Barr, Bryce Bateman, Todd Bayless, Nathan Berg, Ryan Besser, Jessica Bolis, Mandi Brandt, Brad Buckley. Patrick Burke, Braden Burkholder, Paula Capece, Stacey Ciancone, Doug Clark, John DePue, Shana Dunkley, Tim Hanks, Carla Hanson, Dan Haskell, Nick Hatch, Matt Holmes, Andy Jennings, Susan Johnson, Paul Keenlance, Patrick Kolar, Tony Lavictoire, Jenny Lord, Clay Miller, Denny Morris, Kieran O‘Donovan, Gene Orth, Chris Parmater, Jake Powell, Jeremy Rockweit, Jenny Shrum, Josh Smith, Heather Stricker, Adam Strong, Dave Unger, David Waltz, Andy Wastell, Mike Watrobka, Lyle Willmarth, Leslie Witter, Kei Yasuda, Jennifer Zahratka. Research Associates: Bob Dickman, Grant Merrill. Data Analysts: Karin Eichhoff, Joanne Stewart, Anne Trainor. Data Entry: Charlie Blackburn, Patrick Burke, Rebecca Grote, Angela Hill, Mindy Paulek. Mary Schuette and Dave Theobald provided assistance with the GIS analysis and M. Schuette generated the maps used in this report Photographs: Tom Beck, Bruce Gill, Mary Lloyd, Rich Reading, Rick Thompson. Funding: CDOW, Great Outdoors Colorado (GOCO), Turner Foundation, U.S.D.A. Forest Service, Vail Associates, Colorado Wildlife Heritage Foundation. LITERATURE CITED Aubry, K. B., G. M. Koehler, J. R. Squires. 1999. Ecology of Canada lynx in southern boreal forests. Pages 373-396 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. Bartmann, R. M., and Byrne, G. (2001) Analysis and critique of the 1998 snowshoe hare pellet survey. Colorado Division of Wildlife Report No. 20. Fort Collins, Colorado. Byrne, G. 1998. Core area release site selection and considerations for a Canada lynx reintroduction in Colorado. Report for the Colorado Division of Wildlife. Devineau, O., T. M. Shenk, G. C. White, P. F. Doherty Jr., P. M. Lukacs, and R. H. Kahn. 2008. Estimating mortality for a widely dispersing reintroduced carnivore, the Canada lynx (Lynx canadensis). Ecology (in review). Hodges, K. E. 1999. Ecology of snowshoe hares in southern boreal and montane forests. Pages 163206 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires editors. Ecology and Conservation of Lynx in the United States. General 17 �Technical Report for U. S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. Koehler, G. M. 1990. Population and habitat characteristics of lynx and snowshoe hares in north central Washington. Canadian Journal of Zoology 68:845-851. Kolbe, J. A., J. R. Squires, T. W. Parker. 2003. An effective box trap for capturing lynx. Journal of Wildlife Management 31:980-985. Major, A. R. 1989. Lynx, Lynx canadensis canadensis (Kerr) predation patterns and habitat use in the Yukon Territory, Canada. M. S. Thesis, State University of New York, Syracuse. Mowat, G., K. G. Poole, and M. O‘Donoghue. 1999. Ecology of lynx in northern Canada and Alaska. Pages 265-306 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. Poole, K. G., G. Mowat, and B. G. Slough. 1993. Chemical immobilization of lynx. Wildlife Society Bulletin 21:136-140. Shenk, T. M. 1999. Program Narrative Study Plan: Post-release monitoring of reintroduced lynx (Lynx canadensis) to Colorado. Colorado Division of Wildlife, Fort Collins, Colorado. __________. 2001. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report, July: 7- 34. Colorado Division of Wildlife, Fort Collins, Colorado. __________. 2006. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report, July: 1-45. Colorado Division of Wildlife, Fort Collins, Colorado. __________. 2007. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report, July: 1-57. Colorado Division of Wildlife, Fort Collins, Colorado. Silverman, B.W. 1986. Density Estimation for Statistics and Data Analysis. Chapman and Hall. New York, New York, USA. Squires, J. R. and T. Laurion. 1999. Lynx home range and movements in Montana and Wyoming: preliminary results. Pages 337-349 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University Press of Colorado, Boulder, Colorado. U. S. Fish and Wildlife Service. 2000. Endangered and threatened wildlife and plants: final rule to list the contiguous United States distinct population segment of the Canada lynx as a threatened species. Federal Register 65, Number 58. Wild, M. A. 1999. Lynx veterinary services and diagnostics. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, Colorado. Zahratka, J. L. and T. M. Shenk. 2008. Population estimates of snowshoe hares in the southern Rocky Mountains. Journal of Wildlife Management 72: 906-912. Prepared by ___________________________________ Tanya M. Shenk, Wildlife Researcher 18 �Table 1. Number of wild-caught male (M) and female (F) Canada lynx (Lynx canadensis) from Alaska (AK) and Canada (BC = British Columbia, MB = Manitoba, QU = Quebec and YK = Yukon) released in southwestern Colorado per year from 1999–2006. State / Province of Origin Total Year %Released Sex AK BC MB QU YK 1999 19 2000 25 2003 15 2004 17 2005 17 2006 6 Total F 13 5 4 22 M 7 6 6 19 F 6 9 20 35 M 4 9 7 20 F 10 7 17 M 10 5 16 F 7 10 17 M 13 7 20 F 4 M 9 F M 30 1 3 8 3 18 8 3 20 4 3 7 5 2 7 48 218 91 4 45 Table 2. Status of adult Canada lynx (Lynx canadensis) reintroduced to Colorado as of August 27, 2008. Females Lynx Males Unknown TOTALS Released 115 103 218 Known Dead 62 49 1 112 Possible Alive 53 54 106 Missing 27 35 61a Monitoring/tracking 26 19 45 a 1 is unknown mortality Table 3. Causes of death for all Canada lynx (Lynx canadensis) released into southwestern Colorado 1999-2006 as of August 27, 2008. Mortalities Cause of Death Total (%) In Colorado (%) Outside Colorado (%) Unknown 41 (36.6) 27 (32.91) 14 (46.7) Gunshot 15 (13.4) 9 (11.0) 6 (20.0) Hit by Vehicle 14 (12.5) 9 (11.0) 5 (16.7) Starvation 11 (9.8) 10 (12.2) 1 (3.3) Other Trauma 8 (7.1) 7 (8.5) 1 (3.3) Plague 7 (6.3) 7 (8.5) 0 (0) Probable Gunshot 5 (4.5) 4 (4.9) 1 (3.3) Predation 5 (4.5) 5 (6.1) 0 (0) Probable Predation 3 (2.7) 2 (2.4) 1 (3.3) Illness 3 (2.7) 2 (2.4) 1 (3.3) Total Mortalities 112 82 (73.2) 30 (26.8) 19 �Table 4. Known lynx mortalities (n = 30) and causes of death documented by state outside of Colorado from February 1999 – August 27, 2008. Lynx ID State AK99F8 Unknown AK99M11 YK99M06 AK99F13 YK00F04 BC99M04 QU05M01 QU04F05 QU03F07 BC00M04 YK06F01 BC03M08 BC06F07 AK99M06 AK99M01 QU05M08 MB05F02 BC00F14 QU04F07 BC06M10 QU04F02 AK00M03 QU05M03 YK06M01 YK00F07 YK99F01 YK00M03 YK05M03 YK05M02 New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico Nebraska Nebraska Nebraska Nebraska Wyoming Wyoming Wyoming Wyoming Utah Utah Utah Utah Arizona Kansas Montana Iowa Date Mortality Recorded 7/30/1999 2000 1/27/2000 6/19/2000 6/22/2000 4/20/2001 6/7/2002 8/22/2005 8/26/2005 9/15/2005 7/19/2006 10/19/2006 10/19/2006 1/8/2007 11/16/1999 1/11/2005 10/1/2006 2/13/2007 7/28/2004 9/21/2004 8/15/2006 3/14/2007 7/2/2001 10/26/2005 12/4/2006 8/6/2007 9/15/2005 9/30/2005 11/8/2005 8/6/2007 Cause of Death Starvation Hit by Vehicle Unknown Probable Gunshot Unknown Gunshot Gunshot Unknown Hit by Vehicle Unknown Unknown Unknown Unknown Gunshot Gunshot Snared (Other Trauma) Unknown Gunshot Unknown Unknown Vehicle Collision Unknown Unknown Unknown Unknown Unknown Gunshot Vehicle Collision Unknown Vehicle Collision Table 5. Lynx reproduction summary statistics for 1999-2008. No reproduction was expected in 1999 because it was the first year of lynx releases and most animals were released after breeding season. Year Females Tracked 2000 2001 2002 2003 2004 2005 2006 2007 2008 TOTAL 9 25 21 17 26 40 42 34 28 Dens Found in May/June 0 0 0 6 11 17 4 0 0 Percent Tracked Females with Kittens 0.0 0.0 0.0 0.353 0.462 0.425 0.095 0.0 0.0 Additional Litters Found in Winter 0 0 0 0 2 1 0 0 0 20 Mean Kittens/Litte r (SE) 2.67 (0.33) 2.83 (0.24) 2.88 (0.18) 2.75 (0.47) Total Kittens Found Sex Ratio M/F (SE) 0 0 0 16 39 50 11 1.0 1.5 0.8 1.2 0 0 116 1.14 (0.14) �Table 6. Lynx captured because they were in poor body condition or were in atypical habitat and their fates 6 months post re-release and as of August 28, 2008. Lynx ID BC99F6 Date of Capture 3/25/1999 State Where Captured Colorado Reason For Capture Poor body condition Date of Re-release 5/28/1999 Status 6 Months Post Re-release Dead AK99M9 3/24/2000 Colorado 5/3/2000 Missing AK99F2 4/18/2000 Colorado 5/22/2000 BC00F7 2/11/2001 Colorado Alive in Colorado Dead BC00M13 3/21/2001 Wyoming BC03M08 9/5/2003 Colorado QU04M07 2/2/2006 Colorado Poor body condition Poor body condition Poor body condition Poor body condition Poor body condition Poor body condition BC04M01 11/5/2004 Utah QU04F02 4/10/2005 Nebraska QU05M08 11/25/2005 Wyoming QU04M04 12/5/2006 Utah YK00F07 12/12/2006 Utah YK05M02 1/1/2007 Kansas BC04M08 1/22/2007 Wyoming N/A 4/24/2001 1/1/2004 N/A Alive in Colorado Alive in Colorado Dead Atypical habitat Atypical habitat 12/5/2004 Atypical habitat Atypical habitat Atypical habitat Atypical habitat Atypical habitat 4/18/2006 Dead 1/20/2007 1/20/2007 Dead in Colorado Alive in Utah 2/2/2007 Alive in Iowa 2/15/2007 Alive in Colorado 5/7/2005 Alive in Colorado Alive in Wyoming Current Status Died 7/19/1999 in Colorado from vehicle collision Last located 5/3/2000, collar failure Last located 7/30/2003 in Colorado Died at Rehab Center on 2/12/2001 Last located 10/26/2004 in Colorado Died in New Mexico of unknown causes 10/19/06 Died at Rehab Center on 2/5/2006 from hydrocephalous and pneumonia In Colorado as of 8/27/2008 Died 3/14/2007 in Wyoming (good habitat) of unknown causes Died of unknown causes in Nebraska 10/1/2006 Died of starvation in Colorado, found 3/19/07 Died in Utah of unknown causes 8/6/2007 Died in Iowa from vehicle collision 8/6/2007 Died in Colorado from gunshot 1/4/2008 Table 7. Number of kills found each winter field season through snow-tracking of lynx and percent composition of kills of the three primary prey species. Field Season 1999 1999-2000 2000-2001 2001-2002 2002-2003 2003-2004 2004-2005 2005-2006 2006-2007 2007-2008 Total/Mean n 9 83 89 54 65 37 78 50 41 42 548 Snowshoe Hare 55.56 67.47 67.42 90.74 90.77 67.57 83.33 90.00 61.00 59.00 73.29 (SE=4.7) Prey (%) Red Squirrel Cottontail 22.22 0 19.28 1.20 19.10 8.99 5.56 0 6.15 0 27.03 2.70 10.26 0 0.08 0 39.0 0 33.3 0 18.2 (SE=4.2) 1.29 (SE=0.95) 21 Other 22.22 12.05 4.49 3.70 3.08 2.70 6.41 0.02 0 7.4 6.21 (SE=2.22) �Figure 1. Lynx are monitored throughout Colorado and by satellite throughout the western United States. The lynx core release area, where all lynx were released, is located in southwestern Colorado. A lynx-established core use area has developed in the Taylor Park and Collegiate Peak area in central Colorado. 22 �Figure 2. All documented lynx locations (non-truncated datasets) obtained from either aerial (red circles) or satellite (yellow circles) tracking from February 1999 through August 27, 2008. All known lynx mortality locations (n = 112) are displayed as black stars. 23 �Figure 3. Use-density surface for lynx aerial locations (truncated dataset) in Colorado from September 1999-March 2007. 24 �Figure 4. Use-density surface for lynx satellite locations (truncated dataset) in Colorado from September 1999-March 2007. 25 �Colorado Division of Wildlife July 2007 - June 2008 APPENDIX I WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 0670 2 Federal Aid Project No. N/A : : : : : Division of Wildlife Mammals Research Lynx Conservation Density, Demography, and Seasonal Movements Of Snowshoe Hare in Colorado Period Covered: July 1, 2007- June 30, 2008 Author: J. S. Ivan, Ph.D. Candidate, Colorado State University Personnel: Dr. T. Shenk of CDOW and Dr. G. C. White of Colorado State University. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT A program to reintroduce the threatened Canada lynx (Lynx canadensis) into Colorado was initiated in 1997. Analysis of scat collected from winter snow tracking indicates that snowshoe hares (Lepus americanus) comprise 65–90% of the winter diet of reintroduced lynx. Thus, existence of lynx in Colorado and success of the reintroduction hinge at least partly on maintaining adequate and widespread hare populations. Beginning in July 2006, I initiated a study to assess the relative value of 3 stand types for providing hare habitat in Colorado. These types include mature, uneven-aged spruce/fir forests, sapling lodgepole pine forests (―s mall lodgepole‖), and pole-sized lodgepole pine forests (―m edium lodgepole‖). Estimates and comparisons of survival, recruitment, finite population growth rate, and maximum (late summer) and minimum (late winter) snowshoe hare densities for each stand will provide the metrics for assessing these stands. Thus far, snowshoe hare densities on the study area are low compared to densities reported elsewhere. Within the study area, hare densities during summer were highest in small lodgepole stands, followed by mature spruce/fir and medium lodgepole, respectively. This pattern was consistent through the first 2 summers of this project, although absolute hare densities declined considerably in summer 2007. Hare density in small and medium lodgepole stands equalized during both winters of the project. However, as with summer, overall density was much lower during the second winter compared to the first. Hare survival from summer to winter has been relatively high. However the single winter to summer estimate I have to date is quite low. Extension of this time series will help determine whether low winter to summer survival is typical or somehow related to the decline in density. 26 �WILDIFE RESEARCH REPORT DENSITY, DEMOGRAPHY, AND SEASONAL MOVEMENTS OF SNOWSHOE HARES IN COLORADO JACOB S. IVAN P. N. OBJECTIVE Assess the relative value of 3 stand types (mature spruce/fir, sapling lodgepole, pole-sized lodgepole) that purportedly provide high quality hare habitat by estimating survival, recruitment, finite population growth rate, and maximum (late summer) and minimum (late winter) snowshoe hare densities for each type. SEGMENT OBJECTIVES 1. Complete mark-recapture work across all replicate stands during late summer (mid-July through midSeptember) and winter (mid-January through March). 2. Obtain daily telemetry locations on radio-tagged hares for 10 days immediately after capture periods, as well as monthly between primary trapping sessions. 3. Locate, retrieve, and refurbish radio tags as mortalities occur. 4. Summarize initial sampling efforts and provide initial density estimates for Progress Reports for Colorado Division of Wildlife (CDOW). INTRODUCTION NEED A program to reintroduce the threatened Canada lynx (Lynx canadensis) into Colorado was initiated in 1997. Since that time, 218 lynx have been released in the state, and an extensive effort to determine their movements, habitat use, reproductive success, and food habits has ensued (Shenk 2005, Shenk 2007). Analysis of scat collected from winter snow tracking indicates that snowshoe hares (Lepus americanus) comprise 65–90% of the winter diet of reintroduced lynx (T. Shenk, Colorado Division of Wildlife, unpublished data). Thus, as in the far north where the intimate relationship between lynx and snowshoe hares has captured the attention of ecologists for decades, it appears that the existence of lynx in Colorado and success of the reintroduction effort may hinge on maintaining adequate and widespread populations of hares. Colorado represents the extreme southern range limit for both lynx and snowshoe hares (Hodges 2000). At this latitude, habitat for each species is less widespread and more fragmented compared to the continuous expanse of boreal forest at the heart of lynx and hare ranges. Neither exhibits dramatic cycles as occur farther north, and typical lynx ( 2 3 lynx/100km2; Aubry et al. 2000) and hare ( 1 2 hares/ha; Hodges 2000) densities in the southern part of their range correspond to cyclic lows form northern populations (2-30 lynx/100 km2, 1 16 hares/ha; Aubry et al. 2000, Hodges 2000, Hodges et al. 2001). Whereas extensive research on lynx-hare ecology has occurred in the boreal forests of Canada, literature regarding the ecology of these species in the southern portion of their range is relatively sparse. This scientific uncertainty is acknowledged in the ― Canada Lynx Conservation Assessment and Strategy,‖ a formal agreement between federal agencies intended to provide a consistent approach to lynx conservation on public lands in the lower 48 states (Ruediger et al. 2000). In fact, one of the explicit guiding principles of this document is to ― retain future options…until more conclusive information concerning lynx management is developed.‖ Thus, management recommendations in this agreement are decidedly conservative, especially with respect to timber management, and are applied broadly to cover 27 �all habitats thought to be of possible value to lynx and hare. Accurate identification and detailed description of lynx-hare habitat in the southern Rocky Mountains would permit more informed and refined management recommendations. A commonality throughout the snowshoe hare literature, regardless of geographic location, is that hares are associated with dense understory vegetation that provides both browse and protection from elements and predators (Wolfe et al. 1982, Litvaitis et al. 1985, Hodges 2000, Homyack et al. 2003, Miller 2005). In western mountains, this understory can be provided by relatively young conifer stands regenerating after stand-replacing fires or timber harvest (Sullivan and Sullivan 1988, Koehler 1990a, Koehler 1990b, Bull et al. 2005) as well as mature, uneven-aged stands (Beauvais 1997, Griffin 2004). Hares may also take advantage of seasonally abundant browse and cover provided by deciduous, open habitats (e.g., riparian willow [Salix spp.], aspen [Populus tremuloides]; Wolff 1980, Miller 2005). In drier portions of hare range, such as Colorado, regenerating stands can be relatively sparse, and hares may be more associated with mesic, late-seral forest and/or riparian areas than with young stands (Ruggiero et al. 2000). Numerous investigators have sought to determine the relative importance of these distinctly different habitat types with regards to snowshoe hare ecology. Most previous evaluations were based on hare density or abundance (Bull et al. 2005), indices to hare density and abundance (Wolfe et al. 1982, Koehler 1990a, Beauvais 1997, Miller 2005), survival (Bull et al. 2005), and/or habitat use (Dolbeer and Clark 1975). Each of these approaches provides insight into hare ecology, but taken singly, none provide a complete picture and may even be misleading. For example, extensive use of a particular habitat type may not accurately reflect the fitness it imparts on individuals, and density can be high even in ― sink‖ habitats (Van Horne 1983). A more informative approach would be to measure density, survival, and habitat use simultaneously in addition to recruitment and population growth rate through time. Griffin (2004) employed such an approach and found that summer hare densities were consistently highest in young, dense stands. However, he also noted that only dense mature stands held as many hares in winter as in summer. Furthermore hare survival seemed to be higher in dense mature stands, and only dense mature stands were predicted (by matrix projection) to impart a mean positive population growth rate on hares. Griffin‘s (2004) study occurred in the relatively moist forests of Montana, which share many similarities but also many notable differences with Colorado forests including levels of fragmentation, species composition, elevation, and annual precipitation. Density estimation is a key component in assessing the value of a particular stand type and is the common currency by which hare populations are compared across time and space. However, it can be a difficult metric to estimate accurately. Abundance estimation based on capture-recapture methods is a well-developed field (Otis et al. 1978, White et al. 1982), but is often too costly and labor intensive to be implemented on scales necessary to effectively monitor density over a biologically meaningful area. Also, density can be difficult to assess from grid-trapping efforts because it is often unclear how much area was effectively sampled by the grid (Williams et al. 2002:314). Alternate approaches can produce density estimates that differ by an order of magnitude even when calculated from the same data (Zahratka 2004). Indices such as pellet plot counts and distance sampling of pellet groups can be used to estimate density, but each of these has limitations as well (Krebs et al. 1987, Eriksson 2006). The study outlined below is designed principally to evaluate the importance of young, regenerating lodgepole pine (Pinus contorta) and mature Engelmann spruce (Picea engelmannii)/ subalpine fir (Abies lasiocarpa) stands in Colorado by examining density and demography of snowshoe hares that reside in each. Secondarily, I intend to quantify movement between these stands and other seasonally available types (e.g., willow). My hope is that information gathered from this research will be drawn upon as managers make routine decisions, leading to landscapes that include stands capable of 28 �supporting abundant populations of hares. I assume that if management agencies focus on providing habitat, hares will persist. I will use mark-recapture techniques as data from such an approach can provide information on both density and demography. In the future, I will address the ― effective trapping area‖ issue using a new approach that augments mark-recapture data with telemetry locations of animals using the grid. However, for this report I used one of the more popular, traditional techniques. I determined that 2 classes of young, regenerating lodgepole stands could both provide adequate hare habitat. Thus, in addition to older spruce/fir forests, I am sampling ―s mall‖ (2.54-12.69 cm dbh) and ―m edium‖ (12.70-22.85 cm dbh) stands regenerating from clearcutting that took place 20 and 40 years ago, respectively (Figure 1). Additionally, medium lodgepole stands were pre-commercially thinned 20 years ago; small lodgepole stands have not yet been thinned. Hypotheses 1) In general, snowshoe hare density in Colorado will be relatively low ( 0.5 hares/ha) compared to densities reported in northern boreal forests, even immediately post-breeding when an influx of juveniles will bolster hare numbers. 2) Snowshoe hare density will be consistently highest in small lodgepole pine stands, followed by large spruce/fir and medium lodgepole pine, respectively. 3) Survival will generally be highest in mature (large) spruce/fir stands followed by small and medium lodgepole pine, respectively. 4) Finite population growth rate will be consistently at or above 1.0 in mature spruce/fir stands with survival contributing most significantly to the growth rate. Finite growth rates for the lodgepole pine stands will be more variable. 5) Snowshoe hares will significantly shift their home ranges to make use of abundant food and cover provided by riparian willow (and/or aspen) habitats in summer. 6) Snowshoe hare density, survival, and recruitment will be highly correlated with understory cover and stem density. STUDY AREA The study area stretches from Taylor Park to Pitkin in central Colorado (Figure 2). Elevation ranges from 2700 m to 4000 m. Sagebrush (Artemisia spp.) dominates broad, low-lying valleys. Most montane areas are covered by even-aged, large-diameter lodgepole pine forests with sparse understory. Moist, north-facing slopes and areas near tree line are dominated by large-diameter Engelmann spruce/subalpine fir. Interspersed along streams and rivers are corridors of willow. Patches of aspen occur sporadically on southern exposures. This area was chosen over other potential study areas in the state because 1) it contained numerous examples of the 3 stand types of interest (more southern regions lack naturally occurring stands of lodgepole pine), 2) it was not subject to confounding effects of largescale mountain pine beetle outbreak as were more northern stands, and 3) an adequate number of radio frequencies were available to support a large study with hundreds of radio-tagged individuals. Within the study area I selected sample stands based on the following: Potential replicate stands were required to be 1) close enough geographically to minimize differences due to climate, weather, and topography, but are far enough apart to be considered independent, 2) adjacent to one or more riparian willow corridors, 3) within 1 km of an access road for logistical purposes, 4) of suitable size and shape to admit a 16.5-ha trapping grid, and 5) consistent in their management history (i.e., replicate lodgepole pine stands were clear-cut and/or thinned within 1-2 years of each other). I queried the U.S. Forest Service R2VEG GIS database using the criteria listed above to initially develop a suite of potential sample stands. I further narrowed this suite after obtaining updated standlevel information from local USFS personnel (Art Haines, Silviculturalist, USFS Gunnison Ranger 29 �District, personal communication). Finally, I ground-truthed potential stands and qualitatively assessed their representativeness and similarity to other potential replicates. Given the numerous constraints imposed, very few stands met all criteria. Thus, I was unable to randomly select sample stands from a population of suitable stands. Rather, I subjectively chose the ― best‖ stands from among the handful that met my criteria. Small lodgepole stands rarely occur on the landscape in patches large enough to fit a full trapping grid. To accommodate this, I sampled 6 replicate small lodgepole stands (rather than 3) using half-sized trapping grids. METHODS Experimental Design/Procedures Variables.--The response variables of interest for this project include stand-specific snowshoe hare density (D), apparent survival ( ), recruitment (f), finite population growth rate (λ), and a metric of seasonal movement. Density is the number of hares per unit area and is estimated using a conventional techniques in this report. The stand-specific demographic parameters will be estimated primarily from capture-mark-recapture methods. As such, apparent survival is defined as the probability that a marked animal alive and in the population at time i survives and is in the population at time i + 1. Apparent survival encompasses losses due to both death and emigration. Estimates of recruitment, population growth, and seasonal movement are forthcoming and not provided in this report. Potential explanatory variables for snowshoe hare density, demographics, and movement include general species composition and structural stage of each stand in which response variables are measured. Additionally, stem density, horizontal cover, and canopy cover (to a lesser extent) are highly correlated with snowshoe hare abundance and habitat use (Wolfe et al. 1982, Litvaitis et al. 1985, Hodges 2000, Zahratka 2004, Miller 2005). Thus, I further characterized vegetation in each stand by measuring stem density by size class (1-7 cm, 7.1-10 cm, and >10 cm), percent canopy cover, percent horizontal cover of understory and basal area. Basal area is an easily obtainable metric that may be correlated with the other variables and is recorded routinely during timber cruises, whereas the others are not. Thus, it might prove a useful link for biologists designing management strategies for snowshoe hare. Additionally, I recorded physical covariates such as ambient temperature, precipitation, and snow depth at each stand during sampling. These metrics were not included in the current preliminary analyses, but will be used as covariates in future models. Sampling.--All trapping and handling procedures have been approved by the Colorado State University Animal Care and Use Committee and filed with the Colorado Division of Wildlife. Snowshoe hares breed synchronously and generally exhibit 2 birth pulses in Colorado (although in some years, some individuals may have 3 litters), with the first pulse terminating approximately June 5 20 and the second approximately July 15–25 (Dolbeer 1972). To obtain a maximum density estimate, I began data collection on the first suite of sites immediately following the second birth pulse in late July. Along with a crew of 5 technicians, I deployed one 7 12 trapping grid (50-m spacing between traps; grid covers 16.5 ha) in the large spruce/fir and medium lodgepole stands within the first suite, along with 2 6 7 grids in 2 small lodgepole stands. Grid set up and trap deployment followed Griffin (2004) and Zahratka (2004). Grid locations and orientation within each stand were chosen subjectively to accommodate logistical constraints and to ensure that hares using the grid had ample opportunity to use adjacent riparian willow zones. As traps were deployed, they were locked open and ― pre-baited‖ with apple slices, hay cubes, and commercial rabbit chow. Traps were pre-baited in this manner for a total of 3 nights to maximize capture rates when trapping began. This minimized the number of trap-nights needed to capture the desired number of animals which in turn minimized trap-related injuries and minimized problems with predators keying into trap lines. During pilot work in winter 2005, I observed low but increasing capture rates (<0.20) during the first 3 nights of trapping, with higher, more stable capture 30 �probabilities after 3 days (approximately 0.35–0.45). Thus 3 days of pre-baiting seemed reasonable. Traps were set on the afternoon of the 4th day and checked early each morning and again in the evening on days 5–9. By checking traps in both morning and evening I prevented hares from being entrapped >13 hours, which should minimize capture stress. A crew of 2 people worked together on each grid to check traps and process captures as quickly as possible. All captured hares were coaxed out of the trap and into a dark handling bag by blowing quick shots of air on them from behind. Hares remained in the handling bag, physically restrained with their eyes covered, for the entire handling process. Each individual was aged, sexed, marked with a passive integrated transponder (PIT) tag and temporary ear mark (to track PIT tag retention), then released. Aging consisted of assigning each individual as either juvenile (<1 year old, <1000 g) or adult ( 1 year old, 1000 g) based on weight. This criterion is accurate through the end of September at which point juveniles are difficult to distinguish from adults (K. Hodges, University of British Columbia; P. Griffin, University of Montana, personal communication). After the first day of trapping, all captured hares were scanned for a PIT tag prior to any handling and those already marked were recorded and immediately released. Traps and bait were completely removed from the grid on day 10. In addition to PIT tags and ear marks, I radio collared up to 10 hares captured on each grid with a 28-g mortality-sensing transmitter (BioTrack, LTD) to facilitate unbiased density estimation as well as assessment of seasonal movements. I expected heterogeneity in snowshoe hare movements and use of the grid area, with potential bias surfacing due to location at which a hare is captured (e.g., hares captured on the edge of a grid may use the grid area differently than those captured at the center), and differential behavioral responses to trapping (e.g., young individuals may have lower capture probabilities and thus may be more likely to be captured on later occasions). To guard against the first potential bias, I randomly selected a starting trap location each morning and ran the grid systematically from that point. Thus, the first several hares encountered (and collared) were as likely to be from the inner part of the grid as from the edge. To protect against the second potential source of bias, I refrained from deploying the final 3 collars until days 4 and 5 of the trapping session. Immediately following the removal of traps, the field crew began work locating each radiocollared hare 1–2 times per day for 10 days. Most locations were obtained by triangulation from relatively close proximity, but some were obtained by ―hom ing‖ on a signal (Samuel and Fuller 1996, Griffin 2004) taking care not to push hares while approaching them. Because hares are largely nocturnal (Keith 1964, Mech et al. 1966, Foresman and Pearson 1999), I made an effort to conduct telemetry work at various times of the night (safety and logistics permitting) and day to gather a representative sample of locations for each hare. Crews gathered telemetry locations for radio-collared hares on the initial suite of sites for 10 days. Then the 10 day trapping procedure and 8 to 10 day telemetry work were repeated on the grids comprising suites 2 and 3(Figure 3). The entire process was repeated during the winter when densities should have been at a minimum. Thus, during the period covered by this report, sampling occurred from July 16 – September 14 and from January 20 – March 24, 2008. Sampling occurred across similar dates during FY06/07 and will continue during FY08/09. During the interim between intensive trapping and telemetry work, monthly telemetry checks were conducted from the air to track mortalities and facilitate retrieval of collars from dead hares. Telemetry work also occurred during ―pr e-baiting‖ days after the initial summer sampling session to determine which hares were still alive and immediately available to be sampled by the grid during the ensuing trapping period. Vegetation sampling at each stand commenced in June 2008 and is nearly finished. I followed protocols established through previous snowshoe hare and lynx work in Colorado (Zahratka 2004, T. 31 �Shenk, Colorado Division of Wildlife, personal communication). Specifically, on each of the 12 livetrapping grids, I laid out 5 5 grids (3-m spacing) of vegetation sampling points centered on 15 of the 84 trap locations (Figure 4; 9 points were sampled on each of the ½-sized small lodgepole stands). At each of the 25 vegetation sampling points, I recorded canopy cover (present or absent) using a densitometer. I quantified downed coarse wood along the center transect of the 25-point grid following Brown (1974). From the centerpoint (i.e., trap location) I measured 1) distance to the nearest woody stem 1.0 7.0 cm, 7.1 10.0 cm, and >10.0 cm in diameter at heights of 0.1 m and 1.0 m above the ground (to capture both summer and winter stem density; Barbour et al. 1999), 2) horizontal cover in 0.5-m increments above the ground up to 2 m (Nudds 1977), 3) basal area, and 4) slope. Data Analysis Density, Survival, and Population Growth.--I analyzed mark-recapture data in a robust design framework (Williams et al. 2002:523-554) treating summer and winter sampling occasions as primary periods, and the 5-day trapping sessions within each as secondary periods. As such, I assumed hare populations were demographically and geographically closed during the short 5-day mark-recapture sampling periods, but were open to immigration, emigration, births, and deaths between these occasions. I specified the Pradel Robust Design data type in Program MARK (White and Burnham 1999) and chose the Huggins closed capture model (Huggins 1989, 1991) to obtain abundance estimates for each grid from the secondary periods. I obtained estimates of apparent survival ( ˆ i ) between each primary period. I employed a technique known as ½ Mean Maxmimum Distance Moved (MMDM; Wilson and Anderson 1985) to calculate the effective area trapped and obtain a density estimate for each grid from each secondary period. Future density analyses will employ a new estimator that employs telemetry data to correct for bias (Ivan 2005). I used Akaike‘s Information Criterion corrected for small sample size (AICc; Burnham and Anderson 1998) to select appropriate models from alternatives that included all 8 closed capture models (Otis et al. 1978) in combination with models that allowed survival to be constant, vary with time, and/or vary with stand type. RESULTS AND DISCUSSION I captured 30 hares 73 times during July-September 2007. I captured 48 hares 71 times during January-March 2008. During summer, density estimates have thus far followed hypotheses 1) and 2) above (Figure 5). Specifically, hare densities were clearly highest in small lodgepole stands and quite low in medium lodgepole stands. Spruce/fir was intermediate in density. This pattern remained consistent between summer 2006 to summer 2007, although the absolute density of hares dropped considerably during summer 2007. Why this decline occurred is unclear, although disease outbreak, natural population cycles, and response to increased predation due to lynx reintroduction are possibilities. Note that even the highest densities recorded here correspond to low estimates observed in other parts of hare range (Hodges 2000). Hare densities tend to equalize in lodgepole stands during winter (Figure 5). I submit that the interplay between food, cover, and snow depth provides a plausible explanation for this pattern. Medium lodgepole stands apparently provide very little forage/cover for hares during summer as the canopy in these stands is generally ≥1 meter off the ground. However, in winter, accumulated snow may make that canopy available again to hares. Conversely, small lodgepole stands provide abundant food and cover during summer, but accumulated snow during winter brings hares closer to the crowns of the young trees, which then provide less cover. Spruce/fir stands probably provide adequate access to both food and cover during both summer and winter due to their uneven-aged, multi-layered structure. Like the summer estimates, density during the second winter was much lower than during the first winter. 32 �Hare survival from the first sampling season into the first winter was relatively high (Figure 6). However, survival from the first winter to the second summer declined drastically. Survival from the second summer to the second winter was again quite high. Whether this pattern is typical is unclear. Survival from winter to summer is commonly lower than from summer to winter. However, the low survival from the first winter to second summer is coincident with the dramatic decline in hare density observed on spruce/fir and small lodgepole grids. Thus, low survival for this period is possibly reflective of, or maybe even a driver for, the decline in density. Extension of the time series and a breakdown of survival by stand type should provide more evidence for one or the other of these explanations. SUMMARY Snowshoe hare densities on my study sites appear to be relatively low compared to densities reported elsewhere. Densities during summer were highest in small lodgepole stands, followed by spruce/fir and medium lodgepole. During winter, densities equalize in lodgepole stands, possibly due to the interplay between snow depth and canopy height in small and medium lodgepole pine. Hare density declined considerably beginning in summer 2007. Summer to winter hare survival has been consistently high thus far in the study, but the lone winter to summer survival estimate is quite low. It is unclear whether winter to summer survival is typically this low or whether that estimate is related to coincident drop in density. ACKNOWLEDGMENTS Ken Wilson (CSU), Bill Romme (CSU), Paul Doherty (CSU), Dave Freddy (CDOW), Chad Bishop (CDOW), and Paul Lukacs (CDOW) provided helpful insight on the design of this study. We appreciate the invaluable logistical support provided by Mike Jackson (USFS), Art Haines (USFS), Jake Spritzer (USFS), Kerry Spetter (USFS), Margie Michaels (CDOW), Gabriele Engler (USGS), Dana Winkelman (USGS), Brandon Diamond (CDOW), Chris Parmeter (CDOW), Kathaleen Crane (CDOW), Lisa Wolfe (CDOW), and Laurie Baeten (CDOW). Jim Gammonley (CDOW), Dave Freddy (CDOW), Chad Bishop (CDOW), Jack Vayhinger (CDOW), Brandon Diamond (CDOW) assisted with trucks and equipment. The following hardy individuals collected the hard-won data presented in this report: Braden Burkholder, Matt Cuzzocreo, Brian Gerber, Belita Marine, Adam Behney, Pete Lundberg, Katie Yale, Britta Shielke, Cory VanStratt, Mike Watrobka, Meredith Goss, Sidra Blake, Keith Rutz, Rob Saltmarsh, Jennie Sinclair, Evan Wilson, Mat Levine, Matt Strauser, Greg Davidson, Leah Yandow, Renae Sattler, and Caylen Cummins. Funding was provided by the Colorado Division of Wildlife. LITERATURE CITED Aubry, K. B., G. M. Koehler, and J. R. Squires. 2000. Ecology of Canada lynx in southern boreal forests. Pages 373-396 in Ruggiero, L. F., K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, and J. R. Squires, editors. Ecology and conservation of lynx in the United States. Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Collins, Colorado, USA. Barbour, M. G., J. H. Burk, W. E. Pitts, F. S. Gilliam, and M. W. Schwartz. 1999. Terrestrial plant ecology, Addison Wesley Longman, Inc., Menlo Park, California, USA. Beauvais, G. P. 1997. Mammals in fragmented forests in the Rocky Mountains: community structure, habitat selection, and individual fitness. Dissertation, University of Wyoming, Laramie, Wyoming, USA. Brown, J. K. 1974. Handbook for inventorying downed woody material. U.S. Department of Agriculture, Forest Service, Intermountain Forest and Range Experiment Station, General Technical Report 33 �INT-16. Burnham, K. P., and D. R. Anderson. 1998. Model selection and inference: a practical informationtheoretic approach. Springer-Verlag, New York, New York, USA. Bull, E. L., T. W. Heater, A. A. Clark, J. F. Shepherd, and A. K. Blumton. 2005. Influence of precommercial thinning on snowshoe hares. U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station, General Technical Report PNW-RP-562. Dolbeer, R. A. and W. R. Clark. 1975. Population ecology of snowshoe hares in the central Rocky Mountains. Journal of Wildlife Management 39:535-549. Dolbeer, R. A. 1972. Population dynamics of snowshoe hare in Colorado. Dissertation, Colorado State University, Fort Collins, Colorado, USA. Eriksson, H. M. 2006. Snowshoe hare densities in post-fire vegetation. Thesis, University of Alaska Fairbanks, Fairbanks, Alaska, USA. Foresman, K. R. and D. E. Pearson. 1999. Activity patterns of American martens, Martes americana, snowshoe hares, Lepus americanus, and red squirrels, Tamiasciurus hudsonicus, in westcentral Montana. The Canadian Field-Naturalist 113:386-389. Griffin, P. C. 2004. Landscape ecology of snowshoe hares in Montana. Dissertation, University of Montana, Missoula, Montana, USA. Hodges, K. E. 2000. Ecology of snowshoe hares in southern boreal and montane forests. Pages 163-206 in Ruggiero, L. F., K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, and J. R. Squires, editors. Ecology and conservation of lynx in the United States. University Press of Colorado, Boulder, Colorado, USA. Hodges, K. E., C. J. Krebs, D. S. Hik, C. I. Stefan, E. A. Gillis, and C. E. Doyle. 2001. Snowshoe hare demography. Pages 141-178 in Krebs, C. J., S. Boutin, and R. Boonstra, editors. Ecosystem dynamics of the boreal forest: the Kluane project. Oxford University Press, New York, New York, USA. Homyack, J. A., D. J. Harrison, and W. B. Krohn. 2003. Effects of precommercial thinning on select wildlife species in northern Maine, with special emphasis on snowshoe hare. Maine Cooperative Fish and Wildlife Research Unit, Orono, Maine, USA. Huggins, R. M. 1989. On the statistical analysis of capture-recapture experiments. Biometrika 76:133140. Huggins, R. M. 1991. Some practical aspects of a conditional likelihood approach to capture experiments. Biometrics 47:725-732. Ivan, J. S. 2006. Density, demography, and seasonal movements of snowshoe hares in Colorado. Program narrative study plan for mammals research. Pages 27-45 in T.M. Shenk. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report, July: 1-45. Colorado Division of Wildlife, Fort Collins, Colorado, USA. Keith, L. B. 1964. Daily activity pattern of snowshoe hares. Journal of Mammalogy 45:626-627. Koehler, G. M. 1990a. Snowshoe hare, Lepus americanus, use of forest successional stages and population changes during 1985-1989 in North-central Washington. Canadian Field-Naturalist 105:291-293. Koehler, G. M. 1990b. Population and habitat characteristics of lynx and snowshoe hares in north central Washington. Canadian Journal of Zoology 68:845-851. Krebs, C. J., B. Scott Gilbert, S. Boutin, and e. al. R. Boonstra. 1987. Estimation of snowshoe hare population density from turd transects. Canadian Journal of Zoology 65:565-567. Litvaitis, J. A., J. A. Sherburne, and J. A. Bissonette. 1985. Influence of understory characteristics on snowshoe hare habitat use and density. Journal of Wildlife Management 49:866-873. Mech, L. D., K. L. Heezen, and D. B. Siniff. 1966. Onset and cessation of activity in cottontail rabbit and snowshoe hares in relation to sunset and sunrise. Animal Behaviour 14:410-413. Miller, M. A. 2005. Snowshoe hare habitat relationships in northwest Colorado. Thesis, Colorado State University, Fort Collins, Colorado, USA. Nudds, T. D. 1977. Quantifying the vegetative structure of wildlife cover. Wildlife Society Bulletin 34 �5:113-117. Otis, D. L., K. P. Burnham, G. C. White, and D. R. Anderson. 1978. Statisical inference from capture data on closed animal populations. Wildlife Monographs 62: Ruediger, B., J. Claar, S. Gniadek, B. Holt, Lewis Lyle, S. Mighton, B. Naney, G. Patton , T. Rinaldi, J. Trick, A. Vendehey, F. Wahl, N. Warren, D. Wenger, and A. Williamson. 2000. Canada lynx conservation assessment and strategy. U.S. Department of Agriculture, Forest Service, U.S. Department of Interior, Fish and Wildlife Service, Bureau of Land Management, National Park Service R1-00-53 U.S. Department of Agriculture, Forest Service, U.S. Department of Interior, Fish and Wildlife Service, Bureau of Land Management, National Park Service, Missoula, Montana, USA. Ruggiero, L. F., K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, and J. R. Squires. 2000. The scientific basis for lynx conservation: qualified insights. Pages 443-454 in Ruggiero, L. F., K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, and J. R. Squires, editors. Ecology and conservation of lynx in the United States. Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Collins, Colorado, USA. Samuel, M. D. and M. R. Fuller. 1996. Wildlife radiotelemetry. Pages 370-418 in Bookhout, T. A., editors. Research and Management Techniques for Wildlife and Habitats. Allen Press, Inc., Lawrence, Kansas, USA. Shenk, T. M. 2005. General locations of lynx (Lynx canadensis) reintroduced to southwestern Colorado from February 4, 1999 through February 1, 2005. Colorado Division of Wildlife Colorado Division of Wildlife, Fort Collins, Colorado, USA. Shenk, T. M. 2007. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report, July: 1-57. Colorado Division of Wildlife, Fort Collins, Colorado USA. Sullivan, T. P. and D. S. Sullivan. 1988. Influence of stand thinning on snowshoe hare population dynamics and feeding damage in lodgepole pine forest. Journal of Applied Ecology 25:791-805. Van Horne, B. 1983. Density as a misleading indicator of habitat quality. Journal of Wildlife Management 47:893-901. White, G. C., D. R. Anderson, K. P. Burnham, and D. L. Otis. 1982. Capture-recapture and removal methods for sampling closed populations. Los Alamos National Laboratory, Los Alamos, New Mexico, USA. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 Supplement:120-138. Williams, B. K., J. D. Nichols, and M. J. Conroy. 2002. Analysis and management of animal populations, Academic Press, San Diego, California, USA. Wilson, K. R. and D. R. Anderson. 1985. Evaluation of two density estimators of small mammal population size. Journal of Mammalogy 66:13-21. Wolfe, M. L., N. V. Debyle, C. S. Winchell, and T. R. McCabe. 1982. Snowshoe hare cover relationships in northern Utah. Journal of Wildlife Management 46:662-670. Wolff, J. O. 1980. The role of habitat patchiness in the population dynamics of snowshoe hares. Ecological Monographs 50:111-130. Zahratka, J. L. 2004. The population and habitat ecology of snowshoe hares (Lepus americanus) in the southern Rocky Mountains. Thesis, The University of Wyoming, Laramie, Wyoming, USA. Prepared by _________________________________________________ Jacob S. Ivan, Graduate Student, Colorado State University 35 �Figure 1. Purported high quality snowshoe hare habitat in Colorado. From left to right: small lodgepole pine, medium lodgepole pine, and large Engelmann spruce/subalpine fir. Figure 2. Study area near Taylor Park and Pitkin, Colorado including medium lodgepole (squares), small lodgepole (circles), and spruce/fir (triangles) stands selected for mark-recapture sampling. 36 �Jul Aug Sep 1 Jan 1 2 3 4 2 3 4 5 6 7 8 5 6 7 8 9 10 11 9 10 11 12 13 14 15 12 13 14 15 16 17 18 16 17 18 19 20 21 22 19 20 21 22 23 24 25 23 24 25 26 27 28 29 26 27 28 29 30 31 1 30 31 1 2 3 4 5 2 3 4 5 6 7 8 6 7 8 9 10 11 12 9 10 11 12 13 14 15 13 14 15 16 17 18 19 16 17 18 19 20 21 22 20 21 22 23 24 25 26 23 24 25 26 27 28 1 27 28 29 30 31 1 2 2 3 4 5 6 7 8 Feb Mar 3 4 5 6 7 8 9 9 10 11 12 13 14 15 10 11 12 13 14 15 16 16 17 18 19 20 21 22 17 18 19 20 21 22 23 23 24 25 26 27 28 29 24 25 26 27 28 29 30 30 31 Figure 3. Approximate annual data collection schedule for trapping () and telemetry (). Dates and weeks changed depending on calendar year and pay schedule. During telemetry work, the 6-person crew was divided into 2 teams, only one of which worked at any given time. Monthly locations on radiocollared hares were also collected in the interim between the intensive sampling periods indicated here. Figure 4. 15 trap locations ( ) on 7 12 trapping grid where vegetation was sampled by measuring stem density, horizontal cover, downed woody material, and basal area. Additionally, the 25-point grid superimposed on each of the 15 trap locations (inset) was used to quantify canopy covert). 37 �Figure 5. Snowshoe hare density and 95% confidence intervals in 3 types of stands in central Colorado as determined by ½ mean maximum distance moved, summer 2006 through winter 2008. Figure 6. Snowshoe hare survival and 95% confidence intervals across summer (S) and winter (W) sampling seasons in central Colorado as determined by mark-recapture, 2006-2008. 38 �Colorado Division of Wildlife July 2008- Aug 2009 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 0670 1 Federal Aid Project No. N/A : : : : : Division of Wildlife Mammals Research Lynx Conservation Post-Release Monitoring of Lynx Reintroduced to Colorado Period Covered: July 1, 2008 – August 31, 2009 Author: T. M. Shenk Personnel: O. Devineau, R. Dickman, P. Doherty, D. Freddy, L. Gepfert, J. Ivan, R. Kahn, A. Keith, P. Lukacs, G. Merrill, B. Smith, T. Spraker, S. Waters, G. White, L. Wolfe All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT In an effort to establish a viable population of Canada lynx (Lynx canadensis) in Colorado, the Colorado Division of Wildlife (CDOW) initiated a reintroduction effort in 1997 with the first lynx released in February 1999. From 1999-2006, 218 wild-caught lynx from Canada and Alaska were released in Colorado. We documented survival, movement patterns, reproduction, and landscape habitatuse through aerial (n = 11,580) and satellite (n = 29,258) tracking. Most lynx remained near the core release area in southwestern Colorado. From 1999-August 2009, there were 118 mortalities of released adult lynx. Approximately 29.7% were either human-induced or likely human-induced through either collisions with vehicles or shot. Starvation and disease/illness accounted for 18.6% of the deaths while 37.3% of the deaths were from unknown causes. Of these mortalities, 26.3% occurred outside of Colorado. Monthly mortality rate was lower inside the study area than outside, and slightly higher for male than for female lynx, although 95% confidence intervals for sexes overlapped. Mortality was higher immediately after release (first month = 0.0368 [SE = 0.0140] inside the study area, and 0.1012 [SE = 0.0359] outside the study area), and then decreased according to a quadratic trend over time. Reproductive females had the smallest 90% utilization distribution home ranges ( x = 75.2 km2, SE = 15.9 km2), followed by attending males ( x = 102.5 km2, SE = 39.7 km2) and non-reproductive animals ( x = 653.8 km2, SE = 145.4 km2). Reproduction was first documented in 2003 with subsequent successful reproduction in 2004, 2005, 2006 and 2009. No dens were documented in 2007 or 2008. From snow-tracking, the primary winter prey species (n = 604 kills) were snowshoe hare (Lepus americanus, annual x = 69.4%, SE = 5.6, n = 11) and red squirrel (Tamiasciurus hudsonicus, annual x = 22.6%, SE = 5.7, n = 11); other mammals and birds formed a minor part of the winter diet. Lynx usedensity surfaces were generated to illustrate relative use of areas throughout Colorado. Within the areas of high use in southwestern Colorado, site-scale habitat use, documented through snow-tracking, supports 1 �mature Engelmann spruce (Picea engelmannii)-subalpine fir (Abies lasiocarpa) forest stands with 4265% canopy cover and 15-20% conifer understory cover as the most commonly used areas in southwestern Colorado. Little difference in aspect (slight preference for north-facing slopes), slope ( x = 15.7°) or elevation ( x = 3173 m) were detected for long beds, travel and kill sites (n = 1841). Den sites (n = 37) however, were located at higher elevations ( x = 3354 m, SE = 31 m) on steeper ( x = 30°, SE = 2°) and more commonly north-facing slopes with a dense understory of coarse woody debris. Three years of a study to evaluate snowshoe hare densities, demography and seasonal movement patterns among small and medium tree-sized lodgepole pine (Pinus contorta) stands and mature spruce/fir stands have been completed in 2006-2009 (see Appendix I of this report). A pilot study to evaluate the efficacy of using minimally-invasive monitoring techniques was developed to estimate the extent, stability and potential distribution of lynx throughout Colorado. Results to date have demonstrated that CDOW has developed lynx release protocols that ensure high initial post-release survival followed by high long-term survival, site fidelity, reproduction and recruitment of Colorado-born lynx into the Colorado breeding population. What is yet to be demonstrated is whether Colorado can support sufficient recruitment to offset annual mortality for a viable lynx population over time. Monitoring continues in an effort to document such viability. 2 �WILDLIFE RESEARCH REPORT POST RELEASE MONITORING OF LYNX (LYNX CANADENSIS) REINTRODUCED TO COLORADO TANYA M. SHENK P. N. OBJECTIVE The initial post-release monitoring of Canada lynx (Lynx canadensis) reintroduced into Colorado will emphasize 5 primary objectives: 1. Assess and modify release protocols to ensure the highest probability of survival for each lynx released. 2. Obtain regular locations of released lynx to describe general movement patterns and habitats used by lynx. 3. Determine causes of mortality in reintroduced lynx. 4. Estimate survival of lynx reintroduced to Colorado. 5. Estimate reproduction of lynx reintroduced to Colorado. Three additional objectives will be emphasized after lynx display site fidelity to an area: 6. Refine descriptions of habitats used by reintroduced lynx. 7. Refine descriptions of daily and overall movement patterns of reintroduced lynx. 8. Describe hunting habits and prey of reintroduced lynx. Information gained to achieve these objectives will form a basis for the development of lynx conservation strategies in the southern Rocky Mountains. SEGMENT OBJECTIVES 1. Complete winter 2008-09 field data collection on lynx habitat use at the landscape scale, hunting behavior, diet, mortalities, and movement patterns. 2. Complete winter 2008-09 lynx trapping field season to collar Colorado born lynx and re-collar adult lynx. 3. Complete spring 2009 field data on lynx reproduction. 4. Summarize and analyze data and publish information as Progress Reports, peer-reviewed manuscripts for appropriate scientific journals, or CDOW technical publications. 5. Complete the third and final year of field work to evaluate snowshoe hare (Lepus americanus) densities, demography and seasonal movement patterns among small and medium tree-sized lodgepole pine stands and mature spruce/fir stands (see Appendix I). 6. Complete a pilot study to evaluate the efficacy of using minimally-invasive monitoring techniques to estimate the extent, stability and potential distribution of lynx throughout Colorado (see Appendix II). INTRODUCTION The Canada lynx occurs throughout the boreal forests of northern North America. Colorado represents the southern-most historical distribution of lynx, where the species occupied the higher elevation, montane forests in the state. Little was known about the population dynamics or habitat use of this species in their southern distribution. Lynx were extirpated or reduced to a few animals in the state by the late 1970’s due, most likely, to predator control efforts such as poisoning and trapping. Given the isolation of Colorado to the nearest northern populations, the CDOW considered reintroduction as the only option to attempt to reestablish the species in the state. 3 �A reintroduction effort was begun in 1997, with the first lynx released in Colorado in 1999. To date, 218 wild-caught lynx from Alaska and Canada have been released in southwestern Colorado. The goal of the Colorado lynx reintroduction program is to establish a self-sustaining, viable population of lynx in this state. Evaluation of incremental achievements necessary for establishing viable populations is an interim method of assessing if the reintroduction effort is progressing towards success. There are 7 critical criteria for achieving a viable population: 1) development of release protocols that lead to a high initial post-release survival of reintroduced animals, 2) long-term survival of lynx in Colorado, 3) development of site fidelity by the lynx to areas supporting good habitat in densities sufficient to breed, 4) reintroduced lynx must breed, 5) breeding must lead to reproduction of surviving kittens 6) lynx born in Colorado must reach breeding age and reproduce successfully, and 7) recruitment must equal or be greater than mortality over an extended period of time. The post-release monitoring program for the reintroduced lynx has 2 primary goals. The first goal is to determine how many lynx remain in Colorado and their locations relative to each other. Given this information and knowing the sex of each individual, we can assess whether these lynx can form a breeding core from which a viable population might be established. From these data we can also describe general movement patterns and habitat use. The second primary goal of the monitoring program is to estimate survival of the reintroduced lynx and, where possible, determine causes of mortality for reintroduced lynx. Such information will help in assessing and modifying release protocols and management of lynx once they have been released to ensure their highest probability of survival. Documenting reproduction is critical to the success of the program and lynx are monitored intensively to document breeding, births, survival and recruitment of lynx born in Colorado. Site-scale habitat descriptions of den sites are also collected and compared to other sites used by lynx. Lynx populations in Canada and Alaska have long been known to cycle in response to the 10-year snowshoe hare (Lepus americana) cycle (Elton and Nicholson 1942). Northern populations of lynx respond to snowshoe hare lows first through a decline in reproduction followed by an increase in adult mortality; when snowshoe hare populations increase, lynx respond with increased survival and reproduction (O’Donoghue et al. 2001). Therefore, annual survival and reproduction are highly variable but must be sufficient, overall, to result in long-term persistence of the population. It is not known if snowshoe hare populations in Colorado cycle and if so, where in the approximate 10-year cycle we are currently. Given this uncertainty, documenting persistence of lynx in Colorado for a period of at least 1015 years would provide support that a viable population of lynx can be sustained in Colorado even in the event snowshoe hares do cycle in the state. Therefore, to document the continued viability of lynx in Colorado beyond the initial reintroduction period, some form of long-term monitoring must be used to determine whether recruitment exceeds mortality for a period of time long enough to encompass possible snowshoe hare cycles. In addition, a challenge facing CDOW is how efforts should be allocated between focusing on monitoring the persistence of those lynx that have established within the core release area (Shenk 2007, Shenk 2008) and those lynx that may be pioneering and expanding into other portions of the state. Reproduction and known recruitment have been observed to be sporadic in the core area. To continue to document lynx reproduction through den site visits and to document survival of those kittens through tracking the adult females in winter looking for accompanying kittens requires a continued trapping effort to capture and radio-collar adult females. Lynx trapping is typically a time consuming and expensive operation as the lynx are territorial with large home ranges that may be entirely located within or largely comprised of inaccessible areas (e.g., wilderness areas). Alternatively, occupancy modeling using minimally-invasive techniques could be a feasible alternative for ascertaining trends in population status. 4 �Additional goals of the post-release monitoring program for lynx reintroduced to the southern Rocky Mountains included refining descriptions of habitat use and movement patterns and describing successful hunting habitat once lynx established home ranges that encompassed their preferred habitat. Specific objectives for the site-scale habitat data collection include: 1) describe and quantify site-scale habitat use by lynx reintroduced to Colorado, 2) compare site-scale habitat use among types of sites (e.g., kills vs. long-duration beds), and 3) compare habitat features at successful and unsuccessful snowshoe hare chases. The program will also investigate the ecology of snowshoe hare in Colorado. A study comparing snowshoe hare densities among mature stands of Engelmann spruce (Picea engelmannii)/subalpine fir (Abies lasiocarpa), lodgepole pine (Pinus contorta) and Ponderosa pine (Pinus ponderosa) was completed in 2004 with highest hare densities found in Engelmann spruce/subalpine fir stands and no hares found in Ponderosa pine stands. A study to evaluate the importance of young, regenerating lodgepole pine and mature Engelmann spruce/subalpine fir stands in Colorado by examining density and demography of snowshoe hares that reside in each was initiated in 2005 and will continue through 2009 (see Appendix I). Lynx is listed as threatened under the Endangered Species Act (ESA) of 1973, as amended (16 U. S. C. 1531 et. seq.)(U. S. Fish and Wildlife Service 2000). Colorado is included in the federal listing as lynx habitat. Thus, an additional objective of the post-release monitoring program is to develop conservation strategies relevant to lynx in Colorado. To develop these conservation strategies, information specific to the ecology of the lynx in its southern Rocky Mountain range, such as habitat use, movement patterns, mortality factors, survival, and reproduction in Colorado is needed. STUDY AREA Byrne (1998) evaluated five areas within Colorado as potential lynx habitat based on (1) relative snowshoe hare densities (Bartmann and Byrne 2001), (2) road density, (3) size of area, (4) juxtaposition of habitats within the area, (5) historical records of lynx observations, and (6) public issues. Based on results from this analysis, the San Juan Mountains of southwestern Colorado were selected as the core reintroduction area, and where all lynx were reintroduced. Wild Canada lynx captured in Alaska, British Columbia, Manitoba, Quebec and Yukon were transported to Colorado and held at The Frisco Creek Wildlife Rehabilitation Center located within the reintroduction area prior to release. Post-release monitoring efforts were focused in a 20,684 km2 study area which included the core reintroduction area, release sites and surrounding high elevation sites (> 2,591 m). The area encompassed the southwest quadrant of Colorado and was bounded on the south by New Mexico, on the west by Utah, on the north by interstate highway 70, and on the east by the Sangre de Cristo Mountains (Figure 1). Southwestern Colorado is characterized by wide plateaus, river valleys, and rugged mountains that reach elevations over 4,200 m. Engelmann spruce/subalpine fir is the most widely distributed coniferous forest type within the study area. The lynx-established core area is roughly bounded by areas used by lynx in the Taylor Park/Collegiate Peak areas in central Colorado and includes areas of continuous use by lynx, including areas used during breeding and denning (Figure 1). METHODS REINTRODUCTION Effort Wild Canada lynx were captured in Alaska, British Columbia, Manitoba, Quebec and Yukon and transported to Colorado where they were held at the Frisco Creek Wildlife Rehabilitation Center prior to release. All lynx releases were conducted under the protocols found to maximize survival (see Shenk 5 �2001). Estimated age, sex and body condition were ascertained and recorded for each lynx prior to release (see Wild 1999). Lynx were transported from the rehabilitation facility to their release site in individual cages. Specific release site locations were recorded in Universal Transverse Mercator (UTM) coordinates and identification of all lynx released at the same location, on the same day, was recorded. Behavior of the lynx on release and movement away from the release site were documented. Movement, Distribution and Relative Use of Areas by Lynx To monitor lynx movements and thus determine distribution and relative use of areas all released lynx were fitted with radio collars. All lynx released in 1999 were fitted with TelonicsTM radio-collars. All lynx released since 1999, with the exception of 5 males released in spring 2000, were fitted with SirtrackTM dual satellite/VHF radio-collars. These collars have a mortality indicator switch that operated on both the satellite and VHF mode. The satellite component of each collar was programmed to be active for 12 hours per week. The 12-hour active periods for individual collars were staggered throughout the week. Signals from the collars allowed for locations of the animals to be made via Argos, NASA, and NOAA satellites. The location information was processed by ServiceArgos and distributed to the CDOW through e-mail messages. Datasets.-- To determine recent (post-reintroduction) movement and distribution of lynx reintroduced, born or initially trapped in Colorado and relative use of areas by these lynx, regular locations of lynx were collected through a combination of aerial and satellite tracking. Locations were recorded and general habitat descriptions for each aerial location was recorded. The first dataset of lynx locations included all locations obtained from daytime flights conducted with a Cessna 185 or similar aircraft to locate lynx by their VHF collar transmitters (hereafter aerial locations). VHF transmitters have been used on lynx since the first lynx were released in February 1999. The second type of lynx location data was collected via satellite from the satellite collar transmitters placed on the lynx (hereafter satellite locations). Satellite transmitter collars were first used for lynx in April 2000. These satellite collars also contained a VHF transmitter which also allowed locating lynx from the air or ground. All locations were recorded in Universal Transverse Mercator (UTM) coordinates using the CONUS NAD27 datum. Flights to obtain lynx aerial locations were typically conducted on a weekly basis throughout most summer and winter months and twice a week during the den search field season (May 15 – June 30), depending on weather and availability of planes and pilots. Flights were typically concentrated in the high elevation (> 2700 m) southwest quadrant of Colorado which encompasses the core lynx release and research area (Figure 1). Flights during the den seasons were conducted to obtain locations on all female lynx within the state wearing an active VHF transmitter. VHF transmitters were outfitted with sufficient batteries to last 60 months. The satellite transmitters were designed to provide locations on a weekly basis with sufficient batteries to last for 18 months. These data collections remain ongoing and all information will be used for future habitat use and survival analyses. Accuracy of both aerial and satellite locations varied with the environmental conditions at the time the location was obtained. Accuracy of aerial locations was influenced by weather with accuracy ranging from 50 - 500 meters. Satellite location accuracy was also influenced by atmospheric conditions and position of the satellites. Satellite location accuracy ranged from 150 meters -10 km. Movement and Distribution.-- To document all known lynx locations maps were generated with all aerial and satellite locations displayed. Due to lynx movements outside of Colorado, particularly into the states of New Mexico, Utah and Wyoming we further evaluated lynx use throughout those three states, as well as the data would allow. All individual lynx located at least once in these 3 states (nontruncated datasets) were identified and tallied for each year. To document consistency and known use of these states after the initial effect of being reintroduced was minimized (i.e., 180 days post-release), each individual lynx located at least once in these states from the truncated datasets were identified and tallied. 6 �Relative Use.-- To document relative use of areas by lynx, 90% kernel use-density surfaces were calculated for truncated satellite and aerial lynx locations using the ArcGIS Spatial Analyst Kernel Density Tool. Lynx may not be exhibiting typical behavior or habitat use within the first few months after their release in Colorado. Therefore, a subset of each of the aerial and satellite datasets was created that eliminated the first 180 days (approximately 6 months) of locations obtained for each lynx immediately after their initial release. As a result, the truncated aerial location dataset contained lynx locations from September 1999 through April 2009 while the truncated satellite location dataset began October 2000 and extended through April 2009. Due to differences in data collection frequency and accuracy between datasets, the truncated satellite and truncated aerial data were analyzed separately for generating the lynx use-density surfaces. These use-density surfaces fit a smoothly curved surface over each lynx location. The surface value was highest at the location of the point and diminished with increasing distance from the point. A fixed kernel was used with a smoothing parameter of 5 km, reaching 0 at the search radius distance from the point. Only a circular neighborhood was possible. The volume under the surface equaled the total value for the point. The use-density at each output GIS raster cell was calculated by adding the values of all the kernel surfaces from all the lynx point locations that overlaid each raster cell center. The kernel function was based on the quadratic kernel function described in Silverman (1986, p. 76, equation 4.5). The use-density surfaces were calculated at 100 m resolution. To enhance graphic displays of higher usedensity areas, density values representing single locations were not displayed. Home Range Preliminary estimates of annual home ranges were calculated as a 95% utilization distribution using a kernel home-range estimator for each lynx we had at least 30 locations for within a year. A year was defined as March 15 – March 14 of the following year. Locations used in the analyses were collected from September 1999 – January 2006 and all locations obtained for an individual during the first six months after its release were eliminated from any home range analyses as it was assumed movements of lynx initially post-release may not be representative of normal habitat use. Locations were obtained either through aerial VHF surveys or locations or the midpoint (ArcView Movement Extension) of all high quality (accuracy rating of 0-1km) satellite locations obtained within a single 24-hour period. All locations used within a single home range analysis were taken a minimum of 24 hours apart. Home range estimates were classified as being for a reproductive or non-reproductive animal. A reproductive female was defined as one that had kittens with her; a reproductive male was defined as a male whose movement patterns overlapped that of a reproductive female. If a litter was lost within the defined year a home range described for a reproductive animal were estimated using only locations obtained while the kittens were still with the female. Final estimates of annual home range size will completed with the addition of data collected through 2009 and in conjunction with current habitat use analyses and publications to be completed in 2009-2010. Survival Multi-state mark-recapture models were used to estimate monthly mortality rates and described in detail in Devineau et al. 2009a (in review) for the first year post-release and for 10 years post-release in Devineau et al. 2009b (in review). This approach accommodated missing data and allowed exploration of factors possibly affecting lynx survival such as sex, time spent in pre-release captivity, movement patterns, and origin. Mortality Factors When a mortality signal (75 beats per minute [bpm] vs. 50 bpm for the Telonics™ VHF transmitters, 20 bpm vs. 40 bpm for the Sirtrack™ VHF transmitters, 0 activity for Sirtrack™ PTT) was heard during either satellite, aerial or ground surveys, the location (UTM coordinates) was recorded. 7 �Ground crews then located and retrieved the carcass as soon as possible. The immediate area was searched for evidence of other predators and the carcass photographed in place before removal. Additionally, the mortality site was described and habitat associations and exact location were recorded. Any scat found near the dead lynx that appeared to be from the lynx was collected. All carcasses were transported to the Colorado State University Veterinary Teaching Hospital (CSUVTH) for a post mortem exam to 1) determine the cause of death and document with evidence, 2) collect samples for a variety of research projects, and 3) archive samples for future reference (research or forensic). The gross necropsy and histology were performed by, or under the lead and direct supervision of a board certified veterinary pathologist. At least one research personnel from the CDOW involved with the lynx program was also present. The protocol followed standard procedures used for thorough post-mortem examination and sample collection for histopathology and diagnostic testing (see Shenk 1999 for details). Some additional data/samples were routinely collected for research, forensics, and archiving. Other data/samples were collected based on the circumstances of the death (e.g., photographs, video, radiographs, bullet recovery, samples for toxicology or other diagnostic tests, etc.). From 1999–2004 the CDOW retained all samples and carcass remains with the exception of tissues in formalin for histopathology, brain for rabies exam, feces for parasitology, external parasites for ID, and other diagnostic samples. Since 2005 carcasses are disposed of at the CSUVTH with the exception of the lower canine, fecal samples, stomach content samples and tissue or bone marrow samples to be delivered by CDOW to the Center for Disease Control for plague testing. The lower canine, from all carcasses, is sent to Matson Labs (Missoula, Montana) for aging and the fecal and stomach content samples are evaluated for diet. Reproduction Females were monitored for proximity to males during each breeding season. We defined a possible mating pair as any male and female documented within at least 1 km of each other in breeding season through either flight data or snow-tracking data. Females were then monitored for site fidelity to a given area during each denning period of May and June. Each female that exhibited stationary movement patterns in May or June were closely monitored to locate possible dens. Dens were found when field crews walked in on females that exhibited virtually no movement for at least 10 days from both aerial and ground telemetry. Kittens found at den sites were weighed, sexed and photographed. Each kitten was uniquely marked by inserting a sterile passive integrated transponder (PIT, Biomark, Inc., Boise, Idaho, USA) tag subcutaneously between the shoulder blades. Time spent at the den was minimized to ensure the least amount of disturbance to the female and the kittens. Weight, PIT-tag number, sex and any distinguishing characteristics of each kitten was also recorded. Beginning in 2005, blood and saliva samples were collected and archived for genetic identification. During the den site visits, den site location was recorded as UTM coordinates. General vegetation characteristics, elevation, weather, field personnel, time at the den, and behavioral responses of the kittens and female were also recorded. Once the females moved the kittens from the natal den area, den sites were visited again and site-specific habitat data were collected (see Habitat Use section below). Captures Captures were attempted for either lynx that were in poor body condition or lynx that needed to have their radio-collars replaced due to failed or failing batteries or to radio-collar kittens born in Colorado once they reached at least 10-months of age when they were nearly adult size. Methods of recapture included 1) trapping using a Tomahawk™ live trap baited with a rabbit and visual and scent lures, 2) calling in and darting lynx using a Dan-Inject CO2 rifle, 3) custom box-traps modified from those 8 �designed by other lynx researchers (Kolbe et al. 2003) and 4) hounds trained to pursue felids were also used to tree lynx and then the lynx was darted while treed. Lynx were immobilized either with Telazol (3 mg/kg; modified from Poole et al. 1993 as recommended by M. Wild, DVM) or medetomidine (0.09mg/kg) and ketamine (3 mg/kg; as recommended by L. Wolfe, DVM)) administered intramuscularly (IM) with either an extendible pole-syringe or a pressurized syringe-dart fired from a Dan-Inject air rifle. Immobilized lynx were monitored continuously for decreased respiration or hypothermia. If a lynx exhibited decreased respiration 2mg/kg of Dopram was administered under the tongue; if respiration was severely decreased, the animal was ventilated with a resuscitation bag. If medetomidine/ketamine were the immobilization drugs, the antagonist Atipamezole hydrochloride (Antisedan) was administered. Hypothermic (body temperature < 95o F) animals were warmed with hand warmers and blankets. While immobilized, lynx were fitted with replacement SirtrackTM VHF/satellite collar and blood and hair samples were collected. Once an animal was processed, recovery was expedited by injecting the equivalent amount of the antagonist Antisedan IM as the amount of medetomidine given, if medetomodine/ketemine was used for immobilization. Lynx were then monitored while confined in the box-trap until they were sufficiently recovered to move safely on their own. No antagonist is available for Telezol so lynx anesthetized with this drug were monitored until the animal recovered on its own in the box-trap and then released. If captured and in poor body condition, lynx were anesthetized with either Telezol (2 mg/kg) or medetomodine/ketemine and returned to the Frisco Creek Wildlife Rehabilitation Center for treatment. HABITAT USE Gross habitat use was documented by recording canopy vegetation at aerial locations. More refined descriptions of habitat use by reintroduced lynx were obtained through following lynx tracks in the snow (i.e., snow-tracking) and site-scale habitat data collection conducted at sites found through this method to be used by lynx. See Shenk (2006) for detailed methodologies. DIET AND HUNTING BEHAVIOR Winter diet of reintroduced lynx was estimated by documenting successful kills through snowtracking. Prey species from failed and successful hunting attempts were identified by either tracks or remains. Scat analysis also provided information on foods consumed. Scat samples were collected wherever found and labeled with location and individual lynx identification. Only part of the scat was collected (approximately 75%); the remainder was left in place in the event that the scat was being used by the animal as a territory mark. Site-scale habitat data collected for successful and unsuccessful snowshoe hare kills were compared. SNOWSHOE HARE ECOLOGY To further our understanding of snowshoe hare ecology in Colorado, a study was conducted comparing snowshoe hare densities among mature stands of Engelmann spruce/subalpine fir, lodgepole pine (Pinus contorta) and Ponderosa pine (Pinus ponderosa). The highest hare densities were found in Engelmann spruce/subalpine fir stands and no hares found in Ponderosa pine stands (Zahratka and Shenk 2008). A second study was initiated in 2005 to evaluate the importance of young, regenerating lodgepole pine and mature Engelmann spruce / subalpine fir stands in Colorado by examining density and demography of snowshoe hares that reside in each (Ivan 2005). Specifically, this study was designed to evaluate small and medium lodgepole pine stands and large spruce/fir stands where the classes “small”, “medium”, and “large” refer to the diameter at breast height (dbh) of overstory trees as defined in the United States Forest Service R2VEG Database (small = 2.54−12.69 cm dbh, medium = 12.70−22.85 cm, and large = 22.86−40.64 cm dbh; J. Varner, United States Forest Service, personal communication). The study design was also developed to identify which 9 �of the numerous hare density-estimation procedures available perform accurately and consistently using an innovative, telemetry augmentation approach as a baseline. In addition, movement patterns and seasonal use of deciduous cover types such as riparian willow were assessed. Finally, the study was designed to further expound on the relationship between density, demography, and stand-type by examining how snowshoe hare density and demographic rates vary with specific vegetation, physical, and landscape characteristics of a stand. RESULTS REINTRODUCTION Effort From 1999 through 2006, 218 wild-caught lynx were reintroduced into southwestern Colorado (Table 1). No lynx were released in 2007, 2008 or 2009. All lynx were released with either VHF or dual VHF/satellite radio collars so they could be monitored for movement, reproduction and survival. The CDOW does not plan to release any additional lynx in 2010. Movement Patterns and Distribution Numerous travel corridors were used repeatedly by more than one lynx. These travel corridors include the Cochetopa Hills area for northerly movements, the Rio Grande Reservoir-SilvertonLizardhead Pass for movements to the west, and southerly movements down the east side of Wolf Creek Pass to the southeast through the Conejos River Valley. Lynx appear to remain faithful to an area during winter months, and exhibit more extensive movements away from these areas in the summer. A total of 11,580 aerial and 29,258 satellite locations were obtained from the 218 reintroduced lynx, radio-collared Colorado kittens (n = 16) and unmarked lynx captured in Colorado (n = 3) as of August 31, 2009. The majority of these locations were in Colorado (Figure 2). Some reintroduced lynx dispersed outside of Colorado into Arizona, Idaho, Iowa, Kansas, Montana, Nebraska, Nevada, New Mexico, South Dakota, Utah and Wyoming (Figure 2). The majority of surviving lynx from the reintroduction effort currently continue to use high elevation (> 2900 m), forested terrain in an area bounded on the south by New Mexico north to Independence Pass, west as far as Taylor Mesa and east to Monarch Pass. Most movements away from the Core Release Area were to the north. Relative Use The lynx use-density surfaces resulting from the fixed kernel analyses provided relative probabilities of finding lynx in areas throughout their distribution. All 218 lynx released in Colorado, all radio-collared kittens and 3 captured unmarked adults were located at least once in Colorado. The majority of these lynx remained in Colorado. Single use density surfaces were calculated for both truncated aerial and truncated satellite datasets in Colorado up to March 2007 and presented in Shenk (2008). Relative use-density surfaces were also generated for New Mexico, Wyoming and Utah and presented in detail in Shenk (2007). Aerial and satellite use-density surfaces indicated similar high usedensity areas. Satellite locations indicated broader spatial use by lynx because satellite collars provided more locations than flights. A single use-density surface was calculated for the satellite non-truncated dataset from April 2000-April 2009 (n = 18,240). The use-density surface was displayed for the satellite non-truncated dataset in Colorado (Figure 3) and for all documented use (Figure 4). The use-density surface for lynx use in Colorado indicates two primary areas of use. The first is the Core Research Area (see Figure 1) and a secondary core centered in the Collegiate Peaks Wilderness (Figures 1, 3 and 4). High use is also documented for 1) the area east of Dillon, on both the north and south sides of I70 and 2) the area north of Hwy 50 centered around Gunnison and then north to Crested Butte. These last 2 high use areas are smaller in extent than the 2 core areas. 10 �Home Range Reproductive females had the smallest 90% utilization distribution annual home ranges ( x = 75.2 km2, SE = 15.9 km2, n = 19), followed by attending males ( x = 102.5 km2, SE = 39.7 km2, n = 4). Nonreproductive females had the largest annual home ranges ( x = 703.9 km2, SE = 29.8 km2, n = 32) followed by non-reproductive males ( x = 387.0 km2, SE = 73.5 km2, n = 6). Combining all nonreproductive animals yielded a mean annual home range of 653.8 km2 (SE = 145.4 km2, n = 38). Survival Detailed analysis of lynx mortality was completed and described in Devineau et al. 2009a (in review) to evaluate how the different release protocols used to reintroduce lynx in Colorado (Shenk 2001) affected mortality within the first year post-release. Average monthly mortality in the study area during the first year decreased with time in captivity from 0.205 [95% CI 0.069, 0.475] for lynx having spent up to 7 days in captivity to 0.028 [95% CI 0.012, 0.064] for lynx spending > 45 days in captivity before release (Devineau et al. 2009). The results also suggest that keeping lynx in captivity beyond 5 or 6 weeks accrued little benefit in terms of monthly survival. On a monthly average basis, lynx were as likely to move out (probability = 0.196, SE=0.032) as well as back on (probability = 0.143, SE=0.034) the reintroduction area (i.e., study area) during the first year after release. Mortality was 1.6x greater outside of the reintroduction area. Detailed analysis of lynx mortality over the first 10 years post-reintroduction was completed and described in Devineau et al. 2009b (in review). In summary, monthly mortality rate was lower inside the study area than outside, and slightly higher for male than for female lynx, although 95% confidence intervals for sexes overlapped. Mortality was higher immediately after release (first month = 0.0368 [SE = 0.0140]; inside the study area, and 0.1012 [SE = 0.0359] outside the study area), and then decreased according to a quadratic trend over time. As of August 31, 2009, CDOW was actively monitoring/tracking 37 of the 100 lynx still possibly alive (Table 2). There are 61 lynx that we have not heard signals on since at least August 31, 2008 and these animals are classified as ‘missing’ (Table 2). One of these missing lynx is a mortality of unknown identity, thus only 60 are truly missing. Possible reasons for not locating these missing lynx include 1) long distance dispersal, beyond the areas currently being searched, 2) radio failure, or 3) destruction of the radio (e.g., run over by car). CDOW continues to search for all missing lynx during both aerial and ground searches. Two of the missing lynx released in 2000 are thought to have slipped their collars. Mortality Factors Of the total 218 adult lynx released, we have 118 known mortalities as of August 31, 2009 (Table 2). Starvation was a significant cause of mortality in the first year of releases only. The primary known causes of death included 29.7% human-induced deaths which were confirmed or probably caused by collisions with vehicles or gunshot (Table 3). Malnutrition and disease/illness accounted for 18.6% of the deaths. An additional 37.3% of known mortalities were from unknown causes. Mortalities occurred throughout the areas through which lynx moved, with 26.3% (n=31) occurring outside of Colorado. The out of state mortalities included 14 in New Mexico, 5 in Utah, 4 in Wyoming and Nebraska, and 1 each in Arizona, Kansas, Iowa and Montana (Figure 2, Table 4). 11 �Reproduction Reproduction was first documented in 2003 when 6 dens and a total of 16 kittens were found in the lynx Core Release Area in southwestern Colorado. Reproduction was also documented in 2004, 2005, 2006, and 2009. No dens were found in 2007 or 2008 (Table 5). Field crews weighed, photographed, PIT-tagged the kittens and checked body condition. Beginning in 2005, we also collected blood samples from the kittens for genetic work in an attempt to confirm paternity. Kittens were processed as quickly as possible (11-32 minutes) to minimize the time the kittens were without their mother. While working with the kittens the females remained nearby, often making themselves visible to the field crews. The females generally continued a low growling vocalization the entire time personnel were at the den. In all cases, the female returned to the den site once field crews left the area. At all dens the females appeared in excellent condition, as did the kittens. The kittens weighed from 270-500 grams. Lynx kittens weigh approximately 200 grams at birth and do not open their eyes until they are 10-17 days old. The proportion of tracked females found with litters in 2006 was lower (0.095) than in the 3 previous years (0.413, SE = 0.032, Table 5). However, all demographic and habitat characteristics measured at the 4 dens that were found in 2006 were comparable to all other dens found. Mean number of kittens per litter from 2003-2006 was 2.78 (SE = 0.05) and sex ratio of females to males was equal ( x = 1.14, SE = 0.14). More details of reproduction in 2003-06 were presented in Shenk (2007). No dens were found in either 2007 or 2008, even though up to 34 adult females were monitored intensively during the denning period (Table 5). In 2009, 22.7% of females being monitored (n = 22) had dens. Two kittens were found at each of these 5 dens, a decrease in the mean of 2.78 (SE= 0.05) kittens per litter found in other years. Sex ratio was also more biased towards female kittens in 2009 (0.4 males/females) than found in previous years. Den Sites.-- A total of 42 dens were found from 2003-2009. All of the dens except one have been scattered throughout the high elevation areas of Colorado, south of I-70. In 2004, 1 den was found in southeastern Wyoming, near the Colorado border. Habitat measurements conducted through 2006 (n=37) document that dens were located on steep ( x slope = 30o , SE=2o), north-facing, high elevation ( x = 3354 m, SE = 31 m) slopes. The dens were typically in Engelmann spruce/subalpine fir forests in areas of extensive downfall of coarse woody debris (Shenk 2006). All dens (n = 42) were located within the winter use areas used by the females. Captures Two adult lynx were captured in 2001 for collar replacement. One lynx was captured in a tomahawk live-trap, the other was treed by hounds and then anesthetized using a jab pole. Five adult lynx were captured in 2002; 3 were treed by hounds and 2 were captured in padded leghold traps. In 2004, 1 lynx was captured with a Belisle snare and 6 adult lynx were captured in box-traps. Trapping effort was substantially increased in winter and spring 2005 and 12 adult lynx were captured and re-collared. Eight reintroduced lynx were captured in winter and spring 2006. In 2007, 11 reintroduced adult lynx were captured and re-collared; 10 in 2008 and 11 in 2009. All lynx captured in Colorado from 2005-2009were caught in box-traps. In addition, as part of the collaring trapping effort, 16 Colorado-born kittens were captured and collared at approximately 10-months of age. Seven 2004-born kittens were collared in spring 2005; 7 2005-born kittens were collared in spring 2006; and 1 2004- and 1 2005 born kitten were first captured and collared in 2009. We also captured 3 adults (approximate age 2 years old) in winters 2006-09 that had no PIT-tags or radio collars. We assume these 3 lynx were from litters born in Colorado that were 12 �never found at dens (i.e., why there were no PIT-tags). All lynx captured for collaring or re-collaring were fitted with new Sirtrack TM dual VHF/satellite collars and re-released at their capture locations. Seven adult lynx were captured from March 1999-August 31, 2009 because they were in poor body condition (Table 6). Five of these lynx were successfully treated at the Frisco Creek Rehabilitation Center and re-released in the Core Release Area. One lynx, BC00F07, died from starvation and hypothermia within 1 day of capture at the rehabilitation center. Lynx QU04M07 died 3 days after capture at the rehabilitation center. Necropsy results documented starvation as the cause of death for this lynx that was precipitated by hydrocephalus and bronchopneumonia (unpublished data T. Spraker, CSUVTH). There were no apparent commonalities among these animals. Seven lynx were captured (either by CDOW personnel or conservation personnel in other states) because they were in atypical habitat outside the state of Colorado (Table 6). They were held at Frisco Creek Rehabilitation Center for a minimum of 3 weeks, fitted with new Sirtrack TM dual VHF/satellite collars and re-released in the Core Release Area in Colorado. Five of these 7 lynx were still alive 6 months post-re-release but 3 had already dispersed out of Colorado and 1 stayed in Colorado through August 31, 2009. Two of these lynx died within 6 months of re-release: 1 died of starvation in Colorado and the other died of unknown causes in Nebraska. One lynx captured out of state and re-released currently remains in Colorado. HABITAT USE Landscape-scale daytime habitat use was documented from 9496 aerial locations of lynx collected from February 1999-June 30, 2007. Throughout the year Engelmann spruce - subalpine fir was the dominant cover used by lynx. A mix of Engelmann spruce, subalpine fir and aspen (Populus tremuloides) was the second most common cover type used throughout the year. Various riparian and riparian-mix areas were the third most common cover type where lynx were found during the daytime flights. Use of Engelmann spruce-subalpine fir forests and Engelmann spruce-subalpine fir-aspen forests was similar throughout the year. There was a trend in increased use of riparian areas beginning in July, peaking in November, and dropping off December through June. Site-scale habitat data collected from snow-tracking efforts indicate Engelmann spruce and subalpine fir were also the most common forest stands used by lynx for all activities during winter in southwestern Colorado. Comparisons were made among sites used for long beds, dens, travel and where they made kills. Little difference in aspect, mean slope and mean elevation were detected for 3 of the 4 site types including long beds, travel and kills where lynx typically use gentler slopes ( x = 15.7o ) at a mean elevation of 3173 m, and varying aspects with a slight preference for north-facing slopes. See Shenk (2006) for more detailed analyses of habitat use. DIET AND HUNTING BEHAVIOR Winter diet of lynx was documented through detection of kills found through snow-tracking. Prey species from failed and successful hunting attempts were identified by either tracks or remains. Scat analysis also provided information on foods consumed. A total of 604 kills were located from February 1999-April 2009. We collected over 990 scat samples from February 1999-April 2009 that will be analyzed for content. In each winter, the most common prey item was snowshoe hare, followed by red squirrel (Tamiusciurus hudsonicus; Table 7). The percent of snowshoe hare kills found however, varied annually from a low of 30.4% in 2009 to a high of 90.77% in winter 2002-2003. An annual mean of 69.39% (SE = 5.6) snowshoe hare kills in the diet has been documented. A comparison of percent overstory for successful and unsuccessful snowshoe hare chases indicated lynx were more successful at sites with slightly higher percent overstory, if the overstory 13 �species were Englemann spruce, subalpine fir or willow. Lynx were slightly less successful in areas of greater aspen overstory. This trend was repeated for percent understory at all 3 height categories except that higher aspen understory improved hunting success. Higher density of Engelmann spruce and subalpine fir increased hunting success while increased aspen density decreased hunting success. SNOWSHOE HARE ECOLOGY Three years of a 3-year study to evaluate snowshoe hare densities, demography and seasonal movement patterns among small and medium tree-sized lodgepole pine stands and mature spruce/fir stands have been completed and preliminary results presented (see Appendix I). DISCUSSION In an effort to establish a viable population of lynx in Colorado, CDOW initiated a reintroduction effort in 1997 with the first lynx released in winter 1999. From 1999 through spring 2006, 218 lynx were released in the Core Release Area. Locations of each lynx were collected through aerial- or satellite-tracking to document movement patterns and to detect mortalities. Most lynx remain in the high elevation, forested areas in southwestern Colorado. The use-density surfaces for lynx use in Colorado indicate two primary areas of use. The first is the Core Research Area (see Figure 1) and a secondary core centered in the Collegiate Peaks Wilderness (Figures 1, 3, 4). High use is also documented for 1) the area east of Dillon, on both the north and south sides of I70 and 2) the area north of Hwy 50 centered around Gunnison and then north to Crested Butte. These last 2 high use areas are smaller in extent than the 2 core areas. Dispersal movement patterns for lynx released in 2000 and subsequent years were similar to those of lynx released in 1999 (Shenk 2000). However, more animals released in 2000 and subsequent years remained within the Core Release Area than those released in 1999. This increased site fidelity may have been due to the presence of con-specifics in the area on release. Numerous travel corridors within Colorado have been used repeatedly by more than 1 lynx. These travel corridors include the Cochetopa Hills area for northerly movements, the Rio Grande Reservoir-Silverton-Lizardhead Pass for movements to the west, and southerly movements down the east side of Wolf Creek Pass to the southeast to the Conejos River Valley. Lynx appear to remain faithful to an area during winter months, and exhibit more extensive movements away from these areas in the summer. Reproductive females had the smallest 90% utilization distribution home ranges ( x = 75.2 km2, SE = 15.9 km2), followed by attending males ( x = 102.5 km2, SE = 39.7 km2) and non-reproductive animals ( x = 653.8 km2, SE = 145.4 km2). Most lynx currently being tracked are within the Core Release Area. During the summer months, lynx were documented to make extensive movements away from their winter use areas. Extensive summer movements away from areas used throughout the rest of the year have been documented in native lynx in Wyoming and Montana (Squires and Laurion 1999). Current data collection methods used for the Colorado lynx reintroduction program were not specifically designed to address the reintroduced lynx movements or use of areas in other states. In particular, the core research and release area were in Colorado. Therefore, the number of aerial locations obtained would be far fewer in other states than in Colorado which would bias low the number of lynx and intensity of lynx use documented outside the state. In contrast, obtaining satellite locations is not biased by the location of the lynx. Satellite locations are, however, biased by the shorter time the satellite transmitters function, approximately 18 months versus 60 months for the VHF transmitters used to obtain the aerial locations. However, data collected to meet objectives of the lynx reintroduction program were 14 �used to provide information to help address the question of lynx use outside of Colorado. Due to the rarity of flights conducted outside Colorado, only use-density surfaces generated from satellite locations were used to document relative lynx use of areas in New Mexico, Utah and Wyoming. New Mexico and Wyoming have been used continuously by lynx since the first year lynx were released in Colorado (1999) to the present. Lynx reintroduced in Colorado were first documented in Utah in 2000 and are still being documented there to date. In addition, all levels of lynx use-density documented throughout Colorado are also represented in New Mexico, Utah and Wyoming from none to the highest level of use (Shenk 2007). One den was found in Wyoming. Although no reproduction has been documented in New Mexico or Utah to date, documenting areas of the highest intensity of use and the continuous presence of lynx within these states for over six years does suggest the potential for yearround residency of lynx and reproduction in those states. From 1999-August 2009, there were 118 mortalities of released adult lynx. Human-caused mortality factors are currently the highest causes of death with approximately 29.7% attributed to collisions with vehicles or gunshot. Starvation and disease/illness accounted for 18.6% of the deaths while 37.3% of the deaths were from unknown causes. Lynx mortalities were documented throughout all areas lynx used, including 31 (26.3%) occurring in other states (Figure 2, Table 3). Nearly half (14 of 30) of the out-of-state mortalities were documented in New Mexico. Detailed analysis of lynx mortality was completed and described in Devineau et al. 2009a to evaluate how the different release protocols used to reintroduce lynx in Colorado (Shenk 2002) affected mortality within the first year post-release. Average monthly mortality in the study area during the first year decreased with time in captivity from 0.205 [95% CI 0.069, 0.475] for lynx having spent up to 7 days in captivity to 0.028 [95% CI 0.012, 0.064] for lynx spending > 45 days in captivity before release (Devineau et al. 2009a). The results also suggest that keeping lynx in captivity beyond 5 or 6 weeks accrued little benefit in terms of monthly survival. On a monthly average basis, lynx were as likely to move out (probability = 0.196, SE=0.032) as well as back on (probability = 0.143, SE=0.034) the reintroduction area during the first year after release. Mortality was 1.6x greater outside of the study area suggesting that permanent emigration and differential mortality rates on and off reintroduction areas should be factored into sample size calculations for an effective reintroduction effort. A post-release monitoring plan is critical to providing information to assess aspects of release protocols in order to improve the survival of individuals. Future lynx, as well as other carnivore, reintroductions may use our results to help design reintroduction programs including both their release and post-release monitoring protocols. Over the 10 years of the reintroduction effort, monthly mortality rate was lower inside the study area than outside, and slightly higher for male than for female lynx, although 95% confidence intervals for sexes overlapped (Devineau et al. 2009b). Mortality was higher immediately after release (first month = 0.0368 [SE = 0.0140] inside the study area, and 0.1012 [SE = 0.0359] outside the study area), and then decreased according to a quadratic trend over time (Devineau et al. 2009, in review). Reproduction is critical to achieving a self-sustaining viable population of lynx in Colorado. Reproduction was first documented from the 2003 reproduction season and again in 2004, 2005 and 2006. Lower reproduction occurred in 2006 (Table 5) but did include a Colorado-born female giving birth to 2 kittens, documenting the first recruitment of Colorado-born lynx into the Colorado breeding population. No reproduction was documented in 2007 or 2008. The cause of the decreased reproduction from 2006 08 is unknown. One possible explanation would be a decrease in prey abundance. Reproduction was again observed in 2009 with 5 dens and 10 kittens found in Colorado. Litter size was smaller than previously documented with only2 kittens found in each litter in comparison to a mean of 2.78 found in previous years. In addition, a sex bias towards female kittens was evident in 2009 which was not evident 15 �in prior years. Two litters found in 2009 had both parents born in Colorado, resulting in the first documented third generation Colorado lynx from the reintroduction. Additional reproduction is likely to have occurred in all years from females we were no longer tracking, and from Colorado-born lynx that have not been collared. The dens we find are more representative of the minimum number of litters and kittens in a reproduction season. To achieve a viable population of lynx, enough kittens need to be recruited into the population to offset the mortality that occurs in that year and hopefully even exceed the mortality rate to achieve an increasing population. The use-density surfaces depict intensity of use by location. Why certain areas would be used more intensively than others should be explained by the quality of the habitat in those areas. Characteristics of areas used by lynx, as documented through aerial locations and snow-tracking of lynx in the Colorado core research area, include mature Engelmann spruce-subalpine fir forest stands with 4265% canopy cover and 15-20% conifer understory cover (Shenk 2006). Within these forest stand types, lynx appear to have a slight preference for north-facing, moderate slopes ( x = 15.7°) at high elevations ( x = 3173 m; Shenk 2006). Snow-tracking of released lynx also provided information on hunting behavior and diet through documentation of kills, food caches, chases, and diet composition estimated through prey remains. Primary winter prey species (n = 604) were snowshoe hare and red squirrel (Table 7), which comprised 69.4% (SE = 5.6, n = 11) and 22.6.2% (SE = 5.7, n = 11) of the annual diet, respectively. Thus, areas of good habitat must also support populations of snowshoe hare and red squirrel. In winter, lynx reintroduced to Colorado appear to be feeding on their preferred prey species, snowshoe hare and red squirrel in similar proportions as those reported for northern lynx during lows in the snowshoe hare cycle (Aubry et al. 1999). Environmental conditions in the springs and summers of 2003, 2006 and 2008 resulted in high cone crops during their following winters based on field observations, resulting in increased red squirrel abundance. This may partially explain the higher percent of red squirrel kills, and thus a lower percent of snowshoe hare kills, found in winters 2003-04, 2006-07 and 2008-09 (Table 7). Caution must be used in interpreting the proportion of identified kills. Such a proportion ignores other food items that are consumed in their entirety and thus are biased towards larger prey and may not accurately represent the proportion of smaller prey items, such as microtines, in lynx winter diet. Through snow-tracking we have evidence that lynx are mousing and several of the fresh carcasses have yielded small mammals in the gut on necropsy. The summer diet of lynx has been documented to include less snowshoe hare and more alternative prey than in winter (Mowat et al., 1999). All evidence suggests that most reintroduced lynx are finding adequate food resources to survive. Mowat et al. (1999) suggest lynx and snowshoe hare select similar habitats except that hares select more dense stands than lynx. Very dense understory limits hunting success of the lynx and provides refugia for hares. Given the high proportion of snowshoe hare in the lynx diet in Colorado, we might then assume the habitats used by reintroduced lynx also depict areas where snowshoes hare are abundant and available for capture by lynx in Colorado. From both aerial locations taken throughout the year and from the site-scale habitat data collected in winter, the most common areas used by lynx are in stands of Engelmann spruce and subalpine fir. This is in contrast to adjacent areas of Ponderosa pine, pinyon juniper, aspen and oakbrush. The lack of lodgepole pine in the areas used by the lynx may be more reflective of the limited amount of lodgepole pine in southwestern Colorado, the Core Release Area, rather than avoidance of this tree species. Hodges (1999) summarized habitats used by snowshoe hare from 15 studies as areas of dense understory cover from shrubs, stands that are densely stocked, and stands at ages where branches have 16 �more lateral cover. Species composition and stand age appears to be less correlated with hare habitat use than is understory structure (Hodges 1999). The stands need to be old enough to provide dense cover and browse for the hares and cover for the lynx. In winter, the cover/browse needs to be tall enough to still provide browse and cover in average snow depths. Hares also use riparian areas and mature forests with understory. Site-scale habitat use documented for lynx in Colorado indicate lynx are most commonly using areas with Engelmann spruce understory present from the snow line to at least 1.5 m above the snow. The mean percent understory cover within the habitat plots is typically less than 15% regardless of understory species. However, if the understory species is willow, percent understory cover is typically double that, with mean number of shrubs per plot approximately 80, far greater than for any other understory species. In winter, hares browse on small diameter woody stems (<0.25"), bark and needles. In summer, hares shift their diet to include forbs, grasses, and other succulents as well as continuing to browse on woody stems. This shift in diet may express itself in seasonal shifts in habitat use, using more or denser coniferous cover in winter than in summer. The increased use of riparian areas by lynx in Colorado from July to November may reflect a seasonal shift in hare habitat use in Colorado. Major (1989) suggested lynx hunted the edge of dense riparian willow stands. The use of these edge habitats may allow lynx to hunt hares that live in habitats normally too dense to hunt effectively. The use of riparian areas and riparian-Engelmann spruce-subalpine fir and riparian-aspen mixes documented in Colorado may stem from a similar hunting strategy. However, too little is known about habitat use by hares in Colorado to test this hypothesis at this time. Lynx also require sufficient denning habitat. Denning habitat has been described by Koehler (1990) and Mowat et al. (1999) as areas having dense downed trees, roots, or dense live vegetation. We found this to be in true in Colorado as well (Shenk 2006). In addition, the dens used by reintroduced lynx were at high elevations and on steep north-facing slopes. All females that were documented with kittens denned in areas within their winter-use area. FUTURE STUDIES Monitoring of individuals through telemetry continues in an effort to document the viability of the reintroduced lynx population. However, as time since release increases, battery failure of telemetry collars also increases resulting in fewer released animals having working collars. In addition, few Colorado-born lynx have been captured and fitted with telemetry collars. Although trapping efforts have been conducted in earnest since 2003 to capture and fit animals with working telemetry collars, we have not been able to collar a sufficient number of animals throughout the state to document the status and trends of lynx distribution and demography throughout Colorado from these collared animals. The extent of lynx dispersal and current distribution beyond the Core Research Area and the difficulty of trapping lynx in all areas they inhabit, particularly large tracts of wilderness, requires redesigning our sampling and monitoring efforts to provide valid estimates of lynx distribution. Exploring occupancy modeling using non-invasive techniques may be a feasible alternative for ascertaining trends in population status and forming a basis for a large scale area monitoring program Therefore, we propose that monitoring lynx distribution would consist of 3 potential primary objectives to document the extent, stability and potential distribution of lynx (at the species and individual level) in Colorado. To estimate patterns in lynx distribution in Colorado a monitoring program could be developed that will: 1) annually estimate the spatial distribution of lynx in the core area and assess changes in lynx distribution over time; 2) detect colonization or expansion of lynx into other portions of the state, and 3) determine whether distribution or persistence are associated with habitat features, measured at the landscape-scale (stand age or composition). 17 �In order to design the most efficient statewide monitoring program, however, we will first evaluate the detection probabilities and efficacy of 3 methods of detection. These include snow-tracking, hair snares and camera surveillance. All of these methods can be conducted with minimal (camera surveillance or collection of hair) or non-invasive approaches (collection of scat samples) to individual animals. A pilot study will be conducted first to establish the most valid, efficient method to estimate the distribution and persistence of lynx. (see Appendix II for the detailed study plan). Information from the pilot study will then be used to design the most efficient strategy to meet the objectives of larger-scale monitoring programs to detect changes in lynx persistence and distribution as a foundation for assessing whether lynx have become established and will persist in Colorado. First, a minimally invasive monitoring program will be designed and implemented within the Core Research Area to describe lynx distribution and distribution trends in this area. A statewide plan could then be implemented to describe lynx distribution and distribution trends throughout Colorado. This monitoring protocol could result in the development of a standardized methodology that might be used by multiple entities to monitor the status of lynx throughout their range in North America. SUMMARY From results to date it can be concluded that CDOW developed release protocols that ensure high initial post-release survival of lynx, and on an individual level, lynx demonstrated they can survive longterm in areas of Colorado. We also documented that reintroduced lynx exhibited site fidelity, engaged in breeding behavior and produced kittens that were recruited into the Colorado breeding population. What is yet to be demonstrated is whether current conditions in Colorado can support the recruitment necessary to offset annual mortality in order to sustain the population. Monitoring of reintroduced lynx will continue in an effort to document such viability. ACKNOWLEDGMENTS The lynx reintroduction program involves the efforts of literally hundreds of people across North America, in Canada and USA. Any attempt to properly acknowledge all the people who played a role in this effort is at risk of missing many people. The following list should be considered to be incomplete. CDOW CLAWS Team (1998-2001): Bill Andree, Tom Beck, Gene Byrne, Bruce Gill, Mike Grode, Rick Kahn (Program Leader), Dave Kenvin, Todd Malmsbury, Jim Olterman, Dale Reed, John Seidel, Scott Wait, Margaret Wild. CDOW: John Mumma (Director 1996-2000), Russell George (Director 2001-2003), Bruce McCloskey (Director 2004-2007), Conrad Albert, Jerry Apker, Laurie Baeten, Cary Carron, Don Crane, Larry DeClaire, Phil Ehrlich, Lee Flores, Delana Friedrich, Dave Gallegos, Juanita Garcia, Drayton Harrison, Jon Kindler, Ann Mangusso, Jerrie McKee, Gary Miller, Melody Miller, Mike Miller, Kirk Navo, Robin Olterman, Jerry Pacheo, Mike Reid, Tom Remington, Ellen Salem, Eric Schaller, Mike Sherman, Jennie Slater, Steve Steinert, Kip Stransky, Suzanne Tracey, Anne Trainor, Scott Wait, Brad Weinmeister, Nancy Wild, Perry Will, Lisa Wolfe, Brent Woodward, Kelly Woods, Kevin Wright. Lynx Advisory Team (1998-2001): Steve Buskirk, Jeff Copeland, Dave Kenny, John Krebs, Brian Miller (Co-Leader), Mike Phillips, Kim Poole, Rich Reading (Co-Leader), Rob Ramey, John Weaver. U. S. Forest Service: Kit Buell, Joan Friedlander, Dale Gomez, Jerry Mastel, John Squires, Fred Wahl, Nancy Warren. U. S. Fish and Wildlife Service: Lee Carlson, Gary Patton (1998-2000), Kurt Broderdorp. State Agencies: Alaska: ADF&G: Cathie Harms, Mark Mcnay, Dan Reed (Regional Manager), Wayne Reglin (Director), Ken Taylor (Assist. Director), Ken Whitten, Randy Zarnke, Other:Ron Perkins (trapper), Dr. Cort Zachel (veterinarian). Washington: Gary Koehler. 18 �National Park Service: Steve King. Colorado State University: Alan Franklin, Gary White. Colorado Natural Heritage Program: Rob Schorr, Mike Wunder. Canada: British Columbia: Dr. Gary Armstrong (veterinarian), Mike Badry (government), Paul Blackwell (trapper coordinator), Trappers: Dennis Brown, Ken Graham, Tom Sbo, Terry Stocks, Ron Teppema, Matt Ounpuu. Yukon: Government: Arthur Hoole (Director), Harvey Jessup, Brian Pelchat, Helen Slama, Trappers: Roger Alfred, Ron Chamber, Raymond Craft, Lance Goodwin, Jerry Kruse, Elizabeth Hofer, Jurg Hofer, Guenther Mueller (YK Trapper’s Association), Ken Reeder, Rene Rivard (Trapper coordinator), Russ Rose, Gilbert Tulk, Dave Young. Alberta: Al Cook. Northwest Territories: Albert Bourque, Robert Mulders (Furbearer Biologist), Doug Steward (Director NWT Renewable Res.), Fort Providence Native People. Quebec: Luc Farrell, Pierre Fornier. Colorado Holding Facility: Herman and Susan Dieterich, Kate Goshorn, Loree Harvey, Rachel Riling. Pilots: Dell Dhabolt, Larry Gepfert, Al Keith, Jim Olterman, Matt Secor, Brian Smith, Whitey Wannamaker, Steve Waters, Dave Younkin. Field Crews (1999-2009): Steve Abele, Brandon Barr, Bryce Bateman, Todd Bayless, Nathan Berg, Ryan Besser, Jessica Bolis, Mandi Brandt, Keith Bruno, Brad Buckley. Patrick Burke, Braden Burkholder, Paula Capece, Matthew Chappell, Stacey Ciancone, Doug Clark, John DePue, Shana Dunkley, Brady Dunne, Tim Hanks, Carla Hanson, Dan Haskell, Nick Hatch, Matt Holmes, Allie Hunter, Andy Jennings, Susan Johnson, Paul Keenlance, Darrin Kite, Patrick Kolar, Tony Lavictoire, Jenny Lord, Clay Miller, Denny Morris, Kieran O’Donovan, Gene Orth, Chris Parmater, Bob Peterson, Jake Powell, Jeremy Rockweit, Britta Schielke, Jenny Shrum, Josh Smith, Heather Stricker, Adam Strong, Dave Unger, James Waddell, David Waltz, Andy Wastell, Mike Watrobka, Lyle Willmarth, Leslie Witter, Kei Yasuda, Jennifer Zahratka. Research Associates: Bob Dickman, Grant Merrill. Data Analysts: Karin Eichhoff, Joanne Stewart, Anne Trainor. Data Entry: Charlie Blackburn, Patrick Burke, Rebecca Grote, Angela Hill, Mindy Paulek. Mary Schuette , Dave Theobald and Chris Woodward provided assistance with the GIS analysis. . Funding: CDOW, Great Outdoors Colorado (GOCO), Turner Foundation, U.S.D.A. Forest Service, Vail Associates, Colorado Wildlife Heritage Foundation. LITERATURE CITED Aubry, K. B., G. M. Koehler, J. R. Squires. 1999. Ecology of Canada lynx in southern boreal forests. Pages 373-396 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. Bartmann, R. M., and Byrne, G. (2001) Analysis and critique of the 1998 snowshoe hare pellet survey. Colorado Division of Wildlife Report No. 20. Fort Collins, Colorado. Byrne, G. 1998. Core area release site selection and considerations for a Canada lynx reintroduction in Colorado. Report for the Colorado Division of Wildlife. Devineau, O., T. M. Shenk, P. F. Doherty Jr., G. C. White, and R. H. Kahn. 2009. Assessing release protocols for the Colorado Canada lynx (Lynx canadensis) reintroduction. Journal of Wildlife Management (in review). Devineau, O., T. M. Shenk, G. C. White, P. F. Doherty Jr., P. M. Lukacs, and R. H. Kahn. 2009. Estimating mortality for a widely dispersing carnivore, the Canada lynx (Lynx canadensis) reintroduced to Colorado. Journal of Applied Ecology (in review). Elton, C. and M. Nicholson 1942. The ten-year cycle in numbers of lynx in Canada. Journal of Animal Ecology 11: 215-244. Hodges, K. E. 1999. Ecology of snowshoe hares in southern boreal and montane forests. Pages 163206 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, 19 �and J. R. Squires editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. Koehler, G. M. 1990. Population and habitat characteristics of lynx and snowshoe hares in north central Washington. Canadian Journal of Zoology 68:845-851. Kolbe, J. A., J. R. Squires, T. W. Parker. 2003. An effective box trap for capturing lynx. Journal of Wildlife Management 31:980-985. Major, A. R. 1989. Lynx, Lynx canadensis canadensis (Kerr) predation patterns and habitat use in the Yukon Territory, Canada. M. S. Thesis, State University of New York, Syracuse. Mowat, G., K. G. Poole, and M. O’Donoghue. 1999. Ecology of lynx in northern Canada and Alaska. Pages 265-306 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. O’Donoghue, M, S. Boutin, D. L. Murray, C. J. Krebs, E. J. Hofer, U. Breitenmoser, C. BreitenmoserWuersten, G. Zuleta, C. , C. Doyle, and V. O. Nams. 2001. Mammalian predators: Coyotes and lynx. in Ecosystem Dynamics of the Boreal Forest: The Kluane Project. eds. C. J. Krebs, S. Boutin and R. Boonstra. Oxford University Press, Inc. New York, New York Poole, K. G., G. Mowat, and B. G. Slough. 1993. Chemical immobilization of lynx. Wildlife Society Bulletin 21:136-140. Shenk, T. M. 1999. Program Narrative Study Plan: Post-release monitoring of reintroduced lynx (Lynx canadensis) to Colorado. Colorado Division of Wildlife, Fort Collins, Colorado. __________. 2001. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report, July: 7- 34. Colorado Division of Wildlife, Fort Collins, Colorado. __________. 2006. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report, July: 1-45. Colorado Division of Wildlife, Fort Collins, Colorado. __________. 2007. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report, July: 1-57. Colorado Division of Wildlife, Fort Collins, Colorado. __________. 2008. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report, July: 1-57. Colorado Division of Wildlife, Fort Collins, Colorado Silverman, B.W. 1986. Density Estimation for Statistics and Data Analysis. Chapman and Hall. New York, New York, USA. Squires, J. R. and T. Laurion. 1999. Lynx home range and movements in Montana and Wyoming: preliminary results. Pages 337-349 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University Press of Colorado, Boulder, Colorado. U. S. Fish and Wildlife Service. 2000. Endangered and threatened wildlife and plants: final rule to list the contiguous United States distinct population segment of the Canada lynx as a threatened species. Federal Register 65, Number 58. Wild, M. A. 1999. Lynx veterinary services and diagnostics. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, Colorado. Zahratka, J. L. and T. M. Shenk. 2008. Population estimates of snowshoe hares in the southern Rocky Mountains. Journal of Wildlife Management 72: 906-912. Prepared by ___________________________________ Tanya M. Shenk, Wildlife Researcher 20 �Table 1. Number of wild-caught male (M) and female (F) Canada lynx (Lynx canadensis) from Alaska (AK) and Canada (BC = British Columbia, MB = Manitoba, QU = Quebec and YK = Yukon) released in southwestern Colorado per year from 1999–2006. State / Province of Origin Total Year %Released Sex AK BC MB QU YK 1999 19 2000 25 2003 15 2004 17 2005 17 2006 6 Total F 13 5 4 22 M 7 6 6 19 F 6 9 20 35 M 4 9 7 20 F 10 7 17 M 10 5 16 F 7 10 17 M 13 7 20 F 4 M 9 F M 30 1 3 8 3 18 8 3 20 4 3 7 5 2 7 48 218 91 4 45 Table 2. Status of adult Canada lynx (Lynx canadensis) reintroduced to Colorado as of August 31, 2009. Females Lynx Males Unknown TOTALS Released 115 103 218 Known Dead 65 52 1 118 Possible Alive 50 51 100 Missing 27 35 61a Monitoring/tracking 20 17 37 a 1 is unknown mortality Table 3. Causes of death for all Canada lynx (Lynx canadensis) released into southwestern Colorado 1999-2006 as of August 31, 2009. Mortalities Cause of Death Total (%) In Colorado (%) Outside Colorado (%) Unknown 44 (37.3) 29 (24.6) 15 (12.7) Gunshot 16 (13.6) 10 (8.5) 6 (5.1) Hit by Vehicle 14 (11.9) 9 (7.6) 5 (4.2) Starvation 12 (10.2) 11 (9.3) 1 (0.8) Other Trauma 8 (6.8) 7 (5.9) 1 (0.8) Plague 7 (5.9) 7 (5.9) 0 (0) Predation 6 (5.1) 6 (5.1) 0 (0) Probable Gunshot 5 (4.2) 4 (3.4) 1 (0.8) Probable Predation 3 (2.5) 2 (1.7) 1 (0.8) Illness 3 (2.5) 2 (1.7) 1 (0.8) Total Mortalities 118 87 (73.7) 31 (26.3) 21 �Table 4. Known lynx mortalities (n = 31) and causes of death documented by state outside of Colorado from February 1999 – August 31, 2009. Lynx ID AK99F8 Unknown AK99M11 YK99M06 AK99F13 YK00F04 BC99M04 QU05M01 QU04F05 QU03F07 BC00M04 YK06F01 BC03M08 BC06F07 AK99M06 AK99M01 QU05M08 MB05F02 BC00F14 QU04F07 BC06M10 QU04F02 AK00M03 QU05M03 YK06M01 YK00F07 BC06M13 YK99F01 YK00M03 YK05M03 YK05M02 State Date Mortality Recorded Cause of Death New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico New Mexico Nebraska Nebraska Nebraska Nebraska Wyoming Wyoming Wyoming Wyoming Utah Utah Utah Utah Utah Arizona Kansas Montana Iowa 7/30/1999 2000 1/27/2000 6/19/2000 6/22/2000 4/20/2001 6/7/2002 8/22/2005 8/26/2005 9/15/2005 7/19/2006 10/19/2006 10/19/2006 1/8/2007 11/16/1999 1/11/2005 10/1/2006 2/13/2007 7/28/2004 9/21/2004 8/15/2006 3/14/2007 7/2/2001 10/26/2005 12/4/2006 8/6/2007 12/11/08 9/15/2005 9/30/2005 11/8/2005 8/6/2007 Starvation Hit by Vehicle Unknown Probable Gunshot Unknown Gunshot Gunshot Unknown Hit by Vehicle Unknown Unknown Unknown Unknown Gunshot Gunshot Snared (Other Trauma) Unknown Gunshot Unknown Unknown Vehicle Collision Unknown Unknown Unknown Unknown Unknown Unknown Gunshot Vehicle Collision Unknown Vehicle Collision Table 5. Lynx reproduction summary statistics for 1999-2009. No reproduction was expected in 1999 because it was the first year of lynx releases and most animals were released after breeding season. Year Females Tracked 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 TOTAL /MEAN 9 25 21 17 26 40 42 34 28 22 Dens Found in May/June 0 0 0 6 11 17 4 0 0 5 Percent Tracked Females with Kittens 0.0 0.0 0.0 35.3 46.2 42.5 9.5 0.0 0.0 22.7 Additional Litters Found in Winter 0 0 0 0 2 1 0 0 0 - 22 Total Kittens Found Sex Ratio M/F (SE) 2.67 (0.33) 2.83 (0.24) 2.88 (0.18) 2.75 (0.47) 0 0 0 16 39 50 11 1.0 1.5 0.8 1.2 2.00 (0.00) 0 0 10 0.4 2.63(0.16) 126 0.98 (0.18) Mean Kittens/Litter (SE) �Table 6. Lynx captured because they were in poor body condition or were in atypical habitat and their fates 6 months post re-release as of August 31, 2009. Lynx ID Date of Capture State Where Captured Reason For Capture BC99F6 AK99M9 AK99F2 BC00F7 BC00M13 BC03M08 QU04M07 BC04M01 QU04F02 QU05M08 QU04M04 YK00F07 YK05M02 BC04M08 3/25/1999 3/24/2000 4/18/2000 2/11/2001 3/21/2001 9/5/2003 2/2/2006 11/5/2004 4/10/2005 11/25/2005 12/5/2006 12/12/2006 1/1/2007 1/22/2007 Colorado Colorado Colorado Colorado Wyoming Colorado Colorado Utah Nebraska Wyoming Utah Utah Kansas Wyoming Poor body condition Poor body condition Poor body condition Poor body condition Poor body condition Poor body condition Poor body condition Atypical habitat Atypical habitat Atypical habitat Atypical habitat Atypical habitat Atypical habitat Atypical habitat Date of Re-release 5/28/1999 5/3/2000 5/22/2000 N/A 4/24/2001 1/1/2004 N/A 12/5/2004 5/7/2005 4/18/2006 1/20/2007 1/20/2007 2/2/2007 2/15/2007 Status 6 Months Post Re-release Dead Missing Alive in Colorado Dead Alive in Colorado Alive in Colorado Dead Alive in Colorado Alive in Wyoming Dead Dead in Colorado Alive in Utah Alive in Iowa Alive in Colorado Table 7. Number of kills found each winter field season through snow-tracking of lynx and percent composition of kills of the three primary prey species. Field Season 1999 1999-2000 2000-2001 2001-2002 2002-2003 2003-2004 2004-2005 2005-2006 2006-2007 2007-2008 2008-2009 Total/Mean n 9 83 89 54 65 37 78 50 41 42 56 604 Snowshoe Hare 55.56 67.47 67.42 90.74 90.77 67.57 83.33 90.00 61.00 59.00 30.4 69.39 (SE=5.6) Prey (%) Red Squirrel Cottontail 22.22 0 19.28 1.20 19.10 8.99 5.56 0 6.15 0 27.03 2.70 10.26 0 0.08 0 39.0 0 33.3 0 66.1 0 22.55 (SE=5.7) 1.17 (SE=0.82) 23 Other 22.22 12.05 4.49 3.70 3.08 2.70 6.41 0.02 0 7.4 3.5 5.96 (SE=1.92) �Figure 1. Lynx are monitored throughout Colorado and by satellite throughout the western United States. The lynx core release area, where all lynx were released, is located in southwestern Colorado (outlines in white). A lynx-established core use area has developed in the Taylor Park and Collegiate Peak area in central Colorado. 24 �Figure 2. All documented lynx locations (non-truncated datasets) obtained from either aerial (red circles) or satellite (yellow circles) tracking from February 1999 through August 31, 2009 All known lynx mortality locations (n = 112) are displayed as black stars. 25 �Figure 3. Use-density surface for lynx satellite locations (non-truncated dataset) in Colorado from April 2000-April 2009. 26 �Figure 4. Use-density surface for lynx satellite locations (non- truncated dataset) in Colorado from April 2000-April 2009 27 �APPENDIX I Colorado Division of Wildlife August 2009 WILDLIFE RESEARCH REPORT State of Colorado Cost Center_______3430 Work Package_____0670 Task No.___________2 : : : : Federal Aid Project: N/A : Division of Wildlife Mammals Research Lynx Reintroduction Density, Demography, and Seasonal Movements of Snowshoe Hare in Colorado Period Covered: July 1, 2008- June 30, 2009 Author: J. S. Ivan, Ph.D. Candidate, Colorado State University Personnel: Dr. T. Shenk of CDOW and Dr. G. C. White of Colorado State University. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT A program to reintroduce the threatened Canada lynx (Lynx canadensis) into Colorado was initiated in 1997. Analysis of scat collected from winter snow tracking indicates that snowshoe hares (Lepus americanus) comprise 65–90% of the winter diet of reintroduced lynx in most winters. Thus, existence of lynx in Colorado and success of the reintroduction hinge at least partly on maintaining adequate and widespread hare populations. Beginning in July 2006, I initiated a study to assess the relative value of 3 stand types for providing hare habitat in Colorado. These types include mature, uneven-aged Engelmann spruce (Picea engelmannii)-subalpine fir (Abies lasiocarpa) forests, sapling lodgepole pine (Pinus contorta) forests (“small lodgepole”), and pole-sized lodgepole pine forests (“medium lodgepole”). Estimates and comparisons of survival, recruitment, finite population growth rate, and maximum (late summer) and minimum (late winter) snowshoe hare densities for each stand will provide the metrics for assessing these stands. Snowshoe hare densities on the study area are low compared to densities reported elsewhere. Within the study area, hare densities during summer were generally highest in small lodgepole stands, followed by mature spruce/fir and medium lodgepole, respectively. Absolute hare densities declined considerably in summer 2007 and rebounded only slightly during summer 2008. Hare density in small and medium lodgepole stands equalized during winters. However, as with summer, overall density was much lower during the second winter compared to the first and rebounded somewhat during the last winter. Hare survival from summer to winter was relatively high whereas winter to summer survival is quite low. Survival does not appear to differ between stand types or years, although a much more thorough analysis that will include known-fate telemetry data is forthcoming. This combined analysis will provide a final winter-summer estimate, will bring much more information to bear on the estimation process, and should increase precision of all estimates by a fair amount. 28 �WILDIFE RESEARCH REPORT DENSITY AND SURVIVAL OF SNOWSHOE HARES IN TAYLOR PARK AND PITKIN JACOB S. IVAN P. N. OBJECTIVE Assess the relative value of 3 stand types (mature spruce/fir, sapling lodgepole, pole-sized lodgepole) that purportedly provide high quality hare habitat by estimating survival, recruitment, finite population growth rate, and maximum (late summer) and minimum (late winter) snowshoe hare densities for each type. SEGMENT OBJECTIVES 1. Complete mark-recapture work across all replicate stands during late summer (mid-July through midSeptember) and winter (mid-January through March). 2. Obtain daily telemetry locations on radio-tagged hares for 10 days immediately after capture periods, as well as monthly between primary trapping sessions. 3. Locate, retrieve, and refurbish radio tags as mortalities occur. INTRODUCTION A program to reintroduce the threatened Canada lynx (Lynx canadensis) into Colorado was initiated in 1997. Since that time, 218 lynx have been released in the state, and an extensive effort to determine their movements, habitat use, reproductive success, and food habits has ensued (Shenk 2005). Analysis of scat collected from winter snow tracking indicates that snowshoe hares (Lepus americanus) comprise 65–90% of the winter diet of reintroduced lynx during most winters (T. Shenk, Colorado Division of Wildlife, unpublished data). Thus, as in the far north where the relationship between lynx and snowshoe hares has captured the attention of ecologists for decades, it appears that the existence of lynx in Colorado and success of the reintroduction effort may hinge on maintaining adequate and widespread populations of hares. Colorado represents the extreme southern range limit for both lynx and snowshoe hares (Hodges 2000). At this latitude, habitat for each species is less widespread and more fragmented compared to the continuous expanse of boreal forest at the heart of lynx and hare ranges. Neither exhibits dramatic cycles as occur farther north, and typical lynx (≤2−3 lynx/100km2; Aubry et al. 2000) and hare (≤1−2 hares/ha; Hodges 2000) densities in the southern part of their range correspond to cyclic lows form northern populations (2-30 lynx/100 km2, 1−16 hares/ha; Aubry et al. 2000, Hodges 2000, Hodges et al. 2001). Whereas extensive research on lynx-hare ecology has occurred in the boreal forests of Canada, literature regarding the ecology of these species in the southern portion of their range is relatively sparse. This scientific uncertainty is acknowledged in the “Canada Lynx Conservation Assessment and Strategy,” a formal agreement between federal agencies intended to provide a consistent approach to lynx conservation on public lands in the lower 48 states (Ruediger et al. 2000). In fact, one of the explicit guiding principles of this document is to “retain future options…until more conclusive information concerning lynx management is developed.” Thus, management recommendations in this agreement are decidedly conservative, especially with respect to timber management, and are applied broadly to cover all habitats thought to be of possible value to lynx and hare. Accurate identification and detailed 29 �description of lynx-hare habitat in the southern Rocky Mountains would permit more informed and refined management recommendations. A commonality throughout the snowshoe hare literature, regardless of geographic location, is that hares are associated with dense understory vegetation that provides both browse and cover (Wolfe et al. 1982, Litvaitis et al. 1985, Hodges 2000, Homyack et al. 2003, Miller 2005). In western mountains, this understory can be provided by relatively young conifer stands regenerating after stand-replacing fires or timber harvest (Sullivan and Sullivan 1988, Koehler 1990a, Koehler 1990b, Bull et al. 2005) as well as mature, uneven-aged stands (Beauvais 1997, Griffin 2004). Hares may also take advantage of seasonally abundant browse and cover provided by deciduous shrubs (e.g., riparian willow [Salix spp.], aspen [Populus tremuloides]; Wolff 1980, Miller 2005). In drier portions of hare range, such as Colorado, regenerating stands can be relatively sparse, and hares may be more associated with mesic, late-seral forest and/or riparian areas than with young stands (Ruggiero et al. 2000). Numerous investigators have sought to determine the relative importance of these distinctly different habitat types with regards to snowshoe hare ecology. Most previous evaluations were based on hare density or abundance (Bull et al. 2005), indices to hare density and abundance (Wolfe et al. 1982, Koehler 1990a, Beauvais 1997, Miller 2005), survival (Bull et al. 2005), and/or habitat use (Dolbeer and Clark 1975). Each of these approaches provides insight into hare ecology, but taken singly, none provide a complete picture and may even be misleading. For example, extensive use of a particular habitat type may not accurately reflect the fitness it imparts on individuals, and density can be high even in “sink” habitats (Van Horne 1983). A more informative approach would be to measure density, survival, and habitat use simultaneously in addition to recruitment and population growth rate through time. Griffin (2004) employed such an approach and found that summer hare densities were consistently highest in young, dense stands. However, he also noted that only dense mature stands held as many hares in winter as in summer. Furthermore hare survival seemed to be higher in dense mature stands, and only dense mature stands were predicted (by matrix projection) to impart a mean positive population growth rate on hares. Griffin’s (2004) study occurred in the relatively moist forests of Montana, which share many similarities but also many notable differences with Colorado forests including levels of fragmentation, species composition, elevation, and annual precipitation. The study outlined below is designed principally to evaluate the importance of young, regenerating lodgepole pine (Pinus contorta) and mature Engelmann spruce (Picea engelmannii)/ subalpine fir (Abies lasiocarpa) stands in Colorado by examining density and demography of snowshoe hares that reside in each. I determined that 2 classes of regenerating lodgepole could provide adequate hare habitat. Thus, I sampled both “small” (2.54-12.69 cm dbh) and “medium” (12.70-22.85 cm dbh) stands regenerating from clearcutting 20 and 40 years ago, respectively (Figure 1). Medium lodgepole stands were pre-commercially thinned 20 years ago; small lodgepole stands have not yet been thinned. Density and demography will be estimated primarily from mark-recapture techniques as data from such approaches can simultaneously provide information on both aspects of hare ecology. However, I will augment both density and demographic analyses with telemetry data to improve the accuracy and precision of estimates. The estimates reported here do not yet reflect addition of telemetry information. My hope is that information gathered from this research will be drawn upon as managers make routine decisions, leading to landscapes that include stands capable of supporting abundant populations of hares. I assume that if management agencies focus on providing habitat, hares will persist. 30 �Hypotheses 1) In general, snowshoe hare density in Colorado will be relatively low (≤0.5 hares/ha) compared to densities reported in northern boreal forests, even immediately post-breeding when an influx of juveniles will bolster hare numbers. 2) Snowshoe hare density will be consistently highest in small lodgepole pine stands, followed by large spruce/fir and medium lodgepole pine, respectively. 3) Survival will generally be highest in mature (large) spruce/fir stands followed by small and medium lodgepole pine, respectively. 4) Finite population growth rate will be consistently at or above 1.0 in mature spruce/fir stands with survival contributing most significantly to the growth rate. Finite growth rates for the lodgepole pine stands will be more variable. 5) Snowshoe hares will significantly shift their home ranges to make use of abundant food and cover provided by riparian willow (and/or aspen) habitats in summer. 6) Snowshoe hare density, survival, and recruitment will be highly correlated with understory cover and stem density. STUDY AREA The study area stretches from Taylor Park to Pitkin in central Colorado (Figure 2). Elevation ranges from 2700 m to 4000 m. Sagebrush (Artemisia spp.) dominates broad, low-lying valleys. Most montane areas are covered by even-aged, large-diameter lodgepole pine forests with sparse understory. Moist, north-facing slopes and areas near tree line are dominated by large-diameter Engelmann spruce/subalpine fir. Interspersed along streams and rivers are corridors of willow. Patches of aspen occur sporadically on southern exposures. This area was chosen over other potential study areas in the state because 1) it contained numerous examples of the 3 stand types of interest (more southern regions lack naturally occurring stands of lodgepole pine), 2) it was not subject to confounding effects of largescale mountain pine beetle outbreak as were more northern stands, and 3) an adequate number of radio frequencies were available to support a large study with hundreds of radio-tagged individuals. Within the study area I selected sample stands based on the following: Potential replicate stands were required to be 1) close enough geographically to minimize differences due to climate, weather, and topography, but are far enough apart to be considered independent, 2) adjacent to one or more riparian willow corridors, 3) within 1 km of an access road for logistical purposes, 4) of suitable size and shape to admit a 16.5-ha trapping grid, and 5) consistent in their management history (i.e., replicate lodgepole pine stands were clear-cut and/or thinned within 1-2 years of each other). I queried the U.S. Forest Service R2VEG GIS database using the criteria listed above to initially develop a suite of potential sample stands. I further narrowed this suite after obtaining updated standlevel information from local USFS personnel (Art Haines, Silviculturalist, USFS Gunnison Ranger District, personal communication). Finally, I ground-truthed potential stands and qualitatively assessed their representativeness and similarity to other potential replicates. Given the numerous constraints imposed, very few stands met all criteria. Thus, I was unable to randomly select sample stands from a population of suitable stands. Rather, I subjectively chose the “best” stands from among the handful that met my criteria. Small lodgepole stands rarely occur on the landscape in patches large enough to fit a full trapping grid. To accommodate this, I sampled 6 replicate small lodgepole stands (rather than 3) using half-sized trapping grids. 31 �METHODS Experimental Design/Procedures Variables.--The response variables of interest for this project include stand-specific snowshoe hare density (D), apparent survival (φ), recruitment (f), finite population growth rate (λ), and a metric of seasonal movement. Density is the number of hares per unit area and is estimated using conventional “boundary strip” techniques (Wilson and Anderson 1985) in this report. Stand-specific demographic parameters were estimated primarily from capture-mark-recapture methods. As such, apparent survival was defined as the probability that a marked animal alive and in the population at time i survived and was in the population at time i + 1. Apparent survival encompassed losses due to both death and emigration. Estimates of recruitment, population growth, and seasonal movement are forthcoming and not provided in this report. Potential explanatory variables for snowshoe hare density, demographics, and movement include general species composition and structural stage of each stand in which response variables are measured. Additionally, stem density, horizontal cover, and canopy cover (to a lesser extent) are highly correlated with snowshoe hare abundance and habitat use (Wolfe et al. 1982, Litvaitis et al. 1985, Hodges 2000, Zahratka 2004, Miller 2005). Thus, I further characterized vegetation in each stand by measuring stem density by size class (1-7 cm, 7.1-10 cm, and >10 cm), percent canopy cover, percent horizontal cover of understory and basal area. Basal area is an easily obtainable metric that may be correlated with the other variables and is recorded routinely during timber cruises, whereas the others are not. Thus, it might prove a useful link for biologists designing management strategies for snowshoe hare. Additionally, I recorded physical covariates such as ambient temperature, precipitation, and snow depth at each stand during sampling. These metrics were not included in the current preliminary analyses, but will be used as covariates in future models. Sampling.--All trapping and handling procedures have been approved by the Colorado State University Animal Care and Use Committee and filed with the Colorado Division of Wildlife. Snowshoe hares breed synchronously and generally exhibit 2 birth pulses in Colorado (although in some years, some individuals may have 3 litters), with the first pulse terminating approximately June 5−20 and the second approximately July 15–25 (Dolbeer 1972). To obtain a maximum density estimate, I began data collection on the first suite of sites immediately following the second birth pulse in late July. Along with a crew of 5 technicians, I deployed one 7 × 12 trapping grid (50-m spacing between traps; grid covers 16.5 ha) in the large spruce/fir and medium lodgepole stands within the first suite, along with 2 6 × 7 grids in 2 small lodgepole stands. Grid set up and trap deployment followed Griffin (2004) and Zahratka (2004). Grid locations and orientation within each stand were chosen subjectively to accommodate logistical constraints and to ensure that hares using the grid had ample opportunity to use adjacent riparian willow zones. As traps were deployed, they were locked open and “pre-baited” with apple slices, hay cubes, and commercial rabbit chow. Traps were pre-baited in this manner for a total of 3 nights to maximize capture rates when trapping began. This minimized the number of trap-nights needed to capture the desired number of animals which in turn minimized trap-related injuries and minimized problems with predators keying into trap lines. During pilot work in winter 2005, I observed low but increasing capture rates (<0.20) during the first 3 nights of trapping, with higher, more stable capture probabilities after 3 days (approximately 0.35–0.45). Thus 3 days of pre-baiting seemed reasonable. Traps were set on the afternoon of the 4th day and checked early each morning and re-set again in the evening on days 5–9. By checking traps in both morning and evening I prevented hares from being entrapped >13 hours, which minimized capture stress. A crew of 2 people worked together on each grid to check traps and process captures as quickly as possible. All captured hares were coaxed out of the trap and into a dark handling bag by blowing quick shots of air on them from behind. Hares remained in the 32 �handling bag, physically restrained with their eyes covered, for the entire handling process. Each individual was aged, sexed, marked with a passive integrated transponder (PIT) tag and temporary ear mark (to track PIT tag retention), then released. Aging consisted of assigning each individual as either juvenile (<1 year old, <1000 g) or adult (≥1 year old, ≥1000 g) based on weight and development of genitalia. This criterion is accurate through the end of September at which point juveniles are difficult to distinguish from adults (K. Hodges, University of British Columbia; P. Griffin, University of Montana, personal communication). After the first day of trapping, all captured hares were scanned for a PIT tag prior to any handling and those already marked were recorded and immediately released. Traps and bait were completely removed from the grid on day 10. In addition to PIT tags and ear marks, I radio collared up to 10 hares captured on each grid with a 28-g mortality-sensing transmitter (BioTrack, LTD) to facilitate unbiased density estimation as well as assessment of seasonal movements. I expected heterogeneity in snowshoe hare movements and use of the grid area, with potential bias surfacing due to location at which a hare is captured (e.g., hares captured on the edge of a grid may use the grid area differently than those captured at the center), and differential behavioral responses to trapping (e.g., young individuals may have lower capture probabilities and thus may be more likely to be captured on later occasions). To guard against the first potential bias, I randomly selected a starting trap location each morning and ran the grid systematically from that point. Thus, the first several hares encountered (and collared) were as likely to be from the inner part of the grid as from the edge. To protect against the second potential source of bias, I refrained from deploying the final 3 collars until days 4 and 5 of the trapping session. Immediately following the removal of traps, the field crew began work locating each radiocollared hare 1–2 times per day for 10 days. Most locations were obtained by triangulation from relatively close proximity, but some were obtained by “homing” on a signal (Samuel and Fuller 1996, Griffin 2004) taking care not to push hares while approaching them. Because hares are largely nocturnal (Keith 1964, Mech et al. 1966, Foresman and Pearson 1999), I made an effort to conduct telemetry work at various times of the night (safety and logistics permitting) and day to gather a representative sample of locations for each hare. Crews gathered telemetry locations for radio-collared hares on the initial suite of sites for 10 days. Then the 10−day trapping procedure and 8 to 10−day telemetry work were repeated on the grids comprising suites 2 and 3(Figure 3). The entire process was repeated during the winter when densities should have been at a minimum. Thus, during the period covered by this report, sampling occurred between July 16 – September 22 and between January 20−March 26. Telemetry work also occurred during “pre-baiting” days after the initial summer sampling session to determine which hares were still alive and immediately available to be sampled by the grid during the ensuing trapping period. Vegetation sampling was conducted in June and July 2008. I followed protocols established through previous snowshoe hare and lynx work in Colorado (Zahratka 2004, T. Shenk, Colorado Division of Wildlife, personal communication). Specifically, on each of the 12 live-trapping grids, I laid out 5 × 5 grids (3-m spacing) of vegetation sampling points centered on 15 of the 84 trap locations (Figure 4; 9 points were sampled on each of the ½-sized small lodgepole stands). At each of the 25 vegetation sampling points, I recorded canopy cover (present or absent) using a densitometer. I quantified downed coarse wood along the center transect of the 25-point grid following Brown (1974). From the center point (i.e., trap location) I measured 1) distance to the nearest woody stem 1.0−7.0 cm, 7.1−10.0 cm, and >10.0 cm in diameter at heights of 0.1 m and 1.0 m above the ground (to capture both summer and winter stem density; Barbour et al. 1999), 2) horizontal cover in 0.5-m increments above the ground up to 2 m (Nudds 1977), 3) basal area, and 4) slope. 33 �Data Analysis Density, Survival, and Population Growth.--I analyzed mark-recapture data in a robust design framework (Williams et al. 2002:523-554) treating summer and winter sampling occasions as primary periods, and the 5-day trapping sessions within each as secondary periods. As such, I assumed hare populations were demographically and geographically closed during the short 5-day mark-recapture sampling periods, but were open to immigration, emigration, births, and deaths between these occasions. I specified the Robust Design data type in Program MARK (White and Burnham 1999) and used the Huggins closed capture model (Huggins 1989, 1991) for secondary periods. I obtained estimates of apparent survival ( φˆ i )between each primary period. I followed Wilson and Anderson (1985) to calculate the effective area trapped and obtain a density estimate for each grid from each secondary period. Future density analyses will employ a new estimator that employs telemetry data to correct for bias (Ivan 2005). For this report, I used a relatively simple model where capture probability varied by stand type and season (i.e., winter and summer), while survival was allowed to vary by stand type, season, and time. RESULTS AND DISCUSSION During summer, density estimates followed hypotheses 1) and 2) above (Figure 5). Specifically, hare densities were clearly highest in small lodgepole stands and quite low in medium lodgepole stands. Spruce/fir was generally intermediate in density with the exception of the final summer. Telemetry data collected during this last sampling period suggests that many hares were present on spruce/fir sites, but were never caught. Therefore, I believe spruce/fir densities were much higher than actually measured during the final summer. While the relationship in density between stand types remained fairly constant throughout the study, the absolute density of hares dropped considerably from summer 2006 to summer 2007 and rebounded only slightly during summer 2008. It is unclear why this sharp decline occurred, although disease outbreak, natural population cycles, and response to increased predation due to lynx reintroduction are possibilities. Note that even the highest densities recorded here correspond to low estimates observed in other parts of hare range (Hodges 2000). Hare densities tend to equalize in lodgepole stands during winter (Figure 5). I submit that the interplay between food, cover, and snow depth provides a plausible explanation for this pattern. Medium lodgepole stands apparently provide very little forage/cover for hares during summer as the canopy in these stands is generally ≥1 meter off the ground. However, in winter, accumulated snow may make that canopy available again to hares. Conversely, small lodgepole stands provide abundant food and cover during summer, but accumulated snow during winter brings hares closer to the crowns of the young trees, which then provide less cover. Spruce/fir stands probably provide adequate access to both food and cover during both summer and winter due to their uneven-aged, multi-layered structure. Like the summer estimates, density during the second winter was much lower than during the first winter. Hare survival is quite high from summer to winter but very low from winter to summer (Figure 6). However, survival did not appear to differ between stand types or among years of this study. A deeper analysis of these data will occur over the next several months in which known-fate telemetry data will be combined with the current mark-recapture dataset. This combined analysis will bring significantly more information to bear on the process which should improve precision of estimates and may elucidate differences between stands or years that are not yet apparent. A much larger suite of models will be considered in that analysis. Model selection and model averaging (Burnham and Anderson 2002) will be used to more thoroughly assess survival of hares. Additionally, combining telemetry data with the current dataset will allow for another estimate of survival from winter 2009 to summer 2009. Hare recruitment and finite population growth rate will be estimated as derived parameters following the combined survival analysis. 34 �SUMMARY • • • • Snowshoe hare densities on my study sites appear to be relatively low compared to densities reported elsewhere. Densities during summer were highest in small lodgepole stands, followed by spruce/fir and medium lodgepole. During winter, densities equalize in lodgepole stands, possibly due to the interplay between snow depth and canopy height in small and medium lodgepole pine. Hare density declined considerably from winter to summer 2007 but has recovered somewhat since then. Summer to winter hare survival was consistently high but winter to summer survival is quite low. A more thorough analysis including known-fate survival data is forthcoming. This new analysis should improve precision of estimates and will add a sixth survival estimate to the current time series. ACKNOWLEDGMENTS Ken Wilson (CSU), Bill Romme (CSU), Paul Doherty (CSU), Dave Freddy (CDOW), Chad Bishop (CDOW), and Paul Lukacs (CDOW) provided helpful insight on the design of this study. We appreciate the invaluable logistical support provided by Mike Jackson (USFS), Art Haines (USFS), Jake Spritzer (USFS), Kerry Spetter (USFS), Margie Michaels (CDOW), Gabriele Engler (USGS), Dana Winkelman (USGS), Brandon Diamond (CDOW), Chris Parmeter (CDOW), Kathaleen Crane (CDOW), Lisa Wolfe (CDOW), and Laurie Baeten (CDOW). Jim Gammonley (CDOW), Dave Freddy (CDOW), Chad Bishop (CDOW), Jack Vayhinger (CDOW), Brandon Diamond (CDOW), and Brent Bibles (CDOW) assisted with trucks and equipment. The following hardy individuals collected the hard-won data presented in this report: Braden Burkholder, Matt Cuzzocreo, Brian Gerber, Belita Marine, Adam Behney, Pete Lundberg, Katie Yale, Britta Shielke, Cory VanStratt, Mike Watrobka, Meredith Goss, Sidra Blake, Keith Rutz, Rob Saltmarsh, Jennie Sinclair, Evan Wilson, Mat Levine, Matt Strauser, Greg Davidson, Leah Yandow, Renae Sattler, Caylen Cummins, DeVaughn Fraser, Mark Ratchford, Mike Petriello, Cynthia Soria, Roblyn Stitt, Sarah Ryan, Eric Newkirk, Kyle Heinrick, Matt Strauser, Doug Miles, and Cate Brown. Funding was provided by the Colorado Division of Wildlife. LITERATURE CITED Aubry, K. B., G. M. Koehler, and J. R. Squires. 2000. Ecology of Canada lynx in southern boreal forests. Pages 373-396 in Ruggiero, L. F., K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, and J. R. Squires, editors. Ecology and conservation of lynx in the United States. Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Collins, Colorado, USA. Barbour, M. G., J. H. Burk, W. E. Pitts, F. S. Gilliam, and M. W. Schwartz. 1999. Terrestrial plant ecology, Addison Wesley Longman, Inc., Menlo Park, California, USA. Beauvais, G. P. 1997. Mammals in fragmented forests in the Rocky Mountains: community structure, habitat selection, and individual fitness. Dissertation, University of Wyoming, Laramie, Wyoming, USA. BROWN, J. K. 1974. Handbook for inventorying downed woody material. U.S. Department of Agriculture, Forest Service, Intermountain Forest and Range Experiment Station, General Technical Report INT-16. Burnham, K. P., and D. R. Anderson. 1998. Model selection and inference: a practical informationtheoretic approach. Springer-Verlag, New York, New York, USA. 35 �Bull, E. L., T. W. Heater, A. A. Clark, J. F. Shepherd, and A. K. Blumton. 2005. Influence of precommercial thinning on snowshoe hares. U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station, General Technical Report PNW-RP-562. Dolbeer, R. A. and W. R. Clark. 1975. Population ecology of snowshoe hares in the central Rocky Mountains. Journal of Wildlife Management 39:535-549. Dolbeer, R. A. 1972. Population dynamics of snowshoe hare in Colorado. Dissertation, Colorado State University, Fort Collins, Colorado, USA. Foresman, K. R. and D. E. Pearson. 1999. Activity patterns of American martens, Martes americana, snowshoe hares, Lepus americanus, and red squirrels, Tamiasciurus hudsonicus, in westcentral Montana. The Canadian Field-Naturalist 113:386-389. Griffin, P. C. 2004. Landscape ecology of snowshoe hares in Montana. Dissertation, University of Montana, Missoula, Montana, USA. Hodges, K. E. 2000. Ecology of snowshoe hares in southern boreal and montane forests. Pages 163-206 in Ruggiero, L. F., K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, and J. R. Squires, editors. Ecology and conservation of lynx in the United States. University Press of Colorado, Boulder, Colorado, USA. Hodges, K. E., C. J. Krebs, D. S. Hik, C. I. Stefan, E. A. Gillis, and C. E. Doyle. 2001. Snowshoe hare demography. Pages 141-178 in Krebs, C. J., S. Boutin, and R. Boonstra, editors. Ecosystem dynamics of the boreal forest: the Kluane project. Oxford University Press, New York, New York, USA. Homyack, J. A., D. J. Harrison, and W. B. Krohn. 2003. Effects of precommercial thinning on select wildlife species in northern Maine, with special emphasis on snowshoe hare. Maine Cooperative Fish and Wildlife Research Unit, Orono, Maine, USA. Huggins, R. M. 1989. On the statistical analysis of capture-recapture experiments. Biometrika 76:133140. Huggins, R. M. 1991. Some practical aspects of a conditional likelihood approach to capture experiments. Biometrics 47:725-732. Ivan, J. S. 2005. Density, demography, and seasonal movements of snowshoe hares in Colorado. Program narrative study plan for mammals research, Colorado Division of Wildlife, Fort Collins, Colorado, USA. Keith, L. B. 1964. Daily activity pattern of snowshoe hares. Journal of Mammalogy 45:626-627. Koehler, G. M. 1990a. Snowshoe hare, Lepus americanus, use of forest successional stages and population changes during 1985-1989 in North-central Washington. Canadian Field-Naturalist 105:291-293. Koehler, G. M. 1990b. Population and habitat characteristics of lynx and snowshoe hares in north central Washington. Canadian Journal of Zoology 68:845-851. Litvaitis, J. A., J. A. Sherburne, and J. A. Bissonette. 1985. Influence of understory characteristics on snowshoe hare habitat use and density. Journal of Wildlife Management 49:866-873. Mech, L. D., K. L. Heezen, and D. B. Siniff. 1966. Onset and cessation of activity in cottontail rabbit and snowshoe hares in relation to sunset and sunrise. Animal Behaviour 14:410-413. Miller, M. A. 2005. Snowshoe hare habitat relationships in northwest Colorado. Thesis, Colorado State University, Fort Collins, Colorado, USA. Nudds, T. D. 1977. Quantifying the vegetative structure of wildlife cover. Wildlife Society Bulletin 5:113-117. Ruediger, B., J. Claar, S. Gniadek, B. Holt, Lewis Lyle, S. Mighton, B. Naney, G. Patton , T. Rinaldi, J. Trick, A. Vendehey, F. Wahl, N. Warren, D. Wenger, and A. Williamson. 2000. Canada lynx conservation assessment and strategy. U.S. Department of Agriculture, Forest Service, U.S. Department of Interior, Fish and Wildlife Service, Bureau of Land Management, National Park Service R1-00-53 U.S. Department of Agriculture, Forest Service, U.S. Department of Interior, Fish and Wildlife Service, Bureau of Land Management, National Park Service, Missoula, Montana, USA. 36 �Ruggiero, L. F., K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, and J. R. Squires. 2000. The scientific basis for lynx conservation: qualified insights. Pages 443-454 in Ruggiero, L. F., K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, and J. R. Squires, editors. Ecology and conservation of lynx in the United States. Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Collins, Colorado, USA. Samuel, M. D. and M. R. Fuller. 1996. Wildlife radiotelemetry. Pages 370-418 in Bookhout, T. A., editors. Research and Management Techniques for Wildlife and Habitats. Allen Press, Inc., Lawrence, Kansas, USA. Shenk, T. M. 2005. General locations of lynx (Lynx canadensis) reintroduced to southwestern Colorado from February 4, 1999 through February 1, 2005. Colorado Division of Wildlife Colorado Division of Wildlife, Fort Collins, Colorado, USA. Sullivan, T. P. and D. S. Sullivan. 1988. Influence of stand thinning on snowshoe hare population dynamics and feeding damage in lodgepole pine forest. Journal of Applied Ecology 25:791-805. Van Horne, B. 1983. Density as a misleading indicator of habitat quality. Journal of Wildlife Management 47:893-901. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 Supplement:120-138. Williams, B. K., J. D. Nichols, and M. J. Conroy. 2002. Analysis and management of animal populations, Academic Press, San Diego, California, USA. Wilson, K. R. and D. R. Anderson. 1985. Evaluation of two density estimators of small mammal population size. Journal of Mammalogy 66:13-21. Wolfe, M. L., N. V. Debyle, C. S. Winchell, and T. R. McCabe. 1982. Snowshoe hare cover relationships in northern Utah. Journal of Wildlife Management 46:662-670. Wolff, J. O. 1980. The role of habitat patchiness in the population dynamics of snowshoe hares. Ecological Monographs 50:111-130. Zahratka, J. L. 2004. The population and habitat ecology of snowshoe hares (Lepus americanus) in the southern Rocky Mountains. Thesis, The University of Wyoming, Laramie, Wyoming, USA. Prepared by _________________________________________________ Jacob S. Ivan, Graduate Student, Colorado State University 37 �Figure 1. Purported high quality snowshoe hare habitat in Colorado. From left to right: small lodgepole pine, medium lodgepole pine, and large Engelmann spruce/subalpine fir. Figure 2. Study area near Taylor Park and Pitkin, Colorado including medium lodgepole (squares), small lodgepole (circles), and spruce/fir (triangles) stands selected for mark-recapture sampling. 38 �Jul Aug Sep 1 2 3 4 2 3 4 5 6 7 8 1 Jan 5 6 7 8 9 10 11 9 10 11 12 13 14 15 12 13 14 15 16 17 18 16 17 18 19 20 21 22 19 20 21 22 23 24 25 23 24 25 26 27 28 29 26 27 28 29 30 31 1 30 31 1 2 3 4 5 2 3 4 5 6 7 8 6 7 8 9 10 11 12 9 10 11 12 13 14 15 13 14 15 16 17 18 19 16 17 18 19 20 21 22 20 21 22 23 24 25 26 23 24 25 26 27 28 1 27 28 29 30 31 1 2 2 3 4 5 6 7 8 3 4 5 6 7 8 9 9 10 11 12 13 14 15 10 11 12 13 14 15 16 16 17 18 19 20 21 22 17 18 19 20 21 22 23 23 24 25 26 27 28 29 24 25 26 27 28 29 30 30 31 Feb Mar Figure 3. Approximate annual data collection schedule for trapping () and telemetry (). Dates and weeks changed depending on calendar year and pay schedule. During telemetry work, the 6-person crew was divided into 2 teams, only one of which worked at any given time. Monthly locations on radio-collared hares were also collected in the interim between the intensive sampling periods indicated here. Figure 4. 15 trap locations (•) on 7 × 12 trapping grid where vegetation was sampled by measuring stem density, horizontal cover, downed woody material, and basal area. Additionally, the 25-point grid superimposed on each of the 15 trap locations (inset) was used to quantify canopy cover). 39 �Figure 5. Snowshoe hare density and 95% confidence intervals in 3 types of stands in central Colorado as determined by ½ mean maximum distance moved, summer 2006 through winter 2009. Figure 6. Snowshoe hare survival and 95% confidence intervals between summer and winter sampling seasons in 3 types of stands in central Colorado as determined by mark-recapture, 2006-2009. 40 �APPENDIX II PROGRAM NARRATIVE STUDY PLAN FOR MAMMALS RESEARCH FY 2009 – 10 State of: Cost Center: Work Package: Task No.: Colorado 3430 0670 3 Federal Aid Project No. N/A : : : : : Division of Wildlife Mammals Research Lynx Conservation Estimating Potential Changes in Distribution of Canada Lynx in Colorado: A Pilot Study Plan to Estimate Lynx Detection Probabilities ESTIMATING POTENTIAL CHANGES IN DISTRIBUTION OF CANADA LYNX IN COLORADO; A PILOT STUDY PLAN TO ESTIMATE LYNX DETECTION PROBABILITIES Principal Investigator Tanya M. Shenk, Wildlife Researcher, Mammals Research Cooperators Rick H. Kahn, Terrestrial Management Coordinator, CDOW Paul M. Lukacs, Biometrician, CDOW Grant J. Merrill, Research Associate, CSU Cooperative Research Unit Robert D. Dickman, CDOW Mike Miller, Acting Mammals Research Leader, CDOW STUDY PLAN APPROVAL Prepared by: Date: Submitted by; Date: Reviewed by: Date: Date: Date: Biometrician Review Date: Approved by: Date: Mammals Research Leader 41 �PROGRAM NARRATIVE STUDY PLAN FOR MAMMALS RESEARCH FY 2009-10 ESTIMATING THE EXTENT, STABILITY AND POTENTIAL DISTRIBUTION OF CANADA LYNX (LYNX CANADENSIS) IN COLORADO: A PILOT STUDY TO ESTIMATE LYNX DETECTION PROBABILITIES A Research Proposal Submitted By Tanya M. Shenk, Wildlife Researcher, Mammals Research A. Background: The Canada lynx (Lynx canadensis) occurs throughout the boreal forests of northern North America. While Canada and Alaska support healthy populations of the species, the lynx is currently listed as threatened under the Endangered Species Act (ESA) of 1973, as amended (16 U. S. C. 1531 et. seq.; U. S. Fish and Wildlife Service 2000) in the coterminous United States. Colorado represents the southern-most historical distribution of naturally occurring lynx, where the species occupied the higher elevation, montane forests in the state (U. S. Fish and Wildlife Service 2000). Thus, Colorado is included in the federal listing as lynx habitat. Lynx were extirpated or reduced to a few animals in Colorado, however, by the late 1970’s (U. S. Fish and Wildlife Service 2000), most likely due to multiple humanassociated factors, including predator control efforts such as poisoning and trapping (Meaney 2002). Given the isolation of and distance from Colorado to the nearest northern populations of lynx, the Colorado Division of Wildlife (CDOW) considered reintroduction as the only option to attempt to reestablish the species in the state. Therefore, a reintroduction effort was begun in 1997, with the first lynx released in Colorado in 1999. To date, 218 wild lynx were captured in Alaska or Canada and released in southwestern Colorado. The goal of the Colorado lynx reintroduction program is to establish a self-sustaining, viable population of lynx in this state. Evaluation of incremental achievements necessary for establishing viable populations is an interim method of assessing the success of the reintroduction effort. Seven critical criteria were identified that must be met before concluding a viable population had been established: 1) development of release protocols that lead to a high initial post-release survival of reintroduced animals, 2) long-term survival of lynx in Colorado, 3) site fidelity by lynx to areas supporting good habitat and in densities sufficient to breed, 4) reintroduced lynx must breed, 5) breeding must lead to production of surviving kittens, 6) lynx born in Colorado must reach breeding age and reproduce successfully, and 7) recruitment must equal to or be greater than mortality over an extended (~10 year) period of time (Shenk 2006). The fundamental approach taken to evaluate the status of each of these criteria was to PIT-tag and place telemetry collars on every lynx released and as many Colorado-born kittens surviving to adulthood as possible, followed by intensive monitoring of these animals through satellite, aerial and groundtracking. All establishment criteria, except (7) have been achieved. Lynx populations in Canada and Alaska have long been known to cycle in response to the 10-year snowshoe hare (Lepus americana) cycle (Elton and Nicholson 1942). Northern populations of lynx respond to snowshoe hare lows first through a decline in reproduction followed by an increase in adult mortality; when snowshoe hare populations increase, lynx respond with increased survival and reproduction (O’Donoghue et al. 2001). Therefore, annual survival and reproduction are highly variable but must be sufficient, overall, to result in long-term persistence of the population. It is not known if snowshoe hare populations in Colorado cycle and if so, where in the approximate 10-year cycle we are currently. Given this uncertainty, documenting persistence of lynx in Colorado for a period of at least 10- 42 �15 years would provide support that a viable population of lynx can be sustained in Colorado even in the event snowshoe hares do cycle in the state. Therefore, to document viability of the lynx population in Colorado, some form of long-term monitoring must be used to determine whether recruitment exceeds mortality for a period of time long enough to encompass a possible snowshoe hare cycle, and thus, determine the reintroduction a success. A challenge facing CDOW is how efforts should be allocated between focusing on monitoring the persistence of those lynx that have established within the core release area (Shenk 2007, Shenk 2008) and those lynx that may be pioneering and expanding into other portions of the state. Reproduction and known recruitment have been observed to be sporadic in the core area. To continue to document lynx reproduction through den site visits and to document survival of those kittens through tracking the adult females in winter looking for accompanying kittens requires a continued trapping effort to capture and radio-collar adult females. Lynx trapping is typically a time consuming and expensive operation as the lynx are territorial with large home ranges that may be entirely located within or largely comprised of inaccessible areas (e.g., wilderness areas). Alternatively, exploring occupancy modeling using noninvasive techniques may be a feasible alternative for ascertaining trends in population status and forming a basis for a large scale area monitoring program. Monitoring of individuals through telemetry continues in an effort to document the viability of the reintroduced lynx population. However, as time since release increases, battery failure of telemetry collars also increases resulting in fewer released animals having working collars. In addition, few Colorado-born lynx have been captured and fitted with telemetry collars. Although trapping efforts have been conducted in earnest since 2003 to capture and fit animals with working telemetry collars, we have not been able to collar a sufficient number of animals throughout the state to document the status and trends of lynx distribution and demography throughout Colorado from these collared animals. The extent of lynx dispersal and current distribution beyond the Core Research Area and the difficulty of trapping lynx in all areas they inhabit, particularly large tracts of wilderness, requires redesigning our sampling and monitoring efforts to provide valid estimates of lynx distribution. We propose that monitoring lynx distribution would consist of 3 potential primary objectives to document the extent, stability and potential distribution of lynx (at the species and individual level) in Colorado. To estimate patterns in lynx distribution in Colorado a monitoring program could be developed that will: 1) annually estimate the spatial distribution of lynx in the core area and assess changes in lynx distribution over time; 2) detect colonization or expansion of lynx into other portions of the state, and 3) determine whether distribution or persistence are associated with habitat features, measured at the landscape-scale (stand age or composition). A pilot study will be conducted first to establish the most valid, efficient method to estimate the distribution and persistence of lynx. B. Need The primary goal of the Colorado lynx reintroduction program is to establish a self-sustaining, viable population of Canada lynx in Colorado. The approach taken to reach this goal was to initially establish a lynx population within a core reintroduction area in southwestern Colorado. From this core reintroduction area, lynx could disperse on their own throughout the suitable habitat in the state, or additional reintroductions north of the core area could be conducted. The current lynx population in Colorado is comprised of surviving reintroduced adults, lynx born in Colorado from the reintroduced animals and possibly some naturally occurring lynx. Research and monitoring efforts over the last 9 years, since the first lynx were released, have focused primarily on monitoring reintroduced animals through VHF and satellite telemetry and estimating demographic parameters of these animals (e.g., Devineau et al. 2009). However, as more of these animals become unavailable for monitoring due to failed telemetry collars, death or movement out of the Core 43 �Research Area, it becomes more difficult to accurately evaluate the status of the entire lynx population in Colorado, including the Core Research Area. A dual monitoring approach will provide a comprehensive, feasible and valid estimation of the demography of the lynx population throughout the state. The first approach would continue to estimate reproduction within the Core Research Area through the use of telemetry. The second approach would obtain information on the status and trend of the distribution of lynx throughout the high elevation, montane areas of Colorado. Below we first outline the objectives and approach for the statewide distribution study and then propose a pilot study to establish the most valid, efficient methods to estimate the statewide distribution and persistence of lynx. A minimally-invasive monitoring program can be developed to estimate the extent, stability and potential distribution of lynx throughout Colorado. The primary objectives of the monitoring program will be to document the current distribution of lynx throughout Colorado, the stability, growth or shrinkage of this distribution over time, and to identify potential areas lynx may occupy in the future. The proposed goal would be to annually monitor lynx into the long-term future, with regular analyses of change (e.g., every 5 years). The fundamental structure of such a monitoring program will consist of: 1. 2. 3. 4. Creating a sampling frame of all potential lynx home range sized primary sampling units within Colorado. Annually estimating winter site occupancy and persistence within this sampling frame. Measuring key habitat features that have been documented to be important for both snowshoe hare and lynx at the landscape-scale within annually sampled sites. Predicting potential distribution of lynx throughout Colorado based on these habitat relationships. In the past, biologists referred to presence/absence as present/not detected, because absence cannot be absolutely determined. This term, however, confuses the status of being present or not present with the activity of either detecting or not detecting an animal. This monitoring program adopts the term presence/absence with the argument that although absence cannot be determined, it can be estimated statistically using a known or estimated detection probability. The indicator used to determine the distribution of occurrence of lynx is P, the proportion of primary sampling units (PSU’s) (Levy and Lemeshow 1999) with lynx presence. A PSU is a square sampling unit of 75km2, the approximate mean size of a lynx winter home range as estimated by a 90% kernel utilization distribution (Shenk 2007). For the statewide monitoring program, the sampling frame would consist of a grid of PSU’s laid over all areas of Colorado above 2591 meters (8500 feet). We would then estimate P from a random sample of PSU’s, using a sample size that is sufficient for attaining an estimate that is within 10% of the actual frequency 90% of the time (see Table 6.1, pg. 168 in MacKenzie et al. 2006). In order to design the most efficient statewide monitoring program, however, we will first evaluate the detection probabilities and efficacy of 3 methods of detection. These include snow-tracking, hair snares and camera surveillance. All of these methods can be conducted with minimal (camera surveillance or collection of hair) or non-invasive approaches (collection of scat samples) to individual animals. Identification of species will allow us to determine the presence of lynx in a PSU; identifying individual lynx within PSU’s will allow for monitoring individual movement patterns across PSU’s, reproduction, social structure and possibly apparent survival rates. Such non-invasive techniques are widely desirable because they are considered to have a minimal impact on animals and are inexpensive relative to other methods. Methodologies for identifying the species and individual lynx from blood and scat samples has been completed by the USFS Conservation Genetics Laboratory in Missoula, Montana. Thus, development costs have already been expended (by other agencies) and we need only cover the 44 �costs of genetic sample processing and interpretation of results. In order to begin genetic tracking of individual lynx a genetic library should be created from all lynx released in Colorado as part of the Colorado lynx reintroduction program, all documented kittens and lynx of unknown origin captured in Colorado. These samples have already been collected and are currently archived at the CDOW. This genetic library would be used to help determine paternity of Colorado-born kittens for future, detailed reproduction studies, document the dispersal of individuals throughout Colorado and also be available for research conducted on continent-wide studies of Canada lynx (e.g., Schwartz et al. 2002, Schwartz et al. 2003). Collecting scat samples during the pilot study will allow a test of these methodologies for the larger study as well as providing an opportunity to establish the protocols with the conservation genetics lab for collection, transport and analysis of the samples. This pilot study will provide necessary information to (1) identify the most efficient method of detecting lynx in a PSU and (2) provide an estimate of detection probability within a PSU. This detection probability will then be used to design the most efficient strategy to meet the objectives of larger-scale monitoring programs to detect changes in lynx persistence and distribution as a foundation for assessing whether lynx have become established and will persist in Colorado. First, a minimally invasive monitoring program will be designed and implemented within the Core Research Area to describe lynx distribution and distribution trends in this area. A statewide plan could then be implemented to describe lynx distribution and distribution trends throughout Colorado. This monitoring protocol could result in the development of a standardized methodology that might be used by multiple entities to monitor the status of lynx throughout their range in North America. This monitoring design will not provide a means of estimating total population size in the state because detection of a lynx may represent a single territorial animal, a breeding pair or a family unit. To obtain a statewide lynx abundance estimate, further efforts beyond this sampling design would be needed to establish the actual or estimated number of lynx in a PSU. Furthermore, this monitoring program is not designed to provide information on reproductive success or estimate survival. C. Objectives: The primary objectives of this pilot study are to: 1. Provide information needed to estimate the detection probability (p) of 3 different, minimally-invasive methods to detect lynx in a PSU in winter, where lynx are known to occur but in extremely low densities (approximately 1 per 75 km2). 2. Evaluate and compare the efficacy of the 3 methods of lynx detection in winter within a PSU. 3. Develop a standardized, valid methodology for describing various landscape-scale habitat features, including those important to snowshoe hare, within a PSU. D. Expected Results or Benefits: The methodologies developed during this pilot study will be used to develop a valid, non-invasive or minimally invasive inventory and monitoring program to estimate the distribution of Canada lynx in Colorado. The monitoring program will provide information on the annual winter distribution, extent and habitat relationships of these parameters as well as their long-term trend which will be evaluated every 5 years. The protocols developed will be made available to any other agencies or entities that want to monitor lynx. The proposed methodology to estimate and monitor trends in lynx distribution throughout Colorado is designed to make use of technologies (e.g., genetic identification) reliant only on noninvasive or minimally invasive techniques. Such non-invasive techniques are widely desirable because they require minimal impact to the animals and because of their cost efficiencies. 45 �E. Approach The primary objective of the pilot study is to evaluate the efficacy of the proposed sampling techniques for detecting lynx presence. However, the pilot study will also include qualitative evaluation of all design methods that will be employed in a future, larger research area and statewide monitoring efforts, (i.e., the complete sampling frame). Sampling Frame and Primary Sampling Unit Selection The sampling frame will consist of all forested areas in Colorado >2591 m (8500 ft) in elevation. The sampling frame will be randomly overlayed with a contiguous grid of 75 km2 squares. The size of the square reflects a mean annual home range size of a reproducing lynx in Colorado (Shenk 2007) and similar to home range estimates obtained for lynx in Montana (Squires and Laurion 1999). If a grid square is >50% forested it will be identified as a PSU. We will assume the lowest detection probabilities for lynx would occur in a PSU occupied by only 1 lynx. Given that we want to estimate lynx detection probabilities under the worst case scenario, we will eliminate all PSU’s where we know, through VHF or satellite-tracking, there is more than one lynx occupying the area. We will then select 6 PSU’s where we know at least 1 but not likely more than 1 lynx occupies the area. The assumptions that must be met in estimating occupancy are 1) surveyed sites can be occupied by the species of interest throughout the duration of the study, with no sites becoming occupied or unoccupied during the survey period (i.e., the system is closed), 2) species are not falsely detected, but can remain undetected if present, and 3) species detection at a site is assumed to be independent of species detection at other sites (MacKenzie et al. 2006). For this pilot study, there will be 3 different methods of detection (snow-tracking, hair snares and camera surveillance). Snow-tracking and camera surveillance will be evaluated at 2 different levels of effort; hair snares will be evaluated at 3 levels of effort resulting in 7 total detection approaches. In order to meet the assumptions for estimating occupancy and assuming the different detection approaches don’t influence each other, each of the 6 PSU’s will be assigned all detection approaches (except for the higher level of hair-snaring) for 3 weeks, allowing for completing surveys of 2 PSU’s per month. The increased hair snare effort will be conducted on a PSU the month following the initial survey effort (see below). Thus, by the end of four months each PSU will have had each detection approach applied to it. This will result in 6 spatial replications of each of 3 detection approaches applied to a PSU for 3 weeks. Maximum levels of effort will be applied to each PSU and then the data sub-sampled to evaluate lower levels of effort. Field Methods Temporal aspects of the sampling design In order to verify the detection methods being evaluated in this pilot study are effective at detecting lynx when they are present, we need to conduct the study while we have active radio collars on lynx. Currently, we are continuing to monitor and re-collar lynx within the Core Research Area for data on the demography and movement patterns of the reintroduced lynx. Thus, completing this pilot study at the same time that active monitoring is being conducted in the research area eliminates the need for future radio-collaring efforts to conduct this pilot study. All data collection will be conducted from January 1-March 31 (Table 1). This is within the time period (October–April) when lynx typically maintain fidelity to a winter home range and when breeding occurs, the period of interest for document long-term persistence of lynx. 46 �Table 1. Data collection and crew work schedule for the six PSU’s to be sampled. PSU Month Week Crew Activity 1 January 1 I Set-up detection routes and 5 detection stations with hair snares and cameras; Snow-track (2 10-hour days) 2 I Snow-track (4 10-hour days) 3 I Snow-track (4 10-hour days)4 I Snow-track (2 10-hour days); Retrieve cameras and hair snares at the 5 detection stations, place 20 hair snares along the detection route; Travel to next PSU 2 January 1 II Set-up detection routes and stations with hair snares and cameras; Snow-track (2 10-hour days) 2 II Snow-track (4 10-hour days) 3 II Snow-track (4 10-hour days)4 II Snow-track (2 10-hour days); Retrieve cameras and hair snares at the 5 detection stations, place 20 hair snares along the detection route; Travel to next PSU 3 February 1 I Set-up detection routes and stations with hair snares and cameras; Snow-track (2 10-hour days) 2 I Snow-track (4 10-hour days) 3 I Snow-track (4 10-hour days)4 I Snow-track (2 10-hour days); Retrieve cameras and hair snares at the 5 detection stations, place 20 hair snares along the detection route; Travel to next PSU 4 February 1 II Set-up detection routes and stations with hair snares and cameras; Snow-track (2 10-hour days) 2 II Snow-track (4 10-hour days) 3 II Snow-track (4 10-hour days)4 II Snow-track (2 10-hour days); Retrieve cameras and hair snares at the 5 detection stations, place 20 hair snares along the detection route; Travel to next PSU 5 March 1 I Set-up detection routes and stations with hair snares and cameras; Snow-track (2 10-hour days) 2 I Snow-track (4 10-hour days) 3 I Snow-track (4 10-hour days)4 I Snow-track (2 10-hour days); Retrieve cameras and hair snares at the 5 detection stations, place 20 hair snares along the detection route; Travel to next PSU 6 March 1 II Set-up detection routes and stations with hair snares and cameras; Snow-track (2 10-hour days) 2 II Snow-track (4 10-hour days) 3 II Snow-track (4 10-hour days)4 II Snow-track (2 10-hour days); Retrieve cameras and hair snares at the 5 detection stations, place 20 hair snares along the detection route; Travel to next PSU Lynx Detection Data Collection Three methods will be evaluated to determine which is most efficient in detecting the presence of lynx. These methods include 1) documenting the presence of lynx tracks in the snow coupled with a DNA sample collection (hair or scat found through snow-tracking), 2) a photograph of a lynx captured by 47 �a surveillance camera, or 3) documenting the presence of lynx from a hair DNA sample collected on a hair snag at a scent and visual lure station. All methods will be applied to the same stations within a PSU at the same time. Each method will be implemented in the areas within the selected PSU that a lynx would most likely use. Based on lynx habitat use in Colorado (Shenk 2005), this will include areas of mature Engelmann spruce-subalpine fir forest stands with 42-65% canopy cover and 15-20% conifer understory cover, mean slopes of 16° and elevations above 2591 m. In addition, selection of specific detection stations will be based on natural travel routes or the presence of lynx sign (i.e., tracks or scat). Chances of detecting lynx at these locations will be further enhanced by placing scent and visual lures at these sites. Other feline species may be attracted to these same lures, however, the probability will be low as the study will be conducted in winter and the deep snows at these elevations should preclude species such as mountain lion (Puma concolor) and bobcat (Lynx rufus) from using these areas. Different levels of sampling intensity will be evaluated for each method to determine the most efficient sampling design. Establishing Detection Stations & Routes. – To eliminate bias in site selection of detection stations and routes, any known lynx locations in the selected PSU’s will not be made available to the field technicians who will be establishing the detection routes, detection stations and collecting the detection data. Field personnel will be provided information to select routes that are both the most feasible and likely areas to detect lynx within a PSU (see above). Detection stations will be set up in areas along those selected routes in areas of good lynx habitat. Commercial scent lures and visual lures (e.g., CD’s, waterfowl wings) will be used at each detection station to enhance the probability of drawing a lynx into the station. To increase the probability of lynx using the hair snares, the hair snares will be placed on landscape features at the detection station known to be used as scent posts by lynx such as tree stumps, small trees and broken logs protruding from the snow at approximate head height of a lynx (Schmidt and Kowalczyk 2006). Snow-Tracking. –Searches for tracks will be attempted by hiking, driving or snowmobiling detection station routes in the PSU once enough snow has accumulated. Due to the inaccessibility of wilderness and roadless areas after significant snowfall, surveys will be conducted in these areas first, while snow accumulations are great enough to detect tracks but not so great as to preclude human access to the area. Once tracks are observed, personnel will follow the tracks until either lynx hair or scat are found and collected or the distance tracks are followed exceeds 1 km. All hair found in day beds or a single scat will constitute a sample. Because lynx are a federally listed species, which can result in regulatory protection, we will eliminate doubt about the presence of lynx by submitting hair or scat sampled to a conservation genetics lab to confirm species identification (see McKelvey et al. 2006). All hair and fecal samples will be submitted to a conservation genetics lab for identification to species and individual, if possible. The distance a track is followed will be limited to 1 km to increase efficiency in lynx detection within the PSU (i.e., it will be assumed it is quicker to find a new lynx track to follow to locate hair or scat than to pursue a single track for more than 1 km; see McKelvey et al. 2006). Two levels of search effort for lynx tracks will be implemented within a PSU. The first tracking intensity will be 4 consecutive tracking days (although there may be days of no tracking within this period – e.g., days off, cancellation of tracking effort due to weather etc.), the second will be 8 consecutive days of tracking. All PSU’s will be snow-tracked for 12 days (3 week field effort, see Table 1). This will provide 3 replicates of a 4-day tracking session and 2 replicates of an 8-day tracking session (replicating one of the 4-day tracking sessions). Camera Traps. – Digital infrared surveillance cameras (RECONYX RapidFireTM Professional PC85) will be placed at 5 randomly selected detection stations among those that appear the most likely places where lynx would encounter them within the PSU, as defined above. Cameras will be encased in heavy duty 16 gauge steel security enclosure, attached to a tree with a Master Lock TM PythonTM cable lock and powered by 3-volt C-cell lithium batteries. 48 �We will evaluate detection probabilities for 2 levels of camera surveillance, placing either 2 cameras within the grid or 5 cameras. Five cameras will be placed in all PSU’s, a random subset of 2 cameras from these 5 will be selected to evaluate the efficacy of the lesser effort. Cameras will run continuously for the 3.5 week period. We can evaluate the most efficient number of days required to detect a lynx and the interaction between number of cameras and length of time cameras are active. Hair-Snares. - Barbed wire and carpet hair traps, scented with commercial lynx lures as described by McDaniel et al. (2000) will be placed at each of the detection stations within the PSU in areas where lynx would most likely encounter them (see above). A sample will be defined as all hairs from a single hair snare. Each hair sample will be placed in a uniquely numbered paper envelop, and a flame passed under the barbs to remove any genetic material so that the hair snare can be used again without contaminating future samples. All hair samples will be submitted to a conservation genetics lab for identification to species. Hair snares have been shown to be highly reliable for lynx identification to species (Schwartz et al. 2002) but not for individual lynx identification (Lukacs 2005). We will evaluate detection probabilities of lynx for 3 sample intensity levels of hair snares. First, hair snares will be set up within the PSU at each of the 5 detection stations. A the end of the 3.5 week monitoring session of a PSU, 20 hair snares, at least 100 meters apart (McDaniel et al. 2000) will be placed along the detection route (assuming detection routes will be approximately 25 km long) and collected approximately 1 month later (by the crew leader). Both the detection probability for the 20 hair snares and a random subset of 10 hair snares from these 20 will be selected to evaluate the efficacy of the lesser effort. This larger effort of 20 hair snares will be completed in a PSU after the monitoring conducted by snow-tracking and camera traps as the presence of additional scent stations may affect the use of the 5 camera detection stations. Data Analysis We will estimate the probability of detecting a lynx (p) on each of the PSU’s for each of the detection methods and level of effort for each of those methods. Aerial or satellite telemetry will be used to confirm the presence of at least one lynx in each of the six sampled PSU’s. An evaluation of each of the detection methods will be completed to determine the most reliable, efficient (e.g., cost of equipment, labor) and feasible method of detecting a lynx on a PSU when at least one lynx is present. Project Schedule Completed by Dec. 2009 1. Complete sampling frame and selection of primary sampling units. 2. Purchase and test equipment. Jan.–Mar. 2010 1. Set up detection stations. 2. Conduct lynx snow-tracking surveys. 3. Conduct lynx hair snare sampling. 4. Conduct camera surveillance surveys. 5. Process and submit all genetic samples collected during surveys to a genetic conservation lab (e.g., USDAFS Conservation Genetics Lab in Missoula, Montana, USGS Conservation Genetics Lab in Denver, Colorado). Apr.–May 2010 1. Data entry, analyses and complete report. 49 �Personnel: Project Leader: Tanya Shenk, Wildlife Researcher, CDOW Responsibilities: Design study, work with research associate to implement and complete field work and data entry, complete analysis, write report. Crew Leader: Responsibilities: Assist is study design and selection of PSU’s, supervise field technician, complete all data entry, and perform other duties as needed associated with the post-release monitoring program and the reproduction study. Field Technicians Responsibilities. To establish detection routes, detection stations, place hair snags, cameras and conduct all snow-tracking. Data Analysis: Tanya Shenk, Wildlife Researcher, CDOW Paul Lukacs, Biometrician CDOW Gary White, Professor Emeritus, CSU Paul Doherty, Associate Professor, CSU Estimated Annual Budget: January 2009 – April 2010 Salary (Tech III, Jan 2009 –Apr 2010) Salary (4 Field Technicians, Tech II Jan 2010 – Mar 2010) Travel, housing Misc. Supplies/Operating Equipment Repair, maintenance (snowmobiles) Detection cameras (11 @$1,000 each) Processing of genetic samples collected during monitoring Vehicles (3) $ 15,000 $ 36,100 $ 5,000 $ 4,000 $ 5000 $ 11,000 $ 4,000 $ 6,000 TOTAL $86,100 G. Location: Southwestern and central Colorado is characterized by wide plateaus, river valleys, and rugged mountains that reach elevations over 4200 m. Engelmann spruce-subalpine fir is the most widely distributed coniferous forest type at elevations most typically used by lynx (2591-3353 m). The Core Reintroduction Research Area is defined as areas >2591 m in elevation within the area bounded by the New Mexico state line to the south, Taylor Mesa to the west and Monarch Pass on the north and east (Figure 1). Project headquarters will at the Fort Collins CDOW Research Center. H. Literature Cited: Aubry, K. B., G. M. Koehler, J. R. Squires. 1999. Ecology of Canada lynx in southern boreal forests. Pages 373-396 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. 50 �Byrne, G. 1998. Core area release site selection and considerations for a Canada lynx reintroduction in Colorado. Report for the Colorado Division of Wildlife. Curtis, J. T. 1959. The vegetation of Wisconsin. University of Wisconsin Pres, Madison. Devineau, O., T. M. Shenk, G. C. White, P. F. Doherty Jr., P. M. Lukacs, and R. H. Kahn. 2008. Estimating mortality for a widely dispersing reintroduced carnivore, the Canada lynx (Lynx canadensis). Ecology (in review). Elton, C. and M. Nicholson 1942. The ten-year cycle in numbers of lynx in Canada. Journal of Animal Ecology 11: 215-244. Hodges, K. E. 1999. Ecology of snowshoe hares in southern boreal and montane forests. Pages 163206 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. Koehler, G. M. 1990. Population and habitat characteristics of lynx and snowshoe hares in north central Washington. Canadian Journal of Zoology 68:845-851. Kolbe, J. A., J. R. Squires, T. W. Parker. 2003. An effective box trap for capturing lynx. Journal of Wildlife Management 31:980-985. Laymon, S. A. 1988. The ecology of the spotted owl in the central Sierra Nevada, California. PhD Dissertation, University of California, Berkeley, California. Lukacs, P. M. 2005. Statistical aspects of using genetic markers for individual identification in capturerecapture studies. PhD Dissertation, Colorado State University, Fort Collins, Colorado. MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy Estimation and Modeling: Inferring Patterns and Dynamics of Species Occurrence. Elsevier Academic Press. Oxford, UK. Major, A. R. 1989. Lynx, Lynx canadensis canadensis (Kerr) predation patterns and habitat use in the Yukon Territory, Canada. M. S. Thesis, State University of New York, Syracuse. McDaniel, G. W., K. S. McKelvey, J. R. Squires. and L. F. Ruggiero. 2000. Efficacy of lures and hair snares to detect lynx. Wildlife Society Bulletin 28:119-123. McKelvey, K. S., J. von Kienast; K.B. Aubry; G. M. Koehler; B. T. Maletzke; J. R. Squires; E. L. Lindquist; S. Loch; M. K. Schwartz. 2006. DNA analysis of hair and scat collected along snow tracks to document the presence of Canada lynx. Wildlife Society Bulletin 34: 451-455. Meaney C. 2002. A review of Canada lynx (Lynx canadensis) abundance records from Colorado in the first quarter of the 20Th century. Colorado Department of Transportation Report. Mowat, G., K. G. Poole, and M. O’Donoghue. 1999. Ecology of lynx in northern Canada and Alaska. Pages 265-306 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University of Colorado Press, Boulder, Colorado. O’Donoghue, M, S. Boutin, D. L. Murray, C. J. Krebs, E. J. Hofer, U. Breitenmoser, C. BreitenmoserWuersten, G. Zuleta, C. , C. Doyle, and V. O. Nams. 2001. Mammalian predators: Coyotes and lynx. in Ecosystem Dynamics of the Boreal Forest: The Kluane Project. eds. C. J. Krebs, S. Boutin and R. Boonstra. Oxford University Press, Inc. New York, New York. Poole, K. G., G. Mowat, and B. G. Slough. 1993. Chemical immobilization of lynx. Wildlife Society Bulletin 21:136-140. Schmidt, K. and R. Kowalcyzk. 2006. Using scent-marking stations to collect hair samples to monitor Eurasian lynx populations. Wildlife Society Bulletin 34: 462-466. Schwartz, M. K. , L. S. Mills, K. S. McKelvey, L. F. Ruggiero, AND F. W. Allendorf. 2002. DNA reveals high dispersal synchronizing the population dynamics of Canada lynx. Nature 415:520522. Schwartz, M. K., L. S. Mills, Y. Ortega, L. F. Ruggiero, and F. W. Allendorf. 2003. Landscape location affects genetic variation of Canada lynx (Lynx canadensis). Molecular Ecology 12:1807-1816. 51 �Shenk, T. M. 1999. Program narrative Study Plan: Post-release monitoring of reintroduced lynx (Lynx canadensis) to Colorado. Colorado Division of Wildlife, Fort Collins, Colorado. __________. 2001. Post-release monitoring of lynx reintroduced to Colorado. Job Progress Report, Colorado Division of Wildlife, Fort Collins, Colorado. __________. 2006. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, Colorado __________. 2007. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, Colorado Silverman, B.W. 1986. Density Estimation for Statistics and Data Analysis. Chapman and Hall. New York, New York, USA. Squires, J. R. and T. Laurion. 1999. Lynx home range and movements in Montana and Wyoming: preliminary results. Pages 337-349 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University Press of Colorado, Boulder, Colorado. U. S. Fish and Wildlife Service. 2000. Endangered and threatened wildlife and plants: final rule to list the contiguous United States distinct population segment of the Canada lynx as a threatened species. Federal Register 65, Number 58. White, G.C. and K. P. Burnham. 1999. Program MARK: Survival estimation from populations of marked animals. Bird Study 46 Supplement, 120-138. Wild, M. A. 1999. Lynx veterinary services and diagnostics. Job Progress Report for the Colorado Division of Wildlife. Fort Collins, Colorado. 52 �Figure 3. Study area depicting the Core Research Area, Lynx-established Core Area and relative lynx use (red is high intensity use, yellow is low intensity use). 53 �Colorado Division of Wildlife July 2009–June 2010 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 0670 1 Federal Aid Project No. N/A : : : : : Division of Wildlife Mammals Research Lynx Conservation Post-Release Monitoring of Lynx Reintroduced to Colorado Period Covered: July 1, 2009 – June 30, 2010 Author: T. M. Shenk Personnel: O. Devineau, R. Dickman, P. Doherty, L. Gepfert, J. Ivan, R. Kahn, A. Keith, P. Lukacs, G. Merrill, B. Smith, T. Spraker, S. Waters, G. White, L. Wolfe All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT In an effort to establish a viable population of Canada lynx (Lynx canadensis) in Colorado, the Colorado Division of Wildlife (CDOW) initiated a reintroduction effort in 1997 with the first lynx released in February 1999. From 1999-2006, 218 wild-caught lynx from Canada and Alaska were released in Colorado. Post-release monitoring was critical to assess and modify the release protocols as they were implemented to improve the survival of released individuals. Average monthly mortality rate in the reintroduction area during the first year post-release decreased with time in captivity from 0.205 [95% CI 0.069, 0.475] for lynx spending up to 7 days in captivity to 0.028 [95% CI 0.012, 0.064] for lynx spending > 45 days in captivity before release. Under the final release protocol, lynx were held in captivity and fed a high quality diet for a minimum of three weeks before release. Results suggested that keeping lynx in captivity beyond 5 or 6 weeks accrued little benefit in terms of monthly survival. We documented survival, movement patterns, reproduction, and landscape habitat-use through aerial (n = 11,580) and satellite (n = 29,258) tracking. Monthly mortality rate was estimated as lower inside the reintroduction area than outside the reintroduction area, and slightly higher for male than for female lynx, although 95% confidence intervals for sexes overlapped. Mortality was higher immediately after release (first month = 0.0368 [SE = 0.0140] inside the study area, and 0.1012 [SE = 0.0359] outside the study area), and then decreased according to a quadratic trend over time. Given the importance of adult survival in the dynamics of long-lived species, the long-term, high survival rates estimated for the reintroduced lynx both inside (0.9315, SE = 0.0325) and outside (0.8219, SE = 0.0744) the reintroduction area are promising for the establishment of a viable population of lynx in Colorado. From 1999-June 2010, there were 122 known mortalities of released adult lynx. Human-caused mortality factors were the highest causes of death with approximately 29.7% attributed to collisions with vehicles or gunshot. Starvation and disease/illness accounted for 18.6% of the deaths while 37.3% of the deaths were from unknown causes. Reproduction was first documented in 2003 with subsequent successful reproduction 1 �in 2004, 2005, 2006, 2009, and 2010. No dens were documented in 2007 or 2008. Reproduction followed a pattern of good and bad years followed by a return to good years in both the reintroduction area and outside the reintroduction area suggesting there may be a cyclic pattern to reproductive output of lynx in Colorado. If the pattern of annual reproductive and survival parameters estimated to date for lynx within the core reintroduction area would repeat over the next 20 years, the population currently in the core reintroduction area would sustain itself at existing densities. To document the continued viability of lynx in Colorado beyond the reintroduction period, some form of long-term monitoring will be needed. A site-occupancy monitoring program using cost-effective, minimally invasive techniques is currently being developed to estimate the extent, stability and potential distribution of lynx throughout Colorado. 2 �WILDLIFE RESEARCH REPORT POST RELEASE MONITORING OF LYNX (LYNX CANADENSIS) REINTRODUCED TO COLORADO TANYA M. SHENK P. N. OBJECTIVE The post-release monitoring of Canada lynx (Lynx canadensis) reintroduced into Colorado emphasized 5 primary objectives: 1. Assess and modify release protocols to ensure the highest probability of survival for each lynx released. 2. Obtain regular locations of released lynx to describe general movement patterns and habitats used by lynx. 3. Determine causes of mortality in reintroduced lynx. 4. Estimate survival of lynx reintroduced to Colorado. 5. Estimate reproduction of lynx reintroduced to Colorado. Three additional objectives were emphasized after lynx displayed site fidelity to an area: 6. Refine descriptions of habitats used by reintroduced lynx. 7. Refine descriptions of daily and overall movement patterns of reintroduced lynx. 8. Describe hunting habits and prey of reintroduced lynx. Information gained to achieve these objectives will form a basis for the development of lynx conservation strategies in the southern Rocky Mountains. SEGMENT OBJECTIVES 1. Complete winter 2009-10 field data collection on lynx habitat use at the landscape scale, hunting behavior, diet, mortalities, and movement patterns. 2. Complete data collection for the pilot study designed to estimate lynx detection probabilities using non-invasive techniques. 3. Complete spring 2010 field data on lynx reproduction. 4. Summarize and analyze data and publish information as Progress Reports, peer-reviewed manuscripts for appropriate scientific journals, or CDOW technical publications (see Appendix I). 5. Complete field research on the post-release monitoring of lynx reintroduced to Colorado and prepare a final report describing status of the lynx reintroduction. INTRODUCTION The Colorado Division of Wildlife implemented the largest Canada lynx (Lynx canadensis), and one of the largest carnivore, reintroductions programs undertaken to date. Thus, evaluating success of this program is critical, and assessing the methods used may prove useful for other ongoing or future carnivore reintroductions. The reintroduction effort was begun in Colorado in 1997, with the first lynx released in the state in 1999. The goal of the Colorado lynx reintroduction program was to establish a self-sustaining, viable population of lynx in this state. The approach taken to reach this goal was to first establish a viable lynx population within a core reintroduction area in southwestern Colorado. From this core reintroduction area, it was hoped that lynx would remain in this area and disperse on their own into 3 �suitable habitat throughout the state. Thus, 218 wild-caught lynx from Canada and Alaska were reintroduced in the core reintroduction area from 1999-2006. There were 7 critical criteria established for achieving a viable lynx population in Colorado: 1) development of release protocols that lead to a high initial post-release survival of reintroduced animals, 2) long-term survival of lynx in Colorado, 3) development of site fidelity by the lynx to areas supporting good habitat in densities sufficient to breed, 4) reintroduced lynx must breed, 5) breeding must lead to reproduction of surviving kittens 6) lynx born in Colorado must reach breeding age and reproduce successfully, and 7) recruitment must equal or be greater than mortality over an extended period of time. These criteria were evaluated incrementally over time to gauge whether the reintroduction effort was progressing toward success (Shenk and Kahn 2002). All seven criteria have now been met. STUDY AREA Byrne (1998) evaluated five areas within Colorado as potential lynx habitat based on (1) relative snowshoe hare densities (Bartmann and Byrne 2001), (2) road density, (3) size of area, (4) juxtaposition of habitats within the area, (5) historical records of lynx observations, and (6) public issues. Based on results from this analysis, the San Juan Mountains of southwestern Colorado were selected as the core reintroduction area, and where all lynx were reintroduced. Wild Canada lynx captured in Alaska, British Columbia, Manitoba, Quebec and Yukon were transported to Colorado and held at The Frisco Creek Wildlife Rehabilitation Center located within the reintroduction area prior to release. Post-release monitoring efforts were focused in a 20,684 km2 study area which included the core reintroduction area, release sites and surrounding high elevation sites (> 2,591 m). The area encompassed the southwest quadrant of Colorado and was bounded on the south by New Mexico, on the west by Utah, on the north by interstate highway 70, and on the east by the Sangre de Cristo Mountains (Figure 1). Southwestern Colorado is characterized by wide plateaus, river valleys, and rugged mountains that reach elevations over 4,200 m. Engelmann spruce/subalpine fir is the most widely distributed coniferous forest type within the study area. The lynx-established core area is roughly bounded by areas used by lynx in the Taylor Park/Collegiate Peak areas in central Colorado and includes areas of continuous use by lynx, including areas used during breeding and denning (Figure 1). METHODS, RESULTS AND DISCUSSION Development of Release Protocols Post-release monitoring was critical to assess and modify the release protocols as they were implemented to improve the survival of released individuals (Shenk 1999). Under the final release protocol, lynx were held in captivity and fed a high quality diet for a minimum of three weeks before release. Thus, they were released in good body condition and one could expect that the longer the captivity, the lower the post-release mortality. This final protocol resulted in high initial post-release survival. Later, detailed analysis of lynx mortality was completed to evaluate how the different release protocols affected mortality within the first year post-release. From this analysis, it was documented that the average monthly mortality rate in the reintroduction area during the first year post-release decreased with time in captivity from 0.205 [95% CI 0.069, 0.475] for lynx spending up to 7 days in captivity to 0.028 [95% CI 0.012, 0.064] for lynx spending > 45 days in captivity before release (Devineau et al. 2010a). The results also suggested that keeping lynx in captivity beyond 5 or 6 weeks accrued little benefit in terms of monthly survival. On a monthly average basis, lynx were as likely to move out (probability = 0.196, SE=0.032) as to move back on (probability = 0.143, SE=0.034) the reintroduction area during the first year after release. Mortality was 1.6x greater outside of the reintroduction area 4 �suggesting that permanent emigration and differential mortality rates on and off reintroduction areas should be factored into sample size calculations for an effective reintroduction effort. Our results will be useful in the development of release and post-release monitoring protocols for future lynx, as well as other carnivore, reintroductions. Long-Term Survival Viability of a reintroduced population requires long-term survival and site fidelity of individuals to the reintroduction area. Over a 10-year period of the reintroduction effort (1999-2009), monthly mortality rate was estimated as lower inside the reintroduction area than outside the reintroduction area, and slightly higher for male than for female lynx, although 95% confidence intervals for sexes overlapped (Devineau et al. 2010). Mortality was higher immediately after release (first month = 0.0368 [SE = 0.0140] inside the study area, and 0.1012 [SE = 0.0359] outside the study area), and then decreased according to a quadratic trend over time. Given the importance of adult survival in the dynamics of longlived species, the long-term, high survival rates estimated for the reintroduced lynx both inside (0.9315, SE = 0.0325) and outside (0.8219, SE = 0.0744) the reintroduction area are promising for the establishment of a viable population of lynx in Colorado (Figure 2, Devineau et al. 2010b). The higher mortality outside the reintroduction area may have been influenced by habitat fragmentation, increased road density and more opportunities for human interactions. From 1999-June 2010, there were 122 known mortalities of released adult lynx. Human-caused mortality factors are currently the highest causes of death with approximately 29.7% attributed to collisions with vehicles or gunshot. Starvation and disease/illness accounted for 18.6% of the deaths while 37.3% of the deaths were from unknown causes. Lynx mortalities were documented throughout all areas lynx used, including 31 (26.3%) occurring in other states. Reproduction Reproduction is necessary to achieve a self-sustaining viable population of lynx in Colorado. Reproduction was first documented from the 2003 reproduction season and again in 2004, 2005 and 2006. Lower reproduction occurred in 2006, although a Colorado-born female gave birth to 2 kittens, documenting the first recruitment of Colorado-born lynx into the Colorado breeding population. No reproduction was documented in 2007 or 2008. The cause of the decreased reproduction from 2006 -08 is unknown. One possible explanation would be a decrease in prey abundance. Reproduction was again observed in 2009 with 5 dens and 10 kittens found in Colorado. Litter size was smaller than previously documented with only 2 kittens found in each litter in comparison to a mean of 2.8 found in previous years. In addition, a sex bias towards female kittens was evident in 2009 which was not evident in prior years. Two litters found in 2009 had both parents born in Colorado, resulting in the first documented third generation Colorado lynx from the reintroduction. The percent of females having dens increased in 2010 to 33%, similar to the highest years documented in 2004-2005. The average number of kittens per litter also returned to the previously observed mean of 2.8. Breeding males and females in 2010 included Colorado-born lynx that have established territories and are now contributing to the breeding population. Reproduction has followed a pattern of good and bad years followed by a return to good years in both the reintroduction area (Figure 3) and outside the reintroduction area suggesting there may be a cyclic pattern to reproductive output of lynx in Colorado. Such a pattern matches the classic Canada lynx-snowshoe hare (Lepus americanus) cycle (Elton 1942). Long-term studies spanning an additional10-20 years would be required to document such a cycle in Colorado. Viability The current lynx population in Colorado is comprised of surviving reintroduced adults, lynx born in Colorado from the reintroduced animals and their offspring and possibly some naturally occurring lynx. To achieve a self-sustaining, viable population of lynx, enough kittens need to be born and 5 �recruited into this population to offset the mortality that occurs and hopefully even exceed the mortality rate to achieve an increasing population. If the pattern of annual reproductive and survival parameters estimated to date for lynx within the core reintroduction area would repeat over the next 20 years, the population currently in the core reintroduction area would sustain itself at existing densities (Figure 4). FUTURE DIRECTIONS Research and monitoring efforts over the last 11 years, since the first lynx were released, have focused primarily on monitoring reintroduced animals through VHF and satellite telemetry and estimating demographic parameters of these animals. However, as more of these animals become unavailable for monitoring due to failed telemetry collars, death or movement out of the core reintroduction area, it becomes more difficult to accurately evaluate the status of the entire lynx population in Colorado, including the core reintroduction area. To document the continued viability of lynx in Colorado beyond the reintroduction period, some form of long-term monitoring will be needed to determine viability for a period of time long enough to encompass possible snowshoe hare cycles. In addition, a challenge facing Colorado Division of Wildlife is how efforts should be allocated between monitoring persistence of lynx that have established within the core reintroduction area and lynx that may be pioneering and expanding into other portions of the state. A site-occupancy monitoring program using cost-effective, minimally invasive techniques is currently being developed to estimate the extent, stability and potential distribution of lynx throughout Colorado (Shenk 2009, Appendix 2). The primary objectives of this monitoring program would be to document the distribution of lynx throughout Colorado and the stability, growth or shrinkage of this distribution over time, and to identify potential areas lynx may occupy in the future. Minimally invasive techniques (e.g., genetic identification, cameras) would be used to detect changes in lynx persistence and distribution as a foundation for assessing whether lynx continue to persist in Colorado. Such noninvasive techniques are widely desirable because they require minimal impact to the animals and are costeffective. The protocols developed will also be made available to any other agencies or entities that want to monitor lynx. Methods to extend this monitoring effort to estimate lynx density are currently being pursued. ADDITIONAL EFFORTS Additional goals of the post-release monitoring program for lynx reintroduced to the southern Rocky Mountains included refining descriptions of habitat use and movement patterns of lynx once lynx established home ranges that encompassed their preferred habitat. This work is ongoing. The program also investigated the ecology of snowshoe hare in Colorado. A study comparing snowshoe hare densities among mature stands of Engelmann spruce (Picea engelmannii)/subalpine fir (Abies lasiocarpa), lodgepole pine (Pinus contorta) and Ponderosa pine (Pinus ponderosa) was completed in 2004 with highest hare densities found in Engelmann spruce/subalpine fir stands and no hares found in Ponderosa pine stands (Zahratka and Shenk 2008). A study to evaluate the importance of young, regenerating lodgepole pine and mature Engelmann spruce/subalpine fir stands in Colorado by examining density and demography of snowshoe hares that reside in each was completed in 2010. Small lodgepole stands supported the highest densities of hares as well as the highest and most consistent recruitment rates. Hares survived best in spruce/fir stands while density and recruitment in these stands were intermediate. Thus, small lodgepole and mature spruce/fir likely provide the most important hare habitat in Colorado; while thinned, medium lodgepole stands appear to be relatively unimportant based on the density and demography measures in this study (J. Ivan, Colorado State University, unpublished data, Appendix 3). However, within the study area, small lodgepole stands occupied only 10% of the area 6 �covered by mature spruce/fir, and we suspect a similar pattern statewide. Additionally, the structure provided by mature spruce/fir stands is less transient than that provided by regenerating lodgepole. Thus, while density and recruitment estimates in spruce/fir stands were somewhat inferior to those collected in small lodgepole, the areal coverage and longevity of spruce/fir likely renders it as important, if not more important, to snowshoe hare and lynx management in Colorado as regenerating lodgepole (J. Ivan, Colorado State University, unpublished data, Appendix 3). Lynx is listed as threatened under the Endangered Species Act (ESA) of 1973, as amended (16 U. S. C. 1531 et. seq.)(U. S. Fish and Wildlife Service 2000). Colorado is included in the federal listing as lynx habitat. Thus, an additional objective of the post-release monitoring program is to develop conservation strategies relevant to lynx in Colorado. To develop these conservation strategies, information specific to the ecology of the lynx in its southern Rocky Mountain range, such as habitat use, movement patterns, mortality factors, survival, and reproduction in Colorado have been and will continue to be provided to regulatory agencies. SUMMARY From results to date it can be concluded that the Colorado Division of Wildlife developed release protocols that ensured high initial post-release survival of lynx, and on an individual level, lynx demonstrated they can survive long-term in areas of Colorado. We also documented that reintroduced lynx exhibited site fidelity, engaged in breeding behavior and produced kittens that were recruited into the Colorado breeding population. Following the successful reproduction in 2010, we have now documented that if the population would repeat the reproduction and mortality patterns documented over the last 10 years the lynx population would continue into the future at sustainable numbers. Thus, the final criterion of a successful reintroduction, documenting recruitment necessary to offset annual mortality, is now supported. To build upon the success of this reintroduction effort, effective conservation and management strategies will need to be developed and implemented to ensure the long-term viability of Canada lynx in Colorado. ACKNOWLEDGEMENTS The Colorado Lynx Reintroduction Program required the continued efforts of numerous personnel in the Colorado Division of Wildlife, other agencies and the general public. Such sustained dedication has resulted in the successful reintroduction of this species to our ecosystems. Funding for the reintroduction program was provided by Colorado Division of Wildlife, Great Outdoors Colorado (GOCO), Vail Associates, Colorado Wildlife Heritage Foundation, Turner Endangered Species Foundation and the U.S.D.A. Forest Service. LITERATURE CITED Bartmann, R. M., and G. Byrne. 2001. Analysis and critique of the 1998 snowshoe hare pellet survey. Colorado Division of Wildlife Report No. 20. Fort Collins, Colorado. Byrne, G. 1998. Core area release site selection and considerations for a Canada lynx reintroduction in Colorado. Report for the Colorado Division of Wildlife. Devineau, O., T. M. Shenk, P. F. Doherty Jr., G. C. White, and R. H. Kahn. 2010. Assessing release protocols for the Colorado Canada lynx (Lynx canadensis) reintroduction. Journal of Wildlife Management (in review). Devineau, O., T. M. Shenk, G. C. White, P. F. Doherty Jr., P. M. Lukacs, and R. H. Kahn. 2010. Evaluating the Canada lynx reintroduction programme in Colorado: patterns in mortality. Journal of Applied Ecology 47:524-531. 7 �Elton, C. and M. Nicholson 1942. The ten-year cycle in numbers of lynx in Canada. Journal of Animal Ecology 11: 215-244. Shenk, T. M. 2002. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report, July: 7- 34. Colorado Division of Wildlife, Fort Collins, Colorado. __________. 2009. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report, July: 1-57. Colorado Division of Wildlife, Fort Collins, Colorado Shenk, T. M. and R. H. Kahn. Lynx reintroduction: report to wildlife commission. Colorado Division of Wildlife. U. S. Fish and Wildlife Service. 2000. Endangered and threatened wildlife and plants: final rule to list the contiguous United States distinct population segment of the Canada lynx as a threatened species. Federal Register 65, Number 58. Zahratka, J. L. and T. M. Shenk. 2008. Population estimates of snowshoe hares in the southern Rocky Mountains. Journal of Wildlife Management 72:906-912. Prepared by __________________________________________________________ Tanya Shenk, Wildlife Researcher & Jake Ivan, Wildlife Researcher 8 �Figure 1. Lynx are monitored throughout Colorado and by satellite throughout the western United States. The lynx core release area, where all lynx were released, is located in southwestern Colorado (outlines in white). A lynx-established core use area has developed in the Taylor Park and Collegiate Peak area in central Colorado. 9 �Figure 2. Variation of monthly mortality rate with time since release for Canada lynx reintroduced to Colorado, inside and outside of the study area, according to the best-AICc model (from Devineau et al. 2010). Only the first 50 months following release are shown. Figure 3. Percent of tracked Canada lynx females in the reintroduction area found with kittens in May or June from 2003 through 2010. 10 �Figure 4. Projected Canada lynx population trend in the core reintroduction area over 20 years if the pattern of reproductive and survival parameters observed over the last 8 years would repeat. The initial population sizes of 50 males and 50 females for this projection was not based on a current population estimate, however, they are not unreasonable assumptions for the study area. Using alternative initial population sizes would not change the projected pattern. 11 �APPENDIX I STATUS OF PUBLICATIONS ASSOCIATED WITH THE COLORADO LYNX REINTRODUCTION PROGRAM Five papers have been published: Devineau, O., T. M. Shenk, G. C. White, P. F. Doherty, Jr., P. M. Lukacs, and R. H. Kahn. 2010. Evaluating the Canada lynx reintroduction programme in Colorado: patterns in mortality. Journal of Applied Ecology 47:524–531. Shenk, T. M., R. H. Kahn, G. Byrne, D. Kenvin, S. Wait, J. Seidel, and J. Mumma. 2009. Canada lynx (Lynx canadensis) reintroduction in Colorado. Pages 410-421 in A. Vargas, C. Breitenmoser, and U. Breitenmoser, editors. Iberian Lynx Ex situ Conservation: An Interdisciplinary Approach. Fundacion Biodiversidad, Madrid, Spain. Shenk, T. M and, R. H. Kahn. 2009. Reintroduction of the Canada lynx (Lynx canadensis) to Colorado. in Proceedings of the Third Iberian Lynx Symposium. eds. A. Vargas, C. Breitenmoser, U. Breitenmoser, Fundacion Biodiversidad and IUCN Cat Specialist Group. Fundacion Biodiversidad, Spain. Wild, M. A., T. M. Shenk, and T. R. Spraker. 2006. Plague as a mortality factor in Canada lynx (Lynx canadensis) reintroduced to Colorado. Journal of Wildlife Diseases 42:646–650. Zahratka, J. L., and T. M. Shenk. 2008. Population estimates of snowshoe hares in the southern Rocky Mountains. Journal of Wildlife Management 72:906–912. Five additional papers are currently in review: Devineau, O., T. M. Shenk, P. F. Doherty, Jr., G. C. White, and R. H. Kahn. In review. Assessing release protocols used for the Canada lynx (Lynx Canadensis) reintroduction in Colorado: Recommendations for future efforts. Journal of Wildlife Management. Devineau, O., T. M. Shenk, P. F. Doherty, Jr., et al. In review. Modeling known-fate and nest survival data within the multistate framework: increased flexibility for telemetry studies. Journal of Applied Ecology. Wolfe, L. L., T. M. Shenk, B. Powell, and T. E. Rocke. In review. Safety of and serum antibody responses to a recombinant F1-V fusion protein vaccine intended to protect Canada lynx (Lynx Canadensis) from plague. Journal of Wildlife Diseases. Fanson, K., T. M. Shenk, et al. In review. Patterns of testicular activity in captive and wild Canada lynx. General and Comparative Endocrinology. Fanson, K., T. M. Shenk, et al. In review. Patterns of ovarian and luteal activity in captive and wild Canada lynx. General and Comparative Endocrinology. One paper is in the process of being submitted for publication and requires no additional work from CDOW personnel: Fanson, K., T. M. Shenk, et al. In prep. Patterns of stress physiology in reintroduced Canada lynx and implications for reintroduction success. General and Comparative Endocrinology. 12 �Six publications are currently in preparation and require the continued efforts of Tanya Shenk and/or Jake Ivan to complete: Theobald, D., and T. M. Shenk. In prep. Lynx habitat use at site-specific and landscape scales. Shenk, T. M. In prep. Lynx denning habitat and reproduction in Colorado. Ivan, J. S., G. C. White, and T. M. Shenk. In Prep. Using telemetry to correct for bias: an approach to estimating density from trapping grids. Ecology. Ivan, J. S., G. C. White, and T. M. Shenk. In Prep. Comparison of methods for estimating density from capture–recapture data. Journal of Applied Ecology. Ivan, J. S., G. C. White, and T. M. Shenk. In Prep. Density and demography of snowshoe hares in westcentral Colorado. Ecological Monographs. Ivan, J. S., G. C. White, and T. M. Shenk. In Prep. Daily and seasonal movements of snowshoe hares in west-central Colorado. Journal of Mammalogy. 13 � https://cpw.cvlcollections.org/files/original/75bc975c72c10bed56af1c21eebf0f92.pdf 534b399d721b0b4f52d6d0629725c6b8 PDF Text Text 35 Colorado Division of Wildlife Wildlife Research Report July 2001 and July 2002 JOB PROGRESS REPORT State of Colorado Division of Wildlife - Mammals Research Work Package No. -~0~6~7~0_ _ _ _ _ __ Task No. 2 Lynx Conservation Ecology of Snowshoe Hares (Lepus americanus) in Colorado Period Covered: July l? 2000 - June 30, 2002 Author: Steven W. Buskirk and Jennifer L. Zahratka Personnel: T. M. Shenk, Ph.D. ABSTRACT Despite what is known about Canada lynx (Lynx canadensis) and snowshoe hare (Lepus americanus) ecology in Canada and Alaska, a paucity of information exists in the contiguous United States. With the listing of the Canada lynx as threatened under the Endangered Species Act in 2000 the need for more knowledge about lynx and their prey become more pressing. The recent reintroduction of Canada lynx to southwestern Colorado (1999) by the state has furthered increased this need. The development of reliable knowledge about snowshoe hare ecology will be key to the recovery and de listing of lynx. Two habitat factors are generally considered overriding in their importance to the abundance and fitness of snowshoe hares: the density of small-diameter (generally< 5 mm) woody stems within reach of the snow surface for food, and the abundance of somewhat larger-diameter woody structure for overhead cover. This project focuses on two central conceptual issues. First, how do site conditions produce woody stems of suitable diameters and heights above the snow for food for snowshoe hares in late winter, and how do site conditions provide overhead cover suitable for hares? Second, do snowshoe hares in fact attain their highest densities in these presumptive high-quality habitats? Ecological information gained about snowshoe hares will be valuable not only to the recovery of Canada lynx in Colorado, but also throughout the range oflynx in the southern U.S. ��37 Ecology of Snowshoe Hares (Lepus Americanus) in Colorado Steven W. Buskirk and Jennifer L. Zahratka Department of Zoology and Physiology, University ofWyoming Laramie, Wyoming 82071 Introduction The snowshoe hare (Lepus americanus) is a widely distributed and well-studied leporid of North American boreal forests. Scientists have long been interested in the snowshoe hare and its cyclic relationship with the Canada lynx (Lynx canadensis). The snowshoe hare is the obligate primary prey item of the lynx, which was listed as threatened under the Endangered Species Act in 2000. Data dealing with the ecology, particularly the habitat ecology, of southern snowshoe hare populations is lacking, especially in the southern Rocky Mountains. Indeed, only a single study (Dolbeer and Clark 1975) described the habitat associations of hares in the southern Rocky Mountains, but only in the most cursory fashion. The reintroduction of Canada lynx to the southern Rocky Mountains in 1999-2000 has further stimulated the need for understanding the habitat requirements of snowshoe hare populations. Therefore, data from the southern Rocky Mountains is critical for understanding the ecology of snowshoe hares in their southern range. The abundance and fitness of snowshoe hares depend on the protection afforded by plants as well as their suitability as foods for hares. Although food is an obvious requirement for snowshoe hare survival, snowshoe hares rarely starve to death. Instead, predation is the overwhelming proximate cause of death for snowshoe hares (Hodges 2000b) and food shortage only predisposes them to predation. The cover afforded by large-diameter woody structure provid~s horizontal and vertical protection from predators (Wolff 1980). Also, small-diameter(< 5-mm) (Grigal and Moody 1980) woody stems< 45 cm from the snow surface (Bider 1961) are an important food source (Hodges 2000a). Whereas largediameter woody stems presumptively provide protection from predation, small-diameter woody sterns presumptively provide nutrition. Therefore, we assume that woody structure in two different size classes meets the needs of snowshoe hares for habitat. Winter is a critical time of year for snowshoe hare survival because fewer woody stems of either size class are available in winter than in other seasons, and herbaceous plants are not available. Understanding how the density of woody stems of different sizes, tree dominants, and successional stage affect densities of snowshoe hares is key to effective management of snowshoe hare habitats in the southern Rocky Mountains. Therefore, we investigated two conceptual issues relating to snowshoe hare habitat in late winter. First, how do site conditions produce woody stems of suitable diameters and heights above the snow surface for food and how do site conditions provide suitable protective cover for hares? Second, do snowshoe hares in fact attain their highest densities in these presumptive high-quality habitats? Study Area Location The study area was a broad area of southwestern Colorado on the Gunnison and Rio Grande National Forests, which we studied during January-April 2002. Within our study area, we established two study sites: one was a 1963-km2 area centered over Taylor Parle Reservoir on the Gunnison National Forest (39° 50' N, 106° 34' W); the second was the Divide District (4,089 km2) of the Rio Grande National Forest (37° 40' N, 106° 40' W) centered directly north of South Fork, Colorado. The Gunnison study area represents the southernmost extent of naturally occurring lodgepole pine. In coniferous forests of the Rocky Mountains, lodgepole pine is an important habitat type for lynx and �38 snowshoe hares. The Rio Grande study area is lower in elevation and contains ponderosa pine, also widely distributed in the Rocky Mountains, and therefore of interest in our study. Topography The landscape of southwestern Colorado is characterized by high, rugged mountains, wide plateaus, and gaping river valleys. Elevation of the Gunnison study site ranged from 2850 m to 3480 m. On the Rio Grande study site the elevation ranged from 2460 m to 2580 m. Our spruce-fir sites occurred at elevations of 32 l 0 - 3480 m, our lodgepole pine sites occurred at 2850 - 3100 m, and our ponderosa pine sites occurred at 2460 - 2580 m. Climate Southwestern Colorado exhibits an arid and temperate climate; strong local variation responds to elevation and aspect. The mean temperature in Gunnison, Colorado from January -April is -7°C. In South Fork, Colorado the mean temperature from January -April is 0°C (National Weather Service, Gunnison, CO, unpublished data). Unlike areas of the Western Slope where more precipitation falls as winter snow than as summer rain, the monsoon season in southwestern Colorado brings most yearly precipitation in late summer. The mean monthly precipitation in Gunnison, Colorado for January -April is 1.6 cm. In South Fork, Colorado the corresponding mean is 1.5 cm (National Weather Service, Gunnison, CO, unpublished data). Methods Trapping grid selection Our study area comprised the Gunnison and Rio Grande National Forests, within which trapping grids were chosen using a GIS database of national forest lands with Common Vegetation Unit (CVU) coverage using the Integrated Resource Inventory protocol (IRI) made available by each of the forests. Two sets of criteria, applied sequentially, were used to select the site of the trapping grids. The first set of criteria was based upon the CVU coverages using GIS: I. Species dominants represented were lodgepole pine, Engelmann spruce, ponderosa pine and riparian (Salix spp.). 2. For forests, structural stage considered was mature (structural stage 4). 3. Vegetation polygons were candidate if ~30 m, but~ 1 km from an improved road. 4. Vegetation polygons were candidate if ~25 ha. 5. Vegetation polygons were candidate if sufficient to admit a 330 m x 550 m (16.5-ha) trapping grid with a 50-m buffer between the edge of ~e trapping-grid an~ the nearest edge ·of the • polygon. Fifteen of the candidate polygons were selected randomly. Within each of these random polygons a 330 m x 550 m rectangle was placed at a randomly generated orientation (0 - 180°). All potential ponderosa pine sites on the Gunnison National Forest were excluded using these criteria. All potential riparian sites on the Rio Grande were excluded using these criteria and no lodgepole pine was available on the Rio Grande to evaluate by CVU layers. Potential sites were visited in random order, at which time we applied the second set of criteria: 1. Forested sites were excluded if ~40% of the trapping grid was dominated by a cover type other than the nominal species dominant. 2. Candidate sites were excluded if inaccessible by snowmobile and snowshoes. 3. Candidate sites were excluded if they held any unmapped roads. 4. Candidate sites were excluded iflogging or thinning had occurred within them. �39 5. Candidate sites were excluded if avalanche danger was present. 6. Candidate sites were excluded if trapping grids were <500 m from a grid that had already been included. The first three from each species dominant to meet these criteria were included as trapping grids. Because of the availability of suitable sites, and for logistical reasons, all three spruce-fir trapping grids, all three lodgepole pine trapping grids and all three riparian trapping grids were selected on the Gunnison National Forest. Only the three ponderosa pine trapping grids were selected on the Rio Grande National Forest. After visiting fourteen sites mapped as lodgepole pine on the Gunnison National Forest, three were found that met our criteria. Fifteen sites mapped as spruce-fir on the Gunnison National Forest were evaluated before three were found that met our criteria. Ten sites tentatively mapped as riparian on the Gunnison were visited, but none were found that met our criteria. Fifteen sites mapped as ponderosa pine-dominant on the Rio Grande National Forest were visited before three were found that met our criteria. Trapping and handling All methods related to trapping and handling were approved by the University of Wyoming Animal Care and Use Committee and by the Colorado Division of Wildlife Animal Care and Use Committee. Snowshoe hares were trapped using Tomahawk Model 204 live traps (18 cm x 18 cm x 51 cm) placed on trapping grids of 84 traps (7 lines of 12 traps each), with 50-m spacing for a trapping grid size of 16.5 ha (Fig. 1). Three replicates for each species dominant were sampled for 6 trap nights, which we assumed to be a closed population for the purposes of mark-recapture models. No reproduction occurred during our winter field season. The trapping grid size and method were developed by Scott Mills and Paul Griffin, University of Montana; we used these methods to maximize comparability between our study and theirs. Upon visiting a suitable site, the trapping grid was flagged and numbered using the UTM coordinates generated by a GPS receiver and a compass bearing (Fig. 1). Traps were placed in suitable habitat within 2 m of the flagging and if necessary, covered with tree branches to provide cover for captured hares. Traps were baited with a mixture of pellets of Timothy grain, alfalfa, com, and oats (TACO), alfalfa pellets and apples (P. Griffin, pers. comm.). Traps were checked as early as possible each morning and re-baited as needed. Once a snowshoe hare was captured, a pillowcase with a drawstring was placed over the front door of the trap. The hare was persuaded into the bag by gently tipping the trap, blowing on the hare, or making noise. Once the hare was in the bag it was immediately weighed using a 2500-g Pesola spring scale. The hare was then carefully placed between the legs of a ~eeling handler with the head facing towards the handler. The second handler marked the hare using a sterile passive-integrated transponq.er (PIT) tag. One tag was injected subcutaneously with a sterile needle between the shoulder blades. Both ears of the snowshoe hare were also marked using a permanent black marker for short-term identification. After the first day of any trapping session (i.e. on traps days 2-6) every snowshoe hare was scanned with a 125-kHz Mini-portable reader to determine whether the hare was a recapture or a new capture. fu the event the snowshoe hare was preyed upon and partially ingested, the earmarks were checked. Each snowshoe hare was sexed by turning the hare on its dorsal side and protruding the genitalia. The forefinger and middle finger were used to apply slight pressure to the vent area just above the anus. Snowshoe hares were then released away from handlers. Snowshoe hares that suffered severe trap or predation injuries were euthanized with a 1-ml intrathoracic injection of sodium pentobarbital. Each carcass was necropsied and the liver and kidneys preserved for metals analysis. After necropsy and tissue collection, euthanized animals were disposed of by cremation or deposited in a landfill. Any non-target species caught-in traps were immediately released. �40 Diet Within spruce-fir stands, where captures were expected to be more common, we marked trap bait in order to determine whether feces collected from traps contained any bait. TACO and alfalfa pellets were marked with a light dusting of fluorescent, non-toxic powder (DayGlo). Fecal pellets were then collected from the inside of each live-trap where a snowshoe hare was captured. Every fecal pellet within the trap was collected, placed in a brown paper bag and allowed to dry at room temperature. After collection the fecal pellets were placed under an ultraviolet light to show presence or absence of any fluorescent marker ingested by the hares. Samples will be submitted to the Wildlife Habitat and Nutrition Lab at Washington State University, Pullman, WA for analysis. Measurement offecal pellets Each fecal pellet was measured to the nearest 0.1 mm using SPI dial calipers. The sizes of all fecal pellets were recorded, and the mean pellet size for each individual hare was calculated. Vegetation measurements Habitat attributes were estimated at two levels: at each trap site and at each trapping grid. Within each trapping grid, vegetation plots were sampled from 15 trap sites, similar to the design of S. Mills (Fig. 1). Methods developed by T. Shenk (Colorado Division of Wildlife) to monitor habitat use by reintroduced lynx to Colorado were followed, with some minor modifications (Fig. 2). Accordingly, a 12-m x 12-m square of 25 points was placed in 5 rows of 5 (3 m apart), centered over the trap location (Fig. 2). The measurements taken at each of the 25 points included: 1. Snow depth (cm), as measured by a calibrated avalanche probe. 2. Understory measured in a column of 3-cm radius around an avalanche probe. a. All live or dead stems and coarse woody debris (CWD) that fall within the 3-cm radius column using the standardized four-letter genus-species code at 3 height categories (00.5 m, 0.51-1.0 m, 1.01-1.5 m) above the snow surface. b. Each of the above stems classified in 3 different diameter categories(< 5 mm, 5.1-10 mm, 10.1-15 mm) measured at the point where the stem hit the avalanche probe. 3. Overstory was measured using a sighting tube ("moosehom') attached to the avalanche probe. a. Species that hit the crosshairs inside the sighting tube were recorded. Multiple hits by the same species were only recorded once. 4. Every shrub within the plot along with its species and diameter at breast height was recorded (dbh). 5. Every tree within the plot along with its species and dbh was recorded. 6. Every snag within the plot along with its dbh was recorded. 7. Every sapling within the plot along with its species was counted. 8. All coarse woody debris (CWD) deemed usable by snowshoe hare for cover or food (i.e., available above the snow) was recorded along with its diameter. At all of the 84 trap sites within the trapping grid, including the 15 trap sites sampled as described above, the following data were measured: 1. Snow depth (cm), as measured by a calibrated avalanche probe. 2. Species of, dbh of, and distance to the closest woody stem in two categories: ~ 1.0 cm - 7.0 and ~ 7 .1 cm at the snow surface. 3. Canopy cover for the center of the trap site, as estimated by the use of a spherical densiometer, in the four cardinal quadrants (NW, NE, SE, SW). The following rules were used for unusual events: 1. If a point in a vegetation plot lay within a tree bole, the tree species and the dbh was written on the data form. �41 2. A snag was defined as any dead tree bole >45 ° from the horizontal. Dead boles <45 ° vertical angle were considered CWD. 3. The mid-point diameter was measured of exposed CWD partially covered by snow. 4. If a leaning tree fell partially outside the 12 m x 12 m sampling plot it was included if >50% of the tree lay within the sampling plot. Results Captures of snowshoe hares by trapping grid and tree species dominant are summarized in Table 1. A total of 28 hares were captured in 4620 trap nights of effort. Mean dbh and density of trees, by species, for the nine trapping grids in three tree species domiriant categories are summarized in Tables 23. Snow depths, and densities of various vegetative structures for the nine trapping grids ( 15-point protocol) in three tree species dominant categories are summarized in Table 4. Corresponding data for the 84-point protocol are summarized in Table 5. Literature Cited Bider, J. R. 1961. An ecological study of the hare Lepus americanus. Canadian Journal of Zoology 39:81-103. Dolbeer, RA., and W.R. Clark. 1975. Population ecology of snowshoe hares in the central Rocky Mountains. Journal of Wildlife Management 39:535-549. Grigal, D. F., and N. R. Moody. 1980. Estimation of browse by size classes for snowshoe hare. Journal of Wildlife Management 44:34-40. Hodges, K. E. 2000a. The ecology of snowshoe hares in northern boreal forests. Pages 117-161 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, and J. R. Squires, editors. Ecology and conservation oflynx in the United States. University Press of Colorado, Boulder, Colorado. Hodges, K. E. 2000b. Ecology of snowshoe hares in southern boreal and montane forests. Pages 163206 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, and J. R Squires, editors. Ecology and conservation oflynx in the United States. University Press of Colorado, Boulder, Colorado. Hoover, R. L., and D. L. Wills. 1987. Managing forested lands for wildlife. Pages 455-477. Colorado Division of Wildlife, Eastwood Printi~g and Publishing, Denver, Colorado. Wolff, J. 0. 1980. The role of habitat patchiness in the population dynamics of snowshoe hares. Ecological Monographs 50:111-130. �42 Figure 1. Schematic of 300 m x 550 m trapping grid to be used for estimating population density of snowshoe hares in southern Colorado. Asterisks(*) indicate the location of the 15 vegetation plots centered on trapping points. Pound signs(#) indicate where the point-quarter method will be used on all other trap locations. 1# 2# 3# 4# 5# 6# 7# 8# 9* 10# 11* 12# 13* 14# 15# 16# 17# 18# 19# 20# 21# 22# 23* 24# 25* 26# 27* 28# 29# 30# 31# 32# 33# 34# 35# 36# 37* 38# 39* 40# 41* 42# 43# 44# 45# 46# 47# 48# 49# 50# 51* 52# 53* 54# 55* 56# 57# 58# 59# 60# 61# 62# 63# 64# 65* 66# 67* 68# 69* 70# 71# 72# 73# 74# 75# 76# 77# 78# 79# 80# 81# 82# 83# 84# �43 Figure 2. Schematic of 12 m x 12 m vegetation plot centered on each of the 15 trap sites (Fig. 1) used in measuring habitat variables for snowshoe hares in southwestern Colorado, late winter 2002. The trap location is at the center of the vegetation plot. II 12m �44 Table I. The number of snowshoe hare captures (1st, 2nd, and 3rd), and total captures on nine trapping grids in three species dominant categories, and trapping effort (trap-nights), southwestern Colorado, late winter 2002. 3rd 1st 2nd Total TrapTrapping grid number LPl LP2 LP3 SFl SF2 SF3 PPl PP2 PP3 Tree species dominant Pinus contorta Pinus contorta Pinus contorta Picea engelmannii, Abies Iasiocarpa Picea engelmannii, Abies lasiocarpa Picea engelmannii, Abies lasiocarpa Pinus ponderosa Pinus ponderosa Pinus ponderosa capture 1 3 1 5 3 15 0 0 0 capture 0 0 0 2 0 6 0 0 0 capture 0 0 0 0 0 2 0 0 0 capture 1 3 1 7 3 23 0 0 0 nights 504 504 504 504 588 504 504 504 504 Table 2. Mean diameter at breast height (dbh) by tree species for 15 trap locations on nine trapping grids in three species dominant categories, southwestern Colorado, late winter 2002. All measurements are in cm± SE (where n> 1). Abies lasiocarpa Pinus contorta Pinus ponderosa Populus tremuloides NA NA NA NA NA NA 14±1 14±1 15±1 23±3 14±1 20±2 13±2 10±1 11±2 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 27±3 25±3 21±2 11±1 6 3 21±2 18±1 20±3 9±2 Picea Trapping grid number engelmannii LPl LP2 LP3 SFl SF2 SF3 PPl PP2 PP3 MA Psuedotsuga Juniper-us menziesii scopulorum 8 NA Table 3. Mean density by tree species for 15 trap locations on nine trapping grids in three species dominant categories, southwestern Colorado, late winter 2002. All measuremerits are in trees ha- 1•. Picea Trapping grid engelmannii number LPl LP2 LP3 SFl SF2 SF3 PPl PP2 PP3 NA NA 5±5 704±149 1231±159 1194±178 NA NA MA Abies lasiocarpa Pinus contorta NA NA NA 1273±171 1218±231 1310±313 227±64 449±171 449±112 NA NA NA NA NA NA NA NA NA Pinus ponderosa Populus tremuloides NA NA NA NA NA NA 46±19 93±24 120±27 NA NA NA NA NA NA 28±19 5±5 5±5 Psuedotsuga Juniper-us menziesii scopulorum NA NA NA NA NA NA 125±15 14±7 69±22 NA NA NA NA NA NA 51±27 5±5 NA �45 Table 4. Mean snow depth, tree density, sapling density, shrub density, and snag density for 15 trap locations on Colorado, late winter, 2002. All measurements are in cm± SE. Species dominant categories are listed in Table 1. Trapping grid number Snow depth at time of sanipling ± SE Tree density (ha.J) Sapling density Shrub density (ha-I) (ha-I) Snag density (ha- 1) LP I LP2 LP3 SF 1 SF2 SF 3 pp 1 PP2 PP3 37±1 38±2 32±1 70±4 70±3 75±3 0 0 0 1273±171 1218±231 1314±312 931±150 1680±215 1643±180 250±76 116±28 194±31 569±203 333±148 759±203 546±134 749±254 630±154 273±165 481±161 148±67 NA NA NA NA NA NA 315±114 722±305 921±434 620±139 278±95 431±98 162±50 282±44 417±108 162±67 379±125 277±133 Table 5. Mean snow dept~ mean canopy cover, mean diameter at breast height (dbh), and mean distances to nearest stem in two diameter categories for 84 trap locations on nine trapping grids in three species dominant categories, southwestern Colorado, late winter 2002. All measurements are in cm± SE, except canopy cover, which is% ±SE. Trapping grid number LP 1 LP2 LP3 SF 1 SF2 SF 3 pp 1 PP2 pp 3 Species dominant Pinus contorta Pinus contorta Pinus contorta Picea engelmannii, Abies Iasiocarpa Picea engelmannii, Abies Iasiocarpa Picea engelmannii, Abies Iasiocarpa Pinus ponderosa Pinus ponderosa Pinus ponderosa Snow depth at time of sampling Canopy cover dbh of woody stems> 7 cm Distance to nearest stem 1-7 cm dbh Distance to nearest stem > 7 cmdbh 40±1 38±2 40±1 73±2 73±2 79±1 69±2 79±2 15±1 17±1 18±1 28±2 352±35 303±26 351±52 313±38 116±7 146±11 163±14 216±19 66±1 75±2 24±1 238±21 165±16 75±2 85±2 21±1 153±13 145±11 0 0 0 37±4 24±3 48±3 27±1 25±1 26±1 490±41 382±35 479±46 422±37 551±43 291±32 �5 JOB PROGRESS REPORT State of Division of Wildlife- Mammals Research Colorado Work Package No. _ _----"-0-=-67-'--'0"------- Lynx Conservation Task No. Ecology of Snowshoe Hares (Lepus americanus) in Colorado 2 Period Covered: July 1, 2002 - June 30, 2003 Author: Steven W. Buskirk and Jennifer L. Zahratka Personnel: T. M. Shenk Interim Report - Preliminary Results This work continues, and precise analysis of data has yet to be accomplished. Manipulation or interpretation of these data beyond that contained in this report should be labeled as such and is discouraged. ABSTRACT How the densities of woody stems of different sizes, tree dominants, and successional stage affect densities of snowshoe hares is key to effective management of snowshoe hare habitats in the southern Rocky Mountains. Therefore, we investigated two conceptual issues relating to snowshoe hare habitat in late winter. First, how do site conditions produce woody stems of suitable diameters and heights above the snow surface for food and how do site conditions provide suitable protective cover for hares? Second, do snowshoe hares in fact attain their highest densities in these presumptive high-quality habitats? The results in this progress report are preliminary and subject to revision based upon continuing analyses of data. Still, some patterns in the data are apparent. Temperature appeared to have an effect on capture success whereas moon phase, although it has been reported to have an effect, did not. Our preliminary analysis of vegetation data suggests that canopy cover and distance to the nearest 1-7 cm stem also affect capture success. A resource selection model will be generated in the next phase to determine habitat predictors of capture success. Our comparison of diameters of fecal pellets of snowshoe hares and mountain cottontails suggests a difference in size of pellets between the two sympatric lagomorphs that should be useful for identification of pellets to species in the field. �6 Ecology of Snowshoe Hares (Lepus americanus) in Colorado Steven W. Buskirk and Jennifer L. Zahratka Department of Zoology and Physiology University of Wyoming INTRODUCTION The snowshoe hare (Lepus americanus) is a widely distributed and well-studied leporid of North American boreal forests. Scientists have long been interested in the snowshoe hare, its cyclic population fluctuations at high latitudes, and its ecological relationship with the Canada lynx (Lynx canadensis). The snowshoe hare is the obligate primary prey item of the lynx, which was listed as threatened under the Endangered Species Act in 2000 (U.S Fish and Wildlife Service 2000). Data dealing with the ecology, particularly the habitat ecology, of southern snowshoe hare populations are lacking, especially in the southern Rocky Mountains. Indeed, only a single study (Dolbeer and Clark 1975) described the habitat associations of hares in the southern Rocky Mountains, but only in the most cursory fashion. The reintroduction of Canada lynx to the southern Rocky Mountains in 1999-2003 has stimulated the need for understanding the habitat requirements of snowshoe hare populations. Data from the southern Rocky Mountains are critical for managing habitats to conserve lynx and other boreal forest predators at their southernmost limits in the southern Rocky Mountains. The abundance and fitness of snowshoe hares depend on the protection afforded by plants as well as their suitability as food for hares. Although food is an obvious requirement for snowshoe hare survival, snowshoe hares rarely starve to death. Instead, predation is the overwhelming proximate cause of death for snowshoe hares (Hodges 2000a) and food shortage only predisposes them to predation. Largediameter woody structure provides horizontal and vertical protection from predators (Wolff 1980). Also, small-diameter (< 5-mm) (Grigal and Moody 1980) woody stems < 45 cm from the snow surface (Bider 1961) are an important food source (Hodges 2000b). Whereas large-diameter woody stems presumably provide protection from predation, small-diameter woody stems are believed to provide nutrition, particularly in winter. Therefore, we assume that woody structure in two different size classes meets two distinct habitat needs of snowshoe hares. Winter is a critical time of year for snowshoe hare survival because fewer woody stems, large or small, are available than in other seasons, and herbaceous plants are not available. How the densities of woody stems of different sizes, tree dominants, and successional stage affect densities of snowshoe hares is key to effective management of snowshoe hare habitats in the southern Rocky Mountains. Therefore, we investigated two conceptual issues relating to snowshoe hare habitat in late winter. First, how do site conditions produce woody stems of suitable diameters and heights above the snow surface for food and how do site conditions provide suitable protective cover for hares? Second, do snowshoe hares in fact attain their highest densities in these presumptive high-quality habitats? These general questions subsumed more specific ones. 1. In order to understand the links between diet and habitat use in winter, and because diets of snowshoe hares have not been studied in the southern Rocky Mountains, we studied diets of snowshoes hares. 2. Captures of snowshoe hares and non-target leporid species allowed us to collect fecal pellets of known species origin. Because the size of leporid pellets has been used to identify their source to species in the southern Rocky Mountains (Dolbeer and Clark 1975, Hartmann and Byrne 2001) where leporid species are sympatric, we characterized the sizes of fecal pellets of sympatric leporid species, specifically of snowshoe hares and mountain cottontails (Sylvilagus nuttallii). 3. Because various abiotic factors (e.g. air temperature, moon phase) have been reported in the literature (Gilbert and Boutin 1991) or anecdotally to affect capture success of snowshoe hares, we �7 tested for these influences in our data, and accounted for them in our analyses of major treatment effects (e.g. stand type). STUDY AREA Location The study area was a broad area of southwestern Colorado on the Gunnison and Rio Grande National Forests, which we studied during January- April 2002 and January- March 2003. Within our study area, we established two study sites: one was a l 963-km2 area centered over Taylor Park Reservoir on the Gunnison National Forest (39° 50' N, 106° 34' W); the second was the Divide District (4,089 km2) of the Rio Grande National Forest (37° 40' N, 106° 40' W) centered directly north of South Fork, Colorado (Figure 1). Spruce-fir is an important habitat for snowshoe hares throughout its temperate range (Hodges 2000a) and it is the most widely distributed stand type in coniferous forests of Colorado. Approximately 48% of the coniferous forests of Colorado are dominated by spruce-fir (Buttery and Gillam 1987). In Colorado, lodgepole pine accounts for 16% of the coniferous forests (Buttery and Gillam 1987); our Gunnison study area represents the southernmost natural extent of this species. Lodgepole pine is an important habitat type for snowshoe hares in other coniferous forests of the Rocky Mountains (Koehler 1990a, b) and reintroduced lynx have been documented in the Gunnison study area. Therefore, lodgepole pine was included in our study. The Rio Grande study area, although lower in elevation, contains ponderosa pine, also widely distributed in Colorado. About 24% of coniferous forests in Colorado are dominated by ponderosa pine (Buttery and Gillam 1987). Bartmann and Byrne (2001) reported some of their highest densities of lagomorph pellets in ponderosa pine stands. Therefore, it was important for our study to examine the suitability of ponderosa pine stands for snowshoe hares. Topography Southwestern Colorado is characterized by wide plateaus, river valleys, and rugged mountains that reach elevations over 4200 m. Elevations of our Gunnison study site ranged from 2850 m to 3480 m. The Rio Grande study site ranged in elevation from 2460 m to 2580 m. Our spruce-fir sites occurred at elevations of 3210 - 3480 m, our lodgepole pine sites occurred at 2850 - 3100 m, and our ponderosa pine sites occurred at 2460- 2680 m. The overall aspect of each trapping grid varied (Table 1). Climate Southwestern Colorado exhibits an arid and temperate climate; strong local variation reflects elevation and aspect. The mean temperature in Gunnison, Colorado from January - April 2002 was -7°C and in South Fork, Colorado the corresponding mean was 0°C. In 2003 the corresponding mean in Gunnison, Colorado was -5°C and in South Fork was -1°C (Weather Channel web site, unpublished data). Unlike northern Colorado, where more precipitation falls as winter snow than as summer rain, the monsoon season in southwestern Colorado brings most yearly precipitation in late summer. The mean monthly precipitation in Gunnison, Colorado for January - April 2002 was 1.6 cm, whereas in the monsoon months of July and August 2002 the mean was 3.8 cm. In South Fork, Colorado the corresponding means were 1.5 cm and 4.5 cm. �8 METHODS Trapping Grid Selection Our study area comprised the Gunnison and Rio Grande National Forests, within which trapping grids were chosen using a GIS database of national forest lands with Common Vegetation Unit (CVU) coverage using the Integrated Resource Inventory protocol (IRI) made available by each of the forests . Two sets of criteria, applied sequentially, were used to select the site of the trapping grids. The first set of criteria was based upon the CVU coverages using GIS: 1. Stand types included were Engelmann spruce-subalpine fir, lodgepole pine, ponderosa pine and riparian (Salix spp.). 2. Structural stage was mature with canopy cover 2: 40% (SS 4b, 4c) (Buttery and Gillam 1987). 3. Vegetation polygons were considered if 2: 30 m, but :S 1 km from a mapped road, i.e. a highway, paved, graded or gravel road, or a 4-wheel drive road. 4. Vegetation polygons were considered if 2: 25 ha. 5. Vegetation polygons were considered if shaped so as to admit a 330 m x 550 m (16.5-ha) trapping grid with a 50-m buffer between the edge of the trapping grid and the nearest edge of the polygon. 6. Fifteen of the candidate polygons were selected randomly. Within each of these random polygons a 330-m x 550-m rectangle was placed at a randomly generated orientation (0 - 180°). All potential ponderosa pine sites on the Gunnison National Forest were excluded using these criteria. All potential riparian sites on the Rio Grande were excluded using these criteria and no lodgepole pine sites were available on the Rio Grande to evaluate by CVU layers. Potential sites were visited in random order, at which time we applied the second set of criteria: 1. Forested sites were excluded if 2: 40% of the trapping grid was dominated by a cover type other than the nominal species dominant. 2. Candidate sites were excluded if inaccessible by snowmobile and snowshoes. 3. Candidate sites were excluded if they held any unmapped roads. 4. Candidate sites were excluded if logging or thinning had occurred within them. 5. Candidate sites were excluded if avalanche danger was present. 6. Candidate sites were excluded if trapping grids were< 500 m from a grid that had already been included. The first three ~ites from the list of candidates for each stand type to meet these criteria were included as trapping grids. Because of the availability of suitable sites, and for logistical reasons, all spruce-fir trapping grids, all lodgepole pine trapping grids and all riparian trapping grids were evaluated on the Gunnison National Forest. Only the ponderosa pine trapping grids were evaluated on the Rio Grande National Forest. After visiting 14 sites mapped as lodgepole pine on the Gunnison National Forest, three were found that met our criteria. Fifteen sites mapped as spruce-fir on the Gunnison National Forest were evaluated before three were found that met our criteria. Ten sites tentatively mapped as riparian on the Gunnison were visited, but none were found that met our criteria. Fifteen sites mapped as ponderosa pine on the Rio Grande National Forest were visited before three were found that met our criteria. Trapping and Handling All methods related to trapping and handling of animals were approved by the University of Wyoming Animal Care and Use Committee and by the Colorado Division of Wildlife Animal Care and Use Committee. Snowshoe hares were trapped using Tomahawk Model 204 live traps (18 cm x 18 cm x 51 �9 cm) placed on trapping grids of 84 traps (7 lines of 12 traps each), with 50-m spacing for a trapping grid size of 16.5 ha (Figure 2). Three replicates for each stand type were sampled for 6 trap nights, which we assumed to be a closed population for the purposes of mark-recapture models. No reproduction occurred during our winter field season. The trapping grid size and method were developed by Scott Mills and Paul Griffin, University of Montana; we used these methods to maximize comparability between our study and theirs. Upon visiting a suitable site, the trapping grid was flagged and numbered using the UTM coordinates generated by a GPS receiver and a compass bearing (Figure 2). Traps were placed in suitable habitat within 2 m of the flagging and if necessary, covered with tree branches to provide cover for captured hares. Traps were baited with a mixture of pellets of Timothy grain, alfalfa, corn, and oats (TACO), alfalfa pellets and apples (P. Griffin, pers. commun.). Traps were checked as early as possible each morning and re-baited as needed. Once a snowshoe hare was captured, a pillowcase with a drawstring was placed over the front door of the trap. The hare was moved into the bag by gently tipping the trap, blowing on the hare, or making noise. Once the hare was in the bag it was immediately weighed using a 2500-g Pesola spring scale. The hare was then placed between the legs of a kneeling handler with the head facing towards the handler. The second handler marked the hare using a sterile passive-integrated transponder (PIT) tag. One tag was injected subcutaneously with a sterile needle between the shoulder blades. Both ears of the snowshoe hare were also marked using a permanent black marker for short-term identification. After the first day of any trapping session (i.e. on traps days 2-6) every snowshoe hare was scanned with a 125-kHz Miniportable reader to determine whether the hare was a recapture or a new capture. In the event the snowshoe hare was preyed upon and partially ingested, the earmarks were checked. Each snowshoe hare was sexed by turning the hare on its dorsal side and protruding the genitalia. The forefinger and middle finger were used to apply slight pressure to the vent area just above the anus. Snowshoe hares were then released away from handlers. Snowshoe hares that suffered severe trap or predation injuries were euthanized with a 1-ml intrathoracic injection of sodium pentobarbital. Each carcass was necropsied and the liver and kidneys preserved for analysis of metals concentrations. After necropsy and tissue collection, euthanized animals were disposed ofby cremation or deposited in a landfill. Any non-target species caught in traps were immediately released; whole specimens from any mortality of non-target species were donated to the Denver Museum of Nature and Science. Diet In 2003, fecal pellets were collected from the inside of each live-trap where a snowshoe hare was captured and allowed to air dry in kraft brown-paper bags. Fecal pellet samples were randomly selected for diet analyses from 24 individual snowshoe hares: four from each of the three spruce-fir grids and four from each of the three lodgepole pine grids. To reduce the possibility of finding TACO and alfalfa in the diet analyses, only first captures of snowshoe hares were used. Where < 4 snowshoe hares were captured on a trapping grid (e.g. SF 1, LP 2), fecal pellets were collected from fresh snowshoe hare tracks two days after snowfall. Fifteen fecal pellets were required for diet analyses (Bruce Davitt, Washington State University, pers. comm.). If< 15 fecal pellets were collected, a new random sample was chosen. For this reason, one sample (LP 1) was taken from a recaptured snowshoe hare three nights after the initial capture. Fifteen fecal pellets were arbitrarily chosen from each paper bag and transferred to a labeled envelope. Samples were submitted to the Wildlife Habitat and Nutrition Laboratory at Washington State University, Pullman, WA for analysis of diet. �Size of Fecal Pellets We measured snowshoe hare fecal pellets collected in 2002 to 0.1 mm using SPI dial calipers. Fecal pellets were also collected from every mountain cottontail incidentally captured in 2002 and 2003 and measured in the same way. Partial or damaged fecal pellets were eliminated from measurement. We measured the longest diameter for any non-spherical pellets. For snowshoe hares, 32 samples from 23 animals (n = 2374 fecal pellets) were measured. Ten samples from 10 mountain cottontails (n = 655 pellets) were measured. Vegetation Habitat attributes were estimated at two levels: at each trap site and for each trapping grid (Table 2). Within each trapping grid, vegetation was sampled from 15 trap sites, similar to the design of Scott Mills (Figure 2). Methods developed by Tanya Shenk (Colorado Division of Wildlife) to monitor habitat use by reintroduced lynx to Colorado were followed with modification (Figure 3). Accordingly, a 12-m x 12m square of 25 points was placed in 5 rows of 5 (3 m apart), centered over the trap location (Figure 3). The measurements taken at each of the 25 points included: 1. Snow depth (cm), as measured by a calibrated avalanche probe, from the center of each trap location. 2. Understory "hits" measured in a column of 3-cm radius around an avalanche probe. a. All live or dead stems and coarse woody debris (CWD) that fall within the 3-cm radius column using the standardized four-letter genus-species code at 3 height categories (0-0.5 m, 0.51-1.0 m, 1.01 - 1.5 m) above the snow surface. b. Each of the above stems classified in 3 different diameter categories (< 5 mm, 5 .1 - 10 mm, 10.1 - 15 mm) measured at the point where the stem hit the avalanche probe 3. Overstory was measured using a densitometer attached to the avalanche probe. a. Species that hit the crosshairs inside the sighting tube were recorded. Multiple hits by the same species were only recorded once. 4. Every shrub within the plot along with its species and diameter at breast height was recorded (dbh). 5. Every tree within the plot along with its species and dbh was recorded. 6. Every snag within the plot along with its dbh was recorded. 7. Every sapling within the plot along with its species was counted. 8. All coarse woody debris (CWD) deemed usable by snowshoe hare for cover or food (i.e. available above the snow) was recorded along with its diameter. At all of the 84 trap sites within the trapping grid, including the 15 trap sites sampled as described above, the following data were measured: I. Snow depth (cm), as measured by a calibrated avalanche probe. 2. Species of, dbh of, and distance to the closest woody stem in two categories:~ 1.0 cm - 7.0 and~ 7. I cm at the snow surface. 3. Canopy cover for the center of the trap site, as estimated by the use of a spherical densiometer, in the four cardinal quadrants (NW, NE, SE, SW). The following rules were used for unusual events: I. If a point in a vegetation plot lay within a tree bole, the tree species and the dbh was written on the data form. �11 2. A snag was defined as any dead tree bole >45° from the horizontal. Dead boles <45° vertical angle were considered CWD. 3. The mid-point diameter was measured of exposed CWD partially covered by snow. 4. If a leaning tree fell partially outside the 12 m x 12 m sampling plot it was included if>50% of the tree lay within the sampling plot. Temperature and Moon Phase We used daily minimum temperatures recorded by the National Weather Service in Gunnison, Colorado in 2002 and 2003 for each night of trapping. This temperature was intended to represent general weather in the region rather than exact conditions at each trapping grid. We estimated the amount of moonlight for each night of trapping as the percentage of the moon's surface illuminated (Astronomical Applications Department, U.S. Naval Observatory, unpublished data). STATISTICAL ANALYSIS AND PRELIMINARY RESULTS We initially examined our preliminary data for distributional properties and homoscedasticity using SPSS 11. 0. These properties are not important in predictors used in binary logistic regression, but are important in comparisons of means. Where we found substantive violations of assumptions regarding distributional properties, we used the appropriate non-parametric test. Our basic study design involved three stand types as represented by tree species dominants (spruce-fir, lodgepole pine, ponderosa pine). Other predictor variables (e.g. elevation, air temperature, habitat attributes) were highly co-linear with tree stand type and each other (Table 1). Trapping grids in spruce-fir tended to be at higher elevations, and have deeper snow and lower air temperatures (Table 1). Because air temperature (Paul Griffin, University of Montana, pers. comm.) and moon phase (Gilbert and Boutin 1991) have been reported to affect captures of snowshoe hares, we also explored these possible relationships and their relationship to other predictors. We first used binary logistic regression to identify factors measured at the scale of the trapping grid (grid-night= unit of replication) that predict capture success. We included stand type (to include the covariates, elevation and snow depth), percent moon phase, temperature and year as candidate predictors of capture success. In this preliminary analysis stand type and temperature were significant in predicting capture success (Table 3). We then tested how air temperature in Gunnison was related with capture success in spruce-fir and lodgepole pine stands. Although there was no confounding variation with air temperature and stand type (ANOVAF= 98.8, d.f = 2, P = 0.19; spruce-fir = -14°C, lodgepole pine = -I3°C, ponderosa pine = - l l 0 C) we chose to exclude ponderosa pine from this analysis because no hares were captured on the ponderosa pine trapping grids. The relationship between air temperature and captures was significant (t = -3.9, d.f = 45, P < 0.001), with grid-nights for which captures were recorded having mean minimum temperatures of -11 °C, and those for which no captures were recorded having temperatures of -l 8°C. We also examined patterns of captures of snowshoe hares within trapping grids using the response variable of whether a trap location recorded a snowshoe hare capture during either 2002 or 2003. We examined patterns of independence of trap locations within a trapping grid by examining the distribution of trap locations where snowshoe hares were captured, versus those where they were not (Figure 4). We observed no obvious pattern of clumping of successful trap locations, and therefore assumed independence of individual traps as sampling units. When we used the grid-night as the unit of replication (n = 108), and included ponderosa pine trapping grids, trapping success did not differ between years (t = 1.57, df = 106, P = 0.12). However, when trap-night was used as the unit of replication (n = 1512), trapping success did differ between years (t = -3.14, df = 1395, P = 0.002), with more captures in 2003 than 2002. x x x �12 We used binary logistic regression to identify vegetation attributes that predicted capture success for snowshoe hares in an individual trap (trap-night= unit of replication). In this preliminary analysis we found that canopy cover was a significant predictor of capture success at trap locations (Table 5) with successful trap locations ( x = 84% cover) having canopy cover 40% greater than that for unsuccessful trap locations ( x = 60%, Mann-Whitney U = 14086, P < 0.001). The other significant predictor was distance to the nearest woody stem 1-7 cm in diameter, with successful trap locations ( x = 2.0 m) having nearest stems only 56% as far away as unsuccessful trap locations ( x = 3.6 m, M-W U = 21897, P < 0.001). We measured the mean sizes of fecal pellets of snowshoe hares (.x = 8.4 mm) and mountain cottontails ( x = 7 .2 mm) from known species origin and found the means differed (Mann-Whitney U = 26.5, P = 0.001) and 95% confidence intervals did not overlap (Figure 5). DISCUSSION The results in this progress report are preliminary and subject to revision based upon continuing analyses of data. Still, some patterns in the data are apparent. Temperature appeared to have an effect on capture success whereas moon phase, although it has been reported to have an effect, did not. Our preliminary analysis of vegetation data suggests that canopy cover and distance to the nearest 1-7 cm stem also affect capture success. A resource selection model will be generated in the next phase to determine habitat predictors of capture success. Our comparison of diameters of fecal pellets of snowshoe hares and mountain cottontails suggests a difference in size of pellets between the two sympatric lagomorphs that should be useful for identification of pellets to species in the field. �13 LITERATURE CITED Bartmann, R. M. and G. Byrne. 2001. Analysis and critique of the 1998 snowshoe hare pellet survey. Colorado Division of Wildlife, unpublished report no. 20. Bider, J. R 1961. An ecological study of the hare Lepus americanus. Canadian Journal of Zoology 39:81-103. Buttery, R. F. and B. C. Gillam. 1987. Managing forested lands for wildlife. Pages 43-71. Colorado Division of Wildlife, Denver, Colorado. Dolbeer, R. A. and W.R. Clark. 1975. Population ecology of snowshoe hares in the central Rocky Mountains. Journal of Wildlife Management 39:535-549. Gilbert, B. S. and S. Boutin. 1991. Effect of moonlight on winter activity of snowshoe hares. Arctic and Alpine Research. 23:61-65. Grigal, D. F. and N. R Moody. 1980. Estimation of browse by size classes for snowshoe hare. Journal of Wildlife Management 44:34-40. Hodges, K. E. 2000a. The ecology of snowshoe hares in northern boreal forests. Pages 117-161 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, and J. R. Squires, editors. Ecology and conservation oflynx in the United States. University Press of Coloi,:ado, Boulder, Colorado. Hodges, K. E. 2000b. Ecology of snowshoe hares in southern boreal and montane forests. Pages 163206 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, and J. R. Squires, editors. Ecology and conservation of lynx in the United States. University Press of Colorado, Boulder, Colorado. Hoover, R L. and D. L. Wills. 1987. Managing forested lands for wildlife. Pages 455-477. Colorado Division of Wildlife, Eastwood Printing and Publishing, Denver, Colorado. Koehler, G. M. 1990a. Population and habitat characteristics oflynx and snowshoe hares in north-central Washington. Canadian Journal of Zoology 68:845-851. Koehler, G. M. 1990b. Snowshoe hare, Lepus americanus, use of forest successional stages and population changes during 1985-1989 in north-central Washington. Canadian Field-Naturalist 105:291-293. Lemmon, P. E. 1957. A new instrument for measuring forest overstory density. Journal of Forestry 55:667-668. U.S. Fish and Wildlife Service. 2000. Determination of threatened status for the contiguous U.S. distinct population segment of the Canada lynx and related rule; final rule. U.S. Federal Register 65: 1605116086. Wolff, J. 0. 1980. The role of habitat patchiness in the population dynamics of snowshoe hares. Ecological Monographs 50: 111-130. �Table I. Abiotic characteristics of nine trapping grids in three stand types, southwestern Colorado, late winter 2002 and 2003. Snow depth (cm) is the mean (SE) measured at 84 trap locations at each trapping grid. Temperature (0 C) (SE) is the mean low temperature recorded in Gunnison for each grid-night. The aspect of each trapping grid is shown in degrees in their respective order. Trapping grids Stand Type Snow depth Elevation Temperature Aspect PPl,PP2,PP3 Pinus ponderosa 2 (0.4) 2600 -11 (I) 50°, 130°, 130° LP 1, LP 2, LP 3 Pinus contorta 45 (1) 3000 -13 (I) 230°, 90°, 130° SF 1, SF 2, SF 3 Picea engelmanii, Abies lasiocarpa 74 (I) 3400 -14 (I) 310°, 150°, 110° Table 2. Vegetation characteristics for nine trapping grids (see Table I) in three stand types (n = 3 each), southwestern Colorado. Mean tree density, mean sapling density, and mean snag density (number ha- 1) (SE) were counted at 15 trap locations on each grid, late winter 2002. Mean canopy cover(%) (SE) was measured at 84 trap locations on each grid using a densiometer, late winter 2002. The median horizontal cover(%) was measured at 15 trap locations on each grid using a horizontal profile board, late winter 2003. Stand Type Tree density Sapling density Snag density Canopy cover Horizontal cover Pinus ponderosa 187 (29) 301 (81) 273 (65) 36 (2) 0 Pinus contorta 1268 (138) 554 (I 09) 443 (67) 73 (1) Picea engelmanii, Abies lasiocarpa 1418 (116) 642 (107) 287 (44) 79 (1) 65 �Table 3. Preliminary results of binary logistic regression using stand type (excluding ponderosa pine) and abiotic factors as variables to predict capture success (n = 72). Variables are described fully in the methods section. 95% C.I. Coefficient z Temperature 0.169 3.3 Stand type 1.794 2.7 Year -0.210 Moonlight Constant Variable p Odds Ratio Lower Upper 0.001 1.184 1.071 1.309 1 0.006 6.012 NA -0.3 1 0.771 0.811 NA -0.003 -0.2 1 0.808 0.997 -1.146 -0.6 1 0.541 0.318 d.f 0.997 0.808 NA Table 4. Preliminary results of binary logistic regression using vegetation characteristics to predict capture success at trapping locations within trapping grids (n= 108). Variables are described fully in the methods section. 95% C.I. Coefficient -z- d.f p Odds Ratio Lower Upper Canopy cover 0.061 6.10 1 < 0.001 1.063 1.043 1.084 Diameter 1-7 cm -0.005 0.08 0.942 0.995 0.880 l.126 Distance 1-7 cm -0.002 -2.00 0.002 0.998 0.997 0.999 Diameter >7 cm -0.014 -1.17 0.235 0.986 0.963 l.009 Distance >7 cm 0.001 1.00 0.148 1.001 0.999 1.003 Constant -6.058 -6.44 < 0.001 0.002 Variable NA Vl �SF 3 ' LP3 f SF1 Taylor Park Reservoir PP 1-- ', Rio Grande R. PP 2 • r - -- - - - - - - - -- - PP31mil! \~.,., . ------=:~~ <1.~'t ., ., i•., A ~ ~Ol,.. s.fQI" " South Fork 1 ~ I Paved highways 1-- -- - - -I Rivers and streams I I 0 I km N * I I 10 Figure 1. Location of nine trapping grids ( B), southwestern Colorado, 2002 - 2003. SF is spruce-fir, LP is lodgepole pine, and PP is ponderosa pine. Trapping grids are not to scale. �17 1# 2# 3# 4# 5# 6# 7# 8# 9* 10# 11* 12# 13* 14# 15# · 16# 17# 18# 19# 20# 21# 22# 23* 24# 25* 26# 27* 28# 29# 30# 31# 32# 33# 34# 35# 36# 37* 38# 39* 40# 41* 42# 43# 44# 45# 46# 47# 48# 49# 50# 51* 52# 53* 54# 55* 56# 57# 58# 59# 60# 61# 62# 63# 64# 65* 66# 67* 68# 69* 70# 71# 72# 73# 74# 75# 76# 77# 78# 79# 80# 81# 82# 83# 84# Figure 2. Schematic of 300 m x 550 m trapping grid for estimating population density of snowshoe hares in southern Colorado. Asterisks("') indicate the location of the 15 vegetation plots centered on trapping points. Pound signs (#) indicate where the point-quarter method will be used on all other trap locations. �18 11111111111111111:1:111 111111111111 111111111 1 ■ ■ ■ ■ ■ ■ ■ 1 111111111111:1 1::1:1111 1 11 11111:11~1111111:11111 111 1 1 1 1 111 11~1 ~111111111111 trap 11111111:11:~1111111:111 ■ 1 11111111111~11111111111 1 1 1111:1111 :~111111111111 l:lliiii:li~:illllJIIIII ■ ■ 1 11111:1::11~:111 11:1:11 ■ 12 m Figure 3. Schematic of 12 m x 12 m vegetation plot centered on each of the 15 trap sites (Figure 2) used in measuring habitat variables for snowshoe hares in southwestern Colorado, late winter 2002. The trap location is at the center of the vegetation plot. �19 SF 1 •0 00 •• •0 •0 00 • • 0 0 0 0 0 • • 0 0 0 0 0 0 • 0 0 0 0 • • • 0 0 0 0 • • • 0 0 0 0 • • • 0 0 0 0 •0 • 0 0 0 •0 •0 0 0 0 0 0 • • 0 0 0 • • 0 0 0 • • •• •• 0 0 0 0 0 0 • 0 0 0 0 0 0 0 0 0 0 0 0 •0 0 0 0 0 0 0 0 •0 00 00 00 00 00 00 LP 1 •0 00 00 00 00 00 00 0 0 0 •0 •0 0 0 n= 33 SF 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 • 0 0 0 0 0 0 0 0 0 0 •0 • 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 • 0 0 0 0 0 •0 ••• 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 n = 11 0 0 0 0 0 0 0 0 0 0 0 0 0 0 n=4 • LP 2 n=4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 •0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 •0 •0 •0 0 0 0 0 0 0 SF 3 • 00 00 • 00 • • • 0 0 •0 0 •0 • •0 • 0 0 0 •0 •0 •0 0 • •• 0 0 0 0 0 0 0 • 00 00 0 00 •0 •0 •0 0 •0 0 • 0 0 •0 0 • •0 0 0 0 •0 0• • 0 0 0 0 0 • • LP 3 0 • 0 0 0 0 0 0 0 •0 •0 •0 00 •0 00 0 n= 30 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 •0 ••0 • n=9 Figure 4. 12 x 7 trapping grid schematic representing snowshoe hare trap successes(•) for 6 trapping grids in two stand types, spruce-fir and lodgepole pine (snowshoe hares were absent from all pondersosa pine trapping grids), southwestern Colorado, late winter 2002 and 2003. �20 10 9 8 - E E 'Q)"' ..... Q) E ~ ~ 7 6 5 4 3 2 o....__-------~--------.---------' L. americanus S. nuttallii Figure 5. Mean diameters of fecal pellets of Lepus americanus (n = 23 animals) and Sylvilagus nuttallii (n = 10 animals). Error bars indicate 95% confidence interval. �Colorado Division of Wildlife Wildlife Research Report July 2003 – June 2004 JOB PROGRESS REPORT State of Project No. Colorado : : Cost Center 3430 Mammals Research Work Package No. Task No. 0670 2 : Lynx Conservation : Ecology of Snowshoe Hares (Lepus americanus) in Colorado Federal Aid Project: N/A : Period Covered: July 1, 2003 – June 30, 2004 Author: Jennifer L. Zahratka Personnel: S. W. Buskirk , T. M. Shenk All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT A thesis, entitled ‘The population and habitat ecology of snowshoe hares (Lepus americanus) in the southern Rocky Mountains’ was completed and submitted to the University of Wyoming in partial fulfillment of a Master of Science degree. The thesis is available from The Colorado Division of Wildlife Library or the University of Wyoming Library. Included in this report is an abstract of the thesis. 15 �JOB PROGRESS REPORT THE POPULATION AND HABITAT ECOLOGY OF SNOWSHOE HARES (Lepus americanus) IN THE SOUTHERN ROCKY MOUNTAINS Jennifer L. Zahratka ABSTRACT To better understand the population ecology and habitat associations of snowshoe hares (Lepus americanus), I studied snowshoe hares in southwestern Colorado in winters 2002 and 2003. I estimated densities from mark-recapture data and compared vegetative attributes in the mature structural stage (SS 4) among three stand types: Engelmann spruce (Picea engelmannii)–subalpine fir (Abies lasiocarpa), lodgepole pine (Pinus contorta), and ponderosa pine (Pinus ponderosa). I used three methods to calculate a boundary strip width (W) to estimate the effective area trapped Â(Ŵ) in order to illustrate the effect of different methods of estimating Â(Ŵ) on estimates of density. Density estimates [ D̂ = N̂ / Â(Ŵ)] in mature spruce-fir ranged from 0.1 ± 0.03 (SE) hares/ha to 0.9 ± 0.1 hares/ha in 2002 and 0.3 ± 0.05 to 1.0 ± 0.1 hares/ha in 2003. I report only minimum number alive (MNA) in lodgepole pine due to too few captures to estimate density. No snowshoe hares were captured in mature ponderosa pine stands. Model selection based upon the corrected Akaike’s Information Criterion (AICc) showed a strong relationship between MNA and understory cover, density of woody stems 1-7 cm in diameter, and the availability of suitable woody stems for food among the mature stand types I studied (R = 0.91, df = 8, P = 0.008). My empirical data support the assumption that snowshoe hares select habitat with protection from predation. However, the availability of suitable woody stems for food is also an important vegetative attribute for hare habitat. Snowshoe hares selected for spruce-fir among the mature stand types I studied. Mature spruce-fir provided more understory, greater density of woody stems 1-7 cm in diameter, and more woody stems (<1.5 cm) for food. In my study, the winter diet of snowshoe hares was overwhelmingly gymnosperms. Extremely low temperatures affected capture success, but moon phase did not. Counts of fecal pellets are an attractive tool to estimate densities of snowshoe hares because they are less costly and less labor-intensive than conventional mark-recapture techniques. In the southern Rocky Mountains, snowshoe hares and mountain cottontails (Sylvilagus nuttallii) are syntopic. Indeed, I captured two mountain cottontails in two traps in which I captured snowshoe hares. Therefore, distinguishing between fecal pellets is necessary for making inferences specific to these species. Methods to distinguish between the two leporid species have been developed based upon the assumption that the larger snowshoe hare produces larger fecal pellets than the smaller mountain cottontail. In this study, I measured 655 fecal pellets from 10 individual mountain cottontails and 2,374 fecal pellets from 23 individual snowshoe hares: I found no apparent relationship between the body weight of mountain cottontails or snowshoe hares and the size of their fecal pellets (mountain cottontails: r = 0.04, F = 0.01, P = 0.91; snowshoe hares: r = 0.48, F = 9.3, P = 0.005). Although the two species differed in the size of their fecal pellets, the difference (1.2 mm) would be indistinguishable without measuring equipment and is only applicable to adults. While fecal pellet counts may be accurately used to estimate densities of snowshoe hares in the boreal forests of Canada, in the southern Rocky Mountains where leporid species are potentially syntopic, this method may yield misleading results. 16 � https://cpw.cvlcollections.org/files/original/78a2e6cd6a7e6aff02962f3b526d9cfd.pdf 2981cc3bb9387fd35c3e0cb87d77d25d PDF Text Text 115 Colorado Division of Wildlife Wildlife Research Report July 2000 JOB PROGRESS REPORT State of ----~~=~-----Colorado Project No. W-153-R-13 Mammals Research Work Package No. ___ 0_88_0_ _ _ _ _ __ Task No. 1 Black-footed Ferret Recovery Monitoring and Managing Disease in Black-footed Ferrets Period Covered: July. 1, 1999 - June 30, 2000. Authors: M.A. Wild and K. T. Castle Personnel: E. Wheeler, E. Schmal, and S. Kasven. \ ) ABSTRACT \ ) The black-footed ferret is a federally listed endangered species in the United States. Black-footed ferrets have been extirpated from Colorado, but were scheduled to be reintroduced to Moffat County, Colorado in 1999. However, due to high plague activity and low prairie dog densities at the Little Snake Management Area (LSMA), black-footed ferret reintroduction was postponed indefinitely at the site. The secondary site at Coyote Basin, Utah was readied and reintroduction of ferrets was performed there in 1999. The Wolf Creek site in Moffat County, Colorado is also being readied for reintroduction of ferrets, likely in 2001. Our work in support of black-footed ferret reintroduction can be sub-divided into three broad sections: disease monitoring in the proposed release areas in Colorado, care of black-footed ferrets, and flea control as a tool to manage sylvatic plague in prairie dogs and black-footed ferrets. Disease monitoring was performed using collection of coyotes from LSMA in July 1999 and from the Wolf Creek site in February 2000. Two potentially devastating diseases, canine distemper and plague, are present at LSMA. Although prevalence of positive titers to canine distemper virus (CDV) in coyotes ·have been relatively low over the last 3 years (:S33%), the prevalence of positive titers to plague (Yersina pestis) have been high (up to 89% positive). Titers were present in both adult and juvenile animals, suggesting ongoing plague activity in at least some sections of the management area. Samples collected from coyotes at the Wolf Creek site indicated substantially lower disease activity than at LSMA. Prevalence of positive titers to plague and to CDV was 7% of samples collected in February 2000. In general, captive black-footed ferrets maintained at the LSMA breeding and preconditioning pens remained healthy. One ferret died and another was presumed dead in the burrow system after a severe hailstorm. Of 13 kits born at the pens in spring 1999, 12 survived to weaning in fall 1999. These 12 kits in addition to 50 other black-footed ferrets were released at the Coyote Basin site in fall 1999. Twenty ferrets were maintained in the pens overwinter and produced 11 kits in spring 2000. We summarized research results and presented a paper titled "Dose titration and safety of lufenuron fed to captive white-tailed prairie dogs (Cynomys leucurus)" at the 2000 meeting of the �116 Wildlife Disease Association, Jackson, Wyoming. We also performed a trial to test' the efficacy of lufenuron in controlling fleas on captive prairie dogs. One week post-dosing, fleas that fed on prairie dog treated with 500 mg/kg lufenuron prcxiuced a lower proportion (p < 0.05) of viable rggs than those fed on control prairie dogs (mean 0.16 vs. 0.39, respectively). However, the proportion of viable eggs prcxiuced during week 2 post-dosing was low for each group (0.24 treatment vs. 0.17 control). By week 4 postdosing, egg prcxiuction had fallen to nearly zero in each group; however, the three eggs produced by fleas fed on treated prairie dogs were all viable. Although further, similarly controlled investigation is warranted, preliminary results suggest that given current techniques, the likelihood of lufenuron limiting fleas populations and breaking the plague cycle is not extremely promising. �117 MONITORING AND MANAGING DISEASE IN BLACK-FOOTED FERRETS Margaret A. Wild and Kevin T. Castle P. N. OBJECTIVES 1. Monitor disease activity threatening survival of black-footed ferrets reintroduced into the Little Snake Management Area (LSMA). 2. Develop techniques to manage plague in the LSMA using insect growth regulators applied orally to prairie dogs. SEGMENT OBJECTIVES 1. Provide ;ete~ care to captive and reintroduced black-footed ferrets. 2. Monitor and manage plague activity in LSMA. METHODS AND MATERIALS Carnivore Disease Survey Infectious diseases can severely impact the success of black-footed ferret (}Juste/a nigripes) reintroduction efforts. As part of the black-footed ferret reintroduction protocol, we monitored disease activity in carnivores at proposed ferret reintroduction sites: in July 1999 at the Little Snake Management Area (LSMA), Colorado and in February 2000 at the Wolf Creek Management Area (WCMA), Colorado. Coyotes (Canis latrans) were collected for post-mortem examination and samples collected as described in the Program Narrative (Wild and Castle 1998). Black-footed Ferret Reintroduction and Veterinary Care I assisted in preparation of the black-footed ferret allocation request submitted to the US Fish and Wildlife Service by the Colorado-Utah black-footed ferret recovery working team in 2000. I provided veterinary care and consultation on health matters for captive black-footed ferrets and black-footed ferret releases. Animal care was performed in accordance with the established protocol (Wild and Castle 1999). Flea Control In Prairie Dogs We completed the experiment "Bioavailability of lufenuron administered orally to captive white-tailed prairie dogs (Cynomys leucurus)" and performed a follow-up experiment "Efficacy of lufenuron for the control of fleas in white-tailed prairie dogs (Cynomy leucurus)". These studies were outlined in Wild and Castle 1998. The detailed study plan for the lufenuron bioavailability study and e:fp.cacy trial are reported in Wild and Castle 1999 and in Attachment 1, respectively. �118 RESULTS AND DISCUSSION Carnivore Disease Survey With the assistance of USDA Wildlife Services and the Bureau of Land Management (BLM) we collected 16 coyotes from the LSMA between 26-30 July 1999 and 14 coyotes from the Wolf Creek site on 8-9 February 2000. Coyotes were collected using a combination of calling and aerial gunning. Further collections were not possible due to weather constraints and aircraft availability. No lesions indicative of active disease were noted on gross examination of carcasses. Of the coyotes collected from the LSMA, 31 % (5/16) had positive titers to plague (Fig. I) using the standard HA/HI test while one additional coyote was positive using the ELISA test. Interestingly, all juveniles sampled this summer (n = 9) were negative to plague. Because the juvenile coyotes have consumed (sampled) prairie dogs from this summer only, lack of exposure may indicate that the prevalence of plague in prairie dogs in the area is declining. If this is the case, titers in adults would likely be from exposure in previous years. Alternatively, the prey base of the coyotes may have shifted to species that are less commonly infected with plague (e.g., rabbits) if the density of prairie dogs has been greatly reduced by plague. Forty-three percent of coyotes sampled (6/14 usable samples) had positive titers to tularemia. In contrast to plague, all positive titers to tularemia were observed in juvenile coyotes. As previously observed, tularemia appears to elicit a serologic response of short duration in coyotes. The impact of tularemia on black-footed ferrets and prairie dogs is unknown and warrants investigation. A titer to canine distemper virus (CDV) was found in only one of 14 serum samples tested (Fig. 2). Samples collected from coyotes at the Wolf Creek site indicate substantially lower disease activity than at LSMA. A 9-year-old male coyote had positive titers to plague and to CDV, but all other coyotes (n = 13) were negative to plague and CDV (7% positive; Fig. 3). The adult male and one additional juvenile coyote also had titers to tularemia (14% positive). Black-footed Ferret Reintroduction and Veterinary Care In general, captive black-footed ferrets maintained at LSMA remained healthy. One adult female was treated for an apparent mite infection and secondary bacterial pyoderma. Unfortunately, samples to confirm this diagnosis could not be collected prior to treatment; however, the ferret responded quickly and completely to therapy with ivermectin and amoxicillin. Additional cases of crusty skin have been successfully treated with ivermectin but confirmation of the etiology has not yet been made. A I-yr-old male was found dead and a 4-yr-old female was missing and presumed dead after a severe hailstorm. Necropsy results revealed death from blunt trauma. The death(s) occurred in the new pen complex where shallow burrow systems may have been flooded forcing ferrets to the surface in the severe weather. To avoid this problem in the future, additional above ground shelter will be provided. Of 13 kits born at the pens in spring 1999, 12 survived to weaning in fall 1999. One kit disappeared and is presumed dead in the burrow system. Based on results of carnivore disease monitoring over the past 3 years and prairie dog inventories performed in summer 1999, LSMA was determined to be currently unsuitable habitat for the release of black-footed ferrets. Prairie dog inventories performed by Utah Division of Wildlife showed insufficient densities of prairie dogs to support black-footed ferret reintroduction. As a result, ferrets were not released into LSMA but instead were released into Coyote Basin, Utah. Continued monitoring will determine when (if) LSMA can support reintroduction of black-footed ferrets. A secondary site in Colorado, (Wolf Creek) is also being readied for reintroduction of black-footed ferrets in fall 2000 or 2001. _ _, �119 The 12 kits produced onsite, in addition to three 1-yr-old males, and five 3-yr-old females from the LSMA pens were released at Coyote Basin in November 1999. Additionally, 19 kits and five 3-yr-old females from other captive breeding sites were pre-conditioned at the LSMA prior to release at Coyote Basin. The reintroduction was further supplemented with 28 ferrets released immediately upon arrival from other captive breeding sites without pre-conditioning at LSMA. Prior to release, ferrets were trapped for routine examination, treatment, and identification (Wild and Castle 1999) and a health certificate was issued for each individual. All appeared healthy except for the presence of ectoparasites (ticks, mites, fleas). Individual ferrets were treated with ivermectin and pen dusting was advised. In an attempt to meet our ideal age and sex structure of captive black-footed ferrets (Wild and Castle 1999), we retained seven ferrets and supplemented the population with 13 additional ferrets from other captive breeding facilities. Seven black-footed ferrets (five 1-yr-old females and two males) were retained in captivity at the LSMA pens. In November 1999, six adult females,'three female kits, and four male kits were added to this breeding group bringing the total number of ferrets at the LSMA pens to 20. Ferrets were maintained under the standard care protocol (Wild and Castle 1999). Females were paired with males in spring 2000, and four litters resulted. One litter was apparently consumed by the female when about 1 day of age. The other three litters yielded 11 kits. Flea Control In Prairie Dogs We presented results of the lufenuron dose titration study at the 2000 meeting of the Wildlife Disease Association, Jackson, Wyoming. The abstract of that presentation read: DOSE-TITRATION AND SAFETY OF LUFENURON FED TO CAPTIVE WHITE-TAILED PRAIRIE IX>GS (CYNOMYS LEUCURUS) KEVIN T. CASTLE Colorado Division of Wildlife, 317 W. Prospect St. Fort Collins, CO, 80521, MARGARET A. WILD, Colorado Division of Wildlife, 317 W. Prospect St. Fort Collins, CO, 80521, and S. CRAIG PARKS, Novartis Animal Health, P.O. Box 26402, Greensboro, NC 27404. Plague is a zoonotic disease that impacts populations of prairie dogs (Cynomys spp.) and other species, such as black-footed ferrets, that rely on them for food and shelter. Yerstnia pestis, the etiological agent of plague, is transmitted primarily by the bite of an infective flea. Recently developed compounds used to control fleas in pet animals offer a promising alternative to insecticide dusts for the control of fleas in wild rodents. Lufenuron is a lipid-soluble insect growth regulator with ovicidal and larvicidal activity. Lufenuron is efficacious for controlling fleas in cats and dogs at blood concentrations above 50-100 parts per billion (ppb). A single oral dose oflufenuron has been shown to be effective in controlling the cat flea (Ctenocephaltdes felts felts) for at least 30 days in treated cats and dogs. To date, there have been no studies conducted to detennine the duration oflufenuron blood concentrations in any prairie dog species. We compared lufenuron blood concentrations in white-tailed prairie dogs (C. leucurus) during periods of activity (nontorpid group) and hibernation (torpid group) during January-March 1999. We hypothesized that if high serum concentrations of lufenuron could be maintained over winter during hibernation or for > 1 mo in active prairie dogs, the compound may be effective for use in breaking the plague cycle. Thirty captive WTPD were fed 300 rrig/kg lufenuron; half the animals were allowed to become torpid, while the other half were kept awake. All animals remained healthy throughout the 9 week study period. Prairie dogs in the active group gained weight, while those in the torpid group lost weight over the 9 weeks. Blood was drawn from each animal prior to dosing, one week after dosing, then every other week until week 9 post-dosing. Serum was harvested and tested by HPLC for lufenuron concentration. Blood lufenuron concentration did not differ between the groups one week post-dosing. Concentration in both groups decreased over time, but the �120 concentration in torpid animals declined at a more gradual rate; after weeks 3, 5, and 7, lufenuron levels in torpid WfPD were significantly higher than levels in nontorpid WfPD. After nine weeks, blood levels were again similar, and had approached the limit of detection (10 ppb). Blood levels in nontorpid WfPD declined to <50 ppb after 3 weeks, while levels in torpid WfPD declined to <50 ppb after 7 weeks. Future studies will be required to determine efficacy of lufenuron in controlling fleas on WfPD. If effective blood concentrations are similar to dogs and cats, however, frequent dosing would be required to control flea numbers on prairie dogs and thus break the plague cycle. Results from this experiment indicated that blood concentrations of lufenuron did not reach initial levels as high as anticipated, nor were the concentrations maintained above 50 ppb for as long as anticipated (Fig. 4). The most likely cause of these low concentrations was poor absorption of the drug by prairie dogs. lbis may be due to the difference in gut morphology between rodents and carnivores. Alternatively, other aspects of pharmacokenetics may have been responsible for the low serum levels in prairie dogs despite dosing at rates 10-30 times higher than those recommended for cats and dogs, respectively. We followed up the dose titration experiment with an experiment to test the efficacy of oral lufenuron to control fleas on captive prairie dogs. Based on data from the bioavailability study, we increased the lufenuron dose to 500 mg/kg. Serum samples from the study have been submitted for lufenuron assay. Results are pending. Interpretation of efficacy results will rely on these serum lufenuron levels; however, we were able to make some preliminary comparisons between performance of fleas fed on treatment and control prairie dogs. One week post-dosing, fleas that fed on treated prairie dog produced a lower proportion (p < 0.05) of viable eggs than those fed on control prairie dogs (mean 0.16 vs. 039, respectively). Unfortunately, the proportion of viable eggs produced during week 2 post-dosing was low for each group (0.24 treatment vs. 0.17 control). By week 4 post-dosing, egg production had fallen to nearly zero in each group; however, the three eggs produced by fleas fed on treated prairie dogs were all viable. We are uncertain of the cause of the dramatic reduction in egg production and viability observed during the course of the experiment. Anecdotal reports suggest that flea production may be influenced seasonally (or at least cyclically) despite attempts to maintain controlled environmental conditions in the insectary (Metzger, Pers. Comm.). Regardless, it is unfortunate that serum levels oflufenuron decreased more rapidly than we had expected and that data did not support the hypothesis that lufenuron would be an effective means to significantly reduce flea production over the summer. Although further, similarly controlled investigation is warranted, preliminary results suggest that given current techniques, the likelihood of lufenuron limiting fleas populations and breaking the plague cycle is not extremely promising. Therefore, experiments into efficacy of controlling flea infestations in simulated burrow environments and in the field (described in Wild and Castle 1998) will not be performed. LITERATURE CITED Wild, M.A. and K. T. Castle. 1998. Monitoring and managing disease in black-footed ferrets. Colorado Div. Wildl. Res. Rep., 0880-1, Jul 1997 - Jun 1998, Fort Collins. Wild, M.A. and K. T. Castle. 1999. Monitoring and managing disease in black-footed ferrets. Colorado Div. Wildl. Res. Rep., 0880-1, Jul 1998 - Jun 1999, Fort Collins. �121 Attachment l Kevin T. Castle, Margaret A. Wild, and S. Craig Parks Colorado Division of Wildlife Foothills Wildlife Research Facility (KTC and MAW) Novartis Animal Health, Greensboro, NC (SCP) Introduction Black-footed ferret (Mustela nigripes) recovery plans call for the reintroduction of ferrets to sites characterized by the presence of viable populations of prairie dogs (Cynomys spp.), which provide food and shelter for ferrets. Unfortunately, prairie dogs inhabiting many potential reintroduction sites carry fleas that can serve as vectors ofYersinia pestis, the causative agent of plague (Ubico et al, 1988). Prairie dogs and ferrets are both highly susceptible to plague (Barnes, 1993; Williams et al., 1994) so the chances of successful ferret reintroduction will be enhanced if the numbers of fleas infesting a prairie dog colony can be significantly reduced. Lufenuron is a benzoylphenylurea derivative which inhibits formation of chitin in the exoskeleton of insects (Cohen, 1987). A single oral dose oflufenuron has been shown effective in controlling the cat flea (Ctenocephalides felts felis) for at least 30 days in treated cats (Blagburn et al., 1994) and dogs (Hink et al., 1994; Blagburn et al., 1995). No studies on the efficacy oflufenuron have been conducted with prairie dogs; however, Davis ( 1997) reported a significant reduction in fleas on free-ranging ground squirrels (Spermophilus beecheyi) that had been treated with lufenuron. We are performing a series of experiments to test the applicability oflufenuron to control fleas in captive, and ultimately wild, white-tailed prairie dogs (Cynomys leucurus). Thus far in pilot studies we have determined standard husbandry, maintenance, and handling protocols for captive prairie dogs and determined a test dose oflufenuron based upon a bioavailability study. We have also attempted to establish an insectary colony of a flea species (0ropsylla tuberculata) that naturally infests prairie dogs and their burrows. Oropsylla fleas are among the most common prairie dog fleas, and have been implicated in the transmission of plague in prairie dogs (Ubico et al., 1988). Fleas of this genus are nest fleas that infest the host only to obtain a blood meal. At other times the fleas select microenvironments within in the burrow system that are conducive to successful reproduction and survival. In pilot studies we were unable to develop methods to successfully maintain and produce self-sustaining populations of 0. tuberculata in an insectary. However, a related species, 0. montana, which naturally infests ground squirrels (e.g. Spermophilus beechyi) has been successfully maintained in the insectary using methods of M. Metzger and K. Gage (pers. comm). These fleas will feed on prairie dogs and successfully reproduce after ingesting a blood meal. Because of its similarity to 0. tuberculata, we will use 0. montana as a model to determine if flea reproduction can be controlled in lufenuron-treated prairie dogs. A chambered-flea system has been used to test on-host viability and fecundity of fleas on cats (Thomas et al. 1996) and laboratory mice (D. Engelthaller, pers. comm.). In this system, fleas are contained in a chamber attached to the host, and can obtain a blood meal through a mesh screen. This system has several advantages over other methods of artificial infestation. First, because a known number of adults can be placed into the chamber, flea mortality is easily noted, and survivorship can be determined. Second, sex ratios can be adjusted to maximize egg production. Third, eggs can be recovered readily, to determine viability. Fourth, the environment in the chamber will be warm and humid, and therefore conducive to adult and egg survival. Finally, fleas will not be able to leave the prairie dog, and therefore will not be able to infest caretakers or other animals. �122 In this study we will: 1) develop techniques for artificially infesting white-tailed prairie dogs with chambered populations of fleas, 2) test the flea control efficacy oflufenuron fed to white-tailed prairie dogs artificially infested with fleas, and 3) compare efficacy of flea control with serum lufenuron levels. Our working hypothesis is that fleas which feed on lufenuron-dosed prairie dogs will have reduced survival of eggs and larvae compared to fleas which feed on control animals that receive no lufenuron. Methods Prairie Dog Maintenance We will use 31 captive white-tailed prairie dogs maintained at the Foothills Wildlife Research Facility in Ft. Collins in our experiment. Prairie dogs will be housed singly (if used in experiments) or in pairs (if used for flea colony maintenance) in custom-designed cages (100 cm x 500 cm x 600 cm; Wild and Castle 1998). Prairie dogs will be observed daily, and will have ad libitum access to Teklad Rodent Blocks and water. Windows will provide a natural photoperiod, and a combination of timed heaters and air conditioners will be used to keep ambient temperature between 10 and 25° C. Twenty prairie dogs (10 males and 10 females) will be blocked by sex and randomly divided equally into a treatment group and a control group for the experiment. The treatment group will receive lufenuron while the control group will not. Because all animals cannot be housed in one building due to space limitations, the groups will be split evenly between two separate but similar buildings, to help control for any interbuilding differences (e.g. temperature, humidity) that may exist. The sample size of20 prairie dogs will allow us to detect a ~91 % reduction in the number of eggs that successfully hatch in the treated group given alpha = 0.10 and beta = 0 .90. An additional 11 captive prairie dogs will be maintained under similar conditions, but will not participate in the experiment. Instead, they will be used in the maintenance of the flea colony (see below). Lufenuron Dosing Prairie dogs in the experimental group will be fasted for 24 h prior to dosing; water will be available during the fast. After the fasting period (day 0), each experimental animal will be weighed, then offered a bolus dose oflufenuron (300 mg/kg body mass). Lufenuron will be mixed thoroughly in approximately 10 g of highly palatable bait (ground rat chow and molasses); control animals will receive bait only. We previously determined that a majority of fasted prairie dogs consume about 90-95% of their dose in less than 12 h. We will observe the progress of bait ingestion in each animal to determine when the dose is ingested. Remaining bait will be removed after 24 h. We will weigh the bait/lufenuron mixture before and after the prairie dogs are dosed, in order to calculate the actual dose ingested. After dosing, normal feeding will resume. Flea Infestation An initial stock of laboratory-reared, disease-free fleas (0. montana) will be obtained from Dr. Kenneth Gage at the Centers for Disease Control and Prevention (CDC) in Ft. Collins. Fleas will be housed in 500 ml glass jars containing larva-rearing media, to allow the population to be self-sustaining . Rearing media consists ofwheast (Red Star Biologicals), dried beef blood (Monfort Biologicals), powdered dog chow, and sand. Fleas will be maintained in an incubator at about 22-23° C and ~70% relative humidity (M. Metzger, pers. comm.) to ensure optimal reproduction. A natural photoperiod will be approximated within the incubator using fluorescent lights and a timer. Adult fleas must ingest a blood meal in order to reproduce. We will use two methods to provide blood meals to fleas held in the insectary for propagation of the colony. The first 1:11ethod will follow an established protocol (Castle and Wild 1998) to provide blood to the adult fleas in each rearing chamber, using neonatal rodents. Briefly, when fleas are in need of a blood meal (1-2 times per week), we will obtain �123 neonatal rodents from a private colony. The neonates will be placed into the insectaries with fleas for up to 24 h; previous work has shown that over 80% of the neonates are alive after 24 h. No food or water will be provided for the neonates, as they are strictly dependent on nursing. After feeding by the fleas, neonates will be euthanized by an overdose of inhalant anesthetic. Once per week, for 7 weeks, we will utilize the 11 non-experimental prairie dogs as blood sources, using the chambered flea technique described below. While the use of neonatal rodents is an efficient, approved method of providing blood to fleas, neonatal rodents are not always available from private colonies. Prairie dogs will therefore serve the dual purposes of providing blood meals when neonates are unavailable, and minimizing the number of neonatal rodents sacrificed. One week prior to study initiation, and on study days 2, 7, 14, 21, 28, 35, and 42 we will place flea chambers on experimental prairie dogs. Fleas feeding on treated prairie dogs will potentially be exposed to lufenuron from this blood meal. W.e will collect 50 adult female and 30 ad~lt male fleas from the insectary and place them"into a chamber (2.5 cm diameter). Each chamber will be-attached to a prairie dog so the fleas can obtain a blood meal. To attach the chambers, prairie dogs will be anesthetized using isoflurane delivered by a vaporizer. Respiration and depth of anesthesia will be monitored. A patch of fur on the dorsal thorax caudal to the shoulders will be shaved. A flea chamber will be placed on the skin, and taped into place using Elastikon and Vet-rap. The prairie dog will then be placed in a 30 cm x 20 cm x 20 cm holding box to recover from anesthesia, and will be monitored while the chamber is attached. Chambers will remain on each prairie dog for 30 min. At the end of the feeding time, the prairie dog will again be anesthetized with isoflurane, and the tape and chamber will be removed. Prairie dogs will be returned to their cages after recovery from anesthesia. Fleas will be observed for evidence of feeding by observation under a 10-20x ,;nicroscope, and blood-filled fleas will be placed in plastic vials and put into our insectary for egg recovery. 0. montana fleas typically lay eggs within 2-3 days of a blood meal (M. Metzger, pers. comm.), so adult dishes will be monitored every day for 7 days to monitor egg production. Eggs will be removed from the adult dish and placed in new vials inside the insectary; they will be observed every day for larval emergence. The total number of eggs produced by fleas from each prairie dog will be recorded, as will the number of larva that emerge each ·day. Unfed adult fleas will be returned to the insectary (prior to lufenuron dosing) or preserved in alcohol (after lufenuron dosing). Blood Collection Concentration of lufenuron in the blood may be an indicator of efficacy of flea control. To determine the relationship between blood lufenuron concentration and egg viability and larval development, we will collect 3 ml of blood from each anesthetized prairie dog prior to chamber attachment on each sampling day. Blood will be collected by jugular venipuncture and placed into a glass vacutainer blood tube without anticoagulant. Alternatively, ifwe are unable to collect an adequate blood sample peripherally, we will collect blood from the vena cava or directly from the heart. Although cardiac puncture is considered to be a safe procedure for collection of large volumes of blood from laboratory rodents (CCAC, 1984), due to the increased risks involved with cardiac puncture, we will use this technique only to obtain critical samples. Serum will be harvested and frozen within 4 h of collection. Serum lufenuron concentration will i>C? determined by HPLC at Novartis Laboratories. Based on our pilot studies and other literature (Blagbum et al. 1994, 1995; Thomas et al. 1996) we do not anticipate any prairie dog mortality due to the flea infestations, lufenuron dosing regime or blood collection, but if a severe allergic reaction or other health problem associated with our procedures occurs, the prairie dog will be removed from the study and provided veterinary care or euthanized with an overdose of inhalant anesthetic or barbiturate. A complete post-mortem examination will be performed on any animal that may die during the experimental period. �124 Data Analysis Efficacy of lufenuron will be determined by comparing developmental success of eggs produced by fleas exposed to treated prairie dogs vs. eggs produced by fleas exposed to control group prairie dogs. These formulas will be used in efficacy determination: Developmental success = number of larvae hatched x 100 number of eggs collected Percentage efficacy=mean developmental success (control) - mean developmental success (treated) x 100 mean developmental success (control) Mean developmental success values for eggs collected from prairie dogs at each sampling period will be analyzed using analysis of covariance, using blood lufenuron concentration at each time step as the covariate. Group responses will be considered significantly different ifp < 0.10. Literature Cited Barnes, A. M. 1993. A review of plague and its relevance to prairie dog populations and the black-footed ferret. Proceedings of the symposium on the management of prairie dog complexes for the reintroduction of the black-footed ferret. J. L. Oldemeyer, D. E. Biggins, B. J. Miller, and R. Crete, eds. 96pp. Blagburn, B. L., J. L. Vaughan, D.S. Lindsay, and G. L. Tebbitt. 1994. Efficacy dosage titration of lufenuron against developmental stages of fleas (Ctenocephalides felis felis) in cats. American Journal of Veterinary Research 55: 98-101. Blagburn, B. L., C. M. Hendrix, J. L. Vaughan, D.S. Lindsay, and S. H. Barnett. 1995. Efficacy of lufenuron against developmental stages of fleas (Ctenocephalides felis felis) in dogs housed in simulated home environments. American Journal of Veterinary Research 56: 464-467. Castle, K. T. and M.A. Wild. 1998. Protocol for the use of neonatal rodents as a blood source for insectary-reared fleas. Colorado Division of Wildlife Animal Care and Use Committee Study Plan. Canadian Council on Animal Care (CCAC). 1984. Guide to the care and use of experimental animals, Vol. 2. Ottawa, Ont., Canada. Cohen, E. 1987. Interference with chitin biosynthesis in insects. In Chitin and benzoylphenyl ureas, series entomologica, vol. 38, J.E. Wright and A. Retnakaran (eds), Dr. W. Junk, Publishers, Boston, pp.3342. Davis, R. M. 1997. Use of an orally administered insect development inhibitor (lufenuron) as a flea control agent in the California ground squirrel, Spermophilus beecheyi. Fourth International Symposium on Ectoparasites of Pets, pp. 31-35. Hilton, D. F. J. 1971. A method for rearing fleas of ground squirrels. Transactions of the Royal Society of Tropical Medicine and Hygiene. 66: 188-189. Hink, W. F., M. Zakson, and S. Barnett. 1994. Evaluation of a single oral dose oflufenuron to control flea infestations in dogs. American Journal of Veterinary Research 55: 822-824. Thomas, R. E., L. Wallenfels, and I. Popeil. 1996. On-host viability and fecundity of Ctenocephalides felis (Siphonaptera: Pulicidae), using a novel chambered flea technique. Journal of Medical Entomology 33: 250-256. Ubico, S. R., G. 0. Maupin, K. A. Fagerstone, and R. G. McLean. 1988. A plague epizootic in the whitetailed prairie dogs (Cynomys leucurus) of Meeteetse, Wyoming. Journal of Wildlife Diseases 24: 399-406. Williams, E. S., K. Mills, D.R. Kwiatkowski, E.T. Thome, and A. Boerger-Fields. 1994. Plague in a black-footed ferret (Mustela nigripes). Journal ofWildlife Diseases 30:581-585. Wild, M.A. and K. T. Castle 1998. Monitoring and managing disease in black-footed ferrets. Colorado Division of Wildlife Program Narrative, Project# W-153-R. \ ·\,,,_) �125 20 ~ - - - - - - - - - - - - - i ■ Plague-Neg>----------~ □ Plague-Pos 15 L. (1) ..0 E 10 ::::, z 5 1997-W 1997-S 1998-W 1998-S 1999-W 1999-S Fig. la. Prevalence of exposure to plague in juvenile coyotes from the Little Snake Management Area, Colorado, from winter 1997 through summer 1999. II Plague-Neg 20 □ Plague-Pas 15 Q) .0 E 10 :::J z 5 0 -+------'----'-~--'----'----'----------'-----.J'---..L.---'-----'-----'-----'---J 1997-W 1997-S 1998-W 1998-S 1999-W 1999-S Fig. lb. Prevalence of exposure to plague in adult coyotes from the Little Snake Management Area, Colorado, from winter 1997 through summer 1999. Data from winter 1997 (age class unknown) provided by M Albee. �126 25 - . - - - - - - i □ CDV positive II CDV negative 20 t> 15 J 10 5 0 1997-W 1997-S 1998-W 1998-S 1999-W 1999-S Fig. 2. Prevalence of exposure to canine distemper virus (CDV) in coyotes from the Little Snake Management Area, Colorado,' from winter i 997 through smnmer 1999. All positive coyotes were adults with the exception of one juvenile in summer 1998. Data from winter 1997 (age class unknown) provided by M. Albee. 20 , - - - - - - - - - - - , Ill Negative f - - - - - - - , □ Positive 15 .BS 10 z= 5 Plague Fig. 3. Prevalence of exposure to plague and canine distemper virus in adult coyotes from the Wolf Creek site, Colorado in winter 2000. 200 C: 0 175 "ia = 8 1-, 150 . _J25 C: .0 8 ~()() e= 75 C: 50 = 25 ca ...:I • . • • •• • 0 -t-----,--------.-------.----,,---,----,----r----r---, 2 3 4 5 6 7 8 9 10 Week Post-Dosing Fig. 4. Mean serum lufenuron concentrations of active ( ♦) and torpid (■) prairie dogs orally dosed with 300 mg/kg lufenuron. �31 JOB PROGRESS REPORT Stateof _ _ _ _ _ _C.c;c..=ol=o=ra=d=o_ _ Work Package No. 0880 T~k _ _ _ _ _ _ _ _~l_ __ Division of Wildlife - Mammals Research Black-footed Ferret Conservation Disease Monitoring & Management Period Covered: July I 2002 through June 30, 2003 Author: L. L. Wolfe and L.A. Baeten Personnel: D. Finley, P. Schnurr, K. Cramer, H. Edwards, E. Knox, C. T. Larsen, N. Mier, M. W. Miller, K. Taurman, E. S. Williams Interim Report - Preliminary Results This work continues, and precise analysis ofdata has yet to be accomplished. Manipulation or interpretation of these data beyond that contained in this report should be labeled as such and is discouraged. ABSTRACT We continued monitoring carnivores at proposed black-footed ferret reintroduction sites for serological evidence of select disease epidemics. Sampling at the Wolf Creek Management Area (WCMA) in August 2003 revealed little evidence of ongoing epidemics that could impede black-footed ferret restoration efforts. Serology data from culled coyotes showed no evidence of active canine distemper or plague epidemics in the WCMA vicinity. In contrast, serologic evidence of exposure to tularemia w~ relatively high (~30%), consistent with previous observations in this and other monitored areas. We will continue this work as part of the ongoing Colorado-Utah black-footed ferret reintroduction protocol. �32 INTRODUCTION As part of the Colorado-Utah black-footed ferret reintroduction protocol, we continued monitoring carnivores at proposed ferret reintroduction sites for serological evidence of select disease epidemics. Originally, we monitored coyote (Canis latrans) populations at two Colorado sites: the Little Snake Management Area (LSMA) and the Wolf Creek Management Area (WCMA), Colorado. Under this program, >200 coyotes have been collected for post-mortem examination and samples collected as described in established protocols since March 1997. Monitoring has been accomplished via cooperative efforts of Colorado Division of Wildlife, USDA Wildlife Services, and Bureau of Land Management (BLM) personnel. To date, no lesions indicative of active infections with any of the select pathogens (Francisellia tularensis, Yersinia pestis, canine distemper virus [CDV]) have been noted on gross examinations of carcasses. However, relatively high proportions (31-89%) of the coyotes collected from the LSMA had positive titers to plague between March 1997 and July 1999. Although the proportion of plague-positive coyotes declined during the sampling period, evidence of continued exposure and perhaps declining prairie dog abundance led to abandonment of surveillance at LSMA after 1999. Monitoring at the WCMA has continued, and black-footed ferrets were reintroduced at this site in 200 I. METHODS Coyotes were collected using a combination of calling and aerial gunning (USDA-API-IlS-Wildlife Services). In light of ambiguity in results from mid-winter sampling attributable to the inability to accurately estimate ages of coyotes in the field, we began focusing on late summer sampling to monitor epidemic trends. Postmortem examination, sampling, and serological methods were as described previously (Colorado Division of Wildlife, Disease Survey of Carnivores in the Little Snake Area, ACUC 1997-3). RESULTS AND DISCUSSION Initial sampling (February 2000) at WCMA indicated substantially lower exposure rates to select pathogens than observed at LSMA. Data from 200 I surveys indicated a relatively high proportion of adult coyotes exposed to canine distemper virus (CDV)(Figure 1): in February 200 I, about 79% of the coyotes sampled had serum neutralizing titers :::I: 16. Recent sampling revealed lower proportions of CDV-positive coyotes, similar to the initial sampling periods. In contrast to canine distemper, exposure to plague appears relatively rare among coyotes sampled from WCMA (Figure 1). As tularemia is commonly found in rodents in Colorado, a seroprevalence of 20-40% is not surprising in WCMA Date/pathogen ■ Positive El Negative Figure 1. Seroprevalence of presumed tularemia, plague, and canine distemper exposure among coyotes sampled from the Wolf Creek Management Area, Colorado, during February 2000 to August 2003. �Colorado Division of Wildlife Wildlife Research Report July 2003 – June 2004 JOB PROGRESS REPORT State of Project No. Work Package No. Colorado 0880 : Task Federal Aid Project : : : N/A Cost Center 3430 Mammals Research Black-Footed Ferret Conservation Black-Footed Ferret Recovery Program Disease Monitoring & Management : Period Covered: July 1 2003 through June 30, 2004 Author: L. L. Wolfe Personnel: L. A. Baeten, H. Edwards, C. T. Larsen, M. W. Miller, K. Taurman, E. S. Williams All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We continued monitoring carnivores at proposed black-footed ferret reintroduction sites for serological evidence of select disease epidemics. Sampling at the Wolf Creek Management Area (WCMA) in August 2004 revealed little evidence of ongoing epidemics that could impede black-footed ferret restoration efforts. Serology data from culled coyotes showed no evidence of active canine distemper or plague epidemics in the WCMA vicinity. In contrast, serologic evidence of exposure to tularemia continues to be relatively high (~30%), consistent with previous observations in this and other monitored areas. We will continue this work as part of the ongoing Colorado−Utah black-footed ferret reintroduction protocol. 17 �JOB PROGRESS REPORT BLACK-FOOTED FERRET RECOVERY PROGRAM DISEASE MONITORING & MANAGEMENT LISA L. WOLFE INTRODUCTION As part of the Colorado−Utah black-footed ferret reintroduction protocol, we continued monitoring carnivores at proposed ferret reintroduction sites for serological evidence of select disease epidemics. Originally, we monitored coyote (Canis latrans) populations at two Colorado sites: the Little Snake Management Area (LSMA) and the Wolf Creek Management Area (WCMA), Colorado. Under this program, >200 coyotes have been collected for post-mortem examination and samples collected as described in established protocols since March 1997. Monitoring has been accomplished via cooperative efforts of Colorado Division of Wildlife, USDA Wildlife Services, and Bureau of Land Management (BLM) personnel. To date, no lesions indicative of active infections with any of the select pathogens (Francisellia tularensis, Yersinia pestis, canine distemper virus [CDV]) have been noted on gross examinations of carcasses. However, relatively high proportions (31-89%) of the coyotes collected from the LSMA had positive titers to plague between March 1997 and July 1999. Although the proportion of plague-positive coyotes declined during the sampling period, evidence of continued exposure and perhaps declining prairie dog abundance led to abandonment of surveillance at LSMA after 1999. Monitoring at the WCMA has continued, and black-footed ferrets were reintroduced at this site in 2001. RESULTS AND DISCUSSION Disease surveillance As part of the Colorado-Utah black-footed ferret reintroduction protocol, we monitored serological evidence of exposure to select infectious diseases in coyotes at Wolf Creek Management Area (WCMA), Colorado; our strategy was to use exposed coyotes as sentinels for detecting epidemics at restoration and prospective release sites. Over 350 coyotes (Canis latrans) have been collected for postmortem examination and samples collected as described in established protocols since March 1997 via cooperative efforts of Colorado Division of Wildlife, USDA Wildlife Services, and Bureau of Land Management (BLM) personnel. Coyotes were collected using a combination of calling and aerial gunning. In 2004, 20 coyotes were sampled (5 pups, 6 juvenile, 6 adult, 3 not aged) in late July. (Because data from juveniles are most useful in detecting evidence of recent epidemics, we discontinued mid-winter sampling in 2002.) No lesions indicative of active infections with select pathogens (Francisellia tularensis, Yersinia pestis, canine distemper virus [CDV]) were noted on gross examinations of carcasses through 2002; in the absence of meaningful necropsy findings, we discontinued gross examinations of carcasses in 2003. Initial sampling (February 2000) at WCMA indicated substantially lower exposure rates to select pathogens than observed at another site (Little Snake Management Area) monitored in earlier years of this survey. Initial sampling demonstrated 24 percent of the coyotes surveyed with antibody titers suggestive of exposure to CDV. Although seroprevalence increased slightly in 2001, sampling since 2002 revealed much lower proportions of CDV-positive coyotes (Figure 1). There was no serologic evidence of CDV exposure in 2002, and only 1 case each in 2003 and 2004. Exposure to plague still appears relatively rare among coyotes sampled from WCMA (Figure 1). In 2004 one juvenile, out of 21 coyotes sampled, was 18 �“moderately positive” antibody titer to plague. The most significant pathogen exposure noted by seroprevalence is for tularemia. As tularemia is commonly found in rodents in Colorado, seroprevalence of 20–40% is not surprising in carnivores, and very little change in tularemia seroprevalence has been seen over the 5-year sampling period. Prepared by ________________________ Lisa L. Wolfe, Veterinarian 100% 80% 60% 40% 20% Y. pestsis F. tularensis CDV Date/pathogen 08/04 (n=20) 08/03 (n=11) 07/02 (n=13) 07/01 (n=16) 07/00 (n=20) 08/04 (n=20) 08/03 (n=11) 07/02 (n=13) 07/01 (n=16) 07/00 (n=20) 08/04 (n=20) 08/03 (n=11) 07/02 (n=13) 07/01 (n=16) 07/00 (n=20) 0% ■ Positive □ Negative Figure 1. Seroprevalence of presumed tularemia (F. tularensis), plague (Y. pestis), and canine distemper virus (CDV) exposure among coyotes sampled from the Wolf Creek Management Area, Colorado, during summer sampling 2000−2004. 19 � https://cpw.cvlcollections.org/files/original/188a97d1068aa14770546b8a69927369.pdf a9101ca9d2887885c3c9b6fc98a1e885 PDF Text Text 245 Colorado Division of Wildlife Wildlife Research Report July 200 l and July 2002 JOB PROGRESS REPORT State of - - - - - - - - - - -Colorado "-==~----- Mammals Research Work Package No. - - - - ~ l ~ A ~ - - - - - Multispecies Investigations Task No. - - - - - - - - - = 5_ _ _ _ _ __ Consulting Services for Mark-Recapture Analvsis Federal Aid Project No. __W~-1_5_3~-R~-~2_ _ _ __ Research and Development 1 Period Covered: July l, 2000 - June 30, 2001 Author: G C. White, Ph.D. Personnel: C. Bishop, R. B. Gill, D. C. Bowden, R. M. Bartmann, D. J. Freddy, T. M. Shenk, M. M. Conner, M. Post Vieira, A. Dharman, B. Lubow ABSTRACT Progress towards the objectives of this job include: Consulting assistance to CDOW on harvest surveys, terrestrial inventory systems, and population modeling procedures was provided. Estimates of spring and fall turkey, spring snow goose, sharp-tailed and sage grouse, chukars, ptarmigan, Abert's squirrels, and general small game harvest were computed from survey data, and programs and harvest estimates provided to CDOW via email and CD ROM. Input on the design and analysis of the Harvest Information Program was provided on several occasions. The DEAMAN software package for the storage, summary, and analysis of big game population and harvest data was revised further as a Windows 95/98/NT/2000/ME program. The capability to incorporate data on radio-collared animals to estimate survival with the Kaplan-Meier estimator and display movement data was added, and distributed to terrestrial biologist via the WWW at http://www.cnr. colostate. edu/-gw-hite/deaman. A 1-day workshop was conducted with NE region personnel in the use of DEAMAN and population modeling procedures, mainly to instruct region personnel on the use of spreadsheet models for ungulate population dynamics. In addition, numerous questions were answered via meetings with biologists, and via email. A preliminary analysis of the survival rates from the mule deer monitoring data was completed. However, I have not received final data from some of the biologists, so have not been able to complete this analysis. A paper, coauthored with Bruce Lubow, was submitted for publication to the Journal of Wildlife Management on past efforts to develop a realistic mule deer population model based on data collected with current CDOW procedures. Data from the Piceance Basin were used to illustrate the modeling I '; li~ri~r BD0WD16832 �246 technique. In addition, a book chapter on modeling big game populations appeared in print: White, G. C. 2000. Modeling Population Dynamics. Pages 84-107 in S. Demarais and P.R. Krausman, eds. Ecology and Management of Large Mammals in North America. Prentice-Hall, Upper Saddle River, New Jersey, USA. A paper on optimal allocation of resources to sample Colorado mule deer populations was published in the Journal of Wildlife Management: Bowden, D. C., G. C. White, and R. M. Bartmann. 2000. Optimal allocation of sampling effort for monitoring a harvested mule deer population. Journal of Wildlife Management 64:1013-1024. A paper on trends in Colorado mule deer age and sex ratios was published in the Journal of Wildlife Management: White, G. C., D. J. Freddy, R. B. Gill, and J. H. Ellenberger. 2001. Effect of adult sex ratio on mule deer and elk productivity in Colorado. Journal of Wildlife Management 65:436-444. Assistance in the design and analysis of candidate systems to estimate deer abundance in GMU 10 was provided. A research study to examine the impact of nutrition on the decline of mule deer fecundity during the last 20 years was initiated. I have provided input on estimation of the number of deer on the feed sites, and developed an estimator of fawn survival rates based on radio-collared does and fall and spring fawn:doe ratios. Data were collected and analyzed on spatial distribution, movement of radio-collared animals, and population sizes related to estimating the spread and impacts of chronic wasting disease in deer populations. A report summarizing these findings was provided to CDOW personnel involved with the study. A final report on the response of elk to lower numbers of archery licenses in the White River Data Analysis Unit was prepared and submitted to CDOW personnel involved with the project. A paper reporting results of the earlier experiment to detect elk response to the opening of archery season has been accepted for publication in the Journal of Wildlife Management: Conner, M. M., G. C. White, and D. J. Freddy. 2001. Elk movement in response to early-season hunting in Colorado. Journal of Wildlife Management 65. In Press. A graduate research project to evaluate the movements of Preble's meadow jumping mouse populations away from riparian areas was completed. A final report of this project was submitted to CDOW personnel involved with the project. • • In cooperation with CDOW personnel, I developed an analysis of survival of lynx released as part of the reintroduction program. An analysis to estimate the effort required to estimate the percent of eastern Colorado inhabited by blacktailed prairie dogs was completed and results provided to CDOW personnel involved with the effort. �247 CONSULTING SERVICES FOR MARK-RECAPTURE ANALYSES G. C. White P, N. OBJECTIVES Design a sampling scheme to estimate the area of black-tailed prairie dog colonies in eastern Colorado. SEGMENT OBJECTIVES I. Develop a sampling scheme to estimate the area of black-tailed prairie dog colonies in eastern Colorado. 2. Develop estimates of the cost of this sampling scheme as a function of the expected precision. RESULTS AND DISCUSSION Area of black-tailed prairie dog colonies in Wyoming, North and South Dakota, and Nebraska have been sampled successfully with aerial line intercept sampling techniques (Sidle et al. In Press). CDOW is interested in applying this technique to eastern Colorado, and obtaining estimates of the areas occupied by prairie dogs by county. However, there are concerns regarding the cost of the survey and expected precision. In the following, I present an analysis of the expected cost as a function of the precision of the estimates of area of black-tailed prairie dog colonies. To compute the expected precision as a function of the cost of the aerial survey for black-tailed prairie dogs, I went through the following steps. I assumed that the lines to be flown would be stratified by county. Because an infinite number of lines can be flown for each county, the sampling scheme can be viewed as sampling with replacement, and hence, no finite population correction is allowed. Further, I treated each line as providing an estimate of the proportion of the county area in active prairie dog towns, and not as a ratio estimator. Sidle et al. (In Press) compared both types of estimators, and developed a composite of the 2. However, design of a new survey seemed easier to conceptualize with the approach taken. This approach allowed the use of the formulas on pages 341-342 of Thompson et al. ( 1998), ignoring the finite population correction. Other pertinent references are Thompson (1992) and Cochran (1997). From the area of each county and the estimated area of prairie dog towns within the county provided by the EDAW survey, I predicted the proportion of the line in each county that would intersect dog towns (r) as: ActiveTownsArea C r = ---------County Area A From Table 1 of Sidle et al. (In Press) I computed the relationship between the standard deviation of r [SD(r)] and the value of r. To compute SD(r) from Table 1 of Sidle et al., I got the nwnber of lines flown for each of the 8 surveys from Douglas Johnson, Northern Prairie Wildlife Research Center, USGS. A linear relationship of SD(-)= 0.0087 + I.0804r was found to provide a decent fit to the data (Figure 1). �248 0.25 0.2 -- "t::' -..;;;.. 0.15 - Q Cl) 0.1 0.05 0 0 0.05 0.1 0.15 0.2 r I assumed that each county was to be estimated with some relative precision (e) such that the 95% confidence interval for each county would be ± eC, where C is the estimated acres of active towns. This approach is not the optimal for the estimate of the total active town area in the state, but would provide good estimates (i.e., estimates of quality e) for each county. For each county, the standard error of the estimate of the active prairie dog town area [SE(C)] was computed based on the SD(r) and the estimated number oflines (n) to be flown, where SE(r) = SD(r)/ ✓ n, and SE(C) 2 = A SE(r). Given the desired level of precision, I computed the number oflines to fly in a county as: Because some counties had values of C = 0 (and hence the above equation i; undefined), and others have very small values, I assumed that all counties had at least 0.5% area in active towns to compute these samples sizes, although the actual value of r was used to compute the standard deviation. The total length of the lines to be flown in the county is the square root of the county area multiplied by the number of lines to be flown. Cost of the survey for a county was figured as the length of line to be flown plus 2 times the square root of county area in miles (to account for ferry time), all divided by a flight speed of 90 mph, times $180 per hour of flight time. The total acreage of prairie dog towns (Cr) is the sum of the county estimates, with the variance computed as the sum of the variances across the counties. �189 JOB PROGRESS REPORT Stateof _ _ _---'C=o=l=o=ra=d=o_ _ _ _ _ __ Mammals Research Program Work Package No. _ _ _ _ _ _ _ _ _ __ Multispecies Investigations Task No. --~5_ _ _ _ _ _ _ _ __ Consulting Service for Mark-Recapture Analysis Federal Aid Project No. W-153-R-2 Period Covered: July 1, 2002 - June 30, 2003 Author: G. C. White Personnel: C. Bishop, G. Miller, T. E. Remington, D. J. Freddy, T. M. Shenk, L. Stevens, J. Craig, R. Kahn, D. C. Bowden, F. Pusateri, J. Dennis, P. Schnurr, B. Andelt, A Seglund, D. Finley, A Linstrom, D. Walsh, K. Strohm. ABSTRACT Progress towards the objectives of this job include: 1. Consulting assistance to CDOW on harvest surveys, terrestrial inventory systems, and population modeling procedures was provided. Estimates of spring and fall turkey, spring snow goose, sharp-tailed and sage grouse, chukars, ptarmigan, Abert's squirrels, and general small game harvest were computed from survey data, and programs and harvest estimates provided to CDOW via email and CD ROM. Computer code written in SAS to compute these estimates and display results graphically was also provided. Computer code was also written in SAS to estimate the compliance rate of Colorado small game license holders with the Harvest Information Program. 2. The DEAMAN software package for the storage, summary, and analysis of big game population and harvest data was revised further as a Windows 95/98/NT/2000/ME/XP program. A User's Manual was provided to terrestrial biologists on CD and also distributed via the WWW at http://www.cnr.colostate.edu/~gwhite/deaman. 3. Consultation with CDOW Terrestial Biologists in the use of DEAMAN and population modeling procedures continued. Numerous questions were answered via meetings with biologists, and via email. 4. A paper, coauthored with Marilet Zablan and Clait Braun, was published in the Journal of Wildlife Management on past efforts to estimate survival rates of sage grouse in North Park from CDOW banding records. The full citation is: Zablan, M. A, C. E. Braun, and G. C. White. 2003. Estimation of northern sage-grouse survival in North Park, Colorado. Journal of Wildlife Management 67:144-154. �190 5. A paper on the estimation of population size from correlated sampling unit estimates of the variable of interest was published in the Journal of Wildlife Management. The methodology developed in this paper is proposed for use in a joint Colorado/Utah survey of the colony area of whitetailed and Gunnison prairie dogs in western Colorado and eastern Utah. The full citation is: Bowden, D. C., G. C. White, A. B. Franklin, and J. L. Ganey. 2003. Estimating population size with correlated sampling unit estimates. Journal of Wildlife Management 67: 1-10. 6. A paper on the use of lek counts to index prairie grouse populations was published in the Wildlife Society Bulletin: Walsh, D. P., G. C. White, T. E. Remington, and D. C. Bowden. 2003. Evaluation of Lek Count Index for Prairie Grouse Wildlife Society Bulletin. 32:56-68. 7. A paper on the estimation of sage grouse populations was submitted to the Journal of Wildlife Management: Walsh, D. P., G. C. White, T. E. Remington, and D. C. Bowden. 2003. Population Estimation of Greater Sage-Grouse. Journal of Wildlife Management. Submitted. 8. A paper on the effects of early season hunter numbers on elk movement was published in the Journal of Wildlife Management: Vieira, M. E. P., M. M. Conner, G. C. White, and D. J. Freddy. 2003. Relative effects of early season hunter numbers and opening date on elk movement in northwest Colorado. Journal of Wildlife Management. 67:717-728. 9. A paper on the impact of limited antlered harvest on mule deer sex and age ratios was submitted to the Wildlife Society Bulletin: Bishop, C. J., G. C. White, D. J. Freddy, and B. E. Watkins. 2003. Effect of limited antlered harvest on mule deer sex and age ratios. Wildlife Society Bulletin. Submitted. 10. A paper on the survival and recruitment of peregrine falcons was published in the Journal of Wildlife Management: Craig, G. R., G. C. White, and J. H. Enderson. 2004. Survival, recruitment, and rate of population change of the Colorado peregrine falcon population. Journal of Wildlife Management. In Press. 11. A research study to examine the impact of nutrition on the decline of mule deer fecundity during the last 20 years was continued. I have provided input on estimation of the number of deer on the feed sites, and developed an estimator of fawn survival rates based on radio-collared does and fall and spring fawn:doe ratios. 12. A graduate research project by Dan Walsh to evaluate utility oflek counts of Greater Sagegrouse in Middle Park was completed. Mark-resight methods are being used to estimate lek attendance and population size. The thesis citation is: Walsh, D. P. 2002. Population Estimation Techniques for Greater Sage-grouse. M. S. Thesis, Colorado State University, Fort Collins. USA. 158pp. 13. A graduate research project to develop a sage grouse population model, using North Park sage grouse data to develop parameter estimates, was initiated. The graduate student is Kristen Strohm. 14. An analysis to estimate the estimate the percent of eastern Colorado inhabited by black-tailed prairie dogs was completed and results provided to CDOW personnel involved with the effort. Estimates were computed in an Excel spreadsheet, and also verified through a program written in SAS to be sure that no errors in the calculations would be found when the spreadsheet is distributed to interested stakeholders. �191 15. Development of the design of a monitoring system for white-tailed prairie dogs in western Colorado and eastern Utah was started. This effort is in cooperation with Pam Schnurr, Bill Andelt, and Amy Seglund. 16. Development of the design of a monitoring system for swift fox in eastern Colorado was started. This effort is in cooperation with Francie Pusatari and Darby Finley. 17. Two new graduate students have been accepted for my supervision in the Department of Fishery and Wildlife Biology at Colorado State University. Chad Bishop will start a Ph.D. program in Fall, 2003, and Aaron Linstrom will start an M.S. program in Fall, 2003. �192 CONSULTING SERVICES FOR MARK-RECAPTURE ANALYSES G. C. White P. N. OBJECTIVES Assess the status of Colorado swift fox population through an occupancy monitoring approach. SEGMENT OBJECTIVES 1. Develop a monitoring scheme to estimate the occupancy rate of swift fox in eastern Colorado. 2. Determine necessary sample sizes to obtain adequate statistical power to detect biologically important changes in the occupancy rate. RESULTS AND DISCUSSION Estimation of occupancy rate for Swift Foxes (Vulpes velox) in eastern Colorado was based on trapping data provided by Finley (1999). The data consist of 72 randomly selected trapping grids 4 miles by 5 miles in area, with 20 traps set at 1 mile intervals. METHODS The occupancy model of MacKenzie et al. (2002) was fit to the 72 trapping grids using Program MARK (White and Burnham 1999). The model fit included 8 detection probabilities (p) for the 8 trapping occasions plus the probability of occupancy ( f// ). Detection probabilities were predicted with the month that a grid was trapped. Month was modeled with trigometric functions; sin(Monthx2n/12) and cos(Monthx2n/12), and powers of these functions. By using these sin and cosine functions, I can make the capture probability continuous across the December to January interval. Trend models were also used to model capture probabilities across occasions, forcing a linear trend on a logit scale in the capture probabilities. The percentage of each trapping grid comprised of short grass prairie was used as an additional covariate to predict both detection probability and probability of occupancy on a logit scale. Model selection was performed with AICc (Burnham and Anderson 1999). RESULTS Model selection results (Table l)suggest that month is an important predictor of the probability of detecting foxes on a grid. In addition, the top-ranked AICc includes a positive trend effect in the detection probabilities across the occasions, consistent with the results from the population estimation models. Model selection results also suggest that short grass prairie vegetation affects both the detection probability as well as the probability of occupancy. Detection probability is affected by the density of animals on the grid, and the percentage of short grass prairie on a trapping grid correlates (r = 0.375) with estimated population sizes provided in Table 5 of my September 23rd memo. �193 Table 1. Model selection results from fitting the occueancy estimation model of MacKenzie et al. (2002). AJCc ~AJCc AICcNum Weights Par. Deviance {p(T+cosMonth+cosMonth/\2) psi(SGPProp)} 318.6146 0.0000 0.31440 6 305.3223 {p(T+cosMonth+cosMonth/\2+SGPProp) psi(SGPProp)} 319.0596 0.4450 0.25168 7 303.3096 {p(T+cosMonth+cosMonth/\2+SGPProp) psi} 320.1094 1.4948 0.14890 6 306.8171 {p(cosMonth+cosMonth/\2) psi(SGPProp)} 321.2674 2.6528 0.08345 5 310.3583 {p(cosMonth+cosMonth/\2+SGPProp) psi(SGPProp)} 321.3341 2.7195 0.08071 6 308.0418 {p(T+cosMonth+cosMonth/\2) psi} 322.6843 4.0697 0.04109 5 311.7753 {p(cosMonth+cosMonth/\2+cosMonth/\3) psi} 322.7973 4.1827 0.03884 5 311.8882 {p(cosMonth) psi(SGPProp)} 323.6307 5.0161 0.02560 4 315.0337 {p(cosMonth+cosMonth/\2) psi} 325.5083 6.8937 0.01001 4 316.9113 {p(cosMonth) psi} 328.1180 9.5034 0.00272 3 321.7651 {p(cosMonth+sinMonth) psi} 329.1845 10.5699 0.00159 4 320.5875 {p(t+cosMonth+cosMonth/\2) psi} 330.1385 11.5239 0.00099 11 303.7385 {p(.) psi} 340.3683 21.7537 0.00001 2 336.1944 {p(sinMonth) psi} 342.5446 23.9300 0 3 336.1917 {p(t) psi} 343.2296 24.6150 0 9 322.3264 Model Parameter values for the top-ranked AJC 0 model (Table 2) demonstrate the increasing detection probability with occasion. In addition, the estimate of 1// of 0.821 suggests that 59.1 of the 72 grids trapped contained foxes, in contrast to the 51 grids that were observed to have foxes. �194 Table 2. Parameter estimates for the month of March from the top-ranked AI Cc occupancy model {p(T + cos(Month) + cos2(Month)) psi(SGP Proportion)}, where month was set to March (3), and the short grass prairie habitat proportion for the trapping grid was set to 66.9%, the mean of the grids trapped. Parameter Estimate SE LCI UCI P1 0.611647 0.083074 0.442448 0.757627 P2 0.675704 0.066681 0.534363 0.790928 p3 0.733793 0.060627 0.600043 0.835107 p4 0.784792 0.061708 0.640538 0.881835 Ps 0.828306 0.064248 0.665567 0.921227 P6 0.864541 0.065008 0.682552 0.949861 P1 0.894106 0.063204 0.695294 0.968985 pg 0.917831 0.059218 0.705628 0.981150 \jl 0.820811 0.065876 0.655653 0.916806 The effect of month in the top-ranked AI Cc model (Figure 1) is significant, and somewhat consistent with the results obtained with the population estimation models that included the variable month (reported in the memo of September 23). That is, the lowest detection probabilities are during summer. However, the occupancy model results suggest that September through March have the highest detection probabilities. The impact of the percentage of short grass prairie habitat on the estimates of occupancy is strong (Figure 2), with the probability of occupancy estimated at 34% for trapping grids with no short grass prairie habitat up to 93% for grids consisting of 100% short grass prairie. �195 0.9 0.8 >- ~ 0.7 :cn:s 0.6 .c e o.s a. g 0.4 0 0.3 .....Q) cosine quadratic --t--Trend+cosine quadratic '3 0.2 0.1 0 1 2 3 4 5 7 6 8 9 10 11 12 Month Figure 1. Effect of month in the 3 of the models of occupancy considered for detection probability: {p(cosMonth+cosMonth"2+cosMonth1''3) psi}=cosine cubic, {p(cosMonth+cosMonth"2) psi}=cosine quadratic, and {p(T + cos(Month) + cos2(Month)) psi}=Trend+cosine quadratic. The values shown for {p(T + cos(Month) + cos\Month)) psi} are for p 1_so estimates for p 2 throughp 8 increase monotonically from this value. >, 1 g 0.9 +-------- :g_ 0.8 ~ 0.7 + - - - - - - - - - - - - - - - - = ~ - - = - - - - - - - - - - - - - - - - - - - - - - - - - - - - c : ou 0.6 -t------------=_..,...,__ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _----J 0 0.5 - j - - - - - - - - - = ~ , . . . - C - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - c : £ 0.4 +-~-~--------------------------------': .o 0.3 - + - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - c : ~ 0.2 + - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ' : ~ 0--t-----.......,..-----..........,--------r-------,--------....-.....,------------i e 0.1 -+-----------------------------------, 0 20 40 60 80 100 Short Grass Prairie (%) Figure 2. Effect of the percentage of the grid consisting of short grass prairie habitat on the probability of occupancy for the top-ranked AICc model {p(T + cos(Month) + cos2(Month)) psi(SGP Proportion)}. 120 �196 DISCUSSION The high detection probabilities during the September through March period suggests that swift fox monitoring should take place during this period. The increasing detection probability with trapping occasion also suggests that increasing the number of occasions will result in higher detection probabilities on each succeeding occasion. However, this trend effect is relatively minor. That is, the probability of not detecting foxes on a grid with 2 occasions trapped during March with the trend model estimates is (I - 0.610297) □ (1 0.6749937) = 0.126656. With the cosine quadratic model that does not include a trend across occasions, the probability of not detecting foxes is 0.085212. With 3 trapping occasions in March, the corresponding probabilities are 0.041164 and 0.024874, respectively. The strong relationship between the probability of occupancy and the short grass prairie habitat variable suggests that the design of an occupancy monitoring scheme should include this covariate. In particular, a ratio estimator can be developed that predicts the probability of occupancy based on the relationship in Figure 2. FURTHER WORK A reasonable estimate of the number of swift foxes in eastern Colorado can be obtained from the grid trapping scheme analyzed here. The population estimate for each trapping grid within a strata can be used to obtain a naive estimate of population density that will be biased high. However, through the use of radio collars, the proportion of time that marked animals spend on the trapping grid where they were initially captured can be used to correct these naive estimates. That is, the naive estimate multiplied by the proportion of radio locations on the trapping grid gives an unbiased estimate of density. Such a procedure has been used by White and Shenk (2001) to estimate population sizes for Preble's Meadow Jumping Mice, and details are provided in that article. Thus, to obtain an unbiased population estimate, radio-collared animals would be required. LITERATURE CITED Burnham, K. P., and D.R. Anderson. 1998. Model selection and inference: a practical information-theoretic approach. Springer-Verlag, New York, New York, USA. 353 pp. Finley, D. J. 1999. Distribution of the swift fox (Vulpes velox) on the eastern plains of Colorado. M. S. Thesis, University of Northern Colorado, Greeley, USA. 96pp. MacKenzie, D. I., J. D. Nichols, G. B. Lachman, S. Droege, J. A. Royle, and C. A. Langtimm. 2002. Estimating site occupancy when detection probabilities are less than one. Ecology 83:2248-2255. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 Supplement: 120-138. White, G. C., and T. M. Shenk. 2001. Population estimation with radio-marked animals. Pages 329-350 in J. J. Millspaugh and J.M. Marzluff, editors. Design and Analysis of Wildlife Radiotelemetry Studies. Academic Press, San Diego, California, USA. �Colorado Division of Wildlife Wildlife Research Report July 2003 – June 2004 JOB PROGRESS REPORT State of Colorado Project No. Work Package No. Task No. 5 Federal Aid Project: 3001 W-185-R : : : : Cost Center: 3430 Mammals Research Multispecies Investigations Consulting Services for Mark-Recapture Analysis : Period Covered: July 1, 2003 - June 30, 2004 Author: G. C. White Personnel: C. Bishop, G. Miller, T. E. Remington, D. J. Freddy, T. M. Shenk, L. Stevens, J. Craig, R. Kahn, D. C. Bowden, F. Pusateri, J. Dennis, P. Schnurr, B. Andelt, A. Seglund, D. Finley, A. Linstrom, K. Strohm, P. Conn. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Progress towards the objectives of this job include: 1. 2. 3. 4. 5. Consulting assistance to CDOW on harvest surveys, terrestrial inventory systems, and population modeling procedures was provided. Estimates of spring and fall turkey, spring snow goose, sharp-tailed and sage grouse, chukars, ptarmigan, Abert’s squirrels, and general small game harvest were computed from survey data, and programs and harvest estimates provided to CDOW via email and CD ROM. Computer code written in SAS to compute these estimates and display results graphically was also provided. Computer code was also written in SAS to estimate the compliance rate of Colorado small game license holders with the Harvest Information Program. The DEAMAN software package for the storage, summary, and analysis of big game population and harvest data was revised further as a Windows 95/98/NT/2000/ME/XP program. A User’s Manual was provided to terrestrial biologists on CD and also distributed via the WWW at http://www.cnr.colostate.edu/~gwhite/deaman. Consultation with CDOW Terrestrial Biologists in the use of DEAMAN and population modeling procedures continued. Numerous questions were answered via meetings with biologists, and via email. A paper on the use of lek counts to index prairie grouse populations was published in the Wildlife Society Bulletin: Walsh, D. P., G. C. White, T. E. Remington, and D. C. Bowden. 2004. Evaluation of the lek count index for greater sage-grouse. Wildlife Society Bulletin 32: 56-68. A paper on the effects of early season hunter numbers on elk movement was published in the Journal of Wildlife Management: Vieira, M. E. P., M. M. Conner, G. C. White, and D. J. 151 �6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. Freddy. 2003. Effects of archery hunter numbers and opening dates on elk movement. Journal of Wildlife Management. 67:717-728. A paper discussing the implications of the GMU 10 special mule deer surveys was accepted for publication in the Wildlife Society Bulletin: Freddy, D. J., G. C. White, M. C. Kneeland, R. H. Kahn, J. W. Unsworth, W. J. deVergie, V. K. Grahm, J. H. Ellenberger, and C. H. Wagner. 2004. How many mule deer are there? Challenges of credibility in Colorado. Wildlife Society Bulletin. In Press. A paper on the impact of limited antlered harvest on mule deer sex and age ratios was accepted for publication in the Wildlife Society Bulletin: Bishop, C. J., G. C. White, D. J. Freddy, and B. E. Watkins. 2004. Effect of limited antlered harvest on mule deer sex and age ratios. Wildlife Society Bulletin. In Press. A paper on the survival and recruitment of peregrine falcons was accepted for publication in the Journal of Wildlife Management: Craig, G. R., G. C. White, and J. H. Enderson. 2004. Survival, recruitment, and rate of population change of the Colorado peregrine falcon population. Journal of Wildlife Management. In Press. A paper on the estimation of the area of black-tailed prairie dog colonies in eastern Colorado was submitted to the Wildlife Society Bulletin: White, G. C., J. R. Dennis, and F. M. Pusateri. 2004. Area of black-tailed prairie dog colonies in eastern Colorado. Wildlife Society Bulletin. Submitted. A paper on methodologies to obtain more rigorous population monitoring data was submitted to Wildlife Research: White, G. C. 2004. Correcting counts: techniques to de-index. Wildlife Research. Submitted. A paper evaluating methods of estimating the impact of harvest on survival rates was submitted to Animal Diversity and Conservation: Otis, D. L., and G. C. White. 2004. Evaluation of ultrastructure and random effects band recovery models for estimating relationships between survival and harvest rates in exploited populations. Animal Biodiversity and Conservation. Submitted. A paper on the procedures to monitor swift fox populations in eastern Colorado was submitted to the Journal of Wildlife Management: Finley, D. J., G. C. White and J. P. Fitzgerald. 2004. Estimation of swift fox population size and occupancy rates in eastern Colorado. Journal of Wildlife Management. Submitted. A research study to examine the impact of nutrition on the decline of mule deer fecundity during the last 20 years was continued in cooperation with Chad Bishop. Portions of this work will serve as his doctoral dissertation. A graduate research project (M. S.) to develop a sage grouse population model, using North Park sage grouse data to develop parameter estimates, was continued. The graduate student is Kristen Strohm. A graduate research project (M. S.) To evaluate line transect methodology for estimating pronghorn populations in eastern Colorado was initiated. The graduate student is Aaron Linstrom, and the project is in addition to his full-time duties as a terrestrial biologist with CDOW. A graduate research project (Ph. D.) to develop statistical models to monitor puma and black bear populations in Colorado based on checks of harvested animals and DNA and/or radiotracking data was initiated. The graduate student is Paul Conn. Development of the design of a monitoring system for white-tailed prairie dogs in western Colorado and eastern Utah was continued. This effort is in cooperation with Pam Schnurr, Bill Andelt, and Amy Seglund. Development of the design of a monitoring system for swift fox in eastern Colorado was continued. This effort is in cooperation with Francie Pusatari and Darby Finley. 152 �19. A workshop on use of the DEAMAN software for data entry, data summaries, and population modeling was presented to CDOW Terrestrial Biologists on May 20, 2004. A revised edition of the DEAMAN User’s Manual was provided on a CD. 153 �JOB PROGRESS REPORT CONSULTING SERVICES FOR MARK-RECAPTURE ANALYSES G. C. White P. N. OBJECTIVES Extend existing methods to better provide rigorous population monitoring systems. SEGMENT OBJECTIVES 1. 2. Extend a mark-recapture monitoring scheme to estimate population sizes with inadequate data per site to estimate encounter probabilities. Contrast line transect distance sampling approaches with mark-recapture approaches for monitoring populations with inadequate data per site to estimate encounter probabilities. ABSTRACT One of the most pervasive uses of indices of wildlife populations is uncorrected counts of animals. Two examples are the minimum number known alive from capture and release studies, and aerial surveys where the detection probability is not estimated from a sightability model, marked animals, or distance sampling. Both the mark-recapture and distance sampling estimators are techniques to estimate the probability of detection of an individual animal (or cluster of animals), which is then used to correct a count of animals. However, often the number of animals in a survey is inadequate to compute an estimate of the detection probability, and hence correct the count. Modern methods allow sophisticated modeling to estimate the detection probability, including incorporating covariates to provide additional information about the detection probability. Examples from both distance and mark-recapture sampling are presented to demonstrate the approach. RESULTS AND DISCUSSION The practice of using raw counts of animals as an index of the population size is one of the most pervasive uses of indices in wildlife management (Anderson 2003; Engeman 2003; Anderson 2001). Two examples include aerial surveys where no probability of detection is used to correct the count of observed animals, and the use of the minimum number known alive (MNKA) in animal (particularly small mammal) trapping studies (Slade and Blair 2000; McKelvey and Pearson 2001). These counts are known to be biased estimates of population size, and when used as an index, are assumed to be proportional to population size. These uses of uncorrected counts are some of the most perilous uses of an index in the practice of wildlife management because this assumption of proportionality is seldom verified, and is often false. Nichols (1992) appealed to researchers to incorporate capture-recapture estimators into small mammal studies. However, various reasons are given to explain why indices are used in place of more rigorous capture-recapture estimators. The most common reasons (Slade and Blair 2000) include fear of violating assumptions basic to mark-recapture models (Nichols and Pollack 1983), failure to recognize that some of these models are relatively robust to heterogeneity of capture probability and trap response (Carothers 1979), mistaken belief that MNKA suffers less than other models from problems of differential probabilities of capture and survival when capture probabilities are high (Nichols and Pollack 1983; Montgomery 1987), and prevalence of protocols involving fewer than the 5 to 7 trapping occasions recommended for model selection and population estimation (Otis et al. 1978). 154 �McKelvey and Pearson (2001) found that 98% of the samples collected in studies published from 1996 through 2000 they reviewed were too small for reliable selection among models of population estimation. However, their results do not take into account improved model selection methodologies, and new software and estimators that allow combining data across multiple studies and/or sites to provide more reliable model selection and estimation of the nuisance parameters. Their results mainly reflect the capabilities of CAPTURE (Otis et al. 1978; White et al. 1982), a software package developed in the late 1970's. More recent developments are available. The purpose of this presentation is to present the advantages of modern methods of analysis that allow combining data from multiple studies into a type of meta-analysis. Although resulting estimates of population size may not be completely unbiased, these estimates will certainly have less bias than MNKA. As discussed by Eberhardt et al. (1999), status of endangered large-mammal populations may have to depend on indices of population trend, and such indices may be improved by using auxiliary variables. However, in this paper, I go beyond just trying to standardize the counts with auxiliary information, as done by Eberhardt et al. (1999), and present methods incorporating auxiliary covariates that provide estimates of the population size. Modern Methods Correcting counts to produce estimates of population size. Estimators of population size based on counts of animals share a common form. A count, C is corrected for the detection probability, p, to give the population size. Because the detection probability must be estimated as p̂ (or otherwise N would be known), the result is an estimate of population size C Nˆ = (Nichols 1992). The standard methods used in wildlife studies to estimate p are markˆ p encounter methods and distance sampling. Both these seemingly diverse methodologies perform the same function: to correct a count of animals by the probability of detecting an animal. To illustrate, consider the simple Lincoln-Petersen estimator: Nˆ = n1n 2 n n = 2 = 2 , m2 ˆ m2 p n1 where n1 and n2 are the numbers of animals captured on occasions 1 and 2, and m2 is the number of animals marked on occasion 1 that are recaptured on occasion 2. Thus, m2 is an estimate of the capture n1 probability on the second occasion (Nichols 1992), because we know that n1 animals are available for capture on the second occasion, of which m2 were captured. For distance sampling, the estimate of density (Buckland et al. 1993) is: -/ nfˆ(0) n n = = Dˆ = A, ˆ2LW ˆ 2LW p p where n is the count of animals, 2LW is the area surveyed (both sides of the transect line of length L out to ˆ(0) is equivalent to 1/ p̂ . So, the right-hand side of the equation is just the a strip width W), and f ⎛ corrected count ⎜⎜ Nˆ = ⎝ n⎞ ⎟ divided by the area counted to give density. The sightability correction ˆ ⎟⎠ p models of Samuel et al. (1987) also use an equivalent approach. The probability of sighting an animal is computed for each of the groups of animals sighted, and then the number of animals in the group is divided by the estimated sighting probability to estimate the number of animals under the observed conditions that were missed. When these estimates are summed across all groups, an overall estimate of population size is obtained. Although at first glance this estimator appears to be different than the forms shown above, in fact, it is exactly the same idea. Counts are corrected by an estimate of sightability. 155 �However, because the sightability models of Samuel et al. (1987) and their extensions require first developing a model that is then applied to multiple surveys, the protocol deviates from what is the focus of this paper. That is, this paper centers on the idea of combining a number of sparse datasets into one analysis to achieve better inferences. Therefore, sightability models do not particularly fit into this approach, and so will not be discussed further here. The take-home message of this section is that counts are corrected by some probability to achieve an estimate of population size. If this correction is the same when comparing results from two surveys, then comparing just the counts will result in the same proportional change. However, without verification of the assumption that the correction is the same for both surveys, erroneous results may ensue (Nichols 1992). Consider two counts of C1 = C2 = 100, but p̂1 = 0.5 and p̂ 2 = 0.25, resulting in N̂ 1 = 200 and N̂ 2 = 400. Without knowledge of the detection probabilities, the erroneous conclusion that the population had not changed would have been made. Of course, the opposite situation can also occur. Suppose C1 = 200 and C2 = 100, with p̂1 = 0.5 and p̂ 2 = 0.25, resulting in both N̂ 1 = N̂ 2 = 400. Just comparing the counts results in the erroneous conclusion that the population has changed, when in fact, only the detection probability has changed. Thus, comparing counts is dangerous without knowledge of the underlying detection probabilities. In the next section, more advanced approaches to estimation of the detection probability are presented. Improved modeling of data to produce estimates. Earlier approaches to estimation of the size of a closed population only used the information available from the data at hand, e.g., Otis et al. (1978). Program CAPTURE (White et al. 1982) produced separate analyses for each species, sex- and age-class, and trapping grid. However, newer software packages, such as Program MARK (White and Burnham 1999) allow the user to model parameters in user-defined models. As a result, the detection probability for a population estimator can be modeled with group-specific, time-specific, and even individual-specific covariates. These covariates provide additional information with which to improve the estimates of the detection parameters. With this kind of model building capability, multiple sparse (but related) datasets can be combined into models to generate more precise estimates of p. These estimates of detection parameters and population size do not necessarily have to be the same for each of the datasets included in the analysis. For example, suppose capture probabilities are related to habitat quality, with animals in high quality habitat having smaller home ranges, and hence less probability of encountering traps. The approach advocated here is to build a model of detection probabilities by combining the data from multiple study areas and using the information about habitat quality in the model. For example, ˆ) = β 0 + β1Habitat Quality , logit(p ⎛ p ⎞ ⎟⎟ ]. The result is a model predicting ⎝1 − p ⎠ where logit is the logit transformation [logit(p) = log⎜⎜ capture probability as a function of habitat quality, where the parameters β 0 and β1 are estimated from the data and the habitat quality values provided by the user. Instead of estimating a separate value of p for each of the study areas, and likely encountering problems with too small of sample sizes, the researcher obtains an estimate of p specific to each study area based on habitat quality. Besides incorporating covariates into the model, less parameter-rich models that are still biologically realistic can be fitted to the observed data. A more extensive example is provided in White (2001), where models are combined across day- and night-time trapping occasions, gender and age-class. That example demonstrates the capability of additive models, where additive effects in the model (e.g., an effect representing the difference between day and night capture probabilities) provide differences, yet maintain a parallelism between the estimates across time or other categories. Additive models provide a useful alternative to the full multiplicative model. For example, suppose there are 5 trapping grids, each 156 �trapped for 5 nights. The full multiplicative time-specific model would have 5 × 5 = 25 parameters. In contrast, the additive model would still have 5 time parameters, but with only 4 additional parameters to represent the differences between study areas, resulting in only 9 parameters being estimated from the data. Hence, at the expense of possible bias, much improved precision of the estimates will be obtained. Finally, modern approaches provide much more flexibility in exploring alternative models. With CAPTURE, it was all or none. Only 8 models from the Mtbh set were defined, and these were cast in concrete. Only 7 models of this set had estimators (with only 5 estimators in the original program). However, with Program MARK (White and Burnham 1999), all the likelihood models from CAPTURE can be reproduced, plus many additional possibilities are provided by variations of the original 8 models. Included in MARK are the mixture models of Pledger (2000) for modeling individual heterogeneity, and Huggins (1989, 1991) and Alho (1990) versions of the closed capture estimators that allow individual covariates to be used to model initial and recapture probabilities. Thus, approaches available in MARK provide 3 areas of improvement to handle sparse markrecapture datasets. First, covariates can be incorporated into the analysis, bring additional information. Second, flexible modeling structures can provide biologically reasonable models to combine sparse datasets. Third, more flexibility is provided to construct capture-recapture models of individual datasets. Program DISTANCE 3.5 (Thomas et al. 1998; Buckland et al. 2001), and now updated to DISTANCE 4.1, provides similar capabilities for distance sampling data as does MARK for markencounter data. Data are presented to the program in strata, with stratum-specific estimates of density provided from detection probabilities estimated across strata. Although not as flexible at this time in terms of building complex models that incorporate covariates, these capabilities are forthcoming in newer versions of the program. The essence of the approach advocated here is currently available in DISTANCE – multiple datasets can be combined to estimate the sightability parameter, yet stratumspecific estimates of density are achieved. Model selection methods. Another feature of modern methods is that the estimate from a single model is not accepted as the best estimate available from the data. Burnham and Anderson (2002) describe information-theoretic model selection methods, leading to model averaging, where estimates from multiple models are combined to obtain an estimate that is an improvement over estimates from single models. The traditional approach was to find the “best” model, and use that model to make inferences from the data. However, the process of sorting through the available models carries some baggage – multiple decisions are required to decide which model is most appropriate. As a result, a source of variation in the data analysis process is ignored – model selection uncertainty. Simulations have shown that the estimates and their confidence intervals from the “best” model do not perform as hoped (Burnham et al. 1995). In particular, confidence intervals do not cover the parameter value for the expected 95% with α = 0.05. Rather than accept the poor performance because of ignoring model selection uncertainty from using the “best” model, the model-averaging methodology provided by Burnham and Anderson (2002) incorporates the model-selection uncertainty into the estimates and associated confidence intervals. Further, the approach is more biologically satisfying. For example, who really believes that the “best” model for making inferences from a capture-recapture study is something simple like Mt? Rather, we would all suspect some individual heterogeneity to be present, as in Mh. Yet, with the traditional approach of just making inferences from the “best” model, the individual heterogeneity aspect would be completely ignored if Mt was determined to be the “best” model. With model averaging, we incorporate information from all the models that have weight associated with them, with the information provided by each model proportional to its weight. The result is an estimate that reflects more accurately what we know from the data, and that a single model is inadequate for making inferences from the data. 157 �A key part of model averaging is estimating the weight to be associated with each model. The information-theoretic approach presented by Burnham and Anderson (2002) is based on Akaike’s Information Criterion (AIC). Without going into the mathematics (details are presented in Burnham and Anderson 2001; Burnham and Anderson 2002), the general idea behind AIC model selection is to rank models based on the trade-off between bias versus precision of the estimates (Figure 1). Simple models, i.e., models with small numbers of parameters, produce more precise estimates at the expense of potentially biased estimates. In contrast, complex models, i.e., models with large numbers of parameters, will produce generally unbiased estimates, but at the cost of poor precision. That is, the sampling variance of the parameter estimates from complex models will be large compared to simple model. From the AIC value for each model, a weight is computed for each model. These weights are standardized to sum to 1, so that the weight of a model reflects the likelihood of the model. From these weights, the model-averaged estimate of the parameter across all the models considered is computed. Program MARK includes information-theoretic (i.e., AIC) model selection criteria (White and Burnham 1999) and the capability for model averaging population estimates (White et al. 2001). Although DISTANCE includes AIC model selection, the capability to model average is not presently available. However, the calculations for model averaging are simple, and can be easily performed in a spreadsheet given the estimates, standard errors, and AIC values. So what’s the price for using the approach advocated here? For the user, likely a fairly steep learning curve must be climbed. More statistical and computer expertise is required to conduct the analyses described than with traditional approaches. Although likely an excuse, competent scientists will not let this reason keep them from applying better methodology to more fully interpret their data. Data are hard to come by, and deserve full treatment once acquired. Quadrat sampling example I now present an example of estimating the population size of the Mexican spotted owl in the Upper Gila Mountains Recovery Unit. Twenty-five quadrats 50–75 km2 were sampled for owls with a 4pass removal sampling scheme (Ganey et al. 1999). When an owl was detected through night-time calling, it was located the next day and leg banded to individually identify it. Recaptures were obtained when a marked owl was located during a latter pass. These capture-recapture data from banded owls on the 11 quadrats where owls had been banded and subsequently resighted were used to estimate p, the probability of capture on a given trapping occasion (Huggins 1989). To estimate p, a closed capturerecapture modeling procedure developed by Huggins (1989, 1991) that was implemented in Program MARK (White and Burnham 1999) was used. The goal was to estimate p as precisely as possible because the sampling variances of the p’s contribute to the sampling variances of the estimated N’s. In addition to p, the probability of recapture (c) can also be estimated, adding an additional parameter to be modeled. In standard closed capture-recapture models, maximum likelihood estimation is used to estimate both p and N, simultaneously (Otis et al. 1978), i.e., the resulting estimates from standard closed capture-recapture models represent the joint maximum likelihood estimates. The Huggins models differ from the standard models in that only p and c are modeled with N being estimated as a derived parameter (i.e., N is computed algebraically from p). Thus, our initial efforts centered about modeling the capturerecapture data to obtain parsimonious estimates of p. The key point relevant to this paper is that no one quadrat had adequate data to estimate p and/or c. Data were pooled across quadrats to obtain these estimates of detection probabilities, and then used to generate an estimate of N for each of the quadrats. To estimate p, 26 closed-capture models were run in program MARK. The notation used to describe these models follows Lebreton et al. (1992). In this set of models, the effects on p were modeled by sex, road access to the quadrat, occasion-specificity, and behavioral response to initial capture (i.e., inclusion of the recapture parameter c in the model). A bias-corrected version of Akaike’s Information 158 �Criteria, AICc (Burnham and Anderson 2002) was used to rank models with the best model having the lowest AICc. The best model was p = cT+roadless+sex, which constrained p’s equal to c’s, and had a linear occasion effect (T), an effect of roadless quadrats versus non-roadless quadrats and a sex effect on the p’s. The linear occasion, roadless and sex effects were all negative and different from zero (βT = -0.350, 95% CI = -0.637, -0.063; βroadless = -1.614, 95% CI = -2.742, -0.486; βsex = -0.983, 95% CI = -1.764, -0.203). This model indicated that capture probabilities declined over occasions in a linear fashion, roadless quadrats had lower capture probabilities than roaded quadrats, and that females had lower capture probabilities than males. Rather than using the p’s solely from this model, Akaike weights were estimated for each model (Buckland et al. 1997; Burnham and Anderson 2002) which represented the likelihood of a specific model as the best model to explain this particular data set, relative to the other models examined in our set of models. Akaike weights were then used to derive a weighted mean estimate of capture probabilities (pi) (i.e., the pi were “model averaged”) for each occasion for each sex and within roaded and unroaded quadrats across all models (see Stanley 1998a, 1998b). These weighted estimates of pi had estimated standard errors that included a variance component due to model selection uncertainty, i.e., which model was best for providing an adequate structure on the p’s (Buckland et al. 1997; Burnham and Anderson 2002). Thus, we ended up with 16 estimates of p, one for each of four occasions times two types of quadrats (roaded versus unroaded) and for each sex. Based on these estimates of p, a population estimate for the recovery unit was 2173 with SE 520. Had just the raw counts been used, the estimate would have been 1564 with SE 222. This example illustrates an extreme case where each trapping grid (quadrat) contained so little information about detection probabilities that by individual quadrat, the researcher is left with no choice but to use the MNKA value. However, by combining these sparse data, useful estimates were obtained that corrected for the bias of MNKA. CONCLUSIONS Sparse data need not be an impediment to correcting counts of populations to less biased estimates of population size. Modern methods incorporate information from auxiliary variables, build models from multiple sources of information, and build biologically reasonable models with fewer parameters than older approaches. Thus, past justifications of using counts as indices to population levels because of sparse data are no longer defensible. If biologists do not correct counts, we run the risk of drawing erroneous conclusions from our data, and generally losing credibility with our public critics. 159 �LITERATURE CITED Alho, J. M. (1990). Logistic regression in capture-recapture models. Biometrics 46, 623-635. Anderson, D. R. (2003). Response to Engeman, index values rarely constitute reliable information. Wildlife Society Bulletin 31, 288-291. Anderson, D. R. (2001). The need to get the basics right in wildlife field studies. Wildlife Society Bulletin 29, 1294-1297. Buckland, S.T., Anderson, D. R., Burnham, K. P., Laake, J. L, Borchers, D. L., and Thomas, L. (2001). ‘Introduction to distance sampling.’ (Oxford University Press: London) Buckland, S. T., Burnham, K. P. , and Augustin, N. H. (1997). Model selection: an integral part of inference. Biometrics 53, 603-618. Buckland, S.T., Anderson, D.R., Burnham, K. P., and Laake, J.L. (1993). ‘Distance sampling: estimating abundance of biological populations.’ (Chapman and Hall: New York) Burnham, K. P., and Anderson, D. R. (2002). ‘Model selection and multimodel inference: a practical information-theoretic approach.’ 2nd edition. (Springer-Verlag: New York) Burnham, K. P., and Anderson, D. R. (2001). Kullback-Leibler information theory as a basis for strong inference in ecological studies. Wildlife Research 28, 111-119. Burnham, K.P., White, G. C., and Anderson, D. R. (1995). Model selection strategy in the analysis of capture-recapture data. Biometrics 51, 888-898. Carothers, A. D. (1979). Quantifying unequal catchability and its effect on survival estimates in an actual population. Journal of Animal Ecology 48, 863-869. Eberhardt, L. L., Garrott, R. A., and Becker, B. L. 1999. Using trend indices for endangered species. Marine Mammal Science 15: 766-785. Engeman, R. M. (2003). More on the need to get the basics right: population indices. Wildlife Society Bulletin 31, 286-287. Ganey, J. L., Ackers, S. , Fonken, P., Jenness, J. S., Kessler, C. , Nodal, K., Shaklee, P. , and Swarthout, E. (1999). Monitoring populations of Mexican spotted owls in Arizona and New Mexico: 1999 Progress report. USDA Forest Service, Rocky Mountain Research Station, Flagstaff, Arizona, USA. (available at http://www.rms.nau.edu/lab/4251/spowmonitoring.html). Huggins, R. M. (1989). On the statistical analysis of capture-recapture experiments. Biometrika 76, 133-140. Huggins, R. M. (1991). Some practical aspects of a conditional likelihood approach to capture experiments. Biometrics 47, 725-732. Lebreton, J.-D., Burnham, K. P., Clobert, J., and Anderson, D. R. (1992). Modeling survival and testing biological hypotheses using marked animals: a unified approach with case studies. Ecological Monographs 62: 67-118. Montgomery, W. I. (1987). The application of capture-mark-recapture methods to the enumeration of small mammal populations. Symposia of the Zoological Society of London 58, 25-57. Nichols, J. D. (1992). Capture-recapture models: using marked animals to study population dynamics. Bioscience 42, 94-102. Nichols, J. D., and Pollock, K. H. (1983). Estimation methodology in contemporary small mammal capture-recapture studies. Journal of Mammalogy 64, 253-260. McKelvey, K. S., and Pearson, D. E. (2001). Population estimation with sparse data: the role of estimators versus indices revisited. Canadian Journal of Zoology 79, 1754-1765. Otis, D. L., Burnham, K. P., White, G. C., and Anderson, D. R. (1978). Statistical inference from capture data on closed animal populations. Wildlife Monographs 62, 1-135. Pledger, S. (2000). Unified maximum likelihood estimates for closed capture-recapture models using mixtures. Biometrics 56, 434-442. Samuel, M. D., Garton, E. O., Schlegel, M. W., and Carson, R. G. (1987). Visibility bias during aerial surveys of elk in northcentral Idaho. Journal of Wildlife Management 51, 622-630. 160 �Slade, N. A., and Blair, S. M. (2000). An empirical test of using counts of individuals captured as indices of population size. Journal of Mammalogy 81, 1035-1045. Stanley, T. R., and Burnham, K. P. (1998a). Estimator selection for closed-population capture-recapture. Journal of Agricultural, Biological, and Environmental Statistics 3, 131-150. Stanley, T. R., and Burnham, K. P. (1998b). Information-theoretic model selection and model averaging for closed-population capture-recapture studies. Biometrical Journal 40, 475-494. Thomas, L., Laake, J. L. , Derry, J. F., Buckland, S. T., Borchers, D. L., Anderson, D. R., Burnham, K. P., Strindberg, S., Hedley, S. L., Burt, M. L., Marques, F. F. C., Pollard, J. H., and Fewster, R. M. (1998). Distance 3.5. Research Unit for Wildlife Population Assessment, University of St. Andrews, United Kingdom. http://www.ruwpa.st-and.ac.uk/distance/. White, G. C. (2001). Statistical models: keys to understanding the natural world. In ‘Modeling in Natural Resource Management.’ (Ed. T. M. Shenk and A. B. Franklin). pp 35-56. (Island Press: Washington, D. C.) White, G. C., Burnham, K. P., and Anderson, D. R. (2001). Advanced features of Program Mark. In ‘Wildlife, land, and people: priorities for the 21st century. Proceedings of the Second International Wildlife Management Congress.’ (Ed. R. Field, R. J. Warren, H. Okarma, and P. R. Sievert). pp 368-377. (The Wildlife Society: Bethesda, Maryland) White, G. C., and Burnham, K. P. (1999). Program MARK: survival estimation from populations of marked animals. Bird Study 46 Supplement, 120-138. White, G. C., Anderson, D. R., Burnham, K. P., and Otis, D. L. (1982). Capture-recapture and removal methods for sampling closed populations. LA-8787-NERP, Los Alamos National Laboratory, Los Alamos, New Mexico, USA. 235 pp. _________________________ Gary C. White, CSU Professor Variance Bias Squared Prepared by: Few Number of Parameters Many Figure 1. The trade-off between bias2 and variance as a function of the number of parameters (from Burnham and Anderson 2002; 2001). Models with few parameters produce precise estimates that are biased, whereas models with many parameters produce less biased estimates, but imprecise. 161 �Colorado Division of Wildlife July 2004 – June 2005 WILDLIFE RESEARCH REPORT State of Colorado Cost Center 3430 Work Package 3001 Task No. 5 Federal Aid Project: W-185-R : Division of Wildlife : Mammals Research : Deer Conservaton : Multispecies Investigations Consulting Services for Mark-Recapture Analysis : Period Covered: July 1, 2004 - June 30, 2005 Author: G. C. White Personnel: C. Bishop, G. Miller, T. E. Remington, D. J. Freddy, T. M. Shenk, L. Stevens, J. Craig, R. Kahn, D. C. Bowden, F. Pusateri, J. Dennis, P. Schnurr, B. Andelt, A. Seglund, D. Finley, A. Linstrom, K. Strohm, P. Conn. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Progress towards the objectives of this job include: 1. Consulting assistance to CDOW on harvest surveys, terrestrial inventory systems, and population modeling procedures was provided. Assistance with estimation of spring and fall turkey, spring snow goose, sharp-tailed and sage grouse, chukars, ptarmigan, Abert’s squirrels, and general small game harvest was provided, and programs and harvest estimates provided to CDOW via email and CD ROM. Computer code written in SAS to compute these estimates and display results graphically was also provided. Computer code was also written in SAS to estimate the compliance rate of Colorado small game license holders with the Harvest Information Program. 2. The DEAMAN software package for the storage, summary, and analysis of big game population and harvest data was revised further as a Windows XP program. A User’s Manual has been provided to terrestrial biologists via the WWW at http://www.cnr.colostate.edu/~gwhite/deaman. I met with the CDOW software group to discuss conversion of DEAMAN to a central server application. 3. Consultation with CDOW Terrestrial Biologists in the use of DEAMAN and population modeling procedures continued. Numerous questions were answered via meetings with biologists, and via email. 4. A paper on the estimation of mule deer population sizes in GMU 10 was published in the Wildlife Society Bulletin: Freddy, D. J., G. C. White, M. C. Kneeland, R. H. Kahn, J. W. Unsworth, W. J. deVergie, V. K. Grahm, J. H. Ellenberger, and C. H. Wagner. 2004. How many mule deer are there? Challenges of credibility in Colorado. Wildlife Society Bulletin 32:916-927. 5. A paper on the peregrine falcon population dynamics in Colorado was published in the Journal of Wildlife Management: Craig, G. R., G. C. White, and J. H. Enderson. 2004. Survival, recruitment, and rate of population change of the peregrine falcon population in Colorado. Journal of Wildlife Management 68:1032-1038. 67 �6. A paper on the impact of limited antlered harvest on mule deer sex and age ratios was accepted for publication in the Wildlife Society Bulletin: Bishop, C. J., G. C. White, D. J. Freddy, and B. E. Watkins. 2005. Effect of limited antlered harvest on mule deer sex and age ratios. Wildlife Society Bulletin. In Press. 7. A paper on the estimation of the area of black-tailed prairie dog colonies in eastern Colorado was accepted for publication in the Wildlife Society Bulletin: White, G. C., J. R. Dennis, and F. M. Pusateri. 2005. Area of black-tailed prairie dog colonies in eastern Colorado. Wildlife Society Bulletin. In Press. 8. A paper on methodologies to obtain more rigorous population monitoring data was accepted for publication in Wildlife Research: White, G. C. 2004. Correcting counts: techniques to de-index. Wildlife Research. In Press. 9. A paper evaluating methods of estimating the impact of harvest on survival rates was published in Animal Diversity and Conservation: Otis, D. L., and G. C. White. 2004. Evaluation of ultrastructure and random effects band recovery models for estimating relationships between survival and harvest rates in exploited populations. Animal Biodiversity and Conservation 27.1: 157-173. 10. A paper on the procedures to monitor swift fox populations in eastern Colorado was accepted for publication in the Journal of Wildlife Management: Finley, D. J., G. C. White and J. P. Fitzgerald. 2004. Estimation of swift fox population size and occupancy rates in eastern Colorado. Journal of Wildlife Management. In Press. 11. A research study to examine the impact of nutrition on the decline of mule deer fecundity during the last 20 years was continued in cooperation with Chad Bishop. Portions of this work will serve as his doctoral dissertation. 12. A graduate research project (M. S.) to develop a sage grouse population model, using North Park sage grouse data to develop parameter estimates, was completed. The graduate student is Kristen Strohm and her thesis is “Sage Grouse Population Dynamics in North Park, Colorado”. 13. A graduate research project (M. S.) To evaluate line transect methodology for estimating pronghorn populations in eastern Colorado was continued. The graduate student is Aaron Linstrom, and the project is in addition to his full-time duties as a terrestrial biologist with CDOW. 14. A graduate research project (Ph. D.) to develop statistical models to monitor puma and black bear populations in Colorado based on checks of harvested animals and DNA and/or radio-tracking data was continued (with funding for 04-05 through the CSU PRIMES program). The graduate student is Paul Conn. 15. Development of the design of a monitoring system for white-tailed prairie dogs in western Colorado and eastern Utah was continued. This effort is in cooperation with Pam Schnurr, Bill Andelt, and Amy Seglund. 16. Development of the design of a monitoring system for swift fox in eastern Colorado was continued, and data analysis for this project was initiated. This effort is in cooperation with Francie Pusatari and Darby Finley. 68 �WILDLIFE RESEARCH REPORT CONSULTING SERVICES FOR MARK-RECAPTURE ANALYSES GARY C. WHITE P. N. OBJECTIVE Monitor swift fox populations in eastern Colorado. SEGMENT OBJECTIVES 1. Extend a mark-recapture monitoring scheme to estimate occupancy rates of swift foxes (Vulpes velox) on 12-mi2 quadrats in eastern Colorado. 2. Contrast estimates from the current survey with thoses obtained in 1998 and published in Finley et al. (2005). ABSTRACT A randomly selected sample of 15 ~12-mi2 grids in eastern Colorado were trapped with a 4 × 5 grid of traps between August, 2004 and February, 2005. Swift foxes were trapped on 36 of the 51 grids, with 136 total fox captures. Comparison of the estimates of the percent of 12-mi2 grids occupied by swift foxes in eastern Colorado does not appear to have changed since a comparable sample was taken of 72 grids in March, 1995 – January, 1997 (Finley et al. 2005). Using the average percentage of the grids in short grass prairie with the minimum AICc model, the earlier estimate was ψ̂ = 0.821 (SE 0.0659), compared to the current estimate of ψ̂ = 0.777 (SE = 0.0786). The estimated change is −0.044 (SE = 0.103, 95% CI −0.245 – 0.157). Summing the predicted occupancy values across the sampled grids for the respective studies provides a similar conclusion: Finley et al. (2005) found ψ̂ = 0.790 (SE = 0.0574), whereas this study found ψ̂ = 0.742 (SE = 0.0869), providing an estimate of the change of −0.048 (SE = 0.104, 95% CI −0.252 – 0.156). These differences are well within the sampling variation of the estimates, and do not suggest a decline in swift fox populations in eastern Colorado. RESULTS Sample of Grids Finley et al. (2005) found that the covariate percent Short Grass Prairie (SGP) is a good predictor of the presence of foxes in eastern Colorado. The distribution of this covariate is bimodal (Figure 1). To build the best relationship between SGP and fox numbers, we sampled across this continuum of SGP values. Thus, the 2,566 trapping grids considered in the sampling frame of grids (Figure 1) to be trapped were sorted by the percentage of SGP predicted by the CDOW GIS system. Then a random start between 1 and 66 was picked, and every 50th grid was selected to be sampled. This procedure resulted in a sample of 51 blocks. When I multiply the frequency of the sample by 50, I obtain a close relationship between the sampling frame and the grids sampled (Figure 2). Statistical Methods Analysis methods to estimate occupancy rates followed the procedures of Finley et al. (2005), using the occupancy model of MacKenzie et al. (2002) in Program MARK (White and Burnham 1999). I considered a set of a priori models that incorporated month as sine and cosine functions to model detection probabilities (p), and the percentage of short grass prairie on the trapping grid to model both 69 �detection probabilities and probability of occupancy, ψ (psi). Model selection was performed with information-theoretic methods following Burnham and Anderson (2002). Analysis methods to estimate the population of foxes using a trapping grid also followed Finley et al. (2005), using the Huggins estimator (Huggins 1989, 1991) to estimate population size. Model selection was performed with information-theoretic methods following Burnham and Anderson (2002). Occupancy Estimation Model selection results for occupancy estimation are shown in Table 1. The sine and cosine functions for month did not improve model fit of detection probabilities, nor did the percentage of short grass prairie improve estimates of detection probabilities. However the percentage of short grass prairie did provide an important predictor of occupancy (Figure 3) with the logit predictive equation: exp[βˆ 0 + βˆ 1 (SGP%)] Occupancy Probability = , 1+ exp[βˆ 0 + βˆ 1 (SGP%)] where β̂ 0 = -0.287 (SE = 0.624, 95% CI −1.510 – 0.936) and β̂1 = 2.775 (SE = 1.299, 95% CI 0.229 – 5.322). The estimated occupancy rate using the average amount of short grass prairie found on the 51 grids samples was ψ̂ = 0.777 (SE = 0.0786, 95% CI 0.589 – 0.894). When the estimated occupancy was summed across the 51 grids using the observed amount of short grass prairie on each grid, ψ̂ = 0.742 (SE = 0.0869, 95% CI 0.572 – 0.912). Finally, the entire population of grids from which the 51 sampled grids were drawn was used to compute the proportion of eastern Colorado occupied by swift foxes: ψ̂ = 0.711. The amount of short grass prairie for each of the grids in the population was estimated based on a GIS layer. Population Estimation Model selection results for population estimation (Table 2) suggest a behavioral effect in response to initial capture, with capture probabilities a function of month and SGP. Initial capture probabilities (Figure 4) and recapture probabilities (Figure 5) from the minimum AICc model are a function of month through a sin transformation, and SGP. The mean number of animals estimated per grid for all 51 grids was 4.83 (SE = 1.990, 95% CI 0.933 – 8.735), ranging from 0 to 26. DISCUSSION Simulations reported in Finley et al. (2005) reported expected power to detect declines given various combinations of numbers of trapping occasions and numbers of grids trapped. For 50 grids trapped with 3 occasions, their simulation results suggested a SE of about 0.070 for ψ = 0.8. The estimated SEs from this study are slightly greater then this value, likely because of the variation in SGP over the range of the sample. However, the values are close enough to make the simulation results reported in Finley et al. (2005) useful if taken a bit conservatively. The results from this study concerning the importance of SGP in predicting swift fox occupancy compared favorably with the results obtained by Finley et al. (2005) (Figure 6). Basically, the same relationship of SGP to occupancy was found. However, the minimum AICc model for occupancy in this study was much simpler than that of Finley et al. (2005), mainly because grids were trapped only during 70 �the period late August through March when the highest detection probabilities were expected based on Finley et al. (2005) work. When Finley et al. (2005) used the percentage of short grass prairie for each of their sampled grids to estimate a grid-specific ψ value, the sum of ψ̂ values was 56.9 (SE = 4.13), or 56.9 of the 72 ˆ = 0.790, SE = 0.0574). Alternatively, they estimated ψ of 0.821 grids actually contained foxes ( ψ (SE = 0.0659) using the mean (66.9%) of the short-grass prairie habitat for the 72 grids. In either case, their estimates are slightly greater than the values of ψ estimated in this study with the same approaches, but negligibly so when the uncertainty of the estimates is taken into account. As cautioned in Finley et al. (2005), the mean number of animals estimated per grid cannot be extrapolated to a population estimate for eastern Colorado because the grids attract foxes from some unknown distance outside the trapping grid. SUMMARY Comparison of the estimates of the percent of 12-mi2 grids occupied by swift foxes in eastern Colorado does not appear to have changed since a comparable sample was taken of 72 grids in March, 1995 – January, 1997 (Finley et al. 2005). Using the average percentage of the grids in short grass prairie with the minimum AICc model, the earlier estimate was ψ̂ = 0.821 (SE 0.0659), compared to the current estimate of ψ̂ = 0.777 (SE = 0.0786). The estimated change is −0.044 (SE = 0.103, 95% CI −0.245 – 0.157). Summing the predicted occupancy values across the sampled grids for the respective studies provides a similar conclusion: Finley et al. (2005) found ψ̂ = 0.790 (SE = 0.0574), whereas this study found ψ̂ = 0.742 (SE = 0.0869), providing an estimate of the change of −0.048 (SE = 0.104, 95% CI −0.252 – 0.156). These differences are well within the sampling variation of the estimates, and do not suggest a decline in swift fox populations in eastern Colorado. LITERATURE CITED FINLEY, D. J., G. C. WHITE AND J. P. FITZGERALD. 2005. Estimation of swift fox population size and occupancy rates in eastern Colorado. Journal of Wildlife Management. In Press. BURNHAM, K. P., AND D. R. ANDERSON. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd edition. Springer-Verlag, New York, New York, USA. HUGGINS, R. M. 1989. On the statistical analysis of capture-recapture experiments. Biometrika 76: 133– 140. ______________. 1991. Some practical aspects of a conditional likelihood approach to capture experiments. Biometrics 47: 725–732. MACKENZIE, D. I., J. D. NICHOLS, G. B. LACHMAN, S. DROEGE, J. A. ROYLE, AND C. A. LANGTIMM. 2002. Estimating site occupancy when detection probabilities are less than one. Ecology 83: 2248–2255. WHITE, G. C., AND K. P. BURNHAM. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 Supplement: 120–138. Prepared by: ________________________ Dr. Gary C. White, Department Fishery & Wildlife Conservation Biology Colorado State University 71 �Table 1. Occupancy model selection results for 51 swift fox grids trapped in eastern Colorado, August 2004 to February, 2005. AICc ∆AICc 196.785 198.882 198.891 199.133 200.412 201.176 201.242 201.522 203.449 0 2.0969 2.1065 2.3486 3.6277 4.3916 4.4573 4.7372 6.6642 Model {p(.) ψ(SGP)} {p(sinMonth) ψ(SGP)} {p(SGP) ψ(SGP)} {p(cosMonth) ψ(SGP)} {p(.) ψ(.)} {p(sinMonth+cosMonth) ψ(SGP)} {p(T) ψ(.)} {p(cosMonth+cosMonth^2) ψ(SGP)} {p(t) ψ(.)} AICc Model Weights Likelihood 0.39689 1 0.1391 0.3505 0.13843 0.3488 0.12265 0.309 0.0647 0.163 0.04416 0.1113 0.04273 0.1077 0.03715 0.0936 0.01418 0.0357 Num. Deviance Par 3 190.274 4 190.012 4 190.022 4 190.264 2 196.162 5 189.843 3 194.731 5 190.189 4 194.579 Table 2. Closed population estimator model selection results for 51 swift fox grids trapped in eastern Colorado, August 2004 to February, 2005. Model {p(SGP+sinMonth)= c(SGP+sinMonth)+additive effect} {p(SGP+sinMonth+ cosMonth)=c(SGP+ sinMonth+cosMonth)+additiv e effect} {p(SGP)=c(SGP)+ additive effect} {p(sinMonth)= c(sinMonth)+ additive effect} {p(cosMonth)= c(cosMonth)+ additive effect} {p(cosMonth+ sinMonth)= c(cosMonth+ sinMonth)+additive effect} {p(.) c(.)} {p(T)=c(T)} {p(.)=c(.)} {p(T)=c(T)+ additive effect} {p(t)=c(t)+ additive effect} {p(g*t)=c(g*t)} AICc Delta AICc AICc Weights Model Likelihood No. Par Deviance 331.785 0 0.4213 1 4 323.666 333.739 1.9538 0.15861 0.3765 5 323.56 334.006 2.2207 0.13879 0.3294 3 327.935 334.421 2.636 0.11277 0.2677 3 328.35 336.195 4.4094 0.04646 0.1103 3 330.124 336.322 337.474 337.658 338.619 339.211 339.84 537.81 4.5372 5.6892 5.8723 6.8334 7.4255 8.0543 206.024 0.04359 0.0245 0.02236 0.01383 0.01028 0.00751 0 0.1035 0.0582 0.0531 0.0328 0.0244 0.0178 0 4 2 2 1 3 4 108 328.204 333.439 333.622 336.607 333.14 331.721 220.762 72 �350 250 200 150 100 I I I I I I I I I I I I I I 90 -9 5 I 80 -8 5 I 70 -7 5 I 60 -6 5 I 20 -2 5 I 10 -1 5 05 0 - 50 -5 5 - 40 -4 5 50 30 -3 5 Frequency ` 300 Percent Short Grass Prairie Figure 1. Histogram of percentage of short grass prairie on 12-mi2 trapping grids comprising the sampling frame for this study. 350 ■ Frame □ Sample 250 200 150 - ~ 50 - - - ~ ,- - ,- - - ~ r - 50 -5 5 40 -4 5 30 -3 5 20 -2 5 10 -1 5 05 0 - - ,- - - I ~ ,- 90 -9 5 - 80 -8 5 ,- 70 -7 5 100 60 -6 5 Frequency ` 300 Percent Short Grass Prairie Figure 2. Histogram showing the close relationship between the grids included in the sample compared to the sampling frame. A representative sample relative to the availability of the SGP variable was selected. 73 �` Occupancy Probability - -- - --- 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 --- 0% 20% 40% 60% 80% 100% Short Grass Prairie (Percent) Figure 3. Prediction of the probability of occupancy with 95% confidence intervals as a function of the percentage of short grass prairie on the 12-mi2 trapping grid. Ticks on the 0 and 1 lines indicate the status of the 51 trapping grids, with 36 of the grids recording foxes captured. Capture Probability ` 0.6 0.5 0.4 - --- - - 0.3 0.2 - - --------- 0.1 0 0% 20% 40% 60% 80% 100% Short Grass Prairie (Percent) 1- p Sep p Jan -- p Oct p Feb -- p Nov - p Dec Figure 4. Changes in initial capture probability for swift fox trapped in eastern Colorado on 12-mi2 grids, August 2004 – February, 2005. 74 �Reapture Probability 0.25 0.2 0.15 0.1 _,. . - - - - -- - - - . -- - - - - 0.05 0 0% 20% 40% 60% 80% 100% Short Grass Prairie (Percent) c Sep c Jan c Oct c Feb c Nov c Dec Figure 5. Changes in recapture probability for swift fox trapped in eastern Colorado on 12-mi2 grids, August 2004 – February, 2005. - - - .. 1 .- Probability of occupancy 0.9 -. 0.8 0.7 0.6 - . -. - ----- 0.5 - - - - , - -- - ~ - - - - __ __ _ 11111111 _ • - - - - 0.4 # 0.3 - - - -~ - -# -# - - - -# - - - - - - - 0.2 -- 0.1 0 0 20 40 60 80 100 Short grass grairie (%) Figure 6. Effect of the percentage of the 12–mi2 grid consisting of short-grass prairie habitat on the probability of occupancy by swift foxes trapped on 72 grids in eastern Colorado, March, 1995 – January, 1997, for the top-ranked AICc model {p(T + cos(Month) + cos2(Month)) ψ (SGP Proportion)} from Finley et al. (2005). The dashed lines are 95% confidence intervals for the estimated probability of occupancy. Ticks across the 0 and 1 occupancy lines are the observed occupancy values plotted against the percentage of short grass prairie for the 72 grids, with short grass prairie values dithered so that grids would not plot on top of each other. 75 �Colorado Division of Wildlife July 2005 – June 2006 WILDLIFE RESEARCH REPORT State of Cost Center Work Package Task No. Colorado 3430 3001 5 Federal Aid Project: W-185-R : Division of Wildlife : Mammals Research : Deer Conservaton : Multispecies Investigations Consulting Services for Mark-Recapture Analysis : Period Covered: July 1, 2005 - June 30, 2006 Author: G. C. White Personnel: C. Bishop, D. J. Freddy, T. M. Shenk, L. Stevens, R. Kahn, F. Pusateri, E. O’Dell, D. Martin, P. Schnurr, K. Navo, B. Andelt, D. Finley, A. Linstrom, K. Strohm, P. Conn. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Progress towards meeting the objectives of this job include: 1. Consulting assistance to Colorado Division of Wildlife (CDOW) on harvest surveys, terrestrial inventory systems, and population modeling procedures was provided. Assistance with estimation of spring and fall turkey, spring snow goose, sharp-tailed and sage grouse, chukars, ptarmigan, Abert’s squirrels, and general small game harvest was provided, and programs and harvest estimates provided to CDOW via email and CD ROM. Computer code written in SAS to compute these estimates and display results graphically was also provided. Computer code was also written in SAS to estimate the compliance rate of Colorado small game license holders with the Harvest Information Program. 2. The CDOW DEAMAN software package for the storage, summary, and analysis of big game population and harvest data was revised further as a Windows XP program. A User’s Manual has been provided to terrestrial biologists via the WWW at http://www.cnr.colostate.edu/~gwhite/deaman. I met with the CDOW software group to discuss conversion of DEAMAN to a central server application. 3. Consultation with CDOW Terrestrial Biologists in the use of DEAMAN and population modeling procedures continued. Numerous questions were answered via meetings with biologists, and via email. 4. A paper comparing the population levels of swift foxes in eastern Colorado to a previous study in cooperation with CDOW was submitted to Southwestern Naturalist: Martin, D. J., G. C. White, and F. M. Pusateri. 2006. Monitoring swift fox populations in eastern Colorado. Southwestern Naturalist. Submitted. 5. A paper on the use of vaginal implant transmitters in cooperation with CDOW was submitted and accepted for publication in the Journal of Wildlife Management: Bishop, C. J., D. J. Freddy, G. C. 91 �White, B. E. Watkins, T. R. Stephenson, and L. L. Wolfe. 2006. Using vaginal implant transmitters to aid in capture of neonates from marked mule deer. Journal of Wildlife Management. In Press. 6. A paper resulting from Dan Walsh’s M.S. project in cooperation with CDOW was submitted to Ecological Applications: Walsh, D. P., G. C. White, T. E. Remington, and D. C. Bowden. 2006. Population Estimation of Greater Sage-Grouse. Ecological Applications. Submitted. 7. A paper resulting from Sherri Huwer’s M.S. project in cooperation with CDOW was submitted to the Journal of Wildlife Management: Huwer, S. L., D. R. Anderson, T. E. Remington, and G. C. White. 2006. Evaluating the importance of forbs to sage-grouse using human-imprinted chicks. Journal of Wildlife Management. Submitted. 8. A paper on mountain sheep populations in Rocky Mountain National Park was submitted and accepted for publication in the Wildlife Society Bulletin: McClintock, B. T., and G. C. White. 2006. Bighorn sheep abundance following a suspected pneumonia epidemic in Rocky Mountain National Park. Wildlife Society Bulletin. In Press. 9. A paper on extending the mark-resight estimator using a beta-binomial distribution was submitted and accepted in the Journal of Agricultural, Biological, and Ecological Statistics: McClintock, B. T., G. C. White, and K. P. Burnham. 2006. A robust design mark-resight abundance estimator allowing heterogeneity in resighting probabilities. Journal of Agricultural, Biological, and Ecological Statistics. In Press. 10. A paper resulting from the May, 2005 Elk and Deer Workshop was submitted and accepted for publication in the Wildlife Society Bulleting: Mason, J. R., L. H. Carpenter, M. Cox, J. C. Devos, J. Fairchild, D.J. Freddy, J. R. Heffelfinger, R. H. Kahn, S. M. McCorquodale, D. F. Pac, D. Summers, G. C. White, and B. K. Williams. 2006. A case for standardized ungulate surveys and data management in the western United States. Wildlife Society Bulletin. In Press. 11. A paper describing the use of closed captures models to estimate population size with Program MARK was submitted and accepted for publication in Environmental and Ecological Statistics: White, G. C. 2006. Closed population estimation models and their extensions in program MARK. Environmental and Ecological Statistics. In Press. 12. A paper on the application of multistate models in Program MARK was submitted and accepted for publication in the Journal of Wildlife Management: White, G. C., W. L. Kendall, and R. J. Barker. 2006. Multistate survival models and their extensions in program MARK. Journal of Wildlife Management. In Press. 13. A paper on the estimation of female grizzly bears was submitted to the Journal of Agricultural, Biological, and Ecological Statistics: Cherry, S., G. C. White, K. A. Keating, M. A. Haroldson, C. C. Schwartz. 2006. Evaluating estimators of the numbers of females with cubs-of-the-year in the Yellowstone grizzly bear population. Journal of Agricultural, Biological, and Ecological Statistics. Submitted. 14. A paper on the survival of mule deer in the Bridger Mountains, Montana, was submitted and accepted for publication in the Journal of Wildlife Management: Pac, D. F., and G. C. White. 2006. Survival and cause-specific mortality of mule deer in the Bridger Mountains, Montana. Journal of Wildlife Management. In Press. 92 �15. A paper on the impact of limited antlered harvest on mule deer sex and age ratios in cooperation with CDOW was published in the Wildlife Society Bulletin: Bishop, C. J., G. C. White, D. J. Freddy, and B. E. Watkins. 2005. Effect of limited antlered harvest on mule deer sex and age ratios. Wildlife Society Bulletin 33: 662–668. 16. A paper on estimation of nest survival was submitted and accepted for publication in Studies in Avian Biology: Heisey, D. M., T. L. Shaffer, and G. C. White. 2006. The ABCs of nest survival: theory and application from a biostatistical perspective. Studies in Avian Biology. In Press. 17. A paper on the estimation of the area of black-tailed prairie dog colonies in eastern Colorado in cooperation with CDOW was published in the Wildlife Society Bulletin: White, G. C., J. R. Dennis, and F. M. Pusateri. 2005. Area of black-tailed prairie dog colonies in eastern Colorado. Wildlife Society Bulletin 33:265–272. 18. A paper in response to a critique by Sterling Miller was published in the Wildlife Society Bulletin in cooperation with CDOW: White, G. C., J. R. Dennis, and F. M. Pusateri. 2005. Response to: Overestimation bias in estimate of black-tailed prairie dog abundance in Colorado. Wildlife Society Bulletin 33:1452–1455. 19. A paper on methodologies to obtain more rigorous population monitoring data was published in Wildlife Research: White, G. C. 2005. Correcting wildlife counts with detection probabilities. Wildlife Research 32:211–216. 20. A paper on the procedures to monitor swift fox populations in eastern Colorado was published in the Journal of Wildlife Management: Finley, D. J., G. C. White and J. P. Fitzgerald. 2005. Estimation of swift fox population size and occupancy rates in eastern Colorado. Journal of Wildlife Management 69:861–873. 21. A research study to examine the impact of nutrition on the decline of mule deer fecundity during the last 20 years was continued in cooperation with Chad Bishop and CDOW. Portions of this work will serve as his doctoral dissertation in addition to his full-time duties as a researcher with CDOW. 22. A graduate research project (M. S.) was continued in cooperation with CDOW to evaluate line transect methodology for estimating pronghorn populations in eastern Colorado. The graduate student is Aaron Linstrom, and the project is in addition to his full-time duties as a biologist with CDOW. 23. A graduate research project (Ph. D.) in cooperation with CDOW to develop statistical models to monitor puma and black bear populations in Colorado based on checks of harvested animals and DNA and/or radio-tracking data was continued. The graduate student is Paul Conn. 24. A graduate research project (M. S.) in cooperation with CDOW to evaluate methods of redistributing elk in and around Great Sand Dunes National Park was started and then discontinued. The student, Greg Davidson, switched his work to evaluate habitat use by elk on the Grand Mesa. 25. Development of the design of a monitoring system for white-tailed prairie dogs in western Colorado and eastern Utah was continued in cooperation with CDOW with P. Schnurr, K. Navo and B. Andelt. 26. Design of a monitoring system for black-tailed prairie dogs in eastern Colorado in cooperation with CDOW was continued. This effort is in cooperation with Francie Pusateri and Eric O’Dell of CDOW. 93 �WILDLIFE RESEARCH REPORT CONSULTING SERVICES FOR MARK-RECAPTURE ANALYSES GARY C. WHITE P. N. OBJECTIVE Provide expert biostatistical and experimental design services to the Colorado Division of Wildlife, Wildlife Programs Branch. SEGMENT OBJECTIVES 1. Provide biostatisitical support to implement and analyze CDOW hunter harvest surveys. 2. Provide professional oversight, critiques, and analytical support to CDOW terrestrial management and avian and mammals research sections. 3. Convey to CDOW research and management sections new and pertinent information obtained in various collaborative projects conducted with other agencies and entities. RESULTS, DISCUSSION, SUMMARY See ABSTRACT for summary of key activities and publications. Prepared by: Dr. Gary C. White, Department of Fish, Wildlife, and Conservation Biology Colorado State University 94 �Colorado Division of Wildlife July 2006 – June 2007 WILDLIFE RESEARCH REPORT State of Cost Center Work Package Task No. Colorado 3430 3001 5 Federal Aid Project: W-185-R : Division of Wildlife : Mammals Research : Deer Conservation : Multispecies Investigations Consulting Services for Mark-Recapture Analysis : Period Covered: July 1, 2006 - June 30, 2007 Author: G. C. White Personnel: C. Bishop, D. J. Freddy, T. M. Shenk, P. Lukacs, R. Kahn, F. Pusateri, E. O’Dell, D. Martin, P. Schnurr, K. Navo, B. Andelt, A. Linstrom, P. Conn, B. McClintock, G. Davidson, and J. Ivan. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Progress towards meeting the objectives of this job include: 1. Consulting assistance to Colorado Division of Wildlife (CDOW) on harvest surveys, terrestrial inventory systems, and population modeling procedures was provided. Computer code written in SAS to compute these estimates and display results graphically was provided. Specific input involved estimation of variances for design of the E-2 elk aerial survey. 2. Support for the CDOW DEAMAN software package for the storage, summary, and analysis of big game population and harvest data was provided. I met with the CDOW software group to discuss conversion of DEAMAN to a central server application. 3. Consultation with CDOW Terrestrial Biologists in the use of DEAMAN and population modeling procedures continued. Numerous questions were answered via meetings with biologists, and via email. A workshop on modeling Colorado’s deer, elk, and antelope populations was conducted for biologist in Glennwood Springs, August, 2006. 4. A paper comparing the population levels of swift foxes in eastern Colorado to a previous study in cooperation with CDOW was accepted for publication in Southwestern Naturalist: Martin, D. J., G. C. White, and F. M. Pusateri. 2007. Monitoring swift fox populations in eastern Colorado. Southwestern Naturalist. In Press. 5. A paper on estimation of abundance and demography using age-at-harvest and mark-recovery data was submitted to Environmental and Ecological Statistics: Conn, P. B., G. C. White, and J. L. Laake. 2007. Estimating abundance and demography using age-at-harvest and mark-recovery data: a Bayesian approach. Environmental and Ecological Statistics. Submitted. 97 �6. A paper on Bayesian methods to analyze age-at-harvest data was submitted to Biometrics: Conn, P. B., J. L. Laake, D. R. Diefenbach, G. C. White, and M. A. Ternent. 2007. Bayesian analysis of wildlife age-at-harvest data. Biometrics. Submitted. 7. A paper on the use of vaginal implant transmitters in cooperation with CDOW was submitted and accepted for publication in the Journal of Wildlife Management: Bishop, C. J., D. J. Freddy, G. C. White, B. E. Watkins, T. R. Stephenson, and L. L. Wolfe. 2007. Using vaginal implant transmitters to aid in capture of neonates from marked mule deer. Journal of Wildlife Management 71:945–954. 8. A paper resulting from collaboration with Montana colleagues resulted in a publication in the Journal of Wildlife Management: Pac, D. F., and G. C. White. 2007. Survival and cause-specific mortality of male mule deer under different hunting regulations in the Bridger Mountains, Montana. Journal of Wildlife Management 71:816–827. 9. A paper resulting from Sherri Huwer’s M.S. project in cooperation with CDOW was accepted for publication in the Journal of Wildlife Management: Huwer, S. L., D. R. Anderson, T. E. Remington, and G. C. White. 2007. Evaluating the importance of forbs to sage-grouse using human-imprinted chicks. Journal of Wildlife Management. In Press. 10. A paper on mountain sheep populations in Rocky Mountain National Park was published in the Journal of Wildlife Management: McClintock, B. T., and G. C. White. 2007. Bighorn sheep abundance following a suspected pneumonia epidemic in Rocky Mountain National Park. Journal of Wildlife Management 71:183–189. 11. A paper on extending the mark-resight estimator using a beta-binomial distribution was published in the Journal of Agricultural, Biological, and Ecological Statistics: McClintock, B. T., G. C. White, and K. P. Burnham. 2006. A robust design mark-resight abundance estimator allowing heterogeneity in resighting probabilities. Journal of Agricultural, Biological, and Ecological Statistics 11:231–248. 12. A paper resulting from the May, 2005 Elk and Deer Workshop was published in the Wildlife Society Bulletin: Mason, J. R., L. H. Carpenter, M. Cox, J. C. Devos, J. Fairchild, D.J. Freddy, J. R. Heffelfinger, R. H. Kahn, S. M. McCorquodale, D. F. Pac, D. Summers, G. C. White, and B. K. Williams. 2006. A case for standardized ungulate surveys and data management in the western United States. Wildlife Society Bulletin 34:1238–1242. 13. A paper describing the use of closed captures models to estimate population size with Program MARK remains in press in Environmental and Ecological Statistics: White, G. C. 2007. Closed population estimation models and their extensions in program MARK. Environmental and Ecological Statistics. In Press. 14. A paper on the application of multistate models in Program MARK was published in the Journal of Wildlife Management: White, G. C., W. L. Kendall, and R. J. Barker. 2006. Multistate survival models and their extensions in program MARK. Journal of Wildlife Management 70:1521–1529. 15. A paper on the estimation of female grizzly bears was published in the Journal of Agricultural, Biological, and Ecological Statistics: Cherry, S., G. C. White, K. A. Keating, M. A. Haroldson, C. C. Schwartz. 2007. Evaluating estimators of the numbers of females with cubs-of-the-year in the Yellowstone grizzly bear population. Journal of Agricultural, Biological, and Ecological Statistics 12:195–215. 98 �16. A paper on estimation of nest survival was published in Studies in Avian Biology: Heisey, D. M., T. L. Shaffer, and G. C. White. 2007. The ABCs of nest survival: theory and application from a biostatistical perspective. Studies in Avian Biology 34:13–33. 17. A paper on extending the mark-resight population estimation method was submitted to Environmental and Ecological Statistics: McClintock, B.T., G. C. White, K. P. Burnham, and M. A. Pryde. 2007. A robust design mixed effects mark-resight model for estimating abundance when sampling is without replacement. Environmental and Ecological Statistics. Submitted. 18. A paper on estimation of random effects with Bayesian methods was submitted to Environmental and Ecological Statistics: White, G. C., K. P. Burnham, and R. J. Barker. 2007. Evaluation of some Bayesian MCMC random effects inference methodology applicable to bird ringing data. Environmental and Ecological Statistics. Submitted. 19. A paper on analysis of small count data was submitted to Condor: McDonald, T. L., and G. C. White. 2007. A comparison of regression models for small counts. Condor. Submitted. 20. A paper on the impact of previous capture on sampling probabilities with DNA hair-snag grids for grizzly bear populations was submitted to the Journal of Wildlife Management: Boulanger, J., and G. C. White. 2007. Influence of past live captures on detection probabilities of grizzly bears using DNA hair snagging methods. Journal of Wildlife Management. Submitted. 21. A paper on detecting trends in the Yellowstone grizzly bear population was submitted to Ursus: Harris, R. L., G. C. White, C. C. Schwartz, and M. A. Haroldson. 2007. Population growth of Yellowstone grizzly bears: uncertainty and future monitoring. Ursus. Submitted. 22. A graduate research project (Ph. D.) in cooperation with CDOW to develop statistical models to monitor puma and black bear populations in Colorado based on checks of harvested animals and DNA and/or radio-tracking data was completed. The graduate student is Paul B. Conn. The dissertation is: Conn, P. B. 2007. Bayesian Analysis of Age-at-Harvest Data with Focus on Wildlife Monitoring Programs. Ph. D. Dissertation, Colorado State University, Fort Collins. 184 pp. 23. A research study to examine the impact of nutrition on the decline of mule deer fecundity during the last 20 years was continued in cooperation with Chad Bishop and CDOW. Portions of this work will serve as his doctoral dissertation in addition to his full-time duties as a researcher with CDOW. 24. A graduate research project (M. S.) was continued in cooperation with CDOW to evaluate line transect methodology for estimating pronghorn populations in eastern Colorado. The graduate student is Aaron Linstrom, and the project is in addition to his full-time duties as a biologist with CDOW. 25. A graduate research project (M. S.) in cooperation with CDOW to evaluate methods of redistributing elk in and around Great Sand Dunes National Park was started and then discontinued. The student, Greg Davidson, switched his work to evaluate habitat use by elk on the Grand Mesa. A report on the San Luis Valley elk work is nearly completed. 26. A graduate research project (Ph.D.) in cooperation with CDOW to evaluate snowshoe hare densities relative to lodge pole pine and mixed conifer habitats was continued. The graduate student is Jake Ivan. 27. Development of the design of a monitoring system for white-tailed prairie dogs in western Colorado and eastern Utah was continued in cooperation with CDOW with P. Schnurr, K. Navo and B. Andelt. 99 �A final draft of a manuscript on the use of occupancy monitoring for white-tailed and Gunnison prairie dogs was given to B. Andelt for submission to the Journal of Wildlife Management on 20 February, 2007. 28. Preliminary analysis of monitoring data on black-tailed prairie dogs in eastern Colorado in cooperation with CDOW was continued. This effort is in cooperation with Francie Pusateri and Eric O’Dell of CDOW. 100 �WILDLIFE RESEARCH REPORT CONSULTING SERVICES FOR MARK-RECAPTURE ANALYSES GARY C. WHITE P. N. OBJECTIVE Provide expert biostatistical and experimental design services to the Colorado Division of Wildlife, Wildlife Programs Branch. SEGMENT OBJECTIVES 1. Provide biostatistical support to implement and analyze CDOW hunter harvest surveys. 2. Provide professional oversight, critiques, and analytical support to CDOW terrestrial management and avian and mammals research sections. 3. Convey to CDOW research and management sections new and pertinent information obtained in various collaborative projects conducted with other agencies and entities. RESULTS, DISCUSSION, SUMMARY See ABSTRACT for summary of key activities and publications. Prepared by: ___________________________________ Dr. Gary C. White, Department of Fish, Wildlife, and Conservation Biology Colorado State University 101 � https://cpw.cvlcollections.org/files/original/9ac467a63f9bd69b3c8ddb51c0416381.pdf b9da7ec4817212b47fd148dabe2fefee PDF Text Text Colorado Division of Wildlife July 2005 – June 2006 WILDLIFE RESEARCH REPORT State of Cost Center Work Package Colorado 3430 3001 Federal Aid Project W-185-R : Division of Wildlife : Mammals Research : Deer Conservation Program Final Report Deer Conservation Research For 5-Year Federal Aid Grant W-185-R July 2001 – June 2006 : Period Covered: July 1, 2001 – June 30, 2006 Author: David J. Freddy, Mammals Research Leader, 1 June 2006 Principal Investigators: D. L. Baker, C. J. Bishop, E. J. Bergman, D. J. Freddy, and T. M. Pojar, Colorado Division of Wildlife; W. F. Andelt, N. T. Hobbs, and G. C. White, Colorado State University All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT This report highlights the accomplishments of mule deer research and associated activities conducted by the Colorado Division of Wildlife (CDOW) with the funding support of Federal Aid Grant W-185-R during the 5-year grant segment, July 2001-June 2006. Five major multi-year research projects addressing mule deer population limiting factors, habitat status, and habitat enhancements were designed, implemented, completed, and reported upon during this segment in response to addressing stakeholder interests that influenced the direction of mule deer management and research beginning in the late 1990s. Additionally, funding provided critical scientific and technical expertise quality control oversight for statewide deer hunter harvest surveys, statewide deer population databases, mule deer survival and population estimate management surveys, mule deer population modeling, and mule deer research projects. Funding also partially supported research projects addressing chronic wasting disease and fertility control in mule deer. Research experiments provided strong evidence that habitat nutritional quality had a greater impact on net productivity of mule deer than did existing levels of coyote, cougar, and black bear predation and therefore, future research and management efforts should focus on improving the nutritional capabilities of senescent pinyon-juniper winter ranges for deer. Research also provided strong evidence that the timing and rate of breeding were within normal ranges for mule deer and therefore concerns about the breeding cycle could be dismissed as a major contributor to declining performance of mule deer populations. Comparative assessments of vegetation inside and outside sagebrush and mountain brush exclosures indicated that after 40 to 50 years of protection from ungulate herbivory, woody species increased in cover dominance with only minor changes in herbaceous cover. Increasing plant species diversity in these types of winter ranges will probably not be accomplished by excluding herbivory. In a highly scrutinized public experiment, research and management expertise codemonstrated that methods used by Colorado to estimate mule deer population size and to develop 47 �population management models provided reliable information to adequately guide mule deer harvest management decisions. From activities supported by this Grant during this segment, principal investigators published 15 peer-reviewed scientific articles pertaining to mule deer for prominent wildlife research journals with an additional 4 manuscripts currently in review with journals, provided 18 annual CDOW Wildlife Research Reports summarizing yearly progress of projects, and provided 13 presentations at prominent professional workshops or symposia. The cumulative impact of this programmatic effort provides Colorado the basis to progress and proactively sustain the mule deer resource in an increasingly demanding and complex landscape, social, and political environment. The relative success of mule deer management in Colorado reflects the positive synergy between the terrestrial research and management sections in sharing expertise, financial resources, manpower, and common goals. 48 �WILDLIFE RESEARCH REPORT PROGRAM FINAL REPORTDEER CONSERVATION RESEARCH FOR 5-YEAR FEDERAL AID GRANT W-185-R JULY 2001 – JUNE 2006 DAVID J. FREDDY Mammals Research Leader PROGRAM NEED During the late 1990s, the Colorado Division of Wildlife (CDOW) was challenged by sportsmen and other stakeholders to investigate potential causes of declining numbers of mule deer in Colorado. Additionally, sportsmen were critical of methods used to estimate numbers of mule deer and subsequently did not trust the CDOWs assessment of the overall status of mule deer in Colorado. The concerns of stakeholders gained the attention of the Colorado Legislature which directed CDOW to prepare a document to address causes of the mule deer decline and outline a plan of action to reverse the perceived trend in mule deer populations. That document was prepared for the legislature in 1999 (Gill et al. 2001) and established the direction and objectives for mule deer management and research beginning in 1999. Research objectives and program implementation were outlined and initiated in 1999 with most of the research effort directed at the Uncompahgre Plateau mule deer population which was of high concern to various stakeholders. This Federal Aid Grant Final Report highlights the accomplishments of the research pertaining to the mule deer program that was conducted from July 1, 2001 through June 30, 2006 and wholly or partially supported by Federal Aid Grant funds. PROGRAM NARRATIVE OBJECTIVES The primary Program Narrative research objectives were: I. Identify factors limiting the growth of mule deer populations. II. Assess methods to reduce impacts of limiting factors. III. Improve and evaluate statewide systems and technical methods used to determine status of mule deer populations. IV. Assess the impacts of chronic wasting disease on mule deer populations. V. Develop alternative approaches to control over-abundant urban-exurban mule deer populations. RESULTS Objective I. Factors Limiting Growth of Mule Deer Populations. Initially, stakeholders expressed concern that statewide declines in mule deer populations were caused by low pregnancy rates in adult females due to inadequate numbers of mature bucks to breed females, and by low recruitment of neonatal fawns due to excessive predation on neonates. Two primary projects were funded to focus on: 1) estimating pregnancy and fetal rates in adult female mule deer; and, 2) estimating survival rates of neonate fawn mule deer. 49 �Result Highlights: • Pregnancy and fetal rates were determined with ultrasonography and PSPB blood values to be within normal limits for the Poudre River and Uncompahgre Plateau mule deer in 1998 and 1999. Therefore, numbers of mature mule deer bucks were adequate to assure acceptable rates and timing of breeding for adult female deer. • Survival of radio-collared neonatal fawns from birth in June to December averaged 0.50 during 3 years from 1999 through 2001. This rate of neonate survival was only marginally adequate to assure population growth. Primary cause of death in neonates was sick/starve implicating inadequate nutrition for either adult does or neonates. Predation on neonates by canids, ursids, and felids occurred but not at rates considered to be limiting the population. Coyotes were the primary predator accounting for about 13% of the neonate deaths. Resulting Peer-Reviewed Publications: ANDELT, W.F., T.M. POJAR, AND L.W. JOHNSON. 2004. Long-term trends in mule deer pregnancy and fetal rates in Colorado. Journal of Wildlife Management 68:542-549. POJAR, T.M., AND D.C. BOWDEN. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550-560. Associated Annual Wildlife Research Progress Reports Available from the Colorado Division of Wildlife Research Library, Fort Collins, Colorado: POJAR, T.M., AND D.C. BOWDEN. 2002. Mule deer life-cycle-neonatal fawn survival. Colorado Division of Wildlife, Wildlife Research Report July: 47-63. POJAR, T.M. 2003. Mule deer life-cycle-neonatal fawn survival. Colorado Division of Wildlife, Wildlife Research Report July: 55. Objective II. Assess Methods to Reduce Impacts of Limiting Factors. A widespread debate throughout the western states in the late 1990s was whether mule deer were declining primarily due to predation from perceived abundant coyote, cougar, and black bear populations or if the decline was due to long-term losses in habitat quality and availability which negatively affected mule deer nutrition and subsequent recruitment and survival. Both predation and habitat quality were judged by various stakeholders to be the ‘cause’ of declining mule deer in Colorado and specifically the Uncompahgre deer population. Although painting the picture that the mule deer decline was caused by one major factor versus another major factor oversimplified the situation, such a dichotomy of thought quickly helped focus thrusts for potential research and management actions. The case for predation was assessed by Ballard et al. (2001) in their influential overview of predation and deer populations. The potential effects of habitat deterioration resulting from successional senescence of important plant communities and direct losses of habitat space due to human encroachment was argued by deVos, Jr. et al. (2003) in their overview of mule deer conservation strategies. As this debate evolved, Colorado was fortunate to have developed a strong working relationship with the Idaho Department of Fish and Game in our mutual attempts to address causes of the mule deer decline. The research sections of these 2 agencies decided to cooperatively investigate whether predation or habitat was the cause of the mule deer decline. Idaho, because of political, social, and legal aspects, was more capable of addressing the impacts of predation on mule deer than was Colorado and therefore, Idaho designed and implemented an intensive experimental reduction of coyote and cougar populations to measure the impacts of such actions on mule deer net recruitment (Hurley et al. 2002, Hurley et al. 2005). To compliment Idaho’s efforts, Colorado designed and implemented a series of experiments to measure the impacts of improving the nutritional quality of habitats on mule deer net recruitment (Bishop and White 2000). 50 �Three primary projects were funded to focus on: Phase 1A) effect of enhanced nutrition on mule deer population parameters; Phase 1B) long-term effects of herbivory on sagebrush and mountain brush winter ranges; and Phase 2A) effects of landscape habitat alterations within senescent old-growth pinyonjuniper winter ranges to enhance mule deer population parameters. Result Highlights: Phase 1A • Survival of fawns receiving an enhanced nutrition treatment from December through April had an over-winter survival rate of 0.89 which was higher (P < 0.001) than the survival rate of 0.65 for control fawns not receiving enhanced nutrition. The over-winter survival period was 15 December to 15 June during 3 years, 2001-02, 2002-03, 2003-04, and survival rates were based on 240 6-month old fawns with 120 fawns captured and radio-collared in each of the control and treatment areas. The effect of enhanced nutrition was highly evident even with the presence of ongoing predation by coyotes and cougars. • Survival rates of fetuses to neonate and through 1-year of age that were born to adult females receiving enhanced winter nutrition were 0.46 and higher (P < 0.001) than survival rate of 0.28 for fetuses born to adult females not receiving enhanced nutrition. Survival rates were over 3years, 2002-2004, and based on 276 fawns monitored across 293 adult females that were radiocollared of which, 154 adults received vaginal implant transmitters to aid in capture and monitoring neonate survival. Ultrasonography was used to determine in-utero fetal rates. • Body condition on about 1 March, as estimated from percent body fat and depth of longissimus dorsi muscle via ultrasonographpy, was better (P < 0.001) in adult females receiving enhanced winter nutrition (n = 78) than for control adult females not receiving enhanced nutrition (n= 76). Serum thyroid hormone levels were also higher in adult females receiving enhanced nutrition compared to control adult females not receiving better nutrition. Pregnancy and fetal rates were similar (0.95 and 1.80 fetuses per female) for adult females receiving and not receiving enhanced nutrition. • The finite rate of population increase, λ, was 1.20 for those deer receiving enhanced nutrition. For those deer not receiving enhanced nutrition, the finite rate of increase was 1.04 indicating a stable or slightly increasing population. The nutrition enhancement therefore, had a dramatic effect on deer population performance, indicating habitat quality was ultimately limiting the population that was also subject to natural levels of predation. In comparison, intensive control of coyote and cougar populations in Idaho had marginal positive impacts on survival rates of neonate fawns, 6-month old fawns, and adult mule deer and ultimately, net population growth (Hurley et al. 2005). • These enhanced nutrition experimental results provided a foundation for focusing deer management efforts on improving habitat quality in Colorado’s pinyon-juniper mule deer winter ranges rather than trying to intensively control or reduce coyote and/or cougar populations. • Phase 1B Excluding herbivory from semi-arid sagebrush and mountain brush plant communities resulted in increased dominance by shrub species and only minor changes in herbaceous species in nongrazed compared to adjacent grazed areas. Comparisons were based on measurements made in summer 2000 at 17 permanently fenced exclosures in western Colorado where ungulate herbivory was excluded for 40 to 50 years. Improving herbaceous and overall species diversity within established shrub dominated habitats will not likely occur by excluding grazing. 51 �• • Phase 2A Evaluating the effects of landscape alterations within senescent old-growth pinyon-juniper winter ranges on mule deer population performance parameters was initiated in 2004-05 as a pilot study and precursor to full-scale study implementation. Over-winter fawn survival was estimated on 2 critical pinyon-juniper winter range habitat treatment evaluation areas on the Uncompahgre Plateau in 2004-05. Both areas were found to be logistically adequate for future work and fawn survival was 0.84 to 0.96 on both sites (n = 25 radio-collared fawns per site). A project study plan for evaluating landscape habitat treatments was completed in 2005. Fullscale 4-year implementation began during winter 2005-06 as over-winter fawn survival, adult female body condition, and mule deer density were estimated among 8 habitat treatment evaluation areas (each 10-20 km2 in size) on the Uncompahgre Plateau. Pinyon-juniper habitat areas were categorized as controls (non-treated and senescent), pre-treatmeant (treated to reduce density of pinyon-juniper during last 10-15 years), and treatment (receiving additional habitat enhancements during this study). Initial survival rate estimates ranging from 0.76 to 0.88 suggest over-winter fawn survival may vary among habitat treatment levels. Estimates of deer density reaffirmed that deer densities in the northern study areas were lower (4-8 deer/km2) than densities in the southern study areas (19-57 deer/km2). Continued estimation of deer performance parameters over the next 3 years should allow detecting whether altering senescent pinyonjuniper habitats improves mule deer net productivity. Resulting Peer-Reviewed Publications: BISHOP, C.J., G.C. WHITE, D.J. FREDDY, AND B.E. WATKINS. 2003. How habitat quality affects hunting. Mule Deer 8:18-20. BISHOP, C.J., D.J. FREDDY, G.C. WHITE, B.E. WATKINS, T.R. STEPHENSON, AND L.L. WOLFE. 2006 In Review. Using vaginal implant transmitters to aid in capture of neonates from marked mule deer. Journal of Wildlife Management. MANIER, D.J., AND N.T. HOBBS. 2006. Large herbivores influence the composition and diversity of shrub-steppe communities in the Rock Mountains, USA. Oecologia 146:641-651. SCHULTHEISS, P.C., H. VAN CAMPEN, C.J. BISHOP, L.L. WOLFE, AND B. PODELL. 2006 In Review. Malignant catarrhal fever associated with ovine herpesvirus-2 in free-ranging mule deer (Odocoileus hemionus) in Colorado. Journal of Wildlife Disease. Associated Annual Wildlife Research Progress Reports Available from the Colorado Division of Wildlife Research Library, Fort Collins, Colorado: BERGMAN, E.J., C.J. BISHOP, D.J. FREDDY, AND G.C. WHITE. 2005. Pilot evaluation of winter range habitat treatments on over-winter mule deer fawn survival. Colorado Division of Wildlife, Wildlife Research Report July: 24-35. BERGMAN, E.J., C.J. BISHOP, D.J. FREDDY, AND G.C. WHITE. 2006 In Press. Evaluation of winter range habitat treatments on over-winter mule deer fawn survival. Colorado Division of Wildlife, Wildlife Research Report July: Available September 2006. BISHOP, C.J. AND G.C. WHITE. 2002. Effect of nutrition and habitat enhancements on mule deer recruitment and survival rates. Colorado Division of Wildlife, Wildlife Research Report July: 65-79. BISHOP, C.J., D.J. FREDDY, AND G.C. WHITE. 2002. Pilot study: use of ultrasound and vaginal implants. Colorado Division of Wildlife, Wildlife Research Report July: 81-92. BISHOP, C.J. G.C. WHITE, D.J. FREDDY, AND B.E. WATKINS. 2003. Effect of nutrition and habitat enhancements on mule deer recruitment and survival rates. Colorado Division of Wildlife, Wildlife Research Report July: 33-54. 52 �BISHOP, C.J. G.C. WHITE, D.J. FREDDY, AND B.E. WATKINS. 2004. Effect of nutrition and habitat enhancements on mule deer recruitment and survival rates. Colorado Division of Wildlife, Wildlife Research Report July: 21-43. BISHOP, C.J. G.C. WHITE, D.J. FREDDY, AND B.E. WATKINS. 2004. Effect of nutrition and habitat enhancements on mule deer recruitment and survival rates. Colorado Division of Wildlife, Wildlife Research Report July: 21-43. BISHOP, C.J. G.C. WHITE, D.J. FREDDY, AND B.E. WATKINS. 2005. Effect of nutrition and habitat enhancements on mule deer recruitment and survival rates. Colorado Division of Wildlife, Wildlife Research Report July: 37-65. BISHOP, C.J. G.C. WHITE, D.J. FREDDY, AND B.E. WATKINS. 2006 In Press. Effect of nutrition and habitat enhancements on mule deer recruitment and survival rates. Colorado Division of Wildlife, Wildlife Research Report July: Available September 2006. Associated Presentations at Professional Workshops/Symposia: BISHOP, C.J., G.C. WHITE, D.J. FREDDY, AND B.E. WATKINS. 2001. Effects of nutrition and habitat enhancements on mule deer fawn recruitment: preliminary results. Fourth Western States and Provinces Deer and Elk Workshop, August 1-3, Wilsonville, Oregon, USA. BISHOP, C.J., G.C. WHITE, D.J. FREDDY, AND B.E. WATKINS. 2003. Effects of enhanced winter nutrition of free-ranging mule deer on fawn recruitment and recruitment. Fifth Western States and Provinces Deer and Elk Workshop, May 21-24, Jackson, Wyoming, USA. BISHOP, C. J., G. C. WHITE, D. J. FREDDY, AND B. E. WATKINS. 2004. The effect of habitat quality on mule deer fawn survival and recruitment: interim report. Society for Range Management 57th Annual Meeting, January 24−30, Salt Lake City, Utah, USA. BISHOP, C. J., G. C. WHITE, D. J. FREDDY, AND B. E. WATKINS. 2004. Effect of enhanced nutrition of free-ranging mule deer on fawn survival and recruitment rates. The Wildlife Society 11th Annual Conference, September 18−22, Calgary, Alberta, Canada. BISHOP, C. J., G. C. WHITE, D. J. FREDDY, AND B. E. WATKINS. 2005. Effect of enhanced nutrition of free-ranging mule deer on population performance. Sixth Western States and Provinces Deer and Elk Workshop, May 16−18, Reno, Nevada, USA. BISHOP, C. J., G. C. WHITE, D. J. FREDDY, AND B. E. WATKINS. 2005. Effect of enhanced nutrition on mule deer population performance in pinyon-juniper habitat. Ecology and Management of Pinyon-Juniper and Sagebrush Communities Workshop, May 16−19, Montrose, Colorado, USA. III. Improve and Evaluate Statewide Systems and Technical Methods Used to Determine Status of Mule Deer Populations. Monitoring the status of mule deer in Colorado has advanced due to the synergy of the research section developing population monitoring systems and terrestrial management section implementing those monitoring systems as appropriate on a statewide basis. Developing, implementing, and maintaining, statistically rigorous systems to estimate statewide hunter harvest of mule deer, population densities and size for selected deer populations, adult female and fawn survival rates for selected populations, and developing future research projects requires scientific input, oversight and periodic evaluations. Additionally, proper evaluation requires a rigorously maintained and updated database containing statewide mule deer population data. As part of a multi-functional quality control process, the CDOW obtains oversight of key statewide mule deer research and management activities via a contract to a qualified individual through this Federal Aid Grant. Result Highlights: 53 �• Provided annual assistance to maintaining and improving statewide deer hunter harvest survey sampling systems and harvest data acquisition. • Provided annual maintenance and oversight of the DEAMAN (Deer-Elk-Antelope-Management) database representing 20 years of statewide data acquisition and storage. Included updating data files, updating user’s manual, converting DEAMAN operating system to Windows 2000 and then Windows XP, and facilitating the conversion of DEAMAN to a server-based operating system. • Provided 1- and 3-day training workshops in 2002 and 2004 in population modeling and use of DEAMAN for up to 18 terrestrial management biologists. Provided annual support to up to 18 management biologists in their day-to-day use of DEAMAN and associated population modeling spreadsheet analyses. • Provided annual assistance to management biologists in analyzing survival rates of adult female and fawn mule deer and estimates of population density and size within 5 key deer populations used to critically assess the statewide trends mule deer. • Provided critical technical expertise and credibility to designing and implementing a public demonstration experiment to evaluate the reliability of Colorado’s methods to estimate mule deer population size and to model mule deer populations. • Provided scientific and technical expertise annually to all facets of the mule deer research program inclusive of experimental designs and approaches to addressing mule deer population estimation techniques, habitat enhancement studies, and spatial analyses of deer as related to the spread of chronic wasting disease. • Senior or co-authored multiple peer-reviewed publications regarding mule deer research and statewide management in Colorado and provided scientific comment and expertise and several professional workshops pertaining to mule deer and other ungulate research and management. Resulting Peer-Reviewed Publications: BISHOP, C.J., G.C. WHITE, D.J. FREDDY, AND B.E. WATKINS. 2005. Effect of limited antlered harvest on mule deer sex and age ratios. Wildlife Society Bulletin 33:662-668. BOWDEN, D.C., G.C. WHITE, A.B. FRANKLIN, AND J.L. GANEY. 2003. Estimating population size with correlated sampling unit estimates. Journal of Wildlife Management 67:1-10. FREDDY, D.J., G.C. WHITE, M.C. KNEELAND, R.H. KAHN, J.W. UNSWORTH, W.J. DEVERGIE, V.K. GRAHAM, J.H. ELLENBERGER, AND C.H. WAGNER. 2004. How many mule deer are there? Challenges of credibility in Colorado. Wildlife Society Bulletin 32:916-927. MASON. R., L.H. CARPENTER, M. COX, J.C. DEVOS, JR., J. FAIRCHILD, D.J. FREDDY, J.R. HEFFELFINGER, R.H. KAHN, S.M MCCORQUODALE, D.F. PAC, D. SUMMERS, G.C. WHITE, AND B.K. WILLIAMS. 2006 In Press. A case for standardized ungulate surveys and data management in the western United States. Wildlife Society Bulletin. WHITE, G.C. 2004 In Press. Correcting counts: techniques to de-index. Wildlife Research. WHITE, G.C., D.J. FREDDY, R.B. GILL, AND J.H. ELLENBERGER. 2001. Effect of adult sex ratio on mule deer and elk productivity in Colorado. Journal of Wildlife Management 65: 436444. WHITE, G.C., AND B.C. LUBOW. 2002. Fitting spreadsheet population models to multiple sources of observed data. Journal of Wildlife Management 66:300-309. 54 �Associated Annual Wildlife Research Progress Reports Available from the Colorado Division of Wildlife Research Library, Fort Collins, Colorado: FREDDY, D.J. 2002. Deer aerial survey population estimation Rangely deer data analysis unit D6, GMU 10. Colorado Division of Wildlife, Wildlife Research Report July: 117-168. WHITE, G.C. 2002. Improved population modeling-DEAMAN system administration. Colorado Division of Wildlife, Wildlife Research Report July: 93-102. WHITE, G.C. 2003. Multispecies Investigations: consulting services for mark-recapture analysis. Colorado Division of Wildlife, Wildlife Research Report July: 189-196. WHITE, G.C. 2004. Multispecies Investigations: consulting services for mark-recapture analysis. Colorado Division of Wildlife, Wildlife Research Report July: 151-161. WHITE, G.C. 2005. Multispecies Investigations: consulting services for mark-recapture analysis. Colorado Division of Wildlife, Wildlife Research Report July: 67-75. Associated Presentations at Professional Workshops/Symposia: FREDDY, D.J., G.C. WHITE, M.C. KNEELAND, V.K. GRAHAM, W.J. DEVERGIE, J.H. ELLENBERGER, J.W. UNSWORTH, C.H. WAGNER, P.M. SCHNURR, V.W. HOWARD, JR., AND T.S. BICKLE. 2001. Estimating mule deer populatin size using Colorado quadrat system corrected for Idaho mule deer sightability: a sportsmen’s issue. Fourth Western States and Provinces Deer and Elk Workshop, August 1-3, Wilsonville, Oregon, USA. FREDDY, D.J. 2005. Moderator: Session on Representative Strategies. International Association of Fish and Wildlife Agencies Ungulate Data Gathering, Analysis, and Use Workshop, 19 May. Reno, Nevada, USA. WATKINS, B.E., J.H. OLTERMAN, AND T.M. POJAR. 2001. Mule deer survival studies on the Uncompahgre Plateau, Colorado 1997-2001. Fourth Western States and Provinces Deer and Elk Workshop, August 1-3, Wilsonville, Oregon, USA. WAGNER, C.H., B.E. WATKINS, J. VAYHINGER, AND S. STEINERT. 2001. Summary of mule deer survival studies in Colorado, 1997-2001. Fourth Western States and Provinces Deer and Elk Workshop, August 1-3, Wilsonville, Oregon, USA. WHITE, G.C. 2001. Effect of adult sex ratio on mule deer and elk productivity in Colorado. Fourth Western States and Provinces Deer and Elk Workshop, August 1-3, Wilsonville, Oregon, USA. WHITE, G.C. 2005. Featured Speaker: Theoretical considerations and practical implications. International Association of Fish and Wildlife Agencies Ungulate Data Gathering, Analysis, and Use Workshop, 19 May. Reno, Nevada, USA. IV. Assess the Impacts of Chronic Wasting Disease on Mule Deer Populations. Chronic wasting disease (CWD) in mule deer has been a focal point of various research efforts within the CDOW since the early 1990s. Research on CWD was proposed to be funded within this Federal Aid 5-Year Grant. Partial funding from Federal Aid occurred during 2001 but after that year, funding for CWD was obtained from sources other than the Federal Aid Grant. As such, research potentially occurring while Federal Aid funding was in effect was limited to supporting activities associated with publications. Resulting Peer-Reviewed Publications: GROSS, J.E., AND M.W. MILLER. 2001. Chronic wasting disease in mule deer: disease dynamics and control. Journal of Wildlife Management 65:205-215. MILLER, M.W., AND E.S. WILLIAMS. 2002. Detecting PrPCWD in mule deer by immunohistochemistry of lymphoid tissues. Veterinary Record 151:610-612. 55 �WILLIAMS, E.S., AND M.W. MILLER. 2002. Chronic wasting disease in deer and elk in North America. Revue Scientifique et Technique Office International des Epizooties 21:305316. WILLIAMS, E.S., M.W. MILLER, T.J. KREEGER, R.H. KAHN, AND E.T. THORNE. 2002. Chronic wasting disease of deer and elk: a review with recommendations for management. Journal of Wildlife Management 66:551-563. WOLFE, L.L., M.M. CONNER, T.H. BAKER, V.J. DREITZ, K.P. BURNHAM, E.S. WILLIAMS, N.T. HOBBS, AND M.W. MILLER. 2002. Evaluation of antemortem sampling to estimate chronic wasting disease prevalence in free-ranging mule deer. Journal of Wildlife Management 66:564-573. Associated Annual Wildlife Research Progress Reports Available from the Colorado Division of Wildlife Research Library, Fort Collins, Colorado: Miller, M.W. 2002. Chronic wasting disease in mule deer; monitoring and management. Colorado Division of Wildlife, Wildlife Research Report July: 113-116. Associated Presentations at Professional Workshops/Symposia: Conner, M.M. 2005. Increasing the efficacy of chronic wasting disease detection via selective and targeted sampling. Sixth Western States and Provinces Deer and Elk Workshop, May 16−18, Reno, Nevada, USA. V. Develop Alternative Approaches to Control Over-abundant Urban-exurban Mule Deer Populations. An increasing problem with mule deer in Colorado and other states is localized over-abundance of deer in urban-exurban areas. Deer have successfully invaded highly developed human habitats where increasing incidences of browsing damage to lawns, ornamentals, and gardens, and vehicle-deer collisions created the need for some form of deer population control. In these urban-exurban situations, traditional hunting or even highly controlled hunting or culling may not be feasible or socially acceptable. The potential to develop and use hormonal fertility control to reduce net recruitment of deer into these localized populations was recognized by CDOW during the 1990s. Research was initiated to test available hormonal therapies using captive mule deer at the CDOW Foothills Wildlife Research Facility. A portion of this fertility control research was supported by this Federal Aid Grant. After late 2002, other sources of funding were applied to continue this research. Resulting Peer-Reviewed Publications: Baker, D.L., M.A. Wild, M.M. Conner, H.B. Ravivarapu, R.L. Dunn, and T.M. Nett. 2004. Gonadotropin-releasing hormone agonist: a new approach to reversible contraception in female deer. Journal of Wildlife Diseases 40:713-724. Associated Annual Wildlife Research Progress Reports Available from the Colorado Division of Wildlife Research Library, Fort Collins, Colorado: Baker, D.L. 2002. Evaluation of GnRH-PAP as a long-term fertility control agent in female mule deer. Colorado Division of Wildlife, Wildlife Research Report July: 103-112. 56 �SUMMARY Five major multi-year research projects addressing mule deer population limiting factors, habitat status, and habitat enhancements were designed, implemented, completed, and reported upon during this segment. Furthermore, funding partially supported research projects addressing chronic wasting disease and fertility control in mule deer. Additionally, funding provided critical scientific and technical expertise quality control oversight for statewide deer hunter harvest surveys, statewide deer population databases, mule deer survival and population estimate management surveys, mule deer population modeling, and mule deer research projects. LITERATURE CITED BALLARD, W.B., D. LUTZ, T.W. KEEGAN, L.H. CARPENTER, AND J.C. DEVOS, JR. 2001. Deer-predator relationships: a review of recent North American studies with emphasis on mule and black-tailed deer. Wildlife Society Bulletin 29:99-115. BISHOP, C.J., AND G.C. WHITE. 2000. Effects of habitat enrichment on mule deer recruitment and survival rates-a program study plan narrative. Colorado Division of Wildlife, Wildlife Research Report July: 135-180. DEVOS, JR., J.C., M.R. CONOVER, AND N.E. HEADRICK (EDITORS). 2003. Mule deer conservation: issues and management strategies. Berryman Institute Press, Utah State University, Logan, USA. GILL, R.B., T.D.I BECK, C.J. BISHOP, D.J. FREDDY, N.T. HOBBS, R.H. KAHN, M.W. MILLER, T.M. POJAR, AND G.C. WHITE. 2001. Declining mule deer populations in Colorado: reasons and responses. Colorado Division of Wildlife Special Report 77. Fort Collins, Colorado, USA. HURLEY, M.A., M. SCOTT, AND J.W. UNSWORTH. 2002. Influence of predators on mule deer populations. Federal Aid in Wildlife Restoration, Job Progress Report, Project W-160-R-28. Idaho Department of Fish and Game, Boise, USA. HURLEY, M.A., J.W. UNSWORTH, P. ZAGER, E.O. GARTON, AND D.M. MONTGOMERY. 2005. Mule deer survival and population response to experimental reduction of coyotes and mountain lions. Sixth Western States and Provinces Deer and Elk Workshop, May 16−18, Reno, Nevada, USA. Prepared by ______________________________________ David J. Freddy, Mammals Research Leader 57 � https://cpw.cvlcollections.org/files/original/73038235ae1bc33fc5e1a6e2dc9a6dbc.pdf 4555251ee4bed8f36481256f0ae47e88 PDF Text Text 47 Colorado Division of Wildlife Wildlife Research Report July 2001 and July 2002 PROGRESS REPORT State of -------=C=o=lo=r-=a=do=-------- Division of Wildlife-Mammals Research ·work Package No. ___3_0-0-1_ _ _ _ __ Deer Conservation Task _ _ _ _ _ _ _ _.o. . 1_ _ _ _ __ Mule Deer Life Cycle - Neonatal Fawn Survival Federal Aid Project_~W~-~18_5~-~R~---- Research and Development Period Covered: July 1, 2001 - June 30, 2002 Author: T. M. Pojar and D. C. Bowden Personnel: W. Andelt, R Arant, D. Baker, T. Baker, B. Banulis, T. Beck, C. Bishop, G. Bock, D. Bowden, P. Burke, T. Burke, M. Caddy, D. Coven, B. Diamond, B. Dreher, J. Ellenberger, M. Farnsworth, J. Foster, V. Graham, J. Griggs, D. Gustine, P. Hayden, B. Hoffner, B. Lamont, M. King, K. Larsen, M. Mclain, H. McNally, G. Miller, M. W. Miller, E. Myers, J. Olterman, M. Potter, J. Risher, D. Schweitzer, D. Steele, J. Skinner, T. Spraker, D. Swanson, B. Watkins, G. White, S. Znamenacek. . The following is an abstract and manuscript now in preparation for submission to the Journal of Wildlife Management describing the neonatal fawn survival study on the Uncompahgre Plateau. Because ofrequests by reviewers or editors some aspects of the presentation and analysis may be modified. Manipulation or interpretation of these data beyond that contained in this report should be labeled as such, and is discouraged. • ,-/ Abstract: Declining mule deer (Odocoi/eus hemionus) populations resulting from apparent low recruitment brought management and political focus on neonatal fawn survival. Mule deer fawns on the Uncompahgre Plateau (5,957 km 2l in west central Colorado were captured at mean age of 3 days (range from newborn to 6 days) and collared with.mortality sensing drop-off radio collars. Two hundred thirty fawns were radioed with samples of 50, 88, ahd 92 during 1999, 2000, and 2001, respectively. Designated neo~ate survival period was from capture to 14 December. Survival was different among years (X/ = 6.160, P = 0:046) with annual survival (Kaplan-Meier, 95% CL) of0.321 (0.125-0.517); 0:589 (0.474-0.703), and 0.594 (0.472-0.716) for 1999, 2000, and 2001, respectively; the 3-year mean survival was 0.501. Combined 3-year cause-specific mortality (95% CL) was sick/starve 0.171 (0.1160.226), Coyote 0.126 (0.078-0.174), bear 0.040 (0.012-0.068), feline 0.032 (0.006-0.057), trauma 0.043 (0.014-0.072), and unknown 0.047 (0.016-0.077). Neither all predation combined (coyote, bear, and feline) (P = 0.379) nor coyote predation alone (P > 0.989) differed among years. Mortality in the sick/starve category is the only source that approached significance among years (P = 0.070). The major difference was in. 1999 with 0.318 mortality due to sick/starve compared to 0.115 and 0.148 in 2000 and 2001, respecti\'ely. Historic December (1990-99) fawns per 100 does ratios (f:d) were significantly . co.rrelated with the preceding June precipitation (P = 0.004) but not with June temperature (P = 0.441). June precipitation for 1999 was 3 .66 cm and was 1.04 and 0.86 cm in 2000 and 2001, respectively, which BDOW016776 �48 may have contributed directly or indirectly to the differences in sick/starve mortality. Three-fourths of mortality from predation (75.0%) and sick/starve (73.7%) had taken place by 31 July with 76.3% of mortality from all sources occurring by 31 July. Mean fawn weights at capture were 4.35 kg, 4.50 kg, and 4.13 kg for 1999, 2000, 2001, respectively and were different among years (P = 0.044). There was also a difference in hind foot length among years (P = 0.002) with mean length of26.14 cm, 26.62 cm, and 25.63 cm for 1999, 2000, and 2001, respectively. Weight and hind foot means were different between 2000 and 2001 (P > 0.017) with 1999 not different from either 2000 or 2001 (P < 0.017) using mean separation procedure controlled with Bonferonni significance level. Mean capture date was 19 June ( 4.83 days SD) and median capture date was 19 June (range 9 June to 6 July) with 94.78% of all captures occurring between 13-30 June. This implies that most does were bred during their first estrous cycle. Neonatal survival through 14 December does not completely account for observed low f:d ratios. Fetus mortality during late pregnancy or mortality of fawns at birth (before they could be detected for capture) is implicated as a potential cause of poor recruitment. [• ! L L I l I I L . / I ' �49 NEONATAL MULE DEER FAWN SURVIVAL AND CAUSE SPECIFIC MORTALITY T. M. Pojar and D. C. Bowden There is evidence that the mule deer populations in Colorado, as well as other western states, have declined during recent years due mostly to low fawn survival and subsequent low population recruitment (Unsworth et al. 1999). December f:d ratios on the Uncompahgre Plateau (5,957 km2l in west central Colorado have declined (t = -3.41, P = 0.004) by an average of 1.8 fawns:100 does per year from 1982-1998 where December ratios ranged from a high of 79f: 100d in 1982 to a low of 32f: 100d in 1997 (White et al. 2001). Declining deer densities and long-term decline in f:d ratios have resulted is much debate and concern among managers, sportsmen, administrators, and politicians. Gill et al. (2001) offers several potential causes for long-term decline in deer density: 1) habitat deterioration, 2) competition, 3) disease, 4) predation, and 5) hunting. Specifically, low December ratios could be the result of: 1) low pregnancy rate, 2) reduced fetal production, 3) prolonged breeding (fawning) season due to low buck to doe ratios, 4) late term mortality of fetuses and weak or stillborn fawns, or 5) low neonatal fawn survival through December. A popular perception is that predation especially by coyotes (Canis /atrans), but also by black bear (Ursus americanus) and felines (Fe/is conco/or and F. rufus), is a major contributing factor in apparent low neonatal fawn survival. Pregnancy and fetal rates have been relatively constant for this species at population densities encountered during recent decades. During the 1960s and 1970s when deer populations were thriving in Colorado pregnancy rates in the following habitats were: 1) front-range foothills 92.0% (n=163) (Medin and Anderson 1979), 2) high mountain park 94.0% (n=134) (Gill 1971), and western slope pinyonjuniper 89.0% (n = 47) during 1973 and 82.0% (n = 83) during 1978 (Bartmann 1998). During 1963-71, a penned herd that was fed alfalfa hay ad libitum and a 16% protein supplement ranging from 0.23 to 0.90 kg per deer-day averaged 94.7% pregnancy (n=135) (Robinette et al. 1973). Fetal rates, as the other major component of reproductive rates, for adult does were 1.83 (n = 41) during 1961-1965 in frontrange foothills habitat (Medin and Anderson 1979) and 1.82 (n = 114) in high mountain park habitat during 1969-1971 (Gill 1971) in Colorado. To determine if both pregnancy and fetal rates of adult does have not changed since the 1960s and 1970s and to have specific information from the Uncompahgre Plateau, a separate preliminary study examined these factors using transrectal ultrasound. A sample of 40 does was examined in February 1999. Pregnancy rate (93%) did not differ from the historic rate of 94%, (n = 328, X/ = 0.07, P = 0.791) and the fetal rate (1.70) was not less than the historic rate of 1.87 (n = 307, Z = 0.412, P = 0.681) (Andelt, Pojar, and Johnson, unpublished data). Density dependent effects of population growth follows the progression ofreducedjuvenile survival, increased age at first reproduction followed by a decline in reproductive rates of mature females late in the progression to carrying capacity and, finally, reduced adult survival (Eberhardt 1977). Juvenile survival is highly variable and sensitive to population density and stochastic environmental factors whereas, adult female survival is robust against most limiting factors (Gaillard et al. 1998, White and Bartmann 1998). Therefore, the most obvious area of investigation was to examine the survival and mortality sources of neonate fawns. The primary objectives of this study were to estimate neonatal fawn survival from birth (time of capture) to 14 December and cause-specific mortality to determine the contribution of summer fawn mortality to December f:d ratios on the Uncompahgre Plateau. Timing of births (captures), as an index of fawning season compression was a secondary objective; this relates to cohort exposure to predation (predator swamping). Extent and timing of the various mortality sources are described. �50 STUDY AREA The Uncompahgre Plateau was formed by a structural up-lift; it runs generally southeast to northwest with a crest, a_ break in the up-lift, roughly bisecting it forming drainages to the northeast and southwest. The small communities of Gateway and Ridgeway are on the northwest and southeast perimeter, respectively, with the larger communities of Montrose and Delta along the east boundary (Figure 1). The terrain slopes generally to the northwest along the crest with the highest point being Horsefly Peak at 3,147 m and the lowest, 1,389 m, near Gateway. Along the crest, the terrain slopes gently to the northeast where many tributaries to the Gunnison and Uncompahgre rivers have worn deep canyons; the Unaweep Canyon along the northwest boundary of the area averages 914 m deep (Young and Young 1984). Southwest of the crest terrain drops abruptly because the plateau cap lifted and tilted to the northeast giving a gentle slope to the northeast and a steep drop-off to the southwest (Marshall 1998) with drainages flowing to the San Miguel and Dolores rivers. The latitude is from 38° 1.23' to 38° 59_05' and the longitude ranges from 107° 45.52' to 108° 58.80'. Normal annual precipitation (1971-2000 mean) is 29.11 cm with most precipitation in late summer and fall (July-October); the 30-year mean low temperature (-2.78° C) was in January and the high (22.86° C) was in July. These weather statistics were taken from 5 stations on the perimeter of the plateau and would best represent winter range; summer range is at higher elevations and would have lower temperatures and higher precipitation. The study area includes Game Management Units 61 and 62 (5,957 km 2) and is designated as Data Analysis Unit D-19. Vegetation types are agricultural along the major river drainages and as elevation increases the following types are encountered: saltbrush-greaswood (Atriplex canescens and Sarcobatus vermiculatus), mature pinyon-juniper (Pinus edulis and Juniperus osteosperma) forest interspersed with big sagebrush (Artimisia tridentata) parks, Gambel oak (Quercus gambelii), ponderosa pine (Pinus ponderosa), and spruce-fir (Picea-Abies). The oak, pine, and spruce-fir communities are interspersed with aspen (Populus tremuloides) with some areas of pure aspen stands bordered by areas of mountain shrub and high mountain grass-forb meadows. There are vast stands of aspens on the rugged southwest slopes of the Uncompahgre. The oak, pine, spruce-fir, and aspen types are the major summer range and fawning habitat for mule deer and are generally above 2,438 m. Winter range can be from the oakmountain shrub community (depending on winter severity) decreasing in elevation to the agricultural lands along major drainages (Figure 1). Livestock grazing began in 1881 immediately after the Ute Tribe was expelled from the Uncompahgre Plateau (Anderson et al. 1992). Large herds of cattle were brought in from Texas, Kansas, and Mexico and within 20 years the town of Placerville (on the south end of the Uncompahgre Plateau) became " ... the largest cattle-shipping point in the world" (Marshall 1998:59). Severe overuse by livestock continued for 70 plus years, at least until 1951, accompanied by extreme overpopulation of deer beginning in the 1940s until liberal harvests during the 1950s and 1960s reduced the deer population (Kufeld 1979). Range condition around 1900 was described as being eaten " ... down to the nub" (Marshall 1998:59). Still, by about 1944 the range was described as appearing as having " ... had a band of sheep "caked" on it" (anonymous, c.a. 1944:8). Sharp stocking rate declines began in 1948 and " ... have remained relatively constant near their lowest rates from 1951 to the present" (Kufeld 1979: 13 ). The overstocking of livestock from 1881 to 1948 and deer from the 1940s to the 1960s has been alleviated; however, reduction in grazing pressure does not necessarily result in range condition improvement because the range may stabilize at a lower successional state (Laycock 1991 ). Timber harvest has been a factor in the past and continues to the present. Roads and rural subdevelopments are expanding especially in the southern area of the plateau. Summer and winter recreation, including sightseeing, camping, biking, hunting, all terrain vehicle, and snowmobiling and cross-country skiing are common human activities (Uncompahgre Plateau Partners 2002). This study was approved by Colorado State University and Colorado Division of Wildlife animal care and use committees under Protocol Number 99-063A-01. �51 METHODS The fawning area generally included the entire summer range of the plateau above 2,438 m. Mature does (n = 74) that had been radioed during the ultrasound reproductive study and during a separate survival study were tracked to more precisely identify fawning habitat during the first year of capture efforts. Fawns ofradioed does were not necessarily targeted for capture. Searches for parturient does were from the ground, either on foot or by vehicle. Behavior and physical features were clues to identifying parturient does. Does that were alone, had udder development, and sunken flanks were prime candidates for initiating an intense search for bedded fawns within a 50± m radius of where the doe was first spotted. Searches were terminated if young were not found within about a half-hour. Bedded fawns were approached from the rear, avoiding eye contact, quickly approaching for the last couple meters. Upon placing a hand (latex gloved) on it's back, the fawn would freeze. Immediately upon capture, it was blindfolded and hobbled; processing took :s_ 5 minutes and included weighing, hind foot measurement, and attaching the radio collar. Age of the fawn was estimated by condition of the umbilical cord, pelage, hoof condition, and behavior. Before release, the fawn was examined for signs of dehydration, sickness (diarrhea or respiratory discharge), or physical deformities. Attempts were made to leave the fawn at the same site where it was captured. Radio collars (weighing< 110 g) were expandable from 22 cm to 33 cm and designed to drop off at approximately 6 months. Mortality sensor was set at 2 hours. Once radioed, each fawn was tracked to determine live status at least twice a day and sometimes as many as 6 times per day through 1 September and then once a day, excluding weekends, through 14 December. Determination oflive status was done from distances (0.5-5.0+ km) sufficient to minimize disturbance to either the fawn or doe. Mortality signals were investigated immediately upon detection. The site and evidence was surveyed and a determination of cause was assigned as best as evidence offered. If no carcass was present and evidence indicated predation or scavenging, criteria offered in the literature was used to help assign a specific predator (White 1973, Wade and Bowns 1984, Acom and Dorrance 1990, and Andelt et al. 1998). Most of these; however, deal with adult or domestic animals (except White 1973) and other geographic areas but were used anyway to help evaluate on-site evidence. Ground cover and vegetation was too thick to find actual tracks so, in addition to the above, the following general criteria were used. Coyote kill/scavenge sites typically had only shards of bones, tufts of hair/hide, usually> 1 feeding site within 30 m, and sometimes fawn parts buried in mineral soil. Bear sites were identified by one relatively large feeding site (1-5 m diameter) with the fawn hide nearby and usually intact and inverted (Schlegel 1976); bear usually defecate near their feeding site. The presence of hooves and small leg and cranial bones typify black bear kill sites (Bertram and Vivion 2002:751). Felines (Mt. Lion and bobcat) drag their prey from the kill site to a protected feeding area and cover any remains with litter and duff. Deaths due to vehicle collisions, entanglement in fences, accidents, or human caused (poaching) were categorized as trauma deaths. The unknown category included collars that were found with no carcass parts or other evidence of the fawn in the vicinity. It is possible the collar was carried from a kill site by an avian or terrestrial predator/scavenger. This category undoubtedly includes some collars that were slipped either by maternal grooming or a lucky stroke by a hind foot. Even if the collar had bite marks or blood on it, it was classified as unknown because assigning it to any of the 3 major terrestrial predators would be quite uncertain. Ballard et al. ( 1999) chose to construe this line of evidence as coyote predation because coyotes are known to prey on fawns in summer. Whole carcasses that were found were examined for external evidence of sickness such as diarrhea or respiratory discharge or external injuries. Proximate site characteristics were described as with other mortalities. Wh(?le carcasses were field classified as sick/starve but the mortality category was changed to trauma if necropsy revealed internal injury from trauma. During the first year, carcasses were frozen and transported to the necropsy laboratory within 5-7 days. Protocol for 2000 and 2001 provided for the carcass to be iced and transported to the laboratory within 2 hours of discovery. A trained pathologist did all necropsy and tissue collections. Details of laboratory procedures, tissue sample collection, diagnostic techniques and disease detection results are found in Myers (2001 ). �52 Thymus glands were not weighed during the first year of our study but the attending pathologist made a subjective judgment on the condition and size of the thymus at necropsy. During the last 2 years, thymus glands were weighed to the nearest hundredth of a gram. Data in terms of survival time were censored if the radio dropped off before 14 December. If a radio was not heard and could not be accounted for, survival time data were censored on the last day the radio was heard. All survival time data associated with radios recovered and assigned to the "unknown" category were censored. The key assumptions of a survival study are: 1) animals are sampled randomly, 2) experimental unit survival times are independent, 3) capture and carrying a radio package does not affect survival, 4) the censoring mechanism is random, and 5) emigration is zero (Pollock et al. 1989, Tsai et al. 1999). Because there is high probability of mortality soon after birth in mule deer and because the fawning season is temporally compressed (about 2 weeks), staggered entry of subjects was not used. The time origin for the nonparametric Kaplan-Meier (Kaplan and Meier 1958) survival estimator was the date when the first fawn was radioed. Staggered exits from mortalities and censoring were incorporated in the calculations of survival and confidence limits following Pollock et al. ( 1989). Large-sample Chi-square tests were used to compare yearly Kaplan-Meier estimates of survival rates. Fisher's Exact Test was used for tests of association and Log Rank statistic was used to compare mortality distributions among years (Cantor 1997). ANOV A and pair-wise mean comparisons were made following the Bonferroni inequality to compare fawn weight and hind foot measurements. The 0.05 significance level was used for all tests. RESULTS During 3 fawning seasons 230 fawns were radio collared with 50, 88, 92 captured during 1999, 2000, and 2001, respectively (Table 1). Mean capture date was 19 June (4.83 days SD) and median capture date was 19 June (range 9 June to 6 July) with 94.78% of all captures occurring between 13-30 June. Mean fawn weights at capture were 4.35 kg, 4.50 kg, and 4.13 kg for 1999, 2000, 2001, respectively and were different among years (P = 0.044). There was also a difference in hind foot length among years (P = 0.002) with mean length of 26.14 cm, 26.62 cm, and 25.63 cm for 1999, 2000, and 2001, respectively. Both weight and hind foot means were different between 2000 and 2001 (P < 0.017) with 1999 not different from either 2000 or 2001 (P > 0.017). During the first year, the attending pathologist diagnosed 8 of 15 (53%) fawns in the sick/starve mortality category with "severe thymic atrophy". Mean thymus weight was 2.62 g (SD 2.90, n = 10) and 1.96 g (SD 2.40, n = 10) for 2000 and 2001, respectively. These means were not different (P = 0.584) and were combined for a 2-year mean of 2.29 g (SD 2.62, n = 20). Mean June precipitation was 3.66 cm in 1999 and 1.04 and 0.86 in 2000 and 2001, respectively. June precipitation on the Uncompahgre Plateau during 1990-1999 was negatively correlated with subsequent December f:d ratios (P = 0.004) but was not correlated with June temperature (P = 0.441). Survival was different (X22 = 6.160, P = 0.046) among years with annual survival (95% CL) of 0.321 (0.125-0.517), 0.589 (0.474-0.703), and 0.594 (0.472-0.716) for 1999, 2000, and 2001, respectively. Sick/starve was the only cause-specific mortality that approached significance among years (P = 0.070) (Table 2). The major difference was in 1999 with 0.318 mortality due to sick/starve compared to 0.115 and 0.148 in 2000 and 2001, respectively. The 3-year combined mortality due to sick/starve was 0.171 (0.116-0.226). Coyote predation was 0.126 for combined 3-year data and was highly consistent (P = 0.989) among years. Likewise, bear and feline caused mortality was consistent among years (Table 2) and was 0.040 (0.012-0.068) and 0.032 (0.006-0.057), respectively. All predation combined did not differ among years (P = 0.379). Trauma, which included roads, fences, injury, etc. was 0.043 (0.014,..0.072) and unknown causes accounted for 0.047 (0.016-0.077) of the 3-year mortalities. The temporal distribution of mortalities was consistent among years (X/ = 0.680, P = 0.712) with 76.3% • of all mortalities occurring by 31 July. This is the result of the major sources of mortality, sick/starve (73.7%) and predation (75.0%), taking place by 31 July. �53 DISCUSSION Fawn capture effort was focused on the high elevation summer range generally above 2,438 m. How well the sample of fawns captured in this area represents the entire Uncompahgre Plateau population is of major importance. Mule deer tend to be seasonally migratory in the mountainous areas of Colorado (Garrott et al. 1987). However, in the Colorado eastern foothills the majority of the herd may rernain at lower elevations yearlong (Kufeld et al. 1989). In Idaho, 26% of a herd that wintered in broad agricultural valleys and low elevation rangelands stayed on the wintering area yearlong (Brown 1992). Wintering area of the Uncompahgre Plateau herd includes some low elevation valleys but raises quickly into sagebrush and pinyon-juniper habitat (Figure 1). To determine how representative the fawns captured at high elevations were of the entire population we used the elevations of winter-captured does (n = 95) during 1997-2000 and their subsequent aerial relocations during mid-May (n = 64) and late-May (n = 144). The mean elevations for capture, mid-May (May 18-21), and late-May (May 28-31) relocations were 1,927 m, 2,551 m, and 2,603 m, respectively with a highly significant difference (P,:::: 0.001); mean of winter capture locations was different (P < 0.05) from the relocations. The 95% kernel home range of all does relocated during midor late-May closely matches the 95% kernel home range of all fawn capture sites (Figure 1). This indicates this population generally fits the near-total migratory pattern described by Garrott et al. (1987). Does had about 3 more weeks to complete their migration to summer/fawning areas, which would have reduced the May relocation home range size because does do not settle on their fawning area and reduce their individual home range until about 3-5 days of parturition (Haegel et al. 1985). The aerial trapping operation did not attempt to capture does among farmsteads along the river bottoms and agricultural lands. These lands compose 6% of the total area (Figure 1). During the 5 years prior to this study, the sex ratio of the Uncompahgre Plateau deer herd averaged 12.3 bucks per 100 does (Colorado Division of Wildlife data). Later mean parturition date, a less synchronized birthing pulse, and lower pregnancy rates resulting in reduced recruitment are some of the postulated consequences of low sex ratios (Squibb 1985, White et al. 2001). In a controlled experiment with elk, the calving season was later by 17 days and extended by 30 days when breeding was done by yearlings compared to 5-year-old bulls; pregnancy rate was 89% with yearlings and 97% with 5year-olds (Noyes et al. 1996). In a free-ranging elk herd, bulls> 1 year old did 76% of the breeding with yearling bulls making an appreciable contribution to successful breeding in this herd and " ... completely compensating for the absence of older bulls" (Squibb 1985 :750). Low sex ratios could have a greater impact on deer compared to elk because deer have a tending-bond breeding system and elk form harems (Kie and Czech 2000). Examination of 20 years of sex ratio data and fawn to doe ratios provided no evidence to indicate that sex ratios observed across the state affected population productivity in Colorado mule deer (White et al. 2001 ). Our data support the contention that low sex ratios did not adversely affect herd productivity. Mean capture date was 19 June in this study, which is similar to the mean fawning date of 18 June (n=215) for a captive herd in Colorado (Robinette-et al. 1973). Ninety-five percent of our captures were within a 2-week period, between 13 and 30 June, providing evidence that most does were bred during their first estrus. Estrous cycle is 23-29 days for mule deer and 97% conceive during their first cycle (Anderson and Wallmo 1984). Survival of individual fawns is related to the interaction of nutrition, cover, and climate Picton (1979). On northern ranges, deer are frequently subjected to rigorous winters resulting in chronic malnutrition of does and subsequent stillbirths, weak fawns, and lactation failure (Verme 1969). Although the 3 winters encountered during this study were milder than normal, this population seems to fit the description of a herd that is stressed during winter. During our searches of fawning areas we discovered a total of 9 fawns that were stillborn or died within minutes/hours of birth. In addition, we found one doe that was so weakened from trying to deliver dead twin fetuses that she was captured, restrained, and the fetuses delivered; the doe apparently survived as she was not found in the vicinity the next day. Two mature does were killed or scavenged by bear; one was prime aged (3-4 years old) and the other was of unknown age. Both were found during the peak of fawning - 18 and 19 June. Bear are �54 known to prey on both fawns and adult deer (Behrend and Sage 1974, Conger and Giusti 1992, Verspoor 1983) but bear rarely prey upon healthy adult deer (Verspoor 1983 ). The best speculation is that the does were in a difficult delivery, as the above mentioned doe, or had died from delivery complieations and were scavenged. • During extended periods of damp cool weather fawns may have a higher incidence of exposurerelated complications and deaths (Ginnett and Young 2000). Cool dry summer weather in the northern regions of mule deer range enhanced fawn survival (Picton 1979). Mortality due to sick/starve during 1999 was 0.312 compared to 0.115 and 0.148 for 2000 and 2001, respectively. In an attempt to discover possible causes for the higher sick/starve mortality in 1999 compared to 2000 and 2001, we examined June precipitation and temperature and fawn weights and skeletal development as gauged by hind foot length. We speculated that fawning-season weather might be a factor or that fawn robustness, as measured by fawn weights and skeletal development, might differ among years. Since the 1999 fawn size indicators were not different from the other 2 years when sick/starve mortalities were much lower, it cannot be concluded that fawn size affected the proportion of fawns dying of sickness or starvation. Mean birth weight of fawns from a captive herd was 3 .69 kg (n= 172) (Robinette et al. 1973) and is less than means we observed. This is expected because these were nearly true birth weights and our measurements were taken at mean age of 3 days and fawns can gain 0.29 kg per day during their first 12 days (Robinette et al. 1973). Given these weight comparisons, there is no evidence to suggest that fawn weight was a factor in the difference in fawn mortality due to sick/starve among years. Thymus gland development in cervids of similar size to mule deer (fallow deer (Dama dama) and sitka deer (Cervus nippon), follows a pattern of growth during fetal development then remains relatively constant from birth to puberty (Chapman and Twigg 1990). In mule deer (as in other cervids) it then declines into adulthood with seasonal peaks and troughs in summer and winter, respectively (Anderson et al. 1974). Measurements of thymus glands of fawn, yearling, and adult mule deer, Browman and Sears (1956) observed annual cycles of highs in summer and lows in winter with fawns having the highest values of the 3 ages. Thymic atrophy is generally the result of chronic stress and can be seasonal (related to photoperiod or climate) or due to immediate stress such as inanition and disease (Chapman and Twigg 1990). White-tailed deer (0. virginianus) on low energy diets had lower (P < 0.05) thymus weights than deer on high energy diets (Lawrence et al. 1986). Ozoga and Verme (1978) conclude that the thymus provides a reliable index to physiological status and Lawrence et al. (1986) suggest thymus weight of adult does could be used in management decisions. Neonatal white-tailed fawns that were dying of disease or malnutrition had "extremely small" thymus glands averaging 1.3 g (range 0.5-3.0g, n = 14) compared to healthy fawns of similar age (X= 9.7 g, range 4.3-23.7 g, n = 7) (Ozoga and Verme 1978:794). Our combined 2-year mean thymus weight of 2.29 g (SD 2&2, n = 20) is similar to the mean of 1.3 g for fawns near death from disease or starvation observed by Ozoga and Verme (1978). In Colorado Anderson et al. (1974) collected 13 mule deer fawns (6 male and 7 female) from 1 to 5 months old; their mean thymus weight was 9.22 g (SD 2.88, n = 13) and was comparable to the healthy fawns sampled by Ozoga and Verme ( 1978). There were· only 3 measurements in our sample of sick/starve fawns that approached the means observed by Anderson et al. (1974) or Ozoga and Verme (1978) for healthy fawns. In 2000 a fawn 68 days old died (6 September) of a hemorrhagic condition and had thymus weight of 8.00 g and a second fawn 69 days old died (8 August) of pneumonia with a thymus weight of 7.95 g. Both had fat reserves but were judged to be less than optimal. In 2001, a fawn 21 days old died (7 July) of a hemorrhagic condition and had a thymus weight of 8.28 g; it was judged to have poor fat reserves. Excluding these 3 values, the mean for fawn thymus weight in this study was 1.26 g (SD 0.85, n = 17). The obvious reduced thymus size of fawns dying of sickness or starvation in this study should serve as a point of concern for managers. The reduced thymus size was probably initiated during the fetal stage of development and would indicate the stress factor was affecting the dam. Inanition has been shown to result in reduced thymus size in deer (Lawrence et al. 1986, Ozoga and Verme 1978) so the nutritional status of does during pregnancy, and especially during the last trimester, should be investigated. Fawns dying of sickness and starvation in 2000 and 2001 was nearly half the mortalities �55 attributed to this cause in 1999. The weather during fawning seasons of 2000 and 2001 was warm and dry possibly allowing fawns that were not robust to stresses to survive. This study was not designed as a manipulative study where some factor or factors were controlled or manipulated and the impact on fawn survival measured. Coyote predation on neonatal fawns was a popular theory and the opportunity arose to examine the effects of coyote control on a small portion of the study area. The area, 130 km (2% of the total area) included 3 sheep operations on private land. These ranches were used as lambing and summer ranges so coyotes were killed before the sheep were moved onto the area. Coyotes were killed from January trough September with most kills during winter and spring mostly by aerial gunning with a few kills from the ground. There was an active predator (coyotes and bear) control program during 1994 through 2001 on this area with a total of 187 coyotes and 17 bear killed (Animal and Plant Health Inspection Service, Wildlife Services records, Grand Junction, Colorado). During the 3 years of the fawn survival study there were 53 coyotes and 11 bear killed in the predator control area. Forty fawns were collared on this corresponding area allowing a comparison of fawn survival on and off the control area. Seven fawns were killed by predators inside the control area (4, coyote; 1, bear; and 2 feline) and 37 outside the area (24, coyote; 8 bear; and 5 feline). Comparison of predator kills inside and outside the area resulted in a Fisher's Exact Test result with P = 0.830; limiting the test to only coyote kills the results offered no evidence of an association between coyote control and fawn survival (P = 0.794). For fawn survival study results to be comparable they should be similar in the following: 1) fawn age at capture, 2) equipment and handling procedures, 3) tracking frequency, 4) nutritional status of does, 5) vegetation and hiding cover, 6) predator density, and 4) mortality identification criteria. Although it is impossible to match all of the above for comparisons, generalizations may be useful to assess the possible impact of the various mortality sources, particularly predators, on neonatal fawns. In Montana Hamlin et al. (1984) radioed 91 fawns over a 6-year period (1976-1981) and tracked them at 2-3-day intervals. Fawns up to 3 weeks old were included in their sample (Riley and Dood 1984). Mortalities were categorized as either "probable or known coyot~ involved deaths" or "other". They found no whole carcasses, which may be the result of tracking them on 2-3 day intervals allowing scavengers (including coyotes) time to find the carcass. Eighteen of 20 deaths (90%) were attributed to coyotes and 2 (2.2%) were listed as "other". Eighteen mortalities of 91 radioed fawns (19.8%) were assumed to be coyote-caused and total survival was 78.0%, which is higher than we observed. Their sample of fawns was most likely older than our sample. They used aerial observers to spot fawns indicating that the fawns were old enough to be trailing the does and ground crews used long-handled hoop nets to capture the fawns indicating the fawns were no longer in the hiding phase. A sample of older fawns would miss mortalities immediately after parturition and result in a higher survival rate compared to a sample of younger fawns such as ours. A fawn survival study on a 51.8 km Steens Mountain study area in Oregon during 1971-74 had a sample of 106 neonate fawns aged 1-14 days old and were monitored every 3-4 days (Trainer 1975). Mortality attributed to coyotes was 10.3% and for all predators it was 15.1 %. Disease and starvation mortality accounted for 9.4% of the total; survival was 72.6%. Preliminary results of an Idaho study with a sample of 69 fawns during 1998-99 exhibited a loss to coyotes of 13% and total predators (coyotes and lions) of32%. Overall survival was 44.9%. These results are not directly comparable to our study because coyotes and lions were controlled on a portion of the area. Given the shortcomings of comparing results of studies where protocol is not similar, neonatal fawn mortality attributed to coyotes is in the range of 10-20% for the various studies. Survival is highly variable ranging from 44.9% to 78.0%; the range in survival is undoubtedly heavily influenced by differences in age of fawns at capture (beginning of monitoring). Neonatal survival through 14 December does not account for observed low f:d ratios. In addition to pregnancy and fetal rates from the preliminary productivity study in February 1999, data available for this herd included random quadrat-based population size and herd structure estimates in December 1999. Survival estimates for bucks, does, and fawns during winter of 1999-2000 based on radioed animals �56 (Bruce Watkins, Colorado Division of Wildlife, Montrose, personal communication) were available. We used this information and incorporated our observed year 2000 summer fawn survival (0.5887) to calculate the expected f:d ratio for December 2000. Our calculations included I 0% lower fetal rates of primaparous does (Robinette et al. 1973, Trainer et al. 1981) and a differential of viable neonates of 96% for multiparous does and 82% for primaparous does (Robinette et al. 1973). They did not have an estimate of fetal rate, but the birth rate of 1.92 fawns per doe is similar to the maximum fetal rate for mule deer (Jensen and Robinette 1955). Hamlin and Mackie (1989) estimated 80% viable neonates for all-age does; this estimate includes fetal and neonate mortality. Using differential viable rates of Robinette et al. (1973)(96% and 82%) the projected December f:d ratio was 75 and using all-age estimated viable rate of Hamlin and Mackie (1989) (80%) the projected December f:d ratio was 64. Both of these projections were higher than the observed f:d ratio of 51 as estimated by a random quadrat helicopter survey in December 2000. Our data are most comparable to those of Hamlin and Mackie ( 1989) because theirs was a wild population. The herd studied by Robinette et al. ( 1973) was a fed captive population but indicates that a proportion of fawns born are not viable for various reasons even in a well nourished herd. Assuming the f:d ratio of 51 from the helicopter survey is unbiased, mortality of 3 7% from February when fetal rates via ultrasound were taken and June when fawns were captured would be necessary to match the observed f:d ratio. This indicates fetal or early neonate mortality that could be caused by inanition of the does, disease, or effects of poisonous plants. The importance of nutrition in reproductive success and recruitment is well documented. However, in the study by Robinette et al. (1973) fawn weights did not vary with nutrition level of does. The fawn weights in our study were comparable to those of other studies (Robinette et al. 1973, Stieigers and Flinders 1980, Trainer et al. 1981, and others). Apparently, fawn weights do not provide a useful index of doe nutritional status. Although fawns are born at relatively uniform weights the nutritional status of the doe can affect fawn survival through indirect effects such as susceptibility to predation and disease. The dam can be directly affected by failure to conceive, resorption of fetuses, and inability to nourish offspring (Dietz and Nagy 1976). What appears to be excessive fetal and neonate mortality from early pregnancy to a few days postparturition and discovery of 9 under-sized (X = 1.67 kg, n .= 7) stillborn fawns may be indicative of an under nourished adult population. Increased loss to sickness and starvation during 1999 when June precipitation was higher that the other 2 years may also indicate that neonates are in a compromised condition and not robust to stresses. Hemorrhagic diseases (HD), bluetongue (BT) and epizootic hemorrhagic disease (EHD), of the genus Orbivirus are present in the Uncompaghre Plateau mule deer herd (Myers 2001). These diseases are capable of causing significant mortality and Howerth et al. (200 I) cite literature documenting many outbreaks in Western North America dating back to 1886. In temperate regions, mortalities from hemorrhagic disease usually occur in late summer, before first frost, and epidemics can develop when conditions are favorable to the vector, Culicoides spp. These outbreaks are usually sporadic (Howerth et al. 200 I) and localized with total mortality estimated at < I% for mule deer (Thome et al. 1988). Infection with BT or EHD in mule deer may be asymptomatic, result in chronic disease, nonfatal infections, or sudden death (Howerth et al. 2001). Fever, internal bleeding, and shock resulting in death characterize hemorrhagic diseases (Shope 1967). Death may happen so suddenly that some animals may die " ... while walking or running" while others struggle in lateral recumbency position (Thome et al. 1988:115). Because this disease strikes quickly, animals in good physical condition may be found dead from HD (Chalmers et al. 1964). Only I positive result based polymerase chain reaction (PCR) test was obtained for HD during our 3-year study. The low detection rate may be because these are RNA viruses and are very unstable in an open environment (Myers 2001). There may have been other deaths from HD based on the time of year, hemorrhagic condition, and the relatively good condition of the fawn at death indicating a sudden death. Five fawns that died between 18 August and 4 October satisfy the above criteria. An adult female found near (100 m) one of the fawns tested positive by PCR for EHD. �57 Hemorrhagic disease is present on the Uncompahgre Plateau but it is hard to assess the impact on the mule deer population. It is unlikely that an epidemic of HD occurred during this 3-year study. There were no reports of numerous dead deer as was the case in other epidemics (Chalmers et al. 1964, Thome et al. 1988) and field personnel did not observe any abnormal concentrations of mortalities of either radioed fawns or unmarked deer during late summer. Diseases that affect the reproductive capacity of a host population are most liable to have a noticeable impact on that population (Anderson and May 1979). Bovine viral diarrhea virus (BVDV) infections produce abortions, fetal malformations, stillbirths, weakened neonates, and immunosuppression in domestic livestock (Baker 1995, Van Campen et al. 2001a). There is a> 60% prevalence ofBVDV titers in the adult population of deer on the Uncompahgre Plateau (Myers 2001). A mule deer population from northwestern Wyoming, USA, also had a 60% prevalence of BVDV titers (Van Campen et al 2001a), and a serological survey of 4 western national parks resulted in 59% prevalence in mule deer (Aguirre et al. 1995). Viral isolation (VI) is the most reliable method to determine exposure to BVDV and isolations from wild ruminants are rare (Van Campen et al. 2001a, Van Campen et al. 2001b). Isolates were obtained from 2 fawns in our study. These fawns died in the same general area(< 500 m apart) and within 2 days of each other, 17 and 19 July. Diseases that have low mortality and produce immunity with exposure are self-limiting (Myers 2001). In closed herds with no previous exposure, introduction ofBVDV can result in the loss of 75% of the first neonate cohort after exposure through abortions, stillbirths, and compromised immune response (Hana Van Campen, personal communication). With a high proportion of the Uncompahgre Plateau deer herd having titers, and presumed immunity to BVDV, this disease should not have a significant impact on the overall herd performance. However, its presence is certainly a depressant to some unknown degree. Other than the 2 fawns that provided VI ofBVDV, there were 2 other fawns with symptoms of being exposed to BVDV in utero. One was hydrocephalic (1.90 kg) and the other had skeletal deformities (2.09 kg) both characteristic of BVDV exposure. BVDV was isolated from a stillborn fawn from northern New Mexico that had an atrophied thymus and weighed 2.3 kg (Hibler 1981). The high prevalence of BVDV in the 2 above mentioned mule deer populations suggests that this virus circulates in these populations (Van Campen et al. 2001b) without exposure to outside sources such as cattle herds. High prevalence of titers to BVDV does not necessarily mean a population suffers significant consequences. In immunocompetent cattle the majority of infections (70-90%) are subclinical (Baker 1995). So unless the immune response of mule deer is compromised from some other cause, the impact of BVDV may not produce significant or detectable manifestations in population performance. Ingestion of poisonous plants can impair reproductive functions of domestic livestock (Panter et al. 2002). Some of the plants poisonous to livestock are found on the Uncompahgre Plateau and could conceivably also affect the wild ruminants of the plateau. Lupines (Lupinus spp.) can cause skeletal defects through the effects of alkaloids that are toxic to fetuses (Panter et al. 2002). We found a stillborn fawn (2.09 kg) with "Multiple congenital skeletal defects, including flexion contraction, limbs, neck, and thoracic spine" (from lab report) with minimally deformed joints and normal limb bones; all of these symptoms fit lupine poisoning as described by (Panter et al. 2002). In the year following our study, another stillborn fawn with severe skeletal deformities was found (Chad Bishop, Colorado Division of Wildlife, Montrose, personal communication). Both of these fawns came from the same general area of the plateau where lupine is common. Livestock loses to poisonous plants is associated with range in poor condition (Ralphs 2002) and this area is heavily grazed. We have observed some possible pre-natal mortality from poisonous plants; it could be one of many mortality sources but it is unlikely this is a major factor in herd recruitment. MANAGEMENT IMPLICATIONS Conditions of Western ranges have changed dramatically from pristine times (Vale 1975) through an era of extensive overgrazing to the current level of management. The era of unsustainable livestock grazing promoted shrub and forb growth to the benefit of deer (Clements and Young 1997) and led to the �58 eruption of mule deer populations (Gruell 1986). Subsequent management and wild fire control has resulted in over-mature shrubs and invasion of woody species into shrub communities reducing carrying capacity for deer (Gruell 1986). This trend continues. Nutrition is key to recruitment. It is a very common conclusion that although predators can cause a short-term effect on a deer population, alternate prey species abundance dictates the density of coyotes (Hamlin et al. 1984). Increases in vegetation production has a positive effect on abundance of alternate prey species (Hamlin et al. 1984) and may explain why Salwasser (1979) observed that coyote densities and fawn survival trend in unison. Most current land use patterns in the West are detrimental to deer and rather than control of hunting pressure or predators " ... deer numbers are ultimately governed by quality and quantity of habitat" (Connolly 1981 :238). Peek et al. (2002) and Salwasser et al. (1978) suggest that long-term decline of deer populations is not the result of predation but the result of deteriorating forage conditions. Given the high reproductive potential of mule deer, it seem reasonable that improving range conditions on the Uncompahgre Plateau and other mule deer ranges of the West would be the most fruitful for increased and long-lasting improved recruitment. Ballard et al. (2001: 112) state that "The relationship between predators and their prey is a very complex issue". They list numerous possible causes for deer declines including habitat loss (i.e. food and cover), disease, predation, competition, and others. The results of the Uncompahgre Plateau study do not provide evidence to suggest that predators are the cause of low recruitment in this particular herd. Coyote predation accounted for 0.126 of the neonatal mortality with bear and feline predation accounting for 0.040 and 0.032, respectively. Whether or not this degree of predation would warrant a control program would be a societal value judgment based on both the cost and the ethics of killing one species to favor another. Although there have been no studies that demonstrate predator reduction resulted in more mule deer in the possession of hunters (Ballard et al. 2001), that is obviously the ultimate objective of predator control. Predator control is a value judgment and has segments of the public sharply divided on the need or desirability for such a program. This study has provided information for a particular study area on the extent of fawn mortality from various causes that should be of assistance to the entities that make management decisions. ACKNOWLEDGMENTS This study was a contribution of Federal Aid Project W-153-R. We especially thank the fawn capture crew members. They exhibited the patience and persistence necessary to capture, handle, and radio track a reasonable sample of fawns, which contributed to the value of the study: T. Baker, B. Banulis, P. Burke, B. Diamond, B. Dreher, J. Foster, J. Griggs, D. Gustine, B. Hoffner, B. Lamont, H. McNally, J. Risher, E. Scott, J. Skinner, D. Swanson, and S. Znamenacek. We thank W. Andelt for input during the early phases of the study. Local Division of Wildlife field personnel were instrumental in gaining permission on private land and for helpful information on field access in addition to helping capture fawns: R Arant, G. Bock, M. Caddy, D. Coven, J. Ellenberger, V. Graham, M. King, M. Mclain, K. Miller. Others that assisted in fawn capture: T. Burke, and D. Steele. We thank D. Moreno for providing predator kill figures for the area of interest. The following were instrumental in establishing field handling procedures of fawn carcasses and necropsy protocol: D. Gould, K. Larsen, G. Mason, M. W. Miller, E. Myers, B. Powers, T. Spraker, and H. Van Campen. D. Schweitzer did most of the necropsies. M. Farnsworth and S. Strain provided assistance in graphic presentation and analysis. Aircraft piloting for radio tracking, field assistance, and general support of the project provided by J. Olterman is appreciated; R. B. Gill provided administrative support and G. Miller provided administrative support and editorial comments. Planning and statistical consultation by G. White improved the overall results. M. Potter provided late season radio tracking and radio retrieval. We thank colleagues T. Beck, and C. Bishop for many helpful suggestions and discussions and for their field assistance. The interest, support, and commitment for the duration of the project by B. Watkins were valuable contributions. We thank the various reviewers of the manuscript for their constructive criticism and suggestions. �59 LITERATIJRE CITED Acom, R. C., and M. J. Dorrance. 1990. Methods of investigating predation of livestock. Alberta Agriculture Protection Branch. Edmonton. Canada. Aguirre, A. A., D. E. Hansen, E. E. Starkey, R. G. McLean. 1995. Serologic survey of wild cervids for potential disease agents in selected national parks in the United States. Preventive Veterinary Medicine 21:313-322. Andelt, W. F., M. Bruscino, and C. Niemeyer. 1998. Interpreting the physical evidence of predation on big game animals and domestic livestock. Unpublished. Anderson, A. E., D. C. Bowden, and D. M. Kattner. 1992. The puma on Uncompahgre Plateau, Colorado. Colorado Division of Wildlife Technical Publication No. 40. Fort Collins, Colorado, USA. _ _, D. E. Medin, and D. C. Bowden. 1974. Growth and morphometry of the carcass, selected bones, organs, and glands of mule deer. Wildlife Monographs Number 39. _ _, and 0. C. Wallmo. 1984. Odocoileus hemionus. Mammalian Species No. 219. American Society of Mammalogists. Anderson, R. M. and R. M. May. 1979. Population biology of infectious diseases: Part I. Nature 280:361-367. Anonymous, c.a. 1944. Range management plan Escalante Unit, Ouray District, Colorado. Bureau of Land Management files, Delta, Colorado, USA. Baker, J.C. 1995. The clinical manifestations of bovine viral diarrhea infection. Veterinary clinics of North America: Food animal practice 11 :425-445. Ballard, W. B., D. Lutz, T. W. Keegan, L. H. Carpenter, and J.C. deVos, Jr. 2001. Deer-predator relationships: a review of recent North American studies with emphasis on mule and blacktailed deer. Wildlife Society Bulletin 29:99-115. _ _, H. A. Whitlaw, S. J. Young, R. A. Jenkins, and G. J. Forbes. 1999. Predation and survival of white-tailed deer fawns in northcentral New Brunswick. Journal of Wildlife Management 63:574-479. Bartmann, R. M. 1998. A prospectus on mule deer in Colorado. Colorado Division of Wildlife, Fort Collins, Colorado, Unpublished report. Bertram, M. R. and M. T. Vivion. 2002. Moose mortality in eastern interior Alaska. Journal of Wildlife Management. 66:747-756. Browman, L. G. and H. S. Sears. 1956. Cyclic variation in the mule deer thymus. Proceeding of the Society of Experimental Biological Medicine 93:161-162. Brown, C. G. 1992. Movement and migration patterns of mule deer in southeastern Idaho. Journal of Wildlife Management 56:246-253. Cantor, A. B. 1997. Extending SAS survival analysis techniques for medical research. SAS Institute, Cary, North Carolina, USA. Chalmers, G. A., H. N. Vance, and G. J. Mitchell. 1964. An outbreak of epizootic hemorrhagic disease in wild ungulates in Alberta. Wildlife Diseases 42: 1-6. Chapman, N. and G. I. Twigg. 1990. Studies on the thymus gland of British cervidae, particularly muntjac, Munticus reevesi, and fallow, Dama dama, deer. Journal of the Zoological Society of London 222:653-675. Clements, C. D. and J. A. Young. 1997. A viewpoint: Rangeland health and mule deer habitat. Journal of Range Management 50:129-138. Connolly, G. E. 1981. Trends in populations and harvest. Pages 225-243 in 0. C. Wallmo, editor. Mule and black-tailed deer of North America. University of Nebraska Press, Lincoln, Nebraska, USA. Dietz, D.R. and J. G. Nagy. 1976. Mule deer nutrition and plant utilization. Pages 71-78 in G. W. Workman and J.B. Low, editors. Mule deer decline in the West a symposium. Utah State University, Logan. �60 Eberhardt, L. L. 1977. "Optimal" management policies for marine mammals. Wildlife Society Bulletin. 5: 162-169 . . Gaillard, J.M., M. Festa-Bianchet, and N. G. Yoccoz. 1998. Population dynamics of large herbivores: variable recruitment with constant adult survival. Trends in Ecology and Evolution 13:58-63. Garrott, R. A., G. C. White, R. M. Bartmann, L. H. Carpenter, and A. W. Alldredge. 1987. Movements of female mule deer in northwest Colorado. Journal of Wildlife Management 51:634-643. Gill, R. B., T. D. I. Beck, C. J. Bishop, D. J. Freddy, N. T. Hobbs, R.H. Kahn, M. W. Miller, T. M. Pojar, and G. C. White. 2001. Declining mule deer populations in Colorado: Reasons and responses. Colorado Division of Wildlife Special Report Number 77. 1971. Middle Park deer study - productivity and mortality. Colorado Division of Wildlife, Federal Aid in Wildlife Restoration Job Progress Report, Project W-38-R-25. July: 189-207. Ginnett, T. F. and E. L. Butch Young. 2000. Stochastic recruitment in white-tailed deer along an environmental gradient. Journal of Wildlife Management 64:713-720. Gruell, G. E. 1986. Post-1900 mule deer irruptions in the Intermountain West: Principle cause and influences. General Technical Report INT-206. Ogden, Utah, U.S. Department of Agriculture, Forest Service, Intermountain Research Station. Huegel, C. N., R. B. Dahlgren, and H. L. Gladfelter. 1985. Use of doe behavior to capture white-tailed deer fawns. Wildlife Society Bulletin 13:287-289. Hamlin, K. L. and R. J. Mackie. 1989. Mule deer in the Missouri River breaks, Montana. Federal Aid Final Report, Project Number W-120-R-7-18. Montana Department of Fish, Wildlife and Parks. _ _, S. J. Riley, D. Pyrah, A. R. Dood, and R. J. Mackie. 1984. Relationships among mule deer fawn mortality, coyotes and alternate prey species during summer. Journal of Wildlife Management 48:489-499. Hibler, C. P. 1981. Diseases. Pages 129-155 in 0. C. Wallmo, editor. Mule and black-tailed deer of North America. University of Nebraska Press, Lincoln, Nebraska, USA. Howerth, E.W., D. E. Stallknecht, and P. D. Kirkland. 2001. Bluetongue, epizootic hemorrhagic disease, and other orbivirus-related diseases. Pages 77-97, in E. S. Williams and I. K. Barker, editors. Infectious diseases of wild mammals. Iowa State University Press, Ames, Iowa, USA. Jensen, W. and W. L:Robinette. 1955. A high reproductive rate for Rocky Mountain mule deer. Journal of Wildlife Management 19:503. Kaplan, E. L. and P. Meier. 1958. Nonparametric estimation from incomplete observations. American Statistical Association Journal 53:457-481. Kie, J. G. and B. Czech. 2000. Mule and black-tailed deer. Pages 629-657 in S. Demarais and P.R. Krausman, editors. Ecology and management of large mammals in North America. Prentice Hall, Upper Saddle River, New Jersey, USA. Kufeld, R. C., D. C. Bowden, and D. L. Schrupp. 1989. Distribution and movements of female mule deer in the Rocky Mountain foothills. Journal of Wildlife Management 53:871-877. _ _. 1979. History and current status of the mule deer population on the east side of the Uncompahgre Plateau. Division Report No. 11. Colorado Division of Wildlife. Lawrence, R. K., S. Demarias, R. D. Brown, M. Abbott. 1986. Nutritional effects on thyroids, ovaries, and thymuses in white-tailed deer. Proceedings of the Annual Conference of Southeast Association of Fish and Wildlife Agencies 40:416-423. Laycock, W. A. 1991. Stable states and thresholds ofrange condition on North American rangelands: A viewpoint. Journal of Range Management 44:427-433. Marshall, M. 1998. Uncompahgre. Western Reflections, Ouray, Colorado. Medin, D. E., and A. E. Anderson. 1979. Modeling the dynamics of a Colorado mule deer population. Wildlife Monographs 68:1-77. Myers, E. P. 2001. Assessing the role of selected infectious disease agents in neonatal mule deer fawn mortality on the Uncompahgre Plateau of Western Colorado. M. S'. Thesis, Colorado State University, Fort Collins, Colorado, USA. �61 Noyes, J. H., B. K. Johnson, L. D. Bryant, S. L. Findholt, J. W. Thomas. 1996. Effects of bull age on conception dates and pregnancy rates of cow elk. Journal of Wildlife Management 60:508-517. Ozoga, J. J. and L. J. Verme. 1978. The thymus gland as a nutritional status indicator in deer. Journal of Wildlife Management 42:791-798. Panter, K. E., L. F. James, D.R. Gardner, M. H. Ralphs, J. A. Pfister, B. L. Stegelmeier, and S. T. Lee. 2002. Reproductive losses to poisonous plants: Influence of management strategies. Journal of Range Management 55:301-308. Peek, J.M., B. Dennis, and T. Hershey. 2002. Predicting population trends of mule deer. Journal of Wildlife Management 66:729-736. Picton, H. D. 1979. A climate index and mule deer fawn survival in Montana. International Journal of Biometerorology 23(2):115-122. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Ralphs, M. H. 2002. Ecological relationships between poisonous plants and rangeland condition: A review. Journal of Range Management 55:285-290. Riley, S. J. and A. R. Dood. 1984. Summer movements, home range, habitat use, and behavior of mule deer fawns. Journal of Wildlife Management 48:1302-1310. Robinette, W. L., C.H. Baer, R. E. Pillmore, and C. E. Knittle. 1973. Effects of nutritional change on captive mule deer. Journal of Wildlife Management 37:312-326. Salwasser, H. J. 1979. Ecology and management of the Devil's Garden Interstate deer herd and range. Dissertation, University of California, Berkeley, USA. _ _, S. A. Holl, and G. A. Ashcraft. 1978. Fawn production and survival in the North Kings River deer herd. California Fish and Game 64:3 8-52. Schlegel, M. 1976. Factors affecting calf elk survival in north central Idaho: A progress report. Western Association of State Game and Fish Commissioners 56:342-355. Squibb, R. C. 1985. Mating success of yearling and older bull elk. Journal of Wildlife Management. 49:744-750. Thome, E.T., E. S. Williams, T. R. Spraker, W. Helms, and T. Segerstrom. 1988. Bluetongue in freeranging pronghorn antelope (Antilocapra americana) in Wyoming: 1976 and 1984. Journal of Wildlife Diseases 24:113-119. Trainer, C. 1975. Direct causes of mortality in mule deer fawns during summer and winter periods on Steens Mountain, Oregon - a progress report. Western Association of State Game and Fish Commissioners 55: 163-170. Tsai, K., K. H. Pollock, and C. Brownie. 1999. Effects of violation of assumptions for survival analysis methods in radiotelemetry studies. Journal of Wildlife Management 63: 1369-13 75. Uncompahgre Plateau Partners (BLM, CDOW, Public Lands Partnership, USFS). 2002. Uncompahge Plateau Project (UP) Plan: ~ collaborative approach to restore and maintain the ecosystem heath of the Uncompahgre Plateau in western Colorado. Available at the Colorado Division of Wildlife Service Center, Montrose, Colorado, USA. Unsworth, J. W., D. F. Pac, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. Vale, T. R. \975. Presettlement vegetation in the sagebrush-grass area of the Intermountain West. Journal of Range Management 28:32-36. Van Campen, H., K. Frolich, and M. Hofmann. 2001a. Pestivirus infections. Pages 232-244 in E. S. Williams and I. K. Barker, editors. Infectious diseases of wild mammals. Iowa State University Press, Ames, Iowa, USA. _ _, J. Ridpath, E. Williams, J. Cavender, J. Edwards, S. Smith, and H. Sawer. 2001b. Isolation of bovine viral diarrhea virus from a free-ranging mule deer in Wyoming. Journal of Wildlife Diseases 37:306-311. Verme, L. J. 1969. Reproductive patterns of white-tailed deer related to nutritional planes. Journal of Wildlife Management 33:881-887. �62 Wade, D. A., and J.E. Bowns. 1984. Procedures for evaluating predation on livestock and wildlife. Texas Agricultural Experiment Station Bulletin B-1429, College Station. White, G. C. and R. M. Bartmann. 1998. Mule deer management- What _should be monitored? Pages 102-116 in J.C. De Vos, Jr., editor. Proceedings of the 1997 deer-elk workshop, Rio Rico, Arizona. Arizona Game and fish Department, phoenix, Arizona, USA. _ _, D. F. Freddy, R. B. Gill, and J. H. Ellenberger. 2001. Effect of adult sex ratio on mule deer and elk productivity in Colorado. Journal of Wildlife Management 65:543-551. • White, M. 1973. Description of remains of deer fawns killed by coyotes. Journal of Mammalogy 54:291-293. Young, R. G. and J. W. Young. 1984. Colorado West land of geology and wildflowers. Artistic Printing, Salt Lake City, Utah, USA. Table 1. Fate of neonatal mule deer fawns from capture (mean 3 days old) to 14 December, Uncompahgre Plateau, west central Colorado, USA, 1999-2001. Mortality Causes Sick/Starve Coyote Bear Feline Trauma Unknown Total mortality Surviving Total sample Censored 1999 2000 2001 Total 15 6 3 4 1 2 31 19 50 12 10 11 3 2 4 13 11 3 1 4 3 35 57 92 20 38 28 9 7 9 10 101 129 230 43 5 35 53 88 11 Table 2. Chi-square comparison of mortality source proportion by year for mule deer neonates from capture (mean 3 ~~ys old) to 14 December on the Uncompahgre Plateau, west central_Co~orado, USA. Kaplan-Mei_er staggered exits to account for censored subjects were used... •• Mortality Causes 1999 2000 Sick/starve Coyote Bear Feline Trauma Unknown 0.318 0.126 0.060 0.083 0.025 0.042 0.115 0.129 0.034 0.023 0.048 0.058 2001 0.148 0.121 0.033 0.012 0.047 0.037 ~ 5.330 0.023 0.416 2.330 0.568 0.388 p 0.070 0.989 0.812 0.312 0.753 0.824 �63 Figure 1. Uncompahgre Plateau mule deer Data Analysis Unit, D-19, is shown outlined in red. The 95% kernel home range for radioed does aerial located on May 18 and May 31 is outlined in black with black dots representing individual locations. Fawn capture locations are seen as white dots and the 95% kernel home range outlined in white. Blue shading shows pinyon-juniper type and higher elevation vegetative types (summer range) are shown in green, yellow, and brown. Agricultural land is in pink. See text for type descriptions. �55 JOB PROGRESS REPORT State of ------=C~o=lo=r=a=do=------- Division of Wildlife- Mammals Research Work Package No. _ _ _3=0"-'0C...:l~---- Deer Conservation Task _ _ _ _ _ _ _ _...,:le.___ _ _ _ __ Mule Deer Life Cycle- Neonatal Fawn Survival Federal Aid Project _W~--18_5_-R _ _ _ _ __ Period Covered: July I 2002 through June 30, 2003 Author: Thomas M. Pojar Personnel: W. Andelt, R Arant, D. Baker, T. Baker, B. Banulis, T. Beck, C. Bishop, G. Bock, D. Bowden, P. Burke, T. Burke, M. Caddy, D. Coven, B. Diamond, B. Dreher, J. Ellenberger, M. Farnsworth, J. Foster, V. Graham, J. Griggs, D. Gustine, P. Hayden, B. Hoffner, B. Lamont, M. King, K. Larsen, M. Mclain, H. McNally, G. Miller, M. W. Miller, E. Myers, J. Olterman, M. Potter, J. Risher, D. Schweitzer, D. Steele, J. Skinner, T. Spraker, D. Swanson, B. Watkins, G. White, S. Znamenacek. The following is the abstract of the manuscript submitted to the Journal of Wildlife Management describing the neonatal fawn survival study on the Uncompahgre Plateau. Because of requests by reviewers or editors some aspects of the presentation and analysis may be modified. Manipulation or interpretation of these data beyond that contained in this report should be labeled as such, and is discouraged. NEONATAL MULE DEER FAWN SURVIVAL IN WEST-CENTRAL COLORADO ABSTRACT Declining mule deer (Odocoileus hemionus) populations resulting from apparent low recruitment brought management and political focus on neonatal fawn survival. We captured mule deer fawns on the Uncompahgre Plateau (5,957 km2 ) in west-central Colorado, USA, at a mean age of 3 days (range from newborn to 6 days), and we radiomarked them with mortality-sensing drop-off radiocollars. Two hundred thirty fawns were radiomarked with samples of 50 in 1999, 88 in 2000, and 92 in 2001. Designated neonate survival period was from capture to 14 December. Survival was different among years (X/ = 6.160, P = 0.046) with 3-year mean survival of0.501. Cause-specific mortality ordered from highest to lowest was sick/starve, coyote, unknown, trauma, bear, and feline. Neither all predation combined (coyote, bear, and feline; P = 0.379) nor coyote predation alone (P > 0.989) differed among years. By 31 July, 74% of the sick/starve mortality and 75% of the predation mortality had taken place with 76% of mortality from all sources occurring by this date. Mean fawn weights at capture were different among years (P = 0.044). We also found a difference in hind foot length among years (P = 0.002). Weight and hind foot means were different between 2000 and 200 I (P > 0. 017) with 1999 not different from either 2000 or 2001 (P < 0.017). Mean capture date was 19 June (SD= 4.83 days) and median capture date was 19 June (range= 9 Jun to 6 Jul) with 94.78% of all captures occurring between 13 and 30 June. This implies that most does were bred during their first estrous cycle. Neonatal survival through 14 December did not completely account for observed low f:d ratios. We hypothesized fetus mortality during late pregnancy or mortality of fawns at birth (before they could be detected for capture) as potential causes of poor recruitment. � https://cpw.cvlcollections.org/files/original/1933dd5b6643a23a758d154f2f16e804.pdf 682e51cc96e1eec2b8ad71cc0a683102 PDF Text Text Colorado Division of Wildlife July 2007 – June 2008 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 3001 3 Federal Aid Project No. W-185-R : : : : Division of Wildlife Mammals Research Deer Conservation Pilot Evaluation of Predator-Prey Dynamics On the Uncompahgre Plateau Period Covered: July 1, 2007 - June 30, 2008 Authors: M.W. Alldredge, E.J. Bergman, C.J. Bishop, K.A. Logan, D.J. Freddy Personnel: B. Dunne, V. Yovovich, E. Phillips, M. Schuette All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT In an attempt to address predator-prey dynamics, we initiated a pilot study to evaluate cougar predation relative to prey distribution across the southern half of the Uncompahgre Plateau in southwestern Colorado. As part of ongoing mule deer and cougar research in this area, we estimated cougar kill rates and prey selection by sampling different sized clusters of cougar GPS locations across the landscape. Cluster size ranged from 1 location to >30 locations/cluster. In the vicinity of each sampled cluster, we searched for cougar prey items to determine whether a kill had occurred and to classify prey by species and age. This field effort was primarily focused in areas with extensive historical mule deer population and winter range distribution data. Simultaneously, a pilot effort to collect distribution and movement data of elk over this same geographic area was conducted. As predicted, cougar kill sites were associated with deer and elk distribution. The greatest density of kill sites occurred across mid-upper elevation deer winter range where overlap of wintering elk and deer was greatest. We investigated 462 clusters during this pilot study. Kill probability increased as cluster size increased ( /3ˆ = 0.353, SE = 0.0706). Kill probability exceeded 0.9 with ≥ 10 locations/cluster and approached 1 with ≥ 15 locations/cluster. The probability of a kill was high if a cougar spent >2 days in the same general area, and a kill was essentially certain if a cougar spent >3 days in the same general area. There was some probability of a kill at clusters that comprised only 1 location, indicating that isolated cougar locations may periodically be associated with kills and should not be ruled out when using GPS location data to address cougar prey utilization. Our estimates of kill probability are conservative because the estimates assume detection probability was 1, which is unlikely. Cougars killed adult deer, fawn deer, adult elk, and calf elk in roughly equal proportions. Each prey class comprised 0.22 0.24 of the total kill. Kill composition varied as a function of percent vegetative cover and elevation. Future research should evaluate detection probability, which underlies the interpretation of cougar kill rates. 87 �WILDLIFE RESEARCH REPORT PILOT EVALUATION OF PREDATOR-PREY DYNAMICS ON THE UNCOMPAHGRE PLATEAU MATHEW W. ALLDREDGE, ERIC J. BERGMAN, CHAD J. BISHOP, KENNETH A. LOGAN, AND DAVID J. FREDDY P.N. OBJECTIVE To assess if a sampling based approach to collecting cougar predation data can efficiently result in unbiased data. To make a pilot assessment of how cougar kills are spatially distributed over prey winter range. SEGMENT OBJECTIVES 1. Use and evaluate the efficiency of a GPS collar, GIS and statistical sampling based approach to investigate potential cougar kill sites. 2. Estimate mule deer density on three study areas and extrapolate results onto surrounding mule deer range. 3. Overlay locations of 5 elk, collected via GPS collars, on mule deer winter range boundaries to gain preliminary information as to how much spatial overlap occurs between the species and to determine where cougar kills occur in relation to the mule deer and elk space use. INTRODUCTION Predator prey interactions have always been a topic of interest for wildlife managers and ecologists. However, due to the complexities of studying natural systems, behavioral theories pertaining to the subject are often developed in invertebrate, aquatic or small mammal systems, often under controlled laboratory conditions (Mathews et al. 2006, Schmitz 2006, Werner and Peacor 2006). Similarly, many models are developed within theoretical frameworks (Keeling et al. 2000, Mitchell and Lima 2002). While developing theories under these conditions is almost inherently necessary, their subsequent transition to free ranging systems is not frequent (Ryall and Fahrig 2006). Of the free ranging systems where theories are developed and tested, most deal with avian species (Lima and Bednekoff 1999, Roth et al. 2006), where as application to large mammalian systems is less frequent. Of the mammalian predator prey systems that have been studied, most have been conducted in preservation/park settings that largely exclude human influence (Kunkel and Pletscher 1999, Kunkel et al. 1999, Krebs et al. 2001, Creel and Creel 2002, Mao et al. 2005, Wilmers et al. 2006,). Additionally, due to the small number of large scale studies that have been conducted, the ability of managers to draw inference to separate systems (i.e. different species or different ecosystems) is limited. While this existing body of work is invaluable, extrapolation of theories to large mammalian systems could be limited and basing wildlife management decisions on this information may be tenuous. Due to the value of mule deer, elk and cougars as recreationally hunted species in Colorado, there is much interest in understanding the nature and relationship between the population dynamics of these species. However, resulting from the dearth of information pertaining to the interactions of these 3 species, a vast array of opinions and theories pertaining to their impacts on each other have been propagated. As a management agency, the Colorado Division of Wildlife is responsible for supporting or refuting claims with biological data that were collected in a scientifically unbiased manner. To date, these data are largely unavailable. 88 �Currently, the opportunity to develop a predator prey study exists on the Uncompahgre Plateau in southwestern Colorado. Two large scale research programs, independently studying cougar and mule deer, are underway in the same geographic area. Thus, the initial framework to study a top carnivore, and what are thought to be its primary prey species, is in place. However, to date there is little or no information pertaining to elk distribution or population dynamics in this area. The addition of elk spatial data will allow us to assess the feasibility of developing a full study addressing the influence and interactions of cougars, mule deer, and elk. STUDY AREA This pilot study was conducted on the southern half of the Uncompahgre Plateau in southwestern Colorado, near Montrose, Colorado (Figure 1). The study area was defined by the existing boundary for the ongoing cougar research project with prey populations being monitored only in the eastern half of the cougar study area. METHODS Capture and Handling Methods As part of completed, as well as ongoing mule deer research, approximately 75 adult female mule deer were marked with VHF radio collars in the area of interest (Bishop et al. 2005). Additionally, 25 mule deer fawns were captured and radio-collared within the eastern portion of the study area between late-November and late-December 2006 as part of the ongoing mule deer research (Bergman et al. 2005; capture protocols previously approved by CDOW ACUC). All mule deer were captured with baited dropnets (Ramsey 1968, Schmidt et al. 1978, Bartmann et al. 1992) or via helicopter net-gunning (Barrett et al. 1982, van Reenen 1982). As part of the ongoing cougar research project, 19 cougars (15 female, 4 male) were outfitted with GPS collars that allowed on-demand data download interaction with researchers. Cougars were captured primarily via pursuit by dogs as well as in live traps (Logan 2005, capture protocols previously approved by CDOW ACUC). As part of this pilot study, adult female elk (9) were captured via helicopter net-gunning during late-December/January 2006-07 with 5 adult females fitted with drop-off GPS/VHF collars and 4 adult females fitted with VHF permanent collars. Elk were captured on the eastern portion of the study area, directly overlapping areas including radio collared mule deer and cougar. Sample sizes for elk reflected an estimate of what we believed to be an adequate number of elk to provide an initial estimation of elk spatial use in the study area. Ungulate Survival and Location Monitoring On a daily basis, from December through May, we monitored radioed fawns and adult female deer and elk in order to document live/death status. This allowed us to determine accurately the date of death and estimate the proximate cause of death. For animals not heard from the ground, we conducted weekly flights to assess live/death status. Detailed locations of GPS collared elk became available when self-actuating mechanisms caused the GPS collars to drop-off elk in September 2007. Elk GPS collars collected locations every 30 minutes. Identification of Cougar GPS Location Clusters Characteristics of clusters of GPS locations representing cougar-killed ungulate sites (Anderson and Lindzey 2003, Logan 2005) were used to develop a standard algorithm to group GPS points together, to provide a sound sampling frame from which statistical inference could be made about clusters that are not physically investigated. GPS collars collected locations 4 times/day to reflect time periods when cougars are both active and inactive (00:00, 6:00, 12:00 and 19:00). The clustering routine was designed to identify clusters in five unique selection sets in order to identify clusters containing two or more points, those that contained missing GPS locations, and those that were represented by single points. The clustering algorithm was written in Visual Basic and was 89 �designed to run within ARCGIS (Alldredge and Schuette, CDOW unpubl. data 2006). The widths of the spatial and temporal sampling windows were user specified, in order to meet multiple applications and research needs. This also enabled adjustment of the sampling frames to improve cluster specifications as needed. The initial step was to prepare data files for ARCGIS. The main priority was to number all downloaded GPS lat-long location records consecutively to provide a time stamp that could be used in the program. Failed locations were numbered within the data files to maintain the proper time step (i.e. two locations that were separated by a missing location were time stamped in such a way that the clustering algorithm recognized that a missing location existed between the records). At this point data files were imported to ARCGIS and coordinates converted to UTMs. The initial selection set of clusters (S1) were based on clusters consisting of two or more points within a specified distance and time interval. Working with temporal and spatial variables simultaneously is difficult, so we chose to create an association matrix of the combined variables. The units for time were based on GPS locations so that the time between consecutive downloads was one. Cougar locations are attempted 4 times a day, so that one day consisted of 4 time-steps. The association matrix was then constructed as 1 Aij d max e 1 dij ti tj tmax where Aij was the association in time and space between points i and j, dmax was the maximum distance between two points to be considered a cluster, dij was the distance between points i and j, tmax was the maximum number of time steps between points to be considered in a cluster, and ti and tj were the times for locations i and j. This formula weighted the distance between two locations heavier than the time between two locations. It also caused the association Aij to be negative for any locations that were outside the temporal window (separated by more time-steps than tmax). The association between two locations within the specified time interval was greatest for those locations that were spatially closer together. So, the largest value in the association matrix corresponded to the 2 points that were spatially the closest and within the time interval. Initially, dmax was set at 200 m and tmax was set at 16 time steps [4 DAYS] . The initial cluster was selected by choosing the 2 points with the largest association value from the association matrix. The distance was checked to verify that the points were within the specified maximum distance, dmax, and if so, the centroid of the two points was calculated. An association vector Ac was made by calculating the association among the centroid and all other points using the above formula. If all values in Ac were negative, then no points were within the specified time interval, so no additional points were added to the cluster. Then the greatest association value Acmax was selected from Ac and the distance from the centroid to the point corresponding to Acmax was compared to dmax. If the distance was less than dmax then the point was added to the cluster and a new centroid was calculated using all cluster points and a new vector Ac was constructed using the new centroid. This procedure was repeated until no additional points were added to the cluster because either no points were within the specified time interval or the distance from the centroid to all points was greater than dmax. After each cluster was constructed these points were omitted from the association matrix and a new cluster was started by again selecting the greatest value from the matrix and verifying that the distance between points was less than dmax. Points were again added to this cluster as previously described. This entire procedure was repeated until no 2 locations met the temporal or spatial criteria. 90 �All clusters were given a unique identifier, which was based on the animal identification and the Julian date. This completed the selection set for clusters with two or more locations, which were likely to have a high probability of being a kill site. Additional selection sets were constructed from the remaining points as single location clusters. However, not all locations are equal, so the remaining selection sets were created based on whether points were associated with missing locations and based on distance between consecutive locations. The second selection set (S2) of clusters was created from any 2 points that were within a distance dmiss, and were separated by 1 or more missing locations. The cluster was considered to be the area within the distance dmax of each of the known locations (2 areas make up the cluster, and dmiss was initially set at 500 m). The final 2 cluster selection sets consisted of consecutive points that were within the ranges dmax to d2 (S3) and d2 to d3 (S4). To construct these selection sets, the distance between consecutive points was examined and if the distance was within the range dmax to d2 (500 m) then the initial point was added as a cluster to the set S3, or if the distance was within the range d2 to d3 (1000 m) then the initial point was added as a cluster to the set S4. These single-point clusters were assumed to have radius dmax. Points not used in selection sets S1 through S4 were then used in a final selection set S5. These points represented larger movements between consecutive locations and thus were thought to have low probabilities of being associated with a kill site, although these points could be associated with use of small prey items, or kill sites where a cougar was physically disturbed away from a kill site. These single-point clusters were also assumed to have radius dmax. Sampling of Cougar GPS Location Clusters A primary objective of the pilot study was to determine the probability that a given cluster represented a cougar feeding site. Specifically, to evaluate cougar feeding sites as a function of the cluster association matrix. Using the clustering algorithm described above, we attempt to classify each sampled cluster as a cougar feeding site (1) or not a feeding site (0). We expected a high proportion of S1 clusters to represent cougar feeding sites. Conversely, we expected a moderate proportion of S2 and S3 clusters, and a low proportion of S4 and S5 clusters, to represent cougar feeding sites. A secondary objective of the pilot study was to gather preliminary biological data regarding cougar prey utilization, primarily with respect to deer and elk. The secondary objective was most efficiently accomplished by sampling S1 clusters with greater intensity than other clusters. We therefore structured our sampling approach to allow adequate estimation of the proportion of clusters that were cougar feeding sites for each cluster set, while more intensively sampling S1 clusters than all others. With no previous evidence to indicate similarities among individuals based on sex, age, or parental status, sampling was stratified by each individual cougar. GPS collars were downloaded once a month for each cougar and data were analyzed through the clustering algorithm. Clusters within 2 weeks of the download date were selected for the sampling frame, making the maximum time between the predation event and sampling about 1 month by the time field technicians could get to and assess evidence at each cluster site. Clusters were randomly chosen from each selection set for each individual cougar every month in the following manner: S1 = 2 clusters, S2 = 1 cluster, S3 = 1 cluster, and S4 and S5 = 1 cluster on alternating months. Five clusters were sampled each month for each cougar, for a total of 30 clusters per cougar from 1 November 2006, to 15 July 2008. As time allowed, additional clusters were sampled from the selection sets. Our approach forced constant sampling of each cluster set over time regardless of the frequency of clusters within a given set. This prevented a scenario where nearly all sampled clusters in a given month were from sets, S3, S4 and/or S5 (i.e., low probability of finding feeding sites). Our assessment of prey utilization depended on relatively constant detection of cougar feeding sites over time to avoid bias. 91 �However, for each cluster set, the true proportion of clusters representing feeding sites may possibly change over time corresponding to changes in cougar use of feeding sites. If the GPS download data indicated major changes in set-specific cluster frequencies over the sampling period, we maintained the ability to use a proportional-allocation sampling approach if needed. Assuming a binomial distribution and 0.90 of S1 clusters represented cougar feeding sites, our approach enabled us to estimate the true proportion with a 95% confidence interval of +/ 0.07. Assuming 0.5 of S2 clusters represented cougar feeding sites, we were able to estimate the true proportion with a 95% confidence interval of +/ 0.17. Assuming 0.3 of S3 clusters represented feeding sites, we were be able to estimate the true proportion with a 95% confidence interval of +/ 0.15. Finally, assuming 0.1 of S4 and S5 clusters represented feeding sites, we were able to estimate the true proportion with a 95% confidence interval of +/ 0.10. These precision levels were deemed acceptable for the pilot study, and should facilitate development of an optimal sampling scheme in future years for evaluating cougar prey utilization from GPS cluster-location data. Finally, regarding our secondary objective of collecting preliminary prey use data, we were able to estimate the overall proportion of kill sites represented by deer (or the proportion of kill sites represented by elk) with a 95% confidence interval of +/ 0.05 (Anderson and Lindzey 2003, Logan 2005). We used the following protocol to investigate cougar GPS clusters in the field. For S1 clusters, we investigated each cougar GPS location in the cluster by spiraling out a minimum of 20 m from the GPS waypoint while using the GPS unit as a guide, and visually inspecting overlapping view fields in the area for prey remains. Normally, this was sufficient to detect prey remains and other cougar sign (e.g., tracks, beds, toilets) associated with cougar. If prey remains were not detected within 20 m radius of the cluster waypoints, then we expanded our searches to a minimum of 50 m radius around each waypoint. The 20 m and 50 m radius search areas resulted in overlapping view fields of individual waypoints, and took up to 7 hours to complete, depending upon the number of waypoints, topography, and vegetation type and density associated with a cluster. For S2 through S5 clusters, we went to each cougar GPS location and spiraled out 50 m around each waypoint, while using the GPS unit as a guide. Depending on the number of locations, topography, and vegetation type and density, we spent a minimum of 1 hour and up to 3 hours per cluster to judge whether the cluster was a kill site. Estimating Deer, Elk, and Cougar Distributions We examine locations, movements, and kernel home ranges of mule deer, elk, and cougars for spatial overlap and time synchrony using ArcGIS. Our initial analyses are descriptive and should provide insight into patterns of cougar movements and feeding sites in relation to major ungulate species. Based on past observations, we did not expect deer distributions to fluctuate greatly during the winter. However, we did expect elk distributions to fluctuate depending on weather and time. We anticipated being able to generate correlations between species of prey killed by cougars and the relative presence of prey within cougar home ranges. Cougar GPS Cluster Analysis We estimated the probability of locating cougar prey items (i.e., cougar kills) at GPS location clusters using logistic regression in SAS (PROC LOGISTIC; SAS Institute, Cary, NC). We modeled cougar kills as a function of cluster type (S1, S2,…, S5), cluster size (no. locations/cluster), cougar status (adult female with cubs, adult female without cubs, adult male), and season when cluster was investigated (winter, spring, summer, fall). We then analyzed kill composition using a generalized logits model (i.e., multinomial logistic regression) in SAS (PROC LOGISTIC). For this analysis, we used only clusters where prey items were found (i.e., kills). Kill composition was divided into 5 categories: adult deer, fawn deer, adult elk, calf elk, and other (i.e., porcupine, coyote, turkey, unknown). We modeled kill composition as a function of cluster type, cluster size, cougar status, season when kill occurred, elevation, 92 �and percent vegetative cover. We used Akaike‘s information criterion adjusted for sample size (AICc) to select among candidate models in both modeling analyses (Burnham and Anderson 2002). Hypothesis Testing Our preliminary sampling effort of cougar clusters and ungulate distributions provided estimates of cougar kill rates and proportions of deer and elk killed. As data collection continues, we intend to address whether 1) cougar prey mass is positively related to cougar mass (i.e., male cougars kill larger prey than female cougars), 2) cougars prey on deer and elk in proportion to availability (i.e., no selection for prey species), 3) cougars prey on sex or ages of deer or elk populations in proportion to availability (i.e., no selection for prey age classes), 4) cougars alter their use of prey among seasons of the year (i.e., prey-switch between deer and elk, or between juvenile and adult), and 5) maternal cougar home ranges include the highest available densities of ungulate prey. RESULTS AND DISCUSSION Mule Deer Distribution As expected, over the course of the winter, mule deer movements occurred at too fine of spatial and temporal scales to be detected without more intense repeated sampling. However, as they relate to the pilot study, the data gathered are adequate for making basic summaries. Mule deer density appeared to be highly variable across a gradient of winter range (density estimates ranged between 19 and 109 deer/km2 , Figure 1) (Bergman et al. 2008). Relative to the entire Uncompahgre Plateau, the estimates tended to be high, confirming historical information and further justifying the decision to conduct pilot work in this area. Of particular interest in regards to spatial overlap between cougar kill sites and mule deer winter range, the majority of located kill sites were higher in elevation than the greatest concentrations of mule deer. The exception to this trend occurred on the southern most portion of deer winter range where the majority of kill sites were composed of mule deer. As discussed below, an apparent explanation for this may be linked to elk distribution as this area also appeared to be the area of greatest overlap between mule deer and elk. To improve future efforts, several key steps would need to be taken. Mule deer density estimates are relatively course for the majority of winter range included in this pilot study. With the exception of three polygons, deer density estimates were extrapolated from surrounding areas. Furthermore, estimates of deer density for the 3 areas were not collected during the same year and therefore include annual variation. To accurately reflect the conditions, albeit still at a course level, encountered by cougars as they move across mule deer winter range, density estimates should minimally be collected on all segments of winter range on an annual basis. While fine scale movements of deer (i.e. daily movements within winter range) were not incorporated in this study, such data likely would not be hugely beneficial. Fine scale data would be of greatest interest if the focus of the study were shifted to analyze/describe fine scale hunting behavior of individual cougars. Elk Movement and Distribution Elk GPS collar data confirmed our initial expectations that elk movements during winter months were more dynamic than those of deer. The four elk collared with VHF collars left the study area of interest after 7 months and collecting repeated aerial locations was not deemed worthwhile as they were not in areas with radio marked deer or cougar. However, elk did appear to be highly individualistic in regards to space use and movement during winter months. Two elk appeared to concentrate locations over a relatively large geographic area (>75 km2) during the winter months, but restricted movements to stay within these areas. The other 3 elk appeared to utilize relatively small spatial areas (9-10 km2) for 12 week periods before making slightly longer movements (10+ km) to new concentration areas. Plotting known locations for cougar kill sites on elk spatial data suggested that cougar kill sites had a strong correlation to elk distribution (Fig. 2). Based on the more dynamic nature of elk movement during winter, 93 �future efforts to map elk distributions and densities would be better met by saturating the area of interest with GPS collared elk. Annual density estimates, collected via helicopter, would likely only be valid for a relatively short time period (2-4 weeks) due to elk movement and thus making it difficult to track cougar space use and predation patterns in a realistic prey context. By outfitting a large number of elk with GPS collars, resource selection functions for the elk in the area of interest could be built around habitat and elevation selection patterns. Due the large amount of data collected by GPS collars, resource selection functions could justifiably be built at 2-4 week intervals. Kill Probability Associated with Cougar GPS Clusters We investigated 462 clusters during this pilot study (195 S1 clusters, 33 S2 clusters, 71 S3 clusters, 73 S4 clusters, 90 S5 clusters). The probability of locating cougar kills at GPS location clusters varied as a function of cluster type, cluster size, cougar status, and season (Table 1). As expected, S1 clusters were far more likely to be associated with cougar kills than S2 S5 clusters (Figure 3). The probability of a kill at an S1 cluster was 0.505 (95% CI: 0.435, 0.575), whereas kill probability was ≤ 0.12 at all other cluster types. There was some probability of a kill at S4 and S5 clusters, indicating that isolated cougar locations may periodically be associated with kills and should not be ruled out when using GPS location data to address cougar prey utilization. Kill probability increased as cluster size increased ( ˆ = 0.353, SE = 0.0706). Kill probability exceeded 0.9 with ≥ 10 locations/cluster and approached 1 with ≥ 15 locations/cluster (Figure 4). Thus, the probability of a kill was high if a cougar spent >2 days in the same general area, and a kill was essentially certain if a cougar spent >3 days in the same general area. Models receiving the most weight also provided evidence of interactions between cluster size and cougar status and between cluster size and season. The cluster size × cougar status interaction occurred because smaller cluster sizes were more likely to be associated with kills for female cougars than male cougars (Figure 5). For example, female cougars with ≥ 10 locations/cluster indicated a near-certain kill, whereas male cougars with 10 locations/cluster indicated only 0.571 probability of a kill (95% CI: 0.267, 0.830). Adult males were more likely to spend multiple days in an area without a kill than were adult females. The cluster size × season interaction occurred because larger cluster sizes during summer were less likely to indicate a kill than during other seasons (Figure 6). Perhaps cougars were more likely to remain sedentary without a kill nearby during summer months when energetic demands were lower. This result should be interpreted with caution, however, because we collected less data during summer than during other seasons. f/ Our primary reason for including season in the analysis was to evaluate possible differences in detection probability. We expected kills to be difficult to detect during winter and possibly spring months when carcasses and sign would be periodically covered by snow. However, our results did not support this hypothesis, but instead suggested that kills may have been the most difficult to detect during summer. Kills may be difficult to detect in summer range habitats because of extensive foliage or increases in scavenging by bears and/or coyotes. Regardless, carcass detection probability is a significant issue that underlies our entire analysis. That is, it is difficult to fully interpret our findings above without an adequate understanding of detection probability. For example, our summer results could reflect reduced carcass detection probability during summer, or they could reflect changes in cougar behavior during summer as compared to other months. A key point is that our estimates of kill probability for different cluster types and sizes are minimum estimates because these estimates assume detection probability was 1, which is unlikely. Detection probability should be addressed in subsequent research. Cougar Kill Composition Cougars killed adult deer, fawn deer, adult elk, and calf elk in nearly equal proportions (Figure 7). Each prey class comprised 0.22 0.24 of the total kill. Kill composition varied as a function of percent vegetative cover and elevation (Table 2). Adult elk were more likely to be killed in areas with little cover whereas calf elk, adult deer, and other species were more likely to be taken in habitats with heavier cover 94 �(Figure 8). Adult elk and adult deer were more likely to be killed at lower elevations whereas calf elk and other species were more likely to be killed at higher elevations (Figure 9). Unexpectedly, kill composition did not vary in response to cluster type or cluster size (Figure 10). Kill composition could be biased if S1 clusters, or larger cluster sizes, were associated with larger prey items, because it would suggest that larger prey may be more easily detected. However, given that kills of different sized prey occurred in roughly equal probabilities across all cluster sizes, restricting sampling to larger clusters would not necessarily bias kill composition estimates, at least for ungulates. Efficiency would be gained in the field by sampling larger clusters because they are more likely to be associated with kills. Additional data collection will be necessary to determine whether this preliminary finding is valid. Also, we urge caution interpreting this result because it is not biologically intuitive and would lead to biased kill composition data if proven incorrect. SUMMARY Over the past 2 years we have collected data on elk and deer distributions in conjunction with cougar predation data across the southern half of the Uncompahgre Plateau. Part of this effort included the development and implementation of a sampling based approach to estimate cougar kill rates and prey selection from GPS location data. Based on this effort we were able to randomly sample clusters of cougar GPS locations in relation to cluster type/size, which presumably correlates to prey selection and handling time. Mule deer and elk distributions on winter range were as expected with mule deer utilizing lower elevations and elk utilizing both lower and higher elevations with an area of overlap between the two species across deer winter range. Interestingly, cougar kill sites for mule deer generally occurred at midelevations within the range of overlap for deer and elk. Cougar kill sites for elk occurred at all elevations characteristic of elk distribution. As expected, cougar clusters with a large number of points had a high probability of being associated with a predation event and those with few points had a lower probability, especially single point clusters that are spatially distinct from other points. However, evidence of predation was identified at some of the spatially distinct single point clusters, indicating that these types of clusters are important in accurately describing cougar diet composition and predator/prey interactions. The association between cluster size and the probability of a cougar kill was related to season and cougar sex, with larger clusters being less predictive of a kill during summer and for males. Cougars killed elk and deer in approximately equal proportions and killed fawns/calves in equal proportion to adults for both deer and elk. Other prey items that could be detected at GPS locations comprised less than 10% of cougar diets. LITERATURE CITED Anderson, C. R., Jr., and F. G. Lindzey. 2003. Estimating cougar predation rates from GPS location clusters. Journal of Wildlife Management 67:307-316. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108-114. Bartmann, R.M. 1983. Composition and quality of mule deer diets on pinyon-juniper winter range, Colorado. Journal of Range Management 36:534-541. Bartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monographs 121:5-39. Bergman, E.J., C.J. Bishop, D.J. Freddy and G.C. White. 2005. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Wildlife Research Report, July: 23-35. Colorado Division of Wildlife, Fort Collins, USA. 95 �Bergman, E.J., C.J. Bishop, D.J. Freddy and G.C. White. 2008. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Wildlife Research Report, July: In Press. Colorado Division of Wildlife, Fort Collins, USA. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. Second Edition. Springer-Verlag, New York, New York, USA. Creel, S. and N.M. Creel. 2002. The African wild dog: behavior, ecology and conservation. Princeton University Press, New Jersey, USA. Keeling, M.J., H.B. Wilson and S.W. Pacala. 2000. Reinterpreting space, time-lags, and functional responses in ecological models. Science 290:1758-1761. Krebs, C.J., S. Boutin and R. Boonstra, eds. 2001. Ecosystem dynamics of the boreal forest: the Kluane project. Oxford University Press, New York, USA. Kunkel, K. E., and D. H. Pletscher. 1999. Species-specific population dynamics of cervids in a multipredator ecosystem. Journal of Wildlife Management 63:1082-1093. Kunkel, K. E., T. K. Ruth, D. H. Pletscher, and M. G. Hornocker. 1999. Winter prey selection by wolves and cougars in and near Glacier National Park, Montana. Journal of Wildlife Management 63:901-910. Lima, S.L. and P.A. Bednekoff. 1999. Back to the basics of antipredatory vigilance: Can nonvigilant animals detect attack? Animal Behaviour 58:537-543. Logan, K.A. 2005. Cougar population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report, July: 105-126. Colorado Division of Wildlife, Fort Collins, USA. Mathews, C.G., J.A. Lesku, S.L. Lima and C.J. Amlaner. 2006. Asynchronous eye closure as an antipredator behavior in the western fence lizard (Sceloporus occidentalis). Ethology 112:286-292. Mao, J.S., M.S. Boyce, D.W. Smith, F.J. Singer, D.J. Vales, J.M. Vore and E.H. Merrill. 2006. Habitat selection by elk before and after wolf reintroduction in Yellowstone National Park. Journal of Wildlife Management 69:1691-1707. Mitchell, W.A., and S.L. Lima. 2002. Predator-prey shell games: large-scale movement and its implications for decision making by prey. Oikos 99:249-259. Ramsey, C. W. 1968. A drop-net deer trap. Journal of Wildlife Management 32:187-190. Roth, T.C., II, S.L. Lima and W.E. Vetter. 2006. Determinants of predation risk in small wintering birds: the hawk's perspective. Behavioral Ecology and Sociobiology 60:195-204. Ryall, K.L. and L. Fahrig. 2006. Response of predators to loss and fragmentation of prey habitat: a review of theory. Ecology 87:1086-1093. Schmidt, R. L., W. H. Rutherford, and F. M. Bodenham. 1978. Colorado bighorn sheep-trapping techniques. Wildlife Society Bulletin 6:159-163. Schmitz, O.J. 2006. Predators have large effects on ecosystem properties by changing plant diversity, not plant biomass. Ecology 87:1432-1437. van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, Wisconsin, USA. Werner, E.E. and S.D. Peacor. 2006. Lethal and nonlethal predator effects on an herbivore guild mediated by system productivity. Ecology 87:347-361. Wilmers, C.C., E. Post, R.O. Peterson and J.A. Vucetich. 2006. Predator disease out-break modulates top-down, bottom-up and climatic effects on herbivore population dynamics. Ecology Letters 9:383-389. Prepared by Mathew W. Alldredge, Eric J. Bergman and Chad J. Bishop, Wildlife Researchers 96 �Table 1. Model selection results, based on Akaike‘s Information Criterion with small sample size correction (AICc), of an analysis evaluating the probability of locating cougar prey items (i.e., cougar kills) at GPS location clusters. We modeled cougar kills as a function of cluster type (type; S1, S2,…, S5), cluster size (size; no. GPS locations/cluster), cougar age and sex status (status), and season when cluster was investigated (season; spring, summer, fall, winter). Model Type size season size×season Type size status season size×status size×season Type size status season size×season Type size Type size season Size status season size×status size×season Size season size×season Type size status season Size status season size×season Size status size×status No. Parameters AICc Delta AICc Model Weight 12 16 14 6 9 12 8 11 10 6 365.61 367.84 369.68 370.75 371.80 373.04 374.80 375.71 378.87 380.07 0.00 2.22 4.07 5.13 6.19 7.42 9.19 10.10 13.25 14.46 0.615 0.202 0.081 0.047 0.028 0.015 0.006 0.004 0.001 0.000 Table 2. Model selection results, based on Akaike‘s Information Criterion with small sample size correction (AICc), of an analysis evaluating cougar kill composition at GPS location clusters. We modeled kill composition as a function of cluster type (type; S1, S2,…, S5), cluster size (size; no. GPS locations/cluster), cougar age and sex status (status), season when kill occurred (season; spring, summer, fall, winter), elevation (elev), and percent vegetative cover (cover). Model Elevation cover Elevation cover status Cover Elevation Size elevation Status Status season size elevation cover Size Season Status season size elevation No. Parameters AICc Delta AICc Model Weight 12 20 8 8 12 12 36 8 16 32 347.81 353.25 356.35 366.44 371.41 385.84 386.30 386.47 389.36 399.07 0.00 5.45 8.54 18.64 23.60 38.04 38.49 38.67 41.55 51.27 0.926 0.061 0.013 0.000 0.000 0.000 0.000 0.000 0.000 0.000 97 �,.. .v ,ow,o;o~ ........ '!!!I/J , -dT tt0mo1E ~,mm •~ LION CLOSURE " <• tOtOM ~ , ~ DrMSITT •• T Figure 1. Location of pilot predator-prey research on the Uncompahgre Plateau, southwest Colorado. Ongoing deer research study areas are reflected by red and blue polygons with hash marks, as well as by solid yellow polygons. The ongoing lion research study area is designated by the large red polygon. · L ·,:";..t:•: {~;(j ~:.{t;\ ,: l ; ':!\~.~ lJl .~_/>:/j ':~.,.• Figure 2. Distribution of cougar kill sites (red circles) in relation to mule deer winter range on the southeast portion of the Uncompahgre Plateau, Colorado. Black polygons represent segments of mule deer winter range where density estimates were either estimated or extrapolated to by surrounding areas on which estimates were measured. Gray lines represent Game Management Unit boundaries as designated by the Colorado Division of Wildlife. 98 �,:,/. • · ~ ,~~~ • ..,•·-i~·· ~ ~~lft .\=;·--~.}:1:-: ·· ::,_.~:, :·{ \,-; . ~~~7.:, ·~~- ' !:1 : i; :u:~·.: ,. ~ }t ~ f~~f- S1 S2 ., ,·,; ....Jil:::~-: l;Ii Figure 3. Distribution of cougar kill sites (red circles) in relation to GPS collar locations for 5 elk (black circles) on the southeast portion of the Uncompahgre Plateau, Colorado. Gray lines represent Game Management Unit boundaries as designated by the Colorado Division of Wildlife. 0.7 0.6 0.5 :-2 .... 0 > 0.4 ~ :.cro .0 0.3 ~ ci. 0.2 0.1 0 S3 S4 S5 Cluster Type Figure 4. Probability of a cougar kill at different types of GPS location clusters (with 95% CIs), Uncompahgre Plateau, Colorado, 2006 2008. Refer to the Methods section for a detailed explanation of cluster types. 99 �1 0.9 0.8 0.7 :.2 .... 0.6 0 > ~ :.cro 0.5 .0 ... 0.4 0 c.. 0.3 0.2 0.1 0 1 6 11 16 21 26 31 36 No. Locations in Cluste r Figure 5. Probability of a cougar kill as a function of the number of locations in a GPS cluster (with 95% CI), Uncompahgre Plateau, Colorado, 2006 2008. ... -- -- l o.s 0.8 0.7 ....~ 0.6 g 0 .5 0 ii ro .,Q ... 0.4 0 Q,. 0.3 - AdFemSingle 0.2 - - Ad f-emlubs 0.1 -- AdMale 0 0 5 10 15 20 No. Cluste r Locations 25 30 35 Figure 6. Probability of a cougar kill at GPS location clusters relative to sex and reproduction status (with 95% CIs), Uncompahgre Plateau, Colorado, 2006 2008. Cougar status was defined as single adult female (AdFemSingle), adult female with cubs (AdFemCubs), or adult male (AdMale). 100 �1 0.9 0.8 0.7 ~ 0.6 0 ~ 0.5 - - - Spri ng :c10 ..c e 0.4 - ~ 0.3 - Winter - - Fal l 0.2 - 0.1 - Summer - -- -- -- 0 0 5 10 15 20 25 30 35 No. Cluster Locations Figure 7. Probability of a cougar kill at GPS location clusters by season (with 95% CIs), Uncompahgre Plateau, Colorado, 2006 2008. 0 .35 0.3 - ,_ 0.2 5 :;2 - iv 0 I- 0 .2 0 C: 0 ·z..... 0.1 5 0 C. -- ,- -- - 0 .1 - 0 .05 - 0 ..... ~ - - -- 0 AdDeer AdElk FwnDeer CalfElk Other Figure 8. Prey composition of cougar kills (with 95% CIs) on the Uncompahgre Plateau, Colorado, 2006 2008. Prey items included adult deer (AdDeer), ≥ 6-month-old fawn deer (FwnDeer), adult elk (AdElk), calf elk (CalfElk), and other species (e.g., porcupine, turkey, coyote). 101 �0.8 0.7 0.6 I -Adult Elk I 0.5 0.4 0.3 0.2 0.1 0 Predicted probabilities 0 20 40 60 80 100 0.6 0 .5 - Calf Elk 0.4 -Adult Dee~ - 0.3 0.2 0. 1 0 0 20 40 60 80 100 60 80 100 0.8 0.7 0.6 -Fawn Deer 0.5 - Other 0.4 0.3 0.2 0.1 0 0 20 40 Percent cover Figure 9. Predicted prey composition of cougar kills as a function of vegetative cover (with 95% CIs), Uncompahgre Plateau, Colorado, 2006 2008. 102 �0.7 - - - - - - - - - - - - - - - - - - - 0.6 - + - - - - - - > , - - - - - - -- - - - - - - -- - - - -Adult Deer 0.5 -+------'<-- - > o r - - - - - -----< Predicted probabilities - Adult Elk 0 + - - - ~ - - ~ - - - - - ~ _ _ _ _ _ : := - - - - - , r - - - -, - - 1700 1900 2100 2300 2500 2700 2900 1 0.9 -Fa wn Deer , , I 0.8 I - 0.7 I Calf Elk I I - -Other 0.6 0.5 0.4 ,,, 0.3 ~ 0.2 0.1 --- --- 0 1700 1900 2100 2300 2500 2700 2900 Elevation Figure 10. Predicted prey composition of cougar kills as a function of elevation (m) (with 95% CIs), Uncompahgre Plateau, Colorado, 2006 2008. 103 �0.8 0.7 - Fawn Deer 0.6 - Ca lf Elk 5 10 - Adult Deer - Adult Elk 0.5 0.4 0.3 Predicted probabilities 0.2 0.1 0 0 15 20 25 30 35 30 35 0.6 0.5 - - Other 0.4 0.3 0.2 0.1 ------- .......... 0 0 5 10 - --- -- - - --15 20 25 --- No. locations/cluster Figure 11. Predicted prey composition of cougar kills as a function of the number of locations comprising a GPS location cluster (with 95% CIs), Uncompahgre Plateau, Colorado, 2006 2008. 104 � https://cpw.cvlcollections.org/files/original/b4987169ef2fa5ba06e7e0af8454e131.pdf eb60656e45a79c8ce076d06579b82a7c PDF Text Text 65 Colorado Division of Wildlife Wildlife Research Report July 2001 and July 2002 JOB PROGRESS REPORT State of_ _ _ _ _ _C~ol~o~ra~d~o_ _ _ __ Mammals Research Program Work Package No. _ _~30~0~1~----- Deer Conservation Task No._ _ _ _ _ _____,_4_ _ _ _ _ __ Effect of Nutrition and Habitat Enhancements on Mule Deer Recruitment and Survival Rates Research and Development Project No._____W~-=15~3~-=R~---Period Covered: July 1, 2001 -June 30, 2002 Authors: C. J. Bishop and G. C. White Personnel: D. L. Baker, T. D. I. Beck, G. Bock, S. K. Carroll, D. Coven, K. Crane, D. J. Freddy, L. Gepfert, R. B. Gill, R. Harthan, M. McLain, E. P. Myers, G. C. Miller, J. Olterman, J. A. Padia, T. M. Pojar, C. M. Solohub, B. E. Watkins, CDOW; L. H. Carpenter, WMI; J. Sazrna, B. Welch, BLM, Montrose, CO. ABSTRACT To further understand the factors that caused deer numbers to decline in western Colorado during the 1990s, we designed and initiated a field experiment to measure deer population parameters in response to nutrition and habitat enhancement treatments. During November 2000-March 2002, we captured and radio-collared 112 mule deer in a treatment unit and 109 mule deer in a paired control unit during winter · on the Uncompahgre Plateau in southwest Colorado. We enhanced the nutrition of deer in the treatment unit by providing a safe, pelleted supplemental feed on a daily basis from December through April each winter. Early winter fawn:doe ratios were measured using helicopter and ground classification surveys the year following treatment delivery to determine whether fawn production and survival increased as a result of enhanced nutrition of adult females. Based on multiple age classification surveys, we concluded that the winter nutrition enhancement treatment did not cause an increase in neonatal production and survival during 2001. However, fawn production and summer-fall survival were atypically good during 2001, and not representative of most years during the past decade when the population declined. We also measured overwinter fawn survival rates in response to the treatment. The simplest model which effectively explained survival (x\ 1 = 51.87, P = 0.440) included treatment (x21 = 9.95, P = 0.002) and early winter fawn mass (x21 = 8.33, P = 0.004). From December 1, 2001, through May 31, 2002, the survival rate of fawns was significantly greater (x 21 = 13.216, P < 0.001) in the treatment unit (0.865, SE = 0.056) than in the control unit (0.510, SE = 0.080); and fawns that survived the winter averaged 2.9 kg heavier than fawns that died (F1 = 6.11, P = 0.016). Early winter fawn mass was not different among treatment and control fawns (F1 = 0.36, P = 0.550), thus the effect of the treatment was not confounded with fawn mass. Simply, heavier fawns in both experimental units had higher survival probabilities. During•winter2001-02, which was a mild to average winter, the nutrition enhancement treatment clearly improved overwinter fawn surviv;il, and thus yearling recruitment. We will continue this portion of the resea.n;:h for 2 more years. The results reported here are preliminary and should be treated as such. ~.,,.,. 7:,,-. . • • . ' .. :,: '-~:: .' • • . • . ·-- • • r~11Hrnm~i111mm1li BDOWD16779 ��67 EFFECT OF NUTRITION AND HABITAT ENHANCEMENTS ON MULE DEER RECRUITMENT AND SURVNAL RATES C. J. Bishop and G C. White P. N. OBJECTIVES 1. To determine experimentally whether enhancing mule deer nutrition during winter and early spring by supplemental feeding increases December fawn:doe ratios and overwinter fawn survival. 2. To determine experimentally to what extent habitat treatments replicate the effect of enhanced nutrition from supplemental feeding. SEGMENT OBJECTIVES 1. Capture and radio-collar target sample of adult female mule deer and 6-month-old fawns. 2. Deliver nutrition enhancement treatment to all deer occupying the treatment area. 3. Measure overwinter adult female and fawn survival rates and early winter fawn:doe ratios in the treatment and control areas. INTRODUCTION Mule deer numbers apparently declined during the 1990's throughout much of the West, and have clearly decreased since the peak population levels documented in the 1940's-60's (Gill et al. 1999, Unsworth et al. 1999). Biologists and sportsmen alike have concerns as to what factors may be responsible for declining population trends. Although previous and current research indicates that multiple interacting factors are responsible, habitat and predation have received the focus of attention. A number of studies have evaluated whether predator control increases deer survival, yet results are highly variable (Connolly 1981, Ballard et al. 2001). Together, predator control studies with adequate rigor indicate that predation effects on mule deer are variable as a result of time-specific and site-specific factors. Studies which have demonstrated deer population responses to predator control treatments have failed to determine whether predation is ultimately more limiting than habitat. Numerous research studies have evaluated mule deer habitat quality, but virtually no studies have documented population responses to habitat improvements. In many areas where declining deer numbers are of concern, predation is common yet habitat quality appears to have declined. The question remains as to whether predation, habitat, or some other factor is more limiting to mule deer in these situations, and whether habitat quality can be improved for the benefit of deer. It may also be that no single factor is any more or less important than another, and that a more comprehensive understanding of multi-factor interactions is paramount. We designed a field experiment to measure deer population responses to nutrition and habitat enhancement treatments, to further understand the causative factors underlying observed deer population dynamics. We are conducting the study on the Uncompahgre Plateau, where several predator species (i.e. coyotes, mountain lions, and bears) are present in abundant numbers. In addition to predation, myriad diseases in combination proximately affect survival of the Uncompahgre deer population (Pojar 2000, B.E. Watkins, unpublished data). Predator numbers have not and will not be manipulated in any manner during the course of the study. All factors have been left constant with the exception of deer nutrition and habitat. Deer nutrition is being enhanced by providing supplemental feed to deer during the winter. If December fawn:doe ratios and overwinter fawn survival improve as a direct result of the nutrition enhancement treatment, then we can presume that deer nutrition is ultimately more limiting than . predation or disease. The second phase of the field experiment will incorporate habitat manipulation �68 treatments, which will consist of prescribed fire or mechanical techniques to set back succession of pinyon-juniper habitat in an effort to improve the vigor and quality of winter habitat for mule deer. Deer population responses will be measured in relation to the habitat manipulations in the same manner as the supplemental feed. Thus, the experiment allows us to determine whether nutritional quality of habitat is ultimately more limiting than other factors in a late-seral pinyon-juniper/sagebrush landscape, and if so, whether habitat can be effectively improved for mule deer. The results will also advance our current understanding of multi-factor interactions, with direct implications for mule deer management. MATERIALS AND METHODS Experimental Approach Experimental Design and Study Area We non-randomly selected four areas on the Uncompahgre Plateau to create 4 experimental units (A-D) (Fig. I). Treatments were randomly assigned to the experimental units. The following criteria were used to select experimental units: I.). Deer densities (~50-80 deer/mi2): areas were selected where deer densities were sufficient to meet sample size requirements within the experimental unit, while simultaneously selecting areas that would require feeding less than ~500 animals during a normal winter 2.) Buffer zones: areas were selected such that experimental units would be separated by several miles of non-treatment area (buffers) to prevent deer from occupying more than one experimental unit 3.) Similarity: areas were selected that comprise relatively similar habitat complexes and deer densities that are representative of the overall area 4.) Elk pqpulations: areas were selected to minimize the number of elk present during normal winters Units A and B are receiving the nutrition enhancement treatment in a cross-over experimental design, and are being used to address P .N. Objective I. Unit A served as the treatment unit, while Unit B served as the control, for the first 2 years of research (2000 - 2002). Beginning November 2002, Unit B will receive the treatment while Unit A will serve as the control. Upon completion of P.N. Objective I, Units C and D will be used to conduct phase 2 of the research, or P.N. Objective 2. Habitat in one unit will be manipulated to set back plant succession (treatment), while habitat in the other unit will remain unchanged (control) throughout the experiment. Year Unite UnitD 2004-05 Control Control 2005-06 .. Control Fire/Mechanical Treat. 2006-07 • ':- /Control 2000-01 2001-02 2002-03 2003-04 2007-08 Control.· · v~-g Response Yr 1 Veg Response Yr 2 2008-09 ,:.( .Control Veg Response Yr 3 2009-10 Control Veg Response Yr 4 Figure 1. Schematic representation of experimental units and associated treatments. The nutrition enhancement cross-over design will encompass 4 years; monitoring in the habitat manipulation experimental unit and paired control area will encompass approximately 6 years. • �69 The 2 experimental units (A and B) receiving the nutrition enhancement treatment are (Figs. 2 and 3): (1) The Colona Tract ofthe Billy Creek State Wildlife Area (2) Bureau of Land Management lands adjacent to Shavano Valley as defined by the following: Within Dry Creek Basin Quadrangle (USGS 7.5 Minute), includes Sections 6 and 7 in T. 48 N.-R. 10 W. and Sections 1, 2, 10, 11, 12, 13, 14, 15 in T. 48 N.-R. 11 W. This area roughly includes 38°25'00" -38°27'30" Latitude and 108°00'00" - 108°04'30" Longitude. , . r-' esa County Gunnison County Figure 2. Location of Colona and Shavano (Units A and B) experimental units in Grune Management Unit 62 on the Uncompahgre Plateau, southwest Colorado. �70 Figure 3. Colona and Shavano experimental units (Units A and B), located in Game Management Unit 62 on the Uncompahgre Plateau, southwest Colorado. Polygons represent the nucleus of each experimental unit, which is where animals have been collared and the nutrition enhancement treatment delivered. �71 Response Variables The primary response variable is fawn:doe ratios measured during December and January following the previous winter's treatments. The fawns counted during early winter age classification were born the summer following the winter treatments, and classified when they were 6 months of age. Thus, we are measuring the effect of enhanced adult doe nutrition during winter on subsequent fawn production and survival. Fawn:doe ratios are currently being measured in Units A and B corresponding to P.N. Objective I. The second response variable is overwinter fawn survival, measured from radio-collared fawns during the winter in direct response to the enhanced winter nutrition treatment. We are also measuring overwinter and annual survival of adult does as a function of enhanced winter nutrition. Sample Size The primary response variable is the mean fawn:doe ratios of the radio-collared does wintering on the experimental unit of interest. We desired to detect an effect size, i.e., an increase in fawn:doe ratios in response to the treatments, in the range of 15 to 20 fawns per 100 does. These values were based on simple population models with overwinter fawn survival of 0.444, adult female survival of 0.853, and December fawn:doe ratios of 66 fawns per 100 does to obtain a stationary population (Unsworth et al. 1999). Based on surveys of DAU D-19 from 1992-96, the standard deviation of the fawn:doe ratio for groups with at least one adult female was 57, with a mean of 41. Using an expected standard deviation of 57, the standard error of the mean fawn:doe ratio for 40 radio-collared does is 57/(40 112) = 9.0, which is the expected standard deviation of measured fawn:doe ratios on each of the experimental units. We assessed power of the proposed experiment using SAS Analyst®. We used a two-sample t-test with a sample size of 4, representing the years of the study where treatment effects will be measured. The power of the design to detect an increase of 2 0 fawns per I 00 does is about O. 87. A sample size of 40 fawns per experimental unit per year provides a power of 0.81 to detect a difference of0.15 in survival between 2 experimental units if survival on the control unit is 0.40. We expected to see an increase in fawn survival (effect size) of approximately O.15, because this was the difference measured in the density reduction experiment conducted by White and Bartmann (1998). Capture Methods Deer were captured using baited drop nets (Ramsey 1968, Schmidt et al. 1978) and helicopter net guns (Barrett et al. 1982, van Reenen 1982). Drop nets were baited with certified weed-free alfalfa hay and apple pulp. Drop nets were used as the principal capture technique during a 3-week capture period; helicopter net-gunning was used at the end of the capture period to secure the remainder of deer needed to meet our target sample sizes. All deer were hobbled and blind-folded after being captured. All deer captured via drop nets were carried away from the net to an adjacent handling site using stretchers. Deer were fitted with leather radio collars equipped with mortality sensors, which cause an increase in pulse rate after remaining motionless for 4 hours. Permanent collars were placed on adult females, while temporary collars were placed on fawns. To make collars temporary, one end of the collar was cut in half and reattached using rubber surgical tubing; fawns shed the collars ~6 months post-capture. A rectangular piece of flexible plastic (Ritchey® neck band material) engraved with a unique identifier was stitched to the side of each collar. The unique identifier consisted of 2 symbols for adult females, and only 1 symbol on 2 different colors of plastic for fawns. The identifiers were necessary to visually identify deer from the ground. This has allowed us to effectively document use of the treatment, measure fawn:doe ratios from the ground, and assess experimental unit population size via mark-resight estimators. We recorded the weight, hind foot length and chest girth of each deer, and collected blood samples from most does and fawns to evaluate disease prevalence. �72 Measurement of Fawn:Doe Ratios and Overwinter Survival Each winter we used the radio-collared does to measure fawn:doe ratios in each experimental unit. The resulting fawn:doe ·ratio is a measurement of the previous year's treatment effect. We measured fawn:doe ratios using 2 techniques: (1) We located the sample ofradio-collared does in each experimental unit from a fixed-wing airplane, and used the set of locations to define boundaries for the experimental unit. Shortly after (i.e. 1-2 days), we used a helicopter to systematically fly the defined unit and classify all deer groups encountered. For each group, we documented whether a radio-collared doe was present. (2) We located each radio-collared doe by radio telemetry from the ground. The group of deer with the collared doe was counted and classified by age and sex. Both methods have been employed to gather as much information as possible to determine whether there was a treatment effect. The "true" value cannot be measured perfectly because of the inherent biases and potential sources of error associated with each technique. Thus, by employing both techniques, we have a greater chance of fully understanding whether the treatment caused an effect. We measured survival by radio-monitoring collared deer to determine fate (live/mortality). We also attempted to determine the cause of each mortality, with a primary goal of distinguishing between predation and non-predation mortality causes. Deer were radio-monitored on a daily basis during the winter, which typically allowed us to arrive at mortality sites within 24 hours. Treatment Delivery Deer nutrition was enhanced in the treatment area by providing a safe, pelleted supplemental feed. The supplemental feed was developed through extensive testing with both captive and wild deer (Baker and Hobbs 1985, Baker et al. 1998), and has been safely used in both applieq research and management projects. Pellets were distributed daily using 4wd pickup trucks and ATVs on primitive roads throughout the experimental unit to provide a food source for the entire deer population in the treatment unit. Each 501b. bag of pellets was carried :S:;200m from the truck/A TV and distributed by hand in approximately 2030 small piles of feed in a linear fashion. Numerous bags were distributed in successive order allowing us to create a line of feed that spanned most of the treatment area, which prevented animals from concentrating in any single location. This feeding technique also prevented dominant animals from restricting access to the food supply because of the large area over which pellets were distributed. We attempted to supply pellets ad Iibitum such that a small residual remained when the next day's ration was provided. Collared deer were closely monitored to ensure that treatment deer remained in the experimental unit and actually consumed the feed, and to make sure that non-treatment deer remained in the control unit, which they did. Treatment deer that did not regularly consume the feed were withdrawn from the sample for purposes of measuring treatment effects. The pelleted ration was commercially produced in the form of 2x 1x0.5-cm wafers (Baker and Hobbs 1985). Feed constituents (i.e. digestibility, protein, gross energy etc.) vastly exceeded those of typical winter range deer diets; exact constituent values are provided by Baker et al. (1998). When provided ad libitum, the feed should have allowed deer to meet or exceed nutritional requirements for growth and maintenance (Ullrey et al. 1967, Verme and Ullrey 1972, Thompson et al. 1973, Smith et al. 1975, Baker et al. 1979, Holter et al. 1979). The basis for feeding such high quality pellets was to ensure that the treatment (enhanced nutrition) was effectively delivered to the deer. Our intent was not to determine the exact level of nutrition necessary to increase fawn recruitment, but rather to determine if nutrition is a limiting factor to recruitment. If nutrition is in fact limiting, we will rely on the habitat manipulation treatment to evaluate what exactly can be done via management to increase fawn survival and recruitment. �73 Habitat Manipulations In order to accomplish P.N. Objective 2, habitat will be manipulated in experimental unit D through collaboration with the Uncompahgre Ecosystem Restoration Project (UP), which comprises personnel from the Division of Wildlife, U.S. Forest Service, Bureau of Land Management, Public Lands ' Partnership, and a variety of other public and private stakeholders. The UP committee is using an experimental landscape approach to manipulate various habitats in a mosaic pattern throughout the Uncompahgre Plateau. We will focus our intensive deer monitoring on one of these habitat manipulations that will be conducted in experimental unit D. This portion of the research has not yet been initiated. A complete description of our planned protocols to accomplish P.N. Objective 2 is provided in the Program Narrative (Bishop and White 2000). Statistical Methods Once data collection is completed for the full study, we will test for differences in fawn:doe ratios between experimental units and years using the following statistical model: YiJk = µ + a.1 + /3k + a/3.ik + e;Uk), where YiJk = fawn:doe ratio for the ith deer group in treatment combinationjk; i = 1, 2, ... , n1k(deer groups);}= 1, 2, 3, 4 experimental units (control, supplemental feed, habitat manipulation); k= 1, 2, 3, 4 (6) years; a~k = interactions among experimental units and years; and e;Uk) = random error associated with YiJk• A similar model will be used to analyze overwinter fawn survival, but a logit-link function will be used in place of the identity link function in the above general linear model. A similar model will also be used to test for differences in fawn weights, except the response variable will be fawn mass, and sex and fate (i.e. lived or died) will be included in the model as independent variables. For this progress report, a preliminary fawn:doe ratio analysis was completed using PROC MIXED in SAS (SAS Institute 1997). We used a reduced model with experimental unit as the lone independent variable, and considered experimental unit as a fixed effect and radio-collared does within an experimental unit as random effects. Survival rates were calculated using a Kaplan-Meier survival analysis (Kaplan and Meier 1958, Pollock et al. 1989), and contrasted among experimental units and sexes using a chi-square analysis. We modeled winter fawn survival with a product multinomial model (Grizzle et al. 1969) using PROC CATMOD in SAS (SAS Institute 1989a). Survival was modeled as a function of experimental unit, sex, and capture mass. We used a general linear model in PROC GLM in SAS (SAS Institute 1989b) to test for differences in fawn mass between experimental units, sexes, and fates (i.e. lived or died). Other results in this report are presented as data summaries incorporating means and standard errors, or in some cases, raw data values. These results are incomplete and preliminary in nature, and should be treated as such. RESULTS AND DISCUSSION Deer Capture During November-December 2000, we captured and radio-collared 73 adult female mule deer: 37 in the treatment unit and 36 in the control unit. Due to budgeting constraints, we were unable to capture and radio-collar fawns. During November-December 2001, we captured and radio-collared an additional 32 adult females to replace mortalities from the previous year and to buffer our sample size, resulting in a total of 45 radio-collared does in each experimental unit. We also captured and radio-collared 80 fawns: 40 in each experimental unit. During February 28 -March 1, 2002, we captured an additional 36 does (18 in each experimental unit) as part of a related research project (Bishop et al. 2002). In total, we radiomonitored 221 mule deer (141 adult does and 80 fawns) during November 2000-June 2002. �74 Treatment Delivery 2000-01 From December 15, 2000, through April 19, 2001, we distributed 88 tons of the pelleted ration. For most of the winter and spring, on average, we distributed 0.85 tons offeed each day throughout 22 feeding sites across the 2.3 mi 2 treatment unit. Deer were fed ad libitum because there was always residual feed remaining the next day during the feeding routine. Each sack was distributed in approximately 20-30 distinct, small piles, resulting in > 1000 small piles of feed throughout the treatment unit. This effort allowed deer to effectively access the feed in small groups, and no aggression was ever observed among deer seeking access to the feed. By distributing the feed in this manner, we were able to avoid the negative aspects associated with large-scale feeding operations. Deer adapted to the pelleted supplement right away and utilized it extensively throughout the winter. We continually monitored deer use of the feed from ground observation points, where we obtained 440 visual observations of radio-collared does consuming the feed. These observations, coupled with daily radio-monitoring and periodic aerial relocations, indicate 32 of the 37 radio-collared treatment does spent the entire winter and spring within the boundaries of the treatment unit and received the supplement on a daily basis. Mark-resight population estimates from March helicopter (489 deer, SE = 62) and ground (494 deer, SE = 81) surveys, coupled with feed consumption, indicate we fed roughly 450 to 500 deer during most of the winter and spring. Feed consumption declined coincident with spring green-up, although deer continued to use the feed through mid-late April, at which point they began migrating to summer range. We also fed approximately 25 to 30 elk, but the elk did not affect deer access to the feed. Deer in the control experimental unit did not receive feed or any other treatment. Based on helicopter mark-resight surveys, the deer density in the treatment unit in December was 120 deer/mi2 (SE= 9), but increased shortly after and was 213 deer/mi2 (SE= 27) in March. Deer densities in the control unit changed little from 83 deer/mi2 (SE= 12) in December to 101 deer/mi 2 (SE= 14) in March. 2001-02 From December 15, 2001, through April 25, 2002, we distributed 194 tons of the supplement throughout the treatment unit. For most of the winter and spring, we distributed 2.0-2.1 tons of feed each day. The dramatic increase in supplement distribution from the previous year occurred because a large number of elk descended into the Uncompahgre Valley during mid-late fall/early winter. Elk arrived in unusually large numbers throughout much of the valley prior to the onset of treatment delivery. Once feeding was initiated, approximately 300-500 elk adapted to the feed and remained in or around the 2.3 mi2 treatment unit throughout most of the winter. Given myriad logistical and budgetary constraints, 2.1 tons was the maximum amount of feed we could routinely deliver on a daily basis. Feed was not delivered ad libitum to all deer and elk in the treatment unit throughout the winter because residual feed was rarely observed during the next day's distribution. However, daily field observations indicate most deer approached ad libitum consumption of the supplement. In contrast to the previous winter, deer were waiting for the daily supplement to arrive each morning. Deer then consumed the supplement immediately after it was distributed. Elk were rarely observed utilizing th·e feed until late morning or afternoon, and elk continued to forage in fields below the treatment unit, whereas deer did not. We observed numerous radio-collared deer consuming the pelleted supplement each day; not all of these observations were recorded because of time constraints with distributing the feed. Given this time limitation, we still recorded 818 observations of radio-collared deer consuming the supplemental feed (497 collared doe observations and 321 collared fawn observations). Most days, > 100 and sometimes 200-300 deer were observed utilizing the pellets during the course of distributing the supplement. These observations rarely included elk; thus, deer-elk competition was minimized because of temporal differences in feeding, and deer clearly had first access to the feed. �75 Fawn:Doe Ratios In December 2000, at the beginning of the study and prior to the first year's treatment delivery, fawn:doe ratios were similar in the 2 experimental units. Pre-treatment fawn:doe ratios were 52.6 fawns: 100 does (SE= 5.3) in the treatment unit, and 51.6 fawns: 100 does (SE= 5 .0) in the control unit. In late December 2001 and early January 2002, following the first year's treatment, we conducted 2 age classification helicopter surveys in the treatment and control units. On 12/23/01, we observed 52.8 fawns:100 does (SE = 6.7) in the treatment unit, and 36.7 fawns:100 does (SE= 3.8) in the control unit. On 1/8/02, we observed 54. 7 fawns: 100 does (SE= 6.6) in the treatment unit, and 50.5 fawns: 100 does (SE= 6.0) in the control unit. During December 2001 -February 2002, we obtained fawn:doe ratio estimates from ground observations of radio-collared deer groups for both treatment and control deer. This survey resulted in 61.2 fawns:100 does (SE= 7.8) in the treatment unit, and 74.5 fawns:100 does (SE= 8.5) in the control unit, although the result was not statistically significant (t74 = 1.16, P = 0.249). The fawn:doe ratio results are conflicting, and clearly do not provide evidence that there was any treatment effect. In short, we conclude that the nutrition enhancement treatment did not cause an increase in neonatal production and survival during 2001. However, our results, in conjunction with a December estimate of 64 fawns: 100 does for the entire Uncompahgre deer population (B.E. Watkins, unpublished), indicate fawn production and survival was good during 2001. The observed fawn:doe ratios coupled with overwinter fawn survival and annual adult survival rates indicate the deer population is growing. Considering the past 1-2 decades, this was an atypically good year for the Uncompahgre deer population. It would appear that whatever set of environmental conditions have led to a declining deer population were not present during 2001 in the same manner as in the past. Our main interest lies in observing the effect of the treatment on the deer population in a year where fawn:doe ratios are lower for the population as a whole, similar to what they have been much of the past 15 years. Our results point out the inherent difficulties and biases associated with precisely measuring fawn:doe ratios, particularly in this research study. Ratios obtained from helicopter surveys were based on 2 shortduration flights over small spatial units. Helicopter surveys were complicated by high deer densities in heavy cover, making both deer detection and fawn:doe classifications a considerable challenge. There is a variety of potential biases that may have affected the helicopter surveys, including differential sightability of does and fawns, and incorrectly classifying yearling bucks as adult does. These biases are likely real considering the higher ratios measured during the ground classifications based on the radiocollared does. Ground fawn:doe ratio observations of radio-collared doe groups were made using spotting scopes and field glasses, where we commonly studied the deer for some time. Incorrect classifications during these surveys were likely minimal. For example, small-antlered yearling bucks (e.g. 3 - 6" spikes) were detected from the ground, whereas they were clearly missed on occasion during helicopter surveys. The ground classifications were also preferable to the helicopter surveys in that we obtained repeated observations. We recorded as many as 5 separate ratio observations per radio-collared doe. Overall, we believe the ground fawn:doe ratio estimates, based on individual radio-collared does, provided less biased measurements. Given the inherent difficulties of measuring fawn:doe ratios, and the lack of a clear indication as to the effectiveness of the treatment, we initiated a second study using vaginal implant transmitters in order to capture and radio collar newborn fawns from the radio-collared treatment and control does (Bishop et al. 2002). This new aspect of the research will allow us to gain better estimates of the treatment effect on subsequent fawn production and survival, evaluate cause-specific mortality of treatment/control neonates, and simultaneously provide a greater understanding as to the mechanisms affecting the deer population. Survival Adult Fema/,es During winter 2000-01 (Dec 1, 2000 - May 31, 2001 ), the adult doe survival rate of deer in the treatment unit (0.968, SE= 0.032) was greater (x21 = 2.649, P = 0.104) than the survival rate of deer in the control �76 unit (0.861, SE = 0.058). However, annual adult doe survival rates (Dec 1, 2000 - Nov 30, 200 I) were similar among the treatment and control deer (Trt: S(t) = 0.839, SE= 0.066; Control: S(t) = 0.833, SE= 0.062; x21 = 0.004, P = 0.94 7). Thus, mortalities of control deer occurred primarily during the winter months, while treatment does died primarily during the summer and fall months. During winter 2001-02 (Dec I, 2001 - May 31, 2002), the adult doe survival rate of deer in the treatment unit (0.942, SE= 0.030) was once again greater (x2 1 = 3.116,P = 0.078) than the survival rate of deer in the control unit (0.848, SE= 0.044). At this preliminary stage in the research, the nutrition enhancement treatment has apparently increased survival of adult females during the winter, but the overall annual survival among treatment and control does has not varied. The annual survival rate of does measured thus far aligns with expected survival based on other studies (Unsworth et al. 1999, B.E. Watkins, unpublished). Fawns During winter 2001-02 (Dec 1, 2001 - May 31, 2002), the survival rate of fawns was significantly greater (X 21 = 13 .216, P < 0.001) in the treatment unit (0.865, SE= 0.056) than in the control unit (0.510, SE= 0.080) (Fig. 4). The simplest model which effectively explained survival (x\, = 51.87, P = 0.440) included treatment (x2 1 = 9.95, P = 0.002) and early winter mass (x\ = 8.33, P = 0.004). Fawns receiving the nutrition enhancement treatment, and heavier fawns, had higher survival probabilities. During winter 2001-02, which was a mild to average winter, the nutrition enhancement treatment clearly improved overwinter fawn survival, and thus yearling recruitment. 1 -. I I - -- • I - - I I - 0.9 -■ - ---- -- - ---- - - ------ -- - 1~ 0.8 l L_ I I 0.6 7_ -- • Colona (Treatment) 0.5 --Shavano (Control) 0.4 12/1 12/16 12/.H. 1/15 1/30 .2/14 3/1 3/1.6 3/31. 4./15 I I 4/30 5/15 5/30 Figure 4. Overwinter fawn survival (Dec 1, 2001 - May 31, 2002) in a nutrition enhancement treatment unit (S(t) = 0.865, SE= 0.056) and a control unit (S(t) = 0.510, SE= 0.080), Uncompahgre Plateau, southwest Colorado. Causes of Mortality Adult Females During winter 2000-01, only one adult doe from the treatment unit died, which was road-killed in early May. During June-August, 2001, an additional 4 treatment deer died: 2 from unknown causes that were �77 not predator-related, 1 from a prolapsed uterus, and 1 unknown. In contrast, 5 adult does from the control area died during winter 2000-01 : 3 from malnutrition, 1 from mountain lion predation, and 1 was roadkill ed. One additional deer from the control area died during August 2001 from an unknown cause that was not predator-related. During winter 2001-02, 3 adult does from the treatment unit died: 1 from secondary causes related to chronic arthritis, 1 from predation, and 1 road-killed. The predation and road kill mortalities occurred in mid-late May after deer had left the treatment unit. In June, 2 more treatment adult does died during the fawning period. One doe died while giving birth, and the second doe died from an unknown cause seemingly related to birthing. This second doe, based on a field necropsy, had considerable fat and seemed otherwise healthy. On March 1, 2002, we measured 2 fetuses in utero (Bishop et al. 2002), yet the doe had only 1 fetus in utero when she died. Given her good condition, it is unlikely the second fetus was reabsorbed, which indicates she had already passed the first fetus and died prior to giving birth to the second. During winter 2001-02, 7 adult does from the control unit died: 4 died from mountain lion predation, 1 from entanglement in a fence, 1 from an unknown, non-predator mortality, and 1 unknown. No additional control does died during the month of June. Fawns Five fawns in the treatment unit died during winter 2001-02: 2 from malnutrition/sickness and 3 from disease. Of the 2 fawn mortalities caused by malnutrition/sickness, 1 was a result of basic malnutrition and occurred on December 31, 2001. The other fawn had a combination of heavy parasite loads, scours, and general poor condition. Each of the 3 fawns that died from disease had adequate fat stores. At least one of these fawns died as a result of pneumonia. In the control unit, 19 fawns died during the winter: 5 from malnutrition, 6 from mountain lion/bobcat predation, 4 from coyote/canine predation, 3 unknown predation mortalities, and 1 unknown. A majority of the fawns killed by predators had virtually no femur marrow fat remaining, indicating the predation was likely compensatory in nature. Fawn Mass During winter 2001-02, the early winter mass of radio-collared fawns varied significantly between sexes (F1 = 15.32, P < 0.001) and fates (F1 = 6.ll, P = 0.016). Males averaged 3.6 kg heavier than females, and fawns that survived the winter averaged 2.9 kg heavier than fawns that died. Early winter mass was not different among experimental units (F1 = 0.36, P = 0.550), thus the effect of the treatment was not confounded with fawn mass. The interaction of experimental unit x sex x fate was also significant (F1 = 5.80, P = 0.019), while all other 2-way interactions were not significant. The 3-way interaction occurred because in the control experimental unit, female fawns that survived were not heavier than female fawns that died (Survived: x = 31.0 kg, SE= 1.77; Died: x = 31.5 kg, SE= 1.03); whereas male fawns that survived were considerably heavier than male fawns that died (Survived: x = 38.0 kg, SE= 0.83; Died: x = 32.7 kg, SE= 1.35). In contrast, in the treatment experimental unit, weight differences were more pronounced between surviving and non-surviving females (Survived: x = 33.1 kg, SE= 1.00; Died: x = 28.2 kg, SE= 2.75) than between surviving and non-surviving males (Survived: x = 35.0 kg, SE= 0.87; Died: x = 34.5 kg, SE = 1.21 ). The importance of early winter fawn mass as a predictor of overwinter survival has been documented previously (White et al. 1987, Bishop 1998, White and Bartmann 1998, Unsworth et al. 1999). �78 LITERATURE CITED Baker, D. L., and N. T. Hobbs. 1985. Emergency feeding of mule deer during winter: tests of a supplemental ration. Journal of Wildlife Management 49:934-942. Baker, D. L., D. E. Johnson, L. H. Carpenter, 0. C. Wallmo, and R. B. Gill. 1979. Energy requirements of mule deer fawns in winter. Journal of Wildlife Management 43:162-169. Baker, D. L., G. W. Stout, and M. W. Miller. 1998. A diet supplement for captive wild ruminants. Journal of Zoo and Wildlife Medicine 29: 150-156. Ballard, W. B, D. Lutz, T. W. Keegan, L. H. Carpenter, and J.C. deVos, Jr. 2001. Deer-predator relationships: a review ofrecent North American studies with emphasis on mule and black-tailed deer. Wildlife Society Bulletin 29:99-115. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bullet.in I 0: I 08-114. Bishop, C. J. 1998. Mule deer fawn mortality and habitat use, and the nutritional quality of bitterbrush and cheatgrass in southwest Idaho. Thesis, University ofldaho, Moscow, Idaho, USA. Bishop, C. J., D. J. Freddy, and G. C. White. 2002. Effects of enhanced nutrition of adult female mule deer on fetal and neonatal survival rates. Colorado Division of Wildlife, Wildlife Research Report, Federal Aid in Wildlife Restoration Project W-153-R, Progress Report. Fort Collins, CO USA. Bishop, C. J., and G. C. White. 2000. Effects of habitat enrichment on mule deer recruitment and survival rates. Colorado Division of Wildlife, Wildlife Research Report, Federal Aid in Wildlife Restoration Project W-153-R-13, Progress Report. Fort Collins, CO, USA. Connolly, G. E. 1981. Limiting factors and population regulation. Pages 245-285 in 0. C. Wallmo, editor. Mule and black-tailed deer of North America. University of Nebraska Press, Lincoln, Nebraska, USA. Gill, R. B., T. D. I. Beck, C. J. Bishop, D. J. Freddy, N. T. Hobbs, R.H. Kahn, M. W. Miller, T. M. Pojar, and G. C. White. 1999. Declining mule deer populations in Colorado: reasons and responses. A report to the Colorado Legislature. Colorado Division of Wildlife, Denver, Colorado, USA. Grizzle, J. E., C. F. Starmer, and G. G. Koch. 1969. Analysis of categorical data by linear models. Biometrics 25:489-504. Holter, J.B., H. H. Hayes, and S. H. Smith. 1979. Protein requirement of yearling white-tailed deer. Journal of Wildlife Management 43:872-879. Kaplan, E. L., and P. Meier. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 53:457-481. Pojar, T. M. 2000. Investigating factors contributing to declining mule deer numbers. Colorado Division of Wildlife, Wildlife Research Report, Federal Aid in Wildlife Restoration Project W-153-R-13, Progress Report. Fort Collins, CO, USA. Pollock, K. H., S. R. Winterstein, C._M. Bunck, and P. D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Ramsey, C. W. 1968. A drop-net deer trap. Journal of Wildlife Management 32: 187-190. SAS Institute. 1989a. SAS/STAT® user's guide, version 6, fourth edition. Volume I. SAS Institute, Cary, North Carolina, USA. SAS Institute. 1989b. SAS/STAT® user's guide, version 6, fourth edition. Volume 2. SAS Institute, Cary, North Carolina, USA. SAS Institute. 1997. SAS/STAT® Software: Changes and Enhancements through Release 6.12. SAS Institute, Cary, North Carolina, USA. Schmidt, R. L., W. H. Rutherford, and F. M. Bodenham. 1978. Colorado bighorn sheep- trapping techniques. Wildlife Society Bulletin 6: 159-163. Smith, S. H., J.B. Holter, H. H. Hayes, and H. Silver. 1975. Protein requirement of white-tailed deer fawns. Journal of Wildlife Management 39:582-589. Thompson, C. B., J.B. Holter, H. H. Hayes, H. Silver, and W. E. Urban, Jr. 1973. Nutrition ofwhitetailed deer. I. Energy requirements of fawns. Journal of Wildlife Management 37:301-311. �79 Ullrey, D. E., W. G. Youatt, H. E. Johnson, L. D. Fay, and B. L. Bradley. 1967. Protein requirement of white-tailed deer fawns. Journal of Wildlife Management 31 :679-685. Unsworth, J. W., D. F. Pac, G. C. White, and R. M. Bartrnann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-421 in L. Nielsen, J.C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, Wisconsin, USA. Verme, L. J., and D. E. Ullrey. 1972. Feeding and nutrition of deer. Pages 275-291 in D. C. Church, editor. Digestive physiology and nutrition of ruminants. Volume 3 - Practical Nutrition. D. C. Church, Corvallis, Oregon, USA. White, G. C., and R. M. Bartrnann. 1998. Effect of density reduction on overwinter survival of freeranging mule deer fawns. Journal of Wildlife Management 62:214-225. White, G. C., R. A. Garrott, R. M. Bartmann, L. H. Carpenter, and A. W. Alldredge. 1987. Survival of mule deer in northwest Colorado. Journal of Wildlife Management 51 :852-859. �33 JOB PROGRESS REPORT State of Division of Wildlife - Mammals Research Colorado Work Package No. _ _--=-30-=--0=--'1,....__ _ _ __ Deer Conservation Task No. Effect of Nutrition and Habitat Enhancements on Mule Deer Recruitment and Survival Rates 4 Federal Aid Project -----'-W-'---=l-=-8=--5--=R-=------Period Covered: July 1, 2002 - June 30, 2003 Authors: C. J. Bishop, G C. White, D. J. Freddy, and B. E. Watkins Personnel: D. L. Baker, L. Baeten, S. K. Carroll, D. Coven, K. Crane, M. DelTonto, B. Diamond, B. de Vergie, D. Gallegos, J. Garner, L. Gepfert, R. B. Gill, D. Hale, J. Grigg, R. Harthan, W. J. Lassiter, T. Mathieson, J. McMillan, G. C. Miller, M. W. Miller, J. Nicholson, J. A. Padia, T. M. Pojar, J. E. Risher, C. A. Schroeder, W. G. Sinner, C. M. Solohub, J. Thayer, M. A. Thonhoff, L. Wolfe, CDOW; H. VanCampen, CSU; D. Felix, Olathe Spray Service; T. R. Stephenson, California Fish and Game; L. H. Carpenter, WMI; J. SaZina, B. Welch, BLM. ABSTRACT To further understand the factors that caused deer numbers to decline in western Colorado during the 1990s, we designed and initiated a field experiment to measure deer population parameters in response to a nutrition enhancement treatment. During November 2000 - June 2003, we captured and radio-collared 533 individual mule deer evenly distributed among treatment and control units on the Uncompahgre Plateau in southwest Colorado. This included 216 adult females, 94 of which received vaginal implant transmitters (VITs), 160 6-month old fawns, and 157 newborn fawns born from either treatment or control adult does. We enhanced the nutrition of deer in the treatment unit by providing a safe, pelleted supplemental feed on a daily basis from December through April each winter. Early winter fawn:doe ratios were measured using helicopter and ground classification surveys the year following treatment delivery to determine whether fawn production and survival increased as a result of enhanced nutrition of adult females. We also measured overwinter fawn survival rates in response to the treatment. In 2002 and 2003, we measured pregnancy rates, fetus rates, and body condition of treatment and control adult does during late winter using ultrasonography. We also directly measured fetus survival and neonate survival by using VITs to help locate and radio-collar newborn fawns born from treatment and control does. Estimated percent body fat of adult does during late February and early March of 2002 and 2003 was significantly higher (F1. 90 = 108.21, P < 0.001) for treatment deer (10.4%, SE= 0.48, n = 48) than control deer (4.0%, SE= 0.36, n = 46). Serum thyroid hormone concentrations (measured only in 2003) were higher in treatment does than control does as well (F4• 52 = 32 .59, P < 0.001). Pregnancy and fetus rates were similar among treatment and control does. The pregnancy rate of adult does was 0.95 (SE= 0.036, n = 38) and the fetus rate was 1.80 fetuses/doe (SE= 0.10, n = 36) during 2002. Rates were similar in 2003, where we measured a pregnancy rate of 0.92 (SE= 0.034, n =63) and a fetus rate of 1.74 fetuses/doe (SE = 0.069, n = 50) which included 5 yearlings (the fetus rate excluding yearlings was 1.82 fetuses/doe, SE= 0.066, n = 45). The fetus survival rate with treatment and control fetuses combined was 0.86 (SE= 0.073) during 2002 and 0.97 (SE= 0.024) during 2003. Based on multiple early winter age classification surveys, we concluded that the winter nutrition enhancement treatment did not cause an increase in neonatal production and survival during 2001. Neonate survival data coupled with early �34 winter age classification surveys indicated a marginal treatment effect during 2002. However, fawn production and summer-fall survival was relatively good during 2001 and 2002 for the overall population, and not representative of most years during the past decade when the population declined. During 2003, as of late September, survival of newborn treatment fawns was 0.745 (SE= 0.059) and control fawn survival was 0.614 (SE= 0.073). During 2001-02, the overwinter survival rate of fawns was significantly greater (x 21 = 13.216, P < 0.001) in the treatment unit (S(t) = 0.865, SE= 0.056) than in the control unit (S(t) = 0.510, SE= 0.080). Again in 2002-03, the overwinter survival rate of fawns was significantly greater (x 21 = 5.734, P = 0.017) in the treatment unit (S(t) = 0.900, SE= 0.047) than in the control unit (S(t) = 0.691, SE= 0.074). Because of a cross-over over experimental design, the treatment unit during winter 2001-02 became the control unit during winter 2002-03, and vice versa. Thus, the overwinter survival treatment effect was replicated across each experimental unit. Combining both years of data, the best model of overwinter fawn survival (AICc = 148.63) included the treatment effect (x 21 = 14. 71, P < 0.001), early winter fawn mass (x\ = 16.80, P < 0.001), year (x 21 = 3.53, P = 0.060), and sex (x\ = 1.99, P = 0.158). The AIC model selection analysis emphasized the importance of both the treatment effect as well as early winter mass of fawns, because any models without treatment or fawn mass were very poor. Early winter mass was not different among experimental units (F1 = 0.35, P = 0.558), thus the effect of the treatment was not confounded with fawn mass. We will continue this research for 1.5 more years. The results reported here are preliminary and should be treated as such. �35 EFFECT OF NUTRITION AND HABITAT ENHANCEMENTS ON MULE DEER RECRUITMENT AND SURVIVAL RATES C. J. Bishop, G. C. White, D. J. Freddy, and B. E. Watkins P. N. OBJECTIVES 1. To determine experimentally whether enhancing mule deer nutrition during winter and early spring by supplemental feeding increases fetus survival, neonate survival, early winter fawn:doe ratios or overwinter fawn survival. 2. To determine experimentally to what extent habitat treatments replicate the effect of enhanced nutrition from supplemental feeding. SEGMENT OBJECTIVES 1. Capture and radio-collar a target sample of adult female mule deer and 6 month-old fawns during late November through mid-December in a treatment unit and a control unit. 2. Capture a target sample of adult female mule deer in the treatment unit and the control unit to measure pregnancy rates, fetal rates, and body condition during late February to early March, and fit each adult female deer with a radio collar and vaginal implant transmitter. 3. Deliver the nutrition enhancement treatment to all deer occupying the treatment unit from early December through the end of April. 4. Capture and radio-collar a target sample of newborn fawns from treatment and control radio-collared does during June using the vaginal implant transmitters as a technique to determine the timing and location of birth. 5. Measure fetus survival, neonate survival, early winter fawn:doe ratios, overwinter fawn survival, and annual adult female survival based on radio-collared deer from the treatment and control units. INTRODUCTION Mule deer (Odocoileus hemionus) numbers apparently declined during the 1990's throughout much of the West, and have clearly decreased since the peak population levels documented in the 1940's-60's (Gill et al. 1999, Unsworth et al. 1999). Biologists and sportsmen alike have concerns as to what factors may be responsible for declining population trends. Although previous and current research indicates that multiple interacting factors are responsible, habitat and predation have received the focus of attention. A number of studies have evaluated whether predator control increases deer survival, yet results are highly variable (Connolly 1981, Ballard et al. 2001 ). Together, predator control studies with adequate rigor indicate that predation effects on mule deer are variable as a result of time-specific and site-specific factors. Studies which have demonstrated deer population responses to predator control treatments have failed to determine whether predation is ultimately more limiting than habitat. Numerous research studies have evaluated mule deer habitat quality, but virtually no studies have documented population responses to habitat improvements. In many areas where declining deer numbers are of concern, predation is common yet habitat quality appears to have declined. The question remains as to whether predation, habitat, or some other factor is more limiting to mule deer in these situations, and whether habitat quality �36 can be improved for the benefit of deer. It may also be that no single factor is any more or less important than another, and that a more comprehensive understanding of multi-factor interactions is paramount. We designed a field experiment to measure deer population responses to nutrition enhancement treatments, to further understand the causative factors underlying observed deer population dynamics. We are conducting the study on the Uncompahgre Plateau in southwest Colorado, where several predator species are present in abundant numbers: coyotes (Canis latrans), mountain lions (Fe/is concolor), and bears (Ursus americanus). In addition to predation, myriad diseases in combination proximately affect survival of the Uncompahgre deer population (Pojar 2000, B.E. Watkins, unpublished data). Predator numbers have not and will not be manipulated in any manner during the course of the study. All factors have been left constant with the exception of deer nutrition. Deer nutrition is being enhanced by providing supplemental feed to deer occupying a treatment area during the winter. If December fawn recruitment and/or overwinter fawn survival improve as a direct result of the nutrition enhancement treatment, then we can presume that deer nutrition is ultimately more limiting than predation or disease. The second phase of the field experiment, which has not yet been initiated, will incorporate habitat manipulation treatments. The treatments will consist of prescribed fire or mechanical techniques to set back succession of pinyon-juniper (Pinus edulis-Juniperus osteosperma) habitat in an effort to improve the vigor and quality of winter habitat for mule deer. Deer population responses will be measured in relation to the habitat manipulations in the same manner as the supplemental feed. Thus, the experiment allows us to determine whether nutritional quality of winter range habitat is ultimately more limiting than other factors in a late-seral pinyon-juniper/sagebrush (Artemisia spp.) landscape, and if so, whether habitat can be effectively improved for mule deer. The results will also advance our current understanding of multi-factor interactions, with direct implications for mule deer management. MATERIALS AND METHODS Experimental Approach Experimental Design and Study Area We non-randomly selected two areas within mule deer winter range on the Uncompahgre Plateau to create 2 experimental units (A-B) (Fig. 1). The following criteria were used to select experimental units: 1.) Deer densities (~50-80 deer/mi2): areas were selected where deer densities were sufficient to meet sample size requirements within the experimental unit, while simultaneously selecting areas that would require feeding less than ~500-600 animals during a normal winter 2.) Buffer zones: areas were selected such that experimental units would be separated by several miles of non-treatment area (buffers) to prevent deer from occupying more than one experimental unit 3.) Similarity: areas were selected that comprise relatively similar habitat complexes and deer densities that are representative of the overall area 4.) Elk populations: areas were selected to minimize the number of elk present during normal winters �37 Units A and B are receiving the nutrition enhancement treatment in a cross-over experimental design, and are being used to address P.N. Objective 1. Unit A served as the treatment unit, while Unit B served as the control, for the first 2 winters ofresearch (2000 - 2002). Beginning November 2002, Unit B received the treatment while Unit A served as the control. Upon completion of P.N. Objective 1, two additional winter range experimental units will be used to conduct phase 2 of the research, or P.N. Objective 2. Habitat in one unit will be manipulated to set back plant succession (treatment), while habitat in the other unit will remain unchanged (control) throughout the experiment. Year 2000-01 Unit A Unit B 2001-02 2002-03 2003-04 Figure 1. Schematic representation of experimental units and nutrition enhancement treatment allocation. Units A and B are located in winter range habitat on the Uncompahgre Plateau in southwest Colorado. The nutrition enhancement cross-over design will encompass 4 years. The 2 experimental units (A and B) receiving the nutrition enhancement treatment are (Figs. 2 and 3): (1) Experimental unit A includes the Colona Tract of the Billy Creek State Wildlife Area and adjacent land, located approximately 13 km south of Montrose, CO adjacent to U.S. Hwy 550 South. The experimental unit is located within the Colona USGS 7.5 Minute Quadrangle, and roughly includes the polygon defined by the following Zone 13 UTM coordinates: (1) 254000 E, 4250200 N; (2) 252700 E, 4249400 N; (3) 254700 E, 4245600 N; and (4) 256200 E, 4246600 N. (2) Experimental unit B includes Shavano Valley and adjacent land extending west to the Dry Creek Rim. Shavano Valley is located approximately 13 km west of Montrose, CO. The experimental unit is located within the Dry Creek Basin and Montrose West Quadrangles (USGS 7.5 Minute), and roughly includes the polygon defined by the following Zone 13 UTM coordinates: (1) 238400 E, 4262600 N; (2) 232400 E, 4256700 N; (3) 235000 E, 4253600 N; and (4) 239500 E, 4258200 N. In late April and May, prior to fawning, deer from the winter range experimental units migrate to summer range. The summer range study area is defined by movements of the radio-collared deer, which encompass > 1000 mi2 covering the southern portion of the Uncompahgre Plateau and adjacent San Juan Mountains to the south and east (Fig. 2). The summer range study area extends north to the Dry Creek river drainage on the Uncompaghre Plateau, south to Mineral Creek near Silverton, CO, east to the Big Blue river drainage, and west to the San Miguel River canyon. However, a majority of the radio-collared deer summer on the Uncompahgre Plateau between Dry Creek to the north and Horsefly Peak to the south. Winter range elevations range from 1830 m (6000 ft) in Shavano Valley to 2318 m (7600 ft) adjacent to the Dry Creek Rim above Shavano Valley. Winter range habitat is dominated by pinyon-juniper with interspersed sagebrush adjacent to agricultural fields in the Shavano and Uncompahgre Valleys. Summer range elevations occupied by deer range from 1891 m (6200 ft) in the Uncompahgre Valley to 3538 m (11,600 ft) in Imogene Basin southwest of Ouray, CO. Summer range habitats are dominated by sprucefir (Picea spp.-Abies spp.), aspen (Populus tremuloides), ponderosa pine (Pinus ponderosa), Gambel oak (Quercus gambelii), and to a lesser extent, sagebrush and pinyon-juniper at lower elevations. �38 ... ~-· .... ..., ·--.-~ ./ .......a ·<~~~-..:::,,~<•/ ,~---··. i-~~sa County ~- GRAND JUNCTl6N-:'. ---~-:~ ··t Gunnison County Winter Range Exp. Units Summer Ran2e '·· Montrose, County (·'"--''···········•'•""··· .- .•• .... ) :\ ·•. : ·-..t., Figure 2._ Location of Colona and Shavano (Units A and B) experimental units in Game Management Unit 62 on the Uncompahgre Plateau, southwest Colorado; and location of the summe:r range study area throughout the southern Uncompahgre Plateau and adjacent San Juan Mountains �39 Figure 3. Colona and Shavano experimental units (Units A and B), located in Game Management Unit 62 on the Uncompahgre Plateau, southwest Colorado. Polygons represent the nucleus of each experimental unit, which is where animals have been collared and the nutrition enhancement treatment delivered. �40 Response Variables The response variables are fetal and neonatal survival rates, early winter fawn:doe ratios, and overwinter fawn survival rates. The nutrition enhancement treatment is delivered to deer from December through April, fetus survival is assessed during June, neonate survival is measured from June to December, and fawn:doe ratios are measured during the following December and January (1 year after the treatment was initiated). Overwinter fawn survival is measured from December to June as a direct result of the current winter's treatment. We are measuring these response variables in each experimental unit (treatment and control) to determine whether enhanced winter nutrition of adult does increases subsequent newborn fawn production and survival, and whether enhanced winter nutrition of 6-mo. old fawns directly increases overwinter fawn survival. Ultimately, these measurements provide an assessment of the effect of winter range habitat quality on yearling recruitment, and thus population productivity. We are also measuring overwinter and annual survival of adult does as a function of enhanced winter nutrition. Sample Size Fetus/Neonate Survival: We were primarily interested in survival of newborn fawns from radio-collared does that occupy the 2 winter range experimental units. Fetus survival is also important, but difficult to measure. Fetus rates from a sample of radio-collared does can be measured in winter, but the fate of all fetuses cannot be determined the following June because oflogistical constraints. Fetus survival rates can only be measured from some unpredictable fraction of the radio-collared doe sample, making sample size calculations of limited use. Thus, our sample size calculations were based on quantifying neonate survival, not fetus survival. For neonate survival, a sample size of 40 neonates per experimental unit per year provides power of 0.81 to detect a difference of0.15 in survival between 2 experimental units if survival among control fawns is 0.40. We assumed a control survival rate of 0.40 based on neonate survival rates measured recently for the Uncompahgre deer population (Pojar 2000) in combination with December fawn:doe ratios measured during the late 1980's and 1990's, when the Uncompahgre population declined (B. E. Watkins, unpublished data). Based on Bishop et al. (2002), we determined that 60 radio-collared does (30 treatment and 30 control) equipped with vaginal implant transmitters (VITs) would be necessary to capture a minimum of 80 newborn fawns. We also assumed that some fawns would be captured from other treatment and control radio-collared does not equipped with VITs. The 60 radio-collared does with VITs are also being used to evaluate fetus survival; however, logistical constraints limit the power of fetus survival comparisons among experimental units. Early winter fawn:doe ratios: We desired to detect an effect size, i.e., an increase in fawn:doe ratios in response to the treatments, in the range of 15 to 20 fawns per 100 does. These values were based on simple population models with overwinter fawn survival of 0.444, adult female survival of0.853, and December fawn:doe ratios of 66 fawns per 100 does to obtain a stationary population (Unsworth et al. 1999). Based on surveys of the Uncompahgre deer population during the 1990's, the standard deviation of the fawn:doe ratio for groups with at least one adult female was 57, with a mean of 41. Using an expected standard deviation of 57, the standard error of the mean fawn:doe ratio for 40 radio-collared does is 57/(40 112 ) = 9.0, which is the expected standard deviation of measured fawn:doe ratios on each experimental unit. We assessed power using a two-sample t-test with a sample size of 4, representing the 4 years of the study where fawn:doe ratios are being measured in response to enhanced nutrition. Our power to detect an increase of 20 fawns per 100 does based on classification of 40 radio-collared doe groups in each experimental unit is about 0.87. A sample size of 40 fawns per experimental unit per year provides a power of 0.81 to detect a difference of 0.15 in survival between 2 experimental units if survival on the control unit is 0.40. We expected to �41 see an increase in fawn survival (effect size) of approximately 0.15, because this was the difference measured in the density reduction experiment conducted by White and Bartmann (I 998). Adult and 6-month Old Fawn Capture During November and December, adult does and 6-month old fawns were captured using baited drop nets (Ramsey 1968, Schmidt et al. 1978) and helicopter net guns (Barrett et al. 1982, van Reenen 1982). Drop nets were baited with certified weed-free alfalfa hay and apple pulp. Drop nets were used as the principle capture technique for a 3-4 week capture period; helicopter net-gunning was then used at the end of the drop-net capture to secure the remainder of deer needed to meet our target sample sizes. All deer were hobbled and blind-folded after being captured. Deer captured via drop nets were carried away from the net to an adjacent handling site using stretchers. Deer were fitted with leather radio collars equipped with mortality sensors, which cause an increase in pulse rate after remaining motionless for 4 hours. Permanent collars were placed on adult females, while temporary collars were placed on fawns. To make collars temporary, one end of the collar was cut in half and reattached using rubber surgical tubing; fawns shed the collars ~6 months post-capture. A rectangular piece of flexible plastic (Ritchey® neck band material) engraved with a unique identifier was stitched to the side of each collar. The unique identifier consisted of 2 symbols for adult females, and 1 symbol on 2 different colors of plastic for fawns. The identifiers were necessary to visually identify deer from the ground. This allowed us to effectively document use of the treatment, measure fawn:doe ratios from the ground, and assess experimental unit population size via mark-resight estimators. We recorded the weight, hind foot length and chest girth of each deer, and collected blood samples to evaluate disease prevalence. During late February and early March, an additional 30 adult female deer were captured in each experimental unit by net-gunning. Captured deer were ferried by the helicopter to a central processing location, where deer were carried by stretchers to a tent for handling. For each captured deer, we used ultrasonography to measure pregnancy status, fetal rate, and body condition. Only pregnant does were retained and radio-collared. We then inserted a vaginal implant transmitter (VIT) in each doe as a technique for locating the timing and location of her birth site the following June. We also recorded the weight, hind foot length and chest girth of each deer, and collected blood samples to evaluate disease prevalence. Body Condition and Reproductive Status We estimated body fat of treatment and control adult does during mid-late winter using an Aloka 210 (Aloka, Inc., Wallinford, Conn.) portable ultrasound unit with a 5 l\.11-Iz linear transducer. We measured maximum subcutaneous fat thickness on the rump (MAXFAT) following the methodology of Stephenson et al. (1998, 2002). We also measured thickness of the longissimus dorsi muscle via ultrasound (Cook et al. 2001, Stephenson et al. 2002). A small area of hair was shaved to ensure contact between the transducer and the skin. Vegetable oil was applied to the shaved area for conduction purposes and fat/muscle thickness was measured using electronic calipers. We coupled the ultrasound measurements with body condition scores (BCS) obtained from palpation of the ribs, withers, and rump (Cook 2000). MAXF AT and rump BCS measurements were combined into a condition index used to estimate percent body fat (Cook and Cook 2002): % Fat= -6.6387617 + 7.4271417x - l. l 1579443x2 + 0.07733803x3 where x = rLIVINDEX = (MAXFAT-0.15) + rump BCS (ifMAXFAT < 0.15, then rLIVINDEX = rump BCS). The rLIVINDEX and body fat regression was initially developed and validated for elk by Cook et al. (2001 ), and then modified by incorporating a validation of MAXF AT for mule deer performed by Stephenson et al. (2002). �42 During mid-late winter 2003, we also evaluated differences in serum thyroid hormone concentrations between treatment and control adult does. Specifically, we measured total thyroxine (T4), free T4 (Ff4), total tri-iodothyronine (T3), and free T3 (Ff3) following the methodologies of Watkins et al. (1983, 1991). Blood samples were collected at the time of capture, and serum hormone analyses were performed by the Michigan State University Animal Health Diagnostic Laboratory (East Lansing, Michigan). We compared serum thyroid hormone concentrations between treatment and control adult does, and also compared hormone levels to body fat estimates derived from the ultrasonography. We quantified reproductive status (Stephenson et al. 1995, Pojar 2000) with ultrasound via transabdominal scanning using a 3 MHz linear transducer. We searched for fetuses by scanning a portion of the abdomen that was shaved caudal to the last rib and left of the midline. We systematically searched each uterine horn to identify fetal numbers ranging from Oto 3. Whenever possible, we measured eye diameter of each fetus to approximately estimate fetal age and parturition date. Vaginal Implant Transmitters (VITs) We used VITs manufactured by Advanced Telemetry Systems, Inc. (Isanti, MN). The VIT was 76 mm long, excluding antenna length, and had 2 plastic wings with a width of 57 mm when fully spread apart. The plastic wings were used to retain the transmitter in the vagina until parturition. The VIT weighed 15 grams and contained a 10-28 lithium battery programmed to a 12-hour on/off cycle. The diameter of the transmitter/battery was 14 mm, and was encased in an impermeable, water-proof, electrical resin. The transmitter contained an embedded heat-sensor which dictated the frequency pulse rate. When the heat sensor dropped below 90°F, synonymous with transmitter expulsion from the deer, the pulse rate changed from 40 PPM to 80 PPM. VIT batteries were programmed to be active from 0430 to 1630 hrs prior to daylight savings, and thus were active from 0530 to 1730 hrs after daylight savings and during the fawning period. The VIT was inserted into deer using a vaginoscope (Jorgensen Laboratories, Inc., Loveland, CO) and alligator forceps. The vaginoscope was 6" long with a 5/8" internal diameter and had a machined end (smooth surface) to minimize trauma when inserted into the vagina. A discreet mark was placed on the applicator showing the appropriate distance it should be inserted into the deer. The length of a typical mule deer vaginal tract was obtained by taking measurements from road-killed deer and/or other fresh deer carcasses obtained in the study area. Prior to use in the field, VITs were sterilized using a Chlorhexidine solution, air-dried, and sealed in a 3" x 8" sterilization pouch. Sterilization containers with Chlorhexidine solution were used on site during capture to sterilize the vaginoscope and alligator forceps between each use. A new pair of nitrile surgical gloves was used to handle the vaginoscope and VIT for each deer. To insert a VIT, the plastic wings were folded together and placed into the end of the vaginoscope. We then liberally applied sterile KY Jelly to the scope and inserted it into the deer's vagina to the point where the mark on the applicator was reached. The alligator forceps, which extended through the vaginoscope to hold the VIT, was held firmly in place while the scope was pulled out from the vagina. This procedure pushed the VIT out of the scope into the vagina, and the plastic wings spread apart to hold the transmitter in place. The transmitter antenna was typically flush with the vulva, but on occasion extended up to 1 cm beyond the vulva. The tip of the antenna was encapsulated in a wax bead to protect the deer. Neonate Fawn Capture During June we relocated each of the radio-collared does having a VIT each morning using aerial and ground telemetry. Flights began at 0530 hr and were usually completed by 1000 - 1100 hrs. The early flights were crucial for detecting fast signals because shed VITs could exceed 90 °F by mid-day if shed in the open, which caused them to switch back to a slow ("pre-birth") pulse. When a fast ("postpartum") �43 pulse rate was detected, we located the VIT from the ground to determine whether it was shed at the birth site. If the transmitter was located at the birth site, we identified whether any fawn(s) were stillborn. If the fawn(s) were no longer present at the birth site, or could not be found in the vicinity of the birth site, we located the radio-collared doe and searched for fawns at her location. All personnel involved wore surgical gloves to help minimize human scent when handling fawns. For each doe, we attempted to locate each of her fawns and document whether any fawns were stillborn. We attempted to account for each doe's fetuses in order to quantify in utero fetal survival from February to birth. We placed a dropoff radio-collar on each live fawn; radio collars were constructed with elastic neck-band material to facilitate expansion. Hole-punched, leather tabs extended from the end of the elastic and from the transmitter for attachment purposes. Collars were made temporary by cutting the leather tab extending from the elastic and reattaching the leather with latex tubing, which caused the collars to shed from the animal >6 months post-capture. For each fawn, mass and hind foot length were recorded, and a nasal swab sample was collected to screen for Bovine Viral Diarrhea. We then recorded basic vegetation characteristics of the birth site and promptly exited the site. We also routinely located treatment and control radio-collared does not having VITs and attempted to capture their fawns to help achieve our targeted sample size. Each of these does had been previously captured during the research, and were present on either the treatment or control experimental unit during winter. Measurement of Survival Rates and Fawn:Doe Ratios We measured survival rates by radio-monitoring collared deer from the ground and air to determine fate (live/mortality). We also attempted to determine the cause of each mortality, with a primary goal of distinguishing between predation and non-predation mortality causes. Deer were radio-monitored from the ground on a daily basis throughout the year and from the air on approximately a biweekly basis. We were able to detect signals from nearly all radio-collared deer each day during winter, which typically allowed us to arrive at mortality sites within 24 hours of the mortality event. During summer and migration periods, deer were distributed widely and thus were more difficult to radio-monitor. All radiocollared neonates were checked daily throughout the summer and fall, whereas some adult and yearling deer could not be ground-monitored on a routine basis. In result, we typically located neonate mortalities within 24 hours of death, but some adult deer mortalities were not detected for several days, or on rare occasion, for one or more weeks. Fresh, intact neonate carcasses were collected and submitted to the Colorado Division of Wildlife's Wildlife Health Laboratory or the Colorado State University Diagnostic Laboratory for necropsy and tissue analyses. Fresh, intact adult and 6-month old fawn carcasses were also submitted for laboratory necropsy when feasible. Field necropsies were performed on all other deer mortalities, and when appropriate, tissue samples were collected and submitted for analysis. Each winter we used the radio-collared does to measure fawn:doe ratios in each experimental unit. The resulting fawn:doe ratio is a measurement of the previous year's treatment effect. We measured fawn:doe ratios using 2 techniques: (1) We located the sample of radio-collared does in each experimental unit from a fixed-wing airplane, and used the set of locations to define boundaries for the experimental unit. Shortly after (i.e. 1-2 days), we used a helicopter to systematically fly the defined unit and classify all deer groups encountered. For each group, we documented whether a radio-collared doe was present. (2) We located each radio-collared doe by radio telemetry from the ground. The group of deer with the collared doe was counted and classified by age and sex. Both methods were employed to gather as much information as possible to determine whether there was a treatment effect. The "true" value cannot be measured perfectly because of the inherent biases and potential sources of error associated with each technique. Thus, by employing both techniques, we had a greater chance of fully understanding whether the treatment caused an effect. �44 Treatment Delivery Deer nutrition was enhanced in the treatment area by providing a safe, pelleted supplemental feed. The supplemental feed was developed through extensive testing with both captive and wild deer (Baker and Hobbs 1985, Baker et al. 1998), and has been safely used in both applied research and management projects. Pellets were distributed daily using 4wd pickup trucks and ATVs on primitive roads throughout the experimental unit to provide a food source for the entire deer population in the treatment unit. Each 501b. bag of pellets was carried ::;200m from the truck/ATV and distributed by hand in approximately 2030 small piles of feed in a linear fashion. Numerous bags were distributed in successive order allowing us to create linear lines of feed that spanned most of the treatment area, which prevented animals from concentrating in any single location. This feeding technique also prevented dominant animals from restricting access to the food supply because of the large area over which pellets were distributed. We supplied pellets ad libitum such that a small residual remained when the next day's ration was provided. Collared deer were closely monitored to ensure that treatment deer remained in the experimental unit and actually consumed the feed, and to make sure that non-treatment deer remained in the control unit, which they did. The few treatment adult does that moved away from the treatment unit were withdrawn from the sample for purposes of measuring treatment effects. However, to avoid.any biases, all 6-month old fawns captured in the treatment unit were included in survival analyses regardless of whether they accessed the supplement or not. This was because some fawns died shortly after capture (e.g. 2-3 weeks), before we could document whether they had access to the feed. Also, very few fawns that survived more than 2-3 weeks moved away from the treatment unit. The pelleted ration was commercially produced in the form of 2x l x0.5-cm wafers (Baker and Hobbs 1985). Feed constituents (i.e. digestibility, protein, gross energy etc.) vastly exceeded those of typical winter range deer diets; exact constituent values are provided by Baker et al. (1998). When provided ad libitum, the feed should have allowed deer to meet or exceed nutritional requirements for growth and maintenance (Ullrey et al. 1967, Verme and Ullrey 1972, Thompson et al. 1973, Smith et al. 1975, Baker et al. 1979, Holter et al. 1979). The basis for feeding such high quality pellets was to ensure that the treatment (enhanced nutrition) was effectively delivered to the deer. Our intent was not to determine the exact level of nutrition necessary to increase fawn recruitment, but rather to determine if nutrition is a limiting factor to recruitment. If nutrition is in fact limiting, we will rely on habitat manipulation treatments to evaluate what exactly can be done via management to increase fawn survival and recruitment. Statistical Methods A preliminary fawn:doe ratio analysis was completed using PROC MIXED in SAS (SAS Institute 1997). We used a reduced model with experimental unit as the independent variable; we considered experimental unit as a fixed effect and radio-collared does within an experimental unit as random effects. Survival rates were calculated using a Kaplan-Meier survival analysis (Kaplan and Meier 1958, Pollock et al. 1989), and contrasted among experimental units and sexes using a chi-square analysis. For neonate survival analyses, we used a common entry date because a staggered entry would ha.ve biased survival rates low due to early mortalities that occurred before most of the sample was captured. We modeled overwinter fawn survival with a logistic regression model using PROC LOGISTIC in SAS (SAS Institute 1989a); model selection was performed using Akaike's Information Criterion (AlC) (Burnham and Anderson 1998). Survival was modeled as a function of the nutrition enhancement treatment, sex, year, and capture mass. We used a general linear model in PROC GLM in SAS (SAS Institute 1989b) to test for differences in estimated percent body fat between treatment and control adult does and a multivariate model to test for differences in T4, Ff4, T3, and FT3 thryoid hormones between treatment and control does. We then used PROG REG (SAS Institute 1989b) to evaluate the relationship between estimated �45 percent body fat and serum thyroid hormone concentrations. We analyzed fetus survival directly with a binomial survival rate for the subset of fetuses with known fates. We also indirectly analyzed fetus survival by comparing the February fetus rate with the number oflive newborn fawns/doe observed in June using a change-in-ratio estimator (White et al. 1996). Other results in this report are presented as data summaries incorporating means and standard errors, or in some cases, raw data values. These results are incomplete and preliminary in nature, and should be treated as such. RESULTS AND DISCUSSION Deer Capture During November and December 2000-2002, we captured and radio-collared 122 adult female mule deer evenly distributed among the treatment and control units. We also captured and radio-collared 160 6month old fawns during November and December 2001-2002 (40 fawns/unit/year). Due to budgeting constraints, we were unable to radio-collar 6-month old fawns during 2000. We captured an additional 94 adult females during late February and early March 2002-2003 and equipped them with radio collars and VITs. During June 2002-2003, we captured and radio-collared 157 newborn fawns from radio-collared adult females. Thus, the following results are based upon radio-monitoring of 533 individual mule deer evenly distributed among treatment and control units during November 2000-June 2003. Treatment Delivery 2000-01 From December 15, 2000, through April 19, 2001, we distributed 88 tons of the pelleted ration. For most of the winter and spring, on average, we distributed 0.85 tons of feed each day throughout 22 feeding sites across the 2.3 mi2 treatment unit. Deer were fed ad libitum because there was always residual feed remaining the next day during the feeding routine. Each sack was distributed in approximately 20-30 distinct, small piles, resulting in > 1000 small piles of feed throughout the treatment unit. This effort allowed deer to effectively access the feed in small groups, and no aggression was ever observed among deer seeking access to the feed. By distributing the feed in this manner, we were able to avoid the negative aspects associated with large-scale feeding operations. Deer adapted to the pelleted supplement right away and utilized it extensively throughout the winter. We continually monitored deer use of the feed from ground observation points, where we obtained 440 visual observations of radio-collared does consuming the feed. These observations, coupled with daily radio-monitoring and periodic aerial relocations, indicate 32 of the 37 radio-collared treatment does spent the entire winter and spring within the boundaries of the treatment unit and received the supplement on a daily basis. Mark-resight population estimates from March helicopter (489 deer, SE= 62) and ground (494 deer, SE= 81) surveys, coupled with feed consumption, indicate we fed roughly 450 to 500 deer during most of the winter and spring. Feed consumption declined coincident with spring green-up, although deer continued to use the feed through mid-late April, at which point they began migrating to summer range. We also fed approximately 25 to 30 elk, but the elk did not affect deer access to the feed. Deer in the control experimental unit did not receive feed or any other treatment. Based on helicopter mark-resight surveys,· the deer density in the treatment unit in December was 120 deer/mi2 (SE= 9), but increased shortly after and was 213 deer/mi2 (SE= 27) in March. Deer densities in the control unit changed little from 83 deer/mi2 (SE= 12) in December to 101 deer/mi2 (SE= 14) in March. �46 2001-02 From December 15, 2001, through April 25, 2002, we distributed 194 tons of the supplement throughout the treatment unit. For most of the winter and spring, we distributed 2.0-2.1 tons of feed each day. The dramatic increase in supplement distribution from the previous year occurred because a large number of elk descended into the Uncompahgre Valley during mid-late fall/early winter. Elk arrived in unusually large numbers throughout much of the valley prior to the onset of treatment delivery. Once feeding was initiated, approximately 300-500 elk adapted to the feed and remained in or around the 2.3 mi2 treatment unit throughout most of the winter. Given myriad logistical and budgetary constraints, 2.1 tons was the maximum amount of feed we could routinely deliver on a daily basis. Feed was not delivered ad libitum to all deer and elk in the treatment unit throughout the winter because residual feed was rarely observed during the next day's distribution. However, daily field observations indicated most deer approached ad libitum consumption of the supplement. In contrast to the previous winter, deer were waiting for the daily supplement to arrive each morning. Deer then consumed the supplement immediately after it was distributed. Elk were rarely observed utilizing the feed until late morning or afternoon, and elk continued to forage in fields below the treatment unit, whereas deer did not. We observed numerous radio-collared deer consuming the pelleted supplement each day; not all of these observations were recorded because of time constraints with distributing the feed. Given this time limitation, we still recorded 818 observations of radio-collared deer consuming the supplemental feed (497 collared doe observations and 321 collared fawn observations). Most days,> 100 and sometimes 200-300 deer were observed utilizing the pellets during the course of distributing the supplement. These observations rarely included elk; thus, deer-elk competition was minimized because of temporal differences in feeding, and deer clearly had first access to the feed. 2002-03 Beginning December 2002, we switched the treatment and control units consistent with the cross-over experimental design. From December 15, 2002, through April 30, 2003, we distributed 97 tons of the supplement throughout the new treatment unit, which had served as the control unit the previous 2 years. The supplement was distributed daily throughout 29 sites over a larger area (~ 7 mi2) than the first 2 years of research because of the greater size of the experimental unit and broader distribution of radio-collared deer. Residual feed was always present throughout the winter, thus deer were fed ad libitum. Only small groups of elk periodically accessed the supplement, and did not affect deer access. We obtained 286 observations of radio-collared deer consuming the supplement, which were difficult to obtain because the supplement was spread out over a large area and only a single feed site could be observed at any given moment. We also used daily ground radio-monitoring and periodic aerial relocations to document deer access to the supplement. Body Condition Estimated percent body fat of adult does during late February and early March of 2002 and 2003 was significantly higher for treatment deer than control deer {F1, 90 = 108.21, P < 0.001). Over both years combined, mean predicted body fat was 10.4% (SE= 0.48) for treatment adult does and 4.0% (SE= 0.36) for control does. The interaction of experimental unit x year for predicted body fat was also significant (F1, 9o = 21. 79, P < 0.001). This interaction occurred because the difference in body fat between treatment and control deer was greater during 2003 than during 2002. During 2002, mean predicted body fat was 8.2% (SE= 0.92) for treatment adult does and 5.0% (SE= 0.71) for control does, whereas during 2003, mean predicted body fat was 11.7% (SE= 0.35) for treatment does and 3.4% (SE= 0.35) for control does. The body fat estimates reported here should accurately reflect deer, but may be further refined in the �47 future as additional research provides more data on the relationship between body condition indices and estimated percent body fat. In 2003, serum thyroid hormone concentrations were higher in treatment does than control does (F4. 52 = 32.59, P < 0.001). T4 was the most important thyroid hormone in describing the single canonical variable (1.78*T4 - 0.04*T3 + 0.20*FT4 - 0.27*FT3). Not surprisingly, there was a high partial correlation between T4 and FT4 (r = 0.77, P < 0.001) and between T3 and FT3 (r = 0.73, P < 0.001), which has been documented previously (Watkins et al. 1983). When treated as 4 separateANOVAs, T4 (F1,55= 127.45, P < 0.001), FT4 (F1, 55 = 81.72, P < 0.001), and T3 (F1,5 5== 5.39, P== 0.024) were significantly higher in treatment does than control does, whereas FT3 levels were less different among treatment and control deer (F1. 55 == 2.59, P == 0.113). Given these results, we evaluated the relationship between T4 concentrations and estimated percent body fat (derived form ultrasound and BCS indices) using a simple linear regression model(% Fat== -5.114 + 0.106*T4,? = 0.59, P < 0.001). Similar correlations between T4 and actual percent body fat during mid-late winter have been previously documented for white-tailed deer and elk (Watkins et al. 1991, Cook et al. 2001). Fetus Survival and Pregnancy/Fetus Rates We began measuring fetus survival in 2002 as part of our effort to capture and radio-collar newborn fawns born from radio-collared does. Similar numbers of stillborns were observed between treatment and control does during both 2002 and 2003, so all fetus survival analyses reported here represent pooled estimates. In February-March 2002, 36 of 38 adult does captured were pregnant, thus the pregnancy rate was 0.95 (SE= 0.036). We measured an average of 1.80 fetuses/doe (SE== 0.10, n = 36), which included 1.77 fetuses/doe (SE== 0.14, n = 18) in the treatment unit and 1.83 fetuses/doe (SE== 0.15, n == 18) in the control unit. During June 2002, we determined the fate of all fetuses (live or stillborn) from only 14 of the 36 VIT does, largely because of a high VIT battery failure rate. The survival rate of fetuses (n == 22) from these 14 does was 0.86 (SE= 0.073). We also assessed fetus survival using a change-in-ratio estimator between the fetal rate measured in February-March and the observed number of live fawns/doe postpartum in June. In June 2002, considering all does (n == 43) that we located any fawn from, whether live or stillborn, we observed 1.42 (SE== 0.11) live fawns/doe postpartum. This rate should represent a conservative estimate of live fawns/doe postpartum because we inevitably failed to locate all live fawns from each doe. In other words, this estimate would treat any unaccounted fetuses (from the February measurement) as if they were stillborns. For radio-collared does that did not have VITs, and thus we did not have a winter fetus rate measurement, singletons would infer that either the deer only had 1 fetus, or that the other fetus died. It is likely that some of these singletons had a twin that we did not locate. This equates to a conservative fetus survival rate estimate of 0.79 (SE== 0.18). In February-March 2003, 58 of 63 adult does captured were pregnant, resulting in a pregnancy rate of 0.92 (SE== 0.034). Critical personnel and equipment for measuring fetus rates were not continuously available due to capture delays associated with helicopter mechanical problems. Some of the deer fetus counts were performed by inexperienced observers without optimum ultrasound equipment. VITs worked very well, though, allowing us to determine fetus numbers at parturition for many of the deer. Thus, we determined winter fetus rates by using the greatest fetus count for each individual deer, whether obtained using ultrasound during February-March or by locating newborn fawns and stillborns at birthsites during June. We were unable to determine a fetus count for 8 treatment deer because only pregnancy was established with ultrasound and no birthsite assessments were possible in June. These 8 deer were removed from the fetus rate estimates. Of the 50 deer where a fetus count was obtained, 5 were yearlings (2 treatment yearlings, 3 control yearlings). We measured 1.74 fetuses/doe (SE== 0.069, n = 50) overall including yearlings, and 1.82 fetuses/doe (SE== 0.066, n == 45) excluding yearlings. Fetus rates with yearlings included were 1.77 fetuses/doe (SE== 0.091, n == 22) in the treatment unit and 1.70 �48 fetuses/doe (SE= 0.10, n = 28) in the control unit. During June 2003, we determined the fate of all fetuses (live or stillborn) from 33 of the 58 VIT does; the good success was based on VITs commonly being shed at birthsites. The survival rate of fetuses (n = 58) from these 33 does was 0.97 (SE= 0.024). In June 2003, incorporating all does (n = 71) that we located any fawn from, whether live or stillborn, we observed 1.49 (SE= 0.072) live fawns/doe postpartum. Using the change-in-ratio estimator described above, this results in an overall conservative fetus survival rate estimate of 0.86 (SE= 0.15). Neonatal Survival/Fawn: Doe Ratios 2001 In December 2000, at the beginning of the study and prior to the first year's treatment delivery, fawn:doe ratios were similar in the 2 experimental units. Pre-treatment fawn:doe ratios were 52.6 fawns: 100 does (SE= 5.3) in the treatment unit, and 51.6 fawns: 100 does (SE= 5.0) in the control unit. In late December 2001 and early January 2002, following the first year's treatment, we conducted 2 age classification helicopter surveys in the treatment and control units. On 12/23/01, we observed 52.8 fawns: 100 does (SE = 6.7) in the treatment unit, and 36.7 fawns:100 does (SE= 3.8) in the control unit. On 1/8/02, we observed 54. 7 fawns: 100 does (SE = 6.6) in the treatment unit, and 50.5 fawns: 100 does (SE= 6.0) in the control unit. During December 2001 - February 2002, we obtained fawn:doe ratio estimates from ground observations of radio-collared deer groups for both treatment and control deer. This survey resulted in 61.2 fawns:100 does (SE= 7.8) in the treatment unit, and 74.5 fawns:100 does (SE= 8.5) in the control unit, although the result was not statistically significant (h4 = 1.16, P = 0.249). The fawn:doe ratio results are conflicting, and clearly do not provide evidence that there was any treatment effect. In short, we concluded that the nutrition enhancement treatment did not cause an increase in neonatal production and survival during 2001. However, our results, in conjunction with a December estimate of 64 fawns: 100 does for the entire Uncompahgre deer population (B.E. Watkins, unpublished), indicate fawn production and survival was good during 2001. The observed fawn:doe ratios coupled with overwinter fawn survival and annual adult survival rates indicate the deer population was increasing. Considering the past 1-2 decades, this was an atypically good year for the Uncompahgre deer population. It would appear that whatever set of environmental conditions have led to a declining deer population were not present during 2001 in the same manner as in the past. Our main interest lies in observing the effect of the treatment on the deer population in a year where fawn:doe ratios are lower for the population as a whole, similar to what they have been much of the past 15 years. 2002 During June - December 2002, following the second year's treatment, we measured neonate survival directly using radio-collared fawns; however, sample sizes were based on a technique assessment ofVITs and were relatively small for contrasting treatment and control survival of neonates (Bishop et al. 2002). Treatment fawn survival was 0.613 (SE= 0.115, n = 29) and control fawn survival was 0.511 (SE= 0.108, n = 25). In late December 2002 and early January 2003, we once again conducted 2 age classification helicopter surveys in the treatment and control units. On 12/31/02, we observed 91.9 fawns:100 does (SE= 8.4) in the treatment unit, and 52.2 fawns: 100 does (SE= 6.9) in the control unit. On 1/21/03, we observed 52.6 fawns:100 does (SE= 6.4) in the treatment unit, and 36.8 fawns:100 does (SE= 3.9) in the control unit. The combined helicopter survey data indicated 68.1 fawns: 100 does (SE= 5.6) in the treatment unit and 42.8 fawns: 100 does (SE= 3.5) in the control unit. Oppositely, fawn:doe ratio estimates from ground classifications of doe groups during December 2002 - February 2003 were 47.7 fawns: 100 does (SE= 6.3) in the treatment unit, and 63.4 fawns:100 does (SE= 7.5) in the control unit (t108 = 1.61, P = 0.110). As in 2001, fuwn:doe ratio results were conflicting. Helicopter survey data �49 varied between 2 different flights, but consistently indicated a treatment effect. Ground classification data did not indicate a treatment effect. Also, survival data combined with age ratio data indicate neonate production and survival was reasonably favorable during 2002, and not indicative of the low fawn recruitment observed during the late l 980's and l 990's. Our results from 200 I and 2002 point out the inherent difficulties and biases associated with precisely measuring fawn:doe ratios, particularly in this research study. Ratios obtained from helicopter surveys were based on 2 short-duration flights/unit/year over spatially small units. Helicopter surveys were complicated by high deer densities in heavy cover, ma.king both deer detection and fawn:doe classifications a considerable challenge. There is a variety of potential biases that may have affected the helicopter surveys, including differential sightability of does and fawns, double classification of some deer, and incorrectly classifying yearling bucks with small antlers. Ground fawn:doe ratio observations of radio-collared doe groups were made using spotting scopes and field glasses, where we commonly studied the deer for some time. Incorrect classifications during these surveys were likely minimal. For example, small-antlered yearling bucks (e.g. 3 - 6" spikes) were detected from the ground, whereas they were undoubtedly missed on occasion during helicopter surveys. We also obtained repeated observations for some of the radio-collared doe groups from the ground. The main potential bias affecting ground fawn:doe classifications was how observations were made. Many of the ground classifications in the Shavano Valley experimental unit were made by radio-tracking does during the day. On the other hand, a majority of ground classifications in the Colona experimental unit were based on observing deer groups as they entered openings to feed during the late afternoon. Given the inherent difficulties of measuring fawn:doe ratios in the 2 experimental units, and the lack of a clear indication as to the effectiveness of the treatment, we intensified efforts in 2003 to directly measure survival of neonate fawns born from treatment and control radio-collared does. At the completion of the research, we will test whether enhanced winter nutrition of adult does improved newborn fawn survival based on a three-year model of radio-collared neonate survival data. We will continue to measure early winter fawn:doe ratios, but the data will be used cautiously to make inferences regarding treatment effects. 2003 During June 2003, we captured and radio-collared 103 newborn fawns born from treatment and control radio-collared does (55 treatment fawns, 48 control fawns). The VITs worked well; we captured fawns from 41 of the 54 does fitted with VITs. As of late September 2003, treatment fawn survival was 0.745 (SE= 0.059) and control fawn survival was 0.614 (SE= 0.073). Neonate Mortality Causes During 2002, 11 of the 29 treatment fawns died from the following causes: 3 - coyote predation, 2 - bear predation, 1 - felid predation, 1 - predation where the predator was undetermined, 1 - disease/ malnutrition, 1 - abandonment, 1 - road-kill, and 1 - trauma/injury. Twelve of the 25 control fawns died: 6 - malnutrition/disease, 3 - coyote predation, 1 - felid predation, 1 - bear predation, and 1 predation mortality where the predator was undetermined. Thus, 13% of all radio-collared fawns died from malnutrition, 11 % from coyote predation, 6% from bear predation, 4% from felid predation, 4% from predation (unknown predator), and 6% from miscellaneous causes. Currently (June - September 2003), 14 of the 55 treatment fawns have died from the following causes: 6 - disease/malnutrition/starvation, 4 - coyote predation, 3 - predation (unknown predator), and 1 - felid predation. Over the same time period, 18 of the 48 control fawns have died: 8 - coyote predation, 4 - disease/malnutrition/starvation, 3 felid predation, 1 - bear predation, and 2 - unknown. Thus, as of the end of September during 2003, 12% �50 of all radio-collared fawns have died from coyote predation, 10% from disease/malnutrition/starvation, 4% from felid predation, 3% from predation (unknown predator), 1% from bear predation, and 2% from unknown causes. Overwinter Fawn Survival and Mortality Causes During winter 2001-02 (Dec 1, 2001- May 31, 2002), the survival rate of fawns was significantly greater (x\ = 13.216, P < 0.001) in the treatment unit (S(t) = 0.865, SE= 0.056) than in the control unit (S(t) = 0.510, SE= 0.080). Again in 2002-03 (Dec 1, 2002 -May 31, 2003), the overwinter survival rate of fawns was significantly greater (x 21 = 5.734, P = 0.017) in the treatment unit (S(t) = 0.900, SE= 0.047) than in the control unit (S(t) = 0.691, SE= 0.074) (Fig. 4). The treatment unit during winter 2001-02 became the control unit during winter 2002-03, and vice versa. Thus, the overwinter survival treatment effect was replicated across each experimental unit. Combining both years of data, the best model of overwinter fawn survival (AICc = 148.63) included treatment (x\ = 14.71, P < 0.001), early winter fawn mass (x\ = 16.80, P < 0.001), year (x21 = 3.53, P = 0.060), and sex (x2i = 1.99, P = 0.158). The AIC model selection analysis emphasizes the importance of both the treatment effect as well as early winter mass of fawns, because any models without treatment or fawn mass were very poor (Table 1). Survival of fawns receiving the nutrition enhancement treatment was 0.31 higher than survival of control fawns during two mild to average winters, and surviving fawns averaged 2.9 kg heavier than fawns that died. Early winter mass was not different among experimental units (F1 = 0.35, P = 0.558), thus the effect of the treatment was not confounded with fawn mass. Fawn mass was similar between winters as well (F1 = 0.45, P = 0.502). The importance of early winter fawn mass as a predictor of overwinter survival has been documented previously (White et al. 1987, Bishop 1998, White and Bartmann 1998, Unsworth et al. 1999). In summary, the nutrition enhancement treatment improved overwinter fawn survival and thus yearling recruitment, and heavier fawns in each experimental unit had higher survival probabilities. l. •••••, f'" b"'""·•t,:'-«•,.< _.,.•.,,. ~«<: «<~· .,- ««<❖ L ,_.... ·----· ....__. ·---~ <««• ................ .. - - - - - - - - - - - - - - - - - ..,, 0.9 ....... ............. •-: ····t:.:;·· ............. ·.•_ .............................................................................. =: ..............................,, l ;:,;,,;.;.;.;.;.;.;,;,;.;,;,;-.~~ ll..8 • 1~~==-~ .___ _ •••.... ···················v·:;·· ....................................((.........: ..,...:-::0:o:":·:. 0.7 - • • Treatment (2001'"02) -Control (200J.-.02) - " Tttmtrnent. (2002--03) 0.5 · .................. . ()Ji ~··············· .... wmm= Contrul {l002-.03) +--~-~--~-~-~-~~-~-~-~--~-~-__,., 0.4 .U/l l~/1.(i 12.i'.H l.115 l/JO 2l1.4 3,1. ~Vlo ,v..,1 4/lS 4,30 S/15 :5/..~0 Figure 4. Overwinter fawn survival (Dec 1- May 31) in a nutrition enhancement treatment wut (S(t) = 0.865, SE= 0.056, 2001-02; S(t) = 0.900, SE= 0.047, 2002-03) and a control w1.it (S(t) = 0.510, SE= 0.080, 2001-02; S(t) = 0.691, SE= 0.074, 2002-03), Uncompahgre Plateau, southwest Colorado. �51 # Ii Param eters Model Name -2 Log Likelih od (K) AIC AICc C Treatment + Sex + Year + Mass 143.231 5 153.231 148.626 0 Treatment + Year + Mass Treatment + Sex + Year + Trt*Year + Mass Treatment + Sex + Mass 145.286 4 153.286 149.548 0.92 143.059 6 155.059 149.615 0.99 146.898 4 154.898 151.159 2.53 Treatment + Mass 148.957 3 154.957 152.113 3.49 Sex + Year+ Mass 160.345 4 168.345 164.606 15.98 Treatment 165.845 2 169.845 167.922 19.30 Sex +Year 178.195 3 184.195 181.351 32.73 AIC Table 1. Model selection results for a logistic regression analysis of overwinter mule deer fawn survival in southwest Colorado. Enhanced nutrition (freatment) and early winter fawn mass were the critical predictors of survival. Model selection was performed using Akaike's Information Criterion (AIC). During winter 2001-02, five fawns in the treatment unit died: 2 from malnutrition/sickness and 3 from disease. Of the 2 fawn mortalities caused by malnutrition/sickness, I was a result of basic malnutrition and occurred on December 31, 2001, shortly after the treatment was initiated. The other fawn died early as well and had a combination of heavy parasite loads, scours, and general poor condition. Each of the 3 fawns that died from disease had adequate fat stores. At least one of these fawns died as a result of pneumonia. In the control unit, 19 fawns died during the winter: 5 from malnutrition, 6 from mountain lion/bobcat predation, 4 from coyote/canine predation, 3 unknown predation mortalities, and 1 unknown. A majority of the fawns killed by predators had virtually no femur marrow fat remaining, indicating the predation was likely compensatory in nature. During winter 2002-03, where the initial control unit became the treatment following the cross-over, four fawns died in the treatment unit: 3 from coyote predation and 1 unknown mortality. In the control unit, 12 fawns died during the winter: 4 from coyote predation, 2 from malnutrition, 1 from mountain lion predation, 1 was road-killed, and 4 causes were unknown. As in the previous winter, these fawns had virtually no femur marrow fat remaining, indicating very poor condition. Adult Female Survival and Causes of Mortality During winter 2000-01 (Dec 1, 2000 - May 31, 2001 ), the adult doe survival rate in the treatment unit (S(t) = 0.968, SE= 0.032) was greater (x\ = 2.649, P = 0.104) than the survival rate in the control unit (S(t) = 0.861, SE= 0.058). However, annual adult doe survival rates (Dec 1, 2000 - Nov 30, 2001) were similar among the treatment and control deer (Trt: S(t) = 0.839, SE= 0.066; Control: S(t) = 0.833, SE= 0.062; 1 = 0.004, P = 0.947). We observed a similar result the following year. The 2001-02 overwinter adult doe survival rate in the treatment unit (S(t) = 0.942, SE= 0.030) was greater (x 21 = 3.116, P = 0.078) than survival in the control unit (S(t) = 0.848, SE= 0.044), yet annual adult doe survival was similar among treatment and control deer (Trt: S(t) = 0.824, SE= 0.049; Control: S(t) = 0.818, SE= 0.047; x\ = 0.090, P = 0.764). Thus, mortalities of control deer occurred primarily during the winter months, while treatment does died primarily during the summer and fall months. x2 �52 During winter 2002-03, following the treatment cross-over, overwinter adult doe survival rates were similar among treatment and control deer (Trt: S(t) = 0.945, SE= 0.024; Control: S(t) = 0.924, SE= 0.028; x\ = 0.360, P = 0.549). The main difference from the previous 2 years was that overwinter survival of adult does in the Shavano experimental unit increased in 2002-03 upon receiving the treatment. Current annual adult doe survival rates (Dec 1, 2002 - Oct 7, 2003) are 0.888 (SE= 0.034) for treatment does and 0.835 (SE= 0.039) for control does. The treatment has apparently had a minimal impact on annual adult doe survival, and annual survival rates measured thus far align with expected survival based on other studies (Unsworth et al. 1999, B.E. Watkins, unpublished). During 2000-02, when the Colona experimental unit received the treatment and the Shavano experimental unit was the control, 16 treatment and 16 control does died. The 16 treatment does died from the following categories: 4 - road-killed, 3 - while giving birth, 3 - predation (undetermined predator), 2 non-predation unknown (intact carcasses with no evidence of predation or scavenging), 1 - disease (chronic arthritis), 1 - mountain lion predation, and 2 - unknown. Predation was not a major mortality factor for treatment does, and a majority of mortalities were independent of nutrition (does were in good condition). The 16 control doe mortalities included the following causes: 5 - mountain lion predation, 3 - malnutrition, 2 - non-predation unknown, 1 - road-killed, 1 - bear predation, 1 - injury (fence), 1 legal harvest, and 2 - unknown. Predation and malnutrition were the major mortality causes of control deer. Interestingly, during this 2-year period, we did not document any coyote predation on adult does. Thus far during 2003, with Shavano as the treatment and Colona as the control, there have been 9 treatment doe mortalities: 3 - coyote predation, 3 - disease/infection, 1 - road-killed, and 2 unknown. Two of the coyote mortalities, 2 of the disease mortalities, and the road-kill occurred on adult does in good condition. There have been 14 control doe mortalities thus far in 2003: 3 - coyote predation, 3 malnutrition/disease, 3 - non-predation unknown, 1 - mountain lion predation, 1 - road-kill, and 3 unknown. As we saw during 2000 - 2002, malnutrition and predation were the major mortality factors of control does. LITERATURE CITED Baker, D. L., and N. T. Hobbs. 1985. Emergency feeding of mule deer during winter: tests of a supplemental ration. Journal of Wildlife Management 49:934-942. Baker, D. L., D. E. Johnson, L. H. Carpenter, 0. C. Wallmo, and R. B. Gill. 1979. Energy requirements of mule deer fawns in winter. Journal of Wildlife Management 43:162-169. Baker, D. L., G. W. Stout, and M. W. Miller. 1998. A diet supplement for captive wild ruminants. Journal of Zoo and Wildlife Medicine 29: 150-156. Ballard, W. B, D. Lutz, T. W. Keegan, L. H. Carpenter, and J.C. deVos, Jr. 2001. Deer-predator relationships: a review of recent North American studies with emphasis on mule and black-tailed deer. Wildlife Society Bulletin 29:99-115. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10: 108-114. Bishop, C. J. 1998. Mule deer fawn mortality and habitat use, and the nutritional quality ofbitterbrush and cheatgrass in southwest Idaho. Thesis, University of Idaho, Moscow, Idaho, USA. �53 Bishop, C. J., D. J. Freddy, and G. C. White. 2002. Effects of enhanced nutrition of adult female mule deer on fetal and neonatal survival rates: a pilot study to address feasibility. Colorado Division of Wildlife, Wildlife Research Report, Federal Aid in Wildlife Restoration Project W-153-R, Job Final Report. Fort Collins, Colorado, USA. Burnham, K. P., and D.R. Anderson. 1998. Model selection and inference: a practical informationtheoretic approach. Springer-Verlag, New York, New York, USA. Connolly, G. E. 1981. Limiting factors and population regulation. Pages 245-285 in 0. C. Wallmo, editor. Mule and black-tailed deer of North America. University of Nebraska Press, Lincoln, Nebraska, USA. Cook, J. G., and R. C. Cook. 2002. An informal training guide to condition evaluation in elk and deer. National Council for Air and Stream Improvement, Unpublished Report. Cook, R. C. 2000. Studies of body condition and reproductive physiology in Rocky Mountain elk. Thesis, University ofldaho, Moscow, Idaho, USA. Cook, R. C., J. G. Cook, D. L. Murray, P. Zager, B. K. Johnson, and M. W. Gratson. 2001. Development of predictive models of nutritional condition for Rocky Mountain elk. Journal of Wildlife Management 65:973-987. Gill, R. B., T. D. I. Beck, C. J. Bishop, D. J. Freddy, N. T. Hobbs, R.H. Kahn, M. W. Miller, T. M. Pojar, and G. C. White. 1999. Declining mule deer populations in Colorado: reasons and responses. A report to the Colorado Legislature. Colorado Division of Wildlife, Denver, Colorado, USA. Holter, J. B., H. H. Hayes, and S. H. Smith. 1979. Protein requirement of yearling white-tailed deer. Journal of Wildlife Management 43:872-879. Kaplan, E. L., and P. Meier. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 53:457-481. Pojar, T. M. 2000. Investigating factors contributing to declining mule deer numbers. Colorado Division of Wildlife, Wildlife Research Report, Federal Aid in Wildlife Restoration Project W-153-R-13, Progress Report. Fort Collins, CO, USA. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Ramsey, C. W. 1968. A drop-net deer trap. Journal ofWildlife Management 32:187-190. SAS Institute. 1989a. SAS/STAT® user's guide, version 6, fourth edition. Volume 1. SAS Institute, Cary, North Carolina, USA. SAS Institute. 1989b. SAS/STAT® user's guide, version 6, fourth edition. Volume 2. SAS Institute, Cary, North Carolina, USA. SAS Institute. 1997. SAS/STAT® Software: Changes and Enhancements through Release 6.12. SAS Institute, Cary, North Carolina, USA. �54 Schmidt, R. L., W. H. Rutherford, and F. M. Bodenham. 1978. Colorado bighorn sheep-trapping techniques. Wildlife Society Bulletin 6: 159-163. Smith, S. H., J.B. Holter, H. H. Hayes, and H. Silver. 1975. Protein requirement of white-tailed deer fawns. Journal of Wildlife Management 39:582-589. Stephenson, T. R., V. C. Bleich, B. M. Pierce, and G. P. Mulcahy. 2002. Validation of mule deer body composition using in vivo and post-mortem indices of nutritional condition. Wildlife Society Bulletin 30:557-564. Stephenson, T. R., K. J. Hundertmark, C. G. Schwartz, and V. Van Ballenberghe. 1998. Predicting body fat and body mass in moose with ultrasonography. Canadian Journal of Zoology 76:717-722. Stephenson, T. R., J. W. Testa, G. P. Adams, R G. Sasser, C. G. Schwartz, and K. J. Hundertmark. 1995. Diagnosis of pregnancy and twinning in moose by ultrasonography and serum assay. Akes 31:167-172. Thompson, C. B., J.B. Holter, H. H. Hayes, H. Silver, and W. E. Urban, Jr. 1973. Nutrition of whitetailed deer. I. Energy requirements of fawns. Journal of Wildlife Management 37:301-311. Ullrey, D. E., W. G. Youatt, H. E. Johnson, L. D. Fay, and B. L. Bradley. 1967. Protein requirement of white-tailed deer fawns. Journal of Wildlife Management 31 :679-685. Unsworth, J. W., D. F. Pac, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, Wisconsin, USA. Verme, L. J., and D. E. Ullrey. 1972. Feeding and nutrition of deer. Pages 275-291 in D. C. Church, editor. Digestive physiology and nutrition of ruminants. Volume 3 - Practical Nutrition. D. C. Church, Corvallis, Oregon, USA. Watkins, B. E., D. E. Ullrey, R. F. Nachreiner, and S. M. Schmitt. 1983. Effects of supplemental iodine and season on thyroid activity of white-tailed deer. Journal of Wildlife Management 47:45-58. Watkins, B. E., J. H. Witham, D. E. Ullrey, D. J. Watkins, and J.M. Jones. 1991. Body composition and condition evaluation of white-tailed deer fawns. Journal ofWildlife Management 55:39-51. White, G. C., and R. M. Bartmann. 1998. Effect of density reduction on overwinter survival of freeranging mule deer fawns. Journal of Wildlife Management 62:214-225. White, G. C., R. A. Garrott, R. M. Bartmann, L. H. Carpenter, and A. W. Alldredge. 1987. Survival of mule deer in northwest Colorado. Journal of Wildlife Management 51:852-859. White, G. C., A. F. Reeve, F. G. Lindzey, and K. P. Burnham. 1996. Estimation of mule deer winter mortality from age ratios. Journal of Wildlife Management 60:37-44. �Colorado Division of Wildlife Wildlife Research Report July 2003 – June 2004 JOB PROGRESS REPORT State of Project No. Work Package No. Task No. : : : Colorado 3001 : 4 Federal Aid Project: W-185-R Cost Center 3430 Mammals Research Deer Conservation Effect of Nutrition and Habitat Enhancements on Mule Deer Recruitment and Survival Rates : Period Covered: July 1, 2003 - June 30, 2004 Authors: C. J. Bishop, G. C. White, D. J. Freddy, and B. E. Watkins Personnel: D. L. Baker, L. Baeten, T. Banulis, E. J. Bergman, S. K. Carroll, D. Coven, K. Crane, M. DelTonto, B. Diamond, B. deVergie, D. Gallegos, J. Garner, L. Gepfert, R. B. Gill, D. Hale, J. Grigg, H. Halbritter, R. Harthan, M. Johnston, W. J. Lassiter, T. Mathieson, J. McMillan, G. C. Miller, M. W. Miller, J. Nicholson, J. A. Padia, T. M. Pojar, R. Powers, J. E. Risher, C. A. Schroeder, W. G. Sinner, C. M. Solohub, J. Thayer, M. A. Thonhoff, R. Wertsbaugh, L. Wolfe, CDOW; H. VanCampen, CSU; D. Felix, Olathe Spray Service; T. R. Stephenson, California Fish and Game; L. H. Carpenter, WMI; J. Sazma, B. Welch, BLM. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT To further understand the factors that caused deer numbers to decline in western Colorado during the 1990s, we designed and initiated a field experiment to measure deer population parameters in response to a nutrition enhancement treatment. During November 2000 – June 2004, we captured and radio-collared 810 individual mule deer evenly distributed among treatment and control units on the Uncompahgre Plateau in southwest Colorado. This included 293 adult females, 154 of which received vaginal implant transmitters (VITs), 241 6-month old fawns, and 276 newborn fawns born from either treatment or control adult does. We enhanced the nutrition of deer in the treatment unit by providing a safe, pelleted supplemental feed on a daily basis from December through April each winter. Early winter fawn:doe ratios were measured using helicopter and ground classification surveys the year following treatment delivery to determine whether fawn production and survival increased as a result of enhanced nutrition of adult females. We also measured overwinter fawn survival rates in response to the treatment. During 2002 – 2004, we measured pregnancy rates, fetus rates, and body condition of treatment and control adult does during late winter using ultrasonography. We also directly measured fetus survival and neonate survival by using VITs to help locate and radio-collar newborn fawns born from treatment and control does. Estimated percent body fat of adult does during late February and early March, 2002-04, 21 �was significantly higher (F1, 148 = 153.41, P < 0.001) for treatment deer (9.8%, SE = 0.36, n = 78) than control deer (4.3%, SE = 0.26, n = 76). Serum thyroid hormone concentrations, measured only in 2003 and 2004, were higher in treatment does than control does (F4, 108 = 46.59, P < 0.001). Pregnancy and fetus rates were similar among treatment and control does. The pregnancy rate of adult does was 0.95 (SE = 0.036, n = 38) and the fetus rate was 1.80 fetuses/doe (SE = 0.10, n = 36) during 2002. Rates were similar in 2003, where we measured a pregnancy rate of 0.92 (SE = 0.034, n = 63) and a fetus rate of 1.74 fetuses/doe (SE = 0.069, n = 50) which included 5 yearlings (the fetus rate excluding yearlings was 1.82 fetuses/doe, SE = 0.066, n = 45). In 2004, we measured a pregnancy rate of 0.94 (SE = 0.029, n = 66) and the fetus rate was 1.97 fetuses/doe (SE = 0.053, n = 60), which included 4 yearlings (the fetus rate excluding yearlings was 2.00 fetuses/doe, SE = 0.051, n = 56). The fetus survival rate with treatment and control fetuses combined was 0.86 (SE = 0.073) during 2002, 0.97 (SE = 0.024) during 2003, and 0.90 (SE = 0.040) during 2004. Fetus survival was similar among treatment and control deer during 2002 – 2003, but not 2004, where treatment fetus survival was 1.00 (SE = 0.000, n = 33) and control fetus survival was 0.76 (SE = 0.085, n = 25). Based on multiple early winter age classification surveys, we concluded the winter nutrition enhancement treatment did not cause an increase in neonatal production and survival during 2001. However, fawn production and summer-fall survival was relatively good for the overall population, and not representative of most years during the past decade when the population declined. During June – December, 2002−2003, survival of newborn treatment fawns was 0.620 (SE = 0.067) and control fawn survival was 0.493 (SE = 0.070). Survival data coupled with early winter age classification surveys provided evidence the nutrition enhancement treatment increased December fawn recruitment during 2002 and 2003. During December – June, 2001−2004, the overwinter survival rate of fawns was significantly greater (χ21 = 18.781, P < 0.001) in the treatment unit (S(t) = 0.895, SE = 0.029) than in the control unit (S(t) = 0.655, SE = 0.044). Because of a cross-over experimental design, the treatment unit during winter 2001-02 became the control unit during winters 2002-04, and vice versa. Thus, the treatment effect was replicated across each experimental unit. Combining all years of data, the best model of overwinter fawn survival (AICc = 207.65) included the nutrition enhancement treatment (χ21 = 19.04, P < 0.001), early winter fawn mass (χ21 = 23.27, P < 0.001), and year (χ21 = 6.20, P = 0.045). The AIC model selection analysis emphasized the importance of both the treatment effect as well as early winter mass of fawns, because any models without treatment or fawn mass were poor. Early winter mass of control fawns was slightly higher than that of treatment fawns (F1, 231 = 3.00, P = 0.085); thus the effect of the treatment was not confounded with fawn mass. Data collection will not be completed until January 2005. The results reported here are preliminary and should be treated as such. 22 �JOB PROGRESS REPORT EFFECT OF NUTRITION AND HABITAT ENHANCEMENTS ON MULE DEER RECRUITMENT AND SURVIVAL RATES C. J. Bishop, G. C. White, D. J. Freddy, and B. E. Watkins P. N. OBJECTIVES 1. To determine experimentally whether enhancing mule deer nutrition during winter and early spring by supplemental feeding increases fetus survival, neonate survival, early winter fawn:doe ratios or overwinter fawn survival. 2. To determine experimentally to what extent habitat treatments replicate the effect of enhanced nutrition from supplemental feeding. SEGMENT OBJECTIVES 1. Capture and radio-collar a target sample of adult female mule deer and 6 month-old fawns during late November through mid-December in a treatment unit and a control unit. 2. Capture a target sample of adult female mule deer in the treatment unit and the control unit to measure pregnancy rates, fetal rates, and body condition during late February to early March, and fit each adult female deer with a radio collar and vaginal implant transmitter. 3. Deliver the nutrition enhancement treatment to all deer occupying the treatment unit from early December through the end of April. 4. Capture and radio-collar a target sample of newborn fawns from treatment and control radio-collared does during June using the vaginal implant transmitters as a technique to determine the timing and location of birth. 5. Measure fetus survival, neonate survival, early winter fawn:doe ratios, overwinter fawn survival, and annual adult female survival based on radio-collared deer from the treatment and control units. INTRODUCTION Mule deer (Odocoileus hemionus) numbers apparently declined during the 1990’s throughout much of the West, and have clearly decreased since the peak population levels documented in the 1940’s60’s (Unsworth et al. 1999, Gill et al. 2001). Biologists and sportsmen alike have concerns as to what factors may be responsible for declining population trends. Although previous and current research indicates multiple interacting factors are responsible, habitat and predation have received the focus of attention. A number of studies have evaluated whether predator control increases deer survival, yet results are highly variable (Connolly 1981, Ballard et al. 2001). Together, predator control studies with adequate rigor indicate predation effects on mule deer are variable as a result of time-specific and sitespecific factors. Studies which have demonstrated deer population responses to predator control treatments have failed to determine whether predation is ultimately more limiting than habitat. Numerous research studies have evaluated mule deer habitat quality, but virtually no studies have documented population responses to habitat improvements. In many areas where declining deer numbers are of concern, predation is common yet habitat quality appears to have declined. The question remains as to whether predation, habitat, or some other factor is more limiting to mule deer in these situations, and whether habitat quality can be improved for the benefit of deer. It may also be that no single factor is any more or less important than others, and a more comprehensive understanding of multi-factor interactions is needed. We designed a field experiment to measure deer population responses to nutrition enhancement treatments, to further understand the causative factors underlying observed deer population dynamics. We 23 �are conducting the study on the Uncompahgre Plateau in southwest Colorado, where several predator species are present in abundant numbers: coyotes (Canis latrans), mountain lions (Felis concolor), and bears (Ursus americanus). In addition to predation, myriad diseases in combination proximately affect survival of the Uncompahgre deer population (Pojar and Bowden 2004, B.E. Watkins, unpublished data). Predator numbers have not and will not be manipulated in any manner during the course of the study. All factors have been left constant with the exception of deer nutrition. Deer nutrition is being enhanced by providing supplemental feed to deer occupying a treatment area during the winter. If December fawn recruitment and/or overwinter fawn survival improve as a direct result of the nutrition enhancement treatment, then we can presume that deer nutrition is ultimately more limiting than predation or disease. The second phase of the field experiment, which has not yet been initiated, will incorporate habitat manipulation treatments. The treatments will consist of prescribed fire or mechanical techniques to set back succession of pinyon-juniper (Pinus edulis-Juniperus osteosperma) habitat in an effort to improve the vigor and quality of winter habitat for mule deer. Deer population responses will be measured in relation to the habitat manipulations in the same manner as the supplemental feed. Thus, the experiment will evaluate whether nutritional quality of winter range habitat is ultimately more limiting than other factors in a late-seral pinyon-juniper/sagebrush (Artemisia spp.) landscape, and if so, whether habitat can be effectively improved for mule deer. The results will also advance our current understanding of multifactor interactions, with direct implications for mule deer management. MATERIALS AND METHODS Experimental Approach Experimental Design and Study Area We non-randomly selected two areas within mule deer winter range on the Uncompahgre Plateau to create 2 experimental units (A-B) (Fig. 1). The following criteria were used to select experimental units: 1.) Deer densities (~50-80 deer/mi2): areas were selected where deer densities were sufficient to meet sample size requirements within the experimental unit, while simultaneously selecting areas that would require feeding less than ~500-600 animals during a normal winter 2.) Buffer zones: areas were selected such that experimental units would be separated by several miles of non-treatment area (buffers) to prevent deer from occupying more than one experimental unit 3.) Similarity: areas were selected that comprise relatively similar habitat complexes and deer densities that are representative of the overall area 4.) Elk populations: areas were selected to minimize the number of elk present during normal winters Units A and B are receiving the nutrition enhancement treatment in a cross-over experimental design, and are being used to address P.N. Objective 1. Unit A served as the treatment unit, while Unit B served as the control, for the first 2 winters of research (2000 – 2002). Beginning November 2002, Unit B received the treatment while Unit A served as the control. Upon completion of P.N. Objective 1, two additional winter range experimental units will be used to conduct phase 2 of the research, or P.N. Objective 2. Habitat in one unit will be manipulated to set back plant succession (treatment), while habitat in the other unit will remain unchanged (control) throughout the experiment. 24 �Year 2000-01 Unit A Treatment Unit B Control 2001-02 Control 2002-03 Treatment Control Treatment 2003-04 Control Treatment Figure 1. Schematic representation of experimental units and nutrition enhancement treatment allocation. Units A and B are located in winter range habitat on the Uncompahgre Plateau in southwest Colorado. The nutrition enhancement cross-over design will encompass 4 years. The 2 experimental units (A and B) receiving the nutrition enhancement treatment are (Figs. 2 and 3): (1) Experimental unit A includes the Colona Tract of the Billy Creek State Wildlife Area and adjacent land, located approximately 13 km south of Montrose, CO adjacent to U.S. Hwy 550 South. The experimental unit is located within the Colona USGS 7.5 Minute Quadrangle, and roughly includes the polygon defined by the following Zone 13 UTM coordinates: (1) 254000 E, 4250200 N; (2) 252700 E, 4249400 N; (3) 254700 E, 4245600 N; and (4) 256200 E, 4246600 N. (2) Experimental unit B includes Shavano Valley and adjacent land extending west to the Dry Creek Rim. Shavano Valley is located approximately 13 km west of Montrose, CO. The experimental unit is located within the Dry Creek Basin and Montrose West Quadrangles (USGS 7.5 Minute), and roughly includes the polygon defined by the following Zone 13 UTM coordinates: (1) 238400 E, 4262600 N; (2) 232400 E, 4256700 N; (3) 235000 E, 4253600 N; and (4) 239500 E, 4258200 N. In late April and May, prior to fawning, deer from the winter range experimental units migrate to summer range. The summer range study area is defined by movements of the radio-collared deer, which encompass >1000 mi2 covering the southern portion of the Uncompahgre Plateau and adjacent San Juan Mountains to the south and east (Fig. 2). The summer range study area extends north to the Dry Creek river drainage on the Uncompaghre Plateau, south to Mineral Creek near Silverton, CO, east to the Big Blue river drainage, and west to the San Miguel River canyon. However, a majority of the radio-collared deer summer on the Uncompahgre Plateau between Dry Creek to the north and Highway 62 to the south. Winter range elevations range from 1830 m (6000 ft) in Shavano Valley to 2318 m (7600 ft) adjacent to the Dry Creek Rim above Shavano Valley. Winter range habitat is dominated by pinyonjuniper with interspersed sagebrush adjacent to agricultural fields in the Shavano and Uncompahgre Valleys. Summer range elevations occupied by deer range from 1891 m (6200 ft) in the Uncompahgre Valley to 3538 m (11,600 ft) in Imogene Basin southwest of Ouray, CO. Summer range habitats are dominated by spruce-fir (Picea spp.-Abies spp.), aspen (Populus tremuloides), ponderosa pine (Pinus ponderosa), Gambel oak (Quercus gambelii), and to a lesser extent, sagebrush and pinyon-juniper at lower elevations. 25 �Uncompahgre Plateau Mesa County GRAND JUNCTION Delta County U om nc GMU 62 r hg pa e u ea at Pl Montrose County GMU 61 DELTA Shavano E.U. Gunnison County Winter Range Exp. Units MONTROSE Colona Montrose County E.U. Summer Range Ouray County Sanmiguel County Figure 2. Location of Colona and Shavano (Units A and B) experimental units in Game Management Unit 62 on the Uncompahgre Plateau, southwest Colorado; and location of the summer range study area throughout the southern Uncompahgre Plateau and adjacent San Juan Mountains. 26 �Hwy 550 Uncompahgre Valley Colona Exp. Unit Shavano Valley Shavano Exp. Unit Figure 3. Colona and Shavano experimental units (Units A and B), located in Game Management Unit 62 on the Uncompahgre Plateau, southwest Colorado. 27 �Response Variables The response variables are fetal and neonatal survival rates, early winter fawn:doe ratios, and overwinter fawn survival rates. The nutrition enhancement treatment is delivered to deer from December through April, fetus survival is assessed during June, neonate survival is measured from June to December, and fawn:doe ratios are measured during the following December and January (1 year after the treatment was initiated). Overwinter fawn survival is measured from December to June as a direct result of the current winter’s treatment. We are measuring these response variables in each experimental unit (treatment and control) to determine whether enhanced winter nutrition of adult does increases subsequent newborn fawn production and survival, and whether enhanced winter nutrition of 6-mo. old fawns directly increases overwinter fawn survival. Ultimately, these measurements provide an assessment of the effect of winter range habitat quality on yearling recruitment, and thus population productivity. We are also measuring overwinter and annual survival of adult does as a function of enhanced winter nutrition. Sample Size Fetus/Neonate Survival: We were primarily interested in survival of newborn fawns from radiocollared does that occupy the 2 winter range experimental units. Fetus survival is also important, but difficult to measure. Fetus rates from a sample of radio-collared does can be measured in winter, but the fate of all fetuses cannot be determined the following June because of logistical constraints. Fetus survival rates can only be measured from some unpredictable fraction of the radio-collared doe sample, making sample size calculations of limited use. Thus, our sample size calculations were based on quantifying neonate survival, not fetus survival. For neonate survival, a sample size of 40 neonates per experimental unit per year provides power of 0.81 to detect a difference of 0.15 in survival between 2 experimental units if survival among control fawns is 0.40. We assumed a control survival rate of 0.40 based on neonate survival rates measured recently for the Uncompahgre deer population (Pojar and Bowden 2004) in combination with December fawn:doe ratios measured during the late 1980’s and 1990’s, when the Uncompahgre population declined (B. E. Watkins, unpublished data). Based on Bishop et al. (2002), we determined that 60 radio-collared does (30 treatment and 30 control) equipped with vaginal implant transmitters (VITs) would be necessary to capture a minimum of 80 newborn fawns. We also assumed that some fawns would be captured from other treatment and control radio-collared does not equipped with VITs. The 60 radio-collared does with VITs are also being used to evaluate fetus survival; however, logistical constraints limit the power of fetus survival comparisons among experimental units. Early winter fawn:doe ratios: We desired to detect an effect size, i.e., an increase in fawn:doe ratios in response to the treatments, in the range of 15 to 20 fawns per 100 does. These values were based on simple population models with overwinter fawn survival of 0.444, adult female survival of 0.853, and December fawn:doe ratios of 66 fawns per 100 does to obtain a stationary population (Unsworth et al. 1999). Based on surveys of the Uncompahgre deer population during the 1990’s, the standard deviation of the fawn:doe ratio for groups with at least one adult female was 57, with a mean of 41. Using an expected standard deviation of 57, the standard error of the mean fawn:doe ratio for 40 radio-collared does is 57/(401/2) = 9.0, which is the expected standard deviation of measured fawn:doe ratios on each experimental unit. We assessed power using a two-sample t-test with a sample size of 4, representing the 4 years of the study where fawn:doe ratios are being measured in response to enhanced nutrition. Our power to detect an increase of 20 fawns per 100 does based on classification of 40 radio-collared doe groups in each experimental unit is about 0.87. A sample size of 40 fawns per experimental unit per year provides a power of 0.81 to detect a difference of 0.15 in survival between 2 experimental units if survival on the control unit is 0.40. We expected to see an increase in fawn survival (effect size) of approximately 0.15, because this was the difference measured in the density reduction experiment conducted by White and Bartmann (1998). 28 �Adult and 6-month Old Fawn Capture During November and December, adult does and 6-month old fawns were captured using baited drop nets (Ramsey 1968, Schmidt et al. 1978) and helicopter net guns (Barrett et al. 1982, van Reenen 1982). Drop nets were baited with certified weed-free alfalfa hay and apple pulp. Drop nets were used as the principle capture technique for a 3-4 week capture period; helicopter net-gunning was then used at the end of the drop-net capture to secure the remainder of deer needed to meet our target sample sizes. All deer were hobbled and blind-folded after being captured. Deer captured via drop nets were carried away from the net to an adjacent handling site using stretchers. Deer were fitted with leather radio collars equipped with mortality sensors, which cause an increase in pulse rate after remaining motionless for 4 hours. Permanent collars were placed on adult females, while temporary collars were placed on fawns. To make collars temporary, one end of the collar was cut in half and reattached using rubber surgical tubing; fawns shed the collars ≥6 months post-capture. A rectangular piece of flexible plastic (Ritchey® neck band material) engraved with a unique identifier was stitched to the side of each collar. The unique identifier consisted of 2 symbols for adult females, and 1 symbol on 2 different colors of plastic for fawns. The identifiers were necessary to visually identify deer from the ground. This allowed us to effectively document use of the treatment, measure fawn:doe ratios from the ground, and assess experimental unit population size via mark-resight estimators. We recorded the weight, hind foot length and chest girth of each deer, and collected blood samples to evaluate disease prevalence. During late February and early March, an additional 30 adult female deer were captured in each experimental unit by net-gunning. Captured deer were ferried by the helicopter to a central processing location, where deer were carried by stretchers to a tent for handling. For each captured deer, we used ultrasonography to measure pregnancy status, fetal rate, and body condition. Only pregnant does were retained and radio-collared. We then inserted a vaginal implant transmitter (VIT) in each doe as a technique for locating the timing and location of her birth site the following June. We also recorded the weight, hind foot length and chest girth of each deer, and collected blood samples to evaluate disease prevalence. Body Condition and Reproductive Status We estimated body fat of treatment and control adult does during mid-late winter using an Aloka 210 (Aloka, Inc., Wallinford, Conn.) or SonoVet 2000 (Universal Medical Systems, Bedford Hills, NY) portable ultrasound unit with a 5 MHz linear transducer. We measured maximum subcutaneous fat thickness on the rump (MAXFAT) following the methodology of Stephenson et al. (1998, 2002). We also measured thickness of the longissimus dorsi muscle via ultrasound (Cook et al. 2001, Stephenson et al. 2002). A small area of hair was shaved to ensure contact between the transducer and the skin. Vegetable oil was applied to the shaved area for conduction purposes and fat/muscle thickness was measured using electronic calipers. We coupled the ultrasound measurements with body condition scores (BCS) obtained from palpation of the ribs, withers, and rump (Cook 2000). MAXFAT and rump BCS measurements were combined into a condition index used to estimate percent body fat (Cook and Cook 2002): % Fat = -6.6387617 + 7.4271417x – 1.11579443x2 + 0.07733803x3 where x = rLIVINDEX = (MAXFAT – 0.15) + rump BCS (if MAXFAT < 0.15, then rLIVINDEX = rump BCS). The rLIVINDEX and body fat regression was initially developed and validated for elk by Cook et al. (2001), and then modified by incorporating a validation of MAXFAT for mule deer performed by Stephenson et al. (2002). During mid-late winter, we also evaluated differences in serum thyroid hormone concentrations between treatment and control adult does. Specifically, we measured total thyroxine (T4), free T4 (FT4), total tri-iodothyronine (T3), and free T3 (FT3) following the methodologies of Watkins et al. (1983, 1991). Blood samples were collected at the time of capture, and serum hormone analyses were performed by the Michigan State University Animal Health Diagnostic Laboratory (East Lansing, Michigan). We 29 �compared serum thyroid hormone concentrations between treatment and control adult does, and also compared hormone levels to body fat estimates derived from the ultrasonography. We quantified reproductive status (Stephenson et al. 1995, Andelt et al. 2004) with ultrasound via transabdominal scanning using a 3 MHz linear transducer. We searched for fetuses by scanning a portion of the abdomen that was shaved caudal to the last rib and left of the midline. We systematically searched each uterine horn to identify fetal numbers ranging from 0 to 3. Whenever possible, we measured eye diameter of each fetus to approximately estimate fetal age and parturition date. Vaginal Implant Transmitters (VITs) We used VITs manufactured by Advanced Telemetry Systems, Inc. (Isanti, MN). The VIT was 76 mm long, excluding antenna length, and had 2 plastic wings with a width of 57 mm when fully spread apart. The plastic wings were used to retain the transmitter in the vagina until parturition. The VIT weighed 15 grams and contained a 10-28 lithium battery programmed to a 12-hour on/off cycle. The diameter of the transmitter/battery was 14 mm, and was encased in an impermeable, water-proof, electrical resin. The transmitter contained an embedded heat-sensor which dictated the frequency pulse rate. When the heat sensor dropped below 90°F, synonymous with transmitter expulsion from the deer, the pulse rate changed from 40 PPM to 80 PPM. VIT batteries were programmed to be active from 0430 to 1630 hrs prior to daylight savings, and thus were active from 0530 to 1730 hrs after daylight savings and during the fawning period. The VIT was inserted into deer using a vaginoscope (Jorgensen Laboratories, Inc., Loveland, CO) and alligator forceps. The vaginoscope was 6” long with a 5/8” internal diameter and had a machined end (smooth surface) to minimize trauma when inserted into the vagina. A discreet mark was placed on the applicator showing the appropriate distance it should be inserted into the deer. The length of a typical mule deer vaginal tract was obtained by taking measurements from road-killed deer and/or other fresh deer carcasses obtained in the study area. Prior to use in the field, VITs were sterilized using a Chlorhexidine solution, air-dried, and sealed in a 3” x 8” sterilization pouch. Sterilization containers with Chlorhexidine solution were used on site during capture to sterilize the vaginoscope and alligator forceps between each use. A new pair of nitrile surgical gloves was used to handle the vaginoscope and VIT for each deer. To insert a VIT, the plastic wings were folded together and placed into the end of the vaginoscope. We then liberally applied sterile KY Jelly to the scope and inserted it into the deer’s vagina to the point where the mark on the applicator was reached. The alligator forceps, which extended through the vaginoscope to hold the VIT, was held firmly in place while the scope was pulled out from the vagina. This procedure pushed the VIT out of the scope into the vagina, and the plastic wings spread apart to hold the transmitter in place. The transmitter antenna was typically flush with the vulva, but on occasion extended up to 1 cm beyond the vulva. The tip of the antenna was encapsulated in a wax bead to protect the deer. Neonate Fawn Capture During June we relocated each of the radio-collared does having a VIT each morning using aerial and ground telemetry. Flights began at 0530 hr and were usually completed by 1000 – 1100 hrs. The early flights were crucial for detecting fast signals because shed VITs could exceed 90 °F by mid-day if shed in the open, which caused them to switch back to a slow (“pre-birth”) pulse. When a fast (“postpartum”) pulse rate was detected, we located the VIT from the ground to determine whether it was shed at the birth site. If the transmitter was located at the birth site, we identified whether any fawn(s) were stillborn. If the fawn(s) were no longer present at the birth site, or could not be found in the vicinity of the birth site, we located the radio-collared doe and searched for fawns at her location. All personnel involved wore surgical gloves to help minimize human scent when handling fawns. For each doe, we attempted to locate each of her fawns and document whether any fawns were stillborn. We attempted to account for each doe’s fetuses in order to quantify in utero fetal survival from February to birth. We placed a drop-off radio-collar on each live fawn; radio collars were constructed with elastic neck-band 30 �material to facilitate expansion. Hole-punched, leather tabs extended from the end of the elastic and from the transmitter for attachment purposes. Collars were made temporary by cutting the leather tab extending from the elastic and reattaching the leather with latex tubing, which caused the collars to shed from the animal >6 months post-capture. For each fawn, mass and hind foot length were recorded, and a nasal swab sample was collected to screen for Bovine Viral Diarrhea. We then recorded basic vegetation characteristics of the birth site and promptly exited the site. We also routinely located and attempted to capture fawns from treatment and control radiocollared does not having VITs to help achieve our targeted sample size. Each of these does had been previously captured during the research, and were present on either the treatment or control experimental unit during winter. Measurement of Survival Rates and Fawn:Doe Ratios We measured survival rates by radio-monitoring collared deer from the ground and air to determine fate (live/mortality). We also attempted to determine the cause of each mortality, with a primary goal of distinguishing between predation and non-predation mortality causes. Deer were radiomonitored from the ground on a daily basis throughout the year and from the air on approximately a biweekly basis. We were able to detect signals from nearly all radio-collared deer each day during winter, which typically allowed us to arrive at mortality sites within 24 hours of the mortality event. During summer and migration periods, deer were distributed widely and thus were more difficult to radiomonitor. All radio-collared neonates were checked daily throughout the summer and fall, whereas some adult and yearling deer could not be ground-monitored on a routine basis. In result, we typically located neonate mortalities within 24 hours of death, but some adult deer mortalities were not detected for several days, or on rare occasion, for one or more weeks. Fresh, intact neonate carcasses were collected and submitted to the Colorado Division of Wildlife’s Wildlife Health Laboratory or the Colorado State University Diagnostic Laboratory for necropsy and tissue analyses. Fresh, intact adult and 6-month old fawn carcasses were also submitted for laboratory necropsy when feasible. Field necropsies were performed on all other deer mortalities, and when appropriate, tissue samples were collected and submitted for analysis. Each winter we used the radio-collared does to measure fawn:doe ratios in each experimental unit. The resulting fawn:doe ratio is a measurement of the previous year’s treatment effect. We measured fawn:doe ratios using 2 techniques: (1) We located the sample of radio-collared does in each experimental unit from a fixed-wing airplane, and used the set of locations to define boundaries for the experimental unit. Shortly after (i.e. 1-2 days), we used a helicopter to systematically fly the defined unit and classify all deer groups encountered. For each group, we documented whether a radio-collared doe was present. (2) We located each radio-collared doe by radio telemetry from the ground. The group of deer with the collared doe was counted and classified by age and sex. Both methods were employed to gather as much information as possible to determine whether there was a treatment effect. The “true” value cannot be measured perfectly because of the inherent biases and potential sources of error associated with each technique. Thus, by employing both techniques, we had a greater chance of fully understanding whether the treatment caused an effect. Treatment Delivery Deer nutrition was enhanced in the treatment area by providing a safe, pelleted supplemental feed. The supplemental feed was developed through extensive testing with both captive and wild deer (Baker and Hobbs 1985, Baker et al. 1998), and has been safely used in both applied research and management projects. Pellets were distributed daily using 4wd pickup trucks, ATVs, and snowmobiles on primitive roads throughout the experimental unit to provide a food source for the entire deer population in the treatment unit. Each 50lb. bag of pellets was carried ≤200m from the vehicle and distributed by hand in approximately 20-30 small piles of feed in a linear fashion. Numerous bags were 31 �distributed in successive order allowing us to create linear lines of feed that spanned most of the treatment area, which prevented animals from concentrating in any single location. This feeding technique also prevented dominant animals from restricting access to the food supply because of the large area over which pellets were distributed. We supplied pellets ad libitum where a small residual remained when the next day’s ration was provided. Collared deer were closely monitored to ensure that treatment deer remained in the experimental unit and actually consumed the feed, and to make sure that non-treatment deer remained in the control unit, which they did. The few treatment adult does that moved away from the treatment unit were withdrawn from the sample for purposes of measuring treatment effects. However, to avoid any biases, all 6-month old fawns captured in the treatment unit were included in survival analyses regardless of whether they accessed the supplement or not. This was because some fawns died shortly after capture (e.g. 2-3 weeks), before we could document whether they had access to the feed. Also, very few fawns that survived more than 2-3 weeks moved away from the treatment unit. The pelleted ration was commercially produced in the form of 2×1×0.5-cm wafers (Baker and Hobbs 1985). Feed constituents (i.e. digestibility, protein, gross energy etc.) vastly exceeded those of typical winter range deer diets; exact constituent values are provided by Baker et al. (1998). When provided ad libitum, the feed should have allowed deer to meet or exceed nutritional requirements for growth and maintenance (Ullrey et al. 1967, Verme and Ullrey 1972, Thompson et al. 1973, Smith et al. 1975, Baker et al. 1979, Holter et al. 1979). The basis for feeding such high quality pellets was to ensure that the treatment (enhanced nutrition) was effectively delivered to the deer. Our intent was not to determine the exact level of nutrition necessary to increase fawn recruitment, but rather to determine if nutrition is a limiting factor to recruitment. If nutrition is in fact limiting, we will rely on habitat manipulation treatments to evaluate what exactly can be done via management to increase fawn survival and recruitment. Statistical Methods A preliminary fawn:doe ratio analysis was completed using PROC MIXED in SAS (SAS Institute 1997). We used a reduced model with experimental unit as the independent variable; we considered experimental unit as a fixed effect and radio-collared does within an experimental unit as random effects. Survival rates were calculated using a Kaplan-Meier survival analysis (Kaplan and Meier 1958, Pollock et al. 1989), and contrasted among experimental units and sexes using a chi-square analysis. For neonate survival analyses, we used a common entry date because a staggered entry would have biased survival rates low due to early mortalities that occurred before most of the sample was captured. We modeled overwinter fawn survival with a logistic regression model using PROC LOGISTIC in SAS (SAS Institute 1989a); model selection was performed using Akaike’s Information Criterion (AIC) (Burnham and Anderson 1998). Survival was modeled as a function of the nutrition enhancement treatment, sex, year, and capture mass. We used a general linear model in PROC GLM in SAS (SAS Institute 1989b) to test for differences in estimated percent body fat between treatment and control adult does and a multivariate model to test for differences in T4, FT4, T3, and FT3 thryoid hormones between treatment and control does. We then used PROG REG (SAS Institute 1989b) to evaluate the relationship between estimated percent body fat and serum thyroid hormone concentrations. We analyzed fetus survival directly with a binomial survival rate for the subset of fetuses with known fates. We also indirectly analyzed fetus survival by comparing the February fetus rate with the number of live newborn fawns/doe observed in June using a change-in-ratio estimator (White et al. 1996). Other results in this report are presented as data summaries incorporating means and standard errors, or in some cases, raw data values. These results are incomplete and preliminary, and should be treated as such. 32 �RESULTS AND DISCUSSION Deer Capture During November and December 2000-2003, we captured and radio-collared 139 adult female mule deer evenly distributed among the treatment and control units. We also captured and radio-collared 241 6-month-old fawns during November and December 2001-2003 (40 fawns/unit/year). Due to budgeting constraints, we were unable to radio-collar 6-month old fawns during 2000. We captured an additional 154 adult females during late February and early March 2002-2004 and equipped them with radio collars and VITs. During June 2002-2004, we captured and radio-collared 276 newborn fawns from radio-collared adult females. Thus, the following results are based upon radio-monitoring of 810 individual mule deer evenly distributed among treatment and control units during November 2000-June 2004. Treatment Delivery 2000-01 From December 15, 2000, through April 19, 2001, we distributed 88 tons of the pelleted ration. For most of the winter and spring, on average, we distributed 0.85 tons of feed each day throughout 22 feeding sites across the 2.3 mi2 treatment unit. Deer were fed ad libitum because there was always residual feed remaining the next day during the feeding routine. Each sack was distributed in approximately 20-30 distinct, small piles, resulting in >1000 small piles of feed throughout the treatment unit. This effort allowed deer to effectively access the feed in small groups, and no aggression was ever observed among deer seeking access to the feed. By distributing the feed in this manner, we were able to avoid the negative aspects associated with large-scale feeding operations. Deer adapted to the pelleted supplement right away and utilized it extensively throughout the winter. We continually monitored deer use of the feed from ground observation points, where we obtained 440 visual observations of radiocollared does consuming the feed. These observations, coupled with daily radio-monitoring and periodic aerial relocations, indicate 32 of the 37 radio-collared treatment does spent the entire winter and spring within the boundaries of the treatment unit and received the supplement on a daily basis. Mark-resight population estimates from March helicopter (489 deer, SE = 62) and ground (494 deer, SE = 81) surveys, coupled with feed consumption, indicate we fed roughly 450 to 500 deer during most of the winter and spring. Feed consumption declined coincident with spring green-up, although deer continued to use the feed through mid-late April, at which point they began migrating to summer range. We also fed approximately 25 to 30 elk, but the elk did not affect deer access to the feed. Deer in the control experimental unit did not receive feed or any other treatment. Based on helicopter mark-resight surveys, the deer density in the treatment unit in December was 120 deer/mi2 (SE = 9), but increased shortly after and was 213 deer/mi2 (SE = 27) in March. Deer densities in the control unit changed little from 83 deer/mi2 (SE = 12) in December to 101 deer/mi2 (SE = 14) in March. 2001-02 From December 15, 2001, through April 25, 2002, we distributed 194 tons of the supplement throughout the treatment unit. For most of the winter and spring, we distributed 2.0-2.1 tons of feed each day. The dramatic increase in supplement distribution from the previous year occurred because a large number of elk descended into the Uncompahgre Valley during mid-late fall/early winter. Elk arrived in unusually large numbers throughout much of the valley prior to the onset of treatment delivery. Once feeding was initiated, approximately 300-500 elk adapted to the feed and remained in or around the 2.3 mi2 treatment unit throughout most of the winter. Given myriad logistical and budgetary constraints, 2.1 tons was the maximum amount of feed we could routinely deliver on a daily basis. Feed was not delivered ad libitum to all deer and elk in the treatment unit throughout the winter because residual feed was rarely observed during the next day’s 33 �distribution. However, daily field observations indicated most deer approached ad libitum consumption of the supplement. In contrast to the previous winter, deer were waiting for the daily supplement to arrive each morning. Deer then consumed the supplement immediately after it was distributed. Elk were rarely observed utilizing the feed until late morning or afternoon, and elk continued to forage in fields below the treatment unit, whereas deer did not. We observed numerous radio-collared deer consuming the pelleted supplement each day; not all of these observations were recorded because of time constraints with distributing the feed. Given this time limitation, we still recorded 818 observations of radio-collared deer consuming the supplemental feed (497 collared doe observations and 321 collared fawn observations). Most days, >100 and sometimes 200-300 deer were observed utilizing the pellets during the course of distributing the supplement. These observations rarely included elk; thus, deer-elk competition was minimized because of temporal differences in feeding, and deer clearly had first access to the feed. 2002-03 Beginning December 2002, we switched the treatment and control units consistent with the crossover experimental design. From December 15, 2002, through April 30, 2003, we distributed 97 tons of the supplement throughout the new treatment unit, which had served as the control unit the previous 2 years. The supplement was distributed daily throughout 29 sites over a larger area (~7 mi2) than the first 2 years of research because of the greater size of the experimental unit and broader distribution of radiocollared deer. Residual feed was always present throughout the winter, thus deer were fed ad libitum. Only small groups of elk periodically accessed the supplement, and did not affect deer access. We obtained 286 observations of radio-collared deer consuming the supplement, which were difficult to obtain because the supplement was spread out over a large area and only a single feed site could be observed at any given moment. We also used daily ground radio-monitoring and periodic aerial relocations to document deer access to the supplement. 2003-04 From December 10, 2003, through April 30, 2004, we distributed 197 tons of the supplement throughout the treatment unit. The increase in supplement distribution occurred because of an increase in elk on the upper portion of the experimental unit. However, unlike winter 2001-02, residual feed was present throughout the winter and deer were fed ad libitum. By targeting a portion of the daily feed distribution to elk, we restricted elk to the upper extent of the deer winter range for most of the winter. Thus, elk had a minimal affect on deer access to the supplement. We obtained 413 observations of radiocollared deer consuming the supplement. As before, we also used daily ground radio-monitoring and periodic aerial relocations to document deer access to the supplement. Body Condition Estimated percent body fat of adult does during late February and early March, 2002–2004, was significantly higher for treatment deer than control deer (F1, 148 = 153.41, P < 0.001). Over all years combined, mean predicted body fat was 9.8% (SE = 0.36) for treatment adult does and 4.3% (SE = 0.26) for control does. The interaction of experimental unit × year for predicted body fat was also significant (F2, 148 = 14.39, P < 0.001). This interaction occurred because the difference in body fat between treatment and control deer was greater during 2003 than during 2002 or 2004. Mean predicted body fat was 8.2% (SE = 0.92) for treatment adult does and 5.0% (SE = 0.71) for control does during 2002, and 9.0% (SE = 0.53) for treatment does and 4.7% (SE = 0.36) for control does during 2004. The difference was greater during 2003, where mean predicted body fat was 11.7% (SE = 0.35) for treatment does and 3.4% (SE = 0.35) for control does. The body fat estimates reported here should accurately reflect deer, but may be further refined in the future as additional research provides more data on the relationship between body condition indices and estimated percent body fat. Serum thyroid hormone concentrations, measured during 2003 and 2004, were higher in treatment does than control does (F4, 108 = 46.59, P < 0.001) (Table 1). Hormone concentrations also 34 �varied between years (F4, 108 = 14.21, P < 0.001), but the experimental unit × year interaction was not significant (F4, 108 = 1.68, P = 0.160). Thus, each year thyroid hormone concentrations were higher in treatment does than control does. T4 was the most important thyroid hormone in describing the canonical variable for differences between treatment and control does (1.04*T4 − 0.02*T3 + 0.77*FT4 – 0.17*FT3). As expected, there was a high partial correlation between T4 and FT4 (r = 0.67, P < 0.001) and between T3 and FT3 (r = 0.60, P < 0.001), which has been documented previously (Watkins et al. 1983). When treated as 4 separate ANOVAs, T4 (F1, 111 = 165.97, P < 0.001), FT4 (F1, 111 = 144.37, P < 0.001), T3 (F1, 111 = 13.84, P < 0.001), and FT3 (F1, 111 = 8.26, P = 0.005) were significantly higher in treatment does than control does. Given these results, we evaluated the relationship between T4 concentrations and estimated percent body fat (derived from ultrasound and BCS indices) using a simple linear regression model (% Fat = −3.122 + 0.090*T4, r2 = 0.52, P < 0.001). Similar correlations between T4 and actual percent body fat during mid-late winter have been previously documented for white-tailed deer and elk (Watkins et al. 1991, Cook et al. 2001). Table 1. Total thyroxine (T4) and total tri-iodothyronine (T3) concentrations (nmol/l), and free T4 (FT4) and free T3 (FT3) concentrations (pmol/l), measured during late February in adult female mule deer occupying a nutrition enhancement treatment unit and a control unit on the Uncompahgre Plateau in southwest Colorado, 2003-04. Thyroid Hormone T3 (SE) FT3 (SE) 146.6 (3.53) FT4 (SE) 30.0 (1.27) 1.65 (0.058) 4.10 (0.130) Control 92.3 (3.56) 17.1 (0.65) 1.42 (0.080) 3.71 (0.210) Treatment 131.9 (4.48) 24.8 (1.39) 2.08 (0.075) 4.21 (0.154) Control 90.0 (3.54) 12.5 (0.59) 1.70 (0.104) 3.60 (0.188) Year Exp. Unit T4 (SE) 2003 Treatment 2004 Fetus Survival and Pregnancy/Fetus Rates We began measuring fetus survival in 2002 as part of our effort to capture and radio-collar newborn fawns born from radio-collared does. Similar numbers of stillborns were observed between treatment and control does during both 2002 and 2003, so fetus survival estimates for those years are not differentiated by experimental unit. In February-March 2002, 36 of 38 adult does captured were pregnant, thus the pregnancy rate was 0.95 (SE = 0.036). We measured an average of 1.80 fetuses/doe (SE = 0.10, n = 36), which included 1.77 fetuses/doe (SE = 0.14, n = 18) in the treatment unit and 1.83 fetuses/doe (SE = 0.15, n = 18) in the control unit. During June 2002, we determined the fate of all fetuses (live or stillborn) from only 14 of the 36 VIT does, largely because of a high VIT battery failure rate. The survival rate of fetuses (n = 22) from these 14 does was 0.86 (SE = 0.073). We also assessed fetus survival using a change-in-ratio estimator between the fetal rate measured in February-March and the observed number of live fawns/doe postpartum in June. In June 2002, considering all does (n = 43) that we located any fawn from, whether live or stillborn, we observed 1.42 (SE = 0.11) live fawns/doe postpartum. This rate should represent a conservative estimate of live fawns/doe postpartum because we inevitably failed to locate all live fawns from each doe. In other words, this estimate would treat any unaccounted fetuses (from the February measurement) as if they were stillborns. For radio-collared does that did not have VITs, and thus we did not have a winter fetus rate measurement, singletons would infer that either the deer only had 1 fetus, or that the other fetus died. It is likely that some of these singletons had a twin that we did not locate. This equates to a conservative fetus survival rate estimate of 0.79 (SE = 0.18). 35 �In February-March 2003, 58 of 63 adult does captured were pregnant, resulting in a pregnancy rate of 0.92 (SE = 0.034). Critical personnel and equipment for measuring fetus rates were not continuously available due to capture delays associated with helicopter mechanical problems. Some of the deer fetus counts were performed by inexperienced observers without optimum ultrasound equipment. VITs worked very well, though, allowing us to determine fetus numbers at parturition for many of the deer. Thus, we determined winter fetus rates by using the greatest fetus count for each individual deer, whether obtained using ultrasound during February-March or by locating newborn fawns and stillborns at birthsites during June. We were unable to determine a fetus count for 8 treatment deer because only pregnancy was established with ultrasound and no birthsite assessments were possible in June. These 8 deer were removed from the fetus rate estimates. Of the 50 deer where a fetus count was obtained, 5 were yearlings (2 treatment yearlings, 3 control yearlings). We measured 1.74 fetuses/doe (SE = 0.069, n = 50) overall including yearlings, and 1.82 fetuses/doe (SE = 0.066, n = 45) excluding yearlings. Fetus rates with yearlings included were 1.77 fetuses/doe (SE = 0.091, n = 22) in the treatment unit and 1.70 fetuses/doe (SE = 0.10, n = 28) in the control unit. During June 2003, we determined the fate of all fetuses (live or stillborn) from 33 of the 58 VIT does; the good success was based on VITs commonly being shed at birthsites. The survival rate of fetuses (n = 58) from these 33 does was 0.97 (SE = 0.024). In June 2003, incorporating all does (n = 71) that we located any fawn from, whether live or stillborn, we observed 1.49 (SE = 0.072) live fawns/doe postpartum. Using the change-in-ratio estimator described above, this results in an overall conservative fetus survival rate estimate of 0.86 (SE = 0.15). In February 2004, the overall pregnancy rate was 0.94 (SE = 0.029, n = 66) and the fetus rate was 1.97 fetuses/doe (SE = 0.053, n = 60), which included 4 yearlings. Excluding yearlings, the fetus rate was 2.00 fetuses/doe (SE = 0.051, n = 56). Fetus rates were 1.90 fetuses/doe (SE = 0.074, n = 30) in the treatment unit and 2.03 fetuses/doe (SE = 0.076, n = 30) in the control unit with yearlings included, and 1.93 (SE = 0.069, n = 29) in the treatment unit and 2.07 (SE = 0.074, n = 27) in the control unit with yearlings excluded. We determined the fate of all fetuses (live or stillborn) from 31 of the 60 VIT does. The overall fetus survival rate was 0.90 (SE = 0.040, n = 58). Different from 2002 or 2003, each of these stillborns were from control does. The survival rate of control fetuses was 0.76 (SE = 0.085, n = 25) as compared to a survival rate of 1.00 (n = 33) for treatment fetuses. Using data from all does (n = 82) in which we located any fawn, the conservative change-in-ratio fetus survival estimate was 0.79 (SE = 0.13) overall, 0.88 (SE = 0.17) for treatment deer, and 0.69 (SE = 0.14) for control deer. Neonatal Survival/Fawn:Doe Ratios 2001 In December 2000, at the beginning of the study and prior to the first year’s treatment delivery, fawn:doe ratios were similar in the 2 experimental units. Pre-treatment fawn:doe ratios were 52.6 fawns:100 does (SE = 5.3) in the treatment unit, and 51.6 fawns:100 does (SE = 5.0) in the control unit. In late December 2001 and early January 2002, following the first year’s treatment, we conducted 2 age classification helicopter surveys in the treatment and control units. On 12/23/01, we observed 52.8 fawns:100 does (SE = 6.7) in the treatment unit, and 36.7 fawns:100 does (SE = 3.8) in the control unit. On 1/8/02, we observed 54.7 fawns:100 does (SE = 6.6) in the treatment unit, and 50.5 fawns:100 does (SE = 6.0) in the control unit. During December 2001 – February 2002, we obtained fawn:doe ratio estimates from ground observations of radio-collared deer groups for both treatment and control deer. This survey resulted in 61.2 fawns:100 does (SE = 7.8) in the treatment unit, and 74.5 fawns:100 does (SE = 8.5) in the control unit, although the result was not statistically significant (t74 = 1.16, P = 0.249). The fawn:doe ratio results are conflicting, and clearly do not provide evidence that there was any treatment effect. In short, we concluded that the nutrition enhancement treatment did not cause an increase in neonatal production and survival during 2001. However, our results, in conjunction with a December estimate of 64 fawns:100 does for the entire Uncompahgre deer population (B.E. Watkins, 36 �unpublished), indicate fawn production and survival was good during 2001. The observed fawn:doe ratios coupled with overwinter fawn survival and annual adult survival rates indicate the deer population was increasing. Considering the past 1-2 decades, this was an atypically good year for the Uncompahgre deer population. 2002 During June – December 2002, following the second year’s treatment, we measured neonate survival directly using radio-collared fawns; however, sample sizes were based on a technique assessment of VITs and were relatively small for contrasting treatment and control survival of neonates (Bishop et al. 2002). Treatment fawn survival was 0.613 (SE = 0.115, n = 29) and control fawn survival was 0.511 (SE = 0.108, n = 25). In late December 2002 and early January 2003, we once again conducted 2 age classification helicopter surveys in the treatment and control units. On 12/31/02, we observed 91.9 fawns:100 does (SE = 8.4) in the treatment unit, and 52.2 fawns:100 does (SE = 6.9) in the control unit. On 1/21/03, we observed 52.6 fawns:100 does (SE = 6.4) in the treatment unit, and 36.8 fawns:100 does (SE = 3.9) in the control unit. The combined helicopter survey data indicated 68.1 fawns:100 does (SE = 5.6) in the treatment unit and 42.8 fawns:100 does (SE = 3.5) in the control unit. Oppositely, fawn:doe ratio estimates from ground classifications of doe groups during December 2002 – February 2003 were 47.7 fawns:100 does (SE = 6.3) in the treatment unit, and 63.4 fawns:100 does (SE = 7.5) in the control unit (t108 = 1.61, P = 0.110). As in 2001, fawn:doe ratio results were conflicting. Helicopter survey data varied between 2 different flights, but consistently indicated a treatment effect. Ground classification data did not indicate a treatment effect. Also, survival data combined with age ratio data indicate neonate production and survival was reasonably favorable during 2002, and not indicative of the low fawn recruitment observed during the late 1980’s and 1990’s. 2003 During June 2003, we captured and radio-collared 103 newborn fawns born from treatment and control radio-collared does (55 treatment fawns, 48 control fawns). The VITs worked well; we captured fawns from 41 of the 54 does fitted with VITs. Treatment fawn survival (June – Dec) was 0.624 (SE = 0.082) and control fawn survival was 0.483 (SE = 0.093). Final standard errors were larger than expected because a number of fawns shed collars prematurely when crossing fences during fall migration. Using helicopter surveys, we measured 62.4 fawns:100 does (SE = 5.3) in the treatment unit and 50.0 fawns:100 does (SE = 4.9) in the control unit. Estimates from ground classifications of doe groups were 68.0 fawns:100 does (SE = 7.6) in the treatment unit and 62.1 fawns:100 does (SE = 7.6) in the control unit. Age ratio estimates from the helicopter and the ground were more consistent during 2003 than in past years. Overall, observed fawn:doe ratios were consistent with treatment and control fawn survival rates measured from June to December. 2002-03 Survival rate point estimates were very similar during 2002 and 2003. Combined over both years, treatment survival (S(t) = 0.620, SE = 0.067) was higher (P = 0.189) than control survival (S(t) = 0.493, SE = 0.070). The high censor rate due to shed collars during fall affected the p-value. Neonate survival through July 15, 2002 and 2003, was significantly higher (P = 0.006) for treatment fawns (S(t) = 0.833, SE = 0.041) than control fawns (S(t) = 0.634, SE = 0.057). We are currently measuring 2004 neonate survival rates, which will be necessary for final interpretations as to the effectiveness of the treatment. Our results from 2001 and 2002 emphasize the inherent difficulties and biases associated with precisely measuring fawn:doe ratios, particularly in this research study. Ratios obtained from helicopter surveys were based on 2 short-duration flights/unit/year over spatially small units. Helicopter surveys were complicated by high deer densities in heavy cover, making both deer detection and fawn:doe classifications a considerable challenge. There is a variety of potential biases that may have affected the 37 �helicopter surveys, including differential sightability of does and fawns, double classification of some deer, and incorrectly classifying yearling bucks with small antlers. Ground fawn:doe ratio observations of radio-collared doe groups were made using spotting scopes and field glasses, where we commonly studied the deer for some time. Incorrect classifications during these surveys were likely minimal. For example, small-antlered yearling bucks (e.g. 3 – 6” spikes) were detected from the ground, whereas they were undoubtedly missed on occasion during helicopter surveys. We also obtained repeated observations for some of the radio-collared doe groups from the ground. The main potential bias affecting ground fawn:doe classifications was how observations were made. Many of the ground classifications in the Shavano Valley experimental unit were made by radio-tracking does during the day. On the other hand, a majority of ground classifications in the Colona experimental unit were based on observing deer groups as they entered openings to feed during the late afternoon. Our age ratio results were more consistent during 2003. Deer were not as concentrated during helicopter surveys, and unlike previous years, almost all of the ground classification data for the Colona experimental unit was obtained by radio-tracking does during the day. Given the inherent difficulties of measuring fawn:doe ratios in the 2 experimental units, and the lack of a clear indication as to the effectiveness of the treatment, we will only cautiously use fawn:doe ratios to make inferences regarding treatment effects. At the completion of the research, we will test whether enhanced winter nutrition of adult does improved newborn fawn survival based on a three-year model of the radio-collared neonate survival data. Neonate Mortality Causes During June − December of 2002 and 2003, 32 of 84 treatment fawns died: 8 – coyote predation, 2 – bear predation, 2 – felid predation, 3 – predation where the predator was undetermined, 9 – disease/starvation/ malnutrition, 1 – abandonment, 2 – trauma/injury, 1 – road-kill, 2 – unknown, and 2 – poached. The two poached fawns were censored from analyses evaluating the effect of the treatment. Converted to mortality rates based on the Kaplan-Meier survival analysis, 11.4% of all treatment fawns died from disease/starvation/malnutrition, 10.1% from coyote predation, 3.8% from predation where the predator was undetermined, 2.5% each from bear predation, felid predation, injury/trauma, and unknown causes, and 1.3% each from abandonment and road-kill. Simplified, 18.9% of all treatment fawns died from predation, 11.4% died from disease/starvation/malnutrition, and 7.6% died from other or unknown causes. During June – December of 2002 and 2003, 35 of 72 control fawns died: 12 – coyote predation, 4 – felid predation, 2 – bear predation, 1 – predation where the predator was undetermined, 11 – disease/starvation/ malnutrition, 1 – trauma/injury, and 4 – unknown. Converted to mortality rates based on the Kaplan-Meier survival analysis, 17.4% of all control fawns died from coyote predation, 15.9% died from disease/starvation/malnutrition, 5.8% each from felid predation and unknown causes, 2.9% from bear predation, and 1.4% each from trauma/injury and predation where the predator was undetermined. Simplified, 27.5% of all control fawns died from predation, 15.9% from disease/starvation/malnutrition, and 7.2% from other or unknown causes. In summary, mortality rates due to predation and disease/starvation/malnutrition were lower for treatment fawns than control fawns. Overwinter Fawn Survival and Mortality Causes During winter 2001-02 (Dec 10, 2001 – June 15, 2002), the survival rate of fawns was significantly greater (χ21 = 13.216, P < 0.001) in the treatment unit (S(t) = 0.865, SE = 0.056) than in the control unit (S(t) = 0.510, SE = 0.080). Similarly, in 2002-03 (Dec 10, 2002 – June 15, 2003), the overwinter survival rate of fawns was significantly greater (χ21 = 5.734, P = 0.017) in the treatment unit (S(t) = 0.900, SE = 0.047) than in the control unit (S(t) = 0.691, SE = 0.074). Again in 2003-04 (Dec 10, 2003 – June 15, 2004), the overwinter survival rate of fawns was significantly greater (χ21 = 3.852, P = 0.050) in the treatment unit (S(t) = 0.920, SE = 0.045) than in the control unit (S(t) = 0.756, SE = 0.067). Combining survival data across all 3 winters, treatment fawn survival (S(t) = 0.895, SE = 0.029) was 38 �higher (χ21 = 18.781, P < 0.001) than control fawn survival (S(t) = 0.655, SE = 0.044) (Fig. 4). The treatment unit during winter 2001-02 became the control unit during winters 2002-03 and 2003-04, and vice versa. Thus, the overwinter survival treatment effect was replicated across each experimental unit. Combining all years of data, the best model of overwinter fawn survival (AICc = 207.65) included treatment (χ21 = 19.04, P < 0.001), early winter fawn mass (χ21 = 23.27, P < 0.001), and year (χ21 = 6.20, P = 0.045). The AIC model selection analysis emphasizes the importance of both the treatment effect as well as early winter mass of fawns, because any models without treatment or fawn mass were very poor (Table 2). Survival of fawns receiving the nutrition enhancement treatment was 0.24 higher than survival of control fawns during three mild to average winters, and surviving fawns averaged 3.5 kg heavier than fawns that died. Early winter mass of control fawns was slightly higher than that of treatment fawns (F1, 231 = 3.00, P = 0.085); thus the effect of the treatment was not confounded with fawn mass. Fawn mass was similar between winters as well (F2, 231 = 1.31, P = 0.273). The importance of early winter fawn mass as a predictor of overwinter survival has been documented previously (White et al. 1987, Bishop 1998, White and Bartmann 1998, Unsworth et al. 1999). In summary, the nutrition enhancement treatment improved overwinter fawn survival and thus yearling recruitment, and heavier fawns in each experimental unit had higher survival probabilities. 1 ~ -, - -. -- ~ - 0.9 \ 0.8 7 '- - --- -. - ---- \., 'L...- - 0.7 - ------ - Treatment 0.6 - - -Control 0.5 ' 12/1 ' 12/16 12131 ' 1/15 ' 100 ' 2/14 ' ' ' ' 3/1 3/16 3/31 4115 ' 4/30 ' ' 5/15 51.30 Figure 4. Overwinter fawn survival (Dec 10 – June 15, 2001 – 2004) in a nutrition enhancement treatment unit (S(t) = 0.895, SE = 0.029) and a control unit (S(t) = 0.655, SE = 0.044) on the Uncompahgre Plateau, southwest Colorado. 39 ' 6/14 �Table 2. Model selection results for a logistic regression analysis of overwinter mule deer fawn survival in southwest Colorado, 2001− 2004. Enhanced nutrition (Treatment) and early winter fawn mass were the critical predictors of survival. Model selection was performed using Akaike’s Information Criterion (AIC). # -2 Log ∆ Param Likelih AIC eters c (K) AIC AICc Model Name od Treatment + Year + Mass Treatment + Sex + Year + Mass Treatment + Sex + Year + Trt*Year + Mass Treatment + Sex + Mass 202.254 5 212.254 207.649 0 201.227 6 213.227 207.783 0.13 201.060 8 217.060 210.026 2.38 207.179 4 215.179 211.440 3.79 Treatment + Mass 208.556 3 214.556 211.712 4.06 Sex + Year + Mass 223.598 5 233.598 228.993 21.34 Treatment 235.739 2 239.739 237.816 30.17 Sex + Year 248.878 4 256.878 253.139 45.49 During winters 2001-04, 12 of 115 treatment fawns died: 5 from coyote predation, 3 from disease/illness, 2 from malnutrition, 1 from trauma/injury, and 1 unknown. Each of the 3 fawns that died from disease had adequate fat stores. At least one of these fawns died as a result of pneumonia. Converted to mortality rates based on the Kaplan-Meier survival analysis, 4.3% of all treatment fawns died from coyote predation, 2.6% from disease/illness, 1.7% from malnutrition, 0.9% from trauma/injury, and 0.9% from unknown causes. Simplified, 4.3% of all treatment fawns died from predation, 4.3% from disease/malnutrition, and 1.8% from other or unknown causes. During winters 2001-04, 41 of 120 control fawns died: 13 from coyote predation, 8 from mountain lion predation, 8 from malnutrition, 6 from unknown causes, 3 from predation where the predator was undetermined, 2 were road-killed, and 1 from trauma/injury. Converted to mortality rates based on the Kaplan-Meier survival analysis, 10.9% of all control fawns died from coyote predation, 6.7% from mountain lion predation, 6.7% from malnutrition, 5.0% from unknown causes, 2.5% from predation where the predator was undetermined, 1.7% from road-kill, and 0.8% from trauma/injury. Simplified, 20.1% of all control fawns died from predation, 6.7% from malnutrition, and 7.5% from other or unknown causes. Most fawns killed by predators had little or no femur marrow fat remaining, indicating the predation was likely compensatory in nature. Adult Female Survival and Causes of Mortality During winter 2000-01 (Dec 1, 2000 – May 31, 2001), the adult doe survival rate in the treatment unit (S(t) = 0.968, SE = 0.032) was greater (χ21 = 2.649, P = 0.104) than the survival rate in the control unit (S(t) = 0.861, SE = 0.058). However, annual adult doe survival rates (Dec 1, 2000 – Nov 30, 2001) were similar among the treatment and control deer (Trt: S(t) = 0.839, SE = 0.066; Control: S(t) = 0.833, SE = 0.062; χ21 = 0.004, P = 0.947). We observed a similar result the following year. The 2001-02 overwinter adult doe survival rate in the treatment unit (S(t) = 0.942, SE = 0.030) was greater (χ21 = 3.116, P = 0.078) than survival in the control unit (S(t) = 0.848, SE = 0.044), yet annual adult doe survival was similar among treatment and control deer (Trt: S(t) = 0.824, SE = 0.049; Control: S(t) = 0.818, SE = 0.047; χ21 = 0.090, P = 0.764). Thus, mortalities of control deer occurred primarily during the winter months, while treatment does died primarily during the summer and fall months. 40 �During winter 2002-03, following the treatment cross-over, overwinter adult doe survival rates were similar among treatment and control deer (Trt: S(t) = 0.945, SE = 0.024; Control: S(t) = 0.924, SE = 0.028; χ21 = 0.360, P = 0.549). The main difference from the previous 2 years was that overwinter survival of adult does in the Shavano experimental unit increased in 2002-03 upon receiving the treatment. However, annual adult doe survival rates (Dec 1, 2002 – Nov 30, 2003) were higher (χ21 = 2.016, P = 0.156) for treatment does 0.888 (SE = 0.034) than control does 0.813 (SE = 0.041). The main difference from the previous 2 years was overwinter survival of adult does in the Shavano experimental unit increased in 2002-03 upon receiving the treatment. Summer-fall survival was similar in that Colona adult does had higher mortality rates than Shavano adult does. Thus, in 2002-03, there was no difference between survival rates of treatment and control adult does during winter but there was evidence of higher annual survival of treatment adult does. During winter 2003-04, overwinter adult doe survival rates were higher (χ21 = 3.843, P = 0.050) among treatment does (S(t) = 0.979, SE = 0.014) than control does (S(t) = 0.915, SE = 0.027). Thus far in 2004, annual adult doe survival rates (Dec 1, 2003 – 8/31, 2004) are 0.951 (SE = 0.021) for treatment does and 0.896 (SE = 0.029) for control does. Considering all years, the treatment has improved overwinter adult doe survival but had a relatively minor affect on annual survival. Considering only the past 2 years, there is evidence the treatment has had a positive affect on annual survival. Annual survival rates measured in this study align with expected survival based on other studies (Unsworth et al. 1999, B.E. Watkins, unpublished). During 2000-02, when the Colona experimental unit received the treatment and the Shavano experimental unit was the control, 16 treatment and 16 control does died. The 16 treatment does died from the following categories: 4 – road-killed, 3 – while giving birth, 3 – predation (undetermined predator), 2 – non-predation unknown (intact carcasses with no evidence of predation or scavenging), 1 – disease (chronic arthritis), 1 – mountain lion predation, and 2 – unknown. Predation was not a major mortality factor for treatment does, and a majority of mortalities were independent of nutrition (does were in good condition). The 16 control doe mortalities included the following causes: 5 – mountain lion predation, 3 – malnutrition, 2 – non-predation unknown, 1 – road-killed, 1 – bear predation, 1 – injury (fence), 1 – legal harvest, and 2 – unknown. Predation and malnutrition were the major mortality causes of control deer. Interestingly, during this 2-year period, we did not document any coyote predation on adult does. During Dec 2002 – August 2004, with Shavano as the treatment and Colona as the control, there have been 14 treatment doe mortalities: 5 – disease/infection, 5 unknown causes, 3 – coyote predation, and 1 – road-killed. As we saw during 2000-02, predation was not a major mortality factor for treatment does, and a majority of mortalities were independent of nutrition. There have been 26 control adult doe mortalities during this same time period: 7 – malnutrition/disease, 5 – road-kill, 4 – coyote predation, 4 – unknown causes, 3 – mountain lion predation, and 3 – non-predation unknown. Malnutrition, predation, and road-kill were the major mortality factors of control does during 2002-04. SUMMARY We successfully enhanced nutrition of deer occupying the treatment units. There was no evidence the treatment positively influenced fetus survival until 2004, when virtually all stillborn fetuses were from control adult does. We currently have evidence the treatment caused an increase in neonate survival; however, data collection is incomplete. The treatment caused a significant increase in overwinter fawn survival, which is where the greatest differences occurred between treatment and control deer. Overwinter adult doe survival increased as a result of the treatment, but annual survival was more similar among treatment and control adult does. Results reported here are based on preliminary analyses, and in some cases, incomplete data sets. Final analyses will be conducted once data collection is complete. 41 �LITERATURE CITED Andelt, W. F., T. M. Pojar, and L. W. Johnson. 2004. Long-term trends in mule deer pregnancy and fetal rates in Colorado. Journal of Wildlife Management 68:542-549. Baker, D. L., and N. T. Hobbs. 1985. Emergency feeding of mule deer during winter: tests of a supplemental ration. Journal of Wildlife Management 49:934-942. Baker, D. L., D. E. Johnson, L. H. Carpenter, O. C. Wallmo, and R. B. Gill. 1979. Energy requirements of mule deer fawns in winter. Journal of Wildlife Management 43:162-169. Baker, D. L., G. W. Stout, and M. W. Miller. 1998. A diet supplement for captive wild ruminants. Journal of Zoo and Wildlife Medicine 29:150-156. Ballard, W. B, D. Lutz, T. W. Keegan, L. H. Carpenter, and J. C. deVos, Jr. 2001. Deer-predator relationships: a review of recent North American studies with emphasis on mule and black-tailed deer. Wildlife Society Bulletin 29:99-115. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108-114. Bishop, C. J. 1998. Mule deer fawn mortality and habitat use, and the nutritional quality of bitterbrush and cheatgrass in southwest Idaho. Thesis, University of Idaho, Moscow, Idaho, USA. Bishop, C. J., D. J. Freddy, and G. C. White. 2002. Effects of enhanced nutrition of adult female mule deer on fetal and neonatal survival rates: a pilot study to address feasibility. Colorado Division of Wildlife, Wildlife Research Report, Federal Aid in Wildlife Restoration Project W-153-R, Job Final Report. Fort Collins, Colorado, USA. Burnham, K. P., and D. R. Anderson. 1998. Model selection and inference: a practical informationtheoretic approach. Springer-Verlag, New York, New York, USA. Connolly, G. E. 1981. Limiting factors and population regulation. Pages 245-285 in O. C. Wallmo, editor. Mule and black-tailed deer of North America. University of Nebraska Press, Lincoln, Nebraska, USA. Cook, R. C. 2000. Studies of body condition and reproductive physiology in Rocky Mountain elk. Thesis, University of Idaho, Moscow, Idaho, USA. Cook, R. C., and J. G. Cook. 2002. An informal training guide to condition evaluation in elk and deer. National Council for Air and Stream Improvement, Unpublished Report. Cook, R. C., J. G. Cook, D. L. Murray, P. Zager, B. K. Johnson, and M. W. Gratson. 2001. Development of predictive models of nutritional condition for Rocky Mountain elk. Journal of Wildlife Management 65:973-987. Gill, R. B., T. D. I. Beck, C. J. Bishop, D. J. Freddy, N. T. Hobbs, R. H. Kahn, M. W. Miller, T. M. Pojar, and G. C. White. 2001. Declining mule deer populations in Colorado: reasons and responses. Colorado Division of Wildlife Special Report Number 77. Denver, Colorado, USA. Holter, J. B., H. H. Hayes, and S. H. Smith. 1979. Protein requirement of yearling white-tailed deer. Journal of Wildlife Management 43:872-879. Kaplan, E. L., and P. Meier. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 53:457-481. Pojar, T. M., and D. C. Bowden. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550-560. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Ramsey, C. W. 1968. A drop-net deer trap. Journal of Wildlife Management 32:187-190. SAS Institute. 1989a. SAS/STAT® user’s guide, version 6, fourth edition. Volume 1. SAS Institute, Cary, North Carolina, USA. SAS Institute. 1989b. SAS/STAT® user’s guide, version 6, fourth edition. Volume 2. SAS Institute, Cary, North Carolina, USA. SAS Institute. 1997. SAS/STAT® Software: Changes and Enhancements through Release 6.12. SAS Institute, Cary, North Carolina, USA. 42 �Schmidt, R. L., W. H. Rutherford, and F. M. Bodenham. 1978. Colorado bighorn sheep- trapping techniques. Wildlife Society Bulletin 6:159-163. Smith, S. H., J. B. Holter, H. H. Hayes, and H. Silver. 1975. Protein requirement of white-tailed deer fawns. Journal of Wildlife Management 39:582-589. Stephenson, T. R., V. C. Bleich, B. M. Pierce, and G. P. Mulcahy. 2002. Validation of mule deer body composition using in vivo and post-mortem indices of nutritional condition. Wildlife Society Bulletin 30:557-564. Stephenson, T. R., K. J. Hundertmark, C. G. Schwartz, and V. Van Ballenberghe. 1998. Predicting body fat and body mass in moose with ultrasonography. Canadian Journal of Zoology 76:717-722. Stephenson, T. R., J. W. Testa, G. P. Adams, R. G. Sasser, C. G. Schwartz, and K. J. Hundertmark. 1995. Diagnosis of pregnancy and twinning in moose by ultrasonography and serum assay. Alces 31:167-172. Thompson, C. B., J. B. Holter, H. H. Hayes, H. Silver, and W. E. Urban, Jr. 1973. Nutrition of whitetailed deer. I. Energy requirements of fawns. Journal of Wildlife Management 37:301-311. Ullrey, D. E., W. G. Youatt, H. E. Johnson, L. D. Fay, and B. L. Bradley. 1967. Protein requirement of white-tailed deer fawns. Journal of Wildlife Management 31:679-685. Unsworth, J. W., D. F. Pac, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, Wisconsin, USA. Verme, L. J., and D. E. Ullrey. 1972. Feeding and nutrition of deer. Pages 275-291 in D. C. Church, editor. Digestive physiology and nutrition of ruminants. Volume 3 – Practical Nutrition. D. C. Church, Corvallis, Oregon, USA. Watkins, B. E., D. E. Ullrey, R. F. Nachreiner, and S. M. Schmitt. 1983. Effects of supplemental iodine and season on thyroid activity of white-tailed deer. Journal of Wildlife Management 47:45-58. Watkins, B. E., J. H. Witham, D. E. Ullrey, D. J. Watkins, and J. M. Jones. 1991. Body composition and condition evaluation of white-tailed deer fawns. Journal of Wildlife Management 55:39-51. White, G. C., and R. M. Bartmann. 1998. Effect of density reduction on overwinter survival of freeranging mule deer fawns. Journal of Wildlife Management 62:214-225. White, G. C., R. A. Garrott, R. M. Bartmann, L. H. Carpenter, and A. W. Alldredge. 1987. Survival of mule deer in northwest Colorado. Journal of Wildlife Management 51:852-859. White, G. C., A. F. Reeve, F. G. Lindzey, and K. P. Burnham. 1996. Estimation of mule deer winter mortality from age ratios. Journal of Wildlife Management 60:37-44. Prepared by _______________________ Chad J. Bishop, Wildlife Researcher 43 �Colorado Division of Wildlife July 2004 – June 2005 WILDLIFE RESEARCH REPORT State of Cost Center Work Package Task No. Colorado 3430 3001 4 : : : : Federal Aid Project: W-185-R : Division of Wildlife Mammals Research Deer Conservation Effect of Nutrition and Habitat Enhancements on Mule Deer Recruitment and Survival Rate Period Covered: July 1, 2004 - June 30, 2005 Authors: C. J. Bishop, G. C. White, D. J. Freddy, and B. E. Watkins Personnel: D. L. Baker, L. Baeten, T. M. Banulis, E. J. Bergman, S. K. Carroll, M. J. Catanese, D. L. Coven, K. Crane, M. L. DelTonto, B. Diamond, B. deVergie, P. Ehrlich, D. Gallegos, J. Garner, L. Gepfert, R. B. Gill, D. Hale, J. L. Grigg, H. J. Halbritter, R. Harthan, M. D. Johnston, W. J. Lassiter, C. T. Larsen, T. Mathieson, J. W. McMillan, G. C. Miller, M. W. Miller, J. D. Nicholson, J. A. Padia, T. M. Pojar, R. M. Powers, J. E. Risher, C. A. Schroeder, W. G. Sinner, C. M. Solohub, M. H. Swan, K. Taurman, J. A. Thayer, M. A. Thonhoff, C. E. Tucker, R. M. Wertsbaugh, L. L. Wolfe, CDOW; H. VanCampen, CSU; D. Felix, Olathe Spray Service; T. R. Stephenson, California Fish and Game; L. H. Carpenter, WMI; J. Sazma, B. Welch, BLM. Project support received from Federal Aid in Wildlife Restoration, Mule Deer Foundation, and Colorado Habitat Partnership Program. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We measured mule deer (Odocoileus hemionus) population parameters in response to a nutrition enhancement treatment to evaluate the relative importance of habitat quality as a limiting factor of mule deer in western Colorado. During November 2000 – June 2004, we captured and radio-collared 810 individual mule deer evenly distributed among treatment and control units on the Uncompahgre Plateau in southwest Colorado. Our sample included 293 adult females, 154 of which received vaginal implant transmitters (VITs), 241 6-month-old fawns, and 276 newborn fawns born from either treatment or control adult does. We enhanced the nutrition of deer in the treatment unit by providing a safe, pelleted supplemental feed on a daily basis from December through April each winter. The treatment unit during winters 2000−01 and 2001−02 became the control unit during winters 2002−03 and 2003−04, and vice versa. Thus, the treatment effect was replicated across each experimental unit. Early winter fawn:doe ratios were measured using helicopter and ground classification surveys the year following treatment delivery to determine whether fawn production and survival increased as a result of enhanced nutrition of adult females. During winters 2001–02 through 2003−04, we measured pregnancy rates, fetus rates, and body condition of treatment and control adult does using ultrasonography. We measured fetus survival and neonate survival by using VITs to help locate and radio-collar newborn fawns born from treatment 37 �and control does. We also measured overwinter fawn survival rates in response to the treatment. Estimated percent body fat of adult does during late February and early March, 2002−04, was higher (F1, 148 = 153.41, P < 0.001) for treatment deer (9.8%, SE = 0.36, n = 78) than control deer (4.3%, SE = 0.26, n = 76). Serum thyroid hormone concentrations, measured only in 2003 and 2004, were higher in treatment does than control does (F4, 108 = 46.59, P < 0.001). Pregnancy and fetus rates were similar among treatment and control does. The pregnancy rate of adult does was 0.95 (SE = 0.036, n = 38) and the fetus rate was 1.80 fetuses/doe (SE = 0.10, n = 36) during 2002. Rates were similar in 2003, where we measured a pregnancy rate of 0.92 (SE = 0.034, n = 63) and a fetus rate of 1.74 fetuses/doe (SE = 0.069, n = 50) which included 5 yearlings (the fetus rate excluding yearlings was 1.82 fetuses/doe, SE = 0.066, n = 45). In 2004, pregnancy rate was 0.94 (SE = 0.029, n = 66) and fetus rate was 1.97 fetuses/doe (SE = 0.053, n = 60), which included 4 yearlings (fetus rate excluding yearlings was 2.00 fetuses/doe, SE = 0.051, n = 56). Based on multiple early winter age classification surveys, we lacked evidence to determine whether the winter nutrition enhancement treatment had any effect on neonatal production and survival during 2001, which provided additional incentive to directly measure fetus and neonate survival. During 2002−2004, fetus-neonate survival from 1 March−15 December was higher (χ21 = 3.846, P = 0.050) for treatment fawns (S(t) = 0.528, SE = 0.027) than control fawns (S(t) = 0.401, SE = 0.025). Survival data coupled with early winter age classification surveys provided evidence the nutrition enhancement treatment increased December fawn recruitment during 2002−2004. During 15 December– 15 June, 2001−2004, the overwinter survival rate of fawns was greater (χ21 = 18.781, P < 0.001) in the treatment unit (S(t) = 0.895, SE = 0.029) than in the control unit (S(t) = 0.655, SE = 0.044). Using a staggered entry survival process with data combined over years, survival of treatment fetuses to 1 year of age (S(t) = 0.458, SE = 0.031) was 0.18 higher (χ21 = 13.20, P < 0.001) than survival of control fetuses to 1 year of age (S(t) = 0.276, SE = 0.026). The finite rate of population increase, λ, based on our measurements of treatment population parameters was 1.20, which would cause the deer population to double in approximately 4 years. The finite rate of increase calculated from control deer was 1.04, indicating a stable or slightly increasing population. The nutrition enhancement treatment therefore had a dramatic effect on deer population performance, indicating habitat quality was ultimately limiting the population. Our results provide a foundation for focusing deer management efforts on improving habitat quality in western Colorado pinyon-juniper (Pinus edulis-Juniperus osteosperma) ecosystems with corresponding research efforts to quantify the effects of habitat manipulations on deer performance. 38 �WILDLIFE RESEARCH REPORT EFFECT OF NUTRITION AND HABITAT ENHANCEMENTS ON MULE DEER RECRUITMENT AND SURVIVAL RATES CHAD J. BISHOP, GARY C. WHITE, DAVID J. FREDDY, AND BRUCE E. WATKINS P. N. OBJECTIVES 1. To determine experimentally whether enhancing mule deer nutrition during winter and early spring via supplementation increases fetus survival, neonate survival, overwinter fawn survival, or ultimately, population productivity. 2. To determine experimentally to what extent habitat treatments replicate the effect of enhanced nutrition from supplemental feeding. SEGMENT OBJECTIVES 1. Radio-monitor and measure survival of the sample of radio-collared mule deer adult does and fawns. 2. Measure early winter fawn:doe ratios using both aerial helicopter surveys and ground classifications of deer groups associated with radio-collared adult does. 3. Summarize and analyze data and publish information in an annual Job Progress Report. 4. Complete a peer-reviewed manuscript for publication in a scientific journal pertaining to the effectiveness of vaginal implant transmitters for capturing mule deer neonates exclusively from radiomarked adult does (Appendix I). INTRODUCTION Mule deer (Odocoileus hemionus) numbers apparently declined during the 1990s throughout much of the West, and have clearly decreased since the peak population levels documented during the 1940s−1960s (Unsworth et al. 1999, Gill et al. 2001). Biologists and sportsmen alike have concerns as to what factors may be responsible for declining population trends. Although previous and current research indicates multiple interacting factors are responsible, habitat and predation have received the focus of attention. A number of studies have evaluated whether predator control increases deer survival, yet results are highly variable (Connolly 1981, Ballard et al. 2001). Together, predator control studies with adequate rigor and statistical power indicate predation effects on mule deer are variable as a result of time-specific and site-specific factors. Studies which have demonstrated deer population responses to predator control treatments have failed to determine whether predation is ultimately more limiting than habitat when considering long term population changes. Numerous research studies have evaluated mule deer habitat quality, but virtually no studies have documented population responses to habitat improvements. In many areas where declining deer numbers are of concern, predation is common yet habitat quality appears to have declined. The question remains as to whether predation, habitat, or some other factor is more limiting to mule deer in these situations, and whether habitat quality can be improved for the benefit of deer. It may also be that no single factor is responsible for observed deer declines, and a more comprehensive understanding of multi-factor interactions is needed. We designed and implemented a field experiment where we measured deer population responses to a nutrition enhancement treatment to further understand the causative factors underlying observed deer population dynamics. We conducted the study on the Uncompahgre Plateau in southwest Colorado, where several predator species were present in abundant numbers: coyotes (Canis latrans), mountain lions (Felis concolor), and bears (Ursus americanus). In addition to predation, myriad diseases in combination have proximately affected survival of the Uncompahgre deer population (Pojar and Bowden 39 �2004, B. E. Watkins, Colorado Division of Wildlife, unpublished data). Predator numbers were not manipulated in any manner during the course of the study. All factors were left constant with the exception of deer nutrition. Deer nutrition was enhanced by providing supplemental feed to deer occupying a treatment area during winter. We measured December fawn recruitment and overwinter fawn survival in response to the treatment to determine whether deer nutrition was ultimately more limiting than predation or disease. The second phase of the research will incorporate habitat manipulation treatments. The treatments will consist of prescribed fire or mechanical techniques to set back succession of pinyon-juniper (Pinus edulis-Juniperus osteosperma) habitat in an effort to improve the vigor and quality of winter habitat for mule deer. Deer population responses will be measured in relation to the habitat manipulations in the same manner as the supplemental feed. Thus, the experiment evaluates whether nutritional quality of winter range habitat is ultimately more limiting than other factors in a lateseral pinyon-juniper and sagebrush (Artemisia spp.) landscape, and if so, whether habitat can be effectively improved for mule deer. The results advance our understanding of multi-factor interactions, with direct implications for mule deer management. STUDY AREA We non-randomly selected two experimental units (A−B) within mule deer winter range on the Uncompahgre Plateau (Figure 1) to facilitate a cross-over experimental design for evaluating the effects of enhanced deer nutrition during winter on annual population performance. We used the following criteria to select experimental units: 1.) Deer densities (≥80 deer/mi2): we selected areas where deer densities were sufficient to meet sample size requirements within the experimental unit, while simultaneously selecting areas that would require feeding no more than 600−800 animals during a normal winter; 2.) Buffer zones: we selected areas such that experimental units would be separated by several miles of non-treatment area (buffers) to prevent deer from occupying more than one experimental unit; 3.) Similarity: we selected areas that comprised relatively similar habitat complexes and deer densities that were representative of the overall area; 4.) Elk populations: we selected areas in an effort to minimize the number of elk present during normal winters. Units A and B received the nutrition enhancement treatment in a cross-over experimental design to address P.N. Objective 1. Unit A served as the treatment unit, while Unit B served as the control, for the first 2 winters of research (2000 – 2002). During winters 2002−03 and 2003−04, Unit B received the treatment while Unit A served as the control. Upon completion of P.N. Objective 1, additional winter range experimental units will be used to conduct phase 2 of the research, or P.N. Objective 2. Habitat in treatment units will be manipulated to set back plant succession, while habitat in control units will remain unchanged throughout the experiment. Experimental units A and B were defined as follows (Figures 2 and 3): (1) Experimental unit A included the Colona Tract of the Billy Creek State Wildlife Area and adjacent land, located approximately 13 km south of Montrose, CO adjacent to U.S. Hwy 550 South. The experimental unit was located within the Colona USGS 7.5 Minute Quadrangle, and roughly included the polygon defined by the following Zone 13 UTM coordinates: (1) 254000 E, 4250200 N; (2) 252700 E, 4249400 N; (3) 254700 E, 4245600 N; and (4) 256200 E, 4246600 N. (2) Experimental unit B included Shavano Valley and adjacent land extending west to the Dry Creek Rim. Shavano Valley is located approximately 13 km west of Montrose, CO. The experimental unit was located within the Dry Creek Basin and Montrose West Quadrangles (USGS 7.5 Minute), and 40 �roughly included the polygon defined by the following Zone 13 UTM coordinates: (1) 238400 E, 4262600 N; (2) 232400 E, 4256700 N; (3) 235000 E, 4253600 N; and (4) 239500 E, 4258200 N. In late April and May, prior to fawning, deer from the winter range experimental units migrated to summer range. We defined the summer range study area by movements of the radio-collared deer captured on winter range; summer range encompassed >1000 mi2 covering the southern portion of the Uncompahgre Plateau and adjacent San Juan Mountains (Figure 2). The summer range study area extended north to the Dry Creek river drainage on the Uncompaghre Plateau, south to Mineral Creek near Silverton, CO, east to the Big Blue River drainage, and west to the San Miguel River canyon. However, a majority of the radio-collared deer summered on the Uncompahgre Plateau between Dry Creek to the north and Highway 62 to the south. Winter range elevations ranged from 1830 m (6000 ft) in Shavano Valley to 2318 m (7600 ft) adjacent to the Dry Creek Rim above Shavano Valley. Winter range habitat was dominated by pinyonjuniper with interspersed sagebrush adjacent to agricultural fields in the Shavano and Uncompahgre Valleys. Summer range elevations occupied by deer ranged from 1891 m (6200 ft) in the Uncompahgre Valley to 3538 m (11,600 ft) in Imogene Basin southwest of Ouray, CO. Summer range habitats were dominated by spruce-subalpine fir (Picea spp.-Abies lasiocarpa), aspen (Populus tremuloides), sagebrush, ponderosa pine (Pinus ponderosa), Gambel oak (Quercus gambelii), and to a lesser extent, pinyon-juniper at lower elevations. METHODS Response Variables We measured fetal and neonatal survival rates, early winter fawn:doe ratios, and overwinter fawn survival rates of deer occupying the treatment and control units. We delivered the nutrition enhancement treatment to deer from December through April, assessed fetus survival during June, measured neonate survival from June to December, and fawn:doe ratios during December−February (1 year after the treatment was initiated). We measured overwinter fawn survival from December to June in direct response to the current winter’s treatment. Our measurements determined whether enhanced winter nutrition of adult does increased subsequent newborn fawn production and survival, and whether enhanced winter nutrition of 6−12-month old fawns increased overwinter fawn survival. Ultimately, these measurements provided an assessment of the effect of winter range habitat quality on yearling recruitment, and thus population productivity. We also measured overwinter and annual survival of adult does as a function of enhanced winter nutrition. Sample Size Fetus/Neonate Survival: Fetus and neonate sample sizes were not independent of one another because each resulted from the sample size of radio-collared does. We therefore needed a target sample size of either fetuses or neonates to generate our adult doe sample size. We based our sample size calculations on quantifying neonate survival because it was our highest priority and we could generate reliable estimates. Target fetus sample sizes were difficult to estimate because of uncertainty identifying fetus fates. That is, many fetuses measured in utero during winter were not accounted for as live or dead at parturition. Fetus survival rates could only be measured from some unpredictable fraction of the radiocollared doe sample, making sample size calculations of limited use. For neonate survival, a sample size of 40 neonates per experimental unit per year would provide power of 0.81 to detect a difference of 0.15 in survival between the 2 experimental units if survival among control fawns was 0.40. We assumed a control survival rate of 0.40 based on previous neonate survival rates measured on the Uncompahgre (Pojar and Bowden 2004) in combination with December fawn:doe ratios measured during the late 1980s and 1990s, when the Uncompahgre population declined (B. E. Watkins, Colorado Division of Wildlife, 41 �unpublished data). We considered 40 neonates per experimental unit per year a minimum sample size because we ideally wanted to detect a difference in neonate survival of <0.15 between experimental units. Based on Bishop et al. (2002), we determined that 60 radio-collared does (30 treatment and 30 control) equipped with vaginal implant transmitters (VITs) would be necessary to capture a minimum of 80 newborn fawns. We also assumed that some fawns would be captured from other treatment and control radio-collared does not equipped with VITs. The 60 radio-collared does with VITs were also used to evaluate fetus survival; however, logistical constraints limited the power of fetus survival comparisons among experimental units. Early winter fawn:doe ratios: We desired to detect an effect size, i.e., an increase in fawn:doe ratios in response to the treatments, in the range of 15 to 20 fawns per 100 does. These values were based on population models with overwinter fawn survival of 0.444, adult female survival of 0.853, and December fawn:doe ratios of 66 fawns per 100 does to obtain a stationary population (Unsworth et al. 1999). Based on surveys of the Uncompahgre deer population during the 1990s, the standard deviation of the fawns:100 does ratio for groups with at least one adult female was 57, with a mean of 41. Using an expected standard deviation of 57, the standard error of the mean fawns:100 does ratio for 40 radiocollared does is 57/(401/2) = 9.0, which is the expected standard deviation of measured fawns:100 does ratios on each experimental unit. We assessed power using a two-sample t-test with a sample size of 4, representing the 4 years of the study where fawn:doe ratios were measured in response to enhanced nutrition. Our power to detect an increase of 20 fawns per 100 does based on classification of 40 radiocollared doe groups in each experimental unit was about 0.87. Overwinter fawn survival: Our sample size of 40 fawns per experimental unit per year provided a power of 0.81 to detect a difference of 0.15 in survival between the 2 experimental units assuming a control survival rate of 0.40. We expected to see an increase in fawn survival (effect size) of approximately 0.15, because this was the difference measured in the density reduction experiment conducted by White and Bartmann (1998). We assumed a control survival rate of 0.40 based on longterm data from Colorado, Idaho, and Montana (Unsworth et al. 1999). However, recent data from 5 deer populations in Colorado indicates overwinter fawn survival has commonly been ≥70% during the past 6 years (Colorado Division of Wildlife, unpublished data). Adult and 6-month Old Fawn Capture We captured adult does and 6-month-old fawns during November and December using baited drop nets (Ramsey 1968, Schmidt et al. 1978) and helicopter net guns (Barrett et al. 1982, van Reenen 1982). We baited drop nets with certified weed-free alfalfa hay and apple pulp. We used drop nets as the principle capture technique for a 3−4 week capture period; we then used helicopter net-gunning at the end of the drop-net capture to secure the remainder of deer needed to meet our target sample sizes. All deer were hobbled and blind-folded after being captured. We used stretchers to carry deer away from the net when using drop nets. Deer were fitted with nylon-belting radio collars equipped with mortality sensors; pulse rate increased after remaining motionless for 4 hours. We placed permanent collars on adult females and temporary collars on fawns. To make collars temporary, we cut one end of the collar in half and reattached the two ends using rubber surgical tubing; fawns shed the collars ≥6 months post-capture. We stitched a rectangular piece of flexible plastic (Ritchey® neck band material) engraved with a unique identifier to the side of each collar. The unique identifier consisted of 2 symbols for adult females, and 1 symbol on 2 different colors of plastic for fawns. We used the identifiers to visually identify deer from the ground, which allowed us to effectively document use of the treatment, measure fawn:doe ratios, and assess experimental unit population size via mark-resight estimators. We recorded mass (kg), hind foot length (cm), and chest girth (cm) of each deer, and collected blood samples to evaluate disease prevalence. 42 �During late February and early March, we captured an additional 30 adult female deer in each experimental unit by net-gunning. Captured deer were ferried by the helicopter to a central processing location, where deer were carried by stretchers to a tent for handling. We used ultrasonography to measure pregnancy status, fetal rate, and body condition of each captured deer. We retained and radiocollared pregnant does only. We then inserted a vaginal implant transmitter (VIT) in each doe as a technique for locating the timing and location of her birth site the following June. We also recorded the weight (kg), hind foot length (cm), and chest girth (cm) of each deer, and collected blood samples to evaluate disease prevalence. Body Condition and Reproductive Status We estimated body fat of treatment and control adult does during mid-late winter using an Aloka 210 (Aloka, Inc., Wallinford, Conn.) or SonoVet 2000 (Universal Medical Systems, Bedford Hills, NY) portable ultrasound unit with a 5 MHz linear transducer. We measured maximum subcutaneous fat thickness on the rump (MAXFAT) following the methodology of Stephenson et al. (1998, 2002). We also measured thickness of the longissimus dorsi muscle via ultrasound (Cook et al. 2001, Stephenson et al. 2002). A small area of hair was shaved to ensure contact between the transducer and the skin. Lubricant was applied to the shaved area for conduction purposes and fat and muscle thickness were measured using electronic calipers. We coupled the ultrasound measurements with body condition scores (BCS) obtained from palpation of the ribs, withers, and rump (Cook 2000). MAXFAT and rump BCS measurements were combined into a condition index used to estimate percent body fat (Cook and Cook 2002): % Fat = -6.6387617 + 7.4271417x – 1.11579443x2 + 0.07733803x3 where x = rLIVINDEX = (MAXFAT – 0.15) + rump BCS (if MAXFAT < 0.15, then rLIVINDEX = rump BCS). The rLIVINDEX and body fat regression was initially developed and validated for elk by Cook et al. (2001), and then modified by incorporating a validation of MAXFAT for mule deer performed by Stephenson et al. (2002). We also evaluated differences in serum thyroid hormone concentrations between treatment and control adult does during mid-late winter. Specifically, we measured total thyroxine (T4), free T4 (FT4), total tri-iodothyronine (T3), and free T3 (FT3) following the methodologies of Watkins et al. (1983, 1991). Blood samples were collected at the time of capture, and serum hormone analyses were performed by the Michigan State University Animal Health Diagnostic Laboratory (East Lansing, Michigan). We compared serum thyroid hormone concentrations between treatment and control adult does, and also compared hormone levels to body fat estimates derived from the ultrasonography. We quantified reproductive status (Stephenson et al. 1995, Andelt et al. 2004) with ultrasound via transabdominal scanning using a 3 MHz linear transducer. We searched for fetuses by scanning a portion of the abdomen that was shaved caudal to the last rib and left of the midline. We systematically searched each uterine horn to identify fetal numbers ranging from 0 to 3. Whenever possible, we measured eye diameter of each fetus to approximately estimate fetal age and parturition date. Vaginal Implant Transmitters (VITs) We used VITs manufactured by Advanced Telemetry Systems, Inc. (Isanti, MN). The VIT was 76 mm long, excluding antenna length, and had 2 silicone wings with a width of 57 mm when fully spread apart. The silicone wings were used to retain the transmitter in the vagina until parturition. The VIT weighed 15 grams and contained a 10−28 lithium battery programmed to a 12-hour on/off cycle. The diameter of the transmitter (excluding wings) was 14 mm, and was encased in an impermeable, water-proof, electrical resin. The transmitter contained an embedded heat-sensor which dictated the frequency pulse rate. When the heat sensor dropped below 90°F, synonymous with transmitter expulsion from the deer, the pulse rate changed from 40 PPM to 80 PPM. VIT batteries were programmed to be active from 0430 to 1630 hrs prior to daylight savings, and thus were active from 0530 to 1730 hrs after daylight savings and during the fawning period. We inserted VITs into deer using a vaginoscope 43 �(Jorgensen Laboratories, Inc., Loveland, CO) and alligator forceps. The vaginoscope was 6” long with a 5/8” internal diameter and had a machined end (smooth surface) to minimize trauma when inserted into the vagina. A discreet mark was placed on the applicator showing approximate insertion distance. We obtained the length of a typical mule deer vaginal tract by taking measurements from road-killed deer and other fresh deer carcasses obtained in the study area. Prior to use in the field, VITs were sterilized using chlorhexidine, air-dried, and sealed in a 3” × 8” sterilization pouch. We used sterilization containers with diluted chlorhexidine on site during capture to sterilize the vaginoscope and alligator forceps between each use. We used a new pair of nitrile surgical gloves to handle the vaginoscope and VIT for each deer. To insert a VIT, the plastic wings were folded together and placed into the end of the vaginoscope. We liberally applied sterile KY Jelly® to the scope and inserted it into the vaginal canal until the tip of the VIT antenna was approximately flush with the vulva. We used the alligator forceps, which extended through the vaginoscope, to firmly hold the VIT in place while the scope was pulled out from the vagina. The VIT silicone wings spread apart upon removal of the scope to hold the transmitter in place. The transmitter antenna was typically flush with the vulva, but on occasion extended up to 1 cm beyond the vulva. The tip of the antenna was encapsulated in a wax bead to protect the deer. All capture and handling procedures, including VIT techniques, were approved by the Colorado Division of Wildlife’s Animal Care and Use Committee (project protocols 11−2000 and 1−2002). Neonate Fawn Capture All radio-collared adult does were relocated from the air during late May to identify likely fawning areas. During each morning of June we checked VIT signal status by aerially relocating radiocollared does having VITs. Implant radio-signals could not be easily monitored from the ground because of weak signal strength and a large study area. Flights began at 0530 hours and were usually completed by 1000–1100 hours. Early flights were necessary to detect fast signals because temperature sensors of VITs expelled in open habitats and subject to sunlight often exceeded 90°F by mid-day, which caused VITs to switch back to a slow (i.e. prepartum) pulse. When a fast (i.e. postpartum) pulse rate was detected, we ground-tracked both the VIT and radio-collar frequencies simultaneously because the shed VIT and adult doe were typically in close proximity to one another. We attempted to observe behavior of the collared doe, establish whether the VIT was shed at a birth site, and search for fawns in the vicinity of the doe and expelled VIT. In cases where the doe had moved away from the VIT (e.g. >200 m), we located the VIT to determine whether shedding occurred at a birth site and whether any stillborn fawn(s) were present, and subsequently located the collared doe to search for fawns at her location. We attempted to account for each doe’s fetuses as live or stillborn fawns in order to quantify in utero fetus survival from February to birth. All personnel wore surgical gloves when handling fawns to help minimize human scent. We placed a drop-off radio-collar on each live fawn; radio collars were constructed with elastic neck-band material to facilitate expansion. Hole-punched, vinyl-belting tabs extended from the end of the elastic and from the transmitter for attachment purposes. We made collars temporary by cutting the vinyl tab extending from the elastic and reattaching the belting with latex tubing, which caused the collars to shed from the animal >6 months post-capture. Some collars were shed prematurely (i.e. 4−5 months postcapture) in association with fences during fall migration. For each fawn, mass (kg) and hind foot length (cm) were recorded, and a nasal swab sample was collected to screen for Bovine Viral Diarrhea. We then recorded basic vegetation characteristics of the birth site and promptly exited the site. We ground-relocated most of the radio-collared does not receiving VITs approximately every other day during June in an attempt to capture additional fawns from treatment and control does. We did the same for any VIT doe whose implant failed because of premature expulsion or battery failure. We relied on doe behavior and searches in the vicinity of the collared does to locate fawns. We worked in 44 �pairs and partitioned the study area into segments, whereby each 2-person team was responsible for one segment. We used 3−4 teams during 2002 and 5−6 teams during 2003 and 2004. Measurement of Survival Rates and Fawn:Doe Ratios We measured survival rates by radio-monitoring collared deer from the ground and air to determine fate (i.e. lived or died). We also attempted to determine the cause of each mortality, with a primary goal of distinguishing between predation and non-predation mortality causes. We radiomonitored deer from the ground on a daily basis year-round and from the air on approximately a biweekly basis. We detected signals from nearly all radio-collared deer each day during winter, which typically allowed us to arrive at mortality sites within 24 hours of the mortality event. During summer and migration periods, deer were distributed widely and thus were more difficult to radio-monitor. All radiocollared neonates were checked daily throughout the summer and fall, whereas some adult and yearling deer could not be ground-monitored on a routine basis. In result, we typically located neonate mortalities within 24 hours of death, but some adult deer mortalities were not detected for several days, or on rare occasion, for one or more weeks. Fresh, intact neonate carcasses were collected and submitted to the Colorado Division of Wildlife’s Wildlife Health Laboratory or the Colorado State University Diagnostic Laboratory for necropsy and tissue analyses. Fresh, intact adult and 6-month-old fawn carcasses were also submitted for laboratory necropsy when feasible. Field necropsies were performed on all other deer mortalities, and when appropriate, tissue samples were collected and submitted for analysis. Each winter we used the radio-collared does to measure fawn:doe ratios in each experimental unit. The resulting fawn:doe ratio was a measurement of the previous year’s treatment effect. We measured fawn:doe ratios using 2 techniques: (1) We located the sample of radio-collared does in each experimental unit from a fixed-wing airplane, and used the set of locations to define boundaries for the experimental unit. Shortly after (i.e. 1−2 days), we used a helicopter to systematically fly the defined unit and classify all deer groups encountered. For each group, we documented whether a radio-collared doe was present. (2) We located each radio-collared doe by radio telemetry from the ground. The group of deer with the collared doe was counted and classified by age and sex. Both methods were employed to gather as much information as possible to determine whether there was a treatment effect. The “true” value cannot be measured perfectly because of the inherent biases and potential sources of error associated with each technique. Thus, by employing both techniques, we had a greater chance of fully understanding whether the treatment caused an effect. Treatment Delivery We enhanced deer nutrition in the treatment experimental unit by providing a safe, pelleted supplemental feed. The supplemental feed was developed through extensive testing with both captive and wild deer (Baker and Hobbs 1985, Baker et al. 1998), and has been safely used in both applied research and management projects. We distributed pellets daily using 4wd pickup trucks, ATVs, and snowmobiles on primitive roads throughout the experimental unit to provide a food source for the entire deer population in the treatment unit. We carried each 50 lb. bag of pellets ≤200 m from the vehicle and distributed it by hand in approximately 20−30 small piles of feed in a linear fashion. We distributed numerous bags in successive order to create straight lines of feed that spanned most of the treatment area, which prevented animal concentrations. Our feeding technique also prevented dominant animals from restricting access to the food supply because of the large area over which pellets were distributed. We attempted to supply pellets ad libitum such that residual pellets remained when the next day’s ration was provided. We closely monitored collared deer to ensure that treatment deer remained in the experimental unit and actually consumed the feed, and to make sure that non-treatment deer remained in the control unit, which they did. The few treatment adult does that distinctly moved away from the treatment unit were withdrawn from the sample for purposes of measuring treatment effects. However, to avoid any biases, all 6-month-old fawns captured in the treatment unit were included in survival analyses regardless 45 �of whether they accessed the supplement or not. Some fawns died shortly after capture (i.e. 2−3 weeks), before we could document whether they had access to the feed. Censoring these individuals would have biased treatment survival high relative to control survival. Also, very few fawns that survived more than 2−3 weeks moved away from the treatment unit. The pelleted ration was commercially produced in the form of 2×1×0.5-cm wafers (Baker and Hobbs 1985). Feed quality (e.g. digestible energy, protein) vastly exceeded those of typical winter range deer diets; exact constituent values are provided by Baker et al. (1998). When provided ad libitum, the feed should have allowed deer to meet or exceed nutritional requirements for growth and maintenance (Ullrey et al. 1967, Verme and Ullrey 1972, Thompson et al. 1973, Smith et al. 1975, Baker et al. 1979, Holter et al. 1979). The basis for feeding such high quality pellets was to ensure that the treatment (enhanced nutrition) was effectively delivered to the deer. Our intent was not to determine the exact level of nutrition necessary to increase fawn recruitment, but rather to determine if nutrition was a significant limiting factor to recruitment. We will rely on habitat manipulation treatments to evaluate what exactly can be done via management to increase fawn survival and recruitment if nutrition is deemed a critical limiting factor. Statistical Methods We estimated deer numbers in each experimental unit during the first year of research using helicopter and ground mark-resight surveys. We used the joint hypergeometric maximum likelihood estimator for helicopter surveys and the Bowden estimator for ground surveys, and we analyzed data in Program NOREMARK (Neal et al. 1993, Bowden 1993, White 1996). We used a general linear model in PROC GLM in SAS (SAS Institute 1989) to test for differences in estimated percent body fat between treatment and control adult does and a multivariate model to test for differences in T4, FT4, T3, and FT3 thryoid hormones between treatment and control does. We used PROG REG (SAS Institute 1989) to evaluate the relationship between estimated percent body fat and serum thyroid hormone concentrations. We entered all fawn:doe ratios from helicopter surveys into the CDOW Deer, Elk, and Antelope Management (DEAMAN) database (G. C. White, Colorado State University, software) and computed standard errors based on groups (Bowden et al. 1984). We analyzed fawn:doe ratios from ground surveys using PROC MIXED in SAS (SAS Institute 1997). We used a reduced model with experimental unit as the independent variable; we considered experimental unit as a fixed effect and radio-collared does within an experimental unit as random effects. We analyzed fetus survival with a binomial survival rate from the subset of does where all fetuses had known fates. We also indirectly analyzed fetus survival by comparing the February fetus rate with the number of live newborn fawns/doe observed in June using a change-in-ratio estimator (White et al. 1996). We estimated neonate and overwinter fawn survival and adult doe survival using a Kaplan-Meier survival analysis (Kaplan and Meier 1958, Pollock et al. 1989), and we contrasted survival among experimental units using chi-square analyses. We used a common entry date for analyzing neonate survival because staggered entry would have biased survival rates low due to early mortalities that occurred before most of the sample was captured. We analyzed continuous fetus-neonate-overwinter fawn survival from March of one year to June of the following year using a staggered-entry Kaplan-Meier survival analysis (Pollock et al. 1989). All neonates were entered into the survival analysis on a common date rather than the exact date of capture for the same reason mentioned above. We computed the finite rate of increase, λ, for treatment and control deer by constructing a deterministic age-structured population model using measured pregnancy and fetus rates, fetus survival, neonate survival, overwinter fawn survival, and annual adult doe survival. Results are based on preliminary analyses and should be treated as such. Other results are presented as data summaries incorporating means and standard errors, or in some cases, raw data values. 46 �RESULTS AND DISCUSSION Deer Capture During November and December 2000−2003, we captured and radio-collared 139 adult female mule deer evenly distributed among the treatment and control units. We also captured and radio-collared 241 6-month-old fawns during November and December 2001−2003 (40 fawns/unit/year). Due to budgeting constraints, we were unable to radio-collar 6-month old fawns during 2000. We captured an additional 154 adult females during late February and early March 2002−2004 and equipped them with radio collars and VITs. During June 2002−2004, we captured and radio-collared 276 newborn fawns from radio-collared adult females. Thus, the following results are based upon radio-monitoring of 810 individual mule deer evenly distributed among treatment and control units during November 2000−June 2004. Treatment Delivery 2000−01: We distributed 88 tons of supplemental pellets from December 15, 2000, through April 19, 2001. We distributed an average of 0.85 tons of feed each day throughout 22 feeding sites across the 2.3 mi2 treatment unit during most of the winter and spring. Deer were fed ad libitum because there was always residual feed remaining the next day during the feeding routine. We distributed each sack in approximately 20−30 distinct, small piles, resulting in >1000 small piles of feed throughout the treatment unit. Deer were able to effectively access the feed in small groups, and no aggression was ever observed among deer seeking access to the feed. Deer adapted to the pelleted supplement immediately and utilized it extensively throughout the winter. We continually monitored deer use of the feed from ground observation points, where we obtained 440 visual observations of radio-collared does consuming the feed. These observations, coupled with daily radio-monitoring and periodic aerial relocations, indicated 32 of the 37 radio-collared treatment does spent the entire winter and spring within the boundaries of the treatment unit and received the supplement on a daily basis. Mark-resight population estimates from March helicopter (489 deer, SE = 62) and ground (494 deer, SE = 81) surveys, coupled with feed consumption, indicated we fed roughly 450 to 500 deer during most of the winter and spring. Feed consumption declined coincident with spring green-up, although deer continued to use the feed through mid-late April, at which point they began migrating to summer range. We also fed approximately 25 to 30 elk, but the elk did not affect deer access to the feed. Deer in the control experimental unit did not receive feed or any other treatment. Based on helicopter mark-resight surveys, the deer density in the treatment unit in December was 120 deer/mi2 (SE = 9), but increased shortly after and was 213 deer/mi2 (SE = 27) in March. Deer densities in the control unit changed little from 83 deer/mi2 (SE = 12) in December to 101 deer/mi2 (SE = 14) in March. 2001−02: We distributed 194 tons of the supplement throughout the treatment unit from December 15, 2001, through April 25, 2002. We distributed 2.0−2.1 tons of feed each day for most of the winter and spring. The large increase in supplement distribution from the previous year occurred because a large number of elk descended into the Uncompahgre Valley during late fall. Elk arrived in unusually large numbers throughout much of the valley prior to the onset of treatment delivery. Once feeding was initiated, approximately 300−500 elk adapted to the feed and remained in or around the treatment unit throughout most of the winter. We could not deliver >2.1 tons of pellets per day given myriad logistical and budgetary constraints. Feed was not delivered ad libitum to all deer and elk in the treatment unit throughout the winter because residual feed was rarely observed during the next day’s distribution. However, daily field observations indicated most deer approached ad libitum consumption of the supplement. In contrast to the previous winter, deer were waiting for the daily supplement to arrive each morning. Deer then consumed the supplement immediately after it was distributed. Elk were rarely observed utilizing the 47 �feed until late morning or afternoon, and elk continued to forage in fields below the treatment unit, whereas deer did not. We observed numerous radio-collared deer consuming pellets each day; not all of these observations were recorded because of time constraints with distributing the feed. Given this time limitation, we still recorded 818 observations of radio-collared deer consuming the supplemental feed (497 collared doe observations and 321 collared fawn observations). We observed 100−300 deer utilizing the pellets most days during the course of distributing the supplement. These observations rarely included elk; thus, direct deer-elk competition was minimized because of temporal differences in feeding, and deer had first access to the feed. 2002−03: We switched the treatment and control units consistent with the cross-over experimental design in December 2002. We distributed 97 tons of supplement from December 15, 2002 through April 30, 2003 across the new treatment unit, which had been the control unit the previous 2 years. The supplement was distributed daily throughout 29 sites over a larger area (~7 mi2) than the first 2 years of research because of the greater size of the experimental unit and broader distribution of radiocollared deer. Residual feed was always present throughout the winter, thus deer were fed ad libitum. Only small groups of elk periodically accessed the supplement, and did not affect deer access. We obtained 286 observations of radio-collared deer consuming the supplement, which were difficult to obtain because the supplement was spread out over a large area and only a single feed site could be observed at any given moment. We also used daily ground radio-monitoring and periodic aerial relocations to document deer access to the supplement. 2003−04: We distributed 197 tons of pellets throughout the treatment unit from December 10, 2003, through April 30, 2004. The increase in supplement distribution occurred because elk numbers increased on the upper portion of the experimental unit. However, unlike winter 2001−02, residual feed was present throughout the winter and deer were fed ad libitum. We restricted elk to the upper extent of the deer winter range for most of the winter by allocating a portion of the daily feed distribution exclusively to elk. Thus, elk had a minimal affect on deer access to the supplement. We obtained 413 observations of radio-collared deer consuming the supplement. As before, we also used daily ground radio-monitoring and periodic aerial relocations to document deer access to the supplement. Body Condition Estimated percent body fat of adult does during late February and early March, 2002–2004, was higher for treatment deer than control deer (F1, 148 = 153.41, P < 0.001). Over all years combined, mean predicted body fat was 9.8% (SE = 0.36) for treatment adult does and 4.3% (SE = 0.26) for control does. The interaction of experimental unit × year for predicted body fat was also significant (F2, 148 = 14.39, P < 0.001). This interaction occurred because the difference in body fat between treatment and control deer was greater during 2003 than during 2002 or 2004. Mean predicted body fat was 8.2% (SE = 0.92) for treatment adult does and 5.0% (SE = 0.71) for control does during 2002, and 9.0% (SE = 0.53) for treatment does and 4.7% (SE = 0.36) for control does during 2004. The difference was greater during 2003, where mean predicted body fat was 11.7% (SE = 0.35) for treatment does and 3.4% (SE = 0.35) for control does. The body fat estimates reported here should accurately reflect deer, but may be further refined in the future as additional research provides more data on the relationship between body condition indices and estimated percent body fat. Serum thyroid hormone concentrations, measured during 2003 and 2004, were higher in treatment does than control does (F4, 108 = 46.59, P < 0.001) (Table 1). Hormone concentrations also varied between years (F4, 108 = 14.21, P < 0.001), but the experimental unit × year interaction was not significant (F4, 108 = 1.68, P = 0.160). Thus, each year thyroid hormone concentrations were higher in treatment does than control does. T4 was the most important thyroid hormone in describing the canonical variable for differences between treatment and control does (1.04*T4 − 0.02*T3 + 0.77*FT4 – 48 �0.17*FT3). As expected, there was a high partial correlation between T4 and FT4 (r = 0.67, P < 0.001) and between T3 and FT3 (r = 0.60, P < 0.001), which has been documented previously (Watkins et al. 1983). When treated as 4 separate ANOVAs, T4 (F1, 111 = 165.97, P < 0.001), FT4 (F1, 111 = 144.37, P < 0.001), T3 (F1, 111 = 13.84, P < 0.001), and FT3 (F1, 111 = 8.26, P = 0.005) were significantly higher in treatment does than control does. Given these results, we evaluated the relationship between T4 concentrations and estimated percent body fat (derived from ultrasound and BCS indices) using a simple linear regression model (% Fat = −3.122 + 0.090*T4, r2 = 0.52, P < 0.001). Similar correlations between T4 and actual percent body fat during mid-late winter have been previously documented for white-tailed deer and elk (Watkins et al. 1991, Cook et al. 2001). Pregnancy and Fetus Rates 2002: Adult doe pregnancy rate was 0.95 (SE = 0.037, n = 38) in February−March 2002. We measured an average of 1.80 fetuses/doe (SE = 0.10, n = 36), which included 1.77 fetuses/doe (SE = 0.14, n = 18) in the treatment unit and 1.83 fetuses/doe (SE = 0.15, n = 18) in the control unit. 2003: Adult doe pregnancy rate was 0.92 (SE = 0.034, n = 63) in February−March 2003. Critical personnel and equipment for measuring fetus rates were not continuously available due to capture delays associated with helicopter mechanical problems. Some deer fetus counts were performed by inexperienced observers without optimum ultrasound equipment. VITs worked very well, though, allowing us to determine fetus numbers at parturition for many of the deer. Thus, we determined winter fetus rates by using the greatest fetus count for each individual deer, whether obtained using ultrasound during February−March or by locating newborn fawns and stillborns at birth sites during June. We were unable to determine a fetus count for 8 treatment deer because only pregnancy was established with ultrasound and no birth site assessments were possible in June. These 8 deer were removed from the fetus rate estimates. Of the 50 deer where a fetus count was obtained, 5 were yearlings (2 treatment yearlings, 3 control yearlings). We measured 1.74 fetuses/doe (SE = 0.069, n = 50) overall including yearlings, and 1.82 fetuses/doe (SE = 0.066, n = 45) excluding yearlings. Fetus rates with yearlings included were 1.77 fetuses/doe (SE = 0.091, n = 22) in the treatment unit and 1.70 fetuses/doe (SE = 0.10, n = 28) in the control unit. 2004: In February 2004, adult doe pregnancy rate was 0.94 (SE = 0.029, n = 66) and the fetus rate was 1.97 fetuses/doe (SE = 0.053, n = 60), which included 4 yearlings. Excluding yearlings, the fetus rate was 2.00 fetuses/doe (SE = 0.051, n = 56). Fetus rates were 1.90 fetuses/doe (SE = 0.074, n = 30) in the treatment unit and 2.03 fetuses/doe (SE = 0.076, n = 30) in the control unit with yearlings included, and 1.93 (SE = 0.069, n = 29) in the treatment unit and 2.07 (SE = 0.074, n = 27) in the control unit with yearlings excluded. Pregnancy and fetus rates during our study equaled or exceeded other measured rates recorded in Colorado (Andelt et al. 2004), indicating moderate to high innate productivity potential for both treatment and control does. Our data also indicate that adequate numbers of bucks were available to breed does during the years of our study. Fetus and Neonate Survival/Fawn:Doe Ratios 2000: Fawn:doe ratios were similar in the 2 experimental units in December 2000, prior to the first year’s treatment delivery. Pre-treatment fawn:doe ratios were 52.6 fawns:100 does (SE = 5.3) in the Colona experimental unit and 51.6 fawns:100 does (SE = 5.0) in the Shavano experimental unit. 2001: We conducted 2 age classification helicopter surveys in the treatment and control units in late December 2001 and early January 2002, following the first year’s treatment. On 23 December 2001, we observed 52.8 fawns:100 does (SE = 6.7) in the treatment unit and 36.7 fawns:100 does (SE = 3.8) in 49 �the control unit. On 8 January 2002, we observed 54.7 fawns:100 does (SE = 6.6) in the treatment unit and 50.5 fawns:100 does (SE = 6.0) in the control unit. During December 2001 – February 2002, we obtained fawn:doe ratio estimates from ground observations of radio-collared deer groups for both treatment and control deer. This survey resulted in 61.2 fawns:100 does (SE = 7.8) in the treatment unit and 74.5 fawns:100 does (SE = 8.5) in the control unit, although the result was not statistically significant (t74 = 1.16, P = 0.249). Our fawn:doe ratio results were conflicting and did not provide evidence that there was any treatment effect. We could not make any sound conclusions based on the data, although we generally concluded the nutrition enhancement treatment did not cause a substantial increase in neonatal production and survival during 2001. These data provided the incentive to incorporate direct measurements of fetus and neonate survival into our research. 2002: We measured fetus and neonate survival directly during March – December, 2002, following the second year’s treatment; however, sample sizes were based on a technique assessment of VITs and were relatively small for contrasting survival rates among treatment and control fetuses and neonates (Bishop et al. 2002). During June 2002, we determined the fate of all fetuses (live or stillborn) from only 14 of 36 VIT does, largely because of a high VIT battery failure rate. Numbers of stillborns were similar among treatment and control deer, so we did not differentiate by experimental unit. The survival rate of fetuses (n = 22) from the 14 does was 0.86 (SE = 0.073). We also assessed fetus survival using a change-in-ratio estimator between the fetal rate measured in February−March and the observed number of live fawns/doe postpartum in June. In June 2002, considering all does (n = 43) that we located any fawn from, whether live or stillborn, we observed 1.42 (SE = 0.11) live fawns/doe postpartum. This rate should represent a conservative estimate of live fawns/doe postpartum because we inevitably failed to locate all live fawns from each doe. In other words, this estimate would treat any unaccounted fetuses (from the February measurement) as if they were stillborns. For radio-collared does that did not have VITs, and thus we did not have a winter fetus rate measurement, singletons would infer that either the deer only had 1 fetus, or that the other fetus died. It is likely that some of these singletons had a twin that we did not locate. This equates to a conservative fetus survival rate estimate of 0.79 (SE = 0.18). Treatment fawn survival (Jun – Dec) was 0.613 (SE = 0.115, n = 29) and control fawn survival was 0.511 (SE = 0.108, n = 25). In late December 2002 and early January 2003, we once again conducted 2 age classification helicopter surveys in the treatment and control units. On 31 December 2002, we observed 91.9 fawns:100 does (SE = 8.4) in the treatment unit and 52.2 fawns:100 does (SE = 6.9) in the control unit. On 21 January 2003, we observed 52.6 fawns:100 does (SE = 6.4) in the treatment unit and 36.8 fawns:100 does (SE = 3.9) in the control unit. The combined helicopter survey data indicated 68.1 fawns:100 does (SE = 5.6) in the treatment unit and 42.8 fawns:100 does (SE = 3.5) in the control unit. Conversely, fawn:doe ratio estimates from ground classifications of doe groups during December 2002 – February 2003 were 47.7 fawns:100 does (SE = 6.3) in the treatment unit and 63.4 fawns:100 does (SE = 7.5) in the control unit (t108 = 1.61, P = 0.110). As in 2001, fawn:doe ratio results were conflicting. Helicopter survey data varied between 2 different flights, but consistently indicated a treatment effect. Ground classification data did not indicate a treatment effect. 2003: During June 2003, we determined the fate of all fetuses (live or stillborn) from 33 of 58 VIT does; we had better success because VITs commonly shed at birth sites. The survival rate of fetuses (n = 58) from these 33 does was 0.97 (SE = 0.024). In June 2003, incorporating all does (n = 71) from which we located any fawn, whether live or stillborn, we observed 1.49 (SE = 0.072) live fawns/doe postpartum. Using the change-in-ratio estimator described above, this results in an overall conservative fetus survival rate estimate of 0.86 (SE = 0.15). As in 2002, fetus survival was similar among treatment and control deer and not analyzed separately. During June 2003, we captured and radio-collared 103 newborn fawns born from treatment and control radio-collared does (55 treatment fawns, 48 control fawns). The VITs worked well; we captured 50 �fawns from 41 of the 58 does fitted with VITs. Treatment fawn survival (Jun – Dec) was 0.624 (SE = 0.082) and control fawn survival was 0.483 (SE = 0.093). Final standard errors were larger than expected because a number of fawns shed collars prematurely when crossing fences during fall migration. Using helicopter surveys, we measured 62.4 fawns:100 does (SE = 5.3) in the treatment unit and 50.0 fawns:100 does (SE = 4.9) in the control unit. Estimates from ground classifications of doe groups were 68.0 fawns:100 does (SE = 7.6) in the treatment unit and 62.1 fawns:100 does (SE = 7.6) in the control unit. Age ratio estimates from the helicopter and the ground were more consistent during 2003 than in past years. Overall, observed fawn:doe ratios were consistent with treatment and control fawn survival rates measured from June to December. 2004: We determined the fate of all fetuses from 31 of 60 VIT does. The overall fetus survival rate was 0.90 (SE = 0.040, n = 58). Different from 2002 or 2003, all stillborns were from control does. The survival rate of control fetuses was 0.76 (SE = 0.085, n = 25) as compared to a survival rate of 1.00 (n = 33) for treatment fetuses. Using data from all does (n = 82) in which we located any fawn, the conservative change-in-ratio fetus survival estimate was 0.79 (SE = 0.13) overall, 0.88 (SE = 0.17) for treatment deer, and 0.69 (SE = 0.14) for control deer. We captured and radio-collared 119 newborn fawns born from treatment and control radiocollared does during June 2004 (68 treatment fawns, 51 control fawns). Vaginal implants worked well again, and we had a large sample of non-VIT radio-collared does that we could relocate to opportunistically capture additional treatment and control fawns. Treatment fawn survival (Jun – Dec) was 0.438 (SE = 0.068) and control fawn survival was 0.414 (SE = 0.092). As in 2003, final standard errors were larger than expected because fawns shed collars prematurely during fall migration. Although neonate survival rates were similar among treatment and control fawns, fewer control fawns survived to December because of lower fetus survival. The proportion of fetuses measured in March that were born alive and survived to December during 2004 (i.e. fetus-neonate survival) was 0.438 (SE = 0.068) for treatments and 0.304 (0.073) for controls. Similar to 2002 and 2003, we observed higher December fawn recruitment among treatment deer based on measured survival rates. The difference during 2004 was that stillborn deaths factored in as a larger mortality factor among control deer than during 2002 or 2003. We measured 64.6 fawns:100 does (SE = 5.8) in the treatment unit and 52.7 fawns:100 does (SE = 5.1) in the control unit during helicopter surveys in 2004. Our ground classification estimates were 78.5 fawns:100 does (SE = 6.6) in the treatment unit and 68.7 fawns:100 does (SE = 5.1) in the control unit. Similar to 2003, observed fawn:doe ratios were consistent with treatment and control survival rates. 2002−2004 Fetus-Neonate Survival Summary: Fetus-neonate survival combined over all years of study (1 Mar–15 Dec, 2002–2004) was higher (χ21 = 3.089, P = 0.079) for treatment deer (S(t) = 0.519, SE = 0.048) than for control deer (S(t) = 0.409, SE = 0.052). The high censor rate from shed collars during fall reduced power of the analysis and therefore increased standard errors and the resulting Pvalue. However, at roughly the same time neonate radio-collars were being shed, we captured new samples of fawns for measuring overwinter fawn survival. When fawns captured during November and early December were incorporated into the analysis via staggered entry, fetus-neonate treatment survival (S(t) = 0.528, SE = 0.027) and control survival (S(t) = 0.401, SE = 0.025) rates had tighter standard errors, which reduced the p-value associated with the survival rate comparison (χ21 = 3.846, P = 0.050). The nutrition enhancement treatment had a positive effect on fetus and neonate survival through about the first month postpartum, at which point the treatment stopped having an effect (Figure 4). Fetus-neonate survival through 15 July, 2002–2004, was much higher (χ21 = 6.013, P = 0.014) for treatment fawns (S(t) = 0.746, SE = 0.035) than control fawns (S(t) = 0.583, SE = 0.043). In summary, enhanced nutrition of adult does during winter and early spring caused higher survival of fetuses and fawns, resulting in higher December fawn recruitment (Figure 4). 51 �2001−2004 Fawn:Doe Ratio Summary: Our results from 2001 and 2002 emphasize the inherent difficulties and biases associated with precisely measuring fawn:doe ratios, particularly in this research study. Ratios obtained from helicopter surveys were based on 2 short-duration flights/unit/year over spatially small units. Helicopter surveys were complicated by high deer densities in heavy cover, making both deer detection and fawn:doe classifications a considerable challenge. There were a variety of potential biases that may have affected the helicopter surveys, including differential sightability of does and fawns, double classification of some deer, and incorrect classification of yearling bucks with small antlers. Ground fawn:doe ratio observations of radio-collared doe groups were made using spotting scopes and field glasses, where we commonly studied the deer for some time. Incorrect classifications during these surveys were likely minimal. For example, small-antlered yearling bucks (e.g. 3 – 6” spikes) were detected from the ground, whereas they were undoubtedly missed on occasion during helicopter surveys. We also obtained repeated observations for some of the radio-collared doe groups from the ground. The main potential bias affecting ground fawn:doe classifications was how observations were made. Many of the ground classifications in the Shavano Valley experimental unit were made by radiotracking does during the day. On the other hand, a majority of ground classifications in the Colona experimental unit were based on observing deer groups as they entered openings to feed during the late afternoon. Our age ratio results were more consistent with survival data during 2003 and 2004. Deer were not as concentrated during helicopter surveys, and unlike previous years, a majority of the ground classification data for the Colona experimental unit was obtained by radio-tracking does during the day rather than sitting and waiting for deer to emerge from pinyon-juniper hillsides to feed on sagebrush-grass benches. We relied primarily on fetus-neonate survival data to make inferences regarding treatment effects because of the inherent difficulties measuring fawn:doe ratios in the 2 experimental units. However, we plan to compare observed helicopter and ground fawn:doe ratios with predicted ratios based on fetusneonate survival data as a technique assessment of fawn:doe ratio measurements. This analysis will be incorporated into the job completion report. Neonate Mortality Causes 2002−2003: During June − December of 2002 and 2003, 37 treatment fetuses-neonates died: 3 – stillborn, 8 – coyote predation, 2 – bear predation, 2 – felid predation, 3 – predation where the predator was undetermined, 11 – disease-starvation-malnutrition, 1 – abandonment, 3 – trauma-injury, 2 – unknown, and 2 – poached. The two poached fawns were censored from analyses evaluating the effect of the treatment. Converted to mortality rates based on the Kaplan-Meier survival analysis, 11.4% of all treatment fawns died from disease-starvation-malnutrition, 8.3% from coyote predation, 7.7% were stillborn, 3.1% died each from injury-trauma and from predation where the predator was undetermined, 2.1% each from bear predation, felid predation, and unknown causes, and 1.0% from abandonment. Simplified, 15.6% of all treatment fawns died from predation, 11.4% died from disease-starvationmalnutrition, 7.7% were stillborn, and 6.2% died from other or unknown causes. During June – December of 2002 and 2003, 38 control fetuses-neonates died: 2 – stillborn, 12 – coyote predation, 4 – felid predation, 2 – bear predation, 1 – predation where the predator was undetermined, 12 – diseasestarvation-malnutrition, 1 – trauma-injury, and 4 – unknown. Converted to mortality rates based on the Kaplan-Meier survival analysis, 16.0% of all control fawns died from disease-starvation-malnutrition, 16.0% died from coyote predation, 5.3% each from felid predation and unknown causes, 4.9% were stillborn, 2.7% from bear predation, and 1.3% each from trauma-injury and predation where the predator was undetermined. Simplified, 25.3% of all control fawns died from predation, 16.0% from diseasestarvation-malnutrition, 6.7% from other or unknown causes, and 4.9% were stillborn. In summary, mortality rates due to predation and disease-starvation-malnutrition were lower for treatment fawns than control fawns. 52 �2004: During June – December, 2004, 36 treatment neonates died: 0 – stillborn, 13 – coyote or dog predation, 7 – bear predation, 3 – felid predation, 5 – predation where the predator was undetermined, 2 – disease-starvation-malnutrition, 1 – trauma-injury, and 5 – unknown. Converted to mortality rates based on the Kaplan-Meier survival analysis, 20.3% of all treatment fawns died from canid predation, 10.9% died from bear predation, 7.8% each from unknown causes and from predation where the predator was undetermined, 4.7% from felid predation, 3.1% from disease-starvation-malnutrition, and 1.6% from injury-trauma. Simplified, 43.7% of all treatment fawns died from predation, 9.4% died from other or unknown causes, and 3.1% died from disease-starvation-malnutrition. During June – December, 2004, 32 control fetuses-neonates died: 6 – stillborn, 5 – coyote predation, 4 – bear predation, 1 – felid predation, 2 – predation where the predator was undetermined, 4 – disease-starvation-malnutrition, 5 – injurytrauma, and 5 – unknown. We actually observed 9 stillborns from control does with fetus counts, although only 6 were associated with does in which all fetuses were accounted at parturition. Thus, we used only 6 of the stillborns in our estimate of fetus survival, and therefore stillborn mortality. Converted to mortality rates based on the Kaplan-Meier survival analysis, 24.0% of all control neonates were stillborn, 8.8% died each from coyote predation, injury-trauma, and unknown causes, 7.0% each from bear predation and disease-starvation-malnutrition, 3.5% from predation where the predator was undetermined, and 1.8% from felid predation. Simplified, 24.0% of all control fawns were stillborn, 21.0% died from predation, 17.5% died from other or unknown causes, and 7.0% died from diseasestarvation-malnutrition. Mortality causes were much different during 2004 than either 2002 or 2003. Predation rates were high on treatment fawns while stillborn mortality rates were high among control fawns. Several specific observations during 2004 are worthy of note. Three of the treatment fawn mortalities attributed to coyotes or dogs occurred amongst large herds of sheep which had been released to pasture immediately prior to the mortality events. Bear predation was higher among all fawns during 2004, although 3 of the 7 treatment bear mortalities involved triplets that were killed simultaneously by a bear 1−2 days after the fawns were born. Six treatment fawns captured in the same drainage tributary were killed within a 1-mi2 area; the drainage was in a portion of the study area where no control fawns were captured. However, a single animal did not kill each of the fawns because the mortalities encompassed coyote, felid, and bear predation. Finally, we observed more accidental deaths than typical among control fawns. One control fawn drowned in a river, another fell, one became lodged in a water-filled mudhole, and both a control fawn and a treatment fawn died from injuries sustained while stuck in a woven wire fence. 2002−2004 Summary: Combining all years of data, the survival and cause-specific mortality rates of treatment fawns were: 52.8% survived, 27.2% died from predation (i.e. 13.3% canid, 5.7% bear, 3.2% felid, 5.1% undetermined), 8.2% died from disease-starvation-malnutrition, 4.2% were stillborn, and 7.6% died from other or unknown causes. Survival and cause-specific mortality rates of control fawns were: 40.1% survived, 24.3% died from predation (i.e. 12.9% canid, 4.6% bear, 3.8% felid, 3.0% undetermined), 12.1% died from disease-starvation-malnutrition, 12.1% were stillborn, and 11.4% died from other or unknown causes. The relatively high predation rate of treatment fawns was largely explained by 2004 data alone. As a general summary, control fawns suffered higher rates of disease, illness, malnutrition, and stillborn mortality (i.e. non-predator related mortalities) than did treatment fawns, which explains why survival was higher among treatment fawns (Figure 5). Overwinter Fawn Survival and Mortality Causes During winter 2001−02 (10 Dec 2001–15 Jun 2002), the survival rate of fawns was higher (χ21 = 13.216, P < 0.001) in the treatment unit (S(t) = 0.865, SE = 0.056) than in the control unit (S(t) = 0.510, SE = 0.080). Similarly, in 2002−03 (10 Dec 2002–15 June 2003), the overwinter survival rate of fawns was higher (χ21 = 5.734, P = 0.017) in the treatment unit (S(t) = 0.900, SE = 0.047) than in the control unit (S(t) = 0.691, SE = 0.074). Again in 2003−04 (10 Dec 2003–15 June 2004), the overwinter survival 53 �rate of fawns was higher (χ21 = 3.852, P = 0.050) in the treatment unit (S(t) = 0.920, SE = 0.045) than in the control unit (S(t) = 0.756, SE = 0.067). Combining survival data across all 3 winters, treatment fawn survival (S(t) = 0.895, SE = 0.029) was 0.24 higher (χ21 = 18.781, P < 0.001) than control fawn survival (S(t) = 0.655, SE = 0.044) (Figure 6). The treatment unit during winter 2001−02 became the control unit during winters 2002−03 and 2003−04, and vice versa. Thus, the overwinter survival treatment effect was replicated across each experimental unit. Fawn survival also varied as a function of early winter fawn mass (χ21 = 21.19, P < 0.001). Surviving fawns averaged 3.5 kg heavier than fawns that died. The importance of early winter fawn mass as a predictor of overwinter survival has been documented previously (White et al. 1987, Bishop 1998, White and Bartmann 1998, Unsworth et al. 1999). Early winter mass of treatment fawns ( x = 34.2 kg, SE = 0.418) was similar to control fawns ( x = 34.4, SE = 0.423); thus the effect of the treatment was not confounded with pre-treatment fawn mass. It follows that fawns born from treatment does did not arrive to winter heavier than fawns born from control does, which was not necessarily surprising considering the treatment primarily effected neonate survival through about 1 month postpartum. In summary, the nutrition enhancement treatment improved overwinter fawn survival, and heavier fawns in each experimental unit had higher survival probabilities. During winters 2001−04, 12 of 115 treatment fawns died: 5 from coyote predation, 3 from disease/illness, 2 from malnutrition, 1 from trauma-injury, and 1 unknown. Each of the 3 fawns that died from disease had adequate fat stores. At least one of these fawns died as a result of pneumonia. Converted to mortality rates based on the Kaplan-Meier survival analysis, 4.3% of all treatment fawns died from coyote predation, 2.6% from disease-illness, 1.7% from malnutrition, 0.9% from trauma-injury, and 0.9% from unknown causes. Simplified, 4.3% of all treatment fawns died from predation, 4.3% from disease-malnutrition, and 1.8% from other or unknown causes (Figure 7). During winters 2001−04, 41 of 120 control fawns died: 13 from coyote predation, 8 from mountain lion predation, 8 from malnutrition, 6 from unknown causes, 3 from predation where the predator was undetermined, 2 were road-killed, and 1 from trauma-injury. Converted to mortality rates based on the Kaplan-Meier survival analysis, 10.9% of all control fawns died from coyote predation, 6.7% from mountain lion predation, 6.7% from malnutrition, 5.0% from unknown causes, 2.5% from predation where the predator was undetermined, 1.7% from road-kill, and 0.8% from trauma-injury. Simplified, 20.1% of all control fawns died from predation, 6.7% from malnutrition, and 7.5% from other or unknown causes (Figure 7). Most fawns killed by predators had little or no femur marrow fat remaining, indicating the predation was likely compensatory in nature. Fetus-Neonate-Overwinter Fawn Survival We combined the preceding survival data into a single analysis to express the effect of the treatment across all stages of fawn production and survival. Using a staggered entry survival process with data combined over years, we estimated fawn survival from the fetus stage until one year of age, when fawns were recruited to the yearling (adult) age class (Figure 8). Survival of treatment fetuses to the yearling age class (S(t) = 0.458, SE = 0.031) was 0.18 higher (χ21 = 13.20, P < 0.001) than survival of control fetuses to the yearling age class (S(t) = 0.276, SE = 0.026). Adult Female Survival and Causes of Mortality During winter 2000−01 (1 Dec 2000–31 May 2001), the adult doe survival rate in the treatment unit (S(t) = 0.968, SE = 0.032) was greater (χ21 = 2.649, P = 0.104) than the survival rate in the control unit (S(t) = 0.861, SE = 0.058). However, annual adult doe survival rates (1 Dec 2000–30 Nov 2001) were similar among treatment and control deer (Trt: S(t) = 0.839, SE = 0.066; Control: S(t) = 0.833, SE = 0.062; χ21 = 0.004, P = 0.947). We observed a similar result the following year. The 2001−02 overwinter adult doe survival rate in the treatment unit (S(t) = 0.942, SE = 0.030) was greater (χ21 = 3.116, P = 0.078) than survival in the control unit (S(t) = 0.848, SE = 0.044), yet annual adult doe survival was similar among treatment and control deer (Trt: S(t) = 0.824, SE = 0.049; Control: S(t) = 0.818, SE = 54 �0.047; χ21 = 0.090, P = 0.764). Thus, mortalities of control deer occurred primarily during the winter months, while treatment does died primarily during the summer and fall months. During winter 2002−03, following the treatment cross-over, overwinter adult doe survival rates were similar among treatment and control deer (Trt: S(t) = 0.945, SE = 0.024; Control: S(t) = 0.924, SE = 0.028; χ21 = 0.360, P = 0.549). However, annual adult doe survival rates (1 Dec 2002–30 Nov 2003) were higher (χ21 = 2.016, P = 0.156) for treatment does (S(t) = 0.888, SE = 0.034) than control does (S(t) = 0.813, SE = 0.041). The main difference from the previous 2 years was that overwinter survival of adult does in the Shavano experimental unit increased in 2002−03 upon receiving the treatment. Summer-fall survival was similar in that Colona adult does had higher mortality rates than Shavano adult does. Thus, in 2002−03, there was no difference between survival rates of treatment and control adult does during winter but there was evidence of higher annual survival of treatment adult does. During winter 2003−04, overwinter adult doe survival rates were higher (χ21 = 3.843, P = 0.050) among treatment does (S(t) = 0.979, SE = 0.014) than control does (S(t) = 0.915, SE = 0.027). The annual adult doe survival rate (1 Dec 2003–30 Nov 2004) was 0.895 (SE = 0.030) for treatment does and 0.832 (SE = 0.036) for control does, which was marginally different (χ21 = 1.562, P = 0.211). Considering all years, the treatment improved overwinter adult doe survival but had a relatively minor affect on annual survival. Considering only the past 2 years, the treatment had a positive affect on annual survival. Annual survival rates measured in this study align reasonably well with expected survival based on other studies (Unsworth et al. 1999, Bishop et al. 2005, B. E. Watkins, Colorado Division of Wildlife, unpublished data). During 2000−02, when the Colona experimental unit received the treatment and the Shavano experimental unit was the control, 16 treatment and 16 control does died. The 16 treatment does died from the following categories: 4 – road-killed, 3 – while giving birth, 3 – predation (undetermined predator), 2 – non-predation unknown (intact carcasses with no evidence of predation or scavenging), 1 – disease (chronic arthritis), 1 – mountain lion predation, and 2 – unknown. Predation was not a major mortality factor for treatment does, and a majority of mortalities were independent of nutrition (does were in good condition). The 16 control doe mortalities included the following causes: 5 – mountain lion predation, 3 – malnutrition, 2 – non-predation unknown, 1 – road-killed, 1 – bear predation, 1 – fence injury, 1 – legal harvest, and 2 – unknown. Predation and malnutrition were the major mortality causes of control deer. Interestingly, during this 2-year period, we did not document any coyote predation on adult does. During 2002–04, with Shavano as the treatment and Colona as the control, there were 20 treatment doe mortalities: 6 – disease/infection, 3 – coyote predation, 1 – road-killed, 1 – broken jaw which led to starvation, 1 – fence injury, 1 poached, and 7 unknown causes. As we saw during 2000-02, predation was not a major mortality factor for treatment does, and a majority of mortalities were independent of nutrition. We observed 33 control adult doe mortalities during the same time period: 8 – road-kill, 7 – malnutrition-disease, 5 – coyote predation, 3 – mountain lion predation, 3 – non-predation unknown, 1 – bear predation, 1 – predation where the predator was undetermined, and 5 – unknown causes. Road kill, malnutrition-disease, and predation were the major mortality factors of control does during 2002−04. Road kill was a significant mortality factor of Colona adult does but not Shavano adult does, which partially explains why we failed to see a treatment effect during 2000−02 but did see one during 2002−04. If road-killed deer were censored, greater evidence would exist for a treatment effect during 2000−02 while there would be less evidence of a treatment effect during 2002−04. However, road-kill had minimal effect on the overall 4-year interpretation of the treatment effect on adult doe survival 55 �because of the cross-over design. Ignoring road kill, treatment does tended to die of causes unrelated to nutrition whereas control does were more susceptible to malnutrition and predation. Population Growth Rate The finite rate of population increase, λ, based on our measurements of treatment population parameters was 1.20 (Table 2), which would cause the deer population to double in approximately 4 years. The finite rate of increase calculated from control deer was 1.04 (Table 2), indicating a stable or slightly increasing population. The nutrition enhancement treatment therefore had a dramatic effect on deer population performance, indicating habitat quality was ultimately limiting the population. SUMMARY We successfully enhanced nutrition of deer occupying the treatment units based on our body fat estimates of treatment and control does. Pregnancy and fetus rates were similar among treatment and control does. The treatment caused an increase in both fetus-neonate survival and overwinter fawn survival, resulting in higher yearling recruitment. Overwinter adult doe survival increased as a result of the treatment, but annual survival was more similar among treatment and control adult does. Combining all parameter estimates into a deterministic population model, the treatment population indicated an exceptionally high rate of increase (λ = 1.20) while the control population (λ = 1.04) was indicative of the overall Uncompahgre deer population during 2000−2004. The nutrition enhancement treatment was artificial in the sense that we applied it only to test whether habitat quality was ultimately more limiting than predation or other factors. Our results to do not provide support for managing deer populations with nutrition supplements because our treatment delivery approach could not be applied to a large number of animals over a large area. Rather, our results provide a foundation for focusing deer management efforts on improving habitat quality in western Colorado pinyon-juniper ecosystems with corresponding research efforts to quantify the effects of habitat manipulations on deer. LITERATURE CITED ANDELT, W. F., T. M. POJAR, AND L. W. JOHNSON. 2004. Long-term trends in mule deer pregnancy and fetal rates in Colorado. Journal of Wildlife Management 68:542−549. BAKER, D. L., AND N. T. HOBBS. 1985. Emergency feeding of mule deer during winter: tests of a supplemental ration. Journal of Wildlife Management 49:934−942. _____, D. E. JOHNSON, L. H. CARPENTER, O.C. WALLMO, AND R. B. GILL. 1979. Energy requirements of mule deer fawns in winter. Journal of Wildlife Management 43:162−169. _____, G. W. STOUT, AND M. W. MILLER. 1998. A diet supplement for captive wild ruminants. Journal of Zoo and Wildlife Medicine 29:150−156. BALLARD, W. B., D. LUTZ, T. W. KEEGAN, L. H. CARPENTER, AND J. C. DEVOS, JR. 2001. Deer-predator relationships: a review of recent North American studies with emphasis on mule and black-tailed deer. Wildlife Society Bulletin 29:99−115. BARRETT, M. W., J. W. NOLAN, AND L. D. ROY. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108−114. BISHOP, C. J. 1998. Mule deer fawn mortality and habitat use, and the nutritional quality of bitterbrush and cheatgrass in southwest Idaho. Thesis, University of Idaho, Moscow, USA. _____, D. J. FREDDY, AND G. C. WHITE. 2002. Effects of enhanced nutrition of adult female mule deer on fetal and neonatal survival rates: a pilot study to address feasibility. Colorado Division of Wildlife, Wildlife Research Report, Federal Aid in Wildlife Restoration Project W−153−R, Job Final Report. Fort Collins, USA. 56 �_____, J. W. UNSWORTH, AND E. O. GARTON. 2005. Mule deer survival among adjacent populations in southwest Idaho. Journal of Wildlife Management 69:311−321. BOWDEN, D. C. 1993. A simple technique for estimating population size. Department of Statistics, Colorado State University, Fort Collins, USA. _____, A. E. ANDERSON, AND D. E. MEDIN. 1984. Sampling plans for mule deer sex and age ratios. Journal of Wildlife Management 48:500−509. CONNOLLY, G. E. 1981. Limiting factors and population regulation. Pages 245−285 in O. C. Wallmo, editor. Mule and black-tailed deer of North America. University of Nebraska Press, Lincoln, USA. COOK, R. C. 2000. Studies of body condition and reproductive physiology in Rocky Mountain elk. Thesis, University of Idaho, Moscow, USA. _____, AND J. G. COOK. 2002. An informal training guide to condition evaluation in elk and deer. National Council for Air and Stream Improvement, Unpublished Report. _____, _____, D. L. MURRAY, P. ZAGER, B. K. JOHNSON, AND M. W. GRATSON. 2001. Development of predictive models of nutritional condition for Rocky Mountain elk. Journal of Wildlife Management 65:973−987. GILL, R. B., T. D. I. BECK, C. J. BISHOP, D. J. FREDDY, N. T. HOBBS, R. H. KAHN, M. W. MILLER, T. M. POJAR, AND G. C. WHITE. 2001. Declining mule deer populations in Colorado: reasons and responses. Colorado Division of Wildlife Special Report Number 77. Denver, USA. HOLTER, J. B., H. H. HAYES, AND S. H. SMITH. 1979. Protein requirement of yearling white-tailed deer. Journal of Wildlife Management 43:872−879. KAPLAN, E. L., AND P. MEIER. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 53:457−481. NEAL, A. K., G. C. WHITE, R. B. GILL, D. F. REED, AND J. H. OLTERMAN. 1993. Evaluation of markresight model assumptions for estimating mountain sheep numbers. Journal of Wildlife Management 57:436−450. POJAR, T. M., AND D. C. BOWDEN. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550−560. POLLOCK, K. H., S. R. WINTERSTEIN, C. M. BUNCK, AND P. D. CURTIS. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7−15. RAMSEY, C. W. 1968. A drop-net deer trap. Journal of Wildlife Management 32:187−190. SAS INSTITUTE. 1989. SAS/STAT® user’s guide, version 6, fourth edition. Volume 2. SAS Institute, Cary, North Carolina, USA. _____. 1997. SAS/STAT® Software: Changes and Enhancements through Release 6.12. SAS Institute, Cary, North Carolina, USA. SCHMIDT, R. L., W. H. RUTHERFORD, AND F. M. BODENHAM. 1978. Colorado bighorn sheep-trapping techniques. Wildlife Society Bulletin 6:159−163. SMITH, S. H., J. B. HOLTER, H. H. HAYES, AND H. SILVER. 1975. Protein requirement of white-tailed deer fawns. Journal of Wildlife Management 39:582−589. STEPHENSON, T. R., V. C. BLEICH, B. M. PIERCE, AND G. P. MULCAHY. 2002. Validation of mule deer body composition using in vivo and post-mortem indices of nutritional condition. Wildlife Society Bulletin 30:557−564. _____, K. J. HUNDERTMARK, C. G. SCHWARTZ, AND V. VAN BALLENBERGHE. 1998. Predicting body fat and body mass in moose with ultrasonography. Canadian Journal of Zoology 76:717−722. _____, J. W. TESTA, G. P. ADAMS, R. G. SASSER, C. G. SCHWARTZ, AND K. J. HUNDERTMARK. 1995. Diagnosis of pregnancy and twinning in moose by ultrasonography and serum assay. Alces 31:167−172. THOMPSON, C. B., J. B. HOLTER, H. H. HAYES, H. SILVER, AND W. E. URBAN, JR. 1973. Nutrition of white-tailed deer. I. Energy requirements of fawns. Journal of Wildlife Management 37:301−311. 57 �ULLREY, D. E., W. G. YOUATT, H. E. JOHNSON, L. D. FAY, AND B. L. BRADLEY. 1967. Protein requirement of white-tailed deer fawns. Journal of Wildlife Management 31:679−685. UNSWORTH, J. W., D. F. PAC, G. C. WHITE, AND R. M. BARTMANN. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315−326. VAN REENEN, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408−421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. VERME, L. J., AND D. E. ULLREY. 1972. Feeding and nutrition of deer. Pages 275−291 in D. C. Church, editor. Digestive physiology and nutrition of ruminants. Volume 3 – Practical Nutrition. D. C. Church, Corvallis, Oregon, USA. WATKINS, B. E., D. E. ULLREY, R. F. NACHREINER, AND S. M. SCHMITT. 1983. Effects of supplemental iodine and season on thyroid activity of white-tailed deer. Journal of Wildlife Management 47:45−58. _____, J. H. WITHAM, D. E. ULLREY, D. J. WATKINS, AND J. M. JONES. 1991. Body composition and condition evaluation of white-tailed deer fawns. Journal of Wildlife Management 55:39−51. WHITE, G. C. 1996. NOREMARK: Population estimation from mark-resighting surveys. Wildlife Society Bulletin 24:50−52. _____, AND R. M. BARTMANN. 1998. Effect of density reduction on overwinter survival of free-ranging mule deer fawns. Journal of Wildlife Management 62:214−225. _____, R. A. GARROTT, R. M. BARTMANN, L. H. CARPENTER, AND A. W. ALLDREDGE. 1987. Survival of mule deer in northwest Colorado. Journal of Wildlife Management 51:852−859. _____, A. F. REEVE, F. G. LINDZEY, AND K. P. BURNHAM. 1996. Estimation of mule deer winter mortality from age ratios. Journal of Wildlife Management 60:37−44. Prepared by _______________________ Chad J. Bishop, Wildlife Researcher 58 �Table 1. Total thyroxine (T4) and total tri-iodothyronine (T3) concentrations (nmol/l), and free T4 (FT4) and free T3 (FT3) concentrations (pmol/l), measured during late February in adult female mule deer occupying a nutrition enhancement treatment unit and a control unit on the Uncompahgre Plateau in southwest Colorado, 2003−04. Thyroid Hormone T3 (SE) FT3 (SE) 146.6 (3.53) FT4 (SE) 30.0 (1.27) 1.65 (0.058) 4.10 (0.130) Control 92.3 (3.56) 17.1 (0.65) 1.42 (0.080) 3.71 (0.210) Treatment 131.9 (4.48) 24.8 (1.39) 2.08 (0.075) 4.21 (0.154) Control 90.0 (3.54) 12.5 (0.59) 1.70 (0.104) 3.60 (0.188) Year Exp. Unit T4 (SE) 2003 Treatment 2004 Table 2. Population parameter estimates and population finite rate of increase, λ, for treatment deer that received a nutrition enhancement and control deer that accessed existing habitat only, southwest Colorado, 2002−04. Population Parameter Treatment Control 0.937 0.937 Adult doe fetus rate 1.84 1.84 Fetus survival to birth 0.958 0.879 Neonate survival to December 0.551 0.456 Overwinter fawn survival to June 0.895 0.655 Annual adult doe survival 0.860 0.824 Finite Rate of Increase, λ 1.20 1.04 Adult doe pregnancy ratea a a We used overall estimates of pregnancy and fetus rates because we did not detect meaningful differences between treatment and control deer. Year Unit A Unit B 2000-01 Treatment Control 2001-02 Treatment Control 2002-03 Control Treatment 2003-04 Control Treatment Figure 1. Schematic representation of experimental units and nutrition enhancement treatment allocation. Units A and B were located in winter range habitat on the Uncompahgre Plateau in southwest Colorado. The nutrition enhancement cross-over design encompassed 4 years. 59 �Uncompahgre Plateau Mesa County GRAND JUNCTION Delta County m co Un GMU 62 r hg pa e u ea at Pl Montrose County GMU 61 Sanmiguel County Gunnison County DELTA Shavano E.U. Winter Range MONTROSE Colona Montrose County E.U. Summ er Ouray County Figure 2. Location of Colona and Shavano (Units A and B) experimental units on the Uncompahgre Plateau, southwest Colorado; and location of the summer range study area encompassing the southern Uncompahgre Plateau and adjacent San Juan Mountains. 60 �Hwy 550 Uncompahgre Valley Colona Exp. Unit Shavano Valley Shavano Exp. Unit - 6 Miles - - Figure 3. Colona and Shavano experimental units (Units A and B), located in Game Management Unit 62 on the Uncompahgre Plateau, southwest Colorado. 61 �1 \ 0.9 Treatment 0.8 Control 0.7 0.6 -- - 0.5 0.4 0.3 3/1 3/31 4/30 5/30 6/29 7/29 8/28 9/27 10/27 11/26 Figure 4. Survival (1 Mar –15 Dec, 2002–2004) of mule deer fetuses-neonates born from adult does receiving enhanced nutrition during winter (Treatment, S(t) = 0.528, SE = 0.027) and from adult does accessing existing winter habitat only (Control, S(t) = 0.401, SE = 0.025), southwest Colorado. 0.6 Treatment Control 0.5 0.4 0.3 0.2 0.1 0 Survived Predation Illness/Malnutrition Stillborn Other/ Unknow n Figure 5. Survival and cause-specific mortality rates (1 Mar –15 Dec, 2002–2004) of mule deer fetusesneonates born from adult does receiving enhanced nutrition during winter (Treatment) and from adult does accessing existing winter habitat only (Control), southwest Colorado. 62 �- ..... 1 0 .9 0 .8 0 .7 Treatment 0 .6 Control 0 .5 12/1 12/16 12/31 1/15 1/30 2/14 3/1 3/16 3/31 4/15 4/30 5/15 5/30 6/14 Figure 6. Overwinter fawn survival (10 Dec –15 Jun, 2001–2004) in a nutrition enhancement treatment unit (S(t) = 0.895, SE = 0.029) and a control unit (S(t) = 0.655, SE = 0.044) on the Uncompahgre Plateau, southwest Colorado. 0.9 Treatment Control 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Survived Predation Illness/Malnutrition Other/ Unknow n Figure 7. Overwinter fawn survival and cause-specific mortality rates (10 Dec–15 Jun, 2001–2004) in a nutrition enhancement treatment unit and a control unit on the Uncompahgre Plateau, southwest Colorado. 63 �1 Treatment 0.9 Control 0.8 0.7 0.6 0.5 0.4 0.3 0.2 3/1 4/15 5/30 7/14 8/28 10/12 11/26 1/10 2/24 4/10 5/25 Figure 8. Fawn survival from fetus stage (March) to 1 year of age (June of the following year) for deer receiving enhanced nutrition during winter (Treatment, S(t) = 0.458, SE = 0.031) and deer accessing existing winter habitat only (Control, S(t) = 0.276, SE = 0.026), southwest Colorado, 2002−2004. 64 �APPENDIX I We submitted the following manuscript (referenced here by Abstract) to the Journal of Wildlife Management during summer 2005. USING VAGINAL TRANSMITTERS TO CAPTURE NEONATES FROM MARKED MULE DEER CHAD J. BISHOP, DAVID J. FREDDY, GARY C. WHITE, BRUCE E. WATKINS, THOMAS R. STEPHENSON, AND LISA L. WOLFE ABSTRACT Measuring reproductive success of previously-marked, adult female ungulates enables study of certain complex ecological factors limiting populations. We evaluated the effectiveness of using vaginal implant transmitters (VITs, n = 154) in mule deer (Odocoileus hemionus) combined with repeated relocations of other radio-collared deer for capturing effective samples of neonates (e.g. >100/year) from free-ranging, marked females. We also evaluated the effectiveness of VITs, when used in conjunction with in utero fetus counts, for obtaining direct estimates of fetus survival. During 2003 and 2004, when VIT batteries were placed on a 12-hour duty cycle to lower failure rates, the proportion of VITs that shed ≤3 days prepartum or during parturition was 0.623 (SE = 0.0456), and the proportion shed during parturition was 0.447 (SE = 0.0468). Our neonate capture success rate was 0.880 (SE = 0.0359) from does with VITs shed ≤3 days prepartum or during parturition and 0.307 (SE = 0.0235) from radiocollared does without VITs or whose implants failed to function properly. Combining techniques we captured 275 neonates and 21 stillborns during 2002−2004. We accounted for all fetuses at birth (i.e. live or stillborn) from 78 of the 147 does (0.531, SE = 0.0413) with winter fetus counts, which was heavily dependent on VIT retention success. Deer that shed VITs prepartum were larger and older than deer that retained implants to parturition, indicating a need to develop variable-sized VITs which may be individually fitted to deer in the field. We demonstrated that direct estimates of fetus and neonate survival may be obtained from previously-marked female mule deer in free-ranging populations, thus expanding opportunities for conducting field experiments. Resulting neonate survival estimates lacked bias that is typically associated with other neonate capture techniques. However, current vaginal implant failure rates and overall expense limit applicability of the technique to well-funded studies with adequate personnel. 65 �Colorado Division of Wildlife July 2005 − June 2006 WILDLIFE RESEARCH REPORT State of Cost Center Work Package Task No. Colorado 3430 3001 4 Federal Aid Project: W-185-R : Division of Wildlife : Mammals Research : Deer Conservation : Effect of Nutrition and Habitat Enhancements on Mule Deer Recruitment and Survival Rates : Period Covered: July 1, 2005 − June 30, 2006 Authors: C. J. Bishop, G. C. White, D. J. Freddy, and B. E. Watkins All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We measured mule deer (Odocoileus hemionus) population parameters in response to a nutrition enhancement treatment to evaluate the relative importance of habitat quality as a limiting factor of mule deer in western Colorado during November 2000 – January 2005. We conducted preliminary data analyses upon completion of field work. We found strong evidence that enhanced nutrition increased fawn recruitment to the yearling age class. During 2002−2004, fetus-neonate survival from 1 March−15 December was higher (χ21 = 3.846, P = 0.050) for treatment fawns (S(t) = 0.528, SE = 0.027) than control fawns (S(t) = 0.401, SE = 0.025). During 15 December–15 June, 2001−2004, the overwinter survival rate of fawns was greater (χ21 = 18.781, P < 0.001) in the treatment unit (S(t) = 0.895, SE = 0.029) than in the control unit (S(t) = 0.655, SE = 0.044). Using a staggered entry survival process with data combined over years, survival of treatment fetuses to 1 year of age (S(t) = 0.458, SE = 0.031) was 0.18 higher (χ21 = 13.20, P < 0.001) than survival of control fetuses to 1 year of age (S(t) = 0.276, SE = 0.026). The finite rate of population increase, λ, was 1.20 for treatment deer and 1.04 for control deer. Our preliminary results provided a foundation for focusing deer management efforts on improving habitat quality in western Colorado pinyon-juniper (Pinus edulis-Juniperus osteosperma) ecosystems with corresponding research efforts to quantify the effects of habitat manipulations on deer performance. During the past year, we monitored post-treatment adult doe survival, identified a set of publications to be completed for submission to scientific journals, initiated final data analyses corresponding to the set of publications, and worked on or completed several manuscripts. A manuscript on the effectiveness of vaginal implant transmitters was accepted for publication in the Journal of Wildlife Management, and a manuscript documenting malignant catarrhal fever in the deer population was submitted to the Journal of Wildlife Diseases. The lead investigator also wrote a portion of a book chapter regarding the effects of excessive herbivory on mule deer populations. 59 �WILDLIFE RESEARCH REPORT EFFECT OF NUTRITION AND HABITAT ENHANCEMENTS ON MULE DEER RECRUITMENT AND SURVIVAL RATES CHAD J. BISHOP, GARY C. WHITE, DAVID J. FREDDY, AND BRUCE E. WATKINS P. N. OBJECTIVE To determine experimentally whether enhancing mule deer nutrition during winter and early spring via supplementation increases fetus survival, neonate survival, overwinter fawn survival, or ultimately, population productivity. SEGMENT OBJECTIVES 1. Radio-monitor and measure post-treatment survival of the sample of radio-collared mule deer adult does. 2. Identify a set of publications to be generated from the research. 3. Initiate final data analyses to support preparation of manuscripts. 4. Prepare manuscripts for submission to scientific journals for publication. INTRODUCTION Mule deer (Odocoileus hemionus) numbers apparently declined during the 1990s throughout much of the West, and have clearly decreased since the peak population levels documented during the 1940s−1960s (Unsworth et al. 1999, Gill et al. 2001). Biologists and sportsmen alike have concerns as to what factors may be responsible for declining population trends. Although previous and current research indicates multiple interacting factors are responsible, habitat and predation have typically received the focus of attention. A number of studies have evaluated whether predator control increases deer survival, yet results are highly variable (Connolly 1981, Ballard et al. 2001). Together, predator control studies with adequate rigor and statistical power indicate predation effects on mule deer are variable as a result of time-specific and site-specific factors. Studies which have demonstrated deer population responses to predator control treatments have failed to determine whether predation is ultimately more limiting than habitat when considering long term population changes. Numerous research studies have evaluated mule deer habitat quality, but virtually no studies have documented population responses to habitat improvements. In many areas where declining deer numbers are of concern, predation is common yet habitat quality appears to have declined. The question remains as to whether predation, habitat, or some other factor is more limiting to mule deer in these situations, and whether habitat quality can be improved for the benefit of deer. It may also be that no single factor is responsible for observed deer declines, and a more comprehensive understanding of multi-factor interactions is needed. We designed and implemented a field experiment where we measured deer population responses to a nutrition enhancement treatment to further understand the causative factors underlying observed deer population dynamics. We conducted the study on the Uncompahgre Plateau in southwest Colorado, where several predator species were present in abundant numbers: coyotes (Canis latrans), mountain lions (Felis concolor), and bears (Ursus americanus). In addition to predation, myriad diseases in combination have proximately affected survival of the Uncompahgre deer population (Pojar and Bowden 2004, B. E. Watkins, Colorado Division of Wildlife, unpublished data). Predator numbers were not manipulated in any manner during the course of the study. All factors were left constant with the exception of deer nutrition. Deer nutrition was enhanced by providing supplemental feed to deer occupying a treatment area during winter. We measured December fawn recruitment and overwinter fawn 60 �survival in response to the treatment to determine whether deer nutrition was ultimately more limiting than predation or disease. A second phase of research was initiated in 2005 to quantify deer population parameters in response to manipulations of pinyon-juniper (Pinus edulis-Juniperus osteosperma) habitat (Bergman et al. 2006). The objective of this research is to determine whether habitat can be effectively improved for mule deer by introducing disturbance into late-seral pinyon-juniper stands. STUDY AREA We non-randomly selected two experimental units (A−B) within mule deer winter range on the Uncompahgre Plateau (Figure 1) to facilitate a cross-over experimental design for evaluating the effects of enhanced deer nutrition during winter on annual population performance. Unit A received a nutrition enhancement treatment during the first 2 winters of research (2000 – 2002) while Unit B served as a control unit. During winters 2002−03 and 2003−04, Unit B received the treatment while Unit A served as the control. In late April and May, prior to fawning, deer from the winter range experimental units migrated to summer range. We defined the summer range study area by movements of the radio-collared deer captured on winter range; summer range encompassed >1000 mi2 covering the southern portion of the Uncompahgre Plateau and adjacent San Juan Mountains (Figure 2). Winter range elevations ranged from 1830 m (6000 ft) in Shavano Valley to 2318 m (7600 ft) adjacent to the Dry Creek Rim above Shavano Valley. Winter range habitat was dominated by pinyon-juniper with interspersed sagebrush adjacent to agricultural fields in the Shavano and Uncompahgre Valleys. Summer range elevations occupied by deer ranged from 1891 m (6200 ft) in the Uncompahgre Valley to 3538 m (11,600 ft) in Imogene Basin southwest of Ouray, CO. Summer range habitats were dominated by spruce-subalpine fir (Picea spp.-Abies lasiocarpa), aspen (Populus tremuloides), sagebrush, ponderosa pine (Pinus ponderosa), Gambel oak (Quercus gambelii), and to a lesser extent, pinyon-juniper at lower elevations. Bishop et al. (2005) provide a detailed study area description. METHODS Refer to Bishop et al. (2005) for field methodology employed during 2000−2005. During fiscal year 2005-06, we continued to monitor radio-collared adult female deer occupying the two experimental units. Our primary research efforts were focused on data analysis and the preparation of manuscripts for publication in scientific journals. The lead investigator completed additional coursework in mathematical statistics, data analysis, and animal nutrition. We submitted or intend to submit the following manuscripts for publication: 1. Effect of enhanced nutrition on the population performance of free-ranging mule deer. Journal of Wildlife Management. a. A separate publication may be submitted to Science focused on the documentation of coyote predation as a compensatory mortality factor during winter. 2. Using vaginal implant transmitters to aid in capture of neonates from marked mule deer. Journal of Wildlife Management. 3. Evaluation of overdispersion in survival analyses of neonate mule deer associated with sibling fawns. Journal of Wildlife Management. 4. Evaluation of serum thyroid hormone levels as an indicator of body condition during late winter. Journal of Wildlife Management. 5. Evaluation of mule deer age and sex ratios as a response variable in field research. Journal of Wildlife Management. 6. Bovine viral diarrhea isolation and seroprevalence in a free-ranging mule deer (Odocoileus hemionus) population in southwest Colorado. Journal of Wildlife Diseases. 7. Malignant catarrhal fever associated with ovine herpesvirus-2 in free-ranging mule deer (Odocoileus hemionus) in Colorado. Journal of Wildlife Diseases. 61 �8. Spatial patterns in mortality causes of neonatal mule deer across a land use gradient in southwest Colorado. (This could go to several different journals or be published as an internal CDOW publication.) 9. Disease assessment in a Colorado mule deer population following a decline. Journal of Wildlife Diseases (or internal CDOW publication). RESULTS AND DISCUSSION A comprehensive presentation and discussion of preliminary results was provided by Bishop et al. (2005). These results have not changed and therefore we do not repeat them here. The following manuscript was accepted for publication (Appendix I): Bishop, C. J., D. J. Freddy, G. C. White, B. E. Watkins, T. R. Stephenson, and L. L. Wolfe. Using vaginal implant transmitters to aid in capture of neonates from marked mule deer. Journal of Wildlife Management. The following manuscript was submitted for publication (Appendix II): Schultheiss, P. C., H. Van Campen, T. R. Spraker, C. J. Bishop, L. L. Wolfe, and B. Podell. Malignant catarrhal fever associated with ovine herpesvirus-2 in free-ranging mule deer (Odocoileus hemionus) in Colorado. Journal of Wildlife Diseases. The following book chapter was completed and currently undergoing external peer review: Watkins, B. E., C. J. Bishop, E. J. Bergman, A. Bronson, B. Hale, D. W. Lutz, B. F. Wakeling, and L. C. Carpenter. Habitat guidelines for mule deer: Colorado Plateau Ecoregion. Mule Deer Working Group, Western Association of Fish and Wildlife Agencies. The lead investigator completed the following courses to assist with data analysis and manuscript preparation: mathematical statistics (2), population dynamics, population analysis, wildlife nutrition, and animal metabolism. A data bootstrap analysis in SAS was initiated to quantify the degree of overdispersion in our neonate survival data. Overdispersion represents extra-binomial variation in sample data arising from violations of independence. Functionally, undetected overdispersion will result in overly precise variance estimates, and ultimately, incorrect inference. Our neonate samples were subject to independence violations because all captured siblings were routinely radio-collared and treated as independent sample units. A known fates analysis will be conducted using Program MARK to quantify the effect of the nutrition enhancement treatment on various stages of fawn survival while simultaneously accounting for temporal variation and individual heterogeneity (i.e., fawn weight and hind foot length). Once these analyses are completed, we will write and submit manuscripts accordingly. The remaining manuscripts will then be handled in order of priority. Our anticipated timeline is detailed below. Draft manuscripts completed in FY 06-07: 1. Effect of enhanced nutrition on the population performance of free-ranging mule deer. Journal of Wildlife Management. 2. Evaluation of overdispersion in survival analyses of neonate mule deer associated with sibling fawns. Journal of Wildlife Management. 3. Evaluation of serum thyroid hormone levels as an indicator of body condition during late winter. Journal of Wildlife Management. Draft manuscripts completed in FY 07-08: 62 �1. Evaluation of mule deer age and sex ratios as a response variable in field research. Journal of Wildlife Management. 2. Bovine viral diarrhea isolation and seroprevalence in a free-ranging mule deer (Odocoileus hemionus) population in southwest Colorado. Journal of Wildlife Diseases. 3. Spatial patterns in mortality causes of neonatal mule deer across a land use gradient in southwest Colorado. The final proposed manuscript related to disease assessment will be completed as time allows. SUMMARY Enhanced winter nutrition of free-ranging deer caused an increase in both fetus-neonate survival and overwinter fawn survival, resulting in higher yearling recruitment (Bishop et al. 2005). Overwinter adult doe survival increased as a result of the treatment, but annual survival was more similar among treatment and control adult does. Combining all parameter estimates into a deterministic population model, the treatment population indicated an exceptionally high rate of increase (λ = 1.20) while the control population (λ = 1.04) was indicative of the overall Uncompahgre deer population during 2000−2004. The nutrition enhancement treatment was artificial in the sense that we applied it only to test whether habitat quality was ultimately more limiting than predation or other factors. Our results to do not provide support for managing deer populations with nutrition supplements because our treatment delivery approach could not be applied to a large number of animals over a large area. Rather, our results provide a foundation for focusing deer management efforts on improving habitat quality in western Colorado pinyon-juniper ecosystems with corresponding research efforts to quantify the effects of habitat manipulations on deer. We are presently in the process of conducting final data analyses and preparing and submitting manuscripts for publication in scientific journals. LITERATURE CITED BALLARD, W. B., D. LUTZ, T. W. KEEGAN, L. H. CARPENTER, AND J. C. DEVOS, JR. 2001. Deer-predator relationships: a review of recent North American studies with emphasis on mule and black-tailed deer. Wildlife Society Bulletin 29:99−115. BERGMAN, E. J., C. J. BISHOP, D. J. FREDDY, AND G. C. WHITE. 2006. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Federal Aid in Wildlife Restoration Project W−185−R, Job Progress Report. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, USA. BISHOP, C. J., G. C. WHITE, D. J. FREDDY, AND B. E. WATKINS. 2005. Effect of nutrition and habitat enhancements on mule deer recruitment and survival rates. Federal Aid in Wildlife Restoration Project W−185−R, Job Progress Report. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, USA. CONNOLLY, G. E. 1981. Limiting factors and population regulation. Pages 245−285 in O. C. Wallmo, editor. Mule and black-tailed deer of North America. University of Nebraska Press, Lincoln, USA. GILL, R. B., T. D. I. BECK, C. J. BISHOP, D. J. FREDDY, N. T. HOBBS, R. H. KAHN, M. W. MILLER, T. M. POJAR, AND G. C. WHITE. 2001. Declining mule deer populations in Colorado: reasons and responses. Colorado Division of Wildlife Special Report Number 77. Denver, USA. POJAR, T. M., AND D. C. BOWDEN. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550−560. UNSWORTH, J. W., D. F. PAC, G. C. WHITE, AND R. M. BARTMANN. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315−326. Prepared by _______________________ Chad J. Bishop, Wildlife Researcher 63 �Year Unit A Unit B 2000-01 Treatment Control 2001-02 Treatment Control 2002-03 Control Treatment 2003-04 Control Treatment Figure 1. Schematic representation of experimental units and nutrition enhancement treatment allocation. Units A and B were located in winter range habitat on the Uncompahgre Plateau in southwest Colorado. The nutrition enhancement cross-over design encompassed 4 years. Uncompahgre Plateau Mesa County GRAND JUNCTION Delta County m co Un GMU 62 r hg pa e u ea at Pl Montrose County GMU 61 Sanmiguel County Gunnison County DELTA Shavano E. U. Winter Range Exp. Units MONTROSE Colona Montrose County Summer Range E.U. Ouray County Figure 2. Location of Colona and Shavano (Units A and B) experimental units on the Uncompahgre Plateau, southwest Colorado; and location of the summer range study area encompassing the southern Uncompahgre Plateau and adjacent San Juan Mountains. 64 �APPENDIX I The following manuscript (referenced here by Abstract) was accepted for publication by the Journal of Wildlife Management. USING VAGINAL IMPLANT TRANSMITTERS TO AID IN CAPTURE OF NEONATES FROM MARKED MULE DEER CHAD J. BISHOP, DAVID J. FREDDY, GARY C. WHITE, BRUCE E. WATKINS, THOMAS R. STEPHENSON, AND LISA L. WOLFE ABSTRACT Measuring reproductive success of previously marked, adult female ungulates enables study of certain ecological factors that limit populations. We evaluated the feasibility and efficiency of capturing large samples (i.e., >80/year) of neonate mule deer (Odocoileus hemionus) exclusively from free-ranging, marked adult does using vaginal implant transmitters (VITs, n = 154) and repeated locations of radiocollared does without VITs. We also evaluated the effectiveness of VITs, when used in conjunction with in utero fetal counts, for obtaining direct estimates of fetal survival. During 2003 and 2004, after VIT batteries were placed on a 12-hour duty cycle to lower electronic failure rates, the proportion of VITs that shed ≤3 days prepartum or during parturition was 0.623 (SE = 0.0456), and the proportion shed only during parturition was 0.447 (SE = 0.0468). Our neonate capture success rate was 0.880 (SE = 0.0359) from does with VITs shed ≤3 days prepartum or during parturition and 0.307 (SE = 0.0235) from radiocollared does without VITs or whose implants failed to function properly. Using a combination of techniques, we captured 275 neonates and found 21 stillborns during 2002−2004. We accounted for all fetuses at birth (i.e., live or stillborn) from 78 of the 147 does (0.531, SE = 0.0413) having winter fetal counts, and this rate was heavily dependent on VIT retention success. Deer that shed VITs prepartum were larger than deer that retained VITs to parturition, indicating a need to develop variable-sized VITs that may be fitted individually to deer in the field. We demonstrated that direct estimates of fetal and neonatal survival may be obtained from previously marked female mule deer in free-ranging populations, thus expanding opportunities for conducting field experiments. Survival estimates using VITs lacked bias that is typically associated with other neonate capture techniques. However, current vaginal implant failure rates, and overall expense, limit broad applicability of the technique. 65 �APPENDIX II The following manuscript (referenced here by Abstract) was submitted to the Journal of Wildlife Diseases. MALIGNANT CATARRHAL FEVER ASSOCIATED WITH OVINE HERPESVIRUS-2 IN FREERANGING MULE DEER (Odocoileus hemionus) IN COLORADO PATRICIA C. SCHULTHEISS, HANA VAN CAMPEN, TERRY R. SPRAKER, CHAD J. BISHOP, LISA L. WOLFE, AND BRENDAN PODELL ABSTRACT Malignant catarrhal fever (MCF) was diagnosed in 4 free-ranging mule deer (Odocoileus hemionus) in January and February of 2003. Diagnosis was based on typical histologic lesions of lymphocytic vasculitis and PCR identification of ovine herpesvirus-2 (OHV-2) viral genetic sequences in formalin fixed tissues. The animals were from the Uncompahgre Plateau of southwestern Colorado. Deer from these herds occasionally resided in close proximity to domestic sheep (Ovis aries), the reservoir host of OHV-2, in agricultural valleys adjacent to their winter range. These cases indicate that fatal OHV-2 associated MCF can occur in free-ranging mule deer exposed to domestic sheep that overlap their range. 66 �Colorado Division of Wildlife July 2006 − June 2007 WILDLIFE RESEARCH REPORT State of Cost Center Work Package Task No. Colorado 3430 3001 4 Federal Aid Project: W-185-R : Division of Wildlife : Mammals Research : Deer Conservation : Effect of Nutrition and Habitat Enhancements on Mule Deer Recruitment and Survival Rates : Period Covered: July 1, 2006 − June 30, 2007 Authors: C. J. Bishop, G. C. White, D. J. Freddy, and B. E. Watkins All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We measured mule deer (Odocoileus hemionus) population parameters in response to a nutrition enhancement treatment to evaluate the relative importance of habitat quality as a limiting factor of mule deer in western Colorado during November 2000 – January 2005. The nutrition enhancement treatment increased survival of fetuses to the yearling age class by 0.14−0.20 depending on year and fawn sex, although 95% confidence intervals slightly overlapped 0. The nutrition treatment also had a positive effect on annual adult doe survival. Survival of does receiving the treatment (Ŝ = 0.879, SE = 0.0206) was higher than survival of control does (Ŝ = 0.833, SE = 0.0253). Our estimate of the population rate of change, λ̂ , was 1.15−1.17 for treatment deer and 1.02−1.06 for control deer, with some overlap in 95% confidence intervals. The treatment caused λ̂ to increase by 0.139 (95% CI: −0.0152, 0.2941) during 2001−02, 0.113 (95% CI: −0.0009, 0.2279) during 2002−03, and 0.145 (95% CI: 0.0176, 0.2723) during 2003−04.Our results provide a foundation for focusing deer management efforts on improving habitat quality in western Colorado pinyon-juniper (Pinus edulis-Juniperus osteosperma) ecosystems with corresponding research efforts to quantify the effects of habitat manipulations on deer performance. During the past year, we had 2 papers published in peer-reviewed journals, we completed final data analyses, and we prepared 3 other manuscripts for publication. The published manuscripts included: 1) a manuscript on the effectiveness of vaginal implant transmitters (Journal of Wildlife Management 71(3):945−954), and 2) a manuscript documenting malignant catarrhal fever in the Uncompahgre deer population (Journal of Wildlife Diseases 43(3):533−537). We also completed a publication on mule deer habitat guidelines for the Colorado Plateau ecoregion, which will be published by the Western Association of Fish and Wildlife Agencies. The estimated publication date is January 2008. 59 �WILDLIFE RESEARCH REPORT EFFECT OF NUTRITION AND HABITAT ENHANCEMENTS ON MULE DEER RECRUITMENT AND SURVIVAL RATES CHAD J. BISHOP, GARY C. WHITE, DAVID J. FREDDY, AND BRUCE E. WATKINS P. N. OBJECTIVE To determine experimentally whether enhancing mule deer nutrition during winter and early spring via supplementation increases fetal survival, neonatal survival, overwinter fawn survival, or ultimately, population productivity. SEGMENT OBJECTIVES 1. Complete final data analyses to support preparation of manuscripts. 2. Prepare manuscripts for submission to scientific journals for publication. 3. Complete dissertation as part of PhD requirements at Colorado State University INTRODUCTION Mule deer (Odocoileus hemionus) numbers apparently declined during the 1990s throughout much of the West, and have clearly decreased since the peak population levels documented during the 1940s−1960s (Unsworth et al. 1999, Gill et al. 2001). Biologists and sportsmen alike have concerns as to what factors may be responsible for declining population trends. Although previous and current research indicates multiple interacting factors are responsible, habitat and predation have typically received the focus of attention. A number of studies have evaluated whether predator control increases deer survival, yet results are highly variable (Connolly 1981, Ballard et al. 2001). Together, predator control studies with adequate rigor and statistical power indicate predation effects on mule deer are variable as a result of time-specific and site-specific factors. Studies which have demonstrated deer population responses to predator control treatments have failed to determine whether predation is ultimately more limiting than habitat when considering long term population changes. Numerous research studies have evaluated mule deer habitat quality, but virtually no studies have documented population responses to habitat improvements. In many areas where declining deer numbers are of concern, predation is common yet habitat quality appears to have declined. The question remains as to whether predation, habitat, or some other factor is more limiting to mule deer in these situations, and whether habitat quality can be improved for the benefit of deer. It may also be that no single factor is responsible for observed deer declines, and a more comprehensive understanding of multi-factor interactions is needed. We designed and implemented a field experiment where we measured deer population responses to a nutrition enhancement treatment to further understand the causative factors underlying observed deer population dynamics. We conducted the study on the Uncompahgre Plateau in southwest Colorado, where several predator species were present in abundant numbers: coyotes (Canis latrans), mountain lions (Felis concolor), and bears (Ursus americanus). In addition to predation, myriad diseases in combination have proximately affected survival of the Uncompahgre deer population (Pojar and Bowden 2004, B. E. Watkins, Colorado Division of Wildlife, unpublished data). Predator numbers were not manipulated in any manner during the course of the study. All factors were left constant with the exception of deer nutrition. Deer nutrition was enhanced by providing supplemental feed to deer occupying a treatment area during winter. We measured December fawn recruitment and overwinter fawn survival in response to the treatment to determine whether deer nutrition was ultimately more limiting than predation or disease. A second phase of research was initiated in 2005 to quantify deer population 60 �parameters in response to manipulations of pinyon-juniper (Pinus edulis-Juniperus osteosperma) habitat (Bergman et al. 2006). The objective of this research is to determine whether habitat can be effectively improved for mule deer by introducing disturbance into late-seral pinyon-juniper stands. STUDY AREA We non-randomly selected two experimental units (A−B) within mule deer winter range on the Uncompahgre Plateau (Figure 1) to facilitate a cross-over experimental design for evaluating the effects of enhanced deer nutrition during winter on annual population performance. Unit A received a nutrition enhancement treatment during the first 2 winters of research (2000 – 2002) while Unit B served as a control unit. During winters 2002−03 and 2003−04, Unit B received the treatment while Unit A served as the control. In late April and May, prior to fawning, deer from the winter range experimental units migrated to summer range. We defined the summer range study area by movements of the radio-collared deer captured on winter range; summer range encompassed >1000 mi2 covering the southern portion of the Uncompahgre Plateau and adjacent San Juan Mountains (Figure 2). Winter range elevations ranged from 1830 m (6000 ft) in Shavano Valley to 2318 m (7600 ft) adjacent to the Dry Creek Rim above Shavano Valley. Winter range habitat was dominated by pinyon-juniper with interspersed sagebrush adjacent to agricultural fields in the Shavano and Uncompahgre Valleys. Summer range elevations occupied by deer ranged from 1891 m (6200 ft) in the Uncompahgre Valley to 3538 m (11,600 ft) in Imogene Basin southwest of Ouray, CO. Summer range habitats were dominated by spruce-subalpine fir (Picea spp.-Abies lasiocarpa), aspen (Populus tremuloides), sagebrush, ponderosa pine (Pinus ponderosa), Gambel oak (Quercus gambelii), and to a lesser extent, pinyon-juniper at lower elevations. Bishop et al. (2005) provide a detailed study area description. METHODS Refer to Bishop et al. (2005) for field methodology employed during 2000−2005. During fiscal year 2006-07, we had 2 papers published in peer reviewed wildlife journals, which were the result of work completed in the previous year. Our primary research efforts were focused on data analysis and the preparation of manuscripts for publication in scientific journals. We spent much of the year evaluating dependence among deer siblings with respect to fetal and neonatal survival analyses. Essentially all statistical analyses are based on an assumption that sample units are independent. In survival analyses, the assumption pertains to independence of fates. That is, we must assume that the death or survival of one sample unit is not related to the fate of another. Predation and maternal condition are both good examples of mechanisms that could cause sibling neonates to lack independent fates. If a coyote or bear kills twin fawns because both were together, clearly those mortality events were not independent. Similarly, a lack of independence would occur if twin fawns each die of starvation because their dam is in poor condition. Data are considered overdispersed when the independence assumption is violated. Overdispersion does not generally affect point estimates, but rather causes variances to be underestimated. We estimated overdispersion in fetal and neonatal survival datasets and incorporated a data bootstrap procedure into Program MARK (White and Burnham 1999), making it easier for others to conduct similar analyses. The procedure in MARK can be generalized to any situation where multiple individuals are marked from the same litter, clutch, pair, trap site, etc. Once we completed the overdispersion analysis, we spent the remainder of the year conducting final data analyses that quantified the effect of enhanced nutrition on population performance. We then prepared several manuscripts that incorporated the various analyses we conducted. The principal investigator also completed a draft of his PhD dissertation. 61 �RESULTS AND DISCUSSION A comprehensive presentation and discussion of preliminary results was provided by Bishop et al. (2005). These results have not changed and therefore we do not repeat them here. The final results are contained in peer-reviewed manuscripts that have either already been published or will be submitted for publication in 2007 or early 2008. The following manuscripts were published in 2007 (abstracts are provided in Appendix I): Bishop, C. J., D. J. Freddy, G. C. White, B. E. Watkins, T. R. Stephenson, and L. L. Wolfe. 2007. Using vaginal implant transmitters to aid in capture of mule deer neonates. Journal of Wildlife Management 71:945−954. . Schultheiss, P. C., H. Van Campen, T. R. Spraker, C. J. Bishop, L. L. Wolfe, and B. Podell. 2007. Malignant catarrhal fever associated with ovine herpesvirus-2 in free-ranging mule deer in Colorado. Journal of Wildlife Diseases 43:533−537. The following book chapter was completed should be published by January 2008: Watkins, B. E., C. J. Bishop, E. J. Bergman, A. Bronson, B. Hale, B. F. Wakeling, L. H. Carpenter., and D. W. Lutz. 2007. Habitat guidelines for mule deer: Colorado Plateau shrubland and forest ecoregion. Mule Deer Working Group, Western Association of Fish and Wildlife Agencies. The following draft manuscripts were prepared in 2007 and will be submitted for publication in 2007 or early 2008 (abstracts are provided in Appendix II): Bishop, C. J., G. C. White, and P. M. Lukacs. In review. Evaluating dependence among mule deer siblings in fetal and neonatal survival analyses. Journal of Wildlife Management. Bishop, C. J., G. C. White, D. J. Freddy, B. E. Watkins, and T. R. Stephenson. In review. Effect of enhanced nutrition on mule deer population performance. Journal of Wildlife Management OR Wildlife Monographs. Bishop, C. J., B. E. Watkins, L. L. Wolfe, D. J. Freddy, and G. C. White. In review. Evaluating mule deer body condition during late winter using serum thyroid hormone concentrations. Journal of Wildlife Management. The following draft dissertation was prepared in 2007 and submitted for review at Colorado State University (abstract is provided in Appendix III): Bishop, C. J. In review. Effect of enhanced nutrition during winter on the Uncompahgre Plateau mule deer population. Dissertation, Colorado State University, Fort Collins, USA. We intend to pursue several additional manuscripts as time allows, listed below in order of priority. 1. Evaluating dependence of fates among mule deer siblings in Colorado, Idaho, and Montana. Journal of Wildlife Management. 2. Bovine viral diarrhea isolation and seroprevalence in a free-ranging mule deer (Odocoileus hemionus) population in southwest Colorado. Journal of Wildlife Diseases. 3. Spatial patterns in mortality causes of neonatal mule deer across a land use gradient in southwest Colorado. Journal of Wildlife Management. 62 �4. Evaluation of mule deer age and sex ratios as a response variable in field research. Journal of Wildlife Management. SUMMARY Enhanced winter nutrition of free-ranging deer caused an increase in both fetus-neonate survival and overwinter fawn survival, resulting in higher yearling recruitment. Overwinter adult doe survival increased as a result of the treatment, and therefore annual survival was higher among treatment than control adult does. Combining all parameter estimates into a deterministic population model, the treatment population indicated an exceptionally high rate of increase while the control population was stable and indicative of the overall Uncompahgre deer population during 2000−2004. The nutrition enhancement treatment was artificial in the sense that we applied it only to test whether habitat quality was ultimately more limiting than predation or other factors. Our results to do not provide support for managing deer populations with nutrition supplements because our treatment delivery approach could not be applied to a large number of animals over a large area. Rather, our results provide a foundation for focusing deer management efforts on improving habitat quality in western Colorado pinyon-juniper ecosystems with corresponding research efforts to quantify the effects of habitat manipulations on deer. We are presently in the process of conducting final data analyses and preparing and submitting manuscripts for publication in scientific journals. LITERATURE CITED BALLARD, W. B., D. LUTZ, T. W. KEEGAN, L. H. CARPENTER, AND J. C. DEVOS, JR. 2001. Deer-predator relationships: a review of recent North American studies with emphasis on mule and black-tailed deer. Wildlife Society Bulletin 29:99−115. BERGMAN, E. J., C. J. BISHOP, D. J. FREDDY, AND G. C. WHITE. 2006. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Federal Aid in Wildlife Restoration Project W−185−R, Job Progress Report. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, USA. BISHOP, C. J., G. C. WHITE, D. J. FREDDY, AND B. E. WATKINS. 2005. Effect of nutrition and habitat enhancements on mule deer recruitment and survival rates. Federal Aid in Wildlife Restoration Project W−185−R, Job Progress Report. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, USA. CONNOLLY, G. E. 1981. Limiting factors and population regulation. Pages 245−285 in O. C. Wallmo, editor. Mule and black-tailed deer of North America. University of Nebraska Press, Lincoln, USA. GILL, R. B., T. D. I. BECK, C. J. BISHOP, D. J. FREDDY, N. T. HOBBS, R. H. KAHN, M. W. MILLER, T. M. POJAR, AND G. C. WHITE. 2001. Declining mule deer populations in Colorado: reasons and responses. Colorado Division of Wildlife Special Report Number 77. Denver, USA. POJAR, T. M., AND D. C. BOWDEN. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550−560. UNSWORTH, J. W., D. F. PAC, G. C. WHITE, AND R. M. BARTMANN. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315−326. Prepared by _______________________ Chad J. Bishop, Wildlife Researcher 63 �Year Unit A Unit B 2000-01 Treatment Control 2001-02 Treatment Control 2002-03 Control Treatment 2003-04 Control Treatment Figure 1. Schematic representation of experimental units and nutrition enhancement treatment allocation. Units A and B were located in winter range habitat on the Uncompahgre Plateau in southwest Colorado. The nutrition enhancement cross-over design encompassed 4 years. Uncompahgre Plateau Mesa County GRAND JUNCTION Delta County m co Un GMU 62 r hg pa e u ea at Pl Montrose County GMU 61 Sanmiguel County Gunnison County DELTA Shavano E.U. Winter Range Exp. Units MONTROSE Colona Montrose County Summer Range E.U. Ouray County Figure 2. Location of Colona and Shavano (Units A and B) experimental units on the Uncompahgre Plateau, southwest Colorado; and location of the summer range study area encompassing the southern Uncompahgre Plateau and adjacent San Juan Mountains. 64 �APPENDIX I The following manuscript (referenced here by Abstract) was published in the Journal of Wildlife Management in 2007. USING VAGINAL IMPLANT TRANSMITTERS TO AID IN CAPTURE OF MULE DEER NEONATES CHAD J. BISHOP, DAVID J. FREDDY, GARY C. WHITE, BRUCE E. WATKINS, THOMAS R. STEPHENSON, AND LISA L. WOLFE ABSTRACT Estimating survival of the offspring of marked female ungulates has proven difficult in freeranging populations yet could improve our understanding of factors that limit populations. We evaluated the feasibility and efficiency of capturing large samples (i.e., >80/year) of neonate mule deer (Odocoileus hemionus) exclusively from free-ranging, marked adult does using vaginal implant transmitters (VITs, n = 154) and repeated locations of radio-collared does without VITs. We also evaluated the effectiveness of VITs, when used in conjunction with in utero fetal counts, for obtaining direct estimates of fetal survival. During 2003 and 2004, after we placed VIT batteries on a 12-hour duty cycle to lower electronic failure rates, the proportion that shed ≤3 days prepartum or during parturition was 0.623 (SE = 0.0456), and the proportion of VITs shed only during parturition was 0.447 (SE = 0.0468). Our neonate capture success rate was 0.880 (SE = 0.0359) from does with VITs shed ≤3 days prepartum or during parturition and 0.307 (SE = 0.0235) from radio-collared does without VITs or whose implants failed to function properly. Using a combination of techniques, we captured 275 neonates and found 21 stillborns during 2002−2004. We accounted for all fetuses at birth (i.e., live or stillborn) from 78 of the 147 does (0.531, SE = 0.0413) having winter fetal counts, and this rate was heavily dependent on VIT retention success. Deer that shed VITs prepartum were larger than deer that retained VITs to parturition, indicating a need to develop variable-sized VITs that may be fitted individually to deer in the field. We demonstrated that direct estimates of fetal and neonatal survival may be obtained from previously marked female mule deer in free-ranging populations, thus expanding opportunities for conducting field experiments. Survival estimates using VITs lacked bias that is typically associated with other neonate capture techniques. However, current vaginal implant failure rates, and overall expense, limit broad applicability of the technique. Citation: Bishop, C. J., D. J. Freddy, G. C. White, B. E. Watkins, T. R. Stephenson, and L. L. Wolfe. 2007. Using vaginal implant transmitters to aid in capture of mule deer neonates. Journal of Wildlife Management 71:945−954. 65 �The following manuscript (referenced here by Abstract) was published in the Journal of Wildlife Diseases in 2007: MALIGNANT CATARRHAL FEVER ASSOCIATED WITH OVINE HERPESVIRUS-2 IN FREERANGING MULE DEER IN COLORADO PATRICIA C. SCHULTHEISS, HANA VAN CAMPEN, TERRY R. SPRAKER, CHAD J. BISHOP, LISA L. WOLFE, AND BRENDAN PODELL ABSTRACT Malignant catarrhal fever (MCF) was diagnosed in 4 free-ranging mule deer (Odocoileus hemionus) in January and February of 2003. Diagnosis was based on typical histologic lesions of lymphocytic vasculitis and PCR identification of ovine herpesvirus-2 (OHV-2) viral genetic sequences in formalin fixed tissues. The animals were from the Uncompahgre Plateau of southwestern Colorado. Deer from these herds occasionally resided in close proximity to domestic sheep (Ovis aries), the reservoir host of OHV-2, in agricultural valleys adjacent to their winter range. These cases indicate that fatal OHV-2 associated MCF can occur in free-ranging mule deer exposed to domestic sheep that overlap their range. Citation: Schultheiss, P. C., H. Van Campen, T. R. Spraker, C. J. Bishop, L. L. Wolfe, and B. Podell. 2007. Malignant catarrhal fever associated with ovine herpesvirus-2 in free-ranging mule deer in Colorado. Journal of Wildlife Diseases 43:533−537. 66 �APPENDIX II The following draft manuscripts (referenced here by Abstract) were prepared in 2007 and will be submitted to the Journal of Wildlife Management. EVALUATING DEPENDENCE AMONG MULE DEER SIBLINGS IN FETAL AND NEONATAL SURVIVAL ANALYSES CHAD J. BISHOP, GARY C. WHITE, AND PAUL M. LUKACS ABSTRACT The assumption of independent sample units is potentially violated in deer (Odocoileus spp.) fetal and neonatal survival analyses where twin and triplet siblings comprise a high proportion of the sample. Violation of the independence assumption causes sample data to be overdispersed relative to a binomial model, and therefore requires a variance inflation factor, c, to obtain appropriate estimates of sampling variances. We evaluated overdispersion in fetal and neonatal mule deer (O. hemionus) datasets where more than half of the sample units were comprised of siblings. We developed a likelihood function for estimating fetal survival when the fates of some fetuses are unknown, and we used several variations of the binomial model to estimate neonatal survival. We compared theoretical variance estimates obtained from these analyses with empirical variance estimates obtained from data bootstrap analyses to estimate the overdisperion parameter, c. Our estimates of c for fetal survival ranged from 0.678 to 1.118, which provided virtually no evidence of overdispersion. For neonatal survival, 3 different models indicated that ĉ ranged from 1.1 to 1.4 and averaged 1.24−1.26, providing evidence of limited overdispersion (i.e., limited sibling dependence). Our results indicate that fates of sibling mule deer fetuses and neonates may often be independent even though they have the same dam. Predation tends to act independently on sibling neonates because of dam-neonate behavioral adaptations, although we observed several cases of a predator(s) killing siblings. The effect of maternal characteristics on sibling fate dependence is less straightforward and may vary by circumstance. We recommend that future neonatal survival studies incorporate additional sampling intensity to accommodate modest overdispersion (i.e., ĉ = 1.25), which would facilitate a corresponding ĉ adjustment in a model selection analysis using quasi-likelihood without a reduction in power. 67 �EFFECT OF ENHANCED NUTRITION ON MULE DEER POPULATION PERFORMANCE CHAD J. BISHOP, GARY C. WHITE, DAVID J. FREDDY, BRUCE E. WATKINS, AND THOMAS R. STEPHENSON ABSTRACT Concerns over declining mule deer (Odocoileus hemionus) populations during the 1990s prompted research efforts to identify and understand key limiting factors of deer. Similar to past deer declines, a top priority of state wildlife agencies was to evaluate the relative importance of habitat and predation. We therefore evaluated the effect of enhanced nutrition of deer during winter and spring on fecundity and survival rates using a field experiment involving free-ranging mule deer on the Uncompahgre Plateau in southwest Colorado. The nutrition enhancement treatment represented an instantaneous increase in nutritional carrying capacity of a pinyon (Pinus edulis) and Utah juniper (Juniperus osteosperma) winter range, and was intended to simulate optimum habitat quality. Prior studies on the Uncompahgre Plateau indicated predation and disease were the most common proximate causes of deer mortality. By manipulating nutrition and leaving predation as it was, we determined whether habitat quality was ultimately a critical limiting factor of the deer population. We measured fetal, neonatal, and overwinter fawn survival, and annual adult doe survival, which we then used to estimate population rate of change as a function of enhanced nutrition. Pregnancy and fetal rates were high for all deer, regardless of the nutrition treatment. Fetal and neonatal survival rates were higher among deer that received the nutrition enhancement treatment than deer that served as experimental controls. Overwinter fawn survival was considerably higher among treatment deer than control deer. Overwinter survival increased by 0.16−0.31 depending on year and fawn sex, and none of the 95% confidence intervals associated with the effect overlapped 0. The nutrition enhancement treatment increased survival of fetuses to the yearling age class by 0.14−0.20 depending on year and fawn sex, although 95% confidence intervals slightly overlapped 0. The nutrition treatment also had a positive effect on annual adult doe survival. Survival of does receiving the treatment (Ŝ = 0.879, SE = 0.0206) was higher than survival of control does (Ŝ = 0.833, SE = 0.0253). Our estimate of the population rate of change, λ̂ , was 1.15−1.17 for treatment deer and 1.02−1.06 for control deer, with some overlap in 95% confidence intervals. The treatment caused λ̂ to increase by 0.139 (95% CI: −0.0152, 0.2941) during 2001−02, 0.113 (95% CI: −0.0009, 0.2279) during 2002−03, and 0.145 (95% CI: 0.0176, 0.2723) during 2003−04. We documented density dependence in the Uncompahgre deer population because survival of fawns and does increased considerably in response to enhanced nutrition. We found strong evidence that coyote (Canis latrans) predation of ≥6 month old fawns and adult does was compensatory. Our results demonstrate that observed coyote predation is not useful for evaluating whether coyotes are negatively impacting a deer population. We also found evidence that mountain lion (Puma concolor) predation was compensatory. Disease was not compensatory among adult does. We found that winter range habitat quality was a limiting factor of the Uncompahgre Plateau deer population. We recommend the implementation and evaluation of habitat treatments designed to set back succession and increase productivity of late-seral pinyon-juniper habitats that presently dominate the landscape because of the absence of fire. 68 �EVALUATING MULE DEER BODY CONDITION DURING LATE WINTER USING SERUM THYROID HORMONE CONCENTRATIONS CHAD J. BISHOP, BRUCE E. WATKINS, LISA L. WOLFE, DAVID J. FREDDY, AND GARY C. WHITE ABSTRACT Body condition of ungulates is ultimately a determinant of fecundity and survival rates. Researchers have recently developed ultrasonography and body condition scoring techniques that allow reliable estimation of body fat in several ungulate species, but the approach is not feasible to employ in all circumstances, particularly in routine population monitoring programs. There remains a need for a reliable blood chemistry index that could be used to assess relative condition of different deer populations or groups. We evaluated the relationship between estimated body fat of free-ranging mule deer and serum concentrations of total thyroxine (T4), total triiodothyronine (T3), free T4 (FT4), and free T3 (FT3), during late February−early March in southwest Colorado. Deer body fat varied widely because we imposed a nutrition treatment on one-half of our sample. Concentrations of T4 and FT4 were 48.23 nmol/l (SE = 3.88) and 12.69 pmol/l (SE = 1.12) higher, respectively, in deer that received the nutrition treatment than deer that did not receive the treatment. Our optimal model of estimated body fat included T4, T42, FT4 and deer chest girth (%Fât = –4.8015 – 0.0946×T4 + 0.000603×T42 + 0.1474×FT4 + 0.1426×chest girth, r2 = 0.609). Ultrasound and body condition scoring should be used to estimate body fat whenever possible. However, in cases where only a blood sample can be obtained, we documented the potential utility of T4 and FT4 during late winter for evaluating relative body condition of mule deer. 69 �APPENDIX III The following draft dissertation (referenced here by Abstract) was prepared in 2007 and will be submitted to Colorado State University. EFFECT OF ENHANCED NUTRITION DURING WINTER ON THE UNCOMPAHGRE PLATEAU MULE DEER POPULATION CHAD J. BISHOP ABSTRACT Mule deer (Odocoileus hemionus) populations declined across much of the West during the 1990s, prompting state wildlife agencies to pursue explorations of mule deer limiting factors. The greatest concern of agencies and sportsmen was whether declining habitat quality or predation was ultimately responsible for the observed declines. In Colorado, the Uncompahgre Plateau mule deer population received the most attention, having substantially declined from the 1980s through the late 1990s. Biologists hypothesized that poor winter range habitat quality was the primary cause of the observed decline. In contrast, many of the Colorado Division of Wildlife’s (CDOW) external constituents hypothesized that high rates of predation were keeping the mule deer herd below nutritional carrying capacity. The habitat quality hypothesis indicated CDOW should pursue habitat improvements in the pinyon (Pinus edulis) and juniper (Juniperus osteosperma) winter range, whereas the predator hypothesis suggested CDOW should pursue efforts to reduce predator populations, particularly coyote (Canis latrans) populations. The competing hypotheses represented very different paradigms of population limitation. I therefore evaluated the effect of enhanced nutrition during winter on the Uncompahgre deer population as a way to evaluate the importance of habitat quality versus that of predation. I conducted a field study incorporating a crossover experimental design to quantify the effect of enhanced nutrition on fetal, neonatal, overwinter fawn, and annual adult doe survival rates. I captured and radio-collared samples of deer in 2 experimental units (EUs) on winter range. I delivered the nutrition treatment to deer occupying one EU (treatment) and did not administer the treatment to deer in the other EU (control). Established field techniques were not sufficient to allow me to quantify the effect of the treatment on fetal and neonatal survival. I therefore pursued an exploration of vaginal implant transmitters as a mechanism to capture necessary samples of newborn fawns on summer range exclusively from radio-collared does that occupied the winter range EUs (Chapter 1). This effort allowed me to estimate fetal and neonatal survival as a function of the treatment. In broad terms, I demonstrated that direct estimates of fetal and neonatal survival may be obtained from previously marked female mule deer in free-ranging populations, thus expanding opportunities for conducting field experiments. I encountered additional challenges with estimation of fetal and neonatal survival. First, I was unable to determine the fate of all fetuses that I documented in utero. I therefore developed a likelihood function for estimating fetal survival when the fates of some fetuses are unknown (Chapter 2). Second, a majority of my fetal and neonatal samples were comprised of siblings, indicating my data were potentially overdispersed. Overdispersion causes sample variances to be underestimated and requires a variance inflation factor, c. To estimate c, I compared theoretical variance estimates with empirical variance estimates obtained from data bootstrap analyses (Chapter 2). I found little evidence of overdispersion in my fetal survival data, and I found modest overdispersion in my neonatal sample data (ĉ = 1.25). Although some overdispersion was detected, my result indicated that fates of sibling mule deer neonates may often be independent even though they have the same dam and use the environment similarly. I discuss reasons for this in Chapter 2. 70 �After resolving issues with fetal and neonatal survival estimation, I quantified the effect of the nutrition enhancement treatment on fetal, neonatal, overwinter fawn, and annual adult doe survival (Chapter 3). I then used these parameter estimates, along with estimated fecundity rates, in an agestructured, deterministic population model to estimate the effect of the treatment on the population rate of change, λ. The treatment caused λ̂ to increase by an average of 0.133 (SD = 0.0168) during the 3 years of my study. I documented density dependence in the Uncompahgre deer population because survival of fawns and does increased considerably in response to enhanced nutrition. I found strong evidence that coyote predation of ≥6-month-old fawns and adult does was compensatory. Finally, I found that winter range habitat quality was a limiting factor of the Uncompahgre Plateau deer population. I completed my principal study objectives in the first 3 chapters of the dissertation. However, my research afforded the opportunity to evaluate the utility of serum thyroid hormones in mule deer as an index to body condition (Chapter 4). Concentrations of total thyroxine (T4) and free T4 (FT4) were substantially higher in treatment deer than control deer. I also found that serum thyroid hormones were highly correlated with estimated body fat in mule deer during late winter. Concentrations of T4 and FT4 could be useful for evaluating relative condition of different deer groups or populations, and for roughly estimating body fat of individual animals during late winter. n summary, I demonstrated that winter range habitat quality was ultimately limiting the Uncompahgre mule deer population. Observed predation was primarily compensatory, particularly of ≥6month-old fawns and adult does. My findings indicate that CDOW should implement and evaluate habitat treatments in late-seral pinyon-juniper habitat as a means to increase habitat productivity for mule deer. My findings provide no support for predator reduction programs. 71 �Colorado Division of Wildlife July 2007 June 2008 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 3001 4 Federal Aid Project No. W-185-R Period Covered: July 1, 2007 : : : : : Division of Wildlife Mammals Research Deer Conservation Effect of Nutrition and Habitat Enhancements On Mule Deer Recruitment and Survival Rates June 30, 2008 Authors: C. J. Bishop, G. C. White, D. J. Freddy, and B. E. Watkins All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We measured mule deer (Odocoileus hemionus) population parameters in response to a nutrition enhancement treatment to evaluate the relative importance of habitat quality as a limiting factor of mule deer in western Colorado during November 2000 – January 2005. The nutrition enhancement treatment increased survival of fetuses to the yearling age class by 0.14 0.20 depending on year and fawn sex; 95% confidence intervals slightly overlapped 0. Averaged across sexes and years, survival of treatment fetuses to the yearling age class was 0.447 (SE = 0.0519), whereas survival of control fetuses to the yearling age class was 0.271 (SE = 0.0418). The treatment caused fetal to yearling survival to increase by 0.177 (SE = 0.0818, 95% CI: 0.0163, 0.3370). The nutrition treatment also had a positive effect on annual adult female survival. Survival of adult females receiving the treatment (Ŝ = 0.879, SE = 0.0206) was higher than survival of control adult females (Ŝ = 0.833, SE = 0.0253). Our estimate of the population rate of change, ˆ , was 1.165 (SE = 0.0358) for treatment deer and 1.033 (SE = 0.0380) for control deer. The r nutrition treatment caused ˆ to increase by 0.133 (SE = 0.0428). We documented food limitation in the Uncompahgre deer population because survival of fawns and adult females increased considerably in response to enhanced nutrition. Our results provide a foundation for focusing deer management efforts on improving habitat quality in western Colorado pinyon-juniper (Pinus edulis-Juniperus osteosperma) ecosystems with corresponding research efforts to quantify the effects of habitat manipulations on deer performance. During 2007 08, we published one paper from this research in the Journal of Wildlife Management (JWM 72(5):1085 1093), we had another paper accepted for publication in Journal of Wildlife Management, and we had one paper accepted for publication in Wildlife Monographs pending suitable revision. The lead principal investigator published a Dissertation to complete requirements for a Ph.D. at Colorado State University. We previously published a manuscript on the effectiveness of vaginal implant transmitters (VITs) for capturing newborn fawns from specific adult females (Bishop et al. 2007). As a follow-up to this component of our research, we worked with Advanced Telemetry Systems (ATS, Isanti, MN) to develop a VIT with modified retention wings. The modified VIT will be ready for fieldtesting in 2009. r 39 �WILDLIFE RESEARCH REPORT EFFECT OF NUTRITION AND HABITAT ENHANCEMENTS ON MULE DEER RECRUITMENT AND SURVIVAL RATES CHAD J. BISHOP, GARY C. WHITE, DAVID J. FREDDY, AND BRUCE E. WATKINS P. N. OBJECTIVE To determine experimentally whether enhancing mule deer nutrition during winter and early spring via supplementation increases fetal survival, neonatal survival, overwinter fawn survival, or ultimately, population productivity. SEGMENT OBJECTIVES 1. Publish manuscripts in peer-reviewed scientific journals. 2. Publish dissertation as part of Ph.D. requirements at Colorado State University INTRODUCTION Mule deer (Odocoileus hemionus) numbers apparently declined during the 1990s throughout much of the West, and have clearly decreased since the peak population levels documented during the 1940s- 1960s (Unsworth et al. 1999, Gill et al. 2001). Biologists and sportsmen alike have concerns as to what factors may be responsible for declining population trends. Although previous and current research indicates multiple interacting factors are responsible, habitat and predation have typically received the focus of attention. A number of studies have evaluated whether predator control increases deer survival, yet results are highly variable (Connolly 1981, Ballard et al. 2001). Together, predator control studies with adequate rigor and statistical power indicate predation effects on mule deer are variable as a result of time-specific and site-specific factors. Studies which have demonstrated deer population responses to predator control treatments have failed to determine whether predation is ultimately more limiting than habitat when considering long term population changes. Numerous research studies have evaluated mule deer habitat quality, but virtually no studies have documented population responses to habitat improvements. In many areas where declining deer numbers are of concern, predation is common yet habitat quality appears to have declined. The question remains as to whether predation, habitat, or some other factor is more limiting to mule deer in these situations, and whether habitat quality can be improved for the benefit of deer. It may also be that no single factor is responsible for observed deer declines, and a more comprehensive understanding of multi-factor interactions is needed. We designed and implemented a field experiment where we measured deer population responses to a nutrition enhancement treatment to further understand the causative factors underlying observed deer population dynamics. We conducted the study on the Uncompahgre Plateau in southwest Colorado, where several predator species were present in abundant numbers: coyotes (Canis latrans), mountain lions (Felis concolor), and bears (Ursus americanus). In addition to predation, myriad diseases in combination have proximately affected survival of the Uncompahgre deer population (Pojar and Bowden 2004, B. E. Watkins, Colorado Division of Wildlife, unpublished data). Predator numbers were not manipulated in any manner during the course of the study. All factors were left constant with the exception of deer nutrition. Deer nutrition was enhanced by providing supplemental feed to deer occupying a treatment area during winter. We measured December fawn recruitment and overwinter fawn survival in response to the treatment to determine whether deer nutrition was ultimately more limiting 40 �than predation or disease. A second phase of research was initiated in 2005 to quantify deer population parameters in response to manipulations of pinyon-juniper (Pinus edulis-Juniperus osteosperma) habitat (Bergman et al. 2007). The objective of this research is to determine whether habitat can be effectively improved for mule deer by introducing disturbance into late-seral pinyon-juniper stands. STUDY AREA We non-randomly selected two experimental units (A B) within mule deer winter range on the Uncompahgre Plateau (Figure 1) to facilitate a cross-over experimental design for evaluating the effects of enhanced deer nutrition during winter on annual population performance. Unit A received a nutrition enhancement treatment during the first 2 winters of research (2000 – 2002) while Unit B served as a control unit. During winters 2002 03 and 2003 04, Unit B received the treatment while Unit A served as the control. In late April and May, prior to fawning, deer from the winter range experimental units migrated to summer range. We defined the summer range study area by movements of the radio-collared deer captured on winter range; summer range encompassed >1000 mi2 covering the southern portion of the Uncompahgre Plateau and adjacent San Juan Mountains (Figure 2). Winter range elevations ranged from 1830 m (6000 ft) in Shavano Valley to 2318 m (7600 ft) adjacent to the Dry Creek Rim above Shavano Valley. Winter range habitat was dominated by pinyon-juniper with interspersed sagebrush adjacent to agricultural fields in the Shavano and Uncompahgre Valleys. Summer range elevations occupied by deer ranged from 1891 m (6200 ft) in the Uncompahgre Valley to 3538 m (11,600 ft) in Imogene Basin southwest of Ouray, CO. Summer range habitats were dominated by spruce-subalpine fir (Picea spp.-Abies lasiocarpa), aspen (Populus tremuloides), sagebrush, ponderosa pine (Pinus ponderosa), Gambel oak (Quercus gambelii), and to a lesser extent, pinyon-juniper at lower elevations. Bishop et al. (2005) provide a detailed study area description. METHODS Refer to Bishop et al. (2005) or Bishop (2007) for field methodology employed during 2000 2005. During fiscal year 2007 08, we had 1 paper published and 2 papers accepted for publication in peer-reviewed scientific journals. Thus, our primary research efforts were focused on preparation of manuscripts for publication. We completed and published a paper in Journal of Wildlife Management focused on mule deer sibling dependence in context of fetal and neonatal survival analyses. In this paper, we also presented a likelihood function for estimating fetal survival when the fates of some fetuses are unknown. We spent much of the year preparing and submitting a manuscript to Wildlife Monographs. This particular publication documents the effect of enhanced nutrition on all aspects of mule deer productivity, survival, and population rate of change. Finally, we prepared and submitted a manuscript documenting the utility of serum thyroid hormone concentrations for evaluating mule deer body condition in late winter with this manuscript accepted for publication following two substantive revisions. The principal investigator also published his Ph.D. dissertation. A component of this project was an evaluation of vaginal implant transmitters (VITs) as a tool for locating neonatal mule deer fawns from targeted adult females (Bishop et al. 2007). To build on this research, we worked with Advanced Telemetry Systems (ATS, Isanti, MN) to develop a VIT with modified retention wings during 2007 08. We intend to evaluate the modified VIT in conjunction with ongoing mule deer energy development research in northwest Colorado. RESULTS AND DISCUSSION A comprehensive presentation and discussion of all results from this study is provided by Bishop (2007) and is not repeated here. These results and conclusions are being systematically published in peer- 41 �reviewed journals. The following manuscripts were published in 2007 and 2008 (abstracts are provided in Appendix I): Bishop, C. J. 2007. Effect of enhanced nutrition during winter on the Uncompahgre Plateau mule deer population. Dissertation, Colorado State University, Fort Collins, USA. Bishop, C. J., D. J. Freddy, G. C. White, B. E. Watkins, T. R. Stephenson, and L. L. Wolfe. 2007. Using vaginal implant transmitters to aid in capture of mule deer neonates. Journal of Wildlife Management 71:945 954. Bishop, C. J., G. C. White, and P. M. Lukacs. 2008. Evaluating dependence among mule deer siblings in fetal and neonatal survival analyses. Journal of Wildlife Management 72:1085 1093. Schultheiss, P. C., H. Van Campen, T. R. Spraker, C. J. Bishop, L. L. Wolfe, and B. Podell. 2007. Malignant catarrhal fever associated with ovine herpesvirus-2 in free-ranging mule deer in Colorado. Journal of Wildlife Diseases 43:533 537. The following manuscripts were accepted for publication in 2008 and will most likely be published in 2009 (abstracts are provided in Appendix II): Bishop, C. J., B. E. Watkins, L. L. Wolfe, D. J. Freddy, and G. C. White. 2009. Evaluating mule deer body condition using serum thyroid hormone concentrations. Journal of Wildlife Management: In press. Bishop, C. J., G. C. White, D. J. Freddy, B. E. Watkins, and T. R. Stephenson. 2009. Effect of enhanced nutrition on mule deer population rate of change. Wildlife Monographs: in review. (Manuscript has been tentatively accepted pending suitable revision). We intend to pursue several additional manuscripts as time allows, listed below in order of priority. 1. Evaluating dependence of fates among mule deer siblings in Colorado, Idaho, and Montana. Journal of Wildlife Management. 2. Bovine viral diarrhea isolation and seroprevalence in a free-ranging mule deer (Odocoileus hemionus) population in southwest Colorado. Journal of Wildlife Diseases. 3. Spatial patterns in mortality causes of neonatal mule deer across a land use gradient in southwest Colorado. Journal of Wildlife Management. 4. Evaluation of mule deer age and sex ratios as a response variable in field research. Journal of Wildlife Management. SUMMARY Enhanced winter nutrition of free-ranging deer caused an increase in both fetus-neonate survival and overwinter fawn survival, resulting in higher yearling recruitment. Overwinter adult female survival increased as a result of the nutrition treatment, and therefore annual survival was higher among treatment than control adult females. Combining all parameter estimates into a deterministic population model, the treatment population indicated an exceptionally high rate of increase while the control population was stable and indicative of the overall Uncompahgre deer population during 2000 2004. The nutrition enhancement treatment was artificial in the sense that we applied it only to test whether habitat quality was ultimately more limiting than predation or other factors. Our results to do not provide support for managing deer populations with nutrition supplements because our treatment delivery approach could not 42 �be applied to a large number of animals over a large area. Rather, our results provide a foundation for focusing deer management efforts on improving habitat quality in western Colorado pinyon-juniper ecosystems with corresponding research efforts to quantify the effects of habitat manipulations on deer. LITERATURE CITED Ballard, W. B., D. Lutz, T. W. Keegan, L. H. Carpenter, and J. C. DeVos, Jr. 2001. Deer-predator relationships: a review of recent North American studies with emphasis on mule and black-tailed deer. Wildlife Society Bulletin 29:99 115. Bergman, E. J., C. J. Bishop, D. J. Freddy, and G. C. White. 2007. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Wildlife Research Report July: 73-96. Colorado Division of Wildlife, Fort Collins, USA. Bishop, C. J. 2007. Effect of enhanced nutrition during winter on the Uncompahgre Plateau mule deer population. Dissertation, Colorado State University, Fort Collins, USA. Bishop, C. J., D. J. Freddy, G. C. White, B. E. Watkins, T. R. Stephenson, and L. L. Wolfe. 2007. Using vaginal implant transmitters to aid in capture of mule deer neonates. Journal of Wildlife Management 71:945 954. Bishop, C. J., G. C. White, D. J. Freddy, and B. E. Watkins. 2005. Effect of nutrition and habitat enhancements on mule deer recruitment and survival rates. Wildlife Research Report July: 3766. Colorado Division of Wildlife, Fort Collins, USA. Connolly, G. E. 1981. Limiting factors and population regulation. Pages 245 285 in O. C. Wallmo, editor. Mule and black-tailed deer of North America. University of Nebraska Press, Lincoln, USA. Gill, R. B., T. D. I. Beck, C. J. Bishop, D. J. Freddy, N. T. Hobbs, R. H. Kahn, M. W. Miller, T. M. Pojar, And G. C. White. 2001. Declining mule deer populations in Colorado: reasons and responses. Colorado Division of Wildlife Special Report Number 77. Fort Collins, USA. Pojar, T. M., And D. C. Bowden. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550 560. Unsworth, J. W., D. F. Pac, G. C. White, And R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315 326. Prepared by _______________________ Chad J. Bishop, Wildlife Researcher 43 �Year Unit A Unit B 2000-01 Treatment Control 2001-02 Treatment Control 2002-03 Control Treatment 2003-04 Control Treatment Figure 1. Schematic representation of experimental units and nutrition enhancement treatment allocation. Units A and B were located in winter range habitat on the Uncompahgre Plateau in southwest Colorado. The nutrition enhancement cross-over design encompassed 4 years. Uncompahgre Plateau Mesa County GRAND JUNCTION Delta County U nc GMU 62 Gunnison County DELTA om Winter Range Exp. Units pa hg re ea at Pl GMU 61 u Montrose County Sanmiguel County Shavano E.U. MONTROSE Colona Montrose County Summer Range E.U. Ouray County Figure 2. Location of Colona and Shavano (Units A and B) experimental units on the Uncompahgre Plateau, southwest Colorado; and location of the summer range study area encompassing the southern Uncompahgre Plateau and adjacent San Juan Mountains. 44 �APPENDIX I – PUBLISHED PROJECT PAPERS The following Colorado State University dissertation (referenced here by Abstract) was published in 2007. EFFECT OF ENHANCED NUTRITION DURING WINTER ON THE UNCOMPAHGRE PLATEAU MULE DEER POPULATION CHAD J. BISHOP ABSTRACT Mule deer (Odocoileus hemionus) populations declined across much of the West during the 1990s, prompting state wildlife agencies to explore mule deer limiting factors. The greatest concern of agencies and sportsmen was whether declining habitat quality, predation, or both were responsible for the observed declines. In Colorado, the Uncompahgre Plateau mule deer population received the most attention because of a steep population decline from the 1980s through the late 1990s. Biologists hypothesized that poor quality of the pinyon (Pinus edulis) and juniper (Juniperus osteosperma) winter range was the primary cause of the observed decline. In contrast, many of the Colorado Division of Wildlife‘s (CDOW) constituents hypothesized that high predation rates were keeping the mule deer herd below nutritional carrying capacity. These hypotheses represented very different paradigms of population limitation. Perhaps more importantly, the competing views suggested that CDOW should pursue one of two very different management strategies: 1) implement habitat improvements in the pinyon-juniper winter range, or 2) implement efforts to reduce predator populations, particularly coyote (Canis latrans) populations. Information was needed to guide the decision process. I therefore evaluated the effect of enhanced nutrition during winter on the Uncompahgre deer population as a way to evaluate the importance of habitat quality versus that of predation. I conducted a field study incorporating a crossover experimental design to quantify the effect of enhanced nutrition on fetal, neonatal, overwinter fawn, and annual adult doe survival rates. I captured and radio-collared samples of deer in 2 experimental units (EUs) on winter range. I delivered the nutrition treatment to deer occupying one EU (treatment) and did not administer the treatment to deer in the other EU (control). Established field techniques were not sufficient to allow me to quantify the effect of the treatment on fetal and neonatal survival. I therefore pursued an exploration of vaginal implant transmitters as a mechanism to capture necessary samples of newborn fawns on summer range exclusively from radio-collared does that occupied the winter range EUs (Chapter 1). This effort allowed me to estimate fetal and neonatal survival as a function of the treatment. In broad terms, I demonstrated that direct estimates of fetal and neonatal survival may be obtained from previously marked female mule deer in free-ranging populations, thus expanding opportunities for conducting field experiments. I encountered additional challenges with estimation of fetal and neonatal survival. First, I was unable to determine the fate of all fetuses that I documented in utero. I therefore developed a likelihood function for estimating fetal survival when the fates of some fetuses are unknown (Chapter 2). Second, a majority of my fetal and neonatal samples were comprised of siblings, indicating my data were potentially overdispersed. Overdispersion causes sample variances to be underestimated and requires a variance inflation factor, c. To estimate c, I compared theoretical variance estimates with empirical variance estimates obtained from bootstrap analyses of the data (Chapter 2). I found little evidence of overdispersion in my fetal survival data, and I found modest overdispersion in my neonatal sample data (ĉ = 1.25). Although some overdispersion was detected, my results indicated that fates of sibling mule deer neonates may often be independent even though they have the same dam and use the environment similarly. I discuss reasons for this in Chapter 2. 45 �After resolving issues with fetal and neonatal survival estimation, I quantified the effect of the nutrition enhancement treatment on fetal, neonatal, overwinter fawn, and annual adult doe survival (Chapter 3). I then used these parameter estimates, along with estimated fecundity rates, in an agestructured, deterministic population model to estimate the effect of the treatment on the population rate of r change, ˆ . The treatment caused ˆ to increase by an average of 0.133 (SD = 0.0168) during the 3 years of my study. I documented density dependence in the Uncompahgre deer population because survival of fawns and does increased considerably in response to enhanced nutrition. I found strong evidence that coyote predation of ≥6-month-old fawns and adult does was compensatory. Finally, I found that winter range habitat quality was a limiting factor of the Uncompahgre Plateau deer population. r I completed my principal study objectives in the first 3 chapters of the dissertation. However, my research afforded the opportunity to evaluate the utility of serum thyroid hormones in mule deer as an index to body condition (Chapter 4). Concentrations of total thyroxine (T4) and free T4 (FT4) were substantially higher in treatment deer than control deer. I also found that serum thyroid hormones were highly correlated with estimated body fat in mule deer during late winter. Concentrations of T4 and FT4 could be useful for evaluating relative condition of different deer groups or populations, and for roughly estimating body fat of individual animals during late winter. In summary, I demonstrated that winter range habitat quality was ultimately limiting the Uncompahgre mule deer population. Observed predation was primarily compensatory, particularly of ≥6month-old fawns and adult does. My findings indicate that CDOW should evaluate habitat treatments in late-seral pinyon-juniper habitat as a means to increase habitat productivity for mule deer. Citation: Bishop, C. J. 2007. Effect of enhanced nutrition during winter on the Uncompahgre Plateau mule deer population. Dissertation, Colorado State University, Fort Collins, USA. 46 �The following manuscript (referenced here by Abstract) was published in the Journal of Wildlife Management in 2007. USING VAGINAL IMPLANT TRANSMITTERS TO AID IN CAPTURE OF MULE DEER NEONATES CHAD J. BISHOP, DAVID J. FREDDY, GARY C. WHITE, BRUCE E. WATKINS, THOMAS R. STEPHENSON, AND LISA L. WOLFE ABSTRACT Estimating survival of the offspring of marked female ungulates has proven difficult in freeranging populations yet could improve our understanding of factors that limit populations. We evaluated the feasibility and efficiency of capturing large samples (i.e., >80/year) of neonate mule deer (Odocoileus hemionus) exclusively from free-ranging, marked adult does using vaginal implant transmitters (VITs, n = 154) and repeated locations of radio-collared does without VITs. We also evaluated the effectiveness of VITs, when used in conjunction with in utero fetal counts, for obtaining direct estimates of fetal survival. During 2003 and 2004, after we placed VIT batteries on a 12-hour duty cycle to lower electronic failure rates, the proportion that shed 3 days prepartum or during parturition was 0.623 (SE = 0.0456), and the proportion of VITs shed only during parturition was 0.447 (SE = 0.0468). Our neonate capture success rate was 0.880 (SE = 0.0359) from does with VITs shed 3 days prepartum or during parturition and 0.307 (SE = 0.0235) from radio-collared does without VITs or whose implants failed to function properly. Using a combination of techniques, we captured 275 neonates and found 21 stillborns during 2002 2004. We accounted for all fetuses at birth (i.e., live or stillborn) from 78 of the 147 does (0.531, SE = 0.0413) having winter fetal counts, and this rate was heavily dependent on VIT retention success. Deer that shed VITs prepartum were larger than deer that retained VITs to parturition, indicating a need to develop variable-sized VITs that may be fitted individually to deer in the field. We demonstrated that direct estimates of fetal and neonatal survival may be obtained from previously marked female mule deer in free-ranging populations, thus expanding opportunities for conducting field experiments. Survival estimates using VITs lacked bias that is typically associated with other neonate capture techniques. However, current vaginal implant failure rates, and overall expense, limit broad applicability of the technique. Citation: Bishop, C. J., D. J. Freddy, G. C. White, B. E. Watkins, T. R. Stephenson, and L. L. Wolfe. 2007. Using vaginal implant transmitters to aid in capture of mule deer neonates. Journal of Wildlife Management 71:945 954. 47 �The following manuscript (referenced here by Abstract) was published in the Journal of Wildlife Management in 2008. EVALUATING DEPENDENCE AMONG MULE DEER SIBLINGS IN FETAL AND NEONATAL SURVIVAL ANALYSES CHAD J. BISHOP, GARY C. WHITE, AND PAUL M. LUKACS ABSTRACT The assumption of independent sample units is potentially violated in survival analyses where siblings comprise a high proportion of the sample. Violation of the independence assumption causes sample data to be overdispersed relative to a binomial model, which leads to underestimates of sampling variances. A variance inflation factor, c, is therefore required to obtain appropriate estimates of variances. We evaluated overdispersion in fetal and neonatal mule deer (Odocoileus hemionus) datasets where more than half of the sample units were comprised of siblings. We developed a likelihood function for estimating fetal survival when the fates of some fetuses are unknown, and we used several variations of the binomial model to estimate neonatal survival. We compared theoretical variance estimates obtained from these analyses with empirical variance estimates obtained from data bootstrap analyses to estimate the overdisperion parameter, c. Our estimates of c for fetal survival ranged from 0.678 to 1.118, which indicate little to no evidence of overdispersion. For neonatal survival, 3 different models indicated that ĉ ranged from 1.1 to 1.4 and averaged 1.24 1.26, providing evidence of limited overdispersion (i.e., limited sibling dependence). Our results indicate that fates of sibling mule deer fetuses and neonates may often be independent even though they have the same dam. Predation tends to act independently on sibling neonates because of dam-neonate behavioral adaptations. The effect of maternal characteristics on sibling fate dependence is less straightforward and may vary by circumstance. We recommend that future neonatal survival studies incorporate additional sampling intensity to accommodate modest overdispersion (i.e., ĉ = 1.25), which would facilitate a corresponding ĉ adjustment in a model selection analysis using quasi-likelihood without a reduction in power. Our computational approach could be used to evaluate sample unit dependence in other studies where fates of individually marked siblings are monitored. Citation: Bishop, C. J., G. C. White, and P. M. Lukacs. 2008. Evaluating dependence among mule deer siblings in fetal and neonatal survival analyses. Journal of Wildlife Management 72:1085 1093. 48 �The following manuscript (referenced here by Abstract) was published in the Journal of Wildlife Diseases in 2007: MALIGNANT CATARRHAL FEVER ASSOCIATED WITH OVINE HERPESVIRUS-2 IN FREE-RANGING MULE DEER IN COLORADO PATRICIA C. SCHULTHEISS, HANA VAN CAMPEN, TERRY R. SPRAKER, CHAD J. BISHOP, LISA L. WOLFE, AND BRENDAN PODELL ABSTRACT Malignant catarrhal fever (MCF) was diagnosed in 4 free-ranging mule deer (Odocoileus hemionus) in January and February of 2003. Diagnosis was based on typical histologic lesions of lymphocytic vasculitis and PCR identification of ovine herpesvirus-2 (OHV-2) viral genetic sequences in formalin fixed tissues. The animals were from the Uncompahgre Plateau of southwestern Colorado. Deer from these herds occasionally resided in close proximity to domestic sheep (Ovis aries), the reservoir host of OHV-2, in agricultural valleys adjacent to their winter range. These cases indicate that fatal OHV-2 associated MCF can occur in free-ranging mule deer exposed to domestic sheep that overlap their range. Citation: Schultheiss, P. C., H. Van Campen, T. R. Spraker, C. J. Bishop, L. L. Wolfe, and B. Podell. 2007. Malignant catarrhal fever associated with ovine herpesvirus-2 in free-ranging mule deer in Colorado. Journal of Wildlife Diseases 43:533 537. 49 �APPENDIX II PROJECT PAPERS ACCEPTED FOR PUBLICATION The following manuscript (referenced here by Abstract) was accepted for publication by the Journal of Wildlife Management during 2008 but has not yet been published. EVALUATING MULE DEER BODY CONDITION USING SERUM THYROID HORMONE CONCENTRATIONS CHAD J. BISHOP, BRUCE E. WATKINS, LISA L. WOLFE, D. J. FREDDY, AND GARY C. WHITE ABSTRACT Body condition of ungulates is a determinant of fecundity and survival rates. Ultrasonography and body condition scoring techniques allow reliable estimation of body fat but may not be feasible to employ in some circumstances. A reliable blood chemistry index for assessing relative condition of different ungulate populations or groups would be useful in ongoing population monitoring programs. We provided a nutrition supplement (treatment) to a group of free-ranging mule deer (Odocoileus hemionus) during 2 consecutive winters in southwest Colorado. In late February each year, we evaluated whether percent body fat and serum concentrations of total thyroxine (T4), total triiodothyronine (T3), free T4 (FT4), and free T3 (FT3) were higher among treatment deer than an adjacent group of deer that did not receive the treatment (control). As a corroborative analysis, we modeled body fat as a function of thyroid hormone concentrations and morphometric variables. Estimated body fat of treatment deer averaged 12.3% (SE = 0.327), whereas estimated body fat of control deer averaged 7.0% (SE = 0.333), during the 2 winters of study. Concentrations of T4 and FT4 averaged 48.07 nmol/l (SE = 3.80) and 12.61 pmol/l (SE = 1.04) higher, respectively, in treatment deer than control deer. Our optimal model of estimated body fat included T4, T42, FT4, and deer chest girth (%Fât = –4.8015 – 0.0946X T4 + 0.000603X T42 + 0.1474 X FT4 + 0.1426 X chest girth, R2 = 0.609). Serum thyroid hormones effectively discerned treatment deer from control deer and were related to estimated body fat. Ultrasound and body condition scoring should be used to estimate body fat whenever possible. However, in cases where only a blood sample can be obtained, we documented potential utility of T4 and FT4 during late winter for evaluating relative body condition of mule deer. 50 �The following manuscript (referenced here by Abstract) was tentatively accepted for publication by Wildlife Monographs during 2008 and is still in the revision stage. EFFECT OF ENHANCED NUTRITION ON MULE DEER POPULATION RATE OF CHANGE CHAD J. BISHOP, GARY C. WHITE, DAVID J. FREDDY, BRUCE E. WATKINS, AND THOMAS R. STEPHENSON ABSTRACT Concerns over declining mule deer (Odocoileus hemionus) populations during the 1990s prompted research efforts to identify and understand key limiting factors of deer. Similar to past deer decline incidents, a top priority of state wildlife agencies was to evaluate the relative importance of habitat and predation. We therefore evaluated the effect of enhanced nutrition of deer during winter and spring on fecundity and survival rates using a life table response experiment involving free-ranging mule deer on the Uncompahgre Plateau in southwest Colorado. The nutrition enhancement treatment represented an instantaneous increase in nutritional carrying capacity of a pinyon (Pinus edulis) and Utah juniper (Juniperus osteosperma) winter range, and was intended to simulate optimum habitat quality. Prior studies on the Uncompahgre Plateau indicated predation and disease were the most common proximate causes of deer mortality. By manipulating nutrition and leaving natural predation unaltered, we determined whether habitat quality was ultimately a critical factor limiting the deer population. We measured fetal, neonatal, and overwinter fawn survival, and annual adult female survival, which we then used to estimate population rate of change as a function of enhanced nutrition. Pregnancy and fetal rates were high for all deer, regardless of the nutrition treatment. Fetal and neonatal survival rates were higher among deer that received the nutrition enhancement treatment than deer that served as experimental controls. Overwinter fawn survival increased for treatment deer by 0.16 0.31 depending on year and fawn sex, and none of the 95% confidence intervals associated with the effect overlapped 0. Nutrition enhancement increased survival of fetuses to the yearling age class by 0.14 0.20 depending on year and fawn sex, although 95% confidence intervals slightly overlapped 0. Annual survival of adult females receiving the treatment (Ŝ = 0.879, SE = 0.0206) was higher than survival of control adult females (Ŝ = 0.833, SE = 0.0253). Our estimate of the population rate of change, ˆ , was 1.165 (SE = 0.0358) for r treatment deer and 1.033 (SE = 0.0380) for control deer. The nutrition treatment caused ˆ to increase by 0.133 (SE = 0.0428). We documented density dependence in the Uncompahgre deer population because survival of fawns and adult females increased considerably in response to enhanced nutrition. We found strong evidence that coyote (Canis latrans) predation of ≥6-month-old fawns and adult females was compensatory. Our results demonstrate that observed coyote predation, by itself, is not useful for evaluating whether coyotes are negatively impacting a deer population. We also found evidence that mountain lion (Puma concolor) predation was compensatory. Disease mortality was not compensatory among adult females. We found that winter range habitat quality was a limiting factor of the Uncompahgre Plateau mule deer population. Therefore, we recommend evaluating habitat treatments for deer that are designed to set-back succession and increase productivity of late-seral pinyon-juniper habitats that presently dominate the winter range. r 51 � https://cpw.cvlcollections.org/files/original/310a46686431af346629c8bed1afe33b.pdf b28821a288fc91e16e6fbe72992dbdb2 PDF Text Text 81 ! f Colorado Division of Wildlife Wildlife Research Report July 2001 and July 2002 FINAL REPORT I l. State of_______C-=-=-o=lo=r=ad=o=-------- Division of Wildlife - Mammals Research Work Package No . _ _--=3=.0-=-0.::....1_ _ _ _ __ Deer Conservation Task No. _ _ _ _ _ _ _4.c..__ _ _ _ __ Pilot Study- Use of Ultrasound and Vaginal Implants Federal Aid Project_ _W~-1~8~5~-R~----- Research and Development Period Covered: July 1, 2001-June 30, 2002 Authors: C. J. Bishop, D. J. Freddy, and G. C. White, Ph.D. i l. Personnel: D. L. Baker, T. Baker, R. Bavin, T. D. I. Beck, S. K. Carroll, D. Coven, K. Crane, M. Del Tonto, L. Gepfert, J. Grigg, M. McLain, G. C. Miller, M. W. Miller, J. Olterman, J. A. Padia, T. M. Pojar, J. Risher, C. M. Solohub, M. Thonhoff, B. E. Watkins, L. Wolfe, CDOW; T. R. Stephenson, California Dept. of Fish and Game; R. C. Cook, National Council for Air and Stream Improvement. ABSTRACT Field research of mule deer could be greatly enhanced if newborn fawns could be captured from specific adult females from which data has already been collected. We evaluated the logistical feasibility and effectiveness of using vaginal implant transmitters (VITs) to determine the location and timing of birth of specific, radio-collared adult female mule deer. The VITs were manufactured by Advanced Telemetry Systems, Inc. (Isanti, MN). We placed VITs in 36 adult female deer on February 28 and March 1, 2002. At this time, we recorded data such as fetal rate, body condition, body mass, and serology from a blood sample. In June 2002, we intensively radio-monitored the VITs to determine when they were expelled from the deer. When a VIT was shed, we immediately located it to determine whether a birth site was present and to locate/capture neonates. The proportion of VITs that were expelled at or near the birth site with the transmitter functioning correctly was 0.33 (SD= 0.083). Of 36 VIT trials, 3 were censored, 7 were shed prematurely, 15 had battery failures, and 11 were successful (9 of which led to birth sites and subsequent fawn captures). In spite of the high VIT failure rate, we captured and radio-collared a total of 54 fawns from 38 adult does during June 11 - July 1, 2002. Twenty-six fawns were captured from 17 of the 36 VIT does and 28 fawns were captured from 21 radio-collared does that did not have VITs. Contrary to our own expectations, we successfully captured fawns from radio-collared does by relocating the does on a routine basis. However, this technique was inefficient and required a total capture effort of 1700 man-hours (212 man-days) during a 22-day period, or roughly 4 man-days/fawn. The VIT battery failures were clearly the main problem we experienced, which can be corrected. The amount of time and effort saved by the 11 successful VITs justifies their use, particularly with continued refinement of the VIT design. BDOW016782 ��83 EFFECTS OF ENHANCED WINTER NUTRITION OF ADULT FEMALE MULE DEER ON FETAL AND NEONATAL SURVIVAL RATES: A PILOT STUDY TO ADDRESS FEASIBILITY C. J. Bishop, D. J. Freddy, and G. C. White PROJECT OBJECTIVE 1. To evaluate the feasibility of utilizing ultrasound techniques and vaginal implant transmitters in adult female mule deer to measure stillborn fetus mortality and to locate and capture specific neonate fawns. SEGMENT OBJECTIVES 1. • Prepare a ·Program Narrative for a I -year pilot study. 2. Conduct the I-year pilot study to evaluate logistical feasibility of field techniques, collect data necessary for subsequent sample size calculations, and to obtain preliminary biological data. 3. Prepare a Job Final Report for the I-year pilot study. INTRODUCTION Background The Colorado Division of Wildlife initiated 2 studies on the Uncompahgre Plateau in response to chronically low December fawn:doe ratios throughout the 1990's and an overall decline in total deer numbers. In 1997, an ongoing survival study was initiated to quantify overwinter fawn survival and annual adult survival rates, and to identify mortality agents (B.E. Watkins, Colorado Division of Wildlife, unpublished data). In 1999, another study began to quantify pregnancy/fetal rates and to measure survival rates and cause-specific mortality of neonate fawns (Pojar and Andelt 1999, Pojar 2000). These studies have provided several significant findings to date. First, overwi~ter survival rates of fawns, and annual survival rates of does and bucks, are above average when compared to measurements obtained elsewhere (Unsworth et al. 1999). Second, adult doe pregnancy rates (93%) and fetal rates (1.7 fawns/doe) in February are normal (Pojar and Andelt 1999). Third, summer fawn survival has been relatively low overall, with malnutrition/sickness and predation being the primary causes of mortality (Pojar 2000). Based on these findings, in utero fetus survival/summer fawn surviyal is clearly the limiting factor to population growth on the Plateau. Summer fawn survival has been measured directly, while the extent of in utero fetus mortality from February to birth has been back-calculated utilizing expected versus observed December fawn:doe ratios based on the observed summer fawn survival. Although this fetus mortality has not been measured directly, there is considerable evidence that some portion of viable fetuses in February are not surviving to birth. Given the magnitude ofmalnutrition/skkness observed in newborn fawns, the question of prepartum adult doe nutrition is paramount. Summer range habitat quality on the Uncompahgre Plateau is seemingly good, and arguably better than many other deer summer ranges throughout the intermountain West. However, lo'Yer transitional and winter range habitat quality appears to be limited in terms of forage diversity and quality. We understand the inherent limits in nutritional quality of winter range forage, but hypothesize that winter range habitat quality on the Plateau may not be meeting the minimum nutritional requirements of pregnant adult does. �84 In 2000, we initiated a research study on the Uncompahgre Plateau to evaluate the effects of enhanced nutrition of mule deer during winter on fawn production and survival (Bishop and White 2000, Bishop and White 2002). The objectives of this research are twofold: 1) to determine experimentally whether enhancing the nutrition of deer during winter and early spring by supplemental feeding increases overwinter fawn survival and/or December fawn:doe ratios the following December; and 2) to determine experimentally to what extent habitat treatments replicate the effect of enhanced nutrition from supplemental feeding. We are addressing these objectives by radio-collaring adult does and 6-month old fawns in a treatment experimental unit and a control experimental unit. The current phase of the experiment uses supplemental feed as a nutrition enhancement treatment, while the second phase will use habitat manipulations (e.g. prescribed fire, mechanical) as the treatment. The main focus of this research is to determine whether a decline in winter range habitat quality has been a causative factor of poor December fawn recruitment on the Uncompahgre Plateau during the past decade. More specifically, we are determining whether m~trition enhancements and/or habitat treatments on winter range cause an increase in fawn:doe ratios the following December. Our primary response variable is December fawn:doe ratios measured from radio-collared does in the treatment and control units. December fawn:doe ratios represent a combined approximation of fawn production (# fetuses produced and successfully brought to term) and survival (% of newborn fawns surviving from birth to December). Fawn:doe ratios are influenced by the number of yearling females in the population, the proportion of small yearling bucks that may be misidentified as does, doe harvests etc. Irrespective of these inherent biases, low December fawn:doe ratios typically indicate either poor fawn production, low summer fawn survival, or a combination of both. To improve our current study design evaluating the importance of habitat/nutrition, we initiated a I-year pilot study in an attempt to obtain separate, direct measurements of in utero fetus survival and summer fawn survival for treatment and control does. These direct measurements, if obtainable, would be preferable to December fawn:doe ratios, and would provide a better understanding of the effect, if any, of the nutrition enhancement treatment on the deer population. The purpose of the pilot study was to evaluate the logistical feasibility of field techniques necessary to accomplish the research, and to collect data for subsequent sample size calculations assuming the research progressed into a full-scale study. In order to directly measure in utero fetus survival and summer fawn survival of radio-collared does from the treatment and control areas, we needed to record winter fetal rates of the collared does, and then locate and capture the collared does' fawns the following June. Winter fetal rates can be measured using established ultrasound techniques. However, there are no established techniques for locating and capturing a large sample of newborn fawns from specific, individual does. Previous attempts to capture neonates from radio-collared does in forest-shrub habitats have been largely unsuccessful (M.A. Hurley, Idaho Dept. of Fish and Game, pers. comm.; T. M. Pojar, Colorado Division of Wildlife, pers. comm.). There are 2 major problems: 1) there is no effective way to determine when any given doe will give birth to her fawn(s); and 2) it is often very difficult to find a fawn simply based on locating the doe, because the fawns are often bedded in heavy cover some distance from the doe (e.g. 50-100 yards). To overcome these problems, we conducted a I-year research study to evaluate the use of vaginal implant transmitters in adult does as a technique to determine both the timing and location of birthing. Vaginal Implant Transmitters For some time, radio-transmitter implants in the vaginas of deer have been considered as a technique for. locating and capturing newborn fawns from radio-collared does immediately following parturition. Early attempts to employ this technique were largely unsuccessful in terms of both effectiveness and animal �85 welfare concerns (Garrott and Bartmann 1984, Giessman and Dalton 1984, Nelson 1984). This early technique used sutures to partially close the vulva in order to retain the transmitter in the vagina. More recently, Bowman and Jacobsen ( 1998) developed and employed a modified vaginal implant transmitter (VIT) for white-tailed deer, which met better success. This transmitter had plastic wings to retain the transmitter in the vagina until parturition; thus, no sutures were used. They found no indications that animals were negatively impacted by the newly designed VIT; however, retention rate of implants to parturition was only 75%, and sample sizes were small. Within the last 2 years, several studies have been initiated using (modified) VITs to study white-tailed deer (M. Carstenson and G. D. Delguidice, University of Minnesota, pers. comm.), black-tailed deer (N. Pamplin and D. Jackson, Oregon State University and Oregon Dept. offish and Wildlife, pers. comm.), and elk (J. Noyes and B. Johnson, Oregon Dept. offish and Wildlife, pers. comm.; J. Vore, Montana Fish, Wildlife, and Parks, pers. comm.) These ongoing studies have found greater success with VITs in terms ofretention to parturition, and have not documented any detrimental effects to the animals. Given the success at finding birth sites and fawns, these studies do not indicate that vaginal implants cause major problems with in utero fetus survival or birthing. MATERIALS AND METHODS Experimental Design and Study Area This research was conducted in conjunction with our ongoing research study evaluating the effects of enhanced winter nutrition on overwinter mule deer fawn survival and early winter fawn:doe ratios. The research is being conducted in 2 experimental units on winter range on the Uncompahgre Plateau. The 2 units are receiving a nutrition enhancement treatment in a cross-over experimental design. Unit A served as the treatment unit, and Unit B served as the control, for the first 2 years of research (2000 - 2002). Beginning November 2002, Unit B will receive the treatment while Unit A will serve as the control. Bishop and White (2002) provide a complete description of the experimental design and study area. The 2 experimental units (A and B) receiving the nutrition enhancement treatment are (Fig. 1): (1) The Colona Tract of the Billy Creek State Wildlife Area (2) Bureau of Land Management lands adjacent to Shavano Valley as defined by the following: Within Dry Creek Basin Quadrangle (USGS 7.5 Minute), includes Sections 6 and 7 in T. 48 N.-R. 10 W. and Sections 1, 2, 10, 11, 12, 13, 14, 15 in T. 48 N.-R. 11 W. This area roughly includes 38°25'00" 38°27'30" Latitude and 108°00'00" - 108°04'30" Longitude. In late April and May, prior to fawning, deer from the winter range experimental units migrate to summer range. The summer range study area encompasses 800 mi2 covering the southern portion of the Uncompahgre Plateau and adjacent San Juan Mountains to the south and east (Fig. 1). Winter range elevations range from 1830 m (6000 ft) in Shavano Valley to 2318 m (7600 ft) adjacent to the Dry Creek Rim above Shavano Valley. Winter range habitat is dominated by pinyon-juniper with interspersed sagebrush adjacent to agricultural fields in the Shavano and Uncompahgre Valleys. Summer range elevations occupied by deer range from 1891 m (6200 ft) in the Uncompahgre Valley to 3538 m (11,600 ft) in Imogene Basin southwest of Ouray, CO. Summer range habitats are dominated by sprucefir, aspen, ponderosa pine, Gamb~l oak, and to a lesser extent, sagebrush and pinyon-juniper at lower elevations. �86 Sample Size The main objective of the pilot study was to assess "success/failure" ofVITs from an equipment functionality and field logistics standpoint. Our first definition of success/failure was the proportion of vaginal implant transmitters that were expelled at the birth site with the heat sensor functioning correctly (transmitter success rate). We set our sample size based on an initial assumption that 0.90 of the transmitters would function correctly, which necessitated 36 radio-collared does to estimate the transmitter success rate with a 95% confidence interval of+/- 0.10. Our second definition of success/failure was the proportion of successful fawn captures that occurred from the sample of 36 does (fawn capture success rate). We defined a successful fawn capture as locating and capturing at least 1 fawn from a doe equipped with a vaginal implant. Prior to the study, we assumed a 0.80 rate of finding at least 1 fawn/radio-collared doe. With 36 radio-collared does, this would allow us to measure a fawn capture success rate with a 95% confidence interval of+/- 0.13. This level of precision was sufficient for us to evaluate the overall success/failure of vaginal implant transmitters as a field technique for locat_ing and capturing newborn fawns from specific radio-collared does. Captur.e and Handling Techniques On February 28 and March 1, 2002, we captured a total of36 adult female deer utilizing helicopter net guns (Barrett et al. 1982, van Reenen 1982). Eighteen deer were captured in the nutrition enhancement treatment unit, and 18 in the control unit. Captured deer were ferried by the helicopter to a central processing location. Most deer captured in each experimental unit were chemically immobilized using a combination of ketamine and xylazine to facilitate the ultrasound and VIT insertion procedures. Ketamine and xylazine were mixed in a 5: 1 ratio (200:40 mg/ml), and administered intravenously at a dosage rate of approximately 1.5-2.0 ml/45 kg animal body mass. Immediately prior to release, drug effects were (partially) reversed with an intravenous injection of yohimbine at a rate of-12 mg/45 kg animal body mass. Each deer was aged based on tooth replacement and wear, and only deer ~-5 years old were retained. For each captured deer, we used ultrasonography to measure pregnancy status, fetal rate, and body condition. If the doe was pregnant, she was then radio-collared using a fixed length, permanent collar. Each radio collar had Ritchey® neck band material stitched to the left side with a unique identifier engraved on it for visual identification purposes. We then inserted the VIT and released the deer. We performed the ultrasound and VIT insertion procedures in a 10 x 12 ft wall frame tent located at the processing site to avoid problems associated with weather conditions and helicopter rotor wash. We also recorded the weight of each deer, recorded a body condition score, and collected a blood sample for serology tests. Ultrasonography We estimated body fat using an Aloka 210 (Aloka, Inc., Wallinford, Conn.) portable ultrasound unit with a 5 MHz linear transducer. Maximum subcutaneous fat thickness on the rump was measured immediately cranial to the cranial process of the tuber ischium. Proper orientation was assured by scanning along a line between the spine, at its closest point to the tuber coxae (hip bone), and the caudal process of the tuber ischium (pin bone). A small area of hair was shaved to ensure contact between the transducer and the skin. A conducting gel was applied to the shaved area and fat thickness was measured using electronic calipers. We quantified reproductive status using a 3 MHz linear transducer. To permit transabdominal scanning, a portion of the abdomen was shaved caudal to the last rib and -left of the mid line, and gel was applied. Both uterine horns were systematically scanned to identify fetal numbers ranging from Oto 3. Upon identification of a fetus, we measured, whenever possible, eye diameter, crown-rump length, biparietal �87 40 0 40 80 Miles Figure I. Map showing the locations of the Colona and Shavano experimental units (Units A and B) on winter range, and the location of deer summer range, on the Uncompahgre Plateau, southwest Colorado. The summer range study area was defined by the summer distribution of deer that were captured on the winter range experimental units. �88 distance, and skull length. In most cases, we only obtained an eye diameter measurement. Morphometric measurements were collected to estimate fetal age and parturition date. Vaginal Implant Transmitters (VITs) The VITs we used were manufactured by Advanced Telemetry Systems, Inc. (Isanti, MN). The VIT was 76 mm long, excluding antenna length, and had 2 plastic wings with a width of 57 mm when fully spread apart. The plastic wings were used to retain the transmitter in the vagina until parturition. The VIT weighed 15 grams and contained a 10-28 lithium battery. The diameter of the transmitter/battery was 14 mm, and was encased in an impermeable, water-proof, electrical resin. The transmitter contained an embedded heat-sensor which dictated the frequency pulse rate. When the heat sensor dropped below 86°F, synonymous with transmitter expulsion from the deer, the pulse rate changed from 40 PPM to 80 PPM. The VIT was inserted into deer using a vaginoscope (Jorgensen Laboratories, Inc., Loveland, CO) and alligator forceps. The vaginoscope was 6" long with a 5/8" internal diameter and had a machined end (smooth surface) to minimize trauma when inserted into the vagina. A discreet mark was placed on the applicator showing the appropriate distance it should be inserted into the deer. The length of a typical mule deer vaginal tract was obtained by taking measurements from road-killed deer and/or other fresh deer carcasses obtained in the study area. Prior to use in the field, VITs were sterilized using a Chlorhexidine solution, air-dried, and sealed in a 3" x 8" sterilization pouch. Sterilization containers with Chlorhexidine solution were used on site during capture to sterilize the vaginoscope and alligator forceps between each use. A new pair of nitrile surgical gloves was used to handle the vaginoscope and VIT for each deer. To insert a VIT, the plastic wings were folded together and placed into the end of the vaginoscope. We then liberally applied sterile KY Jelly to the scope and inserted it into the deer's vagina to the point where the mark on the applicator was reached. The a11igator forceps, which extended through the vaginoscope to hold the VIT, was held firmly in place while the scope was pulled out from the vagina. This procedure pushed the VIT out of the scope into the vagina, and the plastic wings spread apart to hold the transmitter in place. The transmitter antenna was typically flush with the vulva, but on occasion extended up to 1 cm beyond the vulva. The tip of the antenna was encapsulated is a wax bead to protect the deer. Adult Doe Monitoring From March through May, we regularly monitored the radioed does as part of our current research experimental design, which included daily monitoring for live/death status (Bishop and White 2002). We also used aerial telemetry to relocate each of the does every couple of weeks during the remainder of the winter. Fetus Survival and Neonate Capture During June 10-30, 2002, we relocated each of the radio-collared does having a VIT each morning using a fixed wing aircraft. Flights began at 6:00 AM and were usually completed by 10:00-11:00 AM. The early flights were crucial for detecting fast signals because shed VITs were often warmer than 86 °F by mid-day, which caused them to switch back to a slow ("pre-birth") pulse. When a fast ("postpartum") pulse rate was detected, we located the VIT from the ground to determine whether it was shed at the birth site. If the transmitter was located at the birth site, we identified whether any fawn(s) were stillborn. If the fawn(s) were no longer present at the birth site, or could not be found in the vicinity of the birth site, we located the radio-collared doe and searched for fawns at her location. All personnel involved wore surgical gloves to help minimize human scent when handling fawns. For each doe, we attempted to document whether any fawns were stillborn, locate each of her fawns, radio-collar and weigh the fawns, record basic vegetation characteristics of the birth site, and promptly exit the site. We attempted to �89 account for each doe's fetuses in order to evaluate the efficacy of using this technique to quantify in utero fetal survival from February to birth. We then radio-monitored each of the radio-collared fawns on a daily basis to measure survival rates of treatment and control fawns and to assess cause-specific mortality. We also periodically located other radio-collared does that did not have VITs and attempted to capture their fawns to help achieve our targeted sample size. Each of these does were part of the nutrition enhancement research, and were present on either the treatment or control experimental unit during winter. RESULTS AND DISCUSSION VIT Effectiveness The proportion ofVITs that were expelled at or near the birth site with the transmitter functioning correctly was 0.33 (SD= 0.083). Of36 VIT trials, 3 were censored, 7 were shed prematurely, 15 had battery failures, and 11 were successful. Censors: Two adult does died in May well before fawning and were still carrying the VITs. One doe was never relocated after leaving winter range. These 3 deer were censored because there was no test of whether the VIT functioned correctly or not. Premature Sheds: Three VITs were shed in May or early June well before fawning (May 18, May 19, and June 6). The other 4 VITs were shed during the fawning period, but at least 1-2 days before the respective does gave birth. Battery Failures: With only 1 exception, all battery failures occurred just before or during the fawning period. This was the glaring problem with the VIT success rate. The battery we had hoped to use had a warranty life of 116 days and a capacity of 232 days. Unfortunately, this battery was discontinued by Advanced Telemetry Systems (ATS) just before our research study began due to poor results. The battery we subsequently used in the VITs had a warranty life of only 94 days, and a capacity of 188 days. We needed our batteries to last 120 days for this research. ATS bad found good results with this shorter life battery, and recommended its use because it typically lasted well beyond the warranty battery life. We knew this was a risk at the onset of the research, but had confidence the batteries would last the necessary 120 days based on ATS recommendations. Had the batteries not failed, we likely would have bad a 60-70% success rate. Successes: We had 11 transmitters function correctly. One transmitter was still in the doe and functioning at the end of our capture period. Of the remaining 10 VITs, 9 allowed us to efficiently locate and capture fawns, typically at the birth site, and account for each of the given doe's fetuses measured in February/March. We located 14 fawns, one of which was a stillborn, from these 9 VITs. Our fawn capture success rate for the 33 available does was 0.61 (SD= 0.086), meaning we captured at least 1 fawn from 61 % of the VIT radio-collared does. In 3 instances, we opportunistically located a VIT with a failed battery by radio-tracking the doe and searching for her fawns. In total, we located 30 fawns from VIT does, 3 of which were stillborn, and 1 we weren't able to radio-collar. Fawn Capture We captured and radio-collared a total of 54 fawns from 38 adult does (1.42 fawns/doe) during June 11July 1, 2002 (Fig. 2). We found 4 stillborns at birthsites, and 2 suspected stillborns at or near birthsites which could have been early neonate mortalities. We captured 30 fawns from treatment does and 24 from control does. Twenty-six fawns were captured from 17 of the VIT does (1.53 fawnsNIT doe) and 28 fawns were captured from 21 radio-collared does that did not have VITs (1.33 fawns/non-VIT doe). We documented a total of 62 live fawns from the 38 does (1.63 fawns/doe), although we only captured 54 fawns because 1 fawn from a set of twins escaped on multiple occasions. �90 Capture Effort The 15 VIT battery failures caused considerable problems during the fawn capture. As it turned out, VITs helped us capture only 13 fawns and locate 1 stillborn fawn. The other 41 fawns and 5 stillborns were captured/located by routinely radio-tracking collared does and searching for fawns at their locations. This required an intensive field effort. We worked approximately 1700 man-hours (212 man-days) during a 22-day period to capture the 54 fawns, or roughly 4 man-days/fawn. Our objective pre-fawning was to capture 55 fawns; thus, even with the VIT failures, we were able to capture the necessary fawns from radio-collared does. The field effort would have been considerably less had we not had the battery failures. 8 7 "O "'::I - 6 i... 5 0. C;I u 4 -"' ,.. ~ ~ 3 ~ ::it: 2 1 0 I I 6110/02 6113102 II 6116102 6119102 I 6122/02 6125/02 I I 6128/02 7/1102 Figure 2. Number of newborn fawns captured by day during June 11-July 1, 2002. All fawns were captured from radio-collared does throughout the southern portion of the Uncompahgre Plateau and adjacent San Juan Mountains in southwestern Colorado. Fetus Survival In February-March 2002, we measured an average of 1.80 fetuses/doe (SE= 0.14, n = 36), which included 1.77 fetuses/doe (SE= 0.14, n = 18) in the treatment unit and 1.83 fetuses/doe (SE= 0.15, n = 18) in the control unit. In June 2002, considering all does that we located any fawn from, whether live or stillborn, we observed 1.42 (SE= 0.11, n = 43) live fawns/doe postpartum. This rate includes the 6 stillborns, and should represent a conservative estimate of live fawns/doe postpartum because we inevitably failed to locate all live fawns from each doe. In other words, this estimate would treat any unaccounted fetuses (from the February measurement) as if they were stillborns. For does that did not have VITs, and thus we did not have a winter fetus rate measurement, singletons would infer that either the deer only had I fetus, or that the other fetus died. It is likely that many of these singletons had a twin that we did not locate. This equates to a conservative fetus survival rate estimate of 0.79 (SE= 0.063). We accounted for all the fetuses of 14 VIT does in June 2002, 8 of which were the direct result of the VIT �91 functioning correctly. Of these 14 does, the fetus survival rate was 0.86 (SD= .096). This data point lacks precision and may potentially be biased because we did not account for an adequate number of fetuses due to the VIT failures. Of 10 VITs that functioned correctly and were shed during fawning, we accounted for all recorded fetuses in 8 of the deer. One of the other 2 VITs that functioned correctly allowed us to account for 2 of 3 fetuses measured in February. Clearly, to gain a more reliable estimate of fetus survival, a high percentage of the VITs must function correctly so that more birth sites can be located, and more fetuses can be accounted for. Neonate Survival We accomplished our neonate capture objectives even with the VIT failures; it simply required a greater effort. If this technique were incorporated into the nutrition enhancement experiment at full scale, we would need to capture approximately 80 newborn fawns (40 each from treatment and control does). Assuming we can purchase implants with longer-lived batteries, we could feasibly capture 80 fawns by increasing our sample size ofVIT does. CONCLUSIONS The VITs were largely successful except for the 15 battery failures. The battery problem can easily be corrected by working with Advanced Telemetry Systems to locate and utilize a reliable battery with a longer life. Such batteries exist, and have been used routinely in small mammal and avian radio transmitters. The 7 premature sheds were expected to some extent, and not a major concern. The VITs that did not fail were highly useful for determining the location and timing of birth, which is of critical importance for capturing fawns from individual, radio-collared does. We found the use ofVITs to be a successful field technique given our objectives, assuming a reliable, longer-lived battery can be incorporated into the current design. Literature Cited Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10: 108-114. Bishop, C. J., and G. C. White. 2000. Effects of habitat enrichment on mule deer recruitment and survival rates. Colorado Division of Wildlife, Wildlife Research Report, Federal Aid in Wildlife Restoration Project W-153-R-13, Progress Report. Fort Collins, CO, USA. Bishop, C. J., and G. C.. White. 2002. Effects of nutrition and habitat enhancements on mule deer recruitment and survival rates. Colorado Division of Wildlife, Wildlife Research Report, Federal Aid in Wildlife Restoration Project W-153-R, Progress Report. Fort Collins, CO, USA. Bowman, J. L., and H. A. Jacobson. 1998. An improved vaginal-implant transmitter for locating whitetailed deer birth sites and fawns. Wildlife Society Bulletin 26:295-298. Garrott, R. A, and R. M. Bartmann. 1984. Evaluation of vaginal implants for mule deer. Journal of Wildlife Management 48:646-648. Giessman, N. F., and C. J. Dalton. 1984. White-tailed deer fawn mortality in the southeastern Missouri Ozarks. Missouri Department of Conservation, Jefferson City, Pittman-Robertson Project W-13R-35. �92 Nelson, T. A. 1984. Production and survival of white-tailed deer fawns on Crab Orchard National Wildlife Refuge. Thesis, Southern Illinois University, Carbondale, IL, USA. Pojar, T. M. 2000. Investigating factors contributing to declining mule deer numbers. Colorado Division of Wildlife, Wildlife Research Report, Federal Aid in Wildlife Restoration Project W-153-R-13, Progress Report. Fort Collins, CO, USA. Pojar, T. M., and W. F. Andelt. 1999. Investigating factors contributing to declining mule deer numbers. Colorado Division of Wildlife, Wildlife Research Report, Federal Aid in Wildlife Restoration Project W-153-R-12, Progress Report. Fort Collins, CO, USA. Unsworth, J. W., D. F. Pac, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-421 in L. Nielsen, J.C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, Wisconsin, USA. � https://cpw.cvlcollections.org/files/original/821afeab84e170a005e542df8312032e.pdf 8ab60fcf9e2995d808e819786f0f1e27 PDF Text Text 93 Colorado Division of Wildlife Wildlife Research Report July 2001 and July 2002 PROGRESS REPORT State of_ _ _ _ _ _--=C-=o=lo=r-=a=do-"-------- Division of Wildlife - Mammals Research Work Package_ _ _ _~30~0~1~------ Deer Conservation Task No. 5 -------~------- Federal Aid Project No. Work Package No. Work Package No. Work Package No. Work Package No. Work Package No. Improved Population Modeling DEAMAN System Administration W-185-R Research and Development and the following non-Federal Aid projects 0661 Grouse Conservation 0664 Prairie Dog Conservation 0670 Lynx Conservation 0850 Peregrine Falcon Recovery 3006 Other Small Game Conservation Period Covered: July 1, 2001 - June 30, 2002 Author: G. C. White Personnel: C. Bishop, G. Miller, T. E. Remington, D. J. Freddy, T. M. Shenk, L. Stevens, J. Craig, R. Kahn, D. C. Bowden, F. Pusateri, J. Dennis, M. M. Conner, D. Walsh, B. Lubow. ABSTRACT Progress towards the objectives of this job include: Consulting assistance to CDOW on harvest surveys, terrestrial inventory systems, and population modeling procedures was provided. Estimates of spring and fall turkey, spring snow goose, sharptailed and sage grouse, chukars, ptarmigan, Abert's squirrels, and general small game harvest were computed from survey data, and programs and harvest estimates provided to CDOW via email and CD ROM. Computer code written in SAS to compute these estimates and display results graphically was also provided. Computer code was also written in SAS to estimate the compliance rate of Colorado small game license holders with the Harvest Information Program. The DEAMAN software package for the storage, summary, and analysis of big game population and harvest data was revised further as a Windows 95/98/NT/2000/ME/XP program. The capability to incorporate data on radio-collared animals to estimate survival with the Kaplan-Meier estimator and display movement data was added, and distributed to terrestrial biologist via the WWW at http://www.cnr.colostate.edu/~gwhite/deaman. A 3-day workshop was conducted with CDOW Terrestial Biologists in the use ofDEAMAN and population modeling procedures, mainly to instruct personnel on the use of spreadsheet models for ungulate population dynamics. In addition, numerous questions were answered via meetings with biologists, and via email. A paper, coauthored with Bruce Lubow, was published in the Journal of Wildlife Management on past efforts to develop a realistic mule deer population model based on data collected with current CDOW procedures. Data from the Piceance Basin were used to illustrate the modeling technique. The full '· BDOW016783 �94 citation is: White, G. C., and B. Lubow. 2002. Fitting spreadsheet population models to multiple sources of observed data. Journal of Wildlife Management 66:300-309. A paper on use of population viability analyses applicable to animals monitored with mark-encounter data was published with T. M. Shenk and A. B. Franklin. The full citation is: White, G. C., A. B. Franklin, and T. M. Shenk. 2002. Estimating parameters of PVA models from data on marked animals. Pages 169-190 in S. R. Beissinger and D. R. McCullough, editors. Population Viability Analysis. University of Chicago Press, Chicago, Illinois, USA. A paper on analysis of radio-tracking data for estimation of survival pertinent to monitoring the reintroduced lynx population in Colorado was published with T. M. Shenk: White, G. C., and T. M. Shenk. 2001. Population estimation with radio-marked animals. Pages 329-350 in J. J. Millspaugh and J.M. Marzluff, editors. Design and Analysis of Wildlife Radiotelemetry Studies. Academic Press, San Diego, California, USA A paper on the estimation of population size from correlated sampling unit estimates of the variable of interest was submitted to the Journal of Wildlife Management. The methodology developed in this paper is proposed for use in a joint Colorado/Utah survey of the colony area of white-tailed and Gunnison prairie dogs in western Colorado and eastern Utah. The full citation is: Bowden, D. C., G. C. White, A. B. Franklin, and J. L. Ganey. 2003. Estimating population size with correlated sampling unit estimates. Journal of Wildlife Management 67:1-10. A paper on the estimation of survival of Greater Sage-grouse in North Park, Colorado, was submitted to the Journal of Wildlife Management: Zablan, M.A., C. E. Braun, and G. C. White. 2003. Estimation of northern sage-grouse survival in North Park, Colorado. Journal of Wildlife Management 67:144154. Assistance in the analysis of candidate systems to estimate deer abundance in GMU IO was provided. A research study to examine the impact of nutrition on the decline of mule deer fecundity during the last 20 years was continued. I have provided input on estimation of the number of deer on the feed sites, and developed an estimator of fawn survival rates based on radio-collared does and fall and spring fawn:doe ratios. Data were collected and analyzed on spatial distribution, movement of radio-collared animals, and population sizes related to estimating the spread and impacts of chronic wasting disease in deer populations. A report summarizing these findings was provided to CDOW personnel involved with the study. A graduate research project by Dan Walsh to evaluate utility of Iek counts of Greater Sage-grouse in Middle Park is ongoing. Mark-resight methods are being used to estimate lek attendance and population size. Preliminary results of this work were reported at the Sage-grouse Working Group Meeting on Population Analysis held June 24-25, 2002, in Torrey, Utah. Computer programs to assist researchers with field data collection of feeding rates of tame Greater Sagegrouse were written for the HP Jornada Pocket Computer. In addition, a program for computing triangulation Iocations:ofradioed animals was:wrltte·n for the HP Jorilada Pocket Ccimputer for use in the field to evaluate interactively and graphically the quality of triangulated locations. A model of the Colorado peregrine falcon population was constructed from estimates of survival and reproduction derived from banded birds (1973-2001) and monitored nests (1989-2001). Data were supplied by Jerry Craig. Survival estimates for 0-1, 1-2, and 2+ year old birds were 0.544, 0.670, and 0.800, respectively, with standard errors of 0.0765, 0.091, and 0.0544. Average young produced per pair was 1.660 (SE= 0.0443), but there was considerable variation across years (min =·I.388, SE= 0.1548, in 1995; max= 2.122, SE= 0.1393, in 2000). Based on a population model constructed from these estimates, the annual rate of population increase was 1.028957 for females first reproducing at 3 years of age, and 1.080316 for females first reproducing at 2 years of age. Given these high rates of population increase, some take by falconers could be accommodated by the Colorado peregrine population. An analysis to estimate the effort required to estimate the percent of eastern Colorado inhabited by blacktailed prairie dogs was completed and results provided to CDOW personnel involved with the effort. For the first 5 strata surveyed, I have computed estimates of prairie dog colony areas to assist the CDOW personnel conducting the survey. Results to date suggest the survey is working well. �95 CONSULTING SERVICES FOR MARK-RECAPTURE ANALYSES G. C. White PROJECT OBJECTIVES Assess the status of Colorado peregrine falcon population based on parameters estimated from banded birds and monitored nests. SEGMENT OBJECTIVES 1. Develop a model of the Colorado peregrine falcon population based on survival and reproduction estimates derived from banded birds (1973-2001) and monitored nests (1989-2001). 2. Using this model, determine the impact oflimited take by falconers on Colorado peregrine populations. RESULTS AND DISCUSSION Methods Survival Rate Estimation A total of938 peregrines were banded as nestlings during the interval 1974-2000. From these, 11 live resightings and 53 dead recoveries were obtained. Survival was estimated with Program MARK (White and Burnham 1999) using the joint live and dead recoveries model of Burnham (1993). Program MARK uses the Seber (1970) parameterization for dead recoveries in the Burnham model, so parameters for survival (S), probability that a band from a dead bird is recovered (r), and the probability of a live bird being resighted and the band read (p). The fidelity parameter (F) was fixed to 1 because of the sparseness of the data. Reproduction A total of 142 nesting sites were monitored for the number of young fledged starting in 1973 through 2001, although the number of sites increased with year as the breeding population increased in Colorado. Only data from 1989 through 2001 were used in the analysis presented here because I wanted at least 30 nests per year to estimate the effects modeled. Mean number of young fledged per site was estimated by year. Estimates of the variance components by site and year was est_imated with PROC MIXED of SAS ·(Littell et al. 1996). Structures considered for the variance of the repeated observations within sites were first order autoregressive, first order heterogeneous autoregressive, compound symmetry, heterogeneous compound symmetry, exponential local effects, also known as dispersion effects, in a log-linear variance model, and a null structure with zero covariances and constant variances across years, also commonly known as the variance components structure (Littell et al. 1996). Random effects considered in the models were year and site effects. A fixed effect for even and odd years was also included in the models because there is a defined sequence of a year of high reproductive output, followed by a year of low reproductive output, followed by another year of high reproductive output, etc. The even/odd year fixed effect models this oscillating reproductive output. Selection among models for both survival and reproduction parameter estimation was performed with the AICc criterion recommended by Burnham and Anderson (1998). �96 Population Model A model of the female segment of the Colorado peregrine population based on the estimates of survival and reproduction was constructed in an Excel spreadsheet, and also as a SAS program. The model included 4 age classes (No, N 1, N 2, and N 3), even though the survival estimates used in the model had survival the same for all birds 3+ years old. The parameters in the deterministic model were number of fledglings per reproducing female (F), proportion of fledglings that are female (assumed to be SR= 0.5), survival for 4 age classes (So, S 1, S2, and S3), and the proportion of females that breed on their second birthday (B2). In addition, a parameter for the proportion of fledglings removed by falconers was included in the model to allow the estimation of the effects of human take (1). The difference equations for the transition from year t to year t+ 1 in model were: N1.1+1=No., So, No.1+1=(B2 N2.1+1+ N3.1+1J F SR (1-T). A stochastic model was developed from the above deterministic model by replacing the mean fledglings per female with a value randomly drawn from the observed values for the years 1989 through 2001. The VLOOKUP function of Excel was used to randomly select one of the 13 observed values for each year in the model. That is, the fledgling rate was sampled with replacement from the observed values. This stochastic reproduction model was also extended to incorporate demographic stochasticity in the survival process. Instead of multiplying the population segment by a fixed survival rate to obtain the number of survivors, the process is treated as a binomial process, with the number of survivors drawn from a binomial distribution with the appropriate survival rate. However, given the size of the peregrine population in Colorado, demographic stochasticity had little or no effect on the model predictions. Results Survival Rate Estimation The encounters_ of_:rparked birds were spar!,~, and thus preclude complex models involving both time and age effects. The a priori list of models· co'risidered (Table 1) included models to evaluate age differences in survival for birds during their first 4 years of life, and time-specific effects in survival, live resighting rate, and dead recovery rate. The minimum AICc models all included time-specific variation in the band recovery rate, but did not suggest time-specific variation was required to model the live resighting rates. The minimum AICc model was {S(a2) p(.) r(t)}, with the second best model {S(a3) p(.) r(t)} only 0.388 units above the minimum. I chose to use estimates from the 3-age model because this model is more realistic biologically, and the small difference in AICc values does not suggest that the 2age class model is much better than the 3-age class model. Parameter estimates for the 3-age class model (Table 2) are reasonable in that survival increases with age, but all have large standard errors. Reproduction Estimation Average number of fledglings produced per nest during the interval 1989-2001 was 1.66059 with SE 0.044296 (Table 3). However, as shown in Table 3, there is considerable variation from year to year in the number of fledglings per breeding pair. �97 Model selection results (Table 4) for estimation of the variance components for site and year suggest that a first order autoregressive variance structure is required. Based on the minimum AI Cc model with the AR(l) variance structure, the estimate of the variance component for year was 0.01609, for site was 0.08438, with a residual variance component of 1.6069. The autocorrelation coefficient between consecutive years within a site as 0.1109. Thus, the variance components due to year and site were relatively minor compared to the residual variance in young fledged per breeding pair. The even/odd year effect was estimated as 0.2482 with a SE of0.1065 (P < 0.0403). Thus, the magnitude of the every-other-year oscillation is estimated to be 0.2482 young per nest. Population Model The deterministic population model provided an estimate of A = 1.028957 with T= 0 and F= 0 for the survival parameter values in Table 2 and fledglings per reproducing female of 1.66059, estimated as the mean fledgling rate for the 1989-2001 interval. Thus, the model predicts that the Colorado peregrine population is increasing 2.9% per year, even with no 2-year old birds reproducing. The effect ofreproduction from 2-year birds is shown in Figure 1, and suggests that if all of the 2year old birds breed, A = 1.080316. Results from the model with stochastic reproduction provided estimates of A consistent with the deterministic model, e.g., for 10,000 simulations, A = 1.0290287 with SE= 0.000631839, giving a 95% confidence interval of 1.0277902 to 1.0302673 that encompasses the value estimated from the deterministic model. The large standard errors of the parameter estimates used to build the population model (Tables 2 and 3) suggest that the estimate of A obtained will also have a large standard error. I used Monte Carlo simulation techniques to draw values of each of the survival and reproduction parameters from a normal distribution with mean equal to the parameter estimate and standard deviation equal to the standard error of the parameter estimate. With a single set of parameters so obtained, 10,000 values of A were averaged to obtain a mean for that parameter set. This process was repeated for 1000 parameter sets to obtain a SD of A of 0.0612896. This value can be interpreted as a SE of the estimated A that accounts for the SE of the input parameters. Average standard errors from the 10,000 simulations for the 1000 parameter sets was 0.0015142, suggesting that the preceding SD is not affected by the small amount of variation associated with each of the 1000 estimates. Discussion The best AI Cc model for estimation of survival required 28 parameters to estimate time-specific band recovery rates, r(t). A more parsimonious model would provide estimates of survival with better precision. One approach to obtaining a more parsimonious model would be to include a covariate that models the variation in r(t). The estimated rate of increase from the population projection model is relatively high for a wildlife population. However, the number of nest sites monitored each year (Table 3) provides confirmation of the high estimates of A. An exponential model regression, log.,(No. Nest Monitored)= /3 0 + /3 1 Year, was used to estimate the rate of increase of numbers of nest monitored. From this regression, jJ 1 = 0.07883 with SE= 0.00755 (P < 0.0001), giving an estimate of the annual rate of increase of the population (A) of exp(0.07883) = 1.082. This value exceeds the maximum value predicted from the population model even with 100% of the 2-year old birds breeding. Although the �98 estimate obtained from the number of nests monitored suffers from confounding with monitoring effort and effort to find new nests, the results still suggest that the Colorado peregrine population has had a high annual rate of increase, and that the predictions from the population model described here are consistent with another estimate of A . The high annual rate of increase suggests that moderate take can be accommodated by the Colorado peregrine population without affecting the population growth rate. Based on the population model presented with no 2-year old birds breeding, A = l with 17.5% of the fledged young taken. With 50% of 2-year old birds breeding, A = l with 26. 7% of the fledged young taken, and with 100% of 2year old birds breeding, A = l with 34.0% of the fledged young taken. Summary A model of the Colorado peregrine falcon population was constructed from estimates of survival and reproduction derived from banded birds (1973-2001) and monitored nests (1989-2001). Survival estimates for 0-1, 1-2, and 2+ year old birds were 0.544, 0.670, and 0.800, respectively, with standard errors of 0.0765, 0.091, and 0.0544. Average young produced per pair was 1.660 (SE= 0.0443), but there was considerable variation across years (min= 1.388, SE= 0.1548, in 1995; max= 2.122, SE= 0.1393, in 2000). Based on a population model constructed from these estimates, the annual rate of population increase was 1.028957 for females first reproducing at 3 years of age, and 1.080316 for females first reproducing at 2 years of age. Given these high rates of population increase, some take could be accommodated by the Colorado peregrine population. Literature Cited Burnham, K. P. 1993. A theory for combined analysis of ring recovery and recapture data. Pages 199-213 in J.-D. Lebreton and P. M. North, editors. Marked individuals in the study of bird population. Birkhauser Verlag, Basel, Switzerland. Burnham, K. P., and D.R. Anderson. 1998. Model selection and inference: a practical information-theoretic approach. Springer-Verlag, New York, New York, USA. 353 pp. Littell, R. C., G. A. Milliken, W.W. Stroup, and R. D. Wolfinger. 1996. SAS® System for Mixed Models. SAS Institute Inc., Cary, NC, USA. 633pp. Seber, G. A. F. 1970. Estimating time-specific survival and reporting rates for adult birds from band returns. Biometrika 57:313-318. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 Supplement:120-138. �99 Table 1. Set of a priori models considered in Program MARK (White and Burnham 1999) for estimation of survival of peregrines banded as nestlings with the joint live-dead model of Burnham (1993). Model names include survival (S), live resighting probability (p) and probability that the band from a dead bird is recovered (r). For survival, models with time-specific survival (t), constant survival(.), and 2 (a2), 3 (a3), and 4 (a4) age classes were considered. Constant and time-specific models for both p and r were also considered. The fidelity parameter (F) was fixed to 1 in all models. Model AICc Delta AICc AICc Weights Model Likelihood Num. Par Deviance 31 161.38 {S(a2) p(.) r(t)} 710.841 0 0.46193 {S(a3) p(.) r(t)} 711.229 0.388 0.38047 0.8237 32 159.626 {S(a4) p(.) r(t)} 713.332 2.491 0.13294 0.2878 33 159.583 {S(.) p(.) r(t)} 716.701 5.86 0.02467 0.0534 30 169.377 {S(a3) p(.) r(.)} 742.565 31.724 0 0 5 247.202 {S(a2) p(.) r(.)} 742.803 31.962 0 0 4 249.462 {S(t)p(.) r(.)} 744.811 33.97 0 0 30 197.487 {S(.) p(.) r(.)} 753.754 42.913 0 0 3 262.429 {S(a4)p(.) r(.)} 886.175 175.334 0 0 6 388.786 �100 Table 2. Parameter estimates from the 3-age class model {S( a3) p(.) r(t)} of peregrine falcons in Colorado, estimated from birds banded 1974-2000. Parameter Estimate F (fixed to l) SE LCI UCI 0 Sage 0-1 0.543995 0.076538 0.394531 0.685934 Sage 1-2 0.669762 0.098121 0.459491 0.828723 Sage 2-3+ 0.800291 0.054382 0.672873 0.886454 p 0.005662 0.002265 0.002581 0.012374 r 1974 0 lE-07 -3E-07 3E-07 r 1975 0 5E-07 -l.IE-06 l.IE-06 r 1976 0 3E-07 -7E-07 7E-07 r 1977 0.42494 0.275041 0.07526 0.870289 r 1978 0.781612 0.215533 0.231516 0.977021 r 1979 0.20686 0.101744 0.071797 0.467918 r 1980 0.190922 0.093857 0.066924 0.437053 r 1981 0 0 -lE-07 lE-07 r 1982 0 0 0 0 r 1983 0.074126 0.052037 0.017792 0.261367 r 1984 0.035033 0.034906 0.004775 0.215516 r 1985 0 0 0 0 r 1986 0.042528 0.042355 0.00575 0.254375 r 1987 0 0 0 0 r 1988 0.042219 0.029772 0.010304 0.157271 r 1989 0.016702 0.016695 0.002311 0.11077 r 1990 0.019441 0.01942 0.002685 0.127406 r 1991 0.021935 0.021894 0.003025 0.142181 r 1992 0.142769 0.053181 0.066349 0.280741 r 1993 0.01603 0.01601 0.002223 0.106429 r 1994 0.055134 0.031693 0.017401 0.161262 r 1995 0.083715 0.041537 0.030642 0.208902 r 1996 0.034717 0.024458 0.008529 0.130709 r 1997 0.060956 0.034981 0.019218 0.176984 r 1998 0 0 0 0 r 1999 0.040953 0.041348 0.005395 0.251596 r2000 0.052539 0.053344 0.006742 0.311771 r2001 0.069547 0.07131 0.008547 0.39323 �101 Table 3. Summary of number of ~oung fledged Eer nest for Colorado Eeregrines, 1989-2001. Year Number of Nests Fledglings/Nest SD SE 1989 33 1.90909 1.35471 0.235825 1990 42 1.47619 1.15269 0.177864 1991 52 1.65385 1.16963 0.162198 1992 54 1.62963 1.29289 0.17594 1993 55 1.65455 1.36379 0.183893 1994 64 1.625 1.26617 0.158271 1995 67 1.38806 1.26677 0.154761 1996 83 1. 71084 1.29285 0.141909 1997 71 1.42254 1.26109 0.149664 1998 81 1.95062 1.27379 0.141532 1999 88 1.59091 1.33594 0.142412 2000 98 2.12245 1.37927 0.139327 2001 90 1.35556 1.36003 0.14336 Mean 878 1.66059 1.31254 0.044296 Table 4. Model selection results for estimation of variance components of year and site with the MIXED procedure of SAS (Littell et al. 1996). Variance Structure Fixed Effects Random Effects AICc Delta AICc AR(l) Even/Odd Years Year Site 2950.4 0 AR(l) Even/Odd Years 2951.8 1.4 AR(l) Year Site 2952.3 1.9 AR(l) Year 2953.7 3.3 Default Even/Odd Years Year Site 2954.6 4.2 cs Even/Odd Years Year Site 2954.6 4.2 Default Year Site 2956.5 6.1 cs cs Year 2956.5 6.1 Year Site 2958.5 8.1 AR(l) Site 2961.3 10.9 2962.8 12.4 AR(l) ARH(l) Even/Odd Years Year Site 29702 19.8 EXP(YEAR) Even/Odd Years Year Site 2973.6 23.2 �102 - 1.09 ~ - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ~ _g 1.08 .c ~ 1.07 : 1.06 ra fu 1.05 C: o 1.04 s ~ 1.03 1.02 - + - - - - - - - - . - - - - - - - , - - - - - - ~ - - - - - ~ - - - - ~ - - - - - - - ! 0.6 0 0.2 0.4 0.8 1 1.2 Proportion of 2-year olds Breeding Figure I. Predicted effect from the population model of the proportion of2-year old females breeding on the rate of increase in the population ( A ). � https://cpw.cvlcollections.org/files/original/692941f935a0f48eb05b20d1ed7ee5e6.pdf b9a962c77578f5367ea5688da1291ac8 PDF Text Text 103 Colorado Division of Wildlife Wildlife Research Report July 2001 and July 2002 JOB PROGRESS REPORT State of _ _ _ _ _ ____:C:::.;o""l""'o""'ra~d::..,o"------- Mammals Research Work Package -------=3:....:0:...::0"""1_ _ _ _ _ __ Deer Conservation Study No. _ _ _ _ _ _____::;9_ _ _ _ _ __ Evaluation ofGnRH-PAP as a Long-term Fertility Control Agent in Female Mule Deer (Odocoileus hemionus hemionus) Federal Aid Project No. W-153-R-13 Research and Development Period Covered: July 1, 2001 -June 30, 2002 Author: Dan L. Baker, Ph.D. Personnel: M.A. Wild, T. M. Nett, T. Davis, E. Jones, B. Hochmoth ABSTRACT We evaluated the effects of GnRH-PAP on mule deer pregnancy rates, duration of suppression of luteinizing hormone and progesterone secretion, blood chemistry and hematology, and reproductive behavior during 1 November to 30 December, 2001. Twenty-two adult female mule deer were assigned to one of 3 experimental groups. Nine female mule deer were treated with GnRH-PAP and 9 females served as untreated controls. The dose of GnRH-PAP used in this experiment did not lower pregnancy rates in female mule deer. Treated and control females tested positive for pregnancy specific protein B on all sampling dates and all delivered healthy fawns in July. At 30 days posttreatment, luteinizing hormone and progesterone were not different (P > 0.58) in treated and control mule deer. Reproductive behaviors of GnRH-PAP treated females were not different from controls. We conclude that the dose of GnRH-PAP administered in this experiment was ineffective in suppressing reproduction in female mule deer. ,i·.. • : ( ,.:;._ ... ' �105 EVALUATION OF GnRH-PAP AS A LONG-TERM FERTILITY CONTROL AGENT IN FEMALE MULE DEER (ODOCOILEUS HEM/ONUS HEM/ONUS) Dan L. Baker P. N. OBJECTIVES 1. Develop a practical and acceptable technology for long-term control of fertility in female mule deer. 2. Demonstrate the feasibility of controlling mule deer population growth in a field application. SEGMENT OBJECTIVES 1. Evaluate the effectiveness of GnRH-PAP in preventing pregnancy in captive female mule deer. 2. Evaluate the duration of GnRH-PAP suppression of LH and progesterone secretion in female mule deer. 3. Assess the behavioral and physiological side-effects (if any) of GnRH-PAP in captive female mule deer. INTRODUCTION Controlling the growth of animal populations is fundamental to maintaining proper balance between wildlife and the habitats they occupy. This is particularly true for wild ungulates. Overabundant ungulates can cause serious degradation of plant communities, and preventing such damage requires controlling their numbers. Hunting has traditionally been used to control ungulate populations but there are increasingly more situations where hunting is infeasible. Such areas include urban areas where safety of people and property may be threatened, or national parks and refuges where populations are managed primarily for non-consumptive uses like wildlife viewing and photography or on military installations and industrial parks because of concerns for security. In these situations, alternatives to hunting or culling as a means of controlling ungulate numbers are needed. Fertility control offers a potenti~I alternative for controlling the growth of overabundant ungulate populations when traditional methods are infeasible or unacceptable (Kirkpatrick and Turner, 1985; Bamford, 1990; Garrot, 1995). However, current technology does not provide a means for controlling populations that is practical, economical and without undesirable side-effects (reviewed by Fagerstone et al., 2001). For most free-ranging wild ungulate populations, permanent sterilization has been proposed as the most efficacious approach to population management (Hone 1992, Garrot 1995, Hobbs et al. 2000). A promising new non-steroidal, non-immunological approach to permanent infertility involves analogs of gonadotropin-releasing hormone (GnRH). GnRH is a molecule produced in the hypothalamus of the brain. It directs specific cells in the anterior pituitary gland to synthesize and secrete two important reproductive hormones; follicle stimulating hormone (FSH) and luteinizing hormone (LH). These latter two hormones, known as gonadotropins, control the proper functioning of the ovaries in the female and the testes in the male. �106 Analogs of GnRH have the potential to permanently inhibit reproduction. By coupling a superactive analog of GnRH to a cytotoxin, it should be possible to specifically target that toxin to LH and FSHsecreting cells in the anterior pituitary gland (Collier and Kaplan 1984, Pastan et al. 1986). Therefore, a single GnRH-toxin conjugate has the potential to induce sterility in both sexes and numerous species of animals. There are many natural cytotoxins available for conjugating to GnRH. Toxins are composed of two subunits, a toxic subunit and a binding subunit. In order to target the toxin to specific cell types (rather than all cells) within the body, the binding subunit of the toxin can be removed and replaced by a molecule that will bind to only one cell type, in our case an analog of GnRH. This will target the toxin to gonadotropin secreting cells in the anterior pituitary gland. This approach has several potential advantages over other methods of contraception. These include: 1) a single treatment should permanently sterilize an animal; 2) the same treatment should be effective in both males and females and in different vertebrate species; 3) GnRH-toxin conjugate is a protein and is metabolized from the body within a few days of treatment, therefore it poses no threat to non-target species; 4) the small volume required for contraception facilitates microencapsulation and administration by syringe dart or biodegradable bullet. Proposed Research To our knowledge, only limited investigations have been conducted with GnRH-PAP in wild ungulates (Nett et al. 2001). In order to provide an estimate of the dose ofGnRH-PAP conjugate required for contraception, it is essential that the potency of GnRH analog be determined in each species and at different phases of the reproductive cycle. We addressed this question in a series of GnRH challenge trials with captive mule deer at the Foothills Wildlife Research Facility in Fort Collins, Colorado (Baker et al. 1996). In these experiments, we determined the most effective dose of GnRH analog in female mule deer during the breeding season to be 1 µg/50 kg BW. This is the minimum dose of GnRH analog that will illicit maximum LH secretion. The next questions that needed to be answered were "how effective is GnRH-PAP in preventing pregnancy, how durable are its effects over time, and are there unacceptable side effects? The objective of this experiment is to begin to address these questions. Specifically, our objectives were: 1) to evaluate the effectiveness ofGnRH-PAP in preventing pregnancy in mule deer; 2) to evaluate the duration of GnRH-PAP suppression ofLH and progesterone secretion; 3) to assess the behavioral and physiological side-effects (if any) of GnRH-P AP treatment. MATERIALS AND METHODS Reproductive Biology Mule deer (Odocoileus hemionus hemionus) are polytocous, multiovular, spontaneous ovulators that exhibit highly seasonal patterns of reproduction that are controlled by photoperiod regimens. The onset of the breeding season occurs during decreasing daily photoperiods of autumn and is preceded by a period of deep an estrous in summer (Plotka et al.1977). The first ovulation of the breeding season is usually preceded by one or more silent ovulations associated with the formation of short-lived corpora lutea that serve to synchronize the first overt estrus within a herd (Thomas and McT.Cowan 1975). In temperate North America, the majority of conceptions occur in late November, but recurrent estrous �cycles of 24 -28 days are possible through March if females fail to conceive (Knox et al. 1988). In early spring, coincidental with increasing day length, reproductive cycles cease and females remain anestrous until October. For pregnant females, parturition generally occurs in late May or early June, after a gestation period of about 200 days (Anderson and Medin 1966). Most females produce one fawn when they are two years old, and one or two annually thereafter (Cowan 1956). Experimental Design We evaluated the effects of GnRH-PAP on mule deer pregnancy rates, duration of suppression of LH and progesterone secretion, blood chemistry and hematology, and reproductive behavior during 1 November to 25 December 2002. Twenty-two adult female and 2 adult male mule deer were used in this experiment. Females were assigned to one of 3 experimental groups based on their tractability for handling and blood sampling. Nine female mule deer (Group A) were treated with GnRH-PAP and 9 females (Group B) served as untreated controls and were used to compare pregnancy rates, blood chemistry and hematology, and reproductive behavior to those of treated animals. Immediately following the pretreatment GnRH challenge trial, and before catheters were removed, 9 females in Group A were administered an optimum dose ofGnRH-PAP (lµg/50kgBW) N. Females in Groups A and B were maintained together with 2 adult male mule deer in a 2 hectare pasture. The remaining 4 females (Group C) served as untreated, non-pregnant controls and were placed in a separate pasture (0.5 ha) without direct contact with male deer. We compared LH and progesterone secretion of these females to those treated with GnRH-PAP (Group A). Sample size requirements were based on the variances observed in these measurements from previous studies with captive mule deer and expected effectiveness of treatment (Baker et al. 1996, Nett et al. 2001). Experimental Animals In order to meet sample size requirements calculated for this experiment (see pages 5-7), 5, adult, freeranging female mule deer were captured from urban areas of the front range of Colorado and transported to FWRF. We attempted to capture the most human-habituated deer as possible in order to minimize any stress related to captivity. All deer were captured and handled under the supervision of a veterinarian using one of several previously approved methods (Conner and Miller, CDOW ACUC 12-1999 & addenda); however thiafentanil oxalate (0.1 mg/kg) was substituted for carfentanil citrate as a primary tranquilization drug. Once tranquilized, does were blindfolded, condition and vital signs checked, eartagged, collared, vaccinated with 7-way clostridial v~ccine, treated with Ivermectin (0.1 mg/kg) and long-acting penicillin. Each doe was then placed in a closed vehicle (covered horse trailer), and sedation reversed and blindfold removed. After release at FWRF, deer were observed daily for any signs of post-capture injuries. Response Measurements Hormonal evaluation. Prior to application of GnRH-PAP, we measured the LH response of each female in Groups A and C to a challenge dose of GnRH analog. Results from this trial provided a pretreatment baseline for comparison to future posttreatment LH responses. This and succeeding LH challenge trials were conducted as follows: On day 1 of the trial, deer were moved from 2 ha pastures to individual isolation pens, sedated (4-6 ml, 2:1 ketamine (200 mg/ml):xylazine hydrochloride (100 mg/ml, IM), and fitted nonsurgically with indwelling jugular catheters. Animals were reversed with yohimbine (0.125 mg/kg, N). On day 2, we administered GnRH analog (1µ /50 kg BW) through the cannula and �108 collect blood samples (5 ml) at 0, 60, 120,180,240,300,360, and 480 minutes postinjection. Following the last blood collection, catheters were removed and each animal given an antibiotic (ceftiofur, (1 mg/kg, IV). Animals were then returned to 2 ha pastures. Serum was.stored at -20 °C until analyzed for LH (Niswender et al. 1969). The duration of contraceptive effectiveness was assessed by conducting similar GnRH challenge trials each month from November, 2001 to December 2003. Analysis. Responsiveness of the pituitary to GnRH challenge was assessed in three ways: 1) maximum LH (ng/ml) response achieved postinjection minus baseline, 2) time required to reach maximum LH, and 3) total amount ofLH secreted (ng/ml/min). Pregnancy rates. We assessed contraceptive effectiveness by determining the pregnancy rates of treated (Group A) and control (Group B) deer. A single blood sample (10 ml) was taken via jugular venipuncture from each animal for pregnancy-specific protein B (PSPB) analysis approximately 60, 90, and 220 days post-conception (Willard et al. 1998). Animal handling and blood collections for PSPB followed methods previously described for hormonal assessment and were collected in conjunction with these measurements. Neonates born to any experimental animal were incorporated into the resident FWRF mule deer herd. General health. Limited knowledge of the effects of GnRH-P AP on nutrition, body weight dynamics, blood chemistry and general health of mule deer have been reported in a previous study at this facility (Nett et al. 2001). However, since a different toxin conjugate is being tested in this experiment, we evaluated these potential side-effects here as well. We assessed physiological side-effects of GnRH by comparing serum chemistry, hematology, and body weight dynamics of treated (Group A) and untreated, non-pregnant mule deer (Group C). Blood collections and body weight measurements were made in conjunction with GnRH challenge trials. Blood samples for hematology and serum chemistry analysis were collected prior to treatment and at 90 days posttreatment then submitted for analysis to Colorado State University, Veterinary Teaching Hospital, Clinical Pathology Laboratory, Fort Collins, Colorado, USA. Serum chemistry profiles were obtained using a Hatachi 917 autoanalyzer (Roche/Boehringer Mannheim, Indianapolis, Indiana, USA) for the following parameters: glucose, creatinine, phosphorus, calcium, magnesium, total protein, albumin, globulin, albumin/globulin ratio, bilirubin, creatinine kinase, aspartate aminotransferase, gamma-glutamyltransferase, sorbitol dehydrogenase, sodium, potassium, chloride, and biocarbonate. Values for the following hematological parameters were obtained using an ADVIA 120 autoanalyzer (Bayer Corporation, Tarrytown, New York, USA): nucleated cells, neutrophils, lymphocytes, monocytes, eosinophils, plasma protein, erythrocyte, hemoglobin, packed cell volume, mean corpuscular volume, mean corpuscular hemoglobin concentration, platelets, and fibrinogen. Reproductive behavior. The effectiveness of GnRH-PAP as a, fertility control agent is dependent upon permanent suppression of ovulation and steroidogenesis. Thus, we tested 2 hypotheses relative to the effects of leuprolide on reproductive behavior of mule deer: (1) because GnRH-PAP is expected to suppress gonadotrophin secretion and ovulation, we predicted that sexual interactions during the breeding season (Nov 1 - Dec 20) would be reduced in treated females (Group A) compared to untreated controls (Group B), and (2) once untreated females become pregnant, reproductive behaviors would cease and sexual interactions would be similar between untreated and treated females during the postbreeding season (Jan 10 - Mar 31 ). �109 To test these hypotheses, we examined the effects of GnRH-PAP on reproductive interactions of male and female deer during 2 time periods; breeding season (defined as the period November 1- December 20, 2001) and postbreeding season (defined as the period January 10 - March 27, 2002). On November 1, 2001, female deer in Group A were treated with GnRH-P AP and released with untreated controls (Group B) into 2 ha paddocks. Four days later (November 5), we placed 2 adult male mule deer with these groups and initiate behavioral observations. All females were individually identified with color/numeric-coded neck collars. Animals selected as treatments and controls were unknown to observers. Behavioral measurements were made from a distance of 50-350 m from an elevated tower (10 m) using binoculars and a spotting scope during the day, and a spotlight and night vision scope at night. We recorded selected behaviors using a lap-top computer with a behavioral software program. We used focal animal sampling procedures to sample reproductive behaviors of all experimental animals • over a 24-hour period (Lehner, 1996). Previous studies (Baker et al. 2000) have shown that mule deer are most active in morning (0500-0800), late day (1400-1700) and night (2000-2400). Thus, time-of-day sampling periods was randomly assigned each week using a randomized block design. Each sampling period consisted of at least two hours of continuous observations. Sample size estimation for pregnancy rate and behavior measurements. Sample size calculations were based on results of previous investigations (Baker 2001 ). We used male pre-copulatory behavioral rates to estimate sample size because these rates were much higher than other rates (Table 1). Table 1. Behavior rates for mule deer (Baker 2001 ). Behavior Category Treatment Group Copulatory Control Leuprolide Male Pre-Copulatory Control Leuprolide SQ Female Pre-Copulatory Control Leuprolide SQ General Breeding Control Leuprolide SQ Mean SE (# behaviors/day) 0.20 0.29 0.17 0.24 8.28 22.18 1.77 2.50 0.22 1.35 0.34 0.47 2.00 2.96 0.41 0.58 For control females, we directly bootstrapped a given sample size from the male pre-copulatory rates exhibited towards control does from the Table 1. That is, for a sample size of 6, we randomly selected 6 does (with replacement) for the given observation period, from the 8 control does. For treated does, we followed the same procedure except that we multiplied the response by the effect size. For example, for a +50% effect size, we multiplied the control response by 1.5. We then ran sample sizes for increased behavioral rates because this would be the most problematic to the animal. However, the power should be almost the same for a -50% effect size. We considered the higher behavior rates because increased behavior and corresponding energy output would be the most critical to the animal. This approach captured the day to day variability in behavior rates, because we bootstrapped for each observation period, but it assumes that the variability in behavior is the same for the control and treatment does. Next, we ran the procedure for 52 and 104 observation periods. We assumed 104 observation periods would be acceptable for 2 reasons. First, our proposed collection schedule was: �IIO a. November - 4 weeks x 3 observation periods/day x 5 days/week= 60 obs periods b. December - 4 weeks x 2 observation periods/day x 5 days/week= 40 obs periods Total = I 00 observation periods. Since power was not nearly as sensitive to the number of observation periods as to the number of animals used in the experiment; we decided that ifwe were somewhat below the 104 observations used in sample size simulations, it would not change the power meaningfully. Power results were based on the number of times an effect was detected during 100 simulations. Results from male precopulatory behavioral rates indicated that a sample size of 14 does (7 control and 7 treatment) should provide power of>90% to detect a 50% effect size. Table 2. Sample size calculations for a given effect size using male precopulatory behavior rates. Effec tsize Sample Size* Number of Observation Periods 25% 12 104 16 104 8 104 10 104 12 104 12 52 40% 9 104 50% 7 104 8 104 8 52 30% a. Power 0.05 0.10 0.05 0.10 0.05 0.10 0.05 0.10 0.05 0.10 0.05 0.10 0.05 0.10 0.05 0.10 0.05 0.10 0.05 0.10 57% 74% 81% 91% 68% 70% 72% 83% 90% 95% 68% 81% 91% 97% 94% 96% 97% 100% 74% 88% _ _ 6_· ·- •••• * Sample size for each group, e.g. 12 means 12 treatment and 12 control does. The daily male precopulatory rate for the control group was 8.3 behaviors/day. For a 50% effect size, we could detect a difference between the control mean of 8.3 behaviors/day and a treatment mean of <4.2 and> 12.5 behaviors per day. For a 40% difference in behavior rates, we could detect a difference between the control mean of 8.3 behaviors/day and a treatment mean of <5 and> 11.6 behaviors per day. After estimating the number of animals needed to detect behavioral differences, we calculated power to detect differences in pregnancy rates. If we have 7 control and 7 treatment animals, and 1 treatment animal gets pregnant, and 1 control animal does not get pregnant, we had >95% power to detect this difference. Basically, to detect a difference in pregnancy rates in the case where 1 treatment animal gets pregnant, and 1 control animal does not get pregnant, we could have as low as 5 treatment and 5 control animals and still have >90% power to detect a difference. Thus, the sample size needed to detect �111 differences in behavior rates was sufficient to detect differences in pregnancy rates. Our calculations indicated that the optimum sample size for these measurements to be 14 animals (7 control : 7 treated), however, there is a high probability of losing at least one or two animals per year to non-treatment related mortality. Therefore, in order to insure meaningful measurements over the 2-year investigation, we increased our sample size to 18 animals (9 control: 9 treated). Statistical Analysis We analyzed for differences among hormone levels using least squares analysis of variance for general linear models (SAS Institute 1993). Responses to treatments were analyzed with one-way analysis of variance for a randomized complete block design with repeated measures structure. Treatment effects were tested using the animal-within-treatment variance as the error term. Time was treated as a withinsubject effect using a multivariate approach to repeated measures (Morrison, 1976). A "protected" least significant difference test (Milliken and Johnson 1984) was used to separate means when the overall Ftest indicated significant treatment effects (P < 0.05). We tested specific reproductive behavior hypotheses that mean behavior rate was not different between treatment and control groups for both the breeding and postbreeding seasons using an ANOVA model with a repeated measures structure. Similar to the hormonal analysis, time was be treated as a within subject effect using multivariate approach to repeated measures (Morrison, 1976). To test for treatment effects, we accounted for time-of-day, date effects and their interactions. PROC GENMOD (SAS Institute 1993) was used to estimate and test for differences in mean behavior rate by treatment, time-ofday, and date. Means and standard errors were estimated using least squares, and hypothesis tests were based on type III generalized estimating equations that account for correlation in repeated measurements. RESULTS AND DISCUSSION Pregnancy rates. This dose of GnRH-PAP did not lower pregnancy rates in female mule. Treated and control females tested positive for PSPB on all sampling dates and all delivered healthy fawns in July August, 2002. All fawns were incorporated into the existing captive mule deer herd at the FWRF. Hormonal evaluation. GnRR-PAP did not cause a significant reduction in LH in treated female mule deer. Peak serum LH concentrations for treated animals, after 30 days posttreatment, averaged 8.6 ± 1.97 ng/ml and 9.7 ± 3.0 ng/ml for controls. Based on these responses, GnRH challenge trials were terminated at+ 30 days posttreatment (4 Dec 2001 ). Reproductive behavior. We observed male to male dominance interac~ions immediately following their release into the pastures with treated and untreated females. Within 10 days, one male established dominance. Thereafter, the subdominant male retreated to remote locations within the pastures and rarely interacted with females or the dominant male for the remainder of the experiment. Contrary to our hypothesis, sexual interactions were not different (P > 0.65) during the breeding and post-breeding seasons between GnRH-treated females and untreated controls for any of the breeding behavior categories. We observed almost no sexual interactions between the dominant male and treated or untreated females during the postbreeding season. We conclude :from these measurements that the dose ofGnRH-PAP administered in this experiment was ineffective in suppressing reproduction in female mule deer. Future experiments should investigate the effects of higher levels of GnRH-PAP on reproductive performance in this species. �112 LITERATURE CITED Anderson, A. E., and D. E. Medin. 1966. The breeding season in migratory mule deer. Outdoor Facts, Number 60, Colorado Division of Wildlife. Baker, D. L. 2001. Technical support for deer population management at the Rocky Mountain Arsenal National Wildlife Refuge, Denver, Colorado. Pages 215-232 in Wildlife Research Report, Mammals Research, Federal Aid Projects, Job Progress Report, Project W-153-R-4, SP 1, Jl. Colorado Division of Wildlife, Fort Collins, Colorado, USA. _ _ _. 1996. Regulation of mule deer population growth by fertility control: laboratory, field, and simulation experiments. Pages 81-87 in Wildlife Research Report, Mammals Research, Federal Aid Projects, Job Progress Report, Project W-153-R-4, SP 1, Jl. Colorado Division of Wildlife, Fort Collins, Colorado, USA. Bomford, M. 1990. A role for fertility control in wildlife management? Department of Primary Industries and Energy Bureau of Rural Resources Bulletin No 7 Australian Government Publishing Service Canberra Australia.. Collier, R. J., and D. A. Kaplan. 1984. Immunotoxins. Scientific American 251 :64. Cowan, I. McT. 1956. Life and times of the coast black-tailed deer. Pages 56-78 in The Deer ofNorth America, ed. W. P. Taylor. Wildlife Management Institute, Washington, D. C. Fagerstone K. A., M. Coffey, P. Curtis, R. Dolbeer, G. Killian, L.A. Miller, and L. Wilmot. 2001. Wildlife contraception. Wildlife Society Technical Review. Proceedings of the Wildlife Society 8th Annual Conference, Reno, USA. Garrot, R. A. 1995. Effective management of free-ranging ungulate populations using contraception Wildlife Society Bulletin 23 :445-452. Hone, J. 1992. Rate of increase and fertility control. Journal of Applied Ecology 29:695-698. Hobbs, N. T., D. C. Bowden, and D. L. Baker. 2000. Effects of fertility control on populations of . ungulates: general, stage-structured models. Journal of Wildlife Management 64: 473-491. Knox, W. M., K. V. Miller, and R. L. Marchinton. 1988. Recurrent estrous cycles in white-tailed deer. Journal of Mammalogy 69:384-386. Lehner, P. N. 1996. Handbook of Ethological Methods. Second Edition. Cambridge University Press, . Cambridge, UK. Milliken, G. A., and D. E. Johnson. 1984. Analysis ofMessy Data. Volume I. Designed Experiments. Lifetime Leaming Publications, Belmont,California, USA Morrison, D. F. 1976. Multivariate Statistical Methods. McGraw-Hill Book Co., New York, USA Pastan, I., M. C. Willingham, and D. J. FitzGerald. 1986. Immunotoxins. Cell 47:641-648. Plotka, E. D., U.S. Seal, G. C. Schmoller, P. D. Karns, and K. D. Keenlyne. 1997. Reproductive steroids in the white-tailed deer ( Odocoileus virginianus borealis). l. Seasonal changes in the female. Biology ofReproductiorr 16:340-3_43. ·.' • . Thomas, D. C., and I. McT. Cowan. 1975. The pattern ofreproduction in female Colum.bian black-tailed deer, Odocoileus hemionus columbianus. Journal of Reproduction and Fertility 44:261-272. Nett, T. M., D. L. Baker, and M.A. Wild. 2001. Evaluation ofGnRH-PAP as a chemosterilant in captive mule deer (Odocoileus hemionus hemionus). Proceedings of the 5th International Symposium on Fertility Control in Wildlife, Kruger National ·Park, South Africa, August 19-22. Niswender G.D., L. E. Reichert, Jr., A. R. Midgley, and A. V. Nalbandov. 1969. Radioimmunoassay for bovine and ovine luteinizing hormone. Endocrinology 84: 1166-1173. Willard, S. T., R. G. Sasser, J. T. Jaques, D.R. White, D. A. Neuendorff, and R. D. Randel. 1998. Early pregnancy detection and hormonal characterization of embryonic-fetal mortality in fallow deer (Dama dama). Theriogenology 49:861-869. � https://cpw.cvlcollections.org/files/original/27f8d0d9a75721e44f08512ea663637a.pdf 5daa8aea2df6d3d0287e0e4d503044c1 PDF Text Text 113 Colorado Division of Wildlife Wildlife Research Report July 2001 and July 2002 PROGRESS REPORT State of ------~C~o~lo~r~a=do-'------Work Package No. Division of Wildlife- Mammals Research 3001 Deer Conservation T~k ________~l~0'------ Chronic W~ting Dise~e in Mule Deer Federal Aid Project ___W~-1~8~5~-=R'----- Monitoring & Management Period Covered: July 1, 2000 through June 30, 2001 Author: Michael W. Miller, D.V.M. Personnel: T. R. Davis, L. L. Wolfe, T. H. Baker, K. T. Larsen, E. S. Williams Interim Report - Preliminary Results This work continues, and precise analysis ofdata has yet to be accomplished. Manipulation or interpretation of these data beyond that contained in this report should be labeled as such and is discouraged. ABSTRACT We continued conducting research on various ~pects of chronic w~ting dise~e (CWD) epidemiology and management. Results of original research, ~ well ~ two review articles, were published or accepted for publication during this segment, and citations are included in the body of the report. In addition to published studies, we completed a study of CWD pathogenesis in mule deer. Seven of 10 orally inoculated deer that survived > 12 mo postinoculation (Pl) developed clinical CWD. Five of the seven deer that showed clinical signs either died or were euthanized in end-stage clinical CWD 20-26 mo PI; the other two were euthanized showing mild or marked clinical signs 20 mo PI according to the established sampling schedule. B~ed on observations of the seven deer that developed clinical CWD, earliest signs were first noticed in individuals about 14.5 to 19 mo PI (mean ± SE = 17.3 ± 0. 7 mo PI). Early clinical signs were both subtle and inconsistent. As clinical dise~e progressed, behavioral changes and loss of body condition became more pronounced and more consistent. Ptyalism (drooling), polydypsia (excessive water consumption), and polyuria (excessive urination), widely regarded~ "classic" signs of CWD, occurred relatively late in clinical courses and were not seen in all c~es. Among the five deer that lived long enough to develop terminal CWD, clinical courses ranged from about 3.5 to 9.5 mo (mean± SE = 5.7 ± 1.2 mo); the shortest clinical course (about 3.5 mo) w~ complicated by acute ~piration pneumonia Immunohistochemistry and histopathology results are pending . .\ I I I. ��115 INTRODUCTION We continued conducting research on various aspects of chronic wasting disease (CWD) epidemiology and management. METHODS Epidemiology & Management Two review articles on CWD epidemiology were accepted for publication during this segment in the Journal of Wildlife Management and in the Revue Scientifique et Technique Office international des Epizooties. Results of three earlier studies on CWD epidemiology were published during this segment in the Journal of Wildlife Diseases and the Journal of Wildlife Management. Pathogenesis & Diagnosis Results of two original studies on CWD diagnosis in mule deer were accepted for publication in the Journal of Wildlife Management and in The Veterinary Record. We completed a study of CWD pathogenesis in mule deer. The methods for this project were included in previous progress reports. We also initiated preliminary field work to develop and evaluate reliable methods for collecting tonsil biopsies from live mule deer for use as a diagnostic and management tool. No results are available for reporting in this reporting period. RESULTS AND DISCUSSION Epidemiology & Management Two review articles on CWD epidemiology were published during this segment: Williams, E. S., and M. W. Miller. 2002. Chronic wasting disease in deer and elk in North America. Revue Scientifique et Technique Office international des Epizooties 21:305-316. Williams, E. S., M. W. Miller, T. J. Kreeger, R.H. Kahn, and E.T. Thorne. 2002. Chronic wasting disease of deer and elk: A review with recommendations for management. Journal of Wildlife Management 66:551-563. Results of three studies on CWD epidemiology were published during this segment: Conner, M. M., C. W. McCarty, and M. W. Miller. 2000. Detection of bias in harvest-based estimates of chronic wasting disease prevalence in mule deer. Journal of Wildlife Diseases 36:691-699. Gross, J.E., and M. W. Miller. 2001. Chronic wasting disease in mule deer: disease dynamics and control. Journal of Wildlife Management 65:205-215. Miller, M. W., E. S. Williams, C. W. McCarty, T. R. Spraker, T. J. Kreeger, C. T. Larsen, and E. T. Thorne. 2000. Epizootiology of chronic wasting disease in free-ranging cervids in Colorado and Wyoming. Journal of Wildlife Diseases 36:676-690. Pathogenesis & Diagnosis Nineteen of 20 mule deer orally inoculated with 5 g brain homogenate from CWD-infected mule deer survived ~ 3 months postinoculation (Pl) and were examined as described in the original study plan; �116 one fawn died < I day PI of capture-related complications, and was not evaluated here. Of the 19 . remaining deer, one died from a cervical fracture at 3 mo PI and 12 others were euthanized at 3, 6, 9, 12, 16, or 20 mo PI according to the study schedule; the other 6 were allowed to survive to terminal stages of CWD or study termination. Seven of 10 orally inoculated deer that survived > 12 mo postinoculation (PI) developed clinical CWD; four of these were males and three were females. Of the three that did not show at least early clinical signs, two were euthanized 16 mo PI but the third appeared clinically normal when euthanized 26 mo PL Five of the seven deer that showed clinical signs either died or were euthanized in end-stage clinical CWD 20-26 mo PI; the other two were euthanized showing mild or marked clinical signs 20 mo PI according to the established sampling schedule. Based on observations of the seven deer that developed clinical CWD, earliest signs (dullness in eyes, diminished alertness, misdirected behaviors, piloerection) were first noticed in individuals about 14.5 to 19 mo PI (mean ± SE = 17.3 ± 0. 7 mo PI). Early on, clinical signs were both subtle and inconsistent. As clinical disease progressed, behavioral changes (e.g., blank staring, uncharacteristic or subdued responses to aversive stimuli, lowered head or other unusual postures, ataxia, inefficient foraging activity) and loss of body condition became more pronounced and more consistent. Ptyalism, polydypsia, and polyuria occurred relatively late in clinical courses, and were not seen in all cases. Among the five deer that lived long enough to develop terminal CWD, clinical courses ranged from about 3.5 to 9.5 mo (mean ±SE= 5.7 ± 1.2 mo); the shortest clinical course (about 3.5 mo) was complicated by acute aspiration pneumonia. Immunohistochemistry and histopathology results are pending. Results of two original studies on CWD diagnosis in mule deer were accepted for publication. One of these (Wolfe et al. 2001) represents the first report of a method for detecting CWD infection in live animals. These publications were: Miller, M. W., and E. S. Williams. 2002. Detecting PrPcwn in mule deer by immunohistochemistry of lymphoid tissues. Veterinary Record 151:610-612. Wolfe, L. L., M. M. Conner, T. H. Baker, V. J. Dreitz, K. P. Burnham, E. S. Williams, N. T. Hobbs, and M. W. Miller. 2002. Evaluation of antemortem sampling to estimate chronic wasting disease prevalence in free-ranging mule deer. Journal of Wildlife Management 66:564-573. � https://cpw.cvlcollections.org/files/original/7242157943a6bfd2a6fc01c60a031612.pdf 37f743c448a3d91a449982ea362e42c9 PDF Text Text 117 Colorado Division of Wildlife Wildlife Research Report July 2001 and July 2002 JOB PROGRESS REPORT State of _ _ _ _ _ _ _c-·o-l~o~ra~d_o_ _ _ __ Mammals Research - Terrestrial Section Work Package No. ---~3"""0""-0.e....1_ _ _ __ Task No. _ _ _ _ _ _ _____:A...:.__ _ _ __ Deer Conservation Deer Aerial Survey Populatim1 Estimation Rangely· Deer Data Analysis Unit D~6, GMU 10 Project No. _ _ _ _ _W"---'-1=5....3"""'-....R--=1~4_ _ __ Period Covered: July 1, 2000-June 30, 2001 Author: D. J. Freddy Personnel: V. Graham, W. deVergie, J. Ellenberger, C. Wagner, P. Schnurr, R. Kahn, R. Velarde, G. Miller, T. Wygant, F. Pusateri ofCDOW, Dr. G. White and M. Kneeland of Colorado State University, J. Unsworth, Idaho Department of Fish and Game, consultants V. Howard, Jr. and T. Bickle, Colorado Mule Deer Association, Colorado Bowhunters Association, Dynamic Aviation, and New Air Aviation. ABSTRACT Sportsmen expressed concerns about the credibility of Colorado's survey sampling methodology to estimate numbers of mule deer (Odocoileus hemionus) in specific populations. We therefore conducted an aerial survey in Colorado Deer Analysis Unit D-6 which was an area of concern to sportsmen. We used helicopters from 28 February to 5 March 2001 to count mule deer on randomly selected quadrats 0.25-mi 2 or l .00-mi2 in size distributed within 11 strata encompassing 364 mi 2 of deer winter range composed of sagebrush (Artemisia tridentata) and pinyon-juniper (Pinus edulis-Juniperous osteosperma) habitats. From these counts, we estimated population size using standard stratified random sample estimators and the Idaho mule deer sightability model. Stratified population estimate was 6,782 ± 2,497 (90% CI) deer. Counts corrected for sightability increased the estimate to 11,052 ± 3,503 (90% CI) deer. Both aerial survey estimates buttressed population estimates of 7,000 to 7,300 deer derived from computer models and were substantially greater than sportsmen's estimate of 1,750 deer. Cost of this validation exercise exceeded 50,000 $US. We interpreted this exercise as a forerunner of the public's interest in challenging agency integrity or methods used to estimate status of ungulate populations. We caution agencies to use tested methodology that can withstand dispassionate public scrutiny. Copies of this report containing the original colored versions of the figures are available for review from the Research Center Library, Colorado Division of Wildlife, 317 West Prospect Road, Fort Collins, CO, 80526, USA. All information in this report is preliminary and subject to further evaluation. l~f il~i BD0WD176 □·2 ��119 PROJECT SUMMARY REPORT DEER AERIAL SURVEY POPULATION ESTIMATION RANGELY DEER DATA ANALYSIS UNIT D-6, GAME MANAGEMENT UNIT 10 COLORADO DIVISION OF WILDLIFE PRESENTED APRIL 19, 2001 DOCUMENTPREPAREDTO/NFORM COLORADO DIVISION OF WILDLIFE COLORADO WILDLIFE COMMISSION COLORADO MULE DEER ASSOC/AT/ON COLORADO BOWHUNTERS ASSOC/AT/ON Includes Draft Manuscript Subject to Future Editorial Review© REPORT PREPARED BY DAVID J. FREDDY WILDLIFE RESEARCHER, MAMMALS RESEARCH COLORADO DIVISION OF WILDLIFE, 317 WEST PROSPECT STREET, FORT COLLINS, CO 80526 DEER AERIAL SURVEY POPULATION ESTIMATION RANGELY DEER DATA ANALYSIS UNIT D-6, GAME MANAGEMENT UNIT 10, 28 FEBRUARY- 5 MARCH 2001 REPORT CONTENTS SECTION A ---- EXECUTIVE SUMMARY SECTIONB-- DRAFT TECHNICAL MANUSCRIPT SECTION C ------- 1. STRATIFIED RANDOM SAMPLING CALCULATIONS 2. MODIFICATION OF SAMPLE SIZE DURING PROJECT SECTIOND-- 1. D-6, UNIT 10 DEER WINTER RANGE MAP 2. D-6, UNIT 10 DEER WINTER CONCENTRATION AREA MAP 3. D-6, UNIT 10 DEER WINTER SEVERE Rf'NGE MAP SECTION E --------- 1. SAMPLING FRAME AND SAMPLING STRATA MAP 2. SAMPLING FRAME AND PRIMARY VEGETATION TYPES 3. YAMPA MONUMENT STRATA 1 AND 2 MAP 4. UTAH WHITE RIVER STRATA 3 AND 4 MAP 5. UPPER WHITE RIVER STRATA 5, 6, AND 7 MAP 6. MASSADONA- DINOSAUR STRATA 10, 11, AND 13 MAP 7. TWELVEMILE STRATA 12 MAP (Clarification: No strata numbered 8 & 9; 11 total strata) SECTION F - - - - 1. SURVEY FLIGHT PROTOCOLS 2. SURVEY DATA FORM 3. SURVEY OBSERVER HELP SHEET SECTION G -------- SURVEY FLIGHT QUADRAT SAMPLE UNIT MAP INDEX SECTION H ---- 1. STRATIFIED RANDOM SAMPLE POPULATION ESTIMATE DATA AND CALCULATIONS 2. LETTER FROM IDAHO DEPARTMENT OF FISH & GAME SHOWING POPULATION ESTIMATE USING IDAHO SIGHTABILITY CORRECTIONS 3. COMPLETE DATA LISTING FOR AERIAL SURVEY 4. MAP SHOWING WHERE DEER WERE COUNTED 5. MAP SHOWING WHERE ELK WERE COUNTED 6. SUMMARY OF SPORTSMEN AND CDOW DEER POPULATION ESTIMATES FOR WESTERN COLORADO, DECEMBER 2000 �120 SECTION A - EXECUTIVE SUMMARY DEER AERIAL SURVEY POPULATION ESTIMATION RANGELY DEER DATA ANALYSIS UNIT D-6, GAME MANAGEMENT UNIT 10 A. Sportsmen in Colorado alleged that estimates for numbers of mule deer in western Colorado were substantially over-estimated by the Colorado Division of Wildlife (CDOW). Sportsmen believed there were only 128,000 deer in Colorado in areas west of the Continental Divide where CDOW estimated 409,000 deer. This level of discrepancy also existed for specific deer populations such as in the Rangely Deer Analysis Unit D-6 where sportsmen estimated 1,750 deer compared to CDOW estimates of 7,000 deer. 8. A series of meetings between CDOW and sportsmen from September 2000 through February 2001 did not resolve fundamental issues of sportsmen's mistrust of estimated deer population status. C. On February 16, 2001 CDOW Director Russell George authorized the Terrestrial Section to implement aerial surveys to estimate numbers of deer in Rangely Unit D-6 in accordance with survey methodologies agreed to by all interested parties, including participation in surveys by individuals independently representing sportsmen's concerns. Financial costs for the survey were paid primarily with Wildlife Commission Discretionary Funds with additional contributions from the Colorado Mule Deer Association and Colorado Bowhunters Association. D. CDOW conducted an aerial survey to estimate numbers of deer in D-6 using Colorado quad rat survey techniques that incorporated adjustments in estimated population size based on Idaho mule deer sightability models as requested by sportsmen. The survey was conducted 28 February to 5 March, 2001. E. Estimated numbers of deer in D-6 were 6,782 ± 2,497 based on Colorado quadrat system and 11,052 ± 3,503 when adjusted for the Idaho mule deer sightability model. Population estimates based on CDOW computer models were 7,000 to 7,312 deer. All estimates were substantially higher than the 1,750 deer estimated by sportsmen. F. Financial and personnel costs to design, implement, and analyze survey results likely exceed $50,000. Final costs estimates are not yet available. G. This validation exercise challenged the credibility of CDOW personnel and methodologies and the credibility of sportsmen groups. All parties participated within a certain level of risk. Not to be overlooked was a near fatal helicopter incident that threatened lives of personnel involved in an aerial survey conducted to alleviate mistrust among interested parties. H. We interpret this validation exercise as a potential forerunner of the public's interest in either challenging or understanding methods used to estimate status of wildlife populations. We can only caution that wildlife agencies should gather information using methodology that can withstand public scrutiny. We would hope that this exercise would restore a certain level of public confidence in the CDOW's efforts to manage wildlife in Colorado. �121 SECTION B - DRAFT TECHNICAL MANUSCRIPT April 18, 2001 Draft David J. Freddy Colorado Division of Wildlife 317 West Prospect Road Fort Collins, CO 80526 970-472-4346, FAX 970-472-4457 RH: Deer Population Estimates ESTIMATING MULE DEER POPULATION SIZE USING COLORADO QUAD RAT SYSTEM CORRECTED FOR IDAHO MULE DEER SIGHTABILITY: A SPORTSMEN'S ISSUE. DAVID J. FREDDY1. Colorado Division of Wildlife, Wildlife Research Center, 317 West Prospect Road, Fort Collins, CO 80526, USA GARY C. WHITE, Department of Fishery and Wildlife Biology, Colorado State University, Fort Collins, CO 80523, USA MARY C. KNEELAND, Colorado Division of Wildlife, Wildlife Research Center, 317 West Prospect Road, Fort Collins, CO 80526, USA VAN K. GRAHAM, Colorado Division of Wildlife, 711 Independent Avenue, Grand Junction, CO 81505, USA WILLIAM J. deVERGIE, Colorado Division of Wildlife, P.O. Box 1181, Meeker, CO 81641, USA JOHN H. ELLENBERGER, Colorado Division of Wildlife, 711 Independent Avenue, Grand Junction, CO 81505, USA JAMES W. UNSWORTH, Idaho Department of Fish and Game, 3101 South Powerline Road , Nampa, ID 83686, USA CHARLES H. WAGNER, Colorado Division of Wildlife, 346 Count Road 362, Hot Sulphur Springs, CO 80451 PAMELA M. SCHNURR, Colorado Division of Wildlife, 711 Independent Avenue, Grand Junction, CO 81505, USA V. W. HOWARD, JR., 1025 Hickory Drive, Las Cruces, New Mexico 88005, USA TOMMY S. BICKLE, P.O. Box 750, Hatch, New Mexico 87937, USA 1 Corresponding author. Abstract: Sportsmen expressed concerns about the credibility of Colorado's survey sampling methodology to estimate numbers of mule deer (Odocoileus hemionus) in specific populations. We therefore conducted an aerial survey in Colorado Deer Analysis Unit D-6 which was an area of concern to sportsmen. We used helicopters from 28 February to 5 March 2001 to count mule deer on randomly selected quadrats 0.25-mi 2 or 1.00-mi2 in size distributed within 11 strata encompassing 364 mi2 of deer winter range composed of sagebrush (Artemisia tridentata) and pinyon-juniper (Pinus edulis-Juniperous osteosperma) habitats. From these counts, we estimated population size using standard stratified random sample estimators and the Idaho mule deer sightability model. Stratified population estimate was 6,782 .±. 2,497 (90% Cl) deer. Counts corrected for sightability increased the estimate to 11,052 .±. 3,503 (90% Cl) deer. Both aerial survey estimates buttressed population estimates of 7,000 to 7,300 deer derived from computer models and were substantially greater than sportsmen's estimate of 1,750 deer. Cost of this validation exercise exceeded 50,000 $US. We interpreted this exercise as a forerunner of the public's interest in challenging agency integrity or methods used to estimate status of ungulate populations. We caution agencies to use tested methodology that can withstand dispassionate public scrutiny. Key Words: bias, Colorado, helicopter surveys, Idaho, mule deer, Odocoi/eus hemionus, population estimates, sightability �122 Sportsmen in Colorado alleged that estimates for numbers of mule deer (Odocoileus hemionus) in western Colorado were substantially over-estimated by the Colorado Division of Wildlife (CDOW). For example during post-hunting season 2000, sportsmen believed there were only 128,000 deer in Colorado in areas west of the Continental Divide where CDOW estimated 409,000 deer. This level of discrepancy also existed for specific deer populations such as in the Rangely Deer Analysis Unit D-6, where sportsmen estimated 1,750 deer compared to CDOW estimates of 7,000 deer after hunting season 2000 (Pers. comm. Colorado Mule Deer Association). These 4-fold differences in estimated numbers of deer explained why perceptions about the status of mule deer in Colorado varied between some sportsmen and CDOW. Sportsmen focused their concerns on the credibility of Colorado's quadrat survey sampling methodology to estimate numbers of deer in specific populations. This methodology, based on stratified random sampling theory (Thompson et al. 1998), was initially developed for helicopter counts of mule deer on 1-mi2 sample quadrat units used to estimate total numbers of deer in a population inhabiting extensive sagebrush habitats during winter (Gill 1969). This system was later expanded to estimate size of selected deer populations inhabiting pinyon-juniper habitats in western Colorado where quadrat sample unit size was reduced to 0.25-mi2 to compensate for the detrimental effects that dense pinyonjuniper canopy cover had on detecting and counting mule deer (Kufeld et al. 1980, Bartmann 1983, Bartmann et al. 1986). Aerial counting of deer using random quadrats provided estimates of deer numbers sufficiently suitable for herd management decisions but implementation costs prevented such systems from being employed in most deer management units in western Colorado (Gill et al. 1983). Alternative approaches to estimating trends in numbers of deer in every population included: intensively estimating numbers of deer, age and sex ratios, and survival rates in a few populations whose trends in population parameters could represent many deer populations inhabiting ecologically similar areas (White and Bartmann 1998, Bartmann 2000, Bowden et al. 2000); and, using computer modeling that incorporated measured parameters from appropriately similar ecological core areas in conjunction with less intense measurements of age and sex ratios and hunter harvests that could be obtained yearly for nearly every deer population (CDOW 1991, White and Bartmann 1998, Bartholow 2000,). The discrepancy in perceived numbers of deer in western Colorado more accurately reflected a concern about modeled as opposed to aerial survey estimates of deer population size because only about 10% of the deer populations were monitored using aerial quadrat sampling protocols. Nevertheless, sportsmen focused their concerns on aerial survey sampling fearing that such techniques inflated estimates of deer numbers and therefore, misrepresented the declining plight of mule deer in western Colorado. Furthermore, sportsmen desired to assess the Idaho Mule Deer Sightability survey system (Ackerman 1988, Unsworth et al. 1994) as an alternative to Colorado's approach to estimating numbers of deer on the premise that the Idaho system would provide more acceptable estimates of deer numbers. This project was prompted by sportsmen's concerns about the legitimacy of deer population estimates based upon aerial surveys employing random sampling and counts of deer on sample quad rats. We conducted an aerial quadrat survey in a deer population unit of concern to sportsmen using Colorado quadrat survey techniques incorporating adjustments in estimated population size based on Idaho mule deer sightability models (Unsworth et al. 1994). Our survey and results were monitored by participating individuals independently representing sportsmen's concerns. We then compared estimates based on aerial surveys to ongoing population models used to guide management of deer. STUDY AREA We estimated numbers of mule deer inhabiting winter range in the Rangely Deer Analysis Unit D-6 consisting of Game Management Unit 10 in northwestern Colorado near the town of Rangely. D-6 includes 837 mi2 with large expanses of public lands administered by the U. S. Department of Interior Bureau of Land Management and National Park Service. Deer typically move onto winter range in November and begin returning to summer ranges in April. The project area is semi-desert with yearly precipitation ranging from 8 to 20 inches and winters having moderate temperatures and snow depths. Deer winter range occurs between 5,000 and 7,200 feet elevation and is a mixture of pinyon-juniper, sagebrush, and greasewood (Sarcobatus vermicu/atus) - desert shrub habitats and is shared with domestic sheep, cattle, and elk (Cervus elaphus). During �123 winters with low snow depths, deer distribution could encompass 612 mi2 but under severe snow depth conditions, deer distribution may collapse to 106 mi2 (CDOW 2001 ). Deer have been managed in D-6 under a limited permit hunting system since 1991 resulting in average yearly harvests of 118 bucks (range 70-252) and 60 antlerless does and fawns (range 2-132). Helicopter surveys of population ratios post-hunting season in December have shown 9-26 bucks:100 does and 29-64 fawns: 100 does. Previous efforts to estimate population size of deer in D-6 involved assessing the practicality of using helicopter line transects (White et al. 1989) on a trial basis in 1990 and 1991 with resulting estimates of 21,630 .±. 12,321 and 13,596 .±. 5,427 (90% Cl} deer, respectively (CDOW 1991 ). METHODS Sampling Protocols We estimated deer population size using stratified random sampling and counts of deer on randomly selected quad rat units (quad rats) (Thompson et al. 1998). Counts of animals on random units assume that units are completely searched by observers and all'animals present are detected and counted. These assumptions, therefore, assume 100% sightability of target anim·aIs and resulting estimates would not incorporate correction factors for animals not counted. We delineated our sampling area (frame) based upon the distribution of deer observed during systematic strip-surveys of potential winter range in the project area conducted with a Hiller 12-E Salay helicopter on 6 February 2001. Using ESRI ArcView©, we delineated a frame of 364 mi2 that encompassed the distribution of deer observed during the survey flight. Each cadastral square-mile within the frame was subjectively rated by flight observers as to high, medium, or low expected deer densities. Guidelines for relative deer densities were: >20 for high, 5-20 for medium, and < 4 deer/mi2 for low. We then defined 11 strata based upon expected deer densities for the purpose of distributing quadrats through out the frame. Low, medium, and high density strata encompassed 113 mi2 (31 %}, 157 mi2 (43%}, and 94 mi2 (26%}, respectively, within the frame (Table 1). Proportions of vegetation types within each strata were estimated using ArcView© and Colorado GAP® vegetation coverage (Schrupp et al. 2000). We used 1- mi2 quadrats in strata where open sagebrush-type habitats comprised > 50% of a stratum (Gill 1969) and 0.25-mi2 quadrats where pinyonjuniper habitats comprised > 50% of a stratum recognizing that tree canopy would hinder detection of deer (Bartmann 1983) (Table 1). We allocated quadrats among strata using optimum allocation (Thompson et al. 1998, pages 341342) with estimated variances based upon variance to mean ratios derived from quadrat sample units previously flown in Colorado since 1968 (Expected Standard Deviation = 3.6379 + 1.0891 " [mean deer density], n = 1, 192 qua drats). We calculated number of quad rats needed to achieve precision of.±. 20% of the mean population estimate with a = 0.10 for potential population sizes ranging from 2,000 to 8,000 deer. We selected a sample size of 161 quadrats distributed among 11 strata for an expected population of 6,000 deer. We assumed the cost of flying 1-mi2 and 0.25-mi2 quadrats was the same based on a proportional per area basis (Table 1). We established a grid of point coordinates (UTM [x, y]; NAO 27, all standardized to Zone 13) every 0.25 mi within the frame using ArcView©. We then used the random number option in MS Excel97© to assign a random number to each grid point. Grid point random numbers were then ordered from low to high to initiate the process of randomly selecting locations for quadrats within strata, with quadrat location selection beginning with the lowest ordered grid point and continuing until all quadrats were assigned within each strata. We restricted locations of quadrats by defining a minimum distance between randomly selected grid points of 0.50 mi in strata using 0.25-mi2 quadrats and 1 mi in strata using 1-mi2 quadrats to reduce quadrats having common boundaries and to reduce clustering of quadrats within strata. Quadrats were irregular in shape with boundaries following terrain ridges and gullies or cultural features such as roads or trails that could be discerned on USGS 1:24,000 topographic maps and recognized by observers while flying in a helicopter (Freddy 1994, Unsworth et al. 1994). Quadrat polygons were digitized on digital topographic maps using ArcView© with quadrat area and perimeters calculated by ArcView©. Range of actual areas for 0.25-mi2 units was generally 0.22-0.28 mi2 and for 1mi2 units, 0.9-1.1 mi 2 , shaped with the intent to minimize perimeter to area ratios (Thompson et al. 1998). Randomly selected grid points had to be included within the defined quad rat and preferably centered �124 within the quadrat. Flight path starting latitude-longitude coordinates, back-corrected for UTM Zone 12 or Zone 13 as necessary, were defined for each quadrat and labeled on flight navigation digital topographic maps printed in color using ArcView© layouts and MS PowerPoint97©. Flight Protocols We used Hiller 12-E Salay (Hiller) and Bell Jet Ranger Ill (Ranger) helicopters to count deer on quadrats. While searching for deer, helicopters were flown at 35-50mph at 50-100 feet AGL. Observer, navigator, and pilot comprised flight crews, with observer and navigator having primary responsibilities to detect and count deer with the observer tape-recording all pertinent data. In the Hiller, the observer was seated in the starboard outside seat with the navigator seated in the middle. In the Ranger, the observer was seated in the port outside seat with the navigator in the port rear seat. Crews first flew boundaries of quadrats and then systematically searched the interior of quadrats using strips or strip-contours depending on steepness of terrain following standard procedures (Gill 1969, Kufeld et al. 1980, Freddy 1998, Unsworth et al. 1994). To optimize the visual scanning position of the observer when flying quadrat boundaries, the Hiller crew flew boundaries clock-wise and the Ranger crew flew boundaries counter-clockwise. Navigators and pilots determined proper starting locations of quadrats using previously calculated latitude-longitude coordinates entered into on-board Garmin Pilot Ill© global positioning units (GPS). Navigators then directed pilots along quadrat boundaries and suitable search paths within the quadrat using topographic maps and real-time flight traces recorded on GPS units. Navigators, observers, and pilots constantly adjusted flight speed, altitude and angle of attack to optimize viewing for the observer. Objectives were to fly quadrats to obtain 100% search coverage. Observers and navigators collectively detected and counted groups of deer on quadrats with the highest count by either person recorded by the observer. Observers and navigators collectively made decisions on whether to count deer detected near quadrat boundaries: groups moving onto quadrats when detected were considered outside quadrats; groups moving off quadats when detected were considered on quadrats; one-half of the deer in groups detected on boundaries were considered on quadrats. Observers and navigators also collectively kept mental track of group locations, movements and presence of unique antlered deer in groups to reduce chances of counting groups more than once. Flights were conducted when weather conditions were favorable. Flights were conducted only when wind speeds were low enough in the judgement of pilots to fly safely at desired slow airspeeds and low AGL. Lighting conditions varied from overcast to hazy or bright sunshine while snow cover background varied from Oto 100 percent. Flights continued through short episodes of snow flurries provided safety was not compromised. For each quadrat, observers recorded flight conditions and total flight time. Idaho Sightability Protocols Sightability models correct for undercounting, or negative bias, that is generally associated with counts of ungulates (Caughley 1974, Bartmann et al. 1986, Samuel et al. 1987, Steinhorst and Samuel 1989, Unsworth et al. 1990, Otten et al. 1993, Pojar et al. 1995, White et al. 1989, Anderson and Lindzey 1996, Anderson et al. 1998, Cogan and Diefenbach 1998, Freddy 1998). To correct for potential negative bias in deer detected and counted on quadrats, we obtained values for sighting variables on each group of deer counted following guidelines for the Idaho mule deer sightability model (Unsworth et al. 1994). Sighting variables were total group size, behavior, vegetation type, and percent snow cover. Behavior of the most active deer when a group was first detected was recorded as bedded, standing, or moving. Although deer could have been detected in several vegetation types, we reduced types to broad categories to simplify the process of classifying vegetation: agricultural fields/open meadows; sagebrush, representing all low brush types; and pinyon-juniper, representing all pinyon or juniper dominated areas. Deer were not detected in tall conifer, aspen, or tall mountain brush habitats. Percent snow cover on the ground where each group was detected was classified as a categorical value of low (21-79%) or high~ 80%). The Idaho mule deer model most appropriate to correct undercounting deer was the spring sightability model which contained the following sighting variables (Unsworth et al. 1994: µ = -0.254 +activity+ vegetation class+ snow cover+ 0.047 * (group size) �125 Coefficients for each variable were developed in Idaho in similar but different vegetation and terrain types than might occur in Colorado. Knowing that Idaho coefficients may only approximate coefficients suitable for use in Colorado, we used the following Idaho coefficients for sighting variables (Unsworth et al. 1994): Activity: Bedded = 0.000, Standing = 1.56, Moving = 4.43 Vegetation: Agriculture/meadow= 0.00, Sagebrush = -0.88, Pinyan-Juniper (Idaho Juniper/Mountain mahogany) = -2.383 Snow Cover: Low 21-79% = -1.37, High?_ 80% = -0.60 Estimates of population size based on counts of deer on quadrats were corrected for sightability of each group using Idaho Aerial Survey© program for Windows© beta-version (Unsworth et al. 1994). Program Aerial Survey was limited to accepting only 10 defined strata from which to calculate population size. Our survey design incorporated 11 strata so we therefore, combined strata 1 and strata 2 (Table 1) into 1 strata to accommodate the program. We compared quadrat and quadrat sightability corrected population estimates using a standard z -test (Thompson et al. 1998). Both Idaho and Colorado systems were predicated on using stratified random sampling and thus, these systems complimented each other in conceptual design and application (Gill 1969, Unsworth et al. 1994). Modeling Protocols Estimating trends in deer population size over several years in D-6 was an ongoing CDOW management evaluation process based upon computer modeling using POP-II software (Bartholow 2000). Computer models were constructed independently of data obtained during our aerial survey and by personnel who did not participate in the aerial survey. This model used yearly hunter harvest estimates (Steinert et al. 1994), deer survival rates (White et al. 1987), estimated post-season December doe:fawn and buck:doe ratios collected during yearly helicopter surveys (CDOW 1991 ), and winter severity values to estimate trends in population size. Such models provided an assessment of deer status independent of aerial quadrat surveys, and conversely, aerial quadrat surveys provided point estimates of population size to evaluate models. RESULTS We estimated deer density from 28 February to 5 March 2001 using about 35 hours of helicopter flight time to complete the survey (Table 2). Mechanical malfunctions with the Hiller resulted in using the Ranger more extensively than anticipated, altered availability of survey navigators and observers, and in conjunction with impending unfavorable weather, caused us to reduce sampling intensity in 3 strata in order to complete the survey (Table 1). Adjustments in flight crew members and survey sampling were completed with the approval of independent evaluators. Survey Population Estimates Estimated deer population size was 6,782 with 90% CL of 4,285 to 9,279 for Colorado quadrats assuming 100% sightability of deer (Table 3). Reduced sampling in strata 10, 11, and 13 (Table 1) likely contributed to increasing variability resulting in wide Cl of.:!:. 36% of the mean estimate. Additionally, deer also became more concentrated in their distribution after the ·sampling frame flight of 6 February due to increasing snow depths at upper elevat_ion limits of winter range. This shift in deer distribution contributed markedly to not detecting deer on 66% of the quadrats. Using a 100% Idaho sightability model for deer, a point estimate of population size was 6,481 (Table 5). Colorado and ldhao 100% sightability population estimates would have been equivalent except the Idaho estimate was based only on 1O strata (strata 1 and 2 were combined, Tables 1, 4) instead of 11 Colorado strata due to limitations of program Aerial Survey©. Estimated deer population size was 11,052 with 90% CL of 7,549 to14,555 for Colorado quadrats corrected for the Idaho sightability mule deer model (Table 4). Confidence intervals represented.:!:. 32% of the mean estimate. Idaho sightability increased the standard Colorado quadrat estimate by 1.63x resulting in population estimates tending to be statistically different (z = 1.63, P = 0.103). Within individual strata, sightability increased estimates by 1.37 to 6.26x (Table 4 ). The highest correction 6.26x occurred in strata 11-MDL and should be viewed cautiously and may reflect the sensitivity of sightability correction factors to low counts of deer on few sample units (Table 3). �126 The considerable increase in estimated population size due to sightability corrections reflected that 62% of the deer groups contained ~ 5 deer, 34% of the groups were in pinyon-juniper vegetation and 66% were in sagebrush-type vegetation, and 82% were detected in areas having low and broken snow cover on the ground (Table 2). In essence, many groups of deer were associated with a factor that· decreased the sightability, or probability of detecting a group. Model Population Estimates Computer modeled point estimates of population size for post-season deer populations in 2000 ranged from 7,000 to 7,312 deer. Modeled estimates were similar in magnitude to aerial survey estimates and were within or nearly within the confidence intervals of all aerial survey estimates of population size. Modeled and aerial survey estimates were substantially larger than the population estimate promoted by sportsmen (Table 5). Flight Survey Variables Search times on quadrats were acceptable and comparable to previous surveys in Colorado. Flight crews spent 20.2 :t. 1.1 (SE, n = 38) and 6.4 :t. 0.2 (SE, n = 105) minutes on 1-mi2 and 0.25-mi2 quadrats, respectively. Search times were relatively proportional to total area searched. Wind and lighting conditions were conducive to effectively searching quadrats. Low percent snow cover or broken snow ground cover on many quadrats reduced the probability of detecting deer and made observers more dependent on deer movement to detect groups (Table 2). Compared to the Hiller, flight crews in the Ranger collectively had reduced visibility primarily because the navigator seated in the rear seat had a limited scanning view and could not as effectively help the primary observer detect or count deer. We would expect counts from the Ranger to be more negatively biased than from the Hiller (pers. comm. J. W. Unsworth). Conversely, the 200-shaft horsepower advantage of the Ranger allowed effective slow and low flying in steep and variable terrain. DISCUSSION We conducted an aerial survey in response to demands by sportsmen who strongly believed that methods used to estimate numbers of mule deer over-estimated deer numbers in Colorado. Resulting survey estimates of deer numbers, whether based on the Colorado quadrat system (Gill 1969, Kufeld 1980, Bartmann 1983, Bartmann et al. 1986) or quadrats adjusted for the Idaho mule deer sightablilty model (Unsworth et al. 1994), strongly indicated that sportsmen's estimates of deer numbers were substantially below likely true population size. Furthermore, aerial survey estimates supported population estimates derived in computer population models, and as such, supported the concept that models can provide reasonable estimates of population size to adequately guide decisions for managing mule deer. Colorado quadrats, as expected, provided lower estimates of deer numbers (6,782) than quadrats corrected for sightabiltiy factors (11,052). The Idaho mule deer model (Unsworth et al. 1994) increased estimates by 1.63x compared to the correction factor of 1.51 x developed for deer in pinyon-juniper habitats in Colorado (Bartmann et al. 1986). We are confident that Colorado quadrats, without sightability corrections, will provide conservative estimates of deer numbers· when- proper and adequate sampling procedures and flight protocols are followed. Although applying a calibrated correction factor (Bartmann et al. 1986) would improve accuracy of population estimates, we question whether higher estimates would be more palatable to some sportsmen's groups. We fully recognize the limitations of our population estimates generated from this validation exercise. Our estimates were of low precision for which there a 2 primary reasons. Our efforts to estimate a sampling frame were based on only 1 aerial survey conducted quickly in response to time constraints invoked by pressure to obtain a population estimate. Normally, quadrat sampling frames are determined with several years of deer distribution data obtained when winter sriow conditions would optimize counting deer. In Colorado, quadrat surveys would normally be flown in January when deer distribution and snow cover tend to be stable. In our specific case, major shifts in distribution of deer occurred after the distribution flight and prior to conducting the survey reducing the effectiveness of our sampling allocations. We then reduced sampling intensity in 3 strata to complete the survey with impending unfavorable weather conditions. �127 Observations also suggested that some deer moved off or outside of the frame which would inherently lower estimates of population size. Appropriateness of applying sightability correction factors developed in Idaho for Colorado deer in different habitats can be argued and therefore, the legitimacy of the resulting higher estimates may elicit even less confidence from concerned sportsmen. We foresee worthwhile potential research efforts emanating from this survey effort. Colorado and Idaho both use stratified random sampling procedures in their respective survey systems. However, Idaho often uses sampling units or quadrats having search areas > 3-mi2 while Colorado uses quadrats .'.': 1-mi2 . Cooperative experiments designed to compare effects of sample unit size on population estimates and precision, especially if simultaneously compared against robust mark-resight estimators (Bartmann et al. 1987, Neal et al. 1993, Bowden and Kufeld 1995), may provide valuable insight into designing more efficient aerial survey systems. We believe lessons from this exercise apply more appropriately to human dimension rather than biological issues. Sportsmen demanded a validation process of aerial survey protocols based on their perceptions of deer numbers and not on technical demerits of the survey system in use or reasonably obtained estimates of deer population size. Such demands were not tempered by discussions between CDOW and sportsmen over several months that attempted to resolve mistrust by explaining population estimation procedures, limitations, and likely biases. The result was CDOW spending approximately $50,000 in operating and personnel expenses to estimate numbers of deer in a management unit having low priority for spending limited deer inventory resources. We suspect that our survey exercise minimally mediated concerns of some sportsmen. MANAGEMENT IMPLICATIONS Sportsmen challenged estimates of mule deer populations provided by CDOW and demanded a validation exercise to compare sportsmen's estimates of deer numbers in a specific population with estimates on record with CDOW. Subsequent aerial surveys conducted with sportsmen approval and independent oversight provided deer population estimates that substantiated previous CDOW estimates and that were at least 4x greater than the estimate provided by sportsmen. We interpret this validation exercise as a forerunner of the public's interest in either challenging or understanding methods used to estimate status of ungulate populations. We can only caution that if estimates of population status are part of a routine management process, that estimates should be based on tested methodology that can withstand public scrutiny. ACKNOWLEDGMENTS This project was funded by Colorado Division of Wildlife Federal Aid in Wildlife Restoration Project W153-R, Colorado Wildlife Commission game cash funds, and Colorado Mule Deer Association. We thank R. Kahn, R. Velarde, G. Miller, T. Wygant, and F. Pusateri and their staffs of Colorado Division of Wildlife for project support. We also thank Idaho Department of Fish and Game for allowing the cooperative efforts of J. W. Unsworth. We thank New Air Aviation and Dynamic Aviation for providing survey helicopters. This paper dedicated in memory of M. W. Gratson who contributed significantly to developing helicopter sightability correction protocols for aerial surveys in Idaho. �----- --- 128 Table 1. Characteristics of sampling strata for estimating mule deer population size in Rangely Deer Analysis Unit 0-6, Colorado, Februa~-March 2001. Strata 1 Strata Name No. Yampa Monument 1 Yampa Monument 2 Utah White River 3 Utah White River 4 Upper White River 5 Upper White River 6 Upper White River 7 Massadona Dinosaur 10 Massadona Dinosaur 11 Massadona Dinosaur 13 Twelvemile 12 Totals 11 Density Rank Low Medium Low Medium High Medium Low Medium Low High Medium Strata Area 2 PJ Area 3 Open Mi 2 (%} (%} 28.74 30.40 31.64 22.75 54.16 30.96 31.96 44.09 20.07 40.46 28.93 364.15 66 59 10 35 78 40 20 69 62 54 47 34 41 90 65 22 60 80 31 38 46 53 Sample Unit Size Mi 2 0.25 0.25 1.00 1.00 0.25 1.00 1.00 0.25 0.25 0.25 1.00 Total Quadrat Units 115 122 32 23 217 31 32 176 80 162 29 1018 Sample 4 Quadrats 9 15 5 8 41 10 5 22 6 31 10 161 Sampled 5 Quadrats 9 15 5 8 41 10 5 15 1 21 10 143 Strata numbered 8 and 9 did not exist; there were 11 total strata. Percent of strata area having pinyon-juniper canopy vegetation types. 'Percent of strata area in sagebrush or low brush vegetation types. 'Number of sample quadrat units assigned to each strata based on optimum allocation formulas. 5 Represents number of sample quad rat units actually flown. Quad rats flown in strata 10, 11, and 13 were reduced by random selection to allow completion of aerial survey considering impending weather conditions. 1 2 Table 2. Summary of aerial survey characteristics for Rangely Deer Analysis Unit 0-6, Colorado, February-March 2001. Survey Characteristic Data Summary Aerial Survey Flight Dates Total Sample Quadrats Flown Search Minutes Per Quadrat Observers for Counting Deer Navigators for Counting Deer Flight Wind Speed on Quadrats Flight Lighting on Quadrats Snow Cover on Quad rats Time Period Quadrats Flown 28 February - 5 March 2001; about 35 hours of helicopter flight time. 143; 38 sized 1-mi 2 ; 105 sized 0.25-mi2; 129 flown by Ranger (90%), 14 flown by Hiller (10%) 1-mi2 quadrats = 20.2 .:!: 1.1 (SE); 0.25-mr quadrats = 6.4 .:!: 0.2 (SE) deVergie = 112 quadrats (78%), Graham= 14 (10%), Ellenberger= 17 (12%) Freddy= 82 quadrats (57%), Bickle= 30 (21%), Graham= 17 (12%), Howard= 14 (10%) Low= 136 (95%), Moderate= 7 (5%), High= O (0%) Bright sunshine= 92 (64%), Dull sunshine= 7 (5%), Hazy sunshine= 44 (31%) Fresh snow= 4 (3%}, Old snow= 139 (97%) 7 AM - 12 PM= 65 (45%), 12PM - 5PM = 78 (55%) Total Deer Counted on Quadrats Total Deer Groups Detected Deer Group Size by Quadrat Size Frequency of Group Sizes 1,180 seen on 48 of 143 sample quad rats 179; Average group size= 6.6 .:!: 0.6 (SE), range= 1 - 58 On 1-mi2 quadrats = 6.2 .:!: 0.6 (SE) (n = 94); On 0.25-mi 2 quadrats = 7.1 .:!: 1.0 (SE) (n = 85) (1, n=27, 15%), (2, n=25, 14%}, (3-5, n=59, 33%}, (6-9, n=36, 20%}, (10-19, n=22, 12%), (20-58, n=10, 6%) In sagebrush= 7.2 .:!: 0.7 (SE) (n = 118), In pinyon-juniper = 5.4 =. 0.9 (SE) (n = 61) In Ranger= 7.0 .:!: 0.8 (SE)(n = 127), In Hiller= 5.6 .:!: 0.6 (SE) (n = 52) Group Size by Vegetation Class Group Size by Helicopter Type Deer Group Vegetation Type Deer Group Snow Cover 123 (69%) Moving; 55 (31%) Standing; 1 (<1%) Bedded 118 (66%) sagebrush-type; 61 (34%) pinyon-juniper; (0%) agriculture/meadows 146 (82%) low snow cover; 33 (18%) high snow cover Total Elk Counted 1,297 approximately; seen on 32 of 143 sample quadrats Deer Group Behavior at Detection ---- �129 Table 3. Summary for stratified random sample of mule deer counted on sample unit quadrats in Rangely Deer Analysis Unit D-6, Colorado, February-March 2001. Strata Number With Abbreviated Name and Density Ranking 2 Summary Statistics 3 4 YML YMM UWLUWM Quadrat Sampled Units (uh) 9 Deer Counted Per Stratum(Nh) 114 Mean Deer Per Quadrat (Nhtoa,) 12.67 Quad rat Unit Variance (S 2Nh) 1116 Estimated Deer Per Stratum (N\) 1456 Stratum Variance (Var")(N\) 1510126 Stratum Quadrat Size (Mi 2) 0.25 Total Stratum Quadrat Units (Uh) 115 Stratum Area ( Mi2) 29 Quadrats With 0 Deer Counted 7 Percent Quadrats With 0 Deer 78 5 6 7 10 11 13 12 WRH WRM WRL MDM MDL MDH TMM 15 5 8 41 10 5 15 4 21 10 11 31 66 322 178 112 10 9 133 194 0.73 6.20 17.80 22.40 0.67 2.25 6.33 19.40 54 8.25 181 7.85 3 214 957 760 7 20 274 1020 89 196 188 1702 551 716 118 181 1025 561 2156 9137 7607 198905 62121 131021 12645 31002 297438 55874 0.25 1.00 1.00 0.25 1.00 1.00 0.25 0.25 0.25 1.00 122 32 23 217 31 32 176 80 162 29 30 32 23 54 31 32 44 20 40 29 11 2 5 27 4 1 14 3 15 6 73 40 63 66 40 20 93 75 71 60 Total Deer Counted All Strata (sum Nh) Total Estimated Deer All Strata (sum N\) Total Variance All Strata (sum Var"[N\]) Coefficient of Variation CV"(N") % 90% Confidence Interval for N" (Lower][Upper] 95% Confidence Interval for N" [Lower][Upper] 1,180 6,782 2,318,031 22.45 4,285 9,279 ±. 36% of Population Estimate 3,798 9,766 ±. 44% of Population Estimate Table 4. Estimates of mule deer numbers in individual strata compared between Colorado quadrat and quadrats corrected for Idaho mule deer sightability model and Idaho sightability estimate summary statistics for Rangely deer Analysis Unit D-6, Colorado, February-March 2001. Numbers of Mule Deer Estimated in Each Strata Numbered with Names Abbreviated 1 2 3 4 5 6 7 10 11 13 12 Estimator YML YMM UWL UWM WRH WRM WRL MDM MDL MDH TMM Colorado Quadrats Sightability Corrected Sightability Increase 1456 1422 1 0.92 89 196 293 1.49 188 375 1.99 1702 3183 1.87 551 851 1.54 716 1397 1.95 118 154 1.31 181 1133 6.26 1025 1403 1.37 561 6,782 841 11,052 1.50 1.63 Idaho Sightability Estimate Summary Statistics Total Deer Counted All Strata (sum Nh) Total Estimated Deer All Strata (sum N\) Total Variance All Strata (sum Var"[N\]) 1,180 11,052 4,534,872 Coefficient of Variation CV"(NA) % 90% Confidence Interval for N" [Lower][Upper] 19.27 7,549 Due to Sampling (4,365,014), Sightability (149,173). Model (20,685) 14,555 .:!: 32% of Population Estimate 1Sightability corrected estimate based on pooling strata 1 and 2 resulting in no estimate for strata 2-YMM. All Total �130 Table 5. Summary of computer modeled, aerial helicopter survey, and sportsmen estimates of December post-hunting season mule deer population size in Rangely Deer Analysis Unit D-6, Colorado, 1990 - 2001. Year 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2000 2000 2001 Computer1 Model 8,017 8,016 7,563 7,917 9,141 9,171 8,409 7,801 7,856 8,176 7,312 Computer3 Model 7,0004 Aerial Survey 90 Percent Confidence Interval 21,6305 13,5965 9.309 - 33,951 8,169- 19,023 6,7826 6,481 7 11,0528 4,285 - 9,279 Sportsman Estimate9 1,750 7,549 - 14,555 6,9892 1Computer model constructed in February 2001 with POP-II software (Bartholow 2000) . 2Estimate represents a value projected for December 2001 given assumptions about likely deer recruitment and harvest June to December 2001. 3Computer model constructed in February 2000 with POP-II software (Bartholow 2000). Estimate represented a value projected for December 2000 given assumptions about likely deer recruitment and harvest June to December 2000. Estimate represents value contested by Colorado Mule Deer Association. 5Estimate from helicopter line transects (White et al. 1989) conducted on a trial basis. 6 Estimate from Colorado helicopter quad rat survey technique assuming 100% deer sightability. • 7Estimate from Colorado helicopter quadrat survey technique assuming 100% Idaho mule deer sightability model (Unsworth et al. 1994). 8Estimate from Colorado helicopter quadrat survey technique adjusted for Idaho mule deer sightability model incorporating sightability correction factors (Unsworth et al. 1994). 9Estimate provided by Colorado Mule Deer Association on behalf of sportsmen. 4 �131 LITERATURE CITED Ackerman, B. B. 1988. Visibility bias of mule deer aerial census procedures in southeast Idaho. Dissertation, University of Idaho, Moscow, Idaho USA. Anderson, C.R., Jr., and F. G. Lindzey. 1996. A sightability model for moose developed from helicopter surveys. Wildlife Society Bulletin 24:247-259. Anderson, C.R., Jr., D.S. Moody, 8. L. Smith, F. G. Lindzey, and R. P. Lanka. 1998. Development and evaluation of sightability models for summer elk surveys. Journal of Wildlife Management 62: 1055-1066. Bartholow, J. 2000. Pop-II for Windows©, version 1.0. Fossil Creek software. Fort Collins, CO USA. Bartmann, R.M. 1983. Appraisal of a quad rat census for mule deer in pinyon-juniper vegetation. Colorado Division of Wildlife Game Information Leaflet 109. Colorado Division of Wildlife, Fort Collins, CO USA Bartmann, R.M. 2000. Colorado mule deer population monitoring system procedures, user's manual. Colorado Division of Wildlife, Fort Collins, CO USA. Bartmann, R.M., L.H. Carpenter, R.A. Garrott, and D.C. Bowden. 1986. Accuracy of helicopter counts of mule deer in pinyon-juniper woodland. Wildlife Society Bulletin 14:356-363. Bartmann, R.M., G.C. White, L.H. Carpenter, and R.A. Garrott. 1987. Aerial mark-recapture estimates of confined mule deer in pinyon-juniper woodland. Journal of Wildlife Management 51 :41-46. Bowden, D.C., and R.C. Kufeld. 1995. Generalized mark-sight population size estimation applied to Colorado moose. Journal of Wildlife Management 59:840-851. Bowden, D.C., G.C. White, and R.M. Bartmann. 2000. Optimal allocation of sampling effort for monitoring a harvested mule deer population. Journal of Wildlife Management 64: 1013-1024. Caughley, G. 1974. Bias in aerial survey. Journal of Wildlife Management 38:921-933. Cogan, R.D., and D.R. Diefenbach. 1998. Effect of undercounting and model selection on a sightabilityadjustment estimator for elk. Journal of Wildlife Management 62:269-279. Colorado Division of Wildlife. 1991. POPII simulation model, DEAMAN database manager, and POPMOD parameter estimation workshop manual. Colorado Division of Wildlife. Fort Collins, CO USA Colorado Division of Wildlife. 2001. WRIS database. Colorado Division of Wildlife, Grand Junction, CO,USA Freddy, D.J. 1994. Estimating survival rates of elk and developing techniques to estimate population size. Colorado Division of Wildlife Game Research Report. July: 27-42. Colorado Division of Wildlife, Fort Collins, CO USA. Freddy, D.J. 1998. Estimating survival rates of elk and developing techniques to estimate population size. Colorado Division of Wildlife Game Research Report. July: 177-206. Colorado Division of Wildlife, Fort Collins, CO USA. Gill, R. 8. 1969. A quadrat count system for estimating game populations. Colorado Game, Fish, and Parks Game Information Leaflet 76. Colorado Division of Wildlife, Fort Collins, CO USA Gill, R.8., L.H. Carpenter, and D.C. Bowden. 1983. Monitoring large animal populations: the Colorado experience. Transactions North American Wildlife Conference 48:330-341. Kufeld, R.C., J.H. Olterman, and D.C. Bowden. 1980. A helicopter quadrat census for mule deer on Uncompahgre Plateau, Colorado. Journal of Wildlife Management 44:632-639. Neal, AK., G.C. White, R.B. Gill, D.F. Reed, and J.H. Olterman. 1993. Evaluation of mark-resight model assumptions for estimating mountain sheep numbers. Journal of Wildlife Management 57:436-450. Otten, M. R. M., J. 8. Haufler, S. R. Winterstein, and L. C. Bender. 1993. An aerial censusing procedure for elk in Michigan. Wildlife Society Bulletin 21 :73-80. Pojar, T.M., D.C. Bowden, and R.8. Gill. 1995. Aerial counting experiments to estimate pronghorn density and herd structure. Journal of Wildlife Management 59: 117-128. Samuel, M.D., E.O. Garton, M.W. Schlegel, and R.G. Carson. 1987. Visibility bias during serial sruveys of elk in northcentral Idaho. Journal of Wildlife Management 51 :622-630. Schrupp, D.L., W.A. Reiners, T.G. Thompson, LE. O'Brien, J.A. Kindler, M.B. Wunder, J.F. Lowsky, J.C. Buoy, L. Satcowitz, A.L. Cade, J.D. Stark, K.L. Driese, T.W. Owens, S.J. Russo, and F. D'Erchia. 2000. Colorado Gap Analysis Program: A geographic approach to planning for biological diversity- Final Report. USGS Biological Resources Division, Gap analysis Program and Colorado Division of Wildlife, Denver, CO USA. �132 Steinert, S.F., H.D. Riffel, and G.C. white. 1994. Comparison of big game harvest estimates from check station and telephone surveys. Journal of Wildlife Management 57:336-341. Steinhorst, R.K., and M.D. Samuel. 1989. Sightability adjustment methods for aerial surveys of wildlife populations. Biometrics 45:415-425. Thompson, W.L., G.C. White, and C. Gowan. 1998. Monitoring vertebrate populations. Academic Press, Inc., San Diego, California, USA. Unsworth, J.W., L. Kuck, and E.O. Garton. 1990. Elk sightability model validation at the National Bison Range, Montana. Wildlife Society Bulletin. 18:113-115. Unsworth, J.W., F.A. Leban, D.J. Leptich, E.O. Garton, and P. Zager. 1994. Aerial survey: user's manual, with practical tips for designing and conducting aerial big game surveys. Idaho Department of Fish and Game, Boise, ID USA. White, G.C., and R.M. Bartmann. 1998. Mule deer management-what should be monitored? Pages 104-118 in J.C. deVos, Jr., editor. Proceedings of the 1997 deer/elk workshop, Rio Rico, Arizona. Arizona Game and Fish Department, Phoenix, Arizona, USA. White, G.C., R.A. Garrott, R.M. Bartmann, L.H. Carpenter, and AW. Alldredge. 1987. Survival of mule deer in northwest Colorado. Journal of Wildlife Managemetn 51 :852-859. White, G.C., R.M. Bartmann, L.H. Carpenter, and R.A. Garrott. 1989. Evaluation of aerial line transects for estimating mule deer densities. Journal of Wildlife Management 53:625-635. ��..... .i::. w i i CALCULATIONS FOR TOTAL NUMBER SAMPLE OUAD~ATS AND SAMPLE SIZES FOR EACH STATUM SAMPLE SIZES BASED ON FORMULAS ON PAGES 341 AND 342 FOR STRATIFIED RANDOM SAMPLING IN: THOMP=s'""o;-,.N,....,- - - - - , 1 - - - - + - - - - · · · · · · · •• • • ···-···-t-----l .,,w,.,._,....L.-,o=.-=c,...,.W~H=IT=Ec-A-,-,N'""D=-c-=-.--,G=o=w"'"'A.,..,N..,.._-1~9=98.,.._--,M..,..O=N'""1=To-=-R=1=N=G..,..v=E=R=TE=s=RA=-c=T=E-=p=o-=p.,..,u,...LA'""T"'"1o""N,..,.S"".'--,A'-,,,C',-A-,,,.DE=,M,-,l=c'--P=R=E=s-=-s,-,,,..,.Nc-=-.-.-=s..,..A,-cN-'------+------+-----+-·----··-+------l oiEGo CALIFORNIA. usA. ·- •••• •••• r · · · - - - - · T - · · ' - ' - ' - ' : . C . : . : : . ; _ ; . _ , ; = - . c : : . = ~ . c . . . . c . . : = - ~ - ' - " - ' ~ ~ - - - - - + - - - - - + - - - - - 1 - - - - - - i - - - - - l ! ------1------+----f----··•······l------+---I 1--------1----+----+----+------+-----+- -----+ ESTIMATED SAMPLE VARIANCES FOR STRATA°f3ASECfON REGRESSION OF VARIANCE STANDARD DEVIATION (S) i ON MEAN DEER DENSITY FOR.i°192 QLJADRAT DEER SAMPLE UNITS PREVIOUSLY FLOWN IN COLORADO. ····•·····1-------i'i-----l ANALYSIS CONDUCTED BY G.C. WHITE. STRATA VARIANCE TO MEAN REGRESSION WAS S = 3.638 + 1.089 (MEAN DENSITY). , r········1 r -··· ••••••••• ,cc..:_.::.=..;.;c.;.'--'J..--1:----+------+i-----l----1 1------- ···---t----;----+----t------t------+-----,....---------+---------;------+----l j i BECAUSE SAMPLE SIZE CALCULATIONS ARE BASED ON ESTIMATES OF VARIANCES PER STRATA, ANO VARIANCE IS DEPENDENTLY RELATED TO DEER DENSITY, THEN WE SHOULD CALCULATE SAMPLES SIZES REci"OiRED FOR_A_ _ _ _ _--+1------····-··· ·-·, RANGE OF POPULATION SIZES. SPORTSMEN ESTIMATE ABOUT 2000 DEER IN UNIT 10 WHILE CDOW COMPUTER 1 MODEL ESTIMATES ARE 7-8000DEER.• THEREFORE THIS IS THE RANGE OF POPULATION SIZE T6ESTIMATE · • • - - 1 ; - - - - - + - - - - 1 - - - - - - + - - - - - - - - - < -·- ·~···· ~f.~F'~E S)ZE REQUIREMENTS. j i t-:Sc-:-A-:-:M=P'"'Ll""'NG=-:A=REA:-.,-F=RA=-,-,-M=E,...,B,..,A-=s=Eo·oiiJLF=Es'-"'R""'U"C"CAc=Rc'ccYc--=S,-,.,U=R.,.,VE=Y7 F=L-c,IG""'H-=T07W"'"'A.,..,S=-3c-o6cc-4-c-M=1L=E=sA~2--,0~Fc-!-W~H=1=cH~LO"WD~E~N=s=1TY~S=TRA=--='cT=A----1---------+-----1-----1 ENCOMPASS!:[) 113 MILESA2 (31%), MEDIUM DENSITY 157 MILESA2 (43%) AND HIGH DENSITY 94 MILES"2 (26%). i ESTIMATED DEER DENSITIES BELOW ENCOMPASS RANGES OF DENSITIES SEEN ON PINYON-JUNIPER WINTER RANGES SAMPLED WITH QUADRAT SAMPLE UNITS IN COLORADO. VALUES ALLOW ADJUSTMENTS IN SAMPLE SIZE CALCULATIONS BASED ON BEST ESTIMATES OF VARIANCES (S) ASSOCIATED WITH DIFFERENT EXPECTED DEER DENSITIES AND 2 DIFFERENT SIZED SAMPLE UNITS. DENISITES ON 1/4 Ml"2 UNITS EQUAL DENSITY P;_~c...M ...cl_"2c.../_4-'-.- · · - - - - + - - - - - + - ;- - - - - · 1 - ·----+----•·➔ ··~~~~~~~~~:~~~~~--i :VALUES IN TABLE ARE LINKED AND USED IN CALCULATIONS ONNEXTPAGES • ·\··--..... __,___ _ _ _ _ _ _ _ : ·- ·······-· I '"",_ _ _ __,__ __, 1----------=G""'u=e=ss=E!,,,o=-o=,E=E=,R=-D='cE=N=s'""1TY==,Mc+.-c-A""2..,.,,s=T=RA~TA-,--j ESTIMATED VARIANCES ISi PER SAMPLE UNIT SIZE PER RELATIVE DENSITY LOW MEDIUM HIGH PROJECTED LOW DENSITY ! MEDIUM DENSITY HIGH DENSITY J 1-----------11-:--:3:--::M-:-:11"'"A2:-t-1:--::s=1-=-M::-::1,.c-=2-t-:::9..,.4-=-M::-::1,.c-=2-+=-Po=-p=u-:-:LA_,..,,,T;:-::10,.,.>N=-+.-,-1,"'"'4..,.~.,.,.,A'""2-,-u=N=1r""' .... ,.,.1-=-M"'11Ac-=2-=-u"""N:-:::1T=-'1"c-c1,cc--4"'"'M~1,.-=-2 ON!i' 1 MfA2 UNIT ... 114 .~tz uN11 Ml"2 UN1T_-+_ _ _ _ _ 1- - - - l IF TRUE POP. SIZE = t ··-· 2000 TOTAL DEER 1.5 5 1 10 1895 5.00 6.36 14.53 4.05 5.27 9.08 I I 1-----:-40=0-=-o=rotAL ot:=E=R+------:4--+--1,--,1-+i--=-21-:----+----,4,-,-15=3,----+---,--=---l--,c-:,--=---+--=-=---+--c-=-::-:--t---cc-=c---j-----:,-,-,,..,.-4.73 15.62 9.36 26.51 7.99 6.63 i I 6000 TOTAL DEER 5 16 31 5991 =.=--l--="-'--------+----1 21.06 12.08 5.00 7.99 37.40 9.08 26.51 14.80 48.29 8000 TOTAL DEER 7 21 41 7942 5.54 11.26 9.36 I A ..... ! : �CALCULATION OF TOTAL NUMBER OF SAMPLE U!"'ITl~.-~"=)A::-::S-=-cS:-:::U=M=IN..,.G=-=-=D-:-clFc-=-F:-:::E:-:::R=EN_,.T,,_.,D...,,E:-:::E:-:::R,..,P,..,O.,.,P:-::U,e-LA,,,_,.,T,.,.,IO::-::N,...,S,,..,l=-ZE=S·----+------+-----t------l-----+-----; ~~.J?.J.'.SSOCIATED AVERAGE STRATA DENSITIES AND ESTIMATED VARIANCES (S) PER SAMPLE IJ..flll,T-'-S-'-IZ--'E-;;:--:;;-;--'-----:==----t----.;;---;;-;-'-=~--+----:--..,.-----':,-;;c:-::---t PonSize =2000 PonSlze =4000 PonSlze = 6000 PonSlze= mno STRATA DENSITY TOTAL TOTAL(U) (Uh,SNh) (Uh,S 2Nh) (Uh,SNt,) (Uh,S2Nh) (Uh,SNh) (Uh.s 2;,-· ···cuh.sNhl (Uh,S 2Nh) I i 1---YA-M-PA~A_dl,MR,i;OE,i.NALu-M=E-NT-+iG~l~S_.i..N~O,.,;,!.:!~.h.. .... ~RALA·o·NWK,-f.._.,M...,l..,LE..,A_2__ sA_M_P_L.,.E_U..,N...IT..,s_--"'v=A=LU=E'=---l---"'V..,A=LU=E.._ VALUE t VALUE VALUE 1 29 115 465 1882 --~543 2569 575 YAMPA-MONUMENT 2 MEDIUM 30 122 608 30391 806 5349 972 UTAH-WHITE RIVER 3 LOW 32 32 167 879 253 2022 287 UTAH-WHITE RIVER I 4 MEDIUM 23 23 207 1876! 151 1001 479 UPPER WHITE RIVER ; 5 HIGH 54 217 1378 8765 2027 18962 2617 UPPER WHITE RIVER 6 MEDIUM 31 31 281 f5541 483 7551 652 UPPER WHITE RIVER 7 LOW 32 32 168 888 255 2042 290 MASADONA-DINO 10 MEDIUM 44 176 882- - 4407 1170 7758 1410 MASADONA-DINO 11 LOW 20 80 325 1314 379 1794 401 h;--=rw~E,..,LV""E=M.,..,l'""'LEc--'--+-----..,.12cc---t-.,-,M-c=E~D~IU~M--2~9-+---~2=9--+---~26-,3-j---=23'-,0=7+-----4-5=2+----7-0~56 •••••• --- - 609 • MASADONA-DINO 13 HIGH 40 162 1029 6548 1514 14165 1955 TOTALS SUMS TOTALS SQUARED 11 364 1018 6773 33324611 34540 70268 8036 64653941 f(j:z47 106006872 VALUE ! 2873 7770 2610 10090 31604 13734 2636 11269 2006 12835 23609 VALUE 637 1137 356 603 3207 821 360 1650 445 767 2395 VALUE 3533 10641 4012 15981 47458 21752 4052 15434 2467 20329 35452 121036,· ••••• 12378 - 181112 163217028 i SAMPLE SIZE FORMULA FROM COCHRAN 1977 AND PAGE 341 THOMPSON ET AL. 1998 - - - - - - - - - t - - - - i ; - - - - - - r----1 - - - - - - - - + - - - - - + - - - - ·······--··•····--·-····- + - - - - - - f - - - - - - - + - - - - + - - - - - - + - - - - - - + - - - - - - + - - - - - + - - - - - - l - - - - 1 Total Sample 1-Q-'--u_a_dr_a_tU_n_lts___,_(u__,)_=_ _.,__·_··_··_--_·_··_·_-_-_ ••_••_••_••_••_••_••____~_-··_·_ _ _ _--l-_ _ _ _-+-_ _ _ _ 1--_ _ _ _+-_ _ _ _+ - - - - - -·--········ ··-··-··-··••1------1-----1 2 (eN/t)A2 + [sum1 toH (Uh*S Nh)] - - - - - - - - + - - - - - t - - - - t ' - - · · · · - · · · · - - - + - - - - - - + ' - - - - - t - - - - - t - - - - - - t - - - - - f - - - - •······························--+-----1----- I I .... w u, �w O'I CALCULATION OF J n, "'' •--~- •~ S"""PLE UNITS CONTINUED I I SAMPLES SIZES CALCULATED FOR ESTIMATED ERROR OF +/-20% OF MEAN ESTIMAT~._,E,__ _ _ _- + - - - - - - - - - - - - - - - 1 - - - - - - + - - - - - - - + - - - - - - + AT A.!P.ha = 0.10 T;;·rs45di>:f:fo ••••••• T • I ••••••••••••• "T AT Alpha= 0.05 t• 1.96 df>120 ··'······---t------+l------+-------+-----1----------11------+------+--------+------1 ······!·· ·-· POTENTIAL DEER POPI II T"" S121= ;mt O • 4000 6000 8000 Error B()und(e) +i: 20% 400 800 ·---'!1~2~0~0+-----'!'1~6~00~t===========t=========jf-_-_~----~----~---.,~~---_-_-_-_-_-_-_-_~""+f--~~~----~------"1-------~~~------~-_4,_-::_-::_-::_-::_-::_-::_-::_-::_-::_-::_:-r-----l WHEN t = 1.645 1.645 1.645 1.645 1.645 f--·•Error_Term~...,..(e"'7N./"'t):":2=•~·====-=5_=,9_1~2:'.7~__ 2_3_65_0_9_,__5_3_2_14_6-+--::_~~~~~~9""4"'6""0"'3"'7-t------+-------+------+-------+------1--------1-------il-----l WHENt=1.96 1.96 1.86 1.96 1.96 1---e=rr-o-r=Term ·1eN/-t1-,..-2---+---4-1-e4-9.,..._1_6_6_5_9=7+--3=7-4=s-4-4+----.6.6-'6.:38"'9act--------+-------11-------;--------11-------+------;-------+-----1 I Total Sample Units Cul ·samoleUnlts Aloha=0.10 SampleUnlta AIP!:tlll:"!J,06 1 ··----=+-------1! •-----t------+------t·-----t------>------1-----1 366 2101 ·1·e1 136 I 437 273 212 181 I ! - - - - - - - - - - - - - ; - '- - - - - t - - - - + - - - - + - - - - - - - ; - - - - - - + - - - - - t - - - - - - - 1 - - - - - - t - - - - - - ; - - - - - - - ; - - - - - - - + - - - - - - 1 Where For Each Pop. •s::,.,;;. Bound +/- 2n•L Iii> Aloha ,.n n<: Size Total Sample Quadrat Units (u) • 33324511 i 2 (eN/t)"2 + [sum1 to H (Uh*S Nh)J 41849 + 34640 ,~----------'T-~----,._·-----+------il------b--,~--- ~ - L ___ . . ------; Total Sample Units= 437 i I l ,------+------+-------1-------+-------+------1 I ! I �L __ . ! - i ] j ! CALCULATING NUMBER OF SAMPLE UNITS I l . - FOR INDIVIDUAL STRATA ·! I ' I ASSUME POPULATION OF 6000 DEER, Alpha= 0.10, AND TOTAL SAMPLE UNITS a 161, SEE PREVIOUS PAGE FOR CALCULATIONS I ASSUME COSTS OF FLYING QUADRATS IN EACH STRATUM ARE THE SAME i • I I COST TO FLY 1 Ml"2 QUADRAT IN OPEN VEGETATION ABOUT 25 MINUTES OF HELICOPTER TIME ! f COST TO FLY 1/4 Ml"2 QUADRAT IN PINYON JUNIPER VEGETATION ABOUT 20 MINUTES OF HELICOPTER TIME i I j i i ... _______ ! ·---·· TOTAL (U) SAMPLE I % UNITS i %AREA Ml"2 AREA Ml"2 Strata Test (Uh,SNhl .. __ J STRATA_ I DENSITY' TOTAL ! ' SAMPLED FLOWN MILE"2 !SAMPLE i.TNITS VALUE UNITS Sum Column GI AREA IGIS NO.(hl RANK I SAMPLED 2 115 7.9 9 YAMPA-MONUMENT l 1 LOW 29 575! t 9i 8! 12.6 YAMPA-MONUMENT 2 122 972! 15 4 I MEDIUM 30 13 151 i 51 i 14,3 5 ··--··· ••••••••••••••••••••••.. -·-·········· UTAH-WHITE RIVER 3 LOW i 32 32 2871 ...5 141 I 8 33.11 8 UTAH-WHITE RIVER 4 23 23 479! 33 8 I MEDIUM 10 UPPER WHITE RIVER 217 41' 41 5 HIGH i 54 2617' 19 19.0I I ! 101 I MEDIUM 31 33.1 i 10 10 UPPER WHITE RIVER 6 31 6521 331 ····· I 14.3\ UPPER WHITE RIVER 7 32 32 51 14 5 5 LOW 2901 I 12.6! 22 MASADONA-DINO MEDIUM 44 176 1410 22 13 6 10 i I ! -s'i ! 6 7.9: 2 6 MASADONA-DINO 11 LOW 20 80 40_1 I I I i 609; 10 TWELVEMILE 10i 33 33.1 I 10; 12 MEDIUM 29 29 I ! 19.0 MASADONA-DINO 162 31 I 19 8 31 13 HIGH 40 19551 _, i I j 16 161 ! 162!=actual sum I I ...... ··-----! ianomalv See Note Below ! FORMULA FOR ALLOCATING SAMPLE UNITS T9 .. ::§J.f.3ATAJ'_(N~Y.'!1~~-Allocation after Cochran 1977; In ThCl1_!1pson et al. 199~1 .P.~~e 342) : • I STRATA ALLOCATION FORMULA TOTAL SUMS ! 364.149 11 102471 1018 t Uh = U[ UhSNhl / (sum 1 to H UhSNhJ)] I EXAMPLE CALCULATION! ···-·· jUh -- i 68 162 I I I i I ! I ! i I i' i i I t I I I = 161*[575/ 10247] lfor Yampa-Monument Low Density Strata from above. !sample unlts._in this stratum = 9 i ···-· -----! i I i ' ' i i ' i l ! iI i I i NOTE: Due to weather conpromising ability to comolete necessarv fl'.'.f.ng of units on March 3 and 4, 2001, number of sample units flown In the Massadon-Dlnosaur strata were reduced to 21_, 15, and 4 in the high, m~~l. 1:1.r.n., and low dens._l_l}.t_S.tf!~, res.eectlvelv. This reduced total sample units flown to 143 (162.~~-9}'. __ ! . ____ ._________ L__ The 161 sum in column G Is an arithmetic anomalv that must have somethlnci to due with rounding corrections In automatic calculations .. The strata sample sizes when added as whole numbers in column L sum to 162 which was the number of quadrats org_lnallv drawn to be samoled. This problem discovered when we reduced sampling In Massadona strata by 19 quadrats and the total flown for the entire Unit 10 was 143 lie--161-19 did not eaual 143 flownl. : I J 1···· -·· -·~•-··-···•····· 'I I I I ; i' l ;J -.J �138 2. MODIFICATION OF SAMPLE SIZE DURING PROJECT lRandom Subsam~le of lll!t\.~SADONA-DINO Unit 10 Strata ! ---,-+- ·-----·-- -·· ... ·- --· -·., 1· .. -' iMarch 3, 2001 ··---... ····-----•-'"·--Reduce total sample from 59 quadrat units to 40 quadrat units proportionately across strata. Minimum of 4 sample units in Low Density Strata, reduced from 6 (1/3 reduction) Reduction in Low Densitv Established the Prooortion reduced in Hiah and Medium Densitvl Decisions Made and Techniaue of Random Subsamole comoleted with VW. Howard annroval and sunnort Samole reduction was made to allow completion of a reasonable number of samole units to ----·obtain an estimate within limited time frame and flvable weather patterns .. Used Quattro Sample Tool Random function to select subsamples within each of 3 strata .. • .·, •. •• •Howard, Bickle and Freddv discussed (3/2/01 )options for comoletina the Massadona Unit with reduced • samplina, Jnclusive of droooina entire strata ~md/or orooortional reduction in samoles within • each of 3 strata. Reducinq samples in each strata was considered preferable to droooino a strata. I i Howard, Bickle and Freddv also discussed merits of doina or not doina the Yamoa-Monument Strata tinder aiven time constraints. Howard thoui:Jht this was the least orioritv of strata because • ,. • Sportsmen do not believe deer assoicated with the Monument a part of the- available huntable deer • •. 1oooulation. However. CDOW does include Monument deer in the Unit 10 1deer oooulation model. At this staae of discussion Howard and Bickle comfortable with l not comoletina the Yampa-Monument strata. ·_i i Massadona Hiah Densitv Strata; Units Sorted/Ordered bv Samole Tool Selection Order Samole Tool Set to assian random order to a samole of 3t·and took first 21 because duolicate ·•· numbers will aet assianed so kmored ties due to assianed numbers Quads Selection Order Oriainal First 21 • Hiah Numeric Number Assianed Selected Usinq Quattro ListinQ Bv Random :Densitv !Quadrats Eauivalent SamoleTool Order 20-MDH 20 1 1 26-MDH 26 3 2 15-MDH 15 3 3 ·8,-MDH 8 5 4 10 5 5 • ' 10-MDH II 21-MDH. 21 6 6 7· 30--MDH 30 7" 24--MDH 24 9 8 ,9 31-MDH 31 9 19-MDH 19 11 1-0 7-MDH. 14 7 11 16-MDH 16 15 12 3-MDH 16 3 13 ' 4 .16 14 i 4-MDH • •11-MDH 16 15 11 25 18 16 i 25--MDH 19 22 17 '' 22--MDH 1-MDH 1 21 18 13 21 13-MDH 19 9-MDH 9 23 20 29[ 29-MDH 23 21 17-MDH 17 23 12-MDH 12 24 25 23-MDH 23 14-MDH 14 26 18-MDH 26 18 27 28 ' 28-MDH 2 27 I 2-MDH 5-MDH 5 29 311 6-MDH 6 31, 27-MDH 27 I I ' ., t ; I ' l ; i I I .• l I �139 ! •··-···-----···· ···-······· ·; i ·- ·- ··-·····--------···-·····J-........ ..-------·-·· ···-·--·-·········-······---..,-.......... , ......... __ .................... ···-·· --~-~ ! -~-~ ........ ·······-···· -~---. '----~-----------···L.·--------··· : Massadona Medium Density Strata; Units Sorted/Ordered by Sample Tool Selection Order !Sample Tool Set to a~~igl}_randomorder to a samole of 22 and took first 15 because dur::>Ji<2~'E____ . _ ----····· _ !numbers will oet assioned so ionored ties due to assioned numbers ' ! ----------+-----~- - · - - - - - - - - - - + - - - - - - - - - - - - - - - - + - ~ - - - - + - - - - - - - - - - - - - --~!-------1-----+----1-----·----------·-·-t-------l ; ! ------+-------'---------+-------'--+------!-~- ··+---------+-------< LPrigiJJ.§.I . ·····1-Q~u_a_d_s_ ___,i_S_e_le_c_ti_on_O_rd_e_r_1-F_ir_s_t~15_·- - + · - - - - 1 - - - - - + - - - - - - - - - - - ' ' - - - - ......_ t Hioh Numeric i Number Assianed Selected j Densitv Listino l Usina Quattro Bv Random 1 ;,-.: Q=-=u=a:..;:;d;.;..;ra::.::t-=-s-+.:::;E..c,.ou=i;..;_va=l=ec,c_ntc...+-S-=-a=m.;..c.1£..pl'""e_T'""'o'-'o"-I---1-..::.O..:..;rd;:.::e:_;_r_ _- 1 - - - - - + - - - - + - - - - - 1 - - - - - - + - - - - - • -.....- ....10-MDM 10 1 1. 11-MDM 11 2 2 7-MDM 7i 3 3 -+-----i-·-·--·1 20-MDM 20 4 4 ,-..::::.;;__;=..:.:.;:_--+-_ __ _ _ _ . : : ; c . c . + - - - - - - - - - - - - - ' - + - - - - - ' - - 4 - - - - - l - - - - - l - - - - - + - - - - - 21-MDM 21 _5_ _ _ _ _5-'-+-------l'-----+----1-----+--'--'-----i 14-MDM 14 6 6 3-MDM 3 8 7 I 13-MDM 13 8 8 ' 1-MDM 1 8 9 9-MOM 9 11 10 12-MDM 12 15 11 . •. 18-MDM 18 15 12 17-MDM 17 17 13 22-MDM 22 17 14 4-MDM 4 17 15 16-MDM 16 18 ----=2;:_-=M=D---'M'-'---+------"2'+-------'1..::.8+-------'------1------+-··· ..··.... --+-----~----<\ 1 5-MDM 5 19 , .__----+-----+-------+-----_,__~-------1-----l-----+-----•·--, ~8_-=M=D~M~+-----'-8t-----~2=0-+----.......----+------+----+-----------+---..~ 15-MDM 15 20 ~MOM 6 ~ 19-MDM 19 21 Massadona Low Densitv Strata· Units Sorted/Ordered by Sample Tool Selection Order Sample Tool Set to assion random order to a samole of 6 and took first 4 because duolicate numbers will aet assianed so ianored ties due to assianed numbers I Oriainal Hioh Densitv Quadrats 6-MDL 2-MDL 5-MDL 4-MDL 1-MDL 3-MDL Quads Selection Order First 15 Numeric Number Assioned Selected Listina Usina Quattro BvRandom Eauivalent Samele Tool Order 6 1 1 2 3 2 5 3 3 4 4 4 4i 1 3 5 �140 SECTION D 1. D-6, UNIT 10 DEER WINTER RANGE MAP Vegetation Types Within Mule Deer Winter Range in Game Management Unit 10, DAU D-6, Colorado Vegetation based on Gap Analysis Winter Range Boundaries based on COOW WRJS Data. Gap Vegetatioo Types - - Dry Land Oops Irrigated Q-ops i1M1 Foothills and Mruntains ~ Mesic. Upland Shrub ~ Billerbrush Shrub - Big Sagebrush Saltbrush Fats and Flats - Juniper Woodland Pinycn -Jtmiper - Shrub Dcminaled Wetland N Game Management Unit 10 CJ Greascwood Fans and Flats 2. D-6, UNIT 10 DEER WINTER CONCENTRATION AREA MAP Vegetation Types Within Mule Deer Winter Concentration Areas in Game Management Unit 10, DAU D-6, Colorado Vegetation based on Gap Analysis Winter Range Boundaries based on CDOW WRJS Data GAP Vegetalion 'I)pes g- Bitterbmsh Sbrub - Irrigated Crop Big Sagebrush Sallbrusb Fms and Flals - JuniperWoodlland CJ <hmewood F'"1S and Flals - Pinyon - Juniper - Shrub dominated Wetland N Gane Management Unit 10 �141 3. D-6, UNIT 10 DEER WINTER SEVERE RANGE MAP Vegetation Types Within Mule Deer Severe Winter Range in Game Management Unit 10, DAU D-6, Colorado Vegetation based on Gap Analysis Winter Range Boundaries based on CDOW WRIS Data Gap Vcgctatioo.Typcs -IrrigatedQ-op li!llil Bittabrush Shrub Big Sagebrush Saltbiushfais and Flats D Greasewood Fans and Flats Juniper Woodland Pinyon -Jw,ipcr Elli Shrub Dominated Wetland N Game Management Unit 10 �142 SECTION E 1. SAMPLING FRAME AND SAMPLING STRATA MAP Sampling Frame Area, Sampling Strata, Quadrat Sample Units, and Random Sample Unit Points used or Helicopter Counts of Mule Deer to Esitmate Population Size in Unit 10, DAU D-6, Colorado. Grid Shows Cadastral Square Miles and Subjective Rating of Deer Density Based Upon a Helicopter Survey Flight on 6 February 2001. Sampling Slnla and Sample Quadnint ua;,,, -Yanp••Mooumeol (Low) ■ Yanpa-Moolilileat (Medium) Ulah White River (Low) Ulali White River (Medium) Upper While Rive, (High) Upper While RMr (Medium) -UpperWhileRMo-(Low) Massadona-Dioo (Medium) Mmsadooa-DiDo (Law) ■ Twelvemil.• (M.edium) Massadona- DiDo (High) D Sample Quadrat ua;,,, Sample Quadrat Units Nat Flo"'1 D Cadastnil Square Mile Grid E3:::::i=:==:=3:::=:::::fJ'10 l<lofflltSI A • Sample Unit Points NGaneMaoogementUnit 10 2. SAMPLING FRAME AND PRIMARY VEGETATION TYPES Sampling Frame Area and Distribution of Pinyan Juniper and Jnniper Woodlands Within Strata used for Helicopter Counts of Mule Deer to Estimate Population Size in Unit 10, DAU D-6, Colorado. Sampling Slrala and Sample Qumlnmt ua;,,, -Yanpa-Mooumeol· (Low) ■ Yaopa-Monumeot (Medium) Ulah While River (Low) Ulllh While River (Medium) •upperWhilelliva- (High) Upper White Rive, (Medium) -Upper While Rivei-(Low) Massadona • Dino (Medium) -Massadooa-Dioo (I.aw) ■ T-Jvanile (Medium) Massadona-Dioo (High) EE]! J\nperWoodhmd ~ Playon. JDDiper �143 3. YAMPA MONUMENT STRATA 1 AND 2 MAP Yampa - Monument Sampling Strata Showing Grid of 1/4 Mile Points, .Randomly Selected Points, Sample Unit Points, and Quadrat Sample Units 1/4 Mile Square in Size, Circle 1/2 Mile in Diameter used to Restrict Distances Between Quadrat Units. Sampling Sira!a aad Sample Quadrant Units ■ Yainpa- Mooumerit (Low) Yanpa.-Mooument. (Medium) D • • o • Quadnll Sample Units Sample Unit Points Randomly Selected Points 1/4 Mile Gridded Points 1/2 mile ciamctcr 4. UTAH WHITE RIVER STRATA 3 AND 4 MAP Utah White River Sam piing Strata Showing Grid ot 1/4 Mile Points, Randomly Selected Points, Sample Unit Points, and Quadrat Sample Units 1 Mile Square in Size. Circle 1 Mile in Diameter used to Re.strict Distances Between Quadrat Units. ...... E=:=========i=== Sampling Strala aad. Sample Quadrant Units -Ulsh White River (Low) Ulsh White Riv..- (Medium) D Quadrat Sample Units • Sample Unit Points • Randomly Selected Points o 1/4 Milo Gridded Points N Gano Management Unit 10 �144 5. UPPER WHITE RIVER STRATA 5, 6, AND 7 MAP Upper White River Sampling Strata Showing Grid of 1/4 Mile Points, Randomly Selected Points, Sample Unit Points, and Quadrat Sample Units 1/4 and 1 Mile Square in Size. Circle ·112 and 1 Mile in Diameter .used to Restrict Distances Between Quadrat Units. - - 112 mile diameter I mile di""eter A Sampling Slrala ,ad Sample Quadrant Units Upper White River (High) Upper White River (Medium) Upper While River (Low) c;;J Quadnt-Sample Units • Sample Point Location o Randomly Selected Sample Points • 1/4 &pare Mile Gridded Points Game Management Unit 10 N 6. MASSADONA- DINOSAUR STRATA 10, 11, AND 13 MAP Massadona - Dinosaur Sam piing Strata Showing Grid of 1/4. Mile 'Points, Randomly Selected Points, Sample Unit Points, and QuadratSample Units 1/4 Mile Square in Size. Circle 1/2 Mile in Diameter used to Restrict Distances Between Quadrat Units. e;;;=1:=::iaaaaaaaaaaaaaii===4• Mllel >.. Sampling Slrala mcl Sample Quadrant Uniis Massadooa.-Dino (Medium) ~.a-Dmo (Low) -Massadoaa-Dino (Bish) c:::::J ~ Sample Un~ , ~ Sample Units Not Flown • _Satq,ie Unit Points • Randanly Selected Poinls • 1/4 Mile Gridded Points Game Management Unit 10 N �145 7. TWELVEMILE STRATA 12 MAP (Clarification: No strata numbered 8 & 9;11 total strata) Twelveniile Sampling Strata Showing Grid of 1/4 Mile Points, Randomly Selected Points, Sample Unit Points, and Quadrat Sample Units 1 Mile Square in Size. Circle 1 Mile in Diameter used to Restrict Distances Between Quadrat Units. Sampling Strata and Sample Quadrant Units - D i=======E=====~3 • • o Miles A N Twelvcmile (Medimn) Quadrat Sample Units Sample UnitPoints Randomly Selected Points 1/4 Mile Gridded Points Game Management Unit 10 �146 SECTION F 1. SURVEY FLIGHT PROTOCOLS PROCEDURES FOR FLYING HELICOPTER SAMPLE QUADRATS GMU 10, DEER POPULATION ESTIMATE FEBRUARY 2001 COLORADO QUADRAT WITH IDAHO SIGHTABILITY CORRECTIONS D.J. FREDDY, MAMMALS RESEARCH, COLO. DIV. WILDLIFE J.W. UNSWORTH, IDAHO DEPT. FISH & GAME (PROVIDED SIGHTABILITY TECHNIQUE SUGGESTIONS) FEBRUARY 8, 2001 I. BACKGROUND INFORMATION: AREAS OCCUPIED BY DEER IN UNIT 10 WERE DELINEATED (364 SQUARE MILES) AND STRATIFIED ACCORDING TO RELATIVE DEER DENSITIES INTO HIGH, MEDIUM, AN LOW STRATA (11 STRATA) BASED ON A HELICOPTER SURVEY FLIGHT CONDUCTED BY V. GRAHAM (CDOW) IN FEBRUARY 6, 2001. THE OCCUPIED DEER RANGE DELINEATED REPRESENTS A CONDENSED PORTION OF THE EXTENSIVE WINTER RANGE AREA DELINEATED IN THE CDOW WRIS INVENTORY SYSTEM BECAUSE SNOW-DEPTHS HAVE CONCENTRATED DEER TO SOME EXTENT. SAMPLE UNITS, OR QUADRATS, WITH AN AREA SIZE OF 1 SQUARE MILE (USED IN OPEN VEGETATION HABITATS) OR 1/4 SQUARE MILE (USED IN PINYONJUNIPER FORESTED HABITATS) WERE SELECTED AT RANDOM WITHIN EACH STRATUM ACCORDING TO STANDARD STATISTICAL SAMPLING FORMULAS. BOUNDARIES OF ALL SAMPLE UNITS WERE BASED ON TOPOGRAPHIC OR CULTURAL FEATURES. FROM THE STRATIFIED RANDOM SAMPLE OF QUADRATS WE WILL OBTAIN 2 ESTIMATES OF POPULATION SIZE FOR THE 364 SQUARE MILE AREA SAMPLED: 1) AN ESTIMATE BASED ON UNADJUSTED COUNTS OF DEER ON QUADRATS, AND, 2) AN ESTIMATE BASED ON COUNTS. OF DEER ON EACH QUADRAT ADJUSTED FOR SIGHTING OR DETECTION PROBABILITY OF EACH GROUP OF DEER USING A SIGHTING BIAS CORRECTION FACTOR (IDAHO MULE DEER SIGHTABILITY MODEL). II. AIRCRAFT: THE PRIMARY HELICOPTER WILL BE A HILLER 12E SOLOY AND THE SECONDARY HELICOPTER WILL LIKELY BE FRENCH A-STAR OR TWIN-STAR, ALL TURBINE POWERED AIRCRAFT. THE HILLER PROVIDES THE BEST VISIBILITY PLATFORM FOR COUNTING DEER AND WILL BE USED PRIMARILY IN AREAS OF HIGHER DEER DENSITY. FLIGHT CREWS WILL CONSIST OF A PRIMARY OBSERVER, NAVIGATOR, AND PILOT. ALL 3 PERSONS SHOULD SIT ABREAST IN THE HELICOPTER WITH THE NAVIGATOR POSITIONED IN THE MIDDLE. THE HELICOPTER MUST HAVE FUNCTIONING INTERCOM HEADSETS SO THAT ALL 3 MEMBERS CAN READILY COMMUNICATE VOICE INSTRUCTIONS OR INFORMATION. SUNNY AND COOL DAYS WITH GOOD SNOW BACKGROUND AND LOW WIND SPEEDS ARE THE MOST DESIRABLE FLYING AND COUNTING CONDITIONS. HAZY OR FLAT LIGHTING ON OVERCAST DAYS IS ACCEPTABLE, BUT NOT PREFERRED. CREWS SHOULD AVOID PUSHING TO GET WORK ACCOMPLISHED IF THERE ARE CONSTANT SNOW FLURRIES OR WIND SPEEDS THAT NECESSITATE FLYING AT HIGHER SPEEDS AND ELEVATIONS ABOVE THE GROUND. Ill. SAFETY: HELICOPTERS WILL HAVE A SURVIVAL GEAR BAG FOR SUPPORTING 3 PERSONS, A FUNCTIONING ELT, FIRE EXTINGUISHER, AND PORTABLE PACKSET FOR EMERGENCY COMMUNICATIONS. IT IS RECOMMENDED FLIGHT CREWS PERIODICALLY REPORT THEIR GENERAL LOCATION TO COUNTY SHERIFF DISPATCH VIA HELICOPTER RADIO IF POSSIBLE. FLIGHT CREW MEMBERS ARE ENCOURAGED TO WEAR NOVEX FLIGHT SUITS AND CLOTHING MADE ONLY OF COTTON OR WOOL FIBERS, NO SYNTHETICS. PILOT, NAVIGATOR, AND OBSERVER MUST ALL WORK TOGETHER TO DETECT AND COMMUNICATE THE PRESENCE OF POWER LINES WITHIN THE WORKING AREA. IF BUILDINGS ARE IN THE AREA, ALWAYS ASSUME THAT A POWER LINE IS NEARBY. IF WEATHER CONDITIONS DETERIORATE, CREWS MUST RECOGNIZE THE CHANCE FOR ICING CONDITIONS OR VISIBILITY CONDITIONS THAT CAN GREATLY COMPROMISE SAFETY. PLANNING SHOULD INSURE THAT THE FUEL TRUCK AND DRIVER ARE AT A LOCATION KNOWN TO THE PILOT AND WITHIN 15-20 MINUTES FLIGHT TIME OF THE HELICOPTER'S ANTICIPATED DESTINATION AFTER FLYING FOR 1 HOUR AND 45 MINUTES. �147 IV. PERSONNEL: A. THE PRIMARY OBSERVER SHOULD BE A PERSON EXPERIENCED IN DETECTING AND COUNTING DEER FROM A HELICOPTER, CAPABLE OF CONCENTRATING AND FLYING SEVERAL HOURS/DAY FOR SEVERAL DAYS, CAPABLE OF MAKING RAPID DECISIONS REGARDING GROUP SIZE AND SIGHTABILITY VARIABLES, AND CAPABLE OF ACCURATELY RECORDING DATA ONTO A TAPE RECORDER AND TRANSCRIBING THAT DATA TO DATA FORMS. THE PRIMARY OBSERVER IS PRIMARILY RESPONSIBLE FOR DETECTING GROUPS FORWARD AND TO THE RIGHT OF THE AIRCRAFT AND TOTALLY RESPONSIBLE FOR RECORDING GROUP SIZE AND SIGHTING VARIABLES OF ANY DEER GROUP SEEN. THIS PERSON MUST WORK IN COORDINATION WITH THE NAVIGATOR AND PILOT TO POSITION THE AIRCRAFT AT A FAVORABLE ALTITUDE AND SPEED. THE PRIMARY OBSERVER IS RESPONSIBLE FOR RECORDING ALL DATA PERTINENT TO EACH QUADRAT AND SHOULD REFER TO THE OBSERVER 'CHEAT SHEET' FREQUENTLY TO INSURE THAT PROPER DATA ARE RECORDED. B. THE NAVIGATOR SHOULD BE A PERSON CAPABLE OF NAVIGATING THE HELICOPTER TO THE PROPER LOCATION OF EACH SAMPLE QUADRAT USING LAT/LONG COORDINATES AND TOPOGRAPHIC MAPS PREPARED FOR EACH QUADRAT. THE NAVIGATOR MUST BE ABLE TO DIRECT THE PILOT TO FLY THE CORRECT BOUNDARY OF THE QUADRAT BASED ON VISUAL INTERPRETATION OF THE TOPOGRAPHIC MAP AND THE TERRAIN OVER WHICH THE HELICOPTER IS FLYING. FURTHERMORE, THE NAVIGATOR IS RESPONSIBLE FOR DIRECTING THE FLIGHT PATH TRAVERSED THROUGH THE QUADRAT TO INSURE PROPER 100% COVERAGE AND COUNTING CONDITIONS. THE NAVIGATOR WORKS IN COORDINATION WITH THE PRIMARY OBSERVER AND THE PILOT. NAVIGATOR ASSISTS THE PRIMARY OBSERVER BY DETECTING GROUPS FORWARD AND TO THE LEFT OF THE AIRCRAFT, AND ASSISTS THE PRIMARY OBSERVER IN KEEPING DIFFERENT GROUPS OF DEER SEPARATE. THE NAVIGATOR SHOULD KEEP A RUNNING TALLY OF DEER COUNTED ON EACH QUADRAT USING A TALLY-WHACKER COUNTING DEVICE. THIS PROVIDES A BASIC BACKUP TO THE TAPE RECORDER OF THE PRIMARY OBSERVER. ALTHOUGH THE OBSERVER AND NAVIGATOR WILL USE THE PRESENCE OF DEER TRACKS IN SNOW AS AN INDICATOR THAT DEER ARE PRESENT, THE NAVIGATOR SHOULD MAKE SURE THAT GROUPS OF DEER ARE NOT ACTUALLY FOUND BY TRACKING DOWN INDIVIDUAL ANIMALS BY FOLLOWING SETS OF TRACKS WITH THE HELICOPTER. SIGHTABILITY MODELS ARE BASED ON VISUAL DETECTION OF ANIMALS, NOT FROM TRACKING. C. THE PILOT'S PRIMARY RESPONSIBILITY IS TO CONCENTRATE ON FLYING THE AIRCRAFT SAFELY AND IN A MANNER THAT SUPPORTS THE PRIMARY OBSERVER. THE PILOT WILL DETECT GROUPS OF DEER NOT SEEN BY THE PRIMARY OBSERVER OR NAVIGATOR. THE PILOT WILL RELAY INFORMATION TO THE NAVIGATOR ON GROUPS HE SEES AND THE OBSERVER AND NAVIGATOR WILL COLLECT THE DATA FOR SUCH GROUPS. V. DATA COLLECTED: A. QUADRAT: AT THE BEGINNING OF EACH QUADRAT SAMPLE UNIT, THE PRIMARY OBSERVER WILL RECORD THE QUADRAT IDENTIFICATION NUMBER, APPROXIMATE LAT/LONG STARTING POINT, GENERAL FLIGHT LIGHT, SNOW, AND WIND COUNTING CONDITIONS, AND CLOCK STARTING ENDING TIME FOR EACH QUADRAT. NAVIGATOR SHOULD ASSIST OBSERVER IN MAKING SURE THIS INFORMATION IS RECORDED. B. DEER: THE PRIMARY OBSERVER WILL COUNT EACH GROUP OF DEER DETECTED AND DETERMINE SIGHTING VARIABLES FOR EACH GROUP, REGARDLESS OF WHO FIRST DETECTS THE GROUP. ON THE TAPE RECORDER THE OBSERVER WILL SAY NEXT OR NEW GROUP OF DEER WHEN EACH GROUP IS DETECTED. THE OBSERVER WILL RECORD THE ACTIVITY OF THE DEER WHO IS MOST ACTIVE WHEN THE GROUP IS FIRST DETECTED, THE VEGETATION TYPE AND RELATIVE PERCENT SNOW COVER IN THE CIRCULAR AREA WITHIN 10 METERS (30 FEED OF WHERE THE INITIAL DEER WAS DETECTED, AND THE TOTAL NUMBER OF DEER IN EACH GROUP. THESE 4 VARIABLES CONSTITUTE THE IDAHO SIGHTABILITY MODEL INPUT VARIABLES. SIGHTING VARIABLES: 1. DEER ACTIVITY IS EITHER BEDDED, STANDING, OR MOVING. THE ACTIVITY IS RECORDED FOR THE DEER WITHIN THE GROUP THAT IS MOST ACTIVE WHEN THE GROUP IS FIRST DETECTED. A DEER THAT IS GETTING UP FROM A BEDDED POSITION WHEN FIRST DETECTED IS RECORDED AS A STANDING DEER BECAUSE THE MOVEMENT OF THE DEER GETTING UP IS LIKELY WHY THE DEER WAS DETECTED. A DEER MUST REMAIN BEDDED FOR A BRIEF PERIOD AFTER INITIAL DETECTION TO BE RECORDED AS BEDDED. STANDING DEER MUST REMAIN RELATIVELY �148 MOTIONLESS AFTER INITIAL DETECTED, AND MOVING DEER ARE DEER THAT ARE WALKING TO RUNNING. 2. VEGETATION TYPE WILL BE ONE OF SEVERAL CATEGORIES AND OBSERVERS SHOULD DO THEIR BEST TO CLASSIFY VEGETATION PROPERLY TO THE MOST DOMINANT TYPE AT THE LOCATION WHERE GROUP FIRST DETECTED BUT DO NOT SPEND A GREAT DEAL OF TIME TRYING TO DIFFERENTIATE LOW BRUSH VEGETATION TYPES. IN ALL LIKELIHOOD DURING DATA SUMMARIES AND ANALYSES, VEGETATION TYPES WILL BE POOLED INTO LOW BRUSH TYPES VERSUS PINYON AND JUNIPER TYPES. VEGETATION TYPES WERE BASED ON CLASSIFICATIONS USED IN THE GAP DATABASE FOR UNIT 10. VEGETATION TYPES SAGEBRUSH BITTERBRUSH GREASEWOOD FLATS SALTBUSH FLATS PINYON-JUNIPER WOODLAND JUNIPER WOODLAND AGRICULTURAL AND NATURAL CLEARINGS RIPARIAN SHRUB TALL CONIFER SIGHTABILITY CORRECTIONS FOR VEGETATION TYPES WILL MOST LIKELY BE BASED ON CORRECTION FACTORS DEVELOPED IN IDAHO FOR THEIR VEGETATION CLASSES OF 1) GRASS/OPEN/AGRICULTURE, 2) SAGEBRUSH, 3)JUNIPER/MOUNTAIN MAHOGANY, AND 4), POSSIBLY CONIFER. NO CORRECTION FACTORS HAVE BEEN DIRECTLY DEVELOPED FOR DEER IN OUR PINYON-JUNIPER OR JUNIPER WOODLAND HABITATS SO WE MUST USE IDAHO INFORMATION TO APPROXIMATE OUR VEGETATION CONDITIONS. 3. PERCENT SNOW COVER AT THE LOCATION WHERE EACH GROUP OF DEER IS FIRST DETECTED. SNOW COVER PERCENTAGES WILL BE CLASSIFIED BY THE OBSERVER AS LOW= 079%, AND HIGH= >80% OF THE GROUND COVERED BY SNOW. THESE CLASSIFICATIONS MUST BE USED TO BEST MATCH THE IDAHO SIGHTABILITY MODEL. IDAHO DOES HAVE A SNOW COVER CLASSIFICATION OF 0-19% BUT THIS SITUATION APPLIES TO COUNTS OF DEER DURING SPRING-GREENUP WHEN DEER ARE IN LARGER GROUPS IN OPEN HABITATS. 4. TOTAL NUMBER OF DEER IN EACH GROUP. A NUMERIC COUNT OF ALL DEER SEEN IN EACH GROUP. DEER WILL NOT BE CLASSIFIED TO AGE OR SEX. VI. IN-FLIGHT PROCEDURES 1. NAVIGATOR AND OBSERVER WILL OBTAIN PROPER MAPS, ARRANGE IN ORDER OF NEED, AND DECIDE ON GENERAL ROUTE TO QUADRATS TO BE FLOWN DURING EACH FUEL LOAD. PRIMARY OBSERVER MAKES SURE THAT TAPE RECORDER IS FUNCTIONING AND ADDITIONAL BATTERIES AND TAPES ARE AVAIL.ABLE ..NAVIGATOR SHOULD HAVE A SPARE TAPE ~ECORDER AVAILABLE. . . 2. OBTAIN A GPS POSITION AT STAR"J:ING POINT AND NOTE A GENERAL DESCRIPTION, ie SE OR NW CORNER, MAKE SURE CREW IS ON PROPER QUADRAT SAMPLE UNIT. 3. FLY PERIMETER OF QUAD RAT FIRST IN A CLOCKWISE MANNER SO THE INSIDE OF THE QUAD RAT IS TO THE RIGHT OF THE PRIMARY OBSERVER. FLIGHT SPEEDS SHOULD BE 40-50MPH AT ABOUT 100 FEET ABOVE TERRAIN. IF HIGHER SPEEDS ARE NEEDED TO BE SAFE DUE TO WIND SPEEDS, CONSIDER ABORTING THE FLIGHT. SOME QUADRATS INCLUDE LOWLAND PRIVATE LAND WITH HOUSES, LIVESTOCK, ETC. USE YOUR BEST JUDGEMENT AS TO WHAT AREAS YOU NEED TO FLY TO SEARCH FOR DEER AND AVOID BUILDINGS & LIVESTOCK. 4. DETERMINE STATUS OF GROUPS OF DEER ON PERIMETER. A. DEER MOVING OFF THE QUADRAT WHEN DETECTED ARE CONSIDERED ON THE QUADRAT. B. DEER MOVING ONTO THE QUADRAT WHEN DETECTED ARE CONSIDERED OFF THE QUADRAT. C. IF A GROUP IS STANDING ON THE PERIMETER BOUNDARY, COUNT THOSE DEER INSIDE THE QUADRAT. �149 D. RESIST THE TEMPTATION TO LEAVE THE PERIMETER TO COUNT A GROUP ON THE INSIDE OF THE QUADRAT BEFORE COMPLETING THE PERIMETER. USE YOUR BEST JUDGEMENT AT THE TIME OF DETECTION. 5. FLY INTERIOR OF QUADRAT. A. FLY THE INTERIOR OF THE QUADRAT SYSTEMATICALLY IN STRIPS OR STRIP-CONTOURS AT ABOVE RECOMMEND AIR SPEEDS AND AGL. USE PROMINENT TERRAIN FEATURES TO DIVIDE THE QUADRAT INTO SMALLER COUNTING BLOCKS AND SEARCH EACH SUB-BLOCK INTENSELY. IF YOU DECIDE TO FLY STRIP-CONTOURS, WORKING FROM THE LOWEST TO HIGHEST ELEVATION USUALLY WORKS BEST AS DEER ARE MORE RELUCTANT TO RUN UPHILL THAN DOWNHILL. ON OPEN FLAT TERRAIN, SYSTEMATIC STRIPS WORK WELL WHEN USING LANDMARKS OR GPS LOCATIONS. TRY TO MENTALLY KEEP TRACK OF DEER GROUPS TO AVOID DOUBLE-COUNTING. B. BE PATIENT AND STRIVE FOR 100% COVERAGE OF THE QUADRAT. 6. WHEN MULTIPLE GROUPS OF DEER ARE DETECTED SIM ULTANEOUSLY, FOLLOW THESES GUIDELINES. A. PRIMARY OBSERVER SHOULD BEGIN OBTAINING DATA ON GROUP NEAREST THE HELICOPTER AND USE HELICOPTER TO KEEP GROUPS SEPARATED. B. NAVIGATOR SHOULD FOCUS MOMENTARILY ON SECOND GROUP OBSERVED AND NOTE LOCATION, ACTIVITY AND NUMBER OF ANIMALS FIRST SEEN. C. AFTER COMPLETING DATA COLLECTION ON FIRST GROUP, PROCEED TO LOCATION OF WHERE SECOND GROUP FIRST SEEN, DETERMINE VEGETATION TYPE, AND SNOW COVER% AT SITE OF DETECTION, THEN FIND AND COUNT SECOND GROUP. 7. AFTER EACH GROUP OF DEER IS COUNTED, OBSERVER SHOULD VERBALLY NOTIFY NAVIGATOR OF GROUP TOTAL SO NAVIGATOR CAN RECORD A RUNNING SUM OF DEER COUNTED ON THE TALLY-WHACKER. AT THE END OF THE QUADRAT, OBSERVER SHOULD TAPE RECORD THE TALLY-WHACKER SUM AS A CRUDE CHECK ON THE NUMBER OF DEER DETECTED. NAVIGATOR COULD RECORD TALLY-WHACKER SUM ON THE FLIGHT MAP ON THE QUADRAT. 8. BEFORE LEAVING THE QUADRAT, MAKE SURE YOU HAVE NOT FAILED TO SEARCH ANY OBVIOUS GEOGRAPHIC PORTIONS OF THE QUADRAT. OBSERVER AND NAVIGATOR SHOULD AGREE THEY ARE DONE BEFORE PROCEEDING TO NEXT QUADRAT. VII. DATA TRANSCRIPTION 1. PRIMARY OBSERVERS WILL BE RESPONSIBLE FOR TRANSCRIBING THEIR TAPE-RECORDED DATA ONTO STANDARD DATA FORMS. ALL OBSERVERS WILL LABEL EACH OF THEIR TAPES AS TO THEIR NAME AND DATE. DATA SHOULD BE TRANSCRIBED THE SAME DAY AS COLLECTED. IT MAY BE POSSIBLE TO HAVE AN ADDITIONAL PERSON TRANSCRIBE THE TAPES TO SAVE TIME, AND SUCH PERSON WOULD NOTE ANY QUESTIONS THAT ONLY THE OBSERVER COULD ANSWER REGARDING ANY PROBLEMS ON THE TAPE RECORDINGS. 2. STANDARD DATA FORMS WILL BE AVAILABLE AT PROJECT LOCATION. END C:\DEERCENSUS\QUADPROC.MEM �150 2. SURVEY DATA FORM ~T~TARECXHlf\13 FCH\11 CEER C?MJ 10 2CXJ1 cµid4imPR:c:'deertEr&J (IVVD'Y) 1-«:msllM:: SfAAT _ _ _ _ 8'0 JI-ER: lBIP.(F} _ WN) _ _ _ UGifCXW. Sf\ONTYPE ~ WT------NtMG4.TCR Cl3SERvffi - - - - - '/:>C!E_a=_ I llM=SfAAT: QUAD# _ _ STRA.lllv1 - - - EJ'+l)Cl..W): _ _ _ TOT.TIM:: LCRAN'GPS l.CCA.llOJ arosrARTII\G PONf: TOfAL C:RCl.JP "°· DEER VEC:E foCTlVllY lYPE o/cfN:J/1/ COffi. C:RO.P Sl2E TOfALl:E.ERSEEN IHINllCNS: WNJ. Ligi=l, Mxierae--M, Slrorg=S; LJGif COO. 8igtt=B, Hazy=H, CUl::::O; Sl'ON1YPE: Fresh =F less than 48tvs; Od=Odderlhan 48tTs. DEER ACTIVITY: Bedded=B, Stardng=S, Mlvirlf'IVI; VEGE. 1YPE: Big Sagebrush=SG. 8itlerbrush=8B, Qeasewxxi Ras=GI\( Salbush Ras=SB, Pinyon-Ju,i~J. Juriper Wxxlam=J\N Agirulhre & aeairq;=('A. Riparial Shn.b=RP, Tall Caifer=TC PERCENTSt-ONCCMR L.a.v=L=~79%, Hgi=H=>80% �151 3. SURVEY OBSERVER HELP SHEET OBSERVER CHEAT SHEET QUAD STARTING POINT GPS QUAD NUMBER QUAD TIME START/END QUAD WIND, LIGHT, SNOW TYPE GROUP INITIAL DEER ACTIVITY Bedded, Standing, Moving GROUP VEGETATION TYPE Sagebrush, Bitterbrush, Greasewood, Saltbush, PinyonJuniper, Juniper Woodland, Agriculture & Clearings, Riparian Shrub, Tall Conifer GROUP PERCENT SNOW COVER 0-79%=Low >80%=High TOTAL GROUP SIZE All deer counted in each group �152 SECTIONG SURVEY FLIGHT QUAD RAT SAMPLE UNIT MAP INDEX UNIT 10 DEER POPULATION ESTIMATE FEBRUARY 2001 I INDEX TO QUAD RAT SAMPLE UNIT MAPS /QuadMaollstwb3) I FLIGHT SEQUENCE IS ORDER OF FLYING UNITS Wl11-IIN AN ENTIRE BLOCK AREA INDEPENDENT OF STRATA iOUADRAT 11" X 17" FLIGHT I ~·-STRATUM NAME UNITNO. MAP NO. SEQUENCE LOCALE DESCRIPTION STRATUM Yamoa-Monument Low Density 1YML 5 1 I Bear Draw Mantle Ranch Rd. _1 2 Yamoa-Monument Low Density 2YML 5 Bear Draw Mantle Ranch Rd. 1 3YML 5 3 Drv Woman Canvon Mantle Rd Yamoa-Monument Low Densltv 1 4YML 6 4 Yamoa-Monument Low Densltv Drv Woman Canvon Mantle Rd 1 ~ 5YML 6 6 Schoonover Pasture Mantle Rd 1 ' Yamoa-Monument Low Density t 6YML 6 5 Yampa-Monument Low Density Schoonover Pasture Mantle Rd 1 7YML 9 Yamoa-Monument Low Density 18 Sand Canvon Mantle Rd 1 9 Yamoa-Monument Low Density 8YML 20 Chew Ranch 1 Yamoa-Monument Low Density 9YML 10 22 Trail Draw Chew Ranch 1 L. 9 Units I I I t· 1.. I Medium:Densltv - - -22 - - · Yampa-Monument Yampa-Monument Medium Density Yamoa-Monument Medium Densltv Yampa-Monument Medium Densitv Yampa-Monument Medium Densltv L_2 I Yampa-Monument Medium Density 2 ' Yampa-Monument Medium Densitv 1-- ___ 2 , Yampa-Monument Medium Densitv t 2 r-···-· 2 i Yampa-Monument Medium Densitv Yamoa-Monument Medium Density 2 i 2 i Yamoa-Monument Medium Density j........ _2 ______ i Yampa-Monument Medium Densitv 1 2 I YampaoMonuinent Medium Densitv ,2 Yamoa-Monument Medium D.ensitv 2 ; Yamoa-Monument Medium Densitv 2 2 c-··t ; 1-· ____ L_ __ 10YMM 11YMM 12YMM 13YMM 14YMM 15YMM 16YMM 17YMM 18YMM 19YMM 20YMM 21YMM 22YMM 23YMM 24YMM 15 Units 6 7 6 6 7 8 9 8 8 8 9 9 9 10 9 10 10 8 10 12 11 13 14 15 16 17 21 19 23 24 Johnson Canvon Mantle Rd Johnson Draw Mantle Rd Johnson Draw South Mantle Rd West Serviceberrv Draw Mantle Rd Yampa River Overlook Mantle Rd Marthas Peak Dangerous Mantle Ranch Cave Rock Bench Verv Steep Red Rock Ranch Site Steep Pear1 Park North· Pearl Park South Sand Canvon Rim Lower Sand Canvon Pool Creek Rim Stateline-Pool Creek I - I !INDEX TO QUADRAT SAMPLE UNIT MAPS STRATUM -- ---~ - 3 3 3 3 t ·-l ·----------:- STRATUM NAME Utah White River·Low Densitv utah White River Low Densitv Utah White River Low Densitv . Utah White River Low Densitv Utah White River Low Densitv QUADRAT 11" X 17" FLIGHT UNIT NO. MAP NO. SEQUENCE 9~UL 10-UL 11-UL 12-UL 13-UL 5 Units 11 12 13 13 15 2 5 7 6 13 LOCALE DESCRIPTION Stateline W.Drinnina Rock Ck ·DriDDina Rock Ck Open Flats White River Drinnlna Rock Ridae Stateline Drionin!l Rock Ck White River Cliffs i -I ··-···-·· ------···•-:-- l l 4 4 4 .. 4 ···---· 4 4 4 4 Utah White River Medium Densitv Utah White River Medium Densitv Utah White River Medium Densitv Utah White River Medium Density Utah White River Medium Densitv Utah White River Medium Densitv Utah White Rlver Medium Density Utah. White River Medium Densltv .. 1-UM 2-UM 3-UM 4-UM 5-UM 6-UM 7-UM 8-UM 8 Units 11 12 12 14 14 15 15 15 1 3 4 8 910 11 12 SnakeJohnReef W.Dinosaur Stateline Raven Ridae Raven Ridcie Morman Gap Raven Ridge South Raven Rldae South Raven Ridae Southeast Raven Ridae Hardware Draw RavenRldae White River i i i �153 ·- -·-INDEX TO QUADRAf SAMPLE UNIT MAPS .. .. STRATUM . ------ STRATUM NAME ·--·-··--- -··-·1 IOUADRAT 11"x17" FLIGHT UNIT NO. MAP NO. SEDUENCE _j LOCALE DESCRIPTION -~ ·----,I ···-·--·-··· "'t_j';:;~er White River Hiah Densitv 56 On White River i 1-WRH 24 5 l 2-WRH 55 On White River ___ ':!Peer White River !iiah Densltv 24 5 .I ----··-··--··· .. ___ 5____ . __ UooerWhite River Hiah Densitv 3-WRH 54 North of main ridae 1 24 5 Unner White River Hiah Densitv 4-WRH 23 53 Rldoe North Slone l ! I 5-WRH 23 On Ridae 52 5 ___ J!P.ner White River Hiah Densitv 6-WRH 51 On White River 5 Uooer White River Hiah Densitv 23 7-WRH 22 48 On White River ....... 5 _-.L Ueoer White River Hiah Densltv ·8-WRH 47 On White River 22 ____5_ _ J UQQer White River Hiah Densitv 49 North Slone Road 5 ! UnnerWhite River Hiah Densitv 9-WRH 22 - -5· Unner White River Hiah Densitv 10-WRH 22 46 South Slone Road 11-WRH 45 22 On White River r:-.--5 -···. UeperWhite River Hiah Densitv ---22 44 On White River i- _____5______ ... Uooer White River Hiah Densitv 12-WRH 13-WRH North of Coal Rldae 22 50 ... __ § .. _ __ QP.ner White River Hlah Densitv ------·· I, 21 43 5 Unner White River Hiah Densitv 14-WRH North of White River 1·-5 ·-- • • Unner White River Hiah Densitv 42 On White River Cliff 15-WRH 21 16-WRH 41 On White River Dirt Road 21 ~ 5 __\:!P._eer White River Hiah Densitv Unner White River Hiah Densitv 17-WRH 40 Main Road on Ridae 21 18-WRH __!,Jeoer White River Hiah Densitv 37 On White River Cliff & Mine 19 19-WRH Steen Gullies North of Mine Uooer White River Hiah Densitv 19 38 20-WRH __!d_eper Whlte River Hlah Densitv 19 39 Steen Gullies Northwest of Mine 21-WRH 1 Uooer White River Hiah Densltv 19 32 PinvonJufnner Ridae Road Unner White River Hiah Densitv 22-WRH 19 Chase Ck Steen Cliffs Gullev Jct. ---31 23-WRH Too of'Mesa • • ....J:!.eoer White River Hiah .Densltv 19 33 ·5 Unner White River Hlah Densitv 24-WRH 19 34 Coal Oil Rim Mesa - - -5-- J_Qeoer White River Hiah Densitv 25-WRH 18 29 Coal Oil Rim East ! 19 26-WRH 36 Coal Oil Rim Mesa Northeast ........... 5 ___ --1 U_e:i:i:er White River Hiah Densitv 27-WRH 19 35 ... __ 5 ____ .. _ UQDe"r White River Hiah Densitv ChaseCk Mesa 28-WRH 5 Unner White River Hiah Densltv 28 18 Dead Doa Draw Mesa Cliff 1- • _____ 5 ____ _J!Qner White River Hiah Densitv 29-WRH 18 30 Chase Draw Mesa 30-WRH 18 27 Unner Dead Doa Draw l ........§___ - ___ y_pi:i:er_White River Hiah Densitv -----·. 5 Uooer White River Hlah Densitv 31-WRH 18 26 Dead Doa Reservoir 32-WRH 17 25 Dead Doa Reservoir ... ____ ~- _.. . _...\:!..!?.Per White River Hiah Densitv 33-WRH _ ~ _ ___ Upeer White River Hiah Densitv 17 24 Nate Snrina Reservoir ----5 Uooer White River Hiah Densitv 34-WRH 20 Nate Sorina 17 ------~ 35-WRH 17 21 Nate Sorlna East _ § .. _ ..l:!.2e~.Whlte River High Densitf -·· -·· 36-WRH 17 19 Nate Snrinn North ....§. . . . __ Uei:i:er White Riv~J Hiah Densitv 37-WRH Scullion Gulch Lincoln ReservoirUpeer White River High Densl!Y 18 22 _____ .. 5 . --- -I Upper White. River High Den§Lty . 38-WRH 18 23 Scullion Gulch North 5 ! I 18 17 Nate Snrina North Pioeline 5 i Upp~r White.River High Density 39-WRH 17 • 40-WRH Upp~r White River_ High Densl!Y. Unner Nate S~!lli!...._____ .. _. I 5 .1---1?. 41-WRH 16 Upper Whi~e River High Den~J!Y. __ 5 I --·-·----- ....-···---- .. ··_ -- .. 16 ----· -·----- Upeer Nate Sering 41 Units ... ··-·-·· --J I ----- -· .. i ~ [·! -· =R ···.::_:1 I IINDEX TO QUADRAT SAMPLE UNIT MAPS M-! STRATUM NAME -= ··---··· QUADRAT 11" X 17" FLIGHT UNJ1NO. MAP NO. SEQUENCE jmcerWhit~ River Medium Densitv ~ • 6 UQRer White River Medium Densitv ---s·--- Unner White River Medium Densitv 1-WRM 2-WRM Unner White River Medium Densitv 6 _--6--~-1--_IJ_Qoer White River Medium Densitv 6 __ Uooer White River Medium Densitv 6 : Uooer White River Medium Densitv _ 6 ___ ·1! UpoerWhite River Medium Densitv 6 . Unner White River Medium Densitv Unner White River.Medium Densitv 6 4-WRM 5-WRM 6-WRM 7-WRM 8-WRM 9-WRM 10-WRM .10 Units . ! ---·--- --, i -_ , t--· I E 7 ____ ( __ Unner.White River Low Densitv 7 _.......J __ Upoer White River Low Densitv 7 Uoner White River Low Densitv 7 __IJ_Qner White River Low Densitv 7 ___ _JJp_Q!!r White River Low Densltv ___ --. 3-WRM 1-WRL 2-WRL 3-WRL 4-WRL 5-WRL 5 Units 28A 28A 21 20 20 17 16 16 16 16 --LOCALE DESCRIPTION Hwv40 Box Elder CK Red Wash Box Edler Ck Red Wash Prairie Doa Reservoir Raven Park Dam Scullion Gulch Raven Park Dam Scullion Gulch Rock Shale Reservoir Stlnklno Water Ck Stlnklna Water Ck Stlnklria Water Ck .Stiriklna Water Ck 1 2 4 5 6 9 12 13 14 15 I I 21 29 30-A 30 16 S.Hwv40 Red Wash Hwv 40 Red Wash Red-Wash Reservoir #1 S.H\Al\/40 Skvline Reservoir W. of Stinkina Water Ck 3 7 8 10 11 ---; �154 'INDEX TO OUADRAT_S_A_M_P_L_E_U_N_IT_M_A_P_S------,----,-----.--------.---------------, I .OUADRATI 11" x 17" FLIGHT ! i-=cS-=-T'--'R'--'A-'-T-"-U"'-M'1-,-----"S~T~RA~T~U_M_N~A~M_E~---+-U_N_fr_N_O_.+-M~A~P~N~O-=-.+S~E~,Q=U~E~N~C_E-t----'L~O~C~A~L=E~D_E_S_C_R_IP_T_IO~N--1• ..-------'1--------------+--:----t----t------+---------,-----,---------; L_1_Q_ , Massadon-Dino Medium Density 1-MDM 35 57 K-Creek I 10 Massadon-Dino Medium Density 2-MDM 35 58 Miners Draw Trail Ck ~ 10 l Massadon-Dino Medium Density 3-MDM 35 59 Miners Draw Road L 1o i Massadon-Dino Medium Density 4-MDM 34 54 Middle Ck (Flv Contours 66 & 7700) 10 _ Massadon-Dino Medium Densitv 5-MDM 32 41 Twin Wash Dino Headquarters 'ti:!Q~ .. \_ Massadon-Dino Medium Density 6-MDM 28 22 Skull Ck /Flv Contour 6200) 10 Massadon-Dino Medium Density 7-MDM 28 20 Skull Ck Rim 10 Massadon-Dino Medium Density 8-MDM I 28 19 Skull Ck Rim _ _10___ Massadon-Dino Medium Density 9-MDM 27 14 Three Sprinqs --1-0 Massadon-Dlno Medium Densitv 10-MDM 27 13 Three Sprinas ____.__,.____+....:.:.:=::.==.::...=:..:..:..::c....c.cc-=-=-:=.c-'--=--"-'-'-"-='---t-'c=...cc.;-=:-..:..:.:...+---=..:....-+-......:..:::___f-_ _ _---'...:..:.:...::..::....==,'-"-'-=------7 I r• I l--_ _._10_,.____+----'M'"-a::..:s=-=s:..::a:..::d:..::o-'-'n'--'-D=-i"'ncco....cMc:-=-ed':'.i:-"u~m-'-=:-D-'-en~s=.,i~tv-t--:1:--=1-:·M:-:=D-:--M:-t_-'=":27=--i--1:--=2:--_f--_ _ _-=---"'M'-'a':"s::'-s~a=do-=-n-:-a_____ _J 10 10 Massadon-Dino Medium Densltv 12-MDM 27 11 East Massadona Massadon-Dino Medium Density 13-MDM 27 10 Horse Draw E.Massadona . 10 Massadon-Dino Medium Densitv 14-MDM 25 1 Lower 3 Sprinas Draw 10 . Massadon-Dino Medium Densitv 15-MDM 25 3 Lower Peterson Draw l----'-1o=--.---1-....:.M:;.:;a::..:s:..::s:.;;.a"'-d--'-o-'n_-D'--'l-'n_o_M_e_d_i_u_m~D_e_·n_s...,i~tv-t-...,1_6...,-M..,.D-=--M-+__2_5_--t___ 9 _ - - 1 - - ~ -_P_e_te_rs,....o_n_D_ra_w ___R_e_s_e_rv_o_i_r_ _ 1o Massadon-Dino Medium Densitv 17-MDM 25 8 Peterson Draw 10 Massadon-Dino Medium Densitv 18-MDM 26 7 The Slouqhs 1o i Massadon-Dino Medium Density I 19-MDM 26 15 Skull Ck Rim Road 10 , Massadon-Dino Medium Dens.itv 20-MDM 25 6 Petes Post The Slouahs 10 Massadon-Dino Medium Density 21-MDM 26 16 Bear Canyon Sprina 10 Massadon-Dino Medium Densitv 22-MDM 28 18 Unner Skull Ck 22 Units - ---··-·· -- - - - - - - - - - - - - - - - ~ ~ ~ - - - - - + - - - - - - + - - - - - - - - - - - - - - - ; ,__ -:,:,--· ~· 11 ·- - . ~-----· - -----~-1_ - __ , I, 11 j___ ______11... 11 Il ' ,I Massadon-Dino Low Density 1-MDL 25 2 Peterson Draw Reservoir Massadon-Dino Low Densitv 2-MDL 25 4 North Peterson Reservoir Massadon-Dino Low Densitv 3-MDL 25 5 Wolf Creek Soring Massadon-Dino Low Density I 4-MDL 26 17 Petes Post Wolf Ck __ _ Massadon-Dino Low__ D_e~n~si~ity~--t-5_-_M_D_L_,___3_4_-+--_ _5_3_-+----~B=u~ck~w~at~e~r~D~r~aw~----- _... 35- - - - 1 56 - - - - f - - - -K-Creek - Massadon-Dlno Low Densih, ! . l . -6-MDL ----,~-~---· I -~=~-=~~-~==---=~~-- __________-::__s__u_~_n_,_·t_s_~---t~---_-_-_-_-:_--+-_--_-_-_-_~--~1---_-_-_-_~---_-_-_-_-_-_-_-_-_-_-_-_-_~~--··-~·-____·_-_· __ ~INDEXTO QUADRAT SAMPLE UNIT MAPS STRATUM NAME STRATUM - - - - - - -- 12 12 12 i 12 \i-•.0••--·--··-··· 12 , 12 12 12 12 12 I - Tweivemile Medium Density Twelvemile Medium Densitv Twelvemile Medium Density Twelvemile Medium Density Tweivemile Medium Density Tweivemile Medium Density Twelvemile Medium Density Tweivemile Medium Density Twelvemiie Medium Density Twelvemile Medium Density ·- -1 QUADRAT 11" X 17" FLIGHT UNl'T NO. MAP NO. SEQUENCE 1-TMM 2-TMM 3-TMM 4-TMM 5-TMM 6-TMM 7-TMM 8-TMM 9-TMM 10-TMM 10 Units 1 1 2 2 2 3 3 4 4 4 1 2 3 4 5 6 7 8 9 10 LOCALE DESCRIPTION i ----! I -1 Yamoa River Twelvemile Gulch Rd. N.Hwv 40 Radio Tower Road N.Hwv 40 Radio Tower Road N.Hwv 40 Buffalo Gulch S.Hwv40 Sorinas Ridae N.Hwv 40 Buffalo Gulch I N.ElkSorinas Buffalo Gulch Elk Sorinas Road ··Elk Sorinas Road Bav Gulch NW.Elk Sorinas ......! �155 • --· -•--¥ ., i -- ·------- - - - - - - - - - - ~ - - - - - . - - - - - - - . - - - - - - , c - - - - - - - - - - - - - - - , INDEX TO QUADRAT SAMPLE UNIT MAPS STRATUM~! i- STRATUM NAME QUADRAT 11" x 17" FLIGHT UNIT NO. MAP NO. SEQUENCE LOCALE DESCRIPTION ____, f 13-----Massadon-Dino Hiah Densitv 1-MDH 28 21 SkullCk 13 ; Massadon-Dino Hiah Densitv 2-MDH 28 23 Massadona Hwv 40 13 Massadon-Dino Hiah Oensitv 3-MDH 28 24 South of Skull Ck Town Site I ~-....:.1.::.3---+--'-'-'M""a~ss~a_d_o_n~--D-in_o_H~ici~h-D_e_n_s_itv~-+--4--M-D,,..H--,.--i--....:.28--,1---2~5'---t---~~~S~k~u~l....:.IC"'k~H=w-'--'y4-0~~"---1 j 13 13 __ 13_____ 13 13 .---------· 13 3 Massadon-Dino Hiah Densitv 5-MDH 29 26 Miller Ck Massadon-Dino Hiah Densitv 6-MDH 29 27 Unner Jones Twist Massadon-Dino Hiah Densitv 7-MDH 29 28 Jones Twist Hwv 40 Massadon-Dino Hicih Density 8-MDH 29 29 Martin Gao Massadon-Dlno Hicih Density 9-MDH 29 30 Little Red Wash Massadon-Dino Hiah Densitv 10-MDH 30 31 Red Wash Hwv 40 Massadon-Dino Hiah Densitv 11-MDH 30 32 Martin Gao West ti Massadon-Dino Hiah Density 12-MDH 30 33 Red Wash 13 - - • Massadon-Dlno Hlcih Density 13-MDh 30 35 Skvllne Reservoir Hwy 40 13 i Massadon-Dino Hiah Densltv 14-MDH • 30 34 Uooer Red Wash •~3 Massadon-Dino Hiah bensitv 15-MDH 31 • 36 Blue Mountain T-own Site 13 i Massadon-Dino Hiah Density 16-MDH 31 37 Willow Ck Hwv 40 east ! 13 Massadon-Dino High Densltv 17-MDH 31 38 Spencer-Willow Ck Hwv 40 L_.!~------J Massadon-Dino Hiah Densitv 18-MDH 31 39 Soencer Draw Hwv 40 ! 13 , Massadon-Dino Hiah Densitv 19-MDH 32 40 Soencer Draw West Hwv 40 ___.:11__ Massadon-Dino Hiah Densitv 20-MDH 32 42 Dino Headauarters Hwv 40 13 Massadon-Dino Hiah Densltv 21-MDH 32 43 Dinosaur Quarrv 13 Massadon-Dino Hiah Densitv 22-MDH 32 44 Drionina Rock Ck-Dino Road -·- ·---~- --- ·13 Massadon-Dino Hicih Density 23-MDH 33 45 Sand Ck 1----'-'----t-~~~-----~-=--~--r--------,..,,-t-----t-----+-----,----'-"~~-------' 13 Massadon-Dino Hicih Density 24-MDH 33 46 W. Dinosaur Hwv 40 L 13 ______ Massadon-Dino Hiah Density 25-MDH 33 47 Lower Serina. NW of Dinosaur I 13 Massadon-Dino Hiah Densitv 26-MDH 33 48 Uooer Sorina r·--ii i Massadon-Dino Hlah Densitv 27-MDH 33 49 Bull Canyon Rim 13 •••• I Massadon-Dino Hiah Density 28-MDH 33 50 K-Ranch Massadon-Dlno Hicih Density 29-MDH 33 51 Buckwater Draw -~ Massadon-Dino Hiah Densitv 30-MDH 33 52 K-Creek K l I_·_ l .... j _ _-_-_-I -~~~~~~:.~---D-in_o__H_~_1g:h=D=e:n:s=itv====~=;=~=-~=n=~=~=:===-3~3===~====5=5===~~=======:S:ta:t:el:in:e::K=-C:r:e:e:k~-- �.... V, °' (/) --4 ~ --4 :;; iii C ~ z C 0 s: (/) )> s: ( 17-WRH O t---------------i---t----t---t----t---t----+---t----i---11---+---1----+-:--1---1 :::~=~ :~ I 25-MDH 2 ;::~g~ ~ 20-WRH , 40 30,MDH 0 21-WRH O l1-MDH 0 22-WRH 0 23.WRH 0 24-WRH 0 -·--+----+-2"'5~.w-R~H..-+--eo-~---, .......·-----~-....... ---·-+----;----< 26-WRH 27-WRH 28,WRH 0 0 0 O I >m 0 z --4 i )> ---+---·- m ;:::=~ .~ ◄ O•WRH. ~ - 41,WRH "'O C -:~.,![,. .-==-1-t=--==_-:t+-==--==--=~-1-==--==--=ti--=___ =_-==--=~-1,_=..·=.:~=-_=_,=i+-:-~---.._-_. __ 1 - - - - - - - - - - - - - - + - - + - - - - + - - - t - - - - t - - + - - - - + - - + - - - + - - t - - ~ + - - t - - - - t l - ·--~-Jfllt++t=- ::::j::_ "'O 0 m (/) 1 - - - - - - - - - - - - - - + - - + - - - - + - - + - - - + - - - l - - - - + - - + - - - + - - f - - - +---+---~---+----,..,,~~~:~~-H,.._+ · • - - < , - - - . - - - - ----------+--t-'----t--+---t--+-----+-===::-!·-·~-1--+--- 1---,---t---~: .. m --4 t----------------+---+----+---+----+--+----+--+---+--f---+--f---+---ll---t-~5~ci::ii=:i~-i+"'--..;~c..._+--+,----1--+----I---! ..............31-WRH "'O r- , 18 --·---i·---f---+---4 I I ---+---'""'"" I I --4 C )> ~ z )> C 0 r0 )> C s 0 z (/) (/) 0 --4 0 z :I: �~~- ··:··l _!• - :=: ~WH~~E~R::~~=,,=:3a=1,:=:•=,~=.=~•=.;=!=l•=:.=~=;=~•=.,=:m~rJ=(N=,J===~~~:=~=:=t=d:~•,=~.=~=•:=:=~e-::-i-'r~"-~~;...-o,_·-:-_,q"':"'~"':"1,~;...i;...nu,;..:!"';!"'h;n"'S"-~ac"t.Juc..hm_·;-,~----~~----·_fl---!--:--·+j - ::~ ···-:··-r-------- ~~-• . · - - + ••••••••••• • J ........ t···-t··· ..-11 1-----,-----,,-----c-'--"'----lf--'------,--~..,...---,..:..=;..._;..._,;..;...-'-'-'-''-'-~-----'-----'-----'-----"---•---+----1-----+---+-----1,---+---f..... .___M_•-•~•-Oe_,•,..•_P_,er,..s,.,•_,m,cp,..le_,d,..U_nl-'-l(;...Nc-"''r"'-)---1-•-•""-"'~g_e_nu~m_b_e_,ol_d,..eer_co_u,..n_1,_d_on_aa_c_h_sa_m-'p-ltd_u_n1t_1_n_ea_c_h_St_,a_1_um--';'--•-d_e_e,_c_ou_n_,le_d_/_n_um,,_b_e_,_,,01_1,..am_,_pl_e_un_~_,_,_1_ _ _.i....l_--l'----+---+----!---"~.....,.. f... -~----+---+ -----+---+---·Deer Per Sample Unit Varlanco (S" ,.J • ,tandatd satnple variance of deer counted per sample unk ln 111ch stratum; calculaled using Excel "VAR" function for 11mpla variance. I_ Total Sample Units Per Slfatum (UJ • lolal numbar ol polonllol ,ample unit■ or quadral1 In oach Stratum I I I ·-----~ - - - , - - - ' - - - l r - - - t - - - + - - - - t - - - - t - - - I --+---+---< Estimated Otar Per Stmtum (N",J • average number or deer counted on each 1ampled unit In each Stratum mutllplled b)'potenttal number of sample units In each Stratum 1-----=-c--,-:--,-,-,=~~---1---,:--:-c---,-,c-::-,-:---,---:---,-~-,....,"'."""--,,-----,--~-,~-...,..---r-'----.---..----1---+----+-----l---+--f-f----!---+--- ----------1----s_1ra_1u_m_v._,_1a_nc_,_v_._,.~(N_'~,J---+·-=·'=-"":cd:--aTrd_eslirna1ed Stratum variance for deer counted in each Stratum I I I t----t---+---+---+--- - · - - - - t - - - - - 1 - - - + ' - - - ! - - - f - f - - - - l 1 1 ,_______________,_ •__ : ___,:,..., ___L_ : : : +--·- --t-------++-·-·-=------t:.:.:.-_:.~t-:.:.:.:.~:.:.-_:.:.-=-i+-_:.:.:.-:.-:...::.-:..-:..:.-:..~t----==~======~=====;!::-.·.:.:::~t-=--=--=--=-~~-=--=--=--=--=-~ •_1m_"_ 1---------------!-V-•r'--'('--N---'',+-l•_ _ __,lfU_,J,.,_(1_•---'U,/U,--",J'---·(;...S_••-=-.;--'u,,;,)l'••• Th.,;;;p,on el al. 1998 poge 341 • I i l---=roi---,al""Oee-r"'eou,-n""'lod--,-A'"'"'"'S,,-tt-al,-a.,.N"(-•u-m-N"J"'."""-l-.-,u-,,;-or1art deer coun:td In Ht~ Sl,atum : ••• --- ---- --- ··: : ==r:'---------l'----,,f---+---f----+--f-f---+-------+--+----1---+---f----i,e----J 1 l--=T"ot,..al'"'E'"1'"11m-a""'1ed-:-:o=-.-,-,."g'"'s,,.,ra-1,..a""'N"''c,-(o_u_m...,H"'•-;,J---,;-•. ■u-m-cl=-1-=H-11"'11,..m_a,..le""'d-•u-m"'b,-e-,"'•1,..d=-ae-,.,.,n-•-ac-':h-_s:::,-,.,..,u~m~;-::-lh-,s""'1,-.-.~llm-a-:ltd-:-p-op.u"'11"'11=-on-,""'1z~e"'ro=-,.,.u=-oc-ll•10=--sa-m-p,-le"'d·a,-,-.--+-----+--+---+----lf---+--+------+-- ···_--_-_--_---,_,If---+----, b-c~~=-c:.=.:.;:;.:.~~-"--'~~c-:-:-l....:;:"'--":~~~-:.'-"'c"'-.;:;.;c=-c.c.=:;c-:;.=+=-=r==,:.:;::==,.;.:::.;=:..,-:-:;==;:='-----l----+--l---l---l----l--+---+--1-Tolal Elllmated VaflanceVe,'N'_ (sumVar'(N',J) •sumofolleatimalodSlralumva,lan-11 I I I I 90:;:~:;:,~j~~~~~~~tlr~~~~:1per :.• :Poo'U1ation :~:~:km 95% Conrldenee lritetval for N'· ILO'W1Jr 111 mnerl I i 1----+---+----+-----+---+---a"----!---- I ·----l'----+-----l ~~:: ~~~Atrv1~~:~r•~l~~:X~'•=•re8=~•,~••=•~•~•l!f~°i;•::•l=••~••===:~====t======~!====~~=====~=====t=====j======~======t====j======ti=====t=====--l:'---+----1 N,\l +/. <1.96 ~Dol=M••~r'~N~•-1_ _ 1 ___1~_ __.1_ _ _1 ~ - - - - - - - - - - - . . , __ _~_ ____,__ _ __.__ _.__ _..,__ ___.__ _ _.._1_~1.__ ___, _. V, --.J �158 2. LETTER FROM IDAHO DEPARTMENT OF FISH & GAME SHOWING POPULATION April 5, 200 I V.W. Howard, Jr. 1025 Hickory Drive Las Cruces, NM 88005 Tommy S. Bickle P.O. Box 750 Hatch, NM 87937 Dear Drs. Howard and Bickle: Find enclosed 3 files representing the results of my analysis of the Colorado Unit IO mule deer survey data. Colorado Division of Wildlife personnel flew the survey Feb 25 - Mar 5, 2001. The survey was flown using a stratified random sample. Eleven strata were delineated and sample unit size varied from ¼ - 1 mi. 2 . The ¼ mi. 2 sampling units were used in the high canopy closure habitat types of pinyon-juniper and/or juniper woodland. The l mi. 2 sampling units were used in more open canopy types. One hundred and forty three units were sampled across the 11 strata and 1180 mule deer were observed. Because the Idaho Aerial Survey program has a ten strata maximum, 2 strata were combined for the analysis. Based only on the stratified random sample, without correction for visibility bias, the population estimate was 6481 mule deer. Using the Idaho Department of Fish and Game's Sightability Method the population estimate was 11,052 (90% CI 7549 - 14,555). This represents a sightability factor of 1.7. Bartmann et al. (1986. Wildlife Society Bulletin 14:356-363) recommended a 1.5 .sightability factor for Colorado pinyon-juniper woodland. You can find a copy of the Idaho Aerial Survey software and a downloadable manual on the web at: http://members.nbci.com/fred_leban/survey.html. Please call with any questions. Sincerely, James W. Unsworth Principal Wildlife Research Biologist Cc. G. Miller, CDOW �159 ESTIMATE USING IDAHO SIGHTABILITY CORRECTIONS Aerial Survey for Windows, Version 1.00 Beta 6.1.4 (12-Feb-2000) Thursday, April 05, 2001 09:06 AM Model: Mule Deer, Hiller 12-E, Idaho (Spring) [Files] Title = C:\PROGRAM FILES\IDFG\AERIAL SURVEY\colo2.ttl Summary= C:\PROGRAM FILES\IDFG\AERIAL SURVEY\colo2.sum Colorado Unit 10 Mule Deer Survey, Feb 28 - Mar 5, 2001 Section 1: Summary of Raw Counts Units Stratum Sampled Total ------1 2 3 4 5 6 7 8 9 10 24 5 8 41 10 5 15 4 21 10 125 31 66 322 178 112 10 9 133 194 Total 143 1180 Section 2: Summary of Raw Counts for Perfect Visibility Model This table projects the number of animals that would have been counted if every unit had been flown and visibility bad been perfect (no animals obscured by vegetation, etc.) Strat No of Units Popn Sample --- ---- ------ Total 237 32 23 217 31 32 176 80 162 29 24 5 8 41 10 5 15 4 21 10 1234 198 190 1704 552 717 117 180 1026 563 Total 1019 == 143 6481 1 2 3 4 5 6 7 8 9 10 ---- ---- - - �------------. ---- 160 Section 3: Estimates for Total Number Total Stratum Number of Units Popn. Sample Estimate Sampling Variance Sigbtability Model - - - - -- --- --1 2 3 4 5 6 7 8 9 10 237 32 23 217 31 32 176 80 162 29 24 5 8 41 10 5 15 4 21 10 Bound 90% ---1422 293 375 3183 851 1397 154 1133 1403 841 1158993 27155 25316 654838 154826 388737 21812 1219118 572711 141508 4492 2570 8143 44009 5916 14884 376 53969 10224 4590 349 218 1529 6681 538 1333 17 8787 784 449 1774 285 308 1382 661 1047 245 1862 1257 630 11052 4365014 149173 20685 3503 ----- - - ------Total -- 1019 143 �C') 0 4 . ~ 5 8 K F,,G HIIIJ UNIT 10 OefR PG>PUlATION ESIMATE •EB 28,MARcH •·· 2 WITHCOLORADOCCO)AND.IOAHOflC SIG"1an1L , ,, N M p 0 Q u R V w X fot 10 1.TuA••AftA!lr,1 , Sla"81ntumlDlorCO:st-lor.lD. -0..rActMIYlorCO voo ..v ... TYOO ror co· orlO :Aet•ourAclM ~r or en I I I TlffiPA•M1 or P■ PM when Ouackal nown fo, CO :::! z nl 3 1 ◄' ff ~~oi!ild~-0■ ~o~ldttl!iJ•~-~~•!!!ldL=l====t====r===t===1~==::t:==+===1===r==r==:t:==::t:==+==:+===i:===t===1 rd• or 11 It H• ~Hellco tr for CO; JR•JolR■na11r3, HR•Hillat12E 20 ITel Pflk SHn on Quad rat ror co: 2721 28' 3 31 32 ~ ''35 ie' m 31 )> Colo Quod 1T2YMI.' 3YMI. Idaho SubUnH , z ◄YMI. ◄ 5 I 7 7 7 6YML I a 9 10 11 12 13 12""'-' '""" 14 15 15""'-' 1 Yl,t,1 4 1 ~ 17YMII 45 IIYM\1 ,, ◄2 ◄6 ., 18YM'.1 2DYIIM 21\'M',4 22WM 22WM 2>VM,t 52 2 ◄VPIM 53 9UL ~- .£ 41 48 -SC ii .Colo -.Sta YMI. ·y- 3 5YML l!. - 13 ,. 10 11 0 ::0 21 2◄ ,, G') 18 IS 2S 2a. r m -I m C XP•~T•=OFDATA~l!LOS I I I -auad-au1drat Unit ,o ror CO· -SubUnll•Cuadral Unit 22 03' ~ "C II 17 18. -· 20,a 21 22 22 YMI. Yt,t; YMt YMI. YMI. YMI. Thi.'. "M,1 """' """' """' klaho Sttatum 1 ,, Coio Taeer o. 0 0 0 0 0 1 1 1 .1 1 51 28 . 17 1 13. 1 ,1 0 z 0 2 0 - 0I 2 2 2 \?.t,I YIIM., 2 T- ~ -- ---~ 2 2 0 """' """' ·- 1 0 Colo Idaho . Colo Idaho COio Idaho Coro·, Colo Colo Dtll Act2 , Act 0 0 Snow SnOWCov,r Vlat IVtaCIHI ··uatt.;. Tl Timi 1 1 p a 0 0 0 0 0 58 M M za s 17 13 M 0 0 Q 3 2 3 M , 3 3 M M 3 3 0 0 0 0 0 0 0 0 0 0 3 2 3 2 2 0 24 8 2 3 10 UL 3 UL UL 3 3 0 3 0 0 2 M M 2 0 M 3 SG so 0 PJ 0 60 L 0 H-··· H -· L L .,.._,o 80 80 PJ -- 0 0 0 0 50 50 7>.l --·PJ PJ so·- .. SG 0 L 0 2 2 2 2 0 0 Q p /2001 • 03/0◄/2001 03/0◄/2001 o:w 1 1 1 1 1 1 I 1 p p p p p ' 0 3 1 1 ,1 3 I p p 03/0◄ /2001 1 1 1 p p 12001 03/04/2001 1 I 0 1 1 1 1 0 0 2 03/0◄/2001 001 ,-~~12001 OJI0◄/200 ,umm•OO Colo Dbl 1 1 1 1 1 TimeQ 9 Wind L L L L 1 I 1 1 1 3 •3 3 3 3 1 3 3 3 3 3 3 3 3 3 3 3 3 A p p p p p p A A p 3 - --~ a 8 ◄ 3 8 5 3 3 J A A A A A I I 0310,412001 Colo p 3 0 Colo 3 p 1 0 0 0 3 3 0 0 2 Colo Nav 1 1 p p p 0 3 0 0 PJ ·O 0 0 0 3 0 0 3 .3 S<> SG 60 L 0 0 0 i;o !Kl 0 0 0 M 0 0 so so 0 0 I 0 2 ,_ T- L L L L 0 2 YIIM \'1,/M UL 0 .0 r,vm YM',1 0 0 3 0 0 0 0 0 0 Yl,t,1 YW,1 2 0 0 0 0 2 2 0 0 2 2 23 101ho 3 3 3 8 8 8 8 3 10 3 3 3 3 3 3 3 3 3 3 3 5 11 8 7 7 6 6 6 7 9 3 3 1 ., ..,l,_ 3 3 3 1 1 1 15 16 18 1 23 I L L L L L L L L L L L L L L L L M L M L L L L Colo Ute Colo snw Colo PIiot COio H•llo Colo Tolk B B 0 0 0 0 0 0 0 ET ET, JR JR. JR JR 0 100 100 0 JR JR 100 JR JR JR JR JR JR 0 B B 8 B " H H H B B B B H H H H H H 0 0 0 0 0 0 0 0 0 0 ET •T ET ET ET ET er er er ET er er er ET ET __ ET - _£_ 0 ET ::0 )> r en C ~ m 4 0 -< 20 0 4 0 0 JR JR JR JR JR JR JR JR JR JR JR ·JR JR JR 0 ET 0 ET H 0 ET H 0 ET B 0 ET H 0 ET H 0 ET B 0 ET J/1 ET 8 0 JR JR L H 0 ET L 0 ET JR D ET JR L 8 0 L I ~ 0 !T J/1 L .. ,,~,- H . ' _ O _____ ET. + - ~ " 0 m 0 0 0 0 0 3◄ ·- 14 0 0 I 0 0 0 ·- - so -siJ-:-_1_ .... ,...!._,_ -+-, so .. , ·-50··----··---st :J~ 1--.J½-:.-[:JLLJ:::_~-J-·. ~- F~t~·-::f_- -+--r,::- --··1. ---1r it -· .H1:'L ,13. ., .. Jt~ I ,, -+--· lL. __ L .. . .. ,1, ,L, __ -~--L ~, ~--:-:~i:: -~: --~:-·: -·--o ,.,.. ·;:·: .. --··o·----o·, 60 I-_:-}_::.. -!r:-= ,., t -1--- ,,.,,--- --t f----~ --i---i---, --~r- i·· i--- -·-·o -'t= it , it~ : l ·; -~- ! , : ,, i· -~. I, ~ 51 57 11UL 12UL. 11 12 ..! , M 0 3 0 L_____ __ L L 1, 3. _. , ,,.1..8.. .,_ ,, .•. t ,.,,.,,. 50 """sir" "'"T" ---- 5o-···· --~--- - - · • h • .. -•" 0 • ., ., "'"'ij"' , li:::1----L 1 • 0 , , __1___ 1 1 ___ I • 1 ..... ..... °' �,_. °' t-..J .. 62 .113 A 8 3UM 3 3UM 3UM 3 3 as • s89 70. 71 SUM 7UM 7UM .73. 74 7UM 7UM ·- D 4 C 4 4 4 E 3 5 4 4 2 4 O • 4 UM • .. UM. ,,.. 4 4 7 lJM •- UM UM, UM"' UM, 4 4 4 -Nn WRH 5 5 5 78 ,wnn I "' 2WRH 1 2 11 C UM UM UM UM UM UM UM W 7 7· 7 T 7B 1 ,., Te,, 1\1M i I I ◄ o 0 4 1 2 20 1 1 a 11 0 7 2 F 3 5 4 4 2 O G M M M M N o H 3 3 3 3 3 O I H .L L L L o, 0 4 1 .2 20 1 1 I 11 0 M M M. M s M M B 7 & .J· au 50 !50 oo 50 .0 K PJ PJ PJ PJ PJ o 0 3 3 3 3 2 3 3· 1 0 .L L L l L L L L .2 L 0 50 50 ou 50 !50 oo 90 .., N 1 1 1 1 1 1 1 1 1 I A A P 1 1 1 1 I 1 11. 1 1 1 1 1 1 1 1 4 4 4 1 A 4 1, Lao -,ZOO! 1 031V312G01 03.'V3/2G01' I 1 1 I 03IW/2001 1 A A A A A A 4 4 4 4 4 4 1 1 I 1 1 1 I. L L L L L o· 0 PJ PJ PJ PJ SG PJ PJ fNi 0 50 L 3 3 3 3 3 O· SG ll 3 3 1 3 2 3 3 2 0 z- 02/Zl/200.t •-~ 0310WZ001 SOZWRH 2 2 8 2 I. "° PJ 3 D3oV3l2001 ~a~,~~~i"H-i3-~~•~➔s~-➔-~o-➔-io-~--+~o~+~--~-ot·-➔---~~o_ _ _ -nn 1-1 IH 112 •· 1'1 , 5 • .6 5 6. 9 15 13 10 4 ti 2 16 13 10 4. I 2 M M M M .M.. ..S 3 3 .• 3 3 .3 2 L L L L L L !50 50 !50 50 10 _ .. 50 SG 80 SG 80 88 SG 2 2 2 2 2 2 p P P P P P P P P P A O 1 1 1I I 1 1 1 1 1 1 1 1 1 1I. 1 1 R 23 24 11 19 27 18 I I S L L L L L L I. L L L L L L L L L L L 1' H H H H H B a H II H H H H H H D B B B U 0 O O O O O a O O O O V E ET ET ET ET !T er l!T ET ET ET O E O O O O O ET ET ET eT " l!T ET O O W JR JR JR JR JR JR JR JR JR JR JR JR JR . JR JR JR JR JR X 0 50 0 0 0 0 0 o ETJR 1 ,. • , L " o ET JR 20 I A 4 1. 5 L B. 0 l:T JR 5 +--;:-+--:.,....--,,--~<--+--~:-+---a:--+-----~~-+--..,:~+----~--+--~ETE"T~+--:,..__,_..,~_ __, .,,"'..-1.:.·-=ii-.,.H;;-,1--•.--t-'·r.WRHiif.iH---.sc-.---,1-~lr'-t-T" "-;-----,-+--;",-+---,-+--"'o.,..._+----1--or_ ru -nH 56 IRH 5 o., 0 0 .o; . 0 ~::i-t·r-ia~'-i:R;.;~-I-~ : , : : M : L. : 80 : ~~&6~=~~~~=r==7i==t= ~a7cl--=7W~R;.;H+_7i,--f,-, gg 7WRH 7 WRl:I It 7WRH I 7 -WAH. 90 7WRH I ·, ii 7WRH I ·7 WRH 1 1 1 1 1 1 1 I 0 A A " A A A I B 8 8 B B B O 0 0 O O O ET ET ET ET ET ET JR JR JR JR JR JR �A 123 •=RH 12 ◄ 20WRH Lill RH Lill 8 20 20 20 20 ~~ ffl 20 20 U 2, 23 24. C. -NM WRH.. WRH WRl:I :: • WRlf. -NM D 5 5 6 5• e : 5. 5 5, S 5 5 • F 5 8 5 5 3 5 3 : : ~ 9. 9. M 1 0 O 0 0 1 0 O 0 0 0 ' 0 0 • 0 0 0 0 28 7. 3 4 0 0 G S M M M M 13◄ , 21·. 13 3 139 14 1◄1 1 ◄ 2 ·/32=N 14' ...- N I ◄◄ 3◄WRH W :3◄WRH. 141., 1 ◄7 : H 148 :,:J◄WIUl. · 149"3SWRH.· ""' •38WRH 15 1 21: 20.· 30 31 ,32 33 34 WR.If' • WRl:I :. WRtl WRH WRH. WM·. 3◄ WRH WRH' 34 .>A 3◄ :,S. .. 3S • WRH · -NM WRH 5 5 5 5 5 5 5 5 • 5: 5 5 5 .. • ◄ 5 0 jl: !: ; ~•- ! ; l ll!l 1 1 40 ◄O ~ ◄0 ◄0 157 '-'2 ◄ IWRH .40 40 41 41 41 183 ◄ 1WRH. ◄1 ~15 1il 41WRH 81 ◄ 1WRt! · ~ _◄ tWRH 41 !Wc.JWRM+---· 1 166 2WRM 2 WRH WRH WRH WRH WRH WRH WRH. WRH,. WRH WRH WRH WRM WRM 5 5· 5 6 5 5 ~ S 5 .s 5 6 6 8 .16 1 3 5 5 7 3 1 4 1 0 I ◄ 4. 5· 0 ~ 5 16 1 3 6 5 7 3 1 4 1 D 1 J 50 50 .50 .. 50 K Sc, SG. S<> SG ; 3 3 L L L :~ 50 60 0 O 0 0. :~ PJ SG o 0 0 : Q 0 0 0. 0 28 7• 3 I L L L. L o: I:1 = - :. : : ~1 WRH. H' H R H 2 3 3 3 M M M M' s· u M M' M .S •. S M M. M M M M M M 0.. 0 O 0 0. o .0 3 .. 3 3 3 2 3 o·. ~ 3 3 ' 2 3· 3 3 3· 3 3 3 O 3 : L L L L L PJ S<J S<J SG. 60' PJ L l 50 50 50 60 50 50 50 50 50 50 60 O 60 1 so: pJ pJ: Pl· PJ. SO SO PJ SG SO SO 6G SO PJ O A A A A P ◄ ◄ : : A, ◄ ◄ I I I 1 I 1 A. P Q 1 I R 5 1 S L L 1 L ~ -~ 1 1 1 I I 1 L L T L L L H H· 8 8 8 B o 0 O F t : g O :; E O O· 0. o ET ET ET !T El ET ET.. ◄ ◄ I ◄, P ◄ P P. ◄ - P P P P. P P P A A A ,_ A P. A ◄ ◄ -f -,5 12 I : ; : : 4 1 1 0. O 0 0 0 0 3 2 2 2• 3 2· 0 1 1 1 1 1 1 1 1 1 I 1 J : ~ 3 3 3 2 2 3 2 2 • 2 2 O 3 N 1 1 I 1 ::1 : o· 0 L L. L L' L L .. L L L L L l ~ • 2 0 O 0 0 ·M • : .0 0 O 0 O., D 0 50 50. 80· oa so. L 2 2 • • 2 1 03/0212001 03I02J:1001 03I02/2001 02128/2001 0212812001 1 1 1 I I 1 1 1 1 1 1• A, A A A A A A A A A A A A ◄ 7 8 7 4 A L T L L L L T L L L L L L T H ti ti H V O O O O O B B B B 8 8 B H H H H. H H H 4 1 1 1 1 1 1 1 1 1 1• 1 1 1 ~ -j : t : ◄ 4 ◄• ◄ ◄ ◄ 4 ◄ 4· 4 ◄ 4 ◄ ◄ 4 4 ◄ 4 4 ◄ 4 4 2 2 I I I I 1 1 1 1 1 1· 1 2 2 8 g 7 8 6 9 7 20 37 L L L L L L L T L L L ,_ L.. M M I B· B B B B B B B 8 8 B H O ~ 0. 0 0 O O O O 0 O O O g 0 O O O O O O O O O O Q. O V !T .r 1IT· ET W JR, JR JR JR X 0 ---~~"-',T'-_,_-"'JJ:.,,,__.__ __, l!T ET l!T ET !T JR JR. JR JR JR JR. 0 5 D ~ 0 0 0 0 * ~~· 0 0 0 0 0 0 !l. l!T • ET • . !T ~T. ET ET ET ET JR, JR JR JR JR JR. JR JR JR JR JR JR JR JR. JR JR JR JR . JR er JR ET JR ET JR ET JR ~y JR __ll_ ., ...,Jc,;R_l----1 JR HR 0 JR HR 200 ..... O'I w �,_. 0\ .j>, m m ABC flWRM 6 WRM IIWRM • 18C 191 S 8 OE 6 5 F 5 6 ~ ~ 6 8 6 • 8 8 8 6 8 • 3 2 3 11 5 6 6 G M M HI 3 L 3 H J 50 SO K.. SQ SQ L 2 2 M M M M M M M M 3 • .3 3 3 3 3 3 50 SG sa SG SG SQ SG SQ SQ 2 2 2 2 2 2 2 2 MN 02/291200\ 1 0212812001 1 OP P 2 P 2 QR 2 2 STU L H O L H O V JR JR W HR HR P P P P P P P P 2 2 2 2 2 2 2 2 L L L L L l L L JR JR JR JR JR ~R JR JR HR HR HR HR HR HR HR HR X --·-8WRM IWRM .m IWRM 193 SWRM 19'1 ewRM 8 I WRM WRM WRM v, RM ·~ ,., 1• 8 6 W ••• •~ •~~ -·~- S • ., ~ 3 2 3 11 5 5 L .H L L L L L H •a 50 .OIi 60 50 60 eo .02/28/2001 -02/28/2001 "" 812001 ' ·02/21112001 02/28/2001 02/21112001 02/21112001 02/24/2001 1 1 1 1 1 I 1 2 2 2 2 2 2 2 2 H H H H H H H H O O O O O O O O �246 20 _,H ◄ 7MOH 8MOH 24! 2"' 251 252 2•3 254 255 2.. 257 258 255 280 281 282 1283 284 265 255 28 1288 289 21 2 1 272 21 274 215 • 4 • -·- I5MllH 15MOH 2118 28 28 289 292 293 ~ _,H MOH J,f)H MOH MOH MOH MOH MOH \Thf',4 I .,.,_. • 5 5 6 ~- 5 5 5 7 8 9 10 10 !~: t- T!,N '""' T!,N TMM TMM T!,N. '""' TMM T!,N •= ! I I ! TMM T!,N TMM WM TMM TMM T!,N T!,N TMM ·- ' T!,N TMM ' I TMM TMM TMM T~ 199 300 • 301 302 Total• 303 Subllct 304 30$ 313 F G 3 3S 2 0 0 0 0 3 s 0 1• 1 0 0 3a 2 3 0 0 2 0 0 0 0 4 3 5 & 2 M F G H u 0 0 3a M 7 M 3 0 0 2 0 0 0. 0 M • " " 16 10 3 2 15 10 3 2. 4 6 5 18 23 19 18 23 . 19 8 0 0 32 I 1 3 : .. 9- : M M M M M M ! I ; 1 7 s a • ; 2 0 0 0 0 0 0 2 3 3 0 0 3 3 3 0 0 3 0 0 0 0 3 3 3 2 2 3 3 3 3 3 2 2 2 3 3 3 2 0 0 3 3 3 3 2 0 0 0 0 2 3 a 6 : s s M• .,M ◄ ' 2 0 0 D 0 0 0 14 3 5 7 2 2 I 35 1 0 0 0 0 6 2 H 2 2 a 0 0 32 1 3 9 1 0 0 0 !I s. s s 5 M M. M s M M M M s 0 s 1 L L L l l L J 60 50 50 0 0 0 0 0 0 K SG SG SG 50 80 SG L 60 0 0 50 1. 50 L 80 0 0 50 0 0 0 0 80 50 L H l l l L L H L l l. l l 'H l L L L l L L L L so so PJ PJ PJ PJ PJ PJ PJ 50 50 50 ,SG .60. PJ PJ PJ so so . 60 so. so 50 60 50 80 50 60 so so .PJ PJ SG so SG· 50 0 0 $0 so l L I J' 0"IW2001 03/04/2001 001 0 03/04/2001 - SG SG SG SG PJ .. SG PJ -~- 3 3 3 n 0 3 0 0 0 0 3 3 3 2 2 2 3 1 .3 2 2 2 3. A A 03"0◄/2001 1 p 03/04/2001 03/002001 03/0112001 1· p I 1 A A I A 1 1 1 1 1 1 1 1 1 1 A Ii 1 f/2001 03/0112001 =112001 001 =1 001 1 1 1 1 1 1 1 1 ·,- 03/0 1 03/0 /2001 03/0112001 03/01/2001 03/01/2001 03/0112001 03/111/2001 03/0112001 0 30I12M 1 1 1 1 1 1 1 1 L M 0 0 0 2 A A A A A A A 0"M'"OOI 03/04/2001 03"0◄/2001 1 B 274 Entrlu ,. m Entrlu 274 Entries e 171 1180· 1180 171 Dter OeerGros DeorGms Deer w/Acl w/Act Counted Counted 123.' 121 1711. 179 oroun1 Grouas Grouns Ot0UDI Counltd coun1td Movlna,. 'Movlna II'• II H. II Qu1driti Qu1dr1ts GroUna orou•• w/0dNr Sl1ndlna Stind= w/0 dair 1 1. orou• ·oroun Bedded Beddtd K 179 178 179 179 x,, PHIGmt DeerOms wlSnow DetrGR>I DHrG"'S entrl•s WN""e wN.,,, m. 14'.. 117, Grouo1w/ Low Snow 33 Oroun1W/ HtnhSnow Grouo•wl GroUDI -'"*now lnSa•e 11 ·:. Groiina : In PJ. 1 Grow In 118 Oroufts nS1•tGW w/Snow 33,.. , Grouo1w/ Hl•hlnow Ort■ seWd 11 -a,ou•s 4 4 4 4 4 4 A A p p p p p p p Q I 1 1 1 1 1 1 1 1 1 1 1 4 A A •=001 I p • • A 1 1 1 1 1 1 1 1 03I0◄/2001 2 2 0 0 2 2 2 2 3 0 ·o I I I 03/04/2001 03/04/2001 n•~-1 03l04l2001 •2 N 1 1 1 1 1 1 I 1 l 1 I 2 2 2 0 0 50 50 50 50 60 0 0 0 0 60 50 - M L 2 2 2 0 D 0 0 0 0 4 4 •4• ◄ R 9 a 8 8 5 4 8 1 s 1 1 a ◄ I 5 4 1 5 ◄ I ◄ 4 4 1 1 1 7 1· 4 4 1 1 1 1 1 1 1 1 1 1 I 1 • 1 1 ·1 A A A 1 1 1 1 1 1 1 1 1 1 t , I 1 1 1 1 ·,1 1 1.· 1 1 • p A A A A • A A A A A p p p p 1 I I 1 1 1 ! 1 1 __ 1 1 p p p ' 1 1 1 1 N 0 p 274 Ent~es 151· AMcnts 123 PMcnt, 82 Blcklt p 1 1 1 1 1 1 1 - 11 p e •8 8 1 20 11 15 13 1◄ I I , p p I I ·A ' A 3~ 301. JOI. 309 310 311 312 E 0 ,. D ·'""' T!,N ; 13 13 13 13 13 13 13 13 13 13 13 13 13 13 0 MOH 1Thl',4 1 2Thl',4 2 2Tt.!-I_ ~ 2TMM 2 JT!,N 3 13 Tt.t.l MUH I 0 13 13 13 13 13 13 13 13 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 12 30 1 1 1 1 1 1 7TM\1 295 6Tt.t,4 298 9TMM 207 10T"'4 298 I0T'-'M MOH l,llH MOH I ITMM I -:~ MOH MOH 31 I 1 1 1 1 l'IM',1 1Th'M t'!M',1 1Th'M 1TMM JTMM STh\',! -- 15 18 19 20 20 21 22 24 25 28 29 15MOH lllM:lH 20M)H 20MOH 21MOH 2,-H 24MOH 2SM:>H 25MOH 2&MOH 30MOH 3IMOH 1Th'M lTl,t,I 1Th'M 11""1 1TMM 1TMM '285 MOH MOH MOH MOH 1• 15MOH 277 271 279 280 211 282 283 • 284 MOH 7 8 9 1D 11 13 15 10MOH 11MOH '7! C MOH MOH B A 4MOH 4MOH 245 15 9 14 18 13 1 ...... s L L L L L L L L L L L L L L L L L L L L L l L l L L l l L l .L L L L L L L L. L L l L L L L L L L l l L l L l T B B B 8 8 B e B 8 B B e B 8 B 5. 8. B. 8 B B 8 s· B 8 8 8 B B B u 0 0 0 V !T !T ET 0 E 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 o· 8 8 8 B 8 8 8 8 8 8 0 0 0 0 0 0 0 0 0 0 0 .. 0 0 0 0 0 8 Q R s T u 117 1'3 218 151 . au dtVtral• Qu•drot, llltVlfnd B~ltLII• OldSnow er ET E er E ET ET er ET ET ET ET ET ET ET ET ET eT ET ET er ET ET ET ET ET ET ET ET er ET ET ET ET ET ET ET :~-- w 218 0 0 5 0 0 0 0 0 0 0 0 0 0 0 0 0 30 0 ___l!L - ~ V EdTr■cv 0 a JR ET er )( JR JR JR JR JR JR JR JR JR JR JR JR JR JR JR JR . JR JR JR JR JR JR JR JR JR JR JR JR Jf\ ET ET ET ET !T ET ET ET ET 68" I 68 Flown& I II m. Tlmt ModWlnd HazyLII• P'r1hSnw JlmRlch Howud oraham 0 21 21 StraWtnd OullLlle .ar1h1m Ellenbrar 138 Freddv ' w JR JR JR JR JR JR JR JR JR JR JR JR JR JR JR JR JR JR JR JR JR JR JR -·-w0 ·40 0 0 0 15 X 1297 21B JttRnar ElkCnld H 10 Hlll6r cu1drat1 32 Quadrats with tlk 1 In PJ .. ...... °' Vl �166 4. MAP SHOWING WHERE DEER WERE COUNTED Sampled Quadrats Showing Where Deer Were Observed and Counted During Helicopter Counts of Mule Deer to Estimate .Population Size in Unit 10, DAU D;;.6, Colorado; Sampling Strata anil Sample Quadrant Units -Yampa-Monmneot (Low) ■ Yampa O Monlllllenl (Medium) Utai Wltite River (Low) Utah White River (Medium) -UpperWhiteltivet (High) Upper White River (Medium) UpperWhiteRmr(Low) Massadoiui ~ Dino (Medium) -Massadona- Dirio (Low) • ■ Twelv0111il~ (Medruui) • Mimaoonii:- Dirio (High) CJ Deer Obsavcd and Counted A - NoDeerObsenred �167 5. MAP SHOWING WHERE ELK WERE COUNTED Sampled Quadrats Showing Where Elk Were Observed and Counted During Helicopter Counts of Mule Deer to Estimate Population Size in Unit 10, DAU D-6, Colorado. Sampling Stra!a.and Sample Quadrant Units -Yampa-Monument (Low) -Yampa-Monwnent (Medium) Utah White.River (Low) Utim White River (Medium) Upper White River (High) . -UpperWhiteRiver (Medium) Upper White River.(Low) Massaoona - Dino (Mediuin) Massaoona - Dino (Low) ■ Twelvemilc (Medium) Massaoona • DiDo (High) E:=3::==::E========i:::::::::::::;==:::ie IIMes A r7 Elk Observed. - Ek Not Observed �..... en C: s:: s:: )> DEER POPULATION AND DENSITY ESTIMATES DAU 0-1 0-2 0-3 0-6 0-7 0-8 0-9 D-11 0-43 0-41 D-42 0-18 0-12 D-13 0-14 0-15 Sportsman's Population Estimate 4,100 13,300 2.700 1,750 17,500 8,000 7,800 5,100 1,750 3,700 2,300 Modeled Population Estimate• 13,500 37,800 2,700 7,900 7,300 8,300 2,200 1,200 31,500 10,000 4,900 2,650 1,750 5,900 0-53 1,100 TOTALS 128,420 408,600 0-20 2,200 1,850 0-21 0-23 1,600 0-19 6,100 0-39 0-40 0-25 0-24 0-29 0-52 1,100 0-30 0-35 0-37 900 3,800 3,600 9,200 1,970 2,400 3,300 1,750 750 24.i"98 1,800 7,000 67,000 6,000 13,500 12,000 7,700 13,800 2,400 6,700 5,000 3,900 2,400 31,000 3,000 11,800 5,300 34,000 12,500 9,300 20,800 7,600 2,000 3,200 D-38 0-51 where applicable Quadrat Population Estimate J4,857 11,016 34,619 ~ of Deer Classified 1997 1998 1999 401 3,058 306 623 3,559 2,049 3,835 524 1,963 497 3,839 464 7,076 558 407 5,361 2,099 NIA NIA 844 982 730 1,432 NIA NIA NIA 805 1,355 2,972 926 522 839 5,134 868 586 NIA NIA 2,091 1,071 740 643 1190 1122 473 615 2014 353 425 1005 170 387 187 488 539 156 128 301 18.35 5.07 10.29 9.56 12.30 5.53 15.40 14.10 9.38 8.80 3.57 9.48 14.92 5.71 2.90 4.66 4.78 t4.29 8.31 7.03 :::0 Modeled Densitylrni2 Quadrat Densitymi2 WRIS winter range WRIS winter range 11.34 21.57 33.69 3.81 11.38 17.36 33.27 17,00 33.48 31.76 11.94 45,29 35,66 42.25 14.96 58.44 64.23 64.10 38.28 19.60 4.90 28,88 N/A NIA NIA 395 716 917 352 481 490 232 124 NIA NIA NIA NIA 280 310 1,065 1,344 1,860 365 4,342 784 1,107 2,745 509 76.7 1308 NIA 7.94 50.40 4.80 2.93 3.56 2.11 11.34 18.60 18.47 15.48 5.63 32.99 7.40 23.54 998 NIA NIA 1,190 927 NIA 20,018 NIA 912 1,195 1,055 25,128 NIA 446 5,811 mi2WRIS wl nter range Sportsman's Densitylmi2 WRIS winter range 3.45 11.85 5.71 2.85 8.69 22.66 230 NIA 345 394 266 433 1,309 248 500 1,126 491 355 97 40,568 35,346 36,645 17,352 1,068 4,242 870 1,151 2,992 1,363 226 NIA NIA 1,608 2,684 824 1,202 3,719 773 167 40.32 13.93 7.74 23.70 13.04 44.36 12.24 25.97 -< :E ~ rn en -I "tJ mO ;:o :::0 z cri (") s:: Om •z 0 )> ;;o z ~c on - C ~~ me :S:m co m m ;;o ;;o "tJ NO 0 0 19.21 "tJ C: 0~ -I 0 15.29 z m en =! s:: )> -I m en 'Tl 0 :::0 • LJOW 2000 Post-hunt Population Projections ml.:d?er pop_den ail.xis -11109/00 0\ 00 � https://cpw.cvlcollections.org/files/original/ee5a4eecbb041230ff1ea11948214a12.pdf b51986c61cb4fd9596994dcabd5d57f4 PDF Text Text Colorado Division of Wildlife July 2004 – June 2005 WILDLIFE RESEARCH REPORT State of Cost Center Work Package Task No. Colorado 3430 3002 2 Federal Aid Project: N/A : Division of Wildlife : Mammals Research : Elk Conservation : Evaluation of GnRH Vaccine as a Long-term Contraceptive Agent in Female Elk: Effects on Reproduction and Behavior : Period Covered: July 1, 2004 – June 30, 2005 Author: D. L. Baker and J. G. Powers Personnel: J. Powers, M. Wild (National Park Service), L. Miller, J. Rhyan (National Wildlife Research Center), M. Conner (Utah State University), T. Nett (Colorado State University). All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without the permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We conducted a pilot experiment to evaluate the potential of GnRH vaccine as a long-term contraceptive agent in female elk. The objectives of this preliminary investigation were to characterize the antibody response of captive female elk to GnRH vaccine, evaluate the effectiveness of dart delivery of the agent, and document the presence and severity of systemic reactions (if any) to the treatment. Intramuscular injection of GnRH vaccine was accomplished in 4 female elk. Serum antibody responses were collected each month beginning in February, 2005 and submitted for analysis. Ultrasound imaging of the injection site was conducted in conjunction with monthly blood collections. Analysis of antibody levels have not been completed, however initial results from ultrasound imaging of vaccine injection sites reveal changes in muscle fiber and muscle tissue echogenicity compared to pre-treatment conditions. All animals show some level of disruption of normal muscle fiber patterns and changes in the quality of muscle tissue. These changes began to appear approximately 2 weeks post-treatment, peaked in severity in April then diminished during July, 2005. Based on results of this trial and similar ongoing investigations with captive white-tailed deer (Odocoileus virginianus), we prepared a detailed study plan describing research to evaluate GnRH vaccine as a long-term contraceptive agent in female elk (Appendix I). The objectives of this experiment are to evaluate the effects of this fertility control agent on pregnancy rates, reproductive behavior, and neonatal health and survival. We performed a power analysis to determine the sample sizes needed to detect treatment differences for pregnancy rates and reproductive behavior for captive female elk maintained at the Colorado’s Foothills Wildlife Research Facility in Fort Collins, Colorado. Based on this analysis, a sample size of 18-26 elk (equally divided between control and treatment groups) should provide adequate statistical power to detect treatment differences in pregnancy rates and reproductive behavior. A detailed description of hypotheses, rationale, methods, and statistical analyses are provided in this report. The status of publications in process is also provided (Appendix II). 77 �WILDLIFE RESEARCH REPORT EVALUATION OF GnRH VACCINE AS A LONG-TERM CONTRACEPTIVE AGENT IN FEMALE ELK: EFFECTS ON REPRODUCTION AND BEHAVIOR DAN L. BAKER P. N. OBJECTIVE Evaluate the effects of GnRH vaccine on pregnancy rates, fetal and neonatal growth and development, and reproductive behaviors in captive female elk. SEGMENT OBJECTIVES 1. Conduct a pilot experiment to evaluate individual animal variation in antibody response to GnRH vaccine and assess any side-effects of treatment. 2. Using results from the pilot experiment prepare a study plan program narrative and submit for internal peer review and extramural funding. 3. Summarize and analyze data from previous fertility control experiments and submit manuscripts to appropriate scientific journals. INTRODUCTION Hunting and culling have traditionally been used to regulate ungulate numbers but there are a growing number of situations where these methods are not feasible. Such places include urban and suburban areas where lethal removal is often opposed because of safety concerns or on ethical grounds (Decker and Connelly 1989, McAninch 1993, Wright 1993, McCullough et al. 1997). In addition, there are many conservation areas, and state and national parks where hunting may be inconsistent with other goals of resource management or where it is proscribed by law and policy (Leopold et al.1963, Frost et al.1997, Porter and Underwood 1999). In these situations, fertility control offers a potential alternative for limiting the growth of ungulate populations (Kirkpatrick and Turner 1985, Bomford 1990, Garrott et al. 1993). Additionally, development of fertility control technology may provide resource managers benefits beyond its value as a tool for balancing ungulates and their forage resources. Fertility control may reduce the rate of disease transmission in ungulates by regulating local host densities and pathogen shedding (Rhyan and Drew 2002, Miller et al. 2004). Simulation modeling suggests that, in some situations, fertility control can be as effective as culling in reducing endemic disease or the density of susceptible hosts (Hone 1992, Barlow 1996). Extensive research has been devoted to developing anti-fertility agents that are safe, effective, reversible and economical (Fagerstone et al. 2002) and models have been developed to represent effects of fertility control on population dynamics of wild ungulates (Garrott and Siniff 1992, Seagle and Close 1996, Hobbs et al. 2000). To date, however, only modest successes have been achieved and a practical and acceptable method of controlling reproduction in free-ranging wildlife populations has not yet been attained. In previous research, we administered gonadotropin-releasing hormone (GnRH) agonist (leuprolide acetate) in a biodegradable implant to captive and free-ranging female elk and achieved 100% contraception for one breeding season, without significant behavioral or physiological side-effects (Baker et al. 2002,2004). However, despite the demonstrated efficacy and safety of this approach over existing technology, practical application is compromised by the need for annual treatments in fall, prior to the breeding season, a time when capture efficiency is low compared to winter and early spring. 78 �GnRH Vaccine An alternative approach involves immunization against GnRH. GnRH is a small, 10 amino acid, neuropeptide with an obligatory role in reproduction. It is naturally secreted in a pulsatile pattern from neurons in the hypothalamus and specifically directs gonadotropes in the anterior pituitary gland to synthesize and release luteinizing hormone (LH) and follicle stimulating hormone (FSH). These latter two hormones, in turn, control proper functioning of ovaries in females and testes in males (Hazum and Conn 1998). To successfully immunize an animal against GnRH, it is necessary to make this endogenous protein appear foreign to the host. Therefore, many copies of the peptide are coupled to the highly immunogenic carrier molecule keyhole limpet hemocyanin (KLH). When combined with a potent adjuvant the GnRH-KLH conjugate stimulates the host’s immune system to produce antibodies against GnRH as well as KLH. Anti-GnRH antibodies bind to GnRH in the hypothalamic -pituitary portal vessels and prevent the hormone from attaching to receptors on the gonadotropes. This suppresses secretion of LH and FSH, halting the hormonal cascade that is ultimately responsible for folliculogenesis and ovulation. This condition persists as long as there are sufficient antibodies to bind to all circulating GnRH. The use of GnRH vaccine as a fertility control agent is not new. It has been administered to a variety of domestic ungulates including horses (Rabb et al. 1990), cattle (Adams and Adams 1986), swine (Meloen et al. 1994), and sheep (Brown et al. 1994). It’s use as a contraceptive agent in wild ungulates has been limited, however by the need for multiple initial treatments, annual boosters, and the use of the controversial FCA and FIA to enhance the immune response of the vaccine (Miller et al. 2000b, Curtis et al. 2002). Recently, the impracticality of this approach for wildlife applications has been largely overcome by the development of a new adjuvant by scientists at the National Wildlife Research Center (NWRC) in Fort Collins, Colorado, USA. The alternative adjuvant is thought to be safer and, equally as effective in eliciting an antibody response, as FCA or FIA. The new adjuvant (AdjuVacTM) is derived from a USDAapproved Johne’s disease vaccine (MycoparTM) which has previously been approved for use in food animals by USDA/APHIS(http://www.aphis.usda.gov/ws/nwrc/research/gnrh.html) . A single application of GnRH-KLH and AdjuVacTM) may prove to be a safe, practical, and effective multi-year immunocontraceptive for wild ungulates. This approach has several potential advantages over other methods of contraception. These include: 1) a single treatment may provide long-term (2 + years) of infertility when administered to pregnant animals during winter 2) effectiveness of treatment may be > 90% during the first breeding season following immunization 3) infertility should be reversible 4) the agent should not cause significant behavioral or physiological side-effects 5) the agent should be safe for pregnant animals and the developing fetus 6) the proteinaceous nature of the GnRH-KLH immunogen should eliminate the possibility of passage through the food chain 7) the small volume required for effective contraception should facilitate administration by syringe dart 8) the agent is currently being evaluated for FDA approval as a New Animal Drug and therefore may be available for commercial use in the near future. Preliminary investigations evaluating GnRH-KLH vaccine in captive wild horses (Killian et al. 79 �2005, in preparation), bison (Miller et al. 2004) and white-tailed deer (Miller et al. 2004, unpublished data) are promising and USDA/APHIS is seeking FDA registration of the new vaccine and adjuvant (GonaCon/AdjuVacTM). However, many unanswered questions must be addressed before this potential contraceptive can be considered an effective and acceptable method of population control in free-ranging elk. Research is needed to evaluate the effectiveness and duration of this approach in elk, the effects on elk reproductive physiology, the effect on elk social structure of removing individuals from the breeding population, and the practicality/feasibility of application in wild populations. Captive Elk Experiments Rationale: The efficacy of GnRH-KLH vaccine depends on sufficient stimulation of the immune system and subsequent production of antibodies against this reproductive hormone. Thus, an initial step in assessing the potential of a single application of GnRH-KLH vaccine as a contraceptive agent in elk is to evaluate antibody response to treatment. Such studies have been conducted in wild horses (Killian et al. unpublished data), bison (Miller et al. 2004, in press), and white-tailed deer (Miller et al. unpublished data) but not in elk. Results of these studies indicate that the immunological response to GnRH-KLH vaccine is not uniform across species and highly variable within species. As a consequence, a species specific experiment is required to measure peak antibody response in female elk, time to peak response, and duration of response. Although such titers may not provide a quantitative measure of infertility, their characterization is of interest because sustained elevation of anti-GnRH antibody titers has been consistently associated with infertility in other species. Thus, the primary purpose here was to provide preliminary information on antibody response in elk, to determine optimum sample sizes for future experiments, to assess gross and clinicopathalogical side-effects of treatment (if any), and to evaluate remote delivery of the vaccine. Objectives: We conducted a controlled pilot experiment with captive elk to: 1) characterize serum antibody response of captive female elk to GnRH-KLH vaccine. 2) evaluate the effectiveness of dart delivery of GnRH-KLH vaccine. 3) evaluate presence and severity of systemic reactions or abscesses (if any) to the GnRH-KLH/AdjuVacTM vaccine treatment 4) determine if vaccination with GnRH/AdjuVacTM causes seroconversion to Johne’s disease mycobacteria. This experiment was conducted at the Colorado Division of Wildlife’s Foothills Wildlife Research Facility (FWRF) in Fort Collins, Colorado, USA with the approval of the Colorado Division of Wildlife Animal Care and Use Committee (# 1-2005) and in compliance with U.S. Federal Animal Welfare Act Regulations). METHODS We conducted an experiments with 4, non-pregnant, multiparous adult female elk beginning 7 February 2005. These elk were closely monitored into July 2005 to meet initial objectives of the pilot experiment, but the health of these elk and responses to the vaccine will be monitored until 1 August 2007. The captive elk used in this experiment were permanently maintained at FWRF and were trained to repeated handling, weighing, and blood sampling procedures. On the day before treatment (7 February), elk were moved from holding pastures (5 ha) and placed in individual isolation pens. The next day, each elk received a single injection of 1000µ :g of GnRH-KLH conjugate (0.5 ml aqueous solution) emulsified in 1.0 ml of AdjuVacTM, as a water in oil emulsion. The conjugate was be transferred into single use, 1 ml, 13-mm-diameter, barbless darts equipped with gel-collared 32-mm-long needles (Pneu-dart, Williamsport, Pennsylvania, USA). 80 �Prior to darting, individual elk were placed in a handling chute and lightly sedated with xylazine hydrochloride (15-20 mg/animal, IV). This dose allowed the animal to remain standing in the chute and minimized excitation associated with discharge of the dart gun. We examined of the reproductive tract of each elk using rectal palpation and ultrasonographic techniques, collected blood samples (20-30 ml) and measured body weight (∀ 0.5 kg). Elk were remotely injected in the biceps femoris muscle with a dart fired from a CO2-powered pistol (DanInjectTM, Wildlife Pharmaceuticals, Fort Collins, Colorado, USA) from a distance of approximately 3 meters. In order to accurately determine the precise dose of GnRHKLH delivered to each elk, darts were weighed before and after injection. If a dart failed to discharge or only partially injected the prescribed dose, additional darts were fired until the complete dose was delivered to the animal. Once the vaccine had been administered, sedation was reversed with yohimbine (30 mg, IV) (Antagonil®, Wildlife Laboratories, Fort Collins, Colorado, USA) and elk were returned to holding pastures. One of the elk (F86) used in this experiment was previously used in a Brucella abortus Strain 19 vaccination study (1998). It may still retain antibodies or immune modulation relative to this organism that could influence its immune response to the AdjuVacTM portion of the GnRH vaccine. This elk has not shown any evidence of being affected by Johne’s disease (Mycobacterium avium partuberculosis). However, the AdjuVacTM adjuvant uses small amounts of a remarkably similar killed bacterium (derived from the Johne’s vaccine MycoparTM). This could cause seroconversion indistinguishable from Johne’s disease. Pre-vaccination serum was submitted for a large animal biochemistry profile, Johne’s disease ELISA, and Strain 19 brucellosis vaccination serology. Elk were monitored for local injection site reactions (swelling, erythema, drainage) on a daily basis for 1 week, and on a biweekly basis for the following 2 months. A second biochemistry profile was submitted if elk showed symptoms of local or systemic inflammation. Ultrasound examination of the injection site may was used to evaluate abscess and granuloma formation. Serum anti-GnRH antibody production was monitored on a bimonthly basis until peak anti-body titers were determined, then on a bimonthly basis thereafter until termination of the experiment. Once a measurable (P < 0.05) decrease in anti-body levels is observed, the need to continue monitoring antiGNRH antibodies will be reevaluated. Once peak response in each elk has been achieved, a second reproductive examination will be performed to evaluate any changes in ovarian structures. Analysis: This was a descriptive experiment and no hypotheses were tested. We used descriptive statistics to examine changes in antibody titers over time. Schedule: Date Activity 12 January 2005 Submit study plan for ACUC approval 7 February 2005 Move experimental elk to individual isolation pens 8 February 2005 Perform pre-treatment exams and administer GnRH-KLH conjugate to elk 8 February to 8 March 2005 Intensive health monitoring of elk February 2005 to August 2006 Ongoing health and anti-GnRH antibody monitoring, and compile and analyze data pertinent to Expt. 2 81 �RESULTS AND DISCUSSION Intramuscular injection of GnRH vaccine was accomplished in 4 female elk. Serum antibody responses of experimental elk were collected each month beginning in February, 2005 and submitted for analysis. Initial results from ultrasound imaging of vaccine injection sites reveal changes in muscle fiber and muscle tissue echogenicity compared to pre-treatment conditions. All animals show some level of disruption of normal muscle fiber patterns and changes in the quality of muscle tissue. These changes began to appear approximately 2 weeks post-treatment and have not been resolved to date. SUMMARY Results of the pilot experiment are incomplete at this time. Initial results suggest that GnRH vaccine can be delivered via intramuscular dart injection. However, until laboratory results are completed, it is unknown if the antibody response of elk to GnRH vaccine will be sufficiently high to suppress fertility. Regardless, injection site reaction to the vaccine is a concern and warrants further evaluation. LITERATURE CITED ADAMS, T. H., AND B. M. ADAMS. 1990. Reproductive function and feedlot performance of beef heifers actively immunized against GnRH. Journal of Animal Science 68:2793-2802. BAKER, D. L., M. A. WILD, M. M. CONNER, H. B. RAVIVARAPU, R. L. DUNN, AND T. M. NETT. 2002. Effects of GnRH agonist (leuprolide) on reproduction and behavior in female wapiti (Cervus elaphus nelsoni). Reproduction Supplement 60:155-167. ____________, _________, M. M. CONNER, H. B. RAVIVARAPU, R. L. DUNN, AND T. M. NETT. 2004. Gonadotropin-releasing hormone agonist: a new approach to reversible contraception in female deer. Journal of Wildlife Diseases 40:713-724. BARLOW, N. D. 1996.The ecology of wildlife disease control: simple models revisited. Journal of Applied Ecology 33:303-314. BOMFORD, M. 1990. A role for fertility control in wildlife management. Department of Primary Industries and Energy, Bureau of Rural Resources Bulletin No. 7, Australian Government Publishing Service, Canberra, Australia. BROWN, D. B., T. T. CAI, AND A. DASGUPTA. 2001.Interval estimation for a binomial proportion. 2001. Statistical Science 16:101-133. CURTIS, P. D., R. L. POOLER, M. E. RICHMOND, L. A. MILLER, G. F. MATTFELD, AND F. W. QUIMBY. 2002. Comparative effects of GnRH and porcine zona pellucida (PZP) immunocontraceptive vaccines for controlling reproduction in white-tailed deer (Odocoileus virginianus). Reproduction Supplement 60:131-141. DECKER, D., AND A. N. CONNELLY.1989. Deer in suburbia-pleasure or pets. Conservationist 43:46-49. FAGERSTONE, K. A., M. A. COFFEY, P. D. CURTIS, R. A. DOLBEER, G. J. KILLIAN, L. A. MILLER, AND L. WILMOT. 2002. Wildlife fertility control. Wildlife Society Technical Review 02-2. The Wildlife Society, Bethesda, Maryland, USA. FROST, H. C., G. L. STORN, M. J. BATCHELLER, AND M. J. LOVALLO. 1997. White-tailed deer management at Gettysburg National Military Park and Eisenhower National Historic Site. Wildlife Society Bulletin 25:462-469. GARROTT, R. A., AND D. B. SINIFF. 1992. Limitations of male-oriented contraception for controlling feral horse populations. Journal of Wildlife Management 56:456-464. ______________, P. J. WHITE, AND C. A. VANDERBIL WHITE. 1993. Overabundance: an issue for conservation biologist? Conservation Biology 7:946-949. HAZUM, E., AND P. M. CONN. 1998. Molecular mechanism of gonadotropin releasing hormone (GnRH) action. I. The GnRH receptor. Endocrine Review 9: 379-386. 82 �HOBBS, N. T., D. C. BOWDEN, AND D. L. BAKER. 2000. Effects of fertility control on populations of ungulates: general stage-structured models. Journal of Wildlife Management 64: 473-491. HONE, J. 1992. Rate of increase and fertility control. Journal of Applied Ecology 29:695-698. KIRKPATRICK, J. F., AND J. W. TURNER, JR. 1985. Chemical fertility control and wildlife management. Bioscience 35: 485-491. LEOPOLD, A. S., S. A. CAIN, C. M. COTTAM, AND I. GABRIELSON. 1963. Wildlife management in the national parks. Transactions of the North American Wildlife and Natural Resources Conference 28:28-45. MCCULLOUGH, D. R., K. W. JENNINGS, N. B. GATES, B. G. ELLIOT, AND J. E. DIDONATO. 1997. Overabundant deer populations in California. Wildlife Society Bulletin 25: 478-483. MCANINCH, J. B., editor. 1993. Urban deer: a manageable resource? Proceedings of the 1993 Symposium of the North Central Section, The Wildlife Society, St. Louis, Missouri, USA. MELOEN, R. H., J. A. TURKSTRA, H. LANKHOF, W. C. PUIJK, W. M. M. SCHAAPER, G. DIJKSTRA, C. J. G. WENSING, AND R. B. OONK.1994. Efficient immunocastration of male piglets by immunoneutralization of GnRH using a new GnRH-like peptide. Vaccine 12:741-746. MILLER, L. A., J. C. RHYAN, AND M. DREW. 2004. Contraception of bison by GnRH vaccine: a possible means of decreasing transmission of brucellosis in bison. Journal of Wildlife Diseases 40:725730. __________ , _________, AND ___________ . 2000. Immunocontraception of white-tailed deer with GnRH vaccine. American Journal of Reproductive Immunology 44:266-274. PORTER, W. F., AND B. UNDERWOOD. 1999. Of elephants and blind men: deer management in the U.S. national parks. Ecological Applications 9:3-9. RABB, M. H., D. L. THOMPSON, JR., B. E. BARRY, D. R. COLBORN, K. E. HEHNKE, AND F. GARZA, JR. 1990. Effects of active immunization against GnRH on LH, FSH, and prolactin storage, secretion, and response to their secretagogues in pony geldings. Journal of Animal Science 68:3322-3329. RHYAN, J. C., AND M. D. DREW. 2002. Contraception: A possible means of decreasing transmission of brucellosis in bison. In: Brucellosis in elk and bison in the Greater Yellowstone Area, T.J. Kreeger (ed.). Greater Yellowstone Interagency Brucellosis Committee, Wyoming Game and Fish Department, Cheyenne, Wyoming, 99-108. SEAGLE, S. W., AND J. D. CLOSE.1996. Modeling white-tailed deer population control by contraception. Biological Conservation 76:87-91. WRIGHT, R. G. 1993. Wildlife management in parks and suburbs: alternatives to sport hunting. Renewable Resources Journal 11: 18-22. Prepared by: _____________________ Dan L. Baker, Wildlife Researcher 83 �APPENDIX I PROGRAM NARRATIVE STUDY PLAN FOR MAMMALS RESEARCH FY 2004 – FY 2007 State of: Cost Center: Work Package: Task No.: Colorado 3430 3002 2 Federal Aid Project No.: N/A : Division of Wildlife : Mammals Research : Elk Conservation : Evaluation of GnRH Vaccine as a LongTerm Contraceptive Agent in Female Elk: Effects on Reproduction and Behavior : EVALUATION OF GnRH VACCINE AS A LONG-TERM CONTRACEPTIVE AGENT IN FEMALE ELK: EFFECTS ON REPRODUCTION AND BEHAVIOR Principal Investigator Dan L. Baker, Wildlife Researcher, Mammals Research Cooperators Lowell A. Miller, USDA/APHIS, National Wildlife Research Center Jack C. Rhyan, USDA/APHIS, National Wildlife Research Center Mary M. Conner, Department of Forestry, Range, and Wildlife Science, Utah State University Terry M. Nett, Department of Biomedical Science, Colorado State University Jenny G. Powers, National Park Service Margaret A. Wild, National Park Service STUDY PLAN APPROVAL Prepared by: ____________________________ _ Date: ______________________ Submitted by: _____________________________ Date: ______________________ Reviewed by: _____________________________ Date: _______________________ _____________________________ Date: _______________________ _____________________________ Date: _______________________ Reviewed by: _____________________________ Biometrician Date:________________________ Approved by: _____________________________ Mammals Research Leader Date:_________________________ 84 �PROGRAM NARRATIVE STUDY PLAN FOR MAMMALS RESEARCH State of : Cost Center: Work Package: Study No.: Colorado 3430 3002 : : : : Division of Wildlife Mammals Research Program Elk Conservation Evaluation of GnRH Vaccine as a LongTerm Contraceptive Agent in Female Elk: Effect on Reproduction and Behavior A. STUDY TITLE: Evaluation of GnRH Vaccine as a Long-term Contraceptive Agent in Female Elk: Effects on Reproduction and Behavior B. NEED: Overabundant wild ungulate populations have become a significant problem for natural resource managers in many areas of North America. Unregulated populations can cause adverse effects that are ecological, economic, or political in scope and resolving these problems often requires managing excessive animal numbers (Jewell and Holt 1981, Garrott et al. 1993). Hunting and culling have traditionally been used to regulate ungulate numbers but there are a growing number of situations where these methods are not feasible. Such places include urban and suburban areas where lethal removal is often opposed because of safety concerns or on ethical grounds (Decker and Connelly 1989, McAninch 1993, Wright 1993, McCullough et al. 1997). In addition, there are many conservation areas, and state and national parks where hunting may be inconsistent with other goals of resource management or where it is proscribed by law and policy (Leopold et al.1963, Frost et al.1997, Porter and Underwood 1999). In these situations, fertility control offers a potential alternative for limiting the growth of ungulate populations (Kirkpatrick and Turner 1985, Bomford 1990, Garrott et al. 1993). Additionally, development of fertility control technology may provide resource managers benefits beyond its value as a tool for balancing ungulates and their forage resources. Fertility control may reduce the rate of disease transmission in ungulates by regulating local host densities and pathogen shedding (Rhyan and Drew 2002, Miller et al. 2004). Simulation modeling suggests that, in some situations, fertility control can be as effective as culling in reducing endemic disease or the density of susceptible hosts (Hone 1992, Barlow 1996). Extensive research has been devoted to developing anti-fertility agents that are safe, effective, reversible and economical (Fagerstone et al. 2002) and models have been developed to represent effects of fertility control on population dynamics of wild ungulates (Garrott and Siniff 1992, Seagle and Close 1996, Hobbs et al. 2000). To date, however, only modest successes have been achieved and a practical and acceptable method of controlling reproduction in free-ranging wildlife populations has not yet been attained. GnRH Agonist Gonadotropin-releasing hormone (GnRH) is an endogenous neuropeptide that has an obligatory role in reproduction. It is naturally secreted in a pulsatile pattern from neurons in the hypothalamus and specifically directs gonadotropes in the anterior pituitary gland to synthesize and release luteinizing hormone (LH) and follicle-stimulating hormone (FSH). These latter two hormones, in turn, control proper functioning of the ovaries in females and testes in males (Hazum and Conn 1988). 85 �The chemical structure of endogenous GnRH has been determined (Matsuo et al. 1971) and alterations in the molecule have led to the synthesis of potent GnRH agonist analogs (Karten and Rivier 1986). Long-term treatment with GnRH agonists has been shown to prevent ovulation by decreasing GnRH receptors on gonadotropes, receptor sensitivity to GnRH (Nett et al. 1981), pituitary LH content (Aspden et al. 1996), and by suppressing pulsatile secretion of LH and FSH (D’Occhio et al. 1996). Agonists of GnRH have been used in domestic animals as fertility agents for controlling ovarian activity, gonadal steroidogeneis, and reproduction (McNeilly and Fraser 1987, Montovan et al. 1990, D’Occhio et al. 2002). In previous research, the GnRH agonist, leuprolide, was administered to captive female elk (Cervus elaphus) and mule deer (Odocoileus hemionus) in a controlled release bioimplant and achieved 100% infertility for one breeding season, without significant behavioral or physiological sideeffects (Baker et al. 2002, 2003, 2004). However, despite the demonstrated efficacy and safety of this approach over existing technology, practical application is compromised by the need for annual treatments in fall, prior to the breeding season, a time when capture efficiency is low compared to winter and early spring. Immunocontraception To date, most wildlife contraceptive efforts have been directed toward development of a safe and effective immunocontraceptive vaccine. The immunocontraceptive target antigen that has received the most research and management attention is porcine zona pellucida (PZP). Porcine zona pellucida has been administered experimentally to more than 70 species of wild mammals (Kirkpatrick et al. 1997). This approach relies on host antibodies formed against PZP to block sperm receptor sites on the ovum, thereby preventing fertilization and pregnancy (Dunbar and Schwoebel 1988). The PZP vaccine has been shown to be 85-90% effective in most ungulates, can be administered by syringe dart, is reversible, does not interfere with ongoing pregnancies, and most importantly, the immunogen is proteinaceous and therefore, is not likely to pose a threat to the environment or to non-target species, including humans (Kirkpatrick et al. 1990, Turner et al. 1992, Miller et al. 2000a, Kirkpatrick and Turner 2002, Shideler et al. 2002, Naugle et al. 2002). However, despite these desirable characteristics treatment inefficiency and undesirable sideeffects have limited management application of PZP vaccine (Rudolph et al. 2000, Turner and Kirpatrick 2002, Naugle et al. 2002). Specifically, practical application is compromised by the requirement that the target animal must receive two initial injections within 1-2 months of each other (Walter et al. 2002). Second, with the exception of SpayVacTM which encapsulates PZP within a cholesterol/phospholipid formulation (Fraker et al. 2002), effective duration is typically < 1 year; consequently, annual booster inoculations are required (Kirkpatrick et al. 1996; Turner et al. 1996). Third, while no long-term health effects have been reported for animals treated with PZP (Kirkpatrick et al. 1995, Miller et al. 2000a, Turner and Kirkpatrick 2002), extended estrous cycling and associated breeding behavior have been reported for white-tailed deer (Turner et al. 1992, 1996; McShea et al. 1997), horses (Plotka et al. 1989), elk (Heilmann et al. 1997), and fallow deer (Fraker et al. 2002). By prolonging the breeding season in males and females, PZP vaccine treatments could result in late pregnancies, parturition beyond the normal early summer period, and unpredictable and abnormal behavioral consequences. Finally, for an effective immune response, the PZP antigen must be administered with an adjuvant - a substance that enhances the specific immune response to the antigen. At present, the most effective adjuvants used with PZP are Freund’s complete (FCA) and Freund’s incomplete adjuvant (FIA). In some species, however, this combination has been shown to cause severe systemic reactions, chronic pain, and abscesses at the injection site (Anderson and Alexander 1983, Stills and Bailey 1991, Leenaars et al. 1996) and, as a consequence, it is unlikely that the Food and Drug Administration (FDA) will grant approval for the use of PZP vaccine containing these adjuvants. Thus, in the near future, practical application of immunocontraception for wildlife species will depend on development and use of improved vaccines with different adjuvants. 86 �GnRH Vaccine An alternative to PZP immunocontraception involves immunization against GnRH. To accomplish this, it’s necessary to make this endogenous protein appear foreign to the host. Therefore, many copies of the peptide are typically coupled to a highly immunogenic carrier molecule such as keyhole limpet hemocyanin (KLH) (Levy et al. 2004). When combined with a potent adjuvant, the GnRH-KLH conjugate stimulates the host’s immune system to produce antibodies against GnRH as well as KLH. Anti-GnRH antibodies bind to endogenous GnRH in the hypothalamic - pituitary portal vessels and prevent the hormone from attaching to receptors on the gonadotropes. This mechanism suppresses secretion of LH and FSH and interrupts the normal cascade of hormonal events that are ultimately responsible for folliculogenesis and ovulation. The use of GnRH vaccine as a fertility control agent is not new. It has been administered to a variety of domestic ungulates including horses (Rabb et al. 1990, Turkstra et al. 2005), cattle (Adams and Adams 1990), swine (Meloen et al. 1994), and sheep (Brown et al. 1994). However, its use as a contraceptive agent in wild ungulates has been limited by the need for multiple initial treatments, annual boosters, and the use of the controversial FCA and FIA to enhance the immune response of the vaccine (Miller et al. 2000b, Curtis et al. 2002). Recently, however, the impracticality of this approach for wildlife applications has been largely overcome by the development of a new adjuvant by scientists at the National Wildlife Research Center (NWRC) in Fort Collins, Colorado, USA. This adjuvant is thought to be safer than, and equally as effective, as FCA and FIA in eliciting an antibody response. The new adjuvant (AdjuVacTM) is derived from a United States Department of Agriculture (USDA)-approved Johne’s disease vaccine (MycoparTM) which has previously been approved for use in food animals by the USDA, Animal and Plant Health Inspection Service (APHIS) (http://www.aphis.usda.gov/ws/nwrc/research/gnrh.html). A single application of GnRH-KLH and AdjuVacTM (GonaCon™) has the potential to be a safe, practical, and effective multi-year immunocontraceptive for wild ungulates. As a contraceptive for wildlife, this agent offers the following desirable characteristics: 1) A single treatment should provide long-term infertility (2 + years) when administered either to non-pregnant females prior to the breeding season or to pregnant females during gestation. 2) Treatment effectiveness should be 85-90% the first breeding season. 3) Infertility should be reversible. 4) The agent should be safe for pregnant animals and the developing fetus 5) The agent should not cause significant behavioral or physiological side-effects. 6) The proteinaceous nature of the GnRH-KLH immunogen should eliminate the possibility of passage through the food chain. 7) The small volume required for effective contraception should facilitate administration by syringe dart. 8) The agent is currently being evaluated for FDA approval as a New Animal Drug and therefore, could be available for commercial use in deer and elk. Preliminary investigations evaluating a single application of GnRH-KLH vaccine (GonaCon™) in captive female wild horses (Equus caballus) (Killian et al. 2004), bison (Bison bison) (Miller et al. 2004), white-tailed deer (Odocoileus virginianis) (Miller et al. unpublished data), California ground squirrels (Spermophilus beecheyi) (Nash et al. 2004), New Zealand white rabbits (Oryctolagus cunniculi) (Powers et al. in preparation) and domestic male cats (Felis catus) (Levy et al. 2004) are promising. All female bison (n = 5) treated with a single injection containing 1800µg GnRH-KLH and AdjuVac™ have remained infertile for 3 breeding seasons (Miller et al. 2004, Rhyan and Miller unpublished data). Similarly, mares (n = 18) treated with either 1800µg or 2800µg GnRH-KLH vaccine have been shown to be 100% infertile after one breeding season (Killian et al. 2004). While these results are encouraging, 87 �additional species specific studies are needed to confirm the safety and effectiveness of this contraceptive approach in wildlife. Our goal in this investigation is to conduct controlled experiments with captive elk to investigate the important attributes of this technology prior to management application in free-ranging wild ungulates. C. OBJECTIVES 1. To evaluate the effective duration of a single dose application of GnRH-KLH vaccine in preventing subsequent reproduction in pregnant elk. 2. To evaluate the effect of GnRH-KLH immunization on serum concentrations of LH and progesterone, corpus luteum (CL) function and viability, and neonatal health and survival. 3. To evaluate the effect of the GnRH-KLH vaccine on breeding behavior of captive elk following a contraceptive treatment applied during the second trimester of pregnancy. 4. To evaluate the physiological side-effects (if any) of GnRH-KLH vaccination on female elk, the developing fetus, and/or neonate. D. EXPECTED RESULTS OR BENEFITS The Colorado Division of Wildlife’s Strategic Plan (2002-2007), charges the agency with “finding alternatives for game management when hunting is not a viable option” (H-1.5, p 9). One of the performance measures for accomplishing this objective is to develop alternative methods of population control. Successful development and testing of the fertility control technology described in this proposal has the potential to accomplish this objective and provide resource managers with a non-lethal strategy for controlling the growth of some wild ungulate populations when sport hunting is infeasible. E. APPROACH Proposed Research: Working Hypothesis.: In this investigation, we test the general hypothesis that a single intramuscular application of a novel anti-GnRH vaccine in mid-gestation female elk will prevent pregnancy the following breeding season and may prevent pregnancy for two or more subsequent seasons. The exact duration of infertility is unknown but will be determined in this investigation. However, permanent sterility is not anticipated and we expect treated females to eventually return to normal estrous behavior and fertility as antibody titers decline. Furthermore, we don’t expect immunization against GnRH-KLH to cause substantial negative physiological or behavioral effects in peri-parturient females or neonatal calves. However, since GnRH-KLH vaccine is expected to suppress reproductive hormones, we predict diminished breeding behaviors in treated female elk compared to controls. Design: We will test the effects of GnRH-KLH vaccine treatments on pregnancy rates in elk using a Fisher’s exact test and evaluate serology of reproductive hormones, anti-GnRH antibody titers, and breeding behavior using a one-way ANOVA for a completely randomized design with repeated measures structure. Animals: Approximately, 20 adult female elk (2-12 years of age), 2 mature, and 2 sub-adult male elk will be used in this study. Elk are permanently maintained at the Colorado Division of Wildlife’s Foothills Wildlife Research Facility (FWRF) in Fort Collins, Colorado. The female elk used in these experiments have been previously trained to repeated handling, weighing, ultrasound, and blood sampling procedures. When not involved in periodic intensive sampling procedures required by this study, elk are maintained in fenced paddocks (5 ha) containing native vegetation and fed a diet consisting of ad libitum quantities of grass-alfalfa hay, grain supplement, trace mineral block, and water. 88 �Experiment 1: Effects of GnRH-KLH vaccine on pregnancy rates (objective 1) Hypothesis: One year post-vaccination, female elk vaccinated intramuscularly with 1000µg GnRH-KLH + adjuvant (AdjuVacTM) during the second trimester of gestation will have significantly (P < 0.05) higher anti-GnRH antibody concentrations and lower pregnancy rates than females treated with adjuvant alone. Rationale: Vaccination with GnRH-KLH + AdjuVacTM has successfully stimulated sufficient anti-GnRH antibody production to prevent pregnancy in a wide range of species including many ungulate species (Curtis et al. 2002, Miller et al. 2004, and Killian et al. 2004). Of particular importance to our experiment is the ongoing study with white-tailed deer (Miller, personal communication). Preliminary results suggest that multiple year infertility has been achieved with a single treatment of GnRH-KLH vaccine. Since elk and deer are taxonomically similar and share many common ecological, morphological and physiological traits, we expect to observe a similar contraceptive response for both species. Methods: Many of the measurements in this experiment (i.e. conception/parturition dates, pregnancy rates, luteal function, and hormone concentrations) will be facilitated by controlling the breeding period of female elk. To do this, we will attempt to synchronize estrous cycles of female elk by using progesterone secreting controlled internal drug release (CIDR) implants (Fennessy et al. 1990, Asher et al. 1993, Lucy et al. 2001). The CIDR implants will be placed in female elk during the last week of August 2005 (see appendix A for detailed protocol). Following CIDR removal (approximately the first week of September 2005), reproductively sound male elk will be released into the same pasture as females (see appendix B for breeding soundness exam protocol). During January 2006, we will determine pregnancy status of all females. Once pregnancy status is determined, pregnant elk will be blocked according to age and body condition, and randomly assigned to either a control or treatment group. We will determine pregnancy rates of treatment and control elk each year thereafter until differences in treatment effects can no longer be detected (P > 0.05). Treatment and control formulations will be applied in the following the manner. On the day of application (approximately mid-January, 2006), animals will be moved from paddocks, weighed (± 0.5 kg), and lightly sedated with xylazine hydrochloride (Rompun; Bayer AG, Leverkusen, Germany; 45-55 mg/animal, IM). This dose should allow animals to remain standing in the handling chute and minimize any possible stress or pain associated with blood collection, reproductive tract examination, ultrasound imaging of injection site, and dart delivery of treatments. All elk will be remotely injected in the area of the biceps femoris muscle with 1 ml, 13-mm-diameter, barbless darts equipped with gel-collared 32-mmlong needles (Pneu-dart, Williamsport, Pennsylvania, USA) fired from a CO2 – powered pistol (DanInjectTM, Fort Collins, Colorado, USA). Darts will be fired from approximately 3 m and will contain either GnRH-KLH vaccine + AdjuVacTM) (treatment) or AdjuVacTM alone (control). In order to accurately determine the precise dose delivered to each elk, darts will be weighed (0.001g) before and after injection. Once all elk have been treated, sedation will be reversed with yohimbine (30 mg, IV) (Antagonil®, Wildlife Pharmaceuticals, Fort Collins, Colorado, USA) and animals will be returned to holding pastures. Antibody titers will be measured immediately prior to treatment application and then on a monthly or bimonthly basis until maximal levels are reached. Following peak response, these measurements will be made on a less frequent basis until just prior to subsequent breeding seasons (September 2006, 2007, 2008). At that time, females will be sampled again. Except for this period, monthly sampling will be terminated following October 2006. 89 �The effective duration of GnRH-KLH vaccine in controlling fertility in elk will be determined by comparing pregnancy rates of treated and control elk during January 2007 and 2008. Once pregnancy rates are determined, pregnant elk will be aborted using a combination of prostaglandin F2 α and dexamethasone (Bates et al. 1982) (see appendix C for detailed protocol). Blood sampling procedures for antibody determination, pregnancy rates, hormone concentrations, and serum chemistry and hematology will follow methods previously described. While elk are sedated, blood samples (20-40 ml) will be collected via jugular venipuncture. Serum will be stored at – 70 ºC until analyzed for LH, progesterone, and anti-GnRH antibodies. Following the last blood collection, sedation will be reversed and elk returned to paddocks. Measurements: Anti-GnRH antibodies will be measured using an enzyme linked immunosorbent assay (ELISA) developed by scientists at the NWRC (USDA/APHIS) and/or using radioimmunoassay (RIA) techniques at Colorado State University’s Animal Reproduction and Biotechnology Laboratory (ARBL). The effect of GnRH-KLH vaccine on reproduction will determined in January 2007 and 2008 by measuring pregnancy rates using the presence or absence of pregnancy specific protein B (PSPB) (Huang et al. 2000), rectal palpation (Greer and Hawkins 1967, Hein et al. 1991) and/or ultrasound (Curran et al. 1986). Analysis: To determine the sample sizes needed to detect treatment differences for pregnancy rates, we performed a power-based sample size determination for a one-sided Fisher’s exact test using a software program (NCSS Pass 2000) (Kang and Kim 2004, Krishnamoorthy and Thompson 2002). For this analysis, we used the highest reported pregnancy rate (approximately 30%) for GnRH-KLH vaccine treated white-tailed deer, 1 year post-vaccination (Curtis et al. 2002). To represent the best and worst case scenarios for control elk, we calculated the sample size requirements for a 90% and 100% pregnancy rate. Based on this analysis, between 18−26 female elk (equally divided between control and treatment groups) should provide an adequate sample size to detect expected differences in pregnancy rates (Table 1). Because the pregnancy rates of the control and treatment groups are expected to be close to 1.0 and 0, respectively, the normal approximation invoked for testing the difference between 2 proportions is not valid (Brown et al. 2001). 90 �Table 1. Estimated sample sizes required to detect differences in pregnancy rates, in control and treatment groups, based on a Fisher’s exact test power analysis (NCSS Pass 2000). Power Control Group Pregnancy Rate Treatment Group Pregnancy Rate Total Sample Sizea 0.90 0.90 0.90 0.90 0.90 0.90 0.90 0.90 0.80 0.80 0.80 0.80 0.80 0.80 0.80 0.80 1.0 1.0 1.0 1.0 0.9 0.9 0.9 0.9 1.0 1.0 1.0 1.0 0.9 0.9 0.9 0.9 0.0 0.1 0.2 0.3 0.0 0.1 0.2 0.3 0.0 0.1 0.2 0.3 0.0 0.1 0.2 0.3 8 12 14 18 12 14 20 26 8 10 12 16 10 12 16 22 a Total sample size assumes an equal number for each group, e.g. 18 means 9 treatment and 9 control female elk. Experiment 2: Effects of GnRH-KLH vaccine on luteal function and neonatesurvival (objective 2) Hypothesis: We are uncertain of the effects of GnRH-KLH vaccine treatments on LH secretion, luteal viability and fetal/neonatal survival. Little conclusive research has been conducted on these relationships in wild or domestic ungulates. Limited evidence suggests that GnRH-KLH-induced suppression of LH and progesterone levels in late gestation are not low enough to terminate pregnancy or negatively affect fetal/neonatal survival. Rationale: Corpora lutea (CL) secrete progesterone and are an essential ovarian structure for maintenance of pregenancy in all mammals (Baird 1992). Progesterone is obligatory for early embryonic development and peaks in the blood of pregnant females at different stages of gestation for different species. While progesterone is always produced by the CL in early pregnancy, its role in maintenance of pregnancy varies among species. In some species (i.e. mare, cow, ewe, and women) the CL is not needed for the entire gestational period because the feto-placental unit begins producing sufficient progestins to maintain pregnancy (Squires 1993, Stevenson 1997, Stellflug et al. 1997) In other species (i.e. sow, rabbit, whitetailed deer), surgical removal of the CL will terminate pregnancy regardless of when it occurs during gestation (Plotka et al. 1982, Tomas 1997, Tast et. al 2000). It is well-documented that progesterone secretion is regulated by several hormones, including LH, which plays a principal role in CL function during both the estrus cycle and pregnancy (Niswender et al. 1976, 1994, Rahe et al. 1980, Farin et al. 1990, Okuda et al. 1999). In contrast, however, studies in cattle (Peters et al. 1994), pigs (Buhr 1987), dogs (Onclin et al. 2000), and to some extent sheep (McNeilly and Fraser 1987) provide evidence that LH may not be essential for all aspects of luteal function, including 91 �pregnancy. For wild ungulates, LH suppression due to high doses of the GnRH agonist, leuprolide, were not sufficiently luteolytic to terminate pregnancy when administered to elk during the first 60 days of gestation (Baker et al. 2001). Likewise, bison vaccinated with GnRH-KLH during the second and third trimesters of pregnancy, maintained a viable fetus throughout gestation and delivered healthy calves at parturition (Miller et al. 2004). In this experiment, elk will be vaccinated with GnRH-KLH at approximately 120 days gestation and should develop sufficient antibody titers to suppress LH by 180-200 days of gestation (Miller et al. 2000b).Because the average gestation period in elk is 255 days (Haigh and Hudson 1993), the animals in this experiment will be in the third trimester of pregnancy before the CL is significantly affected by lack of LH. If the elk respond similarly to cattle and bison they will likely retain the pregnancy despite expected declines in progesterone. Alternatively, if elk are highly sensitive to small changes in progesterone concentrations they may abort the fetus. Methods: One day prior to GnRH-KLH vaccine treatments in January, 2006, we will collect blood for antibody titers, LH and progesterone concentrations (Niswender et al. 1969, Niswender 1973), and perform reproductive examinations on all experimental elk. Beginning approximately 4 weeks posttreatment, and in conjunction with scheduled monthly measurements of antibody titers, we will monitor changes in these parameters until 15 April, 2006. Following parturition (approximately June 1-15), we will monitor neonatal health, survival, and growth to 30 days post-parturition. Weaned calves, not needed as replacement animals in other experiments or at other captive research facilities, will be humanely euthanized. At present, we have received a proposal from scientist at USDA/APHIS National Wildlife Research Center to use surplus elk calves in a terminal experiment to develop and test orally active vaccines for managing infectious diseases such as bovine tuberculosis and brucellosis. Analysis: We will use descriptive statistical methods to analyze hormonal data. Hormone concentrations, fetal and CL measurements will be reported as arithmetic means ± ŜE . We will estimate the correlation coefficient between antibody titers, hormone concentrations, CL measurements, and test whether these relationships are significantly different from zero (Zar 1984). We will compare the differences in growth rates (g/da) of calves born to treatment and control females from birth to 30 days of age using a two sample t-test. Experiment 3: Effects of GnRH-KLH vaccine on breeding behavior (objective 3) Hypothesis: The effectiveness of GnRH-KLH vaccine as a contraceptive agent is dependent on the suppression of ovulation and steroidogeneis. Because GnRH-KLH vaccine is expected to suppress estradiol and therefore sexual receptivity during estrus, we predict that 1) rates of male precopulatory, female precopulatory, and copulatory behavior will be lower for treated females compared to untreated controls, and 2) that rates of general breeding behavior (i.e. herding, establishing and/or defending a harem) will be similar for both treated and untreated females. Rationale: In previous research (Baker et al. 2002), we reported that breeding behavior rates of female elk treated with GnRH agonist were not different from those of untreated elk. We attributed this response to basal estradiol concentrations inducing reproductive receptivity in animals that had been exposed to progesterone earlier in the breeding season or during a “silent estrus” (Harder and Moorhead 1980, Asher 1985). However, in the present experiment, GnRH-KLH vaccine should suppress progesterone secretion, 92 �estrogen, folliculogenesis and ovulation well in advance of the onset of the September 2006 breeding season. Therefore, there should be no progesterone “priming” effect and no estrous behavior in treated females. Limited observations of male elk engaged in general breeding behaviors related to establishing and/or defending a harem suggest that they don’t discriminate between cycling and non-cycling females (Baker et al. 2002). If true, general breeding behavior rates of females treated with GnRH-KLH vaccine in the present experiment should not be different from untreated females. Methods: We will test these hypotheses by examining the effects of GnRH-KLH vaccine on reproductive behaviors of female elk during the breeding season (15 September to 31 October 2006). Our experimental unit for analysis will be individual females within each treatment group. On 15 August 2006, two male elk will be placed with treated and control female elk in the same paddocks. All females will be individually identified with color numeric-coded neck collars. Observers will not know which elk are treatments or controls. Behavior observations will be made from a distance of 50-250 m from an elevated tower using binoculars and spotting scope during day, and a spotlight and night vision scope at night. Selected behaviors (Geist 1982) will be recorded using a lap-top computer with a behavioral software program. All-animal all occurrences sampling procedures will be used to sample reproductive behaviors of all experimental animals over a 24 h period (Leaner 1996). Time-of-day sampling periods will be assigned each week using a randomized block design. Each sampling period will consist of at least 2 h of continuous observations. We will group individual sexual behaviors into four general categories (Table 2).Because behavioral interactions are generally short duration (< 30 sec) relative to the sampling interval, we will record the number of occurrences of each behavior rather than the length of time, and calculate rates of sexual interactions as behaviors per animal per hour, then multiply hourly behavior rates by 24 for a daily rate. Table 2. Elk reproductive behavior and associated behavior categories (Baker et al. 2002). Behavior Category General Breeding Male Precopulatory Female Precopulatory Copulatory Reproductive Behavior Male directed behavior related to establishing, maintaining and defending a group or harem of female elk (i.e. Herding, guarding, tending) Male courtship behavior directed toward an individual female to induce or detect estrus or ovulation (i.e. urine testing, flehmen, tongue flick, smell or rub) Female courtship behavior directed toward dominant male to arouse copulatory behavior (i.e. lick and rub male, mount, lordosis, twitch hocks) Male behavior directed toward a receptive female in estrus (i.e. intromission) Analysis: Based on the sample sizes required to detect differences in pregnancy rates (Table 1), we conducted a simulation to estimate the power to detect differences in behavior rates. To complete this simulation, we bootstrapped data from a previous study which examined the effects of an alternate fertility control agent, leuprolide, on female elk reproductive behavior during the breeding season (Baker et al. 2002). Each sample was run through Proc Mixed (SAS 1996) using repeated measures mixed effects structure. The following parameters were used to estimate power based on total experimental sample sizes of 18, 20, or 26 female elk (Table 3). 1. Male pre-copulatory behavior rate was previously shown to be higher than other reproductive behavior rates (Baker et al. 2002). Consequently, this measurement will likely be the most sensitive to detection of treatment effects. Therefore, we used the previously reported male precopulatory behavior rates to estimate power for our simulation. 2. The peak of breeding season is approximately one month in length. Therefore, we estimated power using 60 total observation periods. [4wks x 3 observation periods/da x 5 da/wk = 60 obs. periods] 93 �3. We estimated power for 3 different sample sizes using 60 observation periods and bootstrapping data from the previous leuprolide experiment. Ten control elk were randomly selected (with replacement) from the 5 control female elk of the previous leuprolide experiment (thus some elk were used multiple times in a sample). The behavioral response (male precopulatory behavior rate) for each elk was recorded; thus a complete sample consisted of 10 behavior rates for each of the 60 observation periods. Behavior data from control elk was considered the benchmark for comparison. Hence, to estimate power for treatment elk in the current experiment, we followed the same procedure except that the response was multiplied by the effect size. To represent a 50% decrease in the male precopulatory rate directed toward treatment elk, we multiplied the control response by 0.5. 4. Although our behavioral hypothesis predicts that treated female elk will have reduced reproductive behaviors compared to controls, we also estimated sample size for the possibility of increased behavior rates. Thus, we estimated sample sizes for a 75% and 50% reduction in male precopulatory behavior rates toward treated female elk compared to controls, as well as a 50% increase in male precopulatory behavior rates. 5. Power results are based on the number of times an effect was detected during 100 simulations. Because the variance is larger for higher behavior rates, there is less power to detect a 50% increase compared to a 50% decrease. From this analysis we conclude that a sample size of 20; 10 treatment and 10 control animals, and 60 observation periods would provide adequate power (>90%) to detect most of the differences in reduced reproductive behavior rates as well as reasonable power (>75%) to detect a 50% increase in behavior rates between treated and untreated elk. Table 3. Power results for detecting differences in male precopulatory behavior rates directed toward treated and untreated female elk based on 100 simulations and 60 observation periods. Difference Between Treatment and Controlsa 0.25 0.25 0.25 0.50 0.50 0.50 1.50 1.50 1.50 Total Sample Sizeb 18 20 26 18 20 26 18 20 26 a Power α=0.05 1.00 1.00 1.00 0.99 1.00 1.00 0.70 0.76 0.92 Power α=0.10 1.00 1.00 1.00 0.99 1.00 1.00 0.81 0.84 0.95 Effect size Total sample size assumes an equal number for each group, e.g. 18 means 9 treatment and 9 control female elk. b 94 �Experiment 4: Effects of GnRH-KLH vaccine on maternal behavior, neonatal survival and growth, blood chemistry and hematology (objective 4) Hypothesis: GnRH-KLH vaccine treatments will not result in significant secondary negative behavioral or physiological side-effects in female elk. Rationale: To date, the GnRH-KLH vaccine formulated with AdjuVacTM has produced few reported behavioral or physiological side-effects in any species in which it has been tested (Levy et al. 2004, Miller et al. 2004, Killian et al. 2004). However, it’s not clear from these studies how extensively the side-effects of this agent have been evaluated. In this investigation, we will evaluate the effects of GnRHKLH vaccine on maternal/neonatal behavior, neonatal growth and development, serum biochemistry, and injection site reactions. Methods: Injection site reactions. On the day prior to treatment application (early January 2006) and while elk are lightly sedated (see pages 6-7 for details), we will perform ultrasound examination of the area of the expected injection site. After dart delivery of the vaccine, we will grossly monitor the injection site on a daily basis for one month for signs of inflammation or drainage. In addition, we will use ultrasound imaging each month for 6 months in conjunction with scheduled animal handling and blood collections to monitor changes in muscle echogenicity that would indicate sub-clinical abscesses or granuloma formation. Blood Chemistry and hematology. The physiological side-effects of GnRH-KLH treatment will be assessed by comparing serum chemistry, hematology, and body weight dynamics of treated and control elk. Blood samples will be collected in conjunction with previously described measurements just prior to GnRH-KLH vaccination and one week post-vaccination for evidence of short-term inflammation or infection. At three months post-vaccination, additional blood will be collected and analyzed for biochemistry profiles and evidence of abnormal organ function. These samples will be submitted for analysis to Colorado State University, Veterinary Teaching Hospital, Clinical Pathology Laboratory, Fort Collins, Colorado for analysis. Maternal Bonding, neonatal survival and growth. We will compare maternal/neonatal bonding and neonatal survival and growth of treated and control female elk for 30 days post-parturition during approximately 1 June to 1 July 2006. Parturition behavior of elk will be monitored daily beginning in late May and early June. We will document evidence of dystocia for each adult female, calf birth weight and health, acceptance or rejection by the dam, and growth to 30 days of age. For the purpose of this experiment, we assume that calf survival after 30 days is no longer a function of GnRH-KLH vaccine treatment and multiple factors other than maternal bonding will influence neonatal survival and body weight dynamics. Analysis: Means and standard errors of blood parameters, and neonatal growth rates will be estimated using least–squares ANOVA. Hypothesis tests will be based on type III generalized equations that account for correlations in repeated measures. 95 �Project Schedule: Date 1 May – June 2005 1 Sept 2005 7 Sept 2005 1 Jan 2006 1 Jan 2006 1 Feb – Sept. 2006 1 Feb – June 2006 1 June – July 2006 15 Sept. 2006 – 31 Oct. 2006 Jan 2007 Jan 2008 Jan 2009 Mar – July 2009 Activity Submit study plan for CDOW peer review and ACUC approval Semen evaluation and CIDR’s in all experimental female elk. Remove CIDR and combine males and females. Determine pregnancy status of females and assign to experimental groups. Apply contraceptive and monitor short term health effects. Monitor antibody titers of experimental elk. Monitor hormone levels Monitor birth rates, calf survival, calf weights and cow/calf behavior. Evaluate reproductive behavior Evaluate pregnancy rates 1 year post- vaccination. If funding is available, evaluate pregnancy rates 2 year post-vaccination If funding is available, evaluate pregnancy rates 3 year post-vaccination and/or reversibility of contraceptive treatments. Analyze data and prepare final report. Budget: This research proposal has been submitted to the Morris Animal Foundation for possible funding during the period June 2006 to Jan 2008. Category 2005-‘06 2006-‘07 Total Personal Services 0 0 0 1. Co-Investigator(s) 2. Biometrician 3. Wildlife Technicians (TBA) 5,000 6,675 2,500 11,175 7,500 17,850 Total Salaries and Wages 11,675 13,675 25,350 3,600 3,600 500 2,160 1,600 1,000 1,200 1,200 500 2,160 0 1,000 4,800 4,800 1,000 4,320 1,600 2,000 12,460 6,060 18,520 Operating Supplies & Services 1.Hormone serology LH analysis (240 x $20) Progesterone (240 x $20) PSPB (40 x $25) 2. GnRH Antibody Assays (360 x $12) 3. Biochemistry profile and CBC’s (40 x $40) 4. Miscellaneous veterinary supplies Total Supplies & Expenses Animal Maintenance Total Animal Care 5,760 5,760 11,520 Subtotal of All Categories 29,895 25,495 55,390 2,391 2,039 4,430 32,286 27,534 59,820 *Maximum of 8% - Indirect Costs (University Overhead) Grant Total 96 �F. LOCATION: This study will be conducted at the Colorado Division of Wildlife’s Foothills Wildlife Research Facility in Fort Collins, Colorado, USA. G. RELATED FEDERAL AID PROJECTS: N/A H. LITERATURE CITED ADAMS, T. H., AND B. M ADAMS. 1990. Reproductive function and feedlot performance of beef heifers actively immunized against GnRH. Journal of Animal Science 68:2793-2802. ASHER, G. W., AND J. L. ADAM. 1985. Reproduction of farmed red and fallow deer in northern New Zealand. Pages 217-224 In: P.F. Fennessy and K. R. Drew (eds.), Biology of deer production. Bulletin No. 22, Royal Society of New Zealand, Wellington, New Zealand. __________, M. W. FISHER, P. F. FENNESSY, C. G. MACKINTOSH, H. N. JABBOUR, AND C. J. MORROW. 1993. Oestrous synchronization, semen collection, and artificial insemination of farmed red deer (Cervus elaphus) and fallow deer (Dama dama). Animal Reproduction Science 33:241-265. ANDERSON, D. J., AND N. J. ALEXANDER.1983. A new look at antifertility vaccines. Fertility and Sterility 40: 557-571. ASPDEN, W. J., A. RAO, P. T. SCOTT, I. J. CLARK, T. E. TRIGG, J. WALSH, AND M. J. D’OCCHIO . 1996. Direct actions of the luteinizing hormone-releasing hormone agonist, deslorelin, on anterior pituitary contents of luteinizing hormone (LH) and follicle stimulating hormone (FSH), LH and FSH subunit messenger ribonucleic acid, and plasma concentrations of LH and FSH in castrated male cattle. Biology of Reproduction 55:386-392. BAIRD, D. T. 1992. Luteotrophic control of the corpus luteum. Animal Reproduction Science 28:95-102. BAKER, D. L., M. A. WILD, AND T. M. NETT. 2001. Technical support for elk and vegetation management in Rocky Mountain National Park. Wildlife Research Report. Colorado Division of Wildlife, Fort Collins, Colorado, USA. __________, ________, M. M. CONNER, H. B. RAVIVARAPU, R. L. DUNN, AND T. M. NETT. 2002. Effects of GnRH agonist (leuprolide) on reproduction and behavior in female wapiti (Cervus elaphus nelsoni). Reproduction Supplement 60:155-167. ___________, _________, M. D. HUSSAIN, R. L. DUNN, AND T. M. NETT. 2003. Evaluation of remotely delivered leuprolide formulation as a contraceptive agent in captive female elk. Technical report for elk and vegetation management for Rocky Mountain National Park, Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, Colorado, 125 pp. ____________, _________, M. M. CONNER, H. B. RAVIVARAPU, R. L. DUNN, AND T. M. NETT. 2004. Gonadotropin-releasing hormone agonist: a new approach to reversible contraception in female deer. Journal of Wildlife Diseases 40:713-724. BARLOW, N. D.1996.The ecology of wildlife disease control: simple models revisited. Journal of Applied Ecology 33:303-314. BARTH, A. D. 1986. In: Current Therapy in Theriogenology. D. A. Morrow (ed.) W. B. Saunders Co. pp205-209 ____________1997. Evaluation of potential breeding soundness of the bull. In: Current Therapy in Large Animal Theriogenology. R.S Youngquist (ed.) W.B. Saunders Co. pp222-236 BATES, G. N., J. BROOKS, AND J. CALL.1982.The use of prostaglandin to induce abortion in American elk (Cervus Canadensis). Zoo Animal Medicine 13:125-126 BOMFORD, M. 1990. A role for fertility control in wildlife management. Department of Primary Industries and Energy, Bureau of Rural Resources Bulletin No. 7, Australian Government Publishing Service, Canberra, Australia. BROWJN, B. W., P. E. MATTNER, P. A. CARROLL, E. J. HOLLAND, D. R. PAULL, R. M. HOSKINSON, AND R. D. G. RIGBY.1994. Immunization of sheep against GnRH early in life: Effects on reproductive function and hormones in rams. Journal of Reproduction and Fertility 101:15-21. 97 �BROWN, D. B., T. T. CAI, AND A. DASGUPTA. 2001.Interval estimation for a binomial proportion. 2001. Statistical Science 16:101-133. BUHR, M. M. 1987. Effect of lipoproteins and luteinizing hormone on progesterone production by large and small luteal cells throughout the porcine estrous cycle. Journal of Animal Science 65:10271033. CURRAN, S., R. A. PIERSON, AND O. J. GINTHER.1986.Ultrasonographic appearance of the bovine conceptus from days 20 through 60. Journal of the American Veterinary Medical Association 189:1295-1302. CURTIS, P. D., R. L. POOLER, M. E. RICHMOND, L. A. MILLER, G. F. MATTFELD, AND F. W. QUIMBY. 2002. Comparative effects of GnRH and porcine zona pellucida (PZP) immunocontraceptive vaccines for controlling reproduction in white-tailed deer (Odocoileus virginianus). Reproduction Supplement 60:131-141. DECKER, D., AND A. N. CONNELLY. 1989. Deer in suburbia-pleasure or pets. Conservationist 43:46-49. D’OCCHIO, M. J., W. J. ASPDEN, AND T. R.WHYTE.1996. Controlled, reversible suppression of oestrous cycles in beef heifers and cows using agonist of luteinizing hormone-releasing hormone. Journal of Animal Science 74:218-225. ____________ , G. FORDYCE, T. R. WHYTE, T. F. JUBB, L. A. FITZPATRICK, N. J. COOPER, W. J. ASPDEN, M. J. BOLAM, AND T. E. TRIGG. 2002. Use of GnRH agonist implants for long-term suppression of fertility in extensively managed heifers and cows. Animal Reproduction Science 74: 151-162. DUNBAR, B. S., AND E. SCHWOEBEL.1988. Fertility studies for the benefit of animals and human beings: Development of improved sterilization and contraceptive methods. Journal of the American Veterinary Medical Association 193: 1165-1170. FAGERSTONE, K. A., M. A. COFFEY, P. D. CURTIS, R. A. DOLBEER, G. J. KILLIAN, L. A. MILLER, AND L. WILMOT. 2002. Wildlife fertility control. Wildlife Society Technical Review 02-2. The Wildlife Society, Bethesda, Maryland, USA. FARIN, C. E., T. M. NETT, AND G. D. NISWENDER. 1990. Effects of luteinizing hormone on luteal cell populations in hypophysectomized ewes. Journal of Reproduction and Fertility 88:61-70. FENNESSY, P. F., C. G. MACKINTOSH, S. G. SHACKELL. 1990. Artificial insemination of farmed red deer (Cervus elaphus). Animal Production 51:613-621. FROST, H. C., G. L. STORN, M. J. BATCHELLER, AND M. M. LOVALLO. 1997. White-tailed deer management at Gettysburg National Military Park and Eisenhower National Historic Site. Wildlife Society Bulletin 25:462-469. FRAKER, M. A., R. G. BROWN, G. E. GAUNT, J. A. KEER, AND B. POHAJDAK. 2002. Long-lasting, singledose immunocontraception of feral fallow deer in British Columbia. Journal of Wildlife Management 66:1141-1147. GARROTT, R. A., P. J. WHITE, AND C. A. VANDERBIL WHITE. 1993. Overabundance: an issue for conservation biologist? Conservation Biology 7:946-949. ___________ AND D. B. SINIFF. 1992. Limitations of male-oriented contraception for controlling feral horse populations. Journal of Wildlife Management 56:456-464. GEIST, V. 1982. Adaptive behavioral strategies. In Elk of North America: Ecology and Management pp 219-277. J. W. Thomas and D. E. Toweill (Eds). Stackpole Books, Harrisburg, Pennsylvania, USA. GREER, K. R., AND W. W. HAWKINS. 1967. Determining pregnancy in elk by rectal palpation. Journal of Wildlife Management 31:145-149. HAIGH, J. C. AND R. J. HUDSON. 1993. Farming Wapiti and Red Deer. Mosby-Year Book, Inc. pp.121125. ________, A. D. BARTH, W. F. CATES, AND G. J. GLOVER. 1985. Electro-ejaculation and semen evaluation of wapiti. Royal Society of New Zealand. Bulletin, No. 22. p. 197-203 _________, W. F. CATES, G. J. GLOVER, AND N. C. RAWLINGS. 1984. Relationships between seasonal changes in serum testosterone concentrations, scrotal circumference and sperm morphology of male wapiti (Cervus elaphus). Journal of Reproduction and Fertility 70:413-418 98 �_________, M. CRANFIELD, AND R. G. SASSER. 1988. Estrus synchronization and pregnancy diagnosis in red deer. Journal of Zoo Animal Medication 19202-207. HARDER, J. D., AND D. L. MOORHEAD. 1980. Development of corpora lutea and plasma progesterone levels associated with the onset of the breeding season in white-tailed deer (Odocoileus virginianus). Biology of Reproduction 22:185-191. HAZUM, E., AND P. M. CONN. 1998. Molecular mechanism of gonadotropin releasing hormone (GnRH) action. I. The GnRH receptor. Endocrine Review 9: 379-386. HEILMANN, T. J., R. A. GARROTT, L. L. CADWELL, AND B. L. TILLER. 1998. Behavioral response of freeranging elk treated with an immunocontraceptive vaccine. Journal of Wildlife Management 62:243-250. HEIN, R. G., J. L. MUSSER, AND E. F. BRACKEN. 1991. Serologic, parasitic and pregnancy survey of the Colockum elk herd in Washington. Northwest Science 65:217-222 HOBBS, N. T., D. C. BOWDEN, AND D. L. BAKER. 2000. Effects of fertility control on populations of ungulates: general stage-structured models. Journal of Wildlife Management 64: 473-491. HONE, J. 1992. Rate of increase and fertility control. Journal of Applied Ecology 29:695-698. HUANG, F., D. C. COCKRELL, T. R. STEPHENSON, J. H. NOYES, AND R. G. SASSER. 2000. A serum pregnancy test with a specific radioimmunoassay for moose and elk pregnancy-specific protein B. Journal of Wildlife Management 64:492-499 JEWELL, P. A., AND S. HOLT. 1981. Problems in management of locally abundant wild animals. Academic Press, New York, New York, USA. KAND, S. H., AND S. J. KIM. 2004. A comparison of the three conditional exact tests in two-way contingency tables using the unconditional exact power. Biometrical Journal 46:320−330. KARTEN, M. J., AND J. E. RIVIER. 1986.Gonadotropin-releasing hormone analog design. Structurefunction studies toward the development of agonist and antagonist: rationale and perspectives. Endocrine Review 7:44-46. KEEN, J. E., G. P. RUPP, P. A. WITTENBERG, AND R. E. WALKER. 1999. Breeding soundness examination of North American bison bulls. Journal of the American Veterinary Medical Association 214:1212-1217 KILLIAN, G., L. MILLER, N. K. DIEHL, J. RHYAN, AND D. THAIN. 2004. Evaluation of three contraceptive approaches for population control of wild horses. Proceedings of the Vertebrate Pest Conference 21:263-268 KIRKPATRICK, J. F., AND J. W. TURNER, JR. 1985. Chemical fertility control and wildlife management. Bioscience 35: 485-491. ____________, AND ___________. 1996. Applications of pig zona pellucida immunocontraception to wildlife fertility control. Journal of Reproduction and Fertility 51:183-189. ____________, AND ___________. 2002. Reversibility of action and safety during pregnancy of immunization against porcine zona pellucida in wild mares (Equus caballus). Reproduction Supplement 60:197-202. ____________, ___________, I. K. M. LIU, R. FAYER-HOSKEN, AND A. T. RUTBERG. 1997. Case studies in wildlife immunocontraception: wild and feral equids and white-tailed deer. Reproductive Fertility and Development 9: 105-110. ___________, I. K. M. LIU, AND J. W. TURNER, JR. 1990. Remotely-delivered immunocontraception in feral horses. Wildlife Society Bulletin 18:326-330. ___________, R. NAUGLE, I., K. M. LIU, M. BERNOCO, AND J. TURNER. 1995. Effects of seven consecutive years of porcine zonae pellucidae contraception on ovarian function in feral mares. Biology of Reproduction Monograph Series 1:411-418. KRISHNAMOORTHY, K. AND J. THOMSON. 2002. Hypothesis testing about proportions in two finite populations. American Statistician 56:215−222. LEENAARS, M., E. CLAASSEN, AND C. F. M. HENDRIKSEN.1996. Considering the side-effects of adjuvant products in immunization procedures. Laboratory Animals 30:40-43. LEHNER, P. N. 1996. Handbook of ethological methods. 2nd ed. Cambridge University Press, Cambridge. 99 �LEOPOLD, A. S., S. A. CAIN, C. M. COTTAM, AND I GABRIELSON. 1963. Wildlife management in the national parks. Transactions of the North American Wildlife and Natural Resources Conference 28:28-45. LEVY, J. K., L. A. MILLER, P. C. CRAWFORD, J. W. RITCHEY, M. K. ROSS, AND K. A. FAGERSTONE. 2004. GnRH immunocontraception of male cats. Theriogenology 62(6):1116-1130. LUCY, M. C., H. J. BILLINGS, W. R. BUTLER, L. R. EHNIS, M. J. FIELDS, D. J. KESLER, JR., J. E. KINDER, R. C. MATTOS, R. E. SHORT, W. W. THATCHER, R. P. WETTEMANN, J. V. YELICH, AND H. D. HAFS. 2001. Efficacy of an intravaginal progesterone insert and an injection of PGF2α for synchronizing estrus and shortening the interval to pregnancy in postpartum beef cows, peripubertal beef heifers, and dairy heifers. Journal of Animal Science 79:982-995. MATSUO, H., Y. BABA, R. M. G. NAIR, A. ARIMURA, AND A. V. SCHALLY. 1971. Structure of the porcine LH- and FSH-releasing hormone. I. The proposed amino acid sequence. Biochemical and Biophysical Research Communication 43:1334-1339. MCANINCH, J. B., editor. 1993. Urban deer: a manageable resource? Proceedings of the 1993 Symposium of the North Central Section, The Wildlife Society, St. Louis, Missouri, USA. MCCULLOUGH, D. R., K. W. JENNINGS, N. B. GATES, B. G. ELLIOT, AND J. E. DIDONATO. 1997. Overabundant deer populations in California. Wildlife Society Bulletin 25: 478-483. MCNEILLY, A. S., AND H. M. FRASER.1987. Effect of gonadotrophin-releasing hormone agonist – induced suppression of LH and FSH on follicle growth and corpus luteum function in the ewe. Journal of Endocrinology 115:273-282. MCSHEA, W. J., S. L. MONFORT, S. HAKIM, J. KIRKPATRICK, I. LIU, J. W. TURNER, JR., L. CHASSY, AND L. MUNSON. 1997. The effect of immunocontraception on the behavior and reproduction of white-tailed deer. Journal of Wildlife Management 61:560-569. MELOEN, R. H., J. A. TURKSTRA, H. LANKHOF, W. C. PUIJK, W. M. M. SCHAAPER, G. DIJKSTRA, C. J. G. WENSING, AND R. B. OONK. 1994. Efficient immunocastration of male piglets by immunoneutralization of GnRH using a new GnRH-like peptide. Vaccine 12:741-746. MILLER, L. A., B. E. JOHNS, AND G. J. KILLIN. 2000a. Long-term effects of PZP immunization on reproduction in white-tailed deer. Vaccine 18:568-574. __________ , _________, AND ___________ . 2000b. Immunocontraception of white-tailed deer with GnRH vaccine. American Journal of Reproductive Immunology 44:266-274. __________, J. C. RYAN, AND M. DREW. 2004. Contraception of bison by GnRH vaccine: a possible means of decreasing transmission of brucellosis in bison. Journal of Wildlife Diseases 40:725730. MONTOVAN, S. M., P. P. DAELS, J. RIVIER, J. P. HUGHES, G. H. STABENFELDT, AND B. L. LASLEY. 1990. The effect of potent GnRH agonist on gonadal and sexual activity in the horse. Theriogenology 33: 1305-1321. NASH, P. B., J. K. JAMES, L. T. HUI, AND L. A. MILLER. 2004. Fertility control of California ground squirrels using GnRH immunocontraception. Proceedings of the 21st Vertebrate Pest Conference 274-278. NAUGLE, R. E., A. T. RUTBERG, H. B. UNDERWOOD, J. W. TURNER, JR., AND I. K. M. LIU. 2002. Field testing of immunocontraception on white-tailed deer (Odocoileus virginianus) at Fire Island National Seashore, New York, USA. Reproduction Supplement 60:143-153. NETT, T. M., M. E. CROWDER, G. E. MOSS, AND T. M. DUELLO.1985.GnRH-receptor interaction. V. Down-regulation of the pituitary receptors for GnRH in ovariectomized ewes by infusion of homologous hormone. Biology of Reproduction 24:1145-1155. NISWENDER, G. D. 1973. Influence of the site of conjugation on the specificity of antibodies to progesterone. Steroids 22:413-424. __________, J. L. JUENGEL, W. J. MCGUIRE, C. J. BELFIORE, AND M. C. WILTBANK. 1994. Luteal function: The estrous cycle and early pregnancy. Biology of Reproduction 50:239-247. __________, L. E. REICHERT, JR., A. R. MIDGLEY, AND A. V. NALBANDOV. 1969. Radioimmunoassay for bovine and ovine luteinizing hormone. Endocrinology 84:1166-1173. 100 �___________, T. J., REIMERS, M. A. DIEKMAN, AND T. M. NETT. 1976. Blood flow: a mediator of ovarian function. Biology of Reproduction 14: 64-81. OKUDA, K., Y. UENOYAMA, C. NAITO, Y. SAKABE, AND N. KAWATE. 1999. Luteinizing hormone receptors in the bovine corpus luteum during the oestrous cycle and pregnancy. Reproduction, Fertility, and Development 11:147-151. ONCLIN, K., J. P. VERSTEGEN, AND P. W. CONCANNON. 2000. Time-related changes in canine luteal regulation: in vivo effects of LH on progesterone and prolactin during pregnancy. Journal of Reproduction and Fertility 118:417-424. PETERS, K. E., E. G. BERGFELD, A. S. CUPP.1994. Luteinizing hormone has a role in development of fully functional corpora lutea (CL) but is not required to maintain CL function in heifers. Biology of Reproduction 52 (Suppl) 1:35. PLOTKA, E. D., U. S. SEAL, L. J. VREME, AND J. J. OZOGA. 1982. Reproductive steroids in white tailed deer. IV. Origin of progesterone during pregnancy. Biology of Reproduction 26:258-262 __________, D. N. VEVEA, T. C. EAGLE, D. B. SINIFF, AND J. R. TESTER. 1989. Ovarian function in captive feral mares. Journal of Wildlife Diseases 25:574-579. PORTER, W. F., AND B. UNDERWOOD. 1999. Of elephants and blind men: deer management in the U.S. national parks. Ecological Applications 9:3-9. POWERS, J. G., P. B. NASH, J. C. RHYAN, AND L. A. MILLER. Comparison of AdjuVac™ and Freund’s Complete adjuvants with GnRH-KLH antigen in New Zealand white rabbits. (In preparation). RABB, M. H., D. L. THOMPSON, JR., B. E. BARRY, D. R. COLBORN, K. E. HEHNKE, AND F. GARZA, JR. 1990. Effects of active immunization against GnRH on LH, FSH, and prolactin storage, secretion, and response to their secretagogues in pony geldings. Journal of Animal Science 68:3322-3329. RAHE, C. H., R. E. OWENS, J. L. FLEEGER, H. J. NEWTON, AND P. G. HARMS. 1980. Pattern of plasma luteinizing hormone in the cyclic cow: dependence upon the period of the cycle. Endocrinology 107:498-503. RHYAN, J. C., AND M. D. DREW. 2002. Contraception: A possible means of decreasing transmission of brucellosis in bison. In: Brucellosis in elk and bison in the Greater Yellowstone Area, T.J. Kreeger (ed.). Greater Yellowstone Interagency Brucellosis Committee, Wyoming Game and Fish Department, Cheyenne, Wyoming, 99-108. ROPSTAD, E., AND D. LENVIK. 1991. The use of cloprostenol and prostaglandin F2α to induce luteolysis in reindeer calves (Ranger tarandus). Rangifer 11:13-16. RUDOLPH, B. A., W. F. PORTER, AND H. B. UNDERWOOD. 2000. Evaluating immunocontraception for managing suburban white-tailed deer in Irondequoit, New York. Journal of Wildlife Management 64:463-473. SAS INSTITUTE. 1997. SAS/STAT® users guide 6.03 edition. SAS Institute, Inc., Cary, North Carolina, USA SEAGLE, S. W., AND J. D. CLOSE.1996. Modeling white-tailed deer population control by contraception. Biological Conservation 76:87-91. SHIDELER, S. E., M. A. STOOPS, N. A. GEE, J. A. HOWELL, AND B. L. LASLEY. 2002. Use of porcine zona pellucida (PZP) vaccine as a contraceptive agent in free-ranging tule elk (Cervus elaphus nannodes). Reproduction Supplement 60:169-176. STELLFLUG, J. N. Y. S. WEEMS, AND W. W. WEEMS. 1997. Clinical reproductive physiology of ewes. In: Current Therapy in Large Animal Theriogenology. R.S Youngquist (ed.) W.B. Saunders Co. pp594-598 STEVENSON, J. S. 1997. Clinical reproductive physiology of the cow. In: Current Therapy in Large Animal Theriogenlogy. R.S Youngquist (ed.) W.B. Saunders Co. pp257-267 STILLS, H. F., JR., AND M. Q BAILEY. 1991. The use of Freund’s complete adjuvant. Laboratory Animals 20:25-30. SQUIRES, E. L.1993. Endocrinology of pregnancy. In: Equine Reproduction A. O. McKinnon and J. L. Voss (Eds). Williams and Wilkins, Pennsylvania, USA. pp495-508 TAST, A., R. J. LOVE, I. J. CLARKE, AND G. EVANS. 2000. Effects of active and passive gonadotrophin- 101 �releasing hormone immunization on recognition and establishment of pregnancy in pigs. Reproduction, Fertility and Development 12:277-282. TOMAS, P. G. A. 1997. Induced abortion. In: Current therapy in large animal theriogenology, R. S. Youngquist (ed.).W.B. Saunders Co. 303-306 TURKSTRA, J. A., F. J. U. M. VAN DER MEER, J. KNAAP, P. J. M. ROTTIER, K. J. TEERDS, B. COLENBRANDER, AND R. H. MELOEN. 2005. Effects to GnRH immunization in sexually mature pony stallions. Animal Reproduction Science 86:247-259. TURNER, J. W., JR., I. K. M. LIU, AND J. F. KIRKPATRICK. 1992. Remotely delivered immunocontraception in captive white-tailed deer. Journal of Wildlife Management 56:154-157. __________, J. F. KIRKPATRICK, AND I. K. M. LIU. 1996. Effectiveness, reversibility and serum antibody titers associated with immunocontraception in captive white-tailed deer. Journal of Wildlife Management 60:45-51. __________, AND _________. 2002. Effects of immunocontraception on population longevity and body condition in wild mares (Equus caballus). Reproduction Supplement 60:187-195. WALTER, W. D., P. J. PERKINS, A. T. RUTBERG, AND H. J. KILPATRICK. 2002. Evaluation of immunocontraception in a free-ranging suburban white-tailed deer herd. Wildlife Society Bulletin 30:186-192. WRIGHT, R. G. 1993. Wildlife management in parks and suburbs: alternatives to sport hunting. Renewable Resources Journal 11: 18-22. ZAR J. H. 1984. Biostatistical analysis. Prentice-Hall, Inc. Englewood Cliffs, New Jersey, USA. 102 �APPENDIX II BAKER, D. L., M. A. WILD, M. M. CONNER, H. B. RAVIVARAPU, R. L. DUNN, AND T. M. NETT. 2005. Evaluation of a remotely delivered formulation of leuprolide acetate as a contraceptive agent in female elk (Cervus elaphus nelsoni). Journal of Wildlife Diseases 41: in press. Abstract: Practical application of fertility control technology in free-ranging wild ungulates often requires remote delivery of the contraceptive agent. The objective of this investigation was to evaluate the potential of remote delivery of leuprolide acetate for suppressing fertility in female elk (Cervus elaphus nelsoni). Fifteen captive adult female elk were randomly allocated to one of three experimental groups. Six elk were injected intramuscularly with a dart containing leuprolide, and the remaining nine elk received the same formulation without leuprolide. We determined pregnancy rates, suppression of luteinizing hormone (LH) and progesterone concentrations, and reversibility of treatments during 1 August 2002 to 3 September 2003. Leuprolide formulation caused a decrease in concentrations of LH and progesterone, temporary suppression of ovulation and steroidogenesis, and effective contraception (100%) for one breeding season. These results extend the practical application of this contraceptive agent to include dart delivery, where in the absence of such technology, wild elk must first be captured and restrained prior to treatment. BAKER, D. L., M. A. WILD, M. M. CONNER, M. D. HUSSAIN, R. L. DUNN, AND T. M. NETT. 2006. Evaluation of leuprolide as a contraceptive agent in free-ranging elk in Rocky Mountain National Park, Colorado. Journal of Wildlife Management (in preparation). ___________., _________, M. D. HUSSAIN, R.L. DUNN, AND T. M. NETT. 2006. Leuprolide acetate as a contraceptive agent in female elk: determination of minimum effective dose. Reproduction (in preparation) LUKACS, P., J. GROSS, AND D. BAKER. 2006. Estimating confidence intervals for fawn:doe and buck:doe ratios from counts across days. Journal of Wildlife Management (in preparation). INSLERMAN, R. A., J. E. MILLER, D. L. BAKER, J. E. KENNAMER, R. CUMBERLAND, B. STINSON, P. DOERR, AND S. J. WILLIAMSON. 2005. The Wildlife Society Technical Review Committee on Baiting and Artificial Feeding of Game Wildlife Species (in preparation). 103 � https://cpw.cvlcollections.org/files/original/87f1bec9d928f16c03ecdd82c4ab0334.pdf 66f7fb0598aaede242d73226ee8869e8 PDF Text Text 191 Colorado Division of Wildlife Wildlife Research Report July 2001 and July 2002 JOB PROGRESS REPORT State of ______C=-o=l=o=ra=d...o'--------- Mammals Research Work Package ------=-30-=--0=2"--------- Elk Conservation Task No. 3 ------"---------- Estimating Calf and Adult Survival Rates and Pregnancv Rates of Gunnison Basin Elk Project No. _ _ _W--'-'---"'-1=53=----=-R,._-.:. . 14..:. . . . :1=5_ _ _ __ Research and Development Period Covered: July 1, 2000 - June 30, 2001, and July 1, 2001 - June 30, 2002 Author: D. J. Freddy Personnel: D. Cole, N. Gallowich, D. Masden, L. Gepfert, J. Olterman, R. Basagoitia, L. Spicer, B. Carochi, J. Oulton, P. Mason, W. Brown, V. Organek, J. Young, T. Beck, G. Dekleva, C. Mehaffey, D. Williams, J. Johnston, K. Buffington, K. Fanson, and R. Kahn of CDOW, Dr. G. C. White and Dr. M.C. Conner of Colorado State University, CDOW Gunnison Habitat Partnership Program, and contractors/ cooperators, Helicopters by OZ, Coulter Aviation, USFS, BLM, private land owners, and elk hunters. ABSTRACT We estimated survival rates and pregnancy rates of elk (Cervus elaphpus nelsonii) in the Gunnison Basin of Colorado during 2000 and 2001. During mid-December each year, we captured and radio-collared calves age 6 months and adult females age 2:2 years and during November and December each year, we had hunters collect and submit reproductive organs from female elk harvested during laterifle seasons. During winter-spring, survival rates of calves were 0.89 ± 0.08 (CL) (n = 71), 0.83 ± 0.09 (n = 75), and 0.86 ± 0.06 (n = 146) for 2000-01, 2001-02, and both years combined, respectively. Survival of calves was not different between years (P = 0.2965) or sexes (P = 0.1456) but tended to be different among 3 management DAUs (P = 0.0737) with survival being lowest at 0.78 ± 0.12 in DAU E41. For years combined, 21 calves died with proximate causes of death being 53% predation-related, 24% malnutrition-related, 9% accidents, and 14% unknown causes. In 2000-01, calves tended to die after mid-March while in 2001-02, mortalities occurred from early January through May. Patterns of calf mortalities were not strongly associated with calf body mass. Calf body mass at capture averaged 99.1 ± 2.2 kg, ranged from 52.0 to 133.0 kg, and was not different between years or sexes (P > 0.259), but calves were larger in DAU E-41 (P = 0.003) where calf survival was lowest. Survival rate of adult females age 2:_2 years was 1.00 during winter-spring as no deaths occurred during 2000-01 (n = 39) and 2001-02 (n = 48) and annual survival was 0.92 ± 0.08 (n = 39) including hunting and other causes of death and 0.97 ± 0.05 (n = 37) including only natural deaths. Survival for yearling female elk, age 12-17 months, during summer-fall was 0.89 ± 0. 10 (n = 38) including hunting and other causes of death and 1.00 (n = 34) including only natural deaths. Survival for the same cohort of yearling male elk during summer-fall was 0.86 ± 0.15 (n = 22) including hunting and other causes of death and 0.90 ± 0.13 (n = 21) including only natural deaths. Survival rates for both yearling female and male elk, age 18-23 �192 months, were 1.00 (n = 34 F, n = 19 M) during winter-spring as no deaths occurred. Harvest removal rates during summer-fall 2001 were 0.05 for adult females, 0.11 for yearling females, 0.08 for all adult females age ~12 months, and 0.06 for yearling males. Based on biological collections provided by hunters, pregnancy rate averaged 85% for all adult female elk age ~1 year (n = 89). Conceptions peaked 23 September, spanned 68 days, and followed an expected asymmetrical pattern in timing with 17% of the adult females likely conceiving after 10 October. Litter size was 1 in all uteri with detectable fetuses (n = 69) and female fetuses predominated with fetal sex ratio (37F:21M) deviating from 50:50 (P = 0.036). Estimated percent total body fat based on kidney fat measurements indicated 65% of the adult females age ~1 year were in moderate, 30% in low, and <5 % in very low or very good body condition. Probability (logit(P)) of adult females being pregnant was dependent on estimated percent total body fat (P = 0.033). Measures ofreproductive and survival rate parameters were consistent with predictions of performance outcomes for adult female elk having low to moderate body condition status in the fall. More than likely, marginally deficient levels of seasonal nutrition were depressing optimal reproductive performance of adult female elk. All information in this report is preliminary and subject to further evaluation. �193 JOB PROGRESS REPORT ESTIMATING CALF AND ADULT SURVIVAL AND PREGNANCY RA TES OF GUNNISON BASIN ELK POPULATIONS David J. Freddy P. N. OBJECTIVE Estimate survival rates of calf, adult female, and adult male elk and estimate pregnancy rates of adult female elk in Gunnison Basin elk populations for 3 years. SEGMENT OBJECTIVES 1. Prepare study plan program narrative. 2. Estimate calf, adult female, and adult male survival rates during winter, December-June. 3. Estimate adult male and female survival rates during summer-fall, June-November. 4. Estimate harvest removal rates for yearling and adult males and females. 5. Estimate pregnancy rates, fetal rates, conception dates, and body condition of female elk collected in December. 6. Summarize data in Research Progress reports and prepare peer-reviewed publications. INTRODUCTION The elk (Cervus elaphpus nelsonii) resource has many benefits but frequent social, political, and economic conflicts suggest elk can reach "social" if not "biological" carrying capacities (Freddy et al. 1993). Recent controversy surrounding elk in the Gunnison Basin of Colorado (Roath et al. 1999) exemplifies conflicting social and biological agendas regarding appropriate numbers of elk. The core of conflict in elk management often centers on establishing management objectives for numbers of elk that are agreeable to competing interests and then monitoring elk populations to demonstrate that objectives are achieved. This type of conflict is paramount in Colorado Division of Wildlife (CDOW) elk population Data Analysis Units (DA Us) E-25, E-41, and E-43 in the Gunnison Basin where a combination of resource carrying capacity objectives for elk on winter ranges and difficulties associated with knowingly achieving those objectives has fostered argumentative distrust among public groups and management agencies. Accomplishing management by population objective can depend on reliably estimating elk population size which is expensive and intensive (Samuel et al. 1987, Bear et al. 1989, Unsworth et al. 1990, Anderson et al. 1998, Cogan and Diefenbach 1998, Eberhardt et al. 1998, Freddy 1998). Alternatively, population size and trend can be estimated using computer models that incorporate harvest, age and sex ratios, and survival rates (White 1992, Bartholow 1999). • Model outputs are extremely sensitive to estimates of survival rates such that, reliable measurements of survival can greatly enhance the quality of models (Nelson and Peek 1982). We chose to estimate survival rates of calf and adult elk during winter and adults year-around to aid in developing improved population models for the Gunnison Basin elk. The Gunnison Basin in southcentral Colorado encompasses the entire headwaters of the main Gunnison River and the centrally �194 located town of Gunnison. Between 12-16,000 elk and 8-10,000 mule deer (Odocoileus hemionus) are thought to exist within the Basin. Elk are managed as 3 populations representing DA Us E-25 (Game Management Units [GMU] 66, 67), E-41 (GMU 54), cµid E-43(GMUs 55,551). The 3 DAUs encompass about 9,291 km2 of which 3,648 km2 are considered potential winter range for elk (CDOW WRIS database). DA Us are contiguous with few major geographic barriers separating DA Us that would absolutely prevent interchange of elk among DA Us (see Program Narrative [PN] Appendix I Figure 1). The Basin represents a high altitude, cold winter range for both elk and mule deer which is similar to ecosystems in North Park, Middle Park, and the San Luis Valley, Colorado. The sagebrush (Artemisia tridentata) steppe winter ranges (2,250- 2,700 m elevation) can receive both extreme snow depths and cold temperatures that cause severe mortality among ungulates (Carpenter et al. 1984) while the conifer meadow and alpine summer ranges (3,000 - 4,200 m elevation) can be lush sources of forage subjected to periodic drought. Overall, these ranges collectively are thought to be less productive and nutritious for elk than the milder climate oakbrush-pinyon-juniper winter ranges and aspen and subalpine summer ranges of the Grand Mesa, Colorado where elk survival was measured from 1993-2000 (Freddy 2000). METHODS Capture Adult female (age 2:,2 years) and calf (age 6 months) male and female elk were captured and radiocollared using helicopter net-gunning from 16-22 December, 2000 and 16-20 December 2001 (Freddy 1993, see PN Appendix III). All radio-collars were l 72-l 76 l\.1Hz and contained 4-6 hour mortality sensors (Lotek, Inc., see PN Appendix I). Calf collars were expandable allowing collars to remain on elk as they matured to adults (see PN Appendix I). Objectives were to capture and radio-collar 78 calves each year with 13 calves of each sex in each of the 3 Gunnison Basin DA Us. For adult females, objectives were to capture and radio-collar 39 during the first year with 13 in each DAU and in subsequent years, capture sufficient adult females to maintain 2:,13 adult radioed females in each DAU (see PN, Sample Sizes). Prior to capturing elk, the 3 DA Us demarcating the entire Gunnison Basin were divided into 10 geographic trap-zones (Figure 1 A-J, see PN Figure 1). Numbers of elk to be captured in each trap-zone within a DAU were based upon the proportion of elk observed in each trap-zone within each DAU during elk sex and age composition surveys conducted with a helicopter during December-January post-harvest 1995-1999. We attempted, therefore, to distribute our sample of radioed elk across the landscape in proportion to relative elk numbers during early winter within each DAU. Elk were captured within a 3-km radius of.39 processing sites with some sites common to both years (Figure 1). Capture sites were systematically distributed within trap-zones within DAUs to radio-collar elk representing multiple segments of the entire Gunnison Basin population. Although our capture sites were not based on previously selected random coordinates, we believe we achieved a representative sample of elk from the population to provide relatively unbiased estimates of survival rates. Capture sites were accessed via vehicles when possible or by ferrying capture crews in the helicopter to more remote locations inhabited by elk. Calves were ferried by helicopter from individual elk capture locales to processing sites where body mass (kg), total body length (cm), hind foot length (cm), and rectal body temperature (F) were measured and then calves were radio-collared and released (see PN Appendix ill). All body measurements were made with the same instrumentation by the same individuals both years. Adult females were captured, aged as 2-4 years, 5-9 years, and >9+ years old based on incisor replacement or relative wear, radio-collared, and released at location of capture. We avoided capturing yearling 18-month-old females. Photographs of �195 incisor replacement and wear by elk age-class were provided to handlers responsible for judging the age of adult elk prior to releasing adults. No ear-tags were applied to calf or adult elk. Calf body measurements were compared among years, sexes, and DAUs using SAS (1988, PROC FREQ, PROC GLM[ANOVA]). All capture protocols were approved by the CDOW Animal Care and Use Committee. Telemetry Monitoring We monitored life or death status ofradioed elk daily from December through June from accessible roads using a truck equipped with magnetic-mounted omni-directional and 3-element hand-held Yagi antennas and at 1-4 week intervals from December through November using a Cessna 185 or 182 equipped with strut and/or belly mounted 'H' antennas. We used Lotek® SRX400 and Telonics® TR2 receiverscanners for monitoring telemetry signals. Elk survival data were compiled using the RADIOS module of the CDOW program DEAMAN® (White 1991). Mortality Assessments All suspected mortalities based on telemetry mortality signals were confirmed using ground searches. Once carcasses were located, criteria for assigning probable cause of death followed standardized written procedures that included assessment of body position and body condition, presence of bite or claw marks and sub-dermal hemorrhaging or gunshot wounds, presence of tracks or drag marks, and collection of organ, muscle, and femur marrow samples for laboratory analyses, if available (Wade and Browns 1982, Freddy 1998). Multiple photographs were taken of the carcass along with any potential evidence for assessing cause of death and when appropriate, an outside expert (T. D. I. Beck, CDOW) was consulted to assess evidence. Field necropsies were performed to the extent possible depending on completeness of carcass. We routinely collected muscle samples from large muscle groups in the hind- and forequarters of carcasses when available to assess for evidence of capture myopathy (Lewis et al. 1977, Spraker 1982, Haigh and Hudson 1993). Histopathology assessments of organ and muscle samples were completed by the Colorado State University Veterinary Diagnostic Laboratory and analyses of percent femur marrow fat on a dry-matter basis were conducted by the CDOW research laboratory. Field technicians provided a standardized written summary for each death. The principal investigator made the final assessment for probable cause of death based upon field summaries, photographs, and laboratory analyses. Potential causes of death included malnutrition, predation by black bears (Ursus americanus), mountain lions, (Felis concolor), coyotes (Canis latrans), and domestic dogs (Canis familiaris), legal and illegal hunter harvest, accidental trauma, plant poisoning, capture-induced , and unknown (Freddy 1997). Cause~ of death were broadly summarized as malnutrition, predation, suspected malnutrition, suspected predation, accident,.unknowp., hunter harvest, and capture-induced. Mortalities classed as malnutrition were usually nearly intact carcasses with little or no evidence of predator presence whereas mortalities classed as predation usually had evidence of bite wounds and subdermal hemorrhaging indicating bites were inflicted on a live animal. In those cases classed as suspected malnutrition or suspected predation a preponderance of collective evidence was used to assign cause of death to the most likely class. Telemetry collars that prematurely slipped-off elk causing a mortality signal to be emitted were confirmed by locating and retrieving the collar. Elk were subjected to multiple hunting seasons during fall 2000 and 2001. These seasons with approximate dates were: archery, 25 August-23 September; muzzleloading, 8-16 September; elk-only, 1317 October; deer-elk first combined, 20-26 October; deer-elk second combined, 3-9 November; deer-elk third combined, 10-14 November; late antlerless elk only, 24 November - 16 December in GMUs 54 and 55 (portions ofDAUs E-41 and E-43); and, late antlerless elk only, 1-31 December in GMU 66 (portion of DAU E-25). �196 Survival Rates Survival rates of radioed elk were calculated for this report using the binomial estimator and in final analyses will be calculated using a Kaplan-Meier estimator without staggered entry (White and Garrott 1990). Binomial estimates of survival rates were calculated as mean survival (s ) = [Alive / Alive+Dead collared elk], with a variance of [VAR (s) =(s)*(l- s) / n collars], and 95% confidence intervals of (s) ± [t a.-o.os, n-J aJ *J°( VAR (s))]. Survival rates were estimated for time intervals of winter-spring (15 December - 14 June), summer,-fall (15 June - 14 December), and yearly (15 December - 14 December) which coincided with capturing and radio-collaring elk and thus represented a biological year. By definition, calf elk became 12-month-old yearlings on 15 June and calves surviving to this date were considered to be recruited into the population. For adult elk during time intervals that included hunting seasons, we calculated survival rates inclusive of natural and hunting related mortalities, exclusive of hunting mortalities, and exclusive of natural mortalities. Excluding, or censoring hunting mortalities, provided estimates of natural survival rates, while censoring natural mortalities but including hunting mortalities provided estimates of hunting removal rates calculated as f = (1 - s), with (s) being survival rate with natural mortalities censored. Chi-square contingency tests were initially used for comparing calf survival (alive or dead categories) between sexes, years, and DA Us (White and Garrott 1990, SAS 1988 PROC FREQ). Parameter estimates were expressed as means± 95% confidence limits unless otherwise noted. Elk dying of suspected captured-induced trauma were censored from survival estimates. Deaths of calves or adults occurring within 1 week of capture were likely to be classed as capture-induced deaths unless field evidence strongly suggested a natural cause of death independent of capture. Capture-induced trauma could affect animals for up to 2-4 weeks post-capture so we routinely attempted to assess whether deaths were potentially capture-induced. We also censored elk having telemetry collars that electronically failed or slipped-off the elk (White and Garrott 1990). Elk with failed or slipped collars were censored for an entire seasonal time interval for binomial survival estimates and will be censored on the date they were last known alive based on telemetry signals in Kaplan-Meier estimates of survival. Elk whose telemetry signals disappeared during hunting seasons continued to be monitored for several subsequent months over large geographic areas until such time these elk were judged to have likely been removed during hunting seasons. Radioed elk that disappeared during hunting seasons were assumed to be legally harvested. Reproductive Collections Fecundity of adult female elk (age ,2:1 year) was estimated by examining reproductive organs of antlerless elk harvested during limited-entry late-hunting seasons that occurred from mid-November through December in portions of GMUs 54, 55, and 66. About 2-3 weeks prior to the beginning of seasons, we mailed permitted hunters collection packets· ~xplaining procedures f~r ·obtaining reproductive organs and incisor teeth (for dental cementum aging) from harvested elk as done previously in Colorado (Freddy 1992). Additionally, we asked hunters to collect kidneys with associated fat from these elk to allow calculation of kidney-fat indices and estimates of percent total fat to better assess body condition of adult females in relation to reproductive status (Kohlmann 1999, Cook et al. 2001a, 2001b). Hunters were instructed to place samples in collection bags and leave specimens in protected containers that kept samples cool at several drop-off sites within the Gunnison Basin from which samples were picked-up almost daily by project personnel. Dental cementum aging was completed by the CDOW research laboratory. Pregnancy status of elk was determined as: pregnant was uterus with embryo, fetus or fetal membranes; not pregnant was no evidence of fetus, no active uterine tissue, and no active corpora lutea of pregnancy; suspected pregnant was active corpora lutea, apparently active uterine tissue but no visible embryo or fetus; and unknown was either incomplete sample or no sample available. Fresh fetuses were sexed, �197 weighed, and measured (Armstrong 1950, Morrison et al. 1959). Fetuses and questionable uteri were stored in I 0% buffered formalin for reference examination. Morrison et al. (1959) graphically presented a logarithmic relationship between fetal crown-rump length (Y, dependent variable) and known fetal-age in days (X, independent variable) but did not present a standardized equation. To develop a standardized arithmetic equation for predicting fetal age, traditionally the unknown variable of interest, from fetal crown-rump length, traditionally the variable measured, we first used Morrison et al. data in MS-Excel® curve-fitter to develop 2 predictive polynomial equations. These equations were: (a) y(fetalcrown-nnnpmm) = 0.0085X\fetalageclays) + l.7603X 57.034, r2 = 0.9969, for the complete 8 data points presented by Morrison et al. (1959), and (b) Ycreta1crownrumpmm) = 0.0194X 2creta1agec1ays) + 0.2521X - 17.51, r2 = 0.9987, for 7 Morrison et al. ( 1959) data points with their late March data point excluded. We found equation (b) reduced the error in predicted crown-rump measurements by 2::_50% over equation (a) when predicted crown-rump lengths were compared with Morrison et al. actual crown-rump lengths, especially for the critical 60-90 day fetal-age stage that was associated with fetal collections occurring in November-December. Because fetal age in days is what is estimated from measured crown-rump values, we input polynomial equations (a) and (b) into program DERNE® to solve for fetal-age days (X) in terms of crown-rump measurements (Y). These DERNE® equations were: for polynomial equation (a) Xcre1aJageciays) = {[-f(3400000*Ycreta1crown-rumpmm) + 503781209) -17603] / 170}, and for polynomial equation (b) Xcreia1ase days)= {[-!(7760000* y(fetalcrown-nnnpmm) + 142256321) -2521] / 388}. We used DERNE® equation (b) to estimate elk fetal ages from crown-rump measurements. Fetal age in days was subtracted from date of hunter collection and then converted to calender and Julian dates of estimated conception. We measured total kidney fat mass and trimmed kidney fat mass after Riney ( 1955), Kohlmann ( 1999), and Cook et al. (200la,b). We calculated modified total and trimmed kidney fat indices after Anderson et al. (1990), Kohlmann (1999), and Cook et al. (200la,b). We estimated percent total body fat from measurements of kidney fat using simple linear equations that predicted percent body fat from kidney total fat mass (TFM), full kidney fat index (TF-KFI), and trimmed kidney fat index (TRF-KFI) presented by Cook et al. (2001a). We commonly received only I kidney fat mass submitted with reproductive samples so for those elk for which we received both kidney masses, we averaged kidney fat measurements to produce I value per elk (Cook et al. 2001a). Although TFM was potentially the best predictor of percent body fat of the measurements we made (see Cook et al. 2001a) we had no control on how well hunters collected all fat associated with kidneys so we conservatively viewed percent body fat estimates derived from trimmed kidney fat might be more accurate because we standardized the amount of fat measured among samples. All reproductive measurements were compiled in MS-Excel® and analyzed with SAS (1988, PROC FREQ, PROC UNNARIATE, PROC REG, PROC GLM[ANOVA], PROC LOGISTIC). General Elk Movements During aerial flights to monitor survival status of elk, we interpreted signal strength and direction to judge general locations of telemetry signals for each elk as to primary drainages or topographic features to describe general movements of elk to and from seasonal ranges. Elk that made large or unique movements, such as across main highways or DAU boundaries, were located relatively precisely from the airplane with the radius of location error likely <1,000 m. This location data was not gathered to assess specific habitats used but rather to describe the major movement patterns of these elk. Data will be summarized in future reports using ArcView 3.2®. �198 RESULTS AND DISCUSSION Capture In 2000, we radio-collared 78 calves, of which 38 were males and 40 were females, and 39 adult females age ::::2 years. Frequency of age classes for adult females were: 2-4 years= 3, 5-9 years = 30, and >9 years= 6. We achieved our objective of capturing 13 adult females and 26 calves in each DAU and nearly balanced sex ratios among calves in each DAU (Table 1). Elk were captured at 20 different sites representing a broad geographic area in the Gunnison Basin (Figure 1, Appendix A). There were no acute deaths of calves during capture but 2 adult females died while being blindfolded and hobbled prior to radio-collaring. Upon necropsy, both adults had extensive hemorrhaging in the thoracic cavity but no hemorrhaging in the abdominal cavity and no obvious indications of cervical injuries. We surmised that the heart-lung complex received extensive shock from capture. In neither case did we believe the animals had experienced e>,..1reme physical exertion nor aspiration of rumen contents. At time of capture, snow depths were about 25 cm and chase times appeared reasonable while ambient temperatures were 15 C and -2 C. Therefore, of 41 adult females captured and handled, 2 or 5% died of capture-induced injuries. Both adult elk were donated for human consumption. Subsequent to capture and radio-collaring, 1 male and 1 female calf died likely within 7 days of capture and were classified as capture-induced deaths and censored from the radioed-collared population of calves. Histopathology confirmed capturemyopathy in the male calf which had been killed by a mountain lion (Appendix B). At capture, rectal temperatures were 106.6 F (41.4 C) for the male and 105.5 F (40.8 C) for the female calves. Therefore, of 78 calves captured and handled, 2 or 2.6% died of capture-induced injuries resulting in a net sample of 76 radio-collared calves at the beginning of winter 2000. In 2001, we captured 80 calves, 40 males and 40 females, and 12 adult females age ::::_2 years. Frequency of age classes for adult females were: 2-4 years= 0, 5-9 years= 10, and >9 years= 2. We achieved our objectives of 26 calves of nearly balanced sex ratios in each DAU and maintained ::::_13 radio-collared adult females in each DAU (Table 1). Elk were captured at 19 new and 3 previously used sites (Figure 1, Appendix A). There were no acute deaths of adult females or calves during capture. However, 2 female calves caught from the same group in trap-zone E and radio-collared died within 48-hours of capture and were classified as capture-induced deaths (Appendix C). One calf became entangled in a fence about 1 km from the capture site and the second calf was euthanized by gun-shot about 0.5 km from the capture site because of obvious weakness following capture. At time of capture, snow depths were about 20 cm and capture chase times were < 1 minute at an ambient temperature of -4 C. Rectal temperatures of these calves at capture were 106.7 F (41.5 C) and 105.4 F (40.8 C), respectively. Two additional replacement female calves were captured from the same area prior to completing all capture activities. An additional male calf died within 3 days of capture and was classified as a capture-induced death even though a mountain lion had probably killed the calf (Appendix C). At capture, rectal temperature of-the male calf was 103.8 F (39.9 C), capture chase time seemed reasonable, and snow depths were about 20 cm. All 3 calves were censored from the collared population of calves. Therefore, of 80 calves captured and handled, 3 or 3.8% died of capture-induced injuries resulting in a net sample of 77 radio-collared calves at the beginning of winter 2001. For both years, capture-induced deaths occurred in 3.2% of the calves and 3.8% of the adult females that were captured and handled. Collar Failures We experienced pre-mature expansion of 14 calf collars, 13 males and 1 female, that resulted in collars slipping off calves causing us to censor calves during winter-spring or as yearlings during summer-fall time periods. For calves collared in December 2000, 5 males slipped collars between 30 April and 7 June 2001 and 3 males successfully recruited as yearlings, slipped their collars between 20 June and 20 July 2001. For calves collared in December 2001, 2 males slipped collars between 20 May and 3 June 2002, 3 males recruited as yearlings slipped their collars between 18 June and 17 July 2002, and 1 female recruited as a yearling, slipped her collar between 17 July and 22 August 2002. �199 Evidence suggested that amber latex tubing (3/8" O.D., 3/16" I.D, 3/32" wall thickness) used as a breakaway component to allow collar expansion pre-maturely deteriorated and broke allowing collars to expand and slip over the heads of calf and yearling elk. This calf collar design using the same or similar components had been previously used on 280 calves on the Grand Mesa, Colorado (Freddy 1997) where only I 5-month-old male calf, 1 13-month-old male yearling, and 2 23- to 26-month-old females slipped collars of which none were due to pre-mature breakage of latex tubing. On Grand Mesa, evidence indicated that latex tubing deteriorated as planned 10 to 18 months post-application after males had grown spike antlers or heads of either sex had grown to retard collars from slipping over heads. Deterioration of tubing apparently occurred sooner in the Gunnison Basin, maybe because of colder temperatures or slightly higher effective UV light levels, and males were much more prone than females to slip collars. We also speculated that antlers of yearling males in Gunnison might possibly be shorter than Grand Mesa yearling males during early summer thus allowing collars to slip more readily in Gunnison. After 2000-01, we changed brands oflatex tubing and maintained the same size of tubing for 2001-02 but the problem persisted to a lesser degree. In the future, we will change to a thicker latex tubing on male collars of either 1/8" or 3/16" wall thickness to reduce the problem of pre-mature expansion and the need to censor calves or yearlings from survival estimates but still maintain expansion capability so collars will adequately fit adult elk. We used radio-collar telemetry frequencies between 172 and 17 6 MHz and found an increasing problem with white-noise interference at frequencies> 175 MHz. At times, interference prevented hearing radiocollars except at relatively close distances, especially during aerial surveys when radio-collars or interference could be heard over several kilometers of distance. Interference was most commonly associated with human developments but likely sources could not be identified. We therefore caution project leaders to assess potential interference problems when selecting collar frequencies> 175 MHz. Weather Although official NOAA weather data has not been summarized as yet, winter-spring snow depths and summer-fall precipitation for both 2000-01 and 2001-02 were well below average for the entire Gunnison Basin. Both years were considered to represent drought conditions, not only for the Gunnison Basin, but most of southwestern Colorado. On most segments of winter range, snow depths generally did not exceed 30 cm during either winter with 2001-02 having shallower average snow depths than 2000-01. During both winters, snow had melted from primary winter ranges by late-March to mid-April. Snow depths and persistence of snow cover varied greatly in the Basin. Snow depths tended to decrease from west to east and north to south such that the deepest snow occurred in E-25 (GMU 66), E-4l(GMU 54), and E-43 (GMU 55) and the shallowest snow in E-25 (GMU 67) and in E-43 (GMU 551) (Figure I). Winter temperatures were generally mild for the Gunnison Basin with daily minimums seldom below -26 C and generally >-18 C and daily maximums often >-6 C. Calf Survival Survival rates of all calves pooled among 3 DAUs during winter-spring were 0.89 ± 0.08 (±CL, n = 71 ), 0.83 ± 0.09 (n = 75), and 0.86 ± 0.06 (n = 146) for 2000-01, 2001-02, and both years combined, respectively (Table 2). Survival of all calves was not different between years (P = 0.2965, Table 4). Male calves had lower survival (0.78) than female calves (0.97) in 2000-01 (P = 0.0105) but not in 200102 or when years were combined ( P2:._ 0.1463, Tables 2, 4). The greatest discrepancy between sexes occurred in DAU E-25 where survival of males and females was 0.71 and 0.93, respectively (Table 3). In comparison, yearly calf winter-spring survival on Grand Mesa was 0.86 to 0.92 (n = 69-73) and averaged 0.89 (n = 280) during 4 consecutive winters (1993-94 - 1996-97) with no differences in survival among years or between sexes (Freddy 1997). �200 Among the 3 DA Us, survival of all calves tended to be lower in E-41 (0.78) compared to E-25 (0.84) and E-43 (0.94) (P = 0.0737, Tables 3, 4). In paired comparisons between DA Us, calf survival was lower in E-41(0.78) than E-43 (0.94) (P = 0.0213, Tables 3, 4). During winter-spring, calves died due to predation, malnutrition, suspected predation or malnutrition, accidents, and of unknown causes. In 2000-01, 8 calves died with proximate causes of deaths being 37.5% predator-related and 62.5% malnutrition-related. In 2001-02, 13 calves died with proximate causes of death being 62% predator-related, 15% accidents, and 23% unknown causes. For years combined, 21 calves died with proximate causes of death being 53%predation-related, 24%malnutritionrelated, 9% accidents, and 14% unknown causes (Figure 2). On average then, for each 100 calves entering the population on 15 December, we would expect 86 to survive to the following 15 June with 7 deaths predation-related, 4 deaths malnutrition-related, and 3 deaths from other causes. Mountain lions and black bears predated elk calves and coyotes were suspected predators in one death. Accidental deaths were associated with a haystack collapsing and trapping a calf while elk were feeding on hay and a calf apparently slipped off a deep snow-trail used by elk and became trapped upside down among deadfall trees. In comparison, estimated causes of calf mortalities (n = 31) on Grand Mesa were 65% predation-related, 26% malnutrition-related, and 9% of unknown causes (Freddy 1997). Calf mortalities tended to occur later and primarily after 16 March in the winter-spring of 2000-01 than in 2001-02 (Figure 3, A & B). Timing of deaths in 2000-01 was consistent with deaths directly associated with malnutrition or predation-related deaths of malnourished calves as winter progressed. In contrast, predation-related deaths occurred from January through May in 2001-02 with no deaths directly attributed to malnutrition in 2001-02 (Figures 2, 3). In comparison, calves died on Grand Mesa from January into late May but the majority died in March and April (Freddy 1997). Femur marrow fat of dead calves was 34% ± 25 (SD, n = 8) in 2000-01 and marginally lower (P = 0.11, t-test) than the 59% ± 35 (SD, n = 10) in 2001-02. In general, most calves dying from any cause had femur fat< 50% in 2000-01 and >55% in 2001-02 (Figure 4). For deaths attributed directly to predation, femur fat averaged 16% (n = 3) in 2000-01 and 91 % (n = 4) in 2001-02. In 2001-02, all deaths considered predation-related had femur fat >60% (n = 7), even deaths occurring in mid-May (Figure 4). In contrast, both accidental deaths in 2001-02 had femur fat <5% and likely represented calves already extremely malnourished prior to the end of February (Table 5). Importantly, Cook et al. (2000a) noted that femur fat values <85% in adult female elk were associated with total percent body fat< 5% indicating that nearly any loss of femur fat suggested an animal in poor physical condition. Although there were limited sample sizes both years, data suggested a different dynamic between years of mild winters. in 2000-01, calf survival appeared more infl_uend~d bynutrition andrela.tive.body condition, possibly representing either the previous summer or winter forage production. Predation appeared more compensatory related. In 2001-02, predation appeared more additive than compensatory because calves that died were possibly not predisposed by malnutrition. In both.years, overall calf survival remained high regardless of the proximate cause of mortality. We must also caution that predation-related deaths in early January 2002 could not be totally separated from possible captureinduced deaths as deaths likely occurred <2 weeks post-capture. Adult Survival Survival rates for adult females age ;::2 years were 1.00 during winter-spring as no deaths occurred during 2000-0l(n = 39) and 2001-02 (n = 48). During summer-fall 2001, survival was 0.92 ± 0.08 (n = 39) when natural (1) and hunting deaths (2) were included. The one natural death occurred about July 1 of unknown causes in a female age 19 years based on dental cementum resulting in a natural summer-fall survival rate of 0.97 ± 0.05 (n = 37) (Table 6, Appendix D). Annual survival rates were 0.92 ± 0.08 (n = 39) including all causes of death and 0.97 ± 0.05 (n = 37) including only natural deaths (Table 6). In �201 comparison, natural survival of adult females was ~0.97 in winter-spring and summer-fall during 7 consecutive years on Grand Mesa (1993-94 - 1999-2000) (Freddy 2000). Yearling Survival Survival rate for yearling female elk, age 12-17 months, was 0.89 ± 0.10 (n = 38) during summer-fall 2001 and because all 4 deaths were hunting-related, natural survival during summer-fall was 1.00 (n = 34). Survival of the same cohort of yearling males was 0.86 ± 0.15 (n = 22) during summer-fall when natural (2) and hunting deaths (1) were included. Two yearling males died in July 2001 of suspected predation (Appendix D) resulting in a natural summer-fall survival rate of 0.90 ± 0.13 (n = 21) (Table 7). Survival of all females age ~12 months during summer-fall was 0.91 ± 0.07 (n = 77) inclusive of natural (1) and hunting deaths (6) and natural survival for these females was 0.99 ± 0.03 (n = 71, Table 6). Survival rates for both yearling female and male elk, age 18-23 months, was 1.00 (n = 34 F, n = 19 M) during winter-spring as no deaths occurred during 2001-02 (Table 7). Survival of all females age ~18 months during winter-spring 2001-02 was 1.00 (n = 82) as no deaths occurred (Table 6). Harvest Removal Harvest removal rates (r) during summer-fall 2001 were 0.05 for adult females, 0.11 for yearling females, 0.08 for all adult females age ~12 months, and 0.06 for yearling males. Hunting mortalities for adult females consisted of 3 legally harvested (2 regular rifle, 1 late rifle) and 3 wounding losses (1 archery/muzzleloading, 2 regular rifle). Wounding loss thus equaled the legal harvest in this small sample situation. The one yearling male hunting mortality represented an illegal harvest that occurred during a late-season in December 2001 (Appendix D). With observed calf and adult female natural survival rates, computer models suggest that removal rates for adult females age ~12 months in the Gunnison Basin would need to be ~15% per year to stabilize the population. Calf Body Size Calf body mass averaged 99.1 ±2.2 kg and ranged from 52.0 to 133.0 kg for all calves and years (Table 8) with 7% of the calves having mass <80 kg (Figure 5). There were no effects of capture year, calf sex, or year-sex interaction on body mass (P.:::_ 0.259) although calves were 2.6 kg smaller in mass in 2001 and males were 1 kg larger than females. However, calf mass was different among management DAUs (P = 0.003). In simultaneous paired comparisons, calves were larger in E-41 (104.4 kg) than in E-43 (96.8 kg) and E-25 (95.9kg) with no differences between E-43 and E-25. Calf mass was reasonably consistent within trapzones within DAUs, with mass tending to be larger in trapzones I and J (E-41) than in D (E-25) and F (E-43) (P = 0.090). Similar trends among years, sex, DAUs and trapzones occurred for calf total body length and hind-foot measurements with both measurements supporting that E-41 calves were largest (P < 0.002) (Table 8). Calf mortalities occurred across the range of calf body mass classes (Figure 5). Predation-related mortalities occurred in the most :frequent mass classes between 80 and 119 kg, suggesting predators were taking calves with no particular selectivity. Except in one case, malnutrition-related mortalities occurred in calves< 99 kg in size. Calves <80 kg did not necessarily perish, although 2 of the 3 calves <60 kg died of malnutrition or accident (Figure 5). Survival of calves tended to be lower in E-41 (P = 0.0737), where calf mass was largest, compared to survival in E-25 and E-43 (Tables 3, 4, 8). In E-41, mortalities were predator-related (45%), malnutrition-related (36%, including 1 accident where calf femur marrow was< 2%), and unknown (18%). On Grand Mesa, the larger mass of male calves also did not necessarily translate to higher survival rates compared to smaller female calves (Freddy 1997). In the Gunnison Basin, male and female calves were 15% and 7% smaller, respectively than their counterparts captured and radio-collared on Grand Mesa 1993-94 - 1996-97. On Grand Mesa, body mass was 115 ±2.5 kg for males (n = 138) ·and 106 ± 2.3 kg for females (n = 136) with an overall range in size �202 of 60 to 141 kg (Freddy 1997). Furthermore, unlike Gunnison, males were significantly larger (8%) than females on the Grand Mesa. Unfortunately, we cannot distinguish whether differences in mass reflect population or year effects. Elk Reproduction Samples Participation by successful hunters in providing biological samples from adult female elk harvested during late-seasons was disappointing. Of the estimated 665 adult females harvested during 2000-2001, we received some of the requested biological samples from 19% of the elk with reproductive organ and kidney fat samples representing only 13% of the elk. hnportantly, rates of participation by hunters harvesting elk declined from 28 to 12% from 2000 to 2001 despite attempts to improve collection instructions and packets sent to hunters in 2001 (Table 9). Providing an incisor tooth from harvested elk was the most common biological sample collected by hunters. For many data summaries and analyses, reproductive data from both years was combined because 65% of the samples were obtained in 2000. Furthermore, approximately 80% of the samples came from elk harvested in GMU 66 and 20% from elk harvested in GMUs 41 and 55. Therefore, data summaries were inherently weighted to year 2000 and GMU66. Ages of adult female elk harvested ranged from 1 to 20 years with 63% estimated to be age 3 to 10 years. Yearlings (:::::age 17-18 months) and females age 215 each represented 5% of the harvest. Because hunter selectivity and animal behavior may bias vulnerability of different elk age classes to harvest, the distribution of harvested age classes may biasly represent the age structure of the elk population (Table 10). Pregnancy Rates and Conception Dates Pregnancy rate for all adult females age 21 year was 85% (n = 89, Table 11). Pregnancy rate was 92100% for female age classes 3 to 14 years. Pregnancy rate was 67% in females age 2 and 50% in females age 215. Pregnancy rates across age classes were highly similar to rates measured in the ForbesTrinchera elk population of south-central Colorado (Freddy 1993b). The 100% pregnancy rate in yearlings could be questionable because pregnancy status was unknown in 67% of the submitted yearling samples. For age classes 2 and 3-4, pregnancy status was unknown in up to 44% of the animals. We might expect that in these younger age classes, uteri may be small, in-active and non-pregnant or in the early stages of pregnancy, creating more difficult circumstances for hunters to find and collect specimens. In comparison, for age classes 2 age 5, :::_27% of the specimens were of unknown pregnancy status. Thus, there is the possibility that pregnancy rates for young elk.age 1 to 4 could be ove_restimated due to collections biased against non-pregnant elk. Estimated conception dates followed an expected asymmetrical pattern (Flook 1970, Freddy 1993b,Noyes et al. 1996). Mode, median, and mean days of conception were 23, 26, and 29 September, respectively (n = 72, Figure 6). Conceptions spanned 68 days with 75% occurring in the 26-day interval from 8 September to 3 October. This conception pattern strongly suggested that most adult females conceived during their first estrus cycle at the expected time of year. Females conceiving after 10 October (n = 12, 17%) may have had a delayed first estrus or conceived during their second estrus cycle. Patterns and dates of conception were quite similar to estimates obtained for Forbes-Trinchera elk where post-season mature bull:cow rations commonly exceeded 35:100 (Freddy 1993b). The mirror-image asymmetrical distributions for conception dates (Figure 6) and calf body mass (Figure 5) indirectly suggest that smaller calves (7% < 80 kg) may be associated :with adult females conceiving later in the fall (17% after 10 October). Females conceiving after 10 October were comprised of 9% yearlings, 18% age 3-4, 45% age 5-7, and 27% age 215 years. All pregnant females 2 age 15 conceived after 16 October (n= 3). Later breeding by youngest and oldest age classes would not be unexpected but later conception by females age 5-7 may �203 indicate some nutritional or disturbance stress affecting timing of breeding in about 8%ofthe population. Overall, conception date was not dependent on adult female age (r2 = 0.014, P = 0.173, n = 66). Pregnancy status was associated with ovarian mass of both ovaries combined. Total ovarian mass (g) was 5.37 ± 0.28 (CL, n = 59) and larger in pregnant than the 3.36 ± 1.60 (n = 5) in non-pregnant elk (P = 0.025). Larger ovarian mass reflected the presence of active corpora lutea of pregnancy. Fetal Rates, Sex. Age, and Size Litter size was 1 fetus in uteri with detectable fetuses (n = 69). Fetal sex favored females (37F:21M) for years combined which differed from a 50:50 ratio (i = 4.414, df = I, P = 0.036). Female fetuses also dominated within each year (24F: 15M, 2000; 13F:6M, 2001) but such yearly ratios were not different from 50:50 (P 2: 0.108). Fetal sex could be determined in those fetuses near 2:70 mm crown-rump length based on external genitalia as also found by Morrison et al. (1959) and Kohlmann (1999). In general, fetal sex could not be determined in fetuses estimated to have been conceived after 10 October. Fetal sex tended to be dependently associated with adult female grouped age class (Figure 7 [Right], i = 10.885, df = 5, P = 0.054). Male fetuses predominated in adult females age 8-10 while females fetuses were most common within age classes 3-4, 5-7, and 11-14. Fetal sex ratio was equal in 2-year-old females which may have been conceiving for the first time. Male fetuses were also more predominate in elk age 2:8 years in the Forbes-Trinchera elk (Freddy 1993b). Kohlmann (1999) found male fetuses were more common in adult females having high kidney fat levels, and thus good body condition, and that adult females in good body condition conceived earlier in the breeding season. We could speculate that adult females age 8-10 had male fetuses because they were in better body condition at conception than other age classes due to their age and inherent larger body size that allowed them to withstand the rigors of a previous pregnancy and calf rearing and maintain access to better matriarchal habitats (CluttonBrock et al. 1982). In 2-year-old females, male fetuses may have been more common because these females likely had not gone through a previous pregnancy and subsequent calf-rearing prior to conception and thus were in better body condition. Although male fetuses predominated in those adult females conceiving 16-20 September just prior to the peak of conception, a pattern favoring male fetuses during early conceptions was not clearly evident (Figure 7 [LEFT]). Estimated fetal age averaged 69 and 76 days in 2000 and 200land was not different between years (P = 0.139). We found that fetal age predictive equation (b) (see ME1HODS) provided estimated dates of conception that occurred about 3-days earlier than equation (a) (paired t-test, P < 0.001). Fetal size compared favorably with fetuses measured in the Forbes-Trinchera elk population and appeared to be within an acceptable range of weight and skeletal.dimensions (Freddy 1993b). There was a general pattern of fetuses in 2001 being slightly larger in crown-rump (P < 0.080), hind-leg (P < 0.033), and hind-foot ( P < 0.070) dimensions but not in body mass (P < 0.138) (Table 12). Fetal size is highly dependent on date of collection so absolute comparisons between years or among elk populations must include corrections for date of collection. Body Fat Condition and Reproduction Total fat kidney fat index values (TF-KFI) for adult females age 2:1 year averaged 106 and ranged from 29 to 306 (n = 84) which were similar to values for Oregon elk (Kohlmann 1999) (Table 13, Figure 8[LEFT]). The 25% quantile value was 63 which was slightly higher than the 50 reported by Kohlmann ( 1999). Other kidney fat values in Table 13 were presented for reference as these measurements were the basis for estimating percent total body fat (Cook et al. 200 la), or body condition, in adult female elk. TF-KFI for calves averaged 44 (n = 6). �204 Estimates of percent body fat for all adult females age 2::,1 year based on the 3 kidney fat measurements, TF-KFI, TRF-KFL and TFM, averaged between 11.l and 12.1 % with a combined range of 5.4 to 18.5% (Table 14, Figure 8 [RIGHT]). Percent body fat of all adult females tended to be 1-2% higher in 2001 than 2000 for all estimators of body fat (PS 0.073, Table 14). For yearling females, specifically, percent body fat estimates averaged 10.7 to 13.0% with minimum-maximums of 9 and 14% (n = 4) while body fat in calves was 4-7% (see Table 14 cautionary foot-note). Based on estimates of percent total body fat, about 65% of the adult females age 2::,1 year were in moderate, 30% in low, and <5% in very low or very good body condition, and 0% in excellent condition (Figure 8 [RIGHI']). Relative condition class ratings were very low= <7% body fat, low= 7-10%, moderate= 10-15%, very good= 15-20%, and excellent= 20-25% (Cook, J. G. 2001 unpublished data). Kidney fat measurements provide the best predictive accuracy of percent total body fat at moderate levels of body condition and less accuracy at very high or very low levels of body condition (Cook et al. 2001a) so that our ability to detect outliers in body condition status may have been limited by measuring only kidney fat. Furthermore, percent total body fat levels in the excellent category may be rarely found in wild elk as these values were associated optimum nutrition in captive elk and likely represent the physiological maximums attainable by elk (Cook J. G., 2001 unpublished data). Probability of an adult female elk age 2::,1 year being pregnant was dependent on estimated percent total body fat when body fat was based on TFM (g): (logit (Pregnancy))= -2.2835 + 0.3704*(X-percent body fat); P = 0.033, n = 68). Pregnancy probability was predicted to be 2::,0.90 when percent total body fat was 2::,12%, or 2: moderate body condition (Figure 9). Similar dependent relationships (logit P) could not be detected between pregnancy status and percent total body fat based on TF-KFI (P = 0.130) and TRFKFI (P = 0.099), or on direct TF-KFI (P = 0.206) values. Cook et al. (2001a) indicated that TFM was the superior predictor of body condition within the kidney fat measurement alternatives. Kohlmann ( 1999), however, did find that probability of pregnancy (lo git P) increased with increasing TF-KFI values in Oregon elk (n = I 152). Similarly, Cook et al. (2001c) documented that low quality nutrition prior to breeding prevented or delayed conception in adult female elk and, furthermore, high pregnancy rates could be associated with marginally deficient nutritional conditions. Standard ANOVA results were consistent with logistic regression results. Estimated percent body fat was higher for pregnant than non-pregnant elk for body fat estimates based on TFM (P = 0.038) but not for body fat estimates based on TF-KFI and TRF-KFI (P 2: 0.112, Table 14 ). All estimates of percent body fat were not different among pregnant, non-pregnant and pregnancy status-unknown adult females (P 2: 0 .149, Table 14 ). Furthermore, using linear regression, all estimates of percent body fat were not dependent on adult female age 'Yithin pregnant (r S 0.014, P 2: 0.315, n = 54) or non-pregnant elk (r S 0.008, P 2: 0.543, n = 8). • • Probability of conceiving before or after the median date of conception was not dependent on percent total body fat based on TFM: ( logit (Before) P =0.221, n = 53), indicating there was no detectable increased probability to conceive before the median date based on percent total body fat. Using linear regression, conception date was not dependent on percent body fat based on TF-KFI, TRF-KFI, or TFM (r S 0.025, P > 0.141, df= 49) or dependent on 2-variable combinations of adult female dental cementum age and percent body fat (R2 S 0.031, P > 0.178, df= 49). Estimates of body fat for adult females conceiving after IO October was about 11.1 % for all 3 estimates of body fat. Overall, measured population performance of elk in the Gunnison Basin generally followed the predictions of proposed performance models for adult female elk in moderate or low body fat condition (Band C, below; Cook J.G. 2001 unpublished data). �205 The measured performance of elk in the Gunnison Basin could be summarized as: Pregnancy rates were 85%, about 17% of the females conceived after 10 October, adult female survival during mild winters was 100%, average mass of 6-month old calves was 99 kg with 7% of the calves weighing <80 kg and 21 % weighing> 110 kg, and survival of calves during mild winters was >83%; and compared to: Model A: If adult females in very good body fat condition. then we should expect: Pregnancy rates >90%, significant early breeding, high adult survival in harsh winters, > 110 kg calves in November; <5% of adult female Gunnison elk were classified in very good body fat condition. Model B: If adult females in moderate body fat condition, then we should expect: Pregnancy rates ~90%, some delayed breeding, with high adult winter survival depressed somewhat in harsh winters, 90-110 kg calves in November; 65% ofadult female Gunnison elk were classified in moderate body fat condition. Model C: If adult females in low body fat condition, then we should expect: Pregnancy rates ~70-90%, more delayed breeding, with markedly lower adult winter survival in harsh winters, 70-100 kg calves in November; 30% of adult female Gunnison elk were classified in low body fat condition. Model D: If adult females in very low body fat condition, then we should expect: Pregnancy rates 90%, delayed breeding up to 6 weeks, with low adult winter survival in harsh winters, 6090 kg calves in November; <5% of adult female Gunnison elk were classified in very low body fat condition. General Movements of Elk Insights into the general distribution and movements of elk in Gunnison Basin DA Us (see PN Figure 1,) were obtained from approximate locations of radio-collared elk obtained during 44 aerial survey flights between December 2000 and June 2002. Maps of elk distribution are in process so only verbal descriptions will be presented at this time. Elk wintered in segments of winter range near where they were trapped in December as elk did not usually make large movements during winter. Movement from winter areas towards summer ranges began in April, proceeded in earnest in mid-May after snow had melted at higher elevations, and ended with elk arriving.on highest.elevation summer ranges in July after cc!lving and subsequent to snow melting on alpine ranges. Movements from summer to winter ranges began in early September and continued through November with rates of movement most likely affected by hunting season activities and increasing snow depths. Elk essentially did not cross U.S. Highway 50 (Hy50) which separated the south (trap-zones A-E) and north (trap-zones F-J) (Figure 1) portions of the Gunnison Basin. Only 2 elk were known to cross this highway: an adult female in September 2001 moved from the Tomichi Dome area (trap-zone F) southwest to Sawtooth Mountain; and, a 24-month-old male in June 2002 moved from Tomichi Dome area (trap-zone F) to the southeast onto Sargents Mesa and then proceeded south over the La Garita Mountains to Alder Creek in the Rio Grande River drainage near South Fork, Colorado. Elk did move beyond the boundaries of the Gunnison Basin during winter and summer but only 2, at this time, have likely dispersed from the Gunnison Basin; an 18-month-old female moved from trap-zone H �206 north to Paonia Reservoir in November 200 I and a 24-month-old male moved trap-zone F south into the Rio Grande River drainage during June 2002. During summer, radioed elk were commonly found along the higher elevation divides associated with the boundaries of the Gunnison Basin DAUs, often in subalpine or alpine habitats from which they vacated in September while moving towards their winter ranges within the Gunnison Basin. These boundary areas included: upper and lower Cimarron and Little Cimarron rivers (west of trap-zone A); upper Rio Grande river in Rat Creek and near Continental and Rio Grande reservoirs (south of trap-zones A, B, C); Saguache Park (east of trap-zone D); Sargents MesaCameron Park (east of trap-zone E); upper Chalk creeks (northeast trap-zone F); upper forks of North Fork of Cottonwood, Lake Fork, Clear, and Castle creeks (east-northeast of trap-zone G); Anthracite creeks, Snowshoe and Cliff creeks, Coal Creek Basin, and Willow and Minnesota creeks (north oftrapzones H, I, J); and, upper Smith Fork, Dyer, and Crystal creeks (west-northwest of trap-zone J). The greatest overlap of Gunnison Basin elk with elk from other management DAUs occurred during summer in the Big Blue Wilderness (west trap-zone A), in the upper Rio Grande River near Slumgullion-Spring Creek Pass and west of Continental Reservoir (south trap-zones A, B, C), and in the West Elk Wilderness (north trap-zones I, J). During winter, only a few elk remained outside of the Gunnison Basin DA Us, mainly in lower Cimarron creeks (west trap-zone A), just east of North Pass and Old Cochetopa passes (east of trap-zone E), near Paonia Reservoir (north trap-zone I, J), and in Smith Fork and Doug creeks near Crawford, Colorado (west trap-zone J). Some consideration should be given to re-aligning elk management DA Us in the Gunnison Basin based on observed movements of elk. There was a continuum of elk interchange among trap-zones on an west to east basis, especially during summer and to a lessor degree in winter. South ofHy50, elk in trap-zones A through E interacted with elk in adjacent trap-zones such that there was no clear demarcation of separate elk sub-populations across this area (Figure 1). Similarly, north ofHy50, elk in trap-zones J through F interacted with elk in adjacent trap-zones such that there was no clear demarcation of separate elk sub-populations across this area, although elk in trap-zones G and F only interacted with elk from trap-zone Hin areas near Gothic, Colorado in the upper East and Slate rivers (Figure 1). Therefore, all areas south ofHy50 from Monarch pass on the east to the Cimarron Divide on the west (GMUs-part 551, 67, 66, and adding 65) could be treated as one DAU. Likewise, all areas north ofHy50 from Monarch Pass on the east to at least the Curecanti divide on the west (GMUs- part 551, 55, 54, and potentially adding 53 and 63) could be treated as one DAU. Elk that winter in GMUs 53 and 63 to the northwest of the Gunnison Basin likely have high interchange with elk from GMU 54 during summer in the West Elk Wilderness. SUMMARY In the Gunnison Basin, Colorado during winter-spring 2000-01 and 2001-02, survival rates of calves averaged 83-89% and tended to vary among elk management DAUs while survival rates of all age classes of adults were 100%. During summer-fall, survival rates were ;:::97% for adult females, 100% for yearling females, and 90% for yearling males when hunting deaths were excluded. Survival rates were comparable to survival rates previously estimated for elk inhabiting the Grand Mesa, Colorado. Harvest removal rates during summer-fall 2001 were 5% for adult females, 11 % for yearling females, 8% for all adult females age ;:::12 months, and 6% for yearling males. Measures ofreproductive and survival parameters were consistent with predictions of performance outcomes for adult female elk having low to moderate body condition status during fall. More than likely, marginally deficient levels of seasonal nutrition were depressing optimal reproductive performance of adult female elk. Consideration should be given to re-aligning management DA Us with observed distribution and movements of radioed elk. �207 LITERATIJRE CITED Anderson, A.E., D.C. Bowden, and D.E. Medin. 1990. Indexing the annual fat cycle in a mule deer population. Journal of Wildlife Managment 54:550-556. Anderson, C.R., Jr., D.S. Moody, B.L. Smith, F.G. Lindzey, and R. P Lanka. 1998. Development and evaluation of sightability models for summer elk surveys. Journal of Wildlife Management 62:9933991055-1066. Armstrong, R.A. 1950. Fetal development of northern white-tailed deer. American Midland Naturalist 43:650-666. Bartholow, J. 1999. POP-II system documentation Windows™ Version 1.0. Fossil Creek Software, Fort Collins, Colorado USA. Bear, G.D., G.C. White, L.H. Carpenter, RB. Gill, and DJ. Essex. 1989. Evaluation of aerial markresighting estimates of elk populations. Journal of Wildlife Management 53:908-915. Carpenter, L.H., R. B. Gill, D. L. Baker, and N.T. Hobbs. 1984. Colorado's big game supplemental winter feeding program. Colorado Division of Wildlife, Fort Collins, Colorado USA. Cogan, RD., and D.R. Diefenbach. 1998. Effect ofundercounting and model selection on a sightabilityadjustment estimator for elk. Journal of Wildlife Management 62:269-279. Cook, R..C., J.G. Cook, D.L. Murray, P. Zager, B.K. Johnson, and M.W. Gratson. 2001a. Development of predictive models of nutritional condition for Rocky Mountain elk. Journal of Wildlife Management 65: 973-987. Cook, R..C., J.G. Cook, D.L. Murray, P. Zager, B.K. Johnson, and M.W. Gratson. 2001b. Nutritional condition models for elk: which are the most sensitive, accurate, and precise? Journal of Wildlife Management 65: 988-997. Cook, RC., D.L. Murray, J.G. Cook, P. Zager, and S.L. Monfort. 2001c. Nutritional influences on breeding dynamics in elk. Canadian Journal of Zoology 79:845-853. Clutton-Brock, T.H., F.E. Guinness, and S.D. Alban. 1982. Red deer behavior and ecology of two sexes. University of Chicago Press, Chicago, Illinois, USA. Eberhardt, L.L., RA. Garrott, PJ. White, and PJ. Gogan. 1998. Alternative approaches to aerial censusing of elk. Journal of Wildlife Management 62:1046-1055. Flook, D.R. 1970. Causes and implications of observed sex differential in the survival of wapiti. Canadian Wildlife Service Report Series 11. Ottawa, Ontario, Canada. Freddy, DJ. 1992. Effect of elk harvest systems on elk breeding biology. Colorado Division of Wildlife Research Report July:45-70. Fort Collins, Colorado, USA. July:45-70. Freddy, DJ. I 993. Program Narrative for Estimating survival rates of elk and developing techniques to estimate population size. Colorado Division of Wildlife Research Report July:83-117. Fort Collins, Colorado, USA. Freddy, DJ. 1993b. Effects of elk harvest systems on elk breeding biology. Colorado Division of Wildlife Research Report July:50-65. Fort Collins, Colorado, USA. Freddy, D.J. 1997. Estimating survival rates of elk and developing techniques to estimate population size. Colorado Division of Wildlife Research Report July:47-73. Fort Collins, Colorado, USA. Freddy, DJ. 1998. Estimating survival rates of elk and developing techniques to estimate population size. Colorado Division of Wildlife Research Report July: 177-206. Fort Collins, Colorado, USA. Freddy, DJ. 2000. Estimating survival rates of elk and developing techniques to estimate population size. Colorado Division of Wildlife Research Report July (2):239-258. Fort Collins, Colorado, USA. Freddy, DJ., D.L. Baker, RM. Bartmann, and RC. Kufeld. 1993. Deer and elk management analysis guide, 1992-1994. Colorado Division of Wildlife Division Report 17, Fort Collins, Colorado USA. �208 Haigh, J.C., and R.J. Hudson. 1993. Fanning wapiti and red deer. Mosby, Saint Louis, Missouri, USA, pages 187-192. Kohlmann, S.G. 1999. Adaptive fetal sex allocation in elk: evidence and implications. Journal of Wildlife Managment 63: 1109-1117. Lewis, R.J., G.A. Chalmers, M.W. Barrett, and R Bhatnagar. 1977. Capture myopathy in elk in Alberta, Canada: a report of three cases. Journal of The American Veterinary Medical Association 171:927-932. Morrison, J.A., C.E. Trainer, and P.L. Wright. 1959. Breeding seasons in elk as determined from known-age embryos. Journal of Wildlife Management 23:27-34. Nelson, L.J ., and J.M. Peek. 1982. Effect of survival and fecundity on rate of increase of elk. Journal of Wildlife Management 46:535-540. Noyes, J.H., B.K. Johnson, L.D. Bryant, S.L. Findholt, and J.W. Thomas. 1996. Effects of bull age on conception dates and pregnancy rates of cow elk. Journal of Wildlife Management 60:508-517. Roath, R., L. Carpenter, B. Riebsame, and D. Swift. 1999. Gunnison Basin habitat assessment projectFinal Report March 1999. Colorado State University, Department of Range Science, Fort Collins, Colorado USA. Riney, T. 1955. Evaluating condition of free ranging red deer (Cervus elaphus), with special reference to New Zealand. New Zealand Journal of Sciene and Technology 36:429-463. Samuel, M.D., E.O. Garton, M.W. Schlegel, and R G. Carson. 1987. Visibility bias during aerial surveys of elk in northcentral Idaho. Journal of Wildlife Management 51: 622-630. SAS. 1988. SAS/STAT user's guide, release 6.03. SAS Institute, Inc. Cary, North Carolina USA. Spraker, T.R. 1982. An overview of the pathophysiology of capture myopathy and related conditions that occur at the time of capture of wild animals. Pages 82-118 in L. Nielsen, J.C. Haigh, and M.E. Fowler, editors. Chemical immobilization of North American Wildlife. Wisconsin Humane Society, Inc., Milwaukee, Wisconsin, USA. Unsworth, J.W., L. Kuck, and E.O. Garton. 1990. Elk sightability model validation at the National Bison Range, Montana. Wildlife Society Bulletin 18:113-115. Wade, D.A., and J.E. Browns. 1982. Procedures for evaluating predation on livestock and wildlife. Texas-Agricultural Experiment Station Publication B-1429. White, G.C. 1991. DEAMAN database manager and population modeling procedures; Colorado Division of Wildlife User's manual and reference. Colorado State University, Fort Collins, Colorado USA. White, G.C. and RA. Garrott. 1990. Analysis of wildlife radio-tracking data. Academic Press, Inc., San Diego, California USA �209 Table 1. Number of male (M) and female (F) calf and adult female elk radio-collared in each DAU and tra:e-zone in the Gunnison Basin, December 2000 and 2001. Calf Elk Collared 2000 Adult Female Elk 2001 2000-01 DAU-(GMUs) Trapzone M F Total M F Total M F Total E-25 (66, 67) A- Lake Fork 4 7 11 2 4 6 6 11 17 5 B-Cebolla 3 4 7 2 0 2 5 4 9 3 2 2 2000 2001 6 16 7 5 10 4 5 8 14 3 5 7 19 39 32 3 5 7 4 5 6 6 12 10 7 17 3 2 3 2 4 6 4 5 9 2 ll 26 ll ll 26 ll 27 52 ll 2 3 5• 6 2 7 9 2 6 7 4 11 3 0 3 8 6 8 2 10 26 18 1 16 39 ll 5 10 7 14 7 15 ll E- Razor 4 G-Almont 5 3 3 11 ~ 7 11 18 5 16 18 16 34 12 14 26 ll ll 28 27 27 54 ll H-Flat Top 2 4 6 3 3 6 5 7 12 4 l- Beaver 4 6 10 3 4 7 7 10 17 4 0 4 J - West Elk 7 3 10 7 6 13 14 9 23 5 2 7 15 ll ll 26 ll ll 26 26 26 52 ll 1 li 39 29 38 40 78 40 80 78 80 .ill 39 12 51 ill 92 subtotals E-43 subtotals E-41 Totals Collared 2000-01 D- Sawtooth F-Tomichi E-41 (54) 2000 2001 C-Huntsman subtotals E-25 E-43 (55, 551) Total Elk Collared All Subtotals 40 • Includes 2 female calves that died of capture-induced causes with~- 24 hours of capture for which 2 additional female calves were captured from the same area and radio-collared pnor to completing capture o all elk. The net begmning sample size was therefore I I female calves for estimating survival rates in DAU E-43 in 2001. Table 2. Survival rates of elk calves age 6-11 months for males, females, and sexes combined from 15 December to 14 June in the Gunnison Basin, Colorado, 2000, 2001, and years pooled. Binomial estimator used to calculate survival rates and confidence limits for calves combined among DAUs E-25, E-41, and E-43. Elk Calves Elk Calves 15 Dec 2000- 14 June 2001 Survival Rate All Elk Calves 15 Dec 2001 - 14 June 2002 15 Dec - 14 June 2000 - 2002 Males Females All Males Females All Males Females All 0.78 0.97 0.89 0.84 0.82 0.83 0.81 0.90 0.86 Lower 95%CL 0.63 0.92 0.81 0.71 0.69 0.74 0.72 0.83 0.80 Upper 95%CL 0.93 1.00 0.96 0.96 0.94 0.91 0.91 0.97 0.91 n Collars 32 39 71 37 38 75 69 77 146 Collars Deployed 38 40 78 40 40 80 78 80 158 Collars Censored 6· lb 7 3' 2d 5 9 3 12 Died 7 8 6 7 13 13 8 21 Non-hunting Deaths 7 8 6 7 13 13 8 21 Hunting Deaths 0 0 0 0 0 0 0 0 0 • Male calves censored: for post-capture induced mortality 173.082/00 on 12/29/00 • for slipped-collars, 173.269/00 on 4/30/01, 173.170/00 on 5/7/01, 173.250/00 on 5/25/01, 173.151/00 on 6/7/01, and 173.220/00 on 6/1 /01. Female calves censored: for post-capture induced mortality 172.379/00 on 12/26/00. ' Male calves censored: for post-capture induced. mortality 174.720/01 on 12/19/0 I; for slipped-collars, 174.099/01 on 5/20/02, and 175.221/0 I on 6/3/02. b d Female calves censored: for post-capture induced mortality 173.429/01 on 12/16/01 and 173.740/01 on 12/16/01. �210 Table 3. Survival rates of elk calves age 6-11 months for males, females, and sexes combined from 15 December to 14 JuneforDAUs E-25, E-41, and E-43 in the Gunnison Basin, Colorado, 2000-01 and 2001-02 combined. Binomial estimator used to calculate survival rates and confidence limts. Elk Calves - DAU E-25 Elk Calves - DAU E-43 Elk Calves - DAU E-41 15 Dec - 14 June 00-01, 01-02 15 Dec - 14 June 00-01, 01-02 15 Dec - 14 June 00-01, 01-02 Males Females All Males Females All Males Females All 0.71 0.93 0.84 0.92 0.96 0.94 0.77 0.80 0.78 Lower 95%CL 0.47 0.82 0.73 0.82 0.88 0.87 0.60 0.63 0.67 Upper 95%CL • 0.94 1.00 0.95 1.00 1.00 1.00 0.94 0.97 0.90 n Collars Collars Deployed 17 25 27 27 44 52 26 27 25 27 51 54 26 26 25 26 51 52 Collars Censored 8 0 8 I 2 3 0 1 1 Died 5 2 7 2 1 3 6 5 11 Non-hunting Deaths 5 2 7 2 1 3 6 5 11 Huntin~ Deaths 0 0 0 0 0 0 0 0 0 Survival Rate Table 4. Comparisons of calf survival between years, sexes, and among DAUs in the Gunnison Basin, Colorado, 2000-01 through 2001-02 based upon chi-square (i ) contingency tests. t Calf Survival Rate Comparisons 1.2 Likelihood Ratio Ratio i Probability df 0.2965 1.10 0.2942 1 0.1463 2.12 0.1456 1 0.0105 0.8009 7.07 0.0078 1 0.06 0.8008 1 0.0737 5.71 0.0577 2 Value Probability Calf Sexes & DAUs Pooled 2000-01 vs 2001-02 1.09 Calf Male vs. Female All Years & DAUs pooled 2.11 Calf Male vs. Female in 2000-01 with DAUs pooled 6.56 Calf Male vs. Female in 2001-02 with DAUs pooled 0.06 5.21 Calf Sexes Pooled DAU E-25 vs E-43 vs E-41 & Years Pooled Calf Sexes Pooled DAU E-25 vs E-43 & Years Pooled 2.52 0.1123 2.56 0.1098 1 Calf Sexes Pooled DAU E-25 vs E-41 & Years Pooled 0.49 0.4827 0.50 0.4808 1 Calf Sexes Pooled DAU E-41 vs E-43 & Years Pooled 5.30 0.0213 5.59 0.0181 1 Table 5. Percent (enmr marrow fat in e1k calv.es_dying from estimated causes of mortality during winter-spring 15 .. . December to 14 June, 2000--Gl and 2001-02 in the Gunnison Basin,·Colora~o. - Samples Average Fat n- samples SD SE Min Max Predation Malnutrition 12/15/00- 12/15/0106/14/01 06/14/02 90.7 15.8 45.3 94.4 13.2 83.6 78.5 24.8 86.8 3.0 4.0 17.8 5.5 2.7 10.3 13.2 83.6 45.3 94.4 12/15/00- 12/15/0106/14/01 06/14/02 42.8 42.8 1.0 42.8 42.8 0.0 Estimated Mortali!I Causes Suspected Suspected Predation Malnutrition 12/15/00- 12/15/0112/15/00- 12/15/0106/14/01 06/14/02 06/14/01 06/14/02 68.6 48.7 60.0 27.5 75.6 0.0 ·11.1 68.1 38.5 0.0 3.0 4.0 0.0 7.8 24.4 4.5 12.2 60.0 0.0 15.6 48.7 Accidents Unknown 12/15/00- 12/15/0106/14/01 06/14/02 t.6 5.3 12/15/00- 12/15/0106/14/01 06/14/02 27.1 0.0 3.5 2.0 2.6 1.8 1.6 5.3 0.0 27.1 1.0 27.1 27.1 �2ll Table 6. Survival rates for winter-spring (WS), summer-fall (SF), and annual (Ann) seasonal intervals from 15 December 2000 to 14 June 2002 for adult female elk age ~2 years ~l year radio-collared in December 2000 and 200 I in the Gunnison Basin, Colorado. Binomial estimator used to calculate survival rates and confidence limts for elk combined among DAUs E-25, E-41, and E-43. Adult Female Elk Seasonal Interval and Dates :li:YS SE 6cc :li:YS 12/15/00 06/14/01 06/15/0112/14/01 12/15/0012/14/01 12/15/0106/14/02 1.00 0.92 0.84 1.00 39 39 0 3• 0.92 0.84 1.00 39 39 0 3 1.00 2 2 FEMALES ~2 yrs old) Survival Rate Lower 95%CL Upper 95%CL n Collars Collars Deployed Collars Censored Died Non-hunting Deaths Hunting Deaths 39 39 0 0 0 0 FEMALES ~l yr old) Survival Rate Lower 95%CL Upper 95%CL n Collars Collars Deployed Collars Censored Died Non-hunting Deaths Hunting Deaths 0.91 0.84 0.97 77' 77 0 7 1 6 1.00 39 39 0 0 0 0 48b 48 0 0 0 0 1.00 82 82 0 0 0 0 • Adult female deaths: 172. 758/00 about 7/1/01, 174.478/00 legal rifle harvest, and 172.030/00 archery/muzzleloading wounding loss. bIncludes 12 additional adult females radio-collared 16-20 December 200 l. c Includes 38 yearling females that survived as radio-collared calves. Table 7. Survival rates for winter-spring (WS) and summer-fall (SF) seasonal intervals from 15 December 2000 to 14 June 2002 for the cohort of 6-month old elk calves radio-collared in December 2000 in the Gunnison Basin, Colorado. Binomial estimator used to calculate survival rates and confidence limits for elk combined among DAUs E-25, E-41, and E-43. MALES Survival Rate Lower 95%CL Upper 95%CL n Collars Collars Deployed Collars Censored Died Non-hunting Deaths Hunting Deaths 6-11 mos WS 12/15/0006/14/01 12-17 mos SF 06/15/0112/14/01 0.78 0.63 0.93 32 38 6 7 7 0 0.86 0.71 1.00 22 25 3• 3 2 lb Elk Age (months) and Seasonal Interval and Dates 18-23 mos 6-11 mos ws ws 12/15/0106/14/02 12/15/00 06/14/01 1.00 19 19 0 0 0 0 FEMALES Survival Rate Lower 95%CL Upper 95%CL n Collars Collars Deployed Collars Censored Died Non-hunting Deaths Hunting Deaths 0.97 0.92 1.00 39 40 1 1 1 0 12-17mos SF 06/15/0112/14/01 18-23 mos 0.89 0.79 0.99 38 38 0 4 0 4' 1.00 ws 12/15/0106/14/02 34 34 0 0 0 0 •Yearling males censored: slipped collars, 173.091/00, 173.391/00, and 173. 510/00 between 6/22/0land 7/20/01. b Yearling male illegally wounded and died about 12/10/01 during late-season for antlerless elk. < Yearlin_g females 172.619/00 and 174.360/00 wounded during regular rifle seasons and 174.560/00 173.589/00 disappeared during regular rifle and late rifle seasons respectively, and assumed to be legalfy harvested. �212 Table 8. Body mass (kg), total body length (cm), and hindfoot length (cm) of elk calves captured and radio-collared in mid-December 2000 and 2001 in the Gunnison Basin, Colorado. Summaries include only those calves contributing to estimates of survival during winter from 15 December to 14 June and exclude calves dying from capture-induced causes. Values represent average weight (Mean), sample size (n ), standard error of the mean (SE), confidence interval of the mean (CI), minimum (Min), and maximum (Max). Mass (kg) Calf Groupings Total Body Length (cm) Hindfoot Length (cm) Gunnison Mean n SE 95%CI Min Max Mean n SE 95% CI Min Max Mean n SE 95%CI Min Max All Females 98.6 74 1.54 95.5 -101.7 57.5 1215 179 0 77 115 176.7 - 181.3 1435 196.0 56.0 77 0.22 55.6 - 56.4 49.0 61.5 All Males 99.6 74 1 64 %.3 - 102.8 52.0 133.0 178.2 76 1 12 176.0 -180.4 146.0 196.0 56.4 76 0.30 55.8 - 57.0 47.5 65.0 All Calves 99.1 14 112 %.9-101.3 52.0 133.0 178 6 153 0.80 177.0 -180.2 143.5 196.0 56.2 15 0.19 55.8 - 56.6 47.5 65.0 Females2000 99.9 37 2.14 95.6 - 104.3 57.5 119.0 178.6 39 1.62 175.3 - 181.9 143.5 194.5 55.9 39 0.33 55.2 - 56.5 49.0 59.5 Males2000 100.8 37 2.56 95.6 - 106.0 52.0 124.5 1802 37 1.81 176.4 -183.8 146.0 196.0 56.1 37 0.47 55.2 - 57.1 47.5 60.5 All2000 100.4 74 1.66 971-103.7 52.0 124.5 179.3 76 1.20 176.9 - 181.7 143.5 196.0 56.0 76 0.28 55.4 - 56.6 475 60.5 2.22 92.8 - 101.8 57.5 1215 179.5 38 1.66 176.1 - 182.9 151.5 196.0 56.2 38 0.30 55.5 - 56.8 51.5 61.5 Females 2001 97.3 37 Males 2001 98.3 37 2.05 94.2 - 102.5 78.5 133.0 176 3 39 1.30 173.7 -179 0 163.0 193.5 56.6 39 0.39 55.8 - 57.4 52.5 65.0 All2001 97.8 74 1.50 94.8 - 100.8 57.5 133.0 177.9 77 1.06 175.8 -180.0 151.5 196.0 56.4 77 0.25 55.9 - 56.9 51.5 65.0 DAUE25 All Calves 95.9 48 1.67 92.5 - 99.3 64.5 118.0 176.6 50 125 174.0 - 179.l 151.5 194.5 55.4 50 0.30 54.8 - 56.0 49.0 59.0 DAUE43 All Calves 96.8 50 2.10 92.6 - l 01.0 52.0 119.0 175.4 52 1.43 172.5 - 178.3 143.5 193.5 56.3 52 0.37 55.5 - 57.0 47.5 65.0 DAUE41 All Calves 104 4 50 1.81 ]008-108] 57 5 133 0 184.0 51 ]15 181.6 - 186 3 ]57 0 196.0 569 51 025 56.4- 57.4 52.0 60.5 E25 All Calves Trapzone A 97.6 17 2 47 92.4 - ]02.9 81.0 118.0 179 6 17 1.98 1754-183.8 166.0 194.5 55.8 17 0.50 54.7 - 56.8 51.5 59.0 E25 All Calves Trapzone B 96.2 7 3.27 88.2-1042 86.0 109 0 1771 7 l 66 173.0-18]] ] 70.5 181.5 56.2 7 0.73 54.4- 58.0 530 590 E25 All Calves Trapzone C 95.l 15 3.94 86.7 - 103.6 64.5 117.0 175.8 17 2.38 170.8 - 180.9 153.0 189.0 54.9 17 0.58 53.7 - 56.2 49.0 58.5 E25 All Calves TrapzoneD 93.7 9 3.41 85.8-1015 73.0 105.0 171.8 9 3.23 164.4-179.3 151.5 182.0 54.9 9 0.66 53.4 56.5 51.5 57.5 E43 All Calves Trapzone E 100.4 6 4.68 88.4 - 112.5 87.5 117.0 178.6 7 2.62 172.2 - 185.0 167.5 188.0 55.9 7 0.38 55.0 - 56.9 55.0 58.0 E43 All Calves TrapzoneF 93.l 10 6.17 79.l - 107.l 52.0 117.0 170.8 ']] 3.90 162 l - l 79 5 1460 185.5 56.8 11 1.36 53.8 - 59.8 47.5 65.0 E43 All Calves Trapzone G 97.3 34 2.41 92.4-102.1 57.5 119.0 176.2 34 1.68 172.8 - 179.6 143.5 193.5 56.1 34 0.37 55.4- 56.9 49.0 59.5 E4 l All Calves TrapzoneH 100.] ]] 6.37 86.0 - l 14.4 57.5 133.0 181.6 11 3.56 173.7 - 189.5 157.0 193.5 56.3 11 0.71 54.7 - 57.9 52.0 60.0 E4 l All Calves Trapzone I 104.9 17 2.60 99.4 - 110.5 85.5 121.5 185.0 17 1.86 181.0 - 188.9 1670 196.0 57.3 17 040 56.5 - 58.2 54.5 60.0 E4 l All Calves Trapzone J 106.l 22 179 102.4 - 109.8 89.0 124.5 184.3 23 1.39 181.5 -187.2 169.0 196.0 56.9 23 0.33 56.3 57.6 54.0 60.5 . �213 Table 9. Numbers of adult female elk harvested during late-seasons in the Gunnison Basin, Colorado, NovemberDecember 2000 and 2001 with numbers and percent (%) of harvested adult females from which hunters provided any of the requested biological samples, reproductive organ samples, and kidney fat samples. Samples received for kidney fat expressed as fat with 1 kidney, fat with 2 kidneys; and elk with at least 1 kidney fat sample. Estimates of adult females harvested obtained from CDOW statewide harvest surveys. Late Season Year Adult Females Harvested Adult Females With Any Requested Samples Submitted Adult Females With Reproductive Organs Submitted Adult Females With Kidney Fat Samples Submitted 1 kidney; 2 kidneys; :::_ l kidney 2000 291 (100) 81 (28) 58 (19) 17 (6); 40 (14); 57 (20) 2001 374 (100) 46 (12) 31 (8) 22 (6); 5 (l); 27 (7) All 665 (100) 127 ( 19) 89 (13) 39 (6); 45 (7); 84 (13) Table 10. Frequency (%) of dental cementum ages of adult female elk harvested in the Gunnison Basin, Colorado, November-December 2000 and 2001 based on useable incisor tooth samples submitted by hunters. Age Class of Adult Female Elk Based on Dental Cementum (years) Year 3-4 5-7 8-10 11-14 15-20 All 2000 5 7 12 20 10 14 6 74 2001 1 7 15 8 9 4 0 44 All 6 (5) 14 (12) 27 (23) 28 (24) 19 (16) 18 (15) 6 (5) 118 (100) 2 Table 11. Pregnancy rates(%) by age class of adult female elk in the Gunnison Basin, Colorado, NovemberDecember 2000-2001 based on samples submitted by hunters for years combined. Age of elk based on dental cementum. Pregnancy rates based only on numbers of known pregnant and non-pregnant elk per age class. Age Class of Adult Female Elk Based on Dental Cementum (years) Pregnancy Status 11-14 3-4 5-7 1 14 (93) 0 1 1 3 4 13 21 (100) 13 (93) I2 (92) 3 (50) 5(56) 76 (85) 8-10 15-20 Unknown Adult 2 All Non- Pregnant 0 3 Pregnant 2 (100) 6 (67) Unknown 4 5 12 7 5 5 0 0 38 Total 6 14 27 28 19 18 6 9 127 Table 12. Measurements of elk fetal size in the Gunnison, Basin, Colorado during November-December 2000 and 2001. Values represent average size ( X, mean), sample size (n), standard deviation of the mean (SD), confidence interval of the mean (CI), and minimum (min) and maximum (max) values. November-December 2000 Measurements X (n) SD November-December 2001 95%CI min-max X (n) SD 119.3 (6) 111.0 (13) 4.9 (3) 95%CI min-max 95.9 18.7-219.9 46.4-289.1 85.5 3.2 59.4-162.7 0.0-12.7 22.1-310.5 1.8-8.1 106.1-171.2 116.7-162.2 13.5-75.9 111.6-186.5 Mass (g) Male 64. 7 (15) 51.0 36.5-92.9 12.4-167.0 Female Unknown Sex Crown-Rum!! (mm) Male 86.8 (24) 4.3 (7) 97.8 3.7 45.5-128.0 0.8-7.7 12.3-425.5 2.0-12.6 114.5 (15) 30.4 97.7-131.4 73.6-165.0 138.6 (6) 31.0 125.6 (24) 33.9 (8) 39.5 18.6 108.9-142.3 18.4-49.4 68.5-223.0 0.5-69.0 139.5 (13) 44.7 (3) 37.7 12.6 44.9 (15) 50.7 (24) 24.0 (1) 16.6 35.7-54.0 23.8-73.9 42.2-59.3 23.6-103.9 60.6 (6) 59.4 (13) 15.0 20.3 . 29.2 (15) 10.6 34.7 (24) 16.5 Female Unknown Sex Hind-Leg (mm) Male Female Unknown Sex Hind-Foot (mm) Male Female Unknown Sex 13.9 {12 19.0 44.9-76.4 47.9-70.9 13.4 14.8 30.8-48.7 87.9-202.0 32.3-57.4 47.0-83.4 36.1-91.2 17.4(1) 23.3-35.0 27.7-41.6 14.9-48.1 14.8-79.9 40.4 (6) 39.8 (13) 11.5 {12 26.3-54.5 29.4-61.4 21.4-64.6 �214 Table 13. Summacy values for total fat kidney fat index (TF-KFI), trimmed fat kidney fat index (TRF-KFI), kidney total fat mass (TFM g), and kidney trimmed fat mass (TRFM g) for adult female elk in the Gunnison Basin during November-December 2000-200 I. Kidney mass one and two could represent either the left or right kidney masses with kidney mass two associated with elk for which both kidney masses were collected by hunters during late antlerless-only hunting seasons. Values represent average size (X, mean), sample size (n), standard deviation of the mean (SD), confidence interval of the mean (Cl), and minimum (min) and maximum (max) values. Kidney Mass One Fat Value X" (n) TF-KFI Kidney Mass Two SD 95%CI Min-Max X" (n) SD 95%CI Min-Max 105.8 (84) 54.5 93.9-117.6 29.1-306.1 91.0 (45) 37.5 79.8-102.3 33.4-165.3 TRF-KFI 78.4 (84) 35.5 70.7-86.1 29.1-212.8 70.3 (45) 26.5 62.3-78.3 30.4-136.1 TFM (g) 218.7 (84) 103.9 196.1-241.2 57.0-551.0 202.4 (45) 90.6 175.2-229.6 72.0-486.0 TRFM (g) 163.6 (84) 70.4 148.4-178.9 48.0-383.0 156.8 (45) 66.3 136.9-176.7 69.0-388.0 Table 14. Estimates of percent total body fat in adult female elk by pregnancy status and calf elk (sexes combined) in tl1e Gunnison Basin, Colorado during November-December 2000-200 I. Percent body fat based on total fat kidney fat index (TF-KFI), trimmed fat kidney fat index 9TRF-KFI), and total kidney fat mass (TFM g) after Cook et al 2001a Comparisons among mean values shown by P-values (ANOVA). Values represent average size (X, mean), sample size (n), and confidence interval of the mean (CI). % Body Fat TF-KFI X" (n) 95%CI % Body Fat TRF-KFI % Body Fat TFM X" (n) 95%CI X" (n) 95%CI Percent Body Fat Adult Females Pregnant 11.2 (57) 10.6-11.8 12.1 (57) 11.5-12.8 11.3 (57) 10.8-11.9 7.0 (TF) 18.5 (TRF) Non-Pregnant 10.2 (11) 8.8-11.5 10.9 (11) 9.5-12.2 9.9 (11) 8.9-10.9 8.3 (TFM) 14.8 (TRF) Pregnancy Status Unknown 11.8 (16) 10.4-13.1 12.6 (16) 11.1-14.l 11.1 (16) 9.6-12.5 5.4 (TFM) 15.3 (TRF) Pregnant vs. NonPregnant P=0.143 P = 0.112 P = 0.038 Pregnant vs. NonPregnant vs. Unk. P=0.193 P = 0.198 P = 0.149 All Adult Females 11.2 (84) 10.7-11.7 12.1 (84) 11.5-12.6 11.1 (84) 10.6-11.6 5.4 (TFM) 18.5 (TRF) Aduh Females 2000 10.7 (57) 10.2-11.3 11.6(57) 11.0-12.2 10.8 (57) 10.2-11.4 5.4 (TFM) 18.5 (TRF) AdultFemales2001 12.1 (27) .11.1-13.1 13.1 (27) 12.1-14.1 11.7(27) 10,_8-12.6 7.3 (TF) 17.0 (TRF) 2000vs. 2001 P=0.009 All Calves• P =0.010 P = 0.073 7.1 (6) 4.1-10.1 7.3 (6) 4.1-10.4 4.1 (6) 1.1-7.0 0.2 (TFM) 10.4 (TF) Calves 2000 7.4 (2) 0.3-14.5 7.7 (2) 0.0-23.4 3.8 (2) 0.0-8.2 3.4 (TFM) 9.0 (TRF) Calves 2001 7.0 (4) 1.2-12.8 7.0 (4) 1.1-10.3 4.2 (4) 0.0-9.9 0.2 (TFM) ' 10.4.(TF) • Estimates of percent body fat in calves should be viewed with caution as calibration equations developed by Cook et al. 2001 were based only on adult female elk. b Minimum and maximum estimates of percent body fat obtained from either TF-KFI, TRF-KFI, or TFM estimators. �215 Fig. i. Game Management Units (54-551). trapzones (A-J), and elk capture sites in.the GunnisonBasin .in 2000 arid 2001. C3ptu~ Sites Capture: Site Vie3r AF .. "'tall Fl:,lrct U ... De=mlll:r2CIJD .Al• A.mtOl'\ITllylOr ■ De~trt:r:l.1J1 AT• .---on.1nt11•111: + . IV• lea.in CIC.Ek SWA oC a!fflllC r 2Clla Zll'II 2DD1 cc .. cm1ncreu D "ltl:IPJ:a\C5 co .. oo., Ouldl WIiow □ ouus ~ .. o,:,c~ck e,o .. klt::hO.Jch ,,,.,.,. DV• Dcad'5Cra:k DY• D,r·Creet U~r a: .. a.; 1c;:11n crri:t rT• Flal Too 8ouh IIO • Mom OUICh • H.- N<lll'ICOl.lq"alllmar . • t::K• cat1, ~·111:'_o·tJ<:h l(Z .. IC!:mr lmh ~W N A LC• LM-IOznyCJn 0Uch IP• IG'h PsNn· P~ • P'Olt CteEIC 1 PIil .. l'ctson I\Jdat kl • 111.oi.ndull I :=:in IIIC• kdC.cB. I\D • 111.aldtn Flzal Toa: llV- Ill.GIid 11:an:r Creek 8C • 8oclll Cl'Hk C~I eo--sii:cn:O\lch eu .. ~o- Creek: Tf ■ n~r"""ZOIU~ ,U ■ Ten1111ll1: El:irlrG .'10 • TofficN DCIIIII: TT• 'nlllb: "11)JI WA• IIWH1_Anl:lopc creek UIJC) • Wood:, OUd\ WM• WIii~ M~ ltw ·vuw- -Wll<nu(lrcet:: 1· WY• \l\llmuy Hane "1'0• Yl:;J!JIIU•Oulctl YP • Ye lloWPJnE Ill.II DC 20 Kilometers �216 Calf Elk Winter Mortalities Gunnison Basin, Colorado ■ Lion Predation D Black Bear Predation D Suspected Predation ~ Malnutrition ffl Suspected Malnutrition ~ [I] 2. ated s of calf mortaliti during from 15 mber h 14 in the son Colorado, 2000-01, 2001-02, and years combined. Accident Unknown Figure Estim cause elk es ,..........._~~___.____, winter ,,.....,.~~~~ Dece throug June Gunni Basin, �217 3--r------------,------....--.--------------- Calf Elk Winter Mortalities Gunnison Basin, Colorado 2-1----------.-------r--------1 D Unknown § Accident ■ Predation D Malnutrition 1---------- 0----,---....----Jan 1, 01 Feb 1 Dec16,0D Jan 16 Mar1 Feb 16 JI.DI 1, 01 Apr1 Mar 16 May 16 Apr 16 Beginning Date of Two Week Intervals December 2000 to June 2001 a----- ~ ~ 2-+---- iii (.) 1-+---- Apr1 Dec 18,01 Jan 16 Feb 18 Mar 18 Jun 1, D2 May1 Apr 18 May 18 Beginning Date of Two Week Intervals December 2001 to June 2002 Figure 3. Timing and estimated causes of calf elk mortalities during winter from 15 December through 14 June in the Gunnison Basin, Colorado, 2000-01 and 2001-02. �218 100 ~ ~ I • I x 2001-02 2000-01 : catf Elk Marrow Fat, Gunnison Basin, Colorado: ~ - ~ 'fl' [i] ~ l'u1 D ... Predation Related E Malnutrition Relaled :::, tf. C: 40 a, ea, Accident 0 Unknown 0 0.. 20 ~ \/ '1 'i7 [i] ~ ® (i] [i] ® (it) 0 I 12116 - 1/15 \";:,,.., I I 2114 3116 I t:""' I 4115 6/14 Momh and Day Figure 4. Percent femur marrow fat in calf elk mortalities by estimated cause and timing of deaths during winter from 15 December to 14 June in the Gunnison Basin, Colorado, 2000-01 and 2001-02. 30 1;:::========-=--=-.--------------;::;-------;::==============~7 ■ 2000-01 n=8 X 2001-02 n = 12 Predation Related Accident O Males n=74 0 Malnutrition Related ro ■ Females n=74 Z1 0 Q ~0 1 5 -Unknown -t------------.-,,,---., ~ iz ~ Gil 10--+------------ 50-59 60-69 70-79 80-89 90-99 100-109 calf Body Mass aass (kg) 110-119 120-129 130-139 Figure 5. Distribution of male and female calf body masses and occurrence of calf mortalities by mass class, calf sex, and estimated cause of death during winter from 15 December to 14 June, Gunnison Basin, Colorado, 2000-01 and 2001-02. �219 20----,--------------------------------------. Elk Conception Dates Gunnison Basin, Colorado 2000-2001, n = 72 - - 16-+---------- Mode= Sep23 Median 50% Quantile = Sep 26 ~ 12 - + - - - - - - - - - - Mean= Sep29 C: Q) :::J CT Q) ._ LL 8 -+-------- - 75% Quantile = Oct 3 4-+------ 0 Sep 6-10 Sep 16-20 Sep 26-30 Oct 6-10 Sep 1-5 Sep 11-15 Sep 21-25 Oct 1-5 Oct 11-15 Oct 21-25 Oct 31-Nov4 Nov 10--14 Figure 6. Frequency of estimated conception dates in 5-day intervals for elk fetuses in the Gunnison Basin, Colorado, 2000-2001. Male Elk Fetuses Per Adult Female Age Class, Gunnisoo Basin 2000-2001 Male Elk Fetuses Per Conception Date Interval, Gunnison Basin 2000---2001 Male & Female Felu:.u Perlnl~rv?I! Sho'M1As (11) Male and Femalt": Fetuses Per Age Interval ShCJIMl As (15} 0.6 I 0.6 0.5 I 0.1 01 I Sep&,O s~11.1s S.p21-25 Sepl&-20 I 0<:1-5 I 00;6-10 I I I I ' Adul Female Dental Cemenlum 3---4 S--7 8-10 Ag/: Class 11-14 I 15-20 . Figure 7. Proportions of male elk fetuses per conception date interval (LEF1) and per adult female age class (RIGHI), Gunnison Basin, Colorado, 2000-2001. �220 ., AdlA Fem11~ Elk Total K~y Fat lndn, Guin.I~ Ba~rl. n = 64 17 Esttnt,s Pemrt Body FJIMull r.rm.ie El<, Gwrison Bun.2<m-2001 ~-~ n•Mpcr '-FatTF-KFl TRF-KR. TAI E~ □ ., 16 -· " □ ■ ,0 7 6 3 3 I I ,._,, .,_,, ~0-~ I I 80--g,§I I Hl0---119 140---759 !Xl-139 "' "' 5 0 ,. -"' 8 ' I 100-19'9 HD----179 ::,00.~!1 TOl,Jj KICrleyFS!ln-6e:I: 'watues(Pen::ent) - ~ f-- 10 Jl. 1 0 ,,._,,, ...!....o, I 0--U XD-Jd9 7-U 10---1'.t a o o 1£-11.0 -BolltFJI Figure 8. Frequency of total kidney fat index values (TF-KFI) (LEFT) and percent total body fat estimates (RIGHT) for adult female elk age :::_l year during November-December, Gunnison Basin, Colorado, 2000-2001. Body fat estimates based on TF-KFI, trimmed kidney fat index (TRF-KFI), and total kidney fat mass (TFM g), respectively, after Cook et al. 2001a. Percent body fat classes 0-6.9, 79.9, 10-14.9, 15-19.9, and 20-24.9 represent body condition classes very low, low, moderate, very good, and excellent, respectively, after Cook J.G. (2001 unpublished data). 0.9 Body Condition Upper Confldence Limit VoryGood --- -.... -- 0.8 ----------- 0.7 Body Condfflon ,._ Moderate ~0.6 a, e 1'; 0.5 ,, ,, ~ :c i0.4 , c.. , I ,, , , Body CondltlOII Low 0.3 0.2 0.1 - ..... Lower Conn;i,,"' 0 0 2 4 - L:tt 6 8 W U M Estimated Pen:ent Body Fat in November-December 16 18 20 Figure 9. Probability of adult female elk age 2::.1 year being pregnant as predicted from percent total body fat based on total kidney fat mass (TFM g) measured in November-December, Gunnison Basin, Colorado, 2000-2001. Probablility curve bracketed by approximate 95% confidence limits. Relative body condition rating classes from Cook J.G_ (2001 unpublished data)_ Logistic regression was: logit (P) = -2.2835 + 0.3704 *(X-percent body fat); regression slope significantP = 0.033). �221 Appendix A. Locations of elk capture sites in the Gunnison Basin during December 2000 and 2001. All U'IM coordinates are referenced to NAD 27 datum Erojection. Site Code WY WL WM DV TM WW pp AF TT HG WG cc NP LC AL RD WA BV DC RC CG YG KZ RV SU PR DG HR KK TO yp AT EC RB FT SG DY TF SC Trap Site WINNERY HOME WlLSONGULCH WILLOW MESA BLUE DEVILS CREEK TENMILE SPRING WILLOWCKl POLECKl ALKALIFLYING M TABLE TOP HORN GULCH WOODS GULCH CABIN CK NORTH PARLIN LOST CANYON GULCH ALMONT TAYLOR REDDEN FLATTOP WEST ANTELOPE CK BEAVER CK SWA DRY CREEK RED CREEK COW GULCH WILLOW YEAGER GULCH KEZAR BASIN NW ROAD BEAVER CK SUGARCREEK POISON RIDGE DUTCH GULCH HOME GULCH RAZOR CAMP KETTLE GULCH TOMICID DOME YELLOWPINE RIDGE ALMONT TRIANGLE EAST CABIN CREEK ROUNDUP BASIN FLATTOP SOUTH STEERS GULCH DRY CREEK UPPER TENDERFOOT MESA SOAP CREEK COAL Trapzone A A A B B C C D E F F G G G G H I I J A B B C C D D E E F F G G G H J J Capture UTMx Capture UTMv Capture UTMZone 299842 299987 302286 299S07 307088 319409 329616 343809 3S265S 367S5S 3S7804 342S64 3S1341 343419 343020 331774 32S2S9 321816 313617 30S342 302170 304088 311114 321992 32S267 344624 342179 3S1874 349729 365097 3S9615 342507 344065 348075 3330S1 323101 313563 306929 299839 4229298 4232163 4248791 422124S 42S21S1 42S044S. 4246431 423826S 4249609 42S541S 4263648 4267804 426S7S7 427S048 4287228 4282420 4274696 4266364 4262208 4263S42 4248762 423371S 42S7470 423S039 42S2487 42238S0 42S61S0 4237208 42S0925 4256608 4264943 4287754 4270540 4270061 4281590 4270141 4262643 4264206 4265834 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 Capture Capture 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 2001 2001 2001 2001 2001 2001 2001 2001 2001 2001 2001 2001 2001 2001 2001 2001 2001 2001 2001 2001 2001 2001 2001 �222 • the Gunmson Basm, 15 December to 14 June, 2000-01 AppendixB Summarvofcalf e]k mortali.t:iesm TrapNo Elk ID Sex Mass zone 1 173.082/00 M 92 B Tassue Ferrur Marrow Fat Recovered Samples Parasites Yes CM n/a 2 172.379/00 F 127 H Death Date 12/29/0Q.. 1/4/01 12/1 S-26/00 3-Jan-01 WhiteCreamy 38.7% WhiteFirm 94.7% 3 173.640/00 F 82 A 2/1-7/01 7-Feb-01 Red.Jelly 15.8% NoCM 4 172.959/00 M 125 J 2/15/01 15-Feb-01 No CM 5 173.351/00 M 52 F 3/21/01 24-Mar-01 WateryPink 48.7% Red.Jelly 27.5% NoCM 6 173.160/00 M 94 A 3/25/01 26-Mar-01 Red Jelly 42.8% No CM 7 173.300/00 M 112 I 3131/01 1-Apr-01 Red Jelly 0.0% No CM Carcass Status Death Cause Death Location UTMy Drainage UTMx 303163 4253981 Lake Gulch 335406 4282773 304608 4243376 FlatTop Lake Fork Unk.-Suspect Starvation Unk.-Suspect Starvation Stravation 304753 4261110 Red Ck. 360867 4262422 Yellow Pine 298132 4230866 Dwyer Gulch Unk.-Suspect Starvation Unk.-Suspect Starvation Lion Predation 308379 4263917 Dry Gulch 320104 4270341 Beaver Ck. 311827 4245467 Cebolla Ck. Bear Predation 300715 113 J 4128-5110/01 23-May-01 n/a Scavenged M Mass= Weight of caH (kg) at capturej CM = capture myopathy; NI = no evidence of inllarmtion around sarcocystsj n/a = sanples not available. 4268428 E. Coal Ck. !Nan-01 8 173.041/00 M 97 I 4/13-14/01 15-Apr-01 9 173.949/00 M 107 A 4/20/01 26-Apr-01 10 173.011/00 Red Jelly 77.66% RedCreamy 45.27% Red Jelly 13.15% n/a NoCM No CM n/a Moderate sarcocysts Moderate sarcocysts Nonnal sarcocysts Moderate sarax:ysts/NI Severe sarax:ysts/NI Nearly Complete Scavenged Capture Related / Capture Related Partially Scavenged Carcass Complete Scavenged Lion Predation Carcass Complete Carcass Severe sarcocysts/NI Normal sarcocysts n/a Complete Nearly Complete Scavenged Lion I ! I ! 'I i AnnendixC. sll.lllllUllyofcalf elk mortalities in the Gunnison Basm, 15 December to 14 June, 2001-02. Trap- Death Zone E Date F Mass 100 173.740/01 F 101 E 174.720/01 M 101 B No Elk ID 1 173.429/01 Sex 2 3 Tissue Samples Parasites Firm core, pink;87.45% NoCM unremark- 21-Dec-01 Firm core, pink;66.5% Mild CM n/a 26-Dec-01 Soll core, pink;88.87% NoCM Low Partially Sarcosysts Scavenged n/a Totally Scavenged unremark- Partially able Scavenged Ferrur Recovered 20-Dec-01 Marrow Fat 12/21/ 2001 12/2123/2001 011.212/2002 01/37/2002 14-Jan-02 n/a 8-Jan-02 Firm core, pink;78-50% Firm core, pink;90.71 % Firm, red; 68.56% Jelly, red: 1.57% Firm core, pink:94.36% NoCM 7-Mar-02 Soft core, red;5.27% n/a S-Mar-02 n/a 12/18/ 2001 4 173.269/01 M 98 J 5 172.350/01 F 86 H 6 173.852/01 F 101 I 1/4-9/ 2002 10-Jan-02 7 175.181/01 M 91 A 6-Feb-02 8 172.170/01 F 58 H 1/15-2/5 /2002 1/20-21/ 2002 1/25-28/ 2002 2/21-28/ 2002 2125-316 /2002 2/27.:J/6 /2002 4-Mar-02 9 173.861/01 F 102 J 10 174.770/01 M 92 A 11 172.379/01 F 81 G 12 173.300/01 M none J 13 173.632/01 14 15 16 23-Jan-02 2S-Jan-02 Carcass Status Carcass Complete UTMx 351433 4245033 Prosser Ck Carcass Complete Capture Related eulhanized 353050 4250600 E. TableTop Mt Capture Related / 302744 4230958 Skunk Ck. 300102 4267074 Pearson Pt 334033 4282350 Flat Top Min Partially Lion predation Scavenged Scavenged Link-suspect n/a heavily coyote predation ·Accident-Haystack bronchoCarcass pnuem:,nia Complete collapse 321888 4275810 W.Antelope Ck 297064 4231180 Dwyer Gulch 330243 4280937 Redden's NoCM Moderate sarcocysts Partially Scavenged Lion predation 299250 4267725 Pearson Pt NoCM 294622 4227200 Elk Ck. Unk.nOIMl 342715 4288645 AlrnontTriangle n/a n/a Carcass Complete Totally Scavenged Not FoundSnow Partially Scavenged Totally Scavenged Totally Scavenged Accident-fell, trapped n/a Moderate sarcocysts n/a Unk.nOIM'l 299323 4268454 N.Pearson Pt Lion predation 322925 4232867 RoadBeaver Ck. Unk.nOIMl 306109 4267739 Red Ck. Unk-suspect lion predation 325443 4235881 N. RoadBeaver Ck. Totally Unk-suspect lion Scavenged predation 351363 4240438 Home Gulch able n/a No CM n/a NoCM 96 C I Link-suspect lion predation Bear Predation Fern1r Marrow Fat Tissue Samples Parasites Carcass Recovered >06/22<7 /21 /01 >06/22<7/24/01 21-Jul-01 24-Jul-01 n/a 77.46% n/a n/a n/a n/a G >6/22<7/20/01 16-Aug-01 n/a n/a n/a F C >9/25<10/18/01 19-0ct-01 n/a n/a 16 mos F J > 10/13<10/18/01 20-0ct-01 n/a n/a Co~lete 174.478/00 5-9 yrs F G 10/13/2001 17-0ct-01 v,Me,solid 97.31% pink,crumbles 85.18% n/a n/a n/a Hunter Kill 7 174.360/00 17 mos F G >11/8<11/16/01 17-Nov-01 n/a n/a Co"1)1ete 8 174.140/00 18 mos M C >12/5 <12/28/01 29-Dec-01 n/a n/a Scavenged 9 173.589/00 18 mos F 8 >12/28/01 <1/3/02 n/a white/gray, solid 95.18% white/pink, firm86.27% n/a n/a n/a n/a 10 >10/30/01 <12/31/ n/a n/a 01 Age estimated using dental cementum or know, age as elk collared as calf. n/a n/a n/a Trap- Age@ Death Sex 1 2 172.758/00 173.330/00 19 yrs 12 mos F M H G 3 173.340/00 12 mos M 4 172.030/00 6 yrs 5 172.619/00 6 174.560/00 17 rros F zone Death Date J Status ! ' ' I • the Gunnison Basm 15December2000t0 141une 2002 A,01 :>emdixD S;ummarvof adult elie mortalitiesm No ElklD ' I Lion Low Sarcosysts 26-Mar-02 Firm core. NoCM Moderate 3/20-25 pink: 83.60% sarcocysts /2002 n/a 173.780/01 108 J 4/25-4/30 1-May-02 Firm core. n/a F pink; 27.08% /2002 5/15-20 21-May-02 Red, firm n/a n/a 175.240/01 M 100 C core;60.02% /2002 95 5115-20/ 22-May-02 Firm core, n/a n/a 174.180/01 M E 2002 pink;75.63% Mass = Weight of calf (kg) at capture· CM = capture myopathy; n/a = sanples not available F Death Location UTMy Drainaae Death Cause Capture Related Fence Kill Death Location Death Cause UTMx UTMy Drainage Decomposed Unk.rov,, Mortality Scavenged Un k--Suspect Predation Heavily Unk-Suspect Scavenged PredaUon Heavily Archery/Muzzle Scavenged v.ound loss 321262 354289 4294461 $. Carbon Min. Surrmerville Ck. 384766 4303460 322980 4227638 N. Cottom11oood Ck. Swnehart Gulch Rifie V><>und loss 1st season Rifie 1st season Legal Rifte 4th season w:,und loss Late rifle v.oundrillegal loss Disappear Late Rifle season Legal Disappear 3rd rifle season Legal 299161 4275164 C<mCk. 341000 4242650 Rock Ck. 346979 4279939 E. Beaver Ck. 308573 4242145 Lake City Cut-Off n/a n/a Low Cebolla Ck. n/a n/a West Elk Ck. 4290527 �223 APPENDIX I PROGRAM NARRATIVE STUDY PLAN FOR RESEARCH FY 2000-01 - FY 2003-04 State of: Colorado ProjectNo.: W-153-R-14 Work Package: __3-'0~0~2_ _ Study No.: _ _ __,3'---- Cost Center 3430 Mammals Research Program Elk Conservation Estimating Calf and Adult Survival and Pregnancy Rates of Gunnison Basin Elk Populations ESTIMATING CALF AND ADULT SURVIVAL AND PREGNANCY RATES OF GUNNISON BASIN ELK POPULATIONS Principal Investigators David J. Freddy, Wildlife Researcher, Mammals Research R Bruce Gill, Wildlife Research Leader, Mammals Research Cooperators Rick Kahn, Terrestrial Field State Coordinator John Ellenberger, State Big Game Coordinator Jim Olterman, Senior Biologist, West Region Don Masden, Gunnison Area Terrestrial Biologist Jim Young, Gunnison Area Wildlife Manager Gary C. White, Professor Wildlife Biology, Colo. St. Univ. David C. Bowden, Professor Statistics, Colo. St. Univ. STUDY PLAN APPROVAL Prepared by: _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ Date: _ _ _ _ _ _ __ Submitted by: _ _ _ _ _ _ _ _ _ _ _ _ _ __ Date: _ _ _ _ _ _ _ __ Reviewed by: _ _ _ _ _ _ _ _ _ _ _ _ _ __ Date:. _ _ _ _ _ _ _ __ Date: _ _ _ _ _ _ __ Date: _ _ _ _ _ _ __ Approved by: _ _ _ _ _ _ _ _ _ _ _ _ _ __ Biometrician Date: _ _ _ _ _ _ __ Date: _ _ _ _ _ _ _ __ Research Leader November 2000 Final �224 PROGRAM NARRATIVE STUDY PLAN State of: Colorado Project No.: W-153-R-14 Work Package: 3002 Study No.: 3 Cost Center 3430 Mammals Research Program Elk Conservation Estimating Calf and Adult Survival and Pregnancy Rates of Gunnison Basin Elk Populations A. NEED Elk (Cervus elaphus nelsoni) are a high-profile and highly valued resource throughout much of Colorado because elk provide recreation for persons who hunt, watch, and photograph wildlife (Freddy et al. 1993). The elk resource has many benefits but frequent social, political, and economic conflicts suggest elk can reach "social" if not ''biological" carrying capacities. Recent controversy surrounding management of elk in the Gunnison Basin of Colorado (Roath et al.1999) exemplifies conflicting social and biological agendas regarding appropriate numbers of elk. The core of conflict in elk management often centers on establishing management objectives for numbers of elk that are agreeable to competing interests and then monitoring elk populations to demonstrate that objectives are achieved. This type of conflict is paramount in Colorado Division of Wildlife (CDOW) elk population Data Analysis Units (DA Us) E-25, E-41, and E-43 in the Gunnison Basin (Fig. 1) where a combination of resource carrying capacity objectives for elk on winter ranges and difficulties associated with knowingly achieving those objectives has fostered argumentative distrust among public groups and management agencies. Accomplishing management by population objective can depend on reliably estimating elk population size. Estimating population size is expensive and intensive (Samuel et al. 1987, Bear et al. 1989, Unsworth et al. 1990, Anderson et al. 1998, Cogan and Diefenbach 1998, Eberhardt et al. 1998, Freddy 1998) and these factors often preclude routinely using tested inventory methodologies. Alternatively, population size and trend can be estimated using computer models that incorporate harvest, age and sex ratios, and survival rates (White 1992, Bartholow 1999). Model outputs are extremely sensitive to estimates of survival rates such that, reliable measurements of survival can greatly enhance the quality of models (Nelson and Peek 1982). Thus, estimating survival rates is fundamental to modeling elk pop~la!ions in the absence of routine measurements of population size. Estimating calf and adult female survival during winter and annual rates of survival for adult females are higher priorities than estimating adult male survival primarily because most males are harvested when they reach legal age and contribute little to long-term problems of population growth or decline. Models having valid estimates of survival along with currently obtained precise estimates of harvests and population composition would provide more defensible estimates of population size. Although small changes in adult female survival can have major effects on population growth or decline if compounded for several years, calf survival is likely more variable among years. The ability to detect changes in calf survival should be greater than detecting smaller, but important changes in adult female survival (White et al. 1987, Bartmann et al. 1992, Freddy 1998). Estimates of calf survival in Colorado during winter are limited to the Grand Mesa in west-central Colorado where yearly average survival varied between 0.86 and 0.92 from 1993-1996 (Freddy 1998, n 2::, 67 calves/year). Applying these survival rates to other Colorado elk populations, especially those populations using winter ranges higher November 2000 Final �225 in elevation, colder, and more prone to significant snow depths such as in the Gunnison Basin, may or may not be appropriate. Rates of survival on the Grand Mesa were higher than expected and considerably greater than 0.70-0.72 survival rate estimated for elk calves during winter in Yellowstone National Park (Houston 1982, Singer et al. 1997). Estimates of annual survival for radio-collared adult female elk in Colorado averaged 0.95 and ranged from 0.94-0.99, excluding hunting mortalities, for several populations inhabiting widely differing ecosystems (Petersburg and White 1998, Freddy 1999; n > 1,250 adult female-years). Because of the availability of these adult survival estimates, the need to estimate adult female survival is therefore less than the need to obtain additional estimates of calf survival, but ideally we would measure both calf and adult survival simultaneously to document relative differences in survival. A recent evaluation of existing population models for elk in the Gunnison Basin and subsequent development of new population models using estimates of calf and adult survival measured in Colorado altered population trajectories and relative size (Freddy 2000). Consequently, management objectives for Gunnison elk were amended to continue reducing numbers of elk in all DA Us. Controversy surrounding new models and management decisions reinforced the need to obtain measurements of elk survival specific to the Gunnison Basin. B. OBJECTIVE This project will obtain estimates of population parameters for elk in the Gunnison Basin. Major objectives are: I) Estimate survival rates of elk calves during winter from 15 December-14 June within ±15% of the true survival rate at the 95% confidence interval for 3 consecutive years and identify probable sources of mortality. 2) Estimate winter (15 Dec-14 Jun) and yearly (15 Dec-14 Dec) survival rates of adult females for 3 consecutive years to assess whether the true survival rate is likely ,2:0.95 and identify probable sources of mortality. 3) Estimate pregnancy rates of adult female elk harvested during November-December late hunting seasons for 3 consecutive years if late hunting seasons are scheduled. 4) Estimate hunting removal rates for adult females, yearling males, and when possible, adult males for 3 consecutive years. 5) Evaluate Gunnison elk population models using newly acquired survival rates. C. EXPECTED RESULTS OR BENEFITS This project will provide estimates of survival rates for calf and adult female elk and estimates of hunting removal rates for adult elk in the Gunnison Basin DA Us E-25, E-41, and E-43 for 3 consecutive years. These estimates will immediately assist the CDOW in refining population models for Gunnison elk and provide estimates of survival/removal that may be applicable to modeling other elk populations inhabiting similar habitats. In the process of estimating survival rates, probable causes of mortality will be identified which may provide insight into relative health status of elk. Additionally, estimates of pregnancy rates will provide documentation on the fecundity of these elk in relation to other elk populations in Colorado and other states. November 2000 Final �226 D. APPROACH EXPERIMENTAL DESIGN SURVIVAL RATES Radio-telemetry Equipment Survival rates will be estimated by marking elk with radio-telemetry collars that emit a mortality pulse code when collars remain motionless for 4-6 hours (White et al.1987, Freddy 1993). Radios provide the ability to know the fate of individual animals (alive or dead) over discrete periods of time (White and Garrott 1990). Radio-collaring does not likely bias estimates of survival by jeopardizing or enhancing the welfare of individuals when radio-collars weigh <0.8% of an ungulate's body weight (Garrott et al. 1985, White et al. 1987). Radio-collars similar to those previously designed and successfully used for calf and adult elk on the Grand Mesa, Colorado will be used in this project (Freddy 1993, Appendix I). Collars for male and female calves will allow for expansion to adult size while adult female collars will be of fixed circumference and fitted to each individual. Calf collars weigh 840 gm and represent <l % of expected calf body weight while adult collars weigh 1.1 kg and represent <0.5% of expected body weight. Collars will be white in color, have a unique black colored number/symbol embossed on bright yellow plastic material (Ritchey Manufacturing, Brighton, CO) attached to the dorsal surface of collar to enhance visual identification from helicopters (Appendix II), have unique frequencies between 172-17 6MHz, and a battery life of ,2:4 years. Animal Capture We assume survival of those elk captured provides an unbiased estimate of population survival rates recognizing that individual behavior, social behavior, trapping methods and distribution of trapping effort all potentially bias those individuals actually marked (White et al. 1982). Recognizing these problems, elk will be captured with the intent of systematically marking elk throughout the distribution of elk in the Gunnison Basin. Each of the 3 DAUs, will be divided into trap-zones having multiple trap-sites. Capture quotas for calves and adults in trap-zones within each DAU will be proportional to expected elk density as estimated from yearly sex and age ratio classification flights conducted each January throughout the Gunnison Basin. Trap-zones will be initially defined as: 1) for DAU E-25: Big Blue Creek to Gunnison River (TzA), Gunnison River to Cebolla Creek (TzB), Cebolla Creek to Gold Basin Creek (TzC), and Gold Basin Creek to Cochetopa Creek (TzD), 2) for DAU E-43: Cochetopa Creek to Tomichi Creek (TzE), Tomichi Creek to Quartz Creek (TzF), Quartz Creek to East River (TzG), and 3) for DAU E-41: Ea,st River to Ohio Creek (TzH), Ohio Creek to Dry Creek (Tzl), and Dry Creek to Curecanti Creek (TzJ). • Elk will be captured using a Hughes 500 helicopter and net-guns (contracted services) (Freddy 1994). We will attempt to collar equal numbers of male and female calves. Helicopter trapping will occur in mid-December each year. Capture and handling procedures will follow protocols used to capture 257 calves and 46 adult females on the Grand Mesa (Freddy 1993-1996) and previously approved by CDOW Animal Care and Use Committee (Appendix III). Survival Monitoring Radioed elk will be monitored daily from the ground and bimonthly with aerial surveys (Cessna 185 or equivalent) to determine life/death status of elk. During hunting seasons, aerial surveys will be occur bimonthly in September and weekly during October and November. RADIOS database program will be used to maintain animal records. November 2000 Final �227 Suspected mortalities will be confirmed using ground searches. Criteria for assigning probable cause of death will include body position, presence of bite or claw marks and sub-dermal hemorrhaging, tracks, drag marks, and tissue samples if available (Wade and Browns 1982, Freddy 1998). Potential causes of death include starvation, accidental trauma, plant poisoning, predation by black bears, mountain lions, coyotes, and domestic dogs, and legal and illegal hunter harvest (Freddy 1997). Survival Sample Sizes and Tests Each year we will radio-collar 78 calves (39 male, 39 female) with 26 calves marked in each DAU and during the initial year, 39 adult females will be radio-collared with 13 in each DAU (Table 1). We anticipate >20 radioed female calves will be recruited to yearling adults each year resulting in >50 radioed adult females in the population to estimate adult survival in subsequent years. However, by not collaring known adult females each year, we run the risk of having biased estimates of adult female survival because the age structure of collared adult females will progressively be biased to younger aged females recruited from marked calves. An alternative would be to mark enough adult females in each subsequent year to replace those adult females marked in year 1 that had died the previous year. We anticipate needing to replace 15-20 older adult females per year to achieve this goal which will be dependent on future funding. If adult females could be replaced yearly, we would be able to separate year effects from age effects on survival rates. Approximately 30 yearling males will be available each year, 2001-2003, to estimate percent of yearling males illegally removed under a hunting system using antler-point regulations to protect yearlings. Approximately 30, 2:2-year-old males will be available each year, 2002-2004, to estimate percent of branch-antlered males removed with a hunting system using antler-point regulations. We chose to mark 78 calves per year and 39 adult females during the initial year because we will have acceptable confidence intervals about mean estimates of survival each year for all DA Us pooled into 1 elk population, have the potential to detect major differences in survival between years due to changes in winter severity when all 3 DAUs are pooled, and be able to detect major differences in survival between DA Us when data are pooled within DAUs for 3 years. The ability to detect differences between DA Us within years is desirable but economically prohibitive due to numbers of collared elk required (>47 per DAU per year). We anticipate yearly calf survival to be 0.70 to 0.90 and adult survival, exclusive of hunting-related deaths, to be 0.95 to 0.99. If calf survival is 2:0.70 (n = 78 calves), 95% confidence intervals (Zar 1984, 378) will be ,:s ±15% of the yearly mean survival rate. If adult female survival is 2:0.95 (n 2: 39 adults), 95% confidence intervals will be ,:s ±10% of the yearly mean survival rate. Additionally, if adult female survival= 0.95 we expect to estimate yearly survival within ±5% of the true survival rate at alpha= 0.10 when n 2: 50 adult females. November 2000 Final �228 Table 1. Elk calves (6 months old) and adult females (2:12 months old) captured and radio-collared for Gunnison elk DAUs E-25, E-41, and E-43, December 2000-2002 (shaded cells). Adult females captured only during initial year, 2000. Numbers of radioed adult males and females in December years 2001-05 estimated by assuming survival rates between years: adult females net rate= 0.7, male and female calves to yearling adult age net rate= 0.8, yearling males to adult males net rate = 0.9, adult males net rate= 0.3. DAUE-25 Calves DAU E-41 Adults Calves DAU E-43 Adults C alvcs ALLDAUs Adults Calves Adults Totals Year M F M F M F M F \1 F M F M F M F All 2000 13 13 0 13 13 13 0 13 13 13 0 13 39 39 0 39 117 2001 13 13 10 19 13 13 IO 19 13 13 10 19 39 39 30 57 165 2002 13 13 13 23 13 13 13 23 13 13 13 23 39 39 39 69 186 2003 14 26 14 26 14 26 42 78 120 2004 4 18 4 18 4 18 12 54 66 2005 1 13 1 13 1 13 3 39 42 42' 112' 42' 112' 42' 112' 336' 696' All, -39 39 39 39 39 39 117 126' 117 -~, - ~ ·-- • Represents elk-years and not necessarily numbers of individual radioed adult elk as adults survive between years. Number of collars deployed in combination with actual survival rates determines our ability to detect differences in survival among years, DA Us, or geographic areas. When survival rates are near 0.50, variance, or precision, about the mean survival estimate is largest, and thus the sensitivity to detecting differences in survival rates is least (Zar 1984). As survival rates approach 0.0 or 1.0, precision improves for a fixed sample size of collars, and sensitivity to detecting differences in survival increases. Given our assumptions about expected average survival rates and potential higher calf survival in DAU E-25 based on computer modeling (Freddy 2000), we estimated the statistical power (Snedecor and Cochran 1967; 113, 221, 269; pers. comm. D. Bowden) to detect differences in mean survival rates given specific hypotheses. We consider detecting differences in survival of 0.20 with statistical power of 0.80 at an alpha= 0.10 to be acceptable. Generalized hypotheses (S = survival rate) and power for detecting major differences in survival among years, DA Us, age and sex classes, and geographic areas. •· • (1) Ho: scalvesyearl = scalvesyear2 = scalvesyear3 for DAUs pooled each year. HA, scalvesyearl "F scalvesyear2 "F scalvesyear3 forDAUs pooled each year. Power= 0.80 at alpha= 0.10 to detect differences in yearly survival of 0.15 between pairs of years given 78 collars per year and expected survival rates of 0.90 and 0.75. Power= 0.80 at alpha= 0.10 to detect differences in yearly survival of0.15 between 1 year with lower survival and the average higher survival of the other 2 years given 78 collars per year and expected survival rates of 0.75 and 0.90. (2) Ho: scalves DAUi = scalvos DAU2 = scalves DAUJ for years pooled for each DAU. HA: scalwsDAUl "F scalwsDAU2 "F scalwsDAUJ for years pooled for each DAU. Power= 0.80 at alpha= 0.10 to detect a difference in 3-year average survival of 0.15 between pairs of DAUs given 26 collars per year per DAU and expected yearly survival rates of 0.90 and 0.75. November 2000 Final �229 Power= 0.90 at alpha= 0.10 to detect difference in 3-year average survival of 0.15 between 1 DAU with higher survival and the average lower survival of the other 2 DA Us given 26 collars per year per DAU and expected yearly survival rates of 0.90 and 0.75. Power= 0.90 at alpha= 0.10 to detect difference in 3-year average survival of0.15 between 1 DAU with higher survival and the average lower survival of the other 2 DA Us given 26 collars per year per DAU and expected yearly survival rates of 0.90 and 0.80 and 0.70 amongDAUs. (3) Ho: Smale cah'es = sfem:lle calves for years pooled for each seL HA: Smale calves 'F sfemalecalves for years pooled for each sex. Power= 0.90 at alpha= 0.10 to detect difference in 3-year average survival of 0.15 between sexes of calves given 35 collars per year per calf sex and expected survival rates of 0.75 for one sex and 0.90 for the other sex. (4) Ho: sadultremalesyearl = sadultremalesyear2 = Sadultremalesyear3 for DAUs pooled each year. HA, sadultremalesyearl 'F sadultremalesyear2 'F Sadultremalesyear3 for DAUs pooled each year. Power = 0. 80 at alpha = 0 .10 to detect difference in survival of 0 .15 between pairs of years given 56 collars per year and expected survival rates of 0.95 and 0.80. Power = 0. 80 at alpha = 0 .10 to detect differences in yearly survival of O.15 between 1 year with lower survival and the average higher survival of the other 2 years given 50 collars per year and expected survival rates of 0.80 and 0.95. (5) Ho: scalves = sadult females Ho: Scalves 'F sadult females Power= 0.80 at alpha= 0.10 to detect difference of 0.15 between calf and adult female survival within each year given 56 collars per year per age class and expected survival rates of 0.80 for calves and 0.95 for adult females. Power= 0.90 at alpha= 0.10 to detect difference in 3-year average survival of0.10 between calves and adult females given 51 collars per year per age class and expected survival rates of 0.85 for calves and 0.95 for adult females. (6) Ho: scalves Gunnison = scalves Grand Mesa for years pooled within each area. HA: scalves Gunnison 'F scalves Grand Mesa for years pooled within each area. Power= 0.80 at alpha= 0.05 to detect difference in a 3-year average survival of 0.10 between calf survival in the Gunnison Basin and calf survival on the Grand Mesa given 66 collars per year per area and expected survival of 0.80 in the Gunnison Basin and 0.90 on the Grand Mesa. Survival will be estimated for calves and adults during winter-spring (15 Dec -14 Jun), for adults during summer-fall (15 Jun-14 Dec), and for adults during the year (15 Dec - 14 Dec). The yearly time period, or biological year, initiates with capture and release of marked elk into the population (White et al. 1987). Capture of elk will occur in mid-December instead of early December as on the Grand Mesa (Freddy 1993-1997) to accommodate capture services on other projects. We expect this change in capture dates to have minimal effects on estimates of survival as no natural deaths of calves or adults occurred during December on the Grand Mesa (Freddy 1999). We will use a staggered entry Kaplan-Meier analysis to estimate survival rates (SAS 1988, White and Garrot 1990, Bartmann et al. 1992). We will compare survival rates using chi-square analyses and conduct pair-wise comparisons using log-rank tests to compare survival of calves and adults among years for DA Us combined, between DA Us for years combined, between male and female calves, and between calves and adults. We will assess whether calf survival can be predicted from sex, body weight, hind November 2000 Final �230 foot length, total body length, and mean monthly snow depths and temperature using logistic regression (SAS 1988). Additionally, we will test for differences in survival of calf and adult elk between the Gunnison Basin and the Grand Mesa (Freddy 1998) potentially using beta-binomial distribution approaches outlined by Unsworth et al. 1999. Tests will be significant at alphaP S 0.10. PREGNANCY RA TES Fecundity of adult female elk will be determined by examining reproductive organs of antlerless elk harvested during hunting seasons from mid-November through December. Initially, late seasons are scheduled to occur in 2000, but may continue in subsequent years depending upon population management objectives. Numbers of hunters will be controlled by limited permits issued each year. During 2000, we anticipate >650 hunters will provide ~200 useable reproductive tracts from antlerless elk harvested in portions of DAUs E-25, E-41, and E-43. Hunters will be mailed packets explaining procedures for collecting reproductive organs and incisor teeth from harvested elk as done previously in Colorado for Middle Park and Forbes-Trinchera elk collections (Freddy 1992, pers. comm. C. Wagner, CDOW). Additionally, we will ask hunters to collect kidneys and associated fat from harvested elk to allow calculation of kidney-fat indices to better assess body condition of adult females in relation to reproductive status (Kohlmann 1999). Hunters will be directed to leave collected organs at drop-off sites in Lake City, Colorado, Gunnison CDOW Service Center, Gunnison commercial meat-processors, and at CDOW Roaring Judy Hatchery. Fetuses will be sexed, weighed, and measured (Armstrong 1950) with conception dates estimated from fetal measurements (Morrison et al.1959). Pregnancy status, fetal age, fetal sex, and conception dates will be related to female dental cementum age and kidney fat indices using regression analyses (SAS 1988). Additionally, comparisons to reproductive measurements on elk from Middle Park and ForbesTrinchera will be made. POTENTIAL ADDITIONAL EXPERIMENTS AND APPLICATIONS Management of elk in the Gunnsion Basin has contentiously focused on population status of elk, impacts of elk on plant communities, long-term carrying capacities for wild and domestic ungulates and seasonal patterns of habitat use (Carpenter et al. 1980, Roath et al. 1999). Expanding our understanding of these general topics can be greatly enhanced by effectively utilizing the radio-collared elk that will be available because of this project. Investigations regarding these topics could be initiated with additional funding, personnel, and agency cooperation. . . Potential investigati9ns could address:. a). Management objectives·as of 1999 are to reduce elk populations in DAUs E-25, E-41, and E43. Reductions are projected to be most severe in DAU E-25 and approach 50% over the next 5 years based on computer models. Reductions in DAUs E-41 and E-43 are projected to be <25% and completed in 2-3 years. If elk are indeed at biological carrying capacity and if reductions proceed in E-25, there may be the opportunity to conduct management experiments to assess whether calf survival and/or fecundity increase in response to lowered density. Radio-collaring and monitoring additional calves each year would be required. Estimated additional costs could approach $50,000. b). Population reductions may also create an opportunity to apply sampling systems developed to estimate elk density, including mark-resight estimators (Freddy 1998), to verify modeled population status and achievement of populations goals. Estimated additional costs would be in helicopter hours ($40,000) and additional radio-collared elk ($40,000) for mark-resight surveys. Additionally, sampling systems to estimate sex ratios could be implemented and evaluated in E25 with reallocation of existing survey monies plus an additional $10,000. November 2000 Final �231 c). Patterns of habitat use and forage removal could be investigated utilizing intensive measurements on selected range sites and monitoring ofradioed elk and their associates. This would be a major project and possibly approach $100,000 per year including additional personnel. d). Seasonal movements and patterns of spatial use to document seasonal behavior of elk would require additional personnel and aerial fixed-wing costs of $40,000 per year. PROJECT SCHEDULE Fiscal Year 2000-01 Activity/Objective Complete study plan; purchase radio-collars; Estimate pregnancy/fetal rates; Trap and radio-collar elk and estimate survival. Period Jul-Nov Nov-Dec Dec-Jun 2001-02 Estimate survival and hunting removal rates; Estimate pregnancy/fetal rates; Trap and radio-collar elk and estimate survival. Jul-Dec Nov-Dec Dec-Jun 2002-03 Estimate survival and hunting removal rates; Estimate pregnancy/fetal rates; Trap and radio-collar elk and estimate survival; assess potential for mark-resight estimates of elk density. Jul-Dec Nov-Dec Dec-Jun 2003-04 Estimate survival and hunting removal rates; complete data analyses, initiate manuscripts. Jul-Jun Estimated Annual Costs FTE Requirements PFTE = 1.00 TFfE= 0.83 TOTAL= 1.83 Budget Category (01) Personal Services (21) Operating Supplies and Services (21) Utilities (28) Travel Expenses (31) Capital Outlay Total Costs Costs $102,000 84,000 0 1,000 0 $187.000 Costs anticipated to increase 5% each year in 2001-02, 2002-03, 2003-4 for inflation. November 2000 Final �232 Personnel Program Responsibilities David J. Freddy: Wildlife Researcher, Principal Investigator responsible for final project design, organizing field personnel, obtaining and organizing data, data analyses, financial control, and coordinating publications. R Bruce Gill: Wildlife Research Leader, provides administrative support, input for study design, and liaison with other administrative sections within the Division of Wildlife. Rick Kahn, John Ellenberger, Jim Olterman, Don Masden, Jim Young: Provide coordination and support of Terrestrial managers and biologists and Area management staff and facilities. Gary C. White: Provide input for study design and statistical protocol, conduct data analyses, and provide software support. David C. Bowden: Provide input for study design and statistical protocol. E. LOCATION The Gunnison Basin in south-central Colorado was selected for this project (Fig. 1). The Basin encompasses the entire headwaters of the main Gunnison River and the centrally located town of Gunnison. Between 12-16,000 elk and 8-10,000 mule deer (Odocoileus hemionus) are thought to exist within the Basin. Elk are managed as 3 populations representing DAUs E-25 (Game Management Units [GMU] 66, 67), E-41 (GMU 54), and E43(GMUs 55,551). The 3 DAUs encompass about 9,291 km2 of which 3,648 km2 are considered winter range for elk (CDOW WRIS database). DAUs are contiguous with no major geographic barriers separating DAUs that would prevent interchange of elk among DAUs. t N ···················••••••••••••••••••••• 50 Miles The Basin represents a high altitude, cold winter range for both elk and mule deer which BO Km is similar to ecosystems in North Park, Middl~ Fig. 1. Location of the Gl:llllison Basin and elk Data Park, and the San Luis Valley, Colorado. The Analysis Units E-25, E-41, and E-43 within Colorado. sagebrush steppe winter ranges (2,250- 2,700 m elevation) can receive extreme snow depths and cold temperatures that cause severe mortality among ungulates (Carpenter et al. 1984) while the conifer-alpine summer ranges (3,000 - 4,200 m elevation) can be subjected to drought. Overall, these ranges collectively are thought to be less productive and nutritious for elk than the milder climate oakbrush-pinyon-juniper winter ranges of the Grand Mesa where elk survival was measured from 1993-99. We anticipate dependable access to both private and public lands to conduct research activities and there is local, Area, and Regional CDOW support for conducting the project in this area. Additional financial and logistical support may be available from the Gunnison Habitat Partnership Committee. The airport and other businesses in Gunnison will provide readily accessible support services. November 2000 Final �233 F. RELATED FEDERAL AID PROJECTS Calf and adult elk survival rates were measured on the Grand Mesa, Colorado from 1993-99 under Federal Aid Research Project W-153-R (Freddy 1994-1999). G. LITERATURE CITED Anderson, C.R, Jr., D.S. Moody, B.L. Smith, F.G. Lindzey, and R. P Lanka. 1998. Development and evaluation of sightability models for summer elk surveys. Journal of Wildlife Management 62:1055-1066. Armstrong, R.A. 1950. Fetal development of northern white-tailed de~i:- American Midland Naturalist 43:650-666. Bartholow, L 1999. POP-II system documentation Windows TM Version 1.0. Fossil Creek Software, Fort Collins, Colorado USA. Bear, G.D., G.C. White, L.H. Carpenter, RB. Gill, and DJ. Essex. 1989. Evaluation of aerial markresighting estimates of elk populations. Journal of Wildlife Management 5 3: 908-915. Biutrnann RM., G.C. White, and L.H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monographs 121. Carpenter, L.H., D.L. Baker, and R.B. Gill. 1980. Tests of a nutritionally based big game habitat •· ~valuation system. Colorado Division of Wildlife Unpublished Report. ColoradcfDivision of • -- Wildlife, Fort Collins, Colorado, USA. Carpenter, L.H_, R. B. Gill, D. L. Baker, and N.T. Hobbs. 1984. Colorado's big game supplemental winter feeding program. Colorado Division of Wildlife, Fort Collins, Colorado USA. Conner, M.M. 1999. Elk movement in response to early-season hunting in the White River Area, Colorado. Dissertation, Colorado State University, Fort Collins, Colorado, USA. Cogan, RD_, and D.R. Diefenbach. 1998. Effect ofundercounting and model selection on a sightabilityadjustrnent estimator for elk. Journal of Wildlife Management 62:2~9-279. Eberhardt, L.L., RA. Garrott, PJ. White, and PJ. Gogan. 1998. Altematiye approaches to aerial censusing of elk. Journal ofWildlife Management 62:1046-1055 .. Freddy, DJ. 1992. Effect of elk harvest systems on elk breeding biology. Colorado Division of Wildlife Research Report July: 45-70. Fort Collins, Colorado, USA. July:45-70. . Freddy, D .J. 1993. Program Narrative for Estimating survival rates of elk and developing techniques to estimate population size. Colorado Division of Wildlife Research Report July:83-117. Fort Collins, Colorado, USA. • Freddy, DJ. 1994. Estimating survival rates of elk and developing techniques to estimate population size. Colom.do Division of Wildlife Research Report July: 27-42. Fort Collins, Colorado, USA. Freddy, DJ. 1995. Estimating survival rates of elk and developing_ techniques to estimate population size. Colorado Division of Wildlife Research Report July: 63-79. Fort Collins, Colorado, USA. Freddy, DJ. 1996. Estimating survival rates of elk and developing techniques fo estimate population size. • Colorado Division of Wildlife Research Report July: 87-108. Fort Collins, Colorado, USA. Freddy, DJ. 1997. Estimating survival rates ofelk and developing techniques to estimate population size. Colorado Division of Wildlife Research Report July: 47-73. Fort Collins, Colorado, USA. Freddy, DJ. 1998. Estimating survival rates of elk and developing techniques to estimate population size. Colorado Division of Wildlife Research Report July: 177-206. Fort Collins, Colorado, USA. Freddy, DJ. 1999. Estimating survival rates of elk and developing techniques to estimate population size. Colorado Division of Wildlife Research Report July: (in press). Fort Collins, Colorado, USA. Freddy, DJ. 2000. Modeling elk populations in the Gunnison Basin, Colorado using POPII and POPMOD software (draft in process). Colorado Division of Wildlife Special Report No.??, Fort Collins, Colorado USA. Freddy, DJ., D.L. Baker, R.M. Bartrnann, and RC. Kufeld. 1993. Deer and elk management analysis guide, 1992-1994. Colorado Division ofWildlife Division Report 17, Fort Collins, Colorado USA. November 2000 Final �234 Garrott, RA., R.M. Bartrnann, and G.C. White. 1985. Comparison ofradio-transmitterpackages relative to deer fawn mortality. Journal of Wildlife Management 49:758-759. Houston, D.B.: 1982. _The,nqrthemYellowstone elk, ecology and management. Macmillan Publishing Company, Inc. New York, New York, USA. Kohlinann,'S.G. 1999. Adaptive fetal sex allocation in elk: evidence and implications. Journal of • Wildlife Management 63: 1109-1117. Morrison, J.A., C.E. Trainer, and P.L. Wright. 1959. Breeding seasons in elk as determined from knownage embryos. Journal of Wildlife Management 23:27-34. Nelson, L.J., and J.M. Peek. 1982. Effect of survival and fecundity on rate of increase of elk. Journal of Wildlife Management 46:535-540. _. Petersburg, M., and G. White. 1998. Kaplan-Meier survival estimates for cow elk. Colorado Division of Wildlife Terrestrial Section Unpublished Memorandwn, Fort Collins, Colorado USA. Phillips, G.E. 1998. Effects ofhwnan-induced disturbance during calving season on reproductive success of elk in the upper Eagle River Valley, Colorado. Dissertation, Colorado State University, Fort Collins, Colorado, USA. Quimby, D.C., and J.E. Gaab. 1957. Mandibular dentition as an age indicator in Rocky Mountain elk. Jqurnal of Wildlife Management 21: 134-153. Roath, R)L. Carpenter, :B, R.4epsame, and D. Swift. 1999. Gunnison Basin habitat assessment project. ,Fi¼al Report Mafch i"999. Colorado State University, Department of Range Science, Fort Collins, • 'Colorado USA. Samuel, M.D., E.O. Garton, M.W. Schlegel, and R G. Carson. 1987. Visibility bias during aerial surveys of elk in northcentral Idaho. Journal of Wildlife Management 51: 622-630. SAS. 1988. SAS/STAT user's guide, release 6.03. SAS Institute, Inc. Cary, North Carolina USA. Snedecor, G.W., and W.G. Cochran. 1967. Statistical methods; sixth edition. Iowa State University Press, Ames, Iowa, USA. Singer, F.J., A. Harting, K.K. Symonds, and M.B. Coughenour. 1997. Density dependence, compensation, and environmental effects on elk calf mortality in Yellowstone National Park. Journal of Wildlife Management 61: 12-25. Unsworth, J .W ., L. Kuck, and E. 0. Garton. 1990. Elk sightability model validation at the National Bison Range, Montana. Wildlife Society Bulletin 18:113-115. Unsworth, J.W., D.F. Pac, G.C. White, and R.M. Bartrnann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. Wade, D.A., and J.E. Browns. 1982. Procedures for evaluating predation on livestock and wildlife. Texas Agricultural Experiment Station Publication B-1429. White, G.C. 1992. DEAMAN database manager and populatj.on modeling procedures; Colorado Division of Wildlife User',s manual and reference. Colorado State University, FQrt Collins, Colorado USA. White, G.C. and R.A. Garrott. 1990. Analysis'.ofwildlife radio-tracking data. Academic Press, Inc., San Diego, California USA. White, G.C., D.R. Anderson, K.P. Burnham, and D.L. Otis. 1982. Capture-recapture and removal methods for sampling closed populations. Los Alamos National Laboratory LA-8787-NERP, Los Alamos, New Mexico, USA. White G.C., R.A. Garrott, RM. Bartrnann, L.H. Carpenter, and A.W. Alldrege. 1987. Survival of mule deer in northwest Colorado. Journal of Wildlife Management 51: 852-859. Zar, J.H. 1984. Biostatistical analysis, second edition. Prentice-Hall, Englewood Cliffs, New Jersey, USA. November 2000 Final �235 APPENDIX I SPECIFICATIONS FOR RADIO-COLLARS Manufacturer: Lotek, Inc. Pulse Rate Nonnal: 60-65 ppm Pulse Rate Mortality: 120-130 ppm Motion Sensor Delay: 4-6 hrs Batteries: 4+ year life, 1 lithium D-cell calf collars Antenna: External whip, pvc coated Collar Material: 7.6 cm (3") wide white colored smooth surfaced conveyor belting, 2 layers sewn together, 0.64 cm (1/4") total thickness Additional Material: Bright yellow with black core Ritchie All-Flex plastic material for identification symbol/number placed as a sleeve over top portion of collar (Ritchey Manufacturing, Inc., Brighton, CO). Collar Size: 61-81 cm (24-32") adult females, individually fitted; 56-69 cm (22-27") expandable for female calves; 57-89 cm (22.5-35") expandable for male calves. Collar Weight: 820 gm female calves; 840 gm male calves; 1.1 kg adult females. November 2000 Final �236 APPENDIX II VISUAL IDENTIFICATION SYSTEM FOR RADIO-COLLARS Numbers, symbols, and letters will be used in ordered combinations to quickly allow identification of individual elk primarily during aerial surveys. No more than 2 characters will be used to identify an individual. Characters will be ordered and read from left to right on the collar from the perspective of looking down on the elk from the rear of the animal when approached by an observer in a helicopter. Numbers to be Used (8): 0, 1, 2, 3, 4, 5, 6, and 7 Symbols to be Used (5): solid circle e, solid square ■, solid triangle A, solid hourglass X, plus sign +, (solid diamond potentially ♦ ) Letters to be Used (9): A, C, F, H, J, K, N, P, X (T, V, Ypotentially) Identification Combinations: Number combinations represent 56 individuals 10,20,30,40,50,60, 70 11, 21, 31, 41, 51, 61, 71 12,22,32,42,52,62, 72 13,23,33,43,53,63, 73 14,24,34,44,54,64, 74 15,25,35,45,55,65, 75 16,26,36,46,56,66, 76 17,27,37,47,57,67, 77 Each Symbol paired with each Number represents 16 identification codes when ordered symbol-number and then number-symbol. Five symbols paired with 8 numbers represents 80 individuals. Examples: 7eand e7. Each Letter paired with each Number represents 16 identification codes when ordered letter-number and then number-letter. Nine letters paired with 8 numbers represents 144 individuals. Examples: A 7 and 7A. Each Letter paired with each Symbol represents 10 identification codes when ordered letter-symbol and then symbol-letter. Nine letters paired with 5 symbols represents 90 individuals. Examples: Ae and eA. Therefore, a minimum of370 different animals can be individually marked using this system. November 2000 Final �237 APPENDIX III HELICOPTER NET-GUNNING CAPTURE PROTOCOL FOR ELK Background: Helicopter net-gunning has been successfully and safely used to capture and radio-collar elk in Colorado during both winter and summer. This success has been in part due to following accepted protocols for handling elk (Colorado Division of Wildlife Animal Care and Use Committee Reviews and Approvals). Helicopter capture of elk on the Grand Mesa, Colorado during December 1993-1996, resulted in no acute or post-capture related deaths in 46 adult females and I acute death (broken neck, 0.4%) and 2 post-capture myopathy deaths (0.7%) in 258 calves captured and handled (Freddy 1996, 1997). During early December 1994-96 near Vail, Colorado, 2.2% of 185 adult females died from effects of helicopter capture (Phillips 1998). In the White River, Colorado, <I% of 95 adult female elk captured during July and 4% of 32 adult females captured during August near Vail, Colorado died from effects of helicopter capture (Conner 1999, pers. comm. M. Conner, 1999, Phillips I 998). Capture Protocol: Capture of elk will follow procedures successfully used on the Grand Mesa (Freddy 1995). David J. Freddy is the principal investigator and will coordinate capture of elk. All persons involved in the capture operation, including the helicopter net-gunning crew, will be instructed on proper care and handling of elk to reduce stress and injury to elk. Capture Timing and Conditions: Elk are scheduled to be captured during mid-December in the Gunnsion Basin but there remains the possibility that capture could occur in early January depending on availability of contract helicopter services. During either month, cool ambient temperatures and moderate snow depths (<60cm) contribute to successfully capturing elk by reducing threats of hyperthermia potentially induced by capture chases. We anticipate capturing elk when ambient temperatures are -18 - 3°C. Temperatures < -l 8°C (0° F) may restrict human efficiencies while temperatures >3°C (38° F) may induce hyperthermia in elk. No-fly Zones: Pursuit and capture of elk will not occur within 1,000m (0.5 miles) of human residences or other cultural developments such as well traveled roads, reservoirs, etc .. Notification of Affected Parties: Local residents and federal, state, and local agencies will be notified of the time and general area of capture activities. Notification will be via newspaper articles, public meetings, and other informal verbal communications. Emergency Services: Capture personnel will be instructed that the nearest medical and emergency services are located at the Gunnison Valley Hospital in Gunnison. Capture crews will have communications radio contact to CDOW service centers and emergency Colorado State Patrol. Radio Collars/Ear Tags: Expandable collars will be placed on male and female calves to accommodate neck growth as animals become adults (Appendix I). Collar design was previously used on 285 elk calves on the Grand Mesa with no known cases of expandable collars inducing trauma for up to 4 years of age on males and 7 years of age on females (Freddy 1999). Collars of fixed size will be placed on each adult female and individually fitted usually to 69-74 cm (27-29 in). Fixed collar design was previously used on 82 adult females with no known cases of trauma (Freddy 1999). No ear-tags will be used. Command Post: The principal investigator and handling crew will establish mobile command /handling sites that will be near actual locations of capture. The handling crew will be ferried by the capture helicopter as needed. At these command posts, elk calves will be weighed, measured, collared and released while adult female elk will be captured and released at the point of capture. This will facilitate efforts of the principal investigator to remain in contact with the helicopter net-gunning crew and make all decisions regarding care and welfare of captured elk. Chase Time: The helicopter crew will locate groups of elk and determine if calves and adult females are present. If the group is >20 animals, the helicopter will splinter the group into smaller groups within 1-2 minutes of detecting the initial group. The helicopter will then spend <5 minutes maneuvering a smaller group to a suitable capture site. Once a target animal is selected it will be actively pursued for :::I minute or until active panting is observed at which time the pursuit is terminated. Total time spent disturbing the initial group and target animal should be <10 minutes. No more than 2-3 animals will be taken from an November 2000 Final �238 initial group to avoid unnecessary chase time of non-target animals. Care will be taken by the helicopter crew to avoid chasing animals into fences, roads, rivers, or unfavorable terrain. Animal Care and Handlin2: Elk calves will be hobbled and blindfolded at the point of capture and then slung under the helicopter, one calf per ferry, to a nearby command/handling site where they will be measured, weighed, collared, and released at that site. Capture locales will be within 1-2 minutes of flying time or <3,000m (2 miles) of command/handling sites. Adult females will be blindfolded, hobbled, collared, aged, and then released at the point of capture by the net-gunning crew. Adult females will be assigned to an age class based on relative wear and height of incisors: yearling, 2-4 years, 5-9 years, and >9 years (Quimby and Gaab 1957). At the handling site, 3-4 persons will handle and release calves. Calves will be gently lowered to the ground by the helicopter near the handlers at which time handlers will check calf for injuries, remove netting, and check blindfold and hobbles for proper function. Rectal temperature will then be measured using a digital thermometer (°F) while measurements of total body length (cm), hind foot length (cm) are being obtained. If rectal temperature is 2:41.9°C (107.4°F) and heavy panting evident, the calf will be only collared and released and not weighed to reduce handling time. Previous experience on the Grand Mesa indicated calves survive when rectal temperatures briefly approach 42.2°C (l08°F) (Freddy, unpubl. data). The 2 cases of capture myopathy on the Grand Mesa were males with rectal temperatures of 42.2 and 41.5°C (100 th and 90th quantiles, respectively), ambient air temperatures -2.2 and 3 .9°C (<50 th and 90th quantiles, respectively), and below average body weights <108kg. Assuming acceptable body temperature, the calf will be weighed (kg) by gently sliding the calf into a weigh-bag which will support the entire weight of the calf while the calf is hoisted by a pulley and suspended from a scale affixed to a portable steel quad-pod. Care will be taken to always support the spine and neck of the calf during the weighing process. Once weighed, calves will be lowered to the ground, slid out of the bag, radio-collared, hobbles removed, blindfold removed, and released towards the direction from which they were ferried by the helicopter. Previous experience on the Grand Mesa indicated calves readily find and join elk groups after being released. Total time to process and release calves should be :::8 minutes. If ambient air temperature exceeds 3.3°C (38 °F), capture activities will likely be halted, especially if snow is not present to help cool captured elk. Injured Animals: We expect <3% serious injury/mortality rate. Capture techniques will be constantly monitored and changed if necessary to insure that minor injuries to animals do not chronically occur. However, any debilitating injury or mortality of a captured elk will cause at least temporary suspension of capture activities to assess the cause of injury and if further injuries can be prevented. Animals having a broken leg, neck, pelvis, or other debilitating wound will be euthanized with a gunshot to the head _(0.357 or larger caliber pistol) following euthanasia protocols of the Colorado Division of Wildlife Animal Care and Use Committee. The principal investigator will make decisions regarding euthanasia but all persons involved in capture will be trained to properly euthanize appropriate animals. The helicopter net-gunning crew and ~e handling crew will both have ready access to pistols needed for euthanasia. Euth.anized animals will be processed for human cons~ption ~d donated to social service agencies. Release of Animals: While still blindfolded, hobbled, and prior to release, elk will again be examined for injuries. Superficial injuries such as abrasions and small cuts will be treated with antibiotic ointment. The release sequence will be to place elk in sternal recumbency with head pointed towards direction of capture, remove hobbles, remove blindfold, physically hold elk until elk regains eyesight and orientation, at which time handlers release elk and help elk maintain its balance and upright position. Elk will then be observed for any signs of injury while moving away from handlers. Care will be taken to avoid releasing elk towards fences or unfavorable terrain. Post-Capture Monitorin2: All radioed elk will be monitored for their life/death status 2:2 times within 10 days of capture. If a mortality occurs, the carcass will be located, necropsy performed, and cause of death estimated if possible. If available, muscle tissue samples will be collected and sent to Colorado State University Veterinary Diagnostic Laboratory to detect evidence of capture myopathy. November 2000 Final �71 JOB PROGRESS REPORT Stateof _ _ _ _ _~C~o=lo=r=a=d~o_ _ _ _ __ Work Package -~3C-'0'-"0=2'----------Task No. -----=3_ _ _ _ _ _ _ _ __ Federal Aid Project. Division of Wildlife - Mammals Research Elk Conservation Estimating Calf and Adult Survival Rates and Pregnancy Rates of Gunnison Basin Elk W-153-R-16 Period Covered: July 1, 2002- June 30, 2003 Author: D. J. Freddy Personnel: L. Gepfert, D. Masden, R. Basagoitia, L. Spicer, B. Carochi, J. Oulton, T. Beck, C. Mehaffey, D. Williams, J. Johnston, and R. Kahn of CDOW, Dr. G. C. White Colorado State University, and cooperators/contractors Gunnison Basin Habitat Partnership Program, M. Schuette of MountainScape Imaging, L. Coulter of Coulter Aviation, USFS, BLM, private land owners, and elk hunters. ABSTRACT We used aerial and ground surveys to estimate survival rates and assess sources of mortality for radio-collared adult elk (Cervus elaphpus nelsonii) in the Gunnison Basin of Colorado. Between 15 December 2000 and 14 June 2003, hunting accounted for 94% and 79% of the adult, age ::=:12 months, female and male deaths, respectively, while natural causes were attributed to 6% and 21 % of the adult female and male deaths, respectively. During 3 winter-spring intervals, 15 December - 14 June, natural survival rates for adult females, age ::=:18 months, were::=: 0.98 (n = 39-86 elk, 148-168 elk-winters). During 2 summer-fall intervals, 15 June - 14 December, natural survival rates for adult females, age 2:12 months, were ::=:0.97 (n = 37-86 elk, 98-157 elk-summers). Including hunting mortalities reduced summer-fall female survival to 0.91 ± 0.07 in 2001 (n = 77) and 0.77 ± 0.08 in 2002 (n = 112). During 2 annual intervals, 15 December to next 14 December, natural survival rates for adult females, age::=: 18 months, were ::=:0.97 (n = 33-61). Including hunting mortalities reduced annual female survival to 0.92 ± 0.08 in 2001 (n = 39) and 0.74 ± 0.09 in 2002 (n = 82). Natural survival rates for 2 cohorts of yearlings, age 12-23 months, were 1.00 for females (n = 59) and 0.93 ± 0.08 (n = 43) for males. Including hunting mortalities reduced cohort survival to 0.87 ± 0.08 (n = 68) for females and 0.82 ± 0.11 (n = 49) for males. During summer-fall, natural survival rate for male elk, age 24-29 months, was 1.00 (n = 14) which was reduced to 0.74 ± 0.22 (n = 19) by including hunting mortalities. During winter-spring, natural survival rate for male elk, age 30-35 months, was 1.00 (n = 13). Predation by mountain lions or black bears was suspected in 4 of the 5 adult elk natural deaths. Hunting removal rates for adult females, age ::=:12 months, were 0.08 ± 0.06 (n = 76) in 2001 and lower than the 0.23 ± 0.08 (n = 112) in 2002 (P = 0.006). Removal rates for yearling females, age 12-17 months, averaged 0.13 ± 0.08 (n = 68). Removal rate for yearling • males averaged 0.13 ± 0.10 (n = 48) and for legal branch-antlered males was 0.26 ± 0.22 (n = 19). Wounding loss as a percent of legal harvest was 44 for all adult females and O for branch-antlered males. All hunting deaths of yearling males were illegal harvest/wounding loss while removal rate for branchantlered males was unexpectedly low, likely representing a year effect on elk vulnerability. Apparent differences in survival of adult females between DA Us (P .::S 0.063) likely reflected geographic differences in vulnerability of elk to hunting while differences in male survival between DA Us (P = 0.046) reflected impacts of illegal harvest/wounding loss on removal of yearling males. Adult female elk body condition suggested marginally deficient levels of seasonal nutrition in 2002. �72 Distribution and movements of radio-collared elk during 3 years of monitoring revealed that elk had a relatively high fidelity to the Gunnison Basin as defined by current DAU boundaries but elk also commonly ventured into adjoining GMUs outside the Gunnison Basin. Distribution patterns revealed minimal interchange of elk between areas north and south of U.S. Highway 50 which bisected the Gunnison Basin from east to west. Movements by adult females, young females, and young males (n = 35, 48, and 76) suggested DAU elk population management boundaries might be altered to better represent elk population units. Young male and female elk tended to move greater distances and exhibit higher rates of venturing into adjoining GMUs than adult females. Patterns of dispersion suggested movement corridors that allowed for genetic linkage between Gunnison Basin and other elk populations. All information in this report is preliminary and subject to further evaluation. �Colorado Division of Wildlife Wildlife Research Report July 2003 – June 2004 JOB PROGRESS REPORT State of Project No. Work Package No. Task No. Colorado 3002 3 : : : : Federal Aid Project: N/A : Cost Center 3430 Mammals Research Elk Conservation Estimating Calf and Adult Survival Rates and Pregnancy Rates of Gunnison Basin Elk Period Covered: July 1, 2003- June 30, 2004 Author: D. J. Freddy Personnel: D. Masden, R. Basagoitia, L. Spicer, B. Diamond of CDOW, Dr. G. C. White Colorado State University, and cooperators/contractors Gunnison Basin Habitat Partnership Program, M. Schuette of MountainScape Imaging, private land owners, and elk hunters. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT During this segment, the transition of monitoring the remaining 119 radio-collared adult elk from research to management biologists was facilitated by providing databases, telemetry equipment, and other guidance as needed. Progress reports were completed, peer-reviewed publications on elk survival rates were initiated, and publications were accepted by peer-reviewed journals. 57 �JOB PROGRESS REPORT ESTIMATING CALF AND ADULT SURVIVAL AND PREGNANCY RATES OF GUNNISON BASIN ELK POPULATIONS DAVID J. FREDDY P. N. OBJECTIVE Estimate survival rates of calf, adult female, and adult male elk and estimate pregnancy rates of adult female elk in Gunnison Basin elk populations for 3 years. NOTE: Prioritization of available research funding resulted in discontinuing efforts to estimate calf survival, pregnancy rates and body condition during 2002-03 but allowed for monitoring adult elk survival through June 2003. SEGMENT OBJECTIVES 1. 2. Facilitate the transition of monitoring the remaining 119 radio-collared adult elk from research to management biologists by providing databases, telemetry equipment, and other guidance as needed. Summarize and analyze data and publish information as Progress Reports, peer-reviewed manuscripts for appropriate scientific journals, or Colorado Division of Wildlife (CDOW) technical publications. SUMMARY Progress reports were completed for the Gunnison Basin elk project (Freddy 2002, Freddy 2003) and can be obtained through the CDOW Research Center library in Fort Collins, Colorado. Publications incorporating calf and adult elk survival rates measured in the Gunnison Basin and Grand Mesa, Colorado were initiated. Two publications were accepted by the Wildlife Society Bulletin for publication during this segment with authors and abstracts provided here for reference. How many mule deer are there? Challenges of credibility in Colorado David J. Freddy, Colorado Division of Wildlife, 317 West Prospect Road, Fort Collins, CO 80526, USA, dave.freddy@state.co.us Gary C. White, Department of Fishery and Wildlife Biology, Colorado State University, Fort Collins, CO 80523, USA Mary C. Kneeland, Colorado Division of Wildlife, 317 West Prospect Road, Fort Collins, CO 80526, USA Richard H. Kahn, Colorado Division of Wildlife, 317 West Prospect Road, Fort Collins, CO 80526, USA James W. Unsworth, Idaho Department of Fish and Game, P.O. Box 25, Boise, ID 83707, USA William J. deVergie, Colorado Division of Wildlife, 2300 South Townsend Avenue, Montrose, CO 81401, USA Van K. Graham, Colorado Division of Wildlife, 711 Independent Avenue, Grand Junction, CO 81505, USA John H. Ellenberger, Colorado Division of Wildlife, 711 Independent Avenue, Grand Junction, CO 81505, USA 58 �Charles H. Wagner, Colorado Division of Wildlife, 222 South Road 1 East, Monte Vista, CO 81144, USA Abstract: Conflict resolution between stakeholder groups and management agencies is a problem in wildlife management. We evaluated our success in resolving a conflict between sportsmen and the Colorado Division of Wildlife (CDOW). Sportsmen challenged the credibility of methods used to estimate numbers of mule deer (Odocoileus hemionus) in Colorado and demanded validating surveys to verify numbers of deer. Sportsmen, other interested wildlife stakeholders, and CDOW engaged in a conflict resolution process and designed and implemented an aerial survey to estimate numbers of deer in a specific population whose previous estimated size had been contested by sportsmen. We used helicopters to count mule deer on randomly selected sample units distributed on deer winter range in March 2001. Estimated population size was 6,782 ± 2,497 (90% CL) using stratified random sample estimators and 11,052 ± 3,503 (90% CL) when counts of deer were adjusted using the Idaho mule deer sightability model. Both aerial survey estimates supported computer modeled population estimates of 7,000 to 7,300 deer that had been contested by sportsmen and all estimates were greater than the sportsmen’s estimate of 1,750 deer which was determined from their casual observations. After the survey, sportsmen did not accept survey estimates despite their involvement in design, analysis, and interpretation of the validation survey. By failing to support results of a validation survey they had demanded, the credibility of sportsmen plummeted among other stakeholders, the Colorado Wildlife Commission, and outside public entities while credibility of CDOW managers rose. We contend that CDOW successfully met challenges of sportsmen because the aerial survey systems used to validate deer numbers were founded on credible science and applied within a resolution process that elicited trust from most stakeholders. We caution other agencies facing similar challenges to use tested methods that can withstand public scrutiny, allow ample time for planning and implementing, carefully assess technical and political risks associated with potential outcomes, and engage multiple stakeholders in planning efforts to gain trust of participants. Cost of this resolution process was about 100,000 $US. Key words: Colorado, conflict resolution, credibility, helicopter surveys, human dimensions, mule deer, Odocoileus hemionus, population estimates, sightability Wildlife Society Bulletin 32 (3):00-00. Effect of limited antlered harvest on mule deer sex and age ratios Chad J. Bishop, Colorado Division of Wildlife, 2300 South Townsend Avenue, Montrose, CO 81401, USA, chad.bishop@state.co.us Gary C. White, Department of Fishery and Wildlife Biology, Colorado State University, Fort Collins, CO 80523, USA David J. Freddy, Colorado Division of Wildlife, 317 West Prospect Road, Fort Collins, CO 80526, USA Bruce E. Watkins, Colorado Division of Wildlife, 2300 South Townsend Avenue, Montrose, CO 81401, USA Abstract: During the 1990s, in response to apparent declining mule deer (Odocoileus hemionus) numbers in Colorado, high buck harvest rates were identified as one of several factors that could be negatively affecting population productivity. Some wildlife managers and sportsmen hypothesized that increasing buck:doe ratios by limiting buck harvest would cause an increase in fawn:doe ratios, and hence, population productivity. We evaluated this hypothesis using data collected by the Colorado Division of Wildlife (CDOW) from 1983 to 1998. Beginning in 1991, CDOW reduced buck harvest in 4 deer management units to provide quality hunting opportunities while maintaining high harvests in other management units. We examined effects of limited harvest on December ratios of bucks:100 does and fawns:100 does using data obtained from helicopter surveys in limited and unlimited harvest units. 59 �Annual buck harvest was reduced by 359 bucks (SE = 133) as a result of limiting licenses in the 4 limited harvest units. Fawn:doe ratios declined by 7.51 fawns:100 does (SE = 2.50), total buck:doe ratios increased by 4.52 bucks:100 does (SE = 1.40), and adult buck:doe ratios increased by 3.37 bucks:100 does (SE = 1.04) in response to limited harvest. Evidence suggested that factors other than buck harvest were regulating population productivity with density dependence being a plausible explanation of declining fawn:doe ratios. Limiting buck harvest to enhance fawn recruitment is not justified in Colorado based on our analysis. Management for limited buck harvest should be largely framed as an issue of quality hunting opportunity rather than an issue of deer productivity. Key words: age ratio, buck:doe ratio, Colorado, fawn:doe ratio, limited harvest, mule deer, Odocoileus hemionus, productivity, quality hunting, sex ratio Wildlife Society Bulletin 33 (0):00-00. LITERATURE CITED Freddy, D.J. 2002. Estimating calf and adult survival rates and pregnancy rates of Gunnison Basin elk. Colorado Division of Wildlife Wildlife Research Report July: 191-222. Fort Collins, Colorado, USA. Freddy, D.J. 2003. Estimating calf and adult survival rates and pregnancy rates of Gunnison Basin elk. Colorado Division of Wildlife Wildlife Research Report July: In Press. Fort Collins, Colorado, USA. Prepared by _______________________________ David J. Freddy, Wildlife Researcher 60 � https://cpw.cvlcollections.org/files/original/dbd8334e4e5c3aa4faeda8663eea5d7d.pdf f07997d4a8bbba4f9cb4900c471bf04c PDF Text Text 259 / '\ Colorado Division of Wildlife Wildlife Research Report July2000 JOB PROGRESS REPORT State of Colorado _ _ _ __,,C=o=lo=rad=o=-------- Cost Center 3430 Project _ _ _ _ _ _ _W ..........-=15__3-'-R=-~1~3_ _ __ Mammals Research Work Package _ _ _ _ _-=-30;:;..;0=2'------ Elk Management Study No. ------~RMNP~~---- Technical Support for Elk and Vegetation Management Environmental Impact Statement for Rocky Mountain National Park Period Covered: July 1, 1999 - June 30, 2000 Author: Dan L. Baker \ ) Personnel: M.A. Wild, T. M. Nett, D. Finley, M. Conner, J.C. Ritchie, M, Conner, L. Wheeler ABSTRACT We conducted experiments to evaluate the effectiveness of a GnRH agonist in preventing pregnancy in captive female elk. LUPRON (luprolide acetate), administered as a subcutaneous implant was 100 % effective in preventing pregnancy in female elk treated before the breeding season. Effective duration of . LUPRON was approximately 193 to 225 days. All treated elk regained fertility following contraceptive treatments. There were no significant differences in body condition, hematology, blood chemistry, general health, or breeding behavior of treat~ and untreated elk. 1il1Di1ii1~i1ri1rn BDOWD16790 ��261 TECHNICAL SUPPORT FOR ELK AND VEGETATION MANAGEMENT ENVIRONMENTAL IMPACT STATEMENT FOR ROCKY MOUNTAIN NATIONAL PARK Dan L. Baker P. N. OBJECTIVES 1. Conduct laboratory experiments needed to reliably implement fertility control alternatives for elk management in Rocky Mountain National Park (RMNP). 2. Develop simulation models to evaluate a range of population control alternatives including fertility control. 3. Prepare a discussion of alternatives for inclusion in the Environmental Impact Statement. SEGMENT OBJECTIVES 1. Prepare scientific publications describing results ofGnRH analog and/or GnRH-toxin conjugate experiments in captive elk. 2. Develop a simulation model to evaluate the feasibility of GnRH analogs and/or GnRH-toxin conjugates to regulate elk populations. 3. Develop strategies to integrate informed local, regional, and national stakeholder input and involvement in elk management alternatives in Rocky Mountain National Park. INTRODUCTION Overabundant wild ungulates can do serious and lasting hann to many plant communities, and preventing such damage may require controlling the growth of their populations (Jewell and Holt 1981, Diamond 1992, McCullough et al. 1997). In Rocky Mountain National Park (RMNP), Colorado, the impact of herbivory by elk has emerged as a fundamentally important issue for those who manage the Park and its wildlife (Hess 1993, Singer et al. 1998). In 1968, RMNP adopted a natural-regulation policy for management of ungulates (Cole 1971, Houston 1971). The objective was to allow density dependent processes to regulate the number of ungulates within park boundaries and to use sport hunting to harvest as many elk as possible in areas surrounding the Park. Recently, however, Park managers have become concerned that possible unnatural concentrations of elk may be altering natural plant communities and ecosystem sustainability. Soil conditions and the status of willow and aspen plant communities have declined. Wet meadow, dry grasssland, and alpine and subalpine sites show evidence of deterioration from overgrazing by elk (Singer et al. 1998, White et al. 1998). As a result of the decline in these vegetation types and the diversity of the animal species that are associated with them, the Park is evaluating alternative management strategies for reducing elk densities within RMNP and the surrounding Estes Valley. Acceptable alternatives for managing elk overabundance in RMNP and adjacent lands are limited. Public hunting within National Parks is proscribed by law and policy and is not permitted without Congressional authorization and an amendment to the enabling legislation for the specific park (Wagner et al. 1995). Authorizing legislation does permit professional culling and RMNP has a long history of animal control �262 actions to hold numbers of elk to a desired level. However, opposition from animal welfare organiu1.tions and other public criticism has prevented implementation of control programs for native ungulates in national parks since 1967 (Wright 1992). When lethal control is deemed necessary, the highest priority is given to encouraging public hunting outside Park boundaries. The success of traditional hunting-based elk management depends on access to private lands. In the Estes Valley, the increase in human occupation and recreational developments during the last decade has brought populations of elk in close proximity to high densities of people. Reduced hunter access, resulting from unethical hunter behavior (Wright et al. 1988), private-land recreation liability laws, attitudes toward hunting (Decker and Gaven 1987), and health and safety concerns has limited the use and effectiveness of sport hunting as an alternative for reducing elk densities in the areas surrounding RMNP. Live capture and translocation of elk is a common management technique that is considered by many to be a more humane alternative to hunting and culling. The few attempts at large scale removals have proven to be costly, inefficient, and stressful or lethal to most of the relocated animals (O'Bryan and McCullough 1985). Two additional problems compromise the potential effectiveness of this strategy for overabundant elk in RMNP. First, there are few suitable habitats in the western United States that are not without healthy and productive elk populations. Secondly, even if release sites were available, the prevalence of chronic wasting disease in the RMNP elk herd (Spraker et al. 1997, Miller et al. 1998) and the potential for spreading this disease to uninfected populations effectively eliminates this option from consideration. Elk Fertility Control Experiments The use of fertility control to decrease birth rates is one of the most promising approaches to the long-term control of overabundant wild ungulates. During the past decade, research aimed at developing effective contraceptives for free-ranging wildlife populations has accelerated. These efforts have resulted in development and testing of a wide variety of potential contraceptive agents (Kirkpatrick and Turner 1985, Warren et al. 1995). One of the most promising new non-steroidal, non-vaccine, approaches to contraception involves synthetic analogs of gonadotropin-releasing hormone (GnRH). GnRH is a molecule produced in the hypothalamus of the brain. It directs specific cells in the pituitary gland to synthesize and secrete two important reproductive hormones; follicle stimulating hormone (FSH) and luteinizing hormone (LH). These latter two hormones, known as gonadotrophs, control the proper functioning of the ovaries in the female and testes in the male. Analogs of GnRH have the potential to either permanently or temporarily inhibit reproduction. For most free-ranging wild ungulate applications, permanent sterilization or a combination of permanent sterilization and culling have been proposed as the most efficacious approaches to population management (Hone 1992, Garrot 1995, Hobbs et al. 1999, in press). For this application, superactive analogs ofGnRH are coupled to a cytotoxin. The GnRH-toxin conjugate specifically targets the gonadotroph cells and permanently inhibits the ability of the cell to secrete FSH and LH. This approach has several potential advantages over other methods of contraception. These include: I) a single treatment should permanently sterilize an animal 2) the same treatment should be effective in both males and females and in different mammalian species 3) GnRH-toxin conjugate will be metabolized from the body within a few days of treatment 4) the proteinaceous nature ofGnRH-toxin conjugate eliminates the possibility of passage through the food chain. 5) the small volume required for effective contraception would facilitate microencapsulation and administration by syringe dart or biodegradable projectiles. �263 In other situations where wildlife managers need to maintain flexibility in the use of fertility control, reversible contraception may be desirable. Examples of these situations include I) wild ungulate populations exposed to periodic, severe, unanticipated winter mortality, 2) populations with low genetic variability, 3) populations that cannot be effectively monitored, 4) populations where public attitudes are opposed to permanent contraception, or 5) populations where non-lethal hunting recreation is a primary management objective. In these situations, superactive analogs of GnRH without the toxin subunit would be more appropriate. The inhibitory actions of long-term GnRH analog agonist on the ovulatory cycle of humans and other mammals is well-established (Casper and Yen 1979, Fraser 1983, Fraser etal. 1987, Concannon et al. 1991). Constant administration of high doses ofGnRH agonist results in down regulation of the pituitary GnRH receptors and suppression of secretion of LH and FSH. Continued treatment suppresses LH • secretion, preventing the maintenance of normal luteal function, and thus prevents viable pregnancy. Inhibition of ovulation caused by chronic administration of GnRH agonist has been successful in several species, including dogs (Vickery et al. 1989), cattle (Herschler and Vickery 1981), sheep (McNeilly and Fraser 1987), white-tailed deer (Becker and Katz 1995), and elk (Baker and Nett, unpublished data). Evidence from studies on pituitary receptors and gonadotropin content in experimental animals treated by long-term infusion of GnRH agonist shows that sustained release is the most effective approach for temporarily suppressing pituitary-gonadal function (Clayton 1982, Sandow 1982). The practicality of this approach, however, is dependent upon development of a long-acting, slow-release preparation of agonist that can be remotely delivered. Recently, a practical mode of administralion using subcutaneous implants has overcome the need for constant mechanical infusion of the analog. Slow release formulations of superactive GnRH agonist are now commercially available and have been shown to be effective in suppressing the pituitary ovarian axis for up to 6 months in a variety of mammalian species (Fraser et al. 1987, Asch et al. 1985). Proposed Research To our knowledge, only limited investigations have been conducted with either of these fertility control techniques on wild ungulates (Becker and Katz 1995), however, the minimum dose ofGnRH analog required for maximum pituitary stimulation is known for elk (Baker et al. 1995) and the minimum effective duration of controlled release GnRH agonist implants has been determined (Baker and Nett 1999, unpublished data). Additional research is needed to determine the effectiveness of these contraceptive agents in preventing pregnancy in elk and the maximum effective duration. Research is also needed to identify any nutritional, physiological or behavioral side-effects that may result from treatment. Thus, the objectives of our research are: 1) To evaluate the effectiveness and effective duration of GnRH-toxin conjugate and GnRH agonist in preventing pregnancy in elk. 2) To evaluate the effects ofGnRH-toxin conjugate and GnRH agonist on nutrition, physiology, general health and social behavior of captive elk. We will test the null hypothesis of no effect of GnRH-P AP and GnRH-Agonist on LH levels, pregnancy rates, social behavior, and general health of female elk. �264 MATERIALS AND METHODS Rocky Mountain elk exhibit highly seasonal patterns of reproduction that are controlled by photoperiod regimens. The onset of the breeding season occurs during the decreasing daily photoperiods of autumn and is preceded by a period of deep anestrous/anovulation in summer (Jopson et al. 1990). The first ovulation of the breeding season is usually preceded by one or more silent ovulations associated with the formation of short-lived corpora lutea that serve to synchronize the first overt estrus within a herd. In temperate North America, the majority of conceptions occur in late September, but recurrent estrous cycles of 21 days are possible through February if females fail to conceive. In early spring, coincidental with increasing day length, reproductive cycles cease and females remain anestrous until August. For pregnant females, parturition generally occurs in early June, after a gestation period of about 255 days (Kelly et al. 1982, Adam et al. 1992). Average conception date for elk at the Foothills Wildlife Research Facility has been estimated to be Sept 30 ± 6 days (Sept 24 - Oct 6). We conducted controlled experiments with 16 adult female and 3 adult male elk at the Colorado Division of Wildlife's Foothills Wildlife Research Facility (FWRF), Fort Collins, Colorado during August 1999 to April 2000. Treatments We compared LH responses, pregnancy rates, social behavior, and general health of the following 4 groups of female elk: Group 1: GnRH-PAP - Pre-conception' - Group 1 elk were treated with an irreversible contraceptive (GnRH-PAP) prior to conception. This was accomplished by treating all cows in this group with 3µ /50 kg BW of GnRH-P AP, IM, one week before being exposed to three adult male elk. Group 2: GnRH-PAP - Pregnant - Group 2 elk were treated with 2µ I 50 kg BW ofGnRH-PAP, IM, during the second trimester of pregnancy (approx. January 15). We confirmed pregnancy of the cows in this group using serum pregnancy-specific protein-B (PSPB) (Williard et al. 1994) prior to treatment. Group 3: GnRH-Agonist (Lupron) - Group 3 elk were treated with a reversible contraceptive (GnRHAgonist - Lupron). We administered a subcutaneous implant containing 32.5 mg ofLupron (Baker and Nett 1999, unpublished data) to female elk one week prior to being exposed to adult male elk. All treatments were applied without tranquilization by moving elk from 5 ha pastures to individual isolation pens, then into a restraining chute where treatments were applied, then relecll>ing animals back into 5 ha pastures. Group 4: Control - Group 4 elk are female elk were untreated and nonpregnant for the duration of the experiment. Four elk were assigned to each treatment group. Based on previous studies with captive elk (Baker et al. 1995, Baker and Nett 1999, unpublished data), approximately four elk per treatment is the minimum sample size needed to provide biologically significant differences among treatment means. Measurements Reprodu_ctive Status. Before assigning animals to treatment groups, we determined the reproductive status of all female elk by monitoring serum progesterone levels during late August and early September. On the day of blood sampling, elk were moved from 5 ha pastures to individual isolation pens, sedated with �265 xylazine hydrochloride (50-200 mg/animal IM), and blood samples collected (5ml). Animals were reversed with yohimbine (0.125 mg/kg, IV) and returned to the paddocks. Sedation of elk was done in order to minimize potential adrenal secretion of exogenous progesterone during handling and blood sample collections (Jopson et al. 1990). Analysis: Serum progesterone concentrations were determined using RIA procedures (Niswender 1973). Sensitivity of the progesterone assay is 0.12 ng/ml. Female elk with progesterone levels above I ng/ml were considered reproductively active (Jopson et al. 1990). Elk with progesterone levels below Ing/ml will be sampled 7 days later. If after three consecutive blood collections, progesterone levels remain lower than I ng/ml the elk were removed from the experiment. Hormonal Assessments Prior to application of contraceptive treatments, we measured the LH response of each elk in treatment Groups I, 3, and 4 to GnRH analog (Baker et al. 1995). Results from this trial provided a pretreatment baseline for comparison to future posttreatment LH responses. This and succeeding LH challenge trials were conducted as follows: On Day 1 of the trial, elk were moved from 5 ha pastures to individual isolation pens, sedated with xylazine hydrochloride (50-200 mg/animal, IM), and fitted nonsurgically with indwelling jugular catheters. Animals were reversed with yohimbine (0.125 mg/kg, IV). On Day 2, we will administer GnRH analog (lµ /50 kg BW) through the cannula and collect blood samples (5 ml) at 0, 60, 120, 180,240, 300, 360, and 480 minutes postinjection. Following the last blood collection, catheters were removed and each animal given Naxel (300 mg, I.V.). Animals were then r¢irned to 5 ha pastures. After collection, blood was held at 4 °C for 24 h until serum is obtained by centrifugation. Serum was be stored at - 20 °C until analyzed for LH. Sen.mi concentrations of LH were quantified by means of ovine LH RIA (Niswender et al. 1969). The duration of contraceptive effectiveness was assessed by conducting GnRH challenge trials each month from September 1999 to May 2000. Analysis. Responsiveness of the pituitary to GnRH challenge was assessed in three ways: I) maximum LH (ng/ml) response achieved postinjection minus baseline, 2) time required to reach maximum LH, and 3) total amount ofLH secreted (ng/ml/min). Pregnancy Rates We assessed contraceptive effectiveness by determining the pregnancy status of all experimental elk. A single blood sample (10 ml) was taken via jugular venipuncture from each animal for PSPB analysis approximately 90 days post-conception (Willard et al. 1994). Animal handling and blood collections for PSPB followed methods previously described for hormonal assessment and were collected in conjunction with these measurements. Female neonates born to any experimental cow will be incorporated into the FWRF elk herd. Male neonates born after August 15 were euthanized according to ACUC procedures. Breeding Behavior The effects of the contraceptive treatments on social behavior in elk is not known. However, if gonadotroph cells are destroyed by GnRH-PAP or desensitized by long-acting synthetic GnRH agonist, down-regulation and diminished LH and FSH can be induced. Reduced secretion of LH and FSH could decrease or disrupt ovarian function, estrus cycles and secondary sex characteristics. We provided insight into these potential effects by monitoring maintenance and sexual behavior of male and female elk during the breeding season. Each animal in each treatment and the breeding males were individually identified using color-coded neck bands or ear tags. We tested the null hypothesis that the frequency of sexual interactions between treated females and males would be similar to that of untreated females and males. �266 We tested this hypothesis using focal- animal sampling procedures and discriminant function analysis (Lehner 1987) {Appendix A). General Health Our knowledge of the effects of these contraceptive treatments on general health, nutrition. body weight dynamics, and blood chemistry of elk is limited. Previous experiments with domestic livestock and companion animals (Nett 1999, unpublished data) and captive mule deer and elk (Baker 1999, unpublished data) have shown no measurable short-term effects. We evaluated these potential side-effects by monitoring body weight, blood chemistry, and hematology of all experimental elk. We collected blood (10 ml) for blood chemistry and hematology prior to treatment with contraceptives, at I month posttreatment and at 6 month intervals thereafter. Statistical Analysis We analyzed data using least squares ANOVA for General Linear Models and the SAS Interactive Matrix Language. Response to contraceptive treatments was analyzed with a two-way factorial analysis of variance for a randomized complete block design with repeated measures structure. Levels of GnRH analog were treatments; individual animals were blocks. Factors in the analysis were treatment and time. Treatment was tested using the animal-within-treatment variance as the error term. Time was treated as a within subject effect using a multivariate approach to repeated measures. We used orthogonal contrast to test for differences among individual means (Morrison 1976, Miller 1966). RESULTS AND DISCUSSION Elk Fertility Control Experiments Pregnancy Rates and Hormonal Responses LUPRON (luprolide acetate), administered as a subcutaneous implant, was 100 % effective in preventing pregnancy in female elk treated prior to the breeding season. The antifertility effects ofLUPRON, a GnRH agonist, were associated with luteolysis and accompanied by a marked loss in ovarian luteinizing hormone and serum progesterone levels. Mean serum progesterone levels were reduced in all female elk from an average pretreatment concentration of 1.4 ng/ml (SE= 0.6) to non-detectable levels at 92 days posttreatment. Serum progesterone levels remained at these levels until approximately 193 days posttreatment (Fig. 1). We observed a similar response in serum LH values. Mean serum concentrations of LH began to decline approximately 30 days following treatment and remained at nondetectable levels for 193 days posttreatment (Fig. 2). LUPRON significantly (P ~0.002) reduced serum LH concentrations in treated elk compared to non-pregnant controls from 30 to 193 days posttreatment. Based on these measurements, anitfertility effects of LUPRON in female elk are estimated to be approximately 193 to 225 days posttreatment. We did not observe negative side-effects of LUPRON on nutrition. physiology, or general health of treated females. Breeding Behavior We collected data for 63 sampling periods: 20 morning periods totaling 45.7 hrs, 6 mid-day periods totaling 13 .5 hrs, 20 evening periods totaling 42.8 hrs, and 17 night periods totaling 32.6 hrs, for an overall total of 134.5 hrs of observation. Time of day, date, and the interactions were not significant effects and were dropped from the ANOVA model. Except for general breeding behaviors, the behavior rate of cows treated with Lupron IM was higher than for control cows or for cows treated with Lupron SQ (Table 1). �267 If Lupron had no treatment effect, then the difference between behavior rates of control and treatment groups will equal z.ero. Differences in behavior rates for cows treated with Lupron SQ and control cows were equal to z.ero for all behavior categories (fable 2). The difference in behavior rates for cows treated with Lupron IM and control cows was not equal to z.ero for male pre-copulatory behaviors and was borderline significant at a 0.10 alpha-level for female pre-copulatory behaviors (fable 2). Table 1. Mean rate of behavior and standard error for control and treatment elk groups; mean and standard error were estimated using type III least-squares. Behavior category Treatment Group Copulatory Control Lupron SQ Lupron IM Male Pre-Copulatory Control Lupron SQ Lupron IM Female Pre-Copulatory Control Lupron SQ Lupron IM General Breeding Control Lupron SQ Lupron IM ~ Mean SE (# behviors/hr') 0.022 0.027 0.045 0.014 0.017 0.020 0.263 0.298 0.453 0.058 0.026 0.080 0.047 0.040 0.271 0.033 0.017 0.139 0.324 0.062 0.330 0.279 0.050 0.045 Table 2. Difference in elk behavior rates (# behaviors/hr) between control and treatment groups, standard error of the difference, and P-value of the difference (the probability that the behavior rates were equal); statistics were estimated using type ID least-squares. Behavior category Treatment Group Copulatory Control - Lupron SQ Control - Lupron IM Lupron SQ - Lupron IM Male Pre-Copulatory Control - Lupron SQ Control- Lupron IM Lupron SQ - Lupron IM Female Pre-Copulatory Control - Lupron SQ Control - Lupron IM Lupron SQ - Lupron IM General Breeding Control - Lupron SQ Control- Lupron IM Lupron SQ - Lupron IM Mean Difference SE P-Value -0.005 -0.023 -0.018 0.023 0.024 0.027 0.829 0.346 0.503 -0.035 -0.190 -0.155 0.072 0.099 0.086 0.624 0.054 0.073 0.007 -0.224 -0.231 0.037 0.151 0.148 0.849 0.138 0.119 -0.006 0.045 0.051 0.084 0.080 0.069 0.943 0.574 0.463 �268 -e-- Lupron SC - 0- - Non-Pregnant Control 30 25 20 I \ \ \ \ ::c ....J ~ eel \ \ 15 \ Q) // \ C / \..,..,,,/ ~ eel ~ / \ a.. Q) / \ 10 5 0 Pretrmt 30 92 135 165 193 225 Days Posttreatment Figure 1. Mean serum luteinizing hormone (LH) concentrations of female elk before and after treatment with a subcutaneous implant containing 32.5 mg LUPRON. For all behavior categories, behavior rates were almost identical for the control and Lupron SQ groups. If cows cease cycling once bred, then this similarity suggests that the Lupron SQ group did not cycle or cycled once. In contrast, for all behavior categories except general breeding, behavior rates for the Lupron IM group were 1.5-6.8 times higher than for the control and Lupron SQ groups. The Lupron IM cows appeared to cycle into estrous more than one time, which resulted in higher reproductive behavior rates. There was no statistically significant difference in copulatory behavior rates between control and treatment elk. However the rate of copulatory behavior was about 2 times higher for the Lupron IM group than for the control and Lupron SQ groups. Biologically, this suggests that the Lupron IM group may have cycled and bred more than one time during the experiment. The rate of male pre-copulatory behavior was significantly higher toward the Lupron IM group than toward the control and Lupron SQ treatment groups. Male pre-copulatory behavior rates were 1.5-1. 7 times higher toward the Lupron IM group. Although the rate of female pre-copulatory behavior was 5.8-6.8 times higher for the Lupron IM group compared to the control and Lupron SQ groups, this difference was not significant. The lack of significance probably stems from temporal variations in female pre-copulatory behavior. When not in �269 --- - -0-- - Lupron SC Control 3.00 2.50 ::::::- --E 2.00 0) C -.:;f- a.. 1.50 C co (l) ~ 1.00 l G----.1- 0.50 0.00 Pretrmt 30 92 135 165 193 225 Days Posttreatment Figure 2. Mean serum progesterone (P41 concentrations of female elk before and after treatment with a subcutaneous implant containing 32.5 mg ofLUPRON. estrous, female cows showed no interest in the bulls, resulting in many sampling periods where the rate of behavior was zero. Then, when a cow came into estrous, or cycled in some way, she would spend much of her time soliciting interest from the bull, which resulted in a high rate of female pre-copulatory behavior for that sampling period. Thus, female pre-copulatory behavior rates were spiked through time, resulting in high variation and lack of statistical significance between treatment groups. Although not statistically significant, biologically, the Lupron IM group appeared to continue some sort of cycle associated with estrous. Rates of general breeding behavior were similar for all 3 treatment groups. The general breeding behaviors, approach and herding, are harem gathering and maintenance behaviors. In the wild, elk gather cows, regardless ofwhether they are in estrous at that time. Th.is may explain why the bull did not distinguish between treatment groups; he simply grouped all the cows and kept them from the other 2 bulls in the pen. Elk Fertility Control Mode/ Development of a simulation model representing dynamics of the RMNP elk herd and evaluation of the benefits and liabilities of a range of population control alternatives was not accomplished during this segment and is contingent on data not presently available from RMNP. Th.is objective will be completed during FY 2000-1. �270 Stakeholder Input on Elk Management Alternatives lb.is segment objective was not addressed during FY 1999-0 due to a delay in the initiation of the formal Environmental hnpact Statement process. LITERATORE CITED Adam, C. L., C. E. Moir, and T. Atkinson. 1985. Plasma concentrations of progesterone in female red deer (Cervus elaphus) during the breeding season, pregnancy, and anoestrus. J. Reprod. Fertil. 74:631-636. Asher, G. W. 1985. Oestrous cycle and breeding season of farmed fallow deer, Dama dama. J. Reprod. Fertil. 75:521-529. Asch, R.H., F. J. Rojas, T. R. Tice, and A. V. Schally. 1985. Studies of a controlled - release microcapsule formulation of an LH-RH agonist in the rhesus monkey menstrual cycle. Int. J. Fert. 30:19-26. Baker, D. L., M. W. Miller, and T. M. Nett. 1995. Gonadotropin-releasing hormone analog-induced patterns ofluteinizing hormone secretion in female wapiti (Cervus elaphus nelsoni) during the breeding season, anestrus, and pregnancy. Biol. of Reprod. 52: 1193-1197. Becker, S. E., Katz, L. S. 1995. Effects of gonadotropin - releasing hormone agonist on serum LH concentrations in female white-tailed deer. Small Ruminant Research 18:145-150. Casper, R. F., and S.S. C. Yen. 1979. Induction ofluteolysis in the human with a long-acting analog of luteinizing hormone-releasing factor. Science 205:408-410. Clayton, R. N. 1982. GnRH modulatidn of its own pituitary receptors: evidence ofbiphasic regulation. Endocrinology 111: 152-161. Cole, G. F. 1971. An ecological rationale for the natural or artificial regulation of native ungulates in parks. Transactions of the North American Wildlife and Natural Resources Conference 36:417426. Concannon, P. W., and V. N. Meyers-Wallen. 1991. Current and proposed methods for contraception and termination of pregnancy in dogs and cats. J. Am. Vet. Med. Assoc. 198:1214-1225. Curlewis, J. D., A. S. I. Loudon, and A. P. M. Coleman. 1988. Oestrous cycles and the breeding season of the Pere David's deer hind (Elaphurus davidianus). J. Reprod. Fert. 82: 119-126. Decker, D. J., and Gavin. 1987. Public attitudes toward a suburban deer herd. Wildlife Society Bulletin 16:53-75. Diamond, J. 1992. Must we shoot deer to save nature? Natural History August: 2-8. Fraser, H. M. 1983. Effect of treatment for one year with a luteinizing hormone-releasing hormone agonist on ovarian, thyroidal, and adrenal function and menstruation in the stumptailed monkey (Macaca arctoides). Endocrinology 112:245-253. - - - ~ M., J. Sandow, H. Seidel, and W. von Rechenberg. 1987. An implant ofa gonadotropin releasing hormone agonist (burserelin) which suppresses ovarian :function in the macaque for 3-5 months. Acta Endocrinologica 115: 521-427. Garrot, R. A. 1995. Effective management of free-ranging ungulate populations using contraception. Wildlife Society Bulletin. 23:445-452. Guinness, F. E., Lincoln, G. A., and Short, R. V. 1971. The reproductive cycle of the female red deer, Cervus elaphus L. J. Reprod. Fertil. 27: 427-438. Hess, K. 1993. Rocky times in Rocky Mountain National Park. University_ Press of Colorado, Ni wot, Colorado. Herschler, R. C., and B. H. Vickery. 1981. The effects ofLHRH ethylamide on the estrous cycle, weight gain, and feed efficiency in feedlot heifer. Arner. J. of Vet. Res. 42:1405-1408. �271 Hobbs, N. T., D. L. Baker, and RB. Gill. 1999. A general theory describing effects of fertility control on populations of ungulates. Journal of Wildlife Management (in press). Hone, J. 1992. Rate of increase and fertility control. J. Applied Ecology 29:695-698. Houston, D. B. 1971. Ecosystems of national parks. Science 172:648-65 l. _ _ _. 1982. The Northern Yellowstone Elk: Ecology and Management. MacMillan Publishing Company, New York, New York. Jewell, P. A., and S. Holt. 1981. Problems in management of locally abundant wild animals. Academic Press. 32lpp. Jopson, N. B., M. W. Fisher, and J.M. Suttie. 1990. Plasma progesterone concentrations in cycling and ovariectomiz.ed red deer hinds: the effect of progesterone supplementation and adrenal stimulation. Animal Reprod. Sci. 23:61-73. Kelly, R W., K. P. McNatty, G. H. Moore, D. Ross, and M. Gibb. 1982 . .Plasma concentrations ofLH, prolactin, oestradiol and progesterone in female red deer during pregnancy. Journal of Reproduction and Fertility 64:475-483. Kirkpatrick, J. F., and J. W. Turner, Jr. 1985. Chemical fertility control and wildlife management. Bioscience 35:485-491. Lehner, P. N. 1987. Design and execution of animal behavior research: an overview. Journal of Animal Science 65: 1213-1219. McCullough, D. R, K. W. Jennings, N. B. Gates, B. G. Elliot, and J.E. Didonato. 1997. Overabundant deer populations in California. Wildlife Society Bulletin 25:478-483. McNeilly, A. S., and H. M. Fraser. 1987. Effect of gonadotropin-releasing hormone agonist-induced suppression of LH and FSH on follicle growth and corpus luteum function in the ewe. J. Endocrinology 115:273-282. Miller, M. W., M.A. Wild, and E. S. Williams. 1998. Epidemiology of chronic wasting disease in captive rocky mountain elk. Journal of Wildlife Diseases 34:532-538. Miller, R.G. 1966. Simultaneous statistical inference. New York: McGraw-Hill Book Co. 152-168. Morrison, J. A. 1960. Characteristics of estrus in captive elk. Behavior 16:84-92. Morrison, D. F. 1976. Multivariate statistical methods. New York: McGraw-Hill Book Co. 145- 194. Niswender, G.D., L. E. Reichert, Jr., A. R Midgley, Jr., and A. V. Nalbandov. 1969. Radioimmunoassay for bovine and ovine luteinizing hormone. Endocrinology 84: 1166-1173. ____. 1973. Influence of the site of conjugation on the specificity of antibodies to progesterone. Steroids 22:413-424. _ O'Bryan, M. K., and D.R. McCullough. 1985. Survival of black-tailed deer following relocation in California. Journal of Wildlife Management 49: 115-119. Sandow, J. 1982. Inhibition of pituitary and testicular function by LHRH analogs. Pages 19-39 In: Jeffcoate S. L. and Sandlier (eds). Progress towards a male contraceptive. Wiley and Sons, Chichester. Singer, F. J., L. C. Zeigenfuss, R G. Cates, and D. T. Barnett. 1998. Elk, multiple factors, and persistence of willows in national parks. Wildlife Society Bulletin 26:419-428. Spraker, T. R., M. W. Miller, E. S. Williams, D. M. Get.zy, W. J. Adrian, G. G. Schoonveld, R A. Spowart, K. I. O'Rourke, J.M. Miller, and P.A. Merz. 1997. Spongiform encephalopathy in free-ranging mule deer (Odocoileus hemionus), white-tailed deer (Odocoileus virginianus), and Rocky Mountain elk (Cervus elaphus nelsoni) in northcentral Colorado. Journal of Wildlife Diseases 33:1- 6. Vale, W., C. River, M. Brown, J. Leppaluoto, N. Ling, M. Monaham, and J. River. 1976. Pharmacology of hypothalamic peptides. Clin. Endocrin. 5(suppl): 261-273. Wagner, F. H., R. Foresta, R. B. Gill, D.R. McMullough, M. P. Pelton, W. F. Porter, and H. Salwasser. 1995. Wildlife policies in the U. S. National Parks. Island Press. Washington, D. C. 242 pp. �272 Warren, R. J., L. M. White, W.R. Lance. 1995. Management of urban deer populations with contraceptives: practicality and agency concerns. Wildlife Society Bulletin 23: 441-444. White, C. A, C. E. Olmsted, and C. E. Kay. 1998. Aspen, elk, and fire in the Rocky Mountain national parks of North America. Wildlife Society Bulletin 26:449-462. Willard, S. T., R G. Sasser, J. C. Gillespie, J. T. Jaques, T. H. Welsh, Jr., and R. D. Randel. 1994. Methods for pregnancy determination and the effects of body condition on pregnancy status in Rocky Mountain elk. Theriogenology 42:1095-1102. Wright, R G. 1988. Wildlife issues in national parks. Pp 168-196 in W. J. Chandler (ed) Audubon Wildlife Report, National Audubon Society. _ _ _ _. 1992. Wildlife management and research in the national parks. University of Illinois Press. 224 pp. Prepared by _ _ _ _ _ __ Dan L. Baker Research Biologist �169 Colorado Division of Wildlife Wildlife Research Report July 2001 and July 2002 JOB PROGRESS REPORT State of --------"="-==-----Colorado Division of Wildlife - Mammals Research Work Package No. _____3__0___0=-2_ _ _ _ __ Elk Management Study No. ----~RMN='"'"p~----- Technical Support for Elk and Vegetation Management for Rocky Mountain National Park - Environmental Impact Statement Period Covered: July 1, 1999 -June 30, 2002 Author: Dan L. Baker, Ph.D. Personnel: M. Wild, T. Nett, D. Finley, M. Conner, J. Ritchie, L. Wheeler, E. Jones, D. Hussain ABSTRACT Fertility control offers a potential alternative to traditional methods for regulating the growth of overabundant wild ungulate populations. However, current technology is limited due to practical treatment application, undesirable side-effects, and economic considerations. A promising non-steroidal, non-immunological approach to contraception involves potent GnRH agonist. During 1999-2002, we conducted a series of experiments to evaluate the effectiveness of a GnRH agonist (leuprolide) as a contraceptive agent in captive female elk. In experiment 1, we determined the optimum dose of GnRH agonist treatment by measuring serum luteinizing hormone (LH) and progesterone (P4 ) response of female elk to 4 formulations of leuprolide administered as subcutaneous bioimplants. In experiment 2, we evaluated the effects ofleuprolide on elk pregnancy rates, duration of suppression ofLH and P4 secretion, and short-term behavioral and physiological side-effects. In experiment 3, we evaluated the effects ofleuprolide on pregnant elk, and in experiment 4, assessed the potential for delivering leuprolide remotely in a syringe dart. All concentrations ofleuprolide were equally effective in reducing serum LH and P4 to non-detectable levels for the duration of the 130 day experiment. Leuprolide administered prior to the breeding season was 100% effective in preventing pregnancy in treated females. Serum LH and P4 concentrations were reduced to baseline levels by day 92 and remained at these levels for 195-251 days posttreatment with a return to pretreatment concentrations the following breeding season. Reproductive behavior rates were similar for treated and untreated elk for all behavior categories for both the breeding and postbreeding seasons. Hematology and blood chemistry parameters of treated and untreated females were similar and seasonal intake and body weight dynamics appeared normal. Initial results indicate that leuprolide can be effectively delivered in a syringe dart but additional research is needed to confirm these observations. Thus, we conclude that leuprolide is a safe, effective contraceptive agent and has the potential for suppressing fertility in female wapiti for one breeding season. ��171 TECHNICAL SUPPORT FOR ELK AND VEGETATION MANAGEMENT FOR ROCKY MOUNTAIN NATIONAL PART ENVIRONMENTAL IMPACT STATEMENT Dan L. Baker P. N. OBJECTIVE Conduct captive elk experiments to implement fertility control as an alternative for managing elk in Rocky Mountain National Park. SEGMENT OBJECTIVES 1. Develop and test a reversible contraceptive agent for free-ranging elk. 2. Determine the duration of effectiveness of a selected contraceptive agent in captive elk. 3. Assess contraceptive effects on pregnancy, and behavioral and physiological side-effects in captive elk. 4. Develop and test a remote delivery system for administering the contraceptive agent to free-ranging elk. INTRODUCTION Overabundant wild ungulate populations have become a significant problem for natural resource managers in North America. Unregulated populations can cause adverse effects that are ecological, economic, or political in scope and resolving these issues often requires controlling animal abundance (Jewell and Holt 1981, Garrott et al. 1993, McCullough et al.1997, Smith 2001). In Rocky Mountain National Park (RMNP), Colorado, the impact ofherbivory by elk has emerged as a fundamentally important problem for those who manage the Park and its wildlife (Hess 1993, Zeignefuss et al. 1996). In 1968, RMNP adopted a natural-regulation policy for management of ungulates (Cole 1971, Houston 1971) with the objective of allowing density dependent processes to regulate elk numbers within park boundaries and use sport hunting to harvest as many animals as possible in areas surrounding the Park. Recently, however, Park managers have become concerned that possible unnatural concentrations of elk may be altering natural plant communities and ecosystem sustainability. Soil conditions and the status of willow and aspen plant communities have declined. Wet meadow, dry grasssland, and alpine and subalpine sites show evidence of deterioration from overgrazing by elk (Singer et al. 1998, White et al. 1998). As a result of the decline in these vegetation types and the diversity of the animal species that are associated with them, the Park and other natural resource agencies are evaluating alternative management strategies for reducing elk densities within RMNP and the surrounding Estes Valley. One alternative being considered is controlling the fertility of female elk. Fertility control has been widely advocated as an alternative to lethal methods of population control for wildlife and considerable research has been directed toward development of different contraceptive agents (see reviews by Kirkpatrick and Turner 1985, Fagerstone et al. 2001). Field and laboratory studies have evaluated the efficacy of delivery of contraceptives to ungulates (Jacobsen et al. 1995, DeNicola et al. 1997, �172 Kirkpatrick et al. 1997) and models have been developed to represent effects of fertility control on the population dynamics of individual species and populations (Garrott and Siniff 1992, Seagle and Close 1996, Hobbs et al. 2000). To date, most contraceptive research for wild ungulates has focused on the development of immunocontraceptive vaccines and steroidal hormonal agents. However, after more than 40 years of research, the success of these approaches have been primarily limited to captive wildlife and small localized urban populations of wild ungulates. To meet this challenge, new technologies and approaches are needed if fertility control is to become practical and acceptable management tool for controlling overabundant wildlife species. A promising new non-steroidal, non-immunological approach to contraception involves potent analogs of gonadotropin-releasing hormone (GnRH). GnRH is a molecule produced in the hypothalamus of the brain. It directs specific cells in the pituitary gland to synthesize and secrete two important reproductive hormones; follicle stimulating hormone (FSH) and luteinizing hormone (LH). These latter two hormones, known as gonadotrophs, control the proper functioning of the ovaries in the female and testes in the male. Chronic treatment with continuous, high doses of GnRH agonists results in temporary suppression of pituitary responsiveness and gonadotropin secretion. Resulting decreases in plasma LH and FSH in females leads to suppression of ovulation, estrus cyclicity, and gonadal steroidogenesis (Belchetz et al.1978, Evans and Rawlings 1994). Once GnRH agonist treatments are terminated, normal pituitary function is gradually restored (Bergfeld et al. 1996). GnRH agonists have been shown to inhibit ovulation in several domestic ungulate species including sheep (McNeilly and Fraser 1987), cattle ( D'Occhio et al. 1996, D'Occhio and Aspden 1999), and horses (Montovan et al. 1990). However, studies on wild ungulates are limited (Becker and Katz, 1995; Brown et al. 1999) and none have demonstrated their effectiveness as a contraceptive agent. GnRH agonists provide a potential biotechnology for achieving a controlled, reversible suppression of fertility in both captive and free-ranging female wild ungulates. However, their practicality as a contraceptive agent is dependent on effective inhibition of reproduction without negative behavioral or physiological side-effects. During 1999-2002, we conducted a series of experiments with sustained release formulations of GnRH agonist in captive female elk to evaluate these factors. Specifically, our objectives were: (1) to evaluate the effectiveness of GnRH agonist in preventing pregnancy, (2) to determine the duration of GnRH agonist suppression ofLH and progesterone (P 4) secretion, (3) to assess the behavioral and physiological side-effects (if any) of GnRH agonist treatments, and (4) __to develop a remote ~elivery system for administering the contraceptive agent to free-ranging animals. MATERIALS AND METHODS A. Experiment 1: Dose response 1. Objective Determine the minimum effective dose of GnRH agonist (leuprolide) that will induce halfmaximal release of luteinizing hormone in female elk during estrus and evaluate duration of effectiveness. �173 2. Methods We determined the optimum dose ofleuprolide (desGly1°-D-Leu6 -LH-RH ethylamide acetate) required for suppression of serum LH secretion in 8 female elk (6-12 years of age; 240-300 kg). Females were monitored for occurrence of oestrus cycles by measuring serum progesterone concentrations at weekly intervals beginning l November 1998 and were considered reproductively active when concentrations were greater than 1 ng m1· 1 for two consecutive sampling periods (Adam et al. 1985). Females were randomly selected to receive one of four doses (0, 45, 90, 180 mg leuprolide acetate) of 90 day sustained release leuprolide formulation using the ATRIGEL®drug delivery system (Atrix Laboratories, Inc. Ft. Collins, CO, USA) (Dunn et al.1994). These formulations at lower doses have demonstrated a sustained release and activity in rats and dogs for a period of at least 90-120 days (Ravivarapu et al. 2000). On the day before treatment application, animals were moved from paddocks, weighed(± 0.5 kg), moved to individual isolation pens (5 xl0 m), sedated with xylazine hydrochloride (Rompun; Bayer AG, Leverkusen, Germany; 25-200 mg animat1 i.m.) and fitted nonsurgically with indwelling jugular catheters. Sedation was reversed with yohimbine (30 mg) (Antagonil®, Wildlife Laboratories, Fort Collins, CO, USA). The sampling period began the next day (20 November 1998) at 0900. A patch of hair (3 cm in diameter) was shaved in the shoulder region of each female (controls did not receive a placebo formulation) and leuprolide formulations were injected under the skin using an 18 gauge needle and 3 cc syringe. Blood samples (5 ml) were collected at 0, 60, 120, 180, 240, 300,360,480 min, then at 12, 24, 36, 48, 84, and 240 h postinjection. Catheters were flushed daily with sterile saline solution. Following the last blood collection, catheters were removed and animals returned to 5 ha paddocks. We compared the effective duration ofleuprolide treatments by measuring pituitary responsiveness to an exogenous dose ofGnRH analogue (D-Ala6-GnRH-Pro9-ethylamide; Sigma Chemical Co., St. Louis, MO) administered at 35, 70, 110, and 130 days posttreatment. Animal handling and blood sampling protocols were similar to those previously described. We administered a previously determined dose (Baker et al. 1995) of GnRH analog (1 µg 50 kgBW· 1) through the jugular cannula and collected blood samples at 0, 30, 60, 90, 120, 180,240, 300, 360, 420, and 480 min postinjection. After collection, blood was held at 4 ° C for 24 h until serum was obtained by centrifugation. Serum was then stored at - 20° C until analyzed for LH. B. Experiment 2: Antifertility and behavioral effects on nonpregnant female elk 1. Objectives a. Determine the effectiveness of GnRH agonist in preventing pregnancy in female elk b. Determine the duration of GnRH agonist suppression of LH and P4 secretion c. Evaluate the behavioral and physiological side-effects (if any) of GnRH agonist treatments. 1. Methods We evaluated the effects of the optimum dose of leuprolide formulation established in Experiment 1, on elk pregnancy rates, duration of suppression ofLH and P4 secretion, blood chemistry, and reproductive behavior during 2 November 1999 to 15 May 2000. Fourteen adult female (7-13 years of age; 240-320 kg) and 3 adult male elk (4-13 years of age; 375- 400 kg) �174 were used in this experiment. Females were assigned to one of 3 experimental groups based on their tractability for handling and blood sampling. Four elk cows (Group A) were treated with 32.5 mg of leuprolide and 5 cows (Group B) served as untreated controls and were used to compare pregnancy rates, blood chemistry, and reproductive behavior to those of treated females. These two groups of females were maintained together with 3 adult male elk in adjoining paddocks (2-ha each). The remaining 4 females (Group C) served as untreated, non-pregnant controls and were placed in a separate pasture (1 ha) without direct contact with male elk. We compared LH and progesterone secretion of these females to those treated with leuprolide (Group A). a. Pregnancy rates, hormonal measurements, and blood parameters. We determined the effects ofleuprolide on pregnancy rates of treated and untreated elk by measuring pregnancy-specific protein B (PSPB)(BioTracking, Moscow, Idaho, USA) in serum at about 70, 160, and 215 days of gestation (Huang et al. 2000). We compared the effects of leuprolide on e)l..ient and duration of LH and P 4 suppression in treated and untreated, nonpregnant elk during 2 November 1999 to 11 November 2000. GnRH challenge trials were conducted prior to application ofleuprolide treatments and at 30, 90, 145, 180,225,250, and 373 days posttreatment. The final GnRH challenge trial was conducted to assess reversibility of treatment. Protocol for GnRH challenge trials followed procedures previously described in Experiment 1. We assessed physiological side-effects ofleuprolide by comparing serum chemistry, hematology, and body weight dynamics of treated (Group A) and untreated, non-pregnant elk (Group C). Blood collections and body weight measurement were made in conjunction with GnRH challenge trials. Blood samples for hematology and serum chemistry analysis were collected at 90 days posttreatment then submitted for analysis to Colorado State University, Veterinary Teaching Hospital, Clinical Pathology Laboratory, Fort Collins, Colorado, USA. Serum chemistry profiles were obtained using a Hatachi 917 autoanalyzer (Roche/Boehringer Mannheim, Indianapolis, Indiana, USA) for the following parameters: glucose, creatinine, phosphorus, calcium, magnesium, total protein, albumin, globulin, albumin/globulin ratio, bilirubin, creatinine kinase, aspartate aminotransferase, gamma-glutamyltransferase, sorbitol dehydrogenase, sodium, potassium, chloride, and biocarbonate. Values for the following hematological parameters were obtained using an ADVIA 120 autoanalyzer (Bayer Corporation, Tarrytown, New York, USA): nucleated cells, neutrophils, lymphocytes, monocytes, eosinophils, plasma protein, erythrocyte, hemoglobin, packed cell volume, mean corpuscular volume, mean corpuscular hemoglobin concentration, platelets, and fibrinogen. b. Reproductive behavior. The effectiveness of the leuprolide formulation as a contraceptive agent is dependent upon suppression of ovulation and steroidogenesis for the duration of the breeding season. Thus, we tested 2 hypotheses relative to the effects of leuprolide on reproductive behavior of elk: (I) because leuprolide was expected to suppress gonadotrophin secretion and ovulation, we predicted that sexual interactions during the breeding season would be reduced in leuprolide treated females ( Group A) compared to untreated controls (Group B), and (2) since depletion of the leuprolide implant (90 days) was expected prior to anoestrus (late March), we predicted that behavioral oestrus would resume in treated females ( Group A) and the rate of sexual interactions would be higher than that for untreated controls (Group B) �175 To test these hypotheses, we examined the effects of leuprolide on reproductive interactions of male and female elk during 2 time periods; breeding season (defined as the period 10 November- 23 December 1999) and postbreeding season (defined as the period 7 February 27 March 2000). On 2 November 1999, female elk in Group A were treated with leuprolide and released with untreated controls (Group B) into adjoining paddocks (2 ha each). Seven days later (10 November), we placed 3 adult male elk with these groups and initiated behavioral observations. All females were individually identified with color/numeric-coded neck collars. Animals selected as treatments and controls were unknown to observers. Behavioral measurements were made from a distance of 50-250 m from an elevated tower (10 m) situated between adjacent pastures using binoculars and a spotting scope during the day, and a spotlight and night vision scope at night. We recorded selected behaviors using a laptop computer with a behavioral software program. We used focal animal sampling procedures to sample reproductive behaviors of all experimental animals over a 24 -hour period (Lehner 1996). Preliminary observations indicated that elk were most active in morning (0500-0800), late day ( 1400-1700) and night (2000-2400). Thus, time-of-day sampling periods were randomly assigned each week using a randomized block design. Each sampling period consisted of at least two hours of continuous observations. Based on previously reported elk breeding behavior (Morrison et al. 1960, Geist 1982, Rapley 1985), we identified and recorded 19 sexual interactions. Because sample sizes were small, we grouped individual behaviors into 4 general categories: male copulatory, male precopulatory, female precopulatory, and general breeding (Table 1). Copulatory, male precopulatory, and general breeding were interactions of a male directed toward a specific female, while female precopulatory behaviors were actions of a specific female towards a male. Thus, our experimental unit for analyses was the individual female in each breeding group. Behavioral interactions were generally short duration (<30 sec) relative to sampling interval, therefore we recorded the number of occurrences of each event rather than length of time and calculated sexual interaction rates as acts per animal per hour, then multiplied hourly behavioral rates by 24 for a daily rate. Table 1. Description of elk reproductive behavior and associated behavior categories. Behavior category Reproductive behavior General breeding : Male directed behavior related to establishing, maintaining, and defending a group or harem of female elk (e.g. herding guarding, tending) Male precopulatory Male courtship behavior directed toward an individual female to induce or detect oestrus or ovulation (e.g. urine testing, flehmen, tongue flick, lick, smell, or rub female's body, chivy) Female precopulatory Female courtship behavior directed toward dominant male to arouse copulatory behavior (e.g. lick and rub male, mount, lordosis, twitch hocks) Copulatory Male behavior directed toward a receptive female in oestrus (e.g. precopulatory mounts, intromission, pelvic thrust) �176 c. Hormone radioimmunoassay. Serum concentrations ofLH were quantified by means of an ovine (o) LH RIA (Niswender et al. 1969). Elk serum was demonstrated to inhibit binding of 125 l-oLH to LH antiserum in a parallel manner. Likewise, when varying quantities of oLH standard (NIH-OLH-S24) were added to elk serum and samples were subjected to RIA, the values obtained were increased by the quantity of oLH added (r2 = 0.99, slope = 0.92, SEb = 0.22, P = 0.002). These data indicate that the radioimmunoassay (RIA) provided a quantitative assessment ofLH in elk serum. The limit of sensitivity of the LH assay was 0.4 ng mi- 1 . Serum concentrations of progesterone were determined by RIA (Niswender 1973). Sensitivity of the progesterone assay was 0.12 ng m1· 1 . Intra- and inter-assay coefficients of variation for each of these assays were less than 10%. d. Statistical analysis. Hormone concentrations are reported as untransformed arithmetic means ± standard error of the mean (SE). Responsiveness of the pituitary to GnRH analog challenge was assessed in two ways: 1) maximum response (highest concentration of LH (ng mJ- 1) achieved postinjection minus baseline), and 2) total amount ofLH secreted (ng mJ- 1 min•1) estimated by calculating the area under the LH response curve (Abramowitz and Stegun 1968). We analyzed differences among hormone levels using least squares analysis of variance for general linear models (SAS Institute 1993). Responses to treatments were analyzed with one-way analysis of variance for a randomized complete block design with repeated measures structure. Levels of leuprolide formulations were treatments; individual animals were blocks. Factors in the analysis were dose and time. Treatment effects were tested using the animal-within-treatment variance as the error term. Time was treated as a within-subject effect using a multivariate approach to repeated measures (Morrison 1976). A "protected" least significant difference test (Milliken and Johnson 1984) was used to separate means when the overall F-test indicated significant treatment effects (P < 0.05). We tested specific reproductive behavior hypotheses that mean behavior rate was not different between treatment and control groups for both the breeding and postbreeding seasons using an ANOVA model with a repeated measures structure. Similar to the hormonal analysis, time was treated as a within subject effect using multivariate approach to repeated measures (Morrison 1976). To test for treatment effects, we accounted for time-of-day, date effects and their interactions. PROC GENMOD (SAS Institute 1993) was used to estimate and test for differences in mean behavior rate by treatment, time-ofday, and date. Means and standard errors were estimated using least squares, and hypothesis tests were based on type III generalized ~stimating equations that accounted for correlation in repeated measuremep.ts. D. Experiment 3: Antifertility effects on pregnant elk 1. Objectives a. Evaluate the effects of GnRH agonist (leuprolide)on female elk treated during the first trimester of pregnancy. b. Assess nutritional, physiological, or behavioral side-effects that might result from treatment. �177 2. Methods a. Animals. We conducted controlled experiments with 12 adult female elk and 2 adult male elk at the Colorado Division of Wildlife's Foothills Wildlife Research Facility (FWRF), Fort Collins, Colorado during September 1, 2000 to December 15, 2001. On September 1, 2000, two intact male elk were released with 12 female elk. The purpose of mating was to confirm the reversibility ofleuprolide from previous experimental treatments (Baker et al. 2002, in press). Pregnant animals from this mating would then provide the experimental elk for the experiment described in this study plan. b. Treatments. Approximately 60 days postconception ( 1 December for cows at FWRF), all females were evaluated for pregnancy and fetal age determined using transrectal ultrasonography (Willard et al. 1994). Using ultrasound and selected measurements reported for known-age embryos (Morrison et al. 1959), we estimated fetal ages of all pregnant elk. Eight elk with embryos estimated to be 60-75 days old were randomly selected to receive a subcutaneous implant containing 32.5 mg ofleuprolide formulation. Leuprolide was injected subcutaneously on the lateral thorax using an 18 g x 4 cm needle. The remaining four pregnant elk were designated as untreated controls. Treatment and control elk were maintained in the same pastures, fed similar diets, and handled similarly throughout the study. All treatments were applied without tranquilization by moving elk from 5 ha pastures to individual isolation pens, then into a restraining chute, where treatments were applied, then returning elk back into 5 ha paddocks. Animals were observed daily by trained caretakers for general health and for signs of abortion or parturition. c. Sample size. Based on previous reproduction studies with captive elk at FWRF, 4-6 elk per treatment is the minimum sample size needed to provide biologically significant differences among treatment means ( Baker et al. 1995, Baker et al. 2002, in press). We used an unbalanced experimental design to minimize the number of untreated pregnant control elk, since most neonates will be euthanized. Pregnant, control elk were needed to insure that treatment results were not biased due to handling procedures, and/or other uncontrolled variables. d. Measurements l) Pregnancy Rates. We assessed contraceptive effectiveness by determining pregnancy status of all experimental elk. Using transrectal ultrasonography (Willard et al. 1994, 1998 ), we determined pregnancy rates of treated and control elk prior to treatment, and at 60, and 120 days posttreatment. On the day of pregnancy assessment, elk were moved from 5 ha pastures to a handling chute where they were sedated with xylazine hydrochloride (20-200 mg/animal, IM), then scanned using real-time transrectal ultrasound to determine pregnancy status and/or fetal age. Elk were then reversed with yohimbine (0.125 mg/kg, N) and returned to their original pastures. We determined the reversibility of leuprolide by releasing a epididymectomized male elk with treated female elk in October 2001 and conducting a GnRH challenge trial (Baker et al. 1995) to measure LH and P4 levels. 2) Reproductive Behavior. The effects of leuprolide on the breeding behavior of captive elk treated prior to the breeding season is known (Baker et al. 2002, in press), however, these reported effects may or may not be extended to elk treated during early pregnancy. �178 Down-regulation of gonadotroph cells by the action ofleuprolide and subsequent reduced secretion of LH could effectively inhibit progesterone secretion by the corpus luteum. If the effective action of leuprolide is luteolytic, then early embryonic loss could occur (Plotka et al. 1982, Asher et al. 1988, Flint et al. 1991). However, the efficacy of induction of luteolysis by leuprolide in Cervidae is unknown. Furthermore, it's not known if following early embryonic loss, whether female elk will regain normal estrus cycles and behavior. We evaluated these potential behavioral side-effects by monitoring maintenance and breeding behavior of male and female elk before and after leuprolide treatments. Each animal was individually identified using color-coded neck bands or ear tags. We tested the null hypothesis that the frequency of sexual interactions between males and females is similar before and after contraceptive treatment. e. Statistical Analysis. We analyzed data using least squares ANOVA for General Linear Models and the SAS Interactive Matrix Language. Response to contraceptive treatment was analyzed with a two-way factorial analysis of variance for a randomized complete block design with repeated measures structure. Factors in the analysis were treatment and time. Treatment was tested using the animal-within-treatment variance as the error term. Time was treated as a within subject effect using a multivariate approach to repeated measures. We used orthogonal contrast to test for differences among individual means (Morrison 1976). E. Experiment 4: Development of a remote delivery system 1. Objective. Begin evaluating a remote delivery system by comparing effectiveness of subcutaneous and intramuscular administration of leuprolide formulation in suppressing reproduction in female elk. 2. Methods. a. Animals and treatment. We conducted a controlled experiment with 13 adult female elk (713 years of age; 250-300 kg), lintact male elk, and 1 epididymectomized male elk at the Colorado Division of Wildlife's Foothills Wildlife Research Facility (FWRF), Fort Collins, Colorado during 15 August 2001 to 28 March, 2002. Between 15 August and 1 September, 2001, the epididymectomized male elk was released with 13 adult female elk into 2 adjoining paddocks (2 ha). Females were monitored for occurrence of estrus cycles by measuring progesterone leve'is beginning 1 September 200latid wete ·considered reproductively active when concentrations were greater than 1 ng/ml. Treatments were assigned as follows: 3 females were randomly selected to receive a subcutaneous formulation ofleuprolide (32.5 mg)(ATRIGEL, Atrix Laboratories, Inc. Fort Collins, Colorado, USA) by syringe injection; 3 elk were selected to receive an intramuscular leuprolide formulation (32.5 mg) via syringe injection; 4 elk received the leuprolide formulation via a 1 cc, PneDart dart (16 gauge, 3.39 cm. needle) fired from a CO2 -powered Dan-Inject pistol, and 3 elk were designated as untreated controls. Treatments were applied as follows. On the day before application (6 September 2001), experimental elk were moved from pastures to individual isolation pens (5 m x 10 m), weighed(± 0.5 kg), sedated with xylazine hydrochloride (Rompun; Bayer AG, Leverkusen; 25-200 mg/animal, i.m) and fitted nonsurgically with indwelling jugular catheters. The next day, treatment and placebo treatments were administered. In order to accurately determine the precise dose of leuprolide formulation remotely delivered to each elk, syringe darts were �179 weighed (0.001 g) before and after injection. Prior to darting, elk were placed in a handling chute and lightly sedated with xylazine hydrochloride (15-20 mg/animal, i.v.). This dose allowed animals to remain standing in the chute and minimized excitation associated with discharge of the dart gun. With the exception .of two animals, one dart per animal was fired from approximately 3 meters into the middle gluteus maximus muscle of the standing elk. Once all elk had been treated, sedation was reversed with yohimbine (30 mg) (Antagonil®, Wildlife Laboratories, Fort Collins, Colorado, USA) and animals were returned to individual isolation pens. b. Measurements. Approximately 1 hour following treatment applications, we measured 24hour LH response of elk treated with the leuprolide formulation and untreated control elk. Blood samples (5 ml) were collected via jugular catheters at 0, 120, 180, 240, 300,360,480 min then at 10, 16, and 24 hr after injection. Catheters were flushed after each collection with sterile saline solution. After the last blood collection, catheters were removed and animals returned to 5 ha pastures. The effect of leuprolide formulation on the duration of suppression ofLH was determined by periodically conducting pituitary stimulation trials. These trials were conducted during 29 October 2001 to 28 March 2002 to determine the capability of LH cells to respond to stimulation with an exogenous dose of GnRH analog (D-Ala6-GnRH-Pro 9ethylamide; Sigma Chemical Company, St. Louis, Missouri, USA). Pituitary stimulation trials were conducted with treated and control elk at 30, 60, 90, 120, 160, and 190 days posttreatment. Stimulation trials were conducted according to the following procedures: On the day of testing, treated and control elk were moved from 5 ha pastures to individual isolation pens, weighed, sedated (as previously described), and fitted nonsurgically with indwelling jugular catheters. GnRH analog (1 µg/50 kg body weight) was administered through the cannula and blood samples were collected (5 ml) at 0, 60, 120, 180,240, 300, 360, and 480 minutes posttreatment. After collections, blood was stored at 4° C for 24 hours until serum was obtained by centrifugation (1500 RCF for 15 minutes). Serum samples for progesterone levels were also collected from each elk on each of these trial days. Serum was stored at -20° C until analyzed for LH and progesterone. Following the last blood collection, catheters were removed and elk were returned to the holding pastures. The effect of leuprolide formulation on reproduction in treated and control elk was determined by measuring pregnancy rates using the presence or absence of pregnancy specific protein B (PSPB) (BioTracking, Moscow, Idaho, USA) in serum collected at approximately 100 and 215 days of gestation (Huang et al. 2000). c. Statistical analysis. Responsiveness of the pituitary gland to GnRH analog_ stimulation was assessed in two ways: (1) maximum response (highest concentration of LH (ng/ml) achieved after injection minus baseline), and (2) total amount ofLH secreted (ng/ml/min) estimated by calculating the area under the LH response curve (Abramowitz and Stegun, 1968). Differences among hormone concentrations were tested using least squares ANOV A for general linear models (SAS Institute, 1997). Responses to treatment were analyzed with oneway ANOV A for a randomized complete block design with repeated measures. Treatment effects were determined using the total animal-within-treatment variances as the error term. Time was treated as a within-subject effect, using a multivariate approach to repeated measures (Morrison 1976). A "protected" least significant difference test (Milliken and Johnson, 1984) was used to separate means when the overall F test indicated significant treatment effects (P < 0.05). �180 RESULTS A. Experiment 1: Dose response Administration of sustained release formulations of leuprolide to female elk resulted in an acute, transient rise in serum LH concentrations irrespective of dose. Maximum LH concentrations (15.6 ± 0.93 ng mi-1) occurred approximately 3 hours following treatment and were similar across all treatment levels (Fig. 1). Following peak response, there was a rapid decline in LH to basal levels during the next 24 hours. Total LH secretion (ng m1- 1 min-1) did not differ among treatments and all treatments resulted in higher LH secretion than controls (P s 0.002). Leuprolide reduced serum LH secretion to nondetectable levels in treated females for 130 days posttreatrnent (Fig. 2). Differences in mean maximum serum LH were significantly lower (P ~ 0.031) in treated elk compared to untreated controls at all sampling periods. For untreated females, mean maximum LH fluctuated from a high of 19.3 ± 4.2 to a low of 3.5 ± 0.06 ng m1- 1 . This variation was likely related to the phase of the oestrous cycle when control females were challenged with GnRH and the influence of fluctuating levels of estradiol and progesterone on LH secretion (Goodman and Karsch 1980). B. Experiment 2: Antifertility and behavioral effects on nonpregnant female elk a_ Pregnancy rates, hormonal measurements, and blood parameters. Because Experiment 1 did not establish a minimum effective dose ofleuprolide for LH suppression, we arbitrarily reduced the leuprolide formulation by approximately 20% below the lowest concentration tested in Experiment 1, to 32.5 mg. This dose ofleuprolide prevented pregnancy in all treated females (Group A) while the pregnancy rate of control females (Group B) was 100%. Treated females tested negative and controls positive for PSPB on all sampling dates. Estimated conception dates for pregnant elk ranged from 10 November to 19 November 1999 and parturition occurred between 12 July and 26 July 2000. Leuprolide caused a significant reduction (P ~ 0.035) in mean maximum serum LH (Fig. 3) and P4 (Fig. 4) concentrations in treated females (Group A) with a return to pretreatment levels the following breeding season (11 November 2000). Serum LH was reduced to non-detectable concentrations by 92 days posttreatment and remained at this level until day 225. In one treated female, LH remained at baseline for 250 days posttreatrnent. Maximum LH response was lower (P ~ 0.012) in treated compared to non-pregnant controls (Group C) at 30, 92, 135, 165, and 193 days following treatment. Serum LH of untreated elk declined significantly (P = 0.024) between April and May with the onset of anestrus, then returned to pretreatment levels indicative of estrus in November 2000. • Serum P4 levels of treated females followed a similar pattern to that observed for serum LH (Fig. 4). Progesterone levels were similar in treated and control elk until day 30, thereafter, serum progesterone remained at basal concentrations in treated females until day 225 of the trial, indicating that additional ovulations did not occur. Control females maintained increased serum P4 content, reflecting continued regular estrous cycles within this group until day 165 ( 18 April) when the effects of anestrus reduced P4 to basal levels. Similar to serum LH, P4 content then increased during November 2000 to pretreatment concentrations in both treated and untreated elk (Fig. 4). We evaluated 13 hematology and 19 serum chemistry parameters in treated and untreated elk females. With the exception of creatinine kinase (CK), a muscle derived enzyme, all individuals were clinically similar. Elk in the treatment group showed moderately elevated CK levels (400-702 IU L-1). Creatinine kinase levels can increase in unconditioned animals following vigorous exercise and remain elevated for 4-6 hours (Lefebvre et al. 1994). Handling procedures for blood sampling in treated females were often �181 more physically rigorous than those for controls due to the need to separate females from males. Thus, the elevated CK levels in treated elk compared to controls likely reflect a bias due to a difference in animal handling prior to blood sample collections, rather than a treatment-induced response. b. Reproductive behavior. We observed male to male dominance interactions immediately following their release into the pastures with treated and untreated females. Within 2.5 weeks, one male established dominance over the other two. Thereafter, subdominant males retreated to remote locations in the pastures and rarely interacted with females or the dominant male for the remainder of the experiment. During the breeding season, we observed reproductive interactions of males and females on 34 days during 10 November to 23 December 1999. We analyzed 63 sampling periods (134.5 h): 20 periods at dawn (45.7 h), 6 at mid-day (13.5 h), 20 at dusk (42.8 h), and 17 at night (32.6 h). The average length of the observation periods was 2.1 (SE= 0.10) h. Postbreeding observations occurred on 14 days during 7 February to 27 March. We analyzed data from 16 sampling periods (54.7 h): 6 periods at dawn (22.5 h), 2 at mid-day (7.5 h), 7 at dusk (22.2 h), and 1 at night (2.5 h). Observation periods averaged 3.4 (SE= 0.24) h. Contrary to our first hypothesis, sexual interactions during the breeding season were not diminished in leuprolide-treated females compared to controls. Instead, breeding behavior rates were similar for treated and untreated females for all behavior categories (Fig. 5). Although we did not detect a significant treatment x time interaction, copulatory (P = 0.064), male precopulatory (P = 0.083), and female precopulatory (P = 0.072) behaviors approached significance and are notable. For these 3 behavior categories, the daily behavior rate decreased over time for untreated females, but remained constant for treated elk. We also failed to reject our second hypothesis. Treated females did not resume normal oestrus cycles during the postbreeding season and reproductive behavior rates did not increase compared to untreated controls. We observed almost no sexual interactions between the dominant male and treated or untreated females during the postbreeding season. There were no copulatory or female precopulatory behaviors recorded, and too few male precopulatory (~ 0.17 day·1) and general breeding(~ 0.30 day·1) behaviors to analyze. C. Experiment 3: Antifertility effects on pregnant elk. Leuprolide administered has a 32.5 mg subcutaneous formulation to elk during the first trimester of pregnancy failed to induce fetal loss. Fetal age at the time of treatment of treated females ranged from 30 - 90 days of age and from 50 - 90 days for control elk. Treated and control females .were positive for PSPB at all sampling dates during gestation and all produced a calf at parturition. Dystocia was observed in 3 of 6 females but did not appear to be related to treatment. During the breeding season reproductive behaviors were similar (P = 0.45) for treated and control female elk. We observed almost no sexual interactions during the postbreeding season. D. Experiment 4: Development of a remote delivery system. Administration of a 90-day sustained release formulation ofleuprolide to female elk resulted in an acute, transient rise in serum LH levels irrespective of mode of delivery (Fig. 6). Maximum LH concentrations occurred approximately 3.5-4.5 h following treatment and were highest (84.9 ± 5.3 ng/ml) for the �182 intramuscular syringe treatment, followed by intramuscular dart (42.2 ± 15.8 ng/ml) and subcutaneous syringe (23.0 ± 5.1 ng/ml). Following peak response, there was a rapid decline in LH basal levels during the next 24 hours. Leuprolide reduced serum LH secretion to non-detectable levels in all treatment groups for 120 days posttreatment. Between 120 and 160 days posttreatment, LH levels in the intramuscular syringe and dart treatments increased substantially over the subcutaneous syringe treatment and control females and remained elevated for the duration of the experiment. In contrast, LH levels for the subcutaneous syringe group remained at basal levels (Fig. 6). Serum P4 levels followed a different pattern than that observed for serum LH (Fig. 7). After 60 days posttreatment, P4 concentrations in all treatment groups declined to basal levels and remained at these levels for the remainder of the experiment. Regardless of mode of delivery, leuprolide formulation prevented pregnancy in all treatment groups, whereas pregnancy rate of control females was I 00%. Leuprolide-treated females tested negative and controls positive for PSPB on both sampling dates. DISCUSSION Successful application of fertility control technology for wildlife is dependent on development of contraceptive agents that are safe, practical, and effective. Current technology is limited due to problems of treatment implementation and concerns for the health of target and non-target species. In the present study, we evaluated a promising non-steroidal, non-immunological contraceptive technology for controlling fertility in female elk. Administration of a sustained release formulation containing leuprolide to captive female elk prior to the breeding season, resulted in decreased LH and progesterone secretion, temporary suppression of ovulation and steroidogenesis, and effective contraception without detrimental behavioral or physiological side-effects. The acute increase in serum LH immediately following leuprolide treatment was consistent with previous studies in cattle (D'Occhio et al. 1996), sheep (Nett et al. 1981), horses (Montovan et al. 1990) and African elephants (Loxodonta africana) (Brown et al. 1993). There was little variation among elk in their serum LH response to different doses ofleuprolide, indicating either low variability in the amount and duration of agonist released or doses so high that any variation was masked. The minimum level of leuprolide needed to suppress estrus in female elk was not determined in this study. All doses of leuprolide were equally successful in reducing LH concentrations for the duration of the 130 trial. Additional research to establish a minimum .effective dose ofleuprolide would enhance the economic practicalify of this contracep~v~ agent. The cessation of estrous cycles in females treated with leuprolide and the return to apparently normal ovarian function after depletion of the agonist implant was consistent with findings for females in other species (D'Occhio et al. 1996; Evans and Rawlings, 1994; Fraser et al. 1989). The effectiveness of leuprolide as a contraceptive agent is dependent on suppression of ovulation from the inception of the breeding season to the onset of anestrus, a period of approximately 200 days for elk. Leuprolide inhibited ovulation for > 190 days, 2 times longer that the formulated 90 day delivery period. The prolonged suppression of gonadotrophin secretion may occur for several reasons. Among :these are that release ofleuprolide from the implant may have continued beyond the formulated 90 day period. Certainly, LH secretion remained suppressed for more that 13 0 days in Experiment 1. Likewise, leuprolide treatment may have induced prolonged suppression of gonadotroph function (i.e. extending beyond the duration of the implant). In other ruminants, if gonadotroph function is suppressed for an extended duration, a recovery period of 30-60 days following removal of the suppression is necessary before pituitary content of LH and gonadotropin secretion can return to normal levels (Nett 1987). Thus, �183 if duration of leuprolide release from the implant was 130 days and recovery of gonadotroph function requires approximately 60 days, this would be sufficient to carry the reduced secretion of LH into the normal anoestrous period when secretion of LH would be photoperiodically suppressed. If this is indeed true, then a single treatment should provide a contraceptive effect for approximately one breeding season. The effectiveness ofleuprolide in preventing pregnancy in female elk is conditional. Successful prevention of fertility was achieved by treating elk prior to the breeding season. The use of leuprolide as a contragestive in female elk during early pregnancy was unsuccessful. Since we did not measure LH responses to leuprolide treatment in pregnant elk, the mechanism for failure is unknown. We speculate that complete down-regulation of LH receptors did not occur and LH levels were high enough to stimulate an LH surge and subsequent ovulation. The overall rates of sexual interactions between treated and control elk were not different during the breeding and postbreeding seasons. During the breeding season, the dominant male established and defended a single harem of treated and untreated females. Reproductive behaviors during the breeding season between the dominant male and harem females followed a pattern similar to that described for free-ranging elk (Geist 1982). Treated and untreated females were courted, bred, and defended with equal frequency, however the pattern of reproductive interactions changed over time. Once untreated females became pregnant, reproductive behavior rates decreased, whereas, copulatory, and male and female precopulatory rates remained constant over time in treated females. These extended sexual interactions were generally intermittent and may have been related to fluctuating levels of progesterone and oestradiol. Estrus can occur with relatively low estradiol concentrations, if coupled with low progesterone content. In domestic sheep, pre-exposure to progesterone stimulates estrus behavior at much lower concentrations of estradiol once progesterone is decreased in circulation (Robinson 1954). Therefore, since these animals had ovulated prior to leuprolide treatment, they became very sensitive to low levels of estradiol, and since ovulation and corpus luteum formation were blocked they continued to show estrus behavior with basal estradiol levels. Regardless of the mechanism involved, disruption of normal behavioral patterns are not a desirable sideeffect of contraceptive treatments. However, without carefully designed large-scale investigations with larger sample sizes, and under more natural conditions, we can only speculate on the significance of these behavioral alterations on the health and social organization of treated populations. Before leuprolide can be considered a practical and efficacious approach for wildlife contraception, development of a reliable remote delivery system is needed. Our pilot efforts to develop such a system were promising, however, the small sample size (n = 3) used in this experiment support only guarded optimism. It appears that the rise in LH levels observed in females treated with syringe dart delivery of leuprolide formulation were not high enough to stimulate ovulati.on and conception. Clearly, additional research with larger sample sizes is needed to confirm or reject these findings. CONCLUSION The objective of the work reported here was to evaluate the contraceptive potential of a GnRH agonist (leuprolide) formulation in female elk, provide evidence of physiological and behavioral side effects of treatment (if any), and assess the potential for remote delivery. We conclude that leuprolide administered as a controlled release formulation prior to the breeding season, offers a new approach to reversible contraception in wild ungulates that overcomes problems associated with existing technology. �184 First, leuprolide formulation improves practical application of contraception because a single treatment can induce infertility in females without relocating and treating specific individuals each year. Second, leuprolide acetate is a neuropeptide, thus the proteinaceous nature of this agent eliminates the possibility of passage through the food chain to non-target species. Third, behavioral side-effects were minimal. Sexual interactions of treated females were extended early in the breeding season but recurrent estrous cycling and ovulation did not occur. Fourth, there were no short-term physiological side-effects of treatment. Treated animals appeared healthy and seasonal intake and body weight dynamics normal. However, before this technology can be considered a practical and efficacious approach for wildlife, additional research is needed to ascertain minimum effective dose, verify effective treatment duration, and develop a remote delivery system for administering leuprolide formulation to unrestrained animals. ACKNOWLEDGMENTS Our research was supported by the U. S. National Park Service, Rocky Mountain National Park (Grant 1520-9-9002) and the Colorado Division of Wildlife (Federal Aid to Wildlife Restoration, Project 153R4). We thank Dr. Delwar Hussain and Dr. Richard Dunn at Atrix Laboratories, Ft. Collins, Colorado for the generous donation of the leuprolide acetate formulations used in these investigations and for their technical assistance in delivery technology. We gratefully acknowledge and appreciate the technical assistance of Joan Ritchie in overall organization and execution of blood sampling protocols and animal handling. Darby Finley and Elizabeth Wheeler provided invaluable assistance with behavioral observations and general animal training and husbandry. We thank David Bowden for statistical consultation and analysis. LITERATIJRE CITED Abramowitz, M., and L. A. Stegun. 1968. Handbook ofMathematical Functions. Dover Publications, New York, USA. Adam, C. L., C. E. Moir, and T. Atkinson. 1985. Plasma concentrations of progesterone in female red deer (Cervus elaphus) during the breeding season, pregnancy, and anoestrus. Journal of Reproduction and Fertility 74:631-636. Asher, G. W. 1988. Hormonal changes during the luteal regression in farmed fallow deer, Dama dama. Journal of Reproduction and Fertility 84:379-386. Baker, D. L., M.A. Wild, M. M. Conner, H.B. Ravivarapu, R. L. Dunn, and T. M. Nett. 2002. Effects of GnRH agonist (leuprolide) on reproduction and behayior in fe~ale vyapiti (Cervus elaphu_s nelsoni). Reproduction (Suppl) 60: (in press). • _ , M. W. Miller, and T. M. Nett. 1995. Gonadotropin-releasing hormone analog-induced patterns ofluteinizing hormone secretion in female wapiti (Cervus elaphus nelsoni) during the breeding season, anestrus, and pregnancy. Biol. ofReprod. 52:1193-1197. Becker, S. E., and L. S. Katz. 1995. Effects of gonadotropin-releasing hormone agonist on serum LH concentrations in female white-tailed deer. Small Ruminant Research 18:145-150. Belchetz, P. E., T. M. Plant, Y. Nakai, E. J. Keogh, and E. Knobil. 1978. Hypophysial response to continuous and intermittent delivery of hypothalamic gonadotropin-releasing hormone. Science 202: 631-633. Bergfeld E., M. J. D'Occhio, and J.E. Kinder. 1996. Pituitary function, ovarian follicular growth, and plasma concentrations of l 7P-oestradiol and progesterone in prepubertal heifers during and after treatment with the luteinizing hormone-releasing hormone agonist deslorelin. Biology of Reproduction 54:776-782. Brown, J. L., D. L. Schmitt, A. Bellem, L. H. Graham, and J. Lehnhardt. 1999. Hormone secretion in the Asian elephant (Elaphus marimus): Characterization of ovulatory and anovulatory LH surges. Biology of Reproduction 61: 1294-1299. �185 Cole, G. F. 1971. An ecological rationale for the natural or artificial regulation of native ungulates in parks. Transactions of the North American Wildlife and Natural Resources Conference 36:417-426. D'Occhio, M. H., W. J. Aspden, and T. Whyte. 1996. Controlled reversible suppression of oestrous cycles in beef heifers and cows using agonist ofluteinizing hormone - releasing hormone. Journal of Animal Science 74:218-225. _ _, M. J., and W. J. Aspden. 1999. Endocrine and reproductive responses of male and female cattle to agonist of gonadotrophin-releasing hormone. Journal of Reproduction and Fertility Supplement 54:101-114 DeNicola, A. J., D. J. Kesler, and R. K. Swihart. 1997. Remotely delivered prostaglandin F- 2 alpha implants terminate pregnancy in white-tailed deer. Wildlife Society Bulletin 25:527-531. Dunn, R. L., J.P. English, D.R. Cowan, and D. P. Vanderbilt. 1994. Biodegradable in situ forming implants and methods of producing the same. U. S. patent no. 5,278,201. Evans, A. C. 0., and N. C. Rawlings. 1994. Effects of a long-acting gonadotrophin-releasing hormone agonist (Leuprolide) on ovarian follicular development in prepubertal heifer calves. Canadian Journal of Animal Science 74:649-656. Goodman, R. L., and F. J. Karsch. 1980. Pulsatile secretion ofluteinizing hormone: differential suppression by ovarian steroids. Endocrinology 107: 1286-1290. Fagerstone, K. A, M. Coffey, P. Curtis, R. Dolbeer, G. Killian, L.A. Miller, and L. Wilmot. 2001. Wildlife contraception. Wildlife Society Technical Review. Proceedings of the Wildlife Society 8 th Annual Conference, Reno, USA Flint, A.P.F., E. L. Sheldrick, T. J. McCann, B. R Brinklow, and A.S.I. Loudon. 1991. Prostaglandininduced secretion of oxytocin and prolactin in red deer (Cervus elaphus) and Pere David's (Elaphuro.s davidianus) deer hinds: Evidence for oxytocin ofluteal origin. General and Comparative Endocrinology 83:432-438. Garrott, R. A., P. J. White, and C. A. Vanderbilt-White. 1993. Overabundance: an issue for conservation biologist? Conservation Biology 7:946-949. ~ and D. B. Siniff. 1992. Limitations of male-oriented contraception for controlling feral horse populations. Journal of Wildlife Management 56:456-464. Geist, V. 1982. Adaptive behavioral strategies. Pages 219-277 in Elk of North America: Ecology and Management. J. W. Thomas and D. E. Toweill, editors. Stackpole Books, Harrisburg, USA. Jacobsen, N. K., D. A. Jessup, and D. J. Kesler. 1995. Contraception in captive black-tailed deer by remotely delivered norgestomet ballistic implants. Wildlife Society Bulletin 23:718-722. Hess, K. H. 1993. Rocky Times in Rocky Mountain National Park. University of Colorado Press. Niwot Colorado, USA. Huang, F., D. C. Cockrell, T. R. Stephenson, J. H. Noyes, and R. G. Sasser. 2000. A serum pregnancy test with specific radioimmunoassay for moose and elk pregnancy specific protein B. Journal of Wildlife Management 64:492-499. Hobbs, N. T., D. C. Bowden, and D. L. Baker. 2000. Effects of fertility control on populations of ungulates: general, stage-structured models. Journal of Wildlife Management 64:473-491. Houston, D. B. 1971. Ecosystems of national parks. Science 172:648-651. Jewell, P.A., and S. Holt. 1981. Problems in management oflocally abundant wild animals. Academic Press. 32lpp. Lefebvre, H.P., P. L. Toutain, J.P. Serthelon, V. Lassourd, L. Gartley, and J.P. Braun. 1994. Pharmcokinetic variables and bioavailability from muscle of creatinine kinase in cattle American Journal of Veterinary Research 55:487-497 Lehner, P. N. 1996. Handbook of Ethological Methods. Second Edition. Cambridge University Press, Cambridge, UK. Kirkpatrick, J. F., and J. W. Turner, Jr. 1985. Chemical fertility control and wildlife management. Bioscience 35:485-491. �186 ~ ~ I. K. M. Lui, R. C. Fayrer Hosken, and A. T. Rutberg. 1997. Case studies in wildlife immunocontraception: wild and feral equids and white-tailed deer. Reproduction, Fertility, and Development 9: 105-110. McCullough, D.R., K. W. Jennings, N. B. Gates, B. G. Elliot, and J.E. Didonato. 1997. Overabundant deer populations in California. Wildlife Society Bulletin 25:478-483. McNeilly, A. S., and H. M. Fraser. 1987. Effect of gonadotrophin-releasing hormone agonist-induced suppression of LH and FSH on follicle growth and corpus luteum function in the ewe. Journal of Endocrinology 115:273-282. Milliken, G. A., and D. E. Johnson. 1984. Analysis of Messy Data. Volume I. Designed Experiments. Lifetime Leaming Publications, Belmont,California, USA Montovan, S. M., P. P. Daels, J. Rivier, G. H. Hughes, G. H. Stabenfeldt, and B.L. Lasley. 1990. The effect of potent GnRH agonist on gonadal and sexual activity in the horse. Theriogenology 33: 13051321. Morrison, D. F. 1976. Multivariate Statistical Methods. McGraw-Hill Book Co., New York, USA. Morrison, J. A. 1960. Characteristics of estrus in captive elk. Behavior 16:84-92. Nett, T. M., M. E. Crowder, G. E. Moss, and T. M .. Duello. 1981. GnRH-receptor interaction. V. downregulation of pituitary receptors for GnRH in ovariectomized ewes by infusion of homologous hormone. Biology of Reproduction 24:1145-1155. Nett, T. M. 1987. Function of the hypothalamic-hypophyseal axis during the post-partum period in ewes and cows. Journal of Reproduction and Fertility Supplement 34:210-213. Niswender, G.D., L. E. Reichert, Jr., A. R. Midgley, Jr., and A. V. Nalbandov. 1969. Radioimmunoassay for bovine and ovine luteinizing hormone. Endocrinology 84: 1166-1173. _ _ . 1973. Influence of the site of conjugation on the specificity of antibodies to progesterone. Steroids 22:413-424. Plotka, E. D., U.S. Seal, L. J. Venne, J. J. Ozoga. 1982. Reproductive steroids in white-tailed deer. IV. Origin of progesterone during pregnancy. Biology of Reproduction 26:258-262. Ravivarapu, H.B., K. L. Moyer, and R. L. Dunn. 2000. Sustained activity and release ofleuprolide acetate from an in situ forming polymeric implant. AAPS PharSciTech I :23-26. Rapley, M. D. 1985. Behavior of wapiti during the rut. In Biology of Deer Production Eds. P. F. Fennessy and K. R. Drew. Bulletin No. 22, Royal Society ofNew Zealand, Wellington 357-361. Robinson, T. J. 1954. Relationship of oestrogen and progesterone in oestrous behavior of the ewe. Nature 173:870-874. SAS Institute. 1993. SAS/STAT user's guide. Release 6.03 edition. SAS Institute, Cary, North Carolina, USA. Seagle, S. W., and J. D. Close. 1996. Modeling white-tailed deer population control by contraception. Biological Conservation 76:87-91. _ Singer, F. J., L. C. Zeigenfuss, R G. Cates, and D. T. Barnett. 1998. Elk, multiple factors, and persistence of willows in national parks. Wildlife Society Bulletin 26:419-428. Smith, B. L. 200 I. Winter feeding of elk in western North America. Journal of Wildlife Management 65:173-190. White, C. A., C. E. Olmsted, and C. E. Kay. 1998. Aspen, elk, and fire in the Rocky Mountain national parks of North America. Wildlife Society Bulletin 26:449-462. Willard, S. T., R. G. Sasser, J.C. Gillespie, J. T. Jaques, T. H. Welsh, Jr., and R. D. Randel. 1994. Methods for pregnancy determination and the effects of body condition on pregnancy status in Rocky Mountain elk. Theriogenology 42:1095-1102. ~ R G. Sasser, J. T. Jaques, D. R White, D. A. Neuendorff, and R. D. Randel. 1998. Early pregnancy detection and the hormonal characteristics of embryonic-fetal mortality in fallow deer (Dama dama). Theriogenology 49:861-869. Zeignefuss, L., F. J. Singer, and D. C. Bowden. Vegetation and elk responses following release from artificial controls within Rocky Mountain National Park, 1968-1996. Report to the Superintendent, Rocky Mountain National Park. �187 Fig.1 20 T 15 -1 E .s 0) ---- 0mg -ai.- - 45 mg ----e- - 90mg -T - 180mg 10 :r: _J 5 0 2 1 0 4 3 5 7 6 8 12 24 Time (hours) Figure 1. Twenty-four hour profile of serum LH concentrations in untreated female elk ( •, n = 2) and elk treated with a sustained release formulation containing 45 (A, n = 2), 90 (o, n = 2), and 180 mg (T, n = 2) ofleuprolide. Results are shown as± SE. Fig.2 30 - - - 0 mg 45 mg -Ir- --e-- 90 mg ~ 0) C 20 I _J E E ·x ro :::J ~ C ro Q) 10 I ~ I )\ \ I 180 mg 1 /I' -1 E -T - \ J l \ 0 0 .5 1 35 75 110 130 Time (days) Figure 2. Profiles of 130--day LH concentrations in untreated ( •, n = 2) female elk and elk treated with a sustained release formulation containing 45 (A, n = 2), 90 (o, n = 2), and 180 mg {T, n = 2) of leuprolide. Results are shown as± SE. �188 Frg.l 30 - 25 E 0) C ..__, 20 I I E E ::J ·xCJ ~ T 1 "" e. "e- -- -- --1...... . . . . ...... _J 15 -0- - Control ____.,._ Leuprolide • 'T0 - - - 4\T J\ ..l 10 \ C \ CJ \ Cl) ~ 5 \ "o-.._ .L 0 Pretrmt 30 92 135 165 193 225 373 Time (days) Figure 3. Profiles of mean maximum serum LH concentrations for untreated female elk ( •, n = 4), and elk treated with a sustained release formulation containing 32.5 mg leuprolide ( o, n = 4). Results are shown as ± SE. Fig.4 2.40 -0- - Control ----- Leuprolide ,,......_ E 1.80 I 0) C ...__, Q) Q C ....Q) 0 - 120 t i) Q) ... 0 a.. C a, T T -~ C) 1 ............ 1 ,.,.,.,. . 0.60 Q) l' I /~ 'if 1 I \ I \ /J 1 I \ ~ \ \ 0.00 Pretmrt. 30 92 135 165 193 225 373 Time (days) Figure 4. Profiles of mean maximum serum progesterone concentrations for untreated female elk(•, n = 4), and elk treated with a sustained release formulation containing 32.5 mg leuprolide ( o, n = 4). Results are shown as± SE. �189 Fig 5 10 '>-(U 8 -a ~ 0 ·::;: - 6 (U ..c Q) ..Q Control Ba Leuprolide Q) ro 4 0:::: .._ 0 ·::;: Jg 2 Q) co 0 Gen. Breed. Male Precop. Female Precop. Copulatory Behavior Category Figure 5. Mean(± SE) reproductive behavior rates during the breeding season for untreated (n = 5) and leuprolide-treated (n = 5) female elk. Results are shown as± SE. Fig. 6 90 ~ - 72 1 ---- I;t\\ I \ ---.-- -0-- IM Syringe E C) C "-' I 54 ....J a.. / t\\\ //I,\ I 36 I C (ll (I) ~ SC syringe ----s::J - Control ~ (ll (I) IM Dart 18 ,,510 0 .5 '.I 30 60 90 120 160 190 Time(days) Figure 6. Profiles of mean maximum serum LH concentrations for untreated female elk (e, n = 3), and elk treated with a 32-5 mg leuprolide formulation delivered intermuscularly via syringe dart (0, n = 3), intermuscular syringe (.A., n = 3), and subcutaneous syringe (T, n = 3). Results are shown as± SE. �190 Fig. 7 - - k- • IM syringe _. •• SC syringe --e · Control ---€r IM Dart 2 E ........ C) C ...__,, Cl) 0 I 1.- .... Cl) "' -·• 1 T - C 1 1 1 - T l Ti -V Cl) C) 0 1 1.- a.. C <ti Cl) ~ 0 ---Pretrmt 30 60 90 120 160 190 Time(days) <•. Figure 7. Profiles of mean maximum serum progesterone concentrations for untreated female elk n= 3), and elk treated with a 32.5 mg leuprolide formulation delivered intermuscularly via syringe dart (0, n = 3), intermuscular syringe(.&, n = 3), and subcutaneous syringe (T, n =3). Results are shown as± SE. �57 JOB PROGRESS REPORT State of Division of Wildlife - Mammals Research Colorado Work Package No. --'3'""'0'-"0=2_ _ _ _ _ __ Elk Conservation Task No. Technical Support for Elk and Vegetation Management Environmental Impact Statement for Rocky Mountain National Park Federal Aid Project RMNP W-153-R Period Covered: July 1, 2002 - June 30, 2003 Authors: D. L. Baker, M.A. Wild, and M. M. Conner Personnel: M. Coffey, G. Dodd, B. Gill, T. Johnson, M. Monello, R. Monello, D. Plattner, J. Powers, J. Ritchie, R. Spowert, T. Nett, D. Hussain, R. Dunn, K. Zollers. ABSTRACT Overabundant wild ungulate populations have become a significant concern for natural resource managers in many parts of North America. Wild ungulates can do serious and lasting harm to many plant communities, and preventing such damage requires controlling the growth of their populations. In protected areas such as national parks, traditional methods of population control may not be feasible or publically acceptable. In these situations, alternative methods of population control are needed. One alternative is controlling the fertility of females. In this study, we evaluated the feasibility of using gonadotropin releasing hormone (GnRH) analog to control reproduction in free-ranging female elk in Rocky Mountain National Park. During fall of 2002, we captured, radio-collared and treated 34 adult elk. Seventeen elk were treated subcutaneously with a controlled release bio-implant containing 32.5 mg of leuprolide and seventeen elk were treated with the same formulation without leuprolide. We evaluated the effects ofleuprolide treatments on reproductive rates, body condition, behavior, and daily activity patterns offemale elk during September 2002 to April 2003. Leuprolide administered as a sustained release formulation was 100% effective in preventing pregnancy in female elk. Body condition of all experimental elk declined from fall 2002 to spring 2003. Changes in loin depth and body condition score were similar (P. 0.254) for both treated and control elk, whereas overwinter loss in mean percent rump fat was greater (P. 0.057) for treated elk compared to controls. There were no differences (P = 0.36) in reproductive behavior rates during the breeding season between treated and control elk. �58 TECHNICAL SUPPORT FOR ELK AND VEGETATION MANAGEMENT ENVIRONMENTAL IMPACT STATEMENT FOR ROCKY MOUNTAIN NATIONAL PARK D. L. Baker, M.A. Wild, and M. M. Conner P. N. OBJECTIVE Conduct experiments with captive and free-ranging elk to evaluate fertility control as an management alternative for controlling elk populations in Rocky Mountain National Park (RMNP), Colorado. SEGMENT OBJECTIVES 1. Capture, radio-collar, and apply fertility control treatments to a target sample of free-ranging adult female elk in RMNP during September 2002. 2. Evaluate the effects of fertility control on reproductive rates of treated and non- treated adult female elk and the reversibility of these effects if they occur. 3. Evaluate the effects of fertility control on body condition of treated and non-treated adult female elk. 4. Evaluate the effects of fertility control on reproductive behavior and daily activity patterns of treated and non-treated adult female elk. INTRODUCTION Overabundant wild ungulate populations have become a significant problem for natural resource managers in North America. Unregulated populations can cause adverse effects that are ecological, economic, or political in scope and resolving these issues often requires controlling animal abundance (Jewell and Holt 1981, Garrott et al. 1993, McCullough et al.1997, Smith 2001). In Rocky Mountain National Park (RMNP), Colorado, the impact ofherbivory by elk has emerged as a fundamentally important problem for those who manage the Park and its wildlife (Hess 1993, Zeignefuss et al. 1996). In 1968, RMNP adopted a natural-regulation policy for management of ungulates (Cole 1971, Houston 1971) with the objective of allowing density dependent processes to regulate elk numbers within park boundaries and use sport hunting to harvest as many animals as possible in areas surrounding the Park. Recently, however, Park managers have become concerned that possible unnatural concentrations of elk may be altering natural plant communities and ecosystem sustainability. Soil conditions and the status of willow and aspen plant communities have declined. Wet meadow, dry grasssland, and alpine and subalpine sites show evidence of deterioration from overgrazing by elk (Singer et al. 1998, White et al. 1998). As a result of the decline in these vegetation types and the diversity of the animal species that are associated with them, the Park and other natural resource agencies are evaluating alternative management strategies for reducing elk densities within RMNP and the surrounding Estes Valley. One alternative being considered is controlling the fertility of female elk. Fertility control has been widely advocated as an alternative to lethal methods of population control for wildlife and considerable research �59 has been directed toward development of different contraceptive agents (see reviews by Kirkpatrick and Turner 1985, Fagerstone et al. 200 I). Field and laboratory studies have evaluated the efficacy of delivery of contraceptives to ungulates (Jacobsen et al. 1995, DeNicola et al. 1997, Kirkpatrick et al. 1997) and models have been developed to represent effects of fertility control on the population dynamics of individual species and populations (Garrott and Siniff 1992, Seagle and Close 1996, Hobbs et al. 2000). To date, most contraceptive research for wild ungulates has focused on the development of immunocontraceptive vaccines and steroidal hormonal agents. However, after more than 40 years of research, the success of these approaches have been primarily limited to captive wildlife and small localized urban populations of wild ungulates. To meet this challenge, new technologies and approaches are needed if fertility control is to become practical and acceptable management tool for controlling overabundant wildlife species. A promising new non-steroidal, non-immunological approach to contraception involves potent analogs of gonadotropin-releasing hormone (GnRH). GnRH is a molecule produced in the hypothalamus of the brain. It directs specific cells in the pituitary gland to synthesize and secrete two important reproductive hormones; follicle stimulating hormone (FSH) and luteinizing hormone (LH). These latter two hormones, known as gonadotropes, control the proper functioning of the ovaries in the female and testes in the male. Chronic treatment with continuous, high doses of GnRH agonists results in temporary suppression of pituitary responsiveness and gonadotropin secretion. Resulting decreases in plasma LH and FSH in females leads to suppression of ovulation, estrous cyclicity, and gonadal steroidogenesis (Belchetz et al.1978, Evans and Rawlings 1994). Once GnRH agonist treatments are terminated, normal pituitary function is gradually restored (Bergfeld et al. 1996). GnRH agonists have been shown to inhibit ovulation in several domestic ungulate species including sheep (McNeilly and Fraser 1987), cattle ( D'Occhio et al. 1996; D'Occhio and Aspden 1999), and horses (Montovan et al. 1990). However, studies on wild ungulates are limited (Becker and Katz 1995; Brown et al. 1999) and to our knowledge, and only one study has demonstrated their effectiveness as a contraceptive agent (Baker et al. 2002 ). GnRH agonists provide a potential biotechnology for achieving a controlled, reversible suppression of fertility in both captive and free-ranging female wild ungulates. However, their practicality as a contraceptive agent is dependent on effective inhibition of reproduction without negative behavioral or physiological side-effects, and efficacious application in free-ranging elk. In previous experiments, we determined the effectiveness of GnRH agonist (leuprolide) for controlling fertility in captive female elk and assessed the physiological and behavioral side-effects of treatment (Baker et al. 2002). Leuprolide administered as a subcutaneous, controlled release formulation was 100 % effective in preventing reproduction in elk for one breeding season. Serum LH and progesterone (P 4) concentrations were reduced to baseline levels by day 30 and remained at those levels for 190-252 days posttreatment, with a return to normal fertility the following breeding season. In addition, there were no adverse physiological side-effects and behavioral effects were minimal. However, these results were obtained under controlled conditions with captive animals of known fertility and in excellent body condition. While these results provide strong inference on the potential utility of leuprolide as a contraceptive agent, studies with wild elk are needed to evaluate whether the technique is truly feasible and practical. Thus, the goal of this study was to conduct a field experiment to examine the efficacy of leuprolide as a contraceptive agent and to contribute further understanding of its effects on reproduction and behavior in free-ranging female elk. Our specific objectives were to determine in elk : 1) the effectiveness ofleuprolide in preventing pregnancy, 2) the effects ofleuprolide on reproductive behavior, 3) the effects ofleuprolide on body condition, and 4) the reversibility ofleuprolide treatments. �60 MATERIALS AND METHODS Study Area Investigations were conducted in Rocky Mountain National Park and adjacent Estes Valley on the east slope of the Continental Divide between 2000 and 2800 m elevation. Experimental elk were selected from one of two subpopulations that historically wintered in Moraine Park/Beaver Meadows or Horseshoe Park (Bear 1989). Experimental Procedures During late summer and early fall of 2002, 34 adult female elk were immobilized by darting, from the ground, with 3.0 mg of carfentanil citrate (Wildlife Pharmaceuticals, Fort Collins, Colorado, USA) and 10-20 mg xylazine hydrochloride (Rompun; Bayer AG, Leverkusen, Germany). In order to insure that reproductive failure, if it occurred, was due to contraceptive effects rather than the effects of age or diminished body condition, we attempted to select only adult females of prime reproductive age and in moderate to excellent body condition. We hoped to accomplished this in 2 ways: 1) before immobilization, we made a visual assessment of the target animal using age (calf, yearling, adult) and relative fatness and body musculature (condition). Animal condition was classified as good, medium or poor (Riney 1960) and only medium or good condition females were selected, and 2) once the animal was immobilized we estimated age using tooth wear and replacement (Quimby and Gaab 1957), lactational status, and body condition using ultrasonography (Cook et al. 2001). Captured elk were fitted with frequency-specific transmitters on neck collars containing a plastic identification sleeve marked with a unique alpha-numeric code of 76 mm-high black characters on a colored background (white for controls; yellow for treatment)(Freddy 1993). To meet U.S. Food and Drug Administration regulations, all immobilized animals were marked to prevent human consumption. Radio collars were marked with "Do Not Consume". Once sedated, female elk received a subcutaneous, sustained release leuprolide formulation (32.5 mg) using the ATRIGEL ® drug delivery system (Atrix Laboratories, Inc., Ft. Collins, CO, USA) (Dunn et al. 1994). We reversed the effects of the immobilizing drug with 300 mg of naltrexone HcL (Wildlife Pharmaceuticals, Fort Collins, Colorado, USA). To minimize any possibility of infection from immobilization, each darted elk also received a subcutaneous injection of long-lasting penicillin. We collected blood (20 ml) from each elk as baseline information for health parameters. Blood was archived by veterinarians with the National Park Service (NPS). Measurements Reproductive rates: We assessed the effects of leuprolide treatments on reproduction in elk using 4 methods: pregnancyspecific protein B (PSPB) (Noyes et al. 1997), serum progesterone (P4) (Willard et al. 1994), rectal palpation (Greer and Hawkins 1967) and fecal progesterone metabolites (FPM) (Garrott et al. 1998). We determined pregnancy status of all treated and untreated elk during late gestation (March- April) by relocating animals using radiotelemetry and recapturing them following the immobilization procedures previously described. Once immobilized, a trained wildlife veterinarian, rectally palpated each female and determined the presence or absence of a gravid uterus. A single blood sample ( 10 ml) was collected via jugular venipuncture from each animal for PSPB(BioTracking, Moscow, Idaho, USA) and P4(Niswender 1973) analysis. At the same time, a single fecal sample was collected for fecal P4determination. �61 Females having fecal P4 levels< 0.9 Og/gm were considered nonpregnant and those. 1.0 Og/gm pregnant. Discrimination for samples with concentrations between 0.90-0.99 Og/gm was regarded as inconclusive. We will evaluate the reversibility of leuprolide treatments during March-April 2004 by using the reproductive measurements described above. Reproductive behavior: We examined the effects of leuprolide on reproductive interactions of male and female elk during 2 time periods; breeding season (defined as the period 15 September to 15 November) and postbreeding season (defined as the period 15 January to 15 March). We used focal animal sampling procedures to sample reproductive behaviors of all experimental elk (Lehner 1996). Behavioral measurements were be made by locating a breeding group containing radio collared/marked elk. Depending on the environmental conditions, topography, available cover, and elk viewing restrictions in RMNP, the observer attempted to approach the group undetected to within 150-500 m. Observations were made with the aid of binoculars and 15-60X spotting scope during morning (0500-0800) and late day (1400-1700). Time-of-day sampling periods were randomly assigned each week using a randomized block design. Each sampling period consisted of at least 2 hours of continuous observations. We combined individual behaviors into 4 general categories: male copulatory, male precopulatory, female precopulatory, and general breeding (Table 1). Our experimental unit for analyses was the individually marked female in each breeding group. Because sexual interactions were generally short duration(< 30 sec) relative to sampling interval, we recorded the number of occurrences of each event rather than length of time and calculated sexual interaction rates as behaviors per animal per hour. Body condition: Recent research has correlated measures of body condition, using ultrasonography of body fat deposits, to reproductive success in elk (Cook et al. 2001 ). Using these predictive models, we estimated the body condition of all female elk using body condition scoring and ultrasonography of fat and lean body mass. We classified each female as either excellent, very good, moderate, low, or very low reproductive candidates. We selected only those females that were judged to be, at least, in the moderate (10-15 % body fat;> 90% pregnancy rate) category. Elk that met this criteria were randomly assigned to either treatment or control groups; elk that did not, were rejected from the experiment. Additionally, we measured change in rump fat and lean body of females between fall capture and spring re-capture to evaluate the effects of leuprolide treatments on body condition. Statistical analysis: Reproductive rates. In previous experiments, a sample size of 5 treated and 5 control elk was sufficient to detect significant differences (P. 0.05) in pregnancy rates of captive animals (Baker et al. 2002). However, free-ranging elk are more elusive than their captive counterparts and treatment application and measurements of response variables less certain. Uncontrolled variables such as natural mortality, hunting mortality, low pregnancy rates, relocation success, and transmitter failure increase the need for larger sample sizes. We performed a sample size analysis with Fisher's Exact Test, using a software program (NCSS PASS 2000) to estimate the number of treated and control animals needed to detect treatment differences for PSPB, fecal progesterone metabolites, and calving rates (Table 2). For PSPB and fecal progesterone metabolites, we assumed the lowest reported pregnancy rate (63 %) for elk in RMNP (Johnson and Monello, unpublished data), 90 % recapture of radio collared females, and 100% accuracy of PSPB for pregnancy determination in elk greater than 100 days of gestation (Huang et al. 2000). For estimating �62 sample sizes for calving rates, we assumed 63 % pregnancy rates and an 85 % success in confirming presence or absence of a calf. Results of these analyses indicated that a sample size as low as 10 treated and 10 control females would be sufficient to detect a significant treatment effect using PSPB, and serum and fecal P4 values. Reproductive behavior: We tested specific reproductive behavior hypotheses that mean behavior rate was not different between treatment and control groups for both breeding and postbreeding seasons using an ANOV model with repeated measures structure. Time was treated as a within subject effect using a multivariate approach to repeated measures (Morrison 1976). To test for treatment effects, we accounted for time-of- day effects, date effects, and their interactions. PROC GENMOD (SAS Institute 1993) was used to estimate and test for differences in mean behavior rate by treatment, time- of- day, and date. Means and standard errors were estimated using least squares, and hypothesis tests were be based on type III generalized estimating equations that accounted for correlation in repeated measures. Table 1. Description of elk reproductive behaviors and associated behavior categories. Behavior category Reproductive behavior Reproductive: General Breeding Male directed behavior related to establishing, maintaining, and defending a group or harem of female wapiti Male pre-copulatory Male courtship behavior directed toward an individual female to induce or detect oestrus or ovulation (e.g. urine testing, flehmen, tongue flick, lick , smell, or rub female's body, chivy) Female pre-copulatory Female courtship behavior directed toward dominant male to arouse copulatory behavior (e.g. lick and rub male, mount, lordosis, twitch hocks) Copulatory Male behavior directed toward a receptive female in oestrus (e.g. precopulatory mounts, intromission, pelvic thrust) Non-Reproductive: Feeding Head down in vegetation Idling Bedded or standing upright and not feeding Moving Ambulating �63 Table 2. Sample size estimates and power of the test for measurements of reproductive rates in female elk in RMNP. Measurement Treatment (n) Control (n) a 1-f3 PSPB/Fecal P l. 10 10 0.05 0.9386 2. 10 20 0.05 0.9890 3. 10 120 0.05 0.9996 l. 10 20 0.05 0.8613 2. 20 20 0.05 0.9685 3. 20 25 0.05 0.9829 4. 20 30 0.05 0.9865 Calving rates RESULTS AND DISCUSSION Fall: 2002 We captured, sampled, and radio collared 34 female elk in RMNP during 24 August - 7 September, 2002 Elk were captured from 5 general locations in the RMNP : Kawuneeche Valley (7), alpine tundra areas near Trail Ridge Road (4), Hidden Valley (3), Beaver Meadows (9), and Moraine Park (11) Seventeen females were given a subcutaneous formulation containing 32.5 mg of leuprolide and seventeen a placebo formulation without leuprolide. No capture-related mortalities were observed. Estimated ages of leuprolide-treated females ranged from 1-12 years of age~(= 6.9, SE= 0.82) and 1-10 years of age ( = 6.3, SE= 0 72) for untreated elk Two yearling females were included in both groups. Yearling females were included as experimental animals because they met a priori body composition criteria. and because we wanted additional information on the effects of leuprolide in this age group. Seventy percent of treated females were determined to be lactating when captured compared to 61 % of control females. Fall body condition of leuprolide-treated and control females were similar for rump fat depth ( P = 0 .56), loin depth (P = 0.91), and body condition score (BCS) (P 0.38) (Table 3). Rump fat percent of leuprolide-treated females ranged from 8.8 - 16.3 %~( = 13.1 % , SE= 0.40) and from 10.6 - 15.9 %~( = 12.7 %, SE= 0.38) for control elk. With the exception of one animal, all females in the experiment had a rump fat percentage of greater thanlO % (> 90 % pregnancy rate). �64 Table 3. Mean fat depth, percent rump fat, body condition score, and loin depth of leuprolide-treated and control female elk sampled during Aug-Sept, 2002 and Mar-Apr, 2003, in Rocky Mountain National Park, Colorado. Leuprolide Measurements Control Mean SE Mean SE 2.13 5.43 3.53 13.10 0.18 0.12 0.12 0.40 2.00 5.41 3.38 12.73 0.11 0.10 0.11 0.38 0.37 4.84 2.36 6.90 0.04 0.08 0.12 0.04 0.72 5.00 2.48 8.20 0.12 0.11 0.13 0.49 Fall (Aug-Sept 2002): Rump fat depth (cm) Loin depth (cm) Body condition score Rump fat(%) Spring (Mar-Apr 2003): Rump fat depth (cm) Loin depth (cm) Body condition score Rump fat(%) Fall - Spring ~ Rump fat depth (cm) Loin depth (cm) ~ Body condition score ~ Rump fat (%) ~ - 1.76 - 0.59 - 1.17 -6.20 - 1.28 - 0.41 -0.90 -4.50 We observed reproductive behaviors of treated and control elk in RMNP and Estes Valley during 11 September to 27 November, 2002. We recorded a total of 144, one hour observations for 16 different radio collared female elk (8 treated; 8 control). No copulatory behaviors were observed during this period, thus there was no analysis for this category. There were no differences in reproductive behavior rates (number of behaviors/hour) for general breeding (P = 0.36), female precopulatory (P = 0.13), or male precopulatory (P = 0.70) behaviors (Fig. I). In general, control females showed somewhat higher rates of general breeding (25 % higher than treated females) and male precopulatory (9 % higher than treated females) behaviors, but none of these differences were statistically significant. In addition to reproductive behaviors, we evaluated the effects of leuprolide on the daily activity patterns of treated and control female elk. These data are currently being analyzed. �65 11m Control Leuprolide a '1.00 i ............ ··•.•·· a a ,._ 0,25 0.00 I Behavior category Figure 1. Mean(± SE) reproductive behavior rates during the breeding season for control female elk (n = 8) and females treated with a sustained release implant containing 32.5 mg leuprolide formulation (n = 8), in Rocky Mountain National Park, Colorado. Columns with different lower case letters indicate significant differences between means (P 0.05). Spring 2003 During 24 March to 30 April, 2003, we evaluated the effects ofleuprolide on pregnancy rates, body condition, and reproductive behavior of treated and control female elk. Using the capture methods previously described, we recaptured 15 out of 17 treated elk and 17 out of 17 control elk. Elk were recaptured in 3 general locations: RMNP (13), Estes Park, Colorado area (16), and Loveland, Colorado area (3). Leuprolide, administered as a sustained release formulation, prior to the breeding season, effectively prevented pregnancy in all female elk for one year. Pregnancy rates of untreated females ranged from �66 64. 7 - 78.5 %, depending on the method of determination. Fecal P4 analyses for pregnancy determination have not been completed. Body condition of experimental elk declined for all measures of body composition during fall 2003 and spring 2004 (fable 3). Changes in mean loin depth (P. 0.25) and body condition score (P. 0.08) were similar for both treated and control female elk, whereas, overwinter loss in mean percent rump fat was greater (P. 0.057) for elk treated with leuprolide. Post-breeding season reproductive behaviors and daily activity patterns of control and leuprolide-treated females are currently being analyzed. SUMMARY To date, we have completed or are in the process of completing 3 out of the 4 objectives originally stated for this investigation. First, we have evaluated the effects of leuprolide on pregnancy rates of female elk using rectal palpation, PSPB, and P4 analysis and all methods support the conclusion that leuprolide is 100% effective in preventing pregnancy for at least one breeding season. The only remaining analysis for pregnancy determination is fecal P4, which will be completed during winter 2004. Second, we have evaluated the effects of leuprolide on breeding and post-breeding reproductive behavior of elk. Although neither of these data sets have been completely analyzed, leuprolide does not appear to have deleterious effects on elk reproductive behavior or daily activity patterns. Third, we assessed the effects of leuprolide on body condition dynamics of elk. We observed only minor differences in overwinter body composition changes between treated and control elk. The only objective yet to be completed is to confirm the reversibility of leuprolide treatments. This will be accomplished during March-April 2004 by comparing pregnancy rates of treated and control female elk. LITERATURE CITED Baker, D. L., M. A. Wild, M. M. Conner, H. B. Ravivarapu, R. L. Dunn, and T. M. Nett. 2002. Effects of GnRH agonist (leuprolide) on reproduction and behavior in female wapiti (Cervus elaphus nelsoni). Reproduction (Suppl) 60:155-167. Bear, G.D. 1989. Seasonal distribution and population characteristics of elk in the Estes Valley, Colorado. Colorado Division of Wildlife R-S-65-89, Special Report 65. Becker, S. E., and L. S. Katz. 1995. Effects of gonadotropin-releasing hormone agonist on serum LH concentrations in female white-tailed deer. Small Ruminant Research 18:145-150. Belchetz, P. E., T. M. Plant, Y. Nakai, E. J. Keogh, and E. Knobil. 1978. Hypophysial response to continuous and intermittent delivery of hypothalamic gonadotropin-releasing hormone. Science 202: 631-633. Bergfeld E., M. J. D'Occhio, and J.E. Kinder. 1996. Pituitary function, ovarian follicular growth, and plasma concentrations of 17~-oestradiol and progesterone in prepubertal heifers during and after treatment with the luteinizing hormone-releasing hormone agonist deslorelin. Biology of Reproduction 54: 776-782. Brown, J. L., D. L. Schmitt, A. Bellem, L. H. Graham, and J. Lehnhardt. 1999. Hormone secretion in the Asian elephant (E/aphus maximus): Characterization of ovulatory and anovulatory LH surges. Biology of Reproduction 61:1294-1299. �67 Cole, G. F. 1971. An ecological rationale for the natural or artificial regulation of native ungulates in parks. Transactions of the North American Wildlife and Natural Resources Conference 36:417426. Cook, R. C., J. G. Cook, and D. L. Murray. 2001. Development of predictive models of nutritional condition for Rocky Mountain elk. Journal of Wildlife Management 65:973-987. D'Occhio, M. H., W. J. Aspden, and T. Whyte. 1996. Controlled reversible suppression of oestrous cycles in beef heifers and cows using agonist of luteinizing hormone - releasing hormone. Journal of Animal Science 74:218-225. _ _ __, M. J., and W. J. Aspden. 1999. Endocrine and reproductive responses of male and female cattle to agonist of gonadotrophin-releasing hormone. Journal of Reproduction and Fertility Supplement 54: 101-114 DeNicola, A. J., D. J. Kesler, and R. K. Swihart. 1997. Remotely delivered prostaglandin F- 2 alpha implants terminate pregnancy in white-tailed deer. Wildlife Society Bulletin 25:527-531. Dunn, R. L., J.P. English, D.R. Cowan, and D. P. Vanderbilt. 1994. Biodegradable in situ forming implants and methods of producing the same. U. S. patent no. 5,278,201. Evans, A. C. 0., and N. C. Rawlings. 1994. Effects of a long-acting gonadotrophin-releasing hormone agonist (Leuprolide) on ovarian follicular development in prepubertal heifer calves. Canadian Journal of Animal Science 74:649-656. Fagerstone, K. A, M. Coffey, P. Curtis, R. Dolbeer, G. Killian, L.A. Miller, and L.Wilmot. 2001 Wildlife contraception. Wildlife Society Technical Review. Proceedings of the Wildlife Society 8th Annual Conference, Reno, USA Freddy, D. J. 1993. Estimating survival rates of elk and developing techniques to estimate population size. Colorado Division of Wildlife Report, July. Garrott, P. J. White, and C. A. Vanderbilt White. 1993. Overabundance: an issue for conservation biologist? Conservation Biology 7:946-949. - - - - ~ and D. B. Siniff. 1992. Limitations of male-oriented contraception for controlling feral horse populations. Journal of Wildlife Management 56:456-464. • _ _ _ _ , S. L. Monfort, P. J. White, K. L. Mashburn, and J. G. Cook. 1998. One-sample pregnancy diagnosis in elk using fecal steroid metabolites. Journal of Wildlife Management 34: 126-131. Huang, F., D. C. Cockrell, T. R. Stephenson, J. H. Noyes, and R. G. Sasser. 2000. A serum pregnancy test with specific radioimmunoassay for moose and elk pregnancy specific protein B. Journal of Wildlife Management 64:492-499. Hess, K. H. 1993. Rocky Times in Rocky Mountain National Park. University of Colorado Press. Niwot Colorado, USA. Hobbs, N. T., D. C. Bowden, and D. L. Baker. 2000. Effects of fertility control on populations of ungulates: general, stage-structured models. Journal of Wildlife Management 64: 473-491. �68 Houston, D. B. 1971. Ecosystems of national parks. Science 172: 648-651. Jacobsen, N. K., D. A. Jessup, and D. J. Kesler. 1995. Contraception in captive black-tailed deer by remotely delivered norgestomet ballistic implants. Wildlife Society Bulletin 23:718-722. Jewell, P.A., and S. Holt. 1981. Problems in management of locally abundant wild animals. Academic Press. 32lpp. Lehner, P. N. 1996. Handbook of Ethological Methods. Second Edition. Cambridge University Press, Cambridge, UK. Kirkpatrick, J. F., and J. W. Turner, Jr. 1985. Chemical fertility control and wildlife management. Bioscience 35:485-491. - - - - , I. K. M. Lui, R. C. Fayrer Hosken, and A. T. Rutberg. 1997. Case studies in wildlife immunocontraception: wild and feral equids and white-tailed deer. Reproduction, Fertility, and Development 9: 105-110. McCullough, D.R., K. W. Jennings, N. B. Gates, B. G. Elliot, and J.E. Didonato. 1997. Overabundant deer populations in California. Wildlife Society Bulletin 25:478-483. McNeilly, A. S., and H. M. Fraser HM. 1987. Effect of gonadotrophin-releasing hormone agonist-induced suppression of LH and FSH on follicle growth and corpus luteum function in the ewe. Journal of Endocrinology 115: 273-282. Montovan, S. M., P. P. Daels, J. Rivier, G. H. Hughes, G. H. Stabenfeldt, and B.L. Lasley. 1990. The effect of potent GnRH agonist on gonadal and sexual activity in the horse. Theriogenology 33:1305-1321. Morrison, D. F. 1976. Multivariate Statistical Methods. McGraw-Hill Book Co., New York, USA. Quimby, D. C., and J.E. Gaab. 1957. Mandibular dentition as an age indicator in Rocky Mountain elk. Journal of Wildlife Management 21: 134-15 3. Riney, T. 1960. A field technique for assessing physical condition of some ungulates. Journal of Wildlife Management 24:92-94. SAS Institute. 1993. SAS/STAT user's guide. Release 6.03 edition. SAS Institute, Cary, North Carolina, USA. Seagle, S. W., and J. D. Close. 1996. Modeling white-tailed deer population control by contraception. Biological Conservation 76:87-91. Singer, F. J., L. C. Zeigenfuss, R. G. Cates, and D. T. Barnett. 1998. Elk, multiple factors, and persistence of willows in national parks. Wildlife Society Bulletin 26:419-428. White, C. A., C. E. Olmsted, and C. E. Kay. 1998. Aspen, elk, and fire in the Rocky Mountain national parks of North America. Wildlife Society Bulletin 26:449-462. �69 Willard, S. T., R. G. Sasser, J.C. Gillespie, J. T. Jaques, T. H. Welsh, Jr., and R. D. Randel. 1994. Methods for pregnancy determination and the effects of body condition on pregnancy status in Rocky Mountain elk (Cervus elaphus nelsoni). Theriogenology 42: 1095-1102. White, C. A, C. E. Olmsted, and C. E. Kay. 1998. Aspen, elk, and fire in the Rocky Mountain national parks of North America. Wildlife Society Bulletin 26:449-462. Willard, S. T., R. G. Sasser, J.C. Gillespie, J. T. Jaques, T. H. Welsh, Jr., and R. D. Randel. 1994. Methods for pregnancy determination and the effects of body condition on pregnancy status in Rocky Mountain elk (Cervus elaphus nelsoni). Theriogenology 42: 1095-1102. Zeignefuss, L., F. J. Singer, and D. C. Bowden. Vegetation and elk responses following release from artificial controls within Rocky Mountain National Park, 1968-1996. Report to the Superintendent, Rocky Mountain National Park. �Colorado Division of Wildlife Wildlife Research Report July 2003 – June 2004 JOB PROGRESS REPORT State of Project No. Work Package No. Task No. Colorado W-153-R 3002 RMNP Federal Aid Project: : : : : : Cost Center 3430 Mammals Research Elk Conservation Technical Support for Elk and Vegetation Management Environmental Impact Statement for Rocky Mountain National Park Period Covered : July 1, 2003 - June 30, 2004 Authors : D. L. Baker, M. A. Wild, M. D. Hussain, R. L. Dunn, T. M. Nett Personnel: E. Jones, J. Ritchie, A. Mitchell, X. Sha, M. Allen, J. Powers. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Practical application of fertility control technology in free-ranging wild ungulates requires remote delivery of a safe and efficacious contraceptive agent. The objective of this investigation was to evaluate the potential of a remotely delivered, sustained release, biodegradable implant formulation of leuprolide acetate, to achieve reversible suppression of ovulation and fertility in female elk (Cervus elaphus nelsoni). Fifteen, captive adult female elk were randomly allocated to one of three experimental groups. Six elk were injected intramuscularly with a dart containing the implant formulation of leuprolide, and the remaining nine elk received the same formulation without leuprolide. We measured pregnancy rates, suppression of luteinizing hormone (LH) and progesterone concentrations, and reversibility of leuprolide treatments during 1 August 2002 to 3 September 2003. The sustained release implant formulation, remotely administered by dart, resulted in decreased concentrations of LH and progesterone, temporary suppression of ovulation and steroidogenesis, and effective contraception (100%) for one breeding season. These results extend the potential for practical application of the leuprolide implant as contraceptive agent in female elk, where in the absence of such technology, wild elk must first be captured and restrained prior to treatment. 45 �JOB PROGRESS REPORT TECHNICAL SUPPORT FOR ELK AND VEGETATION MANAGEMENT ENVIRONMENTAL IMPACT STATEMENT FOR ROCKY MOUNTAIN NATIONAL PARK D. L. Baker, M. A. Wild, M. D. Hussain, R. L. Dunn, and T. M. Nett P. N. OBJECTIVE Conduct experiments with captive and free-ranging elk to evaluate fertility control as an management alternative for controlling elk populations in Rocky Mountain National Park (RMNP), Colorado. SEGMENT OBJECTIVES 1. 2. 3. Determine the effectiveness of a remotely delivered intramuscular leuprolide implant in preventing pregnancy in captive female elk. Determine the duration of effectiveness of remotely delivered leuprolide implant (if any) on luteinizing hormone (LH) and progesterone secretion in captive female elk. Determine the reversibility of remotely delivered leuprolide implant on infertility (if achieved) in captive female elk. INTRODUCTION Fundamental to practical application of contraceptives to wildlife, is a safe and effective antifertility agent that can be remotely delivered to the target species. To attain this goal, considerable research has focused on the development and testing of ballistic systems and controlled drug release formulations that can remotely administer contraceptive agents to wild ungulates (Kreeger, 1997). Contraceptive agents have been delivered via projectile dart or biodegradable implant to a variety of wild ungulate species including deer (Odocoileus spp.) (Turner et al., 1992; Jacobsen et al., 1995; DeNicola et al., 1997), elk (Cervus elaphus nannodes) (Shideler et al., 2002), wild horses (Equus caballus) (Kirkpatrick et al.,1990), burros (Equus asinus) (Turner et al., 1996), and elephants (Loxodonta africana) (Delsink et al., 2002). However, to date, no contraceptive agent that possess all of the desired attributes (Fagerstone et al., 2002) has been developed for remote delivery. The use of GnRH agonist implants to suppress short-term ovarian follicular growth and ovulation are well documented for a number of species including cattle (McLeod et al., 1991, D’Occhio et al., 1996), sheep (McNeilly and Fraser,1987), monkeys (Fraser et al., 1987), and humans (Broekmans et al., 1996). However, few studies have established the efficacy of these agents for long-term suppression of ovarian activity and contraception (Trigg et al., 2001; Baker et al., 2002, 2004; D’Occhio et al., 2002) and to our knowledge, none have previously demonstrated effective contraception by dart delivery of the implant. In previous research, we administered gonadotropin releasing hormone (GnRH) agonist leuprolide acetate by hand injection to captive female elk (Cervus elaphus nelsoni) (Baker et al., 2002), and mule deer (Odocoileus hemionus hemionus) (Baker et al. 2004), as a sustained release injectable implant, and achieved 100 % infertility for one breeding season. The implant formulation consisted of 45 % w/w 75/25 poly (DL-lactide-co-glycolide) (PLG) polymer having an intrinsic viscosity of 0.20 dL/g dissolved in N-methyl-2-pyrrolidone (NMP) and containing 6 % w/w leuprolide in the polymer solution. This formulation was designed to release the drug for a period of 3 to 4 months after subcutaneous injection (Ravivarapu et al., 2000). 46 �In these previous studies, the leuprolide formulation was demonstrated to be highly effective when delivered subcutaneously, however, it’s not known if similar effectiveness can be achieved when administered as an intramuscular (IM) injection via dart. Differences in drug pharmacokinetics and metabolism between muscle and subcutaneous tissues could affect release dynamics of the implant and possibly decrease the antifertility properties of leuprolide. Therefore, the objectives of this experiment were to determine in captive female elk (1) the effectiveness of this remotely delivered intramuscular leuprolide implant in preventing pregnancy, (2) the duration of effects (if any) on luteinizing hormone (LH) and progesterone secretion, and (3) the reversibility of infertility (if achieved). MATERIALS AND METHODS Experimental animals During 1 August 2002 to 3 September 2003, we evaluated the effects of remotely delivered leuprolide formulation on pregnancy rates, luteinizing hormone (LH), and progesterone secretion in captive female elk. Controlled experiments were conducted with 15 adult females (2-14 years of age; 220 - 275 kg BW), two intact adult male elk (3 years of age; 350-400 kg BW), and one epididymectomized adult male elk (3 years of age; 340-375 kg BW) at the Colorado Division of Wildlife’s Foothills Wildlife Research Facility in Fort Collins, Colorado, USA. Captive elk used in this experiment were permanently maintained at this facility and were trained to repeated handling, weighing, blood sampling techniques, and isolation pens. When not involved in the periodic intensive sampling procedures required by this study, elk were maintained in fenced pastures (5 ha) containing native vegetation and fed a diet consisting of ad libitum quantities of grass-alfalfa hay, grain supplement, trace mineral block, and water. In an effort to induce normal cyclic ovulatory responses and synchronize estrus, we released an epididymectomized male elk with 15 seasonally anovulatory female elk on 20 July 2002 (McComb, 1987). Four weeks later (21 August) and prior to assigning elk to experimental treatments, we assessed the reproductive status of each female by: 1) manual rectal palpation of the reproductive tract to diagnose ovarian status and identify any abnormalities, and 2) measuring the responsiveness of pituitary gonadotropes to an exogenous dose of GnRH analog. Females showing evidence of reproductive tract abnormalities or suppressed gonadotrope function were excluded from the experiment. Experimental design Fifteen female elk were randomly assigned to one of three experimental groups. Six elk (group A) were injected with a dart containing the polymeric matrix formulation of leuprolide acetate (D-Leu6GnRH-Pro9-ethylamide). Four elk (group B) were designated as pregnant controls. They received the polymer solution without leuprolide and were used to compare the effects of leuprolide formulation on pregnancy rates between treated and untreated elk. These two groups of elk were maintained together in the same pastures with two intact, adult male elk from 13 September 2002 to 10 April 2003. The remaining five elk (group C) served as non-pregnant controls and were placed in a separate pasture (2 ha) without direct contact with male elk. We compared concentrations of LH and progesterone of these females to those treated with leuprolide formulation (group A). Non-pregnant control females (group C) provided a more representative comparison to treated elk for evaluating treatment-induced hormonal responses than potentially pregnant elk, thus the need for two separate control groups. Treatments Leuprolide implant formulation. The polymer, 85/15 poly (DL-lactide-co-glycolide) (PLG) with intrinsic viscosity 0.31 dL/g (Absorbable Polymer Technologies, Pelham, Alabama, USA) and N-methyl2-pyrrolidone (NMP, International Speciality Products, Wayne, New Jersey, USA) were mixed in a ratio of 50:50 in a vial until the polymer was completely dissolved. The polymer solution was sterilized by γirradiation at a dose of approximately 25Gy (Isomedix, Morton Grove, Illinois, USA) and an appropriate amount of the sterilized polymer solution was filled into 1.2 luer-lock female syringes. For the leuprolide 47 �part of the system, calculated volume of filtered aqueous solution of leuprolide acetate (Mallinkrodt, St. Louis, Missouri, USA) was filled in 1-mL male syringe barrels (Becton-Dickenson, Franklin Lakes, New Jersey, USA) and lyophilized. This formulation was designed to deliver a 32.5 mg dose of leuprolide at a controlled rate over a 180-day therapeutic period. A similar formulation was previously shown to suppress ovulation and pregnancy for one breeding season in captive elk when delivered subcutaneously by hand-injection (Baker et al., 2002). Treatment application. On the day before treatment application (6 September 2002), experimental elk were moved from holding paddocks to individual isolation pens (5 m x 10 m), weighed (± 0.5 kg), sedated with xylazine hydrochloride (Rompun; Bayer AG, Leverkusen; 25-200 mg/animal, IM) and fitted nonsurgically with indwelling jugular catheters. The next day, and just prior to injection, separate syringes containing the polymer and the leuprolide were connected and the contents mixed with 60 back and forth mixing cycles. The resulting homogenous dispersion was drawn into the male syringe, and the formulation was transferred into single use, 1 ml, 13-mm-diameter, barb-less darts equipped with gel-collared 32-mm-long needles (Pneu-dart, Williamsport, Pennsylvania, USA). The final concentration of leuprolide was 12 % in the homogenous mixture of polymer solution and leuprolide acetate after mixing and was designed to deliver approximately 32.5 mg of leuprolide acetate to the elk. Control elk received only the polymer solution processed the same way but without leuprolide. Prior to darting, individual elk were placed in a handling chute and lightly sedated with intravenous (IV) xylazine hydrochloride (15-20 mg/animal). This dose allowed animals to remain standing in the chute and minimized excitation associated with discharge of the dart gun. All elk were remotely injected with a dart fired from a CO2-powered pistol (DanInject™ , Wildlife Pharmaceuticals, Fort Collins, Colorado, USA). In order to accurately determine the precise dose of leuprolide formulation delivered to each elk, darts were weighed before and after injection. With the exception of two animals, one dart per animal was fired from approximately 3 meters into the area of the biceps femoris muscle of the standing elk. In two animals, the dart failed to discharge or only partially injected the prescribed dose. In these cases, we re-weighed and fired additional darts until the complete dose was delivered to each animal. Once all elk had been treated, sedation was reversed with yohimbine (30 mg, IV) (Antagonil®, Wildlife Laboratories, Fort Collins, Colorado, USA) and animals were returned to individual isolation pens. Measurements 24 h LH response to leuprolide treatment. Immediately following application of treatments to groups A and group C, we determined the amount of LH released during the initial 24 h of the treatment period. Blood samples (5 ml) were collected via jugular catheters at 0, 120, 180, 240, 300, 360, 480, 600, 960, and 1440 min after drug injection. Catheters were flushed after each collection with sterile saline solution. After the last blood collection, catheters were removed and animals were returned to holding paddocks. Eight days later, two intact male elk were placed into the same pasture with these females. Duration of LH and progesterone response to leuprolide treatment. The effect of leuprolide formulation on the duration of suppression of LH and progesterone was determined by periodically conducting pituitary stimulation trials. These trials were performed prior to treatment application as an aid in the selection of animals for this experiment and periodically during 29 October 2002 to 3 September 2003 to determine pituitary responsiveness to an exogenous dose of GnRH analog (D-Ala6-GnRH-Pro9ethylamide; Sigma Chemical Company, St. Louis, Missouri, USA). Pituitary stimulation trials were conducted with elk in groups A and C elk at 50, 100, 150, 185, 215, and 361 days post-treatment. The final stimulation trial (3 September 2003) provided hormonal evidence of the reversibility of leuprolide treatment. Stimulation trials were conducted according to the 48 �following procedures: On the day of testing, elk from groups A and C were moved from 5 ha pastures to individual isolation pens, weighed, sedated (as previously described), and fitted nonsurgically with indwelling jugular catheters. A bolus dose of GnRH analog (1 g/50 kg body weight) was administered through the cannula and blood samples (5 ml) were collected at 0, 60, 120, 180, 240, 300, 360, and 480 min post-administration. After collections, blood was stored at 4 C for 24 h until serum was obtained by centrifugation (1500 RCF for 15 min). Serum for progesterone analysis was obtained from the 0 h blood sample for each animal on each of the trial days. Serum was stored at -20 C until analyzed for LH and progesterone. Following the last blood collection, catheters were removed, and elk were returned to holding pastures. Reproductive response to leuprolide treatment - The effect of leuprolide formulation on reproduction in groups A and B was determined in two ways : (1) by measuring pregnancy rates using the presence or absence of pregnancy specific protein B (PSPB) (BioTracking, Moscow, Idaho, USA) in serum collected at approximately 100 and 215 days of gestation (Huang et al., 2000), and (2) by observing the presence or absence of calves the following summer. Analyses Serum concentrations of LH were quantified by means of an ovine oLH radioimmunoassay (Niswender et al., 1969). Elk serum was demonstrated to inhibit binding of 125 I-labeled oLH to LH antiserum in a manner that paralleled the standard (NIH- oLH-S24). Similarly, when different quantities of oLH standard were added to elk serum and samples were subjected to radioimmunoassay, the values obtained were increased by the quantity of oLH added (r2 = 0.99, slope = 0.92, β1 = 0.22, P = 0.002). These data indicated that the radioimmunoassay provided a quantitative assessment of LH in elk serum. The limit of sensitivity of the LH assay was 0.02 ng /ml. Serum concentrations of progesterone were also determined by radioimmunoassay (Niswender, 1973). Sensitivity of the progesterone assay was 0.12 ng /ml. Intra-and-inter assay coefficients of variation for each of these assays were < 10 %. Hormone concentrations are reported as untransformed arithmetic means (± SE). Responsiveness of the pituitary gland to GnRH analog stimulation was determined by the total amount of LH secreted (ng /ml/ min) which was estimated by calculating the area under the LH response curve (Abramowitz and Stegun, 1968). Differences among hormone concentrations were tested using least squares ANOVA for general linear models (SAS Institute, 1997). Responses to treatment were analyzed with one-way ANOVA for a randomized complete block design with repeated measures. Treatment effects were determined using the total animal-within-treatment variances as the error term. Time was treated as a within-subject effect, using a multivariate approach to repeated measures (Morrison et al., 1976). A “protected” least significant difference test (Milliken and Johnson, 1984) was used to separate means when the overall F- test indicated significant treatment effects (P < 0.05). RESULTS Intramuscular injection of leuprolide formulation via dart, was 100 % effective in suppressing ovulation and preventing pregnancy in captive female elk for one breeding season. All leuprolide treated females (group A) tested negative and untreated controls (group B) positive for PSPB at approximately 100 and 215 days of gestation. No calves were born to treated elk, whereas the calving rate of untreated elk was 100 %. The amount of leuprolide acetate delivered to each elk ranged from 22. 6 to 38.1 mg ( = 33.1, SE = 2.4). We did not observe any unusual bleeding, swelling or trauma at the injection site nor did any of the elk show evidence of impaired mobility or post-treatment tissue necrosis or abscesses related to dart delivery of the bioimplant. Of particular interest was that the lowest individual dose delivered (22.6 mg) was equally as effective as higher doses in suppressing hormone concentrations 49 �and pregnancy, suggesting that the minimum effective dose in elk could be substantially lower than the estimated dose (32.5 mg) used in this experiment. Mean serum concentrations of LH increased (P = 0.015) in treated elk (group A) within 2 h of drug injection, peaked at 63.12 ± 10.8 ng/ml (mean ± SE) 4.3 ± 0.65 h (mean ± SE) later, then gradually declined to baseline levels by 16 h post-treatment (Fig. 1). Levels of LH in group A were greater (P = 0.032) than those of untreated controls (group C) for 2- 10 h post-treatment, after which, values decreased to baseline levels and were similar (P = 0.285) for both groups. Results of periodic GnRH challenges revealed that the leuprolide formulation reduced pituitary content of LH to basal concentrations for at least 215 days post-treatment, which was 35 days longer than the expected 180-day delivery period (Fig. 2). Concentrations of GnRH analog-induced LH secretion were lower (P = 0.022) in leuprolide - treated elk (group A) compared to non-pregnant controls (group C) at days 50, 150, 185, and 215 days after treatment. Chronic suppression of LH in treated females was followed by a return to pretreatment levels, indicative of estrus, prior to the subsequent breeding season (September 2003, Fig. 3). In contrast to leuprolide-treated elk, pituitary responsiveness of untreated elk (group C) to GnRH analog were elevated and relatively similar (P = 0.64) in magnitude during the first 185 days of the experiment, after which, these levels declined (P = 0.087), presumably with the onset of seasonal anestrus (March). Similar (P = 0.582) to treated elk, pituitary responsiveness in control elk (group C) returned to pretreatment levels in September 20003. Serum concentrations of progesterone in leuprolide - treated females (group A) followed a parallel pattern to that observed for serum LH (Fig. 3). The suppressive effects of leuprolide on corpus luteum formation and steroidogenesis was readily apparent by its effects on serum progesterone concentrations in treated elk compared to controls (group C). Progesterone levels in treated elk declined (P = 0.017) to limits of detection by 50 days post-treatment and remained at those levels for the duration of the breeding period. For untreated elk (group C), serum progesterone was more variable and consistently higher (P = 0.043) than that for treated elk at 50, 100, 150, 185, and 215 days post-treatment. As evidence of normal estrous cycles and contraceptive reversibility, progesterone concentrations in both treated and untreated elk (group C) returned to pretreatment levels (P = 0.435) at the onset of the following breeding season. DISCUSSION In the present experiment, we evaluated the effectiveness of projectile dart delivery of the GnRH agonist, leuprolide, as a potential antifertility agent in female elk. The sustained release polymeric implant formulation of leuprolide acetate, remotely delivered in a projectile dart, resulted in decreased LH and progesterone secretion, presumably suppression of ovulation and steroidogenesis, and effective contraception (100 %) without adverse effects for one breeding season. The contraceptive effects of leuprolide formulation followed a two-phase process. The first phase was characterized by an acute, transient rise in serum LH which gradually declined to basal concentrations about 16 h post-treatment. The second phase was defined by chronic inhibition of LH and progesterone secretion for the duration of the seasonal breeding period. Subsequently, normal ovarian function and fertility were re-established prior to next breeding season. We conclude from these patterns of LH and progesterone in serum that gonadotropes in female elk are down-regulated during treatment with GnRH agonist. As a consequence, long-term exposure to GnRH agonist resulted in reduction in GnRH receptors on gonadotropes (Clayton, 1989), depletion of pituitary LH and FSH content (Aspden et al., 1996), and elimination of the preovulatory LH surge (Gong et al., 1995; D’Occhio et al., 1996). These responses have been shown to result in ovulation failure and infertility which persists as long as the GnRH agonist is present in circulation at therapeutic levels (Melson et al., 1986; D’Occhio et al., 2000). 50 �Our findings here are consistent with previous observations of acute and chronic responses of sheep (Dobson, 1985), cattle (D’Occhio et al., 1989; Gong et al., 1996), horses (Montovan et al., 1990), deer (Becker and Katz, 1995), and elk (Baker et al.,2002) treated with GnRH agonist. Effective contraception in polyestrous, seasonal breeders is dependent on suppression of ovulation from the beginning of the breeding season to the onset of seasonal anestrous, a period of approximately 200 d in elk. Therefore, the timing of treatment application is an important consideration in successful contraception. Because of the acute rise in LH concentrations that occurs following GnRH agonist treatments, ovulation of growing follicles can be induced (Macmillan and Thatcher, 1991, D’Occhio and Aspden, 1999). Therefore, to ensure effective contraception in female elk, leuprolide treatments should be applied prior to the initiation of seasonal estrus. In the present study, leuprolide inhibited LH secretion and ovulation for at least 215 days which is in close agreement with previous research, in which a subcutaneous dose of leuprolide suppressed LH levels for 190-250 days (Baker et al., 2002). In other studies, implants containing GnRH agonist have been shown to suppress ovarian activity for a minimum of 150 days in mule deer and (Baker et al., 2004) and almost 400 days in cattle (D’Occhio et al., 2002). Persistent suppression of ovarian function, beyond the formulated delivery period of the implant, has been reported for a number of different species. Leuprolide suppressed LH and progesterone levels in elk in this experiment for at least 35 days longer (19 %) than the expected six month effective duration and 30 -110 days longer in deer and elk in previous studies (Baker et al., 2002, 2004). Similar observations of extended gonadotrope suppression were reported previously in cattle (Bergfeld et al., 1996; D’Occhio et al., 1996), monkeys (Fraser et al., 1987), men (Hall et al., 1999), and women (Broekmans et al., 1996). The underlying mechanism for this effect is not completely understood, but it is thought to be a associated with prolonged dysfunction of gonadotrope cells rather than direct action on the ovaries (D’Occhio et al., 2000; Aspden et al., 2003). Regardless of the mechanism involved, the extended suppression of ovarian function, as a consequence of GnRH agonist treatment, is fundamentally essential to effective contraception in deer and elk. In conclusion, intramuscular delivery of the sustained release biodegradable polymeric implant formulation of leuprolide via dart resulted in effective suppression of ovarian function and fertility in female elk for one breeding season with a return to normal reproductive function the following year. These results are particularly important for wildlife applications where, in the absence of such technology, animals must first be captured and restrained prior to treatment. LITERATURE CITED Abramowitz, M., and I. A. Stegun. 1968. Handbook of mathematical functions. Dover Publishing, Inc., New York, New York, 343 pp. Aspden, W. J., A. Rao, P. T. Scott, I. J. Clark, T. E. Trigg, J. Walsh, and M. J. D’Occhio. 1996. Direct actions of the luteinizing hormone -releasing hormone agonist, deslorelin, on anterior pituitary contents of luteinizing hormone (LH) and follicle-stimulating hormone (FSH), LH and FSH subunit messenger ribonucleic acid, and plasma concentrations of LH and FSH in castrated male cattle. Biology of Reproduction 55:386-392. __________ , A. Jackson, T. E. Trigg, and M. J. D’occhio. 2003. Pituitary expression of LHβ-and FSHβsubunit mRNA, cellular distribution of Lhβsubunit mRNA and LH and FSH synthesis during and after treatment with a gonadotrophin-releasing hormone agonist in heifers. Reproduction, Fertility and Development 15:149-156. 51 �Baker, D. L., M. A. Wild, M. M. Conner, H. B. Ravivarapu, R. L. Dunn, and T. M. Nett. 2002. Effects of GnRH agonist (leuprolide) on reproduction and behavior in female wapiti (Cervus elaphus nelsoni). Reproduction (Suppl.) 60: 155-167. ___________ , ___________ , ___________ , __________ , _____________ , _________ . 2004. Gonadotropin releasing hormone agonist: a new approach to reversible contraception in female deer. Journal of Wildlife Diseases 40: (in press). Becker, S. E., and L. S. Katz. 1995. Effects of gonadotropin-releasing hormone agonist on serum LH concentrations in female white-tailed deer. Small Ruminant Research 18:145-150. Bergfeld E. G. M., M. J. D’occhio, and J. E. Kinder. 1996. Continued desensitization of the pituitary gland in young bulls after treatment with the luteinizing hormone-releasing hormone agonist deslorelin. Biology of Reproduction 54:769-775. Broekmans, F. J., P. G. Hompes, C. B. Lambalk, E. Broeders, and J. Schoemaker. 1996. Short-term desensitization: effects of different doses of gonadotrophin-releasing hormone agonist triptorelin. Human Reproduction 11:55-60. Clayton, R. N. 1989. Gonadotropin-releasing hormone: its actions and receptors. Journal of Endocrinology 120:11-19. Delsink, A. K., J. J. van Altena, J. J. Kirkpatrick , and R. A. Fayrer-Hosken. 2002. Field application of immunocontraception in African elephants (Loxodonta africana). Reproduction (Suppl.) 60:117124. DeNicola, A. J., D. J. Kesler, and R. K. Swihart. 1997. Remotely delivered prostaglandin F2 implants terminate pregnancy in white-tailed deer. Wildlife Society Bulletin 25: 527-531. Dobson, H. 1985. Effects of chronic treatment with a GnRH agonist on oestrous behavior and on the secretion of LH and progesterone in the ewe. Theriogenology 24: 1-11. D’occhio, M. J., D. R. Gifford, C. R. Earl, T. Weatherly, and W. Von Rechengerg. 1989. Pituitary and ovarian responses of post-partum acyclic beef cows to continuous long-term GnRH and GnRH agonist treatment. Journal of Reproduction and Fertility 85:495-502. ____________, W. J. Aspden, and T. R. Whyte. 1996. Controlled, reversible suppression of estrous cycles in beef heifers and cows using agonist of gonadotropin-releasing hormone. Journal of Animal Science 74:218-225. ____________, and Aspden, W. J. 1999. Endocrine and reproductive responses of male and female cattle to agonist of gonadotrophin-releasing hormone. Journal of Reproduction and Fertility 54 (Suppl.):101-114. ____________ , G. Fordyce, T. R. Whyte, W. J. Aspden, and T. E. Trigg. 2000. Reproductive responses of cattle GnRH agonist. Animal Reproduction Science 60-61: 433-442. ______________ , _____________ , ____________ , L. A. Fitzpatrick, N. J. Cooper, W. J. Aspden, M. J. Bolam, and T. E. Trigg. 2002. Use of GnRH agonist implants for long-term suppression of fertility in extensively managed heifers and cows. Animal Reproduction Science 74: 151-162. Fagerstone, K. A., M. A. Coffey, P. D. Curtis, R. A. Dolbeer, G. J. Killian, L. A. Miller, and D. L. Wilmot. 2002. Wildlife fertility control. Wildlife Society Technical Review 02-02. The Wildlife Society, Bethesda, Maryland, USA. Fraser, H. M, J. Sandow, H. R. Seidel, W. Von Rechenberg. 1987. An implant of gonadotropin releasing hormone agonist (buserelin) which suppresses ovarian function in the macaque for 3-5 months. Acta Endocrinology 121:841-853. Gong, J. G., T. A. Bramley, C. G. Gutierrez, A. R. Peters, R. Webb. 1995. Effects of chronic treatment with gonadotrophin-releasing hormone agonist on peripheral concentrations of FSH and LH, and ovarian function in heifers. Journal of Reproduction and Fertility 105: 263-270. ___________ , B. K. Campbell, T. A. Bramley, C. G. Gutierrez, A. R. Peters, and D R. Webb. 1996. Suppression in the secretion of follicle-stimulating hormone and luteinizing hormone, and ovarian follicle development in heifers continuously infused with a gonadotropin- releasing hormone agonist. Biology of Reproduction 55: 68-74. 52 �Hall, M. C., R. J. Fritzsch, A. I. Sagalowsky, A. Ahrens, B. Petty, and C. G. Roehrborn. 1999. Prospective determination of the hormonal response after cessation of luteinizing hormonereleasing hormone agonist treatment in patients with prostate cancer. Urology 53:898-902. Huang, F., D. C. Cockrell, T. R. Stephenson, J. H. Noyes, and R. G. Sasser. 2000. A serum pregnancy test with a specific radioimmunoassay for moose and elk pregnancy specific protein B. Journal of Wildlife Management 64: 492-499. Jacobson, N. K., D. A. Jessup, and D. J. Kesler. 1995. Contraception in black-tailed deer by remotely delivered norgestomet ballistic implants. Wildlife Society Bulletin 23: 718-722. Kreeger, T. J. 1997. Overview of delivery systems for the administration of contraceptives to wildlife. In Contraception in wildlife management. T. J. Kreeger, (ed.) USDA-APHIS Technical Bulletin 1853, Washington, D. C., pp 29-48. Kirkpatrick, J. F., I. K. M. Liu, and J. W. Turner. 1990. Remotely-delivered immunocontraception in feral horses. Wildlife Society Bulletin 18: 326-330. Macmillan, K. L., and D W. W. Thatcher. 1991. Effects of gonadotrophin-releasing hormone on ovarian follicles in cattle. Biology of Reproduction 45:883-889. McComb, K. 1987. Roaring by red deer stags advances the date of oestrus in hinds. Nature 330:648-649. McCleod, B. J., S. E. Dodson, A. R. Peters, and G. E. Lamming. 1991. Effects of a GnRH agonist (Buserelin) on LH secretion in post-partum beef cows. Animal Reproduction Science 24:1-11. McNeilly, A. S., and H. M. Fraser. 1987. Effect of gonadotrophin-releasing hormone agonist-induced suppression of LH and FSH on follicle growth and corpus luteum function in the ewe. Journal of Endocrinology 115:273-282. Melson, B. E., J. L. Brown, H. M. Schoeneman, G. K. Tarnavsky, and J. J. Reeves. 1986. Elevation of serum testosterone during chronic LHRH agonist treatment in the bull. Journal of Animal Science 62:199-207. Milliken, G. A., and D. E. Johnson. 1984. Analysis of messy data. Volume 1.Designed experiments. Lifetime Learning Publications, Belmont, California, 399 pp. Montovan, S. M., P. P. Daels, J. River, J. P. Hughes, G. H. Stabenfeldt, and B. L. Lasley. 1990. The effect of potent GnRH agonist on gonadal and sexual activity in the horse. Theriogenology 33:1305-1321. Morrison, J. A., C. E. Trainer, and D P. L. Wright. 1976. Multivariate statistical methods. McGraw-Hill, New York, New York, 432 pp. Niswender, G. D, Reichert, L. E., JR., A. R. Midgley, and A.V. Nalbandov. 1969. Radioimmunoassay for bovine and ovine luteinizing hormone. Endocrinology 84:1166-1173. ___________ . 1973. Influence of the site of conjugation on the specificity of antibodies in progesterone. Steriods 22: 413-424. Ravivarapu, H. B., K. L. Moyer, and D. R. Dunn. 2000. Sustained activity and release of leuprolide acetate from an in situ forming polymeric implant. American Association of Pharmaceutical Scientist 1:1-12. SAS Institute. 1997. SAS/STAT ® user’s guide 6.03 edition. SAS Institute Incorporated, Cary, North Carolina, USA. Shideler, S. E., M. A. Stoops, N. A. Gee, J. A. Howell, and D B. L. Lasley. 2002. Use of porcine zona pellucida (PZP) vaccine as a contraceptive agent in free-ranging tule elk (Cervus elaphus nannodes). Reproduction (Suppl) 60: 169-176. Trigg, G. G, T. E., P. J. Wright, A. F. Armout, P. E. Williamson, A. Jundaidi, G. B. Martine, A. G. Doyle, and D. J. Walsh. 2001. Use of GnRH analogue implant to produce reversible long-term suppression of reproductive function in male and female domestic dogs. Journal of Reproduction and Fertility 57:255-261. Turner, J. W., JR., I. K. M. Liu, and J. F. Kirkpatrick. 1992. Remotely delivered immunocontraception in white-tailed deer. Journal of Wildlife Management 56: 154-157. ___________ , ____________ , ____________ . 1996. Remotely delivered 53 �immunocontraception in free-roaming feral burros. Journal of Reproduction and Fertility 107: 31-35. Prepared by ___________________________ Dan L. Baker, Wildlife Researcher Figure 1. Twenty-four hour serum LH concentrations (ng/ml, mean ± SE) for untreated female elk (", n = 5) and female elk (!, n = 6) treated with a 180-day sustained release implant formulation, containing approximately 32.5 mg of leuprolide acetate, remotely delivered via projectile dart. Fig. 1 - Baker et al. - 80 i I 70 50 - 60 - -E 0) C: :r: _J C: (IJ Q) T 40 30 ~ 20 10 0 - l 0 J 1 Treatmert Leuprolide dart (Group A) -e-- Control dart (Group C) l T "" .L _j_ 120 180 240 300 360 Time (min) 54 480 600 960 1444 �Figure 2. Total serum LH concentrations (ng/ml/min, mean ± SE) for GnRH analog-induced release of LH for untreated female elk (", n = 5), and female elk (!, n = 6) treated with a 180 -day sustained release implant formulation, containing approximately 32.5 mg of leuprolide acetate, remotely delivered via projectile dart. Different lower case letters indicate significant differences between means (P 0.05). Fig. 2 - Baker et al. 100 - - - e- - 80 60 I _J C 40 (0 Q) ~ 20 a b b b b b 50 100 150 185 215 0 Pretreatment 361 Time (days) Figure 3. Serum profiles of mean progesterone concentrations (ng/ml, mean ± SE) for untreated female elk (", n = 5) and female elk (!, n = 6) treated with a 180-day sustained release implant formulation, containing approximately 32.5 mg of leuprolide acetate, remotely delivered via projectile dart. Different lower case letters indicate significant differences between means. (P 0.05). Fig. 3 • Baker et al. 1.00 Leuprolide dart (Group A) --- e- - a ...-.. -E 0) C e a a -S Q) 0.60 0) 0.40 C a a l J -- f'',,L_ a ' a (0 Q) ~ i ----- ' ' '" 1 ' ---e- Q) 1n Q) e c.. Control dart (Group C) 0.80 0.20 b b b 1 b b a 50 100 150 185 215 361 000 Pretreatment Time (days) 55 � https://cpw.cvlcollections.org/files/original/f7ad38888bf9552ad312a2744148ccda.pdf 8450b161cafd0b4122e4462f6d54451f PDF Text Text Colorado Division of Wildlife Wildlife Research Report July 2003 – June 2004 JOB PROGRESS REPORT State of Colorado Project No. 1 Work Package No. 3003 Task No. 1 : : : : Federal Aid Project: : N/A Cost Center 3430 Mammals Research Predatory Mammals Conservation Colorado Puma Research & Management Program Period Covered: July 1, 2003― June 30, 2004 Author: K. A. Logan Personnel: J.Apker, J. Kindler, Colorado Division of Wildlife; L. Mundy-Four Corners Houndsmen’s Association; and Safari Club International Foundation. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT The Colorado Puma Research and Management Program started in July 2003. The goal is to improve the scientific foundation of puma management by the Colorado Division of Wildlife. The program was developed with inputs of Division researchers, managers, and biologists, and Colorado citizens interested in wildlife management, hunting, and the environment. Puma population research is scheduled to begin in November 2004. Associated projects to improve puma management in Colorado, also initiated this year include: the Colorado puma data map, prospective work for Front Range pumahuman interaction research, puma technical workshops, and Data Analysis Unit management plans and puma-human conflict guidelines. 61 �JOB PROGRESS REPORT COLORADO PUMA RESEARCH AND MANAGEMENT PROGRAM Kenneth A. Logan INTRODUCTION The Colorado Puma Research and Management Program started in July 2003. The goal is to improve the scientific foundation of puma management in Colorado. A Prospectus was developed with inputs of Division researchers, managers, and biologists, and Colorado citizens interested in wildlife management, hunting, and the environment. The major part of the program is the puma research project in the prospectus, scheduled to begin in November 2004. The initial design of the research will be clarified in a study plan in September 2004 which will pertain to the puma population research on the Uncompahgre Plateau study area. Other work associated with the development of the research program included: visiting with affected publics (private landowners, ranchers, hunters, guides and outfitters) and agency cooperators, surveying potential study areas, and two public meetings for information on the proposed puma research. Associated projects to improve puma management in Colorado included: the Colorado puma data map, puma workshops, technical advice on puma Data Analysis Unit management plans, and puma-human conflict guidelines. COLORADO PUMA DATA MAP The objective of this project is to map and quantify puma data that exists in records of the Colorado Division of Wildlife. This is the first step for Division staff to examine historical and current situations regarding puma management actions and puma mortality patterns state-wide and within management units. The map is intended to be an evolving instrument that allows comparisons with puma data gathered in the future to examine potential effects of changing puma management prescriptions, habitat, ungulate populations, and human developments. Interpretations of the map could be clarified from information on puma populations, movement patterns, habitat use, habitat characteristics, pumaungulate interactions, and hunter access to occupied puma habitat. Reliable interpretations of such maps would be useful to managers. Number and distribution of puma mortality locations and absence of mortality locations may indicate relative puma abundance or hunting pressure. High mortality areas, influenced principally by sport-hunting pressure, may indicate potential areas of high puma densities, puma population sinks (defined as areas where the average population growth rate is negative), areas of facilitated puma hunting conditions (e.g., high road density, consistent snow coverage), and liberal puma harvest objectives. Low puma mortality areas and blank areas on the map may indicate potential puma habitat with puma source populations (defined as areas where mean population growth rate is positive, and which serve as net exporters of dispersing animals), areas where few if any puma live, or areas with low hunter access or good hunting conditions. RESULTS Desired products are maps and associated tabulated data on geographical distribution and intensity of puma mortality. Puma mortality data (including sport-harvest, depredation control, public safety management, and accidental deaths) recorded by the Division on mandatory check forms since 1997 were mapped, state-wide and by Data Analysis Units and Game Management Units, and stratified by year and puma sex and age class (e.g., adult, subadult, cub). This entails about 2,423 data points statewide, from 1997―2002 (2003 data have not been entered, yet). Of that, over 90% of the mortality locations are due to sport-harvest. The remainder is due to depredation or public safety control kills, roadkills, and other recorded deaths. Maps are currently in a preliminary development stage. Mortality 62 �locations of puma will be buffered by average puma home range sizes for adult puma in western North America (195 km2 for females; 357 km2 for males) and overlaid by mule deer, elk, and bighorn sheep winter ranges. In addition, the identity of DAUs with the management objective of a stable or increasing puma population and DAUs with the management objective of a suppressed population also need to be mapped for managers to consider the number, distribution and effects of potential source and sink populations. Other map overlays that may facilitate interpretation of the puma data include: road distribution (i.e., paved, all-weather, dirt), vegetation cover types, elevation, and human developments and density. Puma mortality characteristics (i.e., location, density) might be modeled by using an analytical approach that uses habitat and biological features (e.g., ungulate distribution and relative density, elevation, roads, vegetation, terrain ruggedness) as variables. Another approach might be to distribute puma mortality maps to Division field personnel and to knowledgeable puma hunters to record their explanations about geographical puma distributions, relative densities, mortality patterns, and effects of habitat characteristics (e.g., landownership, snow conditions, access). COLORADO PUMA-HUMAN INTERACTIONS RESULTS Meetings involving Division staff, and individuals from the U.S. Geological Survey Ft. Collins Research Center and Colorado State University were held to discuss potentials for puma-human interaction research on the Colorado Front Range. Meetings were held in conjunction with field trips to explore potential study areas west Ft. Collins on October 6, 2003; west of Boulder on January 28 and February 6, 2004; and west of Colorado Springs on February 5, 2004. Division staff from the southeast and northeast expressed a great deal of interest in developing puma-human interaction of research in the next 1―2 years, and discussions indicated the need to develop a reliable funding base and connections with potential cooperators. The Division of Wildlife and Four Corners Houndsmen’s Association co-hosted four workshops in 2004 to inform hunters and other interested citizens about puma in Colorado. Workshops were held in Grand Junction (July 19), Alamosa (July 17), Denver (July 22), and Canon City (August 14). In all, the workshops were attended by about 60 people. Workshop agenda topics included: puma population characteristics, vital rates, reproductive biology, behavior, prey selection, female and cub vulnerability to hunting, gender identification in the field, aging techniques, Colorado puma data map, puma management, Data Analysis Unit plans, quota setting process, proposed puma research, and puma-human conflict management. A PowerPoint presentation was developed as the main source of information on these topics. Some attendees suggested that such workshops should be held periodically for puma hunters and other people interested in puma. Associated with this effort to bring reliable information on puma to hunters, we also produced printed guidelines for sexing puma in the wild. This information is available on the Division’s webpage: http://wildlife.state.co.us/hunt/BigGame/pdf/MtLionGender.pdf and in APPENDIX I of this progress report. Drafts of guidelines developed by Division managers for addressing incidents when puma conflict with people were discussed and reviewed and a final draft was created for review by Division staff: Draft—Human Mountain-Lion Incidents. 63 �PUMA RESEARCH AND MANAGEMENT PROSPECTUS PROBLEM STATEMENT Division of Wildlife managers need reliable information on puma biology and ecology in Colorado to develop sound management strategies that address diverse public values and the Division objective of actively managing puma while “achieving healthy, self-sustaining populations”(Colorado Div. of Wildlife 2002-2007 Strategic Plan:9). Although 4 puma research efforts have been made in Colorado since the early 1970s and puma harvest data is compiled annually, reliable information on certain aspects of puma biology and ecology, and management tools that may guide managers toward effective puma management is lacking. Members of the Division’s Mammals Research staff met with Division wildlife managers and biologists from the Northwest, Southwest, Southeast, and Northeast Regions regarding puma management issues and the resultant research needs. In addition, we consulted with other agencies, organizations, and interested publics either directly or through other Division employees. In general, Division staff in western Colorado conveyed concern about puma population dynamics, especially as they relate to their abilities to manage puma populations through regulated sport-hunting. Secondarily, (perhaps because of results from recent research findings in western Colorado), they expressed interest in puma-prey interactions. Division managers on the Rocky Mountain’s Front Range placed greater emphasis on puma-human interactions. Staff in both eastern and western Colorado cited information needs regarding effects of puma harvest, puma population monitoring methods, and identifying puma habitat and landscape linkages. Specific management needs and lines of inquiry identified by Division staff and public stakeholders are categorized as follows: Improve our ability to manage puma hunting with enhanced scientific bases, strategies, and tools ● Puma population characteristics (i.e., density, sex and age structure). ● Puma population dynamics and vital rates (i.e., birth rates, survival rates, emigration rates, immigration rates). ● Methods and models for assessing and tracking changes in puma populations. ● Relative vulnerability of puma sex and age classes to hunter harvest. Improve our understanding of puma habitat needs and interrelationships of puma management units ● Puma habitat use, movements, and use of landscape linkages. ● Puma recruitment patterns (i.e., progeny, immigration, emigration). ● Models for identifying puma habitat and landscape linkages. Improve our understanding of the puma’s role in the ecology of other species ● Relationships of puma to mule deer, elk, and other natural prey. ● Relationships of puma to species of special concern, e.g., desert bighorn sheep. Improve our understanding of puma-human interactions and abilities to manage them ● Behavior of puma in relation to people and human facilities. ● Puma predation on domestic animals. ● Effects of translocating nuisance puma. ● Effects of aversive conditioning on puma. Past Puma Research in Colorado Data from past puma research in Colorado that address the topics above are limited. Currier et al. (1977:8) studied 29 captured puma on 2 tracts― one 900 km2, and one 1,950 km2― in Fremont and Custer Counties during 1974―1977. The puma population under study was subject to sport-hunting. Hunters killed a total of 31 puma in 3 winters. Non-systematic puma track counts were used to estimate 64 �the minimum puma population at 11 on the 900 km2 tract (density = 1.2/100 km2) and 25―28 on the 1,950 km2 tract (density = 1.3―1.4/100 km2. The Petersen mark-recapture method was used to estimate 95 puma (95% CI = 35, 155; density = 4.8/100 km2) during 1977―1978; however, researchers probably did not meet 4 of the 6 assumptions needed for valid estimates of puma numbers (Anderson 1983:61, 63). Vital rate data were limited to mean litter size of 2.1 (range = 1―5, n = 14). An effort to estimate puma population density in Game Management Units (GMU) 33 (Garfield and Rio Blanco Counties) and 40 (Mesa County) was made during 1980―1983 (Brent 1981, 1982, 1983). A total of 38 puma were captured: 21 were marked and released; 8 were released unmarked; 9 were killed for livestock depredation control (8) or during handling (1). Twelve puma were captured in GMU 33 and 26 were captured in GMU 40. Crude adult puma density estimates for GMU 33 ranged from 2.7―3.1 puma/100 km2. GMU 40 crude adult puma density estimates ranged from 1.2―3.7 puma/100 km2. Anderson et al. (1992) studied 57 captured puma on 3,426 km2 of the eastern slope of Uncompahgre Plateau in Mesa, Delta, Montrose, and Ouray counties during 1981―1988. Puma density was estimated only for 1987; the minimum density (mean ± SE) of residents was 1.1 ± 0.15 pumas/100 km2. Male to female sex ratios for 26 captured puma 1―12 months old was 1:1; for 19 captured puma ≥24 months old, it was 1:1.4. Age structure in that sample was 66.7% <24 months old and 33.3% ≥24 months old (the class most likely comprised of breeding adults). Vital rates included, mean (± SD) litter size of 2.4 ± 0.80 (n = 17), birth interval of 12 months (n = 2 intervals for 1 female), estimated annual survival rate for 42 puma of both sexes of 88.0 % (90% CI = 83.1, 91.4). Humans caused 18 of 21 deaths in radio-collared puma even though the study population was supposed to be protected from human offtake. Anderson et al. (1992) examined aerial locations of 7 radio-collared puma and subjective estimates of relative deer and elk density categories and could not identify consistent relationships probably because of the small non-random sample of puma, the subjective nature of the ungulate density categories, and other non-quantified factors. Mean annual home range sizes ranged from 436―732 km2 for 3 males and 190―463 km2 for 7 females. All of 9 radio-collared subadult male puma dispersed from natal areas. Two of 6 radio-collared subadult females did not disperse. Means and extremes of dispersal distances were 86.2 km (23―151) for 8 males that were 10―13 months old and 37.0 km (17―54) for 4 females that were 11―31 months old. Data on puma―human interactions were from 17 responses to 40 questionnaires submitted to residents in the housing development on the southeastern extreme of the study area. Seven of 17 respondents reported 25 puma sightings during about 260 months of residence. Ten respondents did not observe puma in about 476 months of residence. Koloski (2002) studied 19 captured puma on the 2,758 km2 Southern Ute Indian Reservation in La Plata, Archuleta, and Montezuma Counties during 1999―2001. The puma population was not subject to sport-hunting at the time. Transect intercept probability sampling was used in 2001 to estimate the puma population at 55 (90% CI = 9.0, 114.4) and a density of independent puma at 2.7/100 km2. Male to female ratio of the captured sample of 14 independent puma was 1:2.8. Of 16 captured pumas that were aged, 31% were <24 months old and 69% ≥24 months old. Vital rates included: litter size (mean ± SD) of 2.5 ± 0.58 (n = 4), birth interval of 16 months (n = 1), annual survival rate for radio-collared males (mean ± SD) of 0.89 ± 0.19 (n = 3) and radio-collared females of 0.72 ± 0.19 (n = 8); earliest age for female reproduction at 2―3 years; annual reproductive rate for resident females (mean ± SD) of 42% ± 12%. Mean home range sizes were 252.4 km2 for 3 radio-collared males and 182.4 km2 for 8 radio-collared females. Road density within puma home ranges and core areas was lower than that on the landscape where pumas occurred (P ≤ 0.002). Distance from puma locations to nearest roads was lower than distance from random points to nearest roads (P = 0.002). 65 �Current Puma Research in Colorado Presently, researchers with the Colorado Division of Wildlife (Mike Miller, Ph.D., DVM) and Colorado State University (Caroline Krumm, Graduate Student and Dr. N. Thompson Hobbs, Advisor) are conducting puma research in Larimer County. The research goal is to test for selective puma predation on mule deer infected with chronic wasting disease (CWD) by comparing CWD prevalence in puma-killed deer to prevalence in harvested deer. The research protocol calls for 6 or more puma fitted with global positioning system (GPS) collars. RESEARCHABLE OBJECTIVES The management issues listed previously in the PROBLEM STATEMENT may be translated into a number of researchable objectives, requiring descriptive studies and field experiments (Fig 1). Our goal is to provide managers with reliable information on puma biology and ecology and to develop and test tools for their efforts to adaptively manage puma in Colorado to maintain healthy, self-sustaining populations. Researchable objectives address managers’ main needs. We propose that the Division begin to address objectives that focus on puma population dynamics, effects of harvest, and estimating puma population abundance with an intensive puma population study on the West Slope. Those objectives include: 1. Describe and quantify puma population characteristics, including: density, sex and age structure. 2. Describe and quantify puma population vital rates, including: birth rates, age or stage-specific survival rates, emigration rates, immigration rates. 3. Describe and quantify agent-specific mortality rates and vulnerability of different classes of puma to hunter harvest and quantify agent-specific mortality rates. 4. Develop and test puma population models using metrics from objectives 1―3. 5. Develop and test indices to puma abundance calibrated on an estimated puma population (i.e., puma track counts, catch per unit effort, DNA genotyping). In addition, other objectives could be partially addressed during the intensive puma population research effort (i.e., objectives 1―5). Those include: 6. Describe and quantify relationships of puma to people and human facilities on the study area. 7. Describe and quantify puma use of habitats and landscape linkages. 8. Describe and quantify relationships of puma to mule deer, elk, and other prey. 9. Describe and quantify responses of puma to aversive conditioning. 10. Describe and quantify behavior and survival of translocated puma. Data collection for primary objectives 1―5 will often have applicability to objectives 6―10. For example, GPS-collared puma will enable us to quantify puma predation rates on ungulate prey, puma use of habitats and landscape linkages, puma-human interactions, and behavior and survival of translocated puma (if puma are removed from a study area as part of an experimental manipulation). However, we cautioned that such opportunistic data gathering likely will not yield the power or confidence levels of studies specifically designed to meet those objectives. Yet, such efforts could function as pilot studies. Additional research efforts can be phased in later in the puma research program. And some, (e.g., puma relationships to people, puma use of habitats and landscape linkages) can be conducted in different areas of Colorado. 66 �GOAL: Strategies, Information, & Tools for Managing Healthy, Self-sustaining Puma Populations in Colorado Puma Population Effects of Harvest & Other Mortality Movements & Corridors Population Dynamics: Density, Sex & Age, Vital Rates, Growth Rates Vulnerability to Harvest Puma Habitat Human Development Habitat Use PumaHuman Relationships Effects of Predation Map Prey Distribution Models for Habitat Indices of Abundance for Monitoring Deer, Elk, Other Natural Prey & Species of Concern Domestic Animals Effects of Evasive Conditioning Effects of Translocation Models for Populations Puma Prey Map Habitat Model PumaPrey Relationships Fig. 1. Conceptual model of the Colorado Puma Research & Management Program. 67 �TESTING ASSUMPTIONS AND HYPOTHESES Hypotheses associated with the main objectives can be structured to test assumptions, information, and methods that may guide puma management in Colorado. 1. Lacking Colorado-specific information, managers might assume that puma population densities in Colorado are within the range of those quantified in other populations studied in Wyoming (Logan et al. 1986), Idaho (Seidensticker et al. 1973), Alberta (Ross and Jalkotzy 1992), and New Mexico (Logan and Sweanor 2001). The Division has used density ranges of 2.0―4.6 puma/100 km2 to extrapolate to Data Analysis Units to estimate a range of 3,000―7,000 puma in Colorado and to guide the quota-setting process. Likewise, managers may assume that the population sex and age structure is similar to puma populations described in the above-mentioned studies. Using capture, mark-recapture (or resight) techniques, a descriptive study will test H1a: The puma population density on the study area will vary within the range of 2.0―4.6 puma/100 km2 and will exhibit a similar sex and age structure to puma populations in Wyoming, Idaho, Alberta, and New Mexico. Yet, an experimental study that allows puma population growth to approach carrying capacity in high quality puma habitat can test if a puma population in Colorado might exceed published density estimates. H1b: Puma population density in high quality puma habitat in Colorado exceeds the high range of 4.6 puma/100 km2. 2. Background material that guides puma management in Colorado assumes a moderate rate of growth of 15% for the adult puma population. Theoretically, consideration of management changes would occur if hunter kill exceeds 15% of the low end adult puma population estimate. A field experiment, involving an increase population growth phase, is required to test H2: The estimated average annual adult puma population growth rate in high quality puma habitat in Colorado (during an increase phase) will match or exceed the hypothetical r = 0.15. 3. The same background material assumes “that when female” puma “comprise greater than 50% of the hunter harvest it is an indicator that hunting may be acting to suppress the population.” An experimental study with a decline puma population growth phase will test H3: The population of harvest-age puma (i.e., adults, subadults) will decline only when 50% or more of the harvest is comprised of harvest-age female puma (i.e., independent subadults ≈≥12―24 mo. old and adults ≥24 mo. old). 4. Colorado puma Data Analysis Units (DAUs) or other management units may behave as a demographic source-sink metapopulation structure where the puma population of a region is comprised of subpopulations each of which may have dynamics that are not necessarily correlated with other subpopulations. Source-sink metapopulation dynamics function as a result of variable puma habitat quality and management practices (e.g., prey population dynamics, harvest pressure). Sources are increasing or stable populations where recruitment via local reproduction and immigration equal or outpace mortality. These source populations produce emigrating progeny that immigrate into other subpopulations, augmenting them numerically and genetically. Comparatively, sink populations are those where mortality exceeds recruitment, and puma numbers are declining or suppressed to a relatively low density. Sink populations contribute few emigrating progeny as potential recruits for other subpopulations. Sink populations are augmented by immigrants from source populations (Sweanor et al. 2000, Logan and Sweanor 2001). This project will examine H4: The study population will exhibit characteristics of a subpopulation in a demographic source-sink metapopulation structure. The following predictions must come true to support this hypothesis. 68 �a. The majority (i.e., >50%) of male recruits in the adult segment of the population on the study area will be immigrants (Logan and Sweanor 2001). Immigrants will not be offspring of puma on the study area as determined by genetic parentage analysis. b. Up to one-third of female recruits will be immigrants (Logan and Sweanor 2001). c. Male and female immigrant recruits will produce viable young. d. The majority of male progeny from the study area will emigrate (Logan and Sweanor 2001). e. About one-third of female progeny from the study area will emigrate (Logan and Sweanor 2001). f. Movements of male and female emigrants will be large enough to carry them to other Data Analysis Units with differing management objectives (i.e., population reduction, population stability). g. Male and female emigrants will establish adult home ranges in other puma habitats in Colorado. H4A: Recruits born in the local population are the largest contributor (i.e., >80%) to the maintenance of the puma population on the study area. 5. In southern Utah, Van Sickle and Lindzey (1992) found a positive relationship (r2 = 0.73) between the number of radio-collared puma known to have home ranges overlapping dirt roads (response variable) and track-finding frequency (explanatory variable). Similarly, researchers in Montana are finding a positive relationship between the number of puma home ranges overlapping search routes and puma track density (DeSimone et al. 2002). This relationship should reflect changes in puma numbers on the survey area, and thus may be useful as an index to relative abundance. A field experiment requiring both increase and decrease puma population growth phases is required to test H5a: Puma track-finding frequency (response variable) is positively correlated to number of puma with home ranges overlapping snow-covered search routes (explanatory variable). H5b: Puma track-finding frequency (response variable) is positively correlated to the density of independent puma (explanatory variable). 6. Theoretically, the amount of effort (i.e., hunting days) that hunters spend in pursuit of finding harvest-age puma (i.e., adults and subadults) should be proportionate to the abundance of puma (Lancia et al. 1996). A relationship should exist between changes in catch (or encounter)-perunit-effort and population changes. A field experiment, involving manipulation of the study population (i.e., increase phase, decline phase), is required to test H6a: Catch-per-unit effort of the research team in the increase phase, teams in the capture-recapture occasions during the increase and decline phases, and puma hunters during the decline phase will reflect the trend in the puma population. There will be an inverse (i.e., negative) correlation between the mean number of days per capture (response variable) and the number of independent puma in the population (explanatory variable) during the increase phase and the decline phase. H6b: During the increase phase, there will be a positive correlation between the mean number of days per capture of unmarked puma (response variable) and the number of marked puma “removed” from the unmarked population per year (explanatory variable). In the decline phase, there will be an inverse correlation between the mean number of days per capture (response variable) and the number of puma killed by hunters per year (explanatory variable). 7. Relative vulnerability of puma to hunters is limited to information from 2 studies on the same area in southern Utah (Van Dyke et al. 1986, Barnhurst 1986). Van Dyke et al. (1986) quantified effort to locate 4 classes of puma by looking for their tracks on dirt roads, a method that hunters use to find puma. He found that cubs and adult females required the least effort, followed by independent subadults and adult males. In contrast, Barnhurst (1986) assessed vulnerability based on the relative road crossing frequencies of radio-collared puma in each of 7 classes that were relocated once per week. He found that the most vulnerable puma were subadult males, followed 69 �by adult resident males, subadult females, and adult females (in 4 classes― females with 0 cubs, females with cubs 0-6 mo. old, females with cubs 7-12 mo. old, females with cubs 13-18 mo. old). Mothers with 0―6 month-old cubs had the lowest road-crossing frequency of all classes. Cubs in this age class are most vulnerable to death if their mothers die (K. Logan, unpublished data). A descriptive study will test H7: Relative vulnerability of GPS-collared puma on the study area, based on road-crossing frequency per day, will reflect the results of Van Dyke et al. (1986). H7A: Relative vulnerability of GPS-collared puma on the study area, based on road-crossing frequency per day, will reflect results of Barnhurst (1986). 8. Studies on the effectiveness of puma translocation, and the behavior, survival, and agent-specific mortality of translocated puma in western North America are limited to 2 studies. Ruth et al. (1998) reported on 14 puma translocated from 338―510 km in New Mexico, and Ross and Jalkotzy (1995) reported on 3 puma that were translocated 51―94 km in Alberta. The New Mexico research found that translocation was most successful for puma that were 12―27 months old, the age at which puma naturally attempt to disperse and search for a home range or establish a home range if they are philopatric. Older adult puma attempt to move back to their original home ranges. They found that mortality rates for translocated puma were significantly higher than mortality rates of non-translocated puma in a reference population. If translocation is used to experimentally reduce the population, this research would test H8: Translocation of puma will exhibit similar characteristics to the New Mexico results. For this hypothesis to be supported, the following predictions must be true. a. Mortality rates of translocated puma will be significantly higher than mortality rates of non-translocated puma. b. Independent puma 12 to about 30 months old will establish home ranges in or near release areas and have relatively greater survival rates than older adult translocated puma. c. Adult puma about 3 years old and older will tend to move back toward their original home ranges. DESIRED OUTCOMES AND MANAGEMENT APPLICATIONS 1. Quantification of variations in puma population density, sex and age structure, growth rates, vital rates, and an understanding of factors affecting them will aid adaptive puma management by yielding population model(s) useful for estimates of puma population abundance and trends, evaluation of management alternatives, and effects of management prescriptions. 2. Indices to puma abundance or trends of known reliability will allow managers to “ground truth” modeled populations and estimate effects of management prescriptions designed to achieve specified puma population objectives. 3. Testing assumptions about puma populations, currently used by Division managers, will help those managers adapt puma management based on Colorado-specific estimated characteristics and dynamics of puma populations. 4. An understanding of relative vulnerability of the various puma sex and age classes to harvest could enable managers to better structure harvest data collection and interpretation, and to develop novel prescriptions to meet management objectives. 5. Functional relationships between population vital rates and population density could be examined. Puma life history traits in Colorado may be used to test generalized hypotheses regarding puma life history strategies in the literature, and inform managers to structure successful management strategies. 70 �6. Determining whether or not the puma population in Colorado has a source-sink demographic sourcesink structure is important to evaluating the current Game Management Unit and Data Analysis Unit structure of puma management and potential effects of the juxtaposition of puma sub-populations in Colorado managed for stability, suppression, or that may function as refugia. 7. Knowledge of the relationships of puma to deer, elk, and species of special concern would allow managers to realistically consider potential effects of puma predation on those prey in the development of management strategies and policy. In addition, such information would enable the testing of scientific hypotheses on relationships of puma to their prey. 8. In the study areas currently being contemplated, some puma home ranges will probably contain human habitations and other facilities. GPS-collared and VHF-collared puma will generate quantitative information on puma behavior in relation to human activity and assist managers to better inform people about ways of reducing potential conflicts between people and puma, and to structure puma conflict policy. 9. Habitat use data gathered during the course of this research could be used to quantify puma habitat characteristics on the study area, as well as habitats and landscape linkages used by dispersing puma. Such information could be used to structure more extensive investigations of puma habitat that contribute to habitat modeling efforts that may help identify puma habitat in Colorado. This would allow a more realistic conceptual inventory of puma habitat in the state. 10. This information could be disseminated to public stakeholders interested in pumas in Colorado, and thus contribute to informed public participation in puma management. STUDY AREAS Three potential study areas were evaluated and are under consideration: the Plateau Creek-toSouth Canyon area (in Garfield, Mesa, Gunnison, and Pitkin counties), the lower Dolores River-toDisappointment Creek area (in Dolores and Montezuma counties), the South Uncompahgre Plateau (in Mesa, Montrose, Ouray, and San Miguel Counties) (Table 1). These areas appear to have attributes conducive to an intensive puma research effort, including sufficient area (400―500 mi.2 = 1,036―1,295 km2) of puma habitat and suitable road access. Preferably, there should be a 300―400 mi.2 buffer zone around the study area to reduce the effect of puma harvest. 71 �Table 1. Potential puma study area locations and characteristics. Location Dolores River-toDisappointment Creek (GMUs 71 & 711) Area ~840 mi.2 = 2,176 km2 Junction of Interstate 70 & State Route 6 northeast to South Canyon (GMU 42) ~400 mi.2 = 1,036 km2 Area can be expanded ~155 mi.2 (~401 km2) by adding northern portion of GMU 421 (north of state routes 6 & 330, the north slope of the Plateau Creek drainage. South Uncompahgre Plateau (southern halves of GMUs 61 & 62) ~870 mi.2 = 2,253 km2 Other Attributes Large enough for core study area and buffer. Ratio of public:private land (mi.2) ~4:1. Town of Dolores is at the south end of this area. Substantial number of people live on the plateau. Domestic sheep and cattle use the area. Puma hunting pressure is moderate. Puma predation on domestic animals is low. Minimum study area size. Ratio of public:private land (mi.2) ~2:1. Substantial number of people live along the I-70 corridor and in lower Mamm, Hollow, Divide, and Battlement Creeks, and on Grass Mesa. Recent research on elk seasonal movements, survival and causespecific mortality rates (Freddy). An unknown number of cattle and horses use the area, but there are no domestic sheep. Gas exploration and development is occurring on the area. Puma hunting pressure is moderate. Puma predation on domestic animals is low. Large enough for core study area and buffer. Ratio of public:private land (mi.2) ~3:1. Ongoing mule deer research (Bishop et al. 2003), substantial “pre-treatment” data on mule deer productivity, survival, and cause-specific mortality (Pojar, Watkins, Bishop). Historical puma research (Anderson et al. 1992). Substantial number of people live in the eastern foothills and along the eastern, western, and southern edges of the plateau. Domestic sheep (~6 operators), cattle, and horses use the area. Puma hunting pressure is moderate. Puma predation on domestic animals is low. Puma can be captured year-round using 4 basic methods: trained dogs, cage traps, foot-hold snares, and hands (for small cubs). Capture efforts with dogs will be conducted mainly during the winter when snow facilitates searches for puma tracks and the ability of dogs to follow puma scent. The study area will be searched systematically multiple times per year by four-wheel-drive trucks, all-terrain vehicles, snow-mobiles, and walking. When puma tracks ≤1 day old are detected, trained dogs will be released to trail puma. Puma usually climb trees to take refuge from the dogs. Adult and subadult puma captured for the first time or requiring a change in telemetry collar will be immobilized with Telazol (tiletamine hydrochloride/zolazepam hydrochloride) dosed at 3.3 mg/kg estimated body mass (Wildlife Restraint Handbook, 1996, California Dep. of Fish and Game, Wildlife Investigation Laboratory, Sacramento). Immobilizing agent will be delivered in a Pneu-Dart® shot from a CO2-powered pistol. Immediately, a 3m-by-3m square nylon net will be deployed beneath the puma to catch it in case it falls from the tree. A researcher will climb the tree, fix a Y-rope to two legs of the puma and lower the cat to the ground with an attached climbing rope. Once the puma is on the ground, its head will be covered, its 72 �legs tethered, and vital signs monitored (Logan et al. 1986). (Normal signs: pulse ≈ 70―80 bpm, respiration ≈ 20 bpm, capillary refill time ≤2 sec., rectal temperature ≈ 101oF average, range = 95―106oF.) (Wildlife Restraint Handbook, 1996, California Dep. of Fish and Game, Wildlife Investigation Laboratory, Sacramento). A cage trap will be used to capture adults, subadults, and large cubs when puma can be lured into the trap using road-killed or puma-killed ungulates (Sweanor et al. 2004). Efficiency of the trap will be enhanced by using an automated digital call box that emits puma vocalizations (Wildlife Technologies, Windham, NH). Researchers will monitor the set cage trap from about 1 km distance by using VHF beacons on the cage and door. This allows researchers to be at the cage to handle captured puma within 30 minutes. Puma will be immobilized with Telazol injected with a pole syringe. Immobilized puma will be restrained and monitored as described above. Foot-hold snares will be used to capture adults, subadults, and large cubs as described by Logan et al. (1999). Puma will be immobilized with Telazol injected with a pole syringe and their vital signs monitored during the handling procedures. Efficiency of snares will also be enhanced with the use of an automated digital call box. Small cubs (≤10 weeks old) will be captured using our hands (covered with clean leather gloves) or with a capture pole. Cubs will be restrained inside new burlap bags during the handling process and will not be administered immobilizing drugs. Cubs at nurseries will be approached when mothers are away from nurseries (as determined by radio-telemetry). Cubs captured at nurseries will be removed from the nursery a distance of ≈100 m to minimize disturbance and human scent at nurseries. Immediately after handling processes are complete, cubs will be returned to nurseries (Logan and Sweanor 2001). All captured puma will be examined thoroughly to ascertain sex and describe physical condition and diagnostic markings. Age of adult puma will be estimated initially by the gum-line recession method (Laundre et al. 2000) and dental characteristics of known-age puma (Logan and Sweanor, unpubl. data). Ages of subadult and cub puma will be estimated initially based on dental and physical characteristics of known-age puma (Logan and Sweanor unpubl. data). Body measurements recorded for each puma will include at a minimum: mass, pinna length, hind foot length, plantar pad dimensions. Tissue collections will include: skin biopsy (from the pinna receiving the 6 mm biopsy punch for the ear-tag ), blood (30 ml from the saphenous or cephalic veins), and hair (from various body regions) for genotyping individuals, parentage analysis and disease screening; fecal for diet analyses. Universal Transverse Mercator Grid Coordinates on each captured puma will be fixed via Global Positioning System (GPS, North American Datum 27). Marking, Global Positioning System and Radio-telemetry- Objectives 1―9 Puma do not possess easily identifiable natural marking, such as tigers (see Karanth and Nichols 1998, 2002), therefore, the capture and marking of individual puma is essential to a number of program objectives. Adult and subadult puma will be marked 3 ways: radio-collar, ear-tag, and tattoo. The identification number tattooed in one pinna is permanent and cannot be lost unless the pinna is severed. A colored, numbered 25 mm diameter ear-tag will be inserted into the other pinna to facilitate individual identification during recaptures and in photos taken by field cameras (see capture-recapture methods below). Adult and subadult puma will be fitted with GPS collars (approximately 400 g each, Lotek Wireless, Canada) programmed to fix and store puma locations at least 4 times per day at 6-hours intervals to sample daytime, nighttime, and crepuscular locations. Each collar will have a color-coded identification number on each side also to facilitate identification during physical recaptures and photographic resightings. GPS locations for puma will provide precise, quantitative data for estimating 73 �puma home ranges, habitat use, quantifying road crossings (an index to vulnerability to hunting), finding ungulate kills (at location clusters), and estimating kill rates on ungulate prey (i.e., days per kill). VHF radio transmitters on GPS collars will enable researchers to find those puma on the ground in real time to acquire remote GPS data reports, facilitate recaptures for re-collaring, and to check on their reproductive and physical status. VHF transmitters will have a mortality mode set to alert researchers when puma have been immobile for at least 4 hours so that dead puma can be found to quantify survival rates and agentspecific mortality rates by gender and age. At least one cub of each sex in each litter will be fitted with small VHF transmitter mounted on an expandable collar (≈100g, MOD 210, Telonics, Inc., Mesa, Arizona). Simultaneous locations of mothers and radioed cubs enable researchers to quantify the frequency that mothers are away from cubs to assess the potential risk of orphaning by hunters and other mortality factors, and quantification of survival rates and agent-specific mortality rates. Attrition of cubs in the remainder of the litter can be estimated by periodic visual checks for other siblings by homing on radioed cubs (Logan and Sweanor 2001). Locations of GPS- and VHF-collared puma will be fixed at least once per week from light fixedwing aircraft (e.g., Cessna 182) fitted with radio signal receiving equipment (Logan and Sweanor 2001). This monitoring will enable researchers to find GPS-collared puma to acquire remote GPS location reports from the ground, monitor the status (i.e., live or dead) of individual puma, and to recover carcasses for necropsy. It will also provide simultaneous location data on mothers and cubs. GPS- and VHF-collared puma will be located from the ground opportunistically using hand-held yagi antenna. At least 3 bearings on peak aural signals will be mapped to fix locations and estimate location error around locations (Logan and Sweanor 2001). Aerial and ground locations will be plotted on 7.5 minute USGS maps and UTMs along with location attributes will be recorded on standard forms. GPS locations will be mapped using ArcGIS 8 software. Puma Abundance― Objectives 1, 4 &5 Capture-recapture estimates 1. Capture-recapture models will be used to estimate the parameters of primary interest― absolute numbers of independent puma (i.e., number of puma present in the survey area) and puma density (i.e., number of puma/100 km2) each winter― Dec. through Mar.― when snow facilitates detection and capture of puma, provided that we meet model assumptions. The Dec.―Mar. period also corresponds with Colorado’s puma hunting season. The population of interest is independent puma (i.e., adults and subadults) because those are the puma of legal harvest age. Furthermore, adults comprise the breeding segment of the population and subadults comprise potential recruits into the adult population in ≤1 year. Thus, the sampling unit is the individual independent puma (≈≥1 yr. old). General assumptions for capture-recapture models are: (1) the population is closed; (2) animals do not lose their marks during the interval; (3) all marks are correctly noted and recorded at each trapping occasion; (4) each animal has a constant and equal probability of capture on each capture occasion. Open population models allow the assumption of closure to be relaxed (Otis et al. 1978, White et al. 1982, Pollock et al. 1990). Marked puma will make it possible to acquire most of the basic statistics needed for capture-recapture models. Those include: nj (number of individually identified puma caught and released on occasion j), mj (number of previously marked puma recaptured in occasion j), uj (number of new unmarked puma captured in occasion j) (Otis et al. 1978, White et al. 1982). Attribute data of captured puma, such as age and sex, will be recorded to stratify the population in case separate analysis of different strata is necessary (if sample size allows) to meet certain assumptions of capture-recapture models (Pollock et al. 1990). Precise estimates of puma population size may also allow analyses of functional relationships between population vital rates and population size. 74 �We anticipate it may take 2 years to capture and mark the large majority of puma in the population. Our operational objective will be to have ≥90% of the independent puma marked before capture-recapture occasions commence. Capturing and marking puma is time consuming, and would lengthen the time to thoroughly search the study area for capturing and marking puma during the capture-recapture occasions, therefore, we will capture and mark puma prior to performing capture-recapture occasions. In addition, by marking puma before capture-recapture occasions begin, we will have opportunities to capture female puma at different stages of their reproductive status, and thus reduce the chance that mothers in a stage with suckling cubs and small activity areas are not detected and marked during the winter. After cubs are weaned, the mothers’ activity area expands (Logan and Sweanor 2001). The probability of females having suckling cubs in winter is naturally small; that season exhibits the lowest rate of births (Logan and Sweanor 2001). Our year-round capture efforts using trained dogs, foot-hold snares, and cage traps should help to reduce biases in capture probabilities attributed to any individual capture method (Miller et al. 1997). Thus, capture-recapture occasions may not begin until the end of the second winter. Capture-recapture occasions performed at that time will be viewed as a pilot study allowing us to examine the logistics of the field methods, the extent to which model assumptions are met, biases in field methods (relative to GPS data on collared puma), and precision of capturerecapture models used to estimate the puma population. Data gathered directly from GPS-collared puma and knowledge of the study area acquired by the research team in years 1―2 will allow us to assess if capture-recapture methods are appropriate (i.e., if basic assumptions can be met), and if they are, facilitate the exact design of the mark-recapture schemes for population and density estimates. Movements of GPS-collared puma in and out of the study area during capture occasions will also allow us to estimate corrections for such movements (White 1996). A composite range (i.e., minimum convex polygon) of all the GPS-collared puma home ranges (i.e., using locations from each of the collared puma) will be estimated and mapped to define the search area (i.e., the area inhabited by the estimated population) and allow mapping of search routes for a thorough systematic search of the area to detect puma for capture (i.e., any individual puma should not have a negligible probability of detection). Density estimates using the generated population estimates will be based on the search area (i.e., N/Area) (Miller et al. 1997). A grid will be constructed on the same search area with cells equal to the minimum home range size. There will be a minimum of 4 search routes per cell, each chosen to sample a quarter of each cell. Any spaces in habitat on the search area not occupied by collared puma will also be sampled. Capture occasions will be repeated 3―6 times each winter (i.e., t1, t2, t3,...t6) to resample the population. Unmarked puma will be marked and returned to the population to increase the precision of population estimates. Teams of trained houndsmen (4―6 teams of at least 2 persons each) will by used to systematically and thoroughly search the study area each occasion. Capture occasions will be about 1―2 weeks apart. Capture occasions will commence 1―2 days following fresh snowfall that covers the study area and last 3―5 days (i.e., this is how long it may take teams to thoroughly search the study area). But if fresh snowfall is lacking, we will attempt capture efforts anyway (although this could increase variation in capture probabilities). At puma captures, the puma identification number, sex, age, and location (U.T.M. coordinate) will be the minimum information recorded. If the same individual puma is caught more than once in the same occasion, each capture will be recorded, but only the first capture will be used for data analysis. All capture-recapture occasions will be conducted within a 2―3 month span in winter to minimize the chance of population changes (i.e., deaths, immigration, emigration). Once capture history data on puma are gathered, estimates of the number of independent puma in the population in winter can be made by using capture-recapture models that deal with variation in capture probability and closed or open populations (Otis et al. 1978, White 75 �et al. 1982, Pollock et al. 1990). In order for closed models to be valid, the population of independent puma cannot change (as a result of death, emigration, or immigration) during the 2―3 month span that contain the capture-recapture occasions. Because the precision of estimates for small populations is sensitive to the probability of capture (White et al. 1982, Pollock et al. 1990), our operational goal will be to achieve capture probabilities of about 0.6 (for 3 occasions) and 0.4 (for 6 occasions) to yield capture probabilities ≥0.9 for individual puma in the population each winter (Trolle and Kery 2003). Theoretically, capture probabilities within this range (i.e., 0.4―0.6) would tend to reduce the coefficient of variation of the estimate to about 0.20 (i.e., increase the precision of the estimate) in small populations where individuals have a survival probability of about 0.90 in 5 samples (Pollock et al. 1990:72), which is realistic for puma. In addition, behavior, movements, and survival rates of GPS-collared puma will allow direct biological examinations of assumptions of geographic and demographic closure (White et al. 1982) and variation in capture probability of individual puma and puma classes (i.e., adult females, adult males, subadult females, subadult males). If capture probabilities vary by puma class, we will examine if data stratification is necessary or possible (depending upon sample size). For example, we expect the larger home ranges of male puma to expose them to more search routes, thus, this may increase their probability of capture. If the assumption of demographic closure cannot be satisfied, then open population models may be used (Pollock et al. 1990). GPS locations (4 fixes/day) on individual puma will provide data on the probability that puma may temporarily move out of and back into the survey area between capture occasions. Unmarked puma that are subsequently GPS-collared should provide such information as well. This will allow us to determine the number of marked puma present in the search area each capturerecapture occasion and the probability that unmarked puma move in and out of the search area during each occasion. 2. Photographic captures and recaptures (i.e., camera traps) could be assessed in a 3-year pilot study (e.g., years 2―4) as an independent method for estimating puma numbers and density on the study area (Mace et al. 1994, Karanth 1995, Karanth and Nichols 1998, Karanth and Nichols 2001, Trolle and Kery 2003). The pilot study could be carried out by a graduate student and begin in the second winter, at which time we may begin estimating the puma population using capturerecapture methods (described above). 3. Genotype captures and recaptures could be assessed in coordination with the physical and photographic capture methods described above (1 and 2). This could also be initiated as a 3-year pilot study (e.g., years 2―4) done by a graduate student. Genotypes can be used to estimate minimum population size directly (Kohn et al. 1999) and in capture-recapture models (e.g., Otis et al. 1978, Boulanger et al. 2002). However, genotyping errors should be estimated and considered in population estimates (Creel et al. 2003). This project could be another independent non-invasive method for estimating puma numbers and density on the study area. Moreover, this method may be useful to monitor puma populations in Colorado (see Indices to Puma Abundance below). Vulnerability of Puma to Hunters― Objective 3 Puma hunting in Colorado normally involves hunters searching for puma tracks while driving snow-covered roads with four-wheel drive vehicles or snowmobiles. Thus, vulnerability of puma to hunters is associated with the frequency that puma cross roads (Murphy 1983, Van Dyke et al. 1986, Barnhurst 1986). Hunters active on snow may successfully catch puma >85% of the time that they release dogs on tracks, and road access influences puma hunting distribution (Murphy 1983). 76 �Road crossing frequencies of GPS-collared puma will be used to assess relative vulnerability of various sex, age, and reproductive classes of puma to detection by puma hunters. Density of roads (i.e., km/100 km2) will be estimated within each GPS-collared puma home range (i.e., 100% minimum convex polygon), each quadrat in the sampling frame, and the entire study area. Road crossings per puma per 24hour periods will be quantified (GPS collars programmed for 4 locations per day, each 6 hours apart). Comparisons will be made between road crossing frequencies of puma classes and road densities in home ranges for puma classes. Puma classes will be: adult females (≥2 yr. old) with no cubs; adult females with cubs ≤2 months old (i.e., nursling cubs), adult females with cubs 3―12 months old (i.e., weaned, carnivorous cubs), subadult females (i.e., independent females <2 yr. old), adult males (i.e., ≥2 years old), and subadult males (i.e., independent males <2 years old) (Logan and Sweanor 2001). Vulnerability will also be quantified using road crossing frequencies of different classes of puma (female, male, divided into adult and subadult age classes if individuals are known) during the track index discussed below. In addition, actual capture rates of those puma will be quantified during capturerecapture occasions (no. of captures per puma class/no. of puma in each class). Frequency of capturing known puma mothers with and without their cubs will also be tallied to quantify vulnerability of puma mothers to legal harvest (i.e., known mothers without cubs by their side). During the puma population decline (i.e., reduction) phase (years 6―10), hunters can be used to kill puma. Relative vulnerability to and selection by hunters will be quantified during hunting seasons by having hunters report number of hunter-days, number of tracks encountered and locations (fixed by GPS), number of times dogs were released on tracks, number of captures, characteristics of killed or captured puma (i.e., hunters may capture puma and release them). At the same time, researchers will have quantitative knowledge of puma available for harvest on the study area (as a result of ongoing capture and marking procedures and GPS data) to estimate capture rate per puma class. The hunter-kill will also allow a direct assessment of puma mothers in the harvest and a comparison of the fates of potential orphaned cubs with cubs in intact families. Indices to Puma Abundance― Objective 5 This project will develop and test both the efficacy and feasibility, including costs and other management considerations, of using indices to monitor changes in puma abundance. Two such indices are track counts and catch-per-unit-effort. These indices will be calibrated with the estimated puma population. Track Counts An index to puma abundance using counts and classification of puma tracks on snow-covered routes will be developed and tested. This will be done simultaneous with obtaining estimates of puma population size using capture-recapture occasions with houndsmen teams (above) and spatial analysis of home ranges of GPS-collared puma. A 3-year pilot study for this index may be started in year 2 and could be conducted by a graduate student. Information on puma gathered during years 1―2 will facilitate the exact research design. Experimental manipulation of the puma population resulting in a 5-year increase phase and a 5-year decline phase will allow testing the index through known puma population changes and assessment of the sensitivity of the index with the parameter in question― puma population size― by analyzing the statistical power of the method to detect population change (up or down) (Kendall et al. 1992, Beier and Cunningham 1996). The main operating assumption is that the frequency of finding puma tracks on snow-covered routes is related to puma numbers. We will examine the number of individual puma track sets/km (differentiated by size and direction of travel) and presence of puma tracks/km of search route to see 77 �which metric better detects actual population change (Kendall et al. 1992). In addition, we will examine how frequency of different classes of tracks (male, female, females with cubs) may relate to the known puma population changes. The most direct relationship will be a linear one between puma numbers and frequency of encountering puma tracks. Snow-tracking conditions (e.g., powder, crusted, slush, continuous, patchy) will be categorized each search day. Tracking teams, different than houndsmen teams used in capture-recapture occasions, will be used to make the puma track surveys 3 to 6 times per winter. Track surveys will be run on 4-wheel drive vehicles or snowmobiles 1―2 days following snowfall that covers the study area. Effort and costs will be quantified. Catch-Per-Unit-Effort The main operating assumption is that the amount of effort to capture puma is related to puma numbers (Lancia et al. 1996). The puma research team(s) will quantify the number of days required to capture individually identified puma with dogs during each year in the population increase phase. We will also quantify the number of days required to capture unmarked puma with dogs each year, and treat marked puma as though they have been removed from the unmarked population. Number of days per capture will also be quantified by capture teams involved in the capture-recapture occasions each winter. During the decline phase, puma hunters used to reduce the puma population will quantify the number of hunt days per puma captured each winter. Theoretically, the number of days per puma capture should increase as the puma population is reduced by 20% increments in years 6 and 7, and in possible further reductions in years 8―10. Hunters will also be asked to record (i.e., GPS location, date) the total number of tracks of independently-traveling puma (classified as male and female), and the number of tracks of female puma and cubs they encounter during their hunting periods. Male and female track categories will be distinguished by the width of the hind foot plantar pads. Hind foot plantar pad widths that are >52 mm will be classed as male; hind foot plantar pad widths ≤52 mm will be classes as female. We will explore functional relationships of these efforts to the estimated puma population on the study area. Quantifying Puma Diet and Ungulate Kill Rates― Objective 8 Data collected on puma diet and ungulate kills are not directly pertinent to a puma population study. However, they would be basic to an integrated study that involves effects of puma predation on mule deer and elk. Location clusters where puma are located for ≥2 nights will be investigated to estimate puma kill rates of ungulate prey (i.e., days/ungulate kill type/puma class) (Anderson and Lindzey 2003). Sex of animals will be determined by secondary sex characteristics and ages will be estimated from tooth eruption patterns (Quimby and Gaab 1952, Robinette et al. 1957, Dimmick and Pelton 1996) and cementum annuli of incisors (Low and Cowan 1963). Necropsies will be performed on all ungulate prey recovered in the field (Roffe et al. 1996), whether killed by puma or not, and data will be recorded on standard forms. If disease is suspected, whole carcasses or vital organ tissues will be collected and preserved by standard procedures (Roffe et al. 1996) and submitted for analysis to the Colorado Division of Wildlife’s Wildlife Health Laboratory or the Colorado State University Diagonostic Laboratory. An index to physical condition of ungulates prior to death will be estimated from percent marrow fat in femurs or metatarsi (depending on presence; femurs are preferred) (Neiland 1970, Mech and DelGiudice 1985, Fuller et al. 1986, Husseman et. al. 2003). Puma feces will be collected opportunistically year-round and stored by either freezing or oven drying (80-85oC, then stored in paper bags with a fumigant) for later macroscopic diet analysis (Big Sky Laboratory, Florence MT) to estimate frequency of occurrence of prey species (Litvaitis et al. 1996). This research component could also be carried out by a graduate student. 78 �Behavior of Puma Subject to Aversive Conditioning― Objective 9 Information on responses of puma to aversive conditioning is lacking. Individual puma with activities in residential areas on the study area might be research subjects on effectiveness of aversive conditioning. GPS collars on puma would be the primary source of behavioral response data before, during, and after aversive conditioning treatments. Behavior and Survival of Translocated Puma― Objective 10 If translocation is chosen as the method of reducing the puma population during the decline phase (years 6―10), then researchers will remove puma at rates needed to test research hypotheses. Prior to translocation, potential puma habitat areas for the release of the puma will need to be identified which are: 1) relatively remote, 2) large enough to accommodate exploratory movements up to 84 km away from release areas, and 3) not near human residential areas, domestic animal operations, or desert bighorn sheep populations (Ruth et al. 1998, Logan and Sweanor 2001). Puma will be captured alive on the study area, fit with new GPS collars, transported in wooden crates, provided food and water, and translocated by truck a minimum of 120 airline km (75 mi.) for females and 220 airline km (137 mi.) for males (Ruth et al. 1998). GPS collar locations will allow researchers to map movements of translocated puma. The VHF transmitters will allow researchers to quantify survival rates and agent-specific mortality rates. This research could also be carried out by a graduate student. ANALYTICAL Puma class survival rates and agent-specific mortality rates will be estimated by using KaplanMeier (Pollock et al. 1989a, b) and Trent and Rongstad procedures (Micromort software, Heisey and Fuller 1985). Cub survival curves for each gender will also be plotted with survival rate on age in months (Logan and Sweanor 2001:119). To analyze capture-recapture, photographic, and genetic capture-recapture data, closed population capture-recapture models are available in program CAPTURE obtainable at www.mbrpwrc.usgs.gov/software.html and program MARK obtainable at www.cnr.colostate.edu/~gwhite.). Closed population model selection can be achieved with the algorithm based on goodness-of-fit tests and between model tests in program CAPTURE (Otis et al. 1978). For open populations, programs JOLLY (for 1 age class), and JOLLYAGE (handles 2 age classes) are available at www.mbr-pwrc.usgs.gov/software.html. Programs JOLLY and JOLLYAGE contain chi-square goodness-of-fit tests of model assumptions and between model tests that enable researchers to choose the most appropriate model for the data (Pollock et al. 1990). NOREMARK (White 1996), also available at www.cnr.colostate.edu/~gwhite, has an extension that accommodates immigration and emigration; thus, it does not assume geographic closure (but demographic closure is still assumed). Finite rates of increase (Nt+1/Nt) between consecutive years and average annual rates of increase (r) for 3- to 5-year periods will be calculated (Caughley 1978, Van Ballenberghe 1983) and plotted. Graphical methods will be used to examine relationships of track counts and catch-per-unit effort (i.e., indices to puma abundance) to changes in the population of independent puma. Linear regression procedures and coefficients of determination will be used to assess functional relationships of track counts and catch-per-unit effort to changes in the population of independent puma if data for the response variable are normally distributed and the variance is the same at each level. If the relationship is not linear, data is non-normal, and variances are unequal, we will consider appropriate transformations of the data for regression procedures (Ott 1993). We will also consider non-parametric correlation methods, such as Spearman’s rank correlation coefficient, to test for a monotonic relationship between the index of abundance and the change in the puma population (Conover 1999). 79 �Statistical analyses will be performed using SYSTAT 10.2 and SAS 6.11. The risk of committing a type I error (i.e., concluding that a population change occurred when it did not) will be controlled at alpha = 0.10 because we will normally have small sample or population sizes (typical of large-carnivore studies). The higher alpha level will increase the probability of detecting a change and reduce the risk of a type II error (i.e., failing to reject a null hypothesis that is false). For managers, the risk of a type II error is probably more important. ArcView 8 geographic information system software will be used to map and analyze puma locations, movements, and home ranges. It will also be used to map and quantify attributes of the study area and sampling frame. PRELIMINARY SCHEDULE Years 1―5 (2005―2009) will be the puma population increase phase. Protecting the puma population from sport-hunting will be vital to allowing the puma population to increase within the bounds of the ecological carrying capacity of the study area. This will allow researchers to quantify baseline demographic data on the puma population and test indices to puma abundance during an increase phase. In this phase, capture-recapture occasions, track counts (for index to abundance), and photographic and genetic capture-recapture efforts will begin in about year 2 (2006). Years 6―10 (2010―2014) will be the puma population decline phase. Puma hunters (or translocation) will be used to experimentally reduce the puma population. The portion of independent puma (i.e., adults and subadults) in the population will be reduced by 20% in year 6 and 20% more in year 7 (i.e., a 40% reduction from year 5). Additional reductions may be made to test the indices to abundance or other hypotheses that may be developed and related to effects of harvest or puma predation on mule deer and elk. Those decisions can be made later in project development and as late as years 8―10. REGULATORY NEEDS Puma on the study area that may be involved in depredation of livestock or human safety incidences will not be treated any differently than other puma in Colorado, whether they are marked or not. Thus, they may be lethally controlled. Researchers that find that GPS-collared puma have killed domestic livestock will record such incidents to facilitate reimbursement to the property owner for loss of the animal(s). The increase phase in years 1―5 will require a temporary interruption of puma sport-hunting on the study area and protection of radio-collared puma that range off the study area. In years 6―10, regulated puma sport-hunting will resume. POTENTIAL COOPERATORS The Colorado Division of Wildlife will be the principal research and regulating agency in this program. The Bureau of Land Management and the Forest Service will be cooperators because the majority of the study area may be on lands under their management jurisdiction. U. S. D. A., A. P. H. I. S. Wildlife Services may provide puma capture assistance. Private landowners on the study area will be asked to cooperate with this effort. Colorado State University and other universities may cooperate by providing graduate research assistants and professors to carry out specific projects of the research program. Private individuals interested in the puma research may be asked to cooperate in puma capture and monitoring operations. 80 �Table 2. Preliminary puma research schedule. Phase Objectives: 1. 2. 3. 4. 5. 6. 7. 8. 9. Increase (Years 1―5) Decline (Years 6―10) Methods & Data: Initial capture & mark efforts of ≥90% of Capture-recapture estimates (yrs. 6―10) using independent puma (yrs. 1―2). data from physical and photographic & Capture-recapture estimates (yrs. 2―5; 3 genotype captures (if reliable). yr. minimum required for trend) using Reduce puma population using hunting or physical, photographic & genotype translocation. Reduce by 20% increments in captures. Quantify puma population sex & years 6 & 7. Puma hunting will continue, and age structure, density, & population there may be additional population reductions growth rate. in subsequent years. Quantify structure of hunter-kill. Quantify puma population vital rates as Quantify puma population vital rates as population increases. population declines. Quantify agent-specific mortality & puma Quantify agent-specific mortality & road-crossing frequency. vulnerability and selectivity of puma to hunters. Develop and test puma population models Develop and test puma population simulation validated by observed increase phase models validated by observed decrease phase puma population. puma population. Use track counts, catch-per-unit effort, & Use track counts, catch-per-unit effort, & genotype capture-recapture methods as genotype capture-recapture methods (if indices to puma abundance (yrs. 2―5). reliable) as indices to puma abundance. Use GPS data to quantify puma activity in Use GPS data to quantify puma activity in relation to people and human facilities on relation to people and human facilities on the the study area. study area. Use GPS data to quantify puma use of Use GPS data to quantify puma use of habitats habitats and landscape linkages. and landscape linkages. Estimate puma kill rates on mule deer and Estimate puma kill rates on mule deer and elk elk using GPS data. Quantify puma diet using GPS data. Quantify puma diet from from feces. feces. Describe & quantify behavior & survival of translocated puma if translocation is used to reduce the puma population. Begin final data analysis & report year 10. POTENTIAL IMPEDIMENTS Because of the relatively low densities of puma, difficulty of capture and research, obtaining needed sample sizes is expensive. Furthermore, multiple years of study are requisite to fulfill objectives. Collared puma that are killed, therefore, represent a significant effort and data loss. Minimizing such losses is a challenge that will improve the efficiency of the study. For certain projects within the program, experimental manipulations of the puma population on the primary study area, possibly ranging from extreme protection to extreme suppression at different stages of the project, are necessary to maximize reliability and scientific defensibility of findings. 81 �FINANCIAL ESTIMATES Conducting intensive puma research requires significant and steady financial support. Yearly costs during years 1-5 are estimated to range between $177,000 and $355,000 (Table 3). LITERATURE CITED Anderson, A. E. 1983. A critical review of literature on puma (Felis concolor). Colorado Division of Wildlife, Special Report No. 54. _____, D. C. Bowden, and D. M. Kattner. 1992. The puma on Uncompahgre Plateau, Colorado. Colorado Division of Wildlife, Technical Publication No. 40. Anderson, C. R., and F. G. Lindzey. 2003. Estimating cougar predation rates from GPS location clusters. Journal of Wildlife Management 67:307-316. Barnhurst, D. 1986. Vulnerability of cougars to hunting. Masters Thesis. Utah State University, Logan. Beier, P., and S. C. Cunningham. 1996. Power of track surveys to detect changes in cougar populations. Wildlife Society Bulletin 24:540-546. Bishop, C. J., G. C. White, D. J. Freddy, and B. E. Watkins. 2003. Effect of nutrition and habitat enhancements on mule deer recruitment and survival rates. Colorado Division of Wildlife, Wildlife Research Report, Federal Aid in Wildlife Restoration Project W-153-R, Progress Report. Ft. Collins, Colorado, U.S.A. Boulanger, J., G. C. White, B. N. McLellan, J. Woods, M. Proctor, and S. Himmer. 2002. A meta-analysis of grizzly bear DNA mark-recapture projects in British Columbia, Canada. Ursus 13:137-152. Brent, J. 1981. Colorado mountain lion population investigations: Game management units 33 and 40. Project Report October 27, 1980 through March 31, 1981. Colorado Division of Wildlife, Grand Junction. ______ _____. 1982. Colorado mountain lion population investigations: Game management units 33 and 40, Phase II. Project Report November 12, 1981 through March 31, 1982. Colorado Division of Wildlife, Grand Junction. _____. 1983. Colorado mountain lion population investigations: Game management units 33 and 40, Phase III. Project Report January 29, 1983 through August 30, 1983. Colorado Division of Wildlife, Grand Junction. Caughley, G. 1978. Analysis of vertebrate populations. John Wiley and Sons, New York. Colorado Division of Wildlife 2002-2007 Strategic Plan. 2002. Colorado Department of Natural Resources, Division of Wildlife. Denver. Conover, W. J. 1999. Practical nonparametric statistics. John Wiley & Sons, Inc., New York. Currier, M. J. P., and K. R. Russell. 1977. Mountain lion population and harvest near Canon City, Colorado, 1974-1977. Colorado Division of Wildlife Special Report No. 42. DeSimone, R., B. Semmens, B. Shinn, T. Chilton, J. Sikich, and B. Weisner. 2002. Garnet Mountains Mountain Lion Research. Progress Report, January 2001 to December 2002. Montana Fish, Wildlife and Parks, Helena. Dimmick, R. W., and M. R. Pelton. 1996. Criteria of sex and age. Pages 169-214 in T. A. Bookhout, editor. Research and management techniques for wildlife and habitats. Fifth edition. The Wildlife Society, Bethesda Maryland. Heisey, D. M., and T. K. Fuller. 1985. Evaluation of survival and cause specific mortality rates using telemetry data. Journal of Wildlife Management 49:668-674. Husseman, J. S., D. L. Murray, G. Power, and C. M. Mack. 2003. Correlation patterns of marrow fat in Rocky Mountain elk bones. Journal of Wildlife Management 67:742-746. Fuller, T. K., P. L. Coy, and W. J. Peterson. 1986. Marrow fat relationships among leg bones of whitetailed deer . Wildlife Society Bulletin 14:73-75. Karanth, K. U. 1995. Estimating tiger Panthera tigris populations from camera-trap data using capturerecapture models. Biological Conservation. 71:333-338. 82 �_____, and J. D. Nichols. 1998. Estimating tiger densities in India from camera trap data using photographic captures and recaptures. Ecology 79:2852-2862. _____, and J. D. Nichols. 2002. Monitoring tigers and their prey: A manual for researchers, managers and conservationists in tropical Asia. Centre for Wildlife Studies, Bangalore, India. Kendall, K. C., L. H. Metzgar, D. A. Patterson, and B. M. Steele. 1992. Power of sign surveys to monitor population trends. Ecological Applications 2:422-430. Koloski, J. H. 2002. Mountain lion ecology and management on the Southern Ute Indian Reservation. M. S. Thesis. Department of Zoology and Physiology, University of Wyoming, Laramie. Kufeld, R. C., J. H. Olterman, and D. C. Bowden 1980. A helicopter quadrat census for mule deer on Uncompaghre Plateau, Colorado. Journal of Wildlife Management 44:632-639. Laundre, J. W., L. Hernandez, D. Streubel, K. Altendorf, and C. L. Lopez Gonzalez. 2000. Aging mountain lions using gum-line recession. Wildlife Society Bulletin 28:963-966. Lancia, R. A., J. D. Nichols, and K. H. Pollock. 1996. Estimating the number of animals in wildlife populations. Pages 215-253 in T. A. Bookhout, editor, Research and management techniques for wildlife and habitats. The Wildlife Society, Bethesda, Maryland. Litvaitis, J. A., K. Titus, and E. M. Anderson. 1996. Measuring vertebrate us of terrestrial habitats and foods. Pages 254-274 in T. A. Bookhout, editor. Research and management techniques for wildlife and habitats. Fifth edition, revised. The Wildlife Society, Bethesda Maryland. Logan, K. A., E. T. Thorne, L. L. Irwin, and R. Skinner. 1986. Immobilizing wild mountain lions (Felis concolor) with ketamine hydrochloride and xylazine hydrochloride. Journal of Wildlife Diseases. 22:97-103. _____, L. L. Sweanor, J. F. Smith, and M. G. Hornocker. 1999. Capturing pumas with foot-hold snares. Wildlife Society Bulletin 27:201-208. _____, and L. L. Sweanor. 2001. Desert puma: evolutionary ecology and conservation of an enduring carnivore. Island Press, Washington, D.C. Mace, R. D., S. C. Minta, T. L. Manley, and K. E. Aune. 1994. Estimating grizzly bear opulation size using camera sightings. Wildlife Society Bulletin 22:74-83. McDaniel, G. W., K. S. McKelvey, J. R. Squires, and L. F. Ruggiero. 2000. Efficacy of lures and hair snares to detect lynx. Wildlife Society Bulletin 28:119-123. Mech, L. D., and G. D. DelGuidice. 1985. Limitations of the marrow fat technique as an indicator of body condition. Wildlife Society Bulletin 13:204-206. Miller, S. D., G. C. White, R. A. Sellers, H. V. Reynolds, J. W. Schoen, K. Titus, V. G. Barnes, Jr., R. B. Smith, R. R. Nelson, W. B. Ballard, and C. C. Schwartz. 1997. Brown and black bear density estimation in Alaska using radiotelemetry and replicated mark-resight techniques. Wildlife Monographs 133:1-55. Murphy, K. M. 1983. Relationships between a mountain lion population and hunting pressure in western Montana. Final Report, Federal Aid in Wildlife Restoration, Project W-120-R13 and 14. Montana Department of Fish, Wildlife, and Parks, Helena. Otis, D. L., K. P. Burnham, G. C. White, and D. R. Anderson. 1978. Statistical inference from capture data on closed animal populations. Wildlife Monographs 62:1-135. Ott, R. L. 1993. An introduction to statistical methods and data analysis. Fourth edition. Wadsworth Publishing Co., Belmont, California. Pojar, T. M. 2000. Investigating factors contributing to declining mule deer numbers. Colorado Division of Wildlife, Wildlife Research Report, Federal Aid in Wildlife Restoration Project W-153-R-13, Progress Report. Fort Collins, Colorado, U.S.A. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989a. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. _____, S. R. Winterstein, and M. J. Conroy. 1989b. Estimation and analysis of survival distributions for radio tagged animals. Biometrics 45:99-109. _____, J. D. Nichols, C. Brownie, and J. E. Hines. 1990. Statistical inference for capture-recapture experiments. Wildlife Monographs 107:1-97. 83 �Quimby, D. C., and J. E. Gaab. 1952. Preliminary report on a study of elk dentition as a means of determining age classes. Proceedings Western Association of State Game and Fish Commissions. 32:225-227. Robinette, W. L., D. A. Jones, G. Rogers, and J. S. Gashwiler. 1957. Notes on tooth development and wear for Rocky Mountain mule deer. Journal of Wildlife Management 21:134-153. Ruth, Tik, K. A. Logan, L. L. Sweanor, M. G. Hornocker, and L. L. Temple (1998) Evaluating cougor translocation in New Mexico. Journal of Wildlife Management 62:1264-1275. Sweanor, L. L., K. A. Logan, and M. G. Hornocker. 2000. Cougar dispersal patterns, metapopulation dynamics, and conservation. Conservation Biology 14:798-808. _____, K. Logan, J. Bauer, and W. Boyce. 2004. Puma and humans in and around Cuyamaca Rancho State Park, San Diego County, California. School of Veterinary Medicine, University of California, Davis. Trolle, M., and M. Kery. 2003. Estimation of ocelot density in the pantanal using capture-recapture analysis of camera-trapping data. Journal of Mammalogy 84:607-614. Van Ballenberghe, V. 1983. Rate of increase of white-tailed deer on the George Reserve: a re-evaluation. Journal of Wildlife Management 47:1245-1247. Van Dyke, F. G., R. H. Brocke, and H. G. Shaw. 1986. Use of road track counts as indices of mountain lion presence. Journal of Wildlife Management 50:102-109. Van Sickle, W. D., and F. G. Lindzey. 1992. Evaluation of road track surveys for cougars (Felis concolor). Great Basin Naturalist 52:232-236. White, G. C., D. R. Anderson, K. P. Burnham, and D. L. Otis. 1982. Capture-recapture and removal methods for sampling closed populations. Los Alamos National Laboratory Publication LA-8787NERP. Los Alamos, NM, U.S.A. _____. 1996. NOREMARK: population estimation from mark-resighting surveys. Wildlife Society Bulletin 24:50-52. Prepared by ______________________ Kenneth A. Logan, Wildlife Researcher 84 �Table 3. Estimated project costs for years 1―5 only. Budget Item Personnel: -DOW Researcher -Houndsman -Project Technician -Temporary Technician Volunteers Support: Lodging, food, fuel for 8―12 Vehicles: -4x4 Trucks (2) -all terrain vehicles (3) -snowmobiles (1)a -utility trailers (1)a -dog sled & trailer Gasoline Vehicle Maintenance GPS- & Radio-telemetry: -GPS-collars -VHF-collars (cub) -cub collar material -command unit (1) -receivers, type (2) -H-antennae (3) -omni antennae (3) -coaxial cables (2) -coaxial cables (2) -antenna switch-box (2) -intercom system (1) -head sets (2) Capture Equipment: -drugs -cage trap -snares -call box -miscellaneous (darts, vials, syringes, needles, envelopes, gloves, tapes, calipers, thermometers, ear-tags, tattoos, etc...) Dog Care: -veterinary -food Aerial Support: Work tack: -backpacks, climbing gear, nets, ropes, office materials, etc... Laboratory: -genetics -carcass or tissue analysis -fecal analysis Photographic: -trail cameras (~40) -film (~200) Total a Year 1 Year 2 Year 3 Year 4 Year 5 K. Logan’s support not incl. 12,500 ($2,500/mo.*5 mo.) 35,000 33,000 12,500 35,000 33,000 13,125 35,000 33,000 13,125 35,000 33,000 13,781 35,000 33,000 0 15,000 15,000 15,000 15,000 50,000 (2*$25,000) 18,600 (3*$6,200) 5,300 1,900 1,000 4,800 (2*$2,400) 3,000 0 0 0 0 0 4,800 3,000 0 0 0 0 0 4,800 3,000 60,000 0 0 0 0 4,800 3,000 0 0 0 0 0 4,800 3,000 114,750 (25*$4,590) 3,192 (12*$266) 100 4,500 5,390 (2*$2,695) 636 (3*$212) 234 (3*$78) 56 (2*$28) 29 (2*$14.50) 112 (2*$56) 285 300 (2*150) 0 0 0 0 0 0 0 0 0 0 0 0 22,950 1,596 50 0 0 0 0 0 29 0 0 0 22,950 0 0 0 0 0 78 56 0 56 0 0 22,950 1,596 50 0 0 0 0 0 29 0 0 0 650 (25*$26/bottle Telazol) 2,000 (2*$1,000) 1,000 850 1,000 650 0 0 0 1,000 650 0 0 0 1,000 650 0 0 0 1,000 650 0 0 0 1,000 2,000 600 (120/mo.*5 mo.) 40,000 ($200/hr x 4 hr x 50) 2,000 600 40,000 2,000 600 40,000 2,000 600 40,000 2,000 600 40,000 2,000 1,000 1,000 1,000 1,000 4,650 (25*$186) 600 4,650 4,650 600 4,650 4,650 600 4,650 4,650 600 4,650 4,650 600 4,650 0 (40*$430) 0 (250*$6) 354,684 17,200 1,500 177,150 4,300 1,500 189,500 0 1,500 243,715 0 1,500 185,856 Two snowmobiles and 1 trailer are already available for the project. 85 �APPENDIX I Sex Determination of Mountain Lions Bayed in Trees With little effort the sex can be determined for a mountain lion bayed in a tree. Refer to the photos of the different lions, 4 males (A―D) 2 females (E, F), attached to these tips. Male adult and subadult lions have a conspicuous black spot of hair, about 1 inch diameter, surrounding the opening to the penis sheath behind the hind legs and about 4 to 5 inches below the anus. In between the black spot and the anus is the scrotum, which is usually covered with silver, light brown, and white hair. Therefore, look for the black spot and scrotum. The anus is usually hidden below the base of the tail. Female adult and subadult lions do not have the black spot or scrotum behind the hind legs and below the base of the tail. There is just white hair there. The anus is directly below the base of the tail, and the vulva is directly below the anus. The anus and vulva are usually hidden by the base of the tail. Teats of females are usually inconspicuous, even of mothers with weaned cubs or mothers that have just finished nursing cubs. So teats are usually not a good indicator of sex in treed lions. Sometimes sex determination of lions can be done with the naked eye. But use a pair of binoculars to make sexing lions easier. If a lion’s position in a tree obscures your view, you can get the lion to move around for a better look. Pick up a baseball-bat-size branch and bang on the trunk of the tree. If there is snow on the ground, throw a few snow balls at the lion. You can even climb the tree toward the lion. These actions usually get the lion to move. When it does, be ready to sex the lion. Also, sometimes the lion urinates when bayed by dogs or when a person climbs the tree toward it. Look for the origin of the urine stream. If the urine stream comes from behind the hind legs about 4 to 5 inches below the anus, then the lion is probably a male. If the urine stream comes from under the base of the tail, then it’s probably a female. Tracks may also be indicative of sex. Adult and large subadult male lions usually have hind foot plantar (“heel”) pad widths that exceed 2 1/16 inches (52 mm). Adult and subadult female lions usually have hind foot plantar pad widths less than or equal to 2 1/16 inches. Carry a small ruler or wind-up metal tape in your pocket to make measurements 86 �Male Mountain Lions (A―D) Penis Spot, Scrotum, Anus. Penis (black) spot ~1 inch dia. is ~4-5 inches below anus. A B K. Logan photo K. Logan photo D K. Logan photo Female Mountain Lions (E, F) Vulva directly below anus, both usually hidden by base of tail. No “black spot” 4-5 inches below anus C E K. Logan photo K. Logan photo F 87 K. Logan photo � https://cpw.cvlcollections.org/files/original/0f0af4b0f2b27135880519535d4be4ba.pdf 3326bfe975ebf1787bb8f268b2de5016 PDF Text Text 239 Colorado Division of Wildlife Wildlife Research Report July 2001 and July 2002 PROGRESS REPORT State of_________C=ol=o=rad=o,___ _ __ Division of Wildlife - Mammals Research Work Package No._ _ _ _. ::;3.. :a.0=03:::___ _ __ Predatory Mammals Conservation Task No_ - - - - - - - - - - ' 3 " - - - - - - - Peiod Covered: Pilot Study - Evaluation of GPS Technology in Measuring Chronic Wasting Disease Prevalence Among Deer Preyed upon by Puma January 1, 2001 - December 31, 2001 Author: C. E. Krumm, T.D.I. Beck, M. W. Miller. Personnel: C. E. Krumm, T.D.I. Beck, M. W. Miller Interim Report - Preliminary Results This work continues, and precise analysis of data has yet to be accomplished. A1anipulation or interpretation of these data beyond that contained in this report should be labeled as such and is discouraged. ABSTRACT A prospectus for a pilot study to ascertain the efficacy and feasibility of using Global Position Systems (GPS) technology to measure chronic wasting disease prevalence among puma prey, as well as in other studies of puma, was developed. Objectives of the pilot study are to: 1) Evaluate the potential utility of Televilt Positioning GPS collars in studies of selective predation in puma under field conditions; and 2) Develop and assess the adequacy of field sampling techniques for studying selective predation on CWD-infected mule deer. Two adult puma are to be captured and fitted with GPS collars for the pilot study. ��241 PILOT STUDY Evaluation of new GPS technology in measuring chronic wasting disease prevalence among deer preyed upon by mountain lions C. E. Krumm, T. D. I. Beck, and M. W. Miller Background As a pilot study to test a new technology in Global Positioning Systems (GPS) and its application to studies of predator-prey relationships, we plan to capture and collar two free-ranging puma (Puma concolor) in the foothills west of Ft. Collins in early April of 2001. Our pilot study will evaluate new GPS technology, as well as the potential utility of data collected with this system in testing hypotheses about selective predation; specifically, we will evaluate the ability to compare chronic wasting disease prevalence among puma-killed deer to prevalence among harvested deer. Chronic wasting disease (CWD) is a naturally occurring spongiform encephalopathy of captive and freeranging deer and elk. CWD has become a concern in managing deer herds in northeastern Colorado. Studies conducted the past several years have provided important data on prevalence of CWD (Miller et al. 2000) and the potential effects of selective population control on affected populations (Gross and Miller 2001). It follows that processes fostering selective removal of affected individuals, like test-andslaughter or predation, should be closely evaluated in the context of disease management. New technology in GPS tracking of animals by Televilt Positioning (Lindesberg, Sweden) allows location data to be downloaded remotely without retrieval of collars. Testing the effectiveness and accuracy of these collars will benefit a suite of studies that are being planned across Colorado to examine the selectivity of puma for prey animals (specifically mule deer) of varying condition. These studies will help to answer a fundamental ecological question: Do puma selectively prey on debilitated or compromised animals rather than healthy ones? Objectives Our specific objectives are to: 1. evaluate the potential utility of Televilt Positioning GPS collars in studies of selective predation in puma under field conditions; and 2. develop and assess the adequacy of field sampling techniques for studying selective predation on CWD-infected mule deer. Study Design Because this is a pilot study, we will capture and collar only two adult puma to evaluate equipment and sampling techniques. We regard two individuals as the fewest needed to adequately assess all aspects of equipment use and performance, sampling techniques, and other logistical facets of larger prospective studies. Capture Methods and Handling We plan to capture adult puma for this study using methods described in Shaw ( 1979). Briefly, a tracker with experience in tracking and handling mountain lions will be hired to facilitate capture and will use trained dogs to track and tree or bay each mountain lion. Field anesthesia will be supervised by an attending veterinarian. Anesthetic drugs will be administered intramuscularly via projectile syringe using a gas-powered projector. For capture, puma will be anesthetized with ketamine (10-1 lmg/kg) and xylazine HCl (1.8-2mg/kg) orketamine (2 mg/kg) and medetomidine (0.075 mg/kg) (Shaw 1979, Kreeger 1996). We will observe darted puma for signs of sedation (salivation, unsteadiness of head and body, and a wide-eyed expression). If the puma is treed, then people and dogs will be removed from the immediate area to give the animal a chance to descend before becoming completely anesthetized. If the puma remains in the tree until almost completely anesthetized, then someone wearing climbing gear will climb to the puma and attach either a chest harness (preferred) or hind leg noose and quickly lower the animal �242 before it falls; others will hold a taut net below to break the puma's fall should it slip before a harness or rope can be secured. If signs of anesthesia are inapparent after 15 minutes, then a second full injection will be given. Upon first approach of an apparently anesthetized puma, a 4-5 foot stick will be used to gently prod the paws and muzzle of the animal; if there is no response (i.e. snarling or biting), then we will assume anesthesia is sufficient for handling. Once anesthetized, we will apply eye ointment and a blindfold to reduce visual stimuli, place gauze pads in the puma's ears to reduce auditory stimuli, and restrain its legs with nylon belts or hobbles. A GPS-Simplex collar (Televilt Positioning; maximum weight 600 g) will be fastened around the puma's neck. The leg restraints will be quickly removed, and the puma will be allowed to recover from the sedation either naturally or with the aid of an antagonist; when prescribed, yohimbine HCl (0.125 mg/kg IV) will be used to antagonize xylazine sedation and atipamezole (0.3 mg/kg) will be used to antagonize medetomidine sedation. Postcapture Monitoring According to the manufacturer, the locations of collared animals can be retrieved and plotted several times a day without removing the collars. Up to 2000 satellite positions can be stored in the memory, allowing us to closely monitor the puma's movement on a daily basis. If a puma remains in one location for several hours, we will assume that it has made a kill. Based on data from studies elsewhere (e.g., Homocker 1970, C. Anderson, personal communication), we anticipate that each collared animal will make an ungulate kill every 7 to 11 days on average. We will locate the prospective kill site using the GPS-Simplex system. We will evaluate whether using this system allows us to locate kill sites quickly enough to retrieve a suitable tissue sample to test for CWD. If the animal killed is a deer, the presence of suitable diagnostic samples (brain stem and tonsil tissues) and overall carcass condition will be noted, and tissues will be taken to test for CWD when available. To evaluate the effect of carcass sampling activities on puma behavior, we will alternate taking the entire head of the kill with sampling only the necessary tissues in the field to compare the effect on the puma's return to the kill. The animals will be monitored closely after the kill has been sampled to ensure our handling does not interfere with their return to the kill site. Generally, researchers' presence at and inspection of a kill site does not dissuade a puma from returning to that site (T. Beck, unpublished data). However, if it becomes apparent that one technique is more disruptive than the other, then we will adopt the least disruptive sampling technique for the remainder of the study. Both puma will remain collared for a period of no less than one month unless the collars appear to be adversely affecting them. We will monitor each animal for changes in behavior like decreased kill rates or mobility that may be attributed to the collars. If the collars seem to have no adverse effects on the puma, then they will remain in place until the batteries must be replaced (about 3-4 mo, depending on final programming configuration). If the collars need to be removed for any reason, the same capture and handling methods as described above will be used for recapture. • Data from this pilot study will be used in designing more comprehensive studies ofpuma~eer relationships in Colorado, and may be of use in other studies of predator-prey ecology. Literature Cited Gross, J. E., and M. W. Miller. 2001. Chronic wasting disease in mule deer: A model of disease dynamics, control options, and population consequences. J. Wildl. Manage. In press. Homocker, M. G. 1970. An analysis of mountain lion predation upon mule deer and elk in the Idaho Primative Area. Wildlife Monographs 21. Kreeger, T. J. 1996. Handbook of wildlife chemical immobilization. International Wildlife Veterinary Sciences, Inc. Laramie, Wyoming, USA. Miller, M. W., E. S. Williams, C. W. McCarty, T. R. Spraker, T. J. Kreeger, C. T. Larsen, and E.T. Thome. 2000. Epizootiology of chronic wasting disease in free-ranging cervids in Colorado and Wyoming. Journal of Wildlife Diseases 38:676-690. Shaw, H.G. 1979. Mountain lion field guide. Fourth edition. Arizona Game and Fish, Phoenix, Arizona, USA. � https://cpw.cvlcollections.org/files/original/2fe1f3f8bb3bc1b59bdade9a25849384.pdf 03f83b1b0aa6b61ba3c15936da81fc1c PDF Text Text Colorado Division of Wildlife Wildlife Research Report July 2003 – June 2004 JOB PROGRESS REPORT State of Project No. Work Package No. Task Colorado Federal Aid Project: N/A 3740 : : : : Cost Center 3430 Mammals Research Wildlife Diseases Pilot evaluation of GPS technology in chronic wasting disease prevalence and management at artificial feeding sites in urban areas. : Period Covered: April 1 2003 through July 31, 2004 Author: Eric J. Bergman, Michael W. Miller and L. L. Wolfe Personnel: M. Sirochman All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT A pilot study for assessing the utility of GPS technology in the evaluation of CWD prevalence and management in urban areas was designed is being implemented. Objectives of this pilot study are to: 1) Evaluate the utility of GPS radio collar technology in identifying artificial feed sites in urban settings, 2) Evaluate if there is evidence that artificial feed sites reduce the size of deer home ranges, 3) Evaluate if deer density is elevated at artificial feed sites, and 4) Evaluate if CWD prevalence is higher at artificial feed sites . 119 �JOB PROGRESS REPORT PILOT EVALUATION OF GPS TECHNOLOGY IN CHRONIC WASTING DISEASE PREVALENCE AND MANAGEMENT AT ARTIFICIAL FEEDING SITES IN URBAN AREAS Eric J. Bergman, Michael W. Miller and L. L. Wolfe INTRODUCTION Analyses of data from recent field studies and from culling have revealed areas of relatively high CWD prevalence associated with urban areas along the northern Front Range (Wolfe et al. 2002, 2004; Conner and Miller 2004; Farnsworth et al. 2004). Within these, artificial and illegal feeding sites may be particularly important because they appear to congregate deer in one location, thereby increasing local deer density and exposure to contaminated environments (Miller et al. 2004). Although the nature of the relationship between disease prevalence and mule deer density has not been definitively identified, it seems likely (Barlow 1996) that CWD prevalence is being indirectly elevated through artificial feeding. The development of global positioning system (GPS) technology and its incorporation into radio collars for wildlife research presents a tool for better understanding CWD in urban areas. We have initiated a pilot field study to: 1) evaluate the effectiveness of different GPS collars in identifying illegal feed sites in urban settings, and 2) develop and evaluate a strategy for utilizing GPS technology in studying and managing CWD in urban mule deer populations. METHODS The study area for this work is located within two subdivisions in Estes Park, Colorado. The subdivisions, separated by approximately 1.6 km, were identified as treatment and control sites based on the presence and absence of known feeding sites (Fig.1, Wolfe et al. 2004). Between five and eight adult (>1 yr old) female deer from each subdivision were captured and collared with one of two different brands of GPS collars (HABIT Research, British Columbia, Canada and LOTEK Wireless, Ontario, Canada). Collars from each company will be evenly distributed between sites. Capture will occur as part of an ongoing "test and cull" research project (Wolfe et al. 2004) during April 2004 and from August to October of 2004 as needed. Deer will be recaptured and collars will be removed prior to battery failure (~220 days service) in order to retrieve GPS data. No specific hypotheses are being tested in this pilot study; rather, we are attempting to determine if GPS radio collar technology is adequate for use as a tool in refining CWD epidemiology and management. We will record and report on the performance of GPS collars, and calculate costs (mean, range per animal tested) associated with our artificial feed site identification strategy as implemented in this pilot study. However, we will compare home range sizes of deer from each site to determine if artificial feeding reduces home range size of deer. We will also incorporate ground survey data (Wolfe et al. 2004) to estimate and compare mule deer density and ultimately CWD prevalence from sampled deer at each site. CWD prevalence will be compared between sites as well as to previous estimates from the greater Estes Park area (Wolfe et al. 2004) to explore future research potential. 120 �RESULTS AND DISCUSSION GPS Collar Comparison A total of 16 GPS collars (10 LOTEK, 6 HABIT) were available for testing in this study. Prior to initiation of this study no HABIT collars were on hand for deployment, rather, all 6 had to be built to specification and delivered. GPS collars from HABIT Research, ~$1,800/unit, were programmed to: collect GPS locations every 2 hours, to transmit GPS data (via VHF signal) over two day intervals every two weeks and to transmit the most recent GPS location (via VHF signal) at the start of each minute. Due to delays in the manufacturing process, no HABIT collars were received in time for spring deployment (≥2 weeks pre-fawning). Additionally, due to programming errors, 0 of 6 HABIT collars were ready for deployment after initial testing. Upon servicing by HABIT Research (~3.5 weeks), 3 of 6 collars appear to be ready for deployment in late summer 2004. The remaining HABIT collars (3 of 6) will be serviced and deployed upon satisfactory performance. All LOTEK collars were on hand prior to initiation of this study. Eight of 10 collars were deployed in spring of 2004, with 1 of 10 needing service. GPS collars from LOTEK Wireless, ~$3,500/unit, were also programmed to collect GPS locations every 2 hours, but did not offer remote download capabilities. All GPS locations collected by LOTEK collars will be acquired upon retrieval of the collar. GPS Collar Performance Data from LOTEK GPS collars continues to be collected and HABIT GPS collars will be deployed between August-September 2004. LITERATURE CITED Barlow, N.D. 1996. The ecology of wildlife disease control: simple models revisited. Journal of Applied Ecology 33:303-314. Conner, M.M., and M.W. Miller. 2004. Spatial epidemiology in natural populations: a case study of movement and prion disease prevalence relationships among mule deer population units. Ecological Applications (in press). Farnsworth, M.L., L.L. Wolfe, N.T. Hobbs, K.P. Burnham, D.M. Theobald, and M.W. Miller. 2004. Human land use influences chronic wasting disease prevalence in mule deer. Ecological Applications: in review. Miller, M.W., E.S. Williams, N.T. Hobbs, and L.L. Wolfe. 2004. Environmental sources of prion transmission in mule deer. Emerging Infectious Diseases: in press. Wolfe, L.L., M.M. Conner, T.H. Baker, V.J. Dreitz, K.P. Burnham, E.S. Williams, N.T. Hobbs, and M.W. Miller. 2002. Evaluation of antemortem sampling to estimate chronic wasting disease prevalence in free-ranging mule deer. Journal of Wildlife Mangement 66:564-573. _________, M.W. Miller, and E.S. Williams. 2004. Feasibility of "'test-and-cull" for managing chronic wasting disease in urban deer. Wildlife Society Bulletin 32:500-505. Prepared by ____________________________ Eric J. Bergman, Wildlife Researcher 121 � https://cpw.cvlcollections.org/files/original/2376790d1e8298352a11238991361be7.pdf 0536b4699cd92aed373d8e6d7594a909 PDF Text Text Colorado Division of Wildlife Wildlife Research Report July 2003 – June 2004 JOB PROGRESS REPORT State of Project No. Work Package No. Task Federal Aid Project Colorado 3740 N/A : : : : Cost Center 3430 Mammals Research Mammals Support Services Veterinary Services – Medical Support : Period Covered: July 1 2003 through June 30, 2004 Author: L. L. Wolfe Personnel: M. W. Miller, L. A. Baeten, M. M. Conner, K. Cramer, T. R. Davis, K. Griffin, D. O. Hunter, J. E. Jewell, E. Knox, C. E. Krumm, C. T. Larsen, J. Rhyan, M. Sirochman, T. Sirochman, E. S. Williams, D. Wroe All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. 123 �JOB PROGRESS REPORT VETERINARY SERVICES – MEDICAL SUPPORT L.L. Wolfe INTRODUCTION Veterinary services are provided as support for a variety of wildlife research projects, transplants and reintroductions conducted by the Colorado Division of Wildlife (CDOW) and its collaborators throughout the year. The following overviews and summarizes key wildlife veterinary medical support services provided during 2003−2004. VETERINARY MEDICAL SUPPORT Location of services & primary investigator CDOW Foothills Wildlife Research Facility (FWRF), Tracy Davis and researchers Rocky Mountain Arsenal, Sherry Skipper Species mule deer, white-tailed deer, elk, bighorn sheep, pronghorn, puma, others mule deer, white-tailed deer Uncompahgre Plateau, Chad Bishop mule deer Pinion Canyon maneuver site, Elizabeth Joyce swift fox Colorado Springs, Brian Dreher mule deer CDOW FWRF, Department of Defense contract in cooperation with Elizabeth Williams mule deer, white-tailed deer Type of medical support Preventive, routine, and emergency medical care for all research animals housed at FWRF for use in ongoing CDOW and research. Chemical immobilization of adult does for survival study and CWD surveillance. Does were ultrasounded, tonsil biopsied, blood was collected, and vaginal implant transmitters (VITs) were inserted. Medical care of injured animals, assisted with ultrasound, VIT, and blood collection for viral serosurvey and thyroid study. Swift fox kits were anesthetized and abodminal radiotransmitters were surgically inserted and blood was collection. Adult deer were captured and radiocollared for CWD surveillance. We tonsil biopsied deer, collected blood, and provided area training for future efforts Provided medical care for hand raised deer fawns, including diarrhea outbreak management and treatment of injured fawns. TRAINING Capture and Sampling A capture and handling training class was provided for the district wildlife manager trainees. A second class was held for researchers and biologists. Capture classes included lectures on drug use regulations and recordkeeping, pharmacology of select capture drugs, dosing, safety and types of equipment. These classes also included “hands on” capture and handling of animals at FWRF. This last year, we also devised and administered a written and practical exam for the DWM trainees. As needed, spaces in these classes also provide opportunities for graduate students and technicians to learn sample collection techniques. 124 �TONSIL BIOPSY Tonsil biopsy training sessions were provided for staff from Wyoming Game and Fish Department and the Wisconsin Department of Natural Resources. In addition, CDOW personnel from the Colorado Springs area were trained on-site in capture and sampling techniques. Tonsil biopsy training sessions included lectures on sampling techniques and recognizing signs of CWD. The training also involves “hands on” time in the necropsy lab for sampling techniques. We also provide hands on training, as scheduling allows, on research animals. TARGETED CWD SURVEILLANCE A training module was developed and used to instruct USFWS and tribal biologists in recognizing signs of CWD as a tool in developing targeted surveillance programs on national wildlife refuges and tribal lands. This training included a lecture and PowerPoint tutorial illustrating clinical signs of CWD, as well as first-hand observation of captive mule deer showing signs of CWD. The tutorial file was subsequently modified and made available to other state and federal agencies as a self-teaching tool to aid in respective CWD management programs. TRANSLOCATION During transplant and translocation operations, we provide emergency medical care or humane euthanasia to injured animals. We also provide blood sampling, health exams, health certificates, vaccinations, anthelmintics, and antibiotics as needed to assure safe transport and improve survival in translocated wildlife. Species BHS swift fox black-footed ferrets Lynx Services Provided Vaccination, anthelmintics, antibiotics, pharyngeal swabs Health exams and health certificates Health exams and health certificates Entry and release exams, medical care for capture injuries Comments Transplanted within Colorado Released in South Dakota Released in Utah Reintroduction project* *Lynx reintroduction: Thirty eight lynx were received for the 2004 release. Most of the lynx were in good condition on arrival at the lynx holding facility in Del Norte. Three lynx required digit amputation due to trapping injuries, but all three healed without complications. One female was euthanized due to a compound fracture of the radius-ulna. In 2003/2004, the anesthetic protocol for transported lynx was changed from 2.0-2.5 mg/kg Telazol delivered intramuscularly (IM) to 20-40 mg (0.02 -0.05 mg/kg) ketamine and 0.6-0.8 mg (0.050.11 mg/kg) medetomidine given IM. No adverse anesthetic reactions were seen. Lynx were given the ketamine/medetomidine by IM injection while held in a squeeze cage. Induction time averaged 5.1 minutes (S.E. 0.4). Anesthetic time (induction to reversal) averaged 24.5 minutes (S.E. 0.9). Lynx were given atipamazole (0.25−0.55 mg/kg) in equal volume to medetomidine by IM injection. Lynx were recovered with minimal stimulation in their den boxes. Recovery time (time from reversal to standing with coordination) averaged 48.5 minutes (S.E. 2.5). There were no poor recoveries or anesthetic reactions. This new drug combination offered substantial reduction in processing and recovery times for lynx being handled at different points in the reintroduction process. ® DRUG DISTRIBUTION Since 2002, there has been extensive reorganization of drug distribution procedures and recordkeeping for chemical capture of wildlife. Overall, there has been a dramatic improvement in drug tracking and accountability. This has resulted in a reduction in wasted expired drugs and improvement in field logs. 125 �Telazol summary 800 700 □ Total bottles Telazol bottles 600 ■ bottles prescribed 500 - - 400 - - 300 200 100 0 □ bottles "out" unknown ~ - - ~ - - □ field use logged - ■ unreported field logs - - - - □ bottles returned expired 1- l r- - 2000 2001 2002 2003 ■ bottles in FW RF safe 2004 year CLINICAL TRIALS The following table summarizes the clinical trials from 2003. These trials were designed and conducted to improve veterinary medical care associated with various research and management programs conducted by CDOW. More complete reports of these trials are in the appendices. Clinical Trial Investigators Clostridium perfringens type A vaccine trial Plague vaccine in Canada lynx Wolfe, Miller, Davis, Ellis BHS, MD A Wolfe, Shenk, Baeten, Miller, Roke Wolfe, Miller lynx B mountain lions, MD, WTD C Wolfe, Miller, Lance MD Published Baker, Wolfe, MD Wolfe, Ryan, Miller fallow deer See progress report for Dan Baker D Chemical immobilization field trial with medetomidine and ketamine combination Chemical immobilization with A-3080 in mule deer Chemosterilization with GnRH toxin in mule deer Comparison of dart injection quality between 2 brands of collared and uncollared darts in fallow deer 126 Species Appendix �APPENDIX A EXPERIMENTAL EVALUATION OF A VACCINE FOR CLOSTRIDIUM PERFRENGENS TYPE A IN CAPTIVE BIGHORN SHEEP (Ovis canadensis) AND CAPTIVE MULE DEER (Odocoileus hemionus) L. L. Wolfe, R. P. Ellis, K. Fox, T. Davis, and M. W. Miller INTRODUCTION Clostridium perfringens is found naturally in the intestines of animals and in the environment. This bacterium possesses the ability to produce heat-resistant endospores and potent extracellular toxins. Isolates of C. perfringens can be subdivided into types based on the production of these exotoxins. The four major toxins implicated in disease are α, β, ε, and ι. Of these four major toxins, type A produces α toxin only, type B produces α, β, and ε, type C produces α and β, type D produces α and ε, and type E produces α and ι toxins. Other minor toxins also exist within the five types of C. perfringens, although they are not used to identify the specific type due to overlap between types. These toxins include δ (found in types B and C), θ (found in all five types), κ (found in all five types), λ (found in types B, D, and E), µ (found in types A, B, C, and D) ν (found in all five types), and neuraminidase or sialidase (found in all five types). In addition, C. perfringens enterotoxin (CPE) is often produced. CPE is most often found occurring with type A, although it has also been documented with all five types of C. perfringens. Many toxins produced by C. perfringens organisms are hydrolytic enzymes, necessary for life as a saprobe found naturally in the soil. Type A, the focus of this study, also possesses enzymes with hydrolytic properties, including phospholipase C and sphingomyelinase activities (from the α toxin). (Petit et al. 1999) Clostridium perfingens type A has recently been implicated as a cause of enterotoxemia in a variety of species including lambs and goats. Tympany, hemorrhagic enteritis and abomasitis, and abomasal ulceration in calves characterize the disease. Lesions include necrotic enteritis in domestic chickens; necrotizing enterocolitis and villous atrophy in suckling and feeder pigs; and hemorrhagic gastroenteritis in dogs. (Bueschel et al. 1998). Other reports of enteric disease associated with C. perfringens type A include enterotoxemia in minks, muskrats, and racing camels, acute toxemia in water buffaloes (Songer, 1996), gastroenteritis in black-footed ferrets (Schulman et al., 1993) and dairy cattle (Dennison et al., 2002). The α toxin in type A C. perfringens acts by way of phospholipase C activity and sphingomyelinase activity, breaking down phosphatidylcholine and sphingomyelin found in the membranes of erythrocytes, platelets, leukocytes and endothelial and muscle cells. By way of this action, α toxin is thought to be responsible for the cytotoxicity, necrosis, and hemolysis observed with type A C perfringens. There is evidence suggesting that minor differences in the amino acid sequence of α toxins exist, creating two strains with different pathways of infection. One strain has an increased resistance to chymotrypsin, allowing survival and multiplication in the gut, followed by entry into circulation. This strain is believed to be the primary cause of type A related enterotoxemia. The other strain, lacking a resistance to chymotrypsin, is believed to have a higher affinity for invasion of muscle tissue, and perhaps the cause of type A related gas gangrene (Songer, 1996). Clinical signs of animals suffering from type A C. perfringens vary from species to species, but consistently include depression, anorexia, diarrhea, bloating in non-avian species, and death. Postmortem findings from these cases varies from species to species and between specific cases, but consistently tend to include gram positive bacilli surrounding necrotic tissue, necrosis, particularly in the small intestine, 127 �and hemorrhage and ulceration, again in the small intestine; in ruminants, abomasitis, tympany, and abomasal hemorrhage and ulceration are also common findings. Infection by type A C. perfringens is believed to occur in a variety of ways. One theory, especially in neonatal ruminant cases, is that engorgement on milk or esophageal groove dysfunction allows milk to spill over into the rumen, providing a substrate for growth of the bacterium, as well as an anaerobic environment in which to proliferate. Another suggested scenario, as found in cattle herds in Nebraska and Wyoming is that bacterial infection is secondary to copper deficiency. The findings of this study indicated that low copper concentrations may have weakened the abomasal mucosa and compromised immune function (Roeder et al., 1988). Environmental contamination may play a role in the acquisition of C. perfringens type A because these toxins are known to exist in the soil and many ruminant species ingest soil in attempts to acquire essential minerals. In addition, α toxin can be detected in the feces of birds with necrotic enteritis (Bueschel et al., 1998), and thus avian vectors may provide an additional method of toxin movement. The presence of C. perfringens type A at the Foothills Wildlife Research Facility (FWRF) in Ft. Collins, Colorado appears to be relatively recent, with the first case identified in 1997. Since then, the number of cases has increased exponentially, approximately doubling each year. A total of 29 cases had been attributed to C. perfringens type A since the first case was diagnosed in 1997. At the FWRF, the disease has affected primarily bighorn sheep and mule deer; these 2 species account for 25 of the 29 cases. In adult animals, bighorn sheep have been the primary species affected (6 out of 10 cases), and sudden death has been common. These animals often exhibited bloating and diarrhea shortly before death, and showed signs of hemorrhage such as bleeding from the mouth or anus. Necropsies of these animals consistently included large amounts of rod-shaped bacteria, especially in the small intestines. Other lesions included necrosis and hemorrhage, particularly in the heart and small intestine as well as intestinal and abomasal ulcers. In neonatal and juvenile animals (<1 yr old), mule deer have been the primary species affected (14 of 19 cases), and chronic symptoms have been most common. These animals consistently exhibited chronic bloating, emaciation, depression, and soft brown diarrhea that was sometimes chronic, usually present early on in the animal’s life, and often never alleviated despite various treatment attempts. Therapies included a barrage of antibiotics (benzathine penicillin and florfenicol appeared most effective), subcutaneous fluids, transfaunation, kaolin pectin with lactobacillus granules, probios powder, electrolytes, and medicated “Deccox” feed distributed by Ranchway Feeds. Despite therapy, most of these cases ended in death -- those that survived exhibited symptoms that were short-lived and often only exhibited a single case of bloating and/or diarrhea. Necropsies of affected animals consistently included: abomasitis; hemorrhage and ulcers in the intestine, abomasum, and lungs; fluid and gas throughout the intestines; watery to frothy green fluid in the rumen, and sometimes extending into other stomachs; necrosis; and rod-shaped bacteria in the abomasum and/or small intestine. Of the 19 neonatal cases, 5 occurred in bighorn sheep, and it is noteworthy that these cases were primarily in lambs born late in the spring, after the majority of lambing had already occurred. These cases were similar to those occurring in mule deer neonates. Mortality caused by C. perfringens type A is a growing impact on FWRF operations and ongoing research: it is the leading cause of death in captive bighorns and second only to chronic wasting disease (CWD) as cause of death in mule deer. Moreover, because infections occur primarily in juvenile animals, many long-term studies (e.g., CWD and fertility control) have been hampered by lack of available animals for planed experiments. Here, we proposed to develop and evaluate efficacy of a vaccine to prevent C. perfringens type A morbidity and mortality in captive bighorn sheep and mule deer. 128 �METHODS Initial vaccine development was pursued in Dr. Robert Ellis’ laboratory in the Department of Microbiology, Immunology, and Pathology at Colorado State University. We used captive Rocky Mountain bighorn sheep (O. canadensis canadensis) and mule deer (Odocoileus hemionus) in this experiment. All animals were housed at the CDOW's Foothills Wildlife Research Facility (FWRF) throughout the study and resided in 3-7 ha pastures. In addition to natural forage, grass/alfalfa hay mix and a pelleted high-energy supplement was provided as prescribed under FWRF feeding protocols for bighorn sheep and mule deer in respective age/sex classes throughout the study; fresh water and mineralized salt blocks was provided ad libitum. The general health of all animals was evaluated immediately after vaccination, as well as daily thereafter, and observations recorded throughout our experiment. Injection sites were also examined weekly for 4 weeks after vaccine administration to assess local reactions to vaccine. Bighorn sheep (n = 19) and Mule deer (n = 10) were randomly assigned to vaccinated or unvaccinated groups. The vaccinated group was injected IM with the C. perfringens vaccine in the right hind leg on day 0 and in the left hind leg 4 weeks later (booster). Blood was collected prevaccination, at the booster injection and 4 weeks after the final booster. The control group was weighed and blood was drawn at the time of the vaccine group’s booster and 4 weeks after the final booster. Serum was separated and stored frozen until it was submitted to Colorado State Diagnostic Lab for antibody titer using enzyme-linked immunosorbent assay (ELISA). As necessary deer were sedated with xylazine HCl (5-20 mg IV or 25-100 mg IM) or immobilized with a cocktail of thiafentanil HCl (8-10 mg), or ketamine HCl (100 mg), and xylazine HCl (20 mg), delivered IM by projectile syringe, to facilitate collections; narcotic effects were be reversed with naltrexone HCl (150 mg SC + 50 mg IV). RESULTS There were no vaccine site reactions or adverse effects from vaccination observed. No serum neutralizing antibody titers to C. perfringens were seen in either the BHS or MD. On follow up evaluation of the vaccine by Colorado State Diagnostic Lab, there was no type A antigen in the vaccine. DISCUSSION The vaccine in this study failed due to lack of quality control by the manufacturer, however, we anticipate that a safe and effective vaccine can be readily developed, and that its incorporation into FWRF’s preventive animal health program will reduce morbidity and mortality associated with C. perfringens type A infection. Managing clostridial enteritis is essential to improving success of preventative health programs at the FWRF and minimizing impacts on planned and ongoing research. 129 �LITERATURE CITED Bueschel, D., R. Walker, L. Woods, J. Kokai-Kun, B. McClane, J. G. Songer. 1998. Enterotoxigenic Clostridium perfringens type A necrotic enteritis in a foal. J. Am. Vet. Med. Assoc. 213(9) 1305—1307. Dennison, A. C., D. C. VanMetre, R. J. Callan, P. Dinsmore, G. L. Mason, R. P. Ellis. 2002. Hemorrhagic bowel syndrome in dairy cattle: 22 cases (1997—2000). J. Am. Vet. Med. Assoc. 221 (5) 686—689. L. Petit, M. Gibert and M. R. Popoff. 1999. Clostridium perfinrgens: toxinotype and genotype. Trends in Microbiology. 7 (3) 104—110. Roeder, M. M. Chengappa, T. G. Nagaraha, T. B. Avery, G. A. Kennedy. 1988. Experimental induction of adominal typany, abomasitis, and abomasal ulceration of intraruminal inoculation of Clostridium perfringens type A in neonatal calves. Am. J. Vet. Res. 49 (2) 201—207. Songer, J. Glenn. 1996. Clostridial enteric diseases of domestic animals. Clinical Microbiology Reviews. 9 (2) 216—234. Tillotson, K., J. Traub-Dargatz, C. E. Dickinson, R. P. Ellis, P. S. Morley, D. R. Hyatt, R. J. Magnuson, W. T. Riddle, M. D. Salman. 2002. Population-based study of fecal shedding of Clostridium perfringens in broodmares and foals. J. Am. Vet. Med. Assoc. 220 (3) 342—348. 130 �APPENDIX B SAFETY AND EFFICACY OF RECOMBINANT F1-V FUSION PROTEIN VACCINE TO PROTECT LYNX FROM PLAGUE L. L. Wolfe1, T. E. Rocke2, S. M. Dieterich3, T. M. Shenk1, A. M. Friedlander4, and M. W. Miller1 1 Colorado Division of Wildlife, Wildlife Research Center, 317 West Prospect Road, Fort Collins, Colorado 80526-2097, USA; 2U.S. Geological Survey, Biological Resources Division, National Wildlife Health Laboratory, 6006 Schroeder Road, Madison, Wisconsin 53711, USA; 3Frisco Creek Wildlife Rehabilitation Center, POB 488, Del Norte, Colorado 81132-0002, USA; 4U.S. Army Medical Research Institute of Infectious Diseases, Bacteriology Division, Fort Detrick, Frederick, Maryland 21702, USA. INTRODUCTION Plague, caused by Yersinia pestis, was introduced into the North American continent in the early 1900s, and its impacts on some native wildlife species since that time have been substantial (Cully 1993, Wuerthner 1997, Gasper and Watson 2001). Epidemics in prairie ecosystems have been well documented, and probably contributed to the marked declines observed in both prairie dogs (Cynomys spp.) and black-footed ferrets (Mustela nigripes) over the last century (Cully 1993). Although less extensively studied, it seems likely that sylvatic plague has impacted other wildlife species as well (Gasper and Watson 2001). Canada lynx (Lynx lynx) resided in Colorado historically (Fitzgerald et al. 1994), but apparently were extirpated by the late 1970s. Whether plague played any role in the disappearance of lynx from Colorado is not known. Regardless of plague’s role in the historical decline, this disease now appears to be an obstacle to ongoing efforts to reestablish lynx in southwestern Colorado. To date, Y. pestis infections have been confirmed in 6 Colorado lynx. Plague was the primary cause of death in 27% (4/15) of the known natural deaths and possibly contributed to 1 of the 6 known hit-by-vehicle deaths in adult lynx released in Colorado since 1999 (Wild 2000, Shenk 2003; T. M. Shenk, Colorado Division of Wildlife, unpublished data). Plague also killed at least 1 kitten born in the wild during the first year of documented natural reproduction in Colorado’s reintroduced lynx population (T. M. Shenk, Colorado Division of Wildlife, unpublished data). Practical tools for preventing plague in reintroduced lynx could benefit species recovery efforts in Colorado and perhaps elsewhere. Effective vaccines for preventing plague in mammalian species, including felids, have been developed only recently (Heath et al. 1998, Gasper and Watson 2001, Creekmore et al. 2002). Of these, a recombinant capsular F1-V fusion protein vaccine (Heath et al. 1998) has shown a promising combination of safety and efficacy in black-footed ferrets (Rocke et al. in press), and could be useful in lynx restoration as well. Here, we propose to (1) evaluate F1-V vaccine in captive lynx being held in southwestern Colorado prior to release as part of an ongoing restoration program and (2) compare number of lynx mortalities caused or complicated by plague in vaccinated and unvaccinated lynx released in Colorado. METHODS Our study was conducted in conjunction with the 2004 release program. All lynx were captured, transported, held, cared for, and handled as described in established protocols for Colorado’s restoration program (Wild 2000). Lynx were held at the Frisco Creek Wildlife Rehabilitation Center (FCWRC) prior to and throughout the study until release. Whenever possible, vaccination and sampling was done in 131 �conjunction with other handling activities to minimize stress that could arise from repeated handling of captive lynx. We initially evaluated safety and efficacy of F1-V vaccine (U.S. Army Medical Research Institute of Infectious Diseases, Fort Detrick, Frederick, MD) in 10 adult lynx; 10 age- and originmatched lynx will remain unvaccinated as controls. Blocks will consist of age and origin: age will be either ≤ 5 years old or ≥ 6 years old; origin will be either British Columbia (where prior exposure to plague is possible) or Manitoba/Quebec (where prior exposure is unlikely). We estimated ages based on tooth wear; animals ≤1 year old were excluded. Within each block (age and origin) of lynx, half were selected at random to receive the vaccine while the remainder will serve as controls. Vaccine was be diluted and combined with Alhydrogel adjuvant (United Vaccines, Madison, WI) as described by Rocke et al. (in press). We administered vaccine via subcutaneous (SQ) injection in the hindquarter on day 0 and a second dose was given 21 days later. Initial vaccine doses were delivered by hand-held syringe when lynx are examined upon entry into FCWRC; booster doses were delivered via hand-held syringe while the lynx was restrained in a squeeze cage. Vaccinated lynx were observed immediately after vaccination, immediately upon recovery from anesthesia (when applicable), and daily thereafter for evidence of adverse effects. To evaluate serological responses of vaccinated lynx as an index of efficacy, we will collected blood (~6 ml) from all captive lynx at each handling during the 2004 season regardless of vaccination status. For the 10 principal vaccinates and controls, at minimum blood will be collected on day 0 and again 42 days later (21 days after the booster vaccination). Serum was harvested and stored frozen until assayed. We will measure antibody titers against F1 and V antigens with phytohemagglutinaiton assay (PHA) at the Center for Disease Control and an enzyme-linked immunosorbent assay (ELISA) using methods of Rocke et al. (in press). For the 10 principal vaccinates and controls, we compared changes in log10 anti-F1 and anti-V antibody titers-1 .. Mortality of vaccinated lynx due to or complicated by plague will be compared to mortality due to or complicated by plague of unvaccinated lynx from previous releases. A suite of models developed a priori will be evaluated through AICc model selection (Burnham and Anderson 2002) to investigate the possible effects of vaccination status, age, location of birth, and time to death on mortality of lynx due to or complicated by plague. RESULTS All PHA prevaccination titers were negative. All vaccinated lynx showed seroconversion on the PHA assay at the after the first and second booster (figure 1.). ELISA results are pending. There were no vaccine site reactions and no adverse side effects were seen. 132 �Vaccine Response 10000 PHA antibody titer □ pre 1000 100 ~ ~ - post 1st dose □ post 2nd dose - - - • - ~ ... 10 1 ~ ~ - - ~ ~ ,- - - ,- ~ ... - QF2 QF3 QF4 QF6 QF7 BF1 animal id BF2 BF3 BF4 BF6 Figure 1. PHA antibody titer for individual lynx. All pre titers were negative. All lynx showed seroconversion following vaccination. DISCUSSION Lynx were examined on entry and 5 females from Quebec and 5 females from British Columbia were randomly chosen for vaccination with F1-V fusion protein plague vaccine. All pre vaccine PHA titers were negative. All vaccinates showed seroconversion but the quantitative titer assays are still pending. The vaccine appears to be safe in lynx; there were no vaccine site reactions or adverse systemic reactions. LITERATURE CITED Burnham, K. P. and D. R. Anderson. 2002. Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach. Second edition. Springer-Verlag. New York. Cully, J. F. 1993. Plague, prairie dogs, and black-footed ferrets. In Management of prairie dog complexes for the reintroduction of the black-footed ferret, J. L. Oldemeyer, D. E. Biggins, B. J. Miller, and R. Crete (eds.). U.S. Fish and Wildlife Service, Biological Report 13, Washington, D.C., pp. 38– 49. Creekmore, T. E., T. E. Rocke, and J. Hurley. 2002. A baiting system for delivery of an oral plague vaccine to black-tailed prairie dogs. Journal of Wildlife Diseases 38: 32–39. Fitzgerald, J. P., C. A. Meaney, and D. M. Armstrong. 1994. Mammals of Colorado. Denver Museum of Natural History and University Press of Colorado, Denver, Colorado, pp. 368–371. Gasper, P. W., and R. P. Watson. 2001. Plague and yersiniosis. In Infectious diseases of wild mammals, 3rd edition, E. S. Williams and I. K. Barker (eds.). Iowa State University Press, Ames, Iowa, pp. 313–329. Heath, D. G., G. W. Anderson, Jr., J. M. Mauro, S. L. Welkos, G. P. Andrews, J. Adamovicz, and A. M. Friedlander. 1998. Protection against experimental bubonic and pneumonic plague by a recombinant capsular F1-V antigen fusion protein vaccine. Vaccine 16: 1131–1137. 133 �Rocke, T. E., J. Mencher, S. R. Smith, A. M. Friedlander, G. P. Andrews, and L. A. Baeten. Recombinant F1-V fusion protein vaccine protects black-footed ferrets (Mustela nigripes) avainst virulent Yersinia pestis infection. Journal of Zoo and Wildlife Medicine, in press.Shenk, T. 2003. Species conservation: Colorado’s lynx. Lynx Home Page, Colorado Division of Wildlife, Denver, Colorado. Accessed 23 October 2003 at http://wildlife.state.co.us/species_cons/lynx.asp. Wild, M. A. 2000. Lynx veterinary services and diagnostics. Federal Aid: Wildlife Research Report for the Colorado Division of Wildlife, pp 47-62. Wuerthner, G. 1997. Viewpoint: The black-tailed prairie dog ⎯ headed for extinction? Journal of Range Management 50: 459–466 APPENDIX C EFFICACY OF KETAMINE MEDETOMIDINE COMBINATION IN MOUTAIN LIONS (Puma concolor) , MULE DEER (Odocoileus hemionus) AND WHITE-TAILED DEER (Odocoileus verginianus) FOR CHEMICAL IMMOBILIZATION IN THE FIELD L. L. Wolfe, W. R. Lance and M. W. Miller In cooperation with Wildlife Pharmaceuticals, Inc. (Fort Collins, CO) we are using ketamine (200mg/ml) and medetomidine (20 mg/ml) compounded at a higher concentration than commercially available. By concentrating the drugs we are able to use an effective dose in a 1 cc dart for mountain lions and a 2 cc dart for deer. To date over 200 deer have been captured and 5 mountain lions using this combination. No adverse side effects have been seen. Evaluation of this drug combination for field capture is ongoing. 134 �APPENDIX D EVALUATION OF COLLARED AND UNCOLLARED DANINJECT AND PNEUDARTS L. L. Wolfe, D. M. Okeson, W. R. Lance, J. Rhyan, and M. W. Miller INTRODUCTION Remote delivery systems, powered by compressed CO2 or blank charge, are an important tool for wildlife immobilization. Darts used with these remote delivery systems are barbed, collared or have uncollared needles. The purpose of the barbs and collars are to hold darts in place long enough to ensure complete drug delivery. However, barbed darts can only be used to deliver anesthetics, thus allowing for dart retrieval. Collared darts and uncollared darts will fall out on their own, but drug delivery may be incomplete. Drugs are delivered from the darts with a powder charge (Pneu-dartTM, Williamsport, PA) or pressurized air (DaninjectTM, Wildlife Pharmaceuticals, Fort Collins, CO). Some level of trauma is inherent, and varies greatly with the type of dart used (Valenburg et al. 1999, Kreeger 2002). Powder charged darts deliver drug rapidly, but are potentially more traumatic than the air pressurized darts; consequently, induction times vary when these darts are used to deliver anesthetic drugs. In this study we compared drug delivery between collared and uncollared darts. We compared dart trauma between collared and uncollared darts and we comopared Daninject darts with Pneu-darts. METHODS This study was conducted in conjunction with a previously-approved terminal study testing fallow deer susceptibility to chronic wasting disease (CWD) (CDOW ACUC 12-2000). Because these animals were already slated for euthanasia, we opportunistically evaluate and compare trauma and drug delivery associated with the respective dart types immediately prior to euthanasia. Deeply anesthetized fallow deer were placed on a stand to facilitate darting the hindquarters. Each animal was darted in the hindquarter with a collared Daninject dart and Pneu-dart dart on one side and uncollared Dainject dart and Pneu-dart dart on the opposite side. Darts were loaded with 2 cc India ink. The animals were then be euthanized by intravenous injection of Euthansol (8.8 mg/kg). Each dart site was evaluated for amount of India ink leakage at dart site, degree of trauma (recorded on a scale of 03. (0= none, 3 = extensive hemorrhage and tissue disruption) and ink injection pattern. All fallow deer were be captured with thiafentanil oxalate (0.1 mg/kg) delivered intramuscularly (IM) via projectile syringe using an adjustable air-powered rifle and xylazine hydrochloride. Anesthetic drugs were delivered to the shoulder and neck to avoid confounding subsequent assessment of dart effects. RESULTS On necropsy the ink pattern at the injection site was evaluated. A common ink pattern noted at necropsy was a “T”shaped pattern. The superficial area of ink (top of T) averaged 20-25 mm in diameter (recorded on line #4- spread of ink in muscle). Note that this refers only to the most superficial spread (horizontally) of the ink in the muscle; ink forming the top part of the T was typically only 2 mm thick. Ink then typically followed the needle track 20-30 mm into muscle. The pattern of “focal” was often recorded, but would probably be more correctly stated as “along needle track” as these 20-30 mm deep tracks were typically only 2 mm wide (roughly the diameter of a needle). 135 �Most injection sites had minimal hemorrhage or trauma this was recorded on a scale of 0-3. (0= none, 3 = extensive hemorrhage and tissue disruption) Overall, most dart sites were rated a “0” (12 of 32 darts) or a “1” (7 of 32 darts). Uncollared PneuDarts Eight of 8 uncollared PneuDarts bounced off the animal immediately after impact. However, ink was delivered into the animals, in all but one case. In 5 cases darts bounced, but no ink was noted on the hair of the animal or was observed spraying from the bounced dart. At necropsy, ink was noted in muscle in all of these cases, indicating that the bounced darts did deliver ink prior to ejecting from animal. One dart that bounced was noted to have sprayed a large amount of ink. However on necropsy, ink was found in muscle, indicating that dart did deliver at least some of the ink into the muscle. One dart bounced was noted as “did inject”, but some ink was noted on hair around injection site. At necropsy, ink was found in muscle, indicating that dart did deliver at least some of the ink into the muscle. One dart bounced was noted as “did inject”, but there was ink running down from the injection site. At necropsy, it was hard to distinguish any ink from bruising, so this dart may have not injected any ink into the muscle. There were 2 of 8 animals scored as “3” (extensive hemorrhage and tissue disruption). However there were also 3 of 8 animals scored as a “0” (no hemorrhage and tissue disruption.). In addition 2 of 8 were NR (not recorded). It is difficult to draw a conclusion as to whether or not there is a tendency for this type of dart to cause more tissue damage. Fallow deer dart study 4/16/03 - Results for smooth Pneudart (symbol +) Deer # Dart stayed Ink on in muscle? hair 1102 N 1 202 N NR 302 N 3 2002 N 1 1002 N 1 Depth of ink in muscle 5 mm Pattern of ink in muscle Superficial spread of ink on muscle Hemorrhage/ trauma 25 mm NR ? 2 unknown 0 ? NR Unable to distinguish ink from hemorrhage/bruising. Did dart inject? 20 mm located 20 mm deep 5 mm deep, focal 20 mm 15 mm focal spot located 20 mm deep dissecting on tendon sheaths 1802 N 0 2502 N 0 < 2 mm only superficially delivered 1202 N 1 20 mm NR spot located 20 mm deep focal, but no needle track 5 mm 60 mm superficial spot plus spot located 20 mm deep into muscle with a 20 mm diameter 20 mm Fits typical T pattern of ink spread? 3 N 0 N 3 Y 0 N NR ? Collared PneuDarts All 8 collared PneuDarts stayed in the animal after impact. In 3 of 8 cases the darts delivered ink only very superficially (not deep in muscle). Overall the collared darts caused very little trauma. There were 2 of 8 animals with a rating of “0”, and 2 of 8 with a rating of “1”. Only 1 of 8 animals had a rating of either “2” or “3”. Not recorded = 2 of 8. 136 �Fallow deer dart study 4/16/03 - Results for collared Pneudarts (symbol #) Deer # Dart stayed Ink on in muscle? hair Depth of ink in muscle Pattern of ink in muscle Superficial spread of ink on muscle Hemorrhage/ trauma only superficially delivered 20 mm 2 5 mm 3 Not sure of pattern 0 N; Ink only superficially delivered with spread along superficial fascial planes 1102 Y 0 "very superficial" (?<2mm) 202 Y 0 20 mm deep, focal 70 mm Fits typical T pattern of ink spread? N; Ink only superficially delivered, minimal muscle penetration of ink. 302 Y 0 < 2mm 70 mm superficial "splotch", dissects along fascial planes 2002 Y 0 20 mm focal 20 mm 1 Y 1002 Y 0 30 mm focal 5 mm NR Y 1802 Y 0 35 mm focal 5 mm 1 Y 2502 Y 0 < 2mm only superficially delivered 60 mm 0 N; Ink only superficially delivered, minimal muscle penetration of ink. 1202 Y 0 15 mm NR 20 mm NR ? Uncollared Dan-Inject Darts Seven of 8 uncollared Dan-Inject darts stayed in the animal’s muscle after impact. The result of one uncollared Dan-Inject dart was not recorded. Overall, the uncollared Dan-inject darts caused very little hemorrhage or trauma. Four of 8 animals had a rating of “0” hemorrage/trauma rating and 2 scored 1 and 1 scored 2. Fallow deer dart study 4/16/03 - Results for smooth DanInject darts (symbol *) Deer # Dart stayed Ink on Depth of ink Pattern of ink in muscle? hair in muscle in muscle 1102 Y 0 < 2mm 202 Y 0 0 Superficial spread of ink on muscle very superficial no ink injected into muscle Hemorrhage/ Fits typical T pattern trauma of ink spread? 15 mm 1 2 smooth Daninjects hit animal. 1st "penetrated leg, injection out back side of leg". 2nd dart also recorded as Yes stayed in muscle & 0 ink on hair; but not sure if depth, pattern, and spread info is for 1st or 2nd dart. ??? 0 0 N; dart went deep into limb but injected ink out medial aspect 302 Y 0 30 mm deep, focal 20 mm 1 Y 2002 Y 0 50 mm deep, focal 5 mm 0 Y 1002 NR 0 15 mm focal 5 mm 0 Y 1802 Y 0 50 mm NR (not recorded) 20 mm 0 Y? 2502 Y 3 75 mm 1202 Y 0 0 NR 10 mm 2 1st smooth Daninject dart went through leg, injected some out caudal aspect (3 for ink on hair refers to ink on back of limb from 1st dart). 2nd dart recorded as - stayed in muscle, no ink on hair; but not sure if 75 mm & 10 mm is for 2nd dart or 1st dart??? only superficially delivered 20 mm NR N; Ink only superficially delivered, no muscle penetration of ink. Collared Dan-Inject Darts Seven of 8 collared Dan-Inject darts stayed in the animal’s muscle after impact. The result of one collared Dan-Inject dart was not recorded. These darts shows a tendency to cause little to no tissue damage. There were 6 of 8 animals with a rating of either “0” or “1”. These darts show a tendency to cause little to no tissue damage. There were 3 of 8 animals with a rating of “0”, and 3 of 8 with a rating of “1”. Only 1 of 8 animals had a rating of “3”. 137 �Fallow deer dart study 4/16/03 - Results for DanInject collared darts (symbol @) Deer # Dart stayed Ink on in muscle? hair Depth of ink Pattern of ink in muscle in muscle Superficial spread of ink on muscle Hemorrhage/ Fits typical T pattern trauma of ink spread? 1102 Y 0 5 mm focal 15 mm 1 Y 202 Y 0 2 mm deep, focal 10 mm 0 N, basically round superficial, spot 302 Y 0 20 mm diffuse ~15 mm 10 mm "deep" (?) 1 ? Not sure of pattern 2002 Y 0 5 mm 3 ? Not sure of pattern 1002 NR 1 25 mm deep, focal 25 mm 1 ? Not sure of pattern 1802 Y 0 40 mm deep, focal 5 mm 0 Y? may be 5 mm wide all the way down 2502 Y 0 10 mm deep, focal 25 mm 0 Y 1202 Y 0 35 mm NR (= not recorded) 35 mm NR ?; Note 0.5 mL of ink left in dart deep 15 mm dissects btw muscle masses DISCUSSION All of the collared darts stayed in the muscle (did not bounce). All of the Daninject uncollared darts also stayed in the muscle. Only the uncollared pneudarts bounced out of the muscle. There was no inject sprayed on the hair of animals darted with collared pneudarts. Only one animal in each group of animals darted with daninjects had ink sprayed on the hair. Five of the animals darted with uncollared pneudarts had ink sprayed on the hair. LITERATURE CITED Kreeger, T. J. 2002. Analyses of immobilizing dart characteristics. Wildlife Society Bulletin. 30(3) 968−970. Valkenburg, P., R. W. Tobey, and D. Kirk. 1999. Velocity of tranquilizer darts and capture mortality of caribou calves. Wildlife Society Bulletin. 27(4) 894−896. Prepared by _______________________________ Lisa L. Wolfe, Veterinarian 138 � https://cpw.cvlcollections.org/files/original/67b18b9a50ae62c96bbae585d027c7e0.pdf 5f4009aa19fc000085191acb4d186423 PDF Text Text 133 JOB PROGRESS REPORT State of -------=C'--"o""'lo=r=a=do=------- : Division of Wildlife - Mammals Research Work Package No. ----=3-'-7--'-4-=-0_ _ _ _ _ : Chronic Wasting Disease and Other Wildlife Disease Management Task _ _ _ _ _ _ _ _ _1_ _ _ _ _ _ _ : Chronic Wasting Disease in Mule Deer Monitoring & Management Period Covered: July 1 2002 through June 30, 2003 Author: Michael W. Miller and L. L. Wolfe Personnel: L.A. Baeten, T. H. Baker, M. M. Conner, K. Cramer, T. R Davis, V. Dreitz, C. P. Hibler, N. T. Hobbs, E. Hoover, D. 0. Hunter, E. Knox, C. E. Krumm, C. T. Larsen, N. Mier, B. E. Powers, J. Rhyan, C. J. Sigurdson, T. R Spraker, K. Taurman, E. S. Williams, D. Wroe Interim Report- Preliminary Results This work continues, and precise analysis ofdata has yet to be accomplished. Manipulation or interpretation of these data beyond that contained in this report should be labeled as such and is discouraged. ABSTRACT We continued conducting research on various aspects of chronic wasting disease (CWD) epidemiology and management. Here, we report progress in ongoing and recently-completed work. Studies focused on improving and expanding surveillance in free-ranging populations, understanding and modeling transmission mechanisms, identifying ecological and anthropogenic factors that may influence epidemic dynamics, and evaluating and applying alternative diagnostic and control strategies. In addition to preliminary findings reported here, eight original studies, as well as one review article, were published publication during this segment; citations are appended to the report. INTRODUCTION We continued conducting research on various aspects of chronic wasting disease (CWD) epidemiology and management. Some parts of this work were conducted in collaboration with investigators at Colorado State University, the University of Wyoming, and elsewhere. Specific projects were supported with various combinations of funds from the Colorado Division of Wildlife (CDOW), Federal Aid in Wildlife Restoration Project W-153-R, the U.S. Department of Agriculture, and National Science Foundation/National Institutes of Health Grant DEB-0091961. �134 METHODS Our work on CWD is both multidisciplinary and multifaceted, but broadly falls under the topics of "epidemiology and management" or "pathogenesis and diagnosis". For simplicity, we describe progress under those headings below. STUDIES OF CWD EPIDEMIOLOGY & MANAGEMENT We continued or initiated studies related to surveillance, transmission mechanisms, epidemic trend forecasting, potential host range and strain variation, risk factors, and management tools and feasibility as aids to understanding and controlling CWD in free-ranging deer and elk in Colorado. Statewide surveillance: The discovery of CWD in northwestern Colorado in January 2002 created a sudden demand for both more widespread surveillance and more rapid turnaround on laboratory results. Consequently, the CDOW's CWD surveillance program was overhauled and its capacity greatly expanded over the summer of 2002 in order to meet anticipated demands for surveillance data, as well as to meet policy-based decisions to provide carcass quality assurance information for individual hunters. The most notable changes were the addition of three regional submission laboratories, streamlining of tissue sampling methods, and incorporation of a rapid screening test for CWD diagnosis. Details of overall programmatic features and changes were described on a new CWD-oriented CDOW web page (http://wildlife.state.co.us/CWD/index.asp): details of the evaluation of modified sampling and testing procedures are described below. Transmission mechanisms: We summarized findings on empirical evidence of animal-animal transmission of CWD and the relative importance of this mechanism in epidemic dynamics. We also completed an experiment comparing the relative contributions of live animals, contaminated environments, and infected carcasses to CWD transmission. In this study, 34 free-ranging mule deer from two sources distant to known endemic foci of chronic wasting disease (Rocky Mountain Arsenal National Wildlife Refuge, US Air Force Academy) were captured for use as experimental subjects during MarchMay 2002. We transported these deer to the Colorado Division of Wildlife's Foothills Wildlife Research Facility (FWRF), where they were placed in paddocks (n = 3 replicates/exposure route; n = 3 deer/paddock). Exposure treatments were: confinement in paddocks housing naturally-infected deer ( l infected deer/paddock), confinement in paddocks where infected deer previously resided, and confinement in paddocks where carcasses from CWD-infected deer have decomposed in situ ( l carcass/paddock); unexposed control paddocks are also being maintained. Entire paddock groups will be sacrificed and examined at the first sign of CWD in any subject deer within a paddock. We compared infection rates within and among treatments to examine which of these may contribute to perpetuation of CWD epidemics. Modeling epidemic dynamics in captive mule deer: Developing detailed, temporally dynamic models of CWD in wild populations is a pressing management need, but available field data are presently insufficient to clearly reveal natural trends in ongoing epidemic dynamics. Moreover, there are several plausible ways to model CWD transmission mechanisms, yet field data will likely not provide sufficient resolution for discerning the most appropriate representation. To begin understanding how to best model CWD transmission, we have undertaken a model selection exercise using a time series of data on prevalence on CWD in captive mule deer. We assembled 26 years of data (1974-2000) from CDOW's Foothills Wildlife Research Facility. These data are being used to evaluate strength of evidence for a set of candidate models involving indirect and direct transmission, as well as with and without latency �135 periods. Estimates of transmission rates derived from these models will provide an upper bound on what could be expected in wild populations and will guide construction of candidate sets for modeling those populations. Host range and strain variation: We continued a series of experimental studies in cattle, fallow deer, and mountain lions to explore potential host range of CWD after intense but natural exposure; these experiments compliment ongoing surveillance for evidence of infection in species not known to be natural hosts of CWD, including moose, mountain lions, and cattle. We also continued work looking for evidence of strain variation in CWD agent from various deer sources using domestic ferrets as a laboratory model. Effects of land use on prevalence: Because land-use changes are likely to shape the spatial and temporal dynamics of CWD, as well as options for its management, we have been working to understand the effect ofland use on patterns of CWD prevalence in free-ranging mule deer. We conducted a study to determine whether CWD prevalence in urban areas is higher than prevalence in non-urban areas. We categorized two land use types: urban areas contained==:, 1 housing unit/IO acres and non-urban areas (e.g., ranch, state, and federal lands) contained< 1 housing unit/10 acres. We compared CWD prevalence between land use types in 3 study areas in northern Colorado (Glacier View Meadows [GVM], Horsetooth [HT], Estes Park [EP]) in which urban and non-urban areas were juxtaposed. In each study area, we delineated urban areas surrounded by a 1-2 km buffer and non-urban areas concentric to the buffer. Deer were sampled in approximately equal numbers from the two land use categories. We used a combination of data collected from mule deer sampled via postmortem (Miller et al., 2000, J. Wildl. Dis. 36:676-690; Miller & Williams, 2002, Vet. Rec. 151:610-612) and antemortem (Wolfe et al., 2002, J. Wildl. Manage. 66:564-573) methods described previously; our target was 210 samples for each land-use category, which provided the ability to detect I 0% differences in prevalence between categories with 90% probability at the 0.05 confidence level. We assumed sampling was normally distributed and tried to balance sampling equally among study areas. Selective predation upon infected mule deer: To test for evidence of selective predation, we began a study to compare prevalence of CWD among puma-killed mule deer to prevalence among mule deer harvested or randomly culled by humans within home ranges of collared mountain lions. Sample size calculations were based on the number of deer samples needed to detect two-fold differences in CWD prevalence: we assume that if the mountain lions are showing selectivity for deer with CWD, then the prevalence in the deer killed by mountain lions will be at least twice the prevalence of CWD in the local deer population. Telemetry-marked mountain lions are being monitored and, when available, brainstem (medulla oblongata at the level of the obex), retropharyngeal lymph nodes, and tonsils are collected from pumakilled mule deer carcasses; where none of these tissues are available, we will try to locate and sample other lymphoid tissues (e.g., submandibular or mesenteric lymph nodes, Peyers patches, etc.). Representative subsamples of collected tissues will be fixed in 10% neutral buffered formalin, and the remainder stored frozen. Tissues will be evaluated for presence of PrPewo accumulations using established irnmunohistochemistry (IHC) techniques; IHC-positive cases will be further evaluated with hematoxylin and eosin staining to stage the duration of CWD infection. We will compare CWD prevalence among puma-killed deer to prevalence among deer harvested by hunters in the same area. Using cumulative location data from each collared puma, home range will be estimated. Data from mule deer harvested and sampled within each home range will be extracted from our harvest survey database, preferentially using data collected during the period of predation sampling where sufficient harvest data are available for that time period. To assess differences between predation- and harvest-associated prevalence, we will calculate the CI on the difference as described above; if the CI does not include 0, then we will conclude that these rates differ. �136 Influence of trace minerals on susceptibility: To investigate the potential influence of trace minerals on CWD susceptibility, we began two independent studies. In a retrospective study, we will use archived tissues to compare tissue levels of copper (Cu), molybdenum (Mo), and manganese (Mn) in mule deer infected with CWD to levels in apparently uninfected deer from the same geographic area. We also started an experiment to examine the effect of Cu supplementation on CWD susceptibility in white-tailed deer, wherein we will compare the natural infection rate and course of CWD in captive deer receiving a sustained-release oral Cu supplement to the rate and course in unsupplemented controls residing in the same paddock. Vaccination as a preventive tool: We collaborated with investigators from Colorado State University to conduct a pilot study evaluating safety and serologic responses of mule deer to an anti-PrP vaccine. Four captive deer (2 vaccinates and 2 controls) were monitored and sampled over a 4-month period for evidence of vaccine effects on health and serum antibody levels. Evaluation of an urban CWD management strategy: Recognizing the need for alternatives to traditional strategies for controlling CWD, we initiated a pilot study to evaluate "test and cull" as an approach for managing CWD in urban habitats. Previously, models exploring probable consequences of various management strategies identified selective removal of infected individuals as a potentially effective method for reducing CWD prevalence in mule deer populations, provided that infected deer were detected early and a large (>50%) proportion of the population could be sampled annually (Gross and Miller, 2001, J. Wildl. Manage. 65:205-215). During November-December 2002, 113 free-ranging mule deer were captured, tested, and marked with timed-release radiocollars in urban areas throughout Estes Park to assess the feasibility of such a management approach. This sampling effort represented testing of about 25% of the adult mule deer residing Estes Park. In January 2003, biopsy-positive deer were culled. Dropped radiocollars were recovered in March-April 2003 for reuse in a second round of sampling planned for April-May 2003. In addition to the primary goal of assessing feasibility, data gathered in the course of this study will also be useful in improving our understanding and modeling of the influences of urban landscapes on CWD epidemiology. STUDIES OF CWD PATHOGENESIS & DIAGNOSIS We continued or initiated studies related to rapid screening test evaluation, pathogenesis in natural hosts, and live-animal diagnostic test refinement and evaluation as aids to improving approaches for CWD surveillance and diagnosis in free-ranging deer and elk in Colorado. Evaluation of a rapid screening test: In conjunction with expanded CWD surveillance in Colorado during Sep-Dec 2002, tissue samples (n = 25,050 total) from 23,256 mule deer, white-tailed deer, and Rocky Mountain elk collected statewide were examined using an ELISA developed by Bio-Rad Laboratories, Inc. (brELISA) in a two-phase study. In the validation phase of this study, a total of 4,175 retropharyngeal lymph node (RLN) or obex (OB) tissue samples were examined independently by brELISA and immunohistochemistry (IHC). There were 137 IHC positive samples and 4,038 IHC negative samples. Optical density (OD) values from brELISA were classified as "not detected" or "suspect" based on recommended cut-off values during the validation phase. Based on the validation phase data, only RLN samples were collected for the field application phase of this study and only samples with brELISA OD values> 0.1 were examined by IHC. We estimated assay performance parameters (sensitivity, specificity, agreement) for brELISA to determine the utility of this rapid screening assay in CWD surveillance programs. �137 Pathogenesis in natural host species: We continued our work studying the pathogenesis of CWD in whitetailed deer after oral inoculation with infectious, conspecific brain tissue. This study will complement studies documenting CWD pathogenesis in mule deer and elk that already have been completed. Evaluation of antemortem diagnostic techniques: In order to better study and manage CWD across landscapes where hunting and culling are not feasible sources of diagnostic samples, we continued working to refine and evaluate techniques for sampling live animals. Previously, we conducted a field study to evaluate tonsil biopsy immunohistochemistry (IHC) as a tool for diagnosing CWD in live, freeranging mule deer and estimating prevalence. Based on our initial success, we have applied these techniques to gather data for new studies related to effects of land use patterns on CWD prevalence and its management, as described elsewhere in this report. We also initiated a study to evaluate a prospective rapid blood test for diagnosing CWD in live deer. A total of 37 samples from 21 different captive mule deer, some infected with CWD, were submitted to a private testing laboratory (GeneThera, Denver, CO) for evaluation using collection materials and instructions provided by the laboratory. In order to objectively assess reliability and repeatability of the candidate assay, the testing laboratory was blinded to the infection status and animal identification for individual samples that we submitted. RESULTS AND DISCUSSION STUDIES OF CWD EPIDEMIOLOGY & MANAGEMENT Statewide CWD surveillance: The CDOW sampled over 26,000 deer and elk harvested or culled in northeastern Colorado and other select locations. Survey results were posted on the Division's CWD web page. Prevalence data also will be used to augment an existing database that is the foundation for ongoing analyses and modeling of temporal and spatial aspects of CWD epidemiology, as well as for evaluating responses to management. This year's data will be particularly useful in further exploring local patterns of disease prevalence related to deer movement, density, and land use patterns. Moreover, the surveillance strategy and methods first devised and implemented in Colorado recently served as a model for developing national recommendations on CWD surveillance in free-ranging populations. Transmission mechanisms: A manuscript describing our findings on the relative importance of animal-animal transmission of CWD, as compared to maternal transmission, was accepted for publication and should appear this fall. Our experiment comparing the relative contributions of live animals, contaminated environments, and infected carcasses to CWD transmission revealed that CWD can be transmitted indirectly, from environments contaminated by excreta or decomposed carcasses to susceptible animals. Under experimental conditions, mule deer became infected in 2 of 3 paddocks containing naturally infected deer, in 2 of 3 paddocks where infected deer carcasses had decomposed in situ ~ 1.8 years earlier, and in 1 of 3 paddocks where infected deer had last resided 2.2 years earlier. Our data suggest that indirect transmission and environmental persistence of infectious prions will complicate efforts to control CWD, and perhaps other animal prion diseases. Modeling epidemic dynamics in captive mule deer: Preliminary analyses suggest that indirect transmission models best represent epidemic data; moreover, our model selection results align well with independent empirical findings on CWD transmission mechanisms. We will continue refining candidate �138 models before making final comparisons and parameter estimations. Findings should be of use in refining epidemic models of CWD in free-ranging mule deer populations. Host range and strain variation: Cattle (n = 11) living in paddocks with naturally-infected mule deer remained healthy through 6 years of exposure; in contrast, only I of 12 mule deer introduced into these same paddocks in 1997 is still alive. Our results are consistent with data from cell-free conversion (Raymond et al., 2000, EMBO 19:4425-4430) and intracerebral (IC) challenge (Hamir et al., 2001, J. Vet. Diag. Invest. 13:91-96) studies that suggest the probability of natural susceptibility to CWD in cattle is extremely low. Similarly, neither signs nor postmortem evidence of infection have been observed in fallow deer (n = 24) exposed to infected mule deer for ~2.5 years, and mountain lions (n = 3) consuming carcasses of CWD-infected deer and elk for > I year also have remained healthy. No evidence of infection has been observed in moose, mountain lions, or cattle examined via ongoing surveillance. Clinical signs and postmortem findings consistent with CWD in ferrets were observed in four of five ICinoculated with tissue from infected deer, but have not been observed in the free-ranging white-tailed deer or control groups. Incidence and incubation periods were consistent among affected groups. Preliminary assessment of Western blots (WB) revealed no apparent differences in glycosylation patterns among WBpositive ferrets. In the absence of changes in status in the unaffected groups, we will terminate this study in the next 6 months and summarize our findings. Effects of land use on prevalence: Preliminary analyses revealed that CWD prevalence was higher among deer sampled from urban areas (12.5%, CI=8.4-16.8%, n=243) than among deer from juxtaposed nonurban areas (7 .3%, CI=4.3-l 0.3%, n=288) (Fisher's exact test, P=0.04). The magnitude of difference between CWD prevalence rates associated with urban and non-urban land use (5.3%, CI=2.4-8.2%) further emphasized the apparent effect of urban land use on CWD prevalence. Although CWD prevalence varied somewhat among study sites, it did not differ (Fisher's exact test, P=0.088). Areaspecific differences may reflect greater risk or exposure among subpopulations. However, the trend of higher CWD prevalence in areas of urban land use was consistent across all three sites. Our findings suggest that urbanization is playing an undesirable role in CWD epidemic dynamics in northcentral Colorado's mule deer populations. The underlying cause of this influence on CWD prevalence remains unclear. Urban landscapes may attract or artificially congregate wild cervids. Supplemental feeding, although illegal in Colorado, occurred throughout urban areas in all 3 of our study sites. Urban areas also may provide refuge from predation. Mountain lions are likely the main predator of deer in this area, but they are reclusive and seldom hunt in urban lands. Deer may become more sedentary in urban areas - in extreme cases, urban development may even promote elimination or modification of seasonal migration patterns made by resident deer. Regardless of the reason(s), urban landscapes clearly cannot be ignored in attempts to manage CWD and perhaps other important wildlife disease problems. Selective predation upon infected mule deer: Our work continues from a pilot study conducted to evaluate available global positioning system (GPS)-based telemetry collars for use in this sampling application. We are now sampling mule deer carcasses to test for evidence of CWD infection by monitoring collared mountain lions 1-3 times/week and locating prospective kill sites using a remotely downloadable GPS telemetry system (Lotek, Inc.; model GPS4000). We will continue refining our monitoring approach to ensure that we find kill sites quickly enough to retrieve a suitable tissue sample to test for CWD. Whether target sample sizes can be attained in the time planned for this work remains to be determined. Influence of trace minerals on susceptibility: Both studies are underway. �139 Vaccination as a preventive tool: We observed no adverse effects of vaccination on captive mule deer; serology results are pending. Evaluation of an urban CWD management strategy: Data from our December pilot trial indicate that testing and culling mule deer appears to be a viable approach for managing CWD in Estes Park. Based on the success of the first round of pilot testing, the CDOW has committed to a 5-year management experiment to evaluate the efficacy of test and cull in lowering CWD prevalence in an urban mule deer population. A manuscript describing the results of our feasibility study is in preparation. STUDIES OF CWD PATHOGENESIS & DIAGNOSIS Evaluation of a rapid screening test: In the validation phase, using II-IC-positive cases as known CWDinfected individuals and assuming II-IC-negative cases were uninfected, the relative sensitivity of brELISA depending on species ranged from 98.3-100% for RLN samples and 92.1-93.3% for OB samples; the relative specificity of brELISA depending on species ranged from 99. 9- l 00% for RLN samples and was 100% for OB samples. Overall agreement between brELISA and IHC was ~97.6% in RLN samples and ~95.7% in OB samples of all species where values could be calculated; moreover, mean brELISA OD values were ~46x higher in II-IC-positive samples than in II-IC-negative samples. Discrepancies were observed only in early-stage cases of CWD. Among 20,875 RLN samples screened with brELISA during the field application phase, 155 of 8,877 mule deer, 33 of l l,73 l elk, and 9 of 267 white-tailed deer samples ( 197 total) had OD values > 0 .1 and were further evaluated by IHC to confirm evidence of CWD infection. Of cases flagged for IHC follow-up, 143 of 155 mule deer, 29 of 33 elk, and all 9 white-tailed deer were confirmed positive. Mean (± SE) OD values for II-IC-positive cases detected during the field application phase were comparable to those measured in RLN tissues during the validation phase. Based on these data, brELISA was determined to be an excellent rapid test for screening large numbers of samples in surveys designed to detect CWD infections in deer and elk populations. Pathogenesis in natural host species: Although our study of CWD pathogenesis in white-tailed deer is ongoing, some white-tailed deer inoculated orally with about 2.5 g of brain tissue homogenate (containing about 15 µg PrPcwo) already developed clinical CWD and were euthanized in end-stage disease 16-30 mo postinoculation (Pl). The clinical course in inoculated white-tailed deer was similar to that previously observed in mule deer inoculated with about 15 µg PrPcwo from infected mule deer. Laboratory evaluations of tissues from both our white-tailed deer and mule deer pathogenesis studies are pending. Evaluation of antemortem diagnostic techniques: Tonsil biopsy is a useful tool for estimating CWD prevalence in nonhunted mule deer populations. In addition to applications in the two field studies described here, the techniqu_es we developed are being used in at least four other field studies of CWD ·epidemiology (WY, NM, WI, SD). Thus far, we have been unable to assess the reliability or repeatability of the "GeneThera test". Over 6 mo have passed since blind samples were submitted, but we have been unable to obtain any test results despite repeated attempts to contact the laboratory. Until such evaluations can be completed, we cannot recommend incorporation of this candidate test into any of our ongoing CWD research or management programs. �140 APPENDIX Publications arising from ongoing CWD work: Gould, D. H., J. L. Voss, M. W. Miller, A. M. Bachand, B. A. Cummings, and A. A. Frank. 2003. Survey of cattle in northeast Colorado for evidence of chronic wasting disease: Geographical and high risk targeted sample. Journal of Veterinary Diagnostic Investigation 15: 274-277. Hibler, C. P., K. L. Wilson, T. R Spraker, M. W. Miller, RR. Zink, L. L. DeBuse, E. Andersen, D. Schweitzer, J. A. Kennedy, L.A. Baeten, J. F. Smeltzer, M. D. Salman, and B. E. Powers. 2003. Field validation and assessment of an enzyme-linked immunosorbent assay for detecting chronic wasting disease in mule deer (Odocoileus hemionus), white-tailed deer (Odocoileus virginianus), and Rocky Mountain elk (Cervus elaphus nelsoni). Journal of Veterinary Diagnostic Investigation 15: 311-319. Race, R. E., A. Raines, T. G. M. Baron, M. W. Miller, A. Jenny, and E. S. Williams. 2002. Comparison of abnormal prion protein glycoform patterns from transmissible spongiform encephalopathy agentinfected deer, elk, sheep, and cattle. Journal of Virology 76(23): 12365-12368. Samuel, M. D., D. 0. Joly, M.A. Wild, S. D. Wright, D. L. Otis, R. W. Werge, and M. W. Miller. 2003. Surveillance strategies for detecting chronic wasting disease in free-ranging deer and elk. Results of a CWD surveillance workshop. USGS, BRD, National Wildlife Health Center, Madison, Wisconsin. Sigurdson, C. J., C. Barillas-Mury, M. W. Miller, B. Oesch, L. J. van Keulen, J.P. Langeveld, and E. A. Hoover. 2002. PrP(CWD) lymphoid cell targets in early and advanced chronic wasting disease of mule deer. Journal of General Virology 83: 2617-2628. Spraker, T. R., K. I. O'Rourke, A. Balachandran, R. R. Zink, B. A. Cummings, M. W. Miller, and B. E. Powers. 2002a. Validation of monoclonal antibody F99/97.6. l for immunohistochemical staining of brain and tonsil in mule deer (Odocoileus hemionus) with chronic wasting disease. Journal of Veterinary Diagnostic Investigation 14:3-7. Spraker, T. R., R.R. Zink, B. A. Cummings, M.A. Wild, M. W. Miller, and K. I. O'Rourke. 2002b. Comparison of histological lesions and immunohistochemical staining of proteinase resistant prion protein in a naturally-occurring spongiform encephalopathy of free-ranging mule deer (Odocoileus hemionus) with those of chronic wasting disease of captive mule deer. Veterinary Pathology 39: 110119. Wild, M. A., T. R. Spraker, C. J. Sigurdson, K. I. O'Rourke, and M. W. Miller. 2002. Preclinical diagnosis of chronic wasting disease in captive mule deer (Odocoileus hemionus) and white-tailed deer (Odocoileus virginianus) using tonsillar biopsy. Journal of General Virology 83: 2629-2634. Williams, E. S., and M. W. Miller. 2003. Transmissible spongiform encephalopathies in non-domestic animals: origin, transmission, and risk factors. In Risk analysis of prion diseases in animals. C. I. Lasmezas and D. B. Adams, (eds.). Revue scientifique et technique Office international des Epizooties 22: 145-156. �Colorado Division of Wildlife Wildlife Research Report July 2003 – June 2004 JOB PROGRESS REPORT State of Project No. Work Package No. Task 1 Colorado Federal Aid Project: N/A : : : : 3740 Cost Center 3430 Mammals Research Wildlife Diseases Chronic Wasting Disease in Mule Deer Research and Development : Period Covered: July 1 2003 through June 30, 2004 Author: Michael W. Miller and L. L. Wolfe Personnel: L. A. Baeten, S. Bender, M. M. Conner, K. Cramer, T. R. Davis, M. Farnsworth, K. Griffin, C. P. Hibler, N. T. Hobbs, D. O. Hunter, J. E. Jewell, E. Knox, C. E. Krumm, C. T. Larsen, B. E. Powers, J. Rhyan, M. Sirochman, T. Sirochman, T. R. Spraker, M. K. Watry, E. S. Williams, D. Wroe All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We continued conducting research on various aspects of chronic wasting disease (CWD) epidemiology and management. Here, we report progress in ongoing and recently-completed work. Studies focused on improving and expanding surveillance in free-ranging populations, understanding and modeling transmission mechanisms, identifying ecological and anthropogenic factors that may influence epidemic dynamics, and evaluating and applying alternative diagnostic and control strategies. In addition to preliminary findings reported here, 12 original studies and review articles were published during this segment; citations are appended to the report. 103 �JOB PROGRESS REPORT CHRONIC WASTING DISEASE IN MULE DEER RESEARCH AND DEVELOPMENT Michael W. Miller and L. L. Wolfe INTRODUCTION We continued conducting research on various aspects of chronic wasting disease (CWD) epidemiology and management. Some parts of this work were conducted in collaboration with investigators at Colorado State University, the University of Wyoming, and elsewhere. Specific projects were supported with various combinations of funds from the Colorado Division of Wildlife (CDOW), Federal Aid in Wildlife Restoration Project W-153-R, the U.S. Department of Agriculture (APHIS/VS, the U.S. Department of Interior (USGS/BRD), and National Science Foundation/National Institutes of Health (NIH) Grant DEB-0091961. METHODS Our work on CWD is both multidisciplinary and multifaceted, but broadly falls under the topics of “epidemiology and management” or “pathogenesis and diagnosis”. For simplicity, we describe progress under those headings below. STUDIES OF CWD EPIDEMIOLOGY & MANAGEMENT We continued or initiated studies related to surveillance, transmission mechanisms, epidemic trend forecasting, potential host range and strain variation, risk factors, and management tools and feasibility as aids to understanding and controlling CWD in free-ranging deer and elk in Colorado. Statewide surveillance: Surveillance for CWD continued throughout Colorado to determine the extent of distribution, to estimate prevalence in affected areas, and to monitor prevalence trends. Surveillance methods were as described elsewhere (Miller et al., 2000, J. Wildl. Dis. 36:676–690; Miller & Williams, 2002, Vet. Rec. 151:610–612; Hibler et al., 2003, J. Vet. Diag. Invest. 15:311−319). In addition to reporting of annual survey findings, we also analyzed cumulative surveillance data to examine the potential influences of demographic, spatial, and temporal factors on observed prevalence patterns. We also began exploring ways of improving the efficiency of our CWD surveillance program. Since 1996, tissue samples have been collected from deer killed in vehicle collisions throughout Colorado as part of our monitoring program for detecting CWD in free-ranging populations. We estimated CWD prevalence among vehicle-killed mule deer statewide and compared this to estimated CWD prevalence among mule deer sampled in the vicinity of these collision sites to determine if CWD-infected individuals were more vulnerable to vehicle collisions than otherwise healthy deer. Transmission mechanisms: We summarized findings on empirical evidence of direct and indirect CWD transmission and the relative importance of these mechanisms in epidemic dynamics. Modeling epidemic dynamics in captive mule deer: We continued analyses of 26 years of data (1974−2000) from CWD epidemics at CDOW’s Foothills Wildlife Research Facility to evaluate strength of evidence for a set of candidate models involving indirect and/or direct transmission, with and without latency periods. Estimates of transmission rates derived from these models will provide an upper bound 104 �on what could be expected in wild populations and will guide construction of candidate sets for modeling those populations. Host range and strain variation: We continued a series of experimental studies in cattle, fallow deer, and mountain lions to explore potential host range of CWD after intense but natural exposure; these experiments compliment ongoing surveillance for evidence of infection in species not known to be natural hosts of CWD, including moose and mountain lions. We also completed work looking for evidence of strain variation in CWD agent from various deer sources using domestic ferrets as a laboratory model. Effects of land use on prevalence: We summarized findings on the apparent effects of urban vs. nonurban land use patterns on CWD prevalence in mule deer. Selective predation upon infected mule deer: We continued a study comparing prevalence of CWD among puma-killed mule deer to prevalence among mule deer harvested or randomly culled by humans within home ranges of collared mountain lions to assess whether predation is selective for CWDinfected mule deer. Methods were as described previously (Miller and Wolfe, 2003, Work Package 3430, Task 7410, Progress Report, Colorado Division of Wildlife, Ft. Collins). A total of eight adult mountain lions have been collared, resulting in 39 collared cat months between February 2001 and May 2004. Sampling of predator-killed deer is ongoing. Influence of trace minerals on susceptibility: We continued two independent studies to investigate the potential influence of trace minerals on CWD susceptibility. In a retrospective study, we completed analyses of archived tissues to compare tissue levels of copper (Cu), molybdenum (Mo), and manganese (Mn) in mule deer infected with CWD to levels in apparently uninfected deer from the same geographic area. We also continued an experiment to examine the effect of Cu supplementation on CWD susceptibility in white-tailed deer, wherein we are comparing the natural infection rate and course of CWD in captive deer receiving a sustained-release oral Cu supplement to the rate and course in unsupplemented controls residing in the same paddock. Genetic influences on susceptibility: We continued collaborating with investigators from the University of Wyoming (UWYO) in studies of genetic influence on CWD susceptibility in mule deer. The main objective of ongoing UWYO research has been to search the DNA sequence of the PrP encoding region in exon 3 of the Prnp gene of mule deer for genetic variations that may influence occurrence of naturally acquired CWD. Recent analyses included samples from 529 free-ranging mule deer from four Colorado DAUs (326 from D-10, 63 from D-19, 71 from D-9, and 69 from D-7). Total genomic DNA was extracted from each sample and the PrP coding region from each deer genome was amplified by polymerase chain reaction (PCR). Genotyping was done using commercial sequencing or a simple restriction enzyme digestion (J. E. Jewell, unpublished data) of the PCR amplified PrP gene. Preventive therapies: We collaborated with investigators from the Rocky Mountain Laboratories, NIH, to conduct a pilot study evaluating safety and efficacy of three prospective therapies for preventing CWD in mule deer. Twenty hand-raised mule deer were randomly divided into groups of 5 and assigned to receive a candidate therapy (coded PP, TA, or TC) or no therapy (control). We administered therapies continuously or 3× daily depending on the drug used; administration began 14 days before inoculation, and continued for 14 days after challenge. All groups received pelleted feed and alfalfa hay from the 105 �same source. We used a novel oral inoculation method (M. W. Miller & L. L. Wolfe, unpublished data) for experimental challenge. We collected tonsil biopsies (Wolfe et al., 2002, J. Wildl. Manage. 66:564– 573) from controls about 4 mo post inoculation (PI) and principals about 5 mo PI to assess efficacy of respective therapies in preventing CWD in mule deer. Evaluation of an urban CWD management strategy: We completed an assessment of the feasibility of “test and cull” as an approach for managing CWD in urban habitats, and continued a 5-year study to evaluate the efficacy of this approach in reducing CWD prevalence among urban mule deer. During October 2003−May 2004, we again captured and tested free-ranging mule deer, and marked them with timed-release radiocollars in urban areas throughout Estes Park; our work was complimented by a parallel, coordinated effort by the US National Park Service (NPS) to capture and test deer inside Rocky Mountain National Park (RMNP). The collective annual goal was to test ≥50% of the adult mule deer residing the Estes Park population unit (Conner and Miller, 2004, Ecol. App. in press); target sample sizes (52 adult males and 153 adult females) were estimated based on a mark-resight inventory conducted in December 2003. Field methods were as previously described (Wolfe et al., 2004, Wildl. Soc. Bull. in press). In addition to the primary goal of assessing the efficacy of test and cull as a management strategy, data gathered in the course of this study will also be useful in improving our understanding and modeling of the influences of urban landscapes on CWD epidemiology. STUDIES OF CWD PATHOGENESIS & DIAGNOSIS We continued or initiated studies related to pathogenesis in natural hosts and live-animal diagnostic test refinement and evaluation as aids to improving approaches for CWD surveillance and diagnosis in free-ranging deer and elk in Colorado. Pathogenesis in natural host species: We completed our work studying the pathogenesis of CWD in white-tailed deer after oral inoculation with infectious, conspecific brain tissue. This study will complement studies documenting CWD pathogenesis in mule deer and elk that already have been completed. Evaluation of antemortem diagnostic techniques: We continued working to refine and evaluate tonsil biopsy techniques for diagnosing CWD in live animals. In light of our continued success in applying established techniques (Wolfe et al., 2002, J. Wildl. Manage. 66:564–573), we have continued using tonsil biopsy to gather data for field studies and epidemiological investigations. We also began using tonsil biopsy IHC as diagnostic benchmark for evaluating other candidate tests for diagnosing CWD in live animals. In conjunction with ongoing studies on CWD transmission, we evaluated a candidate rapid test developed by Prion Developmental Laboratories, Inc. (PDL), modified for potential use under field laboratory conditions (J. E. Jewell, unpubl. data). Initial evaluation of this test on biopsy-sized pieces of tonsil tissue collected postmortem from culled mule deer revealed that sensitivity (about 80%) was near the lower limit of acceptability for field use. Modifications to improve sensitivity were made, and subsequently we evaluated sensitivity of the PDL test under conditions simulating those anticipated in field applications using tonsil biopsies collected from captive mule deer naturally infected with CWD. Tissue samples collected via tonsilar biopsy (Wolfe et al., 2002, J. Wildl. Manage. 66:564–573) were examined within 10 min of collection via the candidate PDL test; details of laboratory techniques were proprietary. Laboratory equipment and conditions simulated those that we anticipated would be available at a field site. Paired biopsies were collected from infected and uninfected deer (n = 16); we randomly assigned one of each pair to PDL and the other to immunohistochemistry (IHC) evaluation. Biopsies were processed, test reactions evaluated, and deer categorized as CWD-positive or -negative based on observed reactions; laboratory personnel were unaware of the infection status of sampled deer. Time 106 �from sample collection to reporting of test result was recorded for each biopsied deer. Sensitivity (estimate, 95% CI) of the PDL test was calculated, using IHC as the reference standard. We compared the proportion of positive deer detected by the PDL test to results from IHC using a one-sided Fishers exact test; we used α = 0.1 for all analyses. In addition, the mean reporting time and range of reporting times was calculated for use in assessing the utility of the PDL test under anticipated field conditions. In conjunction with ongoing studies on CWD prevention, we also reevaluated nictitating membrane (also called the “third eyelid”) biopsy as an approach for detecting CWD in live mule deer. We used a modified technique devised for domestic sheep (S. Bender, unpublished data) to identify lymphoid tissue on the nictitating membrane and adjacent conjunctiva, then collected biopsies using established techniques (O’Rourke et al., 1998, Vet. Rec. 142:489-491). We sampled both eyes of 11 mule deer experimentally infected with CWD and known to be tonsil biopsy positive. Nictitating membrane biopsies were evaluated by IHC using published methods (O’Rourke et al., 1998, Vet. Rec. 142:489-491; O’Rourke et al., 2002, Clin. Diag. Lab. Immunol. 9:966-971). We calculated the proportion (± 95% CI) of usable nictitating membrane biopsies, as well as the sensitivity (± 95% CI) of nictitating membrane biopsy IHC for CWD diagnosis using tonsil biopsy IHC as the reference; criteria for regarding this nictitating membrane biopsy technique as potentially useful in diagnosing CWD in mule deer were ≥ 90% of samples containing usable lymphoid tissue and estimated sensitivity ≥ 95%. We also collaborated in a second study to evaluate a prospective rapid blood test (GeneThera, Denver, CO) for diagnosing CWD in live deer. A total of 10 blood samples from tonsil biopsy-positive, captive mule deer were collected by GeneThera representatives using collection materials and protocols provided by the laboratory; samples were immediately taken to their laboratory for evaluation. By previous agreement, the status of sampled animals was known to GeneThera personnel prior to blood collections. RESULTS AND DISCUSSION STUDIES OF CWD EPIDEMIOLOGY & MANAGEMENT Statewide CWD surveillance: The CDOW sampled over 15,000 deer and elk harvested or culled in northern Colorado and other select locations, as well as smaller numbers of deer and elk submitted as clinical suspects. Surveillance revealed two previously undetected CWD foci in mule deer, one on the Grand Mesa (DAU D-51) and the other in Colorado Springs (DAU D-16). Survey results will be posted on the Division’s CWD web page (). Surveillance data also will be used to augment an existing database that is the foundation for ongoing analyses and modeling of temporal and spatial aspects of CWD epidemiology, as well as for evaluating responses to management. In addition to reaffirming the spatial heterogeneity among wintering mule deer subpopulations observed previously (Miller et al., 2000, J. Wildl. Dis., 36:676–690; Conner & Miller, 2004, Ecol. App., in press), our analyses revealed marked differences in CWD prevalence by sex and age groups, as well as clear local trends of increasing prevalence over a 7-yr period. CWD prevalence differed (P < 0.0001) by age (yearling vs. adult), sex, and geographic area at two different spatial scales (game management unit [GMU] or population unit winter range), and increased over time at both geographic scales (GMU: β = 0.064, 95% CI = 0.009−0.119, P = 0.0219; population unit: β = 0.263, 95% CI = 0.134−0.399, P < 0.0001). Disease status (positive or negative) was not independent of age for males (n = 947, df = 3, χ2 = 459, P < 0.0001) or females (n = 549, df = 4, χ2 = 71, P < 0.0001). For both sexes, prevalence peaked in the 4−6-yr old age class, with the largest increase occurring between the 2−3-yr-old and 4−6-yr-old age classes. This differential was larger for males: prevalence rose from 5.9% (95% CI = 4.9−6.8) among 2−3-yr-olds to 19.4% (95% CI =12.1−26.7) among 4−6-yr-olds (P = 0.0002); for the 4−6 yr age class, prevalence among males (19.4%) was 2.7× greater (P = 0.0006) than among females (7.2%). 107 �Demographic, spatial, and temporal factors all appear to contribute to the marked heterogeneity in CWD prevalence in endemic portions of northcentral Colorado. These factors likely combine in various ways to influence epidemic dynamics on both local and broad geographic scales. A manuscript describing our findings is in review for publication in the Journal of Wildlife Diseases. Sampling of vehicle-killed mule deer may be exploited in increasing the efficiency of surveillance programs designed to detect new foci of CWD infection and direct management actions; however, this differential vulnerability also may bias prevalence estimates in natural populations when data from vehicle-killed deer are included in calculating such estimates. Overall CWD prevalence was 1.66× higher in vehicle-killed deer; prevalence among vehicle-killed deer was 0.101 (95% confidence interval [CI] = 0.064−0.139) compared to 0.061 (95% CI = 0.051−0.072) prevalence among mule deer harvested, culled, or biopsied within 3 km of collision sites. The probability of detecting a CWDinfected, vehicle-killed deer, given that at least one other CWD-infected deer had been detected within a 3 km radius of the vehicle-kill site, was 16.7%. Our data suggest increased susceptibility of CWD-infected individuals to vehicle collisions. Evidence of increased susceptibility to vehicle collisions also may aid in understanding vulnerability of CWD-infected individuals to other forms of death, particularly predation. A manuscript describing our findings is in review for publication in the Journal of Wildlife Diseases. Transmission mechanisms: Manuscripts describing our findings on the relative importance of animal−animal transmission of CWD and on the relative contributions of live animals, contaminated environments, and infected carcasses to CWD transmission were accepted for publication and subsequently published in peer-reviewed journals (see Appendix for citations). Modeling epidemic dynamics in captive mule deer: Preliminary analyses suggest that indirect transmission models best represent epidemic data; moreover, our model selection results align well with independent empirical findings on CWD transmission mechanisms (Miller et al., 2004, Emerg. Inf. Dis. 10:1003−1006). We will continue refining candidate models before making final comparisons and parameter estimations. Findings should be of use in refining epidemic models of CWD in free-ranging mule deer populations. Host range and strain variation: Cattle (n = 11) living in paddocks with naturally-infected mule deer remained healthy through 7 years of exposure; in contrast, only 1 of 12 mule deer introduced into these same paddocks in 1997 is still alive. Our results are consistent with data from cell-free conversion (Raymond et al., 2000, EMBO 19:4425-4430) and intracerebral (IC) challenge (Hamir et al., 2001, J. Vet. Diag. Invest. 13:91–96) studies that suggest the probability of natural susceptibility to CWD in cattle is extremely low. Similarly, neither signs nor postmortem evidence of infection have been observed in fallow deer (n = 24) exposed to infected mule deer for ≤3.5 years, and mountain lions (n = 3) consuming carcasses of CWD-infected deer and elk for >2 years also have remained healthy. No evidence of infection has been observed in moose, mountain lions, or cattle examined via ongoing surveillance. Clinical signs and postmortem findings consistent with CWD in ferrets were observed in four of five IC-inoculated with tissue from infected deer, but were not observed in the free-ranging white-tailed deer or control groups. Incidence and incubation periods were consistent among affected groups. Preliminary assessment of Western blots (WB) revealed no apparent differences in glycosylation patterns among WB-positive ferrets, and no evidence of infection in the unaffected white-tailed deer or control groups. Effects of land use on prevalence: Urban land use appears to affect CWD prevalence: rates were higher in developed areas and among male mule deer, suggesting anthropogenic influences on the 108 �occurrence of CWD. We also observed relatively high variation in prevalence across three study sites (Estes Park, Horsetooth Mountain, Glacier View Meadows), suggesting that spatial patterns may be influenced by other factors operating at a broader, landscape scale. Our results suggest that multiple factors, including changes in land use, differences in exposure risk between sexes, and landscape-scaled heterogeneity, are associated with CWD prevalence in north-central Colorado. A manuscript describing these findings in currently “in press” (see Appendix for citation). Selective predation upon infected mule deer: Our work continues from a pilot study conducted to evaluate available global positioning system (GPS)-based telemetry collars for use in this sampling application. Three collar styles have been deployed, and we are continuing to test and evaluate this new technology; aside from our main objective of data gathering related to CWD ecology, evaluation of this technology should be a substantial contribution to future studies of predator-prey relationships. We have detected and examined over 85 kill sites from radio-collared mountain lions and successfully sampled tissues from 28 sites where adult mule deer carcasses were present; we also have collected 17 samples opportunistically from mule deer killed by mountain lions that were not radio-collared. We will continue capturing mountain lions to reach the objective of six to nine collared cat years, and will continue sampling carcasses of lion-killed mule deer to reach our target sample size (n = 157). We also will continue refining our monitoring approach to ensure that we find kill sites quickly enough to retrieve a suitable tissue sample to test for CWD. Whether target sample sizes can be attained in the time planned for this work remains to be determined. Influence of trace minerals on susceptibility: Both studies are well underway. Laboratory analyses of retrospective samples are complete, and data analysis is underway. Experimentally- treated and control deer are being sampled on a regular schedule, but laboratory analyses are incomplete. Genetic influences on susceptibility: Only four codons in the open reading frame of the Pnrp gene exhibit variation in mule deer, and only one of the four results in a change in the final version of PrP (Brayton et al., 2004, Gene 326:167−173; J. E. Jewell, unpublished data) -- this is a change from the amino acid serine (S), the high frequency allele, to phenylalanine (F) at codon 225. Preliminary results showed that estimated frequency of F allele occurrence in gene pools was similar in Colorado DAUs with (D-10: 0.095; n=652) and without endemic CWD (D-19: 0.111; n=126). However, F225 was not detected in the genomes of CWD-infected deer from D-10 (n = 50), and the F225 gene frequency was lower than in uninfected D-10 deer (0.1; χ c2 = 4.6, P < 0.05). We observed a similar pattern of low F225 gene frequency among mule deer infected with CWD after experimental exposure via direct and indirect routes (Miller et al., 2004, Emerg. Inf. Dis. 10:1003−1006). Whether F225 affects truly affects CWD susceptibility or transmission in mule deer remains to be determined, and is the subject of continued investigation. Preventive therapies: All 4 control deer that survived to 4 mo PI showed evidence of PrPCWD accumulation in tonsil biopsies collected ~4 mo PI. Unfortunately, all but 2 of the 15 treated deer also showed PrPCWD accumulation in tonsil biopsies collected ~5 mo PI; the 2 apparently uninfected deer were both from the same treatment group (PP), but overall infection rate did not differ (P = 0.28) from control. We will continue following these deer to examine potential differences in post-exposure survival that could be attributable to therapies, and to further document the outcomes of the alternative inoculation method used. We also plan to continue this work if other candidate therapies become available. Evaluation of an urban CWD management strategy: Data from the 2002−2003 field season indicated that testing and culling mule deer in Estes Park could be done at rates needed to evaluate the efficacy of this approach in reducing CWD prevalence. A manuscript describing the results of our feasibility study is in preparation is “in press” for publication in the Wildlife Society Bulletin. 109 �0.3 Prevalence Because we were successful in reaching objectives for population-level testing, the 2002−2003 field season became year 1 of a 5-year study to evaluate the efficacy of “test and cull” as a CWD control strategy. In year 2 (2003−2004 field season), we captured and tested 44 adult (≥1.3 yr old) male and 119 adult female mule deer in Estes Park. CWD prevalence was about 13.6% among males and 5% among females tested in --- Males --------------- □ Females 0.2 0.1 0 2002 2003 Year Estes Park (Fig. 1); although no clear evidence of a Figure 2. Chronic wasting disease (CWD) prevalence among male (teal bar) and female (plum treatment effect has emerged (Fig. 1), it is probably bar) mule deer tested in Estes Park, Colorado, unrealistic to expect measurable changes in prevalence after 2002−2004. Prevalence between years did not only 1 year of test and cull management. The combined differ (Fisher exact test P≥0.4). Vertical lines are upper 95% confidence limits on estimated efforts of CDOW and RMNP programs resulted in an prevalence. overall testing rate of 63% of the deer wintering in the Estes Park vicinity, including about 90% of the estimated 103 male and 55% of the estimated 306 female deer in this population unit. STUDIES OF CWD PATHOGENESIS & DIAGNOSIS Pathogenesis in natural host species: White-tailed deer inoculated orally with about 2.5 g of brain tissue homogenate (containing about 15 µg PrPCWD) developed clinical CWD and were euthanized in endstage disease 16−30 mo postinoculation (PI). The clinical course in inoculated white-tailed deer was similar to that previously observed in mule deer inoculated with about 15 µg PrPCWD from infected mule deer. Laboratory evaluations of tissues from both our white-tailed deer and mule deer pathogenesis studies are pending. Evaluation of antemortem diagnostic techniques: Tonsil biopsy is a useful tool for estimating CWD prevalence in nonhunted mule deer populations. In addition to applications in the two field studies described here, the techniques we developed are being used in at least six other field studies of CWD epidemiology (WY, NM, WI, SD, NE, CO). Although the PDL test showed considerable promise as a potential field test, assay performance will need to be improved before it can be incorporated into ongoing CWD research or management programs. We observed good assay sensitivity (1.0; 6/6), but relatively low specificity (0.7; 7/10); overall agreement with IHC was 0.64 (95% CI = 0.29−0.98). There appeared to be an unacceptably high number of “false positive” tests -- application in a low prevalence population (e.g., Estes Park) would likely lead to unnecessary culling of numerous healthy deer, and could erode public support for our field study. Consequently, the PDL test was not incorporated into the 2003−2004 field study in Estes Park. The nictitating membrane biopsy technique provided a high proportion of usable samples: all 22 samples contained at least 1 lymphoid follicle and 12−16/22 (55−73%) samples contained ≥ 9 follicles. Unfortunately, IHC of nictitating membrane biopsies detected PrPCWD accumulation in only 2/22 biopsies, both from the same deer. Because estimated sensitivity (0.09; 95% CI 0.01−0.29) is inadequate, we cannot recommend incorporation of nictitating membrane biopsy IHC into any of our ongoing CWD research or management programs. We remain unable to assess the reliability or repeatability of the “GeneThera test”. No test results were provided on the 10 blood samples from positive mule deer; instead, a company representative indicated that extractions from samples were unsuccessful, and that consequently tests could not be run. 110 �This is our second unsuccessful attempt to obtain results from blood samples submitted to GeneThera for CWD testing. Until an evaluation of their test can be completed, we cannot recommend its incorporation into any of our ongoing CWD research or management programs. APPENDIX Publications arising from ongoing CWD work: Belay, E. D., R. A. Maddox, E. S. Williams, M. W. Miller, P. Gambetti, and L. B. Schonberger. 2004. Chronic wasting disease and potential transmission to humans. Emerging Infectious Diseases 10:977−984. Brayton, K. A., K. I. O’Rourke, A. K. Lyda, M. W. Miller, and D. P. Knowles, Jr. 2004. A processed pseudogene contributes to apparent mule deer prion gene heterogeneity. Gene 326:167−173. Miller, M. W. and M. A. Wild. 2004. Epidemiology of chronic wasting disease in captive white-tailed and mule deer. Journal of Wildlife Diseases 40: 320−327. Miller, M. W., and E. S. Williams. 2003. Horizontal prion transmission in mule deer. Nature 425: 35−36. Miller, M. W., and E. S. Williams. 2003. Chronic wasting disease of cervids. In Mad cow disease and related spongiform encephalopathies. D. A. Harris, (Ed.). Current Topics in Microbiology 284:193−214. Miller, M. W., E. S. Williams, N. T. Hobbs, and L. L. Wolfe. 2004. Environmental sources of prion transmission in mule deer. Emerging Infectious Diseases 10: 1003−1006. Miller, M. W., E. S. Williams, B. E. Powers, L. A. Baeten, L. L. Wolfe, and K. L. Green. 2004. Epidemiology and management of chronic wasting disease in free-ranging cervids. In Proceedings of the One Hundred and Seventh Annual Meeting of the United States Animal Health Association, pp. 60−63. O’Rourke, K. I., D. Zhuang, A. Lyda, G. Gomez, E. S. Williams, W. Tuo, and M. W. Miller. 2003. Abundant PrPCWD in tonsil from mule deer with preclinical chronic wasting disease. Journal of Veterinary Diagnostic Investigation 15: 320−323. Powers, B. E., C. P. Hibler, T. R. Spraker, and M. W. Miller. 2004. Large-scale surveillance for chronic wasting disease: The Colorado laboratory experience. In Proceedings of the One Hundred and Seventh Annual Meeting of the United States Animal Health Association, pp. 64. Sigurdson, C. J., and M. W. Miller. 2003. Other animal prion diseases. In Prions for physicians. C. Weissmann, A. Aguzzi, D. Dormont, and N. Hunter, (Eds.). British Medical Bulletin 66: 199−212. Williams, E., and M. Miller. 2003. Prions in the wild: CWD in deer and elk. Microbiology Today 30: 172−173. Wolfe, L. L., W. R. Lance, and M. W. Miller. 2004. Immobilization of mule deer with thiafentanil (A3080) or thiafentanil plus xylazine. Journal of Wildlife Diseases 40: 282−287. Prepared by ____________________________ Michael W. Miller, Veterinarian 111 � https://cpw.cvlcollections.org/files/original/b5591f5db4ca458a29bccdc7cb2f03a3.pdf e4cdf6bbe2a98490cb52044f13029007 PDF Text Text 141 JOB PROGRESS REPORT State of ______C=ol=o=ra=d=o_ _ _ __ Division of Wildlife - Mammals Research Work Package No. ---=3-'-74-'-0=-------- Chronic Wasting Disease and Other Wildlife Disease Management Chronic Wasting Dise~e Surveillance and Laboratory Support T~k 2 --------~------- Period Covered: July 1 2002 through June 30, 2003 Author: L. A. Baeten Personnel: K. Cramer, K. Green, E. Knox, C. T. Larsen, N. Mier, M. W. Miller, K. Taurman, and L. L. Wolfe Interim Report - Preliminary Results This work continues, and precise analysis ofdata has yet to be accomplished. Manipulation or interpretation of these data beyond that contained in this report should be labeled as such and is discouraged. ABSTRACT We established and staffed a Wildlife Health Laboratory (WHL) to facilitate expanded needs for chronic wasting disease (CWD) surveillance throughout Colorado. WHL activities supported CWD epidemiology and management work, as well as various new and ongoing CWD research projects. INTRODUCTION We established and staffed a Wildlife Health Laboratory (WHL) to facilitate expanded needs for chronic w~ting disease (CWD) surveillance throughout Colorado. WHL activities supported CWD epidemiology and management work, as well~ various new and ongoing CWD research projects. Key contributions are described herein. METHODS Statewide CWD surveillance: The discovery of CWD in northwestern Colorado in January 2002 created a sudden demand for both more widespread surveillance and more rapid turnaround on laboratory results. Consequently, the CDOW's CWD surveillance program was overhauled and its capacity greatly expanded over the summer of 2002 in order to meet anticipated demands for surveillance data, ~ well ~ to meet policy-based decisions to provide carc~s quality ~surance information for individual hunters. The most notable changes were the addition of three regional submission laboratories, streamlining of tissue sampling methods, and incorporation of a rapid screening test for CWD diagnosis. Details of overall programmatic features and changes were described on a new CWD-oriented CDOW web page (http://wildlife.state.co.us/CWD/index.asp); details of the evaluation of modified sampling and testing procedures are described below. Evaluation of a rapid screening test: In conjunction with expanded CWD surveillance in Colorado during Sep-Dec 2002, tissue samples (n = 25,050 total) from 23,256 mule deer, white-tailed deer, and Rocky Mountain elk collected statewide were examined using an ELISA developed by Bio-Rad Laboratories, �142 Inc. (brELISA) in a two-phase study. In the validation phase of this study, a total of 4,175 retropharyngeal lymph node (RLN) or obex (OB) tissue samples were examined independently by brELISA and immunohistochemistry (IHC). There were 137 IHC positive samples and 4,038 IHC negative samples. Optical density (OD) values from brELISA were classified as "not detected" or "suspect" based on recommended cut-off values during the validation phase. Based on the validation phase data, only RLN samples were collected for the field application phase ofthis study and only samples with brELISA OD values> 0.1 were examined by IHC. We estimated assay performance parameters (sensitivity, specificity, agreement) for brELISA to determine the utility of this rapid screening assay in CWD surveillance programs. RESULTS AND DISCUSSION Statewide CWD surveillance: The CDOW sampled over 26,000 deer and elk harvested or culled in northeastern Colorado and other select locations. Survey results were posted on the Division's CWD web page. Prevalence data also will be used to augment an existing database that is the foundation for ongoing analyses and modeling of temporal and spatial aspects of CWD epidemiology, as well as for evaluating responses to management. This year's data will be particularly useful in further exploring local patterns of disease prevalence related to deer movement, density, and land use patterns. Moreover, the surveillance strategy and methods first devised and implemented in Colorado recently served as a model for developing national recommendations on CWD surveillance in free-ranging populations. Evaluation of a rapid screening test: In the validation phase, using II-IC-positive cases as known CWDinfected individuals and assuming II-IC-negative cases were uninfected, the relative sensitivity of brELISA depending on species ranged from 98.3-100% for RLN samples and 92.1-93.3% for OB samples; the relative specificity ofbrELISA depending on species ranged from 99.9-100% for RLN samples and was 100% for OB samples. Overall agreement between brELISA and IHC was ~97.6% in RLN samples and ~95.7% in OB samples of all species where values could be calculated; moreover, mean brELISA OD values were ~46x higher in II-IC-positive samples than in II-IC-negative samples. Discrepancies were observed only in early-stage cases of CWD. Among 20,875 RLN samples screened with brELISA during the field application phase, 155 of8,877 mule deer, 33 of 11,731 elk, and 9 of267 white-tailed deer samples (197 total) had OD values > 0.1 and were further evaluated by IHC to confirm evidence ofCWD infection. Of cases flagged for IHC follow-up, 143 of 155 mule deer, 29 of33 elk, and all 9 white-tailed deer were confirmed positive. Mean (± SE) OD values for II-IC-positive cases detected during the field application phase were comparable to those measured in RLN tissues during the validation phase. Based on these data, brELISA was determined to be an excellent rapid test for screening large numbers of samples in surveys designed to detect CWD infections in deer and elk populations. Publications: Hibler, CP, Wilson, KL, Spraker, TR, Miller, MW, Zink, RR, DeBuse, LL, Andersen, E, Shcweitzer, D, Kennedy, JA, Baeten, LA, Smeltzer, JF, Salman, MD, Powers, BE Field Validation and assessment of an enzyme-linked immunosorbent assay for detecting chronic wasting disease in mule deer (Odocoileus hemionus), white-tailed deer (Odocoileus virginianus), and Rocky Mountain elk (Cervus elaphus nelsoni). 2003 J. Vet Diagn Invest 15:311-319. �143 APPENDIX Publications arising from WHL contributions to ongoing CWD work: Hibler, C. P., K. L. Wilson, T. R. Spraker, M. W. Miller, R.R. Zink, L. L. DeBuse, E. Andersen, D. Schweitzer, J. A. Kennedy, L. A. Baeten, J. F. Smeltzer, M. D. Salman, and B. E. Powers. 2003. Field validation and assessment of an enzyme-linked immunosorbent assay for detecting chronic wasting disease in mule deer (Odocoileus hemionus), white-tailed deer (Odocoileus virginianus), and Rocky Mountain elk (Cervus elaphus nelsoni). Journal of Veterinary Diagnostic Investigation 15: 311-319. Samuel, M. D., D. 0. Joly, M.A. Wild, S. D. Wright, D. L. Otis, R. W. Werge, and M. W. Miller. 2003. Surveillance strategies for detecting chronic wasting disease in free-ranging deer and elk. Results of a CWD surveillance workshop. USGS, BRD, National Wildlife Health Center, Madison, Wisconsin. �Colorado Division of Wildlife Wildlife Research Report July 2003 – June 2004 JOB PROGRESS REPORT State of Project No. Work Package No. Task Colorado : : : : 3740 Federal Aid Project Cost Center 3440 Wildlife Health Program Wildlife Diseases Wildlife Disease Surveillance Technical and Laboratory Support : Period Covered: July 1 2003 through June 30, 2004 Author: L. A. Baeten Personnel: K. Cramer, K. Green, K.A. Griffin, E. Knox, C. T. Larsen, M. W. Miller, L. L. Wolfe and D. Wroe All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT The Wildlife Health Laboratory (WHL) was initially created in 2002 to meet expanded needs for chronic wasting disease (CWD) surveillance throughout Colorado. WHL activities supported CWD epidemiology and management work, as well as various new and ongoing CWD research projects. In addition, the WHL has been able to meet demands for diagnostic and laboratory services related to other wildlife diseases that have come to the forefront of concern in the management of Colorado’s wildlife resources. 113 �JOB PROGRESS REPORT WILDLIFE DISEASE SURVEILLANCE TECHNICAL AND LABORATORY SUPPORT L. A. BAETEN INTRODUCTION The Wildlife Health Laboratory (WHL) was created in 2002 in response to the H-1.1 objective of the Division’s Strategic Plan. The purpose delineated in this objective is to “aggressively research, identify, detect, contain and eliminate, where possible, diseases in free-ranging wildlife and captive wildlife that could negatively impact wildlife populations”. The WHL was developed to meet the expanded needs for chronic wasting disease (CWD) surveillance throughout Colorado. WHL activities supported CWD epidemiology, harvest testing and management work, as well as various new and ongoing CWD research projects. In addition, the WHL has been able to meet demands for monitoring, detection, and diagnostic laboratory services related to other wildlife diseases that have come to the forefront of concern in the management of Colorado’s wildlife resources (i.e. West Nile virus, plague, Pasteurellosis, etc.). SUMMARY Statewide CWD surveillance: The discovery of CWD in northwestern Colorado in January 2002 created a sudden demand for both more widespread surveillance and more rapid turnaround on laboratory results. The CDOW’s CWD surveillance program was modified in 2002-2003 to decrease turnaround time (from initial submission to acquisition and posting of results), improve data collection and quality control. The notable changes in 2003-2004 were the addition of an electronic data collection system that was used statewide for collection of field and laboratory data. The WHL staff was instrumental in helping to delineate system mechanics, provide testing and troubleshooting capabilities and assist with training efforts. Details of overall programmatic features and changes were described on the CWDoriented CDOW web page (http://wildlife.state.co.us/CWD/index.asp); details of the efficiencies in the sampling and testing procedures are described below. Numerous state agencies have request demonstrations of this new system for possible implementation in their CWD surveillance programs. During 2003-2004, the CDOW sampled 17,268 deer, elk and moose harvested or culled in northeastern Colorado and other select locations. Survey results were posted on the Division’s CWD web page (http://wildlife.state.co.us/CWD/index.asp). The data generated provided annual CWD survey results. These data were added to the cumulative surveillance data that is the foundation for ongoing analysis and modeling of temporal and spatial aspects of CWD epidemiology, examining potential influences of demographics, as well as evaluating responses to management. In an effort to improve surveillance efficiencies, tissue samples were collected from deer killed from vehicle collisions throughout the state. Prevalence data from this group were analyzed to determine if CWD-infected individuals were more vulnerable than otherwise healthy animals. Moreover, the surveillance strategy and methods first devised and implemented in Colorado continue to serve as a model for developing national recommendations on CWD surveillance in freeranging populations. 114 �CWD tissue handling and disposal: The WHL staff prepared documents summarizing published literature on appropriate disposal methods for CWD infected tissues (incineration and chemical). These documents were used extensively for public information during the review process for the incinerator proposal for Wellington. The WHL supervisor worked with EPA and national veterinary diagnostic laboratory representatives to develop “Best Management Practices” when handling CWD infected materials. Research projects The WHL lab staff provided technical and diagnostic support for the ongoing DOW research projects listed below. Major accomplishments and contributions for the WHL this fiscal year include: completion of the experimental phase of the “Molecular epidemiology of strain variations in CWD”; evaluation of antemortem diagnostics (described below); the addition of DNA extractions to the list of diagnostic capabilities (supporting the “Genetic influences study and several species conservation projects); initiation of the pilot study on biosolids and wastewater; preliminary results from studies looking at prion inactivation are ready for presentation; and extensive sample collections for the studies of prions in biological excretions and environmental samples. 1. Molecular epidemiology of strain variations in chronic wasting disease (CWD) 2. CWD host range studies 3. Selective predation upon CWD-infected mule deer 4. Trace mineral influences on CWD susceptibility 5. Genetic influences on CWD susceptibility 6. Evaluation of preventative therapies for CWD 7. Evaluation of antemortem diagnostic techniques for CWD 8. Detection of prions in environmental samples 9. Detection of prions in biosolids and waste water 10. Chemical inactivation of prions 11. Detection of prions in biological excretions 12. West Nile virus in black-tailed prairie dogs 13. Prevalence of CWD in ungulates killed via vehicle collisions 14. Evaluation of a recombinant plague vaccine in lynx 15. Invertebrate role in CWD transmission 16. Evaluation of FWRF diarrhea outbreaks 17. Uncompaghre fawn mortality Evaluation of antemortem diagnostic technique for CWD: In a pilot study, the WHL evaluated the use of a lateral flow strip test (Prion Developmental Laboratories, Inc.) to determine its applicability for “live animal testing” in the field. Approximately 40 tonsil samples were used to assess the applicability of this diagnostic test under field conditions. The test kit procedures were manipulated to determine if modifications to the lymph node procedures could be used to accommodate tonsil tissue (and tonsil biopsy sized tissues). It was determined that the assay could be modified to work under field 115 �conditions (i.e. roving lab). However, despite efforts to modify the test parameters, the sensitivity was not acceptable to pursue further field trials with this new diagnostic test system. In addition, the WHL staff provide technical assistance to collaborators interested in archived tissues for ongoing research projects listed in the table below. The WHL staff accomplished this via additional sample collections from hunter harvest, culls and other DOW submissions. These samples are archived then aliquoted and shipped to the collaborators according to specific tissue requests. Collaborative Agreements IND/HPF/USCF USDA/ARS NYSIBR PDL CSU USDA/ARS NIH/RML CWRU NYSIBR CSU IDEXX USU RMNP GeneThera Brain Brain Brain Lymph nodes Multiple tissues Eyelids, blood Multiple tissues Brain, lymph nodes Urine Brain Lymph nodes DNA extracts (blood) DNA extracts (blood) Blood Transgenic mouse development/ host range studies Strain typing, comparison to other TSE strains Transgenic mouse strains Lateral flow strip test Experimental transmission (ante and post mortem collection CWD assay evaluations Evaluate strain variations Transgenic mouse host strain study, cellular prion transport CWD assay evaluations Effects of composting on prion inactivation Validation of diagnostic assay Epidemiology studies Epidemiology studies Antemortem assay evaluation Wildlife Disease Surveillance: The WHL performed necropsies to assist state wildlife managers and biologists in determining the cause of death for wildlife species including: deer, elk, bighorn sheep, mountain goats, bear, various avian species and rodents (See Table 1). This necropsy effort included support for two species conservation projects. The WHL provided technical and diagnostic support for the field projects listed below. This effort included biological sample collection, data collection, sample processing, diagnostic testing, archiving and/or distribution of samples. 1. Evaluation of diagnostic techniques for avian translocations 2. Disease surveillance for Prairie grouse restoration 3. Disease surveillance for Turkey translocation 3. Bighorn sheep translocations: Identification of Pasturella spp. strains 4. Identification of Johne’s disease in BHS and RMG 5. Identification of lungworm larvae in BHS feces 6. Black-footed ferret restoration: carnivore sampling 7. Lynx restoration 8. Winter deer capture: mule deer survival monitoring 9. Elk Fertility control 10. Test and cull evaluation 11. CWD management culling 12. Foot hills wildlife research facility 116 �West Nile Virus: The WHL established in-house testing for West Nile virus (WNV). Fifty-four carcasses were submitted as suspects for necropsy and testing during this fiscal year. Thirteen positives were identified. In conjunction with the DOW WNV testing performed at the WHL, tissue samples collected during necropsies were provided to CDC (Komar) for their use in experimental trials developing a new post mortem assay for WNV. Avian Translocations: The WHL investigated alternative diagnostic testing for avian translocation projects. An in-house assay for Mycoplasma (synoviae, gallisepticum, meleagridis) was determined to be optimal for testing individual birds being translocated. The use of the ELISA assay for these serological tests minimized the cross reactivity effects that were experienced with diagnostic tests used previously. With the use of the ELISA, next-day releases were possible, therefore, decreasing individual stress levels and increasing survivability for birds moved in translocation efforts. The WHL staff provided technical and diagnostic support for three avian translocation projects during this project year (sharp-tailed grouse, turkey and ring-necked pheasant). In combination with the diagnostic necropsy support, the WHL established a database to allow electronic review of these data over time. All historical diagnostic records were incorporated into the database during this fiscal year. To date, this database contains approximately 400 records. From the diagnostic reports database, the WHL prepared wildlife disease summaries for statewide distribution. The wildlife disease summaries for years 2002 and 2003 delineate animal mortality data by species, quarter and region. This data will assist wildlife veterinarians, managers and biologists in future wildlife disease events. During this fiscal year, the WHL established an archive database which includes all of the historical samples collected since the initial establishment of the WHL in the 1990’s. This database allows WHL staff to determine what tissues are available for use in research projects, delineates physical locations where various tissue samples can be found, tracks distribution of tissue samples and contains appropriate animal identification and specifications. At the end of the fiscal year, there were a total of 4,850 entries with approximately 300 of those added for the year. Table 1: Diagnostic Support Species Necropsies Diagnostic Samples (collected, processed, archived) Carnivore 3 40 Deer 30 385 Elk 10 4 Lynx 0 106 Other avian species (WNV) 36 36 Other ungulate 18 107 Prairie grouse 7 22 Small game 10 50 Small mammals 15 35 Total 126 745 117 �Training sessions: The WHL has provided multiple training session for CWD sample collection. The attendees included CDOW employees as well as federal employees from the Rocky Mountain region. In addition, the WHL staff assisted with training sessions for DWM trainees in necropsy techniques, darting and sample collections. Presentations: The WHL staff made various presentations on wildlife disease to various groups including the Wildlife Society, USFWS, CDOW staff, black-footed ferret subcommittee, Colorado Wildlife Federation. The titles of the presentations were: 1. West Nile Surveillance in Colorado 2003 2. The impact of West Nile virus on wildlife populations 3. The significance of West Nile virus in prairie dogs 4. Common diseases in wildlife populations of Colorado APPENDIX Publications arising from WHL contributions to ongoing CWD work: Brayton KA, O’Rourke KI, Lyda AK, Miller MW, Knowles Jr. DP. A processed pseudogene contributes to apparent mule deer gene heterogeneity. Gene 326: 167-173. Miller MW; Williams ES. Horizontal prion transmission in mule deer. Nature 2003 425: 35-36 _________: __________, Hobbs NT; Wolfe LL. Environmental Sources of Prion Transmission in Mule Deer. Emerging Infectious Diseases 2004 10(6): 1003-1007 _________; Wild MA. Epidemiology of Chronic Wasting Disease in Captive White-tailed and Mule deer. J Wildlife Dis 2004 40(2): 320-327 O’Rourke KI; Zhuang D; Lyda A; Gomez G; Williams ES; Tuo W; Miller MW. Abundant PrPCWD in tonsil from mule deer with preclinical chronic wasting disease. J Vet Diagn Invest 2003 13: 320-323 Powers BE, Hibler CP, Spraker TR, Miller MW. Large scale surveillance for chronic wasting disease: The Colorado laboratory experience. Annual proceedings USAHA 2004 pg 64. Prepared by _________________________ Laurie A. Baeten, Veterinarian 118 � https://cpw.cvlcollections.org/files/original/fbee6b213ac31095475b46b4f8398be5.pdf 9da843fa860f7ec35dd3faf7381af65f PDF Text Text 145 JOB PROGRESS REPORT Stateof_ _~C~o~lo~r~ad~o~-----~ Division of Wildlife - Mammals Research Work Package No. 8160 3740 Chronic Wasting Disease and other Willife Disease Management Animal and Pen Support Facilities Task No. -~3_ _ _ _ _ _ _ __ Period Covered: January 1, 2001 - June 30, 2003. Author: T.R. Davis Personnel: 2001: H. Barr, C. Budler, N. Dryer, D. Finley, J. Foster, M. Foster, J. Habel, M. Hanusack, L. Ho, B. Hotchmuth, E. Jones, S. Liss, M. Lowe, M. Miers, A. Mitchell,, T. Petersburg, T. Terrell, C. Weagley, 2001/2002: B. Bates, D. Biggins, E. Berrill, K. Downing, D. Finely, J. Foster, M. Foster, J. Grigg, J. Habel, M. Hanusack, J. Hatch, C. Hernandez, L. Ho, E. Jones, M. Lowe, A. Mitchell, N. Miers, A. Ray, L. Reimer, R. Rhyan, K. Sparks, T. Stout, T. Terrell, R. Thompson, M. Thonhoff, C. Weagley, D. Weaver, B. Williams, T. Zeaman, 2002/2003: M. Anderson, B. Bates, K. Beamer, L. Dahl, J. Fenwick, D. Fox, K. Fox, J. Habel, J. Hatch, T. Halasinski, M. Hanusack, G. Harvey, L. Ho, E. Jones, G. Kyriacou, M. Lowe, N. Miers, A. Mitchell, A. Northrup, R. Rutledge, K. Steffen, T. Stout, D. Thompson, R. Thompson, D. Weaver, ABSTRACT The Colorado Division of Wildlife's Foothills Wildlife Research Facility (FWRF) maintained captive animals (2000/200 l annual total: 262, 2001/2002 annual total: 320, 2002/2003 annual total: 312) and facilities in support of twenty-one captive wildlife research projects. Chronic wasting disease (CWD) pathology, and etiology, in deer and potential transmission to other species was the primary focus of research during this period, however FWRF supported a number of other significant research projects including contraception and reproductive effects, pathogen immunization, foraging behavior, drug delivery systems, and evaluation of wildlife capture pharmaceuticals. Three new species; fallow deer, domestic ferrets, and mountain lions were added to support CWD research as well as additional numbers of mule deer and white-tailed deer. Chronic wasting disease was again a significant source of mortality in mule deer and white-tailed deer and is reflected by the number of CWD research projects conducted at FWRF during this period. The CWD Management Protocol was updated to incorporate new information and early detection techniques, while maintaining the philosophy of managing the disease for research purposes under heightened bio-safety guidelines and intensive herd management. Additionally, a number of other protocols were revised, and new SO P's developed to accommodate the new species, facility improvements, and expanded research. An expanded database, a 5 year facility capitol construction plan, and a draft facility fee schedule were also implemented. The quality of animal care and facility maintenance provided by temporary, work-study, personal service, intern and volunteer employees is in part reflected by the finding of compliance under the Animal Welfare Act during the annual USDA APHIS inspections of FWRF. In addition to routine maintenance, the FWRF team made significant facility improvements including new facilities to accommodate expanded CWD mule deer research, �146 partial completion of a mountain lion holding facility, and support for construction of the new Wildlife Health Lab now located within the FWRF perimeter. Animal Maintenance: Routine animal husbandry including feeding, health observations, training, weighing, and clean-up, was performed primarily by well trained temporary employees, work-study students, and volunteers. FWRF was inspected by USDA APHIS for compliance with federal animal welfare regulations on March 8 200 I, April 12 2002, and April 30 2003. Table I summarizes the species totals reported to USDA animal welfare and includes all neonates born at the facility, transfers into and out of the facility, and all animals that died or were humanely euthanised during the respective fiscal year. Ungulate herd levels at any one time averaged approximately 70 percent of the ungulate total and 60-65 percent of the total number of animals housed at the facility. Table I. Species reported to USDA Animal Welfare Species Bighorn Sheep 2000/2001 Total 57 26 2001/2002 Total 52 22 2002/2003 Total 28 25 25 25 36 74 126 139 21 20 21 24 40 39 227 285 288 159 200 202 11 11 11 21 21 10 3 3 3 262 320 312 Elk Fallow Deer Mule Deer Pronghorn Antelope White-tailed Deer Ungulate Total Ungulate Mean Cattle Domestic Ferrets Mountain Lions Facility Total Herd Management: Three new species; domestic ferrets, fallow deer, and mountain lions were added to the facility in FY 2000/200 l and mule deer and white-tailed deer herd levels were expanded in FY 0 l/02, and 02/03 through herd management practices and incoming transfers. Additional adult animals were brought in to support expanding CWD, fertility control, and brucellosis vaccine research and consisted primarily of free �147 ranging and habituated mule deer obtained from various locations around the state. Captive mule deer, white-tailed deer, and pronghorn antelope were also brought in from out of state to supplement FWRF herds. The bighorn sheep herd was reduced in FY 2001/2002 and FY 2002/2003 through natural mortality and an out-going transfer of excess animals. The Fallow deer herd was allowed to expand naturally as per the study protocol in FY 2001/2002, while the cattle elk, and pronghorn herd levels remained relatively constant for the period. Commission approval was granted in 200 I to transfer excess FWRF captive wildlife, and/or orphaned neonates out of state to support collaborative and non-agency wildlife research projects. In 200 I the excess bighorn sheep were transferred to a research facility in Idaho, and in 2001, 2002, and 2003 orphaned mule deer neonates were transferred to a captive facility in Wyoming. It is important to note that the 2002 and 2003 out of state transfers were not of FWRF origin, but habituated orphaned fawns not suitable for release. Other facility transfers include several excess bighorn weanlings that went to a zoo for display, several pronghorn bucks that were borrowed from (and returned to), another captive wildlife research facility, and several free ranging bull elk brought in for breeding purposes. Breeding was planned annually to maintain optimal population sizes of the various species required to support current and future research projects. Depending on research objectives, some of the offspring from FWRF animals were hand-raised, and various species of wild orphaned neonates were accepted for hand rearing. Habituated weanlings and adult animals were also accepted whenever herd levels would allow. Hand rearing protocols for mule deer are described by Parker and Wong (1987), and by Wild and Miller { 1991) for bighorn sheep, elk, pronghorn antelope, and white-tailed deer. The male cattle, domestic ferrets and mountain lions were castrated at an early age, and the male fallow deer were vasectomized in the summer of 02/03 to prevent further breeding. Table 3 summarizes the breeding and rearing practices of ungulate species for the period: �148 Table 3 Ungulate breeding and rearing practices FWRF Breeding Species 2000 2000/2001 Bighorn Sheep Bred Elk Bred 5 Cows Fallow Deer Yearlings, did not breed Bred Mule Deer Pronghorn Antelope Bred White-tailed Deer Bred 3 yearlings 2001/2002 Bighorn sheep FWRF Breeding 2001 Bred Elk Did not breed Fallow Deer Mule Deer Bred Bred Pronghorn Antelope White-tailed Deer Bred 2002/2003 Bighorn Sheep Pronghorn Antelope White-tailed Deer Orphans 2001 Hand raised 2, dam raised 2, 1 stillborn No offspring 1 weanling Hand raised 4, dam raised others Hand raised 4 , 2 still born, others Euthanized as per study protocol Dam raised 0 0 0 0 13 orphans FWRF Neonate Rearine 2002 Hand raised 5, dam raised others No offspring Orphans 2002 0 3 orphans, 9 weanlings 1 weanling Bred Dam raised Hand raised 20, dam raised others Euthanized as per study protocol Dam raised FWRF Breeding 2002 Not bred FWRF Neonate Rearing 2003 No offspring Orphans 2003 Bred 3 cows 0 Not bred 1 hand raised, 2 dam raised No offspring Bred Dam raised 5 weanlings, Bred Hand raised 1 orphan Bred Dam raised 0 Elk Fallow Deer Mule Deer FWRF Neonate Rearine 2001 Dam raised I orphan I weanling 11 orphans, 2 weanlings 0 0 �149 Nutritional Maintenance: Feeding protocols for ungulates previously housed at the facility were reviewed by Wild ( 1997). The fallow deer were maintained on a high quality grass alfalfa mix hay and Regular Ranch-way deer and elk ration. The domestic ferrets were maintained on a commercial ferret chow, and the mountain lion kittens were initially maintained on Kitten Milk Replacer, Nurtural, and commercial kitten chow. The kittens were switched to a ground commercial feline diet at weaning, and were introduced to chunk deer and elk meat for training purposes at four months of age. A commercial carnivore supplement was added to the training meat to enhance dietary levels of calcium, and vitamins A and E, and was offered several times weekly. At five months of age, the kittens were gradually introduced to whole deer and elk carcasses and carcass portions with the GI tract removed, and are currently maintained on carcass portions, and training meat with supplement. Individuals of all species maintained reasonable body condition on available diets with the exception of some mule deer fawns, and CWD infected animals at the clinical stage of the disease. Fawn mortalities may have been associated with general poor body condition of does infected with chronic wasting disease, the presence of other etiological agents, and/or interspecies competition for space and cover in paddocks housing cattle and fallow deer. Pen Enrichment: In an effort to provide cover and subsequently reduce stress, the mule deer in the cattle pens were provided with a refuge area not accessible to the cattle, and artificial refuge areas were constructed in all paddocks housing semi-wild deer and dam raised neonates. Single piece and "L" shaped hide-outs, were constructed on site, and vegetation ex-closures were added in early spring and removed later to provide seasonal natural cover. Additionally, the Fort Collins Water Treatment Plant donated rock, labor and the use of equipment to construct two rock mountains in the bighorn sheep pens to enhance the natural structure in these areas. In addition to pen structure, behavioral enrichment was offered through training. The mountain lions were trained using operant conditioning; a form of training based on a reward system, and used widely in wildlife display facilities. Using this system, the lion kittens were taught to sit, platform, kennel, and stretch up on the fence for physical exams. Hand raised ungulate neonates were treat trained using the same philosophy, and were taught to follow their human trainers and stand on the scale for physical exams and weighing. Passive training was also used to habituate animals to the scale and alley-way by feeding the animals supplement in these areas, and allowing free exploration without human interference. Health Maintenance: Animal health care was provided as required and as mandated by the preventive medicine program (Wild 1995) and chronic wasting disease protocols. Overall, captive wildlife maintained at FWRF remained healthy throughout the period. Chronic wasting disease (CWD) continues to be a significant source of mortality in captive mule deer and white-tailed deer and is reflected by the number of animals dedicated to CWD research.projects throughout this period. Dystocia was a significant source of mortality in adult pronghorn does, and was associated with a failure of the cervix to dilate at the time of parturition. The underlying cause of the pronghorn dystocia is still unknown, and the collaborative USDA RB5 l brucellosis vaccine project was put on hold in FY 2002/2003 due to the resulting reduced number of adult females available. Other significant etiological agents included Epizootic hemorrhagic disease (EHD), bluetongue virus {BTV), and clostridium perfringens. �150 Standard Operating Procedures: Chronic wasting disease The CWD management protocol was again revised in FY 2002/2003 (Attachment 1). Generally, CWD continues to be managed as described by Wild (1997): to maintain CWD and maximize potential exposure for specific research objectives. The revised protocol was prepared to incorporate new information resulting from recent research findings: increasing bio-security, incorporating early detection techniques, and intensive herd management of CWD infected animals (Wild et. al 2002, Wolfe et. al 2002). All animals at FWRF were monitored closely for clinical signs of CWD, and tissues from all mortalities occurring at FWRF were examined for evidence of infection with CWD. Systems development In addition to the CWD protocol, all animal husbandry, facility security, FWRF management protocols, and veterinary supply inventories were reviewed and updated. Protocols were developed to manage the new species, and an Access database was developed to track additional information such as projects and veterinary treatments. The old paradox database and hard copies of vital records, necropsy, clinical pathology, and transfer information was integrated into the new database. Facility and animal maintenance costs were analyzed and incorporated into a draft fee schedule for use of research animals and FWRF facilities by professional collaborators, and a draft 5 year facility capitol construction plan was developed to address long term planning needs. Educational Contributions FWRF functions primarily to support wildlife research, however when possible and relevant, facility tours were provided to school, university, and professional groups. We emphasized the importance of maintaining captive wildlife to perform controlled experiments, and the contributions made by current and historic research projects conducted at FWRF. FWRF animals and facilities were also used occasionally for hands-on training of CDOW employees, collaborators, and other professional groups in sampling techniques and chemical immobilization. Research Projects: Facility operations offered support for research projects conducted by CDOW personnel and other collaborators that were initiated, conducted, or continued using FWRF animals and facilities. A total of twenty one research projects were supported by FWRF for the period: • • • • • • • • • • • Cattle susceptibility to chronic wasting disease. Mechanisms of CWD transmission in mule deer. Evaluation ofprospective preventative therapies for chronic wasting disease in mule deer. Validation ofa potential blood test for chronic wasting disease (GeneThera test). Prion peptide immunization and challenge. Molecular epidemiology of strain variations in chronic wasting disease. Susceptibility ofMountain Lions to chronic wasting disease. Susceptibility offallow deer to chronic wasting disease. Pathogenesis of chronic wasting disease in white-tailed deer. Effects ofGnRH-PAP on reproduction and behavior in female mule deer. Evaluation ofGnRH agonist (leuprolide) as a reversible contraceptive in mule deer. �151 • • • • • • • • • • Evaluation ofGnRH agonist (Lupron) as a potential contraceptive in rocky mountain elk: Effects on pregnancy. Development ofa remote delivery system for GnRH agonist (leuprolide) in female elk Paradoxical immunosuppression in bighorn lambs as a mechanism for depressed recruitment following pastuerellosis epidemics. Biosafety and reproductive effects ofRB51 (brucellosis) vaccine in pronghorn. Evaluation ofdrug delivery and dart trauma using collared and un-collared pneudart and daninject darts. Evaluation ofA3080 (thiafentanil oxalate) and naltrexone HCLfor the immobilization and reversal of mule deer. Evaluation ofA3080 (thiafentanil oxalate) and naltrexone HCLfor the immobilization and reversal ofpronghorn antelope. Effects of 2% DRC-1339 treated brown rice on non-target species. Testing alternative models ofherbivore foraging in heterogeneous environments. Field Immobilization Training. Facility Improvement Projects: A variety of scheduled and unscheduled maintenance and repair activities were necessary to support facility operation and ongoing research programs. Highlights include construction of the new Wildlife Health Lab (WHL) housing a laboratory, office space, a necropsy lab, and walk in freezer/cooler space, now located within the FWRF perimeter. This project was designed, constructed, and funded by the CDOW engineering and capitol construction team, while FWRF personnel provided support services. Additional facility modifications include twelve new paddocks, associated buildings, alleys and an access road to support the CWD transmission study, and an automatic water system for all paddocks on the east side of the facility. Other improvements included five new isolation pens, perimeter fence and gate upgrades, and construction of compost bins to hold animal waste material generated from CWD research paddocks. A new mountain lion facility including a concrete block building containing 4 indoor dens and a work space, a 50 x 60 foot outdoor pen, and shift containment system is currently under construction. A 2000 gallon vault was installed on the east side, a new pasture was also constructed on the west side, and the old house trailer was demolished. Additionally, the Soldier Canyon Filter Plant donated several culverts and constructed a detention pond on the west side of the facility to better manage natural water run-off and scheduled water releases from the plant. Facility maintenance and construction projects were prioritized based on animal welfare concerns and anticipated research needs. Table 3 summarizes the completed, current, and on-going facility construction maintenance projects for the period. �152 • Table 3. Facility Improvement Projects Details and Information Replace roofs on existing pens, and add 5 additional pens with shelters Remove garage door, replace with permanent wall, add window Automatic waters installed in all existing pens and new paddocks for transmission study. CSU provided the electric contractor, FWRF contracted out plumbing Purchase Tough Sheds, level sites, pour concrete pads for Transmission study, purchased 1 shed, other was supplied by WHL Construct 9 new pens, and split 2 pens into 4 for Transmission study, CSU provided contractor for installation Construct Feed-sheds for north and south transmission study pens Completion Year Construct feed shelters for transmission study pens Construct 400 feet of alley, 16 walkthru gates for transmission study 2001/2002 Completed Construct road, culvert donated by Fort Collins Water Treatment Plant 2001/2002 Completed Purchased all gates < 14 foot long, 14 'gates donated by CDOW game damage, installed gates, added horse fence, add gate opening in D3 contracted out electric fence modifications Construction of D 1 pasture contracted out by NWRC 2001/2002 Install automatic water in D 1, and a water shut off valve in the west hub, contracted out plumbing and excavation Construct l feed shelter, and l 2001/2002 Project Status 1. Improvements to west rearing area 2. East lab improvements 3. Add 13 automatic waters, 4 shut off valves to east side Completed 4. Add pellet feed storage shed, and feed-shed to east side 5. Construct 12 new pens on east side Completed 6. Construct 2 additional feed storage sheds on east side 7. Construct 13 feed shelters on east side 8.Construct alley system and gates for new pens on east side 9. Construct access road to new alley system on east side l 0. Replace all wooden drive through gates with 7 foot metal tubing gates, add gate to south side of D3 11. Construct new pasture on west side (Dl) 12. Plumbing upgrades to west hub area Completed 13. Construct 2 Completed Completed Completed Completed Completed Completed Completed Completed 2001/2002 2001/2002 2001/2002 2001/2002 2001/2002 2001/2002 2001/2002 2001/2002 2001/2002 �153 shelters in D 1 14.East fawn rearing area improvements Completed 15. House trailer demolition, and FWRF site clean-up 16. Construct ram pen exclosure around feed area in E3 17. Reconstruct shed on west side of west scale-room, modify scale 18. Water damage repair to El/E2 feedshed 19. Perimeter Fence upgrades Completed 20. Upgrade 2 perimeter and main east and west gates 21. Add Secondary perimeter gate and 8 foot fence on south side of facility Completed 22. Compost animal waste from CWD paddocks Initial start-up is completed, composting is ongoing Completed 23. Replace east side septic tank 24. Rock mountains constructed in upper sheep pens 25. Construct west detentionpond 26. Construct mountain lion holding facility Completed animal shelter Reconstruct roof structures, repair shelters, double fencing on N side, add 1 alley gate Demo old trailer, clean up, organize FWRF construction materials and supplies, remove waste Purchased range panels, installed panels, added horse fence 2001/2002 2001/2002 2002/2003 Completed Reconstruct shed, modify scale to accommodate access from west side 2002/2003 Completed Remove soil on west side, reconstruct wall, re-grade soil 2002/2003 Completed Replace rotten posts, add V-mesh to lower 4 feet of perimeter fence, contracted out labor on V-mesh Replace 4 old drive thru gates with 8 foot chain link gates 2002/2003 Close off FWRF access road between the Ft. Collins Water Treatment Plant and Soldier Canyon Filter Plant, contracted out time and materials Construct compost bins, purchase bacteria, train personnel to mix and monitor, Contracted out initial bin construction, and start-up Replaced rusted metal l 000 gallon tank with a 2000 gallon concrete vault, Contracted out time and materials Rocks, equipt, and time to construct the mountains donated by the Ft. Collins Water Treatment Plant Construct pond to maintain drainage water inside our perimeter fence, time and equipt. to construct pond was donated by the Soldier Canyon Filter Plant Utilities, concrete block building, 50 x 60 foot outdoor pen, shift containment system, and 4 indoor dens, building slab, and alley concrete, concrete block building, 2002/2003 Completed Completed Completed Current project: planning, utilities, and building construction completed. finish: 2002/2003 2002/2003 2002/2003 2002/2003 2002/2003 Project began 2001/2002, scheduled for completion 2004/2005 �154 outdoor pen, shift containment, indoor dens Current project: 5 completed, finish: 7 plumbing, electrical, and engineering contracted out 28. New roofs/repair structure on old feed-sheds and animal shelters. On-going project 29. Add additional animal shelters On-going project 30. Road Maintenance 3 1. Paint old building exteriors On-going project Approx. ¼ of the old structures and roofs on the facility have been replaced in the last 2 years using treated lumber and long lasting roofing materials Construct additional shelters in pens with heavy stocking rates. (36 unroilate pens on the facility) Road grading and upkeep 32. Repair/replace latches, and broken or water damaged alley-way boards 33. Replace walk thru alley gates 34. Replace old visual barrier fencing and utility wire on metal gates On-going project 27. Reconstruct west isolation pens 35. Animal holding fence upgrades, and repairs 36. Construct artificial refuge areas inside pens for neonates and adults On-going project Demolish old pens and shelters, reconstruct with upgraded design and materials Now using CCA treated lumber, or metal siding for repairs & building replacements to reduce the amount of painting necessary in the future. Now using CCA treated lumber for all repairs Project began 2001/2002, scheduled for completion 2004/2005 Began 2000/2001, as needed Began 200 l/2002, as needed As needed Old structures are on a painting schedule every 3-5 years As needed On-going project Replace old gates as necessary On-going project: most of the old material has been replaced, but this project is on-going due to animal and environmental ~e On-going project: rotten posts have been replaced all over the facility, and many double fences have been constructed to comply with CWD protocols On-going project: completed for all new east side paddocks, maintain existing, construct Old snow fence and construction fence replaced and moved to the outside of the paddock fence (except interior fences), utility wire is systematically being replaced with horse-fence Began 200 l/2002, as needed Replace old range fence and Vmesh, as well as electric fencing in pens that house deer, Construct double fences as required by CWD protocols Began 2002/2003, as needed Construct single and L-shaped, refuge areas to provide refuge and shade, construct hog panel seasonal exclosures to promote vegetation growth in the spring Began 2002/2003, as needed As needed �155 37. Add windscreen to west and south facing fence-lines 38. Mowing and weed control 39.WHL maintenance 40. Unscheduled miscellaneous emergency facility repairs new On-going project On-going project On-going project On-going project Provide additional shaded areas for animals, and maintain existing Began 2002/2003, as needed Seasonal mowing and manual, chemical noxious weed control Provide maintenance assistance to WHL, and support for initial lab construction Emergency repairs to structures, animal holding facilities, perimeter fence, automatic waters, utilities, etc ... As needed Began 2002/2003, as needed As Needed �156 Addendum 1. PROTOCOL FOR MANAGING CHRONIC WASTING DISEASE AT FOOTHILLS WILDLIFE RESEARCH FACILITY Draft Rev. 2003 HISTORY Chronic wasting disease (CWD) is a transmissible spongiform encephalopathy (TSE) or prion disease of cervids (deer and elk). Other TSE's include scrapie of sheep, bovine spongiform encephalopathy (BSE), and Crutzfeld-Jacob disease of humans. The disease causes behavioral changes and loss of body condition and is invariably fatal to infected deer and elk. Despite a comprehensive program initiated in 1985 to eradicate CWD from cervids and the environment at Foothills Wildlife Research Facility (FWRF), CWD remains endemic at the facility. After the 1985 clean-up, CWD was first diagnosed in elk in 1989 and in mule deer in 1994. Natural transmission is now common in mule deer at FWRF and sporadic cases continue to occur in elk. Additionally, natural transmission rates are markedly higher and self-sustaining in paddocks housing infected animals being used in ongoing CWD research studies compared to paddock areas housing animals for other research studies. Based on these observations, guidelines established in 1985 (and revised in 1993 and again in 1997) for maintaining a CWD-free facility are largely obsolete. Here, we provide additional revisions to those guidelines that are directed at maintaining the disease for research purposes in captive deer and elk while minimizing the risk to personnel and the potential spread of CWD outside the facility. OBJECTIVES 1. 2. 3. 4. 5. 6. Prevent transmission or exposure of CWD from FWRF to animals or facilities outside FWRF. Minimize potential for exposing FWRF personnel and visitors to pathogens or potential pathogens including CWD. Maintain endemic CWD in deer at FWRF; however, animals showing end stage clinical signs of CWD will be euthanized to avoid undue suffering, unless directed otherwise by research protocol. Minimize potential spread of CWD among species of captive wildlife (deer, elk and noncervid research animals). Minimize cross contamination between CWD infected and non-targeted research animals. Prevent cross contamination between CWD research treatment groups. ASSUMPTIONS 1. 2. 3. 4. CWD is an infectious disease of deer and elk caused by an abnormally shaped protein prion. CWD is not widespread in free-ranging cervids. Where it occurs, the prevalence of disease varies greatly. Mode of transmission for CWD is not known, and may be direct, via animal/animal contact, or indirect, through contact with excreta (saliva, urine, feces); animate and inanimate objects may serve as fomites (vehicles) in transmitting CWD. Non-cervid wildlife and domestic species are not naturally susceptible to CWD. It is possible that non-cervids could be inapparent carriers of CWD; however, no data have been produced to support this possiblity. Based on patterns seen in other TSE's, it seems likely that if CWD is transmitted to a new host species, then the likelihood of further transmission to others within that species is increased. �157 5. There is no evidence that CWD is transmissible to humans; however, it is prudent to minimize human exposure to CWD as well as animal pathogens known to be transmissible to humans (e.g. Salmonella spp., E. Coli, etc.). APPROACH Overview: 1. Follow established guidelines that prevent contact of captive research animals with animals outside FWRF (wild and domestic). 2. Minimize potential spread of infectious material outside FWRF perimeter. 3. Minimize potential transmission of CWD between species of captive animals, between CWD and non-CWD research animals, between research projects, and between experimental treatment groups where necessary. This includes transmission from mule deer/cattle pens, mule deer/fallow deer pens, therapy mule deer pens, white-tailed deer, and mountain lion pens via contaminated materials or potentially contaminated, equipment, or clothing. 4. Maintain each species of animal in isolation from others, unless directed by research protocol (e.g., mule deer with cattle, mule deer with fallow deer). 5. Educate animal caretakers about CWD (hazards, protocols, and clinical signs exhibited by affected animals). Perform daily animal observations and maintain detailed records of animal health as a portion of the FWRF CWD surveillance program. Animals: 1. 2. 3. 4. 5. Exclude wild or captive cervids from CWD established areas from entering the captive herd, unless directed by a research protocol. Established areas will now include: northeastern, northcentral, and northwestern Colorado, Park, Albany, and surrounding counties in Wyoming, and the Denver Zoo. However established areas are dynamic and may change as surveillance for CWD increases. Therefore, please consult the latest CWD update for guidance. Depending on intended use, orphans, and neonates raised outside FWRF, may be accepted from areas that are CWD established, as well as areas that are not CWD established. These animals will be maintained separately to minimize potential CWD transmission to uninfected neonates that come from sources outside the established area. Raise and maintain each animal species in isolation from others, unless directed by a research protocol. To prevent transmission of CWD from FWRF to facilities where CWD is not established, noncervid species from FWRF will be transferred or donated to other facilities only if the following criteria are met: 1) the transfer location is within the CWD established area, 2) animals are scheduled for a specific research project, 3) the destination is a closed facility (no egress of live animals), 4) animals will not be used in "tame animal trials" in non-confined environments. 5) transfer is approved by the mammal's research leader, 6) recipients will be notified of CWD risks associated with accepting animals from FWRF. Transfers of live cervids from FWRF are prohibited. Animal Maintenance: l . House and maintain each species in isolation from other species, unless directed by a research protocol. 2. House and maintain CWD research animals separate from non-CWD research animals. 3. Maintain accurate records for all animals. This information includes (but is not limited to): birth date, origin, body weights on tractable animals, vaccinations, health problems and treatments, research projects, and movements (intra and inter facility). Additionally, a. Tag all animals for easy individual identification. �158 4. 5. 6. b. Train FWRF personnel to recognize clinical signs of CWD. FWRF personnel will maintain daily animal observation records describing animal status and will report abnormal observations to the facility manager. Where feasible, weigh and/or briefly examine every animal at least once monthly. Wild research animals usually cannot be handled for weighing; these will be visually examined and immobilized via a dart injection for closer examination if necessary. Follow a preventative medicine program that includes routine vaccination, anthelmintic treatment, hoof trimming, nutritional evaluation, and other measures to optimize overall health of research animals. Initiate early detection measures by conducting annual tonsil biopsies on all deer (WTD, MD) housed within the facility. CWD positive animals will be removed at the discretion of the lead project researcher from non-CWD research paddocks and 1) added to the CWD research herd, 2) held in isolation, or 3)humanely euthanized. Use of Research Animals Outside FWRF: l. The transport of non-cervid species from FWRF to facilities or locations, outside the CWD established area is prohibited. 2. The transport of non-cervid species to facilities or locations outside FWRF but within areas where CWD occurs is prohibited unless expressly approved by the mammal's research leader. 3. The transport of cervids outside FWRF is prohibited. 4. Procedures for isolating cervids at other CDOW facilities will be the same as those at FWRF. 5. Animals of any species maintained at FWRF will not be released into the wild. 6. The FWRF Manager is responsible for maintaining accurate records of animals transferred into and out ofFWRF. General Facilities and Equipment: 1. Exclude free-ranging wildlife and livestock from the facility or from contact with captive animals using interior and perimeter fencing. A minimum 4 foot corridor must be maintained between interior pasture fencing and the 8 foot tall perimeter fence surrounding FWRF. The perimeter gates will remain closed at all times, the perimeter fence is inspected monthly, and necessary repairs are made top priority for facility maintenance. 2. Maintain each species of animal separately and allow no direct or fence-line contact unless directed by a research protocol. 3. Minimize runoff between pens housing different species through appropriate pen assignment and drainage control, unless directed by a research protocol. 4. Use drainage control to minimize runoff outside the facility in areas where natural and/or man made drainages occur inside CWD paddocks. 5. Minimize common use of equipment between pens housing different species, between CWD and non-CWD paddock areas, and between CWD treatment groups. When it is necessary to use the same equipment (vehicles) a 20 % chlorine, or 5 % LPH solution can be used to disinfect equipment immediately following the use of equipment inside CWD infected paddock areas. 6. All equipment, materials, organic, inorganic, materials that have been exposed to CWD pathogens must either remain on site or follow EPA treatment guidelines prior to leaving FWRF. 7. Feed and handle animals or clean pens using the following traffic pattern: Clean CWD controls (MD,WTD), non-cervids, elk, non-CWD research mule deer, CWD research mule deer, CWD infected white-tailed deer, mule deer with cattle/fallow deer. Additionally, follow specific protocols for traffic patterns between various CWD research treatment groups. 8. Clean animal pens (especially feed areas and waters) weekly. Dispose of waste from pens housing non-CWD research animals, and clean controls in the main dumpster. Waste from all �159 CWD infected paddocks must never leave the facility. 9. Fecal material and non-palatable feed from CWD research paddocks will be reduced through on-site composting and palatable feed will be recycled to the cattle. 10. Isolation pens, digestion cages, and other areas where animals are held for extended periods, will be cleaned of organic matter and disinfected with a 20% chlorine solution, or 5% LPH solution after use. The researcher last using the area will be responsible for cleanup. Cooperative compliance will be made a condition of all study plans using FWRF ungulates and facilities. 11. Different species may be held concurrently in isolation pens if a buffer zone (empty pen) is used. Feed: 1. Hay will not be accepted from areas where domestic sheep have grazed on cultivated pastures. Personnel: 1. Wash hands before and after handling each species of animal, before and after handling nonCWD and CWD research animals. 2. No eating or drinking allowed in animal areas. 3. Dedicate one pair of shoes/boots to FWRF. Change into/out of this pair of shoes when you arrive at work/when you leave. Alternately, shoes can be sprayed liberally, and/or washed thoroughly in 20% chlorine or 5% LPH solution. 4. Coveralls, boots, and gloves, are required when handling animals showing clinical signs of CWD; and face masks and eye protection are available for use if desired. 5. Coveralls and/or boots are a protocol requirement for CWD infected areas, and CWD control groups. Additionally, each set of treatment groups within a research project may require a separate set of boots and/or coveralls depending on research objectives. Please do not enter a paddock unless you know the protocol. 6. Unsupervised access to FWRF will be limited to authorized personnel. Unauthorized persons will not enter animal pens or be permitted direct contact with research animals. The facility will be locked except when attended during normal business hours. 7. Visitors will be informed that FWRF houses CWD infected animals and is within the CWD established area, and will be given the option of wearing rubber overshoes which will remain on site. 8. All researchers and collaborators and their subordinates will comply with this protocol. All personnel working at FWRF will be required to read this protocol and other appropriate literature and to sign the attached sheet of informed consent. Additional Requirements for CWD research Pens: 1. Protective clothing such as designated boots/shoe covers and/or coveralls and must be worn when entering all pens housing CWD infected animals (currently these are: mule deer/cattle pens, mule deer/fallow deer pens, mule deer therapy pens, infected WTD pens, and mountain lion pens), as well as all CWD control pens. 2. Place waste feed and manure from infected mule deer and white tailed deer pens in the storage compost pile at FWRF (NOT in the dumpster, or working compost piles). Compost will be mixed appropriately and put into composting bins by assigned personnel. Finished compost will be incinerated, or used for topsoil in CWD infected paddocks as needed. 3. Waste feed from the mountain lion pens is disposed of through incineration or sent to CSU for chemical digestion. Fecal material from the lion pens is composted along with other CWD pen waste material. 4. Dedicated (separate) equipment (wheelbarrows, rakes, shovels, water brushes, bucket scrapers, etc.) must be used for cleaning CWD infected vs. CWD control and non-CWD research �160 paddocks. Additionally, separate cleaning equipment may be required for each treatment group within specific research projects. Please ask the facility manager if you are not sure of the cleaning protocol. 5. Vehicles must be cleaned after use in all CWD paddocks. Wash organic material from tires, remove all organic material from the truck bed and disinfect with a 20 % chlorine, or 5% LPH solution. 6. Clean-up procedures following depopulation of a CWD infected paddock: disinfect feed bunks and feed pans in 5% LPH solution and rinse thoroughly, disinfect water receptacle with a 20% bleach solution and rinse thoroughly, rake out all fecal material, spray feed shelter and soil under and around shelter with a 50% bleach solution. Allow all to dry thoroughly before repopulation of paddock. Additional clean-up procedures may be required such as removing the top 6 inches soil around a feed area, soaking with bleach solution, and adding road-base. This will depend on the specific research project. 7. Keep gates to pens, hub/working area, and main east and west gates closed at all times except when passing through. 8. Animal carcasses must be enclosed with a protective cover to contain potentially infectious materials during transportation to the Wildlife Health Lab (WHL) on site, or off site to the CSU Vet Teaching Hospital (VTH) or the Wyoming State Veterinary Lab (WSVL). Alternatively, the truck/equipment could be cleaned with a 20% chlorine solution after use if transported to the necropsy lab on site. 9. Cattle will not leave the facility alive unless transferred to a biosecurity level 2 or greater facility and this requirement is part of a written change to the established research protocol. I 0. Report any abnormalities or accidents immediately to facility supervisor. CWD SURVEILLANCE PROGRAM 1. 2. 3. 4. 5. Euthanize any animal showing clinical signs of CWD and examine tissues grossly and histologically. Perform complete postmortem examination and histologically examine brain tissue of any animal that dies at FWRF. Carcass disposition will be by incineration (required for cattle), chemical digestion, or appropriate burial at the Larimer County Landfill. If CWD is diagnosed in any noncervid species at FWRF, this protocol will be immediately revised and biosecurity at FWRF further increased. The attending veterinarian, facility manager, and Research Facility Animal Care Committee (RFAC) will evaluate and amend this program as necessary. The FWRF CWD PROTOCOL WAS FIRST ESTABLISHED IN 1985 AND REVISED: 1993 1997 2003 INFORMED CONSENT I, - - - - - - - - - - ~ have read the Foothills Wildlife Research Facility (FWRF) protocol concerning chronic wasting disease (CWD) and agree to follow the protocol. Although there is no evidence that CWD is transmissible to humans, I realize that I will be working with research animals and in an environment potentially infected with CWD. I understand that this protocol reflects current knowledge on measures for minimizing exposure to and spread of CWD and other potential pathogens at FWRF. Signature Date �161 LITERATURE CITED Parker, K. L., and B. Wong. 1987. Raising black-tailed deer fawns at natural growth rates. Can. J. Zool. 65:20-23. Wild, M.A., and M. W. Miller 1991. Bottle raising wild ruminants in captivity. Colorado Div. Wild!. Outdoor Facts No. 114. Wild, M.A. 1995. Animal and pen support facilities for mammals research. Colorado Div. Wildl. Res. Rep., WP la, Jl, Jul 1994 - Jun 1995, Fort Collins. Wild, M.A. 1997. Animal and pen support facilities for mammals research. Colorado Div. Wild!. Res. Rep., WP la, Jl, Jul 1996 - Jun 1997, Fort Collins. Wild, M. A., T. R. Spraker, C. J. Sigurdson, K. I. O'Rourke, and M. W. Miller. 2002. Preclinical diagnosis of chronic wasting disease in captive mule deer (Odocoileus hemionus) and white-tailed deer (Odocoileus virgineanus) using tonsillar biopsy. J. General Virol. 83:2629-2634. Wolfe, L. L., M. M. Conner, T. H. Baker, V. J. Dreitz, K. P. Burnham, E. S. Williams, N. T. Hobbs, and M. W. Miller. 2002. Evaluation of antemortem sampling to estimate chronic wasting disease prevalence in free-ranging mule deer. J. Wild!. Manage. 66:564-573 �Colorado Division of Wildlife Wildlife Research Report July 2003 – June 2004 JOB PROGRESS REPORT State of Colorado Project No. Work Package No. 3740 Task No. 3 : : : : Cost Center 3430 Mammals Research Mammals Support Services Animal and Pen Support Facilities for Mammals Research Period Covered: July 1, 2003 – June 30, 2004. Author: T.R. Davis Personnel: M. Anderson, K. Beamer, T. Bogardus, E. Crawford, E. Donegan, M. Dupire, K. Fagerstone, J. Faue, E. Featherman, K. Fox, T. Halasinski, L. Ho, M. Hanusack, G. Harvey, E. Jones, K. Kanapeckas, J. Kint, G. Kyriacou, I. Levan, T. McCollum, A. Mitchell, A. Northrup, M. Paulek, A. Phillips, R. Rhyan, T. Sanders, J. Sirochman, T. Sirochman, J. Stout, T. Stout, D. Thompson, R. Thompson, D. Weaver, A. Wilson All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT The Colorado Division of Wildlife's Foothills Wildlife Research Facility (FWRF) maintained captive animals (2003/2004 annual total: 360) and facilities in support of seventeen captive wildlife research projects. The primary focus of research during this period was chronic wasting disease (CWD) pathology, epidemiology, preventative therapies, sources of transmission in deer and potential transmission to other species. FWRF supported a number of other significant research projects including contraception and reproductive effects, pathogen immunization, evaluation of wildlife capture pharmaceuticals and personnel training in ante mortem sampling and field immobilization. The quality of animal care and facility maintenance provided by temporary, work-study, personal service, intern and volunteer employees is in part reflected by the finding of compliance under the Animal Welfare Act during the annual USDA inspection of FWRF. Herd management practices allowed pronghorn antelope and bighorn sheep herd levels to decline through natural mortality and the remaining domestic ferrets were removed as per study protocols. Alternatively, herd levels of mule deer and white-tailed deer were managed for maximum growth to support on going CWD research. Chronic wasting disease was again a significant source of mortality in mule deer and white-tailed deer and is reflected by the number of CWD research projects conducted during this period. We continue to manage CWD with the philosophy of managing the disease for research purposes under heightened bio-safety guidelines and intensive herd management. Fawn mortalities were higher than expected during the 2004 rearing season however a number of disease causing agents and contributing factors were identified through various diagnostic tests and evaluations. Neonate training was intensified for hand raised animals and training SOP’s developed to accommodate CWD epidemiology research. Administrative actions include compiling a summary of all current and historic FWRF published research, an animal husbandry change order request was 139 �implemented, and the FWRF tour policy revised. New SOP’s were implemented for routine equipment maintenance, seasonal winterizing, tree/shrub care, and the construction maintenance work request forms were revised and reinstated. In addition to routine maintenance, the FWRF team made significant facility improvements including new facilities to accommodate CWD epidemiology research, completion of the mountain lion holding facility, and installation of a perimeter fence around the Wildlife Health Lab. The capitol construction team allocated funds for a new hay barn which is scheduled for replacement in the summer of 2005, and engineering assisted in the development of electronic facility site maps. In addition, the FWRF landowner; Colorado State University approved an easement for the Northern Colorado Water Conservation District to install a 68 inch water pipeline through the center of the facility (north to south). The installation process partially disrupted FWRF management activities for a six week period, but resulted in an upgraded road system, and replacement of existing windrows with five gallon potted trees and shrubs. 140 �JOB PROGRESS REPORT ANIMAL AND PEN SUPPORT FACILITIES FOR MAMMALS RESEARCH T.R. Davis Animal Maintenance: Routine animal husbandry including feeding, health observations, training, weighing, and cleanup, was performed primarily by well trained temporary employees, work-study students, and volunteers. FWRF was inspected by USDA APHIS for compliance with federal animal welfare regulations on July 28, 2004. Table 1 summarizes the number of animals by species reported to USDA animal welfare for the period of October 1, 2003 – September 30, 2004. Table 1. Total number of animals by species reported to USDA Animal Welfare Species Bighorn Sheep Number of animals held, not dedicated to research 11 16 Number of animals dedicated to research 2003/2004 Total 16 15 27 31 0 28 28 75 84 159 19 0 19 1 0 1 45 25 70 0 11 11 167 179 346 201 0 10 10 1 0 1 0 3 3 168 192 360 Elk Fallow Deer Mule Deer Pronghorn Antelope Sika Deer White-tailed Deer Cattle Ungulate Total Ungulate Mean Domestic Ferrets Prairie Dog Mountain Lions Facility Total 141 �The number of animals held but not dedicated to research includes all animals being bred, conditioned, or held for use in research, but not yet used for such purposes. This group consists primarily of breeding animals, and young animals. The relatively high number of ungulates not dedicated to research during this period is the result of a large influx of young animals (born at FWRF, and orphaned neonates) dedicated to the CWD epidemiology study scheduled to begin data collections in the fall of 2004. The total number of animals dedicated to research includes all animals used in experiments at any time during the period. Experiments include those involving no pain, distress, or use of pain relieving drugs, and experiments where pain relieving drugs were necessary to minimize stress on the animal. No animals at FWRF were used in experiments involving pain, without the use of anesthetic, analgesic or tranquilizing drugs. The species total includes all adult animals housed at the facility, neonates born at the facility, transfers into and out of the facility, and all animals that died or were humanely euthanized during the respective fiscal year. It is important to note that ungulate herd levels at any one time averaged approximately 60 percent of the ungulate total and 55% percent of the total number of animals housed at the facility for the entire period. Herd Management: One habituated sika deer and one habituated prairie dog were brought into the facility to support division law enforcement efforts. Mule deer, white-tailed deer, and elk herd levels were expanded through herd management practices and incoming transfers to support CWD and fertility control research. Incoming transfers consisted primarily of habituated adult animals and orphaned neonates obtained from various locations around the state, as well as Wyoming, Nebraska, Iowa, and Kansas. The bighorn sheep and pronghorn antelope herds were reduced through natural mortality and out going transfers as the experiments these animals were dedicated to, reached a stage of completion. Eight fallow deer and the remaining domestic ferrets were removed through planned euthanasia as per the study protocols, while mountain lion and cattle numbers remained constant for the period. Commission approval was granted in 2001 to transfer excess FWRF captive wildlife, and/or orphaned neonates out of state to support collaborative and non-agency wildlife research projects. In 2004 eight pronghorn antelope neonates were transferred to the National Wildlife Research Center (NWRC) in Fort Collins. Two of these animals were orphaned neonates, and six were excess animals of FWRF origin. Other facility transfers include a pronghorn buck that was borrowed from, and returned to, the Sybille Wildlife Research Unit in Wyoming. FWRF herd management practices include planned breeding to maintain optimal population sizes of the various species required to support current and future research projects. Depending on research objectives, some of the offspring from FWRF animals are hand-raised, and various species of wild orphaned neonates are accepted for hand rearing. Habituated weanlings and adult animals are also accepted whenever herd levels will allow. Hand rearing protocols for mule deer are described by Parker and Wong (1987), and by Wild and Miller (1991) for bighorn sheep, elk, pronghorn antelope, and whitetailed deer. Table 3 summarizes the breeding and rearing practices of ungulate species for the period: 142 �Table 3. FWRF Ungulate breeding and rearing practices FWRF Breeding FWRF Neonate Orphan/Transferred 2003 Rearing 2004 Neonates 2004 Hand raised 2 Dam Bighorn Sheep Bred 7 Ewes raised 3 0 Species 2003/2004 Elk Bred 3 Cows Mule Deer Pronghorn Antelope White-tailed Deer Bred 17 Does Bred 4 does Bred 9 does Dam raised 3 Hand raised 16 Dam raised 11 Transferred to NWRC 6 Hand raised 6 Dam raised 7 1 32 0 24 Table 3 does not include three orphan mule deer fawns euthanized on arrival due to severe injuries, or very poor body condition, and five animals (3 mule deer, 2 white-tailed deer) which were still born or died shortly after birth due to parturition complications. The mountain lions, domestic ferrets, and fallow deer are also not included in the above table, as the mountain lions and domestic ferrets were neutered at an early age and the male fallow deer were vasectomies prior to the 2003 breeding season and therefore no breeding occurred in these species. Nutritional Maintenance: Feeding protocols for ungulates previously housed at the facility were reviewed by Wild (1997), and feeding protocols for the fallow deer and mountain lions were described by Davis (2003). The sika deer was maintained on a high quality grass alfalfa mix hay and Regular Ranch-way deer and elk ration. The prairie dog was maintained on Mazuri ADF # 25 herbivore diet, grass/alfalfa mix hay, and fresh vegetables. Individuals of all species maintained reasonable body condition on available diets with the exception of some hand raised neonates (primarily mule deer fawns), and CWD infected animals at the clinical stage of the disease. Fawn mortalities may have been associated with general poor body condition of does infected with chronic wasting disease, the presence of other etiological agents identified (see health maintenance below), and/or interspecies competition for space and cover in paddocks housing cattle and fallow deer. Pen Enrichment: In an effort to provide cover and subsequently reduce stress, additional artificial refuge areas were constructed in paddocks housing semi-wild deer and dam raised neonates. “Y” shaped hide-outs, were constructed on site, vegetation ex-closures were added in early spring and removed later to enhance natural cover, and creep areas with natural cover were provided for dam raised fawns. All pen structure enrichments were readily accepted and utilized by the animals. In addition to pen structure, behavioral enrichment was offered through training. Expanding on the operant conditioning system for mountain lions described by Davis (2003) hand raised ungulate neonates were "treat" trained using the same philosophy. Bighorn sheep, mule deer and white-tailed deer were taught to follow their human trainers and stand on the scale for physical exams, injections, treatments and weighing. Additionally, mule deer and white-tailed deer were gradually conditioned to the metabolic cages in preparation for CWD epidemiology sample collections. Passive training was used in conjunction with the above techniques to habituate animals to the scale and alley-way through supplemental feeding to encourage free exploration without human interference, in these areas. 143 �Health Maintenance: Animal health care was provided as required and as mandated by the preventive medicine program (Wild 1995) and chronic wasting disease protocols. Overall, captive wildlife maintained at FWRF remained healthy throughout the period. Chronic wasting disease (CWD) continues to be a significant source of mortality in captive mule deer and white-tailed deer and is reflected by the number of animals dedicated to CWD research projects throughout this period. Mortality of an adult pronghorn doe was attributed to dystocia, and as described in previous years (Davis 2003) was associated with a failure of the cervix to dilate at the time of parturition. Several cases of dystocia with variable presentations were also observed in mule deer (n=2) and white-tailed deer (n=1). Epizootic hemorrhagic disease (EHD) and bluetongue virus (BTV) were not significant etiological agents during this period and may be associated with a management effort to reduce the quantity of free standing water on the facility, coinciding with the time of documented seasonal peaks of the disease. Mortality rates and disease were higher than expected in hand raised mule deer fawns. Hand raised fawn mortalities were primarily associated with two types of illness: 1.) Intestinal disease resulting in diarrhea, bloating, and/or dehydration accompanied by a general lack of appetite and failure to thrive, and 2.) Respiratory disease (acute bacterial pneumonia) resulting in nasal discharge, coughing, labored breathing, and in some cases, no preliminary signs and acute death. Post mortem sampling and fecal isolation, revealed clostridium perfringens, salmonella, Escherichia coli, and rotovirus. Nasal cultures and post mortem sampling of lung tissue revealed mixed bacterial infections including, Alcaligenes species, Pasteurella species, Pseudomonas aeruginosa, however Arcanobacterium pyogenes was consistently diagnosed and is likely responsible for those cases resulting in acute death. In addition to the etiological agents identified, several management and natural conditions may have contributed to fawn mortalities: 1) Inadequate hospital facilities, and clean isolation areas to separate sick animals (facility carrying capacity), 2) Higher than normal precipitation levels contributing to viable pathogens surviving in the soil for longer periods, and greater exposure to damp/cool conditions, 3) immuno-compromised animals to start with, as FWRF born fawns are exposed to a very pathogen rich environment at birth, and, a high percentage of the hand raised animals were orphans who are often in poor body condition and/or ill when they arrive. Due to animal welfare concerns, management recommends construction of adequate animal holding and hospital facilities prior to hand raising mule deer in the future, as well as a review of the neonate nutrition, health maintenance, and fawn rearing programs. Chronic Wasting Disease: Following the recent revision of the CWD protocol (Davis 2003), we continue to manage CWD with the philosophy of managing the disease for research purposes under heightened bio-safety guidelines and intensive herd management. Intensive herd management is accomplished using the early detection techniques described by Wild et. al (2002) and Wolfe et. al (2002). All animals at FWRF were monitored closely for clinical signs of CWD, and tissues from all mortalities occurring at FWRF were examined for evidence of infection with CWD. Systems Development: Administrative actions include compiling a summary of published articles generated from FWRF research. Hard copies of the 100 + articles filed by date of publication are available at FWRF and the research library. Currently, we are compiling an Access database of the articles to facilitate searches by author, subject, date, etc. In addition, an animal husbandry change order request was implemented as suggested by the Mammals research leader. The change order, modeled after the construction/maintenance work request, was designed to track the origin and justification of facility changes in herd stocking levels, species needs, and basic husbandry techniques including animal care, breeding, rearing, and training practices. 144 �Other administrative actions include the development of new standard operating procedures for routine equipment maintenance, seasonal winterizing, and tree/shrub care. The SOP’s are designed to put all FWRF equipment on a routine maintenance and winterizing schedule, and the schedule is specific to what level of maintenance is necessary at each interval. In the same fashion an SOP was developed for soil moisture testing and watering of tree and shrub windrows. Due to the increasing demands for unscheduled (but necessary and often emergency) construction and maintenance needs, the work request forms were revised and reinstated. The forms were designed to assist in prioritizing and assigning tasks, as well as provide a format for information transfer (an accurate description of the need), and to track labor costs associated with specific projects, routine and emergency maintenance. Educational Contributions: The FWRF tour policy was also revised. The revised policy allows for use of FWRF animals and facilities for hands on training of CDOW employees, collaborators, and other professional groups in sampling techniques and chemical immobilization when pre-approved by the Mammals research leader and/or the Animal Care and Use Committee (ACUC). FWRF functions primarily to support wildlife research, but will no longer function secondarily as an educational facility due to the overwhelming demand for this service. Protecting the integrity of the research, facility management, and increasing animal welfare concerns were sufficient justifications for the policy change. Research Projects: Facility operations offered support for research projects conducted by CDOW personnel and other collaborators that were initiated, conducted, or continued using FWRF animals and facilities. A total of twenty one research projects were supported by FWRF for the period: • • • • • • • • • • • • • • • • • Cattle susceptibility to chronic wasting disease. Susceptibility of fallow deer to chronic wasting disease. Susceptibility of Mountain Lions to chronic wasting disease. Mechanisms of CWD transmission in mule deer. Evaluation of prospective preventative therapies for chronic wasting disease in mule deer. Validation of a potential blood test for chronic wasting disease (GeneThera test). Molecular epidemiology of strain variations in chronic wasting disease. Pathogenesis of chronic wasting disease in white-tailed deer. Effect of copper pathogenesis of CWD in white-tailed deer. Epidemiology of chronic wasting disease: detection of PrPres, shedding, and environmental contamination. Evaluation of third eyelid biopsy for detection of chronic wasting disease infection in mule deer. Survey for chronic wasting disease in cottontail rabbit populations. Leuprolide as a contraceptive agent in female elk: determination of effective minimum dose. Evaluation of GnRH-PAP as a chemosterilant in captive mule deer. I. Effects on animal health. Experimental evaluation of a vaccine for clostridium perfringens type A in captive bighorn sheep (Ovis canadensis) and captive mule deer (Odocoileus hemionus). Training personnel for tonsil biopsy for chronic wasting disease in mule deer. Field Immobilization Training. Facility Improvement Projects: A variety of scheduled and unscheduled maintenance and repair activities were necessary to support facility operation and ongoing research programs. Highlights include construction of new animal holding facilities and restoration of the metabolic cages to accommodate CWD epidemiology research. The mountain lion holding facility was also completed with the exception of the scale and squeeze/treatment area, as we are still in the design phase on this portion of the project. Additional 145 �funding assistance from the Wildlife Health Laboratory (WHL) permitted installation of a separate perimeter fence around WHL. The new fence will allow easier access to the laboratory, while controlling traffic into and out of the animal holding facility. Additional facility modifications include allocation of funding from the capitol construction team to replace the main hay barn. The barn is scheduled for replacement in the summer of 2005, and will be relocated to a central site with better access. The engineering office assisted with the development of electronic site maps of the facility. The site maps show exact locations for buildings and animal holding facilities, as well as locations for all known utilities. The maps will be updated periodically as facility construction and modifications occur. In addition, the FWRF landowner; Colorado State University approved an easement for the Northern Colorado Water Conservation District (NCWCD) to install a 68 inch water pipeline through the center of the facility (north to south). The easement was approved by the CDOW legal staff and included stipulations to maintain the perimeter fence, vehicle access, and keep all excavated soil within the perimeter of FWRF. The installation process partially disrupted FWRF management activities for a six week period, but resulted in an upgraded road system, and replacement of existing windrows with 200 five gallon potted trees and shrubs. NCWCD donated three culverts, built up the road system with excavated dirt, and added road-base which will allow for better water run-off, and should reduce road maintenance costs in the future. Facility maintenance and construction projects were prioritized based on animal welfare concerns and anticipated research needs. Table 3 summarizes the completed, current, and on-going facility construction maintenance projects for the period. 146 �Table 3. Facility Improvement Projects Project 1. CWD Therapy Pens 2. Travel Installation Status Completed Trailer Completed 3. WHL and FWRF Completed Parking Area Improvements 4. D3-6 Feed Area Completed Exclosures /catch areas 5. Electrical Upgrades 6. New Tool Equipment Shed Completed and Completed 7. E4 Site Clean-up Completed 8. Lion Facility Walk-in Completed, Freezer /Cooler Engineering request approved 9. Road Improvements Completed 10. FWRF Electronic Site Maps Completed 11. WHL Water Shut- Completed off Valve 12. Office Septic Pipe Completed Repair 13. WHL Perimeter Completed Fence Details Split 2 pens into 4, Construct 3 new shelters, add 1 automatic water (2 others included in east side plumbing upgrades above) Prepare 3 sites, winterize, hook into electric, water, septic, and purchase propane tank for one, electric and water hook-up for the others, electrical and furnace repair for the third, and misc. repairs for housing, office, and lab space Add road base, gravel, landscaping timbers to expand and improve parking areas Re-set poles, replace range wire and snow fence with Hog panels in MD feed areas and catch areas Increased power needed for expanding facilities- East and West sides, West side upgrades provided by WSVL Provided by NCWCD to replace shed demolished for water pipeline construction Remove 6 top inches of soil, saturate with 20% bleach soln., add 2 inches of road-base to lambing area Add a 10 x 10 cooler, and 10 x 10 freezer unit to the mountain lion complex The road system was built-up with extra dirt to enhance water run-off, 4 new culverts, time and equipt. to build up road system and install culverts were donated by NCWCD Generated electronic site maps from an aerial photo, with all utilities, animal holding facilities, and structures Valve was added to allow shut off the pen and necropsy lab water, while still providing water to the lab Completion Year 2003/2004 2003/2004 2003/2004 2003/2004 2003/2004 2003/2004 2003/2004 2003/2004 2003/2004 2003/2004 2003/2004 Emergency repairs to a cracked septic 2003/2004 pipe Construct a new perimeter fence around 2003/2004 the lab to allow access to the lab without compromising the animal holding facility perimeter fence 147 �Project Status 14. Emergency Fawn Completed Rearing Shelters 15. Equipment storage Completed slab 16. DOD study Completed Facilities 17. Trailer Water Shut- Completed Off Valve Replacement 18. Mountain Lion Completed Facility 19. New roofs/repair On-going project structure on old feedsheds and animal shelters. 20. Add additional On-going project animal shelters 21. Road Maintenance On-going project 22. Paint old building On-going project exteriors 23. Repair/replace On-going project latches, and broken or water damaged alleyway boards 24. Replace walk thru On-going project alley gates Details Dairy calf shelters purchased and installed in waterfowl pen, fences modified to accommodate fawns Pour a concrete slab to store tractor and bobcat attachments out of the mud Convert E7 into 4 pens, add 6 automatic waters, construct 2 alleys, repair N. alley and Dig. Cage ramps, add drainage ditch, double fencing, refurbish the metabolic cages Some materials and labor donated by WSVL Replace leaking water shut-off valve to FWRF travel trailer Utilities, concrete block building, 50 x 60 foot outdoor pen, shift containment system, and 4 indoor dens Approx. ¼ of the old structures and roofs on the facility have been replaced in the last 2 years using treated lumber and long lasting roofing materials Construct additional shelters in pens with heavy stocking rates. (36 ungulate pens on the facility) Road grading and upkeep Now using CCA treated lumber or metal siding for repairs & building replacements to reduce the amount of painting necessary in the future. Now using CCA treated lumber for all repairs Replace old gates as necessary Completion Year 2003/2004 2004/2005 2004/2005 2004/2005 2004/2005 Began 2000/2001, as needed Began 2001/2002, as needed As needed Old structures are on a painting schedule every 3-5 years As needed As needed 25. Replace old visual barrier fencing and utility wire on metal gates On-going project: most of the old material has been replaced, but this project is ongoing due to animal and environmental damage Old snow fence and construction fence Began replaced and moved to the outside of 2001/2002, the paddock fence (except interior as needed fences), utility wire is systematically being replaced with horse-fence 26. Animal holding fence upgrades, and repairs On-going project: rotten posts have been replaced, double fences constructed Replace old range fence and V-mesh, as Began well as electric fencing in pens that 2002/2003, house deer, Construct double fences as As needed required by CWD protocols 148 �Project Status 27. Construct artificial On-going project: refuge areas inside pens completed for all for neonates and adults new east side paddocks, maintain existing, construct new 28. Add windscreen to On-going project west and south facing fence-lines 29. Mowing and weed On-going project control 30.WHL maintenance On-going project 31. Unscheduled On-going project miscellaneous emergency facility repairs Completion Details Year Construct single and L-shaped, refuge Began areas to provide refuge and shade, 2002/2003, construct hog panel seasonal exclosures As needed to promote vegetation growth in the spring Provide additional shaded areas for Began animals, and maintain existing 2002/2003, As needed Seasonal mowing and manual, chemical As needed noxious weed control Provide maintenance assistance to Began WHL, and support for initial lab 2002/2003, construction As needed Emergency repairs to structures, animal As Needed holding facilities, perimeter fence, automatic waters, utilities, etc… LITERATURE CITED Davis, T. R., 2003. Animal and pen support facilities for mammals research. Colorado Div. Wildl. Res. Rep., Jan. 2001 – Jun. 2003, Fort Collins. Parker, K. L., and B. Wong. 1987. Raising black-tailed deer fawns at natural growth rates. Can. J. Zool. 65:20-23. Wild, M. A., and M. W. Miller 1991. Bottle raising wild ruminants in captivity. Colorado Div. Wildl. Outdoor Facts No. 114. Wild, M. A. 1997. Animal and pen support facilities for mammals research. Colorado Div. Wildl. Res. Rep., WP1a, J1, Jul 1996 - Jun 1997, Fort Collins. Wild, M. A., T. R. Spraker, C. J. Sigurdson, K. I. O’Rourke, and M. W. Miller. 2002. Preclinical diagnosis of chronic wasting disease in captive mule deer (Odocoileus hemionus) and white-tailed deer (Odocoileus virgineanus) using tonsillar biopsy. J. General Virol. 83:2629-2634. Wolfe, L. L., M. M. Conner, T. H. Baker, V. J. Dreitz, K. P. Burnham, E. S. Williams, N. T. Hobbs, and M. W. Miller. 2002. Evaluation of antemortem sampling to estimate chronic wasting disease prevalence in free-ranging mule deer. J. Wildl. Manage. 66:564-573 Prepared by ________________________________ Tracy R. Davis, Wildlife Technician 149 � https://cpw.cvlcollections.org/files/original/1a38a7d5bf82e1f047949a7b60aa5f49.pdf 1620fa19343093e23848b9b5c70349a9 PDF Text Text 243 Colorado Division of Wildlife Wildlife Research Report July 2001 and July 2002 PROGRESS REPORT State of~------'C=o=l=ora==d=o_ _ _ __ Division of Wildlife - Mammals Research Work Package No._ _ _3~00~4_ _ _ _ __ Other Ungulate Conservation Task No. - - - - - - - - - - - - - - Annual Winter Count of Middle Park Pronghorn Period Covered: July 1, 2000 - June 30, 2001 Authors: Thomas M. Pojar Personnel: CDOW -T.M. Pojar, R. Firth, I.Claassen, M. Crosby, K. Holinka, T. Kroening, R. Thompson, C. Wagner. Others - C. Cesar (BLM). Volunteers- B. Kraft, R. Nutter, D. O'Sullivan, M. Palowoda. MIDDLE PARK PRONGHORN WINTER 2000-01 COUNT Tom Pojar, February 13, 2001 The annual winter count of the Middle Park pronghorn herd was conducted during February 9 through February 12, 2001. The major effort was done on February 9th when several observation crews from Area 9 assisted and the largest pronghorn groups were counted. The smaller, more isolated groups were counted during subsequent days. Division of Wildlife personnel from Area 9 that participated in the count were: Jerry Claassen, Mike Crosby, Kris Holinka, Tom Kroening, Jim Liewer, Bob Thompson, and Chuck Wagner. Chuck Cesar of BLM assisted as well as the following volunteers: Ben Kraft, Ron Nutter, Dan O'Sullivan, and Marie Palowoda. This year's count was made more interesting with the infusion of animals into the population from the Blue Valley Ranch transplant. The purpose of this transplant was to expand the range of the Middle Park population to wintering areas south of the Colorado River. Throughout the years of habitation, the pronghorn have only wintered north of the Colorado River. BVR pronghorn were released from enclosures they were held in during winter 1999-00 around June 1, 2000. At this point they were free to select the summer, and subsequently, the winter range they found desirable. Of 50 BVR pronghorn radioed, 39 survived to the winter of 2000-01. Twenty-six of these are wintering south of the Colorado River; 15 west ofBVR ~eadquarters west of the Blue River, 2 west ofJim Yust's (west of the Blue River), and 9 east of Junction Butte, which is east of the Blue River. Thirteen of the BVR radioed animals are wintering north of the Colorado River with groups of the "native" pronghorn. COLO ;·;;s·;,;;; •c< :·:: IIIIIIIIIHllllllli BDOW016831 �244 Count North of the Colorado River AREA COUNT Back Troublesome 176 Starr Gulch 221 NE Red Mt. 138 Pinto Ranch 4 TOTAL 539 AREA COUNT Junction Butte - east 13 Blue Valley Ranch - west 28 Jim Yust's ranch - west 2 TOTAL 43 Count south of the Colorado River GRAND TOTAL= 582 The conclusion from this count is that there are at least 582 pronghorn wintering in Middle Park during winter 2000-01. A spreadsheet population model has been maintained for this herd since it was first being tracked in 1986. The projected winter population has corresponded with the actual count quite closely through the years with a mean deviate count of 17. This year, the model projects a population of 675 for the largest discrepancy ever encountered - 93 animals. The model was adjusted for the infusion of BVR animals with the fawn to doe ratio applied to all mature females in the entire Middle Park population. There are several possible scenarios to explain the apparent undercount for this year. No new radios have been deployed on the population north of the Colorado River since 1998. Population growth, radio failure, and harvested radioed animals have contributed to dilution of the proportion of radioed animals. This year about 5% of the population north of the river are carrying working radios. As this percentage decreases the probability of having a group of animals without at least one radio in it to allow detection increases. Finding pronghorn groups with winter conditions featuring a mottled background of snow and sagebrush can be very difficult. Radios are crucial to locating pronghorn groups during winter. All of the radios that were presumed to be working were located except 3. Although these radios were located earlier in the fall during the herd structure survey they may have failed since then or it is possible we missed detecting them during the winter count. The Antelope Creek, Antelope Pass, Cow Gulch, Wolford Mountain, Sulphur Gulch, and Corral Creek areas were searched and radio scanned for these 3 radios. The fact that these radios were not found does not eliminate the possibility that a group (or groups) of pronghorn were missed. Winter conditions are mild thus far this year. The winter began with a much colder than normal November and some snow. However, milder conditions prevailed during December, January, and so far in February. Snow depth where the pronghorn are wintering ranges from 4-10 inches with clear south facing slopes and adequate wind-blown ridges. In brief, the pronghorn should come through this winter in very good condition barring any severe late winter weather. � https://cpw.cvlcollections.org/files/original/ddb68ca712960cb10a55f71e5088ea1c.pdf 113ef8fab45e730df971320245f7827e PDF Text Text Colorado Division of Wildlife July 2009 − June 2010 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 3001 4 Federal Aid Project No. W-185-R : : : : : Division of Wildlife Mammals Research Deer Conservation Development of an Automated Device for Collaring and Weighing Mule Deer Fawns Period Covered: July 1, 2009 − June 30, 2010 Authors: C. J. Bishop, D. P. Walsh, M. W. Alldredge, E. J. Bergman, and C. R. Anderson. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We designed and produced a trap-like device for mule deer that would automatically attach a radio collar to a ≥6-month-old fawn and record the fawn’s weight and sex, without requiring physical restraint or handling of the animal. Our passive collaring device is designed to allow biologists and researchers to radio-collar, weigh, and identify sex of ≥6-month-old mule deer fawns with minimal expense and labor when compared to traditional mule deer capture techniques. This technique should significantly reduce stress that is typically associated with capture and handling and eliminate capturerelated mortality. We collaborated with students and faculty in the Mechanical Engineering Department at Colorado State University to produce a conceptual model and early prototype. We then worked with professional engineers at Dynamic Group Circuit Design in Fort Collins, Colorado, to produce a fullyfunctional prototype of the device. We will conduct an extensive field evaluation of the device with freeranging mule deer during 2010-11. 93 �WILDLIFE RESEARCH REPORT DEVELOPMENT OF AN AUTOMATED DEVICE FOR COLLARING AND WEIGHING MULE DEER FAWNS CHAD J. BISHOP, DANIEL P. WALSH, MATHEW W. ALLDREDGE, ERIC J. BERGMAN, AND CHUCK R. ANDERSON P. N. OBJECTIVE To develop and evaluate a trap-like device for mule deer that would automatically attach a radio collar to a ≥6-month-old deer fawn and record the fawn’s weight and sex, without requiring physical restraint or handling of the animal. SEGMENT OBJECTIVES 1. Work with a professional engineering firm to produce a fully-functional prototype of an automated collaring device for ≥6-month-old mule deer fawns. INTRODUCTION The Colorado Division of Wildlife (CDOW) captures and radio-marks 6-month-old mule deer (Odocoileus hemionus) fawns each year to support research and management of mule deer. Approximately 240 deer fawns are captured annually to monitor survival among 4 populations distributed across western Colorado and an additional 100−350 deer fawns are captured as part of ongoing research studies. Other state agencies in the western United States capture large numbers of mule deer fawns annually also. Most capture is accomplished with net-guns fired from helicopters (Barrett et al. 1982, van Reenen 1982, Webb et al. 2008), which is becoming increasingly expensive (i.e., >$500 per captured deer). Also, net gunning is inherently dangerous with a small market, which at times limits availability of contractors. Drop nets (Ramsey 1968, Schmidt et al. 1978), clover traps (Clover 1956), drive nets (Beasom et al. 1980), and darting (Wolfe et al. 2004) are used occasionally in the western United States to capture deer, but these techniques can be time consuming and labor intensive. Many biologists lack time and resources given other job requirements to conduct such capture operations for any length of time. The increasing cost of helicopter net-gun capture coupled with increasing demand for capturing and radio-collaring 6-month-old fawns has created a need for another capture alternative. Specifically, there is need for a capture technique that is relatively inexpensive to employ considering both operating and personnel costs. In response to CDOW’s capture needs, we conceived the idea of an automated marking device for ≥6-month-old deer fawns that would attach a radio collar and record weight and sex without physically restraining the animal or requiring handling. The idea of automatically attaching radio transmitters to animals is not new, although to our knowledge, there are no proven methods or devices for use on deer or other ungulates. Even a relatively expensive trap or device (e.g., >$5,000 ea.) would reduce CDOW’s capture costs assuming the device could be reused over time with few maintenance expenses. Such a device would enable seasonal wildlife technicians or graduate students to radio-collar samples of deer fawns independently or with little assistance from researchers and biologists because no animal handling would be required. We want the device to record weight and sex because these variables are useful covariates in survival analyses and are typically measured when fawns are captured and handled. A passive marking device would minimize animal stress associated with capture and should have virtually no potential to cause capture-related mortality. The large-mammal capture techniques described 94 �above place considerable, temporary stress on animals as part of netting and handling. Roughly 2-3% of animals typically die from capture-related injuries or stresses under routine capture conditions. Thus, successful development of a passive marking system would reduce CDOW’s operating expenses and improve animal welfare. Therefore, our objective is to design, produce, and evaluate a fully-functional prototype of an automated collaring device for ≥6-month-old mule deer fawns. STUDY AREA We conducted all evaluations with captive deer at the FWRF in Fort Collins, Colorado. We conducted limited evaluations with free-ranging deer near Fort Collins in north-central Colorado. We plan to conduct extensive field evaluations with free-ranging deer in north-central Colorado and elsewhere in Colorado once a fully-functioning device is produced. METHODS We initially wrote a study plan and identified detailed device specifications to guide development of the automated collaring device. We approached Colorado State University’s Mechanical Engineering Department to discuss their interest in helping design such a device. In result, the collaring device became a senior design project for 6 CSU engineering students during the 2008-09 school year. We met with the students weekly and provided them a materials budget of $10,000 to produce a prototype device. We conducted staged evaluations of device components during the year by working with captive deer at FWRF. We also conducted limited evaluations with free-ranging deer near the end of the year. Field evaluations focused primarily on how deer utilized and interacted with the device to guide subsequent design and development decisions. We documented utilization and interactions using direct observation and motion-sensor digital cameras. We relied exclusively on digital cameras when we were not on-site during an evaluation. Automation of the collaring device was disabled any time we were not present to prevent any potential harm to deer. Following preliminary field evaluations, we refined our design specifications and developed a contract with Dynamic Group Circuit Design (DGCD), located in Fort Collins, Colorado, to produce a fully-functional prototype device. We routinely met with electrical engineers from DGCD, and a mechanical engineer subcontracted by DGCD, during the course of the year. These meetings ensured that our device specifications were being satisfactorily met from both engineering and deer biology perspectives. RESULTS AND DISCUSSION We produced a fully-functional prototype device that met our design specifications as set forth in the contract. The prototype device comprises an aluminum cage attached to a bait compartment. Deer enter the device through an adjustable opening at the front of the cage. The adjustable opening can be used to deter entry of larger animals by adjusting both width and height. The sides of the cage comprise one-way gates that prevent entry into the device but allow an animal to exit the device at any point. The bait compartment is accessed through an opening positioned at the rear of the cage. An expandable radio collar is placed in this opening by extending it around four rectangular, aluminum plates that hold the collar in the fully-expanded position (Fig. 1). Radio collars are made expandable by attaching springs to each end of the transmitter; that is, springs are used in place of belting on standard radio collars. Clear plexiglass separates the cage from the bait compartment to maximize visibility. A deer is able to extend its head and neck through the expanded radio collar positioned in the rear opening to access the bait in the bait compartment, which is the only access point to the bait (i.e., it cannot be reached by an animal outside of the device). The floor of the cage is a scale that continuously records weight and informs device operation. Only animals in a specified weight range can be collared, which allows the user to 95 �target fawns and avoid collaring adult deer. Specifically, the mechanism that releases the collar around a deer’s neck will not trigger when an animal is too heavy or too light. Also, an actuator moves a plexiglass plate into the space between the rear cage opening and the bait pan, preventing animals outside of the weight range from accessing the bait. Shortly after a non-target animal exits the device, the collar release mechanism is once again ready to fire and the actuator lowers the plexiglass plate so that the bait is accessible. To prevent an animal from being collared twice, a loop antenna is placed around the entrance to the cage and connected to a radio frequency identification (RFID) reader. All collars used with the device include a small RFID transponder sewn into the collar material. If a previously-collared fawn enters the cage, the RFID transponder is detected, which in turn prevents the collar from being released and activates the actuator to block access to the bait. If a deer enters the cage that is in the specified weight range and has not been previously collared, the collar will release around the deer’s neck once it accesses the bait. The collar release is triggered when a deer’s head breaks an infrared beam positioned immediately above the bait pan. The collar is released by activating a solenoid, which in turn releases a lever that causes the upper 2 aluminum plates holding the expanded collar in place to collapse (Figs. 2 and 3). The collar is then situated around the deer’s neck. When the collar is released, 2 different cameras are immediately activated to take a series of 3 photographs each. One camera is positioned in the back of the bait compartment and set to take a closeup photo of the top of the deer’s head. The second camera is positioned in the floor of the cage and set to take a photo of the deer’s abdomen and groin. These cameras are activated only when a collar is released and facilitate determination of deer sex. Last, when a collar is released, the device records and stores the weight of the deer. An external computer can be hooked up to the device to change program settings, remotely operate the device, and upload weight data. The device is powered by a 12 volt battery that must be recharged every 2-3 days assuming continuous operation. DGCD prepared a user’s manual that explains device operation and detailed schematics to allow future production. We will evaluate effectiveness of the device in the field during 2010-11. Initially, we will only set the device with a collar when we are present and able to directly observe deer interactions with the device. After collaring 5-10 animals in this manner and troubleshooting any problems with the device, we will set the device to operate remotely without an observer on-site, which is how it is intended to be used. SUMMARY We developed a fully-functional prototype of an automated collaring device for mule deer in collaboration with professional engineers. The automated collaring device is designed to allow biologists and researchers to radio-collar portions of their deer samples with minimal time and expense because no animal handling is required and deer can be collared at any time. Primary time commitments include baiting sites, moving device(s) among sites, and adding collars to the devices. The collaring device should also have distinct benefits for studies in urban environments by providing a non-invasive technique for collaring deer. The collaring device should significantly reduce stress that is typically associated with capture and handling and there should be no capture-related mortality. We also have designed the collaring device so that it should be relatively easy to adjust to target adult deer and other ungulate species. Last, the collaring device should have wide applicability for ungulate researchers and managers beyond Colorado. We will be evaluating the device in the field with free-ranging mule deer during the coming year and making additional modifications as necessary. 96 �LITERATURE CITED Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108−114. Beasom, S. L., W. Evans, and L. Temple. 1980. The drive net for capturing western big game. Journal of Wildlife Management 44:478−480. Clover, M. R. 1956. Single-gate deer trap. California Fish and Game 42:199−201. Ramsey, C. W. 1968. A drop-net deer trap. Journal of Wildlife Management 32:187−190. Schmidt, R. L., W. H. Rutherford, and F. M. Bodenham. 1978. Colorado bighorn sheep-trapping techniques. Wildlife Society Bulletin 6:159−163. van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408−421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. Webb, S. L., J. S. Lewis, D. G. Hewitt, M. W. Hellickson, and F. C. Bryant. 2008. Assessing the helicopter and net gun as a capture technique for white-tailed deer. Journal of Wildlife Management 72:310−314. Wolfe, L. L., M. W. Miller, and E. S. Williams. 2004. Feasibility of “test-and-cull” for managing chronic wasting disease in urban mule deer. Wildlife Society Bulletin 32:500−505. Prepared by _______________________ Chad J. Bishop, Mammals Research Leader 97 �Figure 1. View of the radio collar and bait compartment of an automated collaring device for mule deer. To reach bait, deer must extend their head and neck through the expanded radio collar. 98 �Figure 2. View of the collar release mechanism in an automated collaring device for mule deer. 99 �Figure 3. Female mule deer fawn accessing bait by extending her head through an expanded radiocollar. The prototype device will be evaluated extensively in the field with free-ranging deer during 2010-11. 100 � https://cpw.cvlcollections.org/files/original/0e12b1dabc99edcfae1ddf74c39d7285.pdf 52282c9dd1416e8cac8d68b47a07b7d2 PDF Text Text Colorado Division of Wildlife Wildlife Research Report July 2003 – June 2004 JOB PROGRESS REPORT State of Project No. Work Package No. Task Colorado Federal Aid Project: N/A 3740 : : : : Cost Center 3430 Mammals Research Wildlife Diseases Pilot evaluation of GPS technology in chronic wasting disease prevalence and management at artificial feeding sites in urban areas. : Period Covered: April 1 2003 through July 31, 2004 Author: Eric J. Bergman, Michael W. Miller and L. L. Wolfe Personnel: M. Sirochman All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT A pilot study for assessing the utility of GPS technology in the evaluation of CWD prevalence and management in urban areas was designed is being implemented. Objectives of this pilot study are to: 1) Evaluate the utility of GPS radio collar technology in identifying artificial feed sites in urban settings, 2) Evaluate if there is evidence that artificial feed sites reduce the size of deer home ranges, 3) Evaluate if deer density is elevated at artificial feed sites, and 4) Evaluate if CWD prevalence is higher at artificial feed sites . 119 �JOB PROGRESS REPORT PILOT EVALUATION OF GPS TECHNOLOGY IN CHRONIC WASTING DISEASE PREVALENCE AND MANAGEMENT AT ARTIFICIAL FEEDING SITES IN URBAN AREAS Eric J. Bergman, Michael W. Miller and L. L. Wolfe INTRODUCTION Analyses of data from recent field studies and from culling have revealed areas of relatively high CWD prevalence associated with urban areas along the northern Front Range (Wolfe et al. 2002, 2004; Conner and Miller 2004; Farnsworth et al. 2004). Within these, artificial and illegal feeding sites may be particularly important because they appear to congregate deer in one location, thereby increasing local deer density and exposure to contaminated environments (Miller et al. 2004). Although the nature of the relationship between disease prevalence and mule deer density has not been definitively identified, it seems likely (Barlow 1996) that CWD prevalence is being indirectly elevated through artificial feeding. The development of global positioning system (GPS) technology and its incorporation into radio collars for wildlife research presents a tool for better understanding CWD in urban areas. We have initiated a pilot field study to: 1) evaluate the effectiveness of different GPS collars in identifying illegal feed sites in urban settings, and 2) develop and evaluate a strategy for utilizing GPS technology in studying and managing CWD in urban mule deer populations. METHODS The study area for this work is located within two subdivisions in Estes Park, Colorado. The subdivisions, separated by approximately 1.6 km, were identified as treatment and control sites based on the presence and absence of known feeding sites (Fig.1, Wolfe et al. 2004). Between five and eight adult (>1 yr old) female deer from each subdivision were captured and collared with one of two different brands of GPS collars (HABIT Research, British Columbia, Canada and LOTEK Wireless, Ontario, Canada). Collars from each company will be evenly distributed between sites. Capture will occur as part of an ongoing "test and cull" research project (Wolfe et al. 2004) during April 2004 and from August to October of 2004 as needed. Deer will be recaptured and collars will be removed prior to battery failure (~220 days service) in order to retrieve GPS data. No specific hypotheses are being tested in this pilot study; rather, we are attempting to determine if GPS radio collar technology is adequate for use as a tool in refining CWD epidemiology and management. We will record and report on the performance of GPS collars, and calculate costs (mean, range per animal tested) associated with our artificial feed site identification strategy as implemented in this pilot study. However, we will compare home range sizes of deer from each site to determine if artificial feeding reduces home range size of deer. We will also incorporate ground survey data (Wolfe et al. 2004) to estimate and compare mule deer density and ultimately CWD prevalence from sampled deer at each site. CWD prevalence will be compared between sites as well as to previous estimates from the greater Estes Park area (Wolfe et al. 2004) to explore future research potential. 120 �RESULTS AND DISCUSSION GPS Collar Comparison A total of 16 GPS collars (10 LOTEK, 6 HABIT) were available for testing in this study. Prior to initiation of this study no HABIT collars were on hand for deployment, rather, all 6 had to be built to specification and delivered. GPS collars from HABIT Research, ~$1,800/unit, were programmed to: collect GPS locations every 2 hours, to transmit GPS data (via VHF signal) over two day intervals every two weeks and to transmit the most recent GPS location (via VHF signal) at the start of each minute. Due to delays in the manufacturing process, no HABIT collars were received in time for spring deployment (≥2 weeks pre-fawning). Additionally, due to programming errors, 0 of 6 HABIT collars were ready for deployment after initial testing. Upon servicing by HABIT Research (~3.5 weeks), 3 of 6 collars appear to be ready for deployment in late summer 2004. The remaining HABIT collars (3 of 6) will be serviced and deployed upon satisfactory performance. All LOTEK collars were on hand prior to initiation of this study. Eight of 10 collars were deployed in spring of 2004, with 1 of 10 needing service. GPS collars from LOTEK Wireless, ~$3,500/unit, were also programmed to collect GPS locations every 2 hours, but did not offer remote download capabilities. All GPS locations collected by LOTEK collars will be acquired upon retrieval of the collar. GPS Collar Performance Data from LOTEK GPS collars continues to be collected and HABIT GPS collars will be deployed between August-September 2004. LITERATURE CITED Barlow, N.D. 1996. The ecology of wildlife disease control: simple models revisited. Journal of Applied Ecology 33:303-314. Conner, M.M., and M.W. Miller. 2004. Spatial epidemiology in natural populations: a case study of movement and prion disease prevalence relationships among mule deer population units. Ecological Applications (in press). Farnsworth, M.L., L.L. Wolfe, N.T. Hobbs, K.P. Burnham, D.M. Theobald, and M.W. Miller. 2004. Human land use influences chronic wasting disease prevalence in mule deer. Ecological Applications: in review. Miller, M.W., E.S. Williams, N.T. Hobbs, and L.L. Wolfe. 2004. Environmental sources of prion transmission in mule deer. Emerging Infectious Diseases: in press. Wolfe, L.L., M.M. Conner, T.H. Baker, V.J. Dreitz, K.P. Burnham, E.S. Williams, N.T. Hobbs, and M.W. Miller. 2002. Evaluation of antemortem sampling to estimate chronic wasting disease prevalence in free-ranging mule deer. Journal of Wildlife Mangement 66:564-573. _________, M.W. Miller, and E.S. Williams. 2004. Feasibility of "'test-and-cull" for managing chronic wasting disease in urban deer. Wildlife Society Bulletin 32:500-505. Prepared by ____________________________ Eric J. Bergman, Wildlife Researcher 121 � https://cpw.cvlcollections.org/files/original/2c8cfb764ba7ff42f9d3d5a4cdb42e84.pdf d9c11598d5e99b617aeffaeb13dfb64d PDF Text Text Colorado Division of Wildlife Wildlife Research Report July 2003 – June 2004 JOB PROGRESS REPORT State of Project No. Work Package No. Task 2 Colorado Federal Aid Project: N/A : : : : 3004 Cost Center 3430 Mammals Research Other Ungulate Conservation Potential Research Project Assessment : Period Covered: July 1, 2003 through June 30, 2004 Author: Eric J. Bergman and Gary C. Miller Personnel: D. Freddy, C. Bishop, B. Watkins, J. Madison, J. Broderick All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT As part of the research planning process, general assessments were made for the potential to conduct research projects on 3 main topics identified by Colorado Division of Wildlife field management personnel. These topics were: impacts of mule deer/elk interactions on mule deer population performance; improving success of bighorn sheep reintroductions and translocations; and, impacts of natural gas and oil extraction on mule deer. All 3 topics present challenges to conducting successful research endeavors with deer-elk interaction studies potentially providing the most predictable research and funding situations. 89 �JOB PROGRESS REPORT POTENTIAL RESEARCH PROJECT ASSESSMENT Eric J. Bergman INTRODUCTION The Colorado Division of Wildlife is charged with protecting, preserving and enhancing Colorado's ungulate populations for the use, benefit, and enjoyment of the people. The management principles guiding managers in this mission include: wildlife conservation, use and enjoyment, maintaining healthy, diverse and abundant populations and maintaining/conserving habitat quality/quantity through science-based decision making (Colorado Div. of Wildlife 2002-2007 Strategic Plan:9). The objective of ungulate research is to provide information to facilitate the making of these science-based decisions. METHODS Members of the Division’s Mammals Research staff met with wildlife managers and biologists from the Northwest and Southwest Regions to consult on ungulate management issues and research needs. The following topics were identified as the primary statewide issues of concern: ● Negative impacts of deer/elk interactions on deer population performance. ● Variable success of bighorn sheep reintroduction/translocation efforts. ● Impacts of natural gas and oil extraction on deer. In general, research topics identified by wildlife managers favored issues occurring in their respective areas and regions. Topics identified by biologists emphasized need for broad-scale research efforts to address issues that exist throughout the state. The objective of research project selection is to successfully merge these two viewpoints into a study that: 1) Provides clear results through control/treatment experiments. 2) Addresses issues of local and statewide concern. 3) Allows inference to wide spatial/temporal boundaries. RESULTS AND DISCUSSION Impacts of Deer/Elk Interactions on Deer Population Performance A general long-term trend of mule deer population declines has existed both in Colorado and throughout the Western U.S. since the late 1950s (Unsworth et al. 1999, Gill 2001). While insular deer herds have functioned outside of this trend and others have experienced pulses of population growth nested within this long-term decline, declines of as much as 50% have been reported (Colorado Division of Wildlife, unpublished data). Along with these overall population declines, depressed fawn:doe ratios have been simultaneously measured (Gill 2001, White et al. 2001). In an attempt to address concerns over these declines and to develop working hypotheses as to the underlying causes, CDOW hosted a conference for employees of CDOW, CSU, federal agencies, invited publics and deer experts in 1998. A result of this conference was the identification of five potential sources of mule deer population depression: A) habitat deterioration, B) predation, C) competition with elk, D) disease, and E) weather (Fig. 1). Based on a coupling of the ideas from this conference with existing knowledge of mule deer/coyote predation dynamics (Bartmann et al. 1992), CDOW entered a collaborative mule deer 90 �research program with Idaho Fish and Game. Two studies were simultaneously developed. Idaho began a mule deer predation study that is currently assessing the impacts of predator control (both puma and coyote) on mule deer survival. The Division of Wildlife began a nutrition study to simulate and quantify the demographic impacts of habitat enhancement on mule deer fawn survival (Bishop 2003). Through a treatment/control, cross-over study design, CDOW is providing supplemental forage as a mechanism to test the effects of an immediate, short-term enhancement of mule deer habitat (Bishop 2003). Preliminary results of this ongoing research indicate that by enhancing the nutritional value of mule deer diets, overwinter fawn survival can be increased (Bishop 2003). While the methods employed through this research are not an acceptable management strategy to the Division, the pending conclusions exemplify the need for landscape level treatments to enhance mule deer habitat (discussed below). As such, through a series of reductionist experiments, it has been demonstrated that mule deer population performance is directly related to habitat quality, yet functions somewhat independently of coyote predation. However, a confounding factor in these conclusions stems from the fact that concurrent with mule deer declines in Colorado, there has been a dramatic increase in the distribution, number and density of elk. Accordingly, while enhancing the quality of habitat should improve mule deer survival, the ultimate causes of habitat degredation have not been tested. It remains unclear whether interference/exploitation by elk has further reduced carrying capacity for mule deer in late seral stage pinyon-juniper ecosystems. Based on past research, there is variable evidence that competition between mule deer and elk exists. Both interference and exploitation competition have been reported for deer and elk (Singer and Norland 1994, Kirchoff and Larsen 1998, Stewart et al. 2003). However, due to logistical constraints, these studies have been observational and are highly dependent on the specific conditions at which the data was collected. Thus, strong inference is not possible. In order to accurately quantify the impacts of deer/elk competition, a study would need tight control over ungulate densities. Under ideal conditions, this would be done in a controlled setting (i.e. large enclosures). Due to financial constraints and disease potential, this is not possible in Colorado. Additionally, enclosures with controlled densities of deer and elk remove much of the natural variation that exists in nature. Elk are highly mobile animals with weak site fidelity. Thus, while deer may encounter high elk densities one week, they could very easily encounter no elk the following week. Enclosures with static deer and elk densities would remove this variability and subsequent results would be of limited utility. Elk mobility is also the primary reason that density reduction treatments cannot be applied in field settings. A dramatic harvest treatment to reduce local elk density could be erased as elk immigrate into the treatment area once harvest is finished. Given these constraints, the possibility of studying the interactions between deer and elk still exists. Currently, the potential to do this is presenting itself on the Uncompahgre Plateau. The proposed follow-up to the short-term deer nutrition research is the implementation of large scale deer habitat enhancement treatments. However, there is concern that elk will monopolize treatment areas and will either displace deer or deprive deer of receiving the intended treatment. As such, regardless if elk are being studied, mitigation efforts will have to be employed to insure the delivery of treatments to deer. Due to the baseline knowledge on deer in the Uncompahgre system (C. Bishop, unpublished data, B. Watkins, unpublished data), the existing capital investment and given the fact that elk treatments would be an integral part of any deer habitat enhancement study, many of the financial and logistic hurdles to starting a complimentary elk research program would be minimized. 91 �Declining Mule Deer Habitat Quality Bishop Phase I deer/elk research --Collaborative ----- -,- --- --- Enhanced nutrition study showing increased fawn survival l--, NO Are late seral stage pinon/juniper communities conducive to quality deer production? Deer/Elk Interactions I I I enhance nutrition through , Can - -we-habitat - -treatments? -I ........... . ----.l.•······· ........ . =········'·~ Problems: -Will elk displace deer in treatment areas? -Can we successfully apply treatments on large temporal & spatial scales? Has elk population expansion: -displaced deer? -put deer at an exploitation disadvantage? I I I_ - - - - - - - - , . I I No Evidence To Date Is disease depressing deer populations statewide? Have drought conditions negatively impacted deer? NO No clear Evidence Disease Bishop study, 5 statewide study areas Measure species/sex/age class of puma kills using deer & elk radio collars (K. Logan) Do coyotes force poor population performance? Miller, Wolfe, Baeten, Etc. Monitoring of deer mortalities for disease Puma Research & Mgmt. )--------- MULE DEER ISSUES -Low Populations -Poor population parameters Ongoing CWD Research Do puma force poor population performance? Compensatory mortality in a Colorado mule deer population Weather Predation Bartmann et al. (1992) Figure 1. Conceptual model concerning mule deer population declines in Colorado with an emphasis on the relationships between sources of population depression (light gray boxes), current and past research (dark gray boxes) and potential research (dashed lines). A final benefit to conducting a study on the interactions between elk and mule deer on the Uncompahgre Plateau pertains to ecosystem level understanding. As the likelihood of puma research being conducted on the Uncompahgre Plateau increases, the addition of elk research would allow for integrated, broad scale understanding of deer, elk and puma management. Over the past 25 years, wildlife ecologists have strived to understand ecosystem level interactions (Sinclair and Norton-Griffiths 1979, Houston 1982, Jedrzejewska and Jedrzejewski 1998, Krebs et al. 2001). However, a fundamental feature of these studies is that they were conducted in systems that were managed for species conservation (i.e. no hunting), limiting the utility and applicability in systems where species are managed as a consumptive resource. All elements will likely be in place to do ecosystem level research in a system that is managed with a multi-use strategy. The level of interest within the CDOW and other western state agencies in studying deer/elk interactions is high (D. Freddy, unpublished study plan, 1998). Additionally, based on high priority achievements H-1.2 and H-1.3 of the Colorado Division of Wildlife Strategic Plan (2002-2007), research on these issues are highlighted as guiding principles of Division activities. This interest is being echoed by external conservation groups. Over the past decade, the Uncompahgre Project of southwest Colorado has actively pursued habitat enhancement on the Uncompahgre Plateau. This group has expressed interest in coordinating their proposed habitat manipulations with wildlife research, potentially relieving the Division of the financial burden of conducting landscape level treatments (R. Sherman, personal communication). Additionally, it is believed that groups such as the Mule Deer Foundation and the Rocky Mountain Elk Foundation would be willing to invest in research that further explains the impacts of deer/elk interactions as well as the effectiveness of habitat enhancement for these species. 92 �Variable success of Bighorn Sheep Reintroduction/Translocation Efforts During the late 1800's and early 1900's, the rocky-mountain west experienced dramatic declines in bighorn sheep populations (Singer et al. 2000a). Due to the role that bighorns play in ecosystems and their value as both a watched and hunted species, recovery is a high priority. Approximately 55%-58% of the existing populations are a result of reintroduction efforts (Singer et al. 2000b). However, reintroduction efforts have been variable in success (success rates fall between 40% and 58%) (Singer et al. 2000a, Singer et al. 2000b). While these trends and statistics are for the west as a whole, most are indicative of bighorn management efforts within Colorado (Bailey 1990). There are many parameters that influence the success of bighorn sheep reintroduction efforts. Among these, survival, density, metapopulation characteristics, habitat quality/quantity, disease and predation have all been identified as potential limiting factors (Fig. 2). However, due to the lack of experimental manipulation, the relative importance of any single parameter is largely unknown. Bighorn population characteristics are poorly understood, yet are likely important for planning and implementing successful recovery efforts. Many recovery efforts appear to be typified by brief periods of very high population growth, followed shortly by stabilization and population crash (Singer et al. 2000c). In the absence of disease, adult survival is typically high. However, lamb survival can be impacted by density as well as weather (McCarty and Miller 1998, Holl et al. 2004). In the absence of dispersal corridors, and due to potential growth rates as high as λ = 1.30 (Singer et al. 2000c), many recovering bighorn populations suffer from sedentariness. Because bighorns can quickly reach the local carrying capacity of a patch, and due to high adult survival, the instigation of senescence at approximately 7-9 years of age, and low dispersal, it is possible that recovery efforts fail simply because no new individuals are entering the population. A manipulative, research approach to addressing this question is to emulate dispersal through the selective removal of senescent individuals, thereby reducing density and forage pressure such that lamb survival can increase. Bighorn sheep are thought to have traditionally functioned as a metapopulation, with a single population being characterized as a discrete group with limited movement of individuals between groups. While limited in overall quantity, the role of immigration and emigration is vital in metapopulation stability. Dispersal typically occurs in the form of animals leaving one population to join another existing population (i.e. into contiguous, occupied habitat) (Singer et al. 2000a). While providing relief for dense populations, dispersal also provides other populations with new genetic stock. Unfortunately, most recolonization efforts have focused on filling insular habitat patches with a single population and dispersal is not considered. By taking a multi-patch view of the landscape and populating several patches with animals, a metapopulation could be established. Bighorn sheep habitat is typified by open vegetation structures with high visibility and rough terrain to avoid predation (Singer et al. 2000a). Additionally, typical bighorn sheep habitat is composed of climax (late-seral stage) plant communities. These habitats occur naturally (as well as through human influence) in a fragmented fashion across the landscape, further explaining the metapopulation structure. Movement between habitat patches is easily arrested by physical barriers such as rivers and roads (Singer et al. 2000b). Absence of these key habitat characteristics and presence of barriers are all factors that potentially have a negative impact on recovery efforts. Unfortunately, experimental research has not been conducted to determine the actual importance of any single parameter. Disease has also been identified as a potential arresting factor in recolonizing bighorn populations. Epizootic breakouts can reduce bighorn sheep populations by >20%/year (McCarty and Miller 1998). Typically epizootic breakouts are in the form of Pasturella haemolytica, however, parainfluenza-3 and other Mycoplasma outbreaks have also been documented (Singer et al. 2000c). In Colorado, lungworm (Protostrongylus spp.) has also been prevalent. Diseases such as bacterial 93 �pneumonia are often introduced to bighorn populations via exposure to domestic sheep (Goodson 1982). In a massive reintroduction program throughout the west during the 1990’s, a 16km buffer between wild and domestic sheep populations was deemed necessary for habitat to be classified as suitable for bighorn reintroductions (Singer et al. 2000a). Precautionary vaccination programs have been instituted, but reported results have not indicated that such approaches greatly enhance bighorn population recovery (Miller et al. 2000). - - - - - - -, Inoculation Experiment: Harvest I I Experiment: --- I I Can population growth I I be slowed via Can reintroduced animals I be immunized to reduce susceptibility? limited harvest? 1 - - - - - - -1 ~------' ~ Density Does initial population growth exceed carrying capacity at release sites, instituting density dependent limitations prior to initiation of dispersal events? Habitat -------, Livestock Removal Experiment: -- - Disease - Does disease limit growth? - Can disease be eliminated? -Is habitat quality/quantity at reintroduction sites conducive to desired population growth? r----------- Habitat Experiment: 1 -Can we ----I improve habitat by reducing I fragmentation and providing dispersal I corridors? I - - - - - - - - - - _1 Bighorn Issues -Variable success of reintroduction efforts Puma Mgmt. ~-------··► (See Logan’s Program Narrative) ----------, Predator Control Experiment: Metapopulation -Do reintroduction efforts establish metapopulations that function outside of immigration and are thus extremely susceptible to devastational stochastic events -· Can disease vectors be removed? r-----------1 -Does blanket lion removal facilitate re-establishment of bighorn populations? - Is selective removal of offending cats less effective than blanket removal? - - I Reintroduction Experiment: I -- - -long-term --- I -Does a multi-stage, release strategy that emulates immigration increase reintroduction success? I I Predation - Do lions limit population growth in reintroduced populations? Figure 2. Conceptual model concerning the variable success of bighorn sheep reintroduction efforts with emphasis on the relationships between limiting factors (gray boxes) and research possibilities (dashed lines). The role of puma predation on bighorn sheep populations is heavily debated in the literature, but in some circumstances is thought to be the limiting factor in bighorn population growth (Wehausen 1996, Hayes et al. 2000, Holl et al. 2004). In Colorado, the impacts of puma predation on desert bighorn sheep are an issue of concern. For instance, it is believed that recent desert bighorn recovery efforts in the Dolores canyon were unsuccessful due to puma predation. Of the 12 radio-collared animals in this population, 11 have died. Nearly 100% of these deaths were due to predation (those not classified as predation were classified as unknown) (B. Watkins, unpublished data). Due to this preliminary evidence, an experimental research project that addresses the impact of puma predation on desert bighorns would likely be beneficial to future desert bighorn recovery efforts. In summary, there are many management-driven experiments that could be designed to elucidate the effectiveness of different management approaches to the recolonization of bighorn sheep. In fact, many of the potential treatments have been conducted in a non-experimental fashion (see Bailey 1990 for a review). On the western slope of Colorado, the number of potential study sites for bighorn sheep research is high. The Dolores Canyon (west of Durango) has had several failed reintroduction efforts, 94 �despite being classified as viable bighorn habitat. The Escalante, Dominguez and Roubideau canyons east of the Uncompahgre Plateau all have bighorn sheep. However, this population is currently suffering from a Pasturella outbreak (B. Watkins, personal communication). Finally, Colorado National Monument and Debeque Canyon are potential study sites. Because many of these issues take place at the population level, any study would be long-term and broad in scale. Although not explicitly detailed, conservation of bighorn sheep (especially desert bighorn sheep) is loosely prioritized in section S-2 of the CDOW strategic plan (conservation of native species). External funding for this type of research would be difficult to secure. While private groups are interested in bighorn restoration, the level of support needed to address these questions is likely greater than that which they can provide. In terms of logistical support, based on conversations with Division employees from local through the regional levels, interest and support for this research is high. Impacts of Natural Gas and Oil Extraction on Deer and Elk. The impact of natural gas and oil development on deer populations in Colorado appears to be a cyclical issue driven by economics, political policy and current world affairs. Of the impacts that resource extraction can have on deer populations, the two of most immediate concern are space use patterns and population performance (Fig. 3). In terms of space use, deer behavior can shift on broad scales, fine scales or both. For instance, a broad scale shift might manifest itself in the form of animals vacating any area where development is occurring. A fine scale shift might manifest itself in the form of avoiding habitat types, but not abandoning areas of development. Little information for this type of impact has been published. In face of the sparse literature pertaining to the impacts of development on deer and elk, there is information available that pertains to the impacts of development on caribou. Behavioral adaptations similar to those mentioned above have been documented for caribou in response to resource extraction in the arctic (Cameron et al. 1992, Nellemann and Cameron 1998, Dyer et al. 2001, Nellemann et al. 2003). Despite documented shifts in caribou distribution and density, there has been no documentation of negative population level impacts caused by resource extraction. In fact, the number of caribou in the Central Arctic caribou herd increased from 5,000 to 20,000 during oil-field development (between 19751997, Cronin et al. 2000). Similarly, while calving caribou and cow/calf pairs have been observed to avoid roads, a depression of calf survival was not reported (Nelleman and Cameron 1998). While space use behaviors are important, the DOW is primarily concerned about mule deer population performance. Of the population parameters that could potentially be impacted, fawn survival is the most sensitive to disturbance and is subsequently the key parameter for monitoring. Of additional note, measuring changes in the proximate physiological factors that lead to depressed fawn survival may be an avenue for exploring the impacts of resource extraction if monitoring fawn survival is unrealistic. Within the history of Colorado, the impacts of natural resource extraction on wildlife is a subject that has not been ignored. During the 1980's, the U.S. Department of Energy funded research on deer survival on the CA and CB tracts of the Naval Oil Shale Reserve in northwestern Colorado. This research was to be extended to include the impacts of oil-shale development. However, due to financial limitations in the extraction process, development never progressed beyond the initial phase (G. White, personal communication). A pilot study addressing the impacts of natural gas drilling on deer space-use is currently underway in southern Colorado (S. Wait, personal communication). Based on what is known about the current distribution of development, there are essentially two potential study areas in Colorado. There is interest in studying these impacts along the I-70 corridor where gas pad density approaches 1 per 20 acres. Development for natural gas extraction is also either occurring or proposed for the Roan plateau, as well as the Mamm and Divide Creek basins. Unfortunately, the current density of extraction platforms in the Mamm and Divide Creek basins is too high for a controlled experiment. The feasibility of conducting research on the Roan Plateau is unknown. Based on public sentiment reflected in the news media, development on the Roan Plateau is hotly contested. In order for this study to be accomplished, 95 �the Division would likely be placed at odds with this public sentiment due to the need for intense development as a treatment effect. Away from the I-70 corridor, development is also underway in southwestern Colorado (east of Durango, see above). This area likely offers the greatest possibility for advanced research on this issue due to the ongoing dialogue between management agencies, and the advanced stages of research activity. Additionally, because development is progressing at a slower rate in this region, and because much of the proposed development is on the Southern Ute Indian Reservation, there is potential for conducting experimental research. Fawn Survival Is fawn survival negatively impacted by development? -----------:------- ------- I \ _____________ \ Adult Survival :' ' ', ',, ___ Is adult survival negatively impacted by development? • ••. ',,t: ',,_ ', ' \, [ Population Performance Pregnancy Rates What are the population level impacts of development/? Will pregnancy rates be reduced due to stress and/or nutrition? Impacts of Natural Gas and Oil Development on deer and elk populations Temporal Response '' ,, Are the impacts of development permanent, or will deer and elk become accustomed to the change and adapt accordingly? Spatial Response ·········+··················:;:1 _ _ Do deer and elk respond to development by changing space-use patterns? • • ,------ I Resource Selection Do deer and elk respond to development by changing selection patterns? Figure 3. Conceptual model concerning the impacts of oil and gas development on deer and elk populations, with emphasis on potential impacts (gray boxes) and research possibilities (dashed lines). From a greatly simplified viewpoint, natural resource extraction can be broken into two phases. Phase one is primarily composed of the building of infrastructure (i.e. road building, pipeline building, drill pad leveling, pump installation, etc.). The subsequent phase two is composed of steady state resource extraction (Fig. 4). In terms of impacts on deer, phase one is likely to have a strong, negative impact that is relatively short-lived. Infrastructure construction is typically high intensity and may be accompanied by a shift in space use behavior of deer (the longevity of this shift is open to debate). Phase two is typified by the physical extraction of the resource, a highly mechanized process that could require very little human presence on a daily basis. Due to the longevity of phase two, it is likely to have the longest lasting impact on deer (though the impact itself may be more subtle). However, the above described progression of development is an oversimplified scenario of how events could take place. Numerous uncertainties affect the rate of development, many of which are tightly linked to current governing policy and the overall state of the economy. For instance, it is possible that due to economic hardship or due to inefficiencies in the resource refining process, development would be aborted before the completion of phase one (Fig. 5). This sequence of events are similar to those that took place during the 1980's on the CA and CB tracts of the Naval Oil Shale Reserve (G. White, personal communication). 96 �Conversely, it is also possible that regulatory policy could be relaxed during phase one of development. Thus, instead of progressing into phase two, phase one would be closely followed by a period of even more intense development (Fig. 6). The most likely scenario for the rate of development over the next 10-15 years, however, is a merger of both of these last two possibilities (i.e. development that is marked by peaks and valleys driven by the policy and economic events of the time, Fig. 7). As such, any study would have to accommodate a highly unpredictable and sporadic development pattern (Fig. 8). As mentioned above, published experimental research on the impacts of resource extraction has been scarce. The reasons for this can largely be condensed to three primary problems: 1) the lack of experimental control, 2) the necessity for long-term commitment, and 3) cost and logistics. For experimental research to occur, the needed approach would be a treatment/control study design. Due to the fact that the natural process variation of mule deer fawn survival ranges between 0.04 and 0.81 (Unsworth et al. 1999) and to the longevity of this study, a pre/post experimental design would not provide meaningful results. However, through a treatment/control design, the confounding effects of this annual variation would be eliminated. Thus, the issue turns to maintaining clean control/treatment study areas and clean treatment affects. The criteria for a control study area include: 1) quality deer winter range similar to the treatment area (i.e. similar geographic, topographic and botanical composition), 2) close proximity to the treatment study area, such that annual weather patterns are shared between the treatment and control areas, 3) being located far enough from the treatment study area that development activities do not influence deer on the control site, 4) having no pre-existing development, and 5) remaining free of on site and nearby resource extraction/exploration during the 10-15 years covering the experiment. The criteria for an adequate treatment study area are equally complex. In addition to the underlying criteria for a control site, a treatment study area would need: 1) an absence of any pre-existing development, 2) a "phase one" development schedule that is not subject to change (regardless of political, social or economic factors), and 3) a "phase two" period of steady-state extraction and maintenance that is also void of further mining and exploration. Despite great efforts, these criteria would be difficult to meet. Mining companies cannot afford to slow the extraction process if the allowable rate increases, and likewise, they cannot afford to continue pumping if it isn't economically feasible. These economic and social factors that drive production are outside the control of DOW, DNR, high level political officials and the mining companies themselves. Long term commitment is a factor that plagues all long term research programs. However, a failed commitment to see an experiment on oil/gas development to completion would provide very little in terms of knowledge. For many long term research projects the delivery of a treatment occurs in the early stages, only to be monitored in the later years. The treatment in a study concerning oil/gas development would be two-fold, i.e. the treatment (or lack thereof) would be applied every year for 10-15 years. Willingness to maintain this program would need to persist in light of the political and economic changes that will inherently occur. Policy regulating natural resource extraction could become more stringent or more relaxed (discussed above). If the economic potential of development is not realized, long term commitment is not feasible. 97 �Phase Two Phase One HIGH I LEVEL OF DISTURBANCE / . ___..---·-· r-- -·-·- I - ·-· -·- I ·-· -·-·-·- I I I NONE J Figure 4. A conceptual diagram showing the ideal separation between phases of natural resource extraction and subsequent development. Phase one is typified by high intensity, infrastructure construction. Phase two is typified by lower intensity, higher longevity, extraction processes. Phase Two Phase One HIGH I LEVEL OF DISTURBANCE / I .--- --·- ....... \ \ I \ I \ I '--. 1 NONE 1,'_ -..l Figure 5. A conceptual diagram showing potential separation between phases of natural resource extraction and subsequent development. As opposed to ideal conditions, phase one could be cut short due to changes in the economy or changes in regulatory policy. Phase Two Phase One LEVEL OF DISTURBANCE HIGH I I I I NONE Figure 6. A conceptual diagram showing potential separation between phases of natural resource extraction and subsequent development. As opposed to ideal conditions, phase one could be followed immediately by further development, a result of economic incentives or relaxation in regulatory policy. 98 �Phase Two Phase One HIGH LEVEL OF DISTURBANCE / I .---·-·-·-·-·--l I \ I "J I _,..., .T·\ " I. I \ . \ I\ \ \ I \ .. I I\ ·J \: I '· i \\/I \," I NONE V Figure 7. A conceptual diagram showing likely separation between phases of natural resource extraction and subsequent development. As opposed to ideal conditions, phase one will probably be followed by highly fluctuating stages of further development and extraction. The peaks and valleys of this scenario would be driven by normal economic variation and changes in regulatory policy. Phase Two Phase One LEVEL OF DISTURBANCE HIGH NONE Figure 8. A diagram showing the potential development patterns that could be encountered during experimental research on the impacts of resource extraction. The variability shown in phase two is an uncontrollable product of changing economic patterns and changing regulatory policy. The lack of control highlighted by this figure demonstrates the underlying reason that experimental research on this issue is impossible. Cost, the third factor, is also an obstacle encountered during any study. Based on power calculations from other mule deer fawn survival studies, a sample size of 40 marked fawns per study area was deemed necessary to detect a 15% change in fawn survival (Bishop 2000). However, in this example, experimental design increased the power of detecting this difference in excess of 80% because annual results could be pooled over consecutive years. A compiling of consecutive years is not possible when studying the impacts of resource extraction because a carry over effect is present and the impact of development compounds each year. Thus, reductions in fawn survival would need to be captured on a yearly basis. A preliminary power analysis (α=.05, β=.20) using estimates of fawn survival of µ=.444 and SD=.217 (Unsworth et al. 1999) indicated that 60 marked fawns per study area would be needed to have an 80% chance of detecting a 15% change in survival. However, a 15% change in survival is unrealistic in the face of published literature. A more likely expectation is in the realm of a 5%-10% decrease. Thus, more appropriate sample size estimates are between 125 (to detect a 10% change) and 480 (to detect a 5% change) for each of 2 study areas (annual sample sizes would thus range from 250-960 individuals). 99 �Due to a lack of dialogue with personnel from natural gas and oil companies, the DOE, the Colorado Oil and Gas Conservation Commission and the Southern Ute Indian Reservation, the possibility of external funding for this type of research is unknown. Based on past efforts of these entities, it is believed that some level of financial support is possible. The issue of development is addressed by high priority achievement H-1.3 of the CDOW Strategic Plan (2002-2007). While the impacts of natural gas and oil extraction development, per se, are not addressed, they do qualify as a developmental issue of concern. SUMMARY Mule deer/elk interactions, bighorn sheep translocation, and impacts of natural gas and oil extraction are all topics that present suitable and needed research investigations. At this time, moving forward on mule deer/elk interactions appears to be the most reasonable course of action. Cooperative commitments between industry and the State of Colorado appear needed before research could be initiated to meaningfully assess the impacts of natural gas extraction on mule deer populations LITURATURE CITED Bailey, J.A. 1990. Mangement of rocky mountain bighorn sheep herds in Colorado. Colorado Division of Wildlife: Special Report No. 66. Bartmann, R.M., White, G.C. and L.H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monographs No. 121. Bishop, C.J. 2000. Effects of habitat enrichment on mule deer recruitment and survival rates. Program Narrative, Colorado Division of Wildlife. Bishop, C.J. 2003. Effect of nutrition and habitat enhancements on mule deer recruitment and survival rates. Colorado Division of Wildlife: Job Progress Report. Cameron, R.D., Reed, D.J., Dau, J.R., and W.T. Smith. 1992. Redistribution of calving caribou in response to oil field development on the arctic slope of Alaska. Arctic, 45: 338-342. Cronin, M.A., Whitlaw, H.A., and W.B. Ballard. 2000. Northern Alaska oil fields and caribou. Wildlife Society Bulletin, 28: 919-922. Dyer, S.J., O'Neill, J.P., Wasel, S.M. and S. Boutin. 2001. Avoidance of industrial development by woodland caribou. Journal of Wildlife Management, 65: 531-542. Gill, R.B. 2001. Declining mule deer populations in Colorado: Reasons and Responses. Colorado Division of Wildlife: Special Report No. 77. Goodson, N.J. 1982. Effects of domestic sheep grazing on bighorn sheep populations: a review. Proceedings of the biennial symposium of the Northern Wild Sheep and Goat Council 3:287-313. Hayes, C.L., Rubin, E.S., Jorgensen, M.C., Botta, R.A. and W.M. Boyce. 2000. Mountain lion predation of bighorn sheep in the Peninsular Ranges, California. Journal of Wildlife Management 64:954959. Holl, S.A., Bleich, V.C. and S.G. Torres. 2004. Population dynamics of bighorn sheep in the San Gabriel Mountains, California, 1967-2002. Wildlife Society Bulletin 32:412-426. Houston, D.B. 1982. The northern Yellowstone elk herd: ecology and management. Macmillan Publishing Co., New York, New York, USA. Jedrzejewska, B. and W. Jedrzejeswki. 1998. Predation in vertebrate communities: the Bialowieza Primeval Forest as a case study. Springer-Verlag, New York, New York, USA. Kirchoff, M.D. and D.N. Larsen. 1998. Dietary overlap between native Sitka black-tailed deer and introduced elk in southeast Alaska. Journal of Wildlife Management 62:236-242. Krebs, C.J., S. Boutin and R. Boonstra. 2001. Ecosystem dynamics of the boreal forest: the Kluane project. Oxford University Press, Oxford, England. McCarty, C.W. and M.W. Miller. 1998. Modeling the population dynamics of bighorn sheep: a synthesis of literature. Colorado Division of Wildlife: Special Report No. 73. 100 �Miller, M.W., J.E. Vayhinger, D.C. Bowden, S.P. Roush, T.E. Verry, A.N. Torres and V.D. Jurgens. 2000. Drug treatment for lungworm in bighorn sheep: Reevaluation of a 20-year-old management prescription. Journal of Wildlife Management 64:505-512. Nellemann, C. and R.D. Cameron. 1998. Cumulative impacts of an evolving oil-field complex on the distribution of calving caribou. Canadian Journal of Zoology, 76: 1425-1430. Nellemann, C., Vistnes, I., Jordhøy, P., Strand, O., and A. Newton. 2003. Progressive impact of piecemeal infrastructure development on wild reindeer. Biological Conservation, 113: 307-317. Sinclair, A.R.E., and M. Norton-Griffiths. 1979. Serengeti: dynamics of an ecosystem. University of Chicago Press, Chicago, IL, USA. Singer, F.J., Bleich, V.C. and M.A. Gudorf. 2000a. Restoration of bighorn sheep metapopulations in and near western national parks. Restoration Ecology 8(4 special supplement):14-24. Singer, F.J., Moses, M.E., Bellew, S. and W. Sloan. 2000b. Correlates to colonizations of new patches by translocated populations of bighorn sheep. Restoration Ecology 8(4 special supplement):66-74. Singer, F.J. and J.E. Norland. 1994. Niche relationships within a guild of ungulate species in Yellowstone National Park, Wyoming, following release from artificial controls. Canadian Journal of Zoology 72:1383-1394. Singer, F.J., Williams, E., M.W. Miller and L.C. Zeigenfull. 2000c. Population growth, fecundity, and survivorship in recovering populations of bighorn sheep. Restoration Ecology 8(4 special supplement):75-84. Stewart, K.M, Bowyer, R.T., Kie, J.G., Dick, B.L. and M. Ben-David. 2003. Niche partitioning among mule deer, elk, and cattle: Do stable isotopes reflect dietary niche? Ecoscience 10:297-302. Unsworth, J.W., Pac, D.F., White, G.C. and R.M. Bartmann. 1999. Mule deer survival in Colorado, Idaho and Montana. Journal of Wildlife Management 63:315-326. Wehausen, J.D. 1996. Effects of mountain lion predation on bighorn sheep in the Sierra Nevada and Granite Mountains of California. Wildlife Society Bulletin 24:471-479. White, G.C., D.J. Freddy, R.B. Gill and J.H. Ellenberger. 2001. Effect of adult sex ratio on mule deer and elk productivity in Colorado. Journal of Wildlife Management 65:543-551. Prepared by ______________________ Eric J. Bergman, Wildlife Researcher 101 � https://cpw.cvlcollections.org/files/original/dd6011b6a5d8b5f4d9d7bae2220575de.pdf 7fc4c9456bf7e6bb238d96c0e5d4e3a2 PDF Text Text 163 JOB PROGRESS REPORT Stateof_ _ _ _ _--=C=o=lo=r=ad=o=------ Division of Wildlife - Mammals Research Work Package No. _ _ _ _ _ _ _ _ _ __ Multispecies Investigations Task No. _ _ _ _ _ _ _ _ _ _ _ _ __ Prairie Dog Research and Wildlife Extension Period Covered: July 1, 2002 - June 30, 2003 Author: W. F. Andelt, Colorado State University, Dept. Fishery and Wildlife Biology Personnel: M. Christopher, J. Dennis, L. Gepfert, E. Hollowed, BLM, S. M. Quinlivan, P. M. Schnurr, A. Seglund, Utah Division of Wildlife Resources, G. C. White, D. Younkin PRAIRIE DOG AND PREDATOR-GROUSE RESEARCH, AND WILDLIFE EXTENSION W. F. Andelt OBJECTIVES I. Objectively assess and document the current scientific knowledge base about Gunnison's prairie dogs by I September 2002 via a technical review draft publication, submitted to the CDOW research peer review process. 2. Conduct on-the-ground surveys, and collect measurements of key elements of Gunnison's prairie dog colonies at 50 sites in western Colorado by September 30, 2002, and provide a report, including data summaries, by October 30, 2002, to CDOW's project leader. By October 30, 2002, provide a data set that can be used by other investigators to develop a defensible, quantified Gunnison's prairie dog inventory technique. 3. Provide information specifically directed toward chronic wasting disease from DOW/DNR to the public through CSU's Extension network of 57 county Extension offices and provide intensive training to at least 4 offices and I 00 employees/volunteers in key western slope counties by April 30, 2003. 4. Provide general wildlife information and information regarding human-wildlife conflicts from DOW/DNR to the public through the CSU's Extension network of 57 county Extension offices and provide intensive training to at least 4 offices and 100 employees/volunteers by April 30, 2003. 5. Provide analyses of data on the possible role of predators in the sage grouse decline in northwest Colorado. �164 STATUS OF GUNNISON'S PRAIRIE DOGS IN COLORADO W. F. Andelt The Gunnison's prairie dog (Cynomys gunnisoni) occurs in Colorado, Arizona, New Mexico, and Utah. Their geographical range probably has not changed much during the past century (Knowles 2002). However, acreage of Gunnison's prairie dogs within their range likely has contracted during the past century. The extent of decline is unknown because there were no accurate accounts of the abundance of prairie dogs prior to settlement (Clark 1973, Anderson et al. 1986, Knowles 2002), and the abundance of Gunnison's prairie dogs today also is not well known. Approximately 22% of the range of Gunnison's prairie dog occurs in Colorado (Knowles 2002), where it is distributed primarily across the southwestern quarter of the state at elevations of 6,000 to 12,000 feet (Fitzgerald et al. 1994). The Gunnison's prairie dog consists of 2 subspecies (C. g. gunnisoni and C. g. zuniensis). In Colorado C. g. gunnisoni occurs in the Gunnison River drainage, the upper Arkansas and South Platte drainages, and in the San Luis Valley (Fitzgerald et al. 1994). In Colorado, C. g. zuniensis occurs at lower elevations in Montezuma, La Plata, Dolores, San Miguel, and Montrose counties (Fitzgerald et al. 1994). Densities of Gunnison's prairie dogs range from 5 to 10 per acre (Knowles 2002). The primary threat to Gunnison's prairie dogs is plague (Yersinia pestis), whereas poisoning, recreational shooting, agricultural land conversion, and urbanization are of secondary importance (Knowles 2002). Plague became apparent in Gunnison's prairie dog colonies during the late 1940s (Lechleitner et al. 1968, Cully 1993). Plague often kills >99% of Gunnison' s prairie dogs (Lechleitner et al. 1968). South Park, Colorado apparently contained 913,000 acres of Gunnison's prairie dogs in 1941, but an epizootic of sylvatic plague entered this area in 1947, and by 1949 plague reduced the acreage of prairie dogs by 95% (Ecke and Johnson 1952, Fitzgerald 1969, Armstrong 1972). Plague has continued in this area during the 1950s and 1960s (Lechleitner et al. 1962, Fitzgerald and Lechleitner 1974). During the first half of the 20 th century, Gunnison's prairie dogs were mostly eliminated from the major valleys in Colorado (Burnett and McCampbell 1926, Longhurst 1944) due to plague or poisoning (Knowles 2002). Recently, most wildlife biologists interviewed by Knowles (2002) felt that plague was the dominant controlling factor of prairie dogs. Recover of Gunnison's prairie dogs from plague appears to range from no recovery to a pattern where colonies are regularly lost, but new colonies appear and grow in other areas (Knowles 2002). Gunnison' s prairie dogs were subject to poisoning in the higher valleys of Colorado during the 1950s (Lechleitner et al. 1968). Control of Gunnison' s prairie dog continues on private land, but control of prairie dogs on Federal lands currently does not appear to be a conservation issue (Knowles 2002). The current abundance of Gunnison's prairie dog in Colorado is not well known. Some biologists (Fitzgerald 1991), environmental proponents, and other individuals have expressed concern that populations of Gunnison' s prairie dogs have been reduced by epizootics of plague (Lechleitner et al. 1962, 1968; Fitzgerald 1969, 1978, 1993; Rayor 1985), and control of prairie dogs (Fitzgerald 1991) in Colorado. Speculation exists that the Gunnison's prairie dog might be petitioned for listing as threatened or endangered. Decisions to list the Gunnison's prairie dog should be based upon the most accurate and most current data. In this report, I summarize information from various sources about the status of Gunnison's prairie dog in Colorado. �165 Colorado Agricultural Statistics Service (1990) Survey Colorado Agricultural Statistics Service (1990) surveyed 9,046 farmers and ranchers and obtained nearly 3,000 surveys to estimate that 1,553,000 acres were occupied by prairie dogs in Colorado during 1989. This survey estimated acres occupied by prairie dogs in each county, but it did not differentiate between acres occupied by Gunnison's prairie dogs, black-tailed prairie dogs (Cynomys ludovicianus), and white-tailed prairie dogs (Cynomys leucurus). Thus, I used distribution maps in Fitzgerald et al. (1994) to ascertain which counties were occupied by the 3 species of prairie dogs. In counties where Gunnison's prairie dogs overlapped with 1 of the other species of prairie dogs, I estimated the relative proportion of the county that was occupied by Gunnison's prairie dogs. I multiplied that proportion by the acreage reported occupied by all prairie dogs in a county to obtain an estimate of the acreage occupied by Gunnison's prairie dogs for that county. I summed the acres ofreported Gunnison's prairie dogs in each county and obtained an estimated 445,500 acres ofreported Gunnison's prairie dogs in Colorado during 1989 (Table 1). Jim Fitzgerald (1991) letter to Galen Buterbaugh, U.S. Fish and Wildlife Service Fitzgerald (1991) expressed concern about the status of the gunnisoni subspecies of the Gunnison's prairie dog. He indicated that plague and poisoning have eliminated almost all populations in South Park. He also indicated populations appear to be in poor condition in the San Luis Valley, they appear to be gone from the extreme upper Arkansas River valley, and populations appear to be small and patchy in other parts of its historic range in Colorado. He believed Gunnison's prairie dogs are gone from Jefferson, Douglas, and Lake Counties in Colorado. He noted that a large complex exists on the Curecante National Recreation Area west of Gunnison, Colorado. Fitzgerald (1991) sent inquiries to all Colorado Division of Wildlife District Wildlife Managers and Wildlife Biologists and reported that a disappointing number of colonies were identified. He indicated that the low number of reports of colonies sent to him by the Colorado Division of Wildlife and his low estimates are in direct contrast to acreage of Gunnison' s prairie dogs reported by Colorado Agricultural Statistics Service ( 1990). Robert Finley (1991) Survey of Distribution and Status of Gunnison's Prairie Dogs in Colorado Finley ( 1991) conducted a broad reconnaissance survey of the distribution of Gunnison' s prairie dogs by driving some highways and roads and recording observations of prairie dogs. He observed 74 Gunnison's prairie dog colonies, of which 42 were active. He recorded colonies in 10 counties. He reported the largest active colonies were in the Gunnison drainage. He reported that South Park was almost devoid of prairie dogs, but he found a medium sized colony near Hartsel and a few on the periphery. He indicated that some mammalogists suspect that the spread of Wyoming ground squirrels southward through Colorado, after prairie dogs die out from plague, may be preventing prairie dogs from repopulating their former towns east of the Continental Divide and north of the Arkansas River. Finley (1991) concluded that populations of Gunnison's prairie dogs "seem to be far below those reported in the years prior to plague epizootics", "but I do not feel that the present situation is serious enough to warrant protection by Threatened status." �166 Mike Threlkeld, Chief of Rodent Control, Colorado Department of Agriculture (Personal Communication, 11 June 2002) Mike Threlkeld indicated that there are large acreages of Gunnison's prairie dogs around Cortez (perhaps 7,000 acres), Dolores, Montrose (perhaps 7,000 acres), Blue Mesa Reservoir, between Dove Creek and Nucla/Naturita, west of Canyon City, north of Salida, and on the Ute Mountain Indian Reservation (perhaps over 7,000 acres). Colorado Division of Wildlife (2002) Report on Acreage of Gunnison's Prairie dogs Field personnel from the Colorado Division of Wildlife, Forest Service, and the Bureau of Land Management placed Gunnison's prairie dog colonies on 1:50,000 US Geological Survey County sheets during July and August, 2002 (Colorado Division of Wildlife 2002). The colonies were assigned as active (prairie dogs know to be present in the last 3 years) or unknown status (prairie dogs have been active but current presence in the area is unknown and requires field verification). From this exercise, the Colorado Division of Wildlife (2002) reported 85, 795 acres of active and 194,777 of unknown acres of Gunnison's prairie dogs in Colorado (fable 1). In addition, 53,832 acres of active prairie dogs were identified in Delta County, where it was not know if these acres represented Gunnison's or white-tailed prairie dogs. These acreages are considered preliminary minimum estimates of the number of acres occupied by Gunnsion's prairie dogs. Craig Knowles (2002) Report on Status of White-tailed and Gunnison's Prairie Dogs Knowles (2002) primarily summarized Colorado Division of Wildlife (2002) for his assessment of the current status of Gunnison's prairie dogs in Colorado. He criticized the Colorado Agricultural Statistics Service (1990) report of acreage of prairie dogs in Colorado by stating " ... these estimates clearly greatly inflate the acreage at least in some counties." However, it is worth noting that Knowles (1998) reported that there was only 44,000 acres of black-tailed prairie dogs in Colorado during 1998, whereas the Colorado Agricultural Statistics Service (1990) estimated about 930,000 (calculated from their report). Recent aerial surveys by the Colorado Division of Wildlife (following Sidle et al. 2001) indicate that there are about 631,000 acres occupied by black-tailed prairie dogs in Colorado (F. Pusaterie, personal communication). Thus, the estimates provided by Colorado Agricultural Statistics Service (1990) were much closer than Knowles (1998) to the acreage reported by the Colorado Division of Wildlife. Knowles (2002) indicated that Gunnison's prairie dog populations in Colorado were greatly reduced by plague and poisoning during the 1900s and this decline may be continuing, or at best, the populations may be stable. Synthesis of Reports on Abundance of Gunnison's Prairie Dogs in Colorado Abundance of Gunnison's prairie dogs likely has declined in Colorado, particularly starting during the 1940s when plague became endemic. Our best estimates of the acreage ofGunnison's prairie dogs in Colorado seem to be provided by Colorado Division of Wildlife (2002) and Colorado Agricultural Statistics Service (1990). The Colorado Division of Wildlife (2002) reports a preliminary minimum of 85,700 acres of active Gunnison's prairie dogs, another 194,800 acres ofGunnison's prairie dogs where their status is unknown, and another 53,800 acres of prairie dogs in Delta County which are either Gunnison's or white-tailed prairie dogs. The Colorado Agricultural Statistics Service (1990) survey of acreage of prairie dogs in Colorado during 1989, from which I derived 445,500 acres of reported Gunnison's prairie dogs, has been criticized as biased by Knowles (1998, 2002). However, Colorado Agricultural Statistics Service (1990) and Colorado Division of Wildlife (2002) seem to concur at least to some extent. The Colorado Division of Wildlife is assessing the feasibility of aerial surveys for estimating acreage of Gunnison's prairie dogs in Colorado. Pending feasibility, these surveys are needed to provide better estimates of the acreage of Gunnison's prairie dogs in Colorado. �167 LITERATURE CITED Anderson, E. A., S. C. Forrest, T. W. Clark, and L. Richardson. 1986. Paleobiology, biogeography, and systematics of the black-footed ferret, Mustela nigripes (Audubon and Bachman), 1851. Great Basin Naturalist Memoirs 8: 11-62. Armstrong, D. M. 1972. Distribution of mammals in Colorado. Museum of Natural History, University of Kansas Monograph 3. 415pp. Burnett, W. L., and S. C. McCampbell. 1926. The Zuni prairie dog in Montezuma County, Colorado. Office of State Entomologist, Colorado Agricultural College, Fort Collins, Colorado. Circular 49, 16pp. Clark, T. W. 1973. Prairie dogs and black-footed ferrets in Wyoming. Pages 88-101 in Linder, R. L., and C. N. Hillman, editors. Proceedings of the black-footed ferret and prairie dog workshop. South Dakota State University, Brookings, South Dakota. Colorado Agricultural Statistics Service. 1990. Vertebrate rodent infestation survey. Colorado Department of Agriculture, Lakewood, Colorado. Colorado Division of Wildlife. 2002 (September). Report of acreages of active colonies for Gunnison's prairie dogs (Cynomys gunnisoni) and white-tailed prairie dogs (Cynomys leucurus). 2pp. Cully, J. F., Jr. 1993. Plague, prairie dogs, and black-footed ferrets. Pages 38-49 in J. L. Oldemeyer, D. E. Biggins, B. J. Miller, and R. Crete, editors. Proceedings of the symposium on the management of prairie dog complexes for the reintroduction of the black-footed ferret. U.S. Fish and Wildlife Service Biological Report No. 13. Ecke, D. H., and C. W. Johnson. 1952. Plague in Colorado. Part I. Plague in Colorado and Texas. U.S. Public Health Service, Public Health Monograph 6:1-54. Finley, R. B., Jr. 1991. Survey of present distribution and status of Cynomys gunnisoni gunnisoni in Colorado. Unpublished manuscript. 9pp. Fitzgerald, J.P. 1969. Sylvatic plague in Gunnison's prairie dog (Cynomys gunnisoni) and associated mammals in South Park, Colorado. Journal of the Colorado Wyoming Academy of Science 7:45. Fitzgerald, J. P. 1978. Plague (Yersinia pestis) epizootics in introduced Gunnison's prairie dogs: implications for prairie dog management. New Mexico Academy of Science Bulletin 18:40. Fitzgerald, J. P. 1991. Letter to Galen Buterbaugh, Regional Director, U.S. Fish and Wildlife Service. Dated 1 April 1991. Fitzgerald, J. P. 1993. The ecology of plague in Gunnison's prairie dogs and suggestions for the recovery of black-footed ferrets. Pages 50-59 in J. L. Oldemeyer, D. E. Biggins, B. J. Miller, and R. Crete, editors. Proceedings of the symposium on the management of prairie dog complexes for the reintroduction of the black-footed ferret. U.S. Fish and Wildlife Service Biological Report No. 13. Fitzgerald, J.P., and R.R. Lechleitner. 1974. Observations on the biology ofGunnison's prairie dog in central Colorado. American Midland Naturalist 92:146-163. �168 Fitzgerald, J.P., C. A. Meaney, and D. M. Armstrong. 1994. Mammals of Colorado. University Press of Colorado, Niwot, Colorado. 488pp. Knowles, C. J. 1998. Status of the black-tailed prairie dog. Unpublished manuscript prepared for U.S. Fish and Wildlife Service, Pierre, South Dakota. 12pp. Knowles, C. 2002. Status of white-tailed and Gunnison's prairie dogs. National Wildlife Federation, Missoula, Montana and Environmental Defense, Washington, DC. 30pp. Lechleitner, R.R., L. Kartman, M. I. Goldenberg, and B. W. Hudson. 1968. An epizootic of plague in Gunnison's prairie dogs (Cynomys gunnisoni) in south-central Colorado. Ecology 49:734-743. Lechleitner, R. R, J. V. Tileston, and L. Kartman. 1962. Die-off of a Gunnison 's prairie dog colony in central Colorado, I. Ecological observations and description of the epizootic. Zoonoses Research 1:185-199. Longhurst, W. 1944. Observation on the ecology of the Gunnison prairie dog in Colorado. Journal of Mammalogy 25:24-36. Rayor, L. S. 1985. Dynamics of a plague outbreak in Gunnison's prairie dogs. Journal ofMammalogy 66:194-196. Sidle, J. G., D. H. Johnson, and B. R. Euliss. 2001. Estimated areal extent of colonies of black-tailed prairie dogs in the northern great plains. Journal of Mammalogy 82:928-936. �169 Table I. Acres of Gunnison's prairie dogs reported and estimated from the Colorado Agricultural Statistics Service (1990) survey during l 989 and estimated by Colorado Division of Wildlife (2002) during 2002 in Colorado. Colorado Agricultural Statistics Service survey Acres of all County Qrairie dogs Alamosa 6,200 48,900 Archuleta 3,200 Chaffee 20,500 Conejos Costilla 1,600 Custer 5,900 Delta 52,500 Dolores 56,000 Douglas 12,600 16,700 El Paso Fremont 15,300 5,800 Gunnison 300 Hinsdale Huerfano 6,400 Jefferson 1,700 Lake 900 La Plata 80,000 Las Animas 18,500 Mineral 200 Montezuma 92,000 Montrose 52,100 Ouray 7,400 Park 5,100 Rio Grande 14,300 Saguache 13,200 San Juan 13,400 San Miguel Teller 5,200 TOTAL: 555,900 Proportion acres• occupied by Gunnison's Q.dogs 1.0 1.0 1.0 1.0 1.0 1.0 0.12 1.0 0.25 0.05 1.0 1.0 1.0 0.63 0.31 1.0 1.0 0.2 1.0 1.0 0.73 0.5 1.0 1.0 1.0 1.0 1.0 1.0 Acres of Colorado Division Gunnison's of Wildlife Qrairie dogs Active acres Unknown acres 6,200 2 12,220 48,900 18,226 15,978 3,200 2,467 0 20,500 4,707 67,218 1,600 14,948 25,439 5,900 6,300 56,000 3,363 2,549 3,150 58 0 835 15,300 5,800 611 221 300 4,032 527 900 80,000 6,816 619 3,700 200 449 1,221 92,000 12,223 0 38,033 6,482 0 3,700 647 0 5,100 42 3,150 14,300 12,263 2,094 13,200 2,659 58,891 13,400 5,200 448,277 2,017 ____fil_ 85,795 2,927 - -0 194,777 •obtained by estimating the proportion of a county (from Fitzgerald et al. 1994) that was occupied by Gunnison's prairie dogs, white-tailed prairie dogs, and black-tailed prairie dogs, and then dividing the proportion for Gunnison's prairie dogs by the sum of proportions for all 3 species. �170 EVALUATION OF AERIAL SURVEYS FOR ESTIMATING ACREAGE OF GUNNISON'S AND WHITE-TAILED PRAIRIE DOGS IN COLORADO AND UTAH W. F. Andelt, P. M. Schnurr, and A. Seglund During November 2002, we (Andelt and Schnurr 2002) reported our assessment of 3 survey techniques, including ground surveys, interpretation of satellite imagery (Sidle et al. 2002), and aerial surveys (Sidle et al. 2001 ), for obtaining a valid estimate of the distribution and acreage of Gunnison' s prairie dogs (Cynomys gunnisoni) in Colorado. We concluded that ground surveys likely would be very difficult, if not impossible to implement for obtaining a valid scie~tific estimate of acreage of Gunnison' s prairie dogs in Colorado. However, we recognized that ground surveys could be used to provide an estimate of the minimum acreage of Gunnison's prairie dogs in Colorado. We concluded that satellite imagery is very expensive ($2,000 per 36 mi2 or $2,880 per 100 mi 2 of digital imagery [John Norman, Natural Resources Ecology Lab, CSU; personal communication]), the imagery would need to be interpreted and verified, activity of prairie dog towns would need to be ascertained on the ground, and it is unknown if the technology would be suitable in rolling terrain. Aerial surveys, using line intercept methodology, have been used to estimate area occupied by black-tailed prairie dogs (Cynomys ludovicianus) (Sidle et al. 2001, J. Dennis and F. Pusaterie, Colorado Division of Wildlife; personal communication). We concluded that the technique held promise for estimating acreage of Gunnison' s prairie dogs in Colorado. In this paper, we report on our current progress in evaluating aerial surveys for estimating acreage of Gunnison's and white-tailed prairie dogs (Cynomys leucurus) in Colorado and Utah. Initially, on 13 June 2002, William Andelt accompanied Jim Dennis and Dave Younkin on an aerial survey of black-tailed prairie dogs to gain additional familiarity with the technique. On 24 June 2002, William Andelt and Larry Gepfert, CDOW, flew over the 32 active Gunnison's prairie dog colonies reported by Joe Cappodice. With the aid of a GPS unit, all colonies were located, although some of the smaller colonies were somewhat difficult to observe. We ascertained that aerial surveys appear to have potential for establishing distribution of Gunnison' s prairie dogs and that further investigation of the technique was merited. However, because of some difficulty in observing some colonies, we, in collaboration with Gary White, decided that future test flights should also obtain photos of prairie dog colonies; classify colonies as being located in grassland, short shrubs, tall shrubs, or agriculture; rank the colonies as barely detectable, detectable, or highly detectible; and classify colonies as active, inactive, or unknown. Our plans were to use these data to estimate detection probabilities for the various categories of colonies. We then planned to use the detection probabilities to correct acreages of prairie dog colonies observed from the air (White 2002). Subsequently, during summer 2002, Pam Schnurr and Gary White met with Amy Seglund and Bill Bates, biologists with the Utah Division of Wildlife Resources (UDWR). Both states agreed to coordinate and cooperate to further ascertain the feasibility of aerial surveys to estimate acreage of Gunnison's and white-tailed prairie dogs, and to develop detection probabilities for both species. METHODS We entered the boundaries of known Gunnison's and white-tailed prairie dog colonies in both Colorado and Utah into GIS Arc/Info. We established 31, 17, 19, and 11 transects across these Gunnison's and white-tailed prairie dog colonies in Colorado and Utah, respectively. These transects were established across known colonies in both states along with a number of control transects (i.e. transects over areas without colonies). Beginning and ending UTM coordinates were ascertained for each transect and placed in a spreadsheet. We hired and trained a ground crew that verified the distribution of all white-tailed prairie dog colonies on the transects in Colorado. �171 Jim Dennis and Dave Younkin, CDOW, and Brad Crompton and Craig Hunt, from the Utah Division of Wildlife Resources flew all 4 sets of transects and obtained GPS coordinates for the beginning and end of prairie dog colonies on the transects. The crew from Colorado had extensive experience surveying black-tailed prairie dogs, whereas the crew from Utah had extensive experience with aerial surveys of wildlife, other than prairie dogs. The Utah and Colorado survey teams flew the transects in opposite directions. We plotted the endpoints of the prairie dog colonies that were ascertained by both aerial crews on all transects in GIS Arc/Info. We used Arc/Info to determine the lengths of each colony on each transect and then entered these data in a spreadsheet. We summed the lengths of colonies ascertained on the ground and from the air on each transect. We analyzed these data in SAS using Proc GLM to determine the effect of aerial team, rating of colony visibility, and rating of habitat type on the proportion of colonies observed on aerial versus ground surveys. We censored transects without prairie dogs known on ground surveys, and then used Spearman Correlation (Proc CORR) analyses to ascertain correlations for proportion of colonies observed, ratings of visibility, and ratings of habitat types between the 2 aerial crews. We also used Spearman Correlation analyses to ascertain correlations between ratings of visibility of colonies and proportion of colonies detected, and ratings of habitat types and proportion of colonies detected. RESULTS The Colorado and Utah teams overestimated lengths of Gunnison's prairie dog colonies on transects in Colorado and Utah (Table l ). Both teams also overestimated lengths of white-tailed prairie dog colonies on the white-tailed site in Utah. In contrast, the Colorado team underestimated lengths of colonies on the white-tailed site in Colorado. Although the Utah team closely estimated the overall average lengths of colonies on this site, we found considerable variation between total lengths of colonies on transects observed by this team versus those known on the ground. The Utah aerial team (x = 5.3; S.E. = 1. ll), compared to the Colorado team (x = 2.3; S.E. = 0.36), observ'ed a greater proportion oflengths of colonies on transects (Tables 1, 2), however both teams significantly overestimated the lengths of e<;>lonies compared to the lengths ascertained on the ground. The proportion of length of prairie dog colonies observed from the air compared to the lengths ascertained from the ground were not related to ratings of visibility nor to ratings of habitat types observed from the air (Table 2). Proportion of lengths of prairie dog colonies detected by aerial crews from Colorado and Utah were weakly correlated (Table 3). However, ratings of visibility of colonies and ratings of type of habitat found on transects of colonies were not correlated between the Colorado and Utah aerial crews. The 2 crews did not consistently report finding prairie dogs in the same areas along the same transect. This may partially explain the differences between the 2 crews in their ratings of visibility of colonies and rating of habitat types on transects. Proportions of lengths of colonies detected by aerial crews were not correlated with rating of visibility of colonies on transects (Table 4 ). The greatest proportions of lengths of colonies were detected by aerial crews on transects described as grasslands followed by transects described as short shrubs and then followed by transects described as tall shrubs (Table 4). The Colorado team rated prairie dogs on 76% of 51 transects as active, 12% as unknown, and 12% as a combination of active and unknown. The Utah team rated prairie dogs on 28% of 63 transects as active, 2% as inactive, 57% as unknown, and 25% as a combination of active, inactive, and unknown. �172 DISCUSSIONS AND RECOMMENDATIONS We recognize a number of goals when inventorying prairie dogs. We believe the most important goal is to obtain accurate and repeatable estimates (i.e. low variation within and among survey crews) of the acreage of Gunnison' s and white-tailed prairie dogs. Low variation among survey crews is necessary so that differences between estimates of acreage are actually related to increases or decreases in acreage of prairie dogs rather than differences between crews. Another goal for inventorying prairie dogs is to establish minimum acreages of prairie dogs which we can relate to their status and decisions about listing them as threatened or endangered. Our goal has been to ascertain the feasibility of aerial surveys for estimating acreage of Gunnison's and white-tailed prairie dogs in Colorado and Utah. We envisioned this as a multi-step process. We first flew over known Gunnison's prairie dog colonies and noted that many of the colonies were visible from the air. Next, we arranged aerial surveys by crews from Colorado and Utah to estimate . the length of colonies on transects where the distribution of prairie dogs were known to us, but unknown to the crews. Accuracy of aerial surveys was not sufficient to estimate detection probabilities. We found significant variation between the 2 aerial teams in estimates of lengths of prairie dog colonies on transects, however these estimates were weakly correlated between the 2 teams. Shortly after completing the aerial flights and before data were compiled, Jim Dennis noted that his team likely could have more accurately estimated lengths of prairie dog colonies by conducting some flights followed by ground reconnaissance of the same transects to verify what they were observing from the air (see Appendix 1). We anticipate this training would enhance accuracy of estimates. We recommend that training, or other methods to improve estimates between teams, are needed before broad scale surveys are conducted. The large variation between teams in our study indicate that, without improving accuracy and consistency between teams, it would be difficult to ascertain even moderate changes in acreages of prairie dogs. The Colorado and Utah teams surveyed the Colorado white-tailed prairie dog site on 20 September and 28 August 2002, respectively. The Colorado team rated 10 of the transects as active and 2 as unknown. The Utah team rated 4 transects as active, l as inactive, 5 as unknown, and 4 as activeinactive or active-unknown. We surveyed part of the Colorado white-tailed site from the ground on 23 September 2002 and found very little sign of activity by prairie dogs. Thus, we recommend that ground crews verify ratings of activity on a random sample of future transects. If aerial crews are unable to accurately determine activity, a ground crew will need to verify activity on a random portion of transects on future surveys. We reviewed potential causes for why estimates oflengths of prairie dog colonies varied between ground surveys and aerial surveys, and between the 2 aerial crews. We closely surveyed the distribution of prairie dogs on the white-tailed sites in Colorado and Utah, but additional verification on the ground is needed for the 2 Gunnison's prairie dog sites to insure that accuracy of ground surveys is not a cause of error. Coordinates of prairie dog colonies were recorded on the ground and by the Utah team in the NAD27 datum. The Colorado team used the WGS84 datum when they flew the transects. The use of the WGS84 resulted in the Colorado team being 38 to 219 m off the actual transect, depending on the study area and direction of flight (east-west versus north-south). Although we initially suspected that the 38 to 219 m away from transects resulted in some errors, our review of the data suggested that accuracy appeared similar when the airplane was on the transect versus away from the transect. The Utah team strayed over 1,000 m from portions of 4 transects which likely attributed to some errors. �173 We recognize 2 general approaches (ground vs. aerial surveys) for continuing surveys of Gunnison's and white-tailed prairie dogs. To continue aerial surveys, we recommend that the distribution of prairie dogs is more accurately verified on the ground on the 2 Gunnison's prairie dog sites. If distributions are different than what is currently known, the distribution of prairie dogs on aerial and ground surveys should be compared again. Then, we recommend training aerial crews by conducting flights over short transects over some colonies and then surveying the colonies from the ground so that they can better ascertain what they are observing from the air. After this training, we recommend reflying the previous transects to ascertain if accuracy can be improved. If accuracy cannot be improved, we recommend discontinuing aerial surveys. An alternative to surveying prairie dogs from the air would be to continue Pam Schnurr's earlier work of meeting with biologists to plot known distribution of Gunnison' s and white-tailed prairie dogs on maps. A ground crew should then verify a random portion of these distributions. Although this alternative likely would cost less than aerial surveys, it likely would underestimate acreage of prairie dogs and would not provide an adequate and repeatable sample for future comparisons. However, this methodology might be sufficient for considerations of listing prairie dogs as threatened or endangered. LITURATURE CITED Andelt, W. F., and P. Schnurr. 2002. Progress report: inventorying Gunnison's prairie dogs in Colorado. Unpublished Report submitted to the Colorado Division of Wildlife, Denver, Colorado. Sidle, J. G., D. H. Johnson, and B. R. Euliss. 2001. Estimated areal extent of colonies of black-tailed prairie dogs in the northern great plains. Journal of Mammalogy 82:928-936. Sidle, J. G., D. H. Johnson, B. R. Euliss, and M. Tooze. 2002. Monitoring black-tailed prairie dog colonies with high-resolution satellite imagery. Wildlife Society Bulletin 30:405-411. White, G. C. 2002. Memorandum to Pam Schnurr, Bill Bates, and Amy Seglund, Colorado Division of Wildlife and Utah Division of Wildlife Resources. Dated July I, 2002. �174 Table l. Average length (m) of Gunnison's and white-tailed prairie dog colonies, observed from the ground and reported by aerial survey crews from the Colorado Division of Wildlife and the Utah Division of Wildlife Resources, on transects surveyed in Colorado and Utah during August, September, and November 2002. Transects Avg. Avg. length of colonies/transect• Proportion of colony length observedb Team survey N lenlrth Colo 9/19-20 31 8,671 Utah 10/l 31 8,671 Colo 9/20 17 5,446 Utah 8/28 17 5,446 Colo 9/23 19 10,660 19 10,660 Utah 8/28 9/24 Colo 11 40,403 11 40,403 Utah 8/26 Ground Aerial 264 723 246 1,5 l l 1,955 1,202 1,984 1,955 424 1,770 424 3,406 2,912 9,714 2,912 5,418 N 18 18 14 14 11 11 8 8 X 2.6 8.4 0.7 1.8 3.5 7.5 2.7 1.7 S.E. 0.65 2.57 0.16 0.80 0.97 2.09 0.85 0.44 51 51 2.3 5.3 0.36 1.11 Date of Area Colorado Colorado Colorado Colorado Utah Utah Utah Utah S12ecies Gunnison's Gunnison's White-tailed White-tailed Gunnison's Gunnison's White-tailed White-tailed TOTAL: Colo Utah 78 78 12,928 12,928 1,045 1,045 2,350 2,626 aRepresents average length of colonies known primarily from ground reconnaissance, and estimated from aerial surveys on transects with and without prairie dog colonies. bRepresents proportion of length of prairie dog colonies observed from aerial surveys divided by lengths ascertained from ground reconnaissance on transects with prairie dog colonies. Table 2. Effects of aerial teams 3, ratings of visibility of colonies\ and ratings of habitat typesc on proportions oflength ofGunnison's and white-tailed prairie dog colonies observed on aerial transects during August, September, and November 2002. Independent variable Aerial teams Rating of visibility Rating of habitat type 3 df 1 4 5 F 6.79 0.57 0.48 p 0.011 0.684 0.793 Aerial team from Colorado Division of Wildlife and from Utah Division of Wildlife Resources. 113arely detectible, barely detectible-detectible, detectible, detectible-highly detectible, highly detectible. cGrassland, grassland-short shrub, short shrub, short shrub-tall shrub, tall shrub, agricultural. �175 Table 3. Correlations between aerial crews from the Colorado Division of Wildlife and the Utah Division of Wildlife Resources for proportions of lengths of prairie dog colonies detected, ratings of visibility", and ratings of habitat typesb on aerial transects of Gunnison' s and white-tailed prairie dogs observed during August, September, and November 2002. • Colorado team Variable Proportion of colony length detected Rating of visibility of colony Rating of habitat type on colony X N 51 30 22 Utah team X N 51 30 22 S.E. 0.36 0.11 0.12 2.3 2.4 2.2 S.E. 1.11 0.10 0.08 5.3 2.5 1.4 r0.301 -0.020 -0.066 p 0.032 0.916 0.769 •1 = barely detectible, 1.5 = barely detectible-detectible, 2 = detectible, 2.5 = detectible-highly detectible, 3 = highly detectible. bl = grassland, 1.5 = grassland-short shrub, 2 = short shrub, 2.5 = short shrub-tall shrub, 3 = tall shrub. Table 4. Correlations between ratings of visibility" and proportions of prairie dog colony lengths detected, and ratings of habitat typesb and proportions of prairie dog colony lengths detected on transects of Gunnison's and white-tailed prairie dogs combined by aerial crews from the Colorado Division of Wildlife and the Utah Division of Wildlife Resources combined during August, September, and November 2002. Visibility/Habitat Variable Visibility versus proportion of colony length detected Habitat versus proportion of colony length detected Proportion of colony detected N X S.E. N X 77 2.4 0.07 77 4.5 65 1.7 0.07 65 4.3 r p 0.76 0.038 0.742 0.88 -0.246 0.048 S.E. •1 = barely detectible, 1.5 = barely detectible-detectible, 2 = detectible, 2.5 = detectible-highly detectible, 3 = highly detectible. bl = grassland, 1.5 = grassland-short shrub, 2 = short shrub, 2.5 = short shrub-tall shrub, 3 = tall shrub. �176 Appendix 1. Suggestions for Aerial Surveys (from Andelt and Schnurr 2002). Based upon our flight with Larry Gepfert and suggestions from Jim Dennis and Dave Younkin we have developed a number of suggestions for aerial surveys of Gunnison' s prairie dogs and white-tailed prairie dogs: • Elevation and overall range distributions (Armstrong 1972, Fitzgerald et al. 1994) should be ascertained before aerial surveys are conducted to minimize the area that needs to be surveyed. • Flight crews should spend at least I day on the ground in Gunnison's prairie dog and white-tailed prairie dog towns to become more familiar with the towns before they fly transects. The crews should also gain experience by flying over known colonies. After flying over known colonies, the crew should spend some time on the ground in a colony to better ascertain what they have seen from the air. • Transects should be constructed along drainages, instead of across drainages, to minimize changes in elevation while conducting surveys. Further, transects should be flown down the drainage, instead ofup drainages, to maximize aircraft maneuverability while minimizing danger. RECOMMENDED PLANS FOR FUTURE • Complete ground surveys to establish the remaining "known" boundaries for white-tailed prairie dog colony transects already flown in Colorado. Compare known and aerial estimates of the locations of prairie dog colonies to ascertain accuracy of aerial surveys. • Ascertain if a correction for detection probabilities will need to be employed. This will be primarily needed if the aerial crews were unable to observe a significant proportion of the "known" colonies. • Determine strata boundaries utilizing recent WRIS mapped activity areas and elevation limits for prairie dogs to minimize the extent of surveys. • Establish transect lines along drainages and within strata. • Determine who will conduct aerial surveys in Colorado. We suspect that we will need to contract with a commercial company. • Ascertain if prairie dog colony activity can be determined from the air. If colony activity cannot be determined from the air, a subset ground sampling technique will need to be established to determine activity. During September field trips to the white-tailed colony in Colorado, we were unable to ascertain activity of many colonies because many prairie dogs apparently entered hibernation early this year due to the drought (Dean Biggins, personal communication). �177 PRELIMINARY EVALUATION OF SURVEYS OF PLOTS FOR ESTIMATING OCCURRENCE OF WHITE-TAILED PRAIRIE DOGS IN COLORADO AND UTAH W. F. Andelt, G. C. White, P. M. Schnurr, and A. Seglund Our research (see above) indicated that aerial line-intercept surveys likely will not work for reliably estimating acreage of Gunnison's and white-tailed prairie dogs. Thus, during Spring 2003, we established a pilot project and surveyed 19 500 by 500 m plots from the ground and air to ascertain if surveys of plots can be used to ascertain trends in occurrence of white-tailed prairie dogs in Colorado and Utah. We focused on white-tailed prairie dogs because they have been petitioned for listing as a threatened or endangered species, but we also plan to expand this methodology for Gunnison's prairie dogs. METHODS We overlaid 7.5 minute topo maps (NAD27 datum) in GIS with 500 by 500 m grid lines on each of 4 study areas (Wolf Creek and Grand Valley, Colorado, Coyote Basin and Cisco, Utah) where locations of prairie dogs were identified. After reviewing the maps and visiting with colleagues familiar with distributions of prairie dogs in each of the 4 study areas, we visited grids (plots) in the field and choose 6 plots in Wolf Creek and 6 plots Coyote Basin such that 2 had low, 2 had medium, and 2 had relatively high abundance of prairie dogs. Also within this classification, 1 of each of the low, medium, and high abundance grids had low visibility and the other high visibility. We also established 4 plots in the Grand Valley, near Grand Junction, and 3 plots near Cisco, Utah in areas with relatively low abundance of white-tailed prairie dogs. During June 2003, we visited the 4 comers of most study plots 3 times each to establish detection probabilities, with 1 visit during 0700-1100, another visit during 1100-1500, and another visit during 1500-1900 hrs. For each study plot, we recorded the investigator's name, date, time, lITM Zone, GPS coordinates for the lower left (SW) comer of the plot, percent cloud cover, and soil type (from a soil survey map), approximate precipitation during last 24 hours, and approximate precipitation during last 30 minutes. For the 4 comers of each plot, we recorded temperature, wind direction, approximate wind speed, percent of plot that was visible, percent of plot in sunshine, rating of visibility, rating of elevation, number of mounds observed, and groups of prairie dogs observed. On 12 June, William Andelt flew over each study plot to ascertain if prairie dogs could be reliably detected in plots from aircraft. We also hired a commercial company to photograph, with high resolution, 9 by 9 inch, color infrared film, 21 study plots to ascertain its feasibility for establishing occurrence of prairie dogs in plots. RESULTS AND DISCUSSION We are currently analyzing data from our pilot observations of white-tailed prairie dogs within 19 study plots. Initial results indicate that we should be able to reliably monitor occurrence and detect changes in occurrence of white-tailed prairie dogs by visiting plots from the ground. We plan to establish about 300 (based upon computer simulations) random plots within the range of white-tailed prairie dogs in Colorado. We plan to hire a field crew and visit these plots to ascertain occurrence of prairie dogs during spring and summer 2004. Our flight over 21 study plots indicate that an airplane might be used to establish occurrence of prairie dogs in high density plots, especially on warm days with snow cover during spring. We will evaluate aerial photographs after they are developed. We also plan to conduct a pilot study, during spring 2004, of the above methodology for ascertaining occurrence of Gunnison's prairie dogs in Colorado. We are currently writing a proposal which will detail our subsequent work. �178 CHRONIC WASTING DISEASE W. F. Andelt We established links, on my web site {http://www.coopext.colostate.edu/wildlife/, then, go to Diseases), to 9 sites that contain information on chronic wasting disease. I informed all extension personnel, including all county extension agents, in Colorado about the availability of this information on my web site. I also informed 153 Cooperative Extension volunteers at 3 training sessions in Colorado, that information on chronic wasting disease was available on my web site. My web page on Diseases was accessed 701 times during January-June, 2003. EXTENSION'INFORMATION ON RESOLVING HUMAN-WILDLIFE CONFLICTS W. F. Andelt My Cooperative Extension activities included: Refereed Publications: Yoder, C. A., W. F. Andelt, L.A. Miller, J. J. Johnston, and M. J. Goodall. 2003. Effectiveness of twenty, twenty-five diazacholesterol, avian gonadotropin releasing hormone, and chicken riboflavin carrier protein for inhibiting reproduction in Coturnix quail. (Submitted to Poultry Science). Refereed Publications In Preparation: Schwartz, A. M., and W. F. Andelt. Effects of castration on reproduction and social structure in the black-tailed prairie dog (Cynomys ludovicianus). (Manuscript is 95% completed, will be submitted to the Journal of Wildlife Management). Schwartz, A. M., and W. F. Andelt. Effects of castration on body mass and survival in the blacktailed prairie dog (Cynomys ludovicianus). (Manuscript is 95% completed, will be submitted to the Journal of Wildlife Management). Heffernan, D. J., W. F. Andelt, and J. A. Shivik. Coyote exploratory behavior following removal of novel stimuli. (Manuscript is 95% completed, will be submitted to the Journal of Wildlife Management. Book Chapters: Lamb, B. L., R. P. Reading, and W. F. Andelt. 2003. Public attitudes and perceptions toward black-tailed prairie dogs. Pages _to_ in J. L. Hoogland, editor. Conservation and management of prairie dogs. Island Press, Washington, D.C. (Submitted 2nd draft). Andelt, W. F. 2003. Methods and economics of managing prairie dogs. Pages_ to_ in J. L. Hoogland, editor. Conservation and management of prairie dogs. Island Press, Washington, D.C. (Submitted 3rd draft). �179 Extension Publications: Andelt, W. F. 2002. Impacts of drought on wildlife. lpp. (Published at http://drought.colostate.edu/). Andelt, W. F., S.N. Hopper, and M. Cerato. 2002 (revised). Preventing woodpecker damage. Cooperative Extension Bulletin, Colorado State University, Fort Collins. Spp. (Published at http://www.ext.colostate.edu/PUBS/NATRES/pubnatr.html). Andelt, W. F. 2003. Preventing woodpecker damage to trees. The Green Scene (July, In press). Cerato, M., and W. F. Andelt. 2003 (revised). Coping with skunks. Cooperative Extension Bulletin, Colorado State University, Fort Collins. 5pp. (In press; will be published at http://www.ext.colostate.edu/PUBS/NATRES/pubnatr.html). Cerato, M., and W. F. Andelt. 2003 (revised). Coping with snakes. Cooperative Extension Bulletin, Colorado State University, Fort Collins. 6pp. (Published at http://www.ext.colostate.edu/PUBS/NATRES/pubnatr.html). Progress Reports: Andelt, W. F., and P. Schnurr. 2002. Progress report: inventorying Gunnison's prairie dogs in Colorado. Progress report submitted to Gary Miller, Colorado Division of Wildlife, 7 November 2002. 7pp. Andelt, W. F. 2003. Status of Gunnison's prairie dogs in Colorado. Progress report submitted to Gary Miller, Colorado Division ofWildlife, 13 January 2003. !Opp. Andelt, W. F., P. Schnurr, and A. Seglund. 2003. Evaluation of aerial surveys for estimating acreage of Gunnison's and white-tailed prairie dogs in Colorado and Utah. Progress report submitted to Gary Miller, Colorado Division of Wildlife, 24 February 2003. 13pp. Papers Presentation at National. Regional. and State Meetings: Andelt, W. F. 2003. Alternatives to toxicants for managing conflicts with black-tailed prairie dogs. Colorado Prairie Dog Technical Conference, Fort Collins, Colorado (Invited paper). Andelt, W. F. 2003. Behavioral modification of coyotes to reduce predation on livestock. Department of Fisheries and Wildlife, Utah State University (Invited paper). Andelt, W. F. 2003. Evaluation of aerial surveys for estimating acreage ofGunnison's and white-tailed prairie dogs in Colorado and Utah. Colorado Prairie Dog Technical Conference, Fort Collins, Colorado. Andelt, W. F. 2003. Incorporating experimental design in education on managing humanwildlife conflicts at Colorado State University. Tenth Wildlife Damage Management Conference, Hots Springs, Arkansas (Invited paper). Andelt, W. F. 2003. Managing conflicts with coyotes: aversive stimuli, novel stimuli, and livestock guarding dogs. Wyoming Student Chapter of The Wildlife Society, Laramie, Wyoming (Invited paper). �180 Andelt, W. F. 2003. Non-lethal methods for managing conflicts with prairie dogs. Colorado Prairie Dog Technical Conference, Fort Collins, Colorado (Invited paper). Jozwiak, E. A., T. N. Bailey, and W. F. Andelt. 2003. Response of wolves to changing harvest levels on the Kenai NWR, Alaska. The World Wolf Congress 2003 - Bridging Science and Community, The Banff Centre, Banff, Canada (submitted). Analyzed about 200 predator scats to help assess the role of various predators in the decline of sage grouse in northwe~tem Colorado. • Obtained $1,200 from the Renewable Resources Extension Act to revise Cooperative Extension fact sheets on managing conflicts with wildlife. Submitted a research proposal to study Ecology of coyotes and coyote predation on bighorn sheep in Rocky Mountain National Park, Colorado. Project was not funded. Co-coordinator and instructor at 3 2-4 hour workshops for 153 extension volunteers and 3 Colorado Division of Wildlife employees. Speaker at 3 Cooperative Extension meetings with 80 participants. Provided training for 55 biologists and other professionals including wildlife commissioners at l workshop. Presented 3 guest lectures to I 07 students in Colorado State University courses on managing conflicts with wildlife. Advised an M.S. candidate that conducted research on resolving conflicts with prairie dogs, and a Ph.D. candidate that was conducted research on coyotes. Served on 2 M.S. and I PhD. Committees. Evaluated 27 posters for the Cooperative Extension Poster Session at the February 2003 InService training. Served on the Jefferson County Cooperative Extension Natural Resources Agent Search Committee. Served as a Mentor for Thomas Mason, Jefferson County Cooperative Extension Natural Resources County Agent. Served on the Colorado Department of Agriculture Pesticide Review Committee. Commented on impacts of pesticides on wildlife .. Provided extensive reviews of the efficacy data for the Rodex 4000 (an explosive device for killing rodents), and efficacy of 2 repellents (Deer Stopper, Deer Stopper Ready to Use) for deterring deer. Served on the Colorado State University Cooperative Extension, College of Natural Resources, Renewable Resources Extension Act Committee. Served on the Rodent Program Review Panel for the National Wildlife Research Center. �181 Updated my web site on Managing Conflicts with Wildlife at (http://www.coopext.colostate.edu/wildlife/}. Various pages of the web site have been accessed 227 to 3,381 times each during January-June 2003. Provided interviews for 5 newspaper reporters at United Press International, Rocky Mountain News, Denver Post, and others. Provided interviews for 2 radio stations. Wrote 1 news release for CSU Cooperative Extension Agents. Reviewed 2 manuscripts for scientific journals and 1 manuscript for a colleague. Participated in about 75· meetings. Wrote about 50 e-mail messages about conflicts with wildlife. Answered about 50 telephone inquiries about managing conflicts with wildlife. �182 POSSIBLE ROLE OF PREDATORS IN THE SAGE GROUSE DECLINE W. F. Andelt Approximately $5,200 was received from the Moffat County Department of Natural Resources to conduct preliminary research on the possible role of predators in the sage grouse (Centrocercus urophasianus) decline in Moffat County. Red fox (Vulpes vulpes; Flinders [1999]) have been reported as one of the primary mammalian predators of sage grouse, whereas coyotes (Canis la trans; Presnall and Wood [1953]), bobcats (Fe/is rufus; Hartzler [1974], mink (Mustella vison; Hartzler [1974]), badgers (Taxidea taxus; Gill [1965]), and ground squirrels (Spermophilus spp.; Schroeder and Baydack [2001]) also prey on adults or nests of sage grouse. Thus, we obtained data on relative abundance of mammalian carnivores on 2 study areas (immediately northwest of Craig ["Craig"] and north of Maybell ("Maybell"). Sage grouse are scarce on the Craig study area which is fragmented habitat (sagebrush-grassland interspersed with CRP, alfalfa, and wheat), whereas they are moderately abundant on the Maybell study area which is primarily contiguous habitat (mostly sagebrush-grassland). Golden eagles (Aquila chrysaetos; Hartzler [1974]) appear to be the primary avian predator of sage grouse, particularly on leks, whereas prairie falcons (Falco mexicanus; Hartzler [1974]), red-tailed hawks (Buteo jamaicensis), Swainson's hawks (B. swainsoni), ferruginous hawks (B. regalis), northern harriers (Circus cyaneus; references in Schroeder and Baydack 2001) may occasionally kill some sage grouse. Common ravens (Corvus corax; Allred [1942], Autenrieth [1981], Alstatt [1988]) appear to be the primary avian predators of sage grouse nests or simulated nests, whereas black-billed magpies (Pica pica; Autenrieth [1981]) may prey on some nests. Consequently, we collected data on relative abundance of avian predators, and collected carnivore scats on the Craig study area, the Maybell study area, and in the Axial basin (Appendix 1), which consists primarily of contiguous habitat (mostly sagebrushgrassland) where sage grouse are moderately abundant. I also provided information to Dr. Tony Apa and colleagues on identifying which predators killed sage grouse or depredated their nests. Relative Abundance of Mammalian Predators During 5 to IO June 2001, we (Dr. Andelt, 1 graduate student, and 1 technician) set 92 scent stations on the Craig study area and 92 scent stations on the Maybell study area to gain an assessment of general abundance of carnivores on the 2 sites. Scent stations are 1-yard diameter circles of sifted earth with an attractant (fatty acid scent, small traffic cone or both) placed in the center. The scent stations were set in groups of 4 with each station 0.2 miles apart. Each group of 4 scent stations were set at least 2 miles apart to minimize visits to different groups of stations by individual carnivores. The locations for stations were mostly randomly selected from BLM maps. The stations were checked 1 day after they were set. A few of the stations were rendered inoperable by light to moderate rain. I used chi-square tests (PROC FREQ, SAS Inst. Inc. 1988) to analyze the data. Red fox visited more scent stations (X/ = 5 .465; P = 0.0194) and more groups of stations (X/ = 5. 199; P = 0.0226) on the area with few sage grouse compared to the area where grouse were fairly abundant (Table 1). We need to interpret these data with caution. First, we do not know exactly how scent station visitation rates relate to relative abundance of red fox on the 2 sites, but these data do suggest that red fox likely are more abundant on the site where grouse are rare. These data also do not indicate that red fox caused the decline of sage grouse. Surprisingly, we did not positively identify coyote tracks at any of the stations although it is possible that l or 2 of the stations could have been visited by coyotes. �183 Table 1. Visits by red fox to scent stations set in Moffat County, Colorado during 5 to 10 June 2001. Few grouse Grouse moderately abundant (Craig study area) (Maybell study area) 92 92 Number scent stations 78 87 Operable stations 19 21 Operable groups of stations 12 4 Stations visited by red fox 9 3 Groups of stations visited by red fox Stations visited by coyotes 0 0 Raptor Surveys We established 10 I -mile long survey routes on roads on the Craig study area, 10 I -mile long survey routes on the Maybell study area, and 10 I-mile long survey routes in the Axial Basin. We counted raptors (hawks, eagles, magpies), at all distances, along these transects once per month from August 2001 through June 2002 to ascertain if the abundance of raptors differs among the 3 areas. I compared abundance of various raptors on transects with ANOVA (PROC GLM, SAS Inst. Inc. 1988). Abundance of none of the raptors varied among study areas (F2,216 = 0 .15-2. 82; P = 0. 865-0. 062, Table 2). In general, black-billed magpies were the most abundant raptor followed by American crows (Corvus brachyrhynchos; Table 2). Table 2. Average number of raptors observed per month• on the Craig, Maybell, and Axial Basin study areas in Moffat County, Colorado from August 2001 through June 2002. Few grouse Grouse moderately abundant (Craig study area) Maybell study area) Axial Basin Golden eagle 0.7 2.2 1.6 Common raven 0.6 0.9 0.0 Black-billed magpie 7.8 7.8 6.4 Prairie falcon 0.0 0.0 0.0 Red-tailed hawk 0.3 0.5 0.8 Ferruginous hawk 0.0 0.1 0.0 Northern harrier 0.2 0.0 0.1 Bald eagle (Haliaeetus leucocephalus) 0.0 0.2 0.1 American crow 3.7 6.2 9.1 American kestrel (Falco sparverius) 0.1 0.2 0.2 Turkey vulture (Cathartes aura) 0.8 0.0 0.0 Other (unidentified) 0.0 0.2 0.0 aren I-mile long transects were driven once per month and all raptors observed from the vehicle were recorded from August 2001 through June 2002, except observations were not made during January and observations also were not made on the Maybell study area during February due to difficulty traversing roads. �184 Carnivore Food Habits We established 10 I -mile long survey routes on roads on the Craig study area, 10 I -mile long survey routes on roads on the Maybell study area, and 10 I -mile long survey routes on roads in Axial Basin (Appendix 1). We collected carnivore (primarily coyote and red fox) scats along these survey routes once per month from August 2001 through June 2002, except for January when travel was hindered by snow. We measured the diameters of scats with calipers and weighed them on an electronic balance. Green and Flinders (1981) reported that only 5% ofred fox scats are 2:18 mm in maximum diameter and that only 4% of coyote scats were <16 mm in maximum diameter. I extrapolated data from Weaver and Fritts (1979) and Danner and Dodd (1982) which indicated that only about 8 and 11 % of coyote scats are <16 mm in maximum diameter. Thus, I classified scats <16 mm in diameter as red fox and those 2:18 mm in diameter as coyote. Scats that consisted of short segments were classified as bobcat (Murie 1954). We placed these scats in fine-mesh nylon bags and washed and dried them. We visually inspected the scats to determine if they contained sage grouse feathers or egg shells to help ascertain if red fox or coyotes preyed on sage grouse. When feathers were found, we ascertained if they were from sage grouse according to overall size of the feathers, presence and size of quills, presence of aftershafts, and general structure of the feather. Only birds in the order Galliformes, which includes sage grouse, have aftershafts (Elbroch and Marks 2001:235) on their feathers. A total of 224 scats were collected and analyzed (Table 3). Based upon diameter and segmentation of scats, we ascertained that 26 scats were from red fox, 141 from coyotes, 4 from bobcats, and 53 scats could not be assigned to species. Although we collected scats on 10 miles of roads in each study area, the greatest numbers of scats were found on the Maybell and Axial Basin study areas, whereas the fewest scats were found on the Craig study area. Roads on the Craig study area are traveled more frequently by automobiles and are graded more frequently than roads on the other 2 study areas. These activities obliterate scats, thus relative abundance of scats likely is a poor indicator of relative abundance of carnivores on the 3 study areas. We found feathers in only 5 of 224 scats and none of the feathers appeared to be from sage grouse (Table 3). Table 3. Number carnivore scats and presence of feathers in scats found on transects on the Craig, Maybell, and Axial Basin study areas in Moffat County, Colorado from August 2001 through June 2002. Few grouse Grouse moderately abundant (Craig study area) Maybell study area) Axial Basin Total scats 18 101 105 Red fox scats 4 13 9 Red fox scats with feathers 1 1 0 Coyote scats 9 66 66 Coyote scats with feathers 0 1 1 1 3 Bobcat scats 0 Bobcat scats with feathers 0 0 0 Unknown scatsa 5 21 27 Unknown scats with feathers 0 1 0 •Based upon diameter and weight, we could not assign these scats to red fox, coyote, or bobcat. �185 Assistance with Determining which Predators are Responsible for Depredating Sage Grouse and their Nests: I provided Tony Apa and colleagues with information on how to determine which predators killed grouse or depredated their nests. SYNTHESIS OF RESULTS AND DISCUSSION Results of our scent station surveys suggest that red fox are more abundant on the Craig study area, where few sage grouse were present, than on the Maybell study area, where grouse were moderately abundant. The absence of sage grouse feathers in 141 scats, ascertained to be from coyotes, suggests that coyotes perhaps may not be substantial predators of sage grouse. We also did not find grouse feathers in 26 scats ascertained to be from red fox, and 4 scats ascertained to be from bobcats, however these small sample sizes do not allow for strong inferences regarding predation by red fox and bobcats on sage grouse. Even if feathers would have been found in coyote, red fox, or bobcat scats, it would still be difficult to ascertain the impact of either species on sage grouse without knowing densities of these 3 carnivores, densities of sage grouse, carnivore digestion and defecation rates, etc. However, I analyzed carnivore scats in a preliminary attempt to ascertain if either species might be frequently preying on sage grouse. Prior research has indicated that golden eagles and common ravens are the primary avian predators of sage grouse and their nests, respectively. Our raptor surveys indicated that both species were fairly common on most of our study areas. Initially, we expected that we might find more golden eagles and common ravens on the Craig study area, where sage grouse are scarce, if they are having an impact on sage grouse. However, predators are opportunists which often frequent areas of highest prey abundance. Due to these factors, and due to no significant differences in abundance of golden eagles and common raven among the 3 study areas, it is difficult to draw solid inferences from this study about the impact of these species on sage grouse. Ultimately, the best way to ascertain impacts of various predators on adult sage grouse, sage grouse chicks, and sage grouse nests is to monitor survival and causes of mortality for these life stages of sage grouse. ACKNOWLEDGMENTS I thank numerous individuals that assisted with this project. J. Comstock provided continuous encouragement and financial support for the study. G. Miller and the Colorado Division of Wildlife provided financial support while W. Andelt conducted the study. J. Shivik and the National Wildlife Research Center provide salary support for D. Heffernan and D. Martin while they assisted with the study. T. Apa and R. Hoffman provided suggestions for the study. D. Heffernan and D. Martin assisted with scent station surveys. A. Martin and V. Dobrich conducted surveys of raptors and collected carnivore scats. C. Simpson analyzed carnivore scats to determine presence of feathers and egg shells. R. Ryder and R. Hoffman assisted with ascertaining if feathers in carnivore scats were from sage grouse. �186 LITERATURE CITED Allred, W. J. 1942. Predation and the sage grouse. Wyoming Wild Life 71(1):3-4. Alstatt, A. 1988. Sage grouse production and mortality studies. Job performance report, Nevada Department of Wildlife, Reno, Nevada, USA Autenrieth, R. E. 1981. Sage-grouse management in Idaho. Idaho Department of Fish and Game, Wildlife Bulletin 9, Boise, Idaho, USA. Danner, D. A., and N. Dodd. 1982. Comparison of coyote and gray fox scat diameters. Journal of Wildlife Management 46:240-241. Elbroch, M., and E. Marks. 2001. Bird tracks & sign: a guide to North American species. Stackpole Books, Mechanicsburg, Pennsylvania, USA Flinders, J. T. 1999. Restoration of sage-grouse in Strawberry Valley, Utah, 1998-99. Utah Reclamation, Mitigation and Conservation Commission, Progress Report, Brigham Young University, Provo, Utah, USA Gill, R. B. 1965. Distribution and abundance of a population of sage grouse in North Park, Colorado. Thesis, Colorado State University, Fort Collins, Colorado, USA Green, J. S., and J. T. Flinders. 1981. Diameter and pH comparisons of coyote and red fox scats. Journal of Wildlife Management 45:765-767. Hartzler, J.E. 1974. Predation and the daily timing of sage grouse leks. The Auk 91:532-536. Murie, 0. J. 1954. A field guide to animal tracks. The Peterson Field Guide Series. Houghton Mifflin Company, Boston, Massachusetts, USA Presnall, C. C., and A. Wood. 1953. Coyote predation on sage grouse. Journal ofMammalogy 34:127. SAS Institute Inc. 1988. SAS/STAT User's guide, release 6.03 edition, SAS Institute Inc., Cary, North Carolina, USA Schroeder, M. A., and R. K. Baydack. 2001. Predation and the management of prairie grouse. Wildlife Society Bulletin 29:24-32. Weaver, J. L., and S. W. Fritts. 1979. Comparison of coyote and wolf scat diameters. Journal of Wildlife Management 43:786-788. �187 Appendix l. GPS coordinates for transects where carnivore scats were collected and raptors were observed (datum= WGS 84). ----------------------Scat---------------------Start of transect End of transect -------------------Raptor------------------End of transect Start of transect y y X X CRAIG STUDY AREA- FRAGMENTED HABITAT X y X y 4495138 4504934 4498993 4497952 4499870 4503788 4505463 4496358 4493667 4490952 286359 286753 289265 282630 279994 276290 271595 278798 281903 274742 4496685 4503651 4499443 4499345 4500954 4504491 4505789 4496016 4492847 4491859 286417 286993 287842 281182 278429 274892 270687 279930 281766 273755 4498162 4504934 4499021 4499731 4500997 4505050 4504525 4495509 4491265 4492689 Transect# 1 2 3 4 5 6 7 8 9 10 286321 286993 290602 283274 280846 277689 272894 277304 281065 275680 286359 287551 289265 282630 279994 276290 271595 278798 281903 274742 4496685 4506266 4499443 4499345 4500954 4504491 4505789 4496016 4492847 4491859 MAYBELL AREA- UNFRAGMENTED HABITAT 11 12 13 14 15 16 17 18 19 20 747333 747215 749179 745930 742374 739699 745369 743327 749357 248318 4497815 4502458 4508451 4508307 4510170 4510025 4514231 4521876 4520229 4517684 748923 745844 750735 744658 742575 738301 746090 744832 750726 249268 4497867 4501925 4508465 4507713 4511677 4510610 4515477 4522160 4519826 4516416 748923 745844 750735 744658 742575 738301 746090 744832 750726 249268 4497867 4501925 4508465 4507713 4511677 4510610 4515477 4522160 4519826 4516416 750446 744495 752004 744178 741688 737962 ? 746305 752031 249486 4498098 4501667 4509245 4508932 4511365 4511658 ? 4521754 4519026 4514902 4480100 4474342 4476843 4472106 4470006 4472271 4467748 4465355 4470302 4478294 253671 254317 257398 255153 249317 246720 252917 257542 260734 249580 4478589 4474661 4475313 4471305 4471575 4473285 4466698 4464349 4470308 4477016 253671 254317 257398 255153 249317 246720 252917 259268 260734 249580 4478589 4474661 4475313 4471305 4471575 4473285 4466698 4466465 4470308 4477016 253641 255142 257138 254046 249559 247862 251914 258706 260138 248906 4477122 4475609 4473907 4470304 4473119 4474390 4465499 4465355 4468956 4475569 AXIAL BASIN 21 22 23 24 25 26 27 28 29 30 253432 252975 257571 256457 249225 245513 254021 258706 262032 250505 � Dublin Core The Dublin Core metadata element set is common to all Omeka records, including items, files, and collections. For more information see, http://dublincore.org/documents/dces/. Title A name given to the resource Mammals research pdfs Description An account of the resource Pdfs of mammals research projects by title . Pdfs are linked from this page. Includes federal aid progress reports and non federal aid progress reports. If reports cover more than one year they are combined into one pdf. Dublin Core The Dublin Core metadata element set is common to all Omeka records, including items, files, and collections. For more information see, http://dublincore.org/documents/dces/. Title A name given to the resource Mammals Research Projects pdfs (2001-2010) Rights Information about rights held in and over the resource <a href="http://rightsstatements.org/vocab/NoC-NC/1.0/">No Copyright - Non-Commercial Use Only</a> https://cpw.cvlcollections.org/files/original/402b1e7ff9c5fa1a5309cb2080e8660b.pdf d5c955a9e8f9867f14717d551f2af146 PDF Text Text Colorado Division of Parks and Wildlife July 2011–June 2012 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 0670 N/A Federal Aid Project No. N/A : : : : Division of Parks and Wildlife Mammals Research Lynx Conservation Density, demography, and seasonal movements of snowshoe hares in central Colorado. Period Covered: July 1, 2011 – June 30, 2012 Author: J. S. Ivan Personnel: G. White, T. Shenk All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT To improve understanding of snowshoe hare ecology in the southern portion of their range, and enhance the ability of agency personnel to manage subalpine landscapes for snowshoe hares (Lepus americanus) and lynx (Lynx canadensis) in Colorado, we estimated snowshoe hare density, survival, recruitment, and movement in west-central Colorado, USA from July 2006−March 2009. We sampled 3 types of forest stands that purportedly provide good habitat for hares: 1) mature Engelmann spruce (Picea engelmannii)/subalpine fir (Abies lasiocarpa), 2) early seral, even-aged lodgepole pine (Pinus contorta), and 3) mid-seral, even-aged lodgepole pine that had been pre-commercially thinned. In all stand types and all seasons, snowshoe hare densities were <1.0 hares/ha. During summer, hare densities [±SE] were highest in early seral lodgepole pine (0.20 [0.01] to 0.66 [0.07] hares/ha), lowest in mid-seral lodgepole pine (0.01 [0.04] to 0.03 [0.03] hares/ha), and intermediate in mature spruce-fir (0.01 [0.002] to 0.26 [0.08] hares/ha). During winter, densities were similar among the 3 stand types. Annual survival of hares was highest in mature spruce-fir (0.14 [0.05] to 0.20 [0.07]) and similar between the 2 lodgepole stand types (0.10 [0.03] to 0.16 [0.06]). Stand attributes indicative of dense cover were positively correlated with density estimates and explain relatively more process variance in hare densities than other attributes. These same attributes were not positively correlated with hare survival. Both density and survival of hares in early seral lodgepole stands were positively correlated with the occurrence of similar stands in the surrounding landscape. Recruitment of juvenile hares occurred during all 3 summers in early seral lodgepole stands, 2 of 3 summers in mature spruce-fir stands, and in only 1 of 3 summers in mid-seral lodgepole. Within-season movements of hares were larger during winter than during summer and tended to be larger in early seral lodgepole stands. Hares in both early and mid-seral lodgepole stands tended to make larger movements between seasons than hares in spruce-fir stands, possibly reflecting the variable value of these stands as mediated by snow depth. Based on stand-specific estimates of density, demography, and movement, we conclude that thinned, mid-seral lodgepole stands are less important than mature spruce-fir and small lodgepole stand types. Management for snowshoe hares (and lynx) in central 40 �Colorado should focus on maintaining the latter. Given the more persistent nature of spruce-fir compared to early seral lodgepole, and the fact that such stands cover considerably more area, mature spruce-fir may be the most valuable stand type for snowshoe hares in the region. We used simulation to compare relative performance of the method we developed to estimate density for this project (TELEM) to other contemporary methods that are widely used (i.e., spatially explicit capture-recapture (SECR), and mean maximum distance moved (MMDM)). We evaluated performance (percent error) under all combinations of 3 levels of detection probability (0.2, 0.4, 0.6), 3 levels of occasions (5, 7, 10), and 3 levels of abundance (10, 20, 40 animals). We also tested each estimator using 5 different models for animal home ranges. TELEM performed best across most combinations of capture probabilities, sampling occasions, true densities, and home range configurations, and performance was unaffected by home range shape. SECR outperformed MMDM estimators in nearly all comparisons and may be preferable to TELEM at low capture probabilities, but performance varied with home range configuration. MMDM estimators exhibited substantial positive bias for most simulations, but performance improved for elongated or infinite home ranges. 41 �WILDLIFE RESEARCH REPORT DENSITY, DEMOGRAPY, AND SEASONAL MOVEMENTS OF SNOWSHOE HARES IN CENTRAL COLORADO JACOB S. IVAN P. N. OBJECTIVE Assess the relative value of 3 stand types (mature spruce/fir, early seral lodgepole pine, and thinned, midseral lodgepole pine) that purportedly provide high quality hare habitat by estimating density, survival, recruitment, and movements of hares in such stands during summer and winter. SEGMENT OBJECTIVES 1. Publish manuscripts in peer-reviewed scientific journals. INTRODUCTION Snowshoe hares (Lepus americanus), their famous 10-year population cycle, and close association with Canada lynx (Lynx canadensis) have been well-studied in boreal Canada for decades. Snowshoe hare range, however, extends south into the Sierra Nevada, Southern Rockies, upper Lake States, and Appalachian Mountains. Ecology of snowshoe hares in these more southerly regions is not as well understood, though hare research in the U.S. Rocky Mountains has accelerated over the past decade. Through this recent work, biologists have identified stands of young, densely-stocked conifers and those of mature, uneven-aged conifers as primary hare habitat in the region. Both stand types are characterized by dense understory vegetation that provides both browse and protection from elements and predators. From 1999 to 2006, Canada lynx were reintroduced into Colorado in an effort to restore a viable population to the southern portion of their former range. Snow tracking of released individuals and their progeny indicated that the majority of lynx winter diet in Colorado was comprised of snowshoe hares. Thus, long-term success of the lynx reintroduction effort hinges, at least partly, on maintaining adequate and widespread populations of snowshoe hares in the state. To improve our understanding of snowshoe hare ecology in the southern portion of their range, and enhance the ability of agency personnel to manage subalpine landscapes for snowshoe hares and lynx in Colorado, we conducted an observational study to evaluate purported primary hare habitat in the state. Specifically, we estimated snowshoe hare density, survival, recruitment, and movement indices in mature, uneven-aged spruce/fir (Picea engelmannii/Abies lasiocarpa) and 2 classes of young, even-aged lodgepole pine: 1) “small” lodgepole pine (Pinus contorta) stands, which were clear cut 20−25 years prior to this study and had regenerated into densely stocked stands with trees 2.54−12.69 cm in diameter, and 2) “medium” lodgepole pine stands (tree diameter = 12.70−22.85 cm) which were clear cut 40-60 years prior to this study and pre-commercially thinned ~20 years prior. Animal density is one of the most common and fundamental parameters in wildlife ecology and was the first metric we used to evaluate the stand types. However, density can be difficult to estimate from mark-recapture data because animals move on and off of a trapping grid during a sampling session (i.e., lack of geographic closure). Thus, we first developed a density estimator that uses ancillary radio telemetry locations, in addition to mark-recapture information, to account for lack of geographic closure resulting in relatively unbiased estimates of density. We also completed a series of simulations to test the performance of this “telemetry” estimator over a range of sampling parameters (i.e., capture probabilities, sampling occasions, densities, and home range configurations) likely to be encountered in the field, and 42 �compared its performance to two other commonly used, contemporary estimators: spatially explicit capture-recapture (SECR), and mean maximum distance moved (MMDM). STUDY AREA The study area encompassed roughly 1200 km2 around Taylor Park and Pitkin, Colorado, USA (39°50'N, 106°34'W; Figure 1), and included a portion of the “Core Reintroduction Area” occupied by reintroduced Canada lynx (Shenk 2009). Open sagebrush (Artemisia tridentata) parks dissected by narrow riparian zones of willow (Salix spp.) and potentilla (Potentilla spp.) dominated the relatively low elevation (~2800−3000 m) parts of the study area. Extensive stands of lodgepole pine occupied low and mid-elevation slopes (~3000−3300 m), giving way to narrow bands of Engelmann spruce/subalpine fir in the sub-alpine zone (~3200−3600 m). Alpine tundra topped the highest parts of the study area (~3300−4200 m). Moist spruce-fir forests also occurred on north-facing slopes at mid-elevations. Climate was typical of continental, high-elevation zones with relatively short, mild summers and long, harsh winters. Mean July temperature was 14 °C; mean January temperature was −11 °C (Ivan 2011). Maximum snow depth on the study area averaged 80 cm but ranged from 22−163 cm depending on year, elevation, and aspect (Ivan 2011). Snowpack generally persisted from November through May (low elevations) or June (high elevations and north-facing slopes). Some human habitation occurred in the study area, mostly in the form of seasonal residences. Considerable recreational use occurred during summer in the form of dispersed camping and off-highway vehicle traffic. A suite of native predators were present within the study area including lynx, cougar (Puma concolor), coyote (Canis latrans), red fox (Vulpes vulpes), pine marten (Martes Americana), Great Horned Owl (Bubo virginianus) and Northern Goshawk (Accipiter gentilis). METHODS Refer to Ivan (2011) for methods associated with fieldwork conducted during 2006–2009 and subsequent statistical analyses. During fiscal year 2011–2012 we completed work on 2 manuscripts submitted as a pair to the journal Ecology. The first of these manuscripts lays out an approach to estimating animal density using auxiliary telemetry information to improve estimates. The second manuscript uses simulation to compare performance of this new estimator to other contemporary estimators. We have just completed what we believe to be final revisions to these papers. Additionally, we spent much of year combining the demography and movement chapters of the primary author’s dissertation into a single, comprehensive treatment of snowshoe hare ecology in central Colorado that includes analyses on hare density, survival, recruitment, and movement. This manuscript was recently submitted to the Journal of Wildlife Management for consideration as either a research article or monograph. RESULTS AND DISCUSSION A comprehensive treatment of the results is widely available in dissertation form (Ivan 2011), so we do not repeat that here. We are currently in the process of publishing results in the peer-reviewed literature. Below is list of manuscripts that have been submitted for publication (abstracts are provided in Appendix I): Ivan, J. S., G. C. White, and T. M. Shenk. In Review. Using auxiliary telemetry information to estimate animal density from capture-recapture data. Ecology. 43 �Ivan, J. S., G. C. White, and T. M. Shenk. In Review. Using simulation to compare methods for estimating density from capture-recapture data. Ecology. Ivan, J. S., G. C. White, and T. M. Shenk. In Review. Density, Demography, and Seasonal Movements of Snowshoe Hares in Central Colorado. Journal of Wildlife Management. SUMMARY In all stand types and all seasons, snowshoe hare densities were <1.0 hares/ha. During summer, hare densities [±SE] were highest in early seral lodgepole pine (0.20 [0.01] to 0.66 [0.07] hares/ha), lowest in mid-seral lodgepole pine (0.01 [0.04] to 0.03 [0.03] hares/ha), and intermediate in mature spruce-fir (0.01 [0.002] to 0.26 [0.08] hares/ha). During winter, densities were similar among the 3 stand types. Annual survival of hares was highest in mature spruce-fir (0.14 [0.05] to 0.20 [0.07]) and similar between the 2 lodgepole stand types (0.10 [0.03] to 0.16 [0.06]). Stand attributes indicative of dense cover were positively correlated with density estimates and explain relatively more process variance in hare densities than other attributes. These same attributes were not positively correlated with hare survival. Both density and survival of hares in early seral lodgepole stands were positively correlated with the occurrence of similar stands in the surrounding landscape. Recruitment of juvenile hares occurred during all 3 summers in early seral lodgepole stands, 2 of 3 summers in mature spruce-fir stands, and in only 1 of 3 summers in mid-seral lodgepole. Within-season movements of hares were larger during winter than during summer and tended to be larger in early seral lodgepole stands. Hares in both early and mid-seral lodgepole stands tended to make larger movements between seasons than hares in spruce-fir stands, possibly reflecting the variable value of these stands as mediated by snow depth. Based on stand-specific estimates of density, demography, and movement, we conclude that thinned, mid-seral lodgepole stands are less important than mature spruce-fir and small lodgepole stand types. Management for snowshoe hares (and lynx) in central Colorado should focus on maintaining the latter. Given the more persistent nature of spruce-fir compared to early seral lodgepole, and the fact that such stands cover considerably more area, mature spruce-fir may be the most valuable stand type for snowshoe hares in the region. The estimator we developed is based on a modified Huggins closed capture estimator. It directly accounts for lack of geographic closure (animals moving on and off of the sampling grid during the sampling period) using telemetry data, and this auxiliary information is used to compute estimates of density. Contrary to other approaches, this method is free from assumptions regarding the distribution of animals on the landscape, the stationarity of their home ranges, and biases induced by abnormal movements in response to baited detectors. The estimator is freely available in Program MARK. We found that our approach performed best across most combinations of capture probabilities, sampling occasions, true densities, and home range configurations, and performance was unaffected by home range shape. Spatially explicit capture-recapture methods outperformed “mean maximum distance moved” (MMDM) estimators in nearly all comparisons and may be preferable to our telemetry estimator at low capture probabilities, but performance varied with home range configuration. MMDM estimators exhibited substantial positive bias for most simulations, but performance improved for elongated or infinite home ranges. LITERATURE CITED Ivan, J. S. 2011. Density, demography, and seasonal movement of snowshoe hares in central Colorado. Dissertation, Colorado State University, Fort Collins, Colorado, USA. Prepared by ___________________________ Jacob S. Ivan 44 �Figure 1. Study area near Taylor Park and Pitkin, central Colorado. We estimated snowshoe density, demography, and movement in 3 late-seral Engelmann spruce/subalpine fir stands (circles), 3 mid-seral lodgepole stands (squares), and 6 early-seral lodgepole stands (triangles) from summer 2006 through winter 2009. 45 �APPENDIX I PROJECT PAPERS The following manuscript (referenced here by abstract) is currently in review at the journal Ecology. USING AUXILIARY TELEMETRY INFORMATION TO ESTIMATE ANIMAL DENSITY FROM CAPTURE-RECAPTURE DATA JACOB S. IVAN, GARY C. WHITE, AND TANYA M. SHENK ABSTRACT Estimation of animal density is fundamental to ecology, and ecologists often pursue density estimates using grids of detectors (e.g., cameras, traps, hair snags) to sample animals. However, under such a framework, reliable estimates can be difficult to obtain because animals move on and off of the study site during the sampling session (i.e., the site is not closed geographically). Generally, practioners address lack of geographic closure by a) inflating the area sampled by the detectors based on the mean distance individuals moved between trapping events, or b) invoking hierarchical models in which animal density is assumed to be a spatial point process, and detection is modeled as a declining function of distance to a detector. We provide an alternative in which lack of geographic closure is sampled directly using telemetry, and this auxiliary information is used to compute estimates of density based on a modified Huggins closed capture estimator. Contrary to other approaches, this method is free from assumptions regarding the distribution of animals on the landscape, the stationarity of their home ranges, and biases induced by abnormal movements in response to baited detectors. The estimator is freely available in Program MARK. The following manuscript (referenced here by abstract) is currently in review at the journal Ecology. USING SIMULATION TO COMPARE METHODS FOR ESTIMATING DENSITY FROM CAPTURE-RECAPTURE DATA JACOB S. IVAN, GARY C. WHITE, TANYA M. SHENK Estimation of animal density is fundamental to wildlife research and management, but estimation is often complicated by lack of geographic closure of sampling grids. Contemporary methods for estimating density using mark–recapture data include: 1) approximating the effective area sampled by an array of detectors based on the mean maximum distance moved (MMDM) by animals during the sampling session, 2) spatially explicit capture–recapture (SECR) methods that formulate the problem hierarchically with a process model for animal density and an observation model in which detection probability declines with distance from a detector, and 3) a telemetry estimator (TELEM) that uses auxiliary telemetry information to estimate the proportion of animals on the study site. We used simulation to compare relative performance (percent error) of these methods under all combinations of 3 levels of detection probability (0.2, 0.4, 0.6), 3 levels of occasions (5, 7, 10), and 3 levels of abundance (10, 20, 40 animals). We also tested each estimator using 5 different models for animal home ranges. TELEM performed best across most combinations of capture probabilities, sampling occasions, true densities, and home range configurations, and performance was unaffected by home range shape. SECR outperformed MMDM estimators in nearly all comparisons and may be preferable to TELEM at low capture probabilities, but performance varied with home range configuration. MMDM estimators 46 �exhibited substantial positive bias for most simulations, but performance improved for elongated or infinite home ranges. The following manuscript (referenced here by abstract) is currently in review at the Journal of Wildlife Management. Density, Demography, and Seasonal Movements of Snowshoe Hares in Central Colorado JACOB S. IVAN, GARY C. WHITE, TANYA M. SHENK ABSTRACT To improve understanding of snowshoe hare ecology in the southern portion of their range, and enhance the ability of agency personnel to manage subalpine landscapes for snowshoe hares (Lepus americanus) and lynx (Lynx canadensis) in Colorado, we estimated snowshoe hare density, survival, recruitment, and movement in west-central Colorado, USA from July 2006−March 2009. We sampled 3 types of forest stands that purportedly provide good habitat for hares: 1) mature Engelmann spruce (Picea engelmannii)/subalpine fir (Abies lasiocarpa), 2) early seral, even-aged lodgepole pine (Pinus contorta), and 3) mid-seral, even-aged lodgepole pine that had been pre-commercially thinned. In all stand types and all seasons, snowshoe hare densities were <1.0 hares/ha. During summer, hare densities [±SE] were highest in early seral lodgepole pine (0.20 [0.01] to 0.66 [0.07] hares/ha), lowest in mid-seral lodgepole pine (0.01 [0.04] to 0.03 [0.03] hares/ha), and intermediate in mature spruce-fir (0.01 [0.002] to 0.26 [0.08] hares/ha). During winter, densities were similar among the 3 stand types. Annual survival of hares was highest in mature spruce-fir (0.14 [0.05] to 0.20 [0.07]) and similar between the 2 lodgepole stand types (0.10 [0.03] to 0.16 [0.06]). Stand attributes indicative of dense cover were positively correlated with density estimates and explain relatively more process variance in hare densities than other attributes. These same attributes were not positively correlated with hare survival. Both density and survival of hares in early seral lodgepole stands were positively correlated with the occurrence of similar stands in the surrounding landscape. Recruitment of juvenile hares occurred during all 3 summers in early seral lodgepole stands, 2 of 3 summers in mature spruce-fir stands, and in only 1 of 3 summers in mid-seral lodgepole. Within-season movements of hares were larger during winter than during summer and tended to be larger in early seral lodgepole stands. Hares in both early and mid-seral lodgepole stands tended to make larger movements between seasons than hares in spruce-fir stands, possibly reflecting the variable value of these stands as mediated by snow depth. Based on stand-specific estimates of density, demography, and movement, we conclude that thinned, mid-seral lodgepole stands are less important than mature spruce-fir and small lodgepole stand types. Management for snowshoe hares (and lynx) in central Colorado should focus on maintaining the latter. Given the more persistent nature of spruce-fir compared to early seral lodgepole, and the fact that such stands cover considerably more area, mature spruce-fir may be the most valuable stand type for snowshoe hares in the region. 47 � https://cpw.cvlcollections.org/files/original/fdd7b337777b6314a495e4b9bcef9686.pdf 4cce56f36148ff392817e671352029e8 PDF Text Text Colorado Division of Parks and Wildlife July 2010–June 2011 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 0670 N/A Federal Aid Project No. N/A : : : : Division of Parks and Wildlife Mammals Research Lynx Conservation Monitoring Canada Lynx in Colorado Using Occupancy Estimation: Initial Implementation in the Core Lynx Research Area Period Covered: July 1, 2010 – June 30, 2011 Author: J. S. Ivan Personnel: T. Shenk, G. Merrill, E. Newkirk All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT In an effort to restore a viable population of federally threatened Canada lynx (Lynx canadensis) to the southern portion of their former range, 218 individuals were reintroduced into Colorado from 1999−2006 (Devineau et al. 2010). In 2010, the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) determined that the reintroduction effort met all benchmarks of success, and that a viable, self-sustaining population of Canada lynx had been established. The purpose of this project was to develop a scientifically rigorous statewide plan to monitor this newly established population. Occupancy estimation, the use of presence/absence data to estimate the proportion of sample units used by a species within a study area, is appropriate for such a program. To evaluate this approach and provide initial estimates of occupancy and detection probability for planning purposes, we conducted a pilot occupancy estimation project in the core reintroduction area in the San Juan Mountains of southwestern Colorado. Lynx habitat in the study area was divided into 75−km2 sample units (8.66 km x 8.66 km cells), and we stratified the units into those accessible for snow tracking and “inaccessible” units which were sampled via remote cameras. We randomly sampled 30 units from each stratum. Sampling consisted of making multiple visits to each selected unit. We covered 2,178 km during our snow tracking effort (min= 1.4, max = 81.7 per visit) and detected lynx on 12 of the 30 sample units. Estimates of occupancy and detection probability from the top model were 0.62 and 0.37-0.43, respectively. Of the 120 cameras we deployed in late fall to survey the 30 inaccessible units, 113 were still operational when retrieved in early summer; 6 had memory cards that reached capacity in either May or June; 1 was stolen. We obtained 151,191 photos (min = 90, max = 6,948 per camera) from this effort. Work to assign species for each photo is ongoing. These pilot data will be used to conduct simulations and power analyses to determine how many sample units will be required to detect a statewide decline in Canada lynx, assuming that a decline in the actual population will be tied to a decline in the proportion of sample units used by lynx. 11 �WILDLIFE RESEARCH REPORT MONITORING CANADA LYNX IN COLORADO USING OCCUPANCY ESTIMATION: INITIAL IMPLEMENTATION IN THE CORE LYNX RESEARCH AREA JACOB S. IVAN P. N. OBJECTIVE Assess the use of occupancy estimation as a means of monitoring Canada lynx in Colorado using the Core Research Area in the San Juan Mountains as a test site. 1. Obtain initial estimates of occupancy and detection probability based on pilot work. 2. Conduct power analyses using initial estimates to determine the number of sample units, number of visits, and periodicity of sampling required to detect declines of interest in the statewide lynx population. 3. Develop a standardized, statistically rigorous monitoring protocol for estimating the distribution, stability and persistence of Canada lynx in Colorado. SEGMENT OBJECTIVES 1. 2. 3. 4. Assess and suggest modifications to survey protocols. Construct database to store and query survey information. Obtain initial estimates of occupancy and detection probability based on pilot work. Determine covariates and covariate structures that will be most useful for modeling occupancy and detection probability in the future. 5. Determine the efficacy of collecting lynx scat during occupancy surveys and whether such collections can be helpful in identification of putative lynx tracks and/or individuals. INTRODUCTION The Canada lynx (Lynx canadensis) occurs throughout the boreal forests of northern North America. While Canada and Alaska support healthy populations of the species, the lynx is currently listed as threatened under the Endangered Species Act (ESA) of 1973, as amended (16 U. S. C. 1531 et. seq.; U. S. Fish and Wildlife Service 2000) in the conterminous United States. Colorado represents the southern-most historical distribution of naturally occurring lynx, where the species occupied the higher elevation, montane forests in the state (U. S. Fish and Wildlife Service 2000). Lynx were extirpated or reduced to a few animals in Colorado, however, by the late 1970’s (U. S. Fish and Wildlife Service 2000), most likely due to multiple human-associated factors, including predator control efforts such as poisoning and trapping (Meaney 2002). Given the isolation of and distance from Colorado to the nearest northern populations of lynx, the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) considered reintroduction as the best option to reestablish the species in the state. Therefore, a reintroduction effort was begun in 1997, and 218 lynx were released into Colorado from 1999 – 2006 (Devineau et al. 2010). The goal of the Colorado lynx reintroduction program was to establish a selfsustaining, viable population of lynx. Progress toward this goal was tracked via evaluation of critical criteria related to lynx survival, fidelity, and recruitment. Recently, CPW determined that the criteria had been met and a viable Canada lynx population currently exists in Colorado (Shenk and Kahn 2010). In order to track the distribution, stability, and persistence of this new lynx population, a minimally-invasive, long-term, statewide monitoring program is required. Abundance estimation is not 12 �feasible logistically and presents statistical difficulties even when field logistics can be managed. However, occupancy estimation, which uses detection/non-detection survey data to estimate the proportion of area occupied in a study area, is appropriate and feasible. In short, such a monitoring scheme requires multiple visits to a sample of survey units, and on each visit observers record whether a lynx was detected or not. Such information can be used to compute the probability of detecting a lynx given that it is present on a unit, which can in turn be used to estimate the proportion (ψ) of all survey units that are occupied. This metric can be tracked through time and is assumed to be closely tied to the size and extent of the lynx population. That is, if the proportion of survey units occupied by lynx declines through time, we assume this is due to a decline in the lynx population itself. Additionally, occupancy surveys can provide information relative to the distribution of lynx in the state. CPW initiated work to evaluate detection methods for occupancy estimation in 2009-2010 (Shenk 2009). Three methods of detecting lynx were tested in sample units where lynx were known to occur: snow tracking surveys, remote camera surveillance, and hair snags. The best method for detecting lynx was snow-tracking (daily detection probability = 0.70). Camera surveillance was far less efficient (daily detection probability = 0.085), and hair snares were ineffective (daily detection probability = 0.0; Ivan and Shenk 2010). Snow tracking, however, requires safe and extensive access to a survey unit via truck and/or snowmobile. Therefore, it cannot be used in roadless or wilderness areas, which may provide important lynx habitat. Here we build on this work to test occupancy estimation on a large scale using snow tracking where accessibility permitted it, and remote cameras in areas that were not accessible. METHODS Study Area The study area consisted of the 20,684 km2 “Lynx Core Research Area” in southwest Colorado. The Core Research Area is defined as areas >2591 m (>8500 ft) in elevation within the area bounded by New Mexico to the south, Taylor Mesa to the west, and Monarch Pass on the north and east (Figure 1). Topography in this area is characterized by wide plateaus, river valleys, and rugged mountains that reach elevations over 4200 m. Engelmann spruce (Picea engelmanii) − subalpine fir (Abies lasiocarpa) is the most widely distributed coniferous forest type at elevations most typically used by lynx (2591-3353 m, 8500-11,000 ft). Sampling The study area was divided into 75 km2 (8.66 km × 8.66 km) sample units, which reflects the mean annual home range size of reproducing lynx in Colorado (Shenk 2007) and Montana (Squires and Laurion 1999). Sample units that did not meet the following criteria were discarded as they did not represent potential lynx habitat that could be surveyed. ≥ 50 % of the cell contained conifer or montane/alpine habitat, as identified by the SWReGAP LandCover Dataset ( http://earth.gis.usu.edu/swgap/swregap_landcover_report.pdf) and 2. ≥ 50 % of the cell was located on public land (tribal, NGO and city and county lands are considered private) as determined by COMaP (Theobald, D.M., G. Wilcox, S.E. Linn, N. Peterson, and M. Lineal. 2008. Colorado Ownership, Management, and Protection v7 database. Human Dimensions of Natural Resources and Natural Resource Ecology Lab, Colorado State University, Fort Collins, CO, www.nrel.colostate.edu/projects/comap). 1. Each of the remaining sample units was assigned a random number resulting from a spatially balanced sampling scheme (RRQRR; Theobald et al. 2007) and units were stratified by accessibility for snow tracking or camera surveys. The cells with the lowest 30 random numbers for each stratum were 13 �selected for sampling during the pilot work. A few cells in both strata were discarded once field work began due to access issues and these were replaced with cells 31, 32, etc. Snow tracking Surveys Teams of 2 observers generally searched for lynx tracks within a sample unit using snowmobiles, although portions of some units were surveyed via truck or snowshoe. An effort was made to survey all portions of each unit as access allowed. Each of the 30 units selected for sampling was visited 3 times − roughly once per month from January through March. Occasionally a “visit” actually took place over consecutive days as some units could not be covered completely from a single access point. Once tracks were detected in a unit, that visit was considered complete and no further surveying occurred until the next visit. However, observers forward and back-tracked to find a scat sample. For each visit, observers recorded number of kilometers surveyed, tracking conditions (poor, fair, good, excellent), other species detected, location of lynx tracks, and time/distance to scat discovery. Camera Surveys Four remote camera sets (RECONYX RapidFireTM Professional PC85) were placed within each selected “inaccessible” sample unit during September and October. Placement of camera sets was not random within the unit; they were placed strategically on the landscape to maximize coverage of the sample unit and exploit microsites most likely to be used by lynx. Camera sets consisted of 1) a remote camera mounted to a tree using a Master Lock TM PythonTM cable lock, 2) a target tree at which the camera was pointed, generally about 5-10m away, 3) a compact disc strung from a nearby branch to visually attract lynx from a distance, 4) 2 feathers strung up in such a manner as to entice lynx to walk between the camera and the target tree, and 5) wool soaked in commercial scent lure that was packed into the bark of the target tree to hold lynx in front of the camera (Figure 2). Cameras were placed higher than usual, about head-height, and pointed slightly downward at the target tree so photos could be obtained during both snow-free periods and during periods of accumulating snow. Cameras were collected during June and July at which time the number of photos, percent of memory card used, percent battery life remaining, and condition of visual/scent lures was recorded. Analysis Assumptions inherent in occupancy estimation are 1) surveyed sites are either occupied or not occupied by the species of interest throughout the duration of the study; no sites change status during the survey period (i.e., the system is closed), 2) the probability of occupancy is constant across sites or can be modeled using covariates, 3) the probability of detection is constant across sites or can be modeled using site-specific covariates, and 4) species detection at a site is assumed to be independent of species detection at other sites (MacKenzie et al. 2006). Sampling mobile carnivores such as lynx presents a clear violation of the first assumption as individuals undoubtedly move into and out of sample units routinely. Fortunately, estimation can proceed, but the quantities estimated are different from traditional occupancy estimation. Rather than estimating the probability that a unit is occupied by lynx, we now estimate the probability that a sample unit is used by lynx. Also, the estimated detection parameter is not the probability of detection given a site is occupied, it is the product of a) the probability of detection given the species is available for detection, and b) the probability that the species was available. These subtleties aside, the procedure still gives a metric (use) that can be monitored through time to detect trends. We used the “Occupancy Estimation” data type in Program MARK to produce initial estimates of occupancy (i.e., use, ψ) and detection probability (p) for the snow tracking stratum. We hypothesized that some metric of the number of kilometers surveyed, or number that could be surveyed, would be important in explaining variation in detection probability as it should be an indicator of the amount of access to a unit. Surveys on units with more access should stand a better chance of detecting lynx if they are present. We further hypothesized that tracking conditions during a given visit should have an effect on detection 14 �probability. Finally, we did not expect differences among survey teams as both teams were experienced, but we wanted to test that assumption. Therefore, we considered 5 covariates that may explain variation in p: 1) total road length available for surveying in each sampled unit, 2) Kilometers surveyed during each visit, 3) maximum number of kilometers surveyed during any visit to a given unit, 4) tracking conditions during each visit, and 5) observer effect. We hypothesized that the proportion of spruce/fir cover in each unit may affect the probability of use, as might proportion of willow (Salix spp.), and subalpine/alpine meadow. Thus, we considered those 3 covariates as potentially important for explaining variability in ψ. As this analysis is exploratory, we held ψ constant and built an additive model for each detection covariate (one at a time) to determine the best structure for p. Similarly, we held p constant and fit additive models using the 3 covariates for ψ. We combined the best structure as determined by AICc (Burnham and Anderson 2002) for each parameter and used results from that single model to produce initial estimates of p and ψ. We also ran a model where both p and ψ were held constant as a baseline for comparison. Occupancy estimation for the camera stratum will proceed in a similar fashion as above, but data from the photos is incomplete at this time. Photos will be grouped by month (November to March) for each sample unit such that encounter histories will have 5 “visits” rather than 3. Due to this grouping, there are no meaningful covariates for p. Individual cameras recorded moon phase and temperature for each photo, but aggregated over a month, these data are not helpful. Some camera sets used different scent lures than others, but aggregating by unit negates the utility of this information as well. We will consider the same covariates on ψ as listed above. RESULTS On average, we covered 24.71 km per visit to each accessible sample unit (min = 1.40 km, max = 81.67 km) for a total of 2,184 km surveyed. We detected 20 lynx tracks in 12 of the 30 units sampled (i.e., tracks were detected on multiple visits to some units; Figure 1). We were able to collect scat from 13 of the 20 tracks, and mean forward/backtracking distance to scat discovery was 0.65 km (min = 0.05, max = 1.60). According to AICc, the best structures for p and ψ were “kilometers surveyed per visit” and “proportion spruce-fir,” respectively (Table 1). No other structure for either parameter resulted in improvement over constant p and ψ with the exception of modeling ψ as a function of “proportion willow.” In fact, this was the AICc top structure, but the parameters could not be estimated so it was dropped from the model set. Estimates (SE) from the model that combined the best structures were ψ = 0.62 (0.25), p1= 0.37 (0.10), p2= 0.37 (0.10), and p3 = 0.43 (0.10) where pi is the detection probability for visit I (i.e., p1 is the estimated detection probability for January, p2 = February, p3 = March) . As expected, the slope of the spruce-fir effect was highly positive. Probability of use was 0.5 when proportion spruce-fir approached 0.35, and probability of use went to 1.0 when proportion sprucefir approached 0.6 (Figure 3). The relationships between “proportion meadow” and ψ and “proportion willow” and ψ were also positive, but the relationships were weaker as confidence intervals for these slopes covered zero. The relationship between p and kilometers surveyed was negative. Similarly, the relationship between p and visit condition was opposite of our hypothesis (as visit conditions improved, detection probability declined). There was no relationship between “total road length” or “maximum kilometers surveyed” and detection probability. We did not detect differences between teams of observers. 15 �Genetic analysis of scat samples is ongoing. By December 2010, we should be able to assess whether scats were of high enough quality to confirm species and/or individual identification. Of the 120 cameras deployed during Fall 2010, 113 were still operational when retrieved in Summer 2011 after 234-309 days of deployment. Six had memory cards that reached capacity in either May or June, and one camera was stolen. On average, we obtained 1,260 photos per camera (min = 90, max = 6,948) for a total of 151,191 photos. At the time of retrieval, compact discs were still operational for 46% of camera sets, feathers were operational at 64% of sets, and remnants of scent lure were detected at 55% of sets. DISCUSSION Initial results indicate that occupancy (use) can be adequately modeled using data collected via snow tracking. Precision on estimates of ψ and p was relatively poor, but this can be addressed by sampling more units and/or making more visits. Modeling p as a function of the “kilometers surveyed per visit” was a better fit for the data than modeling it as a function of either “total road length within a unit” or “visit conditions.” However, we recommend continuing to record “total road length” and “visit conditions” in future surveys as it seems reasonable that these covariates should impact detection probability, and their effects may show through as sample size increases. Similarly, we recommend retaining all covariates on ψ to assess their performance with a larger dataset. The relationship between p and “kilometers surveyed per visit” was negative, which is likely an artifact of how the units were sampled – when lynx were detected, surveying stopped, so detection probability was higher for visits with few kilometers surveyed. The relationship between p and “visit condition” was opposite of our hypothesis (as visit conditions improved, detection probability declined). Our condition criteria were based largely on the freshness of the snow and degree of melting/crusting where fresh snow was assigned the best condition, and older, crusted snow was assigned the worst. Functionally, this index is an inverse of “time-since-snowfall.” Therefore, it is sensible that “poor” condition indices resulted in higher detection probabilities. While the immediate conditions were poor for tracking, significant time had passed in which lynx could move around and leave tracks to be discovered. We estimated that lynx used approximately 62% of the sample units available in the Core Research Area. However, for this pilot study, lynx habitat was coarsely defined as units with >50% spruce/fir and >50% public land. In several cases, sampled units met these criteria, but field crews that actually made visits indicated these units did not appear to include much lynx habitat. CPW is currently finishing an analysis to produce a map of predicted lynx habitat throughout the state. In the future, we expect to use this map to frame the population of units to sample for lynx monitoring. This more refined population of sample units should reduce time wasted surveying units that do not include good lynx habitat, and will result in an increased estimate of probability of use. Photos from cameras deployed to sample the inaccessible stratum have not been fully processed, therefore we cannot determine whether that portion of the study worked well enough to be included in any future monitoring effort. Roughly half of the visual attractants we used did not operate through the entirety of the study. These attractants are important for drawing lynx to the set from a distance and their failure diminishes the utility of the cameras for detecting lynx. If cameras are to be used in the future, design changes will be necessary to ensure that most of these visual attractants operate throughout the sampling season. 16 �ACKNOWLEDGMENTS We thank Britta Schielke, Cate Brown, Wendy Lanier, Joan Meiners, Shane McKenzie, Nick Burgmeier, Doug Clark, Bob Peterson, Tim Hanks, Kei Yasuda, Ashley Bies, Tyler Kelly, Alyssa Winkler, and Carolyn Shores for their efforts in the field. Dale Gomez and Rhandy Ghormley (USFS) graciously coordinated housing for seasonal crews. We thank various personnel from both the Rio Grande and San Juan National Forests for logistical help in the field. Funding was provided by a U.S. Fish and Wildlife Service Section 6 Grant. LITERATURE CITED Burnham, K. P., and D. R. Anderson. 2002. Model Selection and Multimodel Inference: A practical information-theoretic approach. Springer, New York, New York, USA. Devineau, O., T. M. Shenk, G. C. White, P. F. Doherty Jr., P. M. Lukacs, and R. H. Kahn. 2010. Evaluating the Canada lynx reintroduction programme in Colorado: patterns in mortality. Journal of Applied Ecology 47:524-531. Ivan, J. S., and T. M. Shenk. 2010. Estimating the Extent, Stability and Potential Distribution of Canada Lynx (Lynx canadensis) in Colorado: initial implementation in the core lynx research area. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, Colorado. MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy Estimation and Modeling: Inferring Patterns and Dynamics of Species Occurrence. Elsevier Academic Press, Oxford, UK. McKelvey, K. S., J. von Kienast; K.B. Aubry; G. M. Koehler; B. T. Maletzke; J. R. Squires; E. L. Lindquist; S. Loch; M. K. Schwartz. 2006. DNA analysis of hair and scat collected along snow tracks to document the presence of Canada lynx. Wildlife Society Bulletin 34: 451-455. Meaney C. 2002. A review of Canada lynx (Lynx canadensis) abundance records from Colorado in the first quarter of the 20Th century. Colorado Department of Transportation Report. Shenk, T. M. 2005. Post-release monitoring of lynx reintroduced to Colorado. Job Progress Report, Colorado Division of Wildlife, Fort Collins, Colorado. __________. 2007. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, Colorado. __________. 2009. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, Colorado. Shenk, T.M., and R. H. Kahn. 2003. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, Colorado. __________. 2010. The Colorado lynx reintroduction program. Report to the Colorado Division of Wildlife, Fort Collins, Colorado. Squires, J. R. and T. Laurion. 1999. Lynx home range and movements in Montana and Wyoming: preliminary results. Pages 337-349 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University Press of Colorado, Boulder, Colorado. Theobald, D.M., D.L. Stevens, Jr., D. White, N.S. Urquhart, A.R. Olsen, and J.B. Norman. 2007. Using GIS to generate spatially balanced random survey designs for natural resource applications. Environmental Management 40(1): 134-146. U. S. Fish and Wildlife Service. 2000. Endangered and threatened wildlife and plants: final rule to list the contiguous United States distinct population segment of the Canada lynx as a threatened species. Federal Register 65, Number 58. Prepared by ______________________________________ Jake S. Ivan, Wildlife Researcher 17 �Table 1. Model selection results for estimating occupancy of sample units by Canada Lynx (Lynx canadensis) in the Core Research Area, San Juan Mountains, Colorado, Winter 2010-2011. Model AICc ΔAICc AICc Wt Num Par p(KmSurveyPerVisit)ψ(SprFir) 81.25 0.00 0.78 4 p(.)ψ(SprFir) 84.23 2.98 0.17 3 p(KmSurveyPerVisit)ψ(.) 88.60 7.35 0.02 3 p(.)ψ(.) 89.95 8.70 0.01 2 p(TtlRoadLen)ψ(.) 90.29 9.04 0.01 3 p(.)ψ(Meadow) 91.25 9.99 0.01 3 p(Observer)ψ(.) 92.10 10.85 0.00 3 p(MaxKmSurv)ψ(.) 92.42 11.17 0.00 3 p(VisitCond)ψ(.) 97.77 16.52 0.00 5 Figure 1. Canada lynx Core Research Area in southwest Colorado. Squares are 75km2 sample units available for occupancy surveys. Blue represents the sample of 30 “accessible” units selected for snow tracking surveys. Orange are “inaccessible” units selected for surveys using remote cameras. Crosshatching indicates accessible units where lynx were detected. The data from inaccessible units has not been fully processed and units where lynx were detected are not shown. 18 �Target tree (scern t lure) Feather/wings (visua 11 u re) co (visual lure) Feather/wings (visual lure) Python cable Figure 2. General configuration of remote camera sets for detecting Canada lynx. Four such sets were deployed in each of 30 inaccessible sample units from Fall 2010 to Summer 2011. 19 �1 ➔ -; 0.8 - -- --(/) ::::I 0 -~ 0.6 .c C'O .c 0 o.. 0.4 "C (1) +-' C'O E ',i:i (/) 0.2 w 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0 .7 Proportion spruce/fir in Sample Unit Figure 3. Estimated probability of use (ψ) and 95% confidence intervals plotted against proportion spruce/fir in a sample unit. Relationship is based on snow tracking occupancy surveys completed in southwest Colorado, Winter 2010-2011. 20 0.8 �Colorado Division of Parks and Wildlife July 2011–June 2012 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 0670 N/A Federal Aid Project No. N/A : : : : Division of Parks and Wildlife Mammals Research Lynx Conservation Monitoring Canada Lynx in Colorado Using Occupancy Estimation: Initial Implementation in the Core Lynx Research Area Period Covered: July 1, 2011 – June 30, 2012 Author: J. S. Ivan Personnel: T. Shenk, G. Merrill, E. Newkirk All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT In an effort to restore a viable population of Canada lynx (Lynx canadensis) to the southern portion of their former range, 218 individuals were reintroduced into Colorado from 1999−2006 (Devineau et al. 2010). In 2010, the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) determined that the reintroduction effort met all benchmarks of success, and that a viable, selfsustaining population of Canada lynx had been established. The purpose of this project was to develop a scientifically rigorous statewide plan to monitor this newly established population. Occupancy estimation, the use of presence/absence data to estimate the proportion of sample units used by a species within a study area, is appropriate for such a program. To evaluate this approach and provide initial estimates of occupancy and detection probability for planning purposes, we conducted a pilot occupancy estimation project in the core reintroduction area in the San Juan Mountains of southwestern Colorado. Lynx habitat in the study area was divided into 75-km2 sample units (8.66 km x 8.66 km cells), and we stratified the units into those accessible for snow tracking and “inaccessible” units, which were sampled via remote cameras. We randomly sampled 30 units from each stratum. A summary of snow tracking results can be found in Ivan (2011). Of the 120 cameras we deployed in late fall to survey the 30 inaccessible units, 113 were still operational when retrieved in early summer; 6 had memory cards that reached capacity in either May or June; 1 was stolen. We obtained 151,191 photos (min = 90, max = 6,948 per camera) from this effort. We determined species for each photo and checked our work using multiple observers. Average agreement between observers was 96%. We estimated that approximately 25% of inaccessible cells were used by lynx. Detection probability was 0.43. These pilot data are currently being used to conduct simulations and power analyses to determine how many sample units will be required to detect population changes of interest in Colorado. 26 �WILDLIFE RESEARCH REPORT MONITORING CANADA LYNX IN COLORADO USING OCCUPANCY ESTIMATION: INITIAL IMPLEMENTATION IN THE CORE LYNX RESEARCH AREA JACOB S. IVAN P. N. OBJECTIVE Assess the use of occupancy estimation as a means of monitoring Canada lynx in Colorado using the Core Research Area in the San Juan Mountains as a test site. SEGMENT OBJECTIVES 1. Obtain initial estimates of occupancy and detection probability from units where remote cameras were the primary detection method. 2. Determine covariates and covariate structures that will be most useful for modeling occupancy and detection probability for camera surveys. 3. Combine these results with those obtained via previous work (snow tracking) to inform simulation work aimed at determining the number of sample units, and visits to each unit, required to detect changes of interest in the lynx population in Colorado. INTRODUCTION The Canada lynx (Lynx canadensis) occurs throughout the boreal forests of northern North America. While Canada and Alaska support healthy populations of the species, the lynx is currently listed as threatened under the Endangered Species Act (ESA) of 1973, as amended (16 U. S. C. 1531 et. seq.; U. S. Fish and Wildlife Service 2000) in the conterminous United States. Colorado represents the southern-most historical distribution of naturally occurring lynx, where the species occupied the higher elevation, montane forests in the state (U. S. Fish and Wildlife Service 2000). Lynx were extirpated or reduced to a few animals in Colorado, however, by the late 1970’s (U. S. Fish and Wildlife Service 2000), most likely due to multiple human-associated factors, including predator control efforts such as poisoning and trapping (Meaney 2002). Given the isolation of and distance from Colorado to the nearest northern populations of lynx, the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) considered reintroduction as the best option to reestablish the species in the state. Therefore, a reintroduction effort was begun in 1997, and 218 lynx were released into Colorado from 1999 – 2006 (Devineau et al. 2010). The goal of the Colorado lynx reintroduction program was to establish a selfsustaining, viable population of lynx. Progress toward this goal was tracked via evaluation of critical criteria related to lynx survival, fidelity, and recruitment. Recently, CPW determined that the criteria had been met and a viable Canada lynx population currently exists in Colorado (Shenk and Kahn 2010). In order to track the distribution, stability, and persistence of this new lynx population, a minimally-invasive, long-term, statewide monitoring program is required. Abundance estimation is not feasible logistically and presents statistical difficulties even when field logistics can be managed. However, occupancy estimation, which uses detection/non-detection survey data to estimate the proportion of area occupied in a study area, is appropriate and feasible. In short, such a monitoring scheme requires multiple visits to a sample of survey units, and on each visit observers record whether a lynx was detected or not. Such information can be used to compute the probability of detecting a lynx given that it is present on a unit, which can in turn be used to estimate the proportion (ψ) of all survey units that are occupied. This metric can be tracked through time and is assumed to be closely tied to the 27 �size and extent of the lynx population. That is, if the proportion of survey units occupied by lynx declines through time, we assume this is due to a decline in the lynx population itself. Additionally, occupancy surveys can provide information relative to the distribution of lynx in the state. CPW initiated work to evaluate detection methods for occupancy estimation in 2009-2010 (Shenk 2009). Three methods of detecting lynx were tested in sample units where lynx were known to occur: snow tracking surveys, remote camera surveillance, and hair snags. The best method for detecting lynx was snow-tracking (daily detection probability = 0.70). Camera surveillance was far less efficient (daily detection probability = 0.085), and hair snares were ineffective (daily detection probability = 0.0; Ivan and Shenk 2010). Snow tracking, however, requires safe and extensive access to a survey unit via truck and/or snowmobile. Therefore, it cannot be used in roadless or wilderness areas, which may provide important lynx habitat. Here we build on this work to test occupancy estimation on a large scale using snow tracking where accessibility permitted it, and remote cameras in areas that were not accessible. METHODS Study Area The study area consisted of the 20,684 km2 “Lynx Core Research Area” in southwest Colorado. The Core Research Area is defined as areas >2591 m (>8500 ft) in elevation within the area bounded by New Mexico to the south, Taylor Mesa to the west, and Monarch Pass on the north and east (Figure 1). Topography in this area is characterized by wide plateaus, river valleys, and rugged mountains that reach elevations over 4200 m. Engelmann spruce (Picea engelmanii) - subalpine fir (Abies lasiocarpa) is the most widely distributed coniferous forest type at elevations most typically used by lynx (2591-3353 m, 8500-11,000 ft). Sampling The study area was divided into 75 km2 (8.66 km × 8.66 km) sample units, which reflects the mean annual home range size of reproductively active female lynx in Colorado (Shenk 2007) and Montana (Squires and Laurion 1999). Sample units that did not meet the following criteria were discarded as they did not represent potential lynx habitat that could be surveyed. ≥50 % of the cell contained conifer or montane/alpine habitat, as identified by the SWReGAP LandCover Dataset ( http://earth.gis.usu.edu/swgap/swregap_landcover_report.pdf) and 2. ≥ 50 % of the cell was located on public land (tribal, NGO, city, and county lands were considered private) as determined by COMaP (Theobald, D.M., G. Wilcox, S.E. Linn, N. Peterson, and M. Lineal. 2008. Colorado Ownership, Management, and Protection v7 database. Human Dimensions of Natural Resources and Natural Resource Ecology Lab, Colorado State University, Fort Collins, CO, www.nrel.colostate.edu/projects/comap). 1. Each of the remaining sample units was assigned a random number resulting from a spatially balanced sampling scheme (RRQRR; Theobald et al. 2007) and units were stratified by accessibility for snow tracking or camera surveys. The cells with the lowest 30 random numbers for each stratum were selected for sampling during the pilot work. A few cells in both strata were discarded once field work began due to access issues and these were replaced with cells 31, 32, etc. 28 �Snow tracking Surveys A detailed discussion of the methods and results associated with snow tracking surveys appears in Ivan (2011). We do not repeat that discussion here. Instead we focus on methods and results from the remote cameras, as those data were unavailable for the 2011 report. Camera Surveys Four remote camera sets (RECONYX RapidFireTM Professional PC85) were placed within each selected “inaccessible” sample unit during September and October. Placement of camera sets was not random within the unit; they were placed strategically on the landscape to maximize coverage of the sample unit and exploit microsites most likely to be used by lynx. Camera sets consisted of 1) a remote camera mounted to a tree using a Master Lock TM PythonTM cable lock, 2) a target tree at which the camera was pointed, generally about 5–10m away, 3) a compact disc strung from a nearby branch to visually attract lynx from a distance, 4) 2 feathers strung up in such a manner as to entice lynx to walk between the camera and the target tree, and 5) wool soaked in commercial scent lure that was packed into the bark of the target tree to hold lynx in front of the camera (Figure 2). Cameras were placed higher than usual, about head-height, and pointed slightly downward at the target tree so photos could be obtained during both snow-free periods and during periods of accumulating snow. Cameras were collected during June and July at which time the number of photos, percent of memory card used, percent battery life remaining, and condition of visual/scent lures was recorded. All photo attributes were imported into a database and species was assessed for each photo based on review by at least 2 observers. Analysis Assumptions inherent in occupancy estimation are 1) surveyed sites are either occupied or not occupied by the species of interest throughout the duration of the study; no sites change status during the survey period (i.e., the system is closed), 2) the probability of occupancy is constant across sites or can be modeled using covariates, 3) the probability of detection is constant across sites or can be modeled using site-specific covariates, and 4) species detection at a site is assumed to be independent of species detection at other sites (MacKenzie et al. 2006). Sampling mobile carnivores such as lynx presents a clear violation of the first assumption as individuals undoubtedly move into and out of sample units routinely. Fortunately, estimation can proceed, but the quantities estimated are different from traditional occupancy estimation. Rather than estimating the probability that a unit is occupied by lynx, we now estimate the probability that a sample unit is used by lynx. Also, the estimated detection parameter is not the probability of detection given a site is occupied, it is the product of a) the probability of detection given the species is available for detection, and b) the probability that the species was available. These subtleties aside, the procedure still gives a metric (use) that can be monitored through time to detect trends. We used the “Occupancy Estimation” data type in Program MARK to produce initial estimates of occupancy (i.e., use, ψ) and detection probability (p) for the camera stratum. Photos were grouped by month (November to March) for each sample unit such that encounter histories included 5 “visits.” Due to this grouping, there were no meaningful covariates for p. Individual cameras recorded moon phase and temperature for each photo, but aggregated over a month, these data were not helpful. Some camera sets used different scent lures than others, but aggregating by unit negates the utility of this information as well. We hypothesized that the proportion of spruce/fir and/or willow (Salix spp.) cover in each unit may affect the probability of use and/or probability of detection. Thus, we considered these covariates as potentially important for explaining variability in ψ and p. We held ψ constant and built an additive model for each detection covariate (one at a time) to determine the best structure for p. We then held p at the best structure as determined by AICc (Burnham and Anderson 2002) and fit additive models using the 29 �covariates for ψ. We also ran a model where both p and ψ were held constant as a baseline for comparison. We report estimates of p and ψ from the AICc top model. RESULTS Of the 120 cameras deployed during Fall 2010, 113 were still operational when retrieved in Summer 2011 after 234-309 days of deployment. Six had memory cards that reached capacity in either May or June, and one camera was stolen. On average, we obtained 1,260 photos per camera (min = 90, max = 6,948) for a total of 151,191 photos. At the time of retrieval, compact discs were still operational for 46% of camera sets, feathers were operational at 64% of sets, and remnants of scent lure were detected at 55% of sets. We obtained 445 photos of lynx and detected them in 7 of the 30 units sampled (Figure 1). Average agreement between photo reviewers was 96%. Of the model structures we fit, none was clearly better than the others as AICc weight was distributed fairly evenly (Table 1). Beta estimates for fitted models suggested that ψ was positively associated with both percent spruce/fir and percent willow in a given unit. Spruce/fir was also positively associated with detection probability, whereas willow was negative associated with detection probability. However none of these models were as well supported by the data as the null model in which ψ and p were considered constant across cells. Thus, results generally followed our expectations, but the null model came out on top likely due to sparse data and small samples in this pilot study. Model-averaged estimates for ψ and p were 0.25 and 0.42, respectively. Detection probability using cameras was about the same as for snowtracking (Ivan 2011), but estimated probability of use for inaccessible sampling units was about half that estimated for accessible cells sampled via snow tracking. DISCUSSION Initial results indicate that occupancy (use) can be adequately modeled using data collected via snow tracking. Precision on estimates of ψ and p was relatively poor, but this can be addressed by sampling more units and/or making more visits. Modeling p and ψ as functions of the covariates (spruce/fir and willow) was not as well supported as specifying them to be constant across units. However, we recommend continuing to record and use these covariates and others in future surveys as it seems reasonable that these covariates should impact detection probability and/or use, and their effects may be important as sample size increases. We estimated that lynx used approximately 25% of the sample units available in the Core Research Area. However, for this pilot study, lynx habitat was coarsely defined as units with >50% conifer and/or montane cover and >50% public land. In several cases, sampled units met these criteria, but field crews that actually made visits indicated these units did not appear to include much lynx habitat. CPW recently finished an analysis to produce a map of predicted lynx habitat throughout the state. In the future, we expect to use this map to frame the population of units to sample for lynx monitoring. This more refined population of sample units should reduce time wasted surveying units that do not include good lynx habitat, and will result in an increased estimate of probability of use. Indeed, re-running the analysis using only those cells (n = 24) within the top 40% of predicted lynx habitat in the state increased the occupancy estimate to 0.31. Roughly half of the visual attractants we used did not operate through the entirety of the study. These attractants are important for drawing lynx to the set from a distance and their failure diminishes the utility of the cameras for detecting lynx. If cameras are to be used in the future, design changes will be necessary to ensure that most of these visual attractants operate throughout the sampling season. We suggest that attractants be attached via wire rather than fishing line. We also suggest that auditory 30 �attractants may be helpful. In a recent study on cougars (Puma concolor) in the Front Range of Colorado, visitation rates at camera sites increased dramatically when auditory attractants were used in addition to scent lures and visual attractants (Kirstie Yeager, personal communication). ACKNOWLEDGMENTS We thank Britta Schielke, Cate Brown, Wendy Lanier, Joan Meiners, Shane McKenzie, Nick Burgmeier, Doug Clark, Bob Peterson, Tim Hanks, Kei Yasuda, Ashley Bies, Tyler Kelly, Alyssa Winkler, and Carolyn Shores for their efforts in the field. Dale Gomez and Rhandy Ghormley (USFS) graciously coordinated housing for seasonal crews. We thank various personnel from both the Rio Grande and San Juan National Forests for logistical help in the field. Funding was provided by a U.S. Fish and Wildlife Service Section 6 Grant. LITERATURE CITED Burnham, K. P., and D. R. Anderson. 2002. Model Selection and Multimodel Inference: A practical information-theoretic approach. Springer, New York, New York, USA. Devineau, O., T. M. Shenk, G. C. White, P. F. Doherty Jr., P. M. Lukacs, and R. H. Kahn. 2010. Evaluating the Canada lynx reintroduction programme in Colorado: patterns in mortality. Journal of Applied Ecology 47:524–531. Ivan, J. S., and T. M. Shenk. 2010. Estimating the Extent, Stability and Potential Distribution of Canada Lynx (Lynx canadensis) in Colorado: initial implementation in the core lynx research area. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, Colorado. Ivan, J. S. 2011. Monitoring Canada lynx in Colorado using occupancy estimation: Initial implementation in the Core Lynx Research Area. Wildlife Research Report. Colorado Division of Parks and Wildlife, Fort Collins, CO, USA. Pages 11–20. MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy Estimation and Modeling: Inferring Patterns and Dynamics of Species Occurrence. Elsevier Academic Press, Oxford, UK. McKelvey, K. S., J. von Kienast; K.B. Aubry; G. M. Koehler; B. T. Maletzke; J. R. Squires; E. L. Lindquist; S. Loch; M. K. Schwartz. 2006. DNA analysis of hair and scat collected along snow tracks to document the presence of Canada lynx. Wildlife Society Bulletin 34: 451-455. Meaney C. 2002. A review of Canada lynx (Lynx canadensis) abundance records from Colorado in the first quarter of the 20Th century. Colorado Department of Transportation Report. Shenk, T. M. 2005. Post-release monitoring of lynx reintroduced to Colorado. Job Progress Report, Colorado Division of Wildlife, Fort Collins, Colorado. __________. 2007. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, Colorado. __________. 2009. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, Colorado. Shenk, T.M., and R. H. Kahn. 2003. Post-release monitoring of lynx reintroduced to Colorado. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, Colorado. __________. 2010. The Colorado lynx reintroduction program. Report to the Colorado Division of Wildlife, Fort Collins, Colorado. Squires, J. R. and T. Laurion. 1999. Lynx home range and movements in Montana and Wyoming: preliminary results. Pages 337–349 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S McKelvey, and J. R. Squires, editors. Ecology and Conservation of Lynx in the United States. General Technical Report for U. S. D. A. Rocky Mountain Research Station. University Press of Colorado, Boulder, Colorado. 31 �Theobald, D.M., D.L. Stevens, Jr., D. White, N.S. Urquhart, A.R. Olsen, and J.B. Norman. 2007. Using GIS to generate spatially balanced random survey designs for natural resource applications. Environmental Management 40(1): 134–146. U. S. Fish and Wildlife Service. 2000. Endangered and threatened wildlife and plants: final rule to list the contiguous United States distinct population segment of the Canada lynx as a threatened species. Federal Register 65, Number 58. Prepared by ___________________________ Jacob S. Ivan 32 �Table 1. Model selection results for estimating lynx occupancy of sample units surveyed via remote camera in the Core Research Area, San Juan Mountains, Colorado, Winter 2010–2011. Model ψ(.)p(.) ψ(.)p(willow) ψ(.)p(SprFir) ψ(SprFir)p(.) ψ(willow)p(.) AICc 84.54 85.06 85.37 85.73 85.92 ΔAICc 0.00 0.52 0.83 1.19 1.38 AICc Wt 0.29 0.22 0.19 0.16 0.14 Num Par 2 3 3 3 3 SnowTra Lynx Delee Figure 1. Canada lynx Core Research Area in southwest Colorado. Squares are 75km2 sample units available for occupancy surveys. Blue represents the sample of 30 “accessible” units selected for snow tracking surveys. Orange are “inaccessible” units selected for surveys using remote cameras. Crosshatching indicates units where lynx were detected. 33 �Target tree (scern t lure) Feather/wings (visua 11 u re) co (visual lure) Feather/wings (visual lure) Python cable Figure 2. General configuration of remote camera sets for detecting Canada lynx. Four such sets were deployed in each of 30 inaccessible sample units from Fall 2010 to Summer 2011. 34 � https://cpw.cvlcollections.org/files/original/921c5283d5df51faccf544bbaaddc42d.pdf d07d87bd0ade9817d706c87afe4ae252 PDF Text Text Colorado Parks and Wildlife July 2012–June 2013 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 0670 N/A Federal Aid Project No. N/A : : : : Division of Parks and Wildlife Mammals Research Lynx Conservation Statewide Monitoring of Canada Lynx in Colorado: Evaluation of options Period Covered: July 1, 2012 – June 30, 2013 Author: J. Ivan Personnel: M. Ellis, Alaska Department of Fish and Game, M. Schwartz, U.S. Forest Service Rocky Mountain Research Station All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT In an effort to restore a viable population of Canada lynx (Lynx canadensis) to the southern portion of their former range, 218 individuals were reintroduced into Colorado from 1999−2006. In 2010, the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) determined that the reintroduction effort met all benchmarks of success, and that the population of Canada lynx in the state was apparently viable and self-sustaining. Here we evaluate options for monitoring the long-term success of the reintroduction effort using noninvasive techniques to assess species status and distribution. Ideally, this could be accomplished by estimating abundance of lynx in the state on a recurring basis. However, abundance estimation can be difficult for rare, wide-ranging carnivores because such efforts typically require multiple encounters with a number of individuals. Occupancy estimation may be a useful alternative as sampling under this framework requires only detection or non-detection information at the species level rather than multiple encounters with individuals. Models are fit to the detection data, collected over multiple visits to sampling units, to estimate the proportion of sample units used by the focal species within a study area. The monitoring objective may be to simply track this proportion (ψ) through time. However, ψ and abundance are clearly related; when abundance is zero, ψ is zero, and when the landscape is saturated with animals, ψ = 1.0. Thus, an alternative objective may be to use estimated ψ as a surrogate for abundance, and thus track abundance through time using occupancy estimation. We used a series of simulations based on pilot data to assess the effort required to detect declines (or increases) of interest in abundance and ψ of lynx in Colorado using occupancy estimation. We found that small changes could not be detected even with an enormous amount of effort. Even 50% declines or increases in abundance or ψ would require substantial effort and coordination to implement on a statewide basis. Tracking abundance through time using occupancy required relatively more effort than simply tracking a similar decline in ψ itself. Given these results, perhaps a scaled down approach is most practical. That is, CPW could implement a rigorous occupancy estimation program to track abundance, but only in a portion of the state. Elsewhere, rudimentary presence/absence surveys (i.e., surveys 15 �conducted without repeat visits, and probably on a rotating basis so any given mountain range is only visited every ~5 years) could be conducted to ascertain the distribution of lynx among the major mountain ranges and this distribution would tracked through time as a secondary measure of population performance. WILDLIFE RESEARCH REPORT STATEWIDE MONITORING OF CANADA LYNX IN COLORADO: EVALUATION OF OPTIONS JACOB S. IVAN PROJECT NARRATIVE OBJECTIVE Use simulation to assess occupancy estimation as a means of monitoring Canada lynx in Colorado. SEGMENT OBJECTIVES 1. Complete simulations to assess the effort required to track various declines (or increases) in abundance of lynx using occupancy estimation. 2. Complete simulations to assess the effort required to track various declines (or increases) in occupancy of lynx using occupancy estimation. INTRODUCTION The Canada lynx (Lynx canadensis) occurs throughout the boreal forests of northern North America. While Canada and Alaska support healthy populations of the species, the lynx is currently listed as threatened under the Endangered Species Act (ESA) of 1973, as amended (16 U. S. C. 1531 et. seq.; U.S. Fish and Wildlife Service 2000) in the conterminous United States. Colorado represents the southern-most historical distribution of lynx, where the species occupied the higher elevation, montane forests in the state (U.S. Fish and Wildlife Service 2000). However, lynx were extirpated, or reduced to a few animals, in Colorado by the late 1970’s, (U.S. Fish and Wildlife Service 2000) most likely due to multiple human-associated factors including predator control efforts such as poisoning and trapping (Meaney 2002). Given the isolation of and distance from Colorado to the nearest northern populations of lynx, the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) considered reintroduction as the best option to reestablish the species in the state. Therefore, a reintroduction effort was begun in 1997, and 218 lynx were released into Colorado from 1999 – 2006 (Devineau et al. 2010). The goal of the Colorado lynx reintroduction program was to establish a self-sustaining population of lynx. Progress toward this goal was tracked via evaluation of criteria related to lynx survival, fidelity, and recruitment. Recently, CPW determined that the criteria had been met and an apparently viable Canada lynx population currently exists in Colorado (Shenk and Kahn 2010). In order to track the persistence of this new population and thus determine the long-term success of the reintroduction, a minimally-invasive, statewide monitoring program is required. Ideally, this could be accomplished by estimating abundance of lynx in the state on a recurring basis. However, abundance estimation using traditional mark-recapture methods is difficult for rare, wide-ranging carnivores because such it typically requires multiple encounters with a number of individuals (Lukacs 2013). New advances in spatially explicit capture-recapture (Efford et al. 2009, Royle et al. 2009), use of multiple data sources (Sollmann et al. 2013a), and implementation of mark-resight approaches (Sollmann et al. 2013b) make 16 �the problem more tractable as these approaches generally require less intensive capture efforts than traditional mark-recapture. However, they still require some of this work, which can be both difficult and invasive. Alternatively, occupancy estimation may be a useful approach for monitoring lynx (MacKenzie et al. 2006). Such an approach requires several visits to a set of sampling units, but the data collected for each visit is simply detection or non-detection of the focal species (MacKenzie et al. 2003). There is no need for marking or tallying individuals. The detection information is used to estimate the proportion (ψ) of sample units used by the focal species (i.e., “occupancy”) which can then be monitored through time. The advantage in such an approach is that no individual identification is necessary and the information gathered during sampling is generally easier to obtain, especially for rare carnivores. The disadvantage is that information obtained about the population of interest is less resolute (i.e., knowing the proportion of the landscape used by a species is less informative than knowing the number of individuals within it). Finally, monitoring might be accomplished by simply documenting distribution of lynx in the state. Under this approach, the metric of interest to be tracked through time would be the number of mountain ranges (of the 6−8 main ranges) with evidence of use by lynx. Currently lynx are known to be present in the San Juan, Sawatch, and Elk Mountains where the reintroduction and/or associated research occurred. Expansion into other ranges over the long-term could be considered evidence of a successful reintroduction; recession into only 1 range, or none would indicate failure. Monitoring distribution is the least costly approach considered here, but also the least informative and least rigorous. We assume that estimation of abundance is not a viable option due to cost, although this assumption should be formally tested, especially as new statistical techniques arise. We further assume that documenting distribution is the least costly option and is thus logistically feasible. However, there is no power analysis or other statistical considerations associated with this option. Thus, from here forward, we focus on using simulation to assess the feasibility of using occupancy for monitoring lynx in Colorado. Under an occupancy framework, the monitoring objective may be to declare ψ as the metric of interest and simply track it through time as a means of monitoring the lynx population in a coarse sense. However, ψ is clearly related to abundance; when abundance is zero, ψ = 0, and when the landscape is saturated with animals, ψ = 1. Thus, an alternative objective may be to use estimated ψ as a surrogate for abundance, and thus attempt to track abundance through time using occupancy estimation. This may be a preferable approach because abundance is ultimately the quantity of interest. The utility of this idea relies on the strength of the relationship between ψ and abundance, which is partially dependent on sampling effort and partially dependent on the characteristics of the system under study. That is, if home range size and territorial tendencies of the focal species result in an average of 1 individual per sample unit, then ψ can be expected to mirror abundance quite well. However, if the interaction of these characteristics leads to multiple individuals using a sample unit (on average), then ψ and abundance will be relatively decoupled. Abundance could decline fairly significantly before precipitating any change in ψ. We conducted a series of simulations to assess the effort required for using occupancy estimation to detect declines (or increases) of interest in abundance or ψ of lynx in Colorado. Our simulations were calibrated to reflect estimates of ψ and detection probability (p) collected from pilot work in the state. We compare the various alternatives available for monitoring lynx in Colorado and discuss trade-offs associated with each. 17 �METHODS Pilot Work CPW initiated work to evaluate methods for detecting lynx during winter 2009−2010 (Shenk 2009, Ivan and Shenk 2010). Similar to Squires et al. (2012), the pilot study area was divided into 75km2 sample units (roughly the size of a female home range) and 3 methods of detecting lynx were tested in 6 sample units where lynx were known to occur: snow tracking surveys, remote camera surveillance, and hair snags. The daily probability of detecting a lynx given their presence in the unit was 0.70, 0.09, and 0.00 for snow tracking, remote cameras, and hair snares, respectively (Ivan 2011). During winter 2010−2011, pilot work was expanded to include 30 wilderness sample units surveyed via remote camera and 30 accessible units surveyed via snow-tracking. The status of lynx (present or not) in these randomly selected units was unknown. We fit single-season occupancy models to data from each stratum and found ψ = 0.33 and p = 0.40 for wilderness units (Ivan 2012; camera data were collapsed by month into 5 occasions, p = 0.40 for each occasion), and ψ = 0.65 and p = 0.37−0.43 for accessible units (Ivan 2011; based on 3 occasions of snow-tracking surveys). Thus, overall, ψ ≈ 0.50 and p ≈ 0.40 for the pilot study area. Assessment of using occupancy estimation to track changes in ψ To assess the effort required to detect declines or increases of interest in ψ using an occupancy estimation framework, we conducted a series of analyses using the simulation function in Program MARK (White and Burnham 1999). Within the “robust design occupancy” data type (i.e., multi-season occupancy, MacKenzie et al. 2006), we set up a simulation model in which ψ = 0.5 for year 1 and p = 0.4, thus matching estimates derived from pilot work. We then specified linear declines in ψ to 0.45 (10% decline), 0.40 (20% decline), or 0.25 (50% decline) over a 10-year period. We also specified an increase in ψ to 0.75 (50% increase) over a 10-year period. We generated data from each simulation model for 2, 3, 4, or 5 occasions and N = 25, 50, 75, 100, 125, and 150 units sampled. We also considered that sampling may occur annually or only in alternate years. Thus, there were 192 possible combinations of parameters specifying the simulation model (4 levels of change in ψ, 4 levels of occasions, 6 levels of sample size, 2 levels specifying the survey interval), and we generated 1000 simulated datasets for each of the 192 combinations. To each of the 192,000 data sets, we fit 2 estimation models. The first fixed ψ to be constant across the 10 years represented in each data set. The second specified a linear trend in estimated ψ. The true, data-generating model always included a trend of some type. Thus we defined “power” as the proportion of simulations in which Akaike’s Information Criterion (adjusted for small sample size; AICc) selected the correct (second) model by at least 2 AICc units (Burnham and Anderson 2002). In general, given sparse data (such as that generated from only a few occasions and/or for a small number of sample units) AICc will select the constant model as the one that fits the data best because there is not enough information to support anything but the simplest model. As the data become richer (i.e., more occasions and/or larger sample size), AICc will begin to pick the correct model more often. We adopted the conventional benchmark of 0.8 as a cutoff for adequately identifying declines or increases of interest. That is, combinations of sample size and occasions that resulted in power = 0.80 were deemed adequate to confidently detect declines or increases under consideration. Assessment of using occupancy estimation to track changes in abundance To assess the effort required to detect declines or increases in abundance using an occupancy estimation framework, we conducted a series of analysis using the R (R Development Team 2013) package SPACE (Ellis et al. 2013). Specifically, we provided the package with spatially referenced data representing predicted lynx habitat in Colorado (Ivan et al. 2011). The package then randomly assigned home range centers for 125 males and 125 females on this landscape. Home range centers were only 18 �allowed to occur in cells that had reasonable probability of being lynx habitat. To mimic territoriality, males were not allowed to have a home range center within 6 km of another male; females could not be assigned a home range center within 5 km of another female. Males and females could be any distance from each other. Once home range centers were assigned for all 250 lynx, each individual was temporarily assigned a bivariate normal utilization distribution (i.e., the probability of occurrence for each individual was highest at its home range center and dissipated equally in all directions) appropriately sized for each sex. This simplistic utilization distribution was then weighted by the underlying map of predicted lynx habitat to produce an irregular, realistic utilization distribution that was unique to each individual. Thus, following this first step, a realistic number of virtual lynx were distributed across the state and assigned reasonable utilization distributions that governed their movement across the landscape. We then specified declines (50%, 20%, 10%), and increases (50%) in abundance over a decade by randomly removing or adding the appropriate number of individuals at each of 10 time steps. Next, simulated landscapes were overlaid with a sampling grid consisting of 75-km2 sample units. This was done for each of the 10 time steps. Based on utilization distributions assigned to each individual, SPACE computed the probability of at least 1 animal being present in each unit during each time step (i.e., sampling occasion). It then applied the detection probability specified for the simulation to generate detection/non-detection data for each unit. We generated data sets for a variable number of occasions and sample sizes similar to that described above. As before, each simulated dataset was fitted with 2 competing models, one in which the estimated ψ was fixed to be constant throughout the 10-year period, and second in which it followed a linear trend. We again defined power as the proportion of simulations in which AICc selected the correct model. We also considered the impact of sampling every other year by removing data from even years. On average, estimates of ψ and p for the first year of each simulation were 0.50 and 0.34 respectively, which is close to the values observed from the pilot work. Thus, the model was well calibrated to the field. Sampling Details For each of the monitoring metrics, ψ and abundance, we identified the most plausible scenario that could be implemented in the field by CPW personnel, and further detailed the effort required to complete a survey by selecting a mock sample. To accomplish this, we first defined the population of sample units of interest by overlaying a grid of 75-km2 cells on the predicted lynx habitat layer for Colorado (Ivan et al. 2011). We identified cells as potential sample units if at least 50% of the lynx habitat pixels within them had probability values ≥0.60 (See Ivan et al. 2011 for detailed discussion of these probability values). This resulted in a population of 475 cells from which to draw a sample (Fig. 3). Next, we used the R (R Development Team 2013) package ‘spsurvey’ (Kincaid 2013) to enumerate each sample unit in a spatially balanced random fashion such that a valid sample of any size could be selected by simply ordering the cells by their randomly assigned number and selecting the 1st N cells. For each scenario of interest, we selected an appropriate sample, then summarized the effort required to complete the sample by CPW Area. We assumed 6 person-days would be required to sample each non-wilderness unit and 10 person-days would be required to sample each wilderness unit. These estimates were based on pilot work and assume that for snow-tracking surveys (non-wilderness units), personnel would work in pairs and complete 3 visits per sample unit. For wilderness units, we assumed personnel would work in pairs over 2.5 days to set 4 cameras in each selected unit, then work another 2.5 days per unit to retrieve the cameras after sampling. These represent minimum estimates of cost as any survey effort would also require personnel time to maintain snowmobiles and cameras, enter data, and complete analyses and reports. 19 �RESULTS Regardless of whether the objective was to use occupancy estimation to detect declines in ψ or abundance, power was low (≤ 0.40) for all but the most drastic changes in the lynx population, even with significant survey effort (e.g., N = 125−150; Fig. 1, 2). Fifty percent declines or increases in ψ over a decade could be adequately detected (power = 0.80) with 3 visits to each of 75 sample units if sampling occurred on an annual basis (Fig. 1). Reducing sampling effort to every other year did not impart dramatic changes to the sample size needed to maintain power. In contrast, annual surveys comprising 3 visits to 125 sample units were required to adequately detect 50% declines or increases in abundance over the same time span (Fig. 2). Also, in the case of abundance, reducing survey effort to every other year required ~250 units in order to maintain power. Power curves in the panels representing results for 20% and 10% declines in abundance (Fig. 2) were relatively high at very small sample sizes, then declined with increasing sample size before increasing again at large sample sizes. These counterintuitive results are likely artifacts of fitting models to sparse data. When data are sparse, parameters may not be estimated well and the model may return values at a boundary (i.e., ψ will be estimated as either 0 or 1). If the estimates of ψ near the beginning and/or end of the time series are returned as 0 or 1, then a trend may be detected. In an actual analysis with a single data set, such a phenomenon is easy to diagnose and alternatives are available to tweak the model and prevent this from happening. However, when thousands of datasets and model fits are involved, such tweaking is impossible. Thus, these high initial values and subsequent declines should be ignored. Power to detect trends across this range of sample sizes is likely very low. The most plausible scenarios for monitoring either ψ or abundance were those aimed at detecting a 50% change in either metric. Selection of an actual sample revealed that in both cases, the number of person-hours involved to carry out the sampling was substantial (Fig. 4, 5). For example, the scenario intended to provide an 80% chance of detecting a 50% decline in abundance over 10 years would require making 3 visits to each of 125 sample units on an annual basis. Assuming 2 Biologists, 2 District Wildlife Managers and 2 USFS Biologists were willing the carry out the work in each area, monitoring lynx under this scheme would require on average about 10 days worth of work per person per Area (Fig. 4; on average 64 person-days would be required per area; 64 person-days/6 people ≈10 days). Some Areas would require nearly 3 times that effort (Fig. 4; maximum estimated effort was 184 person-days; 184 person-days/6 people ≈ 30 days of work per person). The scenario intended to provide an 80% chance of detecting a 50% decline in ψ over 10 years was projected to require an average of 38 personhours to complete per Area, or ~6 days per person if the same set of biologists and managers participated. Again, effort in some Areas would be nearly 3 times higher. DISCUSSION We rigorously tested the power to detect various changes in population status of Canada lynx in Colorado using occupancy estimation. Small changes (10% or 20% declines) could not be detected with any reasonable amount of effort. Detection of large changes (50% declines or increases in either abundance or ψ) may be possible but would require considerable investment and coordination among management entities. This was especially true for the scenarios aimed at detecting changes in abundance, which is the more preferable approach as it would be most informative. Detecting large changes in ψ over a 10-year period required just more than half of the effort required to detect the same change in abundance, thus making it more feasible. However, this level of effort would still be costly and the information gained would be of low resolution. That is, by the time ψ declines by 50%, a significant number of individuals would be lost from the landscape, and it may be too late for any action to counter the decline. Monitoring distribution rather than abundance or occupancy would likely be the most 20 �affordable option but it is also least informative and least rigorous. In fact, it was not evaluated in this report because it is completely absent of any statistical underpinnings. Furthermore, the distribution approach provides little opportunity to learn why changes are happening. The multi-season occupancy models employed here to track abundance or ψ include extinction and colonization parameters (which we have largely ignored for the purposes of simulation). Modeling these parameters may provide an opportunity to associate changes on the landscape (e.g., bark beetle outbreaks, wildfire, timber harvest) with changes in ψ, thus providing an opportunity to learn why changes are occurring. Clearly trade-offs exist between containing costs and implementing a program that is meaningful, rigorous, and provides opportunities for continued learning. Perhaps the most practical way forward is a hybrid approach in which CPW implements a rigorous occupancy estimation program to track abundance, but only in a portion of the state, while simultaneously implementing the relatively less rigorous distributional approach statewide. Such an approach would provide detailed information about a (hopefully) representative subpopulation of lynx, but would be easier to implement as it would take less effort it would only be implemented in a portion of the state. Additionally, CPW would still obtain useful information regarding the statewide distribution of animals. If CPW were to adopt such a strategy, we suggest that the rigorous portion of the effort focus on the San Juan Range in the southwest as it provides the bulk of the lynx habitat and has long been considered a core stronghold for the species. Thus, it could be considered a “sentinel” area such that increases in the lynx population there probably bode well for the rest of the state, and declines there probably bode poorly. ACKNOWLEDGMENTS We thank Eric Odell, Gary White, Larissa Bailey, Paul Lukacs, and Fred Allendorf for initial and ongoing discussions relative to monitoring rare carnivores in general and lynx in particular. We thank the Rocky Mountain Research Station and a PECASE award to Mike Schwartz for initial funding toward simulation work. Funding specific to simulations regarding lynx in Colorado was provided by a U.S. Fish and Wildlife Service Section 6 Grant to Colorado Parks and Wildlife. LITERATURE CITED Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd edition. Springer, New York. Devineau, O., T. M. Shenk, G. C. White, P. F. Doherty, P. M. Lukacs, and R. H. Kahn. 2010. Evaluating the Canada lynx reintroduction programme in Colorado: patterns in mortality. Journal of Applied Ecology 47:524-531. Efford, M. G., D. L. Borchers, and A. E. Byrom. 2009. Density estimation by spatially explicit capturerecapture: likelihood based methods. Pages 255-269 in G. P. Patil, D. L. Thomson, E. G. Cooch, andM. J. Conroy, editors. Modeling Demographic Processes in Marked Populations. Springer Science, Boston, Massachusetts, USA. Ellis, M. M., J. S. Ivan, and M. K. Schwartz. 2013. Spatially explicit power analyses for occupancy-based monitoring of wolverine in the U.S. Rocky Mountains. Conservation Biology In Press:xxx-xxx. Ivan, J. S. 2011. Monitoring Canada lynx in Colorado using occupancy estimation: Initial implementation in the Core Lynx Research Area. Colorado Division of Parks and Wildlife. Ivan, J. S. 2012. Monitoring Canada lynx in Colorado using occupancy estimation: Initial implementation in the Core Lynx Research Area. Colorado Division of Parks and Wildlife. Ivan, J. S., M. Rice, P. M. Lukacs, T. M. Shenk, D. M. Theobald, and E. Odell. 2011. Predicted lynx habitat in Colorado. Colorado Division of Parks and Wildlife. Ivan, J. S., and T. M. Shenk. 2010. Estimating the extent, stability and potential distribution of Canada Lynx (Lynx canadensis) in Colorado: initial implementation in the core lynx research area. Colorado Division of Wildlife. 21 �Kincaid, T. 2013. R package 'spsurvey'. Version 2.5:http://www.epa.gov/nheerl/arm/. Lukacs, P. M. 2013. Closed population capture-recapture models. Pages 14.11 - 14.39 in E. Cooch, andG. C. White, editors. Program MARK: A Gentle Introduction MacKenzie, D., J. D. Nichols, J. E. Hines, M. G. Knutson, and A. B. Franklin. 2003. Estimating site occupancy, colonization, and local extinction when a species is detected imperfectly. Ecology 84:2200-2207. MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. Academic Press, Oxford, United Kingdom. Meaney, C. 2002. A review of Canada lynx (Lynx canadensis) abundance records from Colorado in the first quarter of the 20th century. Colorado Department of Transportation Report. Royle, J. A., J. D. Nichols, K. U. Karanth, and A. M. Gopalaswamy. 2009. A hierarchical model for estimating density in camera-trap studies. Journal of Applied Ecology 46:118-127. Service, U. S. F. a. W. 2000. Endangered and threatened wildlife and plants: determination of threatened status for the contiguous U. S. distinct population segment of the Canada lynx and related rule, final rule. Federal Register 65:16052–16086. Shenk, T. M. 2009. Post-Release Monitoring of Lynx Reintroduced to Southwestern Colorado. Colorado Division of Wildlife. Shenk, T. M., and R. H. Kahn. 2010. The Colorado lynx reintroduction program. Colorado Division of Wildlife. Sollmann, R., B. Gardner, R. B. Chandler, D. B. Shindle, D. P. Onorato, J. A. Royle, and A. F. O'Connell. 2013a. Using multiple data sources provides density estimates for endangered Florida panther. Journal of Applied Ecology 00:000-000. Sollmann, R., B. Gardner, A. W. Parsons, J. J. Stocking, B. T. McClintock, T. R. Simons, K. H. Pollock, and A. F. O'Connell. 2013b. A spatial mark–resight model augmented with telemetry data. Ecology 94:553-559. Squires, J. R., L. E. Olson, D. L. Turner, N. J. DeCesare, and J. A. Kolbe. 2012. Estimating detection probability for Canada lynx, Lynx canadensis, using snow-track surveys in the northern Rocky Mountains, Montana, USA. Wildlife Biology 18:215-224. Team, R. D. C. 2013. R Foundation for Statistical Computing, Vienna, Austria. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 Supplement:120-138. Prepared by Jake Ivan, Wildlife Researcher 22 �en 50%_ Decline C 20%_Decline 10%_Decline 50%_Increase 2 1.0 0 0.8 ~ _2l 0.6 - Visits 50 4 - E 02 -0 - 2 ~ 10 ]j 0.8 - - 3 - 4 ~ 06 - - s a, ~ 04 - u 0.2 ~ 0 U I I I 25 50 75 I I I I 100 125 15025 I I 50 ~ I 1001~1~~ ~ ~ I I I I 1001~1~~ I I ~ ~ I I I 1001~1~ Number of cells sampled Figure 1. Power to detect various changes in the proportion (ψ) of sample sites used by lynx in Colorado using occupancy estimation. Changes were assumed to occur over a 10-year period. Power is shown for scenarios in which sample units were sampled annually (top panels) and when sampling occurred only in alternate years (bottom panels). “Visits” corresponds to the number of times selected units would be searched to collect detection/non-detection data. Visits could represent days for units surveyed via snow tracking, or they may represent blocks of time into which continuously collected camera data could be binned (e.g., 1 visit = 1 month of camera sampling). 23 �50% Decline 20% Decline 10% Decline 50% Increase Annual Alternate I I I t I I I I t t I I I I I I I I I 250 500 750 1000 250 500 750 1000 250 500 750 1000 250 500 750 1000 - a.a #visits - 2 3 4 Number of cells sampled Figure 2. Power to detect various changes in abundance of lynx in Colorado using occupancy estimation. Changes were assumed to occur over a 10-year period. Power is shown for scenarios in which cells are sampled annually (top panels) and when sampling occurs only in alternate years (lower panels). “Visits” corresponds to the number of times selected units would be searched to collect detection/non-detection data. Visits could represent days for units surveyed via snow tracking, or they may represent blocks of time into which continuously collected camera data could be binned (e.g., 1 visit = 1 month of camera sampling). 24 �Figure 3. Predicted lynx habitat (red pixels = good, blue pixels = poor) in Colorado overlaid with 75-km2 sample units (black squares, N = 475) from which to select a sample for monitoring declines or increases of interest in ψ or abundance. Only units where at least half of the lynx habitat pixels within them had probability values ≥0.60 were included in the population to sample from. See Ivan et al. (2011) for details regarding construction of the predicted lynx habitat map and interpretation of pixels that comprise it. 25 �Region Northeast Northwest Southeast Southwest Total Area 1 2 6 7 8 9 10 11 13 14 15 16 17 18 #Sample Units 6 1 2 1 13 3 2 1 4 1 8 12 18 3 75 Effort (person-days) 44 10 16 6 102 18 12 6 28 6 52 76 128 22 526 Figure 4. Map and tabular summary of a spatially balanced random sample of N = 75 cells selected for monitoring a 50% decline or increase in ψ over a 10-year period in Colorado, USA using a combination of snow-track surveys and remote camera surveys. Estimated effort accounts for the differential time required to sample wilderness (cameras) and non-wilderness (snow-tracking) units. 26 �Region Northeast Northwest Southeast Southwest Area #Sample Units Effort (person-days) 11 3 3 3 17 5 5 4 6 2 17 19 26 4 125 82 26 22 18 134 30 34 24 44 12 122 130 184 28 890 1 2 6 7 8 9 10 11 13 14 15 16 17 18 Total Figure 5. Map and tabular summary of a spatially balanced random sample of N = 125 cells selected for monitoring a 50% decline or increase in abundance of lynx over a 10-year period in Colorado, USA using a combination of snow-track surveys and remote camera surveys. Estimated effort accounts for the differential time required to sample wilderness (cameras) and non-wilderness (snow-tracking) units. 27 � https://cpw.cvlcollections.org/files/original/0481b19251749efd65d4fa09f8e61bff.pdf 2e369aa296f0897661166329223aaeb4 PDF Text Text Colorado Division of Parks and Wildlife July 2010 – June 2011 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Federal Aid Project No. Colorado 3430 3001 : : : : Division of Parks and Wildlife Mammals Research Deer Conservation Assessment of survival and optimal harvest Strategies of adult male mule deer in Middle Park, Colorado W-185-R Period Covered: July 1, 2010 - June 30, 2011 Author: E.J. Bergman; Project Cooperators; C.J. Bishop, K. Oldham, and L. Sidener Personnel: G. Abram, G. Birch, J. Broderick, M. Crosby, B. Davies, T. Elm, D. Gillham, K. Holinka, A. Holland P. Lukacs, B. Manly, S. Murdoch, S. Schwab, S. Shepherd Colorado Division Parks and Wildlife R. Swisher, S. Swisher, T. McKendrick Quicksilver Air All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We developed a study plan and initiated field work on a study designed to assess the survival and optimal harvest strategies of adult male mule deer in Middle Park, Colorado. Three years of baseline survival data for adult (≥ 1 yr. old) male deer will be collected before implementing a harvest management action that will redistribute hunters within DAU-9. One hundred adult (1.5 years old and older) male deer were captured and radio collared. The survival rate for these deer was estimated at 0.879 (SE = 0.0326) for the first survival period (January through July). 97 �WILDLIFE RESEARCH REPORT ASSESSMENT OF SURVIVAL AND OPTIMAL HARVEST STRATEGIES OF ADULT MALE MULE DEER IN MIDDLE PARK, COLORADO ERIC J. BERGMAN P.N. OBJECTIVES SEGMENT OBJECTIVES 1. Develop a project study plan to address the lack of knowledge regarding survival and harvest strategies of adult mule deer. 2. Initiate field work in the form of capturing and radio collaring animals. 3. Collect survival data on radio collared deer and provide preliminary survival estimates for adult male mule deer. INTRODUCTION Historically, management of big game species has focused on the performance of adult females and the young of the year segments of the population. In the case of mule deer, this has been further refined to the aspects of annual (for adult females) and overwinter (for young of the year) survival. The performance of the male component of populations was deemed less important because it takes few males to provide adequate breeding coverage for the population, and historic harvest management objectives were set to maximize hunting opportunities. As long as sufficient numbers of males were available to breed females there was no desire to restrict hunting opportunity. However, during the past 10-15 years, the management of big game populations, and mule deer populations in particular, has shifted from the objective of providing maximal opportunity towards providing higher quality opportunities (Bishop et al. 2005b, Bergman et al. 2010). High quality opportunities are typically defined by hunters as a combination of the chance to see a greater number of male deer during the hunt, increased potential to harvest an older age class animal (i.e., an animal with more developed antler morphometry), but also reduced interaction and competition with other hunters. In response to this shift in hunter desires and concerns over declining mule deer numbers, the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) implemented a statewide limitation in deer hunting in 1999. This statewide limitation gave CPW the ability to reduce total hunter numbers but also the ability to control the distribution of hunters throughout the state. Since 1999 Colorado’s deer herds have become composed of a greater number of males, yet little biological data on them exist. Also stemming from this change in harvest management was a new responsibility for Colorado’s terrestrial biologists and wildlife managers. Prior to 1999, licenses were sold over-the-counter and were not limited in number (i.e., any hunter who wished to purchase one was able to do so), and the decision of how many licenses to make available did not need to be considered. Since 1999, CPW has the added responsibility of deciding how many licenses should be allocated in each Data Analysis Unit (DAU). This decision must reflect a balance between meeting DAU population performance objectives, but also provide as much hunter opportunity as possible. Big game populations in Colorado are currently modeled using multiple sources of biological data (White and Lubow 2002). Model inputs include harvest estimates, young recruitment to December, and measured rates of survival of adult females and fawns. Also, the ratio of adult males to adult females is estimated and used to align models by minimizing the difference between observed and modeled values. Only rarely have the survival rates of adult males been measured. This gap in knowledge has historically been viewed as trivial and adult male survival rates have been assumed to be similar to the rates of females. Similarly, it has been assumed that natural survival rates (i.e., post hunt survival) of 98 �males do not geographically vary. However, model performance under these assumptions has been poor and the need to measure adult male survival as a parameter has increased. Presently, a number of population models in Colorado suggest that natural adult male survival may be lower than adult female survival, yet empirical data is lacking to verify these suppositions. Despite this apparent lack of information, survival of adult male mule deer and adult male blacktail deer are not completely novel parameters of interest (Pac and White 2007, Bender et al. 2004b, Bleich et al. 2006, Bishop et al. 2005a, McCorquodale 1999). These studies also suggest that adult male mule deer survival tends to be lower than adult female survival when differences occur, further emphasizing the need to rigorously evaluate adult male survival rates. Bishop et al. (2005a) observed lower natural survival rates of adult males than adult females in southwest Idaho: differences were most apparent during winter in 2 of 3 study areas. Pac and White (2007) found that natural survival rates of yearling males in Montana were lower than the average adult female survival rate documented by Unsworth et al. (1999). Finally, Miller et al. (2008) found that adult male survival rates were lower than adult female survival rates in Colorado in response to chronic wasting disease (CWD). In particular to the population modeling interests of Colorado outside the CWD endemic area, the work conducted by Pac and White (2007) has had the greatest utility. This work focused on the survival of males under differing management scenarios and showed a shift in cause-specific mortality of males in areas where harvest was more restricted. It is currently unknown if survival rates would be similar between Montana and Colorado. Similarly, the likelihood of observing shifts in mortality sources is unknown. It has been demonstrated that adult female deer herds in Colorado tend to be habitat limited (Bishop et al. 2009, Bartmann et al. 1992), but the trade-off between harvest, habitat and survival in adult male mule deer has not been explored. An additional need in Colorado pertains to the harvest management of adult male mule deer. As discussed above, a large shift in mule deer herd size and structure occurred as a result of changes in harvest management. Overall, this shift has been viewed as positive by both CPW as well as the public. However, CPW maintains the responsibility of optimally managing the deer of Colorado and maximizing hunting opportunity under this new set of constraints. To date, CPW has had limited biological information and data to guide harvest management decisions. In particular for this issue, as Data Analysis Units (DAUs) reach and surpass their adult male: adult female ratio objectives, CPW typically responds by increasing the number of available hunting licenses. In situations where herds are continually lower than DAU objectives, available hunting licenses are reduced. What remains unknown about survival of adult male deer is at what level natural survival is reduced due to intraspecific competition (i.e., increased density of adult male deer). If, or when deer herds exceed the adult male: adult female objectives for DAUs, it is often assumed that the surplus of male deer remain in the population into perpetuity. However, this assumption is based on the premise that compensatory mortality does not occur. Similarly, it assumes that annual variation in survival is negligible. However, these assumptions are not biologically realistic. It is possible that herds with large post-hunt populations of adult males experience higher levels of non-harvest mortality. Under this scenario, harvest has not been optimized and more hunters could have been afforded the opportunity to hunt with no effect on post hunting season ratios of adult males to adult females. The most effective way to learn about the mortality process is via manipulative experimentation, but to date this topic has not been deemed a high enough priority to pursue. STUDY AREA This study is taking place in Middle Park, Colorado (see Appendix I for discussion of criteria for study area selection). Under the current management structure, Middle Park falls within DAU D-9. Within D-9 are 6 Game Management Units (27,181, 18, 37, 371, and 28; Fig. 1). Due to the geologic and topographical landscape in Middle Park, this area is conducive to splitting the DAU into experimental units (see Appendix I for experimental design). Additionally, from a management perspective, D-9 is 99 �currently managed for 35 adult males per 100 adult females. This ratio objective represents an average “quality” management objective in Colorado (i.e., DAUs with higher or lower objectives exist, thus data from D-9 will be the most universally applicable). Finally, the topography and landscape of Middle Park also makes it prone to periodic, harsh winters. This variability is fundamental to attaining reasonable estimates of process variation in adult male survival. METHODS Capture of adult male deer was initiated in January of 2011. Capture was conducted via helicopter net-gunning (Webb et al. 2008, Potvin and Breton 1988, White and Bartmann 1994, Barrett et al. 1982). All captures occurred after the completion of the 4th rifle hunting season, eliminating conflicts between capture efforts and hunting. Due to the need to generate survival estimates linked to animals of known age, all animals were handled by CPW personnel for aging purposes. Field aging of animals was done by visual inspection of tooth wear patterns (Severinghaus 1949, Robinette et al. 1957, Hamlin et al. 2000). Colorado Parks and Wildlife researchers/biologists were ferried to the general area in which capture was occurring and subsequently ferried the short distance to each capture location after individual animals were captured. Prior to release, all animals had their antlers removed via handsaw to minimize the potential risk of injury as the animal was released. All captures occurred after annual mule deer classification flights had been conducted, alleviating the potential for misclassification of antlerless males as females. All deer were fitted with expandable radio collars (see Appendix I for discussion of radio collar development). All radio collars were equipped with mortality sensors that doubled in pulse rate after remaining motionless for 4 hours. Between the time of capture and mid-June, we used ground based monitoring to determine the live/dead status of deer 3-4 times per week. Additionally, every 5-10 days we conducted a telemetry flight to hear any animals that hadn’t been heard from the ground during the preceding week. A general location was collected for each radio marked deer in early-March to determine if it had departed the GMU in which it had originally been captured. From mid-June through remainder of the summer, deer were monitored from the ground weekly and from the air once per month. When detected, all mortalities were investigated as quickly as possible to determine cause of death and to get an accurate estimate of the date of death. To help evaluate the effects of a changing sex ratio on hunter harvest, we are currently preparing to sample successful hunters to acquire an age of animals harvested in D-9. Ages will be estimated via the cementum aging process of incisors (Hamlin et al. 2000). When possible, a lower incisor was also collected from each radio collared deer that died in order to validate animal ages of captured animals. To acquire teeth for aging purposes, all hunters who have licenses to hunt in any GMU in D-9 will be contacted via mail. Each hunter will be provided with a sampling kit, a pre-posted return envelope and detailed directions on how to extract teeth for aging purposes. These data will help inform terrestrial biologists and wildlife managers if changes in the age of animals harvested occur as populations shift up or down in age structure as sex ratios are increased or decreased. RESULTS AND DISCUSSION In January, 100 deer were captured, aged, weighed, radio collared and released during a 3½ day period. On one occasion, the skull plate of an animal was fractured immediately anterior to the animal’s antler pedestals while being captured. This animal was immediately euthanized via gunshot to the head. No other capture related injuries or mortalities occurred. With the exception of animals falling in the 2 youngest age classes (1½ years old (yearlings) and 2½ years old), the age distribution of captured animals followed the expected age distribution of the 100 �population (Fig. 2). In the case of the 2 youngest age classes, we captured more 2½ year old animals than yearlings. We believe this result was primarily due to misidentification of yearlings as part of the capture process. Small yearlings and particularly those with small antler morphometry had a greater probability of being skipped by the capture crew as they flew over groups of deer. In future years we will make a more concerted effort to increase the number of yearlings in the sample. The mass of adult male deer ranged between 52.3 kg and 106.8 kg, with the average mass being 82.2 kg (Fig. 3). Observationally, the largest animals appeared to be captured in areas in close proximity to irrigated agricultural fields. Survival of adult male deer between the time of capture and the end of July was high. Combined survival for the northern and southern halves of D-9 was 0.879 (SE = 0.0326). When separated, the survival rates for the northern and southern halves of D-9, for the same time period, were 0.858 (SE = 0.0495) and 0.900 (SE = 0.0495). Of the 12 mortalities that occurred, a suite of causes were observed. Six mortalities were attributed to predation (4 coyote, 2 mountain lion), 1 was attributed to starvation, 1 to disease (conjunctivitis that blinded the animal), and 2 were attributed to vehicular collisions (1 automobile and 1 train). The cause of mortality could not be determined for 2 deer. Survival patterns during the winter months during the first year did not demonstrate dramatic swings or mortality pulses during which several animals died. Rather, mortalities tended to occur at a relatively constant interval of approximately 2-3 mortalities per month. However, with the exception of the animal killed by a train, mortalities during the summer months (June and July) were not observed. SUMMARY Project efforts were successful during the first year of the study. Capture and handling of animals was efficient, cost effective and mortality/injury rates were low. The survival rate of adult male mule deer was high. Baseline data collection will continue for 2 additional winters before implementation of the harvest management experiment. LITERATURE CITED Barrett, M.W., J.W. Nolan, and L.D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108-114. Bartmann, R.M., G.C. White, and L.H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monographs 121:1-39. Bender, L.C., J.C. Lewis and D.P. Anderson. 2004a. Population ecology of Columbian black-tailed deer in Urban Vancouver, Washington. Northwestern Naturalist 85:53-59. Bender,L.C., G.A. Schirato, R.D. Spencer, K.R. McAllister, and B.L. Murphie. 2004b. Survival, causespecific mortality, and harvesting of male black-tailed deer in Washington. Journal of Wildlife Management 68:870-878. Bergman, E.J., B.E. Watkins, C.J. Bishop, P.M. Lukacs, and M. Lloyd. 2010. Biological and socioeconomic effects of statewide limitation of deer licenses in Colorado. Journal of Wildlife Management. In Review. Bishop, C.J., J.W. Unsworth and E.O. Garton. 2005a. Mule deer survival among adjacent populations in southwest Idaho. Journal of Wildlife Management 69:311-321. Bishop, C.J., G.C. White, D.J. Freddy and B.E. Watkins. 2005b. Effect of limited antlered harvest on mule deer sex and age ratios. Wildlife Society Bulletin 33:662-668. Bishop, C.J., G.C. White, D.J. Freddy, B.E. Watkins, and T.R. Stephenson. 2009. Effect of enhanced nutrition on mule deer population rate of change. Wildlife Monographs 172:1-28. Bleich, V.C., B.M. Pierce, J.L. Jones and R.T. Bowyer. 2006. Variance in survival of young mule deer in the Sierra Nevada, California. California Fish and Game. 92:24-38. 101 �Hamlin, K.L., D.F. Pac, C.A. Sime, R.M. DeSimone, and G.L. Dusek. 2000. Evaluating the accuracy of ages obtained by two methods for Montana ungulates. Journal of Wildlife Management. 64:441449. Miller, M.W., H.M. Swanson, L.L. Wolfe, F.G. Quatarone, S.L. Huwer, C.H. Southwick and P.M. Lukacs. 2008. Lions and prions and deer demise. Plos One 3:1-7. McCorquodale, S.M. 1999. Movements, survival, and mortality of black-tailed deer in the Klickitat Basin of Washington. Journal of Wildlife Management 63:861-871. Pac, D.F., and G.C. White. 2007. Survival and cause-specific mortality of male mule deer under different hunting regulations in the Bridger Mountains, Montana. Journal of Wildlife Management 71:816-827. Potvin, F., and L. Breton. 1988. Use of a net gun for capturing white-tailed deer, Odocoileusvirginianus, on Anticosti Island, Quebec. Canadian Field Naturalist 102:697-700. Robinette, W.L., J.S. Gashwiler, D.A. Jones, and H.S. Crane. 1957. Notes on tooth development and wear for Rocky Mountain mule deer. Journal of Wildlife Management 21:134-153. Severinghaus, C.W. 1949. Tooth development and wear as criteria of age in white-tailed deer. Journal of Wildlife Management 13:195-216. Underwood, A.J. 1994. On beyond BACI-sampling designs that might reliably detect environmental disturbances. Ecological Applications. 4:3-15. Unsworth, J.W., D.F. Pac, G.C. White, and R.M. Bartmann. 1999. Mule Deer Survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. Webb, S.L., J.S. Lewis, D.G. Wewitt, M.W. Hellickson, and F.C. Bryant. 2008. Assessing the helicopter and net gun as a capture technique for white-tailed deer. Journal of Wildlife Management 72:310-314. White, G.C. and R.M. Bartmann. 1994. Drop nets versus helicopter net guns for capturing mule deer fawns. Wildlife Society Bulletin 22:248-252. White, G.C. and B.C. Lubow. 2002. Fitting population models to multiple sources of observed data. Journal of Wildlife Management 66:300-309. Prepared by Eric J. Bergman, Wildlife Researcher 102 �Figure 1. Data Analysis Unit 9 (D-9) encompasses the Middle Park area of central Colorado. D-9 includes 6 Game Management Units (27, 181 and 18 on the northern half and 37, 371 and 28 on the southern). Current management sex ratio management objectives for D-9 are consistent across GMUs with an overall post hunt objective of 35 adult males per 100 adult females. 103 �30 ... 25 ...,,, 20 u ... "C (I) ::i CL (I) (I) -... Cl 15 0 (I) .0 10 E ::i z 5 0 1.5 2.5 3.5 4.5 5.5 6.5 7.5 8.5 9.5+ Estimated Age of DeerCaptured Figure 2. Frequency of ages of 100 adult male mule deer captured in January (2011) in Middle Park, Colorado. Future capture efforts will be made to increase the frequency of 1½ year old males to accommodate for an expected underrepresentation in the current sample as well as for aging of radio collared animals throughout the study. 16 14 "O 12 ...:::, 10 ..... (1) C. n, u ... (1) (1) 0 8 6 :i:t 4 2 0 so so- s'l> 'o,.. 'o'o 10 10- 1'l> ~,.. ~'o '!Jo '!Jc- C!J'b "01- \.o'o \.\.o Mass(kg) Figure 3. Frequency of masses of 100 adult male mule deer captured in January (2011) in Middle Park, Colorado. Ages of deer captured ranged between 1½ years old and in excess of 9½ years old. 104 �APPENDIX I PROGRAM NARRATIVE STUDY PLAN FOR MAMMALS RESEARCH FY 2010-11 – FY 2015-16 State of: Cost Center: Work Package: Task No. Federal Aid Project No. Colorado 3430 3001 : : : : : Division of Wildlife Mammals Research Deer Conservation Assessment of Survival and Optimal Harvest Strategies of Adult Male Mule Deer in Middle Park, Colorado. W-185-R Assessment of Survival and Optimal Harvest Strategies of Adult Male Mule Deer in Middle Park, Colorado Principal Investigators Eric J. Bergman, Mammals Researcher, Colorado Parks and Wildlife Chad J. Bishop, Mammals Researcher Leader, Colorado Parks and Wildlife Kirk Oldham, Terrestrial Biologist, Colorado Parks and Wildlife Lyle Sidener, Area Wildlife Manager, Colorado Parks and Wildlife Cooperators Andy Holland, Big Game Coordinator, Colorado Parks and Wildlife John Broderick, Terrestrial Management Leader, Colorado Parks and Wildlife Area 9 Personnel, Colorado Parks and Wildlife STUDY PLAN APPROVAL Prepared by: Eric J. Bergman Date: Nov. 2010 Submitted by: Eric J. Bergman Date: Nov. 2010 Reviewed by: Chuck Anderson Date: Nov. 2010 Mike Phillips Date: Biometrician: Paul Lukacs Date: Nov. 2010 Approved by: Chad Bishop Mammals Research Leader Date: Nov. 2010 105 �PROGRAM NARRATIVE STUDY PLAN FOR MAMMALS RESEARCH ASSESSMENT OF SURVIVAL AND OPTIMAL HARVEST STRATEGIES OF ADULT MALE MULE DEER IN MIDDLE PARK, COLORADO A Study Plan Proposal Submitted by: Eric J. Bergman, Mammals Researcher, Colorado Parks and Wildlife Chad J. Bishop, Mammals Research Leader, Colorado Parks and Wildlife Kirk Oldham, Terrestrial Biologist, Colorado Parks and Wildlife Lyle Sidener, Area Wildlife Manager, Colorado Parks and Wildlife Andy Holland, Big Game Coordinator, Colorado Parks and Wildlife John Broderick, Terrestrial Management Leader, Colorado Parks and Wildlife A. Need Historically, management of big game species has focused on the performance of the female and the young of the year components of the population. In the case of mule deer, this has been further refined to the aspects of annual (for adult females) and overwinter (for young of the year) survival. The performance of the male component of populations was deemed less important, primarily due to the fact that it takes relatively few males to provide adequate breeding potential for the population. Additionally, historic harvest management objectives were set to maximize hunting opportunities. Thus, as long as sufficient numbers of males were available to breed females there was no desire to restrict hunting opportunity. However, during the past 10-15 years, the management of big game populations, and mule deer populations in particular, has started to shift away from the objective of providing maximal opportunity towards providing fewer but higher quality opportunities (Bishop et al. 2005b, Bergman et al. 2010). High quality opportunities are typically defined by hunters as a combination of the chance to see a greater number of male deer during the hunt, and the potential to harvest an older age class animal (i.e., an animal with more developed antler morphometry), but also reduced interaction and competition with other hunters. In response to this shift in hunter desires and concerns over declining mule deer numbers, the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) implemented a statewide limitation in deer hunting in 1999. This statewide limitation gave CPW the ability to greatly reduce total hunter numbers but also the ability to control the distribution of hunters throughout the state. Since 1999, a few marked changes in Colorado’s deer herd have occurred. First, due to reduced harvest an overall increase in deer numbers has been observed (Fig. 1). Second, because the reduction in harvest was primarily focused on adult males, a subsequent increase in the ratio of adult males to adult females has resulted (Fig. 2) (Bergman et al. 2010). Stemming from this shift in harvest management and the subsequent changes in herd size and structure, a gap in biological information has been identified. Specifically, Colorado’s deer herds have become composed of a greater number of males, yet little biological data on them exist. Also stemming from this change in harvest management was a new responsibility for Colorado’s terrestrial biologists and wildlife managers. Prior to 1999, licenses were sold over-the-counter and were not limited in number (i.e., any hunter who wished to purchase one was able to do so), and the decision of how many licenses to make available did not need to be considered. Since 1999, CPW has the added responsibility of deciding how many licenses should be allocated in each Data Analysis Unit (DAU). This decision must further reflect a balance between meeting DAU population performance objectives, but also provide as much hunter opportunity as possible. 106 �700000 C: 0 'Z 650000 ..!!! ::::s a. ~ ... 600000 QJ QJ .... 550000 C 0 QJ N 500000 vi "'C ....ro 450000 'Z V'I 400000 QJ E w ..,,_q<o1 ..,,_q<oq ..,,_qq\. ..,,_qq, ..,,_qq? ..,,_qq1 ..,,_q'?>q ,,.oo\. ,_oo, ,,.o o? Year Figure 1. Colorado’s statewide deer herd estimate covering the past 2+ decades. Between 1998 and 1999 CPW implemented a statewide limitation process on the number of deer licenses sold. Since that time, a marked reversal in population trajectory has occurred, largely due to the increase in survival of adult males from reduced hunting license allocation. 80 • .,., 70 • • .!!! ro E 60 • QJ LL ±! ::::s 50 0 0 .... 40 ~ 30 "'i'ii • • "'C <( I • • •• I • • • • ••• •• • • • I •• • • I • • I •• • • • •• •• • • • •• • • • • I I I • .... 20 "'C <( 10 • • 0 1980 I •• I • •• ••• • • •• ~ :i • I I •• •I •• •• • I • I 1984 1988 • • 1992 1996 2000 2004 Year Figure 2. Estimates of adult male: adult female ratios, collected via aerial survey, in the DAUs in western Colorado during the past three decades. Of note, between 1998 and 1999, CPW implemented a statewide limitation on the number of deer hunting licenses sold. The harvest management action brought about a marked increase in estimates of the ratio of adult males to adult females. 107 �Big game populations in Colorado are currently modeled using multiple sources of biological data (White and Lubow 2002). Model inputs include harvest, young recruitment to December, and measured rates of survival of adult females and fawns. Also, the ratio of adult males to adult females is estimated and used to align models by minimizing the difference between observed and modeled values. Very rarely have the survival rates of adult males been measured. This gap in knowledge has historically been viewed as trivial and rates have been assumed to be similar to the rates of females. Similarly, it has been assumed that natural survival rates (i.e., post hunt survival) of males do not geographically vary. However, model performance under these assumptions has been poor and the need to measure adult male survival as a parameter has increased. Presently, a number of population models in Colorado suggest that natural adult male survival may be lower than adult female survival, yet empirical data is lacking to verify these suppositions. Despite this apparent lack of information, survival of adult male mule deer and adult male blacktail deer are not completely novel parameters of interest (Pac and White 2007, Bender et al. 2004b, Bleich et al. 2006, Bishop et al. 2005a, McCorquodale 1999). However, rates of adult male survival reported in the literature are often linked to unique management situations such as variation in harvest structure (Pac and White 2007, Bender et al. 2004b), urban settings (Miller et al. 2008, Bender et al. 2004a) or disease management scenarios (Conner and Miller 2004, Miller et al. 2008). Similarly, most of these studies have been constrained by relatively small sample sizes and were of short duration, making the estimation of the process variation of adult male survival unreliable. However, available data suggest that adult male mule deer survival tends to be lower than adult female survival when differences occur, further emphasizing the need to rigorously evaluate adult male survival rates. Bishop et al. (2005a) observed lower natural survival rates of adult males than adult females in southwest Idaho: differences were most apparent during winter in 2 of 3 study areas. Pac and White (2007) found that natural survival rates of yearling males in Montana were lower than the average adult female survival rate documented by Unsworth et al. (1999). Finally, Miller et al. (2008) found that adult male survival rates were lower than adult female survival rates in Colorado in response to chronic wasting disease (CWD). In particular to the population modeling interests of Colorado outside the CWD endemic area, the work conducted by Pac and White (2007) has had the greatest utility. This work focused on the survival of males under differing management objectives and showed a shift in cause-specific mortality of males in areas where harvest was more restricted. It is currently unknown if survival rates would be similar between Montana and Colorado. Similarly, the likelihood of observing shifts in mortality sources is unknown. It has been demonstrated that adult female deer herds in Colorado tend to be habitat limited (Bishop et al. 2009, Bartmann et al. 1992), but the trade-off between harvest, habitat and survival in adult male mule deer has not been explored. A different, but not unrelated need in Colorado pertains to the harvest management of adult male mule deer. As discussed above, a large shift in mule deer herd size and structure occurred as a result of the change in harvest management. Overall, this shift has been viewed as positive by both CPW as well as the public. However, CPW still maintains the responsibility of optimally managing the deer of Colorado and maximizing hunting opportunity under this new set of constraints. To date, CPW has had limited biological information and data to guide harvest management decisions. In particular for this issue, as DAUs reach and surpass their adult male: adult female ratio objectives, CPW typically responds by increasing the number of available hunting licenses. In situations where herds are continually lower than DAU objectives, available hunting licenses are reduced. What remains unknown about survival of adult male deer is at what level natural survival is reduced due to intraspecific competition (i.e., increased density of adult male deer). If, or when deer herds exceed the adult male: adult female objectives for DAUs, it is often assumed that the surplus of male deer will remain in the population into perpetuity. However, this assumption is based on the premise that compensatory mortality does not occur. Similarly, it assumes that annual variation in survival is negligible. However, this is biologically not realistic. It is very likely that herds with large post-hunt populations of adult males experience higher levels of 108 �mortality. Under this scenario, harvest has not been optimized and more hunters could have been afforded the opportunity to hunt with no effect on post hunting season ratios of adult males to adult females. The most effective way to learn about the mortality process is via manipulative experimentation, but to date this topic has not been deemed a high enough priority to pursue. B. Objectives Our study objective is two-fold. First, we wish to assess annual survival of adult male mule deer. We wish to establish baseline survival and variance estimates for different age classes of deer. Second, we wish to manipulate hunting license allocation within the Game Management Units (GMUs) of D-9 such that adult male: adult female ratios become measurably different between the northern and southern halves of the DAU. Accordingly, we wish to measure and correlate changes in natural survival of adult male deer with this management action. Similarly, as part of this second objective, we will determine if changes in the age structure of harvested animals occur as the sex ratio and age structure of the hunted population changes. While not a direct objective of the study, we will also be able to learn if increasing adult male: adult female ratios causes an increase in the emigration rate of animals from populations composed of a greater proportion of adult male deer. C. Expected Results or Benefits Data and information generated from this study will have immediate use to terrestrial biologists and wildlife managers across the state of Colorado. Survival estimates of adult male deer will be immediately incorporated in the annual population modeling process. As measurements are repeated over years, estimates of process variation will be generated, allowing a refinement of how adult male survival is incorporated into the modeling process. From a general ecology perspective, we will measure the direct and indirect effects of a concerted management action on the male component of the deer population. We expect to detect differences in the harvest rates of radio-collared deer under different hunter/license allocation strategies. We also expect to detect differences in the harvest rate of radio collared deer based on age and antler morphometry. Similarly, we expect to detect a difference in natural survival/mortality rates of deer under differing levels of harvest. Ultimately this will provide information about the additive/compensatory relationship of adult male deer, adult female deer and mule deer fawn survival in Colorado. This information will allow us to directly inform tradeoff decisions between hunting opportunity and hunter desires for various quality standards. Additionally, these data will allow us to identify thresholds where further license restrictions would fail to result in more adult males in the population and fail to increase the mean age or antler structure of males harvested. D. Approach 1. Radio Collar Development Radio collars deployed as part of this project will be permanent (i.e., they will not be fitted with any sort of release mechanism). However, utilizing traditional radio collars with a fixed diameter is not ideal due to the seasonal variation in the size of adult male mule deer necks; as adult male deer enter the breeding period, neck swelling occurs. Researchers have historically addressed this issue with several different approaches. The use of loosely-fitted, fixed diameter radio collars has occurred on several hundred white-tailed deer in Texas with no known incidence of mortality or injury (K. VerCauteren – personal communication). It is unknown if a similar result could be expected for mule deer in Colorado. Researchers in Montana used an expandable radio collar that was made of tubular aircraft grade bungee material to measure survival of 136 adult male mule deer (Pac and White 2007, D. Pac – personal communication). When this research was conducted, expandable radio collars were not commercially 109 �available, so the expansion design was developed and installed on a traditional VHF collar that was produced by Telonics, Inc (Mesa, AZ, USA). This radio collar design alleviated concerns over neck constriction during the breeding period and it sufficiently contracted as neck swelling reduced after the breeding period. However, in a few instances (1%-2% of radio collared population) it was documented that deer were able to get a front hoof/leg between the collar and neck (D. Pac – personal communication). On these occasions, deer were either recaptured or euthanized if recapture was not possible. Researchers in Idaho as well as Colorado used a different expandable collar design, fitted with an expansion device that was made of flat elastic encased in Cordura™ to measure survival of 70 (Idaho) and ~100 (Colorado) adult male mule deer, respectively (Bishop et al. 2005a, Conner and Miller 2004). This collar was also made by Telonics, Inc. This design also alleviated constriction around the neck as deer entered the breeding period, but the contraction properties of the elastic were such that as neck diameter reduced as deer exited the breeding period the collar did not adequately contract to the prebreeding period diameter. This was not ideal as loosely fitting collars had a propensity to slide on the necks of animals and to cause hair loss. Additionally, researchers had concerns over the potential for deer to get hooves caught between the collars and their necks. This event occurred on one occasion with a fawn during the Idaho study (subsequently resulting in the animal’s mortality) and on two occasions in Colorado (both animals were recaptured and collars were removed during the Colorado study). To address the issue of expansion collars failing to contract back to pre-breeding period diameters, a third generation expansion collar was developed by CPW (M. Sirochman – personal communication). This new design incorporated nylon sleeved springs as the expansion device. As was the case in Montana, the spring based expansion collar adequately expanded and contracted through the initial breeding periods. However, on a few occasions the springs in these collars did eventually expand beyond their critical limit and ultimately failed to contract after having been deployed. On these occasions, it appeared that springs had snagged on external features, thereby reducing the integrity of the spring itself. Outside of these external factors, resilience of the spring appeared to be sound. The occurrence of deer getting their hooves caught between the collars and their necks was also documented in this study, but due to the tractability of animals, all were recaptured and radio collars were safely removed (M. Sirochman – personal communication). One additional downfall of the spring based expansion collar was that irritation caused by the pressure of the springs on the dorsal portion of the neck was documented in a few cases. While the irritation did not appear to jeopardize the health of the animal, it was undesirable. For our study, what can be considered a fourth generation expandable radio collar has been designed in collaboration with Advanced Telemetry Systems, Inc. (Isanti, MN, USA) (Figs. 3a and 3b). This newly designed collar closely resembles the earlier generation collars that incorporated flat elastic material. The elastic based expansion collar had fairly high success because only on a single occasion was it documented that a deer had its hoof caught between the collar and its neck. The primary weakness of this design was that the contraction properties of the elastic expansion material were inadequate. This new design incorporates a more robust, high quality, flat bungee material that is sheathed between traditional nylon belting material on the outside and nylon webbing on the inside (Fig. 3b). Due to the sheathing design, only a small portion of the bungee material is exposed, reflecting the desired qualities of the elastic based expansion collar and retaining the reduced potential for deer to get hooves caught in the collar. The higher quality bungee is expected to maintain contraction properties far longer than elastic and thus the potential for loose fitting collars during the later years of the study is reduced, minimizing the opportunity for hair breakage. This new collar design was scrutinized by the researchers who represent the bulk of knowledge on the subject of radio collaring adult male mule deer (D. Pac - retired, MT Fish, Wildlife and Parks; C. Bishop, M. Miller, M. Sirochman and L. Wolfe, CPW). The only additional concern pertained to the orientation and potential wear/irritation of the collar on the dorsal portion of deer necks during the breeding period. However, due to the width of the bungee material, it is expected to be less than that of the spring based expansion design. Concern over the orientation of the collar will be addressed by testing the collar design on a captive animal at CPW wildlife health research facility. 110 �Figure 3a. Figure 3b. Figures 3a and 3b. The newly designed, expandable, VHF radio-collar that will be utilized on adult male mule deer during this study. Collars were designed to meet CPW specifications by Advanced Telemetry Systems, INC. (Isanti, MN, USA). The blue banding material seen in figures 3a and 3b is nylon coated bungee that will allow expansion and contraction, as needed, during the breeding period. To allow 111 �maximal expansion, but to help prevent the opportunity for deer to get hooves and legs caught between the neck and collar, the bungee material is sleeved in nylon webbing (red material visible in figure 3a). 2. Capture Capture of adult male deer for this project will be conducted via helicopter net-gunning (Webb et al. 2008, Potvin and Breton 1988, White and Bartmann 1994, Barrett et al. 1982). All captures will occur after the completion of the 4th rifle hunting season, eliminating potential conflicts between capture efforts and hunting. Typically capture will occur between mid-December and mid-January. Exact timing of capture each year will be dependent on availability of the helicopter net-gunning crew. Due to the need to have survival estimates linked to animals of known age, all animals will be handled by CPW personnel for aging purposes. Depending on situation, captured animals will be handled in one of two ways. When feasible, captured deer will be ferried to a processing area staffed by CPW researchers/biologists who are qualified to age animals according to tooth wear (Severinghaus 1949, Robinette et al. 1957, Hamlin et al. 2000). Deer will subsequently be returned to the capture site for release. In situations when capture locations are too far from processing areas to efficiently ferry animals, CPW researchers/biologists will be ferried to the general area in which capture is occurring and subsequently be ferried the short distance to each capture location to process animals at that site. Regardless of situation, it is possible that a single person will be responsible for collaring, aging and releasing animals. As such, prior to release, all animals will have their antlers removed via handsaw to minimize the potential risk of injury as the animal is released. The removal of antlers from animals at this time of year should have no negative impact on survival as all captures will occur post-rut. Similarly, all legal harvest of animals will have occurred and negative response of hunters should not occur. The only exception to the antler removal process will be if post-hunt sex/age class survey flights have not yet occurred and if the captured animal is located near a survey quadrat. If a deer is captured near a survey quadrat, prior to deer classification flights having been conducted and it is still deemed necessary to remove antlers, these deer will be temporarily marked with livestock marking paint on the back and neck. Marking deer in such a manner will allow biologists to accurately classify those individual deer as adult males, thereby removing any potential bias that may stem from capturing deer prior to classification flights. Whenever possible, capture will be conducted after classification flights to alleviate this problem. All deer will be fitted with expandable radio collars (discussed above). All radio collars will be equipped with mortality sensors which will double in pulse rate after remaining motionless for 4 hours. The desired sample size for each year of this study will be a total of 220 adult ( ≥ 1.5 years old) male deer. One hundred deer will be captured and radio collared during the first year of the study as a pilot assessment of the radio collar design and to test underlying assumptions about deer movement (discussed below). During the second year of the study, 120 additional deer, as well replacements for any deer that die during the first year will be captured and radio collared. Thus, not until the second winter of the study will the full sample size be achieved. For every year thereafter, only enough deer to maintain the 220 animal sample size will be captured. The 220 deer sample will be distributed such that 110 of the radio collared deer are located in the northern half and 110 are located in the southern half of the DAU. 3. Survival/Location Monitoring The primary objective of this study is to generate annual natural survival estimates and harvest rates for adult male deer. While most mortality is expected to occur via rifle harvest between October and November, the bulk of natural mortality is expected to occur between December and May of each year. In order to minimize bias of survival estimates during these periods, we will attempt to monitor the live/dead status of each animal 3-4 times per week. Each year, prior to the start of the archery hunting season, all deer will be located to asses in which half (northern versus southern) of the DAU each animal is located. A similar set of locations will be collected after the 4th rifle hunting season. Between each hunting season, a live/dead flight will be conducted to determine if any animals have disappeared, and subsequently assumed to have been harvested, without having been reported to CPW. Once all hunting 112 �seasons have been completed, we will revert to a weekly flight schedule to assess live/dead status of all animals. A field technician will check live/dead status 2-3 times for each animal between flights. All animals will be located 1 additional time during the winter to confirm that animals have not left the DAU and to determine if any animals have switched between the northern and southern halves of the DAU. Based on historical location data for adult female and fawn mule deer, approximately 10% of deer are expected to cross between northern and southern halves of the DAU (K. Oldham – unpublished data). Any animals switching between halves of the DAU will be censored from the optimal harvest management portion of the analysis. During periods when survival is expected to be higher and less dynamic (June through September), the level of effort of to determine live/dead status will be reduced. Flights to determine live/dead status will occur approximately every 14 days and efforts to hear animals from the ground will occur as time allows. A single location will be collected for each animal after it has arrived on summer range (between late-June and late-July). While not ideal, weekly survival estimates for summer months can be computed from bi-monthly estimates via the delta method (Powell 2007). This approach to survival monitoring will allow us to minimize bias but also minimize costs associated with aircraft and temporary personnel. 4. Harvest Management Experiment We will implement a management experiment to evaluate adult male survival rates under different harvest management strategies. Hunting management in Colorado is partitioned into DAUs. The boundaries of DAUs are intended to reflect the biological boundaries of deer such that deer movement between DAUs is non-existent or infrequent enough to be biologically insignificant. Within DAUs are GMUs. GMU boundaries tend to be highly permeable to deer, but serve to partition DAUs for human oriented management purposes such as survey work and hunter distribution. Typically all GMUs within a DAU have the same management objective. However, this study will deviate from this trend by establishing two different harvest objectives within a single DAU. This approach will help ensure that all deer in the study will experience similar environmental conditions and limiting factors except for different harvest objectives. Thus, any survival differences we observe are likely to be a result of differential harvest as opposed to some other factor. This study will take place in Middle Park, Colorado (see below for rationale). Under the current management structure, Middle Park falls within DAU D-9. Within D-9 are 6 GMUs (27,181, 18, 37, 371, and 28; Fig. 4). D-9 is managed for an adult male: adult female ratio of 35 adult males per 100 adult females. As part of this study, the management of D-9 will be temporarily altered such that it will be viewed as two separate populations (one population will be composed of the northern 3 GMUs (27, 181 and 18) and the other population will be the southern 3 GMUs (37, 371 and 28)). During the 4th-7th years of this study we will redistribute hunters within the DAU via hunting license allocation. During the 1st-3rd years of the study we will monitor survival across the DAU to provide baseline data (Fig. 5). The objective behind the redistribution of hunters will be to increase adult male: adult female ratios in one half of D-9 and to decrease adult male: adult female ratios in the other half of D-9. The current DAU objective of 35 adult males: 100 adult females will not change, but one half of the DAU will be managed for 25 adult males: 100 adult females and the other half will be managed for 45 adult males: 100 adult females. The determination of which half of D-9 will experience higher harvest and which half will experience lower harvest has not yet been made. This decision will ultimately be made by Area 9 and Northwest Region personnel. In the event that there are no overwhelming management concerns about this selection process, the selection will be random. 5. Age at Harvest To help evaluate the effects of a changing sex ratio on hunter harvest, we will attempt to acquire an age for animals harvested in D-9 for years 2-7 of the study. Ages will be estimated via the cementum 113 �aging process of incisors (Hamlin et al. 2000). To acquire teeth for aging purposes, all hunters who have licenses to hunt in any GMU in D-9 will be contacted prior to the archery season via mail. Each hunter will be provided with a sampling kit, a pre-posted return envelope and detailed directions on how to extract teeth for aging purposes. These data will help inform terrestrial biologists and wildlife managers if changes in the age of animals harvested occur as populations shift up or down in age structure as sex ratios are increased or decreased. Figure 4. Data Analysis Unit 9 (D-9) encompasses the Middle Park area of central Colorado. D-9 includes 6 Game Management Units (27, 181 and 18 on the northern half and 37, 371 and 28 on the southern). Current management sex ratio management objectives for D-9 are consistent across GMUs with an overall post hunt objective of 35 adult males per 100 adult females. 6. Data Analysis This study can be structured as a multi-state study (Fig. 6). We are primarily interested in deer that exist in three different states: 1) deer that survive, 2) deer that are harvested, and 3) deer that die due to non-harvest causes. While most multi-state studies include survival, detection and transition probabilities for different states, this study is purely focused on the transition probability of deer that transition from the living state to either one of the two non-living states, or back into the living state. Due to the relatively safe assumptions that deer will not leave study area, that use of radio-telemetry is essentially always detectable and the fates of deer can be readily identified, detection probabilities can be fixed at 1.0 and survival can be artificially set at 1.0. Thus, the transition probabilities between states 114 �becomes a surrogate for survival, thereby allowing us to distinguish and easily measure differences between causes of mortality. Figure 5. For this study, the northern and southern halves of D-9 will managed under different harvest management objectives during years 3-7. One half will be managed under less restrictive objectives, with a post hunt sex ratio objective of 25 adult males per 100 adult females. The other half of the DAU will be managed under more restrictive conditions with a post hunt sex ratio objective of 45 adult males per 100 adult females. For this study, we will have numerous response variables of interest. The basic analysis of this study will follow a before-after-control-impact (BACI) design (Green 1979, Hurlbert 1984, Underwood 1994 and Conner et al. 2007) (Fig. 7). Overall, survival of adult male deer will be analyzed using known-fate models in program MARK (White and Burnham 1999). Survival will be modeled using age of deer, GMU/DAU, year and trophy score. For the purposes of this study, we are primarily interested in weekly survival rates throughout the year. Cause specific mortality will be analyzed under a multi-state modeling framework in which detection probabilities will be fixed to 1.0 based on the known-fate properties of the data for the BACI analysis. We will use mixed models to assess the impact of manipulating harvest management. For the mixed model analyses, we will use adult male: adult female ratio as the response variable for one analysis and hunter success as the response variable in a second analysis. 115 �Survive 4'AB=Natural Mortality Rate 4' AB=Harvest Rate Non-Harvest Mortality Harvest Figure 6. Assessment of survival and mortality causes can be conceptualized as a multi-state analysis with transition rates from the surviving state to the harvest state or the non-harvest related mortality state being the parameters of interest. In this case, transitions represented by black arrows can be estimated via radio collared deer. Transitions represented by gray arrows are not biologically feasible. Deer that are harvested cannot return to the survival state, nor can they enter the non-harvest related mortality. Similarly, deer dying to non-harvest related causes cannot simultaneously survive or be harvested. Under this multi-state framework, all other parameters of interest will be fixed at 1.0. 7. Sample Size Sample size estimates for this study are based on the desire to detect a difference in the nonharvest mortality rates of deer under different harvest management regimes. Best estimates of harvest mortality rates, natural survival rates and the associated variance of each were based on the work published by Pac and White (2007). For our power calculations, baseline/ control harvest rates were set at 0.21 and the associated natural survival was 0.72. For high harvest areas, we set harvest at 0.37 and the associated natural survival at 0.77. For low harvest areas, we set harvest at 0.06 and subsequent natural survival at 0.67. Thus, our power calculation was set up to detect a 10% difference in natural survival under different harvest management regimes. For sample size estimation we chose to fix the number of radio collared deer entering the study each winter at ~200 (~100 animals per area) and then used simulation models to determine the number of releases (i.e., number of winters in the study) that would be needed to detect our desired effect size. Simulations were set up to test the differences in natural survival by comparing survival rates as beta offsets from the expected survival rates under normal conditions (Fig. 8). 116 �Half Sample Full Sampl e Full Sampl e Full Sampl e Full Sampl e Full Sample Full Sample Half Sampl e Full Sampl e Full Sampl e Full Sampl e Full Sampl e Full Sample Full Sample if1:., ., " -~ct 0, .:i: ti "~ ~ I <( ~ 0 ...J Control Period (Years 1-3) Impact Period (Years 4-7) Figure 7. The Before-After-Control-Impact design for this study is based on 6 years of a full sample of deer, with an initial build up year to help offset logistic and financial constraints associated with capturing 220 deer for the full sample (110/area). The control period will experience no change to harvest management, whereas the impact period with experience a concerted effort to redistribute hunters across the DAU to impact post hunt sex ratios. Based on these initial conditions, it appears that 6 winters with a full sample of deer will be needed to reliable detect a 10% difference in natural survival rates. Due to the estimated censoring of 10% of the radio-collared deer, due to movement between the northern and southern halves of the study area, we will inflate the total sample size from 100 animals per area to 110 animals for years 2-7 of the study. Duration of the study was determined by comparison of 95% confidence intervals surrounding the expected difference in natural survival (Fig. 9). Confidence intervals that included 0 were indicative of not enough statistical power to detect a difference. While 5 years may adequately meet the needs of the study, our results indicate that 6 years with a full sample of deer will be substantially more robust. E. Location This work will be conducted in Middle Park, Colorado. Middle Park was selected for this work based on several criteria. First, Middle Park is one of CPW’s mule deer winter survival monitoring areas and has ongoing monitoring of the survival of adult females and fawns. Adding estimates of adult male survival in this area will allow us to compute correlation and covariance between the different sexes through time. Similarly, in the event that changing adult male: adult female ratios affects survival of adult females or fawns, we will have all relevant sex and age classes marked and should be able to detect any changes. Additionally, geological and topographical structure of Middle Park is conducive to splitting the DAU into halves such that few deer migrate from one half to the other during the annual movement cycle. Existing data indicate that 10%-15% of deer cross between halves. As such, the number of deer needing to be censored from the management experiment portion of this study should be minimized. 117 �~ > C: --------------------------------------------------------------- Bo+B2 :::, (.I) ~ t - - - - - - - - - - - - - - - - - - - - - - - - Bo I,,. :::, ~ z Time Figure 8. Non-harvest survival will be analyzed using the log scale comparison of beta estimates. The control period will be the baseline survival estimate (ß0) to which natural survival under differing harvest management efforts will be compared (solid black line). Non-harvest related survival in restrictive harvest units (ß1) is expected to be lower than estimates for both the control phase and more liberal harvest units (long dashed line). Non-harvest related survival in liberal harvest units (ß2) is expected to be higher than estimates for both the control phase and more conservative harvest units (short dashed line). 0 "iii > -~ -0.2 :::, V'I ...:::, "iii - -0.4 ·=u -0.6 ra z Cl) ---------- - -------- · C: ... Cl) Cl) := -0.8 0 -1 5 4 6 Nmnher of Years Figure 9. Power calculation set up to determine the number of years necessary to detect a 10% difference in non-harvest related moratality rates under differing harvest management regimes. Solid lines depict 95% confidence intervals. Adequate power is achieved once 95% confidence interval estimates do not include 0. While statistical power may be adequate after 5 years, addition of a 6th year will make the study more robust to violations or deviations from the underlying parameter estimates used to structure the analysis. 118 �From a management perspective, D-9 is currently managed for 35 adult males per 100 adult females. The management experiment portion of this study will allow the DAU to be split with half of the DAU being managed for 25 adult males per 100 females and the other half being managed for 45 adult males per 100 females. These ratios are largely representative of objectives throughout the state (i.e., management is not purely trophy or opportunity driven) and should allow adequate inference to be drawn. By not altering the overall DAU population objectives, implementing this research will not require that the D-9 management plan be rewritten. Additionally, Middle Park has historically been prone to periodic, harsh winters which are fundamental to getting reasonable estimates of process variation. Lastly, Middle Park has also been the site of numerous deer research and concerted management efforts over the past several decades. Knowledge and information from these past efforts have greatly facilitated the design of this study and historical data are readily available should refinement of study design or objectives become necessary. F. Schedule Of Work Activity Date Design and Purchase Expandable Radio Collars June 2010−Nov 2010 Purchase Field Supplies June 2010−Nov 2010 Capture ½ of initial sample of deer Dec 2010-Jan 2011 Monitor Survival and Movement of Deer Dec 2010−June 2017 Capture remaining sample of deer Dec 2011−Jan 2012 Capture deer to bring sample back to full size Dec 2012−Jan 2013 Implement change in hunter distribution within DAU Feb 2013-Feb 2016 Capture deer to bring sample back to full size Dec 2013−Jan 2014 Capture deer to bring sample back to full size Dec 2014−Jan 2015 Capture deer to bring sample back to full size Dec 2015−Jan 2016 G. Estimated Costs Category Item or Position FY 10-11 Personnel Eric Bergman 0.25 PFTE Chad Bishop 0.25 PFTE Kirk Oldham 0.05 PFTE Lyle Sidener 0.00 PFTE Field Equipment and Capture $100,000 Operating 119 �H. Related Federal Projects Our research will be conducted on federal (i.e., BLM, USFS) and state lands. The study does not involve formal collaboration with any federal agencies, nor does the work duplicate any ongoing federal projects. I. Literature Cited Barrett, M.W., J.W. Nolan, and L.D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108-114. Bartmann, R.M., G.C. White, and L.H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monographs 121:1-39. Bender, L.C., J.C. Lewis and D.P. Anderson. 2004a. Population ecology of Columbian black-tailed deer in Urban Vancouver, Washington. Northwestern Naturalist 85:53-59. Bender,L.C., G.A. Schirato, R.D. Spencer, K.R. McAllister, and B.L. Murphie. 2004b. Survival, causespecific mortality, and harvesting of male black-tailed deer in Washington. Journal of Wildlife Management 68:870-878. Bergman, E.J., B.E. Watkins, C.J. Bishop, P.M. Lukacs, and M. Lloyd. 2010. Biological and socioeconomic effects of statewide limitation of deer licenses in Colorado. Journal of Wildlife Management. In Review. Bishop, C.J., J.W. Unsworth and E.O. Garton. 2005a. Mule deer survival among adjacent populations in southwest Idaho. Journal of Wildlife Management 69:311-321. Bishop, C.J., G.C. White, D.J. Freddy and B.E. Watkins. 2005b. Effect of limited antlered harvest on mule deer sex and age ratios. Wildlife Society Bulletin 33:662-668. Bishop, C.J., G.C. White, D.J. Freddy, B.E. Watkins, and T.R. Stephenson. 2009. Effect of enhanced nutrition on mule deer population rate of change. Wildlife Monographs 172:1-28. Bleich, V.C., B.M. Pierce, J.L. Jones and R.T. Bowyer. 2006. Variance in survival of young mule deer in the Sierra Nevada, California. California Fish and Game. 92:24-38. Conner, M.M., and M.W. Miller. 2004. Movement patterns and spatial epidemiology of a prion disease in mule deer population units. Ecological Applications 14:1870-1881. Conner, M.M., M.W. Miller, M.R. Ebinger and K.P. Burnham. 2007. A meta-BACI approach for evaluating management intervention on chronic wasting disease in mule deer. Ecological Applications 17:140-153. Green, R.H. 1979. Sampling design and statistical methods for environmental biologists. Wiley Interscience, Chichester, England. Hamlin, K.L., D.F. Pac, C.A. Sime, R.M. DeSimone, and G.L. Dusek. 2000. Evaluating the accuracy of ages obtained by two methods for Montana ungulates. Journal of Wildlife Management. 64:441449. Hurlbert, S.H. 1984. Pseudoreplication and the design of ecological field experiments. Ecological Monographs 54:187-211. Miller, M.W., H.M. Swanson, L.L. Wolfe, F.G. Quatarone, S.L. Huwer, C.H. Southwick and P.M. Lukacs. 2008. Lions and prions and deer demise. Plos One 3:1-7. McCorquodale, S.M. 1999. Movements, survival, and mortality of black-tailed deer in the Klickitat Basin of Washington. Journal of Wildlife Management 63:861-871. Pac, D.F., and G.C. White. 2007. Survival and cause-specific mortality of male mule deer under different hunting regulations in the Bridger Mountains, Montana. Journal of Wildlife Management 71:816-827. Potvin, F., and L. Breton. 1988. Use of a net gun for capturing white-tailed deer, Odocoileusvirginianus, on Anticosti Island, Quebec. Canadian Field Naturalist 102:697-700. Powell, L.A. 2007. Approximating variance of demographic parameters using the delta method: a reference for avian biologists. Condor 109:949-954. 120 �Robinette, W.L., J.S. Gashwiler, D.A. Jones, and H.S. Crane. 1957. Notes on tooth development and wear for Rocky Mountain mule deer. Journal of Wildlife Management 21:134-153. Severinghaus, C.W. 1949. Tooth development and wear as criteria of age in white-tailed deer. Journal of Wildlife Management 13:195-216. Underwood, A.J. 1994. On beyond BACI-sampling designs that might reliably detect environmental disturbances. Ecological Applications. 4:3-15. Unsworth, J.W., D.F. Pac, G.C. White, and R.M. Bartmann. 1999. Mule Deer Survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. Webb, S.L., J.S. Lewis, D.G. Wewitt, M.W. Hellickson, and F.C. Bryant. 2008. Assessing the helicopter and net gun as a capture technique for white-tailed deer. Journal of Wildlife Management 72:310-314. White, G.C. and R.M. Bartmann. 1994. Drop nets versus helicopter net guns for capturing mule deer fawns. Wildlife Society Bulletin 22:248-252. White, G.C. and K.P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46:120-139. White, G.C. and B.C. Lubow. 2002. Fitting population models to multiple sources of observed data. Journal of Wildlife Management 66:300-309. 121 �Colorado Parks and Wildlife July 2011 – June 2012 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Federal Aid Project No. Colorado 3430 3001 : : : : Parks and Wildlife Mammals Research Deer Conservation Assessment of survival and optimal harvest Strategies of adult male mule deer in Middle Park, Colorado W-185-R Period Covered: July 1, 2011 - June 30, 2012 Author: E.J. Bergman; Project Cooperators; C.J. Bishop, K. Oldham, and L. Sidener Personnel: G. Abram, G. Birch, J. Broderick, M. Crosby, B. Davies, T. Elm, D. Gillham, K. Holinka, A. Holland P. Lukacs, B. Manly, S. Murdoch, S. Schwab, S. Shepherd Colorado Parks and Wildlife R. Swisher, S. Swisher, T. McKendrick Quicksilver Air All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We continued field work on a study designed to assess the survival and optimal harvest strategies of adult male mule deer in Middle Park, Colorado. Two years of baseline survival data for adult (≥ 1 yr. old) male deer were collected prior to termination of this project. During December (2011), 49 additional adult (≥ 1.5 years old) male deer were captured and radio collared, returning the radio collared sample to 100 animals at the start of the 2012 calendar year. The natural, annual survival rate for all deer for the period ending on the 14th of December (2011) was estimated at 0.820 (SE = 0.0394). From the 15th of December (2011) through the 31st of July (2012), survival was estimated at 0.919 (SE = 0.0275). Due to consternation expressed by a very select group of trophy hunters, the project was terminated in July of 2012. 94 �WILDLIFE RESEARCH REPORT ASSESSMENT OF SURVIVAL AND OPTIMAL HARVEST STRATEGIES OF ADULT MALE MULE DEER IN MIDDLE PARK, COLORADO ERIC J. BERGMAN P.N. OBJECTIVES SEGMENT OBJECTIVES 1. Continue field work in the form of capturing and radio collaring animals. 2. Collect survival data on radio collared deer and provide preliminary survival estimates for adult male mule deer. INTRODUCTION Historically, management of big game species has focused on the performance of adult females and the young of the year segments of the population. In the case of mule deer, this has been further refined to the aspects of annual (for adult females) and overwinter (for young of the year) survival. The performance of the male component of populations was deemed less important because it takes few males to provide adequate breeding coverage for the population, and historic harvest management objectives were set to maximize hunting opportunities. As long as sufficient numbers of males were available to breed females there was no desire to restrict hunting opportunity. However, during the past 10-15 years, the management of big game populations, and mule deer populations in particular, has shifted from the objective of providing maximal opportunity towards providing higher quality opportunities (Bishop et al. 2005b, Bergman et al. 2010). High quality opportunities are typically defined by hunters as a combination of the chance to see a greater number of male deer during the hunt, increased potential to harvest an older age class animal (i.e., an animal with more developed antler morphometry), but also reduced interaction and competition with other hunters. In response to this shift in hunter desires and concerns over declining mule deer numbers, the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) implemented a statewide limitation in deer hunting in 1999. This statewide limitation gave CPW the ability to reduce total hunter numbers and to control the distribution of hunters throughout the state. Since 1999 Colorado’s deer herds have become composed of a greater number of males, yet little biological data on them exist. Also stemming from this change in harvest management was a new responsibility for Colorado’s terrestrial biologists and wildlife managers. Prior to 1999, licenses were sold over-the-counter and were not limited in number (i.e., any hunter who wished to purchase one was able to do so), and the decision of how many licenses to make available did not need to be considered. Since 1999, CPW has the added responsibility of deciding how many licenses should be allocated in each Data Analysis Unit (DAU). This decision must reflect a balance between meeting DAU population performance objectives, and maximizing hunter opportunity. Big game populations in Colorado are currently modeled using multiple sources of biological data (White and Lubow 2002). Model inputs include harvest estimates, young recruitment to December, and measured rates of survival of adult females and fawns. Also, the ratio of adult males to adult females is estimated and used to align models by minimizing the difference between observed and modeled values. Only rarely have the survival rates of adult males been measured. This gap in knowledge has historically been viewed as trivial and adult male survival rates have been assumed to be similar to the rates of females. Similarly, it has been assumed that natural survival rates (i.e., post hunt survival) of males do not geographically vary. However, model performance under these assumptions has been poor and the need to measure adult male survival as a parameter has increased. Presently, a number of 95 �population models in Colorado suggest that natural adult male survival may be lower than adult female survival, yet empirical data is lacking to verify these suppositions. Despite this apparent lack of information, survival of adult male mule deer and adult male blacktail deer are not completely novel parameters of interest (Pac and White 2007, Bender et al. 2004, Bleich et al. 2006, Bishop et al. 2005a, McCorquodale 1999). These studies also suggest that adult male mule deer survival tends to be lower than adult female survival when differences occur, further emphasizing the need to rigorously evaluate adult male survival rates. Bishop et al. (2005a) observed lower natural survival rates of adult males than adult females in southwest Idaho: differences were most apparent during winter in 2 of 3 study areas. Pac and White (2007) found that natural survival rates of yearling males in Montana were lower than the average adult female survival rate documented by Unsworth et al. (1999). Finally, Miller et al. (2008) found that adult male survival rates were lower than adult female survival rates in Colorado in response to chronic wasting disease (CWD). In particular to the population modeling interests of Colorado outside the CWD endemic area, the work conducted by Pac and White (2007) has had the greatest utility. This work focused on the survival of males under differing management scenarios and showed a shift in cause-specific mortality of males in areas where harvest was more restricted. It is currently unknown if survival rates would be similar between Montana and Colorado. Similarly, the likelihood of observing shifts in mortality sources is unknown. It has been demonstrated that adult female deer herds in Colorado tend to be habitat limited (Bishop et al. 2009, Bartmann et al. 1992), but the trade-off between harvest, habitat and survival in adult male mule deer has not been explored. An additional need in Colorado pertains to the harvest management of adult male mule deer. As discussed above, a large shift in mule deer herd size and structure occurred as a result of changes in harvest management. Overall, this shift has been viewed as positive by both CPW as well as the public. However, CPW maintains the responsibility of optimally managing the deer of Colorado and maximizing hunting opportunity under this new set of constraints. To date, CPW has had limited biological information and data to guide harvest management decisions. In particular for this issue, as Data Analysis Units (DAUs) reach and surpass their adult male: adult female ratio objectives, CPW typically responds by increasing the number of available hunting licenses. In situations where herds are continually lower than DAU objectives, available hunting licenses are reduced. What remains unknown about survival of adult male deer is at what level natural survival is reduced due to intraspecific competition (i.e., increased density of adult male deer). If, or when deer herds exceed the adult male: adult female objectives for DAUs, it is often assumed that the surplus of male deer remain in the population into perpetuity. However, this assumption is based on the premise that compensatory mortality does not occur. Similarly, it assumes that annual variation in survival is negligible. However, these assumptions are not biologically realistic. It is possible that herds with large post-hunt populations of adult males experience higher levels of non-harvest mortality. Under this scenario, harvest has not been optimized and more hunters could have been afforded the opportunity to hunt with no effect on post hunting season ratios of adult males to adult females. The most effective way to learn about the mortality process is via manipulative experimentation, but to date this topic has not been deemed a high enough priority to pursue. STUDY AREA This study took place in Middle Park, Colorado, within DAU D-9. Within D-9 are 6 Game Management Units (27,181, 18, 37, 371, and 28; Fig. 1). METHODS Capture of adult male deer was conducted in January and December of 2011. Capture was conducted via helicopter net-gunning (Webb et al. 2008, Potvin and Breton 1988, White and Bartmann 96 �1994, Barrett et al. 1982). All captures occurred after the completion of the 4th rifle hunting season, eliminating conflicts between capture efforts and hunting. All deer were fitted with expandable radio collars. All radio collars were equipped with mortality sensors that doubled in pulse rate after remaining motionless for 4 hours. Between the time of capture and mid-June, we used ground-based monitoring to determine the live/dead status of deer 3-4 times per week. Additionally, every 5-10 days we conducted a telemetry flight to detect any animals that hadn’t been heard from the ground during the preceding week. A general location was collected for each radio marked deer in early-March to determine if it had departed the GMU in which it had originally been captured. From mid-June through remainder of the summer, deer were monitored from the ground weekly and from the air once per month. When detected, all mortalities were investigated as quickly as possible to determine cause of death and to get an accurate estimate of the date of death. RESULTS AND DISCUSSION In December (2011), 49 deer were captured and radio collared. On one occasion, an animal suffered a fractured leg as part of the capture process and was subsequently euthanized at the capture site via gunshot to the head. No other capture related injuries or mortalities occurred, although one animal was killed via vehicular collision 2 days post capture. This animal was subsequently censored from survival analysis due to uncertainty if stress related to the capture process had influenced its fate. Survival of adult male deer between January 2011 and the 14th of December (2011) was estimated to be 0.820 (SE = 0.0394). There was no apparent difference between the north half the D-9 and the south half of D-9 (Table 1), validating the assumptions of the original study plan design. During the 2011 hunting seasons, a total of 31 radio collared bucks were harvested (Fig. 1). Due to mild winter conditions, survival from the 15th of December (2011) through the 31st of July (2012) was very high (0.909, SE = 0.0275). SUMMARY Project efforts were successful during the first two years of the study, although local resistance to the project remained high. Based on public meetings, the majority of hunters in the Middle Park area supported the project, although a very select group of trophy hunters remained opposed. Ultimately, CPW leadership determined that it “could not overcome the current opposition” and the study should not proceed. Thus, this research project has been terminated as of 7/12/2012. LITERATURE CITED Barrett, M.W., J.W. Nolan, and L.D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108-114. Bartmann, R.M., G.C. White, and L.H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monographs 121:1-39. Bender,L.C., G.A. Schirato, R.D. Spencer, K.R. McAllister, and B.L. Murphie. 2004. Survival, causespecific mortality, and harvesting of male black-tailed deer in Washington. Journal of Wildlife Management 68:870-878. Bergman, E.J., B.E. Watkins, C.J. Bishop, P.M. Lukacs, and M. Lloyd. 2010. Biological and socioeconomic effects of statewide limitation of deer licenses in Colorado. Journal of Wildlife Management. In Review. Bishop, C.J., J.W. Unsworth and E.O. Garton. 2005a. Mule deer survival among adjacent populations in southwest Idaho. Journal of Wildlife Management 69:311-321. Bishop, C.J., G.C. White, D.J. Freddy and B.E. Watkins. 2005b. Effect of limited antlered harvest on mule deer sex and age ratios. Wildlife Society Bulletin 33:662-668. 97 �Bishop, C.J., G.C. White, D.J. Freddy, B.E. Watkins, and T.R. Stephenson. 2009. Effect of enhanced nutrition on mule deer population rate of change. Wildlife Monographs 172:1-28. Bleich, V.C., B.M. Pierce, J.L. Jones and R.T. Bowyer. 2006. Variance in survival of young mule deer in the Sierra Nevada, California. California Fish and Game. 92:24-38. Miller, M.W., H.M. Swanson, L.L. Wolfe, F.G. Quatarone, S.L. Huwer, C.H. Southwick and P.M. Lukacs. 2008. Lions and prions and deer demise. Plos One 3:1-7. McCorquodale, S.M. 1999. Movements, survival, and mortality of black-tailed deer in the Klickitat Basin of Washington. Journal of Wildlife Management 63:861-871. Pac, D.F., and G.C. White. 2007. Survival and cause-specific mortality of male mule deer under different hunting regulations in the Bridger Mountains, Montana. Journal of Wildlife Management 71:816-827. Potvin, F., and L. Breton. 1988. Use of a net gun for capturing white-tailed deer, Odocoileusvirginianus, on Anticosti Island, Quebec. Canadian Field Naturalist 102:697-700. Unsworth, J.W., D.F. Pac, G.C. White, and R.M. Bartmann. 1999. Mule Deer Survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315-326. Webb, S.L., J.S. Lewis, D.G. Wewitt, M.W. Hellickson, and F.C. Bryant. 2008. Assessing the helicopter and net gun as a capture technique for white-tailed deer. Journal of Wildlife Management 72:310-314. White, G.C. and R.M. Bartmann. 1994. Drop nets versus helicopter net guns for capturing mule deer fawns. Wildlife Society Bulletin 22:248-252. White, G.C. and B.C. Lubow. 2002. Fitting population models to multiple sources of observed data. Journal of Wildlife Management 66:300-309. Prepared by Eric J. Bergman, Wildlife Researcher 98 �Table 1. Model results for known-fate survival models based on mule deer buck data collected in Middle Park, Colorado. Model comparison is made via Akaike’s Information Criterion corrected for small sample size (AICc). Of interest to the original study design, there was no strong evidence that there was a difference in survival between the northern and southern halves (Area) of the study area. Model ∆AICc AICc Weight Likelihood Parameters Ŝ (Constant) 0.00 0.69 1.00 1 Ŝ (Area) 1.56 0.31 0.46 2 Ŝ (Week) 49.77 0.00 0.00 48 Ŝ (Area + Week) 51.37 0.00 0.00 49 Ŝ (Area*Week) 131.16 0.00 0.00 96 Figure 1. Fate and associated cause of death of 100 mule deer bucks between January 2011 and December 2011 in Middle Park, Colorado. □ Poachin g (n =2) ■ Wounding Loss (11 =1) Coyote (n =2) Survived (n = 53) □ Mt.Lion (n =l) D Unk. Predation (n =3) D Di sease/S tarvat ion (n =2) D Road/Tra in Kill (n =2) D Unknow n (n =3) 99 � https://cpw.cvlcollections.org/files/original/d771b319e26a9695ad1e3481df65876c.pdf c8f4617fa5e0ed7a51399e9f7d22fceb PDF Text Text Colorado Division of Parks and Wildlife July 1, 2010 − June 30, 2011 PROGRAM FINAL REPORT DEER CONSERVATION RESEARCH FOR 5-YEAR FEDERAL AID GRANT W-185-R JULY 2006 – JUNE 2011 State of Colorado Cost Center 3430 Work Package 3001 Federal Aid Project W-185-R : Division of Parks and Wildlife : Mammals Research : Deer Conservation Research : Period Covered: July 1, 2006 – June 30, 2011 Authors: Chad J. Bishop, Charles R. Anderson, Jr., and Eric J. Bergman Principal Investigators: M. W. Alldredge, C. R. Anderson, E. J. Bergman, C. J. Bishop, D. J. Freddy, P. M. Lukacs, D. P. Walsh, and B. E. Watkins. Colorado Division of Wildlife; P. F. Doherty and G. C. White, Colorado State University ABSTRACT This report highlights the accomplishments of mule deer research and associated activities conducted by the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) with the funding support of Federal Aid Grant W-185-R during the 5-year grant segment, July 2006-June 2011. Two major multi-year research projects addressing mule deer population limiting factors and habitat enhancements were completed and reported upon during this segment. Two other major multi-year research projects were designed and implemented during this period. One project is comprehensively addressing approaches to mitigate the impacts of natural gas development on mule deer. The other project is evaluating survival rates and harvest management of adult male mule deer. Several other smaller research projects were designed and implemented, addressing mule deer-elk-cougar interactions and development of techniques for marking and monitoring mule deer. Additionally, funding provided scientific and technical expertise for mule deer population monitoring and analysis. Research experiments provided strong evidence that habitat nutritional quality had a greater impact on net productivity of mule deer than did existing levels of coyote, cougar, and black bear predation and that mechanical habitat treatments in senescent pinyon-juniper winter ranges were an effective strategy for increasing deer survival by increasing the amount of higher-quality forage. These research results provided wildlife managers support and direction for managing pinyon-juniper habitat across western Colorado. These research results also framed the experimental design for evaluating approaches to mitigate impacts of natural gas development on deer. Specifically, a large field experiment was initiated in northwest Colorado to evaluate effectiveness of habitat treatments in late-seral pinyon juniper and mountain shrub habitats that are experiencing high-intensity and low-intensity energy development. From activities supported by this Grant during this segment, principal investigators published 13 peer-reviewed scientific articles for prominent wildlife research journals, provided 21 annual CPW Wildlife Research Reports summarizing yearly progress of projects, provided 34 presentations at professional meetings or workshops, and initiated 2 graduate student projects. The cumulative impact of 37 �this programmatic effort provides Colorado the basis to progress and proactively sustain the mule deer resource in an increasingly complex landscape. The relative success of mule deer management in Colorado reflects the positive synergy between the terrestrial research and management sections in sharing expertise, financial resources, staffing, and common goals. 38 �PROGRAM FINAL REPORT DEER CONSERVATION RESEARCH FOR 5-YEAR FEDERAL AID GRANT W-185-R JULY 2006 – JUNE 2011 CHAD J. BISHOP, CHARLES R. ANDERSON, JR., AND ERIC J. BERGMAN PROGRAM NEED During the late 1990s, CPW was challenged by sportsmen and other stakeholders to investigate potential causes of declining numbers of mule deer in Colorado. The concerns of stakeholders gained the attention of the Colorado Legislature which directed CPW to prepare a document to address causes of the mule deer decline and outline a plan of action to reverse the perceived trend in mule deer populations. That document was prepared for the legislature in 1999 (Gill et al. 2001) and established the direction and objectives for mule deer management and research beginning in 1999. At the same time, the Colorado Wildlife Commission approved statewide limitations on hunting licenses for mule deer, which significantly reduced the number of deer harvested annually in Colorado. Several years later, a sudden and significant increase in natural gas development in the Piceance Basin of northwest Colorado prompted mule deer researchers and managers to initiate a comprehensive effort to mitigate development impacts on deer. The research projects conducted during this 5-year grant period directly or indirectly addressed these various management issues and concerns. This report highlights the accomplishments of research efforts conducted by CPW from July 1, 2006 through June 30, 2011 that were wholly or partially supported by Federal Aid Grant funds. PROGRAM NARRATIVE OBJECTIVES The primary Program Narrative research objectives were divided into two broad categories: 1) managing factors limiting mule deer populations, and 2) monitoring mule deer populations. The specific project objectives were: Managing Factors Limiting Mule Deer Populations Project 1 Objective. Evaluate the impacts of prescribed landscape habitat manipulations in senescent pinyon-juniper habitats on behavior and demographics (survival, reproduction, densities) of mule deer populations. Project 2 Objective. Evaluate approaches to mitigate the impacts of natural gas resource extraction and other related human-caused developments on mule deer habitats and population demographics. Project 3 Objective. Investigate behavioral and spatial relationships between mule deer and elk, and among mule deer, elk, and cougar as these species simultaneously utilize prescribed landscape habitat manipulations. Monitoring Mule Deer Populations Project 4 Objective. Evaluate the technical quality and applications of statewide mule deer research and management systems. Project 5 Objective. Evaluate new approaches to monitoring mule deer population demographics and habitat conditions. 39 �Project 6 Objective. Evaluate hunting systems that could maintain a balance between hunter opportunity and the quality of hunting experience. RESULTS Objective 1. Evaluate the impacts of prescribed landscape habitat manipulations in senescent pinyon-juniper habitats on behavior and demographics (survival, reproduction, densities) of mule deer populations. Project Objective 1 was formulated in response to field research conducted during the previous 5year grant cycle, which indicated that habitat quality was ultimately limiting mule deer population growth in western Colorado. Final data analyses and preparation of publications from this research was completed during 2006-2008, and therefore, are reported here as part of this project objective. We evaluated the effect of enhanced nutrition of deer during winter and spring on fecundity and survival rates of free-ranging mule deer on the Uncompahgre Plateau in southwest Colorado. The treatment represented an instantaneous increase in nutritional carrying capacity of a pinyon (Pinus edulis)−Utah juniper (Juniperus osteosperma) winter range and was intended to simulate optimum habitat quality. Prior studies on the Uncompahgre Plateau indicated predation and disease were the most common proximate causes of deer mortality. By manipulating nutrition and leaving natural predation unaltered, we determined whether habitat quality was ultimately a critical factor limiting the deer population. We measured annual survival and fecundity of adult females and survival of fawns, then estimated population rate of change as a function of enhanced nutrition. Our estimate of the population rate of change was 1.165 (SE = 0.036) for deer receiving the nutrition treatment and 1.033 (SE = 0.038) for control deer. We documented food limitation in the Uncompahgre deer population because survival of fawns and adult females increased considerably in response to enhanced nutrition. We found strong evidence that enhanced nutrition of deer reduced coyote (Canis latrans) and mountain lion (Puma concolor) predation rates of ≥6-month-old fawns and adult females. We concluded that winter-range habitat quality was a limiting factor of the Uncompahgre Plateau mule deer population. We, therefore, recommended evaluating habitat treatments for deer that were designed to set-back succession and increase productivity of late-seral pinyon-juniper habitats that presently dominate the winter range. Pinyon-juniper habitats across western Colorado have been exposed to minimal natural disturbance during recent decades. In particular, the natural role of fire in these systems has been significantly altered through aggressive efforts to extinguish fires ignited by lightning strikes. Fire suppression has become necessary because human dwellings are scattered across pinyon-juniper habitat throughout much of western Colorado. This has caused many mule deer winter ranges to become dominated by late-seral pinyon-juniper, which is unproductive for mule deer. Collaborative management efforts among state and federal agencies, NGOs, and private citizens have been initiated to incorporate disturbance into pinyon-juniper systems through the use of prescribed fire and mechanical treatments that remove or mulch pinyon and juniper trees. We evaluated the effectiveness of these types of habitat treatments on mule deer body condition, survival, and density. Peer-Reviewed Publications: Watkins, B. E., C. J. Bishop, E. J. Bergman, A. Bronson, B. Hale, B. F. Wakeling, L. H. Carpenter, and D. W. Lutz. 2007. Habitat guidelines for mule deer: Colorado Plateau shrubland and forest ecoregion. Mule Deer Working Group, Western Association of Fish and Wildlife Agencies. Schultheiss, P. C., H. Van Campen, T. R. Spraker, C. J. Bishop, L. L. Wolfe, and B. Podell. 2007. Malignant catarrhal fever associated with ovine herpesvirus-2 in free-ranging mule deer in Colorado. Journal of Wildlife Diseases 43:533−537. Bishop, C. J., G. C. White, and P. M. Lukacs. 2008. Evaluating dependence among mule deer siblings in fetal and neonatal survival analyses. Journal of Wildlife Management 72:1085−1093. 40 �Bishop, C. J., B. E. Watkins, L. L. Wolfe, D. J. Freddy, and G. C. White. 2009. Evaluating mule deer body condition using serum thyroid hormone concentrations. Journal of Wildlife Management 73:462−467. Bishop, C. J., G. C. White, D. J. Freddy, B. E. Watkins, and T. R. Stephenson. 2009. Effect of enhanced nutrition on mule deer population rate of change. Wildlife Monographs 172:1−28. Annual Wildlife Research Reports: Bishop, C. J., G. C. White, D. J. Freddy, and B. E. Watkins. 2007. Effect of nutrition and habitat enhancements on mule deer recruitment and survival rates. Colorado Division of Wildlife, Wildlife Research Report July: 59-71. Bergman, E. J., C. J. Bishop, D. J. Freddy, and G. C. White. 2007. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Colorado Division of Wildlife, Wildlife Research Report July: 73-96. Bishop, C. J., G. C. White, D. J. Freddy, and B. E. Watkins. 2008. Effect of nutrition and habitat enhancements on mule deer recruitment and survival rates. Colorado Division of Wildlife, Wildlife Research Report July: 39-51. Bergman, E. J., C. J. Bishop, D. J. Freddy, and G. C. White. 2008. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Colorado Division of Wildlife, Wildlife Research Report July: 53-62. Bergman, E. J., C. J. Bishop, D. J. Freddy, and G. C. White. 2009. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Colorado Division of Wildlife, Wildlife Research Report July: 101-110. Bergman, E. J., C. J. Bishop, D. J. Freddy, G. C. White, and P. F. Doherty. 2010. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Colorado Division of Wildlife, Wildlife Research Report July: 81-91. Bergman, E. J., C. J. Bishop, D. J. Freddy, and G. C. White. 2011. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Colorado Division of Parks and Wildlife, Wildlife Research Report July: in press. Presentations at Professional Meetings/Workshops/Symposia: Bishop, C. J., G. C. White, D. J. Freddy, and B. E. Watkins. 2006. Effect of enhanced nutrition of freeranging mule deer on population performance. The Wildlife Society 13th Annual Conference, September 23−27, Anchorage, Alaska, USA. Bishop, C. J., G. C. White, D. J. Freddy, and B. E. Watkins. 2007. Effect of enhanced nutrition of freeranging mule deer on population performance and effectiveness of vaginal implant transmitters. Colorado State University Student Chapter of The Wildlife Society, February 26, Fort Collins, Colorado, USA. Bishop, C. J. 2007. Capture techniques and radio-telemetry used in wildlife research and management, and an example of technique application using the Uncompahgre deer research study. Colorado State University’s Wildlife Management Short Course, March 26−30, Fort Collins, CO, USA. Bishop, C. J., and E. J. Bergman. 2007. Status of big game habitats and implications for wildlife within the Colorado Plateau. Plant Community Restoration Workshop, September 5−7, Grand Junction, Colorado, USA. Bishop, C. J., G. C. White, and P. M. Lukacs. 2007. Evaluating dependence among mule deer siblings in fetal and neonatal survival analyses. The Wildlife Society 14th Annual Conference, September 22−26, Tucson, Arizona, USA. Bishop, C. J., G. C. White, and P. M. Lukacs. 2008. Evaluating dependence among mule deer siblings in fetal and neonatal survival analyses. Colorado Chapter of The Wildlife Society Annual Meeting, January 23−25, Denver, Colorado, USA. 41 �Bishop, C. J. 2008. Capture techniques and radio-telemetry used in wildlife research and management, and an example of technique application using the Uncompahgre deer research study. Colorado State University’s Wildlife Management Short Course, April 1, Fort Collins, CO, USA. Bergman, E. J., C. J. Bishop, D. J. Freddy, and G. C. White. 2008. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Colorado Division of Wildlife Research Review, August 20-21, Denver, CO, USA. Bergman, E. J. 2009. Monitoring habitat for deer. Joint Meeting of Colorado’s Habitat Partnership Program and the Colorado Chapter of The Wildlife Society, February 5, Grand Junction, CO, USA. Bishop, C. J. 2009. Capture techniques and radio-telemetry used in wildlife research and management, ungulate ecology, and a case study using the Uncompahgre deer research study. Colorado State University’s Wildlife Management Short Course, March 31, Fort Collins, CO, USA. Bergman, E. J., C. J. Bishop, D. J. Freddy, G. C. White, and P. F. Doherty. 2009. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Colorado State University Student Chapter of The Wildlife Society, April, Fort Collins, CO, USA. Bergman, E. J., C. J. Bishop, D. J. Freddy, G. C. White, and P. F. Doherty. 2009. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. 2009 Western States and Provinces Deer and Elk Workshop, May, Spokane, Washington, USA. Bishop, C. J. 2010. Capture techniques and radio-telemetry used in wildlife research and management, ungulate ecology, and a case study using the Uncompahgre deer research study. Colorado State University’s Wildlife Management Short Course, March 30, Fort Collins, CO, USA. Bergman, E. J., C. J. Bishop, D. J. Freddy, G. C. White, and P. F. Doherty. 2010. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Joint Meeting of Colorado’s Habitat Partnership Program and the Colorado Section of the Society of Range Management, December 1, Grand Junction, CO, USA. Bergman, E. J., C. J. Bishop, D. J. Freddy, G. C. White, and P. F. Doherty. 2011. Evaluation of winter range habitat treatments on overwinter survival of mule deer. Northwest Region Biology Days, January 19, Glenwood Springs, CO, USA. Bishop, C. J. 2011. Capture techniques and radio-telemetry used in wildlife research and management, ungulate ecology, and a case study using the Uncompahgre deer research study. Colorado State University’s Wildlife Management Short Course, March 29, Fort Collins, CO, USA. Bishop, C. J., G. C. White, D. J. Freddy, B. E. Watkins, and T. R. Stephenson. 2011. Effect of enhanced nutrition of free-ranging mule deer on population performance. 2011 Western States and Provinces Deer and Elk Workshop, May 17, Santa Ana Pueblo, New Mexico, USA. Bergman, E. J., C. J. Bishop, D. J. Freddy, G. C. White, and P. F. Doherty. 2011. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. 2011 Western States and Provinces Deer and Elk Workshop, May 17, Santa Ana Pueblo, New Mexico, USA. Objective 2. Evaluate approaches to mitigate the impacts of natural gas resource extraction and other related human-caused developments on mule deer habitats and population demographics. We designed and implemented a project to experimentally evaluate habitat treatments and human-activity management alternatives (i.e., best management practices; BMPs) that may benefit mule deer exposed to extensive energy development. The Piceance Basin of northwestern Colorado was selected as the project area due to ongoing natural gas development in one of the most extensive and important mule deer winter and transition range areas within the state. This project was initiated in 2007 and is expected to go to 2016 at a minimum and ideally to 2019. The project timeline was recently extended by 1 year due to a delay in implementing habitat mitigation treatments. The Piceance Basin in northwest Colorado supports one of the largest migratory mule deer populations in North America and also exhibits one of the highest natural gas reserves in North America. 42 �Public stakeholders and CPW are concerned that the cumulative impacts of natural gas extraction will negatively affect mule deer and other wildlife resources in the region. Concern is particularly high for mule deer due to their recreational and economic importance as a principal game species and their ecological importance as one of the primary herbivores of the Colorado Plateau Ecoregion. Extraction of natural gas is directly affecting the potential suitability of the landscape for mule deer by converting native habitat vegetation to drill pads, roads, and noxious weeds, by fragmenting habitat because of drill pads and roads, by increasing noise levels via compressor stations and vehicle traffic, and by increasing the year-round presence of human activities. Extraction is indirectly affecting deer by increasing the human work-force population of the region and the subsequent need for developing additional landscape for human housing, supporting businesses, and upgraded road/transportation infrastructure. Additionally, increased traffic on rural roads is raising the potential for direct mortality from vehicle-animal collisions. Thus, research documenting these impacts and evaluating the most effective strategies for minimizing and mitigating these activities will greatly enhance future management efforts to sustain mule deer populations for future recreational and ecological values. Impacts of natural gas development may be most effectively mitigated for mule deer by restoring or enhancing habitat conditions on or adjacent to disturbed sites and by modifying development practices. However, we presently lack information to appropriately guide the expenditure of mitigation dollars to offset or lessen impacts. The purpose of this project is to address these mitigation questions so that dollars are spent wisely. For example, it remains unknown whether we can effectively mitigate impacts of natural gas development by treating habitat within a developing area. Results from this study will indicate whether mitigation dollars would be better spent enhancing/restoring habitat on-site or enhancing habitat in adjacent, undeveloped areas. Although not hypothesized, there is also the possibility that efforts to enhance habitat within heavily developed areas have a negative impact on deer and other species by causing further disturbance. Thus, this project will scientifically assess approaches for mitigating effects of natural gas development on mule deer to guide future management decisions. From December 2007 to present, we gathered baseline demographic and habitat utilization data from radio-collared deer across the Piceance Basin to allow assessment of mitigation approaches that are presently being implemented. We selected 5 winter range study areas representing varying levels of development to serve as treatment and control sites and recorded habitat use and movement patterns using GPS collars. We also estimated winter fawn survival and annual adult female survival, late winter body condition of adult females using ultrasonography, and deer abundance using helicopter mark-resight surveys. We started with 5 study sites to allow flexibility to respond to changing energy development plans, which can directly affect experimental design. In 2009, we refined our study design using our baseline deer data and current energy development plans of the major companies operating in Piceance Basin. We also eliminated a study site to reduce the annual project budget to the minimum necessary to meet the original research objectives. During December 2010-January 2011, we implemented 100 acres of habitat treatments as a pilot effort to evaluate logistics and effectiveness of habitat treatment strategies. We will implement an additional 1,100 acres of habitat treatments across two of our study sites as a mitigation strategy during 2011-13. ExxonMobil Corporation is directly funding all habitat treatments in this research as part of an agreed-upon mitigation plan with CPW. One study site receiving habitat treatments has undergone extensive energy development whereas the other site receiving treatments is experiencing modest development. We will continue to collect the various population and habitat use data across all study sites in order to evaluate the effectiveness of the habitat treatments. This approach will allow us to determine whether it is possible to effectively mitigate development impacts in highly developed areas, or whether it is better to allocate mitigation dollars toward less-impacted areas. We may also find that habitat mitigation efforts are not effective in developed areas at all, suggesting that habitat enhancement efforts may be only effective in areas that are not impacted by development. In 2010, we initiated a PhD project 43 �in collaboration with Colorado State University and ExxonMobil to evaluate deer behavioral responses to varying levels of development activity and habitat mitigation treatments. ExxonMobil is funding this project via a cooperative funding agreement with Colorado State University and CPW. This will allow us to assess the effectiveness of certain BMPs and habitat manipulations for reducing disturbance to deer. We also initiated a Masters project in collaboration with CSU and funded by ExxonMobil to evaluate vegetation responses to the habitat treatments described above. Danielle Johnston in the Avian Research Section is taking the lead on this project, working in collaboration with Chuck Anderson. Last, we plan to initiate a PhD project in collaboration with CSU during FY 11-12 to measure neonatal deer survival, also funded by ExxonMobil. Through combined funding from Federal Aid and energy companies, we are comprehensively evaluating effects of natural gas development on deer and associated mitigation strategies. Annual Wildlife Research Reports: Anderson, C. R., and D. J. Freddy. 2007. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Colorado Division of Wildlife, Wildlife Research Report July: 103-110. Anderson, C. R., and D. J. Freddy. 2008. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Stage I, Objective 5: patterns of mule deer distribution and movements. Colorado Division of Wildlife, Wildlife Research Report July: 63-86. Anderson, C. R. 2009. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Colorado Division of Wildlife, Wildlife Research Report July: 111-124. Anderson, C. R., and C. J. Bishop. 2010. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Colorado Division of Wildlife, Wildlife Research Report July: 47-62. Anderson, C. R., and C. J. Bishop. 2011. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Colorado Division of Parks and Wildlife, Wildlife Research Report July: in press. Presentations at Professional Meetings/Workshops/Symposia: Anderson, C. R., and D. J. Freddy. 2008. Population performance of Piceance Basin mule deer in relation to natural gas development and mitigation measures to address habitat degradation and human activity management alternatives. Tri-state energy meeting addressing wildlife management in relation to energy development activities, Parachute, CO, USA. Anderson, C. R. 2008. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Energy Company Cooperators Meeting, October, Grand Junction, CO, USA. Anderson, C. R. 2009. Population performance of Piceance Basin mule deer in relation to natural gas development and mitigation measures to address habitat degradation and human activity management alternatives. Graduate-Faculty Seminar Series, Colorado State University, September 18, Fort Collins, CO, USA. Anderson, C. R. 2009. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Energy Company Cooperators Meeting, October, Grand Junction, CO, USA. Anderson, C. R. 2010. Population performance of Piceance Basin mule deer in relation to natural gas development and mitigation measures to address habitat degradation and human activity management alternatives. Faculty-Student Seminar, Western State College, Gunnison, CO, USA. 44 �Anderson, C. R., and C. J. Bishop. 2010. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Energy Company Cooperators Meeting, October, Grand Junction, CO, USA. Anderson, C. R. 2010. Piceance Basin mule deer and energy development: improving winter range habitat as mitigation. Joint Meeting of Colorado’s Habitat Partnership Program and Colorado Section of the Society for Range Management, December 1, Grand Junction, CO, USA. Anderson, C. R. 2011. Piceance Basin mule deer and energy development: improving winter range habitat as mitigation. Northwest Region Biology Training, January 19, Glenwood Springs, CO, USA. Anderson, C. R., and C. J. Bishop. 2011. Current understanding of mule deer-energy development interactions in the western United States. Northwest Region Biology Training, January 19, Glenwood Springs, CO, USA. Northrup, J., G. Wittemyer, and C. R. Anderson. 2011. Behavioral response of mule deer to energy development activities in the Piceance Basin, Colorado. Colorado Chapter of The Wildlife Society Annual Meeting, February 25, Fort Collins, CO, USA. Objective 3. Investigate behavioral and spatial relationships between mule deer and elk, and among mule deer, elk, and cougar as these species simultaneously utilize prescribed landscape habitat manipulations. We capitalized on an opportunity to simultaneously monitor spatial movements and predator-prey dynamics of radio-collared mule deer, elk, and cougars on the Uncompahgre Plateau. Mule deer were marked as part of ongoing research described above under Objective 1. Elk were marked as part of a pilot study to monitor spatial movements of deer and elk on the Uncompahgre Plateau, and cougars were marked with GPS collars as part of a long-term research study (not funded by Federal Aid) evaluating the effects of harvest on cougar populations and the assumptions used by CPW to manage cougar populations. Our primary goal was to improve understanding of cougar-prey dynamics. We investigated GPS location clusters for cougars and assessed if a predation event occurred and what species of prey was involved. Locations of predation events were assessed in relation to vegetation treatments applied to the landscape to benefit mule deer and elk. As predicted, cougar kill sites were associated with deer and elk distribution. The greatest density of kill sites occurred across mid-upper elevation deer winter range where overlap of wintering elk and deer was greatest. We investigated 462 clusters during this pilot study. Kill probability increased as cluster size increased. Kill probability exceeded 0.9 with ≥ 10 locations/cluster and approached 1 with ≥ 15 locations/cluster. The probability of a kill was high if a cougar spent >2 days in the same general area, and a kill was essentially certain if a cougar spent >3 days in the same general area. There was some probability of a kill at clusters that comprised only 1 location, indicating that isolated cougar locations may periodically be associated with kills and should not be ruled out when using GPS location data to address cougar prey utilization. Our estimates of kill probability are conservative because the estimates assume prey detection probability was 1, which is unlikely. Cougars killed adult deer, fawn deer, adult elk, and calf elk in roughly equal proportions. Each prey class comprised 0.22−0.24 of the total kill. Kill composition varied as a function of percent vegetative cover and elevation. In FY 09-10, for logistical and study design reasons, we transitioned all research on this objective to a non-Federal Aid cougar project along the Front Range of Colorado. Annual Wildlife Research Reports: Alldredge, M. W., E. J. Bergman, C. J. Bishop, K. A. Logan, and D. J. Freddy. 2008. Pilot evaluation of predator-prey dynamics on the Uncompahgre Plateau. Colorado Division of Wildlife, Wildlife Research Report July: 87-104. 45 �Objective 4. Evaluate the technical quality and applications of statewide mule deer research and management systems. Considerable progress has been made during recent decades in developing and implementing quality mule deer research and management programs within CPW by enlisting the biostatistical support of faculty at Colorado State University (CSU). This objective has been attained for many years via annual contract for professional services with individual faculty at CSU. Federal Aid grant funding has routinely been used to help fund this contract to support mule deer management and research. Other funds (non-Federal Aid) have also supported this contract, which permits biostatistical support of other research and management functions in CPW as well. During 2006-07, Gary White (CSU faculty) provided support to CPW biologists on designing and implementing harvest surveys, terrestrial inventory systems, and population modeling procedures. Ongoing support was also provided for CPW’s DEAMAN software package, which was used by staff for the storage, summary, and analysis of mule deer and other big game population and harvest data. In July 2007, CPW terminated the annual contract with faculty at CSU after hiring a permanent biometrician within CPW to provide these same services in-house. Peer-Reviewed Publications: McClintock, B. T., G. C. White, and K. P. Burnham. 2006. A robust design mark-resight abundance estimator allowing heterogeneity in resighting probabilities. Journal of Agricultural, Biological, and Ecological Statistics 11:231-248. Martin, D. J., G. C. White, and F. M. Pusateri. 2007. Occupancy rates by swift foxes (Vulpes velox) in eastern Colorado. Southwestern Naturalist 52:541-551. White, G. C. 2008. Closed population estimation models and their extensions in Program MARK. Environmental and Ecological Statistics 15:89-99. Odell, E. A., F. M. Pusateri, and G. C. White. 2008. Estimation of occupied and unoccupied black-tailed prairie dog colony acreage in Colorado. Journal of Wildlife Management 72:1311-1317. Conn, P. B., D. R. Diefenbach, J. L. Laake, M. A. Ternent, and G. C. White. 2008. Bayesian analysis of wildlife age-at-harvest data. Biometrics 64:1170-1177. Annual Wildlife Research Reports: White, G. C. 2007. Multispecies investigations consulting services for mark-recapture analysis. Colorado Division of Wildlife, Wildlife Research Report July: 97-101. Objective 5. Evaluate new approaches to monitoring mule deer population demographics and habitat conditions. We conducted two separate research projects focused on the development and evaluation of new approaches to enhance monitoring of mule deer populations for research and management: 1) modification and evaluation of vaginal implant transmitters in deer, and 2) development of an automated collaring device for mule deer. Redesigned Vaginal Implant Transmitters Our understanding of factors that limit mule deer populations may be improved by evaluating neonatal survival as a function of dam characteristics under free-ranging conditions, which generally requires that both neonates and dams are radiocollared. The only viable technique facilitating capture of neonates from radiocollared adult females is use of vaginal implant transmitters (VITs). To date, VITs have allowed research opportunities that were not possible previously; however, VITs are often expelled from adult females prepartum, which limits their utility. During the previous 5-year Federal Aid Grant Segment, we evaluated effectiveness of VITs in mule deer. Based on this research, during the current grant segment, we redesigned an existing vaginal implant transmitter (VIT) manufactured by Advanced 46 �Telemetry Systems (ATS) by lengthening and widening wings used to retain the VIT in an adult female. Our objective was to increase VIT retention rates to increase likelihood of locating birth sites and newborn fawns. We placed the newly designed VITs in 59 adult female mule deer and evaluated the probability of retention to parturition and the probability of detecting newborn fawns. The probability of a VIT being retained until parturition was 0.766 (SE = 0.0605) and the probability of a VIT being retained to within 3 days of parturition was 0.894 (SE = 0.0441). In our earlier study using the original VIT wings, the probability of a VIT being retained until parturition was 0.447 (SE = 0.0468) and the probability of retention to within 3 days of parturition was 0.623 (SE = 0.0456). Thus, our design modification increased VIT retention to parturition by 0.319 (SE = 0.0765) and VIT retention to within 3 days of parturition by 0.271 (SE = 0.0634). Considering dams that retained VITs to within 3 days of parturition, the probability of detecting at least 1 neonate was 0.952 (SE = 0.0334) and the probability of detecting both fawns from twin litters was 0.588 (SE = 0.0827). Our study expands opportunities for conducting research that links adult female attributes to productivity and offspring survival in mule deer. Automated Collaring Device for Deer We designed and produced a trap-like device for mule deer that would automatically attach a radio collar to a ≥6-month-old fawn and record the fawn’s weight and sex, without requiring physical restraint or handling of the animal. Our passive collaring device is designed to allow biologists and researchers to radio-collar, weigh, and identify sex of ≥6-month-old mule deer fawns with minimal expense and labor when compared to traditional mule deer capture techniques. This technique significantly reduces stress that is typically associated with capture and handling and should eliminate capture-related mortality. We collaborated with students and faculty in the Mechanical Engineering Department at Colorado State University to produce a conceptual model and early prototype. We then worked with professional engineers at Dynamic Group Circuit Design in Fort Collins, Colorado, to produce a fully-functional prototype of the device. We conducted an extensive field evaluation of the device with free-ranging mule deer during winter 2010-11. We successfully collared, weighed, and identified sex of 6 different mule deer fawns across 4 winter range locations along Colorado’s northern Front Range. Collars were purposefully made to shed from deer within several weeks or months of being captured. Two fawns were successfully re-collared after they shed the first collars they received. Thus, we observed 8 successful collaring events involving 6 different fawns. Most fawns demonstrated minimal response to collaring events, either remaining in the device or calmly exiting. Certain components of the collaring device failed to function optimally when temperatures dropped below approximately −15° C, while other components did not adequately withstand mule deer use under field conditions. Also, certain behaviors of mule deer when approaching and using the device created circumstances where it was possible to collar the same animal twice, which happened on one occasion. We identified a series of device modifications that would be necessary to address these various issues. We will modify the device accordingly and conduct a follow-up field evaluation during 2011-12. Peer-Reviewed Publications: Bishop, C. J., D. J. Freddy, G. C. White, B. E. Watkins, T. R. Stephenson, and L. L. Wolfe. 2007. Using vaginal implant transmitters to aid in capture of mule deer neonates. Journal of Wildlife Management 71:945−954. Bishop, C. J., C. R. Anderson, Jr., D. P. Walsh, E. J. Bergman, P. Kuechle, and J. Roth. 2011. Effectiveness of a redesigned vaginal implant transmitter in mule deer. Journal of Wildlife Management 75:1797−1806. Annual Wildlife Research Reports: Bishop, C. J., D. P. Walsh, M. W. Alldredge, E. J. Bergman, and C. R. Anderson. 2009. Development of an automated device for collaring and weighing mule deer fawns. Colorado Division of Wildlife, Wildlife Research Report July: 55-67. 47 �Bishop, C. J., C. R. Anderson, D. P. Walsh, P. Kuechle, J. Roth, and E. J. Bergman. 2009. Effectiveness of a redesigned vaginal implant transmitter in mule deer. Colorado Division of Wildlife, Wildlife Research Report July: 69-99. Bishop, C. J., C. R. Anderson, D. P. Walsh, E. J. Bergman, P. Kuechle, and J. Roth. 2010. Effectiveness of a redesigned vaginal implant transmitter in mule deer. Colorado Division of Wildlife, Wildlife Research Report July: 63-80. Bishop, C. J., D. P. Walsh, M. W. Alldredge, E. J. Bergman, and C. R. Anderson. 2010. Development of an automated device for collaring and weighing mule deer fawns. Colorado Division of Wildlife, Wildlife Research Report July: 93-100. Bishop, C. J., C. R. Anderson, D. P. Walsh, E. J. Bergman, P. Kuechle, and J. Roth. 2011. Effectiveness of a redesigned vaginal implant transmitter in mule deer. Colorado Division of Parks and Wildlife, Wildlife Research Report July: in press. Bishop, C. J., M. W. Alldredge, D. P. Walsh, E. J. Bergman, and C. R. Anderson. 2011. Development of an automated device for collaring and weighing mule deer fawns. Colorado Division of Parks and Wildlife, Wildlife Research Report July: in press. Presentations at Professional Meetings/Workshops/Symposia: Bishop, C. J., D. J. Freddy, G. C. White, B. E. Watkins, T. R. Stephenson, and L. L. Wolfe. 2007. Using vaginal implant transmitters to aid in capture of mule deer neonates. Colorado Chapter of The Wildlife Society Annual Meeting, January 17−19, Glenwood Springs, Colorado, USA. Bishop, C. J., D. J. Freddy, G. C. White, B. E. Watkins, T. R. Stephenson, and L. L. Wolfe. 2007. Using vaginal implant transmitters to aid in capture of mule deer neonates. 7th Western States and Provinces Deer and Elk Workshop, May 13−16, Estes Park, Colorado, USA. Bishop, C. J., M. W. Alldredge, D. P. Walsh, E. J. Bergman, and D. Kilpatrick. 2011. Automated collaring device for mule deer. Colorado Chapter of The Wildlife Society Annual Meeting, February 25, Fort Collins, CO, USA. Objective 6. Evaluate hunting systems that could maintain a balance between hunter opportunity and the quality of hunting experience. Historically, management of big game species has focused on the performance of the female and young of the year components of the population. In the case of mule deer, this has been further refined to the aspects of annual (for adult females) and overwinter (for young of the year) survival. The performance of the male component of populations was deemed less important, primarily due to the fact that it takes relatively few males to provide adequate breeding potential for the population. Additionally, harvest management objectives were to provide maximal hunting opportunity for hunters. Thus, as long as there were adequate numbers of males to breed females there was no need to restrict hunting opportunity. However, during the past 10-15 years, the management of big game populations, and mule deer populations in particular, has started to shift away from the objective of providing maximal opportunity towards providing fewer but higher quality opportunities. High quality opportunities are typically defined by hunters as a combination of the opportunity to see a greater number of male deer during the hunt, the potential to harvest an older age class animal (i.e., an animal with more developed antler morphometry), but also reduced interaction and competition with other hunters. In response to this shift in hunter desires and concerns over declining mule deer numbers, in 1999 CPW implemented a statewide limitation in deer hunting. This statewide limitation gave the CPW the ability to greatly reduce total hunter numbers but also the ability to control the distribution of hunters throughout the state. Since 1999, a few marked changes in Colorado’s deer herd have occurred. First, due to reduced harvest an overall increase in deer numbers has been observed. Second, because the reduction in harvest was primarily focused on adult males, a subsequent increase in the ratio of adult males to adult females has occurred. Stemming from this shift in harvest management and the subsequent changes in herd size and structure, a gap in biological information has been identified. Specifically, Colorado’s deer herds have 48 �become composed of a greater number of males, yet little biological data on them exist. Also stemming from the change in harvest management was a new responsibility for Colorado’s terrestrial biologists and wildlife managers. Prior to 1999, licenses were sold over-the-counter and were not limited in number (i.e., any hunter who wished to purchase one was able to do so). The decision of how many licenses to make available did not need to be considered. Since 1999, the CPW has the added responsibility of deciding how many licenses should be allocated in each Data Analysis Unit (DAU). This decision must further reflect a balance between meeting DAU population performance objectives, but also provide as much hunter opportunity as possible. Big game populations in Colorado are currently modeled using multiple sources of biological data (White and Lubow 2002). Model inputs include harvest, young recruitment to December, and measured rates of survival of adult females and fawns. Also, the ratio of adult males to adult females is estimated and used to align models by minimizing the difference between observed and modeled values. Very rarely have the survival rates of adult males been measured. This gap in knowledge has historically been viewed as trivial and rates have been assumed to not be different from the rates of females. Similarly, it has been assumed that natural survival rates (i.e., post hunt survival) of males do not geographically vary. However, model performance under these assumptions has been poor and the need to measure adult male survival as a parameter has increased. Presently, a number of population models in Colorado suggest that natural adult male survival may be lower than adult female survival, yet empirical data is lacking to verify these suppositions. A different, but not unrelated need in Colorado pertains to the harvest management of adult male mule deer. As discussed above, a large shift in mule deer herd size and structure occurred as a result of the change in harvest management that occurred between 1998 and 1999. Overall, this shift has been viewed as positive by both the CPW as well as the public. However, the CPW still maintains the responsibility of optimally managing the deer of Colorado and providing the maximal amount of hunting opportunity under this new set of constraints. To date, the CPW has had limited biological information and data to guide harvest management decisions. In particular for this issue, as DAUs reach and surpass their adult male: adult female ratio objectives, the CPW typically responds by increasing the number of available hunting licenses. In situations where herds are continually lower than DAU objectives, available hunting licenses are reduced. What remains unknown about survival of adult male deer is at what level natural survival is reduced due to intraspecific competition. If, or when deer herds exceed the adult male: adult female objectives for DAUs, it is often assumed that the surplus of male deer will remain in the population into perpetuity. However, this assumption is based on the premise that compensatory mortality does not occur. Similarly, it assumes that annual variation in survival is negligible. However, this is biologically not realistic. It is very likely that herds with large post-hunt populations of adult males experience higher levels of mortality. Under this scenario, harvest has not been optimized and more hunters could have been afforded the opportunity to hunt with no effect on post hunting season ratios of adult males to adult females. The simplest way to learn about the mortality process is via manipulative experimentation. Our study objective is two-fold. First, we wish to assess annual survival of adult male mule deer. We wish to establish baseline survival estimates, and related estimates of variance, for different age classes of deer. Second, we wish to manipulate hunting license allocation within the Game Management Units (GMUs) of a single DAU such that adult male: adult female ratios become measurably different between two halves of the DAU. Accordingly, we wish to measure and correlate changes in natural survival of adult male deer with this management action. Similarly, as part of this second objective, we will determine if changes in the age structure of harvested animals occur as the sex ratio and age structure of the hunted population changes. We designed the study and wrote a study plan during 2009-10 and initiated field work during 2010-11. 49 �Peer-Reviewed Publications: Bergman, E. J., B. E. Watkins, C. J. Bishop, P. M. Lukacs, and M. Lloyd. 2011. Biological and socioeconomic effects of statewide limitation of deer licenses in Colorado. Journal of Wildlife Management 75:1443−1452. Annual Wildlife Research Reports: Bergman, E. J., C. J. Bishop, K. Oldham, and L. Sidener. 2011. Assessment of survival and optimal harvest strategies of adult male mule deer in Middle Park, Colorado. Colorado Division of Parks and Wildlife, Wildlife Research Report July: in press. Presentations at Professional Meetings/Workshops/Symposia: Bergman, E. J., B. E. Watkins, C. J. Bishop, P. M. Lukacs, and M. Lloyd. 2007. Biological, social, and economic effects of totally limited deer licenses in Colorado. 7th Western States and Provinces Deer and Elk Workshop, May 13−16, Estes Park, Colorado, USA. Bergman, E. J., B. E. Watkins, C. J. Bishop, P. M. Lukacs, and M. Lloyd. 2008. Biological, social, and economic effects of totally limited deer licenses in Colorado. Colorado Chapter of The Wildlife Society Annual Meeting, January 25, Denver, Colorado, USA. Bergman, E. J., C. J. Bishop, L. Sidener, and K. Oldham. 2011. Survival and optimal harvest management of mule deer bucks in Middle Park, CO. Presentation to the Colorado Wildlife Commission, April 7, Meeker, CO, USA. SUMMARY We conducted work on seven research projects addressing mule deer limiting factors, habitat enhancement, mitigation of natural gas development impacts, predator-prey dynamics, buck harvest management, and technique developments. Additionally, funding provided biostatistical support for implementing or maintaining statewide deer harvest surveys, population databases, aerial surveys, population modeling, and research projects. From activities supported by this Grant during this segment, principal investigators published 13 peer-reviewed scientific articles for prominent wildlife research journals, provided 21 annual CPW Wildlife Research Reports summarizing yearly progress of projects, provided 34 presentations at professional meetings, workshops, or symposia, and initiated 2 graduate student projects. The cumulative impact of this programmatic effort provides Colorado the basis to progress and proactively sustain the mule deer resource in an increasingly complex landscape. The relative success of mule deer management in Colorado reflects the positive synergy between the terrestrial research and management sections in sharing expertise, financial resources, staffing, and common goals. LITERATURE CITED Gill, R.B., T.D.I Beck, C.J. Bishop, D.J. Freddy, N.T. Hobbs, R.H. Kahn, M.W. Miller, T.M. Pojar, and G.C. White. 2001. Declining mule deer populations in Colorado: reasons and responses. Colorado Division of Wildlife Special Report 77. Fort Collins, Colorado, USA. White, G.C. and B.C. Lubow. 2002. Fitting population models to multiple sources of observed data. Journal of Wildlife Management 66:300-309. Prepared by ______________________________________ Chad J. Bishop, Mammals Research Leader 50 � https://cpw.cvlcollections.org/files/original/dd84b6ce08e6e3f362fcbeb51c265a5c.pdf c7f2cfcb97d2c58d8dfc8f796f9ef2c5 https://cpw.cvlcollections.org/files/original/365a0eac97ef867b3fbb52c0addf3fe8.pdf 40caf868430af28c09c0218f01f383a6 PDF Text Text Colorado Division of Wildlife July 2008 − June 2009 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 3001 4 Federal Aid Project No. W-185-R : : : : : Division of Wildlife Mammals Research Deer Conservation Development of an Automated Device for Collaring and Weighing Mule Deer Fawns Period Covered: July 1, 2008 − June 30, 2009 Authors: C. J. Bishop, D. P. Walsh, M. W. Alldredge, E. J. Bergman, and C. R. Anderson. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We initiated an effort to design, produce, and evaluate a trap-like device for mule deer that would automatically attach a radio collar to a ≥6-month-old fawn and record the fawn’s weight and sex, without requiring physical restraint or handling of the animal. A passive collaring device would allow biologists and researchers to radio-collar, weigh, and identify sex of ≥6-month-old mule deer fawns with minimal expense and labor when compared to traditional mule deer capture techniques. Such a technique would significantly reduce stress that is typically associated with capture and handling and would eliminate capture-related mortality. We wrote a study plan (Appendix I) and collaborated with students and faculty in the Mechanical Engineering Department at Colorado State University in an attempt to produce a prototype device. We evaluated device components in phases throughout the year using captive deer at the Foothills Wildlife Research Facility (FWRF) in Fort Collins, Colorado. The students did a good job with the mechanical aspects of the design when developing a prototype, but the electrical controls to run the device were too advanced for them. Although the prototype lacked several key components, we were able to evaluate various aspects of the device to guide further development. We tested the device at FWRF and then conducted a field evaluation with free-ranging deer during April and May, 2009. The latter provided extensive information on how deer interacted with the device. Most importantly, we could have collared free-ranging deer without handling them had the device been fully automated. To produce a fully functional device, we are pursuing a contract with a professional engineering firm capable of meeting our detailed device specifications. 55 �WILDLIFE RESEARCH REPORT DEVELOPMENT OF AN AUTOMATED DEVICE FOR COLLARING AND WEIGHING MULE DEER FAWNS CHAD J. BISHOP, DANIEL P. WALSH, MATHEW W. ALLDREDGE, ERIC J. BERGMAN, AND CHUCK R. ANDERSON P. N. OBJECTIVE To develop and evaluate a trap-like device for mule deer that would automatically attach a radio collar to a ≥6-month-old deer fawn and record the fawn’s weight and sex, without requiring physical restraint or handling of the animal. SEGMENT OBJECTIVES 1. Write a study plan to guide development and evaluation of the automated collaring device. 2. Produce a prototype device and conduct a preliminary field evaluation with mule deer. INTRODUCTION The Colorado Division of Wildlife (CDOW) captures and radio-marks 6-month-old mule deer (Odocoileus hemionus) fawns each year to support research and management of mule deer. Approximately 240 deer fawns are captured annually to monitor survival among 4 populations distributed across western Colorado and an additional 100−350 deer fawns are captured as part of ongoing research studies. Other state agencies in the western United States capture large numbers of mule deer fawns annually also. Most capture is accomplished with net-guns fired from helicopters (Barrett et al. 1982, van Reenen 1982, Webb et al. 2008), which is becoming increasingly expensive (i.e., >$500 per captured deer). Also, net gunning is inherently dangerous with a small market, which at times limits availability of contractors. Drop nets (Ramsey 1968, Schmidt et al. 1978), clover traps (Clover 1956), drive nets (Beasom et al. 1980), and darting (Wolfe et al. 2004) are used occasionally in the western United States to capture deer, but these techniques can be time consuming and labor intensive. Many biologists lack time and resources given other job requirements to conduct such capture operations for any length of time. The increasing cost of helicopter net-gun capture coupled with increasing demand for capturing and radio-collaring 6-month-old fawns has created a need for another capture alternative. Specifically, there is need for a capture technique that is relatively inexpensive to employ considering both operating and personnel costs. In response to CDOW’s capture needs, we conceived the idea of an automated marking device for ≥6-month-old deer fawns that would attach a radio collar and record weight and sex without physically restraining the animal or requiring handling. The idea of automatically attaching radio transmitters to animals is not new, although to our knowledge, there are no proven methods or devices for use on deer or other ungulates. Even a relatively expensive trap or device (e.g., $3,000−5,000 ea.) would reduce CDOW’s capture costs assuming the device could be reused over time with few maintenance expenses. Such a device would enable seasonal wildlife technicians or graduate students to radio-collar samples of deer fawns independently or with little assistance from researchers and biologists because no animal handling would be required. We want the device to record weight and sex because these variables are useful covariates in survival analyses and are typically measured when fawns are captured and handled. 56 �A passive marking device would minimize animal stress associated with capture and should have virtually no potential to cause capture-related mortality. The large-mammal capture techniques described above place considerable, temporary stress on animals as part of netting and handling. Roughly 2-3% of animals typically die from capture-related injuries or stresses under routine capture conditions. Thus, successful development of a passive marking system would reduce CDOW’s operating expenses and improve animal welfare. STUDY AREA We conducted all evaluations with captive deer at the FWRF in Fort Collins, Colorado. We conducted limited evaluations with free-ranging deer near Fort Collins in north-central Colorado. We plan to conduct extensive field evaluations in the Piceance Basin in northwest Colorado once a fullyfunctioning device is produced. METHODS We wrote a study plan and identified detailed device specifications to guide development of the automated collaring device (Appendix I). We approached Colorado State University’s Mechanical Engineering Department to discuss their interest in helping design such a device. In result, the collaring device became a senior design project for 6 CSU engineering students during the 2008-09 school year. We met with the students weekly and provided them a materials budget of $10,000 to produce a prototype device. We conducted staged evaluations of device components during the year by working with captive deer at FWRF. We also conducted limited evaluations with free-ranging deer near the end of the year. Field evaluations focused primarily on how deer utilized and interacted with the device to guide subsequent design and development decisions. We documented utilization and interactions using direct observation and motion-sensor digital cameras. We relied exclusively on digital cameras when we were not on-site during an evaluation. Automation of the collaring device was disabled any time we were not present to prevent any potential harm to deer. RESULTS AND DISCUSSION We completed the study plan and detailed device specifications (Appendix I). The student engineers did a good job with the mechanical aspects of the design, but the electrical controls to run the device were too advanced for them. The students therefore approached a private electrical engineering design firm located in Fort Collins – Dynamic Group Circuit Design (DGCD). DGCD donated many hours to the project to help the students produce a prototype. By spring 2009, we were interacting directly with DGCD in an attempt to make the prototype device function. Although the device lacked several key components, a number of aspects were ready for evaluation. We therefore tested the device at FWRF and then conducted a field evaluation with free-ranging deer during April and May, 2009. The latter provided extensive information on how deer interacted with the device. Most importantly, we could have collared free-ranging deer without handling them had the device been fully automated. In order to produce a fully functional device, we are presently pursuing a contract with DGCD because of their capability to incorporate our complete set of design specifications into the device. SUMMARY We made significant progress toward developing an automated collaring device for mule deer. We now depend on services of professional engineers to complete prototype development and evaluation. If we are successful, the automated collaring device would allow biologists and researchers to radio-collar portions of their deer samples with minimal time and expense because no animal handling would be required and deer could be collared at any time. Primary time commitments would include baiting sites, 57 �moving device(s) among sites, and adding collars to the devices. Once design work is completed, the current estimate for producing one fully functional collaring device is $7,000. At the current net-gunning rate of roughly $550/deer, an individual collaring device would be paid off after 13 deer were collared. Over time, as an individual biologist or researcher accumulated several of these devices, it is reasonable to assume they could collar 25-35 deer with a few weeks of limited effort, amounting to a savings of roughly $14,000-$20,000 per study per year once the devices were paid off. The collaring device would also have distinct benefits for studies in urban environments by providing a non-invasive technique for collaring deer. The collaring device would significantly reduce stress that is typically associated with capture and handling and there should be no capture-related mortality. We also have designed the collaring device so that it should be relatively easy to adjust to target adult deer and other ungulate species. Last, the collaring device would have wide applicability for ungulate researchers and managers beyond Colorado. LITERATURE CITED Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108−114. Beasom, S. L., W. Evans, and L. Temple. 1980. The drive net for capturing western big game. Journal of Wildlife Management 44:478−480. Clover, M. R. 1956. Single-gate deer trap. California Fish and Game 42:199−201. Ramsey, C. W. 1968. A drop-net deer trap. Journal of Wildlife Management 32:187−190. Schmidt, R. L., W. H. Rutherford, and F. M. Bodenham. 1978. Colorado bighorn sheep-trapping techniques. Wildlife Society Bulletin 6:159−163. van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408−421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. Webb, S. L., J. S. Lewis, D. G. Hewitt, M. W. Hellickson, and F. C. Bryant. 2008. Assessing the helicopter and net gun as a capture technique for white-tailed deer. Journal of Wildlife Management 72:310−314. Wolfe, L. L., M. W. Miller, and E. S. Williams. 2004. Feasibility of “test-and-cull” for managing chronic wasting disease in urban mule deer. Wildlife Society Bulletin 32:500−505. Prepared by Chad J. Bishop, Wildlife Researcher 58 �APPENDIX I PROGRAM NARRATIVE STUDY PLAN FOR MAMMALS RESEARCH FY 2008-09 – FY 2009-10 State of: Cost Center: Work Package: Task No. Colorado 3430 3001 8 Federal Aid Project No. W-185-R : : : : : Division of Wildlife Mammals Research Deer Conservation Development of an Automated Device for Collaring and Weighing Mule Deer Fawns DEVELOPMENT OF AN AUTOMATED DEVICE FOR COLLARING AND WEIGHING MULE DEER FAWNS Principal Investigators Chad J. Bishop, Mammals Researcher, Colorado Division of Wildlife Daniel P. Walsh, Wildlife Health Researcher, Colorado Division of Wildlife Eric J. Bergman, Mammals Researcher, Colorado Division of Wildlife Mathew W. Alldredge, Mammals Researcher, Colorado Division of Wildlife Chuck R. Anderson, Mammals Researcher, Colorado Division of Wildlife Cooperators Mechanical Engineering Department, Colorado State University Michael Sirochman, Veterinarian Technician, Colorado Division of Wildlife John Broderick, Senior Terrestrial Biologist, Colorado Division of Wildlife Lisa L. Wolfe, Veterinarian, Colorado Division of Wildlife Michael W. Miller, Wildlife Health Leader, Colorado Division of Wildlife Stewart Breck, Research Wildlife Biologist, National Wildlife Research Center STUDY PLAN APPROVAL Prepared by: Chad J. Bishop Date: Nov 2008 Submitted by: Chad J. Bishop Date: Nov 2008 Reviewed by: Date: Date: Biometrician: Approved by: Date: Michael W. Miller Mammals Research Leader, Acting 59 Date: �PROGRAM NARRATIVE STUDY PLAN FOR MAMMALS RESEARCH DEVELOPMENT OF AN AUTOMATED DEVICE FOR COLLARING AND WEIGHING MULE DEER FAWNS A Study Plan Proposal Submitted by: Chad J. Bishop, Mammals Researcher, Colorado Division of Wildlife Daniel P. Walsh, Wildlife Health Researcher, Colorado Division of Wildlife Eric J. Bergman, Mammals Researcher, Colorado Division of Wildlife Mathew W. Alldredge, Mammals Researcher, Colorado Division of Wildlife Chuck R. Anderson, Mammals Researcher, Colorado Division of Wildlife A. Need The Colorado Division of Wildlife (CDOW) captures and radio-marks 6-month-old mule deer (Odocoileus hemionus) fawns each year to support research and management of mule deer. Approximately 240 deer fawns are captured annually to monitor survival among 4 populations distributed across western Colorado and an additional 100−350 deer fawns are captured as part of ongoing research studies. Other state agencies in the western United States capture large numbers of mule deer fawns annually also. Most capture is accomplished with net-guns fired from helicopters (Barrett et al. 1982, van Reenen 1982, Webb et al. 2008), which is becoming increasingly expensive (i.e., >$500 per captured deer). Also, net gunning is inherently dangerous with a small market, which at times limits availability of contractors. Drop nets (Ramsey 1968, Schmidt et al. 1978), clover traps (Clover 1956), drive nets (Beasom et al. 1980), and darting (Wolfe et al. 2004) are used occasionally in the western United States to capture deer, but these techniques can be time consuming and labor intensive. Many biologists lack time and resources given other job requirements to conduct such capture operations for any length of time. The increasing cost of helicopter net-gun capture coupled with increasing demand for capturing and radio-collaring 6-month-old fawns has created a need for another capture alternative. Specifically, there is need for a capture technique that is relatively inexpensive to employ considering both operating and personnel costs. In response to CDOW’s capture needs, we conceived the idea of an automated marking device for ≥6-month-old deer fawns that would attach a radio collar and record weight and sex without physically restraining the animal or requiring handling. The idea of automatically attaching radio transmitters to animals is not new, although to our knowledge, there are no proven methods or devices for use on deer or other ungulates. Even a relatively expensive trap or device (e.g., $3,000−5,000 ea.) would reduce CDOW’s capture costs assuming the device could be reused over time with few maintenance expenses. Such a device would enable seasonal wildlife technicians or graduate students to radio-collar samples of deer fawns independently or with little assistance from researchers and biologists because no animal handling would be required. We want the device to record weight and sex because these variables are useful covariates in survival analyses and are typically measured when fawns are captured and handled. A passive marking device would minimize animal stress associated with capture and should have virtually no potential to cause capture-related mortality. The large-mammal capture techniques described above place considerable, temporary stress on animals as part of netting and handling. Roughly 2-3% of animals typically die from capture-related injuries or stresses under routine capture conditions. Thus, successful development of a passive marking system would reduce CDOW’s operating expenses and improve animal welfare. 60 �B. Objectives Our study objective is to develop and evaluate a trap-like device for mule deer that would automatically attach a radio collar to a ≥6-month-old deer fawn and record the fawn’s weight and sex, without requiring physical restraint or handling of the animal. C. Expected Results or Benefits A passive collaring device, as described above, would allow biologists and researchers to radiocollar, weigh, and identify sex of ≥6-month-old mule deer fawns with minimal expense and labor when compared to traditional mule deer capture techniques. Such a technique would significantly reduce stress that is typically associated with capture and handling and would eliminate capture-related mortality. We do not expect our collaring device to replace other capture techniques. Rather, we expect the device to provide biologists and researchers with an efficient, cost-effective technique to mark a portion of their targeted fawn samples, thereby keeping helicopter net-gunning requirements and associated costs at viable levels. D. Approach 1. Device Specifications We identified an array of specifications to guide design of the automated collaring device, which we divided into 3 categories: 1) collaring device, 2) radio collar, and 3) controls. Collaring device refers to the overall trap-like device and its various components. Our radio collar specifications reflect 6month-old fawn radio collars that are currently used by CDOW. Our intent was to avoid design of a more costly radio collar and to ensure that biologists and researchers could use radio collars readily available on the market without making substantive changes. If radio collar costs increased significantly, the automated collaring device would fail to be cost-effective and have much less utility to biologists and researchers accustomed to using helicopter net-gunning. We were less concerned about cost of the collaring device because it would be a one-time expense that would support repeated fawn captures. Our third specification category, controls, refers to those aspects of the device requiring automation. Collaring Device 1. Device remotely attaches radio collar around the neck of a ≥6-month-old deer fawn; most ≥6month-old fawns range in size from 50−100 lbs. 2. Device deters adult deer or other larger animals from entering but does not deter entry of fawns. 3. Device allows fawns to easily exit in multiple directions at any time. 4. Device must not cause injury to animals. 5. Device incorporates a place for bait, which will lure the animals to the device. 6. The collapsed device should fit in the back of a typical full-size pickup truck. 7. Device should be of a generalized design that could be modified in the future to target different ages and species of animals (e.g., adult deer, calf elk, adult elk, lamb sheep, adult sheep, etc.) Radio collar 1. Collar accommodates fawn neck sizes ranging from 11 to 16 inches in circumference. 2. Width of collar neckband ranges from 0.5 to 3 inches. 3. Collar sheds from the deer 6−12 months after being placed on the animal using surgical tubing or comparable mechanism that does not increase the overall cost of a radio collar. 4. Use existing radio transmitters that are presently available on the market. Controls 1. Restrict collaring to animals that weigh 47−103 lbs (i.e., guarantee that only fawns receive radio collars). 2. Prevent the same fawn from being collared more than once. 3. Measure and record animal weight. 4. Measure and record animal sex. a. Fawn deer sexing options include: 61 �i. Gonads (most reliable) ii. Antler stubs (less reliable) 5. Obtain photo of captured animal. 2. Device Design Working with engineering students and faculty at Colorado State University, we designed the device in stages using a series of prototypes. For example, we initially constructed the device frame out of cheap material and evaluated it using captive deer at the Foothills Wildlife Research Facility in Fort Collins, CO. We observed deer interactions with the prototype to evaluate device dimensions and placement of the radio collar within the device (Figs. 1, 2). We then modified the prototype accordingly and reevaluated until we were comfortable the dimensions were adequate. Once staged prototype testing was completed, we constructed the various device components using materials we believed were suitable for employing the device in winter field conditions. The device frame was constructed from steel and coated to prevent rust and to lessen wear and tear (Fig. 3). The sides of the device comprise one-gay gates, which prevent entry from outside the device yet allow deer to exit the device at any point they choose. The one-way gates were constructed from aluminum and are being mounted with hinges and springs to allow one-way movement. Deer will enter the device through a 14” x 32” opening in the front of the device; entry dimensions were derived from experience feeding deer fawns in Idaho (G. Scholten, Idaho Department of Fish and Game - retired, personal communication). The radio collar and collaring mechanism will be positioned at the rear of the device and in front of the bait compartment (Fig. 4). To access the bait, a deer will be required to extend its head and neck through an expandable collar in the fully expanded position (Fig. 5). The radio collar was made expandable using springs, which was patterned after an expandable adult buck collar designed by Michael Sirochman (Colorado Division of Wildlife, personal communication). The springs prevent the collar from being too loose on a small fawn while not being too tight on a large fawn. Expandable fawn collars are not a new concept and have been commonly used elsewhere on 6-month-old fawns and are sold by telemetry companies. The floor of the device will comprise a scale to estimate the animal’s weight. The animal’s weight will be correctly recorded no matter where the animal stands within the device. A door will close and prevent access to the collaring mechanism/bait compartment if an animal is heavier than 103 lbs, which will allow us to target fawns and prevent older deer from sticking their head through the expanded collar. To be collared, a deer must extend its head through the collar and nudge a joystick positioned in the center of the bait container. The collar will not release unless an animal is heavier than 43 lbs (and less than 103 lbs), which will prevent small animals that may access the bait from triggering the collar. When the joystick is moved and the animal is in the correct weight range, a solenoid will be activated that causes the collar to release around the deer’s neck (Fig. 5). To prevent double-collaring, radio frequency identification (RFID) tags will be attached to all fawn collars. An antenna will be positioned around the opening of the device and connected to an RFID reader. When a previously collared fawn enters the device, the RFID reader will detect the tag and cause the door to the collaring mechanism/bait compartment to close. Digital cameras will be positioned in several locations in the device to photograph the animal when the collar is released. We are currently in the process of assembling the various device components. Once fully assembled and operational, we will evaluate the device with captive deer at FWRF. As necessary, we will make modifications or adjustments to the device until it meets all of our specifications listed above. 62 �3. Field Testing We will evaluate the device with free-ranging deer after we have confirmed the device is working correctly with captive deer. Initially, we will evaluate the device under close supervision in the Fort Collins area to record deer interactions with the trap and to document any problems we may have failed to anticipate. We will be on-site during this initial field testing and we will secure the device entry to prevent access when we’re not present. This will allow us to directly observe how animals interact with the device and to free any animals if there is a problem. If there is a problem, we will use a pole or rod to simultaneously pull back the bars forming the one-way gates on the sides of the trap to encourage the animal to exit and/or to assist the animal with exiting. In the unlikely event we were to seriously injure an animal or kill an animal, we would cease the field study and go back to the design phase to address the problem that caused the animal harm. Animals will be released from the device with functioning radio collars and will be monitored one week post-collaring and every few weeks thereafter. Collars will have surgical tubing between the transmitter and the springs, thereby allowing the collar to drop-off when the surgical tubing degrades. We are using surgical tubing because it is the standard technique used to collar 6-month-old fawns in Colorado, and thus we want to test deployment of collars that will actually be used with this device. However, we will use a knife to make small cuts in the surgical tubing to cause the collars to shed from the animals within a few months of being deployed. Once we have radio-collared several fawns successively without incident and confirmed the device is working correctly, we will begin more widespread testing. During November-December 2009, we will employ ≥1 devices on mule deer winter range to capture fawns as part of ongoing research (Anderson and Freddy 2008). We will document whether the collars cause any ill effects to fawns during the field evaluations by following up on fawns and evaluating whether any mortalities might be related to collaring. We will record numbers of fawns successfully radio-collared and measured relative to personhours expended setting and moving the device. We will then contrast costs and efficiency with other fawn capture techniques. Finally, we will project the cost-savings over a 10-year period associated with using the device for 3 weeks on each deer research and management study in Colorado. It is highly unlikely that an animal would require euthanasia in this study because we will not restrain animals and animals will be able to readily exit the collaring device in any of 3 directions. However, if a deer were to suffer a broken leg, back, neck, pelvis, or other similar wound, it will be euthanized by deep anesthesia with the drug combination of ketamine or Telazol© and xylazine (IV or IM) with dosage based on estimated weight, followed by intravenous administration of KCl (~350 mg KCl/ml sterile water, dosed at >50 mg KCl/kg estimated body mass). In situations where administration of KCl is not feasible, then euthanasia will be performed via a gunshot to the head. E. Location We will conduct all evaluations with captive deer at the FWRF in Fort Collins, CO. We will conduct limited evaluations with free-ranging deer near Fort Collins in north-central Colorado and extensive field evaluations in the Piceance Basin in northwest Colorado. Anderson and Freddy (2008) provided a detailed description of winter range study sites where 6-month-old fawn mule deer will be captured in the Piceance Basin. 63 �F. Schedule Of Work Activity Date Complete Initial Device Specifications Design and Evaluate Prototypes of Device Components Assemble and Evaluate Prototype Device with Captive Deer Initial Evaluation of Device with Free-Ranging Deer Set up Contract with Professional Engineering Firm Complete Design Requirements and Fabricate Working Device Extensive Evaluation of Device with Free-Ranging Deer Prepare Final Report Submit Manuscript to JWM for Publication Sept 2008 Sept 2008−Feb 2009 Mar 2009 Mar−Apr 2009 July−Aug 2009 Sept−Dec 2009 Dec 2009−Feb 2010 Mar−Apr 2010 May−July 2010 G. Estimated Costs Category Item or Position FY 08-09 FY 09-10 Personnel Chad Bishop 0.06 PFTE 0.06 PFTE Dan Walsh 0.06 PFTE 0.04 PFTE Mat Alldredge 0.03 PFTE 0.01 PFTE Eric Bergman 0.03 PFTE 0.01 PFTE Chuck Anderson 0.00 PFTE 0.03 PFTE Device Design and Fabrication $9,000 $22,000 Field Evaluations $1,000 $3,000 Operating H. Related Federal Projects Our research will be conducted on federal (i.e., BLM, USFS) and state lands. The study does not involve formal collaboration with any federal agencies, nor does the work duplicate any ongoing federal projects. I. Literature Cited Anderson, C. R., and D. J. Freddy. 2008. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Study Plan, Colorado Division of Wildlife, Fort Collins, USA. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108−114. Beasom, S. L., W. Evans, and L. Temple. 1980. The drive net for capturing western big game. Journal of Wildlife Management 44:478−480. Clover, M. R. 1956. Single-gate deer trap. California Fish and Game 42:199−201. Ramsey, C. W. 1968. A drop-net deer trap. Journal of Wildlife Management 32:187−190. Schmidt, R. L., W. H. Rutherford, and F. M. Bodenham. 1978. Colorado bighorn sheep-trapping techniques. Wildlife Society Bulletin 6:159−163. 64 �van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408−421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. Webb, S. L., J. S. Lewis, D. G. Hewitt, M. W. Hellickson, and F. C. Bryant. 2008. Assessing the helicopter and net gun as a capture technique for white-tailed deer. Journal of Wildlife Management 72:310−314. Wolfe, L. L., M. W. Miller, and E. S. Williams. 2004. Feasibility of “test-and-cull” for managing chronic wasting disease in urban mule deer. Wildlife Society Bulletin 32:500−505. Figure 1. Prototype evaluation of collar and bait placement, and validation that a deer would extend its head and neck through an expanded collar to access the bait. 65 �Figure 2. Prototype evaluation of entrance and cage dimensions with captive deer. Figure 3. Device frame. The sides of the device will comprise one-way gates that prevent entry to the device yet allow animal to easily exit once inside. Animals will be required to enter the device through a 14” x 32” opening in the front. The rear portion of the device is a bait compartment fabricated from steel. A door on the rear of the bait compartment will allow biologists to easily add bait in the field. 66 �Figure 4. The bait compartment. Deer will be required to extend their head and neck through an outstretched expandable radio collar in order to reach the bait. Figure 5. Radio collar in fully expanded position situated at the entry to the bait compartment. Clear plexi-glass will be placed on either side of the collar to prevent deer from accessing the bait from the side yet will allow visibility. When activated, a solenoid positioned at the top of the collaring device pushes a lever that releases the collar. 67 �Colorado Division of Wildlife July 2009 − June 2010 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 3001 4 Federal Aid Project No. W-185-R : : : : : Division of Wildlife Mammals Research Deer Conservation Development of an Automated Device for Collaring and Weighing Mule Deer Fawns Period Covered: July 1, 2009 − June 30, 2010 Authors: C. J. Bishop, D. P. Walsh, M. W. Alldredge, E. J. Bergman, and C. R. Anderson. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We designed and produced a trap-like device for mule deer that would automatically attach a radio collar to a ≥6-month-old fawn and record the fawn’s weight and sex, without requiring physical restraint or handling of the animal. Our passive collaring device is designed to allow biologists and researchers to radio-collar, weigh, and identify sex of ≥6-month-old mule deer fawns with minimal expense and labor when compared to traditional mule deer capture techniques. This technique should significantly reduce stress that is typically associated with capture and handling and eliminate capturerelated mortality. We collaborated with students and faculty in the Mechanical Engineering Department at Colorado State University to produce a conceptual model and early prototype. We then worked with professional engineers at Dynamic Group Circuit Design in Fort Collins, Colorado, to produce a fullyfunctional prototype of the device. We will conduct an extensive field evaluation of the device with freeranging mule deer during 2010-11. 93 �WILDLIFE RESEARCH REPORT DEVELOPMENT OF AN AUTOMATED DEVICE FOR COLLARING AND WEIGHING MULE DEER FAWNS CHAD J. BISHOP, DANIEL P. WALSH, MATHEW W. ALLDREDGE, ERIC J. BERGMAN, AND CHUCK R. ANDERSON P. N. OBJECTIVE To develop and evaluate a trap-like device for mule deer that would automatically attach a radio collar to a ≥6-month-old deer fawn and record the fawn’s weight and sex, without requiring physical restraint or handling of the animal. SEGMENT OBJECTIVES 1. Work with a professional engineering firm to produce a fully-functional prototype of an automated collaring device for ≥6-month-old mule deer fawns. INTRODUCTION The Colorado Division of Wildlife (CDOW) captures and radio-marks 6-month-old mule deer (Odocoileus hemionus) fawns each year to support research and management of mule deer. Approximately 240 deer fawns are captured annually to monitor survival among 4 populations distributed across western Colorado and an additional 100−350 deer fawns are captured as part of ongoing research studies. Other state agencies in the western United States capture large numbers of mule deer fawns annually also. Most capture is accomplished with net-guns fired from helicopters (Barrett et al. 1982, van Reenen 1982, Webb et al. 2008), which is becoming increasingly expensive (i.e., >$500 per captured deer). Also, net gunning is inherently dangerous with a small market, which at times limits availability of contractors. Drop nets (Ramsey 1968, Schmidt et al. 1978), clover traps (Clover 1956), drive nets (Beasom et al. 1980), and darting (Wolfe et al. 2004) are used occasionally in the western United States to capture deer, but these techniques can be time consuming and labor intensive. Many biologists lack time and resources given other job requirements to conduct such capture operations for any length of time. The increasing cost of helicopter net-gun capture coupled with increasing demand for capturing and radio-collaring 6-month-old fawns has created a need for another capture alternative. Specifically, there is need for a capture technique that is relatively inexpensive to employ considering both operating and personnel costs. In response to CDOW’s capture needs, we conceived the idea of an automated marking device for ≥6-month-old deer fawns that would attach a radio collar and record weight and sex without physically restraining the animal or requiring handling. The idea of automatically attaching radio transmitters to animals is not new, although to our knowledge, there are no proven methods or devices for use on deer or other ungulates. Even a relatively expensive trap or device (e.g., >$5,000 ea.) would reduce CDOW’s capture costs assuming the device could be reused over time with few maintenance expenses. Such a device would enable seasonal wildlife technicians or graduate students to radio-collar samples of deer fawns independently or with little assistance from researchers and biologists because no animal handling would be required. We want the device to record weight and sex because these variables are useful covariates in survival analyses and are typically measured when fawns are captured and handled. A passive marking device would minimize animal stress associated with capture and should have virtually no potential to cause capture-related mortality. The large-mammal capture techniques described 94 �above place considerable, temporary stress on animals as part of netting and handling. Roughly 2-3% of animals typically die from capture-related injuries or stresses under routine capture conditions. Thus, successful development of a passive marking system would reduce CDOW’s operating expenses and improve animal welfare. Therefore, our objective is to design, produce, and evaluate a fully-functional prototype of an automated collaring device for ≥6-month-old mule deer fawns. STUDY AREA We conducted all evaluations with captive deer at the FWRF in Fort Collins, Colorado. We conducted limited evaluations with free-ranging deer near Fort Collins in north-central Colorado. We plan to conduct extensive field evaluations with free-ranging deer in north-central Colorado and elsewhere in Colorado once a fully-functioning device is produced. METHODS We initially wrote a study plan and identified detailed device specifications to guide development of the automated collaring device. We approached Colorado State University’s Mechanical Engineering Department to discuss their interest in helping design such a device. In result, the collaring device became a senior design project for 6 CSU engineering students during the 2008-09 school year. We met with the students weekly and provided them a materials budget of $10,000 to produce a prototype device. We conducted staged evaluations of device components during the year by working with captive deer at FWRF. We also conducted limited evaluations with free-ranging deer near the end of the year. Field evaluations focused primarily on how deer utilized and interacted with the device to guide subsequent design and development decisions. We documented utilization and interactions using direct observation and motion-sensor digital cameras. We relied exclusively on digital cameras when we were not on-site during an evaluation. Automation of the collaring device was disabled any time we were not present to prevent any potential harm to deer. Following preliminary field evaluations, we refined our design specifications and developed a contract with Dynamic Group Circuit Design (DGCD), located in Fort Collins, Colorado, to produce a fully-functional prototype device. We routinely met with electrical engineers from DGCD, and a mechanical engineer subcontracted by DGCD, during the course of the year. These meetings ensured that our device specifications were being satisfactorily met from both engineering and deer biology perspectives. RESULTS AND DISCUSSION We produced a fully-functional prototype device that met our design specifications as set forth in the contract. The prototype device comprises an aluminum cage attached to a bait compartment. Deer enter the device through an adjustable opening at the front of the cage. The adjustable opening can be used to deter entry of larger animals by adjusting both width and height. The sides of the cage comprise one-way gates that prevent entry into the device but allow an animal to exit the device at any point. The bait compartment is accessed through an opening positioned at the rear of the cage. An expandable radio collar is placed in this opening by extending it around four rectangular, aluminum plates that hold the collar in the fully-expanded position (Fig. 1). Radio collars are made expandable by attaching springs to each end of the transmitter; that is, springs are used in place of belting on standard radio collars. Clear plexiglass separates the cage from the bait compartment to maximize visibility. A deer is able to extend its head and neck through the expanded radio collar positioned in the rear opening to access the bait in the bait compartment, which is the only access point to the bait (i.e., it cannot be reached by an animal outside of the device). The floor of the cage is a scale that continuously records weight and informs device operation. Only animals in a specified weight range can be collared, which allows the user to 95 �target fawns and avoid collaring adult deer. Specifically, the mechanism that releases the collar around a deer’s neck will not trigger when an animal is too heavy or too light. Also, an actuator moves a plexiglass plate into the space between the rear cage opening and the bait pan, preventing animals outside of the weight range from accessing the bait. Shortly after a non-target animal exits the device, the collar release mechanism is once again ready to fire and the actuator lowers the plexiglass plate so that the bait is accessible. To prevent an animal from being collared twice, a loop antenna is placed around the entrance to the cage and connected to a radio frequency identification (RFID) reader. All collars used with the device include a small RFID transponder sewn into the collar material. If a previously-collared fawn enters the cage, the RFID transponder is detected, which in turn prevents the collar from being released and activates the actuator to block access to the bait. If a deer enters the cage that is in the specified weight range and has not been previously collared, the collar will release around the deer’s neck once it accesses the bait. The collar release is triggered when a deer’s head breaks an infrared beam positioned immediately above the bait pan. The collar is released by activating a solenoid, which in turn releases a lever that causes the upper 2 aluminum plates holding the expanded collar in place to collapse (Figs. 2 and 3). The collar is then situated around the deer’s neck. When the collar is released, 2 different cameras are immediately activated to take a series of 3 photographs each. One camera is positioned in the back of the bait compartment and set to take a closeup photo of the top of the deer’s head. The second camera is positioned in the floor of the cage and set to take a photo of the deer’s abdomen and groin. These cameras are activated only when a collar is released and facilitate determination of deer sex. Last, when a collar is released, the device records and stores the weight of the deer. An external computer can be hooked up to the device to change program settings, remotely operate the device, and upload weight data. The device is powered by a 12 volt battery that must be recharged every 2-3 days assuming continuous operation. DGCD prepared a user’s manual that explains device operation and detailed schematics to allow future production. We will evaluate effectiveness of the device in the field during 2010-11. Initially, we will only set the device with a collar when we are present and able to directly observe deer interactions with the device. After collaring 5-10 animals in this manner and troubleshooting any problems with the device, we will set the device to operate remotely without an observer on-site, which is how it is intended to be used. SUMMARY We developed a fully-functional prototype of an automated collaring device for mule deer in collaboration with professional engineers. The automated collaring device is designed to allow biologists and researchers to radio-collar portions of their deer samples with minimal time and expense because no animal handling is required and deer can be collared at any time. Primary time commitments include baiting sites, moving device(s) among sites, and adding collars to the devices. The collaring device should also have distinct benefits for studies in urban environments by providing a non-invasive technique for collaring deer. The collaring device should significantly reduce stress that is typically associated with capture and handling and there should be no capture-related mortality. We also have designed the collaring device so that it should be relatively easy to adjust to target adult deer and other ungulate species. Last, the collaring device should have wide applicability for ungulate researchers and managers beyond Colorado. We will be evaluating the device in the field with free-ranging mule deer during the coming year and making additional modifications as necessary. 96 �LITERATURE CITED Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108−114. Beasom, S. L., W. Evans, and L. Temple. 1980. The drive net for capturing western big game. Journal of Wildlife Management 44:478−480. Clover, M. R. 1956. Single-gate deer trap. California Fish and Game 42:199−201. Ramsey, C. W. 1968. A drop-net deer trap. Journal of Wildlife Management 32:187−190. Schmidt, R. L., W. H. Rutherford, and F. M. Bodenham. 1978. Colorado bighorn sheep-trapping techniques. Wildlife Society Bulletin 6:159−163. van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408−421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. Webb, S. L., J. S. Lewis, D. G. Hewitt, M. W. Hellickson, and F. C. Bryant. 2008. Assessing the helicopter and net gun as a capture technique for white-tailed deer. Journal of Wildlife Management 72:310−314. Wolfe, L. L., M. W. Miller, and E. S. Williams. 2004. Feasibility of “test-and-cull” for managing chronic wasting disease in urban mule deer. Wildlife Society Bulletin 32:500−505. Prepared by _______________________ Chad J. Bishop, Mammals Research Leader 97 �Figure 1. View of the radio collar and bait compartment of an automated collaring device for mule deer. To reach bait, deer must extend their head and neck through the expanded radio collar. 98 �Figure 2. View of the collar release mechanism in an automated collaring device for mule deer. 99 �Figure 3. Female mule deer fawn accessing bait by extending her head through an expanded radiocollar. The prototype device will be evaluated extensively in the field with free-ranging deer during 2010-11. 100 �Colorado Division of Parks and Wildlife July 2010 − June 2011 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 3001 4 Federal Aid Project No. W-185-R : : : : : Division of Parks and Wildlife Mammals Research Deer Conservation Development of an Automated Device for Collaring and Weighing Mule Deer Fawns Period Covered: July 1, 2010 − June 30, 2011 Authors: C. J. Bishop, M. W. Alldredge, D. P. Walsh, E. J. Bergman, and C. R. Anderson. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We designed and produced a trap-like device for mule deer that would automatically attach a radio collar to a ≥6-month-old fawn and record the fawn’s weight and sex, without requiring physical restraint or handling of the animal. Our passive collaring device is designed to allow biologists and researchers to radio-collar, weigh, and identify sex of ≥6-month-old mule deer fawns with minimal expense and labor when compared to traditional mule deer capture techniques. This technique should significantly reduce stress that is typically associated with capture and handling and eliminate capturerelated mortality. We collaborated with students and faculty in the Mechanical Engineering Department at Colorado State University to produce a conceptual model and early prototype. We then worked with professional engineers at Dynamic Group Circuit Design in Fort Collins, Colorado, to produce a fullyfunctional prototype of the device. We conducted an extensive field evaluation of the device with freeranging mule deer during winter 2010-11. We successfully collared, weighed, and identified sex of 6 different mule deer fawns across 4 winter range locations along Colorado’s northern Front Range. Collars were purposefully made to shed from deer within several weeks or months of being collared. Two fawns were successfully re-collared after they shed the first collars they received. Thus, we observed 8 successful collaring events involving 6 different fawns. Most fawns demonstrated minimal response to collaring events, either remaining in the device or calmly exiting. Certain components of the collaring device failed to function optimally when temperatures dropped below approximately −15° C, while other components did not adequately withstand mule deer use under field conditions. Also, certain behaviors of mule deer when approaching and using the device created circumstances where it was possible to collar the same animal twice, which happened on one occasion. We identified a series of device modifications that would be necessary to address these various issues. During 2011-12, we will modify the device accordingly and conduct a follow-up field evaluation. 85 �WILDLIFE RESEARCH REPORT DEVELOPMENT OF AN AUTOMATED DEVICE FOR COLLARING AND WEIGHING MULE DEER FAWNS CHAD J. BISHOP, MATHEW W. ALLDREDGE, DANIEL P. WALSH, ERIC J. BERGMAN, AND CHARLES R. ANDERSON, JR. P. N. OBJECTIVE To develop and evaluate a trap-like device for mule deer that will automatically attach a radio collar to a ≥6-month-old deer fawn and record the fawn’s weight and sex, without requiring physical restraint or handling of the animal. SEGMENT OBJECTIVES 7. Evaluate effectiveness and functionality of an automated collaring device for collaring, weighing, and identifying sex of mule deer fawns during winter under free-ranging conditions. INTRODUCTION Colorado Parks and Wildlife (CPW) captures and radio-marks 6-month-old mule deer (Odocoileus hemionus) fawns each year to support research and management of mule deer. Approximately 240−300 deer fawns are captured annually to monitor survival among 4−5 populations distributed across western Colorado and an additional 100−350 deer fawns are captured as part of ongoing research studies. Other state agencies in the western United States capture large numbers of mule deer fawns annually also. Most capture is accomplished with net-guns fired from helicopters (Barrett et al. 1982, van Reenen 1982, Webb et al. 2008), which is becoming increasingly expensive (i.e., >$500 per captured deer). Also, net gunning is inherently dangerous with a small market, which at times limits availability of contractors. Drop nets (Ramsey 1968, Schmidt et al. 1978), clover traps (Clover 1956), drive nets (Beasom et al. 1980), and darting (Wolfe et al. 2004) are used occasionally in the western United States to capture deer, but these techniques can be time consuming and labor intensive. Many biologists lack time and resources given other job requirements to conduct such capture operations for any length of time. The increasing cost of helicopter net-gun capture coupled with increasing demand for capturing and radio-collaring 6-month-old fawns has created a need for another capture alternative. Specifically, there is need for a capture technique that is relatively inexpensive to employ considering both operating and personnel costs. In response to CPW’s capture needs, we conceived the idea of an automated marking device for ≥6-month-old deer fawns that would attach a radio collar and record weight and sex without physically restraining the animal or requiring handling. The idea of automatically attaching radio transmitters to animals is not new, although to our knowledge, there are no proven methods or devices for use on deer or other ungulates. Even a relatively expensive trap or device (e.g., >$5,000 ea.) would reduce CPW’s capture costs assuming the device could be reused over time with few maintenance expenses. Such a device would enable seasonal wildlife technicians or graduate students to radio-collar samples of deer fawns independently or with little assistance from researchers and biologists because no animal handling would be required. We want the device to record weight and sex because these variables are useful covariates in survival analyses and are typically measured when fawns are captured and handled. A passive marking device would minimize animal stress associated with capture and should have virtually no potential to cause capture-related mortality. The large-mammal capture techniques described 86 �above place considerable, temporary stress on animals as part of netting and handling. Roughly 2-3% of animals typically die from capture-related injuries or stresses under routine capture conditions. Thus, successful development of a passive marking system would reduce CPW’s operating expenses and improve animal welfare. Therefore, we designed, produced, and evaluated an automated device for collaring, weighing, and identifying sex of mule deer fawns during winter under free-ranging conditions. STUDY AREA We conducted all evaluations with captive deer at the Foothills Wildlife Research Facility (FWRF) in Fort Collins, Colorado. We conducted field evaluations with free-ranging deer at 5 sites along Colorado’s northern Front Range: 1) Horsetooth Reservoir, west of Fort Collins, private land 2) Masonville, southwest of Fort Collins, private land, 3) Red Feather, northwest of Fort Collins, private land, 4) Hall Ranch, west of Lyons, Boulder County Parks and Open Space, and 5) Heil Valley Ranch, southwest of Lyons, Boulder County Parks and Open Space. We plan to conduct additional field evaluations with free-ranging deer in northwest Colorado during 2011-12. METHODS We identified detailed specifications to guide the design and development of an automated collaring device and sought assistance from Colorado State University’s Mechanical Engineering Department. The collaring device became a senior design project for 6 CSU engineering students during the 2008-09 school year. We met with the students weekly and provided them a materials budget of $10,000 to produce a prototype device. We conducted staged evaluations of device components during the year by working with captive deer at FWRF. We also conducted limited evaluations with freeranging deer during spring 2009. Field evaluations focused primarily on how deer utilized and interacted with the device to guide subsequent design and development decisions. We documented utilization and interactions using direct observation and motion-sensor digital cameras. We relied exclusively on digital cameras when we were not on-site during an evaluation. Automation of the collaring device was disabled any time we were not present to prevent any potential harm to deer. Following preliminary field evaluations, we refined our design specifications and developed a contract with Dynamic Group Circuit Design (DGCD), located in Fort Collins, Colorado, to produce a fully-functional prototype device. We routinely met with electrical engineers from DGCD, and a mechanical engineer subcontracted by DGCD, during 2009-10. These meetings ensured that our device specifications were being satisfactorily met from both engineering and deer biology perspectives. Working with DGCD, we produced a fully-functional prototype device in 2010 that met our design specifications as set forth in the contract. The prototype device comprises an aluminum cage attached to a bait compartment (Fig. 1). Deer enter the device through an adjustable opening at the front of the cage. The adjustable opening can be used to deter entry of larger animals by adjusting both width and height. The sides of the cage comprise one-way gates that prevent entry into the device but allow an animal to exit the device at any point. The bait compartment is accessed through an opening positioned at the rear of the cage. An expandable radio collar is placed in this opening by extending it around four rectangular, aluminum plates that hold the collar in the fully-expanded position (Fig. 2). Radio collars are made expandable by attaching springs to each end of the transmitter; that is, springs are used in place of belting on standard radio collars. Clear plexiglass separates the cage from the bait compartment to maximize visibility. A deer is able to extend its head and neck through the expanded radio collar positioned in the rear opening to access the bait in the bait compartment, which is the only access point to the bait (i.e., it cannot be reached by an animal outside of the device). The floor of the cage is a scale that continuously records weight and informs device operation. Only animals in a specified weight range can be collared, which allows the user to 87 �target fawns and avoid collaring adult deer. Specifically, the mechanism that releases the collar around a deer’s neck will not trigger when an animal is too heavy or too light. Also, an actuator moves a plexiglass plate into the space between the rear cage opening and the bait pan, preventing animals outside of the weight range from accessing the bait. Shortly after a non-target animal exits the device, the collar release mechanism is once again able to be released (when triggered) and the actuator lowers the plexiglass plate so that the bait is accessible. To prevent an animal from being collared twice, a loop antenna is placed around the entrance to the cage and connected to a radio frequency identification (RFID) reader. All collars used with the device include a small RFID transponder sewn into the collar material. If a previously-collared fawn enters the cage, the RFID transponder is detected, which in turn prevents the collar from being released and activates the actuator to block access to the bait. If a deer enters the cage that is in the specified weight range and has not been previously collared, the collar will release around the deer’s neck once it accesses the bait. The collar release is triggered when a deer’s head breaks an infrared beam positioned immediately above the bait pan. The collar is released by activating a solenoid, which in turn releases a lever that causes the upper 2 aluminum plates holding the expanded collar in place to collapse (Figs. 3 and 4). The collar is then situated around the deer’s neck. When the collar is released, 2 different cameras are immediately activated to take a series of 3 photographs each. One camera is positioned in the back of the bait compartment and set to take a closeup photo of the top of the deer’s head. The second camera is positioned in the floor of the cage and set to take a photo of the deer’s abdomen and groin. These cameras are activated only when a collar is released and facilitate determination of deer sex. Last, when a collar is released, the device records and stores the weight of the deer. An external computer can be hooked up to the device to change program settings, remotely operate the device, and upload weight data. The device is powered by a 12 volt battery that must be recharged every 2-3 days assuming continuous operation. DGCD prepared a user’s manual that explains device operation and detailed schematics to allow future production. We evaluated effectiveness of the device in the field during winter 2010-11. Initially, we only set the device with a collar in place when we were present and able to directly observe deer interactions with the device. After collaring several animals in this manner and troubleshooting problems with the device, we set the device to operate remotely without an observer on-site, which is how it was intended to be used. RESULTS AND DISCUSSION We began baiting sites at Horsetooth Reservoir and Masonville on October 21, 2010, to attract deer for evaluating the device. We baited sites with alfalfa hay, apple pulp, dried fruit, and cereal. We baited several other sites briefly but discontinued baiting due to lack of deer use. Deer immediately responded to bait at Horsetooth Reservoir and began accessing the bait daily. On October 26, we placed the collaring device on site and began encouraging deer to walk into the device by placing bait on the scale inside the cage. On October 29, we documented a deer accessing the bait pan within the bait compartment for the first time. In the following weeks, we continued to periodically document deer entering the device and accessing the bait pan, although malfunctioning of the device prevented deer from being collared. One malfunction occurred because an electrical signal emitted from a camera placed at the entry of the device interfered with the RFID reader, which ultimately prevented fawns from being collared. It took roughly a week to diagnose the problem, which was corrected by simply removing the camera from the entry of the device. This particular camera was not wired into the device and was not critical to device functioning. We deemed that this camera was unnecessary and would be more useful if placed approximately 5 meters away from the trap to better document deer use and behavior. A second malfunction occurred because the scale did not have adequate support underneath and touched the 88 �ground, thereby giving inaccurate weight readings, which also prevented deer from being collared. We corrected this particular problem by welding an aluminum frame to better support the scale. Once these problems were corrected and other adjustments were made, we remotely collared our first fawn (female) on November 17, 2010. The fawn showed little reaction to the collaring event, calmly exiting the trap shortly after receiving the collar. The fawn’s weight and sex were successfully recorded. Sex was positively confirmed based on a photograph of the fawn’s head taken by the camera positioned in the bait compartment. We continued to monitor the device at Horsetooth Reservoir because there were adequate numbers of uncollared fawns in the area. However, we continued to encounter various problems with the device that affected functionality. Most notably, the collar release mechanism began failing to release the collar when a fawn was in position. We quickly determined that device controls were working properly and that an electrical signal was successfully being sent to the solenoid when an uncollared fawn was in the proper position accessing the bait. The source of the problem was a mechanical failing associated with the release mechanism itself. When an expanded collar was in place (i.e., in a fully-expanded state), the tension of the collar sometimes prevented the release lever from moving enough to release the aluminum plates holding the collar in position. Once aware of the problem, we began making adjustments to the release mechanism to improve its functionality. Another problem we identified was that fawns were placing their front hooves on a piece of metal trim at the front of the cage when accessing the bait, which led to inaccurate weight readings and missed opportunities to collar fawns. We corrected this problem by placing a plastic shield above the metal trim so that deer could no longer place hooves on the metal trim. Following this modification, the entire floor surface of the cage comprised only the scale. We also noted that small fawns accessing the bait sometimes failed to break the infrared beam extending across the center of the bait pan, thereby failing to be collared. Thus, we adjusted the positioning of the bait pan to make sure that fawns successfully broke the infrared beam when accessing the bait, regardless of size. Once these changes were made, we successfully collared two more fawns (1 male and 1 female) on successive days, December 13 and 14, 2010. Also, the female fawn that was collared on November 17 shed its collar on December 13 and was successfully recollared on December 20. On December 21, the actuator that opens and closes the bait door short-circuited in response to cold, snowy weather and damaged the circuit board that controls operation of the device. The actuator was positioned such that moisture could enter it. The moisture, in combination with cold temperatures, caused the failure. It became evident at this point that future device modifications would likely require a heavier-duty actuator. However, until a new actuator could be researched, tested, and installed, DGCD used the same actuator and positioned it differently so that it was less likely to take on moisture. DGCD also replaced the circuit board to restore functionality of the collaring device. Several weeks were required to make these modifications, causing the device to be inoperable from December 21, 2010, through January 15, 2011. On January 20, we recollared the female fawn that was initially collared on December 14 (it shed the first collar on January 13). We then moved the device to the Masonville bait site on January 21, after documenting 5 successful collaring events at Horsetooth Reservoir. The Masonville bait site was regularly visited by 4 bucks, 3 does, and 2 fawns. The fawns were aggressively chased by the 4 bucks once we put the collaring device in place and restricted the amount of bait available outside of the collaring device. We solved this problem by creating a separate bait site for the bucks a short distance away. It took one week before the fawns at Masonville became comfortable entering the collaring device and accessing the bait in the bait pan. We did not put a collar in place initially because we speculated that the fawns would be more likely to access the bait pan for the first time if they were not required to extend their head through the collar. Once one of the fawns became acclimated and we put a collar in the device, the bait door/actuator began malfunctioning again, preventing the fawn from being collared. The malfunctioning was apparently related to cold temperatures. The bait door/actuator began functioning correctly again several days later and we collared 89 �a male fawn on February 4, 2011. The only other fawn on site showed no interest in accessing the bait in the bait pan during the ensuing week. Thus, we stopped baiting the site on February 12 and moved the device to the Red Feather site on February 14. Several of the gate arms that prevent deer entry into the sides of the device had been damaged by deer over the course of the winter. During February 14−20, as deer became accustomed to the collaring device, we replaced all gate arms with a new, more durable hinge system. We then resumed normal operations and collared our 7th fawn (female) on February 27, 2011. Unfortunately, the RFID reader failed to detect this collared fawn the following day, allowing the fawn to receive a second collar on February 28. We suspended collaring efforts for several days evaluating the RFID failure. It became evident that if a collared fawn entered the device quickly, it could go undetected by the RFID reader. This issue was already understood as a potential problem, but this was the first time a fawn was actually double-collared. We documented no ill effects of the second collar on the fawn. Realizing the odds of a double-collaring event were low, we resumed collaring efforts on approximately March 6. Incidentally, the odds of the double-collared fawn receiving a third collar were essentially zero because the fawn now had two RFID transponders. We made note that the RFID problem would need to be resolved with a device modification during the following year. The other couple of fawns routinely visiting the site were reluctant to access the bait pan. On March 17, we moved the collaring device to the Heil Valley Ranch site on Boulder County Parks and Open Space land. Deer regularly visiting the Heil site included 4 bucks, 2 does, and 1 fawn. We were unable to keep the bucks from being aggressive toward the does and fawn around the collaring device, which prevented the fawn from entering the device. In response, we moved the device to the Hall Ranch bait site on March 24, 2011, where 3-4 bucks, 2-3 does, and 1-3 fawns were using the site. Deer acclimated quickly to the collaring device and we collared our 8th fawn on March 28th, immediately after placing the collar in the device. A few days later we concluded the field evaluation because weather was turning warm, green forage was abundant, and bears were coming out of hibernation. During our field evaluation, we documented a number of issues with the collaring device that need resolved in subsequent design modifications: • The solenoid release mechanism occasionally failed to release the collar even when the solenoid was triggered. We plan to evaluate an alternative release mechanism that uses an archery caliper release instead of the existing metal, latch system. • We documented several scenarios that could allow a fawn to receive a second collar. First, if a collared fawn extends its head through the entry to the device and is detected by the RFID reader but fails to move forward onto the scale for ≥30 seconds, the bait door will move back into the open position. Second, if a collared fawn is on the scale for >15 minutes (i.e., beds down on the scale), the scale will rezero and the door will move back into the open position. At this point another fawn could step into the device, which would indicate a correct weight range, and the collared fawn could receive a second collar if it then accessed the bait. Third, as we directly witnessed, if a collared fawn enters the device quickly, it is possible the RFID reader could fail to detect the RFID transponder in the fawn’s collar. These scenarios, albeit unlikely, can be corrected by changing the device programming and increasing sensitivity of the RFID reader/antenna. • The actuator that controls the bait door commonly malfunctioned in cold temperatures (i.e., ≤ −12 °C). We intend for the device to be fully functional at −32 °C. We plan to research other actuators and evaluate them under controlled temperature settings. A number of actuators are available on the market that meet our temperature specifications, but they range in cost from <$100 to >$1000. The actuator we evaluated was the cheapest available and did not meet its 90 �• stated specifications. Our intent is to find the cheapest actuator that will hold up under field conditions. The camera mounted on the floor of the device commonly failed to provide useful images for identifying sex. We therefore plan to remove the floor-mounted camera. In contrast, the camera in the bait compartment positioned to take pictures of a fawn’s head provided conclusive evidence of sex. The only needed adjustment is to more securely attach the “head camera” to the bait compartment. Working with DGCD, we will research and implement the necessary device modifications to address these issues. We plan to incorporate the design modifications during summer-fall 2011 and conduct a follow-up field evaluation during winter 2011-12. SUMMARY We developed a fully-functional prototype of an automated collaring device for mule deer in collaboration with professional engineers. The automated collaring device is designed to allow biologists and researchers to radio-collar portions of their deer samples with minimal time and expense because no animal handling is required and deer can be collared at any time. Primary time commitments include baiting sites, moving device(s) among sites, and adding collars to the devices. The collaring device should also have distinct benefits for studies in urban environments by providing a non-invasive technique for collaring deer. We successfully collared 6 different fawns during Nov−Mar, 2011−12, along Colorado’s northern Front Range. We recollared 2 of these fawns after they shed their initial collars, resulting in 8 successful collaring events. Fawns generally showed minimal reaction to being collared. It was evident that fawns did not experience the type of stress that is associated with typical capture and handling techniques. We documented a number of functional issues with the collaring device, which we plan to resolve through design modifications during summer-fall 2011. We then plan to conduct a follow-up field evaluation during winter 2011-12. LITERATURE CITED Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108−114. Beasom, S. L., W. Evans, and L. Temple. 1980. The drive net for capturing western big game. Journal of Wildlife Management 44:478−480. Clover, M. R. 1956. Single-gate deer trap. California Fish and Game 42:199−201. Ramsey, C. W. 1968. A drop-net deer trap. Journal of Wildlife Management 32:187−190. Schmidt, R. L., W. H. Rutherford, and F. M. Bodenham. 1978. Colorado bighorn sheep-trapping techniques. Wildlife Society Bulletin 6:159−163. van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408−421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. Webb, S. L., J. S. Lewis, D. G. Hewitt, M. W. Hellickson, and F. C. Bryant. 2008. Assessing the helicopter and net gun as a capture technique for white-tailed deer. Journal of Wildlife Management 72:310−314. Wolfe, L. L., M. W. Miller, and E. S. Williams. 2004. Feasibility of “test-and-cull” for managing chronic wasting disease in urban mule deer. Wildlife Society Bulletin 32:500−505. Prepared by _______________________ Chad J. Bishop, Mammals Research Leader 91 �Figure 1. Automated collaring device for mule deer, comprising an aluminum cage and a bait compartment. Deer become collared by entering the cage and extending their head through an expanded radio collar when accessing bait. 92 �Figure 2. View of the radio collar and bait compartment of an automated collaring device for mule deer. To reach bait, deer must extend their head and neck through the expanded radio collar. 93 �Figure 3. View of the collar release mechanism in an automated collaring device for mule deer. 94 �Figure 4. Female mule deer fawn accessing bait by extending her head through an expanded radiocollar. 95 �Colorado Division of Parks and Wildlife July 2011 − June 2012 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 3001 4 Federal Aid Project No. W-185-R : : : : : Division of Parks and Wildlife Mammals Research Deer Conservation Development of an Automated Device for Collaring and Weighing Mule Deer Fawns Period Covered: July 1, 2011 − June 30, 2012 Authors: C. J. Bishop, M. W. Alldredge, D. P. Walsh, E. J. Bergman, and C. R. Anderson. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We designed and produced a trap-like device for mule deer that would automatically attach a radio collar to a ≥6-month-old fawn and record the fawn’s weight and sex, without requiring physical restraint or handling of the animal. Our passive collaring device is designed to allow biologists and researchers to radio-collar, weigh, and identify sex of ≥6-month-old mule deer fawns with minimal expense and labor when compared to traditional mule deer capture techniques. This technique should significantly reduce stress that is typically associated with capture and handling and eliminate capturerelated mortality. We collaborated with students and faculty in the Mechanical Engineering Department at Colorado State University to produce a conceptual model and early prototype. We then worked with professional engineers at Dynamic Group Circuit Design in Fort Collins, Colorado, to produce a fullyfunctional prototype of the device. We conducted an extensive field evaluation of the device with freeranging mule deer during October-March, 2010-11, and January-March, 2012. We successfully collared, weighed, and identified sex of 6 different mule deer fawns across 4 winter range locations along Colorado’s northern Front Range during winter 2010-11. Collars were purposefully made to shed from deer within several weeks or months of being collared. Two fawns were successfully re-collared after they shed the first collars they received. Thus, we observed 8 successful collaring events involving 6 different fawns in 2010-11. Most fawns demonstrated minimal response to collaring events, either remaining in the device or calmly exiting. We successfully collared, weighed, and identified sex of 2 different mule deer fawns in the Piceance Basin of northwest Colorado during February-March 2012. We collared fewer fawns in winter 2011-12 than the previous winter in part because of a shortened evaluation period (i.e., 3 instead of 6 months). Winter conditions were mild overall during 2011-12, which likely contributed to the lower collaring rate since deer had ample foraging options and may not have been as strongly attracted to bait. During 2010-11, certain components of the collaring device failed to function optimally when temperatures dropped below approximately −15° C, while other components did not adequately withstand mule deer use under field conditions. Also, certain behaviors of mule deer when approaching and using the device created circumstances where it was possible to collar the same animal twice, which happened on one occasion. We incorporated a series of device modifications during summer-fall 2011 necessary to address these various issues. The device functioned well under field 80 �conditions during January-March 2012, indicating the modifications were effective. Our automated collaring device allowed mule deer fawns to be remotely collared, weighed, and sexed with minimal or no stress to the animals. However, fawns typically required one or more weeks of exposure to the device before they entered and accessed the bait. This slow acclimation period limited utility of the device when compared to traditional capture techniques used to collar fawns. Future work will focus on additional device modifications and altered baiting strategies that decrease fawn acclimation period, and in turn, increase collaring rates. 81 �WILDLIFE RESEARCH REPORT DEVELOPMENT OF AN AUTOMATED DEVICE FOR COLLARING AND WEIGHING MULE DEER FAWNS CHAD J. BISHOP, MATHEW W. ALLDREDGE, ERIC J. BERGMAN, DANIEL P. WALSH, AND CHARLES R. ANDERSON, JR. P. N. OBJECTIVE To develop and evaluate a trap-like device for mule deer that will automatically attach a radio collar to a ≥6-month-old deer fawn and record the fawn’s weight and sex, without requiring physical restraint or handling of the animal. SEGMENT OBJECTIVES 1. Evaluate effectiveness and functionality of an automated collaring device for collaring, weighing, and identifying sex of mule deer fawns during winter under free-ranging conditions. INTRODUCTION Colorado Parks and Wildlife (CPW) captures and radio-marks 6-month-old mule deer (Odocoileus hemionus) fawns each year to support research and management of mule deer. Approximately 240−300 fawns are captured annually to monitor survival among 4−5 populations distributed across western Colorado and an additional 100−350 fawns are captured as part of ongoing research studies. Other state agencies in the western United States capture large numbers of mule deer fawns annually also. Most capture is accomplished with net-guns fired from helicopters (Barrett et al. 1982, van Reenen 1982, Webb et al. 2008), which is becoming increasingly expensive (i.e., >$500 per captured deer). Also, net gunning is inherently dangerous with a small market, which at times limits availability of contractors. Drop nets (Ramsey 1968, Schmidt et al. 1978), clover traps (Clover 1956), drive nets (Beasom et al. 1980), and darting (Wolfe et al. 2004) are used occasionally in the western United States to capture deer, but these techniques can be time consuming and labor intensive. Many biologists lack time and resources given other job requirements to conduct such capture operations for any length of time. The increasing cost of helicopter net-gun capture coupled with increasing demand for capturing and radio-collaring 6-month-old fawns has created a need for another capture alternative. Specifically, there is need for a capture technique that is relatively inexpensive to employ considering both operating and personnel costs. In response to CPW’s capture needs, we conceived the idea of an automated marking device for ≥6-month-old deer fawns that would attach a radio collar and record weight and sex without physically restraining the animal or requiring handling. The idea of automatically attaching radio transmitters to animals is not new, although to our knowledge, there are no proven methods or devices for use on deer or other ungulates. Even a relatively expensive trap or device (e.g., >$5,000 ea.) would reduce CPW’s capture costs assuming the device could be reused over time with few maintenance expenses. Such a device would enable seasonal wildlife technicians or graduate students to radio-collar samples of deer fawns independently or with little assistance from researchers and biologists because no animal handling would be required. We want the device to record weight and sex because these variables are useful covariates in survival analyses and are typically measured when fawns are captured and handled. A passive marking device would minimize animal stress associated with capture and should not cause capture-related mortality. The large-mammal capture techniques described above place 82 �considerable, temporary stress on animals as part of netting and handling. Roughly 2-3% of animals typically die from capture-related injuries or stresses under routine capture conditions. Thus, successful development of a passive marking system would reduce CPW’s operating expenses and improve animal welfare. Therefore, we designed, produced, and evaluated an automated device for collaring, weighing, and identifying sex of mule deer fawns during winter under free-ranging conditions. STUDY AREA We worked with captive deer at the Foothills Wildlife Research Facility (FWRF) in Fort Collins, Colorado, when designing the device and evaluating initial prototypes. We conducted subsequent evaluations of the collaring device with free-ranging deer in various field locations. During 2010-11, we conducted field evaluations with free-ranging deer at 5 sites along Colorado’s northern Front Range: 1) Horsetooth Reservoir, west of Fort Collins, private land 2) Masonville, southwest of Fort Collins, private land, 3) Red Feather, northwest of Fort Collins, private land, 4) Hall Ranch, west of Lyons, Boulder County Parks and Open Space, and 5) Heil Valley Ranch, southwest of Lyons, Boulder County Parks and Open Space. During 2012, we conducted field evaluations with free-ranging deer at Hall Ranch (Jan) and in the Piceance Basin southwest of Meeker, Colorado (Feb-Mar). METHODS We identified detailed specifications to guide the design and development of an automated collaring device and sought assistance from Colorado State University’s Mechanical Engineering Department. The collaring device became a senior design project for 6 CSU engineering students during the 2008-09 school year. We met with the students weekly and provided them a materials budget of $10,000 to produce a prototype device. We conducted staged evaluations of device components during the year by working with captive deer at FWRF. We also conducted limited evaluations with freeranging deer during spring 2009. Field evaluations focused primarily on how deer utilized and interacted with the device to guide subsequent design and development decisions. We documented utilization and interactions using direct observation and motion-sensor digital cameras. We relied exclusively on digital cameras when we were not on-site during an evaluation. Automation of the collaring device was disabled any time we were not present to prevent any potential harm to deer. Following preliminary field evaluations, we refined our design specifications and developed a contract with Dynamic Group Circuit Design (DGCD), Fort Collins, Colorado, to produce a fullyfunctional prototype device. We routinely met with electrical engineers from DGCD, and a mechanical engineer subcontracted by DGCD, during 2009-10. These meetings ensured that our device specifications were being satisfactorily met from both engineering and deer biology perspectives. Working with DGCD, we produced a fully-functional prototype device in 2010 that met our design specifications as set forth in the contract. The prototype device comprises an aluminum cage attached to a bait compartment (Fig. 1). Deer enter the device through an adjustable opening at the front of the cage. The adjustable opening can be used to deter entry of larger animals by adjusting both width and height. The sides of the cage comprise one-way gates that prevent entry into the device but allow an animal to exit the device at any point. The bait compartment is accessed through an opening positioned at the rear of the cage. An expandable radio collar is placed in this opening by extending it around four rectangular, aluminum plates that hold the collar in the fully-expanded position (Fig. 2). Radio collars are made expandable by attaching springs to each end of the transmitter; that is, springs are used in place of belting on standard radio collars. Clear plexiglass separates the cage from the bait compartment to maximize visibility. A deer is able to extend its head and neck through the expanded radio collar positioned in the rear opening to access the bait in the bait compartment, which is the only access point to the bait (i.e., it cannot be reached by an animal 83 �outside of the device). The floor of the cage is a scale that continuously records weight and informs device operation. Only animals in a specified weight range can be collared, which allows the user to target fawns and avoid collaring adult deer. Specifically, the mechanism that releases the collar around a deer’s neck will not trigger when an animal is too heavy or too light. Also, an actuator moves a plexiglass plate into the space between the rear cage opening and the bait pan, preventing animals outside of the weight range from accessing the bait. Shortly after a non-target animal exits the device, the collar release mechanism is once again able to be released (when triggered) and the actuator lowers the plexiglass plate so that the bait is accessible. To prevent an animal from being collared twice, a loop antenna is placed around the entrance to the cage and connected to a radio frequency identification (RFID) reader. All collars used with the device include a small RFID transponder sewn into the collar material. If a previously-collared fawn enters the cage, the RFID transponder is detected, which in turn prevents the collar from being released and activates the actuator to block access to the bait. If a deer enters the cage that is in the specified weight range and has not been previously collared, the collar will release around the deer’s neck once it accesses the bait. The collar release is triggered when a deer’s head breaks an infrared beam positioned immediately above the bait pan. The collar is released by activating a solenoid, which in turn releases a lever or trigger that causes the upper 2 aluminum plates holding the expanded collar in place to collapse (Figs. 3 and 4). The collar is then situated around the deer’s neck. In 2011, we replaced the release lever with an archery caliper release in an attempt to improve the release mechanism. When the collar is released, 2 different cameras are immediately activated to take a series of 3 photographs each. One camera is positioned in the back of the bait compartment and set to take a close-up photo of the top of the deer’s head. The second camera is positioned in the floor of the cage and set to take a photo of the deer’s abdomen and groin. These cameras are activated only when a collar is released and facilitate determination of deer sex. In 2011, we removed the floor camera after determining it was not necessary or effective for identifying deer sex. Last, when a collar is released, the device records and stores the weight of the deer. An external computer can be hooked up to the device to change program settings, remotely operate the device, and upload weight data. The device is powered by a 12 volt battery that must be recharged every 2-3 days assuming continuous operation. DGCD prepared a user’s manual that explains device operation and detailed schematics to allow future production. We evaluated effectiveness of the device in the field during October-March 2010-11 and JanuaryMarch 2012. Initially, we only set the device with a collar in place when we were present and able to directly observe deer interactions with the device. After collaring several animals in this manner and troubleshooting problems with the device, we set the device to operate remotely without an observer onsite, which is how it was intended to be used. RESULTS AND DISCUSSION 2010-2011 Field Evaluation We began baiting sites at Horsetooth Reservoir and Masonville on October 21, 2010, to attract deer for evaluating the device. We baited sites with alfalfa hay, apple pulp, dried fruit, and cereal. We baited several other sites briefly but discontinued baiting due to lack of deer use. Deer immediately responded to bait at Horsetooth Reservoir and began accessing the bait daily. On October 26, we placed the collaring device on site and began encouraging deer to walk into the device by placing bait on the scale inside the cage. On October 29, we documented a deer accessing the bait pan within the bait compartment for the first time. In the following weeks, we continued to periodically document deer entering the device and accessing the bait pan, although malfunctioning of the device prevented deer from being collared. One malfunction occurred because an electrical signal emitted from a camera placed at the entry of the device interfered with the RFID reader, which ultimately prevented fawns from being 84 �collared. It took roughly a week to diagnose the problem, which was corrected by simply removing the camera from the entry of the device. This particular camera was not wired into the device and was not critical to device functioning. We deemed that this camera was unnecessary and would be more useful if placed approximately 5 meters away from the trap to better document deer use and behavior. A second malfunction occurred because the scale did not have adequate support underneath and touched the ground, thereby giving inaccurate weight readings, which also prevented deer from being collared. We corrected this particular problem by welding an aluminum frame to better support the scale. Once these problems were corrected and other adjustments were made, we remotely collared our first fawn (female) on November 17, 2010. The fawn showed little reaction to the collaring event, calmly exiting the trap shortly after receiving the collar. The fawn’s weight and sex were successfully recorded. Sex was positively confirmed based on a photograph of the fawn’s head taken by the camera positioned in the bait compartment. We continued to monitor the device at Horsetooth Reservoir because there were adequate numbers of uncollared fawns in the area. However, we continued to encounter various problems with the device that affected functionality. Most notably, the collar release mechanism began failing to release the collar when a fawn was in position. We quickly determined that device controls were working properly and that an electrical signal was successfully being sent to the solenoid when an uncollared fawn was in position accessing the bait. The source of the problem was a mechanical failing associated with the release mechanism itself. When an expanded collar was in place (i.e., in a fully-expanded state), the tension of the collar sometimes prevented the release lever from moving enough to release the aluminum plates holding the collar in position. Once aware of the problem, we began making adjustments to the release mechanism to improve its functionality. Another problem we identified was that fawns were placing their front hooves on a piece of metal trim at the front of the cage when accessing the bait, which led to inaccurate weight readings and missed opportunities to collar fawns. We corrected this problem by placing a plastic shield above the metal trim so that deer could no longer place hooves on the metal trim. Following this modification, the entire floor surface of the cage comprised only the scale. We also noted that small fawns accessing the bait sometimes failed to break the infrared beam extending across the center of the bait pan, thereby failing to be collared. Thus, we adjusted the positioning of the bait pan to make sure that fawns successfully broke the infrared beam when accessing the bait, regardless of size. Once these changes were made, we successfully collared two more fawns (1 male and 1 female) on successive days, December 13 and 14, 2010. Also, the female fawn that was collared on November 17 shed its collar on December 13 and was successfully recollared on December 20. On December 21, the actuator that opens and closes the bait door short-circuited in response to cold, snowy weather and damaged the circuit board that controls operation of the device. The actuator was positioned such that moisture could enter it. The moisture, in combination with cold temperatures, caused the failure. It became evident at this point that future device modifications would likely require a heavier-duty actuator. However, until a new actuator could be researched, tested, and installed, DGCD used the same actuator and positioned it differently so that it was less likely to take on moisture. DGCD also replaced the circuit board to restore functionality of the collaring device. Several weeks were required to make these modifications, causing the device to be inoperable from December 21, 2010, through January 15, 2011. On January 20, we recollared the female fawn that was initially collared on December 14 (it shed the first collar on January 13). We then moved the device to the Masonville bait site on January 21, after documenting 5 successful collaring events at Horsetooth Reservoir. The Masonville bait site was regularly visited by 4 bucks, 3 does, and 2 fawns. The fawns were aggressively chased by the 4 bucks once we put the collaring device in place and restricted the amount of bait available outside of the collaring device. We solved this problem by creating a separate bait site for the bucks a short distance away. It took one week before the fawns at Masonville became comfortable entering the collaring device and accessing the bait in the bait pan. We did not put a collar in place 85 �initially because we speculated that the fawns would be more likely to access the bait pan for the first time if they were not required to extend their head through the collar. Once one of the fawns became acclimated and we put a collar in the device, the bait door/actuator began malfunctioning again, preventing the fawn from being collared. The malfunctioning was apparently related to cold temperatures. The bait door/actuator began functioning correctly again several days later and we collared a male fawn on February 4, 2011. The only other fawn on site showed no interest in accessing the bait in the bait pan during the ensuing week. Thus, we stopped baiting the site on February 12 and moved the device to the Red Feather site on February 14. Several of the gate arms that prevent deer entry into the sides of the device had been damaged by deer over the course of the winter. During February 14−20, as deer became accustomed to the collaring device, we replaced all gate arms with a new, more durable hinge system. We then resumed normal operations and collared our 7th fawn (female) on February 27, 2011. Unfortunately, the RFID reader failed to detect this collared fawn the following day, allowing the fawn to receive a second collar on February 28. We suspended collaring efforts for several days evaluating the RFID failure. It became evident that if a collared fawn entered the device quickly, it could go undetected by the RFID reader. We were aware of this potential problem, but this was the first time it actually occurred. We documented no ill effects of the second collar on the fawn. Realizing the odds of a double-collaring event were low, we resumed collaring efforts on approximately March 6. Incidentally, the odds of the double-collared fawn receiving a third collar were essentially zero because the fawn now had two RFID transponders. We made note that the RFID problem would need to be resolved with a device modification during the following year. The other couple of fawns routinely visiting the site were reluctant to access the bait pan. On March 17, we moved the collaring device to the Heil Valley Ranch site on Boulder County Parks and Open Space land. Deer regularly visiting the Heil site included 4 bucks, 2 does, and 1 fawn. We were unable to keep the bucks from being aggressive toward the does and fawn around the collaring device, which prevented the fawn from entering the device. In response, we moved the device to the Hall Ranch bait site on March 24, 2011, where 3-4 bucks, 2-3 does, and 1-3 fawns were using the site. Deer acclimated quickly to the collaring device and we collared our 8th fawn on March 28th, immediately after placing the collar in the device. A few days later we concluded the field evaluation because weather was turning warm, green forage was abundant, and bears were coming out of hibernation. 2011 Device Modifications During our 2010-11 winter field evaluation, we documented a number of issues with the collaring device that needed resolved. During summer-fall 2011, working with DGCD, we made several modifications to the device to address these issues. • Issue: The solenoid release mechanism occasionally failed to release the collar even when the solenoid was triggered. Modification: We evaluated and incorporated an alternative release mechanism that used an archery caliper release instead of the existing metal, latch system. • Issue: We documented several scenarios that could allow a fawn to receive a second collar. First, if a collared fawn extended its head through the entry to the device and was detected by the RFID reader but failed to move forward onto the scale for ≥30 seconds, the bait door moved back into the open position. Second, if a collared fawn was on the scale for >15 minutes (i.e., bedded down on the scale), the scale re-zeroed and the door moved back into the open position. At this point another fawn could step into the device, which would indicate a correct weight range, and the collared fawn could receive a second collar if it then accessed the bait. Third, as we directly witnessed, if a collared fawn entered the device quickly, the RFID reader sometimes failed to detect the RFID transponder in the fawn’s collar. Modifications: We resolved these issues by reprogramming the device and increasing sensitivity of the RFID reader/antenna. 86 �• • Issue: The actuator that controls the bait door commonly malfunctioned in cold temperatures (i.e., ≤ −12 °C). We intend for the device to be fully functional at −32 °C. Modification: We researched other actuators and selected a higher quality unit that would be more likely to perform adequately under the desired conditions. We then evaluated the actuator under controlled temperature settings in a freezer to confirm functionality before installation in the collaring device. Issue: The camera mounted on the floor of the device commonly failed to provide useful images for identifying sex. The camera in the bait compartment positioned to take pictures of a fawn’s head provided conclusive evidence of sex, indicating the floor camera was unnecessary. Modification: We removed the floor-mounted camera from the device and eliminated the associated wiring and programming. 2012 Field Evaluation We made considerable progress evaluating and subsequently modifying the collaring device during 2010-2011, and therefore, we believed that a 3-month evaluation period during January-March 2012 would be sufficient for a follow-up field evaluation. We initially evaluated the collaring device during January 2012 at Hall Ranch near Lyons, Colorado. We began evaluating the collaring device on January 10. Unfortunately, the bait sites were visited primarily by adult males, limiting any opportunities to collar fawns. These adult males also appeared to prevent regular attendance at the bait site/collaring device by adult females and fawns that were in the area. The problem of adult males dominating a bait site is not unique to this study and has been documented over time when attempting to capture fawns where bait is used to attract animals to a trap (Colorado Parks and Wildlife, unpublished data). Given the challenges posed by adult males near Lyons, we moved the collaring device at the end of January to Piceance Basin, southwest of Meeker, Colorado, where a separate deer study was underway and could benefit from additional collared fawns. We also believed deer densities would be higher near our bait sites in Piceance Basin than our sites along Colorado’s northern Front Range, potentially offering more opportunities to collar fawns. We initiated our evaluation of the collaring device in Piceance Basin on February 5, 2012. We were unable to monitor the collaring device in the field on a daily basis given the distance from the original study area along the northern Front Range. Our first monitoring period occurred during February 5-10. We collared a male fawn weighing 66 lbs on February 7 during early evening. During February 1118, we baited the collaring device but did not monitor deer activity on site. We resumed direct field monitoring of the device during February 19-24. During this time, we consistently observed a mixture of does, fawns, and bucks on site but did not successfully collar a fawn. The previously collared fawn was routinely on site and often entered the collaring device. During February 25-March 3, we again baited the collaring device but did not conduct field observations. We resumed field monitoring during March 4-9. Deer consistently accessed the bait site during this period, typically with a group size of 6-7 deer that included 3-4 fawns. On March 9, we collared a female fawn weighing 63 lbs. We once again ceased direct field monitoring during March 10-18 and completed our final monitoring period during March 1921. Deer were active on site during this final evaluation period, including the previously collared fawns, although we were unable to collar any new fawns. We then ceased our evaluation of the collaring device for 2012, having collared 2 fawns during an approximately 1.5-month evaluation period in Piceance Basin. Our modifications to the collaring device in 2011 appeared to have improved functionality of the device. The only problem we documented during our field evaluation in 2012 was that the bait door often remained open when the first collared fawn reentered the device on subsequent occasions. While initially of concern, the bait door always closed when tested with other collars. Additionally, the bait door closed each time the second collared fawn reentered the device after having been collared. Thus, we concluded 87 �there was a problem with the RFID transponder placed in the first collar rather than a problem with the collaring device itself. Although the device functioned well, the rate at which deer were collared was particularly slow. Our field observations indicate that fawns typically required one or more weeks of exposure before they entered the device and accessed the bait in the bait pan. Some fawns were reluctant to enter the device even after days or weeks of exposure. We tried various baiting strategies in an attempt to maximally encourage fawns to enter the device to access bait in the bait pan. If bait were only placed in the bait pan inside the device, deer groups were not attracted and retained at the site, and therefore, no fawns were present. If too much bait were placed outside the device, there was no incentive for fawns to enter the device and extend their head through the expanded collar to access bait in the bait pan. Generally speaking, we learned that some bait has to be placed outside the device to attract deer groups to the site and that some bait should be placed on the floor/scale to lure fawns into the device. We also tried placing all bait external to the device in buckets to limit the number of animals accessing bait at any given time and to make them more accustomed to placing their heads in an enclosed area to obtain the bait. However, our observations did not suggest this technique was any more effective at encouraging fawns to enter the device and extend their heads through the expanded collar to access bait in the pan. Winter weather conditions were overall mild during our study, particularly during winter 2011-12, which may partly explain the slow rate at which fawns were collared. During winter conditions exhibiting greater snow depths and lower temperatures with less forage available, we would expect fawns to have greater incentive to enter the device and extend their head through the expanded collar to access bait. SUMMARY We developed a fully-functional prototype of an automated collaring device for mule deer in collaboration with professional engineers. The automated collaring device is designed to allow biologists and researchers to radio-collar portions of their deer samples with minimal time and expense because no animal handling is required and deer can be collared at any time. Primary time commitments include baiting sites, moving the device(s) among sites, and adding collars to the device. The collaring device should also have distinct benefits for studies in urban environments by providing a non-invasive technique for collaring deer. We successfully collared 6 different fawns during Nov−Mar, 2011−12, along Colorado’s northern Front Range. We recollared 2 of these fawns after they shed their initial collars, resulting in 8 successful collaring events. Fawns generally showed minimal reaction to being collared. It was evident that fawns did not experience the type of stress that is associated with typical capture and handling techniques. We documented a number of functional issues with the collaring device in 2010-11, which we resolved through design modifications during summer-fall 2011. We conducted a follow-up field evaluation during January-March 2012 and collared 2 additional fawns during February and March in Piceance Basin. The largest drawback of the collaring device is the slow rate at which fawns were collared. Fawns typically required one or more weeks of exposure to the device before fully entering the device and extending their head through the expanded collar to access bait in the bait pan. This slow acclimation period limited utility of the device when compared to traditional capture techniques used to collar fawns. In the future, additional design modifications or more clever baiting strategies will be necessary to improve collaring rates. We also plan to evaluate placement of ≥2 collaring devices at the same site once a second collaring device is produced. With more collaring devices, potentially less bait would need to be placed external to the devices and deer might be more inclined to access bait in the bait pans within the collaring devices. 88 �LITERATURE CITED Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108−114. Beasom, S. L., W. Evans, and L. Temple. 1980. The drive net for capturing western big game. Journal of Wildlife Management 44:478−480. Clover, M. R. 1956. Single-gate deer trap. California Fish and Game 42:199−201. Ramsey, C. W. 1968. A drop-net deer trap. Journal of Wildlife Management 32:187−190. Schmidt, R. L., W. H. Rutherford, and F. M. Bodenham. 1978. Colorado bighorn sheep-trapping techniques. Wildlife Society Bulletin 6:159−163. van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408−421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. Webb, S. L., J. S. Lewis, D. G. Hewitt, M. W. Hellickson, and F. C. Bryant. 2008. Assessing the helicopter and net gun as a capture technique for white-tailed deer. Journal of Wildlife Management 72:310−314. Wolfe, L. L., M. W. Miller, and E. S. Williams. 2004. Feasibility of “test-and-cull” for managing chronic wasting disease in urban mule deer. Wildlife Society Bulletin 32:500−505. Prepared by _______________________ Chad J. Bishop, Mammals Research Leader 89 �Figure 1. Automated collaring device for mule deer, comprising an aluminum cage and a bait compartment. Deer become collared by entering the cage and extending their head through an expanded radio collar when accessing bait. 90 �Figure 2. View of the radio collar and bait compartment of an automated collaring device for mule deer. To reach bait, deer must extend their head and neck through the expanded radio collar. 91 �Figure 3. View of the collar release mechanism in an automated collaring device for mule deer. 92 �Figure 4. Female mule deer fawn accessing bait by extending her head through an expanded radiocollar. 93 �Colorado Parks and Wildlife July 2012 – June 2013 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Colorado 3430 3001 8 Federal Aid Project No. W-185-R : : : : : Parks and Wildlife Mammals Research Deer Conservation Development of an Automated Device for Collaring and Weighing Mule Deer Fawns Period Covered: July 1, 2012 – June 30, 2013 Authors: C. J. Bishop, M. W. Alldredge, D. P. Walsh, E. J. Bergman, and C. R. Anderson. Cooperators: Mechanical Engineering Department, Colorado State University, Michael Sirochman, Veterinarian Technician, Colorado Parks and Wildlife, John Broderick, Senior Terrestrial Biologist, Colorado Parks and Wildlife. Lisa L. Wolfe, Veterinarian, Colorado Parks and Wildlife, Michael W. Miller, Wildlife Health Leader, Colorado Parks and Wildlife, Stewart Breck, Research Wildlife Biologist, National Wildlife Research Center All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We designed and produced a trap-like device for mule deer that would automatically attach a radio collar to a ≥6-month-old fawn and record the fawn’s weight and sex, without requiring physical restraint or handling of the animal. Our passive collaring device is designed to allow biologists and researchers to radio-collar, weigh, and identify sex of ≥6-month-old mule deer fawns with minimal expense and labor when compared to traditional mule deer capture techniques. This technique should significantly reduce stress that is typically associated with capture and handling and eliminate capturerelated mortality. We collaborated with students and faculty in the Mechanical Engineering Department at Colorado State University to produce a conceptual model and early prototype. We then worked with professional engineers at Dynamic Group Circuit Design in Fort Collins, Colorado, to produce a fullyfunctional prototype of the device. We conducted an extensive field evaluation of the device with free-ranging mule deer during OctoberMarch, 2010-11, and January-March, 2012. We successfully collared, weighed, and identified sex of 6 different mule deer fawns across 4 winter range locations along Colorado’s northern Front Range during winter 2010-11. Collars were purposefully made to shed from deer within several weeks or months of being collared. Two fawns were successfully re-collared after they shed the first collars they received. Thus, we observed 8 successful collaring events involving 6 different fawns in 2010-11. Most fawns demonstrated minimal response to collaring events, either remaining in the device or calmly exiting. We successfully collared, weighed, and identified sex of 2 different mule deer fawns in the Piceance Basin of northwest Colorado during February-March 2012. We collared fewer fawns in winter 2011-12 than the 57 �previous winter in part because of a shortened evaluation period (i.e., 3 instead of 6 months). Winter conditions were mild overall during 2011-12, which likely contributed to the lower collaring rate since deer had ample foraging options and may not have been as strongly attracted to bait. During 2010-11, certain components of the collaring device failed to function optimally when temperatures dropped below approximately −15° C, while other components did not adequately withstand mule deer use under field conditions. Also, certain behaviors of mule deer when approaching and using the device created circumstances where it was possible to collar the same animal twice, which happened on one occasion. We incorporated a series of device modifications during summer-fall 2011 necessary to address these various issues. The device functioned well under field conditions during January-March 2012, indicating the modifications were effective. Our automated collaring device allowed mule deer fawns to be remotely collared, weighed, and sexed with minimal or no stress to the animals. However, fawns typically required one or more weeks of exposure to the device before they entered and accessed the bait. This slow acclimation period limited utility of the device when compared to traditional capture techniques used to collar fawns. During 2012-13, focus was on additional device modifications and altered baiting strategies that decrease fawn acclimation period, and in turn, increase collaring rates. 58 �WILDLIFE RESEARCH REPORT DEVELOPMENT OF AN AUTOMATED DEVICE FOR COLLARING AND WEIGHING MULE DEER FAWNS CHAD J. BISHOP, MATHEW W. ALLDREDGE, ERIC J. BERGMAN, DANIEL P. WALSH, AND CHARLES R. ANDERSON, JR. OBJECTIVE Colorado Parks and Wildlife (CPW) captures and radio-marks 6-month-old mule deer (Odocoileus hemionus) fawns each year to support research and management of mule deer. Approximately 240 deer fawns are captured annually to monitor survival among 4 populations distributed across western Colorado and an additional 100−350 deer fawns are captured as part of ongoing research studies. Other state agencies in the western United States capture large numbers of mule deer fawns annually also. Most capture is accomplished with net-guns fired from helicopters (Barrett et al. 1982, van Reenen 1982, Webb et al. 2008), which is becoming increasingly expensive (i.e., >$500 per captured deer). Also, net gunning is inherently dangerous with a small market, which at times limits availability of contractors. Drop nets (Ramsey 1968, Schmidt et al. 1978), clover traps (Clover 1956), drive nets (Beasom et al. 1980), and darting (Wolfe et al. 2004) are used occasionally in the western United States to capture deer, but these techniques can be time consuming and labor intensive. Many biologists lack time and resources given other job requirements to conduct such capture operations for any length of time. The increasing cost of helicopter net-gun capture coupled with increasing demand for capturing and radio-collaring 6-month-old fawns has created a need for another capture alternative. Specifically, there is need for a capture technique that is relatively inexpensive to employ considering both operating and personnel costs. In response to CPW’s capture needs, we conceived the idea of an automated marking device for ≥6-month-old deer fawns that would attach a radio collar and record weight and sex without physically restraining the animal or requiring handling. The idea of automatically attaching radio transmitters to animals is not new, although to our knowledge, there are no proven methods or devices for use on deer or other ungulates. Even a relatively expensive trap or device (e.g., $3,000−5,000 ea.) would reduce CPW’s capture costs assuming the device could be reused over time with few maintenance expenses. Such a device would enable seasonal wildlife technicians or graduate students to radio-collar samples of deer fawns independently or with little assistance from researchers and biologists because no animal handling would be required. We want the device to record weight and sex because these variables are useful covariates in survival analyses and are typically measured when fawns are captured and handled. A passive marking device would minimize animal stress associated with capture and should have virtually no potential to cause capture-related mortality. The large-mammal capture techniques described above place considerable, temporary stress on animals as part of netting and handling. Roughly 2-3% of animals typically die from capture-related injuries or stresses under routine capture conditions. Thus, successful development of a passive marking system would reduce CPW’s operating expenses and improve animal welfare. Our study objective is to develop and evaluate a trap-like device for mule deer that would automatically attach a radio collar to a ≥6-month-old deer fawn and record the fawn’s weight and sex, without requiring physical restraint or handling of the animal. 59 �STUDY AREA We conducted field evaluations with free-ranging deer along Colorado’s Front Range between Boulder and Fort Collins, in the Piceance Basin in northwest Colorado, and on the Uncompahgre Plateau in western Colorado. METHOD Device Specifications We identified an array of specifications to guide design of the automated collaring device, which we divided into 3 categories: 1) collaring device, 2) radio collar, and 3) controls. Collaring device refers to the overall trap-like device and its various components. Our radio collar specifications reflect 6month-old fawn radio collars that are currently used by CPW. Our intent was to avoid design of a more costly radio collar and to ensure that biologists and researchers could use radio collars readily available on the market without making substantive changes. If radio collar costs increased significantly, the automated collaring device would fail to be cost-effective and have much less utility to biologists and researchers accustomed to using helicopter net-gunning. We were less concerned about cost of the collaring device because it would be a one-time expense that would support repeated fawn captures. Our third specification category, controls, refers to those aspects of the device requiring automation. Collaring Device 1. Device remotely attaches radio collar around the neck of a ≥6-month-old deer fawn; most ≥6month-old fawns range in size from 50−100 lbs. 2. Device deters adult deer or other larger animals from entering but does not deter entry of fawns. 3. Device allows fawns to easily exit in multiple directions at any time. 4. Device must not cause injury to animals. 5. Device incorporates a place for bait, which will lure the animals to the device. 6. The collapsed device should fit in the back of a typical full-size pickup truck. 7. Device should be of a generalized design that could be modified in the future to target different ages and species of animals (e.g., adult deer, calf elk, adult elk, lamb sheep, adult sheep, etc.) Radio collar 1. Collar accommodates fawn neck sizes ranging from 11 to 16 inches in circumference. 2. Width of collar neckband ranges from 0.5 to 3 inches. 3. Collar sheds from the deer 6−12 months after being placed on the animal using surgical tubing or comparable mechanism that does not increase the overall cost of a radio collar. 4. Use existing radio transmitters that are presently available on the market. Controls 1. Restrict collaring to animals that weigh 47−103 lbs (i.e., guarantee that only fawns receive radio collars). 2. Prevent the same fawn from being collared more than once. 3. Measure and record animal weight. 4. Measure and record animal sex. a. Fawn deer sexing options include: i. Gonads (most reliable) ii. Antler stubs (less reliable) 5. Obtain photo of captured animal. Device Design Working with engineering students and faculty at Colorado State University, we designed the device in stages using a series of prototypes. For example, we initially constructed the device frame out of cheap material and evaluated it using captive deer at the Foothills Wildlife Research Facility in Fort 60 �Collins, CO. We observed deer interactions with the prototype to evaluate device dimensions and placement of the radio collar within the device (Figs. 1, 2). We then modified the prototype accordingly and reevaluated until we were comfortable the dimensions were adequate. Once staged prototype testing was completed, we constructed the various device components using materials we believed were suitable for employing the device in winter field conditions. The initial device frame was constructed from steel and coated to prevent rust and to lessen wear and tear. We later changed the device frame to aluminum (Fig. 3). The sides of the device comprise one-gay gates, which prevent entry from outside the device yet allow deer to exit the device at any point they choose. The one-way gates were constructed from aluminum and are being mounted with hinges and springs to allow one-way movement. Deer will enter the device through a 14” x 32” opening in the front of the device; entry dimensions were derived from experience feeding deer fawns in Idaho (G. Scholten, Idaho Department of Fish and Game - retired, personal communication). The radio collar and collaring mechanism will be positioned at the rear of the device and in front of the bait compartment. To access the bait, a deer will be required to extend its head and neck through an expandable collar in the fully expanded position (Fig. 4). The radio collar was made expandable using springs, which was patterned after an expandable adult buck collar designed by Michael Sirochman (Colorado Parks and Wildlife, personal communication). The springs prevent the collar from being too loose on a small fawn while not being too tight on a large fawn. Expandable fawn collars are not a new concept and have been commonly used elsewhere on 6-month-old fawns and are sold by telemetry companies. The floor of the device will comprise a scale to estimate the animal’s weight. The animal’s weight will be correctly recorded no matter where the animal stands within the device. A door will close and prevent access to the collaring mechanism/bait compartment if an animal is heavier than 103 lbs, which will allow us to target fawns and prevent older deer from sticking their head through the expanded collar. To be collared, a deer must extend its head through the collar and break an infrared beam positioned immediately above the bait container. The collar will not release unless an animal is heavier than 43 lbs (and less than 103 lbs), which will prevent small animals that may access the bait from triggering the collar. When the IR beam is broken and the animal is in the correct weight range, a solenoid will be activated that causes the collar to release around the deer’s neck (Fig. 4). To prevent double-collaring, radio frequency identification (RFID) tags will be attached to all fawn collars. An antenna will be positioned around the opening of the device and connected to an RFID reader. When a previously collared fawn enters the device, the RFID reader will detect the tag and cause the door to the collaring mechanism/bait compartment to close. Digital cameras will be positioned in several locations in the device to photograph the animal when the collar is released. RESULTS AND BENEFITS A passive collaring device, as described above, would allow biologists and researchers to radiocollar, weigh, and identify sex of ≥6-month-old mule deer fawns with minimal expense and labor when compared to traditional mule deer capture techniques. Such a technique would significantly reduce stress that is typically associated with capture and handling and would eliminate capture-related mortality. We do not expect our collaring device to replace other capture techniques. Rather, we expect the device to provide biologists and researchers with an efficient, cost-effective technique to mark a portion of their targeted fawn samples, thereby keeping helicopter net-gunning requirements and associated costs at viable levels. In winter 2011-12 we completed a second year of field evaluation of a fully-functional prototype device (Figs. 5, 6). During this evaluation, we accumulated hundreds of hours of field observation of mule deer interacting with the device and we noted device components that warranted modification for optimal performance. We incorporated these modifications and conducted a follow-up field evaluation 61 �with free-ranging deer during winter 2012-13 on the Uncompahgre Plateau. We also constructed a second prototype for field testing based on the final design of the first prototype. All animals were released from the device with functioning radio collars and were monitored one week post-collaring and every few weeks thereafter. Collars had surgical tubing between the transmitter and the springs, thereby allowing the collar to drop-off when the surgical tubing degraded. We used surgical tubing because it is the standard technique used to collar 6-month-old fawns in Colorado, and thus we wanted to test deployment of collars that would actually be used with this device. However, we did make small cuts in the surgical tubing to cause the collars to shed from the animals within a few months of being deployed. We designed and fabricated the collaring device in such a manner as to prevent inadvertent collaring of non-target species, thereby preventing any possibility that a threatened, endangered, or candidate species could be harmed. The floor of the collaring device is a scale that continuously records weight and informs the device. The collar can only be released when an appropriately-sized animal is in the device. Animals are attracted to the device with bait, contained in a separate compartment at one end of the device. To access the bait, animals must extend their head and neck through an expanded collar into the bait compartment. The collar can only be released when an animal is accessing the bait, thereby breaking an infrared beam, which further informs the device. We are not familiar with any animals in these study areas that fit the weight range of a deer and could simultaneously access the bait. The only possible animal is a black bear, although it is unlikely the bear could access the bait. However, black bears will be hibernating during the winter months when the collaring device will be employed. Finally, even if a non-target animal accessed the device, there is ample opportunity for the animal to leave the device without being harmed. The sides of the device consist of one-way gates, such that an animal in the device can exit at any time through the entrance or sides. Finally, in the extreme unlikely event that a non-target animal were radio-collared, the expandable collar does not pose a threat to any animal that can fit its head through the expanded collar. The device, therefore, poses no threat to non-target species, including threatened, endangered, and candidate species listed under the Endangered Species Act because none are similar in size or behavior to deer. Also, all travel will occur on established roads throughout the study areas, preventing any chance of damaging a listed plant species. SUMMARY As part of our field evaluation, we recorded numbers of fawns successfully radio-collared and measured relative to person-hours expended setting and moving the device. We planned to contrast costs and efficiency with other fawn capture techniques. However, successful capture of fawns was extremely limited, so at this point other capture techniques would be more efficient. During the final winter of investigation no fawns were collared and only a few actually entered the device. This concludes this project. LITERATURE CITED Anderson, C. R., and D. J. Freddy. 2008. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Study Plan, Colorado Division of Wildlife, Fort Collins, USA. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108−114. Beasom, S. L., W. Evans, and L. Temple. 1980. The drive net for capturing western big game. Journal of Wildlife Management 44:478−480. Clover, M. R. 1956. Single-gate deer trap. California Fish and Game 42:199−201. Ramsey, C. W. 1968. A drop-net deer trap. Journal of Wildlife Management 32:187−190. 62 �Schmidt, R. L., W. H. Rutherford, and F. M. Bodenham. 1978. Colorado bighorn sheep-trapping techniques. Wildlife Society Bulletin 6:159−163. van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408−421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. Webb, S. L., J. S. Lewis, D. G. Hewitt, M. W. Hellickson, and F. C. Bryant. 2008. Assessing the helicopter and net gun as a capture technique for white-tailed deer. Journal of Wildlife Management 72:310−314. Wolfe, L. L., M. W. Miller, and E. S. Williams. 2004. Feasibility of “test-and-cull” for managing chronic wasting disease in urban mule deer. Wildlife Society Bulletin 32:500−505. Prepared by _______________________ Chad J. Bishop, Mammals Research Leader 63 �Figure 1. Prototype evaluation of collar and bait placement, and validation that a deer would extend its head and neck through an expanded collar to access the bait. Figure 2. Prototype evaluation of entrance and cage dimensions with captive deer. 64 �Figure 3. Device frame. The sides of the device comprise one-way gates that prevent entry to the device yet allow animals to easily exit once inside. Animals will be required to enter the device through a 14” x 32” opening in the front, which is adjustable. The rear portion of the device is a bait compartment fabricated from aluminum. A door on the rear of the bait compartment will allow biologists to easily add bait in the field and access controls. Figure 4. The bait compartment. Deer will be required to extend their head and neck through an outstretched expandable radio collar in order to reach the bait. 65 �Figure 5. Mule deer fawn in process of being collared. Figure 6. Mule deer fawn at the moment of the collar being released. 66 � https://cpw.cvlcollections.org/files/original/a917e885d25f935ada35f4868554e1d2.pdf 29d9ebae18e5cd8e0d93069e7080e872 PDF Text Text Colorado Division of Wildlife July 2008 − June 2009 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 3001 4 Federal Aid Project No. W-185-R : : : : : Division of Wildlife Mammals Research Deer Conservation Effectiveness of a Redesigned Vaginal Implant Transmitter in Mule Deer Period Covered: July 1, 2008 − June 30, 2009 Authors: C. J. Bishop, C. R. Anderson, D. P. Walsh, P. Kuechle, J. Roth, and E. J. Bergman. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Our understanding of factors that limit mule deer populations may be improved by evaluating neonatal survival as a function of dam characteristics under free-ranging conditions, which generally requires that both neonates and dams are radiocollared. The only viable technique facilitating capture of neonates from radiocollared adult females is use of vaginal implant transmitters (VITs). To date, VITs have allowed research opportunities that were not possible previously; however, VITs are often expelled from adult females prepartum, which limits their utility. We redesigned an existing vaginal implant transmitter (VIT) manufactured by Advanced Telemetry Systems (ATS) by lengthening and widening wings used to retain the VIT in an adult female. Our objective was to increase VIT retention rates to increase likelihood of locating birth sites and newborn fawns. We placed VITs with modified wings in 59 adult female mule deer and evaluated probability of retention to parturition and probability of locating newborn fawns. Probability of a VIT being expelled during parturition (i.e., success) was 0.766 (SE = 0.0605) and probability of a VIT being expelled ≤3 days prepartum (i.e., partial success) was 0.128 (SE = 0.0477). Thus, probability of a VIT being at least partially successful was 0.894 (SE = 0.0441). Probability of locating at least 1 neonate from successful or partially successful VITs was 0.952 (SE = 0.0333) and probability of locating both fawns from twin litters was 0.588 (SE = 0.0857). We expended approximately 12 person-hours per detected neonate. Our modifications to VIT wings effectively increased VIT retention in mule deer, allowing more neonate fawns to be located per unit cost and effort. Researchers employing VITs with modified wings should require minimal oversampling to offset failures caused by early expulsion. To aid researchers in planning future studies, we developed an equation for determining VIT sample size necessary to achieve a specified sample size of neonates. Our study expands opportunities for conducting research that links adult female attributes to productivity and offspring survival. 69 �WILDLIFE RESEARCH REPORT EFFECTIVENESS OF A REDESIGNED VAGINAL IMPLANT TRANSMITTER FOR CAPTURING MULE DEER NEONATES FROM TARGETED ADULT FEMALES CHAD J. BISHOP, CHUCK R. ANDERSON, DANIEL P. WALSH, PETER KUECHLE, JOHN ROTH, AND ERIC J. BERGMAN P. N. OBJECTIVE To redesign vaginal implant transmitters (VITs) and evaluate their retention in free-ranging mule deer. SEGMENT OBJECTIVES 1. Redesign and manufacture the silicone-covered plastic wings used to retain VITs in deer. 2. Evaluate rates of VIT retention to parturition and fawn capture success using the newly-designed wings in free-ranging mule deer. INTRODUCTION Mule deer (Odocoileus hemionus) fawn production and neonatal survival is influenced by dam characteristics (e.g., body condition, disease status, habitat use). To understand fawn-dam relationships, manipulative field studies are needed that allow fawn production and survival to be estimated as a function of treatments applied to adult females. For example, a study evaluating the effectiveness of winter range habitat treatments on subsequent neonatal survival would require the capture of fawns from marked adult females that verifiably used, or did not use, the habitat treatments the previous winter(s). Such studies depend on a technique that enables newborn fawns to be captured from marked adult females. The most promising technique employed to capture neonates from marked adult females is use of vaginal implant transmitters (VITs), which are placed in the vagina of adult females during early to mid gestation. In theory, adult females retain VITs until parturition, at which point VITs are expelled at birth sites along with newborn fawns. Assuming VITs are routinely monitored, researchers can promptly radio-locate shed VITs and capture the newborn fawns. Recent applications of VITs in white-tailed deer (O. hemionus), black-tailed deer (O. hemionus columbianus), and mule deer have been moderately successful (Bowman and Jacobson 1998, Carstensen et al. 2003, Pamplin 2003, Bishop et al. 2007). Vaginal implant transmitters also permit measurement of fetal survival in free-ranging populations, which has important implications in populations where stillborn mortality occurs (Bishop et al. 2007, 2008, 2009). An additional advantage of using VITs to capture neonates may be a reduction in sample bias when compared to capture techniques that rely on opportunistic fawn capture (White et al. 1972, Ballard et al. 1998, Pojar and Bowden 2004). Opportunistic techniques are susceptible to bias because of unequal capture success among vegetation types, road densities, fawn ages, and stages of fawning. When using VITs, neonate captures should be more random as long as VIT signals are monitored with equal intensity during fawning, and assuming the sample of radio-collared does was captured with minimal bias. Thus, VITs could have broad applicability regardless of whether study objectives require that fawns be captured from previously marked adult females. The most significant problem associated with VITs has been premature expulsion and subsequent failure to locate birth sites or newborn fawns (Bowman and Jacobson 1998, Carstensen et al. 2003, Pamplin 2003, Johnstone-Yellin et al. 2006, Bishop et al. 2007). The VIT has flexible, plastic wings coated with a soft silicone that induce pressure against the vaginal wall to retain the transmitter. The VIT 70 �design facilitates a quick, non-surgical insertion process and is safe for the animal (Johnson et al. 2006), but the current wing design is inadequate with respect to retention. Bishop et al. (2007) found that 43% (SE = 4.7) of VITs in mule deer shed prepartum, although capture success was high when VITs shed only 1−3 days prepartum. More importantly, Bishop et al. (2007) found that 25% (SE = 4.1) of VITs shed >3 days prepartum and that retention probability declined as deer body size increased, indicating the retention wings were too small to be effective in larger deer. Based on these results, considerable oversampling would be required in the design of future projects to achieve a target sample size of fawns. Oversampling is not desirable from an animal care and use perspective or from a cost perspective. Thus, the plastic-silicone retention wings of VITs need to be redesigned to allow maximum retention in deer. To date, the wings used to retain VITs have been purchased from a company in New Zealand (Carter Holt Harvey Plastic Products, Hamilton, New Zealand) that originally produced them for an application in the livestock industry (Bowman and Jacobson 1998). The company manufactured 1 large wing and 1 small wing; the former has been used to produce VITs for bison (Bison bison) and elk (Cervus elaphus) whereas the latter has been used to produce VITs for deer (Advanced Telemetry Systems, Isanti, MN). Advanced Telemetry Systems (ATS), in cooperation with wildlife researchers, made an initial effort in 2004 to lengthen the retention wings by adding resin to the wing tips. Using these VITs and with antennas cut to the appropriate length, S. P. Haskell (Texas Tech University, unpublished data) reported that 81% of VITs (n = 21) in deer were retained until parturition. Although retention improved, this aftermarket modification was not ideal. The modified wing tips were hard because of the resin addition and thus not ideal for placement in the vaginal canal. Also, there remained a need to further increase retention rate. We therefore developed a study plan (Appendix A), redesigned retention wings of VITs used in deer and similar-sized ungulates, fabricated a new production mold, and evaluated retention rates of VITs in free-ranging mule deer. STUDY AREA We conducted our research in Piceance Basin and on Roan Plateau in northwest Colorado (Fig. 1). Our winter range study area comprised 4 study units distributed across much of the Piceance Basin. The 4 units ranged in size from 70 to 130 km2 and are referenced as Magnolia, Story-Sprague, Ryan Gulch, and Yellow Creek (Fig. 2). These study units are part of a larger research study evaluating effects of natural gas development and mitigation on mule deer (Anderson and Freddy 2008). Winter range habitat comprised predominantly pinyon pine (Pinus edulis) and Utah juniper (Juniperus osteosperma) and secondarily big sagebrush (Artemisia tridentata), serviceberry (Amelanchier utahensis), mountain mahogany (Cercocarpus montanus), bitterbrush (Purshia tridentata), and rabbitbrush (Chrysothamnus spp.). Drainage bottoms were characterized by stands of big sagebrush, saltbush (Atriplex spp.), and black greasewood (Sarcobatus vermiculatus), with the majority of the primary drainage bottoms having been converted to irrigated, grass hay fields. Elevations ranged from 1860 m at Piceance Creek in Ryan Gulch to 2280 m in Yellow Creek and Story-Sprague. Our summer range study area comprised roughly 1700 km2 across the Roan Plateau and Piceance Basin (Fig. 1). Principal summer range habitat types included aspen (Populus tremuloides), mountain shrub, oakbrush (Quercus gambellii), big sagebrush, and pinyon-juniper. Serviceberry, snowberry (Symphoricarpos spp.), and chokecherry (Prunus virginiana) were common species in mountain shrub communities. Elevation ranged from 2000 m in Piceance Creek at the mouth of Story Gulch to 2600 m on Roan Plateau. 71 �METHODS VIT Modification We worked with ATS personnel to redesign the M3930 VIT presently manufactured by ATS. The existing M3930 has been described in detail elsewhere (Bowman and Jacobson 1998, Carstensen et al. 2003, Johnstone-Yellin et al. 2006, Bishop et al. 2007). Our redesign included changes to the retention wings and the way in which wings are attached to the transmitter body. Specifically, we extended the length and width of the retention wings and added ridges to the wing surface, both of which were intended to increase probability of retention to parturition (Fig. 3). The wings were made of flexible plastic encased in silicone. We initially produced a small number of the newly-designed wings using a relatively inexpensive prototype mold, which met our target specifications and therefore was deemed acceptable. We then manufactured a production mold, necessary to produce a large number of the wings. We incorporated ejector pins into the VIT design that allow wings to be attached to the VIT transmitter body in the field. In the original design, wings were permanently affixed to the transmitter body during the VIT assembly process. Although we only used one wing size in this study, field-attachment will allow researchers to use more than one wing size or style, without purchasing extra transmitters, if additional production molds are manufactured over time. For each wing design (i.e., production mold), extra wings could be inexpensively purchased and available in the field to affix to the fixed number of transmitter bodies. Researchers could then individually fit VITs to animals in the field much in the same way radiocollars are individually fitted. Deer Capture and VIT Insertion During late February and early March, 2009, we captured 59 adult female deer utilizing helicopter net guns (Barrett et al. 1982, van Reenen 1982) in conjunction with ongoing research addressing other objectives (Anderson and Freddy 2008). We captured 20 deer in Ryan Gulch, 19 deer in Yellow Creek, and 10 deer each in South Magnolia and Story-Sprague study units. Captured deer were hobbled, blind-folded, and ferried ≤5 km by helicopter to a central handling location. For each captured deer, we used transabdominal ultrasonography (SonoVet 2000, Universal Medical Systems, Bedford Hills, NY) to determine pregnancy status and number of fetuses (Stephenson et al. 1995, Bishop et al. 2007, Bishop et al. 2009). We shaved the left caudal abdomen from the last rib and applied lubricant to facilitate transabdominal scanning using a 3-MHz linear transducer. We fitted each pregnant deer with a VIT and a radiocollar equipped with a mortality sensor and store-on-board global position system (GPS). The mortality sensor was programmed to switch signal transmission from a slow pulse to a fast pulse after remaining motionless for 4 hours. We also measured mass, chest girth, and hind foot length of each deer and estimated age by evaluating tooth replacement and wear (Severinghaus 1949, Robinette et al. 1957, Hamlin et al. 2000). We performed the ultrasound and VIT insertion procedures in a wall-frame tent to minimize disturbance from helicopter rotor wash and adverse weather conditions and to create a dim environment to facilitate ultrasonography. We sterilized VITs in a chlorhexidine solution prior to insertion in the field. We inserted VITs using a clear, plastic swine vaginoscope (Jorgensen Laboratories, Inc., Loveland, Colo.) and alligator forceps. The vaginoscope was 15.2 cm long with a 1.59 cm internal diameter and had a smoothed end to minimize vaginal trauma. We placed vaginoscopes and alligator forceps in cold sterilization containers with chlorhexidine solution between each use and used a new pair of surgical gloves to handle the vaginoscope and VIT for each deer, and we applied a lidocaine cream to the deer’s vagina prior to insertion. To insert a VIT, we folded the wings together and placed the VIT into the end of the vaginoscope. We liberally applied sterile KY Jelly to the scope and inserted it into the vaginal canal until the tip of the VIT antenna was approximately flush with the vulva. We used previous field experience to guide insertion distance and antenna length (Bishop et al. 2007). We extended alligator forceps through the vaginoscope to firmly hold the VIT in place while the scope was pulled out from the vagina. Each VIT had a temperature-sensitive switch and a pre-cut antenna (6 cm in length) with antenna 72 �tip encapsulated in a resin bead to eliminate sharp edges. The temperature-sensitive switch caused the VIT to increase pulse rates from 40 pulses to 80 pulses per minute when the temperature dropped below 32° C. A temperature drop below 32° C was indicative of the VIT being expelled from the deer. VIT Monitoring and Success Evaluation We monitored live-dead status and general location of all radiocollared adult females daily from the ground and biweekly from the air during winter and spring. During each morning of June we checked VIT signal status by aerially locating each radio-collared doe having a VIT, weather permitting. We began flights at approximately 0630 hours and completed them by 0900–1100 hours. Early flights were necessary to detect fast signals because temperature sensors of VITs expelled in open habitats and subject to sunlight often exceeded 32° C by mid-day, which caused VITs to switch back to a slow (i.e., prepartum) pulse. When we detected a fast (i.e., postpartum) pulse rate, we used very high frequency (VHF) receivers and directional antennae from the ground to simultaneously locate the VIT and radiocollared doe. We attempted to observe behavior of the collared adult female, establish whether the VIT was shed at a birth site, and search for fawns in the vicinity of the adult female and expelled VIT. In cases where the dam moved away from the VIT (i.e., >200 m), we located the VIT to determine whether shedding occurred at a birth site and whether any stillborn fawn(s) were present and subsequently located the collared dam to search for fawns at her location. We attempted to account for each dam’s fetus(es) as live or stillborn. We typically worked in pairs, which allowed us to effectively partition effort across the study area while maintaining reasonable efficiency when searching for neonates (i.e., two people were more effective locating a hidden neonate than one person). We described effort associated with locating fawns by calculating the number of person-hours per fawn. We also quantified cost per fawn by considering all operating and personnel expenses, including capture and VIT costs for adult females. We assigned the fate of each VIT to one of 4 categories: 1) success (i.e., VIT expelled during parturition), 2) partial success (i.e., VIT expelled ≤3 days prepartum), 3) failure (i.e., VIT expelled >3 days prepartum), or 4) censor. We considered a VIT successful if it was expelled at or near a birth site in conjunction with parturition. For most success events, we located VITs at birth sites and located neonates near the VITs or in close proximity to their dams. In other success cases, we did not locate VITs at birth sites yet we found neonate(s) in close proximity to the dam, sometimes at a birth site a short distance from the expelled VIT. In these cases, we considered a VIT successful if we documented <1-day-old fawn(s) <24 hours after the VIT was expelled. Last, on two occasions, we considered a VIT successful because it was located at an evident birth site even though we could not locate fawns. Birth sites appeared as atypically large deer beds with soil appearing damp and with forbs and grasses flattened and radiating outward, consistent with a deer licking the site clean. On some occasions, fawns and/or placental remains were still present at birth sites when we arrived, providing positive confirmation of birth site characteristics. We considered VITs expelled within 3 days of parturition as partial successes because they provided useful information for locating fawns, consistent with Bishop et al. (2007). We documented such cases by locating a dam’s neonates one or more days after the VIT was expelled and comparing neonate age to VIT expulsion date. We censored VITs when adult females died prior to parturition and when adult females were located on private land that we did not have permission to access. In either case, we were unable to evaluate VIT effectiveness. All females dying prior to parturition were still carrying the VITs upon death. Analysis We modeled VIT success probability using a generalized logits model (i.e., multinomial logistic regression) in PROC LOGISTIC in SAS (SAS Institute, Cary, NC). We considered 3 levels of success consistent with our description above (success, partial success, failure) and we removed all censors from the dataset prior to analysis. We modeled VIT success as a function of adult female age (yr), mass (kg), hind foot length (cm), chest girth (cm), body fat (%), vegetative cover at VIT expulsion site, and study site. The latter two variables were included to evaluate whether locating fawns, and hence VIT success, 73 �was influenced by habitat characteristics. We expressed vegetative cover categorically as low, medium, or high. Low cover class was characterized by limited understory and overstory vegetation with minimal visual obstruction at ground level (e.g., sparsely-vegetated grass, sagebrush, or mountain shrub slopes). Medium cover class was characterized by moderate to heavy vegetative cover within 1 m of the ground but limited cover above 1 m (e.g., typical sagebrush, mountain shrub sites). High cover class comprised moderate to heavy vegetative cover from ground level up to > 1 m with nearly complete visual obstruction (e.g., oakbrush, aspen-mountain shrub, dense serviceberry). We selected among models using Akaike’s information criterion adjusted for sample size (AICc; Burnham and Anderson 2002). We then estimated the probability of locating ≥ 1 fawn, probability of locating both fawns from twin litters, and probability of locating complete litters from adult females with successful or partially successful VITs. Finally, we developed an equation for determining number of VITs necessary to achieve a specified sample of neonates for planning of future neonatal studies. RESULTS AND DISCUSSION We observed 9 adult female mortalities during winter and spring, which was much higher than expected. There was no evidence to suggest VITs were related to the mortality events. Several of the mortalities occurred within 1 week of capture and were likely capture-related. We were unable to groundmonitor 2 other adult females during the fawning period because they were located on private land that we did not have permission to access. One other adult female was inadvertently deleted from the aerial monitoring list due to miscommunication. We censored these 12 deer because they did not permit evaluation of VIT effectiveness, resulting in a sample size of 47 deer. The model of VIT success probability with the lowest AICc included only the intercept (no. parameters = 2, AICc wt = 0.271; Table 1). Probability of a VIT being expelled during parturition (i.e., success) was 0.766 (SE = 0.0605) and probability of a VIT being expelled ≤3 days prepartum (i.e., partial success) was 0.128 (SE = 0.0477). Thus, probability of a VIT being at least partially successful was 0.894 (SE = 0.0441). For comparison, using the original VIT wing design, Bishop et al. (2007) found that probability of VIT expulsion during parturition was 0.447 (SE = 0.0468), and probability of VIT expulsion during parturition or ≤3 days prepartum was 0.623 (SE = 0.0456). We employed the same methodology as Bishop et al. (2007), except for the wing modification. Assuming the 2 studies are comparable, our wing modification increased VIT retention to parturition by 0.319 (SE = 0.0765) and VIT retention to within 3 days of parturition by 0.271 (SE = 0.0634). High VIT success probability may largely explain why VIT retention did not vary as a function of any covariates we evaluated. Bishop et al. (2007) found that larger deer were more likely to expel VITs prematurely, which was the basis for modifying VIT wings. Our results suggest the wing modifications effectively reduced premature expulsion in larger deer. We located 58 neonates and 2 stillborns from 42 adult females with successful or partially successful VITs. For these 42 females, probability of locating at least 1 neonate was 0.952 (SE = 0.0333), probability of locating complete litters was 0.667 (SE = 0.0745), and probability of locating both fawns from twin litters was 0.588 (SE = 0.0857). Fawn location success did not differ between successful and partially successful VITs. Our probability estimate of locating twins is conservative because we did not place radio collars on fawns, and therefore, we could not relocate radiocollared fawns to search for their siblings. The technique of relocating a radiocollared fawn to locate its sibling was found to be successful in a previous study in Colorado (Bishop et al. 2009). During this earlier study, when a dam was known to have twin fetuses yet only one fawn was located and radiocollared during the initial capture attempt, the sibling fawn was found 45% of the time (10/22) by relocating the initial radiocollared fawn 1−2 days post-capture (C. J. Bishop, CDOW, unpublished data). Based on this rate, we would expect our probability of locating both fawns from twin litters to be roughly 0.77 had we radiocollared fawns during our study. 74 �On average, we located 1.3 neonates per VIT excluding censors and 1.0 neonate per VIT including censors. Censors need to be considered when planning VIT sample sizes for neonatal studies. Censored VITs represent the reduction in VIT sample size caused by prepartum mortality of adult females or any factor preventing access to adult females during the fawning period. We developed the following equation for determining the expected number of neonates to be encountered from a sample of VITs: , where = targeted neonate sample size. n N~, = sample size of adult females with VITs. nvns = probability adult female survives to parturition and is accessible. SAdF = probability of VIT retention to within 3 days of parturition. RvIT = probability of detecting ≥1 fawn. 'PPa w-r. = probability adult female has twin fetuses. 1'.4dF PTwms = probability of detecting twin neonates given an adult female has twin fetuses. The purpose of the above equation is to allow determination of VIT sample size once a target neonate sample size has been identified. Thus, it makes more sense to rearrange the equation as: Incorporating our estimates of retention and detection probabilities, we recommend use of the following equation to plan neonatal studies incorporating VITs with our modified wing design: nvrr, = (o.Rs)s [1 + (o.sg )r.◄dF J , AdF We expended roughly 700 person-hours during the fawning period to locate 58 neonates and 2 stillborns, or approximately 12 person-hours per fawn located. This estimate includes hours spent searching for fawns from adult females with failed VITs, although we were never successful in these attempts. Bishop et al. (2007) expended 7 person-hours per captured fawn from adult females with successful VITs, 16 person-hours per fawn from females with partially successful VITs, and 42 personhours per fawn from females with failed VITs and females not receiving VITs. Given their observed VIT success rates, Bishop et al. (2007) would have required approximately 1,315 person-hours to locate 60 neonates, or 22 person-hours per fawn. Assuming these studies are comparable, increased VIT success associated with our modified wing design resulted in a 45% reduction in labor required to locate a fawn from a radiocollared adult female. We expended $31,000 to net-gun our sample of adult females, $15,000 on VITs, $10,000 on fixed wing monitoring, and $20,000 on personnel. Thus, we expended approximately $1,267 per neonate located. We did not include adult female radio collars in our cost estimate because we used GPS collars to meet other research objectives, yet VHF collars would have sufficed for locating neonates. Assuming VHF collars were used on adult females at a rate of $250 per collar, our cost estimate becomes $1,520 per fawn. The VIT technique is therefore effective but expensive to employ. Actual cost of the technique depends on what costs are already incurred to meet other research objectives. For example, in Colorado and elsewhere, researchers have begun estimating late-winter deer body condition as a response variable to accompany survival estimates. In these cases, adult female capture and radio collar costs are already accounted for in the base study, and thus, incorporation of VITs to facilitate neonate capture becomes much more cost-effective. In our study, where adult female capture and collar costs were covered by ongoing research efforts, the added cost of incorporating VITs and neonate capture was $750 per fawn. 75 �SUMMARY Use of VITs in well-designed field studies will increase our understanding of deer limiting factors and population limitation by allowing investigators to link fawn production and survival to dam characteristics under free-ranging conditions. A primary drawback of VITs in deer has been the failure of many adult females to retain VITs to parturition. We increased VIT retention in mule deer by lengthening and widening wings used to retain a VIT in the vaginal canal. Researchers employing VITs with our modified wing design should require minimal oversampling to offset failures caused by early expulsion, thereby rendering the technique more cost-effective and reliable. Our findings provide explicit guidance for planning a fetal-neonatal deer study involving VITs. Improved VIT effectiveness facilitates increased detection of twins, and therefore, increased likelihood of radio-collaring complete litters. Determining fates of complete litters improves our ecological understanding of fawn production and recruitment and allows assessment of individual reproductive fitness if the same females are captured across years. However, it is not reasonable to assume neonatal twins are independent sample units when analyzing survival. A technique is available to quantify the amount of sibling dependence in a sample of radio-collared fawns comprising siblings to correctly estimate variance of survival rates and to improve understanding of sibling relationships (Bishop et al. 2008). Although we significantly increased VIT retention, we cannot explain why 10% of adult females expelled VITs several days or weeks prepartum. These individuals were not older or larger than other deer in our sample, making it difficult to recommend future VIT modifications to further improve retention. We speculate that individual behavior may largely explain early VIT expulsion in this study. That is, some deer may be more inclined to attempt to remove VITs than others, making it difficult to eliminate prepartum shedding altogether without dramatically changing how VITs are retained. LITERATURE CITED Anderson, C. R., and D. J. Freddy. 2008. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Study Plan, Colorado Division of Wildlife, Fort Collins, USA. Ballard, W. B., H. A. Whitlaw, D. L. Sabine, R. A. Jenkins, S. J. Young, and G. J. Forbes. 1998. Whitetailed deer, Odocoileus virginianus, capture techniques in yarding and non-yarding populations in New Brunswick. Canadian Field-Naturalist 112:254−261. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108−114. Bishop, C. J., D. J. Freddy, G. C. White, B. E. Watkins, T. R. Stephenson, and L. L. Wolfe. 2007. Using vaginal implant transmitters to aid in capture of mule deer neonates. Journal of Wildlife Management 71:945−954. Bishop, C. J., G. C. White, D. J. Freddy, B. E. Watkins, and T. R. Stephenson. 2009. Effect of enhanced nutrition on mule deer population rate of change. Wildlife Monographs 172:1−28. Bishop, C. J., G. C. White, and P. M. Lukacs. 2008. Evaluating dependence among mule deer siblings in fetal and neonatal survival analyses. Journal of Wildlife Management 72:1085−1093. Bowman, J. L., and H. A. Jacobson. 1998. An improved vaginal-implant transmitter for locating whitetailed deer birth sites and fawns. Wildlife Society Bulletin 26:295−298. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. Second Edition. Springer-Verlag, New York, New York, USA. 76 �Carstensen, M., G. D. DelGiudice, and B. A. Sampson. 2003. Using doe behavior and vaginal-implant transmitters to capture neonate white-tailed deer in north-central Minnesota. Wildlife Society Bulletin 31:634−641. Hamlin, K. L., D. F. Pac, C. A. Sime, R. M. DeSimone, and G. L. Dusek. 2000. Evaluating the accuracy of ages obtained by two methods for Montana ungulates. Journal of Wildlife Management 64:441−449. Johnson, B. K., T. McCoy, C. O. Kochanny, and R. C. Cook. 2006. Evaluation of vaginal implant transmitters in elk (Cervus elaphus nelsoni). Journal of Zoo and Wildlife Medicine 37:301−305. Johnstone-Yellin, T. L., L. A. Shipley, and W. L. Myers. 2006. Evaluating the effectiveness of vaginal implant transmitters for locating neonatal mule deer fawns. Wildlife Society Bulletin 34:338−344. Pamplin, N. P. 2003. Ecology of Columbian black-tailed deer fawns in western Oregon. Thesis, Oregon State University, Corvallis, USA. Pojar, T. M., and D. C. Bowden. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550−560. Robinette, W. L., D. A. Jones, G. Rogers, and J. S. Gashwiler. 1957. Notes on tooth development and wear for Rocky Mountain mule deer. Journal of Wildlife Management 21:134−153. Severinghaus, C. W. 1949. Tooth development and wear as criteria of age in white-tailed deer. Journal of Wildlife Management 13:195−216. Stephenson, T. R., J. W. Testa, G. P. Adams, R. G. Sasser, C. C. Schwartz, and K. J. Hundertmark. 1995. Diagnosis of pregnancy and twinning in moose by ultrasonography and serum assay. Alces 31:167−172. van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408−421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. White, M., F. F. Knowlton, and W. C. Glazener. 1972. Effects of dam-newborn fawn behavior on capture and mortality. Journal of Wildlife Management 36:897−906. Prepared by Chad J. Bishop, Wildlife Researcher 77 �Figure 1. Location of winter and summer range study areas in Piceance Basin and on Roan Plateau, northwest Colorado. 78 �Figure 2. Location of winter range study units where we captured and radio-marked mule deer in Piceance Basin, northwest Colorado. These study units are part of a larger research study evaluating effects of natural gas development and mitigation on mule deer (Anderson and Freddy 2008). 79 �I -,i 8 \''¼;_ , ~ ii ~ _J 8 SECTION B·B SCALE 2 · 1 • ~TUBlf:~OINl!ll'-.!:OtfttWll'QCJfr"'£/foltW,O,f"2l ~Sm'E'JS,..,_.~t~ ;: 1!1",TE.~fOB!PCl..'110fE>!i• FW'..J,EITl04,IIO~A t•ti01l(;p~ ~a>1!CU\folii.Blf~CWC'U,tlll) l oo..otrtJBE T,IINl9...l,.(:el7 ~ ;•::-=~~l'WE~~Na.£10111£ • m.O'f',"~.:c.•~•roc.,r,r:A(·•~,.. to•lle"r:I • _.,-IPGIIIC]~-C<NlCC,,111'.JftSTIJllt.~II• ~ Q P • t ~ . . •"V'll ...."11 '4'11 ••41.111 "'"J~•-._....,,,., t •~s, ~(MIA ' ~• t•C..-M~l'(JrNi~fW"1<ilV,.IAA. ~ ~•~ ,-;,o.ac io◄ a flC.:rwr.. :n,~• SECTION A-A SCALE 2 · 1 Figure 3. Design and dimensions of a modified retention wing used to retain vaginal implant transmitters in adult female mule deer. The displayed dimensions include a nylon core with an elastomeric overmold that protects deer from any sharp or rigid edges. 80 �APPENDIX A PROGRAM NARRATIVE STUDY PLAN FOR MAMMALS RESEARCH FY 2007-08 – FY 2009-10 State of: Cost Center: Work Package: Task No. Colorado 3430 3001 7 Federal Aid Project No. W-185-R : : : : : Division of Wildlife Mammals Research Deer Conservation Effectiveness of a Redesigned Vaginal Implant Transmitter in Mule Deer EFFECTIVENESS OF A REDESIGNED VAGINAL IMPLANT TRANSMITTER FOR CAPTURING MULE DEER NEONATES FROM TARGETED ADULT FEMALES Principal Investigators Chad J. Bishop, Wildlife Researcher, Mammals Research Chuck R. Anderson, Wildlife Researcher, Mammals Research Daniel P. Walsh, Wildlife Researcher, Wildlife Health Eric J. Bergman, Wildlife Researcher, Mammals Research Peter Kuechle, President, Advanced Telemetry Systems John Roth, Product Consultant, Advanced Telemetry Systems David J. Freddy, Wildlife Research Leader, Mammals Research Cooperators Lisa L. Wolfe, Veterinarian, Colorado Division of Wildlife Darby Finley, Terrestrial Biologist, Colorado Division of Wildlife Jamin Grigg, Terrestrial Biologist, Colorado Division of Wildlife STUDY PLAN APPROVAL Prepared by: Chad J. Bishop Date: July 2008 Submitted by: Chad J. Bishop Date: July 2008 Reviewed by: Danny Martin Date: 11/24/2008 Jon Runge Date: 11/13/2008 Date: Biometrician Review: Paul Lukacs Date: 11/4/2008 Approved by: David J. Freddy Date: Dec. 2008 Mammals Research Leader 81 �PROGRAM NARRATIVE STUDY PLAN FOR MAMMALS RESEARCH EFFECTIVENESS OF A REDESIGNED VAGINAL IMPLANT TRANSMITTER FOR CAPTURING MULE DEER NEONATES FROM TARGETED ADULT FEMALES. A Study Plan Proposal Submitted by: Chad J. Bishop, Wildlife Researcher, Mammals Research Chuck R. Anderson, Wildlife Researcher, Mammals Research Daniel P. Walsh, Wildlife Researcher, Wildlife Health Eric J. Bergman, Wildlife Researcher, Mammals Research Peter Kuechle, President, Advanced Telemetry Systems John Roth, Product Consultant, Advanced Telemetry Systems David J. Freddy, Wildlife Research Leader, Mammals Research A. Need Mule deer (Odocoileus hemionus) fawn production and neonatal survival is influenced by dam characteristics (e.g., body condition, disease status, habitat use). To understand fawn-dam relationships, manipulative field studies are needed that allow fawn production and survival to be estimated as a function of treatments applied to adult females. For example, a study evaluating the effectiveness of winter range habitat treatments on subsequent neonatal survival would require the capture of fawns from marked adult females that verifiably used, or did not use, the habitat treatments the previous winter(s). Such studies depend on a technique that enables newborn fawns to be captured from marked adult females. The most promising technique employed to capture neonates from marked adult females is use of vaginal implant transmitters (VITs), which are placed in the vagina of adult females during early to mid gestation. In theory, adult females retain VITs until parturition, at which point VITs are expelled at birth sites along with newborn fawns. Assuming VITs are routinely monitored, researchers can promptly radio-locate shed VITs and capture the newborn fawns. Recent applications of VITs in white-tailed deer (O. hemionus), black-tailed deer (O. hemionus columbianus), and mule deer have been moderately successful (Bowman and Jacobson 1998, Carstensen et al. 2003, Pamplin 2003, Bishop et al. 2007). Vaginal implant transmitters also permit measurement of fetal survival in free-ranging populations, which has important implications in populations where stillborn mortality is known to occur (Bishop 2007, Bishop et al. 2007, Bishop et al. 2008). An additional advantage of using VITs to capture neonates may be a reduction in sample bias when compared to capture techniques that rely on opportunistic fawn capture (White et al. 1972, Ballard et al. 1998, Pojar and Bowden 2004). Opportunistic techniques are susceptible to bias because of unequal capture success among vegetation types, road densities, fawn ages, and stages of fawning. When using VITs, neonate captures should be more random as long as VIT signals are monitored with equal intensity during fawning, and assuming the sample of radio-collared does was captured with minimal bias. Thus, VITs could have broad applicability regardless of whether study objectives require that fawns be captured from previously marked does. The most significant problem associated with VITs has been premature expulsion and subsequent failure to locate birth sites or newborn fawns (Bowman and Jacobson 1998, Carstensen et al. 2003, Pamplin 2003, Johnstone-Yellin et al. 2006, Bishop et al. 2007). The VIT has flexible, plastic wings coated with silicone that induce pressure against the vaginal wall to retain the transmitter. The VIT design facilitates a quick, non-surgical insertion process and is safe for the animal (Johnson et al. 2006), but the current wing design is inadequate with respect to retention. Bishop et al. (2007) found that 43% (SE = 4.7) of VITs in mule deer shed prepartum, although capture success was high when VITs shed only 1−3 days prepartum. More importantly, Bishop et al. (2007) found that 25% (SE = 4.1) of VITs shed >3 82 �days prepartum and that retention probability declined as deer body size increased, indicating the retention wings were too small to be effective in larger deer. Based on these results, considerable oversampling would be required in the design of future projects to achieve a target sample size of fawns. Oversampling is not desirable from an animal care and use perspective or from a cost perspective. Application of VITs in mule deer costs roughly $1,325 per captured fawn given current rates of premature expulsion (Bishop et al. 2007). Thus, the plastic-silicone retention wings of VITs need to be redesigned to allow maximum retention in deer. To date, the wings used to retain VITs have been purchased from a company in New Zealand (Carter Holt Harvey Plastic Products, Hamilton, New Zealand) that originally produced them for an application in the livestock industry (Bowman and Jacobson 1998). The company manufactures 1 large wing and 1 small wing; the former has been used to produce VITs for bison (Bison bison) and elk (Cervus elaphus) whereas the latter has been used to produce VITs for deer (Advanced Telemetry Systems, Isanti, MN). Advanced Telemetry Systems (ATS), in cooperation with wildlife researchers, made an initial effort in 2004 to lengthen the retention wings by adding resin to the wing tips. Using these VITs and with antennas cut to the appropriate length, S. P. Haskell (Texas Tech University, unpublished data) reported that 81% of VITs (n = 21) in deer were retained until parturition. Although retention improved, this aftermarket modification is not ideal. The modified wing tips are hard because of the resin addition and thus not ideal for placement in the vaginal canal. Also, we desire a VIT design that will provide >0.9 retention rates to parturition. Ideally, any modification to the VIT wings should be incorporated into the manufacturing process. The silicone-covered plastic wings must be manufactured using a production mold that costs roughly $15,000 to fabricate. To date, this cost has deterred design modifications to VIT wings. There is no economic incentive for a company to fabricate wing production molds exclusively for use in wildlife research given the high manufacturing costs and low anticipated return. However, the opportunity exists to redesign VIT retention wings with suitable funding. We propose to redesign the silicone-covered plastic wings, fabricate a new production mold, and conduct a field evaluation. B. Objectives Our study objectives are to (1) redesign and manufacture the silicone-covered plastic wings used to retain VITs in deer, and (2) evaluate rates of VIT retention to parturition and fawn capture rates using the newly designed wings in free-ranging mule deer. C. Expected Results or Benefits A redesigned VIT allowing high rates of retention to parturition (i.e., >0.9) would enable researchers to cost-effectively address complex problems associated with deer reproductive ecology, population productivity, and disease transmission in field studies. This field technique would then be efficacious and directly applicable to research evaluating effects of energy development and associated mitigation strategies, which is presently the highest priority facing Colorado Division of Wildlife and several other state wildlife agencies in the West. D. Approach 1. Hypotheses 1) Redesigned VITs will be retained until parturition in >90% of adult female mule deer. • Redesigning VITs by lengthening and widening the retention wings is expected to increase retention rates based on past research (Bishop et al. 2007; S. P. Haskell, Texas Tech University, unpublished data). 2) Stillborn or neonatal fawns will be located from >85% of adult female mule deer that receive redesigned VITs. • Bishop et al. (2007) captured fawns from 92% (SE = 3.7) of adult female mule deer that retained VITs to parturition. 83 �2. Experimental Design Our study design requires 2 key elements: 1) a minimum sample size of 60 adult female mule deer to guarantee suitable precision of VIT retention estimates, and 2) capture of adult female deer during mid-late winter to facilitate in utero fetus detection and to ensure VIT batteries will be operational throughout the fawning period (i.e., through early July). We will augment existing research efforts by placing VITs in adult female mule deer that will be captured in the Piceance Basin to meet other study objectives (Anderson and Freddy 2008). During 2009−2010, we will place VITs in 60 adult female mule deer each year during late February through early March in the Piceance Basin in northwest Colorado. The adult females will be captured across the Piceance Basin (Anderson and Freddy 2008) and are expected to cover an extensive area during summer (i.e., roughly 3000−4000 mi2) based on past research in this area (White et al. 1987, Bartmann et al. 1992). Assuming a VIT retention rate of 0.9 (i.e., 90% of VITs shed at birth sites), 60 adult females would allow us to estimate a yearly retention rate with a 95% confidence interval (CI) of 0.79−0.96, or a coefficient of variation (CV) of 4.3%. Following the 2-year study, we will be able to estimate retention rate with a 95% CI of 0.83−0.95 (i.e., CV = 3.1%), if there is no significant year effect. If we observe a year effect, we may be able to identify factor(s) that were potentially responsible and improve our understanding of VIT retention. Also, if we experience a problem in the first year, we may be able to correct it prior to the second year. If we experience high success during the first year (e.g., >0.9 retention to parturition), the second year may become part of a biological study to evaluate effects of energy development on fawn production and neonatal survival. 3. Procedures We worked with ATS personnel to redesign the M3930 VIT presently manufactured by ATS. The existing M3930 has been described in detail elsewhere (Bowman and Jacobson 1998, Carstensen et al. 2003, Johnstone-Yellin et al. 2006, Bishop et al. 2007). Our redesign included changes to the retention wings and the way in which wings are attached to the transmitter body. Specifically, we extended the length and width of the retention wings and added ridges to the wing surface, both of which should increase probability of retention to parturition (Figs. 1, 2). The wings are made of flexible plastic encased in silicone. We initially produced a small number of the newly-designed wings using a relatively inexpensive prototype mold (i.e., $1,200). The prototype was acceptable. We will therefore manufacture a production mold (i.e., ~$15,000), which will allow a large number of the wings to be produced. The wings will be inexpensive to manufacture once the production mold is available. We will incorporate ejector pins into the VIT design that will allow wings to be attached to the VIT transmitter body in the field. Previously, wings were permanently affixed to the transmitter body during the VIT assembly process. Field-attachment would allow researchers to use more than one wing size or style, without purchasing extra transmitters, if additional production molds are manufactured over time. For each wing design (i.e., production mold), extra wings could be inexpensively purchased and available in the field to affix to the fixed number of transmitter bodies. Researchers could then individually fit VITs to animals in the field much in the same way radiocollars are individually fitted. In late February or early March each year of study, we will capture a total of 60 adult female deer utilizing helicopter net guns (Barrett et al. 1982, van Reenen 1982) in conjunction with ongoing research (Anderson and Freddy 2008). Captured deer will be hobbled, blind-folded, and ferried ≤3.5 km by helicopter to a central handling location. For each captured deer, we will use transabdominal ultrasonography (SonoVet 2000, Universal Medical Systems, Bedford Hills, NY) to determine pregnancy status and number of fetuses (Stephenson et al. 1995, Bishop 2007, Bishop et al. 2007). We will shave the left caudal abdomen from the last rib and apply lubricant to facilitate transabdominal scanning using a 3-MHz linear transducer. We will fit each pregnant deer with a VIT and a radiocollar equipped with a mortality sensor, which will activate after remaining motionless for 4 hours. 84 �We will also measure mass, chest girth, and hind foot length of each deer and estimate age by evaluating tooth replacement and wear (Severinghaus 1949, Robinette et al. 1957, Hamlin et al. 2000). We will perform the ultrasound and VIT insertion procedures in a wall-frame tent or other structure to minimize disturbance from helicopter rotor wash and adverse weather conditions and to create a dim environment to facilitate ultrasonography. We will sterilize VITs in a chlorhexidine solution prior to insertion in the field. We will insert VITs using a clear, plastic swine vaginoscope (Jorgensen Laboratories, Inc., Loveland, Colo.) and alligator forceps. The vaginoscope is 15.2 cm long with a 1.59 cm internal diameter and has a smoothed end to minimize vaginal trauma. We will place vaginoscopes and alligator forceps in cold sterilization containers with chlorhexidine solution between each use and use a new pair of nitrile surgical gloves to handle the vaginoscope and VIT for each deer, and we will apply a lidocaine cream to the deer’s vagina prior to insertion. To insert a VIT, we will fold the silicone wings together and place the VIT into the end of the vaginoscope. We will liberally apply sterile KY Jelly to the scope and insert it into the vaginal canal until the tip of the VIT antenna is approximately flush with the vulva. We will use previous field experience to guide insertion distance and antenna length (Bishop et al. 2007). We will extend alligator forceps through the vaginoscope to firmly hold the VIT in place while the scope is pulled out from the vagina. Each VIT will have a temperature-sensitive switch, pre-cut antenna (~6 cm in length) with antenna tip encapsulated in a resin bead to eliminate sharp edges, and a 12-hour on-off duty cycle to extend battery life (Bishop et al. 2007). The temperature-sensitive switch will cause the VIT to increase pulse rates from 40 pulses to 80 pulses per minute when the temperature drops below 32° C. A temperature drop below 32° C will be indicative of the VIT being expelled from the deer. We will regularly monitor live-dead status and general location of all radiocollared adult females during winter and spring. During each morning of June we will check VIT signal status by aerially locating each radio-collared doe having a VIT, weather permitting. We will begin flights at approximately 0530 hours and complete them by approximately 1000–1100 hours. Early flights will be necessary to detect fast signals because temperature sensors of VITs expelled in open habitats and subject to sunlight often exceed 32° C by mid-day, which will cause VITs to switch back to a slow (i.e., prepartum) pulse (Bishop et al. 2007). When we detect a fast (i.e., postpartum) pulse rate, we will use very high frequency (VHF) receivers and directional antennae from the ground to simultaneously locate the VIT and radiocollared doe. We will attempt to observe behavior of the collared adult female, establish whether the VIT is shed at a birth site, and search for fawns in the vicinity of the adult female and expelled VIT. In cases where the dam moves away from the VIT (i.e., >200 m), we will locate the VIT to determine whether shedding occurred at a birth site and whether any stillborn fawn(s) are present and subsequently locate the collared dam to search for fawns at her location. We will attempt to account for each dam’s fetus(es) as live or stillborn, which is fundamental to estimating fetal survival (Bishop et al. 2007, 2008). We will wear surgical gloves when handling fawns to help minimize transfer of human scent. We will work in pairs and partition the study area into segments, whereby each 2-person team is responsible for one segment. We anticipate needing 4−5 teams given the expanse of the study area (Fig. 3). We will assign the fate of each VIT to one of 6 categories: 1) parturition shed, 2) late prepartum shed (i.e., ≤3 days prepartum), 3) early prepartum shed (i.e., >3 days prepartum), 4) battery or transmitter failure, 5) migration loss, or 6) censor (Bishop et al. 2007). We will identify parturition sheds based on identification of a birth site where the VIT is shed or location of <1-day-old fawn(s) <24 hours after a VIT is shed. The latter criterion is useful because not all birth sites can be positively identified once the dam has cleaned up afterbirth and moved the fawns. Although our primary objective is to quantify the proportion of VITs shed at parturition, the remaining VIT fate categories will be useful for understanding why VITs failed and should aid additional technique refinements. We will distinguish between early 85 �prepartum sheds and late prepartum sheds because the latter provides useful information for capturing fawns. Neonate capture success rate was 0.792 (SE = 0.0847, n = 24) for dams with VITs shed late prepartum on the Uncompahgre Plateau during 2002−2004 (Bishop et al. 2007). We will document battery failures based on the disappearance of a doe’s VIT signal after having consistently heard the signal on a daily basis. Migration losses refer to any VIT signals that disappear during spring migration. These failures are presumably caused by battery failures or early prepartum sheds between winter and summer range, yet the specific cause cannot be determined (Bishop et al. 2007). We will censor VITs associated with prepartum doe mortalities and missing does (i.e., unable to detect radiocollar signal) because these deer will not provide an adequate test of VIT effectiveness (i.e., the failure is independent of VIT technology). We will quantify the proportion of successful fawn captures associated with VITs shed at parturition as well as those shed ≤3 days prepartum. We will also determine whether we account for the entire litter by comparing the number of fawns located in June to the in utero fetal counts obtained in February−March. We will describe effort associated with fawn capture by calculating the number of person-hours per captured fawn. We will also quantify cost per captured fawn by considering all operating and personnel expenses, including capture and transmitter costs for adult does. 4. Data Analysis Procedures We will use a straight-forward binomial model to estimate the probability of VIT retention until parturition in adult female mule deer. We will contrast this estimate with a previous retention probability estimate (0.447, SE = 0.0468, Bishop et al. 2007) to evaluate the likely effect of our VIT modification. The estimates are not directly comparable because they will not have been measured simultaneously. However, the initial retention estimate measured by Bishop et al. (2007) provides a baseline for evaluating whether our VIT modifications had a positive effect. Ultimately, we will evaluate our retention probability estimate relative to our hypothesized retention rate of 0.9. We will model VIT retention as a function of adult female individual covariates (i.e., age, mass, chest girth, hind foot length) using logistic regression in SAS (SAS Institute, Cary, North Carolina) to improve our understanding of factors related to retention, which will be particularly useful if retention is < 0.9. We will select among models using Akaike’s information criterion adjusted for sample size (AICc; Burnham and Anderson 2002). We will also estimate fawn detection probability associated with adult females receiving VITs. Specifically, we will estimate separate detection probabilities for adult females that shed VITs prepartum and adult females that shed VITs at parturition. We will then use the detection probabilities to estimate the probability of capturing the complete litter for different sized litters. E. Location The proposed research will take place in the vicinity of Piceance Basin and the White River National Forest in northwest Colorado (Fig. 3). Anderson and Freddy (2008) provided a detailed description of winter range study sites where adult female mule deer will be captured. The winter range study area is located primarily within CDOW Game Management Unit (GMU) 22. Summer range will be defined by the movements of the radiocollared adult females captured on winter range. We anticipate the summer range study area will include portions of GMUs 11, 211, 12, 22, 23, 24, 31, 32, and 33 (Fig. 3). 86 �F. Schedule Of Work Activity Complete Initial Draft of Study Plan Manufacture VIT Retention Wing Production Mold Finalize Study Plan and Submit to ACUC Order VITs and Purchase Associated Field Equipment Capture Deer and Insert VITs Periodically Monitor Radiocollared Deer Monitor VITs Daily, Locate Shed VITs, and Conduct Fawn Searches Analyze Data and Prepare Progress Report Analyze Data and Prepare Final Report Submit VIT Techniques Manuscript for Publication Date April−May 2008 May−June 2008 August−October 2008 November 2008−2009 February−March 2009−2010 March−May 2009−2010 June 2009−2010 July−August 2009 July−August 2010 December 2010 G. Estimated Costsa Category Item or Position FY 07-08 FY 08-09 FY 09-10 Personnel Chad Bishop 0.20 PFTE 0.40 PFTE 0.40 PFTE Chuck Anderson 0 0.05 PFTE 0.05 PFTE Eric Bergman 0 0.05 PFTE 0.05 PFTE Dan Walsh 0.05 PFTE 0.05 PFTE 0.05 PFTE 0 6.5 Mo. - $17,186 7.0 Mo. - $18,760 VIT Prototype $2,500 0 0 VIT Production Mold $18,500 0 0 Fixed-wing Monitoring (June) 0 $14,875 $15,750 Field Supplies 0 $5,000 $4,000 60 VITs 0 $13,800 $13,800 Telemetry Equipment 0 $3,000 $1,500 TFTE Operating Total Cost $21,000 $53,861 $53,810 a Study costs were minimized by leveraging existing mule deer capture efforts within the ongoing Piceance Basin deer study (Anderson and Freddy 2008). H. Related Federal Projects Our research will be conducted on federal (i.e., BLM, USFS), state, and private lands. The study does not involve formal collaboration with any federal agencies, nor does the work duplicate any ongoing federal projects. I. Literature Cited Anderson, C. R., and D. J. Freddy. 2008. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Study Plan, Colorado Division of Wildlife, Fort Collins, USA. 87 �Ballard, W. B., H. A. Whitlaw, D. L. Sabine, R. A. Jenkins, S. J. Young, and G. J. Forbes. 1998. Whitetailed deer, Odocoileus virginianus, capture techniques in yarding and non-yarding populations in New Brunswick. Canadian Field-Naturalist 112:254−261. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108−114. Bartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monographs 121:1−39. Bishop, C. J. 2007. Effect of enhanced nutrition during winter on the Uncompahgre Plateau mule deer population. Dissertation, Colorado State University, Fort Collins, USA. Bishop, C. J., D. J. Freddy, G. C. White, B. E. Watkins, T. R. Stephenson, and L. L. Wolfe. 2007. Using vaginal implant transmitters to aid in capture of mule deer neonates. Journal of Wildlife Management 71:945−954. Bishop, C. J., G. C. White, and P. M. Lukacs. 2008. Evaluating dependence among mule deer siblings in fetal and neonatal survival analyses. Journal of Wildlife Management 72:1085−1093. Bowman, J. L., and H. A. Jacobson. 1998. An improved vaginal-implant transmitter for locating whitetailed deer birth sites and fawns. Wildlife Society Bulletin 26:295−298. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. Second Edition. Springer-Verlag, New York, New York, USA. Carstensen, M., G. D. DelGiudice, and B. A. Sampson. 2003. Using doe behavior and vaginal-implant transmitters to capture neonate white-tailed deer in north-central Minnesota. Wildlife Society Bulletin 31:634−641. Hamlin, K. L., D. F. Pac, C. A. Sime, R. M. DeSimone, and G. L. Dusek. 2000. Evaluating the accuracy of ages obtained by two methods for Montana ungulates. Journal of Wildlife Management 64:441−449. Johnson, B. K., T. McCoy, C. O. Kochanny, and R. C. Cook. 2006. Evaluation of vaginal implant transmitters in elk (Cervus elaphus nelsoni). Journal of Zoo and Wildlife Medicine 37:301−305. Johnstone-Yellin, T. L., L. A. Shipley, and W. L. Myers. 2006. Evaluating the effectiveness of vaginal implant transmitters for locating neonatal mule deer fawns. Wildlife Society Bulletin 34:338−344. Pamplin, N. P. 2003. Ecology of Columbian black-tailed deer fawns in western Oregon. Thesis, Oregon State University, Corvallis, USA. Pojar, T. M., and D. C. Bowden. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550−560. Robinette, W. L., D. A. Jones, G. Rogers, and J. S. Gashwiler. 1957. Notes on tooth development and wear for Rocky Mountain mule deer. Journal of Wildlife Management 21:134−153. Severinghaus, C. W. 1949. Tooth development and wear as criteria of age in white-tailed deer. Journal of Wildlife Management 13:195−216. Stephenson, T. R., J. W. Testa, G. P. Adams, R. G. Sasser, C. C. Schwartz, and K. J. Hundertmark. 1995. Diagnosis of pregnancy and twinning in moose by ultrasonography and serum assay. Alces 31:167−172. van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408−421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, USA. White, G. C., R. A. Garrott, R. M. Bartmann, L. H. Carpenter, and A. W. Alldredge. 1987. Survival of mule deer in northwest Colorado. Journal of Wildlife Management 51:852−859. White, M., F. F. Knowlton, and W. C. Glazener. 1972. Effects of dam-newborn fawn behavior on capture and mortality. Journal of Wildlife Management 36:897−906. 88 �J. Figures And Tables D C E 7 B AL("_ - ~ ---J_:_:._JA B ~-.ff -.:::-·:::::::ri::j-• _j _j ---,"C"".. C .. " -~ fll_J\llf.!1 ,.- D '"'.,.,,... . ~f'il;I I I ~ ~,.. ' 'PO'l;rt1- 1,, -I - ,~ ,-- ! ; 2!tJtOH'JC~ _::1 i E - ~Pt "' I ~1:a +- ~ ' 'I ' ,..m-, •~I SECTION C-C SCALE 2 : 1 - .,.,111~, ll'l(krc=,- SECTION D-D SCALE 2 1 SECTION E-E SCALE 2 1 m V --...:: ~ 1N , v- { _,,, / '...-=,. f - O)_ ..,.. ,.. \ Plltlf1 ""' ':I ~ I? ~ •1'1':11:1 ,.,,. W.~IAJ...":)~J#f\.0-.1.-, OCl5UOAtD ,llfE-El~ltlF-BnM'.: J O ~ Wll 7Fl'll •1 Je,,;.~ CD.OA'TOFIE"Qfl.iQJ..:;W,.,W. •1.H;~~F~r...11AJT"1..r...iTOS T«f• I~ ""''" ~ ~~~~J~::t~~~=~FV' CN'ID,111 ... ntnicoa- ~ Dl'l(jQ1D)FtAJM~~l ~1:.S"I ~"Ekl ""' ... ,h, ~ L 1..Qtt..CH>I~ ~Yil• 11~~ SECTION 8-B SCALE 2 : 1 , SECT ION A-A SCALE 2 1 Figure 1. Modified design of the nylon core of retention wings used to retain vaginal implant transmitters in adult female mule deer. We modified the original design by lengthening and widening the wings and modifying the shape. We also incorporated an ejector pin to facilitate attachment of different-sized wings in the field. 89 �/ 8 7 ! -:,- i- \, \, ! : ~-~ _! 8 SECTION B•B SCALE 2 · 1 ...... • ~Tl1111:-...tlcDll\lB1A!:OtEU:'Nl".:..:J,t~-M«:le:I lEI.BiET,o'Si'Sl'U.ISF¥fHJMB' 15f0.I' : ~ c : : ~ = ~ • . w ~ ~ & G I E A T<a!KP~ l.:o.atl'D!!EPl,-,~IT t .L ~"' ~-~;~'} 1 1~-~? \ U~ffl p.- '°'>#IN."""'-LTWOltE:!iS -OBE D,l';I'' ~ ~"~':~~RW'E?"'K-ffU ~ " ' ~ ,oec ___ ,..,_ _ _ _ ,,. _ _ • ~ ~ . . . . -... - ~ , tO 5>,lm'/r.-' ...,., tO• ~ r ~~~~::+t~.~~"r.:,~".'f,=::.i_, S2- 11;.I \ • =;: ,.... J...,_i'.i I \ '~ ,-~~~ SECTION A-A SCALE 2 • 1 ••..,.. ~~.:w flrlf!.'toY'IO,,(W.,.'(UAllt,, ~ ~•-111,10.lilC~ .. fl(,"'1•)1')<1WI. ~ liio!:_ __ ;:o ~nod~~[(,.,;~ ...,~ "154Qe Figure 2. Design and dimensions of a modified retention wing used to retain vaginal implant transmitters in adult female mule deer. The displayed dimensions include the nylon core (Figure 1) with an elastomeric overmold that protects deer from any sharp or rigid edges. 90 �Figure 3. Location of winter range and summer range study areas in the vicinity of Piceance Basin and White River National Forest in northwest Colorado, where we will evaluate the effectiveness of modified vaginal implant transmitters (VITs). Winter and summer range study areas are outlined in white. Mule deer winter range is denoted with dark shading and USFS lands are outlined in black. 91 �K. Appendices APPENDIX I HELICOPTER NET-GUN CAPTURE AND HANDLING PROTOCOL FOR MULE DEER Helicopter net-gunning is a well-established procedure for capturing ungulates (Barrett et al. 1982, van Reenen 1982). Large samples of mule deer and white-tailed deer have been captured using helicopter net-guns with ≤ 1% capture-related mortality (Potvin and Breton 1998, White and Bartmann 1994, Webb et al. 2008). The protocol described below is nearly identical to net-gun protocols approved previously by CDOW’s ACUC (CDOW ACUC Project Protocols 11−2000, 10−2005, 15−2007). Capture-related mortality rates in these projects have ranged from 0 to 3.5%, which includes all animals dying ≤1 week post-capture regardless of cause. A capture mortality rate of 3.5% is higher than the preferred rate of 2% (Spraker 1993) but much lower than what has commonly been experienced in the field using other methods to capture deer (Conner et al. 1987, DelGiudice et al. 2005). The 3.5% capturerelated mortality rate occurred on the Uncompahgre Plateau when large samples of mule deer were captured within small study sites, creating challenging conditions for helicopter net-gunning (Bishop 2007). The overall capture mortality rate in this study was 2% because a majority of deer were captured with drop nets, where capture mortality was 1%. In other recent studies, capture-related mortality rates associated with helicopter net-gunning have been ≤ 2% (Anderson and Freddy 2008; Bergman et al. 2006, 2007, 2008). Net-gunning will be performed by Quicksilver Air, Inc., or other qualified vendor selected by the Colorado Division of Wildlife (CDOW) through a request-for-proposal (RFP) process, which is the required procedure for selecting vendors to conduct helicopter work for CDOW. Quicksilver Air, Inc., has captured large samples of deer in Colorado during the past few years with capture-related mortality rates generally ≤ 2% (Anderson and Freddy 2008; Bergman et al. 2006, 2007, 2008; B. E. Watkins, CDOW, personal communication). Capture and Transport Methods: Wild mule deer will be pursued and netted by the helicopter net-gunning crew. The crew will consist of one pilot, one net-gunner, and ≤2 handlers. Netted animals will immediately be blind-folded and hobbled and transported by the helicopter to a nearby handling site. Deer will be placed inside the helicopter or slung underneath the helicopter during transport. At the handling site, CDOW personnel (i.e., handling crew) will record measurements, affix transmitters, and release each captured deer. Mule deer will be captured within 1−2 miles of the handling site to minimize the distance deer are transported. The handling crew will be ferried to appropriate handling sites by the helicopter pilot if vehicle access is limited in an area. Mule deer will be captured with net-guns in late February or early March in Game Management Unit (GMU) 22 in the Piceance Basin. In Meeker, Colorado, mid-late winter snow depths average roughly 12 cm, and rarely exceed 35 cm, where deer will be captured, and mean daily temperatures during late February have averaged –1 °C (30 °F) during recent decades. Under these conditions, mule deer can be captured safely without undue risk of hyperthermia. Maximum allowable pursuit time, or time necessary to chase and net a target animal, will vary given existing weather conditions and animal behavior. For example, in warmer conditions (e.g. >4°C), pursuit times will be minimized, particularly if unfavorable snow conditions are present. Total pursuit time will not exceed 8−10 minutes regardless of conditions, and will generally be less than 5 minutes. Individual deer will not be repeatedly chased. Large deer groups typically fracture upon the initial pursuit, thereby preventing the need to repeatedly chase the same individuals while still allowing the capture of >1 deer from the initial group. 92 �The helicopter pilot, fuel truck driver, and handling crew will be in radio contact with one another. In the event of an accident, the Meeker CDOW office will be contacted by radio, and necessary emergency services will be sent to the site. The ground crew will have direct radio access to the Rio Blanco County Sheriffs Office, Colorado State Patrol, and other emergency law enforcement channels. Training and Personnel: The helicopter net-gunning crew will be instructed as to procedures for minimizing stress and injury to the animals. Specifically, they will be instructed on pursuit times, transport distances, and safe handling procedures. The handling crew, comprised of CDOW personnel, will be instructed on proper care and handling procedures to minimize stress and risk of injury to the captured deer. Chad Bishop and Chuck Anderson will be ultimately responsible for all animal care and handling during the capture operation. Procedures and Manipulations of Animals: As stated above, netted animals will immediately be blind-folded, hobbled, and transported to the handling site. At the handling site, deer will be removed from the net and/or transport bag if present, and the blind-fold and hobbles will be checked. Deer will be radiocollared and aged by qualitatively evaluating height and wear of incisors and premolars. Radio collars will be of fixed-size and individually fitted to each animal. The following samples will be obtained from each deer: blood, hind foot length, chest girth, and weight. Blood samples will be collected using routine venipuncture for evaluating serum thyroid hormone concentrations and disease serology. Pregnancy status, number of fetuses, and body condition will also be determined using ultrasonography. Please refer to Appendix II for detailed handling procedures (Appendix II. Use of Ultrasonography and Vaginal Implant Transmitters in Adult Female Mule Deer to Capture Neonatal Fawns). If a captured deer suffers a broken leg, back, neck, pelvis, or other similar wound, it will be euthanized by deep anesthesia with the drug combination of ketamine or Telazol© and xylazine (IV or IM) with dosage based on estimated weight, followed by intravenous administration of KCl (~350 mg KCl/ml sterile water, dosed at >50 mg KCl/kg estimated body mass). In situations where administration of KCl is not feasible, then euthanasia will be performed via a gunshot to the head. Radiocollared mule deer will not be handled following capture, although they will be radiomonitored from both the ground and air on a routine basis. Except during the fawning period, deer will not be routinely relocated from the ground using VHF telemetry and therefore will not be regularly disturbed. During fawning in June, deer will be radiomonitored daily to determine when vaginal implant trasmitters are shed (see Appendix II. Use of Ultrasonography and Vaginal Implant Transmitters in Adult Female Mule Deer to Locate Neonatal Fawns). Literature Cited: Anderson, C. R., and D. J. Freddy. 2008. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, USA. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108-114. Bergman, E. J., C. J. Bishop, D. J. Freddy, and G. C. White. 2006. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, USA. Bergman, E. J., C. J. Bishop, D. J. Freddy, and G. C. White. 2007. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, USA. 93 �Bergman, E. J., C. J. Bishop, D. J. Freddy, and G. C. White. 2008. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Wildlife Research Report, Colorado Division of Wildlife, Fort Collins, USA. Bishop, C. J. 2007. Effect of enhanced nutrition during winter on the Uncompahgre Plateau mule deer population. Dissertation, Colorado State University, Fort Collins, USA. Conner, M. C., E. C. Soutiere, and R. A. Lancia. 1987. Drop-netting deer: costs and incidence of capture myopathy. Wildlife Society Bulletin 15:434−438. DelGiudice, G. D., B. A. Sampson, D. W. Kuehn, M. Carstensen Powell, and J. Fieberg. 2005. Understanding margins of safe capture, chemical immobilization, and handling of free-ranging white-tailed deer. Wildlife Society Bulletin 33:677−687. Potvin, F., and L. Breton. 1988. Use of a net gun for capturing white-tailed deer Odocoileus virginianus on Anticosti Island, Quebec. The Canadian Field Naturalist 102:697−700. Spraker, T. R. 1993. Stress and capture myopathy in artiodactylids. Pages 481−488 in M. E. Fowler, editor. Zoo and wild animal medicine: current therapy 3. W. B. Saunders, Philadelphia, Pennsylvania, USA. van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408-421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, Wisconsin, USA. Webb, S. L., J. S. Lewis, D. G. Hewitt, M. W. Hellickson, and F. C. Bryant. 2008. Assessing the helicopter and net gun as a capture technique for white-tailed deer. Journal of Wildlife Management 72:310−314. White, G. C., and R. M. Bartmann. 1994. Drop nets versus helicopter net guns for capturing mule deer fawns. Wildlife Society Bulletin 22:248−252. 94 �APPENDIX II ULTRASONOGRAPHY AND VAGINAL IMPLANT TRANSMITTER PROTOCOLS FOR ADULT FEMALE MULE DEER AND NEONATAL FAWNS Background: For some time, radio-transmitter implants in the vaginas of deer have been considered as a technique for locating and capturing newborn fawns from radio-collared does immediately following parturition. Early attempts to employ this technique were largely unsuccessful in terms of both effectiveness and animal welfare (Garrott and Bartmann 1984, Giessman and Dalton 1984, Nelson 1984). This early technique used sutures to partially close the vulva in order to retain the transmitter in the vagina. Later, Bowman and Jacobsen (1998) developed and employed a modified vaginal implant transmitter (VIT) for white-tailed deer, with better success. This transmitter had plastic wings encased in silicone to retain the transmitter in the vagina until parturition; thus, no sutures were used. They found no indications that animals were negatively impacted by the newly designed VIT. Recent studies employing VITs have not identified any negative impacts to animals receiving VITs (Carstensen et al. 2003, Pamplin 2003, Johnstone-Yellin et al. 2006, Bishop et al. 2007), including a VIT study on elk focused exclusively on animal welfare (Johnson et al. 2006). Also, these studies do not indicate that VITs cause major problems with in utero fetus survival or birthing, particularly given the success of researchers at finding birth sites and fawns, occasionally from the same adult females over consecutive years. Furthermore, farmed deer in New Zealand with vaginal hormone implants with a similar design have not had any major reproduction problems (Asher and Smith 1987, Asher et al. 1988, Mylrea et al. 1992). Although the current VIT design apparently causes no harm to the animal, animals often expel VITs prior to parturition, which greatly reduces their utility. Thus, to achieve target sample sizes of newborn fawns, investigators must oversample adult females, causing excess animals to be captured, handled, and implanted with VITs. To reduce premature VIT expulsion, Advanced Telemetry Systems (ATS), in cooperation with wildlife researchers, lengthened the retention wings in 2004 from 58 mm to 68 mm by adding hard resin to the wing tips, which significantly improved VIT retention (S. P. Haskell, Texas Tech University, unpublished data). Since 2004, researchers employing VITs with the longer wings have not documented any ill effects in deer (ATS, unpublished data). Although retention improved and no ill effects have been observed, this aftermarket modification is not ideal. The modified wing tips are hard because of the resin addition and thus not ideal for placement in the vaginal canal. Ideally, any modification to the VIT wings should be incorporated into the manufacturing process. The retention wings must be manufactured using a production mold that costs a minimum of $15,000 to fabricate. We therefore obtained suitable funding and redesigned the VIT production mold. We lengthened the wing mold from 58 mm to 68 mm, consistent with the aftermarket modifications made to VIT wings beginning in 2004. We also widened the wings from 9 mm to 14 mm to increase the contact surface with the vaginal wall. During spring-summer 2008, we placed 6 prototypes of our newly-manufactured VITs in bighorn sheep ewes at the Foothills Wildlife Research Facility in Fort Collins, CO, where the penned sheep could be closely monitored. We documented no ill effects and all pregnant sheep retained their VITs until parturition. We do not anticipate that our VIT design modifications will pose a risk to animal welfare considering our pilot evaluation in sheep and recent deer studies that employed VITs with aftermarket alterations. In fact, the motivation for developing a new production mold was to improve animal welfare by eliminating the need for aftermarket alterations that create particularly hard wing surfaces. We will monitor fetal survival and neonatal production of all adult female deer receiving VITs to help document whether the newly designed VITs cause any negative effects. We will also monitor survival of the adult females and conduct a thorough necropsy of any deer that die. 95 �Aside from the VIT modifications, the protocols described herein are nearly identical to a protocol approved in the past (CDOW ACUC Project Protocol 1−2002). In this earlier study, we did not document any negative effects to deer associated with ultrasonography or VIT procedures. Also, neonatal fawn survival was higher among fawns captured from adult does that received VITs than fawns captured opportunistically from adult does that did not have VITs (Bishop et al. 2007). Vaginal implants allowed us to remotely monitor adult doe birthing status. If a VIT functioned correctly, we were generally able to capture the adult doe’s fawn(s) with only one disturbance event. In the absence of a VIT, when attempting to capture fawns from a targeted adult doe, we typically had to repeatedly locate and disturb the adult doe during the fawning period to capture her fawn(s). Capture and Transport Technique: Adult female mule deer will be captured in late February and/or early March via helicopter netgunning (Barrett et al. 1982, van Reenen 1982). Please refer to Appendix I. for a detailed helicopter netgunning capture protocol (Appendix I. Helicopter Net-gunning Capture and Handling Protocol for Mule Deer). Net-gunned deer will be blind-folded, hobbled, and ferried a short distance to a handling site. Procedures and Manipulations of Animals: We will use ultrasonography to determine pregnancy status (yes/no), fetal count (# fetuses), and body condition (see below). Additionally, we will measure weight, chest girth, hind foot length, and age (based on tooth replacement and wear). We will collect a blood sample using routine venipuncture. If an adult female is pregnant, we will place a nylon radio-collar around the neck and insert a VIT in the vagina posterior to the cervix. Vaginal implant insertion procedures are explained in detail below. Total handling time for an individual deer will typically be ~15 minutes and will not exceed 25 minutes. We will cease manipulations/data collection at any point the welfare of the deer is in question and immediately begin administering fluids, oxygen, or any other warranted procedure under the guidance of CDOW’s attending veterinarian. Ultrasonography: We will use ultrasonography to determine body condition, diagnose pregnancy, and quantify fetal numbers of each mule deer. Body condition will be measured to meet other research objectives (Anderson and Freddy 2008). Body condition methods are briefly repeated here for completeness. We will measure maximum subcutaneous fat thickness on the rump and thickness of the longissimus dorsi muscle of each doe using a SonoVet 2000 portable ultrasound unit (Universal Medical Systems, Bedford Hills, NY) with a 5 MHz linear transducer (Stephenson et al. 1998, 2002; Cook et al. 2001; Bishop 2007). A small area of hair will be plucked at each measurement point and lubricant will be used to enhance contact between the transducer and skin. The 2 plucked areas will be ≤15 cm long by ≤5 cm wide. We will determine a body condition score (BCS) for each deer by palpating the rump (Cook et al. 2001, 2007). We will combine ultrasound measurements with the BCS score to estimate body fat of each deer (Cook et al. 2007). We will quantify reproductive status using a SonoVet 2000 portable ultrasound unit (Universal Medical Systems, Bedford Hills, NY) with a 3 MHz linear transducer. We will shave the left side of the abdomen and apply lubricant to facilitate transabdominal scanning (Stephenson et al. 1995, Bishop 2007, Bishop et al. 2007). Specifically, we will shave an area covering the haired portion of the left ventral abdomen that is 20 cm wide; the area is bounded by the caudal rib cranially, the inguinal fold caudally, and the ventral midline. Both uterine horns will be systematically scanned to identify fetal numbers ranging from 0 to 3. 96 �Vaginal Implant Transmitter (VIT) and Insertion Technique: Refer to the attached study plan for detailed specifications of VITs to be used in this study. Prior to insertion, we will sterilize VITs in a chlorhexidine solution, rinse them with sterile saline solution, allow them to air-dry, and seal them in air- and water-tight pouches. This will guarantee cleanliness of VITs up until the moment they are placed in deer. We will insert VITs using a clear, plastic swine vaginoscope (Jorgensen Laboratories, Inc., Loveland, Colo.) and alligator forceps. The vaginoscope is 15.2 cm long with a 1.59 cm internal diameter and has a smoothed end to minimize vaginal trauma. We will gauge approximate insertion distance from extensive experience gained on the Uncompahgre Plateau (Bishop et al. 2007). We will place vaginoscopes and alligator forceps in cold sterilization containers with chlorhexidine solution between each use and use a new pair of nitrile surgical gloves to handle the vaginoscope and VIT for each deer, and we will apply a lidocaine cream to the deer’s vagina prior to insertion. To insert a VIT, we will fold the silicone wings together and place the VIT into the end of the vaginoscope. We will liberally apply sterile KY Jelly to the scope and insert it into the vaginal canal until the tip of the VIT antenna is approximately flush with the vulva. We will use the alligator forceps, which extend through the vaginoscope, to firmly hold the VIT in place while the scope is pulled out from the vagina. The tip of the antenna, which may protrude up to 1.5 cm past the vulva, is encapsulated is a resin bead to protect the deer from its sharp edge. Post-Implantation Monitoring: From March through May, we will regularly monitor the radio collar and VIT signals of the adult does in our sample. Monitoring will allow us to document any VITs that shed early and the opportunity to perform a necropsy on mortalities. The latter will allow us to evaluate whether VITs caused any tissue irritation or other impact to the adult doe. Fetus Survival and Neonate Capture: During each morning of June we will check VIT signal status by aerially locating each radiocollared doe having a VIT, weather permitting. We will also radiomonitor VIT signals from the ground as logistically feasible. When we detect a fast (i.e., postpartum) pulse rate, we will use VHF receivers and directional antennae from the ground to simultaneously locate the VIT and radio-collared doe, which should be in proximity to one another. We will attempt to observe behavior of the collared doe, establish whether the VIT is shed at a birth site, and search for fawns in the vicinity of the doe and expelled VIT. If the doe has moved away from the VIT (i.e., >200 m), we will locate the VIT to determine whether shedding occurred at a birth site and whether any stillborn fawn(s) were present and subsequently locate the collared doe to search for fawns at her location. We will attempt to account for each doe’s fetus(es) measured in February as live or stillborn fawns. We will not radiocollar or handle newborn fawns. Thus, once a neonate is located, we will back away and leave the neonate undisturbed. If a VIT is shed prior to parturition, we will radiolocate the adult doe no more than once per day on each successive day and search for fawns in an attempt to determine approximately when the doe actually gives birth. This will allow us to determine how many days a VIT shed prematurely. Neonate searches will typically last up to 30−45 minutes and will not exceed 1 hour. Past deer neonatal studies have reported minimal or no abandonment as a result of neonate capture, handling, and marking (Carstensen et al. 2003, Pojar and Bowden 2004, Bishop 2007). Powell et al. (2005) found no evidence of markinginduced abandonment, and they found that handling time and age-at-capture had no impact on neonatal survival. We therefore do not anticipate that our neonate searches will cause any direct or indirect harm to the neonates or their dams, particularly since we will not be handling fawns. 97 �Literature Cited: Anderson, C. R., and D. J. Freddy. 2008. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Study Plan, Colorado Division of Wildlife, Fort Collins, USA. Asher, G. W., and J. F. Smith. 1987. Induction of oestrus and ovulation in farmed fallow deer (Dama dama) by using progesterone and PMSG treatment. Journal of Reproduction and Fertility 81:113−118. Asher, G. W., J. L. Adam, R. W. James, and D. Barnes. 1988. Artificial insemination of farmed fallow deer (Dama dama): fixed-time insemination at a synchronized oestrus. Animal Production 47:487−492. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108−114. Bishop, C. J. 2007. Effect of enhanced nutrition during winter on the Uncompahgre Plateau mule deer population. Dissertation, Colorado State University, Fort Collins, USA. Bishop, C. J., D. J. Freddy, G. C. White, B. E. Watkins, T. R. Stephenson, and L. L. Wolfe. 2007. Using vaginal implant transmitters to aid in capture of mule deer neonates. Journal of Wildlife Management 71:945−954. Bowman, J. L., and H. A. Jacobson. 1998. An improved vaginal-implant transmitter for locating whitetailed deer birth sites and fawns. Wildlife Society Bulletin 26:295−298. Carstensen, M., G. D. DelGiudice, and B. A. Sampson. 2003. Using doe behavior and vaginal-implant transmitters to capture neonate white-tailed deer in north-central Minnesota. Wildlife Society Bulletin 31:634−641. Cook, R. C., J. G. Cook, D. L. Murray, P. Zager, B. K. Johnson, and M. W. Gratson. 2001. Development of predictive models of nutritional condition for Rocky Mountain Elk. Journal of Wildlife Management 65:973−987. Cook, R. C., T. R. Stephenson, W. L. Myers, J. G. Cook, and L. A. Shipley. 2007. Validating predictive models of nutritional condition for mule deer. Journal of Wildlife Management 71:1934−1943. Garrott, R. A, and R. M. Bartmann. 1984. Evaluation of vaginal implants for mule deer. Journal of Wildlife Management 48:646−648. Giessman, N. F., and C. J. Dalton. 1984. White-tailed deer fawn mortality in the southeastern Missouri Ozarks. Missouri Department of Conservation, Jefferson City, Pittman-Robertson Project W-13R-35. Johnson, B. K., T. McCoy, C. O. Kochanny, and R. C. Cook. 2006. Evaluation of vaginal implant transmitters in elk (Cervus elaphus nelsoni). Journal of Zoo and Wildlife Medicine 37:301−305. Johnstone-Yellin, T. L., L. A. Shipley, and W. L. Myers. 2006. Evaluating the effectiveness of vaginal implant transmitters for locating neonatal mule deer fawns. Wildlife Society Bulletin 34:338−344. Mylrea, G. E., A. W., English, R. C. Mulley, and G. Evans. 1992. Artificial insemination of farmed chital deer. Pages 334−337 in R. D. Brown, editor. The Biology of Deer. Springer-Verlag, New York, New York, USA. Nelson, T. A. 1984. Production and survival of white-tailed deer fawns on Crab Orchard National Wildlife Refuge. Thesis, Southern Illinois University, Carbondale, IL, USA. Pamplin, N. P. 2003. Ecology of Columbian black-tailed deer fawns in western Oregon. Thesis, Oregon State University, Corvallis, USA. Pojar, T. M., and D. C. Bowden. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550−560. Powell, M. C., G. D. DelGiudice, and B. A. Sampson. 2005. Low risk of marking-induced abandonment in free-ranging white-tailed deer neonates. Wildlife Society Bulletin 33:643−655. 98 �Stephenson, T. R., V. C. Bleich, B. M. Pierce, and G. P. Mulcahy. 2002. Validation of mule deer body composition using in vivo and post-mortem indices of nutritional condition. Wildlife Society Bulletin 30:557−564. Stephenson, T. R., K. J. Hundertmark, C. C. Schwartz, and V. Van Ballenberghe. 1998. Predicting body fat and body mass in moose with ultrasonography. Canadian Journal of Zoology 76:717−722. Stephenson, T. R., J. W. Testa, G. P. Adams, R. G. Sasser, C. C. Schwartz, and K. J. Hundertmark. 1995. Diagnosis of pregnancy and twinning in moose by ultrasonography and serum assay. Alces 31:167−172. Van Reenen, G. 1982. Field experience in the capture of red deer by helicopter in New Zealand with reference to post-capture sequela and management. Pages 408−421 in L. Nielsen, J. C. Haigh, and M. E. Fowler, editors. Chemical immobilization of North American wildlife. Wisconsin Humane Society, Milwaukee, Wisconsin, USA. 99 �Colorado Division of Wildlife July 2009 − June 2010 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 3001 4 Federal Aid Project No. W-185-R : : : : : Division of Wildlife Mammals Research Deer Conservation Effectiveness of a Redesigned Vaginal Implant Transmitter in Mule Deer Period Covered: July 1, 2009 − June 30, 2010 Authors: C. J. Bishop, C. R. Anderson, D. P. Walsh, E. J. Bergman, P. Kuechle, and J. Roth. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Our understanding of factors that limit mule deer (Odocoileus hemionus) populations may be improved by evaluating neonatal survival as a function of dam characteristics under free-ranging conditions, which generally requires that both neonates and dams are radiocollared. The most viable technique facilitating capture of neonates from radiocollared adult females is use of vaginal implant transmitters (VITs). To date, VITs have allowed research opportunities that were not previously possible; however, VITs are often expelled from adult females prepartum, which limits their effectiveness. We redesigned an existing VIT manufactured by Advanced Telemetry Systems (ATS; Isanti, MN) by lengthening and widening wings used to retain the VIT in an adult female. Our objective was to increase VIT retention rates and thereby increase likelihood of locating birth sites and newborn fawns. We placed the newly designed VITs in 59 adult female mule deer and evaluated the probability of retention to parturition and the probability of detecting newborn fawns. We also developed an equation for determining VIT sample size necessary to achieve a specified sample size of neonates. The probability of a VIT being retained until parturition was 0.766 (SE = 0.0605) and the probability of a VIT being retained to within 3 days of parturition was 0.894 (SE = 0.0441). In a similar study using the original VIT wings (Bishop et al. 2007), the probability of a VIT being retained until parturition was 0.447 (SE = 0.0468) and the probability of retention to within 3 days of parturition was 0.623 (SE = 0.0456). Thus, our design modification increased VIT retention to parturition by 0.319 (SE = 0.0765) and VIT retention to within 3 days of parturition by 0.271 (SE = 0.0634). Considering dams that retained VITs to within 3 days of parturition, the probability of detecting at least 1 neonate was 0.952 (SE = 0.0334) and the probability of detecting both fawns from twin litters was 0.588 (SE = 0.0827). We expended approximately 12 personhours per detected neonate. As a guide for researchers planning future studies, we found that VIT sample size should approximately equal the targeted neonate sample size. Our study expands opportunities for conducting research that links adult female attributes to productivity and offspring survival in mule deer. 63 �WILDLIFE RESEARCH REPORT EFFECTIVENESS OF A REDESIGNED VAGINAL IMPLANT TRANSMITTER FOR CAPTURING MULE DEER NEONATES FROM TARGETED ADULT FEMALES CHAD J. BISHOP, CHUCK R. ANDERSON, DANIEL P. WALSH, ERIC J. BERGMAN, PETER KUECHLE, AND JOHN ROTH P. N. OBJECTIVE To redesign vaginal implant transmitters (VITs) and evaluate their retention in free-ranging mule deer. SEGMENT OBJECTIVES 1. Evaluate rates of VIT retention to parturition and fawn capture success using the newly-designed wings in free-ranging mule deer. 2. Publish findings in Journal of Wildlife Management. INTRODUCTION Mule deer (Odocoileus hemionus) fawn production and neonatal survival is influenced by dam characteristics (e.g., body condition, disease status, habitat use). To understand fawn-dam relationships, manipulative field studies are needed that allow fawn production and survival to be estimated as a function of treatments applied to adult females. For example, a study evaluating the effectiveness of winter range habitat treatments on subsequent neonatal survival would require the capture of fawns from marked adult females that verifiably used, or did not use, the habitat treatments the previous winter(s). Such studies depend on a technique that enables newborn fawns to be captured from marked adult females. The most promising technique employed to capture neonates from marked adult females is use of vaginal implant transmitters (VITs), which are placed in the vagina of adult females during early to mid gestation. In theory, adult females retain VITs until parturition, at which point VITs are expelled at birth sites along with newborn fawns. Assuming VITs are routinely monitored, researchers can promptly radiolocate shed VITs and capture the newborn fawns. Recent applications of VITs in white-tailed deer (O. virginianus; Carstensen et al. 2003, Haskell et al. 2007, Saalfeld and Ditchkoff 2007), black-tailed deer (O. hemionus columbianus; Pamplin 2003), mule deer (Bishop et al. 2007, Haskell et al. 2007), and elk (Cervus elaphus; Johnson et al. 2006, Barbknecht et al. 2009) have been moderately successful. Vaginal implant transmitters also permit measurement of fetal survival in free-ranging populations, which has important implications in populations where stillborn mortality occurs (Bishop et al. 2007, 2008, 2009). An additional advantage of using VITs to capture neonates may be a reduction in sampling bias when compared to capture techniques that rely on opportunistic fawn capture (White et al. 1972, Ballard et al. 1998, Pojar and Bowden 2004). Opportunistic techniques are susceptible to bias because of unequal capture success among vegetation types, distances to roads, fawn ages, and stages of fawning. For example, if roads are used to conduct opportunistic searches, fawn capture probability will decline with increasing distance from a road and neonates will be disproportionately sampled in areas with high road densities. When using VITs, the distribution of radio-marked adult females carrying VITs determines where neonates are sampled. Inferences will be less biased with VITs than with opportunistic capture techniques if all VITs are monitored with equal intensity during fawning and the sample of radio-marked adult females was captured with minimal bias. Thus, VITs could have broad applicability regardless of whether study objectives require that fawns be captured from previously marked adult females. 64 �The most significant problem associated with VITs has been premature expulsion and subsequent failure to locate birth sites or newborn fawns, especially in mule deer (Johnstone-Yellin et al. 2006, Bishop et al. 2007, Haskell et al. 2007). The VIT has flexible, plastic wings coated with a soft silicone that induce pressure against the vaginal wall to retain the transmitter. The VIT design facilitates a quick, non-surgical insertion process and is safe for the animal (Johnson et al. 2006), but the current wing design is inadequate with respect to retention. Bishop et al. (2007) found that 43% (SE = 4.7) of VITs in mule deer shed prepartum, although the probability of capturing ≥1 fawn was relatively high (0.792, SE = 0.0847) when VITs shed only 1–3 days prepartum. They noted that 25% (SE = 4.1) of VITs shed >3 days prepartum and that retention probability declined as deer body size increased, indicating the retention wings were too small to be effective in larger deer. Based on these results, considerable oversampling of adult females would be required in the design of future projects to achieve a target sample size of fawns. That is, extra adult females would need to be sampled to offset those adult females that shed VITs prematurely. Oversampling, in this instance, is undesirable from an animal care and use perspective and unnecessarily expensive. Thus, our objective was to redesign the plastic-silicone retention wings of VITs to allow maximum retention in larger deer species. To date, the wings used to retain VITs have been purchased from a company in New Zealand (Carter Holt Harvey Plastic Products, Hamilton, New Zealand) that originally produced them for an application in the livestock industry (Bowman and Jacobson 1998). The company manufactured 1 large wing and 1 small wing; the former has been used in production of VITs for bison (Bison bison) and elk (Cervus elaphus) whereas the latter has been used in production of VITs for deer (Advanced Telemetry Systems, Isanti, MN). Advanced Telemetry Systems (ATS), in cooperation with wildlife researchers, made an initial effort in 2004 to lengthen the retention wings by adding resin to the wing tips. Using these VITs with antennas cut to the appropriate length, Haskell et al. (2007) reported that 81% of VITs (n = 21) in deer were retained until parturition. Retention improved but the aftermarket wing-modification was problematic because the wing tips were hard and thus not ideal for placement in the vaginal canal. That study provided justification to pursue further wing development. We therefore redesigned retention wings of VITs used in deer and similar-sized ungulates, fabricated a new production mold, and evaluated retention rates of VITs in free-ranging mule deer. STUDY AREA We conducted our research in Piceance Basin and on the Roan Plateau in northwest Colorado (Fig. 1). Our winter range study area comprised 4 study units distributed across much of the Piceance Basin. The 4 units ranged in size from 70 to 130 km2 and are referenced as South Magnolia, StorySprague, Ryan Gulch, and Yellow Creek (Fig. 1). These study units are part of a larger research study evaluating effects of natural gas development and mitigation on mule deer (Anderson and Freddy 2008). Winter range habitat comprised predominantly pinyon pine (Pinus edulis) and Utah juniper (Juniperus osteosperma) and secondarily big sagebrush (Artemisia tridentata), serviceberry (Amelanchier utahensis), mountain mahogany (Cercocarpus montanus), bitterbrush (Purshia tridentata), and rabbitbrush (Chrysothamnus spp.). Drainage bottoms were characterized by stands of big sagebrush, saltbush (Atriplex spp.), and black greasewood (Sarcobatus vermiculatus), with the majority of the primary drainage bottoms having been converted to irrigated, grass hay fields. Elevations ranged from 1,860 m at Piceance Creek in Ryan Gulch to 2,280 m in Yellow Creek and Story-Sprague study units. Our summer range study area comprised roughly 1,700 km2 across the Roan Plateau and Piceance Basin (Fig. 1). Principal summer range habitat types included aspen (Populus tremuloides), mountain shrub, oakbrush (Quercus gambellii), big sagebrush, and pinyon-juniper. Serviceberry, snowberry (Symphoricarpos spp.), and chokecherry (Prunus virginiana) were common species in mountain shrub communities. Elevation ranged from 2,000 m in Piceance Creek at the mouth of Story Gulch to 2,600 m on Roan Plateau. 65 �METHODS We worked with ATS personnel to redesign the M3930 VIT presently manufactured by ATS. The existing M3930 has been described in detail elsewhere (Bowman and Jacobson 1998, Carstensen et al. 2003, Johnstone-Yellin et al. 2006, Bishop et al. 2007). Our redesign included changes to the retention wings and the means by which they are attached to the transmitter body (Fig. 2). Specifically, we modified dimensions of the retention wings by lengthening them from 57 mm to 68 mm and widening them from 9 mm to 13 mm. We also added ridges to the wing surface as means to increase probability of retention to parturition. The wings were made of flexible plastic encased in silicone. We initially produced a small number of the newly-designed wings using a relatively inexpensive prototype mold, which met our target specifications and therefore was deemed acceptable. We then manufactured a production mold, necessary to produce a large number of the wings. We incorporated ejector pins into the VIT design that allow wings to be attached to the VIT transmitter body in the field. In the original design, wings were permanently affixed to the transmitter body during the VIT assembly process. Although we only used one wing size in this study, field-attachment will allow researchers to use more than one wing size or style, without purchasing extra transmitters, if additional production molds are manufactured over time. For each wing design (i.e., production mold), extra wings could be inexpensively purchased and available in the field to affix to the fixed number of transmitter bodies. Researchers could then individually fit VITs to animals in the field much in the same way radiocollars are individually fitted. During late February and early March, 2009, we captured 60 adult female deer utilizing helicopter net guns (Barrett et al. 1982, Krausman et al. 1985, White and Bartmann 1994) in conjunction with ongoing research addressing other objectives (Anderson and Freddy 2008). We captured 20 deer each in Ryan Gulch and Yellow Creek, and 10 deer each in South Magnolia and Story-Sprague study units (Fig. 1). Captured deer were hobbled, blind-folded, and ferried ≤5 km by helicopter to a central handling location. For each captured deer, we used transabdominal ultrasonography (SonoVet 2000, Universal Medical Systems, Bedford Hills, NY) to determine pregnancy status and number of fetuses (Stephenson et al. 1995, Bishop et al. 2007, Bishop et al. 2009). We also measured rump fat depth of each deer using ultrasonography and estimated a body condition score using palpation to estimate percent body fat (Stephenson et al. 2002, Cook et al. 2007). We measured mass by placing each deer on a stretcher and attaching the stretcher to a scale supported by a steel frame. We measured chest girth by placing a cloth tape around the chest immediately posterior to the front shoulders and recording measurement when deer exhaled. Last, we measured hind foot length of each deer and estimated age by evaluating tooth replacement and wear (Severinghaus 1949, Robinette et al. 1957). This aging technique is susceptible to measurement error (Hamlin et al. 2000). However, two trained observers, each with experience aging >1,000 deer in the field, estimated age of all deer in this study to minimize error and to insure that relative age differences across all deer in our sample were correctly captured in the data. We performed handling procedures in a wall-frame tent to create a dim environment for viewing ultrasound imagery. We fitted each pregnant deer with a radiocollar and VIT. Collar transmitters were turned off on Saturdays and Mondays to extend battery life for meeting other research objectives (Anderson and Freddy 2008). Each collar was equipped with a mortality sensor and store-on-board global positioning system (GPS). Mortality sensors were programmed to switch signal transmission from 60 pulses to 120 pulses per minute after remaining motionless for 8 hours. Each VIT had a temperature-sensitive switch and a pre-cut antenna (6 cm in length) with antenna tip encapsulated in a resin bead to eliminate sharp edges. The temperature-sensitive switch caused the VIT to increase pulse rates from 40 pulses to 80 pulses per minute when the temperature dropped below 32° C, which was indicative of VIT expulsion. We sterilized VITs in a chlorhexidine solution prior to insertion in the field. We inserted VITs using a clear, plastic swine vaginoscope (Jorgensen Laboratories, Inc., Loveland, CO) and alligator forceps. The 66 �vaginoscope was 15.2 cm long with a 1.59 cm internal diameter and had a smoothed end to minimize vaginal trauma. We placed vaginoscopes and alligator forceps in cold sterilization containers with chlorhexidine solution between each use and used a new pair of surgical gloves to handle the vaginoscope and VIT for each deer, and we applied lidocaine topically to the deer’s vagina to minimize irritation during VIT insertion. To insert a VIT, we folded the wings together and placed the VIT into the end of the vaginoscope. We liberally applied sterile KY Jelly to the scope and inserted it into the vaginal canal until the tip of the VIT antenna was approximately flush with the vulva. We used previous field experience to guide insertion distance and antenna length (Bishop et al. 2007). We extended alligator forceps through the vaginoscope to firmly hold the VIT in place while the scope was pulled out from the vagina. During winter and spring, we monitored live-dead status and general location of radiocollared adult females daily from the ground, except when collars were inactive, and biweekly from the air via fixed-wing aircraft. During June, we checked VIT signal status each morning of the week that radiocollars were active by aerially locating each radiocollared doe carrying a VIT. We began flights at approximately 0630 hours and completed them by 0900–1100 hours. Early flights were necessary to detect fast signals because temperature sensors of VITs expelled in open habitats and subject to sunlight often exceeded 32° C by mid-day, which caused VITs to switch back to a slow (i.e., prepartum) pulse (Newbolt and Ditchkoff 2009). When we detected a fast (i.e., postpartum) pulse rate, we ground-located the VIT and radiocollared doe in ≤3 hours using very high frequency (VHF) receivers and directional antennae. We attempted to observe behavior of the collared adult female, establish whether the VIT was shed at a birth site, and search for fawns in the vicinity of the adult female and expelled VIT. In cases where the dam moved away from the VIT (i.e., >200 m), we located the VIT to determine whether shedding occurred at a birth site and whether any stillborn fawns were present and subsequently located the collared dam to search for fawns at her location. We attempted to account for each dam’s fetus(es) as live or stillborn. We typically worked in pairs, which allowed us to effectively partition effort across the study area while maintaining efficiency when searching for neonates (i.e., two people were more effective locating a hidden neonate than one person). We described effort associated with locating fawns by calculating the number of person-hours per fawn. We also quantified cost per fawn by considering all operating and personnel expenses, including capture and VIT costs for adult females. All deer capture and handling procedures and use of VITs were approved by Colorado Division of Wildlife’s (CDOW) Institutional Animal Care and Use Committee (Project # 17-2008). We assigned the fate of each VIT to one of 4 categories: 1) retained (i.e., VIT expelled during parturition), 2) nearly-retained (i.e., VIT expelled ≤3 days prepartum), 3) not retained (i.e., VIT expelled >3 days prepartum), or 4) censored. We considered a VIT to be retained if it was expelled at or near a birth site in conjunction with parturition. For 75% of retention events, we located the VIT at a birth site and located neonate(s) near the VIT or in close proximity to the dam. In other cases, the VIT was not at a birth site but we readily found the dam and her newborn fawn(s) nearby, sometimes at a birth site 10−100 m from the VIT. In these situations, we considered a VIT retained if we documented <1-day-old fawn(s) <24 hours after the VIT was expelled. Finally, on two occasions, we considered a VIT retained because it was located at an evident birth site even though we could not locate fawns. Birth sites appeared as atypically large deer beds with soil appearing damp and with forbs and grasses flattened and radiating outward, consistent with a deer licking the site clean. On some occasions, fawns and/or placental remains were still present at birth sites when we arrived, providing positive confirmation of birth site characteristics. We distinguished VITs expelled ≤3 days prepartum as nearly-retained because they provided useful information for locating fawns, consistent with Bishop et al. (2007). We documented such cases by locating a dam’s neonate(s) one or more days after the VIT was expelled and comparing neonate age to VIT expulsion date. We estimated neonate age using hoof characteristics, condition of the umbilical cord, pelage, and behavior (Haugen and Speake 1958, Robinette et al. 1973, Sams et al. 1996, 67 �Pojar and Bowden 2004). We assumed a VIT was shed >3 days prepartum if the VIT was not at an evident birth site and we documented ≥2 of the following characteristics: 1) the adult female was located with other deer during repeated relocations for >3 days after the VIT was shed, 2) the adult female exhibited no behavioral cues indicating she had a fawn, 3) the adult female was noticeably still pregnant, and 4) we failed to locate a neonate following repeated searches for ≥1 week after the VIT was shed. We censored VITs from our retention analysis when adult females died prior to parturition or when adult females were located on private land that we did not have permission to access. In either case, we were unable to evaluate VIT retention to parturition. All females dying prior to parturition were still carrying the VITs upon death. We modeled VIT retention probability using a generalized logits model (i.e., multinomial logistic regression) in PROC LOGISTIC in SAS (SAS Institute, Cary, NC). We evaluated goodness-of-fit of the global model (i.e., model containing each predictor variable) by dividing model deviance by its degrees of freedom. We considered 3 levels of retention consistent with our description above (i.e., retained, nearlyretained, not retained) and we removed all censors from the dataset prior to analysis. Our primary purpose for this analysis was to evaluate whether our VIT design modifications increased VIT retention probability in larger deer. Our design modifications were based on the observation by Bishop et al. (2007) that VIT retention probability declined as deer body size increased. We modeled VIT retention as a function of mass (kg), hind foot length (cm), chest girth (cm), adult female age (yr), and body fat (%). We considered only linear models because we lacked a rationale for evaluating higher-order polynomial functions. Several of the variables we considered in our analysis were likely correlated because they represented different ways of expressing deer body size. We did not expect models comprising each of these variables to receive more support than simpler models. Thus, we focused our candidate model set on models with one or two variables. We evaluated all single-variable models plus we evaluated twovariable models that included age with each other variable. Age partially related to deer body size but age also related to number of times a female had previously given birth and possibly to behavioral differences among deer, either of which could have influenced retention probability. Thus, age tested hypotheses about retention probability that were not just related to body size or condition. We also considered several models with ≥3 variables to determine whether there was any support for models with higher numbers of parameters. We evaluated 13 models in total and we selected among models using Akaike’s Information Criterion adjusted for sample size (AICc; Burnham and Anderson 2002). We modelaveraged beta parameter estimates to incorporate model selection uncertainty when evaluating whether VIT retention probability varied as a function of the variables in our analysis. We did not model-average real parameter estimates because each of our predictor variables was continuous. We modeled fawn detection probability based on adult females that retained or nearly retained VITs. We planned to conduct separate analyses for singleton and twin litters, but we achieved perfect detection with singleton litters. We therefore modeled fawn detection probability considering only females with twin fetuses using a generalized logits model in SAS, and we evaluated goodness-of-fit by dividing model deviance by its degrees of freedom. We used 3 detection levels (0, 1, 2 fawns) and we modeled detection as a function of VIT retention status (retained vs. nearly-retained), VIT shed-day, adult female age, and vegetative cover at VIT expulsion site. Shed-day distinguished between VITs detected on fast pulse on Sundays and Tuesdays (dummy code = 1) and VITs detected on fast pulse during Wednesday−Friday (dummy code = 0). We used the shed-day variable to evaluate whether delayed response time, caused by our inability to monitor deer on Saturdays and Mondays, influenced our ability to detect fawns. We included adult female age in our analysis to evaluate if older females may have been more experienced at hiding fawns. Last, we used vegetative cover to evaluate if fawns were more difficult to detect in heavier cover. We expressed vegetative cover categorically as low, medium, or high based on a visual assessment at the site. Low cover class was characterized by limited understory and overstory vegetation with minimal visual obstruction at ground level (e.g., sparsely-vegetated grass, sagebrush, or mountain shrub slopes). Medium cover class was characterized by moderate to heavy 68 �vegetative cover within 1 m of the ground but limited cover above 1 m (e.g., typical sagebrush, mountain shrub sites). High cover class comprised moderate to heavy vegetative cover from ground level up to > 1 m with nearly complete visual obstruction (e.g., oakbrush, aspen-mountain shrub, dense serviceberry). We evaluated all single-variable models in addition to 4 models with ≥2 variables to determine whether there was any support for models with higher numbers of parameters. We evaluated 9 models in total and we selected among models using Akaike’s Information Criterion adjusted for sample size (AICc; Burnham and Anderson 2002). We did not model-average parameter estimates because it would have resulted in 10 different estimates of each level of fawn detection probability for a total of 30 probability estimates. These differences were not supported by the model selection results. We used our VIT retention and fawn detection probabilities to guide calculation of VIT sample sizes for planning future neonatal studies. We expressed the expected number of neonates to be encountered from a sample of VITs as: , where n N ,30 n v1r s SAdF RvlT TAdF P 1 1r1.vins P21r ,.vms P1 1Si ngl. = neonate sample size. = sample size of adult females with VITs. = probability an adult female survives to parturition and is accessible. = probability an adult female retains her VIT to within 3 days of parturition given she survives to parturition and is accessible (i.e., VIT is retained or nearly retained). = probability adult female has twin fetuses. = probability 1 fawn is detected given an adult female retains her VIT and has twin fetuses. = probability 2 fawns are detected given an adult female retains her VIT and has twin fetuses. = probability 1 fawn is detected given an adult female retains her VIT and has one fetus. Since we had perfect detection with singleton litters and observed a high probability of detecting at least 1 fawn from twin litters, we simplified the above equation to: where is the probability of detecting at least 1 fawn, irrespective of litter size. Thus, given a targeted sample size of neonates, the estimated number of VITs required can be calculated as: We incorporated our estimates into the above equation to provide guidance for planning future studies. RESULTS AND DISCUSSION A retention wing of 1 VIT snapped at its base when the wings were squeezed together for placement into a vaginoscope, prior to insertion into a deer. No other retention wings exhibited any 69 �cracking or weakness when squeezed together, even after VITs were recovered from animals during spring and summer. Thus, we found this to be an isolated incident, and our resulting sample size was 59 deer with VITs. The probability that an adult female receiving a VIT in winter survived to parturition and was accessible (SAdF) was 0.797 (SE = 0.0529). We observed 9 adult female mortalities during winter and spring, and there was no evidence to suggest VITs were related to the mortality events. Four of the mortalities occurred within 1 week of capture and were likely capture-related. We were unable to groundmonitor 2 other adult females during the fawning period because they were located on private land that we did not have permission to access. One other adult female was inadvertently deleted from the aerial monitoring list due to miscommunication. We censored these 12 deer from our analysis of VIT retention because they did not permit evaluation of VIT retention to parturition, resulting in a sample size of 47 deer. Our global model of VIT retention probability (k = 12) adequately fit the data (deviance/df = 0.670, P = 0.991). The model of VIT retention probability with the lowest AICc included only the intercept (k = 2, ∆AICc = 0.00, wi = 0.331), although the model with deer age received some support (k = 4, ∆AICc = 1.42, wi = 0.163; Table 1). There was slight evidence that retention probability was lower in older deer ( = 0.169, SE = 0.256; Fig. 3). Also, there was slight evidence that retention probability was lower in larger deer ( = 0.086, SE = 0.171; Table 1). Based on the intercept-only model, the probability of a VIT being expelled during parturition (i.e., retained) was 0.766 (SE = 0.0605) and the probability of a VIT being expelled ≤3 days prepartum (i.e., nearly-retained) was 0.128 (SE = 0.0477). Thus, the probability of a VIT being retained to within 3 days of parturition (RVIT) was 0.894 (SE = 0.0441). Our global model of fawn detection probability (k = 12) adequately fit the data (deviance/df = 0.846, P = 0.730). The model of fawn detection probability with the lowest AICc included only the intercept (k = 2, ∆AICc = 0.00, wi = 0.600), whereas the model with the next lowest AICc included the VIT shed-day variable (k = 4, ∆AICc = 1.80, wi = 0.244; Table 2). Thus, we observed some evidence that fawn detection probability was influenced by our inability to monitor deer 2 days of the week (f1sh ,3d - da y,si n9 l ,3ron = 0.537, SE = 0.738). The probability of detecting twins was 0.688 (SE = 0.114) when we located adult females <24 hours after their VITs switched to fast pulse, whereas twin detection probability was 0.500 (SE = 0.115) when our response time was delayed due to irregular monitoring. There was no evidence that probability of fawn detection was influenced by dam age or vegetative cover. Also, fawn detection probability did not meaningfully differ between females with retained and nearlyretained VITs. We detected 58 neonates and 2 stillborns from 42 adult females (1.4 neonates/female) that retained or nearly retained VITs. We detected a neonate from each adult female that had 1 fetus (P11Singl. = 1.0 , n = 8). For adult females with twin fetuses (n = 34), based on the intercept-only model, the probability of detecting 1 neonate (P 1 1Tw i n.s) was 0.353 (SE = 0.0803) and the probability of detecting twins (P21T 1,vin.s) was 0.588 (SE = 0.0827). Combining litter sizes, the probability of detecting at least 1 ) was 0.952 (SE = 0.0334). The probability of an adult female having twin fetuses (TAdF) neonate ( was 0.810 (SE = 0.0613). On average, we located one neonate or stillborn per VIT in our initial sample (nNeo = 60, nVITs = 59). Thus, inputting our estimates into our sample size equation, we found that VIT sample size should roughly equal the targeted neonate sample size: E [n N. ol E [n N. ol n vir , = (0.80 )(0.89) [0.95 + (0.81)(0.59)] = 1.02 70 �We expended roughly 700 person-hours during the fawning period to locate 58 neonates and 2 stillborns, or approximately 12 person-hours per fawn located. This estimate includes hours spent searching for fawns from adult females that expelled VITs >3 days prepartum, although we were never successful in these attempts. We expended $31,000 to net-gun our sample of adult females, $15,000 on VITs, $10,000 on fixed wing monitoring, and $20,000 on personnel. Thus, we expended approximately $1,267 per neonate located. We did not include adult female radio collars in this cost estimate because we used GPS collars to meet other research objectives, yet VHF collars would have sufficed for locating neonates. Assuming VHF collars were used on adult females at a rate of $250 per collar, our cost estimate is approximately $1,520 per fawn. Our wing modification increased VIT retention in adult female mule deer. Our results are consistent with Haskell et al. (2007), who observed 81% retention (17/21) in the final year of their study after lengthening VIT wings and preventing antennas from protruding >1 cm past the vulva. Our study expanded on Haskell et al. (2007) by incorporating VIT wing modifications into the manufacturing process and conducting a focused field evaluation of those modifications. Investigators using the original VIT wing design in mule deer observed much lower rates of retention than we observed (Johnstone-Yellin et al. 2006, Bishop et al. 2007, Haskell et al. 2007). Using the original design, Bishop et al. (2007) found that the probability of VIT expulsion during parturition was 0.447 (SE = 0.0468), and the probability of VIT expulsion during parturition or ≤3 days prepartum was 0.623 (SE = 0.0456). We employed the same methodology as Bishop et al. (2007), except for the wing modification. Our study area was 100 km north of where Bishop et al. (2007) conducted their study. Assuming the 2 studies are comparable, our wing modification increased VIT retention to parturition by 0.319 (SE = 0.0765) and VIT retention to within 3 days of parturition by 0.271 (SE = 0.0634). The intercept-only model of VIT retention probability received the most Akaike weight, which is partly a reflection of our limited sample size. However, overall high rates of retention likely explain why we did not observe any strong relationships between VIT retention and deer body size. Bishop et al. (2007) found that larger deer were more likely to expel VITs prematurely, which was our basis for modifying VIT wings and conducting this study. Our results suggest the wing modifications effectively reduced premature expulsion, particularly in larger deer (Fig. 4). We documented a high probability of detecting at least 1 fawn from adult females that retained or nearly retained VITs, regardless of litter size. When a VIT was shed and evidence suggested the adult female was near parturition or had already given birth, we conducted intense searches up to 1 hour in length for successive days until a fawn was found. Thus, irrespective of vegetative cover or other covariates we assessed, we usually found a fawn when a VIT was adequately retained because it focused our search effort. Our likelihood of detecting twins was somewhat lower, in part because of our irregular monitoring schedule. However, other factors explain why twin detection probability was lower. First, our search intensity decreased when searching for a second fawn. For example, if we had searched most of an hour before detecting the first fawn, we typically limited our search time for a second fawn to minimize our disturbance to the adult female. Second, we did not place radio collars on fawns, and therefore, we could not relocate radiocollared fawns to search for their siblings. The technique of relocating a radiocollared fawn to locate its sibling was found to be successful in a previous study in Colorado (Bishop et al. 2009). During this earlier study, when a dam was known to have twin fetuses yet only one fawn was located and radiocollared during the initial capture attempt, the sibling fawn was found 45% of the time (10/22) by relocating the initial radiocollared fawn 1−2 days post-capture (C. J. Bishop, CDOW, unpublished data). Based on this rate, we would expect our probability of detecting both fawns from twin litters to be roughly 0.77 had we radiocollared fawns during our study. We found that our sample size of detected neonates roughly equaled our sample size of VITs, which provides a useful guide for planning future research using our modified wing design. However, 71 �this recommendation may overestimate VIT sample size because of our lower rate of twin detection and because adult female survival was lower than we anticipated. Fortunately, accessibility of adult females was higher than expected considering we lacked permission to access a large tract of land in the middle of summer range. Bishop et al. (2007) observed 0.97 survival of adult females to parturition and 0.99 were accessible during fawning (SAdF = 0.95). Adult female survival and accessibility is specific to study area. Twinning probability may also vary regionally. We therefore recommend use of the following equation for planning VIT sample size that incorporates information specific to the study area or region of interest: E[n Naol Bishop et al. (2007) expended 7 person-hours per captured fawn from adult females with successful VITs, 16 person-hours per fawn from females with partially successful VITs, and 42 personhours per fawn from females with failed VITs and females not receiving VITs. Given their observed VIT success rates, Bishop et al. (2007) would have required approximately 1,315 person-hours to locate 60 neonates, or 22 person-hours per fawn. Assuming these studies are comparable, increased VIT success associated with our modified wing design resulted in a 45% reduction in labor required to locate a fawn from a radiocollared adult female. The VIT technique is effective but expensive to employ. Actual cost of the technique, however, depends on what costs are already incurred to meet other research objectives. For example, in Colorado and elsewhere, researchers have begun estimating late-winter deer body condition as a response variable to accompany survival estimates. In these cases, adult female capture and radio collar costs are already accounted for in the base study, and thus, incorporation of VITs to facilitate neonate capture becomes much more cost-effective. In our study, where adult female capture and collar costs were covered by ongoing research efforts, the added cost of incorporating VITs and neonate capture was $750 per fawn. SUMMARY Use of VITs in well-designed field studies should increase our understanding of factors limiting deer populations by allowing investigators to link fawn production and survival to dam characteristics under free-ranging conditions. A primary drawback of VITs in deer has been the failure of many adult females to retain VITs to parturition. We increased VIT retention in mule deer by lengthening and widening wings used to retain a VIT in the vaginal canal. Researchers employing VITs with our modified wing design should require minimal oversampling to offset failures caused by early expulsion, thereby rendering the technique more cost-effective and reliable. Our findings provide explicit guidance for planning a fetal-neonatal deer study involving VITs. The question remains as to whether premature expulsion of VITs can be eliminated in mule deer. We observed modest evidence that deer expelling VITs >3 days prepartum were older and larger than deer that retained or nearly-retained VITs. We therefore recommend manufacturing slightly larger wings for large, older mule deer (e.g., >65 kg and >5 yrs old) as a possible strategy to further investigate VIT retention. LITERATURE CITED Anderson, C. R., and D. J. Freddy. 2008. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Study Plan, Colorado Division of Wildlife, Fort Collins, USA. 72 �Ballard, W. B., H. A. Whitlaw, D. L. Sabine, R. A. Jenkins, S. J. Young, and G. J. Forbes. 1998. Whitetailed deer, Odocoileus virginianus, capture techniques in yarding and non-yarding populations in New Brunswick. Canadian Field-Naturalist 112:254−261. Barbknecht, A. E., W. S. Fairbanks, J. D. Rogerson, E. J. Maichak, and L. L. Meadows. 2009. Effectiveness of vaginal-implant transmitters for locating elk parturition sites. Journal of Wildlife Management 73:144−148. Barrett, M. W., J. W. Nolan, and L. D. Roy. 1982. Evaluation of a hand-held net-gun to capture large mammals. Wildlife Society Bulletin 10:108−114. Bishop, C. J., D. J. Freddy, G. C. White, B. E. Watkins, T. R. Stephenson, and L. L. Wolfe. 2007. Using vaginal implant transmitters to aid in capture of mule deer neonates. Journal of Wildlife Management 71:945−954. Bishop, C. J., G. C. White, D. J. Freddy, B. E. Watkins, and T. R. Stephenson. 2009. Effect of enhanced nutrition on mule deer population rate of change. Wildlife Monographs 172:1−28. Bishop, C. J., G. C. White, and P. M. Lukacs. 2008. Evaluating dependence among mule deer siblings in fetal and neonatal survival analyses. Journal of Wildlife Management 72:1085−1093. Bowman, J. L., and H. A. Jacobson. 1998. An improved vaginal-implant transmitter for locating whitetailed deer birth sites and fawns. Wildlife Society Bulletin 26:295−298. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. Second Edition. Springer-Verlag, New York, New York, USA. Carstensen, M., G. D. DelGiudice, and B. A. Sampson. 2003. Using doe behavior and vaginal-implant transmitters to capture neonate white-tailed deer in north-central Minnesota. Wildlife Society Bulletin 31:634−641. Cook, R. C., T. R. Stephenson, W. L. Myers, J. G. Cook, and L. A. Shipley. 2007. Validating predictive models of nutritional condition for mule deer. Journal of Wildlife Management 71:1934−1943. Hamlin, K. L., D. F. Pac, C. A. Sime, R. M. DeSimone, and G. L. Dusek. 2000. Evaluating the accuracy of ages obtained by two methods for Montana ungulates. Journal of Wildlife Management 64:441−449. Haskell, S. P., W. B. Ballard, D. A. Butler, N. M. Tatman, M. C. Wallace, C. O. Kochanny, and O. J. Alcumbrac. 2007. Observations on capturing and aging deer fawns. Journal of Mammalogy 88:1482−1487. Haugen, A. O., and D. W. Speake. 1958. Determining age of young fawn white-tailed deer. Journal of Wildlife Management 22:319−321. Johnson, B. K., T. McCoy, C. O. Kochanny, and R. C. Cook. 2006. Evaluation of vaginal implant transmitters in elk (Cervus elaphus nelsoni). Journal of Zoo and Wildlife Medicine 37:301−305. Johnstone-Yellin, T. L., L. A. Shipley, and W. L. Myers. 2006. Evaluating the effectiveness of vaginal implant transmitters for locating neonatal mule deer fawns. Wildlife Society Bulletin 34:338−344. Krausman, P. R., J. J. Hervert, and L. L. Ordway. 1985. Capturing deer and mountain sheep with a netgun. Wildlife Society Bulletin 13:71−73. Newbolt, C. H., and S. S. Ditchkoff. 2009. Effects of environmental conditions on performance of vaginal implant transmitters. Journal of Wildlife Management 73:303−305. Pamplin, N. P. 2003. Ecology of Columbian black-tailed deer fawns in western Oregon. Thesis, Oregon State University, Corvallis, USA. Pojar, T. M., and D. C. Bowden. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550−560. Robinette, W. L., C. H. Baer, R. E. Pillmore, and C. E. Knittle. 1973. Effects of nutritional change on captive mule deer. Journal of Wildlife Management 37:312−326. Robinette, W. L., D. A. Jones, G. Rogers, and J. S. Gashwiler. 1957. Notes on tooth development and wear for Rocky Mountain mule deer. Journal of Wildlife Management 21:134−153. 73 �Saalfeld, S. T., and S. S. Ditchkoff. 2007. Survival of neonatal white-tailed deer in an exurban population. Journal of Wildlife Management 71:940−944. Sams, M. G., R. L. Lochmiller, E. C. Hellgren, W. D. Warde, and L. W. Varner. 1996. Morphometric predictors of neonatal age for white-tailed deer. Wildlife Society Bulletin 24:53−57. Severinghaus, C. W. 1949. Tooth development and wear as criteria of age in white-tailed deer. Journal of Wildlife Management 13:195−216. Stephenson, T. R., V. C. Bleich, B. M. Pierce, and G. P. Mulcahy. 2002. Validation of mule deer body composition using in vivo and post-mortem indices of nutritional condition. Wildlife Society Bulletin 30:557−564. Stephenson, T. R., J. W. Testa, G. P. Adams, R. G. Sasser, C. C. Schwartz, and K. J. Hundertmark. 1995. Diagnosis of pregnancy and twinning in moose by ultrasonography and serum assay. Alces 31:167−172. White, G. C., and R. M. Bartmann. 1994. Drop nets versus helicopter net guns for capturing mule deer fawns. Wildlife Society Bulletin 22:248−252. White, M., F. F. Knowlton, and W. C. Glazener. 1972. Effects of dam-newborn fawn behavior on capture and mortality. Journal of Wildlife Management 36:897−906. Prepared by _______________________ Chad J. Bishop, Mammals Research Leader 74 �Figure 1. Location of winter and summer range study areas in Piceance Basin and Roan Plateau, northwest Colorado. Winter range study units where we captured and radio-marked mule deer are noted as: YC = Yellow Creek, RG = Ryan Gulch, SM = South Magnolia, and SS = Story-Sprague. 75 �A B 60° (2) R 278 (2) - - - - 1.000 - - --, ,871 0 .581 2'18 (2) ORE 040 ,640 (2) NOMI NAL 743 (2) I R 139 (2) 1 740 :21 17 ~ - - - - - - - - - 2.67-3 f68 ITT[Jlj - - - - - - - - ---1 Figure 2. Three-dimensional view (A) and dimensions (B) of a modified retention wing used to retain vaginal implant transmitters in adult female mule deer. The displayed dimensions at bottom include a nylon core with an elastomeric overmold that protects deer from any sharp or rigid edges. 76 �- - - - - - - - - - -- --- - - - - ----' 1 0.9 C 0 0.8 ' ' +"' C QJ 0.7 +"' QJ I,.. I- 0.6 > 0.5 0 > +"' 0.4 .c re .c 0.3 0 I,.. c.. ' ' ' '' '' 0.2 0.1 '' '' 0 1.5 2.5 3.5 4.5 5.5 6.5 7.5 '' 8.5 Deer age (yr) Figure 3. Estimated probability and 95% confidence interval of adult female mule deer retaining vaginal implant transmitters (VITs) to within 3 days of parturition as a function of deer age in northwest Colorado. 77 9.5 �--- - ----------------~- _______ ... -------- 1 0.9 0.8 - - - -..... ..... ..... C: .2 +C: 0.7 ' ' Q) +- ... Q) 0.6 I- > 0.5 0 >, :!::: .c ('Cl .c ...0 a. 0.4 0.3 0.2 0.1 0 50 55 65 60 70 75 80 Deer body mass (kg) Figure 4. Estimated probabilities and 95% confidence intervals of adult female mule deer retaining vaginal implant transmitters (VITs) to within 3 days of parturition as a function of deer body mass in Colorado using original (solid line, Bishop et al. 2007) and modified (dashed line, this study) VIT retention wings. 78 �Table 1. Model selection results, based on Akaike’s Information Criterion with small sample size correction (AICc), from an analysis of vaginal implant transmitter (VIT) retention in adult female mule deer as a function of adult female age (yr), mass (kg), hind foot length (cm), chest girth (cm), and body fat (%) in northwest Colorado, USA, 2009. Model k AICc ∆AICc wi Intercept only 2 70.58 0.00 0.331 Age 4 72.00 1.42 0.163 Foot length 4 72.88 2.30 0.105 Age, fat 6 72.96 2.39 0.100 Mass 4 73.57 2.99 0.074 Fat 4 73.66 3.08 0.071 Chest girth 4 73.79 3.21 0.066 Age, chest girth 6 75.10 4.52 0.035 Age, foot length 6 75.45 4.88 0.029 Age, mass 6 76.32 5.74 0.019 Age, foot length, chest girth 8 78.53 7.95 0.006 79 �Table 2. Model selection results, based on Akaike’s Information Criterion with small sample size correction (AICc), from an analysis of fawn detection probability associated with adult females that retained or nearly-retained vaginal implant transmitters (VITs) in northwest Colorado, 2009. We modeled detection probability as a function of VIT retention status (retained vs. nearly-retained), adult female age (yr), the day of the week VITs were shed (i.e., shed-day), and amount of vegetative cover at VIT shed sites. We evaluated detection probability relative to shed day because we were unable to monitor radio signals on Saturdays and Mondays. Model k AICc ∆AICc wi Intercept only 2 61.94 0.00 0.600 Shed-day 4 63.74 1.80 0.244 Retention status 4 66.07 4.13 0.076 Age 4 66.26 4.32 0.069 Cover 6 70.46 8.52 0.008 Shed-day, cover 8 73.22 11.28 0.002 Shed-day, cover, retention status 10 79.53 17.59 0.000 Age, shed-day, cover 10 80.24 18.30 0.000 80 �Colorado Division of Parks and Wildlife July 2010 − June 2011 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 3001 4 Federal Aid Project No. W-185-R : : : : : Division of Parks and Wildlife Mammals Research Deer Conservation Effectiveness of a Redesigned Vaginal Implant Transmitter in Mule Deer Period Covered: July 1, 2010 − June 30, 2011 Authors: C. J. Bishop, C. R. Anderson, D. P. Walsh, E. J. Bergman, P. Kuechle, and J. Roth. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT We completed all field work on this project and prepared draft manuscripts for publication prior to FY 10-11. As explained in our Segment Narrative for FY 10-11, our final objective for this project was to publish results of the study in Journal of Wildlife Management (JWM). Our manuscript was accepted for publication in JWM on March 23, 2011. The manuscript will be published in the November 2011 issue of JWM. 71 �WILDLIFE RESEARCH REPORT EFFECTIVENESS OF A REDESIGNED VAGINAL IMPLANT TRANSMITTER FOR CAPTURING MULE DEER NEONATES FROM TARGETED ADULT FEMALES CHAD J. BISHOP, CHARLES R. ANDERSON, JR., DANIEL P. WALSH, ERIC J. BERGMAN, PETER KUECHLE, AND JOHN ROTH P. N. OBJECTIVE To redesign vaginal implant transmitters (VITs) and evaluate their retention in free-ranging mule deer. SEGMENT OBJECTIVES 6. Publish findings in Journal of Wildlife Management. INTRODUCTION Mule deer (Odocoileus hemionus) fawn production and neonatal survival is influenced by dam characteristics (e.g., body condition, disease status, habitat use). To understand fawn-dam relationships, manipulative field studies are needed that allow fawn production and survival to be estimated as a function of treatments applied to adult females. For example, a study evaluating the effectiveness of winter range habitat treatments on subsequent neonatal survival would require the capture of fawns from marked adult females that verifiably used, or did not use, the habitat treatments the previous winter(s). Such studies depend on a technique that enables newborn fawns to be captured from marked adult females. The most promising technique employed to capture neonates from marked adult females is use of vaginal implant transmitters (VITs), which are placed in the vagina of adult females during early to mid gestation. In theory, adult females retain VITs until parturition, at which point VITs are expelled at birth sites along with newborn fawns. Assuming VITs are routinely monitored, researchers can promptly radiolocate shed VITs and capture the newborn fawns. Recent applications of VITs in white-tailed deer (O. virginianus; Carstensen et al. 2003, Haskell et al. 2007, Saalfeld and Ditchkoff 2007), black-tailed deer (O. hemionus columbianus; Pamplin 2003), mule deer (Bishop et al. 2007, Haskell et al. 2007), and elk (Cervus elaphus; Johnson et al. 2006, Barbknecht et al. 2009) have been moderately successful. Vaginal implant transmitters also permit measurement of fetal survival in free-ranging populations, which has important implications in populations where stillborn mortality occurs (Bishop et al. 2007, 2008, 2009). An additional advantage of using VITs to capture neonates may be a reduction in sampling bias when compared to capture techniques that rely on opportunistic fawn capture (White et al. 1972, Ballard et al. 1998, Pojar and Bowden 2004). Opportunistic techniques are susceptible to bias because of unequal capture success among vegetation types, distances to roads, fawn ages, and stages of fawning. For example, if roads are used to conduct opportunistic searches, fawn capture probability will decline with increasing distance from a road and neonates will be disproportionately sampled in areas with high road densities. When using VITs, the distribution of radio-marked adult females carrying VITs determines where neonates are sampled. Inferences will be less biased with VITs than with opportunistic capture techniques if all VITs are monitored with equal intensity during fawning and the sample of radio-marked adult females was captured with minimal bias. Thus, VITs could have broad applicability regardless of whether study objectives require that fawns be captured from previously marked adult females. The most significant problem associated with VITs has been premature expulsion and subsequent failure to locate birth sites or newborn fawns, especially in mule deer (Johnstone-Yellin et al. 2006, 72 �Bishop et al. 2007, Haskell et al. 2007). The VIT has flexible, plastic wings coated with a soft silicone that induce pressure against the vaginal wall to retain the transmitter. The VIT design facilitates a quick, non-surgical insertion process and is safe for the animal (Johnson et al. 2006), but the current wing design is inadequate with respect to retention. Bishop et al. (2007) found that 43% (SE = 4.7) of VITs in mule deer shed prepartum, although the probability of capturing ≥1 fawn was relatively high (0.792, SE = 0.0847) when VITs shed only 1–3 days prepartum. They noted that 25% (SE = 4.1) of VITs shed >3 days prepartum and that retention probability declined as deer body size increased, indicating the retention wings were too small to be effective in larger deer. Based on these results, considerable oversampling of adult females would be required in the design of future projects to achieve a target sample size of fawns. That is, extra adult females would need to be sampled to offset those adult females that shed VITs prematurely. Oversampling, in this instance, is undesirable from an animal care and use perspective and unnecessarily expensive. Thus, our objective was to redesign the plastic-silicone retention wings of VITs to allow maximum retention in larger deer species. Prior to our study, the wings used to retain VITs had been purchased from a company in New Zealand (Carter Holt Harvey Plastic Products, Hamilton, New Zealand) that originally produced them for an application in the livestock industry (Bowman and Jacobson 1998). The company manufactured 1 large wing and 1 small wing; the former has been used in production of VITs for bison (Bison bison) and elk (Cervus elaphus) whereas the latter has been used in production of VITs for deer (Advanced Telemetry Systems, Isanti, MN). Advanced Telemetry Systems (ATS), in cooperation with wildlife researchers, made an initial effort in 2004 to lengthen the retention wings by adding resin to the wing tips. Using these VITs with antennas cut to the appropriate length, Haskell et al. (2007) reported that 81% of VITs (n = 21) in deer were retained until parturition. Retention improved but the aftermarket wingmodification was problematic because the wing tips were hard and thus not ideal for placement in the vaginal canal. That study provided justification to pursue further wing development. We therefore redesigned retention wings of VITs used in deer and similar-sized ungulates, fabricated a new production mold, and evaluated retention rates of VITs in free-ranging mule deer. STUDY AREA We conducted our research in Piceance Basin and on the Roan Plateau in northwest Colorado (Fig. 1). Our winter range study area comprised 4 study units distributed across much of the Piceance Basin. The 4 units ranged in size from 70 to 130 km2 and are referenced as South Magnolia, StorySprague, Ryan Gulch, and Yellow Creek (Fig. 1). METHODS We prepared and submitted a draft manuscript to Journal of Wildlife Management (JWM). Initial reviews were favorable, and thus, we were invited to submit a revised manuscript for further consideration. We prepared a revised manuscript based on comments submitted by peer reviewers and the associate editor. RESULTS AND DISCUSSION Our revised manuscript was accepted for publication on March 23, 2011. The manuscript will be published in the November 2011 issue of JWM. The abstract from this publication follows: Our understanding of factors that limit mule deer (Odocoileus hemionus) populations may be improved by evaluating neonatal survival as a function of dam characteristics under free-ranging conditions, which generally requires that both neonates and dams are radiocollared. The most viable technique facilitating capture of neonates from radiocollared adult females is use of vaginal implant 73 �transmitters (VITs). To date, VITs have allowed research opportunities that were not previously possible; however, VITs are often expelled from adult females prepartum, which limits their effectiveness. We redesigned an existing VIT manufactured by Advanced Telemetry Systems (ATS; Isanti, MN) by lengthening and widening wings used to retain the VIT in an adult female. Our objective was to increase VIT retention rates and thereby increase the likelihood of locating birth sites and newborn fawns. We placed the newly designed VITs in 59 adult female mule deer and evaluated the probability of retention to parturition and the probability of detecting newborn fawns. We also developed an equation for determining VIT sample size necessary to achieve a specified sample size of neonates. The probability of a VIT being retained until parturition was 0.766 (SE = 0.0605) and the probability of a VIT being retained to within 3 days of parturition was 0.894 (SE = 0.0441). In a similar study using the original VIT wings (Bishop et al. 2007), the probability of a VIT being retained until parturition was 0.447 (SE = 0.0468) and the probability of retention to within 3 days of parturition was 0.623 (SE = 0.0456). Thus, our design modification increased VIT retention to parturition by 0.319 (SE = 0.0765) and VIT retention to within 3 days of parturition by 0.271 (SE = 0.0634). Considering dams that retained VITs to within 3 days of parturition, the probability of detecting at least 1 neonate was 0.952 (SE = 0.0334) and the probability of detecting both fawns from twin litters was 0.588 (SE = 0.0827). We expended approximately 12 personhours per detected neonate. As a guide for researchers planning future studies, we found that VIT sample size should approximately equal the targeted neonate sample size. Our study expands opportunities for conducting research that links adult female attributes to productivity and offspring survival in mule deer. The full text publication can be obtained electronically or in hard copy through JWM and WileyBlackwell Publishers. SUMMARY Use of VITs in well-designed field studies should increase our understanding of factors limiting deer populations by allowing investigators to link fawn production and survival to dam characteristics under free-ranging conditions. A primary drawback of VITs in deer has been the failure of many adult females to retain VITs to parturition. We increased VIT retention in mule deer by lengthening and widening wings used to retain a VIT in the vaginal canal. Researchers employing VITs with our modified wing design should require minimal oversampling to offset failures caused by early expulsion, thereby rendering the technique more cost-effective and reliable. Our findings provide explicit guidance for planning a fetal-neonatal deer study involving VITs. The question remains as to whether premature expulsion of VITs can be eliminated in mule deer. We observed modest evidence that deer expelling VITs >3 days prepartum were older and larger than deer that retained or nearly-retained VITs. We therefore recommend manufacturing slightly larger wings for large, older mule deer (e.g., >65 kg and >5 yrs old) as a possible strategy to further investigate VIT retention. An article documenting our research findings will be published in the November 2011 issue of JWM. 74 �LITERATURE CITED Ballard, W. B., H. A. Whitlaw, D. L. Sabine, R. A. Jenkins, S. J. Young, and G. J. Forbes. 1998. Whitetailed deer, Odocoileus virginianus, capture techniques in yarding and non-yarding populations in New Brunswick. Canadian Field-Naturalist 112:254−261. Barbknecht, A. E., W. S. Fairbanks, J. D. Rogerson, E. J. Maichak, and L. L. Meadows. 2009. Effectiveness of vaginal-implant transmitters for locating elk parturition sites. Journal of Wildlife Management 73:144−148. Bishop, C. J., D. J. Freddy, G. C. White, B. E. Watkins, T. R. Stephenson, and L. L. Wolfe. 2007. Using vaginal implant transmitters to aid in capture of mule deer neonates. Journal of Wildlife Management 71:945−954. Bishop, C. J., G. C. White, D. J. Freddy, B. E. Watkins, and T. R. Stephenson. 2009. Effect of enhanced nutrition on mule deer population rate of change. Wildlife Monographs 172:1−28. Bishop, C. J., G. C. White, and P. M. Lukacs. 2008. Evaluating dependence among mule deer siblings in fetal and neonatal survival analyses. Journal of Wildlife Management 72:1085−1093. Bowman, J. L., and H. A. Jacobson. 1998. An improved vaginal-implant transmitter for locating whitetailed deer birth sites and fawns. Wildlife Society Bulletin 26:295−298. Carstensen, M., G. D. DelGiudice, and B. A. Sampson. 2003. Using doe behavior and vaginal-implant transmitters to capture neonate white-tailed deer in north-central Minnesota. Wildlife Society Bulletin 31:634−641. Haskell, S. P., W. B. Ballard, D. A. Butler, N. M. Tatman, M. C. Wallace, C. O. Kochanny, and O. J. Alcumbrac. 2007. Observations on capturing and aging deer fawns. Journal of Mammalogy 88:1482−1487. Johnson, B. K., T. McCoy, C. O. Kochanny, and R. C. Cook. 2006. Evaluation of vaginal implant transmitters in elk (Cervus elaphus nelsoni). Journal of Zoo and Wildlife Medicine 37:301−305. Johnstone-Yellin, T. L., L. A. Shipley, and W. L. Myers. 2006. Evaluating the effectiveness of vaginal implant transmitters for locating neonatal mule deer fawns. Wildlife Society Bulletin 34:338−344. Pamplin, N. P. 2003. Ecology of Columbian black-tailed deer fawns in western Oregon. Thesis, Oregon State University, Corvallis, USA. Pojar, T. M., and D. C. Bowden. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550−560. Saalfeld, S. T., and S. S. Ditchkoff. 2007. Survival of neonatal white-tailed deer in an exurban population. Journal of Wildlife Management 71:940−944. White, M., F. F. Knowlton, and W. C. Glazener. 1972. Effects of dam-newborn fawn behavior on capture and mortality. Journal of Wildlife Management 36:897−906. Prepared by _______________________ Chad J. Bishop, Mammals Research Leader 75 �Figure 1. Location of winter and summer range study areas in Piceance Basin and Roan Plateau, northwest Colorado. Winter range study units where we captured and radio-marked mule deer are noted as: YC = Yellow Creek, RG = Ryan Gulch, SM = South Magnolia, and SS = Story-Sprague. 76 � https://cpw.cvlcollections.org/files/original/7275825e22516664628adf03224d1631.pdf 551222294d28fd64eab2f0081020dea7 PDF Text Text Colorado Division of Parks and Wildlife July 2016 - June 2017 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Federal Aid Project No. Colorado 3430 3002 1 W-242-R2 : : : : : : Division of Parks and Wildlife Mammals Research Pilot Study—Elk Recruitment and Habitat Use in Colorado Period Covered: July 1, 2016 - June 30, 2017 Author: M.W. Alldredge, B. Banulis, and A. Vitt Personnel: R. DeVergie, L. Snobl, T. Stratman, A. Orlando, W. Hiler, A. Meyer, N. Collier, B. Hoffman, K. Botzet, S. Stevens, L. Sweanor, T. Verzuh All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Our principal research objective was to assess elk recruitment rates in southern Colorado, where declining December calf:cow ratios have been observed since the early 2000’s. We initiated this study this year in elk Data Analysis Units (DAUs) E-20 and E-33 as these areas consistently had the lowest calf:cow ratios. Our objective is to determine major factors that may be contributing to low recruitment in order to direct future research activities. Therefore, we are examining this from both the cow productivity standpoint and in terms of calf mortality. In February and March, 2017, 30 adult cow elk were captured in each area (23 collared), to assess pregnancy rates and body condition. Starting in May, 2017, we also captured and fitted at least 40 neonate elk in each study area with GPS collars to assess cause specific mortality rates and recruitment to the yearling age class. Body condition in both study areas was reasonable given the time of year and pregnancy rates were 78.1% in E-20 and 90.0% in E-33. By June 20, 2017, a total of 40 calves had been captured in E-20 and 55 in E-33. 1 �WILDLIFE RESEARCH REPORT PILOT STUDY—ELK RECRUITMENT AND HABITAT USE IN COLORADO MATHEW W. ALLDREDGE PROJECT NARRITIVE OBJECTIVES 1. To assess adult cow elk (Cervus canadensis) body condition and pregnancy rates in two different elk populations, DAUs E-20 and E-33, in southern Colorado. 2. Assess cause specific calf mortality and recruitment into the yearling age class for two different elk populations, E-20 and E-33, in southern Colorado. 3. Assess cow and calf habitat use in relation to body condition and survival. SEGMENT OBJECTIVES Elk Recruitment 1. Measure body condition, pregnancy and fetal rates for cow elk within each study area. 2. Measure calf elk weight at parturition and relate this to the dam’s body condition and whether she had a calf the previous year. 3. Determine cause specific mortality for elk calves from birth to age one. 4. Examine cow elk habitat use, including use of habitat treatments in the study area. ELK RECRUITMENT BY M. ALLDREDGE INTRODUCTION Rocky Mountain Elk (Cervus elaphus) is an iconic species throughout western North America and especially in Colorado, with a high recreational value to hunters, photographers, artists and wildlife enthusiasts in general. Elk populations are known to fluctuate greatly following habitat succession, especially following historic wildfires. Human exploitation, habitat loss, predation and disease are all factors that can lead to population declines. In order to maintain healthy populations, managers must understand these factors and use their best knowledge to set herd objectives, harvest strategies and monitoring programs. Population dynamics of ungulates, including elk, are generally characterized by relatively constant adult survival and variable recruitment (Gaillard et al. 1998, Garrott et al. 2003, White et al. 2010, Proffitt et al. 2014). Eberhardt (2002), when examining large herbivore populations, suggested that juvenile survival is the first parameter expected to change as populations decline. Variability in recruitment rates can manifest in a variety of ways, including pregnancy rates, intra-uterine survival, neonate survival and over-winter survival (Proffitt et al. 2014). Pregnancy rates can be a function of environmental conditions or nutrition and its influence on body condition (Cook et al. 2004). Survival of neonates can be affected by summer temperature, precipitation, predation, population density (White et al. 2010, Griffin et al. 2011) and human disturbance (Phillips and Alldredge 2000). Malnutrition is also a factor that contributes to over-winter mortality (Proffitt et al. 2014). 2 �Concerns about elk calf ratios have been expressed for about a decade, but factors influencing local populations remain largely unknown. During the 1990’s and early 2000’s, elk herds were above objective and efforts were made to reduce elk populations. Calf ratios started declining in the early 2000’s while herds were generally still above objective. Many studies have been conducted to investigate environmental influences on elk, many of which center around juvenile recruitment (Alldredge and Phillips 2000, White et al. 2010, Sargeant et al. 2011, Cook et al. 2013, Proffitt et al. 2014). Colorado is no exception, in many parts of the state recruitment rates are low and declining, which will have long term ramifications on elk populations across the state. Colorado Parks and Wildlife (CPW) has defined numerous elk data analysis units (DAUs) across the state delineating specific elk populations for management objectives and monitoring purposes (Figure 1). Early winter surveys are conducted within each DAU annually in order to obtain population parameters, such as cow:calf:bull ratios, in order to model population size. In general, these data are showing lower and declining cow:calf ratios in the southern portion of the state compared to the northern part of the state since 1994 and especially since the early 2000s (Figures 2 and 3). Annual variation in cow:calf ratios exists but a steady decline for the southern DAUs is also evident with current ratios approaching or below 30 calves per 100 cows (Figure 2). Similar annual variability exists in the northern DAUs, but the declining pattern is not evident and cow:calf ratios are higher, currently above 50 calves per 100 cows in the DAUs presented. Low recruitment rates for elk across the state and potential long term population level ramifications are of great concern to CPW wildlife managers and biologists. If the trend of low recruitment rates continues, resulting declining elk populations will significantly impact both recreational opportunity and economics in Colorado and for CPW. Furthermore, CPW has a statutory responsibility to manage elk. However, very little is known as to the factors driving declining recruitment rates. Research on this topic is vital. A recent study on mule deer (Odocoileus hemionus) has demonstrated a paradigm shift in causes of low recruitment for this species from the historical research demonstrating low over-winter survival (Bartmann et al. 1992) to recent developments suggesting low neonatal survival (Anderson 2015). It is imperative that CPW conduct similar investigations on elk to gain information on factors affecting low recruitment across the state and develop management strategies to mitigate these factors. Because little is known about the factors affecting elk recruitment in the state, we proposed a pilot study designed to identify primary factors. Given that this low recruitment is occurring across a broad spatial scale we also proposed that this work be conducted in multiple study areas exhibiting low recruitment and one study area with higher recruitment as a reference area. The intent of this 2 year pilot study is to determine pregnancy rates, fetal counts, and cause specific mortality of calf elk from birth to age 1. Additional data on cow body condition, birth weights and consecutive year reproduction will allow determination of potential causes of low elk recruitment. Measuring individual body condition of cow elk in the study and then ascertaining the fate of each cow’s calf will provide valuable insights regarding nutritional influences on both calf survival and future pregnancy rates. Examinations of cow elk habitat use will also be conducted, including use of habitat treatments that exist on the landscape, to determine differences in habitat use and the impact that has on pregnancy rates and calf survival. Information from this pilot study will provide insights into factors that are affecting elk recruitment throughout Colorado and potentially throughout many western states. This information will be used to direct future research and management implementation to mitigate factors contributing to low elk recruitment. These factors may include but are not limited to; habitat quality and poor nutrition, neonate predation, disease and/or human disturbance. The results from this pilot project will direct future hypotheses regarding declining calf ratios and future direction of work, such as: 3 �High pregnancy rates: indicates high dam nutrition and reduces evidence of nutritional stress Hypothesis: Poor neonate survival. Low pregnancy rates and poor body condition: indicates nutritional stress for dam, reduced evidence that calf survival is limiting. Hypothesis: Summer nutrition is limiting herd productivity. Low neonatal survival and low predator specific mortality: indicates habitat limitation. Hypothesis: Poor neonate survival is caused by habitat limitation. Low neonatal survival and high predator specific mortality: indicates predator limitation. Hypothesis: High neonatal predation is limiting populations. Results indicating habitat limitations could lead to future studies that examine habitat manipulations/improvement and/or studies that examine reductions in herd size. Results indicating poor neonate survival could lead to studies that will examine cause specific mortality to determine the drivers of low calf survival. If predation appears to be a limiting factor then future studies could examine the effects of predator manipulations on neonate survival. If other factors are limiting calf survival then specific studies will be designed to examine these limitations (disease, habitat, etc.) and what manipulations can be done to improve neonate survival. STUDY AREA This project is designed to examine low elk recruitment issues statewide. Given the broad scale issue, two study areas with low elk recruitment were selected and, if funding permits, one area will be selected as a reference area where elk recruitment is high following the 2-year pilot phase. In conjunction with other work in the state, pregnancy rates are being monitored in areas where recruitment is high. The two study areas in the southern part of the state with low recruitment rates targeted during the pilot phase are DAUs E-20 and E-33. E-20 E-20 is the Uncompahgre Plateau area with game management units (GMU) 61 and 62. These GMUs are in Montrose, Mesa, Delta, San Miguel and Ouray counties encompassing 2,262 square miles. Elevations range from 4,600 feet to 10,300 feet. The area is characterized by flat mesas and deep rugged canyons. Vegetation includes grassland, shrub, pinon/juniper, pine, aspen, and spruce/fir. Land ownership includes private, BLM, US Forest Service and state. E-33 E-33 is the La Veta/Trinidad area with game management units (GMU) 140, 85, and 851. GMU 140 is in Las Animas County on the New Mexico border. Elevations range from 5,414 to 9,544 feet. The area is characterized by gently rolling hills, steep canyons and mesas. Vegetation includes shortgrass prairie, pinion pine, oak and spruce/fir. Approximately 99% of the land is privately owned. GMU 85 is located in Huerfano and Las Animas counties. Elevations range from 6,025 feet to 13,518 feet. The area is characterized by flat valley bottoms, foothills and steep mountains. Vegetation includes grassland, pinon/juniper, pine/oak, spruce/fir, and alpine. Approximately 75% of the area is privately owned. GMU 851 is in Las Animas county and is bordered on the south by New Mexico. Elevations range from 6,025 to 14,000 feet. The area is characterized foothills and steep mountains. Vegetation includes grassland, pine, spruce/fir, aspen, and alpine. Approximately 98% of the area is privately owned. 4 �METHODS Cow Capture and Sampling Cow elk will be captured in January using clover traps, free-range darting, and helicopter net gunning following approved elk capture and handling guidelines (CPW ACUC #09-2008) (Appendix I). Darting and clover traps will be used initially to capture elk and net gunning will be used if necessary to complete the sample of at least 20 collared cow elk per study area and body condition on 30 elk per study area. Fetal counts and body condition will be assessed for all collared elk using ultrasound. Multiple bait sites will be placed within each study area to attract elk prior to capture. Elk will be darted or clover trapped at these sites. Only 2 to 3 elk will be collared at each bait site. Additional animals will be eartagged and blood will be drawn to estimate pregnancy rates before release. Pregnant females equipped with telemetry collars on winter range will also be equipped with vaginal implant transmitters (VITs; Neolink-ITX, Advanced Telemetry Systems, Isanti, MN, USA) to facilitate spring neonate capture and collaring efforts following birth on summer range. Neonate Capture GPS collars on cow elk will send an alert and a location when a VIT is expelled. Once expelled, field crews will be directed to birth site locations to locate and capture newborn calves. In order to minimize human disturbance, neonate searches will typically last 3045 minutes and will not exceed 1 hour. Uncollared pregnant cows will also be observed in an attempt to opportunistically locate and collar additional calves. Each neonate will be handled with sterile nitrile latex gloves to minimize the transfer of human scent, blindfolded, and placed in a cloth bag to measure body mass. Hind foot length, chest girth, age (days), and gender will also be recorded. Each neonate will be fitted with an expandable GPScollar (Wildlink-GTX Globalstar GPS Collar, ATS, Isanti MN, USA) with a 4 hour mortality sensor that is designed to drop off after 12 months. Handling time will be ≤ 5 minutes and neonates will be placed in the precise location where they were located to minimize abandonment. Neonate telemetry signals will be monitored daily using GPS satellite communication. Continuous monitoring will afford us the ability to detect mortalities and assess cause specific mortality within 24 hours. Monitoring of neonate signals will continue throughout the year or until mortality occurs. Once a mortality is detected, neonates and/or collars will be located from the ground and if any part of a carcass is present a thorough field necropsy will be conducted to determine cause of death. Data Analysis Pregnancy rates will be compared among study areas, years and possibly to historic data using a binomial variance. Body condition and age class will be treated as covariates in the analysis. Longitudinal data will also be examined to determine the affects of previous years’ pregnancy status on pregnancy rate, body condition and birth weights. We will also use individual body condition to estimate the animal indicated carrying capacity following Monteith et al. (2014). Low pregnancy rates and/or poor body condition will lead us to consider future habitat manipulations or elk population reductions as a means to improve herd productivity. Pregnancy rates from blood samples of captured cows in the NW, where recruitment is higher, were estimated at 0.92 in December 2015. The binomial power to detect a decrease from this baseline rate of 0.15, using a sample size of 30, is 0.78. A minimum of 20 animals is necessary to detect a 1.5% difference in total body fat based on mule deer with a power of 0.70 (Anderson 2008). Therefore, 30 animals in each study area should give reasonable power to detect differences in body condition between pregnant and non-pregnant cow elk. Neonate survival estimates will be obtained using the Kaplan-Meier (Pollock et al. 1989) approach and examined for seasonal patterns. Cause specific mortality estimates will be examined with 5 �competing risks models using Cox proportional hazards (Heisey and Patterson 2006). These models will include seasonal effects, individual effects (birth weight, gender) and environmental effects to obtain detailed information on the drivers of cause specific neonate mortality. If neonate survival is low then future work will be directed at improving calf survival. This could take the form of predator manipulations if predation rates are high or habitat manipulations if habitat appears to impact calf survival. Bishop et al. (2009) suggested that a sample size of 40 neonates per group per year provided power of 0.81 to detect a difference of 0.15 in survival. Analysis of cause specific mortality of 40 neonates per area over 2 years results in mortality estimates with cv’s less than 0.14, which is sufficient to assess mortality and direct future research efforts. We recognize that cause specific mortality may represent the proximate cause and that the ultimate cause may not be detected as it relates to disease or malnutrition. We will attempt to assess this as well and, when possible, bring calf mortalities to the lab for a more thorough necropsy. Movement models will also be utilized to examine potential drivers of movement for cow elk, including seasonal patterns and responses to environmental variables. Continuous-time-discrete-space models (Hanks et al. 2015) will be used to analyze movement data and RSFs will be used to examine the relationship between resource selection and body condition. Specifically we will use these models to examine animal movement relative to nutritional/energetic demands, response to environmental conditions, seasonal patterns and response to human activity. These movement patterns directly relate to an individual’s resource selection strategy, which are likely to have implications for elk recruitment. These analyses may provide important detail on how elk are utilizing their habitat and what limitations this creates. Additional analyses are currently underway to examine correlations between environmental variables (temperature, moisture) and calf ratios. These data will only provide correlations but will help identify how environmental influences are driving calf ratios on a yearly basis. These data can then be incorporated with data collected in this pilot effort to determine potential interactions among environmental factors and pregnancy rates, body condition, and calf survival. Combining these data will be important to determining the direction of future work. RESULTS AND DISCUSSION Cow elk capture was initiated in late February, 2017, in E-20 and E-33. Weather was hot and dry so baiting elk had limited success as natural forage was starting to develop. A total of 8 elk were caught in E-20 and 5 in E-33 using clover traps. The remaining elk were caught in early March using helicopter net gunning. Body condition was estimated for 32 and 29 elk in E-20 and E-33 respectively and 23 were GPS collared in each area. Body condition of elk, based on loin thickness, rump fat and a body condition score was reasonably good in both study areas (Table 1). Vaginal implant transmitters (VITs) were placed in pregnant elk. Pregnancy rates were 78.1% and 90.0% in E-20 and E-33 respectively (Table 1). Calf capture began in the middle of May. Only 2 of 40 VITs worked so capture was primarily opportunistic. A total of 40 and 55 calves were caught in E-20 and E-33, respectively (Table 2). Average age at capture was estimated at just over 2 days old, although some older calves were caught at a week old. Average capture weight was 17.3 kg for both areas. Collar retention is an issue as numerous collars have been found near fences and the collar belting appears to have broken after getting snagged in the fence. As this is the first year of the study and data is just starting to be collected, no factors have been identified as potentially contributing to low recruitment rates for these elk herds. In the southwest corner of E-20, pregnancy rates were very low. Of the 8 cows captured there, only 4 were pregnant. However, 6 �pregnancy rates in the rest of this unit were high. This may be of interest for further investigation to determine if there are localized low pregnancy rates in this area. Beyond this, the project is on schedule and proceeding as planned. SUMMARY The elk recruitment study was initiated this year in E-20 and E-33. Approximately 30 cow elk were captured in each study area using clover traps and helicopter net gunning and 23 were GPS collared in each study area. Body condition and pregnancy rates were reasonable in both study areas. Calf capture began in May and a total of 40 and 55 calves were captured and collared in E-20 and E-33 respectively. Calf capture concluded at the end of June so no data is available to assess calf survival or cause specific mortality at this time. LITERATURE CITED Anderson, C. R. Jr. 2015. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Federal Aid Project No. W-185-R Annual Report, Colorado Parks and Wildlife, Fort Collins, USA. Bartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monographs No. 121. Cook, J.G., B.K. Johnson, R.C. Cook, R.A. Riggs, T. Delcurto, L.D. Bryant, and L.L. Irwin. 2004. Effects of summer-autumn nutrition and parturition data on reproduction and survival of elk. Wildlife Monographs 155:1-61. Cook, R.C., J.G. Cook, D.J. Vales, B.K. Johnson, S.M. McCorquodale, L.A. Shipley, R.A. Riggs, L.L. Irwin, S.L. Murphie, K.A. Schoenecker, F. Geyer, P.B. Hall, R.D. Spencer, D.A. Immell, D.H. Jackson, B.L. Tiller, P.J. Miller, and L. Schmitz. 2013. Regional and seasonal patterns of nutritional condition and reproduction in elk. Wildlife Monographs 184:1-44. Eberhardt, L.L. 2002. A paradigm for population analysis of long-lived vertebrates. Ecology 83:28412854. Gaillard, J.M., M. Festa-Bianchet, and N.G. Yoccoz. 1998. Population dynamics of large herbivores: variable recruitment with constant adult survival. Trends in Ecology and Evolution. 13:58-63. Garrott, R.A., L.L. Eberhardt, P.J. White, and J. Rotella. 2003 Climate-induced variation in vital rates of an unharvested large-herbivore population. Canadian Journal of Zoology. 81:33-45. Griffin, K.A., M. Hebblewhite, H.S. Robinson, P. Zager, S.M. Barber-Meyer, D. Christianson, S. Creel, N.C. Harris, M.A. Hurley, D.H. Jackson, B.K. Johnson, W.L. Meyers, J.D. Raithel, M. Schlegel, B.L. Smith, C. White, and P.J. White. 2011. Neonatal mortality of elk drivien by climate, predator phenology and predator community composition. Journal of Animal Ecology 80:12461257. Hanks, E.M., M.B. Hooten, and M.W. Alldredge. 2015. Continuous-time discrete-space models for animal movement. The Annals of Applied Statistics. 9:145-165. 7 �Heisey, D.M. and B.R. Patterson. 2006. A review of methods to estimate cause-specific mortality in presence of competing risks. Journal of Wildlife Management. 70:1544-1555. Monteith, K.L., V.C. Bleich, T.R. Stephenson, B.M. Pierce, M.M. Conner, J.G. Kie, and R.T. Bowyer. 2014. Life-history characteristics of mule deer: effects of nutrition in a variable environment. Wildlife Monographs. Phillips, G.E and A.W. Alldredge. 2000. Reproductive success of elk following disturbance by humans during calving season. Journal of Wildlife Management. 64:521-530. Pollock, K.H., S.R. Winterstein, C.M. Bunck, and P.D. Curtis. 1989. Survival analysis in telemetry studies: The staggered entry design. Journal of Wildlife Management. 53:7-15. Proffitt, K.M., J.A. Conningham, K.L. Hamlin, and R.A. Garrott. 2014. Bottom-up and top-down influences on pregnancy rates and recruitment of northern Yellowstone elk. Journal of Wildlife Management 78:1383-1393. Sargeant, G.A., D. C. Weber, and D. E. Roddy. 2011. Implications of chronic wasting disease, cougar predation, and reduced recruitment of elk management. Journal of Wildlife Management 75:171177. White, C.G., P. Zager, and M. W. Gratson. 2010. Influence of predator harvest, biological factors, and landscape on elk calf survival in Idaho. Journal of Wildlife Management 74:355-369. White, P.J., R.A. Garrott, K.L. Hamlin, R.C. Cook, J.G. Cook, and J.A. Cunningham. 2011. Body condition and pregnancy in northern Yellowstone elk: evidence for predation risk effects? Ecological Applications 21:3-8. Prepared by Mathew W. Alldredge, Wildlife Researcher 8 �Table 1: Cow capture statistics for E-20 and E-33. Loin thickness (mm), rump fat thickness (mm), body condition score (BCS) and percent pregnant by year and location. E-20 E-33 Year n Loin Rump BCS % Pregnant 2017 32 48.7 7.1 3.4 78.1 2017 29 52.0 5.7 3.4 90.0 Table 2: Calf capture summary for E-20 and E-33. Sex ratio (female:male), estimated capture age (days) and capture weight (kg). E-20 E-33 Year n F:M Age Weight 2017 40 20:20 2.3 17.3 2017 55 29:26 2.6 17.3 �Figure 1: Colorado’s elk data analysis units (DAUs) used for management and monitoring of elk populations. COLORADO PARKS AND WILDLIFE - Elk DAUs April2015 • �Figure 2: Selected southern Colorado DAUs showing declining calf:cow ratios from 1994 to present. Declining elk calf ratios 1994 - 2014 70 60 [2 -+- r l4 n~ I !; I . . - en [1 S - 111,..n IF1111 u ~ fl :UI 1 J . 11 ar tl U I �Figure 3: Two northern Colorado DAUs showing increasing or slightly decreasing calf:cow ratios from 1994 to present. St bl to Iner fl I ,,, ng elk c If r tios �Colorado Division of Parks and Wildlife July 2017 - June 2018 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Federal Aid Project No. Colorado 3430 3002 1 : : : : : Division of Parks and Wildlife Mammals Research Pilot Study—Elk Recruitment and Habitat Use in Colorado W-242-R2 Period Covered: July 1, 2017 - June 30, 2018 Author: M.W. Alldredge, N. Rayl, B. Banulis, and A. Vitt Personnel: R. DeVergie, L. Snobl, T. Stratman, A. Orlando, W. Hiler, A. Meyer, N. Collier, B. Hoffman, K. Botzet, S. Stevens, L. Sweanor, T. Verzuh, A. Tuck, J. Stanton, A, Kirby, S. Boyle, C. Wallace, A. Hart, A. Larson, R. Coleman, J. Kelley, L. Temple, M. Montoya, K. Logan All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Our principal research objective was to assess elk recruitment rates in southern Colorado, where declining December calf:cow ratios have been observed since the early 2000’s. We initiated this study in 2016 in elk Data Analysis Units (DAUs) E-20 and E-33 as these areas consistently had the lowest calf:cow ratios in the state. Our objective is to determine major factors that may be contributing to low recruitment in order to direct future research activities. Therefore, we are examining this from both the cow productivity standpoint and in terms of calf mortality. In February and March, 2017 and 2018, 30 adult cow elk were captured in each area (approximately 20 collared), to assess pregnancy rates and body condition. Starting in May, 2017 and 2018, we also captured and fitted at least 40 neonate elk in each study area with global positioning system (GPS) collars to assess cause specific mortality rates and recruitment to the yearling age class. Body condition in both study areas was reasonable given the time of year and pregnancy rates have ranged between 77% and 97.0% in E-33. Because a large portion of calf collars fell off prematurely due to manufacture related issued associated with new technology, no calf recruitment estimates were possible for year one and year two estimates are still pending. 1 �WILDLIFE RESEARCH REPORT PILOT STUDY—ELK RECRUITMENT AND HABITAT USE IN COLORADO MATHEW W. ALLDREDGE NATHANIEL RAYL PROJECT NARRITIVE OBJECTIVES 1. To assess adult cow elk (Cervus canadensis) body condition and pregnancy rates in two different elk populations, DAUs E-20 and E-33, in southern Colorado. 2. To assess cause specific calf mortality and recruitment into the yearling age class for two different elk populations, E-20 and E-33, in southern Colorado. 3. To assess cow and calf habitat use in relation to body condition and survival. SEGMENT OBJECTIVES Elk Recruitment 1. Measure body condition, pregnancy and fetal rates for cow elk within each study area. 2. Measure calf elk weight at parturition and relate this to the dam’s body condition. 3. Determine cause specific mortality for elk calves from birth to age one. 4. Examine cow elk habitat use, including use of habitat treatments in the study area. INTRODUCTION Rocky Mountain Elk (Cervus elaphus) is an iconic species throughout western North America and especially in Colorado, with a high recreational value to hunters, photographers, artists and wildlife enthusiasts in general. Elk populations are known to fluctuate greatly following habitat succession, especially following historic wildfires. Human exploitation, habitat loss, predation and disease are all factors that can lead to population declines. In order to maintain healthy populations, managers must understand these factors and use their best knowledge to set herd objectives, harvest strategies and monitoring programs. Population dynamics of ungulates, including elk, are generally characterized by relatively constant adult survival and variable recruitment (Gaillard et al. 1998, Garrott et al. 2003, White et al. 2010, Proffitt et al. 2014). Eberhardt (2002), when examining large herbivore populations, suggested that juvenile survival is the first parameter expected to change as populations decline. Variability in recruitment rates can manifest in a variety of ways, including pregnancy rates, intra-uterine survival, neonate survival and overwinter survival (Proffitt et al. 2014). Pregnancy rates can be a function of environmental conditions or nutrition and its influence on body condition (Cook et al. 2004). Survival of neonates can be affected by summer temperature, precipitation, predation, population density (White et al. 2010, Griffin et al. 2011) and human disturbance (Phillips and Alldredge 2000). Malnutrition is also a factor that contributes to over-winter mortality (Proffitt et al. 2014). Concerns about elk calf ratios have been expressed for about a decade, but factors influencing local populations remain largely unknown. During the 1990’s and early 2000’s, elk herds were above objective and efforts were made to reduce elk populations. Calf ratios started declining in the early 2000’s while herds were generally still above objective. Many studies have been conducted to investigate environmental influences on elk, many of which center around juvenile recruitment (Phillips and Alldredge2000, White et al. 2010, Sargeant et al. 2011, Cook et al. 2013, Proffitt et al. 2014). Colorado is 2 �no exception, in many parts of the state recruitment rates are low and declining, which will have long term ramifications on elk populations across the state. Colorado Parks and Wildlife (CPW) has defined numerous elk data analysis units (DAUs) across the state delineating specific elk populations for management objectives and monitoring purposes (Figure 1). Early winter surveys are conducted within each DAU annually in order to obtain population parameters, such as cow:calf:bull ratios, in order to model population size. In general, these data are showing lower and declining cow:calf ratios in the southern portion of the state compared to the northern part of the state since 1994 and especially since the early 2000s (Figures 2 and 3). Annual variation in cow:calf ratios exists but a steady decline for the southern DAUs is also evident with current ratios approaching or below 30 calves per 100 cows (Figure 2). Similar annual variability exists in the northern DAUs, but the declining pattern is not evident and cow:calf ratios are higher, currently above 50 calves per 100 cows in the DAUs presented. Low recruitment rates for elk across the state and potential long term population level ramifications are of great concern to CPW wildlife managers and biologists. If the trend of low recruitment rates continues, resulting declining elk populations will significantly impact both recreational opportunity and economics in Colorado and for CPW. Furthermore, CPW has a statutory responsibility to manage elk. However, very little is known as to the factors driving declining recruitment rates. Research on this topic is vital. A recent study on mule deer (Odocoileus hemionus) has demonstrated a paradigm shift in causes of low recruitment for this species from the historical research demonstrating low over-winter survival (Bartmann et al. 1992) to recent developments suggesting low neonatal survival (Anderson 2015). It is imperative that CPW conduct similar investigations on elk to gain information on factors affecting low recruitment across the state and develop management strategies to mitigate these factors. Because little is known about the factors affecting elk recruitment in the state, we proposed a pilot study designed to identify primary factors. Given that this low recruitment is occurring across a broad spatial scale we also proposed that this work be conducted in multiple study areas exhibiting low recruitment and one study area with higher recruitment as a reference area. The intent of this 2 year pilot study is to determine pregnancy rates, fetal counts, and cause specific mortality of calf elk from birth to age 1. Additional data on cow body condition, birth weights and consecutive year reproduction will allow determination of potential causes of low elk recruitment. Measuring individual body condition of cow elk in the study and then ascertaining the fate of each cow’s calf will provide valuable insights regarding nutritional influences on both calf survival and future pregnancy rates. Examinations of cow elk habitat use will also be conducted, including use of habitat treatments that exist on the landscape, to determine differences in habitat use and the impact that has on pregnancy rates and calf survival. Information from this pilot study will provide insights into factors that are affecting elk recruitment throughout Colorado and potentially throughout many western states. This information will be used to direct future research and management implementation to mitigate factors contributing to low elk recruitment. These factors may include but are not limited to; habitat quality and poor nutrition, neonate predation, disease and/or human disturbance. The results from this pilot project will direct future hypotheses regarding declining calf ratios and future direction of work, such as: High pregnancy rates: indicates high dam nutrition and reduces evidence of nutritional stress Hypothesis: Poor neonate survival. Low pregnancy rates and poor body condition: indicates nutritional stress for dam, reduced evidence that calf survival is limiting. Hypothesis: Summer nutrition is limiting herd productivity. Low neonatal survival and low predator specific mortality: indicates habitat limitation. Hypothesis: Poor neonate survival is caused by habitat limitation. 3 �Low neonatal survival and high predator specific mortality: indicates predator limitation. Hypothesis: High neonatal predation is limiting populations. Results indicating habitat limitations could lead to future studies that examine habitat manipulations/improvement and/or studies that examine reductions in herd size. Results indicating poor neonate survival could lead to studies that will examine cause specific mortality to determine the drivers of low calf survival. If predation appears to be a limiting factor then future studies could examine the effects of predator manipulations on neonate survival. If other factors are limiting calf survival then specific studies will be designed to examine these limitations (disease, habitat, etc.) and what manipulations can be done to improve neonate survival. STUDY AREA This project is designed to examine low elk recruitment issues statewide. Given the broad scale issue, two study areas with low elk recruitment were selected and, if funding permits, one area will be selected as a reference area where elk recruitment is high following the 2-year pilot phase. In conjunction with other work in the state, pregnancy rates are being monitored in areas where recruitment is high. The two study areas in the southern part of the state with low recruitment rates targeted during the pilot phase are DAUs E-20 and E-33. E-20 E-20 is the Uncompahgre Plateau area with game management units (GMU) 61 and 62. These GMUs are in Montrose, Mesa, Delta, San Miguel and Ouray counties encompassing 2,262 square miles. Elevations range from 4,600 feet to 10,300 feet. The area is characterized by flat mesas and deep rugged canyons. Vegetation includes grassland, shrub, pinon/juniper, pine, aspen, and spruce/fir. Land ownership includes private, BLM, US Forest Service and state. E-33 E-33 is the La Veta/Trinidad area with game management units (GMU) 140, 85, and 851. GMU 140 is in Las Animas County on the New Mexico border. Elevations range from 5,414 to 9,544 feet. The area is characterized by gently rolling hills, steep canyons and mesas. Vegetation includes shortgrass prairie, pinion pine, oak and spruce/fir. Approximately 99% of the land is privately owned. GMU 85 is located in Huerfano and Las Animas counties. Elevations range from 6,025 feet to 13,518 feet. The area is characterized by flat valley bottoms, foothills and steep mountains. Vegetation includes grassland, pinon/juniper, pine/oak, spruce/fir, and alpine. Approximately 75% of the area is privately owned. GMU 851 is in Las Animas county and is bordered on the south by New Mexico. Elevations range from 6,025 to 14,000 feet. The area is characterized foothills and steep mountains. Vegetation includes grassland, pine, spruce/fir, aspen, and alpine. Approximately 98% of the area is privately owned. METHODS Cow Capture and Sampling Cow elk were captured in late February and early March using clover traps and helicopter net gunning following approved elk capture and handling guidelines (CPW ACUC #09-2008) (Appendix I). Clover traps were used for some elk capture in 2017 but was not efficient because of mild winter conditions. The majority of elk in 2017 and all elk in 2018 were captured via net gunning. In this initial phase of the study a target of 20 collared cow elk per study area and body condition on 30 elk per study area was established. Fetal counts and body condition was assessed for all collared elk using ultrasound. Pregnant females equipped with telemetry collars on winter range will also be equipped with vaginal implant transmitters (VITs;) to facilitate spring neonate capture and collaring efforts following birth on summer 4 �range. VITs failed to communicate birth events during the first year of the study, but did reveal that the use of VITs is critical to capturing elk calves at birth and estimating early mortality. Neonate Capture GPS collars on cow elk will send an alert and a location when a VIT is expelled. Once expelled, field crews will be directed to birth site locations to locate and capture newborn calves. In order to minimize human disturbance, neonate searches will typically last 3045 minutes and will not exceed 1 hour. Uncollared pregnant cows will also be observed in an attempt to opportunistically locate and collar additional calves. Each neonate will be handled with sterile nitrile latex gloves to minimize the transfer of human scent, blindfolded, and placed in a cloth bag to measure body mass. Hind foot length, chest girth, age (days), and gender will also be recorded. Each neonate will be fitted with an expandable GPScollar (Wildlink-GTX Globalstar GPS Collar, ATS, Isanti MN, USA) with a 4 hour mortality sensor that is designed to drop off after 12 months. Handling time will be ≤ 5 minutes and neonates will be placed in the precise location where they were located to minimize abandonment. Neonate telemetry signals will be monitored daily using GPS satellite communication. Continuous monitoring will afford us the ability to detect mortalities and assess cause specific mortality within 24 hours. Monitoring of neonate signals will continue throughout the year or until mortality occurs. Once a mortality is detected, neonates and/or collars will be located from the ground and if any part of a carcass is present a thorough field necropsy will be conducted to determine cause of death. Data Analysis Pregnancy rates will be compared among study areas, years and possibly to historic data using a binomial variance. Body condition and age class will be treated as covariates in the analysis. We will also use individual body condition to estimate the animal indicated carrying capacity following Monteith et al. (2014). Low pregnancy rates and/or poor body condition will lead us to consider future habitat manipulations or elk population reductions as a means to improve herd productivity. Pregnancy rates from blood samples of captured cows in the NW, where recruitment is higher, were estimated at 0.91 in December 2015. The binomial power to detect a decrease from this baseline rate of 0.15, using a sample size of 30, is 0.78. A minimum of 20 animals is necessary to detect a 1.5% difference in total body fat based on mule deer with a power of 0.70 (Anderson 2008). Therefore, 30 animals in each study area should give reasonable power to detect differences in body condition between pregnant and non-pregnant cow elk. Neonate survival estimates will be obtained using the Kaplan-Meier (Pollock et al. 1989) approach and examined for seasonal patterns. Cause specific mortality estimates will be examined with competing risks models using Cox proportional hazards (Heisey and Patterson 2006). These models will include seasonal effects, individual effects (birth weight, gender) and environmental effects to obtain detailed information on the drivers of cause specific neonate mortality. If neonate survival is low then future work will be directed at improving calf survival. Bishop et al. (2009) suggested that a sample size of 40 neonates per group per year provided power of 0.81 to detect a difference of 0.15 in survival. Analysis of cause specific mortality of 40 neonates per area over 2 years results in mortality estimates with cv’s less than 0.14, which is sufficient to assess mortality and direct future research efforts. We recognize that cause specific mortality may represent the proximate cause and that the ultimate cause may not be detected as it relates to disease or malnutrition. We will attempt to assess this as well and, when possible, bring calf mortalities to the lab for a more thorough necropsy. Movement models will also be utilized to examine potential drivers of movement for cow elk, including seasonal patterns and responses to environmental variables. Continuous-time-discrete-space models (Hanks et al. 2015) will be used to analyze movement data and resource selection functions (RSFs) will 5 �be used to examine the relationship between resource selection and body condition. Specifically we will use these models to examine animal movement relative to nutritional/energetic demands, response to environmental conditions, seasonal patterns and response to human activity. These movement patterns directly relate to an individual’s resource selection strategy, which are likely to have implications for elk recruitment. These analyses may provide important detail on how elk are utilizing their habitat and what limitations this creates. Additional analyses are currently underway to examine correlations between environmental variables (temperature, moisture) and calf ratios. These data will only provide correlations but will help identify how environmental influences are driving calf ratios on a yearly basis. These data can then be incorporated with data collected in this pilot effort to determine potential interactions among environmental factors and pregnancy rates, body condition, and calf survival. Combining these data will be important to determining the direction of future work. RESULTS AND DISCUSSION Cow elk capture was initiated in late February, 2017, in E-20 and E-33. Weather was hot and dry so baiting elk had limited success as natural forage was starting to develop. A total of 8 elk were caught in E-20 and 5 in E-33 using clover traps. The remaining elk were caught in early March using helicopter net gunning. Body condition was estimated for 32 and 29 elk in E-20 and E-33 respectively and 23 were GPS collared in each area. Body condition of elk, based on loin thickness, rump fat and a body condition score was reasonably good in both study areas (Table 1). Vaginal implant transmitters (VITs) were placed in pregnant elk. Pregnancy rates were 77.4% and 79.3% in E-20 and E-33 respectively (Figure 4). In late February 2018, cow elk capture was conducted for the second year in E-20 and E-33. All capture was conducted using helicopter net gunning. Weather conditions and mechanical issues for the E-20 efforts extended capture over two weeks, but capture in E-33 was done in 3 days. One capture related mortality occurred in each area. Body condition was estimated for 30 elk in E-20 and 31 elk in E-33 and 20 were collared with VITS in each area. Body condition scores, loin thickness and rump fat were measured for all captured elk (Table 1). Ingesta-free body fat was estimated from these measures for each study area and year with an average near 7.5% each year (Figure 5). Pregnancy rates were higher in 2018 at 96.7% and 87.1% in E-20 and E-33 respectively (Table 1). Calf capture began in the middle of May, 2017. Only 2 of 40 VITs worked so capture was primarily opportunistic. A total of 40 and 55 calves were caught in E-20 and E-33, respectively (Table 2). Average age at capture was estimated at just over 2 days old, although some older calves were caught at a week old. Average capture weight was 17.3 kg for both areas. Collar retention was an issue as numerous collars were found near fences and the collar belting appeared to have been broken after getting snagged in the fence. Calf capture was again initiated in the middle of May for the 2018 season. VITs worked reliably this year allowing for numerous captures shortly after parturition for calves from collared cows. Additional calves were also captured opportunistically. A total of 47 and 54 calves were captured in E-20 and E-33 respectively (Table 2). Average age at capture was 1.6 and 2.2 in E-20 and E-33, slightly younger than the previous year because of the use of VITs. The slightly higher age in E-33 is likely due to a greater number of opportunistic captures. Average capture weight was similar across years. Retention issues for calf collars during the first year of the study limits the results for assessing recruitment. Mortality during the first few months (prior to collar retention issues) indicated 10% known predation mortality in E-20 and 27% known predation mortality in E-33. There was an additional mortality from malnutrition and one unknown cause in E-20 and 2 unknown mortalities in E-33. In E-33, 6 �26% of the calf mortality was attributed to bears and 32% to cougars. As of June 30, 2018, similar patterns were being observed during this second year with potentially higher bear predation in E-33, but this only represents a few weeks of data for this year. SUMMARY The elk recruitment study initiated in 2017 continued this year in E-20 and E-33. Approximately 30 cow elk were captured in each study area using helicopter net gunning and 20 were GPS collared in each study area. Body condition was reasonable, although lactation status was unknown so interpretation is limited and pregnancy rates were slightly low in both study areas. Calf capture began in May and a total of 47 and 54 calves were captured and collared in E-20 and E-33 respectively. Calf capture concluded at the end of June so no data are available to assess calf survival or cause specific mortality for 2018. Information on calf recruitment is limited from 2017 efforts because of collar issues, but there is some indication that early mortality from predators does play a significant role in documented December calf:cow ratios. LITERATURE CITED Anderson, C. R. Jr. 2015. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Federal Aid Project No. W-185-R Annual Report, Colorado Parks and Wildlife, Fort Collins, USA. Bartmann, R. M., G. C. White, and L. H. Carpenter. 1992. Compensatory mortality in a Colorado mule deer population. Wildlife Monographs No. 121. Cook, J.G., B.K. Johnson, R.C. Cook, R.A. Riggs, T. Delcurto, L.D. Bryant, and L.L. Irwin. 2004. Effects of summer-autumn nutrition and parturition data on reproduction and survival of elk. Wildlife Monographs 155:1-61. Cook, R.C., J.G. Cook, D.J. Vales, B.K. Johnson, S.M. McCorquodale, L.A. Shipley, R.A. Riggs, L.L. Irwin, S.L. Murphie, K.A. Schoenecker, F. Geyer, P.B. Hall, R.D. Spencer, D.A. Immell, D.H. Jackson, B.L. Tiller, P.J. Miller, and L. Schmitz. 2013. Regional and seasonal patterns of nutritional condition and reproduction in elk. Wildlife Monographs 184:1-44. Eberhardt, L.L. 2002. A paradigm for population analysis of long-lived vertebrates. Ecology 83:28412854. Gaillard, J.M., M. Festa-Bianchet, and N.G. Yoccoz. 1998. Population dynamics of large herbivores: variable recruitment with constant adult survival. Trends in Ecology and Evolution. 13:58-63. Garrott, R.A., L.L. Eberhardt, P.J. White, and J. Rotella. 2003 Climate-induced variation in vital rates of an unharvested large-herbivore population. Canadian Journal of Zoology. 81:33-45. Griffin, K.A., M. Hebblewhite, H.S. Robinson, P. Zager, S.M. Barber-Meyer, D. Christianson, S. Creel, N.C. Harris, M.A. Hurley, D.H. Jackson, B.K. Johnson, W.L. Meyers, J.D. Raithel, M. Schlegel, B.L. Smith, C. White, and P.J. White. 2011. Neonatal mortality of elk drivien by climate, predator phenology and predator community composition. Journal of Animal Ecology 80:1246-1257. 7 �Hanks, E.M., M.B. Hooten, and M.W. Alldredge. 2015. Continuous-time discrete-space models for animal movement. The Annals of Applied Statistics. 9:145-165. Heisey, D.M. and B.R. Patterson. 2006. A review of methods to estimate cause-specific mortality in presence of competing risks. Journal of Wildlife Management. 70:1544-1555. Monteith, K.L., V.C. Bleich, T.R. Stephenson, B.M. Pierce, M.M. Conner, J.G. Kie, and R.T. Bowyer. 2014. Life-history characteristics of mule deer: effects of nutrition in a variable environment. Wildlife Monographs. Phillips, G.E and A.W. Alldredge. 2000. Reproductive success of elk following disturbance by humans during calving season. Journal of Wildlife Management. 64:521-530. Pollock, K.H., S.R. Winterstein, C.M. Bunck, and P.D. Curtis. 1989. Survival analysis in telemetry studies: The staggered entry design. Journal of Wildlife Management. 53:7-15. Proffitt, K.M., J.A. Conningham, K.L. Hamlin, and R.A. Garrott. 2014. Bottom-up and top-down influences on pregnancy rates and recruitment of northern Yellowstone elk. Journal of Wildlife Management 78:1383-1393. Sargeant, G.A., D. C. Weber, and D. E. Roddy. 2011. Implications of chronic wasting disease, cougar predation, and reduced recruitment of elk management. Journal of Wildlife Management 75:171-177. White, C.G., P. Zager, and M. W. Gratson. 2010. Influence of predator harvest, biological factors, and landscape on elk calf survival in Idaho. Journal of Wildlife Management 74:355-369. White, P.J., R.A. Garrott, K.L. Hamlin, R.C. Cook, J.G. Cook, and J.A. Cunningham. 2011. Body condition and pregnancy in northern Yellowstone elk: evidence for predation risk effects? Ecological Applications 21:3-8. Prepared by Mathew W. Alldredge, Wildlife Researcher 8 �Table 1: Cow elk capture statistics for E-20 and E-33. Loin thickness (mm), rump fat thickness (mm), body condition score (BCS) and percent pregnant by year and location. E-20 E-33 Year 2017 2018 2017 2018 n 32 30 29 31 Loin 48.7 53.7 52.0 51.4 Rump 7.1 7.1 5.7 5.2 BCS 3.4 3.9 3.4 3.2 % Pregnant 77.4 96.7 79.3 87.1 Table 2: Calf elk capture summary for E-20 and E-33. Sex ratio (female:male), mean estimated capture age (days) and weight (kg). E-20 E-33 Year 2017 2018 2017 2018 n 40 47 55 54 F:M 20:20 24:23 29:26 25:29 Age 2.3 1.6 2.6 2.2 Weight 17.3 17.9 17.3 18.7 �Figure 1: Colorado’s elk data analysis units (DAUs) used for management and monitoring of elk populations. COLORADO PARKS AND WILDLIFE - Elk DAUs April 2il 15 • �Figure 2: Selected southern Colorado DAUs showing declining calf:cow ratios from 1994 to present. Declining elk calf ratios 1994 - 2014 80.0 70.0 60.0 50.0 - E20 ~ E24 ; - - E34 8 ....._ E33 0 ~ 40.0 .,;,;> -+- E15 - - Linear (E20) "ii u 30.0 - - Unear (E24) - - Linear (E34 ) - - Linear (E33) 20.0 10.0 0.0 - - Linear (E1 5) �Figure 3: Two northern Colorado DAUs showing increasing or slightly decreasing calf:cow ratios from 1994 to present. Stable to increasing elk calf ratios 80.00 70.00 60.00 50.00 ;: 8 -+- E2 0 ~ 40.00 ~ ~ - - - EB - - linear (E2) ;; u - - Lin ear (E13) 30.00 20.00 10.00 0.00 �Figure 4: Pregnancy rates for cow elk in E-20 (Uncompahgre) and E-33 (Trinchera) estimated from late February capture during 2017 and 2018. The sample size is given at the top of the 95% binomial confidence intervals (black lines). 1.0 0.9 29 31 30 31 ,+-,I ffi 0.8 C 0)0.7 Q) c.0.6 Trinchera Uncompahgre 5 0.5 to 0.4 Cl. e o.3 a_ 0.2 0.1 0.0 2017 2018 Year �Figure 5: Ingesta-free body fat for cow elk in E-20 (Uncompahgre) and E-33 (Trinchera) calculated from body condition scores, rump fat, and loin thickness for 2017 and 2018 late February captures. $ Trinchera • ~ 15.0- C)' ._., ..., ~ 12 •5 ~ Uncompahgre • • •• • ""O 0 ..c ID 10.0ID "- '+I .ffi 7.5 - ID 0) C 5.0 - I I 2017 2018 Year I � https://cpw.cvlcollections.org/files/original/35b4be52a8bb45f0431d770854aac773.pdf b0d26be47dac90c63127c16f3862a99c PDF Text Text Colorado Division of Wildlife June 2004 – July 2005 WILDLIFE RESEARCH REPORT State of Cost Center Work Package Task No. Colorado 3430 3003 1 : : : : Federal Aid Project: N/A : Division of Wildlife Mammals Research Predatory Mammals Conservation Puma Population Structure And Vital Rates On The Uncompahgre Plateau, Colorado Period covered: July 1, 2004―June 30, 2005 Author: K. A. Logan. Personnel: S. Waters, T. Murphy, K. Crane, T. Mathieson, M. Caddy, and T. Smith of CDOW, J. Bauer of Colorado Cooperative Fishery and Wildlife Research Unit, J. Kane of U.S.D.A. Wildlife Services, volunteers, cooperators including: private landowners, U.S. Forest Service, Bureau of Land Management, and Colorado State Parks, with project support received from The Howard G. Buffett Foundation and Safari Club International Foundation. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT To begin conducting research on puma population characteristics and dynamics on the Uncompahgre Plateau, public meetings were held, puma hunting regulations were altered for an experimental design, and a study plan was developed and approved along with animal care and handling procedures. Field research began on December 2, 2004. From December 2, 2004 to July 22, 2005 fifteen puma were captured, sampled, tagged and released, including 7 adult pumas (3 males, 4 females) and 8 cubs (3 males, 5 females). Three other pumas were captured with the aid of dogs, but were released without sampling or tagging for safety reasons. One adult female puma was hit and killed by a car on highway 62 at the southern boundary of the study area. The 7 adult pumas wore GPS collars that yielded 355 to 779 locations per puma. GPS locations indicated 139 clusters that were investigated. Prey use was found at 112 clusters, with mule deer (n = 61) and elk (n = 48) comprising the most important items. Tissue samples collected from all puma handled will be used for proposed research on laboratory and field methods to estimate puma numbers using DNA mark-recapture methods. Puma GPS locations will also be used in proposed efforts to develop and test puma habitat suitability models and maps. Information on evaluations of the GPS collar technology and findings at GPS clusters will be used to develop proposed research on puma-prey relationships on the Uncompahgre Plateau. 105 �WILDLIFE RESEARCH REPORT PUMA POPULATION STRUCTURE AND VITAL RATES ON THE UNCOMPAHGRE PLATEAU, COLORADO KENNETH A LOGAN P. N. OBJECTIVE Quantify puma population sex and age structure; estimate puma population vital rates, including: reproduction rates of females, age-stage survival rates, and immigration and emigration rates; quantify agent-specific mortality rates― all to improve the Colorado Division of Wildlife’s (CDOW) model-based approach to managing puma in Colorado. SEGMENT OBJECTIVES 1. Hold public meetings and contact individuals to inform citizens in the area of Uncompahgre Plateau about the CDOW desires to conduct the puma research. 2. Obtain needed regulations from the Wildlife Commission for experimental research on the study area. 3. Develop a peer-reviewed study plan that is authorized by the Leader of Mammals Research in the CDOW. Develop proper procedures for the capture, restraint, handling, and sampling of puma for research which are authorized by the CDOW Animal Care and Use Committee. 4. Begin quantifying puma population sex and age structure. 5. Begin process of estimating female puma reproduction rates. 6. Begin process of estimating puma sex and age-stage survival rates. 7. Begin process of estimating agent-specific mortality rates. 8. Begin gathering quantitative data on puma movements for the development of sampling methods for direct and DNA-genotype mark-recapture population estimates. Begin gathering puma tissue samples for individual puma genotyping procedures. 9. Evaluate other data sources that could come from this research that might be developed into other puma research relevant to CDOW biologists and managers. INTRODUCTION Colorado Division of Wildlife managers need reliable information on puma biology and ecology in Colorado to develop sound management strategies that address diverse public values and the CDOW objective of actively managing puma while “achieving healthy, self-sustaining populations”(CDOW 2002-2007 Strategic Plan:9). Although 4 puma research efforts have been made in Colorado since the early 1970s and puma harvest data is compiled annually, reliable information on certain aspects of puma biology and ecology, and management tools that may guide managers toward effective puma management is lacking. Mammals Research staff held scoping sessions with a number of the CDOW’s wildlife managers and biologists. In addition, we consulted with other agencies, organizations, and interested publics either directly or through other CDOW employees. In general, CDOW staff in western Colorado highlighted concern about puma population dynamics, especially as they relate to their abilities to manage puma populations through regulated sport-hunting. Secondarily, they expressed interest in puma-prey interactions. Staff on the Front Range placed greater emphasis on puma-human interactions. Staff in both eastern and western Colorado cited information needs regarding effects of puma harvest, puma population monitoring methods, and identifying puma habitat and landscape linkages. Management needs identified 106 �by CDOW staff and public stakeholders form the basis of Colorado’s puma research program, with multiple lines of inquiry (i.e., projects): Improve our ability to manage puma hunting with enhanced scientific bases, strategies, and tools ● Puma population characteristics (i.e., density, sex and age structure). ● Puma population dynamics and vital rates (i.e., birth rates, survival rates, emigration rates, immigration rates, population growth rates). ● Field methods and models for assessing and tracking changes in puma populations. ● Relative vulnerability of puma sex and age classes to hunter harvest. Improve our understanding of puma habitat needs and interrelationships of puma management units ● Puma habitat use, movements, and use of landscape linkages. ● Puma recruitment patterns (i.e., progeny, immigration, emigration). ● Models for identifying puma habitat and landscape linkages. Improve our understanding of the puma’s role in the ecology of other species ● Relationships of puma to mule deer, elk, and other natural prey. ● Relationships of puma to species of special concern, e.g., desert bighorn sheep. Improve our understanding of puma-human interactions and abilities to manage them ● Behavior of puma in relation to people and human facilities. ● Puma predation on domestic animals. ● Effects of translocating nuisance pumas. ● Effects of aversive conditioning on pumas. While all projects cannot be addressed concurrently, understanding their relationships to one another is expected to help individual projects maximize their benefits to other projects that will assist the CDOW to achieve its strategic goal in puma management (Fig.1). Management issues identified by managers translate into researchable objectives, requiring descriptive studies and field experiments. Our goal is to provide managers with reliable information on puma population biology and to develop useful tools for their efforts to adaptively manage puma in Colorado to maintain healthy, self-sustaining populations. The highest-priority management needs are being addressed with this intensive population study that focuses on puma population dynamics using sampled, tagged, and GPS/radio-collared puma. Those objectives include: 1. Describe and quantify puma population sex and age structure. 2. Estimate puma population vital rates, including: birth rates, age-stage-specific survival rates, emigration rates, immigration rates. 3. Estimate agent-specific mortality rates. 4. Improve the CDOW’s model-based management approaches with Colorado-specific data from objectives 1―3. Consider other useful models. Concurrently with the tasks associated with the objectives above, significant progress will be made toward a 5th objective, which will initially be subject to pilot study― develop methods that yield reliable estimates of population abundance (i.e., numbers and density) and attendant annual population growth rates, such as, direct capture-resight, and DNA genotype capture-recapture. 107 �TESTING ASSUMPTIONS AND HYPOTHESES Hypotheses associated with main objectives 1―5 of this puma population research are structured to test assumptions guiding puma management in Colorado. 1. Recreational puma hunting management in Colorado Game Management Units (GMUs) is guided by a model to estimate allowable harvest quotas to achieve one of two puma population objectives: 1) maintain puma population stability, or 2) cause puma population decline (CDOW, Draft L-DAU Plans, 2004). Basic model parameters are: puma population density, sex and age structure, and annual population growth rate. Parameter estimates are currently chosen from literature on studies in western states that are deemed to provide reliable information. Background material used in the model assumes a moderate annual rate of growth of 15% (i.e.,λ = 1.15) for the adult and subadult puma population (J. Apker, Carnivore Management Specialist, CDOW, Monte Vista). This assumption is based upon information with variable levels of uncertainty (e.g., small sample sizes, data from habitats dissimilar to Colorado). The key assumption is that the CDOW can manage puma population growth through recreational hunting: for a stable puma population hunting removes the annual increment of population growth (i.e., as estimated from estimates of population density, structure, and λ); for a declining population, hunting removes more than the annual increment of population growth. Parameters influencing λ include population density, sex and age structure, female age-at-first-breeding, age-specific natality, sexand age-specific survival, immigration and emigration. A descriptive study will ascertain these population parameters in an area that appears typical of puma habitat in western Colorado and will yield defensible population parameters based upon contemporary Colorado data. This study will be conducted in a 5-year reference period (i.e., absence of recreational hunting) to allow puma life history traits to interact with the main habitat factors that appear to influence puma population growth (e.g., prey availability and vulnerability, Pierce et al. 2000, Logan and Sweanor 2001). Contingent upon results in the reference period, a subsequent 5-year treatment period is planned. The treatment period will involve the use of controlled recreational hunting to manage the puma population into a decline phase. H1a: Population parameters measured during a 5-year reference period (in absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will yield an estimated annual adult plus subadult population growth rate that will match or exceed λ = 1.15, which is currently assumed in the CDOW’s model-based management. H1aA: Population parameters measured during a 5-year reference period (absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will yield an estimated annual adult plus subadult population growth rate that will be substantially lower (i.e., ≥50% lower, λ≤1.075) than the assumed λ = 1.15. H1b: Population parameters during a 5-year treatment period (controlled puma hunting) will differ substantially from those measured during the preceding 5-year reference period (hunting closure) and will yield an estimated annual adult plus subadult population growth rate that will be approximately λ=0.8 for at least the first 2 years of the treatment period. Hunting-caused mortality will be strongly additive, and will require removal of the annual growth increment (of adults plus subadults) plus 20% (e.g., assume λ = 1.15, so, 0.15 × 0.2 + 0.15 = 0.18; 0.18 × 100 = 18% annual harvest of adults plus subadults). H1bA: Population parameters during a 5-year treatment period (controlled puma hunting) will not differ substantially from those measured during the preceding 5-year reference period (hunting closure), and the adult plus subadult population will not decline on average as a result of hunting mortality. Hunting-caused mortality, reproduction, immigration, and emigration might be compensatory. 108 �2. Considering limitations (i.e., methods, number of years, assumption violations) to the Coloradospecific studies on puma densities cited above (Currier et al. 1977, Anderson et al. 1992, Koloski 2002), managers assume that puma population densities in Colorado are within the range of those quantified in more intensively studied populations in Wyoming (Logan et al. 1986), Idaho (Seidensticker et al. 1973, Alberta (Ross and Jalkotzy 1992, and New Mexico (Logan and Sweanor 2001). The CDOW assumes density ranges of 2.0―4.6 puma/100 km2 to extrapolate to Data Analysis Units to guide the model-based quota-setting process. Likewise, managers assume that the population sex and age structure is similar to puma populations described in the intensive studies. Using capture, mark, re-capture techniques developed and refined during the study to estimate the puma population, the following will be tested: H2a: Puma densities during the 5-year reference period (absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will vary within the range of 2.0―4.6 puma/100 km2 and will exhibit a similar sex and age structure to puma populations in Wyoming, Idaho, Alberta, and New Mexico. 3. The increase and decline phases of the puma population make it possible to test hypotheses related to shifts in the age structure of the population which have been linked to harvest intensity in Wyoming and Utah. H2b: The puma population on the Uncompahgre Plateau study area will exhibit a young age structure after hunting prohibition at the beginning of the reference period. During the 5 years of hunting prohibition, greater survival of independent puma will cause an older age structure in harvest-age puma (i.e., adults and subadults) as suggested by the work of Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. H2c: As hunting is re-instated in the treatment period, the age structure of harvested puma and the harvest-age puma in the population will vary as observed by Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. Desired outcomes and management applications of this research include: 1. Quantification of variations in puma population density, sex and age structure, growth rates, vital rates, and an understanding of factors affecting them will aid adaptive puma management by yielding population parameters useful for estimating puma population abundance, evaluation of management alternatives, and effects of management prescriptions. 2. Testing assumptions about puma populations, currently used by CDOW managers, will help those managers to biologically support and adapt puma management based on Colorado-specific estimated puma population characteristics, parameters, and dynamics. 3. Methods for estimating puma abundance (capture-mark-recapture) of known reliability will allow managers to “ground truth” modeled populations and estimate effects of management prescriptions designed to achieve specified puma population objectives in targeted areas of Colorado. Ascertaining puma numbers and densities during the project will require development of reliable monitoring techniques based on capture-mark-recapture methods and models. Potential methods include direct and DNA genotype capture-recapture. Study plans to develop and test feasible field and analytical methods will be developed in the future after we have learned the logistics of performing those methods, after we have preliminary data on puma demographics and movements which will inform suitable sampling designs, and when we have adequate funding. 4. This information will be disseminated to citizen stakeholders interested in pumas in Colorado, and thus contribute to informed public participation in puma management. 109 �STUDY AREA The study area for the puma population research is on the Uncompahgre Plateau (in Mesa, Montrose, Ouray, and San Miguel Counties, Fig. 2). The study area includes about 2,253 km2 (870 mi.2) of the southern halves of Game Management Units (GMUs) 61 and 62, and about 155 km2 (60 mi.2) of the northern edge of GMU 70 (between state highway 145 and San Miguel River). The area is bounded by state highway 348 at Delta, 25 Mesa road and Forest Service road FS503 to Nucla, state highway 97 to state highway 141 to state highway 145 to Placerville, state highway 62 to Ridgeway, U.S. highway 550 to Montrose, and U.S. highway 50 to Delta. The study area seems typical of puma habitat in Colorado that has vegetation cover that varies from the pinon-juniper covered foothills starting from about 1,700 m elevation to the spruce-fir and aspen forests growing to the highest elevations of about 3,000 m. Mule deer (Odocoileus hemionus) and elk (Cervus elaphus) are the most abundant wild ungulates available for puma prey. There are cattle and domestic sheep raised on summer ranges on the study area. Year-round human residents live along the eastern and western fringe of the area, and there is a growing summer residential presence especially on the southern end of the plateau. A highly developed road system makes the study area well accessible for puma research efforts. A detailed description of the Uncompahgre Plateau is in Pojar and Bowden (2004). METHODS Reference and Experimental Treatment Periods This research is structured in two 5-year periods: a reference period (years 1―5) and a treatment period (years 6―10). The reference period is expected to cause a population increase phase. The treatment period will be managed to cause a population decline phase. In both phases, puma population structure, and vital rates will be quantified, and some management assumptions and hypotheses regarding population dynamics will be tested. Contingent upon results of pilot studies, we will also estimate puma numbers, population growth rates, evaluate enumeration methods, and test other hypotheses (Logan 2004). The reference period, without recreational puma hunting as a major limiting factor, is consistent with the natural history of the current puma species in North America which evolved life history traits during the past 10,000―12,000 years (Culver et al. 2000) that enable puma to survive and reproduce (Logan and Sweanor 2001). In contrast, puma hunting, with its modern intensity and ingenuity, might have influenced puma evolution in western North America for the past 100 years. Hence, the reference period, years 1―5, will provide conditions where individual puma in this population (of estimated sex and age structure) express life history traits interacting with the environment without recreational hunting as a limiting factor. Theoretically, the main limiting factors will be catchable prey abundance (Pierce et al. 2000, Logan and Sweanor 2001). This should allow researchers to understand basic system dynamics before the treatment (i.e., controlled recreational hunting). In the reference period, all puma in the study area will be protected, except for individual puma that might be involved in depredation on livestock or human safety incidents. In addition, all radio-collared and ear-tagged puma that range in a buffer zone, that includes the northern halves of GMUs 61 and 62, will be protected from recreational hunting. The reference period will allow researchers to quantify baseline demographic data on the puma population to estimate parameters for the CDOW’s model-based approach to puma management. Moreover, it will allow researchers to develop and test puma enumeration methods when population growth is known to be in one direction― increasing. Without the hunting closure, pilot data for enumeration methods could be confounded by not knowing if the population was increasing, declining, or stable. The reference period will also facilitate other operational needs (because hunters will not be killing the animals) including the marking of a large proportion of the puma population for capture-mark- 110 �recapture estimates, and the gathering of movement data from GPS-collared puma to help formalize exact sampling designs for enumeration methods. During the treatment period, years 6―10, experimentally structured recreational puma hunting will occur on the same study area with the intent of causing a decline phase in the puma population by using management prescriptions structured from information learned during previous years. Using recreational hunting for the treatment is consistent with the CDOW’s objectives of manipulating natural tendencies of puma populations, particularly survival, to maintain either population stability or population suppression (CDOW, Draft L-DAU Plans, 2004). Theoretically, puma survival will be influenced mainly by recreational hunting, which will be quantified by agent-specific mortality rates of radio-collared puma. The portion of adults and subadults in the population will be reduced by approximately 20% in year 6 and 20% more in year 7. The 20% change was identified by Division managers that requested enumeration tools that might detect 20% changes in puma populations. For managers, detecting the magnitude of puma population decline phases is probably more important that detecting the magnitude of population increase phases. This will also allow quantification of puma population characteristics and vital rates and initial tests of enumeration methods during a decline phase. Additional reductions may be made to test enumeration methods and other hypotheses that may be related to effects of hunting (i.e.,: relative vulnerability of puma sex and age classes to hunting, variations in puma population structure due to hunting) and puma-prey interactions (i.e., lines of research identified in the Colorado Research Program, Fig. 1). Those decisions can be made later in project development and as late as years 8―10. The killing of tagged and collared puma during the treatment period will not hamper operational needs (as it would during the start-up years), because by the beginning of this period, a large majority of independent puma in the population will be marked, and sampling schemes will be formalized. Puma on the study area that may be involved in depredation of livestock or human safety incidences may be lethally controlled. Researchers that find that GPS-collared puma have killed domestic livestock will record such incidents to facilitate reimbursement to the property owner for loss of the animal(s). In addition, researchers will notify the Area Manager of the CDOW of Wildlife if they perceive that an individual puma may be a threat to public safety. Field Methods Puma Capture: Realizing that puma live at low densities and capturing puma is difficult, as a starting point, our logistical aim will be to have a minimum of 6 puma in each of 6 categories (36 total) radio-tagged in any year of the study if those or greater numbers are present. The 6 categories are: adult female, adult male, subadult female, subadult male, female cub, male cub. Our aim is to provide more quantitative and precise estimates of puma demographics than were achieved in earlier Colorado puma studies. This relatively large number of puma might represent the large majority of the puma population on the study area, and will provide the basic data for age- and sex-specific reproductive rates, survival rates, agent-specific mortality rates, emigration rates, and movement data pertinent to sampling designs for various projects. Assuming that the puma population density on the study area is relatively low at the beginning of this study― about 1 adult/100 km2 and the sex ratio is equal (Anderson et al. 1992, Logan and Sweanor 2001:167), then there might be 22 adults, 11 males and 11 females. Also assuming that the total population contains 10% subadults and 34% cubs (Logan and Sweanor 2001), then there might be 4 subadults and 13 cubs with equal sex ratios in a total population of 39 puma. If we achieve our logistical aim in the first 1―2 years (recognizing that the population might grow), then we should be able to quantify population characteristics and vital rates for a majority of the puma population in those years and build upon the tagged number in each subsequent year. Thus, our inferences will pertain to the large 111 �majority of the puma population, if not the population on the study area, instead of a relatively small sample of it. We anticipate it may take 2 years to mark the large majority of puma in the population. In addition, the study area is large and will require some time to learn to access it efficiently. Puma capture and handling procedures have been approved by the CDOW Animal Care and Use Committee (file #08-2004). All captured puma will be examined thoroughly to ascertain sex and describe physical condition and diagnostic markings. Age of adult puma will be estimated initially by the gum-line recession method (Laundre et al. 2000) and dental characteristics of known-age puma (Logan and Sweanor, unpubl. data). Ages of subadult and cub puma will be estimated initially based on dental and physical characteristics of known-age puma (Logan and Sweanor unpubl. data). Body measurements recorded for each puma will include at a minimum: mass, pinna length, hind foot length, plantar pad dimensions. Tissue collections will include: skin biopsy (from the pinna receiving the 6 mm biopsy punch for the ear-tags) and blood (30 ml from the saphenous or cephalic veins) for genotyping individuals, parentage and relatedness analyses; disease screening; hair (from various body regions) and fecal DNA for genotyping tests of field gathered samples. Universal Transverse Mercator Grid Coordinates on each captured puma will be fixed via Global Positioning System (GPS, North American Datum 27). Puma will be captured year-round using 4 methods: trained dogs, cage traps, foot-hold snares, and by hand (for small cubs). Capture efforts with dogs will be conducted mainly during the winter when snow facilitates thorough searches for puma tracks and the ability of dogs to follow puma scent. The study area will be searched systematically multiple times per year by four-wheel-drive trucks, all-terrain vehicles, snow-mobiles, walking, and possibly horse- or mule-back. When puma tracks ≤1 day old are detected, trained dogs will be released to pursue puma to capture. Puma usually climb trees to take refuge from the dogs. Adult and subadult puma captured for the first time or requiring a change in telemetry collar will be immobilized with Telazol (tiletamine hydrochloride/zolazepam hydrochloride) dosed at 5 mg/kg estimated body mass (Lisa Wolfe, DVM, CDOW, attending veterinarian, pers. comm.). Immobilizing agent will be delivered into the caudal thigh muscles via a Pneu-Dart® shot from a CO2-powered pistol. Immediately, a 3m-by-3m square nylon net will be deployed beneath the puma to catch it in case it falls from the tree. A researcher will climb the tree, fix a Y-rope to two legs of the puma and lower the cat to the ground with an attached climbing rope. Once the puma is on the ground, its head will be covered, its legs tethered, and vital signs monitored (Logan et al. 1986). (Normal signs: pulse ~70―80 bpm, respiration ~20 bpm, capillary refill time ≤2 sec., rectal temperature ~101oF average, range = 95―104oF) (Kreeger 1996). A cage trap will be used to capture adults, subadults, and large cubs when puma can be lured into the trap using road-killed or puma-killed ungulates (Sweanor et al. 2005). Efficiency of the trap might be enhanced by using an automated digital call box that emits puma vocalizations (Wildlife Technologies, Manchester, NH). A cage trap will be set only if a target puma scavenges on the lure (i.e., an unmarked puma, or a puma requiring a collar change). Researchers will continuously monitor the set cage trap from about 1 km distance by using VHF beacons on the cage and door. This allows researchers to be at the cage to handle captured puma within 30 minutes. Puma will be immobilized with Telazol injected into the caudal thigh muscles with a pole syringe. Immobilized puma will be restrained and monitored as described above. If non-target animals are caught in the cage trap, we will open the door and allow the animal to leave the trap. Foot-hold snares will be used to capture adults, subadults, and large cubs only when safe snare sites at puma kills can be located as described by Logan et al. (1999). Snares set at puma kills will be monitored continuously with VHF beacons on the snares from about 1 km distance. We will not set snares at sites where tracks indicate that other mammals (e.g., deer, elk, bear, bighorn sheep, livestock) are also active. Puma will be immobilized with Telazol injected into the caudal thigh with a pole syringe. 112 �Vital signs will be monitored during the handling procedures. Efficiency of snares might also be enhanced with the use of an automated call box with puma or prey vocalizations. Small cubs (≤10 weeks old) will be captured using our hands (covered with clean leather gloves) or with a capture pole. Cubs will be restrained inside new burlap bags during the handling process and will not be administered immobilizing drugs. Cubs at nurseries will be approached when mothers are away from nurseries (as determined by radio-telemetry). Cubs captured at nurseries will be removed from the nursery a distance of ~100 m to minimize disturbance and human scent at nurseries. Immediately after handling processes are complete, cubs will be returned to the exact nurseries where they were found (Logan and Sweanor 2001). Marking, Global Positioning System, and Radio-telemetry: Puma do not possess easily identifiable natural marking, such as tigers (see Karanth and Nichols 1998, 2002), therefore, the capture, marking, and GPS- or VHF- collaring of individual puma is essential to a number of project objectives, including estimating vital rates and gathering movement data on puma to formalize designs for developing and testing enumeration methods. Adult, subadult, and cub puma will be marked 3 ways: GPS/VHF- or VHF-collar, ear-tag, and tattoo. The identification number tattooed in the pinna is permanent and cannot be lost unless the pinna is severed. A colored (bright yellow or orange), numbered rectangular (5 cm x 1.5 cm) ear-tag (Allflex USA, Inc., DFW Airport, TX) will be inserted into each pinna to facilitate individual identification during direct recaptures. Cubs ≤10 weeks old will be eartagged in only one pinna. Locations of GPS- and VHF-collared puma will be fixed about once per week from light fixedwing aircraft (e.g., Cessna 182) fitted with radio signal receiving equipment (Logan and Sweanor 2001). This monitoring will enable researchers to find GPS-collared puma to acquire remote GPS location reports from the ground, monitor the status (i.e., live or dead) of individual puma, and to recover carcasses for necropsy. It will also provide simultaneous location data on mothers and cubs. GPS- and VHF-collared puma will be located from the ground opportunistically using hand-held yagi antenna. At least 3 bearings on peak aural signals will be mapped to fix locations and estimate location error around locations (Logan and Sweanor 2001). Aerial and ground locations will be plotted on 7.5 minute USGS maps (NAD 27) and UTMs along with location attributes will be recorded on standard forms. GPS locations will be mapped using ArcGIS software. Adult and subadult female puma will be fitted with GPS collars (approximately 400 g each, Lotek Wireless, Canada). Initially, GPS-collars will be programmed to fix and store puma locations at 4 times per day to sample daytime, nighttime, and crepuscular locations (i.e., 0:00, 06:00, 12:00, 19:00). GPS locations for puma will provide precise, quantitative data on puma movements mainly to provide data to formalize study designs, to test assumptions for capture-mark-recapture methods for this project, and to assess the relevance of puma DAU boundaries. The GPS-collars also will provide basic information on puma movements and locations to design other pilot studies in this program on vulnerability of puma to sport-harvest, habitat use, and predation frequency on mule deer and elk. Subadult male puma will be fitted initially with conventional VHF collars (Lotek, LMRT-3, ~400 g each) with expansion joints fastened to the collars, which allows the collar to expand to the average adult male neck circumference (~46 cm). If subadult male puma reach adulthood on the study area, we will recapture them and fit them with GPS collars. VHF radio transmitters on GPS collars will enable researchers to find those puma on the ground in real time to acquire remote GPS data reports, facilitate recaptures for re-collaring, and to check on their reproductive and physical status. VHF transmitters on GPS- and VHF-collars will have a mortality mode 113 �set to alert researchers when puma have been immobile for at least 3 hours so that dead puma can be found to quantify survival rates and agent-specific mortality rates by gender and age. We will attempt to collar all cubs in observed litters with small VHF transmitter mounted on an expandable collar (~100g, MOD 210, Telonics, Inc., Mesa, Arizona) when cubs weigh 2.3―11 kg (5―25 lb). Cubs with mass ≥11 kg can still wear these small expandable collars until they are about 12 months old. Cubs approaching the age of independence (~11―14 mo. old) may be fit with Lotek LMRT-3 VHF collars (~400 g) with expansion links. Cubs will be recaptured to replace collars as necessary. Monitoring radioed cubs allow quantification of survival rates and agent-specific mortality rates (Logan and Sweanor 2001). Capture-Mark-Recapture: Capture-mark-recapture methods will be evaluated initially as a pilot study. Capturing and marking puma is time consuming, and would lengthen the time to thoroughly search the study area for capturing and marking puma during capture-recapture occasions needed for population estimation. Therefore, we will capture and mark puma prior to performing capture-recapture occasions using houndsmen teams. In addition, by marking puma before capture-recapture occasions begin, we will have opportunities to capture female puma at different stages of their reproductive status, and thus reduce the chance that mothers in a stage with suckling cubs and small activity areas are not detected and marked on the study area. After cubs are weaned, the mothers’ activity area expands (Logan and Sweanor 2001). The probability of females having suckling cubs in winter is naturally small; that season exhibits the lowest rate of births (Logan and Sweanor 2001). Capture-recapture occasions to estimate the population of independent puma may not begin until the end of the second winter or the third winter when we have a large majority of the puma population sampled and marked. Occasions performed at that time will be viewed as a pilot study allowing us to examine the logistics of the field methods, the extent to which model assumptions are met, performance of field methods (e.g., detection differences by sex or life stage as revealed by GPS data on collared puma), and precision of capture-recapture models used to estimate the puma population. Analytical Methods Population Characteristics: Population characteristics each year will be tabulated with the number of individuals in each sex and age category. Age categories, as mentioned, include: adult (puma ≥24 months old, or younger breeders), subadults (young puma independent of mothers, <24 months old that do not breed), cubs (young dependent on mothers, also known as kittens) (Logan and Sweanor 2001). When data allow, age categories may be further partitioned into months (for cubs and subadults) or years (for adults). Reproductive Rates: Reproductive rates will be estimated for GPS- and VHF-collared female puma directly (Logan and Sweanor 2001). Genetic paternity analysis will be used to ascertain paternity for adult male puma (Murphy et al. 1998). Methods will be tested in Dr. M. Douglas’s Laboratory (Colorado State University, Department of Fishery and Wildlife Biology). Survival and Agent-specific Mortality Rates: Radio-collared puma will provide known fate data which can be used to analyze survival rates in program MARK (White and Burnham 1999, Cooch and White 2004). Agent-specific mortality rates will be analyzed using proportions and Trent and Rongstad procedures (Micromort software, Heisey and Fuller 1985). Cub survival curves for each gender will be plotted with survival rate on age in months (Logan and Sweanor 2001:119). Population Estimates: Capture-recapture models will be evaluated initially as a pilot study to estimate the parameters of primary interest― absolute numbers of independent puma (i.e., number of adult and subadult puma present in the survey area) and puma density (i.e., number of independent puma/100 km2) each winter― December through March― when snow facilitates detection and capture of 114 �puma, provided that we meet model assumptions. The December―March period also corresponds with Colorado’s puma hunting season. The population of interest is independent puma (i.e., adults and subadults) because those are the puma that can be legally killed by recreational hunters. Furthermore, adults comprise the breeding segment of the population and subadults are non-breeders that are potential recruits into the adult population in ≤1 year. Thus, the sampling unit is the individual independent puma (~≥1 yr. old). General assumptions for closed capture-recapture models are: (1) the population is closed; (2) animals do not lose their marks during the interval; (3) all marks are correctly noted and recorded at each trapping occasion; (4) each animal has a constant and equal probability of capture on each capture occasion. Open population models allow the assumption of closure to be relaxed (Otis et al. 1978, White et al. 1982, Pollock et al. 1990). The robust design is a combination of closed and open models; thus, assumptions are a combination of the assumptions for closed and open population methods (Kendall 2001). To analyze capture-recapture data, closed, open, and the robust design models are available in program MARK. Akaike’s Information Criterion will be used to select the most parsimonious models based on AICc score ranks and the difference in AIC (∆AIC) between models (Burnham and Anderson 1998). MARK results also include estimates of abundance. Because the precision of estimates for small populations is sensitive to the probability of capture (White et al. 1982, Pollock et al. 1990), our operational goal will be to achieve capture probabilities of at least 0.5 for each animal per capture occasion. Capture simulations using MARK software (Cooch and White 2004) indicate that greater capture probabilities and more capture occasions yield more precise estimates. The capture probability for the simplest closed model [M(o)], which assumes that every member of the population has the same probability of capture (p) for each sampling period, suggest that for a population of 30 animals (i.e., adults plus subadult puma, which might be present by the end of year 2, see Puma Capture above) p must equal 0.5 for 3 capture occasions to attain a coefficient of variation (V) of 0.1. If 6 capture occasions are used, then a p of 0.3 might yield a V of 0.09. In addition, behavior, movements, survival and mortality of GPS- and VHF-collared puma will allow direct biological examinations of assumptions of geographic and demographic closure (White et al. 1982) and variation in capture probability of individual puma and puma classes (i.e., adult females, adult males, subadult females, subadult males). If capture probabilities vary by puma class, we will examine if data stratification is necessary or possible (depending upon sample size). For example, we might expect the larger home ranges of male puma to expose them to more search routes, thus, this may increase their probability of capture. If the assumption of demographic closure cannot be satisfied, then open population models and the robust design would be more appropriate (Pollock et al. 1990, Williams et al. 2001). Collared puma will allow us to determine the number of marked puma present in the search area each capture-recapture occasion. Furthermore, GPS locations (4 fixes/day) on individual puma will provide data on the probability that puma may temporarily move out of and back into the survey area between capture occasions. Unmarked puma that are subsequently GPS-collared should provide such information, too. ArcView geographic information system software will be used to map and analyze puma locations, movements, and home ranges. It will also be used to map and quantify attributes of the study area and sampling frames. Rate of Population Increase: Finite rates of increase (λ=Nt+1/Nt) between consecutive years and average annual rates of increase (r) for 3- to 5-year periods and levels of precision will be calculated (Caughley 1978, Van Ballenberghe 1983) and plotted. 115 �Functional Relationships: Graphical methods will be used to examine functional relationships between puma density and vital rates, relationships between puma density estimated with direct capturerecapture methods (i.e., houndsmen teams) and possibly later (depending upon funding) by using estimates from DNA genotype mark- recapture methods. Linear regression procedures and coefficients of determination can be used to assess these functional relationships if data for the response variable are normally distributed and the variance is the same at each level. If the relationship is not linear, data is non-normal, and variances are unequal, we will consider appropriate transformations of the data for regression procedures (Ott 1993). Non-parametric correlation methods, such as Spearman’s rank correlation coefficient, can also be used to test for monotonic relationships between puma abundance and other parameters of interest (Conover 1999). Statistical analyses will be performed using SYSTAT and SAS software. The risk of committing a type I error (i.e., rejecting a null hypothesis that is actually true) will be controlled at alpha = 0.10 because we will normally have small population sizes (typical of studies of large obligate carnivores). The higher alpha level will increase the probability of detecting a change and reduce the risk of a type II error (i.e., failing to reject a null hypothesis that is false). For managers, the risk of a type II error is probably more important. RESULTS AND DISCUSSION Segment Objective 1 The Division of Wildlife held public meetings in Redvale (August 23, 2004) and Montrose (August 30, 2004) where DOW staff informed attendees from the Uncompahgre Plateau area about the puma research project and addressed their questions. Meetings were held with over 70 private landowners, ranchers, hunters, outfitters, and guides that live and operate on the Uncompahgre Plateau to inform them about the puma research, address questions, and request permission to access private lands for puma research activities. Additional meetings were with representatives of the U.S. Forest Service, Bureau of Land Management, National Park Service, and Colorado State Parks who were also informed about the puma research. Segment Objective 2 The Wildlife Commission passed regulations allowing for the experimental design of this puma research. Their decision resulted in a closure to puma sport-hunting for the first 5-years of the research (Nov. 11, 2004 to Mar. 31, 2009) on the study area. In addition, the Commission allowed the creation of a buffer zone during the same time period comprised of the remaining parts of Game Management Units 61 and 62 north of the 25 Mesa Road (i.e., north of the study area) where pumas tagged on the study area can not be legally taken by puma sport-hunters. The buffer zone is intended to protect puma that are originally captured and sampled in the study area and that range to the north so that pumas in the study population will express life history traits not affected by sport-hunting off-take. A larger buffer zone to protect pumas tagged on the study area was requested of the CDOW Regulations Review Committee. That buffer zone would have protected all puma tagged in the study area even if they ranged off the study area but were west of the continental divide in Colorado. However, that request was denied by the Regulations Review Committee. Segment Objective 3 A study plan was developed, peer-reviewed, modified with the peers’ recommendations (Logan 2004), and then initiated to begin the long-term, experimental research on puma population dynamics on the Uncompahgre Plateau. Procedures for the capture, restraint, handling, and sampling of pumas for this research were reviewed and approved (file #08-2004) by the Colorado Division of Wildlife Animal Care and Use Committee. 116 �Segment Objectives 4―7 Field research to begin quantifying puma population structure, vital rates, and causes of mortality began on Dec. 2, 2004. From December 2, 2004 to May 12, 2005, trained dogs were used as our main method to capture, sample, and mark pumas. Our search efforts on the east slope of the study area were from 25 Mesa Road south to Fisher Creek. On the west slope, our efforts were from the 25 Mesa Road south to Goodenough Gulch. Those efforts resulted in 78 search days, 109 puma tracks, 35 pursuits, and the capture of 14 pumas (Table 1). Eight of those pumas were restrained, sampled, tagged, and released (Table 2). Puma M1 was unintentionally recaptured when we thought we were pursuing an unmarked puma. Puma F3 was recaptured during our effort to capture her male offspring M5 for the first time. Pumas were bayed in trees by dogs on 4 other occasions, but we did not attempt to anesthetize the puma because of concern for the pumas safety on 3 occasions, and concern for the puma’s and the researchers’ safety on one occasion (Table 3). In those cases, a puma was treed in cliffs at night, and on 3 occasions the pumas were bayed in trees too dangerous for researchers to attempt to safely dart the pumas and then climb the trees to retrieve the cats. These pumas included 1 large adult female that was probably caught twice, 1 adult male, and 1 puma that was either a large cub or a subadult (sex undetermined). A summary of capture efforts with dogs is in Table 4. We attempted to capture a female puma on Ridgeway State Park on April 1, 2005. The puma killed an adult mule deer doe, and had begun to eat the deer on the sidewalk beside the Fishing Pond at Pa-Co-Chu-Puk Campground. This location was about 520 m east of where our trained dogs treed, but we could not handle a large female puma on February 1, 2005. We used a cage trap designed for black bears to attempt to capture the puma, but the bear trap was not sufficient. The puma entered the trap, but apparently the cage door did not latch because the puma’s tail was caught in the door jam. The puma did not return to cage trap. We did not pursue the puma with dogs because of the close proximity of highway 550 and private lands directly north and east of the park. We captured 8 cubs from 4 litters born to GPS-collared female pumas. Two litters were born in May, 1 litter was born in June, and 1 litter was born in August. There were 3 males and 5 females (Table 5). One puma death was detected on July 28, 2005. A female, about 49 months old, was hit and killed by a car between 06:00―08:00 on state highway 62 about 10.4 km west of Ridgeway in lower Cottonwood Creek. This location was on the southern boundary of the study area. A necropsy showed that the puma appeared to be in excellent physical condition prior to its death. Her mass was 46 kg; she apparently was not pregnant; and her mammary glands were not producing milk. Segment Objectives 8―9 Seven adult pumas were fit with Lotek 4400S GPS collars programmed to fix 4 locations per day (00:00, 06:00, 12:00, and 19:00). The number of GPS locations per individual puma ranged from 355 to 779 (Table 6). Because none of the puma have yet been monitored for a complete year and the sample is small, annual and seasonal home ranges sizes were not estimated for this report. However, we estimated the activity areas used by the 7 GPS-collared adult pumas (Table 6) during the monitoring periods and overlaid 100% Minimum Convex Polygons on a map of the study area (Fig. 2). In addition, we are collaborating with colleagues at Colorado State University― Dr. K. Crooks, Dr. D. Theobald, and Dr. K. Wilson― to develop a proposal and funding that would allow us to develop and validate puma habitat suitability models and maps for Colorado in which these puma GPS location data will be used. Tissue samples from all of the captured pumas and the unmarked female puma hit and killed by a car have been archived with geneticist Dr. M. Douglas. We are currently collaborating with Dr. Douglas to develop a study plan and funding for the development and assessment of laboratory and field methods 117 �for genotyping pumas and for estimating puma abundance in the wild using DNA mark-recapture techniques. We conducted a preliminary assessment of the usefulness of GPS-collar technology for investigations of puma-prey relationships. The average GPS location fix rate for the 7 GPS-collared pumas was 70.7% (range = 54―87%) (Table 6). We investigated 139 GPS location clusters for 7 adult pumas where individual GPS-collared pumas spent ≥1 day during the span December 26, 2004 to July 31, 2005. The estimated error between 101 collar-fixed GPS locations and prey remains found on the ground averaged 3.2 m (range = 0―50 m, SE = 0.6). Prey remains were found at 112 of the 139 clusters, with mule deer and elk comprising 54% and 43%, respectively (Table 7). The sex and age stage structure of 60 mule deer and 48 elk used by puma at GPS clusters is in Table 8. On average, puma spent 2.3 days on mule deer (range = 1―6, SE = 0.2) and 2.9 days on elk (range = 1―10, SE = 0.3). Ungulate use rates by the GPS-collared pumas estimated from these data are in Table 9. Evidence that black bears (Ursus Americana) used portions of the same ungulates used by GPS-collared pumas was found at remains of 7 mule deer and 10 elk. Evidence that coyotes (Canis latrans) used portions of the same ungulates used by GPS-collared pumas was found at remains of 7 mule deer and 14 elk. We are currently assessing how this GPS-collar capability could be used to structure research on puma-ungulate relationships on the Uncompahgre Plateau and the additional funding and personnel needed to thoroughly execute the research. SUMMARY Experimental, long-term research on puma population dynamics, effects of sport-hunting, and development and testing of puma enumeration methods began in December 2004. From December 2004 to July 2005 fifteen pumas were captured, sampled, marked, and released. The number of pumas handled is partially contingent upon effort, the number of pumas present on the study area, and safety concerns. Individual pumas sampled in the population provide quantitative information on population structure, vital rates, and dynamics over time in reference and treatment periods to improve the CDOW’s puma management. All pumas were sampled as part of developing research for genotype mark-recapture procedures. Seven adult puma were fit with GPS collars, yielding 487―779 locations. Puma GPS location data will be used to: design enumeration methods in the field, develop and test puma habitat suitability models and maps, and develop potential research on puma-ungulate relationships on the Uncompahgre Plateau contingent upon funding and support. Research efforts for year 2 will focus on increasing numbers and distribution of sampled, marked, and GPS/radio-collared puma on the study area for data to address the objectives, management assumptions, and hypotheses in the study plan. We will further develop proposals for the puma genetics research, puma habitat suitability models and maps, and puma-prey relationships. 118 �LITERATURE CITED ANDERSON, A. E., D. C. BOWDEN, AND K. M. KATTNER. 1992. The puma on Uncompahgre Plateau, Colorado. Technical Publication No. 40. Colorado Division of Wildlife, Denver. ANDERSON, C. R., JR., AND F. G. LINDZEY. 2005. Experimental evaluation of population trend and harvest composition in a Wyoming cougar population. Wildlife Society Bulletin 33:179-188. BURNHAM, K. P., AND D. R. ANDERSON. 1998. Model selection and inference: a practical informationtheoretic approach. Springer-Verlag, New York, New York, USA. CAUGHLEY, G. 1978. Analysis of vertebrate populations. John Wiley and Sons, New York. COLORADO DIVISION OF WILDLIFE 2002-2007 Strategic Plan. 2002. Colorado Department of Natural Resources, Division of Wildlife. Denver. CONOVER, W. J. 1999. Practical nonparametric statistics. John Wiley & Sons, Inc., New York. COOCH, E., AND G. WHITE. 2004. Program MARK- a gentle introduction, 3rd edition. Colorado State University, Fort Collins. CULVER, M., W. E. JOHNSON, J. PECON-SLATTERY, AND S. J. O’BRIEN. 2000. Genomic ancestry of the American puma (Puma concolor). The Journal of Heredity 91:186-197. CURRIER, M. J. P., AND K. R. RUSSELL. 1977. Mountain lion population and harvest near Canon City, Colorado, 1974-1977. Colorado Division of Wildlife Special Report No. 42. HEISEY, D. M., AND T. K. FULLER. 1985. Evaluation of survival and cause specific mortality rates using telemetry data. Journal of Wildlife Management 49:668-674. KENDALL, W.L. 2001. The robust design for capture-recapture studies: analysis using program MARK. Pages 357-360 in R. Field, R. J. Warren, H. Okarma, and P. R. Sievert, editors. Wildlife, land, and people: priorities for the 21st century. Proceedings of the Second International Wildlife Management Congress. The Wildlife Society, Bethesda, Maryland, USA. KOLOSKI, J. H. 2002. Mountain lion ecology and management on the Southern Ute Indian Reservation. M. S. Thesis. Department of Zoology and Physiology, University of Wyoming, Laramie. KREEGER, T. J. 1996. Handbook of wildlife chemical immobilization. Wildlife Pharmaceuticals, Inc., Fort Collins, Colorado. LAUNDRE, J. W., L. HERNANDEZ, D. STREUBEL, K. ALTENDORF, AND C. L. LOPEZ GONZALEZ. 2000. Aging mountain lions using gum-line recession. Wildlife Society Bulletin 28:963-966. LOGAN, K. A., E. T. THORNE, L. L. IRWIN, AND R. SKINNER. 1986. Immobilizing wild mountain lions (Felis concolor) with ketamine hydrochloride and xylazine hydrochloride. Journal of Wildlife Diseases. 22:97-103. __________, L. L. SWEANOR, J. F. SMITH, AND M. G. HORNOCKER. 1999. Capturing pumas with foothold snares. Wildlife Society Bulletin 27:201-208. __________, AND L. L. SWEANOR. 2001. Desert puma: evolutionary ecology and conservation of an enduring carnivore. Island Press, Washington, D.C. __________. 2004. Colorado puma research and development program: population characteristics and vital rates study plan. Colorado Division of Wildlife, Ft. Collins Research Center, Ft. Collins. MURPHY, K., M. CULVER, M. MENOTTI-RAYMOND, V. DAVID, M. G. HORNOCKER, AND S. J. O’BRIEN. 1998. Cougar reproductive success in the Northern Yellowstone Ecosystem. Pages 78-112 in The ecology of the cougar (Puma concolor) in the Northern Yellowstone ecosystem: interactions with prey, bears, and humans. Dissertation, University of Idaho, Moscow. OTIS, D. L., K. P. BURNHAM, G. C. WHITE, AND D. R. ANDERSON. 1978. Statistical inference from capture data on closed animal populations. Wildlife Monographs 62:1-135. OTT, R. L. 1993. An introduction to statistical methods and data analysis. Fourth edition. Wadsworth Publishing Co., Belmont, California. PIERCE, B. K., V. C. BLEICH, AND R. T. BOWYER. 2000. Social organization of mountain lions: does land a tenure system regulate population size? Ecology 81:1533-1543. POLLOCK, K. H., S. R. WINTERSTEIN, C. M. BUNCK, AND P. D. CURTIS. 1989a. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. 119 �_____, S. R. WINTERSTEIN, AND M. J. CONROY. 1989b. Estimation and analysis of survival distributions for radio tagged animals. Biometrics 45:99-109. _____, J. D. NICHOLS, C. BROWNIE, AND J. E. HINES. 1990. Statistical inference for capture-recapture experiments. Wildlife Monographs 107:1-97. POJAR, T. M., AND D. C. BOWDEN. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550-560. ROSS, P. I., AND M. G. JALKOTZY. 1992. Characteristics of a hunted population of cougars in southwestern Alberta. Journal of Wildlife Management 56:417-426. STONER, D. C. 2004. Cougar exploitation levels in Utah: implications for demographic structure, metapopulation dynamics, and population recovery. Master of Science Thesis. Utah State University. SWEANOR, L. L., K. A. LOGAN, J. W. BAUER, B. MILSAP, AND W. M. BOYCE. 2005 in review. Pumahuman relationships in Cuyamaca Rancho State Park, California. Wildlife Society Bulletin. VAN BALLENBERGHE, V. 1983. Rage of increasse of white-tailed deer on the George Reserve: a reevaluation. Journal of Wildlife Management 47:1245-1247. WATKINS, B. 2004. Mountain lion data analysis unit L-22 management plan. Colorado Division of Wildlife, Montrose. WILLIAMS, B. K., J. D. NICHOLS, AND M. J. CONROY. 2001. Combining closed and open mark-recapture models: the robust design. Pages 523-554 In Analysis and management of animal populations. Academic Press, New York. WHITE, G. C., D. R. ANDERSON, K. P. BURNHAM, AND D. L. OTIS. 1982. Capture-recapture and removal methods for sampling closed populations. Los Alamos National Laboratory Publication LA-8787NERP. Los Alamos, NM, U.S.A. WORTON, B. J. 1995. Using Monte Carlo simulation to evaluate kernel-based home range estimators. Journal of Wildlife Management 59:794-800. Prepared by: ________________________ Kenneth A. Logan, Wildlife Researcher 120 �Table 1. Puma capture efforts with dogs from December 2, 2004 to May 12, 2005, Uncompahgre Plateau, Colorado. Month No. No. & type of puma No. & type of pumas No. & I.D. or type of Search tracks founda pursued pumas captured Days December 19 20 tracks: 11 male, 9 4 pursuits: 3 males, 1 2 pumas captured: female female M1, 1 female not handled January 15 26 tracks: 9 male, 15 8 pursuits: 4 males, 4 4 pumas captured: female, 2 cub females M1 recaptured, F2, F3, M4 February 17 22-23 tracks: 5 male, 11 pursuits: 2 males, 7 6 pumas captured: 14 female, 2-4 cubs, females, 2 cubs, or 1 1 female not handled, or 2-3 cubs & 1 cub & 1 subadult F3 recaptured, cub subadult M5, M6, 1 cub or subadult, F7 March 11 17 tracks: 8-9 male or 2 pursuits: 2 females 1 puma captured: F8 1 large cub, 7 female, 1 unspecified sex April 9 13 tracks: 10 male, 3 2 pursuits: 2 males 1 puma captured: female 1 male not handled May 7 10 tracks: 4 male, 6 8 pursuits: 3 males, 5 0 pumas captured female females 78 109 tracks found: 47- 35 pursuits: 14 males, 14 captures: (6 males, TOTALS 48 male, 54 female, 19 females, 1 male 6 females, 1 male cub, 4-6 cub or 0-1 cub, 1 cub (unknown 1 cub (unknown sex) subadult, 1 sex) or 1 subadult or 1 subadult unspecified (unknown sex) (unknown sex) a Puma hind-foot tracks with plantar pad widths >52 mm wide are assumed to be male; ≤52 mm are assumed to be female. Table 2. Pumas that were captured with aid of dogs, sampled, tagged, and released from December 2, 2004 to May 12, 2005, Uncompahgre Plateau, Colorado. Puma Sex Estimated Mass Capture Location I.D. Age (mo.) (kg) date M1 male 33 68 12-08-04 Shavano Valley M4 male 25-33 65 01-28-05 McKenzie Butte Mesa M5 male 6 12 02-04-05 Spring Creek M6 male 33 59 02-18-05 Happy Canyon F2 female 49 43 01-07-05 Dolores Canyon F3 female 41 40 01-21-05 Spring Creek F7 female 56-64 32 02-24-05 Dolores Canyon F8 female 21 30 03-21-05 Cottonwood Creek (W) 121 �Table 3. Pumas that were captured with aid of dogs, but were not handled for safety reasons, from December 2, 2004 to May 12, 2005, Uncompahgre Plateau, Colorado. Puma sex Age Capture Location Comments stage date Female adult 12-23-05 McKenzie Butte Mesa Large female. Female adult 02-01-05 South McKenzie Butte Mesa This puma probably same animal caught 12-23-05. Unspecified cub or 02-24-05 Dolores Canyon This puma apparently in subadult association with F7 at an elk kill. Possibly F7’s offspring or an unrelated subadult. Male adult 04-05-05 Horsefly Canyon (E) This puma, or another male, was pursued on 4 other occasions in the San Miguel River-toCottonwood Creek area. Table 4. Summary of puma capture efforts with dogs, December 2004 to May 2005, Uncompahgre Plateau, Colorado. Period Track Pursuit effort Puma capture Effort to capture a detection effort puma for the first time effort 11 pumas captured for 14/78 = 0.18 Dec. 2, 109/78 = 1.40 35/78 = 0.45 first time (minus M1, F3, capture/day 2004 tracks/day pursuit/day & large female) to 78/14 = 5.57 11/78 = 0.14 capture/day 78/35 = 2.23 May 12, day/capture day/pursuit 2005 78/11 = 7.09 day/capture Table 5. Puma cubs sampled on the Uncompahgre Plateau Puma Study area, 2004 to 2005. Cub Sex Estimated Estimated age Mass Mother Estimated age of I.D. birth date at capture (kg) mother at birth of this litter (mo) a M5 male August 2004 6 months 12 F3 36 F9 female May 28, 2005b 31 days 2.27 F2 44 F10 female May 28, 2005b 31 days 2.04 “ “ M11 male May 28, 2005b 31 days 2.27 “ “ 42 days 2.63 F7 59-67 F12 female May 19, 2005b F13 female May 19, 2005b 42 days 1.72 “ “ F14 female June 26, 2005b 26 days 1.90 F8 24 26 days 2.0 “ “ M15 male June 26, 2005b a Estimated age of M5 was based on morphometric comparisons with known-age cubs (Logan and Sweanor 2001, and unpublished data). b Estimated age of cubs sampled at nurseries is based on the starting date for GPS location foci for mothers at nurseries. 122 �Table 6. Numbers of GPS locations for adult puma on the Uncompahgre Plateau, Colorado, December 2004 to August 2005. Puma I.D. Sex Age Dates monitored a No. Acquisition rate Use areas estimated (km2) b stage locations average, range, n with 100% Minimum Convex Polygonc M1 male adult 12-08-04 to 08-19-05 779 76, 73―83, 5 815 M4 male adult 01-28-05 to 07-25-05 487 73, 57―84, 5 254 M6 male adult 02-18-05 to 07-25-05 543 87, 82―93, 5 552 F2 female adult 01-07-05 to 08-10-05 565 65, 43―82, 7 120 F3 female adult 01-21-05 to 08-02-05 586 76, 67―85, 6 174 F7 female adult 02-24-05 to 07-26-05 362 54, 26―78, 5 94 F8 female adult 03-21-05 to 08-08-05 355 64, 48―78, 4 245 a GPS collars on pumas are remotely downloaded at approximately 1-month intervals. The last date in Dates monitored is for the last location from the last GPS data download for an individual puma for this report. b n = number of remote downloads. c Polygons for individual GPS-collared puma are overlaid on a study area map in Figure 2. Table 7. Observations at GPS location clusters for 7 GPS-collared puma on the Uncompahgre Plateau, Colorado, December 2004 to July 2005. Puma No. Dates of GPS clusters Mule Elk Porcupine Beaver Puma Only Only Nothing No. GPS I.D. GPS that were deer scavenge Puma Black bear found clusters clusters investigated or sharea signb signc not visitedd M1 23 12-26-04 to 07-10-05 4 14 1 4 2 M4 16 02-03-05 to 07-12-05 4 7 1 4 2 M6 17 02-18-05 to 07-07-05 3 11 2 1 0 4 F2 26 01-12-05 to 07-26-05 12 9 2 2 1 0 1 F3 27 01-27-05 to 07-31-05 22 5 0 0 F7 18 03-08-05 to 07-22-05 9 1 5 1 2 0 F8 11 03-23-05 to 07-03-05 7 2 1 1 1 0 139 61 48 2 1 4 10 2 11 9 Total a A GPS-collared puma either shared a prey item with another GPS-collared puma (2 instances), or a GPS-collared puma scavenged on remains of prey previously used by another GPS-collared puma (2 instances). b Only puma tracks, feces, and/or beds were found at the GPS cluster. c Only black bear sign (e.g., feces) was found at the puma GPS cluster. d Some puma GPS clusters were not investigated because clusters fell on small private land holdings where we did not have permission for access at the time, or other principal objectives of our research were priority. 123 �Table 8. Sex and age structure of mule deer and elk found at GPS location clusters for 7 GPS-collared adult puma on the Uncompahgre Plateau, Colorado, December 2004 to August 2005. Sex No. Fawn/Calf Yearling 2+ years Unknown age Mule deer Female 26 2 2 20 2 Male 10 0 3 7 0 Unknown 25 13 3 3 6 Total 61 15 8 29 8 Elk Female Male Unknown Total 25 5 18 48 12 0 16 28 1 0 0 1 12 5 1 18 0 0 1 1 Table 9. Estimated ungulate use rates of adult GPS-collared pumas on the Uncompahgre Plateau, Colorado, December 2004 to July 2005. Puma Dates starting with & No. days inclusive in No. Estimated No. I.D. ending with ungulate use date span ungulates ungulates used used per yeara M1 12-26-04 to 07-10-05 196 18 33.5 M4 02-03-05 to 07-04-05 152 11 26.4 M6 02-18-05 to 07-07-05 140 14 36.5 F2 01-12-05 to 07-26-05 195 21 39.3 F3 01-27-05 to 07-31-05 185 27 53.3 F7 03-08-05 to 07-16-05 131 9 25.1 F8 03-23-05 to 07-03-05 103 9 31.9 a Estimated ungulate use rates per year are based on the key assumption that the individual puma would use ungulates throughout the year equal to the same rate recorded during the monitoring span in Dates starting with & ending with ungulate use. This assumption is probably not reliable especially for adult female pumas, because their reproductive status, and thus energetic needs vary throughout the year. For example, F3 was raising cubs born in August 2004; yet, F2, F7, and F8 started raising cubs born in May, May, and June of 2005, respectively. In addition, not all GPS clusters were investigated for M1, M4, M6, and F2 (see Table 7). 124 �GOAL: Strategies, Information, & Tools for Managing Healthy, Self-sustaining Puma Populations in Colorado Puma Population Effects of Harvest & Other Mortality Movements & Corridors Population Dynamics: Sex & Age, Vital Rates, Mortality, Population G h Vulnerability to Harvest Puma Habitat Human Development Habitat Use Effects of Translocation Estimation Methods for Monitoring Deer, Elk, Other Natural Prey & Species of Concern Domestic Animals Puma― Human Relationships Effects of Predation Effects of Evasive Conditioning Map Prey Distribution Habitat Models Population Models Puma Prey Habitat Maps Puma―Prey Relationships Models Figure. 1. An ecologically-based conceptual model of the Colorado Puma Research Program that provides the contextual framework for this and proposed puma research in Colorado. Gray-shaded shapes identify areas of research addressed by this report and the puma management goal (at top). 125 �![ ![ ![ Delta M1 ![ Olathe ![ Montrose F8 F3 ![ Nucla F7 ![ Naturita M4 F2 Norwood ![ M6 L ![ Ridgeway ![ ![ Legend CJ County Boundaries D 0 5 10 20 Kilometers Study Area . Figure 2. The Uncompahgre Plateau Puma Study Area with activity areas of adult GPS-collared pumas depicted with 100% Minimum Convex Polygons, December 2004 to August 2005. 126 �Colorado Division of Wildlife July 2005 – June 2006 WILDLIFE RESEARCH REPORT State of Cost Center Work Package Task No. Colorado 3430 3003 1 Federal Aid Project: N/A : Division of Wildlife : Mammals Research : Predatory Mammals Conservation : Puma Population Structure and Vital Rates on the Uncompahgre Plateau : Period covered: July 1, 2005―June 30, 2006 Author: K. A. Logan. Personnel: K. Logan, S. Waters, B. Bavin, B. Simpson, K. Crane, T. Mathieson, M. Caddy, and T. Smith of CDOW, J. Bauer of Colorado Cooperative Fishery and Wildlife Research Unit, J. Kane, V. Johnson, S. Young, and J. McNamara of U.S.D.A. Wildlife Services, volunteers, cooperators including: private landowners, U.S. Forest Service, Bureau of Land Management, and Colorado State Parks, with financial support received from The Howard G. Buffett Foundation and Safari Club International Foundation. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Research continued on puma population characteristics and dynamics on the Uncompahgre Plateau. Puma capture efforts resulted in a total of 36 puma captures (14 adults [1 adult female captured twice], 4 subadults, 14 cubs, and 2 adult or subadult males [1 captured twice] but not handled) with 20 radio-collared pumas within the study area as of July 2006. Efforts to capture, sample, and mark pumas with the use of trained dogs extended from November 21, 2005 to May 26, 2006. This resulted in 14 puma captures, including 1 adult female, 1 subadult female, 2 adult males, and 1 subadult or adult male captured and processed for the first time. Two other males were captured (one of them twice), but were not handled for safety reasons. The remainder was recaptures of previously marked pumas, including 2 adult females (1 recaptured twice), 1 adult male, 1 subadult male, and 1 male cub. We substantially increased puma capture efforts with ungulate carcasses to bait pumas into cage traps. From August 2, 2005 to June 27, 2006, we used 77 road-killed mule deer, 3 road-killed elk, 3 puma-killed mule deer, and 1 puma killed elk at 23 different sites. This resulted in 11 puma captures, including 4 adult females, 1 adult male, and 1 male cub captured and processed for the first time, and 3 adult females, 1 subadult male, and 1 male cub that were recaptured. Eleven other puma cubs (4 males, 7 females) from 4 litters were captured by hand at nurseries and processed for the first time. We investigated 4 puma mortalities: one adult male was killed by another male puma, 2 cubs (1 male, 1 female) were killed and eaten by other pumas, and 1 female cub died due to the expandable radiocollar she was wearing. To date, 14 pumas (5 males, 9 females) have been monitored with GPS collars, yielding 113 to 1,784 locations per puma, and a total of 13,139 GPS locations. We began quantifying the frequency that puma mothers are away from their cubs during the Colorado puma hunting season (Nov. through Mar.) as a preliminary assessment of potential vulnerability of mothers to harvest. Radio-collared members (mothers and cubs) of 5 families were located 79 times during fixed-wing flights from November 9, 2005 to March 29, 2006. Mothers 95 �were apart from their cubs >600 m during 12 of those occasions (15.2%). Preliminary comparisons between our current puma research on the Uncompahgre Plateau (~21 months duration) and results of the Anderson et al. (1992) puma research on the plateau (~7 years duration 1981-1988) are made where appropriate. We collaborated with colleagues to develop 3 proposals to contribute to the Colorado puma management program. Proposed work includes: testing genetic techniques for non-invasive methods to estimate puma numbers using mark-recapture methods and models, developing state-wide puma habitat models and maps, and assessing puma health. In addition, we will resume quantifying puma use frequencies of ungulates, and considering how research of pumas on developed areas on the Uncompahgre Plateau can contribute to the CDOW’s efforts to study puma-human interactions on the Colorado Front Range. 96 �WILDLIFE RESEARCH REPORT PUMA POPULATION STRUCTURE AND VITAL RATES ON THE UNCOMPAHGRE PLATEAU, COLORADO KENNETH A. LOGAN P. N. OBJECTIVE Quantify puma population sex and age structure; estimate puma population vital rates, including: reproduction rates of females, age-stage survival rates, and immigration and emigration rates; quantify agent-specific mortality rates― all to improve the Colorado Division of Wildlife’s (CDOW) model-based approach to managing pumas in Colorado. SEGMENT OBJECTIVES 1. 2. 3. 4. 5. Continue gathering data on puma population sex and age structure. Continue gathering data for estimates of puma reproduction rates. Continue gathering data to estimate puma sex and age-stage survival rates. Continue gathering data to estimate agent-specific mortality rates. Continue gathering data on puma movements for the development of sampling methods for markresight or recapture population estimates that might involve sampling puma DNA-genotypes, trail cameras, or direct observations. 6. Begin gathering data on spatial relationships of puma mothers to their cubs during the Colorado puma hunting season as a preliminary assessment of the vulnerability of puma mothers to sport-hunting harvest. 7. Evaluate other data sources that could come from this research that can be developed into other puma research relevant to CDOW biologists and managers. INTRODUCTION Colorado Division of Wildlife managers need reliable information on puma biology and ecology in Colorado to develop sound management strategies that address diverse public values and the CDOW objective of actively managing puma while “achieving healthy, self-sustaining populations”(CDOW 2002-2007 Strategic Plan:9). Although 4 puma research efforts have been made in Colorado since the early 1970s and puma harvest data is compiled annually, reliable information on certain aspects of puma biology and ecology, and management tools that may guide managers toward effective puma management is lacking. Mammals Research staff held scoping sessions with a number of the CDOW’s wildlife managers and biologists. In addition, we consulted with other agencies, organizations, and interested publics either directly or through other CDOW employees. In general, CDOW staff in western Colorado highlighted concern about puma population dynamics, especially as they relate to their abilities to manage puma populations through regulated sport-hunting. Secondarily, they expressed interest in puma-prey interactions. Staff on the Front Range placed greater emphasis on puma-human interactions. Staff in both eastern and western Colorado cited information needs regarding effects of puma harvest, puma population monitoring methods, and identifying puma habitat and landscape linkages. Management needs identified by CDOW staff and public stakeholders form the basis of Colorado’s puma research program, with multiple lines of inquiry (i.e., projects): 97 �Improve our ability to manage puma hunting with enhanced scientific bases, strategies, and tools ● Puma population characteristics (i.e., density, sex and age structure). ● Puma population dynamics and vital rates (i.e., birth rates, survival rates, emigration rates, immigration rates, population growth rates). ● Field methods and models for assessing and tracking changes in puma populations. ● Relative vulnerability of puma sex and age classes to hunter harvest. Improve our understanding of puma habitat needs and interrelationships of puma management units ● Puma habitat use, movements, and use of landscape linkages. ● Puma recruitment patterns (i.e., progeny, immigration, emigration). ● Models for identifying puma habitat and landscape linkages. Improve our understanding of the puma’s role in the ecology of other species ● Relationships of puma to mule deer, elk, and other natural prey. ● Relationships of puma to species of special concern, e.g., desert bighorn sheep. Improve our understanding of puma-human interactions and abilities to manage them ● Behavior of puma in relation to people and human facilities. ● Puma predation on domestic animals. ● Effects of translocating nuisance pumas. ● Effects of aversive conditioning on pumas. While all projects cannot be addressed concurrently, understanding their relationships to one another is expected to help individual projects maximize their benefits to other projects that will assist the CDOW to achieve its strategic goal in puma management (Fig.1). Management issues identified by managers translate into researchable objectives, requiring descriptive studies and field experiments. Our goal is to provide managers with reliable information on puma population biology and to develop useful tools for their efforts to adaptively manage puma in Colorado to maintain healthy, self-sustaining populations. The highest-priority management needs are being addressed with this intensive population study that focuses on puma population dynamics using sampled, tagged, and GPS/radio-collared puma. Those objectives include: 1. Describe and quantify puma population sex and age structure. 2. Estimate puma population vital rates, including: birth rates, age-stage-specific survival rates, emigration rates, immigration rates. 3. Estimate agent-specific mortality rates. 4. Improve the CDOW’s model-based management approaches with Colorado-specific data from objectives 1―3. Consider other useful models. Concurrently with the tasks associated with the objectives above, significant progress will be made toward a 5th objective, which will initially be subject to pilot study― develop methods that yield reliable estimates of population abundance (i.e., numbers and density) and attendant annual population growth rates, such as, direct capture-resight, and DNA genotype capture-recapture. TESTING ASSUMPTIONS AND HYPOTHESES Hypotheses associated with main objectives 1―5 of this puma population research are structured to test assumptions guiding puma management in Colorado. 1. Recreational puma hunting management in Colorado Game Management Units (GMUs) is guided by a model to estimate allowable harvest quotas to achieve one of two puma population objectives: 1) maintain puma population stability, or 2) cause puma population decline (CDOW, Draft L-DAU Plans, 2004). Basic model parameters are: puma population density, sex and age structure, and annual 98 �population growth rate. Parameter estimates are currently chosen from literature on studies in western states that are deemed to provide reliable information. Background material used in the model assumes a moderate annual rate of growth of 15% (i.e.,λ = 1.15) for the adult and subadult puma population (J. Apker, Carnivore Management Specialist, CDOW, Monte Vista). This assumption is based upon information with variable levels of uncertainty (e.g., small sample sizes, data from habitats dissimilar to Colorado). The key assumption is that the CDOW can manage puma population growth through recreational hunting: for a stable puma population hunting removes the annual increment of population growth (i.e., as estimated from estimates of population density, structure, and λ ; for a declining population, hunting removes more than the annual increment of population growth. Parameters influencing λ include population density, sex and age structure, female age-at-first-breeding, agespecific natality, sex- and age-specific survival, immigration and emigration. A descriptive study will ascertain these population parameters in an area that appears typical of puma habitat in western Colorado and will yield defensible population parameters based upon contemporary Colorado data. This study will be conducted in a 5-year reference period (i.e., absence of recreational hunting) to allow puma life history traits to interact with the main habitat factors that appear to influence puma population growth (e.g., prey availability and vulnerability, Pierce et al. 2000, Logan and Sweanor 2001). Contingent upon results in the reference period, a subsequent 5-year treatment period is planned. The treatment period will involve the use of controlled recreational hunting to manage the puma population into a decline phase. H1a: Population parameters measured during a 5-year reference period (in absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will yield an estimated annual adult plus subadult population growth rate that will match or exceed λ = 1.15, which is currently assumed in the CDOW’s model-based management. H1aA: Population parameters measured during a 5-year reference period (absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will yield an estimated annual adult plus subadult population growth rate that will be substantially lower (i.e., ≥ 50% lower, λ ≤ 1.075) than the assumed λ = 1.15. H1b: Population parameters during a 5-year treatment period (controlled puma hunting) will differ substantially from those measured during the preceding 5-year reference period (hunting closure) and will yield an estimated annual adult plus subadult population growth rate that will be approximately λ = 0.8 for at least the first 2 years of the treatment period. Hunting-caused mortality will be strongly additive, and will require removal of the annual growth increment (of adults plus subadults) plus 20% (e.g., assume λ = 1.15, so, 0.15 × 0.2 + 0.15 = 0.18; 0.18 × 100 = 18% annual harvest of adults plus subadults). H1bA: Population parameters during a 5-year treatment period (controlled puma hunting) will not differ substantially from those measured during the preceding 5-year reference period (hunting closure), and the adult plus subadult population will not decline on average as a result of hunting mortality. Hunting-caused mortality, reproduction, immigration, and emigration might be compensatory. 2. Considering limitations (i.e., methods, number of years, assumption violations) to the Coloradospecific studies on puma densities cited above (Currier et al. 1977, Anderson et al. 1992, Koloski 2002), managers assume that puma population densities in Colorado are within the range of those quantified in more intensively studied populations in Wyoming (Logan et al. 1986), Idaho (Seidensticker et al. 1973, Alberta (Ross and Jalkotzy 1992, and New Mexico (Logan and Sweanor 2001). The CDOW assumes density ranges of 2.0―4.6 puma/100 km2 to extrapolate to Data Analysis Units to guide the model-based quota-setting process. Likewise, managers assume that the population sex and age structure is similar to puma populations described in the intensive studies. Using capture, 99 �mark, re-capture techniques developed and refined during the study to estimate the puma population, the following will be tested: H2a: Puma densities during the 5-year reference period (absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will vary within the range of 2.0―4.6 puma/100 km2 and will exhibit a similar sex and age structure to puma populations in Wyoming, Idaho, Alberta, and New Mexico. 3. The increase and decline phases of the puma population make it possible to test hypotheses related to shifts in the age structure of the population which have been linked to harvest intensity in Wyoming and Utah. H2b: The puma population on the Uncompahgre Plateau study area will exhibit a young age structure after hunting prohibition at the beginning of the reference period. During the 5 years of hunting prohibition, greater survival of independent puma will cause an older age structure in harvest-age puma (i.e., adults and subadults) as suggested by the work of Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. H2c: As hunting is re-instated in the treatment period, the age structure of harvested puma and the harvest-age puma in the population will vary as observed by Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. Desired outcomes and management applications of this research include: 1. Quantification of variations in puma population density, sex and age structure, growth rates, vital rates, and an understanding of factors affecting them will aid adaptive puma management by yielding population parameters useful for estimating puma population abundance, evaluation of management alternatives, and effects of management prescriptions. 2. Testing assumptions about puma populations, currently used by CDOW managers, will help those managers to biologically support and adapt puma management based on Colorado-specific estimated puma population characteristics, parameters, and dynamics. 3. Methods for estimating puma abundance (capture-mark-recapture) of known reliability will allow managers to “ground truth” modeled populations and estimate effects of management prescriptions designed to achieve specified puma population objectives in targeted areas of Colorado. Ascertaining puma numbers and densities during the project will require development of reliable monitoring techniques based on capture-mark-recapture methods and models. Potential methods include direct and DNA genotype capture-recapture. Study plans to develop and test feasible field and analytical methods will be developed in the future after we have learned the logistics of performing those methods, after we have preliminary data on puma demographics and movements which will inform suitable sampling designs, and when we have adequate funding. 4. This information will be disseminated to citizen stakeholders interested in pumas in Colorado, and thus contribute to informed public participation in puma management. STUDY AREA The study area for the puma population research is on the Uncompahgre Plateau (in Mesa, Montrose, Ouray, and San Miguel Counties, Fig. 2). The study area includes about 2,253 km2 (870 mi.2) of the southern halves of Game Management Units (GMUs) 61 and 62, and about 155 km2 (60 mi.2) of the northern edge of GMU 70 (between state highway 145 and San Miguel River). The area is bounded by state highway 348 at Delta, 25 Mesa road and Forest Service road FS503 to Nucla, state highway 97 to state highway 141 to state highway 145 to Placerville, state highway 62 to Ridgeway, U.S. highway 550 to Montrose, and U.S. highway 50 to Delta. 100 �The study area seems typical of puma habitat in Colorado that has vegetation cover that varies from the pinon-juniper covered foothills starting from about 1,700 m elevation to the spruce-fir and aspen forests growing to the highest elevations of about 3,000 m. Mule deer (Odocoileus hemionus) and elk (Cervus elaphus) are the most abundant wild ungulates available for puma prey. There are cattle and domestic sheep raised on summer ranges on the study area. Year-round human residents live along the eastern and western fringe of the area, and there is a growing summer residential presence especially on the southern end of the plateau. A highly developed road system makes the study area well accessible for puma research efforts. A detailed description of the Uncompahgre Plateau is in Pojar and Bowden (2004). METHODS Reference and Experimental Treatment Periods This research is structured in two 5-year periods: a reference period (years 1―5) and a treatment period (years 6―10). The reference period is expected to cause a population increase phase. The treatment period will be managed to cause a population decline phase. In both phases, puma population structure, and vital rates will be quantified, and some management assumptions and hypotheses regarding population dynamics will be tested. Contingent upon results of pilot studies, we will also estimate puma numbers, population growth rates, evaluate enumeration methods, and test other hypotheses (Logan 2004). The reference period, without recreational puma hunting as a major limiting factor, is consistent with the natural history of the current puma species in North America which evolved life history traits during the past 10,000―12,000 years (Culver et al. 2000) that enable puma to survive and reproduce (Logan and Sweanor 2001). In contrast, puma hunting, with its modern intensity and ingenuity, might have influenced puma evolution in western North America for the past 100 years. Hence, the reference period, years 1―5, will provide conditions where individual puma in this population (of estimated sex and age structure) express life history traits interacting with the environment without recreational hunting as a limiting factor. Theoretically, the main limiting factors will be catchable prey abundance (Pierce et al. 2000, Logan and Sweanor 2001). This should allow researchers to understand basic system dynamics before the treatment (i.e., controlled recreational hunting). In the reference period, all puma in the study area will be protected, except for individual puma that might be involved in depredation on livestock or human safety incidents. In addition, all radio-collared and ear-tagged puma that range in a buffer zone, that includes the northern halves of GMUs 61 and 62, will be protected from recreational hunting. The reference period will allow researchers to quantify baseline demographic data on the puma population to estimate parameters for the CDOW’s model-based approach to puma management. Moreover, it will allow researchers to develop and test puma enumeration methods when population growth is known to be in one direction― increasing. Without the hunting closure, pilot data for enumeration methods could be confounded by not knowing if the population was increasing, declining, or stable. The reference period will also facilitate other operational needs (because hunters will not be killing the animals) including the marking of a large proportion of the puma population for capture-markrecapture estimates, and the gathering of movement data from GPS-collared puma to help formalize exact sampling designs for enumeration methods. During the treatment period, years 6―10, experimentally structured recreational puma hunting will occur on the same study area with the intent of causing a decline phase in the puma population by using management prescriptions structured from information learned during previous years. Using recreational hunting for the treatment is consistent with the CDOW’s objectives of manipulating natural tendencies of puma populations, particularly survival, to maintain either population stability or population suppression (CDOW, Draft L-DAU Plans, 2004). Theoretically, puma survival will be influenced mainly by recreational hunting, which will be quantified by agent-specific mortality rates of radio-collared puma. 101 �The portion of adults and subadults in the population will be reduced by approximately 20% in year 6 and 20% more in year 7. The 20% change was identified by Division managers that requested enumeration tools that might detect 20% changes in puma populations. For managers, detecting the magnitude of puma population decline phases is probably more important that detecting the magnitude of population increase phases. This will also allow quantification of puma population characteristics and vital rates and initial tests of enumeration methods during a decline phase. Additional reductions may be made to test enumeration methods and other hypotheses that may be related to effects of hunting (i.e.,: relative vulnerability of puma sex and age classes to hunting, variations in puma population structure due to hunting) and puma-prey interactions (i.e., lines of research identified in the Colorado Research Program, Fig. 1). Those decisions can be made later in project development and as late as years 8―10. The killing of tagged and collared puma during the treatment period will not hamper operational needs (as it would during the start-up years), because by the beginning of this period, a large majority of independent puma in the population will be marked, and sampling schemes will be formalized. Puma on the study area that may be involved in depredation of livestock or human safety incidences may be lethally controlled. Researchers that find that GPS-collared puma have killed domestic livestock will record such incidents to facilitate reimbursement to the property owner for loss of the animal(s). In addition, researchers will notify the Area Manager of the CDOW of Wildlife if they perceive that an individual puma may be a threat to public safety. Field Methods Puma Capture: Realizing that puma live at low densities and capturing puma is difficult, as a starting point, our logistical aim will be to have a minimum of 6 puma in each of 6 categories (36 total) radio-tagged in any year of the study if those or greater numbers are present. The 6 categories are: adult female, adult male, subadult female, subadult male, female cub, male cub. Our aim is to provide more quantitative and precise estimates of puma demographics than were achieved in earlier Colorado puma studies. This relatively large number of puma might represent the large majority of the puma population on the study area, and will provide the basic data for age- and sex-specific reproductive rates, survival rates, agent-specific mortality rates, emigration rates, and movement data pertinent to sampling designs for various projects. Assuming that the puma population density on the study area is relatively low at the beginning of this study― about 1 adult/100 km2 and the sex ratio is equal (Anderson et al. 1992, Logan and Sweanor 2001:167), then there might be 22 adults, 11 males and 11 females. Also assuming that the total population contains 10% subadults and 34% cubs (Logan and Sweanor 2001), then there might be 4 subadults and 13 cubs with equal sex ratios in a total population of 39 puma. If we achieve our logistical aim in the first 1―2 years (recognizing that the population might grow), then we should be able to quantify population characteristics and vital rates for a majority of the puma population in those years and build upon the tagged number in each subsequent year. Thus, our inferences will pertain to the large majority of the puma population, if not the population on the study area, instead of a relatively small sample of it. We anticipate it may take 2 years to mark the large majority of puma in the population. In addition, the study area is large and will require some time to learn to access it efficiently. Puma capture and handling procedures have been approved by the CDOW Animal Care and Use Committee (file #08-2004). All captured puma will be examined thoroughly to ascertain sex and describe physical condition and diagnostic markings. Age of adult puma will be estimated initially by the gum-line recession method (Laundre et al. 2000) and dental characteristics of known-age puma (Logan and Sweanor, unpubl. data). Ages of subadult and cub puma will be estimated initially based on dental and physical characteristics of known-age puma (Logan and Sweanor unpubl. data). Body measurements 102 �recorded for each puma will include at a minimum: mass, pinna length, hind foot length, plantar pad dimensions. Tissue collections will include: skin biopsy (from the pinna receiving the 6 mm biopsy punch for the ear-tags) and blood (30 ml from the saphenous or cephalic veins) for genotyping individuals, parentage and relatedness analyses; disease screening; hair (from various body regions) and fecal DNA for genotyping tests of field gathered samples. Universal Transverse Mercator Grid Coordinates on each captured puma will be fixed via Global Positioning System (GPS, North American Datum 27). Puma will be captured year-round using 4 methods: trained dogs, cage traps, foot-hold snares, and by hand (for small cubs). Capture efforts with dogs will be conducted mainly during the winter when snow facilitates thorough searches for puma tracks and the ability of dogs to follow puma scent. The study area will be searched systematically multiple times per year by four-wheel-drive trucks, all-terrain vehicles, snow-mobiles, walking, and possibly horse- or mule-back. When puma tracks ≤ 1 day old are detected, trained dogs will be released to pursue puma to capture. Puma usually climb trees to take refuge from the dogs. Adult and subadult puma captured for the first time or requiring a change in telemetry collar will be immobilized with Telazol (tiletamine hydrochloride/zolazepam hydrochloride) dosed at 5 mg/kg estimated body mass (Lisa Wolfe, DVM, CDOW, attending veterinarian, pers. comm.). Immobilizing agent will be delivered into the caudal thigh muscles via a Pneu-Dart® shot from a CO2-powered pistol. Immediately, a 3m-by-3m square nylon net will be deployed beneath the puma to catch it in case it falls from the tree. A researcher will climb the tree, fix a Y-rope to two legs of the puma and lower the cat to the ground with an attached climbing rope. Once the puma is on the ground, its head will be covered, its legs tethered, and vital signs monitored (Logan et al. 1986). (Normal signs: pulse ~70―80 bpm, respiration ~20 bpm, capillary refill time ≤ 2 sec., rectal temperature ~101oF average, range = 95―104oF) (Kreeger 1996). A cage trap will be used to capture adults, subadults, and large cubs when puma can be lured into the trap using road-killed or puma-killed ungulates (Sweanor et al. 2005). Efficiency of the trap might be enhanced by using an automated digital call box that emits puma vocalizations (Wildlife Technologies, Manchester, NH). A cage trap will be set only if a target puma scavenges on the lure (i.e., an unmarked puma, or a puma requiring a collar change). Researchers will continuously monitor the set cage trap from about 1 km distance by using VHF beacons on the cage and door. This allows researchers to be at the cage to handle captured puma within 30 minutes. Puma will be immobilized with Telazol injected into the caudal thigh muscles with a pole syringe. Immobilized puma will be restrained and monitored as described above. If non-target animals are caught in the cage trap, we will open the door and allow the animal to leave the trap. Foot-hold snares will be used to capture adults, subadults, and large cubs only when safe snare sites at puma kills can be located as described by Logan et al. (1999). Snares set at puma kills will be monitored continuously with VHF beacons on the snares from about 1 km distance. We will not set snares at sites where tracks indicate that other mammals (e.g., deer, elk, bear, bighorn sheep, livestock) are also active. Puma will be immobilized with Telazol injected into the caudal thigh with a pole syringe. Vital signs will be monitored during the handling procedures. Efficiency of snares might also be enhanced with the use of an automated call box with puma or prey vocalizations. Small cubs (≤ 10 weeks old) will be captured using our hands (covered with clean leather gloves) or with a capture pole. Cubs will be restrained inside new burlap bags during the handling process and will not be administered immobilizing drugs. Cubs at nurseries will be approached when mothers are away from nurseries (as determined by radio-telemetry). Cubs captured at nurseries will be removed from the nursery a distance of ~100 m to minimize disturbance and human scent at nurseries. Immediately after handling processes are complete, cubs will be returned to the exact nurseries where they were found (Logan and Sweanor 2001). 103 �Marking, Global Positioning System- and Radio-telemetry: Puma do not possess easily identifiable natural marking, such as tigers (see Karanth and Nichols 1998, 2002), therefore, the capture, marking, and GPS- or VHF- collaring of individual puma is essential to a number of project objectives, including estimating vital rates and gathering movement data on puma to formalize designs for developing and testing enumeration methods. Adult, subadult, and cub puma will be marked 3 ways: GPS/VHF- or VHF-collar, ear-tag, and tattoo. The identification number tattooed in the pinna is permanent and cannot be lost unless the pinna is severed. A colored (bright yellow or orange), numbered rectangular (5 cm x 1.5 cm) ear-tag (Allflex USA, Inc., DFW Airport, TX) will be inserted into each pinna to facilitate individual identification during direct recaptures. Cubs ≤ 10 weeks old will be eartagged in only one pinna. Locations of GPS- and VHF-collared puma will be fixed about once per week from light fixedwing aircraft (e.g., Cessna 182) fitted with radio signal receiving equipment (Logan and Sweanor 2001). This monitoring will enable researchers to find GPS-collared puma to acquire remote GPS location reports from the ground, monitor the status (i.e., live or dead) of individual puma, and to recover carcasses for necropsy. It will also provide simultaneous location data on mothers and cubs. GPS- and VHF-collared puma will be located from the ground opportunistically using hand-held yagi antenna. At least 3 bearings on peak aural signals will be mapped to fix locations and estimate location error around locations (Logan and Sweanor 2001). Aerial and ground locations will be plotted on 7.5 minute USGS maps (NAD 27) and UTMs along with location attributes will be recorded on standard forms. GPS locations will be mapped using ArcGIS software. Adult and subadult female pumas will be fitted with GPS collars (approximately 400 g each, Lotek Wireless, Canada). Initially, GPS-collars will be programmed to fix and store puma locations at 4 times per day to sample daytime, nighttime, and crepuscular locations (i.e., 0:00, 06:00, 12:00, 19:00). GPS locations for puma will provide precise, quantitative data on puma movements mainly to provide data to formalize study designs, to test assumptions for capture-mark-recapture methods for this project, and to assess the relevance of puma DAU boundaries. The GPS-collars also will provide basic information on puma movements and locations to design other pilot studies in this program on vulnerability of puma to sport-harvest, habitat use, and predation frequency on mule deer and elk. Subadult male pumas will be fitted initially with conventional VHF collars (Lotek, LMRT-3, ~400 g each) with expansion joints fastened to the collars, which allows the collar to expand to the average adult male neck circumference (~46 cm). If subadult male puma reach adulthood on the study area, we will recapture them and fit them with GPS collars. VHF radio transmitters on GPS collars will enable researchers to find those pumas on the ground in real time to acquire remote GPS data reports, facilitate recaptures for re-collaring, and to check on their reproductive and physical status. VHF transmitters on GPS- and VHF-collars will have a mortality mode set to alert researchers when puma have been immobile for at least 3 hours so that dead puma can be found to quantify survival rates and agent-specific mortality rates by gender and age. We will attempt to collar all cubs in observed litters with small VHF transmitter mounted on an expandable collar (~100g, MOD 210, Telonics, Inc., Mesa, Arizona) when cubs weigh 2.3―11 kg (5―25 lb). Cubs with mass ≥ 11 kg can still wear these small expandable collars until they are about 12 months old. Cubs approaching the age of independence (~11―14 mo. old) may be fit with Lotek LMRT-3 VHF collars (~400 g) with expansion links. Cubs will be recaptured to replace collars as necessary. Monitoring radioed cubs allow quantification of survival rates and agent-specific mortality rates (Logan and Sweanor 2001). 104 �Capture-Mark-Recapture: Capture-mark-recapture methods will be evaluated initially as a pilot study. Capturing and marking puma is time consuming, and would lengthen the time to thoroughly search the study area for capturing and marking puma during capture-recapture occasions needed for population estimation. Therefore, we will capture and mark pumas prior to performing capture-recapture or re-sight occasions using using methods such as houndsmen teams or trail cameras. In addition, by marking puma before capture-recapture occasions begin, we will have opportunities to capture female puma at different stages of their reproductive status, and thus reduce the chance that mothers in a stage with suckling cubs and small activity areas are not detected and marked on the study area. After cubs are weaned, the mothers’ activity area expands (Logan and Sweanor 2001). The probability of females having suckling cubs in winter is naturally small; that season exhibits the lowest rate of births (Logan and Sweanor 2001). Capture-recapture occasions to estimate the population of independent puma may not begin until the end of the second winter or the third winter when we have a large majority of the puma population sampled and marked. Occasions performed at that time will be viewed as a pilot study allowing us to examine the logistics of the field methods, the extent to which model assumptions are met, performance of field methods (e.g., detection differences by sex or life stage as revealed by GPS data on collared puma), and precision of capture-recapture models used to estimate the puma population. Analytical Methods Population Characteristics: Population characteristics each year will be tabulated with the number of individuals in each sex and age category. Age categories, as mentioned, include: adult (puma ≥ 24 months old, or younger breeders), subadults (young puma independent of mothers, < 24 months old that do not breed), cubs (young dependent on mothers, also known as kittens) (Logan and Sweanor 2001). When data allow, age categories may be further partitioned into months (for cubs and subadults) or years (for adults). Reproductive Rates: Reproductive rates will be estimated for GPS- and VHF-collared female puma directly (Logan and Sweanor 2001). Genetic paternity analysis will be used to ascertain paternity for adult male puma (Murphy et al. 1998). Methods will be tested in Dr. M. Douglas’s Laboratory (Colorado State University, Department of Fishery and Wildlife Biology). Survival and Agent-specific Mortality Rates: Radio-collared puma will provide known fate data which can be used to estimate survival rates for each age stage using the binomial survival model (Williams et al. 2001:343-344) or analyzed in program MARK (White and Burnham 1999, Cooch and White 2004). Agent-specific mortality rates can be analyzed using proportions and Trent and Rongstad procedures (Micromort software, Heisey and Fuller 1985). Cub survival curves for each gender will be plotted with survival rate on age in months (Logan and Sweanor 2001:119). Population Estimates: Capture-recapture models will be evaluated initially as a pilot study to estimate the parameters of primary interest― absolute numbers of independent puma (i.e., number of adult and subadult puma present in the survey area) and puma density (i.e., number of independent puma/100 km2) each winter― December through March― when snow facilitates detection and capture of puma, provided that we meet model assumptions. The December―March period also corresponds with Colorado’s puma hunting season. The population of interest is independent puma (i.e., adults and subadults) because those are the puma that can be legally killed by recreational hunters. Furthermore, adults comprise the breeding segment of the population and subadults are non-breeders that are potential recruits into the adult population in ≤ 1 year. Thus, the sampling unit is the individual independent puma (~≥ 1 yr. old). General assumptions for closed capture-recapture models are: (1) the population is closed; (2) animals do not lose their marks during the interval; (3) all marks are correctly noted and recorded at each trapping occasion; (4) each animal has a constant and equal probability of capture on each capture 105 �occasion. Open population models allow the assumption of closure to be relaxed (Otis et al. 1978, White et al. 1982, Pollock et al. 1990). The robust design is a combination of closed and open models; thus, assumptions are a combination of the assumptions for closed and open population methods (Kendall 2001). To analyze capture-recapture data, closed, open, and the robust design models are available in program MARK. Akaike’s Information Criterion will be used to select the most parsimonious models based on AICc score ranks and the difference in AIC (∆AIC) between models (Burnham and Anderson 1998). MARK results also include estimates of abundance. Because the precision of estimates for small populations is sensitive to the probability of capture (White et al. 1982, Pollock et al. 1990), our operational goal will be to achieve capture probabilities of at least 0.5 for each animal per capture occasion. Capture simulations using MARK software (Cooch and White 2004) indicate that greater capture probabilities and more capture occasions yield more precise estimates. The capture probability for the simplest closed model [M(o)], which assumes that every member of the population has the same probability of capture (p) for each sampling period, suggest that for a population of 30 animals (i.e., adults plus subadult puma, which might be present by the end of year 2, see Puma Capture above) p must equal 0.5 for 3 capture occasions to attain a coefficient of variation (V) of 0.1. If 6 capture occasions are used, then a p of 0.3 might yield a V of 0.09. In addition, behavior, movements, survival and mortality of GPS- and VHF-collared puma will allow direct biological examinations of assumptions of geographic and demographic closure (White et al. 1982) and variation in capture probability of individual puma and puma classes (i.e., adult females, adult males, subadult females, subadult males). If capture probabilities vary by puma class, we will examine if data stratification is necessary or possible (depending upon sample size). For example, we might expect the larger home ranges of male puma to expose them to more search routes, thus, this may increase their probability of capture. If the assumption of demographic closure cannot be satisfied, then open population models and the robust design would be more appropriate (Pollock et al. 1990, Williams et al. 2001). Collared puma will allow us to determine the number of marked puma present in the search area each capture-recapture occasion. Furthermore, GPS locations (4 fixes/day) on individual puma will provide data on the probability that puma may temporarily move out of and back into the survey area between capture occasions. Unmarked puma that are subsequently GPS-collared should provide such information, too. ArcView geographic information system software will be used to map and analyze puma locations, movements, and home ranges. It will also be used to map and quantify attributes of the study area and sampling frames. Rate of Population Increase: Finite rates of increase (λ = Nt+1/Nt) between consecutive years and average annual rates of increase (r) for 3- to 5-year periods and levels of precision will be calculated (Caughley 1978, Van Ballenberghe 1983) and plotted. Functional Relationships: Graphical methods will be used to examine functional relationships between puma density and vital rates, relationships between puma density estimated with direct capturerecapture methods (i.e., houndsmen teams) and possibly later (depending upon funding) by using estimates from DNA genotype or other mark- recapture methods. Linear regression procedures and coefficients of determination can be used to assess these functional relationships if data for the response variable are normally distributed and the variance is the same at each level. If the relationship is not linear, data is non-normal, and variances are unequal, we will consider appropriate transformations of the data for regression procedures (Ott 1993). Non-parametric correlation methods, such as Spearman’s rank 106 �correlation coefficient, can also be used to test for monotonic relationships between puma abundance and other parameters of interest (Conover 1999). Statistical analyses will be performed using SYSTAT and SAS software. The risk of committing a type I error (i.e., rejecting a null hypothesis that is actually true) will be controlled at alpha = 0.10 because we will normally have small population sizes (typical of studies of large obligate carnivores). The higher alpha level will increase the probability of detecting a change and reduce the risk of a type II error (i.e., failing to reject a null hypothesis that is false). For managers, the risk of a type II error is probably more important. RESULTS AND DISCUSSION Segment Objective 1 Field research to quantify puma population structure, vital rates, and causes of mortality for this report extended from August 2005 to July 2006. Our searches to detect puma presence covered the entire study area, but, we allocated most of our effort in areas where we consistently found tracks that we thought were of unmarked pumas. Less effort was allocated to the northeast and southwest areas where we found little or no evidence of pumas. We made 36 puma captures during the period (10 adult females [1 adult female captured twice], 4 adult males, 1 subadult female, 2 subadult males, 14 cubs, and 3 adult or subadult males [1 captured twice] but not handled). As our main method to capture, sample, and mark adult and subadult pumas, we used trained dogs from November 21, 2005 to May 26, 2006. Those efforts resulted in 82 search days, 149 puma tracks detected, 43 pursuits, and 14 puma captures (Table 1). Puma capture efforts with dogs in this period was similar to our efforts in the last (first) report period (Table 2). Only the number of pumas captured for the first time is lower in this period (7 vs. 11). These included 2 males (1 of them captured twice) that could not be handled for safety reasons (Table 3). It is possible that we captured 1 or both of those male pumas in subsequent capture efforts. Moreover, we substantially increased our puma capture efforts by using ungulate carcasses and cage traps from August 2005 to June 2006. We used 77 road-killed mule deer, 3 road-killed elk, 3 puma-killed mule deer, and 1 puma killed elk at 23 sites to capture pumas 11 times (Tables 4, 5, 6). Pumas scavenged 16 of 80 (20%) of the roadkilled ungulate carcasses we used for bait. A total of 11 pumas were captured, sampled, and marked for the first time by using dogs and cage traps, (Table 5), including 1 cub caught with its mother in a cage trap (Table 7). Eleven recaptures of 10 marked pumas were made with the use of dogs and cage traps; GPS/VHF collars were replaced as needed (Table 6). We captured, sampled, and marked 11 other cubs in 4 litters that were captured by hand at nurseries (Table 7). Anderson et al. (1992) studied pumas on the east slope of the Uncompahgre Plateau (i.e., GMU 62) during 1981 to 1988. Sport-hunting was banned during that study as it is in this current study. Although our current puma research on the Uncompahgre Plateau has been underway for only about 21 months (compared to 7 years of Anderson et al. 1992), there might be some useful preliminary comparisons between the 2 efforts that we can begin to make in this annual report. As our current effort results in larger samples and progresses in time through the Reference and Treatment periods, similarities and differences in results of the 2 research efforts, now separated by more than 15 years, should become robust, and illuminate new knowledge for pumas in Colorado. In the first 2 winters of puma capture efforts with dogs (1981-82 and 1983), Anderson et al. (1992:33) attempted to capture pumas in 32 and 59 days, respectively, compared to our efforts of 78 and 82 days (2004-05 and 2005-06). In the first winter, they captured 3 female pumas for the first time with an effort of 10.6 days per capture, compared to our 11 pumas (5 males, 6 females) captured for the first time, and an effort of 7.1 days per capture. In the second winter, they captured 7 pumas (4 males, 3 females) for the first time with an effort of 8.4 days per capture, compared to our 7 pumas (5 males, 2 females) captured for the first time with an effort of 11.7 days per capture. In the 7 winters of the 107 �Anderson et al. (1992) study, the average effort was 91.1 days per winter (range = 32 to 136) resulting in average capture effort of 13.9 days per capture. Other capture efforts and results between the 2 studies are not comparable, because Anderson et al. (1992) did not attempt to capture pumas using cage traps or at nurseries like we are (e.g., in about the first 25 months, Anderson et al. captured 11 pumas; we captured 37 pumas in about 20 months). Puma mass recorded by Anderson et al. (1992:86) for puma having an estimated age ≥ 24 months, averaged 61.6 kg for 8 males, (SD = 5.7, range = 51.8 to 70.8) and 44.5 kg for 14 females (SD = 3.6, range = 38.5 to 49.9). So far in the current study, mass for pumas ≥ 24 months old averaged 59.3 kg for 7 males (SD = 9.3, range 40 to 68 kg) and 39.7 kg for 8 females (SD = 4.8, range = 32 to 46). Sexual dimorphism has been described for puma throughout the species range (Young and Goldman 1946) and has been explained as the result of sexual selection (Logan and Sweanor 2001:109). Segment Objective 2 We observed 12 puma cubs produced by 5 females (Table 7). Eleven of the cubs were examined at nurseries when the cubs were 29 to 37 days old; the sexes were 4 males and 7 females. A twelfth cub was caught in a cage trap when he was about 183 days (~6 mo.) old. No evidence of siblings was found during that event. The 5 litters were born in May (1), June (1), August (1), and September (2). Puma F3 has produced 2 litters; 1 in August 2005 and 1 in September 2006; for a birth interval of 13 months. Puma M6 is a candidate sire of F3’s September 2006 litter; he and F3 consorted during June 22―24, 2005 (based on their joint GPS location data). From those consorting dates to the estimated birth date, the estimated gestation period for F3’s litter was 93―95 days. Anderson et al. (1992:47) reported of “17 postnatal litters about 10-240 days in estimated age from 12 individual females, the mean (±SD) and extremes of litter sizes were 2.41 ± 0.8, 1-4)”. “Because most postnatal young were not handled, their sex ratio is unknown” (Anderson et al (1992:48). So far in our current research, for 7 postnatal litters about 26 to 42 days old from 7 individual females, the mean (±SD) and extremes of litter sizes were 2.57 ± 0.79, 2 to 4. Sexes of the 18 cubs we examined in 7 litters aged about 26 to 42 days old were 6 males and 12 females. Anderson et al. (1992:47-48) found that of 10 puma birth dates 7 were during July, August, and September, 2 in October, and 1 in December, with most breeding occurring April through June. So far, the monthly distribution of puma births we have observed in the current study is: May (3), June (2), August (2), September (2). Considering an average 92-day gestation period (Anderson 1983:33, Logan and Sweanor 2001), breeding of pumas that produced these litters occurred from February through June. Anderson’s observation of two 12-month birth intervals for one female (Anderson et al. 1992:48) compares with our sole observation of a 13-month birth interval for F3 (above). Segment Objective 3 & 4 From December 2, 2004 (start of our research) to June 30, 2006, we monitored 7 adult male and 10 adult female pumas to quantify survival and agent-specific mortality rates (Table 8). One adult male is known to have died. M4 was about 37 to 45 months old when he was killed by an unidentified male puma along the southeast boundary of the study area. We lost contact with 2 adult males; 1 due to GPS/VHF collar failure (M6). Evidence in the field suggests that M6 might still be alive. The other male (M31) was classified as an adult at first capture because his estimated age was 25 months. However, he might still be in the latter part of the subadult stage and could have moved away from the study area. Our radiotelemetry flights beyond the boundaries of the study area have yet to locate him. All adult female pumas have survived. Adult pumas with which we have lost contact might be recaptured on the study area as our research efforts continue. 108 �Twenty puma cubs (8 males, 12 females) have been monitored by radiotelemetry (Table 10). Two males (M5, M11) are known to have survived to the subadult stage. Two cubs (F13, M22) were killed and eaten by other pumas. F35 died 1 week after we marked her probably as a result of starvation caused when the radiocollar transmitter box got caught in her mouth. We lost contact with 9 cubs (5 females, 4 males) because they shed their expandable radiocollars. Of those 9, three females (F10, F12, F14) subsequently disappeared from the family groups (i.e., we were unable to find tracks of them with other family members) and are believed to have died. As this study proceeds, some cubs with which we have lost contact will probably be re-captured or re-observed, and thus, provide more complete survival information. Anderson et al. (1992:50) reported on the fates of 21 radio-collared pumas (11 < 24 months old, 10 ≥ 24 months old) from a total of 49 in the previous study where pumas were not hunted. Yet, 19 of those pumas died due to human causes, attributed to: legal kill outside the study area (7), capture-related (6), predator management (3), illegal kill (2), and suspected predacide (1). Other causes of mortality included, intraspecies strife (1) and disease (1). Actual age-stage and annual survival rates and agentspecific survival rates from our current effort will be compared with the Anderson et al. (1992) data set at a later date when we have greater samples, duration in research time, and more complete fate data (i.e., pumas currently without functional collars) to make such comparisons meaningful. Differences might be illuminated. For example, research of a puma population in New Mexico that was not hunted for 10 years indicated that the major cause of death for both sexes and all age stages of pumas was intraspecifies strife, cannibalism, and infanticide (Logan and Sweanor 2001). We have monitored the fates of 3 subadult pumas so far (Table 9). Males M5 and M11 were born on the study area, entered the subadult stage at about 13 months old, and have dispersed from their natal areas. F23 was captured as a subadult, survived to the adult stage, and has given birth to her first litter. Anderson et al. (1992) found that all 9 radio-collared male pumas dispersed from their natal areas, and 2 of 6 radio-collared females did not disperse from their natal areas (A. E. Anderson, Sep. 1993, errata for Anderson et al. 1992:61). Mean ± SD and range of dispersal distances (km) for 8 males, aged 10 to 13 months old at dispersal, were 86.2 ± 51.3, 23 to 151. For 4 females, aged 11 to 31 months old at dispersal, mean ± SD and range of dispersal distances (km) were 37.0 ± 15.3, 17 to 54 (Anderson et al. 1992:63). Although we have observed 2 male pumas disperse from natal areas, and no females disperse, our current research is too short in duration and samples too small yet to make meaningful comparisons with Anderson’s earlier effort, particularly regarding offspring dispersal rates, distances moved, and philopatry. Dispersal and philopatry have been explained as life history strategies in pumas that assist gene flow, colonization, population maintenance, and individual survival and reproductive success (Logan and Sweanor 2001). Thus, such strategies would be expected to be conserved, and thus expressed in puma populations at different times and different locations. In addition, because puma emigration and immigration (i.e., via dispersal) have been shown to be important processes in puma population dynamics (Sweanor et al. 2000), we need larger samples and longer research duration in this study to estimate those parameters. Segment Objective 5 Fourteen adult pumas (5 males, 9 females) were fit with Lotek 4400S GPS collars since field research began in December 2004. The collars are programmed to fix 4 locations per day (00:00, 06:00, 12:00, and 19:00). The number of GPS locations per individual puma ranged from 113―1,784 (Table 11). Activity areas for GPS-collared pumas during this report period were estimated (Table 12) with fixed kernel and minimum convex polygon home range estimators (ArcView 3.2 Animal Movement Extension), and mapped (Fig. 2). In addition, 1 adult female (F30), 1adult male (M32), and 1 independent male (M31, i.e., subadult or adult) were monitored with VHF radiocollars. The number of locations for those 3 pumas were not sufficient to estimate the size of activity areas (Table 13), however, their activity areas or locations are mapped on Fig. 2. 109 �Anderson et al. (1992) provided an exhaustive analysis of seasonal puma home ranges and movements using data collected from VHF-collared animals during 1982 to 1988. We have not yet conducted an exhaustive analysis of adult puma home ranges and movements with the GPS data from our current puma research efforts in the past 21 months. Instead, we provide only limited descriptive information in Table 13 and Fig. 2. Given the different types of location data and analytical methods, only broad descriptive comparisons might be made between the 2 studies at this time. Elemental similarities in home range attributes of pumas in the Anderson et al. (1992) research and our current effort, include: current home ranges of some puma overlap extensively with home ranges of puma documented by Anderson et al (1992), home ranges of male and female pumas are large, male home ranges are larger that female home ranges, male home ranges overlap multiple female home ranges, female home ranges overlap other female home ranges sometimes extensively, male home ranges overlap other male home ranges to a lesser extent than female home ranges. These characteristics are generally similar for pumas in other populations that have been studied with adequate intensity and duration (Beier and Barrett 1993, Logan and Sweanor 2001), and reflect behavioral tactics of male and female pumas that might contribute to individual survival and reproductive success (Logan and Sweanor 2001). Segment Objective 6 To investigate the potential that puma hunters might detect puma mothers away from their cubs, we started gathering data on spatial associations of puma mothers and their cubs during the puma hunting season, which extends from November through March each winter in Colorado. Female pumas are fare game in Colorado, unless they are accompanied by 1 or more cubs. Mothers that are caught away from their cubs could be legally harvested. Such incidents would result in cubs being orphaned. Orphaned cubs that ≤ 6 months old could have a survival rate (to the subadult stage) of < 0.05. Orphaned cubs 7 to 12 months old might have a survival rate (to the subadult stage) of about 0.7 (K. Logan, unpublished data). From November 9, 2005 to March 29, 2006 we located 4 to 5 radio-collared families of puma mothers and cubs from fixed-wing aircraft 79 times (Table 14).To assess whether mothers were apart or in close association with cubs, we needed to consider error in aerial locations. We recovered 7 puma radiocollars that we located from the airplane and fixed with GPS and then fixed the actual locations of collars on the ground with GPS. Range of location error was 20 to 520 m (mean = 282.86, SD = 164.75). We decided to use distances greater than the extreme high range of location error (520 m) as the metric to decide if puma mothers might be detected away from their cubs by hunters. Sixty-seven (84.8%) of observations located mothers and cubs ≤ 500 m apart, within the extreme margin of location error. Mothers were ≥ 650 m from their cubs during 12 (15.2%) of the observations (mean distance = 1,060 m, SD = 325.99, range = 650 to 1,600). Anderson et al. (1992:70-71) recorded 69 instances of simultaneous aerial locations of 7 pairs of puma mothers and dependent young. They reported that mothers and young were together in 21 (30.4%) of those instances, and they were 1 to 2.2 km apart in 48 (69.6%) of those instances. Segment Objective 7 Three proposals were developed with colleagues in the CDOW and Colorado State University to meet some of the objectives of the Uncompahgre Plateau puma population research and to enhance the state-wide puma management program. CDOW and our CSU colleagues are currently seeking funding to support the proposals. A proposal titled: A Non-invasive Method to Estimate Puma Populations based on DNA Genotype Mark-recapture, was developed in collaboration with geneticist Dr. Marlis Douglas (CSU). We propose to use the intensively studied puma population on the Uncompahgre Plateau for gathering genetic material to develop and test molecular techniques as a means of individually genotyping puma. If successful, the methods will be used in the field and laboratory to estimate the puma population on the 110 �Uncompahgre Plateau study area. As part of our current puma capture protocol, we collect puma tissues (i.e., integument, blood, feces, hair) and archive them with Dr. Douglas, who will lead the genetics research. We developed a proposal titled: Colorado Puma Habitat Models and Maps in collaboration with Dr. Kevin Crooks, Dr. Dave Theobald, and Dr. Ken Wilson (CSU) to develop puma habitat models and maps for the entire state of Colorado. Furthermore, we are collaborating with Dr. Crooks to assess if the GPS data currently available on pumas from this project can be used to develop a graduate degree program that investigates puma habitat use on the Uncompahgre Plateau. We collaborated with Dr. Sue VandeWoude (CSU) to develop a pilot study titled: Puma concolor immune health― Relationship to management paradigms and disease. Tissue samples (i.e., blood, saliva, feces) from pumas we capture are collected and shipped to her laboratory for pending analysis. Intensive effort to quantify puma use rates on ungulates by investigating puma GPS clusters was suspended in this report period, because we discovered in our work last year that such effort was time consuming and distracted some members of our research team from our principal objectives pertaining to puma population dynamics. Yet, our work last year proved the reliability of the GPS technology to allow us to gather quantitative information on ungulate prey use rates by pumas. In that effort, 7 GPS-collared adult pumas (3 males, 4 females) used 61 mule deer and 48 elk at 139 puma GPS clusters we investigated. In contrast, when Anderson et al. (1992) studied the pumas during 1981 to 1988, they found 68 mule deer and 3 elk used by pumas. These differences might reflect a greater number and distribution of elk currently on the Uncompahgre Plateau (~1,500 elk in GMU 62 vs. 9,663 elk in E20, sources Anderson et al. 1992:15, CDOW unpubl. 2004 post-hunt elk estimate, respectively), and poses new questions about the impact of puma predation on mule deer as a function of greater availability of elk. Consequently, the CDOW has provided additional support for a 6-month temporary technician to gather such data during the next year. An assessment will be made at the end of that work on whether we should expand the effort to investigate year-round puma use rates of ungulates on the Uncompahgre Plateau. We will evaluate the potential for collaborative research on puma-human relationships on the Uncompahgre Plateau with the developing CDOW puma-human research on the Colorado Front Range. To date, we have gathered location data on 6 (4 adult females, 2 adult males) GPS-collared pumas that have activity areas on the developed southeast portion of our study area, which includes: Fairway Pines, Loghill Village, and Fisher Creek subdivisions, numerous other private homes, Fairway Pines golf course and driving range, all adjacent to Ridgeway State Park (Fig. 3). This is the same area that Anderson et al. (1992:80) received 17 useable questionnaires on puma observations from residents, and also had some radio-collared puma frequenting these same developments. Linking puma-human research on the Uncompahgre Plateau and Front Range provides opportunities for increasing sample size (i.e., puma numbers, study sites) and observing variation in puma-human relationships. SUMMARY Manipulative, long-term research on puma population dynamics, effects of sport-hunting, and development and testing of puma enumeration methods began in December 2004. After 20 months of effort, 36 pumas (14 adults, 3 subadults, 19 cubs) have been captured, sampled, marked, radio-collared and released to quantify vital rates and puma population dynamics in a reference situation (i.e., without sport-hunting off-take). Data on research efforts and puma capture, fates, reproduction, and activity areas are presented. As of July 2006, 20 radio-collared puma are within the study area. Fourteen adult pumas were fit with GPS collars, yielding 113 to 1,784 locations per puma. We started investigating the potential vulnerability of puma mothers to capture by hunters while away from their cubs. Preliminary comparisons of aspects of puma biology and ecology were made between our new research effort on the Uncompahgre 111 �Plateau and that of Anderson et al. (1992) in GMU 62 during 1981 to 1988. Research efforts for year 3 will focus on increasing numbers and distribution of sampled, marked, and GPS/radio-collared pumas on the study area, especially in the northeast and southwest portions of the study area where we have been finding relatively little evidence of pumas, possibly due to low density. Efforts will resume to estimate frequency of puma use of mule deer and elk on the Uncompahgre Plateau. Puma GPS location data will be used to: design enumeration methods in the field, develop and test puma habitat models and maps, and develop research on puma-ungulate relationships on the Uncompahgre Plateau contingent upon funding and support. We will increase our efforts to obtain outside funding for other projects we have proposed on puma genetics, puma habitat use, modeling, and mapping, and puma diseases. We will consider incorporating pumas on the Uncompahgre Plateau to address questions pertaining to research on pumahuman relationships on the Colorado Front Range. All of these projects should enhance the Colorado puma research and management programs. LITERATURE CITED ANDERSON, A. E. 1983. A critical review of literature on puma (Felis concolor). Colorado Division of Wildlife Special Report No. 54. _____, D. C. BOWDEN, AND D. M. KATTNER. 1992. The puma on Uncompahgre Plateau, Colorado. Technical Publication No. 40. Colorado Division of Wildlife, Denver. ANDERSON, C. R., JR., AND F. G. LINDZEY. 2005. Experimental evaluation of population trend and harvest composition in a Wyoming cougar population. Wildlife Society Bulletin 33:179-188. BEIER, P., AND R. H. BARRETT. 1993. The cougar in the Santa Ana Mountain Range, California. Orange County Cooperative Mountain Lion Study Final Report. BURNHAM, K. P., AND D. R. ANDERSON. 1998. Model selection and inference: a practical informationtheoretic approach. Springer-Verlag, New York, New York, USA. CAUGHLEY, G. 1978. Analysis of vertebrate populations. John Wiley and Sons, New York. COLORADO DIVISION OF WILDLIFE 2002-2007 STRATEGIC PLAN. 2002. Colorado Department of Natural Resources, Division of Wildlife. Denver. CONOVER, W. J. 1999. Practical nonparametric statistics. John Wiley & Sons, Inc., New York. COOCH, E., AND G. WHITE. 2004. Program MARK- a gentle introduction, 3rd edition. Colorado State University, Fort Collins. CULVER, M., W. E. JOHNSON, J. PECON-SLATTERY, AND S. J. O’BRIEN. 2000. Genomic ancestry of the American puma (Puma concolor). The Journal of Heredity 91:186-197. CURRIER, M. J. P., AND K. R. RUSSELL. 1977. Mountain lion population and harvest near Canon City, Colorado, 1974-1977. Colorado Division of Wildlife Special Report No. 42. HEISEY, D. M., AND T. K. FULLER. 1985. Evaluation of survival and cause specific mortality rates using telemetry data. Journal of Wildlife Management 49:668-674. KENDALL, W.L. 2001. The robust design for capture-recapture studies: analysis using program MARK. Pages 357-360 in R. Field, R. J. Warren, H. Okarma, and P. R. Sievert, editors. Wildlife, land, and people: priorities for the 21st century. Proceedings of the Second International Wildlife Management Congress. The Wildlife Society, Bethesda, Maryland, USA. KOLOSKI, J. H. 2002. Mountain lion ecology and management on the Southern Ute Indian Reservation. M. S. Thesis. Department of Zoology and Physiology, University of Wyoming, Laramie. KREEGER, T. J. 1996. Handbook of wildlife chemical immobilization. Wildlife Pharmaceuticals, Inc., Fort Collins, Colorado. LAUNDRE, J. W., L. HERNANDEZ, D. STREUBEL, K. ALTENDORF, AND C. L. LOPEZ GONZALEZ. 2000. Aging mountain lions using gum-line recession. Wildlife Society Bulletin 28:963-966. LOGAN, K. A., E. T. THORNE, L. L. IRWIN, AND R. SKINNER. 1986. Immobilizing wild mountain lions (Felis concolor) with ketamine hydrochloride and xylazine hydrochloride. Journal of Wildlife Diseases. 22:97-103. 112 �_____, L. L. SWEANOR, J. F. SMITH, AND M. G. HORNOCKER. 1999. Capturing pumas with foot-hold snares. Wildlife Society Bulletin 27:201-208. _____, AND L. L. SWEANOR. 2001. Desert puma: evolutionary ecology and conservation of an enduring carnivore. Island Press, Washington, D.C. _____. 2004. Colorado puma research and development program: population characteristics and vital rates study plan. Colorado Division of Wildlife, Ft. Collins Research Center, Ft. Collins. MURPHY, K., M. CULVER, M. MENOTTI-RAYMOND, V. DAVID, M. G. HORNOCKER, AND S. J. O’BRIEN. 1998. Cougar reproductive success in the Northern Yellowstone Ecosystem. Pages 78-112 in The ecology of the cougar (Puma concolor) in the Northern Yellowstone ecosystem: interactions with prey, bears, and humans. Dissertation, University of Idaho, Moscow. OTIS, D. L., K. P. BURNHAM, G. C. WHITE, AND D. R. ANDERSON. 1978. Statistical inference from capture data on closed animal populations. Wildlife Monographs 62:1-135. OTT, R. L. 1993. An introduction to statistical methods and data analysis. Fourth edition. Wadsworth Publishing Co., Belmont, California. PIERCE, B. K., V. C. BLEICH, AND R. T. BOWYER. 2000. Social organization of mountain lions: does land a tenure system regulate population size? Ecology 81:1533-1543. POLLOCK, K. H., S. R. WINTERSTEIN, C. M. BUNCK, AND P. D. CURTIS. 1989a. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. _____, S. R. WINTERSTEIN, AND M. J. CONROY. 1989b. Estimation and analysis of survival distributions for radio tagged animals. Biometrics 45:99-109. _____, J. D. NICHOLS, C. BROWNIE, AND J. E. HINES. 1990. Statistical inference for capture-recapture experiments. Wildlife Monographs 107:1-97. POJAR, T. M., AND D. C. BOWDEN. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550-560. ROSS, P. I., AND M. G. JALKOTZY. 1992. Characteristics of a hunted population of cougars in southwestern Alberta. Journal of Wildlife Management 56:417-426. STONER, D. C. 2004. Cougar exploitation levels in Utah: implications for demographic structure, metapopulation dynamics, and population recovery. Master of Science Thesis. Utah State University. SWEANOR, L. L., K. A. LOGAN, AND M. G. HORNOCKER. 2000. Cougar dispersal patterns, metapopulation dynamics, and conservation. Conservation Biology 14:798-808. _____, L. L., K. A. LOGAN, J. W. BAUER, W. M. BOYCE, AND B. MILSAP. 2006 in review. Puma-human relationships in a California, USA state park. Biological Conservation. VAN BALLENBERGHE, V. 1983. Rate of increase of white-tailed deer on the George Reserve: a reevaluation. Journal of Wildlife Management 47:1245-1247. WATKINS, B. 2004. Mountain lion data analysis unit L-22 management plan. Colorado Division of Wildlife, Montrose. WILLIAMS, B. K., J. D. NICHOLS, AND M. J. CONROY. 2001. Combining closed and open mark-recapture models: the robust design. Pages 523-554 In Analysis and management of animal populations. Academic Press, New York. WHITE, G. C., D. R. ANDERSON, K. P. BURNHAM, AND D. L. OTIS. 1982. Capture-recapture and removal methods for sampling closed populations. Los Alamos National Laboratory Publication LA-8787NERP. Los Alamos, NM, U.S.A. WORTON, B. J. 1995. Using Monte Carlo simulation to evaluate kernel-based home range estimators. Journal of Wildlife Management 59:794-800. YOUNG, S. P. AND E. A. GOLDMAN. 1946. The puma: mysterious American cat. The Wildlife Institute, Washington, D. C. Prepared by: Kenneth A. Logan, Wildlife Researcher 113 �Table 1. Summary of puma capture efforts with dogs from November 21, 2005 to May 26, 2006, Uncompahgre Plateau, Colorado. Month November No. Search Days 4 December 18 January No. & type of puma tracks founda No. & type of pumas pursued No. & I.D. or type of pumas captured 12 tracks: 2 male, 9 female, 1 unspecified sex 16 tracks: 10 male, 4 female, 2 unspecified sex 0 pursuits 0 captures 5 pursuits: 4 males, 1 female 19 50 tracks: 15 male, 23 female, 12 cub 11 pursuits: 4 males, 4 females, 3 cubs February 19 9 pursuits: 2 males, 3 females, 4 cubs March 7 39 tracks: 11 male, 14 female, 9 cub, 5 unspecified sex 11 tracks: 2 male, 5 female, 4 cub 11 tracks: 3 male, 5 female, 3 cub 2 pumas captured 3 times: M5 recaptured, 1 male captured twice but not handledb 3 pumas captured: F23, F24, 1 male not handled 1 puma captured: F8 recaptured 7 pursuits: 1 male, 3 females, 3 cubs 9 pursuits: 3 males, 4 females, 2 cubs 2 pumas captured: M27, F8 recaptured April 7 3 puma captured: M29 & M31 captured, M15 recaptured 2 pursuits: 1 male, 1 female 2 pumas captured: F24 May 8 10 tracks: 5 male, 5 female recaptured, M1 recaptured 82 149 tracks found: 48 male, 43 pursuits: 15 males, 16 14 captures: 3 males & 2 TOTALS females, 12 cubs females captured for the 1st 65 female, 28 cub, 8 time, 2 different males unspecified sex captured 3 times but not handled, 1 male recaptured, 2 females recaptured 3 times, 1 subadult male recaptured, 1 subadult or adult male recaptured, 1 male cub recaptured a Puma hind-foot tracks with plantar pad widths >50 mm wide are assumed to be male; ≤50 mm are assumed to be female. b Pumas are not handled for a variety of safety reasons: tree to dangerous to climb for researchers, puma treed near river, creek or cliff, puma might fall from tree after drug induction. Table 2. Summary of puma capture efforts with dogs, December 2004 to May 2006, Uncompahgre Plateau, Colorado. Period Dec. 2, 2004 to May 12, 2005 Nov. 21, 2005 to May 26, 2006 Track detection effort 109/78 = 1.40 tracks/day 149/82 = 1.82 tracks/day Pursuit effort 35/78 = 0.45 pursuit/day Puma capture effort 14/78 = 0.18 capture/day 78/35 = 2.23 day/pursuit 43/82 = 0.52 pursuit/day 78/14 = 5.57 day/capture 14/82 = 0.17 capture/day 82/43 = 1.91 day/pursuit 82/14 = 5.86 day/capture 114 Effort to capture a puma for the first time 11 pumas captured for first time (minus M1, F3, & large female) 11/78 = 0.14 capture/day 78/11 = 7.09 day/capture 7 pumas captured for first time 7/82 = 0.08 capture/day 82/7 = 11.71 day/capture �Table 3. Pumas that were captured with aid of dogs, but were not handled for safety reasons, from December 2005 to January 2006, Uncompahgre Plateau, Colorado. Puma sex Age stage Capture date 12-04-05 Male subadult or adult Male Male Location Roatcap Mesa subadult or adult 12-05-05 Cushman Creek adult 01-22-06 San Miguel River Canyon Comments Puma climbed to top of huge Ponderosa pine tree. This puma might be M32. This puma was the same animal caught 12-04-05. Climbed tall spruce tree on a ledge. Puma climbed Ponderosa pine tree beside the river. This puma might be M29. Table 4. Summary of puma capture efforts with ungulate road-kill baits, puma kills, and cage traps from August 2, 2005 to June 27, 2006, Uncompahgre Plateau, Colorado.a Month August No. of Sites 4 Puma activity & capture effort results Male puma scavenged a mule deer on 8-20-05. Cage trap set 8-20 & 8-21-05; puma did not return. September 7 Male puma scavenged a mule deer on 9-16-05; subadult M5 was recaptured there and was fit with a VHF collar (he had shed the expandable collar he wore as a cub). October 10 Puma F16 captured 10-11-05 at an elk kill. Puma scavenged mule deer on 10-17-05. Cage trap set 10-18 and 10-19-05; puma did not return. November 12 Puma F16 recaptured 11-1-05 at a mule deer kill; her faulty GPS collar was replaced. Puma F16 scavenged a mule deer on 11-27-05. No attempt to recapture her. December 1 No puma activity detected. January 2 No puma activity detected. February 9 Puma F25 and cub M26 captured 2-8-06 at a mule deer kill. Puma F7 scavenged a mule deer on 2-26 & 2-17-06. No attempt to recapture her. Puma F3 scavenged a mule deer on 2-26-06. No attempt to recapture her. March 11 Male puma completely scavenged a mule deer over the weekend of 3-18 & 1906. Female puma completely scavenged a mule deer over the weekend of 3-18 & 19-06. Female puma scavenged a mule deer 3-21 to 3-23-06; puma F28 was captured there 3-23-06. Pumas F3 & cub F21 scavenged a mule deer 3-23 to 3-27-06. No attempt was made to recapture the pumas. Female puma completely scavenged a mule deer over the weekend of 3-25 & 26-06. Puma F7 was recaptured at a mule deer she scavenged on 3-30-06; her GPS collar was replaced. April 11 Puma F30 captured 4-15-06 at a mule deer kill. Male puma scavenged a mule deer on 4-20-06. Cage trap was set 4-20 & 4-2106; puma did not return. Male puma scavenged the same mule deer on 4-26-06; M32 was captured there on 4-26-06. Puma F2 and cubs F9 & M11 scavenged a mule deer on 4-2-06. Cub M11 was recaptured on 4-2-06 and fit with a VHF collar. F2 was recaptured there 4-3-06 and her GPS collar was replaced. May 0 No fresh road-killed ungulate carcasses were available. June 3 No puma activity. a We used 77 road-killed mule deer, 3 road-killed elk, 3 puma-killed mule deer, and 1 puma-killed elk at 23 different sites. Of the road-killed ungulate baits, 16 of 80 (20.0%) were scavenged by pumas. 115 �Table 5. Adult and subadult pumas captured for the first time, sampled, tagged, and released from October 2005 to April 2006, Uncompahgre Plateau, Colorado. Puma I.D. Sex Estimated Age (mo.) Mass (kg) Capture date Capture method F16 F 33 42 10-11-05 Cage trap F23 F24 F25 M27 F28 M29 F30 M31 M32 F F F M F M F M M 17 57 80 55 33 80 33 25 56 42 38 46 61 43 65 34 40 57 01-04-06 01-17-06 02-08-06 03-10-06 03-23-06 04-14-06 04-15-06 04-19-06 04-26-06 Dogs Dogs Cage trap Dogs Cage trap Dogs Cage trap Dogs Cage trap Location Ridgeway Reservoir Dam, Ridgeway State Park San Miguel River Canyon Horsefly Creek (west) Loghill Mesa Big Bucktail Creek Big Bucktail Creek Big Bucktail Creek Wildcat Canyon Craig Draw Spring Creek Table 6. Pumas recaptured with dogs and cage traps, September 2005 to May 2006, Uncompahgre Plateau, Colorado. Puma I.D. Recapture date Mass kg Estimated Age (mo.) M5 F16 M5 F8 F8 F7 M11 F2 M15 F24 M1 09-16-05 11-01-05 12-30-05 02-07-06 03-21-06 03-30-06 04-02-06 04-03-06 04-13-06 05-17-06 05-26-06 39 42 Observed Observed Observed 35 32 43 23 Observed Observed 13 34 16 32 33 69-77 10 64 9.5 61 51 Capture Method Cage trap Cage trap Dogs Dogs Dogs Cage trap Cage trap Cage trap Dogs Dogs Dogs Process Re-collared Re-collared None None None Re-collared Re-collared Re-collared Re-collared None None Table 7. Puma cubs sampled August 2005 to June 2006 on the Uncompahgre Plateau Puma Study area, Colorado. Cub I.D. Sex Estimated birth datea Estimated age at capture (days) Mass (kg) Mother Estimated age of mother at birth of this litter (mo) F17 F Sept. 22, 2005 34 2.5 F16 32 F18 F Sept. 22, 2005 34 2.0 “ “ M19 M Sept. 22, 2005 34 2.0 “ “ M20 M Sept. 22, 2005 34 2.1 “ “ F21b F Sept. 26, 2005 37 2.8 F3c 49 M Sept. 26, 2005 37 2.8 “ “ M22b M26d M Aug. 2005 183 12.0 F25 74 F33 F May 30, 2006 31 1.9 F23 21 F34 F May 30, 2006 31 1.9 “ “ F35 F May 30, 2006 31 2.2 “ “ F36 F June 9, 2006 29 1.9 F28 36 M37 M June 9, 2006 29 2.1 “ “ a Estimated age of cubs sampled at nurseries is based on the starting date for GPS location foci for mothers at nurseries. b Puma M6 is a candidate sire of cubs F21 & M22; he consorted with F3 (based on their joint GPS location data) during June 22―24, 2005. This would indicate a gestation period of 93―95 days. c F3 gave birth to a previous litter in August 2004. From that litter, offspring M5 survived to independence. Birth interval is 13 months (Aug. 2004 to Sept. 2005). d Estimated age of M26 was based on morphometric comparisons with known-age cubs (Logan and Sweanor 2001, and unpublished data, i.e., ~6 mo. ≈183 days). He was initially captured in a cage trap with his mother F25 on Feb. 8, 2006. 116 �Table 8. Summary for individual adult puma survival and mortality, December 2004 to June 2006, Uncompahgre Plateau, Colorado. Puma I.D. Monitoring span No. days Status: Alive/Lost contact/Dead; Cause of death M1 12-08-04 to 06-30-06 569 Alive. M4 01-28-05 to 12-28-05 333 Dead; killed by a male puma.a M6 02-18-05 to 02-22-06 369 Lost contact― failed GPS/VHF collar. M27 03-10-06 to 06-30-06 112 Alive. M29 04-14-06 to 06-30-06 77 Alive. M31 04-19-06 to 04-26-06 7 Lost contact.b M32 04-26-06 to 06-30-06 65 Alive. F2 01-07-05 to 06-30-06 539 Alive. F3 01-21-05 to 06-30-06 525 Alive. F7 02-24-05 to 06-30-06 491 Alive. F8 03-21-05 to 06-30-06 466 Alive. F16 10-11-05 to 06-30-06 262 Alive. F23 02-05-06 to 06-30-06 146 Alive. F24 01-17-06 to 06-30-06 164 Alive. F25 02-08-06 to 06-30-06 142 Alive. F28 03-23-06 to 06-30-06 99 Alive. F30 04-15-06 to 06-30-06 76 Alive. a Puma M4 died at the estimated age of 37―45 months old. b Puma M31 estimated age at capture was 25 months, at the lower margin of puberty. But he might have been a dispersing subadult, instead of an adult. He may have moved away from the study area. No VHF signals have been received of M31 in the area surrounding the study area as of 07-29-06. Table 9. Summary of subadult puma survival and mortality, December 2004 to June 2006, Uncompahgre Plateau, Colorado. Puma Monitoring span No. Status: Alive/Survived to adult stage/ I.D. days Lost contact/Dead; Cause of death M5 09-16-05 to 06-30-06 287 Alive; dispersed from natal area. M11 06-21-06 to 06-30-06 7 Alive; dispersed from natal area. F23 01-04-06 to 02-04-06 31 Alive; survived to adult stage; gave birth to first litter at ~21 months old. 117 �Table 10. Summary for individual puma cub survival and mortality, December 2004 to July 2006, Uncompahgre Plateau, Colorado. Estimated Age at capture (days) Estimated survival span from 1st capture to fate or last monitor date Age to last monitor date alive or at death Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/Dead; Cause of death Mother I.D. M5 183 02-04-05 to 06-30-06 22 mo. F3 F9 31 F10 31 06-27-05 to 4-19-06 06-27-05 to 11-20-05― 12-29-05 329 days 207―246 days M11 31 06-27-05 to 07-11-06 14 mo. F12 42 07-01-05 to 12-08-05― 01-26-06 245―294 days F13 42 F14 26 07-01-05 to 08-28-05 07-22-05 to 02-07-06― 03-10-06 100 days 226―257 days M15 26 F17 34 F18 34 M19 34 M20 34 F21 37 M22 37 07-22-05 to 06-06-06 10-26-05 to 06-06-06 10-26-05 to 06-30-06 10-26-05 to 07-27-06 10-26-05 to 05-24-06 11-02-05 to 06-30-06 11-02-05 to 12-21-05― 12-22-05 345 days 257 days 281 days 306 days 244 days 277 days 86―87 days Survived to subadult stage by 09-16-05; independent at ~13 mo. old. Dispersed from natal area by 09-29-05 at 13 mo. old . Lost contact― shed radiocollar 04-1906―04-26-06. Lost contact― shed radiocollar 08-10-05; last tracks of F10 with mother F2 & siblings F9 & M11 observed 11-20-05. F10 disappeared by 12-30-05. Survived to subadult stage by 06-21-06, independent at 13 mo. old. Dispersed from natal area by 07-11-06 at 14 mo. old. Lost contact― shed radiocollar 07-2805―08-01-05. Tracks of F12 found in association with mother F7 on 12-08-05. F12 disappeared by 01-27-06 when she was not visually observed with F7, and her tracks were not seen in association with F7’s tracks. Dead; killed and eaten by a puma (sex unspecified). Lost contact― shed radiocollar 01-2006―01-25-06. Tracks of F14 were observed with tracks of mother F8 & sibling M15 on 02-07-06. Disappeared by 03-11-06, only tracks of F8 & M15 were found. Lost contact― shed radiocollar 06-0606―06-14-06. Lost contact― shed radiocollar 06-0606―06-14-06. Alive. M26 183 F33 31 F34 31 F35 31 F36 29 M37 29 02-08-06 to 03-21-06 06-30-06 to 07-31-06 06-30-06 to 07-31-06 06-30-06 to 07-07-06 07-08-06 to 07-28-06 07-08-06 to 07-28-06 224 days 62 days 62 days 38 days 49 days 49 days Puma I.D. a F2 F2 F2 F7 F7 F8 F8 F16 F16 Lost contact― shed radiocollar 07-2706―08-02-06. Lost contact― shed radiocollar 05-2406―05-25-06. Alive. F16 Dead; killed and eaten by male puma 12-2105―12-22-05. F3 Lost contact― shed radiocollar 03-2106―03-24-06. Alive. F25 Alive. F23 Dead; research-related fatality.a F23 Alive. F28 Alive. F28 F16 F3 F23 Cub F35 probably starved between 06-30-06 & 07-07-06 after the transmitter on the expandable collar got in its mouth. 118 �Table 11. Numbers of GPS locations for adult puma on the Uncompahgre Plateau, Colorado, December 2004 to June 2006. Puma I.D. M1 M4 M6 M27 M29 F2 F3 F7 F8 F16 F23 Sex Age stage Dates monitored a No. locations M M M M M F F F F F F Acquisition rate average, range, nb 77, 73―84, 13 70, 57―84, 10 84, 73―93, 9 77, 67―84, 3 68, 63―75, 3 75, 43―90, 18 76, 55―88, 17 67, 26―86, 17 70, 48―81, 14 76, 58―90, 10 79, 45―92, 6 adult 12-08-04 to 06-21-06 1,784 adult 01-28-05 to 12-28-05e 910 926 adult 02-18-05 to 11-23-05f adult 03-11-06 to 06-21-06 316 adult 04-14-06 to 07-27-06 287 1,664 adult 01-07-05 to 07-12-06 1,649 adult 01-21-05 to 07-26-06 1,423 adult 02-24-05 to 07-26-06 1,328 adult 03-21-05 to 07-05-06 833 adult 10-12-05 to 07-03-06 113 subadult, 01-04-06 to 02-04-06 02-05-06 to 07-17-06 511 adult 523 88, 86―93, 5 F24 F adult 01-17-06 to 06-14-06 551 78, 68―87, 5 F25 F adult 02-09-06 to 07-12-06 321 74, 61―89, 4 F28 F adult 03-24-06 to 07-07-06 a GPS collars on pumas are remotely downloaded at approximately 1-month intervals. The last date in Dates monitored includes last location from the last GPS data download for an individual puma in this report. b n = number of remote downloads. Table 12. Estimated use areas of GPS-collared pumas on the Uncompahgre Plateau, Colorado, 2005 to 2006.a Puma I.D. No. locations Time span No. months 95% Fixed kernel (km2) 50% Fixed kernel (km2) 100% Minimum convex polygon (km2) 1,129.0 318.5 542.0 504.0 288.3 183.0 194.0 139.0 215.0 74.3 226.0 111.7 115.9 114.8 M1 1,083 07-01-05 to 06-21-06 12 988.1 189.1 M4b 481 07-01-05 to 12-28-05 5 208.8 29.6 465 07-01-05 to 11-23-05 4.8 550.8 67.3 M6c M27 316 03-11-06 to 06-21-06 3.3 452.0 40.3 M29 220 04-14-06 to 06-30-06 2.5 276.1 14.0 F2 1,173 07-01-05 to 06-30-06 12 67.6 6.9 F3 1,079 07-01-05 to 06-30-06 12 84.3 11.7 F7 1,058 07-01-05 to 06-30-06 12 110.4 16.2 F8 1,043 07-01-05 to 06-30-06 12 84.1 7.7 F16 825 10-12-05 to 06-30-06 8.6 39.9 4.8 F23 566 01-04-06 to 06-30-06 5.9 109.2 13.0 F24 574 01-17-06 to 06-30-06 5.5 26.8 2.4 F25 453 02-09-06 to 06-30-06 4.7 105.5 16.1 F28 306 03-24-06 to 06-30-06 3.2 86.2 14.8 a Use areas were estimated by using the Animal Movement extension in ArcView 3.2. b Puma M4 died on 12-28-05; he was killed by a male puma. c Puma M6’s GPS collar malfunctioned on 11-23-05. His last VHF location was fixed on 02-22-06. The VHF beacon failed after that date. Table 13. VHF-radio-collared independent pumas on the Uncompahgre Plateau, Colorado, 2006. Puma I.D. F30 M31 Sex Age stage Dates monitored No. locations F M 04-15-06 to 06-28-06 04-09-06 to 04-26-06 11 2 M32 M Adult Adult or subadult Adult 04-26-06 to 06-28-06 6 119 �Table 14. Summary of puma mother and cub associations by distance (m) during fixed-wing flights, November 9, 2005 to March 29, 2006. Month No. flights No. puma familiesa Ages of cubs (mo.) No. observations with No. observations with mothers & cubs mothers & cubs ≤500 m >600 m apartb apart Nov. 3 4 2―6 10 2 Dec. 4 4 3―7 16 4 Jan. 5 4 4―8 16 4 Feb. 4 5 5―9 16 2 Mar. 2 5 6―10 9 0 Totals 18 4―5 2―10 67 12 a All puma mothers wore GPS-radiocollars. At least 1 cub in the litter wore a VHF radiocollar. b Mean = 1,060 m, SD = 325.99, range = 650―1,600. GOAL: Strategies, Information, & Tools for Managing Healthy, Self-sustaining Puma Populations in Colorado Puma Habitat Puma Population Effects of Harvest & Other Mortality Movements & Corridors Population Dynamics: Sex & Age, Vital Rates, Mortality, Population Growth Rates Vulnerability to Harvest Human Development Habitat Use Effects of Translocation Estimation Methods for Monitoring Domestic Animals Puma― Human Relationships Deer, Elk, Other Natural Prey & Species of Concern Effects of Predation Effects of Evasive Conditioning Map Prey Distribution Habitat Models Population Models Puma Prey Habitat Maps Puma―Prey Relationships Models Figure. 1. An ecologically-based conceptual model of the Colorado Puma Research Program that provides the contextual framework for this and proposed puma research in Colorado. Gray-shaded shapes identify areas of research addressed by this report for the puma management goal (at top). 120 �! ( ! ( ! ( ! ( ! ( Clifton County Boundary Highways Study Area ! ( ! ( ! ( ! ( ! ( Delta ! ( M1 ! ( ! ( M32 Montrose F8 M27 ( ! F23 ! ( ! ( F3 M31 M4 F7 F28 F24 F30 F2 M6 ( ! F16 F25 ! ( Ridgeway M29 ! ( Norwood ! ( ! ( 0 5 10 20 30 40 Kilometers ! ( Figure 2. The Uncompahgre Plateau Puma Study Area with activity areas of GPS- and VHF- radiocollared pumas depicted with 100% Minimum Convex Polygons (for ease of viewing), and 2 locations of one independent puma for which we lost contact (M31), 2005 to 2006. 121 �Figure 3. Locations of 6 GPS-collared pumas on the human-developed southeast portion of the Uncompahgre Plateau puma study area, 2005-2006, intended only to show potential for developing research on puma-human relationships on this study area and the Colorado Front Range. 122 �Colorado Division of Wildlife July 2006 – June 2007 WILDLIFE RESEARCH REPORT State of Cost Center Work Package Task No. Colorado 3430 3003 1 Federal Aid Project: N/A : Division of Wildlife : Mammals Research : Predatory Mammals Conservation : Puma Population Structure and Vital Rates on the Uncompahgre Plateau : Period covered: August 1, 2006―July 30, 2007 Author: K. A. Logan. Personnel: K. Logan, B. Bavin, B. Dunne, J. Mannas, S. Waters, K. Crane, T. Mathieson, M. Caddy, and T. Bonacquista of CDOW; S. Young, and J. McNamara of U.S.D.A. Wildlife Services; volunteers and cooperators including: private landowners, U.S. Forest Service, Bureau of Land Management, and Colorado State Parks with financial support received from The Howard G. Buffett Foundation and Safari Club International Foundation. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Research continued on puma population characteristics and dynamics on the Uncompahgre Plateau. Puma capture efforts resulted in a total of 54 puma captures (9-10 adult females [1 female captured twice], 7 adult males [1 male probably captured 3 times, and another captured twice], 1 subadult female, 0-1 other subadults, and 30 cubs [4 captured twice each]). Efforts to capture, sample, and mark pumas with the use of trained dogs extended from November 13, 2006 to May 11, 2007. This resulted in 22 puma captures, including 1 adult female, 1 adult male, 2 adult males, and 2 male cubs captured and processed for the first time. One female (adult or subadult) and 2 cubs were not handled for safety reasons. Capture efforts with ungulate carcasses and cage traps resulted in 8 puma captures, including 4 adult females, 2 adult males, 1 subadult female, and 1 female cub. Of those animals, 1 adult female, 1 adult male, the subadult female, and the female cub were captured for the first time. Capture efforts during November 2006 through May 2007 enabled us to estimate a minimum count of 24 independent pumas detected on the Uncompahgre Plateau study area during that time. The count included 16 females and 8 males. We captured, sampled, and marked 26 puma cubs produced by 10 females. Twenty-three of the cubs were examined at 8 nurseries when the cubs were 29 to 41 days old. Since the start of this study, 38 cubs from 13 litters aged 29 to 42 days old had a sex ratio of 21 males:17 females. The mean (±SD) and extremes of litter sizes were 2.84 (±0.99), 1 to 4. Eight birth intervals for 7 different females averaged 14.99 months (SD = 3.40), and ranged from 11.7 to 20.5 months. Four gestation periods averaged 92.0 days (SD = 1.68). Of 9 adult males and 12 adult females radio-monitored to quantify survival and agentspecific mortality rates, 1 male and 1 female are known to have died from natural causes. Of 6 subadult pumas monitored via radio-telemetry, none died. Thirty-nine puma cubs (20 males, 19 females) have been monitored by radiotelemetry for varying durations. Among those, 12 deaths were documented, including 7 from intra-species strife, 1 killed by coyotes, 1 killed by a vehicle, and 3 died due to research activities. 111 �Twenty adult pumas (7 males, 13 females) fit with GPS collars since field research began in December 2004 have yielded 113 to 2,759 locations per individual puma. Winter activity areas were estimated for 12 (9 female, 3 male) GPS-collared adult pumas. As an index to the vulnerability of puma mothers to sportharvest we monitored mother-cub distances from an airplane during November to March. Puma mothers were ≥520 m from their cubs during 16.3% of the observations (mean distance = 1,120 m, SD = 1,214.40, range = 616 to 4,101). These results were similar to our results the previous winter (15.2%). A collaborative effort to investigate puma use of ungulates on the Uncompahgre Plateau resumed. GPS clusters were investigated for 13 GPS-collared adult pumas (8 female, 5 male). A total of 257 clusters were investigated. Mule deer and elk were about equally important to pumas as food. Preliminary comparisons between our current puma research on the Uncompahgre Plateau (31 months duration) and results of the Anderson et al. (1992) puma research on the plateau (7 years duration 1981-1988) were made where appropriate. Proposed work includes: continuing investigations of puma use of ungulates, developing and testing methods and models to estimate puma abundance, and collaborating with colleagues to assess puma health. In addition, we will consider how research of pumas on developed areas on the Uncompahgre Plateau can contribute to the CDOW’s efforts to study puma-human interactions on the Colorado Front Range. 112 �WILDLIFE RESEARCH REPORT PUMA POPULATION STRUCTURE AND VITAL RATES ON THE UNCOMPAHGRE PLATEAU, COLORADO KENNETH A. LOGAN P. N. OBJECTIVE Quantify puma population sex and age structure; estimate puma population vital rates, including: reproduction rates of females, age-stage survival rates, and immigration and emigration rates; quantify agent-specific mortality rates― all to improve the Colorado Division of Wildlife’s (CDOW) model-based approach to managing pumas in Colorado. SEGMENT OBJECTIVES 1. 2. 3. 4. 5. Continue gathering data on puma population sex and age structure. Continue gathering data for estimates of puma reproduction rates. Continue gathering data to estimate puma sex and age-stage survival rates. Continue gathering data to estimate agent-specific mortality rates. Continue gathering data on puma movements for the development of sampling methods for markresight or recapture population estimates that might involve sampling puma DNA-genotypes, trail cameras, or direct observations. 6. Begin gathering data on spatial relationships of puma mothers to their cubs during the Colorado puma hunting season as a preliminary assessment of the vulnerability of puma mothers to sport-hunting harvest. 7. Evaluate other data sources that could come from this research that can be developed into other puma research relevant to CDOW biologists and managers. INTRODUCTION Colorado Division of Wildlife managers need reliable information on puma biology and ecology in Colorado to develop sound management strategies that address diverse public values and the CDOW objective of actively managing puma while “achieving healthy, self-sustaining populations”(CDOW 2002-2007 Strategic Plan:9). Although 4 puma research efforts have been made in Colorado since the early 1970s and puma harvest data is compiled annually, reliable information on certain aspects of puma biology and ecology, and management tools that may guide managers toward effective puma management is lacking. Mammals Research staff held scoping sessions with a number of the CDOW’s wildlife managers and biologists. In addition, we consulted with other agencies, organizations, and interested publics either directly or through other CDOW employees. In general, CDOW staff in western Colorado highlighted concern about puma population dynamics, especially as they relate to their abilities to manage puma populations through regulated sport-hunting. Secondarily, they expressed interest in puma―prey interactions. Staff on the Front Range placed greater emphasis on puma―human interactions. Staff in both eastern and western Colorado cited information needs regarding effects of puma harvest, puma population monitoring methods, and identifying puma habitat and landscape linkages. Management needs identified by CDOW staff and public stakeholders form the basis of Colorado’s puma research program, with multiple lines of inquiry (i.e., projects): 113 �Improve our ability to manage puma hunting with enhanced scientific bases, strategies, and tools ● Puma population characteristics (i.e., density, sex and age structure). ● Puma population dynamics and vital rates (i.e., birth rates, survival rates, emigration rates, immigration rates, population growth rates). ● Field methods and models for assessing and tracking changes in puma populations. ● Relative vulnerability of puma sex and age classes to hunter harvest. Improve our understanding of puma habitat needs and interrelationships of puma management units ● Puma habitat use, movements, and use of landscape linkages. ● Puma recruitment patterns (i.e., progeny, immigration, emigration). ● Models for identifying puma habitat and landscape linkages. Improve our understanding of the puma’s role in the ecology of other species ● Relationships of puma to mule deer, elk, and other natural prey. ● Relationships of puma to species of special concern, e.g., desert bighorn sheep. Improve our understanding of puma-human interactions and abilities to manage them ● Behavior of puma in relation to people and human facilities. ● Puma predation on domestic animals. ● Effects of translocating nuisance pumas. ● Effects of aversive conditioning on pumas. While all projects cannot be addressed concurrently, understanding their relationships to one another is expected to help individual projects maximize their benefits to other projects that will assist the CDOW to achieve its strategic goal in puma management (Fig.1). Management issues identified by managers translate into researchable objectives, requiring descriptive studies and field experiments. Our goal is to provide managers with reliable information on puma population biology and to develop useful tools for their efforts to adaptively manage puma in Colorado to maintain healthy, self-sustaining populations. The highest-priority management needs are being addressed with this intensive population study that focuses on puma population dynamics using sampled, tagged, and GPS/radio-collared puma. Those objectives include: 1. Describe and quantify puma population sex and age structure. 2. Estimate puma population vital rates, including: birth rates, age-stage-specific survival rates, emigration rates, immigration rates. 3. Estimate agent-specific mortality rates. 4. Improve the CDOW’s model-based management approaches with Colorado-specific data from objectives 1―3. Consider other useful models. Concurrently with the tasks associated with the objectives above, significant progress will be made toward a 5th objective, which will initially be subject to pilot study― develop methods that yield reliable estimates of population abundance (i.e., numbers and density) and attendant annual population growth rates, such as, direct capture-resight, and DNA genotype capture-recapture. TESTING ASSUMPTIONS AND HYPOTHESES Hypotheses associated with main objectives 1―5 of this puma population research are structured to test assumptions guiding puma management in Colorado. 114 �1. Recreational puma hunting management in Colorado Game Management Units (GMUs) is guided by a model to estimate allowable harvest quotas to achieve one of two puma population objectives: 1) maintain puma population stability, or 2) cause puma population decline (CDOW, Draft L-DAU Plans, 2004). Basic model parameters are: puma population density, sex and age structure, and annual population growth rate. Parameter estimates are currently chosen from literature on studies in western states that are deemed to provide reliable information. Background material used in the model assumes a moderate annual rate of growth of 15% (i.e., λ = 1.15) for the adult and subadult puma population (J. Apker, Carnivore Management Specialist, CDOW, Monte Vista). This assumption is based upon information with variable levels of uncertainty (e.g., small sample sizes, data from habitats dissimilar to Colorado). The key assumption is that the CDOW can manage puma population growth through recreational hunting: for a stable puma population hunting removes the annual increment of population growth (i.e., as estimated from estimates of population density, structure, and λ); for a declining population, hunting removes more than the annual increment of population growth. Parameters influencing λ include population density, sex and age structure, female age-at-first-breeding, agespecific natality, sex- and age-specific survival, immigration and emigration. A descriptive study will ascertain these population parameters in an area that appears typical of puma habitat in western Colorado and will yield defensible population parameters based upon contemporary Colorado data. This study will be conducted in a 5-year reference period (i.e., absence of recreational hunting) to allow puma life history traits to interact with the main habitat factors that appear to influence puma population growth (e.g., prey availability and vulnerability, Pierce et al. 2000, Logan and Sweanor 2001). Contingent upon results in the reference period, a subsequent 5-year treatment period is planned. The treatment period will involve the use of controlled recreational hunting to manage the puma population into a decline phase. H1a: Population parameters measured during a 5-year reference period (in absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will yield an estimated annual adult plus subadult population growth rate that will match or exceed λ = 1.15, which is currently assumed in the CDOW’s model-based management. H1aA: Population parameters measured during a 5-year reference period (absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will yield an estimated annual adult plus subadult population growth rate that will be substantially lower (i.e., ≥50% lower, λ ≤1.075) than the assumed λ = 1.15. H1b: Population parameters during a 5-year treatment period (controlled puma hunting) will differ substantially from those measured during the preceding 5-year reference period (hunting closure) and will yield an estimated annual adult plus subadult population growth rate that will be approximately λ = 0.8 for at least the first 2 years of the treatment period. Hunting-caused mortality will be strongly additive, and will require removal of the annual growth increment (of adults plus subadults) plus 20% (e.g., assume λ = 1.15, so, 0.15 × 0.2 + 0.15 = 0.18; 0.18 × 100 = 18% annual harvest of adults plus subadults). H1bA: Population parameters during a 5-year treatment period (controlled puma hunting) will not differ substantially from those measured during the preceding 5-year reference period (hunting closure), and the adult plus subadult population will not decline on average as a result of hunting mortality. Hunting-caused mortality, reproduction, immigration, and emigration might be compensatory. 2. Considering limitations (i.e., methods, number of years, assumption violations) to the Coloradospecific studies on puma densities cited above (Currier et al. 1977, Anderson et al. 1992, Koloski 2002), managers assume that puma population densities in Colorado are within the range of those 115 �quantified in more intensively studied populations in Wyoming (Logan et al. 1986), Idaho (Seidensticker et al. 1973, Alberta (Ross and Jalkotzy 1992, and New Mexico (Logan and Sweanor 2001). The CDOW assumes density ranges of 2.0―4.6 puma/100 km2 to extrapolate to Data Analysis Units to guide the model-based quota-setting process. Likewise, managers assume that the population sex and age structure is similar to puma populations described in the intensive studies. Using capture, mark, re-capture techniques developed and refined during the study to estimate the puma population, the following will be tested: H2a: Puma densities during the 5-year reference period (absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will vary within the range of 2.0―4.6 puma/100 km2 and will exhibit a similar sex and age structure to puma populations in Wyoming, Idaho, Alberta, and New Mexico. 3. The increase and decline phases of the puma population make it possible to test hypotheses related to shifts in the age structure of the population which have been linked to harvest intensity in Wyoming and Utah. H2b: The puma population on the Uncompahgre Plateau study area will exhibit a young age structure after hunting prohibition at the beginning of the reference period. During the 5 years of hunting prohibition, greater survival of independent puma will cause an older age structure in harvest-age puma (i.e., adults and subadults) as suggested by the work of Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. H2c: As hunting is re-instated in the treatment period, the age structure of harvested puma and the harvest-age puma in the population will vary as observed by Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. Desired outcomes and management applications of this research include: 1. Quantification of variations in puma population density, sex and age structure, growth rates, vital rates, and an understanding of factors affecting them will aid adaptive puma management by yielding population parameters useful for estimating puma population abundance, evaluation of management alternatives, and effects of management prescriptions. 2. Testing assumptions about puma populations, currently used by CDOW managers, will help those managers to biologically support and adapt puma management based on Colorado-specific estimated puma population characteristics, parameters, and dynamics. 3. Methods for estimating puma abundance (capture-mark-recapture) of known reliability will allow managers to “ground truth” modeled populations and estimate effects of management prescriptions designed to achieve specified puma population objectives in targeted areas of Colorado. Ascertaining puma numbers and densities during the project will require development of reliable monitoring techniques based on capture-mark-recapture methods and models. Potential methods include direct and DNA genotype capture-recapture. Study plans to develop and test feasible field and analytical methods will be developed in the future after we have learned the logistics of performing those methods, after we have preliminary data on puma demographics and movements which will inform suitable sampling designs, and when we have adequate funding. 4. This information will be disseminated to citizen stakeholders interested in pumas in Colorado, and thus contribute to informed public participation in puma management. STUDY AREA The study area for the puma population research is on the Uncompahgre Plateau (in Mesa, Montrose, Ouray, and San Miguel Counties, Fig. 2). The study area includes about 2,253 km2 (870 mi.2) of the southern halves of Game Management Units (GMUs) 61 and 62, and about 155 km2 (60 mi.2) of the northern edge of GMU 70 (between state highway 145 and San Miguel River). The area is bounded 116 �by state highway 348 at Delta, 25 Mesa road and Forest Service road FS503 to Nucla, state highway 97 to state highway 141 to state highway 145 to Placerville, state highway 62 to Ridgeway, U.S. highway 550 to Montrose, and U.S. highway 50 to Delta. The study area seems typical of puma habitat in Colorado that has vegetation cover that varies from the pinon-juniper covered foothills starting from about 1,700 m elevation to the spruce-fir and aspen forests growing to the highest elevations of about 3,000 m. Mule deer (Odocoileus hemionus) and elk (Cervus elaphus) are the most abundant wild ungulates available for puma prey. There are cattle and domestic sheep raised on summer ranges on the study area. Year-round human residents live along the eastern and western fringe of the area, and there is a growing summer residential presence especially on the southern end of the plateau. A highly developed road system makes the study area well accessible for puma research efforts. A detailed description of the Uncompahgre Plateau is in Pojar and Bowden (2004). METHODS Reference and Experimental Treatment Periods This research is structured in two 5-year periods: a reference period (years 1―5) and a treatment period (years 6―10). The reference period is expected to cause a population increase phase. The treatment period will be managed to cause a population decline phase. In both phases, puma population structure, and vital rates will be quantified, and some management assumptions and hypotheses regarding population dynamics will be tested. Contingent upon results of pilot studies, we will also estimate puma numbers, population growth rates, evaluate enumeration methods, and test other hypotheses (Logan 2004). The reference period, without recreational puma hunting as a major limiting factor, is consistent with the natural history of the current puma species in North America which evolved life history traits during the past 10,000―12,000 years (Culver et al. 2000) that enable puma to survive and reproduce (Logan and Sweanor 2001). In contrast, puma hunting, with its modern intensity and ingenuity, might have influenced puma evolution in western North America for the past 100 years. Hence, the reference period, years 1―5, will provide conditions where individual puma in this population (of estimated sex and age structure) express life history traits interacting with the environment without recreational hunting as a limiting factor. Theoretically, the main limiting factors will be catchable prey abundance (Pierce et al. 2000, Logan and Sweanor 2001). This should allow researchers to understand basic system dynamics before the treatment (i.e., controlled recreational hunting). In the reference period, all puma in the study area will be protected, except for individual puma that might be involved in depredation on livestock or human safety incidents. In addition, all radio-collared and ear-tagged puma that range in a buffer zone, that includes the northern halves of GMUs 61 and 62, will be protected from recreational hunting. The reference period will allow researchers to quantify baseline demographic data on the puma population to estimate parameters for the CDOW’s model-based approach to puma management. Moreover, it will allow researchers to develop and test puma enumeration methods when population growth is known to be in one direction― increasing. Without the hunting closure, pilot data for enumeration methods could be confounded by not knowing if the population was increasing, declining, or stable. The reference period will also facilitate other operational needs (because hunters will not be killing the animals) including the marking of a large proportion of the puma population for capture-markrecapture estimates, and the gathering of movement data from GPS-collared puma to help formalize exact sampling designs for enumeration methods. During the treatment period, years 6―10, experimentally structured recreational puma hunting will occur on the same study area with the intent of causing a decline phase in the puma population by using management prescriptions structured from information learned during previous years. Using 117 �recreational hunting for the treatment is consistent with the CDOW’s objectives of manipulating natural tendencies of puma populations, particularly survival, to maintain either population stability or population suppression (CDOW, Draft L-DAU Plans, 2004). Theoretically, puma survival will be influenced mainly by recreational hunting, which will be quantified by agent-specific mortality rates of radio-collared puma. The portion of adults and subadults in the population will be reduced by approximately 20% in year 6 and 20% more in year 7. The 20% change was identified by Division managers that requested enumeration tools that might detect 20% changes in puma populations. For managers, detecting the magnitude of puma population decline phases is probably more important that detecting the magnitude of population increase phases. This will also allow quantification of puma population characteristics and vital rates and initial tests of enumeration methods during a decline phase. Additional reductions may be made to test enumeration methods and other hypotheses that may be related to effects of hunting (i.e.,: relative vulnerability of puma sex and age classes to hunting, variations in puma population structure due to hunting) and puma―prey interactions (i.e., lines of research identified in the Colorado Research Program, Fig. 1). Those decisions can be made later in project development and as late as years 8―10. The killing of tagged and collared puma during the treatment period will not hamper operational needs (as it would during the start-up years), because by the beginning of this period, a large majority of independent puma in the population will be marked, and sampling schemes will be formalized. Puma on the study area that may be involved in depredation of livestock or human safety incidences may be lethally controlled. Researchers that find that GPS-collared puma have killed domestic livestock will record such incidents to facilitate reimbursement to the property owner for loss of the animal(s). In addition, researchers will notify the Area Manager of the CDOW of Wildlife if they perceive that an individual puma may be a threat to public safety. Field Methods Puma Capture: Realizing that puma live at low densities and capturing puma is difficult, as a starting point, our logistical aim will be to have a minimum of 6 puma in each of 6 categories (36 total) radio-tagged in any year of the study if those or greater numbers are present. The 6 categories are: adult female, adult male, subadult female, subadult male, female cub, male cub. Our aim is to provide more quantitative and precise estimates of puma demographics than were achieved in earlier Colorado puma studies. This relatively large number of puma might represent the large majority of the puma population on the study area, and will provide the basic data for age- and sex-specific reproductive rates, survival rates, agent-specific mortality rates, emigration rates, and movement data pertinent to sampling designs for various projects. Assuming that the puma population density on the study area is relatively low at the beginning of this study― about 1 adult/100 km2 and the sex ratio is equal (Anderson et al. 1992, Logan and Sweanor 2001:167), then there might be 22 adults, 11 males and 11 females. Also assuming that the total population contains 10% subadults and 34% cubs (Logan and Sweanor 2001), then there might be 4 subadults and 13 cubs with equal sex ratios in a total population of 39 puma. If we achieve our logistical aim in the first 1―2 years (recognizing that the population might grow), then we should be able to quantify population characteristics and vital rates for a majority of the puma population in those years and build upon the tagged number in each subsequent year. Thus, our inferences will pertain to the large majority of the puma population, if not the population on the study area, instead of a relatively small sample of it. We anticipate it may take 2 years to mark the large majority of puma in the population. In addition, the study area is large and will require some time to learn to access it efficiently. Puma capture and handling procedures have been approved by the CDOW Animal Care and Use Committee (file #08-2004). All captured puma will be examined thoroughly to ascertain sex and describe 118 �physical condition and diagnostic markings. Age of adult puma will be estimated initially by the gum-line recession method (Laundre et al. 2000) and dental characteristics of known-age puma (Logan and Sweanor, unpubl. data). Ages of subadult and cub puma will be estimated initially based on dental and physical characteristics of known-age puma (Logan and Sweanor unpubl. data). Body measurements recorded for each puma will include at a minimum: mass, pinna length, hind foot length, plantar pad dimensions. Tissue collections will include: skin biopsy (from the pinna receiving the 6 mm biopsy punch for the ear-tags) and blood (30 ml from the saphenous or cephalic veins) for genotyping individuals, parentage and relatedness analyses; disease screening; hair (from various body regions) and fecal DNA for genotyping tests of field gathered samples. Universal Transverse Mercator Grid Coordinates on each captured puma will be fixed via Global Positioning System (GPS, North American Datum 27). Puma will be captured year-round using 4 methods: trained dogs, cage traps, foot-hold snares, and by hand (for small cubs). Capture efforts with dogs will be conducted mainly during the winter when snow facilitates thorough searches for puma tracks and the ability of dogs to follow puma scent. The study area will be searched systematically multiple times per year by four-wheel-drive trucks, all-terrain vehicles, snow-mobiles, walking, and possibly horse- or mule-back. When puma tracks ≤1 day old are detected, trained dogs will be released to pursue puma to capture. Puma usually climb trees to take refuge from the dogs. Adult and subadult puma captured for the first time or requiring a change in telemetry collar will be immobilized with Telazol (tiletamine hydrochloride/zolazepam hydrochloride) dosed at 5 mg/kg estimated body mass (Lisa Wolfe, DVM, CDOW, attending veterinarian, pers. comm.). Immobilizing agent will be delivered into the caudal thigh muscles via a Pneu-Dart® shot from a CO2-powered pistol. Immediately, a 3m-by-3m square nylon net will be deployed beneath the puma to catch it in case it falls from the tree. A researcher will climb the tree, fix a Y-rope to two legs of the puma and lower the cat to the ground with an attached climbing rope. Once the puma is on the ground, its head will be covered, its legs tethered, and vital signs monitored (Logan et al. 1986). (Normal signs: pulse ~70―80 bpm, respiration ~20 bpm, capillary refill time ≤2 sec., rectal temperature ~101oF average, range = 95―104oF) (Kreeger 1996). A cage trap will be used to capture adults, subadults, and large cubs when puma can be lured into the trap using road-killed or puma-killed ungulates (Sweanor et al. 2005). Efficiency of the trap might be enhanced by using an automated digital call box that emits puma vocalizations (Wildlife Technologies, Manchester, NH). A cage trap will be set only if a target puma scavenges on the lure (i.e., an unmarked puma, or a puma requiring a collar change). Researchers will continuously monitor the set cage trap from about 1 km distance by using VHF beacons on the cage and door. This allows researchers to be at the cage to handle captured puma within 30 minutes. Puma will be immobilized with Telazol injected into the caudal thigh muscles with a pole syringe. Immobilized puma will be restrained and monitored as described above. If non-target animals are caught in the cage trap, we will open the door and allow the animal to leave the trap. Foot-hold snares will be used to capture adults, subadults, and large cubs only when safe snare sites at puma kills can be located as described by Logan et al. (1999). Snares set at puma kills will be monitored continuously with VHF beacons on the snares from about 1 km distance. We will not set snares at sites where tracks indicate that other mammals (e.g., deer, elk, bear, bighorn sheep, livestock) are also active. Puma will be immobilized with Telazol injected into the caudal thigh with a pole syringe. Vital signs will be monitored during the handling procedures. Efficiency of snares might also be enhanced with the use of an automated call box with puma or prey vocalizations. Small cubs (≤10 weeks old) will be captured using our hands (covered with clean leather gloves) or with a capture pole. Cubs will be restrained inside new burlap bags during the handling process and will not be administered immobilizing drugs. Cubs at nurseries will be approached when mothers are 119 �away from nurseries (as determined by radio-telemetry). Cubs captured at nurseries will be removed from the nursery a distance of ~100 m to minimize disturbance and human scent at nurseries. Immediately after handling processes are complete, cubs will be returned to the exact nurseries where they were found (Logan and Sweanor 2001). Marking, Global Positioning System- and Radio-telemetry: Puma do not possess easily identifiable natural marking, such as tigers (see Karanth and Nichols 1998, 2002), therefore, the capture, marking, and GPS- or VHF- collaring of individual puma is essential to a number of project objectives, including estimating vital rates and gathering movement data on puma to formalize designs for developing and testing enumeration methods. Adult, subadult, and cub puma will be marked 3 ways: GPS/VHF- or VHF-collar, ear-tag, and tattoo. The identification number tattooed in the pinna is permanent and cannot be lost unless the pinna is severed. A colored (bright yellow or orange), numbered rectangular (5 cm x 1.5 cm) ear-tag (Allflex USA, Inc., DFW Airport, TX) will be inserted into each pinna to facilitate individual identification during direct recaptures. Cubs ≤10 weeks old will be eartagged in only one pinna. Locations of GPS- and VHF-collared puma will be fixed about once per week from light fixedwing aircraft (e.g., Cessna 182) fitted with radio signal receiving equipment (Logan and Sweanor 2001). This monitoring will enable researchers to find GPS-collared puma to acquire remote GPS location reports from the ground, monitor the status (i.e., live or dead) of individual puma, and to recover carcasses for necropsy. It will also provide simultaneous location data on mothers and cubs. GPS- and VHF-collared puma will be located from the ground opportunistically using hand-held yagi antenna. At least 3 bearings on peak aural signals will be mapped to fix locations and estimate location error around locations (Logan and Sweanor 2001). Aerial and ground locations will be plotted on 7.5 minute USGS maps (NAD 27) and UTMs along with location attributes will be recorded on standard forms. GPS locations will be mapped using ArcGIS software. Adult and subadult female pumas will be fitted with GPS collars (approximately 400 g each, Lotek Wireless, Canada). Initially, GPS-collars will be programmed to fix and store puma locations at 4 times per day to sample daytime, nighttime, and crepuscular locations (i.e., 0:00, 06:00, 12:00, 19:00). GPS locations for puma will provide precise, quantitative data on puma movements mainly to provide data to formalize study designs, to test assumptions for capture-mark-recapture methods for this project, and to assess the relevance of puma DAU boundaries. The GPS-collars also will provide basic information on puma movements and locations to design other pilot studies in this program on vulnerability of puma to sport-harvest, habitat use, and predation frequency on mule deer and elk. Subadult male pumas will be fitted initially with conventional VHF collars (Lotek, LMRT-3, ~400 g each) with expansion joints fastened to the collars, which allows the collar to expand to the average adult male neck circumference (~46 cm). If subadult male puma reach adulthood on the study area, we will recapture them and fit them with GPS collars. VHF radio transmitters on GPS collars will enable researchers to find those pumas on the ground in real time to acquire remote GPS data reports, facilitate recaptures for re-collaring, and to check on their reproductive and physical status. VHF transmitters on GPS- and VHF-collars will have a mortality mode set to alert researchers when puma have been immobile for at least 3 hours so that dead puma can be found to quantify survival rates and agent-specific mortality rates by gender and age. We will attempt to collar all cubs in observed litters with small VHF transmitter mounted on an expandable collar (~100g, MOD 210, Telonics, Inc., Mesa, Arizona) when cubs weigh 2.3―11 kg (5―25 lb). Cubs with mass ≥11 kg can still wear these small expandable collars until they are about 12 months old. Cubs approaching the age of independence (~11―14 mo. old) may be fit with Lotek LMRT-3 VHF 120 �collars (~400 g) with expansion links. Cubs will be recaptured to replace collars as necessary. Monitoring radioed cubs allow quantification of survival rates and agent-specific mortality rates (Logan and Sweanor 2001). Capture-Mark-Recapture: Capture-mark-recapture methods will be evaluated initially as a pilot study. Capturing and marking puma is time consuming, and would lengthen the time to thoroughly search the study area for capturing and marking puma during capture-recapture occasions needed for population estimation. Therefore, we will capture and mark pumas prior to performing capture-recapture or re-sight occasions using using methods such as houndsmen teams or trail cameras. In addition, by marking puma before capture-recapture occasions begin, we will have opportunities to capture female puma at different stages of their reproductive status, and thus reduce the chance that mothers in a stage with suckling cubs and small activity areas are not detected and marked on the study area. After cubs are weaned, the mothers’ activity area expands (Logan and Sweanor 2001). The probability of females having suckling cubs in winter is naturally small; that season exhibits the lowest rate of births (Logan and Sweanor 2001). Capture-recapture occasions to estimate the population of independent puma may not begin until the end of the second winter or the third winter when we have a large majority of the puma population sampled and marked. Occasions performed at that time will be viewed as a pilot study allowing us to examine the logistics of the field methods, the extent to which model assumptions are met, performance of field methods (e.g., detection differences by sex or life stage as revealed by GPS data on collared puma), and precision of capture-recapture models used to estimate the puma population. Analytical Methods Population Characteristics: Population characteristics each year will be tabulated with the number of individuals in each sex and age category. Age categories, as mentioned, include: adult (puma ≥24 months old, or younger breeders), subadults (young puma independent of mothers, <24 months old that do not breed), cubs (young dependent on mothers, also known as kittens) (Logan and Sweanor 2001). When data allow, age categories may be further partitioned into months (for cubs and subadults) or years (for adults). Reproductive Rates: Reproductive rates will be estimated for GPS- and VHF-collared female puma directly (Logan and Sweanor 2001). Genetic paternity analysis will be used to ascertain paternity for adult male puma (Murphy et al. 1998). Methods will be tested in Dr. M. Douglas’s Laboratory (Colorado State University, Department of Fishery and Wildlife Biology). Survival and Agent-specific Mortality Rates: Radio-collared puma will provide known fate data which can be used to estimate survival rates for each age stage using the binomial survival model (Williams et al. 2001:343-344) or analyzed in program MARK (White and Burnham 1999, Cooch and White 2004). Agent-specific mortality rates can be analyzed using proportions and Trent and Rongstad procedures (Micromort software, Heisey and Fuller 1985). Cub survival curves for each gender will be plotted with survival rate on age in months (Logan and Sweanor 2001:119). Population Estimates: Capture-recapture models will be evaluated initially as a pilot study to estimate the parameters of primary interest― absolute numbers of independent puma (i.e., number of adult and subadult puma present in the survey area) and puma density (i.e., number of independent puma/100 km2) each winter― December through March― when snow facilitates detection and capture of puma, provided that we meet model assumptions. The December―March period also corresponds with Colorado’s puma hunting season. The population of interest is independent puma (i.e., adults and subadults) because those are the puma that can be legally killed by recreational hunters. Furthermore, adults comprise the breeding segment of the population and subadults are non-breeders that are potential recruits into the adult population in ≤1 year. Thus, the sampling unit is the individual independent puma (~≥1 yr. old). 121 �General assumptions for closed capture-recapture models are: (1) the population is closed; (2) animals do not lose their marks during the interval; (3) all marks are correctly noted and recorded at each trapping occasion; (4) each animal has a constant and equal probability of capture on each capture occasion. Open population models allow the assumption of closure to be relaxed (Otis et al. 1978, White et al. 1982, Pollock et al. 1990). The robust design is a combination of closed and open models; thus, assumptions are a combination of the assumptions for closed and open population methods (Kendall 2001). To analyze capture-recapture data, closed, open, and the robust design models are available in program MARK. Akaike’s Information Criterion will be used to select the most parsimonious models based on AICc score ranks and the difference in AIC (∆AIC) between models (Burnham and Anderson 1998). MARK results also include estimates of abundance. Because the precision of estimates for small populations is sensitive to the probability of capture (White et al. 1982, Pollock et al. 1990), our operational goal will be to achieve capture probabilities of at least 0.5 for each animal per capture occasion. Capture simulations using MARK software (Cooch and White 2004) indicate that greater capture probabilities and more capture occasions yield more precise estimates. The capture probability for the simplest closed model [M(o)], which assumes that every member of the population has the same probability of capture (p) for each sampling period, suggest that for a population of 30 animals (i.e., adults plus subadult puma, which might be present by the end of year 2, see Puma Capture above) p must equal 0.5 for 3 capture occasions to attain a coefficient of variation (V) of 0.1. If 6 capture occasions are used, then a p of 0.3 might yield a V of 0.09. In addition, behavior, movements, survival and mortality of GPS- and VHF-collared puma will allow direct biological examinations of assumptions of geographic and demographic closure (White et al. 1982) and variation in capture probability of individual puma and puma classes (i.e., adult females, adult males, subadult females, subadult males). If capture probabilities vary by puma class, we will examine if data stratification is necessary or possible (depending upon sample size). For example, we might expect the larger home ranges of male puma to expose them to more search routes, thus, this may increase their probability of capture. If the assumption of demographic closure cannot be satisfied, then open population models and the robust design would be more appropriate (Pollock et al. 1990, Williams et al. 2001). Collared puma will allow us to determine the number of marked puma present in the search area each capture-recapture occasion. Furthermore, GPS locations (4 fixes/day) on individual puma will provide data on the probability that puma may temporarily move out of and back into the survey area between capture occasions. Unmarked puma that are subsequently GPS-collared should provide such information, too. ArcView geographic information system software will be used to map and analyze puma locations, movements, and home ranges. It will also be used to map and quantify attributes of the study area and sampling frames. Rate of Population Increase: Finite rates of increase (λ = Nt+1/Nt) between consecutive years and average annual rates of increase (r) for 3- to 5-year periods and levels of precision will be calculated (Caughley 1978, Van Ballenberghe 1983) and plotted. Functional Relationships: Graphical methods will be used to examine functional relationships between puma density and vital rates, relationships between puma density estimated with direct capturerecapture methods (i.e., houndsmen teams) and possibly later (depending upon funding) by using estimates from DNA genotype or other mark- recapture methods. Linear regression procedures and coefficients of determination can be used to assess these functional relationships if data for the response variable are normally distributed and the variance is the same at each level. If the relationship is not 122 �linear, data is non-normal, and variances are unequal, we will consider appropriate transformations of the data for regression procedures (Ott 1993). Non-parametric correlation methods, such as Spearman’s rank correlation coefficient, can also be used to test for monotonic relationships between puma abundance and other parameters of interest (Conover 1999). Statistical analyses will be performed using SYSTAT and SAS software. The risk of committing a type I error (i.e., rejecting a null hypothesis that is actually true) will be controlled at alpha = 0.10 because we will normally have small population sizes (typical of studies of large obligate carnivores). The higher alpha level will increase the probability of detecting a change and reduce the risk of a type II error (i.e., failing to reject a null hypothesis that is false). For managers, the risk of a type II error is probably more important. RESULTS AND DISCUSSION Segment Objective 1 Field research to quantify puma population structure, vital rates, and causes of mortality for this report extended from August 2006 to July 2007. Our searches to detect puma presence covered the entire study area. We allocated most of our effort in areas where we consistently found tracks that we thought were of unmarked pumas, particularly in the northeast and southwest areas where we found little or no evidence of pumas during the previous 2 years. We made 54 puma captures during the period (9-10 adult females [1 adult female captured twice], 7 adult males [1 adult male probably captured 3 times, another captured twice], 1 subadult female, 0-1 other subadults, and 30 cubs [4 of them captured twice each]). As our main method to capture, sample, and mark adult and subadult pumas, we used trained dogs from November 13, 2006 to May 11, 2007. Those efforts resulted in 78 search days, 177-178 puma tracks detected, 45-47 pursuits, and 22 puma captures (Table 1). Puma capture efforts (i.e., search days) with dogs in this period was similar to our efforts in the 2 previous efforts (Table 2). But, the frequency of pursuits and puma captures has increased over the 2 previous periods. In addition, the number of adult and subadult pumas captured for the first time declined from 11 (Oct. 2005 to Apr. 2006) to 6 (this period). This included 1 adult female or subadult puma that could not be handled for safety reasons (see Tables 3 and 4). Of the pumas we captured, but could not handle, it is probable that we captured and marked 1 adult male (M51) and 1 adult female (F50) in subsequent capture efforts. Our puma capture efforts using ungulate carcasses and cage traps extended from August 2006 to July 2007. We used 64 road-killed mule deer, 7 road-killed elk, 3 puma-killed mule deer, and 1 puma killed elk at 26 sites to capture pumas 8 times (Tables 5). Pumas scavenged 16 of 71 (22.5%) of the roadkilled ungulate carcasses used for bait. This was similar to the results last years (16 of 80, 20%). Five pumas were captured, sampled, and marked for the first time by using dogs and cage traps, (Table 3). Fifteen recaptures of 13 marked pumas were made with the use of dogs and cage traps; GPS/VHF collars were replaced as needed (Table 6). We captured, sampled, and marked 26 cubs in 10 litters that were captured by hand at nurseries (Table 7). Search efforts throughout the study area also revealed the presence of at least 4 other independent females and 1 independent male. The tracks we found of those animals were too old to pursue (i.e., probability of capture with the dogs was negligible). We could separate the activity of those pumas from the GPS- and VHF- collared pumas in time and space. In addition, 2 of the females were in association with cubs. One female was followed by 2 cubs about 5 to 6 months old in December and January, when we captured but could not handle 1 or 2 of the cubs (Table 4). Another female was followed by 1 large cub (probably a male) likely 10 or more months old. And another female on the southwest portion of the 123 �study area might have been an adult if it were associated with a female cub (~6 mo. old) that was hit and killed by a vehicle on highway 62 on January 28, 2007. Our search and capture efforts during November 2006 through May 2007 enabled us to estimate a minimum count of 24 independent pumas detected on the Uncompahgre Plateau study area. The count included 16 females and 8 males. Of those, 12 adult females and 7 adult males were probably marked animals (79% of independent pumas detected). Of the remainder, 2 females were adults because they were followed by cubs, and 2 females and 1 male were of unknown independent status (i.e., either subadult or adult). Figure 2 indicates the estimated use areas of those independent pumas. Some of the animals range outside the borders of the study area, as indicated by movements of GPS- and VHFcollared pumas. There appears to be variation in puma numbers on the west and east slopes of the study area. The west slope count includes 8 independent pumas (5 females― 4 marked, 1 unmarked; 3 males― 2 marked, 1 unmarked). The east slope count includes 16 independent pumas (11 females― 8 marked, 3 unmarked; 5 males― all marked). Female home ranges overlap other female home ranges extensively, and are overlapped by male home ranges. Male home ranges overlap multiple female home ranges, and overlap other male home ranges somewhat. Anderson et al. (1992) studied pumas on the east slope of the Uncompahgre Plateau (i.e., GMU 62) during 1981 to 1988. Sport-hunting was banned during that study as it is in this study Reference Period. As our current effort results in larger samples and progresses in time through the Reference and Treatment periods, similarities and differences in results of the 2 research efforts, now separated by more than 15 years, should illuminate reliable knowledge for puma management in Colorado. Our current puma research on the Uncompahgre Plateau has been underway for 2.7 years (compared to 7 years of Anderson et al. 1992). Our data analysis at this stage of the research is not by any means exhaustive or complete, yet, our data set enables some preliminary comparisons with Anderson’s completed work (Anderson et al. 1992). In the Anderson et al. (1992) study, the average capture effort with dogs was 91.1 days per winter (range = 32 to 136, n = 7) resulting in an average capture effort of 13.9 days per puma. Of 189 pursuits of pumas, 110 (58%) were successful (either of radio-collared or non-collared animals). They captured 47 pumas for an average capture rate of 13.9 days per puma. Eight other pumas, all female cubs ≤7 months old, were caught in steel leg-hold traps by trappers, and were added to the study animal population. So far, in our 3 winters, the average effort is 79.3 days (range = 78 to 82). Of 123 pursuits, 50 (41%) were successful. We captured and GPS- or VHF-collared 25 pumas for the first time, yielding a capture rate of 10.0 days per capture. Other capture efforts and results between the 2 studies are not comparable, because Anderson et al. (1992) did not routinely attempt to capture pumas using cage traps or at nurseries like we are. In their effort, Anderson et al. (1992) captured 57 pumas, of which 49 were radio-collared. In our current effort, we captured, sampled, and marked 68 pumas, of which 61 were radio-collared. Puma mass recorded by Anderson et al. (1992:86) for pumas having an estimated age ≥24 months, averaged 61.6 kg for 8 males, (SD = 5.7, range = 51.8 to 70.8) and 44.5 kg for 14 females (SD = 3.6, range = 38.5 to 49.9). So far in our current study, mass for pumas ≥24 months old averaged 59.8 kg for 9 males (SD = 8.1, range 40 to 68 kg) and 38.4 kg for 11 females (SD = 4.9, range = 31 to 46). Sexual dimorphism has been described for puma throughout the species range (Young and Goldman 1946) and has been explained as a potential result of sexual selection (Logan and Sweanor 2001:109). Segment Objective 2 We captured, sampled, and marked 26 puma cubs produced by 10 females (Table 7). Twentythree of the cubs were examined at 8 nurseries when the cubs were 29 to 41 days old. The sexes were 17 124 �males and 6 females. Four other cubs, including 2 males and 2 females, were caught when they were about 158 to 215 day old. In addition to those offspring, 2 cubs, about 152 to 183 days old, were detected in association with an unmarked female that we pursued. One or 2 of those cubs were captured in different events; the female sex was determined for one of the cubs. But, neither cub could be handled safely for further sampling. The estimated birth month for the 10 litters were April (1), May (1), July (5), August (2), and September (1). During the past 27 months of this work we compiled data on puma reproduction that was heretofore not available for Colorado. We examined 38 cubs from 13 litters aged 29 to 42 days old where we were reasonably sure that we examined all the cubs at the nurseries. The sex ratio of the observed cubs was 21 males:17 females. The mean (±SD) and extremes of litter sizes were 2.84 (±0.99), 1 to 4. The distribution of puma births by month indicate puma births extending from March into September, with 18 of 20 births occurring May to September (Fig. 3). In addition, 8 birth intervals for 7 different female pumas averaged 14.99 months (SD = 3.40), and ranged from 11.7 to 20.5 months (Table 8). Based on observations (from GPS data) of associations between 4 mothers and putative sires, 4 gestation periods averaged 92.0 days (SD = 1.68), which is consistent with average puma gestation reported in literature (i.e., mean ± SD = 91.9 ± 4.1, Anderson 1983:33, mean = 91.5 ± 4.0 Logan and Sweanor 2001:414). Anderson et al. (1992:47) reported of “17 postnatal litters about 10-240 days in estimated age from 12 individual females, the mean (±SD) and extremes of litter sizes were 2.41 ± 0.8, 1-4”. “Because most postnatal young were not handled, their sex ratio is unknown” (Anderson et al (1992:48). In addition, because cubs were first observed at older ages, it is likely that some post-natal mortality had occurred. This is one explanation for smaller litters observed by Anderson et al. (1992). Anderson et al. (1992:47-48) found that of 10 puma birth dates 7 were during July, August, and September, 2 in October, and 1 in December, with most breeding occurring April through June. Data on our 20 litters adds to Anderson’s data (Fig. 3), and indicates puma births in Colorado occurring in every month except January and November (so far). Our data suggests that the majority of puma breeding activity occurs February through June. Anderson’s observation of two 12-month birth intervals for one female (Anderson et al. 1992:48) is at the low range of our observations (above). Segment Objective 3 & 4 From December 2, 2004 (start of our research) to July 31, 2007, we radio-monitored 9 adult male and 12 adult female pumas to quantify survival and agent-specific mortality rates (Table 9). One adult male is known to have died. M4 was about 37 to 45 months old when he was killed by an unidentified male puma along the southeast boundary of the study area. We lost contact with 3 adult males apparently due to GPS/VHF collar failure (M1, M6, M27). Evidence in the field suggests that all 3 males might still be alive. One adult female is known to have died. F50 was about 29 to 31 months old when she apparently died of natural causes (exact agent could not be identified). We have radio-monitored 6 subadult pumas (Table 10). None of those died while we were monitoring them. F23 has become a breeding adult on the study area. M5 dispersed from his natal area and the study area at about 13 months old and went to the northwest slope of the Uncompahgre Plateau where he has apparently established an adult territory. M49 was orphaned at 9 months old when his mother F50 died. He has since dispersed from his natal area and the study area to the northeast slope of the Uncompahgre Plateau. We continue to monitor his status. On the other hand, we have lost contact with 2 subadult males and 1 subadult female. Puma M11 became a subadult at 13 months old and dispersed from his natal area at 14 months old. He was last located in the Dolores River valley between Stapletone and Stoner, Colorado, on December 14, 2006. F52 dispersed from the study area before we lost track of her in the area of the Black Canyon of the Gunnison in mid-May 2007. We lost track of M31 seven days after he was captured. He might have dispersed from the study area. Efforts to locate him by 125 �flying over and around the study area have not been successful. Dispersal rates and distances will be reported after we have compiled more complete data. In addition to the subadults discussed above, a nonmarked female puma about 18 to 24 months old was killed by a vehicle November 4, 2006 on highway 550, which forms the southeast boundary of our study area. The female appeared to be in good health (41 kg), was not pregnant, and was not lactating. Anderson et al. (1992) found that all 9 radio-collared male pumas dispersed from their natal areas, and 2 of 6 radio-collared females did not disperse from their natal areas (A. E. Anderson, Sep. 1993, errata for Anderson et al. 1992:61). Mean ± SD and range of dispersal distances (km) for 8 males, aged 10 to 13 months old at dispersal, were 86.2 ± 51.3, 23 to 151. For 4 females, aged 11 to 31 months old at dispersal, mean ± SD and range of dispersal distances (km) were 37.0 ± 15.3, 17 to 54 (Anderson et al. 1992:63). The current closure on sport-hunting on the study area and protection of marked pumas from sport-harvest on the buffer area on the northern portion of the Uncompahgre Plateau for the Reference Period appears to be operating, so far. None of the adult or subadult pumas wearing functional GPS- or VHF- collars have died due to human causes. This reference condition enables us to quantify puma survival rates and agent-specific mortality rates of adult and subadult pumas (i.e., harvest-age pumas) in the absence of direct human-caused mortality factors related to sport-hunting. So far, survival of radiomonitored adult and subadult pumas in the study and buffer areas appears to be high. In addition, the population sex and age structure can be examined in this reference condition. As indicated in Figure 4, the adult age structure appears to be indicative of high survival rates during the past 3 winters without sporthunting mortality. These data will be valuable in comparisons of sex and age structure during the Treatment Period and with the structure of harvested pumas in other regions of Colorado. But, we will wait for greater sample sizes (i.e., greater numbers of radio-monitored pumas and duration) before we develop more quantitative analyses of survival rates and agent-specific mortality rates and attendant inferences. Thirty-nine puma cubs (20 males, 19 females) have been monitored by radiotelemetry for varying durations (Table 11). Three males (M5, M11, M49) are known to have survived to the subadult or adult stages (Table 10). Seven cubs (F13, F18, M22, F33, F34, F36, and M37) were killed and eaten by other pumas. At least 4 of those were subjects of male-induced infanticide. Sex of the puma involved in each of the other cases could not be determined. In addition, cub F45 was apparently killed by coyotes when she was 280 to 283 days old. F45 was separated from her adopted mother, F2, and also appeared to be emaciated at the time of her death. Cub F17 was killed by a vehicle on highway 550 when she was about 330 days old. She was not radio-collared at the time, but GPS data from her mother, F16, showed her in the vicinity of her offspring. Thus, F17 was probably still dependent on F16. Three cub deaths were due to our research activities, namely problems with the expandable radiocollars. F35 died at 37 days old probably as a result of starvation caused when the transmitter box got caught in her mouth. M42 died at 106 days old apparently from complications of septicemia caused by an infection at the axis of the right foreleg. The cub put his right foreleg through the expandable collar and the collar material lacerated the right underarm as the animal grew, enabling the infection. M60 died at 49 days old, apparently from starvation. He apparently could not keep up with the movements of his mother, because he had put his right foreleg through the expandable collar, restricting his mobility. In addition to these deaths, 1 unmarked female cub (~6 mo. old) was killed by a vehicle on highway 62 on the southwest boundary of the study area on January 28, 2007 (mentioned earlier). We lost contact with a number of cubs because they shed their expandable radiocollars (Table 11). As this study proceeds, some cubs with which we have lost contact will be re-captured, re-observed, or harvested, and thus, provide more complete survival information. 126 �Clearly, data on cub survival and mortality are still preliminary. At this time, we can say that a minimum of 12 deaths occurred in 39 radio-collared cubs that we monitored for varying periods of time. This represents a minimum 0.31 mortality rate (12/39), including research-related causes. Subtracting the 3 research-related deaths, the minimum mortality rate is 0.25 (9/36). The main cause of death is being killed by another puma (0.78, 7/9). These rates should be interpreted as only rudimentary information. More complete data on cub survival and mortality will be forthcoming as our efforts continue. Anderson et al. (1992:50) reported on the fates of 21 radio-collared pumas (11 <24 months old, 10 ≥24 months old) from a total of 49 in the previous study where pumas were not hunted. Yet, 19 of those pumas died due to human causes, attributed to: legal kill outside the study area (7), capture-related (6), predator management (3), illegal kill (2), and suspected predacide (1). Other causes of mortality included, intraspecies strife (1) and disease (1). Actual age-stage and annual survival rates and agentspecific survival rates from our current effort will be compared with the Anderson et al. (1992) data set at a later date when we have greater samples, duration in research time, and more complete fate data (i.e., pumas currently without functional collars) to make such comparisons meaningful. Differences might be illuminated. For example, research of a puma population in New Mexico that was not hunted for 10 years indicated that the major cause of death for both sexes and all age stages of pumas was intraspecifies strife, and male-induced infanticide (Logan and Sweanor 2001). Although we have observed 3 male pumas disperse from natal areas, and no females disperse, our current research is too short in duration and samples too small yet to make meaningful comparisons with Anderson’s earlier effort, particularly regarding offspring dispersal rates, distances moved, and philopatry. Dispersal and philopatry have been explained as life history strategies in pumas that assist gene flow, colonization, population maintenance, and individual survival and reproductive success (Logan and Sweanor 2001). Thus, such strategies would be expected to be conserved, and expressed in puma populations at different times and different locations. In addition, because puma emigration and immigration (i.e., via dispersal) have been shown to be important processes in puma population dynamics (Sweanor et al. 2000), we need larger samples and longer research duration in this study to estimate those parameters. Segment Objective 5 Twenty adult pumas (7 males, 13 females) were fit with Lotek 4400S GPS collars since field research began in December 2004. The collars are programmed to fix 4 locations per day (00:00, 06:00, 12:00, and 19:00). The number of GPS locations per individual puma ranged from 113 to 2,759 (Table 12). Winter activity areas for GPS-collared pumas were estimated (Table 13) with fixed kernel and minimum convex polygon home range estimators (ArcView 3.2 Animal Movement Extension). These estimates are intended for use in developing the sampling frame for the puma population estimation pilot project (see Introduction). In addition, 5 adult and subadult pumas have been monitored with VHF radiocollars (Table 14). Anderson et al. (1992) provided an exhaustive analysis of seasonal puma home ranges and movements using data collected from VHF-collared animals during 1982 to 1988. We have not yet conducted an exhaustive analysis of adult puma home ranges and movements with the GPS data from our current puma research efforts. Instead, we provide only limited descriptive information in Tables 13, 14 and Fig. 2. Given the different types of location data and analytical methods, only broad descriptive comparisons might be made between the 2 studies at this time. Elemental similarities in home range attributes of pumas in the Anderson et al. (1992) research and our current effort, include: current home ranges of some puma overlap extensively with home ranges of puma documented by Anderson et al (1992), home ranges of male and female pumas are large, male home ranges are larger than female home ranges, male home ranges overlap multiple female home ranges, female home ranges overlap other female home ranges sometimes extensively, male home ranges overlap other male home ranges to a lesser 127 �extent than female home ranges. These characteristics are generally similar for pumas in other populations that have been studied with adequate intensity and duration (Beier and Barrett 1993, Logan and Sweanor 2001), and reflect behavioral strategies of male and female pumas that seem to contribute to individual survival and reproductive success (Logan and Sweanor 2001). Segment Objective 6 To investigate the potential that puma hunters might detect puma mothers away from their cubs, we continued gathering data on spatial associations of puma mothers and their cubs during the puma hunting season, which extends from November through March each winter in Colorado. Female pumas are fair game in Colorado, unless they are accompanied by 1 or more cubs. Mothers that are caught away from their cubs could be legally harvested. Such incidents would result in cubs being orphaned. Orphaned cubs that ≤6 months old could have a survival rate (to the subadult stage) of <0.05. Orphaned cubs 7 to 12 months old might have a survival rate (to the subadult stage) of about 0.7 (K. Logan, unpublished data). From November 7, 2006 to March 22, 2007 we located 1 to 4 radio-collared families of puma mothers and cubs from an airplane 49 times (Table 15).To assess whether mothers were apart or in close association with cubs, we needed to consider error in aerial locations. We recovered 7 puma radiocollars that we located from the airplane and fixed with GPS and then fixed the actual locations of collars on the ground with GPS. Range of location error was 20 to 520 m (mean = 282.86, SD = 164.75). We decided to use distances greater than the extreme high range of location error (520 m) as the metric to decide if puma mothers might be detected away from their cubs by hunters. Forty-one (83.7%) of observations located mothers and cubs ≤500 m apart, within the extreme margin of location error. Mothers were ≥520 m from their cubs during 8 (16.3%) of the observations (mean distance = 1,120 m, SD = 1,214.40, range = 616 to 4,101). The results for last winter were similar to our results the previous winter (15.2% and 16.3%, Table 15). Anderson et al. (1992:70-71) recorded 69 instances of simultaneous aerial locations of 7 pairs of puma mothers and dependent young. They reported that mothers and young were together in 21 (30.4%) of those instances, and they were 1 to 2.2 km apart in 48 (69.6%) of those instances. Segment Objective 7 Intensive effort to quantify puma use rates on ungulates by investigating puma GPS clusters continued during this period as an expansion of our pilot effort in the first research year (Logan 2005). That work proved the reliability of the GPS technology to allow us to gather quantitative information on ungulate prey use rates by pumas. In summary, 7 GPS-collared adult pumas (3 males, 4 females) used 61 mule deer, 48 elk, 2 porcupines, and 1 beaver found at 139 puma GPS clusters we investigated. The current work is a collaborative effort among CDOW Mammals Researchers (M. Alldredge, E. Bergman, C. Bishop, D. Freddy, and K. Logan). This was another pilot effort because it involved the development and testing of clustering parameters, clustering routine, associated computer programs, and field investigation protocols. Here we report only the general summary of the pilot field investigations of puma GPS clusters from October 2006 to April 2007. Five types of puma GPS clusters (Bergman et al. 2006) were investigated for 13 GPS-collared adult pumas (8 female, 5 male). The sample unit was the individual puma. The field effort focused on investigating a sample of randomly chosen clusters from each cluster type. In addition, when other non-random S1 clusters (i.e., clusters with the highest probability of ungulate use detection) were conveniently located to random clusters targeted for investigation, field personnel would attempt to investigate those clusters, too. A total of 257 clusters were investigated, including 63 non-random S1 clusters, and 173 random clusters (S1, S2, S3, S4, S5). Mule deer and elk were about equally important to pumas as food (Tables 16, 17, 18). Other mammals were rarely found. The next step in this investigation involves examining the performance of all aspects of the GPS cluster investigations and modifying cluster parameters and field protocols to maximize the 128 �efficiency and reliability of our continuing efforts to quantify ungulate use by pumas on the Uncompahgre Plateau. We will make further progress to designing and implementing a pilot project to investigate puma population estimation methods on the Uncompahgre Plateau. CDOW personnel Mat Alldredge, Chad Bishop, Ken Logan (Mammals Research) and Paul Lukacs (Terrestrial) met with Dr. Gary White (Colorado State University) June 21, 2007 to discuss possible approaches to estimating puma numbers by using capture-recapture methods and models. Another method we will explore, with the collaboration Mammals Researcher Chuck Anderson, is helicopter-based puma track probability sampling. We will evaluate the potential for collaborative research on puma-human relationships on the Uncompahgre Plateau with the developing CDOW puma-human research on the Colorado Front Range. To date, we have gathered location data on 10 (7 adult females, 3 adult males) GPS-collared pumas with activity areas on the developed southeast portion of our study area, which includes: Fairway Pines, Loghill Village, and Fisher Creek subdivisions, numerous other private homes, Fairway Pines golf course and driving range, all adjacent to Ridgeway State Park (Fig. 2). In addition, 2 new subdivisions and golf courses are underdevelopment on the southeast quarter of the Uncompahgre Plateau. This is the same area that Anderson et al. (1992:80) received 17 useable questionnaires on puma observations from residents, and also had some radio-collared pumas frequenting these same developments. Linking puma-human research on the Uncompahgre Plateau and Front Range provides opportunities for increasing sample size (i.e., puma numbers, study sites) and observing variation in puma-human relationships. We collaborated with Dr. Sue VandeWoude (CSU) to develop a pilot study titled: Puma concolor immune health― Relationship to management paradigms and disease. Tissue samples (i.e., blood, saliva, feces) from pumas we capture are collected and shipped to her laboratory for analyses. That project will be expanded to The effects of urban fragmentation and landscape connectivity on disease prevalence and transmission in North American felids. A description of that project and preliminary results on infectious disease surveillance on 21 pumas (13 female, 8 male) sampled on the Uncompahgre Plateau are presented in Appendix I. SUMMARY Manipulative, long-term research on puma population dynamics, effects of sport-hunting, and development and testing of puma enumeration methods began in December 2004. After 31 months of effort, 68 pumas have been captured, sampled, marked, and released. Of those 61 were radio-collared. Age stages we have monitored have included 21 adults, 6 subadults, and 46 cubs. Data from the marked animals are used to quantify vital rates and puma population dynamics in a reference situation (i.e., without sport-hunting off-take). Data on research efforts and puma capture, fates, reproduction, and activity areas are presented. During November 2006 through May 2007 a minimum count of 24 independent pumas were detected on the Uncompahgre Plateau study area. The count included 16 females and 8 males. Of those, 12 adult females and 7 adult males were probably marked animals (79% of independent pumas detected). Our efforts to quantify reproduction are yielding reliable data for Colorado on puma litter sizes, offspring sex ratios, and birth intervals. In this reference period, survival of adult and subadult pumas appears to be high. So far, the main cause of death in puma cubs is infanticide by males. Twenty adult pumas (13 females, 7 males) have been fitted with GPS collars, yielding 113 to 2,759 locations per puma. Our evaluations on the frequency that puma mothers on the Uncompahgre Plateau are away from their cubs >520 meters during the Colorado hunting season is low (15.2 to 16.3%). Intensive efforts to quantify puma use of ungulates on the Uncompahgre Plateau continued. Mule deer and elk appeared to be about equally important as puma food. Preliminary comparisons of aspects of puma biology were made between our new research effort on the Uncompahgre Plateau and that of Anderson et al. (1992) in GMU 62 during 1981 to 1988. Research efforts for year 4 will focus on increasing numbers 129 �and distribution of sampled, marked, and GPS/radio-collared pumas on the study area for the principal objectives of this research. In addition, we will continue to investigate puma use of mule deer and elk, develop a pilot project to estimate pumas, and consider incorporating our data on pumas on the Uncompahgre Plateau to address questions pertaining to research on puma-human relationships in Colorado. All of these efforts should enhance the Colorado puma research and management programs. LITERATURE CITED ANDERSON, A. E. 1983. A critical review of literature on puma (Felis concolor). Colorado Division of Wildlife Special Report No. 54. _____, D. C. BOWDEN, AND D. M. KATTNER. 1992. The puma on Uncompahgre Plateau, Colorado. Technical Publication No. 40. Colorado Division of Wildlife, Denver. ANDERSON, C. R., JR., AND F. G. LINDZEY. 2005. Experimental evaluation of population trend and harvest composition in a Wyoming cougar population. Wildlife Society Bulletin 33:179-188. BEIER, P., AND R. H. BARRETT. 1993. The cougar in the Santa Ana Mountain Range, California. Orange County Cooperative Mountain Lion Study Final Report. BERGMAN, E. J., M. W. ALLDREDGE, K. A. LOGAN, M. SHUETTE, C. J. BISHOP, AND D. J. FREDDY. 2007. Study Plan-Pilot evaluation of predator-prey dynamics on the Uncompahgre Plateau. Wildlife Research Report July: In Press. Colorado Division of Wildlife, Fort Collins. BURNHAM, K. P., AND D. R. ANDERSON. 1998. Model selection and inference: a practical informationtheoretic approach. Springer-Verlag, New York, New York, USA. CAUGHLEY, G. 1978. Analysis of vertebrate populations. John Wiley and Sons, New York. COLORADO DIVISION OF WILDLIFE 2002-2007 STRATEGIC PLAN. 2002. Colorado Department of Natural Resources, Division of Wildlife. Denver. CONOVER, W. J. 1999. Practical nonparametric statistics. John Wiley & Sons, Inc., New York. COOCH, E., AND G. WHITE. 2004. Program MARK- a gentle introduction, 3rd edition. Colorado State University, Fort Collins. CULVER, M., W. E. JOHNSON, J. PECON-SLATTERY, AND S. J. O’BRIEN. 2000. Genomic ancestry of the American puma (Puma concolor). The Journal of Heredity 91:186-197. CURRIER, M. J. P., AND K. R. RUSSELL. 1977. Mountain lion population and harvest near Canon City, Colorado, 1974-1977. Colorado Division of Wildlife Special Report No. 42. HEISEY, D. M., AND T. K. FULLER. 1985. Evaluation of survival and cause specific mortality rates using telemetry data. Journal of Wildlife Management 49:668-674. KENDALL, W.L. 2001. The robust design for capture-recapture studies: analysis using program MARK. Pages 357-360 in R. Field, R. J. Warren, H. Okarma, and P. R. Sievert, editors. Wildlife, land, and people: priorities for the 21st century. Proceedings of the Second International Wildlife Management Congress. The Wildlife Society, Bethesda, Maryland, USA. KOLOSKI, J. H. 2002. Mountain lion ecology and management on the Southern Ute Indian Reservation. M. S. Thesis. Department of Zoology and Physiology, University of Wyoming, Laramie. KREEGER, T. J. 1996. Handbook of wildlife chemical immobilization. Wildlife Pharmaceuticals, Inc., Fort Collins, Colorado. LAUNDRE, J. W., L. HERNANDEZ, D. STREUBEL, K. ALTENDORF, AND C. L. LOPEZ GONZALEZ. 2000. Aging mountain lions using gum-line recession. Wildlife Society Bulletin 28:963-966. LOGAN, K. A., E. T. THORNE, L. L. IRWIN, AND R. SKINNER. 1986. Immobilizing wild mountain lions (Felis concolor) with ketamine hydrochloride and xylazine hydrochloride. Journal of Wildlife Diseases. 22:97-103. _____, L. L. SWEANOR, J. F. SMITH, AND M. G. HORNOCKER. 1999. Capturing pumas with foot-hold snares. Wildlife Society Bulletin 27:201-208. _____, AND L. L. SWEANOR. 2001. Desert puma: evolutionary ecology and conservation of an enduring carnivore. Island Press, Washington, D.C. 130 �_____. 2004. Colorado puma research and development program: population characteristics and vital rates study plan. Colorado Division of Wildlife, Ft. Collins Research Center, Ft. Collins. _____. 2005. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. July: 105-126, Colorado Division of Wildlife, Fort Collins. MURPHY, K., M. CULVER, M. MENOTTI-RAYMOND, V. DAVID, M. G. HORNOCKER, AND S. J. O’BRIEN. 1998. Cougar reproductive success in the Northern Yellowstone Ecosystem. Pages 78-112 in The ecology of the cougar (Puma concolor) in the Northern Yellowstone ecosystem: interactions with prey, bears, and humans. Dissertation, University of Idaho, Moscow. OTIS, D. L., K. P. BURNHAM, G. C. WHITE, AND D. R. ANDERSON. 1978. Statistical inference from capture data on closed animal populations. Wildlife Monographs 62:1-135. OTT, R. L. 1993. An introduction to statistical methods and data analysis. Fourth edition. Wadsworth Publishing Co., Belmont, California. PIERCE, B. K., V. C. BLEICH, AND R. T. BOWYER. 2000. Social organization of mountain lions: does land a tenure system regulate population size? Ecology 81:1533-1543. POLLOCK, K. H., S. R. WINTERSTEIN, C. M. BUNCK, AND P. D. CURTIS. 1989a. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. _____, S. R. WINTERSTEIN, AND M. J. CONROY. 1989b. Estimation and analysis of survival distributions for radio tagged animals. Biometrics 45:99-109. _____, J. D. NICHOLS, C. BROWNIE, AND J. E. HINES. 1990. Statistical inference for capture-recapture experiments. Wildlife Monographs 107:1-97. POJAR, T. M., AND D. C. BOWDEN. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550-560. ROSS, P. I., AND M. G. JALKOTZY. 1992. Characteristics of a hunted population of cougars in southwestern Alberta. Journal of Wildlife Management 56:417-426. STONER, D. C. 2004. Cougar exploitation levels in Utah: implications for demographic structure, metapopulation dynamics, and population recovery. Master of Science Thesis. Utah State University. SWEANOR, L. L., K. A. LOGAN, AND M. G. HORNOCKER. 2000. Cougar dispersal patterns, metapopulation dynamics, and conservation. Conservation Biology 14:798-808. _____, L. L., K. A. LOGAN, J. W. BAUER, W. M. BOYCE, AND B. MILSAP. 2006 in review. Puma-human relationships in a California, USA state park. Biological Conservation. VAN BALLENBERGHE, V. 1983. Rate of increase of white-tailed deer on the George Reserve: a reevaluation. Journal of Wildlife Management 47:1245-1247. WATKINS, B. 2004. Mountain lion data analysis unit L-22 management plan. Colorado Division of Wildlife, Montrose. WILLIAMS, B. K., J. D. NICHOLS, AND M. J. CONROY. 2001. Combining closed and open mark-recapture models: the robust design. Pages 523-554 In Analysis and management of animal populations. Academic Press, New York. WHITE, G. C., D. R. ANDERSON, K. P. BURNHAM, AND D. L. OTIS. 1982. Capture-recapture and removal methods for sampling closed populations. Los Alamos National Laboratory Publication LA-8787NERP. Los Alamos, NM, U.S.A. WORTON, B. J. 1995. Using Monte Carlo simulation to evaluate kernel-based home range estimators. Journal of Wildlife Management 59:794-800. YOUNG, S. P. AND E. A. GOLDMAN. 1946. The puma: mysterious American cat. The Wildlife Institute, Washington, D. C. Prepared by: __________________________________ Kenneth A. Logan, Wildlife Researcher 131 �Table 1. Summary of puma capture efforts with dogs from November 13, 2006 to May 11, 2007, Uncompahgre Plateau, Colorado. No. & type of No. & type of No. & I.D. or type of pumas Month No. puma tracks pumas pursued captured Search founda Days November 8 12 tracks: 3 male, 7 pursuits: 1 3 pumas captured 4 times: 1 male 7 female, 2 cub male, 4 females, (not handledb), F3 twice (not 2 cubs handled once), cub M42 (died) December 16 49 tracks: 7-8 13 pursuits: 3 5 pumas captured 6 times: 1 male male, 19-20 males, 4-5 (not handled), F50, 1 female or female, 22-23 females, 5-6 cubs subadult puma (not handled), cub cub M49 captured twice (not handled once), 1 cub (not handled) 3 pumas captured: 9-10 pursuits: 3 January 19 56-58 tracks: 19 males, 3 females, 1 male (not handled), M51, 1 male, 30 female, female cub (not handled) 3-4 cubs 7-9 cub February 8 4 tracks: 1 male, 3 pursuits: 1 2 pumas captured: 1 female, 2 cub female, 2 cubs cubs M44 & M56 March 14 31-33 tracks: 8 12 pursuits: 4 7 pumas captured: M29, F7, F23 male, 13 female, male, 5 females, (not handled), F24 (not handled), 10-12 cub 3 cubs cubs M43 (not handled), M56 (not handled), & 1 unmarked female cub (not handled) 1 pursuit: 1 male 0 puma captured April 11 23 tracks: 13-16 male, 5-8 female, 2 cub May 2 2 tracks: 2 female 1 pursuit: 1 0 pumas captured female 78 177-178 tracks: 45-47 pursuits: 22 captures of 16 individuals: 7 TOTALS 51-55 male, 7712 males, 18-19 pumas captured for the 1st time81 female, 45-50 females, 15-17 M49, F50, M51, M56, & 1 female cub cubs or subadult (not handled) & 2 female cubs (not handled), 1 adult male caught twice (not handled), 12 marked pumas were recaptured 15 times (including 4 caught for the 1st time this year). a Puma hind-foot tracks with plantar pad widths >50 mm wide are assumed to be male; ≤50 mm are assumed to be female. b Pumas are not handled for a variety of safety reasons: tree to dangerous to climb for researchers, puma treed near river, creek or cliff, puma might fall from tree after drug induction. 132 �Table 2. Summary of puma capture efforts with dogs, December 2004 to May 2007, Uncompahgre Plateau, Colorado. Pursuit effort Puma capture Effort to capture a puma for Period Track effort the first time detection effort Dec. 2, 109/78 = 1.40 35/78 = 0.45 14/78 = 0.18 11 pumas captured for first 2004 tracks/day pursuit/day capture/day time (minus M1, F3, & large to female) May 12, 78/35 = 2.23 78/14 = 5.57 11/78 = 0.14 capture/day 2005 day/pursuit day/capture 78/11 = 7.09 day/capture 7 pumas captured for first time 14/82 = 0.17 149/82 = 1.82 43/82 = 0.52 Nov. 21, 7/82 = 0.08 capture/day capture/day tracks/day pursuit/day 2005 to 82/7 = 11.71 day/capture 82/14 = 5.86 82/43 = 1.91 May 26, day/capture day/pursuit 2006 7 pumas captured for first time 22/78 = 0.28 45/78 to 47/78 177/78 to Nov. 13, 7/78 = 0.09 capture/day capture/day = 0.58-0.60 182/78 = 2.272006 pursuit/day 2.33 to tracks/day May 11, 78/7 = 11.14 day/capture 78/22 = 3.54 78/47 to 78/45 2007 day/capture = 1.66-1.73 day/pursuit Table 3. Adult and subadult pumas captured for the first time, sampled, tagged, and released from December 2006 to January 2007, Uncompahgre Plateau, Colorado. Puma Sex Estimated Mass Capture Capture Location I.D. Age (mo.) (kg) date method F50 F 25-27 31 12-14-06 Dogs West Fork Dry Creek M51 M 44-49 61 01-07-07 Dogs Lindsay Canyon F52 F 18-20 38 01-10-07 Cage trap Paco-Chu-Puk Campground, Ridgway State Park F54 F 30 36 01-12-07 Cage trap Pleasant View, Pleasant Valley M55 M 24-36 62 01-21-07 Cage trap Dallas Creek, Pleasant Valley 133 �Table 4. Pumas that were captured with aid of dogs, but were not handled and marked at that time for either safety reasons or they escaped, November 2006 to May 2007, Uncompahgre Plateau, Colorado. Puma Age Capture Location Comments sex stage date Male adult 11-13-06 East Fork Unmarked puma climbed difficult spruce tree beside Dry Creek creek. This puma is probably M51 (captured & marked 01-07-07, identified with distinguishing notch in margin of right pinna). Female, adult 11-21-07 Dry Creek Puma F3 climbed dangerous tree adjacent to creek. F3 Basin Unknown cub 12-02-06 Dry Creek Unmarked cub was bayed on edge of high cliff. This sex Basin cub was member of family comprised of an adult female & 2 cubs, which was probably pursued again on 01-25-07. Unknown subadult 12-05-06 Dry Creek This unmarked puma was treed on the same day that Basin we captured & handled cub M49. This puma could sex male or have been M49’s mother or a subadult puma (sex female, uncertain). or adult female Male adult 12-18-06 Lower Unmarked puma climbed difficult fir tree on steep East Fork slope. This puma was probably M51 (captured & Dry Creek marked 01-07-07, identified with distinguishing notch in margin of right pinna). Female cub 01-25-07 Piney Unmarked cub climbed tree. Anesthesia was Creek attempted with pole syringe. Cub jumped from tree, apparently with subcutaneous injection. Cub was pursued unsuccessfully by researchers on foot. Dogs were not released on partially sedated cub for safety reasons. This cub was member of family comprised of an adult female & 2 cubs, which was initially pursued on 12-02-06. Female cub 03-01-07 Dolores Unmarked cub associated with puma F2; was Canyon probably her unmarked cub, sibling of M38. Cub climbed difficult spruce tree adjacent to creek. Female, adult 03-07-07 San Puma F23 climbed a cottonwood tree close to the F23 Miguel San Miguel River. We did not attempt to anesthetize River at F23 to replace her non-functional GPS collar for Pinyon safety reasons. 134 �Table 5. Summary of puma capture efforts with ungulate road-kill baits, puma kills, and cage traps from August 2, 2006 to July 26, 2007, Uncompahgre Plateau, Colorado.a Month No. of Puma activity & capture effort resultsb Sites August 5 Puma scavenged a mule deer carcass on 08-07-06. Cage trap set. Black bear caught & released. Puma F16 was in the area. Puma did not return. September 4 No puma activity detected. October 10 Male puma scavenged a mule deer carcass 10-27-06. Cage trap set & monitored 10-27 to 28-06. Puma did not return. November 12 Male puma scavenged a mule deer carcass on 11-06-06. Set & monitored cage trap 11-06 to 10-06. Puma F16 walked around cage trap, but did not enter on 11-09-06. January 3 Subadult female F52 captured at adult female mule deer she killed 01-10-07, Ridgway State Park. Adult female F54 and her cub F53 were captured at an adult female mule deer kill 01-12-07, Pleasant Valley. Adult male puma M55 was captured at a mule deer fawn kill 01-21-07, Dallas Creek. March 7 Adult male puma M29 was temporarily caught in cage trap set on an adult elk cow he had killed 03-15-07. But, M29 escaped out of back of the trap as researchers arrived. An ear-tagged male cub of puma F3 was observed feeding on a mule deer carcass 03-26-07; F3’s family was in the vicinity. Female puma scavenged on a mule deer carcass 03-29-07. Cage trap was set. Puma F30 was recaptured, and her VHF collar was changed to a GPS collar. April 7 Male puma, probably M29, scavenged a mule deer carcass ~04-01-07. Female puma walked by same carcass (as above) on ~04-02-07, but did not feed. During ~04-05 to 08-07 a puma completely scavenged the same mule deer carcass. Puma F3 and her cubs consumed a mule deer carcass 04-06 to 10-07. Puma F30 consumed a mule deer fawn carcass 04-08-07. Puma F30 consumed a mule deer carcass 04-24-07. Male puma scavenged on mule deer carcass 04-10-07. Cage trap set. Male puma M55 walked up to cage trap (GPS data), but did not enter. Pumas F30 & M55 fed on a mule deer carcass 04-17 to 20-07. Female puma scavenged a mule deer carcass 0423-07. Cage trap set. Puma F8 was recaptured; her non-functional GPS collar was replaced with a VHF collar. Male puma M55 scavenged a mule deer carcass 04-30-07. Female puma scavenged a mule deer carcass. 04-27-07. Cage trap set. Puma F16 recaptured, and her GPS collar was changed with a new GPS collar. May 6 Male puma M55 scavenged on a mule deer carcass 05-08-07. Female puma killed a mule deer doe 05-10-07. Cage trap set. Puma did not return or did not enter the trap. Male puma M55 scavenged a mule deer carcass 05-22-07. June 5 Male puma M55 scavenged on an elk carcass 06-06-07. July 4 No puma activity detected. a We used 64 road-killed mule deer, 7 road-killed elk, 3 puma-killed mule deer, and 1 puma-killed elk at 26 different sites. Of the road-killed ungulate baits, 16 of 71 (22.5%) were scavenged by pumas. b Eight pumas were captured, including: 2 adult males (M29, M55), 4 adult females (F54, F30, F8, F16), 1 subadult female (F52), and 1 female cub (F53). 135 �Table 6. Pumas recaptured with dogs and cage traps, November 2006 to April 2007, Uncompahgre Plateau, Colorado. Puma I.D. F3 F3 M42 Recapture date 11-21-06 11-22-06 11-27-06 M49 M44 M43 M56 F7 F23 M29 F24 M29 F30 F8 F16 Mass kg Observed 41 4.8 Estimated Age (mo.) 63 63 3.5 Capture Method Dogs Dogs Dogs 12-12-06 02-14-07 03-01-07 03-01-07 03-03-07 03-07-07 03-05-07 03-22-07 03-27-07 03-29-07 Observed Observed Observed Observed 33 Observed Observed Observed 60 37 5 6 6.5 6.5 88 31 91 71 91 44 Dogs Dogs Dogs Dogs Dogs Dogs Cage trap Dogs Dogs Cage trap 04-23-07 04-28-07 37 48 46 51 Cage trap Cage trap Process None Changed GPS collar Cub died due to infection & stress None None None None Changed GPS collar None None None Changed GPS collar Changed VHF collar to GPS collar Changed GPS collar Changed GPS collar Table 7. Puma cubs sampled July 2006 to August 2007 on the Uncompahgre Plateau Puma Study area, Colorado. Cub I.D. Sex Estimated birth datea Estimated age at capture (days) Mass (kg) Mother Estimated age of mother at birth of this litter (mo) M38 M July 29, 2006 41 2.9 F2 67 Unm.b F “ 215 Observed “ “ M39 M August 13, 2006 29 1.9 F8 37 F40 F “ “ 1.8 “ “ F41 F “ “ 1.3 “ “ M42 M “ “ 1.5 “ “ M43 M August 13, 2006 33 2.4 F7 82 M44 M “ “ 2.5 “ “ F45 F “ “ 1.7 “ “ M56c M “ 185 9.6 “ “ M46 M September 17, 2006 31 2.2 F3 61 M47 M “ “ 2.2 “ “ M48 M “ “ 2.5 “ “ M49c M July 1, 2006 158 10.0 F50 21 c F53 F July 1, 2006 196 15.0 F54 24 F57 F April 16, 2007 35 2.3 F25 94 M58 M May 24, 2007 34 2.3 F16 52 F59 F “ “ 2.2 “ “ M60 M “ “ 2.0 “ “ M61 M “ “ 1.7 “ “ M62 M July 14, 2007 34 1.8 F24 75 M63 M “ “ 2.1 “ “ M64 M “ “ 1.7 “ “ M65 M “ “ 1.9 “ “ F66 F July 17, 2007 37 2.1 F30 48 M67 M “ “ 3.0 “ “ M68 M “ “ 3.3 “ “ a Estimated age of cubs sampled at nurseries is based on the starting date for GPS location foci for mothers at nurseries. b This unmarked female cub was captured on 03-01-07 in association with adult female puma F2. This cub could be the sibling of cub M38, offspring of F2, which we were not able to capture previously with M38 (its tracks were observed). c Estimated ages of M49 and F53 were based on morphometric comparisons with known-age cubs (Logan and Sweanor 2001, and unpublished data). 136 �Table 8. Puma reproduction, Uncompahgre Plateau, Colorado, 2004-2007. Consort pairs and estimated agesa Female Age (mo.) Male Age (mo.) Dates pairs consortedb Estimated birth datec Estimated birth interval (mo.) Estimated gestation Observed number of cubsd F2 53 05/28/05 3 F2 67 07/29/06 14.0 2 F3 36 08/01/04 1 F3 50 M6 37 06/22-24/05 09/26/05 13.8 93-95 2 F3 62 09/17/06 11.7 3 F7 67 05/19/05 2 F7 82 08/13/06 14.9 4 F8 24 06/26/05 2 F8 37 08/13/06 13.4 4 F16 32 09/22/05 4 F16 52 05/24/07 19.9 4 F23 21 05/30/06 3 F24 75 M29 92 04/12-15/07 07/14/07 90-93 4 F25 74 08/01/05 1 F25 94 04/16/07 20.5 1 F28 36 06/09/06 2 F28 48 M29 88 12/27-29/06 03/30/07 11.7 92-93 ≥2 tracks F30 48 M55 34 04/16-20/07 07/17/07 88-92 3 F50 21 07/01/06 1 F54 24 07/01/06 1 a Ages of females were estimated at litter birth dates. Ages of males were estimated around the dates the pairs consorted. b Consort pairs indicate pumas that were observed together based on GPS data. c Estimated birth dates were indicated by GPS data of mothers at nurseries. d Observed number of cubs do not represent litter sizes as some cubs were observed when they were 5 to 6 months old after postnatal mortality could have occurred in siblings. Only cub tracks were observed with F28. 137 �Table 9. Summary for individual adult puma survival and mortality, December 2004 to July 2007, Uncompahgre Plateau, Colorado. Puma I.D. M1 M4 M5 M6 M27 M29 M32 M51 M55 F2 F3 F7 F8 F16 F23 F24 F25 F28 F30 F50 F54 Monitoring span 12-08-04 to 08-1606 01-28-05 to 12-2805 08-01-06 to 07-3107 02-18-05 to 02-2206 03-10-06 to 07-0506 04-14-06 to 07-3107 04-26-06 to 07-3107 01-07-07 to 07-3107 01-21-07 to 07-3107 01-07-05 to 07-3107 01-21-05 to 07-3107 02-24-05 to 07-3107 03-21-05 to 07-3107 10-11-05 to 07-3107 02-05-06 to 03-0707 01-17-06 to 07-3107 02-08-06 to 07-3107 03-23-06 to 07-3107 04-15-06 to 07-3107 12-14-06 to 03-2607 01-12-07 to 07-3107 No. days 616 333 365 369 117 Status: Alive/Lost contact/Dead; Cause of death Lost contact― failed GPS/VHF collar. M1 ranged principally north of the study area. Dead; killed by a male puma. Estimated age at death 37―45 months. Alive. Born on study area; offspring of F3. He was independent of F3 by 13 months old, and dispersed from his natal area at about 14 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months. Lost contact― failed GPS/VHF collar. 473 Lost contact― failed GPS/VHF collar. M27 ranged principally north of the study area. Alive. 461 Alive. 205 Alive. 191 Alive. 935 Alive. 921 Alive. 887 Alive. 862 Alive. 658 Alive. 396 Lost contact― failed GPS/VHF collar. 560 Alive. 538 Alive. 495 Alive. 472 Alive. 102 Died of natural causes; exact agent unknown. 200 Alive. 138 �Table 10. Summary of subadult puma survival and mortality, December 2004 to June 2006, Uncompahgre Plateau, Colorado. Puma Monitoring No. days Status: Alive/Survived to adult stage/ Lost contact/Dead; I.D. span Cause of death M5 09-16-05 to 308 Alive; independent and dispersed from natal area at 13 06-30-06 months old. Established adult territory on northwest slope of Uncompahgre Plateau. M11 06-21-06 to 176 Lost contact. Independent at 13 months old. Dispersed from 12-14-06 natal area at 14 months old. Last location in Dolores River valley Dec. 14, 2006. F23 01-04-06 to 31 Alive; survived to adult stage; gave birth to first litter at ~21 02-04-06 months old. M31 04-19-06 to 7 Lost contact. Probable disperser. M31’s estimated age at 04-26-06 capture was 25 months, at the lower margin of puberty for puma. He may have been a dispersing subadult, and could have moved away from the study area. M49 03-26-07 to 127 M49 was orphaned at about 9 months old, when his mother 07-31-07 F50 died of natural causes. He dispersed from his natal area at about 10 months old and has been ranging on the northeast slope of the Uncompahgre Plateau. F52 01-10-07 to 125 Lost contact. Dispersed from study area as a subadult. F52’s 05-15-07 last location was Crystal Creek, a tributary of the Gunnison River east of the Black Canyon. 139 �Table 11. Summary for individual puma cub survival and mortality, December 2004 to 2007, Uncompahgre Plateau, Colorado. Status: Alive/Survived to subadult stage/ Lost Estimated survival Age to last Puma Estimated contact/Disappeared/ Dead; Cause of death span from 1st monitor date I.D. Age at alive or at capture capture to fate or death (days) (days) last monitor date M5 183 02-04-05 to 07-31-07 907 F9 F10 31 31 329 207-246 M11 31 06-27-05 to 4-19-06 06-27-05 to 11-2005― 12-29-05 06-27-05 to 12-14-06 F12 42 07-01-05 to 12-0805― 01-26-06 245-294 F13 F14 42 26 07-01-05 to 08-28-05 07-22-05 to 02-0706― 03-10-06 100 226-257 M15 F17 26 34 07-22-05 to 06-06-06 10-26-05 to 08-18-06 345 330 F18 34 301-308 M19 M20 F21 M22 34 34 37 37 M26 183 10-26-05 to 0720―27-06 10-26-05 to 07-27-06 10-26-05 to 05-24-06 11-02-05 to 06-30-06 11-02-05 to 12-2105― 12-22-05 02-08-06 to 03-21-06 Mother I.D. F3 306 244 277 86-87 Survived to subadult stage by 09-16-05; independent at ~13 mo. old. Dispersed from natal area by 09-29-05 at 14 mo. old . Lost contact― shed radiocollar 04-19-06―04-26-06. Lost contact― shed radiocollar 08-10-05; last tracks of F10 with mother F2 & siblings F9 & M11 observed 11-20-05. F10 disappeared by 12-30-05. Survived to subadult stage by 06-21-06, independent at 13 mo. old. Dispersed from natal area by 07-11-06 at 14 mo. old. Lost contact― shed radiocollar 07-28-05―08-01-05. Tracks of F12 found in association with mother F7 on 12-08-05. F12 disappeared by 01-27-06 when she was not visually observed with F7, and her tracks were not seen in association with F7’s tracks. Dead; killed and eaten by a puma (sex unspecified). Lost contact― shed radiocollar 01-20-06―01-25-06. Tracks of F14 were observed with tracks of mother F8 & sibling M15 on 0207-06. Disappeared by 03-11-06, only tracks of F8 & M15 were found. Lost contact― shed radiocollar 06-06-06―06-14-06. Dead. Lost contact― shed radiocollar 06-06-06―06-14-06. Killed by a car on highway 550 on 08-18-06. Probably dependent on F16. Dead; probably killed by another puma. Multiple bite wounds to skull. 10 mo. old. Born 9/22/05 Lost contact― shed radiocollar 07-27-06―08-02-06. Lost contact― shed radiocollar 05-24-06―05-25-06. Alive. Dead; killed and eaten by male puma 12-21-05―12-22-05. F16 F16 F3 F3 224 Lost contact― shed radiocollar 03-21-06―03-24-06. F25 535 140 F2 F2 F2 F7 F7 F8 F8 F16 F16 �Puma I.D. Estimated Age at capture (days) Estimated survival span from 1st capture to fate or last monitor date Age to last monitor date alive or at death (days) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death F33 31 06-30-06 to 07-31-06 62 F34 31 06-30-06 to 07-31-06 62 F35 F36 31 29 06-30-06 to 07-07-06 07-08-06 to 07-28-06 38 74 M37 29 07-08-06 to 07-28-06 74 M38 M39 41 29 165 9 226 M40 29 9 226 Lost contact― shed radiocollar by 09-20-06, but seen alive on that date. Tracks of 2 cubs following F8 on 04-25-07. F8 F41 29 09-08-06 to 02-20-07 09-11-06 to 09-20-06 to 04-25-07 09-11-06 to 09-20-06 to 04-25-07 09-11-06 to 10-05-06 Dead. Probably killed and eaten by a male puma 08-01 to 03-06. GPS data on M29 indicate he was not involved. Dead. Probably killed and eaten by a male puma 08-01 to 03-06. GPS data on M29 indicate he was not involved. Dead; research-related fatality.a Dead. Killed and eaten by a male puma 08-22-06. GPS data on M29 indicate he was not involved. Dead. Killed and eaten by a male puma 08-22-06. GPS data on M29 indicate he was not involved. Lost contact― shed radiocollar found 03-06-07. Lost contact― shed radiocollar by 09-20-06, but seen alive on that date. Tracks of 2 cubs following F8 on 04-25-07. 24 F8 M42 M43 M44 29 33 33 09-11-06 to 11-27-06 09-15-06 to 03-01-07 09-15-06 to 02-14-07 77 167 152 F45 33 09-15-06 to 5-20 to 23-07 280-283 M46 31 10-18-06 to 12-15-06 58 M47 31 10-18-06 to 12-15-06 58 M48 31 10-18-06 to 12-15-06 58 Lost Contact― shed radiocollar or died (blood on collar) between 10-05-06 (last live signal) & 10-13-06 (collar found). Dead; research-related fatality.b Treed, visually observed 03-01-07. Treed, visually observed 02-14-07; sibling (?) M56 also captured, sampled, & marked for 1st time. Dead. Multiple puncture wounds on braincase― parietal & occipital regions; consistent with bites from coyote. F45 switched families, moving from F7 to F2 about 12-19 to 20-06. Last date F45 was with F2 was 04-17-07. Lost contact― shed radiocollar. Tracks of all cubs observed following F3 12-15-06. Lost contact― shed radiocollar. Tracks of all cubs observed following F3 12-15-06. Lost contact― shed radiocollar. Tracks of all cubs observed following F3 12-15-06. 141 Mother I.D. F23 F23 F23 F28 F28 F2 F8 F8 F7 F7 F7 F3 F3 F3 �Puma I.D. Estimated Age at capture (days) Estimated survival span from 1st capture to fate or last monitor date Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Age to last monitor date alive or at death (days) Mother I.D. M49 153 12-05-06 to 07-31-07 238 M49 was orphaned when his mother died on about 03-26-07. F50 F53 183 01-12-07 to 02-23-07 42 Lost contact― shed radiocollar 2-23-07. F54 M56c 183 02-14-07 to03-01-07 15 Lost contact― shed radiocollar 2-27-07. M56 observed 03-01-07. F7 (?) F57 35 05-21-07 to 06-06-07 16 Lost contact― shed radiocollar 06-07-07. Live mode 06-06-07. F25 M58 34 06-27-07 Not radio-collared. F16 F59 34 06-27-07 to 08-21-07 55 Alive. F16 M60 34 06-27-07 to 07-11-07 14 Dead; research-related mortality.d F16 F61 34 06-27-07 to 06-29-07 2 Radiocollar malfunction. F16 M62 34 08-17-07 Not radio-collared. F24 M63 34 08-17-07 Not radio-collared. F24 M64 34 08-17-07 Not radio-collared. F24 M65 34 08-17-07 Not radio-collared. F24 F66 37 08-23-07 Radio-collared. F30 M67 37 08-23-07 Not radio-collared. F30 M68 37 08-23-07 Not radio-collared. F30 a Cub F35 probably starved between 06-30-06 & 07-07-06 after the transmitter on the expandable collar got in its mouth. b Cub M42 died after being captured by dogs, probably from stress of capture associated with severe infection of laceration under right foreleg caused by expandable radiocollar. c Cub M56 was captured in association with F7 and her cubs M43 and M44. He may have been missed at the nursery when M43 and M44 were initially sampled and marked. d Cub M60 died probably of starvation. The expandable radiocollar was around the neck and right shoulder, possibly restricting movement. 142 �Table 12. Numbers of GPS locations for pumas captured on the Uncompahgre Plateau, Colorado, December 2004 to July 2007. Puma I.D. M1 M4 M6 M27 M29 M51 M55 F2 F3 F7 F8 F16 F23 Sex Age stage Dates monitored a No. locations M M M M M M M F F F F F F F24 F25 F28 F30 F50 F52 F54 F F F F F F F adult adult adult adult adult adult adult adult adult adult adult adult subadult, adult adult adult adult adult adult subadult adult 12-08-04 to 07-20-06 01-28-05 to 12-28-05 02-18-05 to 11-23-05 03-11-06 to 06-21-06 04-14-06 to 07-30-07 01-07-07 to 07-30-07 01-21-07 to 07-22-07 01-07-05 to 07-11-07 01-21-05 to 07-30-07 02-24-05 to 07-30-07 03-21-05 to 10-04-06 10-12-05 to 06-12-07 01-04-06 to 02-04-06 02-05-06 to 07-17-06 01-17-06 to 07-25-07 02-09-06 to 07-02-07 03-24-06 to 07-11-07 03-30-07 to 07-25-07 12-14-06 to 03-26-07 01-10-07 to 05-08-07 01-12-07 to 07-20-07 1,864 910 926 316 1,165 630 558 2,759 2,474 2,401 1,516 1,797 113 511 1,816 1,408 1,394 381 361 383 615 Acquisition rate average, range, nb 76, 69―84, 14 70, 57―84, 10 84, 73―93, 9 77, 67―84, 3 69, 56―81,13 76, 66―87, 6 79, 68―91, 6 75, 43―91, 30 78, 55―90, 24 68, 26―92, 27 67, 41―81, 17 73, 41―90, 23 79, 45―92, 6 82, 65―93, 18 69, 55―87, 16 74, 53―89, 16 83, 58―94, 4 87, 76―94, 4 83, 70―92, 3 82, 77―86, 6 a GPS collars on pumas are remotely downloaded at approximately 1-month intervals. The last date in Dates monitored includes last location from the last GPS data download for an individual puma in this report. b n = number of remote downloads. Table 13. Estimated use areas of GPS-collared pumas during November through March, Uncompahgre Plateau, Colorado.a Puma I.D. No. locations Time span No. months 95% Fixed kernel (km2) 50% Fixed kernel (km2) F2 F3 F7 F8 F16 F24 F25 F28 F50 M1 M29 M51 151 130 114 147 144 150 147 146 103 149 97 85 11-01-06 to 03-31-07 11-22-06 to 03-31-07 11-01-06 to 03-31-07 11-01-05 to 03-31-06 11-01-06 to 03-31-07 11-01-06 to 03-31-07 11-01-06 to 03-31-07 11-01-06 to 03-31-07 12-14-06 to 03-26-07 11-01-05 to 03-31-06 11-01-06 to 03-31-07 01-07-07 to 03-31-07 5 4.3 3.9b 5 5 5 5 5 3.4 5 3.4c 2.8 78.6 138.9 66.7 33.7 53.6 117.7 52.0 61.8 70.0 1,132.7 349.1 231.0 13.3 12.2 10.6 5.4 7.0 18.9 6.5 6.3 15.8 302.3 25.0 31.0 a 100% Minimum convex polygon (km2) 102.3 164.0 66.8 43.3 59.9 148.9 79.8 105.4 91.2 779.8 379.0 281.2 Use areas were estimated by using the Animal Movement extension in ArcView 3.2. One location per day was randomly chosen from up to 4 locations fixed per day per puma to reduce autocorrelation. b Due to GPS collar failure, GPS locations were not fixed for F7 from 01-30 to 03-02-07. C Due to GPS collar failure, GPS locations were not fixed for M29 from 02-09 to 03-26-07. 143 �Table 14. VHF-radio-collared independent pumas on the Uncompahgre Plateau, Colorado, 2007. Puma I.D. M5 Sex Age stage Dates monitored No. locations M F8 F30 M31 M32 F F M M Subadult Adult Adult Adult Subadult Adult 09-16-05 to 07-31-06 08-01-06 to 07-30-07 04-23-07 to 07-30-07 04-15-06 to 03-29-07 04-09-06 to 04-26-06 04-26-06 to 07-30-07 36 37 14 43 2 50 Table 15. Summary of puma mother and cub associations by distance (m) during airplane flights, November through March each winter. Monitoring period Month Nov. 9, 2005 to March 29, 2006 Nov. Dec. Jan. Feb. Mar. Nov. 7, 2006 to March 22, 2007 Nov. Dec. Jan. Feb. Mar. Totals Totals No. flights No. puma familiesa Ages of cubs (mo.) No. observations with mothers & cubs ≤520 m apart 3 4 5 4 2 18 4 4 5 4 3 20 4 4 4 5 5 4―5 4 4 3 4 1 1―4 2―6 3―7 4―8 5―9 6―10 2―10 2―3 2―5 4―6 5―7 8 2―8 10 16 16 16 9 67 10 11 9 9 2 41 No. observations with mothers & cubs >520 m apart 2 4 4 2 0 12b 1 1 3 2 1 8c a All puma mothers wore GPS-radiocollars. At least 1 cub in the litter wore a VHF radiocollar. b Mean = 1,060 m, SD = 325.99, range = 650―1,600. C Mean = 1,120 m, SD = 1,214.40, range = 616―4,101. Table 16. General results of puma GPS cluster investigations pilot project, October 2006 to April 2007, Uncompahgre Plateau, Colorado. Cluster Types Investigated S1 Non-random S1 Random S2 Random S3 Random S4 Random S5 Random Totals No. 63 84 11 29 30 40 257 Animals found at all clusters Mule deer Elk Beaver Coyote Total 144 No. 63 58 1 2 124 Animals found at random clusters Mule deer Elk Coyote Total No. 33 31 2 66 �Table 17. Sex and age classes of mule deer found at puma GPS cluster investigations, October 2006 to April 2007, Uncompahgre Plateau, Colorado. Sex & age of mule deer Fawn Yearling 2+ year Unknown age 1+ yr. Unknown age Totals Female 0 1 6 1 0 8 All clusters Male Unknown 1 18 6 3 6 1 1 6 0 13 14 41 Female 0 1 5 1 0 7 Random clusters Male Unknown 1 10 3 2 1 1 1 4 0 3 6 20 Table 18. Sex and age classes of elk found at puma GPS cluster investigations, October 2006 to April 2007, Uncompahgre Plateau, Colorado. All clusters Random clusters Sex & age of elk Female Male Unknown Female Male Unknown Calf 3 1 18 1 0 10 Yearling 13 2 3 7 1 3 2+ year 4 3 4 1 2 3 Unknown age 1+ yr. 0 0 4 0 0 2 Unknown age 0 0 3 0 0 1 Totals 20 6 32 9 3 19 145 �GOAL: Strategies, Information, & Tools for Managing Healthy, Self-sustaining Puma Populations in Colorado Puma Habitat Puma Population Effects of Harvest & Other Mortality Movements & Corridors Population Dynamics: Sex & Age, Vital Rates, Mortality, Population Growth Rates Vulnerability to Harvest Human Development Habitat Use Effects of Translocation Domestic Animals Puma― Human Relationships Deer, Elk, Other Natural Prey & Species of Concern Effects of Predation Effects of Evasive Conditioning Map Prey Distribution Habitat Models Population Models Puma Prey Puma―Prey Relationships Models Estimation Methods for Monitoring Habitat Maps Figure. 1. An ecologically-based conceptual model of the Colorado Puma Research Program that provides the contextual framework for this and proposed puma research in Colorado. Gray-shaded shapes identify areas of research addressed by this report for the puma management goal (at top). 146 �! ( ! ( ! ( ! ( ! ( Clifton County Boundary Highways Study Area ! ( ! ( ! ( ! ( ! ( Delta ! ( M1 ! ( UF UF M32 F50 ! ( UF Montrose F8 M27 M51 F3 F23 M29 ! ( ! ( F7 F28 F2 F24 F30 M6 M55 F16 F25 F54 ! ( Norwood ! ( Ridgeway UF ! ( UM ! ( 0 5 10 20 30 40 Kilometers ! ( Figure 2. Schematic of home ranges of GPS-collared (polygons) and non-collared (ellipses) independent pumas (adults and subadults), intended to show the minimum count and location of independent pumas detected on the study area during November to May period, 2006-2007, Uncompahgre Plateau, Colorado. M & F signify male & female, followed by the identification number of the puma. UF and UM signify uncollared and unsampled female and male pumas, respectively. 147 �5 No. Litters 4 3 - + - - - -- - - I - 2 - + - - - -- - - I 1 -+-------- ~ ~ -------, - - - ,~ --------I - - - r----- - - - - - - - ~- - - - - - , - - - - - - - -~ --I Ja n. Fe b. M ar . Ap r. M ay Ju ne Ju ly Au g. Se p. O ct . No v. De c. 0 I □ Births 2005-07 ■ Births 1983-87 Figure 3. Puma births (n = 20 litters) detected by month during the current research effort, 2005 to 2007, and during the earlier effort by Anderson et al. (1992), 1983 to 1987 (n = 10 litters). Age structure of adult pumas captured and sampled on the Uncompahgre Plateau, Colorado, to March 31, 2007. No. Puma 4 3 2 ~.- ~ 1 ~ - ~ Female ..... Male 0 ~ -~ ~ I ~ 11 I 2 to 3 >3 to >4 to >5 to >6 to >7 to >8 to >9 to 10+ 4 5 6 7 8 9 10 Age (Years) Figure 4. Age structure of adult pumas captured and sampled on the Uncompahgre Plateau, Colorado, on March 31, 2007, after 3 winters (Nov. through Mar., 2004-05 to 2006-07) of protection from sporthunting mortality. In addition, no other human-caused mortalities have been documented in the GPS/radio-collared sample of adults. This age structure assumes that puma M1, M6, and M27 (which had non-functional GPS collars) were alive. Evidence was found on the ground that indicated that all 3 of those males were alive. Pumas M1, M5, and M27 range north of the study area and were protected from legal sport-harvest. Mean ± SD of adult female and adult male ages, respectively: 4.90 ± 1.80 yr. (58.82 ± 21.62 mo.), 4.76 ± 1.62 yr. (57.12 ± 19.39 mo.). 148 �APPENDIX I COLLABORATIVE PROJECT ON DISEASE SURVEILLANCE IN WILD FELIDS. College of Veterinary Medicine and Biomedical Sciences Department of Microbiology, Immunology & Pathology 1619 Campus Delivery Fort Collins, CO 80523-1619 970-491-6144 (voice) 970-491-0603 (fax) TO: Ken Logan, Mammals Researcher, Colorado Division of Wildlife, Montrose, CO. FROM: Sue VandeWoude, DVM, Associate Professor, DMIP RE: Disease Seroprevalence in UP Pumas DATE: August 26, 2007 Attached please find the consolidated report on infectious disease surveillance for the mountain lion samples you have provided to our laboratory as an adjunct to your CDOW ongoing studies. Our laboratory has performed puma-lentivirus (PLV) antibody screening using a sensitive western blot assay developed in our laboratory and found 13 of 18 samples conclusively positive (72%), with two additional samples inconclusive and one not tested. Dr. Michael Lappin, a veterinary internal medicine specialist with expertise in feline infectious disease has tested a subset of 6 samples for antibodies to Feline Calicivirus (FCV), Feline Herpes Virus (FHV), Feline parvovirus (FPV), Toxoplasma gondii (IgM, indicating recent infection, IgG indicating past exposure), and Bartonella hensalae (the agent associated with cat scratch disease). At least one of six animals tested has been positive for each of these agents. Further results are pending from the remaining samples you have provided for these 5 assays. In addition, Dr. Martin Scriefer at Fort Collins CDC has also tested 6 animals for evidence of antibodies to the agent responsible for plague (Yersinia pestis). Interestingly, 3 of 6 animals demonstrate significant exposure to this agent as well. These specific agents were selected for analysis in order to provide a variety of types of agents (viruses: PLV, FCV, FHV, FPV; bacteria: Bartonella henselae and Yersinia pestis; and coccidian: T. gondii), a variety of modes of transmission (direct intra-specific contact, PLV; direct contact with domestic cats, FCV, FHV, FPV; arthropod transmission, B. henselae, Y. pestis; prey ingestion, T. gondii, Y. pestis). Further, at least three of these agents (PLV, FCV, B. henselae) result in chronic infections, allowing the possibility of determining genetic relatedness among organisms isolated from different individuals, and three of these agents (B. henselae, Y. pestis, T. gondii) are also potential zoonotic agents. As you are aware, our laboratory has recently been awarded a 5 year NSF Ecology of Infectious Disease grant entitled, “The effects of urban fragmentation and landscape connectivity on disease prevalence and transmission in North American felids”, with co-PI Dr. Kevin Crooks, an associate professor in the Warner College of Natural Resources at CSU. The aims of this grant are to model the effects of urbanization and resultant habitat fragmentation on disease dynamics in large carnivore species as described on the following page. The letter of support provided by you and Dave Freddy were pivotal in demonstrating a large cohort of capable and active field collaborators willing to provide samples to support our studies. The mountain lion field work being led by your team, and the newly initiated studies by your colleague, Dr. Mat Alldredge, have provided us with renewed enthusiasm for developing our collaborations to support the goals of our study. We foresee the opportunity to interact in a mutually beneficial partnership to further the goals of all of our studies, and to maximize the information that can be gleaned about these important and ecologically significant species. We anticipate that the data we are generating will be useful for comparative seroprevalence of different geographic populations of bobcats and pumas, and for genetic phenotyping of pathogens to compare relationships among diseases spread by arthropod vectors, domestic cats, feral rodents, and inter-specific contacts. As we discussed during your recent visit to CSU, these samples are most valuable to us if we can receive them directly as quickly as possible after collection. I have provided an SOP providing information about the types of samples that 149 �will be most valuable, and a draft of a ‘permissions’ document that you can use with each sample submission to provide us with guidance for any testing that is permissible on the materials we receive. This latter document will be filed and recorded electronically. We will continue to provide annual updates and communications about any publications that utilize the data resulting from your samples. Again thank you for providing these extremely valuable samples, and we look forward to our continued collaborations. Sincerely, Sue VandeWoude THE EFFECTS OF URBAN FRAGMENTATION AND LANDSCAPE CONNECTIVITY ON DISEASE PREVALENCE AND TRANSMISSION IN NORTH AMERICAN FELIDS PROJECT SUMMARY Sue VandeWoude (co-PI), Kevin Crooks (co-PI), Michael Lappin, Mo Salman, Walter Boyce, Ken Logan, Mat Alldredge, Carolyn Krumm, Don Hunter, Lisa Lyren, Seth Riley, Jennifer Troyer The objective of this study is to model the effects of urbanization and resultant habitat fragmentation on disease dynamics in large carnivore species--ecologically pivotal organisms that are sensitive to human disturbances. Bobcats, puma, and domestic cats will be evaluated simultaneously in three divergent ecosystems: high mountain desert (Colorado), everglades (Florida), and Mediterranean scrub habitat (California). The research will: 1) assess the relationship between habitat fragmentation and prevalence of viral, bacterial, and parasitic pathogens across a gradient of urbanization, 2) use transmission dynamics of selected disease agents as markers of connectivity of fragmented populations, and 3) evaluate the effect of urbanization on the incidence of cross-species disease transmission. The results of this research will give wildlife managers a better understanding of how urbanization affects their local wildlife and assist them in future disease management planning. The combination of a uniquely qualified, broadly based research team with an extensive dataset on large carnivores from across the country presents an unprecedented opportunity to investigate the disease dynamics in these rare and difficult to study species. The research efforts of each regional team will support and provide new insights for all of the regions involved, not simply their own. Training of graduate students in ecology, infectious disease, and epidemiology will be emphasized, as will training for pre- and post-doctoral veterinarians. Results will be made widely available to other scientists, conservation practitioners, and the general public. This research has a tremendous capacity to broadly impact areas of public and post-graduate education, career development for new investigators and persons from underrepresented groups, and to enhance understanding of complex infectious disease ecological problems using extensive multidisciplinary collaborations. 150 �Table 1. Appendix I. Preliminary results of infectious disease surveillance for puma, Uncompahgre Plateau, Colorado. Puma ID UPCO 3 UPCO 7 UPCO 7 UPCO 7 UPCO 8 UPCO 4 UPCO 5 UPCO 6 UPCO 25 UPCO 28 UPCO 29 UPCO 31 UPCO 23 UPCO 27 UPCO 30 UPCO 50 UPCO 51 UPCO 52 UPCO 54 UPCO 55 UPCO 24 T.g.e IgG B.h. Y.p. FPV T.g. e IgM f g + + - + - ++ + - - - + - +++ + Ph P P P P P P 13S, 247645, 4246097 Ih P P P P P P P 3/21/2005 12S, 727808, 4239029 I - - - - + - ++ M 1/28/2005 13S, 257565, 4239606 + - - - - + + I M 2/4/2005 13S, 240577, 4251037 - - + + - + - I M 2/18/2005 13S, 247399, 4254006 + - - - - + - I F 2/8/2006 13S, 258374, 4230480 + P P P P P P P F 3/23/2006 12S, 722868, 4240115 + P P P P P P P M 4/14/2006 12S, 723458, 4242340 + P P P P P P P M 4/19/2006 12S, 746919, 4225441 + P P P P P P P F 1/4/2006 12S, 730188, 4234861 - P P P P P P P M 3/10/2006 12S, 722339, 4245212 - P P P P P P P F 4/15/2006 13S, 248551, 4242095 - P P P P P P P F 12/14/2006 12S, 753639, 4260149 + P P P P P P P M 1/7/2007 13S, 238783, 4252390 + P P P P P P P F 1/10/2007 13S, 258058, 4236260 I P P P P P P P F 1/12/2007 13S, 252688, 4228050 + P P P P P P P M 1/21/2007 13S, 258133, 4228691 + P P P P P P P F 1/17/2006 12S, 737151, 4233273 % Seroprevalance = No. animals positive/Total animals tested * 100 + P P P P P P P 72 33 33 33 0 100 17 50 Sex Capture Date GPS NAD27 U.T.M.: Zone, E, N PLV FCV FHV F 1/21/2005 13S, 241606, 4251510 - +h F 2/24/2005 13S, 246328, 4244230 + F 3/30/2006 13S, 245901, 4247627 F 3/3/2007 F a a b PLV is Puma Lentivirus. b FCV is Feline Calicivirus. c FHV is Feline Herpesvirus. d FPV is Feline Panleukopenia Virus e T. g. is Toxoplasma gondii. f B. h. is Bartonella hensalae. g Y. p. is Yersinia pestis. h Results: + (positive result), P (Pending result), I (Inconclusive result). 151 c d �Colorado Division of Wildlife July 2007 −June 2008 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 3003 1 Federal Aid Project No. N/A : : : : : Division of Wildlife Mammals Research Predatory Mammals Conservation Puma Population Structure and Vital Rates On the Uncompahgre Plateau Period covered: July 1, 2007−June 30, 2008 Author: K. A. Logan. Personnel: K. Logan, B. Bavin, B. Dunne, J. Timmer, V. Yovovich, S. Waters, K. Crane, T. Mathieson, M. Caddy, and T. Bonacquista of CDOW; S. Young, and J. McNamara of U.S.D.A. Wildlife Services; volunteers and cooperators including: private landowners, Bureau of Land Management, Colorado State Parks, Colorado State University and U.S. Forest Service, with supplemental financial support received in previous years from The Howard G. Buffett Foundation and Safari Club International Foundation. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Research continued on puma population characteristics and dynamics on the Uncompahgre Plateau. All capture efforts in 2007-08 resulted in a total of 41 puma captures (9 adult females [1 adult female captured 3 times], 6 adult males [1 adult male captured 2 times], 1 subadult male, and 21 cubs [4 of them captured twice each]). Two adults, 4 subadults, and 16 cubs were captured for the first time. As of July 2008, there were 18 adults, 1 subadult, and 4 cubs marked with active radio-collars. Efforts to capture, sample, and mark pumas with the use of trained dogs extended from November 19, 2007 to April 24, 2008. Those efforts resulted in 77 search days, 217-218 puma tracks detected, 49 pursuits, and 20 puma captures. In 2007-08, capture efforts with ungulate carcasses and cage traps resulted in 1 adult male being captured twice. One cub was captured for the first time with dogs, and 15 cubs were caught the first time by hand. Capture and search efforts from November 2007 through March 2008 enabled us to estimate a minimum of 33 independent pumas detected on the Uncompahgre Plateau study area during that time, including 21 females and 12 males. Preliminary puma population parameters estimated during the past 3.7 years of research, included: population sex and age structure, reproduction rates, and survival rates. Data on puma reproduction rates included: average litter size = 2.810 ± 0.9808 SD, n = 21; average birth interval (mo.) = 17.969 ± 4.748 SD, n = 13; average proportion of adult females producing cubs each year = 0.65 ± 0.0586 SD, n = 12-13 females for 3 yr.; secondary sex ratio = 33:26, consistent with 1:1; and average gestation length (day) = 91.188 ± 2.3443 SD. Puma births occurred March through September. Survival rates for both adult and subadult pumas in this reference period appear to be high, and might reflect the relatively small samples of individual pumas in each age-stage and sex and years. Cub survival ranged from 0.50 (Kaplan-Meier procedure) to 0.56 (binomial model). The main cause of 105 �mortality in the adults and cubs is caused by male pumas. A puma population model was developed for researchers and wildlife managers to assess scenarios of puma harvest management strategies. Results from a set of scenarios and attendant models are presented. Only 1 puma family with a radio-collared mother and cub could be monitored during the winter to assess association distances during aerial locations. The aggregate data gathered during the past 3 winters generally indicate that mothers were usually within 520 m of their cubs during the day. Preliminary comparisons between our current puma research on the Uncompahgre Plateau (3.7 years duration) and results of the Anderson et al. (1992) puma research on the plateau (7 years duration 1981-1988) were made where appropriate. Proposed work includes: continuing to quantify puma population characteristics and vital rates, with an emphasis on increasing sample sizes on radio-monitored adults, subadults, and cubs, and developing a study plan for the next 6 years of research, which will include the treatment period. We will collaborate with colleagues to assess puma health and model and map puma habitat. 106 �WILDLIFE RESEARCH REPORT PUMA POPULATION STRUCTURE AND VITAL RATES ON THE UNCOMPAHGRE PLATEAU, COLORADO KENNETH A. LOGAN P. N. OBJECTIVE Quantify puma population sex and age structure; estimate puma population vital rates, including: reproduction rates of females, age-stage survival rates, and immigration and emigration rates; quantify agent-specific mortality rates; begin puma population modeling process; and plan for the remaining 6 years of the Uncompahgre Plateau Puma Project― all to improve the Colorado Division of Wildlife‘s (CDOW) model-based approach to managing pumas in Colorado. SEGMENT OBJECTIVES 1. 2. 3. 4. 5. Continue gathering data on puma population sex and age structure. Continue gathering data for estimates of puma reproduction rates. Continue gathering data to estimate puma sex and age-stage survival rates. Continue gathering data to estimate agent-specific mortality rates. Develop a puma population model and parameter estimates useful for guiding decisions about the hunting treatment phase of this project, and for the Data Analysis Unit puma management planning process performed by CDOW biologists and managers. 6. Gather data on spatial relationships of puma mothers to their cubs during the Colorado puma hunting season as a preliminary assessment of the vulnerability of puma mothers to sport-hunting harvest. 7. Develop a study plan for remaining 6 years of puma population research on the Uncompahgre Plateau Study Area. 8. Evaluate other data sources that could come from this research that can be developed into other puma research relevant to CDOW biologists and managers. INTRODUCTION Colorado Division of Wildlife managers need reliable information on puma biology and ecology in Colorado to develop sound management strategies that address diverse public values and the CDOW objective of actively managing puma while ― achieving healthy, self-sustaining populations‖(CDOW 2002-2007 Strategic Plan:9). Although 4 puma research efforts have been made in Colorado since the early 1970s and puma harvest data is compiled annually, reliable information on certain aspects of puma biology and ecology, and management tools that may guide managers toward effective puma management is lacking. Mammals Research staff held scoping sessions with a number of the CDOW‘s wildlife managers and biologists. In addition, we consulted with other agencies, organizations, and interested publics either directly or through other CDOW employees. In general, CDOW staff in western Colorado highlighted concern about puma population dynamics, especially as they relate to their abilities to manage puma populations through regulated sport-hunting. Secondarily, they expressed interest in puma―prey interactions. Staff on the Front Range placed greater emphasis on puma―human interactions. Staff in both eastern and western Colorado cited information needs regarding effects of puma harvest, puma population monitoring methods, and identifying puma habitat and landscape linkages. Management needs identified by CDOW staff and public stakeholders form the basis of Colorado‘s puma research program, with multiple lines of inquiry (i.e., projects): 107 �Improve our ability to manage puma hunting with enhanced scientific bases, strategies, and tools ● Puma population characteristics (i.e., density, sex and age structure). ● Puma population dynamics and vital rates (i.e., birth rates, survival rates, emigration rates, immigration rates, population growth rates). ● Field methods and models for assessing and tracking changes in puma populations. ● Relative vulnerability of puma sex and age classes to hunter harvest. Improve our understanding of puma habitat needs and interrelationships of puma management units ● Puma habitat use, movements, and use of landscape linkages. ● Puma recruitment patterns (i.e., progeny, immigration, emigration). ● Models for identifying puma habitat and landscape linkages. Improve our understanding of the puma’s role in the ecology of other species ● Relationships of puma to mule deer, elk, and other natural prey. ● Relationships of puma to species of special concern, e.g., desert bighorn sheep. Improve our understanding of puma-human interactions and abilities to manage them ● Behavior of puma in relation to people and human facilities. ● Puma predation on domestic animals. ● Effects of translocating nuisance pumas. ● Effects of aversive conditioning on pumas. While all projects cannot be addressed concurrently, understanding their relationships to one another is expected to help individual projects maximize their benefits to other projects that will assist the CDOW to achieve its strategic goal in puma management (Fig.1). Management issues identified by managers translate into researchable objectives, requiring descriptive studies and field experiments. Our goal is to provide managers with reliable information on puma population biology and to develop useful tools for their efforts to adaptively manage puma in Colorado to maintain healthy, self-sustaining populations. The highest-priority management needs are being addressed with this intensive population study that focuses on puma population dynamics using sampled, tagged, and GPS/radio-collared pumas. Those objectives include: Describe and quantify puma population sex and age structure. Estimate puma population vital rates, including: reproduction rates, age-stage-specific survival rates, emigration rates, immigration rates. Estimate agent-specific mortality rates. Improve the CDOW‘s model-based management approaches with Colorado-specific data from objectives 1―3. Consider other useful models. Concurrently with the tasks associated with the objectives above, significant progress will be made toward a 5th objective, which will initially be subject to pilot study― develop methods that yield reliable estimates of population abundance (i.e., numbers and density) and attendant annual population growth rates, such as, direct mark-recapture, and DNA genotype capture-recapture. A descriptive study will estimate population parameters in an area that appears typical of puma habitat in western Colorado and will yield defensible population parameters based upon contemporary Colorado data. This study will be conducted in a 5-year reference period (i.e., absence of recreational hunting) to allow puma life history traits to interact with the main habitat factors that appear to influence puma population growth (e.g., prey availability and vulnerability, Pierce et al. 2000, Logan and Sweanor 2001). Contingent upon results in the reference period, a subsequent 5-year treatment period is planned. The treatment period will involve the use of controlled recreational hunting to manage the puma population. 108 �TESTING ASSUMPTIONS AND HYPOTHESES Hypotheses associated with main objectives 1―5 of this puma population research are structured to test assumptions guiding puma management in Colorado. 1. Recreational puma hunting management in Colorado Data Analysis Units (DAUs) is guided by a model to estimate allowable harvest quotas to achieve one of two puma population objectives: 1) maintain puma population stability or growth, or 2) cause puma population decline (CDOW, Draft LDAU Plans, 2004, CDOW 2007). Basic model parameters are: puma population density, sex and age structure, and annual population growth rate. Parameter estimates are currently chosen from literature on studies in western states that are judged to provide reliable information. Background material used in the model assumes a moderate annual rate of growth of 15% (i.e.,}., = I. 15)  for the adult and subadult puma population (CDOW 2007). This assumption is based upon information with variable levels of uncertainty (e.g., small sample sizes, data from habitats dissimilar to Colorado). Parameters influencing  include population density, sex and age structure, female age-at-first-breeding, reproduction rates, sex- and age-specific survival, immigration and emigration. H1: Population parameters estimated during a 5-year reference period (in absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will yield an estimated annual adult plus subadult population growth rate that will match or exceed = 1.15. 2. The key assumption is that the CDOW can manage puma population growth through recreational hunting: for a stable puma population hunting removes the annual increment of population growth (i.e., from current judgments on population density, structure, and Puma harvest rate formulations for DAUs assumes that total mortality (i.e., harvest plus other detected deaths) in the range of 8 to 15% of the harvest-age population (i.e., independent pumas comprised of adults plus subadults) with the total mortality comprised of 35 to 45% females (i.e., adults and subadults) is acceptable to manage for a stable-to-increasing puma population (CDOW 2007). H2: Total mortality of an estimated 15% of the adults and subadults with no more than 45% of the total mortality comprised of females will not result in a decline of the harvest-age segment of the population by the beginning of the next hunting season. 3. To reduce a puma population, hunting must remove more than the annual increment of population growth. For DAUs with the objective to suppress the puma population, the total mortality guide of greater than 15 to 28% of the harvest-age population with greater than 45% comprised of females is suggested (CDOW 2007). H3: Total mortality of an estimated 16% or greater of the harvestable population with greater than 45% females will cause a decline in the abundance of harvest-age pumas (i.e., adults and subadults). Considering limitations (i.e., methods, number of years, assumption violations) to the Colorado-specific studies on puma densities cited above (Currier et al. 1977, Anderson et al. 1992, Koloski 2002), managers assume that puma population densities in Colorado are within the range of those quantified in more intensively studied populations in Wyoming (Logan et al. 1986), Idaho (Seidensticker et al. 1973, Alberta (Ross and Jalkotzy 1992, and New Mexico (Logan and Sweanor 2001). The CDOW assumes density ranges of 2.0―4.6 puma/100 km2 to extrapolate to DAUs to guide the model-based quota-setting process. Likewise, managers assume that the population sex and age structure is similar to puma populations described in the intensive studies. Using capture, mark, re-capture techniques developed and refined during the study to estimate the puma population, the following will be tested: 109 �H4: Puma densities during the 5-year reference period (absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will vary within the range of 2.0―4.6 puma/100 km2 and will exhibit a sex and age structure similar to puma populations in Wyoming, Idaho, Alberta, and New Mexico. 5. The increase and decline phases of the puma population make it possible to test hypotheses related to shifts in the age structure of the population which have been linked to harvest intensity in Wyoming and Utah. H5: The puma population on the Uncompahgre Plateau study area will exhibit a young age structure after hunting prohibition at the beginning of the reference period. During the 5 years of hunting prohibition, greater survival of independent pumas will cause an older age structure in harvest-age pumas (i.e., adults and subadults) as suggested by the work of Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. As hunting is re-instated in the treatment period, the age structure of harvested pumas and the harvest-age pumas in the population will decline as observed by Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. Desired outcomes and management applications of this research include: 1. Quantification of variations in puma population density, sex and age structure, growth rates, vital rates, and an understanding of factors affecting them will aid adaptive puma management by yielding population parameters and tools useful for assessing puma population dynamics, evaluation of management alternatives, and effects of management prescriptions. 2. Testing assumptions about puma populations, currently used by CDOW managers, will help those managers to biologically support and adapt puma management based on Colorado-specific estimated puma population characteristics, parameters, and dynamics. 3. Methods for estimating puma abundance (e.g., capture-mark-recapture) of known reliability will allow managers to ―g round truth‖ modeled populations and estimate effects of management prescriptions designed to achieve specified puma population objectives in targeted areas of Colorado. Ascertaining puma numbers and densities during the project will require development of reliable monitoring techniques based on capture-mark-recapture methods and models. Potential methods include direct and DNA genotype capture-recapture, and assessments of harvest sex and age structure. Study plans to develop and test feasible field and analytical methods will be developed in the future after we have learned the logistics of performing those methods, after we have preliminary data on puma demographics and movements which will inform suitable sampling designs, and if we have adequate funding. 4. This information will be disseminated to citizen stakeholders interested in pumas in Colorado, and thus contribute to informed public participation in puma management. STUDY AREA The study area for the puma population research is on the Uncompahgre Plateau (in Mesa, Montrose, Ouray, and San Miguel Counties. The study area includes about 2,253 km2 (870 mi.2) of the southern halves of Game Management Units (GMUs) 61 and 62, and about 155 km2 (60 mi.2) of the northern edge of GMU 70 (between state highway 145 and San Miguel River). The area is bounded by state highway 348 at Delta, 25 Mesa road and Forest Service road FS503 to Nucla, state highway 97 to state highway 141 to state highway 145 to Placerville, state highway 62 to Ridgeway, U.S. highway 550 to Montrose, and U.S. highway 50 to Delta. The study area seems typical of puma habitat in Colorado that has vegetation cover that varies from the pinon-juniper covered foothills starting from about 1,700 m elevation to the spruce-fir and aspen forests growing to the highest elevations of about 3,000 m. Mule deer (Odocoileus hemionus) and elk (Cervus elaphus) are the most abundant wild ungulates available for puma prey. There are cattle and 110 �domestic sheep raised on summer ranges on the study area. Year-round human residents live along the eastern and western fringe of the area, and there is a growing residential presence especially on the southern end of the plateau. A highly developed road system makes the study area well accessible for puma research efforts. A detailed description of the Uncompahgre Plateau is in Pojar and Bowden (2004). METHODS Reference and Experimental Treatment Periods This research is structured in two 5-year periods: a reference period (years 1―5) and a treatment period (years 6―10). The reference period is expected to cause a population increase phase. The treatment period will involve structured puma management strategies. In both phases, puma population structure, and vital rates will be quantified, and some management assumptions and hypotheses regarding population dynamics and effects of harvest will be tested. Contingent upon results of pilot studies, we will also estimate puma numbers, population growth rates, evaluate enumeration methods, and test other hypotheses (Logan 2004). The reference period, without recreational puma hunting as a major limiting factor, is consistent with the natural history of the current puma species in North America which evolved life history traits during the past 10,000―12,000 years (Culver et al. 2000) that enable pumas to survive and reproduce (Logan and Sweanor 2001). In contrast, puma hunting, with its modern intensity and ingenuity, might have influenced puma evolution in western North America for the past 100 years. Hence, the reference period, years 1―5, will provide conditions where individual puma in this population (of estimated sex and age structure) express life history traits interacting with the environment without recreational hunting as a limiting factor. Theoretically, the main limiting factors will be catchable prey abundance (Pierce et al. 2000, Logan and Sweanor 2001). This should allow researchers to understand basic system dynamics before the treatment (i.e., controlled recreational hunting). In the reference period, all puma in the study area will be protected, except for individual puma that might be involved in depredation on livestock or human safety incidents. In addition, all radio-collared and ear-tagged puma that range in a buffer zone, that includes the northern halves of GMUs 61 and 62, will be protected from recreational hunting. The reference period will allow researchers to quantify baseline demographic data on the puma population to estimate parameters for the CDOW‘s model-based approach to puma management. Moreover, it will allow researchers to develop and test puma enumeration methods when population growth is known to be in one direction― increasing. Without the hunting closure, pilot data for enumeration methods could be confounded by not knowing if the population was increasing, declining, or stable. The reference period will also facilitate other operational needs (because hunters will not be killing the animals) including the marking of a large proportion of the puma population for capture-markrecapture estimates, and the gathering of movement data from GPS-collared puma to help formalize exact sampling designs for enumeration methods. During the treatment period, years 6―10, experimentally structured recreational puma hunting will occur on the same study area using management prescriptions structured from information learned during previous years. Using recreational hunting for the treatment is consistent with the CDOW‘s objectives of manipulating natural tendencies of puma populations, particularly survival, to maintain either population stability or increase and population suppression (CDOW, Draft L-DAU Plans, 2004). Theoretically, puma survival will be influenced mainly by recreational hunting, which will be quantified by agent-specific mortality rates of radio-collared puma. For managers, demonstrating that they can manage puma populations with hunting and achieve the CDOW strategic objective of managing for a healthy, self- sustainable puma population state-wide is important. Dynamics of the puma population may be manipulated (i.e., increase and decline phases) to evaluate hypotheses that are related to effects of hunting (i.e.,: effects of harvest rates, relative 111 �vulnerability of puma sex and age classes to hunting, variations in puma population structure due to hunting), enumeration methods, and puma―prey interactions (i.e., lines of research identified in the Colorado Research Program, Fig. 1). The killing of tagged and collared puma during the treatment period will not hamper operational needs (as it would during the start-up years), because by the beginning of this period, a large majority of independent puma in the population will be marked, and sampling schemes will be formalized. Puma on the study area that may be involved in depredation of livestock or human safety incidences may be lethally controlled. Researchers that find that GPS-collared puma have killed domestic livestock will record such incidents to facilitate reimbursement to the property owner for loss of the animal(s). In addition, researchers will notify the Area Manager of the CDOW if they perceive that an individual puma may be a threat to public safety. Field Methods Puma Capture: Realizing that puma live at low densities and capturing puma is difficult, as a starting point, our logistical aim will be to have a minimum of 6 puma in each of 6 categories (36 total) radio-tagged in any year of the study if those or greater numbers are present. The 6 categories are: adult female, adult male, subadult female, subadult male, female cub, male cub. Our aim is to provide more quantitative and precise estimates of puma demographics than were achieved in earlier Colorado puma studies. This relatively large number of puma might represent the large majority of the puma population on the study area, and will provide the basic data for age- and sex-specific reproductive rates, survival rates, agent-specific mortality rates, emigration rates, and movement data pertinent to sampling designs for various projects. Assuming that the puma population density on the study area is relatively low at the beginning of this study― about 1 adult/100 km2 and the sex ratio is equal (Anderson et al. 1992, Logan and Sweanor 2001:167), then there might be 22 adults, 11 males and 11 females. Also assuming that the total population contains 10% subadults and 34% cubs (Logan and Sweanor 2001), then there might be 4 subadults and 13 cubs with equal sex ratios in a total population of 39 puma. If we achieve our logistical aim in the first 1―2 years (recognizing that the population might grow), then we should be able to quantify population characteristics and vital rates for a majority of the puma population in those years and build upon the tagged number in each subsequent year. Thus, our inferences will pertain to the large majority of the puma population, if not the population on the study area, instead of a relatively small sample of it. We anticipate it may take 2 years to mark the large majority of puma in the population. In addition, the study area is large and will require some time to learn to access it efficiently. Puma capture and handling procedures have been approved by the CDOW Animal Care and Use Committee (file #08-2004). All captured puma will be examined thoroughly to ascertain sex and describe physical condition and diagnostic markings. Age of adult puma will be estimated initially by the gum-line recession method (Laundre et al. 2000) and dental characteristics of known-age puma (Logan and Sweanor, unpubl. data). Ages of subadult and cub puma will be estimated initially based on dental and physical characteristics of known-age puma (Logan and Sweanor unpubl. data). Body measurements recorded for each puma will include at a minimum: mass, pinna length, hind foot length, plantar pad dimensions. Tissue collections will include: skin biopsy (from the pinna receiving the 6 mm biopsy punch for the ear-tags) and blood (30 ml from the saphenous or cephalic veins) for genotyping individuals, parentage and relatedness analyses, and disease screening; hair (from various body regions) and fecal DNA for genotyping tests of field gathered samples. Universal Transverse Mercator Grid Coordinates on each captured puma will be fixed via Global Positioning System (GPS, North American Datum 27). Puma will be captured year-round using 4 methods: trained dogs, cage traps, foot-hold snares, and by hand (for small cubs). Capture efforts with dogs will be conducted mainly during the winter when 112 �snow facilitates thorough searches for puma tracks and the ability of dogs to follow puma scent. The study area will be searched systematically multiple times per year by four-wheel-drive trucks, all-terrain vehicles, snow-mobiles, walking, and possibly horse- or mule-back. When puma tracks ≤1 day old are detected, trained dogs will be released to pursue puma to capture. Puma usually climb trees to take refuge from the dogs. Adult and subadult puma captured for the first time or requiring a change in telemetry collar will be immobilized with Telazol (tiletamine hydrochloride/zolazepam hydrochloride) dosed at 5 mg/kg estimated body mass (Lisa Wolfe, DVM, CDOW, attending veterinarian, pers. comm.). Immobilizing agent will be delivered into the caudal thigh muscles via a Pneu-Dart® shot from a CO2-powered pistol. Immediately, a 3m-by-3m square nylon net will be deployed beneath the puma to catch it in case it falls from the tree. A researcher will climb the tree, fix a Y-rope to two legs of the puma and lower the cat to the ground with an attached climbing rope. Once the puma is on the ground, its head will be covered, its legs tethered, and vital signs monitored (Logan et al. 1986). (Normal signs: pulse ~70―80 bpm, respiration ~20 bpm, capillary refill time ≤2 sec., rectal temperature ~101oF average, range = 95―104oF) (Kreeger 1996). A cage trap will be used to capture adults, subadults, and large cubs when puma can be lured into the trap using road-killed or puma-killed ungulates (Sweanor et al. 2008). Efficiency of the trap might be enhanced by using an automated digital call box that emits puma vocalizations (Wildlife Technologies, Manchester, NH). A cage trap will be set only if a target puma scavenges on the lure (i.e., an unmarked puma, or a puma requiring a collar change). Researchers will continuously monitor the set cage trap from about 1 km distance by using VHF beacons on the cage and door. This allows researchers to be at the cage to handle captured puma within 30 minutes. Puma will be immobilized with Telazol injected into the caudal thigh muscles with a pole syringe. Immobilized puma will be restrained and monitored as described above. If non-target animals are caught in the cage trap, we will open the door and allow the animal to leave the trap. Foot-hold snares will be used to capture adults, subadults, and large cubs only when safe snare sites at puma kills can be located as described by Logan et al. (1999). Snares set at puma kills will be monitored continuously with VHF beacons on the snares from about 1 km distance. We will not set snares at sites where tracks indicate that other mammals (e.g., deer, elk, bear, bighorn sheep, livestock) are also active. Puma will be immobilized with Telazol injected into the caudal thigh with a pole syringe. Vital signs will be monitored during the handling procedures. Efficiency of snares might also be enhanced with the use of an automated call box with puma or prey vocalizations. Small cubs (≤10 weeks old) will be captured using our hands (covered with clean leather gloves) or with a capture pole. Cubs will be restrained inside new burlap bags during the handling process and will not be administered immobilizing drugs. Cubs at nurseries will be approached when mothers are away from nurseries (as determined by radio-telemetry). Cubs captured at nurseries will be removed from the nursery a distance of ~100 m to minimize disturbance and human scent at nurseries. Immediately after handling processes are complete, cubs will be returned to the exact nurseries where they were found (Logan and Sweanor 2001). Marking, Global Positioning System- and Radio-telemetry: Puma do not possess easily identifiable natural marking, such as tigers (see Karanth and Nichols 1998, 2002), therefore, the capture, marking, and GPS- or VHF- collaring of individual puma is essential to a number of project objectives, including estimating vital rates and gathering movement data on puma to formalize designs for developing and testing enumeration methods. Adult, subadult, and cub puma will be marked 3 ways: GPS/VHF- or VHF-collar, ear-tag, and tattoo. The identification number tattooed in the pinna is permanent and cannot be lost unless the pinna is severed. A colored (bright yellow or orange), numbered rectangular (5 cm x 1.5 cm) ear-tag (Allflex USA, Inc., DFW Airport, TX) will be inserted into each 113 �pinna to facilitate individual identification during direct recaptures. Cubs 10 weeks old will be eartagged in only one pinna. Locations of GPS- and VHF-collared puma will be fixed about once per week (as flight schedules and weather allow) from light fixed-wing aircraft (e.g., Cessna 182) fitted with radio signal receiving equipment (Logan and Sweanor 2001). This monitoring will enable researchers to find GPS-collared puma to acquire remote GPS location reports from the ground, monitor the status (i.e., live or dead) of individual puma, and to recover carcasses for necropsy. It will also provide simultaneous location data on mothers and cubs. GPS- and VHF-collared puma will be located from the ground opportunistically using hand-held yagi antenna. At least 3 bearings on peak aural signals will be mapped to fix locations and estimate location error around locations (Logan and Sweanor 2001). Aerial and ground locations will be plotted on 7.5 minute USGS maps (NAD 27) and UTMs along with location attributes will be recorded on standard forms. GPS locations will be mapped using GIS software. Adult and subadult female pumas will be fitted with GPS collars (approximately 400 g each, Lotek Wireless, Canada). Initially, GPS-collars will be programmed to fix and store puma locations at 4 times per day to sample daytime, nighttime, and crepuscular locations (i.e., 0:00, 06:00, 12:00, 19:00). GPS locations for puma will provide precise, quantitative data on puma movements mainly to provide data to formalize study designs, to test assumptions for capture-mark-recapture methods for this project, and to assess the relevance of puma DAU boundaries. The GPS-collars also will provide basic information on puma movements and locations to design other pilot studies in this program on vulnerability of puma to sport-harvest, habitat use, and predation frequency on mule deer and elk. Subadult male pumas will be fitted initially with conventional VHF collars (Lotek, LMRT-3, ~400 g each) with expansion joints fastened to the collars, which allows the collar to expand to the average adult male neck circumference (~46 cm). If subadult male puma reach adulthood on the study area, we will recapture them and fit them with GPS collars. VHF radio transmitters on GPS collars will enable researchers to find those pumas on the ground in real time to acquire remote GPS data reports, facilitate recaptures for re-collaring, and to check on their reproductive and physical status. VHF transmitters on GPS- and VHF-collars will have a mortality mode set to alert researchers when puma have been immobile for 3 to 24 hours so that dead puma can be found to quantify survival rates and agent-specific mortality rates by gender and age. We will attempt to collar all cubs in observed litters with small VHF transmitter mounted on an expandable collar (Wildlife Materials, Murphysboro, Illinois, HLPM-2160, ~50g, or Telonics, Inc., Mesa, Arizona MOD 210, ~100g,) when cubs weigh 2.3―11 kg (5―25 lb). Cubs with mass ≥11 kg can still wear these small expandable collars until they are about 12 months old. Cubs approaching the age of independence (~11―14 mo. old) may be fit with Lotek LMRT-3 VHF collars (~400 g) with expansion links. Cubs will be recaptured to replace collars as necessary. Monitoring radio-collared cubs allow quantification of survival rates and agent-specific mortality rates (Logan and Sweanor 2001). Capture-Mark-Recapture: Capture-mark-recapture methods will be evaluated initially as a pilot study. Capturing and marking puma is time consuming, and would lengthen the time to thoroughly search the study area for capturing and marking puma during capture-recapture occasions needed for population estimation. Therefore, we will capture and mark pumas prior to performing capture-recapture or re-sight occasions using methods such as houndsmen teams. In addition, by marking puma before capturerecapture occasions begin, we will have opportunities to capture female puma at different stages of their reproductive status, and thus reduce the chance that mothers in a stage with suckling cubs and small activity areas are not detected and marked on the study area. After cubs are weaned, the mothers‘ activity area expands (Logan and Sweanor 2001). The probability of females having suckling cubs in winter is 114 �naturally small; that season exhibits the lowest rate of births (Logan and Sweanor 2001). Capturerecapture occasions to estimate the population of independent puma may not begin until we have a large majority of the puma population sampled and marked. Occasions performed at that time will be viewed as a pilot study allowing us to examine the logistics of the field methods, the extent to which model assumptions are met, performance of field methods (e.g., detection differences by sex or life stage as revealed by GPS data on collared puma), and precision of capture-recapture models used to estimate the puma population. Analytical Methods Population Characteristics: Population characteristics each year will be tabulated with the number of individuals in each sex and age category. Age categories, as mentioned, include: adult (puma ≥24 months old, or younger breeders), subadults (young puma independent of mothers, <24 months old that do not breed), cubs (young dependent on mothers, also known as kittens) (Logan and Sweanor 2001). When data allow, age categories may be further partitioned into months (for cubs and subadults) or years (for adults). Reproductive Rates: Reproductive rates will be estimated for GPS- and VHF-collared female pumas directly (Logan and Sweanor 2001). Genetic paternity analysis will be used to ascertain paternity for adult male puma (Murphy et al. 1998). Survival and Agent-specific Mortality Rates: Radio-collared puma will provide known fate data which can be used to estimate survival rates for each age stage using the Kaplan-Meier procedure to staggered entry (Pollock et al. 1989), binomial survival model (Williams et al. 2001:343-344), or analyzed in program MARK (White and Burnham 1999, Cooch and White 2004). Agent-specific mortality rates can be analyzed using proportions and Trent and Rongstad procedures (Micromort software, Heisey and Fuller 1985). Cub survival curves for each gender will be plotted with survival rate on age in months (Logan and Sweanor 2001:119). Population Estimates: Capture-recapture models will be evaluated initially as a pilot study to estimate the parameters of primary interest― absolute numbers of independent puma (i.e., number of adult and subadult puma present in the survey area) and puma density (i.e., number of independent puma/100 km2) each winter― December through March― when snow facilitates detection and capture of puma, provided that we meet model assumptions. The December―March period also corresponds with Colorado‘s puma hunting season. The population of interest is independent puma (i.e., adults and subadults) because those are the puma that can be legally killed by recreational hunters. Furthermore, adults comprise the breeding segment of the population and subadults are non-breeders that are potential recruits into the adult population in ≤1 year. Thus, the sampling unit is the individual independent puma (~≥1 yr. old). Basic assumptions for closed capture-recapture models are: (1) the population is closed; (2) animals do not lose their marks during the interval; (3) all marks are correctly noted and recorded at each trapping occasion; (4) each animal has a constant and equal probability of capture on each capture occasion. Open population models allow the assumption of closure to be relaxed (Otis et al. 1978, White et al. 1982, Pollock et al. 1990). The robust design is a combination of closed and open models; thus, assumptions are a combination of the assumptions for closed and open population methods (Kendall 2001). To analyze capture-recapture data, closed, open, and the robust design models are available in program MARK. Akaike‘s Information Criterion will be used to select the most parsimonious models based on AICc score ranks and the difference in AIC (∆AIC) between models (Burnham and Anderson 1998). MARK results also include estimates of abundance. 115 �Because the precision of estimates for small populations is sensitive to the probability of capture (White et al. 1982, Pollock et al. 1990), our operational goal will be to achieve capture probabilities of ≥0.4 for each animal per capture occasion (see Results and Discussion, Segment Objective 7). In addition, behavior, movements, survival and mortality of GPS- and VHF-collared puma will allow direct biological examinations of assumptions of geographic and demographic closure (White et al. 1982) and variation in capture probability of individual puma and puma classes (i.e., adult females, adult males, subadult females, subadult males). If capture probabilities vary by puma class, we will examine if data stratification is necessary or possible (depending upon sample size). For example, we might expect the larger home ranges of male puma to expose them to more search routes, thus, this may increase their probability of capture. If the assumption of demographic closure cannot be satisfied, then open population models and the robust design would be more appropriate (Pollock et al. 1990, Williams et al. 2001). Collared puma will allow us to determine the number of marked puma present in the search area each capture-recapture occasion. Furthermore, GPS locations (4 fixes/day) on individual puma will provide data on the probability that puma may temporarily move out of and back into the survey area between capture occasions. Unmarked puma that are subsequently GPS-collared should provide such information, too. ArcView geographic information system software will be used to map and analyze puma locations, movements, and home ranges. It will also be used to map and quantify attributes of the study area and sampling frames. Nt+1/Nt) between consecutive years and Rate of Population Increase: Finite rates of increase ( average annual rates of increase (r) for 3- to 5-year periods and levels of precision will be calculated (Caughley 1978, Van Ballenberghe 1983) and plotted. Functional Relationships: Graphical methods will be used to examine functional relationships among puma population parameters. Linear regression procedures and coefficients of determination can be used to assess functional relationships if data for the response variable are normally distributed and the variance is the same at each level. If the relationship is not linear, data is non-normal, and variances are unequal, we will consider appropriate transformations of the data for regression procedures (Ott 1993). Non-parametric correlation methods, such as Spearman‘s rank correlation coefficient, can also be used to test for monotonic relationships between puma abundance and other parameters of interest (Conover 1999). Statistical analyses will be performed using SYSTAT and SAS software. RESULTS AND DISCUSSION Segment Objective 1 Field research to quantify puma population structure, vital rates, and causes of mortality for this report extended from August 2007 to July 2008. Our searches to detect puma presence covered the entire study area. We made 41 puma captures during the period (9 adult females [1 adult female captured 3 times], 6 adult males [1 adult male captured 2 times], 1 subadult male, and 21 cubs [4 of them captured twice each]), resulting in 2 adults (1 female, 1 male), 4 subadults (2 females, 2 males), and 16 cubs (7 females, 9 males) captured for the first time in 2007-08. Trained dogs were used as our main method to capture, sample, and mark adult and subadult pumas from November 19, 2007 to April 24, 2008. Those efforts resulted in 77 search days, 217-218 puma tracks detected, 49 pursuits, and 20 puma captures (Table 1). Puma capture efforts (i.e., search days) with dogs in this period was similar to our efforts in the 3 previous winters (Table 2). But, the frequency of tracks encountered and pursuits increased over the 3 previous periods. Our capture rate 116 �declined slightly probably due to our ability to identify radio-collared pumas associated with tracks (see later), thus, negating the need to capture the pumas directly with dogs. Seven adult and subadult pumas were captured for the first time (Table 3). This includes 1 adult female puma that could not be handled for safety reasons (Tables 4). One large cub, and 1 adult female were each recaptured twice, but could not be handled for safety reasons. One GPS-collared male puma was visually observed in association with an adult female puma we recaptured with dogs, but the male puma was not be bayed by the dogs (Table 4). The age structure of independent pumas captured for the first time continues to suggest that we have been studying a relatively young age-structured puma population that is increasing in the current reference period (Figure 2). Our puma capture efforts using ungulate carcasses and cage traps extended from August 7, 2007 to July 15, 2008. We used 59 road-killed mule deer, 1 road-killed elk, and 1 puma-killed mule deer as bait at 15 different sites to capture one adult male puma 2 times (Tables 5). The puma-killed deer was used as bait at another site after puma F30 abandoned the carcass after we set a camera trap at her cache to obtain photos of the number of marked cubs in her family to confirm survival data. Pumas scavenged 11 of 60 (18.3%) of the ungulate carcasses used for bait. This was slightly lower than results of the last 2 years (i.e., 20%, 22.5%). Other carnivores that used the ungulate baits included: black bear, coyote, and bobcat. Recaptures of 11 to 12 individual marked pumas were made 17 times with the use of dogs and cage traps; GPS/VHF collars were replaced as needed (Table 6). This included puma M27 (which wore a non-functional GPS collar) that was treed twice north of the study area by a puma hunter (Stan Garvey, Nucla) using dogs. The hunter reported the observation of the tagged animal (including, ear-tag number, and a visible hole in the GPS unit battery box), dates, and locations to principal investigator K. Logan. One recapture was of puma cub M44 (offspring of F7) made by Wildlife Services personnel responding to puma depredation on domestic sheep on the study area. The Wildlife Service houndsman released dogs on the puma tracks, and subsequently treed M44 and shot him to control the depredation. In another instance, a researcher visually observed a GPS-collared male puma in association with puma F23 as we pursued both pumas with dogs. Neither of the pumas had functioning GPS collars at the time. The GPS-collared male puma was either M27 or M29, as those 2 adult males were the only GPS-collared males that ranged in that area. The dogs treed F23, but they did not bay the male puma to enable us to obtain exact identity. We also captured 16 cubs (9 male:7 female) for the first time (Table 7). Seven cubs were radiocollared, including zero to 2 cubs collared in each of 7 litters (Appendix A). One 18 kg female cub was treed by our trained dogs, and immobilized with a pole syringe for safe handling. The other 15 cubs were handled without anesthetics at their nurseries when they were 28 to 40 days old. The litters were produced in May (3), June (2), and July (2). In addition to our direct puma captures, we identified 11 previously marked adult pumas that we detected 34 times initially by snow-tracking (Table 8). Upon detecting puma tracks that were roughly aged at 1 to 2 days old, we followed the tracks with a radio receiver in an effort to detect if the tracks might be of a puma wearing a functional collar. We assigned tracks to a collared individual if we received radio signals from a puma that we judged to be < 1 km from the tracks and in direction of travel of the tracks. GPS data from pumas with functional GPS collars provided confirmatory information about movements of pumas. If GPS data indicated that the puma moved through the area at the time the tracks were made in snow, then we ruled the data were confirmatory. A large majority (i.e., 70%) of confirmatory data is a combination of radio-telemetry and GPS data. One snow track was assigned to a male puma only using GPS data, apparently because he had moved sufficiently far enough away so we did not receive radio signals at the time we found his tracks. If the GPS data did not indicate movement through the area, but the puma probably had sufficient time between fixes to foray to the tracks from proximate GPS locations, then we decided the GPS data were inconclusive. None of the GPS data clearly indicated that an individual puma could not have been the one we initially identified by radio-telemetry. 117 �In one instance, principal investigator K. Logan visually observed puma F25 attack a mule deer after following up her tracks with radio-telemetry. If we could not identify a collared puma in association with 1-day-old puma tracks, then we released the dogs in attempt to capture the puma. This approach allowed us to more efficiently allocate our capture efforts toward pumas of unknown identity on the study area, particularly unmarked pumas or pumas with non-functioning radio- or GPS-collars. This approach would also be useful in a rigorous mark-recapture effort where radio-collared pumas are available. Our search efforts throughout the study area also revealed the presence of at least 9 other independent females and 1 independent male. We could separate the activity of these pumas from the GPS- and VHF- collared pumas in time and space. Moreover, females in association with cubs of different counts and sizes enabled us to separate 5 adult females followed by 1 to 3 medium-to-large-size cubs. One adult female with 1 large dependent cub was treed, but could not be handled safely. She initially treed, which provided us with an excellent visual observation; but, she left the tree and escaped into a system of sink holes that were too unstable (i.e., dangerous) for researchers to enter. Another adult female with 2 medium-size cubs was pursued with dogs, but was not captured. It was the same situation with another adult female with 2 to 3 medium-size cubs in a different area. The tracks we found of the other pumas were too old to pursue (i.e., probability of capture with the dogs was negligible). Our search and capture efforts during November 2007 through April 2008 enabled us to estimate a minimum count of 33 independent pumas detected on the Uncompahgre Plateau study area, up from a minimum count of 24 independent pumas during the November 2006 to May 2007 period. This estimate was based on the number of known radio-collared pumas, the observation of one non-collared puma, and detection of tracks of suspected non-collared pumas on the study area (explained previously). In addition to the independent pumas, we also counted a minimum of 20 to 21 cubs. The sex and age structure of the minimum puma count is in Table 9. Of the 33 independent pumas, 23−24 (70−73%) were marked and 9−10 (27−30%) were assumed to be unmarked animals. Of the expected unmarked pumas, 8−9 were females and 1 was male, which might reflect lower detection rates of females. There appears to be variation in puma numbers on the west and east slopes of the study area. The west slope count includes 12 independent pumas (8 females, 4 males). The east slope count includes 21 independent pumas (13 females, 8 males). We used the minimum puma count and population structure in an effort to develop puma population models to simulate expected puma population dynamics in the remainder of the reference period and expected results of harvest management for the treatment period on the Uncompahgre Plateau Puma Project. Moreover, the models can be used by CODOW wildlife managers and biologists as a tool to explore expected outcomes of puma harvest management strategies in Colorado (see Segment Objective 5). Anderson et al. (1992) studied pumas on the east slope of the Uncompahgre Plateau (i.e., GMU 62) during 1981 to 1988. Sport-hunting was banned during that study to investigate an ― unexploited‖ puma population (Anderson et al. 1992:5). As our current effort results in larger samples and progresses in time through the reference and treatment periods, similarities and differences in results of the 2 research efforts, now separated by more than 15 years, should illuminate reliable knowledge for puma management in Colorado. Our current puma research on the Uncompahgre Plateau has been underway for 3.7 years (compared to 7 years of Anderson et al. 1992). Our data analysis at this stage of the research is not by any means exhaustive or complete because we are still in the intensive data-gathering phase, yet, our data allows some preliminary comparisons with Anderson‘s (1992) completed work. In the Anderson et al. (1992) study, the average capture effort with dogs was 91.1 days per winter (range = 32 to 136, n = 7) resulting in an average capture effort of 13.9 days per puma. Of 189 pursuits of pumas, 110 (58%) were successful (either of radio-collared or non-collared animals). Anderson et al. (1992) focused on capturing pumas >27 kg in body mass while avoiding pumas <27 kg in mass. They captured 47 pumas with dogs for an average capture rate of 13.9 days per puma. Eight other pumas, all 118 �female cubs ≤ 7 months old, were caught in steel leg-hold traps by trappers, and were added to the study animal population. Two other cubs were killed by the dogs. In total, Anderson et al. (1992) captured 57 pumas, of which 49 were radio-collared. So far, in our 4 winters, the average effort to capture pumas with dogs is 78.8 days (range = 77 to 82). Of 172 pursuits, 70 (41%) were successful. We captured 38 individual pumas their first time with dogs (i.e., does not include dog-aided recaptures), yielding an average capture rate of 8.3 days per capture (i.e., 315 days/38 captures). Other capture efforts and results between the 2 studies are not comparable, because Anderson et al. (1992) did not routinely attempt to capture pumas using cage traps or capture cubs at nurseries like we are. In our current effort, we captured, sampled, and marked 90 pumas. Of those animals, 74 were radio-collared, allowing us to monitor fates of pumas in all sexes and age stages, including: 15 adult females, 11 adult males, 2 subadult females, 5 subadult males, 25 female cubs, 22 male cubs (some individuals occur in more than one age-stage). To date, this represents the largest number of individual pumas sampled for population data in Colorado. Mass recorded by Anderson et al. (1992:86) for pumas having an estimated age ≥24 months, averaged 61.6 kg for 8 males, (SD = 5.7, range = 51.8 to 70.8) and 44.5 kg for 14 females (SD = 3.6, range = 38.5 to 49.9). So far in our current study, mass for pumas ≥24 months old averaged 59.4 kg for 11 males (SD = 7.42, range 40 to 68 kg) and 38.4 kg for 14 females (SD = 4.29, range = 31 to 46). Sexual dimorphism is evident in pumas, and has been described for the species throughout its range (Young and Goldman 1946). Sexual dimorphism in puma has been explained as a potential result of sexual selection (Logan and Sweanor 2001:109). Segment Objective 2 During the past 3.7 years of this work we compiled data on puma reproduction that was not previously available on pumas in Colorado. We examined 59 cubs from 21 litters aged 29 to 42 days old where we were reasonably sure that we counted all the cubs at the nurseries (Appendix A). The distribution of puma births by month indicate puma births extending from March into September, with 26 of 28 births occurring May through September (Fig. 3).The secondary sex ratio was 33:26 for 21 litters where all the cubs were sexed. This ratio was not significantly different from 1:1, (X2 = 0.8305 < 3.841, α = 0.05, 1 d.f.). An equal sex ratio at birth is characteristic of other puma populations in North America (Robinette et al. 1961, Logan and Sweanor 2001:69-70). The mean (±SD) and extremes of litter sizes were 2.810 (±0.9808), 1−4 (Table 10). In addition, 13 birth intervals for 8 different female pumas averaged 17.969 months (SD = 4.748), and ranged from 11.7 to 23.9 months (Table 10). During the past 3 biological years (i.e., 2005-06 to 2007-08) when we radio-monitored 12, 13, and 12 adult female pumas respectively, the proportion of adult females that produced cubs each year were 0.67, 0.69, and 0.58, with a mean ± SD of 0.65 ± 0.0586. Based on observations (from GPS and radio-telemetry data) of associations between 7 mothers and putative sires, 8 estimated gestation periods averaged 91.188 days (SD = 2.3443), which is consistent with average puma gestation reported in the technical literature on puma (i.e., mean ± SD = 91.9 ± 4.1, Anderson 1983:33, mean = 91.5 ± 4.0 Logan and Sweanor 2001:414). Anderson et al. (1992:47) reported of ―1 7 postnatal litters about 10-240 days in estimated age from 12 individual females, the mean (±SD) and extremes of litter sizes were 2.41 ± 0.8, 1-4‖. ―Because most postnatal young were not handled, their sex ratio is unknown‖ (Anderson et al.1992:48). In addition, because cubs were first observed at older ages, it is likely that some post-natal mortality had occurred. This is one explanation for smaller litters observed by Anderson et al. (1992). Anderson et al. (1992:47-48) found that of 10 puma birth dates 7 were during July, August, and September, 2 in October, and 1 in December, with most breeding occurring April through June. Data on our 28 litters adds to Anderson‘s data (Fig. 2), and indicates puma births in Colorado occurring in every 119 �month except January and November (so far). Our data suggests that the majority of puma breeding activity occurs February through June. Anderson‘s observation of two 12-month birth intervals for one female (Anderson et al. 1992:48) is at the low range of our observations (see previously). Segment Objective 3 & 4 From December 8, 2004 (capture and collaring of the first adult puma M1) to July 31, 2008, we radio-monitored 11 adult male and 15 adult female pumas to quantify survival and agent-specific mortality rates (Table 11). One adult male is known to have died. M4 was about 37 to 45 months old when he was killed by an unidentified male puma along the southeast boundary of the study area. We lost contact with 3 adult males apparently due to GPS/VHF collar failure: M1, M27, and M29. Direct observations in the field during January 2008 indicated that M27 was alive, and M29 might also be alive. Three adult females are known to have died. F50 was about 29 to 31 months old when she died apparently of natural causes (exact agent could not be identified). Two adult females, F54 and F30, were killed by other pumas. F54 was killed at about 49 months old by a male puma on the southern boundary of the study area while apparently in direct competition for a fawn mule deer. F30 was apparently killed by a puma of unknown sex and for unknown circumstances when she was about 60 months old. Both females died as a result of fatal bites to the head. Preliminary estimates of adult puma survival rates indicate relatively high survival in this reference period (i.e., with no sport-hunting) (Table 12). Survival rates were estimated using the KaplanMeier procedure to staggered entry of animals (Pollock et al. 1989) for the past 2 annual and hunting season periods when samples were ≥ 5 animals in each sex category. The survival rates reflect zero male deaths, and all 3 adult females that occurred in those periods. We need to increase the number of radiomonitored adult males to obtain more realistic survival rates (i.e., other than 1.0). The adult age structure, as indicated in Figure 4, is indicative of high survival rates during the past 4 winters without sporthunting mortality. Research in New Mexico on a non-hunted puma population also indicated higher survival rates for adult male than adult female pumas, with the major cause of death being aggression by male pumas (n = 8 years; Logan and Sweanor 2001:127-138). We have radio-monitored 7 subadult pumas, 5 males and 2 females (Table 13). None of those died while we were monitoring them in the subadult age stage. F23 has become a breeding adult on the study area. M5 dispersed from his natal area and the study area at about 13 months old and went to the northwest slope of the Uncompahgre Plateau where he established an adult territory. M49 was orphaned at 9 months old when his mother F50 died. He dispersed from his natal area and the study area to the northeast slope of the Uncompahgre Plateau, but shed his expandable radio-collar at a fresh elk kill when he was about 15 months old. Puma M11 became a subadult at 13 months old and dispersed from his natal area at 14 months old. He moved to the Dolores River valley between Stapletone and Stoner, Colorado by December 14, 2006. He was legally killed by a puma hunter on December 12, 2007 when he was 30 months old, in the adult age-stage. We need to increase our efforts to acquire larger samples of male and female radio-monitored subadult pumas to acquire more realistic estimates of their survival (i.e., other than 1.0). Contact was lost with 2 subadult males and 1 subadult female. F52 dispersed from the study area before we lost track of her in the area of the Black Canyon of the Gunnison in mid-May 2007. We lost track of M31 seven days after he was captured in April 2006. He might have dispersed from the study area. Efforts to locate him by flying over and around the study area have not been successful. M69 emigrated from the study area in spring 2008 when he was about 16 to 20 months old. We monitored him in the Beaton Creek area east of the Uncompahgre River for awhile until we lost contact with him in April 2008. In addition to the subadults discussed previously, a non-marked female puma about 18 to 24 months old was killed by a vehicle November 4, 2006 on highway 550 (between Colona and Ridgway), 120 �which forms the southeast boundary of our study area. The female appeared to be in good health (41 kg), was not pregnant, and was not lactating. Anderson et al. (1992) found that all 9 radio-collared male pumas dispersed from their natal areas, and 2 of 6 radio-collared females did not disperse from their natal areas (A. E. Anderson, Sep. 1993, errata for Anderson et al. 1992:61). Mean ± SD and range of dispersal distances (km) for 8 males, aged 10 to 13 months old at dispersal, were 86.2 ± 51.3, 23 to 151. For 4 females, aged 11 to 31 months old at dispersal, mean ± SD and range of dispersal distances (km) were 37.0 ± 15.3, 17 to 54 (Anderson et al. 1992:63). Although we have observed 3 male pumas disperse from natal areas, and no females disperse, our current research is too short in duration and samples too small yet to make meaningful comparisons with Anderson‘s earlier effort, particularly regarding offspring dispersal rates, distances moved, and philopatry. Dispersal and philopatry have been explained as life history strategies in pumas that assist gene flow, colonization, population maintenance, and individual survival and reproductive success (Logan and Sweanor 2001). Thus, such strategies would be expected to be conserved, and expressed in puma populations in different locations and at different times. In addition, because puma emigration and immigration (i.e., via dispersal) have been shown to be important processes in puma population dynamics (Sweanor et al. 2000), we need larger samples and longer research duration in this study to estimate those parameters. A preliminary estimate of puma cub survival was made with 38 cubs (21 males, 17 females) that we marked (n = 31 were radio-collared) at nurseries when they were 26 to 42 days old. Only cubs that died of natural causes were used (i.e., 3 capture-related deaths were excluded). All cubs were born from May 2005 to July 2007. Cubs that died included 13 that were radio-collared at nurseries and 3 noncollared cubs that apparently disappeared from families because they were not subsequently observed or track counts indicated attrition in cubs. For the Kaplan-Meier procedure to staggered entry of animals (Pollock et al. 1989), the maximum survival period was assumed to be 365 days (i.e., 12 months) to coincide with the time that puma cubs would usually be expected to become independent from their mothers (Logan and Sweanor 2001). Otherwise, cubs were right censored if they reached independence, or we lost contact before then. Dates that bracketed the deaths or disappearances of cubs were used to estimate minim and maximum survival rates. Maximum estimated cub survival using the Kaplan-Meier procedure was 0.4998 (SE = 0.2499). The estimated minimum survival rate was practically the same, 0.4993 (SE = 0.2498). Cub survival estimated with a binomial model (Williams et al. 2001) was 0.5789 ± 0.1570 (95% C.I.). In order to improve the reliability of puma cub survival data, we will make an effort to increase the number of radio-collared cubs that are monitored. The major natural cause of death in cubs, where cause could be determined, was infanticide and cannibalism by male pumas (Appendix A). Male-caused infanticide and cannibalism, along with aggression-caused mortality in adult (indicated previously) and subadult pumas (Logan and Sweanor 2001) has also been a dominant mortality factor in other puma populations in North America (Logan and Sweanor 2001:115-136). Such male puma behavior has been theorized for being a strong selective force in shaping the evolution of behavioral tactics and life history strategies in pumas (Logan and Sweanor 2001). The current closure on sport-hunting on the study area and protection of marked pumas from sport-harvest on the buffer area on the northern portion of the Uncompahgre Plateau for the reference period appears to be operating, so far. None of the adult or subadult pumas wearing functional GPS- or VHF-collars have died due to human causes. This reference condition enables us to quantify puma population structure, survival rates, and agent-specific mortality rates of pumas in the absence of direct 121 �human-caused mortality factors related to sport-hunting, and allow comparisons with the treatment period when hunting of pumas on the study area resumes. Anderson et al. (1992:50) reported on the fates of 21 radio-collared pumas (11 < 24 months old, 10 ≥ 24 months old) from a total of 49 in his previous study which was intended to ―a ssess the effects of sport-hunting on an unexploited population‖ (Anderson et al. 1992:5). They found 19 (90%) of those pumas died due to human causes, attributed to: legal kill outside the study area (7), capture-related (6), predator management (3), illegal kill (2), and suspected predacide (1). Other causes of mortality included, intraspecies strife (1) and disease (1). Actual age-stage and annual survival rates and agent-specific mortality rates from our current effort cannot be clearly compared with the Anderson et al. (1992:53) data set because they pooled data for male and female pumas in seemingly arbitrary age stages that overlapped puma life history stages (i.e., cubs, subadults, adults). The Anderson et al. (1992:53) estimated survival rates with the Kaplan-Meier procedure (Pollock et al. 1989) for 20 male and 22 female pumas were: 1224 month old = 0.642; 24-36 months old = 0.692, 36 to 48 months old = 0.917, and 48-60 months old = 0.800. Actual sample sizes within each age-stage were not given. There were no quantitative data allowing estimation of survival and agent-specific mortality for cubs less than 12 months old. Segment Objective 5 Cumulative data gathered during the past 3.7 years on the Uncompahgre Plateau Puma Project allowed a minimum count of pumas on the Uncompahgre Plateau Study area, and attendant estimates of population structure, reproduction rates, and survival rates. Those data positioned this project to begin puma population modeling efforts. Such modeling processes are useful for CDOW Mammals Researchers to design the treatment phase of this research project and provide CDOW wildlife biologists and managers with tools to assess current puma harvest management assumptions (previously in Testing Assumptions and Hypotheses) and other conceptual and proposed puma management approaches. A deterministic, discrete time model was developed and created on Excel (Microsoft Office software 2007) by principal investigator K. Logan and CDOW Biometrician P. Lukacs. The model structure has 3 age stages recognized in puma population biology (Logan and Sweanor 2001)− adult, subadult, and juvenile− and which are consistent with parameters we are estimating in this research and available in the technical literature on puma populations: Adult: NAFt+1 = (SF*NAFt + SSF*NSFt)(1-HAFt+1) NAMt+1 = (SM*NAMt + SSM*NSMt)(1-HAMt+1) Subadult: NSFt+1 = ((rSJF*NJt)(1-HSFt+1))PIF/EF NSMt+1 = (((1-r)SJM*NJt)(1-HSMt+1))PIM/EM Juvenile: NJt+1 = RNAFt+1 The model terms are: NAFt+1 = Number of adult females at year t+1. NAMt+1 = Number of adult males at year t+1. NSFt+1 = Number of subadult females at year t+1. NSMt+1 = Number of subadult males at year t+1. NJt+1 = Number of juveniles at year t+1. S = Survival rate for each specified sex and age stage. H = Proportion of the harvest rate comprised by each sex and age stage (e.g., 0.28 harvest rate * 0.40 adult females). PI/E = Ratio of progeny + immigrants/emigrants. R = Reproductive rate for adult females (i.e., average number of cubs per female per year). 122 �r = Proportion of the subadult population that is female (e.g., 0.5; 1-0.5 = proportion of males). These basic assumptions pertain to the use of this model. Expected puma population projections and annual rates of increase (i.e., lambda) generated by the model are conditional on the assigned puma population structure and demographic estimates that parameterize the model. The model structure does not include density dependence, and thus, should not be used to project population trends beyond 10 years. In reality, density dependence probably operates in puma population dynamics, with competition for food expected to regulate independent (i.e., adults and subadults) female density and competition for mates expected to regulate independent male density (Logan and Sweanor 2001). The model structure also assumes that puma harvest is strongly additive mortality, an assumption that is consistent with the current observed adult and subadult (i.e., harvest-age pumas) puma survival rates in the reference period and for adult pumas in other non-hunted puma populations (Logan and Sweanor 2001). We used this model to simulate puma population dynamics to examine a set of scenarios that pertain to current CDOW puma management assumptions and to consider the puma research and management direction for the treatment period. Furthermore, we modeled the potential population impact of the historical puma harvest on the study area prior to the current puma research (i.e., 1994-2003). We parameterized the model with data gathered on the pumas on the study area during the past 3.7 years. The starting population was the minimum count of pumas and attendant estimated sex and age structure made during November 2007 to March 2008 (Table 9). We assumed that all individuals were present in the population during that entire period. No mortalities of independent pumas were detected. But, one radiocollared subadult male emigrated by March 19, 2008. Population parameters included: estimated rates of reproduction and sex and age-stage specific survival, which included data to July 2008 (Table 14). Some sex and age-stage specific estimates of survival (i.e., adult male, subadult male, subadult female) came from the literature (Table 14), because our current sample sizes (i.e., number of individuals and years) were not adequate for realistic estimates (i.e., estimates from our data were 1.0 for adult males and subadults). We did not use actual rates in the literature where estimates involved the pooling of data on sexes and age stages, and where sample sizes for age stages were not presented (e.g., Anderson et al. 1992). In addition, the ratio of progeny and immigrant recruits to emigrants as a model input was from the literature, because such data is scarce and does not exist for Colorado (all references in Table 14). We preferred using the population characteristics and parameter estimates gathered in the current study, because this is the puma population we intend to manipulate in the treatment period to test CDOW puma management strategies. Results of our modeling efforts are presented in Appendix B. This constitutes the first time that current CDOW puma harvest assumptions have been evaluated by using Colorado-specific population data, and thus, is considered to be preliminary. Expected estimates of population growth were generally consistent with the CDOW puma harvest management assumptions that were previously developed from data in the puma population literature to manage for a stable-to-increasing population and for a declining puma population. The results demonstrated the importance of female survival to population dynamics. As more quantitative population data is gathered and the puma population is manipulated during the treatment period, population dynamics can be evaluated further. Results from the model evaluating the historical puma mortality on the study area during 1994 to 2003 indicate the expected outcome is that the puma population on the study area would decline during the treatment years. Segment Objective 6 To investigate the potential that puma hunters might detect puma mothers away from their cubs, we continued gathering data on spatial associations of puma mothers and their cubs during the puma hunting season, which extends from November through March each winter in Colorado. Female pumas are fair game in Colorado, unless they are accompanied by 1 or more cubs. Mothers that are caught away 123 �from their cubs could be legally harvested. Such incidents would result in cubs being orphaned. Orphaned cubs that 6 months old could have a survival rate (to the subadult stage) of < 0.05. Orphaned cubs 7 to 12 months old might have a survival rate (to the subadult stage) of about 0.7 (K. Logan, unpublished data). We monitored only 1 puma family with a radio-collared mother and cub from November 13, 2007 to February 14, 2008 during 8 airplane flights (Table 16).To assess whether mothers were apart or in close association with cubs, we considered error in aerial locations. We recovered 7 puma radiocollars that we located from the airplane and then fixed the actual locations of collars on the ground with GPS. Range of location error was 20 to 520 m (mean = 282.86, SD = 164.75). We decided to use distances greater than the extreme high range of location error (520 m) as the metric to decide if puma mothers might be detected away from their cubs by hunters. Five of 8 (62%) of the observations located the mother and cub :c::::500 m apart, within the extreme margin of location error. In aggregate, the data for the past 3 winters include 136 observations for 1−5 families per winter (Table 15), and generally indicate that puma mothers are more likely to be within 520 m of their cubs during the day in winter. An effort will be made to increase the number of radio-collared family members in subsequent winters. In addition, we will examine variation in mother-cub association distances on an individual female basis. Anderson et al. (1992:70-71) recorded 69 instances of simultaneous aerial locations of 7 pairs of puma mothers and dependent young. They reported that mothers and young were together in 21 (30.4%) of those instances, and they were 1 to 2.2 km apart in 48 (69.6%) of those instances. Segment Objective 7 Principal investigator K. Logan developed 6 drafts study plans pertaining to the next 6 years of puma research on the Uncompahgre Plateau. Three of the drafts were circulated for internal review to obtain comments from CDOW Mammals Research Leader D. Freddy, Carnivore Biologist J. Apker, Area 18 Biologist B. Banulis, Southwest Regional Biologist S. Wait, and Area 18 Wildlife Manager R. Del Piccolo. The planning process involved modeling puma population scenarios (previously in Segment Objective 5) and modeling mark-recapture scenarios in MARK (Cooch and White 2004) with CDOW Biometrician P. Lukacs. The mark-recapture modeling process enabled consideration of effects of puma population size and individual detection rates on the ability to detect changes in puma population abundance that might result from the hunting treatment. Results of the MARK simulations applied to a scenario with 3 capture occasions and puma population abundances that varied from 25 to 50 animals indicated that individual detection rates would need to be 0.4 or greater to be able to detect changes in puma abundance (Table 16). The study plan is expected to be completed in September 2008, with a decision on a course to proceed with the remainder of the research soon thereafter. Segment Objective 8 Data from 23 (7 male, 23 female) GPS-collared pumas, totaling over 31 thousand GPS locations (Table 17) are currently being used in a collaborative study of puma prey use on the Uncompahgre Plateau, carried out by CDOW Mammals Research staff. Plans to use these and other data subsequently gathered, include habitat modeling and mapping for pumas in the western U.S. in collaboration with colleagues at Colorado State University (CSU), and descriptive information on puma behavior in relation to human development on the Uncompahgre Plateau. We are currently collaborating with Dr. Sue VandeWoude and Dr. Kevin Crooks, and postdoctoral and graduate students at CSU to develop a pilot study titled: Puma concolor immune health― Relationship to management paradigms and disease. Tissue samples (i.e., blood, saliva, feces) from pumas we capture are collected and shipped to the Department of Microbiology, Immunology, and Pathology at CSU for analyses. That project will be expanded to The effects of urban fragmentation and landscape connectivity on disease prevalence and transmission in North American felids. A description of 124 �that project and incomplete results on infectious disease surveillance on 27 pumas (16 female, 11 male) sampled on the Uncompahgre Plateau are presented in Appendix C. SUMMARY Manipulative, long-term research on puma population dynamics, effects of sport-hunting, and development and testing of puma enumeration methods began in December 2004. After 3.7 years of effort, 90 pumas have been captured, sampled, marked, and released. Of those, 74 pumas were radiocollared, allowing us to monitor fates of pumas in sexes and age stages, including: 15 adult females, 11 adult males, 2 subadult females, 5 subadult males, 25 female cubs, 22 male cubs. As of July 2008, we were monitoring 18 adults, 1 subadult, and 4 cubs with active radio-collars. Data from the marked animals are used to quantify puma population characteristics and vital rates in a reference situation (i.e., without sport-hunting off-take). During November 2007 through March 2008 a minimum estimate of 33 independent pumas were detected on the Uncompahgre Plateau study area, up from 24 the previous winter, with estimates of sex and age structure. Our efforts to quantify puma population characteristics and vital rates positioned us to begin puma population model development, and to use modeling scenarios to assess potential directions for the remainder of the puma research on the Uncompahgre Plateau. Moreover, our data and model provide tools useful to CDOW wildlife biologists and managers for assessing effects of puma harvest strategies. A study plan for the remainder of the research has been in development and should be completed in September 2008. To improve data on puma population vital rates, attention will be given to increasing sample sizes on radio-collared adult males, subadults, and cubs. Furthermore, data from 23 GPS –collared pumas, totaling over 31 thousand GPS locations enables collaboration on investigations of puma use of prey, puma-human relations on the Uncompahgre Plateau, and puma habitat modeling and mapping with colleagues. All of these efforts should enhance the Colorado puma research and management programs. LITERATURE CITED Anderson, A. E. 1983. A critical review of literature on puma (Felis concolor). Colorado Division of Wildlife Special Report No. 54. _____, D. C. Bowden, and D. M. Kattner. 1992. The puma on Uncompahgre Plateau, Colorado. Technical Publication No. 40. Colorado Division of Wildlife, Denver. Anderson, C. R., Jr., and F. G. Lindzey. 2005. Experimental evaluation of population trend and harvest composition in a Wyoming cougar population. Wildlife Society Bulletin 33:179-188. Beier, P., and R. H. Barrett. 1993. The cougar in the Santa Ana Mountain Range, California. Orange County Cooperative Mountain Lion Study Final Report. Bergman, E. J., M. W. Alldredge, K. A. Logan, M. Shuette, C. J. Bishop, and D. J. Freddy. 2007. Study Plan-Pilot evaluation of predator-prey dynamics on the Uncompahgre Plateau. Pages 83-96 in Bergman, E.J. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Wildlife Research Report July: 73-96. Colorado Division of Wildlife, Fort Collins. Burnham, K. P., and D. R. Anderson. 1998. Model selection and inference: a practical informationtheoretic approach. Springer-Verlag, New York, New York, USA. Caughley, G. 1978. Analysis of vertebrate populations. John Wiley and Sons, New York. Colorado Division of Wildlife 2002-2007 Strategic Plan. 2002. Colorado Department of Natural Resources, Division of Wildlife. Denver. Colorado Division of Wildlife. 2007. Colorado mountain lion management data analysis unit revision and quota development process. Colorado Division of Wildlife, Denver. Conover, W. J. 1999. Practical nonparametric statistics. John Wiley & Sons, Inc., New York. Cooch, E., and G. White. 2004. Program MARK- a gentle introduction, 3rd edition. Colorado State University, Fort Collins. 125 �Culver, M., W. E. Johnson, J. Pecon-Slattery, and S. J. O‘Brien. 2000. Genomic ancestry of the American puma (Puma concolor). The Journal of Heredity 91:186-197. Currier, M. J. P., and K. R. Russell. 1977. Mountain lion population and harvest near Canon City, Colorado, 1974-1977. Colorado Division of Wildlife Special Report No. 42. Heisey, D. M., and T. K. Fuller. 1985. Evaluation of survival and cause specific mortality rates using telemetry data. Journal of Wildlife Management 49:668-674. Kendall, W.L. 2001. The robust design for capture-recapture studies: analysis using program MARK. Pages 357-360 in R. Field, R. J. Warren, H. Okarma, and P. R. Sievert, editors. Wildlife, land, and people: priorities for the 21st century. Proceedings of the Second International Wildlife Management Congress. The Wildlife Society, Bethesda, Maryland, USA. Koloski, J. H. 2002. Mountain lion ecology and management on the Southern Ute Indian Reservation. M. S. Thesis. Department of Zoology and Physiology, University of Wyoming, Laramie. Kreeger, T. J. 1996. Handbook of wildlife chemical immobilization. Wildlife Pharmaceuticals, Inc., Fort Collins, Colorado. Laundre, J. W., L. Hernandez, D. Streubel, K. Altendorf, and C. L. Lopez Gonzalez. 2000. Aging mountain lions using gum-line recession. Wildlife Society Bulletin 28:963-966. Logan, K. A., E. T. Thorne, L. L. Irwin, and R. Skinner. 1986. Immobilizing wild mountain lions (Felis concolor) with ketamine hydrochloride and xylazine hydrochloride. Journal of Wildlife Diseases. 22:97-103. _____, L. L. Sweanor, J. F. Smith, and M. G. Hornocker. 1999. Capturing pumas with foot-hold snares. Wildlife Society Bulletin 27:201-208. _____, and L. L. Sweanor. 2001. Desert puma: evolutionary ecology and conservation of an enduring carnivore. Island Press, Washington, D.C. _____. 2004. Colorado puma research and development program: population characteristics and vital rates study plan. Colorado Division of Wildlife, Ft. Collins Research Center, Ft. Collins. _____. 2005. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. July: 105-126, Colorado Division of Wildlife, Fort Collins. Murphy, K., M. Culver, M. Menotti-Raymond, V. David, M. G. Hornocker, and S. J. O‘Brien. 1998. Cougar reproductive success in the Northern Yellowstone Ecosystem. Pages 78-112 in The ecology of the cougar (Puma concolor) in the Northern Yellowstone ecosystem: interactions with prey, bears, and humans. Dissertation, University of Idaho, Moscow. Otis, D. L., K. P. Burnham, G. C. White, and D. R. Anderson. 1978. Statistical inference from capture data on closed animal populations. Wildlife Monographs 62:1-135. Ott, R. L. 1993. An introduction to statistical methods and data analysis. Fourth edition. Wadsworth Publishing Co., Belmont, California. Pierce, B. K., V. C. Bleich, and R. T. Bowyer. 2000. Social organization of mountain lions: does land a tenure system regulate population size? Ecology 81:1533-1543. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989a. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. _____, S. R. Winterstein, and M. J. Conroy. 1989b. Estimation and analysis of survival distributions for radio tagged animals. Biometrics 45:99-109. _____, J. D. Nichols, C. Brownie, and J. E. Hines. 1990. Statistical inference for capture-recapture experiments. Wildlife Monographs 107:1-97. Pojar, T. M., and D. C. Bowden. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550-560. Ross, P. I., and M. G. Jalkotzy. 1992. Characteristics of a hunted population of cougars in southwestern Alberta. Journal of Wildlife Management 56:417-426. Stoner, D. C. 2004. Cougar exploitation levels in Utah: implications for demographic structure, metapopulation dynamics, and population recovery. Master of Science Thesis. Utah State University. 126 �Sweanor, L. L., K. A. Logan, and M. G. Hornocker. 2000. Cougar dispersal patterns, metapopulation dynamics, and conservation. Conservation Biology 14:798-808. _____, L. L., K. A. Logan, J. W. Bauer, B. Milsap, and W. M. Boyce. 2008. Puma and human spatial and temporal use of a popular California state park. Journal of Wildlife Management 72:1076-1084. Van Ballenberghe, V. 1983. Rate of increase of white-tailed deer on the George Reserve: a re-evaluation. Journal of Wildlife Management 47:1245-1247. Watkins, B. 2004. Mountain lion data analysis unit L-22 management plan. Colorado Division of Wildlife, Montrose. Williams, B. K., J. D. Nichols, and M. J. Conroy. 2001. Combining closed and open mark-recapture models: the robust design. Pages 523-554 In Analysis and management of animal populations. Academic Press, New York. White, G. C., D. R. Anderson, K. P. Burnham, and D. L. Otis. 1982. Capture-recapture and removal methods for sampling closed populations. Los Alamos National Laboratory Publication LA-8787NERP. Los Alamos, NM, U.S.A. Worton, B. J. 1995. Using Monte Carlo simulation to evaluate kernel-based home range estimators. Journal of Wildlife Management 59:794-800. Young, S. P. and E. A. Goldman. 1946. The puma: mysterious American cat. The Wildlife Institute, Washington, D. C. Prepared by: ________________________ Kenneth A. Logan, Wildlife Researcher 127 �Table 1. Summary of puma capture efforts with dogs from November 19, 2007 to April 24, 2008, Uncompahgre Plateau, Colorado. Month November No. Search Days 5 December 18 January 18 69 tracks: 23-27 male, 22-26 female, 20 cub 5 pursuits: 2 males, 3 females February 20 64-65 tracks: 14-15 male, 30-31 female, 19-20 cub 21 pursuits: 9 males, 9 females, 3 cubs March 11 17 tracks: 5-6 male, 9-10 female, 2 cub April 5 15 tracks: 1 male, 6 female, 8 cub 217-218 tracks: 6573 male, 85-93 female, 59-60 cub 11 pursuits: 3-4 males, 4-5 females, 3 cubs 6 pursuits: 2 females, 4 cubs 49 pursuits: 16-17 males, 20-21 females, 12 cubs No. & type of pumas pursued 1 pursuit: 1 male 5 pursuits: 1 male, 2 females, 2 cubs No. & I.D. or type of pumas captured 1 puma recaptured: M55 (not handled). 4 pumas captured 5 times: M32 recaptured (not handled), F25 recaptured (faulty GPS collar changed), cub F57 recaptured twice (not handled), cub M44 recaptured by Wildlife Services & killed for depredation on domestic sheep. 5 pumas captured: M69 & M71 (handled & marked for the first time), F16 recaptured (faulty GPS collar changed), F2 recaptured (faulty GPS collar changed), F70 (handled & marked for the first time). 5 pumas captured 7 times: M73 (handled & marked for the first time), F23 recaptured 3 times (could not be handled safely first 2 times, faulty GPS collar changed the 3rd time), F72 (handled & marked for the first time), 1 radio-collared male puma was visually observed in association with F23 while pursuing a female & male puma with dogs on 2-25-08, but he could not be treed to handle (either M27 or M29, both with non-functional GPS collars), 1 unmarked adult female captured (could not be handled safely). 2 pumas captured: F74 (handled & marked for the first time), F75 (handled & marked for the first time). 0 pumas captured 20 captures of 17 individuals: 7 independent pumas and 1 cub were captured for the 1st time- M69, F70, M71, F72, M73, cub F74, F75, & 1 unmarked adult female (not handled). a Puma hind-foot tracks with plantar pad widths >50 mm wide are assumed to be male; 50 mm are assumed to be female (Logan and Sweanor 2001:399-412). b Pumas are not handled for a variety of safety reasons: tree to dangerous to climb for researchers, puma treed near river, creek or cliff, puma might fall from tree after drug induction. TOTALS 77 No. & type of puma tracks founda 20 tracks: 9 male, 8 female, 3 cub 32 tracks: 13-15 male, 10-12 female, 7 cub 128 �Table 2. Summary of puma capture efforts with dogs, December 2004 to April 2008, Uncompahgre Plateau, Colorado. Period Track detection effort 109/78 = 1.40 tracks/day Dec. 2, 2004 to May 12, 2005 Nov. 21, 2005 to May 26, 2006 Nov. 13, 2006 to May 11, 2007 Nov. 19, 2007 to April 24, 2008 35/78 = 0.45 pursuit/day Puma capture effort 14/78 = 0.18 capture/day Effort to capture an independent puma for the first time 11 pumas captured for first time 11/78 = 0.14 capture/day 78/14 = 5.57 day/capture 14/82 = 0.17 capture/day 78/11 = 7.09 day/capture 149/82 = 1.82 tracks/day 78/35 = 2.23 day/pursuit 43/82 = 0.52 pursuit/day 177/78 to 182/78 = 2.27-2.33 tracks/day 82/43 = 1.91 day/pursuit 45/78 to 47/78 = 0.58-0.60 pursuit/day 82/14 = 5.86 day/capture 22/78 = 0.28 capture/day 78/47 to 78/45 = 1.66-1.73 day/pursuit 49/77 = 0.64 pursuit/day 78/22 = 3.54 day/capture 78/7 = 11.14 day/capture 20/77 = 0.26 capture/day 7 pumas captured for first time 7/77 = 0.09 77/49 = 1.57 day/pursuit 77/20 = 3.85 day/capture 77/7 = 11.00 day/capture 217/77 to 218/77 = 2.82-2.83 tracks/day Pursuit effort 7 pumas captured for first time 7/82 = 0.08 capture/day 82/7 = 11.71 day/capture 7 pumas captured for first time 7/78 = 0.09 capture/day Table 3. Adult and subadult pumas captured for the first time, sampled, tagged, and released from January 2008 to March 2008, Uncompahgre Plateau, Colorado. Puma I.D. M69 F70 M71 F72 M73 F74 F75 Sex M F M F M F F Estimated Age (mo.) 14-18 33 24 24 49 8-9 41 Mass (kg) 42 39 55 43 60 18 39 Capture date 01-11-08 01-14-08 01-29-08 02-12-08 02-21-08 03-12-08 03-26-08 Capture method Dogs Dogs Dogs Dogs Dogs Dogs Dogs 129 Location Dolores Creek Dolores Creek East Fork Dry Creek Loghill Mesa North fork Cottonwood Creek North fork Cottonwood Creek Cottonwood Creek �Table 4. Pumas that were captured and observed with aid of dogs, but were not handled at that time for safety or other reasons, December 2007 to February 2008, Uncompahgre Plateau, Colorado. Puma sex Capture date Location Comments F57 Age stage or months 7 12-03-07 Caterwauler Canyon F57 8 12-19-07 Loghill Mesa Female adult 02-01-08 Cottonwood Canyon F23 49 02-19-08 F23 49 02-20-08 Big Bucktail Canyon San Miguel Canyon M27 or M29 78 107 02-25-08 F57 was previously marked at the nursery when about 35 days old; born ~April 16, 2007. F57 was recaptured high in a tree, too dangerous to attempt to handle her to fit an expandable radio-collar. F57 was recaptured in a tree that did not allow safe immobilization to handle her to fit an expandable radio-collar. Unmarked female was bayed high in a tree out of range of dart gun. The puma left the tree, but escaped into deep system of sink holes too unstable for any research team member to enter. F23 was recaptured in a tree too dangerous to handle her to change the non-functioning GPS collar she wore. F23 was recaptured again in a tree too dangerous to handle her to change the non-functioning GPS collar she wore. She was safely recaptured and handled on 02-25-08, and was fit with a new GPS collar. A radio-collared male puma was visually observed in association with puma F23 when she & a male puma were pursued with dogs. The male puma was either M27 or M29, both of which had over-lapping home ranges in that area, and both had non-functional GPS collars. But, the male puma could not be treed for absolute identity or for handling. Big Bucktail Canyon Table 5. Summary of puma capture efforts with ungulate road-kill baits, puma kills, and cage traps from August 7, 2007 to July 15, 2008, Uncompahgre Plateau, Colorado.a Carnivore activity & capture effort resultsb No puma activity detected. One deer carcass scavenged by a black bear. No puma activity detected. Deer carcasses scavenged by skunk, bobcat, & black bear. Deer carcasses scavenged by male pumas M55 (10-2 to 3-07) and M29 (10-19 to 22-07). Puma M51 walked ~4 m from a deer carcass, but did not feed. An unknown female puma scavenged on a deer carcass 10-16-07; two cage traps were set and monitored for 2 days, but puma did not return. Deer carcasses were also scavenged by bobcat, coyote, and black bear. November 3 An unknown female puma walked past a deer carcass on 11-1&2-07, but did not feed. An unknown female puma walked past another deer carcass on 11-4-07, but did not feed. An unknown male puma walked past a deer carcass on 11-14-07, but did not feed. Deer carcasses were scavenged by bobcat and coyote. December 2 No puma activity detected. March 3 Unknown male puma scavenged a deer carcass 3-15 to 17-08; two cage traps set and monitored 3-18 &19-08, but puma did not return. Unknown male puma (possibly same as above) scavenged deer carcass 3-23 to 24-08; cage trap set and monitored 3-25 to 27-08, but puma did not return. April 5 Male puma M6 recaptured 4-12-08. He had shed his non-functional GPS collar; we fit him with a new one. An unknown female puma scavenged a deer carcass on ~4-10-08, but did not return. A deer carcass was visited by unknown male & a female pumas; one or both scavenged 4-16-08. Two cage traps were set and monitored 4-17 to 19-08, but the pumas did not return. An unknown male puma scavenged a deer carcass 4-19 or 20-08. Cage trap set and monitored 4-21 to 25-08, but the puma did not return. An unknown female puma scavenged a deer carcass 4-23-08. Cage trap was set and monitored 4-23 to 25-08, but the puma did not return. Another unknown female puma walked past a deer carcass without feeding. July 1 Puma M6 was recaptured 7-15-08; his non-functional GPS collar was replaced with a VHF collar. This was the same bait site and cage trap where we recaptured M6 on 4-8-08. a We used 59 road-killed mule deer, 1 road-killed elk, and 1 puma-killed mule deer (abandoned by F30 and used as bait) at 15 different sites. Of the road-killed ungulate baits, 11 of 60 (18.3%) were scavenged by pumas. b One adult male puma, M6, was recaptured twice. Month August September October No. of Sites 3 4 12 130 �Table 6. Pumas recaptured with dogs, cage traps, or visually observed, November 2007 to July 2008, Uncompahgre Plateau, Colorado. Puma I.D. Recapture Date Estimated Age (mo.) 42 76 Capture Method Process 11-28-07 12-02-07 Mass (kg) Observed Observed M55 M27 Dogs Dogs F25 F57 M44 12-03-07 12-03-07 12-05-07 45 Observed 50 102 7.5 15.5 Dogs Dogs Dogs M32 F57 F16 F2 M27 12-12-07 12-19-07 01-01-08 01-08-08 01-22-08 Observed Observed 43 42 Observed 76 8 59 85 77 Dogs Dogs Dogs Dogs Dogs F25 01-26-08 Observed 103 F23 F23 F23 M27 or M29 02-19-08 02-20-08 02-25-08 02-25-08 Observed Observed Observed Observed 42 42 42 78 107 M6 M6 04-12-08 07-15-08 67 63 74 77 Visual observation of F25 attacking a mule deer after detecting tracks on snow, then snow& radio-tracking Dogs Dogs Dogs Visually observed while pursued by dogs Cage Cage None None, treed in E. fork Tabeguache Cr. by outfitter Stan Garvey, Nucla, CO Changed GPS collar None Shot by Wildlife Services for depredation on domestic sheep None None Changed GPS collar Changed GPS collar None, treed in Johnson Cr. by outfitter Stan Garvey, Nucla, CO None None None Changed GPS collar None GPS collar VHF collar Table 7. Puma cubs sampled June 2007 to August 2008 on the Uncompahgre Plateau Puma Study area, Colorado. Cub I.D. Sex Estimated birth datea Estimated age at capture (days) Mass (kg) Mother Estimated age of mother at birth of this litter (mo) F74b F June 1, 2007 267 18 F75 32 M76 M May 19, 2008 30 2.0 F2 89 M77 M ― ― 2.3 ― ― F78 F ― ― 1.2 ― ― M79 M ― ― 2.2 ― ― F80 F May 23, 2008 40 1.1 F23 45 F81 F ― ― 2.8 ― ― M82 M May 29, 2008 37 2.8 F8 58 M83 M ― ― 2.5 ― ― M84 M June 5, 2008 36 2.6 F70 38 F85 F ― ― 1.8 ― ― F86 F ― ― 2.0 ― ― M87 M July 3, 2008 28 1.9 F3 83 M88 M ― ― 1.8 ― ― F89 F ― ― 1.7 ― ― M90 M July 9, 2008 36 2.1 F72 29 a Estimated age of cubs sampled at nurseries is based on the starting date for GPS location and radio-telemetry foci for mothers at nurseries, and development characteristics of cubs with mother only with radio-telemetry. b This unmarked female cub was captured on 03-12-08 in association with an unmarked adult female puma. The adult female puma, F75, was captured and marked 03-26-08 with cub F74 in association. 131 �Table 8. Pumas detected by tracks and identified by radio-telemetry, GPS-collar fixes, and visual observation. Puma I.D.a Date detected Estimated Age of Tracks on Snow (days) Type of IdentificationRadio-telemetry (VHF) and/or GPS fixes, Visual Observation M55 12/2/07 2 GPS M51 12/3/07 1 VHF & GPS F3 12/6/07 1 VHF & GPS M55 12/15/07 1 VHF & GPS M55 12/18/07 1 VHF (GPS inconclusive) F7 12/28/07 1 VHF & GPS M51 12/28/07 1 VHF & GPS M51 1/3/08 1 VHF M51 1/10/08 1 VHF & GPS F2 1/11/08 1 VHF & GPS F16 & cubs 1/15/08 2 VHF & GPS M51 1/17/08 1 VHF & GPS F16 1/17/08 2 VHF & GPS F25 1/17/08 2 VHF & GPS F16 & cubs 1/18/08 1 VHF & GPS F25 & cub F57 1/18/08 1 VHF & GPS F16 & cubs 1/22/08 1 VHF & GPS F16 & cubs 1/24/08 1 VHF & GPS F25 & cub F57 1/26/08 1 VHF & GPS & visual of F25 F16 & cubs 1/26/08 1 VHF & GPS M55 1/26/08 1 VHF & GPS M32 & Unk.F 1/31/08 1 VHF (GPS NA)b M32 2/6/08 1 VHF (GPS NA) F25 & cub F57 2/12/08 2 VHF & GPS F16 & cubs 2/13/08 1 VHF & GPS F16 2/14/08 1 VHF & GPS F16 & 3 cubs 2/15/08 1 VHF & GPS F8 2/21/08 2 VHF (GPS NA) F23 2/28/08 1 VHF & GPS F23 3/12/08 2 VHF & GPS F8 3/12/08 1 VHF (GPS NA) F25 & cub F57 4/12/08 1 VHF (GPS inconclusive) F16 & 3 cubs 4/12/08 1 VHF (GPS inconclusive) F24 & 2 cubs 4/24/08 1 VHF (GPS NA) a Eleven individual adult radio- and/or GPS-collared pumas were first detected by tracks on snow, then identified by radio- and GPS data, including one visual observation, a total of 34 times. b GPS NA means the GPS instrument was non-functional, but the VHF beacon was working. 132 �Table 9. Minimum puma population estimate based on numbers of known radio-collared pumas and track counts of suspected unmarked pumas on Uncompahgre Plateau study area, Colorado, November 2007 to March 2008. Adults Subadults Cubs Female Male Female Male Female Male Unknown sex 10 4 3 4 4 4 7 6 4 2 0 1 2 2-3 16 8 5 4 5 6 9-10 Total Independent Pumas = 33a,b a Of the total, 23−24 (70−73%) independent pumas were marked and 9-10 (27−30%) were assumed to be unmarked. b The unmarked independent pumas included: 1adult female with 2 large cubs in Happy Canyon, 1 adult female with 1 large cub in Potter Creek and 25-mile Mesa, 1 adult female with 2 large cubs in Monitor Creek, 1 adult female with 2 medium-size cubs in Potter Creek, 1 adult female with 2-3 cubs in San Miguel Canyon, and 1 female or F28 with a non-functional collar Big Bucktail Creek to San Miguel Canyon. Region East slope West slope Totals Table 10. Puma reproduction, Uncompahgre Plateau, Colorado, 2005-2008. Consort pairs and estimated agesa Female Age Male Age (mo.) (mo.) F2 F2 F2 F3 F3 F3 F3 F7 F7 F7 F8*e F8 F8 F16 F16 F23* F23 53 67 89 36 50 62 83 67 82 106 24 37 58 32 52 21 45 Dates pairs consortedb M6 37 06/22-24/05 M51 60 03/31/08 M73 M27 or M29f M29 49 78 02/28-29/08 02/19-25/08 Estimated birth datec 05/28/05 07/29/06 05/19/08 08/01/04 09/26/05 09/17/06 07/03/08 05/19/05 08/13/06 07/10/08 06/26/05 08/13/06 05/29/08 09/22/05 05/24/07 05/30/06 05/23/08 Estimated birth interval (mo.) Estimated gestation (days) 14.0 22.0 13.8 11.7 21.5 93-95 94 14.9 23.9 13.4 22.5 90-91 19.9 23.8 87-93 Observed number of cubsd 3 2 4 1 2 3 3 2 4 3 2 4 2 4 4 3 2 107 F24 75 92 04/12-15/07 06/14/07 90-93 4 F25 74 08/01/05 1 F25 94 04/16/07 20.5 1 F28* 36 06/09/06 2 F28 48 M29 88 12/27-29/06 03/30/07 11.7 92-93 ≥2 tracks F30* 48 M55 34 04/16-20/07 07/17/07 88-92 3 F50 21 07/01/06 1 F54 24 07/01/06 1 F70* 38 M51 60 03/10/08 06/05/08 87 3 F72* 29 07/09/08 1 F75 32 06/01/07 1 a Ages of females were estimated at litter birth dates. Ages of males were estimated around the dates the pairs consorted. b Consort pairs indicate pumas that were observed together based on GPS and radio-telemetry data. c Estimated birth dates were indicated by GPS and radio-telemetry data of mothers at nurseries. d Observed number of cubs do not represent litter sizes as some cubs were observed when they were 5 to 6 months old after postnatal mortality could have occurred in siblings. Only cub tracks were observed with F28. e Asterisk (*) indicates first probable litter of the female, based on nipple characteristics noted at first capture of the female. f A radio-collared, ear-tagged male puma was visually observed with F23 on 2/25/08. Both M27 and M29 wore non-functional GPS collars in that area at the time. 133 �Table 11. Summary for individual adult puma survival and mortality, December 8, 2004 to July 31, 2008, Uncompahgre Plateau, Colorado. Puma I.D. M1 Monitoring span 12-08-04 to 08-16-06 No. days 616 M4 M5 01-28-05 to 12-28-05 08-01-06 to 07-31-08 333 730 M6 M27 02-18-05 to 07-31-08 03-10-06 to 01-22-08 1259 683 M29 04-14-06 to 01-11-08 637 M32 M51 M55 M71 M73 F2 F3 F7 F8 F16 F23 F24 F25 F28 F30 04-26-06 to 07-31-08 01-07-07 to 07-31-08 01-21-07 to 07-31-08 01-29-08 to 07-31-08 02-21-08 to 07-31-08 01-07-05 to 07-31-08 01-21-05 to 07-31-08 02-24-05 to 07-31-08 03-21-05 to 07-31-08 10-11-05 to 07-31-08 02-05-06 to 07-31-08 01-17-06 to 07-31-08 02-08-06 to 07-31-08 03-23-06 to 09-25-07 04-15-06 to 07-29-08 827 571 557 184 161 1301 1287 1253 1228 1024 907 926 904 551 836 F50 12-14-06 to 03-26-07 102 F54 01-12-07 to 08-18-07 218 F70 F72 F75 01-14-08 to 07-31-08 02-12-08 to 07-31-08 03-26-08 to 07-31-08 199 170 127 Status: Alive/Lost contact/Dead; Cause of death Lost contact− failed GPS/VHF collar. M1 ranged principally north of the study area. Dead; killed by a male puma. Estimated age at death 37−45 months. Alive. Born on study area; offspring of F3. He was independent of F3 by 13 months old, and dispersed from his natal area at about 14 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months. Alive. Lost contact− failed GPS/VHF collar. Recaptured 12-02-07 & 01-2208 by puma hunter/outfitter north of the study area. Possibly visually observed on study area with F23 on 02-25-08. Lost contact− failed GPS/VHF collar. Possibly visually observed on study area with F23 on 02-25-08. Alive. Alive. Alive. Alive. Alive. Alive. Alive. Alive. Alive. Alive. Alive. Alive. Alive. Lost contact− failed GPS/VHF collar. Died; killed by another puma (sex of puma unknown). Estimated age at death 60 months. Died of natural causes; exact agent unknown. Estimated age at death 30 months. Died; killed by a male puma while in direct competition for prey (i.e., mule deer fawn). Estimated age at death 49 months. Alive. Alive. Alive. Table 12. Preliminary estimated survival rates (S) of adult-age pumas during the reference period (i.e., the study area is closed to puma hunting), Uncompahgre Plateau, Colorado. Survival rates of pumas estimated with the Kaplan-Meier procedure to staggered entry of animals (Pollock et al. 1989). Survival rates are for an annual survival period defined as the biological year (August 1 to July 31) and the hunting season period (November 1 through March 31). Survival rates were estimated only for periods when n ≥ 5 individuals. Period of interest Annual 8/1/2006 to 7/31/2007 Annual 8/1/2007 to 7/31/2008 Hunting season 11/1/2006 to 3/31/2007 Hunting season 11/1/2007 to 3/31/2008 S 0.909 Females SE 0.0867 n 11 S 1.000 Males SE 0.0000 n 5 0.825 0.1041 13 1.000 0.0000 9 0.909 0.0867 11 1.000 0.0000 5 1.000 0.0000 12 1.000 0.0000 9 134 �Table 13. Summary of subadult puma survival and mortality, December 2004 to June 2008, Uncompahgre Plateau, Colorado. Puma I.D. M5 Monitoring span No. days 09-16-05 to 06-3006 308 M11 06-21-06 to 12-0207 529 F23 01-04-06 to 02-0406 04-19-06 to 04-2606 31 M49 03-26-07 to 10-0107 189 F52 01-10-07 to 05-1507 125 M69 01-11-08 to 04-0708 87 M31 7 Status: Alive/Survived to adult stage/ Lost contact/Dead; Cause of death Alive; independent and dispersed from natal area at 13 months old. Established adult territory on northwest slope of Uncompahgre Plateau. Dead. Independent at 13 months old. Dispersed from natal area at 14 months old. Moved to Dolores River valley, CO, by Dec. 14, 2006. Killed by a puma hunter Dec. 12, 2007 when 30 months old. Alive. Captured on the study area when ~17 months old. Survived to adult stage; gave birth to first litter at ~21 months old. Lost contact. Probable disperser. M31‘s estimated age at capture was 25 months, at the lower margin of puberty for puma. He may have been a dispersing subadult, and could have moved away from the study area. Lost contact. M49 was orphaned at about 9 months old, when his mother F50 died of natural causes. Dispersed from his natal area at about 10 months old and ranged on the northeast slope of the Uncompahgre Plateau. When M49 was ~15 months old, he shed his expandable radio-collar on ~10-01-07 at a yearling cow elk kill on the northeast slope of the Uncompahgre Plateau. Lost contact. Dispersed from study area as a subadult by Jan. 16, 2007. F52‘s last location was Crystal Creek, a tributary of the Gunnison River east of the Black Canyon. Lost contact. Captured on the study area when ~14-18 months old. Emigrated from the study area as subadult by Mar. 19, 2008. Last location was in Beaton Creek, east side of Uncompahgre River valley. 135 �Table 14. Summary of preliminary puma population model parameter estimates obtained from the Uncompahgre Plateau Puma Project and from the published literature on puma. Survival Sex and age stage Adult Female Estimate 0.87 Adult Male 0.91 Subadult Female 0.80 Subadult Male 0.60 Cub 0.50 0.90 Reference Estimated average annual survival rate (n = 2 years) for 11−13 adult females on Uncompahgre Plateau study area. Estimated average annual survival rate (n = 8 years) for adult males in a nonhunted New Mexico puma population (Logan and Sweanor 2001:127-128). Estimated annual survival rate (n = 2 years) for 5−9 adult males on Uncompahgre Plateau study area was 1.00. Estimated subadult female survival in New Mexico (0.88, n = 16; Logan and Sweanor 2001:122) adjusted downward for potential lower survival for pumas 12-24 months old on Uncompahgre Plateau (0.642, n = 14 females and 10 males combined, life stages not known or described in Anderson et al. 1992:53). Survival of 7 radio-collared pumas (5 males, 2 females) in the subadult stage in the current Uncompahgre Plateau puma study is 1.00. Estimated subadult male survival in New Mexico (i.e., 0.56, n = 9; Logan and Sweanor 2001:122) adjusted upward for potential slightly higher survival for pumas 12-24 months old (i.e., 0.642) on Uncompahgre Plateau (Anderson et al. 1992:53). Survival of 7 radio-collared pumas (5 males, 2 females) in the subadult stage in the current Uncompahgre Plateau puma study is 1.00. Estimated cub survival rate (n = 38 cubs combined sexes), on Uncompahgre Plateau study area. This survival rate is applied to the model starting with the expected number of cubs from birth in RY5. Estimated cub survival for cubs ≥7 months old, and is applied to RY4 cubs only, because the minimum count of pumas in RY4 was tallied when most cub mortality had already occurred. Survival of cubs ≥7 months old in the literature is about 0.95 (Logan and Sweanor 2001). Here, a more conservative 0.90 is used in this model. Reproduction Parameter Adult age Estimate 2+ years Litter size 2.81 Secondary sex ratio observed at nurseries 1:1 Proportion of adult females producing new litters each year 0.65 Parameter Subadult female Estimated Ratio 1.02 Subadult male 0.94 Reference Assume all females 2 years old and older are adults (Logan and Sweanor 2001: 93-94). Average litter size for 21 litters on the Uncompahgre Plateau study area = 2.810 ± 0.9808SD; litters were examined when the cubs were 26 to 42 days old. Secondary sex ratio was 33:26 for 21 litters examined at 29 to 42 days old on the Uncompahgre Plateau study area (not significantly different from 1:1, (X2 = 0.8305 < 3.841, α = 0.05, 1 d.f.). Also see Robinette et al. 1961, Logan and Sweanor 2001:69-70. Proportion of adult females giving birth each year (n = 3 years for ns = 12, 13, 12 females), Uncompahgre Plateau study area. Proportion for a non-hunted puma population in New Mexico was 0.50 (Logan and Sweanor 2001:98). Progeny + Immigrant Recruits /Emigration Ratio Reference No data for pumas in Colorado exists. Assume the ratio of female immigrants to emigrants = 1.02. This ratio is consistent with estimates for a New Mexico puma population that functioned as a source (Sweanor et al. 2000). No data for pumas in Colorado exists. Assume the ratio of male immigrants to emigrants = 0.94, (i.e., male immigration is half of emigration). This ratio is consistent with estimates for a New Mexico puma population that functioned as a source (Sweanor et al. 2000). 136 �Table 15. Summary of puma mother and cub associations by distance (m) during airplane flights, each winter, Uncompahgre Plateau, Colorado. Monitoring period Month No. flights No. puma familiesa Ages of cubs (mo.) No. observations with mothers & cubs 520 m apart Nov. 9, 2005 to Nov. 3 4 2−6 10 Mar. 29, 2006 Dec. 4 4 3−7 16 Jan. 5 4 4−8 16 Feb. 4 5 5−9 16 Mar. 2 5 6−10 9 Totals 18 4−5 2−10 67 Nov. 7, 2006 to Nov. 4 4 2−3 10 Mar. 22, 2007 Dec. 4 4 2−5 11 Jan. 5 3 4−6 9 Feb. 4 4 5−7 9 Mar. 3 1 8 2 Totals 20 1−4 2−8 41 Nov. 13, 2007 to Nov. 2 1 6 1 Feb. 14, 2008 Dec. 0 1 7 NA Jan. 3 1 8 2 Feb. 3 1 9 2 Totals 8 1 6−9 5 a All puma mothers wore GPS-radiocollars. At least 1 cub in the litter wore a VHF radiocollar. b Mean = 1,060 m, SD = 325.99, range = 650−1,600. c Mean = 1,120 m, SD = 1,214.40, range = 616−4,101. d Mean = 1,317 m, SD = 530, range = 750−1,800. No. observations with mothers & cubs >520 m apart 2 4 4 2 0 12b 1 1 3 2 1 8c 1 NA 1 1 3d Table 16. Results of MARK (Cooch and White 2004) simulations to investigate precision as a function of individual capture probabilities and population size. Expected Standard Error Capture Probability (p) 0.2 0.3 0.4 0.5 Large Population (n = 50) 21 9.6 5.5 3.5 Small Population (n = 25) 13 7.8 4.2 2.5 Confidence Interval width Large Population (n = 50) 84 38.4 22 14 137 Small Population (n = 25) 52 31.2 16.8 10 Large Pop. Lower Bound Small Pop. Upper Bound 8 31 39 43 32 29 27 26 �Table 17. Numbers of GPS locations and spans of monitoring for pumas captured on the Uncompahgre Plateau, Colorado, December 2004 to July 2008. Puma I.D. M1 M4 M6 M27 M29 M51 M55 F2 F3 F7 F8 F16 F23 Sex Age stage Dates monitored a M M M M M M M F F F F F F No. locations adult 12-08-04 to 07-20-06 1,797 adult 01-28-05 to 01-14-06 958 adult 02-18-05 to 05-14-08 1,035 adult 03-12-06 to 06-21-06 313 adult 04-14-06 to 01-01-08 1,599 adult 01-07-07 to 05-17-08 1,464 adult 01-21-07 to 05-01-08 1,334 adult 01-07-05 to 05-07-08 3,239 adult 01-21-05 to 04-01-08 3,205 adult 02-24-05 to 07-30-07 2,401 adult 03-21-05 to 10-10-06 1,541 adult 10-12-05 to 04-01-08 2,089 subadult, 01-04-06 to 02-04-06 113 adult 02-05-06 to 05-07-08 746 F24 F adult 01-17-06 to 07-25-07 1,812 F25 F adult 02-09-06 to 04-07-08 1,854 F28 F adult 03-24-06 to 08-15-07 1,499 F30 F adult 03-30-07 to 02-22-08 1,057 F50 F adult 12-14-06 to 03-26-07 352 F52 F subadult 01-10-07 to 05-08-07 383 F54 F adult 01-12-07 to 08-01-08 686 F70 F adult 01-14-08 to 07-31-08 685 F72 F adult 02-12-08 to 07-31-08 737 F75 F adult 03-26-08 to 07-02-08 287 a GPS collars on pumas are remotely downloaded at approximately 1-month intervals. The last date in Dates monitored includes last location from the last GPS data download acquired for an individual puma in this report period. 138 �GOAL: Strategies, Information, & Tools for Managing Healthy, Self-sustaining Puma Populations in Colorado Puma Habitat Puma Population Effects of Harvest & Other Mortality Movements & Corridors Population Dynamics: Sex & Age, Vital Rates, Mortality, Population Growth Rates Vulnerability to Harvest Human Development Habitat Use Effects of Translocation Estimation Methods for Monitoring Domestic Animals Puma― Human Relationships Deer, Elk, Other Natural Prey & Species of Concern Effects of Predation Effects of Evasive Conditioning Map Prey Distribution Habitat Models Population Models Puma Prey Habitat Maps Puma―Prey Relationships Models Figure. 1. An ecologically-based conceptual model of the Colorado Puma Research Program that provides the contextual framework for this and proposed puma research in Colorado. Gray-shaded shapes identify areas of research addressed by puma research on the Uncompahgre Plateau for the puma management goal in Colorado (at top). 139 �Age sh'ucture ofindeJJendent1mmas rn}Jtured and smnJJled for the first time from December 2004 to March 2008,Uncom}Jahgre Plateau. Colorndo. 7 6 5 ltl § 4 >- - - ~ ■ female ~ 0 0 3 >- -- 2 >- -- 1 >- - 0 >- __._______ Male z ~ I- - ' I 1 to 2 >2 lo 3 >3 to 4 -,4 lo 5 >5 to 6 >6 lo 7 >7 to 8 >8 to 9 Age ( years) >9 to 10 Figure 2. Age structure of independent pumas captured and sampled for the first time on the Uncompahgre Plateau, Colorado, December 2004 to March 2008. Puma births. Uncompabgre Plateau. Colorado 8 7 6 l:'.' 5 ':J ci 4 ~ z 3 2 1 'l 0 Jc1n. 11 I - -- ..,_ Feb. Mai. Apr. May June July ■ Bi rths 2005-2008 i: ~ l Aug. Sep. I Oct. Nov. Dec. Births 1983°1987 Figure 3. Puma births detected by month during the current research effort, 2005 to 2008 (n = 28 litters of 15 females), and during the earlier effort by Anderson et al. (1992; 1983 to 1987, n = 10 litters of 8 females), Uncompahgre Plateau, Colorado. 140 �3 Age sh·ucture of independent pumas in Manh 2008, of ~11rvivingpumas raptured ancl sam11Iecl from December 200-t to l\ifarch 2008. while protected from spoTt-lnmting since Ap1"il 2004. Uncompahgre Plateau, Colornclo ■ Fema·lt- 1 T 0 llol >2to3 Male,: Ir ----.- ?3 to4 >4to5 >5to6 >6to7 ?7to8 >8to9 >9to10 101 Age (years) Figure 4. Age structure of surviving independent pumas captured and sampled on the Uncompahgre Plateau, Colorado, in March 2008, and after protection from sport-hunting mortality since April 2004, which includes 4 hunting seasons (Nov. through Mar., 2004-05 to 2007-08). In addition, no other humancaused mortalities were documented in the radio- and GPS-collared sample of independent pumas. This age structure assumes that puma M27 and M29 were alive on March 31, 2008; they each had nonfunctional GPS collars, and were detected alive on 1-22-08 and 1-11-08, respectively. Pumas M5 and M27 range north of the study area and were protected from legal sport-harvest because they are visually tagged animals. Mean ± SD of adult female and adult male ages, respectively: 5.35 ± 2.11 yr. (64.23 ± 25.36 mo.); 4.79 ± 2.17 yr. (57.50 ± 26.06 mo.). 141 �Appendix A. Summary of individual puma cub survival and mortality, December 2004 to 2008, Uncompahgre Plateau, Colorado. Puma I.D. Est. Birth date Est. survival span from 1st capture to fate or last monitor date M5 Estimated Age at capture (days) 183 ~8-1-04 02-04-05 to 04-07-08 F9 31 5-28-05 F10 31 5-28-05 06-27-05 to 4-19-06 06-27-05 to 11-20-05― 12-29-05 M11 31 5-28-05 06-27-05 to 12-2-07 F12 42 5-19-05 07-01-05 to 12-08-05― 01-26-06 F13 42 5-19-05 F14 26 6-26-05 07-01-05 to 08-28-05 07-22-05 to 02-07-06― 03-10-06 M15 26 6-26-05 F17 34 9-22-05 F18 34 9-22-05 M19 34 9-22-05 M20 34 9-22-05 F21 37 9-26-05 Age to last monitor date alive or at death (days, birth to fate) ~1,345 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. Survived to subadult stage by 09-16-05; independent at ~13 mo. old. Dispersed from natal area by 09-29-05 at 14 mo. old. Established territory on NW U.P. Lost contact― shed radiocollar 04-19-06 to 04-26-06. F3 Lost contact― shed radiocollar 08-10-05; last tracks of F10 with mother F2 & siblings F9 & M11 observed 11-20-05. F10 disappeared by 1230-05. Survived to subadult stage by 06-21-06, independent at 13 mo. old. Dispersed from natal area by 07-11-06 at 14 mo. old. Killed by a hunter in SW CO 12-2-07 at 918 days (30 mo.) old Lost contact― shed radiocollar 07-28-05―08-01-05. Tracks of F12 found in association with mother F7 on 12-08-05. F12 disappeared by 01-27-06 when she was not visually observed with F7, and her tracks were not seen in association with F7‘s tracks. Dead; killed and eaten by a puma (sex unspecified) about 8-28-05. Lost contact― shed radiocollar 01-20-06 to 01-25-06. Tracks of F14 were observed with tracks of mother F8 & sibling M15 on 02-07-06. Disappeared by 03-11-06, only tracks of F8 & M15 were found. Lost contact― shed radiocollar 06-06-06 to 06-14-06. F2 F16 308-314 Dead. Lost contact― shed radiocollar 06-06-06 to 0614-06. Killed by a car on highway 550 on 08-18-06. Probably dependent on F16. Dead; probably killed by another puma. Multiple bite wounds to skull. 10 mo. old. Lost contact― shed radiocollar 07-27-06 to 08-02-06. 244-245 Lost contact― shed radiocollar 05-24-06―05-25-06. F16 324 Lost contact; radiocollar quit. Last aerial location 8-1606, live signal. F3 326-333 176-215 918 203-252 101 226-257 07-22-05 to 06-06 to 14-06 10-26-05 to 08-18-06 345-353 10-26-05 to 07-20 to 27-06 10-26-05 to 07-27 to 08-02-06 10-26-05 to 05-24-06 11-02-05 to 08-16-06 301-308 330 142 F2 F2 F7 F7 F8 F8 F16 F16 �Appendix A continued Puma Estimated I.D. Age at capture (days) M22 37 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 9-26-05 M26 183 8-1-05 F33 31 5-30-06 11-02-05 to 12-21-05― 12-22-05 02-08-06 to 03-21 to 24-06 06-30-06 to 07-31-06 F34 31 5-30-06 06-30-06 to 07-31-06 63-65 F35 31 5-30-06 38 F36 29 6-9-06 M37 29 6-9-06 M38 41 7-29-06 06-30-06 to 07-07-06 07-08-06 to 07-28-06 07-08-06 to 07-28-06 09-08-06 to 07-16 to 17-07 M39 29 8-13-06 F40 29 8-13-06 F41 29 8-13-06 M42 29 8-13-06 M43 33 8-13-06 M44 33 8-13-06 09-11-06 to 09-20-06 to 04-25-07 09-11-06 to 09-20-06 to 04-25-07 09-11-06 to 10-05-06 09-11-06 to 11-27-06 09-15-06 03-01-07 09-15-06 to 02-14-07 Age to last monitor date alive or at death (days, birth to fate) 86-87 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. Dead; killed and eaten by male puma 12-21-05―12-2205. F3 ~232-235 Lost contact― shed radiocollar 03-21-06―03-24-06. F25 63-65 Dead. Probably killed and eaten by a male puma 08-01 to 03-06. GPS data on M29 indicate he was not involved. Dead. Probably killed and eaten by a male puma 08-01 to 03-06. GPS data on M29 indicate he was not involved. Dead; research-related fatality.a F23 Dead. Killed and eaten by a male puma 08-22-06. GPS data on M29 indicate he was not involved. Dead. Killed and eaten by a male puma 08-22-06. GPS data on M29 indicate he was not involved. Lost contact― shed radiocollar found 03-06-07. Photo (trail camera in McKenzie Cr.) of M38 & Unm. F sibling with F2 on 7/16-17/07 at 352-353 days old. F28 F8 53-61 106 Lost contact― shed radiocollar by 09-20-06, but seen alive on that date. Tracks of 2 cubs following F8 on 0425-07. Lost contact― shed radiocollar by 09-20-06, but seen alive on that date. Tracks of 2 cubs following F8 on 0425-07. Assumed dead. Lost Contact― shed radiocollar or died (blood on collar) between 10-05-06 (last live signal) & 10-13-06 (collar found); assumed dead. Dead; research-related fatality.b 200 Treed, visually observed 03-01-07. F7 F7 479 Treed, visually observed 02-14-07; sibling (?) M56 also captured, sampled, & marked for 1st time. Killed by Wildlife Services for depredation control on 12/5/07, for killing 4 domestic sheep. 74 74 352-353 9 255 9 255 143 F23 F23 F28 F2 F8 F8 F8 �Appendix A continued Puma Estimated I.D. Age at capture (days) F45 33 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 8-13-06 09-15-06 to 5-20 to 23-07 M46 9-17-06 10-18-06 to 12-15-06 31 Age to last monitor date alive or at death (days, birth to fate) 280-283 89 360 M47 M48 M49 F53 31 31 153 183 9-17-06 9-17-06 7-1-06 7-1-06 M56c 183 ~8-13-06 F57 35 4-16-07 M58 34 5-24-07 10-18-06 to 12-15-06 to 09-12-07 10-18-06 to 12-15-06 to 09-12-07 89 360 89 360 12-05-06 to 07-31-07 to 01-01-07 01-12-07 to 02-23-07 02-14-07 to 03-01-07 05-21-07 to 06-06-07 06-27-07 ~456 42 ~428 subad. 200 52 324 434 144 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. Dead. Multiple puncture wounds on braincase― parietal & occipital regions; consistent with bites from coyote. F45 switched families, moving from F7 to F2 about 1219 to 20-06. Last date F45 was with F2 was 04-17-07. Lost contact― shed radiocollar. Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-12-07 in Puma Canyon. Lost contact― shed radiocollar. Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-12-07 in Puma Canyon. Lost contact― shed radiocollar. Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-12-07 in Puma Canyon. M49 was orphaned when his mother died on about 0326-07; he was ~268 days old. M49 dispersed from natal area and onto NE slope of U.P. Shed radiocollar at a yearling cow elk kill about 10-01-07; he was ~428 days old. Lost contact― shed radiocollar 2-23-07. F53 visually observed by P. & F. Star, on 9-2-07, when F53 was ~14 months old and an independent subadult. F7 Lost contact― shed radiocollar 2-27-07. M56 observed 03-01-07. Lost contact― shed radiocollar 06-07-07. Live mode 06-06-07. Not radio-collared. Tracks of 3 cubs observed with F16‘s tracks on 04-1208, McKenzie Butte-Pinon Ridge Pass. 3 cubs observed with F16 on 08-08-08 by B. & T. Traegde. F3 F3 F3 F50 F54 F7 (?) F25 F16 �Appendix A continued Puma Estimated I.D. Age at capture (days) F59 34 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 5-24-07 06-27-07 to 08-21-07 Age to last monitor date alive or at death (days, birth to fate) 55 324 434 M60 34 5-24-07 06-27-07 to 07-11 to 12-07 06-27-07 to 06-29-07 F61 34 5-24-07 M62 M63 M64 34 34 34 7-14-07 7-14-07 7-14-07 08-17-07 08-17-07 08-17-07 M65 34 7-14-07 08-17-07 F66 37 7-17-07 08-23-07 to 5-31 to 6-1-08 M67 37 7-17-07 08-23-07 M68 37 7-17-07 08-23-07 F74 259 6-1-07 M76 M77 30 30 5-19-08 5-19-08 03-12-08 to 07-09-08 06-18-08 06-18-08 48-49 324 434 262 262 282-283 403 145 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. Alive. Observed alive 11-20-07 with F16, but without siblings M58 & F61. Tracks of 3 cubs observed with F16‘s tracks on 04-12-08, McKenzie Butte-Pinon Ridge Pass. 3 cubs observed with F16 on 08-08-08 by B. & T. Traegde. Dead; research-related mortality.d F16 Radiocollar malfunction. Tracks of 3 cubs observed with F16‘s tracks on 04-1208, McKenzie Butte-Pinon Ridge Pass. 3 cubs observed with F16 on 08-08-08 by B. & T. Traegde. Not radio-collared. Not radio-collared. Not radio-collared. Two out of potential of 4 of F24‘s male cubs were visually observed with her on 4/1/08. Assume that 2 male cubs died before the age of 8.5 mo. Eartags were seen on both cubs, but the numbers were not. Not radio-collared. Two out of potential of 4 of F24‘s male cubs were visually observed with her on 4/1/08. Assume that 2 male cubs died before the age of 8.5 mo. Eartags were seen on both cubs, but the numbers were not. Radio-collared. Lost contact; last location 11/5/07. No signals after that date. F66 was photographed with one male sibling, either M67 or M68, & F30 on 5/31-6/1/08. Not radio-collared. M67 or M68 was photographed with sibling F66 & mother F30 on 5/31-6/1/08. Not radio-collared. M67 or M68 was photographed with sibling F66 & mother F30 on 5/31-6/1/08. One male cub might have died or was not observed. Radio-collared. Shed radiocollar between 7-9-08 and 715-08, probably while still dependent on mother F75. Not radio-collared. Not radio-collared. F16 F16 F24 F24 F24 F24 F30 F30 F30 F75 F2 F2 �Appendix A continued Puma Estimated I.D. Age at capture (days) F78 30 M79 30 F80 40 F81 40 M82 37 M83 37 M84 36 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 5-19-08 5-19-08 5-23-08 5-23-08 5-29-08 5-29-08 6-5-08 06-18-08 06-18-08 07-02-08 07-02-08 07-05-08 07-05-08 07-11-08 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. Not radio-collared. F2 Not radio-collared. F2 Not radio-collared. F23 Radio-collared. F23 Radio-collared. F8 Not radio-collared. F8 ~69 Radio-collared 7-11-08 to 7-22-08; collar removed F70 because of malfunction. Not radio-collared after 7-22-08. Eartag of M84 was found by E. Phillips on 8-25-08; assuming M84 died, he probably died around 8-13-08 when cub F85 was located ~340m south of the eartag in the East fork Dolores Cyn. F85 36 6-5-08 07-11-08 Radio-collared. F70 F86 36 6-5-08 07-11-08 Radio-collared 7-22-08. F70 M87 28 7-3-08 07-31-08 Not radio-collared. F3 M88 28 7-3-08 07-31-08 Not radio-collared. F3 F89 28 7-3-08 07-31-08 Radio-collared F3 M90 36 7-9-08 08-14-08 Radio-collared F72 7MA 28-35 7-10-08 08-08 to 13-08 Examined, but not tagged. F7 7MB 28-35 7-10-08 08-08 to 13-08 Examined, but not tagged. F7 7FC 28-35 7-10-08 08-08 to 13-08 Examined, but not tagged. F7 a Cub F35 probably starved between 06-30-06 & 07-07-06 after the transmitter on the expandable collar got in its mouth. b Cub M42 died after being captured by dogs, probably from stress of capture associated with severe infection of laceration under right foreleg caused by expandable radiocollar. c Cub M56 was captured in association with F7 and her cubs M43 and M44. He may have been missed at the nursery when M43 and M44 were initially sampled and marked. d Cub M60 died probably of starvation. The expandable radiocollar was around the neck and right shoulder, possibly restricting movement. 146 �Appendix B. Puma population models and simulation results as preliminary assessments of current CDOW puma management assumptions and population manipulations for the treatment period. Modeling Scenarios We modeled a set of scenarios that pertain to current CDOW puma management assumptions and to potential puma research direction on the Uncompahgre Plateau for the treatment period: 1) Puma population dynamics without hunting-caused mortality. 2) Puma harvest that would induce a stable (i.e., no growth) phase to identify a population tipping point induced by harvest mortality. 3) Puma harvest at the upper limit (i.e., 15% of 8-15% range, CDOW 2007) that CDOW assumes would result in a stable to increasing puma population, with harvest apportioned equally among independent males and females. 4) Puma harvest at the upper limit (i.e., 15% of 8-15% range, CDOW 2007) that CDOW assumes would result in a stable to increasing puma population, but with harvest comprised of 40% females and 60% males, which is consistent with the sex composition of puma harvest in Colorado. 5) Puma harvest at the upper limit (i.e., 28% of 16-28% range, CDOW 2007) that CDOW assumes would result in a declining puma population, with harvest apportioned equally among independent males and females. 6) Puma harvest at the upper limit (i.e., 28% of >15-28% range, CDOW 2007) that CDOW assumes would result in a declining puma population, but with harvest comprised of 40% females and 60% males, which is consistent with the sex composition of puma harvest in Colorado. 7) A harvest scenario applied the historic puma harvest on the study area. Puma mortality data for the study area during the 10 years previous (i.e., 1994-2003) to the beginning of the reference period was quantified after carefully geo-referencing mortality locations on the study area (see last table in Appendix B). Model parameters from those data include: mortality rate of 14.3 independent puma mortalities per year (rounded to 14/yr.), and sex proportions of 55% males and 45% females. No other puma population data or parameter estimates were available for the study area at that time. Therefore, the scenario that was modeled pertained to the expected impact of the average annual puma mortality of independent pumas (i.e., adults and subadults) if the hypothetical population was the same as the non-hunted minimum expected puma population in treatment period year 1 (i.e., TY1). A harvest of 14 pumas per year is a 26% harvest rate on the expected TY1 non-hunted minimum independent puma population (i.e., 14/53). Another way of stating this scenario is; what would occur if puma harvest was applied to the puma population on the study area during the treatment period at the average rate of puma mortality that was recorded during 1994 to 2003? Results of Puma Population Simulations The following tables contain the expected minimum population sizes for independent pumas and annual rates of population increase for independent pumas conditional upon the minimum number of independent pumas detected in Reference Year 4 (RY4) and the model input parameters and assumptions (Table 14, this report). Notes below each table explain how results may be interpreted relative to other research results on puma population dynamics and specific CDOW puma management assumptions. The harvest levels for each model are clearly stated in the left column of each table. Simulations involving harvest apply the harvest following reference year 5 (RY5) and starting with treatment year 1 (TY1). 147 �Projected Minimum Puma Population Size Independent Pumas Adult Subadult Cub Year Female Male Female Male F&M Total Lambda RY4 16 8 5 4 20 33 RY5 18 10 9 8 33 45 1.27 TY1 23 14 8 8 42 53 1.14 TY2 27 17 11 10 49 64 1.18 TY3 32 22 12 11 58 77 1.17 TY4 38 27 15 14 69 92 1.17 TY5 44 32 17 16 81 110 1.16 Note: Expected lambda for the modeled non-hunted puma population on the Uncompahgre Plateau approach the high range of observed average annual rates of population increase for a non-hunted puma population in good quality habitat in southern New Mexico (i.e., r = 0.21, n = 4; r = 0.28, n = 4; r = 0.17, n = 4; r = 0.11, n = 7; Logan and Sweanor 2001:169-175). Puma population growth could be higher on the Uncompahgre Plateau because of higher quality habitat (i.e., greater prey biomass), and if puma sources are nearby to the study area. Harvest Level No harvest. Projected Minimum Puma Population Size Independent Pumas Adult Subadult Cub Year Female Male Female Male F&M Total Lambda RY4 16 8 5 4 20 33 RY5 18 10 9 8 33 45 1.27 TY1 19 12 7 6 35 44 0.98 TY2 19 12 8 7 34 45 1.02 TY3 19 13 7 7 34 46 1.01 TY4 19 13 7 7 34 46 1.01 TY5 19 14 7 7 34 46 1.00 Note: The tipping point of population stability and decline is expected to be about 16% harvest of independent male and female pumas, consistent with current CDOW puma harvest assumptions. Harvest Level 16% of independent pumas, sexes are harvested equally; i.e., stable phase model. Projected Minimum Puma Population Size Independent Pumas Adult Subadult Cub Year Female Male Female Male F&M Total Lambda RY4 16 8 5 4 20 33 RY5 18 10 9 8 33 45 1.27 TY1 19 12 7 7 36 45 0.99 TY2 19 12 8 7 35 47 1.03 TY3 19 13 8 7 36 47 1.02 TY4 20 14 8 7 36 48 1.02 TY5 20 14 8 7 36 49 1.01 Note: This result is consistent with the current CDOW puma harvest assumption for a stable-to-increasing population, with very slow growth attributed to equal harvest of females and males. Harvest Level 15% of independent pumas, sexes are harvested equally. 148 �Projected Minimum Puma Population Size Independent Pumas Adult Subadult Cub Year Female Male Female Male F&M Total Lambda RY4 16 8 5 4 20 33 RY5 18 10 9 8 33 45 1.27 TY1 21 11 8 6 38 45 0.99 TY2 22 10 9 7 39 47 1.05 TY3 23 10 9 7 42 50 1.05 TY4 25 11 10 8 45 53 1.05 TY5 26 11 10 8 48 56 1.06 Note: This result is consistent with the current CDOW puma harvest assumption for a stable-to-increasing population, with increased growth due to reduced female mortality. Harvest Level 15% of independent pumas, comprised of 40% females & 60% males. Projected Minimum Puma Population Size Independent Pumas Adult Subadult Cub Year Female Male Female Male F&M Total Lambda RY4 16 8 5 4 20 33 RY5 18 10 9 8 33 45 1.27 TY1 17 10 6 6 30 38 0.81 TY2 14 9 6 5 25 33 0.86 TY3 12 8 5 4 22 29 0.84 TY4 10 7 4 4 18 25 0.84 TY5 9 6 3 3 16 21 0.84 Note: This result is consistent with the current CDOW puma harvest assumption for a declining population. Harvest Level 28% of independent pumas, sexes are harvested equally. Projected Minimum Puma Population Size Independent Pumas Adult Subadult Cub Year Female Male Female Male F&M Total Lambda RY4 16 8 5 4 20 33 RY5 18 10 9 8 33 45 1.27 TY1 19 8 7 4 34 38 0.81 TY2 18 6 7 5 32 35 0.92 TY3 17 5 7 4 31 33 0.93 TY4 16 4 6 4 30 31 0.95 TY5 16 4 6 4 29 30 0.95 Note: This result is consistent with the current CDOW puma harvest assumption for a declining population even with harvest weighted toward males. Harvest Level 28% of independent pumas, comprised of 40% females & 60% males. Projected Minimum Puma Population Size Harvest Independent Pumas Adult Subadult Cub Level Year Female Male Female Male F&M Total Lambda 26% of RY4 16 8 5 4 20 33 independent RY5 18 10 9 8 33 45 1.27 pumas, TY1 18 9 7 6 33 41 0.89 comprised of TY2 17 8 7 6 31 39 0.94 45% females TY3 16 8 7 6 30 36 0.94 & 55% males; TY4 16 7 7 5 28 35 0.95 i.e. historical harvest model TY5 15 7 6 5 27 33 0.95 Note: Results of this model indicate that the expected outcome is that the puma population would decline. 149 �Appendix B (continued). Puma mortality data for portions of Game Management Units (GMUs) 61, 62, 70 that comprise the Uncompahgre Plateau Study Area, 1994-2003. GMU Year Adult Male Subadult Male Adult Female Subadult Female Subtotals 61 2003 4 2 3 0 9 62 2003 1 1 1 3 6 70 2003 0 0 0 0 0 61 2002 1 0 2 0 3 62 2002 0 0 3 1 4 70 2002 1 0 0 0 1 61 2001 4 0 5 0 9 62 2001 2 1 2 1 6 70 2001 1 0 1 0 2 61 2000 5 0 1 2 8 62 2000 0 0 0 0 0 70 2000 0 0 1 1 2 61 1999 3 1 3 0 7 62 1999 2 0 1 0 3 70 1999 2 0 1 0 3 61 1998 3 1 3 1 8 62 1998 3 1 0 0 4 70 1998 1 0 3 0 4 61 1997 5 1 1 0 7 62 1997 2 0 2 1 5 70 1997 1 0 0 0 1 61 1996 3 0 2 0 5 62 1996 2 1 3 0 6 70 1996 1 0 0 0 1 61 1995 6 1 4 0 11 62 1995 9 0 4 0 13 70 1995 1 0 0 0 1 61 1994 2 0 3 0 5 62 1994 3 1 4 0 8 70 1994 0 0 1 0 1 Subtotal 68 11 54 10 143 Total 79 males (55%) 64 females (45%) 14.3/yr. Note: Nine puma records did not designate adult or subadult age stages. Those data were determined with a cointoss for this table, resulting in 6 males designated as 3 adults and 3 subadults, and 3 females designated as 1 adult and 2 subadults. Three mortalities were recorded as ― road-kills‖ (1 subadult male, 2 subadult females). Two adult male deaths were recorded as ― other‖. Two adult male deaths were recorded as ― landowner‖. All other deaths were recorded as ― hunter harvest‖. Source of records: Colorado Division of Wildlife, 6060 Broadway, Denver, CO, and K. Crane, CDOW DWM, Ridgway. 150 �Appendix C. Collaborative project on disease surveillance in wild felids. College of Veterinary Medicine and Biomedical Sciences Department of Microbiology, Immunology & Pathology 1619 Campus Delivery Fort Collins, CO 80523-1619 970-491-6144 (voice) 970-491-0603 (fax) TO: Ken Logan, Mammals Researcher, Colorado Division of Wildlife, Montrose, CO. FROM: Sue VandeWoude, DVM, Associate Professor, DMIP RE: Disease Seroprevalence in UP Pumas DATE: August 26, 2007 Attached please find the consolidated report on infectious disease surveillance for the mountain lion samples you have provided to our laboratory as an adjunct to your CDOW ongoing studies. Our laboratory has performed puma-lentivirus (PLV) antibody screening using a sensitive western blot assay developed in our laboratory and found 13 of 18 samples conclusively positive (72%), with two additional samples inconclusive and one not tested. Dr. Michael Lappin, a veterinary internal medicine specialist with expertise in feline infectious disease has tested a subset of 6 samples for antibodies to Feline Calicivirus (FCV), Feline Herpes Virus (FHV), Feline parvovirus (FPV), Toxoplasma gondii (IgM, indicating recent infection, IgG indicating past exposure), and Bartonella hensalae (the agent associated with cat scratch disease). At least one of six animals tested has been positive for each of these agents. Further results are pending from the remaining samples you have provided for these 5 assays. In addition, Dr. Martin Scriefer at Fort Collins CDC has also tested 6 animals for evidence of antibodies to the agent responsible for plague (Yersinia pestis). Interestingly, 3 of 6 animals demonstrate significant exposure to this agent as well. These specific agents were selected for analysis in order to provide a variety of types of agents (viruses: PLV, FCV, FHV, FPV; bacteria: Bartonella henselae and Yersinia pestis; and coccidian: T. gondii), a variety of modes of transmission (direct intra-specific contact, PLV; direct contact with domestic cats, FCV, FHV, FPV; arthropod transmission, B. henselae, Y. pestis; prey ingestion, T. gondii, Y. pestis). Further, at least three of these agents (PLV, FCV, B. henselae) result in chronic infections, allowing the possibility of determining genetic relatedness among organisms isolated from different individuals, and three of these agents (B. henselae, Y. pestis, T. gondii) are also potential zoonotic agents. As you are aware, our laboratory has recently been awarded a 5 year NSF Ecology of Infectious Disease grant entitled, ― The effects of urban fragmentation and landscape connectivity on disease prevalence and transmission in North American felids‖, with co-PI Dr. Kevin Crooks, an associate professor in the Warner College of Natural Resources at CSU. The aims of this grant are to model the effects of urbanization and resultant habitat fragmentation on disease dynamics in large carnivore species as described on the following page. The letter of support provided by you and Mr. Dave Freddy were pivotal in demonstrating a large cohort of capable and active field collaborators willing to provide samples to support our studies. The mountain lion field work being led by your team, and the newly initiated studies by your colleague, Dr. Mat Alldredge, have provided us with renewed enthusiasm for developing our collaborations to support the goals of our study. We foresee the opportunity to interact in a mutually beneficial partnership to further the goals of all of our studies, and to maximize the information that can be gleaned about these important and ecologically significant species. 151 �We anticipate that the data we are generating will be useful for comparative seroprevalence of different geographic populations of bobcats and pumas, and for genetic phenotyping of pathogens to compare relationships among diseases spread by arthropod vectors, domestic cats, feral rodents, and inter-specific contacts. As we discussed during your recent visit to CSU, these samples are most valuable to us if we can receive them directly as quickly as possible after collection. I have provided an SOP providing information about the types of samples that will be most valuable, and a draft of a ‗permissions‘ document that you can use with each sample submission to provide us with guidance for any testing that is permissible on the materials we receive. This latter document will be filed and recorded electronically. We will continue to provide annual updates and communications about any publications that utilize the data resulting from your samples. Again thank you for providing these extremely valuable samples, and we look forward to our continued collaborations. Sincerely, Sue VandeWoude The effects of urban fragmentation and landscape connectivity on disease prevalence and transmission in North American felids Project Summary Sue VandeWoude (co-PI), Kevin Crooks (co-PI), Michael Lappin, Mo Salman, Walter Boyce, Ken Logan, Mat Alldredge, Carolyn Krumm, Don Hunter, Lisa Lyren, Seth Riley, Jennifer Troyer The objective of this study is to model the effects of urbanization and resultant habitat fragmentation on disease dynamics in carnivore species. Bobcats, puma, and domestic cats will be evaluated simultaneously in three divergent ecosystems: high mountain desert (Colorado), everglades (Florida), and Mediterranean scrub habitat (California). The research will: 1) assess the relationship between habitat fragmentation and prevalence of viral, bacterial, and parasitic pathogens across a gradient of urbanization, 2) use transmission dynamics of selected disease agents as markers of connectivity of fragmented populations, and 3) evaluate the effect of urbanization on the incidence of cross-species disease transmission. The results of this research will give wildlife managers a better understanding of how urbanization affects their local wildlife and assist them in future disease management planning. The combination of a uniquely qualified, broadly based research team with an extensive dataset on carnivores from across the country presents an unprecedented opportunity to investigate the disease dynamics in these rare and difficult to study species. The research efforts of each regional team will support and provide new insights for all of the regions involved, not simply their own. Training of graduate students in ecology, infectious disease, and epidemiology will be emphasized, as will training for pre- and post-doctoral veterinarians. Results will be made widely available to other scientists, conservation practitioners, and the general public. This research has a tremendous capacity to broadly impact areas of public and post-graduate education, career development for new investigators and persons from underrepresented groups, and to enhance understanding of complex infectious disease ecological problems using extensive multi-disciplinary collaborations. 152 �Appendix C (continued). Preliminary results of infectious disease surveillance for puma, Uncompahgre Plateau, Colorado, 2005-2008. Puma ID UPCO2 UPCO3 UPCO7 UPCO7 UPCO7 UPCO8 UPCO4 UPCO5 UPCO6 UPCO6 UPCO23 UPCO25 UPCO28 UPCO29 UPCO31 UPCO23 UPCO27 UPCO30 UPCO50 UPCO51 UPCO52 UPCO54 UPCO55 UPCO24 UPCO69 UPCO70 UPCO71 UPCO72 UPCO73 UPCO74 UPCO75 a Sex F F F F F F M M M M F F F M M F M F F M F F M F M F M F F F F Capture Date 1/8/2008 1/21/2005 2/24/2005 3/30/2006 3/3/2007 3/21/2005 1/28/2005 2/4/2005 2/18/2005 4/12/2008 2/25/2008 2/8/2006 3/23/2006 4/14/2006 4/19/2006 1/4/2006 3/10/2006 4/15/2006 12/14/2006 1/7/2007 1/10/2007 1/12/2007 1/21/2007 1/17/2006 1/11/2008 1/20/2008 1/29/2008 2/12/2008 2/21/2008 3/12/2008 3/26/2008 GPS NAD27 U.T.M.: Zone, E, N 13S, 245722, 4244166 13S, 241606, 4251510 13S, 246328, 4244230 13S, 245901, 4247627 13S, 247645, 4246097 12S, 727808, 4239029 13S, 257565, 4239606 13S, 240577, 4251037 13S, 247399, 4254006 13S, 257516, 4239696 12S, 723304, 4242231 13S, 258374, 4230480 12S, 722868, 4240115 12S, 723458, 4242340 12S, 746919, 4225441 12S, 730188, 4234861 12S, 722339, 4245212 13S, 248551, 4242095 12S, 753639, 4260149 13S, 238783, 4252390 13S, 258058, 4236260 13S, 252688, 4228050 13S, 258133, 4228691 12S, 737151, 4233273 13S, 248191, 4246810 13S, 247122, 4245760 12S, 754611, 4256842 13S, 258294, 4234597 12S, 728576, 4241799 12S, 729678, 4239555 12S, 732894, 4239423 PLV + + + Ih I + + P P + + + + + + I + + + + + P P P P P a FCV +h + Ph P P P P P P P P P P P P P P P P + + P P P P P b PLV is Puma Lentivirus. FCV is Feline Calicivirus. c FHV is Feline Herpesvirus. d FPV is Feline Panleukopenia Virus e T. g. is Toxoplasma gondii. f B. h. is Bartonella hensalae. g Y. p. is Yersinia pestis. h Results: + (positive result), P (Pending result), I (Inconclusive result). b 153 FHV + P P + P P P P P P P P P P P P P P P + + P P P P P c FPV + P P + P P P P P P P P P P P P P P P + + P P P P P d T.g. e IgM + P P P P P P P P P P P P P P P P P P P P P P T.g.e IgG + + P P + + + + P P P P P P P P P P P P P P P + + P P P P P B.h. Y.p. P P + P P P P P P P P P P P P P P P P P P P P + ++ +++ P P ++ I I I + + P P P P P P P P P P P P P + + + f g �Colorado Division of Wildlife July 2008 –July 2009 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 3003 1 Federal Aid Project No. N/A : : : : : Division of Wildlife Mammals Research Carnivore Conservation Puma Population Structure and Vital Rates On the Uncompahgre Plateau Period covered: July 31, 2008−July 31, 2009 Author: K. A. Logan. Personnel: K. Logan, B. Dunne, D. Ranglack, J. Timmer, S. Waters, K. Crane, T. Mathieson, M. Caddy, and T. Bonacquista of CDOW; S. Young and W. Wilson of U.S.D.A. Wildlife Services; houndmen R. Navarette and J. Knight; volunteers and cooperators including: private landowners, Bureau of Land Management, Colorado State Parks, Colorado State University and U.S. Forest Service. Supplemental financial support received in previous years from The Howard G. Buffett Foundation and Safari Club International Foundation. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT This report provides information in the fifth year of the reference period August 2008 through July 2009 on puma population characteristics and dynamics on the Uncompahgre Plateau. Field operations were impacted by a state government issued hiring freeze that did not allow full staffing of 2 puma capture teams during winter 2008-09. All capture efforts involving use of trained dogs, cage traps, and inspections at nurseries in 2008-09 resulted in a total of 37 puma captures (7 adult females [1 adult female captured 3 times, another captured twice], 4 adult males [1 adult male captured 3 times], 1 subadult female, and 18 cubs [2 of them captured twice each]). Five adults (4 females, 1 male) and 14 cubs were captured and marked for the first time. As of July 2009, there were 17 adults (11 females, 6 males), 1 subadult female, and 5 cubs (2 females, 3 males) with active radio-collars. Efforts to capture, sample, and mark pumas with the use of trained dogs extended from December 9, 2008 to April 30, 2009. Those efforts resulted in 71 search days, 198-202 puma tracks detected, 75-78 pursuits, and 24 puma captures. In 2008-09, capture efforts with ungulate carcasses and cage traps resulted in captures of 2 adult females and 1 subadult female. Capture and search efforts from November 2008 through March 2009 enabled us to estimate a minimum of 37 independent pumas detected on the Uncompahgre Plateau study area during that time, including 26 females and 11 males. Preliminary puma population parameters estimated during the past 4.7 years of research, included: population sex and age structure, reproduction rates, and survival rates. Data on puma reproduction rates included: average litter size = 2.77 ± 0.9081 SD, n = 26; average birth interval (mo.) = 18.462 ± 4.6035 SD, n = 16; average proportion of adult females producing cubs each year = 0.598 ± 0.1094 SD, n = 11-13 females per yr. for 4 years; secondary sex ratio = 41:31, consistent with 1:1; and average gestation length (day) = 90.5-92.3(SD = 2.5495, 125 �2.1628, respectively). Puma births occurred March through September, with 24 of 27 occurring May through September. Majority of breeding activity was February through June. Preliminary estimates of survival rates for both adult and subadult pumas in this reference period were high, and may reflect the absence of puma sport-hunting as a mortality factor. An increasing age structure of independent pumas in the reference period reflects the high survival rates. Cub survival was about 0.53 (SE = 0.1623-0.1629; Kaplan-Meier procedure) and 0.58 (± 0.1610 95% CI; binomial model). The main cause of mortality in the adults and cubs was aggression by other pumas. Dispersal from the Uncompahgre Plateau study area was documented for 8 pumas (7 male, 1 female) that dispersed during the subadult stage and moved distances ranging from about 61 to 330 linear km. We monitored 7 puma families with a radio-collared mother and at least one radio-collared cub to assess association distances during aerial locations from November 6, 2008 to March 20, 2009. The aggregate data gathered during the past 4 winters generally indicate that mothers were usually within 660 m of their cubs during the day. Preliminary comparisons between our current puma research on the Uncompahgre Plateau (4.7 years duration) and results of the Anderson et al. (1992) puma research on the plateau (7 years duration 1981-1988) were made where appropriate. Data on puma population characteristics and dynamics gathered during the reference period was used for a preliminary assessment of population-based assumptions used by CDOW to guide puma hunting management and indicated that assumptions pertaining to puma population sex and age structure, density, and expected results from modeled harvest rates are biologically supported. The CDOW structured the puma hunting season for the treatment period. The first hunting season will begin midNovember 2009 and extend to January 31, 2010 unless the quota is filled earlier. The management objective will be to achieve a stable to increasing puma population. Population model simulations indicated a harvest quota of 8 independent pumas to achieve the objective. No limit of hunters on the study area is imposed, but each hunter is required to obtain a hunting permit for the study area. In addition, an effort will be made to survey each hunter obtaining a valid permit. All pumas harvested in and around the study area will be inspected by CDOW personnel. A study plan for the treatment period was submitted for internal review in the CDOW. The plan was substantially modified and received another internal review. That version will be modified and submitted to the Mammals Research leader in fall 2009. Continuing this research includes manipulating the puma population with sport-hunting in the treatment period while also estimating puma population characteristics and vital rates. We are continuing to collaborate with colleagues in Mammals Research and at Colorado State University to assess puma population dynamics and social structure, puma-human interactions, health, habitat use, and we will incorporate a pilot project to examine individual puma detection rates using a camera grid design. 126 �WILDLIFE RESEARCH REPORT PUMA POPULATION STRUCTURE AND VITAL RATES ON THE UNCOMPAHGRE PLATEAU, COLORADO KENNETH A. LOGAN P. N. OBJECTIVE Quantify puma population sex and age structure; estimate puma population vital rates, including: reproduction rates of females, age-stage survival rates, and immigration and emigration rates; quantify agent-specific mortality rates; model puma population dynamics; and plan for the remaining 5 years of the Uncompahgre Plateau Puma Project― all to improve the Colorado Division of Wildlife’s (CDOW) model-based approach to managing pumas in Colorado. SEGMENT OBJECTIVES 1. 2. 3. 4. 5. 6. 7. 8. 9. Continue gathering data on puma population sex and age structure. Continue gathering data for estimates of puma reproduction rates. Continue gathering data to estimate puma sex and age-stage survival rates. Continue gathering data on agent-specific mortality. Gather data on spatial relationships of puma mothers to their cubs during the Colorado puma hunting season as a preliminary assessment of the vulnerability of puma mothers to sport-hunting harvest. Use data on population dynamics for a preliminary evaluation of assumptions used by CDOW biologists and managers in the Data Analysis Unit puma management planning process. Work with CDOW biologists and managers to structure the puma hunting manipulation for the first year of the 5 year treatment phase. Develop a study plan for remaining 5 years of puma population research on the Uncompahgre Plateau Study Area. Collaborate with other researchers and evaluate other data sources that could be relevant to CDOW biologists and managers. INTRODUCTION Colorado Division of Wildlife managers need reliable information on puma biology and ecology in Colorado to develop sound management strategies that address diverse public values and the CDOW objective of actively managing pumas while “achieving healthy, self-sustaining populations”(CDOW 2002-2007 Strategic Plan:9). Although 4 puma research efforts have been made in Colorado since the early 1970s and puma harvest data is compiled annually, reliable information on certain aspects of puma biology and ecology, and management tools that may guide managers toward effective puma management is lacking. Mammals Research staff held scoping sessions with a number of the CDOW’s wildlife managers and biologists. In addition, we consulted with other agencies, organizations, and interested publics either directly or through other CDOW employees. In general, CDOW staff in western Colorado highlighted concern about puma population dynamics, especially as they relate to their abilities to manage puma populations through regulated sport-hunting. Secondarily, they expressed interest in puma―prey interactions. Staff on the Front Range placed greater emphasis on puma―human interactions. Staff in both eastern and western Colorado cited information needs regarding effects of puma harvest, puma population monitoring methods, and identifying puma habitat and landscape linkages. Management needs 127 �identified by CDOW staff and public stakeholders form the basis of Colorado’s puma research program, with multiple lines of inquiry (i.e., projects): Improve our ability to manage puma hunting with enhanced scientific bases, strategies, and tools― ● Puma population characteristics (i.e., density, sex and age structure). ● Puma population dynamics and vital rates (i.e., birth rates, survival rates, emigration rates, immigration rates, population growth rates). ● Field methods and models for assessing and tracking changes in puma populations. ● Relative vulnerability of puma sex and age classes to hunter harvest. Improve our understanding of puma habitat needs and interrelationships of puma management units― ● Puma habitat use, movements, and use of landscape linkages. ● Puma recruitment patterns (i.e., progeny, immigration, emigration). ● Models for identifying puma habitat and landscape linkages. Improve our understanding of the puma’s role in the ecology of other species ● Relationships of puma to mule deer, elk, and other natural prey. ● Relationships of puma to species of special concern, e.g., desert bighorn sheep. Improve our understanding of puma-human interactions and abilities to manage them ● Behavior of puma in relation to people and human facilities. ● Puma predation on domestic animals. ● Effects of translocating nuisance pumas. ● Effects of aversive conditioning on pumas. While all projects cannot be addressed concurrently, understanding their relationships to one another is expected to help individual projects maximize their benefits to other projects that will assist the CDOW to achieve its strategic goal in puma management (Fig.1). Management issues identified by managers translate into researchable objectives, requiring descriptive studies and field manipulations. Our goal is to provide managers with reliable information on puma population biology and to develop useful tools for their efforts to adaptively manage puma in Colorado to maintain healthy, self-sustaining populations. The highest-priority management needs are being addressed with this intensive population study that focuses on puma population dynamics using sampled, tagged, and GPS/radio-collared pumas. Those objectives include: Describe and quantify puma population sex and age structure. Estimate puma population vital rates, including: reproduction rates, age-stage survival rates, emigration rates, immigration rates. Estimate agent-specific mortality rates. Improve the CDOW’s model-based management approaches with Colorado-specific data from objectives 1―3. Consider other useful models. Concurrently with the tasks associated with the objectives above, significant progress will be made toward a 5th objective, which will initially be subject to pilot study― develop methods that yield reliable estimates of puma population abundance. A descriptive study will estimate population parameters in an area that appears typical of puma habitat in western Colorado and will yield defensible population parameters based upon contemporary Colorado data. This study will be conducted in a 5-year reference period (i.e., absence of recreational hunting) to allow puma life history traits to interact with the main habitat factors that appear to influence puma population growth (e.g., prey availability and vulnerability, Pierce et al. 2000, Logan and Sweanor 2001). Contingent upon results in the reference period, a subsequent 5-year treatment period is planned. 128 �The treatment period will involve the use of controlled recreational hunting to manipulate the puma population. TESTING ASSUMPTIONS AND HYPOTHESES Hypotheses associated with main objectives 1―5 of this puma population research are structured to test assumptions guiding puma management in Colorado. 1. Considering limitations (i.e., methods, number of years, assumption violations) to the previous Colorado-specific studies on puma populations (Currier et al. 1977, Anderson et al. 1992, Koloski 2002), managers assume that puma population densities in Colorado are within the range of those quantified in more intensively studied populations in Wyoming (Logan et al. 1986), Idaho (Seidensticker et al. 1973), Alberta (Ross and Jalkotzy 1992, and New Mexico (Logan and Sweanor 2001). The CDOW assumes density ranges of 2.0−4.6 puma/100 km2 (i.e., includes pumas of all age stages- adults, subadults, and cubs, J. Apker, CDOW Carnivore Biologist, person. commun. Nov. 19, 2003) to extrapolate to DAUs to guide the model-based quota-setting process. Likewise, managers assume that the population sex and age structure is similar to puma populations described in the intensive studies. Using intensive efforts to capture, mark, and estimate non-marked animals developed and refined during the study to estimate the minimum puma population, the following will be tested: H1: Puma densities during the 5-year reference period (absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will vary within the range of 2.0 to 4.6 puma/100 km2 and will exhibit a sex and age structure similar to puma populations in Wyoming, Idaho, Alberta, and New Mexico. 2. Recreational puma hunting management in Colorado Data Analysis Units (DAUs) is guided by a model to estimate allowable harvest quotas to achieve one of two puma population objectives: 1) maintain puma population stability or growth, or 2) cause puma population decline (CDOW, Draft L-DAU Plans, 2004, CDOW 2007). Basic model parameters are: puma population density, sex and age structure, and annual population growth rate. Parameter estimates are currently chosen from literature on studies in western states that are judged to provide reliable information. Background material used in the model assumes a moderate annual rate of growth of 15% (i.e., λ = 1.15) for the adult and subadult puma population (CDOW 2007). This assumption is based upon information with variable levels of uncertainty (e.g., small sample sizes, data from habitats dissimilar to Colorado). Parameters influencing λ include population density, sex and age structure, female age-at-firstbreeding, reproduction rates, sex- and age-specific survival, immigration and emigration. H2: Population parameters estimated during a 5-year reference period (in absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will yield an estimated annual adult plus subadult population growth rate that will match or exceed λ = 1.15. 3. The key assumption is that the CDOW can manage puma population growth through recreational hunting on the basis that for a stable puma population hunting removes the annual increment of population growth (i.e., from current judgments on population density, structure, and λ). Puma harvest rate formulations for DAUs assumes that total mortality (i.e., harvest plus other detected deaths) in the range of 8 to 15% of the harvest-age population (i.e., independent pumas comprised of adults plus subadults) with the total mortality comprised of 35 to 45% females (i.e., adults and subadults) is acceptable to manage for a stable-to-increasing puma population (CDOW 2007). 129 �H3: Total mortality of an estimated 15% of the adults and subadults with no more than 45% of the total mortality comprised of females will not result in a decline of the harvest-age segment of the population by the beginning of the next hunting season. 4. To reduce a puma population, hunting must remove more than the annual increment of population growth. For DAUs with the objective to suppress the puma population, the total mortality guide of greater than 15 to 28% of the harvest-age population with greater than 45% comprised of females is suggested (CDOW 2007). H4: Total mortality of an estimated 16% or greater of the harvestable population with greater than 45% females will cause a decline in the abundance of harvest-age pumas (i.e., adults and subadults). 5. The increase and decline phases of the puma population make it possible to test hypotheses related to shifts in the age structure of the population which have been linked to harvest intensity in Wyoming and Utah. H5: The puma population on the Uncompahgre Plateau study area will exhibit a young age structure after hunting prohibition at the beginning of the reference period. During the 5 years of hunting prohibition, greater survival of independent pumas will cause an older age structure in harvest-age pumas (i.e., adults and subadults) as suggested by the work of Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. As hunting is re-instated in the treatment period, the age structure of harvested pumas and the harvest-age pumas in the population will decline as observed by Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. Desired outcomes and management applications of this research include: 1. Quantification of variations in puma population density, sex and age structure, growth rates, vital rates, and an understanding of factors affecting them will aid adaptive puma management by yielding population parameters and tools useful for assessing puma population dynamics, evaluation of management alternatives, and effects of management prescriptions. 2. Testing assumptions about puma populations, currently used by CDOW managers, will help managers to biologically support and adapt puma management based on Colorado-specific estimated puma population characteristics, parameters, and dynamics. 3. Methods for assessing puma population dynamics will allow managers to evaluate modeled populations and estimate effects of management prescriptions designed to achieve specified puma population objectives in targeted areas of Colorado. Ascertaining puma numbers and densities during the project will allow assessment of monitoring techniques. Potential methods include use of harvest sex and age structure and photographic and DNA genotype capture-recapture. Study plans to develop and test feasible field and analytical methods will be developed in the future after we have learned the logistics of performing those methods, after we have preliminary data on puma demographics and movements which will inform suitable sampling designs, and if we have adequate funding. 4. This information will be disseminated to citizen stakeholders interested in pumas in Colorado, and thus contribute to informed public participation in puma management. STUDY AREA The study area for the puma population research is on the Uncompahgre Plateau (in Mesa, Montrose, Ouray, and San Miguel Counties; Fig. 2). The study area includes about 2,253 km2 (870 mi.2) of the southern halves of Game Management Units (GMUs) 61 and 62, and about 155 km2 (60 mi.2) of the northern edge of GMU 70 (between state highway 145 and San Miguel River). The area is bounded by state highway 348 at Delta, 25 Mesa road and Forest Service road FS503 to Nucla, state highway 97 to state highway 141 to state highway 145 to Placerville, state highway 62 to Ridgeway, U.S. highway 550 to Montrose, and U.S. highway 50 to Delta. 130 �The study area seems typical of puma habitat in Colorado that has vegetation cover that varies from the pinion-juniper covered foothills starting from about 1,700 m elevation to the spruce-fir and aspen forests growing to the highest elevations of about 3,000 m. Mule deer (Odocoileus hemionus) and elk (Cervus elaphus) are the most abundant wild ungulates available for puma prey. There are cattle and domestic sheep raised on summer ranges on the study area. Year-round human residents live along the eastern and western fringe of the area, and there is a growing residential presence especially on the southern end of the plateau. A highly developed road system makes the study area well accessible for puma research efforts. A detailed description of the Uncompahgre Plateau is in Pojar and Bowden (2004). METHODS Reference and Treatment Periods This research was structured in two 5-year periods: a reference period (years 1―5) and a treatment period (years 6―10). The reference period was closed to puma hunting on the study area and was expected to cause a population increase phase. The treatment period (starting in November 2009) involves manipulation of the puma population with sport-hunting structured to achieve a management objective for a stable to increasing population. In both phases, puma population structure, and vital rates will be quantified, and management assumptions and hypotheses regarding population dynamics and effects of harvest will be tested. Contingent upon results of pilot studies, we will also assess enumeration methods for estimating puma population abundance. The reference period, without recreational puma hunting as a major limiting factor, was consistent with the natural history of the current puma species in North America which evolved life history traits during the past 10,000 to 12,000 years (Culver et al. 2000) that enable pumas to survive and reproduce (Logan and Sweanor 2001). In contrast, puma hunting, with its modern intensity and ingenuity, might have influenced puma selection pressures in western North America for the past 100 years. Hence, the reference period, years 1―5, would provide conditions where individual pumas in this population (of estimated sex and age structure) express life history traits interacting with the environment without recreational hunting as a limiting factor. Theoretically, the main limiting factor will be catchable prey abundance (Pierce et al. 2000, Logan and Sweanor 2001). This should allow researchers to understand basic system dynamics before manipulating the population with controlled recreational hunting. In the reference period, all pumas in the study area were protected, except for individual pumas that might be involved in depredation on livestock or human safety incidents. In addition, all radio-collared and eartagged pumas that ranged in a buffer zone, that includes the northern halves of GMUs 61 and 62, were protected from recreational hunting mortality. The reference period allowed researchers to quantify baseline demographic data on the puma population to estimate parameters useful for assessing the CDOW’s assumptions for its model-based approach to puma management. The reference period also facilitated other operational needs (because hunters did not kill the animals) including the marking of a large proportion of the puma population for parameter estimates and gathering movement data from GPS-collared pumas. During the treatment period, years 6―10, recreational puma hunting will occur on the same study area using management prescriptions structured from information learned during previous years. Using recreational hunting for the treatment is consistent with the CDOW’s objectives of manipulating natural tendencies of puma populations, particularly survival, to maintain either population stability or increase or suppression (CDOW, Draft L-DAU Plans, 2004). Theoretically, survival of independent pumas will be influenced mainly by recreational hunting, which will be quantified by agent-specific mortality rates of radio-collared pumas. For managers, demonstrating that they can manage puma populations with hunting and achieve the CDOW strategic objective of managing for a healthy, selfsustainable puma population state-wide is important to their mandated responsibility. Dynamics of the 131 �puma population will be manipulated to evaluate hypotheses that are related to effects of hunting (i.e.,: effects of harvest rates, relative vulnerability of puma sex and age classes to hunting, variations in puma population structure due to hunting). The killing of tagged and collared pumas during the treatment period will not hamper operational needs (as it would during the start-up years), because by the beginning of this period, a majority of independent pumas in the population should be marked, and sampling methods formalized. Pumas on the study area that may be involved in depredation of livestock or human safety incidences may be lethally controlled. Researchers that find that GPS-collared pumas have killed domestic livestock will record such incidents to facilitate reimbursement to the property owner for loss of the animal(s). In addition, researchers will notify the Area Manager of the CDOW if they perceive that an individual puma may be a threat to public safety. Field Methods Puma Capture: Realizing that pumas live at low densities and capturing pumas is difficult, as a starting point, our logistical aim was have a minimum of 6 puma in each of 6 categories (36 total) radiotagged in any year of the study if those or greater numbers are present. The 6 categories are: adult female, adult male, subadult female, subadult male, female cub, male cub. Our aim was to provide more quantitative and precise estimates of puma demographics than were achieved in earlier Colorado puma studies. This relatively large number of pumas might represent the majority of the puma population on the study area, and would provide the basic data for age- and sex-specific reproductive rates, survival rates, agent-specific mortality rates, emigration, and other movement data. Assuming that the puma population density on the study area was relatively low at the beginning of this study― about 1 adult/100 km2 and the sex ratio was equal (Anderson et al. 1992, Logan and Sweanor 2001:167), then there might have been 22 adults, 11 males and 11 females. Also assuming that the total population contained 10% subadults and 34% cubs (Logan and Sweanor 2001), then there might have been 4 subadults and 13 cubs with equal sex ratios in a total population of 39 pumas. If we achieved our logistical aim, then we should be able to quantify population characteristics and vital rates of the puma population based on a sample that includes a majority of individuals in the population. Recognizing that the population may grow, we will build upon the tagged number in each subsequent year to maintain a high proportion of marked individuals in the population. Puma capture and handling procedures were approved by the CDOW Animal Care and Use Committee (file #08-2004). All captured pumas were examined thoroughly to ascertain sex and describe physical condition and diagnostic markings. Ages of adult pumas were estimated initially by the gum-line recession method (Laundre et al. 2000) and dental characteristics of known-age puma (Logan and Sweanor, unpubl. data). Ages of subadult and cub pumas were estimated initially based on dental and physical characteristics of known-age pumas (Logan and Sweanor unpubl. data). Body measurements recorded for each puma included at a minimum: mass, pinna length, hind foot length, plantar pad dimensions. Tissue collections included: skin biopsy (from the pinna receiving the 6 mm biopsy punch for the ear-tags), and blood (30 ml from the saphenous or cephalic veins) for genotyping individuals, parentage and relatedness analyses, and disease screening; hair (from various body regions) and when available fecal DNA for genotyping tests of field gathered samples. Universal Transverse Mercator Grid Coordinates on each captured puma were fixed via Global Positioning System (GPS, North American Datum 27). Pumas were captured year-round using 4 methods: trained dogs, cage traps, foot-hold snares, and by hand (for small cubs). Capture efforts with dogs were conducted mainly during the winter when snow facilitates thorough searches for puma tracks and the ability of dogs to follow puma scent. The study area was searched systematically multiple times per year by four-wheel-drive trucks, all-terrain vehicles, 132 �snow-mobiles, and walking. When puma tracks ≤1 day old were detected, trained dogs were released to pursue pumas for capture. Pumas usually climbed trees to take refuge from the dogs. Adult and subadult pumas captured for the first time or requiring a change in telemetry collar were immobilized with Telazol (tiletamine hydrochloride/zolazepam hydrochloride) dosed at 5 mg/kg estimated body mass (Lisa Wolfe, DVM, CDOW, attending veterinarian, pers. comm.). Immobilizing agent was delivered into the caudal thigh muscles via a Pneu-Dart® shot from a CO2-powered pistol. Immediately, a 3m-by-3m square nylon net was deployed beneath the puma to catch it in case it fell from the tree. A researcher climbed the tree, fixed a Y-rope to two legs of the puma and lowered the cat to the ground with an attached climbing rope. Once the puma was on the ground, its head was covered, its legs tethered, and vital signs monitored (Logan et al. 1986). Normal signs include: pulse ~70 to 80 bpm, respiration ~20 bpm, capillary refill time ≤2 sec., rectal temperature ~101oF average, range = 95 to 104oF (Kreeger 1996). A cage trap was used to capture adults, subadults, and large cubs when pumas were lured into the trap using road-killed or puma-killed ungulates (Sweanor et al. 2008). A cage trap was set only if a target puma scavenged on the lure (i.e., an unmarked puma, or a puma requiring a collar change). Researchers continuously monitored the set cage trap from about 1 km distance by using VHF beacons on the cage and door. This allowed researchers to be at the cage to handle captured pumas within 30 minutes. Puma were immobilized with Telazol injected into the caudal thigh muscles with a pole syringe. Immobilized pumas were restrained and monitored as described previously. If non-target animals were caught in the cage trap, we opened the door and allowed the animal to leave the trap. Small cubs (≤10 weeks old) were captured using our hands (covered with clean leather gloves) or with a capture pole. Cubs were restrained inside new burlap bags during the handling process and were not administered immobilizing drugs. Cubs at nurseries were approached when mothers are away from nurseries (as determined by radio-telemetry). Cubs captured at nurseries were removed from the nursery a distance of ~100 m to minimize disturbance and human scent at nurseries. Immediately after handling processes were completed, cubs were returned to the exact nurseries where they were found (Logan and Sweanor 2001). Marking, Global Positioning System- and Radio-telemetry: Pumas do not possess easily identifiable natural marking, such as tigers (see Karanth and Nichols 1998, 2002), therefore, the capture, marking, and GPS- or VHF- collaring of individual pumas was essential to a number of project objectives, including estimating vital rates and gathering movement data relevant to population dynamics (i.e., emigration and Data Analysis Unit boundaries). Adult, subadult, and cub pumas were marked 3 ways: GPS/VHF- or VHF-collar, ear-tag, and tattoo. The identification number tattooed in the pinna was permanent and could not be lost unless the pinna is severed. A colored (bright yellow or orange), numbered rectangular (5 cm x 1.5 cm) ear-tag (Allflex USA, Inc., DFW Airport, TX) was inserted into each pinna to facilitate individual identification during direct recaptures. Cubs ≤10 weeks old were eartagged in only one pinna. Adult and subadult female pumas were fitted with GPS collars (approximately 400 g each, Lotek Wireless, Canada) if available. Initially, GPS-collars were programmed to fix and store puma locations at 4 times per day to sample daytime, nighttime, and crepuscular locations (i.e., 0:00, 06:00, 12:00, 19:00). GPS locations for pumas would provide precise, quantitative data on movements to assess the relevance of puma DAU boundaries, our search efforts, and to evaluate puma behavior and social structure. The GPS-collars also provided basic information on puma movements and locations to design other pilot studies in this program on vulnerability of puma to sport-harvest, habitat use, and enumeration methods (e.g., photographic or DNA mark-recapture). 133 �Subadult male pumas were fitted initially with conventional VHF collars (Lotek, LMRT-3, ~400 g each) with expansion joints fastened to the collars, which allows the collar to expand to the average adult male neck circumference (~46 cm). If subadult male pumas reached adulthood on the study area, we would recapture them and fit them with GPS collars. In addition, other adult and female subadult pumas were fitted with VHF collars when GPS collars are not available. VHF radio transmitters on GPS collars enabled researchers to find those pumas on the ground in real time to acquire remote GPS data reports, facilitate recaptures for re-collaring, and to check on their reproductive and survival status. VHF transmitters on GPS- and VHF-collars had a mortality mode set to alert researchers when puma was immobile for 3 to 24 hours so that dead pumas could be found to quantify survival rates and agent-specific mortality rates by gender and age. Locations of GPS- and VHFcollared pumas were fixed about once per week (as flight schedules and weather allow) from light fixedwing aircraft (e.g., Cessna 182) fitted with radio signal receiving equipment (Logan and Sweanor 2001). Aerial locations also provided simultaneous location data on mothers and cubs. GPS- and VHF-collared pumas were located from the ground opportunistically using hand-held yagi antenna. At least 3 bearings on peak aural signals were mapped to fix locations and estimate location error around locations (Logan and Sweanor 2001). Aerial and ground locations were plotted on 7.5 minute USGS maps (NAD 27) and UTMs along with location attributes recorded on standard forms. GPS and aerial locations were mapped using GIS software. We attempted to collar all cubs in observed litters with small VHF transmitter mounted on an expandable collar that can expand to adult neck size (Wildlife Materials, Murphysboro, Illinois, HLPM2160, ~50g, or Telonics, Inc., Mesa, Arizona MOD 210, ~100g,) when cubs weighed 2.3―11 kg (5―25 lb). Cubs with mass ≥11 kg could wear these small expandable collars until they are over 12 months old. Cubs were recaptured to replace collars as opportunities allowed. Monitoring radio-collared cubs allowed quantification of survival rates and agent-specific mortality rates (Logan and Sweanor 2001). Analytical Methods Population Characteristics: Population characteristics each year were tabulated with the number of individuals in each sex and age category. Age categories, as mentioned, include: adult (puma ≥24 months old, or younger breeders), subadults (young puma independent of mothers, <24 months old that do not breed), cubs (young dependent on mothers, also known as kittens) (Logan and Sweanor 2001). When data allowed, age categories were further partitioned into months (for cubs and subadults) or years (for adults). Reproductive Rates: Reproductive rates were estimated for GPS- and VHF-collared female pumas directly (Logan and Sweanor 2001). Genetic paternity analysis will be used to ascertain paternity for adult male pumas (Murphy et al. 1998). Survival and Agent-specific Mortality Rates: Radio-collared pumas provided known fate data used to estimate survival rates for each age stage using the Kaplan-Meier procedure to staggered entry (Pollock et al. 1989). A binomial survival model was also used for crude estimates of survival during the cub age stage (Williams et al. 2001:343-344). In addition, when data collection is complete, survival rates will be modeled in program MARK (White and Burnham 1999, Cooch and White 2004) where effects of individual (e.g., sex, age stage, reproductive stage) and temporal (i.e., reference period, treatment period) covariates to survival can be examined. Agent-specific mortality rates can also be analyzed using proportions and Trent and Rongstad procedures (Micromort software, Heisey and Fuller 1985). Population Inventory: The population of interest was independent pumas (i.e., adults and subadults) November to March which corresponds with Colorado’s puma hunting season. Independent pumas were those that could be legally killed by recreational hunters. Initially, we estimated the minimum 134 �number of independent pumas and puma density (i.e., number of independent puma/100 km2) each winter. The minimum number of independent pumas included all marked pumas known to be present on the study area during the period, plus individuals thought to be non-marked and detected by visual observation or tracks that were separated from locations of radio-collared pumas. Furthermore, adults comprised the breeding segment of the population and subadults were non-breeders that are potential recruits into the adult population in ≤1 year. The sampling unit was the individual independent puma (~≥1 yr. old). Puma Population Dynamics: A deterministic, discrete time model parameterized with population characteristics and vital rates from this research was used to assess puma population dynamics (Logan 2008). Functional Relationships: Once data collection is complete, a variety of analyses will be conducted to estimate parameters and examine functional relationships. Graphical methods will be used to initially examine functional relationships among puma population parameters. Linear regression procedures and coefficients of determination will be used to assess functional relationships if data for the response variable are normally distributed and the variance is the same at each level. If the relationship is not linear, data is non-normal, and variances are unequal, we will consider appropriate transformations of the data for regression procedures (Ott 1993). Non-parametric correlation methods, such as Spearman’s rank correlation coefficient, will also be used where appropriate to test for monotonic relationships between puma abundance and other parameters of interest (Conover 1999). Relationships of explanatory variables to survival parameters will be modeled in MARK. Statistical analyses can be performed using SYSTAT, R, and MARK software. RESULTS AND DISCUSSION Segment Objective 1 Field research to quantify puma population structure, vital rates, and causes of mortality for this report extended from August 2008 to July 2009. Our plan was to use 2 fully-staffed puma capture teams with dogs November through April, with each team operating on half the study area, with the intent of substantially boosting puma capture and sampling efforts. But, field operations were impacted by a state government mandated hiring freeze. We were limited to the principal investigator and 2 houndmen teams from October 2008 through April 2009. The principal investigator operated with the 2 houndmen teams for a single expanded moving search footprint and performed all immobilization and sampling procedures during winter and spring capture efforts. Our searches to detect puma presence covered the entire study area. By May 2009 technicians could be hired again and assisted in puma captures in cage traps and at nurseries. In addition, the Colorado State University bobcat research team facilitated the recapture of an adult female puma. We made 37 puma captures during the period (7 adult females [1 adult female captured 3 times, another captured twice], 4 adult males [1 adult male captured 3 times], 1 subadult female, and 18 cubs [2 of them captured twice each]). Five adults (4 females, 1 male) and 14 cubs were captured and marked for the first time in 2008-2009. One adult female and 2 cubs were visually observed at capture efforts, but could not be handled. A total of 39 pumas were monitored with radiotelemetry from August 2008 to July 2009 (some of these had been collared during previous years). Trained dogs were used as our main method to capture, sample, and mark adult and subadult pumas from December 9, 2008 to April 30, 2009. Those efforts resulted in 71 search days, 198-202 puma tracks detected, 75-78 pursuits, and 24 puma captures (Table 1). Puma capture efforts (i.e., search days) with dogs in this period was slightly less than our efforts in the 4 previous winters (Table 2). But, the frequency of tracks encountered was about the same as the previous winter. The pursuits increased over the 4 previous periods, as did our capture rate. The later 2 statistics were probably the result of using 2 houndmen teams. Four adult and 7 cubs were captured for the first time by using dogs (Tables 1 and 3). 135 �This included 2 non-marked cubs that could not be handled for safety reasons. Three adult male pumas and 1 large male cub were captured with dogs but could not be handled for safety reasons, and 1 adult female and her cub were visually observed but could not be caught for marking and sampling (Table 4). Two adult females (1 recaptured twice) and an adult male were recaptured and observed, but there was no need to handle them (Table 5). Puma capture efforts using ungulate carcasses and cage traps extended from August 20, 2008 to July 20, 2009. We used 36 road-killed mule deer at 17 different sites to capture one adult female and one subadult female (Tables 6). In addition, the Colorado State University bobcat research team recaptured an adult female in a trap set for bobcats, thus, providing an opportunity to change a failing GPS collar. Pumas scavenged 7 of 36 (19.4%) of the ungulate carcasses used for bait. Percentages of puma scavenging ungulate carcasses in the previous 3 years were 20%, 22.5%, and 18.3%. Other carnivores that used the ungulate baits included: black bear, bobcat, gray fox, and domestic dogs. We captured 14 cubs (8 male:6 female) for the first time (Table 7), and fit 11 of them with radiocollars (Appendix A). Three cubs were not radio-collared. In 1 case the mother returned to the nursery while we were sampling the cubs so we quickly returned the cubs to the nursery, leaving 1 collared and 1 not collared. In the other case, 2 cubs in a litter of 3 were too small to wear the collar design. Three of the cubs were bayed by our dogs and were large enough to require anesthetics for safe handling. The other 11 cubs were handled without anesthetics at their nurseries when they were 34 to 38 days old. Litters bearing these cubs were produced in August (2), September (1), April (1), and May (3). In addition to our direct puma captures with dogs December through April, we detected 10 independent pumas that we were able to identify with GPS or VHF telemetry 12 times, thus, negating the need to capture those pumas directly with dogs (Table 1). Upon detecting puma tracks that were aged at 1 day old, we followed the tracks with a radio receiver in an effort to detect if the tracks might be of a puma wearing a functional collar. We assigned tracks to a collared individual if we received radio signals from a puma that we judged to be <1 km from the tracks and in direction of travel of the tracks. GPS data from pumas with functional GPS collars provided confirmatory information about movements of pumas. If GPS data indicated that the puma moved through the area at the time the tracks were made, then we ruled the data were confirmatory. This approach allowed us to more efficiently allocate our capture efforts toward pumas of unknown identity on the study area, particularly unmarked pumas or pumas with nonfunctioning GPS- or radiocollars. Our search efforts throughout the study area also revealed the presence of at least 14 other independent pumas, we classified as 12 females and 2 males. We could separate the activity of these pumas from the GPS- and VHF- collared pumas in time, space, and track size differences between females and males. Moreover, females in association with cubs of different numbers, sizes, and locations enabled us to separate 5 adult females followed by 1 to 3 medium-to-large-size cubs. One of the adult females was visually observed with 2 of her 3 cubs, 2 of which we captured and marked. The tracks we found of the other pumas were too old to pursue (i.e., probability of capture with the dogs was negligible). It is also possible that 2 of the adult females were previously marked animals wearing non-functional GPS collars (Table 8). Our search and capture efforts during November 2008 through April 2009 enabled us to estimate a minimum count of 37 independent pumas detected on the Uncompahgre Plateau study area, up from a minimum count of 33 independent pumas during the November 2007 to March 2008 (Table 8). This estimate was based on the number of known radio-collared pumas, the observation of one non-collared female puma, and detection of tracks of suspected non-collared pumas or pumas with non-functional GPS collars on the study area (explained previously). In addition to the independent pumas, we also counted a minimum of 21 cubs. Of the 37 independent pumas, 23 to 25 (62-68%) were marked and 12 to 14 (32- 136 �38%) were assumed to be unmarked animals. Of the expected unmarked pumas, 12 were females and 2 were males, which might reflect lower detection rates of females, making it more difficult for us to capture and mark females. Although, we would have expected to capture, sample, and mark a larger portion of those animals had we fielded the 2 complete capture teams in winter 2008 to 2009 as previously planned. There may be variation in puma numbers on the west and east slopes of the study area. The west slope count includes 16 independent pumas (11 females, 5 males). The east slope count includes 21 independent pumas (15 females, 6 males). We used the minimum puma counts in the past 2 periods, (i.e., 33 independent pumas for November 2007 to March 2008 and 37 independent pumas for November 2008 to April 2009) to calculate preliminary minimum densities for the winter puma habitat area estimated at 1,671 km2 on the Uncompahgre Plateau study area. The minimum densities ranged from 2.0 to 2.2 independent pumas/100 km2. Anderson et al. (1992) studied pumas on the east slope of the Uncompahgre Plateau (i.e., GMU 62) during 1981 to 1988. Sport-hunting was banned during that study to investigate an “unexploited” puma population (Anderson et al. 1992:5). As our current effort results in larger samples and progresses in time through the reference and treatment periods, similarities and differences in results of the 2 research efforts, now separated by more than 15 years, should illuminate reliable knowledge for puma management in Colorado. Our current puma research on the Uncompahgre Plateau has been underway for 4.7 years (compared to 7 years of Anderson et al. 1992). Our data analysis at this stage of the research is not by any means exhaustive or complete because we are still in the intensive data-gathering phase, yet, our data allows some preliminary comparisons with Anderson’s (1992) completed work. In the Anderson et al. (1992) study, the average capture effort with dogs was 91.1 days per winter (range = 32 to 136, n = 7) resulting in an average capture effort of 13.9 days per puma. Of 189 pursuits of pumas, 110 (58%) were successful (either of radio-collared or non-collared animals). Anderson et al. (1992) focused on capturing pumas >27 kg in body mass while avoiding pumas <27 kg in mass. They captured 47 pumas with dogs for an average capture rate of 13.9 days per puma. Eight other pumas, all female cubs ≤7 months old, were caught in steel leg-hold traps by trappers in pursuit of furbearers, and were added to the study animal population. Two other cubs were killed by the dogs. In total, Anderson et al. (1992) captured 57 pumas, of which 49 were radio-collared. Anderson et al. (1992:49) estimated a minimum density of “resident” pumas (equivalent to our independent pumas) at 1.1 pumas/100 km2. This was practically half the density of our current preliminary minimum density estimates for independent pumas (see previously). So far, in our 5 winters, the average effort per winter to capture pumas with dogs is 77.2 days (range = 71 to 82). Of 247 pursuits, 94 (38%) were successful. We captured 41 individual pumas their first time with dogs (i.e., does not include dog-aided recaptures), yielding an average capture rate of 9.4 days per capture (i.e., 386 days/41 captures). Other capture efforts and results between the 2 studies are not comparable, because Anderson et al. (1992) did not routinely attempt to capture pumas using cage traps or capture cubs at nurseries like we are. In our current effort, we captured, sampled, and marked 109 pumas. Of those animals, 91 were radiocollared, allowing us to monitor fates of pumas in all sexes and age stages, including: 19 adult females, 12 adult males, 4 subadult females, 5 subadult males, 30 female cubs, 30 male cubs (some individuals occur in more than one age-stage). To date, this represents the largest number of individual pumas sampled for population data in Colorado. Mass recorded by Anderson et al. (1992:86) for pumas having an estimated age ≥24 months, averaged 61.6 kg for 8 males, (SD = 5.7, range = 51.8 to 70.8) and 44.5 kg for 14 females (SD = 3.6, range = 38.5 to 49.9). So far in our current study, mass for pumas ≥24 months old and weighed for the first time averaged 61.3 kg for 10 males (SD = 3.72, range 55 to 68 kg) and 38.3 kg for 18 females (SD = 137 �4.01, range = 31 to 45). Sexual dimorphism is evident in pumas, and has been described for the species throughout its range (Young and Goldman 1946). Sexual dimorphism in the puma has been explained as a potential result of sexual selection (Logan and Sweanor 2001:109). Segment Objective 2 During the past 4.7 years of this work we compiled data on puma reproduction that was not previously available on pumas in Colorado. We examined 72 cubs from 26 litters aged 26 to 42 days old where we were reasonably sure that we counted all the cubs at the nurseries (Table 9, Appendix A). Using those litters and 1 other litter confirmed by nursling cub tracks with a GPS-collared female (i.e., n = 27 litters with approximately known birth dates), the distribution of puma births by month indicate puma births extending from March into September, with 24 of 27 births occurring May through September (Fig. 4). Our data suggests that the majority of puma breeding activity occurs February through June. The secondary male:female sex ratio was 41:31 for 26 litters where all the cubs were sexed. This ratio was not significantly different from 1:1, (X2 = 1.389 < 3.841, α = 0.05, 1 d.f.). An equal sex ratio at birth is characteristic of other puma populations in North America (Robinette et al. 1961, Logan and Sweanor 2001:69-70). The mean (±SD) and extreme sizes of the 26 litters examined at nurseries were 2.77 (±0.9081), 1 to 4 (Table 9). In addition, 16 birth intervals for 9 different female pumas averaged 18.462 months (SD = 4.6035), and ranged from 11.7 to 23.9 months (Table 9). During the past 4 biological years (i.e., 2005-06 to 2008-09) when we radio-monitored 12, 13, 12, and 11 adult female pumas per year, respectively, the proportion of adult females that produced cubs each year were 0.67, 0.69, 0.58, 0.45 with a mean ± SD of 0.598 ± 0.1094. Based on observations (from GPS and radio-telemetry data) of associations between 9 mothers and putative sires (Table 9), 10 estimated gestation periods, considering a range of days for 7 observations, averaged 90.5 to 92.3 days (SD = 2.5495, 2.1628, respectively), which is consistent with average puma gestation reported in the technical literature on puma (i.e., mean ± SD = 91.9 ± 4.1, Anderson 1983:33, mean = 91.5 ± 4.0 Logan and Sweanor 2001:414). Anderson et al. (1992:47) reported of “17 postnatal litters about 10-240 days in estimated age from 12 individual females, the mean (±SD) and extremes of litter sizes were 2.41 ± 0.8, 1-4”. “Because most postnatal young were not handled, their sex ratio is unknown” (Anderson et al.1992:48). In addition, because cubs were first observed at older ages, it is likely that some post-natal mortality had occurred. This is one explanation for smaller litters observed by Anderson et al. (1992). Anderson et al. (1992:47-48) found that of 10 puma birth dates 7 were during July, August, and September, 2 in October, and 1 in December, with most breeding occurring April through June. Data on our 27 litters adds to Anderson’s data (Fig. 4), and indicates puma births in Colorado occurring in every month except January and November (so far). Anderson’s observation of two 12-month birth intervals for one female (Anderson et al. 1992:48) is at the low range of our observations (Table 9). Segment Objectives 3 & 4 From December 8, 2004 (capture and collaring of the first adult puma M1) to July 31, 2009, we radio-monitored 12 adult male and 19 adult female pumas to quantify survival and agent-specific mortality rates (Table 10). One adult male is known to have died of natural causes. M4 was about 37 to 45 months old when he was killed by an unidentified male puma along the southeast boundary of the study area. One adult male, M5, lived in the buffer zone north of the study area where all marked pumas were protected from sport-hunting. However, M5 was killed at 54 months old by a puma hunter when M5 left the buffer zone and ranged into eastern Utah. We lost contact with 3 adult males apparently due to GPS/VHF collar failure: M1, M27, and M29. Direct observations in the field indicated that M27 was alive on 05-07-09 (camera photo), and M29 was alive on 02-25-09 (recapture). Four adult females are known to have died of natural causes. F50 was about 29 to 31 months old when she died apparently of natural causes (exact agent could not be identified). Three adult females, F54, F30, and F2 were killed by other pumas. F54 was killed at about 49 months old by a male puma on the southern boundary of the 138 �study area while apparently in direct competition for a fawn mule deer. F30 was killed by a puma of unknown sex and for unknown circumstances when she was about 60 months old. F2 was killed when she was about 92 months old by a puma of unknown sex (but thought to be a male based on presence of 8 scrapes), as was at least one of her four 87-day-old cubs M79 (Appendix A). All 3 adult females appeared to have fatal bites to the head, with canine punctures that penetrated the skull. One adult female, F7, was killed for depredation control purposes when she was about 107 months old. Preliminary estimates of adult puma survival rates indicate relatively high survival in this reference period (i.e., with no sport-hunting) (Table 11). Survival rates were estimated using the KaplanMeier procedure to staggered entry of animals (Pollock et al. 1989) for the past 4 annual and hunting season periods when samples were ≥ 5 animals in each sex category. The survival rates reflect 1 male death and 4 adult female deaths from natural causes. Data on M5 (killed by a hunter) and F7 (killed for depredation control) were right censored after the date they died. In general, adult male puma survival is higher than adult female survival in this non-hunted population state. The adult age structure, as indicated in Figure 4, is indicative of high survival rates during the past 5 winters without sport-hunting mortality. Research in New Mexico on a non-hunted puma population also indicated high adult survival rates with survival rates of adult males higher than adult females and the major cause of death being aggression by male pumas (n = 8 years; Logan and Sweanor 2001:127-138). We have radio-monitored 9 pumas, 5 males and 4 females, in the subadult age-stage (independent pumas <24 months old) (Table 12). One of those, F66, died of natural causes. F66 died at 23 months old of trauma to internal organs that caused massive bleeding attributed to trampling by an elk or mule deer. We need to increase our efforts to acquire larger samples of male and female radio-monitored subadult pumas to acquire reliable estimates of their survival. Data from puma hunters provided additional information on fates of 8 pumas, 7 males and 1 female, initially captured and marked as cubs (5 males) or subadults (2 males, 1 female) on the Uncompahgre Plateau puma study area (Table 13). All 7 of the males were killed away from the study area by hunters at linear distances (i.e., from initial capture sites to kill sites) ranging from about 66 to 370 km. Two males with extreme moves were killed in the Snowy Range of southeastern Wyoming (369.6 km) and the Cimarron Range of north-central New Mexico (329.8 km). The female (F52) was treed and released by hunters. These pumas represent dispersal moves from the Uncompahgre Plateau. All of the pumas, except for 1 (M68, 17 months old) had reached adult ages ranging from 24 to 54 months old. Our current research effort is still too short in duration and samples too small to make meaningful comparisons with evidence in the literature regarding puma offspring dispersal rates, distances moved, and philopatry. Dispersal and philopatry have been explained as life history strategies in pumas that assist gene flow, colonization, population maintenance, and individual survival and reproductive success (Logan and Sweanor 2001). Thus, such strategies would be expected to be conserved, and expressed in puma populations in different locations. In addition, because puma emigration and immigration (i.e., via dispersal) have been shown to be important processes in puma population dynamics (Sweanor et al. 2000), we need larger samples and longer research duration in this study to understand the significance of those parameters in our study population. A preliminary estimate of puma cub survival was made with 36 radio-collared cubs (16 males, 20 females) that we marked at nurseries when they were 26 to 42 days old. Only cubs that died of natural causes were used (i.e., 3 capture-related deaths were excluded). All cubs were born from May 2005 to July 2007. For the Kaplan-Meier procedure to staggered entry of animals (Pollock et al. 1989), the maximum survival period was assumed to be 365 days after capture (i.e., ~13-14 months old) to coincide with the age that puma cubs would normally be expected to become independent from their mothers (Logan and Sweanor 2001). In this preliminary estimate, observations of siblings are assumed to be 139 �independent (i.e., distribution of mortalities among litters is random), but that assumption might not be reliable (Bishop et al. 2008; an overdispersion parameter will need to be estimated). We omitted 3 radiocollared cubs that died as a result of the expandable radiocollars (Appendix A). Otherwise, cubs were right censored when they reached independence, or from the date after we lost contact. Dates that bracketed the deaths of cubs were used to estimate minimum and maximum survival rates. The estimated minimum survival rate using the Kaplan-Meier procedure was 0.5285 (SE = 0.1623). The maximum estimated cub survival was practically the same, 0.5328 (SE = 0.1629). Cub survival estimated with a binomial model (Williams et al. 2001) for the same sample was 0.5833 ± 0.1610 (95% C.I.). In order to improve the reliability of puma cub survival data, we will make an effort to increase the number of radiocollared cubs that are monitored. The major natural cause of death in cubs, where cause could be determined, was infanticide and cannibalism by other, especially male, pumas. We attributed 8 cub mortalities to infanticide, and it is probable that 5 other cubs died directly from infanticide or because their mother was killed when her 4 cubs were at an age (87 days) when they could not survive without her (Appendix A). Male-caused infanticide, along with aggression-caused mortality in adult (indicated previously) and subadult pumas (Logan and Sweanor 2001) has also been a dominant mortality factor in other puma populations in North America (Logan and Sweanor 2001:115-136). Such male puma behavior has been theorized for being a strong selective force in shaping the evolution of behavioral tactics, social structure, and life history strategies in pumas (Logan and Sweanor 2001). The closure on sport-hunting on the study area and protection of marked pumas from sportharvest on the buffer area on the northern portion of the Uncompahgre Plateau for the reference period operated as designed to remove sport-hunting as a cause of death in the study population. Of the adult and subadult pumas wearing a functional GPS/VHF-collar, only 1 adult puma died due to human causes on the study or buffer areas (F7 killed for depredation control, mentioned previously). This reference condition enabled us to quantify puma population structure, survival rates, and agent-specific mortality rates of pumas in the absence of direct human-caused mortality by sport-hunting, and will allow comparisons with the treatment period when puma hunting manipulates the puma population on the study area. Furthermore, we recorded deaths of 7 non-marked pumas that died since 2004, mainly from human causes (Table 14). Six non-marked pumas (2 males, 4 females) were struck by vehicles on highways or a county road along boundaries of the study area. In addition, 2 marked female cubs (mentioned previously) were killed in vehicle collisions on a highway. Both of those cubs were offspring of F16 which has a home range straddling highway 550 south of Montrose. Of the 8 pumas killed by vehicles, 5 were dependent cubs, 2 were probably subadults, and 1 was an adult female. A bizarre natural mortality case we documented was of a non-marked adult female found in Roubideau Canyon that was lodged in a narrow fork of an aspen tree and probably died of asphyxia due to compression of the thorax. Anderson et al. (1992:50) reported on the fates of 21 radio-collared pumas (11 pumas <24 months old, and 10 ≥ 24 months old) from a total of 49 in his previous study which was intended to “assess the effects of sport-hunting on an unexploited population” (Anderson et al. 1992:5). They found 19 of the 21 deaths (i.e., 90%) were due to human causes, attributed to: legal kill outside the study area (7), research capture-related (6), predator management (3), illegal kill (2), and suspected predacide (1). Other causes of mortality included, intraspecies strife (1) and disease (1). Actual age-stage and annual survival rates and agent-specific mortality rates from our current effort cannot be clearly compared with the Anderson et al. (1992:53) effort because they pooled data for male and female pumas in seemingly arbitrary age stages that overlapped puma life history stages (i.e., cubs, subadults, adults). The Anderson et al. (1992:53) estimated survival rates with the Kaplan-Meier procedure (Pollock et al. 1989) for 20 male and 22 female pumas were: 12-24 month old = 0.642; 24-36 months old = 0.692, 36 to 48 months old = 0.917, and 48- 140 �60 months old = 0.800. Actual sample sizes within each age category were not given. There were no quantitative data allowing estimation of survival and agent-specific mortality for cubs less than 12 months old. Anderson et al. (1992) found that all 9 radio-collared male pumas dispersed from their natal areas, and 2 of 6 radio-collared females did not disperse from their natal areas (A. E. Anderson, Sep. 1993, errata for Anderson et al. 1992:61). Mean ± SD and range of dispersal distances (km) for 8 males, aged 10 to 13 months old at dispersal, were 86.2 ± 51.3, 23 to 151. For 4 females, aged 11 to 31 months old at dispersal, mean ± SD and range of dispersal distances (km) were 37.0 ± 15.3, 17 to 54 (Anderson et al. 1992:63). Segment Objective 5 To investigate the potential that puma hunters might detect puma mothers away from their cubs, we continued gathering data on spatial associations of puma mothers and their cubs during the puma hunting season, which extends from November through March each winter in Colorado. Female pumas are fair game in Colorado, unless they are accompanied by 1 or more cubs. Mothers that are caught away from their cubs could be legally harvested. Such incidents would result in cubs being orphaned. Orphaned cubs that are ≤6 months old could have a survival rate (to the subadult stage) of < 0.05. Orphaned cubs 7 to 12 months old might have a survival rate (to the subadult stage) of about 0.7 (K. Logan, unpublished data). We monitored 7 puma families with a radio-collared mother and at least 1 radio-collared cub from November 6, 2008 to March 20, 2009 during 11 airplane flights (Table 15). To assess whether mothers were apart or in close association with cubs, we considered error in aerial locations. We recovered 28 puma radiocollars (i.e., of dead pumas or shed collars from cubs) that we located from the airplane and then fixed the actual locations of collars on the ground with hand-held GPS receivers. Range of location error was 5 to 660 m (mean = 260, SD = 179.73). We used distances greater than the extreme high range of location error (660 m) as the metric to decide if puma mothers might be detected away from their cubs by hunters. In aggregate, the data for the past 4 winters include 171 observations for 1−7 families per winter (Table 15), and generally indicate that puma mothers are more likely (87% of observations) to be within 660 m of their cubs during the day in winter. An effort will be made to increase the number of radio-collared family members in subsequent winters. If the total sample size allows, we want to examine variation in mother-cub association distances on an individual female basis. Moreover, we will gather direct information on the frequency that cubs are orphaned and their survival during the treatment period when the pumas are hunted. Anderson et al. (1992:70-71) recorded 69 instances of simultaneous aerial locations of 7 pairs of puma mothers and dependent young. They reported that mothers and young were together in 21 (30.4%) of those instances, and they were 1 to 2.2 km apart in 48 (69.6%) of those instances. Segment Objective 6 We used the data gathered so far in the reference period for a preliminary evaluation of 5 assumptions used by CDOW biologists and managers to manage puma populations with sport-hunting. Assumption 1: The CDOW assumes density ranges of 2.0 to 4.6 puma/100 km2 (i.e., includes pumas of all age stages- adults, subadults, and cubs, J. Apker, CDOW Carnivore Biologist, person. commun. Nov. 19, 2003) to extrapolate to DAUs to guide the model-based quota-setting process. Assuming that on average 66% of the population is comprised of adults and subadults (previously), then the range of density for independent pumas would be 1.3 to 3.0/100 km2. The population sex and age structure is also assumed to be similar to puma populations described in the intensive studies in the literature on puma populations (CDOW 2007). 141 �H1: Puma densities during the reference period and the treatment period will vary within the range of 2.0 to 4.6 puma/100 km2 and will exhibit a similar sex and age structure to puma populations studied intensively in Wyoming, Idaho, Alberta, and New Mexico (CDOW 2007). We have partially addressed H1 with a preliminary minimum estimated density of 2.0 to 2.2 independent pumas/100 km2 of estimated winter habitat on the Uncompahgre Plateau study area in RY4 (i.e., 33 minimum independent pumas/1,671 km2) and RY5 (i.e., 37 minimum independent pumas/1,671 km2). These minimum density estimates represent the mid-to-high range of density for independent resident pumas in some North American populations (i.e., range 0.3-2.2/100 km2, Logan and Sweanor 2001:167), but lower than higher estimates for independent pumas in more recent studies in Wyoming (3.4/100 km2, Anderson and Lindzey 2005) and Utah (3.2/100 km2, Choate et al. 2006). Moreover, the sex and age structure of the minimum population observed in winter of reference year 4 (i.e., RY4) is similar to descriptions of other puma populations in western states (Logan and Sweanor 2001:167). Assumption 2: The adult plus subadult (i.e., harvest-age pumas) segment of the population exhibit a moderate annual rate of growth of 15% (i.e., λ = 1.15, CDOW 2007). H2: Population parameters estimated during a 5-year reference period (in absence of recreational puma hunting) will yield an estimated annual adult plus subadult population growth rate that will match or exceed λ = 1.15. Puma population modeling using population characteristics and vital rates from this current research effort supports this assumption (Appendix B). Expected lambda (i.e., finite rate of population change of independent pumas) ranges from 1.17 to 1.22 and an average of 1.20 ± 0.0182 SD (n = 5; TY1-TY5) for the no harvest model (Appendix B, Table B.7). Expected lambda for the modeled non-hunted puma population on the Uncompahgre Plateau are consistent with the high range of observed average annual rates of population increase for a non-hunted puma population in good quality habitat in southern New Mexico (i.e., r = 0.21, n = 4 yr.; r = 0.28, n = 4 yr.; r = 0.17, n = 4 yr.; r = 0.11, n = 7 yr.; Logan and Sweanor 2001:169-175). Puma population growth might be higher on the Uncompahgre Plateau because of higher quality habitat (i.e., greater vulnerable prey biomass), and if puma sources are nearby to the study area which provide immigrants. Assumption 3: Puma harvest rate formulations for DAUs assume that total mortality (i.e., harvest plus other natural deaths) in the range of 8 to 15% of the harvest-age population (i.e., independent pumas comprised of adults plus subadults) with the total mortality comprised of 35 to 45% females (i.e., adults and subadults) is acceptable to manage for a stable-to-increasing puma population (CDOW 2007). Harvest is assumed to be additive to natural mortality. H3a: The puma population is not expected to decline, therefore, we should be observing puma population parameters characteristic of a stable or increasing hunted puma population. Preliminary modeling results with 15% and 16% mortality in the harvest-age population indicates expected stable or increase population phases, with additive harvest mortality (Appendix B, Tables B.3, B.4, B.5, B.8, B.9, Fig. B.2). H3b: Harvest mortality of 15% of the adults and subadults will be strongly additive to other natural causes of mortality. Preliminary survival rates for annual and shorter-term hunting season periods for adult-age pumas in the reference period indicate high survival (Table 11). Similarly, a course survival rate for 9 subadult radiocollared pumas in the reference period is also high (finite rate of survival during the subadult stage: 8/9 = 0.89). These rates partially support the assumption that additive mortality caused by hunting can be expected. A direct test of this assumption will develop in the treatment period. Assumption 4: To reduce a puma population, hunting must remove more than the annual increment of population growth. For DAUs with the objective to suppress the puma population, the total mortality 142 �guide of greater than 15 to 28% of the harvest-age population with greater than 45% comprised of females is suggested (CDOW 2007). H4: Total mortality of an estimated 16% or greater of the harvestable population with greater than 45% females will cause a decline in the abundance of harvest-age pumas (i.e., adults and subadults). Preliminary modeling results with 16% mortality or greater in the harvest-age population and with greater than 45% of the harvest comprised of females indicates expected puma population declines (Appendix B, Tables B.6, B.10, B.12−B.16, Figs. B.2−B.4). Assumption 5: The increase and decline phases of the puma population make it possible to test hypotheses related to shifts in the age structure of the population which have been linked to harvest intensity in Wyoming and Utah. H5: The puma population on the Uncompahgre Plateau study area will exhibit a young age structure after hunting prohibition at the beginning of the reference period. During the 5 years of hunting prohibition, greater survival of independent pumas will cause an older age structure in harvest-age pumas (i.e., adults and subadults) as suggested by the work of Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. Preliminary results as indicated by the age structure of independent pumas captured for the first time in 2004-05 (Logan 2005), at first capture (Fig. 3), and the age structure of the independent puma population in March 2009 (Fig. 5), and apparently high adult and subadult survival rates during the reference period support the hypothesis for a young age structure early in the reference period with an aging structure later at the end of the period. Segment Objective 7 Principal investigator K. Logan with CDOW biologists and managers developed by consensus a preliminary structure (i.e., official approval pending Wildlife Commission decision in September 2009) to manipulate the puma population with sport-hunting on the study area during the treatment period. The hunting season will begin in mid-November and extend to January 31, unless the last puma on the design quota is killed before January 31, which will effectively close the season on the study area. The harvest quota will be 8 pumas (i.e., 15% harvest of the estimated minimum number of independent pumas), with the objective to manage for a stable to increasing population. The quota of 8 is based on the projected minimum number of independent pumas expected on the study area in winter 2009-10, modeled from a minimum count during winter 2007-08 (see Appendix B, Table B.7). No assumptions about additional pumas on the study area are made or contrived. The quota of 8 is expected to allow the population to achieve a stable or increase phase even if the quota is exceeded due to potential ideal snow-tracking conditions that could result in multiple pumas being killed within a mandatory 48-hour reporting period. Such an overharvest might be expected to reach 20 to 30% over the design harvest (in this case ~2 pumas killed over the harvest; J. Apker, Carnivore Biologist, CDOW, person. comm. June 8, 2009). The number of hunters on the study area at any particular time each hunting season will not be limited. However, each hunter on the study area will be required to obtain a hunting permit from the CDOW Montrose Service Center. Permits will be free and unlimited. Each permit will allow the individual hunter with a legal puma hunting license in Colorado to hunt in the puma study area for up to 14 days from the issue date. Unsuccessful hunters that wish to continue hunting past the permit expiration date can request a new permit for another 14 days or until the hunting season on the study area closes due to the quota being reached or the end of the hunting season. (The number of pumas killed on the study area each winter will be regulated by the design quota, discussed previously.) This permit system is expected to allow the CDOW to monitor the number of hunters on the study area and to contact each hunter for survey information. All pumas harvested on the study area will be subject to the examination check and seal mandated by the State of Colorado. Hunters must report their puma kill to CDOW within 48 hours of harvest and 143 �present the puma carcass for inspection within 5 days of harvest. At the time of carcass check-in a biologist with the puma research team will inspect the puma to assist in recording information on the CDOW puma harvest data form and to collect an upper premolar tooth for aging (i.e., mandatory) and a tissue sample using a 6 mm biopsy punch (i.e., voluntary) for DNA genotyping. Each successful hunter will also be asked at that time to complete a one-page hunter survey form. All other hunters that do not report a puma kill on the study area will be contacted and asked to complete the survey. Hunter harvest will provide direct evidence of removal rates of marked puma for survival and agent-specific mortality data, and to help evaluate the relative vulnerability of pumas to harvest and potential for hunter selectivity. Hunter harvest will also reveal availability and sex and age classes of unmarked pumas on the study area. After the design quota is filled or January 31 (whichever comes first), puma research teams will immediately activate for capture operations with trained dogs. Two fully-staffed capture teams, one detailed on the east slope and one detailed on the west slope, will systematically and thoroughly search the study area to capture, sample, and GPS/VHF radiocollar pumas the remainder of winter and early spring when snow-tracking conditions can facilitate those efforts. These efforts are necessary to maintain samples to quantify population sex and age structure and estimate minimum population size and other population parameters. Segment Objective 8 Principal investigator K. Logan developed another draft study plan pertaining to the next 5 years of puma research on the Uncompahgre Plateau. The draft plan was subjected to an internal review by researchers and was circulated for review to Carnivore Biologist J. Apker, Area 18 Biologist B. Banulis, Southwest Regional Biologist S. Wait, and Area 18 Wildlife Manager R. Del Piccolo. Comments were incorporated into a substantially modified study plan which was reviewed by Mammals Researcher Dr. Chad Bishop (now the Mammals Research Leader). That study plan will be modified to address new considerations and will be submitted to Mammals Research Leader Chad Bishop in fall 2009. Segment Objective 9 Data from 26 (8 male, 18 female) GPS-collared pumas, totaling over 39 thousand GPS locations (Table 16) will be used to examine the social structure of the puma population on the Uncompahgre Plateau and to examine movements of pumas relative to Game and Data Analysis Unit boundaries. Those data will also be used in a set of collaborative projects, including: examination of puma behavior in relation to human development with Mammals Researcher Dr. Mat Alldredge, who is studying pumahuman interactions on the Colorado Front Range; modeling and mapping puma habitat in Colorado and other western states with Dr. Kevin Crooks and Dr. Chris Burdett (Department of Fish, Wildlife and Conservation Biology, Colorado State University- DFWCB, CSU); evaluation of puma detection rates using camera grids with Dr. Kevin Crooks and Ph.D. candidate Jesse Lewis (DFWCB, CSU). Furthermore, puma population and genetic data from the Uncompahgre Plateau can be used in collaboration with Dr. Alldredge’s puma research efforts on the Front Range to examine similarities or differences in puma population dynamics and social structure between the 2 environments. We are currently collaborating with Dr. Sue VandeWoude and Dr. Kevin Crooks, and postdoctoral and graduate students at the College of Veterinary Medicine and Biomedical Sciences, Department of Microbiology, Pathology, and Immunology, Colorado State University in a pilot study titled: Puma concolor immune health― Relationship to management paradigms and disease. Tissue samples (i.e., blood, saliva, feces) from pumas we capture are collected and shipped to the DMIP for analyses. That project has been expanded to The effects of urban fragmentation and landscape connectivity on disease prevalence and transmission in North American felids. A description of that project and incomplete results on infectious disease surveillance on 35 individual pumas (22 independent 144 �females, 12 independent males, and 1 male cub) sampled on the Uncompahgre Plateau are presented in Appendix C. Those data contributed to a publication in Emerging Infectious Diseases (accepted), titled Plague and wild felids: zoonotic disease in the western US , a paper on seroprevalence in populations of pumas and bobcats in the western United States by collaborators: Sarah N. Bevins1, Jeff A. Tracey1, Sam P. Franklin1, Virginia L. Schmit1, Martha L. MacMillan1, Kenneth L. Gage2, Martin E. Schriefer2, Kenneth A. Logan3, Linda L. Sweanor1, Mat W. Alldredge3, Karoline Krumm1, Walter M. Boyce4, Winston Vickers4, Seth P.D. Riley5, Lisa M. Lyren6, Erin E. Boydston6, Melody E. Roelke7, Robert Fischer6, Kevin R. Crooks1, and Sue VandeWoude1 (1Colorado State University, USA; 2DVBID Centers for Disease Control, USA; 3Colorado Division of Wildlife, USA; 4University of California, Davis, USA; 5 National Park Service, USA; 6United States Geological Survey, USA; 7 National Cancer Institute, USA). SUMMARY Manipulative, long-term research on puma population dynamics, effects of sport-hunting, and development and testing of puma enumeration methods began in December 2004. After 4.7 years of effort in a reference period, 109 pumas have been captured, sampled, marked, and released. Of those animals, 91 were radio-collared, allowing us to monitor fates of pumas in all sexes and age stages, including: 19 adult females, 12 adult males, 4 subadult females, 5 subadult males, 30 female cubs, 30 male cubs (some individuals occur in more than one age-stage). Data from the marked animals are used to quantify puma population characteristics and vital rates in a reference situation without sport-hunting off-take as a mortality factor. Our efforts to quantify puma population characteristics and vital rates in a reference condition positioned us to develop a puma population model, and to use the population data and modeling scenarios to conduct a preliminary assessment of CDOW puma management assumptions and to guide directions for the remainder of the puma research on the Uncompahgre Plateau. Moreover, our data and model provide tools currently useful to CDOW wildlife biologists and managers for assessing puma harvest strategies. To improve data on puma population vital rates, attention will be given to increasing radio-collared sample sizes on life stages and sexes. The treatment period, scheduled to begin winter 2009-10 and to extend the following 5 years, will be a population-wide evaluation of sport-hunting impacts on a puma population. Furthermore, we will continue collaboration efforts with colleagues on investigations of puma population parameter estimation, puma-human relations, puma habitat modeling and mapping, wild felid disease surveillance, and individual puma detection rates in camera grid designs. All of these efforts should enhance the Colorado puma research and management programs. 145 �LITERATURE CITED Anderson, A. E. 1983. A critical review of literature on puma (Felis concolor). Colorado Division of Wildlife Special Report No. 54. _____, D. C. Bowden, and D. M. Kattner. 1992. The puma on Uncompahgre Plateau, Colorado. Technical Publication No. 40. Colorado Division of Wildlife, Denver. Anderson, C. R., Jr., and F. G. Lindzey. 2005. Experimental evaluation of population trend and harvest composition in a Wyoming cougar population. Wildlife Society Bulletin 33:179-188. Beier, P., and R. H. Barrett. 1993. The cougar in the Santa Ana Mountain Range, California. Orange County Cooperative Mountain Lion Study Final Report. Bergman, E. J., M. W. Alldredge, K. A. Logan, M. Shuette, C. J. Bishop, and D. J. Freddy. 2007. Study Plan-Pilot evaluation of predator-prey dynamics on the Uncompahgre Plateau. Pages 83-96 in Bergman, E.J. Evaluation of winter range habitat treatments on over-winter survival and body condition of mule deer. Wildlife Research Report July: 73-96. Colorado Division of Wildlife, Fort Collins. Bishop, C.J., G. C. White, and P.M. Lukacs. Evaluating dependence among mule deer siblings in fetal and neonatal survival analyses. Journal of Wildlife Management 72:1085-1093. Burnham, K. P., and D. R. Anderson. 1998. Model selection and inference: a practical informationtheoretic approach. Springer-Verlag, New York, New York, USA. Caughley, G. 1978. Analysis of vertebrate populations. John Wiley and Sons, New York. Colorado Division of Wildlife 2002-2007 Strategic Plan. 2002. Colorado Department of Natural Resources, Division of Wildlife. Denver. Colorado Division of Wildlife. 2007. Colorado mountain lion management data analysis unit revision and quota development process. Colorado Division of Wildlife, Denver. Conover, W. J. 1999. Practical nonparametric statistics. John Wiley & Sons, Inc., New York. Cooch, E., and G. White. 2004. Program MARK- a gentle introduction, 3rd edition. Colorado State University, Fort Collins. Culver, M., W. E. Johnson, J. Pecon-Slattery, and S. J. O’Brien. 2000. Genomic ancestry of the American puma (Puma concolor). The Journal of Heredity 91:186-197. Currier, M. J. P., and K. R. Russell. 1977. Mountain lion population and harvest near Canon City, Colorado, 1974-1977. Colorado Division of Wildlife Special Report No. 42. Heisey, D. M., and T. K. Fuller. 1985. Evaluation of survival and cause specific mortality rates using telemetry data. Journal of Wildlife Management 49:668-674. Kendall, W.L. 2001. The robust design for capture-recapture studies: analysis using program MARK. Pages 357-360 in R. Field, R. J. Warren, H. Okarma, and P. R. Sievert, editors. Wildlife, land, and people: priorities for the 21st century. Proceedings of the Second International Wildlife Management Congress. The Wildlife Society, Bethesda, Maryland, USA. Koloski, J. H. 2002. Mountain lion ecology and management on the Southern Ute Indian Reservation. M. S. Thesis. Department of Zoology and Physiology, University of Wyoming, Laramie. Kreeger, T. J. 1996. Handbook of wildlife chemical immobilization. Wildlife Pharmaceuticals, Inc., Fort Collins, Colorado. Laundre, J. W., L. Hernandez, D. Streubel, K. Altendorf, and C. L. Lopez Gonzalez. 2000. Aging mountain lions using gum-line recession. Wildlife Society Bulletin 28:963-966. Logan, K. A., E. T. Thorne, L. L. Irwin, and R. Skinner. 1986. Immobilizing wild mountain lions (Felis concolor) with ketamine hydrochloride and xylazine hydrochloride. Journal of Wildlife Diseases. 22:97-103. _____, L. L. Sweanor, J. F. Smith, and M. G. Hornocker. 1999. Capturing pumas with foot-hold snares. Wildlife Society Bulletin 27:201-208. _____, and L. L. Sweanor. 2001. Desert puma: evolutionary ecology and conservation of an enduring carnivore. Island Press, Washington, D.C. 146 �_____. 2004. Colorado puma research and development program: population characteristics and vital rates study plan. Colorado Division of Wildlife, Ft. Collins Research Center, Ft. Collins. _____. 2005. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. July: 105-126, Colorado Division of Wildlife, Fort Collins. _____. 2008. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Wildlife, Fort Collins. Murphy, K., M. Culver, M. Menotti-Raymond, V. David, M. G. Hornocker, and S. J. O’Brien. 1998. Cougar reproductive success in the Northern Yellowstone Ecosystem. Pages 78-112 in The ecology of the cougar (Puma concolor) in the Northern Yellowstone ecosystem: interactions with prey, bears, and humans. Dissertation, University of Idaho, Moscow. Otis, D. L., K. P. Burnham, G. C. White, and D. R. Anderson. 1978. Statistical inference from capture data on closed animal populations. Wildlife Monographs 62:1-135. Ott, R. L. 1993. An introduction to statistical methods and data analysis. Fourth edition. Wadsworth Publishing Co., Belmont, California. Pierce, B. K., V. C. Bleich, and R. T. Bowyer. 2000. Social organization of mountain lions: does land a tenure system regulate population size? Ecology 81:1533-1543. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989a. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. _____, S. R. Winterstein, and M. J. Conroy. 1989b. Estimation and analysis of survival distributions for radio tagged animals. Biometrics 45:99-109. _____, J. D. Nichols, C. Brownie, and J. E. Hines. 1990. Statistical inference for capture-recapture experiments. Wildlife Monographs 107:1-97. Pojar, T. M., and D. C. Bowden. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550-560. Ross, P. I., and M. G. Jalkotzy. 1992. Characteristics of a hunted population of cougars in southwestern Alberta. Journal of Wildlife Management 56:417-426. Stoner, D. C. 2004. Cougar exploitation levels in Utah: implications for demographic structure, metapopulation dynamics, and population recovery. Master of Science Thesis. Utah State University. Sweanor, L. L., K. A. Logan, and M. G. Hornocker. 2000. Cougar dispersal patterns, metapopulation dynamics, and conservation. Conservation Biology 14:798-808. _____, L. L., K. A. Logan, J. W. Bauer, B. Milsap, and W. M. Boyce. 2008. Puma and human spatial and temporal use of a popular California state park. Journal of Wildlife Management 72:1076-1084. Van Ballenberghe, V. 1983. Rate of increase of white-tailed deer on the George Reserve: a re-evaluation. Journal of Wildlife Management 47:1245-1247. Watkins, B. 2004. Mountain lion data analysis unit L-22 management plan. Colorado Division of Wildlife, Montrose. Williams, B. K., J. D. Nichols, and M. J. Conroy. 2001. Combining closed and open mark-recapture models: the robust design. Pages 523-554 In Analysis and management of animal populations. Academic Press, New York. White, G. C., D. R. Anderson, K. P. Burnham, and D. L. Otis. 1982. Capture-recapture and removal methods for sampling closed populations. Los Alamos National Laboratory Publication LA-8787NERP. Los Alamos, NM, U.S.A. Worton, B. J. 1995. Using Monte Carlo simulation to evaluate kernel-based home range estimators. Journal of Wildlife Management 59:794-800. Young, S. P. and E. A. Goldman. 1946. The puma: mysterious American cat. The Wildlife Institute, Washington, D. C. Prepared by: Kenneth A. Logan, Wildlife Researcher 147 �Table 1. Summary of puma capture efforts with dogs from December 9, 2008 to April 30, 2009, Uncompahgre Plateau, Colorado. Month No. Search Days 11 No. & type of puma tracks founda 16 tracks: 6 male, 6 female, 4 cub No. & type of pumas pursued 10 pursuits: 5 males, 5 females, 4 cubs No. & I.D. or type of pumas captured, observed, or identified December 6 pumas captured 8 times: M71 recaptured (not handled), M55 recaptured twice (not handled to change faulty GPS collar due to dangerous tree & cliffs), F93 captured twice- the first time, then with her 2 large cubs F95 and a male cub that could not be handled in a hole, F94 captured for the first time. In addition, male puma tracks found and attributed to M32 by VHF telemetry (no pursuit with hounds). January 17 38 tracks: 17 male, 17 pursuits: 6 males, 5 pumas captured 6 times: M55 (faulty GPS 10 female, 11 cub 4 females, 7 cubs collar changed), F93 recaptured while cub F95 and unmarked male cub escaped, F16 recaptured (faulty GPS collar changed) while M6 (consort) escaped, F96 captured for first time while 2 cubs escaped, F96 recaptured while 2 cubs escaped, cub M84 recaptured (handled to fit with new expandable cub collar). In addition, M6 and F16 were detected by tracks and identified with VHF telemetry on 2 other occasions. M51 was detected by tracks and identified with VHF telemetry and pursued, but was not caught to change his GPS collar on low battery. F93 and F95 were detected by tracks with non-marked cub and identified with VHF telemetry. February 15 64-65 tracks: 12-17 26 pursuits: 3-4 5 pumas captured: cub F97 captured for the first male, 26-31 female, males, 7-8 females, time while mother F23 & sibling F81 escaped. 24-27 cub 15 cubs Cub M82 recaptured and fit with new VHF collar, while mother F8 escaped and confirmed with VHF telemetry. Cub F98 captured for the first time; one of three cubs of an unmarked adult female puma visually observed with F98 on 217-09. M29 recaptured, but could not be handled in dangerous cliffs to replace faulty GPS collar. M99 captured for first time; sibling of F98. March 15 56 tracks: 24-26 15 pursuits: 4-5 4 pumas captured 5 times: F98 recaptured while male, 21-23 female, 9 males, 3-4 females, mother and 2 sibling cubs escaped, M99 cub 7 cubs captured for first time while mother and siblings F98 and non-marked cub escaped, M99 and nontagged cub visually observed, M100 captured for the first time. April 13 24-27 tracks: 17 0 pumas captured. One male pursued identified 7-10 pursuits: 4 male, 6 female, 1-4 males, 2 females, 1- as M55 with GPS data. Another male pursued cub identified as M100 with GPS data. Two females 4 cubs and their cubs pursued identified as F70 and F96 with 1-4 cubs with VHF telemetry. 71 198-202 tracks: 75-78 pursuits: 24 captures of 17 individuals: 4 independent TOTALS 76-83 male, 22-24 males, pumas (F93, F94, F96, M100) and 4 marked 69-76 female, 21-23 females, (F95, F97, F98, M99) and 2 non-marked cubs 49-55 cub 34-37 cubs were captured for the 1st time. 10 independent pumas were detected by tracks and identified with GPS/VHF telemetry 12 times: M6 (twice), F8, F16 (twice), M32, M51, M55, F70, F93, F96, M100. a Puma hind-foot tracks with plantar pad widths >50 mm wide are assumed to be male; ≤50 mm are assumed to be female (Logan and Sweanor 2001:399-412). b Pumas are not handled for a variety of safety reasons: tree too dangerous to climb for researchers, puma treed near river, creek or cliff, puma might fall from tree after drug induction. 148 �Table 2. Summary of puma capture efforts with dogs, December 2004 to April 2009, Uncompahgre Plateau, Colorado. Period Track detection effort 109/78 = 1.40 tracks/day Dec. 2, 2004 to May 12, 2005 Nov. 21, 2005 to May 26, 2006 Nov. 13, 2006 to May 11, 2007 Nov. 19, 2007 to April 24, 2008 Dec. 9, 2008 to April 30, 2009 35/78 = 0.45 pursuit/day Puma capture effort 14/78 = 0.18 capture/day Effort to capture an independent puma for the first time 11 pumas captured for first time 11/78 = 0.14 capture/day 78/14 = 5.57 day/capture 14/82 = 0.17 capture/day 78/11 = 7.09 day/capture 149/82 = 1.82 tracks/day 78/35 = 2.23 day/pursuit 43/82 = 0.52 pursuit/day 177/78 to 182/78 = 2.27-2.33 tracks/day 82/43 = 1.91 day/pursuit 45/78 to 47/78 = 0.58-0.60 pursuit/day 82/14 = 5.86 day/capture 22/78 = 0.28 capture/day 78/47 to 78/45 = 1.66-1.73 day/pursuit 49/77 = 0.64 pursuit/day 78/22 = 3.54 day/capture 78/7 = 11.14 day/capture 20/77 = 0.26 capture/day 7 pumas captured for first time 7/77 = 0.09 capture/day 77/20 = 3.85 day/capture 24/71 = 0.34 capture/day 77/7 = 11.00 day/capture 217/77 to 218/77 = 2.82-2.83 tracks/day 198/71 to 202/71 = 2.79-2.84 tracks/day Pursuit effort 77/49 = 1.57 day/pursuit 75/71 to 78/71 = 1.06-1.10 pursuit/day 71/75 to 71/78 = 0.91-0.95 day/pursuit 7 pumas captured for first time 7/82 = 0.08 capture/day 82/7 = 11.71 day/capture 7 pumas captured for first time 7/78 = 0.09 capture/day 9 pumas captured for first time 9/71 = 0.13 capture/day 71/24 = 2.96 day/capture 71/9 = 7.89 day/capture Table 3. Adult and subadult pumas captured for the first time, sampled, tagged, and released from December 2008 to May 2009, Uncompahgre Plateau, Colorado. Puma I.D. F93 F94 F96 M100 Sex F F F M Estimated Age (mo.) 72 41 36 72 Mass (kg) Capture date 12-15-08 12-19-08 01-28-09 03-27-09 32 36 40 64-68 estimated* F104 F 96 40 05-21-09 *M100 could not be weighed by scale due to steepness of terrain. 149 Capture method Dogs Dogs Dogs Dogs Location Coal Bank Canyon Shavano Valley Dolores Canyon San Miguel Canyon Cage Trap Roubideau Canyon �Table 4. Pumas that were captured and observed with aid of dogs, or observed in association with another radio-collared puma, but were not handled at that time for safety or other reasons, December 2008 to March 2009, Uncompahgre Plateau, Colorado. Puma sex Capture date Location Comments M55 Age stage or months 42 12-12-08 Dolores Canyon M55 Male 42 16 12-21-08 12-29-08 Spring Creek Dry Creek Basin Female Unk. adult 02-19-09 San Miguel Canyon Unknown 5 02-19-09 San Miguel Canyon M29 129 02-25-09 Unknown 6 03-11-09 Big Bucktail Canyon San Miguel Canyon M55 bayed in a hole then climbed a tree too dangerous for handling to replace non-functioning GPS collar. M55 bayed on cliffs too dangerous for handling. Non-marked male cub of F93 and sibling of F95 took refuge in narrow hole; unable to handle him. Non-marked adult female puma was visually observed with radio-collared cub F98 and a non-marked cub (either M99 later marked or non-marked sibling below), but could not be caught with dogs. Non-marked cub was visually observed with radio-collared cub F98 and non-marked adult puma, but could not be caught with dogs. M29 bayed in cliffs too dangerous for handling. Non-marked cub- sibling of F98 & M99- visually observed with radio-collared cub M99, but could not be caught with dogs. Table 5. Pumas recaptured with dogs, cage traps, or visually observed, December 2008 to January 2009, Uncompahgre Plateau, Colorado. Puma I.D. Recapture Date 12-08-08 Mass (kg) Observed Estimated Age (mo.) 35 Capture Method/ Location Dogs/Shavano Mesa M71 F93 12-29-08 Observed 72 Dogs/Dry Creek Basin F93 01-08-09 Observed 72 Dogs/Shavano Valley F96 01-29-09 Observed 36 Dogs/Dolores Canyon 150 Process M71 wore a functioning vhf collar; no need to handle him. F93 wore a functioning GPS collar; no need to handle her. F93 wore a functioning GPS collar; no need to handle her. F96 wore a functioning GPS collar; no need to handle her. �Table 6. Summary of puma capture efforts with ungulate road-kill baits and cage traps from August 20, 2008 to July 20, 2009, Uncompahgre Plateau, Colorado.a Carnivore activity & capture effort resultsb No puma activity detected. One deer carcass scavenged by coyotes. Deer carcasses scavenged by male puma 9-14-08; likely M55 (trail camera photos). Set cage trap 9-15-08. Puma did not return. Bobcat fed on deer carcass in cage trap. A bobcat, a black bear and domestic dogs scavenged 3 different deer carcasses. October 3 Deer carcass scavenged by bobcat. November 6 Female puma scavenged a deer carcass 11-21 to 22-08. Cage trap set 11-23-08; but, female puma did not return. Male puma scavenged a deer carcass 11-24-08, and cage trap set 11-2408. The male puma returned, walked around the trap, but did not enter. Female puma and bobcat scavenged a carcass 11-24-08. Cage trap set 11-24-08. Bobcat captured and released 1124-08. Subadult female puma F66 recaptured and radio-collared 11-25-08. December 2 No puma activity detected. February 2 A female or small male puma walked ~20 m past a deer carcass but did not feed. Another deer carcass was scavenged by a bobcat. March 4 A male puma walked ~5 m past 2 different deer carcasses but did not feed. Three deer carcasses were scavenged by 2 gray foxes and 2 bobcats. April 1 Male puma M55 scavenged a deer carcass 5-6-09. No need to recapture him. May 4 Female puma fed on a deer carcass 5-8 to 10-09. Set cage trap 5-11-09. Female puma returned but did not enter cage trap. Set 2 cage traps 5-12-09; but female puma did not return. Female puma (possibly same as previous) scavenged deer carcass 5-21-09. Cage trap set 5-21-09. F104 captured. A black bear scavenged one deer carcass. July 2 Puma F72b was recaptured 7-20-09 in cage trap set for bobcat study. Her malfunctioning GPS collar was replaced. A non-marked puma was photographed at one deer bait 7-17-09; but it did not feed. Same deer bait was scavenged by ~5 different black bears. a We used 36 road-killed mule deer at 17 different sites. Of the road-killed ungulate baits, 7 of 36 (19.4%) were scavenged by pumas. b Adult female puma F72 was recaptured in a bobcat cage trap baited with a predator call box and visual attractant. Month August September No. of Sites 2 5 Table 7. Puma cubs sampled September 2008 to June 2009 on the Uncompahgre Plateau Puma Study area, Colorado. a Cub I.D. Sex Estimated birth datea Estimated age at capture (days) Mass (kg) Mother Estimated age of mother at birth of this litter (mo) M91 M92 F95 F97 F98 M99 M101 M102 F103 M105 F106 M107 F108 M109 M M F F F M M M F M F M F M August 19, 2008 August 19, 2008 August 2007 May 23, 2008 Sep.-Oct. 2008 Sep.-Oct. 2008 April 15, 2009 April 15, 2009 April 15, 2009 May 7, 2009 May 7, 2009 May 25, 2009 May 25, 2009 May 25, 2009 35 35 488 257 122-152 152 35 35 35 38 38 34 34 34 2.5 2.8 33 20 9.5 13.6 2.8 2.5 2.1 2.6 2.6 2.0 1.75 1.75 F25 F25 F93 F23 Fb Fb F16 F16 F16 F75 F75 F94 F94 F94 110 110 56 45 Unk. Unk. 75 75 75 55 55 46 46 46 Estimated age of cubs sampled at nurseries is based on the starting date for GPS location and radio-telemetry foci for mothers at nurseries, and development characteristics of cubs with mother only with radio-telemetry. b F98 and M99 were captured in association with an non-marked adult female puma and another non-marked cub. 151 �Table 8. Minimum puma population estimate based on numbers of known radio-collared pumas, visual observations of non-marked pumas, and track counts of suspected non-marked pumas on the study area during the past 2 winters, November 2007 to March 2008 and November 2008 to April 2009, Uncompahgre Plateau study area, Colorado. Adults Winter & Region Nov.07-Mar.08 East slope West slope Totals Female Male Subadults Female Male 10 4 3 6 4 2 16 8 5 Total Independent Pumas = 33a,b 4 0 4 Nov.08-Apr.09 East slope West slope Totals Female Cubs Male Unknown sex 4 1 5 4 2 6 7 2-3 9-10 11-13 5-6 2-4 0-1 2 5 5 9-10 4 1-2 1 3 2 4 20-23 9-10 3-6 1-2 5 7 9 Total Independent Pumas = 37c,d a Of the total, 23−24 (70−73%) independent pumas were marked and 9-10 (27−30%) were assumed to be non-marked, but some might have ear-tags, tattoos, or non-functional GPS/VHF collars. b The non-marked independent pumas included: 1adult female with 2 large cubs in Happy Canyon, 1 adult female with 1 large cub in Potter Creek and 25-mile Mesa, 1 adult female with 2 large cubs in Monitor Creek, 1 adult female with 2 medium-size cubs in Potter Creek, 1 adult female with 2-3 cubs in San Miguel Canyon, and 1 female or F28 with a non-functional collar Big Bucktail Creek to San Miguel Canyon. c Of the total, 23−25 (62−68%) independent pumas were marked and 12-14 (32−38%) were assumed to be non-collared, but some might have ear-tags, tattoos, or non-functional GPS/VHF collars. d The non-marked independent pumas included: 1 adult female with 2 cubs on N. McKenzie Mesa, 1 subadult or adult female in Linscott Creek, 1 adult female in Monitor Creek, 1 subadult or adult female in Roubideau Canyon, 1 subadult or adult male in Monitor Creek, 1 adult female with 3 cubs in San Miguel Canyon, 1 adult female with ≥1 cub or F28 with a non-functional GPS collar in Big Bucktail Canyon to N. Fork Cottonwood Creek, 1 adult female or F24 with non-functional GPS collar in Horsefly Creek to Dead Horse Mesa, 1 adult female or F28 with non-functional GPS collar in San Miguel Canyon W of Pinion, 1 adult female with ≥1 cub on Mailbox Park, 1 adult female with 1 cub from McKenzie Creek to Iron Springs Mesa. 1 subadult or adult female on Iron Springs Mesa, 1 subadult female in Big Bucktail Canyon to ridge E of Nucla, 1 subadult male from Pinion across Big Bucktail Canyon and ridge E of Nucla. 152 �Table 9. Individual puma reproduction histories, Uncompahgre Plateau, Colorado, 2005-2009. Consort pairs and estimated agesa Dates pairs Estimated Estimated Estimated Observed consortedb birth datec birth interval gestation number of Female Age Male Age (mo.) (days) cubsd (mo.) (mo.) F2 53 05/28/05 3 F2 67 07/29/06 14.0 2 F2 89 05/19/08 22.0 4 F3 36 08/01/04 1 F3 50 M6 37 06/22-24/05 09/26/05 13.8 93-95 2 F3 62 09/17/06 11.7 3 F3 84 M51 60 03/31/08 07/03/08 21.5 94 3 F7 67 05/19/05 2 F7 82 08/13/06 14.9 4 F7 106 07/10/08 23.9 3 F8*e 24 06/26/05 2 F8 37 08/13/06 13.4 4 F8 60 M73 49 02/28-29/08 05/29/08 22.5 90-91 2 F16 32 09/22/05 4 F16 52 05/24/07 19.9 4 F16 75 M6 80 01/13-14/09 04/15/09 22.7 91-92 3 F23* 21 05/30/06 3 F23 45 M27 or 78 02/19-25/08 05/23/08 23.8 87-93 3 M29f 107 F24 75 M29 92 04/12-15/07 06/14/07 90-93 4 F25 74 08/01/05 1 F25 94 04/16/07 20.5 1 F25 110 08/19/08 16.1 2 F28* 36 06/09/06 2 F28 48 M29 88 12/27-29/06 03/30/07 11.7 92-93 ≥2 tracks F30* 48 M55 34 04/16-20/07 07/17/07 88-92 3 F50 21 07/01/06 1 F54 24 07/01/06 1 F70* 38 M51 60 03/10/08 06/05/08 87 3 F72* 28 07/09/08 1 F75 32 06/01/07 1 F75 55 M73 61 02/11/09 05/07/09 23 93 2 F93 56 08/07 2 F94* 46 05/27/09 3 a Ages of females were estimated at litter birth dates. Ages of males were estimated around the dates the pairs consorted. b Consort pairs indicate pumas that were observed together based on GPS data or VHF location data. c Estimated birth dates were indicated by GPS data of mothers at nurseries or by back-aging cubs to approximate birth date. d Observed number of cubs do not represent litter sizes as some cubs were observed when they were 5 to 16 months old after postnatal mortality could have occurred in siblings. Only cub tracks were observed with F28. e Asterisk (*) indicates first probable litter of the female, based on nipple characteristics noted at first capture of the female. f A radio-collared, ear-tagged male puma was visually observed with F23 on 2/25/08. Both M27 and M29 wore non-functional GPS collars in that area at the time. 153 �Table 10. Summary for individual adult puma survival and mortality, December 8, 2004 to July 31, 2009, Uncompahgre Plateau, Colorado. Puma I.D. M1 Monitoring span 12-08-04 to 08-16-06 No. days 616 M4 M5 01-28-05 to 12-28-05 08-01-06 to 02-20-09 333 934 M6 M27 02-18-05 to 07-31-09 03-10-06 to 05-07-09 1,624 1,154 M29 04-14-06 to 02-25-09 1,048 M32 M51 M55 M71 M73 M100 F2 04-26-06 to 07-31-09 01-07-07 to 03-20-09 01-21-07 to 07-31-09 01-29-08 to 07-31-09 02-21-08 to 07-31-09 03-27-09 to 07-31-09 01-07-05 to 08-14-08 1,192 803 922 549 526 126 1,315 F3 F7 01-21-05 to 01-15-09 02-24-05 to 08-03-08 1,455 1,256 F8 F16 F23 F24 F25 F28 F30 03-21-05 to 07-31-09 10-11-05 to 07-31-09 02-05-06 to 07-31-09 01-17-06 to 09-03-08 02-08-06 to 07-31-09 03-23-06 to 09-25-07 04-15-06 to 07-29-08 1,593 1,389 1,272 960 1,269 551 836 F50 12-14-06 to 03-26-07 102 F54 01-12-07 to 08-18-07 218 F70 F72 F75 F93 F94 F96 F104 01-14-08 to 07-31-09 02-12-08 to 07-31-09 03-26-08 to 07-31-09 12-05-08 to 07-31-09 12-19-08 to 07-31-09 01-28-09 to 07-31-09 05-21-09 to 07-31-09 564 535 492 238 224 184 71 Status: Alive/Lost contact/Dead; Cause of death Lost contact− failed GPS/VHF collar. M1 ranged principally north of the study area as far as Unaweep Canyon. Dead; killed by a male puma. Estimated age at death 37−45 months. Dead. Born on study area; offspring of F3. M5 was independent of F3 by 13 months old, and dispersed from his natal area at about 14 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months (protected from hunting mortality in buffer area) and ranged into the eastern edge of Utah (vulnerable to hunting). Killed by a puma hunter on 02-20-09 in Beaver Creek, Utah at age 54 months. Alive. Lost contact− failed GPS/VHF collar. Recaptured 12-02-07 & 01-2208 by puma hunter/outfitter north of the study area. Possibly visually observed on study area with F23 on 02-25-08. Recaptured by a puma hunter/outfitter 12-11-08 & 12-28-08 north of the study area. Photographed by a trail camera on the study area (Big Bucktail Canyon) on 5 occasions: 03-27-09, 04-02-09, 04-15-09, 04-24-09, & 05-07-09. Lost contact− failed GPS/VHF collar. Possibly visually observed on study area with F23 on 02-25-08. Recaptured on study area 02-25-09, but could not be safely handled to change faulty GPS collar. Alive. Alive. Lost contact− failed GPS/VHF collar. Alive. Alive. Alive. Alive. Dead; killed by another puma (sex of puma unknown; male suspected) 08-14-08. Estimated age at death 92 months. Lost contact− failed GPS/VHF collar. Dead 08-03-08; killed by U.S., W.S. agent for predator control of depredation on domestic sheep. Estimated age at death 107 months. Alive. Alive. Alive. Lost contact− failed GPS/VHF collar. Alive. Lost contact− failed GPS/VHF collar. Dead; killed by another puma (sex of puma unknown) 07-29-08. Estimated age at death 60 months. Dead of natural causes 03-26-07; probably injury or illness-related; exact agent unknown. Estimated age at death 30 months. Dead; killed by a male puma while in direct competition for prey (i.e., mule deer fawn) 08-18-07. Estimated age at death 49 months. Alive. Alive. Alive. Alive. Alive. Alive. Alive. 154 �Table 11. Preliminary estimated survival rates (S) of adult-age pumas during the reference period (i.e., the study area is closed to puma hunting), Uncompahgre Plateau, Colorado. Survival rates of pumas estimated with the Kaplan-Meier procedure to staggered entry of animals (Pollock et al. 1989). Survival rates are for an annual survival period defined as the biological year (August 1 to July 31) and the hunting season period (November 1 through March 31). Survival rates were estimated only for periods when n ≥ 5 individual pumas were monitored in the interval. Puma deaths in this analysis pertained only to pumas that died of natural causes. Pumas that were killed by people, a non-natural cause (i.e., F7 for depredation control 8/3/2008 and M5 killed by a puma hunter off the protected study area and buffer zone 2/20/2009) were right censored. Period of interest Annual 8/1/2005 to 7/31/2006 Annual 8/1/2006 to 7/31/2007 Annual 8/1/2007 to 7/31/2008 Annual 8/1/2008 to 7/31/2009 Hunting season 11/1/2005 to 3/31/2006 Hunting season 11/1/2006 to 3/31/2007 Hunting season 11/1/2007 to 3/31/2008 Hunting season 11/1/2008 to 3/31/2009 S 1.000 Females SE 0.0000 n 10 S 0.667* Males SE 0.2222* n 6* 0.909 0.0867 11 1.000 0.0000 5 0.831 0.0986 14 1.000 0.0000 7 0.875 0.1031 13 1.000 0.0000 8 1.000 0.0000 6 na na 4 0.909 0.0867 11 1.000 0.0000 5 1.000 0.0000 12 1.000 0.0000 9 1.000 0.0000 11 1.000 0.0000 8 Adult male annual S 2005 to 2006 is probably underestimated with poor precision because 3 of the 6 pumas were GPS/VHF-monitored for 4 to 5 months at the end of the interval; 1 of 6 adult males died. 155 �Table 12. Summary of subadult puma survival and mortality, December 2004 to July 2009, Uncompahgre Plateau, Colorado. Puma I.D. M5 Monitoring span 09-16-05 to 06-30-06 No. days 308 M11 06-21-06 to 12-02-07 529 F23 01-04-06 to 02-04-06 04-19-06 to 04-26-06 31 M49 03-26-07 to 10-01-07 189 F52 01-10-07 to 05-15-07 125 F66 08-23-07 to 11-05-07 11-25-08 to 06-03-09 74 M31 7 190 M69 01-11-08 to 04-07-08 87 F95 12-29-08 to 07-31-09 214 Status M5 was offspring of F3, born August 2004. Independent and dispersed from natal area at 13 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months (protected from hunting mortality in buffer area) and ranged into the eastern edge of Utah (vulnerable to hunting). Killed by a puma hunter on 02-20-09 in Beaver Creek, Utah at about 54 months old. M11 was offspring of F2, born May 2005. Independent at 13 months old. Dispersed from natal area at 14 months old. Moved to Dolores River valley, CO, by 12-14-06. Killed by a puma hunter on 12-02-07 when about 30 months old. Alive. Captured on the study area when about 17 months old. Survived to adult stage; gave birth to first litter at about 21 months old. M31’s estimated age at capture was 20 months. Dispersed to northern New Mexico and was killed by a puma hunter on 12-11-08 in Middle Ponil Creek, Cimarron Range. He was about 52 months old. M49 was offspring of F50, born July 2006. Orphaned at about 9 months old, when F50 died of natural causes. Dispersed from his natal area at about 10 months old and ranged on the northeast slope of the Uncompahgre Plateau. When M49 was about 15 months old, he shed his expandable radiocollar on about 10-01-07 at a yearling cow elk kill on the northeast slope of the Uncompahgre Plateau. He was killed by a puma hunter in Blue Creek in the protected buffer zone north of the study area on 01-24-09; he was about 29 months old. F52 dispersed from study area as a subadult by Jan. 16, 2007. F52’s last VHF aerial location was Crystal Creek, a tributary of the Gunnison River east of the Black Canyon 05-15-07. She was treed by puma hunters on 12-29-08 on east Huntsman Mesa, southeast of Powderhorn, CO. She was about 41-43 months old and could have been in her adultstage home range. GPS collar nonfunctional. F66 was offspring of F30, born July 2007. Lost contact; her cub collar quit after 11-05-07. Recaptured as an independent subadult on her natal area 11-25-08 when 16 months old. F30 was killed by a puma when F66 was 12 months old, within the age range of normal independence. F66 died of injuries to internal organs that caused massive bleeding attributed to trampling by an elk or mule deer on about 05-28-09 when she was 23 months old. Her range partially overlapped her natal area. M69 was captured on the study area when about 14-18 months old. Emigrated from the study area as subadult by 03-19-08. Last VHF aerial location was southwest of Waterdog Peak, east side of Uncompahgre River Valley on 04-07-08. M69 was killed by a puma hunter on 11-0608 in Pass Creek in the Snowy Range, WY when he was 24 to 28 months old. Alive. F95 is the offspring of F93, born about August 2007. She became an independent subadult by about 18 months old (02-11-09 aerial location). She has been ranging adjacent to and overlapping the northern portion of her natal area. 156 �Table 13. Records of pumas that dispersed from the Uncompahgre Plateau study area, December 2004 to July 2009. Puma I.D. 1st capture date on study area 1st capture location→kill or resite location (UTM, NAD27) M5 02-04-05 13S,240577Ex 4251037N→ 12S,665853Ex 4277125N M11 06-27-05 13S,248278Ex 4239858N→ 12S,741882Ex 4161575N 84.8 M31 04-19-06 329.8 M43 09-15-06 12S,746919Ex 4225441N→ 13S,500000Ex 4050000N 12S,760177Ex 4242995N→ 12S,739859Ex 4308557N M49 12-05-06 12S,757241Ex 4258259N→ 12S,693350Ex 4274559N 66.1 M68 08-23-07 80.7 M69 01-11-08 13S,257371Ex 4235231N→ 12S,711262Ex 4198681N 13S,248191Ex 4246810N→ 13T,378900Ex 4591990N F52 01-10-07 13S,258058Ex 4236260N→ 13S,319217Ex 4240467N Estimated linear dispersal distance (km)* 102.2 68.6 369.6 61.1 Puma Information M5 was offspring of F3, born August 2004. Independent and dispersed from natal area at 13 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months (protected from hunting mortality in buffer area) and ranged into the eastern edge of Utah (vulnerable to hunting). Killed by a puma hunter on 02-20-09 in Beaver Creek, Utah at about 54 months old. M11 was offspring of F2, born May 2005. Shed expandable radiocollar 10-24 to 11-08-05. Recaptured and re-collared 04-0206. Independent at 13 months old. Dispersed from natal area at 14 months old. Moved to Dolores River valley, CO, by 12-14-06. Killed by a puma hunter on 12-02-07 when about 30 months old. M31’s estimated age at capture was 20 months. Dispersed to northern New Mexico and was killed by a puma hunter on 12-1108 in Middle Ponil Creek, Cimarron Range. He was about 52 months old. M43 was offspring of F7, born August 2006. He shed the expandable radiocollar 11-7 to 17-06, after which direct contact was lost. M43 was killed by a puma hunter 01-28-09 in Deer Creek, west slope of Grand Mesa, CO when he was 29 months old. M49 was offspring of F50, born July 2006. Orphaned at about 9 months old, when F50 died of natural causes. Dispersed from his natal area at about 10 months old and ranged on the northeast slope of the Uncompahgre Plateau. When M49 was about 15 months old, he shed his expandable radiocollar on about 10-01-07 at a yearling cow elk kill on the northeast slope of the Uncompahgre Plateau. He was killed by a puma hunter in Blue Creek in the protected buffer zone north of the study area on 0124-09; he was about 29 months old. M68 was offspring of F30, born July 2007. He was orphaned at 12 months old when his mother was killed by a puma. He was killed by a puma hunter in the Disappointment Valley in southwest CO on 12-30-08; he was 17 months old. M69 was captured on the study area when about 14-18 months old. Emigrated from the study area as subadult by 03-19-08. Last VHF aerial location was southwest of Waterdog Peak, east side of Uncompahgre River Valley on 04-07-08. M69 was killed by a puma hunter on 11-06-08 in Pass Creek in the Snowy Range, WY when he was 24 to 28 months old. F52 was captured on the study area when about 18-20 months old. Dispersed from study area as a subadult by Jan. 16, 2007. F52’s last VHF aerial location was Crystal Creek, a tributary of the Gunnison River east of the Black Canyon 05-15-07. She was treed by puma hunters on 12-29-08 on east Huntsman Mesa, southeast of Powderhorn, CO. She was about 41-43 months old and could have been in her adult-stage home range. *Estimated linear dispersal distance (km) from initial capture site on Uncompahgre Plateau study area to hunter kill or recapture site. 157 �Table 14. Recorded deaths of non-marked pumas and of marked pumas struck by vehicles, in chronological order, on the Uncompahgre Plateau puma study area, Colorado, from 2004 to 2009. Puma sex & ID if marked M Estimated age (mo) Date recorded Cause of death General physical condition Location & UTM NAD27 12 09-24-04 Good F 49 07-28-05 Vehicle collision Vehicle collision Pleasant Valley, County Road 24 13S,252870Ex4227520N Highway 62 east of Dallas divide 13S,250000Ex4222500N F F17 F 11 08-18-06 18-24 11-06-06 F 6 01-30-07 F 36 09-16-08 M 12-24 08-13-08 F F61 18 11-13-08 F 12 08-10-09 Vehicle collision Vehicle collision Vehicle collision Asphyxia, lodged in fork of tree Vehicle collision Vehicle collision Good Not pregnant or lactating Good Good Good Unknown Good Good Vehicle collision Good 158 Highway 550 south of Colona 13S,257602Ex4242185N Highway 550 east of Ridgway State Park 13S,259843Ex4235985N Highway 62 west of Dallas divide 12S,762286Ex4218992N Davis Point, Roubideau Canyon 12S, 743718Ex4255277N Highway 145 west of Placerville 13S,756490Ex4212336N Highway 550 east of Ridgway State Park 13S,259843Ex4235985N Highway 145 east of Norwood 12S,745739Ex4222548N �Table 15. Summary of puma mother and cub associations by distance (m) during airplane flights, each winter, Uncompahgre Plateau, Colorado. Monitoring period Month No. flights No. puma familiesa Ages of cubs (mo.) No. observations with mothers & cubs ≤660 m apart Nov. 9, 2005 to Nov. 3 4 2−6 9 Mar. 29, 2006 Dec. 4 4 3−7 16 Jan. 5 4 4−8 17 Feb. 4 5 5−9 16 Mar. 2 5 6−10 9 Totals 18 4−5 2−10 67 Nov. 7, 2006 to Nov. 4 4 2−3 11 Mar. 22, 2007 Dec. 4 4 2−5 11 Jan. 5 3 4−6 10 Feb. 4 4 5−7 10 Mar. 3 1 8 2 Totals 20 1−4 2−8 44 Nov. 13, 2007 to Nov. 2 1 6 1 Feb. 14, 2008 Dec. 0 1 7 NA Jan. 3 1 8 2 Feb. 3 1 9 2 Totals 8 1 6-9 5 Nov. 6, 2008 to Nov. 3 5 3-6 10 Mar. 20, 2009 Dec. 1 4 4-7 4 Jan. 2 6 5-17 8 Feb. 2 4 7-9 6 Mar. 3 2 7-10 5 Totals 11 2-6 3-17 33 a All puma mothers wore GPS-radiocollars. At least 1 cub in the litter wore a VHF radiocollar. b Mean = 1,097 m, SD = 313.95, range = 670−1,600. c Mean = 1,606 m, SD = 1,665.39, range = 678−4,101. d Mean = 1,341 m, SD = 542.34, range = 759−1,832. e Mean = 2,608 m, SD = 3,360.56, range = 799-7,641. 159 No. observations with mothers & cubs >660 m apart 2 4 3 2 0 11b 0 0 2 1 1 4c 1 NA 1 1 3d 0 0 3 0 1 4e �Table 16. Numbers of GPS locations and spans of monitoring for pumas captured on the Uncompahgre Plateau, Colorado, December 2004 to July 2009. Puma I.D. M1 M4 M6 M27 M29 M51 M55 M100 F2 F3 F7 F8 F16 F23 Sex Age stage Dates monitored a M M M M M M M M F F F F F F No. locations adult 12-08-04 to 07-20-06 1,797 adult 01-28-05 to 01-14-06 958 adult 02-18-05 to 05-14-08 1,035 adult 03-12-06 to 06-21-06 313 adult 04-14-06 to 01-01-08 1,599 adult 01-07-07 to 07-15-08 1,643 adult 01-21-07 to 04-22-09 1,887 adult 03-27-09 to 06-30-09 318 adult 01-07-05 to 08-14-08 3,516 adult 01-21-05 to 05-14-08 3,344 adult 02-24-05 to 08-03-08 3.922 adult 03-21-05 to 10-10-06 1,541 adult 10-12-05 to 05-13-09 3,157 subadult, 01-04-06 to 02-04-06 113 adult 02-05-06 to 04-22-09 1,083 F24 F adult 01-17-06 to 07-25-07 1,812 F25 F adult 02-09-06 to 06-26-09 3,398 F28 F adult 03-24-06 to 08-15-07 1,499 F30 F adult 03-30-07 to 02-22-08 1,057 F50 F adult 12-14-06 to 03-26-07 352 F52 F subadult 01-10-07 to 05-08-07 383 F54 F adult 01-12-07 to 08-18-08 723 F70 F adult 01-14-08 to 04-29-09 1,486 F72 F adult 02-12-08 to 06-23-09 1,186 F75 F adult 03-26-08 to 06-03-09 1,112 F96 F adult 01-28-09 to 04-29-09 235 F104 F adult 05-29-09 to 08-19-09 274 a GPS collars on pumas were remotely downloaded at approximately 1-month intervals, except during winter 20082009 to summer 2009 due to shortage of technicians during hiring freeze to assist in airplane flights to obtain downloads and to capture pumas to replace GPS collars (lengthening the download interval saved battery power). The last date in Dates monitored includes last location from the last GPS data download acquired for an individual puma. 160 �GOAL: Strategies, Information, & Tools for Managing Healthy, Self-sustaining Puma Populations in Colorado Puma Habitat Puma Population Effects of Harvest & Other Mortality Movements & Corridors Population Dynamics: Sex & Age, Vital Rates, Mortality, Population Growth Rates Vulnerability to Harvest Human Development Habitat Use Effects of Translocation Domestic Animals Puma― Human Relationships Deer, Elk, Other Natural Prey & Species of Concern Effects of Predation Effects of Evasive Conditioning Map Prey Distribution Habitat Models Population Models Puma Prey Puma―Prey Relationships Models Estimation Methods for Monitoring Habitat Maps Figure. 1. An ecologically-based conceptual model of the Colorado Puma Research Program that provides the contextual framework for this and proposed puma research in Colorado. Grayshaded shapes identify areas of research addressed by puma research on the Uncompahgre Plateau for the puma management goal in Colorado (at top). 161 �Figure 2. The puma study area on the southern half of the Uncompahgre Plateau, Colorado (shaded in gray) comprising the southern portions of Game Management Units (GMUs) 61 and 62 and a northern portion of GMU 70. Agt': :(U·nr:tnrt nf indtpnul,uttrnmas ('apnn·~d and ~•nn1ll~d fo1· the fil':(f. t.im~ from Det t.tnl>er 2tl04 fo Pvfa1·d1 20ftR. l :n tom)lahgl't': Plat.e~m, g Colnl'Mlo. 7 f, " s !: ~ a. 0 ■ ~@rn;ilf-: 4 .... ci 3 z 2 l 0 lto 2 >2to3 >3to 4 >4 to 5 >~ toG >Gto 7 >7 to8 >8 to9 >9 to Age {vears) 10+ :ID Figure 3. Age structure of independent pumas captured and sampled for the first time on the Uncompahgre Plateau, Colorado, December 2004 to May 2009. 162 �Puma births, Uncompahgre Plateau, Colorado 10 9 8 7 6 5 4 3 2 1 0 Jan . Feb. M ar. Apr. May June ■ Births 2005-2009 July Aug. Sep. Oct. Nov. Dec. ■ Births 1982-1987 Figure 4. Puma births detected by month during the reference period (i.e., no puma hunting), 2005 to 2009 (n = 27 litters of 14 females; 26 of the litters were examined at nurseries when cubs were 26-42 days old and 1 litter confirmed by tracks of ≥2 cubs following GPS-collared mother F28 when cubs were ~42 days old), and during the earlier effort by Anderson et al. (1992:48; 1982 to 1987, n = 10 litters of 8 females, examined when cubs were <1-8 months old), Uncompahgre Plateau, Colorado. Age structure of independent.pumas in March 2009, of surviving pumas captured and sampled from December 2004 , while protected from sport-hunting since April 2004, UncompahgrePlateau, Colorado. 4 ,,, 3 v> E ::i c.. 2 0 ■ Fema le ci z 1 ■ M ale 0 lto2 >2to 3 >3to 4 >4 to5 >5to6 >6to7 >7to8 >8to9 >9to10 10+ Age {years) Figure 5. Age structure of surviving independent pumas captured and sampled on the Uncompahgre Plateau, Colorado in March 2009, and after protection from sport-hunting mortality since April 2004, which includes 5 hunting seasons (Nov. through Mar., 2004-05 to 2008-09). One human-caused mortality (F7 killed for depredation control 08-03-08) was documented in the radio- and GPS-collared sample of independent pumas on the study area. This age structure assumes that pumas F3, M29, and M51 were alive on March 31, 2009; they each had non-functional GPS collars and were detected alive as late as 1-15-09, 02-25-09, and 03-20-09, respectively. Mean ± SD of adult female and adult male ages, respectively: 5.21 ± 2.29 yr. (62.54 ± 27.42 mo.); 6.31 ± 1.87 yr. (75.67 ± 22.45 mo.). 163 �APPENDIX A Appendix A. Summary of individual puma cub survival and mortality, 2005 to 2009, Uncompahgre Plateau, Colorado. Puma I.D. Estimated Est. Est. survival span Age to last monitor date Status: Alive/Survived to subadult stage/ from 1st capture to Age at Birth alive or at death (days, Lost contact/Disappeared/ fate or last monitor capture date birth to fate) Dead; Cause of death (days) date M5 183 ~8-1-04 02-04-05 to ~1,345 Survived to subadult stage by 09-16-05; independent at ~13 04-07-08 mo. old. Dispersed from natal area by 09-29-05 at 14 mo. old. Established territory on NW U.P. Killed by hunter in Beaver Creek, UT 02-20-09 at 4 ½ years old. F9 31 5-28-05 06-27-05 to 326-333 Lost contact― shed radiocollar 04-19-06 to 04-26-06. 4-19-06 F10 31 5-28-05 06-27-05 to 176-215 Lost contact― shed radiocollar 11-20-05― 08-10-05; last tracks of F10 with mother F2 & siblings F9 & 12-29-05 M11 observed 11-20-05. F10 disappeared by 12-30-05. M11 31 5-28-05 06-27-05 to Survived to subadult stage by 12-2-07 06-21-06, independent at 13 mo. old. Dispersed from natal area by 07-11-06 at 14 mo. old. Killed by a hunter in SW 918 CO 12-2-07 at 918 days (30 mo.) old F12 42 5-19-05 07-01-05 to 203-252 Lost contact― shed radiocollar 07-28-05―08-01-05. 12-08-05― Tracks of F12 found in association with mother F7 on 1201-26-06 08-05. F12 disappeared by 01-27-06 when she was not visually observed with F7, and her tracks were not seen in association with F7’s tracks. F13 42 5-19-05 07-01-05 to 101 Dead; killed and eaten by a puma (sex unspecified) about 808-28-05 28-05. F14 26 6-26-05 07-22-05 to 226-257 Lost contact― shed radiocollar 01-20-06 to 01-25-06. 02-07-06― Tracks of F14 were observed with tracks of mother F8 & 03-10-06 sibling M15 on 02-07-06. Disappeared by 03-11-06, only tracks of F8 & M15 were found. M15 26 6-26-05 07-22-05 to 345-353 Lost contact― shed radiocollar 06-06-06 to 06-14-06. 06-06 to 14-06 F17 34 9-22-05 10-26-05 to 330 Dead. Lost contact― shed radiocollar 06-06-06 to 06-14-06. 08-18-06 Killed by a car on highway 550 on 08-18-06. Probably dependent on F16. F18 34 9-22-05 10-26-05 to 301-308 Dead; probably killed by another puma. Multiple bite 07-20 to 27-06 wounds to skull. 10 mo. old. M19 34 9-22-05 10-26-05 to 308-314 Lost contact― shed radiocollar 07-27-06 to 08-02-06. 07-27 to 08-02-06 M20 34 9-22-05 10-26-05 to 244-245 Lost contact― shed radiocollar 05-24-06―05-25-06. 05-24-06 F21 37 9-26-05 11-02-05 to 324 Lost contact; radiocollar quit. Last aerial location 8-16-06, 08-16-06 live signal. 164 Mother I.D. F3 F2 F2 F2 F7 F7 F8 F8 F16 F16 F16 F16 F3 �Appendix A continued Puma I.D. Estimated Age at capture (days) M22 37 9-26-05 M26 183 8-1-05 F33 31 5-30-06 F34 31 5-30-06 F35 31 5-30-06 F36 29 6-9-06 M37 29 6-9-06 M38 41 7-29-06 M39 29 8-13-06 F40 29 8-13-06 F41 29 8-13-06 M42 29 8-13-06 M43 33 8-13-06 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 11-02-05 to 12-21-05― 12-22-05 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 86-87 Dead; killed and eaten by male puma 12-21-05―12-22-05. F3 02-08-06 to 03-21 to 24-06 06-30-06 to 07-31-06 06-30-06 to 07-31-06 ~232-235 Lost contact― shed radiocollar 03-21-06―03-24-06. F25 F23 06-30-06 to 07-07-06 07-08-06 to 07-28-06 07-08-06 to 07-28-06 09-08-06 to 07-16 to 17-07 38 Dead. Probably killed and eaten by a male puma 08-01 to 03-06. GPS data on M29 indicate he was not involved. Dead. Probably killed and eaten by a male puma 08-01 to 03-06. GPS data on M29 indicate he was not involved. Dead; research-related fatality.a Dead. Killed and eaten by a male puma 08-22-06. GPS data on M29 indicate he was not involved. Dead. Killed and eaten by a male puma 08-22-06. GPS data on M29 indicate he was not involved. Lost contact― shed radiocollar found 03-06-07. Photo (trail camera in McKenzie Cr.) of M38 & Unm. F sibling with F2 on 07-16 to 17-07 at 352-353 days old. F28 09-11-06 to 09-20-06 to 04-25-07 09-11-06 to 09-20-06 to 04-25-07 09-11-06 to 10-05-06 09-11-06 to 11-27-06 09-15-06 03-01-07 63-65 63-65 74 74 352-353 9 255 9 255 53-61 106 200 165 Mother I.D. F23 F23 F28 F2 Lost contact― shed radiocollar by 09-20-06, but seen alive on that date. Tracks of 2 cubs following F8 on 04-25-07. F8 Lost contact― shed radiocollar by 09-20-06, but seen alive on that date. Tracks of 2 cubs following F8 on 04-25-07. F8 Assumed dead. Lost Contact― shed radiocollar or died (blood on collar) between 10-05-06 (last live signal) & 1013-06 (collar found). Dead; research-related fatality.b F8 Lost contact− shed radiocollar by 11-7 to 17-06. Treed, visually observed 03-01-07. Killed by a puma hunter 01-2809 in Deer Creek, west slope of Grand Mesa, CO at 29 months old. F7 F8 �Appendix A continued Puma I.D. Estimated Age at capture (days) M44 33 Est. Birth date 8-13-06 Est. survival span from 1st capture to fate or last monitor date 09-15-06 to 02-14-07 Age to last monitor date alive or at death (days, birth to fate) 479 F45 33 8-13-06 09-15-06 to 5-20 to 23-07 280-283 M46 31 9-17-06 10-18-06 to 12-15-06 89 360 M47 M48 M49 31 31 153 9-17-06 9-17-06 7-1-06 10-18-06 to 12-15-06 to 09-12-07 10-18-06 to 12-15-06 to 09-12-07 183 7-1-06 M56c 183 ~8-13-06 F57 35 4-16-07 360 89 360 12-05-06 to 07-31-07 to 01-01-07 F53 89 01-12-07 to 02-23-07 02-14-07 to 03-01-07 05-21-07 to 06-06-07 ~456 42 ~428 subad. 200 52 166 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. Lost contact− shed radiocollar by 10-27-06. Treed, visually observed 02-14-07; sibling (?) M56 also captured, sampled, & marked for 1st time. Killed by Wildlife Services for depredation control on 12-05-07, for killing 4 domestic sheep. Dead. Multiple puncture wounds on braincase― parietal & occipital regions; consistent with bites from coyote. F45 switched families, moving from F7 to F2 about 12-19 to 2006. Last date F45 was with F2 was 04-17-07. Lost contact― shed radiocollar. Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-1207 in Puma Canyon. Lost contact― shed radiocollar. Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-1207 in Puma Canyon. Lost contact― shed radiocollar. Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-1207 in Puma Canyon. M49 was orphaned when his mother died on about 03-2607; he was ~268 days old. M49 dispersed from natal area and onto NE slope of U.P. Shed radiocollar at a yearling cow elk kill about 10-01-07; he was ~428 days old. Killed by a puma hunter in Blue Creek, northwest Uncompahgre Plateau 01-24-09 when ~29 months old. Lost contact― shed radiocollar 2-23-07. F53 visually observed by P. & F. Star, on 9-2-07, when F53 was ~14 months old and an independent subadult. F7 Lost contact― shed radiocollar 2-27-07. M56 observed 0301-07. Lost contact― shed radiocollar 06-07-07. Live mode 06-0607. F7 (?) F7 F3 F3 F3 F50 F54 F25 �Appendix A continued Puma I.D. Estimated Age at capture (days) M58 34 Est. Birth date 5-24-07 Est. survival span from 1st capture to fate or last monitor date 06-27-07 Age to last monitor date alive or at death (days, birth to fate) 324 F59 34 5-24-07 06-27-07 to 08-21-07 434 55 324 M60 34 5-24-07 F61 34 5-24-07 06-27-07 to 07-11 to 12-07 06-27-07 to 06-29-07 434 48-49 324 434 538 M62 M63 M64 34 34 34 7-14-07 7-14-07 7-14-07 08-17-07 08-17-07 08-17-07 262 M65 34 7-14-07 08-17-07 262 167 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. Not radio-collared. Tracks of 3 cubs observed with F16’s tracks on 04-12-08, McKenzie Butte-Pinon Ridge Pass. 3 cubs observed with F16 on 08-08-08 by B. & T. Traegde. F16 Alive. Observed alive 11-20-07 with F16, but without siblings M58 & F61. Tracks of 3 cubs observed with F16’s tracks on 04-12-08, McKenzie Butte-Pinon Ridge Pass. 3 cubs observed with F16 on 08-08-08 by B. & T. Traegde. F16 Dead; research-related mortality.d F16 Radiocollar malfunction. Tracks of 3 cubs observed with F16’s tracks on 04-12-08, McKenzie Butte-Pinon Ridge Pass. 3 cubs observed with F16 on 08-08-08 by B. & T. Traegde. Dead. Died probably as independent subadult at 538 days old; struck by car on Hwy 550 mi. marker 111 N. of Ridgway, CO, euthanized by gunshot on 11/13/08. Not radio-collared. Not radio-collared. Not radio-collared. Two out of potential of 4 of F24’s male cubs were visually observed with her on 4/1/08. Assume that 2 male cubs died before the age of 8.5 mo. Eartags were seen on both cubs, but the numbers were not. Not radio-collared. Two out of potential of 4 of F24’s male cubs were visually observed with her on 4/1/08. Assume that 2 male cubs died before the age of 8.5 mo. Eartags were seen on both cubs, but the numbers were not. F16 F24 F24 F24 F24 �Appendix A continued Puma I.D. Estimated Age at capture (days) F66 37 Est. Birth date 7-17-07 Est. survival span from 1st capture to fate or last monitor date 08-23-07 to 11-05-07 Age to last monitor date alive or at death (days, birth to fate) Radio-collared. Lost contact; last location 11/5/07. No signals after that date. F66 was photographed with one male sibling, either M67 or M68, & F30 on 5/31-6/1/08. F66 was recaptured and radio-collared as a subadult on 11/25/08. She died from massive trauma & bleeding of internal organs possibly resulting from being trampled by an elk or mule deer on about 05-28-09 as an independent subadult 23 months old. Not radio-collared. M67 or M68 was photographed with sibling F66 & mother F30 on 5/31-6/1/08. Not radio-collared. M67 or M68 was photographed with sibling F66 & mother F30 on 05-31 to 06-01-08. Killed by a puma hunter in Disappointment Valley, CO 12-30-08 at 17 months old. Radio-collared. Shed radiocollar between 7-9-08 and 7-1508, probably while still dependent on mother F75. Not radio-collared. Probably dead; if not killed when sibling M79 was killed, then probably would starve to death. Not radio-collared. Probably dead; if not killed when sibling M79 was killed, then probably would starve to death. Not radio-collared. Probably dead; if not killed when sibling M79 was killed, then probably would starve to death. Not radio-collared. Dead. Chewed-off anterior portions of the nasals, maxilla, palate, dentaries, and pieces of the braincase, with 6 or 9 portion of yellow ear-tag and intestines and bits of skin found ~45 m from mother F2’s death site on 8/14/08. Cub death probably due to puma-caused infanticide with cannibalism at ~87 days old. Male puma scrapes, about 8, under a rock rim ~50m distance from cub remains, and made ~ time of pumas’ deaths. Not radio-collared. Apparently died before 2-4-09; no tracks found in association with F23 & siblings F81 & F97. 111 M67 37 7-17-07 08-23-07 M68 37 7-17-07 08-23-07 F74 259 6-1-07 M76 30 5-19-08 03-12-08 to 07-09-08 06-18-08 ~87 M77 30 5-19-08 06-18-08 ~87 F78 30 5-19-08 06-18-08 ~87 M79 30 5-19-08 06-18-08 87 F80 40 5-23-08 07-02-08 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 403 168 Mother I.D. F30 F30 F30 F75 F2 F2 F2 F2 F23 �Appendix A continued Puma I.D. Estimated Age at capture (days) F81 40 F97 8 ½ mo. 5-23-08 5-23-08 M82 37 5-29-08 M83 37 M84 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 07-02-08 to 07-29-09 02-04-09 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 424 354 F23 F23 295-308 5-29-08 07-05-08 to 03-20-09 or 04-02-09 07-05-08 Radio-collared. Last live location 7-29-09. Radio-collared. Lost contact after 05-12-09; shed collar at elk kill cache on Mailbox Park. Radio-collared. F8 36 6-5-08 07-11-08 to 02-11-09 251 Not radio-collared. Apparently died; no tracks found in association with F8 & sibling M82 2-10-09. Radio-collared 7-11-08 to 7-22-08; collar removed because of malfunction. Not radio-collared after 7-22-08. Eartag of M84 was found by E. Phillips on 8-25-08 when mother F70’s GPS locations located here on either side of the eartag in the East fork Dolores Cyn. M84 recaptured radiocollared again 1-29-09 in Dolores Cyn. in association with F70 & F96’s family. Shed radiocollar again about 211-09. F85 36 6-5-08 07-11-08 F70 F86 36 6-5-08 07-11-08 to 07-23 to 08-03-08 M87 M88 F89 M90 Male 7A 28 28 28 36 28-35 7-3-08 7-3-08 7-3-08 7-9-08 7-10-08 07-31-08 07-31-08 07-31-08 08-14-08 ~08-07-08 to 08-14-08 Male 7B 28-35 7-10-08 ~08-07-08 to 08-14-08 28 to 35 Female 7C 28-35 7-10-08 ~08-07-08 to 08-14-08 28 to 35 Radio-collared. Dead. Probably died of predation or infanticide about 10-108 near elk calf kill. Radio-collared 7-22-08. Dead. Radio-collar, orange ear-tag #86 with pinna with green tattoo #86 found by J. Timmer 9-1-08. F86 died ~7-23 to 8-3-08 when mother F70’s GPS locations located her at F86 remains. Probable predation. Not radio-collared. Not radio-collared. Radio-collared Radio-collared Not radio-collared. F7’s cubs died from starvation after they were orphaned. F7 was shot on 8-3-08 for depredating on domestic sheep. Not radio-collared. F7’s cubs died from starvation after they were orphaned. F7 was shot on 8-3-08 for depredating on domestic sheep. Not radio-collared. F7’s cubs died from starvation after they were orphaned. F7 was shot on 8-3-08 for depredating on domestic sheep. ~48-59 28 to 35 169 Mother I.D. F8 F70 F70 F3 F3 F3 F72 F7 F7 F7 �Appendix A continued Puma I.D. Estimated Age at capture (days) M91 35 M92 35 F95 16 mo. F98 4-5 mo. Est. Birth date Est. survival span from 1st capture to fate or last monitor date 09-29-08 09-29-08 12-29-08 2-12-09 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. 8-19-08 Radio-collared. F25 8-19-08 Radio-collared. F25 June-07 Radio-collared. Survived to subadult stage. F93 Sep-Oct23-24 Radio-collared. Died, probably killed by male puma Unm.F 08 (infanticide). M99 5 mo. Sep-Oct2-27-09 Radio-collared. Last location 4-22-09 on Paterson Mt. Unm.F 08 M101 35 4-15-09 05-20-09 Radio-collared. F16 M102 35 4-15-09 05-20-09 Radio-collared. F16 F103 35 4-15-09 05-20-09 Radio-collared. F16 M105 38 5-7-09 06-14-09 Radio-collared F75 F106 38 5-7-09 06-14-09 Not radio-collared; F75 returned to nursery during handling. F75 M107 34 5-25-09 06-28-09 Not radio-collared; too small. F94 F108 34 5-25-09 06-28-09 Shed radiocollar; fastener failed. F94 M109 34 5-25-09 06-28-09 Not radio-collared; too small. F94 a Cub F35 probably starved between 06-30-06 & 07-07-06 after the transmitter on the expandable collar got in its mouth. b Cub M42 died after being captured by dogs, probably from stress of capture associated with severe infection of laceration under right foreleg caused by expandable radiocollar. c Cub M56 was captured in association with F7 and her cubs M43 and M44. He may have been missed at the nursery when M43 and M44 were initially sampled and marked. d Cub M60 died probably of starvation. The expandable radiocollar was around the neck and right shoulder, possibly restricting movement. 170 �APPENDIX B Puma Population Models and Simulations. Research on the Uncompahgre Plateau Puma Project from December 2004 to July 2009 provides estimates of puma population structure and parameters for a model-based approach developed by CDOW biometrician P. Lukacs and Mammals Researcher K. Logan to examine options for the design of the remainder of this research, and as a preliminary assessment of the CDOW puma management assumptions. Puma Population Modeling Our puma population projections for the study area involved an age-structured, deterministic, discrete time model. The additive puma population model structure is: Nt+1 = Adult Females = (SAF * NAFt + SSF * NSFt) * (1 – HAFt+1) + Adult Males = (SAM * NAMt + SSM * NSMt) * (1 – HAMt+1) + Subadult Females = ((r * SC * NCt) * (1 – HSFt+1)) * PISF/ESF + Subadult Males = (((1 − r) * SC * NCt) * (1 – HSMt+1)) * PISM/ESM + Cubs = Lỹ * AFR * NAFt+1 Terms: NAFt+1 = Number of adult females at year t+1. NAMt+1 = Number of adult males at year t+1. NSFt+1 = Number of subadult females at year t+1. NSMt+1 = Number of subadult males at year t+1. NJt+1 = Number of juveniles at year t+1. S = Survival rate for each specified sex and age stage. H = Proportion of the harvest rate comprised by each sex and age stage (e.g., 0.28 harvest rate * 0.40 adult females). r = Proportion of the subadult population that is female (e.g., 0.5; 1-0.5 = proportion of males). PI/E = Ratio of progeny + immigrants/emigrants. Lỹ = Average litter size. AFR = Proportion of adult females giving birth to new litters each year. Basic assumptions of the model include: 1) expected puma population projections and annual rates of increase (i.e., lambda) are conditional on the assigned puma population structure and demographic estimates, 2) no density dependent responses are built into the model. Density dependence might operate in puma population dynamics, with competition for food regulating adult female density and competition for mates regulating adult male density (Logan and Sweanor 2001), and 3) harvest is additive mortality. We parameterized the model with data gathered on the pumas on the study area during the first 3.7 years. (Data from this past year, 2008-09 could not be used because decisions about harvest structure for the treatment period needed to be made June of that biological year). The starting population was the minimum count of pumas and attendant estimated sex and age structure made during November 2007 to March 2008 (Table B.1). We assumed that all individuals were present in the population during that entire period. No mortalities of independent pumas were detected. But, one radio-collared subadult male emigrated by March 19, 2008. Population parameters included: estimated rates of reproduction and sex and age-stage specific survival, which included data to July 2008 (Table B.2). Some sex and age-stage specific estimates of survival (i.e., adult male, subadult male, subadult female) came from the literature 171 �(Table B.2), because our current sample sizes (i.e., number of individuals and years) may not be adequate for realistic estimates (i.e., adult males and subadults). We did not use actual rates in the literature where estimates involved the pooling of data on sexes and age stages, and where sample sizes for age stages were not presented (e.g., Anderson et al. 1992). In addition, the ratio of progeny and immigrant recruits to emigrants as a model input was from the literature, because such data is scarce and does not exist for Colorado (all references in Table B.2). We preferred using the population characteristics and parameter estimates gathered in the current research effort, because this is the puma population we intend to manipulate to assess current CDOW puma management strategies. Table B.1. Minimum puma population count on Uncompahgre Plateau study area, Colorado, November 2007 to March 2008 (RY4). The minimum count involves counting all radio- and GPS-collared pumas, all other marked pumas, and all presumably unmarked pumas detected on the study area during the period. Presumed unmarked pumas could be marked with ear-tags and tattoos. Their tracks and movements could be separated from movements of radio- and GPS-collared pumas. Or they exhibited evidence that could separate them from other local marked pumas from their tracks (i.e., distinguishable by sex, number of cubs and/or relative size of cubs varied). Region East slope West slope Totals Adults Subadults Female Male Female Male 10 4 3 4 6 4 2 0 16 8 5 4 Total Independent Pumas = 33a,b Female 4 1 5 Cubs Male 4 2 6 Unknown sex 7 2-3 20-21 Of the total, 23−24 (70−73%) independent pumas were marked and 9-10 (27−30%) were assumed to be unmarked. a 172 �Table B.2. Summary of preliminary puma population model parameter estimates obtained from the Uncompahgre Plateau Puma Project and from the literature on puma. Survival Sex and age stage Adult Female Estimate 0.87 Adult Male 0.91 Subadult Female 0.80 Subadult Male 0.60 Cub 0.50 0.90 Reference Estimated average annual survival rate (n = 2 years) for 11−13 adult females on Uncompahgre Plateau study area. Estimated average annual survival rate (n = 8 years) for adult males in a nonhunted New Mexico puma population (Logan and Sweanor 2001:127-128). Estimated annual survival rate (n = 2 years) for 5−9 adult males on Uncompahgre Plateau study area was 1.00. Estimated subadult female survival in New Mexico (0.88, n = 16; Logan and Sweanor 2001:122) adjusted downward for potential lower survival for pumas 12-24 months old on Uncompahgre Plateau (0.642, n = 14 females and 10 males combined, life stages not known or described in Anderson et al. 1992:53). Survival of 7 radio-collared pumas (5 males, 2 females) in the subadult stage in the current Uncompahgre Plateau puma study is 1.00. Estimated subadult male survival in New Mexico (i.e., 0.56, n = 9; Logan and Sweanor 2001:122) adjusted upward for potential slightly higher survival for pumas of both sexes 12-24 months old (i.e., 0.642) on Uncompahgre Plateau (Anderson et al. 1992:53). Survival of 7 radiocollared pumas (5 males, 2 females) in the subadult stage in the current Uncompahgre Plateau puma study is 1.00. Estimated cub survival rate (n = 38 cubs combined sexes), on Uncompahgre Plateau study area. This survival rate is applied to the model starting with the expected number of cubs from birth in RY5. Estimated cub survival for cubs ≥7 months old, and is applied to RY4 cubs only, because the minimum count of pumas in RY4 was tallied when most cub mortality had already occurred. Survival of cubs ≥7 months old in the literature is about 0.95 (Logan and Sweanor 2001). Here, a more conservative 0.90 is used in this model. Reproduction Parameter Adult age Estimate 2+ years Litter size 2.81 Secondary sex ratio observed at nurseries 1:1 Proportion of adult females producing new litters each year 0.65 Parameter Subadult female Estimated Ratio 1.02 Subadult male 0.94 Reference Assume all females 2 years old and older are adults (Logan and Sweanor 2001: 93-94). Average litter size for 21 litters on the Uncompahgre Plateau study area = 2.810 ± 0.9808SD; litters were examined when the cubs were 26 to 42 days old. Secondary sex ratio was 33:26 for 21 litters examined at 29 to 42 days old on the Uncompahgre Plateau study area (not significantly different from 1:1, (X2 = 0.8305 < 3.841, α = 0.05, 1 d.f.). This result supported Logan and Sweanor 2001:69, n = 148). Proportion of adult females giving birth each year (n = 3 years for n = 12, 13, 12 females), Uncompahgre Plateau study area. Proportion for a non-hunted puma population in New Mexico was 0.50 (Logan and Sweanor 2001:98). Progeny + Immigrant Recruits/Emigration Ratio Reference No data for pumas in Colorado exists. Assume the ratio of female immigrants to emigrants = 1.02. This ratio is consistent with estimates for a New Mexico puma population that functioned as a source (Sweanor et al. 2000). No data for pumas in Colorado exists. Assume the ratio of male immigrants to emigrants = 0.94, (i.e., male immigration is half of emigration). This ratio is consistent with estimates for a New Mexico puma population that functioned as a source (Sweanor et al. 2000). 173 �Puma Population Simulations We used this model to simulate puma population dynamics to examine a set of scenarios that pertain to current CDOW puma management assumptions and to the puma research and management direction on the Uncompahgre Plateau for the treatment period: 1) Puma population dynamics without hunting-caused mortality. 2) Puma harvest that would induce a stable (i.e., no growth) phase to identify a population tipping point induced by harvest mortality, expected to be 16% harvest of independent pumas. Various sex ratios of harvest composition were examined. 3) Puma harvest at the upper limit (i.e., 15% of 8-15% range, CDOW 2007) that CDOW assumes would result in a stable to increasing puma population. Various sex ratios of harvest composition were examined. 4) Puma harvest at the upper limit (i.e., 28% of >15-28% range, CDOW 2007) that CDOW assumes would result in a declining puma population. Various sex ratios of harvest composition were examined. 5) Puma harvest at a 20% harvest level intermediate to the 16% stable growth and 28% decline phase with varying female to male sex structure of the harvest. 6) Puma harvest at the historic harvest level of 26% and sex ratio of 45 females:55 males on the study area during 1994-2003. Results of Puma Population Simulations The following tables contain the expected minimum population sizes for independent pumas and annual rates of population increase conditional upon the minimum number of independent pumas detected in Reference Year 4 (RY4) and the model input parameters and assumptions (given in Tables B.1 and B.2). The total number of independent pumas is probably higher in any particular scenario because we probably did not detect all of the independent pumas in RY4. Simulations involving harvest apply the harvest following reference year 5 (RY5) and starting with treatment year 1 (TY1) to assess what might be expected to occur within the current research structure on the Uncompahgre Plateau. Our puma population simulation modeling suggest strategies to achieve increasing and declining puma populations contingent upon the set of assumptions and input demographic data. Moreover, results of this modeling effort constitute the first time that CDOW puma harvest assumptions have been evaluated by using Colorado-specific population data. Results could change as more quantitative population data are gathered and the puma population is manipulated during this research. Expected estimates of population growth were generally consistent with the current CDOW puma harvest management assumptions that were previously developed from data in the puma population literature to manage for a stable-to-increasing population, and for a declining puma population. The following series of tables (B.3 – B.16) indicate results of the individual models, followed by notes on how results may be interpreted relative to other research results on puma population dynamics and specific CDOW puma management assumptions. The harvest levels for each model are clearly stated in the left column of each table. 174 �Table B.3. Harvest Level 16% of independent pumas, sexes are harvested equally; i.e., stable phase model. Projected Minimum Puma Population Size Adult Subadult Female Male Female Male 16 8 5 4 18 10 9 8 19 12 7 6 19 12 8 7 19 13 7 7 19 13 7 7 19 14 7 7 Year RY4 RY5 TY1 TY2 TY3 TY4 TY5 Independent Pumas Cub 20 33 35 34 34 34 34 Total 33 45 44 45 46 46 46 Lambda* 1.37 0.98 1.02 1.01 1.01 1.00 Note: The tipping point of population stability and decline is expected to be about 16% harvest of independent male and female pumas, consistent with current CDOW puma harvest assumptions. *Lambda is the finite rate of population growth (Williams et al. 2002:136): λ = 1 + (N t+1 – N t) / N t Expected Minimum No. Pumas 50 "'n:, 45 E 40 ::I c.. 35 C: 30 (IJ "C 25 C: (IJ C. 20 (IJ "C 15 -=ci 10 z 5 0 ... .,,, - - - 11;:: /' .,.. 33 RY4 RYS TYl TY2 TY3 TY4 TYS Year Figure B.1. Expected minimum number of independent pumas based on population simulations with 16% harvest of independent pumas comprised of 50% males and 50% females in the harvest in TY1 to TY5. Table B.4. Harvest Level 16% of independent pumas, harvest comprised of 40% females:60% males. Year RY4 RY5 TY1 TY2 TY3 TY4 TY5 Projected Minimum Puma Population Size Adult Subadult Female Male Female Male 16 8 5 4 18 10 9 8 20 11 7 6 21 10 9 7 23 10 9 7 24 10 9 7 25 10 10 8 Note: The puma population is expected to increase. 175 Independent Pumas Cub 20 33 37 39 41 44 46 Total 33 45 44 46 48 51 53 Lambda 1.37 0.98 1.05 1.04 1.05 1.05 �Table B.5. Harvest Level 16% of independent pumas, harvest comprised of 45% females:55% males. Year RY4 RY5 TY1 TY2 TY3 TY4 TY5 Projected Minimum Puma Population Size Adult Subadult Female Male Female Male 16 8 5 4 18 10 9 8 20 11 7 6 20 11 8 7 21 11 8 7 21 12 8 7 22 12 9 7 Independent Pumas Cub 20 33 36 37 38 39 40 Total 33 45 45 46 47 49 50 Lambda 1.37 0.98 1.04 1.03 1.03 1.03 Note: The puma population is expected to increase. Table B.6. Harvest Level 16% of independent pumas, harvest comprised of 55% females:45% males. Year RY4 RY5 TY1 TY2 TY3 TY4 TY5 Projected Minimum Puma Population Size Adult Subadult Female Male Female Male 16 8 5 4 18 10 9 8 18 12 7 7 17 13 7 7 17 14 6 6 16 14 6 6 15 15 6 6 Independent Pumas Cub 20 33 34 31 30 29 27 Total 33 45 44 44 43 42 41 Lambda 1.37 0.97 1.00 0.98 0.98 0.97 Note: The puma population is expected to decline slowly. Expected No. Independent Pumas at 16% harvest with varying female:male ratio 60 V, Ill - so E ::i c... .... 40 ~ QJ "ti ~ 30 / / F:M ratlo - 40:60 - 45 :55 50:50 QJ C. - QJ "ti -=ci z 20 55:45 10 0 RY4 RYS TY1 TY2 Ye ars TY3 TY4 TYS Figure B.2. Expected minimum number of independent pumas based on population simulations with 16% harvest of independent pumas comprised of varying female to male sex ratios in the harvest in TY1 to TY5. See tables B.3-6 (above) for quantities of results for each model. In reality, the ratio of females to males in the harvest may vary randomly on an annual basis, and the expected annual numbers of independent pumas may fall within the lower and upper population trend lines. 176 �Table B.7. Harvest Level No harvest. Year RY4 RY5 TY1 TY2 TY3 TY4 TY5 Projected Minimum Puma Population Size Adult Subadult Female Male Female Male 16 8 5 4 18 10 9 8 23 14 8 8 27 17 11 10 32 22 12 11 38 27 15 14 44 32 17 16 Independent Pumas Cub 20 33 42 49 58 69 81 Total 33 45 53 64 77 92 110 Lambda 1.37 1.17 1.22 1.20 1.20 1.19 Note: Expected lambda for the modeled non-hunted puma population on the Uncompahgre Plateau are consistent with the high range of observed average annual rates of population increase for a non-hunted puma population in good quality habitat in southern New Mexico (i.e., r = 0.21, n = 4 yr.; r = 0.28, n = 4 yr.; r = 0.17, n = 4 yr.; r = 0.11, n = 7 yr.; Logan and Sweanor 2001:169-175). Puma population growth could be higher on the Uncompahgre Plateau because of higher quality habitat (i.e., greater vulnerable prey biomass), and if puma sources are nearby to the study area. Table B.8. Harvest Level 15% of independent pumas, sexes are harvested equally. Year RY4 RY5 TY1 TY2 TY3 TY4 TY5 Projected Minimum Puma Population Size Adult Subadult Female Male Female Male 16 8 5 4 18 10 9 8 19 12 7 7 19 12 8 7 19 13 8 7 20 14 8 7 20 14 8 7 Independent Pumas Cub 20 33 36 35 36 36 36 Total 33 45 45 47 47 48 49 Lambda 1.37 0.99 1.04 1.02 1.02 1.01 Note: This result is consistent with current the CDOW puma harvest assumption for a stable-to-increasing population, with slow growth attributed to equal harvest of females and males. Table B.9. Harvest Level 15% of independent pumas, comprised of 40% females & 60% males. Year RY4 RY5 TY1 TY2 TY3 TY4 TY5 Projected Minimum Puma Population Size Adult Subadult Female Male Female Male 16 8 5 4 18 10 9 8 21 11 8 6 22 10 9 7 23 10 9 7 25 11 10 8 26 11 10 8 Independent Pumas Cub 20 33 38 39 42 45 48 Total 33 45 45 47 50 53 56 Lambda 1.37 0.99 1.06 1.05 1.06 1.06 Note: This result is consistent with the current CDOW puma harvest assumption for a stable-to-increasing population, with increased growth due to reduced female mortality. 177 �Table B.10. Puma population simulation results, based on the minimum number of detected independent pumas in RY4, and harvest rate of 20% of independent pumas comprised of 50% females and 50% males applied to independent pumas as a treatment during TY1-TY5. Harvest Level 20% of independent pumas, comprised of 50% females & 50% males. Year RY4 RY5 TY1 TY2 TY3 TY4 TY5 Projected Minimum Puma Population Size Adult Subadult Female Male Female Male 16 8 5 4 18 10 9 8 18 11 7 6 17 11 7 6 16 11 6 6 15 11 6 6 15 11 6 5 Independent Pumas Cub 20 33 34 31 30 28 27 Total 33 45 42 41 40 38 36 Lambda* 1.37 0.93 0.97 0.96 0.96 0.96 Note: The puma population would be expected to decline. Table B.11. Puma population simulation results, based on the minimum number of detected independent pumas in RY4, and harvest rate of 20% of independent pumas comprised of 40% females and 60% males applied to independent pumas as a treatment during TY1-TY5. Harvest Level 20% of independent pumas, comprised of 40% females & 60% males. Year RY4 RY5 TY1 TY2 TY3 TY4 TY5 Projected Minimum Puma Population Size Adult Subadult Female Male Female Male 16 8 5 4 18 10 9 8 20 10 7 5 20 8 8 6 21 8 8 6 21 8 8 6 22 7 9 6 Independent Pumas Cub 20 33 36 37 38 39 40 Total 33 45 42 42 43 43 44 Lambda 1.37 0.93 1.01 1.00 1.01 1.02 Note: The puma population would be expected to increase slowly. Table B.12. Puma population simulation results, based on the minimum number of detected independent pumas in RY4, and harvest rate of 20% of independent pumas comprised of 45% females and 55% males applied to independent pumas as a treatment during TY1-TY5. Harvest Level 20% of independent pumas, comprised of 45% females & 55% males. Year RY4 RY5 TY1 TY2 TY3 TY4 TY5 Projected Minimum Puma Population Size Adult Subadult Female Male Female Male 16 8 5 4 18 10 9 8 19 10 7 6 19 10 7 6 19 10 7 6 18 9 7 6 18 9 7 6 Independent Pumas Cub 20 33 35 34 34 34 33 Total 33 45 42 42 41 41 40 Lambda 1.37 0.94 0.99 0.98 0.99 0.99 Note: The puma population would be expected to decline slowly. The ratio of 45% females and 55% males in the harvest is the average harvest sex ratio during 1994-2003. 178 �Table B.13. Puma population simulation results, based on the minimum number of detected independent pumas in RY4, and harvest rate of 20% of independent pumas comprised of 55% females and 45% males applied to independent pumas as a treatment during TY1-TY5. Harvest Level 20% of independent pumas, comprised of 55% females & 45% males. Projected Minimum Puma Population Size Adult Subadult Female Male Female Male 16 8 5 4 18 10 9 8 17 12 6 6 15 12 6 6 14 12 5 5 12 12 5 5 11 12 4 4 Year RY4 RY5 TY1 TY2 TY3 TY4 TY5 Independent Pumas Cub 20 33 32 28 25 22 20 Total 33 45 42 40 37 34 31 Lambda 1.37 0.94 0.99 0.98 0.99 0.99 Note: The puma population would be expected to decline more rapidly. Expected No. Independent Pumas at 20% harvest with varying female:male ratio so 45 "'ro 40 ::I c.. 35 - 40:60 C: 30 - 45:55 "C 25 E ..... QJ C: 50:50 QJ Q. QJ "C -=ci z 20 - 15 55:45 10 5 0 RY4 RYS TYl TY2 TY3 TY4 TVS Years Figure B.3. A harvest level of 20% of independent pumas is expected to result in a declining population, except in the scenario consistently weighted heavily toward male harvest (i.e., 60%). Table B.14. Harvest Level 28% of independent pumas, sexes are harvested equally. Year RY4 RY5 TY1 TY2 TY3 TY4 TY5 Projected Minimum Puma Population Size Adult Subadult Female Male Female Male 16 8 5 4 18 10 9 8 17 10 6 6 14 9 6 5 12 8 5 4 10 7 4 4 9 6 3 3 Independent Pumas Cub 20 33 30 25 22 18 16 Total 33 45 38 33 29 25 21 Lambda 1.37 0.84 0.88 0.86 0.86 0.86 Note: This result is consistent with the current CDOW puma harvest assumption for a declining population. 179 �Table B.15. Harvest Level 28% of independent pumas, comprised of 40% females & 60% males. Year RY4 RY5 TY1 TY2 TY3 TY4 TY5 Projected Minimum Puma Population Size Adult Subadult Female Male Female Male 16 8 5 4 18 10 9 8 19 8 7 4 18 6 7 5 17 5 7 4 16 4 6 4 16 4 6 4 Independent Pumas Cub 20 33 34 32 31 30 29 Total 33 45 38 35 33 31 30 Lambda 1.37 0.84 0.93 0.93 0.95 0.95 Note: This result is consistent with the current CDOW puma harvest assumption for a declining population even with harvest weighted toward males. Yet another harvest scenario to consider for the treatment period is application of the historic puma harvest on the study area. Puma mortality data for the study area during the 10 years previous 19942003 prior to the beginning of the study reference period was tabulated after carefully geo-referencing mortality locations on the study area (Logan 2008). Model parameters from those data include: mortality rate of 14.3 independent puma mortalities per year (rounded to 14/yr.), and sex proportions of 55% males and 45% females. No other puma population data or parameter estimates were available for the study area at that time. Therefore, the scenario that was modeled pertained to the expected impact of the average annual puma mortality of independent pumas (i.e., adults and subadults) if the hypothetical population was the same as the minimum expected puma population after year 5 of the reference period (i.e., RY5). A harvest of 14 pumas/yr. is a 26% harvest rate of the expected minimum independent puma population at the start of TY1. Table B.16. Harvest Level 26% of independent pumas at start of TY1, comprised of 45% females & 55% males. Year RY4 RY5 TY1 TY2 TY3 TY4 TY5 Projected Minimum Puma Population Size Adult Subadult Female Male Female Male 16 8 5 4 18 10 9 8 18 9 7 5 17 8 7 5 15 7 6 5 14 6 6 5 13 6 5 4 Independent Pumas Cub 20 33 33 30 28 26 25 Total 33 45 39 36 34 31 29 Lambda 1.27 0.87 0.93 0.92 0.93 0.93 Note: As expected, results of this model indicate puma population decline. This simulation demonstrates the negative cost of uncertainty in puma management; in this case a puma population would decline where the intended management objective was for a stable-to-increasing population. 180 �Expected No. Independent Pumas 50 45 V, re 40 E 35 ::J c.. .... C: 30 29 Q) "C C: 25 Q) Q. Q) "C 20 -=ci 15 z 10 5 0 RY4 RYS TYl TY2 TY3 TY4 TVS Years Figure B.4. Expected dynamics of a puma population with the historical harvest (1994-2003) rate on the Uncompahgre Plateau study area of 26% of the independent puma and sex ratio of 45% females to 55% males (see Logan 2008 for historical harvest data on the study area). 181 �APPENDIX C Collaborative project on disease surveillance in wild felids with College of Veterinary Medicine and Biomedical Sciences, Department of Microbiology, Pathology, and Immunology, Colorado State University. College of Veterinary Medicine and Biomedical Sciences Department of Microbiology, Immunology & Pathology 1619 Campus Delivery Fort Collins, CO 80523-1619 970-491-6144 (voice) 970-491-0603 (fax) TO: Ken Logan, Mammals Researcher, Colorado Division of Wildlife, Montrose, CO. FROM: Sue VandeWoude, DVM, Associate Professor, DMIP RE: Disease Seroprevalence in UP Pumas DATE: August 26, 2007 These specific agents were selected for analysis in order to provide a variety of types of agents (viruses: PLV, FCV, FHV, FPV; bacteria: Bartonella henselae and Yersinia pestis; and coccidian: T. gondii), a variety of modes of transmission (direct intra-specific contact, PLV; direct contact with domestic cats, FCV, FHV, FPV; arthropod transmission, B. henselae, Y. pestis; prey ingestion, T. gondii, Y. pestis). Further, at least three of these agents (PLV, FCV, B. henselae) result in chronic infections, allowing the possibility of determining genetic relatedness among organisms isolated from different individuals, and three of these agents (B. henselae, Y. pestis, T. gondii) are also potential zoonotic agents. As you are aware, our laboratory has recently been awarded a 5 year NSF Ecology of Infectious Disease grant entitled, “The effects of urban fragmentation and landscape connectivity on disease prevalence and transmission in North American felids”, with co-PI Dr. Kevin Crooks, an associate professor in the Warner College of Natural Resources at CSU. The aims of this grant are to model the effects of urbanization and resultant habitat fragmentation on disease dynamics in large carnivore species as described on the following page. The letter of support provided by you and Mr. Dave Freddy were pivotal in demonstrating a large cohort of capable and active field collaborators willing to provide samples to support our studies. The mountain lion field work being led by your team, and the newly initiated studies by your colleague, Dr. Mat Alldredge, have provided us with renewed enthusiasm for developing our collaborations to support the goals of our study. We foresee the opportunity to interact in a mutually beneficial partnership to further the goals of all of our studies, and to maximize the information that can be gleaned about these important and ecologically significant species. We anticipate that the data we are generating will be useful for comparative seroprevalence of different geographic populations of bobcats and pumas, and for genetic phenotyping of pathogens to compare relationships among diseases spread by arthropod vectors, domestic cats, feral rodents, and interspecific contacts. As we discussed during your recent visit to CSU, these samples are most valuable to us if we can receive them directly as quickly as possible after collection. I have provided an SOP providing information about the types of samples that will be most valuable, and a draft of a ‘permissions’ document that you can use with each sample submission to provide us with guidance for any testing that is permissible on the materials we receive. This latter document will be filed and recorded electronically. We will continue to provide annual updates and communications about any publications that utilize the data resulting from your samples. Again thank you for providing these extremely valuable samples, and we look forward to our continued collaborations. Sincerely, Sue VandeWoude 182 �The effects of urban fragmentation and landscape connectivity on disease prevalence and transmission in North American felids Project Summary Sue VandeWoude (co-PI), Kevin Crooks (co-PI), Michael Lappin, Mo Salman, Walter Boyce, Ken Logan, Mat Alldredge, Carolyn Krumm, Don Hunter, Lisa Lyren, Seth Riley, Jennifer Troyer The objective of this study is to model the effects of urbanization and resultant habitat fragmentation on disease dynamics in carnivore species. Bobcats, puma, and domestic cats will be evaluated simultaneously in three divergent ecosystems: high mountain desert (Colorado), everglades (Florida), and Mediterranean scrub habitat (California). The research will: 1) assess the relationship between habitat fragmentation and prevalence of viral, bacterial, and parasitic pathogens across a gradient of urbanization, 2) use transmission dynamics of selected disease agents as markers of connectivity of fragmented populations, and 3) evaluate the effect of urbanization on the incidence of cross-species disease transmission. The results of this research will give wildlife managers a better understanding of how urbanization affects their local wildlife and assist them in future disease management planning. The combination of a uniquely qualified, broadly based research team with an extensive dataset on carnivores from across the country presents an unprecedented opportunity to investigate the disease dynamics in these rare and difficult to study species. The research efforts of each regional team will support and provide new insights for all of the regions involved, not simply their own. Training of graduate students in ecology, infectious disease, and epidemiology will be emphasized, as will training for pre- and post-doctoral veterinarians. Results will be made widely available to other scientists, conservation practitioners, and the general public. This research has a tremendous capacity to broadly impact areas of public and postgraduate education, career development for new investigators and persons from underrepresented groups, and to enhance understanding of complex infectious disease ecological problems using extensive multidisciplinary collaborations. 183 �Appendix C (continued). Preliminary results of infectious disease surveillance for puma, Uncompahgre Plateau, Colorado, 2005-2009. Puma ID UPCO2 UPCO3 UPCO7 UPCO7 UPCO7 UPCO8 UPCO4 UPCO5 UPCO6 UPCO6 UPCO23 UPCO25 UPCO28 UPCO29 UPCO31 UPCO23 UPCO27 UPCO30 UPCO50 UPCO51 UPCO52 UPCO54 UPCO55 UPCO24 UPCO69 UPCO70 UPCO71 UPCO72 UPCO73 UPCO74 UPCO75 UPCO72 Sex F F F F F F M M M M F F F M M F M F F M F F M F M F M F F F F F Capture Date 1/8/2008 1/21/2005 2/24/2005 3/30/2006 3/3/2007 3/21/2005 1/28/2005 2/4/2005 2/18/2005 4/12/2008 2/25/2008 2/8/2006 3/23/2006 4/14/2006 4/19/2006 1/4/2006 3/10/2006 4/15/2006 12/14/2006 1/7/2007 1/10/2007 1/12/2007 1/21/2007 1/17/2006 1/11/2008 1/20/2008 1/29/2008 2/12/2008 2/21/2008 3/12/2008 3/26/2008 7/20/2009 UPCO104 UPCO55 UPCOF16 UPCO66 UPCO94 UPCO96 UPCO100 UPCO82 UPCO93 UPCO71 UPCO72 UPCO73 UPCO74 F M F F F F M M F M F F F 5/21/2009 1/5/2009 1/14/2009 11/25/2008 12/19/2008 1/28/2009 3/27/2009 2/10/2009 12/15/2008 1/29/2008 2/12/2008 2/21/2008 3/12/2008 GPS NAD27 U.T.M.: Zone, E, N 13S, 245722, 4244166 13S, 241606, 4251510 13S, 246328, 4244230 13S, 245901, 4247627 13S, 247645, 4246097 12S, 727808, 4239029 13S, 257565, 4239606 13S, 240577, 4251037 13S, 247399, 4254006 13S, 257516, 4239696 12S, 723304, 4242231 13S, 258374, 4230480 12S, 722868, 4240115 12S, 723458, 4242340 12S, 746919, 4225441 12S, 730188, 4234861 12S, 722339, 4245212 13S, 248551, 4242095 12S, 753639, 4260149 13S, 238783, 4252390 13S, 258058, 4236260 13S, 252688, 4228050 13S, 258133, 4228691 12S, 737151, 4233273 13S, 248191, 4246810 13S, 247122, 4245760 12S, 754611, 4256842 13S, 258294, 4234597 12S, 728576, 4241799 12S, 729678, 4239555 12S, 732894, 4239423 13S, 255400, 4229658 12S, 745118, 4264721N 13S, 239076, 4248637 13S, 256528, 4235500 13S, 245901, 4247627 12S, 758531, 4259824 13S, 247764, 4246239 12S, 749832, 4217148 12S, 726732, 4243782 12S, 751445, 4265985 12S, 754611, 4256842 13S, 258294, 4234597 12S, 728576, 4241799 12S, 729678, 4239555 PLV + + + + I + + + + + + + + + + + + + + P P P P + P + P P P + P a a b FCV +h + + + + + + No swab - c FHV + + + + + + + No swab - FPV + + + NA NA + + + + + + + + NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA d T.g.e IgM + P - T.g.e IgG + + + + + + + + + + + + + + + + + + + + + + + + P + + B.h.f + + Y.p.g + ++ +++ ++ ++ ++ + + ++ + + + NA P + + + + + + + + + + + P + - NA NA NA NA NA NA NA NA NA + - PLV is Puma Lentivirus. FCV is Feline Calicivirus. c FHV is Feline Herpesvirus. d FPV is Feline Panleukopenia Virus e T. g. is Toxoplasma gondii. f B. h. is Bartonella hensalae. g Y. p. is Yersinia pestis. h Results: + (positive result), P (Pending result), I (Inconclusive result), NA (not applicable). b 184 �Colorado Division of Wildlife July 2009 –July 2010 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 3003 1 Federal Aid Project No. N/A : : : : : Division of Wildlife Mammals Research Carnivore Conservation Puma Population Structure and Vital Rates On the Uncompahgre Plateau Period covered: July 31, 2009−July 31, 2010 Author: K. A. Logan. Personnel: K. Logan, C. Burnett, B. Dunne, A. Greenleaf, J. Knight, R. Navarrete, J. Waddell, S. Waters, K. Crane, T. Mathieson, J. Koch, and T. Bonacquista of CDOW; S. Young and W. Wilson of U.S.D.A. Wildlife Services; volunteers and cooperators including: private landowners, Bureau of Land Management, Colorado State Parks, Colorado State University and U.S. Forest Service. Supplemental financial support received in previous years from The Howard G. Buffett Foundation and Safari Club International Foundation. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT The Colorado Division of Wildlife initiated a 10-year study on the Uncompahgre Plateau in 2004 to quantify puma population characteristics in the absence (reference period, yrs 1-5) and presence (treatment period, yrs 6-10) of hunting. The purpose of the study is to evaluate assumptions underlying the Colorado Division of Wildlife’s model-based approach to managing pumas with sport-hunting in Colorado. The reference period began December 2004 and ended July 2009, during which we captured, sampled, and marked 109 pumas for population research purposes on the Uncompahgre Plateau (Logan 2009). This report informs on the first year of the treatment period (TY1), August 2009 through July 2010, on puma population characteristics and dynamics with hunting as a mortality factor. Puma sporthunting opened November 16 and closed December 11, 2009 after a quota of 8 independent pumas was harvested. The harvest was designed to test the management assumption that a 15% harvest of independent pumas results in a stable-to-increasing population. A total of 9 pumas were killed: 2 adult females, 1 subadult female, 5 adult males, and 1 dependent cub. The harvest of 8 independent pumas represented 15% of the expected (i.e., modeled) 53 independent pumas and 14.5% of the minimum number of 55 independent pumas counted 2009-10. Independent females and males comprised 37.5% and 62.5% of the harvest, respectively. Four other radio-collared pumas, 1 adult female and 3 adult male, in the study area population were killed in GMUs adjacent to the study area. The total harvest of 12 independent pumas represented 21.8% of the minimum count of independent pumas. Eight independent pumas will be the harvest quota for the 2010-11 hunting season (TY2). Seventy-nine hunters requested mandatory permits with an attached voluntary hunter survey in TY1. Seventy-one of the hunters provided responses to written (n = 43) or telephone call follow-up contact (n = 28). An estimated 67 hunters 101 �actually hunted on the study area, of which 13% harvested pumas and 24% captured pumas (i.e., harvested plus treed and released). All hunters responded that they were selective hunters, and the capture and population data indicated that most successful hunters practiced selection. From August 2009 to July 2010 thirty-three individual pumas were captured 38 times. Two capture teams with dogs operated over 86 search days from December 2009 through April 2010 to find 266 puma tracks, pursue pumas 93 times, and capture 21 pumas 26 times. Capture efforts with cage traps resulted in the recapture of 2 adult pumas and 1 cub. Nine cubs were observed for the first time at nurseries. A total of 42 pumas were monitored by radiotelemetry. Search efforts also revealed the presence of at least 15 other independent pumas. Our minimum count of independent pumas from September 2009 to April 2010 was 55, including 31 females and 24 males. A preliminary minimum estimated density of independent pumas was 3.29/100 km2. The proportion of radio-collared adult females giving birth in the August 2009 to July 2010 biological year was 0.42 (8/19). Seven litters that could be dated to month of birth were produced in June (4), July (2), and August (1). We monitored 19 female and 8 male adult radio-collared pumas for survival and agentspecific mortality. Survival rates in TY1 with hunting were generally lower than in the reference period without hunting. Causes of mortality were vehicle strikes and hunting. In addition, all 5 adult males with malfunctional radiocollars since the beginning of this study were harvested by hunters in TY1. Two radiomonitored subadult males died apparently due to natural causes. Of 19 cubs monitored with radiotelemetry, 5 died, all associated with infanticide. A non-marked adult male was also killed by a vehicle on the boundary of the study area. Puma harvest data also provided information on dispersals of 12 male and 1 female puma initially marked on the study area. Those pumas moved from about 60 to 370 km from initial capture sites. A pilot study on detection probabilities of pumas using a camera grid for a mark-recapture design was conducted in collaboration with Colorado State University Researchers J. Lewis and K. Crooks as they studied bobcats on the east slope of our study area. Two camera grids, Area 1 and Area 2, were on the east slope of the study area. Each grid was 80 square kilometers in size and contained 20 cells which were each 4 square kilometers. Cameras operated for 108 days from August 21 to December 7, 2009. Detection probabilities for 4 adult radio-collared pumas on Area 1 and 5 adult pumas on Area 2 were 0.75 and 0.80, respectively. Those pumas were photographed a total of 51 times: 17 times in Area 1 and 34 times in Area 2. Males were detected more frequently than females. Four other marked pumas without functioning collars were also detected 7 times. Non-marked pumas were photographed 31 times, representing 2 to 4 individuals in Area 1 and 3 to 5 individuals in Area 2. The next step in this collaboration is to conduct an intensive evaluation of pilot study data to model detection probability, estimate precision, and define the survey area for a camera grid design specifically for puma. Data are continued to be gathered for other collaborative projects with Mammals Research and CSU investigators on puma behavior, social organization, population dynamics, and habitat use. 102 �WILDLIFE RESEARCH REPORT PUMA POPULATION STRUCTURE AND VITAL RATES ON THE UNCOMPAHGRE PLATEAU, COLORADO KENNETH A. LOGAN P. N. OBJECTIVE Quantify puma population sex and age structure; estimate puma population vital rates, including: reproduction rates of females, age-stage survival rates, and immigration and emigration rates; quantify agent-specific mortality rates; model puma population dynamics; develop and execute the puma harvest manipulation to begin the population-wide test of Colorado Division of Wildlife (CDOW) puma management assumptions in the first year of a five-year Treatment Period of the Uncompahgre Plateau Puma Project― all to improve the CDOW model-based approach to managing pumas in Colorado. SEGMENT OBJECTIVES 1. Execute the first year of the five-year treatment period by working with CDOW biologists and managers to manipulate the puma population with sport-hunting and to survey hunters. 2. Continue gathering data on puma population sex and age structure. 3. Continue gathering data for estimates of puma reproduction rates. 4. Continue gathering data to estimate puma sex and age-stage survival rates. 5. Continue gathering data on agent-specific mortality. 6. Collaborate with Colorado State University (CSU) researchers on a pilot project to assess puma detection probability in a camera grid design. 7. Collaborate with other researchers involved with puma biology and ecology. INTRODUCTION Colorado Division of Wildlife managers need reliable information on puma biology and ecology in Colorado to develop sound management strategies that address diverse public values and the CDOW objective of actively managing pumas while “achieving healthy, self-sustaining populations”(CDOW 2002-2007 Strategic Plan:9). Although 4 puma research efforts have been made in Colorado since the early 1970s and puma harvest data is compiled annually, reliable information on certain aspects of puma biology and ecology, and management tools that may guide managers toward effective puma management is lacking. Mammals Research staff held scoping sessions with a number of the CDOW’s wildlife managers and biologists prior to initiating the project. In addition, we consulted with other agencies, organizations, and interested publics either directly or through other CDOW employees. In general, CDOW staff in western Colorado highlighted concern about puma population dynamics, especially as they relate to their abilities to manage puma populations through regulated sport-hunting. Secondarily, they expressed interest in puma―prey interactions. Staff on the Front Range placed greater emphasis on puma―human interactions. Staff in both eastern and western Colorado cited information needs regarding effects of puma harvest, puma population monitoring methods, and identifying puma habitat and landscape linkages. Management needs identified by CDOW staff and public stakeholders form the basis of Colorado’s puma research program, with multiple lines of inquiry (i.e., projects): Improve our ability to manage puma hunting with enhanced scientific bases, strategies, and tools― ● Puma population characteristics (i.e., density, sex and age structure). 103 �● Puma population dynamics and vital rates (i.e., birth rates, survival rates, emigration rates, immigration rates, population growth rates). ● Field methods and models for assessing and tracking changes in puma populations. ● Relative vulnerability of puma sex and age classes to hunter harvest. Improve our understanding of puma habitat needs and interrelationships of puma management units― ● Puma habitat use, movements, and use of landscape linkages. ● Puma recruitment patterns (i.e., progeny, immigration, emigration). ● Models for identifying puma habitat and landscape linkages. Improve our understanding of the puma’s role in the ecology of other species ● Relationships of puma to mule deer, elk, and other natural prey. ● Relationships of puma to species of special concern, e.g., desert bighorn sheep. Improve our understanding of puma-human interactions and abilities to manage them ● Behavior of puma in relation to people and human facilities. ● Puma predation on domestic animals. ● Effects of translocating nuisance pumas. ● Effects of aversive conditioning on pumas. While all projects cannot be addressed concurrently, understanding their relationships to one another is expected to help individual projects maximize their benefits to other projects that will assist the CDOW to achieve its strategic goal in puma management (Fig.1). This project has been addressing all of the gray-shaded components on the left side of the conceptual model in Figure 1. Management issues identified by managers translate into researchable objectives, requiring descriptive studies and field manipulations. Our goal is to provide managers with reliable information on puma population biology and to develop useful tools for their efforts to adaptively manage puma in Colorado to maintain healthy, self-sustaining populations. The highest-priority management needs are being addressed with this intensive population study that focuses on puma population dynamics using sampled, tagged, and GPS/VHF-radio-collared pumas. Those objectives include: Describe and quantify puma population sex and age structure. Estimate puma population vital rates, including: reproduction rates, age-stage survival rates, emigration rates, immigration rates. Estimate agent-specific mortality rates. Improve the CDOW’s model-based management approaches with Colorado-specific data from objectives 1―3. Consider other useful models. Concurrently with the tasks associated with the objectives above, significant progress will be made toward a 5th objective, which will initially be subject to pilot study― develop methods that yield reliable estimates of puma population abundance. A descriptive and manipulative study will estimate population parameters in an area that appears typical of puma habitat in western Colorado and will yield defensible population parameters based upon contemporary Colorado data. This study will be conducted in two 5-year periods. A completed 5-year reference period, 2004-09, (i.e., absence of recreational hunting) allowed puma life history traits to interact with the main habitat factors that influenced puma population growth (e.g., prey availability and vulnerability, Pierce et al. 2000, Logan and Sweanor 2001, Logan 2009). A subsequent 5-year treatment period started in 2009-10 will involve the use of controlled recreational hunting to manipulate the puma population. 104 �TESTING ASSUMPTIONS AND HYPOTHESES Hypotheses associated with main objectives 1―5 of this puma population research are structured to test assumptions guiding puma management in Colorado. 1. Considering limitations (i.e., methods, number of years, assumption violations) to the previous Colorado-specific studies on puma populations (Currier et al. 1977, Anderson et al. 1992, Koloski 2002), managers assume that puma population densities in Colorado are within the range of those quantified in more intensively studied populations in Wyoming (Logan et al. 1986), Idaho (Seidensticker et al. 1973), Alberta (Ross and Jalkotzy 1992, and New Mexico (Logan and Sweanor 2001). The CDOW assumes density ranges of 2.0−4.6 puma/100 km2 (i.e., includes pumas of all age stages- adults, subadults, and cubs, J. Apker, CDOW Carnivore Biologist, person. commun. Nov. 19, 2003) to extrapolate to DAUs to guide the model-based quota-setting process. Likewise, managers assume that the population sex and age structure is similar to puma populations described in the intensive studies. Using intensive efforts to capture, mark, and estimate non-marked animals developed and refined during the study to estimate the puma population, the following will be tested: H1: Puma densities during the 5-year reference period (absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will vary within the range of 2.0 to 4.6 puma/100 km2 and will exhibit a sex and age structure similar to puma populations in Wyoming, Idaho, Alberta, and New Mexico. 2. Recreational puma hunting management in Colorado Data Analysis Units (DAUs) is guided by a model to estimate allowable harvest quotas to achieve one of two puma population objectives: 1) maintain puma population stability or growth, or 2) cause puma population decline (CDOW, Draft L-DAU Plans, 2004, CDOW 2007). Basic model parameters are: puma population density, sex and age structure, and annual population growth rate. Parameter estimates are currently chosen from literature on studies in western states that are judged to provide reliable information. Background material used in the model assumes a moderate annual rate of growth of 15% (i.e.,λ = 1.15) for the adult and subadult puma population (CDOW 2007). This assumption is based upon information with variable levels of uncertainty (e.g., small sample sizes, data from habitats dissimilar to Colorado). Parameters influencing λ include population density, sex and age structure, female age-at-firstbreeding, reproduction rates, sex- and age-specific survival, immigration and emigration. H2: Population parameters estimated during a 5-year reference period (in absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will yield an estimated annual adult plus subadult population growth rate that will match or exceed λ = 1.15. 3. The key assumption is that the CDOW can manage puma population growth through recreational hunting on the basis that for a stable puma population hunting removes the annual increment of population growth (i.e., from current judgments on population density, structure, and λ). Puma harvest rate formulations for DAUs assumes that total mortality (i.e., harvest plus other detected deaths) in the range of 8 to 15% of the harvest-age population (i.e., independent pumas comprised of adults plus subadults) with the total mortality comprised of 35 to 45% females (i.e., adults and subadults) is acceptable to manage for a stable-to-increasing puma population (CDOW 2007). H3: Total mortality of an estimated 15% of the adults and subadults with no more than 45% of the total mortality comprised of females will not result in a declining trend of the harvest-age segment of the population. 4. To reduce a puma population, hunting must remove more than the annual increment of population growth. For DAUs with the objective to suppress the puma population, the total mortality guide of 105 �greater than 15 to 28% of the harvest-age population with greater than 45% comprised of females is suggested (CDOW 2007). H4: Total mortality of an estimated 16% or greater of the harvestable population with greater than 45% females will cause a declining trend in the abundance of harvest-age pumas (i.e., adults and subadults). 5. The increase and decline phases of the puma population make it possible to test hypotheses related to shifts in the age structure of the population which have been linked to harvest intensity in Wyoming and Utah. H5: The puma population on the Uncompahgre Plateau study area will exhibit a young age structure after hunting prohibition at the beginning of the reference period. During the 5 years of hunting prohibition, greater survival of independent pumas will cause an older age structure in harvest-age pumas (i.e., adults and subadults) as suggested by the work of Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. As hunting is re-instated in the treatment period, the age structure of harvested pumas and the harvest-age pumas in the population will decline as observed by Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. Researchers in Wyoming (Anderson and Lindzey 2005) concluded that sex and age composition of the harvest varies predictably with puma population size because the likelihood of a specific sex or age class of puma being harvested (with the use of hounds) is a product of the relative abundance of particular sex and age classes in the population and their relative vulnerability to harvest. Results of that study suggest that managers could use sex and age composition of the harvest to infer puma population changes (Anderson and Lindzey 2005). The CDOW currently uses this approach as one tool to infer potential DAU puma population dynamics (CDOW 2008). This assumes no purposeful selection by hunters for any particular sex or age-stage other than the puma must be legal (i.e., independent subadult or adult, not a lactating female or a female in association with spotted cubs) and that changes in the sex and age structure of the harvested pumas is due solely to changes in the relative abundance of particular sex and age classes in the population and their relative vulnerability to harvest. Theoretically, pumas that travel longer distances with movements that intercept access routes used by hunters (i.e., roads, trails) should be more exposed to detection by hunters and thus vulnerable to harvest. A key assumption to this method is that pumas are killed as they are encountered and the harvest sex and age composition will reliably indicate whether a population is stable, increasing, or declining even if harvest intensity does not vary. Thus, an alternate view is that a population segment, such as independent females, may be more abundant and have shorter movement lengths, yet be detected more frequently by hunters. However, because the same intensively studied Wyoming puma population was manipulated over 6 years with varying intensities of harvest (Anderson and Lindzey 2005), variations in harvest structure using the same harvest level over a period of years could not be examined. This is a property we will investigate during the treatment period on the Uncompahgre Plateau puma study. Moreover, we will directly evaluate to what extent puma harvest might be influenced by hunter selection. A hunter survey is intended to reveal puma hunter behavior, detection of different classes of pumas, and lack of or presence of hunter selection. We want to examine the usefulness of this approach in Colorado. CDOW managers attempt to weight sport-harvest toward male pumas in GMUs with the stable-to-increasing population objective with an active educational program (i.e., mandatory hunter exam, brochure, workshops). Thus, there is a need to test assumptions associated with the Anderson and Lindzey (2005) method. H6: No hunter selection is practiced so that the sex and age structure of pumas harvested by hunters in this population protected from hunting during a 5-year reference period and subsequently managed for stability or increase with conservative harvest levels will reflect the relative vulnerabilities to detection and capture with dogs during each year in the 5-year treatment period in this order from high to low vulnerabilities: subadult males, adult males, subadult females, adult females without cubs or with cubs >6 months old, and adult females with cubs ≤6 106 �months old (Barnhurst 1982, Anderson and Lindzey 2005). In each of the 5 years of the treatment period, subadults and adult males should comprise the majority of the harvest and reflect the assumed sex and age structure (Anderson and Lindzey 2005) of a puma population managed for a stable to increasing phase and not hunted for 5 previous years (i.e., a puma population source). Desired outcomes and management applications of this research include: 1. Quantification of variations in puma population density, sex and age structure, growth rates, vital rates, and an understanding of factors affecting them will aid adaptive puma management by yielding population parameters and tools useful for assessing puma population dynamics, evaluation of management alternatives, and effects of management prescriptions. 2. Testing assumptions about puma populations, currently used by CDOW managers, will help managers to biologically support and adapt puma management based on Colorado-specific estimated puma population characteristics, parameters, and dynamics. 3. Methods for assessing puma population dynamics will allow managers to evaluate modeled populations and estimate effects of management prescriptions designed to achieve specified puma population objectives in targeted areas of Colorado. Ascertaining puma numbers and densities during the project will allow assessment of monitoring techniques. Potential methods include use of harvest sex and age structure and photographic and DNA genotype capture-recapture. Study plans to develop and test feasible field and analytical methods will be developed as we learn the logistics of performing those methods, after we have preliminary data on puma demographics and movements which will inform suitable sampling designs, and if we have adequate funding. 4. This information will be disseminated to citizen stakeholders interested in pumas in Colorado, and thus contribute to informed public participation in puma management. STUDY AREA The study area for the puma population research is on the Uncompahgre Plateau (in Mesa, Montrose, Ouray, and San Miguel Counties; Fig. 2). The study area includes about 2,253 km2 (870 mi.2) of the southern halves of Game Management Units (GMUs) 61 and 62, and about 155 km2 (60 mi.2) of the northern edge of GMU 70 (between state highway 145 and San Miguel River). The area is bounded by state highway 348 at Delta, 25 Mesa road and Forest Service road FS503 to Nucla, state highway 97 to state highway 141 to state highway 145 to Placerville, state highway 62 to Ridgeway, U.S. highway 550 to Montrose, and U.S. highway 50 to Delta. The study area seems typical of puma habitat in Colorado that has vegetation cover that varies from the pinion-juniper covered foothills starting from about 1,700 m elevation to the spruce-fir and aspen forests growing to the highest elevations of about 3,000 m. Mule deer (Odocoileus hemionus) and elk (Cervus elaphus) are the most abundant wild ungulates available for puma prey. Cattle and domestic sheep are raised on summer ranges on the study area. Year-round human residents live along the eastern and western fringe of the area, and there is a growing residential presence especially on the southern end of the plateau. A highly developed road system makes the study area highly accessible for puma research efforts. A detailed description of the Uncompahgre Plateau is in Pojar and Bowden (2004). METHODS Reference and Treatment Periods This research was structured in two 5-year periods: a reference period (years 1―5) and a treatment period (years 6―10). The reference period was closed to puma hunting on the study area and was expected to cause a population increase phase. The treatment period (starting in November 2009) involves manipulation of the puma population with sport-hunting structured to achieve a management objective for a stable to increasing population. In both phases, puma population structure, and vital rates 107 �will be quantified, and management assumptions and hypotheses regarding population dynamics and effects of harvest will be tested. Contingent upon results of pilot studies, we will also assess enumeration methods for estimating puma population abundance. The reference period, without recreational puma hunting as a major limiting factor, was consistent with the natural history of the current puma species in North America which evolved life history traits during the past 10,000 to 12,000 years (Culver et al. 2000) that enable pumas to survive and reproduce (Logan and Sweanor 2001). In contrast, puma hunting, with its modern intensity and ingenuity, might have influenced puma selection pressures in western North America for at least the past 100 years. Hence, the reference period, years 1―5, provided conditions where individual pumas in this population (of estimated sex and age structure) expressed life history traits interacting with the environment without recreational hunting as a limiting factor. Theoretically, the main limiting factor was vulnerable prey abundance (Pierce et al. 2000, Logan and Sweanor 2001). This allowed researchers to understand basic system dynamics before manipulating the population with controlled recreational hunting. In the reference period, all pumas in the study area were protected, except for individual pumas involved in depredation on livestock or human safety incidents. In addition, all radio-collared and ear-tagged pumas that ranged in a buffer zone in the northern halves of GMUs 61 and 62 were protected from recreational hunting mortality. The reference period allowed researchers to quantify baseline demographic data on the puma population to estimate parameters useful for assessing the CDOW’s assumptions for its model-based approach to puma management. The reference period also facilitated other operational needs (because hunters did not kill the animals) including the marking of a large proportion of the puma population for parameter estimates and gathering movement data from GPS-collared pumas. During the treatment period, years 6―10, recreational puma hunting will occur on the same study area using management prescriptions structured from information learned during previous years. Using recreational hunting for the treatment is consistent with the CDOW’s objectives of manipulating natural tendencies of puma populations, particularly survival, to maintain either population stability or increase or suppression (CDOW, Draft L-DAU Plans, 2004). Theoretically, survival of independent pumas will be influenced mainly by recreational hunting, which will be quantified by agent-specific mortality rates of radio-collared pumas. Dynamics of the puma population will be manipulated to evaluate hypotheses that are related to effects of hunting (i.e.,: effects of harvest rates, relative vulnerability of puma sex and age classes to hunting, variations in puma population structure due to hunting). The killing of tagged and collared pumas during the treatment period is not hampering operational needs (as it would have during the start-up years), because a majority of independent pumas in the population have already been marked, and sampling methods formalized. Pumas on the study area that may be involved in depredation of livestock or human safety incidences may be lethally controlled. Researchers that find that GPS-collared pumas have killed domestic livestock will record such incidents to facilitate reimbursement to the property owner for loss of the animal(s). In addition, researchers will notify the Area Manager of the CDOW if they perceive that an individual puma may be a threat to public safety. Field Methods Puma Capture: Realizing that pumas live at low densities and capturing pumas is difficult, as a starting point, our logistical aim was to have a minimum of 6 puma in each of 6 categories (36 total) radio-tagged in any year of the study if those or greater numbers are present. The 6 categories are: adult female, adult male, subadult female, subadult male, female cub, male cub. Our aim was to provide more quantitative and precise estimates of puma demographics than were achieved in earlier Colorado puma studies. This relatively large number of pumas might represent the majority of the puma population on the 108 �study area, and would provide the basic data for age- and sex-specific reproductive rates, survival rates, agent-specific mortality rates, emigration, and other movement data. Puma capture and handling procedures were approved by the CDOW Animal Care and Use Committee (file #08-2004). All captured pumas were examined thoroughly to ascertain sex and describe physical condition and diagnostic markings. Ages of adult pumas were estimated initially by the gum-line recession method (Laundre et al. 2000) and dental characteristics of known-age puma (Logan and Sweanor, unpubl. data). Ages of subadult and cub pumas were estimated initially based on dental and physical characteristics of known-age pumas (Logan and Sweanor unpubl. data). Body measurements recorded for each puma included at a minimum: mass, pinna length, hind foot length, plantar pad dimensions. Tissue collections included: skin biopsy (from the pinna receiving the 6 mm biopsy punch for the ear-tags), and blood (30 ml from the saphenous or cephalic veins) for genotyping individuals, parentage and relatedness analyses, and disease screening; hair (from various body regions) for genotyping tests of field gathered samples. Universal Transverse Mercator Grid Coordinates on each captured puma were fixed via Global Positioning System (GPS, North American Datum 27). Pumas were captured year-round using 4 methods: trained dogs, cage traps, foot-hold snares, and by hand (for small cubs). Capture efforts with dogs were conducted mainly during the winter when snow facilitated thorough searches for puma tracks and the ability of dogs to follow puma scent. The study area was searched systematically multiple times per winter by four-wheel-drive trucks, all-terrain vehicles, snow-mobiles, and walking. When puma tracks ≤1 day old were detected, trained dogs were released to pursue pumas for capture. Pumas usually climbed trees to take refuge from the dogs. Adult and subadult pumas captured for the first time or requiring a change in telemetry collar were immobilized with Telazol (tiletamine hydrochloride/zolazepam hydrochloride) dosed at 5 mg/kg estimated body mass (Lisa Wolfe, DVM, CDOW, attending veterinarian, pers. comm.). Immobilizing agent was delivered into the caudal thigh muscles via a Pneu-Dart® shot from a CO2-powered pistol. Immediately, a 3m-by-3m square nylon net was deployed beneath the puma to catch it in case it fell from the tree. A researcher climbed the tree, fixed a Y-rope to two legs of the puma and lowered the cat to the ground with an attached climbing rope. Once the puma was on the ground, its head was covered, its legs tethered, and vital signs monitored (Logan et al. 1986). Normal signs include: pulse ~70 to 80 bpm, respiration ~20 bpm, capillary refill time ≤2 sec., rectal temperature ~101oF average, range = 95 to 104oF (Kreeger 1996). Pumas that climbed trees too dangerous for the pumas or researchers were released without handling, or we encourage the animals to leave the tree by heaving snowballs toward them. If the pumas climbed a safe tree, then we handled them as described above. A cage trap was used to capture adults, subadults, and large cubs when pumas were lured into the trap using road-killed or puma-killed ungulates (Sweanor et al. 2008). A cage trap was set only if a target puma scavenged on the lure (i.e., an unmarked puma, or a puma requiring a collar change). Researchers continuously monitored the set cage trap from about 1 km distance by using VHF beacons on the cage and door. Researchers handled captured pumas within 30 minutes of capture. Puma were immobilized with Telazol injected into the caudal thigh muscles with a pole syringe. Immobilized pumas were restrained and monitored as described previously. If non-target animals were caught in the cage trap, we opened the door and allowed the animal to leave the trap. Small cubs (≤10 weeks old) were captured using our hands (covered with clean leather gloves) or with a capture pole. Cubs were restrained inside new burlap bags during the handling process and were not administered immobilizing drugs. Cubs at nurseries were approached when mothers were away from nurseries (as determined by radio-telemetry). Cubs captured at nurseries were removed from the nursery a distance of 30 to 100 m to minimize disturbance and human scent at nurseries. Immediately after handling 109 �processes were completed, cubs were returned to the exact nurseries where they were found (Logan and Sweanor 2001). Marking, Global Positioning System- and Radio-telemetry: Pumas did not possess easily identifiable natural marking, such as tigers (see Karanth and Nichols 1998, 2002), therefore, the capture, marking, and GPS- or VHF- collaring of individual pumas was essential to a number of project objectives, including estimating numbers, vital rates, and gathering movement data relevant to population dynamics (i.e., emigration and Data Analysis Unit boundaries). Adult, subadult, and cub pumas were marked 3 ways: GPS/VHF- or VHF-collar, ear-tag, and tattoo. The identification number tattooed in the pinna was permanent and could not be lost unless the pinna was severed. A colored (bright yellow or orange), numbered rectangular (5 cm x 1.5 cm) ear-tag (Allflex USA, Inc., DFW Airport, TX) was inserted into each pinna to facilitate individual identification during direct recaptures. Cubs ≤10 weeks old were ear-tagged in only one pinna. Adult and subadult female pumas were fitted with GPS collars (approximately 400 g each, Lotek Wireless, Canada) if available. Initially, GPS-collars were programmed to fix and store puma locations at 4 times per day to sample daytime, nighttime, and crepuscular locations (i.e., 0:00, 06:00, 12:00, 19:00). GPS locations for pumas provided precise, quantitative data on movements to assess the relevance of puma DAU boundaries, our search efforts, and to evaluate puma behavior and social structure. The GPScollars also provided basic information on puma movements and locations to design other pilot studies in this program on vulnerability of puma to sport-harvest, habitat use, and enumeration methods (e.g., photographic or DNA mark-recapture). Subadult male pumas were fitted initially with conventional VHF collars (Lotek, LMRT-3, ~400 g each) with expansion joints fastened to the collars, which allowed the collar to expand to the average adult male neck circumference (~46 cm). If subadult male pumas reached adulthood on the study area, we would recapture them and fit them with GPS collars. In addition, other adult and female subadult pumas were fitted with VHF collars when GPS collars were not available. VHF radio transmitters on GPS collars enabled researchers to find those pumas on the ground in real time to acquire remote GPS data reports, facilitate recaptures for re-collaring, and to determine their reproductive and survival status. VHF transmitters on GPS- and VHF-collars had a mortality mode set to alert researchers when puma was immobile for 3 to 24 hours so that dead pumas could be found to quantify survival rates and agent-specific mortality rates by gender and age. Locations of GPS- and VHFcollared pumas were fixed about once per week (as flight schedules and weather allowed) from light fixed-wing aircraft (e.g., Cessna 182) fitted with radio signal receiving equipment (Logan and Sweanor 2001). GPS- and VHF-collared pumas were located from the ground opportunistically using hand-held yagi antenna. At least 3 bearings on peak aural signals were mapped to fix locations and estimate location error around locations (Logan and Sweanor 2001). Aerial and ground locations were plotted on 7.5 minute USGS maps (NAD 27) and UTMs along with location attributes recorded on standard forms. GPS and aerial locations were mapped using GIS software. We attempted to collar all cubs in observed litters with small VHF transmitter mounted on an expandable collar that can expand to adult neck size (Wildlife Materials, Murphysboro, Illinois, HLPM2160, 47g, Telonics, Inc., Mesa, Arizona MOD 080, 62g, or Telonics MOD 205, 90g,) when cubs weighed 2.3―11 kg (5―25 lb). Cubs could wear these small expandable collars until they are over 12 months old. Cubs were recaptured to replace collars as opportunities allowed. Monitoring radio-collared cubs allowed quantification of survival rates and agent-specific mortality rates (Logan and Sweanor 2001). 110 �Analytical Methods Population Characteristics: Population characteristics each year were tabulated with the number of individuals in each sex and age category. Age categories, as mentioned, include: adult (puma ≥24 months old, or younger breeders), subadults (young puma independent of mothers, <24 months old that do not breed), cubs (young dependent on mothers, also called kittens) (Logan and Sweanor 2001). When data allowed, age categories were further partitioned into months or years. Reproductive Rates: Reproductive rates were estimated for GPS- and VHF-collared female pumas directly (Logan and Sweanor 2001). Genetic paternity analysis will be used to ascertain paternity for adult male pumas (Murphy et al. 1998). Survival and Agent-specific Mortality Rates: Radio-collared pumas provided known fate data used to estimate survival rates for each age stage using the Kaplan-Meier procedure to staggered entry (Pollock et al. 1989). A binomial survival model was also used for crude estimates of survival during the subadult age stage (Williams et al. 2001:343-344). In addition, when data collection is complete, survival rates will be modeled in program MARK (White and Burnham 1999, Cooch and White 2004) where effects of individual (e.g., sex, age stage, reproductive stage) and temporal (i.e., reference period, treatment period) covariates to survival can be examined. Agent-specific mortality rates can also be analyzed using proportions and Trent and Rongstad procedures (Micromort software, Heisey and Fuller 1985). Population Inventory: The population of interest was independent pumas (i.e., adults and subadults) mainly during November to March which corresponds with Colorado’s puma hunting season. Independent pumas were those that could be legally killed by recreational hunters. Initially, we estimated the minimum number of independent pumas and puma density (i.e., number of independent puma/100 km2) each winter. The minimum number of independent pumas included all marked pumas known to be present on the study area during the period, plus individuals thought to be non-marked and detected by visual observation or tracks that were separated from locations of radio-collared pumas. Furthermore, adults comprised the breeding segment of the population and subadults were non-breeders that are potential recruits into the adult population in ≤1 year. The sampling unit was the individual independent puma (~≥1 yr. old). Puma Population Dynamics: A deterministic, discrete time model parameterized with population characteristics and vital rates from this research was used to assess puma population dynamics (Logan 2008). Functional Relationships: Once data collection is complete, a variety of analyses will be conducted to estimate parameters and examine functional relationships. Graphical methods will be used to initially examine functional relationships among puma population parameters. Linear regression procedures and coefficients of determination will be used to assess functional relationships if data for the response variable are normally distributed and the variance is the same at each level. If the relationship is not linear, data is non-normal, and variances are unequal, we will consider appropriate transformations of the data for regression procedures (Ott 1993). Non-parametric correlation methods, such as Spearman’s rank correlation coefficient, will also be used where appropriate to test for monotonic relationships between puma abundance and other parameters of interest (Conover 1999). Relationships of explanatory variables to survival parameters will be modeled in MARK. Statistical analyses can be performed in a variety of software (e.g., SYSTAT, R, and MARK). 111 �RESULTS AND DISCUSSION Segment Objective 1 Puma harvest: This biological year, August 2009 to July 2010, was the first year of the treatment period in this study of puma population dynamics on the Uncompahgre Plateau. Principal investigator K. Logan with CDOW biologists and managers developed a structure (i.e., officially approved by Wildlife Commission decision in September 2009) to manipulate the puma population with sport-hunting and to survey hunters. The hunting season on the study area began on November 16, 2009 and was scheduled to extend to January 31, 2010, unless the harvest quota was taken before then. The design harvest quota was 8 pumas (i.e., 15% harvest of the estimated minimum number of independent pumas), with the objective to manage for a stable to increasing population. This design harvest tests the CDOW’s current assumption that total mortality (i.e., harvest plus other natural deaths) in the range of 8 to 15% of the harvest-age population (i.e., independent pumas comprised of adults plus subadults) with the total mortality comprised of 35 to 45% females (i.e., adults and subadults) is acceptable to manage for a stable-toincreasing puma population (Assumption and Hypothesis 3 p.5 this report). The quota of 8 was based on the projected minimum number of 53 independent pumas expected on the study area in winter 2009-10, modeled from a minimum count of pumas during winter 2007-08 (Table 1). We relied on the count data from 2007-08 because that was the last year in the reference period in which a fully staffed research team was able to adequately survey the study area in winter capture operations. The next year, 2008-09 (i.e., the last year of the reference period), a state government-mandated hiring freeze contributed to subpar winter capture operations, and thus, an inadequate minimum count effort. The number of puma hunters on the study area was not limited. Each hunter on the study area was required to obtain a hunting permit from the CDOW Montrose Service Center. Permits were free and unlimited. Each permit allowed the individual hunter with a legal puma hunting license in Colorado to hunt in the puma study area for up to 14 days from the issue date. Unsuccessful hunters that wanted to continue hunting past the permit expiration date requested a new permit for another 14 days or until the hunter killed a puma within the season, or the season on the study area closed due to the quota being reached, or the end of the hunting season. This permit system allowed the CDOW to monitor the number of hunters on the study area and to contact each hunter for survey information (see later). All pumas harvested on the study area were examined by principal investigator K. Logan and sealed as mandated by Colorado statute. All successful hunters reported their puma kill and presented the puma carcass for inspection by CDOW within 48 hours of harvest. Upon inspection data was recorded on the puma harvested, including: sex, age, and location of harvest. In addition, an upper premolar tooth was collected for aging (i.e., mandatory) and a tissue sample was collected for DNA genotyping. Each successful hunter was also asked at that time to complete a one-page hunter survey form. All other hunters that did not report a puma kill on the study were asked to complete the survey form and return it in a stamped envelope that was provided. An attempt was made to contact other hunters by telephone if they did not mail in surveys. The puma hunting season occurred on the study area from November 16 to December 11, 2009, taking 26 days to fill the quota of 8 pumas. Nine pumas were killed, including: 2 adult females, 1 subadult female, 5 adult males, and 1 dependent male cub (Table 2). Three of the pumas were killed on the last day, resulting in the quota being exceeded by 1 puma. Of the harvested pumas, 3 were marked: dependent male cub M91 (offspring of F25), and 2 adult males M51 and M71. In addition to the pumas killed on the study area, 1 adult female (F110) and 3 adult males (M27, M29, M100) that had home ranges overlapping the study area were killed off the study area on adjoining GMUs (Table 3). The harvest of 8 independent pumas on the study area was 14.5% (8/55*100) of the minimum count of 55 independent pumas, including 31 females and 24 males, estimated by the research team 112 �during September 2009 to April 2010 (Table 4). Independent females and males comprised 37.5% (3/8*100) and 62.5% (5/8*100) of the harvest, respectively. This harvest structure was 9.7% (3/31*100) of the independent females and 20.8% (5/24*100) of the independent males. Considering the harvest of 4 other radio-collared adults (F110, M27, M29, M100) off the study area, which had home ranges overlapping on and off the study area, a harvest of 12 independent pumas was 21.8% (12/55*100) of the minimum number of independent pumas. The harvest composition of 4 females and 8 males was comprised of 33.3% (4/12*100) females and 66.7% (8/12*100) males. This harvest structure was 12.9% (4/31*100) of the independent females and 33.3% (8/24*100) of the independent males in the minimum count. The minimum count of independent pumas in 2009-10 was highly consistent with the expected number and sex structure of independent pumas projected by the deterministic, discrete time model (see Tables 1 and 4. Minimum count 2009-10 = 55 independent pumas, including 31 females, 24 males. Model projected independent pumas = 53, including 31 females, 22 males). Therefore, we used the model to guide the decision to manipulate the puma population with a harvest of 8 independent pumas in the 201011 hunting season to emulate an approximate 15% harvest of independent pumas to achieve a stable to increasing population objective while also considering that a number of independent pumas in the study area population will probably be killed outside of the study area as in the 2009-10 hunting season (Fig. 3). The projected population trends are stable-to-increasing. Hunter permits and survey: Mandatory permits with the voluntary survey attached were requested by 79 individual hunters. Thirty-three of the hunters requested a second permit after the first one expired after 14 days. Seventy-one hunters (90%) provided responses to the voluntary survey either by turning in the survey (i.e., n = 43) or providing information during follow-up telephone calls (i.e., n = 28) by principal investigator K. Logan. The remaining 8 hunters could not be contacted, because either they did not have working phone numbers or they did not return calls. Of the respondents, 11 hunters indicated that they did not hunt on the study area. As a proportion of the 71 respondents, the number that hunted extrapolated to the total of 79 hunters (60/71 = 0.845) indicated that about 67 hunters took to the field for pumas on the study area during the 26-day hunting season. Considering that 67 hunters were estimated to be afield, then 13% harvested pumas (9/67*100) and 24% of individual hunters captured pumas (16/67*100; see captured and released pumas below and in Table 5). In response to the survey question, “Do you consider yourself a selective or non-selective hunter?” all the respondents that hunted on the study area indicated that they were selective hunters. (A selective hunter is one that purposely is hunting for a specific type of legal puma, such as a male, large male, or large female. A non-selective hunter is one that intends to take whatever legal puma is first encountered or caught, with no desire for sex or size.) Yet, selective hunter was indicated by the 3 hunters that killed a subadult female, a lactating female, and a dependent male cub, which may indicate that in fact not all the hunters are selective or some cannot distinguish types (i.e., sex, age stage) of pumas in the field to practice selection. On the other hand, hunter surveys also revealed that hunters treed pumas on the study area, but they chose not to kill them (Table 5). Those hunters reported they treed pumas 14 times, including 9 females and 5 males. All 9 females were described by the hunters as adult age; 2 males were described as adult age, and 3 males were described as subadults. Five of the treed pumas were marked, including adult female F8 treed twice, adult female F74, and 2 yellow ear-tagged subadult males (numbers could not be distinguished). Hunters gave various reasons for not wanting to kill the pumas, including reasons based on puma sex and size (Table 5). These preliminary survey and harvest data indicate independent females were probably captured slightly more frequently than independent males (i.e., ratio 12 females:10 males; females = 3 harvested + 9 captured and released; males = 5 harvested + 5 captured and released). This sex structure was consistent with the sex structure of the independent pumas in the minimum count (Table 4). Yet, the harvest was comprised of mostly males (3 females, 5 males). 113 �This preliminary assessment from TY1 puma harvest and hunter survey data suggests that most hunters that captured pumas were selective and influenced harvest sex and age composition. Segment Objective 2 After the design quota was filled, puma research teams immediately activated for capture operations with trained dogs. Two fully-staffed capture teams, one detailed on the east slope and one detailed on the west slope, systematically and thoroughly searched the study area to capture, sample, and GPS/VHF radiocollar pumas the remainder of winter and early spring 2009-10. These efforts along with cage trap efforts and hand-capturing cubs at nurseries maintained samples to quantify population sex and age structure, survival, and agent-specific mortality, and allowed determination of minimum population size on the study area. We made 34 puma captures of 28 individuals from August 2009 to July 2010 (Tables 6-11). Twenty-one individual pumas were captured with dogs 26 times. Three pumas were captured in cage traps. Cubs were captured at nurseries 5 times. A total of 42 pumas were monitored with radiotelemetry from August 2009 to July 2010 (some of these had been collared in previous years). In addition, 2 cubs were monitored from birth to death at the nursery by monitoring the GPS and VHF data of their mother. Trained dogs were used as our main method to capture, sample, and mark adult and subadult pumas from December 15, 2009 to April 30, 2010. Those efforts resulted in 86 search days, 266 puma tracks detected, 93 pursuits, and 26 puma captures (Table 6). Search days with dogs in this period was greater than our efforts in the 4 previous winters by 4 to 15 days (Table 12). In addition, this was the first year we deployed 2 fully-staffed hound capture teams. The frequency of tracks (tracks/day) encountered was higher than the previous 5 winters. The pursuits increased over all previous years by 18 to 58, with the lowest number of pursuits occurring in the first year of this study (2004-05). The capture rate was also the highest by 2 to 12 captures. Increased capture efforts and captures were probably the result of using 2 fully-staffed houndsmen teams even though the puma population had been reduced due to harvest just before our capture operations. Researchers also recorded instances when the first tracks ≤1 day old of independent pumas were encountered on each search route each day to represent encounters with puma tracks that could be pursued by houndsmen. The count was: 37 tracks of females, including 5 associated with cubs; 21 tracks of males; and 2 tracks of unspecified sex. The ratio of female to male tracks was consistent with the sex structure of independent pumas in our minimum count (Table 4). Puma capture efforts using ungulate carcasses and cage traps extended from September 11, 2009 to May 17, 2010 (Table 10). We used 21 road-killed mule deer at 17 different sites, but did not capture any pumas. However, 2 adult pumas (M55, F94) were each recaptured in cage traps at mule deer kills they made. Pumas scavenged at 3 of 17 (17.65%) sites where ungulate carcasses were used for bait. A bobcat trapper inadvertently caught male cub M112 (offspring of F70) in a cage trap. The trapper notified us, and we sampled, tagged, radio-collared, and released the cub. The cub successfully rejoined his family. We captured 5 cubs, all males for the first time (Table 11), and fit all with radio-collars (Appendix A). Two cubs of F3 were captured at nurseries, 2 were bayed by hounds (M115 of F28, M117 of F119), and 1 was caught in a bobcat cage trap (M112 of F70, see above). In addition, we found 2 male cubs (P1016, P1017) of F72 that were killed by male puma M32 on the day we investigated the nursery to sample and tag the cubs (see later). Two cubs of F93 were observed in the nursery at about 28 days old, but they could not be handled because the rock structure of the nursery afforded them complete protection from capture. 114 �In addition to our direct puma captures with dogs December through April, we detected 16 pumas that we were able to identify with GPS or VHF telemetry 38 times, thus, negating the need to capture those pumas directly with dogs (Table 6). Upon detecting puma tracks that were aged at ≤1 day old, we followed the tracks with a radio receiver in an effort to detect if the tracks might be of a puma wearing a functional collar. We assigned tracks to a collared individual if we received radio signals from a puma that we judged to be <1 km from the tracks and in direction of travel of the tracks. GPS data from pumas with functional GPS collars provided confirmatory information about movements of pumas. If GPS data indicated that the puma moved through the area at the time the tracks were made, then we ruled the data were confirmatory. This approach allowed us to more efficiently allocate our capture efforts toward pumas of unknown identity on the study area, particularly unmarked pumas or pumas with nonfunctioning GPS- or radiocollars. Our search efforts throughout the study area also revealed the presence of at least 15 other independent pumas, we classified as 9 females and 6 males. Two of the males were treed by our hounds, but we could not handle the pumas because they climbed dangerous trees (Table 7). We could separate the activity of the other pumas from the GPS- and VHF- collared pumas in time, space, and track size differences between females and males. Moreover, females in association with cubs of different numbers, sizes, and locations enabled us to separate 2 adult females followed by 2 to 3 medium-to-large-size cubs. The tracks we found of the other pumas were too old to pursue (i.e., probability of capture with the dogs was negligible). One of the adult females was likely F74, which was also treed and observed by a puma hunter on December 9, 2009. It is also possible that 1 of the adult females was previously marked animal F24 wearing non-functional GPS collar. Our search and capture efforts during September 2009 through April 2010 enabled us to quantify a minimum count of 55 independent pumas detected on the Uncompahgre Plateau study area, including 31 independent females and 24 independent males (Table 4). This count was based on the number of known radio-collared pumas, non-marked pumas harvested by hunters on the study area, observations of marked and non-marked pumas observed by researchers or treed and released by hunters on the study area, and fresh puma tracks (i.e., ≤ 1 days old) observed by researchers that could not be attributed to pumas with functioning radiocollars. The estimated age structure of independent pumas in November 2009 at the beginning of the puma hunting season in Treatment Year 1 (TY1) on the Uncompahgre Plateau study area is depicted in Figure 4. In addition to the independent pumas, we also counted a minimum of 20 to 25 cubs. Of the 55 independent pumas, 34 to 35 (62-64%) were marked and 20 to 21 (36-38%) were assumed to be unmarked animals. Of the expected unmarked pumas, 10 to 11 were females and 10 were males. The abundance and sex structure of independent pumas on the east and west slopes of the study area were similar. The east slope count included 28 independent pumas (17 females, 11 males). The west slope count included 27 independent pumas (14 females, 13 males). Considering the minimum count of 55 independent pumas, a preliminary minimum density for the winter puma habitat area estimated at 1,671 km2 on the Uncompahgre Plateau study area was 3.29 independent pumas/100 km2. Segment Objective 3 During the past 5.7 years of this work we compiled data on puma reproduction that was not previously available on pumas in Colorado. Puma reproduction data (i.e., litter size, sex structure, gestation, birth interval, proportion of females giving birth per year) were summarized for the reference period in Logan (2009). We observed 6 litters born in June (3), July (2), and August (1) 2010, each with 1 to 3 cubs each, born to radio-collared females. We found sign (i.e., nurseries, tracks) of a fourth litter born in June to a GPS-collared female (F111); but, we could not catch the cubs before they developed well enough to escape us (about 6 weeks old). Data on reproduction observed in this first year of the treatment period were added to Table 13, but will not be summarized again until the end of the period. The proportion of radio-collared adult females giving birth from August 2009 to July 2010 biological year was 0.53 (8/15). 115 �Considering our 32 total observed litters and 2 other litters confirmed by nurseries and nursling cub tracks with GPS-collared females, all with cubs 26 to 42 days old, the distribution of puma births by month indicate births extending from March into September, with 24 of 34 births (70.6%) occurring May through July (Fig. 5). Our data suggests that the large majority of puma breeding activity occurred February through April. In contrast, Anderson et al. (1992:47-48) found on the Uncompahgre Plateau that of 10 puma birth dates 7 were during July, August, and September, 2 in October, and 1 in December, with most breeding occurring April through June. Data on our 34 litters added to Anderson’s data (Fig. 5), and indicated puma births on the Uncompahgre Plateau occurred in every month except January and November (so far). Segment Objectives 4 & 5 From December 8, 2004 (capture and collaring of the first adult puma M1) to July 31, 2010, we radio-monitored 14 adult male and 26 adult female pumas to quantify survival and agent-specific mortality rates (Table 14). Survival and agent-specific mortality of adult pumas were summarized for the reference period in Logan (2009). Preliminary estimates of adult puma survival rates in the absence of sport-hunting indicated relatively high survival, with adult male survival generally higher than adult female survival (Table 15). For this first year of the treatment period, we monitored 19 adult radio-collared females and 8 radio-collared adult males. The initial indication is that adult survival rates declined for adult females and males (Table 15). But, no conclusions should be drawn with only 1 year in the treatment period (TY1). The primary interest is the magnitude of reduction in survival, and the implications of those survival rates for population growth rate. This is what ultimately allows us to evaluate the effect of this harvest level for our population management assumptions when the goal is a stable to increasing population. Causes of mortality for adult pumas with functioning radiocollars in TY1 were due to vehicle strikes on roadways (2 females, 1 male) and hunting (1 female, 1 male). In addition, all 5 adult males which developed non-functional radiocollars (M1, M27, M29, M51; Table 3) or shed a collar (M71) since the beginning of this study were harvested by hunters in TY1. Inclusion of those adult males in the survival estimate indicated a substantially lower adult male survival rate in TY1 (Table 15). We have radio-monitored 11 pumas, 4 females and 7 males, in the subadult age-stage (independent pumas <24 months old) (Table 16). Three died before reaching adulthood, indicating a preliminary finite survival rate of 0.727. All 3 subadults apparently died of natural causes. F66 died at 23 months old of trauma to internal organs that caused massive bleeding attributed to trampling by an elk or mule deer. M99 died at about 16 months old due to unknown causes; but, punctures in the skull suggested strife with another puma. M115 died at about 14 months old due to complications of a broken left foreleg, cause unknown. This injury probably affected his ability to efficiently kill prey. We need to increase our efforts to acquire larger samples of male and female radio-monitored subadult pumas to acquire reliable estimates of their survival. Data from puma hunters provided additional information on fates of 13 pumas, 12 males and 1 female, initially captured and marked as cubs (10 males) or subadults (2 males, 1 female) on the Uncompahgre Plateau puma study area (Table 17). All 12 of the males were killed away from the study area by hunters at linear distances (i.e., from initial capture sites to kill sites) ranging from about 60 to 370 km. Two males with extreme moves were killed in the Snowy Range of southeastern Wyoming (369.6 km) and the Cimarron Range of north-central New Mexico (329.8 km). The female (F52) was treed and released by hunters in December 2008 and 2009 south of Powderhorn, Colorado, indicating that she probably established an adult home range there. These pumas represent dispersal moves from the 116 �Uncompahgre Plateau. Eleven of the 13 pumas (except M68, 17 months old and M82, 19 months old) had reached adult ages ranging from 24 to 55 months old. A preliminary estimate of cub survival during the reference period was summarized in Logan 2009. In that summary 36 radio-collared cubs (16 males, 20 females) marked at nurseries when they were 26 to 42 days old were used for a Kaplan Meier procedure cub survival estimate of about 0.53 to one year of age. The major natural cause of death in cubs, where cause could be determined, was infanticide and cannibalism by other, especially male, pumas. In this first year of the treatment period, we monitored the fates of 19 cubs (Appendix A). Five of the cubs were known to have died, all of them associated with infanticide. Two (M101, F103) were orphaned at 149 days old when their mother (F16) was hit by a vehicle on County Road 1 on September 11, 2009. The 2 cubs were killed and partially eaten by adult male puma M55 on September 17 and 19, 2009. Fate of their sibling M102 was unknown because of a failed radiocollar after September 4, 2009. But M102 probably would have died of starvation if he was not killed by M55. F72’s 2 male cubs were killed, and 1 partially eaten, by adult male puma M32 at the nursery when the cubs were 39 days old on July 21, 2010. Mother F72 was about 2 km away from the nursery at the time the cubs met their fate. A greater number of cubs over a longer period of time must be sampled before estimating cub survival and agent-specific mortality rates in the treatment period. In addition, a 2-year-old non-marked male puma was struck and killed by a vehicle on highway 62 in Leopard Creek on the south boundary of the study area on August 25, 2010. This mortality made the twelfth puma death recorded due to vehicle collision on the study area since 2004 (Table 18). Five of the 12 pumas were marked, including 3 adults with GPS/VHF collars. Those 3 adults died during the first year of the treatment period. Segment Objective 6 We wanted to enhance this project with reliable estimates of puma abundance and density (see Objective 5, page 4). Because a majority of independent pumas were individually marked on the study area, we decided to explore the potential of using a camera grid mark-recapture structure to derive puma abundance estimates by first examining detection probabilities in a pilot effort. This effort is an attempt to develop puma population monitoring methods (Fig. 1). A camera grid mark-recapture approach is a method for counting pumas independent from our main method of capturing pumas with searches on snow-covered routes and dogs and thus has the potential of providing unbiased estimates. For this pilot project, we collaborated with Colorado State University Researchers Jesse Lewis (Ph.D. candidate) and Dr. Kevin Crooks (Dep. of Fish, Wildlife, and Conservation Biology) who studied bobcat distribution, abundance, and behaviors on the eastern slope of our Uncompahgre Plateau puma study area. Because those researchers used a camera grid design for bobcats where we also had GPS/VHF- collared pumas, this gave our project an opportunity to evaluate puma detection probability on a small scale. This was a first step in considering the usefulness of a camera grid design for puma abundance estimates. We established 2 camera grids on the east slope of the study area (Fig. 6). Each grid was 80 square kilometers in size and contained 20 cells which were each 4 square kilometers. We searched each grid for potential camera sites with the intention to maximize the encounter of a puma or bobcat with a camera. We used our general knowledge about puma and bobcat behavior to place the cameras and did not use any GPS/VHF data on puma locations. Felid sign on the ground (i.e., tracks, feces, scrapes) helped to guide our camera placement. Initially we placed 1 Cuddeback Capture digital camera (Park Falls, WI) in each cell at the site we deemed best to intercept wild felids, and did not use scent or sight lures in an attempt to attract the felids. One alternate camera site was placed in Area 1 and 5 alternate camera sites were placed in Area 2 to increase the sample effort in canyon bottoms relative to canyon rims. All cameras were set at the highest design setting of 1 photo per 30 seconds if the passive infrared 117 �sensor were activated and serviced every 2 weeks to ensure operation and fresh batteries. Cameras operated for 108 days from August 21 to December 7, 2009. During the period that the cameras operated, 4 adult (2 females, 2 males) GPS/VHF-collared pumas ranged on Area 1 and 5 adult (3 females, 2 males) GPS/VHF-collared pumas ranged on Area 2. Those pumas were photographed a total of 51 times: 17 times in Area 1 and 34 times in Area 2. Three of 4 adult pumas (probability 0.75) on Area 1 and 4 of 5 adult pumas (probability 0.80) were detected 2 to 19 times each in the 108 day period. Daily detection rates ranged from 0.02 to 0.18 (Table 19). Detection rates varied among individuals, and were the highest for adult males. Both adult pumas that were not detected were females. One, F16, died on September 11, so was available for 21 days. The other, F70, had a new litter of cubs on August 31 at a nursery in a canyon between the 2 camera grids where she focused her activities. Then on September 23 her GPS collar quit functioning and we were unaware of her movements. In addition, 4 other marked pumas without functioning collars were detected by cameras a total of 7 times. Those pumas were: adult F3 (detected 3 times; non-functional GPS collar), adult M71 (detected twice; eartags, shed expandable VHF collar), a subadult female detected once (orange eartag right pinna), and a male cub detected once (yellow eartag left pinna). Non-marked pumas were photographed 31 times on the camera grids. In Area 1 non-marked pumas were photographed 20 times at primary cameras and once at the alternate camera. We estimated the photos represented 2 to 4 individual independent pumas. In Area 2 non-marked pumas were photographed 8 times at primary cameras and twice by alternate cameras. We estimated the photos represented 3 to 5 independent pumas. Any of the non-marked pumas could have ranged on both camera grid areas. Our next step in this collaborative process is to analyze the photographic data on the 2 grids, including modeling detection probabilities with landscape and puma covariates and to examine expected estimates of precision. We also will examine population closure and investigate methods for defining the survey area by using the GPS and VHF locations of pumas with functioning collars that used the camera grid areas. This information will be used to assess the feasibility of designing a camera grid specifically to obtain accurate and precise estimates of puma abundance and density on a portion of the Uncompahgre Plateau study area. This phase is expected to be completed by July 2011. Segment Objective 7 Data from 28 (8 male, 20 female) GPS-collared pumas, totaling over 48 thousand GPS locations (Table 20) will be used to examine behaviors and social structure of the puma population on the Uncompahgre Plateau, including movements of pumas relative to Game and Data Analysis Unit boundaries and vulnerability to hunter detection. Those data will also be used in a set of collaborative projects, including: examination of puma behavior in relation to human development with Mammals Researcher Dr. Mat Alldredge, who is studying puma-human interactions on the Colorado Front Range and modeling and mapping puma habitat in Colorado and other western states with Dr. Kevin Crooks and Dr. Chris Burdett (Department of Fish, Wildlife and Conservation Biology, Colorado State UniversityDFWCB, CSU). Furthermore, puma population and genetic data from the Uncompahgre Plateau can be used in collaboration with Dr. Alldredge’s puma research efforts on the Front Range to examine similarities or differences in puma population dynamics and behaviors between the 2 environments. 118 �SUMMARY Manipulative, long-term research on puma population dynamics, effects of sport-hunting, and development and testing of puma enumeration methods began in December 2004. After 5.7 years of effort 125 pumas have been captured, sampled, marked, and released. Of those animals, 107 were radiomonitored, allowing us to monitor fates of pumas in all sexes and age stages, including: 25 adult females, 13 adult males, 4 subadult females, 7 subadult males, 32 female cubs, 39 male cubs (some individuals occur in more than one age-stage). Data from the marked animals were used to quantify puma population characteristics and vital rates in a reference period without sport-hunting off-take as a mortality factor from December 2004 to July 2009. Puma population characteristics and vital rates in a reference condition allowed us to develop a puma population model, and to use population data and modeling scenarios to conduct a preliminary assessment of CDOW puma management assumptions and guide directions for the remainder of the puma research on the Uncompahgre Plateau. Moreover, our data and model provide tools currently useful to CDOW wildlife biologists and managers for assessing puma harvest strategies. The first year of the 5-year treatment period was August 2009 to July 2010 in which sport-hunting is a mortality factor. The treatment period will be a population-wide test of CDOW puma management assumptions. The puma harvest quota for TY2 will be 8 independent pumas, and the hunters will be surveyed again. To improve data on puma population vital rates, attention will be given to increasing radio-collared sample sizes on life stages and sexes. Furthermore, we will continue collaboration efforts with colleagues on investigations of puma population parameter estimation, pumahuman relations, puma habitat modeling and mapping, and individual puma detection rates in camera grid designs. All of these efforts should enhance the Colorado puma research and management programs. 119 �LITERATURE CITED Anderson, A. E. 1983. A critical review of literature on puma (Felis concolor). Colorado Division of Wildlife Special Report No. 54. _____, D. C. Bowden, and D. M. Kattner. 1992. The puma on Uncompahgre Plateau, Colorado. Technical Publication No. 40. Colorado Division of Wildlife, Denver. Anderson, C. R., Jr., and F. G. Lindzey. 2005. Experimental evaluation of population trend and harvest composition in a Wyoming cougar population. Wildlife Society Bulletin 33:179-188. Colorado Division of Wildlife 2002-2007 Strategic Plan. 2002. Colorado Department of Natural Resources, Division of Wildlife. Denver. Colorado Division of Wildlife. 2007. Colorado mountain lion management data analysis unit revision and quota development process. Colorado Division of Wildlife, Denver. Conover, W. J. 1999. Practical nonparametric statistics. John Wiley & Sons, Inc., New York. Cooch, E., and G. White. 2004. Program MARK- a gentle introduction, 3rd edition. Colorado State University, Fort Collins. Culver, M., W. E. Johnson, J. Pecon-Slattery, and S. J. O’Brien. 2000. Genomic ancestry of the American puma (Puma concolor). The Journal of Heredity 91:186-197. Currier, M. J. P., and K. R. Russell. 1977. Mountain lion population and harvest near Canon City, Colorado, 1974-1977. Colorado Division of Wildlife Special Report No. 42. Heisey, D. M., and T. K. Fuller. 1985. Evaluation of survival and cause specific mortality rates using telemetry data. Journal of Wildlife Management 49:668-674. Karanth, K. U., and J. D. Nichols. 2002. Monitoring tigers and their prey: A manual for researchers, managers and conservationists in tropical Asia. Centre for Wildlife Studies, Bangalore, India. Koloski, J. H. 2002. Mountain lion ecology and management on the Southern Ute Indian Reservation. M. S. Thesis. Department of Zoology and Physiology, University of Wyoming, Laramie. Kreeger, T. J. 1996. Handbook of wildlife chemical immobilization. Wildlife Pharmaceuticals, Inc., Fort Collins, Colorado. Laundre, J. W., L. Hernandez, D. Streubel, K. Altendorf, and C. L. Lopez Gonzalez. 2000. Aging mountain lions using gum-line recession. Wildlife Society Bulletin 28:963-966. Logan, K. A., E. T. Thorne, L. L. Irwin, and R. Skinner. 1986. Immobilizing wild mountain lions (Felis concolor) with ketamine hydrochloride and xylazine hydrochloride. Journal of Wildlife Diseases. 22:97-103. _____, L. L. Sweanor, J. F. Smith, and M. G. Hornocker. 1999. Capturing pumas with foot-hold snares. Wildlife Society Bulletin 27:201-208. _____, and L. L. Sweanor. 2001. Desert puma: evolutionary ecology and conservation of an enduring carnivore. Island Press, Washington, D.C. _____. 2009. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Wildlife, Fort Collins. Murphy, K., M. Culver, M. Menotti-Raymond, V. David, M. G. Hornocker, and S. J. O’Brien. 1998. Cougar reproductive success in the Northern Yellowstone Ecosystem. Pages 78-112 in The ecology of the cougar (Puma concolor) in the Northern Yellowstone ecosystem: interactions with prey, bears, and humans. Dissertation, University of Idaho, Moscow. Ott, R. L. 1993. An introduction to statistical methods and data analysis. Fourth edition. Wadsworth Publishing Co., Belmont, California. Pierce, B. K., V. C. Bleich, and R. T. Bowyer. 2000. Social organization of mountain lions: does land a tenure system regulate population size? Ecology 81:1533-1543. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Pojar, T. M., and D. C. Bowden. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550-560. 120 �Ross, P. I., and M. G. Jalkotzy. 1992. Characteristics of a hunted population of cougars in southwestern Alberta. Journal of Wildlife Management 56:417-426. Seidensticker, J. C., M. G. Hornocker, W. V. Wiles, and J. P. Messick. 1973. Mountain lion social organization in the Idaho Primitive Area. Wildlife Monographs No. 35. Stoner, D. C. 2004. Cougar exploitation levels in Utah: implications for demographic structure, metapopulation dynamics, and population recovery. Master of Science Thesis. Utah State University. Sweanor, L. L., K. A. Logan, J. W. Bauer, B. Milsap, and W. M. Boyce. 2008. Puma and human spatial and temporal use of a popular California state park. Journal of Wildlife Management 72:10761084. Williams, B. K., J. D. Nichols, and M. J. Conroy. 2001. Combining closed and open mark-recapture models: the robust design. Pages 523-554 In Analysis and management of animal populations. Academic Press, New York. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 (Suppl):S120-S139. Prepared by: ________________________ Kenneth A. Logan, Wildlife Researcher 121 �Table 1. Projected puma population growth modeled from a minimum count of independent pumas during winter 2007-08 reference period year 4 (RY4). Treatment period year 1 (TY1), shaded in gray, indicates the results used to derive a quota of 8 independent pumas, representing 15% of the independent pumas (from Logan 2009). Harvest Level No harvest. Projected Minimum Puma Population Size Adult Subadult Female Male Female Male 16 8 5 4 18 10 9 8 23 14 8 8 27 17 11 10 32 22 12 11 38 27 15 14 44 32 17 16 Year RY4 RY5 TY1 TY2 TY3 TY4 TY5 Independent Pumas Cub 20 33 42 49 58 69 81 Total 33 45 53 64 77 92 110 Lambda 1.37 1.17 1.22 1.20 1.20 1.19 Table 2. Pumas harvested by sport-hunters in Treatment Year 1 (TY1) on the Uncompahgre Plateau Study Area, Colorado, November 16 to December 11, 2009. Puma sex/age/mark M (cub of F25) M Age (yr.) 1.25 2-3 11/21/2009 F 4 12/9/2009 F 1.5-2 12/9/2009 M 4 M 4 12/9/2009 F (lactating) M 2 12/11/2009 M 2-3 7 Previous I.D. M91 M71 M51 Date of kill Location/UTM Hunter/status 11/17/2009 Pleasant Valley/ 13S,247640E,4228470N Little Bucktail Creek/ 12S,726165E,4240290N San Miguel Canyon/ 12S,732268E,4234711N Pinyon Ridge/ 13S,256380E,4241740N Spring Creek/ 12S,762033E,4248487N Horsefly Canyon (E)/ 13S,249114E,4240143N Roubideau Canyon/ 12S,746670E,4254762N Shavano Valley/ 12S,761117E,4256800N Mailbox Park/ 12S, 726524E,4234984N Jack Flowers/ Resident Ty Spangler/ Resident Larry McPeak/ Non-resident M. Ryan Hatter/ Resident Caleb Marquardt/ Resident Darren Reed/ Resident Brian Coe/ Non-resident Darrel Moberly/ Resident Donald Gambril/ Non-resident 12/9/2009 12/11/2009 12/11/2009 122 �Table 3. Five other independent pumas from the Uncompahgre Plateau Study Area killed by hunters off of the study area. Four adult pumas– F110, M27, M29, M100– were in the minimum count on the study area in winter 2009-10.a Adult male M1 probably no longer ranged on the study area. Puma sex/age/mark M29/adult Date of kill 11/16/2009 Place of kill/UTM Hunter/status Beaver Creek (GMU70east) Syver Bicknase/ 12S,745500E,4219660N Resident M27/adult 12/9/2009 N. Fork Mesa Creek (GMU61north) Kevin Thornton/ 12S,693422E,4266607N Non-resident M1/adult 1/2/2010 West Bang’s Canyon (GMU40) Outfitter Steve Biggerstaff 12S,710656E,4314243N M100/adult 1/16/2010 Naturita Canyon Outfitter Wade Wilson 12S,734604E,4216634N F110/adult 2/25/2010 Naturita Creek Alex Sokolik/ 12S,721010E,4230929N Resident a All five adult male pumas with non-functioning (4) or shed (1) radiocollars were killed during TY1 either on (M51, M71) or off (M1, M27, M29) of the UP Study Area. Table 4. Minimum count of pumas based on numbers of known radio-collared pumas, visual observations of non-marked pumas, harvested non-marked pumas, and track counts of suspected non-marked pumas on the study area during September 2009 to April 2010, Uncompahgre Plateau study area, Colorado. Study Area Adults Subadults region Female Male Female Male East slope 16 10 1 1 West slope 14 10 0 3 subtotals 30 20 1 4 Total Independent Pumas = 55, including 31 females, 24 males Female 1 3 4 Cubs Male 4 3 7 Unknown sex 4-8* 5-6 9-14 *One adult non-marked female puma was killed by a hunter in Roubideau Canyon. The female puma was lactating, indicating she had nurslings. Up to 4 cubs were assumed to be in the litter. 123 �Table 5. Pumas captured and released by sport-hunters in Treatment Year 1 (TY1) on the Uncompahgre Plateau Study Area, Colorado, November 16 to December 11, 2009. Data are from puma hunter responses in 71 voluntary surveys, including: 43 original surveys on mandatory permits and 28 telephone contacts with hunters that did not return surveys on permits. Total response rate from 79 individual hunters was 90% (71/79 = 0.899*100). Puma sex/age stage/mark F/adult/none Date of capture 12/1/2009 Capture location Hunter name Preston Joseph 12/8/2009 N. Fork Cottonwood Creek N. Fork Cottonwood Creek DeVinney Canyon F/adult/F8 collar & eartag M/subadult/ yellow eartags in both ears (numbers not distinguished) F/adult/F74 orange eartags 12/7/2009 12/9/2009 Cottonwood Creek 11/30 to 12/7/2009 11/30 to 12/7/2009 Loghill Mesa, Fisher Creek area Loghill Mesa, Fisher Creek area Larry McPeak, guided by Stan Garvey Zachary Prock & Dustin Braiser Zachary Prock F/adult/none M/subadult/none 12/11/2009 Big Bucktail Canyon Brian Hibbert F/adult/F8 collar & eartag 11/23 to 12/11/2009 N. Fork Cottonwood Creek Gerald Sickels, Jr. F/adult/none 11/23 to 12/11/2009 East of Nucla Gerald Sickels, Jr. F/adult/none 11/23 to 12/11/2009 Pinyon, Cottonwood Creek Gerald Sickels, Jr. M/subadult/yellow eartag 11/23 to 12/11/2009 San Miguel Canyon below Pinyon Gerald Sickels, Jr. M/adult/none 11/23 to 12/11/2009 Mailbox Park Gerald Sickels, Jr. M/adult/none 11/23 to 12/11/2009 Dead Horse Mesa Gerald Sickels, Jr. F/adult/none Late 11/2009 Pinyon Ridge Micah Brogden F/adult/none 124 Ryan Weimer Gary Gleason Reason for releasing the puma given by hunter Did not want to kill a female puma. Outfitter R. Weimer did not want hunter to kill a female puma. Did not want to kill a small male puma. Estimated ~125 lb. Did not want to kill a female. L. McPeak later in same day killed another adult female puma. Hunters will not kill a female puma.* Will not kill a female puma. *These 2 females treed ~4 days apart. One seemed younger than the other, so thought to be different females. But, could have been same puma. Did not want to kill a small male puma. B. Hibbert estimated puma about 1.5 years old. Likes to look at the pumas and train his dogs. Does not want to kill a female puma. Likes to look at the pumas and train his dogs. Does not want to kill a female puma. Likes to look at the pumas and train his dogs. Does not want to kill a female puma. Likes to look at the pumas and train his dogs. Does not want to kill a small male. Wants to kill a big male puma. Likes to look at the pumas and train his dogs. Does not want to kill an average male. Wants to kill a big male puma. Likes to look at the pumas and train his dogs. Does not want to kill an average male. Wants to kill a big male puma. Not interested in killing any puma. Likes to hunt pumas with dogs. �Table 6. Summary of puma capture efforts with dogs from December 15, 2009 to April 30, 2010, Uncompahgre Plateau, Colorado. Month December No. Search Days 10 No. & type of puma tracks founda,b 27 tracks: 12 male, 15 female, 0 cub Tracks ≤1 day old: 5 male, 8 female, 0 cub No. & type of pumas pursued 10 pursuits: 4 males, 6 females , 0 cubs January 20 80 tracks: 24 male, 35 female, 21 cub Tracks ≤1 day old: 11 male, 15 female, 10 cub 23 pursuits: 7 males, 10 females, 6 cubs February 22 77 tracks: 19-20 male, 36-37 female, 20 cub; 1 unknown sex Tracks ≤1 day old: 11 male, 24 female, 12 cub 36 pursuits: 7 males, 17 females, 12 cubs March 23 58 tracks: 16 male, 26 female, 16 cub Tracks ≤1 day old: 7 male, 14 female, 10 cub 18 pursuits: 4 males, 8 females, 6 cubs April 19 No. & I.D. or type of pumas captured, observed, or identified 2 pumas captured 3 times: F3 recaptured (nonfunctioning GPS collar replaced). One adult male puma ~2-3 yr. old captured twice, but not handled due to dangerous trees. In addition, adult F93 associated once with tracks by VHF telemetry (no pursuit with hounds). 6 pumas captured 9 times: M55 recaptured twice; F70 recaptured once; F111, F115, & F116 captured for first time. Then M115 & F116 recaptured. One adult male ~2-3 yr. old was captured, but not handled due to dangerous tree. In addition, 5 adult pumas were associated with tracks 6 times with VHF or GPS telemetry: M55 twice (VHF), F70 (GPS), F72 (GPS), F93 (VHF), F111 (GPS). 10 pumas captured 12 times: F23 recaptured 3 times, but in trees too dangerous for handling to replace her non-functional GPS collar. F28 recaptured in a tree too dangerous for handling to replace her non-functional GPS collar. M32 recaptured (VHF collar replaced). F72 recaptured (non-functional GPS collar replaced). Cubs F106, M107 & F108 recaptured (expandable radiocollars fitted on F106 & F108). M114, M117, F118 captured for the first time. In addition, 7 adult pumas were associated with tracks 9 times via VHF or GPS telemetry: M32 (VHF), F70 (GPS), F95 (VHF), F111 three times (GPS), F113 (VHF), F116 (VHF), F118 (VHF). 3 pumas captured: F96 and M115 recaptured. F119 captured for first time. In addition, 8 pumas were associated with tracks 16 times via VHF telemetry and/or GPS: F3 three times (GPS, VHF & GPS, VHF), M6 twice (VHF), M55 four times (VHF, GPS, VHF & GPS twice), F70 three times (VHF, GPS twice), cub M112 once (VHF), F93 once (VHF), F96 once (GPS), and cub M115 (VHF). 0 pumas captured physically, but F95 identified in one pursuit with VHF telemetry. In addition, 3 adult pumas associated with tracks with VHF telemetry: F93, F104, F118. 24 tracks: 11-12 male, 6 pursuits: 2-3 12-13 female, males, 3-4 females, 0 cub 0 cubs Tracks ≤1 day old: 3-4 male, 6-7 female, 0 cub 86 266 tracks: 93 pursuits: 21 individual pumas were captured 26 times with TOTALS 82-84 male, 24-24 males, aid of dogs. In addition, 16 radio-collared pumas 124-126 female, 44-45 females, were detected 38 times by tracks and identified 57 cub, 24 cubs with VHF and/or GPS telemetry. 1 unknown sex Tracks ≤1 day old: 37-38 male 67-68 female 32 cub a Puma hind-foot tracks with plantar pad widths >50 mm wide are assumed to be male; ≤50 mm are assumed to be female (Logan and Sweanor 2001:399-412). b Researchers also recorded instances when the first puma tracks ≤1 day old were encountered on each search route each day. The count was: 37 tracks of females, including 5 associated with cubs; 21 tracks of males; and 2 tracks of unspecified sex. 125 �Table 7. Adult and subadult pumas captured for the first time, sampled, tagged, and released from December 2009 to April 2010, Uncompahgre Plateau, Colorado. Puma I.D. MA* MB* F111 M112 F113 M114 M115 F116 M117 F118 F119 Sex Estimated Age (mo.) 24-36 24-36 24-27 4.7 36 36 14 36-48 6 18-24 60-72 M M F M F M M F M F F Mass (kg) Unknown Unknown 35 10 47 63 39 49 12 38 46 Capture date 12-16-09 01-03-10 01-01-10 01/23/10 01/26/10 02-27-10 01-13-10 01-20-10 02-05-10 02-25-10 03-25-10 Capture method Dogs Dogs Dogs Cage trap Cage trap Dogs Dogs Dogs Dogs Dogs Dogs Location West Fork Dry Creek Basin East Fork Dry Creek Basin Cushman Canyon Horsefly Canyon (east slope) McKenzie Butte McKenzie Butte San Miguel Canyon San Miguel Canyon San Miguel Canyon Big Bucktail Canyon San Miguel Canyon * Pumas MA and MB were adult males that could not be handled because they climbed dangerous trees. Table 8. Pumas that were captured and observed with aid of dogs, or observed in association with another radio-collared puma, but were not handled at that time for safety reasons, December 2009 to April 2010, Uncompahgre Plateau, Colorado. Puma sex & I.D. Capture date Location Comments MA Age stage or months 24-36 12-16-09 West Fork Dry Creek Basin MB 24-36 01-03-10 East Fork Dry Creek Basin F23 72 02-23-10 San Miguel Canyon F23 72 02-24-10 Big Bucktail Creek F23 72 02-25-10 San Miguel Canyon F28 89 02-01-10 Tomcat Creek Puma climbed dangerous tree, not handled. No noticeable marks. Puma climbed dangerous tree, not handled. This puma obviously larger than MA (above). F23 climbed dangerous tree, not handled to change non-functional GPS collar. F23 climbed dangerous tree, not handled to change non-functional GPS collar. F23 climbed dangerous tree, not handled to change non-functional GPS collar. F28 climbed dangerous tree, not handled to change non-functional GPS collar. F28 was in association with M115, apparently her offspring. 126 �Table 9. Pumas recaptured with dogs, cage traps, or visually observed, November 2009 to May 2010, Uncompahgre Plateau, Colorado. Puma I.D. Recapture Date 12-23-09 Mass (kg) Not weighed Estimated Age (mo.) 101 M32 02-04-10 54 100 Capture Method/ Location Dogs/East Fork Dry Creek Basin Dogs/Dry Creek Basin F3 M55 11-06-09 70 66 Cage trap/Puma Canyon M55 01-07-10 Observed 68 Spring Creek Canyon M55 01-24-10 Observed 68 Linscott Canyon M67 02-24-10 73 31 Dogs/Tomcat Creek F70 01-19-10 Not weighed 63 F72 02-09-10 Observed 47 Dogs/Horsefly Canyon (east slope) Dogs/Loghill Mesa F94 05-13-10 Not weighed 58 F96 03-11-10 43 50 Cage trap/Pinyon Hills west of Happy Canyon Dogs/Happy Canyon F106 02-10-10 20 9 Dogs/Dry Park M107 02-24-10 Observed 9 Dogs/Spring Creek Canyon F108 02-24-10 20 9 Dogs/Spring Creek Canyon M115 01-21-10 Observed 14 Dogs/San Miguel Canyon M115 03-18-10 34 16 Dogs/North Fork Cottonwood Creek F116 01/21/10 Observed 36-48 Dogs/San Miguel Canyon 127 Process Non-functional GPS collar replaced. M32’s old VHF collar was replaced. M55’s old GPS collar was replaced. M55 was wearing a functional GPS collar. No need to handle. M55 was wearing a functional GPS collar. No need to handle. M67 fitted with VHF collar. Offspring of F30, born July 17, 2007. Non-functional GPS collar replaced. F72 wore functional GPS collar, no need to handle. F94’s VHF collar changed to GPS collar. F96’s old GPS collar was replaced. F106 fitted with expandable VHF collar. Offspring of F75, born May 7, 2009. M107 captured with sibling F108, offspring of F94, born May 25, 2009. F108 captured with sibling M107, offspring of F94, born May 25, 2009. F108 fitted with expandable VHF collar. Attempted to capture female puma with M115. Dogs got on M115’s tracks. M115 handled to examine draining wound to left foreleg that occurred about 1-2 weeks prior to this capture; cause unknown. Broken bone detected by palpation. Left ulna was broken (examined later at mortality 08/06/10). F116 wore functional VHF collar, no need to handle. �Table 10. Summary of puma capture efforts with cage traps from September 11, 2009 to May 17, 2010, Uncompahgre Plateau, Colorado.* Month September No. of Sites 2 Carnivore activity & capture effort results Set cage trap with mule deer and predator call box on east rim Roubideau Canyon 09-11-09 to 09-15-09. Adult female puma with 2 large cubs visited 09-13-09; clawed at deer carcass, but did not feed; clawed at call box. Puma family did not return. October 1 A non-collared puma (probably subadult or adult female) visited the fawn mule deer carcass 10-17-10, but did not feed (Reconyx camera photos). A black bear walked ~10 m from the carcass, but did not feed. Mule deer carcasses scavenged by bobcats and magpies. November 2 Cage trap set with catnip oil and K-9 call scent bait and predator call box and stuffed toy rabbit 11-03-09 to 11-06-09. Cage trap closed due to proximity of puma F72 to trap. Puma M55 was recaptured at a mule deer buck he killed 11-06-09. January 2 Set cage trap with mule deer buck killed by male puma 01-04 to 08-10. Male puma was treed by dogs on 01-03-10, but could not be safely handled in East Fork Dry Creek. Puma did not return to its deer kill and cage trap. Bobcat trapper inadvertently captured cub M112 in cage trap on west rim Horsefly Canyon 0123-10. M112 offspring of F70. February 2 A bobcat and Golden Eagle scavenged mule deer carcasses. March 12 Puma F94 and cubs walked with 10 m of a mule deer carcass with predator call box, but did not feed 03-29-10. A male puma walked by mule deer carcass with predator call box, but did not feed 03-18-10. A male puma scraped 2 m from mule deer carcass, but did not feed 03-2310. Puma F96 investigated a predator call box set about 10 m from a mule deer carcass and clawed the call box, but did not feed on the deer. April 6 Puma F94 and cubs M107, F108 consumed a mule deer carcass 04-03 to 07-10. A male puma scavenged a mule deer carcass sometime during 04-05 to 13-10, possibly M55. M55 scavenged from another mule deer carcass on 04-05-10. May 3 Cage trap set 05-13-10 with mule deer doe killed by a female puma in Pinyon Hills; recaptured F94. Tracks indicated a male puma walked ~15 m from 2 cage traps with call boxes and scent lures, but did not go to cage traps to investigate. * We used 21 road-killed mule deer at 17 different sites. Of the road-killed deer baits, 3 of 17 (17.65%) were scavenged by pumas. Table 11. Puma cubs sampled July 2009 to July 2010 on the Uncompahgre Plateau Puma Study area, Colorado. a Cub I.D. Sex Estimated birth datea Estimated age at capture (days) Mass (kg) Mother Estimated age of mother at birth of this litter (mo) M112 M115 M117 M120 M121 P1016b P1017b M M M M M M M August 31, 2009 November 2008 August 2009 June 28, 2010 June 28, 2010 June 12, 2010 June 12, 2010 145 427 183 30 30 39 39 10 39 12 2.5 2.2 2.1 half eaten F70 F28 F119 F3 F3 F72 F72 52 68 66 107 107 51 51 Estimated age of cubs sampled at nurseries is based on the starting date for GPS location and radio-telemetry foci for mothers at nurseries, and development characteristics of cubs caught with mothers without radiocollars or mothers with non-functioning radiocollars. b Cubs P1016 and P1017 were monitored from birth via F72’s GPS data and visual of her nursery to the day of their death; but the cubs were not individually marked. Individual identification of non-marked pumas were designated with P one thousand series numbers (e.g., P1016). On the day we investigated F72’s nursery, male adult puma M32 was at the nursery; he had killed both cubs and half-consumed one about 3 to 6 hours before our arrival. 128 �Table 12. Summary of puma capture efforts with dogs, December 2004 to April 2010, Uncompahgre Plateau, Colorado. Period Dec. 2, 2004 to May 12, 2005 Nov. 21, 2005 to May 26, 2006 Nov. 13, 2006 to May 11, 2007 Nov. 19, 2007 to April 24, 2008 Dec. 9, 2008 to April 30, 2009 Dec. 15, 2009 to April 30, 2010 Track detection effort 109/78 = 1.40 tracks/day 35/78 = 0.45 pursuit/day Puma capture effort 14/78 = 0.18 capture/day Effort to capture an independent puma for the first time 11 pumas captured for first time 11/78 = 0.14 capture/day 78/14 = 5.57 day/capture 14/82 = 0.17 capture/day 78/11 = 7.09 day/capture 149/82 = 1.82 tracks/day 78/35 = 2.23 day/pursuit 43/82 = 0.52 pursuit/day 177/78 to 182/78 = 2.27-2.33 tracks/day 82/43 = 1.91 day/pursuit 45/78 to 47/78 = 0.58-0.60 pursuit/day 82/14 = 5.86 day/capture 22/78 = 0.28 capture/day 78/47 to 78/45 = 1.66-1.73 day/pursuit 49/77 = 0.64 pursuit/day 78/22 = 3.54 day/capture 78/7 = 11.14 day/capture 20/77 = 0.26 capture/day 7 pumas captured for first time 7/77 = 0.09 capture/day 77/20 = 3.85 day/capture 24/71 = 0.34 capture/day 77/7 = 11.00 day/capture 217/77 to 218/77 = 2.82-2.83 tracks/day Pursuit effort 198/71 to 202/71 = 2.79-2.84 tracks/day 77/49 = 1.57 day/pursuit 75/71 to 78/71 = 1.06-1.10 pursuit/day 266/86 = 3.09 tracks/day 71/75 to 71/78 = 0.91-0.95 day/pursuit 93/86 = 1.08 pursuit/day 86/93 = 0.92 day/pursuit 7 pumas captured for first time 7/82 = 0.08 capture/day 82/7 = 11.71 day/capture 7 pumas captured for first time 7/78 = 0.09 capture/day 9 pumas captured for first time 9/71 = 0.13 capture/day 71/24 = 2.96 day/capture 71/9 = 7.89 day/capture 26/86 = 0.30 capture/day 9 pumas captured for first time 9/86 = 0.11 capture/day 86/26 = 3.31 day/capture 86/9 = 9.56 day/capture 129 �Table 13. Individual puma reproduction histories, Uncompahgre Plateau, Colorado, 2005-2010. Consort pairs and estimated agesa Dates pairs Estimated Estimated Estimated Observed consortedb birth datec birth interval gestation number of Female Age Male Age (mo.) (days) cubsd (mo.) (mo.) F2 53 05/28/05 3 F2 67 07/29/06 14.0 2 F2 89 05/19/08 22.0 4 F3 36 08/01/04 1 F3 50 M6 37 06/22-24/05 09/26/05 13.8 93-95 2 F3 62 09/17/06 11.7 3 F3 84 M51 60 03/31/08 07/03/08 21.5 94 3 F3 107 M55 69 03/28-31/10 06/28/10 23.8 89-92 2 F7 67 05/19/05 2 F7 82 08/13/06 14.9 4 F7 106 07/10/08 23.9 3 F8*e 24 06/26/05 2 F8 37 08/13/06 13.4 4 F8 60 M73 49 02/28-29/08 05/29/08 22.5 90-91 2 F16 32 09/22/05 4 F16 52 05/24/07 19.9 4 F16 75 M6 80 01/13-14/09 04/15/09 22.7 91-92 3 F23* 21 05/30/06 3 F23 45 M27 or 78 02/19-25/08 05/23/08 23.8 87-93 3 M29f 107 F24 75 M29 92 04/12-15/07 06/14/07 90-93 4 F25 74 08/01/05 1 F25 94 04/16/07 20.5 1 F25 110 08/19/08 16.1 2 F28* 36 06/09/06 2 F28 48 M29 88 12/27-29/06 03/30/07 11.7 92-93 ≥2 tracks F28 68 11/08 1 F30* 48 M55 34 04/16-20/07 07/17/07 88-92 3 F50 21 07/01/06 1 F54 24 07/01/06 1 F70* 38 M51 60 03/10/08 06/05/08 87 3 F70 52 08/31/09 14.8 3 F72* 28 07/09/08 1 F72 51 06/12/10 23.1 2 F75 32 06/01/07 1 F75 55 M73 61 02/11/09 05/07/09 23.2 93 2 F93 56 08/07 2 F93 90 06/16/10 2 F94* 46 05/27/09 3 F94 60 M55 70 04/15/10 07/15/10 13.3 91 3 F104 110 07/08/10 1 F111* 32 06/16/10 ≥1 tracks F116g 36-48 2009 2 F119 66 08/09 2 a Ages of females were estimated at litter birth dates. Ages of males were estimated around the dates the pairs consorted. b Consort pairs indicate pumas that were observed together based on GPS data or VHF location data. c Estimated birth dates were indicated by GPS data of mothers at nurseries or by back-aging cubs to approximate birth date. d Observed number of cubs do not represent litter sizes as some cubs were observed when they were 5 to 16 months old after postnatal mortality could have occurred in siblings. Only cub tracks were observed with F28. e Asterisk (*) indicates first probable litter of the female, based on nipple characteristics noted at first capture of the female. f A radio-collared, ear-tagged male puma was visually observed with F23 on 2/25/08. Both M27 and M29 wore non-functional GPS collars in that area at the time. g When captured on 1/20/10, puma F116 was in association with 2 large cubs which were not captured. 130 �Table 14. Summary for individual adult puma survival and mortality, December 8, 2004 to July 31, 2010, Uncompahgre Plateau, Colorado. Puma I.D. M1 Monitoring span 12-08-04 to 08-16-06 M4 M5 01-28-05 to 12-28-05 08-01-06 to 02-20-09 M6 02-18-05 to 05-21-10 M27 03-10-06 to 05-07-09 M29 04-14-06 to 02-25-09 M32 M51 04-26-06 to 07-31-10 01-07-07 to 03-20-09 M55 M67 M71 01-21-07 to 07-31-10 08-23-07 to 07-31-10 01-29-08 to 11-12-09 M73 M100 02-21-08 to 07-31-10 03-27-09 to 07-31-09 M114 02-27-10 to 06-23-10 F2 01-07-05 to 08-14-08 F3 F7 01-21-05 to 07-31-10 02-24-05 to 08-03-08 F8 F16 03-21-05 to 07-31-10 10-11-05 to 09-11-09 F23 02-05-06 to 02-25-10 F24 F25 01-17-06 to 09-03-08 02-08-06 to 09-04-09 Status: Alive/Lost contact/Dead; Cause of death Dead. Lost contact− failed GPS/VHF collar. M1 ranged principally north of the study area as far as Unaweep Canyon. M1 was killed by a puma hunter on 01-02-10 west of Bang’s Canyon, north of Unaweep Canyon, GMU 40. M1 was about 97 months old at death. Dead; killed by a male puma. Estimated age at death 37−45 months. Dead. Born on study area; offspring of F3. M5 was independent of F3 by 13 months old, and dispersed from his natal area at about 14 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months (protected from hunting mortality in buffer area) and ranged into the eastern edge of Utah (vulnerable to hunting). Killed by a puma hunter on 02-20-09 in Beaver Creek, Utah at age 54 months. Dead. M6 was struck and killed by a vehicle on highway 550 south of Colona, CO on 05-21-10. M6 was about 99 months old at death. Dead. Lost contact− failed GPS/VHF collar. Recaptured 12-02-07 & 01-22-08 by puma hunter/outfitter north of the study area. Possibly visually observed on study area with F23 on 02-25-08. Recaptured by a puma hunter/outfitter 12-11-08 & 12-28-08 north of the study area. Photographed by a trail camera on the study area (Big Bucktail Canyon) on 5 occasions: 03-27-09, 04-02-09, 04-15-09, 04-24-09, & 05-07-09. M27 was killed by a puma hunter on 12-09-09 in the North Fork Mesa Creek, Uncompahgre Plateau, GMU 61 North. M27 was about 100 months old at death. Dead. Lost contact− failed GPS/VHF collar. Possibly visually observed on study area with F23 on 02-25-08. Recaptured on study area 02-25-09, but could not be safely handled to change faulty GPS collar. M29 was killed by a puma hunter on 11-16-09 in Beaver Canyon, GMU 70 East. M29 was about 121 months old at death. Alive. Dead. Lost contact− failed GPS/VHF collar after 03-20-09. Killed by a puma hunter on 12-11-09 in Shavano Valley, Uncompahgre Plateau study area. M51 was about 77 months old at death. Alive. Alive. M67 is offspring of F30. Dead. Lost contact– M71 shed his VHF collar with an expansion link on about 11-12-09. He was killed by a puma hunter on 12-09-09 on the west rim of Spring Creek Canyon, Uncompahgre Plateau study area. M71 was about 47 months old at death. Alive. Dead. M100 was killed by a puma hunter on 01-16-10 in Naturita Canyon, GMU 70 East. M100 was about 63 months old at death. Lost contact– after 06-23-10. VHF collar may have failed or puma dispersed. Dead; killed by another puma (sex of puma unknown; male suspected) 08-14-08. F2 was about 92 months old at death. Lost contact− failed GPS/VHF collar. Dead. Killed by U.S. Wildlife Services agent 08-03-08 for predator control of depredation on domestic sheep. F7 was about 107 months old at death. Alive. Dead. F16 was struck and killed by a vehicle on Ouray County Road 1 southwest of Colona, CO on 09-11-09. F16 was about 80 months old at death. Lost radio contact after12-02-09. Recaptured F23 on the study area 02-25-10, but could not be handled to replace non-functional GPS collar. Lost radio contact after 09-03-08− failed GPS/VHF collar. Lost radio contact after 09-04-09– failed GPS/VHF collar. 131 �Puma I.D. F28 Monitoring span 03-23-06 to 02-01-10 F30 04-15-06 to 07-29-08 F50 12-14-06 to 03-26-07 F54 01-12-07 to 08-18-07 F70 F72 F75 01-14-08 to 07-31-10 02-12-08 to 07-31-10 03-26-08 to 02-10-10 F93 F94 F95 F96 F104 F110 12-05-08 to 07-31-10 12-19-08 to 07-31-10 08-01-09 to 07-31-10 01-28-09 to 07-31-10 05-21-09 to 07-31-10 09-21-09 to 02-25-10 F111 F113 01-01-10 to 07-31-10 01-26-10 to 06-06-10 F116 F118 F119 01-20-10 to 07-31-10 02-25-10 to 07-31-10 03-25-10 to 07-31-10 Table 14 continued. Status: Alive/Lost contact/Dead; Cause of death Lost radio contact after 09-25-07− failed GPS/VHF collar. Recaptured on the study area 02-01-10, but could not be handled to replace nonfunctional GPS/VHF collar. Dead. Killed by another puma (sex of puma unknown) 07-29-08. F30 was about 60 months old at death. Dead of natural causes 03-26-07; probably injury or illness-related; exact agent unknown. F50 was about 30 months old at death. Dead; killed by a male puma while in direct competition for prey (i.e., mule deer fawn) 08-18-07. F54 was about 49 months old at death. Alive. Alive. Lost radio contact after 09-29-09– failed GPS/VHF collar. F75 in association with her cubs M105 and F106 when F106 was recaptured on 02-10-10 on the study area. Alive. Alive. Alive. Alive. Alive. Dead. Killed by a puma hunter on 02-25-10 in GMU 70 East. F110 was about 41 months old at death. Alive. Dead. F113 died 06-06-10 of injuries consistent with being struck by a vehicle. GPS data indicated that F113 had crossed highway 550 and roads on Loghill Mesa north of Ridgway 24-30 hours before she died in McKenzie Creek. F113 was about 42 months old at death. Alive. Alive. Alive. 132 �Table 15. Preliminary estimated survival rates (S) of adult-age pumas during the 4 years in the reference period (i.e., the study area is closed to puma hunting) and 1 year in the treatment period, Uncompahgre Plateau, Colorado. Survival rates of pumas estimated with the Kaplan-Meier procedure to staggered entry of animals (Pollock et al. 1989). Survival rates are for an annual survival period defined as the biological year (August 1 to July 31). Survival rates were estimated only for periods when n ≥ 5 individual pumas were monitored in the interval. Puma survival in the reference period pertained only to pumas that died of natural causes. Pumas that were killed by people in the reference period, a non-natural cause (i.e., F7 for depredation control 8/3/2008 and M5 killed by a puma hunter off the protected study area and buffer zone 2/20/2009) were right censored. In the treatment period all sources of natural and human-caused mortality are considered in the survival estimates. Period of interest S 1.000 Females SE 0.0000 S 0.667a Males SE 0.2222a n 6a Reference Annual 8/1/2005 to 7/31/2006 0.909 0.0867 11 1.000 0.0000 5 Reference Annual 8/1/2006 to 7/31/2007 0.831 0.0986 14 1.000 0.0000 7 Reference Annual 8/1/2007 to 7/31/2008 0.875 0.1031 13 1.000 0.0000 8 Reference Annual 8/1/2008 to 7/31/2009 0.784 0.1011 19 0.667 0.1924 8 Treatment Annual 8/1/2009 to 7/31/2010 Treatment Annualb 0.333b 0.1361b 12b 8/1/2009 to 7/31/2010 With mortalities of all marked adult males a Adult male annual S 2005 to 2006 is probably underestimated with poor precision because 3 of the 6 pumas were GPS/VHF-monitored for 4 to 5 months at the end of the interval; 1 of 6 adult males died. b This second estimate of adult male puma survival includes 5 males that had non-functional (4) or shed (1) radiocollars. All adult males with non-functional or shed radiocollars in this study survived into treatment year 1 (TY1), which was expected considering adult male survival in 3 previous years. All 5 of those adult males were detected and killed by hunters in TY1. 133 n 10 �Table 16. Summary of subadult puma survival and mortality, December 2004 to July 2010, Uncompahgre Plateau, Colorado. Puma I.D. M5 Monitoring span 09-16-05 to 06-30-06 No. days 308 M11 06-21-06 to 12-02-07 529 F23 01-04-06 to 02-04-06 04-19-06 to 04-26-06 31 M49 03-26-07 to 10-01-07 189 F52 01-10-07 to 05-15-07 125 F66 08-23-07 to 11-05-07 11-25-08 to 06-03-09 74 M31 7 190 M69 01-11-08 to 04-07-08 87 F95 12-29-08 to 07-31-09 214 M99 02-27-09 to 04-22-09 54 Status M5 was offspring of F3, born August 2004. Independent and dispersed from natal area at 13 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months (protected from hunting mortality in buffer area) and ranged into the eastern edge of Utah (vulnerable to hunting). Killed by a puma hunter on 02-20-09 in Beaver Creek, Utah at about 54 months old. M11 was offspring of F2, born May 2005. Independent at 13 months old. Dispersed from natal area at 14 months old. Moved to Dolores River valley, CO, by 12-14-06. Killed by a puma hunter on 12-02-07 when about 30 months old. Alive. Captured on the study area when about 17 months old. Survived to adult stage; gave birth to first litter at about 21 months old. M31’s estimated age at capture was 20 months. Dispersed to northern New Mexico and was killed by a puma hunter on 12-11-08 in Middle Ponil Creek, Cimarron Range. He was about 52 months old. M49 was offspring of F50, born July 2006. Orphaned at about 9 months old, when F50 died of natural causes. Dispersed from his natal area at about 10 months old and ranged on the northeast slope of the Uncompahgre Plateau. When M49 was about 15 months old, he shed his expandable radiocollar on about 10-01-07 at a yearling cow elk kill on the northeast slope of the Uncompahgre Plateau. He was killed by a puma hunter in Blue Creek in the protected buffer zone north of the study area on 01-24-09; he was about 29 months old, a young adult. F52 dispersed from study area as a subadult by Jan. 16, 2007. F52’s last VHF aerial location was Crystal Creek, a tributary of the Gunnison River east of the Black Canyon 05-15-07. She was treed by puma hunters on 12-29-08 on east Huntsman Mesa, southeast of Powderhorn, CO. She was about 41-43 months old and could have been in her adultstage home range. GPS collar nonfunctional. F66 was offspring of F30, born July 2007. Lost contact; her cub collar quit after 11-05-07. Recaptured as an independent subadult on her natal area 11-25-08 when 16 months old. F30 was killed by a puma when F66 was 12 months old, within the age range of normal independence. F66 died of injuries to internal organs that caused massive bleeding attributed to trampling by an elk or mule deer on about 05-28-09 when she was 23 months old. Her range partially overlapped her natal area. M69 was captured on the study area when about 14-18 months old. Emigrated from the study area as subadult by 03-19-08. Last VHF aerial location was southwest of Waterdog Peak, east side of Uncompahgre River Valley on 04-07-08. M69 was killed by a puma hunter on 11-0608 in Pass Creek in the Snowy Range, WY when he was 24 to 28 months old. Alive. F95 is the offspring of F93, born about August 2007. She became an independent subadult by about 18 months old (02-11-09 aerial location) and an adult by about 24 month old (Aug. 2009). F95 established an adult home range adjacent to and overlapping the northern portion of her natal area. M99 died on unknown causes; but, possibly killed by another puma (holes in skull) in Jan. 2010 when he was about 16 months old. His radiocollar quit after 54 days. 134 �Puma I.D. M115 Monitoring span 01-13-10 to 07-21-10 No. days 189 Table 16 continued. Status M115 was offspring of F28, born in Nov. 2008. He was about 14 months old when first captured on Jan. 13, 2010. When he was recaptured on Mar. 18, 2010, he had previously suffered a broken left ulna. M115 was probably independent by July15, 2010 when he was located outside of his natal area on a probably dispersal move. M115 died on about July 21, 2010 apparently from complications of his broken left foreleg; possibly not allowing him to kill prey sufficiently for survival. M115 was about 20 months old at death. 135 �Table 17. Records of pumas that dispersed from the Uncompahgre Plateau study area, December 2004 to July 2010. Puma I.D. 1st capture date on study area 1st capture location→kill or resight location (UTM, NAD27) M5 02-04-05 13S,240577E, 4251037N→ 12S,665853Ex 4277125N M11 06-27-05 13S,248278E, 4239858N→ 12S,741882Ex 4161575N 84.8 M31 04-19-06 329.8 M39 09-11-06 M43 09-15-06 12S,746919E, 4225441N→ 13S,500000Ex 4050000N 12S,724270E, 4243610N→ 12S,709889E, 4313490N 12S,760177E, 4242995N→ 12S,739859E, 4308557N M48 10-18-06 52.0 M49 12-05-06 12S,756676E, 4247777N→ 12S,704982E, 4248998N 12S,757241E, 4258259N→ 12S,693350E, 4274559N M58 06-27-07 73.2 M65 08-17-07 M68 08-23-07 13S,258543E, 4238071N→ 13S,274670E, 4309488N 12S,738144E, 4233628N→ 12S,684084E, 4314200N 13S,257371E, 4235231N→ 12S,711262E, 4198681N Estimated linear dispersal distance (km)* 102.2 71.3 68.6 66.1 Puma Information M5 was offspring of F3, born August 2004. Independent and dispersed from natal area at 13 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months (protected from hunting mortality in buffer area) and ranged into the eastern edge of Utah (vulnerable to hunting). Killed by a puma hunter on 02-20-09 in Beaver Creek, Utah at about 54 months old. M11 was offspring of F2, born May 2005. Shed expandable radiocollar 10-24 to 11-08-05. Recaptured and re-collared 04-0206. Independent at 13 months old. Dispersed from natal area at 14 months old. Moved to Dolores River valley, CO, by 12-14-06. Killed by a puma hunter on 12-02-07 when about 30 months old. M31’s estimated age at capture was 20 months. Dispersed to northern New Mexico and was killed by a puma hunter on 12-1108 in Middle Ponil Creek, Cimarron Range. He was about 52 months old. M39 was offspring of F8, born August 2006. M39 was killed by a puma hunter in Bangs Canyon, GMU 40 on 03-12-10 when he was 43 months old. M43 was offspring of F7, born August 2006. He shed the expandable radiocollar 11-7 to 17-06, after which direct contact was lost. M43 was killed by a puma hunter 01-28-09 in Deer Creek, west slope of Grand Mesa, CO when he was 29 months old. M48 was the offspring of F3, born September 2006. M48 was killed by a puma hunter in Tabeguache Creek, GMU 61 North on 12-27-09 when he was 39 months old. M49 was offspring of F50, born July 2006. Orphaned at about 9 months old, when F50 died of natural causes. Dispersed from his natal area at about 10 months old and ranged on the northeast slope of the Uncompahgre Plateau. When M49 was about 15 months old, he shed his expandable radiocollar on about 10-01-07 at a yearling cow elk kill on the northeast slope of the Uncompahgre Plateau. He was killed by a puma hunter in Blue Creek in the protected buffer zone north of the study area on 0124-09; he was about 29 months old. M58 was offspring of F16, born May 2007. M58 was killed by a puma hunter on 12-27-09 in the North Fork of the Gunnison River north of Paonia, GMU 521; he was 31 months old. 97.0 M65 was offspring of F24, born July 2007. M65 was killed by a U.S. Wildlife Service agent for depredation on llamas in the Little Dolores River on 11-07-09. M65 was 28 months old. 80.7 M68 was offspring of F30, born July 2007. He was orphaned at 12 months old when his mother was killed by a puma. He was killed by a puma hunter in the Disappointment Valley in southwest CO on 12-30-08; he was 17 months old. 136 �Puma I.D. 1st capture date on study area 1st capture location→kill or resight location (UTM, NAD27) M69 01-11-08 13S,248191E, 4246810N→ 13T,378900E, 4591990N M82 07-05-08 F52 01-10-07 12S,726901E, 4243463N→ 13S,255316E, 4216768N 13S,258058E, 4236260N→ 13S,319217E, 4240467N Table 17 continued. Estimated Puma Information linear dispersal distance (km)* 369.6 M69 was captured on the study area when about 14-18 months old. Emigrated from the study area as subadult by 03-19-08. Last VHF aerial location was southwest of Waterdog Peak, east side of Uncompahgre River Valley on 04-07-08. M69 was killed by a puma hunter on 11-06-08 in Pass Creek in the Snowy Range, WY when he was 24 to 28 months old. 60.5 M82 was offspring of F8, born May 2008. M82 was killed by a hunter on 12-10-09 in the Beaver Creek fork of East Dallas Creek, GMU 65. M82 was 19 months old. 61.1 F52 was captured on the study area when about 18-20 months old. Dispersed from study area as a subadult by Jan. 16, 2007. F52’s last VHF aerial location was Crystal Creek, a tributary of the Gunnison River east of the Black Canyon 05-15-07. She was treed by puma hunters on 12-29-08 on east Huntsman Mesa, southeast of Powderhorn, CO. She was about 41-43 months old . F52 was treed again by puma hunters on about 12-16-09 south of Powderhorn: 13S,319480E,4233219N. F52 was about 53-55 months old. This suggests that F52 has an adult home range in that area. *Estimated linear dispersal distance (km) from initial capture site on Uncompahgre Plateau study area to hunter kill or recapture site. 137 �Table 18. Recorded deaths of non-marked and marked pumas struck by vehicles and other unusual causes, in chronological order, on the Uncompahgre Plateau puma study area, Colorado, from 2004 to 2010. Puma sex & ID if marked M Estimated age (mo) Date recorded Cause of death General physical condition Location & UTM NAD27 12 09-24-04 Good F 49 07-28-05 Vehicle collision Vehicle collision Pleasant Valley, County Road 24 13S,252870E,4227520N Highway 62 east of Dallas divide 13S,250000E,4222500N F17a 11 08-18-06 F 18-24 11-06-06 F 6 01-30-07 F 36 09-16-08 M 12-24 08-13-08 F61a 18 11-13-08 F 12 08-10-09 F16b 80 09-11-09 M6b 99 05-21-0 F113b 42 06-06-10 Vehicle collision Vehicle collision Vehicle collision Asphyxia, lodged in fork of tree Vehicle collision Vehicle collision Vehicle collision Vehicle collision Vehicle collision Vehicle collision Good Not pregnant or lactating Good Good Good Unknown, decomposed Good Good Good Good Good Good Not pregnant or lactating M 24 08-25-10 Vehicle Excellent P1018c collision a Subadult marked (i.e., tattoos, eartags), but not radio-collared. b Adult GPS/VHF-collared pumas. c Non-marked puma with P one-thousand number designation. 138 Highway 550 south of Colona 13S,257602E,4242185N Highway 550 east of Ridgway State Park 13S,259843E,4235985N Highway 62 west of Dallas divide 12S,762286Ex4218992N Davis Point, Roubideau Canyon 12S, 743718E,4255277N Highway 145 west of Placerville 13S,756490E,4212336N Highway 550 east of Ridgway State Park 13S,259843E,4235985N Highway 145 east of Norwood 12S,745739E,4222548N Ouray County Road 1 13S,253733E,4240060N Highway 550 south of Colona 13S,258610E,4236805N F113 crossed Highway 550 and roads on Loghill Mesa 24-30 hours before she died in McKenzie Creek 13S,257272E,4238435N Highway 62 Leopard Creek 12S,237747E,4220330N �Table 19. GPS- and VHF-collared pumas with functioning collars using two camera grids (Area 1Loghill, Area 2- Delores Creek to Spring Creek) during August 21 to December 7, 2009 (i.e., 108 days), Uncompahgre Plateau, Colorado. Area 1- Loghill Puma Sex Estimated Collar & Number of Capture rate per day for ID Age (mo.) data type detections by primary camera configuration Aug.-Dec. cameras in grid (photo per puma/no. days) 2009 primary/alternate cameraa F16 female 79-83 GPS 0 0 F72 female 41-45 GPS 4/0 4/108 = 0.04 M6 male 90-94 VHF 6/0 6/108 = 0.06 M55 male 50-54 GPS 7/0 7/108 = 0.06 Area 2- Delores Creek to Spring Creek Puma Sex Estimated Collar & Number of Capture rate per day for ID Age (mo.) data type detections by primary camera configuration Aug.-Nov. cameras in grid (photo per puma/no. days) 2009 primary/alternate camerab F70 female 52-56 GPS 0 0 F94 female 49-53 VHF 3/1 3/108 = 0.03 F96 female 43-49 GPS 2/0 2/108 = 0.02 M32 male 96-100 VHF 4/0 4/108 = 0.04 M55 male 50-54 GPS 19/5 19/108 = 0.18 a Aug. 21 to Nov. 2 (74 days) to detect 3 of 4 adult pumas with functioning collars for first time. Aug. 21 to Oct. 20 (61 days) to detect 4 of 5 adult pumas with functioning collars for first time. It took 88 days (Aug. 21 to Nov. 16) to also detect 2 adult pumas with non-functioning collars. b 139 �Table 20. Numbers of GPS locations and spans of monitoring for pumas captured on the Uncompahgre Plateau, Colorado, December 2004 to July 2010. Puma I.D. M1 M4 M6 M27 M29 M51 M55 M100 F2 F3 F7 F8 F16 F23 Sex Age stage Dates monitored a M M M M M M M M F F F F F F No. locations adult 12-08-04 to 07-20-06 1,797 adult 01-28-05 to 01-14-06 958 adult 02-18-05 to 05-14-08 1,035 adult 03-12-06 to 06-21-06 313 adult 04-14-06 to 01-01-08 1,599 adult 01-07-07 to 07-15-08 1,643 adult 01-21-07 to 08-09-10 3,226 adult 03-27-09 to 01-16-10 923 adult 01-07-05 to 08-14-08 3,516 adult 01-21-05 to 05-14-08 3,344 adult 02-24-05 to 08-03-08 3,922 adult 03-21-05 to 10-10-06 1,541 adult 10-12-05 to 09-10-09 3,801 subadult, 01-04-06 to 02-04-06 113 adult 02-05-06 to 09-04-09 2,281 F24 F adult 01-17-06 to 07-25-07 1,812 F25 F adult 02-09-06 to 06-26-09 3,398 F28 F adult 03-24-06 to 08-15-07 1,499 F30 F adult 03-30-07 to 02-22-08 1,057 F50 F adult 12-14-06 to 03-26-07 352 F52 F subadult 01-10-07 to 05-08-07 383 F54 F adult 01-12-07 to 08-18-08 723 F70 F adult 01-14-08 to 07-01-10 2,429 F72 F adult 02-12-08 to 07-07-10 2,842 F75 F adult 03-26-08 to 06-03-09 1,112 F96 F adult 01-28-09 to 08-08-10 1,061 F104 F adult 05-29-09 to 08-09-10 1,349 F111 F adult 01-01-10 to 08-02-10 488 F113 F adult 01-27-10 to 06-06-10 445 a GPS collars on pumas were remotely downloaded at approximately 1-month intervals, except during winter 20082009 to summer 2009 due to shortage of technicians during hiring freeze to assist in airplane flights to obtain downloads and to capture pumas to replace GPS collars (lengthening the download interval saved battery power). The last date in Dates monitored includes last location from the last GPS data download acquired for an individual puma. 140 �GOAL: Strategies, Information, & Tools for Managing Healthy, Self-sustaining Puma Populations in Colorado Puma Population Puma Habitat Effects of Harvest & Other Mortality Movements & Corridors Population Dynamics: Sex & Age, Vital Rates, Mortality, Population Growth Rates Vulnerability to Harvest Human Development Habitat Use Effects of Translocation Methods for Monitoring Populations Domestic Animals Puma― Human Relationships Deer, Elk, Other Natural Prey & Species of Concern Effects of Predation Effects of Evasive Conditioning Map Prey Distribution Habitat Models Population Models Puma Prey Puma―Prey Relationships Models Habitat Maps Figure. 1. An ecologically-based conceptual model of the Colorado Puma Research Program that provides the contextual framework for this and proposed puma research in Colorado. Gray-shaded shapes identify areas of research addressed by this puma research on the Uncompahgre Plateau for the puma management goal in Colorado (at top). 141 �Figure 2. The puma study area on the southern half of the Uncompahgre Plateau, Colorado (shaded in gray) comprising the southern portions of Game Management Units (GMUs) 61 and 62 and a northern portion of GMU 70. Expected No. Independent Pumas ~ 70 ml .i-0 "'E 40 :, "- 0 z .,n 20 In 0 RY4 T\'1 T\'2 T'/3 Year Figure 3. Expected (i.e., modeled) number of independent pumas on the Uncompahgre Plateau Study area after the harvest of 14.5% and 21.8% of independent pumas observed in the 2009-10 hunting season. The 14.5% harvest rate represents 8 independent pumas (3 females, 5 males) killed inside the study area. The 21.8% harvest represents 12 independent pumas (4 females, 8 males), including 4 pumas (1 female, 3 males) killed outside of the study area in addition to 8 killed inside the study area. The projected lines represent the expected population trends resulting from the observed harvest rates and sex structure. 142 �Age structure of independent pumas in November 2009 at beginning ofthe puma hunting season in Treatment Year 1, Uncompahgre Plateau, Colorado. 4 "' Ill E 3 ::J '.:: 2 0 1 0 z >--- - - - f- II 0 I ■ Female I ■ Male lto 2 >2 to >3 to >4 to >5 to >6 to >7 to >8 to >9 to 10+ 3 4 5 6 7 8 9 10 Age {years) Figure 4. Estimated age structure of independent pumas in November 2009 at the beginning of the puma hunting season in Treatment Year 1 (TY1) on the Uncompahgre Plateau, Colorado. All these pumas were captured and sampled by researchers or harvested by hunters and examined by researchers. Mean ± SD of female and male ages, respectively: 4.55 ± 2.11 yr. (54.63 ± 25.29 mo.), n = 19; 5.48 ± 2.57 yr. (65.71 ± 30.88 mo.), n = 14. Puma births, Uncompahgre Plateau, Colorado 10 9 8 7 6 5 4 3 2 - 1 11 0 Jan. I I Feb. M ar. Apr. - - - ~ f- M ay June July ■ Births 2005-2010 IT H1 Aug. Sep. Oct. 11 Nov. Dec. ■ Bi rths 1982-1987 Figure 5. Puma births (black bars) detected by month during 2005 to 2010 (n = 34 litters of 17 females; 32 of the litters were examined at nurseries when cubs were 26-42 days old and 2 litters confirmed by tracks of ≥1 cubs following GPS-collared mothers F28 and F111 when cubs were ≤42 days old). Also shown (gray bars) are results of the earlier effort by Anderson et al. (1992:48; 1982 to 1987, n = 10 litters of 8 females, examined when cubs were <1 to 8 months old), Uncompahgre Plateau, Colorado. 143 �2 AREAS FOR CAMERA GRIDS Montrose N A 0 2.5 5 1O Kilometers Figure 6. Layout of 2 camera grids on the east slope of the Uncompaghre Plateau Puma Study Area. Each grid was 80 square kilometers in size and contained 20 cells which were each 4 square kilometers. Area 1 was the south grid that covered Loghill Mesa to upper Horsefly Canyon. Area 2 was the north grid that covered from Dolores Canyon to Spring Creek Canyon. 144 �Appendix A. Summary of individual puma cub survival and mortality, 2005 to 2010, Uncompahgre Plateau, Colorado. Puma I.D. M5 Estimated Age at capture (days) 183 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 02-04-05 to 04-07-08 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death ~8-1-04 F9 31 5-28-05 F10 31 5-28-05 ~1,345 F3 06-27-05 to 4-19-06 06-27-05 to 11-20-05― 12-29-05 06-27-05 to 12-2-07 326-333 Survived to subadult stage by 09-16-05; independent at ~13 mo. old. Dispersed from natal area by 09-29-05 at 14 mo. old. Established territory on NW U.P. Killed by hunter in Beaver Creek, UT 02-20-09 at 4.5 years old. Lost contact― shed radiocollar 04-19-06 to 04-26-06. Lost contact― shed radiocollar 08-10-05; last tracks of F10 with mother F2 & siblings F9 & M11 observed 11-20-05. F10 disappeared by 12-30-05. Survived to subadult stage by 06-21-06, independent at 13 mo. old. Dispersed from natal area by 07-11-06 at 14 mo. old. Killed by a hunter in SW CO 12-2-07 at 918 days (30 mo.) old Lost contact― shed radiocollar 07-28-05―08-01-05. Tracks of F12 found in association with mother F7 on 1208-05. F12 disappeared by 01-27-06 when she was not visually observed with F7, and her tracks were not seen in association with F7’s tracks. Dead; killed and eaten by a puma (sex unspecified) about 828-05. Lost contact― shed radiocollar 01-20-06 to 01-25-06. Tracks of F14 were observed with tracks of mother F8 & sibling M15 on 02-07-06. Disappeared by 03-11-06, only tracks of F8 & M15 were found. Lost contact― shed radiocollar 06-06-06 to 06-14-06. F2 M11 31 5-28-05 F12 42 5-19-05 07-01-05 to 12-08-05― 01-26-06 F13 42 5-19-05 F14 26 6-26-05 07-01-05 to 08-28-05 07-22-05 to 02-07-06― 03-10-06 M15 26 6-26-05 F17 34 9-22-05 F16 308-314 Dead. Lost contact― shed radiocollar 06-06-06 to 06-14-06. Killed by a car on highway 550 on 08-18-06. Probably dependent on F16. Dead; probably killed by another puma. Multiple bite wounds to skull. 10 mo. old. Lost contact― shed radiocollar 07-27-06 to 08-02-06. F18 34 9-22-05 M19 34 9-22-05 M20 34 9-22-05 244-245 Lost contact― shed radiocollar 05-24-06―05-25-06. F16 F21 37 9-26-05 Lost contact; radiocollar quit. Last aerial location 8-16-06, live signal. F3 176-215 918 203-252 101 226-257 07-22-05 to 06-06 to 14-06 10-26-05 to 08-18-06 345-353 10-26-05 to 07-20 to 27-06 10-26-05 to 07-27 to 08-02-06 10-26-05 to 05-24-06 11-02-05 to 08-16-06 301-308 330 324 145 Mother I.D. F2 F2 F7 F7 F8 F8 F16 F16 �Appendix A continued Puma I.D. Estimated Age at capture (days) M22 37 9-26-05 M26 183 8-1-05 F33 31 5-30-06 F34 31 5-30-06 F35 31 5-30-06 F36 29 6-9-06 M37 29 6-9-06 M38 41 7-29-06 M39 29 8-13-06 Est. Birth date F40 29 8-13-06 F41 29 8-13-06 M42 29 8-13-06 M43 33 8-13-06 Est. survival span from 1st capture to fate or last monitor date 11-02-05 to 12-21-05― 12-22-05 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 86-87 Dead; killed and eaten by male puma 12-21-05―12-22-05. F3 02-08-06 to 03-21 to 24-06 06-30-06 to 07-31-06 06-30-06 to 07-31-06 ~232-235 Lost contact― shed radiocollar 03-21-06―03-24-06. F25 F23 06-30-06 to 07-07-06 07-08-06 to 07-28-06 07-08-06 to 07-28-06 09-08-06 to 07-16 to 17-07 38 Dead. Probably killed and eaten by a male puma 08-01 to 03-06. GPS data on M29 indicate he was not involved. Dead. Probably killed and eaten by a male puma 08-01 to 03-06. GPS data on M29 indicate he was not involved. Dead; research-related fatality.a Dead. Killed and eaten by a male puma 08-22-06. GPS data on M29 indicate he was not involved. Dead. Killed and eaten by a male puma 08-22-06. GPS data on M29 indicate he was not involved. Lost contact― shed radiocollar found 03-06-07. Photo (trail camera in McKenzie Cr.) of M38 & Unm. F sibling with F2 on 07-16 to 17-07 at 352-353 days old. F28 Lost contact― shed radiocollar by 09-20-06, but seen alive on that date. Tracks of 2 cubs following F8 on 04-25-07. Survived to adult stage; dispersed from natal area. Killed by a puma hunter 03-12-10 in GMU 40 when 43 months old. Lost contact― shed radiocollar by 09-20-06, but seen alive on that date. Tracks of 2 cubs following F8 on 04-25-07. F8 Assumed dead. Lost Contact― shed radiocollar or died (blood on collar) between 10-05-06 (last live signal) & 1013-06 (collar found). Dead; research-related fatality.b F8 Lost contact− shed radiocollar by 11-7 to 17-06. Treed 0301-07. Killed by a puma hunter 01-28-09 in Deer Creek, west slope of Grand Mesa, CO at 29 months old. Survived to adult stage; dispersed from natal area. Killed by a puma hunter 01-28-09 in GMU 41 when 29 months old. F7 09-11-06 to 09-20-06 to 04-25-07 09-11-06 to 09-20-06 to 04-25-07 09-11-06 to 10-05-06 09-11-06 to 11-27-06 09-15-06 03-01-07 63-65 63-65 74 74 352-353 9 255 9 255 53-61 106 200 146 Mother I.D. F23 F23 F28 F2 F8 F8 �Appendix A continued Puma I.D. Estimated Age at capture (days) M44 33 Est. Birth date 8-13-06 Est. survival span from 1st capture to fate or last monitor date 09-15-06 to 02-14-07 Age to last monitor date alive or at death (days, birth to fate) 479 F45 33 8-13-06 09-15-06 to 5-20 to 23-07 280-283 M46 31 9-17-06 10-18-06 to 12-15-06 89 360 M47 M48 M49 31 31 153 9-17-06 9-17-06 7-1-06 10-18-06 to 12-15-06 to 09-12-07 10-18-06 to 12-15-06 to 09-12-07 183 7-1-06 360 89 360 12-05-06 to 07-31-07 to 01-01-07 F53 89 01-12-07 to 02-23-07 ~456 42 ~428 subad. 147 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Lost contact− shed radiocollar by 10-27-06. Treed, visually observed 02-14-07; sibling (?) M56 also captured, sampled, & marked for 1st time. Killed by Wildlife Services for depredation control on 12-05-07, for killing 4 domestic sheep. He was still dependent on F7. Dead. Multiple puncture wounds on braincase― parietal & occipital regions; consistent with bites from coyote. F45 switched families, moving from F7 to F2 about 12-19 to 2006. Last date F45 was with F2 was 04-17-07. Lost contact― shed radiocollar. Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-1207 in Puma Canyon. Lost contact― shed radiocollar. Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-1207 in Puma Canyon. Lost contact― shed radiocollar. Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-1207 in Puma Canyon. Survived to adult stage; dispersed from natal area. Killed by a puma hunter 12-27-09 in GMU 61 when 39 months old. M49 was orphaned when his mother died on about 03-2607; he was ~268 days old. M49 dispersed from natal area and onto NE slope of U.P. Shed radiocollar at a yearling cow elk kill about 10-01-07; he was ~428 days old. Killed by a puma hunter in Blue Creek, northwest Uncompahgre Plateau (GMU 61 N) 01-24-09 when ~29 months old. Lost contact― shed radiocollar 2-23-07. F53 visually observed by P. & F. Star, on 9-2-07, when F53 was ~14 months old and an independent subadult. Mother I.D. F7 F7 F3 F3 F3 F50 F54 �Appendix A continued Puma I.D. Estimated Age at capture (days) M56c 183 ~8-13-06 F57 35 4-16-07 M58 34 5-24-07 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 02-14-07 to 03-01-07 05-21-07 to 06-06-07 06-27-07 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. 200 Lost contact― shed radiocollar 2-27-07. M56 observed 0301-07. Lost contact― shed radiocollar 06-07-07. Live mode 06-0607. Not radio-collared. Tracks of 3 cubs observed with F16’s tracks on 04-12-08, McKenzie Butte-Pinon Ridge Pass. 3 cubs observed with F16 on 08-08-08 by B. & T. Traegde. Survived to adult stage; dispersed from natal area. Killed by a puma hunter 12-27-09 in GMU 521 when 31 months old. Alive. Observed alive 11-20-07 with F16, but without siblings M58 & F61. Tracks of 3 cubs observed with F16’s tracks on 04-12-08, McKenzie Butte-Pinon Ridge Pass. 3 cubs observed with F16 on 08-08-08 by B. & T. Traegde. F7 (?) Dead; research-related mortality.d F16 Radiocollar malfunction. Tracks of 3 cubs observed with F16’s tracks on 04-12-08, McKenzie Butte-Pinon Ridge Pass. 3 cubs observed with F16 on 08-08-08 by B. & T. Traegde. Dead. Died probably as independent subadult at 538 days old; struck by car on Hwy 550 mi. marker 111 N. of Ridgway, CO, euthanized by gunshot on 11/13/08. Not radio-collared. Not radio-collared. Not radio-collared. Two out of potential of 4 of F24’s male cubs were visually observed with her on 4/1/08. Assume that 2 male cubs died before the age of 8.5 mo. Eartags were seen on both cubs, but the numbers were not. Not radio-collared. Two out of potential of 4 of F24’s male cubs were visually observed with her on 4/1/08. Assume that 2 male cubs died before the age of 8.5 mo. Eartags were seen on both cubs, but the numbers were not. Survived to adult stage; dispersed from natal area. Killed by Wildlife Services for depredation control on 11-07-09 when 28 months old. F16 52 324 434 F59 34 5-24-07 06-27-07 to 08-21-07 55 324 M60 34 5-24-07 F61 34 5-24-07 06-27-07 to 07-11 to 12-07 06-27-07 to 06-29-07 434 48-49 324 434 538 M62 M63 M64 34 34 34 7-14-07 7-14-07 7-14-07 08-17-07 08-17-07 08-17-07 262 M65 34 7-14-07 08-17-07 262 148 F25 F16 F16 F24 F24 F24 F24 �Appendix A continued Puma I.D. Estimated Age at capture (days) F66 37 Est. Birth date 7-17-07 Est. survival span from 1st capture to fate or last monitor date 08-23-07 to 11-05-07 Age to last monitor date alive or at death (days, birth to fate) Radio-collared. Lost contact; last location 11/5/07. No signals after that date. F66 was photographed with one male sibling, either M67 or M68, & F30 on 5/31-6/1/08. F66 was recaptured and radio-collared as a subadult on 11/25/08. She died from massive trauma & bleeding of internal organs possibly resulting from being trampled by an elk or mule deer on about 05-28-09 as an independent subadult 23 months old. Not radio-collared. M67 or M68 was photographed with sibling F66 & mother F30 on 5/31-6/1/08. Not radio-collared. M67 or M68 was photographed with sibling F66 & mother F30 on 05-31 to 06-01-08. Survived to subadult stage; dispersed from natal area. Killed by a puma hunter in Disappointment Valley, CO (GMU 71) 12-30-08 at 17 months old. Radio-collared. Shed radiocollar between 7-9-08 and 7-1508, probably while still dependent on mother F75. Not radio-collared. Probably dead; if not killed when sibling M79 was killed, then probably would starve to death. Not radio-collared. Probably dead; if not killed when sibling M79 was killed, then probably would starve to death. Not radio-collared. Probably dead; if not killed when sibling M79 was killed, then probably would starve to death. Not radio-collared. Dead. Chewed-off anterior portions of the nasals, maxilla, palate, dentaries, and pieces of the braincase, with 6 or 9 portion of yellow ear-tag and intestines and bits of skin found ~45 m from mother F2’s death site on 8/14/08. Cub death probably due to puma-caused infanticide with cannibalism at ~87 days old. Male puma scrapes, about 8, under a rock rim ~50m distance from cub remains, and made ~ time of pumas’ deaths. Not radio-collared. Apparently died before 2-4-09; no tracks found in association with F23 & siblings F81 & F97. 111 M67 37 7-17-07 08-23-07 M68 37 7-17-07 08-23-07 F74 259 6-1-07 M76 30 5-19-08 03-12-08 to 07-09-08 06-18-08 ~87 M77 30 5-19-08 06-18-08 ~87 F78 30 5-19-08 06-18-08 ~87 M79 30 5-19-08 06-18-08 87 F80 40 5-23-08 07-02-08 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 403 149 Mother I.D. F30 F30 F30 F75 F2 F2 F2 F2 F23 �Appendix A continued Puma I.D. Estimated Age at capture (days) F81 40 F97 8 ½ mo. 5-23-08 5-23-08 M82 37 M83 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 07-02-08 to 07-29-09 02-04-09 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 424 354 F23 F23 5-29-08 07-05-08 to 03-20-09 or 04-02-09 295-308 37 5-29-08 07-05-08 M84 36 6-5-08 07-11-08 to 02-11-09 Radio-collared. Last live location 7-29-09. Radio-collared. Lost contact after 05-12-09; shed collar at elk kill cache on Mailbox Park. Radio-collared. Survived to subadult stage; dispersed from natal area. Killed by a puma hunter in 12-10-09 GMU 65 when 19 months old. Not radio-collared. Apparently died; no tracks found in association with F8 & sibling M82 2-10-09. Radio-collared 7-11-08 to 7-22-08; collar removed because of malfunction. Not radio-collared after 7-22-08. Eartag of M84 was found by E. Phillips on 8-25-08 when mother F70’s GPS locations located here on either side of the eartag in the East fork Dolores Cyn. M84 recaptured radiocollared again 1-29-09 in Dolores Cyn. in association with F70 & F96’s family. Shed radiocollar again about 211-09. F85 36 6-5-08 07-11-08 F70 F86 36 6-5-08 07-11-08 to 07-23 to 08-03-08 M87 M88 F89 M90 Male 7A 28 28 28 36 28-35 7-3-08 7-3-08 7-3-08 7-9-08 7-10-08 07-31-08 07-31-08 07-31-08 08-14-08 ~08-07-08 to 08-14-08 Male 7B 28-35 7-10-08 ~08-07-08 to 08-14-08 28 to 35 Female 7C 28-35 7-10-08 ~08-07-08 to 08-14-08 28 to 35 Radio-collared. Dead. Probably died of predation or infanticide about 10-108 near elk calf kill. Radio-collared 7-22-08. Dead. Radio-collar, orange ear-tag #86 with pinna with green tattoo #86 found by J. Timmer 9-1-08. F86 died ~7-23 to 8-3-08 when mother F70’s GPS locations located her at F86 remains. Probable predation. Not radio-collared. Not radio-collared. Radio-collared Radio-collared Not radio-collared. F7’s cubs died from starvation after they were orphaned. F7 was shot on 8-3-08 for killing domestic sheep. Not radio-collared. F7’s cubs died from starvation after they were orphaned. F7 shot on 8-3-08 for killing domestic sheep. Not radio-collared. F7’s cubs died of starvation after orphaned. F7 shot on 8-3-08 for killing domestic sheep. 251 ~48-59 28 to 35 150 Mother I.D. F8 F8 F70 F70 F3 F3 F3 F72 F7 F7 F7 �Appendix A continued Puma I.D. Estimated Age at capture (days) M91 35 M92 35 F95 16 mo. Est. Birth date Age to last monitor date alive or at death (days, birth to fate) 8-19-08 8-19-08 June-07 Est. survival span from 1st capture to fate or last monitor date 09-29-08 09-29-08 12-29-08 Sep-Oct08 Sep-Oct08 02-12-09 to 03-08-09 2-27-09 to 01-2010 146-176 05-20-09 to 09-19-09 05-20-09 157 05-20-09 to 09-17-09 06-14-09 to 02-09-10 06-14-09 to 03-16-10 159 241 F98 4-5 mo. M99 5 mo. M101 35 4-15-09 M102 35 4-15-09 F103 35 4-15-09 M105 38 5-7-09 F106 38 5-7-09 M107 34 5-25-09 F108 34 5-25-09 M109 M112 34 145 5-25-09 8-31-09 06-28-09 to 02-24-10 06-28-09 to 03-05-10 06-28-09 05-04-10 M115 14 mo. Nov.-08 07-21-10 610 M117 6 mo. Aug.-09 02-05-10 275 P1016(M) 39 6-12-10 06-12-10 to 07-21-10 39 P1017(M) 39 6-12-10 06-12-10 to 07-21-10 39 488 278 275 250 246 151 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. Radio-collared. Radio-collared. Radio-collared. Survived to adult stage. Established adult home range overlapping F93’s home range. Radio-collared. Died, probably killed by male puma (infanticide). Radio-collared. Last location 4-22-09 on Paterson Mt. Died as 16-month old subadult in San Miguel Canyon. Cause of death unknown, possibly killed by another puma. Radio-collared. Died; killed by puma M55 after cub was orphaned due to death of mother F16 by vehicle strike. Radio-collared. Lost contact after 9-4-09. Did not find evidence of M102 associated with deaths of siblings M101 and F103. But M102 probably died. Radio-collared. Died; killed by puma M55 after cub was orphaned due to death of mother F16 by vehicle strike. Radio-collared. Lost contact after 2-9-10 due to shed collar. F25 F25 F93 Not radio-collared at nursery; F75 returned to nursery during handling. Radio-collared later on 2-10-10. Lost contact due to shed collar 3-16 to 29-10. Not radio-collared; too small. Recaptured 2-24-10; not collared. Shed radiocollar at nursery; fastener failed. Recaptured and re-collared 2-24-10. Shed collar ~3-5-10. Not radio-collared; too small. Radio-collared. Lost contact after 5-4-10 (last live signal) possibly due to failed transmitter. Radio-collared. M115 died as a subadult (~20 mo. old) due to complications of a broken left foreleg (natural cause). Radio-collared. Lost contact after 5-14-10 (last live signal); shed collar found on 7-15-10 in the natal area. Not radio-collared. Monitored at nursery via mother’s GPS/VHF collar. Found dead at nursery due to infanticide by puma M32 on same day as our investigation of nursery. Not radio-collared. Monitored at nursery via mother’s GPS/VHF collar. Found dead at nursery due to infanticide by puma M32 on same day as our investigation of nursery. F75 Unm.F Unm.F F16 F16 F16 F75 F94 F94 F94 F70 F28 F119 F72 F72 �Appendix A continued Puma I.D. Estimated Est. Est. survival span Age to last monitor date Status: Alive/Survived to subadult stage/ Mother Age at Birth from 1st capture to alive or at death (days, Lost contact/Disappeared/ I.D. birth to fate) capture date fate or last monitor Dead; Cause of death (days) date M120 30 6-28-10 07-28-10 Radio-collared. F3 M121 30 6-28-10 07-28-10 Radio-collared. F3 M122 35 7-8-10 08-12-10 Radio-collared. F104 F123 29 7-15-10 08-13-10 Radio-collared. F94 F124 29 7-15-10 08-13-10 Radio-collared. F94 M125 29 7-15-10 08-13-10 Radio-collared. F94 a Cub F35 probably starved between 06-30-06 & 07-07-06 after the transmitter on the expandable collar got in its mouth. b Cub M42 died after being captured by dogs, probably from stress of capture associated with severe infection of laceration under right foreleg caused by expandable radiocollar. c Cub M56 was captured in association with F7 and her cubs M43 and M44. He may have been missed at the nursery when M43 and M44 were initially sampled and marked. d Cub M60 died probably of starvation. The expandable radiocollar was around the neck and right shoulder, possibly restricting movement 152 �Colorado Division of Parks and Wildlife July 2010 –June 2011 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 3003 1 : : : : Division of Parks and Wildlife Mammals Research Carnivore Conservation Puma Population Structure and Vital Rates On the Uncompahgre Plateau Federal Aid Project No. Period covered: July 31, 2010−June 30, 2011 Author: K. A. Logan. Personnel: K. Logan, A. Butler, B. Dunne, W. Hollerman, C. Jacobs, W. Jesson, J. Knight, B. Nay, R. Navarrete, J. Waddell, S. Waters, T. Bonacquista, K. Crane, J. Koch, and G. Watson of CPW; volunteers and cooperators including: private landowners, Bureau of Land Management, Ridgway State Park, Colorado State University, and U.S. Forest Service. Supplemental financial support received in previous years from The Howard G. Buffett Foundation and Safari Club International Foundation. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT The Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) initiated a 10-year study on the Uncompahgre Plateau in 2004 to quantify puma population characteristics in the absence (reference period, yrs 1-5) and presence (treatment period, yrs 6-10) of hunting. The purpose of the study is to evaluate assumptions underlying the Colorado Parks and Wildlife’s model-based approach to managing pumas with sport-hunting in Colorado. The reference period began December 2004 and ended July 2009, during which we captured, sampled, and marked 109 pumas for population research purposes on the Uncompahgre Plateau (Logan 2009). This report informs on the second year of the treatment period (TY2), August 2010 through July 2011, on puma population characteristics and dynamics with hunting as a mortality factor. Puma sport-hunting opened November 22 and closed December 12, 2010 after a quota of 8 independent pumas was harvested. The harvest was designed to test the management assumption that a 15% harvest of independent pumas results in a stable-to-increasing population. A total of 8 pumas were killed: 2 subadult females, 5 adult males, and 1 subadult male. The harvest of 8 independent pumas represented 15.4% of the 52 independent pumas in our minimum count during November 2010 to April 2011. Independent females and males comprised 25.0% and 75.0% of the harvest, respectively. Three other radio-collared independent pumas in the study area population were killed during the Colorado puma hunting season; 2 adult females killed on the study area for depredation control and 1 adult male in a GMU adjacent to the study area. The total mortality of 11 independent pumas during the hunting season represented 21.2% of the minimum count of independent pumas. Eight independent pumas will be the harvest quota for the 2011-12 hunting season (TY3), based on an expectation of a stable-to-increasing population. Sixty-four hunters requested mandatory permits with an 177 �attached voluntary hunter survey in TY2. Fifty-four of the hunters provided responses to written (n = 42) or telephone call follow-up contact (n = 12). An estimated 42 hunters actually hunted on the study area, of which about 19% harvested pumas and 38% captured pumas (i.e., harvested plus treed and released). Thirty-three hunters responded that they were selective hunters, and the capture, tracking, and population data indicated that most hunters practiced selection. Puma tracks < 1 day old encountered by hunters and pumas captured by hunters indicated that independent female pumas were more vulnerable than males to detection by hunters. From August 2010 to July 2011 54-55 individual pumas were captured 70 times. Two capture teams with dogs operated over 81 search days from November 16 and December 14, 2010 through April 22, 2011 to find 291 puma tracks, pursue pumas 99 times, and capture 36-37 pumas 52 times. Capture efforts with cage traps resulted in the capture of 1 adult male and 1 subadult male for the first time and the recapture of 2 adult female pumas. Fourteen cubs were observed for the first time at nurseries. A total of 53 pumas were monitored by radiotelemetry in TY2. Search efforts also revealed the presence of at least 15 other independent pumas. Our minimum count of independent pumas from November 2010 to April 2011 was 52, including 35 females and 17 males. A preliminary minimum estimated density of independent pumas was 3.11/100 km2. The proportion of radio-collared adult females giving birth in the August 2010 to July 2011 biological year was 0.56 (9/16). Six litters that could be dated to month of birth were produced in April (2), July (2), and August (2). Since 2005 a birth peak has occurred from May through August, involving 80% of nursling litters. We monitored 19 female and 9 male adult radio-collared pumas for survival and agent-specific mortality. Survival rates in TY2 for adult females were within the range during the reference period, but substantially lower for males. Causes of mortality were hunting and depredation control. One subadult female was killed and eaten by a male puma during competition for an elk carcass. Of 23 cubs monitored with radiotelemetry, 6 died, 3 from natural causes (including 2 infanticide and cannibalism) and 3 from depredation control. A non-marked female cub was also killed by a vehicle on the boundary of the study area. Puma harvest, capture, and radiotelemetry data provided information on dispersals of 26 pumas initially marked on the study area. Those pumas moved from about 20 to 370 km from initial capture sites. We explored the feasibility of attracting pumas to rub stations to obtain tissue non-invasively for potential use in a genotype markrecapture structure for estimating abundance. Nine sites with trail cameras, rub devices, and 6 scents produced 39 puma visit events. Puma behavior toward the scents was highly variable. Beaver castorium produced the highest maximum detection probability. Data continue to be gathered for other collaborative projects with Mammals Research and CSU investigators on puma behavior, social organization, population dynamics, and habitat use. 178 �WILDLIFE RESEARCH REPORT PUMA POPULATION STRUCTURE AND VITAL RATES ON THE UNCOMPAHGRE PLATEAU, COLORADO KENNETH A. LOGAN P. N. OBJECTIVE Quantify puma population sex and age structure; estimate puma population vital rates, including: reproduction rates of females, age-stage survival rates, and immigration and emigration rates; quantify agent-specific mortality rates; model puma population dynamics; develop and execute the puma harvest manipulation to begin the population-wide test of Colorado Parks and Wildlife (CPW) puma management assumptions in the first year of a five-year Treatment Period of the Uncompahgre Plateau Puma Project― all to improve the CPW model-based approach to managing pumas in Colorado. SEGMENT OBJECTIVES 1. Execute the second year of the five-year treatment period by working with CPW biologists and managers to manipulate the puma population with sport-hunting and to survey hunters. 2. Continue gathering data on puma population sex and age structure. 3. Continue gathering data for estimates of puma reproduction rates. 4. Continue gathering data to estimate puma sex and age-stage survival rates. 5. Continue gathering data on agent-specific mortality. 6. Explore feasibility of attracting pumas to a rub station and obtaining tissue for potential use in a noninvasive genotype mark-recapture structure. INTRODUCTION Colorado Parks and Wildlife managers need reliable information on puma biology and ecology in Colorado to develop sound management strategies that address diverse public values and the CPW objective of actively managing pumas while “achieving healthy, self-sustaining populations”(Colorado Division Of Wildlife 2002-2007 Strategic Plan:9). Although 4 puma research efforts have been made in Colorado since the early 1970s and puma harvest data is compiled annually, reliable information on certain aspects of puma biology and ecology, and management tools that may guide managers toward effective puma management is lacking. Mammals Research staff held scoping sessions with a number of the CPW’s wildlife managers and biologists prior to initiating the project. In addition, we consulted with other agencies, organizations, and interested publics either directly or through other CPW employees. In general, CPW staff in western Colorado highlighted concern about puma population dynamics, especially as they relate to their abilities to manage puma populations through regulated sport-hunting. Secondarily, they expressed interest in puma―prey interactions. Staff on the Front Range placed greater emphasis on puma―human interactions. Staff in both eastern and western Colorado cited information needs regarding effects of puma harvest, puma population monitoring methods, and identifying puma habitat and landscape linkages. Management needs identified by CPW staff and public stakeholders form the basis of Colorado’s puma research program, with multiple lines of inquiry (i.e., projects): Improve our ability to manage puma hunting with enhanced scientific bases, strategies, and tools― ● Puma population characteristics (i.e., density, sex and age structure). ● Puma population dynamics and vital rates (i.e., birth rates, survival rates, emigration rates, immigration rates, population growth rates). 179 �● Field methods and models for assessing and tracking changes in puma populations. ● Relative vulnerability of puma sex and age classes to hunter harvest. Improve our understanding of puma habitat needs and interrelationships of puma management units― ● Puma habitat use, movements, and use of landscape linkages. ● Puma recruitment patterns (i.e., progeny, immigration, emigration). ● Models for identifying puma habitat and landscape linkages. Improve our understanding of the puma’s role in the ecology of other species ● Relationships of puma to mule deer, elk, and other natural prey. ● Relationships of puma to species of special concern, e.g., desert bighorn sheep. Improve our understanding of puma-human interactions and abilities to manage them ● Behavior of puma in relation to people and human facilities. ● Puma predation on domestic animals. ● Effects of translocating nuisance pumas. ● Effects of aversive conditioning on pumas. While all projects cannot be addressed concurrently, understanding their relationships to one another is expected to help individual projects maximize their benefits to other projects that will assist the CPW to achieve its strategic goal in puma management (Fig.1). This project has been addressing all of the gray-shaded components on the left side of the conceptual model in Figure 1. Management issues identified by managers translate into researchable objectives, requiring descriptive studies and field manipulations. Our goal is to provide managers with reliable information on puma population biology and to develop useful tools for their efforts to adaptively manage puma in Colorado to maintain healthy, self-sustaining populations. The highest-priority management needs are being addressed with this intensive population study that focuses on puma population dynamics using sampled, tagged, and GPS/VHF-radio-collared pumas. Those objectives include: Describe and quantify puma population sex and age structure. Estimate puma population vital rates, including: reproduction rates, age-stage survival rates, emigration rates, immigration rates. Estimate agent-specific mortality rates. Improve the CPW’s model-based management approaches with Colorado-specific data from objectives 1―3. Consider other useful models. Concurrently with the tasks associated with the objectives above, significant progress will be made toward a 5th objective, which will initially be subject to pilot study― develop methods that yield reliable estimates of puma population abundance. A descriptive and manipulative study will estimate population parameters in an area that appears typical of puma habitat in western Colorado and will yield defensible population parameters based upon contemporary Colorado data. This study will be conducted in two 5-year periods. A completed 5-year reference period, 2004-09, (i.e., absence of recreational hunting) allowed puma life history traits to interact with the main habitat factors that influenced puma population growth (e.g., prey availability and vulnerability, Pierce et al. 2000, Logan and Sweanor 2001, Logan 2009). A subsequent 5-year treatment period started in 2009-10 will involve the use of controlled recreational hunting to manipulate the puma population. 180 �TESTING ASSUMPTIONS AND HYPOTHESES Hypotheses associated with main objectives 1―5 of this puma population research are structured to test assumptions guiding puma management in Colorado. 1. Considering limitations (i.e., methods, number of years, assumption violations) to the previous Colorado-specific studies on puma populations (Currier et al. 1977, Anderson et al. 1992, Koloski 2002), managers assume that puma population densities in Colorado are within the range of those quantified in more intensively studied populations in Wyoming (Logan et al. 1986), Idaho (Seidensticker et al. 1973), Alberta (Ross and Jalkotzy 1992, and New Mexico (Logan and Sweanor 2001). The CPW assumes density ranges of 2.0−4.6 puma/100 km2 (i.e., includes pumas of all age stages- adults, subadults, and cubs, J. Apker, CPW Carnivore Biologist, person. commun. Nov. 19, 2003) to extrapolate to Data Analysis Units (DAUs) to guide the model-based quota-setting process. Likewise, managers assume that the population sex and age structure is similar to puma populations described in the intensive studies. Using intensive efforts to capture, mark, and estimate non-marked animals developed and refined during the study to estimate the puma population, the following will be tested: H1: Puma densities during the 5-year reference period (absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will vary within the range of 2.0 to 4.6 puma/100 km2 and will exhibit a sex and age structure similar to puma populations in Wyoming, Idaho, Alberta, and New Mexico. 2. Recreational puma hunting management in Colorado DAUs is guided by a model to estimate allowable harvest quotas to achieve one of two puma population objectives: 1) maintain puma population stability or growth, or 2) cause puma population decline (CDOW, Draft L-DAU Plans, 2004, CDOW 2007). Basic model parameters are: puma population density, sex and age structure, and annual population growth rate. Parameter estimates are currently chosen from literature on studies in western states that are judged to provide reliable information. Background material used in the model assumes a moderate annual rate of growth of 15% (i.e.,λ = 1.15) for the adult and subadult puma population (CDOW 2007). This assumption is based upon information with variable levels of uncertainty (e.g., small sample sizes, data from habitats dissimilar to Colorado). Parameters influencing λ include population density, sex and age structure, female age-at-first-breeding, reproduction rates, sex- and age-specific survival, immigration and emigration. H2: Population parameters estimated during a 5-year reference period (in absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will yield an estimated annual adult plus subadult population growth rate that will match or exceed λ = 1.15. 3. The key assumption is that the CPW can manage puma population growth through recreational hunting on the basis that for a stable puma population hunting removes the annual increment of population growth (i.e., from current judgments on population density, structure, and λ). Puma harvest rate formulations for DAUs assumes that total mortality (i.e., harvest plus other detected deaths) in the range of 8 to 15% of the harvest-age population (i.e., independent pumas comprised of adults plus subadults) with the total mortality comprised of 35 to 45% females (i.e., adults and subadults) is acceptable to manage for a stable-to-increasing puma population (CPW 2007). H3: Total mortality of an estimated 15% of the adults and subadults with no more than 45% of the total mortality comprised of females will not result in a declining trend of the harvest-age segment of the population. 4. To reduce a puma population, hunting must remove more than the annual increment of population growth. For DAUs with the objective to suppress the puma population, the total mortality guide of 181 �greater than 15 to 28% of the harvest-age population with greater than 45% comprised of females is suggested (CDOW 2007). H4: Total mortality of an estimated 16% or greater of the harvestable population with greater than 45% females will cause a declining trend in the abundance of harvest-age pumas (i.e., adults and subadults). 5. The increase and decline phases of the puma population make it possible to test hypotheses related to shifts in the age structure of the population which have been linked to harvest intensity in Wyoming and Utah. H5: The puma population on the Uncompahgre Plateau study area will exhibit a young age structure after hunting prohibition at the beginning of the reference period. During the 5 years of hunting prohibition, greater survival of independent pumas will cause an older age structure in harvest-age pumas (i.e., adults and subadults) as suggested by the work of Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. As hunting is re-instated in the treatment period, the age structure of harvested pumas and the harvest-age pumas in the population will decline as observed by Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. Researchers in Wyoming (Anderson and Lindzey 2005) concluded that sex and age composition of the harvest varies predictably with puma population size because the likelihood of a specific sex or age class of puma being harvested with the use of hounds is a product of the relative abundance of particular sex and age classes in the population and their relative vulnerability to harvest. Results of that study suggest that managers could use sex and age composition of the harvest to infer puma population changes (Anderson and Lindzey 2005). The CPW currently uses this approach as one tool to infer potential DAU puma population dynamics (CDOW 2008). This assumes no purposeful selection by hunters for any particular sex or age-stage other than the puma must be legal (i.e., independent subadult or adult, not a lactating female or a female in association with spotted cubs) and that changes in the sex and age structure of the harvested pumas is due solely to changes in the relative abundance of particular sex and age classes in the population and their relative vulnerability to harvest. Theoretically, pumas that travel longer distances with movements that intercept access routes used by hunters (i.e., roads, trails) should be more exposed to detection by hunters and thus more vulnerable to harvest. A key assumption to this method is that pumas are killed as they are encountered and the harvest sex and age composition will reliably indicate whether a population is stable, increasing, or declining even if harvest intensity does not vary. Thus, an alternate view is that a population segment, such as independent females, may be more abundant and have shorter movement lengths, yet be detected more frequently by hunters. However, because the same intensively studied Wyoming puma population was manipulated over 6 years with varying intensities of harvest (Anderson and Lindzey 2005), variations in harvest structure using the same harvest level over a period of years could not be examined. This is a property we will investigate during the treatment period on the Uncompahgre Plateau puma study. Moreover, we will directly evaluate to what extent puma harvest might be influenced by hunter selection. A hunter survey is intended to reveal puma hunter behavior, detection of different classes of pumas, and lack of or presence of hunter selection. These data should allow us to examine the credibility of the assumption of non-selection by hunters and the robustness of this technique in gauging puma population dynamics relative to harvest. We want to examine the usefulness of this approach in Colorado. CPW managers attempt to weight sport-harvest toward male pumas in GMUs with the stable-to-increasing population objective with an active educational program (i.e., mandatory hunter exam, brochure, workshops). Thus, there is a need to test assumptions associated with the Anderson and Lindzey (2005) method. H6: No hunter selection is practiced so that the sex and age structure of pumas harvested by hunters in this population protected from hunting during a 5-year reference period and subsequently managed for stability or increase with conservative harvest levels will reflect the relative vulnerabilities to detection and capture with dogs during each year in the 5-year treatment 182 �period in this order from high to low vulnerabilities: subadult males, adult males, subadult females, adult females without cubs or with cubs >6 months old, and adult females with cubs ≤6 months old (Barnhurst 1982, Anderson and Lindzey 2005). In each of the 5 years of the treatment period, subadults and adult males should comprise the majority of the harvest and reflect the assumed sex and age structure (Anderson and Lindzey 2005) of a puma population managed for a stable to increasing phase and not hunted for 5 previous years (i.e., a puma population source). Desired outcomes and management applications of this research include: 1. Quantification of variations in puma population density, sex and age structure, growth rates, vital rates, and an understanding of factors affecting them will aid adaptive puma management by yielding population parameters and tools useful for assessing puma population dynamics, evaluation of management alternatives, and effects of management prescriptions. 2. Testing assumptions about puma populations, currently used by CPW managers, will help managers to biologically support and adapt puma management based on Colorado-specific estimated puma population characteristics, parameters, and dynamics. 3. Methods for assessing puma population dynamics will allow managers to evaluate modeled populations and estimate effects of management prescriptions designed to achieve specified puma population objectives in targeted areas of Colorado. Ascertaining puma numbers and densities during the project will allow assessment of monitoring techniques. Potential methods include use of harvest sex and age structure and photographic and DNA genotype capture-recapture. Study plans to develop and test feasible field and analytical methods will be developed as we learn the logistics of performing those methods, after we have preliminary data on puma demographics and movements which will inform suitable sampling designs, and if we have adequate funding. 4. This information will be disseminated to citizen stakeholders interested in pumas in Colorado, and thus contribute to informed public participation in puma management. STUDY AREA The study area for the puma population research is on the Uncompahgre Plateau (in Mesa, Montrose, Ouray, and San Miguel Counties; Fig. 2). The study area includes about 2,253 km2 (870 mi.2) of the southern halves of Game Management Units (GMUs) 61 and 62, and about 155 km2 (60 mi.2) of the northern edge of GMU 70 (between state highway 145 and San Miguel River). The area is bounded by state highway 348 at Delta, 25 Mesa road and Forest Service road FS503 to Nucla, state highway 97 to state highway 141 to state highway 145 to Placerville, state highway 62 to Ridgeway, U.S. highway 550 to Montrose, and U.S. highway 50 to Delta. The study area seems typical of puma habitat in Colorado that has vegetation cover that varies from the pinion-juniper covered foothills starting from about 1,700 m elevation to the spruce-fir and aspen forests growing to the highest elevations of about 3,000 m. Mule deer (Odocoileus hemionus) and elk (Cervus elaphus) are the most abundant wild ungulates available for puma prey. Cattle and domestic sheep are raised on summer ranges on the study area. Year-round human residents live along the eastern and western fringe of the area, and there is a growing residential presence especially on the southern end of the plateau. A highly developed road system makes the study area highly accessible for puma research efforts. A detailed description of the Uncompahgre Plateau is in Pojar and Bowden (2004). METHODS Reference and Treatment Periods This research was structured in two 5-year periods: a reference period (years 1―5) and a treatment period (years 6―10). The reference period was closed to puma hunting on the study area and was expected to cause a population increase phase. The treatment period (starting in November 2009) 183 �involves manipulation of the puma population with sport-hunting structured to achieve a management objective for a stable to increasing population. In both phases, puma population structure, and vital rates are being quantified, and management assumptions and hypotheses regarding population dynamics and effects of harvest are being tested. Contingent upon results of pilot studies, we will also assess enumeration methods for estimating puma population abundance. The reference period, without recreational puma hunting as a major limiting factor, was consistent with the natural history of the current puma species in North America which evolved life history traits during the past 10,000 to 12,000 years (Culver et al. 2000) that enable pumas to survive and reproduce (Logan and Sweanor 2001). In contrast, puma hunting, with its modern intensity and ingenuity, might have influenced puma selection pressures in western North America for at least the past 100 years. Hence, the reference period, years 1―5, provided conditions where individual pumas in this population (of estimated sex and age structure) expressed life history traits interacting with the environment without recreational hunting as a limiting factor. Theoretically, the main limiting factor was vulnerable prey abundance (Pierce et al. 2000, Logan and Sweanor 2001). This allowed researchers to understand basic system dynamics before manipulating the population with controlled recreational hunting. In the reference period, all pumas in the study area were protected, except for individual pumas involved in depredation on livestock or human safety incidents. In addition, all radio-collared and ear-tagged pumas that ranged in a buffer zone in the northern halves of GMUs 61 and 62 were protected from recreational hunting mortality. The reference period allowed researchers to quantify baseline demographic data on the puma population to estimate parameters useful for assessing the CPW’s assumptions for its model-based approach to puma management. The reference period also facilitated other operational needs (because hunters did not kill the animals) including the marking of a large proportion of the puma population for parameter estimates and gathering movement data from GPS-collared pumas. During the treatment period, years 6―10, recreational puma hunting is occurring on the same study area using management prescriptions structured from information learned during previous years. Using recreational hunting for the treatment is consistent with the CPW’s objectives of manipulating natural tendencies of puma populations, particularly survival, to maintain either population stability or increase or suppression (CDOW, Draft L-DAU Plans, 2004). Theoretically, survival of independent pumas is being influenced mainly by recreational hunting, which is being quantified by agent-specific mortality rates of radio-collared pumas. Dynamics of the puma population are being manipulated to evaluate hypotheses that are related to effects of hunting (i.e.,: effects of harvest rates, relative vulnerability of puma sex and age classes to hunting, variations in puma population structure due to hunting). The killing of tagged and collared pumas during the treatment period is not hampering operational needs (as it would have during the start-up years), because a majority of independent pumas in the population have already been marked, and sampling methods formalized. Pumas on the study area that may be involved in depredation of livestock or human safety incidences may be lethally controlled. Researchers that find that GPS-collared pumas have killed domestic livestock will record such incidents to facilitate reimbursement to the property owner for loss of the animal(s). In addition, researchers will notify the Area Manager of the CPW if they perceive that an individual puma may be a threat to public safety. 184 �Field Methods Puma Capture: Realizing that pumas live at low densities and capturing pumas is difficult, as a starting point, our logistical aim was to have a minimum of 6 puma in each of 6 categories (36 total) radio-tagged in any year of the study if those or greater numbers are present. The 6 categories are: adult female, adult male, subadult female, subadult male, female cub, male cub. Our aim was to provide more quantitative and precise estimates of puma demographics than were achieved in earlier Colorado puma studies. This relatively large number of pumas might represent the majority of the puma population on the study area, and would provide the basic data for age- and sex-specific reproductive rates, survival rates, agent-specific mortality rates, emigration, and other movement data. Puma capture and handling procedures were approved by the CPW Animal Care and Use Committee (file #08-2004). All captured pumas were examined thoroughly to ascertain sex and describe physical condition and diagnostic markings. Ages of adult pumas were estimated initially by the gum-line recession method (Laundre et al. 2000) and dental characteristics of known-age puma (Logan and Sweanor, unpubl. data). Ages of subadult and cub pumas were estimated initially based on dental and physical characteristics of known-age pumas (Logan and Sweanor unpubl. data). Body measurements recorded for each puma included at a minimum: mass, pinna length, hind foot length, plantar pad dimensions. Tissue collections included: skin biopsy (from the pinna receiving the 6 mm biopsy punch for the ear-tags), and blood (30 ml from the saphenous or cephalic veins) for genotyping individuals, parentage and relatedness analyses, and disease screening; hair (from various body regions) for genotyping tests of field gathered samples. Universal Transverse Mercator Grid Coordinates on each captured puma were fixed via Global Positioning System (GPS, North American Datum 27). Pumas were captured year-round using 4 methods: trained dogs, cage traps, foot-hold snares, and by hand (for small cubs). Capture efforts with dogs were conducted mainly during the winter when snow facilitated thorough searches for puma tracks and the ability of dogs to follow puma scent. The study area was searched systematically multiple times per winter by four-wheel-drive trucks, all-terrain vehicles, snow-mobiles, and walking. When puma tracks ≤1 day old were detected, trained dogs were released to pursue pumas for capture. Pumas usually climbed trees to take refuge from the dogs. Adult and subadult pumas captured for the first time or requiring a change in telemetry collar were immobilized with Telazol (tiletamine hydrochloride/zolazepam hydrochloride) dosed at 5 mg/kg estimated body mass (Lisa Wolfe, DVM, CPW, attending veterinarian, pers. comm.). Immobilizing agent was delivered into the caudal thigh muscles via a Pneu-Dart® shot from a CO2-powered pistol. Immediately, a 3m-by-3m square nylon net was deployed beneath the puma to catch it in case it fell from the tree. A researcher climbed the tree, fixed a Y-rope to two legs of the puma and lowered the cat to the ground with an attached climbing rope. Once the puma was on the ground, its head was covered, its legs tethered, and vital signs monitored (Logan et al. 1986). Normal signs include: pulse ~70 to 80 bpm, respiration ~20 bpm, capillary refill time ≤2 sec., rectal temperature ~101oF average, range = 95 to 104oF (Kreeger 1996). Pumas that climbed trees too dangerous for the pumas or researchers were released without handling, or we encourage the animals to leave the tree by heaving snowballs toward them. If the pumas climbed a safe tree, then we handled them as described above. A cage trap was used to capture adults, subadults, and large cubs when pumas were lured into the trap using road-killed or puma-killed ungulates (Sweanor et al. 2008). A cage trap was set only if a target puma scavenged on the lure (i.e., an unmarked puma, or a puma requiring a collar change). Researchers continuously monitored the set cage trap from about 1 km distance by using VHF beacons on the cage and door. Researchers handled captured pumas within 30 minutes of capture. Puma were immobilized with Telazol injected into the caudal thigh muscles with a pole syringe. Immobilized pumas were 185 �restrained and monitored as described previously. If non-target animals were caught in the cage trap, we opened the door and allowed the animal to leave the trap. Small cubs (≤10 weeks old) were captured using our hands (covered with clean leather gloves) or with a capture pole. Cubs were restrained inside new burlap bags during the handling process and were not administered immobilizing drugs. Cubs at nurseries were approached when mothers were away from nurseries (as determined by radio-telemetry). Cubs captured at nurseries were removed from the nursery a distance of 30 to 100 m to minimize disturbance and human scent at nurseries. Immediately after handling processes were completed, cubs were returned to the exact nurseries where they were found (Logan and Sweanor 2001). Marking, Global Positioning System- and Radio-telemetry: Pumas did not possess easily identifiable natural marking, such as tigers (see Karanth and Nichols 1998, 2002), therefore, the capture, marking, and GPS- or VHF- collaring of individual pumas was essential to a number of project objectives, including estimating numbers, vital rates, and gathering movement data relevant to population dynamics (i.e., emigration and Data Analysis Unit boundaries). Adult, subadult, and cub pumas were marked 3 ways: GPS/VHF- or VHF-collar, ear-tag, and tattoo. The identification number tattooed in the pinna was permanent and could not be lost unless the pinna was severed. A colored (bright yellow or orange), numbered rectangular (5 cm x 1.5 cm) ear-tag (Allflex USA, Inc., DFW Airport, TX) was inserted into each pinna to facilitate individual identification during direct recaptures. Cubs ≤10 weeks old were ear-tagged in only one pinna. Adult and subadult female pumas were fitted with GPS collars (approximately 400 g each, Lotek Wireless, Canada) if available. Initially, GPS-collars were programmed to fix and store puma locations at 4 times per day to sample daytime, nighttime, and crepuscular locations (i.e., 0:00, 06:00, 12:00, 19:00). GPS locations for pumas provided precise, quantitative data on movements to assess the relevance of puma DAU boundaries, our search efforts, and to evaluate puma behavior and social structure. The GPScollars also provided basic information on puma movements and locations to design other pilot studies in this program on vulnerability of puma to sport-harvest, habitat use, and enumeration methods (e.g., photographic or DNA mark-recapture). Subadult male pumas were fitted initially with conventional VHF collars (Lotek, LMRT-3, ~400 g each) with expansion joints fastened to the collars, which allowed the collar to expand to the average adult male neck circumference (~46 cm). If subadult male pumas reached adulthood on the study area, we would recapture them and fit them with GPS collars. In addition, other adult and female subadult pumas were fitted with VHF collars when GPS collars were not available. VHF radio transmitters on GPS collars enabled researchers to find those pumas on the ground in real time to acquire remote GPS data reports, facilitate recaptures for re-collaring, and to determine their reproductive and survival status. VHF transmitters on GPS- and VHF-collars had a mortality mode set to alert researchers when pumas were immobile for 3 to 24 hours so that dead pumas could be found to quantify survival rates and agent-specific mortality rates by gender and age. Locations of GPS- and VHFcollared pumas were fixed about once per week (as flight schedules and weather allowed) from light fixed-wing aircraft (e.g., Cessna 185) fitted with radio signal receiving equipment (Logan and Sweanor 2001). GPS- and VHF-collared pumas were located from the ground opportunistically using hand-held yagi antenna. At least 3 bearings on peak aural signals were mapped to fix locations and estimate location error around locations (Logan and Sweanor 2001). Aerial and ground locations were plotted on 7.5 minute USGS maps (NAD 27) and UTMs along with location attributes recorded on standard forms. GPS and aerial locations were mapped using GIS software. 186 �We attempted to collar all cubs in observed litters with small VHF transmitter mounted on an expandable collar that can expand to adult neck size (Wildlife Materials, Murphysboro, Illinois, HLPM2160, 47g, Telonics, Inc., Mesa, Arizona MOD 080, 62g, or Telonics MOD 205, 90g,) when cubs weighed 2.3―11 kg (5―25 lb). Cubs could wear these small expandable collars until they were over 12 months old. Cubs were recaptured to replace collars as opportunities allowed. Monitoring radio-collared cubs allowed quantification of survival rates and agent-specific mortality rates (Logan and Sweanor 2001). Analytical Methods Population Characteristics: Population characteristics each year were tabulated with the number of individuals in each sex and age category. Age categories, as mentioned, include: adult (puma ≥24 months old, or younger breeders), subadults (young puma independent of mothers, <24 months old that do not breed), cubs (young dependent on mothers, also called kittens) (Logan and Sweanor 2001). When data allowed, age categories were further partitioned into months or years. Reproductive Rates: Reproductive rates were estimated for GPS- and VHF-collared female pumas directly (Logan and Sweanor 2001). Genetic paternity analysis will be used to ascertain paternity for adult male pumas (Murphy et al. 1998). Survival and Agent-specific Mortality Rates: Radio-collared pumas provided known fate data used to estimate survival rates for each age stage using the Kaplan-Meier procedure to staggered entry (Pollock et al. 1989). A binomial survival model was also used for crude estimates of survival during the subadult age stage (Williams et al. 2001:343-344). In addition, when data collection is complete, survival rates will be modeled in program MARK (White and Burnham 1999, Cooch and White 2004) where effects of individual (e.g., sex, age stage, reproductive stage) and temporal (i.e., reference period, treatment period) covariates to survival can be examined. Agent-specific mortality rates can also be analyzed using proportions and Trent and Rongstad procedures (Micromort software, Heisey and Fuller 1985). Population Inventory: The population of interest was independent pumas (i.e., adults and subadults) mainly during November to March which corresponds with Colorado’s puma hunting season. Independent pumas were those that could be legally killed by recreational hunters. Initially, we estimated the minimum number of independent pumas and puma density (i.e., number of independent puma/100 km2) each winter. The minimum number of independent pumas included all marked pumas known to be present on the study area during the period, plus individuals thought to be non-marked and detected by visual observation or tracks that were separated from locations of radio-collared pumas. Furthermore, adults comprised the breeding segment of the population and subadults were non-breeders that are potential recruits into the adult population in ≤1 year. The sampling unit was the individual independent puma (~≥1 yr. old). Puma Population Dynamics: A deterministic, discrete time model parameterized with population characteristics and vital rates from this research was used to assess puma population dynamics (Logan 2008). Functional Relationships: Once data collection is complete, a variety of analyses will be conducted to estimate parameters and examine functional relationships. Graphical methods will be used to initially examine functional relationships among puma population parameters. Linear regression procedures and coefficients of determination will be used to assess functional relationships if data for the response variable are normally distributed and the variance is the same at each level. If the relationship is not linear, data is non-normal, and variances are unequal, we will consider appropriate transformations of the data for regression procedures (Ott 1993). Non-parametric correlation methods, such as Spearman’s 187 �rank correlation coefficient, will also be used where appropriate to test for monotonic relationships between puma abundance and other parameters of interest (Conover 1999). Relationships of explanatory variables to survival parameters will be modeled in MARK. Statistical analyses can be performed in a variety of software (e.g., SYSTAT, R, and MARK). RESULTS AND DISCUSSION Segment Objective 1 Puma harvest: This biological year, August 2010 to July 2011, was the second year of the treatment period (TY2) in this study of puma population dynamics on the Uncompahgre Plateau. The hunting season on the study area began on November 22, 2010 and was scheduled to extend to January 31, 2011, unless the harvest quota was taken before then. The design harvest quota was 8 pumas (i.e., 15% harvest of the estimated minimum number of independent pumas), with the objective to manage for a stable to increasing population. This design harvest tests the CPW’s current assumption that total mortality (i.e., harvest plus other natural deaths) in the range of 8 to 15% of the harvest-age population (i.e., independent pumas comprised of adults plus subadults) with the total mortality comprised of 35 to 45% females (i.e., adults and subadults) is acceptable to manage for a stable-to-increasing puma population (Assumption and Hypothesis 3 p.5 this report). The initial quota of 8 pumas for TY1 was based on the projected minimum number of 53 independent pumas expected on the study area in winter 2009-10, modeled from a minimum count of pumas during winter 2007-08 (Table 1; Logan 2010). The quota of 8 pumas for TY2 was based on the observed minimum count of 55 independent pumas during September 2009 to April 2010 in TY1 and that approximately the same number of independent pumas were expected during the puma hunting season for TY2 (an expectation consistent with our observed minimum count of 52 independent pumas for TY2, see later in Segment Objective 2). The hunting structure in TY2 was the same as in TY1. The number of puma hunters on the study area was not limited. Each hunter on the study area was required to obtain a hunting permit from the CPW Montrose Service Center. Permits were free and unlimited. Each permit allowed the individual hunter with a legal puma hunting license in Colorado to hunt in the puma study area for up to 14 days from the issue date. Unsuccessful hunters that wanted to continue hunting past the permit expiration date requested a new permit for another 14 days, or until the hunter killed a puma within the season, or the season on the study area closed due to the quota being reached, or the end of the hunting season. This permit system allowed the CPW to monitor the number of hunters on the study area and to contact each hunter for survey information (see later in this section). All pumas harvested on the study area were examined by principal investigator K. Logan or a wildlife research technician and sealed as mandated by Colorado statute. All successful hunters reported their puma kill and presented the puma carcass for inspection by CPW within 48 hours of harvest. Upon inspection data were recorded on the puma harvested, including: sex, age, and location of harvest. In addition, an upper premolar tooth was collected for aging (i.e., mandatory) and a tissue sample was collected for DNA genotyping. Each successful hunter was also asked at that time to complete a one-page hunter survey form. All other hunters that did not report a puma kill on the study area were asked to complete the survey form and return it in a stamped envelope that was provided. An attempt was made to contact other hunters by telephone if they did not mail in surveys. The puma hunting season occurred on the study area from November 22 to December 12, 2010, taking 21 days to fill the quota of 8 pumas. This was 5 days less than it took to harvest 8 pumas in TY1 (i.e., 26 days, Nov. 16 to Dec. 11, 2009). Eight pumas were killed, including: 2 subadult females, 5 adult males, and 1 subadult male (Table 2). Of the 8 harvested pumas, 4 were marked: M32, M55, M90, and F108. In addition to the pumas killed on the study area during the Colorado puma hunting season, adult 188 �male M133 was killed by a hunter in north GMU62 and adult females F25 and F94 were killed for depredation control reasons on the study area (Table 3). The harvest of 8 independent pumas on the study area was 15.4% (8/52*100) of the minimum count of 52 independent pumas, including 31 females and 24 males, determined by the research team during November 2010 to April 2011 (Table 4). Independent females and males comprised 25.0% (2/8*100) and 75.0% (6/8*100) of the harvest, respectively. This harvest structure was 5.7% (2/35*100) of the independent females and 35.3% (6/17*100) of the independent males. Considering the mortality of 3 other radio-collared adults (F25, F94, M133, Table 3), a harvest of 11 independent pumas was 21.2% (11/52*100) of the minimum number of independent pumas. The harvest composition of 4 females and 8 males was comprised of 36.4% (4/11*100) females and 63.6% (7/11*100) males. This harvest structure was 11.4% (4/35*100) of the independent females and 41.2% (7/17*100) of the independent males in the minimum count. The minimum count of 52 independent pumas in TY2 was slightly lower than the minimum count of 55 independent pumas in TY1 (Table 4). Minimum count TY2 = 52 independent pumas, including 35 females and 17 males. This count reflected the relatively high adult female survival rate and low adult male survival rate in TY1 (Logan 2010). Because the harvest quota of 8 independent pumas in TY1 resulted in a minimum count of 52 independent pumas in TY2 and is expected to result in a stable-toincreasing population trend, we decided to set the quota to harvest 8 independent pumas in the TY3 (2011-12) hunting season to emulate an approximate 15% harvest of independent pumas to achieve a stable to increasing population objective while also considering that a number of independent pumas in the study area population might be killed outside of the study area as in the TY1 and TY2 hunting seasons (Fig. 3). It is still too early in this research to tell if this harvest structure is resulting in a declining, stable, or increasing population trend. Hunter permits and survey: In TY2 mandatory permits with the voluntary survey attached were requested by 64 individual hunters, down from 79 individual hunters in TY1. Seventeen of the hunters requested a second permit after the first one expired after 14 days. Fifty-four hunters (84.4%) provided responses to the voluntary survey either by turning in the printed survey (n = 42) or providing information during follow-up telephone calls (n = 12) by principal investigator K. Logan. The remaining 10 hunters could not be contacted because either they did not have working phone numbers or they did not return calls. Of the respondents, 19 hunters indicated that they did not hunt on the study area. The proportion of the 54 respondents that hunted extrapolated to the total of 64 hunters (35/54 = 0.648) indicated that about 42 hunters took to the field for pumas on the study area during the 21-day hunting season. This was down from 67 hunters that probably hunted in TY1 (Logan 2010). Considering that 42 hunters were estimated to be afield, then 19% of the hunters harvested pumas (8/42*100) and 38% of hunters captured pumas (16/42*100; see captured and released pumas below and in Table 5). The 42 puma hunters that turned in the written volunteer survey were asked to answer, “Do you consider yourself a selective or non-selective hunter?” A selective hunter is one that purposely is hunting for a specific type of legal puma, such as a male, large male, or large female. A non-selective hunter is one that intends to take whatever legal puma is first encountered or caught, with no desire for sex or size. Selective hunter was indicated by 33 respondents. Of the remaining 9 hunters, 5 did not answer the question because they indicated that they did not hunt on the study area and 1 was an outfitter that did not hunt on the study area for himself (i.e., he hunted for his clients). One hunter indicated he was nonselective, and he killed a subadult female puma. Another hunter that did not answer the question killed a subadult female puma, too. The volunteer hunter survey also revealed that hunters treed pumas on the study area, but they chose not to kill them (Table 5). Those hunters reported they treed pumas 8 times, including 7 females and 1 subadult male. Of the 7 females 6 were described as adult, including 1 with at 189 �least 1 cub. Two of the adult females were marked with GPS collars (F3, F96). One female was either an adult or subadult. Hunters gave various reasons for not wanting to kill the pumas, including reasons based on puma sex, reproductive status, and size (Table 5). In an effort to better ascertain the vulnerability of sexes and age-stages (i.e., adult, subadult) of independent pumas to detection by puma hunters to address assumption 6 and hypothesis 6 (previously), the survey was changed in TY2 to ask hunters, “What was the sex of the lion that made the first set of tracks you encountered that were less than one day old?”. This question pertained to tracks that could be pursued by dogs and captured with a relatively high probability to allow the hunter an opportunity to harvest the puma. Associated with the question, we asked, “Did you pursue the lion to harvest it?” Hunters responses showed they encountered 30 puma tracks less than one day old. Of those, 20 tracks were of females, and 10 tracks were of males, indicating that during the hunting season females are more detectable than males by a ratio of 2:1, and similar to the sex structure of independent pumas in the minimum count on the study area which was 35 females and 17 males (ratio 2.06:1, Table 4). Of the female tracks, 3 female pumas were pursued by hunters with intent to harvest, of which 2 females were actually killed. Seventeen hunters indicated they did not pursue female tracks with intent to harvest; but, hunters captured and released 7 female pumas. Of the male tracks, 7 were pursued by hunters with intent to harvest, of which 6 were actually killed. Three hunters indicated they did not pursue to harvest 3 male tracks; but, 1 subadult male puma was captured and released. These preliminary survey and harvest data for TY2 indicate independent females were captured by hunters slightly more frequently than independent males by 9 to 7 (i.e., females = 2 harvested + 7 captured and released; males = 6 harvested + 1 captured and released). Moreover, hunters are choosing to kill males more frequently than females. This result is consistent with TY1 where hunters caught females slightly more frequently than males (i.e., 12 females, 10 males; females = 3 harvested + 9 captured and released; males = 5 harvested + 5 captured and released). Also in TY1, hunters indicated a preference to harvest males over females. This preliminary assessment from years TY1 and TY2 puma harvest and hunter survey data suggests that most hunters that captured pumas were selective and influenced harvest sex and age composition and that independent female pumas were detected by hunters at a higher rate than were independent male pumas. Segment Objective 2 After the design quota was filled, puma research teams immediately activated for capture operations with trained dogs. Two fully-staffed capture teams, one each detailed on the east and west slopes, systematically and thoroughly searched the study area to capture, sample, and GPS/VHF radiocollar pumas the remainder of winter and early spring 2010-11. These efforts along with cage trap efforts and hand-capturing cubs at nurseries maintained samples to quantify population sex and age structure, survival, and agent-specific mortality, and allowed determination of minimum population size on the study area. We made 70 puma captures of 54 to 55 individuals from August 2010 to July 2011 (Tables 6-11); 36 to 37 individual pumas were captured with dogs 52 times. Four pumas were captured in cage traps. Cubs were captured at nurseries 14 times. A total of 53 pumas were monitored with radio-telemetry from August 2010 to July 2011 (some of these had been collared in previous years). Trained dogs were used as our main method to capture, sample, and mark pumas on November 16, 2010 and from December 14, 2010 to April 22, 2011. Those efforts resulted in 81 search days, 291 total puma tracks detected of which 157 were ≤1 day old, 99 pursuits, and a total of 52 puma captures of 36-37 individual pumas (Table 6). This was the second year we deployed 2 fully-staffed hound capture teams in the treatment period. Search days with dogs was similar in both TY1 (86) and TY2 (81; Table 12). The frequency of tracks (tracks/day) encountered was higher in TY2 than the previous 6 winters. 190 �Also, pursuits increased over all previous years by 6 to 58, with the lowest number of pursuits occurring in the first year of this study (2004-05) when the puma population was probably at its lowest abundance on the study area. The capture rate was also the highest by 26 to 38 captures. Increased capture efforts and captures were probably the result of using 2 fully-staffed relatively more efficient houndsmen teams in TY2 even though the puma population had been reduced due to harvest just before our capture operations. Researchers in the two hound capture teams on November 16, 2010 and from December 14, 2010 to April 22, 2011 also recorded instances when the first tracks ≤1 day old of independent pumas were encountered on each search route each day to represent encounters with puma tracks that could be detected by houndsmen. The count was: 47 tracks of females, including 11 associated with cubs; 21 tracks of males; 4 tracks of cubs, and 1 track of unspecified sex. Except for 1 female and 1 male track ≤ 1 day old found on November 16, 2010, all other tracks ≤ 1 day old were found after the TY2 puma hunting season when 6 independent males and 2 independent females were harvested. Therefore, the harvested pumas were not present to make tracks for our researchers to observe. The loss of the 6 males and 2 females may reflect the slightly higher ratio of female:male tracks post-hunting season, 2.2:1 than was reported by hunters during the hunting season, 2:1 (previously, Segment Objective 1). Still, the ratios are similar and reflect the greater likelihood of encountering females than males. Puma capture efforts using ungulate carcasses and cage traps was sporadic from November 8, 2010 to April 18, 2011 (Table 10). We used 12 road-killed mule deer at 10 different sites. Two independent male pumas (M133, M153) were captured for the first time, and 2 adult females (F70, F137) were recaptured and re-collared. Pumas scavenged at 5 of 12 (41.66%) sites where deer carcasses were used for bait. We sampled 24 cubs, including 10 females and 14 males (Table 11). Nine females and 14 males were captured by us, of which 21 (7 females, 14 males) were radio-collared to monitor survival and agent-specific mortality (Appendix A). Female cub P1026 was sampled with a bio-dart only because she climbed a dangerous tree. Another female cub, P1030, was found dead, hit by a vehicle on state highway 62 in Leopard Creek. In addition to our direct puma captures with dogs November through April, we detected 18 radiocollared pumas that we were able to identify with GPS or VHF telemetry 28 times, thus, negating the need to capture those pumas directly with dogs (Table 6). Upon detecting puma tracks that were aged at ≤1 day old, we followed the tracks with a radio receiver in an effort to detect if the tracks might be of a puma wearing a functional collar. We assigned tracks to a collared individual if we received radio signals from a puma that we judged to be <1 km from the tracks and in direction of travel of the tracks. This approach allowed us to more efficiently allocate our capture efforts toward pumas of unknown identity on the study area, particularly unmarked pumas or pumas with non-functioning GPS- or VHF- radiocollars. Our search efforts throughout the study area from December 2010 to April 2011 also revealed the presence of at least 13 other independent pumas, which we classified as 9 females and 4 males. Three females and 2 males were treed by our hounds, but we could not handle the pumas because they climbed dangerous trees (Table 8). Of those, 2 females and 2 males were sampled with biodarts to obtain a tissue sample for genotyping the individuals. We could separate the activity of the other pumas from the GPSand VHF- collared pumas in time, space, and track size differences between females and males. One puma might have been F75 with a non-functional GPS collar. Moreover, females in association with cubs of different numbers, sizes, and locations enabled us to distinguish 4 adult females followed by 1 to 2 medium-to-large-size cubs. Some tracks we found of these pumas were too old to pursue (i.e., 2+ days old; probability of capture with the dogs was negligible). 191 �Our search and capture efforts during November 2010 through April 2011 and information from the puma hunting season in TY2 enabled us to quantify a minimum count of 52 independent pumas detected on the Uncompahgre Plateau study area, including 35 independent females and 17 independent males (Table 4). This count was based on the number of known radio-collared pumas, non-marked pumas harvested by hunters on the study area, observations of marked and non-marked pumas observed by researchers or treed and released by hunters on the study area, and puma tracks observed by researchers that could not be attributed to pumas with functioning radiocollars. The estimated age structure of independent pumas in November 2010 at the beginning of the puma hunting season in TY2 on the Uncompahgre Plateau study area is depicted in Figure 4. In addition to the independent pumas, we also counted a minimum of 39 cubs. Of the 52 independent pumas, 36 to 37 (69-71%) were marked and 15 to 16 (29-31%) were assumed to be unmarked animals. The abundance and sex structure of independent pumas on the east and west slopes of the study area were similar. The east slope count included 25 independent pumas (18 females, 7 males). The west slope count included 27 independent pumas (17 females, 10 males). Considering the minimum count of 52 independent pumas, a preliminary minimum density for the winter puma habitat area estimated at 1,671 km2 on the Uncompahgre Plateau study area was 3.11 independent pumas/100 km2. Segment Objective 3 During the past 6.7 years of this work we compiled data on puma reproduction that was not previously available on pumas in Colorado (Table 13). Puma reproduction data (i.e., litter size, sex structure, gestation, birth interval, proportion of females giving birth per year) were summarized for the reference period in Logan (2009). In TY2 we directly observed 6 litters in nurseries which were born in April (2), July (2), and August (2) 2010, each with 1 to 4 cubs, born to radio-collared females. Data on reproduction we observed in TY1 and TY2 were added to Table 13 which gives the reproductive chronology and information on mates of reproducing females. But those data will not be summarized again until the end of the treatment period. The proportion of radio-collared adult females giving birth from August 2010 to July 2011 biological year (TY2) was 0.56 (9/16), similar to TY1 (0.53, 8/15). Considering our 38 total observed litters with cubs 26 to 42 days old and 2 other litters confirmed by nurseries and nursling cub tracks with GPS-collared females (the latter include F111’s cubs caught later when 8.5 months old) (Table 13), the distribution of puma births by month since 2005 indicate births extending from March into September (Fig. 5). Births peak during May, June, July, and August involving 80% of the births (Fig. 5). The data indicate that the large majority of puma breeding activity occurred February through May (i.e., gestation averages about 90-92 days, Logan 2009). In comparison, Anderson et al. (1992:47-48) found on the Uncompahgre Plateau during 1982-1987 that of 10 puma birth dates 7 were during July, August, and September, 2 in October, and 1 in December, with most breeding occurring April through June. The 2 data sets indicated puma births on the Uncompahgre Plateau have occurred in every month except January and November (so far). As we gather more data on the puma births during the treatment period, we will examine the distributions in the reference and treatment periods separately. Segment Objectives 4 & 5 From December 8, 2004 (capture and collaring of the first adult puma M1) to July 31, 2011, we radio-monitored 19 adult male and 30 adult female pumas to quantify survival and agent-specific mortality rates (Table 14). Survival and agent-specific mortality of adult pumas were summarized for the reference period in Logan (2009). Preliminary estimates of adult puma survival rates in the absence of sport-hunting during the reference period indicated high survival, with adult male survival generally higher than adult female survival (Table 15). Preliminary adult puma survival for TY1 and TY2 are also shown in Table 15. So far, adult male survival is substantially lower in the treatment period than in the reference period and adult female 192 �survival may be similar in both periods. These characteristics may be indicative of hunter selection for male pumas (previously in Segment Objective 1). But, no conclusions should be drawn with results from only 2 years in the treatment period. The primary research interests include how survival rates influence population growth rates and the strength of factors associated with survival and mortality. This is what ultimately allows us to evaluate the effect of a 15% harvest level on independent pumas for our population management assumptions when the goal is a stable to increasing population. Human-related causes of mortality dominated deaths of marked adult pumas in TY2, including: sport-hunting harvest (4 males- M32, M55, M90, M133) and depredation control (1 male- M134; 2 females- F25, F94) (Table 14). We have radio-monitored 19 pumas, including 6 females and 13 males, in the subadult age-stage (independent pumas <24 months old) (Table 16). Four died before reaching adulthood, indicating a preliminary finite survival rate of 0.789 (i.e., 15/19). All 4 subadults apparently died of natural causes. F66 died at 23 months old of trauma to internal organs that caused massive bleeding attributed to trampling by an elk or mule deer. M99 died at about 16 months old; punctures to his skull were consistent with canine bites from another puma and suggested intra-species strife as cause of death. M115 died at about 14 months old due to complications of a broken left foreleg, cause unknown. This injury probably affected his ability to efficiently kill prey. F143 was killed and eaten by a male puma while in competition for an elk carcass that one of the pumas killed. We need to increase our efforts to acquire larger samples of male and female radio-monitored subadult pumas to acquire reliable estimates of their survival. Harvest data along with our capture and radiotelemetry data provided additional information on fates of 26 marked pumas, 22 males and 4 females. Of those, 21 (2 females, 19 males) were initially captured and marked as cubs, and 5 (2 females, 3 males) were captured and marked as subadults on the Uncompahgre Plateau puma study area (Table 17). Twenty males were killed away from the study area by hunters at linear distances (i.e., from initial capture sites to kill sites) ranging from about 20 to 370 km. Two males with extreme moves were killed in the Snowy Range of southeastern Wyoming (369.6 km) and the Cimarron Range of north-central New Mexico (329.8 km). Female F52 was treed and released by hunters in December 2008 and 2009 south of Powderhorn, Colorado, indicating that she probably established an adult home range there. Three males marked initially as cubs born on the study area (M67, M87, M92) dispersed from their natal ranges and were recaptured as adults on the study area. All were born on the east slope of the Uncompahgre Plateau and moved to the west slope. These pumas represent dispersal moves on and from the Uncompahgre Plateau. Eighteen of the 26 pumas had reached adult ages ranging from 24 to 55 months old. A preliminary estimate of cub survival during the reference period was summarized in Logan 2009 using 36 radio-collared cubs (16 males, 20 females) marked at nurseries when they were 26 to 42 days old. In that summary, estimated survival of cubs to one year of age was 0.53.. The major natural cause of death in cubs, where cause could be determined, was infanticide and cannibalism by other, especially male, pumas. In TY2 we monitored the fates of 23 radio-collared cubs (Appendix A). Six of the cubs (3 females, 3 males) were known to have died. Three cubs with their mother F94 were killed for depredation control to protect a commercial domestic elk operation. Three other cubs died of natural causes. M130 died from a cause associated with injury to his right shoulder during the first move away from his nursery with F96 and 3 other siblings. Two cubs, M139 and F148 (offspring of F8), died of infanticide and cannibalism by a female or subadult male puma. A greater number of cubs over a longer period of time must be sampled before estimating cub survival and agent-specific mortality rates in the treatment period. 193 �In addition, a non-marked female puma cub was struck and killed by a vehicle on state highway 62 in Leopard Creek on the south boundary of the study area on February 16, 2011. This mortality made the thirteenth puma death recorded due to vehicle collision on the study area since 2004 (Table 18). Five of the 13 pumas were marked, including 3 adults with GPS/VHF collars. Those 3 adults died during the first year of the treatment period. Thirty-two adult pumas (23 females, 9 males) have worn GPS collars since this project began in 2004 (Table 19). Over 55 thousand GPS locations have been obtained for studies on puma behavior, social organization, population dynamics, movements, habitat use and puma-human relations in collaboration with colleagues in Mammals Research and Colorado State University. Segment Objective 6 As an extension of our pilot puma camera grid project in 2009 (Logan 2010), we decided to explore the feasibility of attracting wild pumas to a rub station to obtain tissue non-invasively for potential use in a genotype mark-recapture structure for estimating abundance. Our question was basic to such a structure. What might be expected detection probabilities for wild pumas at scent/rub stations? This work operated on minimal resources consisting of 9 trail cameras, opportunistically available scents, and the field work was done primarily by volunteer Linda Sweanor. Thus, we consider this work exploratory to inform how we might continue in future efforts. Our approach was simple, reflecting available resources. We placed cameras and scent stations with hair capture devices at sites where we thought we could maximize encounters with pumas. Cameras were Reconyx ™ with passive infrared motion detectors and night time infrared illumination each set to take photos each second after the camera was triggered. Our previous approach to locating stations using only trail cameras in a grid resulted in very high detection probabilities of marked pumas during our pilot camera grid project in 2009 (Logan 2010). This allowed us to photographically record behavior of pumas at scent/rub stations. Scents used included: beaver castorium, catnip oil, MT Lynx ™, Obsession for Men ™, Spotted Fever ™, and one combination of catnip oil and Spotted Fever™. Scent/rub stations, camera operation, and camera digital data were examined at approximately 2 to 4 week intervals. At those times, each rub pad (i.e., rub device and carpet swatch) was treated with a different available scent if a puma had visited the scent/rub station and regardless of the puma’s response to the scent/rub station. If no pumas visited the rub/scent station, then the carpet swatch was re-treated with the same scent used the previous weeks. Our aim was to expose as many individual pumas as possible to different scents and record their behaviors. We defined the sampled population of pumas to include only those pumas recorded by the cameras. All pumas photographed passed ≤5 m of the scent/rub station. We defined a maximum detection probability for a particular scent as the number of individual pumas that were photographically recorded at scent/rub stations with a particular scent that rubbed and deposited hair that could be collected divided by the total number of individual pumas that were photographically recorded at scent/rub stations with a particular scent. We did not have resources to attempt to assess quality of the DNA and individual puma genotype accuracy; thus, detection was considered to be maximum for this exploratory assessment only. In addition, this design did not consider other pumas in the environment that were not detected by the camera/scent/rub stations. Non-detected pumas in the area of the camera/scent/rub stations and DNA that provided inaccurate genotypes would lower the detection probability. Detailed notes were kept on visits and behaviors of all pumas and other wildlife that were recorded by cameras. Camera scent/rub stations were maintained from November 20, 2010 to August 14, 2011. A total of 9 stations were used. All information in Tables 20, 21 and Appendix B should be considered exploratory and preliminary. Thirty-nine puma visit events were photographed, including one family of 4 194 �pumas (i.e., mother with 3 cubs). Beaver castorium produced the highest maximum detection probability, 0.667, (Table 20). Detection was variable among the scents used and among pumas and appeared to be substantially lower for male than for female pumas (Table 21). These results indicate that more work needs to be done in a more structured manner to sample a greater number of known individual wild pumas, a variety of scents, and with an analysis of DNA quality and genotype accuracy. SUMMARY Manipulative, long-term research on puma population dynamics, effects of sport-hunting, and development and testing of puma enumeration methods began in December 2004. After 6.7 years of effort 153 unique pumas have been captured, sampled, marked, and released. Using these animals, we monitored fates of pumas in all sexes and age stages, including: 30 adult females, 19 adult males, 6 subadult females, 12 subadult males, 39 female cubs, 53 male cubs (some individuals occur in more than one age-stage). Data from the marked animals were used to quantify puma population characteristics and vital rates in a reference period without sport-hunting off-take as a mortality factor from December 2004 to July 2009. Puma population characteristics and vital rates in a reference condition allowed us to develop a puma population model, and to use population data and modeling scenarios to conduct a preliminary assessment of CPW puma management assumptions and guide directions for the remainder of the puma research on the Uncompahgre Plateau. Moreover, our data and model provide tools currently useful to CPW wildlife biologists and managers for assessing puma harvest strategies. The 5-year treatment period began August 2009 in which sport-hunting is a mortality factor. The treatment period will be a population-wide test of CPW puma management assumptions. Now 2 years of the treatment period are complete (TY1, TY2). Although some data support CPW puma management assumptions, it is still too early in this research to adequately test the assumptions and attendant hypotheses. Although the assumption and hypothesis on harvest structure and hunter selection is not supported with the first 2 years of data in the treatment period, this could change with a substantial change in abundance and sex structure of independent pumas available for hunting in TY3 to TY5. The puma harvest quota for TY3 will be 8 independent pumas, and the hunters will be surveyed again. To improve data on puma population vital rates, attention will be given to increasing radio-collared sample sizes across the various life stages and sexes. We will continue to explore methods for estimating puma abundance with accurate and affordable methods. Furthermore, we will continue collaboration with colleagues on investigations of puma population parameter estimation, abundance estimation, puma movements, puma habitat modeling and mapping, and puma-human relations. All of these efforts should enhance the Colorado puma research and management programs. 195 �LITERATURE CITED Anderson, A. E. 1983. A critical review of literature on puma (Felis concolor). Colorado Division of Wildlife Special Report No. 54. _____, D. C. Bowden, and D. M. Kattner. 1992. The puma on Uncompahgre Plateau, Colorado. Technical Publication No. 40. Colorado Division of Wildlife, Denver. Anderson, C. R., Jr., and F. G. Lindzey. 2005. Experimental evaluation of population trend and harvest composition in a Wyoming cougar population. Wildlife Society Bulletin 33:179-188. Colorado Division of Wildlife 2002-2007 Strategic Plan. 2002. Colorado Department of Natural Resources, Division of Wildlife. Denver. Colorado Division of Wildlife. 2007. Colorado mountain lion management data analysis unit revision and quota development process. Colorado Division of Wildlife, Denver. Conover, W. J. 1999. Practical nonparametric statistics. John Wiley & Sons, Inc., New York. Cooch, E., and G. White. 2004. Program MARK- a gentle introduction, 3rd edition. Colorado State University, Fort Collins. Culver, M., W. E. Johnson, J. Pecon-Slattery, and S. J. O’Brien. 2000. Genomic ancestry of the American puma (Puma concolor). The Journal of Heredity 91:186-197. Currier, M. J. P., and K. R. Russell. 1977. Mountain lion population and harvest near Canon City, Colorado, 1974-1977. Colorado Division of Wildlife Special Report No. 42. Heisey, D. M., and T. K. Fuller. 1985. Evaluation of survival and cause specific mortality rates using telemetry data. Journal of Wildlife Management 49:668-674. Karanth, K. U., and J. D. Nichols. 2002. Monitoring tigers and their prey: A manual for researchers, managers and conservationists in tropical Asia. Centre for Wildlife Studies, Bangalore, India. Koloski, J. H. 2002. Mountain lion ecology and management on the Southern Ute Indian Reservation. M. S. Thesis. Department of Zoology and Physiology, University of Wyoming, Laramie. Kreeger, T. J. 1996. Handbook of wildlife chemical immobilization. Wildlife Pharmaceuticals, Inc., Fort Collins, Colorado. Laundre, J. W., L. Hernandez, D. Streubel, K. Altendorf, and C. L. Lopez Gonzalez. 2000. Aging mountain lions using gum-line recession. Wildlife Society Bulletin 28:963-966. Logan, K. A., E. T. Thorne, L. L. Irwin, and R. Skinner. 1986. Immobilizing wild mountain lions (Felis concolor) with ketamine hydrochloride and xylazine hydrochloride. Journal of Wildlife Diseases. 22:97-103. _____, L. L. Sweanor, J. F. Smith, and M. G. Hornocker. 1999. Capturing pumas with foot-hold snares. Wildlife Society Bulletin 27:201-208. _____, and L. L. Sweanor. 2001. Desert puma: evolutionary ecology and conservation of an enduring carnivore. Island Press, Washington, D.C. _____. 2009. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Wildlife, Fort Collins. _____. 2010. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Wildlife, Fort Collins. Murphy, K., M. Culver, M. Menotti-Raymond, V. David, M. G. Hornocker, and S. J. O’Brien. 1998. Cougar reproductive success in the Northern Yellowstone Ecosystem. Pages 78-112 in The ecology of the cougar (Puma concolor) in the Northern Yellowstone ecosystem: interactions with prey, bears, and humans. Dissertation, University of Idaho, Moscow. Ott, R. L. 1993. An introduction to statistical methods and data analysis. Fourth edition. Wadsworth Publishing Co., Belmont, California. Pierce, B. K., V. C. Bleich, and R. T. Bowyer. 2000. Social organization of mountain lions: does land a tenure system regulate population size? Ecology 81:1533-1543. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. 196 �Pojar, T. M., and D. C. Bowden. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550-560. Ross, P. I., and M. G. Jalkotzy. 1992. Characteristics of a hunted population of cougars in southwestern Alberta. Journal of Wildlife Management 56:417-426. Seidensticker, J. C., M. G. Hornocker, W. V. Wiles, and J. P. Messick. 1973. Mountain lion social organization in the Idaho Primitive Area. Wildlife Monographs No. 35. Stoner, D. C. 2004. Cougar exploitation levels in Utah: implications for demographic structure, metapopulation dynamics, and population recovery. Master of Science Thesis. Utah State University. Sweanor, L. L., K. A. Logan, J. W. Bauer, B. Milsap, and W. M. Boyce. 2008. Puma and human spatial and temporal use of a popular California state park. Journal of Wildlife Management 72:10761084. Williams, B. K., J. D. Nichols, and M. J. Conroy. 2001. Combining closed and open mark-recapture models: the robust design. Pages 523-554 In Analysis and management of animal populations. Academic Press, New York. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 (Suppl):S120-S139. Prepared by: ________________________ Kenneth A. Logan, Wildlife Researcher 197 �Table 1. Projected puma population growth modeled from a minimum count of independent pumas during winter 2007-08 reference period year 4 (RY4). Treatment period year 1 (TY1), shaded in gray, indicates the results used to derive a quota of 8 independent pumas, representing 15% of the independent pumas (from Logan 2009). Harvest Level No harvest. Projected Minimum Puma Population Size Adult Subadult Female Male Female Male 16 8 5 4 18 10 9 8 23 14 8 8 27 17 11 10 32 22 12 11 38 27 15 14 44 32 17 16 Year RY4 RY5 TY1 TY2 TY3 TY4 TY5 Independent Pumas Cub 20 33 42 49 58 69 81 Total 33 45 53 64 77 92 110 Lambda 1.37 1.17 1.22 1.20 1.20 1.19 Table 2. Pumas harvested by sport-hunters in Treatment Year 2 (TY2) on the Uncompahgre Plateau Study Area, Colorado, November 22 to December 12, 2010. Puma sex Age (yr.) Date of kill Location/UTM 1.5 Previous M/F I.D. or specimen P no. if not marked P1020 F 11/22/2010 M 2.3 M90 11/23/2010 M 6.3 M55 11/25/2010 M 3.5 P1023 11/26/2010 F 1.5 F108 11/29/2010 M 3 P1032 12/1/2010 M 9.2 M32 12/2/2010 M 1.5 P1024 12/12/2010 McKenzie Butte/ 13S,255947E,4238054N McKenzie Creek/ 13S,257237E,4238244N Spring Creek Canyon/ 13S,239181E,4248300N San Miguel River Canyon/ 12S,736610E,4230762N Cushman Creek/ 12S,752013E,4263883N San Miguel Canyon (E)/ 12S,729439E,4236264N McKenzie Creek/ 13S,257722E,4239169N Tabeguache Creek/ 12S,735100E,4249600N 198 Hunter/status Micah Brogden/ Resident Jack Flowers/ Resident Dennis Rawley/ Non-resident Michael Compton/ Resident Richard Fischer/ Resident Nathan Nickle/ Non-resident Mat Iverson/ Resident Mark Puerschner/ Non-resident �Table 3. Three other independent GPS-collared adult pumas in the minimum count for the Uncompahgre Plateau Study Area that died during the 2010-11 Colorado puma hunting season. Puma sex (M or F) Date of kill Place of kill/UTM Hunter/status/other cause M133 Age (yr.) 3.5 12/1/2010 F94 5 2/1/2011 F25 10 2/5/2011 Dry Fork Escalante Canyon 12S,731720E,4278128N Happy Canyon 13S,246976E,4255108N Pleasant Valley 13S,252703E,4225101N Trent Schloegel/ Non-resident Killed by A.P.H.I.S.W.S. agent for depredation on domestic elk Killed by ranch-hand because puma was seen in vicinity of cattle Table 4. Minimum count of pumas based on numbers of known radio-collared pumas, visual observations of non-marked pumas, harvested non-marked pumas, and track counts of suspected non-marked pumas on the study area during September 2009 to April 2010 of Treatment Year 1 (TY1) and November 2010 to April 2011 (TY2), Uncompahgre Plateau study area, Colorado. Treatment Year (TY) Study Area region TY1 East slope West slope subtotals TY2 Adults Female Male Subadults Female Male Female Cubs Male 16 10 1 1 1 4 14 10 0 3 3 3 30 20 1 4 4 7 Total Independent Pumas = 55, including 31 females, 24 males. Cubs = 20-25 East slope 15 5 3 2 7 9 West slope 15 7 2 3 2 5 subtotals 30 12 5 5 9 14 Total Independent Pumas = 52, including 35 females, 17 males. Cubs = 39 Unknown sex 4-8* 5-6 9-14 7 9 16 *One adult non-marked female puma was killed by a hunter in Roubideau Canyon. The female puma was lactating, indicating she had nurslings. Up to 4 cubs were assumed to be in the litter. 199 �Table 5. Pumas captured and released by sport-hunters in Treatment Year 2 (TY2) on the Uncompahgre Plateau Study Area, Colorado, November 22 to December 12, 2010. Data are from puma hunter responses in 54 voluntary surveys, including: 42 original surveys on printed voluntary permits and 12 telephone contacts with hunters that did not return printed surveys on permits. Total response rate from 64 individual hunters was 84.4% (54/64 = 0.894*100). Puma sex/age stage/mark F/adult/F3 by collar, no eartags, confirmed with GPS and VHF data F/adult/F96 by GPS collar, confirmed with GPS data F/adult/none Date of capture 11/25/2010 Capture location Hunter name Spring Creek Canyon Justin Hill 11/27/2010 Dolores Canyon Justin Hill Did not pursue the female puma with intent to harvest it. 11/23 to 27/2010 McKenzie Creek (west) Tommie Buckington guided by Ryan Weimer F /adult and cub/none F/adult or subadult/none 11/22 to 30/2010 11/30/2010 Dolores River Canyon Dolores Creek (east) Ryan Weimer Female puma with evidence of suckling on nipples. Did not want to kill a female puma with cubs. Cubs not actually seen. Not legal to kill a female puma with cubs. Did not pursue the female puma with intent to harvest it. F/adult/none 12/11/2010 F/adult/none 11/22 to 12/12/2010 11/22 to 12/12/2010 Sims Mesa to Happy Canyon Dry Park to Big Bucktail Creek San Miguel Canyon above Goodenough Gulch M/subadult/none John Akerberg & Kris Brown guided by Ben Harris Wade Wilson Sam Sickels Ty Sickels 200 Reason for releasing the puma given by hunter Did not pursue the female puma with intent to harvest it. Did not pursue the female puma with intent to harvest it. Did not pursue the female puma with intent to harvest it. Did not want to harvest a subadult male; guessed weight 125 lb. �Table 6. Summary of puma capture efforts with dogs from November 16, 2010 to April 22, 2011, Uncompahgre Plateau, Colorado. Month November No. Search Days 1 December 11 January 22 February 20 March 21 No. & type of puma tracks founda,b 2 tracks: 1 male, 1female, 0 cub Tracks ≤1 day old: 1 male, 1 female, 0 cub 35 tracks: 7 male, 17 female, 9 cub, 2 undetermined independent pumas Tracks ≤1 day old: 2 male, 3 female, 2 cub 109 tracks: 15 male, 60 female, 30 cub, 4 undetermined independent pumas Tracks ≤1 day old: 5 male, 25 female, 24 cub No. & type of pumas pursued 1 pursuits: 1 male, 0 female , 0 cub No. & I.D. or type of pumas captured, observed, or identified 1 puma captured: M90 recaptured and fit with adult-size VHF collar (cub collar had quit/shed a long time previously). 5 pursuits: 1 male, 3 female, 1 cub 3 pumas captured 3 times: AFP1025 (biodart, dangerous tree), Adult F (not handled due to dangerous tree), M134 cub. In addition, adult female F118, her 3 cubs M126, M127, M128, and adult male M67 were associated with tracks by VHF telemetry. 29 pursuits: 5 male, 14 female, 10 cub 65 tracks: 13 male, 28 female, 24 cub Tracks ≤1 day old: 10 male, 21 female, 22 cub 30 pursuits: 9 male, 11 female, 10 cub 18-19 pumas captured 20 times: F135, F104, AFP1029 (bio-darted, dangerous tree), F136, F137, F28, F23 captured twice, Sub./AMP1028 (possibly M138), M138, and cubs F111's two cubs (not handled, dangerous trees), M112, FP1026, MP1027, M134, M112, F140, M141, M142. In addition, adult females F111, F3 (twice), F96, F136, F116 (twice), and cubs F140, M141, M142 were associated with tracks by VHF telemetry. 14 pumas captured 15 times: F137, F70, F23, F143, adult F (not handled, dangerous tree), F24 (twice), independent M (not handled, dangerous tree), M138, M87, subMP1031 (bio-darted, dangerous tree), and cubs M150 (twice), P1026, M151. In addition, adult females F96, F70 (twice), F23, F118, F143, adult male M67, and cubs M141, M142 were associated with tracks by VHF telemetry. 7 pumas captured 7 times: F111, F3, F72, F145, F146, M144, and cub F152. In addition, subadults M144, F145, and cub M142 were associated with tracks by VHF telemetry. 5 pumas captured 5 times: F24, M92, and cubs F140, M141, F147. In addition, adult M67 was associated with tracks with VHF telemetry. 73 tracks: 26 male, 22 pursuits: 4 male, 30 female, 17 cub 11 female, 7 cub Tracks ≤1 day old: 9 male, 12 female, 7 cub April 6 16 tracks: 3 male, 12 pursuits: 6 female, 7 cub 2 male, 3 female, Tracks ≤1 day old: 7 cub 2 male, 4 female, 7 cub 81 300 tracks: 99 pursuits: 36 to 37 individual pumas were captured 52 TOTALS 65 male, 22 male, times with aid of dogs. In addition, 18 radio142 female, 42 female, collared pumas were detected 28 times by tracks 87 cub, 35 cub and identified with VHF telemetry ≤1 km from 6 undetermined the tracks. Tracks ≤1 day old: 29 male 68 female 62 cub a Puma hind-foot tracks with plantar pad widths >50 mm wide are assumed to be male; ≤50 mm are assumed to be female (Logan and Sweanor 2001:399-412). b Researchers also recorded instances when the first puma tracks ≤1 day old were encountered on each search route each day. The count was: 47 tracks of females, including 11 associated with cubs; 21 tracks of males; 4 tracks of cubs, and 1 track of undetermined sex. 201 �Table 7. Adult and subadult pumas captured for the first time, sampled, tagged, and released from November 2010 to April 2011, Uncompahgre Plateau, Colorado. Puma I.D. M133 F135 F136 F137 M138 F143 M144 F145 F146 M153 Sex M F F F M F M F F M Estimated Age (mo.) 42 27 30 24 18 24 18 18 18 18 Mass (kg) Capture date 11/12/2010 1/1/2011 1/20/2011 1/21/2011 1/26/2011 2/15/2011 3/7/2011 3/8/2011 3/8/2011 4/12/2011 70 38 41 35 50 45 63 42 36 55 Capture method Cage trap Dogs Dogs Dogs Dogs Dogs Dogs Dogs Dogs Cage trap Location Roubideau Canyon Dry Creek Basin McKenzie Creek (east) Dry Creek Basin Spring Creek Canyon San Miguel Canyon Little Big Bucktail Creek North Fork Cottonwood Creek Tomcat Creek McKenzie Mesa Table 8. Pumas that were captured and observed with aid of dogs, some of which were biopsy-darted and given specimen numbers (e.g., P1025), but were not handled at that time for safety reasons, December 2010 to April 2011, Uncompahgre Plateau, Colorado. Puma sex & I.D. F P1025 Age stage or months adult Capture date 12/14/2010 Location Comments Monitor Mesa, Roubideau Canyon Puma climbed dangerous tree. Biopsy-darted to obtain tissue sample for genotype. Apparent mother of cub M134. Puma climbed dangerous tree momentarily, then left the tree and took refuge in a deep narrow hole where we could not gain access to her to change the non-functional GPS collar. Puma climbed dangerous tree. Cub of F111. Two cub tracks found; one was M151 marked 2/24/2011. Puma climbed dangerous tree. Cub of F111. Two cub tracks found; one was M151 marked 2/24/2011. Puma climbed dangerous tree. Biopsy-darted to obtain tissue sample for genotype. Probably offspring of F70; sibling of M112 and M150. Puma climbed dangerous tree. Biopsy-darted to obtain tissue sample for genotype. Probably M150, offspring of F70; sibling of M112 and P1026. Puma climbed dangerous tree. Biopsy-darted to obtain tissue sample for genotype. Possibly M138. Puma climbed dangerous tree. Biopsy-darted to obtain tissue sample for genotype. Puma climbed dangerous tree. Too high to biopsy dart. Puma climbed dangerous trees. Biopsy-darted to obtain tissue sample for genotype. First dart missed puma; puma left tree and evaded dogs on bare ground. Puma climbed dangerous tree. Identified by eartag. F28 adult 1/1/2011 San Miguel Canyon Unknown none cub 7 1/2/2011 Piney Creek Unknown none cub 7 1/2/2011 Piney Creek F P1026 cub 18 1/6/2011 Happy Canyon M P1027 cub 18 1/7/2011 Happy Canyon M P1028 F P1029 M none M P1031 F none M92 adult 1/12/2011 Roubideau Canyon adult 1/15/2011 Dolores Canyon (E) adult 2/3/2011 West Fork Dry Creek Basin subadult 2/17/2011 adult 2/21/2011 adult 4/22/2011 North Fork Cottonwood Creek San Miguel Canyon above Horsefly Creek McKenzie Canyon (W) 202 �Table 9. Pumas recaptured with dogs and cage traps January 2011 to April 2011, Uncompahgre Plateau, Colorado. Puma I.D. F28 Recapture Date 1/1/2011 Mass (kg) Observed Estimated Age (mo.) 94 Capture Method/ Location Dogs/East Fork Dry Creek Basin F23 1/6/2011 Observed 77 Dogs/San Miguel Canyon above Pinyon M112 1/6/2011 Observed 17 Dogs/Happy Canyon M134 F104 M112 1/8/2011 1/11/2011 1/24/2011 Observed 36 42 19 116 17 Dogs/Potter Basin Dogs/Roatcap Canyon Dogs/Horsefly Canyon F23 1/26/2011 45 77 F137 2/1/2011 Observed 25 F23 M87 2/8/2011 2/9/2011 Observed 65 78 31 M138 F70 2/9/2011 2/18/2011 Observed Observed 19 70 F70 2/21/2011 46 70 F24 2/22/2011 38 119 F24 2/24/2011 Observed 119 Dogs/San Miguel Canyon below Pinyon Dogs/East Fork Dry Creek Dogs/Tomcat Creek Dogs/Big Bucktail Creek Dogs/Roatcap Canyon Dogs/Spring Creek Canyon Cage trap/Pinyon Hills, Happy Canyon Dogs/Dry Park, Cottonwood Creek Dogs/San Miguel Canyon above Pinyon F111 3/4/2011 41 41 F3 3/15/2011 Observed 116 F72 3/18/2011 Observed 60 Dogs/Fisher Creek F140 4/1/2011 8 Dogs/Coal Canyon Dogs/Cushman Canyon Dogs/Spring Creek Canyon 22 M141 4/1/2011 Observed 8 Dogs/Coal Canyon F137 4/11/2011 42 27 F24 4/21/2011 Observed 121 Cage trap/Dry Creek Basin Dogs/McKenzie Canyon (west) M92 4/22/2011 Observed 32 Dogs/McKenzie Canyon (west) 203 Process F28 first climbed dangerous tree, left the tree, then entered deep narrow hole; could not be handled to replace non-functional GPS collar. F23 took refuge in elevated crevice on canyon wall; could not be handled to replace non-functional GPS collar. Observed puma bayed on the ground, fighting the dogs. Dogs caught and puma allowed to escape. Not handled. GPS collar replaced with VHF radiocollar. M112 fit with VHF radiocollar with expansion link. Replaced non-functional GPS collar with new VHF radiocollar. Observed and released. Observed and released. M87 fit with VHF radiocollar. Observed and released. F70 climbed dangerous tree; could not be handled. Old GPS collar replaced with new GPS collar. Replaced non-functional GPS collar with new VHF radiocollar. F24 observed and released. Effort to capture 2 cubs failed; lost tracks on bare ground in ledges. Old GPS collar replaced with new GPS collar. F3 climbed dangerous tree. Could not be handled to replace old, working GPS collar. F72 climbed dangerous tree. Could not be handled to replace non-functional GPS collar. Recollared with large expandable cub collar to replace the collar that was shed earlier. M141 left tree before we could handle him; escaped the dogs on bare ground. Replaced VHF radiocollar with GPS collar. F24 observed and released. Captured, sampled, and radio-collared cub F147 (one of two cubs). M92 climbed dangerous tree. Could not be handled to fit with radiocollar. �Table 10. Summary of puma capture efforts with cage traps from November 8, 2010 to April 18, 2011, Uncompahgre Plateau, Colorado.* Month November No. of Sites 6 Carnivore activity & capture effort results Captured adult male puma M133 that scavenged mule deer doe carcass in Roubideau Canyon 11/12/2010. Set cage trap in mouth Linscott Canyon on 11/18/2010 in effort to capture male puma that scavenged mule deer carcass; but, the male puma did not return. January 0 All capture efforts with dogs. February 1 Puma F70 was recaptured at a mule deer kill on 2/21/2011 on Pinyon Hills, Happy Canyon. March 3 No pumas scavenged the mule deer carcasses. April 4 Puma F137 was recaptured when she returned to scavenge on a mule deer buck carcass in Dry Creek Basin on 4/11/2011. Puma M153 was captured when he returned to scavenge a mule deer doe carcass on McKenzie Mesa on 4/12/2011. Puma F70 scavenged a mule deer buck carcass on 4/16-17/2011; no effort was made to recapture her. * We used 12 road-killed mule deer at 10 different sites. Of the road-killed deer baits, 5 of 12 (41.66%) were scavenged by pumas. Table 11. Puma cubs sampled August 2010 to July 2011 on the Uncompahgre Plateau Puma Study area, Colorado. a Cub I.D. Sex Estimated birth datea Estimated age at capture (days) Mass (kg) Mother Estimated age of mother at birth of this litter (mo) M122b F123 F124 M125 M126 M127 M128 F129 M130 M131 F132 M134 M139 F148 F140 M141 M142 F147c F149 M150d P1026d M151e F152f P1030 M F M M M M M F M M F M M F F M M F F M F M F F 7/8/2010 7/15/2010 7/15/2010 7/15/2010 8/8/2010 8/8/2010 8/8/2010 8/21/2010 8/21/2010 8/21/2010 8/21/2010 6/2009 4/18/2011 4/18/2011 8/2010 8/2010 8/2010 9/2010 4/22/2011 8/31/2009 8/31/2009 6/16/2010 6/16/2010 8/2010 35 29 29 29 28 28 28 35 35 35 35 547 36 36 152 152 152 214 45 547 516 253 261 183 2.2 1.8 1.9 2.0 1.6 1.9 2.0 1.6 1.9 1.8 1.6 64 2.25 2.25 13 15 14 16 2.9 53 NH 23 25 21 F104 F94 F94 F94 F118 F118 F118 F96 F96 F96 F96 Unknown F8 F8 Unknown Unknown Unknown F24 F23 F70 F70 F111 F93 Unknown 110 60 60 60 27 27 27 55 55 55 55 Unknown 95 95 Unknown Unknown Unknown 114 80 52 52 32 90 Unknown Estimated age of cubs sampled at nurseries is based on the starting date for GPS location and radio-telemetry foci for mothers at nurseries, and development characteristics of cubs caught with mothers without radiocollars or mothers with non-functioning radiocollars. b Three sets of cub tracks (including M122) observed in association with F104 when she was recaptured 1/11/2011 in Roatcap Canyon. c Three sets of cub tracks (including F147) observed in association with F24. d Cubs M150 and P1026 are siblings of M112. F70 had at least 3 cubs in the litter. Birth date based on GPS data on F70’s collar. e Two cubs were observed in association of F111. f F93 had two cubs in this litter. 204 �Table 12. Summary of puma capture efforts with dogs, December 2004 to April 2011, Uncompahgre Plateau, Colorado. Period Dec. 2, 2004 to May 12, 2005 Nov. 21, 2005 to May 26, 2006 Nov. 13, 2006 to May 11, 2007 Nov. 19, 2007 to April 24, 2008 Dec. 9, 2008 to April 30, 2009 Dec. 15, 2009 to April 30, 2010 Nov. 16 and Dec. 14, 2010 to April 22, 2011 Track detection effort 109/78 = 1.40 tracks/day 35/78 = 0.45 pursuit/day Puma capture effort 14/78 = 0.18 capture/day Effort to capture an independent puma for the first time 11 pumas captured for first time 11/78 = 0.14 capture/day 78/14 = 5.57 day/capture 14/82 = 0.17 capture/day 78/11 = 7.09 day/capture 149/82 = 1.82 tracks/day 78/35 = 2.23 day/pursuit 43/82 = 0.52 pursuit/day 177/78 to 182/78 = 2.27-2.33 tracks/day 82/43 = 1.91 day/pursuit 45/78 to 47/78 = 0.58-0.60 pursuit/day 82/14 = 5.86 day/capture 22/78 = 0.28 capture/day 78/47 to 78/45 = 1.66-1.73 day/pursuit 49/77 = 0.64 pursuit/day 78/22 = 3.54 day/capture 78/7 = 11.14 day/capture 20/77 = 0.26 capture/day 7 pumas captured for first time 7/77 = 0.09 capture/day 77/20 = 3.85 day/capture 24/71 = 0.34 capture/day 77/7 = 11.00 day/capture 217/77 to 218/77 = 2.82-2.83 tracks/day Pursuit effort 198/71 to 202/71 = 2.79-2.84 tracks/day 77/49 = 1.57 day/pursuit 75/71 to 78/71 = 1.06-1.10 pursuit/day 266/86 = 3.09 tracks/day 71/75 to 71/78 = 0.91-0.95 day/pursuit 93/86 = 1.08 pursuit/day 300/81 = 3.70 tracks/day 7 pumas captured for first time 7/82 = 0.08 capture/day 82/7 = 11.71 day/capture 7 pumas captured for first time 7/78 = 0.09 capture/day 9 pumas captured for first time 9/71 = 0.13 capture/day 71/24 = 2.96 day/capture 71/9 = 7.89 day/capture 26/86 = 0.30 capture/day 9 pumas captured for first time 9/86 = 0.11 capture/day 86/93 = 0.92 day/pursuit 99/81 = 1.22 pursuit/day 86/26 = 3.31 day/capture 52/81 = 0.64 capture/day 86/9 = 9.56 day/capture 81/99 = 0.82 day/pursuit 81/52 = 1.56 day/capture 205 15 pumas captured for first time 15/81 = 0.18 capture/day 81/15 = 5.40 day/capture �Table 13. Individual puma reproduction histories, Uncompahgre Plateau, Colorado, 2005-2011. Consort pairs and estimated agesa Female Age Male Age (mo.) (mo.) F2 53 F2 67 F2 89 F3 36 F3 50 M6 37 F3 62 F3 84 M51 60 F3 107 M55 69 F7 67 F7 82 F7 106 F8*e 24 F8 37 F8 60 M73 49 F8 95 F16 32 F16 52 F16 75 M6 80 F23* 21 F23 45 M27 or 78 M29f 107 F23 80 F24 75 M29 92 F24 114 F25 74 F25 94 F25 110 F25 129 F28* 36 F28 48 M29 88 F28 68 F30* 48 M55 34 F50 21 F54 24 F70* 38 M51 60 F70 52 F72* 28 F72 51 F75 32 F75 55 M73 61 F93 56 F93 90 F94* 46 F94 60 M55 70 F96 55 M55 71 F104 110 F111* 32 F116g 36-48 F118 27 F119 66 Dates pairs consortedb 06/22-24/05 03/31/08 03/28-31/10 02/28-29/08 01/13-14/09 02/19-25/08 04/12-15/07 12/27-29/06 04/16-20/07 03/10/08 02/11/09 04/15/10 05/21/10 Estimated birth datec 05/28/05 07/29/06 05/19/08 08/01/04 09/26/05 09/17/06 07/03/08 06/28/10 05/19/05 08/13/06 07/10/08 06/26/05 08/13/06 05/29/08 04/18/11 09/22/05 05/24/07 04/15/09 05/30/06 05/23/08 04/22/11 06/14/07 09/10 08/01/05 04/16/07 08/19/08 3/10 06/09/06 03/30/07 11/08 07/17/07 07/01/06 07/01/06 06/05/08 08/31/09 07/09/08 06/12/10 06/01/07 05/07/09 08/07 06/16/10 05/27/09 07/15/10 08/21/10 07/08/10 06/16/10 2009 08/08/10 08/09 a Estimated birth interval (mo.) Estimated gestation (days) 14.0 22.0 13.8 11.7 21.5 23.8 93-95 94 89-92 14.9 23.9 13.4 22.5 34.7 90-91 19.9 22.7 91-92 23.8 87-93 Non-funct.GPS 90-93 Non-funct.GPS 20.5 16.1 Non-funct.GPS 11.7 92-93 88-92 87 14.8 23.1 23.2 93 13.3 91 Observed number of cubsd 3 2 4 1 2 3 3 2 2 4 3 2 4 2 4 4 3 3 3 1 4 3 1 1 2 3 2 ≥2 tracks 1 3 1 1 3 3 1 2 1 2 2 2 3 3 4 3 2 2 3 2 Ages of females were estimated at litter birth dates. Ages of males were estimated around the dates the pairs consorted. b Consort pairs indicate pumas that were observed together based on GPS data or VHF location data. c Estimated birth dates were indicated by GPS data of mothers at nurseries or by back-aging cubs to approximate birth date. 206 �d Observed number of cubs do not represent litter sizes as some cubs were observed when they were 5 to 16 months old after postnatal mortality could have occurred in siblings. Only cub tracks were observed with F28. e Asterisk (*) indicates first probable litter of the female, based on nipple characteristics noted at first capture of the female. f A radio-collared, ear-tagged male puma was visually observed with F23 on 2/25/08. Both M27 and M29 wore nonfunctional GPS collars in that area at the time. g When captured on 1/20/10, puma F116 was in association with 2 large cubs which were not captured. 207 �Table 14. Summary for individual adult puma survival and mortality, December 8, 2004 to July 31, 2011, Uncompahgre Plateau, Colorado. Puma I.D. M1 Monitoring span 12-08-04 to 08-16-06 M4 M5 01-28-05 to 12-28-05 08-01-06 to 02-20-09 M6 02-18-05 to 05-21-10 M27 03-10-06 to 05-07-09 M29 04-14-06 to 02-25-09 M32 04-26-06 to 12-02-10 M51 01-07-07 to 03-20-09 M55 01-21-07 to 07-31-10 M67 M71 08-23-07 to 07-31-11 01-29-08 to 11-12-09 M73 M87 M90 02-21-08 to 07-31-11 02-09-11 to 07-31-11 11-16-10 to 11-23-10 M100 03-27-09 to 07-31-09 M114 M133 02-27-10 to 06-23-10 11-12-10 to 12-01-10 Status: Alive/Lost contact/Dead; Cause of death Dead. Lost contact− failed GPS/VHF collar. M1 ranged principally north of the study area as far as Unaweep Canyon. M1 was killed by a puma hunter on 01-02-10 west of Bang’s Canyon, north of Unaweep Canyon, GMU 40. M1 was about 97 months old at death. Dead; killed by a male puma. Estimated age at death 37−45 months. Dead. Born on study area; offspring of F3. M5 was independent of F3 by 13 months old, and dispersed from his natal area at about 14 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months (protected from hunting mortality in buffer area) and ranged into the eastern edge of Utah (vulnerable to hunting). Killed by a puma hunter on 02-20-09 in Beaver Creek, Utah at age 54 months. Dead. M6 was struck and killed by a vehicle on highway 550 south of Colona, CO on 05-21-10. M6 was about 99 months old at death. Dead. Lost contact− failed GPS/VHF collar. Recaptured 12-02-07 & 01-22-08 by puma hunter/outfitter north of the study area. Possibly visually observed on study area with F23 on 02-25-08. Recaptured by a puma hunter/outfitter 12-11-08 & 12-28-08 north of the study area. Photographed by a trail camera on the study area (Big Bucktail Canyon) on 5 occasions: 03-27-09, 04-02-09, 04-15-09, 04-24-09, & 05-07-09. M27 was killed by a puma hunter on 12-09-09 in the North Fork Mesa Creek, Uncompahgre Plateau, GMU 61 North. M27 was about 100 months old at death. Dead. Lost contact− failed GPS/VHF collar. Possibly visually observed on study area with F23 on 02-25-08. Recaptured on study area 02-25-09, but could not be safely handled to change faulty GPS collar. M29 was killed by a puma hunter on 11-16-09 in Beaver Canyon, GMU 70 East. M29 was about 121 months old at death. Dead. Killed by a puma hunter on 12-02-10 in McKenzie Creek on the Uncompahgre Plateau study area. M32 was about 112 months old at death. Dead. Lost contact− failed GPS/VHF collar after 03-20-09. Killed by a puma hunter on 12-11-09 in Shavano Valley, Uncompahgre Plateau study area. M51 was about 77 months old at death. Dead. Killed by a puma hunter on 11-25-10 in Spring Creek Canyon on the Uncompahgre Plateau study area. M55 was about 77 months old at death. Alive. M67 is offspring of F30. Dead. Lost contact– M71 shed his VHF collar with an expansion link on about 11-1209. He was killed by a puma hunter on 12-09-09 on the west rim of Spring Creek Canyon, Uncompahgre Plateau study area. M71 was about 47 months old at death. Alive. Alive. M87 is offspring of F3. Dead. M90 was killed by a hunter on 11-23-10 on McKenzie Butte. M90 was offspring of F72, born 07-09-08. He was 28 months old at death. Dead. M100 was killed by a puma hunter on 01-16-10 in Naturita Canyon, GMU 70 East. M100 was about 63 months old at death. Lost contact– after 06-23-10. VHF collar may have failed or puma dispersed. Dead. M133 was killed by a puma hunter on 12-01-10 in Dry Fork Escalante Canyon north of the study area. M133 was about 43 months old at death. 208 �Puma I.D. M134 Monitoring span 06-01-11 to 06-10-11 M138 F2 07-01-11 to 07-31-11 01-07-05 to 08-14-08 F3 F7 01-21-05 to 07-31-10 02-24-05 to 08-03-08 F8 F16 03-21-05 to 07-31-11 10-11-05 to 09-11-09 F23 02-05-06 to 07-31-11 F24 01-17-06 to 07-31-11 F25 02-08-06 to 02-03-11 F28 03-23-06 to 01-01-11 F30 04-15-06 to 07-29-08 F50 12-14-06 to 03-26-07 F54 01-12-07 to 08-18-07 F70 F72 01-14-08 to 07-31-11 02-12-08 to 03-18-11 F75 03-26-08 to 02-10-10 F93 F94 12-05-08 to 07-31-11 12-19-08 to 02-01-11 F95 F96 F104 F110 08-01-09 to 07-31-11 01-28-09 to 07-31-11 05-21-09 to 07-31-11 09-21-09 to 02-25-10 F111 01-01-10 to 07-31-11 Table 14 continued. Status: Alive/Lost contact/Dead; Cause of death Dead. M134 was offspring of unmarked female puma in Roubideau Canyon. Independent by about 03-28-11. Shot dead by USDA, APHIS, WS agent while in the act of attacking domestic sheep on 06-10-11 when he was 24 months old at start of adult life stage. Alive. Dead; killed by another puma (sex of puma unknown; male suspected) 08-14-08. F2 was about 92 months old at death. Lost contact− failed GPS/VHF collar. Dead. Killed by U.S. Wildlife.Services agent 08-03-08 for predator control of depredation on domestic sheep. F7 was about 107 months old at death. Alive. Dead. F16 was struck and killed by a vehicle on Ouray County Road 1 southwest of Colona, CO on 09-11-09. F16 was about 80 months old at death. Alive. Lost radio contact after12-02-09. F23 recaptured on the study area 01-26-11; her non-functional GPS collar was replaced with a VHF radiocollar. Alive. Lost radio contact after 09-03-08− failed GPS/VHF collar. F24 recaptured on 0222-11; her non-functional GPS collar was replaced with a VHF radiocollar. Dead. Lost radio contact after 09-04-09– failed GPS/VHF collar. Photographed alive with three ~9 month old cubs on 12-03-10 on Loghill Mesa. F25 shot dead by a ranch hand on 02-03-11 in Pleasant Valley, Dallas Creek because she was seen among cattle. F25 was about 138 months old at death and in excellent physical condition (49 kg). Lost radio contact after 09-25-07− failed GPS/VHF collar. Recaptured F28 on the study area 02-01-10 and 01-01-11, but could not be handled to replace non-functional GPS collar. Dead. Killed by another puma (sex of puma unknown) 07-29-08. F30 was about 60 months old at death. Dead of natural causes 03-26-07; probably injury or illness-related; exact agent unknown. F50 was about 30 months old at death. Dead; killed by a male puma while in direct competition for prey (i.e., mule deer fawn) 08-18-07. F54 was about 49 months old at death. Alive. Lost radio contact after 12-02-10. F72 recaptured in Fisher Creek on 03-18-11, but could not be handled to replace non-functional GPS collar. Lost radio contact after 09-29-09– failed GPS/VHF collar. F75 in association with her cubs M105 and F106 when F106 was recaptured on 02-10-10 on the study area. Alive. Dead. Shot dead on 02-01-11 by USDA, APHIS, WS agent for predation on domestic elk in Happy Canyon. F94 was about 74 months old at death. Alive. Alive. Alive. Dead. Killed by a puma hunter on 02-25-10 in GMU 70 East. F110 was about 41 months old at death. Alive. 209 �Puma I.D. F113 Monitoring span 01-26-10 to 06-06-10 F116 F118 F119 F135 F136 F137 F143 01-20-10 to 07-31-11 02-25-10 to 07-31-11 03-25-10 to 07-31-11 01-01-11 to 07-31-11 01-20-11 to 07-31-11 01-21-11 to 07-31-11 02-15-11 to 07-31-11 Table 14 continued. Status: Alive/Lost contact/Dead; Cause of death Dead. F113 died 06-06-10 of injuries consistent with being struck by a vehicle. GPS data indicated that F113 had crossed highway 550 and roads on Loghill Mesa north of Ridgway 24-30 hours before she died in McKenzie Creek. F113 was about 42 months old at death. Alive. Alive. Alive. Alive. Alive. Alive. Alive. 210 �Table 15. Preliminary estimated survival rates (S) of adult-age pumas during the 4 years in the reference period (i.e., the study area is closed to puma hunting) and 2 years in the treatment period, Uncompahgre Plateau, Colorado. Survival rates of pumas estimated with the Kaplan-Meier procedure to staggered entry of animals (Pollock et al. 1989). Survival rates are for an annual survival period defined as the biological year (August 1 to July 31). Survival rates were estimated only for periods when n ≥ 5 individual pumas were monitored in the interval. Puma survival in the reference period pertained only to pumas that died of natural causes. Pumas that were killed by people in the reference period, a non-natural cause (i.e., two adult pumas: F7 for depredation control 8/3/2008 and M5 killed by a puma hunter off the protected study area and buffer zone 2/20/2009) were right censored. In the treatment period all sources of natural and human-caused mortality are considered in the survival estimates. Biological Year S 1.000 Females SE 0.0000 S 0.667a Males SE 0.2222a n 6a Reference Annual 2 8/1/2005 to 7/31/2006 0.909 0.0867 11 1.000 0.0000 5 Reference Annual 3 8/1/2006 to 7/31/2007 0.831 0.0986 14 1.000 0.0000 7 Reference Annual 4 8/1/2007 to 7/31/2008 0.875 0.1031 13 1.000 0.0000 8 Reference Annual 5 8/1/2008 to 7/31/2009 0.784 0.1011 19 0.667 0.1924 8 Treatment Annual 1 8/1/2009 to 7/31/2010 Treatment Annualb 0.333b 0.1361b 12b 8/1/2009 to 7/31/2010 With mortalities of all marked adult males 0.947c 0.0568 19 0.250 0.1082 9 Treatment Annual 2 8/1/2010 to 7/31/2011 a Adult male annual S 2005 to 2006 is probably underestimated with poor precision because 3 of the 6 pumas were GPS/VHF-monitored for 4 to 5 months at the end of the interval; 1 of 6 adult males died. b This second estimate of adult male puma survival 8/1/2009 to 7/31/2010 includes 5 males that had non-functional (4) or shed (1) radiocollars. All adult males with non-functional or shed radiocollars in this study survived into treatment year 1 (TY1), which was expected considering adult male survival in 3 previous years. All 5 of those adult males were detected and killed by hunters in TY1. c Only 1 of 2 adult female puma mortalities is represented in this survival analysis for 8/1/2010 to 7/31/2011, that of F94 killed for depredation control. One other adult female mortality, F25, is not represented because she wore a nonfunctional GPS collar making it impossible for us to monitor her survival. F25 was shot by a ranch hand on 2/3/2011 when he saw her among cattle. 211 n 10 �Table 16. Summary of subadult puma survival and mortality, December 2004 to July 2011, Uncompahgre Plateau, Colorado. Puma I.D. M5 Monitoring span 09-16-05 to 06-30-06 No. days 308 M11 06-21-06 to 12-02-07 529 F23 01-04-06 to 02-04-06 04-19-06 to 04-26-06 31 M49 03-26-07 to 10-01-07 189 F52 01-10-07 to 05-15-07 125 F66 08-23-07 to 11-05-07 11-25-08 to 06-03-09 74 M31 M69 01-11-08 to 04-07-08 7 190 87 Status Survived to adult stage. M5 was offspring of F3, born August 2004. Independent and dispersed from natal area at 13 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months (protected from hunting mortality in buffer area) and ranged into the eastern edge of Utah (vulnerable to hunting). Killed by a puma hunter on 02-20-09 in Beaver Creek, Utah at about 54 months old. Survived to adult stage. M11 was offspring of F2, born May 2005. Independent at 13 months old. Dispersed from natal area at 14 months old. Moved to Dolores River valley, CO, by 12-14-06. Killed by a puma hunter on 12-02-07 when about 30 months old. Alive. Captured on the study area when about 17 months old. Survived to adult stage; gave birth to first litter at about 21 months old. Survived to adult stage. M31’s estimated age at capture was 20 months. Dispersed to northern New Mexico and was killed by a puma hunter on 12-11-08 in Middle Ponil Creek, Cimarron Range. He was about 52 months old. Survived to adult stage. M49 was offspring of F50, born July 2006. Orphaned at about 9 months old, when F50 died of natural causes. Dispersed from his natal area at about 10 months old and ranged on the northeast slope of the Uncompahgre Plateau. When M49 was about 15 months old, he shed his expandable radiocollar on about 10-01-07 at a yearling cow elk kill on the northeast slope of the Uncompahgre Plateau. He was killed by a puma hunter in Blue Creek in the protected buffer zone north of the study area on 01-24-09; he was about 29 months old, a young adult. Survived to adult stage. F52 dispersed from study area as a subadult by 01-16-07. F52’s last VHF aerial location was Crystal Creek, a tributary of the Gunnison River east of the Black Canyon 05-15-07. She was treed by puma hunters on 12-29-08 on east Huntsman Mesa, southeast of Powderhorn, CO. She was about 41-43 months old and could have been in her adult-stage home range. GPS collar nonfunctional. Dead. F66 was offspring of F30, born July 2007. Lost contact; her cub collar quit after 11-05-07. Recaptured as an independent subadult on her natal area 11-25-08 when 16 months old. F30 was killed by a puma when F66 was 12 months old, within the age range of normal independence. F66 died of injuries to internal organs that caused massive bleeding attributed to trampling by an elk or mule deer on about 05-28-09 when she was 23 months old. Her range partially overlapped her natal area. Survived to adult stage. M69 was captured on the study area when about 14-18 months old. Emigrated from the study area as subadult by 03-1908. Last VHF aerial location was southwest of Waterdog Peak, east side of Uncompahgre River Valley on 04-07-08. M69 was killed by a puma hunter on 11-06-08 in Pass Creek in the Snowy Range, WY when he was 24 to 28 months old. 212 �Puma I.D. F95 Monitoring span 12-29-08 to 07-31-09 No. days 214 M99 02-27-09 to 04-22-09 54 M112 02-10-11 to 04-18-11 67 M115 01-13-10 to 07-21-10 189 M134 03-28-11 to 06-10-11 74 M138 01-26-11 to 06-30-11 03-07-11 to 07-13-11 03-08-11 to 04-28-11 03-08-11 to 03-23-11 155 M150 03-28-11 to 04-11-11 14 M153 04-12-11 to 07-31-11 110 M144 F145 F146 128 51 15 Table 16 continued Status Alive. F95 is the offspring of F93, born about August 2007. She became an independent subadult by about 18 months old (02-11-09 aerial location) and an adult by about 24 month old (Aug. 2009). F95 established an adult home range adjacent to and overlapping the northern portion of her natal area. Dead. M99 probably killed by another puma (canine punctures in skull including braincase) in Jan. 2010 when he was about 16 months old. His radiocollar quit after 54 days. M112 was offspring of F70. Lost contact of M112 after 04-18-11; he may have dispersed or radiocollar quit. M112 associated with F96 and her two radio-collared cubs F129 and M130 during 02-10-11 to 04-1811. Dead. M115 was offspring of F28, born in Nov. 2008. He was about 14 months old when first captured on Jan. 13, 2010. When he was recaptured on 03-18-10, he had previously suffered a broken left ulna. M115 was probably independent by 07-15-10 when he was located outside of his natal area on a probably dispersal move. M115 died on about 07-21-10 apparently from complications of his broken left foreleg; probably not allowing him to kill prey sufficiently for survival. M115 was about 20 months old at death. M134 was offspring of unmarked female puma in Roubideau Canyon. Independent by about 03-28-11. Shot dead by USDA, APHIS, WS agent while in the act of attacking domestic sheep on 06-10-11 when he was 24 months old at start of adult life stage. Alive on the study area. Entered adult life stage 07-01-11. Dispersed. Last contact on 07-13-11 in Blue Creek, northwest Uncompahgre Plateau. Dispersed. Last contact on 04-28-11 in UC Creek, Deep Canyon, northwest Uncompahgre Plateau. Dead. F146 was killed and eaten by a male puma while in competition for an adult bull elk carcass that one of the pumas killed in Coal Canyon on the study area. F146 was about 19 months old at death. Dispersed. M150 was offspring of F111, born on 08-31-09. He was independent by 03-28-11 when he was 19 months old. Lost contact after 04-11-11 when M150 was in Cow Creek southeast of the study area. Alive on the study area. 213 �Table 17. Records of pumas that dispersed from the Uncompahgre Plateau study area, December 2004 to July 2011. Puma I.D. 1st capture date on study area 1st capture location→kill or resight location (UTM, NAD27) M5 02-04-05 13S,240577E, 4251037N→ 12S,665853Ex 4277125N M11 06-27-05 13S,248278E, 4239858N→ 12S,741882Ex 4161575N 84.8 M31 04-19-06 329.8 M38 09-08-06 12S,746919E, 4225441N→ 13S,500000Ex 4050000N 13S,249200E, 4239703N→ 12S,703371E, 4316856N M39 09-11-06 71.3 M43 09-15-06 12S,724270E, 4243610N→ 12S,709889E, 4313490N 12S,760177E, 4242995N→ 12S,739859E, 4308557N M48 10-18-06 52.0 M49 12-05-06 12S,756676E, 4247777N→ 12S,704982E, 4248998N 12S,757241E, 4258259N→ 12S,693350E, 4274559N M58 06-27-07 13S,258543E, 4238071N→ 13S,274670E, 4309488N 73.2 Estimated linear dispersal distance (km)* 102.2 104.1 68.6 66.1 Puma Information M5 was offspring of F3, born August 2004. Independent and dispersed from natal area at 13 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months (protected from hunting mortality in buffer area) and ranged into the eastern edge of Utah (vulnerable to hunting). Killed by a puma hunter on 02-20-09 in Beaver Creek, Utah at about 54 months old. M11 was offspring of F2, born May 2005. Shed expandable radiocollar 10-24 to 11-08-05. Recaptured and re-collared 04-0206. Independent at 13 months old. Dispersed from natal area at 14 months old. Moved to Dolores River valley, CO, by 12-14-06. Killed by a puma hunter on 12-02-07 when about 30 months old. M31’s estimated age at capture was 20 months. Dispersed to northern New Mexico and was killed by a puma hunter on 12-1108 in Middle Ponil Creek, Cimarron Range. He was about 52 months old. M38 was offspring of F2, born July 29, 2006. Shed his expandable radiocollar by 03-06-07. Photographs by trail camera in McKenzie Cr. of M38 & Unm. F sibling with F2 on 07-16 to 17-07 at 352-353 days old. M38 was killed by a hunter in Ladder Creek southwest of Grand Junction, CO on 01-07-11. He was 54 months old at death. M39 was offspring of F8, born August 2006. M39 was killed by a puma hunter in Bangs Canyon, GMU 40 on 03-12-10 when he was 43 months old. M43 was offspring of F7, born August 2006. He shed the expandable radiocollar 11-7 to 17-06, after which direct contact was lost. M43 was killed by a puma hunter 01-28-09 in Deer Creek, west slope of Grand Mesa, CO when he was 29 months old. M48 was the offspring of F3, born September 2006. M48 was killed by a puma hunter in Tabeguache Creek, GMU 61 North on 12-27-09 when he was 39 months old. M49 was offspring of F50, born July 2006. Orphaned at about 9 months old, when F50 died of natural causes. Dispersed from his natal area at about 10 months old and ranged on the northeast slope of the Uncompahgre Plateau. When M49 was about 15 months old, he shed his expandable radiocollar on about 10-01-07 at a yearling cow elk kill on the northeast slope of the Uncompahgre Plateau. He was killed by a puma hunter in Blue Creek in the protected buffer zone north of the study area on 0124-09; he was about 29 months old. M58 was offspring of F16, born May 2007. M58 was killed by a puma hunter on 12-27-09 in the North Fork of the Gunnison River north of Paonia, GMU 521; he was 31 months old. 214 �Puma I.D. 1st capture date on study area 1st capture location→kill or resight location (UTM, NAD27) M63 08-17-07 M65 08-17-07 M67 08-23-07 12S,738144E, 4233628N→ 12S,689111E, 4277908N 12S,738144E, 4233628N→ 12S,684084E, 4314200N 13S,257371E, 4235231N→ 12S,725113E, 4242447N M68 08-23-07 M69 01-11-08 M82 07-05-08 M83 07-05-08 M87 07-31-08 M88 13S,257371E, 4235231N→ 12S,711262E, 4198681N 13S,248191E, 4246810N→ 13T,378900E, 4591990N Table 17 continued. Estimated Puma Information linear dispersal distance (km)* 66.1 M63 was offspring of F24, born July 14, 2007. He was not radiocollared as a cub. M63 was killed by a hunter in Calamity Creek on northwest Uncompahgre Plateau on 01-01-11. M63 was 42 months old at death. 97.0 M65 was offspring of F24, born July 2007. M65 was killed by a USDA, APHIS, WS agent for depredation on llamas in the Little Dolores River on 11-07-09. M65 was 28 months old. 57.7 80.7 369.6 12S,726901E, 4243463N→ 13S,255316E, 4216768N 12S,726901E, 4243463N→ 12S,670949E, 4314779N 13S,239006E, 4248601N→ 12S,724325E, 4244118N 60.5 07-31-08 13S,239006E, 4248601N→ 12S,704835E, 4197839N 77.6 M92 09-29-08 13S,246359E, 4226949N→ 12S,750871E, 4222921N 21.9 M107 06-28-09 13S,242359E, 4252618N→ 12S,754886E, 4341330N 89.2 90.7 39.2 M67 was offspring of F30, born July 17, 2007 in Fisher Creek on the east slope of the study area. He was not radiocollared as a cub. M67 dispersed from the natal area and was recaptured in Tomcat Creek on the west slope of the study area on 02-24-10 when he was 31 months old. M67 is a resident adult in that area (07-3111). M68 was offspring of F30, born July 2007. He was orphaned at 12 months old when his mother was killed by a puma. He was killed by a puma hunter in the Disappointment Valley in southwest CO on 12-30-08; he was 17 months old. M69 was captured on the study area when about 14-18 months old. Emigrated from the study area as subadult by 03-19-08. Last VHF aerial location was southwest of Waterdog Peak, east side of Uncompahgre River Valley on 04-07-08. M69 was killed by a puma hunter on 11-06-08 in Pass Creek in the Snowy Range, WY when he was 24 to 28 months old. M82 was offspring of F8, born May 29, 2008; sibling of M83 below. He shed his expandable cub radiocollar after 03-20-09. M82 was killed by a hunter on 12-10-09 in the Beaver Creek fork of East Dallas Creek, GMU 65. M82 was 19 months old. M83 was offspring of F8, born May 29, 2008; sibling of M82 above. He was not radiocollared as a cub. M82 was killed by a hunter on 01-18-11 in Coates Creek west of Glade Park, CO. He was 30 months old at death. M87 was offspring of F3, born July 3, 2008 on the east slope of the study area; sibling of M88 below. He was not radiocollared as a cub. M87 dispersed from the natal area. He was recaptured on the west slope of the study area on 02-09-11 when he was 31 months old. M87 is a resident adult on the west slope of the study area to 07-31-11. M87 was offspring of F3, born July 3, 2008 on the east slope of the study area; sibling of M87 above. He was not radiocollared as a cub. M87 dispersed from the natal area. He was killed by a hunter in Dawson Creek, Disappointment Valley on 11-30-10 when he was 29 months old. M92 was offspring of F25, born August 19, 2008. He was radiocollared as a cub; last contact on 12-12-08. M92 dispersed from the natal area and was recaptured in McKenzie Creek, west slope of the study area on 04-22-11 when he was 32 months old. He could not be handled to fit a new radiocollar because of a dangerous tree. M107 was offspring of F94, born May 25, 2009; sibling of F108 below. He was not radiocollared as a cub. M107 dispersed from the nata area. He was killed by a hunter in Cottonwood Creek near Molina, CO on 12-09-10 when he was 19 months old. 215 �Puma I.D. 1st capture date on study area 1st capture location→kill or resight location (UTM, NAD27) M117 02-05-10 12S,731840E, 4232346N→ 12S,743909E, 4216633N M144 03-07-33 12S,727173E, 4242012N→ 12S,696439E, 4276888N F52 01-10-07 13S,258058E, 4236260N→ 13S,319217E, 4240467N F106 06-14-09 12S,736451E, 4240278N→ 13S,258089E, 4235866N F108 06-28-09 13S,242359E, 4252618N→ 12S,752013E, 4263883N F145 03-18-11 12S,727181E, 4241468N→ 12S,701196E, 4270127N Table 17 continued. Estimated Puma Information linear dispersal distance (km)* 19.7 M117 was offspring of F119. He wore an expandable cub collar, but shed the collar by 07-15-10 on the natal area when about 11 months old. M117 was killed by a puma hunter in Beaver Creek, San Miguel River at the southern extreme of his natal area on 0101-11. He was 17 months old at death. It is unknown if M117 was independent from his mother F119 at the time of his death. 46.6 M144 was initially captured as an independent subadult in association with subadults F145 and F146 on the study area. Mother is unknown. He moved off the study area on 03-15-11. M144’s last aerial radio location was in Blue Creek on northwest Uncompahgre Plateau on 07-13-11; he was about 22 months old. 61.1 F52 was captured on the study area when about 18-20 months old. Dispersed from study area as a subadult by Jan. 16, 2007. F52’s last VHF aerial location was Crystal Creek, a tributary of the Gunnison River east of the Black Canyon 05-15-07. She was treed by puma hunters on 12-29-08 on east Huntsman Mesa, southeast of Powderhorn, CO. She was about 41-43 months old . F52 was treed again by puma hunters on about 12-16-09 south of Powderhorn: 13S,319480E,4233219N. F52 was about 53-55 months old. This suggests that F52 has an adult home range in that area. 46.9 F106 was offspring of F75, born May 7, 2009. She wore an expandable cub collar, but shed it about 03-23-10. F106 dispersed from the natal area and moved to the east slope of the study area where she was photographed at one of our scent station cameras at the mouth of Fisher Creek from 02-27-11 to 03-03-11. She was identified by her eartag. F106 was 21 months old. F108 was offspring of F94, born May 25, 2009; sibling of M107 18.2 above. She was fitted with an expandable cub collar; but, shed the collar in the original nursery due to failure of the fastener. F108 dispersed from the natal area. She was killed by a hunter on the study area on 11-29-10 when she was 17 months old. 38.6 F145 was originally captured in association of M144 and F146; they may be siblings. Mother unknown. She moved off the study area with M144 on 03-15-11. F145’s last aerial radio location was in UC Creek, Deep Canyon, North Fork Mesa Creek on northwest Uncompahgre Plateau on 04-28-11. She was about 19 months old. *Estimated linear dispersal distance (km) from initial capture site on Uncompahgre Plateau study area to hunter kill, or last recapture, radio location, or observation site. 216 �Table 18. Recorded deaths of non-marked and marked pumas struck by vehicles and other unusual causes, in chronological order, on the Uncompahgre Plateau puma study area, Colorado, from 2004 to 2011. Puma sex & ID if marked M Estimated age (mo) Date recorded Cause of death General physical condition Location & UTM NAD27 12 09-24-04 Good F 49 07-28-05 Vehicle collision Vehicle collision Pleasant Valley, County Road 24 13S,252870E,4227520N Highway 62 east of Dallas divide 13S,250000E,4222500N F17a 11 08-18-06 F 18-24 11-06-06 F 6 01-30-07 F P1005 36 09-16-08 M 12-24 08-13-08 F61a 18 11-13-08 F 12 08-10-09 F16b 80 09-11-09 M6b 99 05-21-0 F113b 42 06-06-10 Vehicle collision Vehicle collision Vehicle collision Asphyxia, lodged in fork of tree Vehicle collision Vehicle collision Vehicle collision Vehicle collision Vehicle collision Vehicle collision Good Not pregnant or lactating Good Good Good Unknown, decomposed Good Good Good Good Good Good Not pregnant or lactating M 24 08-25-10 Vehicle Excellent P1018c collision F 6 2/16/2011 Vehicle Good P1030c collision a Subadult marked (i.e., tattoos, eartags), but not radio-collared. b Adult GPS/VHF-collared pumas. c Non-marked puma with P one-thousand number designation. 217 Highway 550 south of Colona 13S,257602E,4242185N Highway 550 east of Ridgway State Park 13S,259843E,4235985N Highway 62 west of Dallas divide 12S,762286Ex4218992N Davis Point, Roubideau Canyon 12S, 743718E,4255277N Highway 145 west of Placerville 13S,756490E,4212336N Highway 550 east of Ridgway State Park 13S,259843E,4235985N Highway 145 east of Norwood 12S,745739E,4222548N Ouray County Road 1 13S,253733E,4240060N Highway 550 south of Colona 13S,258610E,4236805N F113 crossed Highway 550 and roads on Loghill Mesa 24-30 hours before she died in McKenzie Creek 13S,257272E,4238435N Highway 62 Leopard Creek 12S,237747E,4220330N Highway 62 Leopard Creek 12S,760953E,4216683N �Table 19. Numbers of GPS locations and spans of monitoring for pumas captured on the Uncompahgre Plateau, Colorado, December 2004 to August 2011. Puma I.D. M1 M4 M6 M27 M29 M51 M55 M100 M133 F2 F3 F7 F8 F16 F23 Sex Age stage Dates monitored a M M M M M M M M M F F F F F F No. locations adult 12-08-04 to 07-20-06 1,797 adult 01-28-05 to 01-14-06 958 adult 02-18-05 to 05-14-08 1,035 adult 03-12-06 to 06-21-06 313 adult 04-14-06 to 01-01-08 1,599 adult 01-07-07 to 07-15-08 1,643 adult 01-21-07 to 11-25-10 3,523 adult 03-27-09 to 01-16-10 923 adult 11-12-10 to 12-01-10 45 adult 01-07-05 to 08-14-08 3,516 adult 01-21-05 to 04-19-11 4,862 adult 02-24-05 to 08-03-08 3,922 adult 03-21-05 to 10-10-06 1,541 adult 10-12-05 to 09-10-09 3,801 subadult, 01-04-06 to 02-04-06 113 adult 02-05-06 to 09-04-09 2,281 F24 F adult 01-17-06 to 07-25-07 1,812 F25 F adult 02-09-06 to 09-09-09 3,653 F28 F adult 03-24-06 to 08-15-07 1,499 F30 F adult 03-30-07 to 02-22-08 1,057 F50 F adult 12-14-06 to 03-26-07 352 F52 F subadult 01-10-07 to 05-08-07 383 F54 F adult 01-12-07 to 08-18-08 723 F70 F adult 01-14-08 to 06-09-11 3,359 F72 F adult 02-12-08 to 07-07-10 2,842 F75 F adult 03-26-08 to 06-03-09 1,112 F96 F adult 01-28-09 to 04-20-11 1,619 F104 F adult 05-29-09 to 11-04-10 1,632 F111 F adult 01-01-10 to 07-12-11 1174 F113 F adult 01-27-10 to 06-06-10 445 F135 F adult 01-01-11 to 08-15-11 787 F136 F adult 01-20-11 to 08-08-11 649 F137 F adult 04-12-11 to 08-15-11 235 a GPS collars on pumas were remotely downloaded at approximately 1-month intervals, except during winter 20082009 to summer 2009 due to shortage of technicians during hiring freeze to assist in airplane flights to obtain downloads and to capture pumas to replace GPS collars (lengthening the download interval saved battery power). The last date in Dates monitored includes last location from the last GPS data download acquired for an individual puma. 218 �Table 20. Summary results of exploratory use of scents and hair snags to detect individual wild pumas, November 2010 to August 2011, Uncompahgre Plateau, Colorado. Scent used No times scent used at 9 sites No. puma visits No. individual puma visits . No. times pumas rubbed Beaver castorium Catnip oil 16 8 5 5 2 0 Catnip/Spotted Fever MT Lynx 1 1 7 8 Obsession for Men 11 16 Spotted Fever 7 4 Totals 3 (Unm F, F72, F106) 2 (unm M, M153) 1 (unk sex & age) 4-5 (M153, 12 of unk sex & age, 2 unm M) 5-6 (F72, F106, F136, M153, unm M, unidentifiable) 4 (F3, F25 & 3 cubs, F96, M32) No. times hair was collected from device 5 No. individual pumas detected Max. detection probability (defined in text) 0 2 (Unm F, F106) 0 0.667 (2/3) 0.0 0 0 0 0.0 1 1 1 (unm, unk sex and age) 0.200-0.250 (1/5 to 1/4) 3 3 2(F106, F136) 0.333-0.400 (2/6 to 2/5) 1 1 1 (F25 & cubs) 0.250 (1/4) 39 10 10 Table 21. Variation in individual puma response to scents, November 2010 to August 2011, Uncompahgre Plateau, Colorado. Individual F3 F25 (& 3 cubs) F72 F72 F96 F106 F106 F136 Unmarked Female, unk age M32 M153 M153 M153 Unmarked Male, unk age Unmarked Male, unk age Unmarked Male, unk age Unmarked Male, unk age Unmarked, unk sex and age Unmarked, unk sex and age Unmarked, unk sex and age Unknown if marked, unk sex and age Scent Spotted Fever Spotted Fever Beaver Castorium Obsession for Men Spotted Fever Beaver Castorium Obsession for Men Obsession Beaver Castorium Spotted Fever Obsession for Men Catnip MT Lynx Obsession MT Lynx MT Lynx Catnip Spotted Fever & Catnip MT Lynx MT Lynx Obsession 219 No. times rubbed/ No. of visits 0/1 1/1 0/2 0/3 0/1 4/5 1/1 2/7 1/1 0/1 0/2 0/1 0/2 0/2 0/2 0/1 0/1 0/1 0/2 1/1 0/1 �GOAL: Strategies, Information, & Tools for Managing Healthy, Self-sustaining Puma Populations in Colorado Puma Population Puma Habitat Effects of Harvest & Other Mortality Movements & Corridors Population Dynamics: Sex & Age, Vital Rates, Mortality, Population Growth Rates Vulnerability to Harvest Human Development Habitat Use Effects of Translocation Methods for Monitoring Populations Domestic Animals Puma― Human Relationships Deer, Elk, Other Natural Prey & Species of Concern Effects of Predation Effects of Evasive Conditioning Map Prey Distribution Habitat Models Population Models Puma Prey Puma―Prey Relationships Models Habitat Maps Figure. 1. An ecologically-based conceptual model of the Colorado Puma Research Program that provides the contextual framework for this and proposed puma research in Colorado. Gray-shaded shapes identify areas of research addressed by this puma research on the Uncompahgre Plateau for the puma management goal in Colorado (at top). 220 �Figure 2. The puma study area on the southern half of the Uncompahgre Plateau, Colorado (shaded in gray) comprising the southern portions of Game Management Units (GMUs) 61 and 62 and a northern portion of GMU 70. Puma Po11ulation Trend, U.P., CO ~ 60 "E so 40 ,,.," 30 ".,.,,,_ 20 4~ ~ :11 52 ,,4 43 41 1'(1 I Y2 ll. ~ ,, lO .!: 0 z 0 KY4 Years Figure 3. Trends in the population of independent pumas on the Uncompahgre Plateau Puma Study Area, including Reference Years 4 and 5 (RY4, RY5) and Treatment Years 1 and 2 (TY1, TY2). Numbers represent minimum counts that include all pumas from known radio-collared pumas, visual observations of non-marked pumas, harvested non-marked pumas, and track counts of suspected non-marked pumas on the study area during fall to spring hunting and research capture seasons, except RY5 (45), which had to be modeled from RY4 observation data (33) because the hiring freeze that year affected search and capture efforts. The actual minimum count for RY5 was 37 independent pumas. The quota of 8 pumas for TY1 represented a 15% harvest of the model projected 53 independent pumas expected in TY1 and was used to set the quota ahead of the hunting season. Starting in TY1, two capture teams were deployed to 221 �count pumas on the study area because the hunting season shortened our fall-winter-spring research period. We deployed a team on each the east and west sides of the study area. The minimum count for TY1 was actually 55 independent pumas, consistent with the model expected 53. We made further team changes for TY2, which made our efforts more efficient and successful. Yet, in TY2 we counted slightly less (52) independent pumas than in TY1 (55). Post-harvest high trend line represents the population of independent pumas after pumas harvested only on the study area by hunters. This trend line represents 14.5% to 15.4% harvest of independent pumas. Post-harvest low trend line represents the population of independent pumas after pumas harvested on the study area and pumas harvested when they ranged onto adjacent GMUs open to hunting. The TY2 postharvest low also includes 2 adult female pumas killed February 1, 5, 2011 on the study area to protect livestock (F25 killed while seen by a ranch hand among cattle; F94 killed for preying on domestic elk). This trend line represents 21.2% to 21.8% harvest of independent pumas. Age structure of independent pumas in November 2010 at beginning of the puma hunting season in Treatment Year 2, Uncompahgre Plateau, Colorado. 7 6 f-f-- 1 0 f-- ~ f-- ~ f-- ~ ~ ~ I r1 11 11 11 ■ Female I 11 ■ Male lto2>2to >3to >4 to >S to >6to >7 to >8to >9to 10+ 3 4 5 6 7 8 9 10 Age(years) Figure 4. Estimated age structure of independent pumas in November 2010 at the beginning of the puma hunting season in Treatment Year 2 (TY2) on the Uncompahgre Plateau, Colorado. All these pumas were captured and sampled by researchers or harvested by hunters and examined by researchers. Mean ± SD of female and male ages, respectively: 4.87 ± 3.11 yr. (58.40 ± 37.26 mo.), n = 25; 3.51 ± 2.59 yr. (42.07 ± 31.08 mo.), n = 14. 222 �Puma births, Uncompahgre Plateau, Colorado. 10 9 "' ....a1 = z 0 8 7 6 5 4 3 2 1 0 I Ja n. 11 I I I I I I I l I H: I I I 11 I Feb. M ar. Apr. M ay June July Aug. Sep. Oct. Nov. Dec. ■ Bi1ths 2005-2011 ■ Bi1t hs 1982-1 987 Figure 5. Puma births (black bars) detected by month from May 19, 2005 to April 22, 2011 (n = 40 litters of 21 females; 38 of the litters were examined at nurseries when cubs were 26-42 days old and 2 litters confirmed by tracks of ≥1 cubs following GPS-collared mothers F28 and F111 when cubs were ≤42 days old). Also shown (gray bars) are results of the earlier effort by Anderson et al. (1992:48; 1982 to 1987, n = 10 litters of 8 females, examined when cubs were <1 to 8 months old), Uncompahgre Plateau, Colorado. 223 �Appendix A. Summary of individual puma cub survival and mortality, 2005 to 2010, Uncompahgre Plateau, Colorado. Puma I.D. M5 Estimated Age at capture (days) 183 Est. Birth date ~8-1-04 Est. survival span from 1st capture to fate or last monitor date 02-04-05 to 04-07-08 Age to last monitor date alive or at death (days, birth to fate) 31 5-28-05 F10 31 5-28-05 M11 31 5-28-05 F12 42 F13 Mother I.D. Survived to subadult stage by 09-16-05; independent at ~13 mo. old. Dispersed from natal area by 09-29-05 at 14 mo. old. Established territory on NW U.P. Killed by hunter in Beaver Creek, UT 02-20-09 at 4.5 years old. Lost contact― shed radiocollar 04-19-06 to 04-26-06. F3 Lost contact― shed radiocollar 08-10-05; last tracks of F10 with mother F2 & siblings F9 & M11 observed 11-20-05. F10 disappeared by 12-30-05. Survived to subadult stage by 06-21-06, independent at 13 mo. old. Dispersed from natal area by 07-11-06 at 14 mo. old. Killed by a hunter in SW CO 12-2-07 at 918 days (30 mo.) old. Lost contact― shed radiocollar 07-28-05―08-01-05. Tracks of F12 found in association with mother F7 on 1208-05. F12 disappeared by 01-27-06 when she was not visually observed with F7, and her tracks were not seen in association with F7’s tracks. Dead; killed and eaten by a puma (sex unspecified) about 828-05. Lost contact― shed radiocollar 01-20-06 to 01-25-06. Tracks of F14 were observed with tracks of mother F8 & sibling M15 on 02-07-06. Disappeared by 03-11-06, only tracks of F8 & M15 were found. Lost contact― shed radiocollar 06-06-06 to 06-14-06. F2 F16 308-314 Dead. Lost contact― shed radiocollar 06-06-06 to 06-14-06. Killed by a car on highway 550 on 08-18-06. Probably dependent on F16. Dead; probably killed by another puma. Multiple bite wounds to skull. 10 mo. old. Lost contact― shed radiocollar 07-27-06 to 08-02-06. 244-245 Lost contact― shed radiocollar 05-24-06―05-25-06. F16 Lost contact; radiocollar quit. Last aerial location 8-16-06, live signal. F3 ~1,345 F9 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 06-27-05 to 4-19-06 06-27-05 to 11-20-05― 12-29-05 06-27-05 to 12-2-07 326-333 5-19-05 07-01-05 to 12-08-05― 01-26-06 203-252 42 5-19-05 101 F14 26 6-26-05 07-01-05 to 08-28-05 07-22-05 to 02-07-06― 03-10-06 M15 26 6-26-05 F17 34 9-22-05 F18 34 9-22-05 M19 34 9-22-05 M20 34 9-22-05 F21 37 9-26-05 176-215 918 226-257 07-22-05 to 06-06 to 14-06 10-26-05 to 08-18-06 345-353 10-26-05 to 07-20 to 27-06 10-26-05 to 07-27 to 08-02-06 10-26-05 to 05-24-06 11-02-05 to 08-16-06 301-308 330 324 224 F2 F2 F7 F7 F8 F8 F16 F16 �Appendix A continued Puma I.D. Estimated Age at capture (days) M22 37 9-26-05 M26 183 8-1-05 F33 31 5-30-06 F34 31 5-30-06 F35 31 5-30-06 F36 29 6-9-06 M37 29 6-9-06 M38 41 7-29-06 M39 29 8-13-06 Est. Birth date F40 29 8-13-06 F41 29 8-13-06 M42 29 8-13-06 M43 33 8-13-06 Est. survival span from 1st capture to fate or last monitor date 11-02-05 to 12-21-05― 12-22-05 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 86-87 Dead; killed and eaten by male puma 12-21-05―12-22-05. F3 02-08-06 to 03-21 to 24-06 06-30-06 to 07-31-06 06-30-06 to 07-31-06 ~232-235 Lost contact― shed radiocollar 03-21-06―03-24-06. F25 F23 06-30-06 to 07-07-06 07-08-06 to 07-28-06 07-08-06 to 07-28-06 09-08-06 to 07-16 to 17-07 38 Dead. Probably killed and eaten by a male puma 08-01 to 03-06. GPS data on M29 indicate he was not involved. Dead. Probably killed and eaten by a male puma 08-01 to 03-06. GPS data on M29 indicate he was not involved. Dead; research-related fatality.a Dead. Killed and eaten by a male puma 08-22-06. GPS data on M29 indicate he was not involved. Dead. Killed and eaten by a male puma 08-22-06. GPS data on M29 indicate he was not involved. Lost contact― shed radiocollar found 03-06-07. Photo (trail camera in McKenzie Cr.) of M38 & Unm. F sibling with F2 on 07-16 to 17-07 at 352-353 days old. F28 Lost contact― shed radiocollar by 09-20-06, but seen alive on that date. Tracks of 2 cubs following F8 on 04-25-07. Survived to adult stage; dispersed from natal area. Killed by a puma hunter 03-12-10 in GMU 40 when 43 months old. Lost contact― shed radiocollar by 09-20-06, but seen alive on that date. Tracks of 2 cubs following F8 on 04-25-07. F8 Assumed dead. Lost Contact― shed radiocollar or died (blood on collar) between 10-05-06 (last live signal) & 1013-06 (collar found). Dead; research-related fatality.b F8 Lost contact− shed radiocollar by 11-7 to 17-06. Treed 0301-07. Killed by a puma hunter 01-28-09 in Deer Creek, west slope of Grand Mesa, CO at 29 months old. Survived to adult stage; dispersed from natal area. Killed by a puma hunter 01-28-09 in GMU 41 when 29 months old. F7 09-11-06 to 09-20-06 to 04-25-07 09-11-06 to 09-20-06 to 04-25-07 09-11-06 to 10-05-06 09-11-06 to 11-27-06 09-15-06 03-01-07 63-65 63-65 74 74 352-353 9 255 9 255 53-61 106 200 225 Mother I.D. F23 F23 F28 F2 F8 F8 �Appendix A continued Puma I.D. Estimated Age at capture (days) M44 33 Est. Birth date 8-13-06 Est. survival span from 1st capture to fate or last monitor date 09-15-06 to 02-14-07 Age to last monitor date alive or at death (days, birth to fate) 479 F45 33 8-13-06 09-15-06 to 5-20 to 23-07 280-283 M46 31 9-17-06 10-18-06 to 12-15-06 89 360 M47 M48 M49 31 31 153 9-17-06 9-17-06 7-1-06 10-18-06 to 12-15-06 to 09-12-07 10-18-06 to 12-15-06 to 09-12-07 183 7-1-06 360 89 360 12-05-06 to 07-31-07 to 01-01-07 F53 89 01-12-07 to 02-23-07 ~456 42 ~428 subad. 226 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Lost contact− shed radiocollar by 10-27-06. Treed, visually observed 02-14-07; sibling (?) M56 also captured, sampled, & marked for 1st time. Killed by Wildlife Services for depredation control on 12-05-07, for killing 4 domestic sheep. He was still dependent on F7. Dead. Multiple puncture wounds on braincase― parietal & occipital regions; consistent with bites from coyote. F45 switched families, moving from F7 to F2 about 12-19 to 2006. Last date F45 was with F2 was 04-17-07. Lost contact― shed radiocollar. Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-1207 in Puma Canyon. Lost contact― shed radiocollar. Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-1207 in Puma Canyon. Lost contact― shed radiocollar. Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-1207 in Puma Canyon. Survived to adult stage; dispersed from natal area. Killed by a puma hunter 12-27-09 in GMU 61 when 39 months old. M49 was orphaned when his mother died on about 03-2607; he was ~268 days old. M49 dispersed from natal area and onto NE slope of U.P. Shed radiocollar at a yearling cow elk kill about 10-01-07; he was ~428 days old. Killed by a puma hunter in Blue Creek, northwest Uncompahgre Plateau (GMU 61 N) 01-24-09 when ~29 months old. Lost contact― shed radiocollar 2-23-07. F53 visually observed by P. & F. Star, on 9-2-07, when F53 was ~14 months old and an independent subadult. Mother I.D. F7 F7 F3 F3 F3 F50 F54 �Appendix A continued Puma I.D. Estimated Age at capture (days) M56c 183 ~8-13-06 F57 35 4-16-07 M58 34 5-24-07 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 02-14-07 to 03-01-07 05-21-07 to 06-06-07 06-27-07 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. 200 Lost contact― shed radiocollar 2-27-07. M56 observed 0301-07. Lost contact― shed radiocollar 06-07-07. Live mode 06-0607. Not radio-collared. Tracks of 3 cubs observed with F16’s tracks on 04-12-08, McKenzie Butte-Pinon Ridge Pass. 3 cubs observed with F16 on 08-08-08 by B. & T. Traegde. Survived to adult stage; dispersed from natal area. Killed by a puma hunter 12-27-09 in GMU 521 when 31 months old. Alive. Observed alive 11-20-07 with F16, but without siblings M58 & F61. Tracks of 3 cubs observed with F16’s tracks on 04-12-08, McKenzie Butte-Pinon Ridge Pass. 3 cubs observed with F16 on 08-08-08 by B. & T. Traegde. F7 (?) Dead; research-related mortality.d F16 Radiocollar malfunction. Tracks of 3 cubs observed with F16’s tracks on 04-12-08, McKenzie Butte-Pinon Ridge Pass. 3 cubs observed with F16 on 08-08-08 by B. & T. Traegde. Dead. Died probably as independent subadult at 538 days old; struck by car on Hwy 550 mi. marker 111 N. of Ridgway, CO, euthanized by gunshot on 11/13/08. Not radio-collared. Not radio-collared. Not radio-collared. Two out of potential of 4 of F24’s male cubs were visually observed with her on 4/1/08. Assume that 2 male cubs died before the age of 8.5 mo. Eartags were seen on both cubs, but the numbers were not. Not radio-collared. Two out of potential of 4 of F24’s male cubs were visually observed with her on 4/1/08. Assume that 2 male cubs died before the age of 8.5 mo. Eartags were seen on both cubs, but the numbers were not. Survived to adult stage; dispersed from natal area. Killed by Wildlife Services for depredation control on 11-07-09 when 28 months old. F16 52 324 434 F59 34 5-24-07 06-27-07 to 08-21-07 55 324 M60 34 5-24-07 F61 34 5-24-07 06-27-07 to 07-11 to 12-07 06-27-07 to 06-29-07 434 48-49 324 434 538 M62 M63 M64 34 34 34 7-14-07 7-14-07 7-14-07 08-17-07 08-17-07 08-17-07 262 M65 34 7-14-07 08-17-07 262 227 F25 F16 F16 F24 F24 F24 F24 �Appendix A continued Puma I.D. Estimated Age at capture (days) F66 37 Est. Birth date 7-17-07 Est. survival span from 1st capture to fate or last monitor date 08-23-07 to 11-05-07 Age to last monitor date alive or at death (days, birth to fate) Radio-collared. Lost contact; last location 11/5/07. No signals after that date. F66 was photographed with one male sibling, either M67 or M68, & F30 on 5/31-6/1/08. F66 was recaptured and radio-collared as a subadult on 11/25/08. She died from massive trauma & bleeding of internal organs possibly resulting from being trampled by an elk or mule deer on about 05-28-09 as an independent subadult 23 months old. Not radio-collared. M67 or M68 was photographed with sibling F66 & mother F30 on 5/31-6/1/08. Dispersed from natal area. Established adult home range on west side of Uncompahgre Plateau. Alive as of 07-31-11. Not radio-collared. M67 or M68 was photographed with sibling F66 & mother F30 on 05-31 to 06-01-08. Survived to subadult stage; dispersed from natal area. Killed by a puma hunter in Disappointment Valley, CO (GMU 71) 12-30-08 at 17 months old. Radio-collared. Shed radiocollar between 7-9-08 and 7-1508, probably while still dependent on mother F75. Not radio-collared. Probably dead; if not killed when sibling M79 was killed, then probably would starve to death. Not radio-collared. Probably dead; if not killed when sibling M79 was killed, then probably would starve to death. Not radio-collared. Probably dead; if not killed when sibling M79 was killed, then probably would starve to death. Not radio-collared. Dead. Chewed-off anterior portions of the nasals, maxilla, palate, dentaries, and pieces of the braincase, with 6 or 9 portion of yellow ear-tag and intestines and bits of skin found ~45 m from mother F2’s death site on 8/14/08. Cub death probably due to puma-caused infanticide with cannibalism at ~87 days old. Male puma scrapes, about 8, under a rock rim ~50m distance from cub remains, and made ~ time of pumas’ deaths. 111 681 M67 37 7-17-07 08-23-07 M68 37 7-17-07 08-23-07 1475 532 F74 259 6-1-07 5-19-08 03-12-08 to 07-09-08 06-18-08 M76 30 ~87 M77 30 5-19-08 06-18-08 ~87 F78 30 5-19-08 06-18-08 ~87 M79 30 5-19-08 06-18-08 87 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 403 228 Mother I.D. F30 F30 F30 F75 F2 F2 F2 F2 �Appendix A continued Puma I.D. Estimated Age at capture (days) F80 40 F81 F97 Est. Birth date Age to last monitor date alive or at death (days, birth to fate) 5-23-08 Est. survival span from 1st capture to fate or last monitor date 07-02-08 40 8 ½ mo. 5-23-08 5-23-08 07-02-08 to 07-29-09 02-04-09 424 354 M82 37 5-29-08 07-05-08 to 03-20-09 or 04-02-09 295-308 M83 37 5-29-08 07-05-08 M84 36 6-5-08 07-11-08 to 02-11-09 F85 36 6-5-08 07-11-08 F86 36 6-5-08 07-11-08 to 07-23 to 08-03-08 ~48-59 M87 28 7-3-08 07-31-08 1123 M88 F89 M90 28 28 36 7-3-08 7-3-08 7-9-08 07-31-08 07-31-08 08-14-08 251 867 229 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. Not radio-collared. Apparently died before 2-4-09; no tracks found in association with F23 & siblings F81 & F97. Radio-collared. Last live location 7-29-09. Radio-collared. Lost contact after 05-12-09; shed collar at elk kill cache on Mailbox Park. Radio-collared. Survived to subadult stage; dispersed from natal area. Killed by a puma hunter in 12-10-09 GMU 65 when 19 months old. Not radio-collared. Apparently died; no tracks found in association with F8 & sibling M82 2-10-09. Radio-collared 7-11-08 to 7-22-08; collar removed because of malfunction. Not radio-collared after 7-22-08. Eartag of M84 was found by E. Phillips on 8-25-08 when mother F70’s GPS locations located here on either side of the eartag in the East fork Dolores Cyn. M84 recaptured radiocollared again 1-29-09 in Dolores Cyn. in association with F70 & F96’s family. Shed radiocollar again about 211-09. F23 Radio-collared. Dead. Probably died of predation or infanticide about 10-108 near elk calf kill. Radio-collared 7-22-08. Dead. Radio-collar, orange ear-tag #86 with pinna with green tattoo #86 found by J. Timmer 9-1-08. F86 died ~7-23 to 8-3-08 when mother F70’s GPS locations located her at F86 remains. Probable predation. Not radio-collared. Dispersed from natal area. Recaptured as adult on west slope of study area on 02-09-11. Alive as of 07-31-11. Not radio-collared. Radio-collared. Radio-collared. Recaptured as young adult on study area, adjacent to natal area, on 11-16-10. Killed by a puma hunter during TY2 on 11-23-10. F70 F23 F23 F8 F8 F70 F70 F3 F3 F3 F72 �Appendix A continued Puma I.D. Estimated Age at capture (days) Male 7A 28-35 7-10-08 Male 7B 28-35 Female 7C Est. Birth date Est. survival span from 1st capture to fate or last monitor date ~08-07-08 to 08-14-08 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 28 to 35 F7 7-10-08 ~08-07-08 to 08-14-08 28 to 35 28-35 7-10-08 28 to 35 M91 M92 35 35 8-19-08 8-19-08 ~08-07-08 to 08-14-08 09-29-08 09-29-08 F95 16 mo. June-07 12-29-08 F98 4-5 mo. 5 mo. 02-12-09 to 03-08-09 2-27-09 to 01-2010 146-176 M99 Sep-Oct08 Sep-Oct08 M101 35 4-15-09 157 M102 35 4-15-09 05-20-09 to 09-19-09 05-20-09 F103 35 4-15-09 159 M105 38 5-7-09 F106 38 5-7-09 05-20-09 to 09-17-09 06-14-09 to 02-09-10 06-14-09 to 02-27-11 Not radio-collared. F7’s cubs died from starvation after they were orphaned. F7 was shot on 8-3-08 for killing domestic sheep. Not radio-collared. F7’s cubs died from starvation after they were orphaned. F7 shot on 8-3-08 for killing domestic sheep. Not radio-collared. F7’s cubs died of starvation after orphaned. F7 shot on 8-3-08 for killing domestic sheep. Radio-collared. Radio-collared. Lost contact after 12-12-08. Dispersed from natal area. Recaptured in McKenzie Creek, west slope of study area on 04-22-11 when 32 months old. Radio-collared. Survived to adult stage. Established adult home range overlapping mother F93’s home range. Radio-collared. Died, probably killed by male puma (infanticide). Radio-collared. Last location 4-22-09 on Paterson Mt. Died as 16-month old subadult in San Miguel Canyon. Probably killed by another puma. Radio-collared. Died; killed by puma M55 after cub was orphaned due to death of mother F16 by vehicle strike. Radio-collared. Lost contact after 9-4-09. Did not find evidence of M102 associated with deaths of siblings M101 and F103. But M102 probably died. Radio-collared. Died; killed by puma M55 after cub was orphaned due to death of mother F16 by vehicle strike. Radio-collared. Lost contact after 2-9-10 due to shed collar. F75 M107 34 5-25-09 Not radio-collared at nursery; F75 returned to nursery during handling. Radio-collared later on 2-10-10. Lost contact due to shed collar 3-16 to 29-10. F106 dispersed from natal area and was photographed at 21 months old at camera and scent-rub station on east slope of Uncompahgre Plateau on 02-27-11. Not radio-collared; too small. Recaptured 2-24-10; not collared. 06-28-09 to 02-24-10 976 488 278 275 661 241 230 Mother I.D. F7 F7 F25 F25 F93 Unm.F Unm.F F16 F16 F16 F75 F94 �Appendix A continued Puma I.D. Estimated Age at capture (days) F108 34 Est. Birth date 5-25-09 Est. survival span from 1st capture to fate or last monitor date 06-28-09 to 03-05-10 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. M115 14 mo. Nov.-08 07-21-10 610 M117 6 mo. Aug.-09 02-05-10 275 P1016(M) 39 6-12-10 06-12-10 to 07-21-10 39 P1017(M) 39 6-12-10 06-12-10 to 07-21-10 39 M120 30 6-28-10 157 M121 30 6-28-10 273 Radio-collared. Lost radio contact after 03-28-11. F3 M122 35 7-8-10 07-28-10 to 12-02-10 07-28-10 to 03-28-11 08-12-10 to 04-28-11 Shed radiocollar at nursery; fastener failed. Recaptured and re-collared 2-24-10. Shed collar ~3-5-10. Dispersed from natal area. Killed by a puma hunter on the study area during TY2 on 11-29-11. Not radio-collared; too small. Radio-collared. Lost contact after 5-4-10 (last live signal) possibly due to failed transmitter. Recaptured and re-radiocollared on 01-24-11. Independent subadult during 02-10-11 to 04-18-11. Lost contact after 04-18-11; he may have dispersed or radiocollar quit. Radio-collared. M115 died as a subadult (~20 mo. old) due to complications of a broken left foreleg (natural cause). Radio-collared. Lost contact after 5-14-10 (last live signal); shed collar found on 7-15-10 in the natal area. Not radio-collared. Monitored at nursery via mother’s GPS/VHF collar. Found dead at nursery due to infanticide by puma M32 on same day as our investigation of nursery. Not radio-collared. Monitored at nursery via mother’s GPS/VHF collar. Found dead at nursery due to infanticide by puma M32 on same day as our investigation of nursery. Radio-collared. Lost radio contact after 12-02-10. 274 F104 F123 29 7-15-10 F124 29 7-15-10 M125 29 7-15-10 M126 28 08-08-10 M127 28 08-08-10 Radio-collared. Lost radio contact after 04-28-11. Tracks of 2 other siblings of M122 observed on 01-11-11 (neither cub marked). Radio-collared. Killed on 02-17-11 for depredation control on domestic elk by Wildlife Services agent. Radio-collared. Killed on 02-16-11 for depredation control on domestic elk by elk farm manager. Radio-collared. Killed on 02-01-11 for depredation control on domestic elk by Wildlife Services agent. Radio-collared. Lost radio contact after 03-17-11; shed his radiocollar at a mule deer cache. Radio-collared. Lost radio contact after 07-01-11; shed his radiocollar about 07-01-11. 553 M109 M112 34 145 5-25-09 8-31-09 06-28-09 05-04-10 528 595 08-13-10 to 02-17-11 08-13-10 to 02-16-11 08-13-10 to 02-01-11 09-05-10 to 03-17-11 09-05-10 to 07-01-11 217 216 201 221 327 231 F94 F94 F70 F28 F119 F72 F72 F3 F94 F94 F94 F118 F118 �Appendix A continued Puma I.D. Estimated Age at capture (days) M128 28 08-08-10 F129 35 08-21-10 M130 35 08-21-10 09-25-10 to 10-23-10 M131 35 08-21-10 F132 35 08-21-10 09-25-10 to 07-21-11 09-25-10 M134 ~18 mo. ~June-09 12-14-10 to 06-10-11 731 M139 36 04-18-11 05-24-11 to 07-29-11 102 F148 36 04-18-11 05-24-11 to 07-29-11 102 F140 ~5 mo. ~Aug.10 01-02-11 to 04-18-11 258 M141 ~5 mo. ~Aug.10 01-02-11 to 04-01-11 241 M142 ~5 mo. ~ 6 mo. 01-02-11 to 04-18-11 02-16-11 258 P1030 F147 ~7 mo. ~Aug.10 ~Aug.10 ~Sep.-10 F149 45 04-22-11 M150 525 08-31-09 04-21-11 to 07-31-11 06-06-11 to 07-31-11 02-07-11 to 04-11-11 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 09-05-10 to 02-22-11 09-25-10 to 04-28-11 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. 198 F118 315 Radio-collared. Lost radio contact after 02-22-11; radiocollar probably quit. Radio-collared. Fate unknown. Transmitter on mortality mode on 04-28-11. Unable to get to collar until 06-23-11 due to high spring run-off, by then the transmitter had quit. Radio-collared. Died of natural causes associated with injury to right shoulder during first move away from nursery about 10-23-10. Radio-collared. Lost contact after 07-21-11. Shed his radiocollar about 07-27-11. Not radio-collared. Too small for collar design. Fate unknown. Radiocollared as dependent large cub. Independent by about 03-28-11. Dead; killed for depredation control by Wildlife Services agent on 06-10-11. Radio-collared. Dead of infanticide and cannibalism along with sibling F148; killed and eaten by female or subadult male puma about 07-29-11. Radio-collared. Dead of infanticide and cannibalism along with sibling M139; killed and eaten by female or subadult male puma about 07-29-11. Radio-collared. Lost contact. Shed first collar about 01-2411. Recaptured and re-collared on 04-01-11. Shed second collar after 04-18-11. Radio-collared. Lost contact; shed radiocollar about 03-2911. Recaptured, but could not be handled safely on 04-0111. Radio-collared. Lost contact after 04-18-11 due to shed collar. Struck by vehicle and killed on state highway 62 in Leopard Creek, south boundary of study area on 02-16-11. Radio-collared. 100 Radio-collared. F23 588 Radio-collared. M151 was independent by 03-28-11 at 19 mo. old. He dispersed from the natal area by 04-11-11 at 19.5 mo. old. Contact lost after 04-11-11. F70 250 63 334 35 183 232 F96 F96 F96 F96 Unm. F F8 F8 Unk./ F28? Unk./ F28? Unk./ F28? Unk. F24 �Appendix A continued Puma I.D. Estimated Age at capture (days) M151 253 06-16-10 F152 06-06-10 271 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 02-24-11 to 03-07-11 03-14-11 to 03-21-11 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. 264 Radio-collared. Lost contact after 03-07-11 (GPS location of mother F111 at shed collar of M151). Radio-collared. Lost contact after 03-21-11; shed collar. F111 271 a F93 Cub F35 probably starved between 06-30-06 & 07-07-06 after the transmitter on the expandable collar got in its mouth. Cub M42 died after being captured by dogs, probably from stress of capture associated with severe infection of laceration under right foreleg caused by expandable radiocollar. c Cub M56 was captured in association with F7 and her cubs M43 and M44. He may have been missed at the nursery when M43 and M44 were initially sampled and marked. d Cub M60 died probably of starvation. The expandable radiocollar was around the neck and right shoulder, probably restricted movement. b 233 �Appendix B. Summary of exploratory use of scents and hair snags to detect individual pumas, 2010 to 2011, Uncompahgre Plateau, Colorado. Details on behaviors of pumas and other wildlife that visited the camera-scent stations are not included in this appendix, but are in original data file. Time puma was at site (min.) No. photos of puma Time lapse between scent treatment and puma visit (days) Scent type/name Closest puma distance estimate to scent pad-hair snag (m) Rub response by puma (yes, no) Date MS Time Puma Sex Age stage Female Reproductive Status HS01 11/27/2010 9:39 F96 F Adult Cubs 6 mo. old 1 2 7 Spotted Fever 0.3 no HS01 12/8/2010 16:40 unmarked F Adult Unk 5 114 2 Beaver Castor 0 yes HS02 11/29/2010 15:40 F3 F Adult Cubs 5 mo. old 1 6 9 Spotted Fever 0.6 no HS03 2/27/2011 6:10 F106 F Adult No cubs 3 95 12 Beaver Castor 0 yes yes HS03 2/27/2011 17:45 F106 F Adult No cubs 5 48 12 Beaver Castor 0 yes yes HS03 2/28/2011 18:38 F106 F Adult No cubs 74 572 13 Beaver Castor 0 yes yes HS03 3/1/2011 6:31 F106 F Adult No cubs 1 15 14 Beaver Castor 0 no HS03 3/2/2011 18:37 F106 F Adult No cubs 13 297 15 Beaver Castor 0 yes yes HS03 3/3/2011 18:23 F106 F Adult No cubs 4 107 16 Obsession 0 yes yes HS03 3/7/2011 20:26 F136 F Adult No cubs 1 18 4 Obsession 3 no HS03 3/11/2011 4:41 F136 F No cubs 5 54 8 Obsession 0 yes HS03 4/11/2011 0:31 unmarked M Adult 1 3 3 Catnip 1.5 no HS03 4/13/2011 22:18 M153 M Sub-adult 1 3 5 Catnip 0.6 no HS03 5/16/2011 21:54 unmarked Unk Unk 1 4 4 Catnip/Spotted Fever 0.6 no HS03 6/11/2011 22:56 unmarked M Adult 1 9 2 Obsession 0.3 no HS03 7/6/2011 17:59 unmarked Unk Unk 1 11 0 MT Lynx 0.1 no HSO4 11/22/2010 5:34 M32 M Adult 1 10 4 Spotted Fever 0.6 no HS04 12/3/2010 17:40 F25 & 3 cubs F Adult 7 243 3 Spotted Fever 0 yes Camera site I.D. Cubs 8-9 mo. old 234 Hair on snag collected yes yes yes �Appendix B continued. Time puma was at site (min.) Time lapse between scent treatment and puma visit (days) Closest puma distance estimate to scent pad-hair snag (m) Date MS Time Puma Sex Age stage Female Reproductive Status HS04 2/24/2011 16:04 F72 F Adult No cubs 3 44 9 Beaver Castor 0 no HS04 2/25/2011 15:36 F72 F Adult No cubs 1 15 10 Beaver Castor 0 no HS04 3/8/2011 6:33 F72 F Adult No cubs 1 21 5 Obsession 0 no HS04 3/8/2011 20:51 F72 F Adult No cubs 1 21 5 Obsession 3.5 no HS04 3/16/2011 3:29 F72 F Adult No cubs 1 13 Obsession 3.5 no HS04 3/18/2011 21:32 Not identifiable Unk Unk Unk 1 9 15 Obsession 3 no HS04 4/14/2011 2:03 F136 F Adult No cubs 1 9 6 Obsession 3.5 no HS04 4/15/2011 4:40 F136 F Adult No cubs 1 15 7 Obsession 0.3 no HS04 4/16/2011 4:40 F136 F Adult No cubs 1 6 8 Obsession 3 no HS04 4/18/2011 18:20 F136 F Adult Pregnant 1 15 10 Obsession 0 yes HS04 4/25/2011 23:52 M153 M Sub-adult 1 15 17 Obsession 0.3 no HS04 5/2/2011 19:31 unmarked M Adult 1 27 24 Obsession 0 no M Sub-adult 1 9 31 Obsession 3 no Adult 1 12 32 Obsession 2.3 no MT Lynx 1.5 no Camera site I.D. HS04 5/9/2011 15:36 VHF male 150 or M153 HS04 5/10/2011 1:07 F136 F HS04 7/15/2011 unmarked M HS06 6/29/2011 1:39 M153 M HS06 7/4/2011 18:21 unmarked HS06 7/15/2011 19:36 unmarked M HS07 7/8/2011 3:02 M153 M HS09 7/24/2011 21:01 unmarked Unk Pregnant No. photos of puma 4 Sub-adult Sub-adult Scent type/name Rub response by puma (yes, no) 1 33 7 MT Lynx 0.3 no 1 6 12 MT Lynx 0.3 no 1 3 23 MT Lynx 0.3 no 1 15 16 MT Lynx 0 no 1 9 25 MT Lynx 1.3 no 235 Hair on snag collected yes no �Appendix B continued. Camera site I.D. HS09 Date MS Time Puma Sex 8/6/2011 5:44 unmarked Unk Age stage Female Reproductive Status Time puma was at site (min.) No. photos of puma Time lapse between scent treatment and puma visit (days) 2 21 38 236 Scent type/name Closest puma distance estimate to scent pad-hair snag (m) Rub response by puma (yes, no) Hair on snag collected MT Lynx 0 yes yes �Colorado Division of Parks and Wildlife July 2011 –June 2012 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 3003 1 Federal Aid Project No. W-204-R1 : : : : : Division of Parks and Wildlife Mammals Research Carnivore Conservation Puma Population Structure and Vital Rates on the Uncompahgre Plateau Period covered: July 31, 2011−June 30, 2012 Author: Kenneth A. Logan. Personnel: K. Logan, S. Bard, B. Dunne, W. Hollerman, W. Jesson, R. Navarrete, B. Nay, H. Taylor, S. Waters, B. Banulis, T. Bonacquista, K. Crane, J. Koch, E. Phillips, and G. Watson of CPW; volunteers and cooperators including: private landowners, Bureau of Land Management, Ridgway State Park, Colorado State University, Oklahoma State University, and U.S. Forest Service. Supplemental financial support received in previous years from The Howard G. Buffett Foundation and Safari Club International Foundation. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT The Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) initiated a 10-year study on the Uncompahgre Plateau in 2004 to quantify puma population characteristics in the absence (reference period, years 1-5) and presence (treatment period, years 6-10) of sport-hunting. The purpose of the study is to evaluate assumptions underlying the Colorado Parks and Wildlife model-based approach to managing pumas with sport-hunting in Colorado. The reference period began December 2004 and ended July 2009, during which we captured, sampled, and marked 109 pumas for population research purposes on the Uncompahgre Plateau (Logan 2009). This report provides information on the third year of the treatment period (TY3), August 2011 through July 2012, on puma population characteristics and dynamics with hunting as a mortality factor. Puma sport-hunting opened November 21 and closed December 23, 2011 after a quota of 8 independent pumas was harvested. The harvest was designed to test the management assumption that a 15% harvest of independent pumas results in a stable-to-increasing population. A total of 8 pumas were killed: 3 adult females, 1 adult male, and 4 subadult males. The harvest of 8 independent pumas represented 16.7% of the 48 independent pumas in our minimum count during November 2011 to April 2012. Independent females and males comprised 37.5% and 62.5% of the harvest, respectively. Four other radio-collared independent pumas (2 adult females, 2 adult males) and 3 non-collared adults (1 female, 2 males) in the study area population died during the Colorado puma hunting season. Of those, 2 adult females died of natural causes and the remainder was killed by puma hunters in GMUs adjacent to 134 �the study area. The total mortality of 15 independent pumas during the TY3 hunting season represented 31.2% of the 48 minimum count of independent pumas on the study area. Seventy-four hunters requested mandatory permits with an attached voluntary hunter survey in TY3. Thirty-six of the hunters provided responses to written (n = 31) or telephone call follow-up contact (n = 5). An estimated 49 hunters actually hunted on the study area, of which about 16.3% harvested pumas and 26.5% captured pumas (i.e., harvested plus treed and released). Twenty-four of 26 answering hunters responded that they were selective hunters, and the capture, tracking, and population data indicated that most hunters practiced selection. Puma tracks < 1 day old encountered by hunters and pumas captured by hunters indicated that independent female pumas were detected more frequently than males by hunters. From August 2011 to July 2012 twenty-eight individual pumas were captured 35 times by research teams. Two capture teams with dogs operated over 79 search days from December 27, 2011 through April 12, 2012 to find 268 puma tracks, pursue pumas 89 times, and capture 21 pumas 26 times. Capture efforts with cage traps resulted in the capture of 1 adult female for the first time. Nine new cubs were captured and radio-collared. A total of 42 pumas were monitored by radio-telemetry in TY3. Search efforts also revealed the presence of at least 26 other independent pumas. Our minimum count of 48 independent pumas from November 2011 to April 2012 included: 31 females and 17 males. The minimum count of 48 independent pumas in TY3 was lower than 52 in TY2 and 55 in TY1. A preliminary minimum estimated density of independent pumas was 2.87/100 km2. The proportion of radio-collared adult females giving birth in the August 2011 to July 2012 biological year was 0.19 (3/16). Three litters that could be dated to month of birth were produced in August. Since 2005 a birth peak has occurred from May through August, involving 86% of births. We monitored 20 female and 7 male adult radio-collared pumas for survival and agent-specific mortality. Survival rates in TY3 for adult females (0.548, SE=0.1063) and males (0.167, SE=0.1076) were lower than in TY1 and TY2. A preliminary assessment is that hunting mortality is additive to natural mortality. Of 12 cubs monitored with radiotelemetry in TY3, 6 died. Three died of starvation after their mothers were killed by puma hunters. Three others died of natural-related causes, including 2 that starved after their mother died of a natural cause. One non-marked male cub was struck and killed by a vehicle on state highway 62. Puma harvest, capture, and radio-telemetry data from the beginning of this study to the present provided information on dispersals of 33 pumas initially marked on the study area. Those pumas moved from about 18.2 to 370 km from initial capture sites. We investigated the prevalence of Trichinella spp. in pumas killed in southwest Colorado in collaboration with Dr. Mason Reichard, Oklahoma State University. Twelve of 14 (85.7%) puma tongues were infected with Trichinella. The apparent decline in the puma population on the study area during TY1 to TY3 necessitates a reduction in the harvest quota to continue to test the harvest assumption for a stable-to-increasing puma population. This change will be pursued for TY4 and the results of the harvest monitored through the end of the treatment period. 135 �WILDLIFE RESEARCH REPORT PUMA POPULATION STRUCTURE AND VITAL RATES ON THE UNCOMPAHGRE PLATEAU, COLORADO KENNETH A. LOGAN P. N. OBJECTIVE Quantify puma population sex and age structure; estimate puma population vital rates, including: reproduction of females, stage-specific survival, and immigration and emigration; quantify agent-specific mortality rates; model puma population dynamics; develop and execute the puma harvest manipulation to begin the population-wide test of Colorado Parks and Wildlife (CPW) puma management assumptions in the third year of a five-year Treatment Period of the Uncompahgre Plateau Puma Project― all to improve the CPW model-based approach to managing pumas in Colorado. SEGMENT OBJECTIVES 1. Execute the third year of the five-year treatment period by working with CPW biologists and managers to manipulate the puma population with sport-hunting and to survey hunters. 2. Continue gathering data on puma population sex and age structure. 3. Continue gathering data for estimates of puma reproduction rates. 4. Continue gathering data to estimate puma sex and stage-specific survival rates. 5. Continue gathering data on agent-specific mortality. 6. Explore frequency of Trichinella ssp. in pumas harvested in southwest Colorado in collaboration with Dr. Mason Reichard, Veterinary Health Science, Oklahoma State University. INTRODUCTION Colorado Parks and Wildlife managers need reliable information on puma biology and ecology in Colorado to develop sound management strategies that address diverse public values and the CPW objective of “achieving healthy, self-sustaining populations” through management (Colorado Division Of Wildlife 2002-2007 Strategic Plan:9). Although 4 puma research efforts have been made in Colorado since the early 1970s and puma harvest data is compiled annually, reliable information on certain aspects of puma biology and ecology, and management tools that may guide managers toward effective puma management is lacking. Mammals Research staff held scoping sessions with a number of the CPW’s wildlife managers and biologists prior to initiating the project. In addition, we consulted with other agencies, organizations, and interested publics either directly or through other CPW employees. In general, CPW staff in western Colorado highlighted concern about puma population dynamics, especially as they relate to their abilities to manage puma populations through regulated sport-hunting. Secondarily, they expressed interest in puma―prey interactions. Staff on the Front Range placed greater emphasis on puma―human interactions. Staff in both eastern and western Colorado cited information needs regarding effects of puma harvest, puma population monitoring methods, and identifying puma habitat and landscape linkages. Management needs identified by CPW staff and public stakeholders form the basis of Colorado’s puma research program, with multiple lines of inquiry (i.e., projects): Improve our ability to manage puma hunting with enhanced scientific bases, strategies, and tools― ● Puma population characteristics (i.e., density, sex and age structure). ● Puma population dynamics and vital rates (i.e., birth rates, survival rates, emigration rates, immigration rates, population growth rates). 136 �● Field methods and models for assessing and tracking changes in puma populations. ● Relative vulnerability of puma sex and age classes to hunter harvest. Improve our understanding of puma habitat needs and interrelationships of puma management units― ● Puma habitat use, movements, and use of landscape linkages. ● Puma recruitment patterns (i.e., progeny, immigration, emigration). ● Models for identifying puma habitat and landscape linkages. Improve our understanding of the puma’s role in the ecology of other species ● Relationships of puma to mule deer, elk, and other natural prey. ● Relationships of puma to species of special concern, e.g., desert bighorn sheep. Improve our understanding of puma-human interactions and abilities to manage them ● Behavior of puma in relation to people and human facilities. ● Puma predation on domestic animals. ● Effects of translocating nuisance pumas. ● Effects of aversive conditioning on pumas. While all projects cannot be addressed concurrently, understanding their relationships to one another is expected to help individual projects maximize their benefits to other projects that will assist the CPW to achieve its strategic goal in puma management (Fig.1). This project has been addressing all of the grayshaded components on the left side of the conceptual model in Figure 1. Management issues identified by managers translate into researchable objectives, requiring descriptive studies and field manipulations. Our goal is to provide managers with reliable information on puma population biology and to develop useful tools for their efforts to adaptively manage puma in Colorado to maintain healthy, self-sustaining populations. The highest-priority management needs are being addressed with this intensive population study that focuses on puma population dynamics using sampled, tagged, and GPS/VHF-radio-collared pumas to investigate the effects of sport-hunting and other causes of mortality on puma population dynamics. Those objectives include: Describe and quantify puma population sex and age structure. Estimate puma population vital rates, including: reproduction rates, age-stage survival rates, emigration rates, immigration rates. Estimate agent-specific mortality rates. Improve the CPW’s puma model-based management and attendant assumptions with Coloradospecific data from objectives 1―3. Consider other useful models. Conduct a pilot study to develop methods that yield reliable estimates of puma population abundance. Investigate diseases in pumas. A descriptive and manipulative study will estimate population parameters in an area that appears typical of puma habitat in western Colorado and will yield defensible population parameters based upon contemporary Colorado data. This study will be conducted in two 5-year periods. A completed 5-year reference period, 2004-09, (i.e., absence of recreational hunting) allowed puma life history traits to interact with the main habitat factors that influenced puma population growth (e.g., prey availability and vulnerability, Pierce et al. 2000, Logan and Sweanor 2001, Logan 2009). A subsequent 5-year treatment period started in 2009-10 which involves the use of controlled recreational hunting to manipulate the puma population. 137 �TESTING ASSUMPTIONS AND HYPOTHESES Hypotheses associated with main objectives 1―5 of this puma population research are structured to test assumptions guiding puma management in Colorado. 1. Considering limitations (i.e., methods, number of years, assumption violations) to the previous Colorado-specific studies on puma populations (Currier et al. 1977, Anderson et al. 1992, Koloski 2002), managers assume that puma population densities in Colorado are within the range of those quantified in more intensively studied populations in Wyoming (Logan et al. 1986), Idaho (Seidensticker et al. 1973), Alberta (Ross and Jalkotzy 1992, and New Mexico (Logan and Sweanor 2001). The CPW assumes density ranges of 2.0−4.6 puma/100 km2 (i.e., includes pumas of all stage classes - adults, subadults, and cubs, J. Apker, CPW Carnivore Biologist, person. commun. Nov. 19, 2003) to extrapolate to Data Analysis Units (DAUs) to guide the model-based quota-setting process. Likewise, managers assume that the population sex and age structure is similar to puma populations described in the intensive studies. Using intensive efforts to capture, mark, and estimate non-marked animals developed and refined during the study to estimate the puma population, the following will be tested: H1: Puma densities during the 5-year reference period (absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will vary within the range of 2.0 to 4.6 puma/100 km2 and will exhibit a sex and age structure similar to puma populations in Wyoming, Idaho, Alberta, and New Mexico. 2. Recreational puma hunting management in Colorado DAUs is guided by a model to estimate allowable harvest quotas to achieve one of two puma population objectives: 1) maintain puma population stability or growth, or 2) cause puma population decline (CDOW, Draft L-DAU Plans, 2004, CDOW 2007). These objectives are expected to provide both the capacity for puma population resiliency to achieve a state-wide goal of a healthy, self-sustaining puma population while managing the puma population to provide sport-hunting opportunity and population control in some DAUs (even though puma population dynamics in any DAUs are not known). Basic model parameters are: puma population density, sex and age structure, annual population growth rate, and relative puma habitat quality and quantity. Parameter estimates are currently chosen from literature on studies in western states that are judged to provide reliable information. Background material used in the model assumes a moderate annual rate of growth of 15% (i.e., λ = 1.15) for the adult and subadult puma population (CDOW 2007). This assumption is based upon information with variable levels of uncertainty (e.g., small sample sizes, data from habitats dissimilar to Colorado). Parameters influencing λ include population density, sex and age structure, female age-at-first-breeding, reproduction rates, sex- and age-specific survival, immigration and emigration. H2: Population parameters estimated during a 5-year reference period (in absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will yield an estimated annual adult plus subadult population growth rate that will match or exceed λ = 1.15. 3. An assumption is that the CPW can manage puma population growth through recreational hunting on the basis that for a stable puma population hunting removes the annual increment of population growth (i.e., from current judgments on population density, structure, and λ) Puma harvest rate formulations for DAUs assumes that total mortality (i.e., harvest plus other detected deaths) in the range of 8 to 15% of the harvest-age population (i.e., independent pumas comprised of adults plus subadults) with the total mortality comprised of 35 to 45% females (i.e., adults and subadults) is acceptable to manage for a stable-to-increasing puma population (CDOW 2007). This assumption is 138 �vital to providing the capacity for resiliency in the state-wide puma population which is hunted by applying this assumption to about three-quarters of the puma GMUs in the state. H3: Total mortality of an estimated 15% of the adults and subadults with no more than 45% of the total mortality comprised of females will not result in a declining trend of the harvest-age segment of the population. 4. To reduce a puma population, hunting must remove more than the annual increment of population growth. For DAUs with the objective to suppress the puma population, the total mortality guide of greater than 15 to 28% of the harvest-age population with greater than 45% comprised of females is suggested (CDOW 2007). This assumption is applied to about one-quarter of the GMUs in the state. H4: Total mortality of an estimated 16% or greater of the harvestable population with greater than 45% females will cause a declining trend in the abundance of harvest-age pumas (i.e., adults and subadults). 5. The increase and decline phases of the puma population make it possible to test hypotheses related to shifts in the age structure of the population which have been linked to harvest intensity in Wyoming and Utah. H5: The puma population on the Uncompahgre Plateau study area will exhibit a young age structure after hunting prohibition at the beginning of the reference period. During the 5 years of hunting prohibition, greater survival of independent pumas will cause an older age structure in harvest-age pumas (i.e., adults and subadults) as suggested by the work of Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. As hunting is re-instated in the treatment period, the age structure of harvested pumas and the harvest-age pumas in the population will decline as observed by Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. Researchers in Wyoming (Anderson and Lindzey 2005) concluded that sex and age composition of the harvest varies predictably with puma population size because the likelihood of a specific sex or age class of puma being harvested with the use of hounds is a product of the relative abundance of particular sex and age classes in the population and their relative vulnerability to harvest. Results of that study suggest that managers could use sex and age composition of the harvest to infer puma population changes (Anderson and Lindzey 2005). The CPW currently uses this approach as one tool to infer potential DAU puma population dynamics (CDOW 2007). This assumes no purposeful selection by hunters for any particular sex or age-stage other than the puma must be legal (i.e., independent subadult or adult, not a lactating female or a female in association with spotted cubs) and that changes in the sex and age structure of the harvested pumas is due solely to changes in the relative abundance of particular sex and age classes in the population and their relative vulnerability to harvest. Theoretically, pumas that travel longer distances with movements that intercept access routes used by hunters (i.e., roads, trails) should be more exposed to detection by hunters and thus more vulnerable to harvest. A key assumption to this method is that pumas are killed as they are encountered and the harvest sex and age composition will reliably indicate whether a population is stable, increasing, or declining even if harvest intensity does not vary. Thus, an alternate view is that a population segment, such as independent females, may be more abundant and have shorter movement lengths, yet be detected more frequently by hunters. However, because the same intensively studied Wyoming puma population was manipulated over 6 years with varying intensities of harvest (Anderson and Lindzey 2005), variations in harvest structure using the same harvest level over a period of years could not be examined. This is a property we will investigate during the treatment period on the Uncompahgre Plateau puma study. Moreover, we will directly evaluate to what extent puma harvest might be influenced by hunter selection. A hunter survey is intended to reveal puma hunter behavior, detection of different classes of pumas, and lack of or presence of hunter selection. These data should allow us to examine the credibility of the assumption of non-selection by hunters and the robustness of this technique in gauging puma population dynamics relative to harvest. 139 �We want to examine the usefulness of this approach in Colorado. CPW managers attempt to weight sport-harvest toward male pumas in GMUs with the stable-to-increasing population objective with an active educational program (i.e., mandatory hunter exam, brochure, workshops). Thus, there is a need to test assumptions associated with the Anderson and Lindzey (2005) method. H6: No hunter selection is practiced so that the sex and age structure of pumas harvested by hunters in this population protected from hunting during a 5-year reference period and subsequently managed for stability or increase with conservative harvest levels will reflect the relative vulnerabilities to detection and capture with dogs during each year in the 5-year treatment period in this order from high to low vulnerabilities: subadult males, adult males, subadult females, adult females without cubs or with cubs >6 months old, and adult females with cubs ≤6 months old (Barnhurst 1986, Anderson and Lindzey 2005). In each of the 5 years of the treatment period, subadults and adult males should comprise the majority of the harvest and reflect the assumed sex and age structure (Anderson and Lindzey 2005) of a puma population managed for a stable to increasing phase and not hunted for 5 previous years (i.e., a puma population source). Desired outcomes and management applications of this research include: 1. Quantification of variations in puma population density, sex and age structure, growth rates, vital rates, and an understanding of factors affecting them will aid adaptive puma management by yielding population parameters and tools useful for assessing puma population dynamics, evaluation of management alternatives, and effects of management prescriptions. 2. Testing assumptions about puma populations, currently used by CPW managers, will help managers to biologically support and adapt puma management based on Colorado-specific estimated puma population characteristics, parameters, and dynamics. 3. Methods for assessing puma population dynamics will allow managers to evaluate modeled populations and estimate effects of management prescriptions designed to achieve specified puma population objectives in targeted areas of Colorado. Ascertaining puma numbers and densities during the project will allow assessment of monitoring techniques. Potential methods include use of harvest sex and age structure and photographic and DNA genotype capture-recapture. Study plans to develop and test feasible field and analytical methods will be developed as we learn the logistics of performing those methods, after we have preliminary data on puma demographics and movements which will inform suitable sampling designs, and if we have adequate funding. 4. Information which will be disseminated to citizen stakeholders interested in pumas in Colorado, and thus contribute to informed public participation in puma management. STUDY AREA The study area for the puma population research is on the Uncompahgre Plateau (in Mesa, Montrose, Ouray, and San Miguel Counties; Fig. 2). The study area includes about 2,253 km2 (870 mi.2) of the southern halves of Game Management Units (GMUs) 61 and 62, and about 155 km2 (60 mi.2) of the northern edge of GMU 70 (between state highway 145 and San Miguel River). The area is bounded by state highway 348 at Delta, 25 Mesa road and Forest Service road FS503 to Nucla, state highway 97 to state highway 141 to state highway 145 to Placerville, state highway 62 to Ridgeway, U.S. highway 550 to Montrose, and U.S. highway 50 to Delta. The study area seems typical of puma habitat in Colorado that has vegetation cover that varies from the pinion-juniper covered foothills starting from about 1,700 m elevation to the spruce-fir and aspen forests growing to the highest elevations of about 3,000 m. Mule deer (Odocoileus hemionus) and elk (Cervus elaphus) are the most abundant wild ungulates available for puma prey. Cattle and domestic sheep are raised on summer ranges on the study area. People reside year-round along the eastern and western fringe of the area, and there is a growing residential presence especially on the southern end of 140 �the plateau. A highly developed road system makes the study area easily accessible for puma research efforts. A detailed description of the Uncompahgre Plateau is in Pojar and Bowden (2004). METHODS Reference and Treatment Periods This research was structured in two 5-year periods: a reference period (years 1―5) and a treatment period (years 6―10). The reference period was closed to puma hunting on the study area and was expected to cause a population increase phase. The treatment period (starting in November 2009) involves manipulation of the puma population with sport-hunting structured to achieve a management objective for a stable to increasing population. In both phases, puma population structure, and vital rates are being quantified, and management assumptions and hypotheses regarding population dynamics and effects of harvest are being tested. Contingent upon results of pilot studies, we will also assess enumeration methods for estimating puma population abundance. The reference period, without recreational puma hunting as a major limiting factor, was consistent with the natural history of the current puma species in North America which evolved life history traits during the past 10,000 to 12,000 years (Culver et al. 2000) that enable pumas to survive and reproduce (Logan and Sweanor 2001). In contrast, puma hunting, with its modern intensity and ingenuity, might have influenced puma populations in western North America for at least the past 100 years. Hence, the reference period, years 1 to 5, provided conditions where individual pumas in this population expressed life history traits interacting with the environment without recreational hunting as a limiting factor. Theoretically, the main limiting factor was vulnerable prey abundance (Pierce et al. 2000, Logan and Sweanor 2001). This allowed researchers to understand basic system dynamics before manipulating the population with controlled recreational hunting. In the reference period, all pumas in the study area were protected, except for individual pumas involved in depredation on livestock or human safety incidents. In addition, all radio-collared and ear-tagged pumas that ranged in a buffer zone in the northern halves of GMUs 61 and 62 were protected from recreational hunting mortality. The reference period allowed researchers to quantify baseline demographic data on the puma population to estimate parameters useful for assessing the CPW’s assumptions for its model-based approach to puma management. The reference period also facilitated other operational needs (because hunters did not kill the animals) including the marking of a large proportion of the puma population for parameter estimates and gathering movement data from GPS-collared pumas. During the treatment period, years 6 to10, recreational puma hunting is occurring on the same study area using management prescriptions structured from information learned during previous years. Using recreational hunting for the treatment is consistent with the CPW’s objectives of manipulating natural tendencies of puma populations, particularly survival, to maintain either population stability or increase or suppression (CDOW, Draft L-DAU Plans, 2004). Theoretically, survival of independent pumas is being influenced mainly by recreational hunting, which is being quantified by agent-specific mortality rates of radio-collared pumas. Dynamics of the puma population are being manipulated to evaluate hypotheses that are related to effects of hunting (i.e.,: effects of harvest rates, relative vulnerability of puma sex and age classes to hunting, variations in puma population structure due to hunting). The killing of tagged and collared pumas during the treatment period is not hampering operational needs (as it would have during the start-up years), because a majority of independent pumas in the population have already been marked, and sampling methods formalized. Pumas on the study area that may be involved in depredation of livestock or human safety incidences may be lethally controlled. Researchers that find that GPS-collared pumas have killed domestic livestock will record such incidents to facilitate reimbursement to the property owner for loss of 141 �the animal(s). In addition, researchers will notify the Area Manager of the CPW if they perceive that an individual puma may be a threat to public safety. Field Methods Puma Capture: Realizing that pumas live at low densities and capturing pumas is difficult, as a starting point, our logistical aim was to have a minimum of 6 puma in each of 6 categories (36 total) radio-tagged in any year of the study if those or greater numbers are present. The 6 categories are: adult female, adult male, subadult female, subadult male, female cub, male cub. Our aim was to provide more quantitative and precise estimates of puma demographics than were achieved in earlier Colorado puma studies. This relatively large number of pumas might represent the majority of the puma population on the study area, and would provide the basic data for age- and sex-specific reproductive rates, survival rates, agent-specific mortality rates, emigration, and other movement data. Puma capture and handling procedures were approved by the CPW Animal Care and Use Committee (file #08-2004). All captured pumas were examined thoroughly to ascertain sex and describe physical condition and diagnostic markings. Ages of adult pumas were estimated initially by the gum-line recession method (Laundre et al. 2000) and dental characteristics of known-age puma (Logan and Sweanor, unpubl. data). Ages of subadult and cub pumas were estimated initially based on dental and physical characteristics of known-age pumas (Logan and Sweanor unpubl. data). Body measurements recorded for each puma included at a minimum: mass, pinna length, hind foot length, plantar pad dimensions. Tissue collections included: skin biopsy (from the pinna receiving the 6 mm biopsy punch for the ear-tags), and blood (30 ml from the saphenous or cephalic veins) for genotyping individuals, parentage and relatedness analyses, and disease screening; hair (from various body regions) for genotyping tests of field gathered samples. Universal Transverse Mercator Grid Coordinates on each captured puma were recorded via Global Positioning System (GPS, North American Datum 27). Pumas were captured year-round using 4 methods: trained dogs, cage traps, foot-hold snares, and by hand (for small cubs). Capture efforts with dogs were conducted mainly during the winter when snow facilitated thorough searches for puma tracks and enabled dogs to follow puma scent. The study area was searched systematically multiple times per winter by four-wheel-drive trucks, all-terrain vehicles, snowmobiles, and on foot. When puma tracks ≤1 day old were detected, trained dogs were released to pursue pumas for capture. Pumas usually climbed trees to take refuge from the dogs. Adult and subadult pumas captured for the first time or requiring a change in telemetry collar were immobilized with Telazol (tiletamine hydrochloride/zolazepam hydrochloride) dosed at 5 mg/kg based on estimated body mass (Lisa Wolfe, DVM, CPW, attending veterinarian, pers. comm.). The immobilizing agent was delivered into the caudal thigh muscles via a Pneu-Dart® shot from a CO2-powered pistol. Immediately, a 3m-by-3m square nylon net was deployed beneath the puma to catch it in case it fell from the tree. A researcher climbed the tree, fixed a Y-rope to two legs of the puma and lowered the cat to the ground with an attached climbing rope. Once the puma was on the ground, its head was covered, its legs tethered, and vital signs monitored (Logan et al. 1986). Normal signs include: pulse ~70 to 80 bpm, respiration ~20 bpm, capillary refill time ≤2 sec., rectal temperature ~101oF average, range = 95 to 104oF (Kreeger 1996). Pumas that climbed trees too dangerous for the pumas or researchers for capture were released without handling, or we encourage the animals to leave the tree by heaving snowballs toward them. If the pumas climbed a safe tree, then we handled them as described above. A cage trap was used to capture adults, subadults, and large cubs when pumas were lured into the trap using road-killed or puma-killed ungulates (Sweanor et al. 2008). A cage trap was set only if a target puma scavenged on the lure (i.e., an unmarked puma, or a puma requiring a collar change). Researchers continuously monitored the set cage trap from about 1 km distance by using VHF beacons on the cage 142 �and door. Researchers handled captured pumas within 30 minutes of capture. Puma were immobilized with Telazol injected into the caudal thigh muscles with a pole syringe. Immobilized pumas were restrained and monitored as described previously. If non-target animals were caught in the cage trap, we opened the door and allowed the animal to leave the trap. Small cubs (≤10 weeks old) were captured using our hands (covered with clean leather gloves) or with a capture pole. Cubs were restrained inside new burlap bags during the handling process and were not administered immobilizing drugs. Cubs at nurseries were approached when mothers were away from nurseries (as determined by radio-telemetry). Cubs captured at nurseries were removed from the nursery a distance of 30 to 100 m to minimize disturbance and human scent at nurseries. Immediately after handling processes were completed, cubs were returned to the exact nurseries where they were found (Logan and Sweanor 2001). Marking, Global Positioning System- and Radio-telemetry: Pumas did not possess easily identifiable natural marking, such as tigers (see Karanth and Nichols 2002), therefore, the capture, marking, and GPS- or VHF- collaring of individual pumas was essential to a number of project objectives, including estimating numbers, vital rates, and gathering movement data relevant to population dynamics (i.e., emigration and movement across Data Analysis Unit boundaries). Adults, subadults, and cubs were marked 3 ways: GPS/VHF- or VHF-collar, ear-tag, and tattoo. The identification number tattooed in the pinna was permanent and could not be lost unless the pinna was severed. A colored (bright yellow or orange), numbered rectangular (5 cm x 1.5 cm) ear-tag (Allflex USA, Inc., DFW Airport, TX) was inserted into each pinna to facilitate individual identification during direct recaptures. Cubs ≤10 weeks old were ear-tagged in only one pinna. Adult and subadult female pumas were fitted with GPS collars (approximately 400 g each, Lotek Wireless, Canada) if available. Initially, GPS-collars were programmed to fix and store puma locations at 4 times per day to sample daytime, nighttime, and crepuscular locations (i.e., 0:00, 06:00, 12:00, 19:00). GPS locations for pumas provided precise, quantitative data on movements to assess the relevance of puma DAU boundaries, our search efforts, and to evaluate puma behavior and social structure. The GPScollars also provided basic information on puma movements and locations to design other pilot studies in this program on vulnerability of puma to sport-harvest, habitat use, and enumeration methods (e.g., photographic or DNA mark-recapture). Subadult male pumas were fitted initially with conventional VHF collars (Lotek, LMRT-3, ~400 g each) with expansion joints fastened to the collars, which allowed the collar to expand to the average adult male neck circumference (~46 cm). If subadult male pumas reached adulthood on the study area, we would recapture them and fit them with GPS collars. In addition, other adult and female subadult pumas were fitted with VHF collars when GPS collars were not available. VHF radio transmitters on GPS collars enabled researchers to find those pumas on the ground in real time to acquire remote GPS data reports, facilitate recaptures for re-collaring, and to determine their reproductive and survival status. VHF transmitters on GPS- and VHF-collars had a mortality mode set to alert researchers when pumas were immobile for 3 to 24 hours so that dead pumas could be found to quantify survival rates and agent-specific mortality rates by gender and age. Locations of GPS- and VHFcollared pumas were identified about once per week (as flight schedules and weather allowed) from light fixed-wing aircraft (e.g., Cessna 185) fitted with radio signal receiving equipment (Logan and Sweanor 2001). GPS- and VHF-collared pumas were located from the ground opportunistically using a hand-held yagi antenna. At least 3 bearings on peak aural signals were mapped to fix locations and estimate location error around those locations (Logan and Sweanor 2001). Aerial and ground locations were plotted on 7.5 minute USGS maps (NAD 27) and UTMs along with location attributes were recorded on standard forms. GPS and aerial locations were mapped using GIS software. 143 �We attempted to collar all cubs in observed litters. Cubs were fit with small VHF transmitters mounted on expandable collars that expand to adult neck size (Wildlife Materials, Murphysboro, Illinois, HLPM-2160, 47g, Telonics, Inc., Mesa, Arizona MOD 080, 62g, or Telonics MOD 205, 90g,) when cubs weighed 2.3―11 kg (5―25 lb). Cubs could wear these small expandable collars until they were over 12 months old. Cubs were recaptured to replace collars as opportunities allowed. Monitoring radio-collared cubs allowed quantification of survival rates and agent-specific mortality rates (Logan and Sweanor 2001). Analytical Methods Population characteristics each year were tabulated with the number of individuals in each sex and age category. Age categories, as mentioned, include: adult (puma ≥24 months old, or younger breeders), subadults (young puma independent of mothers, <24 months old that do not breed), cubs (young dependent on mothers, also called kittens) (Logan and Sweanor 2001). When data allowed, age categories were further partitioned into months or years. Reproductive Rates: Reproductive rates were estimated for GPS- and VHF-collared female pumas directly (Logan and Sweanor 2001). Genetic paternity analysis will be used to ascertain paternity for adult male pumas (Murphy et al. 1998). Survival and Agent-specific Mortality Rates: Radio-collared pumas provided known fate data used to estimate survival rates for each age stage using the Kaplan-Meier procedure to staggered entry (Pollock et al. 1989). A binomial survival model was also used for crude estimates of survival during the subadult age stage (Williams et al. 2001:343-344). In addition, when data collection is complete, survival rates will be modeled in program MARK (White and Burnham 1999, Cooch and White 2004) where effects of individual (e.g., sex, age stage, reproductive stage) and temporal (i.e., reference period, treatment period) covariates to survival can be examined. Agent-specific mortality rates can also be analyzed using proportions and Trent and Rongstad procedures (Micromort software, Heisey and Fuller 1985). Population Inventory: The population of interest was independent pumas (i.e., adults and subadults) mainly during November to March which corresponds with the Colorado puma hunting season. Independent pumas were those that could be legally killed by recreational hunters. Initially, we estimated the minimum number of independent pumas and puma density (i.e., number of independent puma/100 km2) each winter. The minimum number of independent pumas included all marked pumas known to be present on the study area during the period, plus individuals thought to be non-marked and detected by visual observation or tracks that were separated from locations of radio-collared pumas. Furthermore, adults comprised the breeding segment of the population and subadults were non-breeders that are potential recruits into the adult population in ≤1 year. The sampling unit was the individual independent puma (~≥1 yr. old). Puma Population Dynamics: A deterministic, discrete time model parameterized with population characteristics and vital rates from this research was used to assess puma population dynamics (Logan 2008). Functional Relationships: Once data collection is complete, a variety of analyses will be conducted to estimate parameters and examine functional relationships. Graphical methods will be used to initially examine functional relationships among puma population parameters. Linear regression procedures and coefficients of determination will be used to assess functional relationships if data for the response variable are normally distributed and the variance is the same at each level. If the relationship is not linear, data is non-normal, and variances are unequal, we will consider appropriate transformations of 144 �the data for regression procedures (Ott 1993). Non-parametric correlation methods, such as Spearman’s rank correlation coefficient, will also be used where appropriate to test for monotonic relationships between puma abundance and other parameters of interest (Conover 1999). Relationships of explanatory variables to survival parameters will be modeled in MARK. Statistical analyses can be performed in a variety of software (e.g., SYSTAT, R, and MARK). RESULTS AND DISCUSSION Segment Objective 1 Puma harvest: This biological year, August 2011 to July 2012, was the third year of the treatment period (TY3) in this study of puma population dynamics on the Uncompahgre Plateau. The hunting season on the study area began on November 21, 2011 and was scheduled to extend to January 31, 2011, unless the harvest quota was taken before then. The harvest design quota was 8 pumas (i.e., 15% harvest of the estimated minimum number of independent pumas), with the objective to manage for a stable to increasing population. This harvest design tests the CPW’s current assumption that total mortality (i.e., harvest plus other natural deaths) in the range of 8 to 15% of the harvest-age population (i.e., independent pumas comprised of adults plus subadults) with the total mortality comprised of 35 to 45% females (i.e., adults and subadults) is acceptable to manage for a stable-to-increasing puma population (Assumption and Hypothesis 3 p.5-6 this report). The initial quota of 8 pumas for TY1 was based on the projected minimum number of 53 independent pumas expected on the study area in winter 2009-10, modeled from a minimum count of pumas during winter 2007-08 (Table 1; Logan 2010). The quota of 8 pumas for TY3 was based on the observed minimum count of 52 independent pumas during November 2010 to April 2011 in TY2 and that approximately the same number of independent pumas was expected during the puma hunting season for TY3. The hunting structure in TY3 was the same as in TY1 and TY2. The number of puma hunters on the study area was not limited. Each hunter on the study area was required to obtain a hunting permit from the CPW Montrose Service Center. Permits were free and unlimited. Each permit allowed the individual hunter with a legal puma hunting license in Colorado to hunt in the puma study area for up to 14 days from the issue date. Unsuccessful hunters that wanted to continue hunting past the permit expiration date requested a new permit for another 14 days, or until the hunter killed a puma within the season, or the season on the study area closed due to the quota being reached, or the end of the hunting season. This permit system allowed the CPW to monitor the number of hunters on the study area and to contact each hunter for survey information (see later in this section). All pumas harvested on the study area were examined by principal investigator K. Logan or a wildlife research technician and sealed as mandated by Colorado statute. All successful hunters reported their puma kill and presented the puma carcass for inspection by CPW within 48 hours of harvest. Upon inspection, the following data were recorded: sex, age, and location of harvest. In addition, an upper premolar tooth was collected for aging (i.e., mandatory) and a tissue sample was collected for DNA genotyping. Each successful hunter was also asked at that time to complete a one-page hunter survey form. All other hunters that did not report a puma kill on the study area were asked to complete the survey form and return it in a stamped envelope that was provided. An attempt was made to contact other hunters by telephone if they did not mail in surveys. The puma hunting season occurred on the study area from November 21 to December 23, 2011, taking 33 days to fill the quota of 8 pumas. This was 12 days more than it took to harvest 8 pumas in TY2 (i.e., 21 days, Nov. 22 to Dec. 12, 2010) and 7 more days than it took to harvest 8 pumas in TY1 (i.e., 26 days, Nov. 16 to Dec. 11, 2009). Eight pumas were killed on the study area, including: 3 adult females, 1 adult male, and 4 subadult males (Table 2). Of the 8 harvested pumas, 6 were marked: F3, F70, F75, M120, M138, and M141. In addition to the pumas killed on the study area during the Colorado puma 145 �hunting season, adult males M67 and M87 were killed by hunters in north GMU61, and adult females F104 and F119 died of natural causes. In addition, 3 non-marked adult pumas that apparently ranged on the study area were killed by hunters (Table 3). Of those, one adult male was pursued across 25 Mesa Road, the north study area boundary, and killed in north GMU62; another adult male was snow-tracked across Colorado state highway 145, a south study area boundary, and was killed in east GMU70; and an adult female with a radio-collared cub that ranged on the study area was killed adjacent to the study area in north GMU61. All these pumas were included in the minimum count of pumas for TY3. The harvest of 8 independent pumas on the study area was 16.7% (8/48*100) of the minimum count of 48 independent pumas counted on the study area, including 31 females and 17 males, determined by the research team during November 2011 to April 2012 (Table 4). Independent females and males comprised 37.5% (3/8*100) and 62.5% (5/8*100) of the harvest, respectively. This harvest structure was 9.7% (3/31*100) of the independent females and 29.4% (5/17*100) of the independent males. Considering the mortality of 4 other radio-collared adults (F104, F119, M67, M87) and 3 noncollared adults (1 female, 2 males) (Table 3), the mortality of 15 independent pumas was 31.2% (15/48*100) of the minimum number of independent pumas. The mortality composition of 6 females and 9 males was comprised of 40.0% (6/15*100) females and 60.0% (9/15*100) males. This harvest structure was 19.4% (6/31*100) of the independent females and 52.9% (9/17*100) of the independent males in the minimum count. The minimum count of 48 independent pumas in TY3 was lower than the minimum count of 52 independent pumas in TY2 and 55 independent pumas in TY1 (Table 4, Fig.3.). Minimum count TY3 = 48 independent pumas, including 31 females and 17 males. This count reflected the relatively low adult female and low adult male survival rates (see Table 15, later). Because the harvest quota of 8 independent pumas in TY1 resulted in a minimum count of 52 independent pumas in TY2 and was expected to result in a stable-to-increasing population trend, we decided to set the quota to harvest 8 independent pumas in the TY3 (2011-12) hunting season to emulate an approximate 15% harvest of independent pumas to achieve a stable to increasing population objective while also considering that a number of independent pumas in the study area population might be killed outside of the study area as in the TY1 and TY2 hunting seasons. However, the additional pumas killed by hunters outside of the study area and natural mortality occurring during the hunting season and other parts of the biological year has apparently resulted in a declining population trend (Fig. 3). Hunter permits and survey: In TY3 mandatory permits with the voluntary survey attached were requested by 74 individual puma hunters. This number is up from 64 hunters in TY2 and down from 79 individual hunters in TY1. Twenty-three of the hunters requested a second permit, 13 hunters requested a third permit, and one hunter requested a fourth permit after a previous permit expired after 14 days. Thirty-six hunters (48.6%) provided responses to the voluntary survey either by turning in the printed survey (n = 31) or providing information during follow-up telephone calls (n = 5) by principal investigator K. Logan. The remaining 38 hunters could not be contacted because either they did not have working phone numbers or they did not return calls. Of the respondents, 12 hunters indicated that they did not hunt on the study area. The proportion of the 36 respondents that hunted extrapolated to the total of 74 hunters (24/36 = 0.666) indicated that about 49 hunters took to the field for pumas on the study area during the 33-day TY3 hunting season. This was up from 42 hunters in TY2, but down from 67 hunters in TY1 (Logan 2010, 2011). Considering that 49 hunters were estimated to be afield, then 16.3% of the hunters harvested pumas (8/49*100) and 26.5% of hunters captured pumas (13/49*100; see captured and released pumas below and in Table 5). The 31 puma hunters that turned in the written volunteer survey were asked to answer, “Do you consider yourself a selective or non-selective hunter?” A selective hunter is one that purposely is hunting 146 �for a specific type of legal puma, such as a male, large male, or large female. A non-selective hunter is one that intends to take whatever legal puma is first encountered or caught, with no desire for sex or size. Selective hunter was indicated by 24 respondents (92.3%; 24/26 = 0.923). Of the remaining 7 hunters, 2 indicated they were non-selective (7.7%), and 5 did not answer the question because they indicated that they did not hunt on the study area. The volunteer hunter survey also revealed that hunters treed pumas on the study area, but they chose not to kill them (Table 5). Those hunters reported they treed pumas 4 times and observed one, including 2 adult females (1 of them twice), 1 female of unspecified age-class, and 2 “young” males (1 male treed by 2 hunters). Two of the females were marked with collars and ear-tags. Hunters gave various reasons for not wanting to kill the pumas, including reasons based on puma sex, size, and one hunter did not want to kill a puma (Table 5). In an effort to better ascertain the vulnerability of sexes and age-stages (i.e., adult, subadult) of independent pumas to detection by puma hunters and hunter selection to address assumption 6 and hypothesis 6 (previously), the survey was changed in TY2 to ask hunters, “What was the sex of the lion that made the first set of tracks you encountered that were less than one day old?”. This question pertained to pumas that could be pursued by dogs and captured with a relatively high probability to allow the hunter an opportunity to harvest the puma. Associated with the question, we asked, “Did you pursue the lion to harvest it?” Hunters’ responses in TY3 showed they encountered 21 puma tracks less than one day old. Of those, 15 tracks were of females, and 6 tracks were of males, indicating that during the hunting season females are more detectable than males by a ratio of 2.5:1, consistent with the sex structure of independent pumas in the minimum count on the study area which was 31 females and 17 males (Table 4). Of the 15 female tracks, 1 female puma was pursued by a hunter with intent to harvest it, and that female was killed. Nine hunters indicated they observed female tracks as their first tracks <1day old, but did not pursue the puma with intent to harvest it. Another 4 hunters did not answer the question, “Did you pursue the lion to harvest it?” Six hunters indicated they observed male tracks as their first track <1 day old; 4 indicated they pursued the puma to harvest it, and 3 male pumas were killed. Two hunters indicated they did not pursue male pumas to harvest them. These preliminary survey and harvest data for TY3 indicate that hunters detect independent females more frequently than male pumas and females were captured by hunters slightly less than or about the same frequency as independent males by 6 to 7 (i.e., females = 3 harvested + 3 captured and released; males = 5 harvested + 2 captured and released). Moreover, hunters were choosing to kill males more frequently than females. Results in TY3 indicated selection for male pumas by hunters was consistent with TY1 and TY2 results where hunters caught females slightly more frequently than males, yet the males were selected for harvest. This preliminary assessment from years TY1, TY2, and TY3 puma harvest and hunter survey data suggests that female pumas were detected by hunters more frequently than male pumas, most puma hunters were selective, and hunter choices influenced harvest sex and age composition. Segment Objective 2 After the harvest quota was filled, puma research teams immediately initiated capture operations with trained dogs. Two fully-staffed capture teams, one each detailed on the east and west slopes, systematically and thoroughly searched the study area to capture, sample, and GPS/VHF radio-collar pumas the remainder of winter and early spring 2011-12. These efforts along with cage trap efforts and hand-capturing cubs at nurseries maintained samples to quantify population sex and age structure, survival, and agent-specific mortality, and allowed determination of minimum population size on the study area. We made 35 puma captures of 28 individuals from August 2011 to July 2012 (Tables 6-11); 21 individual pumas were captured with dogs 26 times. One puma was captured in a cage trap. Six cubs were captured at nurseries by hand. A total of 42 individual pumas were monitored with radio-telemetry from 147 �August 2011 to July 2012 (some of these had been collared in previous years), representing sex and age classes including: 19 adult females, 7 adult males, 4 subadult females, 4 subadult males, and 12 cubs (i.e., 1 cub and 2 subadult males survived to older age classes during the biological year). Trained dogs were used as our main method to capture, sample, and mark pumas from December 27, 2011 to April 12, 2012. Those efforts resulted in 79 search days, 268 total puma tracks detected of which 138 were ≤1 day old, 89 pursuits, and a total of 26 puma captures of 21 individual pumas (Table 6). This was the third year we deployed 2 fully-staffed hound capture teams in the treatment period. Search days with dogs in TY3 (79 days) were similar to TY2 (81 days), but slightly lower than TY1 (86 days) (Table 12). The frequency of tracks (tracks/day) encountered in TY3 was slightly lower than TY2, but slightly higher than TY1. The number of pursuits in TY3 was 10 less than in TY2 and 4 less than in TY1. The capture rate in TY3 was less than half that in TY2, but similar to TY1. The number of new pumas captured for the first time in TY3 was 4 less than TY2, but 2 more than TY1 (Table 12). Researchers in the two hound capture teams from December 27, 2011 to April 12, 2012 also recorded instances when the first tracks ≤1 day old of independent pumas were encountered on each search route each day to represent encounters with puma tracks that could be detected and pursued by puma hunters. The count was: 70 tracks of females, including 17 associated with cubs; 2 of 2 orphaned cubs; 12 tracks of males; and 2 tracks of unspecified sex. These tracks ≤ 1 day old were found after the TY3 puma hunting season when 3 independent females and 5 independent males were harvested (Table 2). Therefore, the harvested pumas were not present to make tracks for our researchers to observe. The loss of the 3females and 5 males may be reflected in the substantially higher ratio of female:male tracks post-hunting season. By comparison, the number of female to male tracks reported by puma hunters in TY3 was 15 females and 6 males (Segment Objective 1 above). Puma capture efforts using ungulate carcasses and cage traps was sporadic from October 5, 2011 to April 11, 2012 (Table 10). We used 21 road-killed mule deer at 18 different sites. One independent adult female puma, F172, was captured for the first time. Pumas scavenged at 3 of 21 (14.29%) sites where deer carcasses were used for bait. We sampled 9 new cubs, including 2 females and 7 males (Table 11). All were radio-collared to monitor survival and agent-specific mortality (Appendix A). In addition to our direct puma captures with dogs December through April, we detected 17 radiocollared pumas that we were able to identify with GPS or VHF telemetry 40 times, thus, negating the need to capture those pumas directly with dogs (Table 6). Upon detecting puma tracks that were aged at ≤1 day old, we followed the tracks with a radio receiver in an effort to detect if the tracks might be of a puma wearing a functional collar. We assigned tracks to a collared individual if we received radio signals from a puma that we judged to be <1 km from the tracks and in direction of travel of the tracks. This approach allowed us to more efficiently allocate our capture efforts toward pumas of unknown identity on the study area, particularly unmarked pumas or pumas with non-functioning GPS- or VHF- radiocollars. In addition to the harvest and capture data (previously), our search efforts also revealed the presence of at least 26 other pumas which we included in our minimum count November 2011 through April 2012 (Table 4). We classified those pumas as: 10 adult females, 4 adult males, 1 subadult female, 1 subadult male, and 10 cubs. Two adult females and 2 cubs were treed by our hounds, but we could not handle the pumas because they climbed dangerous trees (Table 8). Of those, 2 adult females were sampled with biopsy darts to obtain a tissue sample for genotyping the individuals. We could separate the activity of the other pumas from the GPS- and VHF- collared pumas in time, space, and track size differences between females, males, and numbers of cubs with females. Moreover, of the 26, 4 non- 148 �marked independent pumas (3 females, 1 male) and 4 non-marked cubs were confirmed with photographs from digital trail cameras. Our search and capture efforts during December 2011 through April 2012 and information from the puma hunting season in TY3 enabled us to quantify a minimum count of 48 independent pumas detected on the Uncompahgre Plateau study area, including 31 independent females and 17 independent males (Table 4). This count was based on the number of known radio-collared pumas, non-marked pumas harvested by hunters on the study area, observations of marked and non-marked pumas observed by researchers or pursued, treed and released by hunters on and adjacent to the study area, and puma tracks observed by researchers that could not be attributed to pumas with functioning radiocollars. Of the 48 independent pumas, 27 (56%) were marked and 21 (44%) were assumed to be non-marked animals (i.e., some may have ear-tags and tattoos). The abundance and sex structure of independent pumas on the east and west slopes of the study area were similar. The east slope count included 21 independent pumas (14 females, 7 males). The west slope count included 27 independent pumas (17 females, 10 males). A decline in the study area puma population is evident on the east slope. Considering the minimum count of 48 independent pumas, a preliminary minimum density for the winter puma habitat area estimated at 1,671 km2 on the Uncompahgre Plateau study area was 2.87 independent pumas/100 km2. The TY3 minimum count of 48 independent pumas is lower than the two previous treatment years TY1 and TY2 and appears to signal a declining trend in the puma population on the Uncompahgre Plateau study area (Fig. 3). The declining trend is further supported by declining survival rates of adult pumas on the study area (see Segment Objective 4&5 below). Taking into account the apparent declining trend in the number of independent pumas, a simple linear regression model of minimum counts of independent pumas in TY1, TY2, and TY3 on year projected that a minimum of 45 independent pumas could be expected in TY4 if the population decline continues. The recommended puma harvest for TY4 will be 5 pumas, representing 11.1% of the 45 expected number of independent pumas. This harvest rate is in the mid-range of the 8-15% test assumption for a stable to increasing population. The estimated age structure of independent pumas in November 2011 at the beginning of the puma hunting season in TY3 on the Uncompahgre Plateau study area is depicted in Figure 4. The male age structure has declined when compared with TY1 and TY2 (Logan 2010, 2011). The female age structure is more evenly distributed and does not yet reflect a decline in survival rates of adult females in TY3 (Logan 2010, 2011). In addition to the independent pumas, we counted a minimum of 19 cubs in TY3 (Table 4). Segment Objective 3 During the past 7.7 years of this work we compiled data on puma reproduction that was not previously available on pumas in Colorado (Table 13). Puma reproduction data (i.e., litter size, sex structure, gestation, birth interval, proportion of females giving birth per year) were summarized for the reference period in Logan (2009). In TY3 we directly observed 4 litters in nurseries which were born in June 2012 (1; F118’s cub not marked), July 2011 (1) and August 2011 (2), each with 1 to 3 cubs born to radio-collared females. Data on reproduction we observed in TY1, TY2, and TY3 were added to Table 13 which gives the reproductive chronology and information on mates of reproducing females. But those data will not be summarized again until the end of the treatment period. The proportion of radio-collared adult females giving birth from August 2011 to July 2012 biological year (TY3) was 0.19 (3/16), substantially lower than TY1 (0.53, 8/15) and TY2 (0.56, 9/16), further evidence for a declining puma population. 149 �Considering our 46 total litters from 24 females, including 44 observed with cubs 26 to 42 days old and 2 other litters confirmed by nurseries and nursling cub tracks with GPS-collared females (the latter include F111’s cubs caught later when 8.5 months old) (Table 13), the distribution of puma births by month since 2005 indicate births extending from March into September (Fig. 5). Births are high in May and June, peak in July, and decline in August and September. Births during late spring to late summer (May to August) involve 86% of the births (Fig. 5). The data indicate that the large majority of puma breeding activity occurred February through May (i.e., gestation averages about 90-92 days, Logan 2009). In comparison, Anderson et al. (1992:47-48) found on the Uncompahgre Plateau during 19821987 that of 10 puma birth dates 7 were during July, August, and September, 2 in October, and 1 in December, with most breeding occurring April through June. The 2 data sets indicated puma births on the Uncompahgre Plateau have occurred in every month except January and November (so far). As we gather more data on the puma births during the treatment period, we will examine the distributions of births in the reference and treatment periods separately for a treatment effect on timing of breeding and births. Segment Objectives 4 & 5 From December 8, 2004 (capture and collaring of the first adult puma M1) to July 31, 2012, we radio-monitored 21 adult male and 34 adult female pumas to quantify survival and agent-specific mortality rates (Table 14). Survival and agent-specific mortality of adult pumas were summarized for the reference period in Logan (2009). Preliminary estimates of adult puma survival rates in the absence of sport-hunting during the reference period indicated high survival, with adult male survival generally higher than adult female survival (Table 15). We monitored 20 adult females and 7 adult males for annual survival and agent-specific mortality in TY3. Annual survival rate for adult females was 0.548 (SE=0.1063) and for males was 0.167 (SE=0.1076). Preliminary adult puma survival for TY1, TY2, and TY3 are also shown in Table 15. So far, adult male survival is substantially lower in the treatment period than in the reference period. Adult female survival is lower in TY1 and TY3, with marked decline in TY3. Yet, female survival is generally higher than male survival. These characteristics are probably indicative of hunter selection for male pumas (previously in Segment Objective 1). The lower adult puma survival rates were consistent with an observed decline in the puma population on the study area (see Segment Objective 2, previously). Human-related factors caused 8 deaths of radio-marked adult pumas in TY3, including: sporthunting harvest (3 males- M67, M87, M138; 3 females- F3, F70, F75), illegal shooting (M73), and depredation control (1 male- M153) (Tables 2, 3, 14). In addition, 6 adult female pumas died of natural causes: F23 and F24 were killed by a male puma; F104 apparently died of starvation associated with senescence; F116 apparently died of complications associated with pregnancy and parturition; F119 died of a ruptured uterus and internal bleeding associated with pregnancy, and F135 died of unknown natural cause (Table 14). The occurrence of an increasing frequency of natural deaths and declining adult survival rates in this hunted puma population suggests that sport-hunting causes additive mortality. We have radio-monitored 27 subadult pumas (i.e., independent pumas <24 months old), including 11 females and 16 males (Table 16). We lost contact with 2 males that probably dispersed from the study area unknown distances. Of the remaining 25 subadults (females and males combined), 6 (2 females, 4 males) died before reaching adulthood, indicating a preliminary binomial survival rate of 0.76 (i.e., 19/25). F66 died at 23 months old of trauma to internal organs that caused massive bleeding attributed to trampling by an elk or mule deer. M99 died at about 16 months old; punctures to his skull were consistent with canine bites from another puma and suggested intra-species strife as cause of death. M115 died at about 14 months old due to complications of a broken left foreleg, cause unknown. This injury probably affected his ability to efficiently kill prey. F143 was killed and eaten by a male puma while in competition for an elk carcass that one of the pumas killed. Two subadult males were killed by puma hunters. We 150 �need to increase our efforts to acquire larger samples of male and female radio-monitored subadult pumas to acquire reliable estimates of their survival. Harvest data along with our capture and radiotelemetry data provided dispersal and fate information on 33 marked pumas, 25 males and 8 females. Of those, 25 (4 females, 21 males) were initially captured and marked as cubs, and 8 (4 females, 4 males) were captured and marked in the subadult life-stage on the Uncompahgre Plateau puma study area (Table 17). Twenty males were killed away from the study area by hunters at linear distances (i.e., from initial capture sites to kill sites) ranging from about 20 to 370 km. Two males with extreme moves were killed in the Snowy Range of southeastern Wyoming (369.6 km) and the Cimarron Range of north-central New Mexico (329.8 km). Four females were killed by puma hunters; 3 off the study area ranging from 24.0 to 74.5 km from initial capture sites; 1 on the study area 18.2 km from her initial capture site. Female F52 was treed and released by hunters in December 2008 and 2009 south of Powderhorn, Colorado, indicating that she established an adult home range there before she was killed by a puma hunter in that area on Jan. 9, 2012. Three males marked initially as cubs born on the study area (M67, M87, M92) dispersed from their natal ranges and were recaptured as adults on the study area. All were born on the east slope of the Uncompahgre Plateau and moved to the west slope. Twenty-three of the 33 pumas had reached adult ages ranging from 24 to 79 months old. A preliminary estimate of cub survival during the reference period was summarized in Logan 2009 using 36 radio-collared cubs (16 males, 20 females) marked at nurseries when they were 26 to 42 days old. In that summary, estimated survival of cubs to one year of age was 0.53. [The estimated minimum survival rate using the Kaplan-Meier procedure was 0.5285 (SE = 0.1623). The maximum estimated cub survival was practically the same, 0.5328 (SE = 0.1629).] The major natural cause of death in cubs, where cause could be determined, was infanticide and cannibalism by other, especially male, pumas. In TY3 we monitored the fates of 12 radio-collared cubs (Appendix A). We lost contact with one (M156) after he shed his expandable radio-collar; he was 59 days old. Of the remaining 11 collared cubs, 6 died. Cubs M154 and M155 died probably of starvation after their mother died of an unknown natural cause; they were 77-81 days old. M159 died of an unknown natural cause when he was about 105 days old. His siblings F157 and F158 died of starvation after their mother F70 was killed by a puma hunter; they were 150 days old. M162 died of starvation after his mother was killed by a puma hunter; he was about 10.6 months old. Three other cubs that were orphaned at older ages survived to the subadult life stage. F147 was orphaned at 12 months old when her mother F24 was killed by a male puma. F147 continued to range on her natal area until her radiocollar quit functioning when she was 19 months old. Siblings F149 and M161 were orphaned at 13.5 months old when their mother F23 was killed by a male puma. Both siblings dispersed to the east slope of the study area when they were 14 to 15 months old. Another cub, F152, offspring of F93, survived to at least 25 month old in July 2012 and ranged on her natal area. A greater number of cubs over a longer period of time must be sampled before estimating cub survival and agent-specific mortality rates in the treatment period. In addition, a non-marked male puma cub was struck and killed by a vehicle on state highway 62 in Leopard Creek on the south boundary of the study area on October 7, 2011. This mortality made the fourteenth puma death recorded due to vehicle collision on the study area since 2004 (Table 18). Five of the 14 pumas were marked, including 3 adults with GPS/VHF collars. Those 3 adults died during the first year of the treatment period. Thirty-five adult pumas (26 females, 9 males) have worn GPS collars since this project began in 2004 (Table 19). Over 60 thousand GPS locations have been obtained and will be used for studies on 151 �puma behavior, social organization, population dynamics, movements, habitat use and puma-human relations in collaboration with colleagues in Mammals Research and Colorado State University. Segment Objective 6 A pilot survey of prevalence of Trichinella spp. in puma from southwest Colorado was initiated by Mammals Researcher Ken Logan and Dr. Mason Reichard of Center for Veterinary Health Science, Oklahoma State University, Stillwater, OK. Summary: The current pilot study documented the occurrence and high prevalence of Trichinella spp. in Puma concolor from Colorado. Twelve of 14 (85.7%) puma tongues were infected with Trichinella. The high prevalence of the zoonotic nematode in Colorado pumas justifies expansion of the sampling area to include pumas from a broader geographical scale. Background: Trichinella spp. are zoonotic nematodes capable of infecting humans and other animals. Wild animals and humans throughout the world become infected when they ingest infected tissue containing the parasite. Infection in humans of Trichinella spp. may result in nausea, diarrhea, vomiting, fatigue, fever, abdominal discomfort, headaches, chills, cough, eye swelling, and may even lead to heart and breathing problems. In severe cases, infection of Trichinella spp. may result in death. Hunting of pumas in Colorado has substantial historical, cultural, recreational, and economic importance. However, little current research and literature (either public or peer-reviewed) is available regarding the prevalence of Trichinella in Colorado puma and the potential for human infection. In 1995, an outbreak of trichinellosis in 10 people from Idaho County, Idaho was reported from the consumption of improperly prepared cougar jerky (Vollbrecht et al. 1996). The outbreak of trichinellosis in Idaho stresses the importance of wild carnivores as reservoirs of Trichinella spp. infections to humans (Kennedy et al. 2009). In addition to Idaho, pumas infected with Trichinella spp. have been reported from Montana (Worley et al. 1974; Winters 1969), Oregon (Rausch et al. 1983), Wyoming (Worley et al. 1974), and British Columbia, Canada (Gajadhar and Forbes 2010). The purpose of the current pilot study was to determine if puma from southwest Colorado were infected with Trichinella spp. The specific objectives were to: 1. Determine the prevalence of Trichinella spp. in P. concolor from southwest Colorado. 2. Determine which species of Trichinella is/are present in P. concolor from southwest Colorado. 3. Establish baseline data on the occurrence, prevalence, and distribution of Trichinella spp. in southwest Colorado. Pilot Project Design: Tongues from hunter-killed pumas were artificially digested to detect Trichinella spp. larvae. Infection with Trichinella spp. was assessed according to sex, age class, and geographic location of capture. Collection of Tissue from Pumas Tongues from dead pumas were collected by Mammals Researcher Ken Logan from pumas that were killed by sport-hunters (n = 12) and for depredation control (n = 2) in GMUs 61, 62, 64, 65, 66, and 521 representing Delta, Gunnison, Montrose, and Ouray counties in southwest Colorado. Jaws of the cats were opened, tongue firmly grasped, and pulled out of the mouth. One-half to three-quarters of the puma’s tongue was cut from the carcass using a clean knife or sterile scalpel. Excised tongues were placed in zip-top bags, labelled with sex, age estimate, and unique identifiers according to the host puma and location of where and when the sample was collected. Tongues samples were then frozen (-20 C) until they were shipped to Oklahoma State University for analysis. 152 �Determination of Trichinella Infection Infection with Trichinella sp. was determined by tissue digestion of tongues from puma (Webster et al., 2006). Puma tongues were weighed and homogenized in a Polytron (Kinematica GmbH, Kriens-Luzern, Switzerland). Ground samples were mixed with 10 ml of artificial digestive fluid (1% pepsin [1:3,000 IU] and 1% hydrochloric acid) per 1 gram of tissue. Digests were then mixed vigorously on magnetic stir plats at 37° C for 3 hours. Digests were allowed to settle for 20 min and the sediment containing Trichinella larvae were washed with tap water and enumerated under 40x magnification. Results were recorded as the number of larvae recovered per gram of tongue tissue digested. Results: Fourteen pumas were tested for infection with Trichinella. Twelve of the 14 (85.7%) were infected with Trichinella (Table 20). Because the prevalence of Trichinella infection was high and the sample size was relatively small, additional statistical comparisons of sex and age classes were not made as they were unlikely to yield useful information. Based on previously published data on the prevalence of Trichinella spp. in pumas from other locations (Winters 1969; Worley et al. 1974; Vollbrecht et al. 1996), we anticipated that approximately 50% of the pumas from southwest Colorado would have been infected. However, the prevalence of Trichinella in pumas was much higher than originally thought. The common occurrence of the zoonotic parasites in pumas from southwest Colorado coupled with the fact that consumption of improperly prepared meat from wild felids can infect humans (Vollbrecht et al. 1996) necessitates continued sampling from a broader geographical area in Colorado to determine infection risk to humans. Project Continuation: Trichinella larvae recovered from these twelve pumas will be submitted to the International Trichinella Reference Center (ITRC, www.iss.it/site/Trichinella/) in Rome, Italy for genotyping to identify the species of Trichinella. Individual Trichinella larvae will be identified by a multiplex PCR analysis following the protocol described by Zarlenga et al. (1999) and modified by Pozio and La Rosa (2003). Briefly, DNA will be extracted from individual worms and PCR will be performed using ExTaq DNA polymerase (Takara) in 50 ml containing 1.5 mM MgCl2, 200 mM dNTPs, 50 pmol of each primer and 0.5 unit of ExTaq DNA polymerase. The PCR-amplified fragments from purified DNA will be visualised by agarose gel electrophoresis (2.0% standard agarose). When Trichinella larvae were counted from infected pumas, we noticed that the majority of the worms were still alive after being frozen for at least 6 months or longer. The trait of freeze resistance suggests the Trichinella in pumas from southwestern Colorado are either T. nativa or Trichinella genotype T6 (i.e., the two freeze resistant species in North America). However, T. murrelli, not freeze resistant, is the species of Trichinella most commonly recovered from wild animals in temperate areas across North America. Fifteen additional puma tongues from southwest Colorado were collected to accumulate a larger sample size during the 2011 to 2012 puma hunting season. Those tissues will be analyzed for prevalence of Trichinella by Dr. Mason’s laboratory in 2013. SUMMARY Manipulative, long-term research on puma population dynamics, effects of sport-hunting, and development and testing of puma enumeration methods began in December 2004. After 7.7 years of effort 168 unique pumas have been captured, sampled, marked, and released. Using these animals, we monitored fates of pumas in all sexes and age stages, including: 34 adult females, 21 adult males, 11 subadult females, 16 subadult males, 45 female cubs, 71 male cubs, and 1 cub of undetermined sex (some individuals occur in more than one stage class). Data from marked animals were used to quantify puma population characteristics and vital rates in a reference period without sport-hunting off-take as a mortality factor from December 2004 to July 2009. Puma population characteristics and vital rates in a 153 �reference condition allowed us to develop a puma population model, and to use population data and modeling scenarios to conduct a preliminary assessment of CPW puma management assumptions and guide directions for the remainder of the puma research on the Uncompahgre Plateau. Moreover, our data and model provide tools currently useful to CPW wildlife biologists and managers for assessing puma harvest strategies. The 5-year treatment period began August 2009 in which sport-hunting is a mortality factor. The treatment period will be a population-wide test of CPW puma management assumptions. Now 3 years of the treatment period are complete (TY1, TY2, TY3). Although data support some CPW puma management assumptions (e.g., population structure, density, how sport-harvest can cause population decline), it is still too early in this research to adequately test all the assumptions and attendant hypotheses. Although the assumption and hypothesis on harvest structure and hunter selection is not supported with the first 3 years of data in the treatment period, this could change with a substantial change in abundance and sex structure of independent pumas available for hunting in TY4 and TY5. The puma harvest quota for TY4 is recommended to be 5 independent pumas to align with the research design and harvest objective, and the hunters will be surveyed again. To improve data on puma population vital rates, attention will be given to increasing radio-collared sample sizes across the various life stages and sexes. We will continue to explore methods for estimating puma abundance with accurate and affordable methods. Furthermore, we will continue collaboration with colleagues on investigations of puma population parameter estimation, puma movements, puma habitat modeling and mapping, puma-human relations, and Trichinella prevalance. All of these efforts should enhance the Colorado puma research and management programs. 154 �LITERATURE CITED Anderson, A. E, D. C. Bowden, and D. M. Kattner. 1992. The puma on Uncompahgre Plateau, Colorado. Technical Publication No. 40. Colorado Division of Wildlife, Denver. Anderson, C. R., Jr., and F. G. Lindzey. 2005. Experimental evaluation of population trend and harvest composition in a Wyoming cougar population. Wildlife Society Bulletin 33:179-188. Barnhurst, D. 1986. Vulnerability of cougars to hunting. Master’s Thesis. Utah State University. Colorado Division of Wildlife 2002-2007 Strategic Plan. 2002. Colorado Department of Natural Resources, Division of Wildlife. Denver. Colorado Division of Wildlife. 2007. Colorado mountain lion management data analysis unit revision and quota development process. Colorado Division of Wildlife, Denver. Conover, W. J. 1999. Practical nonparametric statistics. John Wiley & Sons, Inc., New York. Cooch, E., and G. White. 2004. Program MARK- a gentle introduction, 3rd edition. Colorado State University, Fort Collins. Culver, M., W. E. Johnson, J. Pecon-Slattery, and S. J. O’Brien. 2000. Genomic ancestry of the American puma (Puma concolor). The Journal of Heredity 91:186-197. Currier, M. J. P., and K. R. Russell. 1977. Mountain lion population and harvest near Canon City, Colorado, 1974-1977. Colorado Division of Wildlife Special Report No. 42. Gajadhar, A. A., and L. B. Forbes. 2010. A 10-year wildlife survey of 15 species of Canadian carnivores identifies new hosts or geographical locations for Trichinella genotypes T2, T4, T5, and T6. Veterinary Parasitology 168: 78-83. Heisey, D. M., and T. K. Fuller. 1985. Evaluation of survival and cause specific mortality rates using telemetry data. Journal of Wildlife Management 49:668-674. Karanth, K. U., and J. D. Nichols. 2002. Monitoring tigers and their prey: A manual for researchers, managers and conservationists in tropical Asia. Centre for Wildlife Studies, Bangalore, India. Kennedy, E. D., R. L. Hall, S. P. Montgomery, D. G. Pyburn, and J. L. Jones. 2009. Trichinellosis surveillance – United States, 2002-2007. Morbidity and Mortality Weekly Report 58: 1-7. Koloski, J. H. 2002. Mountain lion ecology and management on the Southern Ute Indian Reservation. M. S. Thesis. Department of Zoology and Physiology, University of Wyoming, Laramie. Kreeger, T. J. 1996. Handbook of wildlife chemical immobilization. Wildlife Pharmaceuticals, Inc., Fort Collins, Colorado. Laundre, J. W., L. Hernandez, D. Streubel, K. Altendorf, and C. L. Lopez Gonzalez. 2000. Aging mountain lions using gum-line recession. Wildlife Society Bulletin 28:963-966. Logan, K. A., E. T. Thorne, L. L. Irwin, and R. Skinner. 1986. Immobilizing wild mountain lions (Felis concolor) with ketamine hydrochloride and xylazine hydrochloride. Journal of Wildlife Diseases. 22:97-103. _____, and L. L. Sweanor. 2001. Desert puma: evolutionary ecology and conservation of an enduring carnivore. Island Press, Washington, D.C. _____. 2008. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Wildlife, Fort Collins. _____. 2009. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Wildlife, Fort Collins. _____. 2010. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Wildlife, Fort Collins. _____. 2011. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Parks and Wildlife, Fort Collins. Murphy, K., M. Culver, M. Menotti-Raymond, V. David, M. G. Hornocker, and S. J. O’Brien. 1998. Cougar reproductive success in the Northern Yellowstone Ecosystem. Pages 78-112 in The ecology of the cougar (Puma concolor) in the Northern Yellowstone ecosystem: interactions with prey, bears, and humans. Dissertation, University of Idaho, Moscow. 155 �Ott, R. L. 1993. An introduction to statistical methods and data analysis. Fourth edition. Wadsworth Publishing Co., Belmont, California. Pierce, B. K., V. C. Bleich, and R. T. Bowyer. 2000. Social organization of mountain lions: does land a tenure system regulate population size? Ecology 81:1533-1543. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Pojar, T. M., and D. C. Bowden. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550-560. Pozio, E., and G. La Rosa. 2003. PCR-derived methods for the identification of Trichinella parasites from animal and human samples. Methods in Molecular and Cellular Biology 216: 299-309. Rausch, R. L., C. Maser, and E. P. Hoberg. 1983. Gastrointestinal helminths of the cougar, Felis concolor L. in northeastern Oregon. Journal of Wildlife Diseases 19: 14-19. Ross, P. I., and M. G. Jalkotzy. 1992. Characteristics of a hunted population of cougars in southwestern Alberta. Journal of Wildlife Management 56:417-426. Seidensticker, J. C., M. G. Hornocker, W. V. Wiles, and J. P. Messick. 1973. Mountain lion social organization in the Idaho Primitive Area. Wildlife Monographs No. 35. Stoner, D. C. 2004. Cougar exploitation levels in Utah: implications for demographic structure, metapopulation dynamics, and population recovery. Master of Science Thesis. Utah State University. Sweanor, L. L., K. A. Logan, J. W. Bauer, B. Milsap, and W. M. Boyce. 2008. Puma and human spatial and temporal use of a popular California state park. Journal of Wildlife Management 72:10761084. Vollbrecht, A., D. Sokolowski, W. Hollipeter, R. Sigler, J. Greenblatt, D. E. Anderson, P. J. Tennican, H. R. Gamble, and D. Zarlenga. 1996. Oubreak of trichinellosis associated with eating cougar jerky – Idaho, 1995. Morbidity and Mortality Weekly Report 45: 205-206. Webster, P., C. Maddox-Hyttel, K. Nockler, A. Malajauskas, J. van der Giessen, E. Pozio, P. Boireau, C. M. O. Kapel. 2006. Meat inspection for Trichinella in pork, horsemeat and game within the EU: available technology and its present implementation. Eurosurveillance 11:50-55. Williams, B. K., J. D. Nichols, and M. J. Conroy. 2001. Combining closed and open mark-recapture models: the robust design. Pages 523-554 In Analysis and management of animal populations. Academic Press, New York. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 (Suppl):S120-S139. Winters, J. B. 1969. Trichiniasis in Montana mountain lions. Bulletin of the Wildlife Disease Association 5: 400. Worley, D. E., J. C. Fox, J. B. Winters, and K. R. Greer. 1974. Prevalence and distribution of Trichinella spiralis in carnivorous mammals in the United States northern Rocky Mountain region. Proceedings of the 3rd International Conference on Trichinellosis, Florida, USA, November 2-4, 1972. Zarlenga, D.S., Chute, M.B., Martin, A. and Kapel, C.M. 1999. A multiplex PCR for unequivocal differentiation of all encapsulated and nonencapsulated genotypes of Trichinella. International Journal for Parasitology 29:1859–1867. Prepared by: ________________________ Kenneth A. Logan, Wildlife Researcher 156 �Table 1. Projected puma population growth modeled from a minimum count of independent pumas during winter 2007-08 reference period year 4 (RY4). Treatment period year 1 (TY1), shaded in gray, indicates the results used to derive a quota of 8 independent pumas, representing 15% of the independent pumas (from Logan 2009). Harvest Level No harvest. Year RY4 RY5 TY1 Projected Minimum Puma Population Size Adult Subadult Female Male Female Male 16 8 5 4 18 10 9 8 23 14 8 8 Independent Pumas Cub 20 33 42 Total 33 45 53 Lambda 1.37 1.17 Table 2. Pumas harvested by sport-hunters in Treatment Year 3 (TY3) on the Uncompahgre Plateau Study Area, Colorado, November 21 to December 23, 2011. Puma sex Age (yr.) Date of kill Location/UTM NAD27 Zone, Easting, Northing 1.6 Previous M/F I.D. or specimen P no. if not marked P1038 M 12/5/2011 M 1.4 M120 12/6/2011 F 9.5 F3 12/11/2011 F 8 F75 12/14/2011 F 7 F70 12/22/2011 M 1.1 P1049 12/23/2011 M 2.5 M138 12/23/2011 M 1.3 M141 12/23/2011 Cottonwood Fork of Dry Creek 12, 756280, 4250547 Spring Creek 13, 238681, 4249866 Lindsay Creek 13, 238911, 4252542 Cottonwood Creek (W) 12, 732894, 4239423 Spring Creek, Puma Fork 13, 239323, 4243719 Hills west of Colona, CO 13, 256132, 4245751 Horsefly Creek (E) 13, 249592, 4240770 Little Bucktail Creek 12, 752201, 4239371 157 Hunter/status Ray David/resident Gary Gleason/resident Kari McClanahan/ resident Joe Gray/non-resident Dustin Gleason/resident Dawson Flowers/ resident Darren Reed/resident Kenneth Sowell/ resident �Table 3. Four other independent VHF/GPS-collared and 3 non-collared adult pumas in the minimum count for the Uncompahgre Plateau Study Area that died during the 2011-2012 Colorado puma hunting season. Puma sex (M or F) M87 Age (yr.) 3.4 Date of kill/death 12/6/2011 M67 4.4 12/18/2011 F119 7.5 1/28/2012 F104 11 1/31/2012 M 5.5 12/6/2011 M 3 1/17/2012 F 6 1/18/2012 Place of kill/UTM NAD27 Zone, Easting, Northing Forty-seven Canyon, Tabaguache Canyon 6.5 km N of study area Lower Tabaguache Canyon 12.61 km NW of study area 12, 707031, 4247827 Clay Creek 12, 743719, 4228535 Lower Roubideau Creek, died 1.73 km N of study area 12, 748282, 4288223 Cottonwood Creek north of Roubideau Canyon 0.66 km N of study area 12, 736764, 4274349 Specie Creek 1.65 km S of study area 12, 752861, 4211534 Pinto Mesa 1.02 km N of study area 12, 721658, 4247479 Hunter/status/other cause John Elmer/rresident/S. Garvey Outfitter Karl Red/resident Ruptured uterus and blood loss associated with pregnancy Starvation, probably associated with senescence Brett Merritt/non-resident Trailed from study area/R. Navarrete Outfitter Alan Hatfield/resident Trailed from study area James Williams/nonresident/S. Garvey Outfitter Radio-collared cub M162 ranged on study area Table 4. Minimum count of pumas based on numbers of known radio-collared pumas, visual observations of non-marked pumas, harvested non-marked pumas, and track counts of suspected non-marked pumas on the study area during September 2009 to April 2010 of Treatment Year 1 (TY1), November 2010 to April 2011 (TY2), and November 2011 to April 2012 (TY3) Uncompahgre Plateau study area, Colorado. Treatment Year (TY) Study Area region TY1 East slope West slope subtotals TY2 TY3 Adults Female Male Subadults Female Male Female Cubs Male 16 10 1 1 1 4 14 10 0 3 3 3 30 20 1 4 4 7 Total Independent Pumas = 55, including 31 females, 24 males. Cubs = 20-25 East slope 15 5 3 2 7 9 West slope 15 7 2 3 2 5 subtotals 30 12 5 5 9 14 Total Independent Pumas = 52, including 35 females, 17 males. Cubs = 39 East slope 13 4 1 3 4 2 West slope 14 5 3 5 1 2 subtotals 27 9 4 8 5 4 Total Independent Pumas = 48, including 31 females, 17 males. Cubs = 19 Unknown sex 4-8* 5-6 9-14 7 9 16 4 6 10 *One adult non-marked female puma was killed by a hunter in Roubideau Canyon. The female puma was lactating, indicating she had nurslings. Up to 4 cubs were assumed to be in the litter. 158 �Table 5. Pumas captured and released by sport-hunters in Treatment Year 3 (TY3) on the Uncompahgre Plateau Study Area, Colorado, November 21 to December 23, 2011. Data are from puma hunter responses in 36 voluntary surveys, including: 31 original surveys on printed voluntary permits and 5 telephone contacts with hunters that did not return printed surveys on permits. Total response rate from 74 individual permitted hunters was 48.6 % (36/74 = 0.486*100). Puma sex/age stage/mark F/adult/collar/eartags Date of capture 12/4/2011 M/young/no marks 11/30/2011 to 12/5/2011 12/14 to 16/2011 F/adult/F75/ collar/eartags/caught twice M/young/no marks F/no marks 11/30/2011 to 12/5/2011 12/20 to 23/2011 Capture location Hunter name Transfer Rd., Roubideau Cyn. Dry Park. George Quintana Cottonwood Cr. Thomas Barnes Dry Park. Ross Ward Loghill Mesa Zachary Prock Eric Franklin 159 Reason for releasing the puma given by hunter Did not want to kill a female. Observed the puma on the road. Did not want to kill a small male. Caught F75 twice. Did not want to kill a puma. Wanted to take photos. Did not want to kill a small male. Same male caught with Eric Franklin (above). Did not want to kill a female. �Table 6. Summary of puma capture efforts with dogs from December 27, 2011 to April 12, 2012, Uncompahgre Plateau, Colorado. Month December No. Search Days 4 January 25 February 20 March 22 No. & type of puma tracks founda,b 16 tracks: 4 male, 10 female, 1 cub, 1 undetermined independent puma Tracks ≤1 day old: 2 male, 8 female, 0 cub 103 tracks: 12 male, 56 female, 31 cub, 4 undetermined independent pumas Tracks ≤1 day old: 4 male, 31 female, 23 cub, 1 undetermined 71 tracks: 9 male, 44 female, 17 cub, 1 undetermined independent puma Tracks ≤1 day old: 7 male, 26 female, 9 cub, 1 undetermined No. & type of pumas pursued 4 pursuits: 2 male, 2 female, 0 cub No. & I.D. or type of pumas captured, observed, or identified 2 pumas captured 2 times: adult female F137 and dependent young F152 (of F93). In addition, adult females F93 and F111 were associated with tracks by VHF telemetry. 39 pursuits: 3 male, 19 female, 16 cub, 1 undetermined 12 pumas captured 14 times: F93, F96, M170 (cub of F171), F171, cub (not handled, of F171), F8, F140, F149 (cub of F23), M160, M161 twice (cub of F23), F163 twice, M162 (orphaned cub). In addition, adult females F23 (3 times), F93 (2 times), F136, F137, F149 (2 times), F152 (2 times), M170, and F171 were located by VHF telemetry in association with tracks. 9 pumas captured 9 times: F28 (not handled in hole), F129 (not handled, dangerous tree), M131 (not handled, dangerous tree), F163, M164, M165, PF1051 (biodarted, not handled in dangerous tree), PF1052 (biodarted, not handled in dangerous tree), cub (not handled in dangerous tree, of non-marked female). In addition, adult females F23 (2 times), F93, F95, F96 (2 times), F136, F137, F171, adult male M164, subadult females F147 and F163 (3 times), and cubs F152 and M162 (2 times) were associated with tracks by VHF telemetry. 2 pumas captured 2 times: F149, M161. In addition, adult females F23, F95, F96, F171, subadult females F140 and F149, cubs M161 and M170 were associated with tracks by VHF telemetry. 27 pursuits: 6 male, 13 female, 8 cub 66 tracks: 12 male, 18 pursuits: 2 male, 39 female, 14 cub, 1 12 female, 4 cub undetermined independent puma Tracks ≤1 day old: 2 male, 16 female, 4 cub April 8 12 tracks: 6 male, 1 pursuit: 0 pumas captured. None associated with tracks 5 female, 1 cub 1 male with VHF telemetry. Tracks ≤1 day old: 1 male, 2 female, 1 cub 79 268 tracks: 89 pursuits: 21 individual pumas were captured 26 times with TOTALS 43 male, 14 male, aid of dogs. In addition, 17 radio-collared pumas 154 female, 46 female, were detected 40 times by tracks and identified 64 cub, 28 cub with VHF telemetry ≤1 km from the tracks. 7 undetermined 1 undetermined Tracks ≤1 day old: 16 male 83 female 37 cub 2 undetermined a Puma hind-foot tracks with plantar pad widths >50 mm wide are assumed to be male; ≤50 mm are assumed to be female (Logan and Sweanor 2001:399-412). b Each capture season researchers also recorded instances when the first puma tracks ≤1 day old were encountered on each search route each day to gather data on vulnerability to detection using methods similar to puma hunters. For 2011-2012 (TY3) the count was: 70 tracks of females, including 17 of those associated with cubs; 2 tracks of 2 orphaned cubs; 12 tracks of males; and 2 tracks of undetermined sex. 160 �Table 7. Adult and subadult pumas captured for the first time, sampled, tagged, and released from December 2011 to April 2012, Uncompahgre Plateau, Colorado. Puma I.D. M160 F163 M164 M165 F171 F172 Sex Estimated Age (mo.) 19 18 19 19 27 33 M F M M F F Mass (kg) 46 43 56 56 45 NM* Capture date 1/18/2012 1/26/2012 2/14/2012 2/24/2012 1/20/2012 3/28/2012 Capture method Dogs Dogs Dogs Dogs Dogs Cage trap Location Sanborn Park, head of Albin Draw San Miguel Canyon E of Pinyon Pinto Mesa, moved from Big Bucktail Canyon Head of Coal Canyon McKenzie Butte Monitor Canyon, Roubideau Canyon *Not measured. Table 8. Pumas that were captured and observed with aid of dogs, some of which were biopsy-darted and given specimen numbers (e.g., P1051), but were not handled at that time for safety reasons, and a puma killed for depredation control, December 2011 to April 2012, Uncompahgre Plateau, Colorado. Puma sex & I.D. F P1035 F P1051 Age stage or months 18 Capture date 10/22/2011 adult 2/13/2012 F P1052 Unknown none Unknown none adult 2/29/2012 cub 4 to 5 cub 8 to 10 1/12/2012 2/29/2012 Location Comments Dallas Creek, Pleasant Valley. Potter Canyon, Roubideau Canyon. Puma killed by Wildlife Services agent for killing a domestic llama. Puma not previously marked. Puma climbed dangerous tree. Biopsy-darted to obtain tissue sample for genotype. In association with a single cub about 8 to 10 months old, which also could not be handled due to dangerous tree. Puma climbed dangerous tree Biopsy-darted to obtain tissue sample for genotype. Puma cub climbed high in dangerous tree. Probably 1 of 2 cubs of F171. Puma cub climbed high in dangerous tree. Not handled. In association with P1051 above. Monitor Canyon, Roubideau Canyon. E Loghill Mesa. Monitor Canyon, Roubideau Canyon. Table 9. Pumas recaptured with dogs (none in cage traps) December 2011 to April 2012, Uncompahgre Plateau, Colorado. Puma I.D. F152 Recapture Date 12/27/2011 Mass (kg) Observed Estimated Age (mo.) 18 Capture Method/ Location Dogs/Dry Cr. Basin F137 F96 F140 F93 12/28/2011 1/9/2012 1/13/2012 1/17/2012 Observed 44 47 Observed 35 72 17 109 F8 F152 F149 M161 F163 F129 1/17/2012 1/18/2012 1/24/2012 1/24/2012 1/27/2012 2/2/2012 Observed 43 Observed Observed Observed Observed 104 19 9 9 18 18 Dogs/W Fk. Dry Cr. Dogs/lower Delores Cr. Dogs/Tomcat Cr. Dogs/Lower Linscott Cyn. Dogs/Coal Canyon Dogs/Shavano Mesa Dogs/Big Bucktail Cyn. Dogs/Big Bucktail Cyn. Dogs/Maverick Draw Dogs/Dolores Cyn. M131 2/2/2012 Observed 18 Dogs/Dolores Cyn. F163 F28 2/9/2012 2/16/2012 Observed Observed 19 107 Dogs/San Miguel Cyn. Dogs/San Miguel Cyn. F149 M161 3/5/2012 3/5/2012 29 Observed 11 11 Dogs/Tomcat Cr. Dogs/Tomcat Cr. 161 Process F152 climbed dangerous tree. Could not be handled to fit with radiocollar. None. Replaced faulty GPS collar. Fitted with VHF collar. None. None. Fitted with GPS collar. None. None. None. F129 climbed dangerous tree. Could not be handled to fit with radiocollar. M131 climbed dangerous tree. Could not be handled to fit with radiocollar. None. F28 took refuge in hole. Could not be handled. Replaced VHF collar on F149. None. �Table 10. Summary of puma capture efforts with cage traps from October 5, 2011 to April 11, 2012, Uncompahgre Plateau, Colorado.* Month October November No. of Sites 6 7 March 2 April 5 Carnivore activity & capture effort results No pumas scavenged 7 mule deer carcasses used at the sites. Unknown male puma scavenged deer carcass on SE Loghill Rim 11/9/2011; attempted capture with cage trap; male puma did not return. Unknown male puma walked ~100 m from deer carcass at same bait site above 11/22/2011, but did not scavenge the bait. Unknown male puma walked ~1.5 m from a deer carcass at same bait site as above 11/29/2012, but did not scavenge the bait. Puma F172 captured in cage trap baited with mule deer carcass 3/28/2012. Unknown female puma and lone cub scavenged mule deer carcass at another bait site 4/68/2012. The pumas were pursued with dogs on 4/9/2012, but were not captured. No pumas scavenged 5 mule deer carcasses used at the sites. * We used 21 road-killed mule deer at 18 different sites. Of the road-killed deer baits, 3 of 21 (14.29%) were scavenged by pumas. Table 11. Puma cubs sampled August 2011 to July 2012 on the Uncompahgre Plateau Puma Study area, Colorado. Cub I.D. Sex Estimated birth datea Estimated age at capture (days) Mass (kg) Mother Estimated age of mother at birth of this litter (mo) M154 M155 M156b F157 F158 M159 M161c M162d M170e M M M F F M M M M 7/6/2011 7/6/2011 8/20/2011 8/18/2011 8/18/2011 8/18/2011 4/22/2011 7/2011 8/2011 42 42 43 40 40 40 276 183 137 2.6 3.0 3.25 2.5 2.5 2.5 24 12 9 F135 F135 F137 F70 F70 F70 F23 Nonmarked F171 33 33 30 76 76 76 80 Adult 22 a Estimated age of cubs sampled at nurseries is based on the starting date for GPS location and radio-telemetry foci for mothers at nurseries, and development characteristics of cubs caught with mothers without radiocollars or mothers with non-functioning radiocollars. b Probably more than one cub in F137’s litter; others probably hiding in a hole at the nursery. c M161 is sibling of F149; birth date known from radio-telemetry on mother F23. d M162 was observed with one non-marked sibling on 2/7/2012. Both cubs were orphans; their mother non-marked mother apparently killed by a hunter on 1/18/ 2012 on Pinto Mesa. e M170 was observed with one sibling on 1/12,13/2012. Mother F171 was captured for first time on 1/20/2012. 162 �Table 12. Summary of puma capture efforts with dogs, December 2004 to April 2012, Uncompahgre Plateau, Colorado. Period Dec. 2, 2004 to May 12, 2005 Nov. 21, 2005 to May 26, 2006 Nov. 13, 2006 to May 11, 2007 Nov. 19, 2007 to April 24, 2008 Dec. 9, 2008 to April 30, 2009 Track detection effort 109/78 = 1.40 tracks/day 35/78 = 0.45 pursuit/day Puma capture effort 14/78 = 0.18 capture/day Effort to capture an independent puma for the first time 11 pumas captured for first time 11/78 = 0.14 capture/day 78/14 = 5.57 day/capture 14/82 = 0.17 capture/day 78/11 = 7.09 day/capture 149/82 = 1.82 tracks/day 78/35 = 2.23 day/pursuit 43/82 = 0.52 pursuit/day 177/78 to 182/78 = 2.27-2.33 tracks/day 82/43 = 1.91 day/pursuit 45/78 to 47/78 = 0.58-0.60 pursuit/day 82/14 = 5.86 day/capture 22/78 = 0.28 capture/day 78/47 to 78/45 = 1.66-1.73 day/pursuit 49/77 = 0.64 pursuit/day 78/22 = 3.54 day/capture 78/7 = 11.14 day/capture 20/77 = 0.26 capture/day 7 pumas captured for first time 7/77 = 0.09 capture/day 77/20 = 3.85 day/capture 24/71 = 0.34 capture/day 77/7 = 11.00 day/capture 217/77 to 218/77 = 2.82-2.83 tracks/day Pursuit effort 198/71 to 202/71 = 2.79-2.84 tracks/day 77/49 = 1.57 day/pursuit 75/71 to 78/71 = 1.06-1.10 pursuit/day Dec. 15, 2009 to April 30, 2010 Nov. 16 and Dec. 14, 2010 to April 22, 2011 266/86 = 3.09 tracks/day 71/75 to 71/78 = 0.91-0.95 day/pursuit 93/86 = 1.08 pursuit/day 300/81 = 3.70 tracks/day Dec. 27, 2011 To April 12, 2012 268/79 = 3.39 tracks/day 7 pumas captured for first time 7/82 = 0.08 capture/day 82/7 = 11.71 day/capture 7 pumas captured for first time 7/78 = 0.09 capture/day 9 pumas captured for first time 9/71 = 0.13 capture/day 71/24 = 2.96 day/capture 71/9 = 7.89 day/capture 26/86 = 0.30 capture/day 9 pumas captured for first time 9/86 = 0.11 capture/day 86/93 = 0.92 day/pursuit 99/81 = 1.22 pursuit/day 86/26 = 3.31 day/capture 52/81 = 0.64 capture/day 86/9 = 9.56 day/capture 81/99 = 0.82 day/pursuit 81/52 = 1.56 day/capture 81/15 = 5.40 day/capture 89/79 = 1.13 pursuit/day 26/79 = 0.28 capture/day 11 pumas captured for first time 11/79 = 0.14 capture/day 79/89 = 0.89 day/pursuit 79/26 = 3.04 day/capture 79/11 = 7.18 day/capture 163 15 pumas captured for first time 15/81 = 0.18 capture/day �Table 13. Individual puma reproduction histories, Uncompahgre Plateau, Colorado, 2005-2012. Consort pairs and estimated agesa Female Age (mo.) Male Age (mo.) F2 F2 F2 F3 F3 F3 F3 F3 F7 F7 F7 F8*e F8 F8 F8 F16 F16 F16 F23* F23 53 67 89 36 50 62 84 107 67 82 106 24 37 60 95 32 52 75 21 45 F23 80 F24 F24 75 114 F25 F25 F25 F25 74 94 110 129 F28* F28 F28 F30* F50 F54 F70* F70 F70 F72* F72 F72 36 48 68 48 21 24 38 52 76 28 51 64 F75 F75 F93 F93 F94* F94 F96 F104 F111* F116g 32 55 56 90 46 60 55 110 32 36 Dates pairs consortedb Estimated birth datec M73 49 02/28-29/08 M6 80 01/13-14/09 M27 or M29f M67 78 107 53 02/19-25/08 05/28/05 07/29/06 05/19/08 08/01/04 09/26/05 09/17/06 07/03/08 06/28/10 05/19/05 08/13/06 07/10/08 06/26/05 08/13/06 05/29/08 04/18/11 09/22/05 05/24/07 04/15/09 05/30/06 05/23/08 01/28-31/11 04/22/11 M29 92 04/12-15/07 06/14/07 09/10 M6 37 06/22-24/05 M51 M55 60 69 03/31/08 03/28-31/10 08/01/05 04/16/07 08/19/08 3/10 M29 88 12/27-29/06 M55 34 04/16-20/07 M51 60 03/10/08 M73 61 02/11/09 M55 M55 70 71 04/15/10 05/21/10 164 06/09/06 03/30/07 11/08 07/17/07 07/01/06 07/01/06 06/05/08 08/31/09 08/18/11 07/09/08 06/12/10 07/15/11 06/01/07 05/07/09 08/07 06/16/10 05/27/09 07/15/10 08/21/10 07/08/10 06/16/10 2009 Estimated birth interval (mo.) Estimated gestation (days) Observed number of cubsd 19.9 22.7 91-92 23.8 87-93 3 2 4 1 2 3 3 2 2 4 3 2 4 2 2 4 4 3 3 3 Nonfunct.GPS 84-86 2 90-93 4 3 14.0 22.0 13.8 11.7 21.5 23.8 93-95 94 89-92 14.9 23.9 13.4 22.5 34.7 90-91 Nonfunct.GPS 1 1 2 3 20.5 16.1 Nonfunct.GPS 11.7 92-93 88-92 87 14.8 23.6 23.1 13 23.2 93 13.3 91 2 ≥2 tracks 1 3 1 1 3 3 3 1 2 3 photographed 1 2 2 2 3 3 4 3 2 2 �Consort pairs and estimated agesa Female Age (mo.) Male Age (mo.) Table 13 continued. Dates pairs Estimated consortedb birth datec Estimated birth interval (mo.) 23 Estimated gestation (days) Observed number of cubsd ≥1 observed F119 66 08/09 2 F119i 96 02/12 29 1 plus 1-2 expected expected expected uterine scars F135 33 07/06/11 2 F136j 39 07/10/11 ≥1 remains F136 51 07/05/12 12 2 F137 30 07/08/11 ≥1 F171 22 08/11 2 a Ages of females were estimated at litter birth dates. Ages of males were estimated around the dates the pairs consorted. b Consort pairs indicate pumas that were observed together based on GPS data or VHF location data. c Estimated birth dates were indicated by GPS data of mothers at nurseries or by back-aging cubs to approximate birth date. d Observed number of cubs do not represent litter sizes as some cubs were observed when they were 5 to 16 months old after postnatal mortality could have occurred in siblings. Only cub tracks were observed with F28. e Asterisk (*) indicates first probable litter of the female, based on nipple characteristics noted at first capture of the female. f A radio-collared, ear-tagged male puma was visually observed with F23 on 2/25/08. Both M27 and M29 wore non-functional GPS collars in that area at the time. g When captured on 1/20/10, puma F116 was in association with 2 large cubs which were not captured. h One cub observed with F118 in Maverick Draw 7/19/2012. i F119 died of a ruptured uterus and internal bleeding on 1/28/12. Cub in uterus in third trimester; 1-2 uterine scars indicated expulsion of 1-2 fetuses. j Remains of F136’s cubs found 8/9/11. Cause of death predation by puma or black bear. F118h 50 06/20/2012 165 �Table 14. Summary for individual adult puma survival and mortality, December 8, 2004 to July 31, 2012, Uncompahgre Plateau, Colorado. Puma I.D. M1 Monitoring span 12-08-04 to 08-16-06 M4 M5 01-28-05 to 12-28-05 08-01-06 to 02-20-09 M6 02-18-05 to 05-21-10 M27 03-10-06 to 05-07-09 M29 04-14-06 to 02-25-09 M32 04-26-06 to 12-02-10 M51 01-07-07 to 03-20-09 M55 01-21-07 to 07-31-10 M67 08-23-07 to 12-18-11 M71 01-29-08 to 11-12-09 M73 02-21-08 to 10-26-11 M87 02-09-11 to 12-06-11 M90 11-16-10 to 11-23-10 M100 03-27-09 to 07-31-09 M114 02-27-10 to 03-10-12 M133 11-12-10 to 12-01-10 Status: Alive/Lost contact/Dead; Cause of death Dead. Lost contact− failed GPS/VHF collar. M1 ranged principally north of the study area as far as Unaweep Canyon. M1 was killed by a puma hunter on 01-02-10 west of Bang’s Canyon, north of Unaweep Canyon, GMU 40. M1 was about 97 months old at death. Dead; killed by a male puma. Estimated age at death 37−45 months. Dead. Born on study area; offspring of F3. M5 was independent of F3 by 13 months old, and dispersed from his natal area at about 14 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months (protected from hunting mortality in buffer area) and ranged into the eastern edge of Utah (vulnerable to hunting). Killed by a puma hunter on 02-20-09 in Beaver Creek, Utah at age 54 months. Dead. M6 was struck and killed by a vehicle on highway 550 south of Colona, CO on 05-21-10. M6 was about 99 months old at death. Dead. Lost contact− failed GPS/VHF collar. Recaptured 12-02-07 & 01-22-08 by puma hunter/outfitter north of the study area. Possibly visually observed on study area with F23 on 02-25-08. Recaptured by a puma hunter/outfitter 12-11-08 & 12-28-08 north of the study area. Photographed by a trail camera on the study area (Big Bucktail Canyon) on 5 occasions: 03-27-09, 04-02-09, 04-15-09, 04-24-09, & 05-07-09. M27 was killed by a puma hunter on 12-09-09 in the North Fork Mesa Creek, Uncompahgre Plateau, GMU 61 North. M27 was about 100 months old at death. Dead. Lost contact− failed GPS/VHF collar. Possibly visually observed on study area with F23 on 02-25-08. Recaptured on study area 02-25-09, but could not be safely handled to change faulty GPS collar. M29 was killed by a puma hunter on 11-16-09 in Beaver Canyon, GMU 70 East. M29 was about 121 months old at death. Dead. Killed by a puma hunter on 12-02-10 in McKenzie Creek on the Uncompahgre Plateau study area. M32 was about 112 months old at death. Dead. Lost contact− failed GPS/VHF collar after 03-20-09. Killed by a puma hunter on 12-11-09 in Shavano Valley, Uncompahgre Plateau study area. M51 was about 77 months old at death. Dead. Killed by a puma hunter on 11-25-10 in Spring Creek Canyon on the Uncompahgre Plateau study area. M55 was about 77 months old at death. Dead. M67 is offspring of F30. Dispersed natal area. Established territory on W side U.P. study area. Killed by a puma hunter in Tabaguache Creek 12-18-2011 at age 52.9 months. Dead. Lost contact– M71 shed his VHF collar with an expansion link on about 11-1209. He was killed by a puma hunter on 12-09-09 on the west rim of Spring Creek Canyon, Uncompahgre Plateau study area. M71 was about 47 months old at death. Dead. Illegally killed 10-26-2011 in Bear Pen Gulch, upper East Fork Escalante Canyon; shot through abdomen during second rifle season. M73 was about 80 months old at death. Dead. M87 is offspring of F3. Dispersed from natal area. Established territory on W side of U.P. study area. Killed by a puma hunter in 47 Canyon, Tabaguache Canyon 12-06-2011. M87 was 41 months old at death. Dead. M90 was killed by a hunter on 11-23-10 on McKenzie Butte. M90 was offspring of F72, born 07-09-08. He was 28 months old at death. Dead. M100 was killed by a puma hunter on 01-16-10 in Naturita Canyon, GMU 70 East. M100 was about 63 months old at death. Dispersed from U.P. study area after 06-23-10. Killed by a puma hunter in Beaver Creek, NE of Canyon City, GMU59, 03-10-12. M114 was about 55 months old at death. Dead. M133 was killed by a puma hunter on 12-01-10 in Dry Fork Escalante Canyon north of the study area. M133 was about 43 months old at death. 166 �Puma I.D. M134 Monitoring span 06-01-11 to 06-10-11 M138 07-01-11 to 12-23-11 M144 09-01-11 to 07-31-12 M153 M165 09-01-11 to 09-13-11 07-01-12 to 07-31-12 F2 01-07-05 to 08-14-08 F3 01-21-05 to 12-11-11 F7 02-24-05 to 08-03-08 F8 F16 03-21-05 to 07-31-12 10-11-05 to 09-11-09 F23 02-05-06 to 06-06-12 F24 01-17-06 to 07-31-11 F25 02-08-06 to 02-03-11 F28 03-23-06 to 02-16-12 F30 04-15-06 to 07-29-08 F50 12-14-06 to 03-26-07 F54 01-12-07 to 08-18-07 F70 01-14-08 to 12-22-11 F72 02-12-08 to 12-21-11 F75 03-26-08 to 12-13-11 F93 F94 12-05-08 to 07-31-12 12-19-08 to 02-01-11 F95 F96 F104 08-01-09 to 07-31-12 01-28-09 to 07-31-12 05-21-09 to 01-31-12 Table 14. Continued. Status: Alive/Lost contact/Dead; Cause of death Dead. M134 was offspring of unmarked female puma in Roubideau Canyon. Independent by about 03-28-11. Shot dead by USDA, APHIS, WS agent while in the act of attacking domestic sheep on 06-10-11 when he was 24 months old at start of adult life stage. Dead. Killed by a puma hunter in Horsefly Canyon (E) 12/23/11. M138 was about 29 months old at death. Alive. Initially captured as 18 mo. old subadult on W side U.P. study area 03-07-11. Established adult territory on NW U.P. Dead. Killed for depredation control; killed an alpaca in Pleasant Valley 09-13-11. Alive. Initially captured as 19 mo. old subadult on W side U.P. study area 02-24-12. Moved to Escalante Creek drainage by adult age 07-31-12. Dead; killed by another puma (sex of puma unknown; male suspected) 08-14-08. F2 was about 92 months old at death. Dead. Killed by a puma hunter in Lindsay Creek 12-11-11. F3 was about 120 months old at death. Dead. Killed by U.S. Wildlife.Services agent 08-03-08 for predator control of depredation on domestic sheep. F7 was about 107 months old at death. Alive. Dead. F16 was struck and killed by a vehicle on Ouray County Road 1 southwest of Colona, CO on 09-11-09. F16 was about 80 months old at death. Dead. Killed by a male puma about 06-06-12. F23 was about 94 months old at death. F23 may have attempted to defend 2 cubs (F149, M161; 13.5 months old) and/or calf elk kill. Dead. Killed by a male puma in Logging Camp Draw about 09-16-11. F24 was about 126 months old at death. F24 may have attempted to defend ≥2 cubs (F147, nonmarked siblings; 12 mo. old). Dead. Lost radio contact after 09-04-09– failed GPS/VHF collar. Photographed alive with three ~9 month old cubs on 12-03-10 on Loghill Mesa. F25 shot dead by a ranch hand on 02-03-11 in Pleasant Valley, Dallas Creek because she was seen among cattle. F25 was about 138 months old at death and in excellent physical condition (49 kg). Lost radio contact after 09-25-07− failed GPS/VHF collar. Recaptured F28 on the study area 02-01-10 and 01-01-11 and 02-16-12, but could not be handled to replace non-functional GPS collar. Dead. Killed by another puma (sex of puma unknown) 07-29-08. F30 was about 60 months old at death. Dead of natural causes 03-26-07; probably injury or illness-related; exact agent unknown. F50 was about 30 months old at death. Dead; killed by a male puma while in direct competition for prey (i.e., mule deer fawn) 08-18-07. F54 was about 49 months old at death. Dead. Killed by a puma hunter Spring Creek 12-22-11. F70 was 80 months old at death. Her death orphaned 2 cubs, F157 and F158, at 4 months old; both starved to death about 01-15-12 at about 5 months old. Lost radio contact after 12-02-10. F72 recaptured in Fisher Creek on 03-18-11, but could not be handled to replace non-functional GPS collar. Photographed on Miller Mesa S of U.P. study area on 12-18 to 21-11 with 3 new cubs born about July 2012. Dead. Killed by a puma hunter in North Fork Cottonwood Creek 12-13-11. F75 was about 98 months old at death. Alive. Dead. Shot dead on 02-01-11 by USDA, APHIS, WS agent for predation on domestic elk in Happy Canyon. F94 was about 74 months old at death. Alive. Alive. Dead. Died probably of starvation associated with senescence in lower Roubideau Creek 01-31-12. F104 was about 132 months old at death. 167 �Puma I.D. F110 Monitoring span 09-21-09 to 02-25-10 F111 F113 01-01-10 to 07-31-12 01-26-10 to 06-06-10 F116 01-20-10 to 09-20-11 F118 F119 02-25-10 to 07-31-12 03-25-10 to 01-28-12 F135 01-01-11 to 09-20-11 F136 F137 F143 F152 F163 F171 F172 01-20-11 to 07-31-12 01-21-11 to 07-31-12 02-15-11 to 07-31-12 06-16-12 to 07-31-12 07-01-12 to 07-31-12 01-20-12 to 07-31-12 03-28-12 to 07-31-12 Table 14 continued. Status: Alive/Lost contact/Dead; Cause of death Dead. Killed by a puma hunter on 02-25-10 in GMU 70 East. F110 was about 41 months old at death. Alive. Dead. F113 died 06-06-10 of injuries consistent with being struck by a vehicle. GPS data indicated that F113 had crossed highway 550 and roads on Loghill Mesa north of Ridgway 24-30 hours before she died in McKenzie Creek. F113 was about 42 months old at death. Dead. Died about 09-20-11 of unknown natural cause associated with pregnancy and birth of new litter of cubs. F116 was about 60 months old at death. Alive. Dead. Died of ruptured uterus and internal bleeding associated with pregnancy in Clay Creek Canyon 01-28-12. F119 was about 95 months old at death. Dead. Died of unknown natural cause in E Fork Dry Creek 09-20-11. Her death orphaned cubs M154 and M155 at 76 days old; both died of starvation or disease when 77 (M154) and 81 (M155) days old. Alive. Alive. Alive. Alive. Alive. Alive. Alive. 168 �Table 15. Preliminary estimated survival rates (S) of adult-age pumas during the 4 years in the reference period (i.e., the study area is closed to puma hunting) and 2 years in the treatment period, Uncompahgre Plateau, Colorado. Survival rates of pumas estimated with the Kaplan-Meier procedure to staggered entry of animals (Pollock et al. 1989). Survival rates are for an annual survival period defined as the biological year (August 1 to July 31). Survival rates were estimated only for periods when n ≥ 5 individual pumas were monitored in the interval. Puma survival in the reference period pertained only to pumas that died of natural causes. Pumas that were killed by people in the reference period, a non-natural cause (i.e., two adult pumas: F7 for depredation control 8/3/2008 and M5 killed by a puma hunter off the protected study area and buffer zone 2/20/2009) were right censored. In the treatment period all sources of natural and human-caused mortality are considered in the survival estimates. Biological Year S 1.000 Females SE 0.0000 S 0.667a Males SE 0.2222a n 6a Reference Annual 2 8/1/2005 to 7/31/2006 0.909 0.0867 11 1.000 0.0000 5 Reference Annual 3 8/1/2006 to 7/31/2007 0.831 0.0986 14 1.000 0.0000 7 Reference Annual 4 8/1/2007 to 7/31/2008 0.875 0.1031 13 1.000 0.0000 8 Reference Annual 5 8/1/2008 to 7/31/2009 0.784 0.1011 19 0.667 0.1924 8 Treatment Annual 1 8/1/2009 to 7/31/2010 Treatment Annual 1b 0.333b 0.1361b 12b 8/1/2009 to 7/31/2010 With mortalities of all marked adult males 0.947c 0.0568 19 0.250 0.1082 9 Treatment Annual 2 8/1/2010 to 7/31/2011 0.548d 0.1063 20 0.167 0.1076 7d Treatment Annual 3 8/1/2011 to 7/31/2012 a Adult male annual S 2005 to 2006 is probably underestimated with poor precision because 3 of the 6 pumas were GPS/VHF-monitored for 4 to 5 months at the end of the interval; 1 of 6 adult males died. b This second estimate of adult male puma survival 8/1/2009 to 7/31/2010 includes 5 males that had non-functional (4) or shed (1) radiocollars. All adult males with non-functional or shed radiocollars in this study survived into treatment year 1 (TY1), which was expected considering adult male survival in 3 previous years. All 5 of those adult males were detected and killed by hunters in TY1. c Only 1 of 2 adult female puma mortalities is represented in this survival analysis for 8/1/2010 to 7/31/2011, that of F94 killed for depredation control. One other adult female mortality, F25, is not represented because she wore a nonfunctional GPS collar making it impossible for us to monitor her survival. F25 was shot by a ranch hand on 2/3/2011 when he saw her among cattle. d Sample includes M144, ranges on NW Uncompahgre Plateau N of the study area but not on the U.P. study area, vulnerable to annual hunting. 169 n 10 �Table 16. Summary of subadult puma survival and mortality, December 2004 to July 2012, Uncompahgre Plateau, Colorado. Puma I.D. M5 Monitoring span 09-16-05 to 06-30-06 No. days 308 M11 06-21-06 to 12-02-07 529 F23 01-04-06 to 02-04-06 31 M31 04-19-06 to 04-26-06 7 M49 03-26-07 to 10-01-07 189 F52 01-10-07 to 05-15-07 125 F66 08-23-07 to 11-05-07 11-25-08 to 06-03-09 74 Status Survived to adult stage. M5 was offspring of F3, born August 2004. Independent and dispersed from natal area at 13 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months (protected from hunting mortality in buffer area) and ranged into the eastern edge of Utah (vulnerable to hunting). Killed by a puma hunter on 02-20-09 in Beaver Creek, Utah at about 54 months old. Survived to adult stage. M11 was offspring of F2, born May 2005. Independent at 13 months old. Dispersed from natal area at 14 months old. Moved to Dolores River valley, CO, by 12-14-06. Killed by a puma hunter on 12-02-07 when about 30 months old. Survived to adult stage. Captured on the study area when about 17 months old. Survived to adult stage; gave birth to first litter at about 21 months old. Killed by a male puma about 06-06-12. F23 was about 94 months old at death. M69 01-11-08 to 04-07-08 190 87 Survived to adult stage. M31’s estimated age at capture was 20 months. Dispersed to northern New Mexico and was killed by a puma hunter on 12-11-08 in Middle Ponil Creek, Cimarron Range. He was about 52 months old. Survived to adult stage. M49 was offspring of F50, born July 2006. Orphaned at about 9 months old, when F50 died of natural causes. Dispersed from his natal area at about 10 months old and ranged on the northeast slope of the Uncompahgre Plateau. When M49 was about 15 months old, he shed his expandable radiocollar on about 10-01-07 at a yearling cow elk kill on the northeast slope of the Uncompahgre Plateau. He was killed by a puma hunter in Blue Creek in the protected buffer zone north of the study area on 01-24-09; he was about 29 months old, a young adult. Survived to adult stage. F52 dispersed from study area as a subadult by 01-16-07. F52’s last VHF aerial location was Crystal Creek, a tributary of the Gunnison River east of the Black Canyon 05-15-07. She was treed by puma hunters on 12-29-08 on east Huntsman Mesa, southeast of Powderhorn, CO. She was about 41-43 months old and could have been in her adult-stage home range. GPS collar nonfunctional. F52 was killed by a puma hunter on 01-09-12 in North Beaver Creek SE of Powederhorn, CO. She was about 79 months old at death. Died in subadult stage. F66 was offspring of F30, born July 2007. Lost contact; her cub collar quit after 11-05-07. Recaptured as an independent subadult on her natal area 11-25-08 when 16 months old. Mother F30 was killed by a puma when F66 was 12 months old, within the age range of normal independence. F66 died of injuries to internal organs that caused massive bleeding attributed to trampling by an elk or mule deer on about 05-28-09 when she was 23 months old. Her range partially overlapped her natal area. Survived to adult stage. M69 was captured on the study area when about 14-18 months old. Emigrated from the study area as subadult by 03-1908. Last VHF aerial location was southwest of Waterdog Peak, east side of Uncompahgre River Valley on 04-07-08. M69 was killed by a puma hunter on 11-06-08 in Pass Creek in the Snowy Range, WY when he was 24 to 28 months old. 170 �Puma I.D. F95 Monitoring span 12-29-08 to 07-31-12 No. days 214 M99 02-27-09 to 04-22-09 54 M112 02-10-11 to 04-18-11 67 M115 01-13-10 to 07-21-10 189 M120 12-06-12 1 M134 03-28-11 to 06-10-11 74 M138 01-26-11 to 06-30-11 155 F140 01-13-12 to 07- 31-12 200 M141 12-23-11 1 M144 03-07-11 to 09-08-11 185 F145 03-08-11 to 09-08-11 184 F146 03-08-11 to 03-23-11 15 F147 09-16-11 to 04-12-12 209 Table 16 continued Status Alive. F95 is the offspring of F93, born about August 2007. She became an independent subadult by about 18 months old (02-11-09 aerial location) and an adult by about 24 month old (Aug. 2009). F95 established an adult home range adjacent to and overlapping the northern portion of her natal area. Died in subadult stage. M99 probably killed by another puma (canine punctures in skull including braincase) in Jan. 2010 when he was about 16 months old. His radiocollar quit after 54 days. M112 was offspring of F70. Lost contact of M112 after 04-18-11; he may have dispersed or radiocollar quit. M112 associated with F96 and her two radio-collared cubs F129 and M130 during 02-10-11 to 04-1811. Died in subadult stage. M115 was offspring of F28, born in Nov. 2008. He was about 14 months old when first captured on Jan. 13, 2010. When he was recaptured on 03-18-10, he had previously suffered a broken left ulna. M115 was probably independent by 07-15-10 when he was located outside of his natal area on a probably dispersal move. M115 died on about 07-21-10 apparently from complications of his broken left foreleg; probably not allowing him to kill prey sufficiently for survival. M115 was about 20 months old at death. Died in subadult stage. M120 was offspring of F3. M120 was killed by a puma hunter 12-06-12 in his natal area in Spring Creek. He was 17 months old at death. Survived to adult stage (barely). M134 was offspring of unmarked female puma in Roubideau Canyon. Independent by about 03-28-11. Shot dead by USDA, APHIS, WS agent while in the act of attacking domestic sheep on 06-10-11 when he was 24 months old at start of adult life stage. Survived to adult stage. Entered adult life stage 07-01-11. Killed by a puma hunter 12-23-11 in Horsefly Canyon. M138 was about 29 months old at death. Survived to late subadult stage. Will turn adult in Aug. 2012. Probably offspring of F28. Has established a home range adjacent to natal area where she was initially captured at 5 months old on 01-02-11. Died in subadult stage. M141 was killed by a puma hunter on 12-23-11 in Little Bucktail Creek. He was 16 months old at death. Survived to adult stage. Emigrated from U.P. study area. Established adult territory on northwest Uncompahgre Plateau. M144 is sibling of F145 below. Survived to adult stage. Emigrated from U.P. study area and to Colorado Mesa. Killed by a puma hunter 01-23-12 in West Bangs Canyon. F145 was 28 months old at death. Died in subadult stage. F146 was killed and eaten by a male puma while in competition for an adult bull elk carcass that one of the pumas killed in Coal Canyon on the study area. F146 was about 19 months old at death. Lost contact; radiocollar quit after 04-12-12. F147 orphaned at about 12 months old when her mother F24 was killed by a male puma on 09-1611. 171 �Puma I.D. F149 Monitoring span 06-06-12 to 07-31-12 No. days 55 M150 03-28-11 to 04-11-11 14 F152 05-04-12 to 06-16-12 44 M153 04-12-11 to 09-06-11 147 M161 06-06-12 to 07-31-12 55 F163 01-26-12 to 07-01-12 157 Table 16 continued. Status F149 (sibling of M161 below) was orphaned at 13.5 months old when her mother F23 was killed by a male puma. F149 dispersed from the natal area by 07-16-12 to E side U.P. study area when she was 14.8 months old. Dispersed. M150 was offspring of F111, born on 08-31-09. He was independent by 03-28-11 when he was 19 months old. Lost contact after 04-11-11 when M150 was in Cow Creek southeast of the study area. Survived to adult stage. F152 was independent from her mother F93 by 05-04-12 when about 23 months old. She ranged as a subadult and adult on the natal area (07-31-12). Survived to adult stage. Consorted with F137 when 23 months old on 09-07-2011. Killed by Wildlife Services agent for depredation on an alpaca in Dallas Creek on 09-13-11. M153 was 23 months old at death. M161 (sibling of F149 above) was orphaned at 13.5 months old when his mother F23 was killed by a male puma. M161 dispersed from the natal area by 06-29-12 to E side U.P. study area when he was 14 months old. F163 was captured at about 18 months old on the study area. She emigrated from the study area and may have established an adult home range on the N portion of the Uncompahgre Plateau as of July 2012 (07-16-12 most recent location). 172 �Table 17. Records of pumas that dispersed from the Uncompahgre Plateau study area, December 2004 to July 2012. Puma I.D. 1st capture date on study area 1st capture location→kill or resight location (UTM, NAD27) M5 02-04-05 13S,240577E, 4251037N→ 12S,665853Ex 4277125N M11 06-27-05 13S,248278E, 4239858N→ 12S,741882Ex 4161575N 84.8 M31 04-19-06 329.8 M38 09-08-06 12S,746919E, 4225441N→ 13S,500000Ex 4050000N 13S,249200E, 4239703N→ 12S,703371E, 4316856N M39 09-11-06 71.3 M43 09-15-06 12S,724270E, 4243610N→ 12S,709889E, 4313490N 12S,760177E, 4242995N→ 12S,739859E, 4308557N M48 10-18-06 52.0 M49 12-05-06 12S,756676E, 4247777N→ 12S,704982E, 4248998N 12S,757241E, 4258259N→ 12S,693350E, 4274559N M58 06-27-07 13S,258543E, 4238071N→ 13S,274670E, 4309488N 73.2 Estimated linear dispersal distance (km)* 102.2 104.1 68.6 66.1 Puma Information M5 was offspring of F3, born August 2004. Independent and dispersed from natal area at 13 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months (protected from hunting mortality in buffer area) and ranged into the eastern edge of Utah (vulnerable to hunting). Killed by a puma hunter on 02-20-09 in Beaver Creek, Utah at about 54 months old. M11 was offspring of F2, born May 2005. Shed expandable radiocollar 10-24 to 11-08-05. Recaptured and re-collared 04-0206. Independent at 13 months old. Dispersed from natal area at 14 months old. Moved to Dolores River valley, CO, by 12-14-06. Killed by a puma hunter on 12-02-07 when about 30 months old. M31’s estimated age at capture was 20 months. Dispersed to northern New Mexico and was killed by a puma hunter on 12-1108 in Middle Ponil Creek, Cimarron Range. He was about 52 months old. M38 was offspring of F2, born July 29, 2006. Shed his expandable radiocollar by 03-06-07. Photographs by trail camera in McKenzie Cr. of M38 & Unm. F sibling with F2 on 07-16 to 17-07 at 352-353 days old. M38 was killed by a puma hunter in Ladder Creek southwest of Grand Junction, CO on 01-07-11. He was 53.2 months old at death. M39 was offspring of F8, born August 2006. M39 was killed by a puma hunter in Bangs Canyon, GMU 40 on 03-12-10 when he was 42.8 months old. M43 was offspring of F7, born August 2006. He shed the expandable radiocollar 11-7 to 17-06, after which direct contact was lost. M43 was killed by a puma hunter 01-28-09 in Deer Creek, west slope of Grand Mesa, CO when he was 29.5 months old. M48 was the offspring of F3, born September 2006. M48 was killed by a puma hunter in Tabeguache Creek, GMU 61N on 1227-09 when he was 38.9 months old. M49 was offspring of F50, born July 2006. Orphaned at about 9 months old, when F50 died of natural causes. Dispersed from his natal area at about 10 months old and ranged on the northeast slope of the Uncompahgre Plateau. When M49 was about 15 months old, he shed his expandable radiocollar on about 10-01-07 at a yearling cow elk kill on the northeast slope of the Uncompahgre Plateau. He was killed by a puma hunter in Blue Creek GMU 61N in the protected buffer zone north of the study area on 01-24-09; he was about 29 months old. M58 was offspring of F16, born May 2007. M58 was killed by a puma hunter on 12-27-09 in the North Fork of the Gunnison River north of Paonia, GMU 521; he was 31 months old. 173 �Table 17 continued. Estimated Puma Information linear dispersal distance (km)* 66.1 M63 was offspring of F24, born July 14, 2007. He was not radiocollared as a cub. M63 was killed by a puma hunter in Calamity Creek on northwest Uncompahgre Plateau on 01-01-11. M63 was 41.5 months old at death. 97.0 M65 was offspring of F24, born July 2007. M65 was killed by a USDA, APHIS, WS agent for depredation on llamas in the Little Dolores River on 11-07-09. M65 was 27.8 months old. Puma I.D. 1st capture date on study area 1st capture location→kill or resight location (UTM, NAD27) M63 08-17-07 M65 08-17-07 M67 08-23-07 12S,738144E, 4233628N→ 12S,689111E, 4277908N 12S,738144E, 4233628N→ 12S,684084E, 4314200N 13S,257371E, 4235231N→ 12S,725113E, 4242447N M68 08-23-07 M69 01-11-08 M82 07-05-08 12S,726901E, 4243463N→ 13S,255316E, 4216768N 60.5 M83 07-05-08 90.7 M87 07-31-08 12S,726901E, 4243463N→ 12S,670949E, 4314779N 13S,239006E, 4248601N→ 12S,724325E, 4244118N M88 07-31-08 13S,239006E, 4248601N→ 12S,704835E, 4197839N 77.6 M92 09-29-08 13S,246359E, 4226949N→ 12S,750871E, 4222921N 21.9 13S,257371E, 4235231N→ 12S,711262E, 4198681N 13S,248191E, 4246810N→ 13T,378900E, 4591990N 57.7 80.7 369.6 39.2 M67 was offspring of F30, born July 17, 2007 in Fisher Creek on the east slope of the study area. He was not radiocollared as a cub. M67 dispersed from the natal area and was recaptured in Tomcat Creek on the west slope of the study area on 02-24-10 when he was 31 months old. M67 is a resident adult in that area (07-3111). Killed by puma hunter in GMU61N on 12-18-11 when 52.9 months old. M68 was offspring of F30, born July 2007. He was orphaned at 12 months old when his mother was killed by a puma. He was killed by a puma hunter in the Disappointment Valley in southwest CO on 12-30-08; he was 17 months old. M69 was captured on the study area when about 14-18 months old. Emigrated from the study area as subadult by 03-19-08. Last VHF aerial location was southwest of Waterdog Peak, east side of Uncompahgre River Valley on 04-07-08. M69 was killed by a puma hunter on 11-06-08 in Pass Creek in the Snowy Range, WY when he was 24 to 28 months old. M82 was offspring of F8, born May 29, 2008; sibling of M83 below. He shed his expandable cub radiocollar after 03-20-09. M82 was killed by a puma hunter on 12-10-09 in the Beaver Creek fork of East Dallas Creek, GMU 65. M82 was 19 months old. M83 was offspring of F8, born May 29, 2008; sibling of M82 above. He was not radiocollared as a cub. M82 was killed by a puma hunter on 01-18-11 in Coates Creek west of Glade Park, CO. He was 30 months old at death. M87 was offspring of F3, born July 3, 2008 on the east slope of the study area; sibling of M88 below. He was not radiocollared as a cub. M87 dispersed from the natal area. He was recaptured on the west slope of the study area on 02-09-11 when he was 31 months old. M87 is was resident adult on the west slope of the study area. He was killed by a puma hunter on 12-06-11 at 41 months old north of the study area. M87 was offspring of F3, born July 3, 2008 on the east slope of the study area; sibling of M87 above. He was not radiocollared as a cub. M87 dispersed from the natal area. He was killed by a puma hunter in Dawson Creek, Disappointment Valley on 11-3010 when he was 29 months old. M92 was offspring of F25, born August 19, 2008. He was radiocollared as a cub; last contact on 12-12-08. M92 dispersed from the natal area and was recaptured in McKenzie Creek, west slope of the study area on 04-22-11 when he was 32 months old. He could not be handled to fit a new radiocollar because of a dangerous tree. 174 �Puma I.D. 1st capture date on study area 1st capture location→kill or resight location (UTM, NAD27) M107 06-28-09 M114 02-27-10 M117 02-05-10 13S,242359E, 4252618N→ 12S,754886E, 4341330N 13S,256933E, 4237862N→ 13S,492615E, 4266192N 12S,731840E, 4232346N→ 12S,743909E, 4216633N M126 09-05-10 M144 03-07-33 M161 01-23-12 F52 01-10-07 F97 02-04-09 12S,727529E, 4237648N→ 12S,705930E, 4227299N F106 06-14-09 12S,736451E, 4240278N→ 13S,258089E, 4235866N 12S,734503E, 4224636N→ 12S, 710850E, 4239350N 12S,727173E, 4242012N→ 12S,696439E, 4276888N 12S,727932E, 4239430N→ 12S,750473E, 4247250N 13S,258058E, 4236260N→ 13S,319217E, 4240467N Table 17 continued. Estimated Puma Information linear dispersal distance (km)* 89.2 M107 was offspring of F94, born May 25, 2009; sibling of F108 below. He was not radiocollared as a cub. M107 dispersed from the nata area. He was killed by a puma hunter in Cottonwood Creek near Molina, CO on 12-09-10 when he was 19 months old. 237.5 M114 was initially captured at about 30 months old. Emigrated from the U.P. study area. He was killed by a puma hunter on 0310-12 in Beaver Creek, GMU59. He was about 55 months old at death. 19.7 M117 was offspring of F119. He wore an expandable cub collar, but shed the collar by 07-15-10 on the natal area when about 11 months old. M117 was killed by a puma hunter in Beaver Creek, San Miguel River at the southern extreme of his natal area on 0101-11. He was 17 months old at death. It is unknown if M117 was independent from his mother F119 at the time of his death. 27.7 M126 was offspring of F118, born Aug. 8, 2010. Lost radio contact after 03-17-11; shed his radiocollar at a mule deer cache. Dispersed from natal area. Killed by a puma hunter on 01-08-12 in Tuttle Draw WNW of Nucla, CO as 17-month-old subadult. 46.6 M144 was initially captured as an independent subadult in association with subadults F145 and F146 on the study area. Mother is unknown. He moved off the study area on 03-15-11. M144’s last aerial radio location was in Blue Creek on northwest Uncompahgre Plateau on 07-13-11; he was about 22 months old. M144 established his adult territory on northwest Uncompahgre Plateau and upper Unaweep Canyon from Sep. 2011 to July 2012. 23.9 M161 (sibling of F149) was orphaned when his mother F23 was killed by a male puma on 06-06-12; he was 411 days (13.5 mo.) old. M161 dispersed from the natal area by 06-29-12 when he was 14 months old and moved to the east slope of the U.P. study area. 61.1 F52 was captured on the study area when about 18-20 months old. Dispersed from study area as a subadult by Jan. 16, 2007. F52’s last VHF aerial location was Crystal Creek, a tributary of the Gunnison River east of the Black Canyon 05-15-07. She was treed by puma hunters on 12-29-08 on east Huntsman Mesa, southeast of Powderhorn, CO. She was about 41-43 months old . F52 was treed again by puma hunters on about 12-16-09 south of Powderhorn: 13S,319480E,4233219N. F52 was about 53-55 months old. This suggests that F52 has an adult home range in that area. F52 was killed by a puma hunter on 01-09-12 in North Beaver Creek SE of Powederhorn, CO. She was about 79 months old at death. 24.0 F97 was offspring of F23, born May 23, 2008. She was radiocollared at 8.5 month old in San Miguel Canyon; but, lost contact on 05-12-09 after F97 shed the radiocollar at an elk cache. F97 dispersed from the U.P. study area. She was killed by a puma hunter on 01-22-12 in Dry Creek west of the U.P. study area when she was 43.9 months old. 46.9 F106 was offspring of F75, born May 7, 2009. She wore an expandable cub collar, but shed it about 03-23-10. F106 dispersed from the natal area and moved to the east slope of the study area where she was photographed at one of our scent station cameras at the mouth of Fisher Creek from 02-27-11 to 03-03-11. She was identified by her eartag. F106 was 21 months old. 175 �Puma I.D. 1st capture date on study area 1st capture location→kill or resight location (UTM, NAD27) F108 06-28-09 13S,242359E, 4252618N→ 12S,752013E, 4263883N F143 02-15-11 F145 03-18-11 12S,723748E, 4238579N→ 12S,721795, 4264246 12S,727181E, 4241468N→ 12S,705833E, 4312909N F149 06-06-11 F163 01-26-12 12S,729993E, 4242329N→ 12S,715551E, 4285489N 12S,732153E, 4232452N→ 12S,695407E, 4280753N Table 17 continued. Estimated Puma Information linear dispersal distance (km)* 18.2 F108 was offspring of F94, born May 25, 2009; sibling of M107 above. She was fitted with an expandable cub collar; but, shed the collar in the original nursery due to failure of the fastener. F108 dispersed from the natal area. She was killed by a puma hunter on the study area on 11-29-10 when she was 17 months old. 25.7 F143 was captured on the study area when about 24 months old. Dispersed N on the Uncompahgre Plateau and established an adult home range on the NW portion of the Uncompahgre Plateau (most recent location 07-16-12). 74.5 F145 was originally captured in association of M144 and F146; they may be siblings. Mother unknown. She moved off the study area with M144 on 03-15-11. F145 emigrated to Colorado Mesa. She was killed by a puma hunter 01-23-12 in West Bangs Canyon. F145 was 28 months old at death. 45.5 F149 (sibling of M161) was orphaned when her mother F23 was killed by a male puma on 06-06-12; she was 411 days (13.5 mo.) old. F149 dispersed from the natal area by 07-16-12 when she was 14.8 months old and moved to the NE Uncompahgre Plateau. 60.7 F163 was initially captured at about 18 months old. She emigrated from the study area and may have established an adult home range on the N portion of the Uncompahgre Plateau as of July 2012 (0716-12 most recent location). *Estimated linear dispersal distance (km) from initial capture site on Uncompahgre Plateau study area to hunter kill, or last recapture, radio location, or observation site. 176 �Table 18. Recorded deaths of non-marked and marked pumas struck by vehicles and other unusual causes, in chronological order, on the Uncompahgre Plateau puma study area, Colorado, from 2004 to 2012. Puma sex & ID if marked M Estimated age (mo.) Date recorded Cause of death General physical condition Location & UTM NAD27 12 09-24-04 Good F 49 07-28-05 Vehicle collision Vehicle collision Pleasant Valley, County Road 24 13S,252870E,4227520N Highway 62 east of Dallas divide 13S,250000E,4222500N F17a 11 08-18-06 F 18-24 11-06-06 F 6 01-30-07 F P1005 36 09-16-08 M 12-24 08-13-08 F61a 18 11-13-08 F 12 08-10-09 F16b 80 09-11-09 M6b 99 05-21-0 F113b 42 06-06-10 Vehicle collision Vehicle collision Vehicle collision Asphyxia, lodged in fork of tree Vehicle collision Vehicle collision Vehicle collision Vehicle collision Vehicle collision Vehicle collision Good Not pregnant or lactating Good Good Good Unknown, decomposed Good Good Good Good Good Good Not pregnant or lactating M 24 08-25-10 Vehicle Excellent P1018c collision F 6 02-16-11 Vehicle Good P1030c collision M 4 10-07-11 Vehicle Fair P1034 collision a Subadult marked (i.e., tattoos, eartags), but not radio-collared. b Adult GPS/VHF-collared pumas. c Non-marked puma with P one-thousand number designation. 177 Highway 550 south of Colona 13S,257602E,4242185N Highway 550 east of Ridgway State Park 13S,259843E,4235985N Highway 62 west of Dallas divide 12S,762286Ex4218992N Davis Point, Roubideau Canyon 12S, 743718E,4255277N Highway 145 west of Placerville 13S,756490E,4212336N Highway 550 east of Ridgway State Park 13S,259843E,4235985N Highway 145 east of Norwood 12S,745739E,4222548N Ouray County Road 1 13S,253733E,4240060N Highway 550 south of Colona 13S,258610E,4236805N F113 crossed Highway 550 and roads on Loghill Mesa 24-30 hours before she died in McKenzie Creek 13S,257272E,4238435N Highway 62 Leopard Creek 12S,237747E,4220330N Highway 62 Leopard Creek 12S,760953E,4216683N Highway 62 Leopard Creek 12S,762806E,4219531N �Table 19. Pumas monitored with GPS collars on the Uncompahgre Plateau, Colorado, December 2004 to July 2012. Puma I.D. M1 M4 M6 M27 M29 M51 M55 M100 M133 F2 F3 F7 F8 F16 F23 Sex M M M M M M M M M F F F F F F F24 F25 F28 F30 F50 F52 F54 F70 F72 F75 F96 F104 F111 F113 F135 F136 F137 F152 F F F F F F F F F F F F F F F F F F F171 F172 F F Age stage adult adult adult adult adult adult adult adult adult adult adult adult adult adult subadult adult adult adult adult adult adult subadult adult adult adult adult adult adult adult adult adult adult adult subadult adult adult adult Dates monitored 12-08-04 to 07-20-06 01-28-05 to 01-14-06 02-18-05 to 05-14-08 03-12-06 to 06-21-06 04-14-06 to 01-01-08 01-07-07 to 07-15-08 01-21-07 to 11-25-10 03-27-09 to 01-16-10 11-12-10 to 12-01-10 01-07-05 to 08-14-08 01-21-05 to 12-11-11 02-24-05 to 08-03-08 03-21-05 to 10-10-06 10-12-05 to 09-10-09 01-04-06 to 02-04-06 02-05-06 to 09-04-09 01-17-06 to 07-25-07 02-09-06 to 09-09-09 03-24-06 to 08-15-07 03-30-07 to 02-22-08 12-14-06 to 03-26-07 01-10-07 to 05-08-07 01-12-07 to 08-18-08 01-14-08 to 12-22-11 02-12-08 to 07-07-10 03-26-08 to 06-03-09 01-28-09 to 07-31-12 05-29-09 to 01-31-12 01-01-10 to 07-31-12 01-27-10 to 06-06-10 01-01-11 to 09-20-11 01-20-11 to 07-31-12 04-12-11 to 07-31-12 01-18-12 to 06-15-12 06-16-12 to 07-31-12 01-20-12 to 07-31-12 03-28-12 to 07-31-12 178 �Table 20. Number of Trichinella larvae recovered from puma tongues, southwest Colorado, 2010-2011. Puma Seal Sex and/or I.D. Estimate Date Location: UTM Trichinella Larvae d Age collected NAD27 Per Gram (LPG) of Zone, Easting, Tongue Tissue (years) Number Northing F94 F 5 2/1/2011 13S,246976E,4255108N 1.2 12301 M 1.5 12/12/2010 12S,735100E,4249600N 5.1 6266 F 11-12 2/3/2011 13S,252703E,4225101N 0.4 (F25) 12039 F 4-5 11/22/2010 13S,283349E,4234088N 2.0 12042 M 3-4 11/26/2010 12S,736610E,4230762N 3.2 12045 M 2-3 12/1/2010 13S,283888E,4310965N 8.4 12046 M 3 12/1/2010 12S,729439E,4236264N 5.4 12047 M 9-10 12/2/2010 13S,257722E,4239169N 12048 M 2 12/3/2010 13S,261946E,4241911N 7.6 12302 M 2.5 12/17/2010 13S,316520E,4228320N 5.1 12314 F 5 1/13/2011 13S,305193E,4247057N 1.4 12317 M 1.5 1/17/2011 12044 F 1.5 11/29/2010 12S,752013E,4263883N M 6-7 11/25/2010 13S,239181E,4248300N (M32) 2.8 (F108) 12041 1.0 (M55) 179 0.0 0.0 �GOAL: Strategies, Information, & Tools for Managing Healthy, Self-sustaining Puma Populations in Colorado Puma Population Puma Habitat Effects of Harvest & Other Mortality Movements & Corridors Population Dynamics: Sex & Age, Vital Rates, Mortality, Population Growth Rates Vulnerability to Harvest Human Development Habitat Use Effects of Translocation Methods for Monitoring Populations Domestic Animals Puma― Human Relationships Deer, Elk, Other Natural Prey & Species of Concern Effects of Predation Effects of Evasive Conditioning Map Prey Distribution Habitat Models Population Models Puma Prey Puma―Prey Relationships Models Habitat Maps Figure. 1. An ecologically-based conceptual model of the Colorado Puma Research Program that provides the contextual framework for this and proposed puma research in Colorado. Gray-shaded shapes identify areas of research addressed by this puma research on the Uncompahgre Plateau for the puma management goal in Colorado (at top). 180 �Figure 2. The puma study area on the southern half of the Uncompahgre Plateau, Colorado (shaded in gray) comprising the southern portions of Game Management Units (GMUs) 61 and 62 and a northern portion of GMU 70. Pun1a Population Trend, U. P., CO 60 0 RY4 IY.! I Yl YEARS - - \ i ir. imum Cour t - -Post Ha·vest Lov,.r - Post harv<::st Hi~n Figure 3. Trends in the population of independent pumas on the Uncompahgre Plateau Puma Study Area, including Reference Years 4 and 5 (RY4, RY5) and Treatment Years 1, 2, and 3 (TY1, TY2, TY3). Numbers represent minimum counts that include all pumas from known radio-collared pumas, visual observations of non-marked pumas, harvested non-marked pumas, and track counts of suspected nonmarked pumas on the study area during fall to spring hunting and research capture seasons, except RY5 (45), which had to be modeled from RY4 observation data (33) because the state government hiring freeze that year affected search and capture efforts. The actual minimum count for RY5 was 37 independent pumas. The quota of 8 pumas for TY1 represented a 15% harvest of the model projected 53 181 �independent pumas expected in TY1 and was used to set the quota ahead of the hunting season. Starting in TY1, two capture teams were deployed to count pumas on the study area because the hunting season shortened our fall-winter-spring research period. We deployed a team on each the east and west sides of the study area. The minimum count for TY1 was actually 55 independent pumas, consistent with the model expected 53. Post-harvest high trend line represents the population of independent pumas after pumas harvested only on the study area by hunters. This trend line represents 14.5% to 16.7% harvest of independent pumas. Post-harvest low trend line represents the population of independent pumas after pumas harvested on the study area and pumas harvested when they ranged onto adjacent GMUs open to hunting and other mortalities are subtracted from the minimum count. TY1 post-harvest low includes 1 adult female and 3 adult males killed off the study area. The TY2 post- harvest low includes 1 adult male killed off the study area and 2 adult female pumas killed in February 2011 on the study area to protect livestock. The TY3 post-harvest low includes 1 adult female and 4 adult males harvested off the study area and 2 adult females that died of natural causes on the study area. This trend line represents 21.2% to 31.2% harvest of independent pumas. Age structure of independent pumas in November 2011 at beginning of the puma hunting season in Treatment Year 3, Uncompahgre Plateau, Colorado. 8 7 ~ 6 8 s .... ~ 4 0 3 0 ■ Female Z 2 ■ Ma l e 1 0 lto 2 >2 to >3 to >4 to >5 to >6 to >7 to >8 to >9 to 10+ 3 4 5 6 7 8 9 10 Age (years) Figure 4. Estimated age structure of independent pumas in November 2011 at the beginning of the puma hunting season in Treatment Year 3 (TY3) on the Uncompahgre Plateau study area, Colorado. All these pumas were captured and sampled by researchers or harvested by hunters and examined by researchers. Mean ± SD of female and male ages, respectively: 5.85 ± 3.05 yr. (70.19 ± 36.57 mo.), n = 16; 2.25 ± 1.58 yr. (27.00 ± 18.95 mo.), n = 10. 182 �Puma births, Uncompahgre Plateau, Colorado. 14 12 10 ""~ ..,; ..,; 8 ;:i 0 z 6 4 ■ 7 2 0 I Jan. n ■ I I I I I I 17 11 11 I I I I I n I Feb. M ar. Apr. M ay Ju ne July Au g. Sep. Oct. Nov. Dec. ■ Bi1ths 2005-2012 ■ Births 1982-1 987 Figure 5. Puma births (black bars) detected by month from May 19, 2005 to July 5, 2012 (n = 46 litters of 24 females; 44 of the litters were examined at nurseries when cubs were 26-42 days old and 2 litters confirmed by tracks of ≥1 cubs following GPS-collared mothers F28 and F111 when cubs were ≤42 days old). Also shown (gray bars) are results of the earlier effort by Anderson et al. (1992:48; 1982 to 1987, n = 10 litters of 8 females, examined when cubs were <1 to 8 months old), Uncompahgre Plateau, Colorado. 183 �Appendix A. Summary of individual puma cub survival and mortality, 2005 to 2012, Uncompahgre Plateau, Colorado. Puma I.D. M5 Estimated Age at capture (days) 183 Est. Birth date ~8-1-04 Est. survival span from 1st capture to fate or last monitor date 02-04-05 to 04-07-08 Age to last monitor date alive or at death (days, birth to fate) ~1,664 F9 31 5-28-05 06-27-05 to 4-19-06 06-27-05 to 11-20-05― 12-29-05 06-27-05 to 12-2-07 326-333 F10 31 5-28-05 M11 31 5-28-05 F12 42 5-19-05 07-01-05 to 12-08-05― 01-26-06 203-252 F13 42 5-19-05 101 F14 26 6-26-05 07-01-05 to 08-28-05 07-22-05 to 02-07-06― 03-10-06 M15 26 6-26-05 F17 34 9-22-05 F18 34 9-22-05 M19 34 9-22-05 M20 34 9-22-05 F21 37 9-26-05 176-215 918 226-257 07-22-05 to 06-06 to 14-06 10-26-05 to 08-18-06 345-353 10-26-05 to 07-20 to 27-06 10-26-05 to 07-27 to 08-02-06 10-26-05 to 05-24-06 11-02-05 to 08-16-06 301-308 330 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. Radio-collared. Survived to subadult stage by 09-16-05; independent at ~13 mo. old. Dispersed from natal area by 09-29-05 at 14 mo. old. Established territory on NW U.P. Killed by hunter in Beaver Creek, UT 02-20-09 at 54.6 months old. Radio-collared. Shed radiocollar 04-19-06 to 04-26-06. F3 Radio-collared. Dhed radiocollar 08-10-05; last tracks of F10 with mother F2 & siblings F9 & M11 observed 11-2005. F10 disappeared by 12-30-05. Radio-collared. Shed collar 10-24 to 11-08-05. Recollared on 04-02-06. Survived to subadult stage by 06-21-06, independent at 13 mo. old. Dispersed from natal area by 0711-06 at 14 mo. old. Killed by a hunter in SW CO 12-2-07 at 918 days (30 mo.) old. Radio-collared. Shed radiocollar 07-28-05―08-01-05. Tracks of F12 found in association with mother F7 on 1208-05. F12 disappeared by 01-27-06 when she was not visually observed with F7, and her tracks were not seen in association with F7’s tracks. Radio-collared. Killed and eaten by a puma possibly M5 (13 mo. old) about 08-28-05. Radio-collared. Shed radiocollar 01-20-06 to 01-25-06. Tracks of F14 were observed with tracks of mother F8 & sibling M15 on 02-07-06. Disappeared by 03-11-06, only tracks of F8 & M15 were found. Radio-collared. Shed radiocollar 06-06-06 to 06-14-06. F2 F2 F2 F7 F7 F8 F8 F16 308-314 Radio-collared. Shed radiocollar 06-06-06 to 06-14-06. Killed by a car on highway 550 on 08-18-06. Probably dependent on F16. Died at 10.8 months old Radio-collared. Probably killed by another puma. Multiple bite wounds to skull. Died at 10 months old. Radio-collared. Shed radiocollar 07-27-06 to 08-02-06. 244-245 Radio-collared. Shed radiocollar 05-24-06―05-25-06. F16 324 Radio-collared. Lost contact; radiocollar quit. Last aerial location 8-16-06, live signal. F3 184 F16 F16 �Appendix A continued Puma I.D. Estimated Age at capture (days) M22 37 9-26-05 M26 183 8-1-05 F33 31 5-30-06 F34 31 F35 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 11-02-05 to 12-21-05― 12-22-05 02-08-06 to 03-21 to 24-06 06-30-06 to 07-31-06 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 86-87 Radio-collared. Killed and eaten by male puma 12-21-05 to 12-22-05. F3 ~232-235 Radio-collared. Shed radiocollar 03-21-06 to 03-24-06. F25 63-65 F23 5-30-06 06-30-06 to 07-31-06 63-65 31 5-30-06 38 F36 29 6-9-06 29 6-9-06 M38 41 7-29-06 Radio-collared. Killed and eaten by a male puma 08-22-06. GPS data on M29 indicate he was not involved. Radio-collared. Killed and eaten by a male puma 08-22-06. GPS data on M29 indicate he was not involved. Radio-collared. Shed radiocollar found 03-06-07. Photo (trail camera in McKenzie Cr.) of M38 & Unm. F sibling with F2 on 07-16 to 17-07 at 352-353 days old. Killed by puma hunter 01-07-11 in GMU40 Ladder Creek when he was 53.2 months old. Radio-collared. Shed radiocollar by 09-20-06, but seen alive on that date. Tracks of 2 cubs following F8 on 04-25-07. Survived to adult stage; dispersed from natal area. Killed by a puma hunter 03-12-10 in GMU 40 when 42.8 months old. Radio-collared. Shed radiocollar by 09-20-06, but seen alive on that date. Tracks of 2 cubs following F8 on 04-25-07. F28 M37 06-30-06 to 07-07-06 07-08-06 to 07-28-06 07-08-06 to 07-28-06 09-08-06 to 07-16 to 17-07 Radio-collared. Probably killed and eaten by a male puma 08-01 to 03-06. GPS data on M29 indicate he was not involved. Radio-collared. Probably killed and eaten by a male puma 08-01 to 03-06. GPS data on M29 indicate he was not involved. Dead; research-related fatality.a Radio-collared. Assumed dead. Shed radiocollar or died (blood on collar) between 10-05-06 (last live signal) & 1013-06 (collar found). Dead; research-related fatality.b F8 Radio-collared. Shed radiocollar by 11-7 to 17-06. Killed by a puma hunter 01-28-09 in Deer Creek, west slope of Grand Mesa, CO GMU41 at 29.5 months old. F7 74 74 352-353 M39 29 8-13-06 F40 29 8-13-06 F41 29 8-13-06 M42 29 8-13-06 M43 33 8-13-06 09-11-06 to 09-20-06 to 04-25-07 09-11-06 to 09-20-06 to 04-25-07 09-11-06 to 10-05-06 09-11-06 to 11-27-06 09-15-06 03-01-07 1623 9 255 1307 9 Mother I.D. F23 F23 F28 F2 F8 F8 255 53-61 106 200 899 185 F8 �Appendix A continued Puma I.D. Estimated Age at capture (days) M44 33 Est. Birth date 8-13-06 Est. survival span from 1st capture to fate or last monitor date 09-15-06 to 02-14-07 Age to last monitor date alive or at death (days, birth to fate) 479 F45 33 8-13-06 09-15-06 to 5-20 to 23-07 280-283 M46 31 9-17-06 10-18-06 to 12-15-06 89 360 M47 M48 M49 31 31 153 9-17-06 9-17-06 7-1-06 10-18-06 to 12-15-06 to 09-12-07 10-18-06 to 12-15-06 to 09-12-07 to 12-27-09 89 360 89 360 1187 12-05-06 to 07-31-07 to 01-01-07 ~456 186 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Radio-collared. Shed radiocollar by 10-27-06. Treed, visually observed 02-14-07; sibling (?) M56 also captured, sampled, & marked for 1st time. M44 killed by Wildlife Services for depredation control on 12-05-07, for killing 4 domestic sheep. He was still dependent on F7. He was 15.7 months old. Radio-collared. Multiple puncture wounds on braincase― parietal & occipital regions; consistent with bites from coyote. F45 switched families, moving from F7 to F2 about 12-19 to 20-06. Last date F45 was with F2 was 04-17-07. Died 05-20 to 23-07 when she was 9.2 months old. Radio-collared. Shed collar by 12-14-06. Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-1207 in Puma Canyon. Radio-collared. Shed collar . Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-1207 in Puma Canyon. Radio-collared. Shed radiocollar. Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-1207 in Puma Canyon. Survived to adult stage; dispersed from natal area. Killed by a puma hunter 12-27-09 in GMU 61N when 38.9 months old. Radio-collared. M49 was orphaned when his mother died on about 03-26-07; he was ~268 days old. M49 dispersed from natal area and onto NE slope of U.P. Shed radiocollar at a yearling cow elk kill about 10-01-07; he was ~428 days old. Killed by a puma hunter in Blue Creek, northwest Uncompahgre Plateau (GMU 61N) 01-24-09 when ~29 months old. Mother I.D. F7 F7 F3 F3 F3 F50 �Appendix A continued Puma I.D. Estimated Age at capture (days) F53 183 Est. Birth date 7-1-06 M56c 183 ~8-13-06 F57 35 4-16-07 M58 34 5-24-07 Est. survival span from 1st capture to fate or last monitor date 01-12-07 to 02-23-07 to 09-02-07 02-14-07 to 03-01-07 05-21-07 to 06-06-07 06-27-07 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 42 Radio-collared. Shed radiocollar 02-23-07. F53 visually observed by P. & F. Star (Loghill Mesa), on 09-02-07, when F53 was ~14 months old and an independent subadult. F54 Radio-collared. Shed radiocollar 2-27-07. M56 observed 0301-07. Radio-collared. Shed radiocollar 06-07-07. Live mode 0606-07. Not radio-collared. Tracks of 3 cubs observed with F16’s tracks on 04-12-08, McKenzie Butte-Pinon Ridge Pass. 3 cubs observed with F16 on 08-08-08 by B. & T. Traegde. Survived to adult stage; dispersed from natal area. Killed by a puma hunter 12-27-09 in GMU 521 when 31 months old. Radio-collared. Shed collar about 02-14-08. Observed with 11-20-07 with F16, but without siblings M58 and F61. Tracks of 3 cubs observed with F16’s tracks on 04-12-08, McKenzie Butte-Pinon Ridge Pass. Three cubs observed with F16 on 08-08-08 by B. & T. Traegde. Dead; research-related mortality.d F7 (?) Radio-collared. Radiocollar malfunction. Tracks of 3 cubs observed with F16’s tracks on 04-12-08, McKenzie Butte-Pinon Ridge Pass. 3 cubs observed with F16 on 08-08-08 by B. & T. Traegde. Dead. Died probably as independent subadult at 538 days old; struck by car on Hwy 550 mi. marker 111 N. of Ridgway, CO, euthanized by gunshot on 11-13-08. Not radio-collared. Not radio-collared. Dispersed from study area. Killed by a puma hunter 01-01-11 in Calamity Creek, GMU61N when he was 41.5 months old. Not radio-collared. Two out of potential of 4 of F24’s male cubs were visually observed with her on 4/1/08. Assume that 2 male cubs died before the age of 8.5 mo. Eartags were seen on both cubs, but the numbers were not. F16 ~428 subad. 200 52 324 434 F59 34 5-24-07 06-27-07 to 08-21-07 55 324 M60 34 5-24-07 F61 34 5-24-07 06-27-07 to 07-11 to 12-07 06-27-07 to 06-29-07 434 48-49 324 434 538 M62 M63 M64 34 34 34 7-14-07 7-14-07 7-14-07 08-17-07 08-17-07 to 01-01-11 1267 08-17-07 262 187 Mother I.D. F25 F16 F16 F16 F24 F24 F24 �Appendix A continued Puma I.D. Estimated Age at capture (days) M65 34 Est. Birth date 7-14-07 Est. survival span from 1st capture to fate or last monitor date 08-17-07 to 11-07-09 Age to last monitor date alive or at death (days, birth to fate) Not radio-collared. Two out of potential of 4 of F24’s male cubs were visually observed with her on 4/1/08. Assume that 2 male cubs died before the age of 8.5 mo. Eartags were seen on both cubs, but the numbers were not. Survived to adult stage; dispersed from natal area. Killed by Wildlife Services for depredation control on 11-07-09 when 27.8 months old. Radio-collared. Lost contact; last location 11/5/07. No signals after that date. F66 was photographed with one male sibling, either M67 or M68, & F30 on 5/31-6/1/08. F66 was recaptured and radio-collared as a subadult on 11/25/08. She died from massive trauma & bleeding of internal organs possibly resulting from being trampled by an elk or mule deer on about 05-28-09 as an independent subadult 23 months old. Not radio-collared. M67 or M68 was photographed with sibling F66 & mother F30 on 5/31-6/1/08. Dispersed from natal area. Established adult home range on west side of Uncompahgre Plateau. Killed by puma hunter in GMU61N on 12-18-11 when 52.9 months old. Not radio-collared. M67 or M68 was photographed with sibling F66 & mother F30 on 05-31 to 06-01-08. Survived to subadult stage; dispersed from natal area. Killed by a puma hunter in Disappointment Valley, CO (GMU 71) 12-30-08 at 17.5 months old. Radio-collared. Shed radiocollar between 7-9-08 and 7-1508, probably while still dependent on mother F75. Not radio-collared. Probably dead; if not killed when sibling M79 was killed, then probably would starve to death. Not radio-collared. Probably dead; if not killed when sibling M79 was killed, then probably would starve to death. Not radio-collared. Probably dead; if not killed when sibling M79 was killed, then probably would starve to death. 262 847 F66 37 7-17-07 08-23-07 to 11-05-07 111 681 M67 37 7-17-07 08-23-07 to 12-18-11 M68 37 7-17-07 08-23-07 to 12-30-08 1615 532 F74 259 6-1-07 5-19-08 03-12-08 to 07-09-08 06-18-08 M76 30 ~87 M77 30 5-19-08 06-18-08 ~87 F78 30 5-19-08 06-18-08 ~87 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 403 188 Mother I.D. F24 F30 F30 F30 F75 F2 F2 F2 �Appendix A continued Puma I.D. Estimated Age at capture (days) M79 30 5-19-08 Est. survival span from 1st capture to fate or last monitor date 06-18-08 F80 40 5-23-08 07-02-08 F81 40 5-23-08 F97 257 5-23-08 07-02-08 to 07-29-09 02-04-09 to 01-22-12 M82 37 5-29-08 07-05-08 to 12-10-09 560 M83 37 5-29-08 07-05-08 to 01-18-11 964 M84 36 6-5-08 07-11-08 to 02-11-09 251 F85 36 6-5-08 07-11-08 to 10-01-08 118 Est. Birth date Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 87 Not radio-collared. Dead. Chewed-off anterior portions of the nasals, maxilla, palate, dentaries, and pieces of the braincase, with 6 or 9 portion of yellow ear-tag and intestines and bits of skin found ~45 m from mother F2’s death site on 08/14/08. Cub death probably due to puma-caused infanticide with cannibalism at ~87 days old. Male puma scrapes, about 8, under a rock rim ~50m distance from cub remains, and made ~ time of pumas’ deaths. Not radio-collared. Apparently died before 02-04-09; no tracks found in association with F23 & siblings F81 & F97. Radio-collared. Last live location 7-29-09. 424 1339 Radio-collared. Lost contact after 05-12-09; shed collar at elk kill cache on Mailbox Park. Dispersed from study area. Killed by a puma hunter 01-22-12 in Dry Creek when 43.9 months old. Radio-collared. Survived to subadult stage; dispersed from natal area. Killed by a puma hunter in 12-10-09 GMU 65 when 18.4 months old. Not radio-collared. Survived; dispersed from study area. Killed by a puma hunter 01-18-11 on Glade Park, GMU40. He was 31.6 months old. Radio-collared 7-11-08 to 7-22-08; collar removed because of malfunction. Not radio-collared after 7-22-08. Eartag of M84 was found by E. Phillips on 8-25-08 when mother F70’s GPS locations located her on either side of the eartag in the East fork Dolores Cyn. M84 recaptured radiocollared again 1-29-09 in Dolores Cyn. in association with F70 & F96’s family. Shed radiocollar again about 0214-09. Radio-collared. Dead. Probably died of predation or infanticide about 10-108 near elk calf kill at age 3.9 months. 189 Mother I.D. F2 F23 F23 F23 F8 F8 F70 F70 �Appendix A continued Puma I.D. Estimated Age at capture (days) F86 36 6-5-08 M87 28 M88 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 07-11-08 to 07-23 to 08-03-08 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death ~48-59 7-3-08 07-31-08 to 12-06-11 1251 28 7-3-08 07-31-08 to 11-30-10 880 F89 M90 28 36 7-3-08 7-9-08 07-31-08 08-14-08 Male 7A 28-35 7-10-08 ~08-07-08 to 08-14-08 28 to 35 Male 7B 28-35 7-10-08 ~08-07-08 to 08-14-08 28 to 35 Female 7C 28-35 7-10-08 28 to 35 M91 35 8-19-08 ~08-07-08 to 08-14-08 09-29-08 M92 35 8-19-08 09-29-08 976 F95 16 mo. June-07 12-29-08 F98 4-5 mo. Sep-Oct08 02-12-09 to 03-08-09 Radio-collared 7-22-08. Dead. Radio-collar, orange ear-tag #86 with pinna with green tattoo #86 found by J. Timmer 9-1-08. F86 died ~7-23 to 8-3-08 when mother F70’s GPS locations located her at F86 remains. Probable predation. Not radio-collared. Dispersed from natal area. Recaptured as adult on west slope of study area on 02-09-11. Alive as of 07-31-11. Killed by puma hunter on 12-06-11 at 41 months old north of the study area. Not radio-collared. Dispersed. Killed by a puma hunter in Disappointment Valley, GMU711 on 11-30-10 when 28.9 months old. Radio-collared. Radio-collared. Recaptured as young adult on study area, adjacent to natal area, on 11-16-10. Killed by a puma hunter during TY2 on 11-23-10. Not radio-collared. F7’s cubs died from starvation after they were orphaned. F7 was shot on 8-3-08 for killing domestic sheep. Not radio-collared. F7’s cubs died from starvation after they were orphaned. F7 shot on 8-3-08 for killing domestic sheep. Not radio-collared. F7’s cubs died of starvation after orphaned. F7 shot on 8-3-08 for killing domestic sheep. Radio-collared. Killed by a puma hunter on study area during TY1 as dependent cub on 11-17-09 at age 14.9 months. Radio-collared. Lost contact after 12-12-08. Dispersed from natal area. Recaptured in McKenzie Creek, west slope of study area on 04-22-11 when 32 months old. Radio-collared. Survived to adult stage. Established adult home range overlapping mother F93’s home range. To date, July 2012, F95’s home range mainly adjacent to N side of natal area. Radio-collared. Died; probably killed by male puma (infanticide). 867 455 146-176 190 Mother I.D. F70 F3 F3 F3 F72 F7 F7 F7 F25 F25 F93 Unm.F �Appendix A continued Puma I.D. Estimated Age at capture (days) M99 5 mo. Est. Birth date Sep-Oct08 M101 35 4-15-09 M102 35 4-15-09 F103 35 4-15-09 M105 38 5-7-09 F106 38 5-7-09 M107 34 5-25-09 F108 34 5-25-09 Est. survival span from 1st capture to fate or last monitor date 2-27-09 to 01-2010 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. 488 Unm.F 05-20-09 to 09-19-09 05-20-09 157 05-20-09 to 09-17-09 06-14-09 to 02-09-10 06-14-09 to 02-27-11 159 Radio-collared. Last location 4-22-09 on Paterson Mt. Died as 16-month old subadult in San Miguel Canyon. Probably killed by another puma. Radio-collared. Died; killed by puma M55 after he was orphaned due to death of mother F16 by vehicle strike. Radio-collared. Lost contact after 09-04-09. Did not find evidence of M102 associated with deaths of siblings M101 and F103. But M102 probably died. Radio-collared. Died; killed by puma M55 after she was orphaned due to death of mother F16 by vehicle strike. Radio-collared. Lost contact after 02-09-10 due to shed collar. Not radio-collared at nursery; F75 returned to nursery during handling. Radio-collared later on 2-10-10. Lost contact due to shed collar 3-16 to 29-10. F106 dispersed from natal area and was photographed at 21 months old at camera and scent-rub station on east slope of Uncompahgre Plateau on 02-27-11. Not radio-collared; too small. Recaptured 2-24-10; not collared. Shed radiocollar at nursery; fastener failed. Recaptured and re-collared 2-24-10. Shed collar ~3-5-10. Dispersed from natal area. Killed by a puma hunter on the study area during TY2 on 11-29-11 at 18.1 months old. Not radio-collared; too small. Radio-collared. Lost contact after 5-4-10 (last live signal) possibly due to failed transmitter. Recaptured and re-radiocollared on 01-24-11. Independent subadult during 02-10-11 to 04-18-11. Lost contact after 04-18-11; he may have dispersed or radiocollar quit. Radio-collared. M115 died as a subadult (~20 mo. old) due to complications of a broken left foreleg (natural cause). Radio-collared. Lost contact after 5-14-10 (last live signal); shed collar found on 7-15-10 in the natal area. Killed by a puma hunter on the natal area in Beaver Creek, off the U.P. study area on 01-01-11 when he was 17 months old. Not radio-collared. Monitored at nursery via mother’s GPS/VHF collar. Found dead at nursery due to infanticide by puma M32 on same day as our investigation of nursery. 06-28-09 to 02-24-10 06-28-09 to 03-05-10 278 275 661 241 553 M109 M112 34 145 5-25-09 8-31-09 06-28-09 05-04-10 528 595 M115 14 mo. Nov.-08 07-21-10 610 M117 6 mo. Aug.-09 02-05-10 to 01-01-11 518 P1016(M) 39 6-12-10 06-12-10 to 07-21-10 39 191 F16 F16 F16 F75 F75 F94 F94 F94 F70 F28 F119 F72 �Appendix A continued Puma I.D. Estimated Age at capture (days) P1017(M) 39 6-12-10 M120 30 M121 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 06-12-10 to 07-21-10 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 39 F72 6-28-10 07-28-10 to 12-02-10 526 30 6-28-10 273 M122 35 7-8-10 07-28-10 to 03-28-11 08-12-10 to 04-28-11 Not radio-collared. Monitored at nursery via mother’s GPS/VHF collar. Found dead at nursery due to infanticide by puma M32 on same day as our investigation of nursery. Radio-collared. Lost radio contact after 12-02-10. Killed by a puma hunter on his natal area on 12-06-11 when he was 17.2 months old. Radio-collared. Lost radio contact after 03-28-11. F104 F123 29 7-15-10 217 F124 29 7-15-10 M125 29 7-15-10 M126 28 08-08-10 08-13-10 to 02-17-11 08-13-10 to 02-16-11 08-13-10 to 02-01-11 09-05-10 to 01-08-12 M127 28 08-08-10 09-05-10 to 09-10-11 398 M128 28 08-08-10 198 F129 35 08-21-10 09-05-10 to 02-22-11 09-25-10 to 02-02-12 M130 35 08-21-10 09-25-10 to 02-02-12 M131 35 08-21-10 09-25-10 to 07-21-11 Radio-collared. Lost radio contact after 04-28-11. Tracks of 2 other siblings of M122 observed on 01-11-11 (neither cub marked). Radio-collared. Killed on 02-17-11 for depredation control on domestic elk by Wildlife Services agent. Radio-collared. Killed on 02-16-11 for depredation control on domestic elk by elk farm manager. Radio-collared. Killed on 02-01-11 for depredation control on domestic elk by Wildlife Services agent. Radio-collared. Lost radio contact after 03-17-11; shed his radiocollar at a mule deer cache. Dispersed from natal area. Killed by a puma hunter on 01-08-12 in Tuttle Draw WNW of Nucla, CO as 17-month-old subadult. Radio-collared. Lost radio contact after 07-01-11; shed his radiocollar about 07-01-11. Found dead 09-14-11 on natal area; killed by another puma on about 09-10-11 at age 13 months. Radio-collared. Lost radio contact after 02-22-11; radiocollar probably quit. Radio-collared. Fate unknown. Transmitter on mortality mode on 04-28-11. Unable to get to collar until 06-23-11 due to high spring run-off, by then the transmitter had quit. Survived to recapture on 02-02-12 at 17.4 months old, with sibling M131; neither handled due to dangerous trees. Radio-collared. Died of natural causes associated with injury to right shoulder during first move away from nursery about 10-23-10. Radio-collared. Lost contact after 07-21-11. Shed his radiocollar about 07-27-11. Survived to recapture on 02-0212 at 17.4 months old, with sibling F129; neither handled due to dangerous trees. 274 216 201 221 530 530 334 192 Mother I.D. F3 F3 F94 F94 F94 F118 F118 F118 F96 F96 F96 �Appendix A continued Puma I.D. Estimated Age at capture (days) F132 35 08-21-10 M134 ~18 mo. M139 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 09-25-10 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 35 ~June-09 12-14-10 to 06-10-11 731 36 04-18-11 05-24-11 to 07-29-11 102 F148 36 04-18-11 05-24-11 to 07-29-11 102 F140 ~5 mo. ~Aug.10 01-02-11 to 04-18-11 258 M141 ~5 mo. ~Aug.10 01-02-11 to 04-01-11 241 M142 ~5 mo. ~ 6 mo. 01-02-11 to 04-18-11 02-16-11 258 P1030 F147 ~7 mo. ~Aug.10 ~Aug.10 ~Sep.-10 04-21-11 to 07-31-11 315 F149 45 04-22-11 06-06-11 to 07-16-12 451 M150 525 08-31-09 02-07-11 to 04-11-11 588 M151 253 06-16-10 02-24-11 to 03-07-11 264 Not radio-collared. Too small for collar design. Fate unknown. Apparently died; not with F96 and siblings F129 and M130 on 02-02-12. Radiocollared as dependent large cub. Independent by about 03-28-11. Dead; killed for depredation control by Wildlife Services agent on 06-10-11. Radio-collared. Dead of infanticide and cannibalism along with sibling F148; killed and eaten by female or subadult male puma about 07-29-11. Radio-collared. Dead of infanticide and cannibalism along with sibling M139; killed and eaten by female or subadult male puma about 07-29-11. Radio-collared. Lost contact. Shed first collar about 01-2411. Recaptured and re-collared on 04-01-11. Shed second collar after 04-18-11. Recaptured and re-collared 01-12-12 as 17-month-old subadult on natal range. Radio-collared. Lost contact; shed radiocollar about 03-2911. Recaptured, but could not be handled safely on 04-0111. Radio-collared. Lost contact after 04-18-11 due to shed collar. Struck by vehicle and killed on state highway 62 in Leopard Creek, south boundary of study area on 02-16-11. Radio-collared. Orphaned at about 12 months old when her mother F24 was killed by a male puma on 09-16-11. She ranged in her natal area until her radiocollar quit after 0412-12. Radio-collared. F149 (sibling of M161) was orphaned when her mother F23 was killed by a male puma on 06-06-12; she was 411 days (13.5 mo.) old. F149 dispersed from the natal area by 07-16-12 when she was 14.8 months old. Radio-collared. M151 was independent by 03-28-11 at 19 mo. old. He dispersed from the natal area by 04-11-11 at 19.5 mo. old. Contact lost after 04-11-11. Radio-collared. Lost contact after 03-07-11 (GPS location of mother F111 at shed collar of M151). 183 193 Mother I.D. F96 Unm. F F8 F8 Unk./ F28? Unk./ F28? Unk./ F28? Unk. F24 F23 F70 F111 �Appendix A continued Puma I.D. Estimated Age at capture (days) F152 271 06-16-10 M154 42 07-06-11 M155 42 07-06-11 M156 43 07-08-11 F157 40 08-18-11 F158 40 08-18-11 09-27-11 to 01-15-12 150 M159 40 08-18-11 09-27-11 to 12-01-11 105 M161 276 04-22-11 01-23-12 to 07-16-12 451 M162 183 07-25-11 01-25-12 to 06-11-12 322 M170 137 08-29-11 01-13-12 to 03-12-12 199 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 03-14-11 to 07-31-12 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 776 08-16-11 to 09-21-11 08-16-11 to 09-25-11 08-20-11 to 09-05-11 09-27-11 to 01-15-12 77 Radio-collared. Lost contact after 03-21-11; shed collar. Recaptured 01-18-12; fit with GPS collar at 19 months old; currently (July 31, 2012) a 25-month-old adult ranging on her natal area (philopatric). Radio-collared. M154 probably died of starvation following natural death of his mother F135. Sibling M155 also died. Radio-collared. M155 died of starvation following death of his mother F135. Sibling M154 also died. Radio-collared. M156 shed the collar about 09-05-11. He was 59 days old. F157 with sibling F158 died of starvation following death of his mother F70 due to hunter harvest on 12-22-11. Cubs died 24 days after their mother died. The cubs were 150 days old. F158 with sibling F157 died of starvation following death of his mother F70 due to hunter harvest on 12-22-11. Cubs died 24 days after their mother died. The cubs were 150 days old. M159 probably died about 12-01-11 when he was located with his family (F70, siblings F157, F158). He was not located with them on 12-12-11 and was not observed with them on 12-13-11. He was 105 days old on 12-01-11. M161 (sibling of F149) was orphaned when his mother F23 was killed by a male puma on 06-06-12; he was 411 days (13.5 mo.) old. M161 dispersed from the natal area by 0629-12 when he was 14 months old. M162 probably was orphaned cub of non-marked adult female puma killed on Pinto Mesa 01-18-12. M162 died of starvation on 06-11-12 when he was 322 days (10.6 mo.) old. M170 died about 03-15-12 of unknown natural cause. He was 199 days (6.5 mo.) old. 81 56 150 194 Mother I.D. F93 F135 F135 F137 F70 F70 F70 F23 F171 �Appendix A continued Puma I.D. Estimated Age at capture (days) P1033 22 Est. Birth date 07-10-11 Est. survival span from 1st capture to fate or last monitor date NA Age to last monitor date alive or at death (days, birth to fate) 22 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. Cub P1033 was offspring of F136. It died of predation, F136 probably killed by a puma or black bear in the nursery when about 22 days old, before researchers could examine the entire litter to sample and mark the cubs. a Cub F35 probably starved between 06-30-06 & 07-07-06 after the transmitter on the expandable collar got in its mouth. b Cub M42 died after being captured by dogs, probably from stress of capture associated with severe infection of laceration under right foreleg caused by expandable radiocollar. c Cub M56 was captured in association with F7 and her cubs M43 and M44. He may have been missed at the nursery when M43 and M44 were initially sampled and marked. d Cub M60 died probably of starvation. The expandable radiocollar was around the neck and right shoulder, probably restricted movement. 195 � https://cpw.cvlcollections.org/files/original/bc30ee992c3291ecd16386953db83616.pdf 59e0ab0a34962daefadfa42d0affc1cd PDF Text Text Colorado Division of Parks and Wildlife July 2012 –June 2013 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 3003 1 Federal Aid Project: W-204-R1 : : : : Division of Parks and Wildlife Mammals Research Carnivore Conservation Puma Population Structure and Vital Rates on the Uncompahgre Plateau : Period covered: July 31, 2012−June 30, 2013 Author: Kenneth A. Logan. Personnel: K. Logan, R. Alonso, C. Anton, S. Bard, B. Dunne, W. Hollerman, W. Jesson, R. Navarrete, B. Nay, H. Taylor, S. Waters, B. Banulis, T. Bonacquista, K. Crane, J. Koch, E. Phillips, and G. Watson of CPW; volunteers and cooperators including: private landowners, Bureau of Land Management, Ridgway State Park, Colorado State University, and U.S. Forest Service. Supplemental financial support received in previous years from The Howard G. Buffett Foundation, Safari Club International Foundation, and The Summerlee Foundation. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT The Colorado Parks and Wildlife (CPW) initiated a 10-year study on the Uncompahgre Plateau in 2004 to quantify puma population characteristics in the absence (reference period, years 1-5) and presence (treatment period, years 6-10) of sport-hunting. The purpose of the study is to evaluate assumptions underlying the CPW model-based approach to managing pumas with sport-hunting in Colorado. The reference period began December 2004 and ended July 2009, during which we captured, sampled, and marked 109 pumas for population research purposes on the Uncompahgre Plateau (Logan 2009). This report provides information on the fourth year of the treatment period (TY4), August 2012 through July 2013, on puma population characteristics and dynamics with hunting as a mortality factor. Puma sport-hunting opened November 19 and closed December 29, 2012 after a quota of 5 independent pumas was harvested. The harvest was designed to test the management assumption that an 8-15% harvest of independent pumas results in a stable-to-increasing population. The decline in the puma population on the study area during TY1 to TY3 necessitated a reduction in the harvest quota from 8 to 5 to continue to test the harvest assumption for a stable-to-increasing puma population. A total of 5 pumas were killed: 2 adult females, 2 adult males, and 1 subadult male. The harvest of 5 independent pumas represented 11.9% of the 42 independent pumas in our minimum count during November 2012 to April 2013. Independent females and males comprised 40.0% and 60.0% of the harvest, respectively. Four other radio-collared independent pumas (3 adult females, 1 adult male) in the study area population died during the Colorado puma hunting season. Of those, 1 adult female died of natural cause and the remainder was killed by puma hunters in GMUs adjacent to the study area. The total mortality of 9 independent pumas during the TY4 hunting season represented 21.4% of the 42 minimum count of independent pumas on the study area. �Seventy hunters requested mandatory permits with an attached voluntary hunter survey in TY4. Forty-two of the hunters provided written responses on the surveys. An estimated 40 hunters actually hunted on the study area, of which about 12.5% harvested pumas and 15.0% captured pumas (i.e., harvested plus treed and released). Twenty of 24 answering hunters responded that they were selective hunters, and the capture, tracking, and population data indicated that most hunters practiced selection. Puma tracks < 1 day old encountered by hunters and pumas captured by hunters indicated that independent male pumas were detected more frequently than females in TY4. Researchers captured forty-nine individual pumas captured 62 times from August 2012 to July 2013. Two capture teams with dogs operated over 74 search days from January 1, 2013 through April 18, 2013 to find 229 puma tracks, pursue pumas 82 times, and capture 29 pumas 42 times. Capture efforts with cage traps resulted in the capture of 4 independent pumas and 1 cub for the first time, and the recapture of 2 adult females. Twenty-one new cubs were captured and radio-collared. A total of 55 pumas were monitored by radio-telemetry in TY4. Search efforts also revealed the presence of at least 8 other independent pumas. Our minimum count of 42 independent pumas from November 2012 to April 2013 included: 31 females and 11 males. The minimum count of 42 independent pumas in TY4 was lower than 48 in TY3, 52 in TY2 and 55 in TY1, indicating a steadily declining population. A preliminary minimum estimated density of independent pumas was 2.51/100 km2. The proportion of radio-collared adult females giving birth in the August 2012 to July 2013 biological year was 0.60 (9/15). Since 2005 a birth peak has occurred from May through August, involving 84.9% of births. We monitored 19 female and 8 male adult radio-collared pumas for survival and agent-specific mortality. Adult puma survival rates in TY4 for adult females and males were 0.819 (SE=0.0931) and 0.188 (SE=0.0845), respectively. Sporthunting mortality was the major cause of death. Of 21 cubs monitored with radio-telemetry in TY4, 7 died and 1 orphaned cub was removed from the wild. Six died of infanticide and cannibalism by male pumas and 1 was killed by puma hunting dogs. One subadult male was killed by a hunter, and 1 subadult male was struck and killed by a vehicle on state highway 62. Puma harvest, capture, and radio-telemetry data from the beginning of this study to the present provided information on dispersals of 38 pumas initially marked on the study area. Those pumas moved from about 18 to 370 km from initial capture sites. Since the start of this study 45 adult pumas have been monitored with GPS collars and have yielded over 70,000 locations. Efforts to develop and test puma population survey methods continued with a collaborative effort with M.S. student K. Yeager and Mammals Researcher Dr. Mat Alldredge. This involved 54 sites in randomly selected cells in a grid system with predator call boxes, digital cameras, and hair gathering devices from December 2012 to March 2013. Pumas were photographed at the sites 18 times with all the photos depicting GPS/VHF-collared pumas. Seven of 11 collared pumas that used the grid were detected by photographs (p = 0.64). Six hair samples were acquired from 4 to 5 individual pumas. The quality of the hair samples for accurate genotypes has yet to be analyzed. �WILDLIFE RESEARCH REPORT PUMA POPULATION STRUCTURE AND VITAL RATES ON THE UNCOMPAHGRE PLATEAU, COLORADO KENNETH A. LOGAN PROJECT NARRATIVE OBJECTIVE Quantify puma population sex and age structure; estimate puma population vital rates, including: reproduction of females, stage-specific survival, and immigration and emigration; quantify agent-specific mortality rates; model puma population dynamics; develop and execute the puma harvest manipulation to begin the population-wide test of Colorado Parks and Wildlife (CPW) puma management assumptions in the third year of a five-year Treatment Period of the Uncompahgre Plateau Puma Project― all to improve the CPW model-based approach to managing pumas in Colorado. SEGMENT OBJECTIVES 1. Execute the fourth year of the five-year treatment period by working with CPW biologists and managers to manipulate the puma population with sport-hunting and to survey hunters. 2. Continue gathering data on puma population sex and age structure. 3. Continue gathering data for estimates of puma reproduction rates. 4. Continue gathering data to estimate puma sex and stage-specific survival rates. 5. Continue gathering data on agent-specific mortality. 6. Explore non-invasive methods for sampling pumas to estimate abundance in collaboration with Dr. Mat Alldredge (Mammals Researcher, CPW) and Master of Science graduate student Kirstie Yeager, Colorado Cooperative Fish and Wildlife Research Unit, Colorado State University. INTRODUCTION Colorado Parks and Wildlife managers need reliable information on puma biology and ecology in Colorado to develop sound management strategies that address diverse public values and the CPW objective of “achieving healthy, self-sustaining populations” through management (Colorado Division Of Wildlife 2002-2007 Strategic Plan:9). Although 4 puma research efforts have been made in Colorado since the early 1970s and puma harvest data is compiled annually, reliable information on certain aspects of puma biology and ecology, and management tools that may guide managers toward effective puma management is lacking. Mammals Research staff held scoping sessions with a number of the CPW’s wildlife managers and biologists prior to initiating the project. In addition, we consulted with other agencies, organizations, and interested publics either directly or through other CPW employees. In general, CPW staff in western Colorado highlighted concern about puma population dynamics, especially as they relate to their abilities to manage puma populations through regulated sport-hunting. Secondarily, they expressed interest in puma−prey interactions. Staff on the Front Range placed greater emphasis on puma−human interactions. Staff in both eastern and western Colorado cited information needs regarding effects of puma harvest, puma population monitoring methods, and identifying puma habitat and landscape linkages. Management needs identified by CPW staff and public stakeholders form the basis of Colorado’s puma research program, with multiple lines of inquiry (i.e., projects): Improve our ability to manage puma hunting with enhanced scientific bases, strategies, and tools― ● Puma population characteristics (i.e., density, sex and age structure). �● Puma population dynamics and vital rates (i.e., birth rates, survival rates, emigration rates, immigration rates, population growth rates). ● Field methods and models for assessing and tracking changes in puma populations. ● Relative vulnerability of puma sex and age classes to hunter harvest. Improve our understanding of puma habitat needs and interrelationships of puma management units― ● Puma habitat use, movements, and use of landscape linkages. ● Puma recruitment patterns (i.e., progeny, immigration, emigration). ● Models for identifying puma habitat and landscape linkages. Improve our understanding of the puma’s role in the ecology of other species ● Relationships of puma to mule deer, elk, and other natural prey. ● Relationships of puma to species of special concern, e.g., desert bighorn sheep. Improve our understanding of puma-human interactions and abilities to manage them ● Behavior of puma in relation to people and human facilities. ● Puma predation on domestic animals. ● Effects of translocating nuisance pumas. ● Effects of aversive conditioning on pumas. While all projects cannot be addressed concurrently, understanding their relationships to one another is expected to help individual projects maximize their benefits to other projects that will assist the CPW to achieve its strategic goal in puma management (Fig.1). This project has been addressing all of the grayshaded components on the left side of the conceptual model in Figure 1. Management issues identified by managers translate into researchable objectives, requiring descriptive studies and field manipulations. Our goal is to provide managers with reliable information on puma population biology and to develop useful tools for their efforts to adaptively manage puma in Colorado to maintain healthy, self-sustaining populations. The highest-priority management needs are being addressed with this intensive population study that focuses on puma population dynamics using sampled, tagged, and GPS/VHF-radio-collared pumas to investigate the effects of sport-hunting and other causes of mortality on puma population dynamics. Those objectives include: 1. Describe and quantify puma population sex and age structure. 2. Estimate puma population vital rates, including: reproduction rates, age-stage survival rates, emigration rates, immigration rates. 3. Estimate agent-specific mortality rates. 4. Improve the CPW’s puma model-based management and attendant assumptions with Coloradospecific data from objectives 1−3. Consider other useful models. 5. Conduct a pilot study to develop methods that yield reliable estimates of puma population abundance. 6. Investigate diseases in pumas. A descriptive and manipulative study will estimate population parameters in an area that appears typical of puma habitat in western Colorado and will yield defensible population parameters based upon contemporary Colorado data. This study will be conducted in two 5-year periods. A completed 5-year reference period, 2004-09, (i.e., absence of recreational hunting) allowed puma life history traits to interact with the main habitat factors that influenced puma population growth (e.g., prey availability and vulnerability, Pierce et al. 2000, Logan and Sweanor 2001, Logan 2009). A subsequent 5-year treatment period started in 2009-10 which involves the use of controlled recreational hunting to manipulate the puma population. �TESTING ASSUMPTIONS AND HYPOTHESES Hypotheses associated with main objectives 1−5 of this puma population research are structured to test assumptions guiding puma management in Colorado. 1. Considering limitations (i.e., methods, number of years, assumption violations) to the previous Colorado-specific studies on puma populations (Currier et al. 1977, Anderson et al. 1992, Koloski 2002), managers assume that puma population densities in Colorado are within the range of those quantified in more intensively studied populations in Wyoming (Logan et al. 1986), Idaho (Seidensticker et al. 1973), Alberta (Ross and Jalkotzy 1992), and New Mexico (Logan and Sweanor 2001). The CPW assumes density ranges of 2.0−4.6 puma/100 km2 (i.e., includes pumas of all stage classes - adults, subadults, and cubs, J. Apker, CPW Carnivore Biologist, person. commun. Nov. 19, 2003) to extrapolate to Data Analysis Units (DAUs) to guide the model-based quota-setting process. Likewise, managers assume that the population sex and age structure is similar to puma populations described in the intensive studies. Using intensive efforts to capture, mark, and estimate non-marked animals developed and refined during the study to estimate the puma population, the following will be tested: H1: Puma densities during the 5-year reference period (absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will vary within the range of 2.0 to 4.6 puma/100 km2 and will exhibit a sex and age structure similar to puma populations in Wyoming, Idaho, Alberta, and New Mexico. 2. Recreational puma hunting management in Colorado DAUs is guided by a model to provide allowable harvest quotas in an effort to achieve one of two puma population objectives: 1) maintain puma population stability or growth, or 2) cause puma population decline (CDOW, Draft L-DAU Plans, 2004, CDOW 2007). These objectives are expected to provide both the capacity for puma population resiliency to achieve a state-wide goal of a healthy, self-sustaining puma population while managing the puma population to provide sport-hunting opportunity and population control in some DAUs (even though puma population dynamics in any DAUs are not known). Basic model parameters assigned to the model are: puma population density, sex and age structure, annual population growth rate, and relative puma habitat quality and quantity. Parameter quantities are currently chosen from literature on studies in western states that are judged to provide reliable information. Background material used in the model assumes a moderate annual rate of growth of 15% (i.e., λ = 1.15) for the adult and subadult puma population (CDOW 2007). This assumption is based upon information with variable levels of uncertainty (e.g., small sample sizes, data from habitats dissimilar to Colorado). Parameters influencing λ include population density, sex and age structure, female age-at-first-breeding, reproduction rates, sex- and age-specific survival, immigration and emigration. H2: Population parameters estimated during a 5-year reference period (in absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will yield an estimated annual adult plus subadult population growth rate that will match or exceed λ = 1.15. 3. An assumption is that the CPW can manage puma population growth through recreational hunting on the basis that for a stable puma population hunting removes the annual increment of population growth (i.e., from current judgments on population density, structure, and λ). Puma harvest rate formulations for DAUs assume that total mortality (i.e., harvest plus other detected deaths) in the range of 8 to 15% of the harvest-age population (i.e., independent pumas comprised of adults plus �subadults) with the total mortality comprised of 35 to 45% females (i.e., adults and subadults) is acceptable to manage for a stable-to-increasing puma population (CDOW 2007). This assumption is vital to providing the capacity for resiliency in the state-wide puma population which is hunted by applying this assumption to about three-quarters of the puma GMUs in the state. H3: Total mortality of an estimated 15% of the adults and subadults with no more than 45% of the total mortality comprised of females will not result in a declining trend of the harvest-age segment of the population. 4. To reduce a puma population, hunting must remove more than the annual increment of population growth. For DAUs with the objective to suppress the puma population, the total mortality guide of greater than 15 to 28% of the harvest-age population with greater than 45% comprised of females is suggested (CDOW 2007). This assumption is applied to about one-quarter of the GMUs in the state. H4: Total mortality of an estimated 16% or greater of the harvestable population with greater than 45% females will cause a declining trend in the abundance of harvest-age pumas (i.e., adults and subadults). 5. The increase and decline phases of the puma population make it possible to test hypotheses related to shifts in the age structure of the population which have been linked to harvest intensity in Wyoming and Utah. H5: The puma population on the Uncompahgre Plateau study area will exhibit a young age structure after hunting prohibition at the beginning of the reference period. During the 5 years of hunting prohibition, greater survival of independent pumas will cause an older age structure in harvest-age pumas (i.e., adults and subadults) as suggested by the work of Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. As hunting is re-instated in the treatment period, the age structure of harvested pumas and the harvest-age pumas in the population will decline as observed by Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. 6. Researchers in Wyoming (Anderson and Lindzey 2005) concluded that sex and age composition of the harvest varies predictably with puma population size because the likelihood of a specific sex or age class of puma being harvested with the use of hounds is a product of the relative abundance of particular sex and age classes in the population and their relative vulnerability to harvest. Results of that study suggest that managers could use sex and age composition of the harvest to infer puma population changes (Anderson and Lindzey 2005). The CPW currently uses this approach as one tool to infer potential DAU puma population dynamics (CDOW 2007). This assumes no purposeful selection by hunters for any particular sex or age-stage other than the puma must be legal (i.e., independent subadult or adult, not a lactating female or a female in association with spotted cubs) and that changes in the sex and age structure of the harvested pumas is due solely to changes in the relative abundance of particular sex and age classes in the population and their relative vulnerability to harvest. Theoretically, pumas that travel longer distances with movements that intercept access routes used by hunters (i.e., roads, trails) should be more exposed to detection by hunters and thus more vulnerable to harvest. A key assumption to this method is that pumas are killed as they are encountered and the harvest sex and age composition will reliably indicate whether a population is stable, increasing, or declining even if harvest intensity does not vary. Thus, an alternate view is that a population segment, such as independent females, may be more abundant and have shorter movement lengths, yet be detected more frequently by hunters. However, because the same intensively studied Wyoming puma population was manipulated over 6 years with varying intensities of harvest (Anderson and Lindzey 2005), variations in harvest structure using the same harvest level over a period of years could not be examined. This is a property we will investigate �during the treatment period on the Uncompahgre Plateau puma study. Moreover, we will directly evaluate to what extent puma harvest might be influenced by hunter selection. A hunter survey is intended to reveal puma hunter behavior, detection of different classes of pumas, and lack of or presence of hunter selection. These data should allow us to examine the credibility of the assumption of non-selection by hunters and the robustness of this technique in gauging puma population dynamics relative to harvest. We want to examine the usefulness of this approach in Colorado. CPW managers attempt to weight sport-harvest toward male pumas in GMUs with the stable-to-increasing population objective with an active educational program (i.e., mandatory hunter exam, brochure, workshops). Thus, there is a need to test assumptions associated with the Anderson and Lindzey (2005) method. H6: No hunter selection is practiced so that the sex and age structure of pumas harvested by hunters in this population protected from hunting during a 5-year reference period and subsequently managed for stability or increase with conservative harvest levels will reflect the relative vulnerabilities to detection and capture with dogs during each year in the 5-year treatment period in this order from high to low vulnerabilities: subadult males, adult males, subadult females, adult females without cubs or with cubs >6 months old, and adult females with cubs ≤6 months old (Barnhurst 1986, Anderson and Lindzey 2005). In each of the 5 years of the treatment period, subadults and adult males should comprise the majority of the harvest and reflect the assumed sex and age structure (Anderson and Lindzey 2005) of a puma population managed for a stable to increasing phase and not hunted for 5 previous years (i.e., a puma population source). Desired outcomes and management applications of this research include: 1. Quantification of variations in puma population density, sex and age structure, growth rates, vital rates, and an understanding of factors affecting them will aid adaptive puma management by yielding population parameters and tools useful for assessing puma population dynamics, evaluation of management alternatives, and effects of management prescriptions. 2. Testing assumptions about puma populations, currently used by CPW managers, will help managers to biologically support and adapt puma management based on Colorado-specific estimated puma population characteristics, parameters, and dynamics. 3. Methods for assessing puma population dynamics will allow managers to evaluate modeled populations and estimate effects of management prescriptions designed to achieve specified puma population objectives in targeted areas of Colorado. Ascertaining puma numbers and densities during the project will allow assessment of monitoring techniques. Potential methods include use of harvest sex and age structure and photographic and DNA genotype capture-recapture. Study plans to develop and test feasible field and analytical methods will be developed as we learn the logistics of performing those methods, after we have preliminary data on puma demographics and movements which will inform suitable sampling designs, and if we have adequate funding. 4. Information which will be disseminated to citizen stakeholders interested in pumas in Colorado, and thus contribute to informed public participation in puma management. STUDY AREA The study area for the puma population research is on the Uncompahgre Plateau (in Mesa, Montrose, Ouray, and San Miguel Counties; Fig. 2). The study area includes about 2,253 km2 (870 mi.2) of the southern halves of Game Management Units (GMUs) 61 and 62, and about 155 km2 (60 mi.2) of the northern edge of GMU 70 (between state highway 145 and San Miguel River). The area is bounded by state highway 348 at Delta, 25 Mesa road and Forest Service road FS503 to Nucla, state highway 97 to state highway 141 to state highway 145 to Placerville, state highway 62 to Ridgeway, U.S. highway 550 to Montrose, and U.S. highway 50 to Delta. �The study area seems typical of puma habitat in Colorado that has vegetation cover that varies from the pinion-juniper covered foothills starting from about 1,700 m elevation to the spruce-fir and aspen forests growing to the highest elevations of about 3,000 m. Mule deer (Odocoileus hemionus) and elk (Cervus elaphus) are the most abundant wild ungulates available for puma prey. Cattle and domestic sheep are raised on summer ranges on the study area. People reside year-round along the eastern and western fringe of the area, and there is a growing residential presence especially on the southern end of the plateau. A highly developed road system makes the study area easily accessible for puma research efforts. A detailed description of the Uncompahgre Plateau is in Pojar and Bowden (2004). METHODS Reference and Treatment Periods This research was structured in two 5-year periods: a reference period (years 1―5) and a treatment period (years 6―10). The reference period was closed to puma hunting on the study area and was expected to cause a population increase phase. The treatment period (starting in November 2009) involves manipulation of the puma population with sport-hunting structured to achieve a management objective for a stable to increasing population. In both phases, puma population structure, and vital rates are being quantified, and management assumptions and hypotheses regarding population dynamics and effects of harvest are being tested. Contingent upon results of pilot studies, we will also assess enumeration methods for estimating puma population abundance. The reference period, without recreational puma hunting as a major limiting factor, was consistent with the natural history of the current puma species in North America which evolved life history traits during the past 10,000 to 12,000 years (Culver et al. 2000) that enable pumas to survive and reproduce (Logan and Sweanor 2001). In contrast, puma hunting, with its modern intensity and ingenuity, might have influenced puma populations in western North America for at least the past 100 years. Hence, the reference period, years 1 to 5, provided conditions where individual pumas in this population expressed life history traits interacting with the environment without recreational hunting as a limiting factor. Theoretically, the main limiting factor was vulnerable prey abundance (Pierce et al. 2000, Logan and Sweanor 2001). This allowed researchers to understand basic system dynamics before manipulating the population with controlled recreational hunting. In the reference period, all pumas in the study area were protected, except for individual pumas involved in depredation on livestock or human safety incidents. In addition, all radio-collared and ear-tagged pumas that ranged in a buffer zone in the northern halves of GMUs 61 and 62 were protected from recreational hunting mortality. The reference period allowed researchers to quantify baseline demographic data on the puma population to estimate parameters useful for assessing the CPW’s assumptions for its model-based approach to puma management. The reference period also facilitated other operational needs (because hunters did not kill the animals) including the marking of a large proportion of the puma population for parameter estimates and gathering movement data from GPS-collared pumas. During the treatment period, years 6 to10, recreational puma hunting is occurring on the same study area using management prescriptions structured from information learned during previous years. Using recreational hunting for the treatment is consistent with the CPW’s objectives of manipulating natural tendencies of puma populations, particularly survival, to maintain either population stability or increase or suppression (CDOW, Draft L-DAU Plans, 2004). Theoretically, survival of independent pumas is being influenced mainly by recreational hunting, which is being quantified by agent-specific mortality rates of radio-collared pumas. Dynamics of the puma population are being manipulated to evaluate hypotheses that are related to effects of hunting (i.e., effects of harvest rates, relative vulnerability of puma sex and age classes to hunting, variations in puma population structure due to �hunting). The killing of tagged and collared pumas during the treatment period is not hampering operational needs (as it would have during the start-up years), because a majority of independent pumas in the population have already been marked, and sampling methods formalized. Pumas on the study area that may be involved in depredation of livestock or human safety incidences may be lethally controlled. Researchers that find that GPS-collared pumas have killed domestic livestock will record such incidents to facilitate reimbursement to the property owner for loss of the animal(s). In addition, researchers will notify the Area Manager of the CPW if they perceive that an individual puma may be a threat to public safety. Field Methods Puma Capture: Realizing that pumas live at low densities and capturing pumas is difficult, as a starting point, our logistical aim was to have a minimum of 6 puma in each of 6 categories (36 total) radio-tagged in any year of the study if those or greater numbers are present. The 6 categories are: adult female, adult male, subadult female, subadult male, female cub, male cub. Our aim was to provide more quantitative and precise estimates of puma demographics than were achieved in earlier Colorado puma studies. This relatively large number of pumas might represent the majority of the puma population on the study area, and would provide the basic data for age- and sex-specific reproductive rates, survival rates, agent-specific mortality rates, emigration, and other movement data. Puma capture and handling procedures were approved by the CPW Animal Care and Use Committee (file #08-2004). All captured pumas were examined thoroughly to ascertain sex and describe physical condition and diagnostic markings. Ages of adult pumas were estimated initially by the gum-line recession method (Laundre et al. 2000) and dental characteristics of known-age puma (Logan and Sweanor, unpubl. data). Ages of subadult and cub pumas were estimated initially based on dental and physical characteristics of known-age pumas (Logan and Sweanor unpubl. data). Body measurements recorded for each puma included at a minimum: mass, pinna length, hind foot length, plantar pad dimensions. Tissue collections included: skin biopsy (from the pinna receiving the 6 mm biopsy punch for the ear-tags), and blood (30 ml from the saphenous or cephalic veins) for genotyping individuals, parentage and relatedness analyses, and disease screening; hair (from various body regions) for genotyping tests of field gathered samples. Universal Transverse Mercator Grid Coordinates on each captured puma were recorded via Global Positioning System (GPS, North American Datum 27). Pumas were captured year-round using 4 methods: trained dogs, cage traps, foot-hold snares, and by hand (for small cubs). Capture efforts with dogs were conducted mainly during the winter when snow facilitated thorough searches for puma tracks and enabled dogs to follow puma scent. The study area was searched systematically multiple times per winter by four-wheel-drive trucks, all-terrain vehicles, snowmobiles, and on foot. When puma tracks ≤1 day old were detected, trained dogs were released to pursue pumas for capture. Pumas usually climbed trees to take refuge from the dogs. Adult and subadult pumas captured for the first time or requiring a change in telemetry collar were immobilized with Telazol (tiletamine hydrochloride/zolazepam hydrochloride) dosed at 5 mg/kg based on estimated body mass (Lisa Wolfe, DVM, CPW, attending veterinarian, pers. comm.). The immobilizing agent was delivered into the caudal thigh muscles via a Pneu-Dart® shot from a CO2-powered pistol. Immediately, a 3m-by-3m square nylon net was deployed beneath the puma to catch it in case it fell from the tree. A researcher climbed the tree, fixed a Y-rope to two legs of the puma and lowered the cat to the ground with an attached climbing rope. Once the puma was on the ground, its head was covered, its legs tethered, and vital signs monitored (Logan et al. 1986). Normal signs include: pulse ~70 to 80 bpm, respiration ~20 bpm, capillary refill time ≤2 sec., rectal temperature ~101oF average, range = 95 to 104oF (Kreeger 1996). Pumas that climbed trees too dangerous for the pumas or researchers for capture were released without handling, or we encourage �the animals to leave the tree by heaving snowballs toward them. If the pumas climbed a safe tree, then we handled them as described above. A cage trap was used to capture adults, subadults, and large cubs when pumas were lured into the trap using road-killed or puma-killed ungulates (Sweanor et al. 2008). A cage trap was set only if a target puma scavenged on the lure (i.e., an unmarked puma, or a puma requiring a collar change). Researchers continuously monitored the set cage trap from about 1 km distance by using VHF beacons on the cage and door. Researchers handled captured pumas within 30 minutes of capture. Puma were immobilized with Telazol injected into the caudal thigh muscles with a pole syringe. Immobilized pumas were restrained and monitored as described previously. If non-target animals were caught in the cage trap, we opened the door and allowed the animal to leave the trap. Small cubs (≤10 weeks old) were captured using our hands (covered with clean leather gloves) or with a capture pole. Cubs were restrained inside new burlap bags during the handling process and were not administered immobilizing drugs. Cubs at nurseries were approached when mothers were away from nurseries (as determined by radio-telemetry). Cubs captured at nurseries were removed from the nursery a distance of 30 to 100 m to minimize disturbance and human scent at nurseries. Immediately after handling processes were completed, cubs were returned to the exact nurseries where they were found (Logan and Sweanor 2001). Marking, Global Positioning System- and Radio-telemetry: Pumas do not possess easily identifiable natural marking, such as tigers (see Karanth and Nichols 2002), therefore, the capture, marking, and GPS- or VHF- collaring of individual pumas was essential to a number of project objectives, including estimating numbers, vital rates, and gathering movement data relevant to population dynamics (i.e., emigration and movement across Data Analysis Unit boundaries). Adults, subadults, and cubs were marked 3 ways: GPS/VHF- or VHF-collar, ear-tag, and tattoo. The identification number tattooed in the pinna was permanent and could not be lost unless the pinna was severed. A colored (bright yellow or orange), numbered rectangular (5 cm x 1.5 cm) ear-tag (Allflex USA, Inc., DFW Airport, TX) was inserted into each pinna to facilitate individual identification during direct recaptures. Cubs ≤10 weeks old were ear-tagged in only one pinna. Adult and subadult female pumas were fitted with GPS collars (approximately 400 g each, Lotek Wireless, Canada) if available. Initially, GPS-collars were programmed to fix and store puma locations at 4 times per day to sample daytime, nighttime, and crepuscular locations (i.e., 0:00, 06:00, 12:00, 19:00). GPS locations for pumas provided precise, quantitative data on movements to assess the relevance of puma DAU boundaries, our search efforts, and to evaluate puma behavior and social structure. The GPScollars also provided basic information on puma movements and locations to design other pilot studies in this program on vulnerability of puma to sport-harvest, habitat use, and enumeration methods (e.g., photographic or DNA mark-recapture). Subadult male pumas were fitted initially with conventional VHF collars (Lotek, LMRT-3, ~400 g each) with expansion joints fastened to the collars, which allowed the collar to expand to the average adult male neck circumference (~46 cm). If subadult male pumas reached adulthood on the study area, we would recapture them and fit them with GPS collars. In addition, other adult and female subadult pumas were fitted with VHF collars when GPS collars were not available. VHF radio transmitters on GPS collars enabled researchers to find those pumas on the ground in real time to acquire remote GPS data reports, facilitate recaptures for re-collaring, and to determine their reproductive and survival status. VHF transmitters on GPS- and VHF-collars had a mortality mode set to alert researchers when pumas were immobile for 3 to 24 hours so that dead pumas could be found to quantify survival rates and agent-specific mortality rates by gender and age. Locations of GPS- and VHF- �collared pumas were identified about once per week (as flight schedules and weather allowed) from light fixed-wing aircraft (e.g., Cessna 185) fitted with radio signal receiving equipment (Logan and Sweanor 2001). GPS- and VHF-collared pumas were located from the ground opportunistically using a hand-held yagi antenna. At least 3 bearings on peak aural signals were mapped to fix locations and estimate location error around those locations (Logan and Sweanor 2001). Aerial and ground locations were plotted on 7.5 minute USGS maps (NAD 27) and UTMs along with location attributes were recorded on standard forms. GPS and aerial locations were mapped using GIS software. We attempted to collar all cubs in observed litters. Cubs were fit with small VHF transmitters mounted on expandable collars that expand to adult neck size (Wildlife Materials, Murphysboro, Illinois, HLPM-2160, 47g, Telonics, Inc., Mesa, Arizona MOD 080, 62g, or Telonics MOD 205, 90g,) when cubs weighed 2.3−11 kg (5−25 lb). Cubs could wear these small expandable collars until they were over 12 months old. Cubs were recaptured to replace collars as opportunities allowed. Monitoring radio-collared cubs allowed quantification of survival rates and agent-specific mortality rates (Logan and Sweanor 2001). Analytical Methods Population characteristics each year were tabulated with the number of individuals in each sex and age category. Age categories, as mentioned, include: adult (puma ≥24 months old, or younger breeders), subadults (young puma independent of mothers, <24 months old that do not breed), cubs (young dependent on mothers, also called kittens) (Logan and Sweanor 2001). When data allowed, age categories were further partitioned into months or years. Reproductive Rates: Reproductive rates were estimated for GPS- and VHF-collared female pumas directly (Logan and Sweanor 2001). Genetic paternity analysis will be used to ascertain paternity for adult male pumas (Murphy et al. 1998). Survival and Agent-specific Mortality Rates: Radio-collared pumas provided known fate data used to estimate survival rates for each age stage using the Kaplan-Meier procedure to staggered entry (Pollock et al. 1989). A binomial survival model was also used for crude estimates of survival during the subadult age stage (Williams et al. 2001:343-344). In addition, when data collection is complete, survival rates will be modeled in program MARK (White and Burnham 1999, Cooch and White 2004) where effects of individual (e.g., sex, age stage, reproductive stage) and temporal (i.e., reference period, treatment period) covariates to survival can be examined. Agent-specific mortality rates can also be analyzed using proportions and Trent and Rongstad procedures (Micromort software, Heisey and Fuller 1985). Population Inventory: The population of interest was independent pumas (i.e., adults and subadults) mainly during November to March which corresponds with the Colorado puma hunting season. Independent pumas were those that could be legally killed by recreational hunters. Initially, we estimated the minimum number of independent pumas and puma density (i.e., number of independent puma/100 km2) each winter. The minimum number of independent pumas included all marked pumas known to be present on the study area during the period, plus individuals thought to be non-marked and detected by visual observation or tracks that were separated from locations of radio-collared pumas. Furthermore, adults comprised the breeding segment of the population and subadults were non-breeders that are potential recruits into the adult population in ≤1 year. The sampling unit was the individual independent puma (~≥1 yr. old). Puma Population Dynamics: A deterministic, discrete time model parameterized with population characteristics and vital rates from this research was used to assess puma population dynamics (Logan 2008). �Functional Relationships: Once data collection is complete, a variety of analyses will be conducted to estimate parameters and examine functional relationships. Graphical methods will be used to initially examine functional relationships among puma population parameters. Linear regression procedures and coefficients of determination will be used to assess functional relationships if data for the response variable are normally distributed and the variance is the same at each level. If the relationship is not linear, data is non-normal, and variances are unequal, we will consider appropriate transformations of the data for regression procedures (Ott 1993). Non-parametric correlation methods, such as Spearman’s rank correlation coefficient, will also be used where appropriate to test for monotonic relationships between puma abundance and other parameters of interest (Conover 1999). Relationships of explanatory variables to survival parameters will be modeled in MARK. Statistical analyses can be performed in a variety of software (e.g., SYSTAT, R, and MARK). RESULTS AND DISCUSSION Segment Objective 1 Puma harvest: This biological year, August 2012 to July 2013, was the fourth year of the treatment period (TY4) in this study of puma population dynamics on the Uncompahgre Plateau. The hunting season on the study area began on November 19, 2012 and was scheduled to extend to January 31, 2013, unless the harvest quota was taken before then. The harvest design quota was 5 pumas. This represented an 11.1% harvest of the projected minimum number of 45 independent pumas expected on the study area during November through March in TY4 (Logan 2012). The expected number of 45 was derived with a simple linear model that regressed the number of independent pumas observed in TY1 (55), TY2 (52, and TY3 (48) on year and projected to TY4 (i.e., 45 expected). We reduced the harvest rate because the population of independent pumas was declining during TY1 to TY3, contrary to the expected population trend in this research project and the realization that the population would probably continue to decline with a 15% harvest rate. This harvest design still tests the CPW’s current assumption that total mortality (i.e., harvest plus other natural deaths) in the range of 8 to 15% of the harvest-age population (i.e., independent pumas comprised of adults plus subadults) with the total mortality comprised of 35 to 45% females (i.e., adults and subadults) is acceptable to manage for a stable-toincreasing puma population (Assumption and Hypothesis 3 p.5-6 this report). The 11% harvest in TY4 is in the middle of the 8 to 15% harvest range we are testing. The initial quota of 8 pumas for TY1, TY2, and TY3 was based on the projected minimum number of 53 independent pumas expected on the study area in winter 2009-10, modeled from a minimum count of pumas during winter 2007-08 (Table 1; Logan 2010). The quota of 8 pumas for TY3 was based on the observed minimum count of 52 independent pumas during November 2010 to April 2011 in TY2 and that approximately the same number of independent pumas was expected during the puma hunting season for TY3. The hunting structure in TY4 was the same as in TY1 to TY3, except for the reduction in quota (see above). The number of puma hunters on the study area was not limited. Each hunter on the study area was required to obtain a hunting permit from the CPW Montrose Service Center. Permits were free and unlimited. Each permit allowed the individual hunter with a legal puma hunting license in Colorado to hunt in the puma study area for up to 14 days from the issue date. Unsuccessful hunters that wanted to continue hunting past the permit expiration date requested a new permit for another 14 days, or until the hunter killed a puma within the season, or the season on the study area closed due to the quota being reached, or the end of the hunting season. This permit system allowed the CPW to monitor the number of hunters on the study area and to contact each hunter for survey information (see later in this section). All pumas harvested on the study area were examined by principal investigator K. Logan or a wildlife research technician and sealed as mandated by Colorado statute. All successful hunters reported their puma kill and presented the puma carcass for inspection by CPW within 48 hours of harvest. Upon �inspection, the following data were recorded: sex, age, and location of harvest. In addition, an upper premolar tooth was collected for aging (i.e., mandatory) and a tissue sample was collected for DNA genotyping. Each successful hunter was also asked at that time to complete a one-page hunter survey form. All other hunters that did not report a puma kill on the study area were asked to complete the survey form and return it in a stamped envelope that was provided. The puma hunting season occurred on the study area from November 19 to December 29, 2012, taking 41 days to fill the quota of 5 pumas; the longest hunting season yet on the study area during this treatment period. This was 8 more days than it took to harvest 8 pumas in TY3 (i.e., 33 days, Nov. 21 to Dec. 23, 2011), 20 days more than it took to harvest 8 pumas in TY2 (i.e., 21 days, Nov. 22 to Dec. 12, 2010) and 15 more days than it took to harvest 8 pumas in TY1 (i.e., 26 days, Nov. 16 to Dec. 11, 2009). Five pumas were killed on the study area, including: 2 adult females, 2 adult males, and 1 subadult male (Table 2). Of the 5 harvested pumas, 3 were marked: F152, M156 and M179. In addition to the pumas killed on the study area during the Colorado puma hunting season, 4 other marked independent pumas died (Table 3): adult female F93 was killed by another puma on the study area, and 3 pumas were killed by hunters on adjacent GMUs, including subadult female F149 (GMU 70), adult female F177 (GMU 65) and adult male M178 (GMU 65). All these pumas were included in the minimum count of pumas for TY4 because they were initially captured on the study area and were present in the population during the TY4 survey period (Nov.–Apr.). The harvest of 5 independent pumas on the study area was 11.9% (5/42*100) of the minimum count of 42 independent pumas counted on the study area, including 31 females and 11 males, determined by the research team during November 2012 to April 2013 (Table 4). Independent females and males comprised 40.0% (2/5*100) and 60.0% (3/5*100) of the harvest, respectively. This harvest structure was 6.4% (2/31*100) of the independent females and 27.3% (3/11*100) of the independent males. Considering the mortality of 4 other radio-collared independent pumas (F93, F149, F177, M178; Table 3), the mortality of 9 independent pumas was 21.4% (9/42*100) of the minimum number of 42 independent pumas. The mortality composition of 5 females and 4 males was comprised of 55.6% (5/9*100) females and 44.4% (4/9*100) males. This harvest structure was 16.1% (5/31*100) of the independent females and 36.4% (4/11*100) of the independent males in the minimum count. The minimum count of 42 independent pumas in TY4 was lower than the minimum count of 48 independent pumas in TY3, 52 independent pumas in TY2, and 55 in TY1 (Table 4) and indicates a consistently declining population (Fig.3.). The population decline is explained mainly by the declining number of independent male pumas (Table 4) and the relatively low adult male survival rates (see Table 15, later). The number of adult females have also declined, but to a lesser extent (Table 4). Hunter permits and survey: In TY4 mandatory permits with the voluntary survey attached were requested by 70 individual puma hunters. This number is slightly down from 74 in TY3, up from 64 hunters in TY2, and down from 79 hunters in TY1. Twenty-three of the hunters requested a second permit, 7 hunters requested a third permit, and zero hunters requested a fourth permit after a previous permit expired after 14 days. Forty-two hunters (60.0%; 42/70*100) provided responses to the voluntary survey by turning in the printed survey. Of the respondents, 18 hunters indicated that they did not hunt on the study area. The proportion of the 42 respondents that hunted extrapolated to the total of 70 hunters (24/42 = 0.571) indicated that about 40 hunters took to the field for pumas on the study area during the 41-day TY4 hunting season. This was down from 49 hunters in TY3, 42 hunters in TY2 and 67 hunters in TY1 (Logan 2010, 2011). Considering that 40 hunters were estimated to be afield, then 12.5% of the hunters harvested pumas (5/40*100) and 15.0% of hunters captured pumas (6/40*100; see captured and released pumas below and in Table 5). �The 42 puma hunters that turned in the written volunteer survey were asked to answer, “Do you consider yourself a selective or non-selective hunter?” A selective hunter is one that purposely is hunting for a specific type of legal puma, such as a male, large male or large female. A non-selective hunter is one that intends to take whatever legal puma is first encountered or caught, with no desire for sex or size. Selective hunter was indicated by 20 respondents that answered the question (83.3%; 20/24 = 0.833). Of the remaining hunters, 4 indicated they were non-selective (16.7%). Eighteen hunters that returned surveys did not answer the question. The voluntary hunter survey also revealed that one hunter treed a puma on the study area, but chose not to kill it (Table 5). The hunter reported he treed a puma he believed to be a female. But his description of a yellow ear-tag in the puma indicated that it was instead a subadult male. The hunter’s reason for not wanting to kill the puma was he did not want to kill a female puma. In an effort to better ascertain the vulnerability of sexes and age-stages (i.e., adult, subadult) of independent pumas to detection by puma hunters and hunter selection to address assumption 6 and hypothesis 6 (previously), the survey was changed in TY2 to ask hunters, “What was the sex of the lion that made the first set of tracks you encountered that were less than one day old?”. This question pertained to pumas that could be pursued by dogs and captured with a relatively high probability to allow the hunter an opportunity to harvest the puma. Associated with the question, we asked, “Did you pursue the lion to harvest it?” Hunters’ responses in TY4 showed they encountered 19 puma tracks less than one day old. Of those, 8 tracks were of females, and 11 tracks were of males, indicating that during the TY4 hunting season males were more detectable than females even though independent females outnumbered independent males by 31 females and 11 males based on the minimum count (Table 4). In comparison with the previous 2 treatment years (these data were not gathered in the survey for TY1) tracks < 1 day old reported by puma hunters consistently favored females (TY2: 20 female, 10 male; TY3: 15 female, 6 male). Of the 8 female tracks less than one day old, 7 hunters that encountered them said they had no intent to harvest the puma and one hunter did not indicate his intent. Of the 11 male tracks less than one day old, 10 of the hunters that encountered them indicated intent to harvest the pumas and in fact did harvest 3 of them. One hunter did not pursue the male puma with intent to harvest it. These preliminary survey and harvest data for TY4 indicate that hunters detected independent male pumas more frequently than females and males were captured by hunters more frequently than females by 2 to 1 (i.e., males = 3 harvested + 1 captured and released; females = 2 harvested). Moreover, hunters were choosing to kill males more frequently than females. Results in TY4 indicated selection for male pumas by hunters was consistent with TY1, TY2, and TY3 results, except in those 3 previous treatment years hunters caught females slightly more frequently than males, and males were selected for harvest. This preliminary assessment from years TY1, TY2, TY3, and TY4 puma harvest and hunter survey data suggests that female pumas were detected by hunters more frequently than male pumas, except for TY4, the large majority of puma hunters were selective, and hunter choices influenced harvest sex and age composition. Segment Objective 2 After the harvest quota was filled, puma research teams immediately initiated capture operations with trained dogs. Two fully-staffed capture teams, one each detailed on the east and west slopes of the study area, systematically and thoroughly searched the study area to capture, sample, and GPS/VHF radio-collar pumas the remainder of winter and early spring 2012-13. These efforts along with cage trap efforts and hand-capturing cubs at nurseries maintained samples to quantify population sex and age structure, survival, and agent-specific mortality, and allowed determination of minimum population size on the study area during November to April. �We made 62 puma captures of 49 individuals from August 2012 to July 2013 (Tables 6-11); 29 individual pumas were captured with dogs 42 times. Seven pumas were captured in cage traps. Thirteen cubs were captured at nurseries by hand. A total of 55 individual pumas were monitored with radiotelemetry from August 2012 to July 2013 (some of these had been collared in previous years), representing sex and age classes including: 19 adult females, 8 adult males, 5 subadult females, 2 subadult males, and 21 cubs (i.e., the 2 subadult males survived to adult age during the biological year). Trained dogs were used as our main method to capture, sample, and mark pumas from January 1, 2013 to April 18, 2013. Those efforts resulted in 74 search days, 229 total puma tracks detected of which 125 were ≤1 day old, 82 pursuits, and a total of 42 puma captures of 29 individual pumas (Table 6). Search days with dogs in TY4 (74) were slightly lower than TY3 (79 days) and lower than TY1 (86 days) and TY2 (81 days)(Table 12). The frequency of tracks (tracks/day) encountered in TY4 was equivalent to TY1 and lower than TY2 and TY3. The number of pursuits in TY4 was 7 less than TY3 was 17 less than in TY2 and 11 less than in TY1. The capture rate in TY4 was substantially higher than TY1 and TY3 but somewhat less than TY2. The number of new pumas captured for the first time in TY4 was 3 higher than TY1, 3 lower than TY2 and 1 more than TY3 (Table 12). Researchers in the two hound capture teams also recorded instances when the first tracks ≤1 day old of independent pumas were encountered on each search route each day to represent encounters with puma tracks that could be detected and pursued by puma hunters. The count was: 46 tracks of females, including 9 associated with cubs; 23 tracks of males; and 1 track of unspecified sex. These tracks ≤ 1 day old were found after the TY4 puma hunting season when 4 independent females and 4 independent males were harvested (Tables 2 and 3). Therefore, the harvested pumas were not present to make tracks for our researchers to observe. By comparison, the number of first tracks <1 day hunters reported by puma hunters in TY4 was 8 females and 11 males (Segment Objective 1 above). Puma capture efforts using ungulate carcasses and cage traps occurred from September 18, 2012 to May 22, 2013 with the main efforts in the fall and spring (Table 10). We used 50 road-killed mule deer and one road-killed elk at 28 different sites. Two adult females (F176, F177), 2 adult males (M178, M179, 1 subadult female (F186), and 1 cub (F186 were captured for the first time. Two adult females (F93, F95) were recaptured and re-collared. Pumas scavenged at 12 of 51 (23.53%) of the ungulate carcasses used for bait. Pumas sometimes walked past the ungulate baits but did not feed (Table 10). We sampled 23 new cubs, including 12 females and 11 males (Table 11). All except 2 were radio-collared to monitor survival and agent-specific mortality (Appendix A). One non-marked female cub (PF1062) climbed an electrical utility pole and was electrocuted on 12/18/2012. A previously nonmarked cub (PM1068) was found dead, killed and partially consumed by a male puma; the same fate that befell his sibling M191. Besides our direct puma captures with dogs January through April, we detected 10 radio-collared pumas that we were able to identify with GPS or VHF telemetry 12 times, thus, negating the need to capture those pumas directly with dogs (Table 6). Upon detecting puma tracks that were aged at ≤1 day old, we followed the tracks with a radio receiver in an effort to detect if the tracks might be of a puma wearing a functional collar. We assigned tracks to a collared individual if we received radio signals from a puma that we judged to be <1 km from the tracks and in direction of travel of the tracks. This approach allowed us to more efficiently allocate our capture efforts toward pumas of unknown identity on the study area, particularly unmarked pumas or pumas with non-functioning GPS- or VHF- radiocollars. In addition to the harvest and capture data (previously), our search efforts revealed the presence of at least 22 other pumas which we included in our minimum count November 2012 through April 2013 (Table 4). We classified those pumas as: 7 adult females, 2 adult males, 1 subadult female, and 12 cubs. �Three adult females, 1 adult male and 1 cub were treed by our hounds, but we could not handle the pumas because they climbed dangerous trees (Table 8). Four of those were bio-darted for genotyping. Also, 1 cub jumped from a tree and was briefly caught by dogs (P1073). We collected a hair sample from it. We collected tissue samples from 1 cub that was electrocuted (previously), 1 cub killed and partially eaten by a male puma, and 1 subadult female shot by a bobcat hunter (Table 8). We could separate the activity of the other pumas from the GPS- and VHF- collared pumas in time, space, and track size differences between females, males, and numbers of cubs with females. Also, 1 non-marked adult male was photographed by a digital trail camera while consorting with 2 adult GPS-collared females (F136, F182) at the same time. Our search and capture efforts during January through April 2013 and information from the puma hunting season in TY4 enabled us to quantify a minimum count of 42 independent pumas detected on the Uncompahgre Plateau study area, including 31 independent females and 11 independent males (Table 4). This count was based on the number of known radio-collared pumas, non-marked pumas killed by hunters on the study area, observations of marked and non-marked pumas observed by researchers or pursued, treed and released by hunters on the study area, and puma tracks observed by researchers that could not be attributed to pumas with functioning radiocollars. Of the 42 independent pumas, 29 (69.0%) were marked and 13 (31.0%) were assumed to be non-marked animals (i.e., some may have ear-tags and tattoos). Our observed minimum count of 42 independent pumas for TY4 was close to the expected model projected 45 independent pumas that we used to reset the harvest quota for TY4 (see Segment Objective 1, Puma harvest). The abundance was higher on the east slope of the study area compared to the west slope. But the sex structure of independent pumas on the east and west slopes was similar. The east slope count included 24 independent pumas (18 females, 6 males). The west slope count included 18 independent pumas (13 females, 5 males). A decline in the study area puma population was most evident on the west slope. Considering the minimum count of 42 independent pumas in TY4, a preliminary minimum density for the winter puma habitat area estimated at 1,671 km2 on the Uncompahgre Plateau study area was 2.51 independent pumas/100 km2. The TY4 minimum count of 42 independent pumas is lower than the 3 previous treatment years TY1, TY2, and TY3, which indicated a steadily declining trend in the puma population on the Uncompahgre Plateau study area (Fig. 3). The declining trend was further reflected by declining survival rates of adult pumas on the study area (see Segment Objective 4&5 below). The major cause of death in the independent pumas was sport-hunting mortality (Logan 2010, 2011, 2012, this report). The estimated age structure of independent pumas in November 2012 at the beginning of the puma hunting season in TY4 on the Uncompahgre Plateau study area is depicted in Figure 4. The male age structure has declined when compared with TY1, TY2, and TY3 (Logan 2010, 2011, 2012) with the oldest males about 4 years old. The female age structure is also distributed to the younger ages with a few reaching 9 or 10 years (Logan 2010, 2011). In addition to the independent pumas, we counted a minimum of 24 cubs in TY4 (Table 4). Segment Objective 3 During the past 8.7 years of this work we compiled data on puma reproduction that was not previously available on pumas in Colorado (Table 13). Puma reproduction data (i.e., litter size, sex structure, gestation, birth interval, proportion of females giving birth per year) were summarized for the reference period in Logan (2009). In TY4 we directly observed 6 litters in nurseries of which 1 was born in May, 2 in June, 2 in July, and 1 in August (Table 11), each with 1 to 3 cubs born to radio-collared females. Data on reproduction we observed in TY1, TY2, TY3, and TY4 were added to Table 13 which gives the reproductive chronology and information on mates (if known) of reproducing females. Those �data will not be summarized again until the end of the treatment period. The proportion of radio-collared adult females giving birth from August 2012 to July 2013 biological year (TY4) was 0.53 (8/15). For the previous 3 treatment years the proportion was TY1=0.53 (8/15), TY2=0.53 (9/17), and TY3=0.29 (5/17). Considering our 53 total litters from 27 females, including 51 observed with cubs 26 to 42 days old and 2 other litters confirmed by nurseries and nursling cub tracks with GPS-collared females (the latter include F111’s cubs caught later when 8.5 months old) (Table 13), the distribution of puma births by month from 2005 to 2013 indicate births extending from March into September (Fig. 5). Births are high in May and June, peak in July, high in August and decline in September. Births during late spring to late summer (May to August) involve 84.9% (45/53*100) of the births (Fig. 5). The data indicate that the large majority of puma breeding activity occurred February through May (i.e., gestation averages about 90-92 days, Logan 2009). In comparison, Anderson et al. (1992:47-48) found on the Uncompahgre Plateau during 1982-1987 that of 10 puma birth dates 7 were during July, August, and September, 2 in October, and 1 in December, with most breeding occurring April through June. The 2 data sets indicated puma births on the Uncompahgre Plateau have occurred in every month except January and November (so far). As we gather more data on the puma births during the treatment period, we will examine the distributions of births in the reference and treatment periods separately for a treatment effect on timing of breeding and births. Segment Objectives 4 & 5 From December 8, 2004 (capture and collaring of the first adult puma M1) to July 31, 2013, we radio-monitored 28 adult male and 42 adult female pumas to quantify survival and agent-specific mortality rates (Table 14). Survival and agent-specific mortality of adult pumas were summarized for the reference period in Logan (2009). Preliminary estimates of adult puma survival rates in the absence of sport-hunting during the reference period indicated high survival, with adult male survival generally higher than adult female survival (Table 15). We monitored 19 adult females and 8 adult males for annual survival and agent-specific mortality in TY4. Annual survival rate for adult females was 0.819 (SE=0.0931) and for males was 0.188 (SE=0.0845). Preliminary adult puma survival for TY1, TY2, and TY3 are also shown in Table 15. So far, adult male survival is substantially lower in the treatment period than in the reference period. Adult female survival is lower in TY1 and TY3, with marked decline in TY3. Yet, female survival is generally higher than male survival. These characteristics are indicative of hunter selection for male pumas (previously in Segment Objective 1). The lower adult puma survival rates, particularly of males, were consistent with an observed decline in the puma population on the study area (see Segment Objective 2, previously). Human-related factors caused 4 deaths of radio-marked adult pumas in TY4, including: sporthunting harvest (2 males- M178, M179; 2 females- F152, F177) (Tables 2, 3, 14). In addition, 1 adult female puma died of natural causes: F93 was killed by another puma (Table 14). We have information on 35 subadult pumas (i.e., independent pumas <24 months old), including 14 females and 21 males (Table 16). We lost radio contact with 2 male and 2 females that probably dispersed from the study area unknown distances. Of the remaining 31 subadults (females and males combined), 8 (3 females, 5 males) died before reaching adulthood, indicating a rough preliminary binomial survival rate of 0.74 (i.e., 23/31) for subadults surviving to the adult age stage (i.e., 24 mo. old). Of the 8 subadults that died, 4 deaths were from natural causes, 3 were from sport-hunting, and 1 was from a vehicle strike (Table 16).We need to increase our efforts to acquire larger samples of male and female radio-monitored subadult pumas to acquire more reliable estimates of their survival. �Harvest data along with our capture and radiotelemetry data provided dispersal and fate information on 38 marked pumas, 29 males and 9 females. Of those, 28 (4 females, 24 males) were initially captured and marked as cubs, and 10 (5 females, 5 males) were captured and marked in the subadult life-stage on the Uncompahgre Plateau puma study area (Table 17). Twenty-three males were killed by hunters away from the study area at linear distances (i.e., from initial capture sites to kill sites) ranging from about 20 to 370 km. Two males with extreme moves were killed in the Snowy Range of southeastern Wyoming (369.6 km) and the Cimarron Range of north-central New Mexico (329.8 km). One male was killed by a hunter on the study area 12.9 km from his original capture site. Four females were killed by puma hunters off the study area ranging from 20.7 to 74.5 km from initial capture sites. One female was killed by a hunter on the study area 18.2 km from her initial capture site. Female F52 was treed and released by hunters in December 2008 and 2009 south of Powderhorn, Colorado, indicating that she established an adult home range there before she was killed by a puma hunter in that area on Jan. 9, 2012. Three males (M67, M87, M92) that were marked initially as cubs born on the east slope of the study area, dispersed from their natal ranges and were recaptured as adults on the west slope of the study area. Two of those (M67, M87) were killed on their adult territories by hunters. One (M92) is of unknown fate as of July 2013. A preliminary estimate of cub survival during the reference period was summarized in Logan 2009 using 36 radio-collared cubs (16 males, 20 females) marked at nurseries when they were 26 to 42 days old. In that summary, estimated survival of cubs to one year of age was 0.53. [The estimated minimum survival rate using the Kaplan-Meier procedure was 0.5285 (SE = 0.1623). The maximum estimated cub survival was practically the same, 0.5328 (SE = 0.1629).] The major natural cause of death in cubs, where cause could be determined, was infanticide and cannibalism by other, especially male, pumas. In TY4 we monitored the fates of 21 radio-collared cubs (Table 11, Appendix A). We lost contact with 2 (F185, F195) after they shed their expandable radio-collars prematurely. Of the remaining 19 collared cubs, 7 died and 1 was orphaned and removed from the wild to be rehabilitated to the subadult stage. One non-marked cub in association with a radio-collared cub was also found dead. Eight cubs from 3 litters (1 of those litters with the radio-collared and a non-collared cub) died from infanticide and cannibalism by male pumas. One cub (M175) died when it was apparently mauled by puma-hunting dogs. Later his mother (F152) was killed by a puma hunter. Her death orphaned the remaining cub (M174) and he was recaptured and removed from the wild to be rehabilitated at the CPW Wildlife Rehabilitation Center in Del Norte, Colorado. Of the 11 remaining live radio-collared cubs 4 survived to the subadult stage and 7 were being monitored in association with their mothers as of July 31, 2013. A greater number of cubs over a longer period of time must be sampled before estimating cub survival and agent-specific mortality rates in the treatment period. Subadult male puma M161 was struck and killed by a vehicle on state highway 62 at Dallas Divide on the south boundary of the study area in October 2012 (Tables 17, 18, Appendix A). This mortality made the fifteenth puma death recorded due to vehicle collision on the study area since 2004 (Table 18). Six of the 15 pumas were marked, including 3 adults with GPS/VHF collars. Those 3 adults died during the first year of the treatment period. Forty-five adult pumas (33 females, 12 males) have worn GPS collars since this project began in 2004 (Table 19). Over 70 thousand GPS locations have been obtained and will be used for studies on puma behavior, social organization, population dynamics, population genetics, movements, population survey methods, habitat use and puma-human relations in collaboration with colleagues in Mammals Research, Colorado State University, and Arizona State University. �Segment Objective 6 We continued to explore non-invasive methods for sampling pumas to estimate abundance by collaborating with Dr. Mat Alldredge (Mammals Researcher, CPW) and Master of Science graduate student Kirstie Yeager, Colorado Cooperative Fish and Wildlife Research Unit and Colorado State University. Here only a brief summary will be presented. M.S. student Kirstie Yeager is currently in the process of analyzing data. For a detailed report refer to Yeager (2013). A grid of 2 km x 2 km (4 sq. km) cells was established on the east slope of the Uncompahgre Plateau study area (Fig. 6). Eighteen cells were identified randomly for each of 3 survey periods each lasting about 28 days. A total of 54 random cells were surveyed during December 2012 to March 2013. Within each random cell M.S. student Kirstie Yeager subjectively chose the “best” site to attract pumas by using vocal baits each consisting of a Fur-Finder ® (Magna, UT) electronic predator call of a distressed deer fawn. Each site also had a Reconyx ® (Holmen, WI) PC900 Hyperfire camera to record animal activity and hair-sampling devices (i.e., barbed-wire strands, sticky rollers) to attempt to acquire hair. This was an effort to evaluate these methods for a non-invasive survey of puma abundance by using tissue to genetically identify individuals in a mark-recapture structure. During the survey spanning December 2012 to March 2013 eleven GPS and VHF collared pumas were known to use the survey grid for varying amounts of time, including 7 adult females, 1 subadult female, 2 adult males, and 1 subadult male. During the survey a total of 18 photographs of pumas visiting the sites were acquired, and all 18 of the photographs depicted GPS or VHF collared pumas. No noncollared pumas were photographed. Of the 11 collared pumas known to use the grid, at least 7 of them were photographed 1 to 4 times each, including 5 adult females, 1 subadult female, and 1 subadult male. Probability of detecting the 11 collared pumas available during the entire survey time was 0.64 (p = 7/11). Six hair samples were acquired from 4 to 5 individual pumas. The quality of the tissues for accurate genotypes of the individual will be determined by K. Yeager later by comparing with genotypes derived from skin and hair samples acquired from the individual pumas at capture and handling events. SUMMARY Manipulative, long-term research on puma population dynamics, effects of sport-hunting, and development and testing of puma enumeration methods began in December 2004. After 8.7 years of effort 202 unique pumas have been captured, sampled, marked, and released. Using these animals, we monitored fates of pumas in all sexes and age stages, including: 42 adult females, 28 adult males, 14 subadult females, 21 subadult males, 56 female cubs, 82 male cubs, and 1 cub of undetermined sex (some individuals occur in more than one stage class). Data from marked animals were used to quantify puma population characteristics and vital rates in a reference period without sport-hunting off-take as a mortality factor from December 2004 to July 2009. Puma population characteristics and vital rates in a reference condition allowed us to develop a puma population model, and to use population data and modeling scenarios to conduct a preliminary assessment of CPW puma management assumptions and guide directions for the remainder of the puma research on the Uncompahgre Plateau. Moreover, our data and model provide tools currently useful to CPW wildlife biologists and managers for assessing puma harvest strategies. The 5-year treatment period began August 2009 in which sport-hunting is a mortality factor. The treatment period will be a population-wide test of CPW puma management assumptions. Now 4 years of the treatment period are complete (TY1, TY2, TY3, TY4). Although data support some CPW puma management assumptions (e.g., population structure, density, how sport-harvest can cause population decline), it is still too early in this research to adequately test all the assumptions and attendant hypotheses. Although the assumption and hypothesis on harvest structure and hunter selection is not supported with the first 4 years of data in the treatment period, this could change with a substantial change in abundance and sex structure of independent pumas available for hunting in TY5. The puma harvest quota for TY5 is recommended to be 5 independent pumas to align with the research design and �harvest objective, and the hunters will be surveyed again. Since the beginning of this study 2 efforts have been made to develop and test non-invasive methods for estimating puma abundance. These efforts were in collaboration with Colorado State University in a Ph.D. program (Jesse Lewis) and a M.S. program (Kirstie Yeager). To improve data on puma population vital rates, attention will be given to increasing radio-collared sample sizes across the various life stages and sexes. Furthermore, we will continue collaboration with colleagues on investigations of puma population parameter estimation, population genetics, puma movements, puma habitat modeling and mapping, puma-human relations, and disease prevalence. All of these efforts should enhance the Colorado puma research and management programs. LITERATURE CITED Anderson, A. E, D. C. Bowden, and D. M. Kattner. 1992. The puma on Uncompahgre Plateau, Colorado. Technical Publication No. 40. Colorado Division of Wildlife, Denver. Anderson, C. R., Jr., and F. G. Lindzey. 2005. Experimental evaluation of population trend and harvest composition in a Wyoming cougar population. Wildlife Society Bulletin 33:179-188. Barnhurst, D. 1986. Vulnerability of cougars to hunting. Master’s Thesis. Utah State University. Colorado Division of Wildlife 2002-2007 Strategic Plan. 2002. Colorado Department of Natural Resources, Division of Wildlife. Denver. Colorado Division of Wildlife. 2007. Colorado mountain lion management data analysis unit revision and quota development process. Colorado Division of Wildlife, Denver. Conover, W. J. 1999. Practical nonparametric statistics. John Wiley & Sons, Inc., New York. Cooch, E., and G. White. 2004. Program MARK- a gentle introduction, 3rd edition. Colorado State University, Fort Collins. Culver, M., W. E. Johnson, J. Pecon-Slattery, and S. J. O’Brien. 2000. Genomic ancestry of the American puma (Puma concolor). The Journal of Heredity 91:186-197. Currier, M. J. P., and K. R. Russell. 1977. Mountain lion population and harvest near Canon City, Colorado, 1974-1977. Colorado Division of Wildlife Special Report No. 42. Gajadhar, A. A., and L. B. Forbes. 2010. A 10-year wildlife survey of 15 species of Canadian carnivores identifies new hosts or geographical locations for Trichinella genotypes T2, T4, T5, and T6. Veterinary Parasitology 168: 78-83. Heisey, D. M., and T. K. Fuller. 1985. Evaluation of survival and cause specific mortality rates using telemetry data. Journal of Wildlife Management 49:668-674. Karanth, K. U., and J. D. Nichols. 2002. Monitoring tigers and their prey: A manual for researchers, managers and conservationists in tropical Asia. Centre for Wildlife Studies, Bangalore, India. Kennedy, E. D., R. L. Hall, S. P. Montgomery, D. G. Pyburn, and J. L. Jones. 2009. Trichinellosis surveillance – United States, 2002-2007. Morbidity and Mortality Weekly Report 58: 1-7. Koloski, J. H. 2002. Mountain lion ecology and management on the Southern Ute Indian Reservation. M. S. Thesis. Department of Zoology and Physiology, University of Wyoming, Laramie. Kreeger, T. J. 1996. Handbook of wildlife chemical immobilization. Wildlife Pharmaceuticals, Inc., Fort Collins, Colorado. Laundre, J. W., L. Hernandez, D. Streubel, K. Altendorf, and C. L. Lopez Gonzalez. 2000. Aging mountain lions using gum-line recession. Wildlife Society Bulletin 28:963-966. Logan, K. A., E. T. Thorne, L. L. Irwin, and R. Skinner. 1986. Immobilizing wild mountain lions (Felis concolor) with ketamine hydrochloride and xylazine hydrochloride. Journal of Wildlife Diseases. 22:97-103. _____, and L. L. Sweanor. 2001. Desert puma: evolutionary ecology and conservation of an enduring carnivore. Island Press, Washington, D.C. _____. 2008. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Wildlife, Fort Collins. _____. 2009. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Wildlife, Fort Collins. �_____. 2010. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Wildlife, Fort Collins. _____. 2011. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Parks and Wildlife, Fort Collins. _____. 2012. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Parks and Wildlife, Fort Collins. Murphy, K., M. Culver, M. Menotti-Raymond, V. David, M. G. Hornocker, and S. J. O’Brien. 1998. Cougar reproductive success in the Northern Yellowstone Ecosystem. Pages 78-112 in The ecology of the cougar (Puma concolor) in the Northern Yellowstone ecosystem: interactions with prey, bears, and humans. Dissertation, University of Idaho, Moscow. Ott, R. L. 1993. An introduction to statistical methods and data analysis. Fourth edition. Wadsworth Publishing Co., Belmont, California. Pierce, B. K., V. C. Bleich, and R. T. Bowyer. 2000. Social organization of mountain lions: does land a tenure system regulate population size? Ecology 81:1533-1543. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Pojar, T. M., and D. C. Bowden. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550-560. Pozio, E., and G. La Rosa. 2003. PCR-derived methods for the identification of Trichinella parasites from animal and human samples. Methods in Molecular and Cellular Biology 216: 299-309. Rausch, R. L., C. Maser, and E. P. Hoberg. 1983. Gastrointestinal helminths of the cougar, Felis concolor L. in northeastern Oregon. Journal of Wildlife Diseases 19: 14-19. Ross, P. I., and M. G. Jalkotzy. 1992. Characteristics of a hunted population of cougars in southwestern Alberta. Journal of Wildlife Management 56:417-426. Seidensticker, J. C., M. G. Hornocker, W. V. Wiles, and J. P. Messick. 1973. Mountain lion social organization in the Idaho Primitive Area. Wildlife Monographs No. 35. Stoner, D. C. 2004. Cougar exploitation levels in Utah: implications for demographic structure, metapopulation dynamics, and population recovery. Master of Science Thesis. Utah State University. Sweanor, L. L., K. A. Logan, J. W. Bauer, B. Milsap, and W. M. Boyce. 2008. Puma and human spatial and temporal use of a popular California state park. Journal of Wildlife Management 72:10761084. Vollbrecht, A., D. Sokolowski, W. Hollipeter, R. Sigler, J. Greenblatt, D. E. Anderson, P. J. Tennican, H. R. Gamble, and D. Zarlenga. 1996. Oubreak of trichinellosis associated with eating cougar jerky – Idaho, 1995. Morbidity and Mortality Weekly Report 45: 205-206. Webster, P., C. Maddox-Hyttel, K. Nockler, A. Malajauskas, J. van der Giessen, E. Pozio, P. Boireau, C. M. O. Kapel. 2006. Meat inspection for Trichinella in pork, horsemeat and game within the EU: available technology and its present implementation. Eurosurveillance 11:50-55. Williams, B. K., J. D. Nichols, and M. J. Conroy. 2001. Combining closed and open mark-recapture models: the robust design. Pages 523-554 In Analysis and management of animal populations. Academic Press, New York. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 (Suppl):S120-S139. Winters, J. B. 1969. Trichiniasis in Montana mountain lions. Bulletin of the Wildlife Disease Association 5: 400. Worley, D. E., J. C. Fox, J. B. Winters, and K. R. Greer. 1974. Prevalence and distribution of Trichinella spiralis in carnivorous mammals in the United States northern Rocky Mountain region. Proceedings of the 3rd International Conference on Trichinellosis, Florida, USA, November 2-4, 1972. �Yeager, K., B. Kendall, M. Alldredge. 2013. The Use of Lures, Hair Snares, and Snow Tracking as NonInvasive Sampling Techniques to Detect and Identify Cougars. Colorado Cooperative Fish and Wildlife Research Unit, Ft. Collins. Zarlenga, D.S., Chute, M.B., Martin, A. and Kapel, C.M. 1999. A multiplex PCR for unequivocal differentiation of all encapsulated and nonencapsulated genotypes of Trichinella. International Journal for Parasitology 29:1859–1867. Prepared by: ________________________ Kenneth A. Logan, Wildlife Researcher �Table 1. Projected puma population growth modeled from a minimum count of independent pumas during winter 2007-08 reference period year 4 (RY4). Treatment period year 1 (TY1), shaded in gray, indicates the results used to derive a quota of 8 independent pumas, representing 15% of the independent pumas (from Logan 2009). Harvest Level No harvest. Year RY4 RY5 TY1 Projected Minimum Puma Population Size Adult Subadult Female Male Female Male 16 8 5 4 18 10 9 8 23 14 8 8 Independent Pumas Cub 20 33 42 Total 33 45 53 Lambda 1.37 1.17 Table 2. Pumas harvested by sport-hunters in Treatment Year 4 (TY4) on the Uncompahgre Plateau Study Area, Colorado, November 19 to December 29, 2012. Puma sex Age (yr.) F M M F M 2.5 1.5 2.5 2.5 2.5 Previous M/F I.D. or specimen P no. if not marked P1058 M156 P1066 F152 M179 Date of kill Location/UTM NAD27 Zone, Easting, Northing 12/10/2012 12/10/2012 12/21/2012 12/23/2012 12/29/2012 12, 756299, 4250598 12, 753851, 4249709 12, 741039, 4236392 13, 239123, 4248299 12, 759988, 4250158 Hunter/status Mark Rackay/resident Dustin Gleason/resident Mia Enstrom/resident Jared Roberts/resident Gary Gleason/resident Table 3. Four other independent VHF/GPS-collared pumas in the minimum count November 2012 to April 2013 (TY4) for the Uncompahgre Plateau Study Area that also died during November to April 2012-2013 period coinciding with the Colorado puma hunting season. Puma sex (M or F) F93 Age (yr.) 11 Date of kill/death 11/11/2012 F149 1.7 12/31/2012 F177 2.5 12/10/2012 M178 3 12/11/2012 Place of kill/UTM NAD27 Zone, Easting, Northing Linscott Canyon on study area 12, 761904, 4253939 GMU 70W, Dry Creek, south of Naturita, CO 12, 713658, 4229703 GMU 65, Tommy Creek fork of Cow Creek, 13, 263944, 4233691 GMU 65, Uncompahgre River, trailed off of study area at McKenzie Buttes, 13, 258413, 4239129 Hunter/status/other cause Killed by another puma Duane Pool/nonresident/Bobby Starks Outfitter Scott Hill/resident/Matt Iverson Outfitter Michael Delfino/nonresident/Ben Harris-Needle Rock Outfitter �Table 4. Minimum count of pumas based on numbers of known radio-collared pumas, visual observations of non-marked pumas, harvested non-marked pumas, and track counts of suspected non-marked pumas on the study area during September 2009 to April 2010 of Treatment Year 1 (TY1), November 2010 to April 2011 (TY2), November 2011 to April 2012 (TY3), and November 2012 to April 2013 (TY4), Uncompahgre Plateau study area, Colorado. Treatment Year (TY) Study Area region TY1 East slope West slope subtotals TY2 TY3 TY4 Adults Female Male Subadults Female Male Female Cubs Male 16 10 1 1 1 4 14 10 0 3 3 3 30 20 1 4 4 7 Total Independent Pumas = 55, including 31 females, 24 males. Cubs = 20-25 East slope 15 5 3 2 7 9 West slope 15 7 2 3 2 5 subtotals 30 12 5 5 9 14 Total Independent Pumas = 52, including 35 females, 17 males. Cubs = 39 East slope 13 4 1 3 4 2 West slope 14 5 3 5 1 2 subtotals 27 9 4 8 5 4 Total Independent Pumas = 48, including 31 females, 17 males. Cubs = 19 East slope 15 4 3 2 4 4 West slope 10 5 3 0 2 5 subtotals 25 9 6 2 6 9 Total Independent Pumas = 42, including 31 females, 11 males. Cubs = 24 Unknown sex 4-8* 5-6 9-14 7 9 16 4 6 10 3 6 9 *One adult non-marked female puma was killed by a hunter in Roubideau Canyon. The female puma was lactating, indicating she had nurslings. Up to 4 cubs were assumed to be in the litter. Table 5. Pumas captured and released by sport-hunters in Treatment Year 4 (TY4) on the Uncompahgre Plateau Study Area, Colorado, November 19 to December 29, 2012. Data are from puma hunter responses in 42 original voluntary surveys on printed permits. Total response rate from 70 individual permitted hunters was 60% (42/70 = 0.60*100). Puma sex/age stage/mark M/subadult/ eartags Date of capture 12/23/2012 Capture location Hunter name Sim’s Mesa Jeremiah Wheeler Reason for releasing the puma given by hunter Hunter thought the puma was a female. Yellow ear-tag indicated male puma. Number on ear-tag not noted by hunter. �Table 6. Summary of puma capture efforts with dogs from January 1, 2013 to April 18, 2013, Uncompahgre Plateau, Colorado. Month January No. Search Days 23 No. & type of puma tracks founda,b 93 tracks: 13 male, 50 female, 19 cub, 11 undetermined independent pumas Tracks ≤1 day old: 5 male, 22 female, 10 cub No. & type of pumas pursued 32 pursuits: 5 male, 18 female, 9 cub February 23 69 tracks: 15 male, 34 female, 14 cub, 6 undetermined independent puma Tracks ≤1 day old: 12 male, 27 female, 14 cub, 2 undetermined independent puma 29 pursuits: 9 male, 11 female, 7 cub, 2 undetermined independent puma March 21 No. & I.D. or type of pumas captured, observed, or identified 12 pumas captured 19 times: M180, M190 (twice), F129, F181, F136 (twice), F137 (4 times), F194, F74 (twice), PM1067 (twice; cub of F171; bio-darted; not handled in dangerous trees), M191 (probably cub of F28), M192 and M193 (cubs of F118). In addition, adult females F74 , F136, F137, F171 and subadult female F194 were associated with tracks by VHF telemetry. 13 pumas captured 13 times: M183, M196, PM1072 (bio-darted; not handled in dangerous trees), F182, PF1070 (bio-darted; not handled in dangerous trees), PF1071 (bio-darted; not handled in dangerous trees), cub unknown sex of PF1071 bayed on dangerous ledge but could not be handled safely, F136, F171, F197, F28 (nonfunctional collar; bio-darted; not handled in dangerous trees), F195 (cub of F118), M192 (cub of F118). In addition, adult females F140 (2 times), F74, F111 and subadult female F181 were associated with tracks by VHF telemetry. 4 pumas captured 4 times: PM1072 (not handled in dangerous trees), F137, F28 (non-functional collar; not handled in dangerous trees) F184 (cub of F111), P1073 (sex undetermined; hair collected, escaped). In addition, adult male M190 was associated with tracks by VHF telemetry. 4 pumas captured 5 times: F111, PF1074 (biodarted; not handled in dangerous trees), M198 (twice; cub of PF1074), F199 (cub of PF1074). In addition, subadult female F186 was associated with tracks by VHF telemetry. 29 individual pumas were captured 42 times with aid of dogs. In addition, 10 radio-collared pumas were detected 12 times by tracks and identified with VHF telemetry ≤1 km from the tracks. 12 independent pumas (adults, subadults) were captured with dogs for the first time (refer to Tables 7 and 8). 48 tracks: 12 male, 11 pursuits: 4 male, 24 female, 11 cub, 1 4 female, 3 cub undetermined independent puma Tracks ≤1 day old: 5 male, 7 female, 3 cub April 7 19 tracks: 1 male, 10 pursuits: 8 female, 10 cub 4 female, 6 cub Tracks ≤1 day old: 0 male, 8 female, 10 cub 74 229 tracks: 82 pursuits: TOTALS 41 male, 18 male, 116 female, 37 female, 54 cub, 25 cub 18 undetermined 2 undetermined Tracks ≤1 day old: 23 male 64 female 37 cub 2 undetermined a Puma hind-foot tracks with plantar pad widths >50 mm wide are assumed to be male; ≤50 mm are assumed to be female (Logan and Sweanor 2001:399-412). b Each capture season researchers also recorded instances when the first puma tracks ≤1 day old were encountered on each search route each day to gather data on vulnerability to detection using methods similar to puma hunters. For 2012-2013 (TY4) the count was: 46 tracks of females, including 9 of those associated with cubs; 23 tracks of males; and 1 track of undetermined sex. �Table 7. Adult and subadult pumas captured for the first time, sampled, tagged, and released from October 2012 to April 2013, Uncompahgre Plateau, Colorado. Puma I.D. F176 F177 M178 M179 M180 F181 F182 M183 F186 M190 F194 M196 F197 Sex Estimated Age (mo.) 27 28 29 25 18 21 48 54 29 36 26 45 18 F F M M M F F M F M F M F Mass (kg) 42 44 65 54 45 36 55 72 38 57 40 78 49 Capture date 10/17/2012 10/27/2012 11/13/2012 11/18/2012 1/1/2013 1/15/2013 2/4/2013 2/14/2013 3/30/2013 1/1/2013 1/29/2013 2/5/2013 2/14/2013 Capture method Cage trap Cage trap Cage trap Cage trap Dogs Dogs Dogs Dogs Cage trap Dogs Dogs Dogs Dogs Location North of Norwood Hill, San Miguel Canyon North McKenzie Mesa North McKenzie Mesa East rim Dry Creek Basin Dolores Creek Happy Creek Fisher Canyon Roubideau Canyon 7N Mesa, Roubideau Canyon San Miguel Canyon San Miguel Canyon San Miguel Canyon San Miguel Canyon Table 8. Pumas that were captured and observed with aid of dogs, some of which were biopsy-darted or hair was collected and given specimen numbers (e.g., PM1067, M for male, F for female), but were not handled at that time for safety reasons, and a puma killed by a bobcat hunter, January 2013 to March 2013, Uncompahgre Plateau, Colorado. Puma sex & I.D. PM1067 PF1069 Age stage or months 17 18 Capture date 1/25/2013 1/11/2013 PF1070 Adult 2/11/2013 PF1071 Adult 2/25/2013 PM1072 Adult 2/27/2013 P1073 6 3/15/2013 PF1074 Adult 4/12/2013 Location Horsefly Creek Lower Maverick Draw North Fork Cottonwood Creek Potter Creek, Roubideau Canyon North Fork Cottonwood Creek Monitor Creek, Roubideau Canyon Craig Draw Comments Cub of F171, sibling of M170. Puma shot by a bobcat hunter that thought the cat was a bobcat. Puma not previously marked. Mother of 3 cubs ~5-6 mo. old; one of which was bayed on a ledge but not handled. Mother of 1 male cub ~18 mo. old (not handled). Puma naturally marked with abbreviated right pinna with 2 notches and left nostril pad removed. Puma cub was mauled by dogs and escaped. Hair left at scene was collected. Mother of cubs M198, F199. �Table 9. Pumas recaptured October 2012 to April 2013, Uncompahgre Plateau, Colorado. Puma I.D. F93 Recapture Date 10/3/2012 Mass (kg) 39 Estimated Age (mo.) 132 Capture Method/ Location Cage trap/Happy Canyon Dogs/Horsefly Creek Dogs/McKenzie Creek F129 1/2/2013 1/17/2013 43 Observed 28 53 F136 1/18/2013 Observed 53 2/7/2013 50 54 1/4/2013 Observed 48 1/9/2013 Observed 48 Dogs/Caterwauler Canyon, SE Loghill Mesa Dogs/south rim Loghill Mesa Dogs/West Fork Dry Creek Dogs/Piney Creek 1/13/2013 Observed 48 Dogs/Dry Creek Forks 1/31/2013 Observed 48 Dogs/Dry Creek 3/6/2013 Observed 48 Dogs/Lower Dry Creek 1/15/2013 34 65 1/30/2013 Observed 65 2/5/2013 2/21/2013 Observed Observed 43 120 3/1/2013 Observed 121 Dogs/Lower Clay Creek Dogs/Lower Cottonwood Creek Dogs/Horsefly Creek Dogs/East Fork Big Bucktail Canyon Dogs/North Fork Cottonwood Creek F95 3/14/2013 40 67 F111 M190 4/12/2013 1/1/2013 35 Observed 60 36 PM1067 1/29/2013 Observed 17 M192 PM1072 2/1/2013 3/12/2013 Observed Observed 7 Adult M198 4/18/2013 Observed 9 F137 F74 F171 F28 Cage trap/Roubideau Canyon Dogs/Piney Creek Dogs/San Miguel Canyon Dogs/Horsefly Creek Dogs/Mailbox Park Dogs/Big Bucktail Canyon Dogs/upper Horsefly Creek Process Replaced VHF collar with GPS collar. Fit with GPS collar. F136 climbed dangerous trees; not handled. F136 climbed dangerous trees; not handled. Replaced non-functional GPS collar with a new one. F137 climbed dangerous tree; not handled. F137 climbed dangerous tree; not handled. F137 climbed dangerous tree; not handled. F137 climbed dangerous tree; not handled. F137 climbed dangerous tree; bio-darted for tissue sample, but not handled. F74 fit with new radiocollar. None. None. F28 climbed dangerous tree; not handled to replace non-functional GPS collar. F28 climbed dangerous tree; bio-darted for tissue sample, but not handled to replace non-functional GPS collar. Replaced VHF collar with GPS collar. Replaced GPS collar. M190 took refuge in dangerous ledges; not handled. PM1067 climbed dangerous tree; not handled. None. PM1072 climbed dangerous tree; not handled. None. �Table 10. Summary of puma capture efforts with cage traps from September 18, 2012 to May 22, 2013, Uncompahgre Plateau, Colorado.* Month September No. of Sites 4 Carnivore activity & capture effort results Female puma walked ~5-10 m from mule deer bait 6 days old, but did not feed, East McKenzie Mesa bait site. Black bears, bobcats, coyotes scavenged some mule deer carcasses. October 13 Adult puma F93 recaptured in cage trap baited with mule deer 10/3/2012. Adult puma F176 captured for the first time in cage trap baited with mule deer 10/17/2012. Adult puma F177 captured for first time in cage trap baited with mule deer 10/27/2012. Non-marked female puma scavenged mule deer bait at SE Loghill Mesa rim 10/30/2012; set cage trap; but, puma did not return. Adult puma F136 walked past same mule deer bait, 8 days old, at SE Loghill Mesa Rim on 10/31/2012, but did not feed. Unknown puma walked past mule deer bait in mouth of Clay Creek, but did not feed. Bobcats and gray foxes scavenged from some of the mule deer carcasses. November 7 Adult puma M178 captured for first time in cage trap baited with mule deer 11/13/2012. Adult puma M179 captured for first time in cage trap baited with mule deer 11/18/2012. Non-marked female puma scavenged mule deer bait at SE Loghill Mesa rim 11/5/2012; set cage trap; but, puma did not return (probably same as in October). Puma M178 walked by mule deer bait 5 days old on SE Loghill Mesa 11/10/2012, but did not feed. March 14 Adult puma F95 recaptured in cage trap baited with mule deer 3/14/2013. Puma cub F185 captured for the first time in cage trap baited with mule deer 3/23/2013. Subadult puma F186 captured for the first time in a cage trap baited with mule deer 3/30/2013. Puma, probably F95, fed on mule deer bait on east Roubideau Canyon rim, cage trap set; puma did not return. Adult puma M183 visited mule deer bait on 7N Mesa, but did not feed. Female puma, probably F118, walked ~3 m from mule deer bait on N Norwood Hill, but did not feed. Coyotes, bobcats and black bear scavenged some of the mule deer carcasses. April 3 Adult F171 fed on elk bait at Horsefly Canyon on 4/23/2013; no capture effort needed. Black bears and coyotes scavenged on elk bait. May 2 Black bears scavenged mule deer baits. * We used 50 road-killed mule deer and 1 road-killed elk at 28 different sites. Of the road-killed baits, 12 of 51 (23.53%) were scavenged by pumas. Table 11. Puma cubs sampled August 2012 to July 2013 on the Uncompahgre Plateau Puma Study area, Colorado. Cub I.D. Sex Estimated birth datea Estimated age at capture (days) Mass (kg) Mother Estimated age of mother at birth of this litter (mo) PF1062 F 6/2012 183 13.5 Nonmarked Adult M166 M 7/5/2012 33 2.2 F136 51 M167 M 7/5/2012 33 2.1 M168 M 7/27/2012 37 2.3 F169 F 7/27/2012 37 2.2 F96 78 F173 F 7/27/2012 37 2.5 M174 M 8/8/2012 32 1.9 F152 25.7 M175 M 8/8/2012 32 1.8 F184 F 8/25/2012 208 13.0 F111 58 F185 F 9/2012 183 12.0 Nonmarked Adult F187 F 5/14/2013 31 2.3 F96 88 F188 F 5/14/2013 31 2.5 b M191 M 7/2012 183 14.0 F28 probably 112 PM1068 M 7/2012 183 Unknown M192 M 6/20/2012 199 21.0 M193 M 6/20/2012 199 20.0 F118 50 F195 F 6/20/2012 227 20.0 M198 M 6/2012 274 30 PF1074 Adult F199 F 6/2012 282 25 F189 F 6/18/2013 38 2.6 F200 F 6/18/2013 38 2.6 F136 62 M201 M 6/18/2013 38 2.8 F202 F 6/25/2013 35 2.5 F172 48 a Estimated age of cubs sampled at nurseries is based on the starting date for GPS location and radio-telemetry foci for mothers at nurseries, and development characteristics of cubs caught with mothers without radiocollars or mothers with non-functioning radiocollars. b F28 had a non-functional GPS collar, but recapture sites and tracked travel routes were consistent with associations with cubs M191 and PM1068. Another non-marked cub was in association, making the total number of cubs = 3. �Table 12. Summary of puma capture efforts with dogs, December 2004 to April 2013, Uncompahgre Plateau, Colorado. Period Dec. 2, 2004 to May 12, 2005 Nov. 21, 2005 to May 26, 2006 Nov. 13, 2006 to May 11, 2007 Nov. 19, 2007 to April 24, 2008 Dec. 9, 2008 to April 30, 2009 Track detection effort 109/78 = 1.40 tracks/day 35/78 = 0.45 pursuit/day Puma capture effort 14/78 = 0.18 capture/day Effort to capture an independent puma for the first time 11 pumas captured for first time 11/78 = 0.14 capture/day 78/14 = 5.57 day/capture 14/82 = 0.17 capture/day 78/11 = 7.09 day/capture 149/82 = 1.82 tracks/day 78/35 = 2.23 day/pursuit 43/82 = 0.52 pursuit/day 177/78 to 182/78 = 2.27-2.33 tracks/day 82/43 = 1.91 day/pursuit 45/78 to 47/78 = 0.58-0.60 pursuit/day 82/14 = 5.86 day/capture 22/78 = 0.28 capture/day 78/47 to 78/45 = 1.66-1.73 day/pursuit 49/77 = 0.64 pursuit/day 78/22 = 3.54 day/capture 78/7 = 11.14 day/capture 20/77 = 0.26 capture/day 7 pumas captured for first time 7/77 = 0.09 capture/day 77/20 = 3.85 day/capture 24/71 = 0.34 capture/day 77/7 = 11.00 day/capture 217/77 to 218/77 = 2.82-2.83 tracks/day Pursuit effort 198/71 to 202/71 = 2.79-2.84 tracks/day 77/49 = 1.57 day/pursuit 75/71 to 78/71 = 1.06-1.10 pursuit/day Dec. 15, 2009 to April 30, 2010 Nov. 16 and Dec. 14, 2010 to April 22, 2011 266/86 = 3.09 tracks/day 71/75 to 71/78 = 0.91-0.95 day/pursuit 93/86 = 1.08 pursuit/day 300/81 = 3.70 tracks/day Dec. 27, 2011 to April 12, 2012 268/79 = 3.39 tracks/day Jan. 1, 2013 to April 18, 2013 229/74 = 3.09 tracks/day 7 pumas captured for first time 7/82 = 0.08 capture/day 82/7 = 11.71 day/capture 7 pumas captured for first time 7/78 = 0.09 capture/day 9 pumas captured for first time 9/71 = 0.13 capture/day 71/24 = 2.96 day/capture 71/9 = 7.89 day/capture 26/86 = 0.30 capture/day 9 pumas captured for first time 9/86 = 0.11 capture/day 86/93 = 0.92 day/pursuit 99/81 = 1.22 pursuit/day 86/26 = 3.31 day/capture 52/81 = 0.64 capture/day 86/9 = 9.56 day/capture 81/99 = 0.82 day/pursuit 81/52 = 1.56 day/capture 81/15 = 5.40 day/capture 89/79 = 1.13 pursuit/day 26/79 = 0.28 capture/day 11 pumas captured for first time 11/79 = 0.14 capture/day 79/89 = 0.89 day/pursuit 79/26 = 3.04 day/capture 79/11 = 7.18 day/capture 82/74 = 1.11 pursuit/day 42/74 = 0.57 capture/day 12 pumas captured for the first time 12/74 = 0.16 capture/day 74/82 = 0.90 day/pursuit 74/42 = 1.76 day/capture 74/12 = 6.17 day/capture 15 pumas captured for first time 15/81 = 0.18 capture/day �Table 13. Individual puma reproduction histories, Uncompahgre Plateau, Colorado, 2005-2013. Consort pairs and estimated agesa Female Age (mo.) Male Age (mo.) F2 F2 F2 F3 F3 F3 F3 F3 F7 F7 F7 F8*e F8 F8 F8 F16 F16 F16 F23* F23 53 67 89 36 50 62 84 107 67 82 106 24 37 60 95 32 52 75 21 45 F23 80 F24 F24 75 114 F25 F25 F25 F25 74 94 110 129 F28* F28 F28 F28 F30* F50 F54 F70* F70 F70 F72* F72 F72 36 48 68 112 48 21 24 38 52 76 28 51 64 F75 F75 F93 F93 F94* F94 F96 F96 F96 32 55 56 90 46 60 55 78 88 Dates pairs consortedb Estimated birth datec M73 49 02/28-29/08 M6 80 01/13-14/09 M27 or M29f M67 78 107 53 02/19-25/08 05/28/05 07/29/06 05/19/08 08/01/04 09/26/05 09/17/06 07/03/08 06/28/10 05/19/05 08/13/06 07/10/08 06/26/05 08/13/06 05/29/08 04/18/11 09/22/05 05/24/07 04/15/09 05/30/06 05/23/08 01/28-31/11 04/22/11 M29 92 04/12-15/07 06/14/07 09/10 M6 37 06/22-24/05 M51 M55 60 69 03/31/08 03/28-31/10 08/01/05 04/16/07 08/19/08 3/10 M29 88 12/27-29/06 M55 34 04/16-20/07 M51 60 03/10/08 M73 61 02/11/09 M55 M55 70 71 04/15/10 05/21/10 06/09/06 03/30/07 11/08 07/12 07/17/07 07/01/06 07/01/06 06/05/08 08/31/09 08/18/11 07/09/08 06/12/10 07/15/11 08/07 05/07/09 08/07 06/16/10 05/27/09 07/15/10 08/21/10 07/27/12 05/14/13 Estimated birth interval (mo.) Estimated gestation (days) Observed number of cubsd 19.9 22.7 91-92 23.8 87-93 3 2 4 1 2 3 3 2 2 4 3 2 4 2 2 4 4 3 3 3 Nonfunct.GPS 84-86 2 90-93 4 3 14.0 22.0 13.8 11.7 21.5 23.8 93-95 94 89-92 14.9 23.9 13.4 22.5 34.7 90-91 Nonfunct.GPS 1 1 2 3 20.5 16.1 Nonfunct.GPS 11.7 92-93 88-92 87 14.8 23.6 23.1 13 23.2 93 13.3 91 23.2 9.6 2 ≥2 tracks 1 3 3 1 1 3 3 3 1 2 3 photographed 1 2 2 2 3 3 4 3 2 �Consort pairs and estimated agesa Female Age (mo.) Male Age (mo.) Table 13 continued. Dates pairs Estimated consortedb birth datec Estimated birth interval (mo.) F104 F111* F111 F116g F118 F118h F119 F119i 110 32 58 36 27 50 66 96 expected 07/08/10 06/16/10 08/25/12 2009 08/08/10 06/20/2012 08/09 02/12 expected 29 expected F135 F136j F136 F136 33 39 51 62 07/06/11 07/10/11 07/05/12 06/18/13 12 11 Nonmarkedl Unk. 03/19/13 Estimated gestation (days) Observed number of cubsd 92 3 2 2k 2 3 3 2 1 plus 1-2 uterine scars 2 ≥1 remains 2 3 26.3 22.4 F137 30 07/08/11 ≥1 F137 54 07/12/2013 3 F152* 25.7 08/08/2012 2 F171 22 08/11 2 F171 45 07/31/2013 4 F172 48 06/25/2013 1 a Ages of females were estimated at litter birth dates. Ages of males were estimated around the dates the pairs consorted. b Consort pairs indicate pumas that were observed together based on GPS data or VHF location data. c Estimated birth dates were indicated by GPS data of mothers at nurseries or by back-aging cubs to approximate birth date. d Observed number of cubs do not represent litter sizes as some cubs were observed when they were 5 to 16 months old after postnatal mortality could have occurred in siblings. Only cub tracks were observed with F28. e Asterisk (*) indicates first probable litter of the female, based on known history or nipple characteristics noted at first capture of the female. f A radio-collared, ear-tagged male puma was visually observed with F23 on 2/25/08. Both M27 and M29 wore non-functional GPS collars in that area at the time. g When captured on 1/20/10, puma F116 was in association with 2 large cubs which were not captured. h Two cubs observed with F118 south of Norwood 9/24/2012. i F119 died of a ruptured uterus and internal bleeding on 1/28/12. Cub in uterus in third trimester; 1-2 uterine scars indicated expulsion of 1-2 fetuses. j Remains of F136’s cubs found 8/9/11. Cause of death predation by puma or black bear. k Tracks evidence of one other cub in association with F111 and cub F184, but not captured and marked. l A non-marked adult male puma was photographed consorting with adult female pumas F136 and F182 at the same time on the NE rim of Loghill Mesa on 03/19-20/13. �Table 14. Summary for individual adult puma survival and mortality, December 8, 2004 to July 31, 2013, Uncompahgre Plateau, Colorado. Puma I.D. M1 Monitoring span 12-08-04 to 08-16-06 M4 M5 01-28-05 to 12-28-05 08-01-06 to 02-20-09 M6 02-18-05 to 05-21-10 M27 03-10-06 to 05-07-09 M29 04-14-06 to 02-25-09 M32 04-26-06 to 12-02-10 M51 01-07-07 to 03-20-09 M55 01-21-07 to 07-31-10 M67 08-23-07 to 12-18-11 M71 01-29-08 to 11-12-09 M73 02-21-08 to 10-26-11 M87 02-09-11 to 12-06-11 M90 11-16-10 to 11-23-10 M100 03-27-09 to 07-31-09 M114 02-27-10 to 03-10-12 M133 11-12-10 to 12-01-10 Status: Alive/Lost contact/Dead; Cause of death Dead. Lost contact− failed GPS/VHF collar. M1 ranged principally north of the study area as far as Unaweep Canyon. M1 was killed by a puma hunter on 01-02-10 west of Bang’s Canyon, north of Unaweep Canyon, GMU 40. M1 was about 97 months old at death. Dead; killed by a male puma. Estimated age at death 37−45 months. Dead. Born on study area; offspring of F3. M5 was independent of F3 by 13 months old, and dispersed from his natal area at about 14 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months (protected from hunting mortality in buffer area) and ranged into the eastern edge of Utah (vulnerable to hunting). Killed by a puma hunter on 02-20-09 in Beaver Creek, Utah at age 54 months. Dead. M6 was struck and killed by a vehicle on highway 550 south of Colona, CO on 05-21-10. M6 was about 99 months old at death. Dead. Lost contact− failed GPS/VHF collar. Recaptured 12-02-07 & 01-22-08 by puma hunter/outfitter north of the study area. Possibly visually observed on study area with F23 on 02-25-08. Recaptured by a puma hunter/outfitter 12-11-08 & 12-28-08 north of the study area. Photographed by a trail camera on the study area (Big Bucktail Canyon) on 5 occasions: 03-27-09, 04-02-09, 04-15-09, 04-24-09, & 05-07-09. M27 was killed by a puma hunter on 12-09-09 in the North Fork Mesa Creek, Uncompahgre Plateau, GMU 61 North. M27 was about 100 months old at death. Dead. Lost contact− failed GPS/VHF collar. Possibly visually observed on study area with F23 on 02-25-08. Recaptured on study area 02-25-09, but could not be safely handled to change faulty GPS collar. M29 was killed by a puma hunter on 11-16-09 in Beaver Canyon, GMU 70 East. M29 was about 121 months old at death. Dead. Killed by a puma hunter on 12-02-10 in McKenzie Creek on the Uncompahgre Plateau study area. M32 was about 112 months old at death. Dead. Lost contact− failed GPS/VHF collar after 03-20-09. Killed by a puma hunter on 12-11-09 in Shavano Valley, Uncompahgre Plateau study area. M51 was about 77 months old at death. Dead. Killed by a puma hunter on 11-25-10 in Spring Creek Canyon on the Uncompahgre Plateau study area. M55 was about 77 months old at death. Dead. M67 is offspring of F30. Dispersed natal area. Established territory on W side U.P. study area. Killed by a puma hunter in Tabaguache Creek 12-18-2011 at age 52.9 months. Dead. Lost contact– M71 shed his VHF collar with an expansion link on about 11-1209. He was killed by a puma hunter on 12-09-09 on the west rim of Spring Creek Canyon, Uncompahgre Plateau study area. M71 was about 47 months old at death. Dead. Illegally killed 10-26-2011 in Bear Pen Gulch, upper East Fork Escalante Canyon; shot through abdomen during second rifle season. M73 was about 80 months old at death. Dead. M87 is offspring of F3. Dispersed from natal area. Established territory on W side of U.P. study area. Killed by a puma hunter in 47 Canyon, Tabaguache Canyon 12-06-2011. M87 was 41 months old at death. Dead. M90 was killed by a hunter on 11-23-10 on McKenzie Butte. M90 was offspring of F72, born 07-09-08. He was 28 months old at death. Dead. M100 was killed by a puma hunter on 01-16-10 in Naturita Canyon, GMU 70 East. M100 was about 63 months old at death. Dispersed from U.P. study area after 06-23-10. Killed by a puma hunter in Beaver Creek, NE of Canyon City, GMU59, 03-10-12. M114 was about 55 months old at death. Dead. M133 was killed by a puma hunter on 12-01-10 in Dry Fork Escalante Canyon north of the study area. M133 was about 43 months old at death. �Puma I.D. M134 Monitoring span 06-01-11 to 06-10-11 M138 07-01-11 to 12-23-11 M144 09-01-11 to 02-25-13 M153 M165 09-01-11 to 09-13-11 07-01-12 to 02-17-12 M178 11-13-12 to 12-11-12 M179 M180 M183 M190 M196 F2 11-18-12 to 12-29-12 07-01-13 to 07-31-13 02-14-13 to 07-31-13 01-02-13 to 07-31-13 02-05-13 to 07-31-13 01-07-05 to 08-14-08 F3 01-21-05 to 12-11-11 F7 02-24-05 to 08-03-08 F8 03-21-05 to 12-17-12 F16 10-11-05 to 09-11-09 F23 02-05-06 to 06-06-12 F24 01-17-06 to 07-31-11 F25 02-08-06 to 02-03-11 F28 03-23-06 to 02-16-12 F30 04-15-06 to 07-29-08 F50 12-14-06 to 03-26-07 F54 01-12-07 to 08-18-07 F70 01-14-08 to 12-22-11 F72 02-12-08 to 12-21-11 F74 01-15-13 to 5-16-13 Table 14. Continued. Status: Alive/Lost contact/Dead; Cause of death Dead. M134 was offspring of unmarked female puma in Roubideau Canyon. Independent by about 03-28-11. Shot dead by USDA, APHIS, WS agent while in the act of attacking domestic sheep on 06-10-11 when he was 24 months old at start of adult life stage. Dead. Killed by a puma hunter in Horsefly Canyon (E) 12/23/11. M138 was about 29 months old at death. Dead. Initially captured as 18 mo. old subadult on W side U.P. study area 03-07-11. Dispersed from study area. Established adult territory on NW U.P. Killed by puma hunter 2-25-2013 in GMU 40, North Fork West Creek, Unaweep Canyon. Dead. Killed for depredation control; killed an alpaca in Pleasant Valley 09-13-11. Alive. Initially captured as 19 mo. old subadult on W side U.P. study area 02-24-12. Moved to Escalante Creek drainage by adult age 07-31-12. Killed by puma hunter 1217-2012 in GMU 62N, Dry Fork Escalante Canyon. Dead. Originally captured on the study area 11-13-12. Killed by puma hunter 12-1112 after tracking M178 off the study area and onto adjacent GMU 65. Dead. Killed by puma hunter on study area 12-29-12. Alive. Alive. Alive. Alive. Dead; killed by another puma (sex of puma unknown; male suspected) 08-14-08. F2 was about 92 months old at death. Dead. Killed by a puma hunter in Lindsay Creek 12-11-11. F3 was about 120 months old at death. Dead. Killed by U.S. Wildlife.Services agent 08-03-08 for predator control of depredation on domestic sheep. F7 was about 107 months old at death. Lost radio contact. Last live signal heard 12/17/2012 in Big Bucktail Canyon on study area. Fate unknown; was not recaptured on study area Jan. to April 2013. Dead. F16 was struck and killed by a vehicle on Ouray County Road 1 southwest of Colona, CO on 09-11-09. F16 was about 80 months old at death. Dead. Killed by a male puma about 06-06-12. F23 was about 94 months old at death. F23 may have attempted to defend 2 cubs (F149, M161; 13.5 months old) and/or calf elk kill. Dead. Killed by a male puma in Logging Camp Draw about 09-16-11. F24 was about 126 months old at death. F24 may have attempted to defend ≥2 cubs (F147, nonmarked siblings; 12 mo. old). Dead. Lost radio contact after 09-04-09– failed GPS/VHF collar. Photographed alive with three ~9 month old cubs on 12-03-10 on Loghill Mesa. F25 shot dead by a ranch hand on 02-03-11 in Pleasant Valley, Dallas Creek because she was seen among cattle. F25 was about 138 months old at death and in excellent physical condition (49 kg). Lost radio contact after 09-25-07− failed GPS/VHF collar. Recaptured F28 on the study area 02-01-10 and 01-01-11 and 02-16-12, but could not be handled to replace non-functional GPS collar. Dead. Killed by another puma (sex of puma unknown) 07-29-08. F30 was about 60 months old at death. Dead of natural causes 03-26-07; probably injury or illness-related; exact agent unknown. F50 was about 30 months old at death. Dead; killed by a male puma while in direct competition for prey (i.e., mule deer fawn) 08-18-07. F54 was about 49 months old at death. Dead. Killed by a puma hunter Spring Creek 12-22-11. F70 was 80 months old at death. Her death orphaned 2 cubs, F157 and F158, at 4 months old; both starved to death about 01-15-12 at about 5 months old. Lost radio contact after 12-02-10. F72 recaptured in Fisher Creek on 03-18-11, but could not be handled to replace non-functional GPS collar. Photographed on Miller Mesa S of U.P. study area on 12-18 to 21-11 with 3 new cubs born about July 2012. Lost radio contact after 5-16-13; radiocollar fell off after canvas breakaway tab broke; detected 6-10-13. �Puma I.D. F75 Monitoring span 03-26-08 to 12-13-11 F93 F94 12-05-08 to 11-11-12 12-19-08 to 02-01-11 F95 F96 F104 08-01-09 to 07-31-13 01-28-09 to 07-31-13 05-21-09 to 01-31-12 F110 09-21-09 to 02-25-10 F111 F113 01-01-10 to 07-31-13 01-26-10 to 06-06-10 F116 01-20-10 to 09-20-11 F118 F119 02-25-10 to 07-31-13 03-25-10 to 01-28-12 F135 01-01-11 to 09-20-11 F136 F137 F140 F143 F152 F163 F171 F172 F176 F177 F181 F182 F186 F194 01-20-11 to 07-31-13 01-21-11 to 07-31-13 08-01-12 to 07-31-13 02-15-11 to 07-31-13 06-16-12 to 12-23-12 07-01-12 to 07-31-13 01-20-12 to 07-31-13 03-28-12 to 07-31-13 10-17-12 to 07-31-13 10-27-12 to 12-10-12 04-01-13 to 07-31-13 02-04-13 to 07-31-13 03-30-13 to 07-31-13 01-29-13 to 06-17-13 Table 14 continued. Status: Alive/Lost contact/Dead; Cause of death Dead. Killed by a puma hunter in North Fork Cottonwood Creek 12-13-11. F75 was about 98 months old at death. Dead. Killed by another puma 11-11-12. Fatal bite wounds to the skull. Dead. Shot dead on 02-01-11 by USDA, APHIS, WS agent for predation on domestic elk in Happy Canyon. F94 was about 74 months old at death. Alive. Alive. Dead. Died probably of starvation associated with senescence in lower Roubideau Creek 01-31-12. F104 was about 132 months old at death. Dead. Killed by a puma hunter on 02-25-10 in GMU 70 East. F110 was about 41 months old at death. Alive. Dead. F113 died 06-06-10 of injuries consistent with being struck by a vehicle. GPS data indicated that F113 had crossed highway 550 and roads on Loghill Mesa north of Ridgway 24-30 hours before she died in McKenzie Creek. F113 was about 42 months old at death. Dead. Died about 09-20-11 of unknown natural cause associated with pregnancy and birth of new litter of cubs. F116 was about 60 months old at death. Alive. Dead. Died of ruptured uterus and internal bleeding associated with pregnancy in Clay Creek Canyon 01-28-12. F119 was about 95 months old at death. Dead. Died of unknown natural cause in E Fork Dry Creek 09-20-11. Her death orphaned cubs M154 and M155 at 76 days old; both died of starvation or disease when 77 (M154) and 81 (M155) days old. Alive. Alive. Alive. Alive. Dead. Killed by puma hunter on study area, Spring Creek Canyon. Alive. Alive. Alive. Alive. Dead. Killed by puma hunter 12-10-12 in GMU 65 adjacent to study area. Alive. Alive. Alive. Dispersed, exhibited subadult behavior. Fate unknown. Censor. �Table 15. Preliminary estimated survival rates (S) of adult-age pumas during the 4 years in the reference period (i.e., the study area is closed to puma hunting) and 4 years in the treatment period, Uncompahgre Plateau, Colorado. Survival rates of pumas estimated with the Kaplan-Meier procedure to staggered entry of animals (Pollock et al. 1989). Survival rates are for an annual survival period defined as the biological year (August 1 to July 31). Survival rates were estimated only for periods when n ≥ 5 individual pumas were monitored in the interval. Puma survival in the reference period pertained only to pumas that died of natural causes. Pumas that were killed by people in the reference period, a non-natural cause (i.e., two adult pumas: F7 for depredation control 8/3/2008 and M5 killed by a puma hunter off the protected study area and buffer zone 2/20/2009) were right censored. In the treatment period all sources of natural and human-caused mortality are considered in the survival estimates. Biological Year Reference Annual 2 8/1/2005 to 7/31/2006 Reference Annual 3 8/1/2006 to 7/31/2007 Reference Annual 4 8/1/2007 to 7/31/2008 Reference Annual 5 8/1/2008 to 7/31/2009 Treatment Annual 1 8/1/2009 to 7/31/2010 Treatment Annual 1b 8/1/2009 to 7/31/2010 With mortalities of all marked adult males Treatment Annual 2 8/1/2010 to 7/31/2011 Treatment Annual 3 8/1/2011 to 7/31/2012 Treatment Annual 4 8/1/2012 to 7/31/2013 a S 1.000 Females SE 0.0000 n 10 S 0.667a Males SE 0.2222a n 6a 0.909 0.0867 11 1.000 0.0000 5 0.831 0.0986 14 1.000 0.0000 7 0.875 0.1031 13 1.000 0.0000 8 0.784 0.1011 19 0.667 0.1924 8 NA (see rates above) NA NA 0.333b 0.1361b 12b 0.947c 0.0568 19 0.250 0.1082 9 0.548d 0.1063 20 0.167 0.1076 7d 0.819 0.0931 19 0.188 0.0845 8e Adult male annual S 2005 to 2006 is probably underestimated with poor precision because 3 of the 6 pumas were GPS/VHFmonitored for 4 to 5 months at the end of the interval; 1 of 6 adult males died. b This second estimate of adult male puma survival 8/1/2009 to 7/31/2010 includes 5 males that had non-functional (4) or shed (1) radiocollars. All adult males with non-functional or shed radiocollars in this study survived into treatment year 1 (TY1), which was expected considering adult male survival in 3 previous years. All 5 of those adult males were detected and killed by hunters in TY1. c Only 1 of 2 adult female puma mortalities is represented in this survival analysis for 8/1/2010 to 7/31/2011, that of F94 killed for depredation control. One other adult female mortality, F25, is not represented because she wore a non-functional GPS collar making it impossible for us to monitor her survival. F25 was shot by a ranch hand on 2/3/2011 when he saw her among cattle. d Sample included F143, F163, M144, ranged on NW Uncompahgre Plateau N of the study area but not on the U.P. study area, vulnerable to annual hunting. e Sample includes F143, F163, M144, M165 that ranged on north half of the Uncompahgre Plateau north of the study area (not on the study area) and were at risk to annual sport-hunting mortality. �Table 16. Summary of subadult puma survival and mortality, December 2004 to July 2013, Uncompahgre Plateau, Colorado. Puma I.D. M5 Monitoring span 09-16-05 to 06-30-06 No. days 308 M11 06-21-06 to 12-02-07 529 F23 01-04-06 to 02-04-06 31 M31 04-19-06 to 04-26-06 7 M49 03-26-07 to 10-01-07 189 F52 01-10-07 to 05-15-07 125 F66 08-23-07 to 11-05-07 11-25-08 to 06-03-09 74 Status Survived to adult stage. M5 was offspring of F3, born August 2004. Independent and dispersed from natal area at 13 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months (protected from hunting mortality in buffer area) and ranged into the eastern edge of Utah (vulnerable to hunting). Killed by a puma hunter on 02-20-09 in Beaver Creek, Utah at about 54 months old. Survived to adult stage. M11 was offspring of F2, born May 2005. Independent at 13 months old. Dispersed from natal area at 14 months old. Moved to Dolores River valley, CO, by 12-14-06. Killed by a puma hunter on 12-02-07 when about 30 months old. Survived to adult stage. Captured on the study area when about 17 months old. Survived to adult stage; gave birth to first litter at about 21 months old. Killed by a male puma about 06-06-12. F23 was about 94 months old at death. M69 01-11-08 to 04-07-08 190 87 Survived to adult stage. M31’s estimated age at capture was 20 months. Dispersed to northern New Mexico and was killed by a puma hunter on 12-11-08 in Middle Ponil Creek, Cimarron Range. He was about 52 months old. Survived to adult stage. M49 was offspring of F50, born July 2006. Orphaned at about 9 months old, when F50 died of natural causes. Dispersed from his natal area at about 10 months old and ranged on the northeast slope of the Uncompahgre Plateau. When M49 was about 15 months old, he shed his expandable radiocollar on about 10-01-07 at a yearling cow elk kill on the northeast slope of the Uncompahgre Plateau. He was killed by a puma hunter in Blue Creek in the protected buffer zone north of the study area on 01-24-09; he was about 29 months old, a young adult. Survived to adult stage. F52 dispersed from study area as a subadult by 01-16-07. F52’s last VHF aerial location was Crystal Creek, a tributary of the Gunnison River east of the Black Canyon 05-15-07. She was treed by puma hunters on 12-29-08 on east Huntsman Mesa, southeast of Powderhorn, CO. She was about 41-43 months old and could have been in her adult-stage home range. GPS collar nonfunctional. F52 was killed by a puma hunter on 01-09-12 in North Beaver Creek SE of Powederhorn, CO. She was about 79 months old at death. Died in subadult stage. F66 was offspring of F30, born July 2007. Lost contact; her cub collar quit after 11-05-07. Recaptured as an independent subadult on her natal area 11-25-08 when 16 months old. Mother F30 was killed by a puma when F66 was 12 months old, within the age range of normal independence. F66 died of injuries to internal organs that caused massive bleeding attributed to trampling by an elk or mule deer on about 05-28-09 when she was 23 months old. Her range partially overlapped her natal area. Survived to adult stage. M69 was captured on the study area when about 14-18 months old. Emigrated from the study area as subadult by 03-1908. Last VHF aerial location was southwest of Waterdog Peak, east side of Uncompahgre River Valley on 04-07-08. M69 was killed by a puma hunter on 11-06-08 in Pass Creek in the Snowy Range, WY when he was 24 to 28 months old. �Table 16 continued Puma I.D. F95 Monitoring span 12-29-08 to 07-31-12 No. days 214 M99 02-27-09 to 04-22-09 54 M112 02-10-11 to 04-18-11 67 M115 01-13-10 to 07-21-10 189 M120 12-06-11 1 M122 08-12-10 to 04-18-11 250 M131 09-25-10 to 04-18-11 206 M134 03-28-11 to 06-10-11 74 M138 01-26-11 to 06-30-11 155 F140 01-13-12 to 07- 31-12 200 M141 12-23-11 1 M144 03-07-11 to 09-08-11 185 F145 03-08-11 to 09-08-11 184 Status Alive. F95 is the offspring of F93, born about August 2007. She became an independent subadult by about 18 months old (02-11-09 aerial location) and an adult by about 24 month old (Aug. 2009). F95 established an adult home range adjacent to and overlapping the northern portion of her natal area. Died in subadult stage. M99 probably killed by another puma (canine punctures in skull including braincase) in Jan. 2010 when he was about 16 months old. His radiocollar quit after 54 days. M112 was offspring of F70 born August 2009. M112 associated with F96 and her two radio-collared cubs F129 and M130 during 02-10-11 to 04-18-11. Lost contact of M112 after 04-18-11. Dispersed. M112 was killed by a puma hunter 01-06-2013, GMU 73, SE of Dolores, CO; UTM: 12S, 732863E, 4146772N; age 41 months, adult stage. Died in subadult stage. M115 was offspring of F28, born in Nov. 2008. He was about 14 months old when first captured on Jan. 13, 2010. When he was recaptured on 03-18-10, he had previously suffered a broken left ulna. M115 was probably independent by 07-15-10 when he was located outside of his natal area on a probably dispersal move. M115 died on about 07-21-10 apparently from complications of his broken left foreleg; probably not allowing him to kill prey sufficiently for survival. M115 was about 20 months old at death. Died in subadult stage. M120 was offspring of F3. M120 was killed by a puma hunter 12-06-11 in his natal area in Spring Creek. He was 17 months old at death. M122 was offspring of F104, born 07-08-10. Lost contact after 04-1811 when radio-collar malfunctioned. Dispersed. Killed by puma hunter in GMU 62, Tatum Draw, Dry Fork Escalante Creek, N of natal area 01-23-13; UTM: 12S, 735353E, 4283455N; age 30 months, adult stage. M131 was offspring of F96, born 08-21-10. Lost contact after 04-18-11 when collar malfunctioned. Dispersed. Killed by puma hunter in GMU 60, Lion Creek, extreme W CO 01-17-13; UTM: 12S, 670829E, 4246980N; age 29 months old, adult stage. Survived to adult stage (barely). M134 was offspring of unmarked female puma in Roubideau Canyon. Independent by about 03-28-11. Shot dead by USDA, APHIS, WS agent while in the act of attacking domestic sheep on 06-10-11 when he was 24 months old at start of adult life stage. Survived to adult stage. Entered adult life stage 07-01-11. Killed by a puma hunter 12-23-11 in Horsefly Canyon. M138 was about 29 months old at death. Survived to adult stage. Turned adult in Aug. 2012. Probably offspring of F28. Has established a home range adjacent to natal area where she was initially captured at 5 months old on 01-02-11. Died in subadult stage. M141 was killed by a puma hunter on 12-23-11 in Little Bucktail Creek. He was 16 months old at death. Survived to adult stage. Emigrated from U.P. study area. Established adult territory on northwest Uncompahgre Plateau. M144 is sibling of F145 below. Killed by puma hunter 2/25/2013 at ~41 mo. old. Survived to adult stage. Emigrated from U.P. study area and to Colorado Mesa. Killed by a puma hunter 01-23-12 in West Bangs Canyon. F145 was 28 months old at death. �Table 16 continued Puma I.D. F146 Monitoring span 03-08-11 to 03-23-11 No. days 15 F147 09-16-11 to 04-12-12 209 F149 06-06-11 to 12-31-12 575 M150 03-28-11 to 04-11-11 14 F152 05-04-12 to 06-16-12 44 M153 04-12-11 to 09-06-11 147 M161 06-06-12 to 08-03-12 59 F163 01-26-12 to 07-01-12 157 M164 02-14-12 to 02-26-12 02-24-12 to 12-17-12 12 M165 M180 F181 F194 F197 01-01-13 to 07-01-13 01-15-13 to 07-01-13 01-29-13 to 6-17-13 02-13-13 to 07-01-13 298 182 Status Died in subadult stage. F146 was killed and eaten by a male puma while in competition for an adult bull elk carcass that one of the pumas killed in Coal Canyon on the study area. F146 was about 19 months old at death. Lost contact; radiocollar quit after 04-12-12. F147 orphaned at about 12 months old when her mother F24 was killed by a male puma on 09-1611. Died in subadult stage. F149 was offspring of F23, born 04-22-11. F149 (sibling of M161 below) was orphaned at 13.5 months old when her mother F23 was killed by a male puma. F149 dispersed from the natal area by 07-16-12 to E side U.P. study area when she was 14.8 months old; onto Bostwick Park, then W to Dry Creek. Killed by a puma hunter 12-31-12 in GMU 70W, Dry Creek; UTM: 12S, 713658E, 4229703N; age 20 months. Dispersed. M150 was offspring of F111, born on 08-31-09. He was independent by 03-28-11 when he was 19 months old. Lost contact after 04-11-11 when M150 was in Cow Creek southeast of the study area. Survived to adult stage. F152 was independent from her mother F93 by 05-04-12 when about 23 months old. She ranged as a subadult and adult on the natal area (07-31-12). Survived to adult stage. Consorted with F137 when 23 months old on 09-07-2011. Killed by Wildlife Services agent for depredation on an alpaca in Dallas Creek on 09-13-11. M153 was 23 months old at death. Died in subadult stage. M161 (sibling of F149 above) was orphaned at 13.5 months old when his mother F23 was killed by a male puma. M161 dispersed from the natal area by 06-29-12 to E side U.P. study area when he was 14 months old. He shed his expandable cub collar about 08-03-12. M161 was struck and killed by a vehicle on Dallas Divide, HWY 62 in October 2012 when he was 18 months old. Survived to adult stage. F163 was captured at about 18 months old on the study area. She emigrated from the study area and established an adult home range on the NW Uncompahgre Plateau as of July 2012 (0716-12 location). Lost contact after 02-26-12. M164 may have dispersed a long distance. Fate unknown. M165 moved from W to E side of the study area. Appeared to establish adult home range on NE Uncompahgre Plateau. Killed by a puma hunter 12-17-12 in GMU 62N, Dry Fork Escalante Creek; UTM: 12S, 730184E, 4272500N; age about 29 months, adult stage. M181 moved to NE Uncompahgre Plateau, ranging N of the study area. Turned to adult age (24 mo.) July 2013. 168 F181 moved from E to W side of study area. Turned to adult age (24 mo.) April 2013. 140 Lost contact after 06-17-13. F194 dispersed S, last location on North Mt., head of Naturita Creek. Estimated age 30 months in June 2013. 139 F197 ranges on W side of the study area. Turns to adult age (24 mo.) August 2013. �Table 17. Records of pumas that dispersed from the Uncompahgre Plateau study area, December 2004 to July 2013. Puma I.D. 1st capture date on study area 1st capture location→kill or resight location (UTM, NAD27) M5 02-04-05 13S,240577E, 4251037N→ 12S,665853Ex 4277125N M11 06-27-05 13S,248278E, 4239858N→ 12S,741882Ex 4161575N 84.8 M31 04-19-06 329.8 M38 09-08-06 12S,746919E, 4225441N→ 13S,500000Ex 4050000N 13S,249200E, 4239703N→ 12S,703371E, 4316856N M39 09-11-06 71.3 M43 09-15-06 12S,724270E, 4243610N→ 12S,709889E, 4313490N 12S,760177E, 4242995N→ 12S,739859E, 4308557N M48 10-18-06 52.0 M49 12-05-06 12S,756676E, 4247777N→ 12S,704982E, 4248998N 12S,757241E, 4258259N→ 12S,693350E, 4274559N M58 06-27-07 13S,258543E, 4238071N→ 13S,274670E, 4309488N 73.2 Estimated linear dispersal distance (km)* 102.2 104.1 68.6 66.1 Puma Information M5 was offspring of F3, born August 2004. Independent and dispersed from natal area at 13 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months (protected from hunting mortality in buffer area) and ranged into the eastern edge of Utah (vulnerable to hunting). Killed by a puma hunter on 02-20-09 in Beaver Creek, Utah at about 54 months old. M11 was offspring of F2, born May 2005. Shed expandable radiocollar 10-24 to 11-08-05. Recaptured and re-collared 04-0206. Independent at 13 months old. Dispersed from natal area at 14 months old. Moved to Dolores River valley, CO, by 12-14-06. Killed by a puma hunter on 12-02-07 when about 30 months old. M31’s estimated age at capture was 20 months. Dispersed to northern New Mexico and was killed by a puma hunter on 12-1108 in Middle Ponil Creek, Cimarron Range. He was about 52 months old. M38 was offspring of F2, born July 29, 2006. Shed his expandable radiocollar by 03-06-07. Photographs by trail camera in McKenzie Cr. of M38 & Unm. F sibling with F2 on 07-16 to 17-07 at 352-353 days old. M38 was killed by a puma hunter in Ladder Creek southwest of Grand Junction, CO on 01-07-11. He was 53.2 months old at death. M39 was offspring of F8, born August 2006. M39 was killed by a puma hunter in Bangs Canyon, GMU 40 on 03-12-10 when he was 42.8 months old. M43 was offspring of F7, born August 2006. He shed the expandable radiocollar 11-7 to 17-06, after which direct contact was lost. M43 was killed by a puma hunter 01-28-09 in Deer Creek, west slope of Grand Mesa, CO when he was 29.5 months old. M48 was the offspring of F3, born September 2006. M48 was killed by a puma hunter in Tabeguache Creek, GMU 61N on 1227-09 when he was 38.9 months old. M49 was offspring of F50, born July 2006. Orphaned at about 9 months old, when F50 died of natural causes. Dispersed from his natal area at about 10 months old and ranged on the northeast slope of the Uncompahgre Plateau. When M49 was about 15 months old, he shed his expandable radiocollar on about 10-01-07 at a yearling cow elk kill on the northeast slope of the Uncompahgre Plateau. He was killed by a puma hunter in Blue Creek GMU 61N in the protected buffer zone north of the study area on 01-24-09; he was about 29 months old. M58 was offspring of F16, born May 2007. M58 was killed by a puma hunter on 12-27-09 in the North Fork of the Gunnison River north of Paonia, GMU 521; he was 31 months old. �Table 17 continued. Estimated Puma Information linear dispersal distance (km)* 66.1 M63 was offspring of F24, born July 14, 2007. He was not radiocollared as a cub. M63 was killed by a puma hunter in Calamity Creek on northwest Uncompahgre Plateau on 01-01-11. M63 was 41.5 months old at death. 97.0 M65 was offspring of F24, born July 2007. M65 was killed by a USDA, APHIS, WS agent for depredation on llamas in the Little Dolores River on 11-07-09. M65 was 27.8 months old. Puma I.D. 1st capture date on study area 1st capture location→kill or resight location (UTM, NAD27) M63 08-17-07 M65 08-17-07 M67 08-23-07 12S,738144E, 4233628N→ 12S,689111E, 4277908N 12S,738144E, 4233628N→ 12S,684084E, 4314200N 13S,257371E, 4235231N→ 12S,725113E, 4242447N M68 08-23-07 M69 01-11-08 M82 07-05-08 12S,726901E, 4243463N→ 13S,255316E, 4216768N 60.5 M83 07-05-08 90.7 M87 07-31-08 12S,726901E, 4243463N→ 12S,670949E, 4314779N 13S,239006E, 4248601N→ 12S,724325E, 4244118N M88 07-31-08 13S,239006E, 4248601N→ 12S,704835E, 4197839N 77.6 M92 09-29-08 13S,246359E, 4226949N→ 12S,750871E, 4222921N 21.9 13S,257371E, 4235231N→ 12S,711262E, 4198681N 13S,248191E, 4246810N→ 13T,378900E, 4591990N 57.7 80.7 369.6 39.2 M67 was offspring of F30, born July 17, 2007 in Fisher Creek on the east slope of the study area. He was not radiocollared as a cub. M67 dispersed from the natal area and was recaptured in Tomcat Creek on the west slope of the study area on 02-24-10 when he was 31 months old. M67 is a resident adult in that area (07-3111). Killed by puma hunter in GMU61N on 12-18-11 when 52.9 months old. M68 was offspring of F30, born July 2007. He was orphaned at 12 months old when his mother was killed by a puma. He was killed by a puma hunter in the Disappointment Valley in southwest CO on 12-30-08; he was 17 months old. M69 was captured on the study area when about 14-18 months old. Emigrated from the study area as subadult by 03-19-08. Last VHF aerial location was southwest of Waterdog Peak, east side of Uncompahgre River Valley on 04-07-08. M69 was killed by a puma hunter on 11-06-08 in Pass Creek in the Snowy Range, WY when he was 24 to 28 months old. M82 was offspring of F8, born May 29, 2008; sibling of M83 below. He shed his expandable cub radiocollar after 03-20-09. M82 was killed by a puma hunter on 12-10-09 in the Beaver Creek fork of East Dallas Creek, GMU 65. M82 was 19 months old. M83 was offspring of F8, born May 29, 2008; sibling of M82 above. He was not radiocollared as a cub. M82 was killed by a puma hunter on 01-18-11 in Coates Creek west of Glade Park, CO. He was 31.6 months old at death. M87 was offspring of F3, born July 3, 2008 on the east slope of the study area; sibling of M88 below. He was not radiocollared as a cub. M87 dispersed from the natal area. He was recaptured on the west slope of the study area on 02-09-11 when he was 31 months old. M87 is was resident adult on the west slope of the study area. He was killed by a puma hunter on 12-06-11 at 41 months old north of the study area. M88 was offspring of F3, born July 3, 2008 on the east slope of the study area; sibling of M87 above. He was not radiocollared as a cub. M87 dispersed from the natal area. He was killed by a puma hunter in Dawson Creek, Disappointment Valley on 11-3010 when he was 29 months old. M92 was offspring of F25, born August 19, 2008. He was radiocollared as a cub; last contact on 12-12-08. M92 dispersed from the natal area and was recaptured in McKenzie Creek, west slope of the study area on 04-22-11 when he was 32 months old. He could not be handled to fit a new radiocollar because of a dangerous tree. �Puma I.D. 1st capture date on study area 1st capture location→kill or resight location (UTM, NAD27) M107 06-28-09 M112 01-23-10 13S,242359E, 4252618N→ 12S,754886E, 4341330N 13S,248567E, 4240108N→ 12S,732863E, 4146772N M114 02-27-10 M117 02-05-10 M126 09-05-10 M144 03-07-33 M161 01-23-12 Table 17 continued. Estimated Puma Information linear dispersal distance (km)* 89.2 M107 was offspring of F94, born May 25, 2009; sibling of F108 below. He was not radiocollared as a cub. M107 dispersed from the nata area. He was killed by a puma hunter in Cottonwood Creek near Molina, CO on 12-09-10 when he was 19 months old. 102.5 M112 was initially captured 4.7 mo. old in his natal area while dependent on his mother F70 on 01-23-10. He was recaptured 0124-11 in the natal area at 17 months old, independent of F70. M112 associated with F96 and her two radio-collared cubs F129 and M130 during 02-10-11 to 04-18-11 when he was 18-20 mo. old. Lost contact of M112 after 04-18-11. Dispersed and emigrated from the U.P. study area. M112 was killed by a puma hunter 01-06-2013, GMU 73, SE of Dolores, CO; UTM: 12S, 732863E, 4146772N; age 41 months. 13S,256933E, 4237862N→ 13S,492615E, 4266192N 12S,731840E, 4232346N→ 12S,743909E, 4216633N 237.5 12S,734503E, 4224636N→ 12S, 710850E, 4239350N 12S,727173E, 4242012N→ 12S,696439E, 4276888N 27.7 12S,727932E, 4239430N→ 13S,247567E, 4220129N 49.2 19.7 46.6 M114 was initially captured at about 30 months old. Emigrated from the U.P. study area. He was killed by a puma hunter on 0310-12 in Beaver Creek, GMU59. He was about 55 months old at death. M117 was offspring of F119. He wore an expandable cub collar, but shed the collar by 07-15-10 on the natal area when about 11 months old. M117 was killed by a puma hunter in Beaver Creek, San Miguel River at the southern extreme of his natal area on 0101-11. He was 17 months old at death. It is unknown if M117 was independent from his mother F119 at the time of his death. M126 was offspring of F118, born Aug. 8, 2010. Lost radio contact after 03-17-11; shed his radiocollar at a mule deer cache. Dispersed from natal area. Killed by a puma hunter on 01-08-12 in Tuttle Draw WNW of Nucla, CO as 17-month-old subadult. M144 was initially captured as an independent subadult in association with subadults F145 and F146 on the study area. Mother is unknown. He moved off the study area on 03-15-11. M144 established his adult territory on northwest Uncompahgre Plateau and upper Unaweep Canyon from Sep. 2011 to 02-25-13. M144 was killed by a puma hunter 02-25-13 in GMU 40, North Fork West Creek, Unaweep Canyon. M161 (sibling of F149) was orphaned when his mother F23 was killed by a male puma on 06-06-12; he was 411 days (13.5 mo.) old. M161 dispersed from the natal area by 06-29-12 when he was 14 months old and moved to the east slope of the U.P. study area. M161 shed his expandable cub collar about Aug. 3, 2012 in head of E Fk. Dry Creek. He was struck and killed by a vehicle on highway 62 at Dallas Divide in October 2012; he was 18 mo. old. �Puma I.D. 1st capture date on study area 1st capture location→kill or resight location (UTM, NAD27) F52 01-10-07 13S,258058E, 4236260N→ 13S,319217E, 4240467N F97 02-04-09 12S,727529E, 4237648N→ 12S,705930E, 4227299N F106 06-14-09 12S,736451E, 4240278N→ 13S,258089E, 4235866N F108 06-28-09 13S,242359E, 4252618N→ 12S,752013E, 4263883N M122 08-12-2010 M131 09-25-10 12S,746164E, 4276613N→ 12S,735353E, 4283455N 12S,760695E, 4243505N→ 12S,670829E, 4246980N F143 02-15-11 F145 03-18-11 12S,723748E, 4238579N→ 12S,721795, 4264246 12S,727181E, 4241468N→ 12S,705833E, 4312909N Table 17 continued. Estimated Puma Information linear dispersal distance (km)* 61.1 F52 was captured on the study area when about 18-20 months old. Dispersed from study area as a subadult by Jan. 16, 2007. F52’s last VHF aerial location was Crystal Creek, a tributary of the Gunnison River east of the Black Canyon 05-15-07. She was treed by puma hunters on 12-29-08 on east Huntsman Mesa, southeast of Powderhorn, CO. She was about 41-43 months old . F52 was treed again by puma hunters on about 12-16-09 south of Powderhorn: 13S,319480E,4233219N. F52 was about 53-55 months old. This suggests that F52 has an adult home range in that area. F52 was killed by a puma hunter on 01-09-12 in North Beaver Creek SE of Powederhorn, CO. She was about 79 months old at death. 24.0 F97 was offspring of F23, born May 23, 2008. She was radiocollared at 8.5 month old in San Miguel Canyon; but, lost contact on 05-12-09 after F97 shed the radiocollar at an elk cache. F97 dispersed from the U.P. study area. She was killed by a puma hunter on 01-22-12 in Dry Creek west of the U.P. study area when she was 43.9 months old. 46.9 F106 was offspring of F75, born May 7, 2009. She wore an expandable cub collar, but shed it about 03-23-10. F106 dispersed from the natal area and moved to the east slope of the study area where she was photographed at one of our scent station cameras at the mouth of Fisher Creek from 02-27-11 to 03-03-11. She was identified by her eartag. F106 was 21 months old. 18.2 F108 was offspring of F94, born May 25, 2009; sibling of M107 above. She was fitted with an expandable cub collar; but, shed the collar in the original nursery due to failure of the fastener. F108 dispersed from the natal area. She was killed by a puma hunter on the study area on 11-29-10 when she was 17 months old. 12.9 M122 was offspring of F104, born July 8, 2010. Fitted with expandable cub collar 08-12-10. Lost contact 04-28-11 due to transmitter malfunction. Killed by puma hunter N of natal area 01-23-13 at 30 mo. old. 90.1 M131 was offspring of F96, born August 21, 2010. Lost contact after 07-21-11. Shed his radiocollar about 07-27-11. Survived to recapture on 02-02-12 at 17.4 months old, with sibling F129; neither handled due to dangerous trees. Emigrated from U.P. study area. Killed by a puma hunter 01-17-13 at 29 mo. old in GMU 60 in western Colorado near border with Utah. 25.7 F143 was captured on the study area when about 24 months old. Dispersed N on the Uncompahgre Plateau and established an adult home range on the NW portion of the Uncompahgre Plateau (most recent location 07-16-12). 74.5 F145 was originally captured in association of M144 and F146; they may be siblings. Mother unknown. She moved off the study area with M144 on 03-15-11. F145 emigrated to Colorado Mesa. She was killed by a puma hunter 01-23-12 in West Bangs Canyon. F145 was 28 months old at death. �Puma I.D. 1st capture date on study area 1st capture location→kill or resight location (UTM, NAD27) F149 06-06-11 12S,729993E, 4242329N→ 12S,713658E, 4229703N F163 01-26-12 M165 02-24-12 12S,732153E, 4232452N→ 12S,695407E, 4280753N 12S,722816E, 4246926N→ 12S,730814E, 4272500N F194 01-29-13 12S,742443E, 4225259N→ 12S,729101E, 4201962N Table 17 continued. Estimated Puma Information linear dispersal distance (km)* 20.7 F149 (sibling of M161) was orphaned when her mother F23 was killed by a male puma on 06-06-12; she was 411 days (13.5 mo.) old. F149 dispersed from the natal area by 07-16-12 when she was 14.8 months old and moved to the NE Uncompahgre Plateau, onto Bostwick Park, then back across Uncompahgre Plateau. She emigrated from the U.P. study area and was killed by a puma hunter 12-31-12 at 20 mo. old 60.7 F163 was initially captured at about 18 months old. She emigrated from the study area and may have established an adult home range on the N portion of the Uncompahgre Plateau as of July 2012 (0716-12 most recent location). 26.9 M165 was first captured 02-24-12 at ~19 mo. old. His origin unknown. He moved from the west slope of the U.P. study area to the east slope of the U.P. north of the study area between 05-042012 and 06-15-12. He was killed by a puma hunter in GMU 62N on 12-17-12 when he was ~29 mo. old. 26.9 F194 was first captured at ~24 mo. old on W slope of U.P. study area on 01-29-13. Her origin unknown. She emigrated from the U.P. study area heading S. Her last aerial location was 06-17-13 on North Mt. in the SW head of Naturita Creek. *Estimated linear dispersal distance (km) from initial capture site on Uncompahgre Plateau study area to hunter kill, or last recapture, radio location, or observation site. �Table 18. Recorded deaths of non-marked and marked pumas struck by vehicles and other unusual causes, in chronological order, on the Uncompahgre Plateau puma study area, Colorado, from 2004 to 2012. Puma sex & ID if marked M Estimated age (mo.) Date recorded Cause of death General physical condition Location & UTM NAD27 12 09-24-04 Good F 49 07-28-05 Vehicle collision Vehicle collision Pleasant Valley, County Road 24 13S,252870E,4227520N Highway 62 east of Dallas divide 13S,250000E,4222500N F17a 11 08-18-06 F 18-24 11-06-06 F 6 01-30-07 F P1005 36 09-16-08 M 12-24 08-13-08 F61a 18 11-13-08 F 12 08-10-09 F16b 80 09-11-09 M6b 99 05-21-0 F113b 42 06-06-10 M P1018c F P1030c M P1034 M161 24 08-25-10 6 02-16-11 4 10-07-11 18 06-17-13 a Vehicle collision Vehicle collision Vehicle collision Asphyxia, lodged in fork of tree Vehicle collision Vehicle collision Vehicle collision Vehicle collision Vehicle collision Vehicle collision Vehicle collision Vehicle collision Vehicle collision Vehicle collision Good Not pregnant or lactating Good Good Good Unknown, decomposed Good Good Good Good Good Good Not pregnant or lactating Excellent Good Fair Unknown, decomposed Subadult marked (i.e., tattoos, eartags), but not radio-collared. Adult GPS/VHF-collared pumas. c Non-marked puma with P one-thousand number designation. b Highway 550 south of Colona 13S,257602E,4242185N Highway 550 east of Ridgway State Park 13S,259843E,4235985N Highway 62 west of Dallas divide 12S,762286Ex4218992N Davis Point, Roubideau Canyon 12S, 743718E,4255277N Highway 145 west of Placerville 13S,756490E,4212336N Highway 550 east of Ridgway State Park 13S,259843E,4235985N Highway 145 east of Norwood 12S,745739E,4222548N Ouray County Road 1 13S,253733E,4240060N Highway 550 south of Colona 13S,258610E,4236805N F113 crossed Highway 550 and roads on Loghill Mesa 24-30 hours before she died in McKenzie Creek 13S,257272E,4238435N Highway 62 Leopard Creek 12S,237747E,4220330N Highway 62 Leopard Creek 12S,760953E,4216683N Highway 62 Leopard Creek 12S,762806E,4219531N Highway 62 Dallas Divide 13S,2475674220129 �Table 19. Pumas monitored with GPS collars on the Uncompahgre Plateau, Colorado, December 2004 to July 2013. Puma I.D. M1 M4 M6 M27 M29 M51 M55 M100 M133 M178 M179 M183 F2 F3 F7 F8 F16 F23 Sex M M M M M M M M M M M M F F F F F F F24 F25 F28 F30 F50 F52 F54 F70 F72 F75 F93 F95 F96 F104 F111 F113 F129 F135 F136 F137 F152 F F F F F F F F F F F F F F F F F F F F F F171 F172 F177 F181 F F F F F182 F186 F F Age stage adult adult adult adult adult adult adult adult adult adult adult adult adult adult adult adult adult subadult adult adult adult adult adult adult subadult adult adult adult adult adult adult adult adult adult adult adult adult adult adult subadult adult adult adult adult subadult adult adult adult Dates monitored 12-08-04 to 07-20-06 01-28-05 to 01-14-06 02-18-05 to 05-14-08 03-12-06 to 06-21-06 04-14-06 to 01-01-08 01-07-07 to 07-15-08 01-21-07 to 11-25-10 03-27-09 to 01-16-10 11-12-10 to 12-01-10 11-13-12 to 12-11-12 11-18-12 to 12-29-12 02-14-13 to 07-31-13 01-07-05 to 08-14-08 01-21-05 to 12-11-11 02-24-05 to 08-03-08 03-21-05 to 10-10-06 10-12-05 to 09-10-09 01-04-06 to 02-04-06 02-05-06 to 09-04-09 01-17-06 to 07-25-07 02-09-06 to 09-09-09 03-24-06 to 08-15-07 03-30-07 to 02-22-08 12-14-06 to 03-26-07 01-10-07 to 05-08-07 01-12-07 to 08-18-08 01-14-08 to 12-22-11 02-12-08 to 07-07-10 03-26-08 to 06-03-09 10-03-12 to 11-11-12 03-14-13 to 07-31-13 01-28-09 to 07-31-12 05-29-09 to 01-31-12 01-01-10 to 07-31-13 01-27-10 to 06-06-10 01-02-13 to 07-31-13 01-01-11 to 09-20-11 01-20-11 to 07-31-13 04-12-11 to 07-31-13 01-18-12 to 06-15-12 06-16-12 to 12-23-12 01-20-12 to 07-31-13 03-28-12 to 07-31-13 10-27-12 to 12-10-12 01-15-13 to 04-15-13 04-16-13 to 07-31-13 02-04-13 to 07-31-13 03-30-13 to 07-31-13 �GOAL: Strategies, Information, & Tools for Managing Healthy, Self-sustaining Puma Populations in Colorado Puma Habitat Puma Population Effects of Harvest & Other Mortality Movements & Corridors Population Dynamics: Sex & Age, Vital Rates, Mortality, Population Growth Rates Vulnerability to Harvest Human Development Habitat Use Effects of Translocation Domestic Animals Puma― Human Relationships Deer, Elk, Other Natural Prey & Species of Concern Effects of Predation Effects of Evasive Conditioning Map Prey Distribution Habitat Models Population Models Puma Prey Puma―Prey Relationships Models Methods for Monitoring Populations Habitat Maps Figure. 1. An ecologically-based conceptual model of the Colorado Puma Research Program that provides the contextual framework for this and proposed puma research in Colorado. Gray-shaded shapes identify areas of research addressed by this puma research on the Uncompahgre Plateau for the puma management goal in Colorado (at top). �Figure 2. The puma study area on the southern half of the Uncompahgre Plateau, Colorado (shaded in gray) comprising the southern portions of Game Management Units (GMUs) 61 and 62 and a northern portion of GMU 70. Puma Population Trend, U.P., CO GO 47 0 IY2 IYl YEARS Figure 3. Trends in the population of independent pumas on the Uncompahgre Plateau Puma Study Area, including Reference Years 4 and 5 (RY4, RY5) and Treatment Years 1, 2, 3, and 4 (TY1, TY2, TY3, TY4). Numbers represent minimum counts that include all pumas from known radio-collared pumas, visual observations of non-marked pumas, harvested non-marked pumas, and track counts of suspected �non-marked pumas on the study area during fall to spring hunting and research capture seasons, except RY5 (45), which had to be modeled from RY4 observation data (33) because the state government hiring freeze that year affected search and capture efforts. The actual minimum count for RY5 was 37 independent pumas. The quota of 8 pumas for TY1 represented a 15% harvest of the model projected 53 independent pumas expected in TY1 and was used to set the quota ahead of the hunting season. Starting in TY1, two capture teams were deployed to count pumas on the study area because the hunting season shortened our fall-winter-spring research period. We deployed a team on each the east and west sides of the study area. The minimum count for TY1 was actually 55 independent pumas, consistent with the model expected 53. Post-harvest high trend line represents the population of independent pumas after pumas harvested only on the study area by hunters. This trend line represents 11.9% to 16.7% harvest of independent pumas. Post-harvest low trend line represents the population of independent pumas after pumas harvested on the study area and pumas harvested when they ranged onto adjacent GMUs open to hunting and other mortalities are subtracted from the minimum count. TY1 post-harvest low includes 1 adult female and 3 adult males killed off the study area. The TY2 post- harvest low includes 1 adult male killed off the study area and 2 adult female pumas killed in February 2011 on the study area to protect livestock. The TY3 post-harvest low includes 1 adult female and 4 adult males harvested off the study area and 2 adult females that died of natural causes on the study area. The TY4 post-harvest low includes 1 adult female and 1 adult male harvested off the study area and 1 adult female that died of natural cause. This trend line represents 21.2% to 31.2% harvest of independent pumas. Age structure of independent pumas in November 2012 at beginning of puma hunting season in Treatment Year 4, Uncompahgre Plateau, Colorado. 8 7 ~ 6 55 ....if 4 0 ■ Female 0 3 Z 2 ■ Male 1 0 lto 2 >2 to >3 to >4 to >5 to >6 to >7 to >8 to >9 to 10+ 3 4 5 6 7 8 9 10 Age (.y ears) Figure 4. Estimated age structure of independent pumas in November 2012 at the beginning of the puma hunting season in Treatment Year 4 (TY4) on the Uncompahgre Plateau study area, Colorado. All these pumas were captured and sampled by researchers or harvested by hunters and examined by researchers. Mean ± SD of independent female and male ages, respectively: 4.29 ± 2.69 yr. (51.48 ± 32.29 mo.), n = 21; 2.51 ± 0.86 yr. (30.12 ± 10.37 mo.), n = 8. �Puma births, Uncompahgre Plateau, Colorado. 16 14 12 ........"'~ 10 8 ;:i 0 z 6 4 17 2 0 I n ■ I ■ j I I I I I 17 I II Jan. Feb. Mar. Apr. May June Ju ly Aug. Sep. ■ Births 2005-2013 I 11 I I n I Oct. Nov. Dec. ■ Births 1982-1987 Figure 5. Puma births (black bars) detected by month from May 19, 2005 to July 31, 2013 (n = 53 litters of 27 females; 51 of the litters were examined at nurseries when cubs were 26-42 days old and 2 litters confirmed by tracks of ≥1 cubs following GPS-collared mothers F28 and F111 when cubs were ≤42 days old). Also shown (gray bars) are results of the earlier effort by Anderson et al. (1992:48; 1982 to 1987, n = 10 litters of 8 females, examined when cubs were <1 to 8 months old), Uncompahgre Plateau, Colorado. �UP Study Area_ Period 1 Locations ~ N BlM SOR coow CITY COl»ITY FWS lNCJTRIJST NPS """"TE USFS- GMUG U SFS - s.-.N JIJ,,Vf 0 3.75 7.5 15 Kilometers Figure 6. The grid on the east slope of the Uncompahgre Plateau Puma Project study area indicating the 18 camera/call box sites (red dots) in sample period 1. A total of 3 sample periods were used, each 28 days long and each with 18 sites, for a total of 54 random cells surveyed December 2012 to March 2013 to test non-invasive survey methods. Image by M.S. student Kirstie Yeager. �Appendix A. Summary of individual puma cub survival and mortality, 2005 to 2013, Uncompahgre Plateau, Colorado. Puma I.D. M5 Estimated Age at capture (days) 183 Est. Birth date ~8-1-04 Est. survival span from 1st capture to fate or last monitor date 02-04-05 to 02-20-09 Age to last monitor date alive or at death (days, birth to fate) ~1,664 F9 31 5-28-05 06-27-05 to 4-19-06 06-27-05 to 11-20-05― 12-29-05 06-27-05 to 12-02-07 326-333 F10 31 5-28-05 M11 31 5-28-05 F12 42 5-19-05 07-01-05 to 12-08-05― 01-26-06 203-252 F13 42 5-19-05 101 F14 26 6-26-05 07-01-05 to 08-28-05 07-22-05 to 02-07-06― 03-10-06 M15 26 6-26-05 F17 34 9-22-05 F18 34 9-22-05 M19 34 9-22-05 M20 34 9-22-05 176-215 918 226-257 07-22-05 to 06-06 to 14-06 10-26-05 to 08-18-06 345-353 10-26-05 to 07-20 to 27-06 10-26-05 to 07-27 to 08-02-06 10-26-05 to 05-24-06 301-308 330 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. Radio-collared. Survived to subadult stage by 09-16-05; independent at ~13 mo. old. Dispersed from natal area by 09-29-05 at 14 mo. old. Established territory on NW U.P. Killed by hunter in Beaver Creek, UT 02-20-09 at 54.6 months old. Radio-collared. Shed radiocollar 04-19-06 to 04-26-06. F3 Radio-collared. Dhed radiocollar 08-10-05; last tracks of F10 with mother F2 & siblings F9 & M11 observed 11-2005. F10 disappeared by 12-30-05. Radio-collared. Shed collar 10-24 to 11-08-05. Recollared on 04-02-06. Survived to subadult stage by 06-21-06, independent at 13 mo. old. Dispersed from natal area by 0711-06 at 14 mo. old. Moved to Dolores River valley in SW CO by 12-14-06. Killed by a hunter in SW CO 12-2-07 at 918 days (30 mo.) old. Radio-collared. Shed radiocollar 07-28-05―08-01-05. Tracks of F12 found in association with mother F7 on 1208-05. F12 disappeared by 01-27-06 when she was not visually observed with F7, and her tracks were not seen in association with F7’s tracks. Radio-collared. Killed and eaten by a puma possibly M5 (13 mo. old) about 08-28-05. Radio-collared. Shed radiocollar 01-20-06 to 01-25-06. Tracks of F14 were observed with tracks of mother F8 & sibling M15 on 02-07-06. Disappeared by 03-11-06, only tracks of F8 & M15 were found. Radio-collared. Shed radiocollar 06-06-06 to 06-14-06. F2 F2 F2 F7 F7 F8 F8 308-314 Radio-collared. Shed radiocollar 06-06-06 to 06-14-06. Killed by a car on highway 550 on 08-18-06. Probably dependent on F16. Died at 10.8 months old Radio-collared. Probably killed by another puma. Multiple bite wounds to skull. Died at 10 months old. Radio-collared. Shed radiocollar 07-27-06 to 08-02-06. F16 F16 244-245 Radio-collared. Shed radiocollar 05-24-06―05-25-06. F16 F16 �Appendix A continued Puma I.D. Estimated Age at capture (days) F21 37 9-26-05 M22 37 9-26-05 M26 183 8-1-05 F33 31 5-30-06 F34 31 F35 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 11-02-05 to 08-16-06 11-02-05 to 12-21-05― 12-22-05 02-08-06 to 03-21 to 24-06 06-30-06 to 07-31-06 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 324 Radio-collared. Lost contact; radiocollar quit. Last aerial location 8-16-06, live signal. Radio-collared. Killed and eaten by male puma 12-21-05 to 12-22-05. F3 ~232-235 Radio-collared. Shed radiocollar 03-21-06 to 03-24-06. F25 63-65 F23 5-30-06 06-30-06 to 07-31-06 63-65 31 5-30-06 38 F36 29 6-9-06 29 6-9-06 M38 41 7-29-06 Radio-collared. Killed and eaten by a male puma 08-22-06. GPS data on M29 indicate he was not involved. Radio-collared. Killed and eaten by a male puma 08-22-06. GPS data on M29 indicate he was not involved. Radio-collared. Shed radiocollar found 03-06-07. Photo (trail camera in McKenzie Cr.) of M38 & Unm. F sibling with F2 on 07-16 to 17-07 at 352-353 days old. Dispersed. Killed by puma hunter 01-07-11 in GMU40 Ladder Creek, SW of Grand Junction, CO when he was 53.2 months old. Radio-collared. Shed radiocollar by 09-20-06, but seen alive on that date. Tracks of 2 cubs following F8 on 04-25-07. Survived to adult stage; dispersed from natal area. Dispersed. Killed by a puma hunter 03-12-10 in GMU 40, Bangs Canyon, when 42.8 months old. Radio-collared. Shed radiocollar by 09-20-06, but seen alive on that date. Tracks of 2 cubs following F8 on 04-25-07. F28 M37 06-30-06 to 07-07-06 07-08-06 to 07-28-06 07-08-06 to 07-28-06 09-08-06 to 07-16 to 17-07 Radio-collared. Probably killed and eaten by a male puma 08-01 to 03-06. GPS data on M29 indicate he was not involved. Radio-collared. Probably killed and eaten by a male puma 08-01 to 03-06. GPS data on M29 indicate he was not involved. Dead; research-related fatality.a Radio-collared. Assumed dead. Shed radiocollar or died (blood on collar) between 10-05-06 (last live signal) & 1013-06 (collar found). Dead; research-related fatality.b F8 86-87 74 74 352-353 M39 29 8-13-06 F40 29 8-13-06 F41 29 8-13-06 M42 29 8-13-06 09-11-06 to 09-20-06 to 04-25-07 09-11-06 to 09-20-06 to 04-25-07 09-11-06 to 10-05-06 09-11-06 to 11-27-06 1623 9 255 1307 9 Mother I.D. F3 F23 F23 F28 F2 F8 F8 255 53-61 106 F8 �Appendix A continued Puma I.D. Estimated Age at capture (days) M43 33 8-13-06 M44 8-13-06 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 09-15-06 03-01-07 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 200 Radio-collared. Shed radiocollar by 11-7 to 17-06. Dispersed. Killed by a puma hunter 01-28-09 in Deer Creek, west slope of Grand Mesa, CO GMU41 at 29.5 months old. Radio-collared. Shed radiocollar by 10-27-06. Treed, visually observed 02-14-07; sibling (?) M56 also captured, sampled, & marked for 1st time. M44 killed by Wildlife Services for depredation control on 12-05-07, for killing 4 domestic sheep. He was still dependent on F7. He was 15.7 months old. Radio-collared. Multiple puncture wounds on braincase― parietal & occipital regions; consistent with bites from coyote. F45 switched families, moving from F7 to F2 about 12-19 to 20-06. Last date F45 was with F2 was 04-17-07. Died 05-20 to 23-07 when she was 9.2 months old. Radio-collared. Shed collar by 12-14-06. Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-1207 in Puma Canyon. Radio-collared. Shed collar . Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-1207 in Puma Canyon. Radio-collared. Shed radiocollar. Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-12-07 in Puma Canyon. Dispersed. Survived to adult stage; dispersed from natal area. Killed by a puma hunter 12-27-09 in Tabaguache Creek, GMU 61N when 38.9 months old. Radio-collared. M49 was orphaned when his mother died on about 03-26-07; he was ~268 days old. M49 dispersed from natal area and onto NE slope of U.P. Shed radiocollar at a yearling cow elk kill about 10-01-07; he was ~428 days old. Dispersed from natal area. Killed by a puma hunter in Blue Creek, northwest Uncompahgre Plateau (GMU 61N) 01-2409 when ~29 months old. 899 33 09-15-06 to 02-14-07 479 F45 33 8-13-06 09-15-06 to 5-20 to 23-07 280-283 M46 31 9-17-06 10-18-06 to 12-15-06 89 360 M47 M48 M49 31 31 153 9-17-06 9-17-06 7-1-06 10-18-06 to 12-15-06 to 09-12-07 10-18-06 to 12-15-06 to 09-12-07 to 12-27-09 89 360 89 360 1187 12-05-06 to 07-31-07 to 01-24-09 939 Mother I.D. F7 F7 F7 F3 F3 F3 F50 �Appendix A continued Puma I.D. Estimated Age at capture (days) F53 183 Est. Birth date 7-1-06 M56c 183 ~8-13-06 F57 35 4-16-07 M58 34 5-24-07 Est. survival span from 1st capture to fate or last monitor date 01-12-07 to 02-23-07 to 09-02-07 02-14-07 to 03-01-07 05-21-07 to 06-06-07 06-27-07 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 42 Radio-collared. Shed radiocollar 02-23-07. F53 visually observed by P. & F. Star (Loghill Mesa), on 09-02-07, when F53 was ~14 months old and an independent subadult. F54 Radio-collared. Shed radiocollar 2-27-07. M56 observed 0301-07. Radio-collared. Shed radiocollar 06-07-07. Live mode 0606-07. Not radio-collared. Tracks of 3 cubs observed with F16’s tracks on 04-12-08, McKenzie Butte-Pinon Ridge Pass. 3 cubs observed with F16 on 08-08-08 by B. & T. Traegde. Dispersed. Survived to adult stage. Killed by a puma hunter 12-27-09 in GMU 521, North Fork Gunnison River, when 31 months old. Radio-collared. Shed collar about 02-14-08. Observed with 11-20-07 with F16, but without siblings M58 and F61. Tracks of 3 cubs observed with F16’s tracks on 04-12-08, McKenzie Butte-Pinon Ridge Pass. Three cubs observed with F16 on 08-08-08 by B. & T. Traegde. Dead; research-related mortality.d F7 (?) Radio-collared. Radiocollar malfunction. Tracks of 3 cubs observed with F16’s tracks on 04-12-08, McKenzie Butte-Pinon Ridge Pass. 3 cubs observed with F16 on 08-08-08 by B. & T. Traegde. Dead. Died probably as independent subadult at 538 days old; struck by car on Hwy 550 mi. marker 111 N. of Ridgway, CO, euthanized by gunshot on 11-13-08. Not radio-collared. Not radio-collared. Dispersed from study area. Killed by a puma hunter 01-01-11 in Calamity Creek, GMU61N when he was 41.5 months old. Not radio-collared. Two out of potential of 4 of F24’s male cubs were visually observed with her on 4/1/08. Assume that 2 male cubs died before the age of 8.5 mo. Eartags were seen on both cubs, but the numbers were not. F16 ~428 subad. 200 52 324 434 F59 34 5-24-07 06-27-07 to 08-21-07 55 324 M60 34 5-24-07 F61 34 5-24-07 06-27-07 to 07-11 to 12-07 06-27-07 to 06-29-07 434 48-49 324 434 538 M62 M63 M64 34 34 34 7-14-07 7-14-07 7-14-07 08-17-07 08-17-07 to 01-01-11 1267 08-17-07 262 Mother I.D. F25 F16 F16 F16 F24 F24 F24 �Appendix A continued Puma I.D. Estimated Age at capture (days) M65 34 Est. Birth date 7-14-07 Est. survival span from 1st capture to fate or last monitor date 08-17-07 to 11-07-09 Age to last monitor date alive or at death (days, birth to fate) 262 847 F66 37 7-17-07 08-23-07 to 05-28-09 M67 37 7-17-07 08-23-07 to 12-18-11 M68 37 7-17-07 08-23-07 to 12-30-08 682 1615 532 F74 259 6-1-07 5-19-08 03-12-08 to 07-09-08 06-18-08 M76 30 M77 30 403 ~87 5-19-08 06-18-08 ~87 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Not radio-collared. Two out of potential of 4 of F24’s male cubs were visually observed with her on 04-01-08. Assume that 2 male cubs died before the age of 8.5 mo. Eartags were seen on both cubs, but the numbers were not. Dispersed. Survived to adult stage. Killed by Wildlife Services for depredation control for predation on llamas in Little Dolores River, on 11-07-09 when 27.8 months old. Radio-collared. Lost contact; last location 11/5/07. No signals after that date. F66 was photographed with one male sibling, either M67 or M68, & F30 on 5/31-6/1/08. F66 was recaptured and radio-collared as a subadult on 11/25/08. She died from massive trauma & bleeding of internal organs possibly resulting from being trampled by an elk or mule deer on about 05-28-09 as an independent subadult 23 months old. Her range overlapped her natal area. Not radio-collared. M67 or M68 was photographed with sibling F66 & mother F30 on 5/31-6/1/08. Dispersed from natal area. Established adult home range on west side of Uncompahgre Plateau study area. Killed by puma hunter in GMU61N on 12-18-11 when 52.9 months old. Not radio-collared. M67 or M68 was photographed with sibling F66 & mother F30 on 05-31 to 06-01-08. Survived to subadult stage. Dispersed. Killed by a puma hunter in Disappointment Valley, CO (GMU 71) 12-30-08 at 17.5 months old. Radio-collared. Shed radiocollar between 7-9-08 and 7-1508, probably while still dependent on mother F75. Not radio-collared. Probably dead; if not killed when sibling M79 was killed, then probably would starve to death. Not radio-collared. Probably dead; if not killed when sibling M79 was killed, then probably would starve to death. Mother I.D. F24 F30 F30 F30 F75 F2 F2 �Appendix A continued Puma I.D. Estimated Age at capture (days) F78 30 M79 Est. Birth date Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 5-19-08 Est. survival span from 1st capture to fate or last monitor date 06-18-08 ~87 F2 30 5-19-08 06-18-08 87 F80 40 5-23-08 07-02-08 F81 40 5-23-08 F95 ~488 ~Aug. 2007 07-02-08 to 07-29-09 12-29-08 to 07-31-13 Not radio-collared. Probably dead; if not killed when sibling M79 was killed, then probably would starve to death. Not radio-collared. Dead. Chewed-off anterior portions of the nasals, maxilla, palate, dentaries, and pieces of the braincase, with 6 or 9 portion of yellow ear-tag and intestines and bits of skin found ~45 m from mother F2’s death site on 08/14/08. Cub death probably due to puma-caused infanticide with cannibalism at ~87 days old. Male puma scrapes, about 8, under a rock rim ~50m distance from cub remains, and made ~ time of pumas’ deaths. Not radio-collared. Apparently died before 02-04-09; no tracks found in association with F23 & siblings F81 & F97. Radio-collared. Last live location 7-29-09. F93 F97 257 5-23-08 02-04-09 to 01-22-12 1339 M82 37 5-29-08 07-05-08 to 12-10-09 560 M83 37 5-29-08 07-05-08 to 01-18-11 964 Radio-collared. F95 was offspring of F93. She survived the subadult stage and into the adult stage. Her home range overlapped her natal area. Radio-collared. Lost contact after 05-12-09; shed collar at elk kill cache on Mailbox Park. Dispersed from study area. Killed by a puma hunter 01-22-12 in Dry Creek when 43.9 months old. Radio-collared. Shed radiocollar after 03-20-09. Survived to subadult stage. Dispersed. Killed by a puma hunter in 1210-09 GMU 65, Beaver Creek fork of East Dallas Creek, when 18.4 months old. Not radio-collared. Survived; dispersed from study area. Killed by a puma hunter 01-18-11 in Coates Creek west of Glade Park, GMU40. He was 31.6 months old. 424 2,196 Mother I.D. F2 F23 F23 F23 F8 F8 �Appendix A continued Puma I.D. Estimated Age at capture (days) M84 36 6-5-08 F85 36 F86 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 07-11-08 to 02-11-09 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 251 6-5-08 07-11-08 to 10-01-08 118 36 6-5-08 07-11-08 to 07-23 to 08-03-08 ~48-59 M87 28 7-3-08 07-31-08 to 12-06-11 1251 M88 28 7-3-08 07-31-08 to 11-30-10 880 F89 M90 28 36 7-3-08 7-9-08 07-31-08 08-14-08 Male 7A 28-35 7-10-08 ~08-07-08 to 08-14-08 28 to 35 Male 7B 28-35 7-10-08 ~08-07-08 to 08-14-08 28 to 35 Female 7C 28-35 7-10-08 ~08-07-08 to 08-14-08 28 to 35 Radio-collared 7-11-08 to 7-22-08; collar removed because of malfunction. Not radio-collared after 7-22-08. Eartag of M84 was found by E. Phillips on 8-25-08 when mother F70’s GPS locations located her on either side of the eartag in the East fork Dolores Cyn. M84 recaptured radiocollared again 1-29-09 in Dolores Cyn. in association with F70 & F96’s family. Shed radiocollar again about 0214-09. Radio-collared. Dead. Probably died of predation or infanticide about 10-108 near elk calf kill at age 3.9 months. Radio-collared 7-22-08. Dead. Radio-collar, orange ear-tag #86 with pinna with green tattoo #86 found by J. Timmer 9-1-08. F86 died ~7-23 to 8-3-08 when mother F70’s GPS locations located her at F86 remains. Probable predation. Not radio-collared. Dispersed from natal area. Recaptured as adult on west slope of study area on 02-09-11 at 31 months old. Killed by puma hunter on 12-06-11 at 41 months old in GMU61N north of the study area. Not radio-collared. Dispersed. Killed by a puma hunter in Dawson Creek, Disappointment Valley, GMU711 on 11-3010 when 28.9 months old. Radio-collared. Radio-collared. Recaptured as young adult on study area, adjacent to natal area, on 11-16-10. Killed by a puma hunter during TY2 on 11-23-10. Not radio-collared. F7’s cubs died from starvation after they were orphaned. F7 was shot on 8-3-08 for killing domestic sheep. Not radio-collared. F7’s cubs died from starvation after they were orphaned. F7 shot on 8-3-08 for killing domestic sheep. Not radio-collared. F7’s cubs died of starvation after orphaned. F7 shot on 8-3-08 for killing domestic sheep. 867 Mother I.D. F70 F70 F70 F3 F3 F3 F72 F7 F7 F7 �Appendix A continued Puma I.D. Estimated Age at capture (days) M91 35 M92 Est. Birth date Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 8-19-08 Est. survival span from 1st capture to fate or last monitor date 09-29-08 455 35 8-19-08 09-29-08 976 F95 16 mo. June-07 12-29-08 F98 4-5 mo. 158 02-12-09 to 03-08-09 2-27-09 to 01-2010 146-176 M99 Sep-Oct08 Sep-Oct08 M101 35 4-15-09 157 M102 35 4-15-09 05-20-09 to 09-19-09 05-20-09 F103 35 4-15-09 159 M105 38 5-7-09 F106 38 5-7-09 05-20-09 to 09-17-09 06-14-09 to 02-09-10 06-14-09 to 02-27-11 M107 34 5-25-09 Radio-collared. Killed by a puma hunter on study area during TY1 as dependent cub on 11-17-09 at age 14.9 months. Radio-collared. Lost contact after 12-12-08. Dispersed from natal area. Recaptured in McKenzie Creek, west slope of study area on 04-22-11 when 32 months old. Due to dangerous tree, could not handle him safely to fit new radiocollar. Radio-collared. Survived to adult stage. Established adult home range overlapping mother F93’s home range. To date, July 2012, F95’s home range mainly adjacent to N side of natal area. Radio-collared. Died; probably killed by male puma (infanticide). Radio-collared. Offspring of non-marked female. Last location 4-22-09 on Paterson Mt. Died as 16-month old subadult in San Miguel Canyon. Probably killed by another puma; apparent canine punctures to braincase. Radio-collared. Died; killed by puma M55 after he was orphaned due to death of mother F16 by vehicle strike. Radio-collared. Lost contact after 09-04-09. Did not find evidence of M102 associated with deaths of siblings M101 and F103. But M102 probably died. Radio-collared. Died; killed by puma M55 after she was orphaned due to death of mother F16 by vehicle strike. Radio-collared. Lost contact after 02-09-10 due to shed collar. Not radio-collared at nursery; F75 returned to nursery during handling. Radio-collared later on 2-10-10. Lost contact due to shed collar 3-16 to 29-10. F106 dispersed from natal area and was photographed at 21 months old at camera and scent-rub station on east slope of Uncompahgre Plateau on 02-27-11. Not radio-collared; too small. Recaptured 02-24-10; not collared. Dispersed. Killed by a puma hunter in Cottonwood Creek near Molina, CO on 12-09-10 when he was 19 months old. 06-28-09 to 02-24-10 488 278 275 661 241 Mother I.D. F25 F25 F93 Unm.F Unm.F F16 F16 F16 F75 F75 F94 �Appendix A continued Puma I.D. Estimated Age at capture (days) F108 34 Est. Birth date 5-25-09 Est. survival span from 1st capture to fate or last monitor date 06-28-09 to 03-05-10 Age to last monitor date alive or at death (days, birth to fate) 553 M109 M112 34 145 5-25-09 8-31-09 06-28-09 05-04-10 to 01-06-13 1,225 M115 427 Nov.-08 07-21-10 610 M117 193 Aug.-09 02-05-10 to 01-01-11 518 P1016(M) 39 6-12-10 06-12-10 to 07-21-10 39 P1017(M) 39 06-12-10 06-12-10 to 07-21-10 39 M120 30 06-28-10 07-28-10 to 12-02-10 526 M121 30 06-28-10 273 M122 35 07-8-10 07-28-10 to 03-28-11 08-12-10 to 04-28-11 F123 29 07-15-10 217 F124 29 07-15-10 08-13-10 to 02-17-11 08-13-10 to 02-16-11 931 216 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. Shed radiocollar at nursery; fastener failed. Recaptured and re-collared 2-24-10. Shed collar ~3-5-10. Dispersed from natal area. Killed by a puma hunter on the study area during TY2 on 11-29-11 at 18.1 months old. Not radio-collared; too small. Radio-collared. Lost contact after 05-4-10 (last live signal) possibly due to failed transmitter. Recaptured and re-radiocollared on 01-24-11. Independent subadult during 02-10-11 to 04-18-11. Lost contact after 04-18-11. Dispersed. Killed by a puma hunter 01-06-13 in GMU73 SE of Dolores, CO; age 41 months. Radio-collared. M115 died as a subadult (~20 mo. old) due to complications of a broken left foreleg (natural cause). Radio-collared. Lost contact after 5-14-10 (last live signal); shed collar found on 7-15-10 in the natal area. Killed by a puma hunter on the natal area in Beaver Creek, GMU70E, off the U.P. study area on 01-01-11 when he was 17 months old. Not radio-collared. Monitored at nursery via mother’s GPS/VHF collar. Found dead at nursery due to infanticide by puma M32 on same day as our investigation of nursery. Not radio-collared. Monitored at nursery via mother’s GPS/VHF collar. Found dead at nursery due to infanticide by puma M32 on same day as our investigation of nursery. Radio-collared. Lost radio contact after 12-02-10. Killed by a puma hunter on his natal area on 12-06-11 when he was 17.2 months old. Radio-collared. Lost radio contact after 03-28-11. F94 Radio-collared. Lost radio contact after 04-28-11. Tracks of 2 other siblings of M122 observed on 01-11-11 (neither cub marked). M122 killed by a puma hunter in Tatum Draw, Dry Fk. Escalante Cr., GMU62N, 01-23-13; age 30 months. Radio-collared. Killed on 02-17-11 for depredation control on domestic elk by Wildlife Services agent. Radio-collared. Killed on 02-16-11 for depredation control on domestic elk by elk farm manager. F104 F94 F70 F28 F119 F72 F72 F3 F3 F94 F94 �Appendix A continued Puma I.D. Estimated Age at capture (days) M125 29 07-15-10 M126 28 08-08-10 M127 28 M128 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 08-13-10 to 02-01-11 09-05-10 to 01-08-12 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 201 08-08-10 09-05-10 to 09-10-11 398 28 08-08-10 198 F129 35 08-21-10 09-05-10 to 02-22-11 09-25-10 to 02-02-12 M130 35 08-21-10 09-25-10 to 02-02-12 M131 35 08-21-10 09-25-10 to 07-21-11 F132 35 08-21-10 09-25-10 35 M134 ~18 mo. ~June-09 12-14-10 to 06-10-11 740 M139 36 04-18-11 05-24-11 to 07-29-11 102 F148 36 04-18-11 05-24-11 to 07-29-11 102 Radio-collared. Killed on 02-01-11 for depredation control on domestic elk by Wildlife Services agent. Radio-collared. Lost radio contact after 03-17-11; shed his radiocollar at a mule deer cache. Dispersed from natal area. Killed by a puma hunter on 01-08-12 in Tuttle Draw WNW of Nucla, CO, GMU61N, as 17-month-old subadult. Radio-collared. Lost radio contact after 07-01-11; shed his radiocollar about 07-01-11. Found dead 09-14-11 on natal area; killed by another puma on about 09-10-11 at age 13 months. Radio-collared. Lost radio contact after 02-22-11; radiocollar probably quit. Radio-collared. Fate unknown. Transmitter on mortality mode on 04-28-11. Unable to get to collar until 06-23-11 due to high spring run-off, by then the transmitter had quit. Survived to recapture on 02-02-12 at 17.4 months old, with sibling M131; neither handled due to dangerous trees. Radio-collared. Died of natural causes associated with injury to right shoulder during first move away from nursery about 10-23-10. Radio-collared. Lost contact after 07-21-11. Shed his radiocollar about 07-27-11. Survived to recapture on 02-0212 at 17.4 months old, with sibling F129; neither handled due to dangerous trees. Dispersed. Killed by a puma hunter in Lion Cr., extreme western CO, GMU60; age 29 months. Not radio-collared. Too small for collar design. Fate unknown. Apparently died; not with F96 and siblings F129 and M130 on 02-02-12. Radiocollared as dependent large cub. Independent by about 03-28-11. Dead; killed for depredation control by Wildlife Services agent on 06-10-11. He was about 24 mo. old Radio-collared. Dead of infanticide and cannibalism along with sibling F148; killed and eaten by female or subadult male puma about 07-29-11. Radio-collared. Dead of infanticide and cannibalism along with sibling M139; killed and eaten by female or subadult male puma about 07-29-11. 221 530 530 334 Mother I.D. F94 F118 F118 F118 F96 F96 F96 F96 Unm. F F8 F8 �Appendix A continued Puma I.D. Estimated Age at capture (days) F140 ~5 mo. Est. Birth date Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. ~Aug.10 Est. survival span from 1st capture to fate or last monitor date 01-02-11 to 07-31-13 1,096 Radio-collared. Lost contact. Shed first collar about 01-2411. Recaptured and re-collared on 04-01-11. Shed second collar after 04-18-11. Recaptured and re-collared 01-12-12 as 17-month-old subadult on natal range. Survived to adult stage. Radio-collared. Lost contact; shed radiocollar about 03-2911. Recaptured, but could not be handled safely on 04-0111. Killed by a puma hunter on 12-23-11 in his natal area; age 16 months. Radio-collared. Lost contact after 04-18-11 due to shed collar. Struck by vehicle and killed on state highway 62 in Leopard Creek, south boundary of study area on 02-16-11. Radio-collared. Orphaned at about 12 months old when her mother F24 was killed by a male puma on 09-16-11. She ranged in her natal area until her radiocollar quit after 0412-12. Radio-collared. F149 (sibling of M161) was orphaned when her mother F23 was killed by a male puma on 06-06-12; she was 411 days (13.5 mo.) old. F149 dispersed from the natal area by 07-16-12 when she was 14.8 months old. Radio-collared. M151 was independent by 03-28-11 at 19 mo. old. He dispersed from the natal area by 04-11-11 at 19.5 mo. old. Contact lost after 04-11-11. Radio-collared. Lost contact after 03-07-11 (GPS location of mother F111 at shed collar of M151). Radio-collared. Lost contact after 03-21-11; shed collar. Recaptured 01-18-12; fit with GPS collar at 19 months old. Ranged on natal area as adult (philopatric). First litter on 0808-12 at 26 mo. old. Killed by puma hunter on 12/23/12. Radio-collared. M154 probably died of starvation following natural death of his mother F135. Sibling M155 also died. Radio-collared. M155 died of starvation following death of his mother F135. Sibling M154 also died. Radio-collared. M156 shed the collar about 09-05-11. He was 59 days old. Unk./ F28? M141 ~5 mo. ~Aug.10 01-02-11 to 04-01-11 509 M142 ~5 mo. ~ 6 mo. 01-02-11 to 04-18-11 02-16-11 258 P1030 F147 ~7 mo. ~Aug.10 ~Aug.10 ~Sep.-10 04-21-11 to 07-31-11 315 F149 45 04-22-11 06-06-11 to 07-16-12 451 M150 525 08-31-09 02-07-11 to 04-11-11 588 M151 253 06-16-10 264 F152 271 06-16-10 02-24-11 to 03-07-11 03-14-11 to 12-23-12 M154 42 07-06-11 77 M155 42 07-06-11 M156 43 07-08-11 08-16-11 to 09-21-11 08-16-11 to 09-25-11 08-20-11 to 09-05-11 183 776 81 56 Unk./ F28? Unk./ F28? Unk. F24 F23 F70 F111 F93 F135 F135 F137 �Appendix A continued Puma I.D. Estimated Age at capture (days) F157 40 08-18-11 F158 40 M159 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 09-27-11 to 01-15-12 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 150 08-18-11 09-27-11 to 01-15-12 150 40 08-18-11 09-27-11 to 12-01-11 105 M161 276 04-22-11 01-23-12 to 10-15-12 543 M162 183 07-25-11 01-25-12 to 06-11-12 322 M168 37 07-27-12 09-02-12 to 09-12-12 47 F169 37 07-27-12 09-02-12 to 09-12-12 47 M170 137 08-29-11 199 P1033 22 07-10-11 01-13-12 to 03-12-12 NA Radio-collared. F157 with sibling F158 died of starvation following death of his mother F70 due to hunter harvest on 12-22-11. Cubs died 24 days after their mother died. The cubs were 150 days old. Radio-collared. F158 with sibling F157 died of starvation following death of his mother F70 due to hunter harvest on 12-22-11. Cubs died 24 days after their mother died. The cubs were 150 days old. Radio-collared. M159 probably died about 12-01-11 when he was located with his family (F70, siblings F157, F158). He was not located with them on 12-12-11 and was not observed with them on 12-13-11. He was 105 days old on 12-01-11. Radio-collared. M161 (sibling of F149) was orphaned when his mother F23 was killed by a male puma on 06-06-12; he was 411 days (13.5 mo.) old. M161 dispersed from the natal area by 06-29-12 when he was 14 months old. Shed his expandable collar about 08-03-12. Was struck and killed by a vehicle on Dallas Divide, Hwy 62 in October 2012 when 18 mo. old. Radio-collared. M162 probably was orphaned cub of nonmarked adult female puma killed on Pinto Mesa 01-18-12. M162 died of starvation on 06-11-12 when he was 322 days (10.6 mo.) old. Radio-collared. Cub M168 was offspring of F96; sibling of F169 & F173. It died of infanticide, probably of a male puma, based on track sizes (fhpw = 60 mm; hhpw = 50 mm). Radio-collared. Cub F169 was offspring of F96; sibling of M168 & F173. It died of infanticide, probably of a male puma, based on track sizes (fhpw = 60 mm; hhpw = 50 mm). Radio-collared. M170 died about 03-15-12 of unknown natural cause. He was 199 days (6.5 mo.) old. Radio-collared. Cub P1033 was offspring of F136. It died of predation, probably killed by a puma or black bear in the nursery when about 22 days old, before researchers could examine the entire litter to sample and mark the cubs. 22 Mother I.D. F70 F70 F70 F23 Unm.F F96 F96 F171 F136 �Appendix A continued Puma I.D. Estimated Age at capture (days) F173 37 07-27-12 M174 32 M175 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 09-02-12 to 09-12-12 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 47 08-08-12 09-11-12 to 03-10-13 181 32 08-08-12 09-11-12 to 12-11-12 126 F184 208 08-25-12 03-20-13 to 07-29-13 339 F185 ~183 ~Oct.2012 03-23-12 to 03-27-13 190 F187 31 05-14-13 77 F188 31 05-14-13 F189 38 06-18-13 M191 ~183 ~July 2012 06-14-13 to 07-29-13 06-14-13 to 07-29-13 07-26-13 to 07-31-13 01-03-13 to 01-20-13 PM1068 ~183 ~July 2012 01-03-13 to 01-20-13 ~210 M192 199 06-20-12 01-04-13 to 07-01-13 376 Radio-collared. Cub F173 was offspring of F96; sibling of M168 & F169. It died of infanticide, probably of a male puma, based on track sizes (fhpw = 60 mm; hhpw = 50 mm). Radio-collared. Cub M174 was offspring of F152; sibling of M175. He was orphaned after his mother was killed by a hunter on 12-23-12. He was 137 days old. M174 was recaptured at 181 days old and removed from the wild to be rehabilitated at the CPW Del Norte wildlife center for rerelease to the wild at a later date. Radio-collared. Cub M175 was offspring of F152; sibling of M174. He was mauled to death probably by puma hunting dogs on about 12-11-12 when he was 126 days old. Radio-collared. Cub F184 was offspring of F111; one other sibling track was observed, but the puma was not captured. F184 still dependent on F111 on 07-29-13. Radio-collared. Cub F185 was offspring of a non-marked female puma in Roubideau Cyn. F185 shed her expandable collar about 7 days after initial capture. Lost contact. Fate unknown. Radio-collared. Cub F187 was offspring of F96; sibling of F188. Radio-collared. Cub F188 was offspring of F96; sibling of F187. Radio-collared. Cub F189 was offspring of F136; sibling of F200 and M201. Radio-collared. Cub M191 apparently was offspring of F28 (with non-functional GPS collar). He was sibling of PM1068 and one other non-marked cub. M191 was killed by a non-marked male puma on about 01-20-13 along with PM1068. PM1068 was not captured and tagged. It was apparently offspring of F28; sibling of M191 and one other nonmarked cub. PM1068 was killed and partially eaten by a non-marked male puma. Radio-collared. M192 was offspring of F118; sibling of M193 & F195. M192 was independent of F118 at ~11.7 mo. old. He shed his expandable collar at a mule deer kill after 07-01-13. 77 44 ~210 Mother I.D. F96 F152 F152 F111 Unm.F F96 F96 F136 F28 F28 F118 �Appendix A continued Puma I.D. Estimated Age at capture (days) M193 199 Est. Birth date 06-20-12 Est. survival span from 1st capture to fate or last monitor date 01-04-13 to 07-01-13 Age to last monitor date alive or at death (days, birth to fate) 376 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. Radio-collared. M193 was offspring of F118; sibling of F118 M192 & F195. M192 was independent of F118 at ~11.7 mo. old. He shed his expandable collar at a mule deer kill after 07-01-13 like sibling M192, but the siblings were not associating (kills were at different locations). F195 227 06-20-12 02-01-13 to 258 Radio-collared. F195 was offspring of F118; sibling of F118 03-04-13 M192 & M193. F195 shed her expandable radiocollar at an elk kill on about 03-04-13; contact lost afterwards. M198 274 ~June 04-10-13 to 417 Radio-collared. M198 was offspring of non-marked female PF1074 2012 07-31-13 PF1074 (sampled by bio-dart). He was sibling of F199. F199 292 ~June 04-18-13 to 417 Radio-collared. F199 was offspring of non-marked female PF1074 2012 07-31-13 PF1074 (sampled by bio-dart). She was sibling of M198. F200 38 06-18-13 07-26-13 to 44 Radio-collared. Cub F200 was offspring of F136; sibling of F136 07-31-13 F189 and M201. M201 38 06-18-13 07-26-13 to 44 Radio-collared. Cub M201 was offspring of F136; sibling of F136 07-31-13 F189 and F200. F202 35 06-25-13 07-30-13 to 36 Radio-collared. Cub F202 was offspring of F172. No F172 07-31-13 siblings were observed at the nursery; but some could have hidden. a Cub F35 probably starved between 06-30-06 & 07-07-06 after the transmitter on the expandable collar got in its mouth. b Cub M42 died after being captured by dogs, probably from stress of capture associated with severe infection of laceration under right foreleg caused by expandable radiocollar. c Cub M56 was captured in association with F7 and her cubs M43 and M44. He may have been missed at the nursery when M43 and M44 were initially sampled and marked. d Cub M60 died probably of starvation. The expandable radiocollar was around the neck and right shoulder, probably restricted movement. �Colorado Division of Parks and Wildlife July 2013 –June 2014 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 3003 1 Federal Aid Project: W-204-R1 : : : : : Division of Parks and Wildlife Mammals Research Carnivore Conservation Puma Population Structure and Vital Rates on the Uncompahgre Plateau Period covered: July 31, 2013−June 30, 2014 Author: Kenneth A. Logan. Personnel: K. Logan, R. Alonso, S. Bard, Yasaman Shakeri, W. Hollerman, W. Jesson, R. Navarrete, B. Nay, S. Waters, B. Banulis, T. Bonacquista, M. Caddy, K. Crane, E. Phillips, and G. Watson of CPW; volunteers and cooperators including: private landowners, Bureau of Land Management, Ridgway State Park, Colorado State University, and U.S. Forest Service. Supplemental financial support received in previous years from The Howard G. Buffett Foundation, Safari Club International Foundation, and The Summerlee Foundation. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT The Colorado Parks and Wildlife (CPW) initiated a 10-year study on the Uncompahgre Plateau in 2004 to quantify puma population characteristics in the absence (reference period, years 1-5) and presence (treatment period, years 6-10) of sport-hunting. The purpose of the study is to evaluate the assumptions underlying the CPW puma management program with sport-hunting in Colorado. The reference period began December 2004 and ended July 2009, during which we captured, sampled, and marked 109 pumas for research purposes on the Uncompahgre Plateau (Logan 2009). This report provides information on the fifth year of the treatment period (TY5), August 2013 through July 2014, on puma population characteristics and dynamics with hunting as a mortality factor. Puma sport-hunting opened November 18, 2013 and closed January 10, 2014 after a quota of 5 independent pumas were killed. The harvest was designed to test the management assumption that an 815% harvest of independent pumas results in a stable-to-increasing population. The decline in the puma population on the study area during TY1 to TY3 necessitated a reduction in the harvest quota from 8 to 5 to continue to test the harvest assumption for a stable-to-increasing puma population. The five pumas killed included, 1 adult female, 3 adult males, and 1 subadult male. The harvest of five independent pumas represented 11.4% of the 44 independent pumas in our minimum count during November 2013 to April 2014. Independent females and males comprised 20.0% and 80.0% of the harvest, respectively. One other radio-collared adult female in the study area population died of natural cause in November 2013. The total mortality of 6 independent pumas during the TY5 minimum population count (November through April) represented 13.6% of the 44 minimum count of independent pumas on the study area. Sixty-five hunters requested mandatory permits with an attached voluntary hunter survey in TY5. Forty-four of the hunters provided responses on the surveys. An estimated 37 hunters actually hunted on the study area, of which about 13.5% harvested pumas and 27.0% captured pumas (i.e., harvested plus 1 �treed and released). Twenty-seven of 29 answering hunters responded that they were selective hunters, and the capture, tracking, and population data indicated that most hunters practiced selection. Puma tracks <1 day old encountered by hunters and pumas captured by hunters indicated that independent female pumas were detected more frequently than males in TY5. Researchers captured 38 individual pumas 43 times from August 2013 to July 2014. Two capture teams with dogs operated during 75 search days from January 9, 2014 through April 24, 2014 to find 361 puma tracks, pursue pumas 82 times, and capture 24 pumas 29 times. Capture efforts with cage traps resulted in the capture of 2 independent pumas for the first time, and the recapture of 1 adult male. Twelve new cubs were captured and marked. A total of 56 pumas were monitored by radio-telemetry in TY5. Search efforts also revealed the presence of at least 13 other independent pumas. Our minimum count of 44 independent pumas from November 2013 to April 2014 included: 27 females and 17 males. A preliminary minimum estimated density of independent pumas was 2.63/100 km2. The TY5 minimum count of 44 was up slightly from 42 in TY4 and could be attributed partially to the reduction in harvest from 15% to 11% in TY4 and to immigration especially of subadult male pumas on the east slope of the study area. The proportion of radio-collared adult females giving birth in the August 2012 to July 2013 biological year was 0.33 (5/15). Since 2005 puma births occurred mainly from May through August, involving 86% of births. We monitored 22 female and 7 male adult radio-collared pumas for survival and agent-specific mortality. Adult puma survival rates in TY5 for adult females and males were 0.678 (SE=0.0934) and 0.667 (SE=0.1721), respectively. Sport-hunting mortality was the major cause of death. One adult female was struck and killed by a vehicle on state highway 550. One adult female was killed for depredation control. Of 21 radio-collared cubs in TY5, 14 were monitored continuously. Of those 14, six died. Causes of death included: starvation (3), infanticide or predation (1), and unknown natural causes (2). Puma harvest, capture, and radio-telemetry data from the beginning of this study to the present provided information on dispersals of 41 pumas initially marked on the study area. Those pumas moved from about 18 to 370 km from initial capture sites. Since the start of this study 48 adult pumas have been monitored with GPS collars and have yielded over 70,000 locations. In addition, 45 subadult pumas have been monitored for survival and fate data. Efforts to develop and test puma population survey methods continued with a pilot project started in June 2014. We are evaluating the use of temporary foot-hold devices and break-away neck snares as passive devices that non-invasively collect puma hair for DNA genotyping and mark-recapture population estimation methods. Those efforts will continue into November 2014. 2 �WILDLIFE RESEARCH REPORT PUMA POPULATION STRUCTURE AND VITAL RATES ON THE UNCOMPAHGRE PLATEAU, COLORADO KENNETH A. LOGAN PROJECT NARRATIVE OBJECTIVE Quantify puma population sex and age structure; estimate puma population vital rates, including: reproduction of females, stage-specific survival, and immigration and emigration; quantify agent-specific mortality rates; model puma population dynamics; develop and execute the puma harvest manipulation to begin the population-wide test of Colorado Parks and Wildlife (CPW) puma management assumptions in the fifth year of a five-year Treatment Period of the Uncompahgre Plateau Puma Project― all to improve the CPW model-based approach to managing pumas in Colorado. SEGMENT OBJECTIVES 1. Execute the fifth year of the five-year treatment period by working with CPW biologists and managers to manipulate the puma population with sport-hunting and to survey hunters. 2. Continue gathering data on puma population sex and age structure. 3. Continue gathering data for estimates of puma reproduction rates. 4. Continue gathering data to estimate puma sex and stage-specific survival rates. 5. Continue gathering data on agent-specific mortality. 6. Explore non-invasive methods for sampling pumas to estimate abundance. INTRODUCTION Wildlife managers need reliable information on puma biology and ecology in Colorado to develop sound management strategies that address diverse public values and the CPW objective of “achieving healthy, self-sustaining populations” through management (Colorado Division Of Wildlife 2002-2007 Strategic Plan:9). Although 4 puma research efforts have been conducted in Colorado since the early 1970s and puma harvest data is compiled annually, additional information on certain aspects of puma biology and ecology, and management tools that may guide managers toward effective puma management is needed. Mammals Research staff held scoping sessions with a number of the CPW’s wildlife managers and biologists prior to initiating the project. In addition, we consulted with other agencies, organizations, and interested publics either directly or through other CPW employees. In general, CPW staff in western Colorado highlighted concern about puma population dynamics, especially as they relate to their abilities to manage puma populations through regulated sport-hunting. Secondarily, they expressed interest in pumaprey interactions. Staff on the Front Range placed greater emphasis on pumahuman interactions. Staff in both eastern and western Colorado cited information needs regarding effects of puma harvest, puma population monitoring methods, and identifying puma habitat and landscape linkages. Management needs identified by CPW staff and public stakeholders form the basis of Colorado’s puma research program, with multiple lines of inquiry (i.e., projects): Improve our ability to manage puma hunting with enhanced scientific bases, strategies, and tools― ● Puma population characteristics (i.e., density, sex and age structure). ● Puma population dynamics and vital rates (i.e., birth rates, survival rates, emigration rates, immigration rates, population growth rates). 3 �● Field methods and models for assessing and tracking changes in puma populations. ● Relative vulnerability of puma sex and age classes to hunter harvest. Improve our understanding of puma habitat needs and interrelationships of puma management units― ● Puma habitat use, movements, and use of landscape linkages. ● Puma recruitment patterns (i.e., progeny, immigration, emigration). ● Models for identifying puma habitat and landscape linkages. Improve our understanding of the puma’s role in the ecology of other species ● Relationships of puma to mule deer, elk, and other natural prey. ● Relationships of puma to species of special concern, e.g., desert bighorn sheep. Improve our understanding of puma-human interactions and abilities to manage them ● Behavior of puma in relation to people and human facilities. ● Puma predation on domestic animals. ● Effects of translocating nuisance pumas. ● Effects of aversive conditioning on pumas. While all projects cannot be addressed concurrently, understanding their relationships to one another is expected to help individual projects maximize their benefits to other projects that will assist the CPW to achieve its strategic goal in puma management (Fig.1). This project has been addressing all of the grayshaded components on the left side of the conceptual model in Figure 1 that pertain to the puma population, including: effects of harvest and other mortality, movements and corridors, population dynamics, vulnerability to harvest, population models, and methods for monitoring populations. Management issues identified by managers translate into researchable objectives, requiring descriptive studies and field manipulations. Our goal is to provide managers with reliable information on puma population biology and to develop useful tools for their efforts to adaptively manage puma in Colorado to maintain healthy, self-sustaining populations. The highest-priority management needs are being addressed with this intensive population study that focuses on puma population dynamics using sampled, tagged, and GPS/VHF-radio-collared pumas to investigate the effects of sport-hunting and other causes of mortality on puma population dynamics. Those objectives include: 1. Describe and quantify puma population sex and age structure. 2. Estimate puma population vital rates, including: reproduction rates, age-stage survival rates, emigration rates, immigration rates. 3. Estimate agent-specific mortality rates. 4. Improve the CPW’s puma model-based management and attendant assumptions with Coloradospecific data from objectives 13. Consider other useful models. 5. Conduct a pilot study to develop methods that yield reliable estimates of puma population abundance. 6. Investigate diseases in pumas. A descriptive and manipulative study will estimate population parameters in an area that appears typical of puma habitat in western Colorado and will yield defensible population parameters based upon contemporary Colorado data. This study will be conducted in two 5-year periods. A completed 5-year reference period, 2004-09, (i.e., absence of recreational hunting) allowed puma life history traits to interact with the main habitat factors that influenced puma population growth (e.g., prey availability and vulnerability, Pierce et al. 2000, Logan and Sweanor 2001, Logan 2009). A subsequent 5-year treatment period started in 2009-10 which involves the use of controlled recreational hunting to manipulate the puma population. 4 �TESTING ASSUMPTIONS AND HYPOTHESES Hypotheses associated with main objectives 15 of this puma population research are structured to test assumptions guiding puma management in Colorado. 1. Considering limitations (i.e., methods, number of years, assumption violations) to the previous Colorado-specific studies on puma populations (Currier et al. 1977, Anderson et al. 1992, Koloski 2002), managers assume that puma population densities in Colorado are within the range of those quantified in more intensively studied populations in Wyoming (Logan et al. 1986), Idaho (Seidensticker et al. 1973), Alberta (Ross and Jalkotzy 1992), and New Mexico (Logan and Sweanor 2001). The CPW assumes density ranges of 2.0−4.6 puma/100 km2 (i.e., includes pumas of all age classes - adults, subadults, and cubs, J. Apker, CPW Carnivore Biologist, person. commun. Nov. 19, 2003) to extrapolate to Data Analysis Units (DAUs) to guide the model-based quota-setting process. Likewise, managers assume that the population sex and age structure is similar to puma populations described in the intensive studies. Using intensive efforts to capture, mark, and estimate non-marked animals developed and refined during the study to estimate the puma population, the following will be tested: H1: Puma densities during the 5-year reference period (absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will vary within the range of 2.0 to 4.6 puma/100 km2 and will exhibit a sex and age structure similar to puma populations in Wyoming, Idaho, Alberta, and New Mexico. 2. Recreational puma hunting management in Colorado DAUs is guided by a model to provide allowable harvest quotas in an effort to achieve one of two puma population objectives: 1) maintain puma population stability or growth, or 2) cause puma population decline (CDOW, Draft L-DAU Plans, 2004, CDOW 2007). These objectives are expected to provide both the capacity for puma population resiliency to achieve a state-wide goal of a healthy, self-sustaining puma population while managing the puma population to provide sport-hunting opportunity and population control in some DAUs (even though puma population dynamics in any DAUs are not known). Basic model parameters assigned to the model are: puma population density, sex and age structure, annual population growth rate, and relative puma habitat quality and quantity. Parameter quantities are currently chosen from literature on studies in western states that are judged to provide reliable information. Background material used in the model assumes a moderate annual rate of growth of 15% (i.e., for the adult and subadult puma population (CDOW 2007). This assumption is based upon information with variable levels of uncertainty (e.g., small sample sizes, data from habitats dissimilar to Colorado). Parameters influencing  include population density, sex and age structure, female age-at-first-breeding, reproduction rates, sex- and age-specific survival, immigration and emigration. H2: Population parameters estimated during a 5-year reference period (in absence of recreational puma hunting) in conifer and oak communities with deer, elk and other prey populations typical of those communities in Colorado will yield an estimated annual adult plus subadult population growth rate that will match or exceed  = 1.15. 3. An assumption is that the CPW can manage puma population growth through recreational hunting on the basis that for a stable puma population hunting removes the annual increment of population growth (i.e., from current judgments on population density, structure, and Puma harvest rate formulations for DAUs assume that total mortality (i.e., harvest plus other detected deaths) in the range of 8 to 15% of the harvest-age population (i.e., independent pumas comprised of adults plus 5 �subadults) with the total mortality comprised of 35 to 45% females (i.e., adults and subadults) is acceptable to manage for a stable-to-increasing puma population (CDOW 2007). This assumption is vital to providing the capacity for resiliency in the state-wide puma population which is hunted by applying this assumption to about three-quarters of the puma GMUs in the state. H3: Total mortality of an estimated 15% of the adults and subadults with no more than 45% of the total mortality comprised of females will not result in a declining trend of the harvest-age segment of the population. 4. To reduce a puma population, hunting must remove more than the annual increment of population growth. For DAUs with the objective to suppress the puma population, the total mortality guide of greater than 15 to 28% of the harvest-age population with greater than 45% comprised of females is suggested (CDOW 2007). This assumption is applied to about one-quarter of the GMUs in the state. H4: Total mortality of an estimated 16% or greater of the harvestable population with greater than 45% females will cause a declining trend in the abundance of harvest-age pumas (i.e., adults and subadults). 5. The increase and decline phases of the puma population make it possible to test hypotheses related to shifts in the age structure of the population which have been linked to harvest intensity in Wyoming and Utah. H5: The puma population on the Uncompahgre Plateau study area will exhibit a young age structure after hunting prohibition at the beginning of the reference period. During the 5 years of hunting prohibition, greater survival of independent pumas will cause an older age structure in harvest-age pumas (i.e., adults and subadults) as suggested by the work of Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. As hunting is re-instated in the treatment period, the age structure of harvested pumas and the harvest-age pumas in the population will decline as observed by Anderson and Lindzey (2005) in Wyoming and Stoner (2004) in Utah. 6. Researchers in Wyoming (Anderson and Lindzey 2005) concluded that sex and age composition of the harvest varies predictably with puma population size because the likelihood of a specific sex or age class of puma being harvested with the use of hounds is a product of the relative abundance of particular sex and age classes in the population and their relative vulnerability to harvest. Results of that study suggest that managers could use sex and age composition of the harvest to infer puma population changes (Anderson and Lindzey 2005). The CPW currently uses this approach as one tool to infer potential DAU puma population dynamics (CDOW 2007). This assumes no purposeful selection by hunters for any particular sex or age-stage other than the puma must be legal (i.e., independent subadult or adult, not a lactating female or a female in association with spotted cubs) and that changes in the sex and age structure of the harvested pumas is due solely to changes in the relative abundance of particular sex and age classes in the population and their relative vulnerability to harvest as predicated by puma movement patterns. It was assumed that pumas that traveled longer distances with movements that intercept access routes used by hunters (i.e., roads, trails) should be more exposed to detection by hunters and thus more vulnerable to harvest. A key assumption to this method is that pumas are killed as they are encountered and the harvest sex and age composition will reliably indicate whether a population is stable, increasing, or declining even if harvest intensity does not vary. Thus, an alternate view is that a population segment, such as independent females, may be more abundant and have shorter movement lengths, yet be detected more frequently by hunters. However, because the same intensively studied Wyoming puma population was manipulated over 5 years with varying intensities of harvest (Anderson and Lindzey 2005), variations in harvest structure using the same harvest level over a period of years could not be examined. This is a property we will 6 �investigate during the treatment period on the Uncompahgre Plateau puma study. Moreover, we will directly evaluate to what extent puma harvest might be influenced by hunter selection. A hunter survey is intended to reveal puma hunter behavior, detection of different classes of pumas, and lack of or presence of hunter selection. These data should allow us to examine the credibility of the assumption of non-selection by hunters and the robustness of this technique in gauging puma population dynamics by using harvest composition as an indicator. We want to examine the usefulness of this approach in Colorado. CPW managers attempt to weight sport-harvest toward male pumas in GMUs with the stable-to-increasing population objective with an active educational program (i.e., mandatory hunter exam, brochure, workshops). Thus, there is a need to test assumptions associated with the Anderson and Lindzey (2005) method. H6: No hunter selection is practiced so that the sex and age structure of pumas harvested by hunters in this population protected from hunting during a 5-year reference period and subsequently managed for stability or increase with conservative harvest levels will reflect the relative vulnerabilities to detection and capture with dogs during each year in the 5-year treatment period in this order from high to low vulnerabilities: subadult males, adult males, subadult females, adult females without cubs or with cubs >6 months old, and adult females with cubs ≤6 months old (Barnhurst 1986, Anderson and Lindzey 2005). In each of the 5 years of the treatment period, subadults and adult males should comprise the majority of the harvest and reflect the assumed sex and age structure (Anderson and Lindzey 2005) of a puma population managed for a stable to increasing phase and not hunted for 5 previous years (i.e., a puma population source). Desired outcomes and management applications of this research include: 1. Quantification of variations in puma population density, sex and age structure, growth rates, vital rates, and an understanding of factors affecting them will aid adaptive puma management by yielding population parameters and tools useful for assessing puma population dynamics, evaluation of management alternatives, and effects of management prescriptions. 2. Testing assumptions about puma populations, currently used by CPW managers, will help managers to biologically support and adapt puma management based on Colorado-specific estimated puma population characteristics, parameters, and dynamics. 3. Methods for assessing puma population dynamics will allow managers to evaluate modeled populations and estimate effects of management prescriptions designed to achieve specified puma population objectives in targeted areas of Colorado. Ascertaining puma numbers and densities during the project will allow assessment of monitoring techniques. Potential methods include use of harvest sex and age structure and photographic and DNA genotype capture-recapture. Study plans to develop and test feasible field and analytical methods will be developed as we learn the logistics of performing those methods, after we have preliminary data on puma demographics and movements which will inform suitable sampling designs, and if we have adequate funding. 4. Information which will be disseminated to citizen stakeholders interested in pumas in Colorado, and thus contribute to informed public participation in puma management. STUDY AREA The study area for the puma population research is on the Uncompahgre Plateau (in Mesa, Montrose, Ouray, and San Miguel Counties; Fig. 2). The study area includes about 2,253 km2 (870 mi.2) of the southern halves of Game Management Units (GMUs) 61 and 62, and about 155 km2 (60 mi.2) of the northern edge of GMU 70 (between state highway 145 and San Miguel River). The area is bounded by state highway 348 at Delta, 25 Mesa road and Forest Service road FS503 to Nucla, state highway 97 to state highway 141 to state highway 145 to Placerville, state highway 62 to Ridgeway, U.S. highway 550 to Montrose, and U.S. highway 50 to Delta. 7 �The study area seems typical of puma habitat in Colorado that has vegetation cover that varies from the pinion-juniper covered foothills starting from about 1,700 m elevation to the spruce-fir and aspen forests growing to the highest elevations of about 3,000 m. Mule deer (Odocoileus hemionus) and elk (Cervus elaphus) are the most abundant wild ungulates available for puma prey. Cattle and domestic sheep are raised on summer ranges on the study area. People reside year-round along the eastern and western fringe of the area, and there is a growing residential presence especially on the southern end of the plateau. A highly developed road system makes the study area easily accessible for puma research efforts. A detailed description of the Uncompahgre Plateau is in Pojar and Bowden (2004). METHODS Reference and Treatment Periods This research was structured in two 5-year periods: a reference period (years 1―5) and a treatment period (years 6―10). The reference period was closed to puma hunting on the study area and was expected to cause a population increase phase. The treatment period (starting in November 2009) involves manipulation of the puma population with sport-hunting structured to achieve a management objective for a stable to increasing population. In both phases, puma population structure, and vital rates are being quantified, and management assumptions and hypotheses regarding population dynamics and effects of harvest are being tested. Contingent upon results of pilot studies, we will also assess enumeration methods for estimating puma population abundance. The reference period, without recreational puma hunting as a major limiting factor, was consistent with the natural history of the current puma species in North America which evolved life history traits during the past 10,000 to 12,000 years (Culver et al. 2000) that enable pumas to survive and reproduce (Logan and Sweanor 2001). In contrast, puma hunting, with its modern intensity and ingenuity, might have influenced puma populations in western North America for at least the past 100 years. Hence, the reference period, years 1 to 5, provided conditions where individual pumas in this population expressed life history traits interacting with the environment without recreational hunting as a limiting factor. Theoretically, the main limiting factor was vulnerable prey abundance (Pierce et al. 2000, Logan and Sweanor 2001). This allowed researchers to understand basic system dynamics before manipulating the population with controlled recreational hunting. In the reference period, all pumas in the study area were protected, except for individual pumas involved in depredation on livestock or human safety incidents. In addition, all radio-collared and ear-tagged pumas that ranged in a buffer zone in the northern halves of GMUs 61 and 62 were protected from recreational hunting mortality. The reference period allowed researchers to quantify baseline demographic data on the puma population to estimate parameters useful for assessing the CPW’s assumptions for its model-based approach to puma management. The reference period also facilitated other operational needs (because hunters did not kill the animals) including the marking of a large proportion of the puma population for parameter estimates and gathering movement data from GPS-collared pumas. During the treatment period, years 6 to10, recreational puma hunting is occurring on the same study area using management prescriptions structured from information learned during previous years. Using recreational hunting for the treatment is consistent with the CPW’s objectives of manipulating natural tendencies of puma populations, particularly survival, to maintain either population stability or increase or suppression (CDOW, Draft L-DAU Plans, 2004). Theoretically, survival of independent pumas is being influenced mainly by recreational hunting, which is being quantified by agent-specific mortality rates of radio-collared pumas. Dynamics of the puma population are being manipulated to evaluate hypotheses that are related to effects of hunting (i.e., effects of harvest rates, relative vulnerability of puma sex and age classes to hunting, variations in puma population structure due to 8 �hunting). The killing of tagged and collared pumas during the treatment period is not hampering operational needs (as it would have during the start-up years), because a majority of independent pumas in the population have already been marked, and sampling methods formalized. Pumas on the study area that may be involved in depredation of livestock or human safety incidences may be lethally controlled. Researchers that find that GPS-collared pumas have killed domestic livestock will record such incidents to facilitate reimbursement to the property owner for loss of the animal(s). In addition, researchers will notify the Area Manager of the CPW if they perceive that an individual puma may be a threat to public safety. Field Methods Puma Capture: Realizing that pumas live at low densities and capturing pumas is difficult, as a starting point, our logistical aim was to have a minimum of 6 puma in each of 6 categories (36 total) radio-tagged in any year of the study if those or greater numbers are present. The 6 categories are: adult female, adult male, subadult female, subadult male, female cub, male cub. This relatively large number of pumas might represent the majority of the puma population on the study area, and would provide the basic data for age- and sex-specific reproductive rates, survival rates, agent-specific mortality rates, emigration, and other movement data. Puma capture and handling procedures were approved by the CPW Animal Care and Use Committee (file #08-2004). All captured pumas were examined thoroughly to ascertain sex and describe physical condition and diagnostic markings. Ages of adult pumas were estimated initially by the gum-line recession method (Laundre et al. 2000) and dental characteristics of known-age puma (Logan and Sweanor, unpubl. data). Ages of subadult and cub pumas were estimated initially based on dental and physical characteristics of known-age pumas (Logan and Sweanor unpubl. data). Body measurements recorded for each puma included at a minimum: mass, pinna length, hind foot length, plantar pad dimensions. Tissue collections included: skin biopsy (from the pinna receiving the 6 mm biopsy punch for the ear-tags), and blood (30 ml from the saphenous or cephalic veins) for genotyping individuals, parentage and relatedness analyses, and disease screening; hair (from various body regions) for genotyping tests of field gathered samples. Universal Transverse Mercator Grid Coordinates on each captured puma were recorded via Global Positioning System (GPS, North American Datum 27). Pumas were captured year-round using 4 methods: trained dogs, cage traps, foot-hold snares, and by hand (for small cubs). Capture efforts with dogs were conducted mainly during the winter when snow facilitated thorough searches for puma tracks and enabled dogs to follow puma scent. The study area was searched systematically multiple times per winter by four-wheel-drive trucks, all-terrain vehicles, snowmobiles, and on foot. When puma tracks ≤1 day old were detected, trained dogs were released to pursue pumas for capture. Pumas usually climbed trees to take refuge from the dogs. Adult and subadult pumas captured for the first time or requiring a change in telemetry collar were immobilized with Telazol (tiletamine hydrochloride/zolazepam hydrochloride) dosed at 5 mg/kg based on estimated body mass (Lisa Wolfe, DVM, CPW, attending veterinarian, pers. comm.). The immobilizing agent was delivered into the caudal thigh muscles via a Pneu-Dart® shot from a CO2-powered pistol. Immediately, a 3m-by-3m square nylon net was deployed beneath the puma to catch it in case it fell from the tree. A researcher climbed the tree, fixed a Y-rope to two legs of the puma and lowered the cat to the ground with an attached climbing rope. Once the puma was on the ground, its head was covered, its legs tethered, and vital signs monitored (Logan et al. 1986). Normal signs include: pulse ~70 to 80 bpm, respiration ~20 bpm, capillary refill time ≤2 sec., rectal temperature ~101oF average, range = 95 to 104oF (Kreeger 1996). Pumas that climbed trees too dangerous for the pumas or researchers for capture were released without handling, or we encourage 9 �the animals to leave the tree by heaving snowballs toward them. If the pumas climbed a safe tree, then we handled them as described above. A cage trap was used to capture adults, subadults, and large cubs when pumas were lured into the trap using road-killed or puma-killed ungulates (Sweanor et al. 2008). A cage trap was set only if a target puma scavenged on the lure (i.e., an unmarked puma, or a puma requiring a collar change). Researchers continuously monitored the set cage trap from about 1 km distance by using VHF beacons on the cage and door. Researchers handled captured pumas within 30 minutes of capture. Puma were immobilized with Telazol injected into the caudal thigh muscles with a pole syringe. Immobilized pumas were restrained and monitored as described previously. If non-target animals were caught in the cage trap, we opened the door and allowed the animal to leave the trap. Small cubs (≤10 weeks old) were captured using our hands (covered with clean leather gloves) or with a capture pole. Cubs were restrained inside new burlap bags during the handling process and were not administered immobilizing drugs. Cubs at nurseries were approached when mothers were away from nurseries (as determined by radio-telemetry). Cubs captured at nurseries were removed from the nursery a distance of 30 to 100 m to minimize disturbance and human scent at nurseries. Immediately after handling processes were completed, cubs were returned to the exact nurseries where they were found (Logan and Sweanor 2001). Marking, Global Positioning System- and Radio-telemetry: Pumas do not possess easily identifiable natural marking, such as tigers (see Karanth and Nichols 2002), therefore, the capture, marking, and GPS- or VHF- collaring of individual pumas was essential to a number of project objectives, including estimating numbers, vital rates, and gathering movement data relevant to population dynamics (i.e., emigration and movement across Data Analysis Unit boundaries). Adults, subadults, and cubs were marked 3 ways: GPS/VHF- or VHF-collar, ear-tag, and tattoo. The identification number tattooed in the pinna was permanent and could not be lost unless the pinna was severed. A colored (bright yellow or orange), numbered rectangular (5 cm x 1.5 cm) ear-tag (Allflex USA, Inc., DFW Airport, TX) was inserted into each pinna to facilitate individual identification during direct recaptures. Cubs 10 weeks old were ear-tagged in only one pinna. Adult and subadult female pumas were fitted with GPS collars (approximately 400 g each, Lotek Wireless, Canada) if available. Initially, GPS-collars were programmed to fix and store puma locations at 4 times per day to sample daytime, nighttime, and crepuscular locations (i.e., 0:00, 06:00, 12:00, 19:00). GPS locations for pumas provided precise, quantitative data on movements to assess the relevance of puma DAU boundaries, our search efforts, and to evaluate puma behavior and social structure. The GPScollars also provided basic information on puma movements and locations to design other pilot studies in this program on vulnerability of puma to sport-harvest, habitat use, and enumeration methods (e.g., photographic or DNA mark-recapture). Subadult male pumas were fitted initially with conventional VHF collars (Lotek, LMRT-3, ~400 g each) with expansion joints fastened to the collars, which allowed the collar to expand to the average adult male neck circumference (~46 cm). If subadult male pumas reached adulthood on the study area, we would recapture them and fit them with GPS collars. In addition, other adult and female subadult pumas were fitted with VHF collars when GPS collars were not available. VHF radio transmitters on GPS collars enabled researchers to find those pumas on the ground in real time to acquire remote GPS data reports, facilitate recaptures for re-collaring, and to determine their reproductive and survival status. VHF transmitters on GPS- and VHF-collars had a mortality mode set to alert researchers when pumas were immobile for 3 to 24 hours so that dead pumas could be found to quantify survival rates and agent-specific mortality rates by gender and age. Locations of GPS- and VHF10 �collared pumas were identified about once per week (as flight schedules and weather allowed) from light fixed-wing aircraft (e.g., Cessna 185) fitted with radio signal receiving equipment (Logan and Sweanor 2001). GPS- and VHF-collared pumas were located from the ground opportunistically using a hand-held yagi antenna. At least 3 bearings on peak aural signals were mapped to fix locations and estimate location error around those locations (Logan and Sweanor 2001). Aerial and ground locations were plotted on 7.5 minute USGS maps (NAD 27) and UTMs along with location attributes were recorded on standard forms. GPS and aerial locations were mapped using GIS software. We attempted to collar all cubs in observed litters. Cubs were fit with small VHF transmitters mounted on expandable collars that expand to adult neck size (Wildlife Materials, Murphysboro, Illinois, HLPM-2160, 47g, Telonics, Inc., Mesa, Arizona MOD 080, 62g, or Telonics MOD 205, 90g,) when cubs weighed 2.311 kg (525 lb). Cubs could wear these small expandable collars until they were over 12 months old. Cubs were recaptured to replace collars as opportunities allowed. Monitoring radio-collared cubs allowed quantification of survival rates and agent-specific mortality rates (Logan and Sweanor 2001). Analytical Methods Population characteristics each year were tabulated with the number of individuals in each sex and age category. Age categories, as mentioned, include: adult (puma ≥24 months old, or younger breeders), subadults (young puma independent of mothers, <24 months old that do not breed), cubs (young dependent on mothers, also called kittens) (Logan and Sweanor 2001). When data allowed, age categories were further partitioned into months or years. Reproductive Rates: Reproductive rates were estimated for GPS- and VHF-collared female pumas directly (Logan and Sweanor 2001). Genetic paternity analysis will be used to ascertain paternity for adult male pumas (Murphy et al. 1998). Survival and Agent-specific Mortality Rates: Radio-collared pumas provided known fate data used to estimate survival rates for each age stage using the Kaplan-Meier procedure to staggered entry (Pollock et al. 1989). A binomial survival model was also used for crude estimates of survival during the subadult age stage (Williams et al. 2001:343-344). In addition, when data collection is complete, survival rates will be modeled in program MARK (White and Burnham 1999, Cooch and White 2004) where effects of individual (e.g., sex, age stage, reproductive stage) and temporal (i.e., reference period, treatment period) covariates to survival can be examined. Agent-specific mortality rates can also be analyzed using proportions and Trent and Rongstad procedures (Micromort software, Heisey and Fuller 1985). Population Inventory: The population of interest was independent pumas (i.e., adults and subadults) mainly during November to March which corresponds with the Colorado puma hunting season. Independent pumas were those that could be legally killed by recreational hunters. Initially, we estimated the minimum number of independent pumas and puma density (i.e., number of independent puma/100 km2) each winter. The minimum number of independent pumas included all marked pumas known to be present on the study area during the period, plus individuals thought to be non-marked and detected by visual observation or tracks that were separated from locations of radio-collared pumas. This minimum count is achieved by very intensive field operations and should be considered close to a complete enumeration of independent pumas counted during our annual November to April search period. Furthermore, adults comprised the breeding segment of the population and subadults were non-breeders that are potential recruits into the adult population in ≤1 year. The sampling unit was the individual independent puma (~≥1 yr. old). 11 �Puma Population Dynamics: A deterministic, discrete time model parameterized with population characteristics and vital rates from this research was used to assess puma population dynamics (Logan 2008). Functional Relationships: Once data collection is complete, a variety of analyses will be conducted to estimate parameters and examine functional relationships. Graphical methods will be used to initially examine functional relationships among puma population parameters. Linear regression procedures and coefficients of determination will be used to assess functional relationships if data for the response variable are normally distributed and the variance is the same at each level. If the relationship is not linear, data is non-normal, and variances are unequal, we will consider appropriate transformations of the data for regression procedures (Ott 1993). Non-parametric correlation methods, such as Spearman’s rank correlation coefficient, will also be used where appropriate to test for monotonic relationships between puma abundance and other parameters of interest (Conover 1999). Relationships of explanatory variables to survival parameters will be modeled in MARK. Statistical analyses can be performed in a variety of software (e.g., SYSTAT, R, and MARK). RESULTS AND DISCUSSION Segment Objective 1 Puma harvest: This biological year, August 2013 to July 2014, was the fifth year of the treatment period (TY5) in this study of puma population dynamics on the Uncompahgre Plateau. The hunting season on the study area began on November 18, 2013 and was scheduled to extend to January 31, 2014, unless the harvest quota was taken before then. The harvest design quota was 5 pumas, the same as in TY4 in an effort see if the lower harvest rate of about 11% (reduced from a desired 15%) would result in a stable or increased number of independent pumas (Logan 2013). We reduced the harvest rate because the population of independent pumas was declining during TY1 to TY3, contrary to the expected population trend in this research project and the realization that the population would probably continue to decline with a 15% harvest rate. This harvest design still tests the CPW’s current assumption that total mortality (i.e., harvest plus other natural deaths) in the range of 8 to 15% of the harvest-age population (i.e., independent pumas comprised of adults plus subadults) with the total mortality comprised of 35 to 45% females (i.e., adults and subadults) is acceptable to manage for a stable-to-increasing puma population (Assumption and Hypothesis 3 p.5-6 this report). The 11% harvest is in the middle of the 8 to 15% harvest range we are testing. The initial quota of 8 pumas for TY1, TY2, and TY3 was based on the projected minimum number of 53 independent pumas expected on the study area in winter 2009-10, modeled from a minimum count of pumas during winter 2007-08 (Table 1; Logan 2010). The quota of 8 pumas for TY3 was based on the observed minimum count of 52 independent pumas during November 2010 to April 2011 in TY2 and that approximately the same number of independent pumas was expected during the puma hunting season for TY3. The hunting structure in TY5 was the same as in TY1 to TY4, except for the reduction in quota (see above). The number of puma hunters on the study area was not limited. Each hunter on the study area was required to obtain a hunting permit from the CPW Montrose Service Center. Permits were free and unlimited. Each permit allowed the individual hunter with a legal puma hunting license in Colorado to hunt in the puma study area for up to 14 days from the issue date. Unsuccessful hunters that wanted to continue hunting past the permit expiration date requested a new permit for another 14 days, or until the hunter killed a puma within the season, or the season on the study area closed due to the quota being reached, or the end of the hunting season. This permit system allowed the CPW to monitor the number of hunters on the study area and to contact each hunter for survey information (see later in this section). 12 �All pumas harvested on the study area were examined by principal investigator K. Logan and sealed as mandated by Colorado statute. All successful hunters reported their puma kill and presented the puma carcass for inspection by CPW within 48 hours of harvest. Upon inspection, the following data were recorded: sex, age, and location of harvest. In addition, an upper premolar tooth was collected for aging (i.e., mandatory) and a tissue sample was collected for DNA genotyping. Each successful hunter was also asked at that time to complete a one-page hunter survey form. All other hunters that did not report a puma kill on the study area were asked to complete the survey form and return it in a stamped envelope that was provided. The puma hunting season occurred on the study area from November 18, 2013 to January 10, 2014, taking 54 days to fill the quota of 5 pumas; the longest hunting season yet on the study area during this treatment period. This was 13 more days than it took to harvest 5 pumas in TY4 (i.e., 41 days, Nov. 19 to Dec. 29, 2012), 21 more days than it took to harvest 8 pumas in TY3 (i.e., 33 days, Nov. 21 to Dec. 23, 2011), 33 days more than it took to harvest 8 pumas in TY2 (i.e., 21 days, Nov. 22 to Dec. 12, 2010) and 28 more days than it took to harvest 8 pumas in TY1 (i.e., 26 days, Nov. 16 to Dec. 11, 2009). Five pumas were killed on the study area, including: 1 adult female, 3 adult males, and 1 subadult male (Table 2). Of the 5 harvested pumas, 3 were marked: F111, M92 and M180. In addition to the pumas killed on the study area during the Colorado puma hunting season, 1 other marked adult female puma died during November 3 to 5, 2013 (Table 3). Adult female F140 died from an unknown natural cause. This puma was included in the minimum count of pumas for TY5 because she was initially captured on the study area and was present in the population during the TY5 survey period (Nov.–Apr.). The harvest of 5 independent pumas on the study area was 11.4% (5/44*100) of the minimum count of 44 independent pumas counted on the study area, including 27 females and 17 males, determined by intensive searches of the research team during November 2013 to April 2014 (Table 4). Independent females and males comprised 20.0% (1/5*100) and 80.0% (4/5*100) of the harvest, respectively. This harvest structure was 3.7% (1/27*100) of the independent females and 23.5% (4/17*100) of the independent males. Considering the mortality of 1 other radio-collared independent puma (F140; Table 3), the mortality of 6 independent pumas was 13.6% (6/44*100) of the minimum number of 44 independent pumas. The mortality composition of 2 females and 4 males was comprised of 33.3% (2/6*100) females and 66.7% (4/6*100) males. This mortality structure was 7.4% (2/27*100) of the independent females and 23.5% (4/17*100) of the independent males in the minimum count. The minimum count of 44 independent pumas in TY5 was slightly higher than the 42 minimum count in TY4, but was lower than the minimum count of 48 independent pumas in TY3, 52 independent pumas in TY2, and 55 in TY1 (Table 4). The slightly higher number of independent pumas in TY5 compared to TY4 suggests that the puma population responded in a positive way to the lower harvest rate (11%) applied in TY4 (Fig.3.) and to immigration of subadults. Nevertheless, the population decline caused by the higher harvest rate during TY1 to TY3 was still evident in a low population phase, and was particularly represented in losses to the adult female and adult male segments (Table 4, Segment Objectives 4 & 5 below). Hunter permits and survey: In TY5 mandatory permits with the voluntary survey attached were requested by 65 individual puma hunters. This number is lower than 70 hunters in TY4, and continued a downward trend from 74 hunters in TY3. It was close to 64 hunters in TY2, yet down substantially from 79 hunters in TY1. Thirteen of the hunters requested a second permit (after 14 days of the first permit had expired). Eleven hunters requested three permits, and one hunter requested four permits. Forty-four individual hunters (67.7%; 44/65*100) provided responses to the voluntary survey by turning in the 13 �printed survey. Of the respondents, 19 hunters indicated that they did not hunt on the study area. Twentyfive hunters indicated they did hunt at least one day. The proportion of the 44 respondents that hunted (25/44 = 0.568) extrapolated to the total of 65 hunters (0.568*65) indicated that about 37 hunters took to the field for pumas on the study area during the 54-day TY5 hunting season. This was down from 40 hunters in TY4, 49 hunters in TY3, 42 hunters in TY2 and 67 hunters in TY1 (Logan 2010, 2011, Logan 2012, Logan 2013). Considering that 37 hunters were estimated to be afield, then 13.5% of the hunters harvested pumas (5/37*100) and 27.0% of hunters captured pumas (10/37*100; see captured and released pumas below and in Table 5). The 44 puma hunters that turned in the volunteer survey were asked to answer, “Do you consider yourself a selective or non-selective hunter?” A selective hunter is one that purposely is hunting for a specific type of legal puma, such as a male, large male or large female. A non-selective hunter is one that intends to take whatever legal puma is first encountered or caught, with no desire for sex or size. Selective hunter was indicated by 27 respondents that answered the question (93.1%; 27/29 = 0.931). Of the remaining hunters, 2 indicated they were non-selective (6.9%). Fifteen hunters that returned surveys did not answer the question. The voluntary hunter survey also revealed that four hunters treed other pumas on the study area, but chose not to kill them (Table 5). The hunters’ reasons for not wanting to kill the pumas included they were not legal game (2 cubs), the size was too small to harvest (1 subadult males, 1 adult male). No reason was given for not harvesting one other subadult male. In an effort to better ascertain the vulnerability of sexes and age-stages (i.e., adult, subadult) of independent pumas to detection by puma hunters and hunter selection to address assumption 6 and hypothesis 6 (previously), the survey was changed in TY2 to ask hunters, “What was the sex of the lion that made the first set of tracks you encountered that were less than one day old?”. This question pertained to pumas that could be pursued by dogs and captured with a relatively high probability to allow the hunter an opportunity to harvest the puma. Associated with the question, we asked, “Did you pursue the lion to harvest it?” Hunters’ responses in TY5 showed they encountered 17 puma tracks less than one day old. Of those, 11 tracks were of females, and 6 tracks were of males, indicating that during the TY5 hunting season females were more detectable than males. These data from TY5 were consistent with data from two previous treatment years TY2 and TY3 (these data were not gathered in the survey for TY1) where tracks less than 1 day old reported by puma hunters consistently favored females (TY2: 20 female, 10 male; TY3: 15 female, 6 male). But, they were at variance with TY4 (8 female, 11 male). Of the 11 female tracks less than one day old, 10 hunters that encountered them said they had no intent to harvest the puma and one hunter did not indicate his intent. However, one of those hunters killed adult female F111 because he could not clearly sex the puma and thought it was a male. Of the 6 male tracks less than one day old, 3 of the hunters that encountered them indicated intent to harvest the pumas and in fact they did harvest 2 of them. One hunter that encountered both male and female tracks together did not intend to harvest the pumas he pursued. Two other hunters that encountered male tracks did not indicate their intent to harvest them. These preliminary survey and harvest data for TY5 indicate that hunters detected independent female pumas more frequently than males and males were captured by hunters more frequently than females by 7 to 1 (i.e., males = 4 harvested + 3 captured and released; females = 1 harvested). Moreover, hunters chose to kill males instead of females. Results in TY5 indicated selection for male pumas by hunters were consistent with TY1, TY2, TY3 and TY4 results. In addition, hunters in TY1, TY2, and TY3 treatment years caught females slightly more frequently than males, and males were selected for harvest. But, in TY4 hunters said they detected independent male pumas more frequently than females, and males were captured by hunters more frequently than females (Logan 2013). This preliminary assessment from the treatment period (i.e., TY1-TY5) puma harvest and hunter survey data suggests that generally female pumas were detected by hunters more frequently than male pumas, except for TY4, the 14 �large majority of puma hunters were selective, and hunter choices influenced harvest sex and age composition. Harvest composition was not simply determined by hunter detections of pumas as determined by puma movement patterns (Hypothesis 6, previously). These results do not support assumptions in Hypothesis 6. Independent male pumas on our study area were more vulnerable to harvest mainly due to hunter selection (not to greater detection), while independent female pumas were more detectable. In addition, the puma population declined even though independent male pumas dominated the harvest all 5 treatment years, with percentages in harvest ranging from 60% in TY4 to 75% in TY2 during the decline years, to 80% in TY5 when the population was still in low phase. Moreover, adult males dominated the harvest in TY1 (62.5%) when the population was highest, in TY2 (62.5%) as the population declined, and in TY5 (60%) when the puma population was still in low phase, contrary to predictions. Subadult males comprised a large percentage of the harvest (50%) only in TY3 when the population was well into a decline phase. Segment Objective 2 After the harvest quota was filled, puma research teams immediately initiated capture operations with trained dogs. Two fully-staffed capture teams, one each detailed on the east and west slopes of the study area, systematically and thoroughly searched the study area to capture, sample, and GPS/VHF radio-collar pumas the remainder of winter and early spring 2013-14. These efforts along with cage trap efforts and hand-capturing cubs at nurseries maintained samples to quantify population sex and age structure, survival, and agent-specific mortality, and allowed determination of minimum population size on the study area during November to April. We made 43 captures of 38 individual pumas from August 2013 to July 2014 (Tables 6-11); 24 individual pumas were captured with dogs 29 times, including one that was originally captured in a cage trap earlier. Three pumas were captured in cage traps. Twelve cubs were captured at nurseries by hand. A total of 56 individual pumas were monitored with radio-telemetry from August 2013 to July 2014 (some of these had been collared in previous years), representing sex and age classes including: 23 adult females, 7 adult males, 4 subadult females, 5 subadult males, and 21 cubs (13 female, 8 male). Three subadult females and 2 subadult males that were monitored survived to adult age during the biological year. Trained dogs were used as our main method to capture, sample, and mark pumas from January 9, 2014 to April 24, 2014. Those efforts resulted in 75 search days, 361 total puma tracks detected of which 109 were ≤1 day old, 82 pursuits, and a total of 29 puma captures of 24 individual pumas (Table 6). Search days with dogs in TY5 (75) was almost equal to TY4 (74), slightly lower than TY3 (79 days) and lower than TY1 (86 days) and TY2 (81 days) (Table 12). The frequency of tracks (tracks/day) encountered in TY5 was the highest of all the treatment years (Table 12). The capture rate (capture/day) in TY5 fell at the mid-range of the capture rates for TY1 to TY5 (Table 12). The number of new pumas captured for the first time in TY5 was the second lowest, exceeding only TY1 (Table 12). Researchers in the two hound capture teams also recorded instances when the first tracks ≤1 day old of independent pumas were encountered on each search route each day to represent encounters with puma tracks that could be detected and pursued by puma hunters. The count was: 37 tracks of females, including 10 associated with cubs; 11 tracks of males; 1 track of a lone cub; and 3 track of unspecified sex. These tracks ≤ 1 day old were found by the researchers after the TY5 puma hunting season when one independent female and four independent males were harvested and one adult female died natural causes earlier in November (Tables 2 and 3). Therefore, those dead pumas were not present to make tracks for our researchers to observe. By comparison, the number of first tracks <1 day reported by puma hunters in TY5 was 11 females and 6 males (Segment Objective 1 above). 15 �Puma capture efforts using ungulate carcasses and cage traps occurred from October 8, 2013 to July 21, 2014 with the main efforts in the fall and spring (Table 10). We used 50 road-killed mule deer and one road-killed elk at 24 different sites. One subadult female (F210) and 1 adult male (M211) were captured for the first time. One adult male (M196) was recaptured and re-collared. Pumas scavenged at 11 of 51 (21.6%) of the ungulate carcasses used for bait. We sampled 12 new cubs, including 5 females and 7 males (Table 11). All except 1 were radiocollared to monitor survival and agent-specific mortality (Appendix A). One other non-marked female cub (PF1094; offspring of F176) was mauled to death by our dogs on 2/4/2014 when the cub was about 7 months old. Besides our direct puma captures with dogs January through April, we detected 13 radio-collared pumas that we were able to identify with GPS or VHF telemetry 18 times, thus, negating the need to capture those pumas directly with dogs (Table 6). Upon detecting puma tracks that were aged at ≤1 day old, we followed the tracks with a radio receiver in an effort to detect if the tracks might be of a puma wearing a functional collar. We assigned tracks to a collared individual if we received radio signals from a puma that we judged to be <1 km from the tracks and in direction of travel of the tracks. This approach allowed us to more efficiently allocate our capture efforts toward pumas of unknown identity on the study area, particularly unmarked pumas or pumas with non-functioning GPS- or VHF- radiocollars. In addition to the harvest and capture data (previously), our search efforts revealed the presence of at least 26 to 28 non-marked pumas which we included in our minimum count November 2013 through April 2014 (Table 4). We classified those pumas as: 6 adult females, 4 adult males, 3 subadult males, and 13 to 15 cubs. Of these pumas, one adult female and one adult male were treed by our hounds, but we could not handle the pumas because they climbed dangerous trees (Table 8). In addition, one cub was mauled to death by our dogs (PF1094, previously). The two adults were bio-darted for genotyping and we collected tissue from the dead cub. We could separate the activity of these other pumas from the GPSand VHF- collared pumas in time, space, track size differences between females and males, and by the numbers of cubs following females. Also, one non-marked adult male was photographed by a digital trail camera with two notches in his right pinna that distinguished him from M211 caught for the first time. An adult female was photographed by a trail camera while associating with M196. A subadult male was photographed twice while scavenging on mule deer bait. Our intensive search and capture efforts during January through April 2014, information from pumas previously marked and GPS/VHF-monitored pumas, and information from the puma hunting season in TY5 enabled us to quantify a minimum count of 44 independent pumas detected on the Uncompahgre Plateau study area, including 27 independent females and 17 independent males (Table 4). This count was based on the number of known radio-collared pumas, non-marked pumas killed by hunters on the study area, observations of marked and non-marked pumas observed by researchers or pursued, treed and released by hunters on the study area, puma tracks observed by researchers that could not be attributed to pumas with functioning radiocollars, and distinctive pumas photographed by trail cameras. Of the 44 independent pumas, 31 (70.5%) were marked and 13 (29.5%) were non-marked animals (i.e., some may have ear-tags and tattoos). The abundance of independent pumas in TY5 on the east slope (25; 13 females, 12 males) was greater than the west slope (19; 14 females, 5 males) of the study area. The largest difference was the higher number of subadult males detected on the east slope (Table 4). Otherwise, the sex structure of adult pumas on the east and west slopes was similar, with substantially more females than males. The slight increase in independent pumas in TY5 compared to TY4 was attributed to the addition of one adult male and an increase in subadult males. Considering the minimum count of 44 independent pumas in TY5, a preliminary minimum density for the winter puma habitat area estimated at 1,671 km2 on the 16 �Uncompahgre Plateau study area was 2.63 independent pumas/100 km2. The adult puma density was 1.97 adult pumas/100 km2. These preliminary densities contrast with the highest densities for the study area in TY1, in which there were 3.29 independent pumas/100 km2 and 2.99 adult pumas/100 km2. The TY5 minimum count of 44 independent pumas was a slight increase over 42 minimum independent pumas counted in TY4 (Table 4). This signaled stabilization if not slight increase in the puma population compared to the steady decline phase that occurred during the 4 previous treatment years TY1 to TY4; however, the population was still in low phase (Fig. 3). A stabilization or slight increase can be attributed to the lower harvest rate in TY4 (Fig. 3) in which the harvest rate was reduced from a designed 15% to 11% of independent pumas (i.e., quota reduced from 8 to 5 pumas) and to immigration of subadults (Table 4). Clearly, a 15% harvest rate of independent pumas on the Uncompahgre Plateau study area, in addition to other sources of mortality, reduced the abundance of independent and adult pumas (Fig. 3, Table 4). Between TY1 and TY5 minimum counts indicated that the adult female segment of the population had declined from 30 to 23 females, a 23% decline. During this decline the percent of adult females in the study area sport harvest ranged from zero to 40%, with zero adult females killed in TY2, and 25% in TY1, 37.5% in TY3, 40% in TY4, and 20% in TY5. The percent of independent females (adults plus subadults) in the harvest ranged from 20% (TY5) to 37.5% (TY1 and TY3). The puma population declined with 25% to 37.5% independent females in the harvest from TY1 to TY3. The adult male segment had declined from 20 to 10 adults, a 50% decline which is indicative of the strong hunter selection for male pumas (previously in Segment Objective 1). The number of independent pumas (adults and subadults) declined from 55 to 44 pumas, a 20% decline. The largest decline in the number of independent pumas occurred between TY1 and TY4 after three years of an intended 15% harvest rate of independent pumas and before the harvest rate was reduced to 11%. The independent pumas declined from 55 to 42, a 24% decline. These results from this population test do not support Hypothesis 3 (previously). The estimated age structure of independent pumas in November 2013 at the beginning of the puma hunting season in TY5 on the Uncompahgre Plateau study area is depicted in Figure 4. The male age structure is young and declined primarily due to harvest during TY1, TY2, TY3, and TY4 (Logan 2010, 2011, 2012, 2013) and immigration of subadults with the oldest males about 4.75 years old. The female age structure was also distributed mainly to the younger ages but with several distributed among the older ages, including two aged 10+ years. This age structure is a characteristic of the decline in the puma population from TY1 to TY5 (Fig. 3) and attendant with reduced adult annual survival rates, especially of males, when compared with the reference period (Table 15). Segment Objective 3 During the past 9.7 years of this work we compiled data on puma reproduction that was not previously available on pumas in Colorado (Table 13). In TY5 we directly observed four litters in nurseries and inferred the birth of one other litter from the GPS data of the mother (F129). In the latter case lack of private land access impeded our ability to observe the litter. Of those 5 litters, four were born in August 2013 and one was born in June 2014 (Table 11, Table 13). In addition, one litter identified by one observed male cub that was captured and marked (M225) was born to an unmarked female in September 2013. Data on reproduction we observed in TY1, TY2, TY3, TY4 and TY5 were added to Table 13 which gives the reproductive chronology and information on mates (if known) of reproducing females. The proportion of radio-collared adult females giving birth from August 2013 to July 2014 biological year (TY5) was 0.33 (5/15). For the previous 4 treatment years the proportion was TY1=0.53 (8/15), TY2=0.53 (9/17), TY3=0.29 (5/17), TY4=0.67 (10/15). 17 �Considering our 57 total litters from 31 females, including 53 observed with cubs 25 to 45 days old and 4 other litters confirmed by nurseries and ≥1 nursling cub tracks or dead cub remains with GPScollared females (Table 13), the distribution of puma births by month from 2005 to 2014 indicate births extending from March into September (Fig. 5). Births began in spring and were high in May through August with a peak in July. Our data from the past 10 years indicated no observed live births from October through February. Births during late spring to late summer (May to August) involved 86% (49/57*100) of the births (Fig. 5). The data indicated that the large majority of puma breeding activity occurred February through May (i.e., gestation averages about 90-92 days, Logan 2009). In comparison, Anderson et al. (1992:47-48) found on the Uncompahgre Plateau during 19821987 that of 10 puma birth dates 7 were during July, August, and September, 2 in October, and 1 in December, with most breeding occurring April through June. The 2 data sets indicated puma births on the Uncompahgre Plateau have occurred in every month except January and November (so far). As we gather more data on the puma births during the treatment period, we will examine the distributions of births in the reference and treatment periods separately for a treatment effect on timing of breeding and births. Segment Objectives 4 & 5 From December 8, 2004 (capture and collaring of the first adult puma M1) to July 31, 2014, we monitored 31 adult male and 47 adult female pumas to quantify survival and agent-specific mortality rates (Table 14). Preliminary estimates of adult puma survival rates in the absence of sport-hunting during the reference period indicated high survival, with adult male survival generally higher than adult female survival (Table 15). Annual survival rates for adult pumas generally declined during the treatment period (Table 15), with the exception of adult females in TY2. Otherwise, adult puma survival rates declined substantially in the treatment period and the greater mortality was associated with the population to decline (Table 4, Fig. 3). The major cause of mortality was sport-hunting (Table 14). In TY5 we monitored 22 adult females and 7 adult males for annual survival and agent-specific mortality in TY5. Annual survival rate for adult females was 0.678 (SE=0.0934) and for males was 0.667(SE=0.1721). Sport-hunting was the only cause of death for adult males. However, one adult female was killed by a hunter, two died of other human causes, and one died of natural cause. Adult F182 was struck and killed by a vehicle on highway 550 August 25, 2013. She was about 48 months old at death and was pregnant with two fetuses that probably would have been born in September. This mortality made the sixteenth puma death recorded due to vehicle collision on the study area since 2004 (Table 18). Seven of the 16 pumas were marked, including 4 adults with GPS/VHF collars. Adult F136 was shot on September 30, 2013 by a U.S.D.A., A.P.H.I.S.,Wildlife Services agent because she killed a domestic goat. Her three 3.4 month-old cubs (F189, F200, M201) starved to death about 20 to 23 days later. F136 was about 65 months old at death. We recorded two natural deaths of adult female pumas, exact causes unknown. Adult F140 died on November 4, 2013. F140 was found laying in a space under a boulder over-hang, completely intact, no visible injuries, but dehydrated and emaciated. Enlarged nipples, red-stained hair around the nipples, and uterine scars indicated that F140 probably had recently given birth to cubs. GPS location data on F140 indicated the probable birth date of the cubs was June 25, 2013. Fate of the cubs was unknown. F140 was 39 months old at death. Adult F171 died on June 1, 2014 according to GPS data. Her death orphaned 3 cubs (M206, F207, F208) when they were 10 months old. We have information on 45 subadult pumas (i.e., independent pumas <24 months old), including 19 females and 26 males (Table 16). We lost radio contact with 4 male and 2 female that probably dispersed from the study area unknown distances. The fate subadult M198 is currently unknown. His 18 �radiocollar was detected on mortality mode on July 30, 2014; but, we could not investigate the fate of M198 because the collar was in dangerous cliffs. Of the remaining 38 subadults (females and males combined), 8 (3 females, 5 males) died before reaching adult stage, indicating a rough preliminary binomial survival rate of 0.79 (i.e., 30/38) for subadults surviving to the adult age stage (i.e., 24 mo. old). Of the eight subadults that died, four deaths were from natural causes (1 trampled by elk, 2 killed by other pumas, 1 broken leg), three were from sport-hunting, and one was from a vehicle strike (Table 16). Harvest data along with our capture and radiotelemetry data provided dispersal and fate information on 41 marked pumas, 30 males and 11 females. Of those, 30 (5 females, 25 males) were initially captured and marked as cubs, and 11 (5 females, 6 males) were captured and marked in the subadult life-stage on the Uncompahgre Plateau puma study area (Table 17). Twenty-three males were killed by hunters away from the study area at linear distances (i.e., from initial capture sites to kill sites) ranging from about 20 to 370 km. Two males with extreme moves were killed in the Snowy Range of southeastern Wyoming (369.6 km) and the Cimarron Range of north-central New Mexico (329.8 km). Two males were killed by a hunter on the study area 12.9 and 21.9 km from their original capture sites. Four females were killed by puma hunters off the study area ranging from 20.7 to 74.5 km from initial capture sites. One female was killed by a hunter on the study area 18.2 km from her initial capture site. Female F52 was treed and released by hunters in December 2008 and 2009 south of Powderhorn, Colorado, indicating that she established an adult home range there before she was killed by a puma hunter in that area on Jan. 9, 2012. Three males (M67, M87, M92) that were marked initially as cubs born on the east slope of the study area, dispersed from their natal ranges and were recaptured as adults on the west slope of the study area. All three of the males were killed by hunters on their adult territories. A preliminary estimate of cub survival during the reference period was summarized in Logan 2009 using 36 radio-collared cubs (16 males, 20 females) marked at nurseries when they were 26 to 42 days old. In that summary, estimated survival of cubs to one year of age was 0.53. [The estimated minimum survival rate using the Kaplan-Meier procedure was 0.5285 (SE = 0.1623). The maximum estimated cub survival was practically the same, 0.5328 (SE = 0.1629).] The major natural cause of death in cubs, where cause could be determined, was infanticide and cannibalism by other, especially male, pumas. In TY5 we monitored the fates of 21 radio-collared cubs (Table 11, Appendix A). We lost contact with 7 (F184, F187, F188, F203, F204, F208, M221) after radiocollars apparently quit operating and one collar was shed prematurely. However, one of those, F184, was later recaptured as a subadult. Of the remaining 14 collared cubs, 6 died. Three siblings (F189, F200, M201) died of starvation at 93 to 96 days old after their mother F136 was killed for depredation control purposes (i.e., she killed a goat). One cub (F202) died from predation or infanticide at about 59 days old. One cub (M205) died of an unknown natural cause when about 190 days old. One cub, F207, died an unknown natural cause when about 330 days old and about 24 days after her mother F171 died of an unknown natural cause. F207 had numerous porcupine quills stuck in her face, mouth, left shoulder, and both hind feet. F207 also had suffered a broken left humerus earlier in life, which had mended. Of the 8 remaining live radio-collared cubs 3 survived to the subadult stage, including one, M206, that was orphaned at 10 months old (sibling of F207). Five dependent cubs were being monitored as of July 31, 2014. Two non-marked cubs also died. Nursling P1076 was found dead and mostly consumed due to infanticide or predation in the nursery of F181. PF1094 was killed by our hunting dogs during a capture effort. Forty-eight adult pumas (35 females, 13 males) have worn GPS collars since this project began in 2004 (Table 19). Over 70 thousand GPS locations have been obtained and will be used for studies on puma behavior, social organization, population dynamics, population genetics, movements, population 19 �survey methods, habitat use and puma-human relations in collaboration with colleagues in Mammals Research, Colorado State University, and Arizona State University. Segment Objective 6 We continued to explore non-invasive methods for sampling pumas to estimate abundance. We began a pilot project in June 2014 to evaluate modified foot-hold devices and break-away neck snares as passive devices to snag hair from passing pumas. Hair obtained from the pumas will be tested to determine the quality of the DNA for genotyping individuals for mark-recapture methods for estimating abundance. This effort will continue into early November 2014. SUMMARY Manipulative, long-term research on puma population dynamics, effects of sport-hunting, and development and testing of puma enumeration methods began in December 2004. After 9.7 years of effort 224 unique pumas have been captured, sampled, marked, and released. Using these animals, we monitored fates of pumas in all sexes and age stages, including: 47 adult females, 31 adult males, 19 subadult females, 26 subadult males, 62 female cubs, 90 male cubs, and 2 cubs of undetermined sex (some individuals occur in more than one age stage). Data from marked animals were used to quantify puma population characteristics and vital rates in a reference period without sport-hunting off-take as a mortality factor from December 2004 to July 2009. Puma population characteristics and vital rates in a reference condition allowed us to develop a puma population model, and to use population data and modeling scenarios to conduct a preliminary assessment of CPW puma management assumptions and guide directions for the remainder of the puma research on the Uncompahgre Plateau. Moreover, our data and model provide tools for CPW wildlife biologists and managers to assess puma harvest strategies. The 5-year treatment period began August 2009 in which sport-hunting is a mortality factor. The treatment period will be a population-wide test of CPW puma management assumptions. Now almost 5 years of the treatment period are complete. The data support some CPW puma management assumptions, such as expected population structure and density ranges, and sport-hunting can cause population decline. Other assumptions are being challenged, such as 15% harvest rate with 35 to 45% females in the harvest can result in population stability or growth (15% harvest with 25-40% females caused population decline), male pumas are more detectable by hunters (females were more detectable), and puma harvest composition is not influenced by hunter selection (hunter selection strongly influenced harvest composition). Another underlying assumption of the GMU structure is that pumas assumed to occupy a GMU are only subject to harvest within that GMU. The data indicated that assumption to be unreliable. Pumas moved among adjacent GMUs and were at risk to harvest after the study area GMU was closed (i.e., the quota was filled) even though the study area GMU is among the largest in the state. The additional mortality that occurred contributed to population decline. Since the beginning of this study 2 previous efforts have been made to develop and test non-invasive methods for estimating puma abundance. These efforts were in collaboration with Colorado State University in a Ph.D. program (Jesse Lewis) and a M.S. program (Kirstie Yeager). A third pilot effort was begun in June 2014 and will go to November 2014. Monitoring of the radio-collared pumas will continue through December 2014 to gather data on puma reproduction, survival and causes of mortality. In January 2015, the principal investigator will start an intensive, thorough data analysis, modeling, and writing phase. Furthermore, we will continue collaboration with colleagues on investigations of puma population parameter estimation, population genetics, puma movements, puma habitat modeling and mapping, puma-human relations, and disease prevalence. All of these efforts should enhance the Colorado puma research and management programs. LITERATURE CITED Anderson, A. E, D. C. Bowden, and D. M. Kattner. 1992. The puma on Uncompahgre Plateau, Colorado. Technical Publication No. 40. Colorado Division of Wildlife, Denver. 20 �Anderson, C. R., Jr., and F. G. Lindzey. 2005. Experimental evaluation of population trend and harvest composition in a Wyoming cougar population. Wildlife Society Bulletin 33:179-188. Barnhurst, D. 1986. Vulnerability of cougars to hunting. Master’s Thesis. Utah State University. Colorado Division of Wildlife 2002-2007 Strategic Plan. 2002. Colorado Department of Natural Resources, Division of Wildlife. Denver. Colorado Division of Wildlife. 2007. Colorado mountain lion management data analysis unit revision and quota development process. Colorado Division of Wildlife, Denver. Conover, W. J. 1999. Practical nonparametric statistics. John Wiley & Sons, Inc., New York. Cooch, E., and G. White. 2004. Program MARK- a gentle introduction, 3rd edition. Colorado State University, Fort Collins. Culver, M., W. E. Johnson, J. Pecon-Slattery, and S. J. O’Brien. 2000. Genomic ancestry of the American puma (Puma concolor). The Journal of Heredity 91:186-197. Currier, M. J. P., and K. R. Russell. 1977. Mountain lion population and harvest near Canon City, Colorado, 1974-1977. Colorado Division of Wildlife Special Report No. 42. Heisey, D. M., and T. K. Fuller. 1985. Evaluation of survival and cause specific mortality rates using telemetry data. Journal of Wildlife Management 49:668-674. Karanth, K. U., and J. D. Nichols. 2002. Monitoring tigers and their prey: A manual for researchers, managers and conservationists in tropical Asia. Centre for Wildlife Studies, Bangalore, India. Koloski, J. H. 2002. Mountain lion ecology and management on the Southern Ute Indian Reservation. M. S. Thesis. Department of Zoology and Physiology, University of Wyoming, Laramie. Kreeger, T. J. 1996. Handbook of wildlife chemical immobilization. Wildlife Pharmaceuticals, Inc., Fort Collins, Colorado. Laundre, J. W., L. Hernandez, D. Streubel, K. Altendorf, and C. L. Lopez Gonzalez. 2000. Aging mountain lions using gum-line recession. Wildlife Society Bulletin 28:963-966. Logan, K. A., E. T. Thorne, L. L. Irwin, and R. Skinner. 1986. Immobilizing wild mountain lions (Felis concolor) with ketamine hydrochloride and xylazine hydrochloride. Journal of Wildlife Diseases. 22:97-103. _____, and L. L. Sweanor. 2001. Desert puma: evolutionary ecology and conservation of an enduring carnivore. Island Press, Washington, D.C. _____. 2008. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Wildlife, Fort Collins. _____. 2009. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Wildlife, Fort Collins. _____. 2010. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Wildlife, Fort Collins. _____. 2011. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Parks and Wildlife, Fort Collins. _____. 2012. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Parks and Wildlife, Fort Collins. _____. 2013. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Parks and Wildlife, Fort Collins. Murphy, K., M. Culver, M. Menotti-Raymond, V. David, M. G. Hornocker, and S. J. O’Brien. 1998. Cougar reproductive success in the Northern Yellowstone Ecosystem. Pages 78-112 in The ecology of the cougar (Puma concolor) in the Northern Yellowstone ecosystem: interactions with prey, bears, and humans. Dissertation, University of Idaho, Moscow. Ott, R. L. 1993. An introduction to statistical methods and data analysis. Fourth edition. Wadsworth Publishing Co., Belmont, California. Pierce, B. K., V. C. Bleich, and R. T. Bowyer. 2000. Social organization of mountain lions: does land a tenure system regulate population size? Ecology 81:1533-1543. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. 21 �Pojar, T. M., and D. C. Bowden. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550-560. Rausch, R. L., C. Maser, and E. P. Hoberg. 1983. Gastrointestinal helminths of the cougar, Felis concolor L. in northeastern Oregon. Journal of Wildlife Diseases 19: 14-19. Ross, P. I., and M. G. Jalkotzy. 1992. Characteristics of a hunted population of cougars in southwestern Alberta. Journal of Wildlife Management 56:417-426. Seidensticker, J. C., M. G. Hornocker, W. V. Wiles, and J. P. Messick. 1973. Mountain lion social organization in the Idaho Primitive Area. Wildlife Monographs No. 35. Stoner, D. C. 2004. Cougar exploitation levels in Utah: implications for demographic structure, metapopulation dynamics, and population recovery. Master of Science Thesis. Utah State University. Sweanor, L. L., K. A. Logan, J. W. Bauer, B. Milsap, and W. M. Boyce. 2008. Puma and human spatial and temporal use of a popular California state park. Journal of Wildlife Management 72:10761084. Williams, B. K., J. D. Nichols, and M. J. Conroy. 2001. Combining closed and open mark-recapture models: the robust design. Pages 523-554 In Analysis and management of animal populations. Academic Press, New York. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 (Suppl):S120-S139. Prepared by: ________________________ Kenneth A. Logan, Wildlife Researcher 22 �Table 1. Projected puma population growth modeled from a minimum count of independent pumas during winter 2007-08 reference period year 4 (RY4). Treatment period year 1 (TY1), shaded in gray, indicates the results used to derive a quota of 8 independent pumas, representing 15% of the independent pumas (from Logan 2009). Harvest Level No harvest. Year RY4 RY5 TY1 Projected Minimum Puma Population Size Adult Subadult Female Male Female Male 16 8 5 4 18 10 9 8 23 14 8 8 Independent Pumas Cub 20 33 42 Total 33 45 53 Lambda 1.37 1.17 Table 2. Pumas harvested by sport-hunters in Treatment Year 5 (TY5) on the Uncompahgre Plateau Study Area, Colorado, November 18, 2013 to January 10, 2014 (54 days). Puma sex Age (yr.) Date of kill Location/UTM NAD27 Zone, Easting, Northing Hunter/status 1.5 Previous M/F I.D. or specimen P no. if not marked PM1080 M 11/29/2013 13S, 239533, 4249632 M M 3.5 2.5 PM1089 M180 12/14/2013 12/16/2013 12S, 755418, 425910 12S, 737774, 4271802 F M 6.5 5.4 F111 M92 12/21/2013 1/10/2014 12S, 750802, 4261687 12S, 736233, 4238645 Brandon Owens/ Resident Shamus Howe/Resident John Watson/Nonresident Jacob Owens/Resident Gary Heubel/Nonresident Table 3. One other independent VHF-collared puma in the minimum count November 2013 to April 2014 (TY5) for the Uncompahgre Plateau Study Area that also died during November to April 2013-2014 period coinciding with the Colorado puma hunting season. Puma sex (M or F) F140 Age (yr.) 3.25 Date of kill/death 11/3-5/2013 Place of kill/UTM NAD27 Zone, Easting, Northing Study Area, San Miguel River Canyon 12S, 719342, 4236621 23 Hunter/status/other cause Unknown natural cause �Table 4. Minimum count of pumas based on numbers of known radio-collared pumas, visual observations of non-marked pumas, harvested non-marked pumas, and track counts of suspected non-marked pumas on the study area during September 2009 to April 2010 of Treatment Year 1 (TY1), November 2010 to April 2011 (TY2), November 2011 to April 2012 (TY3), November 2012 to April 2013 (TY4), and November 2013 to April 2014 (TY5), Uncompahgre Plateau study area, Colorado. Treatment Year (TY) Study Area region TY1 East slope West slope subtotals TY2 TY3 TY4 TY5 Adults Female Male Subadults Female Male Female Cubs Male 16 10 1 1 1 4 14 10 0 3 3 3 30 20 1 4 4 7 Total Independent Pumas = 55, including 31 females, 24 males. Cubs = 20-25 East slope 15 5 3 2 7 9 West slope 15 7 2 3 2 5 subtotals 30 12 5 5 9 14 Total Independent Pumas = 52, including 35 females, 17 males. Cubs = 39 East slope 13 4 1 3 4 2 West slope 14 5 3 5 1 2 subtotals 27 9 4 8 5 4 Total Independent Pumas = 48, including 31 females, 17 males. Cubs = 19 East slope 15 4 3 2 4 4 West slope 10 5 3 0 2 5 subtotals 25 9 6 2 6 9 Total Independent Pumas = 42, including 31 females, 11 males. Cubs = 24 East slope 10 6 3 6 6-7 2 West slope 13 4 1 1 1 3 subtotals 23 10 4 7 7-8 5 Total Independent Pumas = 44, including 27 females, 17 males. Cubs = 25-28 Unknown sex 4-8* 5-6 9-14 7 9 16 4 6 10 3 6 9 2 11-13 13-15 *One adult non-marked female puma was killed by a hunter in Roubideau Canyon. The female puma was lactating, indicating she had nurslings. Up to 4 cubs were assumed to be in the litter. Table 5. Pumas captured and released by sport-hunters in Treatment Year 5 (TY5) on the Uncompahgre Plateau Study Area, Colorado, November 18, 2013 to January 10, 2014. Data are from puma hunter responses in 44 original voluntary surveys on printed permits. Total response rate from 65 individual permitted hunters was 68% (44/65 = 0.68*100). Puma sex/age stage/mark M/subadult/ no tags 2 M/cubs/no tags M/subadult/no tags M/adult/no tags Date of capture 11/26/2013 11/28 or 12/13/2013 12/7,8,14,15/2013 12/7,8,14,15/2013 Capture location Sim’s Mesa San Miguel River Sim’s Mesa Sim’s Mesa Hunter name Chad Black John Martinez & Gerald Sickels, Jr. Jeremy Wheeler Jeremy Wheeler 24 Reason for releasing the puma given by hunter None given. Cubs. Too small to harvest. Too small to harvest. �Table 6. Summary of puma capture efforts with dogs from January 9, 2014 to April 24, 2014, Uncompahgre Plateau, Colorado. Month No. Search Days 16 No. & type of puma No. & type of No. & I.D. or type of pumas captured, tracks founda,b pumas pursued observed, or identified January 95 tracks: 14 male, 19 pursuits: 4 male, 7 pumas captured 7 times: F95, F118, F171, 51 female, 19 cub, 10 female, 5 cub M211, F212, M220, M221 (cub of F197). In 11 undetermined addition, adult females F137, F171, F176, F181, independent pumas F197, subadult male M220, and cubs F203, F204 Tracks ≤1 day old: and M206 were associated with tracks by VHF 4 male, 14 female, telemetry. 7 cub, 1 undetermined independent puma February 23 144 tracks: 27 male, 41 pursuits: 7 male, 12 pumas captured 16 times: F8, F28, F176, 71 female, 38 cub, 19 female, 13 cub, F184 (2 times), M213 (3 times), F214, F222, 8 undetermined 2 undetermined M223 (cub of F28), F224, PM1095 (2 times; bioindependent puma independent puma darted; not handled due to dangerous trees), Tracks ≤1 day old: PF1094 (cub of F176; killed by dogs). In 9 male, 26 female, addition, adult females F95, F118, and cub M223 16 cub, 2 (2 times) were associated with tracks by VHF undetermined telemetry, and adult M211 and subadult M220 independent puma were identified with tracks at trail cameras. March 19 78 tracks: 23 male, 16 pursuits: 2 male, 4 pumas captured 4 times: M215, M223 (cub of 29 female, 24 cub, 7 female, 7 cub F28), M225 (cub of unidentified female), M226. 2 undetermined In addition, adult female F176 and adult male independent puma M196 were associated with tracks by VHF Tracks ≤1 day old: telemetry. 3 male, 8 female, 8 cub April 17 44 tracks: 10 male, 6 pursuits: 3 pumas captured 3 times: F96, F186 and F188 24 female, 10 cub 3 female, 3 cub (cub of F96). In addition, adult F118 and adult Tracks ≤1 day old: M196 (2 times) were associated with tracks by 2 male, 4 female, VHF telemetry, and adult M190 was identified at 3 cub a trail camera. 75 361 tracks: 82 pursuits: 24 individual pumas were captured 29 times with TOTALS 74 male, 13 male, aid of dogs. In addition, 13 radio-collared pumas 175 female, 39 female, were detected 18 times by tracks and identified 91 cub, 28 cub with VHF telemetry ≤1 km from the tracks and 3 21 undetermined 2 undetermined marked pumas were identified at trail cameras. Tracks ≤1 day old: 10 independent pumas (adults, subadults) were 18 male captured with dogs for the first time (refer to 52 female Tables 7 and 8). 34 cub 5 undetermined a Puma hind-foot tracks with plantar pad widths >50 mm wide are assumed to be male; 50 mm are assumed to be female (Logan and Sweanor 2001:399-412). b Each capture season project researchers also recorded instances when the first puma tracks ≤1 day old were encountered on each search route each day to gather data on vulnerability to detection using methods similar to puma hunters (i.e., using roads, twotracks, ATV trails, searching canyon rims on snow). For 2013-2014 (TY5) the count was: 37 tracks of females, including 10 of those associated with cubs; 11 tracks of males; 1 track of a lone cub; and 3 tracks of undetermined sex. 25 �Table 7. Adult and subadult pumas captured for the first time, marked, collared (with canvass breakaways), and released from November 2013 to March 2014, Uncompahgre Plateau, Colorado. Puma I.D. F210 M211 F212 M213 F214 M215 M220 F222 F224 M226 Sex F M F M F M M F F M Estimated Age (mo.) 16 45 35 17 17 19 20 69 20 20 Mass (kg) 38 65 53 45 32 61 57 46 40 68 est.* Capture date 11/14/2013 1/3/2014 1/23/2014 2/5/2014 2/20/2014 3/8/2014 1/16/2014 2/1/2014 2/12/2014 3/8/2014 Capture method Cage trap Cage trap Dogs Dogs Dogs Dogs Dogs Dogs Dogs Dogs *Mass was estimated. Scales not available. Location Loghill Mesa Loghill Mesa Roubideau Canyon Cushman Creek Loghill Mesa McKenzie Mesa San Miguel Canyon Big Bucktail Canyon San Miguel Canyon Uncompahgre River Canyon Table 8. Pumas that were captured and observed with aid of dogs, biopsy-darted and given specimen numbers (e.g., PM1067, M for male, F for female), but were not handled at that time for safety reasons, February 2014, Uncompahgre Plateau, Colorado. Puma sex & I.D. PF1093 Age stage or months Adult Capture date 2/1/2014 PM1095 Adult 2/11/2014 Location Comments Monitor Creek of Roubideau Creek Spring Creek Notch in apex of left pinna. Estimated mass 41kg. Probable abdominal hernia. Recaptured 2/12/214 and could not be handled due to dangerous tree. 26 �Table 9. Pumas recaptured January 2014 to March 2014, Uncompahgre Plateau, Colorado. Puma I.D. F171 Recapture Date 1/14/2014 Mass (kg) 41 Estimated Age (mo.) 54 Capture Method/ Location Dogs/Dolores Canyon F95 1/24/2014 43 77 Dogs/Shavano Valley F118 1/24/2014 44 69 M211 F176 1/29/2014 2/4/2014 Observed Observed 45 43 2/5/2014 Observed 17 Dogs/San Miguel Canyon Dogs/Loghill Mesa Dogs/Horsefly Canyon (West) Dogs/Cushman Creek 2/8/2014 Observed 17 Dogs/Dry Creek Basin F28 2/8/2014 2/28/2014 2/11/2014 Observed Observed Observed 17 18 132 Dogs/Dry Creek Basin Dogs/Piney Creek Dogs/Tomcat Creek PM1095 2/12/2014 Observed Adult Dogs/Spring Creek F8 2/20/2014 44 129 M220 2/27/2014 Observed 21 M223 3/13/2014 Observed 9 M196 3/27/2014 71 58 Dogs/Big Bucktail Creek Dogs-Photo/Big Bucktail Creek Dogs/San Miguel Canyon Cage trap/Clay Creek F96 4/3/2014 45 99 Dogs/Dolores Canyon F188 4/3/2014 Observed 11 Dogs/Dolores Canyon F186 4/11/2014 Observed 42 Dogs/McKenzie Canyon (West) F184 M213 27 Process Replaced GPS collar and inserted canvass break-away. Inserted canvass break-away on GPS collar. Replaced VHF collar and inserted canvass break-away. None. F176 climbed dangerous tree; not handled. F118 climbed dangerous tree; not handled. Could not change nonfunctional expandable cub collar. F118 climbed dangerous tree; not handled. Could not change nonfunctional expandable cub collar. None. None. F28 climbed dangerous tree: not handled to replace her non-functional GPS collar. PM1095 climbed dangerous tree; not handled. Non-functional VHF collar replaced and canvass break-away inserted. Trail camera identified the male puma we pursued as M220. None. Replaced VHF collar and inserted canvass break-away. Replaced non-functional GPS collar and inserted canvass break-away. F188 climbed a dangerous tree; not handled. Cub of F96. Tracks of sibling F187 also in association with F96 and F188, but was not captured. F186 climbed a dangerous tree; not handled. �Table 10. Summary of puma capture efforts with cage traps from October 8, 2013 to July 21, 2014, Uncompahgre Plateau, Colorado.* Month October No. of Sites 4 Carnivore activity & capture effort results Unknown puma scavenged mule deer bait on east rim Dry Creek Basin site 11/2/2013. Nonmarked male puma passed by mule deer bait at southeast Loghill Mesa site 10/19-20/2013 (possibly M211, captured 1/3/2013). Bobcats and coyotes scavenged from some of the mule deer carcasses. November 7 Subadult puma F210 captured for first time in cage trap baited with mule deer at southeast Loghill Mesa site 11/14/2013. Non-marked male puma walked past mule deer bait at southeast Loghill Mesa site 11/14/2013 (possibly M211, captured 1/3/2013). Non-marked male puma scavenged mule deer bait at southeast Loghill Mesa site 11/16/2013; set cage trap and monitored it 11/16-17/2013; but, puma did not return (possibly M211, captured 1/3/2013). Bobcats, coyotes, and a bear scavenged from some of the mule deer carcasses. December 1 Bobcat scavenged the mule deer carcass. January 1 Adult male puma M211 captured for the first time in cage trap baited with mule deer at southeast Loghill Mesa site 1/3/2014. F210 scavenged mule deer bait at southeast Loghill Mesa site 1/13/2014. Non-collared female puma scavenged mule deer bait at southeast Loghill Mesa site 1/13-14/2014 (probably F214 captured with dogs on southeast Loghill Mesa 2/20/2014). February 1 No puma visits. March 11 Adult male puma M196 recaptured in cage trap baited with mule deer at north rim Clay Creek site 3/27/2014. A subadult male puma was photographed scavenging on a mule deer carcass on McKenzie Mesa on 3/9/2014. An unknown male puma walked about 10 m from a mule deer bait at the Monitor Mesa Rim site on 3/11/2014. April 8 Subadult female F214 fed on mule deer bait and was photographed at east rim McKenzie Mesa site on 4/19 and 21/2014; no capture effort needed. Adult male M196 associated with a nonmarked female puma at an elk carcass on Iron Springs Mesa. A subadult male puma was photographed scavenging on a mule deer carcass on McKenzie Mesa on 4/4/2014. Black bears scavenged on elk bait. July 1 No puma or other carnivore visits. * We used 50 road-killed mule deer and 1 road-killed elk (at 2 sites) at 24 different sites. Of the road-killed baits, 11 of 51 (21.6%) were scavenged by pumas. Table 11. Puma cubs sampled August 2013 to July 2014 on the Uncompahgre Plateau Puma Study area, Colorado. a Cub I.D. Sex Estimated birth datea Estimated age at capture (days) Mass (kg) F203 F204 M205 M206 F207 F208 F209b M216 M217 M221 M223 M225 F F M M F F F M M M M M 7/12/2013 7/12/2013 7/12/2013 7/31/2013 7/31/2013 7/31/2013 7/31/2013 6/12/2014 6/12/2014 8/2013 6/2013 9/2013 35 35 35 28 28 28 28 25 25 167 244 183 1.95 2.2 2.5 1.8 1.7 1.7 1.5 1.9 1.75 11.0 32 12 Mother Estimated age of mother at birth of this litter (mo) F137 54 F171 45 F210c F197 F28 Not captured 23 24 123 Adult Estimated age of cubs sampled at nurseries is based on the starting date for GPS location and radio-telemetry foci for mothers at nurseries, and development characteristics of cubs caught with mothers without radiocollars or mothers with non-functioning radiocollars. b Cub F209 was tattooed and ear-tagged, but was not radio-collared. c A third cub was in F210’s litter, sibling of M216 and M217, but it could not be handled (crawled into a hole). 28 �Table 12. Summary of puma capture efforts with dogs, December 2004 to April 2014, Uncompahgre Plateau, Colorado. Period Dec. 2, 2004 to May 12, 2005 Nov. 21, 2005 to May 26, 2006 Nov. 13, 2006 to May 11, 2007 Track detection effort 109/78 = 1.40 tracks/day Pursuit effort 35/78 = 0.45 pursuit/day Puma capture effort 14/78 = 0.18 capture/day Effort to capture an independent puma for the first time 11 pumas captured for first time 11/78 = 0.14 capture/day 78/14 = 5.57 day/capture 14/82 = 0.17 capture/day 78/11 = 7.09 day/capture 149/82 = 1.82 tracks/day 78/35 = 2.23 day/pursuit 43/82 = 0.52 pursuit/day 177/78 to 182/78 = 2.27-2.33 tracks/day 82/43 = 1.91 day/pursuit 45/78 to 47/78 = 0.58-0.60 pursuit/day 82/14 = 5.86 day/capture 22/78 = 0.28 capture/day 78/47 to 78/45 = 1.66-1.73 day/pursuit 49/77 = 0.64 pursuit/day 78/22 = 3.54 day/capture 78/7 = 11.14 day/capture 20/77 = 0.26 capture/day 7 pumas captured for first time 7/77 = 0.09 capture/day 77/20 = 3.85 day/capture 24/71 = 0.34 capture/day 77/7 = 11.00 day/capture Nov. 19, 2007 to April 24, 2008 Dec. 9, 2008 to April 30, 2009 217/77 to 218/77 = 2.82-2.83 tracks/day 198/71 to 202/71 = 2.79-2.84 tracks/day 77/49 = 1.57 day/pursuit 75/71 to 78/71 = 1.06-1.10 pursuit/day Dec. 15, 2009 to April 30, 2010 Nov. 16 and Dec. 14, 2010 to April 22, 2011 266/86 = 3.09 tracks/day 71/75 to 71/78 = 0.91-0.95 day/pursuit 93/86 = 1.08 pursuit/day 300/81 = 3.70 tracks/day Dec. 27, 2011 to April 12, 2012 268/79 = 3.39 tracks/day Jan. 1, 2013 to April 18, 2013 229/74 = 3.09 tracks/day 7 pumas captured for first time 7/82 = 0.08 capture/day 82/7 = 11.71 day/capture 7 pumas captured for first time 7/78 = 0.09 capture/day 9 pumas captured for first time 9/71 = 0.13 capture/day 71/24 = 2.96 day/capture 71/9 = 7.89 day/capture 26/86 = 0.30 capture/day 9 pumas captured for first time 9/86 = 0.11 capture/day 86/93 = 0.92 day/pursuit 99/81 = 1.22 pursuit/day 86/26 = 3.31 day/capture 52/81 = 0.64 capture/day 86/9 = 9.56 day/capture 81/99 = 0.82 day/pursuit 81/52 = 1.56 day/capture 81/15 = 5.40 day/capture 89/79 = 1.13 pursuit/day 26/79 = 0.28 capture/day 11 pumas captured for first time 11/79 = 0.14 capture/day 79/89 = 0.89 day/pursuit 79/26 = 3.04 day/capture 79/11 = 7.18 day/capture 82/74 = 1.11 pursuit/day 42/74 = 0.57 capture/day 12 pumas captured for the first time 12/74 = 0.16 capture/day 74/82 = 0.90 day/pursuit 74/42 = 1.76 day/capture 74/12 = 6.17 day/capture 29 15 pumas captured for first time 15/81 = 0.18 capture/day �Table 12. continued. Period Track detection effort Pursuit effort Puma capture effort Effort to capture an independent puma for the first time Jan. 9, 2014 To April 31, 2014 361/75 = 4.81 tracks/day 82/75 = 1.09 pursuit/day 29/75 = 0.39 capture/day 10 pumas captured for the first time 10/75 = 0.13 capture/day 75/82 = 0.91 day/pursuit 75/29 = 2.59 day/capture 75/10 = 7.50 day/capture 30 �Table 13. Individual puma reproduction histories, Uncompahgre Plateau, Colorado, 2005-2014. Consort pairs and estimated agesa Female Age (mo.) Male Age (mo.) F2 F2 F2 F3 F3 F3 F3 F3 F7 F7 F7 F8*e F8 F8 F8 F16 F16 F16 F23* F23 53 67 89 36 50 62 84 107 67 82 106 24 37 60 95 32 52 75 21 45 F23 80 F24 F24 75 114 F25 F25 F25 F25 74 94 110 129 F28* F28 F28 F28 F28 F30* F50 F54 F70* F70 F70 F72* F72 F72 36 48 68 112 123 48 21 24 38 52 76 28 51 64 F75 F75 F93 F93 F94* F94 F95 32 55 56 90 46 60 58 Dates pairs consortedb Estimated birth datec M73 49 02/28-29/08 M6 80 01/13-14/09 M27 or M29f M67 78 107 53 02/19-25/08 05/28/05 07/29/06 05/19/08 08/01/04 09/26/05 09/17/06 07/03/08 06/28/10 05/19/05 08/13/06 07/10/08 06/26/05 08/13/06 05/29/08 04/18/11 09/22/05 05/24/07 04/15/09 05/30/06 05/23/08 01/28-31/11 04/22/11 M29 92 04/12-15/07 06/14/07 09/10 M6 37 06/22-24/05 M51 M55 60 69 03/31/08 03/28-31/10 08/01/05 04/16/07 08/19/08 3/10 M29 88 12/27-29/06 M55 34 04/16-20/07 M51 60 03/10/08 M73 61 02/11/09 M55 70 04/15/10 31 06/09/06 03/30/07 11/08 07/12 06/13 07/17/07 07/01/06 07/01/06 06/05/08 08/31/09 08/18/11 07/09/08 06/12/10 07/15/11 08/07 05/07/09 08/07 06/16/10 05/27/09 07/15/10 06/17/13 Estimated birth interval (mo.) Estimated gestation (days) Observed number of cubsd 19.9 22.7 91-92 23.8 87-93 3 2 4 1 2 3 3 2 2 4 3 2 4 2 2 4 4 3 3 3 Nonfunct.GPS 84-86 2 90-93 4 3 14.0 22.0 13.8 11.7 21.5 23.8 93-95 94 89-92 14.9 23.9 13.4 22.5 34.7 Nonfunct.GPS 90-91 1 1 2 3 20.5 16.1 Nonfunct.GPS 11.7 92-93 88-92 14.8 23.6 87 23.1 13 23.2 93 13.3 91 2 ≥2 tracks 1 3m 2 3 1 1 3 3 3 1 2 3 photographed 1 2 2 2 3 3 Tracks/GPSn �Table 13 continued. Dates pairs Estimated consortedb birth datec Consort pairs and estimated agesa F96 F96 F96 F104 F111* F111 F116g F118 F118h F118 F119 F119i 55 78 88 110 32 58 36 27 50 64 66 96 expected F129* F135 F136j F136 F136 23.4 33 39 51 62 F137 F137 F140* 30 54 34 M55 Nonmarkedl 71 Unk. 05/21/10 03/19/13 08/21/10 07/27/12 05/14/13 07/08/10 06/16/10 08/25/12 2009 08/08/10 06/20/12 08/05/13 08/09 02/12 expected 08/03/13 07/06/11 07/10/11 07/05/12 06/18/13 07/08/11 07/12/13 06/25/13 Estimated birth interval (mo.) Estimated gestation (days) Observed number of cubsd 92 4 3 2 3 2 2k 2 3 3 1 2 1 plus 1-2 uterine scars GPS datap 2 ≥1 remains 2 3 23.2 9.6 26.3 22.4 13.5 29 expected 12 11 ≥1 3 GPS data, uterine scarsq F152* 25.7 08/08/12 2 F171 22 08/11 2 F171 45 07/31/13 4 F172 48 06/25/13 1 F176 35 06/13 3 F181* 28 08/05/13 ≥1o F197* 24 08/13/13 2 F210* 23 06/12/13 3 a Ages of females were estimated at litter birth dates. Ages of males were estimated around the dates the pairs consorted. b Consort pairs indicate pumas that were observed together based on GPS data or VHF location data. c Estimated birth dates were indicated by GPS data of mothers at nurseries or by back-aging cubs to approximate birth date. d Observed number of cubs do not represent litter sizes as some cubs were observed when they were 5 to 16 months old after postnatal mortality could have occurred in siblings. Only cub tracks were observed with F28. e Asterisk (*) indicates first probable litter of the female, based on known history or nipple characteristics noted at first capture of the female. f A radio-collared, ear-tagged male puma was visually observed with F23 on 2/25/08. Both M27 and M29 wore non-functional GPS collars in that area at the time. g When captured on 1/20/10, puma F116 was in association with 2 large cubs which were not captured. h Two cubs observed with F118 south of Norwood 9/24/2012. i F119 died of a ruptured uterus and internal bleeding on 1/28/12. Cub in uterus in third trimester; 1-2 uterine scars indicated expulsion of 1-2 fetuses. j Remains of F136’s cubs found 8/9/11. Cause of death predation by puma or black bear. k Tracks evidence of one other cub in association with F111 and cub F184, but not captured and marked; probably M213. l A non-marked adult male puma was photographed consorting with adult female pumas F136 and F182 at the same time on the NE rim of Loghill Mesa on 03/19-20/13. m F28 entire litter lost winter 2012-13; two were killed by a male puma (infanticide). n GPS data and tracks at a nursery site indicated that F95 had a litter of cubs on 6/17/13. The entire litter of cubs was apparently lost (died) by January 2014 (no evidence of cubs were found in association with F95 during winter capture efforts). 32 �o F181 had her first litter on 8/5/13 based on GPS data. When we investigated the nursery on 9/6/13, we found one cub that died of predation or infanticide. No evidence of additional cubs were found; but, they may have been completely consumed. P GPS data indicated F129 probably had her first litter of cubs on 08/03/ 13. But we could not investigate due to private land restrictions to access. F129 apparently lost the entire litter by 11/02/13. This data point is not used in Fig. 5. q We could not view F140’s cubs. But uterine scars present on F140 during necropsy and GPS location data indicated that F140 had give birth to cubs on 06/25/2013. Fate of the cubs was unknown. This data point not used in Fig. 5. 33 �Table 14. Summary for individual adult puma survival and mortality, December 8, 2004 to July 31, 2014, Uncompahgre Plateau, Colorado. Puma I.D. M1 Monitoring span 12-08-04 to 08-16-06 M4 M5 01-28-05 to 12-28-05 08-01-06 to 02-20-09 M6 02-18-05 to 05-21-10 M27 03-10-06 to 05-07-09 M29 04-14-06 to 02-25-09 M32 04-26-06 to 12-02-10 M51 01-07-07 to 03-20-09 M55 01-21-07 to 07-31-10 M67 08-23-07 to 12-18-11 M71 01-29-08 to 11-12-09 M73 02-21-08 to 10-26-11 M87 02-09-11 to 12-06-11 M90 11-16-10 to 11-23-10 M92 04-22-11 to 01-10-14 M100 03-27-09 to 07-31-09 M114 02-27-10 to 03-10-12 M133 11-12-10 to 12-01-10 Status: Alive/Lost contact/Dead; Cause of death Dead. Lost contact− failed GPS/VHF collar. M1 ranged principally north of the study area as far as Unaweep Canyon. M1 was killed by a puma hunter on 01-02-10 west of Bang’s Canyon, north of Unaweep Canyon, GMU 40. M1 was about 97 months old at death. Dead; killed by a male puma. Estimated age at death 37−45 months. Dead. Born on study area; offspring of F3. M5 was independent of F3 by 13 months old, and dispersed from his natal area at about 14 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months (protected from hunting mortality in buffer area) and ranged into the eastern edge of Utah (vulnerable to hunting). Killed by a puma hunter on 02-20-09 in Beaver Creek, Utah at age 54 months. Dead. M6 was struck and killed by a vehicle on highway 550 south of Colona, CO on 05-21-10. M6 was about 99 months old at death. Dead. Lost contact− failed GPS/VHF collar. Recaptured 12-02-07 & 01-22-08 by puma hunter/outfitter north of the study area. Possibly visually observed on study area with F23 on 02-25-08. Recaptured by a puma hunter/outfitter 12-11-08 & 12-28-08 north of the study area. Photographed by a trail camera on the study area (Big Bucktail Canyon) on 5 occasions: 03-27-09, 04-02-09, 04-15-09, 04-24-09, & 05-07-09. M27 was killed by a puma hunter on 12-09-09 in the North Fork Mesa Creek, Uncompahgre Plateau, GMU 61 North. M27 was about 100 months old at death. Dead. Lost contact− failed GPS/VHF collar. Possibly visually observed on study area with F23 on 02-25-08. Recaptured on study area 02-25-09, but could not be safely handled to change faulty GPS collar. M29 was killed by a puma hunter on 11-16-09 in Beaver Canyon, GMU 70 East. M29 was about 121 months old at death. Dead. Killed by a puma hunter on 12-02-10 in McKenzie Creek on the Uncompahgre Plateau study area. M32 was about 112 months old at death. Dead. Lost contact− failed GPS/VHF collar after 03-20-09. Killed by a puma hunter on 12-11-09 in Shavano Valley, Uncompahgre Plateau study area. M51 was about 77 months old at death. Dead. Killed by a puma hunter on 11-25-10 in Spring Creek Canyon on the Uncompahgre Plateau study area. M55 was about 77 months old at death. Dead. M67 is offspring of F30. Dispersed natal area. Established territory on W side U.P. study area. Killed by a puma hunter in Tabaguache Creek 12-18-2011 at age 52.9 months. Dead. Lost contact– M71 shed his VHF collar with an expansion link on about 11-1209. He was killed by a puma hunter on 12-09-09 on the west rim of Spring Creek Canyon, Uncompahgre Plateau study area. M71 was about 47 months old at death. Dead. Illegally killed 10-26-2011 in Bear Pen Gulch, upper East Fork Escalante Canyon; shot through abdomen during second rifle season. M73 was about 80 months old at death. Dead. M87 is offspring of F3. Dispersed from natal area. Established territory on W side of U.P. study area. Killed by a puma hunter in 47 Canyon, Tabaguache Canyon 12-06-2011. M87 was 41 months old at death. Dead. M90 was killed by a puma hunter on 11-23-10 on McKenzie Butte. M90 was offspring of F72, born 07-09-08. He was 28 months old at death. Dead. M92 was killed by a puma hunter on 01-10-14 on the study area in upper Cottonwood Creek, near Ute. He was 65 months old. M92 was the offspring of F25, born on 08-19-08 in Pleasant Valley Canyon. He dispersed to the west side of the study area and was recaptured there on 04-22-11 when 32 months old. M92 was recaptured 2 more times on the west slope of the study area on 02-27-13 and 03-12-13; but, we could not handle him to radio-collar him because he climbed dangerous tree. Dead. M100 was killed by a puma hunter on 01-16-10 in Naturita Canyon, GMU 70 East. M100 was about 63 months old at death. Dispersed from U.P. study area after 06-23-10. Killed by a puma hunter in Beaver Creek, NE of Canyon City, GMU59, 03-10-12. M114 was about 55 months old at death. Dead. M133 was killed by a puma hunter on 12-01-10 in Dry Fork Escalante Canyon north of the study area. M133 was about 43 months old at death. 34 �Puma I.D. M134 Monitoring span 06-01-11 to 06-10-11 M138 07-01-11 to 12-23-11 M144 09-01-11 to 02-25-13 M165 07-01-12 to 12-17-12 M178 11-13-12 to 12-11-12 M179 M180 11-18-12 to 12-29-12 07-01-13 to 12-16-13 M183 02-14-13 to 11-11-13 M190 M196 M211 M220 M226 F2 01-02-13 to 07-31-13 02-05-13 to 07-31-13 01-03-14 to 07-31-14 05-01-14 to 07-31-14 07-01-14 07-31-14 01-07-05 to 08-14-08 F3 01-21-05 to 12-11-11 F7 02-24-05 to 08-03-08 F8 F16 03-21-05 to 07-31-14 10-11-05 to 09-11-09 F23 02-05-06 to 06-06-12 F24 01-17-06 to 07-31-11 F25 02-08-06 to 02-03-11 F28 03-23-06 to 08-08-14 F30 04-15-06 to 07-29-08 F50 12-14-06 to 03-26-07 F54 01-12-07 to 08-18-07 F70 01-14-08 to 12-22-11 Table 14. Continued. Status: Alive/Lost contact/Dead; Cause of death Dead. M134 was offspring of unmarked female puma in Roubideau Canyon. Independent by about 03-28-11. Shot dead by USDA, APHIS, WS agent while in the act of attacking domestic sheep on 06-10-11 when he was 24 months old at start of adult life stage. Dead. Killed by a puma hunter in Horsefly Canyon (E) 12/23/11. M138 was about 29 months old at death. Dead. Initially captured as 18 mo. old subadult on W side U.P. study area 03-07-11. Dispersed from study area. Established adult territory on NW U.P. Killed by puma hunter 2-25-2013 in GMU 40, North Fork West Creek, Unaweep Canyon. Dead. Initially captured as 19 mo. old subadult on W side U.P. study area 02-24-12. Moved to Escalante Creek drainage by adult age 07-31-12. Killed by puma hunter 1217-2012 in GMU 62N, Dry Fork Escalante Canyon. Dead. Originally captured on the study area 11-13-12. Killed by puma hunter 12-1112 after tracking M178 off the study area and onto adjacent GMU 65. Dead. Killed by puma hunter on study area 12-29-12 at about 30 months old. Dead. Killed by a puma hunter on the study area in Monitor Creek on 12-16-13 at about 30 months old. Lost radio contact. Unknown status. GPS collar may have malfunctioned. Last live radio signal on the study area in Spring Creek on 11-11-13. Alive. Alive. Alive. Alive. M220 originally captured as a subadult male about 20 months old on 01-16-14. Alive. M226 originally captured as a subadult male about 20 months old on 03-08-14. Dead; killed by another puma (sex of puma unknown; male suspected) 08-14-08. F2 was about 92 months old at death. Dead. Killed by a puma hunter in Lindsay Creek 12-11-11. F3 was about 120 months old at death. Dead. Killed by U.S. Wildlife.Services agent 08-03-08 for predator control of depredation on domestic sheep. F7 was about 107 months old at death. Alive. Dead. F16 was struck and killed by a vehicle on Ouray County Road 1 southwest of Colona, CO on 09-11-09. F16 was about 80 months old at death. Dead. Killed by a male puma about 06-06-12. F23 was about 94 months old at death. F23 may have attempted to defend 2 cubs (F149, M161; 13.5 months old) and/or calf elk kill. Dead. Killed by a male puma in Logging Camp Draw about 09-16-11. F24 was about 126 months old at death. F24 may have attempted to defend ≥2 cubs (F147, nonmarked siblings; 12 mo. old). Dead. Lost radio contact after 09-04-09– failed GPS/VHF collar. Photographed alive with three ~9 month old cubs on 12-03-10 on Loghill Mesa. F25 shot dead by a ranch hand on 02-03-11 in Pleasant Valley, Dallas Creek because she was seen among cattle. F25 was about 138 months old at death and in excellent physical condition (49 kg). Alive as of 08-08-14. Lost radio contact after 09-25-07− failed GPS/VHF collar. Recaptured F28 on the study area on 02-11-14, but could not be handled to replace non-functional GPS collar and on 08-08-14 photographed on trail camera (P. Joseph, Nucla). Dead. Killed by another puma (sex of puma unknown) 07-29-08. F30 was about 60 months old at death. Dead of natural causes 03-26-07; probably injury or illness-related; exact agent unknown. F50 was about 30 months old at death. Dead; killed by a male puma while in direct competition for prey (i.e., mule deer fawn) 08-18-07. F54 was about 49 months old at death. Dead. Killed by a puma hunter Spring Creek 12-22-11. F70 was 80 months old at death. Her death orphaned 2 cubs, F157 and F158, at 4 months old; both starved to death about 01-15-12 at about 5 months old. 35 �Puma I.D. F72 Monitoring span 02-12-08 to 12-21-11 F74 01-15-13 to 5-16-13 F75 03-26-08 to 12-13-11 F93 F94 12-05-08 to 11-11-12 12-19-08 to 02-01-11 F95 F96 F104 08-01-09 to 07-31-14 01-28-09 to 07-31-14 05-21-09 to 01-31-12 F110 09-21-09 to 02-25-10 F111 01-01-10 to 12-21-13 F113 01-26-10 to 06-06-10 F116 01-20-10 to 09-20-11 F118 F119 02-25-10 to 07-31-13 03-25-10 to 01-28-12 F135 01-01-11 to 09-20-11 F136 01-20-11 to 09-30-13 F137 F140 01-21-11 to 01-09-14 08-01-12 to 11-04-13 F143 F152 F163 F171 02-15-11 to 07-31-14 06-16-12 to 12-23-12 07-01-12 to 07-31-14 01-20-12 to 06-01-14 F172 03-28-12 to 02-08-14 F176 F177 F181 F182 10-17-12 to 07-31-14 10-27-12 to 12-10-12 04-01-13 to 07-31-14 02-04-13 to 08-25-13 F186 F194 F197 03-30-13 to 07-31-14 01-29-13 to 06-17-13 08-01-13 to 07-31-14 Table 14 continued. Status: Alive/Lost contact/Dead; Cause of death Lost radio contact after 12-02-10. F72 recaptured in Fisher Creek on 03-18-11, but could not be handled to replace non-functional GPS collar. Photographed on Miller Mesa S of U.P. study area on 12-18 to 21-11 with 3 new cubs born about July 2012. Lost radio contact after 5-16-13; radiocollar fell off after canvas breakaway tab broke; detected 6-10-13. Dead. Killed by a puma hunter in North Fork Cottonwood Creek 12-13-11. F75 was about 98 months old at death. Dead. Killed by another puma 11-11-12. Fatal bite wounds to the skull. Dead. Shot dead on 02-01-11 by USDA, APHIS, WS agent for predation on domestic elk in Happy Canyon. F94 was about 74 months old at death. Alive. Alive. Dead. Died probably of starvation associated with senescence in lower Roubideau Creek 01-31-12. F104 was about 132 months old at death. Dead. Killed by a puma hunter on 02-25-10 in GMU 70 East. F110 was about 41 months old at death. Dead. Killed by a puma hunter on12-21-13 on the study area in Cushman Creek. F111 was about 74 months old at death. Dead. F113 died 06-06-10 of injuries consistent with being struck by a vehicle. GPS data indicated that F113 had crossed highway 550 and roads on Loghill Mesa north of Ridgway 24-30 hours before she died in McKenzie Creek. F113 was about 42 months old at death. Dead. Died about 09-20-11 of unknown natural cause associated with pregnancy and birth of new litter of cubs. F116 was about 60 months old at death. Alive. Dead. Died of ruptured uterus and internal bleeding associated with pregnancy in Clay Creek Canyon 01-28-12. F119 was about 95 months old at death. Dead. Died of unknown natural cause in E Fork Dry Creek 09-20-11. Her death orphaned cubs M154 and M155 at 76 days old; both died of starvation or disease when 77 (M154) and 81 (M155) days old. Dead. F136 was killed by A.P.H.I.S., Wildlife Services Agent for depredation control. She killed one goat. F136 was about 65 months old at death. Lost radio contact. Unknown status. GPS collar malfunctioned after 01-09-13. Dead. F140 died on an unknown natural cause on the study area in lower San Miguel Canyon. She was 39 months old known age. Alive. Dead. Killed by puma hunter on study area, Spring Creek Canyon. Alive. Dead. F171 died probably of an unknown natural cause on 06-01-14. She was about 56 months old at death. Dead. Killed by a puma hunter on 02-08-14 off the study area on Dry Fork Escalante Creek. She was about 56 months old at death. Alive. Dead. Killed by puma hunter 12-10-12 in GMU 65 adjacent to study area. Alive. Dead. Killed by a vehicle on highway 550 on 08-25-13. She was about 48 months old at death. F182 was pregnant with 2 fetuses; projected birth September 2013. Alive. Dispersed, exhibited subadult behavior. Fate unknown. Censor. Alive. F197 was originally captured on the study area as a subadult about 18 months old on 02-14-03. 36 �Puma I.D. F199 Monitoring span 06-01-14 to 07-31-14 F210 03-01-14 to 07-31-14 F212 F222 F224 01-23-04 to 07-31-14 02-01-14 to 07-31-14 06-1-14 to 07-31-14 Table 14 continued. Status: Alive/Lost contact/Dead; Cause of death Alive. F199 was originally captured on the study area as a cub (born June 2012) of PF1074, and is sibling of M198. F199 separated from M198 by July 29, 2013 when 13 months old, and dispersed from her natal area by August 14, 2013 when 14 months old. She established a home range in upper Dallas Creek-to-Miller Mesa area. Alive. F210 was originally captured on the study area as a subadult about 16 months old. She was impregnated in March 2014 at 20 months old and had her first litter on 06-12-14 at 23 months old. Alive. Alive. Alive. F224 was originally captured as a subadult about 20 months old on 02-12-14. 37 �Table 15. Preliminary estimated survival rates (S) of adult-age pumas during the 4 years in the reference period (i.e., the study area is closed to puma hunting) and 5 years in the treatment period, Uncompahgre Plateau, Colorado. Survival rates of pumas estimated with the Kaplan-Meier procedure to staggered entry of animals (Pollock et al. 1989). Survival rates are for an annual survival period defined as the biological year (August 1 to July 31). Survival rates were estimated only for periods when n ≥ 5 individual pumas were monitored in the interval. Puma survival in the reference period pertained only to pumas that died of natural causes. Pumas that were killed by people in the reference period, a non-natural cause (i.e., two adult pumas: F7 for depredation control 8/3/2008 and M5 killed by a puma hunter off the protected study area and buffer zone 2/20/2009) were right censored. In the treatment period all sources of natural and human-caused mortality are considered in the survival estimates. Biological Year a Reference Annual 2 8/1/2005 to 7/31/2006 Reference Annual 3 8/1/2006 to 7/31/2007 Reference Annual 4 8/1/2007 to 7/31/2008 Reference Annual 5 8/1/2008 to 7/31/2009 Treatment Annual 1 8/1/2009 to 7/31/2010 Treatment Annual 1b 8/1/2009 to 7/31/2010 With mortalities of all marked adult males Treatment Annual 2 8/1/2010 to 7/31/2011 Treatment Annual 3 8/1/2011 to 7/31/2012 Treatment Annual 4 8/1/2012 to 7/31/2013 Treatment Annual 5 8/1/2013 to 7/31/2014 S 1.000 Females SE 0.0000 n 10 S 0.667a Males SE 0.2222a n 6a 0.909 0.0867 11 1.000 0.0000 5 0.831 0.0986 14 1.000 0.0000 7 0.875 0.1031 13 1.000 0.0000 8 0.784 0.1011 19 0.667 0.1924 8 NA (see rates above) NA NA 0.333b 0.1361b 12b 0.947c 0.0568 19 0.250 0.1082 9 0.548d 0.1063 20 0.167 0.1076 7d 0.819 0.0931 19e 0.188 0.0845 8e 0.678 0.0934 22f 0.667 0.1721 7 Adult male annual S 2005 to 2006 is probably underestimated with poor precision because 3 of the 6 pumas were GPS/VHF-monitored for 4 to 5 months at the end of the interval; 1 of 6 adult males died. b This second estimate of adult male puma survival 8/1/2009 to 7/31/2010 includes 5 males that had nonfunctional (4) or shed (1) radiocollars. All adult males with non-functional or shed radiocollars in this study survived into treatment year 1 (TY1), which was expected considering adult male survival in 3 previous years. All 5 of those adult males were detected and killed by hunters in TY1. c Only 1 of 2 adult female puma mortalities is represented in this survival analysis for 8/1/2010 to 7/31/2011, that of F94 killed for depredation control. One other adult female mortality, F25, is not represented because she wore a non-functional GPS collar making it impossible for us to monitor her survival. F25 was shot by a ranch hand on 2/3/2011 when he saw her among cattle. d Sample included F143, F163, M144, ranged on NW Uncompahgre Plateau N of the study area but not on the U.P. study area, vulnerable to annual hunting. e Sample includes F143, F163, M144, M165 that ranged on north half of the Uncompahgre Plateau north of the study area (not on the study area) and were at risk to annual sport-hunting mortality. f Sample includes F143, F163, F172 that ranged on north half of the Uncompahgre Plateau north of the study area (not on the study area) and were at risk to annual sport-hunting mortality. 38 �Table 16. Summary of subadult puma survival and mortality, December 2004 to July 2014, Uncompahgre Plateau, Colorado. Puma I.D. M5 Monitoring span 09-16-05 to 06-30-06 No. days 308 M11 06-21-06 to 12-02-07 529 F23 01-04-06 to 02-04-06 31 M31 04-19-06 to 04-26-06 7 M49 03-26-07 to 10-01-07 189 F52 01-10-07 to 05-15-07 125 F66 08-23-07 to 11-05-07 11-25-08 to 06-03-09 74 M69 01-11-08 to 04-07-08 190 87 Status Survived to adult stage. M5 was offspring of F3, born August 2004. Independent and dispersed from natal area at 13 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months (protected from hunting mortality in buffer area) and ranged into the eastern edge of Utah (vulnerable to hunting). Killed by a puma hunter on 02-20-09 in Beaver Creek, Utah at about 54 months old. Survived to adult stage. M11 was offspring of F2, born May 2005. Independent at 13 months old. Dispersed from natal area at 14 months old. Moved to Dolores River valley, CO, by 12-14-06. Killed by a puma hunter on 12-02-07 when about 30 months old. Survived to adult stage. Captured on the study area when about 17 months old. Survived to adult stage; gave birth to first litter at about 21 months old. Killed by a male puma about 06-06-12. F23 was about 94 months old at death. Survived to adult stage. M31’s estimated age at capture was 20 months. Dispersed to northern New Mexico and was killed by a puma hunter on 12-11-08 in Middle Ponil Creek, Cimarron Range. He was about 52 months old. Survived to adult stage. M49 was offspring of F50, born July 2006. Orphaned at about 9 months old, when F50 died of natural causes. Dispersed from his natal area at about 10 months old and ranged on the northeast slope of the Uncompahgre Plateau. When M49 was about 15 months old, he shed his expandable radiocollar on about 10-01-07 at a yearling cow elk kill on the northeast slope of the Uncompahgre Plateau. He was killed by a puma hunter in Blue Creek in the protected buffer zone north of the study area on 01-24-09; he was about 29 months old, a young adult. Survived to adult stage. F52 dispersed from study area as a subadult by 01-16-07. F52’s last VHF aerial location was Crystal Creek, a tributary of the Gunnison River east of the Black Canyon 05-15-07. She was treed by puma hunters on 12-29-08 on east Huntsman Mesa, southeast of Powderhorn, CO. She was about 41-43 months old and could have been in her adult-stage home range. GPS collar nonfunctional. F52 was killed by a puma hunter on 01-09-12 in North Beaver Creek SE of Powederhorn, CO. She was about 79 months old at death. Died in subadult stage. F66 was offspring of F30, born July 2007. Lost contact; her cub collar quit after 11-05-07. Recaptured as an independent subadult on her natal area 11-25-08 when 16 months old. Mother F30 was killed by a puma when F66 was 12 months old, within the age range of normal independence. F66 died of injuries to internal organs that caused massive bleeding attributed to trampling by an elk or mule deer on about 05-28-09 when she was 23 months old. Her range partially overlapped her natal area. Survived to adult stage. M69 was captured on the study area when about 14-18 months old. Emigrated from the study area as subadult by 03-1908. Last VHF aerial location was southwest of Waterdog Peak, east side of Uncompahgre River Valley on 04-07-08. M69 was killed by a puma hunter on 11-06-08 in Pass Creek in the Snowy Range, WY when he was 24 to 28 months old. 39 �Puma I.D. F95 Monitoring span 12-29-08 to 07-31-12 No. days 214 M99 02-27-09 to 04-22-09 54 M112 02-10-11 to 04-18-11 67 M115 01-13-10 to 07-21-10 189 M120 12-06-11 1 M122 08-12-10 to 04-18-11 250 M131 09-25-10 to 04-18-11 206 M134 03-28-11 to 06-10-11 74 M138 01-26-11 to 06-30-11 155 F140 01-13-12 to 07- 31-12 200 M141 12-23-11 1 M144 03-07-11 to 09-08-11 185 Table 16 continued Status Survived to adult stage. Alive. F95 is the offspring of F93, born about August 2007. She became an independent subadult by about 18 months old (02-11-09 aerial location) and an adult by about 24 month old (Aug. 2009). F95 established an adult home range adjacent to and overlapping the northern portion of her natal area. Died in subadult stage. M99 probably killed by another puma (canine punctures in skull including braincase) in Jan. 2010 when he was about 16 months old. His radiocollar quit after 54 days. Survived to adult stage. M112 was offspring of F70 born August 2009. M112 associated with F96 and her two radio-collared cubs F129 and M130 during 02-10-11 to 04-18-11. Lost contact of M112 after 04-1811. Dispersed. M112 was killed by a puma hunter 01-06-2013, GMU 73, SE of Dolores, CO; UTM: 12S, 732863E, 4146772N; age 41 months. Died in subadult stage. M115 was offspring of F28, born in Nov. 2008. He was about 14 months old when first captured on Jan. 13, 2010. When he was recaptured on 03-18-10, he had previously suffered a broken left ulna. M115 was probably independent by 07-15-10 when he was located outside of his natal area on a probably dispersal move. M115 died on about 07-21-10 apparently from complications of his broken left foreleg; probably not allowing him to kill prey sufficiently for survival. M115 was about 20 months old at death. Died in subadult stage. M120 was offspring of F3. M120 was killed by a puma hunter 12-06-11 in his natal area in Spring Creek. He was 17 months old at death. Survived to adult stage. M122 was offspring of F104, born 07-08-10. Lost contact after 04-18-11 when radio-collar malfunctioned. Dispersed. Killed by puma hunter in GMU 62, Tatum Draw, Dry Fork Escalante Creek, N of natal area 01-23-13; UTM: 12S, 735353E, 4283455N; age 30 months. Survived to adult stage. M131 was offspring of F96, born 08-21-10. Lost contact after 04-18-11 when collar malfunctioned. Dispersed. Killed by puma hunter in GMU 60, Lion Creek, extreme W CO 01-1713; UTM: 12S, 670829E, 4246980N; age 29 months old. Survived to adult stage (barely). M134 was offspring of unmarked female puma in Roubideau Canyon. Independent by about 03-28-11. Shot dead by USDA, APHIS, WS agent while in the act of attacking domestic sheep on 06-10-11 when he was 24 months old at start of adult life stage. Survived to adult stage. Entered adult life stage 07-01-11. Killed by a puma hunter 12-23-11 in Horsefly Canyon. M138 was about 29 months old at death. Survived to adult stage. Turned adult in Aug. 2012. Probably offspring of F28. Has established a home range adjacent to natal area where she was initially captured at 5 months old on 01-02-11. Died in subadult stage. M141 was killed by a puma hunter on 12-23-11 in Little Bucktail Creek. He was 16 months old at death. Survived to adult stage. Emigrated from U.P. study area. Established adult territory on northwest Uncompahgre Plateau. M144 is sibling of F145 below. Killed by puma hunter 2/25/2013 at ~41 mo. old. 40 �Puma I.D. F145 Monitoring span 03-08-11 to 09-08-11 No. days 184 F146 03-08-11 to 03-23-11 15 F147 09-16-11 to 04-12-12 209 F149 06-06-11 to 12-31-12 575 M150 03-28-11 to 04-11-11 14 F152 05-04-12 to 06-16-12 44 M153 04-12-11 to 09-06-11 147 M161 06-06-12 to 08-03-12 59 F163 01-26-12 to 07-01-12 157 M164 02-14-12 to 02-26-12 02-24-12 to 12-17-12 12 M165 M180 01-01-13 to 07-01-13 298 182 Table 16 continued Status Survived to adult stage. Emigrated from U.P. study area and to Colorado Mesa. Killed by a puma hunter 01-23-12 in West Bangs Canyon. F145 was 28 months old at death. Died in subadult stage. F146 was killed and eaten by a male puma while in competition for an adult bull elk carcass that one of the pumas killed in Coal Canyon on the study area. F146 was about 19 months old at death. Lost contact; radiocollar quit after 04-12-12. F147 orphaned at about 12 months old when her mother F24 was killed by a male puma on 09-1611. Died in subadult stage. F149 was offspring of F23, born 04-22-11. F149 (sibling of M161 below) was orphaned at 13.5 months old when her mother F23 was killed by a male puma. F149 dispersed from the natal area by 07-16-12 to E side U.P. study area when she was 14.8 months old; onto Bostwick Park, then W to Dry Creek. Killed by a puma hunter 12-31-12 in GMU 70W, Dry Creek; UTM: 12S, 713658E, 4229703N; age 20 months. Lost contact; probably dispersed. M150 was offspring of F111, born on 08-31-09. He was independent by 03-28-11 when he was 19 months old. Lost contact after 04-11-11 when M150 was in Cow Creek southeast of the study area. Survived to adult stage. F152 was independent from her mother F93 by 05-04-12 when about 23 months old. She ranged as a subadult and adult on the natal area (07-31-12). Survived to adult stage. Associated with F137 when 23 months old on 09-07-2011. Killed by Wildlife Services agent for depredation on an alpaca in Dallas Creek on 09-13-11. M153 was 23 months old at death. Died in subadult stage. M161 (sibling of F149 above) was orphaned at 13.5 months old when his mother F23 was killed by a male puma. M161 dispersed from the natal area by 06-29-12 to E side U.P. study area when he was 14 months old. He shed his expandable cub collar about 08-03-12. M161 was struck and killed by a vehicle on Dallas Divide, HWY 62 in October 2012 when he was 18 months old. Survived to adult stage. F163 was captured at about 18 months old on the study area. She emigrated from the study area and established an adult home range on the NW Uncompahgre Plateau as of July 2012 (0716-12 location). Lost contact after 02-26-12. M164 may have dispersed a long distance. Fate unknown. Survived to adult stage. M165 moved from W to E side of the study area. Appeared to establish adult home range on NE Uncompahgre Plateau. Killed by a puma hunter 12-17-12 in GMU 62N, Dry Fork Escalante Creek; UTM: 12S, 730184E, 4272500N; age about 29 months, adult stage. Survived to adult stage. M180 moved to NE Uncompahgre Plateau, ranging N of the study area. Turned to adult age (24 mo.) July 2013. Killed by a puma hunter on the study area 12-16-13 when 29 mo. old adult. 41 �Puma I.D. F181 Monitoring span 01-15-13 to 04-01-13 ~12-21-13, 02-05-14, 02-08-14 07-09-14 No. days 168 F194 01-29-13 to 6-17-13 140 F197 02-13-13 to 08-01-13 07-29-13 to 06-01-2014 171 F199 07-29-13 to 06-01-2014 308 F210 11-14-13 to 03-12-14 02-05-14 to 07-31-14 122 02-20-14 to 07-31-14 03-08-14 to 03-13-14 01-16-14 to 05-15-14 162 F184 M198 M213 F214 M215 M220 308 177 5 120 Table 16 continued Status Survived to adult stage. F181 moved from E to W side of study area. Turned to adult age (24 mo.) April 2013. Had first litter 08-05-13 at 28 mo. old. Lost contact. F184, offspring of F111, born 08-25-12. Probably was orphaned and became independent subadult when mother F111 was killed by a puma hunter on 12-21-13. Previous radio signal with F111 was on 10-10-13, before F184’s collar malfunctioned. F184 recaptured with subadult M213 (probably sibling) on 02-05-14 and 02-08-14. F184 probably photographed by trail camera on Horsefly Creek on 07-09-14 (dispersed from natal area). Survived to adult stage. F194 dispersed S to North Mt., head of Naturita Creek by 06-17-13. Estimated age 30 months in June 2013. Located again 03-31-14 on Hamilton Mesa, head of Hamilton Creek. Probably established adult home range there. Survived to adult stage. F197 ranges on W side of the study area. Turned to adult age (24 mo.) August 2013. Had her first litter 08-13-13 when 24 mo. old. Fate unknown. M198 was originally captured on the study area as a cub on 04-18-2013 (born June 2012) of PF1074, and is sibling of F199. M198 separated from F199 by 07-29-13 when 13 months old, and dispersed from his natal area and the study area by 10-23-13 when 15 months old. His radio-collar was located on mortality mode on 07-30-14 in Bear Creek San Miguel River above Sawpit. Field investigation indicated that the collar is in very steep cliffs and site could not be examined due to danger. M198 was 25 months old. Survived to adult stage. Alive. F199 was originally captured on the study area as a cub on 04-18-2013 (born June 2012) of PF1074, and is sibling of M198. F199 separated from M198 by 07-29-13 when 13 months old, and dispersed from her natal area by 08-14-13 when 14 months old. She established a home range in upper Dallas Creek-toMiller Mesa area. Survived to adult stage. F210 was first captured at about 16 mo. old on 11-14-13. Turned adult age by 20 mo. old, when she was impregnated. F210 gave birth to her first litter on about 06-12-14. Lost contact. M213 was captured on 02-05-14 in the company of subadult female F184, offspring of F111 when F184 was 17.3 mo. old. M213 probably sibling of F111 (we previously captured one, F184, of 2 cubs in the litter). If so, M213 will reach adult age (24 mo.) on 08-2514, birthday of the F111’s litter. Dispersed north of suspected natal area by 05-27-14 at age 21 months. Lost contact after 06-09-14; probably dispersed. Alive. F214 was captured on 02-20-14 at about 17 mo. old. She will turn adult age (24 mo.) in September 2014. Lost contact. M215 was captured on 03-08-14 at about 19 mo. old. Last radio location was on 03-14-14. He apparently dispersed from the study area and surrounding area. Survived to adult stage. M220 was captured on 01-16-14 at about 20 mo. old. He turned adult age (24 mo.) by about 05-01-14 and is an adult male on the study area. 42 �Puma I.D. F224 Monitoring span 02-12-14 to 06-01-14 No. days 110 M226 03-08-14 to 07-01-14 116 Table 16 continued Status Survived to adult stage. F224 was captured on 02-12-14 at about 20 mo. old. She dispersed from the southwest of the study area by 03-31-14, and since has ranged south of the study area. F224 turned adult age (24 mo.) by about 06-01-14 and has an early adult home range from Mud Spring Draw to Naturita Creek. Survived to adult stage. M226 was captured 03-08-14 at about 20 mo. old. He turned adult age (24 mo.) by about 07-01-14. He has an adult home range on the study area. 43 �Table 17. Records of pumas that dispersed from the Uncompahgre Plateau study area, December 2004 to July 2014. Puma I.D. 1st capture date on study area 1st capture location→kill or resight location (UTM, NAD27) M5 02-04-05 13S,240577E, 4251037N→ 12S,665853Ex 4277125N M11 06-27-05 13S,248278E, 4239858N→ 12S,741882Ex 4161575N 84.8 M31 04-19-06 329.8 M38 09-08-06 12S,746919E, 4225441N→ 13S,500000Ex 4050000N 13S,249200E, 4239703N→ 12S,703371E, 4316856N M39 09-11-06 71.3 M43 09-15-06 12S,724270E, 4243610N→ 12S,709889E, 4313490N 12S,760177E, 4242995N→ 12S,739859E, 4308557N M48 10-18-06 52.0 M49 12-05-06 12S,756676E, 4247777N→ 12S,704982E, 4248998N 12S,757241E, 4258259N→ 12S,693350E, 4274559N M58 06-27-07 13S,258543E, 4238071N→ 13S,274670E, 4309488N 73.2 Estimated linear dispersal distance (km)* 102.2 104.1 68.6 66.1 Puma Information M5 was offspring of F3, born August 2004. Independent and dispersed from natal area at 13 months old. Established adult territory on northwest slope of Uncompahgre Plateau at the age of 24 months (protected from hunting mortality in buffer area) and ranged into the eastern edge of Utah (vulnerable to hunting). Killed by a puma hunter on 02-20-09 in Beaver Creek, Utah at about 54 months old. M11 was offspring of F2, born May 2005. Shed expandable radiocollar 10-24 to 11-08-05. Recaptured and re-collared 04-0206. Independent at 13 months old. Dispersed from natal area at 14 months old. Moved to Dolores River valley, CO, by 12-14-06. Killed by a puma hunter on 12-02-07 when about 30 months old. M31’s estimated age at capture was 20 months. Dispersed to northern New Mexico and was killed by a puma hunter on 12-1108 in Middle Ponil Creek, Cimarron Range. He was about 52 months old. M38 was offspring of F2, born July 29, 2006. Shed his expandable radiocollar by 03-06-07. Photographs by trail camera in McKenzie Cr. of M38 & Unm. F sibling with F2 on 07-16 to 17-07 at 352-353 days old. M38 was killed by a puma hunter in Ladder Creek southwest of Grand Junction, CO on 01-07-11. He was 53.2 months old at death. M39 was offspring of F8, born August 2006. M39 was killed by a puma hunter in Bangs Canyon, GMU 40 on 03-12-10 when he was 42.8 months old. M43 was offspring of F7, born August 2006. He shed the expandable radiocollar 11-7 to 17-06, after which direct contact was lost. M43 was killed by a puma hunter 01-28-09 in Deer Creek, west slope of Grand Mesa, CO when he was 29.5 months old. M48 was the offspring of F3, born September 2006. M48 was killed by a puma hunter in Tabeguache Creek, GMU 61N on 1227-09 when he was 38.9 months old. M49 was offspring of F50, born July 2006. Orphaned at about 9 months old, when F50 died of natural causes. Dispersed from his natal area at about 10 months old and ranged on the northeast slope of the Uncompahgre Plateau. When M49 was about 15 months old, he shed his expandable radiocollar on about 10-01-07 at a yearling cow elk kill on the northeast slope of the Uncompahgre Plateau. He was killed by a puma hunter in Blue Creek GMU 61N in the protected buffer zone north of the study area on 01-24-09; he was about 29 months old. M58 was offspring of F16, born May 2007. M58 was killed by a puma hunter on 12-27-09 in the North Fork of the Gunnison River north of Paonia, GMU 521; he was 31 months old. 44 �Table 17 continued. Estimated Puma Information linear dispersal distance (km)* 66.1 M63 was offspring of F24, born July 14, 2007. He was not radiocollared as a cub. M63 was killed by a puma hunter in Calamity Creek on northwest Uncompahgre Plateau on 01-01-11. M63 was 41.5 months old at death. 97.0 M65 was offspring of F24, born July 2007. M65 was killed by a USDA, APHIS, WS agent for depredation on llamas in the Little Dolores River on 11-07-09. M65 was 27.8 months old. Puma I.D. 1st capture date on study area 1st capture location→kill or resight location (UTM, NAD27) M63 08-17-07 M65 08-17-07 M67 08-23-07 12S,738144E, 4233628N→ 12S,689111E, 4277908N 12S,738144E, 4233628N→ 12S,684084E, 4314200N 13S,257371E, 4235231N→ 12S,725113E, 4242447N M68 08-23-07 M69 01-11-08 M82 07-05-08 12S,726901E, 4243463N→ 13S,255316E, 4216768N 60.5 M83 07-05-08 90.7 M87 07-31-08 12S,726901E, 4243463N→ 12S,670949E, 4314779N 13S,239006E, 4248601N→ 12S,724325E, 4244118N M88 07-31-08 13S,239006E, 4248601N→ 12S,704835E, 4197839N 77.6 13S,257371E, 4235231N→ 12S,711262E, 4198681N 13S,248191E, 4246810N→ 13T,378900E, 4591990N 57.7 80.7 369.6 39.2 M67 was offspring of F30, born July 17, 2007 in Fisher Creek on the east slope of the study area. He was not radiocollared as a cub. M67 dispersed from the natal area and was recaptured in Tomcat Creek on the west slope of the study area on 02-24-10 when he was 31 months old. M67 is a resident adult in that area (07-3111). Killed by puma hunter in GMU61N on 12-18-11 when 52.9 months old. M68 was offspring of F30, born July 2007. He was orphaned at 12 months old when his mother was killed by a puma. He was killed by a puma hunter in the Disappointment Valley in southwest CO on 12-30-08; he was 17 months old. M69 was captured on the study area when about 14-18 months old. Emigrated from the study area as subadult by 03-19-08. Last VHF aerial location was southwest of Waterdog Peak, east side of Uncompahgre River Valley on 04-07-08. M69 was killed by a puma hunter on 11-06-08 in Pass Creek in the Snowy Range, WY when he was 24 to 28 months old. M82 was offspring of F8, born May 29, 2008; sibling of M83 below. He shed his expandable cub radiocollar after 03-20-09. M82 was killed by a puma hunter on 12-10-09 in the Beaver Creek fork of East Dallas Creek, GMU 65. M82 was 19 months old. M83 was offspring of F8, born May 29, 2008; sibling of M82 above. He was not radiocollared as a cub. M82 was killed by a puma hunter on 01-18-11 in Coates Creek west of Glade Park, CO. He was 31.6 months old at death. M87 was offspring of F3, born July 3, 2008 on the east slope of the study area; sibling of M88 below. He was not radiocollared as a cub. M87 dispersed from the natal area. He was recaptured on the west slope of the study area on 02-09-11 when he was 31 months old. M87 is was resident adult on the west slope of the study area. He was killed by a puma hunter on 12-06-11 at 41 months old in GMU 61N north of the study area. M88 was offspring of F3, born July 3, 2008 on the east slope of the study area; sibling of M87 above. He was not radiocollared as a cub. M87 dispersed from the natal area. He was killed by a puma hunter in Dawson Creek, Disappointment Valley on 11-3010 when he was 29 months old. 45 �Puma I.D. 1st capture date on study area 1st capture location→kill or resight location (UTM, NAD27) M92 09-29-08 13S,246359E, 4226949N→ 12S,750871E, 4222921N M107 06-28-09 M112 01-23-10 13S,242359E, 4252618N→ 12S,754886E, 4341330N 13S,248567E, 4240108N→ 12S,732863E, 4146772N M114 02-27-10 M117 02-05-10 M126 09-05-10 M144 03-07-33 13S,256933E, 4237862N→ 13S,492615E, 4266192N 12S,731840E, 4232346N→ 12S,743909E, 4216633N 12S,734503E, 4224636N→ 12S, 710850E, 4239350N 12S,727173E, 4242012N→ 12S,696439E, 4276888N Table 17 continued. Estimated Puma Information linear dispersal distance (km)* 21.9 M92 was offspring of F25, born August 19, 2008. He was radiocollared as a cub; last contact on 12-12-08. M92 dispersed from the natal area and was recaptured in McKenzie Creek, west slope of the study area on 04-22-11 when he was 32 months old. He could not be handled to fit a new radiocollar because of a dangerous tree. He was recaptured twice more on 02-27-13 and 03-12-13 also on the SW side of the study area, but could not be collared because he climbed dangerous trees. M92 was killed by a puma hunter on 01-10-14 in upper Cottonwood Creek; he was 65 mo. old. M92 had established an adult territory on the SW side of the study area. 89.2 M107 was offspring of F94, born May 25, 2009; sibling of F108 below. He was not radiocollared as a cub. M107 dispersed from the nata area. He was killed by a puma hunter in Cottonwood Creek near Molina, CO on 12-09-10 when he was 19 months old. 102.5 M112 was initially captured 4.7 mo. old in his natal area while dependent on his mother F70 on 01-23-10. He was recaptured 0124-11 in the natal area at 17 months old, independent of F70. M112 associated with F96 and her two radio-collared cubs F129 and M130 during 02-10-11 to 04-18-11 when he was 18-20 mo. old. Lost contact of M112 after 04-18-11. Dispersed and emigrated from the U.P. study area. M112 was killed by a puma hunter 01-06-2013, GMU 73, SE of Dolores, CO; UTM: 12S, 732863E, 4146772N; age 41 months. 237.5 M114 was initially captured at about 30 months old. Emigrated from the U.P. study area. He was killed by a puma hunter on 0310-12 in Beaver Creek, GMU59. He was about 55 months old at death. 19.7 M117 was offspring of F119. He wore an expandable cub collar, but shed the collar by 07-15-10 on the natal area when about 11 months old. M117 was killed by a puma hunter in Beaver Creek, San Miguel River at the southern extreme of his natal area on 0101-11. He was 17 months old at death. It is unknown if M117 was independent from his mother F119 at the time of his death. 27.7 M126 was offspring of F118, born Aug. 8, 2010. Lost radio contact after 03-17-11; shed his radiocollar at a mule deer cache. Dispersed from natal area. Killed by a puma hunter on 01-08-12 in Tuttle Draw WNW of Nucla, CO as 17-month-old subadult. 46.6 M144 was initially captured as an independent subadult in association with subadults F145 and F146 on the study area when ~18 months old. Mother is unknown. He moved off the study area on 03-15-11. M144 established his adult territory on northwest Uncompahgre Plateau and upper Unaweep Canyon from Sep. 2011 to 02-25-13. M144 was killed by a puma hunter 02-25-13 in GMU 40, North Fork West Creek, Unaweep Canyon. 46 �Puma I.D. 1st capture date on study area 1st capture location→kill or resight location (UTM, NAD27) M161 01-23-12 12S,727932E, 4239430N→ 13S,247567E, 4220129N F52 01-10-07 13S,258058E, 4236260N→ 13S,319217E, 4240467N F97 02-04-09 12S,727529E, 4237648N→ 12S,705930E, 4227299N F106 06-14-09 12S,736451E, 4240278N→ 13S,258089E, 4235866N F108 06-28-09 13S,242359E, 4252618N→ 12S,752013E, 4263883N M122 08-12-2010 M131 09-25-10 12S,746164E, 4276613N→ 12S,735353E, 4283455N 12S,760695E, 4243505N→ 12S,670829E, 4246980N F143 02-15-11 12S,723748E, 4238579N→ 12S,721795, 4264246 Table 17 continued. Estimated Puma Information linear dispersal distance (km)* 49.2 M161 (sibling of F149) was orphaned when his mother F23 was killed by a male puma on 06-06-12; he was 411 days (13.5 mo.) old. M161 dispersed from the natal area by 06-29-12 when he was 14 months old and moved to the east slope of the U.P. study area. M161 shed his expandable cub collar about Aug. 3, 2012 in head of E Fk. Dry Creek. He was struck and killed by a vehicle on highway 62 at Dallas Divide in October 2012; he was 18 mo. old. 61.1 F52 was captured on the study area when about 18-20 months old. Dispersed from study area as a subadult by Jan. 16, 2007. F52’s last VHF aerial location was Crystal Creek, a tributary of the Gunnison River east of the Black Canyon 05-15-07. She was treed by puma hunters on 12-29-08 on east Huntsman Mesa, southeast of Powderhorn, CO. She was about 41-43 months old . F52 was treed again by puma hunters on about 12-16-09 south of Powderhorn: 13S,319480E,4233219N. F52 was about 53-55 months old. This suggests that F52 has an adult home range in that area. F52 was killed by a puma hunter on 01-09-12 in North Beaver Creek SE of Powederhorn, CO. She was about 79 months old at death. 24.0 F97 was offspring of F23, born May 23, 2008. She was radiocollared at 8.5 month old in San Miguel Canyon; but, lost contact on 05-12-09 after F97 shed the radiocollar at an elk cache. F97 dispersed from the U.P. study area. She was killed by a puma hunter on 01-22-12 in Dry Creek west of the U.P. study area when she was 43.9 months old. 46.9 F106 was offspring of F75, born May 7, 2009. She wore an expandable cub collar, but shed it about 03-23-10. F106 dispersed from the natal area and moved to the east slope of the study area where she was photographed at one of our scent station cameras at the mouth of Fisher Creek from 02-27-11 to 03-03-11. She was identified by her eartag. F106 was 21 months old. 18.2 F108 was offspring of F94, born May 25, 2009; sibling of M107 above. She was fitted with an expandable cub collar; but, shed the collar in the original nursery due to failure of the fastener. F108 dispersed from the natal area. She was killed by a puma hunter on the study area on 11-29-10 when she was 17 months old. 12.9 M122 was offspring of F104, born July 8, 2010. Fitted with expandable cub collar 08-12-10. Lost contact 04-28-11 due to transmitter malfunction. Killed by puma hunter N of natal area 01-23-13 at 30 mo. old. 90.1 M131 was offspring of F96, born August 21, 2010. Lost contact after 07-21-11. Shed his radiocollar about 07-27-11. Survived to recapture on 02-02-12 at 17.4 months old, with sibling F129; neither handled due to dangerous trees. Emigrated from U.P. study area. Killed by a puma hunter 01-17-13 at 29 mo. old in GMU 60 in western Colorado near border with Utah. 25.7 F143 was captured on the study area when about 24 months old. Dispersed N on the Uncompahgre Plateau and established an adult home range on the NW portion of the Uncompahgre Plateau, GMU61N. 47 �Puma I.D. 1st capture date on study area 1st capture location→kill or resight location (UTM, NAD27) F145 03-18-11 12S,727181E, 4241468N→ 12S,705833E, 4312909N F149 06-06-11 12S,729993E, 4242329N→ 12S,713658E, 4229703N F163 01-26-12 M165 02-24-12 12S,732153E, 4232452N→ 12S,695407E, 4280753N 12S,722816E, 4246926N→ 12S,730814E, 4272500N F194 01-29-13 12S,742443E, 4225259N→ 12S,729101E, 4201962N M198 04-10-13 12S,749316, 4222763N→ 13S,238781E, 4204762N F199 04-10-13 12S,749316, 4222763N→ 13S,252306E, 4214655N F224 02-12-14 12S,722381E 4237049N→ 12S,733636E, 4218890N Table 17 continued. Estimated Puma Information linear dispersal distance (km)* 74.5 F145 was originally captured in association of M144 and F146 when ~18 months old; they may be siblings. Mother unknown. She moved off the study area with M144 on 03-15-11. F145 emigrated to Colorado Mesa. She was killed by a puma hunter 0123-12 in West Bangs Canyon. F145 was 28 months old at death. 20.7 F149 (sibling of M161) was orphaned when her mother F23 was killed by a male puma on 06-06-12; she was 411 days (13.5 mo.) old. F149 dispersed from the natal area by 07-16-12 when she was 14.8 months old and moved to the NE Uncompahgre Plateau, onto Bostwick Park, then back across Uncompahgre Plateau. She emigrated from the U.P. study area and was killed by a puma hunter 12-31-12 at 20 mo. old 60.7 F163 was initially captured at about 18 months old. She emigrated from the study area and may have established an adult home range on the N portion of the Uncompahgre Plateau as of July 2012 (0716-12 most recent location). 26.9 M165 was first captured 02-24-12 at ~19 mo. old. His origin unknown. He moved from the west slope of the U.P. study area to the east slope of the U.P. north of the study area between 05-042012 and 06-15-12. He was killed by a puma hunter in GMU 62N on 12-17-12 when he was ~29 mo. old. 26.9 F194 was first captured at ~24 mo. old on W slope of U.P. study area on 01-29-13. Her origin unknown. She emigrated from the U.P. study area heading S. An aerial location on 06-17-13 located her on North Mt. in the SW head of Naturita Creek. Another aerial location on 03-31-14 located her on Hamilton Mesa. F194 apparently has an adult home range in that area. 24.5 Fate unknown. M198 was originally captured on the study area as a cub on 04-18-2013 (born June 2012) of PF1074, and is sibling of F199. M198 separated from F199 by July 29, 2013 when 13 months old, and dispersed from his natal area and the study area by 10-23-13 when 15 months old. His radio-collar was located on mortality mode on 07-31-14 in Bear Creek San Miguel River above Sawpit. Field investigation indicated that the collar is in very steep cliffs and site could not be examined due to danger. 30.4 F199 was originally captured on the study area as a cub on 04-182013 (born June 2012) of PF1074, and is sibling of M198. F199 separated from M198 by July 29, 2013 when 13 months old, and dispersed from her natal area by August 14, 2013 when 14 months old. She established a home range in upper Dallas Creek-to-Miller Mesa area. 21.2 F224 was first captured at ~20 mo. old on the W slope of the U.P. study area on 02-12-14. She emigrated from the study area heading S by 03-31-14 when she was located in Hamilton Creek, N of Hamilton Mesa. She has ranged in that area and Naturita Creek up to 07-14-14. *Estimated linear dispersal distance (km) from initial capture site on Uncompahgre Plateau study area to hunter kill, or last recapture, radio location, or observation site. 48 �Table 18. Recorded deaths of non-marked and marked pumas struck by vehicles and other unusual causes, in chronological order, on the Uncompahgre Plateau puma study area, Colorado, from 2004 to 2013. Puma sex & ID if marked M Estimated age (mo.) Date recorded Cause of death General physical condition Location & UTM NAD27 12 09-24-04 Good F 49 07-28-05 Vehicle collision Vehicle collision Pleasant Valley, County Road 24 13S,252870E,4227520N Highway 62 east of Dallas divide 13S,250000E,4222500N F17a 11 08-18-06 F 18-24 11-06-06 F 6 01-30-07 F P1005 36 09-16-08 M 12-24 08-13-08 F61a 18 11-13-08 F 12 08-10-09 F16b 80 09-11-09 M6b 99 05-21-0 F113b 42 06-06-10 M P1018c F P1030c M P1034 M161 24 08-25-10 6 02-16-11 4 10-07-11 18 06-17-13 F182 48 08-25-13 a Vehicle collision Vehicle collision Vehicle collision Asphyxia, lodged in fork of tree Vehicle collision Vehicle collision Vehicle collision Vehicle collision Vehicle collision Vehicle collision Vehicle collision Vehicle collision Vehicle collision Vehicle collision Vehicle collision Good Not pregnant or lactating Good Good Good Unknown, decomposed Good Good Good Good Good Good Not pregnant or lactating Excellent Good Fair Unknown, decomposed Fair, pregnant with 2 fetuses Subadult marked (i.e., tattoos, eartags), but not radio-collared. Adult GPS/VHF-collared pumas. c Non-marked puma with P one-thousand number designation. b 49 Highway 550 south of Colona 13S,257602E,4242185N Highway 550 east of Ridgway State Park 13S,259843E,4235985N Highway 62 west of Dallas divide 12S,762286Ex4218992N Davis Point, Roubideau Canyon 12S, 743718E,4255277N Highway 145 west of Placerville 13S,756490E,4212336N Highway 550 east of Ridgway State Park 13S,259843E,4235985N Highway 145 east of Norwood 12S,745739E,4222548N Ouray County Road 1 13S,253733E,4240060N Highway 550 south of Colona 13S,258610E,4236805N F113 crossed Highway 550 and roads on Loghill Mesa 24-30 hours before she died in McKenzie Creek 13S,257272E,4238435N Highway 62 Leopard Creek 12S,237747E,4220330N Highway 62 Leopard Creek 12S,760953E,4216683N Highway 62 Leopard Creek 12S,762806E,4219531N Highway 62 Dallas Divide 13S,2475674220129 Highway 550 Cow Creek 13S,258958E,4236541N �Table 19. Pumas monitored with GPS collars on the Uncompahgre Plateau, Colorado, December 2004 to July 2014. Puma I.D. M1 M4 M6 M27 M29 M51 M55 M100 M133 M178 M179 M183 M211 F2 F3 F7 F8 F16 F23 Sex M M M M M M M M M M M M M F F F F F F F24 F25 F28 F30 F50 F52 F54 F70 F72 F75 F93 F95 F96 F104 F111 F113 F129 F135 F136 F137 F152 F F F F F F F F F F F F F F F F F F F F F F171 F172 F177 F181 F F F F F182 F186 F210 F F F F214 F Age stage adult adult adult adult adult adult adult adult adult adult adult adult adult adult adult adult adult adult subadult adult adult adult adult adult adult subadult adult adult adult adult adult adult adult adult adult adult adult adult adult adult subadult adult adult adult adult subadult adult adult adult subadult adult subadult Dates monitored 12-08-04 to 07-20-06 01-28-05 to 01-14-06 02-18-05 to 05-14-08 03-12-06 to 06-21-06 04-14-06 to 01-01-08 01-07-07 to 07-15-08 01-21-07 to 11-25-10 03-27-09 to 01-16-10 11-12-10 to 12-01-10 11-13-12 to 12-11-12 11-18-12 to 12-29-12 02-14-13 to 11-11-13 01-03-14 to 07-31-14 01-07-05 to 08-14-08 01-21-05 to 12-11-11 02-24-05 to 08-03-08 03-21-05 to 10-10-06 10-12-05 to 09-10-09 01-04-06 to 02-04-06 02-05-06 to 09-04-09 01-17-06 to 07-25-07 02-09-06 to 09-09-09 03-24-06 to 08-15-07 03-30-07 to 02-22-08 12-14-06 to 03-26-07 01-10-07 to 05-08-07 01-12-07 to 08-18-08 01-14-08 to 12-22-11 02-12-08 to 07-07-10 03-26-08 to 06-03-09 10-03-12 to 11-11-12 03-14-13 to 07-31-14 01-28-09 to 07-31-14 05-29-09 to 01-31-12 01-01-10 to 12-21-13 01-27-10 to 06-06-10 01-02-13 to 07-31-14 01-01-11 to 09-20-11 01-20-11 to 09-30-13 04-12-11 to 01-09-13 01-18-12 to 06-15-12 06-16-12 to 12-23-12 01-20-12 to 06-01-14 03-28-12 to 02-08-14 10-27-12 to 12-10-12 01-15-13 to 04-15-13 04-16-13 to 07-31-14 02-04-13 to 08-25-13 03-30-13 to 07-31-14 11-14-13 to 03-01-14 03-01-14 to 07-31-14 02-20-14 to 07-31-14 50 �GOAL: Strategies, Information, & Tools for Managing Healthy, Self-sustaining Puma Populations in Colorado Puma Population Puma Habitat Effects of Harvest & Other Mortality Movements & Corridors Population Dynamics: Sex & Age, Vital Rates, Mortality, Population Growth Rates Vulnerability to Harvest Human Development Habitat Use Effects of Translocation Domestic Animals Puma― Human Relationships Deer, Elk, Other Natural Prey & Species of Concern Effects of Predation Effects of Evasive Conditioning Map Prey Distribution Habitat Models Population Models Puma Prey Habitat Maps Methods for Monitoring Populations Puma―Prey Relationships Models Figure. 1. An ecologically-based conceptual model of the Colorado Puma Research Program that provides the contextual framework for this and proposed puma research in Colorado. Gray-shaded shapes identify areas of research addressed by this puma research on the Uncompahgre Plateau for the puma management goal in Colorado (at top). 51 �Figure 2. The puma study area on the southern half of the Uncompahgre Plateau, Colorado (shaded in gray) comprising the southern portions of Game Management Units (GMUs) 61 and 62 and a northern portion of GMU 70. 52 �Puma Population Trend, Uncompahgre Plateau, Colorado ra "' E ~ ..., 40 + - - -~ ' - - - - - - - - -,.--,--- ------""'- -...-- -----"4 ~ - - - -------,,,,,-------- C QJ -0 ~ 30 c. + - - - - - - - - - - - - - - -~ ~- -~ ---- QJ -0 C ci z RY4 RYS TYl TY2 TY3 TY4 TYS YEARS ~ Minimum Count Harvest reduction - All mortalities reduction Figure 3. Trends in the population of independent pumas on the Uncompahgre Plateau Puma Study Area, including Reference Years 4 and 5 (RY4, RY5) and Treatment Years 1, 2, 3, 4, and 5 (TY1, TY2, TY3, TY4, TY5). Numbers represent minimum counts that include all pumas from known radio-collared pumas, visual observations of non-marked pumas, harvested non-marked pumas, and track counts of suspected non-marked pumas on the study area during fall to spring hunting and research capture seasons, except RY5 (45), which had to be modeled from RY4 observation data (33) because the state government hiring freeze that year affected search and capture efforts. The actual minimum count for RY5 was 37 independent pumas. The quota of 8 pumas for TY1 represented a 15% harvest of the model projected 53 independent pumas expected in TY1 and was used to set the quota ahead of the hunting season. Starting in TY1, two capture teams were deployed to count pumas on the study area because the hunting season shortened our fall-winter-spring research period. We deployed a team on each the east and west sides of the study area. The minimum count for TY1 was actually 55 independent pumas, consistent with the model expected 53. Harvest reduction trend line represents the population of independent pumas after pumas harvested only on the study area by hunters. This trend line represents 11.4% to 16.7% harvest of independent pumas. All mortalities trend line represents the population of independent pumas after pumas harvested on the study area and pumas harvested when they ranged onto adjacent GMUs open to hunting 53 �and other mortalities are subtracted from the minimum count. TY1 datum includes 1 adult female and 3 adult males killed off the study area. The TY2 datum includes 1 adult male killed off the study area and 2 adult female pumas killed in February 2011 on the study area to protect livestock. The TY3 datum includes 1 adult female and 4 adult males harvested off the study area and 2 adult females that died of natural causes on the study area. The TY4 datum includes 1 adult female and 1 adult male harvested off the study area and 1 adult female that died of natural cause. The TY5 datum includes 1 adult female that died of natural cause on the study area. This trend line represents 13.6% to 31.2% harvest of independent pumas. Age structure of independent pmm1s in November 2013 at beginning of puma hunting season in Treatment Year 5. Uncompahgre Plateau. Coloi'aclo. 6 ---------------------- 5 "' 84 E 3 0 Female o2 z; ■ Ma le 1 0 1 to2 >2 to >3 to >4 to >5 to >6 to >7 to >8 to >9 to 10+ 3 4 5 6 7 Age(vears) 8 9 10 Figure 4. Estimated age structure of independent pumas in November 2013 at the beginning of the puma hunting season in Treatment Year 5 (TY5) on the Uncompahgre Plateau study area, Colorado. All these pumas were captured and sampled by researchers or harvested by hunters and examined by researchers. Mean ± SD of independent female and male ages, respectively: 4.59 ± 2.85 yr. (55.11 ± 34.17 mo.), n = 18; 2.66 ± 1.40 yr. (31.91 ± 16.74 mo.), n = 11. 54 �Puma births by month, Uncolilpahgre Phlteau, Colorado. 16 14 12 "'a,; 10 :: ;:i 8 .....0 o 6 z. 4 2 0 - I r :-1· n Ji!in. Feb. M ai·. Apr. M ay Jun!:! July AL1g. Sep. Oct. Nov. Dec;. ■ Births 2005-1014 Births1 982-l 987 Figure 5. Puma births (black bars) detected by month from May 19, 2005 to July 31, 2014 (n = 57 litters of 31 females; 53 of the litters were examined at nurseries when cubs were 25-42 days old and 4 litters confirmed by tracks of ≥1 cubs following GPS-collared mothers F28, F95 and F111 and VHF-collared mother F197 when cubs were ≤42 days old). Also shown (gray bars) are results of the earlier effort by Anderson et al. (1992:48; 1982 to 1987, n = 10 litters of 8 females, examined when cubs were <1 to 8 months old), Uncompahgre Plateau, Colorado. 55 �Appendix A. Summary of individual puma cub survival and mortality, 2005 to 2014, Uncompahgre Plateau, Colorado. Puma I.D. M5 Estimated Age at capture (days) 183 Est. Birth date ~8-1-04 Est. survival span from 1st capture to fate or last monitor date 02-04-05 to 02-20-09 Age to last monitor date alive or at death (days, birth to fate) 31 5-28-05 F10 31 5-28-05 M11 31 5-28-05 F12 42 F13 Mother I.D. Radio-collared. Survived to subadult stage by 09-16-05; independent at ~13 mo. old. Dispersed from natal area by 09-29-05 at 14 mo. old. Established territory on NW U.P. Killed by hunter in Beaver Creek, UT 02-20-09 at 54.6 months old. Radio-collared. Shed radiocollar 04-19-06 to 04-26-06. F3 Radio-collared. Dhed radiocollar 08-10-05; last tracks of F10 with mother F2 & siblings F9 & M11 observed 11-2005. F10 disappeared by 12-30-05. Radio-collared. Shed collar 10-24 to 11-08-05. Recollared on 04-02-06. Survived to subadult stage by 06-21-06, independent at 13 mo. old. Dispersed from natal area by 0711-06 at 14 mo. old. Moved to Dolores River valley in SW CO by 12-14-06. Killed by a hunter in SW CO 12-2-07 at 918 days (30 mo.) old. Radio-collared. Shed radiocollar 07-28-05―08-01-05. Tracks of F12 found in association with mother F7 on 1208-05. F12 disappeared by 01-27-06 when she was not visually observed with F7, and her tracks were not seen in association with F7’s tracks. Radio-collared. Killed and eaten by a puma possibly M5 (13 mo. old) about 08-28-05. Radio-collared. Shed radiocollar 01-20-06 to 01-25-06. Tracks of F14 were observed with tracks of mother F8 & sibling M15 on 02-07-06. Disappeared by 03-11-06, only tracks of F8 & M15 were found. Radio-collared. Shed radiocollar 06-06-06 to 06-14-06. F2 F16 308-314 Radio-collared. Shed radiocollar 06-06-06 to 06-14-06. Killed by a car on highway 550 on 08-18-06. Probably dependent on F16. Died at 10.8 months old Radio-collared. Probably killed by another puma. Multiple bite wounds to skull. Died at 10 months old. Radio-collared. Shed radiocollar 07-27-06 to 08-02-06. 244-245 Radio-collared. Shed radiocollar 05-24-06―05-25-06. F16 ~1,664 F9 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 06-27-05 to 4-19-06 06-27-05 to 11-20-05― 12-29-05 06-27-05 to 12-02-07 326-333 5-19-05 07-01-05 to 12-08-05― 01-26-06 203-252 42 5-19-05 101 F14 26 6-26-05 07-01-05 to 08-28-05 07-22-05 to 02-07-06― 03-10-06 M15 26 6-26-05 F17 34 9-22-05 F18 34 9-22-05 M19 34 9-22-05 M20 34 9-22-05 176-215 918 226-257 07-22-05 to 06-06 to 14-06 10-26-05 to 08-18-06 345-353 10-26-05 to 07-20 to 27-06 10-26-05 to 07-27 to 08-02-06 10-26-05 to 05-24-06 301-308 330 56 F2 F2 F7 F7 F8 F8 F16 F16 �Appendix A continued Puma I.D. Estimated Age at capture (days) F21 37 9-26-05 M22 37 9-26-05 M26 183 8-1-05 F33 31 5-30-06 F34 31 F35 Est. survival span from 1st capture to fate or last monitor date 11-02-05 to 08-16-06 11-02-05 to 12-21-05― 12-22-05 02-08-06 to 03-21 to 24-06 06-30-06 to 07-31-06 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 324 Radio-collared. Lost contact; radiocollar quit. Last aerial location 8-16-06, live signal. Radio-collared. Killed and eaten by male puma 12-21-05 to 12-22-05. F3 ~232-235 Radio-collared. Shed radiocollar 03-21-06 to 03-24-06. F25 63-65 F23 5-30-06 06-30-06 to 07-31-06 63-65 31 5-30-06 38 F36 29 6-9-06 29 6-9-06 M38 41 7-29-06 Radio-collared. Killed and eaten by a male puma 08-22-06. GPS data on M29 indicate he was not involved. Radio-collared. Killed and eaten by a male puma 08-22-06. GPS data on M29 indicate he was not involved. Radio-collared. Shed radiocollar found 03-06-07. Photo (trail camera in McKenzie Cr.) of M38 & Unm. F sibling with F2 on 07-16 to 17-07 at 352-353 days old. Dispersed. Killed by puma hunter 01-07-11 in GMU40 Ladder Creek, SW of Grand Junction, CO when he was 53.2 months old. Radio-collared. Shed radiocollar by 09-20-06, but seen alive on that date. Tracks of 2 cubs following F8 on 04-25-07. Survived to adult stage; dispersed from natal area. Dispersed. Killed by a puma hunter 03-12-10 in GMU 40, Bangs Canyon, when 42.8 months old. Radio-collared. Shed radiocollar by 09-20-06, but seen alive on that date. Tracks of 2 cubs following F8 on 04-25-07. F28 M37 06-30-06 to 07-07-06 07-08-06 to 07-28-06 07-08-06 to 07-28-06 09-08-06 to 07-16 to 17-07 Radio-collared. Probably killed and eaten by a male puma 08-01 to 03-06. GPS data on M29 indicate he was not involved. Radio-collared. Probably killed and eaten by a male puma 08-01 to 03-06. GPS data on M29 indicate he was not involved. Dead; research-related fatality.a Radio-collared. Assumed dead. Shed radiocollar or died (blood on collar) between 10-05-06 (last live signal) & 1013-06 (collar found). Dead; research-related fatality.b F8 M39 29 Est. Birth date 8-13-06 F40 29 8-13-06 F41 29 8-13-06 M42 29 8-13-06 09-11-06 to 09-20-06 to 04-25-07 09-11-06 to 09-20-06 to 04-25-07 09-11-06 to 10-05-06 09-11-06 to 11-27-06 86-87 74 74 352-353 1623 9 255 1307 9 255 53-61 106 57 Mother I.D. F3 F23 F23 F28 F2 F8 F8 F8 �Appendix A continued Puma I.D. Estimated Age at capture (days) M43 33 8-13-06 M44 8-13-06 33 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 09-15-06 03-01-07 09-15-06 to 02-14-07 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 200 Radio-collared. Shed radiocollar by 11-7 to 17-06. Dispersed. Killed by a puma hunter 01-28-09 in Deer Creek, west slope of Grand Mesa, CO GMU41 at 29.5 months old. Radio-collared. Shed radiocollar by 10-27-06. Treed, visually observed 02-14-07; sibling (?) M56 also captured, sampled, & marked for 1st time. M44 killed by Wildlife Services for depredation control on 12-05-07, for killing 4 domestic sheep. He was still dependent on F7. He was 15.7 months old. Radio-collared. Multiple puncture wounds on braincase― parietal & occipital regions; consistent with bites from coyote. F45 switched families, moving from F7 to F2 about 12-19 to 20-06. Last date F45 was with F2 was 04-17-07. Died 05-20 to 23-07 when she was 9.2 months old. Radio-collared. Shed collar by 12-14-06. Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-1207 in Puma Canyon. Radio-collared. Shed collar . Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-1207 in Puma Canyon. Radio-collared. Shed radiocollar. Tracks of all cubs observed following F3 12-15-06. Tracks & GPS data indicated that F3 apparently with ≥1 of her male cubs (M46, M47, M48) at 360 days old on 09-12-07 in Puma Canyon. Dispersed. Survived to adult stage; dispersed from natal area. Killed by a puma hunter 12-27-09 in Tabaguache Creek, GMU 61N when 38.9 months old. Radio-collared. M49 was orphaned when his mother died on about 03-26-07; he was ~268 days old. M49 dispersed from natal area and onto NE slope of U.P. Shed radiocollar at a yearling cow elk kill about 10-01-07; he was ~428 days old. Dispersed from natal area. Killed by a puma hunter in Blue Creek, northwest Uncompahgre Plateau (GMU 61N) 01-2409 when ~29 months old. 899 479 F45 33 8-13-06 09-15-06 to 5-20 to 23-07 280-283 M46 31 9-17-06 10-18-06 to 12-15-06 89 360 M47 M48 M49 31 31 153 9-17-06 9-17-06 7-1-06 10-18-06 to 12-15-06 to 09-12-07 10-18-06 to 12-15-06 to 09-12-07 to 12-27-09 89 360 89 360 1187 12-05-06 to 07-31-07 to 01-24-09 939 58 Mother I.D. F7 F7 F7 F3 F3 F3 F50 �Appendix A continued Puma I.D. Estimated Age at capture (days) F53 183 Est. Birth date 7-1-06 M56c 183 ~8-13-06 F57 35 4-16-07 M58 34 5-24-07 Est. survival span from 1st capture to fate or last monitor date 01-12-07 to 02-23-07 to 09-02-07 02-14-07 to 03-01-07 05-21-07 to 06-06-07 06-27-07 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 42 Radio-collared. Shed radiocollar 02-23-07. F53 visually observed by P. & F. Star (Loghill Mesa), on 09-02-07, when F53 was ~14 months old and an independent subadult. F54 Radio-collared. Shed radiocollar 2-27-07. M56 observed 0301-07. Radio-collared. Shed radiocollar 06-07-07. Live mode 0606-07. Not radio-collared. Tracks of 3 cubs observed with F16’s tracks on 04-12-08, McKenzie Butte-Pinon Ridge Pass. 3 cubs observed with F16 on 08-08-08 by B. & T. Traegde. Dispersed. Survived to adult stage. Killed by a puma hunter 12-27-09 in GMU 521, North Fork Gunnison River, when 31 months old. Radio-collared. Shed collar about 02-14-08. Observed with 11-20-07 with F16, but without siblings M58 and F61. Tracks of 3 cubs observed with F16’s tracks on 04-12-08, McKenzie Butte-Pinon Ridge Pass. Three cubs observed with F16 on 08-08-08 by B. & T. Traegde. Dead; research-related mortality.d F7 (?) Radio-collared. Radiocollar malfunction. Tracks of 3 cubs observed with F16’s tracks on 04-12-08, McKenzie Butte-Pinon Ridge Pass. 3 cubs observed with F16 on 08-08-08 by B. & T. Traegde. Dead. Died probably as independent subadult at 538 days old; struck by car on Hwy 550 mi. marker 111 N. of Ridgway, CO, euthanized by gunshot on 11-13-08. Not radio-collared. Not radio-collared. Dispersed from study area. Killed by a puma hunter 01-01-11 in Calamity Creek, GMU61N when he was 41.5 months old. Not radio-collared. Two out of potential of 4 of F24’s male cubs were visually observed with her on 4/1/08. Assume that 2 male cubs died before the age of 8.5 mo. Eartags were seen on both cubs, but the numbers were not. F16 ~428 subad. 200 52 324 434 F59 34 5-24-07 M60 34 5-24-07 F61 34 5-24-07 06-27-07 to 08-21-07 06-27-07 to 07-11 to 12-07 06-27-07 to 06-29-07 55 324 434 48-49 324 434 M62 M63 34 34 7-14-07 7-14-07 08-17-07 08-17-07 to 01-01-11 M64 34 7-14-07 08-17-07 538 1267 262 59 Mother I.D. F25 F16 F16 F16 F24 F24 F24 �Appendix A continued Puma I.D. Estimated Age at capture (days) M65 34 Est. Birth date 7-14-07 Est. survival span from 1st capture to fate or last monitor date 08-17-07 to 11-07-09 Age to last monitor date alive or at death (days, birth to fate) Not radio-collared. Two out of potential of 4 of F24’s male cubs were visually observed with her on 04-01-08. Assume that 2 male cubs died before the age of 8.5 mo. Eartags were seen on both cubs, but the numbers were not. Dispersed. Survived to adult stage. Killed by Wildlife Services for depredation control for predation on llamas in Little Dolores River, on 11-07-09 when 27.8 months old. Radio-collared. Lost contact; last location 11/5/07. No signals after that date. F66 was photographed with one male sibling, either M67 or M68, & F30 on 5/31-6/1/08. F66 was recaptured and radio-collared as a subadult on 11/25/08. She died from massive trauma & bleeding of internal organs possibly resulting from being trampled by an elk or mule deer on about 05-28-09 as an independent subadult 23 months old. Her range overlapped her natal area. Not radio-collared. M67 or M68 was photographed with sibling F66 & mother F30 on 5/31-6/1/08. Dispersed from natal area. Established adult home range on west side of Uncompahgre Plateau study area. Killed by puma hunter in GMU61N on 12-18-11 when 52.9 months old. Not radio-collared. M67 or M68 was photographed with sibling F66 & mother F30 on 05-31 to 06-01-08. Survived to subadult stage. Dispersed. Killed by a puma hunter in Disappointment Valley, CO (GMU 71) 12-30-08 at 17.5 months old. Radio-collared. Shed radiocollar between 7-9-08 and 7-1508, probably while still dependent on mother F75. Not radio-collared. Probably dead; if not killed when sibling M79 was killed, then probably would starve to death. Not radio-collared. Probably dead; if not killed when sibling M79 was killed, then probably would starve to death. 262 847 F66 37 7-17-07 08-23-07 to 05-28-09 M67 37 7-17-07 08-23-07 to 12-18-11 M68 37 7-17-07 08-23-07 to 12-30-08 F74 259 6-1-07 M76 30 M77 30 682 1615 532 5-19-08 03-12-08 to 07-09-08 06-18-08 ~87 5-19-08 06-18-08 ~87 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 403 60 Mother I.D. F24 F30 F30 F30 F75 F2 F2 �Appendix A continued Puma I.D. Estimated Age at capture (days) F78 30 M79 Est. Birth date Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 5-19-08 Est. survival span from 1st capture to fate or last monitor date 06-18-08 ~87 F2 30 5-19-08 06-18-08 87 F80 40 5-23-08 07-02-08 F81 40 5-23-08 F95 ~488 ~Aug. 2007 07-02-08 to 07-29-09 12-29-08 to 07-31-13 Not radio-collared. Probably dead; if not killed when sibling M79 was killed, then probably would starve to death. Not radio-collared. Dead. Chewed-off anterior portions of the nasals, maxilla, palate, dentaries, and pieces of the braincase, with 6 or 9 portion of yellow ear-tag and intestines and bits of skin found ~45 m from mother F2’s death site on 08/14/08. Cub death probably due to puma-caused infanticide with cannibalism at ~87 days old. Male puma scrapes, about 8, under a rock rim ~50m distance from cub remains, and made ~ time of pumas’ deaths. Not radio-collared. Apparently died before 02-04-09; no tracks found in association with F23 & siblings F81 & F97. Radio-collared. Last live location 7-29-09. F93 F97 257 5-23-08 02-04-09 to 01-22-12 1339 M82 37 5-29-08 07-05-08 to 12-10-09 560 M83 37 5-29-08 07-05-08 to 01-18-11 964 Radio-collared. F95 was offspring of F93. She survived the subadult stage and into the adult stage. Her home range overlapped her natal area. Radio-collared. Lost contact after 05-12-09; shed collar at elk kill cache on Mailbox Park. Dispersed from study area. Killed by a puma hunter 01-22-12 in Dry Creek when 43.9 months old. Radio-collared. Shed radiocollar after 03-20-09. Survived to subadult stage. Dispersed. Killed by a puma hunter in 1210-09 GMU 65, Beaver Creek fork of East Dallas Creek, when 18.4 months old. Not radio-collared. Survived; dispersed from study area. Killed by a puma hunter 01-18-11 in Coates Creek west of Glade Park, GMU40. He was 31.6 months old. 424 2,196 61 Mother I.D. F2 F23 F23 F23 F8 F8 �Appendix A continued Puma I.D. Estimated Age at capture (days) M84 36 6-5-08 F85 36 F86 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 07-11-08 to 02-11-09 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 251 6-5-08 07-11-08 to 10-01-08 118 36 6-5-08 07-11-08 to 07-23 to 08-03-08 ~48-59 M87 28 7-3-08 07-31-08 to 12-06-11 1251 M88 28 7-3-08 07-31-08 to 11-30-10 880 F89 M90 28 36 7-3-08 7-9-08 07-31-08 08-14-08 Male 7A 28-35 7-10-08 ~08-07-08 to 08-14-08 28 to 35 Male 7B 28-35 7-10-08 ~08-07-08 to 08-14-08 28 to 35 Female 7C 28-35 7-10-08 ~08-07-08 to 08-14-08 28 to 35 Radio-collared 7-11-08 to 7-22-08; collar removed because of malfunction. Not radio-collared after 7-22-08. Eartag of M84 was found by E. Phillips on 8-25-08 when mother F70’s GPS locations located her on either side of the eartag in the East fork Dolores Cyn. M84 recaptured radiocollared again 1-29-09 in Dolores Cyn. in association with F70 & F96’s family. Shed radiocollar again about 0214-09. Radio-collared. Dead. Probably died of predation or infanticide about 10-108 near elk calf kill at age 3.9 months. Radio-collared 7-22-08. Dead. Radio-collar, orange ear-tag #86 with pinna with green tattoo #86 found by J. Timmer 9-1-08. F86 died ~7-23 to 8-3-08 when mother F70’s GPS locations located her at F86 remains. Probable predation. Not radio-collared. Dispersed from natal area. Recaptured as adult on west slope of study area on 02-09-11 at 31 months old. Killed by puma hunter on 12-06-11 at 41 months old in GMU61N north of the study area. Not radio-collared. Dispersed. Killed by a puma hunter in Dawson Creek, Disappointment Valley, GMU711 on 11-3010 when 28.9 months old. Radio-collared. Radio-collared. Recaptured as young adult on study area, adjacent to natal area, on 11-16-10. Killed by a puma hunter during TY2 on 11-23-10. Not radio-collared. F7’s cubs died from starvation after they were orphaned. F7 was shot on 8-3-08 for killing domestic sheep. Not radio-collared. F7’s cubs died from starvation after they were orphaned. F7 shot on 8-3-08 for killing domestic sheep. Not radio-collared. F7’s cubs died of starvation after orphaned. F7 shot on 8-3-08 for killing domestic sheep. 867 62 Mother I.D. F70 F70 F70 F3 F3 F3 F72 F7 F7 F7 �Appendix A continued Puma I.D. Estimated Age at capture (days) M91 35 M92 Est. Birth date Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 8-19-08 Est. survival span from 1st capture to fate or last monitor date 09-29-08 455 35 8-19-08 09-29-08 976 F95 16 mo. June-07 12-29-08 F98 4-5 mo. 158 02-12-09 to 03-08-09 2-27-09 to 01-2010 146-176 M99 Sep-Oct08 Sep-Oct08 M101 35 4-15-09 157 M102 35 4-15-09 05-20-09 to 09-19-09 05-20-09 F103 35 4-15-09 159 M105 38 5-7-09 F106 38 5-7-09 05-20-09 to 09-17-09 06-14-09 to 02-09-10 06-14-09 to 02-27-11 M107 34 5-25-09 Radio-collared. Killed by a puma hunter on study area during TY1 as dependent cub on 11-17-09 at age 14.9 months. Radio-collared. Lost contact after 12-12-08. Dispersed from natal area. Recaptured in McKenzie Creek, west slope of study area on 04-22-11 when 32 months old. Due to dangerous tree, could not handle him safely to fit new radiocollar. Radio-collared. Survived to adult stage. Established adult home range overlapping mother F93’s home range. To date, July 2012, F95’s home range mainly adjacent to N side of natal area. Radio-collared. Died; probably killed by male puma (infanticide). Radio-collared. Offspring of non-marked female. Last location 4-22-09 on Paterson Mt. Died as 16-month old subadult in San Miguel Canyon. Probably killed by another puma; apparent canine punctures to braincase. Radio-collared. Died; killed by puma M55 after he was orphaned due to death of mother F16 by vehicle strike. Radio-collared. Lost contact after 09-04-09. Did not find evidence of M102 associated with deaths of siblings M101 and F103. But M102 probably died. Radio-collared. Died; killed by puma M55 after she was orphaned due to death of mother F16 by vehicle strike. Radio-collared. Lost contact after 02-09-10 due to shed collar. Not radio-collared at nursery; F75 returned to nursery during handling. Radio-collared later on 2-10-10. Lost contact due to shed collar 3-16 to 29-10. F106 dispersed from natal area and was photographed at 21 months old at camera and scent-rub station on east slope of Uncompahgre Plateau on 02-27-11. Not radio-collared; too small. Recaptured 02-24-10; not collared. Dispersed. Killed by a puma hunter in Cottonwood Creek near Molina, CO on 12-09-10 when he was 19 months old. 06-28-09 to 02-24-10 488 278 275 661 241 63 Mother I.D. F25 F25 F93 Unm.F Unm.F F16 F16 F16 F75 F75 F94 �Appendix A continued Puma I.D. Estimated Age at capture (days) F108 34 Est. Birth date 5-25-09 Est. survival span from 1st capture to fate or last monitor date 06-28-09 to 03-05-10 Age to last monitor date alive or at death (days, birth to fate) 553 M109 M112 34 145 5-25-09 8-31-09 06-28-09 05-04-10 to 01-06-13 1,225 M115 427 Nov.-08 07-21-10 610 M117 193 Aug.-09 02-05-10 to 01-01-11 518 P1016(M) 39 6-12-10 06-12-10 to 07-21-10 39 P1017(M) 39 06-12-10 06-12-10 to 07-21-10 39 M120 30 06-28-10 07-28-10 to 12-02-10 526 M121 30 06-28-10 273 M122 35 07-8-10 07-28-10 to 03-28-11 08-12-10 to 04-28-11 F123 29 07-15-10 217 F124 29 07-15-10 08-13-10 to 02-17-11 08-13-10 to 02-16-11 931 216 64 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. Shed radiocollar at nursery; fastener failed. Recaptured and re-collared 2-24-10. Shed collar ~3-5-10. Dispersed from natal area. Killed by a puma hunter on the study area during TY2 on 11-29-11 at 18.1 months old. Not radio-collared; too small. Radio-collared. Lost contact after 05-4-10 (last live signal) possibly due to failed transmitter. Recaptured and re-radiocollared on 01-24-11. Independent subadult during 02-10-11 to 04-18-11. Lost contact after 04-18-11. Dispersed. Killed by a puma hunter 01-06-13 in GMU73 SE of Dolores, CO; age 41 months. Radio-collared. M115 died as a subadult (~20 mo. old) due to complications of a broken left foreleg (natural cause). Radio-collared. Lost contact after 5-14-10 (last live signal); shed collar found on 7-15-10 in the natal area. Killed by a puma hunter on the natal area in Beaver Creek, GMU70E, off the U.P. study area on 01-01-11 when he was 17 months old. Not radio-collared. Monitored at nursery via mother’s GPS/VHF collar. Found dead at nursery due to infanticide by puma M32 on same day as our investigation of nursery. Not radio-collared. Monitored at nursery via mother’s GPS/VHF collar. Found dead at nursery due to infanticide by puma M32 on same day as our investigation of nursery. Radio-collared. Lost radio contact after 12-02-10. Killed by a puma hunter on his natal area on 12-06-11 when he was 17.2 months old. Radio-collared. Lost radio contact after 03-28-11. F94 Radio-collared. Lost radio contact after 04-28-11. Tracks of 2 other siblings of M122 observed on 01-11-11 (neither cub marked). M122 killed by a puma hunter in Tatum Draw, Dry Fk. Escalante Cr., GMU62N, 01-23-13; age 30 months. Radio-collared. Killed on 02-17-11 for depredation control on domestic elk by Wildlife Services agent. Radio-collared. Killed on 02-16-11 for depredation control on domestic elk by elk farm manager. F104 F94 F70 F28 F119 F72 F72 F3 F3 F94 F94 �Appendix A continued Puma I.D. Estimated Age at capture (days) M125 29 07-15-10 M126 28 08-08-10 M127 28 M128 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 08-13-10 to 02-01-11 09-05-10 to 01-08-12 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 201 08-08-10 09-05-10 to 09-10-11 398 28 08-08-10 198 F129 35 08-21-10 09-05-10 to 02-22-11 09-25-10 to 02-02-12 M130 35 08-21-10 09-25-10 to 02-02-12 M131 35 08-21-10 09-25-10 to 07-21-11 F132 35 08-21-10 09-25-10 35 M134 ~18 mo. ~June-09 12-14-10 to 06-10-11 740 M139 36 04-18-11 05-24-11 to 07-29-11 102 F148 36 04-18-11 05-24-11 to 07-29-11 102 Radio-collared. Killed on 02-01-11 for depredation control on domestic elk by Wildlife Services agent. Radio-collared. Lost radio contact after 03-17-11; shed his radiocollar at a mule deer cache. Dispersed from natal area. Killed by a puma hunter on 01-08-12 in Tuttle Draw WNW of Nucla, CO, GMU61N, as 17-month-old subadult. Radio-collared. Lost radio contact after 07-01-11; shed his radiocollar about 07-01-11. Found dead 09-14-11 on natal area; killed by another puma on about 09-10-11 at age 13 months. Radio-collared. Lost radio contact after 02-22-11; radiocollar probably quit. Radio-collared. Fate unknown. Transmitter on mortality mode on 04-28-11. Unable to get to collar until 06-23-11 due to high spring run-off, by then the transmitter had quit. Survived to recapture on 02-02-12 at 17.4 months old, with sibling M131; neither handled due to dangerous trees. Radio-collared. Died of natural causes associated with injury to right shoulder during first move away from nursery about 10-23-10. Radio-collared. Lost contact after 07-21-11. Shed his radiocollar about 07-27-11. Survived to recapture on 02-0212 at 17.4 months old, with sibling F129; neither handled due to dangerous trees. Dispersed. Killed by a puma hunter in Lion Cr., extreme western CO, GMU60; age 29 months. Not radio-collared. Too small for collar design. Fate unknown. Apparently died; not with F96 and siblings F129 and M130 on 02-02-12. Radiocollared as dependent large cub. Independent by about 03-28-11. Dead; killed for depredation control by Wildlife Services agent on 06-10-11. He was about 24 mo. old Radio-collared. Dead of infanticide and cannibalism along with sibling F148; killed and eaten by female or subadult male puma about 07-29-11. Radio-collared. Dead of infanticide and cannibalism along with sibling M139; killed and eaten by female or subadult male puma about 07-29-11. 221 530 530 334 65 Mother I.D. F94 F118 F118 F118 F96 F96 F96 F96 Unm. F F8 F8 �Appendix A continued Puma I.D. Estimated Age at capture (days) F140 ~5 mo. Est. Birth date Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. ~Aug.10 Est. survival span from 1st capture to fate or last monitor date 01-02-11 to 07-31-13 1,096 Radio-collared. Lost contact. Shed first collar about 01-2411. Recaptured and re-collared on 04-01-11. Shed second collar after 04-18-11. Recaptured and re-collared 01-12-12 as 17-month-old subadult on natal range. Survived to adult stage. Radio-collared. Lost contact; shed radiocollar about 03-2911. Recaptured, but could not be handled safely on 04-0111. Killed by a puma hunter on 12-23-11 in his natal area; age 16 months. Radio-collared. Lost contact after 04-18-11 due to shed collar. Struck by vehicle and killed on state highway 62 in Leopard Creek, south boundary of study area on 02-16-11. Radio-collared. Orphaned at about 12 months old when her mother F24 was killed by a male puma on 09-16-11. She ranged in her natal area until her radiocollar quit after 0412-12. Radio-collared. F149 (sibling of M161) was orphaned when her mother F23 was killed by a male puma on 06-06-12; she was 411 days (13.5 mo.) old. F149 dispersed from the natal area by 07-16-12 when she was 14.8 months old. Radio-collared. M151 was independent by 03-28-11 at 19 mo. old. He dispersed from the natal area by 04-11-11 at 19.5 mo. old. Contact lost after 04-11-11. Radio-collared. Lost contact after 03-07-11 (GPS location of mother F111 at shed collar of M151). Radio-collared. Lost contact after 03-21-11; shed collar. Recaptured 01-18-12; fit with GPS collar at 19 months old. Ranged on natal area as adult (philopatric). First litter on 0808-12 at 26 mo. old. Killed by puma hunter on 12/23/12. Radio-collared. M154 probably died of starvation following natural death of his mother F135. Sibling M155 also died. Radio-collared. M155 died of starvation following death of his mother F135. Sibling M154 also died. Radio-collared. M156 shed the collar about 09-05-11. He was 59 days old. Unk./ F28? M141 ~5 mo. ~Aug.10 01-02-11 to 04-01-11 509 M142 ~5 mo. ~ 6 mo. 01-02-11 to 04-18-11 02-16-11 258 P1030 F147 ~7 mo. ~Aug.10 ~Aug.10 ~Sep.-10 04-21-11 to 07-31-11 315 F149 45 04-22-11 06-06-11 to 07-16-12 451 M150 525 08-31-09 02-07-11 to 04-11-11 588 M151 253 06-16-10 264 F152 271 06-16-10 02-24-11 to 03-07-11 03-14-11 to 12-23-12 M154 42 07-06-11 77 M155 42 07-06-11 M156 43 07-08-11 08-16-11 to 09-21-11 08-16-11 to 09-25-11 08-20-11 to 09-05-11 183 776 81 56 66 Unk./ F28? Unk./ F28? Unk. F24 F23 F70 F111 F93 F135 F135 F137 �Appendix A continued Puma I.D. Estimated Age at capture (days) F157 40 08-18-11 F158 40 M159 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 09-27-11 to 01-15-12 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 150 08-18-11 09-27-11 to 01-15-12 150 40 08-18-11 09-27-11 to 12-01-11 105 M161 276 04-22-11 01-23-12 to 10-15-12 543 M162 183 07-25-11 01-25-12 to 06-11-12 322 M168 37 07-27-12 09-02-12 to 09-12-12 47 F169 37 07-27-12 09-02-12 to 09-12-12 47 M170 137 08-29-11 199 P1033 22 07-10-11 01-13-12 to 03-12-12 NA Radio-collared. F157 with sibling F158 died of starvation following death of his mother F70 due to hunter harvest on 12-22-11. Cubs died 24 days after their mother died. The cubs were 150 days old. Radio-collared. F158 with sibling F157 died of starvation following death of his mother F70 due to hunter harvest on 12-22-11. Cubs died 24 days after their mother died. The cubs were 150 days old. Radio-collared. M159 probably died about 12-01-11 when he was located with his family (F70, siblings F157, F158). He was not located with them on 12-12-11 and was not observed with them on 12-13-11. He was 105 days old on 12-01-11. Radio-collared. M161 (sibling of F149) was orphaned when his mother F23 was killed by a male puma on 06-06-12; he was 411 days (13.5 mo.) old. M161 dispersed from the natal area by 06-29-12 when he was 14 months old. Shed his expandable collar about 08-03-12. Was struck and killed by a vehicle on Dallas Divide, Hwy 62 in October 2012 when 18 mo. old. Radio-collared. M162 probably was orphaned cub of nonmarked adult female puma killed on Pinto Mesa 01-18-12. M162 died of starvation on 06-11-12 when he was 322 days (10.6 mo.) old. Radio-collared. Cub M168 was offspring of F96; sibling of F169 & F173. It died of infanticide, probably of a male puma, based on track sizes (fhpw = 60 mm; hhpw = 50 mm). Radio-collared. Cub F169 was offspring of F96; sibling of M168 & F173. It died of infanticide, probably of a male puma, based on track sizes (fhpw = 60 mm; hhpw = 50 mm). Radio-collared. M170 died about 03-15-12 of unknown natural cause. He was 199 days (6.5 mo.) old. Radio-collared. Cub P1033 was offspring of F136. It died of predation, probably killed by a puma or black bear in the nursery when about 22 days old, before researchers could examine the entire litter to sample and mark the cubs. 22 67 Mother I.D. F70 F70 F70 F23 Unm.F F96 F96 F171 F136 �Appendix A continued Puma I.D. Estimated Age at capture (days) F173 37 07-27-12 M174 32 M175 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 09-02-12 to 09-12-12 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death 47 08-08-12 09-11-12 to 03-10-13 181 32 08-08-12 09-11-12 to 12-11-12 126 F184 208 08-25-12 03-20-13 to 02-08-14 533 F185 ~183 ~Oct.2012 03-23-12 to 03-27-13 190 F187 31 05-14-13 06-14-13 to 04-03-14 325 F188 31 05-14-13 06-14-13 to 04-03-14 325 F189 38 06-18-13 07-26-13 to 10-19 to 20-13 93-94 Radio-collared. Cub F173 was offspring of F96; sibling of M168 & F169. It died of infanticide, probably of a male puma, based on track sizes (fhpw = 60 mm; hhpw = 50 mm). Radio-collared. Cub M174 was offspring of F152; sibling of M175. He was orphaned after his mother was killed by a hunter on 12-23-12. He was 137 days old. M174 was recaptured at 181 days old and removed from the wild and was rehabilitated at the CPW Del Norte wildlife center for re-release to the wild. He was released to the wild on NW Uncompahgre Plateau on 05-08-14 at 21 mo. old. He was shot for killing 2 domestic goats in Nucla, CO on 08-24-14. Radio-collared. Cub M175 was offspring of F152; sibling of M174. He was mauled to death probably by puma hunting dogs on about 12-11-12 when he was 126 days old. Radio-collared. Cub F184 was offspring of F111; one other sibling track was observed, but the puma was not captured (probably M213 captured with F184 02-05-14 and 02-0814. F184 still dependent on F111 on 10-10-13. F184 orphaned when mother F111 was killed by a hunter on 1221-13. F184 collar malfunctioned; lost contact. She was probably photographed on N rim Horsefly Canyon 07-09-14 and tissue sampled in Belisle device. Radio-collared. Cub F185 was offspring of a non-marked female puma in Roubideau Cyn. F185 shed her expandable collar about 7 days after initial capture. Lost contact. Fate unknown. Radio-collared. Cub F187 was offspring of F96; sibling of F188. Collar quit; but tracks of F187 present with F96 and sibling F188 when recaptured 04-03-14 at 10.5 months old. Radio-collared. Cub F188 was offspring of F96; sibling of F187. Collar quit; but F188 recaptured 04-03-14 at 10.5 mo. old. Radio-collared. Cub F189 was offspring of F136; sibling of F200 and M201. Died of starvation at 93-94 days old after F136 was killed for depredation control. 68 Mother I.D. F96 F152 F152 F111 Unm.F F96 F96 F136 �Appendix A continued Puma I.D. Estimated Age at capture (days) M191 ~183 Est. Birth date Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death ~July 2012 Est. survival span from 1st capture to fate or last monitor date 01-03-13 to 01-20-13 ~210 Radio-collared. Cub M191 apparently was offspring of F28 (with non-functional GPS collar). He was sibling of PM1068 and one other non-marked cub. M191 was killed by a non-marked male puma on about 01-20-13 along with PM1068. PM1068 was not captured and tagged. It was apparently offspring of F28; sibling of M191 and one other nonmarked cub. PM1068 was killed and partially eaten by a non-marked male puma. Radio-collared. M192 was offspring of F118; sibling of M193 & F195. M192 was independent of F118 at ~11.7 mo. old. He shed his expandable collar at a mule deer kill after 07-01-13. Radio-collared. M193 was offspring of F118; sibling of M192 & F195. M192 was independent of F118 at ~11.7 mo. old. He shed his expandable collar at a mule deer kill after 07-01-13 like sibling M192, but the siblings were not associating (kills were at different locations). Radio-collared. F195 was offspring of F118; sibling of M192 & M193. F195 shed her expandable radiocollar at an elk kill on about 03-04-13; contact lost afterwards. Radio-collared. M198 was offspring of non-marked female PF1074 (sampled by bio-dart). He was sibling of F199. Survived to subadult stage. Dispersed. Last live signal 1023-13 at 15 mo. old. Collar on mortality mode 07-31-14; fate undetermined/unknown; collar in dangerous cliffs. Radio-collared. F199 was offspring of non-marked female PF1074 (sampled by bio-dart). She was sibling of M198. Dispersed. Survived to subadult and adult stages. Radio-collared. Cub F200 was offspring of F136; sibling of F189 and M201. Died of starvation at 96 days old after F136 was killed for depredation control. Radio-collared. Cub M201 was offspring of F136; sibling of F189 and F200. Died of starvation at 93-94 days old after F136 was killed for depredation control. PM1068 ~183 ~July 2012 01-03-13 to 01-20-13 ~210 M192 199 06-20-12 01-04-13 to 07-01-13 376 M193 199 06-20-12 01-04-13 to 07-01-13 376 F195 227 06-20-12 02-01-13 to 03-04-13 258 M198 274 ~June 2012 04-10-13 to 10-23-13 ~510 F199 292 ~June 2012 04-18-13 to 07-31-14 ~790 F200 38 06-18-13 07-26-13 to 10-22-13 96 M201 38 06-18-13 07-26-13 to 10-19 to 20-13 93-94 69 Mother I.D. F28 F28 F118 F118 F118 PF1074 PF1074 F136 F136 �Appendix A continued Puma I.D. Estimated Age at capture (days) F202 35 06-25-13 F203 35 F204 Est. Birth date Est. survival span from 1st capture to fate or last monitor date 07-30-13 to 08-22-13 Age to last monitor date alive or at death (days, birth to fate) Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. 59 F172 07-12-13 08-16-13 to 03-13-14 245 35 07-12-13 08-16-13 to 03-13-14 245 M205 35 07-12-13 08-16-13 to 01-17-14 190 M206 28 07-31-13 08-28-13 to 07-30-14 365 F207 28 07-31-13 08-28-13 to 06-25-14 330 F208 28 07-31-13 08-28-13 to 06-09-14 314 F209 28 07-31-13 08-28-13 1 Radio-collared. Cub F202 was offspring of F172. No siblings were observed at the nursery; but some could have hidden. Died of predation or infanticide about 08-22-13. Radio-collared. Cub F203 offspring of F137; sibling of F204, M205. Last live signal 03-13-14 in association with F204. Radiocollar probably quit after. Radio-collared. Cub F203 offspring of F137; sibling of F203, M205. Last live signal 03-13-14 in association with F203. Radiocollar probably quit after. Radio-collared. Cub M205 offspring of F137; sibling of F203, F204. Died apparently of natural cause when about 6.2 mo. old. Radio-collared. Cub M206 offspring of F171; sibling of F207, F208, F209. Orphaned at 10 mo. old when mother died of natural cause on 06-01-14. Separated from siblings. Dispersed from natal area. Alive on 07-30-14 in Uncompahgre River bottom between Colona and Billy Cr. Radio-collared. Cub F207 offspring of F171; sibling of M206, F208, F209. Orphaned at 10 mo. old when mother died of natural cause on 06-01-14. Separated from siblings. Died apparently of natural cause about 06-25-14 at 10.8 mo. old. Radio-collared. Cub F208 offspring of F171; sibling of M206, F207, F209. Orphaned at 10 mo. old when mother died of natural cause on 06-01-14. Separated from siblings. Lost contact; last live signal 06-09-14. Radiocollar may have quit. Fate unknown as of 07-23-14. Not radio-collared; too small. Cub F209 offspring of F171; sibling of M206, F207, F208. Fate unknown. 70 F137 F137 F137 F171 F171 F171 F171 �Appendix A continued Puma I.D. Estimated Age at capture (days) M213 550 Est. Birth date 08-25-12 Est. survival span from 1st capture to fate or last monitor date 02-05-14 to 06-09-14 Age to last monitor date alive or at death (days, birth to fate) 654 Status: Alive/Survived to subadult stage/ Lost contact/Disappeared/ Dead; Cause of death Mother I.D. Radio-collared. Caught in association with subadult F184 on F111 02-05-14 and 02-08-14; probably her sibling, offspring of F111. Previously we had not captured the second cub in F111’s litter, of which F184 was one. Last live signal of M213 on 06-09-14, on NW edge of natal area. He probably dispersed. M216 25 06-12-14 07-08-14 to 49 Radio-collared. Offspring of F210; sibling of M217 and one F210 07-30-14 non-marked cub. Alive on 07-30-14. M217 25 06-12-14 07-08-14 to 49 Radio-collared. Offspring of F210; sibling of M216 and one F210 07-30-14 non-marked cub. Alive on 07-30-14. M221 168 08-13-13 01-27-14 to 18 Radio-collared. Offspring of F197; sibling of one other non- F197 02-13-14 marked cub. Shed expandable cub collar by 03-04-14 at bull elk cache. Last live signal 02-13-14. M223 244 June 02-05-14 to 425 Radio-collared. Radio-monitored in association with F28 at F28 2013 07-30-14 recapture 02-11-14. Alive on 07-30-14. M225 183 Sep. 03-05-14 to 333 Radio-collared. Offspring of non-marked adult female. Unm.F 2013 07-30-14 Possibly one sibling in association at capture 03-05-14. Alive on 07-30-14. P1076 30-31 08-05-13 30-31 30-31 Found dead and mostly consumed in the nursery. Cause F181 probably infanticide or predation. PF1094 214 July 02-04-13 1 Died; mauled by our capture dogs. Offspring of F176. Born F176 2013 in July 2014. Had 2 siblings; 3 cubs total. Other 2 cubs not captured and marked. a Cub F35 probably starved between 06-30-06 & 07-07-06 after the transmitter on the expandable collar got in its mouth. b Cub M42 died after being captured by dogs, probably from stress of capture associated with severe infection of laceration under right foreleg caused by expandable radiocollar. c Cub M56 was captured in association with F7 and her cubs M43 and M44. He may have been missed at the nursery when M43 and M44 were initially sampled and marked. d Cub M60 died probably of starvation. The expandable radiocollar was around the neck and right shoulder, probably restricted movement. 71 � https://cpw.cvlcollections.org/files/original/4fc5497ee0d163186822c8afe502ebe8.pdf f8715034a7046db12521fe8b782b314e https://cpw.cvlcollections.org/files/original/4900dd59efbf86b63b17352444f5ce31.pdf e007376ef770594d046293b4ebaf3730 PDF Text Text Colorado Parks and Wildlife July 1, 2014  June 30, 2015 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 3003 3 Federal Aid Project No. W-204-R4 : : : : : Parks and Wildlife Mammals Research Carnivore Conservation Assessing Effects of Hunting on a Puma Population on the Uncompahgre Plateau, Colorado Period Covered: July 1, 2014  June 30, 2015 Author: K. A. Logan Personnel: J. Runge, C. Anderson, Colorado Parks and Wildlife All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the authors. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Colorado Parks and Wildlife (CPW) conducted a 10-year (2004-2014) puma study on the Uncompahgre Plateau to quantify puma population dynamics in the absence (reference period, years 1-5) and presence (treatment period, years 6-10) of sport-hunting. The purpose of the study was to evaluate the assumptions underlying the CPW puma management program with sport-hunting in Colorado. The reference period began December 2004 and ended October 2009 and the treatment period began November 2009 and ended all data collection in December 2014. 109 pumas were captured and marked in the reference period and 115 pumas were captured and marked in the treatment period. Those animals produced known-fate data for 75 adults, 75 subadults, and 118 cubs. This report summarizes results of early preliminary stages of data analyses. In the absence of sport-hunting, the puma population increased and exhibited high survival rates and a high fecundity rate. There was a clear treatment effect with sporthunting. The puma population declined substantially after 3 hunting seasons at a 15% design harvest of independent pumas. An 11-12% design harvest of independent pumas was applied in the final 2 years of the treatment period. The treatment period puma population also exhibited substantially lower survival rates of adults, subadults, and cubs, and a lower fecundity rate. Data analyses are ongoing and will ultimately inform future puma management in Colorado. 1 �WILDLIFE RESEARCH REPORT ASSESSING EFFECTS OF HUNTING ON A PUMA POPULATION ON THE UNCOMPAHGRE PLATEAU, COLORADO Kenneth A. Logan PROJECT NARRATIVE OBJECTIVES 1. Gather data on puma population abundance, sex and age structure, vital rates (i.e., reproduction, age-stage survival rates, and emigration and immigration rates if possible), and agent-specific mortality in a non-hunted puma population phase and a hunted puma population phase for use in modeling puma population dynamics and evaluating and structuring puma harvest management and research approaches. 2. Test current CPW puma harvest-related assumptions that are applied to puma populations in DAUs, and arrive at acceptable harvest levels intended to achieve population objectives, including increases, stability, and reductions in the puma population. 3. Apply a hunting treatment to the puma population on the Uncompahgre Plateau study area designed to test CPW harvest-related assumptions and learn about impacts of hunting on pumas. 4. Develop methods that detect changes in puma population abundance on the Uncompahgre Plateau study area that might be useful for monitoring changes in puma abundance in other puma habitats. SEGMENT OBJECTIVES 1. Complete data collection of the fifth and final year of the five-year treatment period by working with CPW biologists and managers to manipulate the puma population with sport-hunting and to survey hunters. 2. Complete gathering data on puma population sex and age structure. 3. Complete gathering data for estimates of puma reproduction rates. 4. Complete gathering data to estimate puma sex and stage-specific survival rates. 5. Complete gathering data on agent-specific mortality. 6. Begin data analysis phase working with CPW Biometrician, Jon Runge. Introduction Colorado Parks and Wildlife (CPW) managers need reliable information on puma population biology to develop sound management strategies that address diverse public values and the CPW objective of actively managing pumas while “achieving healthy, self-sustaining populations” (Colorado Division of Wildlife 2002-2007 Strategic Plan:9). Active management of pumas includes managing for sustained populations to provide sport-hunting opportunity, and reducing puma populations to suppress depredation on livestock, predation on mule deer, and enhance public safety. Thus, sport-hunting is intended as a tool for puma management in addition to recreation. Because sport-hunting is a major cause of death for pumas in hunted populations (Murphy 1983, Logan et al. 1986, Anderson et al. 1992, Ross and Jalkotzy 1992, Lambert et al. 2006, Stoner et al. 2006, Laundré et al. 2007), managers need information to better understand how hunting impacts puma populations and methods to monitor changes in puma abundance to assess how management actions are working to meet management objectives. To improve the biological basis for managing pumas, the CPW began a process in year 2000 to develop puma Data Analysis Unit (DAU) plans (Colorado Division of Wildlife 2007). The DAU plans 2 �involved a formulation or deductive model to project an expected number of pumas on available habitat and the level of sport-harvest deemed acceptable to achieve one of two management objectives for each DAU: 1) to maintain a stable or increasing puma population, or 2) to suppress the puma population. A series of “best judgments” and assumptions by CPW managers on puma populations in DAUs was necessary because reliable and affordable methods for estimating puma population abundance in habitat were not available, and there was no information on impacts of hunting on Colorado puma populations. Consequently, managers that developed DAU plans mostly used data from intensive puma population studies from other western states that were published in the literature and from information from studies of puma in Colorado (Anderson et al. 1992) as a guide. The information included estimates of puma population density, sex and age structure, population rates of increase, and expected impacts of harvest rates, and that information was extrapolated to expected puma habitat across Colorado. Current CPW (CDOW 2007) management assumptions include: 1) Puma densities range from 2.0−4.6 pumas/100 km2. 2) A moderate annual rate of growth of 15% (i.e., for the adult and subadult puma population. 3) For DAUs managed for a stable or increasing puma population, acceptable total mortality could fall in the range of 8 to 15% of the projected huntable population (i.e., adult plus subadult pumas). 4) In addition, for DAUs with a management objective for a stable-to-increasing puma population, acceptable female (i.e., adults and subadults) mortality could fall in the range of 35 to 45% of the total mortality. 5) For DAUs managed to suppress the puma population, total mortality could fall in the range of >15 to 28% of the projected huntable population of adult plus subadult pumas. Prior to the current puma research described in this report, none of these demographic prescriptions had been tested for their validity on a puma population in Colorado. Such testing is prudent because some biological judgments made for DAU management plans might be in error and cause unintended consequences to puma populations, such as cause puma populations to decline where the management objective is for stable or increasing puma populations- the critical component for providing resiliency in the puma population to effects of hunting mortality. Metrics from research in other western states support or are at variance with current CPW puma harvest guidelines for a stable to increasing population. Recent research in Wyoming indicated that a puma population could sustain a harvest comprised of 10 to 15% adult females, and population decline occurred when about 25% of adult females comprised the harvest (Anderson and Lindzey 2005:187). A Utah study found that a puma population declined when harvest exceeded 30% of the adults and subadults and comprised 42% females for 3 years (Choate et al. 2006). Another study in southern Idaho and northern Utah suggested that a harvest that included 15 to 20% of resident females probably would not reduce a puma population (Laundré et al. 2007). More recently, researchers in Washington modeled puma population dynamics and indicated that a 14% harvest of adult pumas was expected to result in a stable population and age structure (Beausoleil et al. 2013). Thus considering any of this information, if adult females comprise the majority of the current acceptable level of female harvest (i.e., 35-45%) in a Colorado DAU, and there is substantial error in the population projection, the puma population could decline. This result is possible because actual puma population estimates are not available for any DAUs, in fact numbers used are at best educated guesses or biological judgments extrapolated over huge nonsurveyed areas, and would be problematic if errors occurred for a substantial number of DAUs where the management objective is for a stable-to-increasing population. Thus, the state-wide strategic objective of managing for a healthy, self-sustaining puma population could be in jeopardy. This emphasizes the need to quantify impacts of puma harvest on population parameters to structure guidelines that will likely achieve population objectives. This current study serves as an empirical test of the more theoretical guidelines that could be derived from the literature (previously cited). To gauge the impact of management prescriptions on puma populations, wildlife managers need reliable, affordable methods to apply to representative DAUs. Already the CPW gathers information 3 �useful for guiding puma management through mandatory puma harvest reports and records on other detected mortality (e.g., road-kill, depredation control). Those data include sex, age-stage, location, and cause of death. In 2007, new efforts to improve the quality of the data included aging harvested pumas by tooth cementum-annuli and assessing population structure of pumas in Colorado using population genetics (J. Apker, Carnivore Management Coordinator, M. Alldredge, Mammals Researcher, personal commun.). Yet, the CPW needs to link those data to puma population dynamics influenced by harvest. To address information needs, the CPW began this research in 2004 on the Uncompahgre Plateau to better understand puma population dynamics and effects of sport-harvest. The study was designed in two 5-year periods: a reference period (years 1 to 5) and a treatment period (years 6 to 10). The reference period provided baseline estimates on puma population abundance, sex and age structure, reproduction, survival, agent-specific mortality, and dynamics in representative puma habitat in Colorado where sporthunting was not a cause of mortality. The treatment period occurred on the same study area and included manipulation of the puma population through the use of sport-hunting to provide information on the impact of hunting on a puma population and to evaluate methods intended to detect changes in the puma population. Study Area The study area for the puma population research is on the Uncompahgre Plateau (in Mesa, Montrose, Ouray, and San Miguel Counties; Fig. 1). The study area includes about 2,253 km2 (870 mi.2) of the southern halves of Game Management Units (GMUs) 61 and 62, and about 155 km2 (60 mi.2) of the northern edge of GMU 70 (between state highway 145 and San Miguel River). The area is bounded by state highway 348 at Delta, 25 Mesa road and Forest Service road FS503 to Nucla, state highway 97 to state highway 141 to state highway 145 to Placerville, state highway 62 to Ridgeway, U.S. highway 550 to Montrose, and U.S. highway 50 to Delta. This area comprised a Game Management Unit (GMU), the basic spatial unit used to manage pumas by CPW, for the purpose of this study on effects of sport-hunting on a puma population. The study area size represented the very largest GMUs for puma management in Colorado, therefore inferences from the study could be interpreted at the GMU population level. The study area was typical of puma habitat in Colorado that has vegetation cover varying from pinion-juniper covered foothills starting from about 1,700 m elevation to the spruce-fir and aspen forests growing to the highest elevations of about 3,000 m. Mule deer (Odocoileus hemionus) and elk (Cervus elaphus) were the most abundant wild ungulates available for puma prey. Cattle and domestic sheep were raised on summer ranges on the study area. People reside year-round along the eastern and western fringe of the area, and there is a growing residential presence especially on the southern end of the plateau. A highly developed road system makes the study area easily accessible for puma research efforts. A detailed description of the Uncompahgre Plateau is in Pojar and Bowden (2004). 4 �Figure 1. The puma study area on the southern half of the Uncompahgre Plateau, Colorado (shaded in gray) comprising the southern portions of Game Management Units (GMUs) 61 and 62 and a northern portion of GMU 70. Expected Results Results of this study inform CPW biologists and managers about expected puma population dynamics and biological impacts of sport-hunting and other causes of mortality (e.g., intra-specific strife, disease, poaching, vehicle collisions, depredation control) on a puma population in Colorado. The study reveal puma life history traits and management effects useful for developing sound management strategies. Moreover, this study evaluated the current puma management structure and assumptions used in puma harvest management through the examination of data gathered directly from a population-level manipulation. This study also assessed one method to detect changes in puma populations on a local intensive scale and for potential application on large representative management areas in collaboration with Mammals Researcher Dr. Mat Alldredge and colleagues. This information should assist the CPW to improve puma management in Colorado. Approach The puma population on the Uncompahgre Plateau study area was studied during a 5-year reference period (i.e., reference period years RY1-RY5) and manipulated during a 5-year treatment period (i.e., treatment period years TY1-TY5). The reference period provided baseline data on puma population dynamics (i.e., abundance, sex and age structure, survival, reproduction, agent-specific mortality, immigration and emigration) without puma sport-hunting as a cause of mortality. The study area was closed to puma hunting from November 2004 to October 2009. In addition, any radio-collared or ear-tagged pumas in the two Game Management Units (GMUs 61 north, 62 north) adjacent to the study area to the north were also protected from sport-hunting. This was an unreplicated case study on one geographic area having a before and after treatment effect design. This effort represented the most pumas ever studied in a population in Colorado and an unprecedented opportunity for CPW to learn about puma population dynamics and effects of sport-hunting. 5 �Field Methods Puma Population Status Puma capture, marking, and sampling: The capture, marking, and GPS- or VHF- collaring of individual pumas and subsequent monitoring was essential to a number of project objectives, including obtaining: population counts, sex and age structure, estimates of vital rates, estimating detection probabilities per individual in camera grids, and movement data to evaluate emigration, vulnerability to hunters, and GMU and DAU boundaries. Pumas were captured year-round using 3 methods: trained dogs, cage traps, and by hand (for small cubs). All captured pumas were examined to ascertain sex and describe physical condition and diagnostic markings. Ages of adult puma were estimated initially by the gum-line recession method (Laundre et al. 2000) and dental characteristics of known-age puma (Logan and Sweanor, unpubl. data). Ages of subadult and cub puma were estimated initially based on dental and physical characteristics of known-age puma (Logan and Sweanor unpubl. data). Ages of nurslings were estimated from apparent birthing dates indicated by GPS- and VHF-location data of collared mothers. Metric scale body measurements were recorded for each puma included: mass (kg), pinna length, hind foot length, plantar pad dimensions, total length and tail length. Tissue collections of adult and subadult pumas included: skin biopsy (from the pinna receiving the 6 mm biopsy punch for the ear-tags) and blood (30 ml from the saphenous or cephalic veins) and hair for genotyping individuals, parentage and relatedness analyses, and disease screening. Only skin and hair samples were collected from cubs  10 weeks old. Universal Transverse Mercator Grid Coordinates on each captured puma were fixed via Global Positioning System (GPS, North American Datum 27). All pumas were handled in accordance with approved Animal Care and Use Committee (ACUC) capture and handling protocols in ACUC file #08-2004 (Appendix I) and ACUC protocol #03-2007 titled, Mountain Lion Capture and Handling Guidelines. Captured and handled adult, subadult, and cub pumas were marked 3 ways: GPS/VHF- or VHFcollar, ear-tag, and tattoo. The identification number tattooed in the pinna was permanent and could not be lost unless the pinna was severed. A colored (bright yellow or orange), numbered rectangular (5 cm x 1.5 cm) ear-tag (Allflex USA, Inc., DFW Airport, TX) was inserted into at least one pinna to facilitate individual identification during direct recaptures and when pumas were harvested. Pumas captured by dogs usually climbed trees to take refuge. Adult and subadult pumas captured for the first time or requiring a change in telemetry collar were immobilized with Telazol (tiletamine hydrochloride/zolazepam hydrochloride) dosed at 5 mg/kg estimated body mass. The drug was delivered into the caudal thigh or shoulder muscles via a Pneu-Dart® shot from a CO2-powered pistol (Pneu-Dart X-Caliber, Pneu-Dart Inc., Williamsburg, PA). A 3m-by-3m square nylon net was deployed beneath the puma to catch it in case it fell. A researcher climbed the tree, fixed a rope to two legs of the puma and lowered the cat to the ground with an attached climbing rope. Once on the ground, the puma’s head was covered, its legs tethered, and vital signs monitored. (Normal signs: pulse ~70―80 bpm, respiration ~20 bpm, capillary refill time ≤2 sec., rectal temperature ~101oF average, range = 95―104oF) (Kreeger 1996). Treed pumas that could not be safely immobilized and handled were shot with a biopsy dart (8 mm long x 3 mm dia., Pneu-Dart Inc., Williamsburg, PA) fired from a CO2-powered pistol to obtain a skin sample from the caudal thigh or shoulder. This sample was used in a study of puma population genetics. Cage traps were used to capture adults, subadults, and large cubs. Pumas were lured into the trap using road-killed or puma-killed ungulates (Bauer et al. 2005, Sweanor et al. 2008). A cage trap was set only if a target puma (i.e., an unmarked puma, or a puma requiring a collar change) scavenged on the lure. Researchers continuously monitored the set cage trap from about 0.5 to 1 km distance by using VHF beacons on the cage and door. This allowed researchers to respond to the captured puma within 30 6 �minutes. Pumas were immobilized with Telazol injected into the caudal thigh or shoulder muscles with a pole or hand syringe. Immobilized pumas were restrained and monitored as described above. Small cubs (≤10 weeks old) were captured using our hands (covered with clean gloves) or with a catch pole. Cubs were restrained inside new burlap bags during the handling process and were not administered drugs. Cubs at nurseries were approached when mothers were away from nurseries as determined by radio-telemetry. Cubs captured at nurseries were removed from the nursery a distance of ~20-100 m to minimize disturbance and human scent at nurseries. Cubs were returned to the exact nurseries immediately after sampling processes were completed (Logan and Sweanor 2001). Adult and subadult pumas were fitted with GPS collars (approximately 400 g each) or VHF collars (approximately 300 g each (Lotek Wireless, Newmarket, Ontario, Canada). Budget constraints limited the number of GPS collars (~10-15) available annually. Therefore, GPS collars were fitted to primarily adult pumas. GPS collars were programmed to fix and store puma locations at 4 times per day to sample daytime, nighttime, and crepuscular locations (i.e., 0:00, 06:00, 12:00, 19:00). This schedule seems optimal for sampling different parts of the day and to extend battery life (~18 months). Other adult and subadult pumas were fitted with VHF collars. Our efforts were to locate all collared pumas once per week from fixed wing aircraft and as weather and scheduling conditions allowed, for data on survival, agent-specific mortality, and location. We checked the live signal/mortality signal status from collared pumas from the ground opportunistically when we operated within their home ranges. VHF and GPS collars had mortality modes set to alert researchers when puma were immobile for 3 hours (VHF collars) to 24 hours (GPS collars) so that dead pumas could be found for data on survival and agent-specific mortality. Because subadult male pumas were not fully grown, they also received leather expansion links in their collars. The expansion links added 10-12 cm when open to allow the collars to be worn safely into the adult stage. We attempted to collar all cubs in each observed litter with a small VHF transmitter mounted on an expandable collar (62 g, model 080, Telonics Inc., Mesa, AZ) when cubs weighed 1.3―10 kg. The collars were designed to operate for 10−12 months, and expanded to 54 cm circumference to accommodate growth. Cubs with mass ≥7 kg were fitted with a larger expandable collar (90 g, model 210, Telonics Inc., Mesa, AZ). The collars were designed to operate for 12−18 months and could expand to 54 cm circumference to accommodate growth. Cubs approaching the age of independence (~11−14 mo. old) were fitted with Lotek LMRT-3 VHF collars (~400 g) with leather expansion links that add 10−14 cm to the collar circumference to accommodate the adult puma neck size. These collars operated for 2−3 years. Cubs were recaptured when possible to replace collars as necessary. Monitoring collared cubs allowed quantification of survival rates and agent-specific mortality rates. Radio-collared offspring also provided information on movements, age of independence, recruitment, and emigration. Puma population sampling considerations: The puma is one of the most difficult large mammals to study in North America because of its relatively low abundance on the landscape and its highly cryptic behavior. These characteristics were expected to influence the ability to sample individuals in the study population. The most efficient technique for locating and capturing pumas is detecting their tracks in snow and using trained dogs to pursue and secure them for sampling purposes. Hunters use the same technique to harvest pumas, which creates potential for biased survival rate estimates if researchers and hunters use similar strategies to detect and capture pumas. That is, with similar sampling strategies, pumas that are most vulnerable to being captured and radio-collared might also be more vulnerable to harvest, resulting in survival rates that are biased low. Hunters’ detection of puma tracks is heavily influenced by road access. To minimize bias potential, we attempted to intensively search the entire study area for puma tracks, irrespective of road characteristics, thereby equally detecting puma with both higher and lower hunter-detection probabilities. Thus, our approach was to apply roughly equal (i.e., intensive, uniform) searching intensity across the study area and apply an alternative capture technique with bait and 7 �cage traps that did not rely on track detection to capture pumas, and attempt to directly monitor via VHF telemetry a large proportion of the population in the study area in order to reduce heterogeneity in sampling individuals. Capture efforts to sample the adult and subadult pumas (i.e., independent pumas) subject to sporthunting mortality in the study area population was conducted mainly during winter when snow cover maximized the detection and capture probability of pumas. Snow provided a continuous or almost continuous substrate that registered tracks of terrestrial mammals. Puma tracks were highly distinctive and at ground level could be accurately and consistently visually identified and distinguished from tracks of all other mammals by trained personnel in a variety of snow and weather conditions and in the variety of terrains and vegetation communities. This characteristic was the reason why most intensive puma population studies in the West have been conducted during winter- to maximize detection, quantification, classification and monitoring of animals in the populations (Hornocker 1970, Logan et al. 1986, Lindzey et al. 1992, Ross and Jalkotzy 1992, Spreadbury 1996, Anderson and Lindzey 2005, Lambert et al. 2006, Laundre et al. 2007, Cooley et al. 2008). Puma population research in winter also more directly linked the puma population investigated with animals killed during the hunting season, which in Colorado occurred annually during mid-November through March to facilitate the detection of pumas by hunters which mainly use trained dogs to capture the pumas. Snow also maximized the ability of trained dogs to follow scent in tracks and capture the pumas. In addition, during spring and fall and opportunistically in winter, we attempted to capture pumas in cage traps where pumas were attracted to road-killed deer baits, and puma prey kills. Individuals caught in cage traps were available to move about the study area during winter and be exposed to hunters. The Uncompahgre Plateau study area was highly roaded, and from those roads branched ATV trails that further facilitated thorough searches of the study area to detect pumas. Still, the road system was not uniform, with some areas densely roaded, others moderately roaded, and one area in particular that did not allow motorized vehicles. The area is the combined Camelback Wilderness Study Area (BLM portion) and Roubideau Special Management Area (U.S. Forest Service portion) in the main fork of Roubideau Canyon. That non-roaded area was about 109 km2 (42 mi.2). Yet a system of roads and trails we used surrounded this area. We routinely handled this area by hiking up the lower reaches of Roubideau Canyon and onto upper benches and canyons to search for puma tracks. A puma capture team, involving 4 people on separate search routes, was detailed to search this region on the surrounding roads, ATV/snowmobile trails, and hiking paths. By visiting this area repeatedly each winter we expected to detect some pumas that used the canyon and that might not have been detected in the canyon in other search days. Pumas were expected to move out of the non-roaded portion of the canyon periodically during the winter and be exposed by their movements. Thus, periodic searches of any of the search routes was expected to increase exposure of the pumas to detection. The study area was partitioned into search areas that a capture team could search within 1-2 days to detect puma tracks on snow within each area (Table 1). The intent was to structure a thorough, relatively uniform, systematic search effort across the study area and to repeat it multiple times during winter and spring. To cover the areas efficiently, we used four-wheel-drive trucks, all-terrain vehicles, snow-mobiles, and walking. When puma tracks ≤1 day old were detected, trained dogs were released to pursue the puma to capture, sample, and mark it. When puma tracks 1-2 days old were detected, we searched in the direction of travel of the puma in an effort to find ≤ 1 day old tracks that would facilitate pursuit of the puma. This sometimes lengthened our search within any particular area by another 1-2 days. When a GPS/VHF-collared puma was detected with radiotelemetry within 1 km (usually < 0.5 km) of the tracks and the direction of the tracks indicate that the puma was likely the collared individual, then we directed our efforts away from those tracks to focus our efforts on non-collared (i.e., non-sampled) pumas in the population to use our time more efficiently. 8 �Table 1. Puma search areas on the Uncompahgre Plateau Study area. West Slope East Slope 25 Mesa Road to Cottonwood Creek and San 25 Mile Mesa Road to east rim of Roubideau Miguel Canyon (west reach) Canyon and Ben Lowe Mesa ---------------------------------------------------------------------------------------------------+-------------------------------Cottonwood Creek to Horsefly Canyon Roubideau Canyon to Transfer Road ---------------------------------------------------------------------------------------------------+-------------------------------San Miguel Canyon (mid reach) to Maverick Draw Transfer Road to east rim of Dry Creek Basin ---------------------------------------------------------------------------------------------------+-------------------------------Horsefly Canyon and San Miguel Canyon (mid East rim of Dry Creek Basin to east rim of Spring reach) to Clay Creek Canyon ---------------------------------------------------------------------------------------------------+-------------------------------Clay Creek and San Miguel Canyon (upper reach) Spring Canyon to Happy Canyon to McKenzie Creek ---------------------------------------------------------------------------------------------------+-------------------------------McKenzie Creek and San Miguel Canyon (upper Happy Canyon to Horsefly Canyon reach) to Leopard Creek ---------------------------------------------------------------------------------------------------+-------------------------------Horsefly Canyon to McKenzie Butte ---------------------------------------------------------------------------------------------------+-------------------------------McKenzie Butte to Loghill Mesa Reliability of population count methods: The approach described previously was expected to enable us to monitor a large proportion of the independent puma population on the study area in winter for reliable counts. On this point, we wanted direct evidence on the reliability of our field methods to study the puma population and make our winter counts. An opportunity for a one-time independent evaluation on the proportion of independent pumas on the study area that we might have marked was provided by an independent camera grid study conducted on our study area by Master of Science graduate student Kirsti Yeager (Colorado State Univ., Dep. of Fish, Wildlife, and Conservation Biology) and advisors Dr. William Kendall (Colorado Cooperative Fish and Wildlife Research Unit, Colorado State University) and Dr. Mat Alldredge (Mammals Researcher, CPW). This collaboration was part of a larger CPW supported project on the Colorado Front Range that evaluated the use of a grid with cameras, predator call boxes and DNA collection methods as a noninvasive method to collect puma tissue for capture-recapture models for estimates of puma abundance (Yeager et al. 2013). A grid of 2 km x 2 km (4 sq. km) cells covering 540 km2 was established on the east slope of the Uncompahgre Plateau study area. Eighteen cells were identified randomly for each of 3 survey periods each lasting about 28 days. Therefore, a total of 54 random cells were surveyed during December 2012 to March 2013. This period was in treatment year 4 (TY4). Within each random cell K. Yeager subjectively chose the “best” site to attract pumas by using vocal baits each consisting of a Fur-Finder ® (Magna, UT) electronic predator call of a distressed deer fawn. Each site also had a Reconyx ® PC900 Hyperfire camera (Holmen, WI) to record animal activity and hair-sampling devices (i.e., barbed-wire strands, sticky rollers) to attempt to acquire hair. This effort evaluated these methods for a non-invasive survey of puma abundance by using tissue to genetically identify individuals in a capture-recapture structure (Yeager et al. in development). This also allowed us to evaluate our field methods and proportion of independent pumas that were marked in the population. Population Manipulation: The puma population on the Uncompahgre Plateau Study Area was manipulated by sport hunting after the 5-year reference period with no hunting. The hunting season was from mid-November and extended to January 31, unless the last puma on the design quota was killed before January 31, which effectively closed the season on the study area. The initial harvest quota was 8 pumas which represented a 15% harvest of the expected number of independent pumas in treatment year 1 (TY1). The predicted effect was that the 15% harvest of independent pumas would result in a stable or increasing population, an expectation that mangers used to guide puma harvest rates in Colorado. The quota of 8 was based on the projected number of 52 independent pumas expected on the study area in winter 2009-10 (TY1), modeled from count data in winter 2007-08 (RY4) (see Appendix II, Table AI.3). After it was evident that the number of independent pumas had declined during TY1 to TY3, we adjusted 9 �the harvest quota down to 5 pumas to represent an 11% harvest of the projected 45 independent pumas expected in TY4 in an effort to find a sustainable harvest rate useful to managers. The harvest quota of 5 was continued in TY5. The number of hunters on the study area each winter was not limited. Each hunter on the study area was required to obtain a hunting permit from the CPW Montrose Service Center. Permits were free and unlimited. Each permit allowed the individual hunter with a legal puma hunting license in Colorado to hunt in the puma study area for 14 days from the issue date. Unsuccessful hunters that wanted to continue hunting past the permit expiration date could get a new permit for another 14 days or until the hunting season on the study area closed due to the quota being reached or the end of the hunting season. This permit system enabled CPW to estimate the number of hunters that actually hunted on the study area each season. In addition, a voluntary survey questionnaire (see Appendix III) was attached to each puma hunting permit issued to each hunter with a stamped envelope addressed to the CPW principal investigator. Hunters were asked to complete the survey as soon as possible for each hunting period associated with the permit in an effort to have hunters report the information while it was still fresh in their minds. All pumas harvested on the study area were visually examined and sealed by the principal investigator as mandated by CPW. Hunters reported their puma kill to CPW within 48 hours of harvest and presented the puma carcass for inspection within 5 days of harvest. At the time of carcass check-in a mandatory CPW harvest check form was completed. In addition, an upper premolar tooth (i.e., PM2) was extracted for aging by the cementum annuli method and a tissue sample was collected. Each successful hunter was asked to fill-out the one-page hunter survey form. All other hunters that did not report a puma kill on the study area were contacted by phone or in person and asked to complete the survey form as well. Many hunters returned the surveys on their own volition. Mandatory hunter harvest checks provided accurate data on pumas removed from the study population for estimates of survival and agent-specific mortality. Hunters also provided data to evaluate the relative vulnerability of pumas to harvest and potential for hunter selectivity. Hunter harvest and capture events also revealed availability and sex and age classes of unmarked pumas on the study area during the hunting season and before our capture teams operated after the season to quantify the population. After the design quota was filled and the study area closed to hunting, two puma research teams activated for capture operations with trained dogs and cage traps. One team operated on the east slope and one operated on the west slope of the study area to systematically and thoroughly search the study area to capture, sample, and GPS/VHF radiocollar pumas the remainder of winter and early spring. Population Monitoring: This monitoring plan enabled us to estimate the winter puma population abundance and sex and age structure each hunting season. Researchers monitored other population parameters, including: reproduction, survival and agent-specific mortality year-round. Movements of VHF- and GPS-collared pumas were also monitored. GPS- and VHF-collared pumas were fixed about once per week from light fixed-wing aircraft (e.g., Cessna 185) fitted with radio signal receiving equipment (Logan and Sweanor 2001). This monitoring enabled researchers to find GPS-collared pumas to acquire remote GPS location reports, monitor the status (i.e., live or dead) of individual pumas, and to locate carcasses for necropsy. Status of GPS- and VHF-collared pumas were monitored from the ground opportunistically using handheld yagi antenna. GPS-collared pumas were monitored for survival status daily by using GPS-data that attempt location fixes 4 times daily (00:00, 06:00, 12:00, and 19:00). Cessation of activity (i.e., due to death) around those times allow a more accurate identification of time of death. Puma births occurred from March through September, but potentially they could happen any month of the year (Anderson et al. 10 �1992, Laundre and Hernandez 2007, Logan 2008). Researchers estimated reproduction data on birth interval, litter size, sex ratio, survival, agent-specific mortality, and recruitment to later life stages (i.e., subadult, adult). Emigration was revealed with a sample of radio-collared or ear-tagged marked offspring that left the study area. Analytical methods: The population of interest to managers was the independent pumas (i.e., adults and subadults) in winter, which coincided with the puma hunting season in Colorado when snow cover maximized the vulnerability of pumas to hunting. As indicated previously, our winter research was designed to be thorough to search the study area to maximize the number of marked pumas while trying to reduce heterogeneity in sampling for population parameter estimation. This along with harvest statistics we attempted to obtain a complete count with attendant sex and age structure of the population that represented the Colorado hunting season from November through March. The counts consisted of the sum total of all known marked (i.e., radio-collared and ear-tagged) pumas on the study area, and non-marked harvested pumas plus any other pumas detected on the study area whose movements did not match movements of collared pumas and exhibited diagnostic characteristics of unique individuals (e.g., tracks distinguishing sex from hind-foot plantar pad measurements, counts of cub tracks with female tracks). In addition, we wanted to maximize the number of radio-collared pumas in all sex and age classes and to obtain a uniform distribution of those animals across the study area to represent animals in the population exposed to all causes of mortality. Puma Population Trends Puma population change was quantified by winter population counts. Data on agent-specific mortality, survival and reproduction rates were used to evaluate changes in population associated with the reference and treatment periods. Puma Survival and Mortality Analysis Adult puma survival and mortality was examined from data on radio-collared pumas that provided known-fate data (i.e., monitoring dates, estimated dates of death, cause of death). We used program MARK (White and Burnham 1999) (accessed January 12, 2015), the known fates data type and the logit link function to model survival rates with a candidate set of models structured to investigate factors that might explain variation in survival. MARK estimated survival rates, standard errors, and 95% confidence intervals for each model. Our main interest was the effect of the hunting treatment as partitioned among the reference and treatment periods on survival, because our research focus was to examine effects of sport-hunting on a puma population. Radio-location records for each adult puma were converted to monthly encounter histories. MARK estimated monthly survival rates using the modified Kaplan and Meier (1958) estimator that allowed staggered entry based on when we collared individuals and censoring of individuals if we lost contact with them (Pollock et al. 1989). We used data from year 2 of the reference (RY2) period to year 5 of the treatment period (TY5) (i.e., a 9 year span). We did not use data from reference year 1 (RY1) because we had just started the study and had collared only 7 adult pumas (3 males and 5 females). Encounter histories of individual adult pumas started on the day of capture, because no pumas died as a result of capture, or the beginning of RY2 (November 1, 2005). We censored individuals in the data if we did not receive its signal after the month of its last location. Individuals re-entered the data set if we recaptured them and fit them with a new collar. Death dates for puma were assigned to pumas with GPS collars based on the first day that GPS locations indicated that the pumas were immobile. Death dates for VHF-collared pumas were estimated based on previous live signal data and the mid-point of the span of days the puma was estimated to have died based on carcass decomposition. Causes of death were categorized to known human causes (e.g., harvest, depredation control, vehicle strike, poached), to known natural causes (e.g., intraspecific strife, injury), or to unknown natural causes. 11 �Subadults puma survival and mortality was estimated for all known radio-collared and ear-tagged and tattooed pumas with known fates that spanned a 12-month subadult stage from 13 months up to 24 months of age. We did not know with certainty when all of the pumas in this 12-month age stage became independent, therefore some of the pumas may have been dependent for a period of time. Encounter histories for the pumas started as marked pumas entered the age stage. Histories started on the first day of capture for subadults caught and marked for the first time, because no subadults died as a result of capture. All histories were converted to monthly encounter histories. Death dates were assigned to harvest, depredation control, and observed vehicle strike dates. For other VHF-collared pumas where dates were not observed, dates were estimated as the mid-point of the span of days the puma was estimated to have died based on previous live signal data and carcass decomposition. The encounter histories were treated as known-fate data and entered into program MARK to model subadult puma survival rates using a candidate set of models that might explain the variation in survival rates. Examining survival rates of adults and subadults, the legal harvest-age pumas, in the reference and treatment periods with contrasting models with and without the hunting treatment allowed us to assess changes in survival associated with the treatment effect. A treatment effect supported an inference that sport-hunting mortality was an important factor explaining the variation in puma survival and a factor that was largely additive if survival declined in association with the treatment. However, if models lacking the treatment effect received the most support, this would indicate that hunting mortality was primarily compensatory or that statistical power was insufficient to detect a treatment effect. If population growth of independent pumas also declined in association with treatment effect on survival, this change would further support that hunting-caused mortality was mostly an additive factor. Cub survival and mortality was estimated for all radio-collared pumas 1 month to 12 months of age representing a stage when the pumas were dependent on their mothers. The large majority of the cubs in this data set were initially radio-collared as nurslings 1-2 months old. But we also included cubs collared at older ages, because we entered data so MARK would estimate monthly survival rates. In this way, use of data on the older cubs only added to the sample of older cubs and did not bias estimates because older cubs have a tendency to exhibit higher survival (Logan and Sweanor 2001, Ruth et al. 2011). Encounter histories for the cubs started on the first day they were collared. Three nursling cubs that died as a result of malfunctions of the design of the expandable radiocollars early in the study in the reference period were removed from the analysis. After the collars were modified, no other cub mortalities from the collars occurred. Causes of cub deaths were assigned after dead cubs were examined directly. Dates of death were estimated as the mid-point of the span of days the puma was estimated to have died based on dates of previous live signal data and carcass decomposition. The assumption that each radio-collared cub was an independent random sample (i.e., distribution of mortalities among litters is random) may be violated because multiple cubs were often collared in litters and the fates of cubs within litters may be linked. For example, sometimes more than 1 or all cubs in a litter may die from the same proximate cause (e.g., infanticide by a male puma) or the survival of surviving cubs in a litter may be linked to death of siblings (e.g., resulting from greater individual maternal care ). Violation of the independence assumption can result in unbiased survival point estimates, however, sample variances are expected to be underestimated, (i.e., overdispersion, Bishop et al. 2008). Therefore, we will examine validity of the independence assumption in data by estimating an over dispersion parameter, c-hat (Cooch and White 2015). Model selection and parameter estimates The candidate models were considered in importance in an information-theoretic approach (Burnham and Anderson 1998) using Akaike’s Information Criterion adjusted for small sample sizes (AICc) to rank the models. We considered the models with the most support as those with the lowest AICc scores, high AICc weights (wi), and models with ∆AICc ≤2 as having similar support (Burnham and 12 �Anderson 2002). Survival estimates reported here were estimates in the top model and other supported models. Average monthly survival rates for adults were converted to annual survival rates (i.e., Saveragemonthly12), standard errors, and 95% confidence intervals. Stage survival parameters for subadults and cubs were derived estimates calculated in MARK from monthly survival rates that produced average stage (i.e., 12-month) survival rates, standard errors, and 95% confidence intervals. Reproduction Female pumas with GPS/VHF collars were monitored the year round. Data from those pumas provided information on fecundity (i.e., proportion of adult females giving birth each year), litter size, secondary sex ratio (i.e., sex ratio of cubs born), birth intervals, and age at first breeding. Reproduction was verified by direct observations of cubs in nurseries and in direct association of adult females during capture efforts. Fecundity, defined as the proportion of adult female pumas giving birth each year, was estimated annually from reference year 2 (RY2) to through treatment year 5 (TY5) when we had ≥12 adult females in annual samples (there were only 4 adult females in RY1). Data on each adult female each year was coded with the individual identification number and as producing a litter of cubs (1) or not producing a litter (0) and whether the individual female produced a litter each year in the reference period (1) or the treatment period (2). Because adult females comprising the samples within each year were not independent of other years (i.e., some of the same females were monitored in a series of years within and among periods) mean period fecundity rates were modeled by using the generalized linear mixed model procedure (PROC GLIMMIX) in SAS (Version 9.3, 2010, SAS Institute) where the period was the fixed effect and individual puma identification was the random effect. We used the binomial distribution and logit link. We also investigated all adult females that exhibited extremely constrained GPS and VHF location clusters or movements that might indicate the birth of a litter for data on numbers and gender of cubs. When the cubs were 25 to 45 days old we entered the nurseries when the mothers were absent to examine the cubs and to mark them (previously described). We coded the data with each adult female identification number, the period in which the litter was produced (reference=1, treatment=2) and the number of cubs observed in each litter (1, 2, 3, 4). Similarly, because adult females comprising the samples within each year were not independent of other years and some occurred in both periods, we modeled period mean litter size using the mixed linear model procedure (PROC MIXED) in SAS, where period was the fixed effect and individual puma identification was the random effect. The sex ratio of cubs produced in the reference and treatment periods was compared to an expected 1:1 sex ratio by using the Goodness of fit Chi-square procedure in Zar (1984). PRELIMINARY FINDINGS Puma Capture From December 2, 2004 to October 30, 2014 we captured ~256 individual pumas a total of 440 times on the Uncompahgre Plateau study area. None of the adult or subadult pumas died from capture procedures. However, 3 cubs died as a result of premature expansions of the radiocollars (indicated previously in Field Methods) and 1 cub was killed by our tracking dogs. We individually marked 226 pumas: 109 in the reference period and 115 in the treatment period. The number of radio-collared pumas monitored each year ranged from 16 to 56 and averaged 40. Marked pumas provided known-fate data on 75 adults, 75 subadults, and 118 cubs. About 30 individuals were captured with dogs, but were not handled due to dangerous positions in trees. Of those pumas not handled, 11 were captured in the reference period and 19 were captured in the treatment period. Six of 11 pumas not handled in the reference period were associated with marked family members (i.e., mothers, siblings, cubs). Likewise, 8 of 19 pumas not handled in the treatment period were associated with marked family members. Reliability of Population Count Methods The camera grid survey by K. Yeager in treatment year 4 (TY4) spanned 102 days from December 2012 to March 2013. The survey time overlapped 3 months of our capture efforts on the study 13 �area from January 1, 2013 to April 18, 2013. Eleven GPS and VHF collared pumas were known to use the survey grid for varying amounts of time, including 7 adult females, 1 subadult female, 2 adult males, and 1 subadult male. During the survey 18 photographs of pumas visiting the sites were acquired, and all 18 of the photographs depicted GPS or VHF collared pumas. The photographed pumas included 1 subadult female (captured 1/15/2013) and 1 subadult male (captured 1/1/2013) that we captured and marked during the camera survey for the first time before cameras subsequently detected them 5 and 3 times each, respectively. In addition, we captured and marked 1 adult male (2/14/2013) for the first time during the camera survey that was not detected by the cameras. Of the 11 collared pumas known to use the grid, 7 were photographed 1 to 5 times each, including 5 adult females, 1 subadult female, and 1 subadult male. Probability of detecting the 11 collared pumas available during the entire survey period was 0.64 (p= 7/11). Because no non-collared pumas were photographed and we detected, captured and marked 3 new pumas before the cameras detected them during the survey, the data indicated that our field methods produced reliable winter population counts. Puma Population Counts The number of days we spent each winter and early spring searching for pumas with dogs in each period was similar (reference mean=77.2, SD=4.0, range 71-82; treatment mean=79, SD=4.8, range 7486). We believe we had a thorough knowledge of the study area and search routes for reliable counts of the winter population of independent pumas on the study area by reference year 4 (RY4) and throughout the treatment period. However, in RY5 a state-mandated hiring freeze (in response to economic recession) resulted in insufficient personnel for thorough searches of the study area for a reliable winter count of independent pumas. Therefore, the count in RY5 is biased low, but is still larger than RY4 (Table 2, Fig. 2). Puma Population Trends The population of independent pumas increased during the reference period without hunting as a mortality factor and the population declined substantially during the treatment period when hunting was restored (Table 2, Fig. 2). The increasing population during the reference period was the first indication that hunting mortality might have a population affect. The highest number of independent pumas was counted in winter of treatment year 1 (TY1) which was preceded by 5 previous years without hunting. The hunting treatment during TY1 to TY3 consisted of a designed 15% harvest rate on the independent pumas with an expectation that the puma population would remain stable or increase. The quota to represent a 15% harvest was 8 pumas based on a model that projected 53 independent pumas expected in TY1 (Appendix II). However, the puma population declined, therefore, a 15% design harvest (actual harvest averaged 16.1% TY1-TY3, Table 3) of independent pumas was not supported for managing toward a stable or increasing population. Because the population declined with a 16.1% actual harvest rate, we wanted to find a harvest that might be sustainable. Therefore, in TY4 and TY5 the quota was reduced to 5 pumas constituting 11-12% design harvest of independent pumas. The population reached a minimum of 42 independent pumas in TY4, a 25% decline (Fig. 3), and was affected mainly by hunting mortality from TY1 to TY3. The population increased slightly in TY5 and was associated with the lower design harvest rate. During the hunting treatment the number of adult pumas declined to the lowest number in TY5, a 34% decline (Fig. 3). Puma Mortality The regulations implemented for eliminating sport-hunting as a mortality factor in pumas on the study area in the reference period were effective. Of the 32 (21 females, 11 males at risk) adult radiocollared pumas we monitored, 7 adult pumas died; but none from hunting (Table 4). Causes of death were attributed to: 5 natural causes (4 intraspecific strife, 1 unknown), 1 vehicle strike, and 1 depredation control. Of the 22 subadults (8 females, 14 males at risk) providing known-fate data in the reference period, 3 died. One male was killed by a hunter after he dispersed from the study area. The other causes of death in subadults were 1 natural cause (trampled by elk) and 1 vehicle strike. Of 55 radio-collared 14 �cubs (28 females, 27 males at risk) monitored in the reference period, 16 died. Causes included: 13 infanticide, 1 predation, 1 natural, and 1 vehicle strike. In the reference period natural causes dominated deaths of adults and cubs, but of the 3 subadult deaths 2 were from human-causes. In the treatment period a total of 35 pumas were killed by hunters on the study area (Table 3), including: 8 adult females (22.8%), 16 adult males (45.7%), 3 subadult females 8.6%, and 8 subadult males (22.9%). This harvest structure was associated with a declining puma population. The ratio of marked to non-marked pumas killed by hunters on the study area before we started our winter capture operations during the treatment period was 19:16. Moreover, another 12 radio-collared independent pumas that ranged on the study area were killed by hunters when those pumas moved onto adjacent GMUs open to puma hunting, including: 4 adult females, 7 adult males, and 1 subadult female. Sport-hunting in the treatment period changed mortality for independent pumas (Table 4). Of the 61 adults we monitored (39 females, 22 males at risk), 37 died. Hunting caused 21 adult deaths (14 males, 7 females). Other adult deaths were attributed to: 10 natural (7 unknown probably disease related, 3 strife), 3 vehicle strike, 2 depredation control, 1 illegal kill. Of the 53 subadults (19 females, 34 males at risk) providing known-fate data in the treatment period, 20 died. Eleven were killed by hunters. Other deaths in subadults were: 3 strife, 2 other natural, 1 vehicle strike, and 3 depredation control. Of the 63 radio-collared cubs (27 females, 36 males at risk), 27 died. Mortality causes in the cubs included: 8 infanticide, 4 other natural, 2 vehicle strike, 3 depredation control, and 9 starvation. The 9 cubs starved after the deaths of 5 mothers due to: hunting (2 mothers involving 3 cubs), depredation control (1 mother with 3 cubs), and natural causes (2 mothers involving 3 cubs). Human caused mortality, particularly from hunting, dominated adult and subadult puma deaths in the treatment period. Natural mortality comprised the majority of cubs deaths (15/27*100=55.6%). But, human-caused cub deaths in the treatment period increased to 44.4% (12/27*100=44.4%) from 6.2% in the reference period. In addition to these deaths revealed by the radio-collared cubs, we observed deaths of 4 entire litters on the day we entered nurseries to examine cubs for the first time. These cubs were not part of the radio-collared cub population used to model or estimate cub survival (see below in Puma Survival, Cubs). One litter of 3 nursling cubs starved to death in the reference period after the mother was killed for depredation control. In the treatment period, we observed that three entire litters died: one litter with 2 cubs and one litter with ≥1 cubs died of infanticide. A third litter with ≥1 cub died due to black bear predation. Puma Survival Adults Our adult survival sample included 75 radio-collared individuals, with 32 monitored in the reference period and 61 monitored in the treatment period. The most parsimonious survival model included gender interacting with period (i.e., reference, treatment periods, indicating a treatment effect) as factors that best explained variation in adult puma survival rates (Table 5). The evidence ratio using AICc weights (wi) indicated very strong support for the top model with 10.10 times the support of the secondranked model with gender in an additive effect with period. Moreover, >4 ∆AICc separated the top model from the second-ranked model. The remainder of the models in the 8 model candidate set had weak to no support. Clearly, hunting-caused mortality negatively affected adult male and female puma survival, and was particularly strong on adult males. Adult male survival declined from 0.96 in the reference period to 0.40 in the treatment period, and adult female survival declined from 0.86 to 0.74 in those respective periods (Table 6). 15 �Subadults Our subadult survival sample included 75 individuals with known-fates: 22 in the reference period and 53 in the treatment period. For subadult pumas the modeling results indicated period as an important factor influencing survival as indicated by the two top-ranked models ≤2 ∆AICc points (Table 7), which together accounted for 0.77 of the model weights (wi). Evidence ratios using AICc weights (wi) indicated the top-ranked model with interaction of gender with period had weak support for the best model with 1.7 times the support of the second-ranked model with period (i.e., genders combined). Subadult males were strongly negatively affected by hunting-caused mortality, similar to adult males. Subadult male survival declined from 0.92 in the reference period to 0.43 in the treatment period. However, subadult female survival indicated no substantial change from 0.63 to 0.70 with overlapping 95% CIs in those respective periods (Table 7). The second-ranked model Speriod (with genders combined) had reasonable support as the best model (i.e., 1.0567 ∆AICc, wi=0.28631) for explaining variation in subadult survival which declined from 0.84 (95% CI 0.60, 0.95) in the reference period to 0.52 (95% CI 0.37, 0.66) in the treatment period (Table 8). Independent pumas Hunting-caused mortality in the treatment period was the single-most important cause and it was strongly additive. Had hunting mortality been strongly compensatory, the population of independent pumas was expected to be relatively stable or increase (consistent with the reference period population trend), meaning hunting mortality would have compensated for other causes of mortality. Adequate immigration would have also compensated for some mortality. But, this was not the case. The number of independent pumas increased in the reference period without hunting mortality and it declined substantially in the treatment period in association with hunting as the major cause. Moreover, other independent pumas that ranged on and off the study area were killed by hunters and other causes of mortality continued to materialize in the treatment period, including natural and other human causes, all of which contributed to population decline. Cubs Our cub survival sample included 118 cubs: 55 cubs from 32 litters in the reference period, 63 cubs from 45 litters in the treatment period. Modeling results indicated that hunting treatment as a factor explaining puma cub survival variation was less conclusive. This age stage was not expected to be directly affected by hunting mortality because cubs were not legal game. The two top-ranked models indicated period as a factor that influenced cub survival (Table 9) and together accounted for 0.54 of the model weights (wi). Evidence ratios using AICc weights (wi) indicated the top-ranked model with an interaction of gender with period had weak support as the best model with 1.6 times the support of the second-ranked model with period. Moreover, the top model had 1.7 times the support of the third-ranked continuous survival model with no treatment effect. The top model was separated from the next two models by ≤1.0174 ∆AICc, and with the evidence ratios this indicated that all 3 models had substantial support as best models for explaining the variation in cub survival. The top model with gender interacting with period indicated no substantial change in female cub survival between the reference (0.34) and treatment (0.39) periods (Table 5). But male cub survival declined from 0.71 to 0.30 in those respective periods. The second-ranked model (Speriod) also had substantial support as the best model to explain changes in cub survival (females and males combined) which declined from 0.50 (95% CI 0.34, 0.66) in the reference period to 0.34 (95% CI 0.22, 0.48) in the treatment period (Table 10). The third-ranked model S(.) (i.e., no treatment effect, combined genders) estimated the cub survival rate 0.41 (95% CI 0.31, 0.52) for the entire study duration. Note: - - c-hat (i.e., over-dispersion parameter) has yet to be estimated for cub survival, and adjustments will be made if necessary. 16 �Puma Reproduction Adult female pumas on the Uncompahgre Plateau produced litters of cubs between the months of March to September, spanning the early spring to early fall seasons. Data on 66 birth dates revealed that births increased rapidly in May and June, peaked in July, followed by a slight decline in August and a rapid decline in September. No live births were detected in the months of October, November, December, January, or February (Fig. 4). Assuming a 92 day gestation period (Anderson 1983, Logan and Sweanor 2001, this study), the distribution of birth months indicated that puma breeding activity spanned the months of December to June, with a rapid increase in February and peaking March through May. We estimated gestation for 17 litters by 13 females based on GPS- or VHF- location data of females with prospective sires that produced minimum and maximum estimates. Gestation lengths averaged 90.4min-91.8max days (SDmin=2.6, 95% CImin 89.1, 91.6; SDmax=1.0, 95% CImax 90.7, 92.8). Birth intervals for 18 adult females that produced 33 litters averaged 18.5 months (SD=5.9, 95% CI 16.4, 24.4). We estimated the age of 13 adult females when they produced their first litters based on estimated ages (n=11) or known-ages (n=2) of pumas at previous captures and nipple characteristics (i.e., tiny, pink or white color) and associated reproduction histories. The average age at first litter was 32.2 months (SD=8.4, 95% CI 27.6, 36.8, range=21-48). This meant those females conceived at the average age of 29.2 months (SD=8.4, 95% CI 24.6, 33.8, range=18-45) assuming an average 92 day gestation period. Litter sizes were determined for 26 litters produced by 14 females in the reference period where we were reasonably certain we counted all the cubs in nurseries when the cubs were 26 to 42 days old. Likewise, we determined litter sizes for 21 litters of 16 females in the treatment period for nursling cubs 25 to 45 days old. Average litter sizes for each period estimated using linear mixed models were 2.76 (SE=0.1806, 95% CI 2.41, 3.12) for the reference period and 2.38 (SE=0.1972, 95% CI 1.99, 2.76) for the treatment period. Change in the average litter sizes in the two periods was small and the averages were not significantly different because the 95% confidence intervals on the slope for period included zero. The male:female sex ratio for 72 nursling cubs in the reference period was 41:31 and was not significantly different from an expected 1:1 ratio (χ2=1.388, 1 d.f., 0.10<P<0.25). Similarly, the 27:22 sex ratio for 49 nurslings in the treatment period was not significantly different from an expected 1:1 ratio (χ2=0.51, 1 d.f., 0.25<P<0.50). With all the cubs pooled from both periods, the sex ratio 68:53 was not significantly different from parity (χc2=1.860, 1 d.f., 0.10<P<0.25). Fecundity, defined here as the proportion of adult female pumas giving birth each year, was determined for reference period years 2 to 5 (i.e., RY2-RY5) when we radio-monitored 12 to 13 individual females per year and treatment period years 1 to 5 (i.e., TY1-TY5) when we radio-monitored 15 to 17 individual females per year. Average fecundity per year for each period estimated using generalized linear mixed models were 0.63 (SE=0.068, 95% CI 0.49, 0.75) for the reference period and 0.48 (SE=0.057, 95% CI 0.37, 0.59) for the treatment period. The average fecundity rates for the periods were not statistically different because the 95% confidence interval for the slope included zero. However, a lower average fecundity of 0.48 over 5 years was expected to produce a substantially lower population growth (~6% less per year) than an average fecundity of 0.63 in our deterministic discrete time model (Appendix II) with zero harvest and all other population parameters being equal. Therefore, decline in fecundity from the reference to treatment periods was biologically significant, especially because it occurred with lower survival rates. Management Implications 1) A 15% design harvest (actual harvest averaged ~16%) of independent pumas in this population manipulation was associated with a substantial population decline in 3 years. Up to ~12% harvest of independent pumas may be sustainable. 2) Human causes of mortality, especially hunting, affected survival of independent pumas in the treatment period. 17 �3) Natural causes of mortality were undetected by managers, as were some vehicle strikes and illegal killing. 4) The GMU puma management unit structure inadequately fitted the scale of hunting-caused mortality due to the movements of pumas beyond GMU boundaries. Literature Cited Anderson, A. E. 1983. A critical review of literature on puma (Felis concolor). Special Report No. 54, Colorado Division of Wildlife, Ft. Collins. Anderson, A. E., D. C. Bowden, and D. M. Kattner. 1992. The puma on the Uncompahgre Plateau, Colorado. Colorado Division of Wildlife Technical Publication No. 40. Anderson, C. R., and F. G. Lindzey. 2005. Experimental evaluation of population trend and harvest composition in a Wyoming cougar population. Wildlife Society Bulletin 33:179-188. Bauer, J. W., K. A. Logan, L. L. Sweanor, and W. M. Boyce. 2005. Scavenging behavior in puma. The Southwestern Naturalist 50:466-471. Beausoleil, R. A., G. M. Koehler, B. T. Maletzke, B. N. Kertson, and R. B. Wielgus. 2013. Research to regulation: cougar social behavior as a guide for management. Wildlife Society Bulletin 37:680688. Burnham, K. P., and D. R. Anderson. 1998. Model selection and inference: a practical informationtheoretic approach. Springer-Verlag, New York, New York, USA. Choate, D. M., M. L. Wolfe, and D. C. Stoner. 2006. Evaluation of cougar population estimators in Utah. Wildlife Society Bulletin 34:782-799. Colorado Division of Wildlife 2002-2007 Strategic Plan. 2002. Colorado Department of Natural Resources, Division of Wildlife, Denver. Colorado Division of Wildlife. 2007. Colorado mountain lion management data analysis unit revision and quota development process. Colorado Division of Wildlife, Denver. Cooch, E., and G. White. 2015. Program MARK- a gentle introduction, 14th edition. Colorado State University, Fort Collins. Cooley, H. S. 2008. Effects of hunting on cougar population demography. Ph.D. Dissertation. Washington State University, Pullman. Hornocker, M. G. 1970. An analysis of mountain lion predation upon mule deer and elk in the Idaho Primitive Area. Wildlife Monographs No. 21. Kaplan, E. L., and P. Meier. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 53:457-481. Kreeger, T. J. 1996. Handbook of wildlife chemical immobilization. Wildlife Pharmaceuticals, Inc., Fort Collins, Colorado. Lambert, C. M. S., R. B. Wielgus, H. S. Robinson, D. D. Katnik, H. S. Cruickshank, R. Clarke, and J. Almack. 2006. Cougar population dynamics and viability in the Pacific Northwest. Journal of Wildlife Management 70:246-254. Laundre, J. W., L. Hernandez, D. Streubel, K. Altendorf, and C. L. Lopez Gonzalez. 2000. Aging mountain lions using gum-line recession. Wildlife Society Bulletin 28:963-966. Laundré, J. W., L. Hernández, and S. G. Clark. 2007. Numerical and demographic responses of lions to changes in prey abundance: Testing current predictions. Journal of Wildlife Management 71:345355. Laundré, J. W., and L. Hernández. 2007. Do female pumas (Puma concolor) exhibit a birth pulse? Journal of Mammalogy 88:1300-1304. Lindzey, F. G., W. D. Van Sickle, S. P. Laing, and C. S. Mecham. 1992. Cougar population response to manipulation in southern Utah. Wildlife Society Bulletin 20:224-227. Logan, K. A., L. L. Irwin, and R. Skinner. 1986. Characteristics of a hunted mountain lion population in Wyoming. Journal of Wildlife Management 50:648-654. 18 �Logan, K. A., and L. L. Sweanor. 2001. Desert puma: Evolutionary ecology and conservation of an enduring carnivore. Island Press, Washington, D. C. Logan, K. A. 2004. Colorado lion research and development program: population characteristics and vital rates study plan. Colorado Division of Wildlife, Ft. Collins Research Center. Logan, K. A. 2008. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Wildlife, Fort Collins. Murphy, K. M. 1983. Relationships between a mountain lion population and hunting pressure in western Montana. Master’s Thesis, University of Montana. Pojar, T. M., and D. C. Bowden. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550-560. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Ross, P. I., and M. G. Jalkotzy. 1992. Characteristics of a hunted population of cougars in southwestern Alberta. Journal of Wildlife Management 56:417-426. Ruth, T. K., M. A. Haroldson, K. M. Murphy, P. C. Buotte, M. G. Hornocker, H. B. Quigley. 2011. Cougar survival and source-sink structure on Greater Yellowstone’s Northern Range. Journal of Wildlife Management 75:1381-1398. Spreadbury, B. R., K. Musil, J. Musil, C. Kaisner, and J. Kovak. 1996. Cougar population characteristics in southeastern British Columbia. Journal of Wildlife Management 60:962-969. Stoner, D. C., M. L. Wolfe, and D. M. Choate. 2006. Cougar exploitation levels in Utah: Implications for demographic structure, population recovery, and metapopulation dynamics. Journal of Wildlife Management 70:1588-1600. Sweanor, L. L., K. A. Logan, and M. G. Hornocker. 2000. Cougar dispersal patterns, metapopulation dynamics, and conservation. Conservation Biology 14:798-808. Sweanor, L. L., K. A. Logan, J. W. Bauer, B. Milsap, and W. M. Boyce. 2008. Puma and human spatial and temporal use of a popular California state park. Journal of Wildlife Management 72:10761084. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 (Suppl):S120-S139. Yeager, K., B. Kendall, M. Alldredge. 2013. The Use of Lures, Hair Snares, and Snow Tracking as NonInvasive Sampling Techniques to Detect and Identify Cougars. Colorado Cooperative Fish and Wildlife Research Unit, Ft. Collins. Zar, J. H. 1984. Biostatistical analysis. Second edition. Prentice Hall, Englewood Cliffs, New Jersey. Prepared by: ________________________ Kenneth A. Logan, Wildlife Researcher 19 �Table 2. Count of pumas based on numbers of known radio-collared pumas, visual observations of non-marked pumas, harvested non-marked pumas, and track counts of suspected non-marked pumas on the study area during reference years 4 and 5 (RY4, RY5) and treatment years 1-5 (TY1-TY5). Also indicated * is the population projection for RY5 due to lack of a reliable count (see text), Uncompahgre Plateau study area, Colorado. Period & Year RY4 RY5 TY1 TY2 TY3 TY4 TY5 Study Area region Adults Female Male Subadults Female Male Female East slope 10 4 3 4 4 West slope 6 4 2 0 1 subtotals 16 8 5 4 5 Total Independent Pumas = 33: 21 females, 12 males. Cubs = 20-21 East slope 11-13 5-6 2-4 0-1 2 West slope 9-10 4 1-2 1 3 subtotals 20-23 9-10 3-6 1-2 5 Total Independent Pumas = 37, *45 East slope 16 10 1 2 1 West slope 14 10 0 3 3 subtotals 30 20 1 4 4 Total Independent Pumas = 56: 31 females, 25 males. Cubs = 19-24 East slope 15 5 3 2 7 West slope 15 7 2 3 2 subtotals 30 12 5 5 9 Total Independent Pumas = 52: 35 females, 17 males. Cubs = 39 East slope 13 4 1 3 4 West slope 14 5 3 5 1 subtotals 27 9 4 8 5 Total Independent Pumas = 48: 31 females, 17 males. Cubs = 19 East slope 15 4 3 2 4 West slope 10 5 3 0 2 subtotals 25 9 6 2 6 Total Independent Pumas = 42: 31 females, 11 males. Cubs = 24 East slope 10 6 3 6 6-7 West slope 13 4 1 1 1 subtotals 23 10 4 7 7-8 Total Independent Pumas = 44: 27 females, 17 males. Cubs = 25-28 Cubs Male 4 2 6 Unknown sex 7 2-3 9-10 5 2 7 5 4 9 3 3 7 4-8* 5-6 9-14 9 5 14 7 9 16 2 2 4 4 6 10 4 5 9 3 6 9 2 3 5 2 11-13 13-15 Table 3. Pumas killed by hunters on the study area during the treatment period, Uncompahgre Plateau, Colorado. Actual No. of No. Independent Percent harvest of Treatment Adult Adult Subadult Subadult pumas pumas in Independent Period Year Female Male Female Male Quota killed count pumas 8 9 56 16.1 TY1 2 5 1 1 8 8 52 15.4 TY2 0 5 2 1 8 8 48 16.7 TY3 3 1 0 4 5 5 42 11.9 TY4 2 2 0 1 5 5 44 11.4 TY5 1 3 0 1 subtotals 8 16 3 8 20 �Table 4. Causes of death in adult, subadult, and cub pumas in the reference and treatment periods, Uncompahgre Plateau, Colorado. Reference Period Adults (21F,11M at Females Females Males Males Total Total risk Number Percent Number Percent Number Percent Strife 3 50 1 100 4 57 Other natural 1 16.7 0 0 1 14.3 Vehicle strike 1 16.7 0 0 1 14.3 Depredation control 1 16.7 0 0 1 14.3 Illegal kill 0 0 0 0 0 0 Hunting 0 0 0 0 0 0 Subadults (8F,14M at risk) Strife 0 0 0 0 0 0 Other natural 1 50 0 0 1 33.3 Vehicle strike 1 50 0 0 1 33.3 Depredation control 0 0 0 0 0 0 Illegal kill 0 0 0 0 0 0 Hunting 0 0 1 100 1 33.3 Cubs (28F,27M at risk) Infanticide 9 75 4 100 13 81.3 Predation 1 8.3 0 0 1 6.2 Other unknown natural 1 8.3 0 0 1 6.2 Starvation 0 0 0 0 0 0 Vehicle strike 1 8.3 0 0 1 6.2 Depredation control 0 0 0 0 0 0 Illegal kill 0 0 0 0 0 0 Hunting 0 0 0 0 0 0 21 �Table 4. Continued Adults (39F,22M at risk) Strife Other natural Vehicle strike Depredation control Illegal kill Hunting Subadults (19F,34M at risk) Strife Other natural Vehicle strike Depredation control Illegal kill Hunting Cubs (28F,36M at risk) Infanticide Predation Other unknown natural Starvation Vehicle strike Depredation control Illegal kill Hunting Mauled by dogs Treatment Period Females Females Males Number Percent Number 3 14.3 0 7 33 0 2 9.5 1 2 9.5 0 0 0 1 7 33 14 Males Percent 0 0 0 0 6.7 93.3 Total Number 3 7 3 2 1 21 Total Percent 8.1 18.9 8.1 5.4 2.7 56.8 1 0 0 1 0 2 25 0 0 25 0 50 2 2 1 2 0 9 12.5 12.5 6.2 12.5 0 56.2 3 2 1 3 0 11 15 10 5 15 0 55 3 0 0 5 0 2 0 0 0 30 0 0 50 0 20 0 0 0 5 0 4 4 2 1 0 0 1 29.4 0 23.5 23.5 11.8 5.9 0 0 5.9 8 0 4 9 2 3 0 0 1 29.6 0 14.8 33.3 7.4 11.1 0 0 3.7 22 �Table 5. Adult puma survival modeling results, Uncompahgre Plateau, Colorado. Model Number ∆ AICc AICc wi Likelihood Parameters Deviance Model AICc {S(gender*period)} 396.9874 0 0.84055 1 4 162.0375 {S(gender+period)} 401.613 4.6256 0.0832 0.099 3 168.6719 {Sgender*year} 402.1608 5.1734 0.06327 0.0753 14 147.0023 {S{period)} 405.339 8.3516 0.01291 0.0154 2 174.4044 {S(gender)} 416.7478 19.7604 0.00004 0 2 185.8131 {S(.))} 417.3778 20.3904 0.00003 0 1 188.4475 {S(month)} 509.8345 112.8471 0 0 108 53.2893 {S(gender*month)} 716.7591 319.7717 0 0 216 0 Table 6. Puma adult, subadult, and cub annual survival rates, estimated from top model Sgender*period for each stage, Uncompahgre Plateau, Colorado. Adults (≥24 months old) Period Gender Average annual Lower 95% CI Upper 95% CI Survival estimate Reference Female 0.8599 0.7153 0.9345 Male 0.9593 0.7459 0.9942 Treatment Female 0.7415 0.6324 0.8230 Male 0.3971 0.2232 0.5692 Subadults (13-24 months old) Period Gender Survival estimate Lower 95% CI Upper 95% CI Reference Female 0.6303 0.2320 0.9058 Male 0.9233 0.6106 0.9893 Treatment Female 0.7026 0.4247 0.8832 Male 0.4272 0.2651 0.6071 Cubs (1-12 months old) Period Gender Survival estimate Lower 95% CI Upper 95% CI Reference Female 0.3439 0.1727 0.5683 Male 0.7132 0.4393 0.8875 Treatment Female 0.3906 0.2048 0.6147 Male 0.3020 0.1606 0.4944 * Over-dispersion parameter c-hat has to be estimated for the cub survival data. Table 7. Subadult puma modeling results, Uncompahgre Plateau, Colorado. Model Number Model AICc ∆ AICc AICc wi Likelihood Parameters Deviance {Sgender*period} 190.0683 0 0.48562 1 4 39.4874 {Speriod} 191.125 1.0567 0.28631 0.5896 2 44.5933 {Sgender+period} 192.1299 2.0616 0.17323 0.3567 3 43.5771 {S.} 195.2243 5.156 0.03687 0.0759 1 50.7065 {Sgender} 196.6757 6.6074 0.01784 0.0367 2 50.1439 {Smonth*period} 206.5767 16.5084 0.00013 0.0003 24 13.888 {Smonth*period*gender} 266.4878 76.4195 0 0 48 0 23 �Table 8. Subadult puma survival rates estimated with second-ranked model Speriod, females and males combined, Uncompahgre Plateau, Colorado. Subadults (13-24 months old) Period Survival estimate Lower 95% CI Reference 0.8371 0.5991 Treatment 0.5152 0.3685 Upper 95% CI 0.9464 0.6594 Table 9. Cub puma modeling results, Uncompaghre Plateau, Colorado. Model Number Model AICc ∆ AICc AICc wi Likelihood Parameters Deviance {Sgender*period} 317.1119 0 0.3335 1 4 309.0458 {S{period} 318.0671 0.9552 0.20686 0.6203 2 314.0473 {S{.} 318.1293 1.0174 0.20053 0.6013 1 316.1227 {Sgender+period} 319.4672 2.3553 0.10272 0.308 3 313.4276 {S{gender} 319.6242 2.5123 0.09496 0.2847 2 315.6045 {S{gender+Birthmonth} 320.8604 3.7485 0.05118 0.1535 3 314.8208 {Sgender*period}+birthmonth} 324.8551 7.7432 0.00695 0.0208 5 314.7558 {Smonth*period} 326.3474 9.2355 0.00329 0.0099 24 276.2961 {Smonth*period*gender} 367.9388 50.8269 0 0 48 263.5538 Table 10. Cub puma survival rates estimated with second-ranked model Speriod, females and males combined, Uncompahgre Plateau, Colorado. Over-dispersion parameter c-hat has to be estimated for the cub survival data. Cubs (1-12 months old) Period Survival estimate Lower 95% CI Reference 0.4999 0.3363 Treatment 0.3392 0.2187 24 Upper 95% CI 0.6635 0.4849 �Independent Pumas 60 - - , - - - - - - - - -~ - -.....----- - - - - - - - 50 + - - - - - ---F-- --=--....a;;::. - ------,,.,---- ----44- :Q 40 E ~ 30 + - - - - - - - - - - - - - - - - - - - - - 10 - + - - - - - - - - - - - - - - - - - - - - 0 +-----~-----~--~--~-~ RY4 RYS TY2 TY1 TY4 TY3 TY5 Study Years Figure 2. Counts of independent pumas, Uncompahgre Plateau, Colorado. Counts in RY4 and TY1 to TY5 are from ground surveys and capture efforts. The count for RY5 is biased low because capture efforts were insufficient due to lack of personnel to thoroughly search the study area (see text). Change in Number of Pumas 60 50 ~ 44 40 E :::J 36 .... c.. 30 33 0 0 Z 20 10 0 TY1 TY2 TY3 TY4 Treatment Years - Independent Pumas ...,._Adult Pumas Figure 3. Change in numbers of independent and adult pumas, Uncompahgre Plateau, Colorado. 25 �20 18 16 1.4 ~ .~~10 .2-J 12 ,- I- - f-- f--- - 0 0 % 8 6 4 2 0 - - -,- Jan. Feb. Mar. Apr. May June July Aug, Sep. Oct. lfov. Dec. Figure 4. Puma births (black bars) detected by month from May 19, 2005 to September 30, 2014 (n = 66 litters of 33 females; 60 litters were examined at nurseries when cubs were 25-45 days old, 4 litters were confirmed by tracks of ≥1 cubs following GPS- and VHF-collared mothers and 2 litters by remains of cubs of 2 GPS-collared mothers when cubs were ≤45 days old, Uncompahgre Plateau, Colorado. 26 �APPENDICES Appendix I. ACUC Capture and Handling Forms and Protocols File # _________________ Revised Date ______________ (ACUC Secretary will supply) COLORADO PARKS AND WILDLIFE ANIMAL CARE AND USE COMMITTEE (CPW ACUC) FORM FOR REVIEW OF NEW RESEARCH PROJECTS 1. Principal Investigator (s): Dr. Kenneth A. Logan, Mammals Researcher, CPW. Phone: 970-252-6013(o) or 970-275-3227(c) E-mail: ken.logan@state.co.us 2. 3. All investigators (including all individuals involved in implementing research: Principal investigator Ken Logan (CPW), all CPW technicians and other houndsmen. Location of facility or study area: The study area is on the Uncompahgre Plateau in western 4. Beginning date: December 1, 2008. 5. Ending date: April 1, 2014. 6. Title of project: Assessing Effects of Hunting on a Puma Population on the Uncompahgre Plateau, Colorado. 7. Species of animal (s): Puma concolor 8. A study Plan or Prospectus describing each research or pilot project is required with this form. Is the Study Plan attached? Yes _X_ No ___ 9. Rationale for use of this animal model: a. Explain why other models (e.g. nonanimal models, in vitro techniques) are inappropriate. This study pertains specifically to puma population dynamics and attendant effects of hunting off-take. It is intended to provide wildlife managers with useful information for the management of pumas in Colorado. b. If not a species specific study, why is this the most appropriate species for this research? Colorado in areas west and southwest of Montrose. The study area is the South Uncompahgre Plateau (in Mesa, Montrose, Ouray, and San Miguel Counties). The study area includes about 2,200 km2 of the southern halves of GMUs 61 and 62, and about 155 km2 of the northern edge of GMU 70. The area is bounded by state highway 348 at Delta, 25 Mesa road and Forest Service road FS503 to Nucla, state highway 97 to state highway 141 to state highway 145 to Placerville, state highway 62 to Ridgeway, U.S. highway 550 to Montrose, and U.S. highway 50 to Delta. c. If capturing wild animals for pen research, why is this source most appropriate? 27 �10. 11. If the study will use wild animals, describe capture and transport methods: Please refer to the attached study plan Puma capture and marking (pages 13-15, 18, 24) and the Mountain Lion Capture and Handling Guidelines. Location of capture: Pumas will be captured on the study area described in question 3 above. 12. Indicate number of animals to be used _20-30 pumas/year_. Provide a brief justification (or page reference in Study Plan) for sample size selected: Please see study plan sections Puma capture and marking, Population monitoring, and Population Size (pages 13-19). 13. By signing this form, you are verifying that all persons involved in this project are adequately trained. Briefly describe the training process(es) and list personnel responsible for animal care and handling: K. Logan, S. Young, have all been trained in and have directly captured, immobilized, and sampled pumas. All new technicians and houndsmen will receive training on pumas by principal investigator Ken Logan. 14. Provide a detailed description of the procedures and manipulations of animals, including an end point (if necessary) at which animals will be removed from experiment or be euthanized. (If described in Study Plan or Prospectus, provide reference page numbers.) If administration of anesthesia and /or surgery is part of the procedure, identify who will perform these tasks: Please see study plan section Puma capture and marking (pages 13-19). 15. Are the levels of pain and suffering, stress, discomfort, deprivation, etc., to be experienced by experimental animals greater than normally associated with handling, administration of therapeutics by commonly used methods, or routine venipuncture? Yes __ No _X_ If answered yes, attach a detailed justification and indicate here the date of search, some of literature search, date range searched, and key words and combination of key words searched to document the lack of alternative methods: 16. 17. Will pain and suffering be controlled? Yes ____ No ____ N/A _X__ If answered no, attach a detailed justification. Describe how pain and suffering not associated with routine handling will be controlled. a. Methods and dosage of anesthesia to be used: N/A b. Methods and dosage of analgesia to be used: N/A c. Methods and dosage of tranquilization to be used: N/A The attending veterinarian must be consulted when planning projects where handling of any animal will occur. Do this prior to submitting this application. Date consulted_________ Does the proposed project include planned euthanasia of animals? Yes _____ No _X__ 28 �Signing below assures that all investigators have reviewed the CPW ACUC Euthanasia Guidelines and that investigators will use appropriate methods for humanely destroying animals involved in their study. Please indicate the criteria for and methods of euthanasia to be used in this study: Date: ________ 18. Signed: ___________________________________ Principal Investigator Signing below assures that the planned research does not unnecessarily duplicate previous research on the subject and species proposed for study. Date: ________ Signed: ___________________________________ Principal Investigator 29 �Appendix II. Puma Population Model and Simulations Research on the Uncompahgre Plateau Puma Project from December 2004 to July 2007 provided estimates of puma population structure and parameters for a model-based approach developed by CPW biometrician Dr. P. Lukacs and Mammals Researcher Dr. K. Logan to examine options for the design of the remainder of this research, and as a preliminary assessment of the CPW puma management assumptions. Puma Population Modeling Our puma population projections for the study area involved an age-structured, deterministic, discrete time model. The additive puma population model structure is: Nt+1 = Adult Females = (SAF * NAFt + SSF * NSFt) * (1 – HAFt+1) + Adult Males = (SAM * NAMt + SSM * NSMt) * (1 – HAMt+1) + Subadult Females = ((r * SC * NCt) * (1 – HSFt+1)) * PISF/ESF + Subadult Males = (((1 − r) * SC * NCt) * (1 – HSMt+1)) * PISM/ESM + Cubs = Lỹ * AFR * NAFt+1 Terms: NAFt+1 = Number of adult females at year t+1. NAMt+1 = Number of adult males at year t+1. NSFt+1 = Number of subadult females at year t+1. NSMt+1 = Number of subadult males at year t+1. NJt+1 = Number of juveniles at year t+1. S = Survival rate for each specified sex and age stage. H = Proportion of the harvest rate comprised by each sex and age stage (e.g., 0.28 harvest rate * 0.40 adult females). r = Proportion of the subadult population that is female (e.g., 0.5; 1-0.5 = proportion of males). PI/E = Ratio of progeny + immigrants/emigrants. Lỹ = Average litter size. AFR = Proportion of adult females giving birth to new litters each year. Basic assumptions of the model include: 1) expected puma population projections and annual rates of increase (i.e., lambda) are conditional on the assigned puma population structure and demographic estimates, and 2) no density dependent responses are built into the model. In reality, density dependence probably operates in puma population dynamics, with competition for food regulating adult female density and competition for mates regulating adult male density (Logan and Sweanor 2001). We parameterized the model with data gathered on the pumas on the study area during the previous 3.7 years. The starting population was the minimum count of pumas and attendant estimated sex and age structure made during November 2007 to March 2008 (Table AI.1). We assumed that all individuals were present in the population during that entire period. No mortalities of independent pumas were detected. But, one radio-collared subadult male emigrated by March 19, 2008. Population parameters included: estimated rates of reproduction and sex and age-stage specific survival, which included data to July 2008 (Table I.2). Some sex and age-stage specific estimates of survival (i.e., adult male, subadult male, subadult female) came from the literature (Table 2), because our current sample sizes (i.e., number of individuals and years) were not adequate for realistic estimates (i.e., estimates from our data were 1.0 for adult males and subadults). We did not use actual rates in the literature where estimates involved the pooling of data on sexes and age stages, and where sample sizes for age stages were not presented (e.g., Anderson et al. 1992). In addition, the ratio of progeny and immigrant recruits to 30 �emigrants as a model input was from the literature, because such data were scarce and does not exist for Colorado (all references in Table AI.2). We preferred using the population characteristics and parameter estimates gathered in the current research effort, because this is the puma population we intend to manipulate to assess current CPW puma management strategies. Table AI.1. Minimum puma population count on Uncompahgre Plateau study area, Colorado, November 2007 to March 2008 (RY4). The minimum count involves counting all radio- and GPS-collared pumas, all other marked pumas, and all presumably unmarked pumas detected on the study area during the period. Presumed unmarked pumas could be marked with ear-tags and tattoos. Their tracks and movements could be separated from movements of radio- and GPS-collared pumas. Or they exhibited evidence that could separate them from other local marked pumas from their tracks (i.e., distinguishable by sex, number of cubs and/or relative size of cubs varied). Region East slope West slope Totals a Adults Subadults Female Male Female Male 10 4 3 4 6 4 2 0 16 8 5 4 Total Independent Pumas = 33a,b Female 4 1 5 Cubs Male 4 2 6 Unknown sex 7 2-3 20-21 Of the total, 23−24 (70−73%) independent pumas were marked and 9-10 (27−30%) were assumed to be unmarked. Table AI.2. Summary of preliminary puma population model parameter estimates obtained from the Uncompahgre Plateau Puma Project and from the literature on puma. Survival Sex and age stage Adult Female Estimate 0.87 Adult Male 0.91 Subadult Female 0.80 Subadult Male 0.60 Cub 0.50 0.90 Parameter Adult age Estimate 2+ years Reference Estimated average annual survival rate (n = 2 years) for 11−13 adult females on Uncompahgre Plateau study area. Estimated average annual survival rate (n = 8 years) for adult males in a nonhunted New Mexico puma population (Logan and Sweanor 2001:127-128). Estimated annual survival rate (n = 2 years) for 5−9 adult males on Uncompahgre Plateau study area was 1.00. Estimated subadult female survival in New Mexico (0.88, n = 16; Logan and Sweanor 2001:122) adjusted downward for potential lower survival for pumas 12-24 months old on Uncompahgre Plateau (0.642, n = 14 females and 10 males combined, life stages not known or described in Anderson et al. 1992:53). Survival of 7 radio-collared pumas (5 males, 2 females) in the subadult stage in the current Uncompahgre Plateau puma study is 1.00. Estimated subadult male survival in New Mexico (i.e., 0.56, n = 9; Logan and Sweanor 2001:122) adjusted upward for potential slightly higher survival for pumas of both sexes 12-24 months old (i.e., 0.642) on Uncompahgre Plateau (Anderson et al. 1992:53). Survival of 7 radiocollared pumas (5 males, 2 females) in the subadult stage in the current Uncompahgre Plateau puma study is 1.00. Estimated cub survival rate (n = 38 cubs combined sexes), on Uncompahgre Plateau study area. This survival rate is applied to the model starting with the expected number of cubs from birth in RY5. Estimated cub survival for cubs ≥7 months old, and is applied to RY4 cubs only, because the minimum count of pumas in RY4 was tallied when most cub mortality had already occurred. Survival of cubs ≥7 months old in the literature is about 0.95 (Logan and Sweanor 2001). Here, a more conservative 0.90 is used in this model. Reproduction Reference Assume all females 2 years old and older are adults (Logan and Sweanor 2001: 93-94). 31 �Litter size 2.81 Secondary sex ratio observed at nurseries 1:1 Proportion of adult females producing new litters each year 0.65 Parameter Subadult female Estimated Ratio 1.02 Subadult male 0.94 Average litter size for 21 litters on the Uncompahgre Plateau study area = 2.810 ± 0.9808SD; litters were examined when the cubs were 26 to 42 days old. Secondary sex ratio was 33:26 for 21 litters examined at 29 to 42 days old on the Uncompahgre Plateau study area (not significantly different from 1:1, (X2 = 0.8305 < 3.841, α = 0.05, 1 d.f.). This result supported Logan and Sweanor 2001:69, n = 148). Proportion of adult females giving birth each year (n = 3 years for n = 12, 13, 12 females), Uncompahgre Plateau study area. Proportion for a non-hunted puma population in New Mexico was 0.50 (Logan and Sweanor 2001:98). Progeny + Immigrant Recruits/Emigration Ratio Reference No data for pumas in Colorado exists. Assume the ratio of female immigrants to emigrants = 1.02. This ratio is consistent with estimates for a New Mexico puma population that functioned as a source (Sweanor et al. 2000). No data for pumas in Colorado exists. Assume the ratio of male immigrants to emigrants = 0.94, (i.e., male immigration is half of emigration). This ratio is consistent with estimates for a New Mexico puma population that functioned as a source (Sweanor et al. 2000). Results of Puma Population Simulations Expected minimum population sizes for independent pumas for RY5 and TY1 conditional upon the number of independent pumas counted in Reference Year 4 (RY4) and the model input parameters and assumptions (given in Tables AI.1 and AI.2). Table AI.3. Year RY4 RY5 TY1 Adult Female 16 18 23 Puma Population Size Subadult Male Female 8 5 10 9 14 8 Male 4 8 8 Cub 20 33 42 32 Independent Pumas Total 33 count 45 projected 53 projected �Appendix III. MOUNTAIN LION HUNTER SURVEY MOUNTAIN LION HUNTER SURVEY EXPERIMENTAL LION HARVEST UNCOMPAHGRE PLATEAU STUDY AREA- GMUs 61, 62, and 70 Hunter Name: ___ License No.: CID No.: 1. Please circle the days on which you hunted (please count partial days hunting as full days) November: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30 December: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31 January: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31 2. Name the drainages and mesas where you hunted __________________________________________________ ____________________________________________________________________________________________ ____________________________________________________________________________________________ 3. Did you hunt with hounds? YES or NO (circle one) 4. Did you hunt with an outfitter? YES or NO (circle one) 5. Do you consider yourself to be a SELECTIVE hunter or a NON-SELECTIVE hunter? (read explanation below, then circle one) A SELECTIVE hunter is one that purposely is hunting for a specific type of legal lion, such as a male, large male, or large female, and therefore attempts to distinguish between male and female tracks, large and small males or females before taking the animal, and is willing to pass up lions that are detected from tracks or when treed. A NON-SELECTIVE hunter is one that intends to take whatever legal lion is first encountered or caught, with no desire for sex or size. 6. What was the sex of the lion that made the first set of tracks you encountered that were less than one day old? FEMALE  MALE  Did you pursue the lion to harvest it? YES  NO  NOTE: Adult & subadult male lions usually have hindfoot heel pad widths greater than or equal to 2 1/16 in. (52mm) wide. Adult & subadult female lions usually have hindfoot heel pad widths less than or equal to 1 15/16 in. (50 mm) wide. 7. Of the total tracks you encountered that were less than one day old, how many were male (_____) and female (_____) lions? (write number on the blank) 8. How many tracks were of females followed by cubs? _________ 9. How many times did you pursue lions with dogs? _________ 10. How many times did you tree or bay lions with dogs? _________ 11. How many of the lions treed and bayed were males (________), females (________), and cubs (________)? 12. Were any of the lions marked with a visible collar or ear-tags? YES or NO (circle one) If YES, describe the collar color, ear-tag color and number on each lion and its sex & age (i.e., male or female; adults ≥2 yrs. or subadults ~1-2 yrs.; indicate male or female and adult or subadult for each) ___________________________________________________________________________________________ 13. Describe the non-marked lions you caught (e.g., adult male, adult female, subadult male, subadult female) and list here: ___________________________________________________________________________________ 14. Did you harvest a lion? YES or NO (circle one) If YES, what was it? MALE or FEMALE (circle one). ADULT (≥2 yrs.) or SUBADULT (~1-2 yrs.) (circle one) 15. What was the seal number? ____________________ 16. Did marks (e.g., collar, ear-tag) on the lion influence your decision to harvest or not harvest the animal? (check one)  TO HARVEST  NOT TO HARVEST  NO INFLUENCE AT ALL 17. Did snow facilitate your harvest? YES if the puma was tracked on snow. NO if the puma was tracked on ground without snow. (circle one) 33 �Compliance Endangered Species Act This research will involve trapping mountain lions using hounds, cage traps and snares. It is extremely unlikely that any listed species under the Endangered Species Act will be inadvertently captured. However, in the unlikely event that a lynx or wolverine was captured, we will immediately release the animal unharmed. We will utilize existing roadways on public and private lands to access areas for running hounds and setting traps. Other field work on this project will comprise telemetry monitoring primarily from roads and fixed wing aircraft, minimizing potential for disturbing any listed species. No activities associated with this project pose a threat to the well-being of any listed species in Colorado. Animal Welfare Act The project is approved through Colorado Division of Wildlife’s Animal Care and Use Committee (Project #08-2004 and #03-2007). NEPA This research falls under a Categorical Exclusion as set forth in Title 40, Section 1508.4 of the Code of Federal Regulations (i.e., 40 CFR 1508.4) because the actions in this research do not involve significant environmental impacts. Other Landscape-Oriented Federal Acts This research will have no impact on the landscape, and therefore, will not violate provisions of other Federal Legislation governing floodplains and wetlands, historical sites, and prime and unique farmlands. Americans With Disabilities Act When hiring personnel as part of this project, qualified individuals will not be discriminated against based on disability. No structures or access points will be constructed as part of this research, and thus accessibility is not applicable. 34 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY State of: Cost Center: Work Package: Task No.: Colorado 3430 3003 3 Federal Aid Project No. W-204-R5 : : : : : Parks and Wildlife Mammals Research Carnivore Conservation Assessing Effects of Hunting on a Puma Population on the Uncompahgre Plateau, Colorado Period Covered: July 1, 2015  June 30, 2016 Mountain lion population responses to sport-hunting on the Uncompahgre Plateau, Colorado Principal Investigator: Kenneth A. Logan, Ken.Logan@state.co.us All information in this project summary is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. Colorado Parks and Wildlife (CPW) conducted a 10-year (2004−2014) study on effects of sport-hunting on a mountain lion population on the Uncompahgre Plateau. The purpose was to examine effects of hunting on a lion population, to evaluate assumptions used by CPW in lion management, and learn how lion hunter behavior may influence harvest. This report summarizes the latest analysis of the effects of hunting and other causes of mortality on a lion population. Analyses are ongoing and are expected to provide reliable information for application in lion management in Colorado. The study was designed with a reference period (years 1−5, RY1−RY5) without mountain lion hunting, and a treatment period (years 6−10, TY1−TY5) with lion hunting. The reference period began December 2004 and ended October 2009. The treatment period began November 2009 and all data collection ended in December 2014. The study area was on the Uncompahgre Plateau in Mesa, Montrose, Ouray, and San Miguel Counties. The 2,996 km2 (1,157 mi.2) study area included the southern halves of Game Management Units (GMUs) 61 and 62, and the northern edge of GMU 70. The Uncompahgre Plateau Study Area GMU (UPSA from here on) was in the largest 8% of the 185 GMUs used to manage lions in Colorado (average = 1,457 km2, range = 71−4,460 km2). Because this study was designed to represent a lion population segment on a Colorado GMU scale, the study area was managed as its own GMU so that inferences from the study could be interpreted at the GMU scale. 1 �From December 2, 2004 to October 30, 2014 we captured about 256 individual lions a total of 440 times on the UPSA. We individually marked 226 lions: 109 in the reference period and 115 in the treatment period. Marked lions provided known-fate data on 75 adults, 75 subadults, and 118 cubs. In addition to the lions captured by our research team during the treatment period, lion hunters captured and killed a total of 35 lions, including 8 adult females, 16 adult males, 3 subadult females, and 8 subadult males. Lion hunters also reported having captured and released 30 independent lions, with their reported gender identification of 19 females and 11 males. During the reference period without sport-hunting as a mortality factor the population of independent lions comprised of adults and subadults increased from a low of 33 lions counted in RY4 to a high of 56 lions counted in TY1 (Fig. 1). This indicated that lion management on the study area before this study probably suppressed the lion population. Along with the population increase during the reference period, adult lion survival was high and the age structure of independent lions increased. In the reference period, of the 32 (21 females, 11 males) adult radio-collared lions we monitored 7 adult lions died but none from hunting. Causes of death were attributed to: 5 natural causes (4 intra-specific strife, 1 unknown), 1 vehicle strike, and 1 depredation control. Of the 22 subadults (8 females, 14 males) providing known-fate data, 3 died. One male that had dispersed from the study area was killed by a hunter that did not see the tags (tagged lions that ranged north of the study area were protected from hunting during the reference period). Other causes of death in subadults were 1 natural cause (trampled by elk) and 1 vehicle strike. Of 55 radiocollared cubs (28 females, 27 males) monitored in the reference period, 16 died. Causes included: 13 infanticide, 1 predation, 1 unknown natural, and 1 vehicle strike. In the reference period natural causes dominated deaths of adults and cubs (71.3% and 93.8%, respectively), but 2 of 3 subadult deaths were from human causes. The treatment period was managed with mountain lion sport-hunting. TY1 was the first year that hunting influenced the lion population after 5 years of no hunting, and it was marked with the highest count of independent lions (56) on the study area. TY1−TY3, the lion harvest rate was set with a design quota of 8 lions to test if a 15% harvest of independent lions with 35−45% independent females in the harvest would result in a stable-to-increasing population. However, the expectation that a 15% harvest results in a stable-to-increasing population was not supported as the population of independent lions declined steadily from 56 in TY1 to 42 by TY4 (Fig. 1). Results from TY1−TY4 indicated that reducing a lion population with hunting is achievable at a 15% harvest rate with other human-caused and natural mortality operating on the population. The lion population in the treatment period was expected to continue to decline if the quota remained at 8 lions. Therefore, in an effort to find a harvest rate useful to managers that might result in a stable-to-increasing population for the remainder of the study, the quota was reduced to 5 lions. This quota represented about 11−12% harvest rate of independent lions for TY4 and TY5. 2 �Sport-hunting was the most important cause of death for independent lions during the treatment period. Of the 61 adults (39 females, 22 males) we radio-monitored during the period, 37 died. Hunting caused 56.8% of adult deaths (n = 21: 14 males, 7 females), followed by natural causes (27%; n = 10: 7 unknown with 6 probably disease-related and 1 due to starvation with senescence, 3 intra-specific strife), and other human causes (16.2%; n = 6: 3 vehicle strike, 2 depredation control, and 1 illegal kill). Of the 53 subadults (19 females, 34 males) providing known-fate data, 20 died. Hunting caused 55% of the subadult deaths (n = 11: 9 males, 2 females). Natural mortality followed in importance with 25% (n = 5: 3 intra-specific strife, 2 other natural), then closely by 20% other human causes of death (n = 4: 3 depredation control, 1 vehicle strike). Combining adult and subadult lion deaths in the treatment period, human causes were 73.7 % (i.e., 42/57*100), of which hunting caused 76.2% (i.e., 32/42*100) and other human causes comprised 23.8% (10/42*100). Of the 63 radio-collared cubs (27 females, 36 males) monitored, 27 died. Mortality causes in the cubs included: 9 infanticide, 4 other natural, 2 vehicle strike, 3 depredation control, and 9 starvation. The 9 cubs starved after the deaths of 5 mothers due to: hunting (2 mothers involving 3 cubs), depredation control (1 mother with 3 cubs), and natural causes (2 mothers involving 3 cubs). Natural mortality comprised the majority of cubs deaths (15/27*100 = 55.6%). But, human-caused cub deaths in the treatment period increased to 44.4% (12/27*100 = 44.4%) from 6.2% in the reference period. In the treatment period, the population of independent lions declined from a total count of 56 in TY1 to a low of 42 in TY4, a 25% decline after three hunting seasons (Fig. 1). The abundance of adult females declined 23.3% by TY5. Adult males declined 55% by TY3 and TY4, and 50% by TY5. The percentage of females in the harvest TY1−TY5 was 31.6%; comprised of 23% adult females and 8.6% subadult females. The remainder of the harvest was comprised of adult males (45.7%) and subadult males (22.9%). After we reduced the quota to 5 for TY4 and TY5, the abundance of independent pumas seemed to stay in a low phase and may have slightly increased (Fig. 1). Hunting in surrounding GMUs also contributed to the decline in the abundance of independent lions on UPSA. Ten radio-collared independent lions (2 adult females, 7 adult males, 1 subadult female) included in treatment year winter counts were killed by hunters in adjoining GMUs 61 North, 62 North, 65, and 70 because those lions had home ranges that extended beyond the boundaries of UPSA. Those lions were counted in the hunting quota in the adjoining GMUs, not UPSA. Including these deaths off the study area, the percent of hunting kill from TY1−TY5 ranged from 11.4%−25% (average = 18.2%) of independent lions in winter counts on the UPSA. The actual hunter-kill of the number of independent lions during TY1−TY3 ranged from 17.3−25% (average = 21.8%), and was associated with the population decline phase. During TY4−TY5 the actual hunter-kill was 11.4−19.0% (average = 15.2%), and was associated with the low population phase. We used an information-theoretic approach and Akaike’s Information Criterion to rank survival models with and without the treatment effect for adults, subadults, and cubs. The hunting treatment was indicated as an important factor explaining variation in adult and subadult 3 �male lion survival rates. Average annual survival rates of adult male lions declined significantly from 0.96 in the reference period to 0.40 in the treatment period. Likewise, subadult male lion survival rates declined significantly from 0.92 in the reference period to 0.43 in the treatment period. Although average annual adult female lion survival in the reference period, 0.86, was not statistically different than in the treatment period, 0.74, the decline in the abundance of adult females by 23.3% from TY1−TY5 suggested that the lower survival rate during the treatment period was biologically significant. Subadult female survival in the reference period, 0.63, was not statistically different from survival in the treatment period, 0.70. For cubs, models indicated that whether the dam lived or died was the single most important factor affecting cub survival. For the entire study period, the rate of cub survival to the subadult stage was 0.45. Female cub survival, 0.42, was not statistically different than male cub survival, 0.48. Age structure of independent lions declined from TY1−TY5. After 5 years of no hunting, the younger and up to middle aged (i.e., 1−5 years old) lions comprised the majority of the population and with both adult females and males being represented up to the oldest ages (i.e., >5−10+ years old). After 5 years of hunting, adult males >5 years old were eliminated from the age structure. Average litter size in the reference period, 2.76, was not statistically different from the treatment period, 2.38. Likewise, parturition rate for adult females in the reference period, 0.63, was not statistically different from the treatment period, 0.48. Sex ratio of cubs born in the reference or treatment periods, and in the study overall was not statistically different from parity. Management Implications 1) In the GMU-based mountain lion management structure in Colorado, a design harvest of ≥15% of independent lions with an average of ≥20% adult (i.e., 2+ years old) females in the harvest, and with other human and natural causes of mortality operating on a relatively high density lion population, can cause population decline in as few as 3 years. Managers should consider accounting for all detectable (i.e., recorded) human-caused mortality in quotas when setting removal rates in respect to lion population management objectives. Other human causes of death comprised about 24% of the total human-caused mortality with the remaining 76% of deaths due to hunting in the treatment period on the UPSA GMU when the lion population declined and reached a low phase. 2) It can take up to 5 years for a lion population previously reduced to a low density to recover to a relatively high density after hunting has been eliminated. 3) Design harvests of up to 11−12% of independent lions with an average of <20% adult females in the harvest is expected to result in a stable, possibly increasing population, considering that other human- and natural-causes of mortality operate on the population. 4) Lion population segment management objectives and attendant harvest rates can affect lion abundance in the particular GMUs of interest and adjacent GMUs where lions have home ranges overlapping GMU boundaries because GMUs in connected lion habitat are not closed lion populations. 5) Lion harvest management structure which includes provisions for reducing lion population segments to achieve specified management objectives (e.g., reduce predation 4 �on livestock or mule deer populations) should also provide for lion population segments managed with conservative harvest rates to allow for stable or increasing lion population segments (i.e., source-sink management) to ensure overall lion population resiliency because of all the unknowns and uncertainties associated with lion population management, including lion abundance and effects of harvest and other human and natural causes of lion mortality in GMUs. 6) Management experiments and research involving lion population segments should consider potential effects of historical lion hunting on and around the study areas. When experimental designs require reference conditions, human-caused mortality to lions should be limited or eliminated if possible. Final publications from this work are in preparation and will be submitted to the USFWS Wildlife & Sport Fish Restoration Program upon completion. ~ 50 _ _ _ _ _ __ ,_:..___ _ _~ = --..-==--=--......,..,,..------- ~ ~ .......lo-'!!l~- · - 44 ~ 30 - -~ ----------------------- c... 40 - - -~ L -- - - - - - - - - - - - -~ ~---- ~ 4> Q. 4> -,::, 20 - - - - - - - - - - - - - - - - - - - - - - - - £ 0 z 10 - - - - - - - - - - - - - - - - - - - - - - - - 0 RY4 RY5 TY1 TY2 Study Year TY3 TY4 TY5 Figure 1. Trends in the population of independent mountain lions associated with no sporthunting in the reference period years 4 and 5 (RY4, RY5) and with sport-hunting in the treatment period years 1 through 5 (TY1−TY5), Uncompahgre Plateau, Colorado. The count data were gathered from November through April each winter in efforts to canvass the study area thoroughly to count the number of independent lions in addition to the lion harvest. These data represent the number of independent lions expected to have been at risk to hunting during the Colorado lion hunting season November through March each year. 5 �Colorado Parks and Wildlife July 1, 2016  June 30, 2017 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 3003 3 Federal Aid Project No. W-204-R4 : : : : : Parks and Wildlife Mammals Research Carnivore Conservation Effects of Hunting on a Puma Population on the Uncompahgre Plateau, Colorado Period Covered: July 1, 2016  June 30, 2017 Author: K. A. Logan Personnel: J. Runge, Colorado Parks and Wildlife All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the authors. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Colorado Parks and Wildlife (CPW) conducted a 10-year (2004−2014) study on effects of sporthunting on a puma population on the Uncompahgre Plateau. The purpose was to examine effects of sporthunting on a puma population and evaluate assumptions used by CPW in puma management. The study design had a reference period (years 1−5) without puma hunting, and a treatment period (years 6−10) with puma hunting. The reference period began December 2004 and ended October 2009 and the treatment period began November 2009 and all data collection ended in December 2014. Counts of pumas in the study area population segment were made each winter coinciding with the Colorado puma hunting season from reference year 4 to treatment year 5. In the reference and treatment periods, 109 and 115 pumas were captured and marked, respectively. Those animals produced known-fate data for 75 adults, 75 subadults, and 118 cubs. Another 30 pumas were captured, but not handled for safety reasons. During the treatment period, hunters killed 35 independent pumas and captured and released 30 independent pumas. Responses of the puma population to sport-hunting and other causes of mortality were based on changes in four variables: 1) abundance of independent pumas, 2) survival of adult, subadult, and cub pumas, 3) reproduction rates, and 4) age structure of independent pumas. This report summarizes results of data analyses for use in puma management in Colorado and for preparation of manuscripts for publication on this research project. In the absence of sport-hunting in the reference period, the population of independent pumas increased by at least 70% and exhibited relatively high survival of independent pumas. There were clear effects of sport-hunting in the treatment period. Sporthunting was the major cause of death to independent pumas which added to other human-caused and natural mortality. The population of independent pumas on the study area declined 25% after the first 3 hunting seasons with a 15% design harvest of independent pumas on the study area. Actual harvests ranged from 15.4−16.7% of independent pumas and total independent puma mortality ranged from 16.1−20.8%. The sport-hunting harvest was reduced to 11−12% of independent pumas in the final two 1 �years of the treatment period with total independent puma mortality ranging 13.6−14.3% in which the population decline ceased and the population remained in a low phase. By the fifth year of the treatment period, the abundance of independent pumas had declined by 21%. The number of adult females and males declined by16.7% and 55%, respectively, by the fourth treatment year. By the the fifth year of the treatment period, the number of adult females and males had declined by 23.3% and 50%, respectively. The abundance of independent pumas on the study area was also influenced by sport-hunting off-take of pumas with home ranges that overlapped the study area and adjacent GMUs. During the treatment period, other human causes of death (excluding hunting) comprised about 24−29% of the total human-caused mortality (i.e., 74−81% of all deaths). Hunting comprised 71−76% of all human-caused deaths (depending on accounting method). Survival modeling results indicated that hunting was an important factor that explained adult and subadult male survival rates. The treatment period puma population exhibited statistically significant declines in survival rates of male adults (reference period S = 0.96 [95% CI = 0.746−0.994] , treatment period S = 0.40 [95% CI = 0.223−0.569]) and subadults (reference period S = 0.92 [95% CI = 0.611−0.989], treatment period S = 0.43 [95% CI = 0.265−0.607]). Changes in adult female (reference period S = 0.86 [95% CI = 0.715−0.935], treatment period S = 0.74 [95% CI = 0.632−0.823]) and subadult female (reference period S = 0.63 [95% CI = 0.232−0.906], treatment period S = 0.70 [95% CI = 0.425−0.883]) survival rates were not significantly different. However, the 23.3% decline in the abundance of adult females in the treatment period exhibited the biological significance of lower survival. The age structure for independent males declined in the treatment period, and reflected the lower survival and a selection of male pumas by hunters. There was no evidence of adequate compensatory immigration and local recruitment to offset mortalities of adult pumas. Puma cub survival was determined primarily by natural causes of mortality in the reference and treatment periods. The most important factor influencing cub survival was the fate of dams while cubs were dependent. Estimated probability of cub survival to the adult stage (i.e., 2 yr. old) in the reference period was 0.39. Estimated probability of cub survival to the adult stage in the treatment period was 0.22. Average litter sizes (reference period = 2.76 [SE = 0.1806, 95% CI = 2.41−3.12], treatment period = 2.38 [SE = 0.1972, 95% CI = 1.99−2.76]) and parturition rates (reference period = 0.63 [SE = 0.068, 95% CI = 0.49−0.75], treatment period = 0.48 [SE = 0.057, 95% CI 0.37−0.59]) were not statistically different in the reference and treatment periods. There was no evidence of compensatory reproduction in the treatment period. Therefore the most important factor associated with the decline in the abundance of independent pumas on the study area was hunting-caused mortality, which can be regulated by management. Management implications are listed at the end of this report to inform future puma management and research in Colorado. Final publications from this work are in preparation and peer-review and the final report will be submitted to the USFWS Wildlife & Sport Fish Restoration Program by the next reporting period. 2 �WILDLIFE RESEARCH REPORT EFFECTS OF HUNTING ON A PUMA POPULATION ON THE UNCOMPAHGRE PLATEAU, COLORADO Kenneth A. Logan PROJECT NARRATIVE OBJECTIVES 1. Gather data on puma population abundance, sex and age structure, vital rates (i.e., reproduction, age-stage survival rates, and emigration and immigration rates if possible), and agent-specific mortality in a non-hunted puma population phase and a hunted puma population phase for use in understanding puma population dynamics and evaluating and structuring puma harvest management and research approaches. 2. Test current CPW puma harvest-related assumptions that are applied to puma population segments in Colorado, and arrive at acceptable harvest levels intended to achieve population objectives, including increases, stability, and reductions in puma population segments. 3. Apply a hunting treatment to the puma population on the Uncompahgre Plateau study area designed to test CPW puma population management assumptions and learn about impacts of hunting on pumas. 4. Develop methods that detect changes in puma population abundance on the Uncompahgre Plateau study area that might be useful for monitoring changes in puma abundance in other puma habitats. SEGMENT OBJECTIVES 1. Arrange all the puma population data gathered during the 10-year study for data analysis. 2. Complete analysis of the puma population data and present it in written form for preparation for internal peer-review and for manuscripts submitted for publication. Introduction Colorado Parks and Wildlife (CPW) managers need reliable information on puma population biology to develop sound management strategies that address diverse public values and the CPW objective of actively managing pumas while “achieving healthy, self-sustaining populations” (Colorado Division of Wildlife 2002−2007 Strategic Plan:9). Moreover, the CPW 2015 Strategic Plan states Goal I to “Conserve wildlife and habitat to ensure healthy sustainable populations and ecosystems”. Active management of pumas includes managing for sustained populations to provide sport-hunting opportunity, and reducing puma populations to achieve specified management objectives, such as to reduce depredation on livestock and predation on mule deer, and to enhance public safety (Cooley et al. 2011). Thus, regulated sport-hunting is a main component of puma conservation and in addition to providing recreation it is intended to be used as a tool for managing pumas. Because sport-hunting is a major cause of death for pumas in hunted populations (Murphy 1983, Logan et al. 1986, Anderson et al. 1992, Ross and Jalkotzy 1992, Lambert et al. 2006, Stoner et al. 2006, Laundré et al. 2007), managers need information to better understand how hunting impacts puma populations to assess how management actions are working to meet management objectives. To improve the biological basis for managing pumas, the CPW began a process in 2000 to develop puma Data Analysis Unit (DAU) plans (Colorado Division of Wildlife 2007). DAUs are each comprised of several smaller Game Management Units (GMUs) and cover areas ranging 4,048−21,054 3 �km2. Conceptually, GMUs are the basic unit of puma harvest that is regulated with quotas on the number of pumas that are allowed to be killed by hunters each year and used to distribute harvest to achieve the management objective at the DAU level. The DAU plans involved a formulation to extrapolate an expected number of pumas on available habitat and the level of sport-harvest assumed to be acceptable to achieve one of two management objectives for each DAU: 1) to maintain a stable or increasing puma population, or 2) to suppress (i.e., reduce) the puma population. A series of “best biological judgments” and assumptions by CPW biologists on puma populations in DAUs was necessary because reliable and affordable methods for estimating puma population abundance in habitat were not available, and there was no information on impacts of hunting on Colorado puma populations. Consequently, biologists that developed DAU plans mostly used data from intensive puma population studies from other western states that were published in the literature and from information from studies of puma in Colorado (Anderson et al. 1992) as a guide. The information included estimates of puma population density, sex and age structure, population rates of increase, and expected impacts of harvest rates, and that information was extrapolated to expected puma habitat across Colorado. Colorado Parks and Wildlife (CDOW 2007) puma management assumptions included: 1) For areas managed for a stable or increasing puma population, acceptable total mortality (i.e., natural and human-caused) could fall in the range of 8 to 15% of the projected huntable population (i.e., adult plus subadult pumas) with female (i.e., adults and subadults) mortality in the range of 35 to 45% of the deaths. 2) For areas managed to suppress the puma population, total mortality could fall in the range of >15 to 28% of the projected huntable population with >45% of the deaths comprised of females. Other unknowns associated with these assumptions, in addition to the ones previously mentioned (i.e., puma abundance, effects of harvest) included undetected human-caused deaths and effects of natural mortality. All of these unknowns were consistently represented in the 19 puma DAU management plans developed by CPW biologists. The management objectives were to manage for stable or increasing puma populations in 17 of the 19 DAU plans. In plans with this objective, non-hunting mortality was assumed to be negligible, ranging from 0 to 7 pumas per year per DAU, with a median 0.5 pumas/year/DAU and an average 1.4 pumas/year/DAU for 17 of the 19 DAU plans that provided this information. Therefore, the total allowable mortality was generally considered to be comprised of sport-hunting off-take. Considering this, our focus was to investigate effects of sport-hunting off-take on a puma population when managed for a stable or increasing population. Specifically, we chose to investigate the effects that a 15% design harvest of independent pumas might have on a puma population, representing the maximum range of the assumption expected to result in a stable-to-increasing puma population (i.e., 8−15% harvest of the projected huntable population). Prior to the current puma research described in this report, none of the demographic prescriptions for management had been tested for their validity on a puma population segment at the GMU level in Colorado. Such testing is prudent because some assumptions made for management plans might be in error and cause puma populations to decline where the management objective is for stable or increasing puma populations. This objective is critical to providing resiliency in the puma population due to effects of hunting-caused mortality, which previous research has revealed was not compensatory (Cooley et al. 2009), and because other puma population segments would purposely be managed for suppression. Metrics from research in other western states support or are at variance with current CPW puma harvest guidelines for a stable to increasing population. Recent research in Wyoming indicated that a puma population could sustain a harvest comprised of 10 to 15% adult females, and population decline occurred when about 25% of adult females comprised the harvest (Anderson and Lindzey 2005:187). A Utah study found that a puma population declined when harvest exceeded 30% of the adults and subadults and comprised 42% females for 3 years (Choate et al. 2006). Another study in southern Idaho and 4 �northern Utah suggested that a harvest that included 15 to 20% of resident females probably would not reduce a puma population (Laundré et al. 2007). More recently, researchers in Washington modeled puma population dynamics and indicated that a 14% harvest of adult pumas was expected to result in a stable population and age structure (Beausoleil et al. 2013, Wielgus et al. 2013). A traditionally used reference claiming a sustainable puma harvest up to an extreme of 30% does not present any data to support that notion (Ashman et al. 1983). Another more recent reference has been used to support up to a 21.1% harvest rate on a puma population; but those authors clearly cautioned against using this metric because “potential effects of this harvest rate were offset by [three] interceding years when no cougars were shot. It is unknown what annual harvest rate could be sustained and still allow for stability or growth in the population size” (Ross and Jalkotzy 1992:424). Thus considering any of this information including our CPW assumptions, a population decline could occur if there is substantial over-projection error in the assumed puma abundance for an area and/or the applied harvest is excessive. This result is possible because actual puma population estimates are not available for any non-surveyed areas. In fact numbers used are at best educated guesses or biological judgments extrapolated over huge non-surveyed areas. This would be especially problematic if errors occurred for a substantial number of areas where the management objective is for a stable-to-increasing population. Thus, the state-wide strategic objective of managing for a healthy, self-sustaining puma population could be in jeopardy. This emphasized the need to quantify impacts of puma hunting on population parameters to structure guidelines that will likely achieve population objectives. Our current study serves as a field-based population-level test of the theoretical guidelines that could be derived from the literature and our management assumptions (previously cited). To address these information needs, CPW began this research in 2004 on the Uncompahgre Plateau to better understand puma population dynamics and effects of sport-hunting. The study was designed in two 5-year periods: a reference period (years 1−5) and a treatment period (years 6−10). The reference period provided baseline estimates on puma population abundance, sex and age structure, reproduction, survival, agent-specific mortality, and dynamics in representative puma habitat in Colorado where sport-hunting was not a cause of mortality. The treatment period occurred on the same study area and included manipulation of the puma population through the use of sport-hunting to provide information on the impact of hunting on a puma population. Study Area The study area for our puma research was on the Uncompahgre Plateau (in Mesa, Montrose, Ouray, and San Miguel Counties) in southwestern Colorado (Fig. 1). The 2,996 km2 (1,157 mi.2) study area included the southern halves of Game Management Units (GMUs) 61 and 62, and the northern edge of GMU 70. The Uncompahgre Plateau Study Area GMU (UPSA from here on) was in the largest 8% of the 185 GMUs used to manage pumas in Colorado (average = 1,457 km2, range = 71−4,460 km2). Because this study was designed to represent a puma population segment on a Colorado GMU scale, the study area was managed as its own GMU so that inferences from the study could be interpreted at the GMU scale. Highly developed road and all-terrain vehicle trail systems made the study area relatively highly accessible for puma research efforts. The UPSA was typical of puma habitat in Colorado. Vegetation cover transitioned from pinion-juniper (Pinus edulis-Juniperus spp.) covered foothills starting at about 1,700 m elevation to a Ponderosa pine (Pinus ponderosa) dominated woodland at mid-elevation, and up to the spruce-fir (Picea-Abies) and aspen (Populus tremuloides) forests at the highest elevations of 5 �about 3,000 m. Mid-elevation forests were interspersed with oak-serviceberry (Quercus spp.Amelancheir spp.) brushlands. Expansive sagebrush-steppe (Artemesia-grass) meadows and basins occupied mid-to-high-elevations, especially in the southern portion of the area. Mule deer (Odocoileus hemionus) and elk (Cervus elaphus) were the most abundant wild ungulates and were used as prey by pumas. Both mule deer and elk migrated from higher to lower elevations during winter due to deep snow at higher elevations. These deer also were subject to annual fall big game hunting seasons. Cattle and domestic sheep grazed on summer ranges and fed in low-elevation pastures in winter on UPSA. Sheep in particular were sometimes prey for pumas. People resided year-round along the eastern and western fringe of the area, and there was a growing residential presence especially on the southern end. Hobby animals kept by people, including alpacas, llamas, and goats, were also sometimes prey for pumas. Prior to our research, the puma population on the UPSA was subject to sport-hunting during a legal hunting season that extended from mid-November through March. The number of pumas allowed to be killed by hunters was limited by a quota. During the 5 previous years (1999-2003) an average of 12 independent pumas were reported killed by hunters on the study area (range = 9−17/yr.; unpublished puma mortality records, Colorado Parks and Wildlife, Denver). Based on the records of the gender and age stage (i.e., adults 2+ years old, subadults between 1−2 years old) of the pumas killed, 34% were classified as adult females; the rest were adult males and subadult females and males. During our research, portions of four abutting GMUs (61 North, 62 North, 65, and 70) were included in home ranges of adult radio-collared pumas that lived on the UPSA. Those GMUs also were subject to an annual puma sport-hunting season regulated by a quota. Consequently, those pumas with home ranges that overlapped the UPSA and adjacent GMUs were at risk of hunting mortality whether on or off the study area. As a result, some radiocollared pumas were killed off the study area during the treatment period (see later). During the 10 years of our research a total of 235 independent pumas were killed by hunters in those four surrounding GMUs, with a range of 14 to 29 pumas killed each year. As part of a puma management planning process in Colorado in 2004, the stated puma population management objectives were to manage for a stable puma population (i.e., not intended to increase or decline) in GMUs 61 North, 62 North, and 65, and a stable-to-increasing population in GMU 70 (Watkins 2004, Colorado Division of Wildlife 2004). Expected Results Information from our study will assist the Colorado Parks and Wildlife to improve puma management in Colorado. Results of the study will inform CPW biologists and managers about expected puma population dynamics and biological impacts of sport-hunting and other human-caused and natural mortality on a puma population in Colorado. The study also reveals puma life history traits and management effects useful for developing sound management strategies. Moreover, this study evaluated the current puma management structure and assumptions used in puma hunting management through the examination of data gathered directly from a GMU-level puma population manipulation. 6 �Approach The puma population on the Uncompahgre Plateau was studied for a total of 10 years divided into a 5-year reference period (i.e., reference years RY1−RY5) and a 5-year treatment period (i.e., treatment years TY1−TY5). In the reference period (November 2004 to October 2009) the study area was closed to puma hunting to provide baseline data on puma population dynamics without puma sport-hunting as a mortality factor, including: abundance, sex and age structure, survival, reproduction, agent-specific mortality, immigration and emigration. In addition, any radio-collared or ear-tagged pumas that ranged onto GMUs 61 North and 62 North (along the northwest boundary of the study area) were also protected from sport-hunting. The study area puma population was manipulated with sport-hunting during the treatment period (November 2009 to October 2014). This was an un-replicated case study on one geographic area having a before and after treatment effect design. This effort represented the largest number of pumas ever studied in a population segment in Colorado and an unprecedented opportunity for CPW to learn about puma population dynamics and effects of sport-hunting to apply to puma management. Field Methods Puma capture, marking, and sampling The capture, marking, and GPS- or VHF- collaring of individual pumas and subsequent monitoring was essential to a number of project objectives, including obtaining data on: population abundance, sex and age structure, vital rates, proportion of independent pumas marked in a camera grid, and puma movements to evaluate emigration and validity of GMUs. Pumas were captured year-round using 3 methods: trained dogs, cage traps, and by hand (for small cubs). All captured pumas were examined to ascertain gender and age, and describe physical condition and diagnostic markings. Ages of adult pumas were estimated initially by the gum-line recession method (Laundre et al. 2000) and dental characteristics of known-age pumas from New Mexico (Logan and Sweanor, unpubl. data) and later from known-age pumas in this study. Ages of subadult and cub pumas were estimated initially based on dental and physical characteristics of known-age pumas from New Mexico (Logan and Sweanor unpubl. data) and later from known-age pumas in this study. Ages of nurslings were estimated from apparent birthing dates indicated by GPS- and VHF-location data of collared mothers. Metric scale body measurements recorded for each puma included: mass (kg), pinna length, hind foot length, plantar pad dimensions, total length and tail length. Puma tissues were collected for genotyping individuals for parentage and relatedness analyses and disease screening, and included: skin biopsy (from the pinna receiving the 6 mm biopsy punch for the ear-tags), hair, and blood (30 ml from the saphenous or cephalic veins; only collected from pumas >10 weeks old). Universal Transverse Mercator Grid Coordinates on each captured puma were fixed via Global Positioning System (GPS, North American Datum 27). All pumas were handled in accordance with approved Animal Care and Use Committee (ACUC) capture and handling protocols in ACUC file #08-2004 (Appendix I) and ACUC protocol #03-2007 titled, Mountain Lion Capture and Handling Guidelines. Captured and handled adult, subadult, and cub pumas were marked 3 ways: GPS/VHF- or VHFcollar, ear-tag, and tattoo. The identification number tattooed in the pinna was permanent and could not be lost unless the pinna was severed. A colored (bright yellow or orange), numbered rectangular (5 cm x 1.5 cm) ear-tag (Allflex USA, Inc., DFW Airport, TX) was inserted into at least one pinna to facilitate individual identification during direct recaptures and when retrieved dead pumas were inspected. We captured pumas with dogs trained by experienced houndsmen whose research activities were guided directly in the field by the principal investigator. During the reference period when no sporthunting was allowed, our capture team could operate usually from early snow accumulation in November 7 �until high temperatures and black bear emergence from hibernation impacted the dogs’ effectiveness in April or May. However, during the treatment period we inserted our dog-assisted capture operations after the UPSA puma-hunting quota was reached so as not to interfere with hunters’ activities, harvest preferences, and influence on the puma population. Our shorter capture time-span usually went from midDecember through April, but we made up for that by deploying two capture teams, one on the west-side and one on the east-side of the UPSA. None of the houndsmen involved in our capture teams were allowed to hunt pumas for sport on the UPSA during the treatment period. Pumas captured by dogs usually climbed trees to take refuge. Adult and subadult pumas captured for the first time or requiring a change in telemetry collar were immobilized with Telazol (tiletamine hydrochloride/zolazepam hydrochloride) dosed at 5 mg/kg estimated body mass. The drug was delivered into the caudal thigh or shoulder muscles via a Pneu-Dart® shot from a CO2-powered pistol (Pneu-Dart X-Caliber, Pneu-Dart Inc., Williamsburg, PA). A 3m-by-3m square nylon net was deployed beneath the puma to catch it in case it fell. A researcher climbed the tree, fixed a rope to two legs of the puma and lowered the cat to the ground with an attached climbing rope. Once on the ground, the puma’s head was covered, its legs tethered, and vital signs monitored. Normal signs were considered: pulse ~70−80 bpm, respiration ~20 bpm, capillary refill time ≤2 sec., rectal temperature ~101oF average, range = 95−104oF (Kreeger 1996). Treed pumas that could not be safely immobilized and handled were shot with a biopsy dart (8 mm long x 3 mm dia., Pneu-Dart Inc., Williamsburg, PA) fired from the CO2-powered pistol to obtain a skin sample from the caudal thigh or shoulder. These samples were used in a study of puma population genetics and genomics. Cage traps were used to capture adults, subadults, and large cubs. Pumas were lured into traps using road-killed or puma-killed ungulates (Bauer et al. 2005, Sweanor et al. 2008). A cage trap was set only if a target puma (i.e., an unmarked puma, or a puma requiring a collar change) scavenged on the lure. Researchers continuously monitored a set cage trap from about 0.5−1 km distance by using VHF beacons on the cage and door. This allowed researchers to respond to the captured puma within 30 minutes. Pumas were immobilized with Telazol injected into the caudal thigh or shoulder muscles with a pole or hand syringe. Immobilized pumas were restrained and monitored as described above. Small cubs (≤10 weeks old) were captured using our hands (covered with clean gloves) or with a catch pole. Cubs were restrained inside new burlap bags during the handling process and were not administered drugs. Cubs at nurseries were approached when mothers were away from nurseries as determined by radio-telemetry. Cubs captured at nurseries were removed from the nursery a distance of ~20–100 m to minimize disturbance and human scent at nurseries. Cubs were returned to the exact nurseries immediately after sampling processes were completed (Logan and Sweanor 2001). Adult and subadult pumas were fitted with GPS (approximately 400 g each) or VHF collars (approximately 300 g each (Lotek Wireless, Newmarket, Ontario, Canada). Budget constraints limited the number of GPS collars (~10−15) available annually. Therefore, GPS collars were fitted to primarily adult pumas. Other adult and subadult pumas were fitted with VHF collars. Our efforts were to locate all collared pumas once per week from fixed wing aircraft and as weather and scheduling conditions allowed for data on survival, agent-specific mortality, and location. We checked the live/dead signal status from collared pumas from the ground opportunistically when we operated within their home ranges. VHF and GPS collars had mortality modes set to alert researchers when pumas were immobile for 3 hours (VHF collars) to 24 hours (GPS collars) so that dead pumas could be found for data on survival and agentspecific mortality. Because subadult male pumas were not fully grown, they also received leather expansion links in their collars. The expansion links added 10−12 cm when open and allowed the collars to be worn safely into the adult stage. 8 �We attempted to collar all cubs in each observed litter with a small VHF transmitter mounted on an expandable collar (62 g, model 080, Telonics Inc., Mesa, AZ) when cubs weighed 1.3−10 kg. The collars were designed to operate for 10−12 months, and expanded to 54 cm circumference to accommodate growth. Cubs with mass ≥7 kg were fitted with a larger expandable collar (90 g, model 210, Telonics Inc., Mesa, AZ). The collars were designed to operate for 12−18 months and could expand to 54 cm circumference to accommodate growth. Cubs approaching the age of independence (~11−14 mo. old) were fitted with Lotek LMRT-3 VHF collars (~400 g) with leather expansion links that add 10−14 cm to the collar circumference to accommodate the adult puma neck size. These collars operated for 2−3 years. Cubs were recaptured when possible to replace collars as necessary. Puma population sampling considerations The puma is one of the most difficult large, North American mammals to study because of its relatively low abundance on the landscape and its highly cryptic behavior. These characteristics were expected to influence the ability to sample individuals in the study population. The most efficient technique for locating and capturing pumas is detecting their tracks in snow and using trained dogs to pursue and secure them for sampling purposes. Hunters use the same technique to harvest pumas, which creates potential for biased survival rate estimates if researchers and hunters use similar strategies to detect and capture pumas. That is, with similar sampling strategies, pumas that are most vulnerable to being captured and radio-collared might also be more vulnerable to harvest, resulting in survival rates that are biased low. Hunters’ detection of puma tracks is heavily influenced by road access. To minimize bias potential, we attempted to intensively search the entire study area for puma tracks, irrespective of road characteristics, thereby equally detecting puma with both higher and lower hunter-detection probabilities. Thus, our approach was to apply roughly equal (i.e., intensive, uniform) searching intensity across the study area and apply an alternative capture technique with bait and cage traps that did not rely on track detection to capture pumas, and attempt to directly monitor via VHF telemetry a large majority of the population in the study area. Capture efforts to sample the adult and subadult pumas (i.e., independent pumas) subject to sporthunting mortality in the UPSA population were conducted mainly during winter when snow cover maximized the detection and capture probability of pumas. Snow provided a continuous or almost continuous substrate that registered tracks of terrestrial mammals. Puma tracks were highly distinctive and at ground level could be accurately and consistently visually identified and distinguished from tracks of all other mammals by trained personnel in a variety of snow and weather conditions and in the variety of terrains and vegetation communities. This characteristic was the reason why most intensive puma population studies in western North America have been conducted during winter to maximize detection, quantification, classification and monitoring of animals in the populations (Hornocker 1970, Logan et al. 1986, Lindzey et al. 1992, Ross and Jalkotzy 1992, Spreadbury 1996, Anderson and Lindzey 2005, Lambert et al. 2006, Laundre et al. 2007, Cooley et al. 2008). Puma population research in winter also directly linked the puma population investigated with animals killed during the hunting season, which in Colorado occurred annually during mid-November through March when snow facilitated the detection of pumas by hunters and maximized the ability of their dogs to follow scent in tracks to capture the pumas. In addition, during spring and fall and opportunistically in winter, we attempted to capture pumas in cage traps. Individuals caught in cage traps were available to move about the study area during winter and be exposed to hunters. The Uncompahgre Plateau study area was highly roaded. From those roads branched ATV trails that further facilitated thorough searches of the study area to detect pumas (Fig. 2). Still, the road system was not uniform, with some areas densely roaded, others moderately roaded, and one area in particular that did not allow motorized vehicles−the combined Camelback Wilderness Study Area (BLM jurisdiction) and Roubideau Special Management Area (U.S. Forest Service jurisdiction) in the main fork of Roubideau Canyon. That non-roaded area was about 109 km2 (42 mi.2). Because a system of roads and 9 �trails surrounded this area we were able to address this problem by hiking up the lower reaches of Roubideau Canyon and onto upper benches and canyons to search for puma tracks. A puma capture team, involving 4 people on separate search routes, was detailed to search this region on the surrounding roads, ATV/snowmobile trails, and hiking paths. By visiting this area repeatedly each winter we expected to detect some pumas that used the canyon and that might not have been detected in the canyon in other search days. Pumas were expected to move out of the non-roaded portion of the canyon periodically during the winter and be exposed by their movements. Thus, periodic searches of any of the search routes were expected to increase exposure of the pumas to our detection. The UPSA was partitioned into smaller search areas that a capture team could cover within 1−2 days to detect puma tracks on snow within each area each winter to spring (Table 1). The West-side search area was about 721 sq. km and the East-side search area was about 980 sq. km, with each area limited at the higher elevations by deep snow and lack of puma activity during winter. However, search routes at higher elevations enabled us to explore areas until deep snow limited our mobility and absence of puma activity was apparent, usually by the end of December each year. By this time, pumas and their ungulate prey mule deer and elk were concentrated on their winter range. The intent was to structure a thorough, relatively uniform, systematic search effort across the study area and to repeat it multiple times each winter−spring. To cover the areas efficiently, we traveled by four-wheel-drive trucks, all-terrain vehicles, snow-mobiles, and foot. When puma tracks ≤1 day old were detected, trained dogs were released to pursue the puma to capture, sample, and mark it. When puma tracks 1−2 days old were detected, we searched in the direction of travel of the puma in an effort to find ≤ 1 day old tracks that would facilitate pursuit of the puma. This sometimes lengthened our search within any particular area by another 1−2 days. When a GPS/VHF-collared puma was detected with radiotelemetry ≤1 km distance (usually <0.5 km) of the tracks and the direction of the tracks indicate that the puma was likely the collared individual, then we directed our efforts away from those tracks and redirected them to finding non-collared (i.e., non-sampled) pumas. Reliability of population count methods We expected the approach just described would enable us to monitor a large majority of the independent pumas on the UPSA from November through March and consequently gather reliable population data. The main way we indexed puma abundance was to count all the individual independent pumas we detected on the study area each winter. We were able to gauge the reliability of our field methods in Treatment Year 4 (TY4), when our resources allowed a one-time independent evaluation on the proportion of independent pumas on the study area that we captured and marked. This evaluation was provided by a camera grid study conducted by Colorado State University researcher K. Yeager (2016). K. Yeager established a 540 km2 grid comprised of 2 km x 2 km (4 sq. km) cells on the East-side of the UPSA. She randomly identified 18 cells for each of 3 survey periods, each lasting about 28 days during December 2012 to March 2013. Therefore, a total of 54 random cells were surveyed. Within each random cell Yeager subjectively chose the “best” site to attract pumas by using vocal baits each consisting of a Fur-Finder ® (Magna, UT) electronic recording of a distressed deer fawn. Each site also had a Reconyx ® PC900 Hyperfire camera (Holmen, WI) to record animal activity and hair-sampling devices (i.e., barbed-wire strands, sticky rollers) to attempt to acquire hair. This effort allowed us to evaluate our field methods intended to mark a large majority of independent pumas on the UPSA. Population Manipulation The puma population on the UPSA was manipulated by sport hunting after the 5-year reference period with no hunting. The hunting season was from mid-November to January 31, or until the last puma on the design quota was killed if it was before January 31. We initially tested the assumption that a 15% harvest of independent pumas comprised of 35−45% females would result in a stable-to-increasing population in the UPSA GMU. 10 �The initial harvest quota was 8 pumas which represented a 15% harvest of the expected number of independent pumas in treatment year 1 (TY1) on UPSA. Therefore, the predicted effect was that the 15% harvest of independent pumas would result in a stable or increasing population in subsequent years of the treatment period. The quota of 8 was based on the projected number of 52 independent pumas expected on the study area in winter 2009−10 (TY1), modeled from count data in winter 2007−08 (RY4) (see Appendix II, Table AI.3). After it was evident that the number of independent pumas had declined during TY1−TY3, we adjusted the harvest quota down to 5 pumas to represent an 11% harvest of the projected 45 independent pumas expected in TY4 in an effort to find a sustainable harvest rate useful to managers. The harvest quota of 5 was continued in TY5. The number of hunters on the study area each winter was not limited to be consistent with the Colorado puma hunting structure. Each hunter on the study area was required to obtain a hunting permit from the CPW Montrose Service Center. Permits were free and unlimited. Each permit allowed the individual hunter with a legal Colorado puma hunting license to hunt in the UPSA for 14 days from the issue date. Unsuccessful hunters that wanted to continue hunting past the permit expiration date could get serial 14-day permits until they harvested a puma or until the hunting season on the study area closed due to the quota being reached or the end of the hunting season. This permit system enabled CPW to estimate the number of hunters that actually hunted on the study area each season. In addition, a voluntary survey questionnaire (see Appendix III) was attached to each puma hunting permit issued to each hunter with a stamped envelope addressed to the CPW principal investigator. Hunters were asked to complete the survey as soon as possible for each hunting period associated with the permit in an effort to have hunters report the information while it was still fresh in their minds. All pumas harvested on the study area were visually examined and sealed by the principal investigator or project biologists just as is mandated by CPW for all pumas killed by hunters in Colorado. Hunters reported their puma kill to CPW within 48 hours of harvest and presented the puma carcass for inspection within 5 days of harvest. At the time of carcass check-in a mandatory CPW harvest check form was completed, which included gender, age estimate, and location of capture for each puma. Each successful hunter was asked to fill-out the one-page hunter survey form. Most other hunters either mailedin or handed-in their surveys on their own volition. If hunters did not respond, they were contacted by telephone or in person, if possible, and asked to complete the survey and return it. Mandatory hunter harvest checks provided accurate data on the gender and sex of pumas removed from the study population. Hunters also provided data to evaluate the relative vulnerability of pumas to harvest and potential for hunter selectivity. Hunter harvest and capture events also revealed availability and sex and age classes of non-marked pumas on the study area during the hunting season and before our capture teams operated after the season to quantify the population. Population Monitoring This monitoring plan enabled us to index the abundance of pumas with our counts and estimate the sex and age structure during each hunting season period November through March. Radio-collared pumas monitored year-round enabled us to gather population data, including: fecundity, birth interval, litter size, sex ratio, survival, agent-specific mortality, and recruitment to the adult stage. Emigration was revealed with a sample of radio-collared or ear-tagged marked offspring that left the study area. GPS- and VHF-collared pumas were located about once per week from light fixed-wing aircraft (e.g., Cessna 185) fitted with radio signal receiving equipment (Logan and Sweanor 2001). This monitoring enabled researchers to find GPS-collared pumas to acquire remote GPS location reports, monitor the status (i.e., live or dead) of individual pumas, and to locate carcasses for necropsy. Life status of GPS- and VHFcollared pumas were monitored from the ground opportunistically using hand-held yagi antenna. GPScollared pumas were monitored for survival status by using data from GPS collars. 11 �Analytical Methods Response variables We quantified and estimated responses of the puma population to sport-hunting and other causes of mortality based on changes in four variables: 1) abundance of independent pumas, 2) survival of adult, subadult, and cub pumas, 3) reproduction, including litter size and parturition rates, and 4) age structure of independent pumas. Puma population counts and trends The parameter of interest to managers was the abundance of independent pumas (i.e., adults and subadults) in winter, which coincided with the puma hunting season in Colorado when snow cover maximized the vulnerability of pumas to hunting. Our winter research effort was designed to be thorough to maximize the proportion of marked pumas in the population to sample for other parameter estimates, including survival, agent-specific mortality, reproduction rates, and age structure. Using this design, we attempted to obtain a reliable index to puma abundance with as complete of a count as we could achieve along with the attendant sex and age structure of the population that occurred from November through March (i.e., winter counts). We assumed the winter counts of adult and subadult pumas to be a census of the population without estimates of variance of independent pumas that used the UPSA each winter. We used the annual winter counts of independent pumas as one level of gauging effects of hunting and other causes of mortality at the population level. The winter counts consisted of the sum total of all known, marked (i.e., radio-collared and eartagged) pumas, non-marked pumas we captured but could not safely handle, non-marked harvested pumas and non-marked pumas reported captured and released by hunters on the UPSA. In addition, our counts included any other pumas detected by their tracks as recorded by our capture teams on the study area whose track locations and movements fit these criteria: 1) did not match known movements and locations of collared pumas, 2) exhibited diagnostic characteristics of unique individuals (e.g., tracks distinguishing sex from hind-foot plantar pad width measurements ≥52 mm classified as male, ≤50 mm classified as female; ≥2 mm difference in hind-foot plantar pad widths, 3) variation in counts of cub tracks with female tracks. We believe these counts accurately quantified puma abundance and sex and age structure. Yet, the counts may conservatively estimate population size due to a probability of us missing tracks of individuals and a probability that similar sized tracks may have been made by more than one individual. If our counts were conservative, then the design quotas we used and the observed mortality were a lower percentage of the actual population size. Puma density We used the winter population counts to calculate the density of independent pumas based on two methods of assessing area: 1) density based on the winter search areas, and 2) density based on mapped strata of winter puma habitat (December through February) as defined by a resource selection function (RSF) developed in 2015 by the CPW GIS Unit and the Avian Research, Mammals Research, and Terrestrial Sections (Colorado Parks and Wildlife 2015, Ft. Collins). The first approach made our density estimates comparable to other previous puma density estimates in western North America in the literature. The RSF approach allowed us to estimate puma density in a context of defined puma habitat that can be used by wildlife managers in Colorado. The RSF model was developed using 2,470 male and 1,603 female puma mortality locations documented in Colorado through mandatory checks. Validation of the model was assessed using 164 documented winter puma predation sites on radio-collared mule deer in Colorado, 14,793 winter GPS locations (i.e., Dec−Feb.) from 33 female and 9 male independent pumas on the Uncompahgre Plateau from 2004−2015, and 58,593 winter GPS locations from 45 female and 32 male pumas on the northern Colorado Front Range from 2007−2015. The assessment indicated that 86% of the puma locations on the Uncompahgre Plateau were within strata 3 and 4 of the model. 12 �Puma survival and mortality analysis Adult pumas participate in breeding behavior and were considered to be animals that were aged 2+ years old (Logan and Sweanor 2001:93-95, Lindzey et al. 1994). We chose 2+ years as the lower limit of age for adult pumas, because that age assignment was consistent with more complete data from New Mexico (Logan and Sweanor 2001:93-95) in which 12 known-age female pumas conceived successfully for the first time at an average age of 26.1 months old, which was close to the average age of 29.2 months (median = 27 months, range = 18−43 months) of first conception for a less-known sample of females in this study (see Puma reproduction section). Furthermore, Lindzey et al. (1994) estimated 23 months as the average age of first breeding for 6 known-age female pumas in Utah. Males in the New Mexico study were estimated to reach sexual maturity at about 24 months old. We did not have comparable data for this study. Adult puma survival and mortality was examined from data on radio-collared pumas that provided known-fate data (i.e., monitoring dates, estimated date of death, cause of death). We used program MARK (White and Burnham 1999) (accessed January 12, 2015), the known fates data type and the logit link function to model survival rates with a candidate set of models structured to investigate factors that might explain variation in survival. MARK estimated survival rates, standard errors, and 95% confidence intervals for each model. Our main interest was the effect of the hunting treatment as partitioned among the reference and treatment periods on survival, because our research focus was to examine effects of sport-hunting on a puma population. As such, our biological year spanned from November (the month that puma hunting seasons began) to October to encompass complete hunting seasons within each 12month period. Radio-location records for each adult puma were converted to monthly encounter histories. MARK estimated monthly survival rates using the modified Kaplan and Meier (1958) estimator that allowed staggered entry based on when we collared individuals and censoring of individuals if we lost contact with them (Pollock et al. 1989). We used data from year 2 of the reference (RY2) period to year 5 of the treatment period (TY5) (i.e., a 9 year span). We did not use data from reference year 1 (RY1) because we had just started the study and had collared only 7 adult pumas (3 males and 4 females). Encounter histories of individual adult pumas started on the day of capture because no pumas died as a result of capture, or the beginning of RY2 (November 1, 2005) for surviving pumas that were captured previous to that date. We censored individuals in the data if we did not receive its signal after the month of its last location. Individuals re-entered the data set if we recaptured them and fit them with a new collar. Death dates for puma were assigned to pumas with GPS collars based on the first day that GPS locations indicated that the pumas were immobile. Death dates for VHF-collared pumas were estimated based on previous live signal data and the mid-point of the span of days the puma was estimated to have died based on carcass decomposition. Death dates of hunter-killed pumas were reported by the hunters and recorded on mandatory check forms. Causes of death were categorized to known human causes (e.g., hunting, depredation control, vehicle strike, poached), to known natural causes (e.g., intraspecific strife, injury), or to unknown natural causes. Subadult pumas in our study were considered to be animals known or estimated to be 13−24 months old. This is a life stage where pumas are independent of their mothers and usually not participating in breeding behavior (Logan and Sweanor 2001:146). Subadult puma survival and mortality was estimated for all known radio-collared and ear-tagged and tattooed pumas with known fates that spanned the 12-month subadult stage. We did not know with certainty when all of the pumas in this stage became independent, therefore some of the pumas may have been dependent cubs for a period of time longer than 13 months. Furthermore, at the upper end of this stage we could not determine when most pumas were transitioning into adulthood. Encounter histories for the pumas started as marked pumas entered the stage and on the first day of capture for subadults caught and marked for the first time because no subadults died as a result of capture. All histories were converted to monthly encounter histories. Death dates were assigned to reported and observed harvest, depredation control, and vehicle strike dates. For VHF-collared pumas that died of natural causes where mortality dates were not observed, dates were 13 �estimated as the mid-point of the span of days the puma was estimated to have died based on previous radio-telemetry live signal data and carcass decomposition. The encounter histories were treated as known-fate data and entered into program MARK to model subadult puma survival rates using a candidate set of models that might explain the variation in survival rates. Cubs were dependent upon provisioning from their mothers. In our data set, these included animals between 1 to 12 months old. Survival and mortality was estimated for all radio-collared pumas in this stage. The large majority of the cubs in this data set were initially radio-collared as nurslings 1−2 months old. But we also included cubs collared at older ages, because we entered data to estimate monthly survival rates. In this way, use of data on the older cubs only added to the sample of older cubs and did not bias estimates because older cubs have a tendency to exhibit higher survival (Logan and Sweanor 2001, Ruth et al. 2011). Monthly survival rates were estimated using MARK and the known-fate data type. Encounter histories for the cubs started on the first day they were collared. Three nursling cubs that died as a result of malfunctions of the design of the expandable radiocollars early in the study in the reference period were removed from the analysis. After the collars were modified, no other cub mortalities from the collars occurred. Causes of cub deaths were assigned after dead cubs were physically examined. Dates of death were estimated as the mid-point of the span of days the puma was estimated to have died based on dates of previous radio-telemetry live signal data and carcass decomposition. The assumption that each radio-collared cub was an independent random sample (i.e., distribution of mortalities among litters is random) may be violated because we collared one-to-three cubs in litters, and often two-to-three siblings per litter, and the fates of siblings might be linked. For example, sometimes more than 1 or all cubs in a litter may die from the same proximate cause (e.g., infanticide by a male puma) or the survival of surviving cubs in a litter may be linked to death of siblings (i.e., resulting from greater individual maternal care). Violation of the independence assumption can result in unbiased survival point estimates, however, sample variances are expected to be underestimated. The data are said to be over-dispersed (Bishop et al. 2008). Therefore, we examined validity of the independence assumption in data by estimating an over dispersion parameter, c-hat (Cooch and White 2015). Covariate selection, model selection and inferences We developed sets of covariates that we hypothesized might affect survival of adult, subadult, and cub pumas. Because our research investigated effects of sport-hunting on a puma population which involved adult and subadult pumas as the huntable life stages, we developed models that included individual covariates for gender and period (i.e., reference and treatment periods). For adults, we also used a covariate for the abundance of mule deer and elk combined because these ungulates were the primary prey of puma on UPSA and because of a hypothesis that prey abundance ultimately limits puma populations (Pierce et al. 2000, Laundre et al. 2007, Logan and Sweanor 2010). Mule deer and elk abundance was indexed by modeled population estimates in December of each year that puma survival was assessed (Terrestrial Section, Colorado Parks and Wildlife, unpublished data). The mule deer and elk covariate was not used for subadult pumas because most of the subadults had dispersed from the UPSA in various directions and at varying distances when they were subject to hunting. Individual cub covariates included gender and period. We used a covariate to indicate if a cub’s dam lived or died during the 12month age stage of cub’s dependency. We used abundance of mule deer and elk combined as individual covariates to assess variation of key prey abundance to cub survival. We also explored whether birth month was associated with cub survival. We included time-varying models using year as a covariate for adults and month for subadults and cubs. We modeled puma survival for all three stages including constant (.) and additive (+) and interactive (*) combinations of covariates; for cubs we also included some quadratic terms. 14 �We evaluated the importance of candidate models in an information-theoretic approach (Burnham and Anderson 1998). For adults and subadults, we used Akaike’s Information Criterion adjusted for small sample sizes (AICc) to rank the models. We considered the models with the most support as those with the lowest AICc scores, high AICc weights (wi), and models with ∆AICc <2 as having similar support (Burnham and Anderson 2002). Survival estimates reported here were estimates in the top model and other supported models. Average monthly survival rates for adults were converted to annual survival rates (i.e., Saveragemonthly12), standard errors, and 95% confidence intervals. Survival parameters for subadults were derived estimates calculated in MARK from monthly survival rates that produced average stage (i.e., 12-month) survival rates, standard errors, and 95% confidence intervals. For cub survival, we estimated the overdispersion parameter c-hat in MARK to test for violation of the independence assumption. Likewise, we considered 1.0 <c-hat ≤1.2 as weak evidence of overdispersion as suggested by Bishop et al. (2008) and Ruth et al. (2011). We followed the method of Cooch and White (2015) and used the Tests option in program MARK to run 1000 bootstrap simulations on our cub data set in the most parameterized model we could use. We then estimated c-hat by dividing the observed c-hat in the original model estimates by the mean simulated c-hat. If the results indicated non-independence in the cub fates, then we used the Adjustments option for c-hat in MARK and entered in the estimated c-hat to adjust for the quasi-likelihood estimate (QAICc) in program MARK. We considered the models with the lowest QAICc and <2 as having the most support. Survival parameters for cubs were derived estimates calculated in MARK from monthly survival rates that produced average stage (i.e., 12-month) survival rates, standard errors, and 95% confidence intervals. Examining survival rates of adults, subadults, and cubs in the reference and treatment periods with contrasting models with and without the hunting treatment allowed us to assess changes in survival that might be associated with the treatment effect. The reference period would include all detectable natural and human causes of mortality in the population, sans hunting mortality. The treatment period would include all detectable causes of mortality, with the addition of hunting mortality. A treatment effect supported an inference that sport-hunting mortality was an important factor explaining the variation in puma survival and a factor that was to some extent additive if survival declined in association with the treatment. However, if models lacking the treatment effect received the most support, this would indicate that hunting mortality was primarily compensatory and that survival was influenced mainly by some other factors, or that statistical power was insufficient to detect a treatment effect. Reproduction Female pumas with GPS/VHF collars were monitored year round. Data from those pumas provided information on parturition, litter size, sex ratio of cubs observed in nurseries, birth intervals, and age at first breeding. Reproduction was verified by direct observations of cubs in nurseries and in direct association of adult females during capture efforts. Parturition rate, defined as the proportion of adult female pumas giving birth each year, was estimated annually from reference year 2 (RY2) through treatment year 5 (TY5) when we had ≥12 adult females in annual samples (there were only 4 adult females in RY1). Data on each adult female each year was coded with the individual identification number and as producing a litter of cubs (1) or not producing a litter (0) and whether the individual female produced a litter each year in the reference period (1) or the treatment period (2). Because adult females comprising the samples within each year were not independent of other years (i.e., some of the same females were monitored in a series of years within and among periods) mean period parturition rates were modeled by using the generalized linear mixed model procedure (PROC GLIMMIX) in SAS (Version 9.3, 2010, SAS Institute) where the period was the fixed effect and individual puma identification was the random effect. We used the binomial distribution and logit link. We also investigated all adult females that exhibited extremely constrained GPS and VHF location clusters or movements that might indicate the birth of a litter. When the cubs were 25 to 45 days 15 �old we entered the nurseries when the mothers were absent to examine the cubs and to mark them. We coded the data with each adult female identification number, the period in which the litter was produced (reference = 1, treatment = 2) and the number of cubs observed in each litter (1, 2, 3, 4). Similarly, because adult females comprising the samples within each year were not independent of other years and some occurred in both periods, we modeled period mean litter size using the mixed linear model procedure (PROC MIXED) in SAS, where period was the fixed effect and individual puma identification was the random effect. The sex ratio of cubs produced in the reference and treatment periods was compared to an expected 1:1 sex ratio by using the Goodness of fit Chi-square procedure (Zar 1984). This procedure was also used to test for a difference in the ratio of litters subject to infanticide in the reference and treatment periods and determined significance at alpha = 0.05. Sex and age structure We looked for changes in the sex and age structure of the population of independent pumas by graphically analyzing the population at the beginning of treatment years TY1 and TY5. The age structure at the beginning of TY1 represented ages of the independent pumas in this population after 5 years of protection from hunting and just before any pumas were removed in the first treatment year. The age structure at the beginning of TY5 represented the ages of independent pumas in the population after 4 years of hunting (TY1−TY4) and other causes of mortality operated on the population up to the start of TY5. PRELIMINARY FINDINGS Puma capture From December 2, 2004 to October 30, 2014 we captured about 256 individual pumas a total of 440 times on the UPSA. None of the adult or subadult pumas died from capture procedures. However, 3 cubs died as a result of premature expansions of the radiocollars (indicated previously in Field Methods) and 1 cub was killed by our tracking dogs. We individually marked 226 pumas: 109 in the reference period and 115 in the treatment period. The number of radio-collared pumas monitored each year ranged from 16 to 56 and averaged 40. Marked pumas provided known-fate data on 75 adults, 75 subadults, and 118 cubs. About 30 individuals were captured with dogs, but were not handled due to dangerous positions in trees. Of those pumas not handled, 11 were captured in the reference period and 19 were captured in the treatment period. Six of 11 pumas not handled in the reference period were associated with marked family members (i.e., mothers or siblings). Similarly, 8 of 19 pumas not handled in the treatment period were associated with marked family members. By the end of the study, we could account for the fates of all of the radio-collared adults (i.e., we determined they either survived or died), including those with failed radiocollars, except for one adult male. In addition to the pumas captured by our research team during the treatment period, puma hunters captured and killed a total of 35 pumas, including 8 adult females, 16 adult males, 3 subadult females, and 8 subadult males. Puma hunters also reported having captured and released 30 independent pumas, with their reported gender identification of 19 females and 11 males. The sex ratio of independent pumas captured and released by hunters (1.7 females:1 male) fell into the range of our winter counts of independent pumas (1.2 females: 1 male to 2.8 females: 1 male; Table 2) in the treatment period. The sex ratio of pumas killed by hunters (1 female: 2.2 males) reflected a hunter selection bias toward males reported in hunter surveys during the treatment period (K. Logan, unpublished data). Reliability of population count methods The camera grid survey by K. Yeager (2016) in TY4 spanned 102 days from December 2012 to March 2013. The survey partially overlapped with 3 months of our winter capture efforts which spanned from January 1 to April 18, 2013. Eleven GPS and VHF collared pumas were known to use the survey grid for varying amounts of time, including 7 adult females, 1 subadult female, 2 adult males, and 1 16 �subadult male. During the survey cameras acquired 18 photographs of pumas visiting the sampling sites, and all 18 of the photographs depicted GPS or VHF collared pumas. The photographed pumas included 1 subadult female and 1 subadult male that we captured and marked for the first time in January 2013, before cameras subsequently detected them 5 and 3 times each, respectively. An adult male that we captured and marked for the first time February 14, 2013 was not detected by the cameras. Of the 11 collared pumas known to use the grid, 7 were photographed 1 to 5 times each, including 5 adult females, 1 subadult female, and 1 subadult male. The results of the camera grid survey indicated that our field methods could produce reliable counts of independent pumas for our index to abundance because: 1) no non-collared pumas were detected during the survey; and 2) we detected, captured and marked 3 new pumas before the cameras detected any of them. Furthermore, the ratio of marked-to-non-marked independent pumas killed by hunters during the treatment period before our research teams inserted to search for non-marked pumas was 19:16. The hunters’ survey responses indicated that marks on pumas did not influence their decisions to harvest a puma (K. Logan, unpublished data). A preponderance of this evidence indicated that our field operations thoroughly accounted for the abundance of independent pumas on the UPSA and enabled us to sample a majority of the independent pumas for estimates of survival, agent-specific mortality, reproduction and age structure. Puma population counts The number of days we spent each winter and early spring searching for pumas with dogs was similar in each period (reference mean = 77.2, SD = 4.0, range 71−82; treatment mean = 79, SD = 4.8, range 74−86). We believe we had a thorough knowledge of the study area and search routes for reliable counts of the winter population of independent pumas on the study area by reference year 4 (RY4) and throughout the treatment period. However, in RY5 a Colorado state government-mandated hiring freeze (in response to economic recession) resulted in insufficient personnel for thorough searches of the study area for a reliable winter count of independent pumas. Therefore, the count in RY5 (n = 37) was probably biased low, but was still larger than RY4 (n = 33). Recognizing this probable bias, we modeled the population by using the count data in winter 2007−08 (RY4) and previous reference period parameter estimates (see Appendix II, Table AI.3), which projected an expected 45 independent pumas in RY5 (Table 2, Fig. 3). The hunting treatment The hunting treatment on the puma population during TY1−TY3 consisted of a designed 15% harvest rate of the independent pumas in UPSA. The quota to represent a 15% harvest was 8 pumas based on a model projected 53 independent pumas expected in TY1 (Appendix II). The model projected 53 independent pumas turned out to be very close to the 56 pumas we actually counted in TY1. The graphed results of the treatment period counts indicated a non-ambiguous monotonic decline in the number of independent pumas from TY1−TY4 (Fig. 3). Clearly, the population declined when subjected to a 15% design harvest for the UPSA GMU (actual average 16.1% harvest of independent pumas, range = 15.4−16.7%) from TY1−TY3 and when other human and natural causes of mortality operated on the population. Of the 25 independent pumas killed by hunters in TY1−TY3, 32% of them were females with 20% adult females. Therefore, a 15% design harvest of independent pumas comprised of 35−45% females was not supported for managing toward a stable or increasing puma population in GMU-based management when other human and natural causes of mortality were operating. The actual average harvest rate of 16.1% (Table 3A) of independent pumas from TY1−TY3 exceeded the 15% design harvest because the quota was exceeded by one puma in TY1, and as the population declined the quota exceeded the 15% harvest in TY2 and TY3. Because the population declined with an average 16.1% actual harvest rate, we wanted to find a harvest rate of independent pumas that might be sustainable with other human and natural causes of mortality operating on the 17 �population. Therefore, in TY4 and TY5 the quota was reduced to 5 pumas constituting 11−12% design harvest of independent pumas. However, additional independent pumas died of other causes during the treatment period hunting seasons (i.e., Nov.−Mar., Table 3B). Additional pumas deaths each hunting season ranged from 0−2 each season. All of the deaths were adult females. Adding these puma deaths to the harvest the total mortality as a percentage of the winter counts of independent pumas ranged from 16.1−20.8% and averaged 18.7% during TY1−TY3 when the puma population declined. In TY4 and TY5 the total mortality of independent pumas was 14.3% and 13.6%, respectively (Table 3B). In addition, hunting in surrounding GMUs contributed to the decline in the abundance of independent pumas on UPSA. Ten radio-collared independent pumas (2 adult females, 7 adult males, 1 subadult female) included in treatment year winter counts were killed by hunters in adjoining GMUs 61 North, 62 North, 65, and 70 because those pumas had home ranges that extended beyond the boundaries of UPSA. Two of the adult radio-collared males were actually trailed by hunter’s dogs off of UPSA and were caught and killed in adjacent GMUs 65 and 70. In addition, as reported by the hunters, 2 noncollared adult males were trailed with dogs off of the study area and killed in adjacent GMUs 62 North and 70. The number of study animals killed by hunters outside of UPSA ranged from 0 to 4 and averaged 2 each year from TY1−TY5, and those pumas were counted in the hunting quota in the adjoining GMUs, not UPSA. Including these deaths off the study area, the percent of hunting kill from TY1−TY5 ranged from 11.4%−25% (average = 18.2%) of independent pumas in winter counts on the UPSA. The actual hunter-kill of the number of independent pumas during TY1−TY3 ranged from 17.3−25% (average = 21.8%), and was associated with the population decline phase. During TY4−TY5 the actual hunter-kill was 11.4−19.0% (average = 15.2%), and was associated with the low population phase. Therefore, hunting mortality in the adjacent GMUs added to the overall mortality impacting the UPSA puma population. This indicated that GMU-based puma management must consider that GMUs do not represent closed puma population segments and that harvest in adjacent GMUs can confound the expected management results for any particular GMU. Changes in puma abundance The population of independent pumas increased by about 70% (i.e., 33 in RY4 to 56 in TY1) during the reference period without hunting as a mortality factor (Fig. 3). We counted the highest number of independent pumas during winter of TY1 which was preceded by 5 years without hunting. The increasing puma population during the reference period was an indication that previous hunting mortality was probably a major factor that reduced the population. In the treatment period, the population of independent pumas declined from a total count of 56 in TY1 to a low of 42 in TY4, a 25% decline after three hunting seasons (Fig. 3). The abundance of adult females declined 23.3% by TY5. Adult males declined 55% by TY3 and TY4, and were 50% lower by TY5 (Table 2, Fig. 4). The percentage of independent females in the harvest TY1−TY5 was 31.6%; comprised of 23% adult females and 8.6% subadult females. The remainder of the harvest was comprised of adult males (45.7%) and subadult males (22.9%). After we reduced the quota to 5 for TY4 and TY5, the decline in abundance of independent pumas stopped, stayed in a low phase, and may have slightly increased. An increase in subadult males also contributed to a slight increase in independent pumas by TY5 (Table 2, Fig. 4). Considering the total independent puma mortality that occurred in treatment period hunting seasons (Table 3B), the percentage of independent females in the mortality TY1−TY5 was 41.5%; comprised of 34.1% adult females and 7.3% subadult females. In TY3 and TY4 the percent independent and adult females in the total mortality was 50% and 33.3%, respectively (Tables 3A, 3B) and were associated with the population low phase. 18 �These results were evidence that a designed hunting off-take of 15% of independent pumas plus other human- and natural-caused mortality could result in a decline in abundance of adult and independent pumas at the GMU level. Furthermore, data on the puma population in the UPSA GMU indicated that the puma population responded to our hunting closure and treatment, and thus GMU-based hunting management could be an effective way of managing localized puma population segments. Mortality in marked pumas The regulations implemented to eliminate sport-hunting as a mortality factor on pumas on the UPSA in the reference period were effective. The puma population responded to this regulation change on the UPSA GMU by increasing by about 70% (previously). Of the 32 (21 females, 11 males) adult radio-collared pumas we monitored, 7 adult pumas died but none from hunting (Table 4A). Causes of death were attributed to: 5 natural causes (4 intra-specific strife, 1 unknown), 1 vehicle strike, and 1 depredation control. Of the 22 subadults (8 females, 14 males) providing known-fate data in the reference period, 3 died. One male that had moved off the study area was killed by a hunter that did not see the tags. The other causes of death in subadults were 1 natural cause (trampled by elk) and 1 vehicle strike. Of 55 radio-collared cubs (28 females, 27 males) monitored in the reference period, 16 died. Causes included: 13 infanticide, 1 predation, 1 unknown natural, and 1 vehicle strike. In the reference period natural causes dominated deaths of adults and cubs (71.3% and 93.8%, respectively), but 2 of 3 subadult deaths were from human causes. Sport-hunting was the most important cause of death for independent pumas during the treatment period (Table 4B). Of the 61 adults (39 females, 22 males) we radio-monitored during the period, 37 died. Hunting caused 56.8% of adult deaths (n = 21: 14 males, 7 females), followed by natural causes (27%; n = 10: 7 unknown with 6 probably disease-related and 1 due to starvation with senescence, 3 intra-specific strife), and other human causes (16.2%; n = 6: 3 vehicle strike, 2 depredation control, and 1 illegal kill). Of the 53 subadults (19 females, 34 males) providing known-fate data in the treatment period, 20 died. Hunting caused 55% of the subadult deaths (n = 11: 9 males, 2 females). Natural mortality followed in importance with 25% (n = 5: 3 intra-specific strife, 2 other natural), then closely by 20% other human causes of death (n = 4: 3 depredation control, 1 vehicle strike). Combining adult and subadult puma deaths in the treatment period, human causes were 73.7 % (i.e., 42/57*100), of which hunting caused 76.2% (i.e., 32/42*100) and other human causes comprised 23.8% (10/42*100). Of the 63 radio-collared cubs (27 females, 36 males) monitored in the treatment period, 27 died (Table 4B). Mortality causes in the cubs included: 9 infanticide, 4 other natural, 2 vehicle strike, 3 depredation control, and 9 starvation. The 9 cubs starved after the deaths of 5 mothers due to: hunting (2 mothers involving 3 cubs), depredation control (1 mother with 3 cubs), and natural causes (2 mothers involving 3 cubs). Natural mortality comprised the majority of cubs deaths (15/27*100 = 55.6%). But, human-caused cub deaths in the treatment period increased to 44.4% (12/27*100 = 44.4%) from 6.2% in the reference period. There was no apparent compensation to hunting mortality by a reduction of other causes of death for most independent puma categories in the treatment period. Hunting-caused deaths were additive mortality to adult male pumas, because percentages of other causes of death in the treatment period were similar to the reference period (Figs. 5A, 5B). Hunting and other human-caused deaths added to natural mortality in adult females in the treatment period, which was higher than in the reference period (Figs. 5A, 5B). Similarly, hunting and other human-caused deaths to subadult male pumas appeared to be additive mortality (Figs. 6A, 6B). But, the few hunting deaths to subadult females appeared to be partially compensatory, as there was a decline in the percentage of non-hunting deaths during the treatment period (Figs. 6A, 6B). The additive quality of hunting and other human-caused deaths was further reflected in significantly lower adult and subadult male survival rates and a decline in abundance of independent and adult pumas on UPSA during the treatment period. As we previously indicated, the abundance of 19 �independent pumas declined 25% by TY4, adult females declined 23% by TY5, and adult males declined 55% by TY3 and TY4 (Fig. 4). Infanticide is thought to be sexually selected and a means by which adult male pumas compete for reproductive success (Hrdy 1979, Logan and Sweanor 2010). It has been theorized that periods of male territory instability contribute to reduced cub survival via increased infanticide as immigrant males and shifting neighbor territorial males move into vacated territories (Logan and Sweanor 2001, Ruth et al. 2011). Conditions on UPSA were not sufficient for testing this hypothesis, because there was not an adequate duration of time for adult male abundance and their territories to stabilize. The population of independent pumas increased at least by 70% during the reference period. During this growth period, the abundance of adult males increased as a result of high survival. Theoretically, we would expect competition among the males to increase as well. The population of independent pumas declined by 21.4% during the hunting treatment period and included a 50−55% reduction in adult males. Although territory stability was put into flux again due to high mortality in adult males during the treatment period, there also were 55% less adult males in the population to compete for mates by TY3 and TY4 and 50% less by TY5. These conditions might explain why during the reference period that 8 of 32 litters involving 13 cubs of the radio-collared litters we monitored were impacted by infanticide, and during the treatment period 5 of 45 litters involving 9 cubs of the radio-collared litters we monitored were impacted by infanticide. The ratio of litters subject to infanticide in the reference period was significantly different than the ratio of litters subject to infanticide in the treatment period (χ2 = 4.571, 1 d.f., 0.025 < P < 0.05). In addition to deaths revealed by the radio-collared cubs, we observed deaths of 4 entire litters on the day we entered nurseries to examine cubs for the first time. These cubs were not part of the radiocollared cub population used to model or estimate cub survival (see below in Puma Survival, Cubs). One litter of 3 nursling cubs starved to death in the reference period after the mother was killed for depredation control. In the treatment period, we observed that three entire litters died; 2 litters (1 with 2 cubs and 1 litter with ≥1 cubs) died of infanticide, and the third litter (with ≥1 cub) died due to black bear predation. Mortality in independent marked and non-marked pumas We calculated the percentage of detected marked and non-marked independent puma deaths on UPSA each 12-month study year RY4−TY5 attributed to natural-, other human-, total human-, and hunting-causes, and the portion of total human-caused deaths attributed to hunting on UPSA (Table 5). In this way, the percentages for each type of death could relate to the trends in the population for each period (Figs. 3, 4). Total detected deaths of independent pumas on UPSA for reference years RY4 and RY5 ranged from 3−5 independent pumas each year. Natural causes of death ranged from 33.3−60% and averaged 46.7% per year. Other human-caused deaths ranged from 40−66.7% and averaged 53.4% per year. There were no hunting deaths in RY4−RY5 by design. During the treatment years TY1−TY5, total detected deaths ranged from 9−17 per year. Natural causes of death ranged from 11.1−25% and averaged 18.6% per year. Other human-caused deaths (excluding hunting) ranged from 8.3−36.4% and averaged 24.6% per year. Hunting-caused deaths ranged from 45.5−69.2% and averaged 56.8% per year. Total human-caused deaths (with hunting) ranged from 75−88.9% and averaged 81.4% per year. Of the total human-caused deaths, hunting ranged from 55.6−88.9% and averaged 70.6%. Puma density The density of independent pumas based on our defined winter search area (1,701 km2, 657 mi.2) ranged from 1.9/100 km2 in RY4 to a high of 3.3/100 km2 in TY1 (Table 6). During the treatment period, density declined from 3.3/100 km2 to a low of 2.5/100 km2 in TY4 and increased only slightly in RY5 to 2.6/100 km2. For broader management considerations in Colorado (see Analytical Methods, Puma density, previously), the density of independent pumas based on puma habitat as defined by the Colorado puma RSF model was based on two strata groupings (Table 6). In the first, we calculated density for strata 20 �3 and 4 combined (1,636 km2). Strata 3 and 4 represented 0.50 to 0.749 and 0.75 to 0.99 probability of puma presence, respectively. These density estimates were very similar to our estimates based on winter search area because of the similarity in the areas (i.e., total km2) where pumas were concentrated with their ungulate prey in winter. In the second grouping of RSF strata, we included the area of stratum 2, which represented 0.25 to 0.49 probability of puma presence. The combined area of strata 2, 3, and 4 was 2,426 km2, and therefore produced lower density estimates (Table 6). We did not include stratum 1 because it was comprised primarily of non-puma habitat, including towns, reservoirs, farmlands, and high elevation areas with deep snow that probably could not sustain pumas in winter. Expected presence of pumas in stratum 1 was 0.001 to 0.249. Puma harvest density We assumed that Colorado RSF strata 3 and 4 combined represented mule deer, elk, and puma winter range, which was consistent with our direct observations on the UPSA. Puma harvest densities on the UPSA during the puma population decline (TY1−TY3) ranged from 4.9−5.5 and averaged 5.1 independent pumas/1,000 km2 of RSF strata 3 and 4 combined (Table 7A). During the entire treatment period TY1−TY5, which included the population decline and low phases, harvest density ranged from 3.1−5.5 and averaged 4.3 pumas/1,000 km2. Considering that the UPSA puma population was partially affected by hunting in surrounding GMUs, we calculated harvest densities for those same RSF strata in GMUs adjacent to the UPSA, including 61 North, 62 North, 65, and 70 during the treatment period (CPW puma harvest data, Denver, CO). Generally, harvest densities all five years in each of the GMUs, except for 62 North, was within or greater than the range of harvest densities on the UPSA during the population decline phase during TY1−TY3 (Table 7A). In all four of the adjacent GMUs combined, harvest densities ranged from 3.2−5.4 and averaged 4.8 pumas/1,000 km2, and three of the five years exhibited harvest densities within the range of UPSA harvest densities during the decline phase TY1−TY3 (Table 7A). The percentage of adult females in the combined harvest for all 4 of the GMUs during TY1−TY5 ranged from 10.3−35.3% and averaged 18%. Assuming that the puma harvest densities and the percent of adult females in the harvest (i.e., 20%) with other natural and human causes of death on the UPSA during TY1−TY3 following a 5-year hiatus from hunting mortality relate to population decline, then one interpretation is that it is probable that puma abundance in the area of these adjacent GMUs declined or was in a low phase if they also experienced similar rates of natural and other human causes of mortality (Fig. 7A). We also presented puma harvest densities on the UPSA and the same adjacent GMUs based on RSF strata 2, 3, and 4 areas for consideration by managers (Table 7B, Fig. 7B). Our interpretation of these results was similar to results of the previous analysis using RSF strata 3 and 4, except that four of the five years exhibited harvest densities within the range of UPSA harvest densities during the decline phase TY1−TY3. Puma age structure The age structure of independent male pumas we examined in the UPSA declined from TY1 to TY5 (Fig. 8). After 5 years of no hunting, the younger and up to middle aged (i.e., 1−5 years old) pumas comprised a majority (55.3%) of the population and with both adult females and males being represented up to the oldest ages (i.e., >5−10+ years old). After 5 years of hunting, adult males >5 years old were eliminated from the age structure. Previously at TY1 adult males >5−10+ years old had comprised 57% of the adult male (>2 yr. old) population. There was an increase in the abundance of subadult males by TY5, but recruitment of males into the adult stage during the treatment period was not sufficient to numerically replace the adult males that died. There was no noticeable change in the female age structure in the treatment period although there was a 23.3% decline in the abundance of adult females from TY1−TY5 (previously). The changing age structure and abundance of independent pumas at each age could be explained by patterns of puma survival associated with the hunting treatment of the population. Puma survival Adults 21 �Our adult survival sample included 75 radio-collared individuals, with 32 monitored in the reference period and 61 monitored in the treatment period. The most parsimonious survival model indicated a treatment effect on gender with gender interacting with period (i.e.,{S(gender*period)}; Table 8). The evidence ratio for this model using AICc weights (wi) indicated it had only 1.3 times the support of the second-ranked model with gender interacting with combined mule deer and elk abundance. These two top models explained over 86% of the variation in adult puma survival. However, when we graphed the estimates of deer and elk abundance with model generated estimates of annual adult female and male puma survival rates (converted from monthly rates), a serendipitous relationship was revealed (Fig. 9). That relationship was clearly and directly influenced by our manipulation of the puma population which resulted in the high adult puma survival in the reference period and the decline in adult male puma survival during the treatment period, mainly caused by our hunting treatment. Mule deer and elk abundance generally declined during the reference period and the first 3 years of the treatment period. But then ungulate abundance seemed to stabilize or increase slightly in TY4 and TY5 at the same time we purposely reduced the harvest rate of pumas which also was reflected in the adult survival rate estimates. Therefore, we believe the model of gender interacting with deer and elk abundance was primarily coincidental with our study design and attendant manipulation rather than a biological explanation for variation in adult puma survival. We did not observe malnutrition or starvation as a common malady in adult pumas we examined. Only one adult female puma died apparently due to starvation associated with senescence; she was about 11 years old when she died in TY3 of the treatment period. Considering this, we reduced the model set by excluding the models with deer and elk abundance as a covariate (Table 9). The reduced set of adult puma survival models indicated gender interacting with period as the most parsiminious model (Table 9). The evidence ratio using AICc weights (wi) indicated very strong support for the top model with 10.10 times the support of the second model with gender in an additive effect with period. Moreover, >4 ∆AICc separated the top model from the second-ranked model. Thus, the remainder of the models in the 6 model candidate set had weak to no support. Adult male survival declined significantly (i.e., 95% confidence intervals did not overlap) from 0.96 (95% CI = 0.746−0.994) in the reference period to 0.40 (95% CI = 0.223−0.569) in the treatment period. The decline in adult male survival was also indicated by the 50% decline in the abundance of adult males in the treatment period (Fig. 4). The adult female survival estimate was higher for the reference period (i.e., 0.86; 95% CI = 0.715−0.935) than the treatment period (i.e., 0.74; 95% CI = 0.632−0.823), but the difference was not statistically significant (i.e., 95% confidence intervals overlapped; Table 10). However, the 23.3% decline in the abundance of adult females in the population during the treatment period indicated that lower survival was biologically significant, as we did not observe radio-collared adult females emigrating from the population and female recruitment was not sufficient to replace the adult females that died (Fig. 4). Subadults Our subadult survival sample included 75 individuals with known-fates: 22 in the reference period and 53 in the treatment period. For subadult pumas the modeling results indicated period as an important factor influencing survival as indicated by the two top-ranked models <2 ∆AICc points (Table 11), which together accounted for 77% of the model weights (wi). Evidence ratios using AICc weights (wi) indicated the top-ranked model of gender interacting with period had weak support for the best model with 1.7 times the support of the second-ranked model with period as the main explanatory factor. Subadult males were negatively affected by hunting-caused mortality, similar to adult males. Subadult male survival declined significantly from 0.92 (95% CI = 0.611−0.989) in the reference period to 0.43 (95% CI = 0.265−0.607) in the treatment period. However, subadult female survival estimates increased in those respective periods from 0.63 (95% CI = 0.232−0.906) to 0.70 (95% CI = 0.425−0.883), but the difference was not statistically significant (Table 10). The second-ranked model Speriod (with genders combined) had reasonable support as the best model (i.e., 1.0567 ∆AICc, wi=0.28631) for explaining variation in subadult survival which declined from 22 �0.84 (95% CI = 0.599−0.946) in the reference period to 0.52 (95% CI = 0.37−0.659) in the treatment period, but the difference was not statistically significant (Table 12). Recruitment of subadults was not sufficient to offset the mortalities of adult males and females, as the abundance of both declined during the treatment period (previously). Three models not in Table 11 included: [{S(month*period)}, {S(month*gender)}, {S(month*period*gender)}] in which MARK could not estimate all the parameters because of the sparseness of the monthly data. Cubs Our cub survival sample included 118 radio-collared cubs: 55 cubs from 32 litters in the reference period, and 63 cubs from 45 litters in the treatment period. We could not use the most parameterized model {S(month*period*gender)} or model {S(month*period)} to estimate c-hat because MARK could not estimate all of the parameters due to the sparseness of the monthly data. Therefore, we used the next most parameterized model {S(period*gender)}. The estimated c-hat for kitten survival was 1.545, indicating that the fates of siblings were not independent. We documented numerous occasions of this phenomenon. In the reference period 9 radio-collared siblings in 4 litters died at the same time due to infanticide. In addition 3 non-collared cubs in 1 litter starved after the dam was killed for depredation control. In the treatment period 19 radio-collared siblings in 8 litters died at the same time due to a variety of causes, including: depredation control (3 cubs in 1 litter), vehicle strike (2 cubs in 1 litter), infanticide (7 cubs in 3 litters), and starvation (7 cubs in 3 litters). Of the 3 litters that starved, the dams died due to a natural cause, hunting, and depredation control. In addition, 2 non-collared cubs in 1 litter died from infanticide. Modeling results indicated three models with <2 QAICc with the covariate for dam’s status (i.e., damld) that accounted for 66% of the model weights (wi) (Table 13). These models indicated that whether the dam lived or died [{S(damld)}] was the single most important factor affecting cub survival. Evidence ratios using QAICc weights (wi) indicated the top model with the covariate damld alone had 2.5 times the support of the second-ranked model [{S(gender+damld)}] and 2.7 times the support of the third-ranked model {{S(period+damld)}]. However, the second- and third-ranked models provided additional information about how cub survival varied with gender and period (Table 14). Cub survival considering dam status alone was 0.45 (95% CI = 0.34−0.57) for the entire study duration. Considering gender additive to dam status, female cub survival was 0.42 (95% CI = 0.27−0.58) and similar to male cub survival of 0.48 (95% CI = 0.33−0.63). Considering period additive to dam status, there was no statistically significant change in cub survival in the reference period (S = 0.47; 95% CI = 0.31−0.64) and the treatment period (S = 0.43; 95% CI = 0.28−0.59). These top models and data on fates of dams indicated that any factors that contributed to a dam’s death while cubs were dependent resulted in lower cub survival. There was no support for period alone explaining variation in cub survival (∆QAICc = 5.8). Likewise, there was no support for mule deer and elk abundance influencing cub survival (∆QAICc >4.5 for all models with the covariate). Starvation in cubs that we observed occurred because their mothers were no longer alive to provision them, not because of a lack of deer and elk in the environment. We estimated the probability of cub survival to the adult stage (i.e., 2 yr. old) by using survival estimates from the set of top-ranked survival models (<2 QAICc) from this study (see Tables 11−14). For cubs we used model S(period+damld) (third-ranked model) and for subadults we used model S(period) (second-ranked model) for both genders. The estimated probability of cub survival to the adult stage in the reference period was 0.39 (1*0.471cubS*0.837subadultS = 0.394). The estimated probability of cub survival to the adult stage in the treatment period was 0.22 (1*0.431cubS*0.515subadultS = 0.222). Puma reproduction Adult female pumas on the Uncompahgre Plateau produced litters of cubs between the months of March to September, spanning the early spring to early fall seasons. Data on 66 birth dates revealed that births increased rapidly in May and June, peaked in July, followed by a slight decline in August and a 23 �rapid decline in September. No live births were detected in the months of October, November, December, January, or February (Fig. 10). Considering a 92-day gestation period (Anderson 1983, Logan and Sweanor 2001, this study), the distribution of birth months indicated that puma breeding activity would have spanned the months of December to June, with a rapid increase in February and peaking March through May. We estimated gestation for 17 litters by 13 females based on GPS- or VHF- location data of females with prospective sires that produced minimum and maximum estimates. Gestation lengths averaged 90.4min−91.8max days (SDmin = 2.6, 95% CImin = 89.1−91.6; SDmax = 1.0, 95% CImax = 90.7−92.8). Birth intervals for 18 adult females that produced 33 litters averaged 18.5 months (SD = 5.9, 95% CI = 16.4−24.4). We estimated the age of 13 adult females when they produced their first litters that we were able to observe based on estimated ages (n = 11) or known-ages (n = 2) of pumas at previous captures and nipple characteristics (i.e., tiny, pink or white color) and associated reproduction histories. As a practical matter, we probably would not have observed births of litters if the cubs died before the 2−4 weeks it took us to confirm the production of litters by observation. Therefore, some of the ages we recorded may be biased high. The median age at first litter was 30 months (range = 21−48); the average was 32.2 months (SD = 8.4, 95% CI = 27.6−36.8). This meant those females conceived at the median age of 27 months (range = 18−45); the average age was 29.2 months (SD = 8.4, 95% CI = 24.6−33.8) assuming an average 92 day gestation period. Litter sizes were determined for 26 litters produced by 14 females in the reference period where we were reasonably certain we counted all the cubs in nurseries when the cubs were 26 to 42 days old. Likewise, we determined litter sizes for 21 litters of 16 females in the treatment period for nursling cubs 25 to 45 days old. Average litter sizes for each period estimated using linear mixed models were 2.76 (SE = 0.1806, 95% CI = 2.41−3.12) for the reference period and 2.38 (SE = 0.1972, 95% CI = 1.99−2.76) for the treatment period. Change in the average litter sizes in the two periods was small and not statistically significant because the 95% confidence intervals on the slope for period included zero. The male:female sex ratio for 72 nursling cubs in the reference period was 41:31 and was not significantly different from an expected 1:1 ratio (χ2 = 1.388, 1 d.f., 0.10 < P < 0.25). Similarly, the 27:22 sex ratio for 49 nurslings in the treatment period was not significantly different from an expected 1:1 ratio (χ2 = 0.51, 1 d.f., 0.25 < P < 0.50). With all the cubs pooled from both periods, the sex ratio 68:53 was not significantly different from parity (χc2 = 1.860, 1 d.f., 0.10 < P < 0.25). Parturition rate, defined here as the proportion of adult female pumas giving birth each year, was determined for reference period years RY2−RY5 when we radio-monitored 12 to 13 individual females per year and treatment period years TY1−TY5 when we radio-monitored 15 to 17 individual females per year. Average parturition rate per year for each period estimated using generalized linear mixed models was 0.63 (SE = 0.068, 95% CI = 0.49−0.75) for the reference period and 0.48 (SE = 0.057, 95% CI 0.37−0.59) for the treatment period. Although the average parturition rate in the reference period was higher than in the treatment period, the change was not statistically significant because the 95% confidence interval for the slope included zero. Conclusions The general purpose of this research was to investigate potential effects of sport-hunting on a puma population in Colorado, because the puma population in the state is managed and conserved primarily by regulating puma sport-hunting with kill quotas. More specifically, the purpose was to test puma management assumptions used by CPW to manage puma population segments with sport-hunting. The preponderance of evidence from the data indicated that the addition of hunting as a mortality factor limited the puma population. In the context of the GMU-based puma management structure in Colorado, 24 �elimination of puma hunting in a reference period resulted in at least a 70% increase in the abundance of independent pumas and high adult and subadult male survival rates. However, in a treatment period a 15% design harvest including 32% females in the harvest in addition to other human-caused and natural mortality during TY1−TY3 resulted in a 25% decline in independent puma abundance by TY4 in the UPSA GMU. Declines in the abundance of independent pumas overall and adult female and male pumas in annual winter counts, survival of adult and subadult male pumas, and male age structure all were biologically significant changes associated with sport-hunting mortality with other human and natural causes of mortality operating on the puma population. There was no evidence of biologically significant compensatory reproduction or immigration with recruitment to offset the mortalities in adult pumas during the treatment period. Hunting-caused mortality to independent pumas on UPSA in the treatment period was the single-most important cause of death ranging from 11.4−16.7% of independent pumas per winter. Additional hunting mortality of independent pumas that ranged onto adjacent GMUs contributed to a decline and low phase in abundance of UPSA pumas from TY1−TY5, with total hunting mortality (i.e., occurring on and around UPSA) ranging from 11.4−25% (average = 18.2%) of independent pumas on the UPSA each winter. Responses of the UPSA puma population to our designed manipulations in both the reference and treatment periods indicated that puma population segments at the GMU-size represented by UPSA can be managed for specific population segment objectives as long as managers understand the potential effects of sport-hunting, other human caused-, and natural mortality on puma abundance on the GMU of interest and the surrounding GMUs. Management Implications 1) In the GMU-based puma management structure in Colorado, a design hunting-caused mortality of ≥15% of independent pumas with an average of ≥20% adult (i.e., 2 years old and older) females in the harvest, and with other human and natural causes of mortality ranging from 16.1−20.8% each hunting season operating on a relatively high density puma population can cause 25% reduction in the independent puma population after 3 hunting seasons. Hunting harvest is the only component of puma mortality that can be managed by regulation to determine population trend, as the other causes of mortality (natural and other human causes) occur randomly and vary in amount annually. Moreover, natural causes of death are rarely observed by managers, but other human causes (e.g., depredation control kills, some vehicle strikes) are observed and recorded and occur throughout any year. In our study, other human causes of death comprised about 24−29% of the total human-caused mortality in independent pumas (i.e., 74−81% of all deaths) with 71−76% of deaths (depending on accounting method) due to hunting in the treatment period on the UPSA GMU when the puma population declined to a low phase. 2) It can take up to 5 years for a puma population previously reduced to a low density to recover to a relatively high density after hunting has been eliminated. 3) Design hunting harvests of up to 11−12% of independent pumas with an average of <20% adult (i.e., 2 years old and older) females in the harvest and with total mortality to independent pumas ranging from 13.6−14.3% is expected to result in a stable, possibly increasing population, considering that other human- and natural-causes of mortality will operate on the population. 4) Total mortality metrics in each hunting season treatment year associated with the decline in the independent puma population on UPSA during TY1−TY3 averaged 18.7% (range = 16.1−20.8%). During TY4 and TY5 when the population decline stopped and possibly increased slightly total mortality to independent pumas were 14.3% and 13.6%, respectively. If total mortality metrics are used by managers to set kill quotas, all detected mortalities of independent pumas occurring during hunting seasons should be discounted from quotas. 5) Other metrics that might be useful to managers for gauging potential effects of hunting mortality include harvest density and percentage of adult females in the harvest. On the UPSA, we observed a declining abundance of independent pumas when harvest density was 4.9−5.5 independent pumas/1000 km2 of RSF strata 3 and 4 combined and the percentage of adult females 25 �6) 7) 8) 9) in the harvest was 20%. For RSF strata 2, 3, & 4 combined, the abundance of independent pumas declined when the harvest density was 3.3−3.7 independent pumas/1000 km2. Puma population segment management objectives can be achieved at the GMU level as exhibited by the puma population responses to our research manipulations on UPSA GMU-size study area (2,996 km2, 1,157 mi.2). Puma hunting harvest can affect puma abundance in the particular GMUs of interest and adjacent GMUs where puma home ranges overlap GMU boundaries because GMUs in connected puma habitat are not closed puma populations. Puma harvest management structure which includes provisions for reducing puma population segments to achieve specified management objectives (e.g., reduce predation on livestock or mule deer populations) should also provide for puma population segments managed with conservative harvest rates to allow for stable or increasing puma population segments (i.e., source-sink management) to ensure overall puma population resiliency because of all the unknowns and uncertainties associated with puma population management, including puma abundance, and effects of harvest and other human and natural causes of mortality in GMUs. Management experiments and research involving puma population segments should consider potential population effects of historical puma hunting on and around the study areas. When experimental designs require reference conditions, human-caused mortality to pumas should be limited or eliminated if possible. Reference areas could be part of a state-wide puma management design to provide reference conditions for scientific research and management experiments as science on natural phenomena is an ongoing process. Literature Cited Anderson, A. E. 1983. A critical review of literature on puma (Felis concolor). Special Report No. 54, Colorado Division of Wildlife, Ft. Collins. Anderson, A. E., D. C. Bowden, and D. M. Kattner. 1992. The puma on the Uncompahgre Plateau, Colorado. Colorado Division of Wildlife Technical Publication No. 40. Anderson, C. R., and F. G. Lindzey. 2005. Experimental evaluation of population trend and harvest composition in a Wyoming cougar population. Wildlife Society Bulletin 33:179-188. Ashman, D.L., G. C. Christensen, M. L. Hess, G. K. Tsukamoto, and M. S. Wickersham. 1983. The mountain lion in Nevada. Nevada Department of Wildlife, Federal Aid in Wildlife Restoration Project W-48-15, Final Report. Bauer, J. W., K. A. Logan, L. L. Sweanor, and W. M. Boyce. 2005. Scavenging behavior in puma. The Southwestern Naturalist 50:466-471. Beausoleil, R. A., G. M. Koehler, B. T. Maletzke, B. N. Kertson, and R. B. Wielgus. 2013. Research to regulation: cougar social behavior as a guide for management. Wildlife Society Bulletin 37:680688. Burnham, K. P., and D. R. Anderson. 1998. Model selection and inference: a practical informationtheoretic approach. Springer-Verlag, New York, New York, USA. Choate, D. M., M. L. Wolfe, and D. C. Stoner. 2006. Evaluation of cougar population estimators in Utah. Wildlife Society Bulletin 34:782-799. Colorado Division of Wildlife 2002-2007 Strategic Plan. 2002. Colorado Department of Natural Resources, Division of Wildlife, Denver. Colorado Division of Wildlife. 2004. Puma management plan Data Analysis Unit L-23, Game Management Units 70, 71 & 711 Dolores-Norwood Area of Southwest Colorado. Colorado Division of Wildlife, Denver, Colorado. Colorado Division of Wildlife. 2007. Colorado mountain lion management data analysis unit revision and quota development process. Colorado Division of Wildlife, Denver. Colorado Parks and Wildlife. 2015. Strategic Plan. Colorado Parks and Wildlife, Denver. 26 �Cooch, E., and G. White. 2015. Program MARK- a gentle introduction, 14th edition. Colorado State University, Fort Collins. Cooley, H. S. 2008. Effects of hunting on cougar population demography. Ph.D. Dissertation. Washington State University, Pullman. Cooley, H. S., R. B. Wielgus, G. M. Koehler, H. S. Robinson, and B. T. Maletzke. 2009. Does hunting regulate cougar populations? A test of the compensatory mortality hypothesis. Ecology. 90:29132921. Cooley, H. S., K. D. Bunnell, D. C. Stoner, and M. L. Wolfe. 2011. Population management: Cougar hunting. Pages 111-133 in J. A. Jenks, editor. Managing cougars in North America. Jack H. Berryman Institute, Utah State Univ., Logan, Utah, USA. Hornocker, M. G. 1970. An analysis of mountain lion predation upon mule deer and elk in the Idaho Primitive Area. Wildlife Monographs No. 21. Kaplan, E. L., and P. Meier. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 53:457-481. Kreeger, T. J. 1996. Handbook of wildlife chemical immobilization. Wildlife Pharmaceuticals, Inc., Fort Collins, Colorado. Lambert, C. M. S., R. B. Wielgus, H. S. Robinson, D. D. Katnik, H. S. Cruickshank, R. Clarke, and J. Almack. 2006. Cougar population dynamics and viability in the Pacific Northwest. Journal of Wildlife Management 70:246-254. Laundre, J. W., L. Hernandez, D. Streubel, K. Altendorf, and C. L. Lopez Gonzalez. 2000. Aging mountain lions using gum-line recession. Wildlife Society Bulletin 28:963-966. Laundré, J. W., L. Hernández, and S. G. Clark. 2007. Numerical and demographic responses of lions to changes in prey abundance: Testing current predictions. Journal of Wildlife Management 71:345355. Laundré, J. W., and L. Hernández. 2007. Do female pumas (Puma concolor) exhibit a birth pulse? Journal of Mammalogy 88:1300-1304. Lindzey, F. G., W. D. Van Sickle, S. P. Laing, and C. S. Mecham. 1992. Cougar population response to manipulation in southern Utah. Wildlife Society Bulletin 20:224-227. Lindzey, F. G., W. D. Van Sickle, B. B. Ackerman, D. Barnhurst, T. P. Hemker, and S. P. Laing. 1994. Cougar population dynamics in southern Utah. Journal of Wildlife Management 58:619-624. Logan, K. A., L. L. Irwin, and R. Skinner. 1986. Characteristics of a hunted mountain lion population in Wyoming. Journal of Wildlife Management 50:648-654. Logan, K. A., and L. L. Sweanor. 2001. Desert puma: Evolutionary ecology and conservation of an enduring carnivore. Island Press, Washington, D. C. Logan, K. A. 2004. Colorado lion research and development program: population characteristics and vital rates study plan. Colorado Division of Wildlife, Ft. Collins Research Center. Logan, K. A. 2008. Puma population structure and vital rates on the Uncompahgre Plateau, Colorado. Wildlife Research Report. Colorado Division of Wildlife, Fort Collins. Murphy, K. M. 1983. Relationships between a mountain lion population and hunting pressure in western Montana. Master’s Thesis, University of Montana. Pierce, B. M., V. C. Bleich, and R. T. Bowyer. 2000. Social organization of mountain lions: Does a landtenure system regulate population size? Ecology 81:1533-1543. Pojar, T. M., and D. C. Bowden. 2004. Neonatal mule deer fawn survival in west-central Colorado. Journal of Wildlife Management 68:550-560. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7-15. Ross, P. I., and M. G. Jalkotzy. 1992. Characteristics of a hunted population of cougars in southwestern Alberta. Journal of Wildlife Management 56:417-426. Ruth, T. K., M. A. Haroldson, K. M. Murphy, P. C. Buotte, M. G. Hornocker, H. B. Quigley. 2011. Cougar survival and source-sink structure on Greater Yellowstone’s Northern Range. Journal of Wildlife Management 75:1381-1398. 27 �Spreadbury, B. R., K. Musil, J. Musil, C. Kaisner, and J. Kovak. 1996. Cougar population characteristics in southeastern British Columbia. Journal of Wildlife Management 60:962-969. Stoner, D. C., M. L. Wolfe, and D. M. Choate. 2006. Cougar exploitation levels in Utah: Implications for demographic structure, population recovery, and metapopulation dynamics. Journal of Wildlife Management 70:1588-1600. Sweanor, L. L., K. A. Logan, and M. G. Hornocker. 2000. Cougar dispersal patterns, metapopulation dynamics, and conservation. Conservation Biology 14:798-808. Sweanor, L. L., K. A. Logan, J. W. Bauer, B. Milsap, and W. M. Boyce. 2008. Puma and human spatial and temporal use of a popular California state park. Journal of Wildlife Management 72:10761084. Watkins, B. 2004. Mountain lion Data Analysis Unit L-22 Management Plan. Colorado Division of Wildlife, Denver, Colorado. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 (Suppl):S120-S139. Wielgus, R. B., D. E. Morrison, H. S. Cooley, and B. Maletzke. 2013. Effects of male trophy hunting on female carnivore population growth and persistence. Biological Conservation 167:69-75. Yeager, K. 2016. A noninvasive method using auditory predator calls and hair snares to detect and genetically sample cougars (Puma concolor). Master’s Thesis. Department of Fish, Wildlife, and Conservation Biology, Colorado State University, Ft. Collins. Zar, J. H. 1984. Biostatistical analysis. Second edition. Prentice Hall, Englewood Cliffs, New Jersey. Prepared by: ________________________ Kenneth A. Logan, Wildlife Researcher 28 �Table 1. Puma search areas on the Uncompahgre Plateau Study area. West-side East-side 25 Mesa Road to Cottonwood Creek and San 25 Mile Mesa Road to east rim of Roubideau Miguel Canyon (west reach) Canyon and Ben Lowe Mesa ---------------------------------------------------------------------------------------------------+-------------------------------Cottonwood Creek to Horsefly Canyon Roubideau Canyon to Transfer Road ---------------------------------------------------------------------------------------------------!-------------------------------San Miguel Canyon (mid reach) to Maverick Draw Transfer Road to east rim of Dry Creek Basin ---------------------------------------------------------------------------------------------------!-------------------------------Horsefly Canyon and San Miguel Canyon (mid East rim of Dry Creek Basin to east rim of Spring reach) to Clay Creek Canyon ---------------------------------------------------------------------------------------------------!-------------------------------Clay Creek and San Miguel Canyon (upper reach) Spring Canyon to Happy Canyon to McKenzie Creek ---------------------------------------------------------------------------------------------------!-------------------------------McKenzie Creek and San Miguel Canyon (upper Happy Canyon to Horsefly Canyon reach) to Leopard Creek ---------------------------------------------------------------------------------------------------+-------------------------------Horsefly Canyon to McKenzie Butte McKenzie Butte to Loghill Mesa Table 2. Winter count of pumas based on numbers of known radio-collared pumas, visual observations of non-marked pumas, harvested non-marked pumas, and track counts of suspected non-marked pumas on the study area during reference years 4 and 5 (RY4, RY5) and treatment years 1-5 (TY1─TY5). Also indicated * is the population projection for RY5 due to lack of a reliable count (see text), Uncompahgre Plateau study area, Colorado. Period & Year RY4 RY5 TY1 TY2 TY3 TY4 TY5 Study Area region Adults Female Male Subadults Female Male Female East slope 10 4 3 4 4 West slope 6 4 2 0 1 subtotals 16 8 5 4 5 Total Independent Pumas = 33: 21 females, 12 males. Cubs = 20-21 East slope 11-13 5-6 2-4 0-1 2 West slope 9-10 4 1-2 1 3 subtotals 20-23 9-10 3-6 1-2 5 Total Independent Pumas = 37 counted, *45 modeled estimated East slope 16 10 1 2 1 West slope 14 10 0 3 3 subtotals 30 20 1 5 4 Total Independent Pumas = 56: 31 females, 25 males. Cubs = 18-23 East slope 15 5 3 2 7 West slope 15 7 2 3 2 subtotals 30 12 5 5 9 Total Independent Pumas = 52: 35 females, 17 males. Cubs = 39 East slope 13 4 1 3 4 West slope 14 5 3 5 1 subtotals 27 9 4 8 5 Total Independent Pumas = 48: 31 females, 17 males. Cubs = 19 East slope 15 4 3 2 4 West slope 10 5 3 0 2 subtotals 25 9 6 2 6 Total Independent Pumas = 42: 31 females, 11 males. Cubs = 24 East slope 10 6 3 6 6-7 West slope 13 4 1 1 1 subtotals 23 10 4 7 7-8 Total Independent Pumas = 44: 27 females, 17 males. Cubs = 25-28 29 Cubs Male 4 2 6 Unknown sex 7 2-3 9-10 5 2 7 5 4 9 3 3 6 4-8* 5-6 9-14 9 5 14 7 9 16 2 2 4 4 6 10 4 5 9 3 6 9 2 3 5 2 11-13 13-15 �Table 3A. Independent pumas killed by hunters on the study area during each of the treatment period hunting seasons, Uncompahgre Plateau, Colorado. Adult Treatment Period Year Female Male TY1 2 5 TY2 0 5 TY3 3 1 TY4 2 2 TY5 1 3 subtotals 8 16 Subadult Female 1 2 0 0 0 3 Male 1 1 4 1 1 8 Quota 8 8 8 5 5 Actual No. pumas killed 9 8 8 5 5 No. of Independent pumas in count 56 52 48 42 44 Percent harvest of Independent pumas 16.1 15.4 16.7 11.9 11.4 Table 3B. Independent pumas that died of all causes (i.e., total mortality) on the study area during each of the treatment period hunting seasons, Uncompahgre Plateau, Colorado. Treatment Hunting Vehicle Depredation Natural Total Winter % yr. season strike control mortalities count mortality TY1 9 0 0 9 56 16.1 TY2 8 0 2 10 52 19.2 TY3 8 0 0 2 10 48 20.8 TY4 5 0 0 1 6 42 14.3 TY5 5 0 0 1 6 44 13.6 30 �Table 4A. Causes of death in adult, subadult, and cub pumas in the reference and treatment periods, Uncompahgre Plateau, Colorado. Reference Period Adults (21F,11M at Females Females Males Males Total Total risk Number Percent Number Percent Number Percent Strife 3 50 1 100 4 57 Other natural 1 16.7 0 0 1 14.3 Vehicle strike 1 16.7 0 0 1 14.3 Depredation control 1 16.7 0 0 1 14.3 Illegal kill 0 0 0 0 0 0 Hunting 0 0 0 0 0 0 Subadults (8F,14M at risk) Strife 0 0 0 0 0 0 Other natural 1 50 0 0 1 33.3 Vehicle strike 1 50 0 0 1 33.3 Depredation control 0 0 0 0 0 0 Illegal kill 0 0 0 0 0 0 Hunting 0 0 1 100 1 33.3 Cubs (28F,27M at risk) Infanticide 9 75 4 100 13 81.3 Predation 1 8.3 0 0 1 6.2 Other unknown natural 1 8.3 0 0 1 6.2 Starvation 0 0 0 0 0 0 Vehicle strike 1 8.3 0 0 1 6.2 Depredation control 0 0 0 0 0 0 Illegal kill 0 0 0 0 0 0 Hunting 0 0 0 0 0 0 31 �Table 4B. Continued Adults (39F,22M at risk) Strife Other natural Vehicle strike Depredation control Illegal kill Hunting Subadults (19F,34M at risk) Strife Other natural Vehicle strike Depredation control Illegal kill Hunting Cubs (27F,36M at risk) Infanticide Predation Other unknown natural Starvation Vehicle strike Depredation control Illegal kill Hunting Mauled by dogs Treatment Period Females Females Males Number Percent Number 3 14.3 0 7 33 0 2 9.5 1 2 9.5 0 0 0 1 7 33 14 Males Percent 0 0 0 0 6.7 93.3 Total Number 3 7 3 2 1 21 Total Percent 8.1 18.9 8.1 5.4 2.7 56.8 1 0 0 1 0 2 25 0 0 25 0 50 2 2 1 2 0 9 12.5 12.5 6.2 12.5 0 56.2 3 2 1 3 0 11 15 10 5 15 0 55 3 0 0 5 0 2 0 0 0 30 0 0 50 0 20 0 0 0 5 0 4 4 2 1 0 0 1 29.4 0 23.5 23.5 11.8 5.9 0 0 5.9 8 0 4 9 2 3 0 0 1 29.6 0 14.8 33.3 7.4 11.1 0 0 3.7 32 �Table 5. Percent of deaths by cause of marked and non-marked independent pumas by study year RY4−TY5, Uncompahgre Plateau study area, Colorado. Study RY4 RY5 Average % of TY1 TY2 TY3 TY4 TY5 Average % of Year deaths in deaths in reference years 13 treatment years 5 3 17 12 9 11 Total no. RY4−RY5 TY1−TY5 deaths 60 33.3 46.7 15.4 23.5 25.0 11.1 18.2 18.6 % Natural deaths 40 66.7 53.4 15.4 29.4 8.3 33.3 36.4 24.6 % Other humancaused deaths 0 0 0 69.2 47.1 66.7 55.6 45.5 56.8 % Hunting deaths 40 66.7 53.4 84.6 76.5 75.0 88.9 81.8 81.4 % Total humancaused deaths 0 0 0 81.8 61.5 88.9 62.5 55.6 70.6 Hunting as % of humancaused deaths 33 �Table 6. Density* of independent pumas per 100 km2 on the Uncompahgre Plateau, Colorado. Study Year Estimated Estimated density of Estimated density of Estimated density of No. independent independent pumas/100 independent pumas/100 independent pumas/100 km2 of km2 of RSF strata 3 and km2 of RSF strata 2, 3 pumas winter search area 4 (1,636 km2) and 4 (2,426 km2) 2 (1,701 km ) RY4 33 1.9 2.0 1.4 RY5 45 2.6 2.8 1.9 TY1 56 3.3 3.4 2.3 TY2 52 3.1 3.2 2.1 TY3 48 2.8 2.9 2.0 TY4 42 2.5 2.6 1.7 TY5 44 2.6 2.7 1.8 *Density/100 km2=winter count/area*100. Table 7A. Harvest density of independent pumas per 1,000 km2a in combined Colorado puma RSF strata 3 & 4 combined during the treatment period on the UPSA and surrounding GMUs 61 North, 62 North, 65, and 70. GMU puma harvest density/1000 km2 RSF strata 3 & 4 combined Study UPSA 61 North 62 North 65 70 GMUs 61 North, 62 North, Year (1,635)b (1,204) (1,043) (912) (2,211) 65 & 70 combined (5,369) TY1 5.5 5.8 1.0 4.4 6.3 4.8 TY2 4.9 2.5 2.9 5.5 2.7 3.2 TY3 4.9 5.8 6.7 6.6 3.6 5.2 TY4 3.1 5.0 3.8 7.7 5.4 5.4 TY5 3.1 3.3 5.8 6.6 5.4 5.2 a Harvest density=No. independent pumas reported killed by hunters/GMU strata area km2*1000. b Numbers in parenthesis=area in GMU comprised of Colorado puma RSF strata 3 & 4. Table 7B. Harvest density of independent pumas per 1,000 km2a in combined Colorado puma RSF strata 2, 3, & 4 combined during the treatment period on the UPSA and surrounding GMUs 61 North, 62 North, 65, and 70. GMU puma harvest density/1000 km2 RSF strata 2, 3 & 4 combined Study UPSA 61 North 62 North 65 70 GMUs 61 North, 62 North, Year (2,426)b (1,419) (1,477) (1,443) (3,362) 65 & 70 combined (7,701) TY1 3.7 4.9 0.7 2.8 4.2 3.4 TY2 3.3 2.1 2.0 3.5 1.8 2.2 TY3 3.3 4.9 4.7 4.2 2.4 3.6 TY4 2.1 4.2 2.7 4.9 3.6 3.8 TY5 2.1 2.8 4.1 4.2 3.6 3.6 a Harvest density=No. independent pumas reported killed by hunters/GMU strata km2*1000. b Numbers in parenthesis=area in GMU comprised of Colorado puma RSF strata 2, 3, & 4. 34 �Table 8. Adult puma survival modeling results, Uncompahgre Plateau, Colorado. AICc Model Number Model AICc ∆AICc Weights Likelihood Parameters Deviance {S(gender*period)} 396.9874 0 0.49405 1 4 162.0375 {S(gender*deerelk)} 397.5432 0.5558 0.37418 0.7574 4 162.5933 {S(gender+period)} 401.613 4.6256 0.0489 0.099 3 168.6719 {S(gender*year)} 402.1608 5.1734 0.03719 0.0753 14 147.0023 {S(gender+deerelk)} 402.4021 5.4147 0.03296 0.0667 3 169.461 {S(period)} 405.339 8.3516 0.00759 0.0154 2 174.4044 {S(deerelk)} 406.1359 9.1485 0.0051 0.0103 2 175.2013 {S(gender)} 416.7478 19.7604 0.00003 0.0001 2 185.8131 {S(.)} 417.3778 20.3904 0.00002 0 1 188.4475 {S(month)} 509.8345 112.8471 0 0 108 53.2893 {S(gender*month)} 716.7591 319.7717 0 0 216 0 Table 9. Reduced model set for adult puma survival, Uncompahgre Plateau, Colorado. Model Number ∆ AICc AICc wi Likelihood Parameters Deviance Model AICc {S(gender*period)} 396.9874 0 0.84055 1 4 162.0375 {S(gender+period)} 401.613 4.6256 0.0832 0.099 3 168.6719 {S(gender*year)} 402.1608 5.1734 0.06327 0.0753 14 147.0023 {S(period)} 405.339 8.3516 0.01291 0.0154 2 174.4044 {S(gender)} 416.7478 19.7604 0.00004 0 2 185.8131 {S(.)} 417.3778 20.3904 0.00003 0 1 188.4475 Table 10. Puma adult and subadult annual survival rates, estimated from top model Sgender*period for each stage, Uncompahgre Plateau, Colorado. Adults (≥24 months old) Period Gender Average annual Lower 95% CI Upper 95% CI Survival estimate Reference Female 0.8599 0.7153 0.9345 Male 0.9593 0.7459 0.9942 Treatment Female 0.7415 0.6324 0.8230 Male 0.3971 0.2232 0.5692 Subadults (13-24 months old) Period Gender Survival estimate Lower 95% CI Upper 95% CI Reference Female 0.6303 0.2320 0.9058 Male 0.9233 0.6106 0.9893 Treatment Female 0.7026 0.4247 0.8832 Male 0.4272 0.2651 0.6071 35 �Table 11. Subadult puma modeling results, Uncompahgre Plateau, Colorado. Model Number Model AICc ∆ AICc AICc wi Likelihood Parameters Deviance {S(gender*period)} 190.0683 0 0.48562 1 4 39.4874 {S(period)} 191.125 1.0567 0.28631 0.5896 2 44.5933 {S(gender+period)} 192.1299 2.0616 0.17323 0.3567 3 43.5771 {S(.)} 195.2243 5.156 0.03687 0.0759 1 50.7065 {S(gender)} 196.6757 6.6074 0.01784 0.0367 2 50.1439 Table 12. Subadult puma survival rates estimated with second-ranked model Speriod, females and males combined, Uncompahgre Plateau, Colorado. Subadults (13─24 months old) Period Survival estimate Lower 95% CI Upper 95% CI Reference 0.8371 0.5991 0.9464 Treatment 0.5152 0.3685 0.6594 Table 13. Cub puma modeling results, Uncompaghre Plateau, Colorado. Delta AICc Model Number Model QAICc QAICc Weights Likelihood Parameters QDeviance {S(damld)} 201.4726 0 0.37516 1 2 197.4528 {S(gender+damld)} 203.3312 1.8586 0.14813 0.3948 3 197.2916 {S(period+damld)} 203.426 1.9534 0.14127 0.3766 3 197.3864 {S(gender*damld)} 204.6335 3.1609 0.07724 0.2059 4 196.5674 {S(period*damld)} 205.0772 3.6046 0.06187 0.1649 4 197.0111 {S(gender*deerelk)} 205.9641 4.4915 0.03971 0.1058 4 197.898 {S(.)} 206.6167 5.1441 0.02865 0.0764 1 204.6102 {S(deerelk)} 206.7965 5.3239 0.02619 0.0698 2 202.7767 {S(period)} 207.2866 5.814 0.0205 0.0546 2 203.2669 {S(birthmonth)} 208.0169 6.5443 0.01423 0.0379 2 203.9971 {S(period*gender)} 208.1739 6.7013 0.01315 0.0351 4 200.1078 {S(gender)} 208.3153 6.8427 0.01226 0.0327 2 204.2955 {S(gender+deerelk)} 208.3844 6.9118 0.01184 0.0316 3 202.3448 {S(gender+period)} 208.9231 7.4505 0.00904 0.0241 3 202.8835 {S(birthmonth^2)} 209.6091 8.1365 0.00642 0.0171 3 203.5694 {S(gender*period)+birthmonth} 209.6737 8.2011 0.00621 0.0166 5 199.5744 {S(gender+birthmonth)} 209.8226 8.35 0.00577 0.0154 3 203.783 {S(gender*period)+birthmonth^2} 211.6137 10.1411 0.00236 0.0063 6 199.4744 36 �Table 14. Cub puma survival rates estimated with top-ranked models, Uncompaghre Plateau, Colorado. Cubs (1─12 months old) Model Survival estimate Lower 95% CI Upper 95% CI {S(damld)} 0.450 0.338 0.567 {S(gender+damld)} 0.421 0.274 0.584 {S(period+damld)} I Reference period 0.471 0.306 0.641 0.282 0.594 I Treatment period 0.431 37 �43 ' 40 Figure 1. The puma study area on the southern half of the Uncompahgre Plateau, Colorado (shaded in gray) comprising the southern portions of Game Management Units (GMUs) 61 and 62 and a northern portion of GMU 70, with surrounding GMUs referred to in the text. 38 �Figure 2. The Uncompahgre Plateau Study Area Game Management Unit (UPSA) indicating all the puma search routes and the more intensively used search routes on the east and west winter search areas. 39 �)Q 50 - - - - - - - - z-----:.111 f '-,:____ _ _____,,= ----=-....aa =------=-=-------§ c.. 40 - - -~ _;., 45 -.............. ..... _. . .- • - 44 -~ - - - - - - - - - - - - - - • - ,,< ~ 4 _2 _ _ __ ~ 30 - - - - - - - - - - - - - - - - - - - - - - - - 4t Q. 4t "0 20 - - - - - - - - - - - - - - - - - - - - - - - - - .E 0 :z: 10 - - - - - - - - - - - - - - - - - - - - - - - - - 0 RYS RY4 TY1 TY2 TY3 TY4 Study Ye ar Figure 3. Annual winter counts of independent pumas, Uncompahgre Plateau, Colorado. Counts in RY4 and TY1 to TY5 are from ground surveys and capture efforts. RY5 is a modeled data point (see text). The count for RY5 is biased low because capture efforts were insufficient due to lack of personnel to thoroughly search the study area (see text). 60 50 )8 ~.., 40 C Cl.I ~ 30 11,1 Q. • • TY1 TY2 ] 20 0 :z; • • TY3 TY4 4 10 --, 0 Treatment Period Year l1idependent PL1111as ..,_No. Adu lt F ...... No. Adult M Figure 4. Change in numbers of independent and adult pumas, Uncompahgre Plateau, Colorado. Abundance of independent pumas declined 25% by TY4, adult females declined 23% by TY5, and adult males declined 55% by TY4 and 50% by TY5. See text for puma harvest rates. 40 �100.0 90.Q 80.0 70.0 60 .0 C: (II ...u(II 50.0 CL 40.0 ... ■ Ad ult F 30.0 2.0.0 10.0 ■ Ad ult M 0.0 Hunting Other human Natural Unknown Survived fate period Cause.s of Adt.dl Puma Mortality in Reference Period Figure 5A. Categories of causes of death in radio-collared adult pumas monitored during the reference period, Uncompahgre Plateau, Colorado. 70.0 60.0 50.0 ...; 40.0 ~ £ 30. 0 ■ Adult F 20.0 ■ Adult M 10.0 0.0 Hunting • Other human tlatural Unl-.nown Survived fat e penod Causes of Adult Puma Mortality In Treatment Period Figure 5B. Categories of causes of death in radio-collared adult pumas monitored during the treatment period, Uncompahgre Plateau, Colorado. 41 �100.0 90.0 80.0 70. 0 ....C: 60 .0 Cl.I u 50.0 Cl.I Q. 40.0 30.0 20 .0 ... ■ Subadlllt F ■ Subadlllt M 10.0 0. 0 Hunting Other Natural Unknown Survived human fate period Causes of Subadult Puma Mortality in Reference Period Figure 6A. Categories of causes of death in known-fate subadult pumas during the reference period, Uncompahgre Plateau, Colorado. 80~0 ~ - - - - - - - - - - - - - - - - 70.0 60.0 - + - - - - - - - - - - - - - - .., 50.0 - + - - - - - - - - - - - - - - - c: Cl.I l: 40.0 Cl.I c. 30.0 - + - - - - - - - - - - - - - - - - - ■ Subadult F ■ Subadlllt M 20.0 10.0 0.0 Hunting Other human l~atural Unknown Sllrvived fate period Causes of Subadult Pum'?I Mortality in Treatment Period Figure 6B. Categories of causes of death in known-fate subadult pumas during the treatment period, Uncompahgre Plateau, Colorado. 42 �<( 60 0 ~ 1 .! lQ 45 2 "' ... 3 "+l ~ 35 C: 4 la i "' s 5 ~ 55 § so i::i- 5 o. 40 §.... ~ C :c ! 30 25 Q. ~ 20 6 TYl TY2 1Y4 rY3 TVS Treatment Period Year No. Independent pumas on UPSA ...- Purn a h arvest / 1000 sq . km RSF Strata. 39:4 111 GMU s 61 N ,62M,65 1 70 combined .....,UPSA Puma harvest/ 1000 sq . lm1 RSF Strata 3&4 Figure 7A. Relationships of the abundance of independent pumas on the UPSA, puma harvest density on the UPSA, and harvest density on adjacent GMUs 61 North, 62 North, 65, and 70 in RSF strata 3 and 4 combined within each GMU during the treatment period, Uncompahgre Plateau Puma Project, Colorado. 0 60 0. 5 E ,:,&, 1 .,,er 3 "'E :::, 3. 5 Go 20 4 TY1 TY2 TY4 TY3 Treatment Period Vear TY5 llo. Independen t pumas on UPSA - Puma harvest / 1000 sq. km RSF St rata 2, l , 4 1n GMU$ 6111,6211 ,651 70 combined UPSA Puma h arvest / 1000 sq . km RSF Strata 2 ,3.4 Figure 7B. Relationships of the abundance of independent pumas on the UPSA, puma harvest density on the UPSA, and harvest density on adjacent GMUs 61 North, 62 North, 65, and 70 in RSF strata 2, 3 and 4 combined within each GMU during the treatment period, Uncompahgre Plateau Puma Project, Colorado. 43 �Agestn1 ctu1 e o r i udep endtni pum::1si11 Non•mbc1· 21109 at lt~ lnnhtg of lht> puma hnnUug :W:J'lOU Ill Tl t'a lmt'IIC Yea r J l lncom1n1hgre Plakan. C'olo1ad o. G 5 "' ~ 4 if ._ :3 ., i 0 1 0 I I- Female I r 1 to 2 >2 to >3 to -.4 lCl ,5 lo >6 lo ,7 ft) >8 lO >9 tCI 3 4 ~ fi 7 R ll LO ■ M.ile 101 A u•(yl":m ;I Agestrncture of independen t p umas in November 1013 at beginning of puma hunting season in Treatment Year 5. Uncompahgre Plateau. Colol'aclo. 6 ---------------------- "' 5 84 E3 0 Female o2 z. 1 t-----------=--~ r ------l r - ■ Ma l e 0 1 lo 2 >l' to >3 lo >4 to >5 to >.6 lo >7 to >8 to >9 to 10+ 3 4 5 6 7 8 9 10 Age (years) Figure 8. Age structures of independent pumas at the start of TY1 (top) and TY5 (bottom), Uncompahgre Plateau, Colorado. The top graph represents the age structure after 5 years of no sport-hunting and ages of independent pumas just before the first treatment hunting season on the study area. Therefore, it represents the oldest age structure after 5 years with no hunting. The bottom graph represents the change in age structure at the start of TY5 after 4 years of sport-hunting pumas (TY1─TY4) and other causes of mortality operated on the population. 44 �45000 "" 40000 ~ 35000 :, 1.200 +l~ii:::~::=;,;;;ii======================J ....recu 1.000 a:: iii > ·;;: ... c. ... 30000 0.800 Cl ~ 20000 0 .600 E :, {l 15000 OAOO .::: :, ~ 25000 re c.. C "Cl ] 10000 <t: :, Vl <( 0.200 iii 5000 ::, C C 0 .000 <( 0 RY2 RY3 RY4 RYS TYl TY2 TY3 TY4 TVS Study Year - DeerElk - AMS - AFS Figure 9. Mule deer and elk abundance combined estimates graphed with the modeled adult puma monthly survival rates converted to annual point estimates to illustrate this relationship. The relationship of deer and elk abundance to adult puma survival rates was influenced directly by our manipulation of the puma population that caused it to increase in the reference period and to decline in the treatment period. Therefore we excluded this model from the model set for adult puma survival (see text). 20 18 16 1.4 ~ ...$ 12 .2 10 0 0 z 8 6 I- f-- 4 2 0 ■ -,Jan. Feb. Mar. Apr. May June July Aug, Sep. Oct. lfov. Dec. Figure 10. Puma births (black bars) detected by month from May 19, 2005 to September 30, 2014 (n = 66 litters of 33 females; 60 litters were examined at nurseries when cubs were 25-45 days old, 4 litters were confirmed by tracks of ≥1 cubs following GPS- and VHF-collared mothers and 2 litters by remains of cubs of 2 GPS-collared mothers when cubs were ≤45 days old, Uncompahgre Plateau, Colorado. 45 �APPENDICES Appendix I. ACUC Capture and Handling Forms and Protocols File # _________________ Revised Date ______________ (ACUC Secretary will supply) COLORADO PARKS AND WILDLIFE ANIMAL CARE AND USE COMMITTEE (CPW ACUC) FORM FOR REVIEW OF NEW RESEARCH PROJECTS 1. Principal Investigator (s): Dr. Kenneth A. Logan, Mammals Researcher, CPW. Phone: 970-252-6013(o) or 970-275-3227(c) E-mail: ken.logan@state.co.us 2. 3. All investigators (including all individuals involved in implementing research: Principal investigator Ken Logan (CPW), all CPW technicians and other houndsmen. Location of facility or study area: The study area is on the Uncompahgre Plateau in western 4. Beginning date: December 1, 2008. 5. Ending date: April 1, 2014. 6. Title of project: Assessing Effects of Hunting on a Puma Population on the Uncompahgre Plateau, Colorado. 7. Species of animal (s): Puma concolor 8. A study Plan or Prospectus describing each research or pilot project is required with this form. Is the Study Plan attached? Yes _X_ No ___ 9. Rationale for use of this animal model: a. Explain why other models (e.g. nonanimal models, in vitro techniques) are inappropriate. This study pertains specifically to puma population dynamics and attendant effects of hunting off-take. It is intended to provide wildlife managers with useful information for the management of pumas in Colorado. b. If not a species specific study, why is this the most appropriate species for this research? Colorado in areas west and southwest of Montrose. The study area is the South Uncompahgre Plateau (in Mesa, Montrose, Ouray, and San Miguel Counties). The study area includes about 2,200 km2 of the southern halves of GMUs 61 and 62, and about 155 km2 of the northern edge of GMU 70. The area is bounded by state highway 348 at Delta, 25 Mesa road and Forest Service road FS503 to Nucla, state highway 97 to state highway 141 to state highway 145 to Placerville, state highway 62 to Ridgeway, U.S. highway 550 to Montrose, and U.S. highway 50 to Delta. c. If capturing wild animals for pen research, why is this source most appropriate? 46 �10. 11. If the study will use wild animals, describe capture and transport methods: Please refer to the attached study plan Puma capture and marking (pages 13-15, 18, 24) and the Mountain Lion Capture and Handling Guidelines. Location of capture: Pumas will be captured on the study area described in question 3 above. 12. Indicate number of animals to be used _20-30 pumas/year_. Provide a brief justification (or page reference in Study Plan) for sample size selected: Please see study plan sections Puma capture and marking, Population monitoring, and Population Size (pages 13-19). 13. By signing this form, you are verifying that all persons involved in this project are adequately trained. Briefly describe the training process(es) and list personnel responsible for animal care and handling: K. Logan, S. Young, have all been trained in and have directly captured, immobilized, and sampled pumas. All new technicians and houndsmen will receive training on pumas by principal investigator Ken Logan. 14. Provide a detailed description of the procedures and manipulations of animals, including an end point (if necessary) at which animals will be removed from experiment or be euthanized. (If described in Study Plan or Prospectus, provide reference page numbers.) If administration of anesthesia and /or surgery is part of the procedure, identify who will perform these tasks: Please see study plan section Puma capture and marking (pages 13-19). 15. Are the levels of pain and suffering, stress, discomfort, deprivation, etc., to be experienced by experimental animals greater than normally associated with handling, administration of therapeutics by commonly used methods, or routine venipuncture? Yes __ No _X_ If answered yes, attach a detailed justification and indicate here the date of search, some of literature search, date range searched, and key words and combination of key words searched to document the lack of alternative methods: 16. 17. Will pain and suffering be controlled? Yes ____ No ____ N/A _X__ If answered no, attach a detailed justification. Describe how pain and suffering not associated with routine handling will be controlled. a. Methods and dosage of anesthesia to be used: N/A b. Methods and dosage of analgesia to be used: N/A c. Methods and dosage of tranquilization to be used: N/A The attending veterinarian must be consulted when planning projects where handling of any animal will occur. Do this prior to submitting this application. Date consulted_________ Does the proposed project include planned euthanasia of animals? Yes _____ No _X__ 47 �Signing below assures that all investigators have reviewed the CPW ACUC Euthanasia Guidelines and that investigators will use appropriate methods for humanely destroying animals involved in their study. Please indicate the criteria for and methods of euthanasia to be used in this study: Date: ________ 18. Signed: ___________________________________ Principal Investigator Signing below assures that the planned research does not unnecessarily duplicate previous research on the subject and species proposed for study. Date: ________ Signed: ___________________________________ Principal Investigator 48 �Appendix II. Puma Population Model and Simulations Research on the Uncompahgre Plateau Puma Project from December 2004 to July 2007 provided estimates of puma population structure and parameters for a model-based approach developed by CPW biometrician Dr. P. Lukacs and Mammals Researcher Dr. K. Logan to examine options for the design of the remainder of this research, and as a preliminary assessment of the CPW puma management assumptions. Puma Population Modeling Our puma population projections for the study area involved an age-structured, deterministic, discrete time model. The additive puma population model structure is: Nt+1 = Adult Females = (SAF * NAFt + SSF * NSFt) * (1 – HAFt+1) + Adult Males = (SAM * NAMt + SSM * NSMt) * (1 – HAMt+1) + Subadult Females = ((r * SC * NCt) * (1 – HSFt+1)) * PISF/ESF + Subadult Males = (((1 − r) * SC * NCt) * (1 – HSMt+1)) * PISM/ESM + Cubs = Lỹ * AFR * NAFt+1 Terms: NAFt+1 = Number of adult females at year t+1. NAMt+1 = Number of adult males at year t+1. NSFt+1 = Number of subadult females at year t+1. NSMt+1 = Number of subadult males at year t+1. NJt+1 = Number of juveniles at year t+1. S = Survival rate for each specified sex and age stage. H = Proportion of the harvest rate comprised by each sex and age stage (e.g., 0.28 harvest rate * 0.40 adult females). r = Proportion of the subadult population that is female (e.g., 0.5; 1-0.5 = proportion of males). PI/E = Ratio of progeny + immigrants/emigrants. Lỹ = Average litter size. AFR = Proportion of adult females giving birth to new litters each year. Basic assumptions of the model include: 1) expected puma population projections and annual rates of increase (i.e., lambda) are conditional on the assigned puma population structure and demographic estimates, and 2) no density dependent responses are built into the model. In reality, density dependence probably operates in puma population dynamics, with competition for food regulating adult female density and competition for mates regulating adult male density (Logan and Sweanor 2001). We parameterized the model with data gathered on the pumas on the study area during the previous 3.7 years. The starting population was the minimum count of pumas and attendant estimated sex and age structure made during November 2007 to March 2008 (Table AI.1). We assumed that all individuals were present in the population during that entire period. No mortalities of independent pumas were detected. But, one radio-collared subadult male emigrated by March 19, 2008. Population parameters included: estimated rates of reproduction and sex and age-stage specific survival, which included data to July 2008 (Table I.2). Some sex and age-stage specific estimates of survival (i.e., adult male, subadult male, subadult female) came from the literature (Table 2), because our current sample sizes (i.e., number of individuals and years) were not adequate for realistic estimates (i.e., estimates from our data were 1.0 for adult males and subadults). We did not use actual rates in the literature where estimates involved the pooling of data on sexes and age stages, and where sample sizes for age stages were not presented (e.g., Anderson et al. 1992). In addition, the ratio of progeny and immigrant recruits to 49 �emigrants as a model input was from the literature, because such data were scarce and does not exist for Colorado (all references in Table AI.2). We preferred using the population characteristics and parameter estimates gathered in the current research effort, because this is the puma population we intend to manipulate to assess current CPW puma management strategies. Table AI.1. Minimum puma population count on Uncompahgre Plateau study area, Colorado, November 2007 to March 2008 (RY4). The minimum count involves counting all radio- and GPS-collared pumas, all other marked pumas, and all presumably unmarked pumas detected on the study area during the period. Presumed unmarked pumas could be marked with ear-tags and tattoos. Their tracks and movements could be separated from movements of radio- and GPS-collared pumas. Or they exhibited evidence that could separate them from other local marked pumas from their tracks (i.e., distinguishable by sex, number of cubs and/or relative size of cubs varied). Region East slope West slope Totals a Adults Subadults Female Male Female Male 10 4 3 4 6 4 2 0 16 8 5 4 Total Independent Pumas = 33a,b Female 4 1 5 Cubs Male 4 2 6 Unknown sex 7 2-3 20-21 Of the total, 23−24 (70−73%) independent pumas were marked and 9-10 (27−30%) were assumed to be unmarked. Table AI.2. Summary of preliminary puma population model parameter estimates obtained from the Uncompahgre Plateau Puma Project and from the literature on puma. Survival Sex and age stage Adult Female Estimate 0.87 Adult Male 0.91 Subadult Female 0.80 Subadult Male 0.60 Cub 0.50 0.90 Parameter Adult age Estimate 2+ years Reference Estimated average annual survival rate (n = 2 years) for 11−13 adult females on Uncompahgre Plateau study area. Estimated average annual survival rate (n = 8 years) for adult males in a nonhunted New Mexico puma population (Logan and Sweanor 2001:127-128). Estimated annual survival rate (n = 2 years) for 5−9 adult males on Uncompahgre Plateau study area was 1.00. Estimated subadult female survival in New Mexico (0.88, n = 16; Logan and Sweanor 2001:122) adjusted downward for potential lower survival for pumas 12-24 months old on Uncompahgre Plateau (0.642, n = 14 females and 10 males combined, life stages not known or described in Anderson et al. 1992:53). Survival of 7 radio-collared pumas (5 males, 2 females) in the subadult stage in the current Uncompahgre Plateau puma study is 1.00. Estimated subadult male survival in New Mexico (i.e., 0.56, n = 9; Logan and Sweanor 2001:122) adjusted upward for potential slightly higher survival for pumas of both sexes 12-24 months old (i.e., 0.642) on Uncompahgre Plateau (Anderson et al. 1992:53). Survival of 7 radiocollared pumas (5 males, 2 females) in the subadult stage in the current Uncompahgre Plateau puma study is 1.00. Estimated cub survival rate (n = 38 cubs combined sexes), on Uncompahgre Plateau study area. This survival rate is applied to the model starting with the expected number of cubs from birth in RY5. Estimated cub survival for cubs ≥7 months old, and is applied to RY4 cubs only, because the minimum count of pumas in RY4 was tallied when most cub mortality had already occurred. Survival of cubs ≥7 months old in the literature is about 0.95 (Logan and Sweanor 2001). Here, a more conservative 0.90 is used in this model. Reproduction Reference Assume all females 2 years old and older are adults (Logan and Sweanor 2001: 93-94). 50 �Litter size 2.81 Secondary sex ratio observed at nurseries 1:1 Proportion of adult females producing new litters each year 0.65 Parameter Subadult female Estimated Ratio 1.02 Subadult male 0.94 Average litter size for 21 litters on the Uncompahgre Plateau study area = 2.810 ± 0.9808SD; litters were examined when the cubs were 26 to 42 days old. Secondary sex ratio was 33:26 for 21 litters examined at 29 to 42 days old on the Uncompahgre Plateau study area (not significantly different from 1:1, (X2 = 0.8305 < 3.841, α = 0.05, 1 d.f.). This result supported Logan and Sweanor 2001:69, n = 148). Proportion of adult females giving birth each year (n = 3 years for n = 12, 13, 12 females), Uncompahgre Plateau study area. Proportion for a non-hunted puma population in New Mexico was 0.50 (Logan and Sweanor 2001:98). Progeny + Immigrant Recruits/Emigration Ratio Reference No data for pumas in Colorado exists. Assume the ratio of female immigrants to emigrants = 1.02. This ratio is consistent with estimates for a New Mexico puma population that functioned as a source (Sweanor et al. 2000). No data for pumas in Colorado exists. Assume the ratio of male immigrants to emigrants = 0.94, (i.e., male immigration is half of emigration). This ratio is consistent with estimates for a New Mexico puma population that functioned as a source (Sweanor et al. 2000). Results of Puma Population Simulations Expected minimum population sizes for independent pumas for RY5 and TY1 conditional upon the number of independent pumas counted in Reference Year 4 (RY4) and the model input parameters and assumptions (given in Tables AI.1 and AI.2). Table AI.3. Year RY4 RY5 TY1 Adult Female 16 18 23 Puma Population Size Subadult Male Female 8 5 10 9 14 8 Male 4 8 8 Cub 20 33 42 51 Independent Pumas Total 33 count 45 projected 53 projected �Appendix III. MOUNTAIN LION HUNTER SURVEY MOUNTAIN LION HUNTER SURVEY EXPERIMENTAL LION HARVEST UNCOMPAHGRE PLATEAU STUDY AREA- GMUs 61, 62, and 70 Hunter Name: ___ License No.: CID No.: 1. Please circle the days on which you hunted (please count partial days hunting as full days) November: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30 December: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31 January: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31 2. Name the drainages and mesas where you hunted __________________________________________________ ____________________________________________________________________________________________ ____________________________________________________________________________________________ 3. Did you hunt with hounds? YES or NO (circle one) 4. Did you hunt with an outfitter? YES or NO (circle one) 5. Do you consider yourself to be a SELECTIVE hunter or a NON-SELECTIVE hunter? (read explanation below, then circle one) A SELECTIVE hunter is one that purposely is hunting for a specific type of legal lion, such as a male, large male, or large female, and therefore attempts to distinguish between male and female tracks, large and small males or females before taking the animal, and is willing to pass up lions that are detected from tracks or when treed. A NON-SELECTIVE hunter is one that intends to take whatever legal lion is first encountered or caught, with no desire for sex or size. 6. What was the sex of the lion that made the first set of tracks you encountered that were less than one day old? FEMALE  MALE  Did you pursue the lion to harvest it? YES  NO  NOTE: Adult & subadult male lions usually have hindfoot heel pad widths greater than or equal to 2 1/16 in. (52mm) wide. Adult & subadult female lions usually have hindfoot heel pad widths less than or equal to 1 15/16 in. (50 mm) wide. 7. Of the total tracks you encountered that were less than one day old, how many were male (_____) and female (_____) lions? (write number on the blank) 8. How many tracks were of females followed by cubs? _________ 9. How many times did you pursue lions with dogs? _________ 10. How many times did you tree or bay lions with dogs? _________ 11. How many of the lions treed and bayed were males (________), females (________), and cubs (________)? 12. Were any of the lions marked with a visible collar or ear-tags? YES or NO (circle one) If YES, describe the collar color, ear-tag color and number on each lion and its sex & age (i.e., male or female; adults ≥2 yrs. or subadults ~1-2 yrs.; indicate male or female and adult or subadult for each) ___________________________________________________________________________________________ 13. Describe the non-marked lions you caught (e.g., adult male, adult female, subadult male, subadult female) and list here: ___________________________________________________________________________________ 14. Did you harvest a lion? YES or NO (circle one) If YES, what was it? MALE or FEMALE (circle one). ADULT (≥2 yrs.) or SUBADULT (~1-2 yrs.) (circle one) 15. What was the seal number? ____________________ 16. Did marks (e.g., collar, ear-tag) on the lion influence your decision to harvest or not harvest the animal? (check one)  TO HARVEST  NOT TO HARVEST  NO INFLUENCE AT ALL 17. Did snow facilitate your harvest? YES if the puma was tracked on snow. NO if the puma was tracked on ground without snow. (circle one) 52 �Compliance Endangered Species Act This research will involve trapping mountain lions using hounds, cage traps and snares. It is extremely unlikely that any listed species under the Endangered Species Act will be inadvertently captured. However, in the unlikely event that a lynx or wolverine was captured, we will immediately release the animal unharmed. We will utilize existing roadways on public and private lands to access areas for running hounds and setting traps. Other field work on this project will comprise telemetry monitoring primarily from roads and fixed wing aircraft, minimizing potential for disturbing any listed species. No activities associated with this project pose a threat to the well-being of any listed species in Colorado. Animal Welfare Act The project is approved through Colorado Division of Wildlife’s Animal Care and Use Committee (Project #08-2004 and #03-2007). NEPA This research falls under a Categorical Exclusion as set forth in Title 40, Section 1508.4 of the Code of Federal Regulations (i.e., 40 CFR 1508.4) because the actions in this research do not involve significant environmental impacts. Other Landscape-Oriented Federal Acts This research will have no impact on the landscape, and therefore, will not violate provisions of other Federal Legislation governing floodplains and wetlands, historical sites, and prime and unique farmlands. Americans With Disabilities Act When hiring personnel as part of this project, qualified individuals will not be discriminated against based on disability. No structures or access points will be constructed as part of this research, and thus accessibility is not applicable. 53 �ÿ ÿ ÿ ÿ 34546784ÿ76 ÿ78ÿ��585���ÿ ÿ ��������ÿ������3�ÿ����3�ÿ������ ÿ ÿ �!75"7#�4ÿ4�ÿ$4" #7�ÿ5�4ÿ$77%�$�#ÿ�ÿ34546784ÿ ÿ &'()*+ÿ,*-'('+.ÿ/012ÿ134ÿ56378/09'ÿ164ÿ563:ÿ ÿ ;<<'=>?ÿ*<ÿ&0@AÿB09>)9Cÿ*9ÿAÿ&0@Aÿ&*D01A>)*9ÿ)9ÿ,*1*(A+*ÿ ÿ &()9=)DA1ÿE9-'?>)CA>*(.ÿF'99'>GÿHIÿJ*CA94ÿF'9IJ*CA9K?>A>'I=*I0?4ÿ/*9ÿL09C'4ÿ/*9IL09C'K?>A>'I=*I0?ÿ ÿ �55ÿ��46$7#�4ÿ�ÿ#M�ÿN64O�P#ÿ"$$76Qÿ�ÿN6�5�$�76Qÿ78ÿ"RO�P#ÿ#4ÿ�"6#M�6ÿ�!75"7#�4Sÿ ��46$7#�4ÿ�� ÿT��ÿU�ÿ�U������ÿ��ÿV�����ÿW�#M4"#ÿN�6$��4ÿ4�ÿ#M�ÿ7"#M46Sÿ �7�N"57#�4 ÿ4�ÿ#M��ÿ87#7ÿR�Q48ÿ#M7#ÿP4#7��8ÿ�ÿ#M�ÿ6�N46#ÿ�ÿ8�P4"67%�8Sÿ ÿ X'ÿ'YA@)9'+ÿ'<<'=>?ÿ*<ÿ('C01A>'+ÿG09>)9Cÿ*9ÿAÿD0@AÿZ[\]^ÿ̀ab`acadeÿD*D01A>)*9ÿ*9ÿ>G'ÿ f9=*@DAGC('ÿ&1A>'A04ÿ,*1*(A+*ÿ<*(ÿ36ÿ2'A(?ÿg566285632hIÿiG'ÿ?>0+2ÿjA?ÿ+'?)C9'+ÿj)>GÿAÿ('<'('9='ÿ D'()*+ÿg2'A(?ÿ38khÿj)>G*0>ÿD0@AÿG09>)9CÿA9+ÿAÿ>('A>@'9>ÿD'()*+ÿg2'A(?ÿl836hÿj)>GÿD0@AÿG09>)9CIÿX'ÿ =AD>0('+ÿA9+ÿ@A(m'+ÿD0@A?ÿ*9ÿ>G'ÿ?>0+2ÿA('Aÿgf&nHhÿA9+ÿ@*9)>*('+ÿ>G'@ÿ2'A(o(*09+ÿ<*(ÿ+A>Aÿ*9ÿD0@Aÿ +'@*C(ADG)=?IÿHp09+A9='ÿ*<ÿ)9+'D'9+'9>ÿD0@A?ÿjA?ÿ'?>)@A>'+ÿ*9ÿ>G'ÿf&nHÿ'A=Gÿj)9>'(ÿ+0()9Cÿ>G'ÿ ,*1*(A+*ÿD0@AÿG09>)9Cÿ?'A?*9ÿ<(*@ÿ('<'('9='ÿ2'A(ÿ2ÿgLq2hÿ>*ÿ>('A>@'9>ÿ2'A(ÿkÿgiqkhÿ0?)9Cÿ>G'ÿJ)9=*19o &'>'(?'9ÿ@'>G*+IÿE9ÿA++)>)*94ÿj'ÿ?0(-'2'+ÿD0@AÿG09>'(?ÿ>*ÿ1'A(9ÿG*jÿG09>'(ÿp'GA-)*(ÿ)9<10'9='+ÿGA(-'?>ÿ A9+ÿ>G'ÿD0@AÿD*D01A>)*9IÿE9ÿ>G'ÿ('<'('9='ÿA9+ÿ>('A>@'9>ÿD'()*+?4ÿ336ÿA9+ÿ33lÿ)9+)-)+0A1ÿD0@A?ÿj'('ÿ =AD>0('+ÿA9+ÿ@A(m'+4ÿ('?D'=>)-'124ÿ+0()9Cÿ226ÿ>*>A1ÿ=AD>0('ÿ'-'9>?IÿrA>AÿA9A12?'?ÿGA-'ÿp''9ÿ=*@D1'>'+ÿ A9+ÿ('?01>?ÿA9+ÿ=*9=10?)*9?ÿGA-'ÿp''9ÿ?0p@)>>'+ÿ<*(ÿ?=)'9>)<)=ÿD''(o('-)'jÿA9+ÿD0p1)=A>)*9ÿ>*ÿstcuctvwÿ xabayd^z{|Iÿ ÿ XGA>ÿJ)@)>?ÿA9+ÿL'C01A>'?ÿ&0@Aÿ&*D01A>)*9?}ÿXGA>ÿ~A>>'(?}ÿ ÿ &()9=)DA1ÿE9-'?>)CA>*(.ÿF'99'>GÿHIÿJ*CA94ÿF'9IJ*CA9K?>A>'I=*I0?ÿ ÿ X)1+1)<'ÿ@A9AC'(?ÿ('0)('ÿ('1)Ap1'ÿ)9<*(@A>)*9ÿ*9ÿ<A=>*(?ÿ>GA>ÿ)9<10'9='ÿA9)@A1ÿD*D01A>)*9?ÿ)9ÿ *(+'(ÿ>*ÿ+'-'1*Dÿ?0=='??<01ÿ@A9AC'@'9>ÿD(*C(A@?Iÿn0=Gÿ)?ÿ>G'ÿ=A?'ÿj)>Gÿ>G'ÿD0@Aÿg[\]^ÿ̀ab`acadhÿ)9ÿ j'?>'(9ÿ€*(>GÿH@'()=A4ÿjG'('ÿD*D01A>)*9?ÿGA-'ÿ('=*-'('+ÿ)9ÿ('='9>ÿ+'=A+'?ÿD()@A()12ÿ+0'ÿ>*ÿ('?>()=>)*9?ÿ *9ÿG0@A9o=A0?'+ÿ@*(>A1)>2Iÿ~A9AC'(?ÿ9''+ÿAÿ=1'A(ÿ09+'(?>A9+)9Cÿ*<ÿ>G'ÿ<A=>*(?ÿ>GA>ÿ1)@)>ÿ*(ÿ('C01A>'ÿ D0@AÿD*D01A>)*9?ÿA9+ÿG*jÿ>G*?'ÿ<A=>*(?ÿ@)CG>ÿp'ÿ@A9)D01A>'+ÿ>*ÿA=G)'-'ÿ@A9AC'@'9>ÿ*p'=>)-'?4ÿ )9=10+)9C.ÿ?0?>A)9)9CÿD0@AÿA9+ÿ*>G'(ÿj)1+1)<'ÿD*D01A>)*9?4ÿD(*-)+)9CÿG09>)9Cÿ*DD*(>09)>24ÿA9+ÿ('+0=)9Cÿ D0@Aÿ=*9<1)=>?ÿj)>GÿD'*D1'IÿE9ÿ>G)?ÿDAD'(4ÿEÿD('?'9>ÿ('1'-A9>ÿ>'=G9)=A1ÿ1)>'(A>0('ÿ*9ÿD0@AÿD*D01A>)*9?4ÿ p'GA-)*(4ÿA9+ÿ('1A>)*9?G)D?ÿj)>GÿD('2ÿ>GA>ÿGA?ÿ=*9>()p0>'+ÿ>*ÿ'?>Ap1)?G)9CÿG2D*>G'?'?ÿ*9ÿD0@AÿD*D01A>)*9ÿ 1)@)>A>)*9ÿA9+ÿ('C01A>)*9Iÿ,0(('9>ÿG2D*>G'?'?ÿ)9=10+'.ÿ>G'ÿ?*=)A1ÿ1)@)>A>)*9ÿG2D*>G'?)?4ÿA9+ÿ>G'ÿ<**+ÿ 1)@)>A>)*9ÿG2D*>G'?)?IÿH??*=)A>'+ÿj)>Gÿ'A=Gÿ*<ÿ>G*?'ÿA('ÿ5ÿG2D*>G'?'?ÿ*9ÿD0@AÿD*D01A>)*9ÿ('C01A>)*9.ÿ>G'ÿ ?*=)A1ÿ('C01A>)*9ÿG2D*>G'?)?ÿA9+ÿ>G'ÿ('C01A>)*9ÿp2ÿ=*@D'>)>)*9ÿG2D*>G'?)?IÿEÿ+)?=0??ÿ@A9AC'@'9>ÿ )@D1)=A>)*9?ÿ<(A@'+ÿp2ÿ>G'?'ÿG2D*>G'?'?ÿA?ÿ>G'2ÿD'(>A)9ÿ>*ÿD0@AÿG09>)9C4ÿ=*9<1)=>?4ÿA9+ÿ('1A>)*9?G)D?ÿj)>Gÿ 09C01A>'ÿD('2Iÿ‚)9A1124ÿEÿ+)?=0??ÿ'9-)(*9@'9>A1ÿA9+ÿ?*=)'>A1ÿ)??0'?ÿ>GA>ÿ+'>'(@)9'ÿD0@Aÿ@A9AC'@'9>ÿ *p'=>)-'?ÿA9+ÿ=GA11'9C'ÿD0@Aÿ=*9?'(-A>)*9IÿE9>'(9A1ÿ,&XÿD''(ÿ('-)'jÿGA?ÿp''9ÿ=*@D1'>'+ÿ<*(ÿ>G)?ÿ @A90?=()D>ÿA9+ÿ>G'ÿ('-)?'+ÿ@A90?=()D>ÿ)?ÿD1A99'+ÿ>*ÿp'ÿ?0p@)>>'+ÿ<*(ÿ?=)'9>)<)=ÿD''(o('-)'jÿA9+ÿ D0p1)=A>)*9ÿ>*ÿstcuctvwÿƒtacay„Iÿ ÿ ÿ ÿ 12ÿ ÿ � https://cpw.cvlcollections.org/files/original/314efd6d7ee2e73f5cda68806763627f.pdf 8230c6efcc93354742be6e6e1180bf8f PDF Text Text .... Colorado Division of Parks and Wildlife Ju ly2010 - June20 1 l WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Federal Aid Proj ect No. Colorado 3430 3003 Division of Parks and Wildlife Mammals Research Predatory Mammal Conservation Black bear exploitation of urban environments: finding management solutions and assessing regional population effects Period Covered: July I, 20 lO - June 30, 2011 Author: H.E. Johnson; project cooperators, C. Bishop, J. Brodrick, J. Apker, M. Alldredge, S. Breck, J. Beckmarm, K. Wilson, M. Reynolds-Hogland, T. Speeze, and P. Dorsey. - All information in this report is preliminary and subject to further evaluation. Infor·mation IVIA Y NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT -..., - -- -- Across the country conflicts among people and black bears are increasing in number, frequency and severity, and have become a high priority wildl ife management issue. Whether increases in conflicts reflect recent changes in bear population trends or just bear behavioral shifts to anthropogenic food resources, is largely unk nown, with key implications for bear management. This issue has generated a pressing need for bear research in Colorado and has resu lted in a unique collaboration that builds on the resources and abil ities of personnel from 5 entities: the Colorado Di vision of Wildlife (now Colorado Parks and Wi ldl ife [CPW]), the National Wildlife Research Center, Colorado State University, Wildlife Conservation Soc iety, and Bear Trust fnternational. Collectively, we are implementing a 5-year study on black bears that I) tests management strategies for reducing bear-human conflicts, including a large-scale treatment/control urban-food- removal experiment; 2) determines the consequences of bear-use of urban environments on regional bear population dynamics; and 3) develops population and habitat models to support the sustainable monitoring and management of bears in Colorado. We initiated this project in FYI 0- 11 by developing a research proposal, selecting a field site for detailed data collection (Durango, CO), coordinating with numerous entities (non-profit organizations, private citizens, and personnel from city, county, state, and federal government agenc ies) on fie ld logistics, and commencing several aspects of data collection (trapping and collaring bears, monitoring human-related bear mortalities, implementing DNA hair-snare protocols, monitoring garbage-related bear-human conflicts, and conducting mast surveys). Project collaborators will continue to seek additional funding to implement the remaining activities outlined in the research proposal (i.e., conduct an urban-food- removal experiment, increase the sample size of OPS collared bears, and acquire telemetry collars to test a translocation model). lnformation from this study will provide solutions for sustainably managing black bears outside urban environments, while reducing bear-human conflicts within urban environments; knowledge that is critical for wildl ife managers in Colorado and across the country. 139 �WILDLIFE RESEARCH REPORT BLACK BEAR EXPLOITATION OF URBAN ENVIRONMENTS: FINDING MANAGEMENT SOLUTIONS AND ASSESSING REGIONAL POPULATION EFFECTS - - HEATHER E. JOHNSON P.N. OBJECTIVES To conduct a study on black bears in Colorado that 1) tests management strategies for reducing bearhuman conflicts, including a large-scale treatment/control urban-food-removal experiment; 2) determines the consequences of bear-use of urban environments on regional bear population dynamics; and 3) develops population and habitat models to support the sustainable monitoring and management of bears in the state. SEGMENT OBJECTIVES 1. Develop a research proposal for internal CP W peer review and funding solicitation. 2. Consult with CPW personnel on potential study sites and compile key information about those sites including num bers ofreporied bear-human conflicts, publ ic land access, urban sanitation practices, harvest data, and urban development statistics. 3. Work with personnel from the City of Durango, La Plata County, the San Juan Public Lands Office (USFS/BLM), the Columbine and Pagosa USFS/BLM Ranger Districts, Bear Smart Durango, CPW South west Region, CPW Area 15, and private landowners on logistical fi eld considerations. 4. Initiate black bear capture and GPS coll aring efforts to collect data on bear movements, habitatuse patterns, and vital rates. 5. Track human-related bear mortalities and removals around Durango from tra nslocations, vehicle collisions, conflict mortalities and harvest. 6. Deploy bear hair-snares in an "urban" Durango sampling grid and a "wildland" Piedra sampling grid to obtain DNA for genetic mark-recapture analyses. Genotyped hair samples will be used to estimate population densities. 7. Collect data on natural food ava ilability for bears based on the mast abundance of gambel oak, serviceberry, chokecherry, hawthorne, pinyon pine and squaw apple. 8. M onitor the frequency of garbage-related bear-human conflicts within proposed treatment and control areas for an urban-food-removal experiment. - -- - V ~ -- 140 �'Cl INTRODUCTION 'cl Conflicts among people and black bears (Ursus americanus) are increasing nationwide (Hristienko and McDonald 2007), as the human population grows and urban development expands in and around bear habitat. State and federal wildlife agencies are responsible for both minimizing bear-human conflicts and maintaining and monitoring viable bear populations; two mandates that are proving to be incredibly challenging. Conflicts between bears and people can result in human injuries, property damage, and bear mortality (i.e. euthanasia), but despite increasing efforts from wildlife agencies to reduce conflicts, rates have been on the rise (Baruch-Mordo et al. 2008). Meanwhile, bear population parameters have been exceedingly difficult to estimate across large spatial scales (Garshelis and Hristienko 2006), and current population sizes and trends are largely unknown. As a result, management agencies are uncertain whether recent increases in bear-human conflicts reflect increases in the bear population or just bear behavioral shifts to anthropogenic food resources. Without a thorough understanding of the factors that drive nuisance bear behavior, and the relationship between conflict rates and bear dynamics, it has been difficult for wildlife agencies to initiate effective management practices. This issue has generated a pressing need for comprehensive bear research in Colorado, and resulted in the development of a detailed study proposal by Johnson et al. (2011 ; Appendix I). The proposal outlines a 5-year project on black bears that 1) tests management strategies for reducing bearhuman conflicts, including a large-scale treatment/control urban-food-removal experiment; 2) determines the consequences of bear-use of urban environments on regional bear population dynamics; and 3) develops population and habitat models to support the sustainable monitoring and management of bears in Colorado. Overall, this study should explicitly link bear movement and resource-use to population parameters, while rigorously testing an array of management techniques to reduce conflicts. This information should provide solutions for sustainably managing black bears outside urban environments, while reducing bear-human conflicts within urban environments; knowledge that is critical for wildlife managers in Colorado and across the west. During FY 10-11 we developed a research proposal, identified a study site, determined the logistics of collecting field data at that site, and initiated data collection. Field efforts focused largely on meeting research objectives 1 and 2, which will yield data that will eventually be used to address objective 3. Specifically, we captured and collared adult female bears, collected data on human-related bear mortalities, deployed and monitored hair-snares to collect DNA for estimating population size, tracked garbage-related bear-human conflicts, and collected data on natural food availability for bears. We report general summary information from recent fieldwork (1 May- 15 Sept 2011) in this progress report; detailed analyses of field data will occur during FY 11-12. While we have initiated data collection for several aspects of this project, collaborators will need to generate additional funding to conduct all research activities outlined in the proposal (i.e., conduct the urban-food-removal experiment, increase the sample size of GPS collared bears, and obtain telemetry collars to test a translocation model). STUDY AREA \id 'el To meet study objectives, a combination of detailed, site-specific field data, and statewide data will be required. For the information presented in this progress report, we focus specifically on the selection of a site for detailed data collection on bear resource-use, demography, and the effectiveness of urban bear-proofing. To make this determination we evaluated a suite of factors. We first identified urban areas in Colorado that reported the highest numbers of conflict-related bear mortalities, translocations, and public calls. From those cities, we then considered the quality and history of bear-human conflict reporting, current bear-proofing infrastructure, the feasibility of conducting a large-scale human-foodremoval experiment (based on current city waste management practices), the size of the urban-wildland interface, harvest management, and public land accessibility. Based on those factors, project collaborators 141 �decided that field efforts should be initiated around the urban center of Durango, Colorado (La Plata County). Durango consistently exhibits some of the highest numbers of bear-human conflicts in the state, conflict reports are regularly monitored by CPW Area 15 and Bear Smart Durango (a local non-profit organization), and unlike other areas experiencing high conflict rates, bear harvest was expected to be maintained at similar levels for the foreseeable future. Durango also had limited bear-proofing infrastructure, was the only city with a coordinated residential waste management system (all residential waste is removed by the city), and is largely surrounded by public land (USFS, BLM, CPW, City of Durango and La Plata County; Fig. 1). The city of Durango contains~ 17,000 people (within city limits) and sits at 1,985 m along the Animas river valley. The town is surrounded by mountainous terrain ranging in elevation from~ 1,930 to ~3,600 m, and is generally characterized by mild winters and warm summers that experience monsoon rains. Vegetation in the region is dominated by ponderosa pine, oak, pinyon-juniper, aspen, mountain shrub, and agricultural communities. Key forage species for black bears include gambel oak (Quercus gambelii), chokecherry (Padus virginiana), serviceberry (Ame/anchier alnifolia), hawthorne (Crataegus spp), squaw apple (Peraphyllum ramosissimum ), angelica (Angelica spp), sweet cicily (Osmorhiza spp ), cow parsnip (Herac/eum sphondylium) and waterleaf (Hydrophy//um spp ). Public land in the region is primarily managed by the San Juan National Forest, the Bureau of Land Management, Colorado Parks and Wildlife, La Plata County and the City of Durango. 'wl METHODS Logistical Considerations During fall and early winter FY l 0-11, we developed a research proposal for internal CPW review and identified a field site for collecting detailed bear habitat-use and demography data. In late winter and spring we worked with various entities around Durango to prepare to conduct fieldwork. We presented our research proposal to personnel from the U.S. Forest Service and BLM (San Juan Public Lands Office, Columbine Ranger District, and Pagosa Ranger District) and worked to develop an operating plan for capturing bears and deploying hair-snares on federal land. We also presented our proposal to staff from the City of Durango and La Plata County, and discussed access to their respective lands for meeting research objectives. Within CPW, we worked with personnel from Area 15 and the Southwest Region to identify initial capture and hair-snare sites, create a bear-human conflict mailbox for recording public calls, and clarify the research objectives relative to local management actions. Additionally, we solicited various entities for financial contributions to the project. Bear trapping and collaring, tracking of humanrelated bear mortalities, DNA hair-snare surveys, garbage-related conflict monitoring, and mast surveys were all initiated during summer 2011; the study proposal (Appendix I) provides detailed descriptions of these methods so we only briefly describe them below. Bear Trapping and Collaring To relate the habitat-use patterns of bears to their demographic trends, we captured and collared adult female bears. We specifically targeted adult females as they represent the reproductive segment of the population and should provide reliable inference to general demographic trends. Additionally, we can obtain information on multiple key vital rates from collaring a single sex-stage class, because, in addition to adult female survival (the vital rate with the greatest elasticity), collared females allow us to track fecundity and cub survival from winter den checks. While our long-term goal is to collar ~50 adult females (Appendix I), in the first year of the study we had the resources available to deploy 25 GPS collars (20 new Vectronics collars, 5 used Lotek collars). We targeted our trapping efforts within~ 12 km of the center of Durango to capture a cohort of bears that experience similar natural food availability, have anthropogenic food resources readily available, and encompass a range of habitat-use patterns relative to the urban-wildland interface. 142 'wl �From May through 15 September we used a combination of box traps and leg-hold snares to capture black bears (Jonkel 1993). We built smaller box traps than those previously used for bear research in Colorado (previously built traps are 0.91 x 0.91 x 1.83 m and weigh ~205 kg; newly constructed traps are 0. 71 x 0.66 x 1.83 m and weigh ~ 125 kg), allowing for increased mobility and flexibility in placement (Fig. 2). A detailed description of the capture and handling procedures is available in Appendix II. Traps and snares were baited with fish, fruit, human foods (at urban locations) and manufactured scents; they were set in the evening and checked the following morning. Adult female bears were fitted with a GPS collar, marked with a PIT tag, and had a tooth pulled for age verification. All other bears (except cubs) were uniquely marked with a PIT and ear-tag (a single small black tag). Bears were weighed, measured, and sampled for blood and hair. GPS locations from Vectronics collars were programmed to upload 4 locations/day through a satellite system, while locations from Lotek collars were manually downloaded in the field using a hand-held device from the ground or air (fixed-wing aircraft). Monitoring Human-Related Bear Mortalities Between I May and I 5 September 2011 we recorded all human-related black bear mortalities and removals in the vicinity of Durango. Mortalities and removals occurred from translocations, vehicle collisions, conflict-related euthanasia and harvest. For all bears removed from the study area we collected a hair and tooth sample and recorded the date, mortality/removal cause, location, bear age, sex, weight, and morphological measurements. Tooth samples will be used to age and genotype these bears so they can be incorporated into population density analyses. Hair-Snare Surveys To estimate the density of black bears around Durango we used a DNA hair-snare sampling scheme (Woods et al. 1999, Mowat and Strobeck 2000). We centered a 36 cell grid (576 kni) over Durango where each cell was 4 x 4 km in size and contained one snare. We sampled a total of3 l grid cells, dropping 5 cells along the outer edge of the grid where public or motorized access was prohibited (Fig. 3 ). Snares consisted of a scented bait hanging high in a tree, surrounded by barbed wire around a cluster of trees encircling the bait; when the bears climbed over or under the wire to investigate the bait, they left a hair sample on the barbed wire. On half of the snares we hung a single strand of barbed wire (50 cm high), and on the other half of the snares we hung two strands (50 and 20 cm high). Our goal with this design was to determine whether the additional strand of wire increased capture probability. Snares were deployed from June I to 14, and we conducted 6 weekly sampling occasions thereafter. On each occasion, we re-baited the snare (randomly baited with anise, strawberry, fish, or maple), and collected hair samples off all barbs. Each hair sample was uniquely catalogued according to the site, date, occasion, and barb number. Samples will be sent to the laboratory at Wildlife Genetics International for genotyping during fall 2011 and we will use the pattern of genotypes to estimate density using mark-recapture statistics. In addition to implementing the Durango hair-snare grid, we also conducted a pilot grid in the Piedra watershed (located between Durango and Pagosa Springs; see Appendix I Figure 7). This site was chosen as high quality ''wildland" bear habitat, reflecting representative densities of bears in the region in the absence of urban development and human food resources. Initially, we intended to deploy and monitor ~32 snares in both the Durango and Piedra grids, however, lack of motorized access in the Piedra watershed inhibited field crews from constructing and checking all snares in a timely fashion. As a result, we opted to run a subset of 9 snare sites in the Piedra to determine whether twice/month sampling (as opposed to weekly) would have significant impacts on DNA quality, DNA contamination (hair samples from > I bear/barb), and recapture rate. These samples will be genotyped this fall. Depending on the results, we will design an appropriate sampling scheme to estimate the wildland bear density in FYI 1-12. 143 �Mast Surveys Bear-human conflicts and bear-use of urban environments may increase when natural foods are in short supply (Zack et al. 2003, Baruch-Morda 2007, Baruch-Morda et al. 2010). To quantify the role of natural food availability on bear habitat selection, we initiated weekly surveys of the local soft and hard mast. In the Durango region, the key mast species for bears are gambel oak, chokecherry, serviceberry, hawthome, squaw apple, and pinyon pine (Beck 1991, Tom Beck, personal communication). Although the phenology of these species is variable throughout the late summer/early fall, they generally reach peak fruit or nut maturation between mid-August and mid-September. We randomly selected 12 transects throughout the 576 km 2 hair-snare study area to evaluate bear natural food availability (Fig. 3). Each transect was 1 km in length and ran along an existing public trail or public-accessible stream drainage. Field technicians walked vegetation transects each week between 15 August and 15 September and for each species, recorded the phenological stage and the percentage of plants that exhibited mast in different abundance categories (mast failure, <25% of plants with mast, etc). Conffict Monitoring One management strategy proposed for reducing bear-human conflicts is removing access to human foods for bears (Peine 200 I, Spencer et al. 2007). Given the high price to operationally "bearproof' a community, municipalities must have definitive evidence that such an effort would significantly decrease conflict activity before initiating major changes to waste storage and collection practices. As part of this study we plan on implementing the first rigorous scientific evaluation of the efficiency of widescale urban bear-proofing for minimizing bear-human conflicts. Although this portion of the project has not yet been funded, we conducted pre-treatment monitoring in proposed treatment and control areas (Fig. 4). During July and August, the months that experience the highest numbers of bear-human conflicts (CPW unpublished data) we patrolled each street within proposed treatment/control areas on the day that waste removal was scheduled to occur (when maximum human food was assumed to be available to bears). Patrols were conducted from 06:00- 07:00 AM; for all locations where there was evidence that bears had obtained garbage we recorded UTM coordinates and the trash container type. RESULTS AND DISCUSSION During the summer 2011 field season we conducted 92 total bear captures; 71 captures were unique individuals and 21 were recaptures (see map of capture locations in Fig. 3). Of the unique individuals captured, there were 30 females, 38 males, and three cubs of unidentified sex (cubs were released without being immobilized and thus, gender was not determined; Table I). The mean age of captured bears ~l year old was 4.9, and the mean weight was 80.9 kg (60.0 kg for females and 97.4 kg for males). In total, we placed traps/snares at 105 different locations and we had 1,253 trap nights. Across all bear captures (new captures and re-captures), 86 bears were captured using box traps (I, 119 box trap nights) and 6 with leg-hold snares (134 snare nights). Generally capture success peaked during the first couple weeks of June and again in mid-August; capture success was low during July. We modified our newly constructed, smaller box traps to have a locking mechanism on the door that, once triggered, only allowed the door to close shut and not re-open. This was a critical design element, and allowed us to use the smaller box traps to catch bears::; 214 kg. Generally, we found these traps to be convenient to place in the field and successful in safely capturing and holding bears until they were immobilized. We collared a total of 26 female bears, however two bears slipped out of their collars and were not recaptured leaving us with 24 collared bears at the end of the field season. During the trapping season, Vectronics collars successfully uploaded >5,000 GPS locations through the satellite system, and we downloaded an additional 1,500 locations from Lotek co11ars (Fig. 5). One Vectronics collar prematurely switched to low-battery mode in August; we are currently attempting to recapture the bear to replace the collar. Although we have not yet conducted any formal movement analyses, one collared female moved 144 �~50 km southwest from Perins Peak State Wildlife Area (adjacent to Durango), eventually moving back after several weeks. The second longest movement by a collared bear was ~ 16 km. Between I May and 15 September, 23 bears were removed from the greater Durango area due to human-related causes. Of those bears that were removed, three were translocated due to conflicts with people, seven were killed i•n vehicle collisions, one was killed during research trapping, and 12 were euthanized due to conflicts with people (breaking into house, killing livestock, etc). There were three cubs, two yearling females, five yearling males, six adult females and nine adult males that were removed. Until bears begin hibernating, additional mortalities and removals are expected to occur. Field crews collected a total of 998 individual bear hair samples, 743 samples from the Durango grid and 255 samples from the pilot Piedra grid. Over the 6 sampling occasions from 31 snares around Durango we collected 224, 167, 138, 77, 68, and 69 hair samples, respectively. Over the three sampling occasions from nine snares in the Piedra we collected 127 samples; 46, 50, and 31 samples/occasion, respectively. We also collected 128 additional samples from snares in the Piedra watershed that were only checked on a single occasion. Samples will be sent to Wildlife Genetics International for genotyping in the fall, and results will allow us to estimate bear density. ....,, 'Cl Within the proposed treatment and control areas for the urban bear-proofing experiment, we observed 129 incidences of bears accessing human garbage during July and August; incidences peaked during the first week of August. Of those events, I 0% were wildlife-resistant garbage containers and 90% were regular containers. Bears accessed human food from wildlife-resistant containers when they were not closed properly or could break the locking mechanism on the lid. In assessing the availability of garbage to bears, we recorded the location and container type of 1,167 garbage cans in the proposed treatment and control areas (Fig. 4 ). Of those containers, 14% were wildlife resistant and 86% were regular (non-wildlife resistant). This demonstrates the limited residential bear-proofing that currently exists in Durango, and the relevance of conducting an experimental test of wide-scale urban bear-proofing in this community. Mast surveys are currently ongoing; results will be in the annual report for FY 11-12. SUMMARY AND FUTURE PLANS -.;I -.I During FY 10- l l we successfully developed a research proposal addressing bear-human conflict issues in Colorado, selected a field site, coordinated with numerous entities (non-profit organizations, private citizens, and personnel from city, county, state, and federal government agencies) on field logistics, and initiated several aspects of data collection (trapping and collaring bears, tracking humanrelated bear mortalities, implementing DNA hair-snare protocols, monitoring garbage-related bear-human conflicts, and conducting mast surveys). We will continue these field activities during summers 20122015. Additionally, we will begin winter den checks in January 2012 to track fecundity and cub survival, and ensure that collars are fitting appropriately. Project collaborators will continue to seek additional funding to implement the remaining activities outlined in the research proposal. These activities include the implementation of an urban bear-proofing experiment, increasing the number of GPS collared female bears, and purchasing telemetry collars for a translocation study. In addressing the objectives of this project we hope to better understand the influence of urban environments on bear populations, elucidate the relationship between bear-human conflicts and bear population trends, develop tools to promote the sustainable management of bears in Colorado, and ultimately, identify solutions for reducing bear-human conflicts in urban environments. 145 �LITERATURE CITED Baruch-Mordo, S. 2007. B1ack-bear human conflicts in Colorado: spatiotemporal patterns and predictors. Thesis, Colorado State University, Fort Collins, Colorado. Baruch-Mordo, S., S. W. Breck, K.R. Wilson, and D.M. Theobald. 2008. Spatiotemporal distribution of black bear-human conflicts in Colorado, USA. Journal of Wildlife Management 72: 1853-1862. Baruch-Mordo, S., K.R. Wilson, D. Lewis, J. Broderick, J. Mao, and S.W. Breck. 2010. Roaring Fork Valiey urban black bear ecology study: progress report to the Colorado Division of Wildlife. Contact Sharon Baruch-Mordo for copy. Email: sharonb m@yahoo.com Beck, T.D.I. 1991. Black bears of west-central Colorado. Technical Publication No. 39, Colorado Division of Wildlife, Colorado. Garshelis, D.L., and H. Hristienko. 2006. State and provincial estimates of American black bear numbers versus agency assessments of population trend. Ursus 17: 1-7. Hristienko, H., and J.E. McDonald Jr. 2007. Going into the 21 st century: a perspective on trends and controversies in the management of the American black bear. Ursus 18:72-88. Johnson, H.E, C.J. Bishop, M.W. Alldredge, J. Brodrick, J. Apker, S. Breck, K. Wilson, and J. Beckmann. 2011. Black bear exploitation of urban environments: finding management solutions and assessing regional population effects. Research Proposal, Colorado Division of Parks and Wildlife, Fort Collins, USA. Jonkel, J.J. 1993. A manual for handling bears for managers and researchers. Office of Grizzly Bear Recovery, United States Fish and Wildlife Service, Missoula, Montana. Mowat, G., and C. Strobeck. 2000. Estimating population size of grizzly bears using hair capture, DNA profiling, and mark-recapture analysis. Journal of Wildlife Management 64: 183-193. Peine, J.D. 2001. Nuisance bears in communities: Strategies to reduce conflicts. Human Dimensions of Wildlife 6:223-237. Spencer, R.D., R.A. Beausoleil, and D.A. Martore11o. 2007. How agencies respond to human-bear conflicts: a survey of wildlife species in North America. Ursus 18: 217-229. Woods, J.G., D. Paetkau, D. Lewis, B.N. McLellan, M. Proctor, and C. Strobeck. 1999. Genetic tagging of free-ranging black and brown bears. Wildlife Society Bulletin 27 :616-627. Zack, C.S., B.T. Milne, and W.C. Dunn. 2003. Southern oscillation index as an indicator of encounters between humans and black bears in New Mexico. Wildlife Society Bulletin 31 :517-520. Prepared by _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ Heather E. Johnson, Wildlife Researcher 146 �..._ -al I.a/ ..., ... .... '-.I 'el '-.I 'cl '-.I .I .I ...,; '...,/ -.._; \SI "'111 -.I '-.I 'SI -..,) ...; '-' -..I -..I 'cl ,..,; 1..,1 '-' I._/ '-' le/ ...., 'cl '-' 'c;/ -al ~ -el \_,I ~ .._ Table 1. Capture information for 65 bears ~ l year old in the vicinity of Durango, CO . Unigue ID Ca~ture Date Bl 82 B3 84 B5 86 B7 88 89 810 Bl 1 B12 813 814 B15 B16 B17 B18 B19 820 B21 B22 823 B24 B25 826 827 B28 B29 830 B31 B32 833 834 835 836 837 838 B39 840 B41 B42 843 844 B45 846 847 5/10/201 I 5/12/2011 5/13/2011 5/16/2011 5/16/2011 5/17/2011 5/17/2011 5/18/2011 5/26/2011 5/26/2011 6/3/2011 6/2/2011 6/3/2011 6/6/2011 6/6/2011 6/7/2011 6/7/2011 6/8/2011 6/9/2011 6/9/2011 6/10/2011 6/10/2011 6/13/2011 6/14/2011 6/15/2011 6/15/2011 6/16/2011 6/16/2011 6/21/2011 6/22/2011 6/24/2011 6/24/2011 6/28/2011 6/28/2011 7/5/2011 7/6/201 I 7/7/2011 7/13/2011 7/13/2011 7/21/2011 7/22/2011 7/26/2011 8/3/2011 8/3/2011 9/3/2011 9/5/2011 8/8/2011 Sex M M M M M F F F M F. M M M F M M F F M M F M M F F M F M M F M F M M F M M M M F M F F M M F F Estimated Age Weight {k2) I 35 144 130 84 135 63 64 52 35 81 130 103 59 58 58 117 52 62 147 132 69 5 5 3 6 3 6 3 4 8 5 3 6 3 7 6 6 9 10 7 8 3 8 4 10 88 65 65 64 109 75 101 49 60 85 19 35 85 44 67 39 145 150 81 67 70 85 35 176 58 54 IO 6 4 4 1 3 3 4 1 8 6 5 3 6 8 2 6 3 8 147 Capture Location UTM Easting UTM Northing 246233 271495 271495 270950 270227 243210 243225 271478 238803 269869 252163 253216 253216 252157 253216 253216 256936 256918 235193 243258 252298 252163 246350 243252 239003 252164 243252 253233 239840 235911 239840 243252 239294 239001 246350 239840 243252 243236 251222 248550 237368 245945 246183 756124 245965 243435 251783 4142768 4130889 4130894 4127914 4139984 4128716 4133053 4130892 4126790 4139040 4137968 4137387 4138868 4137967 4138868 4138868 4134633 4134625 4128894 4133040 4136435 4137968 4135617 4133030 4134158 4137966 4133030 4138873 4126949 4128916 4126949 4133030 4133260 4134154 4135617 4126949 4133030 4128710 4133120 4131645 4132272 4141391 4142791 4132494 4139587 4128720 4131581 �Table I-Continued Unigue ID Ca2ture Date B48 B49 B50 B51 B52 B53 B54 B55 B56 B57 B58 B59 B60 B61 B62 B63 B64 B65 B66 B67 B68 8/10/2011 8/11/2011 8/11/2011 8/12/2011 8/12/2011 8/15/2011 8/16/2011 8/18/2011 8/29/2011 8/30/2011 8/31/2011 9/1/2011 9/2/2011 9/3/2011 9/6/2011 9/7/2011 8/6/2011 9/15/2011 9/20/2011 9/21/2011 9/21/2011 Sex F F F F F M M F M F M M F M Estimated Age We~ht{ke} 1 3 7 12 4 7 2 3 26 55 101 62 65 163 53 49 167 46 48 153 35 214 23 37 30 91 41 54 209 10 3 3 15 2 7 M M 2 M F F F M l 5 1 3 8 148 Capture Location UTMEasting UTM Northing 245914 243435 245965 249049 245965 243435 251898 251464 246321 243374 243374 243952 242187 244602 245790 248612 245850 243948 240731 256930 249067 4139620 4128720 4139587 4130370 4139587 4128720 4130516 4134423 4132993 4135903 4135903 4132935 4133020 4130321 4128530 4131251 4141969 4134848 4130163 4134626 4133006 �Figure. I. Land ownership in the vicinity of Durango, CO. - - LANDOWNER ) c=J USFS --- - C=1 BLM c=J corw c=J La Plata County 11111 City of Durango c=J State Land Board c=J Private Ute Tribe N 0 1.5 3 6 Kilometers A 149 �Figure 2. Photos of a newly designed box trap to capture black bears. 1w - - 150 �Figure 3. Location of bear hair-snare sites, mast survey transects, and capture sites around Durango, CO. - "1111 -Masting Plant Surveys -- N A 0 2 4 Hair Snare Locations 0 Capture Locations 8 Kilometers -- - 0 15 l �Fig ure 4. Proposed treatment and control areas for an urban bear-proofing experiment and observations of garbage-related confl icts from pre-treatment monitoring. Red stars indicate evidence of bears foraging on human garbage, circ les indicate the availability of human food for bears (green circles represent regular garbage containers and yellow circles represent wildlife-resistant containers). Garbage Conflict 0 Regular Container 0 Wildlife Resistant Container i ol..,__ ___ o.,_ .5_ _ _ _.1,___ _ __ ....___ _ ___,2 Kilometers 152 - �Fig ure 5. Ad ul t female black bear GPS locations collected between May and September 20 11 in the vicinity of Durango, CO (different colored circles represent different individual bears). N A 0 5 10 153 20 Kilo meters �APPENDIX I PROGRAM NARRATIVE STUDY PLAN FOR MAMMALS RESEARCH FY 2011-12 - FY 2015-16 State of: Cost Center: Work Package: Task No. Federal Aid Project No. Colorado --------3430 --------3003 --------- Division of Parks and Wildlife Mammals Research Predatory Mammal Conservation Black bear exploitation of urban environments: finding management solutions and assessing regional population effects Black bear exploitation of urban environments: finding management solutions and assessing regional population effects Principal Investigators Heather Johnson, Mammals Researcher, Colorado Parks and Wildlife Chad Bishop, Mammals Researcher Leader, Colorado Parks and Wildlife Mathew Alldredge, Mammals Researcher, Colorado Parks and Wildlife John Broderick, Terrestrial Programs Leader, Colorado Parks and Wildlife Jerry Apker, Carnivore Coordinator, Colorado Parks and Wildlife Stewart Breck, Research Wildlife Biologist, National Wildlife Research Center Kenneth Wilson, Professor, Colorado State University Jon Beckmann, Associate Conservation Scientist, Wildlife Conservation Society w '_.I 'w Cooperators Melissa Reynolds-Hogland, Executive Director, Bear Trust International Tom Spezze, Southwest Regional Manager, Colorado Parks and Wildlife Patt Dorsey, Area Wildlife Manager, Colorado Parks and Wildlife STUDY PLAN APPROVAL Prepared by: Heather Johnson Date: 2/15/2011 Submitted by: Heather Johnson Date: 3/12/2011 Reviewed by: Jon Runge Date: 3/25/2011 Chuck Anderson Date: 3/14/2011 Danny Martin 4/4/2011 Biometrician: Paul Lukacs Date: 3/10/2011 Approved by: Chad Bishop Mammals Research Leader Date: 3/10/2011 'wl 154 �-- 1w - PROGRAM NARRATIVE STUDY PLAN FOR MAMMALS RESEARCH .... Black Bear Exploitation of Urba n Environments: Finding Management Solutions and Assessing Regional Population Effects A Research Proposal Submitted bv: Heather Johnson. Mammals Researcher. Colorado Parks and Wildlife Chad Bishop, Ma111111als Researcher Leader, Colorado Parks and Wildl{fe Mathe w Alldredge, Mammals Researcher, Colorado Parks and Wildlife John Broderick. Terrestrial Programs Leader, Colorado Parks and Wilr/lffe Jerry Apker, Carnivore Coordinator, Colorado Parks and Wildlife Stewart Breck. Research Wildlife Biologist, National Wildlife Research Center Kenneth Wilson, Professor, Colorado State Universily Jon Beckmann, Associate Conservation Scientist, Wildl{le Conservation Society J. Need Conflicts among people and black bears (Ursus americanus) are increasing nationwide, as the human population grows and urban development expands in and around black bear habitat. In a survey of 4 1 state wild Ii Fe agencies that manage black bears, 30 repo1ied increasing numbers of bear-human confli cts in recent decades (Hristienko and McDonald 2007). Wh ile state and federal wildlife agencies are responsible for minimizing bear-human conflicts, they are also responsible fo r maintai ning viable bear populations. Achieving this balance is proving to be difficult, as agencies struggle to find effective management solutions while conflict rates continue to rise, particularly around urban areas (Tavss 2005, Baruch-Mordo et al. 2008). Whethe r increases in conflicts reflect recent changes in bear population trends or j ust behavioral sh ifts to anthropogenic food resources, is largely unknown, as bear population parameters have been exceedingly difficult to estimate (Garshelis and Hristienko 2006). The primary cause of bl ack bear-human confli cts along the urban-wildland inte rface has been attributed to the availability of anthropogenic food resources to bears (Fig. I; Spencer et al. 2007, Beckmann et al. 2008, Greenleaf et al. 2009). Urban areas contain a wealth of reliable, high-calorie foods, in the form of garbage, fru it trees, vegetable gardens, pet food, and bird feeders. As opportunistic foragers, bears readi ly ex ploit these resources, resulti ng in negative interactions with people. These interacti ons, however, have been highly temporally and spatially variable (Baruch- Morda et al. 20 I 0), generating uncertainty about the relative infl uence of natural food availabil ity, confl ict management, harvest, and bear population trends on driving annual variation in rates of bear-human con fl icts. Without a thorough understand ing of the factors that exacerbate nuisance bear behavior, and uncertainty about the relationship between conflict rates and bear dynam ics, it has been difficul t for w ildlife agencies to initiate effective management practices. -~---. -- -- - Bear use of the urban environment has serious consequences for people, bears, and wildlife managers. For people, bear-human conflicts lead to increased public safety concerns, property damage, and high management costs, while fo r bears they lead to increased mortal ity (Beckmann and Berger 2003, Beckmann et a l. 2008, Hostetler et al. 2009). For 155 Figure I. Black bearforaging on urban food resources. �example, in 2007 Colorado data analysis unit (DAU) 8-11 reported >500 public safety and property damage conflicts with bears, resulting in >$500,000 expended by the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) in bear management. This is one of 19 bear DAUs in Colorado, and encompasses the towns of Aspen and Vail, which have been hotspots of bear-human conflicts. That year, in 8-11 alone, 44 bears were euthanized for conflict control, 25 were translocated for nuisance behavior, 27 died of road kill, and 30 were legally harvested. Overall, this resulted in >75% of bear mortality attributed to conflicts with people, with unknown consequences for local bear populations. In addition to high management costs, these extreme conflict solutions have critical repercussions for wildlife management agencies. Because managers are obligated to respond to conflict calls, conflict management usurps limited resources and radically reduces those available for other programs. High conflict rates and unpopular management activities (i.e. lethal bear removals) also degrade the credibility of wildlife agencies to the general public, and ultimately reduce the inherent value of black bears in the public eye (Will 1980). Given expected changes in both human development and climate patterns, bear-human conflicts should rise in the future. As the human population grows, development will continue to permeate bear habitat, creating additional opportunities for conflicts with bears (Kretser et al. 2008). This situation will likely be exacerbated by anticipated changes in annual weather patterns. Drought conditions reduce the availability of natural foods for bears and are associated with an increase in bear-human conflicts (Zack et al. 2003, Baruch-Mordo 2007). Drier, wanner weather, as predicted with climate change, is expected to escalate conflicts with bears in the coming years. ► Identify management strategies to reduce bear-1,uman conflicts Ultimately, the public will not tolerate ever-increasing conflicts with bears and wildlife agencies must find effective solutions to resolve this pressing problem. Yet, despite the trajectory of increasing black bear-human conflicts, and the severe consequences of those conflicts for both people and bears, best management practices for reducing conflicts remain unclear. Managers commonly employ education (Gore et al. 2008, Baruch-Mordo et al. 2011), aversive conditioning (Beckmann et al. 2004, Mazur 2010), and increased harvest (Treves et al. 20 I0) to curb conflict rates, yet when the effectiveness of these strategies has been scientifically tested, they have been found to be largely ineffective as implemented. Investigators have suggested alternative approaches for reducing conflicts, such as reducing the availability of anthropogenic food for bears, using models to increase translocation success of nuisance bears, and altering public hunting programs to be spatially or temporally aligned to remove nuisance bears. These techniques may be useful for reducing conflicts, but their efficacy has not been rigorously tested. Removing anthropogenic food - Given that bears are attracted to anthropogenic food it is believed that eliminating the availability of this resource will dramatically reduce nuisance bear behavior (Peine 2001, Beckmann et al. 2004, Gore et al. 2005, Lyons 2005, Spencer et al. 2007). This strategy has had some success within national parks (Greenleaf et al. 2009), and anecdotally in some communities (Mammoth Lakes CA, Juneau AK, Whistler BC), but no research has ever scientifically tested the costs and benefits of"cleaning up" a town. Given the high price to operationally "bear-proof' a community, municipalities must have definitive evidence that such an effort would significantly decrease conflict activity before initiating major changes to waste storage and collection practices. A thorough, rigorous evaluation of this approach would provide guidance to wildlife agencies and municipalities on the benefit of investing in bear-proofing infrastructure. Trans location Suitability Modeling - Translocation of nuisance black bears is another common management technique that has been applied with varied results (Rogers 1986, Linnell et al. 1997, Landriault et al. 2009). Often bear translocation decisions are handled by field managers without formal guidance. These professionals are knowledgeable on bear capture and transport techniques, but often lack the flexibility to release bears in other management areas without obtaining approvals from different managers, who are often also experiencing nuisance bear problems. Limitations in selecting a 156 �translocation site and the profound movement ability of bears can result in an unsuccessful translocation where the bear continues to cause conflicts either in its new location or after returning to the capture site. To improve bear management, a strategic translocation approach is needed that applies the best available science on bear habitat quality, conflict potential, and harvest in the selection of bear release sites, while incorporating statewide collaboration among managers. Targeted Bear Hunting - Wildlife managers frequently increase harvest quotas to reduce bearhuman conflicts, but the scientific literature has been equivocal on the effectiveness of this approach (Obbard et al. 1997, Hristienko and McDonald 2007, Treves et al. 2010). Hristienko and McDonald (2007) found that states with higher harvest rates reported fewer conflicts, while other studies evaluating elevated harvest on region-specific spatial scales have concluded either no effect or increases in numbers of conflicts (Obbard et al. 1997, Tavss 2005, Treves et al. 2010). Lack of harvest success has been largely attributed to a mismatch between the timing and location of bear-human conflicts and the timing and location of the hunt, as bear-human conflicts peak during summer months along the urban interface while public hunting occurs during the fall in areas away from development (Treves et al. 20 I 0). As a result, a general increase in harvest likely translates into a reduction in the population at large, not necessarily the removal of nuisance bears. This strategy also inherently assumes that conflict rates reflect bear population sizes, an untested assumption that could potentially lead to overexploitation. To determine whether public harvest can successfully curb conflict rates, hunts need to be spatially and/or temporally coordinated with conflicts as they occur. While this is a strategy that has the potential to reduce management-related conflict mortality, it has yet to be thoroughly evaluated. ► 'Cl IC,/ >cl 'ell -..I Elucidate tl,e dynamics of bear populations along tl,e wild/and-urban interface To sustainably manage bear populations in the face of a growing human population and changing landscape conditions, it is critical to elucidate the dynamics and drivers of bear populations. Of those factors that influence bear dynamics, the contribution of urban environments is the least understood, most contentious, and has the greatest potential to elicit major population change. While urban environments offer bears the benefit of anthropogenic food, they also inflict the cost of increased mortality from lethal removals, translocations, and other urban factors (i.e. road kills), yielding uncertainty about whether urban environments contribute to the growth or decline of local bear populations. In the two studies that have evaluated bear populations along the wildland-urban interface, bears experienced reduced survival with population-level consequences (Beckmann and Berger 2003, Hostetler et al. 2009). In Florida, Hostetler et al. (2009) found that reduced adult survival caused the "urban" bear population to decrease in size, while the adjacent "wild" population increased, demonstrating the possibility of source-sink dynamics. Meanwhile, in Nevada, Beckmann and Berger (2003) found that bears around urban development were present at higher densities and had greater reproductive rates, but cubs had exceedingly low survival. The researchers suggested that urban areas did not just operate as a sink but as an ecological "trap" as human food attracted bears into town only to lead to their demise and depopulate the adjacent wildlands. While these studies suggest that urban environments may reduce bear populations, many management agencies have assumed that increasing conflicts reflect increasing populations, and that the availability of anthropogenic foods has bolstered demographic rates. So, do urban areas serve as population sources or sinks for bears, and are these impacts static or do they vary under different conditions? Do urban environments operate as ecological traps, attracting bears into habitat that is maladaptive when suitable conditions exist elsewhere? This question is complicated by the influence of annual variation in natural foods, or environmental stochasticity, on bear behavior and demography. While Beckmann et al. (2004) and McCarthy and Seavoy (1994) report that bears habituated to anthropogenic foods regularly return to them, preliminary data from Aspen, Colorado also suggests that bears increase time spent in urban environments in years of natural food failure and decrease that use when natural foods are readily abundant (Fig. 2; Baruch-Morda et al. 20 I 0). This pattern implies that bears may avoid urban environments when conditions allow, despite the common assumption that a bear savvy to anthropogenic 157 �Figure 2. Annual distances between the home range and the center of town.for a collared adult female bear in a good natural f ood year when she had no cubs (2005), a had natural.food year when she had no cubs (2007), and a good natural food year when she had cubs (2008; from Baruch-Morda et al. 2010). foods w ill consistently be a "conflict bear." In a state like Colorado, where human development has effectively permeated almost all tracks of prime bear habitat, the consistency of bear foraging behaviors has key implications for managers. For example, ifa small subset of bears consistently causes a majority of the conflicts with people, then the removal ofa few key indi viduals should alleviate the problem. If, however, high rates of conflict coincide with years of natural food fa ilure because a large proportion of the population is seeking alternative food resources, such a removal strategy may be ineffective. O r perhaps a combination of these hypotheses are true, that a subset of bears cause a majori ty of confli cts unti l a food-failure " teaches" a new group of bears to use human foods, a pattern that is then repeated in subsequent years, despite natural food conditions. Currently, managers have no information about the proportion of bears that cause conflicts, how the use of urban resources varies among individuals, and how variation in the availability of natural foods drives temporal variation in urban resource-use. As agencies struggle to de fine conflict management practices with minimal information on population trajectories, unde rstanding the effects of urban e nvironments on bear demography is critica l. Currently, conflict bear management practices (lethal removal and trans locations) are based on several inherent assumptions such that I) there is a correlation between bear-human conflicts and bear population size, 2) conflicts are caused by a few individual bears and their removal will alleviate local problems, and 3) management removals do not signi licantly influence regional bear dynamics or local harvest opportun ities. The validity o f these assumptions have yet to be dete1111ined, despite their importance for bear management. To develop sustainable management practices for black bears, we must tease apart the relati ve influences of annual variation in natural bear foods, the availability of an thropogenic foods, conflict-management (lethal remo vals and translocations) and harvest on bear dynamics and bear-h uman conflicts. ► Develop better tools to m onitor the dynamics and drivers of bear pop11/atio11s Despite the need to understand the drivers and trends of bear populations to direct management, Garshelis and Hristien ko (2006) fo und that most states have limited data from which to make sound decisions. As a result, state agencies rely on coarse harvest indices that yield little power for detecting population change, and no ability to di still the underlying causes of change. New tools that increase the scientific rigor in monitoring bear populations are desperately needed, so that harvest quotas are biologically-based and designed to meet population objectives. Recent advances in wildli fe statistics have foc used on max imizing the use of traditional age/sexat-harvest data, that which is routinely collected during mandatory harvest reporting. New techniques are available to more effectively extract information about population trend from harvest data (Skalski et al. 158 - �2007) and can be augmented with mark-recapture or radio-telemetry data to increase precision in parameter estimation (Fieberg et al. 20 I0, Johnson et al. 2010). While these approaches hold tremendous promise for supporting biologically-based bear monitoring and management, they are still in their infancy and have yet to be widely implemented. These techniques could be used to identify the value of different data types for tracking populations and to allocate field efforts that most efficiently determine bear population trends across a region of interest. Such information could also be used to inform annual harvest recommendations, elucidate statewide bear dynamics, and reconcile the relationship between bear population trends and conflict rates. -...I K. Objectives 1) Test management strategies to red11ce bear-1,uman conflicts. Bear-human conflicts in urban areas of Colorado echo nationwide trends, as they are increasing in number, frequency, and severity, and have become a high priority management issue in all regions of the state (Baruch-Mordo et al. 2008, Colorado Division of Wildlife unpublished data). In evaluating strategies to reduce conflicts we will: 1A) Experimentally reduce the availability of anthropogenic food to bears in an urban environment to assess the effect on bear-human conflicts and bear behavior. 18) Develop and evaluate a strategic statewide plan for the translocation of nuisance black bears. 1C) Assess a spatially-targeted bear harvest program designed to reduce the nwnber of nuisance animals. '..al "el 2) Determine the influence of urban environments on regional bear population dynamics. According to the 2010 U.S. Census, Colorado is the ninth fastest growing state in the country, with associated increases in housing and development (Mackun and Wilson 2011 ). Despite these trends, there is substantial uncertainty about the effects of urban habitats on bear habitat selection and population dynamics. To elucidate the effects of urban environments on bears we will: 2A) Evaluate the role of annual variation in natural foods on bear movement and resource-use. 2B) Estimate vital rates of urban and wildland bears relative to their resource-use patterns. 2C) Quantify the effects of resource-use, conflict bear management (lethal removals and translocations) and harvest on bear demography. 3) Develop pop11lation and habitat models to support the sustainable management of black bears in Colorado. Bear populations have been notoriously difficult to monitor for state wildlife agencies (Garshelis and Hristienko 2006). While meeting other project objectives we will obtain key biological data on bears from which we can: 3A) Use multiple data sources (harvest, DNA mark-recapture, and telemetry data) to develop improved bear population models to guide harvest regulations and inform estimates of population size and trend. 3B) Build regional habitat models to better predict bear density, direct the location of future monitoring efforts, and identify key seasonal resource areas. L. Expected Results or Benefits 'Cl This will be one of the most comprehensive studies to date on bear-human conflicts and the ecology of urban and wildland bears, resulting in crucial information that will be used to manage black bears in Colorado and across the country. Results from this study will: • Quantify the relative effectiveness of different management strategies (anthropogenic food removal, translocations, and spatially-targeted harvest) for reducing bear-human conflicts, information which will be broadly used by wildlife managers. A reduction in bear-human conflicts will ultimately increase public safety, reduce property damage, decrease wildlife management costs, and gain management credibility for collaborating agencies. 159 �• Identify key differences in the de mographic and behavioral patterns of urban and wildland bears to better in form managers about the efficacy of conflict-bear management (lethal removals and translocations) on bear behavior and population dynamics. For example, tl1is study will e lucidale the proportion o f bears using urban food resources, how that proportion varies due to natural food conditions, the relati onship between population perfomiance and conflict rates, and whether "town" serves as a source, sink, or ecological trap. • Provide robust, data-driven population and habitat model s to guide the monitoring and management of bears in Colorado. These models will be used to inform annual harvest regulations, revise statewide estimates of population size and trend, and direct the location of future data collecti on efforts. S uch information will increase the scientific rigor that is applied to the management of bears in Colorado and ensure that management actions to minimize conflicts are consistent with population o bjecti ves. • Advance theory and statistical methodology for linking resource-use patterns of animals to their demographic rates, and ultimately, population growth. To date, ha bitat and de mographic analyses have been largely conducted independently of one another, with a re lationship that is often inferred rather than directly measured. Using intensive field populati on data and GPS collar locations, this study will expli citly link space-use, resource acq uis ition, and demographic patterns, exploring new concephral and statistical avenues to elucidate their relationshi ps. M. Approach :.-; ( 'J , '..::/', i.. I r- :/ ' IA) Reducing the availability ofanthropogenic joods to bears in an urban environment to assess the effect 011 bear-human conflicts and bear behavior. . ,:;:. To test the effectiveness of reducing the availability o f human food on reducing bear-human conflic ts, we will conduct a large-scale experiment. We will drastically reduce the accessibili ty o f anthropogenic foods known to attract bears (garbage, bird-feeders, pet food, etc) within a designated ' treatment' area, while simultaneously monitoring comparable 'control' areas where no action w ill occur. We w ill perfom1 this experi ment in Durango, Colorado, a town with one of the highest bearhuma n conflict rates Ln the state as 200-900 conflicts were annually reported between 2007 and 20 IO (see Fig. 3). This town has abundant human food resources available to bears and a definitive urbanwi ldland interface, where urban developme nt is juxtaposed Lo high quality bear habitat. - Within Durango we will specify a treatment area and 2 control areas focused on the core zones of bear-human conflicts (Fig. 3). Each area wi ll contain Figure 3. locations ofbear-human conflicts in Durango, Colorado from 2007-2010 are shown with yellow circles and proposed treatment and control areas are represented by black boxes. approx imately 500 structures (residences and businesses) and be roughly the equivalent in size 2 (0.6 km ). T he treatment will occur in north west section of town, where the highest numbers of conflicts have been reported. In the treatment zone we wi ll provide bear-proof garbage containers, 160 -- �canvass citizens to discourage food availability outside of secure structures (bird-feeders, pet food, etc), conduct daily patrols to remove human food and provide strict enforcement. Our primary control area will occur on the south side of the Animas River (a moderate barrier to bear movement), to facilitate independence among experimental units. Additionally, we wiH monitor a second "spil1over" control area, adjacent to the treatment (north of the river) to measure the influence of the treatment on human behavior in adjacent neighborhoods. We wiH monitor treatment and control areas for 1 pre-treatment year and 4 post-treatment years, measuring changes in two key response variables: bear-human conflicts and bear behavior. We will define a "conflict" as any bear-human interaction that results in property damage or a threat to public safety, and compare the number of conflicts and their severity (i.e., a bear in a garbage can versus a bear breaking into a house) between treatment and control areas. Currently, citizens report conflicts to the Colorado Division of Wildlife, the non-profit organization BearSmart Durango, and the city newspaper; we wiH compile data from all sources for analysis. During the months bears are active, we will also conduct weekly patrols of treatment and control areas. Patrols will occur the morning that residential trash is collected, with an observer recording visible human food resources available to bears and evidence that bears have obtained human foods (i.e. trash cans knocked over and strewn garbage). We will use conflicts from treatment and control areas, and from pre- and post-experiment implementation to measure the effect of bear-proofing on the number and severity of urban bear-human conflicts. Additionally, we will monitor the influence of the food removal treatment on bear behavior. Bears Jiving on the urban-wildland interface wi11 be collared with global positioning system (GPS) satellite technology (see Objective 2 for capture and collaring details). GPS collars will automatically record the location of each bear every 4 hours, and we will use locations to conduct detailed resource selection analyses (Manly et al. 2002). Using selection indices from "in town" bear locations, we will test for differences in bear use among treatment and control areas (Blomquist and Hunter 2010, Boyce et al. 2010), and whether such use varies over the course of the active bear season. By tracking bear locations relative to our treatment and control sites, we should be able to quantify the benefits of 'cleaning-up' a community for reducing conflicts and modifying bear behavior in urban environments. JB) Developing and evaluating a strategic plan for the translocation ofnuisance black bears. To develop a strategic, statewide translocation plan, we will use existing information on black bears to map relative habitat quality, resource selection, nuisance potential, and hunter harvest potential across Colorado. These factors will be combined to generate a single layer depicting overall translocation suitability. Nuisance bears will then be allocated to release sites based on this suitability rating and their respective age, sex, reproductive status, management history (i.e. whether the bear was hazed), and distance to capture site. We will compare the success rates of bears translocated using the strategic approach with those of bears translocated following existing procedures, with success defined as a bear that does not engage in new conflict behavior. In all cases, bears will be marked using very high frequency (VHF) or GPS collars to quantify movements and fates following translocation. Additiona1ly, we will augment information from newly captured bears with data from >80 bear translocations that have already occurred in Colorado. Translocation success will be analyzed in a known-fate, time-to-failure framework (Hosmer et al. 2008), where the translocation outcome is modeled as a function of the relevant covariates. If our strategic approach increases translocation success our model will be incorporated into a user-friendly, internet-based tool for wildlife managers to assist with translocation decisions in the field. Specifically, a wildlife manager would enter the bear characteristics and capture site into the internet program and be given a set of optimal release sites. When a bear is released, the wildlife manager would enter the date and location of release into the program, which would be used to update subsequent releasesite decisions. 161 �1C) Assessing a spatially-targeted bear harvest strategy designed to reduce nuisance animals. Managers in southeast Colorado are responding to high numbe rs of conflicts by increasing harvest rates, however, they are using a novel approach. Rather than implement unit-wide increases in ha rvest quotas, managers will be spatially targeting hunting pressure in zones adj acent to conflict hotspots. These new harvest management zones are ex pected to be implemented in fa ll 2 011 with the goal of reducing bear densities in a reas bordering the urban interface (see example in Fig. 4). We w ill measure the success of this strategy for reducing conflicts in the communities of Colorado Springs, Pueblo, and Colorado City; cities which re po11 hundreds of conflicts/year. Using nuisance reports from pre- and postimplementation of this stra tegy, we wil l compare the number of conflicts, conflict severity, and numbers o f trans located and euthanized bears. Colorado Division of Wildlife has recorded these metrics for the past 16 years, a nd w ill continue to collect this data in the future. In addition, we will compare harvested numbers of bears in the DA Us in which these cities are located (82 and 8 7) pre- and post-implementation of the new strategy, to determine its effect on meeting annual bear harvest obj ectives. With 2:3 replications of this approach (around different urban centers) we will examine whether a spatiall ytargeted harvest approach, executed by the public, significantly decreases urban bear conflic ts w hil e increasing hunting opportuni ties. 2A) Evaluating rhe role ofannual variation in natural.foods on bear movement and resource-use. While anthropogen ic food is consistently avai lable to bears in urban e nvironments, the availability of natural foods can dramatically fluctuate based on annual patterns in temperahire and precipitat ion. For example, late frosts and summer droughts can cause fai lures in the local berry and acorn resources, fo rcing bears to expand their search for calories and potentially increase their use of urban environments (Zach et a l. 2003, 8 aruch-Mordo 2007, 8aruch-Mordo et al. 2010). To determine the influence o f annual variation in natural foods, or e nvironmental stochastic ity, on bear habitat-use we will evaluate location data from GP S collared adult females. From June through September, we will capture bears using c ulvert traps, box traps and A ldrich snares following the techniques described in Jonke! ( 1993, Appendix 2). Captured adult famles will be fitted with a Vectronics collar with a degradable spacer, ear-tagged, weighed, and measured fo r morphometric characteristi cs. Additionally, we will pull a tooth for age determination and obtain a blood sample for DNA. See Detailed capture and handling protocols are provided in Appendix 2. Each year we will attempt to maintain a sample of 50 collared females, with approximately ha!f collared in and arotmd the town of Durango ( La Plata county), and the other half in the surrounding w ildlands (La Plata, Hinsdale, and Archueta counties). This ~ Bear Manogemenl Ama Figure 4. The hatched-blue area represents the proposed bear conflict harvest zone on the wildland-urban inte,:face near Pueblo, CO. 162 - �collaring strategy will allow us to track a range of resource-selection patterns of bears, from those that are heavily dependent on human foods to those that rely exclusively on natural foods, quantifying the proportion of bears using urban resources and their frequency of urban habitat-use. -.,I We will also use GPS location data to examine resource selection and movement patterns in response to temporal variation in natural food availability (see Figs. 2 and 6). To do this we will partition location data into weekly intervals and use a repeated-measures resource selection function (RSF) approach (Manly et al. 2002, Borger et al. 2006, Kie et al. 2010, Mcloughlin et al. 2010). We will determine· those factors that drive temporal resource selection, evaluating the availability of natural foods, changes in weather patterns, distance to town, reproductive status, and conflict management history (i.e. whether the animal was hazed, trapped, etc). We will also evaluate the influence of these covariates on the size ofbear home-ranges and their rates of seasonal movements (Jonsen et al. 2005, Morales et al. 2010). Additionally, we will employ a time-to-failure analysis to examine those covariates (listed above) that predict when a bear will "fail" and use urban resources (Cook and Lawless 2007, Hosmer et al. 2008). We will work with colleagues in the Remote Sensing/Ecology program at Colorado State University to develop satellite image signatures to track annual vegetation productivity for natural bear foods. To quantify weather patterns, we will use PRISM spatial data (http://www.prism.oregonstate.edu/) which interpolates monthly temperature and precipitation patterns across landscapes, accounting for elevation and topography. All covariates related to human development will be extracted from existing CPW digital data layers. Ultimately, these analyses will not only allow us to summarize patterns of movement and resource-use, but elucidate the underlying drivers of bear behavior, providing insight for the design of better management strategies to minimize conflicts. 2B) Estimating the vital rates of urban and wild/and bears relative to their resource-use patterns. To assess differences in the population dynamics of those bears that use urban food resources versus those that do not, we will track the demographic trends of female bears collared in adjacent urban and wildland habitats. We are concerned with the vital rates (survival and reproductive rates) of female bears, as they represent the reproductive segment of the population and should provide reliable inference to the population at large. We will monitor ~50 GPS collared bears each year for their annual survival, fecundity, and the survival of their cubs; collecting this data for a total of 5 years. Survival of adult females will be tracked with real-time GPS locations, and all mortalities will be immediately investigated. To estimate annual fecundity and cub survival, we will inspect the winter natal dens of collared females for the presence of newborn and yearling cubs. If a newborn cub is observed with an adult female in year t, but is not observed in the den with that female in year t+ I, we will assume the cub is dead (Obbard and Howe 2008). Based on power analyses, our target sample size and study timeframe should allow us to detect biologically significant differences among the demographic rates of bears that use urban and wildland habitats, while still being logistically feasible (Fig. 5). In conducting power analyses to detennine samples Adult Female Survival Fecundity l 09 • 1 ~ 0.9 · 0.8 0.7 a: 0.6 11,1 ~ 0.5 0 g_ 0.4 0.3 0.2 0.1 o:s j 0.7 ffi o.6 1 ~ 0.5 1 -a-20% Decrease -.-1S%Decrease -10¾0K-rP;m• 0 15 25 35 45 ~ 0.4 1 -,!,-40%1ncrease 0.3 7 -11-30% Increase 0.2 1 0.1 ~ ~200/21ncrease 0 .,_!----,-----.----......----, 40 30 Sample Size/Group so 60 Sample Size/Group 163 70 Figure 5. Power to detect significant differences (alpha = 0.05) in vital rates between bears using urban habitats and those that do not, based on the sample sizes ofeach group. �sizes for adult female survival, we assumed a baseline survival rate of 0.90 with a standard deviation of 0.20 (Beck 1991, Koehler and Pierce 2005, Obbard and Howe 2008, Hostetler et al. 2009). The only study that has measured differences in adult female survival between urban and wildland bears found a 20% reduction in the survival of"urban bears" (Hostetler et al. 2009). With a sample of ~50 collared bears/year (~25 in urban habitat and ~25 in wildland habitat) for 5 years, we should have power~ 0.8 to detect at least a 15% difference in the survival of those bears that use urban habitats and those that do not. Similarly, assuming that adult female fecundity is 0.44 (Beck 1991) with a standard deviation of0.25 (Hebblewhite et al. 2003), we expect to observe >150 cubs in dens over the course of the study. This number will yield power~ 0.9 to detect a significant difference of~30% in the fecundity rates of urban and wildland bears; Beckmann and Berger (2003) reported >60% difference in fecundity rates of bears in these different habitats. Using GPS location data, we will model the demographic rates of individual bears as a continuous function of how they use urban and wildland habitat, explicitly linking habitat-use to population performance (Mcloughlin et al. 2007, Gaillard et al. 2010). To estimate annual adult female survival we will use Cox proportional hazard models (Therneau and Grambsch 2000, Murray 2006), which allow for staggered entry, continuous-time data collection, and the evaluation of different covariates. We will use multinomial and binomial logistic regression to model fecundity and cub survival rates, respectively (Obbard and Howe 2008), which will rely on annual counts of juvenile bears in winter dens. In these models, we will insert random effects to account for fecundity rates of individual females measured over multiple years, and for the survival of cubs born in the same litter. With all vital rate models we will use GPS data to specifically test whether time in urban habitats, annual availability of natural foods, or density of urban development influences bear population parameters (McLoughlin et al. 2007, Gaillard et al. 20 I0). Annual variation in natural foods will be tracked with satellite imagery (Pettorelli et al. 2005) and information on urban development will be obtained from existing digital data layers. We will also test for the covariate effects of year (to account for variation in natural foods), age (for adult survival and fecundity models), season (for adult survival models) and reproductive status (for adult survival models). We will build a set of apriori candidate models for each vital rate from our covariate set, and identify the best models using model selection (Burnham and Anderson 2002). Additionally, for adult female bears, we will evaluate cause-specific mortality. We will use competing risks analyses (Heisey and Patterson 2006) to examine the differential sources of mortality and their relative importance in urban and wildland habitats. 2C) Quantifying the relative influence ofresource-use, conflict bear management practices (lethal removals and translocations) and harvest on bear demography. Vital rate means and variances measured from Objective 2B will then be inserted into stagestructured matrix projection models (Caswell 2001, Morris and Doak 2002) to assess differences in the population growth rate among those bears that use urban food resources ("urban") and those that do not ("wildland"; Hostetler et al. 2009). The wildland model will serve as a baseline, representing bear demography in the absence of urban food or conflict management, and in the presence of natural food variability. Of those bears that use urban environments, we will then simulate a suite of scenarios to tease apart the inherent effects of anthropogenic food, management-related conflict mortality (i.e. lethal removals and translocations), other urban-related mortality (i.e. road kill, electrocution, etc), and harvest on vital rates, and ultimately, population growth (Fig. 6). First, we will project a matrix with vital rates from bears that used town to estimate the actual ( or realized) growth rate. This model will allow us to compare harvest rates among bears that use urban versus wildland habitats. Second, we will quantify the inherent benefit of anthropogenic food for bears in the absence ofall harvest and urban conflict mortality. To do this we will re-calculate adult female survival censoring all harvest and conflict-related deaths/removals (management and non-management related). We will use the updated values, along with cub survival rates from wildland bears (conservatively assuming that in the absence of human-related mortality "town" cubs would have survival~ than those in the wild) to re-project population growth rates (Hostetler et al. 2009). This will allow us to assess the inherent, but hypothetical, benefit of human food 164 �.... - -- .._J - -- - on local bear demography without urba n- related motiality. Third, we will isolate the impacts of conflict management removals ( lethal removals and translocations) on bear populations. For this scena rio, we will re-calculate adult and c ub survival by censo1ing all management- related removals (but maintaining harvest and non-management mortal ity), and insert these new values into a projection matri x. T his will a llow us to estimate the cha nge in population growth associated with current conflict management practices and estimate their c umulati ve impacts on local populations. Additionall y, for all scenario matrices, we will identi fy those vital rates with the highest elasticity and those dri ving overall growth rates (Wisdom et al. 2000, Caswell 2001). T his wi ll allow us to better understand how patterns of population g rowth respond to vital rate-specific changes in natural and human food availabil ity, confli ct management, and harvest. (' -Individual Attributes - -~ eprloducdv e ~ __ Natural Food Human Food Availabil~...--_) ~ __..,/ I ~. - Resource-Use ••., __c--::o-nf-lic_t__, Removals Stage-Specific Vital Rates In addition to tracking the dri vers of indi vidual Population bear vital rates, we will also assess changes in population Growth Rate density. Density w ill be estimated from hair-snare grids using mark-recapture techniq ues (Woods et al. 1999, Mowat Figure 6. Conceptual model depicting /) the and Strobeck 2000). Bear DNA will be extracted and d(fj'erentfactors that c![fect bear resourcegenotyped from hair to effectively " mark" ind iv id ual bears use, 2) that resource-use influences bear and the pattern of"recaptured" anirua ls will be used to susceptibility to harvest and conflict estimate population s ize. We will set up one hair-snare grid removals, and. 3) how the combined impacts aro und the town of Durango and another grid in adjacent ofresource-use, harvest, and cor1/lict wildland habitat, moni toring each grid for 4 yea rs (Fig. 7). removals determine stage-specific vital Each grid wi ll be composed of 36 cells that are 4km x 4km rates, and ultimately. population growt/1. in size. We will collect bear hair fro m two different sampling sources within each cell , a baited scent trap and a natural rub tree. Baited scent stations will be sunounded by barbed w ire to collect hair fro m bears as they climb around the wire to investigate the bait. We will use multiple bait scents, randomly assigned to different traps each sampling occasion to maintain a hig h hair recapture rate. Additionally, we will attempt to identify l nah1ral rub tree/cell. Rub trees wi ll not be baited, but affixed with a piece of barbed wi re to faci litate hair collection. B y collecting hair fro m both these sources (baited traps and rub trees) we should increase recapture rates and reduce indi vidual heterogeneity in capture response (Boulanger et al. 2008). We will conduct 6 sampling occasions/summer (mid-June through July), checking baited traps and rub trees for hair once/week, and re-baiting scent traps. At the encl of the sampling season hair samples will be sent to the Wildlife Genetics International Laboratory for microsatellite genoty ping. We will use genotype data to estimate density using a spatiallyexplicit Bayesian model for open populations (Gardner e t al. 2009, Gardner et al. 20 I 0). Addi tionally, we use the genotype data to interpolate a spatia l density surface that will allow us to identify habitat covariates associated with high and low bear dens ities in both urban and wild land sampling g rids. We w ill compare densities among sites to detennine whether the availability of human food increases bear density adjacent to town (Beckmann and Berger 2003). Over the course of the study we will also estimate the annual variability in density among urban and wildland habitats. This will e lucidate whether dens ities in each habitat type vary in association with natural food producti on, and the reliability of the hair-snare technique for "snapshot" density measures for statewide mon itoring purposes. Additionally, hair-snare grids will allow us to in fer m ovement of bears from wildland to urban habitats. For example, if high bear densities are maintained along the urban interface despite negative population growth rates (as projected from individual vital rates), it wi ll be suggestive that bears are moving into 165 �Figure 7. Location of urban and wildand DNA hair snare grids. Red circles represent sampling baited traps sites within each cell. Yellow circles represent conflicts reported around Durango, 2007-2010. town from adjace nt wildl ands ( Robinson et al. 2008). Ultimately, using data on both the vital rates from collared animals and density from hair-snares, we w ill be able to discri minate whether town serves as source or sink for local bear populati ons, whether this influence varies under different env ironmental conditions. 3A) Using multiple data sources to build bear population models to inform annual harvest management and elucidate population trajectories. We will use ind ividual vital rate and mark-recapture data fro m O bjective 2, in conjunction w ith annual harvest data, to develop more precise population models for the management of black bears in Colorado. ClUTentl y, it is mandatory fo r hunters to report all harvested bears to Colorado Division of Wildlife and submit a tooth for age estimation (Willey 1974, Stoneberg and Jonke! 1996). Combin ing the three different data types we w ill have available on bears around Durango (sex/age-at-harvest, indi vidual demography from collared bears, and mark-recapture data) we will first estimate baseline population parameters, dramaticall y increasing precis ion in those estimates ( Fiebe rg 20 l 0, Johnson et al. 20 I 0). Then, we will identify the value of each data ty pe (based on sample s ize and years of data collection) for mode ling bear dynamics according to the precision required for making management decisions. Thi s info rmation will be used to generate a parsimonious model that adequate ly describes changes in bear population trends while minimizing unnecessary field data. In do ing this, we hope to provide guidance to the Colorado Division of Wildlife on the allocation of field efforts for effectively monitoring populations, and allow managers to set biologically-based harvest quotas. We will test the accuracy and effecti veness of population mode ls using data collected around Durango, Trinidad, and Aspen (all areas where multipl e bear data types are available), and simulated data, all owing us to furt her validate model structure and precision (Fiebe rg et al. 20 I0). These models w ill be used to inform annual harvest regulations, update population traj ectories, and revise statewide estimates of population s ize. 38) Developing regional habitat models .from GPS collar location data. We will use the wealth o f GPS collar data that we will collect around Durango and which is available from --40 bears around Aspe n (CPW, unpublished data) to build detailed regional habitat models. Currently, bear habitat models for Colorado are deri ved from the perceived value of different vegetation types, as determined by Colorado Division of Wildl ife managers. We hope to enhance regional models through analyses of thousands of bear GPS locations, using additional information on elevation, topography, satellite m easures of annual primary producti vity, and human development variables (i.e. road density, distance to town, etc). We will use a mixed-effects RSF approach lo identify habitat characteristics associated with bear occupancy, appl ying a use-availability design (Manly et al. 2002, Gillies et al. 2 006). We w ill specifically identify second-order habitat selection (Johnson 1980), the 166 - �-..I conditions under which bears establish their home-ranges, using established model selection procedures (Burnham and Anderson 2002). To test the predictive power of the habitat model, we will use crossvalidation (Boyce et al. 2002) and then map expected relative probabilities of selection across the landscape. Additionally, we will use harvest locations and bear sightings from other geographic regions to test the validity of our models for application in other parts of the state. These data-driven habitat models can then be used to provide better estimates of statewide bear density, design more efficient monitoring strategies (Allen et al. 2008), and to identify critical seasonal resources and movement corridors for bears. -....I ._, 'cl N. Location Data used to meet different objectives of this study will be obtained from various parts of Colorado. The anthropogenic food removal experiment (Objective IA) and the demography/resource-use portion of the study (Objective 2A-C) will be conducted in the vicinity of Durango, Colorado (La Plata, Hinsdale, and Archuleta counties). Durango was selected as the focal urban environment based on several factors including a history of high bear-human conflicts (Fig. 3), a good record of recent conflict reporting, the feasibility of conducting the food-removal experiment (based on city waste management practices), and minimal city-wide bear-proofing infrastructure. Tracking bear population parameters in this region will require that trapping and hair-snaring will occur on a combination of USFS, BLM, state, city, and private lands. We will test the effectiveness of a spatially-targeted harvest program along the southern Front Range (Objective lC), and opportunistically throughout the state as changes occur in harvest management. The strategic translocation model will be developed on a statewide basis (Objective 1B), along with population and habitat models (Objectives 3A-B). 0. Schedule of Work Activity Trap and collar bears Monitor bear survival Conduct DNA hair-snare grids Genotype hair samples Distribute bear-resistant containers Monitor human-food-removal experiment Translocation modeling and evaluation Implement spatially-targeted harvest program Evaluate spatially-targeted harvest program Conduct winter den checks (reproduction) Estimate population parameters (individual vital rates, and population density) Develop and test population and habitat models 167 Timeline Summer 2011-2015 Summer 2011-2016 Summer 2011-2014 Fall 2011-2014 Spring 2012 Spring-Fall 2012-2015 Summer-Fall 2012-2015 Fall 2012 2012-2015 Winter 2012-2016 Winter 2012-2016 Winter 2013-2017 �P. Estimated Costs NEED INDIVIDUAL DEMOGRAPHY (5 Yrs) 50 OPS Collars ( 10 Purchased) GPS Battery Replacements (2/ea) Telemetry Receivers/Ant (3) Traps (20) Snares (10) Jab Stick ( 1) Misc Equipment Snowmobiles Field Technicians Spring Trapping Yr I (3.5mo) Spring Trapping Yrs 2-5 (3 .5 mo) Winter Dens Yrs 1-5 (3 mo) COST/UNIT FY201 l-12 $4,800 $300 $695 $1,000 $100 $800 $192,000 3 & Maintenance Tech! (3)/TechII (I) Tech I (I )/Techll (I) Techl(3)/TechII(l) 2 DNA HAIR-SNARE GRIDS (4 Yrs) Field Equipment Field Technicians (2.5 mo) Tech! (2) Genetic Analysis $20,000/Grid GARBAGE EXPERIMENT (4 Yrs) Bear-resistant containers Residental/Commerical Field Technicians (5 mo) Tech! (1) TRANSLOCATIONPLAN (4 Yrs) Store-on-Board OPS Collars (50) $1,500 Web Prot:[ammer {1 mo} Programmer {1} PROJECT TOTAL FY2012-13 FY2013-14 FY2014-15 FY2015-16 TOTAL $5,000 $5,000 $5,000 $5,000 $5,000 $5,000 $5,000 $5,000 $192,000 $30,000 $2,085 $20,000 $1,000 $800 $30,000 $40,000 $31,984 $19,301 $31,984 $19,301 $31,984 $19,301 $31,984 $19,301 $31,984 $37,209 $77,204 $159,920 $1,200 $12,792 $40,000 $250 $12,792 $40,000 $250 $12,792 $40,000 $250 $12,792 $40,000 $250,000 $12,792 $12,792 $12,792 $30,000 $2,085 $20,000 $1,000 $800 $10,000 $20,000 $37,209 $1,950 $51,168 $160,000 $12,792 $75,000 $75,000 $3,200 $3,200 $369,070 $452,119 $157,119 $130,319 $250,000 $51,168 $74,077 $1,182,704 168 ( ( { ( ( { { ( { { ( { ( { ( ( C{ C( CC( ( C{ ( ( CC( ( C( ( CC( C{ C( { �-..I Q. Related Federal Projects There are no related federal projects. R. Literature Cited Allen, J.R., L.E. Mclnenly, E.H. Merrill, and M.S. Boyce. 2008. Using resource selection functions to improve estimation of elk population numbers. Journal of Wildlife Management 72: 1798-1804. Baruch-Mordo, S. 2007. Black-bear human conflicts in Colorado: spatiotemporal patterns and predictors. Thesis, Colorado State University, Fort Collins, Colorado. Baruch-Mordo, S., S. W. Breck, K.R. Wilson, and D.M. Theobald. 2008. Spatiotemporal distribution of black bear-human conflicts in Colorado, USA. Journal of Wildlife Management 72: 1853-1862. Baruch-Mordo, S., S.W. Breck, K.R. Wilson, and J. Broderick. 2011. The carrot or the stick? Evaluation of education and enforcement as management tools for human-wildlife conflicts. Plos One 6: el 5681. Baruch-Mordo, S., K.R. Wilson, D. Lewis, J. Broderick, J. Mao, and S.W. Breck. 2010. Roaring Fork Valley urban black bear ecology study: progress report to the Colorado Division of Wildlife. Contact Sharon Baruch-Mordo for copy. Email: sharonb m@yahoo.com Beck, T.D.I. 1991. Black bears of west-central Colorado. Technical Publication No. 39, Colorado Division of Wildlife, Colorado. Beckmann, J.P., and J. Berger. 2003. Using black bears to test ideal-free distribution models experimentally. Journal ofMammalogy 84:594-606. Beckmann, J.P., L. Karasin, C. Costello, S. Matthews, and Z. Smith. 2008. Coexisting with black bears: perspectives from four case studies across North America, WCS Working Paper No. 33, Wildlife Conservation Society, New York, New York. Beckmann, J.P., C.W. Lackey, and J. Berger. 2004. Evaluation of deterrent techniques and dogs to alter behavior of "nuisance" black bears. Wildlife Society Bulletin 32: 1141-1146. Blomquist, S.M., and M.L. Hunter. 2010. A multi-scale assessment of amphibian habitat selection: wood frog response to timber harvesting. Ecoscience 17 :251-264. Boulanger, J., K.C. Kendall, J.B. Stetz, D.A. Roon, L.P. Waits, and D. Paetkau. 2008. Mulitple data sources improve DNA-based mark-recapture population estimates of grizzly bears. Ecological Applications 18:577-589. Borger, L., N. Franconi, G. De Michele, A. Gantz, F. Meschi, and A. Manica. 2006. Effects of sampling regime on the mean and variance of home range estimates. Journal of Animal Ecology 75:13931405. Boyce, M.S., J. Pitt, J.M. Northrup, A.T. Morehouse, K.H. Knopff, B. Dristesu, and G.B. Stenhouse. 2010. Temporal autocorrelation functions for movement rates from global positioning system radiotelemetry data. Philosophical Transactions of the Royal Society Series B 365 :2213-2219. Boyce, M.S., P.R. Vernier, S.E. Nielsen, and F.K.A. Schmiegelow. 2002. Evaluating resource selection functions. Ecological Modelling 157 :281-300. Burnham, K.P., and D.R. Anderson. 2002. Model selection and inference: a practical informationtheoretic approach. Springer-Verlag, New York, New York. Caswell, H. 2001. Matrix population models: construction, analysis, and interpretation. Second Edition, Sinauer Associates, Sunderland, Massachussetts. Cook, R.J., and J.F. Lawless. 2007. The statistical analysis ofrecurrent events. Springer Science and Business Media, New York, New York. Fieberg, J.R., K.W. Shertzer, P.B. Conn, K.V. Noyce, and D.L. Garshelis. 2010. Integrated population modeling of black bears in Minnesota: implications of monitoring and management. Plos One 5:el2114. Gaillard, J.-M., M. Hebblewhite, A. Loison, M. Fuller, R. Powell, M. Basille, and B. Van Mooter. 2010. Habitat-performance relationships: finding the right metric at a given spatial scale. Philosophical Transactions of the Royal Society Series B 365:2255-2265. 169 �Gardner, B., J. Reppucci, M. Lucherini, and J.A. Royle. 2010. Spatially explicit inference for open populations: estimating demographic parameters from camera trap studies. Ecology 91 :33763383. Gardner, B., J.A. Royle, M.T. Wegan, R.E. Rainbolt, and P.O. Curtis. 2009. Estimating black bear density using DNA data from hair snares. Journal of Wildlife Management 74:318-325. Garshelis, D.L., and H. Hristienko. 2006. State and provincia·l estimates of American black bear numbers versus agency assessments of population trend. Ursus 17:1-7. Gillies, C.S., M. Hebblewhite, S.E. Nielsen, M.A. Krawchuk, C.L. Aldridge, J.L. Frair, D.J. Saher, C.E. Stevens, and C.L. Jerde. 2006. Application of random effects to the study ofresource selection by animals. Journal of Animal Ecology 75:887-898. Gore, M.L., W.F. Siemer, J.E. Shanahan, D. Schuefele, and DJ. Decker. 2005. Effects on risk perception of media coverage of a black bear-related human fatality. Wildlife Society Bulletin 33:507-516. Gore, M.L., B.A. Knuth, C. W. Scherer, and P.O. Curtis. 2008. Evaluating a conservation investment designed to reduce human-wildlife conflicts. Conservation Letters 1:136-145. Greenleaf, S.S., S.M. Matthews, R.G. Wright, J.J. Beecham, and H.M. Leithead. 2009. Food habitat of American black bears as a metric for direct management of human-bear conflict in Yosemite Valley, Yosemite National Park, California. Ursus 20:94-101. Hebblewhite, M., M. Percy, and R. Serrouya. 2003. Black bear (Ursus americanus) survival and demography in the Bow Valley of BanffNational Park, Alberta. Biological Conservation 112:415-425. Heisey, D.M., and B.R. Patterson. 2006. A review of methods to estimate cause-specific mortality in the presence of competing risks. Journal of Wildlife Management 70: 1544-1555. Hosmer, D.W., S. Lemeshow, and S. May. 2008. Applied survival analysis: regression modeling of time to event data. Jon Wiley and Sons, Hoboken, New Jersey. Hostetler, J.A., J.W. McCown, E.P. Garrison, A.M Neils, M.A. Barrett, M.E. Sunquist, S.L. Simek, and M.K. Oli. 2009. Demographic consequences of anthropogenic influences: Florida black bears in north-central Florida. Biological Conservation l 42:2456-2463. Hristienko, H., and J.E. McDonald Jr. 2007. Going into the 21 st century: a perspective on trends and controversies in the management of the American black bear. Ursus 18:72-88. Johnson, D.H. 1980. The comparison of usage and availability measurements for evaluating resource preference. Ecology 61 :65-71. Johnson, H.E, L.S. Mills, J.D. Wehausen, and T.R. Stephenson. 2010. Combining ground count, telemetry, and mark-resight data to infer population dynamics in an endangered species. Journal of Applied Ecology 47: I 083-1093. Jonkel, J J. 1993. A manual for handling bears for managers and researchers. Office of Grizzly Bear Recovery, United States Fish and Wildlife Service, Missoula, Montana. Jonsen, l.D., J.M. Flemming, and R.A. Myers. 2005. Robust state-space modeling of animal movement data. Ecology 86:2874-2880. Kie, J., J. Matthiopoulos, J. Fieberg, R.A. Powell, F. Cagnacci, M.S. Mitchell, J.-M. Gaillard, and P.R. Moorcroft. 2010. The home-range concept: are traditional estimators still relevant with modern teleme~ry technology? Philosophical Transactions of the Royal Society Series B 365 :2221-2231. Koehler, G.M., and DJ. Pierce. 2005. Survival, cause-specific mortality, sex, and ages of American black bears in Washington State, USA. Ursus 16:157-166. Kretser, H., P.J. Sullivan, and B.A. Knuth. 2008. Housing density as an indicator of spatial patterns of reported human-wildlife interactions in Northern New York. Landscape and Urban Planning 84:282-292. Landriault, L.J., G.S. Brown, J. Hamr, and F.F. Mallory. 2009. Age, sex and relocation distance as predictors of return for relocated nuisance black bears Ursus americanus in Ontario, Canada. Wildlife Biology 15:155-164. Linnell, J.D.C., R. Aanes, and J.E. Swenson. 1997. Translocation of carnivores as a method for managing problem animals: a review. Biodiversity and Conservation 6:1245-1257. 170 �..,I '¥1 'cl 'cl Lyons, A.J. 2005. Activity patterns of urban American black bears in the San Gabriel Mountains of southern California. Ursus I 6: 255-262. Mackun, P., and S. Wilson. 2011. Population distribution and change: 2000 to 2010. U.S. Census Bureau, 20 IO Census Briefs: http://www.census.gov/prod/cen20 I 0/briefs/c20 I 0br-0 l .pdf Manly, B.F.J., L.L. McDonald, D.L. Thomas, T.L. McDonald, and W.P. Erickson. 2002. Resource selection by animals; statistical design and analysis for field studies. Second Edition. Kluwer Academic Publishers, Boston, Massachussetts. Mazur, R.L. 2010. Does aversive conditioning reduce human-black bear conflict? Journal of Wildlife Management 74:48-54. McCarthy, T., and R. Seavoy. 1994. Reducing non-sport losses attributable to food conditioning: human and bear behavior modification in an urban environment. International Conference on Bear Research and Management 9:7 5-84. Mcloughlin, P.O., J.-M. Gaillard, M.S. Boyce, C. Bonenfant, F. Messier, P. Duncan, D. Delonne, B. Van Moorter, S. Said, and F. Klein. 2007. Lifetime reproductive success and composition ofthe home range in a large herbivore. Ecology 88:3192-3201. McLoughlin, P.O., D.W. Morris, D. Fortin, E. Vander Wal, and A.L. Contasti. 2010. Considering ecological dynamics in resource selection functions. Journal of Animal Ecology 79:4-12. Morales, J.M., P.R. Moorcroft, J. Matthiopoulos, J.L. Frair, J.G. Kie, R.A. Powell, E.H. Merrill, and D.T. Haydon. 20 I 0. Building the bridge between animal movement and population dynamics. Philosophical Transactions of the Royal Society Series B 365:2289-2301. Morris, W.F., and D.F. Doak. 2002. Quantitative Conservation Biology: Theory and Practice of Population Viability Analysis. Sinauer, Sunderland, Massachussetts. Mowat, G., and C. Strobeck. 2000. Estimating population size of grizzly bears using hair capture, DNA profiling, and mark-recapture analysis. Journal of Wildlife Management 64: 183-193. Murray, D.L. 2006. On improving telemetry-based survival estimation. Journal of Wildlife Management 70: 1530-1543. Obbard, M.E., and E.J. Howe. 2008. Demography of black bears in hunted and unhunted areas of the boreal forest of Ontario. Journal of Wildlife Management 72:869-880. Obbard, M.E., B.A. Pond, and E.J. Howe. 1997. Analysis of relationships among black bear nuisance activity, food availability, and harvest in Ontario. Nuisance Bear Review Committee, Appendix I 0. Ministry of Natural Resources, Ontario, Canada. Peine, J.D. 2001. Nuisance bears in communities: Strategies to reduce conflicts. Human Dimensions of Wildlife 6:223-237. Pettorelli, N., J.O. Vik, A. Mysterud, J.-M. Gaillard, C.J. Tucker, and N.C. Stenseth. 2005. Using the satellite-derived NOVI to assess ecological responses to environmental change. Trends in Ecology and Evolution 20:503-510. Robinson, H.S., R.B. Wielgus, H.S. Cooley, and S. W. Cooley. 2008. Sink populations in carnivore management: cougar demography and immigration in a hunted population. Ecological Applications 18: 1028-103 7. Rogers, L. L. 1986. Effects of trans location distance on frequency of return by adult black bears. Wildlife Society Bulletin 14:76-80. Skalski, J.R., R.L. Townsend, and G.A. Gilbert. 2007. Calibrating statistical population reconstruction methods using catch-effort and index data. Journal of Wildlife Management 71: 1309-1316. Spencer, R.D., R.A. Beausoleil, and D.A. Martorello. 2007. How agencies respond to human-bear conflicts: a survey of wildlife species in North America. Ursus 18: 217-229. Stoneberg, R.P., and C.J. Jonkel. 1966. Age determination of black bears by cementum layers. Journal of Wildlife Management 30:411-4 I 4. Tavss, E.A. 2005. Correlation of reduction in nuisance black bear complaints with implementation of (a) a non-violent program and (b) a hunt. Final Report, New Jersey Fish and Game Council, New Jersey Public Hearing on the Comprehensive Black Bear Management Policy, New Jersey. Themeau, T.M., and P.M. Grambsch. 2000. Modeling survival data: extending the Cox model. Springer, 171 �New York, New York. Treves, A., K.J. Knapp, and D.M. Macfarland. 2010. American black bear nuisance complaints and hunter take. Ursus 21 :30-42. Will, G.B. 1980. Black bear-human conflicts and management considerations to minimize and correct these problems. Proceedings Eastern Workshop on Black Bear Management and Research 5:7587. Willey, C.H. 1974. Aging black bears from first premolar tooth sections. Journal of Wildlife Management 38:97-100. Wisdom, M.J., L.S. Mills, and D.F. Doak. 2000. Life stage simulation analysis: estimating vital-rate effects on population growth for conservation. Ecology 81 :628-641. Woods, J.G., D. Paetkau, D. Lewis, B.N. Mclellan, M. Proctor, and C. Strobeck. 1999. Genetic tagging of free-ranging black and brown bears. Wildlife Society Bulletin 27:616-627. Zac~ C.S., B.T. Milne, and W.C. Dunn. 2003. Southern oscillation index as an indicator of encounters between humans and black bears in New Mexico. Wildlife Society Bulletin 31 :517-520. 172 �\&I APPENDIX II CAPTURE AND HANDLING PROCEDURES FOR FREE-RANGING BLACK BEARS Black bears will be initially captured and collared during the summer months and annually recaptured in their dens during winter months to obtain reproductive information. Summer We will capture and collar adult female black bears during summer months (May-Sept) using cage traps and foot snares. We will use cage traps in areas close to Durango or with high human activity, and where there is good road access. Snares will be used for more remote trapping locations, away from human activity and where vehicle access is limited. Once a bear has been captured using either method, field crews will use an identical protocol to process animals. ..., wl Cage Traps We will capture bears with two different trap designs, a cage trap designed and used extensively by Beck ( 1993), and a newly designed trap to specifically target female bears. The trap developed by Beck is 1.8 m long and l .O m in height and width. The frame is constructed of angle iron, all side and top panels are wire mesh of 1.9 x 1.9 in size, and the trap has a floor that is 16-gauge steel. A spring-powered, solid aluminum door is mounted on a full-length hinge at one end and a latching mechanism holds the door closed. The door is triggered via a treadle pedal on the floor, and a standard garage door coil spring provides closing power. A hinged panel along the back of the trap allows access for administering immobilizing drugs via jabpole. In total, the trap weighs approximately 236 kg. In the first study in which these traps were used, only 1 bear in 134 captures was injured, as the individual broke a canine on the wire mesh. Because we are specifically interested in capturing and collaring female black bears, we worked with Mat Alldredge, Tom Davies, Lye) Willmarth and others to design a smaller, lighter trap that would discourage the capture of large males and increase portability in the field. These traps were built to be slightly larger than those that have been successfully used for cougars (Alldredge et al. personal communication) and are 34in high, 60in long, and 25in wide. The frame is built with lxl in heavy gauge steel, covered with 1x 1in heavy gauge, high tinsel, steel mesh. The smaller dimensions of the mesh will reduce the possibility that animals will break their teeth on the cage. The sides of the trap have additional braces to increase overall strength and support. The door of the trap comprises one end of the structure and is designed drop and latch to the bottom of the frame. Bait is hung from a cable attached to an archery trigger, and the door falls shut when the trigger is released. Due to the smaller size of the trap, it only weighs approximately 60 kg. Cage traps will be positioned so they are in the shade, and exposure to sun and precipitation is minimized. All cage traps will be clearly marked with warning signs. Cages will be baited with rotting fish, fruit, or road kill. They will be set in the late afternoon or evening and checked by field crews the following morning to minimize the time an animal spends in a trap. If the bear can be clearly identified as a male in the trap, or the bear is a cub or yearling (too small for a GPS collar), it will be released without being immobilized. If the bear is an adult female, or there is uncertainty in the sex of the adult bear, it will be immobilized following procedures described below. Bears will be immobilized with ajabpole, syringe pole, or syringe (hand injection), with the injection targeted into muscle tissue along the shoulder or thigh. Aldrich Foot Snares Aldrich foot snares were specifically developed to capture bears and have proven to be safe and effective (Jonke) 1993 ). The spring activated snare secures a ¼ inch steel cable around the foot of the bear, closing tight with the action of a small piece of angle iron fashioned into a sliding lock mechanism. 173 �The inside of the snare loop is wrapped with duct tape to minimize surface abrasion on the skin of the foot. We will modify snares with additional duct tape and/or surgical tubing over the cable to serve as a "cub stopper" such that small bears (cubs and yearlings) have a low probability of being captured (Jonkel 1993). An in-line swivel is placed in the cable to avoid torsion of the foot and a potential bone fracture. A short lead is attached to the snare to further minimize stress to the leg. The lead is then secured to an anchor tree at least IO inches in diameter with a ¼ in steel cable clamped and stapled to the base of the tree so the bear cannot climb it. Branches of the tree are lopped off with a saw or axe about 8 ft up, so the bear cannot hang itself from a branch by the snare cable. An area of 2::5 meters is cleared around the snare site to eliminate potential that the bear is able to twist the snare loop around any obstacles (saplings, brush, etc). Large branches will be angled over the snare to force ungulates to step over or go around it, minimizing the possibility of catching non-target animals. Additional details of setting snares can be found in Jonkel ( 1993). A disadvantage of using foot snares is that all bears that are caught (even if they are a male bear or too small to collar) must be immobilized to be released. Other non-target animals that are caught (i.e. mountain lions, coyotes, etc) will be immobilized with Telazol and released. Snares will be set in the evening and checked in the morning, operated when ambient temperatures are between 32 and 90°F. Snared bears will be immobilized using a jabpole or CO2 dart gun with the injection targeted into muscle tissue along the shoulder or thigh. "W' Animal Processing During summer months bears will be anesthetized with butorphanol, azaperone, and medetomidine (BAM), a drug combination that has been successful immobilizing black bears and is reversible with atipamezole (a medetomidine antagonist), allowing a faster and safer release of animals around urban environments (Wolfe et al. 2008). BAM will be administered at a volume of 0.4ml/23kg (50 lbs) with a dosage of 0.26mg/kg for butorphaneol, 0.22mg/kg for azaperone and 0.09mg/kg for medetomidine. We will initially give the recommended dose based on estimated animal weight and boost as necessary by½ and¼ of the original dose for the first and second boosters, respectively. To reverse immobilization we will intravenously administer atipamezole. We will dispense a volume of 1ml/1ml at a dosage of 5mg/I mg of medetomidine or 0.45mg/kg. One dose should be sufficient to reverse BAM. Bears immobilized with BAM should not be consumed for 45 days afterward, information which will be printed on collars and ear-tags (see below). Following the injection of BAM, field personnel will approach and gently prod the bear to ensure that the animal is fully anesthetized, administering additional doses as needed. Once anesthetized, the bear will be removed from the trap or snare and placed in a stemally recumbent position with front and rear legs extended. If the bear will not be collared (either because it is a male or too young) it will be subcutaneously injected with a passive integrated transponder (PIT) tag and marked with a single black or brown ear-tag that is labeled with the appropriate consumption date information. Afterwards, the bear will be administered atipamezole and released. Adult female bears will be discriminated from subadults based on weight, and nipple size and coloration (Beck 1991 ). Adult female bears will be fully processed. They will immediately be treated with eye ointment and blindfolded to reduce visual stimuli and protect the eyes from debris and bright light. Throughout the time a bear is anesthetized, its vital signs (heart rate, respiration and temperature) will be monitored. Normal ranges for vital rates of adult bears: heart rate = 60-90 beats/minute, respiration = 15-20 breaths/minute, and temperature= 99.6 - 101.0°F (Jonkel 1993). If a bear's body temperature exceeds the normal range, field staff will cool the underside of the bear with water, particularly the arm pits, groin and stomach. If heart rate and respiration values fall outside normal expectations we will reverse the anesthesia and release the bear. 174 ~ �...,; cl 'Cl In processing female bears, we will check each animal for any lacerations that occurred in the capture process and treat them with topical antibiotics. Additionally, bears will be given an injection of Oxytetracycline (9mg/1b) or Baytril (7 .5 mg/kg) to reduce chances of infection from darting and tooth extraction (described below). Adult female bears will be subcutaneously injected with a PIT tag. If the individual has been identified by CPW Area staff as a hconflict" bear it will be marked in accordance with CPW Administrative Directive W-2. Individuals will be weighed using a portable spring scale and pulley system and their breeding status will be recorded (lactating, cubs present, evidence of suckling, etc). We will take multiple body size measurements including total length, chest girth and neck girth. During winter months we will also use bioelectrical impedance analysis to measure bear body fat (Farley and Robbins 1994, Hilderbrand et al. 1998). Additionally we will draw blood and collect a hair sample. These samples will be used for gepetic, stable isotope, and telomere analysis. To age captured bears using tooth cementum annuli counts (Stoneberg and Jonkel 1966, Willey 1974), we will remove the first vestigial premolar ( or if unavailable the lower first premolar) using a dental elevator. For tooth extraction, we will topically apply Lidocaine and subcutaneously administer Ketofen for analgesia ( 1cc/ I 00lb ). A piece of foam gel will then be placed on the removal site and left for adhesion and filling of the wound. We will attach a GPS collar (~700 g) with a ~2 year life expectancy. Collars will be programmed to collect ?:3 locations/day, and will be labeled with the appropriate conswnption date based on immobilization. The GPS collar will include a VHF transmitter that allows tracking via standard telemetry equipment and the retrieval of collars. We will recapture each collared female each winter to assess fecundity and cub survival. If we are unable to recapture a bear, however, each collar will have a degradable canvas spacer that should break-down within 1-2 years and allow the collar to fall off. GPS collars will upload the location of each individual every day via a satellite system and the location will be available to researchers in real-time. When animal processing procedures are completed, the blindfold will be removed and the immobilization reversal will be administered. Field staff will observe the bear from a safe distance to ensure that the animal recovers to a standing position (Wolfe et al. 2008). ._ Winter Den Checks To assess fecundity and cub survival we will recapture collared female bears each winter. Bears will be tracked to their dens using GPS collar locations, and researchers will dig through the snow as needed to access the den~ Adult female bears and accompanying yearlings will be anesthetized with Telazol using ajabpole or CO2 dart gun. Telazol will be administered intramuscularly with a dose of 1.5 2.5mg/1b at a lower concentration (5cc at 100mg/ml). Bears will be immobilized at a higher concentration (3cc at 166 mg/ml) if they are particularly agitated or large. We will initially give the recommended dose based on estimated animal weight and boost as necessary by ½ and ¼ of the original dose for the first and second boosters, respectively. Unlike BAM, there is no reversal drug for Telazol. That said, an immobilized bear can be returned to its den for recovery, reducing animal stress and increasing researcher safety. 'el Once immobilized, bears will be removed from the den, placed on blanket, and processed in a similar manner to that described above. Field staff will check the fit of the GPS collar and make any necessary modifications, and clean up any neck wounds with saline solution. Newborn cubs in the den will be tucked inside the jacket of a field crew member, next to their body, so that the cub says warm and quiet. After processing, bears will be returned to the den; adults and yearlings will be positioned on their side and newborn cubs will be placed on their mother's back. The den entrance will be covered with sticks and boughs and a layer of snow to discourage the bear from leaving the den. We will retain a small opening in the snow to ensure that the bear has a fresh supply of air (Jonkel 1993). 175 �Injuries and Euthanasia If an animal is seriously injured (e.g. fractured or broken appendage, vertebrae, pelvis, or jaw, severe dislocation, laceration or any other injury that severely compromises its ability to survive and/or causes severe pain or distress) during capture, it will be quickly and humanely euthanized. Bears will be deeply anesthetized with BAM or Telazol and euthanized via a intravenous potassium chloride (KCI; 400-800 mEq) injection or gunshot to the head or neck. Carcasses that are euthanized will be disposed of in a landfill or left in an area appropriate for scavengers. LITERATURE CITED Beck, T.D.l. 1991. Black bears of west-central Colorado. Technical Publication 39, Colorado Division of Wildlife, Fort Collins, Colorado. Beck, T.D.l. 1993. Development of black bear inventory techniques;job progress report. Colorado Division of Wildlife, Project Number W-153-R-6. Farley, S.D., and C.T. Robbins. 1994. Development of two methods to estimate body composition of bears. Canadian Journal of Zoology 72:220-226. Hilderbrand, G.V., S.D. Farley, and C.T. Robbins. 1998. Predicting body condition of bears via two field methods. Journal of Wildlife Management 62:406-409. Jonkel, J.J. 1993. A manual for handling bears for managers and researchers. Office of Grizzly Bear Recovery, US Fish and Wildlife Service, University of Montana, Missoula, Montana. Stoneberg, R.P ., and C.J. Jonke I. 1966. Age determination of black bears by cementum layers. Journal of Wildlife Management 30:411-414. Willey, C.H. 1974. Aging black bears from first premolar tooth sections. Journal of Wildlife Management 38:97-100. Wolfe, L.L., C.T. Goshorn, and S. Baruch-Mordo. 2008. Immobilization ofblack bears (Ursus americanus) with a combination ofbutorphanol, azaperone, and medetomidine. Journal of Wildlife Diseases 44:748-752. 176 ..,, �-..,, Colorado Division of Parks and Wildlife July 2011 - June 201 2 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Colorado 3430 3003 Division of Wildlife Mammals Research Predatory Mammal Conservation Black bear exploitation of urban envirnrunents: finding management solutions and assessing regional population effects Federal Aid Project No. Period Covered: July 1, 2011 - June 30, 201 2 .,.,, Author: H.E. Johnson; proj ect cooperators, C. Bishop, J. Broderick, J. Apker, S. Lischke, M. Alldredge, S. Breck, J. Beckmann, K. Wilson, M. Reynolds-Hogland, T. Spezze, and P. Dorsey. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT ..,,JI .,,,,,, ...,,,, ..,.; Across the country conflicts among people and black bears are increasing in frequency and severity, and have become a high priority wildlife management issue. Whether increases in conflicts reflect recent changes in bear population trends or just bear behavioral shifts to anthropogenic food resources, is largely unknown, with key implications for bear management. This issue has generated a pressing need for bear research in Colorado and has resulted in a unique collaboration that builds on the resources and abilities of personnel from 5 entities: the Colorado Parks and Wildlife (CPW), the USDA National Wildlife Research Center, Colorado State University, Wildlife Conservation Society, and Bear Trust Internationa l. Collectively, we completed year 1 of a 5-year study on black bears that I) tests management strategies for reducing bear-human conflicts, 2) determines the influence of urban environments on bear habitat-use patterns and demography, 3) identifies public attitudes and perceptions about bears, bear management and bear-human encounters, and 4) develops population and habitat models to support the sustainable monitoring and management of bears in Colorado. This project was initiated in FY 10-11; during this past fiscal year we have primarily focused on coordinating research logistics and collecting field data in the vicinity of Durango, Colorado. Specifica lly, we obtained data on garbage-related bear-human conflicts, trapped and marked black bears, monitored the vital rates of collared bears (surviva l, fecundity and cub survival) through telemetry and winter den visits, collected data on the availabi lity of late summer/fall mast, tracked human-related bear mortalities and removals, performed non-invasive genetic mark-recapture surveys, and conducted a survey of public attitudes and perceptions about bear-human encounters. Project collaborators will continue to seek additional fundi ng to implement the remaining activities outlined in the research proposal (i.e., purchase additional containers for an urban-food-removal experiment, increase the sample size of collared bears, and acquire telemetry collars to test a translocation model). Information from this study will provide solutions for sustainably managing black bears outside urban environments, while reducing bear-human conflicts within urban environments; knowledge that is critica l for wildlife managers in Colorado and across the country. 114 �WILDLIFE RESEARCH REPORT BLACK BEAR EXPLOITATION OF URBAN ENVIRONMENTS: FINDING MANAGEMENT SOLUTIONS AND ASSESSING REGIONAL POPULATION EFFECTS HEATHER E. JOHNSON P.N. OBJECTIVES To conduct a study on black bears in Colorado that l ) tests management strategies for reducing bearhuman conflicts, 2) determines the influence of urban environments on bear habitat-use patterns and demography, 3) identifies public attitudes and perceptions about bears, bear management and bear-human encounters, and 4) develops population and habitat models to support the sustainable monitoring and management of bears. SEGMENT OBJECTIVES 1. Work with personnel from CPW Area 15, CPW Southwest Region, the City of Durango, La Plata County, US Forest Service (Columbine and Pagosa Ranger Districts), Bureau of Land Management (BLM; Tres Rio Field Office), and private landowners on field research logistics. 2. Collect pre-treatment data on the frequency of bears accessing human garbage in preparation for an urban bear-proofing experiment. 3. Trap and collar adult female black bears in the vicinity of Durango to collect data on bear habitatuse patterns and demography. 4. Monitor bear survival via global position system (GPS) collar locations. 5. Obtain data on summer/fall natural food availabi lity for bears based on the phenology and abundance of gambel oak, serviceberry, chokecherry, hawthorne, pinyon pine and squaw apple. 6. Investigate the winter dens of collared female bears to collect data on fecundity and cub survival, inspect collar fit, and replace collar spacers and batteries. 7. Track human-related bear mortalities and removals around Durango from lethal conflict mortalities, vehicle collisions, harvest, and translocations. 8. Perform non-invasive genetic mark-recapture surveys to estimate bear density and population size around Durango (urban site) and in the Piedra watershed (wildland site). 9. Conduct a survey of public attitudes and perceptions about bears, the local bear population, bear management and bear-human encounters. INTRODUCTION In Colorado and across the country, conflicts among people and black bears (Ursus americanus) appear to be increasing in number and severity (Hristienko and McDonald 2007, Baruch-Mordo et al. 2008, CPW unpublished data). Bear-human confl icts can result in public safety concerns, property damage, bear mortaUty (i.e. euthanasia), and high management costs, and thus, have become a critical wildlife management issue. While wildlife agencies have used a variety of tools to try to minimize bearhuman confl icts (i.e., education, aversive conditioning of bears, and modifications to harvest), conflict rates have continued to rise. Whether increases in bear-human conflicts reflect recent changes in the bear population or just behavioral shifts to anthropogenic food resources, is largely unknown, as bear population parameters have been exceeding difficult to estimate (Garshelis and Hristienko 2006). Without a thorough understanding of the relationship between conflict rates and bear behavior and population dynamics, it has been difficult for wildlife agencies to successfully reduce conflicts through bear management. 11 5 - ...., --...,, -- �While there is uncertainty about how to reduce bear-human conflicts, two key factors thought to exacerbate this problem are expanding human development and climatic variation. Colorado has had one of the highest rates of exurban development in the nation (Theobald and Romme 2007), and this development has resulted in additional human food on the landscape in the form of garbage, agricultural resources, fruit trees, etc. The availability of human food to bears has been identified as the primary cause of bear-human conflicts (Spencer et al. 2007, Beckmann et al. 2008, Greenleaf et al. 2009), as bears are opportunistic foragers that will readily take advantage of this resource. Bear-use of human food not only increases interactions between bears and people but has been found to alter bear activity patterns, foraging behavior, movement rates, and even survival and reproductive rates (Beckmann and Berger 2003a, Beckmann and Berger 2003b, Hostetler et al. 2009), having the potential to significantly alter both bear behavior and demography. This phenomenon is further complicated by variation in annual weather patterns, as bear-use of human development appears to increase when natural foods are in short supply (Zack et al. 2003, Baruch-Morda et al. 2010). Because bears predominately consume vegetation, recent patterns of drought in Colorado have caused natural food failures in some years. As a result, bears may be increasing their reliance on human foods, with associated behavioral and demographic impacts. While the effects of urbanization and climate have critical implications for ·modifying bear-habitat relationships, they also have critical implications for increasing rates of bear-human conflicts. To develop successful strategies to reduce conflicts while maintaining viable bear populations, wildlife agencies must understand how factors such as climate, natural food availability, human food ability, and management influence the behavior and dynamics of bear populations. To address these questions, Colorado Parks and Wildlife has partnered with the USDA National Wildlife Research Center, Wildlife Conservation Society, Colorado State University, and Bear Trust International. Collectively, we initiated a project in FYl0-11 to 1) test management strategies for reducing bear-human conflicts, 2) determine the influence of urban environments on bear habitat-use patterns and demography, 3) identify public attitudes and perceptions about bears, bear management and bear-human encounters, and 4) develop population and habitat models to support the sustainable monitoring and management of bears in Colorado (Johnson et al. 2011). This information should provide solutions for sustainably managing black bears outside urban environments, while reducing bear-human conflicts within urban environments; knowledge that is critical for wildlife managers in Colorado and across the west. \cl During FY 11-12 _we worked with internal and external stakeholders on field research logistics, obtained data on garbage-related bear-human conflicts, trapped and marked black bears, monitored the vital rates of collared bears (survival, fecundity and cub survival) through telemetry and winter den visits, collected data on the availability of late summer/fall mast, tracked human-related bear mortalities and removals, performed non-invasive genetic mark-recapture surveys, and conducted a survey of public attitudes and perceptions about bear-human encounters. Our efforts focused largely on collecting field data to meet research objectives 1-3, information which will eventually be used to address objective 4. We report general summary information from field activities over the past year; detailed analyses of field data will occur in future years. STUDY AREA To meet study objectives, a combination of site-specific field data and statewide data will be required. Site-specific field data is being collected in the vicinity of Durango, and is the focus of this progress report. Regional and statewide analyses will be conducted in future years. The town of Durango contains ~ 17,000 people ( within city limits) and sits at 1,985 m along the Animas river valley. The town is surrounded by mountainous terrain ranging in elevation from~ 1,930 to ~3,600 m, and is generally characterized by mild winters and warm summers that experience monsoon rains. Vegetation in the region is dominated by ponderosa pine, oak, pinyon-juniper, aspen, mountain shrub, and agricultural 116 �communities. Key forage species for black bears include gambel oak (Quercus gambelii), chokecherry (Padus virginiana), serviceberry (Amelanchier alnifolia), hawthome (Crataegus spp), squaw apple (Peraphyllum ramosissimum), angelica (Angelica spp), sweet cicily (Osmorhiza spp), cow parsnip (Herac/eum sphondylium) and waterleaf (Hydrophyllum spp ). Durango is predominately surrounded by public land managed by the San Juan National Forest, BLM, CPW, La Plata County and the City of Durango. The vicinity of Durango is considered high quality bear habitat, and the town has consistently experienced high rates of bear-human conflicts. METHODS Objective 1: Testing management strategies to reduce bear-human conflicts Given that the primary cause of black bear-human conflicts has been attributed to the availability of human foods to bears, it has been suggested that the most effective strategy to reduce conflicts is to reduce the availability of that food source (Peine 2001, Beckmann et al. 2004, Gore et al. 2005, Spencer et al. 2007). This strategy has had some success within national parks (Greenleaf et al. 2009), and anecdotally in some communities (Mammoth Lakes CA, Juneau AK, Whistler BC), but no research has ever scientifically tested the benefits of "cleaning up" a town. Given the high price to operationally "bearproof' a community, municipalities must have definitive evidence that such an effort would significantly decrease conflict activity before initiating major changes to waste storage and collection practices. As part of this project we will be implementing the first experimental test of wide-scale urban bear-proofing for reducing bear-human conflicts. To do this, we will drastically reduce the accessibility of anthropogenic foods known to attract bears (garbage, bird-feeders, pet food, etc) within 2 designated 'treatment' areas, while simultaneously monitoring 2 comparable 'co~trol' areas where no action will occur. In the treatment areas we will provide bear-proof garbage containers, canvass citizens to discourage food outside of secure structures (bird-feeders, pet food, etc), conduct daily patrols to remove human foods, and provide strict enforcement. Each area will contain approximately 500 homes in residential neighborhoods. Treatment and control areas will be monitored for 3 years after the experiment has commenced, and we will track the number of conflicts and their severity among our experimental units. Conflicts will be recorded from weekly monitoring and from calls received by CPW, the City of Durango, and Bear Smart Durango (local non-profit organization). During summer 201 l project personnel collected pre-treatment data (data collection for 2012 is ongoing) on bears accessing garbage in Durango. In July and August, months that experience the highest numbers of bear-human conflicts (CPW unpublished data), technicians patrolled each street within proposed treatment/control areas on the day waste removal was scheduled to occur (when maximum human food was assumed to be available to bears). Technicians conducted patrols from ~05:30 - 06:30 AM and recorded the locations where there was evidence that bears had obtained garbage or other human food sources. Additionally, during late July, we quantified the "availability" of garbage to bears, by • documenting the location and container type (wildlife-resistant or regular) of every garbage receptacle in the survey area. These data will allow us to track changes in the number of wildlife-resistant containers over the course of the study, and provide an estimate of the amount of human food available to bears in town. In addition to collecting pre-treatment data, we worked with the City of Durango to coordinate the logistics of implementing the bear-proofing experiment in spring 2013. Objective 2: Determining the influence of urban environments on bear behavior and demography To sustainably manage bears in the face of a growing human population and changing landscape conditions, it is critical to elucidate the drivers and dynamics of bear populations. Of those factors that influence bear populations, the expansion of human development is the least understood, most contentious, and has the greatest potential to elicit major population change. To elucidate the influence of human development on bear habitat-use patterns and demography, we are collecting a suite of data types 117 �'-' including locations from collared bears on the urban-wildland interface, survival and reproductive rates of those bears in conjunction with their habitat-use patterns, information on annual summer/fall mast production, and genetic data to estimate bear density in urban and wildland habitat types. We briefly describe data collection methods for this portion of the study below; detailed information is available in Johnson et al. (2011 ). Collaring and Marking Bears - To assess bear movement and habitat-use patterns with respect to human development, we are capturing and collaring adult female bears. We are specifically targeting adult females as they represent the reproductive segment of the population and allow us to obtain information on multiple key vital rates. For example, in addition to being able to track adult female survival, the vital rate with the highest elasticity (Beston 2011), we can use collared females to track fecundity and cub survival, vital rates that are often associated with variation in bear population growth rates (Mitchell et al. 2009, Beston 2011). We have targeted summer trapping efforts within ~ l 0 km of the center of Durango to collar a cohort of bears that experience similar natural food availability, have anthropogenic food resources readily available, and encompass a range of behaviors and habitat-use patterns relative to the urbanwildland interface. Bears are trapped with box traps, which are baited with fish, fruit, human foods (at urban locations) and manufactured scents. Traps are set in the evening and checked the following morning. Adult female bears are fitted with a GPS collar manufactured by Vectronics, and a tooth is pulled for age verification. A collar records a bear's location every hour, and uploads a location to a central database via satellite system every 6 hours. Although trapping efforts are focused on adult females, all bears that are trapped (i.e., males, subadults, yearlings) are uniquely marked with a PIT and ear-tag and are weighed, measured, and sampled for blood and hair.. Evaluating Bear Movement and Habitat-Use Relative to the Urban-Wild/and Interface-To examine movement and habitat-use patterns of bears along the urban-wildland interface we will use GPS collar location data from adult female bears. We will assess the influence of factors such as natural food availability, human food availability, weather, habitat covariates, and individual bear attributes (i.e., age, reproductive status) on bear behavior. During winter 2012, we downloaded hourly GPS location data from the collars during winter den checks, and will continue to download and process this data on an annual basis. We will use locations in conjunction with various types of spatial data to conduct a suite of movement and resource selection analyses (Manly et al. 2002, Mcloughlin et al. 20 l 0, Morales et al. 2010). In terms of spatial data, we will use satellite imagery to track annual spring/early summer forage availability, and ground surveys to track late summer/fall mast availability (see details below). Weather information will be modeled using PRISM spatial data (www.prism.oregonstate.edu/) which interpolates monthly temperature and precipitation patterns across landscapes, accounting for elevation and topography. Covariates related to human development will be derived from existing CPW digital data layers such as parcel density, road density, and census population size. 'Cl ,...; While most habitat and human development information can be extracted from existing spatial data sources, there is no existing data layer that tracks annual variation in late summer/fall hard and soft mast for bears. The abundance of acorn and berry resources for bears is known to be highly variable, depending on annual trends in precipitation and temperature (Noyce and Coy 1989). To account for variation in the availability of natural forage for bears around Durango we conducted weekly mast surveys. Surveys were performed between mid-August and mid-September in 2011, when fruits and nuts should reach peak maturation. In the Durango region, the key mast species for bears are gambel oak, chokecherry, serviceberry, hawthome, squaw apple, and pinyon pine (Beck 1991, Tom Beck, personal communication). We randomly selected 12 transects on public lands to evaluate bear natural food availability. Each transect was l km in length and was situated along an existing trail or stream drainage. For each transect, field technicians recorded the phenological stage and the percentage of plants of each 118 'c,/ �species that exhibited mast in different abundance categories (mast failure, <25% of plants with mast, 25 - 50% of plants with mast, etc). Estimating Demographic Rates-To assess the influence of human development on bear demographic rates and population trends we are using the following data types: 1) survival and reproduction of collared adult female bears, 2) mortalities and removals from marked and unmarked bears in the vicinity of Durango, and 2) samples from non-invasive genetic surveys of bears around Durango and in the Piedra watershed. Collared female bears allow us to track annual survival, fecundity and cub survival, parameters we monitored in FYl 1-12 and which we will continue to monitor for the next 4 years. We used real-time GPS collar locations to assess adult female survival, investigating mortalities and slipped collars when GPS locations were stationary for multiple sampling points. Fecundity and cub survival were monitored from den checks of collared females. Numbers of newborn cubs provide information to estimate fecundity rates, while repeated annual den checks of collared females allow us to estimate cub survival. Yearlings hibernate with their mothers, so we can observe the number of cubs alive in the den in year t that have survived their first year of life to t+ 1. Adult female survival, fecundity and cub survival will be collectively used in projection models to assess population performance in future analyses (Caswell 2001). In addition to tracking survival and reproduction of collared bears, we are also tracking survival and cause-specific mortality of marked and unmarked bears in the study area. All bears that are trapped are marked with an ear-tag and PIT tag, unique identifiers that we are using to collect data on humanrelated bear mortalities and removals. Mortalities and removals primarily occur from translocations, vehicle collisions, conflict-related euthanasia and harvest. For all bears removed from the study area we collected a hair and tooth sample and recorded the date, mortality/removal cause, location, bear age, sex, weight, and morphological measurements. We will use mark-recapture and recovery analyses to estimate adult male survival and subadult survival, while also gaining valuable information on cause-specific bear mortality around human development. To better understand the influence of urban environments on bear density and population sizes, we are employing non-invasive genetic sampling (Woods et al. 1999, Mowat and Strobeck 2000) to compare these parameters between a bear population around the urban center of Durango and in a nearby "wildland" area. For each area we identified a 36 cell grid (576 km2) where each cell was 4 x 4 km in size; we constructed 1 snare in each cell. Snares consisted of a scented bait hanging high in a tree, surrounded barbed wire around a cluster of trees encircling the bait. When the bears climbed over or under the wire to investigate the bait, they left a hair sample on the barbed wire. In summer 2011 we hung a single strand of barbed wire (50 cm high), and on the other half of the snares we hung two strands (50 and 20 cm high). Our goal with this design was to determine whether the additional strand of wire increased capture probability. In summer 2012 all sites were strung with a single strand of wire. Snares were deployed during the first 2 weeks of June, and we conducted 6 weekly sampling occasions thereafter. On each occasion, we randomly re-baited the snare with anise, strawberry, fish, maple or bacon scent, and collected hair samples off all barbs. Each hair sample was uniquely catalogued according to the site, date, occasion, and barb number. In 2011, we sampled a total of 31 grid cells in Durango (dropping 5 cells where public or motorized access was prohibited) and 9 cells in the Piedra watershed. We did not have the logistical capability to sample both grids in their entirety, so we ran a pilot study on the Piedra to determine whether twice/month sampling (as opposed to weekly) would have significant impacts on DNA quality, DNA contamination (hair samples from> 1 bear/barb), and recapture rate. In 2012, we constructed 35 snares in the Durango grid and 34 snares in the Piedra grid. The layout of the Piedra grid had to be modified in 119 �th 2012 to account for closures associated with the Little Sand fire, which began burning on May 13 2012 (Figure 1). This modification can be easily accounted for in future analyses with spatially-explicit mark recapture statistics (Efford et al. 2009, Gardner et al. 2010). In fall 2011, all hair samples were sent to the laboratory at Wildlife Genetics International for genotyping; genetic results were returned at the end of June 2012. Summary data from the Durango grid is provided, and the remainder of the analyses will occur during FY12-13. Samples collected in 2012 will be sent to the laboratory this fall. 'Cl 'Cl Objective 3: Identifying public attitudes about bear-human encounters Wildlife management agencies must identify the biological factors driving increases in bearhuman conflicts, but they also must identify and incorporate human attitudes and perceptions about this issue into management strategies. This is particularly critical for black bears, as increasing bear-human conflicts around urban development have simulated significant public interest and concern. It is also critical because bear-human conflicts typically arise over bear-use of human foods, prompting investigators to suggest that a critical component of reducing conflicts is managing human behavior (Beckmann et al. 2004, Gore et al. 2008, Baruch-Mordo et al. 2011 ). Thus, in conjunction with Stacy Lischka, Human Dimensions Specialist for CPW, we have initiated a public survey to 1) better understand public perceptions about bears, bear management, and bear-human encounters and 2) explore motivations for compliance and non-compliance with wildlife ordinances designed to reduce bear-human conflicts. To meet those objectives, we developed a three part public mail survey to be conducted in conjunction with our urban bear-proofing experiment. Residents will be surveyed pre-, during, and postimplementation of the experiment, in treatment and control areas, as well as across a larger portion of the community. Surveys will be mailed to all residents within Durango city limits, and a subset of La Plata county residents within the study area. Survey responses will allow us to quantify current public attitudes and perceptions about bears, and how those perceptions change over time in association with a management effort such as wide-scale urban bear-proofing. The survey will also determine the number of residents that have had interactions with bears, the acceptability of management actions by CPW, and factors that promote or inhibit residents from complying with wildlife ordinances. The pre-treatment survey was mailed to 5,852 residents; 4,352 residents in Durango city limits and 1,500 in surrounding areas of La Plata county (Appendix 1). The total valid sample, once surveys mailed to incorrect addresses were returned, was 5,329. Surveys were mailed on January 17th 2012, a reminder postcard was mailed on February 2 nd 2012, and a second survey was mailed to non-respondents on February 29 th 2012. For those people that did not send back a completed survey, we mailed a nonresponse postcard on May l 8th 2012. The postcard had a few background questions so that any systematic biases in respondents could be assessed and incorporated into analyses (Appendix 2). RESULTS AND DISCUSSION 'Cl Objective 1: Testing management strategies to reduce bear-human conflicts During summer 2011 we collected pre-treatment data for the proposed bear-proofing experiment. We observed 129 instances of bears accessing garbage during our weekly surveys in July and August; observations peaked the first week of August. Of those garbage containers accessed by bears, 10% were wildlife-resistant and 90% were regular containers. Bears accessed human food from wildlife-resistant containers when they were not closed properly or the locking mechanism on the lid was broken. In quantifying the availability of garbage to bears, we recorded the location and container type of 1,167 garbage cans in the proposed treatment and control areas. Of those containers, 14% were wildlife resistant and 86% were regular (non-wildlife resistant). This demonstrates the limited residential bear-proofing that currently exists in Durango, and the relevance of conducting an experimental test of wide-scale urban bear-proofing in this community. 120 �This past year, we have worked on the logistics of conducting a wide-scale urban bear-proofing experiment that should commence in spring 2013. A majority of the necessary funds were secured through CPW and the Summerlee Foundation; we are still seeking funds to complete project needs. With funds currently dedicated to the project, we purchased 760 wildlife-resistant containers from Solid Waste Systems (Parker, CO), a company that manufactures products certified by the Living with Wildlife Foundation. This fall, those containers will be fitted with electronic chips and entered into the Durango's Solid Waste Program database. Because all residential waste is removed by the City of Durango, city staff will replace regular garbage containers with the newly purchased wildlife-resistant containers according to CPW's study design. The wildlife-resistant containers will be distributed in late fall and winter after the bears have hibernated so that they are in place for the experiment in spring 2013. Objective 2: Determining the influence of urban environments on bear behavior and demography Between May 15t1t 2011 and August 15 th 2012, a total of 162 different bears were marked as part of this study, during 287 bear captures. Information about these captures is described below for each discrete capture season: summer 2011, winter 2012, and summer 2012 (ongoing; Table 1). During summer 2011 we conducted 92 total bear captures; 71 captures were unique individuals and 21 were recaptures. Of the unique individuals captured, there were 30 females, 38 males, and 3 cubs of unidentified sex (cubs were released without being immobilized and thus, gender was not determined; Table 1). We collared a total of 26 adult females, however two bears slipped out of their collars and were not recaptured, leaving 24 collared bears at the end of the field season. The mean estimated age of bears ~1 year-old on their initial capture date was 5.3 (5.7 for females and 5.1 for males), and the mean weight was 80.8 kg (59.9 kg for females and 97.4 kg for males). The mean age of collared females, based on tooth cementum, was 6 years, and estimated ages ranged from 2.5 to 23. In total, we placed traps/snares at 102 different locations (26 on public land and 76 on private land) and we had 1,253 trap nights. Capture success generally peaked during the first couple weeks of June and was highly variable throughout the remainder of the summer (Figure 2). We visited the winter dens of22 collared females between January and March 2012. Although we had 24 adult female bears collared in fall 2011, 1 female was harvested (B49), and we could not locate the den of 1 bear wearing a Lotek collar (B51 ). Nine females did not have any cubs or yearlings, 3 bears had yearlings (6 yearlings in total) and 10 had newborn cubs (21 cubs in total; 11 females and 10 males). Of those females with yearlings, 1 bear had 1 yearling, 1 bear had 2 yearlings, and 1 bear had 3 yearlings. Of those females with newborn cubs, 1 had only 1 cub, 7 bears had twins, and 2 bears had triplets. We PIT and ear-tagged yearlings in the den, recorded information on weight and body size, and collected hair and blood samples. We also PIT tagged newborn cubs, and recorded their sex and weight. One collared bear (B43) died during the immobilization process in the den. Between May 15 th and August 15 th 2012, we conducted summer captures to obtain a sample of 40 GPS collared adult females (captures are currently ongoing). During that time there were 153 total captures; 74 were unique individuals and 79 were recaptures (Table 1). Of the unique individuals captures, 33 bears were females, 3 7 were males, and 4 cubs were of unidentified sex (cubs were not immobilized). The mean estimated age of bears ~1 year-old on their initial capture date was 4.7 (5.3 for females and 4.3 for males) and the mean weight was 72.3 kg (58.3 kg for females and 83.5 kg for males). This summer, to date, 22 new adult females have been collared. Given mortalities and slipped collars (3 collars were slipped in spring/summer 2012), 37 females were collared as of August 15t\ and trapping will continue through mid-September or until 40 GPS collars have been deployed. To date, traps have been placed in ~78 different trap locations (26 on public land and 52 on private land) for approximately 1,024 trap nights. Capture success generally climbed each week until the second week of July, and has remained high (Figure 2). The increase in capture success in 2012 is likely due to extra trapping effort, as 121 �we increased our weekly trap nights from 5 nights/week to 7 nights/week and had a higher number of traps that were baited and set on a consistent basis. Although we are still working to deploy collars for the study, the Vectronics GPS technology has been highly efficient at tracking collared bears for movement rates, habitat-use patterns, den site locations, and daily survival. To date, we have obtained >60,000 locations from 48 different female bears (Figure 3). Additionally, the GPS collar technology has allowed us to observed long-distance movements by females, particularly during the estrous period; data which has been rarely collected and reported. For example, this past June, 3 different collared bears traveled between ~60 and ~320 km in different directions from the study area (Figure 4). Two of the bears returned to their original home ranges, and I died in a vehicle collision as she appeared to be returning to Durango (B35). In 2011, mast surveys revealed that the peak timing for serviceberry maturation was in midAugust, for chokecherries it was during the last week of August and first week of September, for squaw apples it was around September I s1, and for acorns it was during in the first two weeks of September. Hawthorne berries and pinyon cones were only observed on 2 of 12 transects; neither had reached peak maturation by mid-September. Across the transects, on average, <25% of gambel oak, chokecherry, squaw apple, and hawthorn plants had mast production. Serviceberry and pinyon production was categorized as a complete failure for the year. Between I May 2011 and August 15 th 2012, 25 bears were removed from the vicinity of Durango due to non-harvest, human-related causes. Of those bears that were removed, 9 were lethally removed due to nuisance behavior (breaking into houses, killing livestock, etc), IO were killed in vehicle collisions (including 2 collared females), 4 were translocated due to conflicts with people (including I collared female), and 2 died from research activities (including I collared female). Of those mortalities and removals, 17 bears were unmarked and 8 were marked/collared for the research project ( I marked bear was a lethal conflict removal outside the study area); there were 8 adult females, 4 adult males, 2 subadult females, 7 subadult males, and 4 cubs. In addition, approximately 20 bears were harvested in the greater Durango area (GMUs 74, 75, and 751), three of which were marked by the research project (1 collared female and 2 adult males). ....,I 'Cl In summer 2011, we collected 998 hair samples from the Durango and Piedra hair-snare grids; 743 samples from Durango and 255 from Piedra. Over the 6 sampling occasions from 31 snares around Durango we collected 224, 167, 138, 77, 68, and 69 hair samples, respectively. Over the 3 sampling occasions from 9 snares in the Piedra we collected 127 samples; 46, 50, and 31 samples/occasion, respectively. We also collected 128 additional samples from 10 snares in the Piedra watershed that were only checked on a single occasion. We received the genetic results back from Wildlife Genetics International at the end of June 2012, and have summarized the Durango data. Of the 743 hair samples submitted to the laboratory, good genotypes were obtained for 438 samples. Of the remaining samples that did not produce a valid genotype, 193 did not contain enough genetic material, 104 failed during analyses for other reasons, 4 samples were not black bear, and 2 were contaminated (hair from > l bear in the sample). Across the 438 valid samples there were 107 different individuals (61 females and 47 males) detected during 192 "captures" (multiple hair samples from a single bear during I sampling occasion were considered l "capture"). Of the different individuals, 21 were only detected in I sampling occasion and 86 were detected in> I occasion (recaptures). The probability of detecting a bear within any single sampling occasion was ~0.21, and across all sampling occasions was ~O. 76. More detailed analyses of these data will be included in the FY12-13 report. In summer 2012, we collected 1,367 hair samples from the Durango and Piedra grids; 586 samples from Durango and 781 samples from Piedra. Over the 6 sampling occasions from 35 snares around Durango we collected 92, 136, 59, 55, 142, and I 02 samples, respectively. Over the 6 sampling 122 �occasions from 34 sites in the Piedra watershed we collected 73, 135, 142, 118, 144, and 169 samples respectively. Samples will be sent to Wildlife Genetics International this fall for genetic analysis. Objective 3: Identifying public attitudes about bear-human encounters Of the 5,334 valid surveys that were mailed to residents, we received 2,947 completed surveys; 2, 170 from Durango residents and 777 from La Plata county residents. The overall response rate was 55%. Non-response postcards were mailed to 2,375 residents and 354 postcards were returned (15%). Survey results are being electronically recorded so this data can be analyzed in FY12-13. SUMMARY AND FUTURE PLANS During FYl 1-12 we successfully coordinated field logistics and conducted several aspects of data collection (monitoring garbage-related bear-human conflicts, trapping and collaring bears, tracking human-related bear mortalities, assessing summer/fall forage availability, implementing DNA hair-snare surveys, and conducting a public survey). We will continue these field activities through 2015, and begin data analyses as field data are compiled. Project collaborators will continue to seek additional funding to implement the remaining activities outlined in the research proposal. These activities include the implementation of an urban bear-proofing experiment, increasing the number of GPS collared female bears, and purchasing telemetry collars for a translocation study. In addressing the objectives of this project we hope to better understand the influence of urban environments on bear populations, elucidate the relationship between bear-human conflicts and bear behavior and population trends, develop tools to promote the sustainable management of bears in Colorado, and ultimately, identify solutions for reducing bear-human conflicts in urban environments. LITERATURE CITED Baruch-Mordo, S., S.W. Breck, K.R. Wilson, and J. Broderick. 2011. The carrot or the stick? Evaluation of education and enforcement as management tools for human-wildlife conflicts. Plos One 6: e15681. Baruch-Mordo, S., S.W. Breck, K.R. Wilson, and D.M. Theobald. 2008. Spatiotemporal distribution of black bear-human conflicts in Colorado, USA. Journal of Wildlife Management 72:1853-1862. Baruch-Mordo, S., K.R. Wilson, D. Lewis, J. Broderick, J. Mao, and S.W. Breck. 2010. Roaring Fork Valley urban black bear ecology study: progress report to the Colorado Division of Wildlife. Contact Sharon Baruch-Mordo for copy. Email: sharonb m@yahoo.com Beck, T.D.I. 1991-. Black bears of west-central Colorado. Technical Publication No. 39, Colorado Division of Wildlife, Colorado. Beckmann, J.P., and J. Berger. 2003a. Using black bears to test ideal-free distribution models experimentally. Journal of Mammalogy 84:594-606. Beckmann, J.P., and J. Berger. 2003b. Rapid ecological and behavioural changes in carnivores: the response of black bears (Ursus americanus) to altered food. Journal of Zoology 261:207-212. Beckmann, J.P., L. Karasin, C. Costello, S. Matthews, and Zoe Smith. 2008. Coexisting with black bears: perspectives from four case studies across North America. Wildlife Conservation Society Working Paper No 33. Beckmann, J.P., C.W. Lackey, and J. Berger. 2004. Evaluation of deterrent techniques and dogs to alter behavior of"nuisance" black bears. Wildlife Society Bulletin 32: 1141-1146. Beston, J.A. 2011. Variation in life history and demography of the American black bear. Journal of Wildlife Management 75:1588-1596. Caswell, H. 200 I. Matrix population models: construction, analysis, and interpretation. Second Edition, Sinauer Associates, Sunderland, Massachussetts. Efford, M.G., D.K. Dawson, and D.L. Borchers. 2009. Population density estimated from locations of individuals on a passive dector array. Ecology 90:2676-2682. 123 �'-' 'Cl ...,; 'Cl Garshelis, D.L., and H. Hristienko. 2006. State and provincial estimates of American black bear numbers versus agency assessments of population trend. Ursus 17: 1-7. Gardner, B., J.A. Royle, M.T. Wegan, R.E. Rainbolt, P.O. Curtis. 2010. Estimating black bear density using DNA data from hair snares. Journal of Wildlife Management 74:318-325. Gore, M.L., W.F. Siemer, J.E. Shanahan, D. Schuefele, and D.J. Decker. 2005. Effects on risk perception of media coverage of a black bear-related human fatality. Wildlife Society Bulletin 33 :507-516. Gore, M.L., B.A. Knuth, C.W. Scherer, and P.O. Curtis. 2008. Evaluating a conservation investment designed to reduce human-wildlife conflicts. Conservation Letters 1: 136-145. Greenleaf, S.S., S.M. Matthews, R.G. Wright, J.J. Beecham, and H.M. Leithead. 2009. Food habitat of American black bears as a metric for direct management of human-bear conflict in Yosemite Valley, Yosemite National Park, California. Ursus 20:94-101. Hostetler, J.A., J.W. McCown, E.P. Garrison, A.M Neils, M.A. Barrett, M.E. Sunquist, S.L. Simek, and M.K. Oli. 2009. Demographic consequences of anthropogenic influences: Florida black bears in north-central Florida. Biological Conservation 142:2456-2463. Hristienko, H., and J.E. McDonald Jr. 2007. Going into the 21 st century: a perspective on trends and controversies in the management of the American black bear. Ursus 18:72-88. Johnson, H.E, C.J. Bishop, M.W. Alldredge, J. Brodrick, J. Apker, S. Breck, K. Wilson, and J. Beckmann. 2011. Black bear exploitation of urban environments: finding management solutions and assessing regional population effects. Research Proposal, Colorado Division of Parks and Wildlife, Fort Collins, USA. Manly, B.F.J., L.L. McDonald, D.L. Thomas, T.L. McDonald, and W.P. Erickson. 2002. Resource selection by animals; statistical design and analysis for field studies. Second Edition. Kluwer Academic Publishers, Boston, Massachussetts. McLoughlin, P.O., D.W. Morris, D. Fortin, E. Vander Wal, and A.L. Contasti. 2010. Considering ecological dynamics in resource selection functions. Journal of Animal Ecology 79:4-12. Mitchell, M.S., L.B. Pacifici, J.B. Grand, and R.A. Powell. 2009. Contributions of vital rates to growth of a protected population of American black bears. Ursus 20:77-84. Morales, J.M., P.R. Moorcroft, J. Matthiopoulos, J.L. Frair, J.G. Kie, R.A. Powell, E.H. Merrill, and D.T. Haydon. 2010. Building the bridge between animal movement and population dynamics. Philosophical Transactions of the Royal Society Series B 365:2289-2301. Mowat, G., and C. Strobeck. 2000. Estimating population size of grizzly bears using hair capture, DNA profiling, and mark-recapture analysis. Journal of Wildlife Management 64: 183-193. Noyce, K.V., and P.L. Coy. 1989. Abundance and productivity of bear food species in different forest types of northcentral Minnesota. International Conference on Bear Research and Management 8: 169-181. Peine, J.D. 2001. Nuisance bears in communities: Strategies to reduce conflicts. Human Dimensions of Wildlife 6:223-237. Spencer, R.D., R.A. Beausoleil, and D.A. Martorello. 2007. How agencies respond to human-bear conflicts: a survey of wildlife species in North America. Ursus 18: 217-229. Theobald, D.M., and W.H. Romme. 2007. Expansion of the US wildland-urban interface. Landscape and Urban Planning 83 :340-354. Woods, J.G., D. Paetkau, D. Lewis, B.N. McLellan, M. Proctor, and C. Strobeck. 1999. Genetic tagging of free-ranging black and brown bears. Wildlife Society Bulletin 27:616-627. Zack, C.S., B.T. Milne, and W.C. Dunn. 2003. Southern oscillation index as an indicator of encounters between humans and black bears in New Mexico. Wildlife Society Bulletin 31 :517-520. Prepared by _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ Heather E. Johnson, Wildlife Researcher 124 �wt _,I _,i Table 1. Capture information for black bears that have been marked in the vicinity of Durango, CO (collared adult females are identified with an"*"). Only information from the initial capture of each individual is shown {no recaEtures}. Bear ID Bl B2 B3 B4 B5 B6* B7* B8* B9 BIO* Bl I B12 B13 B14* B15 B16 B17* B18* B19 B20 B21* B22 B23 B24* B25* B26 B27* B28 B29 B30* B31 B32 B33 B34 B35* B36 B37 B38 B39 B40* Capture Date 5/10/2011 5/12/2011 5/13/2011 5/16/2011 5/16/2011 5/17/2011 5/17/2011 5/18/2011 5/26/2011 5/26/2011 6/3/2011 6/2/2011 6/3/2011 6/6/2011 6/6/2011 6/7/2011 6/7/2011 6/8/2011 6/9/2011 6/9/2011 6/10/2011 6/10/2011 6/13/2011 6/14/2011 6/15/2011 6/15/2011 6/16/2011 6/16/2011 6/21/2011 6/22/2011 6/24/2011 6/24/2011 6/28/2011 6/28/2011 7/5/2011 7/6/2011 7/7/2011 7/13/2011 7/13/2011 7/21/2011 UTM Easting 246233 271495 271495 270950 270227 243210 243225 271478 238803 269869 252163 253216 253216 252157 253216 253216 256936 256918 235193 243258 252298 252163 246350 243252 239003 252164 243252 253233 239840 235911 239840 243252 239294 239001 246350 239840 243252 243236 251222 248550 UTM Northing 4142768 4130889 4130894 4127914 4139984 4128716 4133053 4130892 4126790 4139040 4137968 4137387 4138868 4137967 4138868 4138868 4134633 4134625 4128894 4133040 4136435 4137968 4135617 4133030 4134158 4137966 4133030 4138873 4126949 4128916 4126949 4133030 4133260 4134154 4135617 4126949 4133030 4128710 4133120 4131645 Sex M M M M M F F F M F M M M F M M F F M M F M M F F M F M M F M F M M F M M M M F Age 1 9 6 3 6 4 4 4 1 7 8 5 3 7 3 7 4 8 9 10 8 8 3 7 4 10 23 6 1 6 4 1 1 3 3 4 1 8 6 4 Kg 35.4 144.2 130.2 84.4 135.2 62.6 63.5 51.7 34.9 80.7 130.2 103.4 59.0 58.1 58.1 117.0 51.7 61.7 146.5 131.5 69.4 87.5 64.9 64.9 63.5 108.9 75.3 101.2 49.0 60.3 85.3 19.1 34.9 85.3 44.5 67.1 39.0 145.1 149.7 80.7 \_.I _,, 1w 1w' w ~ .._, ~ 1w ~ 'w .._, 1w ~ _,i '-' 1w 1w 1w' ~ '-" \wl 1w '-' 'w I-, ~ ~ w' 1w ~ ._, ~ '-" 125 _, ~ '-' �'<al ,_, ..., '.-i '-.I "-' .._ '..,I .,,.; 'Cl ~ '-' .I I,.) 'cl 'el '«;I ....._-__, .I ... ~ \al ...., 'el ~ ..._ 'cl -..I 'Cl ..., ...,, "1111 1.'661 .,I ._ 'SI ._ ,.,,, B41 B42* B43* B44 B45 B46* B47* B48 B49* B50* B51* B52* B53 B54 B55* B56 B57* B58 B59 B60 B61 B62 B63 B64 B65* B66 B67* B68 B95 B96 B97 B98 B99 BIOO BIOi Bl02 Bl03 Bl04 Bl05 Bl06 Bl07 Bl08 Bl09 BllO Bl 11 7/22/2011 7/26/2011 8/3/2011 8/3/2011 9/3/2011 8/5/2011 8/8/2011 8/10/2011 8/11/2011 8/11/2011 8/12/201 I 8/12/2011 8/15/2011 8/16/2011 8/18/2011 8/29/2011 8/30/201 I 8/31/2011 9/1/2011 9/2/2011 9/3/2011 9/6/2011 9/7/2011 8/6/2011 9/15/2011 9/20/2011 9/21/2011 9/21/2011 1/19/2012 1/19/2012 1/19/2012 1/26/2012 2/27/2012 2/27/2012 2/29/2012 2/29/2012 3/1/2012 3/1/2012 3/6/2012 3/8/2012 3/8/2012 3/14/2012 3/14/2012 3/15/2012 3/15/2012 4132272 4141391 4142791 4132487 4139587 4128720 4131581 4139620 4128720 4139587 4130370 4139587 4128720 4130516 4134423 4132993 4135903 4135903 4132935 4133020 4130321 4128530 4131251 4141969 4134848 4130163 4134626 4133006 4142276 4142276 4142276 4134879 4137615 4137615 4126214 4126214 4120831 4120831 4132714 4139774 4139774 4138519 4138519 4140021 4140021 237368 245945 246183 765141 245965 243435 251783 245914 243435 245965 249049 245965 243435 251898 251464 246321 243374 243374 243952 242187 244602 245790 248612 245850 243948 240731 256930 249067 247647 247647 247647 257183 249929 249929 248090 248090 243713 243713 240146 268055 268055 245785 245785 244547 244547 126 M F F M M F F F F F F F M M F M F M M F M M M M F F F M M M M M M F M F M M F F F F M M F 3 8 11 I 6 4 6 I 4 11 12 4 7 2 3 IO 3 3 15 2 7 I 2 I 4 I 3 8 I I I I cub cub cub cub cub cub cub cub cub cub cub cub cub 67.1 69.9 85.3 35.4 176.0 58.1 53.5 26.3 55.3 101.2 61.7 65.3 163.3 52.6 49.0 166.9 46.3 48.1 153.3 35.4 213.6 22.7 37.2 28.6 89.8 39.9 52.6 207.7 44.0 44.0 49.0 20.9 I.I 1.2 1.8 1.7 1.9 1.7 2.0 2.7 2.7 1.4 1.4 2.5 2.5 �~ w B112 B113 B114 B115 B116 B117 B118 B119 B120 B121 * B122* B123 B124* B125* B126 B127* B128* B129* B130 B131 B132 B133* B134* B135 B136 B137 B138 B139 B140 B141* B142 B143* B144* B145* B146 B147 B148 B149 B150 B151 B152* B153 B154 B155 B156 3/15/2012 3/17/2012 3/17/2012 3/17/2012 3/20/2012 3/20/2012 3/22/2012 3/22/2012 5/27/2012 5/29/2012 5/30/2012 6/5/2012 6/6/2012 6/8/2012 6/8/2012 6/10/2012 6/11/2012 6/14/2012 6/22/2012 6/23/2012 6/28/2012 6/29/2012 6/30/2012 6/30/2012 7/1/2012 7/5/2012 7/5/2012 7/5/2012 7/5/2012 7/5/2012 7/6/2012 7/6/2012 7/7/2012 7/7/2012 7/7/2012 7/10/2012 7/11/2012 7/15/2012 7/16/2012 7/26/2012 7/17/2012 7/17/2012 7/17/2012 7/19/2012 7/19/2012 244547 245228 245228 245228 240909 240909 243524 243524 254732 251670 249059 240102 249158 244618 251670 239005 239005 254576 250152 765047 765047 765932 765932 252014 765047 249059 254997 238245 763921 765132 254997 241210 238245 763921 254739 241334 255983 244618 241334 243888 241210 249059 253439 241334 252621 4140021 4143164 4143164 4143164 4134002 4134002 4146585 4146585 4133249 4132767 4132998 4128939 4127065 4132132 4132767 4134459 4134159 4135043 4127691 4131635 4131635 4127651 4127651 4133509 4131635 4132998 4135825 4131204 4132873 4132506 4135825 4137115 4131204 4132873 4133234 4138018 4135921 4132132 4138018 4129546 4137114 4132998 4134693 4138018 4130532 127 F F M M M F M F F F F M F F F F F F M M M F F M F M F M M F M F F F M M M M M M F M M F F cub cub cub cub cub cub cub cub 1 4 5 2 7 8 1 10 8 6 1 6 1 3 8 1 2 2 2 1 3 3 3 3 9 6 5 1 3 1 2 3 8 2 3 2 1 2.6 2.5 2.0 2.0 2.0 2.0 2.5 2.9 20.9 76.2 66.2 48.1 80.7 98.9 15.9 58.1 56.2 54.4 12.7 111.6 20.4 49.0 90.7 20.9 46.3 26.8 30.8 30.4 67.1 55.3 37.2 45.4 72.6 70.3 110.2 43.5 73.9 45.4 53.5 60.3 99.8 49.9 63.5 30.8 26.3 'w ~ 'w ~ 1-, ~ 1w ~ 'w ~ 1w' l_.i ~ 1w' w '- 1w, '-' 'wl w 'w 'w' wl ._ W' iw 1w' ~ l.,J '-' -' ~ 1w ., 'WI Iv �..., \ell 'Cl ...., ...., "Cl "el ,.-I .._ 1,./ _, "161 'Cl ...,, ~ '411 'Cl ._ ,._ '-.I 'Cl ..., le/ 'cl 'Cl \(SJ/ -el ...., val >,cl B157 B158 B159 B160 B161 * B162 B163 B164 B165* B166 B167* B168 B169 B170 B171 B172 B173* B174* B175* B176 B177 B178 B179 B180* B181 * B182 B190 B191 B192 B193 B194 B195 B196 7/19/2012 7/20/2012 7/21/2012 7/24/2012 7/25/2012 7/25/2012 7/26/2012 7/28/2012 7/29/2012 7/29/2012 7/31/2012 7/31/2012 7/31/2012 8/1/2012 8/2/2012 8/2/2012 8/3/2012 8/3/2012 8/3/2012 8/3/2012 8/4/2012 8/4/2012 8/5/2012 8/5/2012 8/5/2012 8/8/2012 8/9/2012 8/11/2012 8/11/2012 8/12/2012 8/12/2012 8/13/2012 8/14/2012 4144294 4134789 4127920 4132998 4134789 4132998 4134875 4133863 4133706 4130532 4139143 4134693 4132998 4132998 4137051 4139143 4139143 4128740 4128609 4130532 4139143 4132988 4139143 4141533 4138557 4132998 4128959 4132998 4128959 4129360 4128712 4132998 4132998 248417 252546 242236 249059 242546 249059 243954 242611 251815 252621 248578 253439 249059 249059 248192 248578 248578 253341 254916 252621 248578 249059 248578 248939 247127 259049 245293 249059 245293 243652 243218 259049 249059 'SI 'cl 'ct/ - ..I ..., 'cl 'CJ/ le/ ._ 'ct/ .I ..,., . 128 M M F M F M F M F M F M F M M F F F F M M M M F F M M M M M M M F 6 3 2 2 5 10 1 6 12 8 18 4 2 4 2 2 5 3 10 2 7 2 1 3 3 2 12 4 9 5 5 6 2 136.1 50.8 39.9 64.0 76.2 108.0 37.6 117.0 85.3 121.6 58.1 87.5 44.5 119.3 26.8 28.1 73.9 43.5 76.2 54.4 93.9 35.4 35.4 71.7 58.1 70.8 153.3 60.3 148.8 87.5 137.0 151.0 42.2 �Figure I. Map of the locations of the 2012 hair snare sites for the Durango and Piedra non-invasive genetic sampling grids. ,·~~·~:- :.w .- ' .· -.r~!:~:~>-~. -~ \"' ,'·, ,. •• ,,. '( .. ,r. )~~ ,_,-~ -~ :~ ' ~ . ,±.,~ · :; • . ~ ~ \·---=· ' : 7 < ' t ['-'t ~ ·, .,·1t .L.' __, ·:'·• ~ -~ .... ! '. -• • :·;-!! • 2012 Hair s nare Locations Major Roads Perimeter of Little Sand Fire 3.75 7.5 '.:'-:-, , .1 -~ • . ',\ I J~~•••·, . ·~· ,::;,-;:- f"'!""'":"""~~...rt ... :.~ 'L;.,. .~ 129 �Figure 2. Number of black bear captures by week from May 15 th through September 3rd for the 2011 and 2012 summer trapping seasons (2012 is currently ongoing). 20 --t--2011 18 -4--2012 u, 16 cu ...a.~ 14 '-' ,, " a... 12 I \ I \ "' -... ~ 10 0 cu .a E :::, z ,..-,) 8 6 4 2 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Week (from May 15th through August) 'Cl ...._ 130 16 17 18 �Figure 3. Locations from 48 adult female black bears collected with GPS collars from May 20 11 to present in the vicin ity of Durango, CO (different colored clusters of points represent di fferent individual bears). t ·, ~ 131 t ,- ..., ~1,} • ·P ( : -~ -!; �Figure 4. Long distance movements of three collared adult female bears during the breeding season in June 20 12. -~~,,. ·-"·"' A~ ~o .,, e .... 0 <¢ ~ ~ ' ~ (.)(t) (D ..,,,o oeo 0 "'""'(t) 0 "' ., ~~ • 1-..I i~----, Bear ID B67 • B57 ,..___ 0 !.1'11tg.1,11,r.- -. ,.,er,.,, 0 ~-- 0 B35 0 l:,{t',lUVff ".£, 5 . 10 '.'I 20 Kilometers ; " f 1HOIJ 1.14•.;. If) ;H .lOC.., K•rtt~J -- • Bloomfield , ...~lt.1. £ lllf,;_•:t.l t 4 Al I r •1t, l Ah.4 IJJ.C.rl{l'IJ.,.\. :r4'.l. ~r,t')fJ H ;1,~ .;.t fll,.J ..c~u, Ii• ,.., .. 21:: m ...IJ'J. · ~4w.f 132 I �Colorado Division of Parks and Wildlife July 2012 – June 2013 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Colorado 3430 3003 3 Federal Aid Project No. W-204-R1 : : : : : Division of Wildlife Mammals Research Predatory Mammal Conservation Black bear exploitation of urban environments: finding management solutions and assessing regional population effects Period Covered: July 1, 2012 – June 30, 2013 Author: H.E. Johnson; project cooperators, C. Bishop, J. Broderick, J. Apker, S. Lischke, S. Breck, J. Beckmann, K. Wilson, M. Reynolds-Hogland, and P. Dorsey. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Across the country conflicts among people and black bears are increasing in frequency and severity, and have become a high priority wildlife management issue. Whether increases in conflicts reflect recent changes in bear population trends or just bear behavioral shifts to anthropogenic food resources, is largely unknown, with key implications for bear management. This issue has generated a pressing need for bear research in Colorado and has resulted in a unique collaboration that builds on the resources and abilities of personnel from 4 entities: Colorado Parks and Wildlife (CPW), the USDA National Wildlife Research Center, Colorado State University, and Wildlife Conservation Society. Collectively, we have designed and implemented a study on black bears that 1) determines the influence of urban environments on bear habitat-use patterns and demography, 2) tests a management strategy for reducing bear-human conflicts, 3) examines public attitudes and behaviors related to bear-human encounters, and 4) develops population and habitat models to support the sustainable monitoring and management of bears in Colorado. This project was initiated in FY10-11; during this past fiscal year we have primarily focused on collecting field data in the vicinity of Durango, Colorado. Specifically, we worked with collaborators and stakeholders on research logistics, trapped and marked black bears, collected GPS collar data on bear locations, monitored demographic rates (adult female survival, adult female fecundity and cub survival) through telemetry and winter den visits, collected data on the availability of late summer/fall mast, tracked human-related bear mortalities and removals from the study area, performed non-invasive genetic mark-recapture surveys, deployed 900 bear-resistant containers for an experiment on the effectiveness of urban-bear-proofing, obtained data on garbage-related bear-human conflicts, and specified a sampling design to assess human compliance with city ordinances. Information from this study will provide solutions for sustainably managing black bears outside urban environments, while reducing bear-human conflicts within urban environments; knowledge that is critical for wildlife managers in Colorado and across the country. 119 �WILDLIFE RESEARCH REPORT BLACK BEAR EXPLOITATION OF URBAN ENVIRONMENTS: FINDING MANAGEMENT SOLUTIONS AND ASSESSING REGIONAL POPULATION EFFECTS HEATHER E. JOHNSON PROJECT NARRATIVE OBJECTIVES To conduct a study on black bears in Colorado that 1) determines the influence of urban environments on bear habitat-use patterns and demography, 2) tests a management strategy for reducing bear-human conflicts, 3) examines public attitudes and behaviors related to bear-human encounters, and 4) develops population and habitat models to support the sustainable monitoring and management of bears. SEGMENT OBJECTIVES 1. Work with personnel from CPW Area 15, CPW Southwest Region, the City of Durango, La Plata County, US Forest Service (Columbine and Pagosa Ranger Districts), Bureau of Land Management (BLM; Tres Rio Field Office), Southern Ute Tribe, and private landowners on field research logistics. 2. Trap and collar adult female black bears in the vicinity of Durango to collect data on bear habitatuse patterns and demography. 3. Monitor bear locations and survival via global position system (GPS) collar locations. 4. Monitor bear fecundity and cub survival through winter den investigations of collared adult female bears. 5. Obtain data on summer/fall natural food availability for bears based on the phenology and abundance of gambel oak, serviceberry, chokecherry, hawthorne, pinyon pine and squaw apple. 6. Track human-related bear mortalities and removals around Durango from lethal conflict managment, vehicle collisions, harvest, and translocations. 7. Perform non-invasive genetic mark-recapture surveys to estimate bear density and population size around Durango (urban site) and in the Piedra watershed (wildland site). 8. Deploy 900 bear-resistant garbage containers for an experiment on the effectiveness of wide-scale urban bear-proofing for reducing bear-human conflicts. 9. Collect data on the frequency of bears accessing human garbage in treatment and control areas for an urban bear-proofing experiment. 10. Specify a sampling design to quantify compliance of human behavior with wildlife ordinances. INTRODUCTION In Colorado and across the country, conflicts among people and black bears (Ursus americanus) appear to be increasing in number and severity (Hristienko and McDonald 2007, Baruch-Mordo et al. 2008, CPW unpublished data). Bear-human conflicts can result in public safety concerns, property damage, bear mortality (i.e. euthanasia), and high management costs, and thus, have become a critical wildlife management issue. While wildlife agencies have used a variety of tools to try to minimize bearhuman conflicts (i.e., education, aversive conditioning of bears, and modifications to harvest), conflict rates have continued to rise. Whether increases in bear-human conflicts reflect recent changes in the bear population or just behavioral shifts to anthropogenic food resources, is largely unknown, as bear population parameters have been exceeding difficult to estimate (Garshelis and Hristienko 2006). Without a thorough understanding of the relationship between conflict rates and bear behavior and population 120 �dynamics, it has been difficult for wildlife agencies to successfully reduce conflicts through bear management. While there is uncertainty about how to reduce bear-human conflicts, two key factors thought to exacerbate this problem are expanding human development and climatic variation. Colorado has had one of the highest rates of exurban development in the nation (Theobald and Romme 2007), and this development has resulted in additional human food on the landscape in the form of garbage, agricultural resources, fruit trees, etc. The availability of human food to bears has been identified as the primary cause of bear-human conflicts (Spencer et al. 2007, Beckmann et al. 2008, Greenleaf et al. 2009), as bears are opportunistic foragers that will readily take advantage of this resource. Bear-use of human food not only increases interactions between bears and people but has been found to alter bear activity patterns, foraging behavior, movement rates, and even survival and reproductive rates (Beckmann and Berger 2003a, Beckmann and Berger 2003b, Hostetler et al. 2009), having the potential to significantly influence both bear behavior and demography. This phenomenon is further complicated by variation in annual weather patterns, as bear-use of human development appears to increase when natural foods are in short supply (Zack et al. 2003, Baruch-Mordo et al. 2010). Because bears predominately consume vegetation, recent patterns of drought in Colorado have caused natural food failures for bears in some years. As a result, bears may be increasing their reliance on human foods, with associated behavioral and demographic impacts. While the effects of urbanization and climate have critical implications for modifying bearhabitat relationships, they also have critical implications for increasing rates of bear-human conflicts. To develop successful strategies to reduce conflicts while maintaining viable bear populations, wildlife agencies must understand how factors such as climate, natural food availability, human food ability, and management influence the behavior and dynamics of bear populations. To address these questions, Colorado Parks and Wildlife has partnered with the USDA National Wildlife Research Center, Wildlife Conservation Society and Colorado State University. Collectively, we initiated a project in FY10-11 to 1) determine the influence of urban environments on bear habitat-use patterns and demography, 2) test a management strategy for reducing bear-human conflicts, 3) examine public attitudes and behaviors related to bear-human encounters, and 4) develop population and habitat models to support the sustainable monitoring and management of bears in Colorado (Johnson et al. 2011). This information should provide solutions for sustainably managing black bears outside urban environments, while reducing bear-human conflicts within urban environments; knowledge that is critical for wildlife managers in Colorado and across the west. During FY12-13, we worked with collaborators and stakeholders on research logistics, trapped and marked black bears, collected GPS collar data on bear locations, monitored demographic rates (adult female survival, adult female fecundity and cub survival) through telemetry and winter den visits, collected data on the availability of late summer/fall mast, tracked human-related bear mortalities and removals from the study area, performed non-invasive genetic mark-recapture surveys, deployed 900 bear-resistant containers for an experiment on the effectiveness of urban-bear-proofing, obtained data on garbage-related bear-human conflicts, and specified a sampling design to assess human compliance with city ordinances. Our efforts focused largely on collecting field data to meet research objectives 1-3, information which will eventually be used to address objective 4. We report general summary information from field activities over the past year; detailed analyses of field data will occur in future years. STUDY AREA To meet study objectives, a combination of site-specific field data and statewide data will be required. Site-specific field data is being collected in the vicinity of Durango, and is the focus of this progress report. Regional and statewide analyses will be conducted in future years. The town of Durango contains ~17,000 people (within city limits) and sits at 1,985 m along the Animas river valley. The town 121 �is surrounded by mountainous terrain ranging in elevation from ~1,930 to ~3,600 m, and is generally characterized by mild winters and warm summers that experience monsoon rains. Vegetation in the region is dominated by ponderosa pine, oak, pinyon-juniper, aspen, mountain shrub, and agricultural communities. Key forage species for black bears include gambel oak (Quercus gambelii), chokecherry (Padus virginiana), serviceberry (Amelanchier alnifolia), hawthorne (Crataegus spp), squaw apple (Peraphyllum ramosissimum) and pinyon pine (Pinus edulis). Durango is predominately surrounded by public land managed by the San Juan National Forest, BLM, CPW, La Plata County and the City of Durango. The vicinity of Durango is considered high quality bear habitat, and the town has consistently experienced high rates of bear-human conflicts (Baruch-Mordo et al. 2008, CPW unpublished data). METHODS Objective 1: Determining the influence of urban environments on bear behavior and demography To sustainably manage bears in the face of a growing human population and changing landscape conditions, it is critical to elucidate the drivers and dynamics of bear populations. Of those factors that influence bear populations, the expansion of human development is the least understood, most contentious, and has the greatest potential to elicit major population change. To elucidate the influence of human development on bear habitat-use patterns and demography, we are collecting a suite of data types including locations from collared bears on the urban-wildland interface, survival and reproductive rates of those bears in conjunction with their habitat-use patterns, information on annual summer/fall mast production, and genetic data to estimate bear density in urban and wildland habitat types using markrecapture methods. We briefly describe data collection methods for this portion of the study below; detailed information is available in Johnson et al. (2011). Collaring and Marking Bears – To assess bear habitat-use patterns and demographic rates with respect to human development, we are capturing and collaring adult female bears. We are specifically targeting adult females as they represent the reproductive segment of the population and allow us to obtain information on multiple key vital rates that drive population growth. For example, in addition to being able to track adult female survival, the vital rate with the highest elasticity (Beston 2011), we can use collared females to track fecundity and cub survival, vital rates that are often associated with variation in bear population trends (Mitchell et al. 2009, Beston 2011). We have targeted summer trapping efforts within ~10 km of the center of Durango to collar a cohort of bears that experience similar natural food availability, have anthropogenic food resources readily available, and encompass a range of behaviors and habitat-use patterns relative to the urbanwildland interface. Bears are trapped with box traps, which are baited with fish, fruit, human foods (at urban locations) and manufactured scents. Traps are set in the evening and checked the following morning. Adult female bears are fitted with a GPS collar (manufactured by Vectronics), and a tooth (first pre-molar) is pulled for age verification. GPS collars record bear locations every hour, and upload realtime locations to a central database via satellite system every 6 hours. Although trapping efforts are focused on adult females, all bears that are trapped (i.e., males, subadults, yearlings) are uniquely marked with a PIT and ear-tag and are weighed, measured, and sampled for blood and hair. Evaluating Bear Movement and Habitat-Use Relative to the Urban-Wildland Interface – To examine movement and habitat-use patterns of bears along the urban-wildland interface, we are using GPS collar location data from adult females. Hourly GPS data are downloaded from the collars in the field on a biannual basis (during early fall and winter den checks). We will use those locations to assess the influence of factors such as natural food availability, human food availability, weather, habitat covariates, and individual bear attributes (i.e., age, reproductive status) on bear movement and resource selection patterns (Manly et al. 2002, McLoughlin et al. 2010, Morales et al. 2010). For spatial data, we will use satellite imagery to track annual spring/early summer forage availability, and ground surveys to 122 �track late summer/fall mast availability (see details below). We will obtain information on elevation, aspect, slope and terrain ruggedness information from digital elevation models. Weather information will be acquired from PRISM spatial data (www.prism.oregonstate.edu/) which interpolates monthly temperature and precipitation patterns across landscapes, accounting for elevation and topography. We will derive spatial models on distances to perennial water sources and watershed drainages from the National Hydrology Dataset. Vegetation type and cover layers will be generated from the USFS LandFire datasets (http://www.landfire.gov/vegetation.php). Covariates related to human development (e.g., density of human structures and paved roads) will be derived from existing CPW and La Plata County digital data layers on locations of human structures, roads, and census information. While most habitat and human development information can be extracted from existing spatial data sources, there is no existing data layer that tracks annual variation in late summer/fall hard and soft mast for bears. The abundance of acorn and berry resources for bears is known to be highly variable, depending on annual trends in precipitation and temperature (Noyce and Coy 1989). To account for variation in the availability of natural fall forage for bears around Durango, we conducted bimonthly mast surveys. Surveys were performed from early August through mid-September in 2011 and 2012, when fruits and nuts should reach peak maturation and bears are in their hyperphagia stage prior to hibernation. In the Durango region, key mast species for bears are gambel oak, chokecherry, serviceberry, hawthorne, squaw apple, and pinyon pine (Beck 1991, Tom Beck, personal communication). We randomly selected 16 transects on public lands to evaluate bear mast availability. Each transect was 1 km in length and was situated along an existing trail or stream drainage. For each transect, field technicians recorded the phenological stage and the percentage of plants of each species that exhibited mast in different abundance categories (mast failure, <25% of plants with mast, 25 – 50% of plants with mast, etc). Estimating Demographic Rates – To assess the influence of human development on bear demographic rates and population trends we are using the following data types: 1) survival and reproduction of collared adult female bears, and cub survival, 2) mortalities and removals of marked and unmarked bears in the vicinity of Durango, and 3) non-invasive genetic surveys to estimate density and abundance of bears around urban and wildland sites. Collared female bears allow us to track annual survival, fecundity and cub survival (of their offspring); parameters we have monitored since summer 2011 and which we will continue to monitor for the next 3 years. We used real-time GPS collar locations to assess adult female survival, investigating mortalities and slipped collars when GPS locations were stationary for multiple sampling points. Fecundity and cub survival were monitored from den checks of collared females. Numbers of newborn cubs provide information on fucundity, while repeated annual den checks of collared females allow us to estimate cub survival. Yearlings hibernate with their mothers, so we can observe the number of cubs alive in the den in year t that survived their first year of life to t+1. Adult female survival, fecundity and cub survival will be used in projection models to assess population performance (Caswell 2001), particularly in relation to habitat selection. In addition to tracking survival and reproduction of collared bears, we are also tracking survival and cause-specific mortality of marked (i.e., males, subadults) and unmarked bears in the study area. All bears that are trapped are marked with an ear-tag and PIT tag, unique identifiers that we are using to collect data on human-related bear mortalities and removals. Mortalities and removals primarily occur from translocations, vehicle collisions, conflict-related euthanasia and hunter harvest. For all bears removed from the study area we collect a hair and tooth sample and recorded the date, mortality/removal cause, location, bear age, sex, weight, and morphological measurements. We will use mark-recapture and recovery analyses to estimate adult male and subadult survival, while also gaining valuable information on cause-specific bear mortality within the study system. 123 �To better understand the influence of urban environments on bear density and abundance, we are employing non-invasive genetic sampling (Woods et al. 1999, Mowat and Strobeck 2000) to compare these parameters between a bear population around the urban center of Durango and in a nearby “wildland” area. For each area we identified a 36 cell grid (576 km2) where each cell was 4 x 4 km in size, and within each cell we constructed 1 snare site. Snares consisted of a scented bait hanging high in a tree, surrounded barbed wire around a cluster of trees encircling the bait (wire was strung 50 cm above ground). When bears climb over or under the wire to investigate the bait, they leave a hair sample on the barbed wire. During summers 2011 through 2013, snares were deployed during the first 2 weeks of June, and we conducted 6 weekly sampling occasions thereafter. On each occasion, we randomly re-baited the snare with anise, berry, fish, maple or bacon scent, and collected hair samples from all barbs. Each hair sample was uniquely catalogued according to the site, date, occasion, and barb number. In summer 2012, we constructed 35 snares in the Durango grid and 34 snares in the wildland grid. The layout of the wildland grid had to be modified to account for closures associated with the Little Sand fire, which began burning on the San Juan National Forest on May 13th 2012. This modification can be easily accounted for in future analyses with spatially-explicit mark recapture statistics (Efford et al. 2009, Gardner et al. 2010) which increase flexibility with sampling designs. In fall 2012, all hair samples were sent to the laboratory at Wildlife Genetics International (Nelson, British Columbia, Canada) for genotyping; genetic results were returned at the end of July 2012. In summer 2013, we constructed 34 snare sites in the Durango grid and 35 sites in the wildland grid (Figure 1). Samples collected in 2013 will be sent to the laboratory this fall and results are expected in summer 2014. Objective 2: Testing a management strategy to reduce bear-human conflicts Given that the primary cause of black bear-human conflicts has been attributed to the availability of human foods to bears, it has been suggested that the most effective strategy to reduce conflicts is to reduce the availability of that food source (Peine 2001, Beckmann et al. 2004, Gore et al. 2005, Spencer et al. 2007). This strategy has had some success within national parks (Greenleaf et al. 2009), and anecdotally in some communities (Mammoth Lakes CA, Juneau AK, Whistler BC), but no research has ever scientifically tested the benefits of “cleaning up” a town. Given the high price to operationally “bearproof” a community, municipalities must have definitive evidence that such an effort would significantly decrease conflict activity before initiating major changes to waste storage and collection practices. As part of this project, we are implementing the first experimental test of wide-scale urban bearproofing for reducing bear-human conflicts. As part of the experiment we have designated 2 residential ‘treatment’ areas and 2 paired ‘control’ areas, consisting of a total of ~2,000 homes (Figure 2). In spring and early summer 2013 we deployed ~900 bear-resistant garbage containers within the treatment areas (approximately 100 homes already had these containers), such that all residents had a bear-resistant container. We also canvassed homes within the treatment areas, talking with residents about methods to bear-proof their properties, reminding them to lock their garbage containers, and asking that they remove bird feeders, outdoor pet food, and other bear attractants (no action occurred in control areas). Additionally, we increased enforcement of wildlife ordinances within treatment areas, providing official warnings and notifying City Code Enforcement when wildlife ordinances were violated. To track the effectiveness of these efforts in reducing bear-human conflicts we have planned to collect pre- and post-treatment data. For 2 years pre-treatment (summers 2011 and 2012), field technicians patrolled each street within proposed treatment/control areas on the day waste removal was scheduled to occur (when maximum human food was assumed to be available to bears). Technicians conducted patrols from ~05:30 - 06:30 AM and recorded locations where there was evidence that bears had obtained garbage or other human food sources. Monitoring occurred from early July through midSept, months that experience the highest numbers of bear-human conflicts in Durango (CPW unpublished 124 �data). During summer 2013 project personnel have been collecting the first year of post-treatment data (currently ongoing); post-treatment data will be collected for a minimum of 3 years. Each summer, in addition to collecting information on bears accessing human foods, we have quantified the “availability” of garbage to bears, by documenting the location and container type (wildlife-resistant or regular) of every garbage receptacle in the survey area accessible to bears the night prior to garbage pick-up. These data will allow us to track changes in the number of wildlife-resistant containers in the study area over the course of the experiment, and provide an estimate of the amount of human food available to bears in town. Once the experiment is complete, we will use pre- and posttreatment data collected during morning patrols and from calls received by CPW and the City of Durango to quantify the effectiveness of residential bear-proofing. Objective 3: Identifying public attitudes and behaviors related to bear-human encounters Wildlife management agencies must identify the biological factors driving increases in bearhuman conflicts, but they also must identify and incorporate human attitudes and perceptions about this issue into management strategies. This is particularly critical for black bears, as increasing bear-human conflicts around urban development have stimulated significant public interest and concern. It is also critical because bear-human conflicts typically arise over bear-use of human foods, prompting investigators to suggest that a critical component of reducing conflicts is managing human behavior (Beckmann et al. 2004, Gore et al. 2008, Baruch-Mordo et al. 2011). Thus, in conjunction with Stacy Lischka, Human Dimensions Specialist for CPW, we have initiated efforts to better understand human attitudes and behaviors in the context of our ecological data on bears. To assess data on human attitudes we are using public surveys to 1) quantify perceptions about bears, bear management, and bear-human encounters, and 2) explore motivations for compliance and noncompliance with wildlife ordinances designed to reduce bear-human conflicts. To meet those objectives, we developed a three part public mail survey to be conducted in conjunction with our urban bear-proofing experiment. Residents will be surveyed pre-, during, and post-implementation of the experiment, in treatment and control areas, as well as across a larger portion of the community. Surveys will be mailed to all residents within Durango city limits, and a subset of La Plata county residents within the study area. Survey responses will allow us to quantify current attitudes and perceptions about bear-human interactions, and how those perceptions change over time in association with a management effort such as wide-scale urban bear-proofing. The survey will also determine the number of residents that have had interactions with bears, the acceptability of management actions by CPW, and factors that promote or inhibit residents from complying with wildlife ordinances. The first (pre-treatment) public survey was implemented during winter 2012 (see Johnson et al. 2012 for details). The second survey will be conducted during fall 2013 or winter 2014. In addition to collecting data on human attitudes, we will also collect data on human behavior as part of an effort that was initiated this past year. Data collection will occur in conjunction with the treatment and control areas of the bear-proofing experiment starting summer 2013 (mid-July through midSept). Using a random stratified sampling design we will monitor human compliance with wildlife ordinances at residences throughout the conflict season. Houses will be surveyed on the morning of garbage pick-up (5:30 – 7:00 AM) to record whether those residences have secured their garbage the night prior (locked wildlife-resistant container or in a garage or shed that is not visible from the street) or have garbage available to bears. Compliance data will be analyzed in conjunction with survey information, spatial covariates, and bear activity to better understand how factors such as management actions and rates of wildlife-human interactions influence human behavior. The first year of data collected on human compliance will be summarized in the annual report for FY2013-14. 125 �RESULTS AND DISCUSSION Objective 1: Determining the influence of urban environments on bear behavior and demography Between May 20th 2012 and August 26th 2013, an additional 140 unique bears were marked during 327 bear captures; on the project to date there have been 232 different individuals marked during 435 captures. Information about these captures is described below for each discrete capture season: summer 2012, winter 2013, and summer 2013 (ongoing; Table 1). During summer 2012 we conducted 179 total bear captures; 86 captures were unique individuals and 93 were recaptures. Of the unique individuals captured, there were 37 females and 49 males (Table 1). We placed collars on 25 new adult females, and with those that had been previously collared in 2011, had 40 collars deployed by the end of August. The mean estimated age of bears ≥1 year-old on their initial capture date was 4.8 (5.7 for females and 4.2 for males), and mean weight was 73.2 kg (61.4 kg for females and 82.4 kg for males). The mean age of females that were newly collared in 2012, based on tooth cementum, was 8.6 years, with ages ranging from 3 to 24 years. In total, we placed traps at 90 different locations and conducted 1,114 trap nights. Capture success generally climbed each week until the second week of July, and remained high except for the second week of August (Figure 3). High trapping rates in 2012 were likely due to a combination of extra effort (we increased weekly trap nights from 5 nights/week to 7 nights/week and had a higher number of traps that were baited and set on a consistent basis) and a poor natural food year that brought additional bears into the urban-wildland interface around Durango. We visited the winter dens of 27 collared females between January and March 2013. Although we had 40 adult female bears collared at the end of summer 2012 there were 4 mortalities in fall: 1 female was legally harvested (B173), 1 was killed in a vehicle collision (B35), one was illegally shot (B134) and 1 died of unknown causes (B174). Additionally, in fall 2012, 9 GPS collars on collared females prematurely failed (B14, B18, B21, B24, B42, B55, B121, B122, and B144) and we could not locate their dens via VHF or GPS signals. Of the 27 adult females that we processed last winter, 13 did not have any cubs or yearlings, 6 had yearlings (6 yearlings in total, all bears had only one surviving yearling), and 8 had newborn cubs (14 cubs in total; 5 females and 9 males). Of those females with newborn cubs, 2 bears had only 1 cub and 6 bears had twins. We PIT and ear-tagged yearlings in the den, recorded information on weight, body size, body condition, and collected hair and blood samples. We also PIT tagged newborn cubs, and recorded their sex and weight. We found that reproductive success, measured as the number of cubs/adult female/year was 0.52 (SE = 0.16) for winter 2013, almost half of the reproductive rate observed in 2012 (0.95, SE = 0.24). Cub survival for 2013 (survival from newborn to 1 year) was ~40% (we do not have these data for 2012 as 2 sequential years are required for estimation). Between June 1st and August 26th 2012, we conducted summer captures with the goal of obtaining a sample of 40 GPS collared adult females (captures are currently ongoing). During that time there were 114 total captures; 37 were unique individuals and 77 were recaptures (Table 1). Of the unique individuals captures, 17 bears were females and 20 were males (there were also 9 cubs caught of unidentified sex; cubs were not immobilized or processed). The mean estimated age of bears ≥1 year-old on their initial capture date was 5.0 (4.9 for females and 5.0 for males) and the mean weight was 77.5 kg (56.9 kg for females and 92.1 kg for males). This summer, to date, 7 new adult females have been collared, and 5 females were recaptured that had previously slipped collars or had a malfunctioning collar. Given malfunctioning collars and 1 mortality (B65, vehicle collision), 35 females were collared as of August 26th, and trapping will continue through mid-September or until working GPS collars have been deployed. To date, traps have been placed in 93 different trap locations (30 on public land and 63 on private land) for 1,124 trap nights. Thus far, capture success has been fairly steady throughout the summer (Figure 3). 126 �To date, we have obtained >183,000 locations from GPS collars on 56 different female bears; 44 different bears collected location data in 2012 (Figure 4). We will start analyzing habitat-use data in the coming year and have generated spatial covariate data for elevation, slope, aspect, terrain ruggedness, distance to perennial water, distance to drainage, vegetation type, vegetation cover, distance to human structure, density of human structures, distance to paved road, and density of paved roads. The availability of natural mast foods was extremely limited in 2012, likely due to late freezes in June that destroyed berry and acorn flowers, and due to extreme summer drought. Of the few berries and acorns observed that summer, most were shriveled and dried up during the peak period of hyperphagia for bears (late summer/early fall). For example, gambel oak can be observed along all vegetation transects, but acorns were only observed on 5 of 15 transects, with limited production. Those acorns that did reach maturation were at their peak in mid-Sept. Chokecherry and serviceberry plants were observed on 13 of 15 transects, but no chokecherry fruits were seen (complete mast failure), and only a few shriveled serviceberries were seen. Pinyon pine nuts were fairly abundant on 2 transects, with mast production peaking in mid-Sept. Squaw apple fruits were only seen on 2 transects, and there were limited fruits that were mostly dried up. Hawthorne fruits were not found on any transect (complete mast failure). Between May 1st 2011 and November 1st 2012, a total of 59 bears were removed from the vicinity of Durango. Of those bears, 21 were killed in vehicle collisions, 18 were legally harvested, 10 were lethally removed due to nuisance behavior (breaking into houses, killing livestock, etc.), 3 died from nonharvest related gunshots, 2 were translocated due to conflicts with people, 1 was electrocuted, and 4 died from unknown causes. Of those mortalities and removals there were 17 adult females, 14 adult males, 11 subadult females, 9 subadult males, 6 cubs, and 2 bears of unknown sex/age class. Forty-one bears were unmarked and 17 were marked or collared for the research project. Additionally, 5 marked bears were reported killed outside the study area; 3 died from lethal conflict management and 2 died from vehicle collisions. In summer 2012, we collected 1,367 hair samples from the Durango and wildland grids; 586 samples from Durango and 781 samples from the wildland site. Over the 6 sampling occasions from 35 snares around Durango, we collected 92, 136, 59, 55, 142, and 102 samples, respectively. Over the 6 sampling occasions from 34 sites in the wildland grid, we collected 73, 135, 142, 118, 144, and 169 samples, respectively. We received the genetic results back from Wildlife Genetics International at the end of July 2013. Of the 1,367 hair samples submitted to the laboratory, good genotypes were obtained for 707 samples (52%). Of the remaining samples that did not produce a valid genotype, 363 (27%) did not contain enough genetic material, 274 (20%) failed during analyses for other reasons, and 23 (2%) samples were not black bear. Across the 707 valid samples there were 303 genotypes generated from the Durango grid and 404 generated from the wildland grid. In the Durango grid, 97 different individuals were detected during 138 “captures” (multiple hair samples from a single bear during 1 sampling occasion were considered 1 “capture”). Of those individuals, 71 were only detected in 1 sampling occasion and 26 were detected in >1 occasion (recaptures). The probability of detecting a bear within any single sampling occasion was 0.14, and across all sampling occasions was 0.58. In the wildland grid, 55 different individuals were detected during 71 “captures.” Of those individuals, 44 were only detected in 1 sampling occasion and 11 were detected in >1 occasion (recaptures). The probability of detecting a bear within any single sampling occasion was 0.09, and across all sampling occasions was 0.42. Detailed mark-recapture analyses of these data will be conducted in the future to estimate annual density and abundance at each site. In summer 2013, we collected 1,365 hair samples from the Durango and wildland grids; 680 samples from Durango and 685 samples from the wildland site. Over the 6 sampling occasions from 34 snares around Durango we collected 106, 151, 131, 62, 106, and 124 samples, respectively. Over the 6 127 �sampling occasions from 35 sites in the wildland grid we collected 112, 83, 141, 100, 126, and 123 samples, respectively. The number of samples/snare ranged from 0 to 121 in the Durango grid and from 1 to 78 in the wildland grid. Samples will be sent to Wildlife Genetics International this fall for genetic analysis. Objective 2: Testing management strategies to reduce bear-human conflicts During summer 2012 (July through mid-August) we collected a second year of pre-treatment data for the bear-proofing experiment. Within proposed treatment and control areas we observed 177 instances of bears accessing residential garbage during our morning patrols (Figure 2); observations peaked in early September. Of those garbage containers accessed by bears, 94% were regular containers or unsecured trash bags and 6% were wildlife-resistant containers. Bears accessed human food from wildlife-resistant containers when they were not properly latched. In quantifying the availability of garbage to bears, we recorded the location and container type of 1,530 garbage cans in proposed treatment and control areas. Of those containers, 86% were regular (non-wildlife resistant) and 14% were wildlife-resistant. This demonstrates the limited residential bear-proofing that currently exists in Durango, and the relevance of conducting an experimental test of wide-scale urban bear-proofing in this community. This past year, we worked on the logistics of implementing the urban bear-proofing experiment. Final funds were secured through CPW, the Summerlee Foundation, and the International Bear Association to purchase the remaining containers needed for the study. Wildlife-resistant containers were acquired through Solid Waste Systems (Parker, CO), a company that manufactures containers certified by the Living with Wildlife Foundation. Containers were delivered to Durango, entered into the City of Durango’s Solid Waste Program database, and distributed by the City in spring and early summer 2013 to residences within treatment areas. Starting in mid-July 2013, we initiated the first year of post-treatment conflict monitoring. Data collection is currently ongoing, but as of August 26th we had recorded 153 incidences of bears accessing residential garbage; 75 conflicts in the treatment areas and 78 conflicts in control areas. Of those conflicts, 71% occurred with regular garbage containers or unsecured trash, and 29% occurred with wildliferesistant containers. In quantifying the availability of garbage to bears, we recorded the location and container type of 1,678 garbage cans in treatment and control areas. Within the northern control area 72% of containers were regular and 28% were wildlife-resistant, in the southern control area 91% were regular and 9% were wildlife-resistant, in the northern treatment area 11% of containers were regular (residents that refused a bear-resistant container or have kept additional regular containers on their property) and 89% were wildlife-resistant, and in the southern treatment area 27% were regular and 73% were wildliferesistant. SUMMARY AND FUTURE PLANS During FY12-13 we successfully coordinated field logistics and conducted several aspects of data collection (trapping and collaring bears, tracking human-related bear mortalities, assessing summer/fall forage availability, implementing DNA hair-snare surveys, and monitoring garbage-related bear-human conflicts). We will continue these field activities through 2015, and begin data analyses as field data are completed. In addressing the objectives of this project we hope to better understand the influence of urban environments on bear populations, elucidate the relationship between bear-human conflicts and bear behavior and population trends, develop tools to promote the sustainable management of bears in Colorado, and ultimately, identify solutions for reducing bear-human conflicts in urban environments. 128 �LITERATURE CITED Baruch-Mordo, S., S.W. Breck, K.R. Wilson, and J. Broderick. 2011. The carrot or the stick? Evaluation of education and enforcement as management tools for human-wildlife conflicts. Plos One 6: e15681. Baruch-Mordo, S., S.W. Breck, K.R. Wilson, and D.M. Theobald. 2008. Spatiotemporal distribution of black bear-human conflicts in Colorado, USA. Journal of Wildlife Management 72:1853-1862. Baruch-Mordo, S., K.R. Wilson, D. Lewis, J. Broderick, J. Mao, and S.W. Breck. 2010. Roaring Fork Valley urban black bear ecology study: progress report to the Colorado Division of Wildlife. Contact Sharon Baruch-Mordo for copy. Email: sharonb_m@yahoo.com Beck, T.D.I. 1991. Black bears of west-central Colorado. Technical Publication No. 39, Colorado Division of Wildlife, Colorado. Beckmann, J.P., and J. Berger. 2003a. Using black bears to test ideal-free distribution models experimentally. Journal of Mammalogy 84:594-606. Beckmann, J.P., and J. Berger. 2003b. Rapid ecological and behavioural changes in carnivores: the response of black bears (Ursus americanus) to altered food. Journal of Zoology 261:207-212. Beckmann, J.P., L. Karasin, C. Costello, S. Matthews, and Zoë Smith. 2008. Coexisting with black bears: perspectives from four case studies across North America. Wildlife Conservation Society Working Paper No 33. Beckmann, J.P., C.W. Lackey, and J. Berger. 2004. Evaluation of deterrent techniques and dogs to alter behavior of “nuisance” black bears. Wildlife Society Bulletin 32: 1141-1146. Beston, J.A. 2011. Variation in life history and demography of the American black bear. Journal of Wildlife Management 75:1588-1596. Caswell, H. 2001. Matrix population models: construction, analysis, and interpretation. Second Edition, Sinauer Associates, Sunderland, Massachussetts. Efford, M.G., D.K. Dawson, and D.L. Borchers. 2009. Population density estimated from locations of individuals on a passive dector array. Ecology 90:2676-2682. Garshelis, D.L., and H. Hristienko. 2006. State and provincial estimates of American black bear numbers versus agency assessments of population trend. Ursus 17:1-7. Gardner, B., J.A. Royle, M.T. Wegan, R.E. Rainbolt, P.D. Curtis. 2010. Estimating black bear density using DNA data from hair snares. Journal of Wildlife Management 74:318-325. Gore, M.L., W.F. Siemer, J.E. Shanahan, D. Schuefele, and D.J. Decker. 2005. Effects on risk perception of media coverage of a black bear-related human fatality. Wildlife Society Bulletin 33:507-516. Gore, M.L., B.A. Knuth, C.W. Scherer, and P.D. Curtis. 2008. Evaluating a conservation investment designed to reduce human-wildlife conflicts. Conservation Letters 1:136–145. Greenleaf, S.S., S.M. Matthews, R.G. Wright, J.J. Beecham, and H.M. Leithead. 2009. Food habitat of American black bears as a metric for direct management of human-bear conflict in Yosemite Valley, Yosemite National Park, California. Ursus 20:94-101. Hostetler, J.A., J.W. McCown, E.P. Garrison, A.M Neils, M.A. Barrett, M.E. Sunquist, S.L. Simek, and M.K. Oli. 2009. Demographic consequences of anthropogenic influences: Florida black bears in north-central Florida. Biological Conservation 142:2456-2463. Hristienko, H., and J.E. McDonald Jr. 2007. Going into the 21st century: a perspective on trends and controversies in the management of the American black bear. Ursus 18:72-88. Johnson, H.E, C.J. Bishop, M.W. Alldredge, J. Brodrick, J. Apker, S. Breck, K. Wilson, and J. Beckmann. 2011. Black bear exploitation of urban environments: finding management solutions and assessing regional population effects. Research Proposal, Colorado Division of Parks and Wildlife, Fort Collins, USA. Johnson, H.E. 2012. Black bear exploitation of urban environments: finding management solutions and assessing regional population effects. Federal Aid Annual Report, Colorado Division of Parks and Wildlife, Fort Collins, USA. 129 �Manly, B.F.J., L.L. McDonald, D.L. Thomas, T.L. McDonald, and W.P. Erickson. 2002. Resource selection by animals; statistical design and analysis for field studies. Second Edition. Kluwer Academic Publishers, Boston, Massachussetts. McLoughlin, P.D., D.W. Morris, D. Fortin, E. Vander Wal, and A.L. Contasti. 2010. Considering ecological dynamics in resource selection functions. Journal of Animal Ecology 79:4-12. Mitchell, M.S., L.B. Pacifici, J.B. Grand, and R.A. Powell. 2009. Contributions of vital rates to growth of a protected population of American black bears. Ursus 20:77-84. Morales, J.M., P.R. Moorcroft, J. Matthiopoulos, J.L. Frair, J.G. Kie, R.A. Powell, E.H. Merrill, and D.T. Haydon. 2010. Building the bridge between animal movement and population dynamics. Philosophical Transactions of the Royal Society Series B 365:2289-2301. Mowat, G., and C. Strobeck. 2000. Estimating population size of grizzly bears using hair capture, DNA profiling, and mark-recapture analysis. Journal of Wildlife Management 64:183-193. Noyce, K.V., and P.L. Coy. 1989. Abundance and productivity of bear food species in different forest types of northcentral Minnesota. International Conference on Bear Research and Management 8:169-181. Peine, J.D. 2001. Nuisance bears in communities: Strategies to reduce conflicts. Human Dimensions of Wildlife 6:223-237. Spencer, R.D., R.A. Beausoleil, and D.A. Martorello. 2007. How agencies respond to human–bear conflicts: a survey of wildlife species in North America. Ursus 18: 217–229. Theobald, D.M., and W.H. Romme. 2007. Expansion of the US wildland-urban interface. Landscape and Urban Planning 83:340-354. Woods, J.G., D. Paetkau, D. Lewis, B.N. McLellan, M. Proctor, and C. Strobeck. 1999. Genetic tagging of free-ranging black and brown bears. Wildlife Society Bulletin 27:616-627. Zack, C.S., B.T. Milne, and W.C. Dunn. 2003. Southern oscillation index as an indicator of encounters between humans and black bears in New Mexico. Wildlife Society Bulletin 31:517-520. Prepared by _______________________________________ Heather E. Johnson, Wildlife Researcher 130 �Table 1. Capture information for black bears that have been marked in the vicinity of Durango, CO since May 2012 (collared adult females are identified with an “*”). Only information from the initial capture of each individual is shown (no recaptures). Bear ID B120 B121* B122* B123 B124* B125* B126 B127* B128* B129* B130 B131 B132 B133* B134* B135 B136 B137 B138 B139 B140 B141* B142 B143* B144* B145* B146 B147 B148 B149 B150 B151 B152* B153 B154 B155 B156 B157 B158 Capture Date 5/27/2012 5/29/2012 5/30/2012 6/5/2012 6/6/2012 6/8/2012 6/8/2012 6/10/2012 6/11/2012 6/14/2012 6/22/2012 6/23/2012 6/28/2012 6/29/2012 6/30/2012 6/30/2012 7/1/2012 7/5/2012 7/5/2012 7/5/2012 7/5/2012 7/5/2012 7/6/2012 7/6/2012 7/7/2012 7/7/2012 7/7/2012 7/10/2012 7/11/2012 7/15/2012 7/16/2012 7/26/2012 7/17/2012 7/17/2012 7/17/2012 7/19/2012 7/19/2012 7/19/2012 7/20/2012 UTM Easting 254732 251670 249059 240102 249158 244618 251670 239005 239005 254576 250152 765047 765047 765932 765932 252014 765047 249059 254997 238245 763921 765132 254997 241210 238245 763921 254739 241334 255983 244618 241334 243888 241210 249059 253439 241334 252621 248417 252546 UTM Northing 4133249 4132767 4132998 4128939 4127065 4132132 4132767 4134459 4134159 4135043 4127691 4131635 4131635 4127651 4127651 4133509 4131635 4132998 4135825 4131204 4132873 4132506 4135825 4137115 4131204 4132873 4133234 4138018 4135921 4132132 4138018 4129546 4137114 4132998 4134693 4138018 4130532 4144294 4134789 131 Sex F F F M F F F F F F M M M F F M F M F M M F M F F F M M M M M M F M M F F M M Age 1 4 5 2 7 8 1 10 8 6 1 6 1 3 8 1 2 2 2 1 3 3 3 3 9 6 5 1 3 1 2 3 8 2 3 2 1 6 3 Kg 20.9 76.2 66.2 48.1 80.7 98.9 15.9 58.1 56.2 54.4 12.7 111.6 20.4 49.0 90.7 20.9 46.3 26.8 30.8 30.4 67.1 55.3 37.2 45.4 72.6 70.3 110.2 43.5 73.9 45.4 53.5 60.3 99.8 49.9 63.5 30.8 26.3 136.1 50.8 �B159 B160 B161* B162 B163 B164 B165* B166 B167* B168 B169 B170 B171 B172 B173* B174* B175* B176 B177 B178 B179 B180* B181* B182 B190 B191 B192 B193 B194 B195 B196 B197 B198* B199 B200 B201 B202 B203 B204 B205* B206 B207 B212 B213* B214 7/21/2012 7/24/2012 7/25/2012 7/25/2012 7/26/2012 7/28/2012 7/29/2012 7/29/2012 7/31/2012 7/31/2012 7/31/2012 8/1/2012 8/2/2012 8/2/2012 8/3/2012 8/3/2012 8/3/2012 8/3/2012 8/4/2012 8/4/2012 8/5/2012 8/5/2012 8/5/2012 8/8/2012 8/9/2012 8/11/2012 8/11/2012 8/12/2012 8/12/2012 8/13/2012 8/14/2012 8/16/2012 8/16/2012 8/17/2012 8/17/2012 8/18/2012 8/19/2012 8/21/2012 8/21/2012 8/22/2012 8/22/2012 8/23/2012 8/24/2012 8/24/2012 8/24/2012 242236 249059 242546 249059 243954 242611 251815 252621 248578 253439 249059 249059 248192 248578 248578 253341 254916 252621 248578 249059 248578 248939 247127 259049 245293 249059 245293 243652 243218 259049 249059 242772 247295 244208 245293 244600 249059 244174 243652 243652 247295 249059 243274 244174 249059 4127920 4132998 4134789 4132998 4134875 4133863 4133706 4130532 4139143 4134693 4132998 4132998 4137051 4139143 4139143 4128740 4128609 4130532 4139143 4132988 4139143 4141533 4138557 4132998 4128959 4132998 4128959 4129360 4128712 4132998 4132998 4129388 4132138 4129996 4128959 4132218 4132998 4130027 4129360 4129360 4132138 4132998 4129124 4130027 4132998 132 F M F M F M F M F M F M M F F F F M M M M F F M M M M M M M F M F M M M M M M F M M M F M 2 2 5 10 1 6 12 8 18 4 2 4 2 2 5 3 10 2 7 2 1 3 3 2 12 4 9 5 5 6 2 1 10 5 5 5 3 1 3 4 2 3 2 6 2 39.9 64.0 76.2 108.0 37.6 117.0 85.3 121.6 58.1 87.5 44.5 119.3 26.8 28.1 73.9 43.5 76.2 54.4 93.9 35.4 35.4 71.7 58.1 70.8 153.3 60.3 148.8 87.5 137.0 151.0 42.2 34.0 75.3 140.6 91.6 99.8 47.6 24.5 71.7 49.9 46.3 59.0 45.4 61.2 63.5 �B215 B216 B244 B245* B246 B259 B247 B248 B249 B250 B251 B252 B253 B254 B255 B256 B257 B258 B294 B295 B260 B261* B262 B263 B265 B266 B267 B268 B269 B274 B275 B276 B277 B278 B280 B281* B282 B283 B284 B285* B287 B289 B290 B291* B292 8/24/2012 8/28/2012 1/25/2013 2/14/2013 2/14/2013 6/3/2013 3/10/2013 3/11/2013 3/11/2013 3/14/2013 3/15/2013 3/15/2013 3/16/2013 3/16/2013 3/19/2013 3/19/2013 3/27/2013 3/27/2013 4/19/2013 4/19/2013 6/4/2013 6/7/2013 6/10/2013 6/11/2013 6/14/2013 6/17/2013 6/19/2013 6/21/2013 6/21/2013 6/26/2013 6/27/2013 6/29/2013 6/30/2013 7/1/2013 7/8/2013 7/11/2013 7/16/2013 7/16/2013 7/19/2013 7/19/2013 7/23/2013 7/25/2013 8/1/2013 8/2/2013 8/4/2013 247295 249059 240578 243288 243288 251514 243920 247275 247275 245069 255660 255660 240747 240747 255951 255951 240819 240819 268453 268453 253231 253231 251343 254584 251343 251933 249872 251817 249153 256824 256937 237708 238206 256763 240340 255385 244631 242022 247443 254750 248532 764348 246591 237566 246591 4132138 4132998 4137131 4123754 4123754 4137313 4135766 4126961 4126961 4137542 4131489 4131489 4132514 4132514 4141487 4141487 4154867 4154867 4207298 4207298 4138879 4138879 4134446 4134994 4134446 4137246 4130099 4131555 4132855 4134340 4134617 4130726 4130573 4134317 4131577 4133334 4132166 4127361 4137388 4133273 4139266 4132592 4135689 4124276 4135689 133 M M F F F M M M M F M M F F M M M F F M M F M M F M F F M M F F M M F F M M F F M M M F M 5 3 1 15 1 1 Cub Cub Cub Cub Cub Cub Cub Cub Cub Cub Cub Cub Cub Cub 6 8 1 6 1 8 2 2 4 6 2 2 10 3 1 13 3 2 0 8 3 3 4 4 3 127.0 59.9 15.0 83.0 26.3 32.7 2.4 2.6 4.3 2.4 1.9 2.2 2.0 2.0 1.5 1.4 2.7 2.7 2.4 2.4 102.1 78.0 30.8 83.9 38.6 125.6 30.8 31.3 54.4 116.6 34.9 57.2 181.4 79.8 25.9 63.5 73.0 33.6 10.4 66.2 75.3 88.5 93.4 64.4 56.7 �B293* B296 B297 B298* B299* B300 B301 B302 B303* B304 8/6/2013 8/10/2013 8/10/2013 8/14/2013 8/16/2013 8/17/2013 8/20/2013 8/23/2013 8/23/2013 8/23/2013 245913 245913 246383 245848 245848 243934 243934 248263 240787 243934 4139623 4139623 4142011 4141980 4141980 4134857 4134793 4136448 4130376 4134857 134 F M M F F M F M F M 3 2 4 5 3 10 1 8 7 1 49.4 45.8 99.3 51.7 61.7 122.0 33.1 133.8 93.0 42.2 �Figure 1. Locations of the 2013 hair snare sites (red dots) for the Durango and Wildland genetic sampling grids. -- 0 1.5 3 N 6 9 + .• ~ ! 135 �Figure 2. Location of garbage-related conflicts observed during morning patrols and garbage containers (by type) available to bears during summer 2012. Treatment and control areas for the bear-proofing experiment (implemented in summer 2013) are also shown. Garbage Conflicts and Availability 2012 Summary * 2012 Conflicts ., 2012 Garbage Availabiity o Regular can • Wildlrte-Resistant Can Experimental Zone 8 0 Treatment Cootrol 0. 125 0.25 0.5 Mi5es s OOpens:reetMap(.nl) comJ10UICG. cc-ev.-s,,.. 136 �Figure 3. Number of black bear captures by week from May 15th through September 15th for bear captures during the 2011 through 2013 summer trapping seasons (2013 is currently ongoing). Number of Bear Captures 25 2011 2012 2013 20 15 10 5 0 1 3 5 7 9 11 13 Week (May 15th through Sept 15th) 137 15 17 �Figure 4. GPS collar locations from 44 adult female black bears from April 2012 through December 2012 in the vicinity of Durango, CO (different colored clusters of points represent different individual bears): A) an overview of all locations and B) locations around the town of Durango. B A ·•• : ~ ~-. i ;,, ·. ·...... ·--~~1,:.· :,,a u1 1M rn + -- 0 10 • \ ui. ln~ 1;2 n N 20 R•-rvahon ,. 2588m 2:,o5m ,. 2.Je'- .A'. - ~ .... ,,J" Du lce ~· 2 ~~ ~ ·· A >l ec 5 40 -, 60 g1 80 Ki lometers ... ' ·-,. . ·-· 14.lO .••"':•,... -- 0 J a:;; aolla Ap.,h• lrn::l ,;2 r, Hl111Gt1Lakt. 2.5 138 5 Sot.1 th •r n Lit• lnd111-n ~;.,, rv iu>o n �Colorado Division of Parks and Wildlife July 2013 – June 2014 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Colorado 3430 3003 3 Federal Aid Project No. W-204-R1 : : : : : Division of Wildlife Mammals Research Predatory Mammal Conservation Black bear exploitation of urban environments: finding management solutions and assessing regional population effects Period Covered: July 1, 2013 – June 30, 2014 H.E. Johnson, S.A. Lischka, J. Broderick, J. Apker, S. Breck, J. Beckmann, K. Wilson, and P. Dorsey. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Across the country conflicts among people and black bears are increasing in frequency and severity, and have become a high priority wildlife management issue. Whether increases in conflicts reflect recent changes in bear population trends or just bear behavioral shifts to anthropogenic food resources, is largely unknown, with key implications for bear management. This issue has generated a pressing need for bear research in Colorado and has resulted in a unique collaboration that builds on the resources and abilities of personnel from 4 entities: Colorado Parks and Wildlife (CPW), the USDA National Wildlife Research Center, Wildlife Conservation Society and Colorado State University. Collectively, we have designed and implemented a study on black bears that 1) determines the influence of urban environments on bear behavior and demography, 2) tests a management strategy for reducing bear-human conflicts, 3) examines public attitudes and behaviors related to bear-human interactions, and 4) develops population and habitat models to support the sustainable monitoring and management of bears in Colorado. This project was initiated in FY2010-11; during this past fiscal year we have primarily focused on collecting field data in the vicinity of Durango, Colorado. Specifically, we worked with collaborators and stakeholders on research logistics, trapped and marked black bears, collected GPS collar location data on bears along the urban-wildland interface, monitored bear demographic rates (adult female survival, adult female fecundity and cub survival) through telemetry and winter den visits, collected data on the availability of late summer/fall mast, tracked human-related bear mortalities and removals from the study area, performed non-invasive genetic mark-recapture surveys, deployed ~150 bear-resistant containers for an experiment on the effectiveness of urban-bear-proofing, obtained data on garbagerelated bear-human conflicts, monitored resident use of project-supplied bear-resistant garbage containers, and conducted a survey assessing resident attributes about bears and bear-human interactions. Additionally, we conducted an initial analysis of bear selection of human development. Information from this study will provide solutions for sustainably managing black bears outside urban environments, while reducing bear-human conflicts within urban environments; knowledge that is critical for wildlife managers in Colorado and across the country. 1 �PROJECT OBJECTIVES FY13-14 1. Work with personnel from CPW Area 15, CPW Southwest Region, City of Durango, La Plata County, US Forest Service (Columbine and Pagosa Ranger Districts), Bureau of Land Management (BLM; Tres Rio Field Office), Southern Ute Tribe, and private landowners on field research logistics. 2. Trap and collar adult female black bears in the vicinity of Durango to collect data on bear behavior and demography. 3. Track bear locations and survival via global position system (GPS) collar locations. 4. Monitor bear fecundity and cub survival through winter den investigations of collared adult female bears. 5. Obtain data on summer/fall natural food availability for bears based on the abundance of gambel oak, serviceberry, chokecherry, hawthorne, pinyon pine and squaw apple. 6. Track human-related bear mortalities and removals around Durango from lethal conflict managment, vehicle collisions, harvest, and translocations. 7. Perform non-invasive genetic mark-recapture surveys to estimate bear density and population size around Durango (urban site) and in the Piedra watershed (wildland site). 8. Deploy ~150 bear-resistant garbage containers to “clean up” treatment areas for an experiment on the effectiveness of wide-scale urban bear-proofing for reducing bear-human conflicts. 9. Collect data on the frequency of bears accessing garbage in treatment and control areas of the bear-proofing experiment. 10. Estimate appropriate use of project-supplied bear-resistant garbage containers by residents in treatment areas. 11. Assess attitudes about bears and bear-human interactions among residents of Durango and La Plata county. INTRODUCTION In Colorado and across the country, conflicts among people and black bears (Ursus americanus) appear to be increasing in number and severity (Hristienko and McDonald 2007, Baruch-Mordo et al. 2008, CPW unpublished data). Bear-human conflicts can result in public safety concerns, property damage, bear mortality (i.e., euthanasia), and high management costs, and thus, have become a critical wildlife management issue. While wildlife agencies have used a variety of tools to try to minimize bearhuman conflicts (i.e., education, aversive conditioning of bears, and increased harvest), conflict rates have continued to rise. Whether increases in bear-human conflicts reflect recent changes in the bear population or just behavioral shifts to anthropogenic food resources, is largely unknown, as bear population parameters have been exceedingly difficult to estimate (Garshelis and Hristienko 2006). Without a thorough understanding of the relationship between conflict rates and bear behavior and population dynamics, it has been difficult for wildlife agencies to successfully reduce conflicts through bear management. 2 �While there is uncertainty about how to reduce bear-human conflicts, two key factors thought to exacerbate this problem are expanding human development and climatic variation. Colorado has had one of the highest rates of exurban development in the nation (Theobald and Romme 2007), and this development has resulted in additional human food on the landscape in the form of garbage, fruit trees, livestock, birdfeeders, etc. The availability of human food to bears has been identified as the primary cause of bear-human conflicts (Spencer et al. 2007, Beckmann et al. 2008, Greenleaf et al. 2009), as bears are opportunistic foragers that will readily take advantage of additional resources. Bear-use of human food not only increases interactions between bears and people but has been found to alter bear activity patterns, foraging behavior, movement rates, and even survival and reproductive rates (Beckmann and Berger 2003a, Beckmann and Berger 2003b, Hostetler et al. 2009), having the potential to significantly influence both bear behavior and demography. This phenomenon is further complicated by variation in annual weather patterns, as bear-use of human development appears to increase when natural foods are in short supply (Zack et al. 2003, Baruch-Mordo et al. 2010). Because bears predominately consume vegetation, recent patterns of drought in Colorado have caused natural food failures for bears in some years. As a result, bears may be increasing their reliance on human foods, with associated behavioral and demographic impacts. While the effects of urbanization and climate have critical implications for modifying bear-habitat relationships, they also have critical implications for increasing rates of bearhuman conflicts. To develop successful strategies to reduce conflicts while maintaining viable bear populations, wildlife agencies must understand how factors such as climate, natural food availability, human food ability, and management influence the behavior and dynamics of bear populations. To address these questions, Colorado Parks and Wildlife has partnered with the USDA National Wildlife Research Center, Wildlife Conservation Society and Colorado State University. Collectively, we initiated a project in FY10-11 to 1) determine the influence of urban environments on bear behavior and demography, 2) test a management strategy for reducing bear-human conflicts, 3) examine public attitudes and behaviors related to bear-human interactions, and 4) develop population and habitat models to support the sustainable monitoring and management of bears in Colorado (Johnson et al. 2011). This information should provide solutions for sustainably managing black bears outside urban environments, while reducing bear-human conflicts within urban environments; knowledge that is critical for wildlife managers in Colorado and across the west. During FY13-14, we worked with collaborators and stakeholders on research logistics, trapped and marked black bears, collected GPS collar data on bear locations, monitored bear demographic rates (adult female survival, adult female fecundity and cub survival) through telemetry and winter den visits, collected data on the availability of late summer/fall mast, tracked human-related bear mortalities and removals from the study area, performed non-invasive genetic mark-recapture surveys, deployed ~150 bear-resistant containers for an experiment on the effectiveness of urban-bear-proofing, obtained data on garbage-related bear-human conflicts, monitored resident use of project-supplied bear-resistant garbage containers, and conducted a survey assessing resident attributes about bears and bear-human interactions. Additionally, we performed an initial analysis of bear selection of human development using GPS collar data. Our efforts focused largely on collecting field data to meet research objectives 1-3, information which will eventually be used to address objective 4. We report general summary information from field activities over the past year; detailed analyses of field data will occur in future years. STUDY AREA To meet study objectives, a combination of site-specific field data and statewide data will be required. Site-specific field data is being collected in the vicinity of Durango, and is the focus of this progress report. The town of Durango contains ~17,000 people (within city limits) and sits at 1,985 m along the Animas river valley. The town is surrounded by mountainous terrain ranging in elevation from ~1,930 to ~3,600 m, and is generally characterized by mild winters and warm summers that experience 3 �monsoon rains. Vegetation in the region is dominated by ponderosa pine, oak, pinyon pine, juniper, aspen, mountain shrubs, and agriculture. Key forage species for black bears include gambel oak (Quercus gambelii), chokecherry (Padus virginiana), serviceberry (Amelanchier alnifolia), hawthorne (Crataegus spp), wild crabapple (Peraphyllum ramosissimum) and pinyon pine (Pinus edulis). Durango is predominately surrounded by public land managed by the San Juan National Forest, BLM, CPW, La Plata County and the City of Durango. The vicinity of Durango is considered high quality bear habitat, and the town has consistently experienced high rates of bear-human conflicts (Baruch-Mordo et al. 2008, CPW unpublished data). METHODS Objective 1: Determining the influence of urban environments on bear behavior and demography To sustainably manage bears in the face of a growing human population and changing landscape conditions, it is critical to elucidate the drivers and dynamics of bear populations. Of those factors that influence bear populations, the expansion of human development is the least understood, most contentious, and has the greatest potential to elicit major population change. To elucidate the influence of human development on bear behavior and demography, we are collecting a suite of data types including locations from collared bears on the urban-wildland interface, survival and reproductive rates of those bears in conjunction with their habitat-use patterns, information on annual summer/fall mast production, and genetic data to estimate bear density in urban and wildland habitats using mark-recapture methods. We briefly describe data collection methods for this portion of the study below; detailed information is available in Johnson et al. (2011). Collaring and Marking Bears – To assess bear behavior and demographic rates with respect to human development, we are capturing and collaring adult female bears. We are specifically targeting adult females as they represent the reproductive segment of the population and allow us to obtain information on multiple key vital rates that drive population growth. For example, in addition to being able to track adult female survival, the vital rate with the highest elasticity (Beston 2011), we can use collared females to track fecundity and cub survival, vital rates that are often associated with variation in bear population trends (Mitchell et al. 2009, Beston 2011). We have targeted summer trapping efforts within ~10 km of Durango to collar a cohort of bears that experience similar natural food availability, have anthropogenic food resources readily available, and encompass a range of behaviors and habitat-use patterns relative to the urban-wildland interface. Bears are trapped with box traps, which are baited with fish, fruit, human foods (at urban locations) and manufactured scents. Traps are set in the evening and checked the following morning. Adult female bears are fitted with a GPS collar (manufactured by Vectronics), and a tooth (first pre-molar) is pulled for age verification. GPS collars record bear locations every hour, and upload a real-time location to a central database via a satellite system every 6 hours. Although trapping efforts are focused on adult females, all bears that are trapped (i.e., males, subadults, yearlings) are uniquely marked with a PIT and ear-tag and are weighed, measured, and sampled for blood and hair. Evaluating Bear Movement and Habitat-Use Relative to the Urban-Wildland Interface – To examine movement and habitat-use patterns of bears along the urban-wildland interface, we are using GPS location data from collared females. Hourly GPS data are downloaded from the collars in the field on a biannual basis (fall and winter). Locations are being used to assess the influence of factors such as natural food availability, human food availability, weather, habitat covariates, and individual bear attributes (i.e., age, reproductive status) on bear movement and resource selection patterns (Manly et al. 2002, McLoughlin et al. 2010, Morales et al. 2010). For spatial covariate data, we have generated rasters representing elevation, aspect, slope and terrain ruggedness using digital elevation models. We also created rasters depicting distances to drainages and perennial water using the National Hydrology Dataset, 4 �and have estimated the proportion of different vegetation types using the USFS LandFire dataset (http://www.landfire.gov/vegetation.php). We derived rasters depicting human structure and road densities using data from La Plata county and CPW. Weather information will be acquired from local weather stations and PRISM datasets (www.prism.oregonstate.edu/). While most habitat and human development information can be extracted from existing spatial data sources, there is no existing data layer that tracks annual variation in late summer/fall hard and soft mast for bears. The abundance of berry and nut resources for bears is known to be highly variable, depending on annual trends in precipitation and temperature (Noyce and Coy 1989). To account for variation in the availability of natural forage for bears around Durango, we conducted bimonthly mast surveys. Surveys were performed from late July through mid-September, when berries and nuts should reach peak maturation and bears are experiencing hyperphagia prior to hibernation. Key mast species for bears around Durango are gambel oak, chokecherry, serviceberry, hawthorne, wild crabapple, and pinyon pine (Beck 1991, Tom Beck, personal communication). We randomly selected 16 transects on public lands to evaluate bear mast availability. Each transect was 1 km in length and was situated along an existing trail or stream drainage. For each transect, field technicians recorded the phenological stage and the percentage of plants of each species that exhibited mast in different abundance categories (mast failure, <25% of plants with mast, 25 – 50% of plants with mast, etc). During FY13-14 we conducted an initial analysis of bear selection for human development around Durango, examining temporal patterns of selection both within and across years. We used GPS collar data collected in 2011, a good natural food year for bears, and in 2012, a poor natural food year. We used mixed-effects resource selection models (RSFs) following a use-availability design (Manly et al. 2002), using random intercepts account for differences in sample sizes among individuals and autocorrelation within animal datasets (Gillies et al. 2006). “Used” locations were evaluated on an animal-year-specific basis, collected from May – Oct and “available” locations were randomly generated from animal-specific 95% minimum convex polygons. We used 80% of the animal-year datasets for developing models and withheld 20% for model validation. We first ran a “base” model that accounted for topography, vegetation, and distance-to-drainage, which are covariates consistently associated with black bear habitat (Clevenger et al. 2002; Carter et al. 2010; Sadeghpour and Ginnett 2011). We then compared the base model to a series of models that included density of human development (HD), based on point data of human structures in La Plata county. We evaluated the following models: base + HD, base + HD * Food Year (temporal variation in selection for HD between years), base + HD * Week (temporal variation in selection for HD within the active bear year) and base + HD * Food Year * Week (both across and within year variation in selection for development; models included all relevant main effects and interactions). We fit models using maximum likelihood estimation and used minimum AIC scores to assess the relative support for models with different fixed effects (Burnham and Anderson 2002). We assessed the predictive power of our top model with cross-validation using hold-out data (Boyce et al. 2002). Estimating Demographic Rates – To assess the influence of human development on bear demographic rates we have been collecting the following data types: 1) survival and reproduction of collared adult female bears, 2) cub survival monitored during annual winter den checks of collared females, 3) mortalities and removals of marked and unmarked bears in the vicinity of Durango, and 4) non-invasive genetic surveys that estimate density and abundance of bears around urban and wildland sites. Collared female bears allow us to estimate annual adult female survival, fecundity (number of cubs born/adult female) and cub survival (survival from newborn cub to yearling); parameters we have monitored since summer 2011 and which we will continue to monitor through winter 2017. We use realtime GPS collar locations to assess adult female survival, investigating mortalities and slipped collars when GPS locations are stationary during multiple fixes. Fecundity and cub survival are monitored from 5 �winter den checks of collared females. Numbers of newborn cubs provide information on fecundity, while consecutive annual den checks of collared females allow us to estimate cub survival. Because yearlings hibernate with their mothers, we can observe the number of cubs alive in the den in year t that survived their first year of life to t+1. Adult female survival, fecundity and cub survival will be used in matrix projection models to assess population performance (Caswell 2001), particularly in relation to use of human development. In addition to tracking survival and reproduction of collared bears, we are also tracking survival and cause-specific mortality of marked (i.e., males, subadults) and unmarked bears in the study area. All bears that are trapped are marked with an ear-tag and PIT tag, unique identifiers that we are using to collect data on human-related bear mortalities and removals. Mortalities and removals primarily occur from translocations, vehicle collisions, conflict-related euthanasia and hunter harvest. For all bears removed from the study area we collect a hair and tooth sample and recorded the date, mortality/removal cause, location, bear age, sex, weight, and morphological measurements. We will use mark-recapture and recovery data to estimate adult male and subadult survival, while also gaining valuable information on cause-specific bear mortality within the study system. To better understand the influence of urban environments on bear density and abundance, we are employing non-invasive genetic sampling (Woods et al. 1999, Mowat and Strobeck 2000) to compare these parameters between the bear population around Durango and for a nearby “wildland” area. For each area we identified a 36 cell grid (576 km2) where each cell was 4 x 4 km in size. We constructed and monitored 1 snare site within each cell. Snares consisted of a scented bait hanging high in a tree, surrounded by barbed wire around a cluster of trees encircling the bait (wire was strung 50 cm above ground). When bears climb over or under the wire to investigate the bait, they leave a hair sample on the barbed wire. During summers 2011 through 2014, we deployed snares during the first 2 weeks of June, and conducted 6 weekly sampling occasions thereafter. On each occasion, we randomly re-baited the snare with a scent (anise, berry, fish, maple or bacon), and collected hair samples from all barbs. Each hair sample was uniquely catalogued according to the site, date, occasion, and barb number. In summer 2013, we constructed and monitored 34 snares in the Durango grid and 35 snares in the wildland grid. All hair samples were sent to the laboratory at Wildlife Genetics International (Nelson, British Columbia, Canada) for genotyping; genetic results were returned in July 2014. In summer 2014, we conducted the final year of hair-snare data collection. We constructed and monitored 35 snare sites in the Durango grid and 34 sites in the wildland grid (Figure 1). Samples collected in 2014 will be sent to the laboratory this fall and results are expected in summer 2015. Once genotypes are returned, we will use spatially-explicit mark recapture statistics (Efford et al. 2009, Gardner et al. 2010) to estimate abundance and density for the urban and wildland grids in each year of the study. Objective 2: Testing a management strategy to reduce bear-human conflicts Given that the primary cause of black bear-human conflicts has been attributed to the availability of human foods to bears, it has been suggested that the most effective strategy to reduce conflicts is to reduce the availability of that resource (Peine 2001, Beckmann et al. 2004, Gore et al. 2005, Spencer et al. 2007). This strategy has had some success within national parks (Greenleaf et al. 2009), and anecdotally in some communities (Mammoth Lakes CA, Juneau AK, Whistler BC), but no research has ever scientifically tested the benefits of “cleaning up” a town. Given the high price to operationally “bearproof” a community, municipalities must have definitive evidence that such an effort would significantly decrease conflict activity before initiating major changes to waste storage and collection practices. As part of this project, we are implementing the first experimental test of wide-scale urban bearproofing for reducing bear-human conflicts. As part of the experiment we have designated 2 residential ‘treatment’ areas and 2 paired ‘control’ areas, consisting of a total of ~2,000 homes (Figure 2). In spring and early summer 2013 we deployed ~900 bear-resistant garbage containers within the treatment areas 6 �(approximately 100 homes already had these containers) with the goal that regular receptacles were exchanged with bear-resistant containers for all residents. In spring and early summer 2014 we deployed an additional ~150 containers to “clean-up” treatment areas, ensuring that all residences had a bearresistant container. This second deployment was necessary due to errors in the City of Durango database and logistical challenges with the initial container deployment. In July 2013 and 2014 we also canvassed homes within treatment areas, reminding residents to lock their bear-resistant garbage containers and asking that they bear-proof their properties (remove bird feeders, outdoor pet food, and other bear attractants); no canvassing occurred in control areas. Additionally, we increased enforcement of wildlife ordinances within treatment areas, providing official warnings at residences with bear-strewn trash and notifying City Code Enforcement. To track the effectiveness of these efforts in reducing bear-human conflicts we are collecting preand post-treatment data. For 2 years pre-treatment, summers 2011 and 2012, field technicians patrolled streets within proposed treatment and control areas on the day waste removal was scheduled to occur (when maximum human food was assumed to be available to bears). Technicians conducted patrols from 5:30 – 7:30 am and recorded locations of bear-strewn trash. Monitoring occurred from July through September, months that experience the highest numbers of bear-human conflicts in Durango (CPW unpublished data). During summers 2013 and 2014 project personnel have been collecting post-treatment data (currently ongoing); post-treatment data will be collected for a minimum of 3 years. Each summer, in addition to collecting information on bears accessing human foods, we have quantified the “availability” of garbage to bears, by documenting the location and container type (bearresistant or regular) of every garbage receptacle in the survey area accessible to bears the night prior to garbage pick-up. These data provide an index of the garbage available to bears in town and allow us to track changes in the number of bear-resistant containers in the study area over the course of the experiment. Once the experiment is complete, we will use data from pre- and post-treatment years, and from treatment and control areas, to quantify the effectiveness of residential bear-proofing. Additionally, we will use information from conflict reports to the CPW Area 15 Office to examine differences in conflict rates between treatment and control areas during post-treatment years. Objective 3: Identifying public attitudes and behaviors related to bear-human encounters Wildlife management agencies must identify the biological factors driving increases in bearhuman conflicts, but they also must identify and incorporate human attitudes and perceptions about this issue into management strategies. This is particularly critical for black bears, as increasing bear-human conflicts around urban development have stimulated significant public interest and concern. It is also critical because bear-human conflicts typically arise over bear-use of human foods, prompting investigators to suggest that a critical component of reducing conflicts is managing human behavior (Beckmann et al. 2004, Gore et al. 2008, Baruch-Mordo et al. 2011). Thus, we have initiated efforts to better understand human attitudes related to bears and bear-human interactions, and human behaviors related to the appropriate use of bear-resistant garbage containers. To assess data on human attitudes we are using public mail surveys to 1) quantify perceptions about bears, bear management, and bear-human interactions, and 2) explore motivations for compliance and non-compliance with wildlife ordinances designed to reduce bear-human conflicts. To meet these objectives, we developed a three-part mail survey, conducted in conjunction with our urban bear-proofing experiment. Residents are being surveyed pre- (2012), during (2014), and post-implementation (2016) of the experiment, in treatment and control areas, as well as across a larger portion of the community. Surveys are mailed to all residents within Durango city limits, and a subset of La Plata county residents within the study area. Survey responses will allow us to quantify current attitudes and perceptions about bear-human interactions, and how those perceptions change over time in association with a management effort such as wide-scale urban bear-proofing. Survey data will also identify the number of residents that 7 �have had interactions with bears, the acceptability of management actions by CPW, and factors that promote or inhibit residents from complying with wildlife ordinances. The first (pre-treatment) public survey was conducted during winter 2012 (see Johnson et al. 2012 for details) and the second survey was conducted during winter 2014, following the first year of the bear-proofing experiment. In addition to collecting data on human attitudes and perceptions, we are also collecting data on human behavior through direct observations. Using a random, stratified sampling design we monitored human compliance with wildlife ordinances at residences in treatment and control areas. Durango city ordinances specify that garbage can only be accessible after 6:00 am on the morning of pick-up; therefore, we define compliance as having garbage adequately secured so that bears cannot access it, either through appropriate use of a bear-resistant garbage container (e.g. latching both latches and having all garbage completely inside the can) or by keeping garbage enclosed in a garage or shed until the morning of trash pick-up. Non-compliance is defined as allowing garbage to be accessible to bears by not using a bearresistant can appropriately (leaving it unlatched) or putting a regular garbage container at the curb the night before garbage pickup. To assess compliance, houses were surveyed on the morning of pick-up (between 5:00 and 6:00 am) between July and September 2013. In each treatment and control area, a sample of 40 randomly selected blocks were monitored (total of 160 blocks) such that the number and type of cans (regular or bear-resistant) and compliance status were recorded. Each block was surveyed twice, once during midsummer and again during late summer. In the north experimental area, compliance was recorded for each parcel, but in the south experimental area, compliance was recorded per block because garbage containers are stored along alleys and cannot be easily tracked to parcel. This protocol was developed, field tested and modified during summer 2013 and data collected this year should be considered pilot. Data collections methods were further modified in summer 2014. Compliance data will be analyzed in conjunction with mail survey data, spatial covariates, and conflict activity to better understand how factors such as management actions and rates of bear-human interactions influence human behavior. A predictive model of compliance behavior will be developed to determine how CPW may tailor education and communication efforts to be most effective at achieving public compliance with wildlife ordinances. RESULTS AND DISCUSSION Objective 1: Determining the influence of urban environments on bear behavior and demography Between 6 June 2013 and 25 March 2014 (the 2013-2014 capture year), an additional 75 unique bears were marked during 206 bear captures. To date on the project there have been 280 different individuals marked during 601 captures. Information about these captures is described below for summer 2013 and winter 2014. During summer 2013 we conducted 129 total bear captures; 42 captures were newly marked unique individuals and 87 were recaptures. Of the unique individuals captured, there were 18 females and 24 males (Table 1). We placed collars on 9 new adult females. With those bears that had been previously collared in 2011 and 2012, this resulted in 39 collars deployed by mid-September, the end of the summer capture season. The mean estimated age of bears ≥1 year-old on their initial capture date was 5.5 (5.7 for females and 5.4 for males), and mean weight was 81.7 kg (57.6 kg for females and 96.6 kg for males). In total, we placed traps at 94 different locations and conducted 1,403 trap nights. Capture success was fairly consistent throughout the summer with peaks in late June and late August (Figure 3). Trapping success was less than in 2012, when limited natural foods likely brought additional bears into the urban-wildland interface around Durango. Between January and March 2014, we visited the winter dens of 35 collared females. Although we had 39 adult female bears collared at the end of summer 2013 there were 4 mortalities in fall: 2 females were legally harvested (B25 and B143), 1 was lethally removed by a landowner (B21), and 1 died 8 �of unknown causes (B167). Of the 35 adult females that we processed, 13 did not have any cubs or yearlings, 9 had yearlings (13 yearlings in total), and 13 had newborn cubs (27 cubs in total, but one was dead in the den; 14 females and 13 males). Of those females with newborn cubs, 1 bear had only 1 cub, 10 bears had twins, and 2 bears had triplets. We PIT and ear-tagged yearlings in the den, recorded information on weight, body size, body condition, and collected hair and blood samples. We also PIT tagged newborn cubs, and recorded their sex and weight. We found that reproductive success, measured as the number of live cubs/adult female was 0.74 (SE = 0.18) for winter 2014, compared to 0.95 (SE = 0.24) in 2012 and 0.52 (SE = 0.16) in 2013. Cub survival for 2014 (survival from newborn to 1 year) was 50% (based on 12 cubs), compared to 40% in 2013. To date, we have obtained >318,000 locations from GPS collars on 67 different adult female bears; 42 different bears provided location data during the summer of 2013 (Figure 4). While most locations were in close proximity to Durango, a few animals ventured outside the primary study area, including a sow that moved to New Mexico (Figure 4). In our initial analyses of bear habitat-use patterns we found that female bears around Durango generally selected for lower elevations, steeper slopes, northern aspects (avoiding southern and western aspects), and alpine, aspen, oak, and riparian vegetation. Bears avoided mixed conifer, meadow/grassland, pinyon-juniper, and shrub vegetation. In testing a series of models describing bear selection for HD, we found that bears selected positively for development, and that the best model included both between- and within-year temporal variation in selection for development (second best model had ∆AIC score = 1911). Bears exhibited greater selection for HD during the poor natural food year (Figure 5, Figure 6). When limited natural food was available, bear selection for HD increased throughout the active season (May-Oct), while in good years, selection for HD decreased throughout the active season (Figure 5). Individual bears also displayed a bimodal pattern, either selecting for very high or very low HD (Figure 5), although selection was predominately for high HD. Our top model had high predictive power when tested against hold-out data. With 10 equal-area bins of predicted resource use, Spearman rank correlations of selection probabilities between training and testing data were 0.99 (p < 0.001). The availability of natural mast foods was generally good in late summer and fall 2013. Mast surveys revealed that the peak time for serviceberry and wild crabapple maturation was early August, for chokecherry and hawthorne was mid-August, for gambel oak was the end of August, and for pinyon pine was early September. On transects that had those species, mast was present for about 10-15% of serviceberry plants, about 25% of pinyon pines and wild crabapples, and for about 50% of hawthorns and chokecherries. Presence of acorns on gambel oaks was highly variable on different transects, ranging from complete failure on some transects to 75% of the plants with acorns on other transects. This was the first year that all mast foods were observed in the study area during the survey period. Between 1 April and 1 December 2013 (the active bear season of 2013), a total of 37 bears were removed from the vicinity of Durango. Of those bears, 15 were killed in vehicle collisions, 14 were legally harvested, 5 were lethally removed due to nuisance behavior (breaking into houses, killing livestock, etc), 2 died from unknown causes, and 1 died when trapped for research purposes. Of those mortalities and removals there were 8 adult females, 3 adult males, 2 subadult females, 5 subadult males, 7 female cubs, and 12 male cubs. Twenty-four bears were unmarked and 13 were marked for the research project. Additionally, 7 marked bears died outside the study area; 4 were legally harvested, 2 died from lethal conflict management (including a bear that died in New Mexico) and 1 was killed in a vehicle collision. In summer 2013, we collected 1,365 hair samples from the Durango and wildland grids; 680 samples from Durango and 685 samples from the wildland site. Over the 6 sampling occasions from 34 snares around Durango we collected 106, 151, 131, 62, 106, and 124 samples, respectively. Over the 6 sampling occasions from 35 sites in the wildland grid we collected 112, 83, 141, 100, 126, and 123 samples, respectively. The number of samples/snare ranged from 0 to 121 in the Durango grid and from 1 9 �to 78 in the wildland grid. We received the genetic results back from Wildlife Genetics International in July 2014. Of the 1,365 hair samples submitted to the laboratory, good genotypes were obtained for 693 samples (51%). Of those samples that did not produce a valid genotype, 389 (28%) did not contain enough genetic material, 273 (20%) failed during analyses for other reasons, and 12 (0.9%) samples were not black bear. Across the 693 valid samples there were 334 genotypes generated from the Durango grid and 359 generated from the wildland grid. In the Durango grid, 86 different individuals were detected during 160 “captures” (multiple hair samples from a single bear during 1 sampling occasion were considered 1 “capture”). Of those individuals, 50 were only detected in 1 sampling occasion and 36 were detected in >1 occasion (recaptures). The probability of detecting a bear within any single sampling occasion was 0.24, and across all sampling occasions was 0.80. In the wildland grid, 110 different individuals were detected during 183 “captures.” Of those individuals, 67 were only detected in 1 sampling occasion and 43 were detected in >1 occasion (recaptures). The probability of detecting a bear within any single sampling occasion was 0.18, and across all sampling occasions was 0.69. Detailed mark-recapture analyses of these data will be conducted in the future to estimate annual density and abundance at each site. Objective 2: Testing management strategies to reduce bear-human conflicts During summer 2013 (July through mid-August) we collected our first year of post-treatment data on the bear-proofing experiment. Within proposed treatment and control areas we observed 330 instances of bears accessing residential garbage during our morning patrols; observations peaked in early September. Of those garbage containers accessed by bears, 84% were regular containers and 16% were bear-resistant containers. Bears accessed human food from bear-resistant containers when they were not properly latched. In quantifying the number of garbage containers out the night before garbage pick-up, we recorded information on 1,678 garbage cans in experimental areas. Within the northern treatment area 89% of the containers were bear-resistant and 11% were regular, and in the southern treatment area 73% of containers were bear-resistant and 27% were regular. In the northern control area 28% of containers were bear-resistant and 72% were regular, and in the southern control area 9% of containers were bearresistant and 91% were regular. Given that we wanted >95% of residents to have a bear-resistant container in the treatment areas, we worked on “cleaning up” the bear-proofing experiment treatment areas this past year. Our intention was to outfit every resident with a bear-resistant container in 2013, but about 150 regular containers were not exchanged due primarily to errors in the city waste management database and oversights by city personnel during deployment. During fall 2013 funds were secured to purchase additional containers through the USDA National Wildlife Research Center and the City of Durango. Containers were purchased from Solid Waste Systems (Parker, CO), a company that manufactures bear-resistant containers certified by the Living with Wildlife Foundation. Containers were distributed to the reminder of residences by City of Durango and CPW staff in spring and early summer 2014. The second year of post-treatment monitoring began in July 2014. Data collection is currently ongoing, but as of August 22nd we had recorded 146 incidences of bears accessing residential garbage; 38 conflicts in treatment areas and 108 conflicts in control areas. In quantifying the number and type of containers in experimental areas (n = 1,483) early the morning of pick-up, we found that our clean-up efforts were a success. Within the northern treatment area 98% of containers were bear-resistant and 2% were regular, and in the southern treatment area 95% were bear-resistant and 5% were regular. Within the northern control area 37% of containers were bear-resistant and 63% were regular, and in the southern control area 11% were bear-resistant and 89% were regular. We will continue working with the City of Durango to replace regular containers with bear-resistant ones in treatment areas. 10 �Objective 3: Identifying public attitudes and behaviors related to bear-human encounters Between January and April 2014, we administered the second mail survey of resident attitudes about bears. Surveys were sent to all residents within Durango city limits and a random sample of 1,500 residents outside city limits but within our study area in La Plata County. A total of 5,853 residents were sent mailings, including a pre-notification letter asking them to complete the survey online, 3 printed copies of the survey, 2 reminder postcards and 1 non-respondent survey. As individuals responded to the survey, they were removed from the mailing list and received no subsequent mailings. We removed a total of 698 bad addresses from the original mailing list, for an adjusted sample of 5,155 residences. We received 2,310 responses (782 online, 1,528 paper) to the survey, for an adjusted response rate of 45%. Non-response bias is currently being assessed. Detailed analysis of tolerance for black bears, compliance behaviors and perceived risk of bear-human conflicts will be conducted in future years. During the 2013 mid-summer compliance survey we found that compliance in the north treatment area was 51%, in the south treatment area was 27%, in the north control area was 12%, and in the south control area was 2%. During the late summer survey, we found that compliance was 39% in the north treatment area, 31% in the south treatment area, 14% in the north control area, and 4% in the south control area. We expected that compliance would increase over the course of the summer as conflicts increased. This pattern appeared evident in all areas except the north treatment area, which demonstrated a notable decrease in compliance (51% to 39%). Because observations sometimes occurred after 6:00 am, residents may have unlatched cans to allow for garbage collection, causing our compliance estimates to be biased low. To account for this, compliance surveys were modified in 2014 to be completed before 6:00 am. Additionally, in 2013, we did not determine if unlatched cans actually contained garbage, therefore making them truly non-compliant, as defined by the city ordinance. As a result, we have updated field data collections methods for 2014 to inspect a sub-sample of compliance sampling blocks to determine the proportion of empty, unlatched cans in our sample. This information will be used to correct bias for data on non-compliance in the future. SUMMARY AND FUTURE PLANS During FY13-14 we successfully coordinated field logistics and conducted several aspects of data collection (trapping and collaring bears, tracking human-related bear mortalities, assessing summer/fall mast availability, implementing DNA hair-snare surveys, monitoring garbage-related bear-human conflicts, assessing resident use of project-supplied bear-resistant containers, etc). Data collection will continue on most aspects of the study through winter 2017, with the collection of some data types concluding earlier (i.e., DNA hair-snare surveys and compliance data will be completed after summer 2014). We have initiated data analysis on bear resource selection of human development, and will continue to analyze data on various aspects of the project in the coming years. In addressing our research objectives we hope to better understand the influence of urban environments on bear populations, elucidate the relationship between bear-human conflicts and bear behavior and demography, understand the effect of bear-human interactions on human attitudes and actions, develop tools to promote the sustainable management of bears in Colorado, and ultimately, identify solutions for reducing bear-human conflicts in urban environments. LITERATURE CITED Baruch-Mordo, S., S.W. Breck, K.R. Wilson, and J. Broderick. 2011. The carrot or the stick? Evaluation of education and enforcement as management tools for human-wildlife conflicts. Plos One 6: e15681. Baruch-Mordo, S., S.W. Breck, K.R. Wilson, and D.M. Theobald. 2008. Spatiotemporal distribution of black bear-human conflicts in Colorado, USA. Journal of Wildlife Management 72:1853-1862. 11 �Baruch-Mordo, S., K.R. Wilson, D. Lewis, J. Broderick, J. Mao, and S.W. Breck. 2010. Roaring Fork Valley urban black bear ecology study: progress report to the Colorado Division of Wildlife. Contact Sharon Baruch-Mordo for copy. Email: sharonb_m@yahoo.com Beck, T.D.I. 1991. Black bears of west-central Colorado. Technical Publication No. 39, Colorado Division of Wildlife, Colorado. Beckmann, J.P., and J. Berger. 2003a. Using black bears to test ideal-free distribution models experimentally. Journal of Mammalogy 84:594-606. Beckmann, J.P., and J. Berger. 2003b. Rapid ecological and behavioural changes in carnivores: the response of black bears (Ursus americanus) to altered food. Journal of Zoology 261:207-212. Beckmann, J.P., L. Karasin, C. Costello, S. Matthews, and Zoë Smith. 2008. Coexisting with black bears: perspectives from four case studies across North America. Wildlife Conservation Society Working Paper No 33. Beckmann, J.P., C.W. Lackey, and J. Berger. 2004. Evaluation of deterrent techniques and dogs to alter behavior of “nuisance” black bears. Wildlife Society Bulletin 32: 1141-1146. Beston, J.A. 2011. Variation in life history and demography of the American black bear. Journal of Wildlife Management 75:1588-1596. Boyce, M.S., P.R. Vernier, S.E. Nielsen, and F.K.A. Schmiegelow. 2002. Evaluating resource selection functions. Ecological Modelling 157:281-300. Burnham, K.P., and D.R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. Second Edition, Springer, New York, New York. Carter, N.H., D.G. Brown, D.R. Etter, and L.G. Visser. 2010. American black bear habitat selection in northern Lower Peninsula, Michigan, USA, using discrete-choice modeling. Ursus 21:57-71. Caswell, H. 2001. Matrix population models: construction, analysis, and interpretation. Second Edition, Sinauer Associates, Sunderland, Massachussetts. Clevenger, A.P., J. Wierzchowski, B. Chruszcz, and K. Gunson. 2002. GIS-generated, expert-based models for identifying wildlife habitat linkages and planning mitigation passages. Conservation Biology 16:503-514. Efford, M.G., D.K. Dawson, and D.L. Borchers. 2009. Population density estimated from locations of individuals on a passive dector array. Ecology 90:2676-2682. Garshelis, D.L., and H. Hristienko. 2006. State and provincial estimates of American black bear numbers versus agency assessments of population trend. Ursus 17:1-7. Gardner, B., J.A. Royle, M.T. Wegan, R.E. Rainbolt, P.D. Curtis. 2010. Estimating black bear density using DNA data from hair snares. Journal of Wildlife Management 74:318-325. Gillies, C.S., M. Hebblewhite, S.E. Nielsen, M.A. Krawchuk, C.L. Aldridge, J.L. Frair, D.J. Saher, C.E. Stevens, and C.L. Jerde. 2006. Application of random effects to the study of resource selection by animals. Journal of Animal Ecology 75:887-898. Gore, M.L., W.F. Siemer, J.E. Shanahan, D. Schuefele, and D.J. Decker. 2005. Effects on risk perception of media coverage of a black bear-related human fatality. Wildlife Society Bulletin 33:507-516. Gore, M.L., B.A. Knuth, C.W. Scherer, and P.D. Curtis. 2008. Evaluating a conservation investment designed to reduce human-wildlife conflicts. Conservation Letters 1:136–145. Greenleaf, S.S., S.M. Matthews, R.G. Wright, J.J. Beecham, and H.M. Leithead. 2009. Food habitat of American black bears as a metric for direct management of human-bear conflict in Yosemite Valley, Yosemite National Park, California. Ursus 20:94-101. Hostetler, J.A., J.W. McCown, E.P. Garrison, A.M Neils, M.A. Barrett, M.E. Sunquist, S.L. Simek, and M.K. Oli. 2009. Demographic consequences of anthropogenic influences: Florida black bears in north-central Florida. Biological Conservation 142:2456-2463. Hristienko, H., and J.E. McDonald Jr. 2007. Going into the 21st century: a perspective on trends and controversies in the management of the American black bear. Ursus 18:72-88. 12 �Johnson, H.E, C.J. Bishop, M.W. Alldredge, J. Brodrick, J. Apker, S. Breck, K. Wilson, and J. Beckmann. 2011. Black bear exploitation of urban environments: finding management solutions and assessing regional population effects. Research Proposal, Colorado Division of Parks and Wildlife, Fort Collins, USA. Johnson, H.E. 2012. Black bear exploitation of urban environments: finding management solutions and assessing regional population effects. Federal Aid Annual Report, Colorado Division of Parks and Wildlife, Fort Collins, USA. Manly, B.F.J., L.L. McDonald, D.L. Thomas, T.L. McDonald, and W.P. Erickson. 2002. Resource selection by animals; statistical design and analysis for field studies. Second Edition. Kluwer Academic Publishers, Boston, Massachussetts. McLoughlin, P.D., D.W. Morris, D. Fortin, E. Vander Wal, and A.L. Contasti. 2010. Considering ecological dynamics in resource selection functions. Journal of Animal Ecology 79:4-12. Mitchell, M.S., L.B. Pacifici, J.B. Grand, and R.A. Powell. 2009. Contributions of vital rates to growth of a protected population of American black bears. Ursus 20:77-84. Morales, J.M., P.R. Moorcroft, J. Matthiopoulos, J.L. Frair, J.G. Kie, R.A. Powell, E.H. Merrill, and D.T. Haydon. 2010. Building the bridge between animal movement and population dynamics. Philosophical Transactions of the Royal Society Series B 365:2289-2301. Mowat, G., and C. Strobeck. 2000. Estimating population size of grizzly bears using hair capture, DNA profiling, and mark-recapture analysis. Journal of Wildlife Management 64:183-193. Noyce, K.V., and P.L. Coy. 1989. Abundance and productivity of bear food species in different forest types of northcentral Minnesota. International Conference on Bear Research and Management 8:169-181. Peine, J.D. 2001. Nuisance bears in communities: Strategies to reduce conflicts. Human Dimensions of Wildlife 6:223-237. Sadeghpour, M.H., and T.F. Ginnett. 2011. Habitat selection by female American black bears in northern Wisconsin. Ursus 22:159-166. Spencer, R.D., R.A. Beausoleil, and D.A. Martorello. 2007. How agencies respond to human–bear conflicts: a survey of wildlife species in North America. Ursus 18: 217–229. Theobald, D.M., and W.H. Romme. 2007. Expansion of the US wildland-urban interface. Landscape and Urban Planning 83:340-354. Woods, J.G., D. Paetkau, D. Lewis, B.N. McLellan, M. Proctor, and C. Strobeck. 1999. Genetic tagging of free-ranging black and brown bears. Wildlife Society Bulletin 27:616-627. Zack, C.S., B.T. Milne, and W.C. Dunn. 2003. Southern oscillation index as an indicator of encounters between humans and black bears in New Mexico. Wildlife Society Bulletin 31:517-520. 13 �Table 1. Capture information for black bears that were marked in the vicinity of Durango, CO between 1 June 2013 and 1 April 2014 (collared adult females are identified with an “*”). Only information from the initial capture of each individual is shown (no recaptures). Bear ID B259 B260 B261* B262 B263 B265 B266 B267 B268 B269 B274 B275 B276 B277 B278 B280 B281* B282 B283 B284 B285* B287 B289 B290 B291* B292 B293* B296 B297 B298* B299* B300 B301 B302 B303* B304 B306 B307 B308 Capture Date 6/3/2013 6/4/2013 6/7/2013 6/10/2013 6/11/2013 6/14/2013 6/17/2013 6/19/2013 6/21/2013 6/21/2013 6/26/2013 6/27/2013 6/29/2013 6/30/2013 7/1/2013 7/8/2013 7/11/2013 7/16/2013 7/16/2013 7/19/2013 7/19/2013 7/23/2013 7/25/2013 8/1/2013 8/2/2013 8/4/2013 8/6/2013 8/10/2013 8/10/2013 8/14/2013 8/16/2013 8/17/2013 8/20/2013 8/23/2013 8/23/2013 8/23/2013 8/29/2013 8/29/2013 8/31/2013 UTM Easting 251514 253231 253231 251343 254584 251343 251933 249872 251817 249153 256824 256937 237708 238206 256763 240340 255385 244631 242022 247443 254750 248532 764348 246591 237566 246591 245913 245913 246383 245848 245848 243934 243934 248263 240787 243934 248263 250971 244631 UTM Northing 4137313 4138879 4138879 4134446 4134994 4134446 4137246 4130099 4131555 4132855 4134340 4134617 4130726 4130573 4134317 4131577 4133334 4132166 4127361 4137388 4133273 4139266 4132592 4135689 4124276 4135689 4139623 4139623 4142011 4141980 4141980 4134857 4134793 4136448 4130376 4134857 4136448 4132450 4132166 14 Sex M M F M M F M F F M M F F M M F F M M F F M M M F M F M M F F M F M F M F M M Age 1 6 8 1 6 1 8 2 2 4 6 2 2 10 3 1 13 3 2 0 8 3 3 4 4 3 3 2 4 5 3 10 1 8 7 1 1 6 4 Kg 32.7 102.1 78.0 30.8 83.9 38.6 125.6 30.8 31.3 54.4 116.6 34.9 57.2 181.4 79.8 25.9 63.5 73.0 33.6 10.4 66.2 75.3 88.5 93.4 64.4 56.7 49.4 45.8 99.3 51.7 61.7 122.0 33.1 133.8 93.0 42.2 33.1 87.1 70.8 �B311 B312 B313* B315 B316 B317 B318 B319 B320 B321 B322 B324 B325 B326 B327 B328 B329 B330 B331 B332 B333 B334 B335 B336 B337 B338 B339 B340 B341 B342 B343 B344 B345 B346 B347 9/3/2013 9/6/2013 9/12/2013 1/24/2014 1/24/2014 2/11/2014 2/11/2014 2/25/2014 2/25/2014 2/25/2014 2/25/2014 2/26/2014 2/26/2014 2/27/2014 2/27/2014 2/28/2014 2/28/2014 2/28/2014 2/28/2014 3/4/2014 3/4/2014 3/5/2014 3/5/2014 3/7/2014 3/7/2014 3/7/2014 3/11/2014 3/11/2014 3/11/2014 3/13/2014 3/13/2014 3/15/2014 3/15/2014 3/17/2014 3/17/2014 243225 248460 248263 257532 257532 253426 253426 248885 248885 248878 248878 242007 242007 249876 249876 245874 245874 250426 250426 244707 244707 236152 236152 237280 237280 237280 241748 241748 241748 250685 250685 256122 256122 763528 763528 4129043 4135188 4136448 4132989 4132989 4131665 4131665 4136736 4136736 4137006 4137006 4135878 4135878 4131660 4131660 4138878 4138878 4130202 4130202 4140518 4140518 4148573 4148573 4133531 4133531 4133531 4133069 4133069 4133069 4138783 4138783 4132990 4132990 4133211 4133211 15 M M F M M M F M F M F M F M M F F M F F F F M M M F F F M F M F F M M 8 10 15 1 1 1 1 1 1 cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub 142.4 179.2 77.1 16.3 16.3 17.2 16.8 15.0 15.9 1.4 1.3 1.9 2.0 1.9 1.8 1.6 1.6 1.6 1.1 1.7 1.8 1.5 1.9 1.9 1.9 1.8 2.9 3.4 2.7 1.5 1.3 2.6 2.6 2.4 2.6 �Figure 1. Locations of the 2014 hair snare sites (red dots) for the Durango and wildland genetic sampling grids. 16 �Figure 2. Change in garbage containers (regular to bear-resistant) at residences in experimental areas pretreatment (2012) and post-treatment (2014), Durango, Colorado. Locations of garbage-related conflicts for each year are also displayed. Garbage Conflicts & Availabitty • Conllicls • R-1...-Gart>age Can • WildMfe-Resistant Gabage Can C Trutment Area C Contn,IA,ea -- 0 0.1 0.2 ' ~+ f ' 0.4 0.6 mer,enl PCCjp, USGS. lnla'lllilp, E5'I Japan, IETI, e.. 0.8 (Hang l<a,g). - ~Rm'lllnl. Miles Cllna l&apny1nlla, OOpenSIIM!Map ~ aid llo 0 0 .1 02 0 .4 O.tl =~ S<luf1lKC DeLam10. USGS.lrltMnal>, n:n,n,nt1>cap. ~CAN. METl El'1 M~s CN"' (ttorg Ka,g~ art (1)l&nlJ. llln'lllnl. """"""'..00pensneu.iap-., an<11ne 0.8 GIS lMr C0nmlnlly GIS UW COMrru1ly 17 �Figure 3. Number of weekly black bear captures from May 15th through September 15th during the 2011 through 2013 summer trapping seasons. 25 2011 Number of Bear Captures 2012 20 2013 15 10 5 0 0 2 4 6 8 10 12 14 Week (May 15th through Sept 15th) 18 16 �Figure 4. GPS collar locations from 42 adult female black bears collected during 1 January – 31 December 2013 in the vicinity of Durango, Colorado (different colored clusters of points represent different individual bears): A) an overview of all locations and B) locations around the town of Durango. B A 21 IM7m F.uu11ng1ou 172m z/plm Ao~--~=-= N on,1"o"a 20 40 60 ...... '00Kilometers 19 �Figure 5. Probabilities of selection by black bears for density of human development from June through October in Durango, Colorado (selection is depicted on a monthly basis). Warm colors depict selection during a poor natural food year (2012) and cooler colors depict selection in a good natural food year (2011). 1 Jun1/Poor Jul1/Poor Aug1/Poor Sep1/Poor Oct1/Poor Jun1/Good Jul1/Good Aug1/Good Sep1/Good Oct1/Good Probability of Selection 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 100 200 300 Density of Human Development/km2 20 �Figure 6. Spatial predictions of resource selection from female black bears in Durango, Colorado, for a good (A) and poor (B) natural food year during fall (Oct 1st). B A 21 �Colorado Division of Parks and Wildlife July 2014 – June 2015 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Colorado 3430 3003 3 Federal Aid Project No. W-204-R4 : : : : : Division of Wildlife Mammals Research Predatory Mammal Conservation Black bear exploitation of urban environments: finding management solutions and assessing regional population effects Period Covered: July 1, 2014 – June 30, 2015 Author: H. E. Johnson Personnel: H.E. Johnson, S.A. Lischka, J. Broderick, J. Apker, S. Breck, J. Beckmann, K. Wilson, and P. Dorsey. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Across the country conflicts among people and black bears are increasing in frequency and severity, and have become a high priority wildlife management issue. Whether increases in conflicts reflect recent changes in bear population trends or just bear behavioral shifts to anthropogenic food resources, is largely unknown, with key implications for bear management. This issue has generated a pressing need for bear research in Colorado and has resulted in a collaborative study involving Colorado Parks and Wildlife (CPW; lead agency), the USDA National Wildlife Research Center, Wildlife Conservation Society and Colorado State University. Collectively, we have designed and implemented a study on black bears that 1) determines the influence of urban environments on bear demography and behavior, 2) tests a management strategy for reducing bear-human conflicts, 3) examines public attitudes and behaviors related to bear-human interactions, and 4) develops population and habitat models to support the sustainable monitoring and management of bears in Colorado. This project was initiated in FY2010-11; during this past fiscal year we focused on collecting field data in the vicinity of Durango, and publishing papers on bear selection for human development and on new immobilization drugs for black bears (Johnson et al. 2015, Wolfe et al. 2014). With respect to data collection, we worked with collaborators and stakeholders on research logistics, trapped and marked black bears, monitored bear demographic rates (adult female survival, adult female fecundity and cub survival) through telemetry and winter den visits, tracked human-related bear mortalities and removals from the study area, performed non-invasive genetic mark-recapture surveys to estimate bear density, collected GPS collar location data on bears along the urban-wildland interface, monitored the availability of late summer/fall mast, obtained data on garbage-related bear-human conflicts, and assessed resident use of project-supplied bear-resistant garbage containers. Information from this study will provide solutions for sustainably managing black 1 �bears outside urban environments, while reducing bear-human conflicts within urban environments; knowledge that is critical for wildlife managers in Colorado and across the country. 2 �WILDLIFE RESEARCH REPORT BLACK BEAR EXPLOITATION OF URBAN ENVIRONMENTS: FINDING MANAGEMENT SOLUTIONS AND ASSESSING REGIONAL POPUATION EFFECTS HEATHER E. JOHNSON PROJECT NARRATIVE OBJECTIVES 1. Determine the influence of urban environments on black bear demography and behavior. 2. Test a management strategy for reducing bear-human conflicts. 3. Examine public attitudes and behaviors related to bear-human interactions. 4. Develop population and habitat models to support the sustainable monitoring and management of bears in Colorado. SEGMENT OBJECTIVES 1. Work with personnel from CPW Area 15, CPW Southwest Region, City of Durango, La Plata County, US Forest Service (Columbine and Pagosa Ranger Districts), Bureau of Land Management (BLM; Tres Rio Field Office), Southern Ute Tribe, and private landowners on field research logistics. 2. Trap and collar adult female black bears in the vicinity of Durango to collect data on bear demography and behavior. 3. Track bear movements and survival via global position system (GPS) collar locations. 4. Monitor bear fecundity and cub survival through winter den investigations of collared adult female bears. 5. Obtain data on natural food conditions for bears based on the abundance of mast from gambel oak, serviceberry, chokecherry, hawthorne, pinyon pine and wild crabapple. 6. Track human-related bear mortalities and removals around Durango from lethal conflict management, vehicle collisions, harvest, and translocations. 7. Perform non-invasive genetic mark-recapture surveys to estimate bear density and population size around Durango (urban site) and in the Piedra watershed (wildland site). 8. Assess the efficacy of wide-scale urban bear-proofing for reducing bear-human conflicts by quantifying conflicts in areas with and without bear-resistant containers. 9. Examine human behavior by monitoring resident compliance with wildlife ordinances in neighborhoods that were provided with bear-resistant garbage containers. 10. Complete an analysis and publication on bear selection for human development. Collaborate with colleagues to test new immobilization drugs for black bears, and disseminate results (Wolfe et al. 2014). 3 �INTRODUCTION In Colorado and across the country, conflicts among people and black bears (Ursus americanus) appear to be increasing in number and severity (Hristienko and McDonald 2007, Baruch-Mordo et al. 2008, CPW unpublished data). Bear-human conflicts can result in public safety concerns, property damage, bear mortality (i.e., euthanasia), and high management costs, and thus, have become a critical wildlife management issue. While wildlife agencies have used a variety of tools to try to minimize bearhuman conflicts (i.e., education, aversive conditioning of bears, and increased harvest), conflict rates have continued to rise. Whether increases in bear-human conflicts reflect recent changes in the bear population or just behavioral shifts to anthropogenic food resources, is largely unknown, as bear population parameters have been exceedingly difficult to estimate (Garshelis and Hristienko 2006). Without a thorough understanding of the relationship between conflict rates and bear behavior and population dynamics, it has been difficult for wildlife agencies to successfully reduce conflicts through bear management. While there is uncertainty about how to reduce bear-human conflicts, two key factors thought to exacerbate this problem are expanding human development and climatic variation. Colorado has had one of the highest rates of exurban development in the nation (Theobald and Romme 2007), and this development has resulted in additional human food on the landscape in the form of garbage, fruit trees, livestock, birdfeeders, etc. The availability of human food to bears has been identified as the primary cause of bear-human conflicts (Spencer et al. 2007, Beckmann et al. 2008, Greenleaf et al. 2009), as bears are opportunistic foragers that will readily take advantage of novel resources. Bear-use of human food not only increases interactions between bears and people but has been found to alter bear activity patterns, foraging behavior, movement rates, and even survival and reproductive rates (Beckmann and Berger 2003a, Beckmann and Berger 2003b, Hostetler et al. 2009), having the potential to significantly influence both bear demography and behavior. This phenomenon is further complicated by variation in annual weather patterns, as bear-use of human development appears to increase when natural foods are in short supply (Zack et al. 2003, Baruch-Mordo et al. 2010). Because bears predominately consume vegetation, recent patterns of drought in Colorado have caused natural food failures for bears in some years. As a result, bears may be increasing their reliance on human foods, with associated behavioral and demographic impacts. While the effects of urbanization and climate have critical implications for modifying bear-habitat relationships, they also have critical implications for increasing rates of bearhuman conflicts. To develop successful strategies to reduce conflicts while maintaining viable bear populations, wildlife agencies must understand how factors such as climate, natural food availability, human food ability, and management influence the behavior and dynamics of bear populations. To address these questions, Colorado Parks and Wildlife has partnered with the USDA National Wildlife Research Center, Wildlife Conservation Society and Colorado State University. Collectively, we initiated a project in FY10-11 to 1) determine the influence of urban environments on bear behavior and demography, 2) test a management strategy for reducing bear-human conflicts, 3) examine public attitudes and behaviors related to bear-human interactions, and 4) develop population and habitat models to support the sustainable monitoring and management of bears in Colorado (Johnson et al. 2011). This information should provide solutions for sustainably managing black bears outside urban environments, while reducing bear-human conflicts within urban environments; knowledge that is critical for wildlife managers in Colorado and across the west. During FY14-15, we worked with collaborators and stakeholders on research logistics, trapped and marked black bears, monitored bear demographic rates (adult female survival, adult female fecundity and cub survival) through telemetry and winter den visits, tracked human-related bear mortalities and removals from the study area, performed non-invasive genetic mark-recapture surveys to estimate bear density, collected GPS collar location data on bears along the urban-wildland interface, monitored the 4 �availability of late summer/fall mast, obtained data on garbage-related bear-human conflicts, and assessed resident use of project-supplied bear-resistant garbage containers. Our efforts focused largely on collecting field data to meet research objectives 1-3, information which will be used to address objective 4. Additionally, we used GPS collar data to perform an analysis of bear selection of human development (Johnson et al. 2015, Appendix 1) and collaborated on a study to develop and test new immobilization drugs for black bears and other species (Wolfe et al. 2014, Appendix 2). We report general summary information from field activities over the past year; detailed analyses of field data are ongoing. STUDY AREA To meet study objectives, we are using a combination of site-specific field data and statewide data. Site-specific field data are being collected in the vicinity of Durango, and are the focus of this progress report. The town of Durango contains ~17,000 people (within city limits) and sits at 1,985 m along the Animas river valley. The town is surrounded by mountainous terrain ranging in elevation from ~1,930 to ~3,600 m, and is generally characterized by mild winters and warm summers that experience monsoon rains. Vegetation in the region is dominated by ponderosa pine, oak, pinyon pine, juniper, aspen, mountain shrubs, and agriculture. Key forage species for black bears include gambel oak (Quercus gambelii), chokecherry (Prunus virginiana), serviceberry (Amelanchier alnifolia), hawthorne (Crataegus spp), wild crabapple (Peraphyllum ramosissimum) and pinyon pine (Pinus edulis). Durango is predominately surrounded by public land managed by the San Juan National Forest, BLM, CPW, La Plata County and the City of Durango. The vicinity of Durango is considered high quality bear habitat, and the town has consistently experienced high rates of bear-human conflicts (Baruch-Mordo et al. 2008, CPW unpublished data). METHODS Objective 1: Determining the influence of urban environments on bear demography and behavior To sustainably manage bears in the face of a growing human population and changing landscape conditions, it is critical to elucidate the drivers and dynamics of bear populations. Of those factors that influence bear populations, the expansion of human development is the least understood, most contentious, and has the greatest potential to elicit major population change. To elucidate the influence of human development on bear demography and behavior, we are collecting a suite of data types including survival and reproductive rates of bears in conjunction with their habitat-use patterns, information on annual summer/fall mast production, and genetic data to estimate bear density in urban and wildland habitats using mark-recapture methods. We briefly describe data collection methods for this portion of the study below; detailed information is available in Johnson et al. (2011). Collaring and Marking Bears – To assess bear demographic rates and behavior with respect to human development, we are capturing and collaring adult female bears. We are specifically targeting adult females as they represent the reproductive segment of the population and allow us to obtain information on multiple key vital rates that drive population growth. For example, in addition to being able to track adult female survival, the vital rate with the highest elasticity (Beston 2011), we can use collared females to track fecundity and cub survival, the vital rates that are typically associated with variation in bear population trends (Mitchell et al. 2009, Beston 2011). We have focused summer trapping efforts within ~10 km of Durango to collar a cohort of bears that experience similar natural food conditions, have anthropogenic food resources readily available, and encompass a range of behaviors and habitat-use patterns relative to the urban-wildland interface. Bears are trapped with box traps, which are baited with fish, road kill, fruit, human foods (at urban locations) and manufactured scents. Traps are set in the evening and checked the following morning. Adult female bears are fitted with a GPS collar (manufactured by Vectronics), and a tooth (first pre-molar) is pulled for 5 �age verification. GPS collars record bear locations every hour, and upload a real-time location to a central database via satellite every 6 hours. Although trapping efforts are focused on adult females, all bears that are trapped (i.e., males, subadults, yearlings) are uniquely marked with a PIT and ear-tag and are weighed, measured, and sampled for blood and hair. Estimating Demographic Rates – To assess the influence of human development on bear demographic rates we have been collecting the following data types: 1) survival and reproduction of collared adult female bears, 2) cub survival monitored during annual winter den checks of collared females, 3) mortalities and removals of marked and unmarked bears in the vicinity of Durango, and 4) non-invasive genetic surveys that estimate density and abundance of bears around urban and wildland sites. Collared female bears allow us to estimate annual adult female survival, fecundity (number of cubs born/adult female) and cub survival (survival from newborn cub to yearling); parameters we have monitored since summer 2011 and which we will continue to monitor through winter 2017. We use realtime GPS collar locations to assess adult female survival, investigating mortalities and slipped collars when GPS locations are stationary during multiple fixes. Fecundity and cub survival are monitored from winter den checks of collared females. Numbers of newborn cubs provide information on fecundity, while consecutive annual den checks of collared females allow us to estimate cub survival. Because yearlings hibernate with their mothers, we can observe the number of cubs alive in the den in year t that survived their first year of life to t+1. Adult female survival, fecundity and cub survival will be used in matrix projection models to assess population performance (Caswell 2001), particularly in relation to use of human development. In addition to tracking survival and reproduction of collared bears, we are also tracking survival and cause-specific mortality of marked (i.e., males, subadults) and unmarked bears in the study area. All bears that are trapped are marked with an ear-tag and PIT tag, unique identifiers that we are using to collect data on human-related bear mortalities and removals. Mortalities and removals primarily occur from translocations, vehicle collisions, conflict-related euthanasia and hunter harvest. For all bears that are removed from the study area we collect a hair and tooth sample and recorded the date, mortality/removal cause, location, bear age, sex, weight, and morphological measurements. We will use mark-recapture and recovery data to estimate adult male and subadult survival, while also gaining valuable information on cause-specific bear mortality. To better understand the influence of urban environments on bear density and abundance, we are employing non-invasive genetic sampling (Woods et al. 1999, Mowat and Strobeck 2000) to compare these parameters between the bear population around Durango and for a nearby “wildland” area. For each area we identified a 36 cell grid (576 km2) where each cell was 4 x 4 km in size. We constructed and monitored 1 snare site within each cell. Snares consisted of a scented bait hanging high in a tree, surrounded by barbed wire around a cluster of trees encircling the bait (wire was strung 50 cm above ground). When bears climb over or under the wire to investigate the bait, they leave a hair sample on the barbed wire. During summers 2011 through 2014, we deployed snares during the first 2 weeks of June, and conducted 6 weekly sampling occasions thereafter. On each occasion, we randomly re-baited the snare with a scent (anise, berry, fish, maple or bacon), and collected hair samples from all barbs. Each hair sample was uniquely catalogued according to the site, date, occasion, and barb number. In summer 2014 we conducted the final year of hair-snare data collection. We constructed and monitored 35 snare sites in the Durango grid and 34 sites in the wildland grid (Figure 1). All hair samples collected in the field were sent to Wildlife Genetics International (Nelson, British Columbia, Canada) for genotyping; genetic results will be returned during summer 2015. Once genotypes are returned, we will use spatially-explicit mark recapture statistics (Efford et al. 2009, Gardner et al. 2010) to estimate abundance and density for the urban and wildland grids in each year of the study. 6 �Evaluating Bear Movement and Habitat-Use Relative to the Urban-Wildland Interface – To examine movement and habitat-use patterns of bears along the urban-wildland interface, we are using GPS location data from collared females. Hourly GPS data are downloaded from the collars in the field on a biannual basis (fall and winter). Locations are being used to assess the influence of factors such as natural food availability, human food availability, weather, habitat covariates, and individual bear attributes (i.e., age, reproductive status) on bear movement and resource selection patterns (Manly et al. 2002, McLoughlin et al. 2010, Morales et al. 2010). For spatial covariate data, we have generated rasters representing elevation, aspect, slope and terrain ruggedness using digital elevation models. We also created rasters depicting distances to drainages and perennial water using the National Hydrology Dataset, and have estimated the proportion of different vegetation types using the USFS LandFire dataset (http://www.landfire.gov/vegetation.php). We derived rasters depicting human structure and road densities using data from La Plata county and CPW. Weather information will be acquired from local weather stations and PRISM datasets (www.prism.oregonstate.edu/). While most habitat and human development information can be extracted from existing spatial data sources, there is no existing data layer that tracks annual variation in late summer/fall hard and soft mast for bears. The abundance of berry and nut resources for bears is known to be highly variable, depending on annual trends in precipitation and temperature (Noyce and Coy 1989). To account for variation in the availability of natural forage for bears around Durango, we conduct bimonthly mast surveys. Surveys are performed from late July through mid-September, when berries and nuts should reach peak maturation. Key mast species for bears around Durango are gambel oak, chokecherry, serviceberry, hawthorne, wild crabapple, and pinyon pine (Beck 1991, Tom Beck, personal communication). We randomly selected 15 transects on public lands to evaluate bear mast availability. Each transect is 1 km in length and situated along an existing trail or stream drainage. For each species, along each transect, field technicians qualitatively assess the phenological stage (immature fruits/nuts, peak maturation, etc) and abundance of mast (proportion of plants with no mast, scarce fruits/nuts, moderate fruits/nuts, etc). During FY14-15 we used GPS collar data to conduct an analysis examining bear selection for human development (Johnson et al. 2015, Appendix 1). In addition to using data from bears around Durango, we collaborated with researchers that have collected similar data around Aspen, Colorado and Lake Tahoe, Nevada, to assess bear selection for development across these 3 sites. I have provided a brief summary of the methods and results in this report, and have included the paper as Appendix 1. Our objectives were to 1) identify temporal patterns of selection for human development within a year and across years based on natural food conditions, 2) compare spatial patterns of selection for development across systems, and 3) examine individual characteristics associated with increased selection for development. We used generalized linear mixed-effects models (GLMMs) following a use-availability design (Manly et al. 2002), using random intercepts account for differences in sample sizes among individuals and autocorrelation within animal datasets (Gillies et al. 2006). “Used” locations were evaluated on an animal-year-specific basis, collected from May – Oct and “available” locations were randomly generated from animal-specific 95% minimum convex polygons. We used 80% of the animal-year datasets for developing models and withheld 20% for model validation. We first ran a “base” model that accounted for topography, vegetation, and distance-to-drainage, which are covariates consistently associated with black bear habitat (Clevenger et al. 2002; Carter et al. 2010; Sadeghpour and Ginnett 2011). We then compared the base model to a series of models that included density of human development (HD), based on point data of human structures in La Plata county. We evaluated the following models: base + HD, base + HD * Food Year (temporal variation in selection for HD between years), base + HD * Week (temporal variation in selection for HD within the active bear year) and base + HD * Food Year * Week (both across and within year variation in selection for development; models included all relevant main 7 �effects and interactions). We fit models using maximum likelihood estimation and used minimum AIC scores to assess the relative support for models with different fixed effects (Burnham and Anderson 2002). We assessed the predictive power of our top model with cross-validation using hold-out data (Boyce et al. 2002). Model results were used to make inferences about bear selection for human development across and within years, and among study sites. To determine whether individual covariates were correlated with bear selection for development we estimated individual selection coefficients (random slope) from GLMMs (Hebblewhite and Merrill, 2008 and Wagner et al., 2011). We restricted this analysis to locations collected during Aug-Sep, allowing us to assess bear selection for human development during the time period of peak conflict activity across sites. Individual selection coefficients were then used as the response variable in a linear regression to test whether bear selection for HD was associated with several covariates. Covariates included the bear’s age, maternal status (cubs or no cubs based on den visits the previous winter), mean HD within the total known 95% MCP for an individual (HDall), and mean HD within the year-specific Aug-Sep 95% MCP (HDhyperphagia). Objective 2: Testing a management strategy to reduce bear-human conflicts Given that the primary cause of black bear-human conflicts has been attributed to the availability of human foods to bears, it has been suggested that the most effective strategy to reduce conflicts is to reduce the availability of that resource (Peine 2001, Beckmann et al. 2004, Gore et al. 2005, Spencer et al. 2007). This strategy has had some success within national parks (Greenleaf et al. 2009), and anecdotally in some communities (Mammoth Lakes CA, Juneau AK, Whistler BC), but no research has ever scientifically tested the benefits of “cleaning up” a town. Given the high price to operationally “bearproof” a community, municipalities must have definitive evidence that such an effort would significantly decrease conflict activity before initiating major changes to waste storage and collection practices. As part of this project, we are implementing the first experimental test of wide-scale urban bearproofing for reducing bear-human conflicts. As part of the experiment we have designated 2 residential ‘treatment’ areas and 2 paired ‘control’ areas, consisting of a total of ~2,000 homes. In spring and early summer 2013 we deployed ~900 bear-resistant garbage containers within the treatment areas (approximately 100 homes already had these containers) with the goal that regular receptacles were exchanged with bear-resistant containers for all residents. In spring and early summer 2014 we deployed an additional ~150 containers to “clean-up” treatment areas, ensuring that all residences had a bearresistant container. In July 2013 and 2014 we also canvassed homes within treatment areas, reminding residents to lock their bear-resistant garbage containers and asking that they bear-proof their properties (remove bird feeders, outdoor pet food, and other bear attractants); no canvassing occurred in control areas. Additionally, we increased enforcement of wildlife ordinances within treatment areas, providing official warnings at residences with bear-strewn trash and notifying city code enforcement for subsequent ticketing. To track the effectiveness of these efforts in reducing bear-human conflicts we are collecting preand post-treatment data. For 2 years pre-treatment, summers 2011 and 2012, field technicians patrolled streets within proposed treatment and control areas on the day waste removal was scheduled to occur (when maximum human food was assumed to be available to bears). Technicians conducted patrols from 5:00 – 7:00 am and recorded locations of bear-strewn trash. Monitoring occurred from July through September, months that experience the highest numbers of bear-human conflicts in Durango (CPW unpublished data). During summers 2013 and 2014 project personnel collected post-treatment data; posttreatment data will be collected through 2016. Each summer, in addition to collecting information on bears accessing human foods, we have quantified the “availability” of garbage to bears, by documenting the location and container type (bearresistant or regular) of every garbage receptacle in the survey area accessible to bears the night prior to 8 �garbage pick-up. These data provide an index of the garbage available to bears in town and allow us to track changes in the number of bear-resistant containers in the study area over the course of the experiment. Once the experiment is complete, we will use data from pre- and post-treatment years, and from treatment and control areas, to quantify the effectiveness of residential bear-proofing. In addition to our observations of bear-strewn trash, we will use conflict calls to the CPW Area 15 Office to examine differences in conflict rates pre- and post-treatment, and across treatment and control areas. Objective 3: Identifying public attitudes and behaviors related to bear-human encounters Wildlife management agencies must identify the biological factors driving increases in bearhuman conflicts, but they also must identify and incorporate human attitudes and perceptions about this issue into management strategies. This is particularly critical for black bears, as increasing bear-human conflicts around urban development have stimulated significant public interest and concern. It is also critical because bear-human conflicts typically arise over bear-use of human foods, prompting investigators to suggest that a critical component of reducing conflicts is managing human behavior (Beckmann et al. 2004, Gore et al. 2008, Baruch-Mordo et al. 2011). Thus, we have initiated efforts to better understand human attitudes related to bears and bear-human interactions, and human behaviors related to the appropriate use of bear-resistant garbage containers. To assess data on human attitudes we are using public mail surveys to 1) quantify perceptions about bears, bear management, and bear-human interactions, and 2) explore motivations for compliance and non-compliance with wildlife ordinances designed to reduce bear-human conflicts. To meet these objectives, we developed a three-part mail survey, conducted in conjunction with our urban bear-proofing experiment. Residents are being surveyed pre- (2012), during (2014), and post-implementation (2016) of the experiment, in treatment and control areas, as well as across a larger portion of the community. Surveys are mailed to all residents within Durango city limits, and a subset of La Plata county residents within the study area. Survey responses will allow us to quantify current attitudes and perceptions about bear-human interactions, and how those perceptions change over time in association with a management effort such as wide-scale urban bear-proofing. Survey data will also identify the number of residents that have had interactions with bears, the acceptability of management actions by CPW, and factors that promote or inhibit residents from complying with wildlife ordinances. The first (pre-treatment) public survey was conducted during winter 2012 (see Johnson et al. 2012 for details) and the second survey was conducted during winter 2014 (see Johnson et al. 2014 for details), following the first year of the bearproofing experiment. In addition to collecting data on human attitudes and perceptions, we are also collecting data on human behavior through direct observations. Using a random, stratified sampling design we are monitoring human compliance with wildlife ordinances at residences in treatment and control areas. Durango city ordinances specify that garbage can only be accessible after 6:00 am on the morning of pick-up; therefore, we define compliance as having garbage adequately secured so that bears cannot access it, either through appropriate use of a bear-resistant garbage container (e.g. latched lid) or by keeping garbage enclosed in a garage or shed until the morning of trash pick-up. Non-compliance is defined as allowing garbage to be accessible to bears by not latching a bear-resistant container or putting a regular garbage container at the curb the night before garbage pickup. To assess compliance, we survey residences on the morning of garbage pick-up (5:00-6:00 am) between July and September. Compliance monitoring began in 2013 and will continue through 2016. In each treatment and control area, a sample of 40 randomly selected blocks are monitored (a total of 160 blocks) such that the number and type of cans (regular or bear-resistant) and compliance status are recorded. Each block is surveyed three times/summer. In the north experimental area, compliance is recorded for each parcel, but in the south experimental area, compliance is recorded per block because garbage containers are stored along alleys and cannot be easily tracked to parcel. Compliance data will be 9 �analyzed in conjunction with mail survey data, spatial covariates, and conflict activity to better understand how factors such as management actions and rates of bear-human interactions influence human behavior. This should help CPW tailor education and communication efforts to be more effective at achieving public compliance with wildlife ordinances. In monitoring compliance, we expected that some residents likely had unsecured containers that did not contain any food for bears, potentially biasing our estimate. We examined this source of bias in summer 2014 by looking in garbage cans to develop a correction factor for “non-compliant” but empty containers. We divided the study area into three smaller areas: the north experimental area with parcels without alleys, the north experimental area with parcels with alleys, and the south experimental area (all parcels had alleys). We aimed to examine ~50 garbage cans/area to determine the proportion that were unsecured but empty. In the north experimental area without alleys, we visually identified all containers that were inaccessible to the technician (those visible but not placed on the street). Those homes were revisited by the technician later in the day, and homeowners were asked if there was garbage in the container the night before their garbage pick-up day. RESULTS AND DISCUSSION Objective 1: Determining the influence of urban environments on bear behavior and demography Between 8 July 2014 and 31 March 2015 (the 2014-2015 capture year), an additional 63 unique bears were marked during 147 bear captures. To date on the project there have been 327 different individuals marked during 717 captures. Information about these captures is described below for summer 2014 and winter 2015. During summer 2014 we conducted 56 total bear captures; 20 captures were newly marked unique individuals and 36 were recaptures. Of the unique individuals captured, there were 7 females and 13 males (Table 1). We placed collars on 5 new adult females. Including bears that were already collared at the start of the summer, this resulted in 42 collars deployed by mid-September, the end of the summer capture season. The mean estimated age of bears ≥1 year-old on their initial capture date was 4.0 (5.9 for females and 3.0 for males), and the mean weight was 65.9 kg (64.2 kg for females and 66.8 kg for males). In total, we placed traps at 63 different locations and conducted 1,021 trap nights. Capture success was fairly consistent throughout the summer, dropping off at the end of the season (Figure 2). Capture effort was slightly reduced from previous years, as we only needed to collar a few additional female bears. Between January and March 2015, we visited the winter dens of 37 collared females. Although we had 42 female bears collared at the end of summer 2014 there were 2 mortalities in fall (B417 was harvested and B161 was shot but not recovered by a hunter), and 3 dens that could not be located. Of the 37 adult females that we processed, 11 did not have any cubs or yearlings, 8 had yearlings (12 yearlings in total; 6 females and 6 males), and 19 had newborn cubs (41 cubs in total; 25 females and 16 males). Additionally, we were able to confirm that B67, a sow that we could not locate during den work, had 2 yearlings in spring of 2015. Of those females with newborn cubs, 1 bear had only 1 cub, 14 bears had twins, and 4 bears had triplets. We PIT and ear-tagged yearlings in the den, recorded information on weight, body size, body condition, and collected hair and blood samples. We also PIT tagged newborn cubs, and recorded their sex and weight. We found that reproductive success, measured as the number of cubs/adult female ≥4 years old was 1.15 (SE=0.20) for winter 2015. This was the highest fecundity rate observed since the study commenced, as previous fecundity rates were between 0.52 (SE=0.16) and 0.95 (SE=0.24). Cub survival for 2015 (survival from newborn to 1 year) was 58% (SE=0.10; based on 24 cubs), compared to 50% in 2014 and 40% in 2013. Between 1 April and 1 December 2014 (the active bear season of 2014), a total of 32 bears were removed from the study area. Of those bears, 22 were legally harvested, 3 were killed in vehicle collisions, 2 were lethally removed for nuisance behavior, 2 were electrocuted (from bears climbing 10 �power poles), 2 died of unknown causes, and 1 was killed by gunshot but not recovered by a hunter. Of those mortalities there were 12 adult males, 7 adult females, 3 subadult males, 5 subadult females, 2 male cubs, 2 female cubs, and 1 male of unknown age. Twenty-two of those bears were unmarked and 10 had been marked by research personnel. Additionally, 6 marked bears died outside the study area; 4 were legally harvested and 2 died in vehicle collisions. Survival of collared adult female bears was high in 2014. The known fate Kaplan Meier estimate of survival was 0.95 (SE=0.04), compared to 0.85 (SE=0.07) in 2013 and 0.82 (SE=0.08) in 2012. Survival in 2011 was similar to 2014 at 0.96 (SE=0.04). In summer 2014, we collected 1,209 hair samples from the Durango and wildland grids; 551 samples from Durango and 658 samples from the wildland site. Over the 6 sampling occasions from 35 snares around Durango we collected 82, 122, 87, 91, 88, and 81 hair samples, respectively. Over the 6 sampling occasions from 34 sites in the wildland grid we collected 159, 134, 127, 81, 61, and 96 hair samples, respectively. The number of samples/snare ranged from 0 to 66 in the Durango grid and from 2 to 50 in the wildland grid. Genotype results should be returned from Wildlife Genetics International during summer 2015. Detailed mark-recapture analyses of these data will be conducted in FY15-16 to estimate annual density and abundance at each site. The availability of natural mast foods was generally very good in late summer and fall 2014. Mast surveys demonstrated that the peak time for maturation of wild crabapple was late July, serviceberry was the first half of August, chokecherry was mid-August, hawthorne was late August, gambel oak was early September, and pinyon pines had cones developing in mid-September. On transects that had those species, mast was present on about 40% of wild crabapple shrubs, 70% of hawthorne shrubs and trees, and 50% of chokecherry, serviceberry oak and pinyon pine shrubs and trees. This was the second year that all mast foods were observed in the study area during the survey period. To date, we have obtained >496,000 locations from GPS collars on 70 different adult female bears; 42 different bears provided location data during the summer of 2014 (Figure 3). While most locations were in close proximity to Durango, a few animals ventured outside the primary study area. For example, B67 had moved to New Mexico in 2013, but moved back to Durango in fall of 2014 (with 2 cubs in tow). Another sow, B57, left her home range in lower Junction Creek (just north of Durango) to travel north to Hamilton Mesa (just south of Norwood; Figure 3), before returning to her original range. Using mixed effects resource selection models we found that bear selection for human development was highly dynamic, varying as a function of changing environmental and physiological conditions (See Appendix 1 for details). Bears increased use of development in years when natural foods were scarce, throughout the summer-fall, as they aged, and as a function of gender, with males exhibiting greater use of development. While patterns were similar across systems, bears at sites with poorer quality habitat selected development more consistently than bears at sites with higher quality habitat. Black bears appear to use development largely for food subsidy, suggesting that conflicts with bears will increase when the physiological demand for resources outweighs risks associated with human activity. These results have key implications for bear management. For example, many bears may be considered “conflict” individuals in a poor natural food year that otherwise exhibit natural foraging behavior. Wildlife agencies often assume that increases in human-bear conflicts reflect increases in bear populations, but our work suggests that bear selection for development may be increasing as animals age and gain more experience with anthropogenic foods. This behavior may then be the source of additional conflicts without an associated increase in population size. Objective 2: Testing management strategies to reduce bear-human conflicts During summer 2014 we collected our second year of post-treatment data on the bear-proofing experiment. To ensure that >95% of residences in treatment areas had bear-resistant containers, we surveyed each treatment and control area during mid-August to quantify the number and type of containers that were visible from the street (n = 1,678). We found that our efforts to “clean up” treatment 11 �areas were a success. Within the northern treatment area 98% of containers were bear-resistant and 2% were regular, and in the southern treatment area 95% were bear-resistant and 5% were regular. Within the northern control area 37% of containers were bear-resistant and 63% were regular, and in the southern control area 11% were bear-resistant and 89% were regular. We will continue working with the City of Durango to replace regular containers with bear-resistant containers in treatment areas. Within proposed treatment and control areas we observed 202 instances of bears accessing residential garbage during morning patrols, 40 conflicts in treatment areas and 162 in control areas (Figure 4). Of those conflicts, 25 were in the north treatment area, 45 were in the north control area, 15 were in the south treatment area and 117 were in the south control area. The number of trash-related conflicts in 2014 was lower than during the previous 2 years and generally peaked in mid-August. Of those garbage containers accessed by bears, 79% were regular containers and 21% were bear-resistant containers. Bears accessed human food from bear-resistant containers when they were not properly latched. While monitoring garbage-related conflicts, we issued 16 notices of violation in treatment areas. One resident was issued a second notice and a ticket from city code enforcement. Objective 3: Identifying human behaviors related to bear-human encounters During summer 2014 we found that the average compliance of residents to wildlife ordinances was 52% in the north treatment area and 34% in the south treatment area. “Compliance” was defined as having a container that was properly locked (both latches clipped) or secured in a garage or shed before 6:00 am. Across all sampling periods, compliance was generally higher in the northern experimental area than in the southern area. In the northern area, compliance increased from ~45% in 2013 to ~55% in 2014. In the southern area compliance remained ~45% in both years. When we surveyed residences to assess the proportion of containers that we labeled “non-compliant” but were devoid of any trash, we found that 7% met that description in the northern experimental area without alleys, 2% in the northern experimental area with alleys, and 14% in the southern experimental area. Future estimates of compliance will be corrected based on these numbers. SUMMARY AND FUTURE PLANS During FY14-15 we successfully coordinated field logistics and conducted several aspects of data collection (trapping and collaring bears, tracking human-related bear mortalities, collecting bear locations on the urban-wildland interface, assessing summer/fall mast availability, implementing DNA hair-snare surveys, monitoring garbage-related bear-human conflicts, etc.). Data collection will continue on most aspects of the study through winter 2017, with the collection of some data types concluding earlier (i.e., DNA hair-snare surveys are now complete). We have initiated data analysis on bear demographic rates and population dynamics, and will continue to analyze data on various aspects of the project in the coming years. In addressing our research objectives we hope to better understand the influence of urban environments on bear populations, elucidate the relationship between bear-human conflicts and bear behavior and demography, understand the effect of bear-human interactions on human attitudes and actions, develop tools to promote the sustainable management of bears in Colorado, and ultimately, identify solutions for reducing bear-human conflicts in urban environments. LITERATURE CITED Baruch-Mordo, S., S.W. Breck, K.R. Wilson, and J. Broderick. 2011. The carrot or the stick? Evaluation of education and enforcement as management tools for human-wildlife conflicts. PLoS One 6: e15681. Baruch-Mordo, S., S.W. Breck, K.R. Wilson, and D.M. Theobald. 2008. Spatiotemporal distribution of black bear-human conflicts in Colorado, USA. Journal of Wildlife Management 72:1853-1862. 12 �Baruch-Mordo, S., K.R. Wilson, D. Lewis, J. Broderick, J. Mao, and S.W. Breck. 2010. Roaring Fork Valley urban black bear ecology study: progress report to the Colorado Division of Wildlife. Contact Sharon Baruch-Mordo for copy. Email: sharonb_m@yahoo.com Beck, T.D.I. 1991. Black bears of west-central Colorado. Technical Publication No. 39, Colorado Division of Wildlife, Colorado. Beckmann, J.P., and J. Berger. 2003a. Using black bears to test ideal-free distribution models experimentally. Journal of Mammalogy 84:594-606. Beckmann, J.P., and J. Berger. 2003b. Rapid ecological and behavioural changes in carnivores: the response of black bears (Ursus americanus) to altered food. Journal of Zoology 261:207-212. Beckmann, J.P., L. Karasin, C. Costello, S. Matthews, and Zoë Smith. 2008. Coexisting with black bears: perspectives from four case studies across North America. Wildlife Conservation Society Working Paper No 33. Beckmann, J.P., C.W. Lackey, and J. Berger. 2004. Evaluation of deterrent techniques and dogs to alter behavior of “nuisance” black bears. Wildlife Society Bulletin 32: 1141-1146. Beston, J.A. 2011. Variation in life history and demography of the American black bear. Journal of Wildlife Management 75:1588-1596. Boyce, M.S., P.R. Vernier, S.E. Nielsen, and F.K.A. Schmiegelow. 2002. Evaluating resource selection functions. Ecological Modelling 157:281-300. Burnham, K.P., and D.R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. Second Edition, Springer, New York, New York. Carter, N.H., D.G. Brown, D.R. Etter, and L.G. Visser. 2010. American black bear habitat selection in northern Lower Peninsula, Michigan, USA, using discrete-choice modeling. Ursus 21:57-71. Caswell, H. 2001. Matrix population models: construction, analysis, and interpretation. Second Edition, Sinauer Associates, Sunderland, Massachussetts. Clevenger, A.P., J. Wierzchowski, B. Chruszcz, and K. Gunson. 2002. GIS-generated, expert-based models for identifying wildlife habitat linkages and planning mitigation passages. Conservation Biology 16:503-514. Efford, M.G., D.K. Dawson, and D.L. Borchers. 2009. Population density estimated from locations of individuals on a passive dector array. Ecology 90:2676-2682. Garshelis, D.L., and H. Hristienko. 2006. State and provincial estimates of American black bear numbers versus agency assessments of population trend. Ursus 17:1-7. Gardner, B., J.A. Royle, M.T. Wegan, R.E. Rainbolt, P.D. Curtis. 2010. Estimating black bear density using DNA data from hair snares. Journal of Wildlife Management 74:318-325. Gillies, C.S., M. Hebblewhite, S.E. Nielsen, M.A. Krawchuk, C.L. Aldridge, J.L. Frair, D.J. Saher, C.E. Stevens, and C.L. Jerde. 2006. Application of random effects to the study of resource selection by animals. Journal of Animal Ecology 75:887-898. Gore, M.L., W.F. Siemer, J.E. Shanahan, D. Schuefele, and D.J. Decker. 2005. Effects on risk perception of media coverage of a black bear-related human fatality. Wildlife Society Bulletin 33:507-516. Gore, M.L., B.A. Knuth, C.W. Scherer, and P.D. Curtis. 2008. Evaluating a conservation investment designed to reduce human-wildlife conflicts. Conservation Letters 1:136–145. Greenleaf, S.S., S.M. Matthews, R.G. Wright, J.J. Beecham, and H.M. Leithead. 2009. Food habitat of American black bears as a metric for direct management of human-bear conflict in Yosemite Valley, Yosemite National Park, California. Ursus 20:94-101. Hebblewhite, M., and E. Merrill. 2008. Modelling wildlife – human relationships for social species with mixed-effects resource selection models. Journal of Applied Ecology 45:834-844. Hostetler, J.A., J.W. McCown, E.P. Garrison, A.M Neils, M.A. Barrett, M.E. Sunquist, S.L. Simek, and M.K. Oli. 2009. Demographic consequences of anthropogenic influences: Florida black bears in north-central Florida. Biological Conservation 142:2456-2463. Hristienko, H., and J.E. McDonald Jr. 2007. Going into the 21st century: a perspective on trends and controversies in the management of the American black bear. Ursus 18:72-88. 13 �Johnson, H.E, C.J. Bishop, M.W. Alldredge, J. Brodrick, J. Apker, S. Breck, K. Wilson, and J. Beckmann. 2011. Black bear exploitation of urban environments: finding management solutions and assessing regional population effects. Research Proposal, Colorado Division of Parks and Wildlife, Fort Collins, USA. Johnson, H.E. 2012. Black bear exploitation of urban environments: finding management solutions and assessing regional population effects. Federal Aid Annual Report, Colorado Division of Parks and Wildlife, Fort Collins, USA. Johnson, H.E., S.A. Lischka, J. Broderick, J. Apker, S. Breck, J. Beckmann, K. Wilson, and P. Dorsey. 2014. Black bear exploitation of urban environments: finding management solutions and assessing regional population effects. Federal Aid Annual Report, Colorado Parks and Wildlife, Fort Collins, USA. Johnson, H.E., S.W. Breck, S. Baruch-Mordo, D.L. Lewis, C.W. Lackey, K.R. Wilson, J. Broderick, J.S. Mao, and J.P. Beckmann. 2015. Shifting perceptions of risk and reward: dynamic selection for human development by black bears in the western United States. Biological Conservation 187:164-172. Manly, B.F.J., L.L. McDonald, D.L. Thomas, T.L. McDonald, and W.P. Erickson. 2002. Resource selection by animals; statistical design and analysis for field studies. Second Edition. Kluwer Academic Publishers, Boston, Massachussetts. McLoughlin, P.D., D.W. Morris, D. Fortin, E. Vander Wal, and A.L. Contasti. 2010. Considering ecological dynamics in resource selection functions. Journal of Animal Ecology 79:4-12. Mitchell, M.S., L.B. Pacifici, J.B. Grand, and R.A. Powell. 2009. Contributions of vital rates to growth of a protected population of American black bears. Ursus 20:77-84. Morales, J.M., P.R. Moorcroft, J. Matthiopoulos, J.L. Frair, J.G. Kie, R.A. Powell, E.H. Merrill, and D.T. Haydon. 2010. Building the bridge between animal movement and population dynamics. Philosophical Transactions of the Royal Society Series B 365:2289-2301. Mowat, G., and C. Strobeck. 2000. Estimating population size of grizzly bears using hair capture, DNA profiling, and mark-recapture analysis. Journal of Wildlife Management 64:183-193. Noyce, K.V., and P.L. Coy. 1989. Abundance and productivity of bear food species in different forest types of northcentral Minnesota. International Conference on Bear Research and Management 8:169-181. Peine, J.D. 2001. Nuisance bears in communities: Strategies to reduce conflicts. Human Dimensions of Wildlife 6:223-237. Sadeghpour, M.H., and T.F. Ginnett. 2011. Habitat selection by female American black bears in northern Wisconsin. Ursus 22:159-166. Spencer, R.D., R.A. Beausoleil, and D.A. Martorello. 2007. How agencies respond to human–bear conflicts: a survey of wildlife species in North America. Ursus 18: 217–229. Theobald, D.M., and W.H. Romme. 2007. Expansion of the US wildland-urban interface. Landscape and Urban Planning 83:340-354. Wagner, T., D.R. Diefenbach, S.A. Christensen, and A.S. Norton. 2011. Using multilevel models to quantify heterogeneity in resource selection. Journal of Wildlife Management 75:1788-1796. Wolfe, L.L., H.E. Johnson, M.C. Fisher, M.A. Sirochman, B. Kraft, and M.W. Miller. 2014. Use of Acepromazine and Medetomidine in combination for sedation and handling of Rocky Mountain elk (Cervus elaphus nelsoni) and black bears (Ursus americanus). Journal of Wildlife Diseases 50:979-981. Woods, J.G., D. Paetkau, D. Lewis, B.N. McLellan, M. Proctor, and C. Strobeck. 1999. Genetic tagging of free-ranging black and brown bears. Wildlife Society Bulletin 27:616-627. Zack, C.S., B.T. Milne, and W.C. Dunn. 2003. Southern oscillation index as an indicator of encounters between humans and black bears in New Mexico. Wildlife Society Bulletin 31:517-520. 14 �Prepared by Heather E. Johnson, Mammals Researcher 15 �Table 1. Capture information for black bears that were newly marked in the vicinity of Durango, CO between 1 June 2014 and 1 March 2015 (collared adult females are identified with an “*”). Only information from the initial capture of each individual is shown (no recaptures). Bear ID B393* B394 B401 B402 B403 B404 B405 B406 B407* B408* B409 B410* B411 B412 B413 B414 B415 B416 B417* B422 B425 B424 B429 B428 B431 B430 B436 B435 B432 B434 B433 B438 B437 B442 B441 B444 B443 B446 B445 B451 B450 B449 B448 B447 B454 Capture Date 7/8/2014 7/11/2014 7/14/2014 7/17/2014 7/24/2014 7/26/2014 7/28/2014 7/30/2014 7/30/2014 8/1/2014 8/8/2014 8/9/2014 8/18/2014 8/18/2014 8/20/2014 8/23/2014 8/26/2014 8/28/2014 8/30/2014 9/14/2014 2/18/2015 2/18/2015 3/4/2015 3/4/2015 3/4/2015 3/4/2015 3/5/2015 3/5/2015 3/5/2015 3/6/2015 3/6/2015 3/9/2015 3/9/2015 3/9/2015 3/9/2015 3/10/2015 3/10/2015 3/11/2015 3/11/2015 3/12/2015 3/12/2015 3/12/2015 3/12/2015 3/12/2015 3/13/2015 UTM Easting 237643 239018 242541 243210 238209 237184 237184 237184 245924 245967 243625 249071 251815 249071 243210 251902 249071 240554 250671 243443 260637 260637 245250 245250 245342 245342 247265 247265 241748 249168 249168 238077 238077 231755 231755 248984 248984 252802 252802 244240 244240 244240 239542 239542 249644 UTM Northing 4124202 4134402 4128373 4128717 4130574 4131686 4131686 4131656 4139622 4141967 4129374 4133006 4131559 4133006 4128717 4130512 4133006 4131370 4132079 4131784 4145413 4145413 4137726 4137726 4137855 4137855 4140832 4140832 4135821 4127890 4127890 4110252 4110252 4115088 4115088 4137372 4137372 4134347 4134347 4133716 4133716 4133716 4134694 4134694 4136076 16 Sex F M M M F M M M F F M F M F M M M M F M F M M F F F F F M F F F F F M F F M M M F F M F F Estimated Age 14 2 2 2 2 2 3 2 8 3 5 5 4 1 4 3 3 4 8 0 cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub Weight (kg) 16.2 44.1 28.4 84.6 44.6 52.2 72.9 41.4 80.1 66.2 95.4 73.4 64.8 30.2 107.1 81.0 43.2 80.1 75.6 30.6 1.0 1.0 2.5 2.2 1.4 1.2 1.5 1.4 1.4 3.4 3.5 2.8 2.5 1.7 1.8 1.6 1.7 2.4 2.5 2.6 2.9 3.4 2.4 2.2 2.8 �B453 B452 B457 B456 B455 B459 B458 B462 B461 B460 B464 B463 B466 B465 B468 B467 3/13/2015 3/13/2015 3/16/2015 3/16/2015 3/16/2015 3/17/2015 3/17/2015 3/18/2015 3/18/2015 3/18/2015 3/19/2015 3/19/2015 3/23/2015 3/23/2015 3/30/2015 3/30/2015 249644 249644 257293 257293 257293 252458 252458 255500 255500 255500 242198 242198 243371 243371 243306 243306 4136076 4136076 4133607 4133607 4133607 4139329 4139329 4133416 4133416 4133416 4131891 4131891 4153025 4153025 4124104 4124104 17 F M F M M M F M M F M F M F F F cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub cub 2.5 2.8 1.8 1.6 1.6 2.7 2.8 2.0 2.0 2.3 2.2 2.2 3.3 3.2 3.5 3.9 �Figure 1. Locations of the 2014 hair snare sites (red dots) for the Durango and wildland genetic sampling grids. 8 12 16 -==-□•••===-••■ Kilometers F ORIDA Mic 18 �Figure 2. Number of weekly black bear captures from May 15th through September 15th during the 2011 through 2014 summer trapping seasons. Note: trapping did not commence until July in 2014. 25 2011 Number of Bear Captures 2012 20 2013 2014 15 10 5 0 0 2 4 6 8 10 12 Week (May 15th through Sept 15th) 19 14 16 �Figure 3. GPS collar locations from 42 adult female black bears collected during 1 January – 31 December 2014 in the vicinity of Durango, Colorado (different colored clusters of points represent different individual bears): A) an overview of all locations and B) locations around the town of Durango. A B .... 20 �Figure 4. Garbage-related black bear-human conflicts observed during July through September 2014. Red lines indicate treatment areas and black lines indicate control areas. Green circles represent conflicts with regular residential garbage containers and yellow circles represent conflicts with wildlife-resistant containers. 231 A o-__;. 02 ic:::::: 50:::i5_ _ _1Kilometers 21 �Colorado Division of Parks and Wildlife July 2015 – June 2016 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Colorado 3430 3003 3 Federal Aid Project No. W-204-R5 : : : : : Division of Wildlife Mammals Research Predatory Mammal Conservation Black bear exploitation of urban environments: finding management solutions and assessing regional population effects Period Covered: July 1, 2015 – June 30, 2016 Author: H. E. Johnson Personnel: H.E. Johnson, S.A. Lischka, J. Broderick, J. Apker, S. Breck, J. Beckmann, K. Wilson, and P. Dorsey. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Across the country conflicts among people and black bears are increasing and have become a high priority for wildlife management agencies. Whether increases in conflicts reflect recent changes in bear population trends or just bear behavioral shifts to anthropogenic food resources, is largely unknown, with key implications for bear management. This issue has generated a pressing need for bear research in Colorado and has resulted in a collaborative study involving Colorado Parks and Wildlife (CPW; lead agency), the USDA National Wildlife Research Center, Wildlife Conservation Society and Colorado State University. Collectively, we have implemented a study on black bears that 1) determines the influence of human development on bear demography and behavior, 2) tests a management strategy for reducing bearhuman conflicts, 3) examines public attitudes and behaviors related to bear-human interactions, and 4) develops population and habitat models to support the sustainable management of bears in Colorado. This project was initiated in FY2010-11; during this past fiscal year we focused on collecting field data in the vicinity of Durango and modeling demographic parameters from known-fate and mark-recapture data. With respect to data collection, we worked with collaborators and stakeholders on research logistics, trapped and marked black bears, monitored bear demographic rates through telemetry and winter den visits, tracked human-related bear mortalities and removals from the study area, collected GPS collar location data from bears along the urban-wildland interface, monitored the availability of summer/fall mast, obtained data on garbage-related bear-human conflicts, assessed resident use of project-supplied bear-resistant containers, and surveyed residents about their attitudes and behaviors with respect to bears. Information from this study will provide solutions for sustainably managing black bears outside urban environments, while reducing bear-human conflicts within urban environments; knowledge that is critical for wildlife managers in Colorado and across the country. 1 �WILDLIFE RESEARCH REPORT BLACK BEAR EXPLOITATION OF URBAN ENVIRONMENTS: FINDING MANAGEMENT SOLUTIONS AND ASSESSING REGIONAL POPUATION EFFECTS HEATHER E. JOHNSON PROJECT NARRATIVE OBJECTIVES The objectives of this project are to 1) determine the influence of urban environments on bear demography and behavior, 2) test a management strategy for reducing bear-human conflicts, 3) examine public attitudes and behaviors related to bear-human interactions, and 4) develop population and habitat models to support the sustainable management of bears in Colorado. SEGMENT OBJECTIVES 1. Work with personnel from CPW Area 15, CPW Southwest Region, City of Durango, La Plata County, US Forest Service, Bureau of Land Management, Southern Ute Tribe, and private landowners on field research logistics. 2. Trap and collar adult female black bears in the vicinity of Durango to collect data on bear demography and behavior. 3. Track bear movements and survival via global position system (GPS) collar locations. 4. Monitor bear fecundity and cub survival through winter den investigations of collared adult female bears. 5. Obtain data on natural food conditions for bears based on the abundance of mast from gambel oak, serviceberry, chokecherry, hawthorne, pinyon pine and native crabapple. 6. Track human-related bear mortalities and removals around Durango from lethal conflict management, vehicle collisions, harvest, and translocations. 7. Assess the efficacy of wide-scale urban bear-proofing for reducing bear-human conflicts by quantifying conflicts in areas with and without bear-resistant containers. 8. Examine human behavior by monitoring resident compliance with wildlife ordinances in neighborhoods that were provided with bear-resistant garbage containers. 9. Survey residents in the study area about their attitudes and behaviors with respect to black bears. INTRODUCTION In Colorado and across the country, conflicts among people and black bears (Ursus americanus) appear to be increasing in number and severity (Hristienko and McDonald 2007, Baruch-Mordo et al. 2008, CPW unpublished data). Bear-human conflicts can result in public safety concerns, property damage, bear mortality (i.e., euthanasia), and high management costs, and thus, have become a critical wildlife management issue. While wildlife agencies have used a variety of tools to try to minimize bearhuman conflicts (i.e., education, aversive conditioning of bears, and increased harvest), conflict rates have continued to rise. Whether increases in bear-human conflicts reflect recent changes in the bear population or just behavioral shifts to anthropogenic food resources, is largely unknown, as bear population parameters have been exceedingly difficult to estimate (Garshelis and Hristienko 2006). Without a �thorough understanding of the relationship between conflict rates and bear behavior and population dynamics, it has been difficult for wildlife agencies to successfully reduce conflicts through bear management. While there is uncertainty about how to reduce bear-human conflicts, two key factors thought to exacerbate this problem are expanding human development and climatic variation. Colorado has had one of the highest rates of exurban development in the nation (Theobald and Romme 2007), and this development has resulted in additional human food on the landscape in the form of garbage, fruit trees, livestock, birdfeeders, etc. The availability of human food to bears has been identified as the primary cause of bear-human conflicts (Spencer et al. 2007, Beckmann et al. 2008, Greenleaf et al. 2009), as bears are opportunistic foragers that will readily take advantage of novel resources. Bear-use of human food not only increases interactions between bears and people but has been associated with changes in bear activity patterns, foraging behavior, movement rates, and even survival and reproductive rates (Beckmann and Berger 2003a, Beckmann and Berger 2003b, Hostetler et al. 2009), having the potential to significantly influence both bear demography and behavior. This phenomenon is further complicated by variation in annual weather patterns, as bear-use of human development appears to increase when natural foods are in short supply (Zack et al. 2003, Baruch-Mordo et al. 2010). Because bears predominately consume vegetation, recent patterns of drought in Colorado have caused natural food failures for bears in some years. As a result, bears may be increasing their reliance on human foods, with associated behavioral and demographic impacts. While the effects of urbanization and climate have critical implications for modifying bear-habitat relationships, they also have critical implications for increasing rates of bearhuman conflicts. To develop successful strategies to reduce conflicts while maintaining viable bear populations, wildlife agencies must understand how factors such as climate, natural food availability, human food ability, and management influence the behavior and dynamics of bear populations. To address these questions, Colorado Parks and Wildlife has partnered with the USDA National Wildlife Research Center, Wildlife Conservation Society and Colorado State University. Collectively, we initiated a project in FY10-11 to 1) determine the influence of urban environments on bear behavior and demography, 2) test a management strategy for reducing bear-human conflicts, 3) examine public attitudes and behaviors related to bear-human interactions, and 4) develop population and habitat models to support the sustainable management of bears in Colorado (Johnson et al. 2011). This information should provide solutions for sustainably managing black bears outside urban environments, while reducing bear-human conflicts within urban environments; knowledge that is critical for wildlife managers in Colorado and across the west. During FY14-15, we worked with collaborators and stakeholders on research logistics, trapped and marked black bears, monitored bear demographic rates (adult female survival, adult female fecundity and cub survival) through telemetry and winter den visits, tracked human-related bear mortalities and removals from the study area, collected GPS collar location data on bears along the urban-wildland interface, monitored the availability of summer/fall mast, obtained data on garbage-related bear-human conflicts, assessed resident use of project-supplied bear-resistant garbage containers, and surveyed residents about their attitudes and behaviors with respect to bears. Our efforts focused largely on collecting field data to meet research objectives 1-3, and initiating the development of bear population models to meet objective 4. We report general summary information from field activities over the past year; detailed analyses of field data are ongoing. STUDY AREA To meet study objectives, we are using a combination of site-specific field data and statewide data. Site-specific field data are being collected in the vicinity of Durango, Colorado and are the focus of �this progress report. The town of Durango contains ~17,000 people (within city limits) and sits at 1,985 m along the Animas river valley. The town is surrounded by mountainous terrain ranging in elevation from ~1,930 to ~3,600 m, and is generally characterized by mild winters and warm summers that experience monsoon rains. Vegetation in the region is dominated by ponderosa pine, oak, pinyon pine, juniper, aspen, mountain shrubs, and agriculture. Key forage species for black bears include gambel oak (Quercus gambelii), chokecherry (Prunus virginiana), serviceberry (Amelanchier alnifolia), hawthorne (Crataegus spp), native crabapple (Peraphyllum ramosissimum) and pinyon pine (Pinus edulis). Durango is predominately surrounded by public land managed by the San Juan National Forest, Bureau of Land Management, Colorado Parks and Wildlife, La Plata County and the City of Durango. The vicinity of Durango is considered high quality bear habitat, and the town has consistently experienced high rates of bear-human conflicts (Baruch-Mordo et al. 2008, CPW unpublished data). METHODS Objective 1: Determining the influence of urban environments on bear demography and behavior To sustainably manage bears in the face of a growing human population and changing landscape conditions, it is critical to elucidate the drivers and dynamics of bear populations. Of those factors that influence bear populations, the expansion of human development is the least understood, most contentious, and has the greatest potential to elicit major population change. To elucidate the influence of human development on bear demography and behavior, we are collecting a suite of data types including survival and reproductive rates of bears in conjunction with their habitat-use patterns, information on annual summer/fall mast production, and genetic data to estimate bear density in urban and wildland habitats using mark-recapture methods. We briefly describe data collection methods for this portion of the study below; detailed information is available in Johnson et al. (2011). Collaring and Marking Bears – To assess bear demographic rates and behavior with respect to human development, we are capturing and collaring adult female bears. We are specifically targeting adult females as they represent the reproductive segment of the population and allow us to obtain information on multiple key vital rates that drive population growth. For example, in addition to being able to track adult female survival, the vital rate with the highest elasticity (Beston 2011), we can use collared females to track fecundity and cub survival, the vital rates that are typically associated with variation in bear population trends (Mitchell et al. 2009, Beston 2011). We have focused summer trapping efforts within ~10 km of Durango to collar a cohort of bears that experience similar natural food conditions, have anthropogenic food resources readily available, and encompass a range of behaviors and habitat-use patterns relative to the urban-wildland interface. Bears are trapped with box traps, which are baited with fish, road kill, fruit, human foods (at urban locations) and manufactured scents. Traps are set in the evening and checked the following morning. Adult female bears are fitted with a GPS collar (manufactured by Vectronics), and a tooth (first pre-molar) is pulled for age verification. GPS collars record bear locations every hour, and upload a real-time location to a central database via satellite every 6 hours. Although trapping efforts are focused on adult females, all bears that are trapped (i.e., males, subadults, yearlings) are uniquely marked with a PIT and ear-tag and are weighed, measured, and sampled for blood and hair. Estimating Demographic Rates – To assess the influence of human development on bear demographic rates we have been collecting the following data types: 1) survival and reproduction of collared adult female bears, 2) cub survival monitored during annual winter den checks of collared females, 3) mortalities and removals of marked and unmarked bears in the vicinity of Durango, and 4) non-invasive genetic surveys that estimate density and abundance of bears around urban and wildland sites. Collared female bears allow us to estimate annual adult female survival, fecundity (number of cubs born/adult female) and cub survival (survival from newborn cub to yearling); parameters we have �monitored since summer 2011 and which we will continue to monitor through winter 2017. We use realtime GPS collar locations to assess adult female survival, investigating mortalities and slipped collars when GPS locations are stationary during multiple fixes. Fecundity and cub survival are monitored from winter den checks of collared females. Numbers of newborn cubs provide information on fecundity, while consecutive annual den checks of collared females allow us to estimate cub survival. Because yearlings hibernate with their mothers, we can observe the number of cubs alive in the den in year t that survived their first year of life to t+1. Adult female survival, fecundity and cub survival will be used in matrix projection models to assess population performance (Caswell 2001), particularly in relation to bear use of human development. In addition to tracking survival and reproduction of collared bears, we are also tracking survival and cause-specific mortality of marked (i.e., males, subadults) and unmarked bears in the study area. All bears that are trapped are marked with an ear-tag and PIT tag, unique identifiers that we are using to collect data on human-related bear mortalities and removals. Mortalities and removals primarily occur from translocations, vehicle collisions, conflict-related euthanasia and hunter harvest. For all bears that are removed from the study area we collect a hair and tooth sample and record the date, mortality/removal cause, location, bear age, sex, weight, and morphological measurements. We will use mark-recapture and recovery data to estimate adult male and subadult survival, while also gaining valuable information on cause-specific bear mortality. To better understand the influence of urban environments on bear density and abundance, we have employed non-invasive genetic sampling (Woods et al. 1999, Mowat and Strobeck 2000) to compare these parameters between the bear population around Durango and for a nearby “wildland” area. For each area we identified a 36 cell grid (576 km2) where each cell was 4 x 4 km in size. We constructed and monitored 1 snare site within each cell. Snares consisted of a scented bait hanging high in a tree, surrounded by barbed wire around a cluster of trees encircling the bait (wire was strung 50 cm above ground). When bears climb over or under the wire to investigate the bait, they leave a hair sample on the barbed wire. During summers 2011 through 2014, we deployed snares during the first 2 weeks of June, and conducted 6 weekly sampling occasions thereafter. On each occasion, we randomly re-baited the snare with a scent (anise, berry, fish, maple or bacon), and collected hair samples from all barbs. All hair samples were sent to Wildlife Genetics International (Nelson, British Columbia, Canada) for genotyping. This past year, we used genotype data to estimate female bear abundance and density around Durango. We used an integrated modeling approach that combined spatially-explicit capture-markrecapture data (SCR) from non-invasive hair snags and location data from GPS-collared females into a single unified analysis (Royle et al. 2013). This approach provided annual estimates of female population abundance, density, and population growth rate and annual estimates of resource selection parameters at the 2nd and 3rd order (Johnson 1980). Between 2011 and 2014, during June and July, non-invasive genetic sampling resulted in the annual detection of 41–61 females and the annual monitoring of 12–34 GPScollared females. We modeled 3rd-order resource selection as a function of 15 spatial covariates previously identified as important predictors of black bear space use (Johnson et al. 2015) and as a function of distance from a bear’s summer home range center. We modeled spatial variation in black bear density (i.e., 2nd-order resource selection) as a function of 4 spatial covariates including elevation, human development, stream density, and a forest classification that included a mixture of aspen, mesic montane, and mixed-conifer forest types. We fit 15 models that included all combinations of the 4 density covariates and a null model that assumed constant density across space. We also fitted 4 additional models that added an interaction term between forest and development to models in the previous model set that contained both covariates. We added a second-order polynomial term for elevation to all models that contained that covariate. We used AIC-based model selection and multi-model inference to rank candidate models and derive model-averaged parameter estimates. To evaluate the potential benefits of integrating GPS data in to our analysis, we also fit the same set of candidate models for abundance and density within a standard SCR framework (no GPS data) that assumes a bivariate-normal space-use model for comparison. �Evaluating Bear Movement and Habitat-Use Relative to the Urban-Wildland Interface – To examine movement and habitat-use patterns of bears along the urban-wildland interface, we are using GPS location data from collared females. Hourly GPS data are downloaded from the collars in the field on a biannual basis (fall and winter). Locations are being used to assess the influence of factors such as natural food availability, human food availability, weather, habitat covariates, and individual bear attributes (i.e., age, reproductive status) on bear movement and resource selection patterns (Manly et al. 2002, McLoughlin et al. 2010, Morales et al. 2010, Johnson et al. 2015). For spatial covariate data, we have generated rasters representing elevation, aspect, slope and terrain ruggedness using digital elevation models. We also created rasters depicting distances to drainages and perennial water using the National Hydrology Dataset, and have estimated the proportion of different vegetation types using the USFS LandFire dataset (http://www.landfire.gov/vegetation.php). We derived rasters depicting human structure and road densities using data from La Plata county and CPW. Weather information has been acquired from local weather stations and from PRISM nationwide datasets (www.prism.oregonstate.edu/). While most habitat and human development information can be extracted from existing spatial data sources, there is no existing data layer that tracks annual variation in late summer/fall hard and soft mast for bears. The abundance of berry and nut resources for bears is known to be highly variable, depending on annual trends in precipitation and temperature (Noyce and Coy 1989). To account for variation in the availability of natural forage for bears around Durango, we conduct bimonthly mast surveys. Surveys are performed from late July through mid-September, when berries and nuts should reach peak maturation. Key mast species for bears around Durango are gambel oak, chokecherry, serviceberry, hawthorne, native crabapple, and pinyon pine (Beck 1991, Tom Beck, personal communication). We randomly selected 15 transects on public lands to evaluate bear mast availability. Each transect is 1 km in length and situated along an existing trail or stream drainage. For each species, along each transect, field technicians qualitatively assess the phenological stage (immature fruits/nuts, peak maturation, etc) and abundance of mast (proportion of plants with no mast, scarce fruits/nuts, moderate fruits/nuts, etc). Objective 2: Testing a management strategy to reduce bear-human conflicts Given that the primary cause of black bear-human conflicts has been attributed to the availability of human foods to bears, it has been suggested that the most effective strategy to reduce conflicts is to reduce the availability of that resource (Peine 2001, Beckmann et al. 2004, Gore et al. 2005, Spencer et al. 2007). This strategy has had some success within national parks (Greenleaf et al. 2009), and anecdotally in some communities (Mammoth Lakes CA, Juneau AK, Whistler BC), but no research has ever scientifically tested the benefits of “cleaning up” a town. Given the high price to operationally “bearproof” a community, many municipalities must have definitive evidence that such an effort would significantly decrease conflict activity before initiating major changes to waste storage and collection practices. As part of this project, we are implementing the first experimental test of wide-scale urban bearproofing for reducing bear-human conflicts. As part of the experiment we have designated 2 residential ‘treatment’ areas and 2 paired ‘control’ areas, consisting of a total of ~2,000 homes. In spring and early summer 2013 we deployed ~900 bear-resistant garbage containers within the treatment areas (approximately 100 homes already had these containers) with the goal that regular receptacles were exchanged with bear-resistant containers for all residents. In spring and early summer 2014 we deployed an additional ~150 containers to “clean-up” treatment areas, ensuring that all residences had a bearresistant container. In July 2013, 2014 and 2015 we also canvassed homes within treatment areas, reminding residents to lock their bear-resistant garbage containers and asking that they bear-proof their properties (remove bird feeders, outdoor pet food, and other bear attractants); no canvassing occurred in control areas. Additionally, we increased enforcement of wildlife ordinances within treatment areas, providing official warnings at residences with bear-strewn trash and notifying city code enforcement for subsequent ticketing. �To track the effectiveness of these efforts in reducing bear-human conflicts we are collecting preand post-treatment data. For 2 years pre-treatment, summers 2011 and 2012, field technicians patrolled streets within proposed treatment and control areas on the day waste removal was scheduled to occur (when maximum human food was assumed to be available to bears). Technicians conducted patrols from 5:00 – 7:00 am and recorded locations of bear-strewn trash. Monitoring occurred from July through September, months that experience the highest numbers of bear-human conflicts in Durango (CPW unpublished data). During summers 2013-2015 project personnel collected post-treatment data, conducting surveys twice/week; post-treatment data will be collected through 2016. Once the experiment is complete, we will use data from pre- and post-treatment years, and from treatment and control areas, to quantify the effectiveness of residential bear-proofing. In addition to our observations of bear-strewn trash, we will use conflict calls to the CPW Area 15 Office to examine differences in conflict rates preand post-treatment, and across treatment and control areas. Objective 3: Identifying public attitudes and behaviors related to bear-human encounters Wildlife management agencies must identify the biological factors driving increases in bearhuman conflicts, but they also must identify and incorporate human attitudes and perceptions about this issue into management strategies. This is particularly critical for black bears, as increasing bear-human conflicts around urban development have stimulated significant public interest and concern. It is also critical because bear-human conflicts typically arise over bear-use of human foods, prompting investigators to suggest that a critical component of reducing conflicts is managing human behavior (Beckmann et al. 2004, Gore et al. 2008, Baruch-Mordo et al. 2011). Thus, we have initiated efforts to better understand human attitudes related to bears and bear-human interactions, and human behaviors related to the appropriate use of bear-resistant garbage containers. To assess data on human attitudes, we are using public mail surveys to 1) quantify perceptions about bears, bear management, and bear-human interactions, and 2) explore motivations for compliance and non-compliance with wildlife ordinances designed to reduce bear-human conflicts. To meet these objectives, we developed a three-part mail survey, conducted in conjunction with our urban bear-proofing experiment. Residents were surveyed pre- (2012), during (2014), and post-implementation (2016) of the experiment, in treatment and control areas, as well as across a larger portion of the community. Johnson et al. (2012) and (2014) provide detailed information about the 2012 and 2014 surveys, respectively. The 2016 survey occurred between January and June 2016 (Appendix 1), where all residents within the city limits of Durango and a sample of residents within the county were asked about their interactions with bears, perceptions of management actions to reduce conflicts, and household actions to reduce conflict. The survey was mailed to a total of 6,566 individuals and we had a valid sample of 5,449 (1,117 surveys were invalid because they could not be delivered to the intended recipients). Responses are currently being electronically recorded. Survey responses will allow us to quantify current attitudes and perceptions about bear-human interactions, and how those perceptions have changed over time in association with a management effort such as wide-scale urban bear-proofing. Survey data will also identify the number of residents that have had interactions with bears, the acceptability of management actions by CPW, and factors that promote or inhibit residents from complying with wildlife ordinances. In addition to collecting data on human attitudes and perceptions, we are also collecting data on human behavior through direct observations. Using a random, stratified sampling design we are monitoring human compliance with wildlife ordinances at residences in treatment and control areas. Durango city ordinances specify that garbage can only be accessible after 6:00am on the morning of pickup; therefore, we define compliance as having garbage adequately secured so that bears cannot access it, either through appropriate use of a bear-resistant garbage container (e.g. latched lid) or by keeping garbage enclosed in a garage or shed until the morning of trash pick-up. Non-compliance is defined as allowing garbage to be accessible to bears by not latching a bear-resistant container or putting a regular garbage container at the curb the night before garbage pickup. �To assess compliance, we observe residences on the morning of garbage pick-up (5:00-6:00 am) between July and September. Compliance monitoring began in 2013 and will continue through 2016. In each treatment and control area, a sample of 40 randomly selected blocks are monitored (a total of 160 blocks) such that the number and type of cans (regular or bear-resistant) and compliance status are recorded. Each block is surveyed three times/summer. In the north experimental area, compliance is recorded for each parcel, but in the south experimental area, compliance is recorded per block because garbage containers are stored along alleys and cannot be easily tracked to parcel. Compliance data will be analyzed in conjunction with mail survey data, spatial covariates, and conflict activity to better understand how factors such as management actions and rates of bear-human interactions influence human behavior. This should help CPW tailor education and communication efforts to be more effective at achieving public compliance with wildlife ordinances. RESULTS AND DISCUSSION Objective 1: Determining the influence of urban environments on bear behavior and demography Between 5 July 2015 and 23 March 2016 (the 2015-2016 capture year), an additional 54 unique bears were marked during 136 bear captures (Table 1). To date on the project there have been 380 different individuals marked during 891 captures. Information about these captures is described below for summer 2015 and winter 2016. During summer 2015 we conducted 56 total bear captures; 29 captures were newly marked unique individuals and 27 were recaptures. Of the unique individuals captured, there were 7 females and 20 males (Table 1). We placed collars on 5 new adult females. Including bears that were already collared at the start of the summer, this resulted in 38 collars deployed by mid-September, the end of the summer capture season. The mean estimated age of bears ≥1 year-old on their initial capture date was 5.0 (7.0 for females and 4.6 for males), and the mean weight was 78.4 kg (73.5 kg for females and 79.6 kg for males). In total, we placed traps at 43 different locations and conducted 767 trap nights. Capture success peaked in late August and early September (Figure 1). Capture effort was slightly reduced from previous years, as we only needed to collar a few additional female bears to maintain our target sample size. Between January and March 2016, we visited the winter dens of 34 collared females. Although we had 38 female bears collared at the end of the trapping season in mid-September 2015, 4 bears lost their collars in September and October due to faulty ‘spacers’. In case we cannot recapture a bear, we always attach collars using a biodegradable spacer (designed to rot off >12 months post-deployment). Fabricon, our manufacuture, had given us a new spacer design this past year that was prematurely rotting off; the problem was subsequently rectified. Of those 34 dens that we visited, we processed bears in 30 dens; 2 dens were too dangerous to enter and 2 bears left the dens when we approached and never redenned during the field season (both were barren). We obtained reproductive information from all 34 collared bears (trail cameras were used on the dens that were too dangerous to enter): 7 were barren, 15 had yearlings (24 yearlings in total; 13 females, 10 males, and 1 unknown [confirmed on trail camera]), and 12 had newborn cubs (25 cubs in total; 10 females and 15 males). Of those females with newborn cubs, 3 bears had only 1 cub, 5 bears had twins, and 4 bears had triplets. We PIT and ear-tagged yearlings in the den, recorded information on weight, body size, body condition, and collected hair and blood samples. We also PIT tagged newborn cubs, and recorded their sex and weight. We found that reproductive success, measured as the number of cubs/adult female ≥4 years old was 0.74 (SE=0.15) for winter 2016; previous fecundity rates have varied between 0.58 and 1.28. Annual cub survival (survival from newborn to 1 year) was 0.66 (SE=0.08; based on 33 cubs) which was the highest rate observed during the study. Previous annual values have varied from 0.42 to 0.54. Between 1 April and 30 March 2016 (based on when bears emerge from their dens each spring), annual survival of collared adult female bears was 0.88 (SE=0.05), which is close to the 5 year study average (range: 0.82 – 0.94). Four collared bears died during the year: 2 died in vehicle collisions, 1 was harvested and 1 died of unknown causes (the bear was estimated to be 16+ years old). Throughout the �study area, a total of 41 bears (marked and unmarked) died or were translocated. Sixteen bears were killed in vehicle collisions, 14 were legally harvested, 6 were lethally removed for nuisance behavior, 2 died of unknown causes, 2 were translocated, and 1 was electrocuted (cub climbing a power pole). Of those mortalities there were 9 adult males, 8 adult females, 6 subadult males, 2 subadult females, 2 male cubs, 2 female cubs, and 1 cub of unknown gender. Seventeen of those bears were unmarked and 13 had been marked by research personnel. Additionally, 2 marked bears died outside the study area; both were males that were legally harvested. To date, we have obtained >705,000 locations from GPS collars on 83 different adult female bears; 46 different bears provided 113,973 GPS locations during the summer of 2015 (Figure 2). Collared bears generally stayed within the vicinity of Durango; there were no extraordinary movements recorded this past year. The furthest a bear traveled to the north was up Hermosa Creek, to the east was Vallecito Reservoir, to the south was the Colorado-New Mexico border, and to the west was the La Plata River. The availability of natural mast foods was generally moderate in late summer and fall 2015 (Figure 3). Surveys demonstrated that the peak time for mast maturation of native crabapple was early August, serviceberry was between mid-August and mid-September (depending on transect location), chokecherry was early September, hawthorne was mid-September, gambel oak was mid-September, and pinyon pines was in mid- to late-September. Generally, the maturation of soft and hard mast occurred later in 2015 than in previous years. On transects that had key mast species, mast was present on about 25% of chokecherry, 15% of native crabapple, and 10% of oak and serviceberry shrubs, while approximately 30% of pinyon pines produced moderate to abundant cones. Hawthorne berries were only observed on 1 transect, but production was abundant on 80% of those plants. While mast from important species like oak and chokecherry were relatively low in 2015, mast from native crabapple and pinyon pines were quite high; pinyon pines had >3 times the mast that had been observed during any previous year of the study (Figure 3). Based on a study area size of 840km2, integrated spatially-explicit capture-mark-recapture models (IntSCR) estimated that female bear abundance in the vicinity of Durango was 156.6 (SE = 22.2) in 2011, 182.7 (SE = 35.7) in 2012, 83.7 (SE = 9.8) in 2013, and 76.2 (SE = 11) in 2014. Density estimates ranged from 0.09 (SE = 0.01) to 0.22 (SE = 0.04; Figure 4). Model averaged estimates of abundance and density based on standard SCR models (using only hair-snare data, no GPS data) were typically greater than the integrated-SCR estimates and were generally less precise (Figure 4). We identified the following models as the top-ranked model for each year, respectively; 2011: forest-only model, 2012: development-only model, 2013: development and elevation model, and 2014: elevation-only model. Predicted density surfaces derived from model-averaged estimates are provided in Figure 5. Abundance and density estimates were dramatically lower in 2013 and 2014, which followed a severe natural food failure in late summer/fall of 2012. Objective 2: Testing management strategies to reduce bear-human conflicts During summer 2015 we collected our third year of post-treatment data on the bear-proofing experiment. To ensure that >95% of residences in treatment areas had bear-resistant containers, we surveyed each treatment and control area during early-August to quantify the number and type of containers that were visible from the street (n = 1,341). We found that our efforts to “clean up” treatment areas were a success. Within the northern treatment area 98% of containers were bear-resistant and 2% were regular, and in the southern treatment area 95% were bear-resistant and 5% were regular. We will continue working with the City of Durango to replace regular containers with bear-resistant containers in treatment areas. Within the northern control area 40% of containers were bear-resistant and 60% were regular, and in the southern control area 24% were bear-resistant and 76% were regular. The proportions of bear-resistant containers within control areas have increased over the course of the study as residents have purchased them from the City. For example, when the study was initiated in 2011, only 28% of �residences in the northern control area had bear-resistant containers and only 9% had bear-resistant containers in the southern control area. Within treatment and control areas we observed 473 instances of bears accessing residential garbage during morning patrols, 115 conflicts in treatment areas and 358 in control areas (Figure 6). Of those conflicts, 47 were in the north treatment area, 103 were in the north control area, 33 were in the south treatment area and 290 were in the south control area. The number of trash-related conflicts in 2015 was higher than during the previous 3 years and peaked in late-August. Of those garbage containers accessed by bears, 76% were regular containers and 24% were bear-resistant containers. Bears accessed human food from bear-resistant containers when they were not properly latched or when trash was stored outside of the cans. We used kernel density functions (Worton 1987) with an href value (Gitzen et al. 2006) to spatially estimate the probability of trash-related bear conflicts before and after the distribution of bear resistant containers. We found that since the implementation of the bear-proofing experiment in 2013, trash conflicts have been significantly reduced in the northern experimental unit, and have shifted to the control area in the south experimental unit (Figure 7). While monitoring garbage-related conflicts, we issued 31 notices of violation in treatment areas. Objective 3: Identifying human behaviors and attitudes related to bear-human encounters We received a total of 2,432 valid mail survey responses from residents in Durango and La Plata county, which resulted in a 45% survey response rate. Of those surveys, 1,681 residents completed paper surveys and 751 submitted online responses. Survey data is currently being electronically recorded for future analysis. During summer 2015 we found that the average compliance of residents to wildlife ordinances was 59% in the north treatment area and 35% in the south treatment area. “Compliance” was defined as having a container that was properly locked (both latches clipped) or secured in a garage or shed before 6:00am. Across all sampling periods, compliance was higher in the northern experimental area than in the southern area. In the northern area, compliance increased from 45% in 2013 to 52% in 2014, to 59% in 2015. In the southern area compliance increased from 29% in 2013, to 34% in 2014, to 35% in 2015. When we surveyed residences to assess the proportion of containers that we labeled “non-compliant” (clips unlatched) but were devoid of any trash, we found that 4% met that description in the northern experimental area, and 26% in the southern experimental area (which has alleys). Future estimates of compliance will be corrected based on these numbers. SUMMARY AND FUTURE PLANS During FY15-16 we successfully coordinated field logistics and conducted several aspects of data collection (trapping and collaring bears, tracking human-related bear mortalities, collecting bear locations on the urban-wildland interface, assessing summer/fall mast availability, monitoring garbage-related bearhuman conflicts, conducting mail surveys, etc.) and initiated demographic analyses. Data collection will continue through winter 2017, and we will continue to analyze data and prepare research publications. In the coming year, we will be finalizing demographic estimates from the non-invasive genetic markrecapture data, and developing integrated population models which can be used to better track trends in bear population dynamics. In addition, we will be identifying factors affecting driving tolerance for black bears, compliance behaviors related to bear-proofing, and the effects of bear-proofing efforts on risk of conflict with bears. Once data collection is complete, we will then be able to conduct the remainder of the analyses needed to meet project goals. By addressing our research objectives we hope to better understand the influence of urban environments on bear populations, elucidate the relationship between bear-human conflicts and bear behavior and demography, understand the effect of bear-human interactions on human attitudes and actions, develop tools to promote the sustainable management of bears in Colorado, and ultimately, identify solutions for reducing bear-human conflicts in urban environments. �LITERATURE CITED Baruch-Mordo, S., S.W. Breck, K.R. Wilson, and J. Broderick. 2011. The carrot or the stick? Evaluation of education and enforcement as management tools for human-wildlife conflicts. PLoS One 6: e15681. Baruch-Mordo, S., S.W. Breck, K.R. Wilson, and D.M. Theobald. 2008. Spatiotemporal distribution of black bear-human conflicts in Colorado, USA. Journal of Wildlife Management 72:1853-1862. Baruch-Mordo, S., K.R. Wilson, D. Lewis, J. Broderick, J. Mao, and S.W. Breck. 2010. Roaring Fork Valley urban black bear ecology study: progress report to the Colorado Division of Wildlife. Contact Sharon Baruch-Mordo for copy. Email: sharonb_m@yahoo.com Beck, T.D.I. 1991. Black bears of west-central Colorado. Technical Publication No. 39, Colorado Division of Wildlife, Colorado. Beckmann, J.P., and J. Berger. 2003a. Using black bears to test ideal-free distribution models experimentally. Journal of Mammalogy 84:594-606. Beckmann, J.P., and J. Berger. 2003b. Rapid ecological and behavioural changes in carnivores: the response of black bears (Ursus americanus) to altered food. Journal of Zoology 261:207-212. Beckmann, J.P., L. Karasin, C. Costello, S. Matthews, and Zoë Smith. 2008. Coexisting with black bears: perspectives from four case studies across North America. Wildlife Conservation Society Working Paper No 33. Beckmann, J.P., C.W. Lackey, and J. Berger. 2004. Evaluation of deterrent techniques and dogs to alter behavior of “nuisance” black bears. Wildlife Society Bulletin 32: 1141-1146. Beston, J.A. 2011. Variation in life history and demography of the American black bear. Journal of Wildlife Management 75:1588-1596. Boyce, M.S., P.R. Vernier, S.E. Nielsen, and F.K.A. Schmiegelow. 2002. Evaluating resource selection functions. Ecological Modelling 157:281-300. Burnham, K.P., and D.R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. Second Edition, Springer, New York, New York. Carter, N.H., D.G. Brown, D.R. Etter, and L.G. Visser. 2010. American black bear habitat selection in northern Lower Peninsula, Michigan, USA, using discrete-choice modeling. Ursus 21:57-71. Caswell, H. 2001. Matrix population models: construction, analysis, and interpretation. Second Edition, Sinauer Associates, Sunderland, Massachussetts. Clevenger, A.P., J. Wierzchowski, B. Chruszcz, and K. Gunson. 2002. GIS-generated, expert-based models for identifying wildlife habitat linkages and planning mitigation passages. Conservation Biology 16:503-514. Garshelis, D.L., and H. Hristienko. 2006. State and provincial estimates of American black bear numbers versus agency assessments of population trend. Ursus 17:1-7. Gillies, C.S., M. Hebblewhite, S.E. Nielsen, M.A. Krawchuk, C.L. Aldridge, J.L. Frair, D.J. Saher, C.E. Stevens, and C.L. Jerde. 2006. Application of random effects to the study of resource selection by animals. Journal of Animal Ecology 75:887-898. Gitzen, R.A., J.J. Millspaugh, and B.J. Kernohan. 2006. Bandwidth selection for fixed-kernal analysis of animal utilization distributions. Journal of Wildlife Management 70:1334-1344. Gore, M.L., W.F. Siemer, J.E. Shanahan, D. Schuefele, and D.J. Decker. 2005. Effects on risk perception of media coverage of a black bear-related human fatality. Wildlife Society Bulletin 33:507-516. Gore, M.L., B.A. Knuth, C.W. Scherer, and P.D. Curtis. 2008. Evaluating a conservation investment designed to reduce human-wildlife conflicts. Conservation Letters 1:136–145. Greenleaf, S.S., S.M. Matthews, R.G. Wright, J.J. Beecham, and H.M. Leithead. 2009. Food habitat of American black bears as a metric for direct management of human-bear conflict in Yosemite Valley, Yosemite National Park, California. Ursus 20:94-101. Hebblewhite, M., and E. Merrill. 2008. Modelling wildlife – human relationships for social species with mixed-effects resource selection models. Journal of Applied Ecology 45:834-844. Hostetler, J.A., J.W. McCown, E.P. Garrison, A.M Neils, M.A. Barrett, M.E. Sunquist, S.L. Simek, and �M.K. Oli. 2009. Demographic consequences of anthropogenic influences: Florida black bears in north-central Florida. Biological Conservation 142:2456-2463. Hristienko, H., and J.E. McDonald Jr. 2007. Going into the 21st century: a perspective on trends and controversies in the management of the American black bear. Ursus 18:72-88. Johnson, D.H. 1980. The comparison of usage and availability measurements for evaluating resource preference. Ecology 61:65-71. Johnson, H.E, C.J. Bishop, M.W. Alldredge, J. Brodrick, J. Apker, S. Breck, K. Wilson, and J. Beckmann. 2011. Black bear exploitation of urban environments: finding management solutions and assessing regional population effects. Research Proposal, Colorado Division of Parks and Wildlife, Fort Collins, USA. Johnson, H.E. 2012. Black bear exploitation of urban environments: finding management solutions and assessing regional population effects. Federal Aid Annual Report, Colorado Division of Parks and Wildlife, Fort Collins, USA. Johnson, H.E., S.A. Lischka, J. Broderick, J. Apker, S. Breck, J. Beckmann, K. Wilson, and P. Dorsey. 2014. Black bear exploitation of urban environments: finding management solutions and assessing regional population effects. Federal Aid Annual Report, Colorado Parks and Wildlife, Fort Collins, USA. Johnson, H.E., S.W. Breck, S. Baruch-Mordo, D.L. Lewis, C.W. Lackey, K.R. Wilson, J. Broderick, J.S. Mao, and J.P. Beckmann. 2015. Shifting perceptions of risk and reward: dynamic selection for human development by black bears in the western United States. Biological Conservation 187:164-172. Manly, B.F.J., L.L. McDonald, D.L. Thomas, T.L. McDonald, and W.P. Erickson. 2002. Resource selection by animals; statistical design and analysis for field studies. Second Edition. Kluwer Academic Publishers, Boston, Massachussetts. McLoughlin, P.D., D.W. Morris, D. Fortin, E. Vander Wal, and A.L. Contasti. 2010. Considering ecological dynamics in resource selection functions. Journal of Animal Ecology 79:4-12. Mitchell, M.S., L.B. Pacifici, J.B. Grand, and R.A. Powell. 2009. Contributions of vital rates to growth of a protected population of American black bears. Ursus 20:77-84. Morales, J.M., P.R. Moorcroft, J. Matthiopoulos, J.L. Frair, J.G. Kie, R.A. Powell, E.H. Merrill, and D.T. Haydon. 2010. Building the bridge between animal movement and population dynamics. Philosophical Transactions of the Royal Society Series B 365:2289-2301. Mowat, G., and C. Strobeck. 2000. Estimating population size of grizzly bears using hair capture, DNA profiling, and mark-recapture analysis. Journal of Wildlife Management 64:183-193. Noyce, K.V., and P.L. Coy. 1989. Abundance and productivity of bear food species in different forest types of northcentral Minnesota. International Conference on Bear Research and Management 8:169-181. Peine, J.D. 2001. Nuisance bears in communities: Strategies to reduce conflicts. Human Dimensions of Wildlife 6:223-237. Royle, J.A., R.B. Chandler, C.C. Sun, and A.K. Fuller. 2013. Integrating resource selection information with spatial capture-recapture. Methods in Ecology and Evolution 4:520-530. Sadeghpour, M.H., and T.F. Ginnett. 2011. Habitat selection by female American black bears in northern Wisconsin. Ursus 22:159-166. Spencer, R.D., R.A. Beausoleil, and D.A. Martorello. 2007. How agencies respond to human–bear conflicts: a survey of wildlife species in North America. Ursus 18: 217–229. Theobald, D.M., and W.H. Romme. 2007. Expansion of the US wildland-urban interface. Landscape and Urban Planning 83:340-354. Wagner, T., D.R. Diefenbach, S.A. Christensen, and A.S. Norton. 2011. Using multilevel models to quantify heterogeneity in resource selection. Journal of Wildlife Management 75:1788-1796. Wolfe, L.L., H.E. Johnson, M.C. Fisher, M.A. Sirochman, B. Kraft, and M.W. Miller. 2014. Use of �Acepromazine and Medetomidine in combination for sedation and handling of Rocky Mountain elk (Cervus elaphus nelsoni) and black bears (Ursus americanus). Journal of Wildlife Diseases 50:979-981. Woods, J.G., D. Paetkau, D. Lewis, B.N. McLellan, M. Proctor, and C. Strobeck. 1999. Genetic tagging of free-ranging black and brown bears. Wildlife Society Bulletin 27:616-627. Worton, B.J. 1987. A review of models of home range for animal movement. Ecological Modelling 38:277-298. Zack, C.S., B.T. Milne, and W.C. Dunn. 2003. Southern oscillation index as an indicator of encounters between humans and black bears in New Mexico. Wildlife Society Bulletin 31:517-520. Prepared by Heather E. Johnson, Wildlife Researcher �Table 1. Capture information for black bears that were newly marked in the vicinity of Durango, CO during summer 2015 and winter 2016 (collared adult females are identified with an “*”). Only information from the initial capture of each individual is shown (no recaptures). Bear ID B470 B469 B471 B472 B473 B474 B475 B476 B477 B478 B479 B480* B481* B482 B483* B484 B485 B486 B487 B488 B489 B490 B491* B492* B493 B514 B515 B531 B547 B532 B533 B534 B535 B536 B537 B538 B539 B540 B541 B542 B543 B544 B545 B546 Capture Date 7/15/2015 7/21/2015 7/28/2015 8/4/2015 8/10/2015 8/19/2015 8/21/2015 8/22/2015 8/25/2015 8/27/2015 8/27/2015 8/28/2015 8/29/2015 8/31/2015 8/31/2015 9/2/2015 9/3/2015 9/3/2015 9/9/2015 9/10/2015 9/11/2015 9/11/2015 9/12/2015 9/14/2015 9/16/2015 1/28/2016 1/28/2016 2/18/2016 3/15/2016 3/1/2016 3/1/2016 3/1/2016 3/3/2016 3/3/2016 3/4/2016 3/7/2016 3/8/2016 3/10/2016 3/10/2016 3/10/2016 3/11/2016 3/11/2016 3/14/206 3/14/2016 UMT Easting 243847 243861 243861 244219 242544 246530 239245 239992 246530 246530 243944 238871 246530 244608 242766 238209 239245 244608 249044 243215 249035 249065 248536 248536 248536 240613 240613 239003 236722 236340 236340 236340 252235 252235 256994 248987 249773 240429 240429 240429 247170 247170 257188 257188 UTM Northing 4122598 4122743 4122743 4122743 4128370 4135648 4128553 4128359 4135648 4135648 4134850 4126931 4135648 4125554 4133049 4130562 4128553 4125554 4131886 4128740 4131895 4133012 4139267 4139267 4139267 4125119 4125119 4137869 4133888 4132122 4132122 4132122 4138922 4138922 4140745 4140248 4129514 4104602 4104602 4104602 4134166 4134166 4134879 4134879 14 Sex Estimated Age M 3 M 12 M 2 M 3 M 2 M 5 M 8 M 10 M 10 M 1 M 2 F 5 F 8 M 4 F 4 M 10 M 1 M 1 M 1 M 2 M 3 M 1 F 6 F 12 M 10 M 1 F 1 M 1 F 1 M cub F cub F cub F cub M cub F cub M cub M cub M cub M cub M cub M cub M cub F cub F cub Weight (kg) 74.8 199.6 40.4 64.0 52.6 97.1 119.7 122.5 152.4 35.8 56.7 60.8 78.9 74.8 61.2 131.1 46.7 42.6 34.0 43.1 62.1 41.3 80.7 73.5 137.9 40.8 36.3 30.8 11.8 1.1 1.2 1.2 2.3 2.4 0.8 1.6 3.2 2.3 1.9 2.4 2.2 2.5 2.9 2.8 �B548 B549 B550 B551 B552 B553 B554 B555 B556 B557 3/16/2016 3/16/2016 3/17/2016 3/17/2016 3/17/2016 3/18/2016 3/18/2016 3/19/2016 3/19/2016 3/19/2016 245703 245703 252210 252210 252210 235030 235030 763412 763412 763412 4141023 4141023 4132171 4132171 4132171 4150681 4150681 4133906 4133906 4133906 M M F F M F M M M F cub cub cub cub cub cub cub cub cub cub 1.8 1.8 2.9 2.3 2.3 1.6 1.8 2.9 2.9 2.6 �Figure 1. Number of weekly black bear captures from May 15th through September 15th during the 2011 through 2015 summer trapping seasons. Note: trapping did not commence until July in 2014 and 2015. Number of Bear Captures 25 ..-·•·. 20 2011 2012 2013 2014 2015 j\ I \ I \. I 15 10 5 0 0 2 4 6 8 10 12 Week (May 15th through Sept 15th) 14 16 �Figure 2. GPS collar locations from 46 adult female black bears collected during 1 January – 31 December 2015 in the vicinity of Durango, Colorado (different colored clusters of points represent different individual bears). ~~ 0 ol - 00 0 00 0 0 0 ¢ () Q --~ ..o .....i • 0 0 ,.. • -.· ...... ..-1 . .•.••• ;'.- ~,-.. ,j'10 • • •• • • 6:·~ ' 0 0, r~,,, , . •e,:,,ffib!d ,U1tte ♦ O 0 0 0. (;,.J. i j ,., 0 N Q L JP1 8:.t*'O 1 A :D'l 1 Ill 0 3 6 12 Kilometers I sou t h111 n L!J " r,..-.h ,." n .. .. .., , t• •n• lrfbaX 17 �Figure 3. Mean abundance of soft and hard mast observed on vegetation transects from 2011-2015. Mast species included gamble oak, chokecherry, serviceberry, native crabapple and pinyon pine. Abundance reflects the proportion of plants observed with mast. Gamble Oak Choke Cherry g Service Berry g g -, I I I I I I I I lil lil lil lil -, I I .,. I I 8 -r BT .□ -'- 0 2011 2013 -'- 2013 2015 year Pinyon Native Crab Apple lil - -, -, I I .., o· i ~□~□~ I 2011 I lil - I 2013 year I I 2015 1s: o - I I I I I I -, I ~-□□B I 2011 I I 2013 year 18 I I 2015 I I I I I ~□ 2011 2013 year lil - I I I I I -, BT 0 year -, 0 • -'- 2011 .,. -'I g _ lil - - I I I I I 0 2015 g - 1s: - 1s: I I I I I I lil iI gi LJ I 0 1s: -, -, I I I I I I .,. 0 0 -, 2015 �Figure 4. Model averaged density estimates based on integrated spatially-explicit capture-mark-recapture models (solid lines; using both hair-snare and GPS collar data) and standard spatially-explicit capturemark-recapture models (dashed lines; using hair-snare data only) for female black bears near Durango, Colorado, USA from 2011 to 2014. Density 0.5 0.4 cu 0.3 , iii E z; 1/) , w 0.2 , , , ,, , ~ . . .. ,, ', ' • I I I ~ \ \ ~ \ '\ 'I 1 r· -, I 0 .1 l 0.0 2011 2012 2013 Year 19 201 �Figure 5. Predicted density surfaces for female black bears/km2 near Durango, Colorado, USA from 2011 to 2014. Surfaces were derived from year-specific model-averaged estimates. 2011 0 g ... ;"' .,.+ + + + ..."':::! i-T 230000 + Oura.,,go + 0 0 + + + + + + -+ • g + + + + 0 0 0 + ++ 8 "' "':;; 2012 + -1- + + + + + t J0,O.9.8 0 .7 ,6 80.4.5 0 ,3 8? 0,0 ... 0 8 "' "';; 250000 260000 230000 240000 2013 0 + ++ "';!;... ++ + + + + + + + + + 0 8 'il ;; + + + 80 + !!... + .. Durango + g t + + + + + l LO .9 80.7.8 - g:i.4 0 ,3 800,2, 1 ++ + + + "'...;!; - + - -+ 240000 260000 0 8 ..."'p 0 250000 - - 0 !!... 0 260000 + ++ ..+ '-8 0. 0.8 0.7 ++ + + + + + + + + + + + + + Ourangi + +I• ++ + + + + ++ + I 230000 250000 2014 0 8 1.0 80.7:t 80.4.5,6 0.3 80,0.? • g 0 :i1 ... + 240000 + + + + + ++ + + + + + + + + + + + + + + + +Dutaogo t + + + + + ..+ + + + + "' ;!; 230000 I I 240000 I I 250000 I 80.4:i 0,3 800,2, 1 I 260000 Figure 6. Garbage-related black bear-human conflicts observed during July through September 2015. Red lines indicate treatment areas and black lines indicate control areas. Green circles represent conflicts with �regular residential garbage containers and purple circles represent conflicts with wildlife-resistant containers. e;,.f ..~.JQ,'>- 1 .,.-f .~ ~ oj t:,.. ~, '?~-- 3 M ~.,e1\\~r,tJ~, ~'JJra-s, . ~ 'll x; ,...,, , I I Garbage Container Type o Residential Container 0 Wildlife Resistant Container f- StudyArea D Treatment □ control N ----- 0 0.25 0.5 Kilometers ,,,, A Sot.rce-;; Esri. H ERE,. Q8'L,✓ me. USGS, ln-iermap, increment P Corp., NRCAN. E.sri-Japan, METI, Esri China {Hong Kcr.g) , Esr i (Thailand). Mapmylndis. ® O~nSlle<!tMap conllil>llto,;; ano th~ GIS Use, Community Figure 7. ‘Hot spots’ of black-bear human trash conflicts pre- and post-distribution of bear-resistant trash containers in Durango, Colorado. All residents in treatment areas (outlined in red) were given bearresistant trash containers in 2013; residents in the control areas (outlined in black) did not receive bear- �resistant containers. Pre-treatment data were collected 2011-2012, and post-treatment data were collected 2013-2015. Hot spots were identified as those areas with the highest probabilites of conflict from kernal density functions of all observed trash conflicts. Pre-Treatment Post-Treatment Treatment Control Treatment Control �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Black bear exploitation of urban environments: finding management solutions and assessing regional population effects Period Covered: July 1, 2016 - June 30, 2017 Principal Investigator: Heather E. Johnson, heatherjohnson@usgs.gov Project Collaborators: S.A. Lischka, S. Breck, J. Beckmann, J. Apker, K. Wilson, and P. Dorsey All information in this project summary is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the principal investigator. Manipulation of these data beyond that contained in this summary is discouraged. Across the country conflicts among people and black bears are increasing in frequency and severity, and have become a high priority wildlife management issue. Whether increases in conflicts reflect recent changes in bear population trends or bear behavioral shifts to anthropogenic food resources, is largely unknown, with key implications for bear management. This issue has generated a pressing need for bear research in Colorado and has resulted in a collaborative study involving Colorado Parks and Wildlife (CPW; lead agency), the USDA National Wildlife Research Center, Wildlife Conservation Society and Colorado State University. Collectively, we have designed and implemented a study on black bears that 1) determines the influence of urban environments on bear behavior and demography, 2) tests a management strategy for reducing bear-human conflicts, 3) examines public attitudes and behaviors related to bear-human interactions, and 4) develops population and habitat models to support the sustainable monitoring and management of bears in Colorado. Field data collection for this project was initiated spring 2011 and completed spring 2016. Several publications from this work are in various stages of analyses, peer-review and publication. Publications in progress and published abstracts are listed below: Publications in Progress: Laufenberg, J., H.E. Johnson, S. Breck, and P. Doherty. Using integrated population models to understand spatio-temporal dynamics in Colorado black bear populations. In Preparation for Ecological Applications. Kirby, R., H.E. Johnson, M.W. Alldredge, and J.N. Pauli. The tension between foraging and hibernation shapes biological aging in bears. In Preparation for Journal ofAnimal Ecology. Lischka, S., T. Teel, H. E. Johnson, S. Breck, and K. Crooks. Factors associated with public compliance of wildlife ordinances. In Preparation for Journal of Wildlife Management. Johnson, H.E., S.W. Breck, and D.L. Lewis. The effects of human development on black bear survival and fecundity. In Preparation for Journal ofAnimal Ecology. Lischka, S. T. Teel, H.E. Johnson, S. Breck, and K. Crooks. What drives real and perceived risk of human-wildlife conflict? In Preparation for Human Dimensions of Wildlife. 28 �Johnson, H.E., D.L. Lewis, S. Lischka, and S. W. Breck. Bear-resistant containers reduce human-black bear conflicts and improve public perceptions. Journal of Wildlife Management, In Press. Wibur, R.C., S.A. Lischka, J.R. Young and H.E. Johnson. 2017. Experience, attitudes and demographic factors influence the probability ofreporting human-black bear interactions. Wildlife Society Bulletin, In Press. Published Abstracts: Shifting perceptions of risk and reward: Dynamic selection for human development by black bears in the western United States H.E. Johnson1, S.W. Breck2, S. Baruch-Mordo3, D.L. Lewis4, C.W. Lackey5, K.R. Wilson4, J. Broderick6, J.S. Mao', J.P. Beckmann 8 1 Colorado Parks and Wildlife, 415 Turner Drive, Durango, CO 81303, USA USDA-Wildlife Services, National Wildlife Research Center, 4101 La Porte Ave, Fort Collins, CO 80521, USA 3 The Nature Conservancy, 117 E Mountain Ave, Suite 20 I, Fort Collins, CO 80524, USA 4 Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA 5Nevada Department of Wildlife, 2788 Esaw Street, Minden, NV 89423, USA 6 Colorado Parks and Wildlife, 317 West Prospect Road, Fort Collins, CO 80526, USA 7 Colorado Parks and Wildlife, 0088 Wildlife Way, Glenwood Springs, CO 81601, USA 8 Wildlife Conservation Society, 301 North Willson Ave, Bozeman, MT 59715, USA 2 Citation: Johnson, H. E., Breck, S. W., Baruch-Mordo, S., Lewis, D. L., Lackey, C. W., Wilson, K. R., Broderick, J., Mao, J. S., & Beckmann, J. P. 2015. Shifting perceptions of risk and reward: Dynamic selection for human development by black bears in the western United States. Biological Conservation 178: 164-172. Abstract As landscapes across the globe experience increasing human development, it is critical to identify the behavioral responses of wildlife to this change given associated shifts in resource availability and risk from human activity. This is particularly important for large carnivores as their interactions with people are often a source of conflict, which can impede conservation efforts and require extensive management. To examine the adaptations of a large carnivore to benefits and risks associated with human development we investigated black bear behavior in three systems in the western United States. Our objectives were to (I) identify temporal patterns of selection for development within a year and across years based on natural food conditions, (2) compare spatial patterns of selection for development across systems, and (3) examine individual characteristics associated with increased selection for development. Using mixed effects resource selection models we found that bear selection for development was highly dynamic, varying as a function of changing environmental and physiological conditions. Bears increased their use of development in years when natural foods were scarce, throughout the summer-fall, as they aged, and as a function of gender, with males exhibiting greater use of development. While patterns were similar across systems, bears at sites with poorer quality habitat selected development more consistently than bears at sites with higher quality habitat. Black bears appear to use development largely for food subsidy, suggesting that conflicts with bears, and potentially other large carnivores, will increase when the physiological demand for resources outweighs risks associated with human activity. 29 �Human development and climate affect hibernation in a large carnivore with implications for human-carnivore conflicts Heather E. Johnson1, David L. Lewis1, Tana L. Verzuh1, Cody F. Wallace1, Rebecca M. Much1, Lyle K. Willmarth 1, Stewart W. Breck2 1 Colorado Parks and Wildlife, Durango CO, USA USDA National Wildlife Research Center, Fort Collins, CO, USA 2 Citation: Johnson, H. E., D. L. Lewis, T. L. Verzuh, C. F. Wallace, R. M. Much, L. K. Willmarth and S. W. Breck. 2017. Human development and climate affect hibernation in a large carnivore with implications for human-carnivore conflicts. Journal of Applied Ecology, DOI:10.l l l l/1365-2664.13021 Abstract 1. Expanding human development and climate change are dramatically altering habitat conditions for wildlife. While the initial response of wildlife to changing environmental conditions is typically a shift in behaviour, little is known about the effects of these stressors on hibernation behaviour, an important lifehistory trait that can subsequently affect animal physiology, demography, interspecific interactions and human-wildlife interactions. Given future trajectories of land use and climate change, it is important that wildlife professionals understand how animals that hibernate are adapting to altered landscape conditions so that management activities can be appropriately tailored. 2. We investigated the influence ofhuman development and weather on hibernation in black bears (Ursus americanus), a species of high management concern, whose behaviour is strongly tied to natural food availability, anthropogenic foods around development and variation in annual weather conditions. Using GPS collar data from 131 den events of adult female bears (n = 51 ), we employed fine-scale, animalspecific habitat information to evaluate the relative and cumulative influence of natural food availability, anthropogenic food and weather on the start, duration and end of hibernation. 3. We found that weather and food availability (both natural and human) additively shaped black bear hibernation behaviour. Of the habitat variables we examined, warmer temperatures were most strongly associated with denning chronology, reducing the duration of hibernation and expediting emergence in the spring. Bears appeared to respond to natural and anthropogenic foods similarly, as more natural foods, and greater use of human foods around development, both postponed hibernation in the fall and decreased its duration. 4. Synthesis and applications. Warmer temperatures and use of anthropogenic food subsides additively reduced black bear hibernation, suggesting that future changes in climate and land use may further alter bear behaviour and increase the length of their active season. We speculate that longer active periods for bears will result in subsequent increases in human-bear conflicts and human-caused bear mortalities. These metrics are commonly used by wildlife agencies to index trends in bear populations, but have the potential to be misleading when bear behaviour dynamically adapts to changing environmental conditions, and should be substituted with reliable demographic methods. 30 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Black bear exploitation of urban environments: finding management solutions and assessing regional population effects Period Covered: July I, 2017-June 30, 2018 Principal Investigator: Heather E. Johnson, heatherjohnson@usgs.gov Project Collaborators: S.A. Lischka, S. Breck, J. Beckmann, J. Apker, K. Wilson, and P. Dorsey Across the country conflicts among people and black bears are increasing in frequency and severity, and have become a high priority wildlife management issue. Whether increases in conflicts reflect recent changes in bear population trends or bear behavioral shifts to anthropogenic food resources, is largely unknown, with key implications for bear management. This issue has generated a pressing need for bear research in Colorado and has resulted in a collaborative study involving Colorado Parks and Wildlife (CPW; lead agency), the U.S. Department of Agriculture (USDA) National Wildlife Research Center, Wildlife Conservation Society and Colorado State University. Collectively, we have designed and implemented a study on black bears that I) determines the influence of urban environments on bear behavior and demography, 2) tests a management strategy for reducing bear-human conflicts, 3) examines public attitudes and behaviors related to bear-human interactions, and 4) develops population and habitat models to support the sustainable monitoring and management of bears in Colorado. Field data collection for this project was initiated spring 2011 and completed spring 2016. Several publications from this work are in various stages of analyses, peer-review and publication. Publications in progress and abstracts from published manuscripts during the reporting period are listed below: Publications in Progress: Kirby, R., H.E. Johnson, M.W. Alldredge, and J.N. Pauli. The tension between foraging and hibernation shapes biological aging in bears. In Review with Scientific Reports. Lischka, S., T. Teel, H.E. Johnson, S. Breck, and K. Crooks. Factors associated with public compliance of wildlife ordinances. In Preparation for Journal of Wildlife Management. Johnson, H.E., S.W. Breck, and D.L. Lewis. The effects of human development on black bear survival and fecundity. In Preparation for Journal ofAnimal Ecology. Published Abstracts: Human development and climate affect hibernation in a large carnivore with implications for human-carnivore conflicts Heather E. Johnson', David L. Lewis', Tana L. Verzuh 1, Cody F. Wallace', Rebecca M. Much 1, Lyle K. Willmarth', Stewart W. Breck2 1 Colorado Parks and Wildlife, Durango CO, USA USDA National Wildlife Research Center, Fon Collins, CO, USA 2 29 �Citation: Johnson. H. E., D. L. Lewis, T. L. Vcrzuh, C. F. Wallace, R. M. Much, L. K. Willmarth and S. W. Breck. 2017. Human development and climate affect hibernation in a large carnivore with implications for human-carnivore conflicts. Journal of Applied Ecology, DOl:10.I I I l/1365-2664.13021 Abstract 1. Expanding human development and climate change are dramatically altering habitat conditions for wildlife. While the initial response of wildlife to changing environmental conditions is typically a shift in behaviour, little is known about the effects of these stressors on hibernation behaviour, an important lifehistory trait that can subsequently affect animal physiology, demography, interspecific interactions and human-wildlife interactions. Given future trajectories of land use and climate change, it is important that wildlife professionals understand how animals that hibernate are adapting to altered landscape conditions so that management activities can be appropriately tailored. 2. We investigated the influence of human development and weather on hibernation in black bears ( Ursus americanus), a species of high management concern, whose behaviour is strongly tied to natural food availability, anthropogenic foods around development and variation in annual weather conditions. Using GPS collar data from 131 den events of adult female bears (n = 51 ), we employed fine-scale, animalspecific habitat information to evaluate the relative and cumulative influence of natural food availability, anthropogenic food and weather on the start, duration and end of hibernation. 3. We found that weather and food availability (both natural and human) additively shaped black bear hibernation behaviour. Of the habitat variables we examined, warmer temperatures were most strongly associated with denning chronology, reducing the duration of hibernation and expediting emergence in the spring. Bears appeared to respond to natural and anthropogenic foods similarly, as more natural foods, and greater use of human foods around development, both postponed hibernation in the fall and decreased its duration. 4. Synthesis and applications. Warmer temperatures and use of anthropogenic food subsides additively reduced black bear hibernation, suggesting that future changes in climate and land use may further alter bear behaviour and increase the length of their active season. We speculate that longer active periods for bears will result in subsequent increases in human-bear conflicts and human-caused bear mortalities. These metrics are commonly used by wildlife agencies to index trends in bear populations, but have the potential to be misleading when bear behaviour dynamically adapts to changing environmental conditions, and should be substituted with reliable demographic methods. Experience, attitudes, and demographic factors influencing the probability of reporting human-black bear interactions Ryan C. Wilbert, Stacy A. Lischka 2, Jessica R. Young 1, Heather E. Johnson 3 'Western State Colorado University, 600 N. Adams St., Gunnison, CO 81231, USA 2Colorado Parks and Wildlife, 317 W. Prospect Ave., Fort Collins, CO 80526, USA 3Colorado Parks and Wildlife, 415 Turner Drive, Durango, CO 81303, USA Citation: Wilber, R. C., S. A. Lischka, J. R. Young and H. E. Johnson. 2018. Experience, attitudes, and demographic factors influencing the probability of reporting human-black bear interactions. Wildlife Society Bulletin; DOI: 10.1002/wsb.854 ABSTRACT Interactions between people and American black bears (Ursus americanus) have been increasing throughout the United States, with negative interactions becoming a major management challenge for wildlife agencies. To monitor the number, location, and severity of these conflicts, wildlife agencies typically rely on voluntary public reports. Although trends in voluntary reports are commonly assumed to reflect actual trends in human-bear interactions, recent research suggests an individual's likelihood ofreporting interactions may be biased, influenced by attitudes toward the species and its management, previous experiences with wildlife, or demographic factors. During 2012, we used a mail survey of residents in the vicinity of Durango, Colorado, USA, (n = 1,667) to explore the relative importance of tolerance for black bears, satisfaction with bear management, personal experience with 30 �bears, and demographic traits as predictors of a resident's decision to report interactions to the authorities. We found that residents' experiences with bears were most important in predicting reporting behavior, followed closely by attitudes related to tolerance for bears, and satisfaction with management; demographic factors had relatively little influence. Respondents were more likely to report when they had seen black bears near their homes, had been threatened by bears, were intolerant of bears, dissatisfied with management, and were female. Although several variables in our analyses were influential in explaining reporting behavior, the overall predictive power of our models was low (R 2 = 0.17), suggesting future investigations of reporting behavior should include a broader set of covariates. Our results indicate that public reports represent a biased measure of human-bear interactions, and management agencies should either account for bias, or collect different types of interaction data, when assessing patterns of bear activity. Compounding effects of human development and a natural food shortage on a black bear population along a human development-wildland interface Jared S. Laufenberg0 , Heather E. Johnsonh, Paul F. Doherty Jr.a, Stewart W. Breckc 3 Department of Fish, Wildlife, and Conservation Biology, Colorado State University, Fort Collins, CO 80523, USA bColorado Parks and Wildlife, 415 Turner Drive, Durango, CO 82303, USA cusDA-Wildlife Services, National Wildlife Research Center, 4101 La Porte Ave., Fort Collins, CO 80521, USA Citation: Laufenberg, J. S., H. E. Johnson, P. F. Doherty Jr., and S. W. Breck.2018. Compounding cfTects of human development and a natural food shortage on a black bear population along a human-wildland interfact. Biological Conservation 224:188-198; doi.org/l0.1016/j.biocon.2018.05.004 Abstract Human development and climate change are two stressors that threaten numerous wildlife populations, and their combined effects are likely to be most pronounced along the human development-wildland interface where changes in both natural and anthropogenic conditions interact to affect wildlife. To better understand the compounding influence of these stressors, we investigated the effects of a climate-induced natural food shortage on the dynamics of a black bear population in the vicinity of Durango, Colorado. We integrated 4 years of DNA based capture-mark-recapture data with OPS-based telemetry data to evaluate the combined effects of human development and the food shortage on the abundance, population growth rate, and spatial distribution offemale black bears. We documented a 57% decline in female bear abundance immediately following the natural food shortage coinciding with an increase in human-caused bear mortality (e.g., vehicle collisions, harvest and lethal removals) primarily in developed areas. We also detected a change in the spatial distribution of female bears with fewer bears occurring near human development in years immediately following the food shortage, likely as a consequence of high mortality near human infrastructure during the food shortage. Given expected future increases in human development and climate-induced food shortages, we expect that bear dynamics may be increasingly influenced by human-caused mortality, which will be difficult to detect with current management practices. To ensure long-term sustainability of bear populations, we recommend that wildlife agencies invest in monitoring programs that can accurately track bear populations, incorporate non-harvest humancaused mortality into management models, and work to reduce human-caused mortality, particularly in years with natural food shortages. Assessing ecological and social outcomes of a bear-proofing experiment Heather E. Johnson', David L. Lewis', Stacy A. Lischka2, Stewart W. Breck3 1 Colorado Parks and Wildlife, 415 Turner Drive, Durango, CO 81303, USA 2Colorado Parks and Wildlife, 317 W. Prospect Ave., Fort Collins, CO 80526, USA 3 U. S. Department of Agriculture National Wildlife Research Center, 4101 LaPorte, Ave., Fort Collins, CO 80521, USA 31 �Citation: Johnson, M. E., D. L. Lewis, S. A. Lischka, and S. W. Breck. 2018. Assessing ecological and social outcomes ofa bearproofing experiment. Journal of Wildlife Management 82: 1102-1114; DOI: I 0.1002/jwmg.21472 ABSTRACT Human-black bear conflicts within urban environments have been increasing throughout North America, becoming a high priority management issue. The main factor influencing these conflicts is black bears foraging on anthropogenic foods within areas of human development, primarily on residential garbage. Wildlife professionals have advocated for increased bear-proofing measures to decrease the accessibility of garbage to bears, but little research has been conducted to empirically test the effectiveness of this approach for reducing conflicts. Between 2011 and 2016, we conducted a beforeafter-control-impact experiment in Durango, Colorado where we distributed 1,110 bear-resistant trash containers, enhanced education, and increased enforcement to residents in 2 treatment areas, and monitored 2 paired control areas. We examined the ecological and social outcomes of this experiment, assessing whether bear-resistant containers were effective at reducing conflicts; the level of public compliance (i.e., properly locking away garbage) needed to reduce conflicts; whether the effectiveness of bear-resistant containers increased over time; and if the distribution of bear-resistant containers changed residents' attitudes about bear management, support for ordinances that require bear-proofing, or perceptions of their future risk of garbage-related conflicts. After the bear-resistant containers were deployed, trash-related conflicts (i.e., observations of strewn trash) were 60% lower in treatment areas than control areas, resident compliance with local wildlife ordinances (properly locking away trash) was 39% higher in treatment areas than control areas, and the effectiveness of the new containers was immediate. Conflicts declined as resident compliance with wildlife ordinances increased to approximately 60% (by using a bear-resistant container or locking trash in a secure location), with minor additional declines in conflicts at higher levels of compliance. In addition to these ecological benefits, public mail surveys demonstrated that the deployment of bear-resistant containers was associated with increases in the perceived quality of bear management and support for ordinances that require bear-proofing, and declines in the perceived risk of future trash-related conflicts. Our results validate efforts by wildlife professionals and municipalities to reduce black bear access to human foods, and should encourage other entities of the merits of bear-proofing efforts for reducing human-bear conflicts and improving public attitudes about bears and their management. A conceptual model for the integration of social and ecological information to understand human-wildlife interactions Stacy A. Lischka"· h, Tara L. Teele, Heather E. Johnsond, Sarah E. Reedb,e, Stewart Breck', Andrew Don CarlosC, Kevin R. Crooks., Research, Policy, and Planning Branch, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526, USA hDcpartment ofFish, Wildlife, and Conservation Biology, Colorado State University, Fort Collins, CO 80523, USA cDepartmcnt of Murnan Dimentions of Natural Resources, Colorado State University, Fort Collins, CO 80523, USA dResearch, Policy, and Planning Branch, Colorado Parks and Wildlife, 415 Turner Dr., Durango, CO 81303, USA eAmericas Program, Wildlife Conservation Society, 2300 Southern Blvd., Bronx, NY I 0460, USA i'National Wildlife Research Center, USDA Wildlife Services, 4101 Laporte Ave., Fort Collins, CO 80521, USA 3 Citation: Lischka, S. A., T. L. Teel, II. E. Johnson, S. E. Reed, S. Breck, A. D. Carlos, and K. R. Crooks. 2018. A conceptual model for the integration of social and ecological information to understand human-wildlife interactions. Biological Conservation 225:80-87: doi.org/l0.1016/j.biocon.2018.06.020 Abstract There is growing recognition that interdisciplinary approaches that account for both ecological and social processes are necessary to successfully address human-wildlife interactions. However, such approaches are hindered by challenges in aligning data types, communicating across disciplines, and applying social science information to conservation actions. To meet these challenges, we propose a conceptual model that adopts a social-ecological systems approach and integrates social and ecological theory to identify the 32 �multiple, nested levels of influence on both human and animal behavior. By accounting for a diverse array of influences and feedback mechanisms between social and ecological systems, this model fulfills a need for approaches that treat social and ecological processes with equal depth and facilitates a comprehensive understanding of the drivers of human and animal behaviors that perpetuate human-wildlife interactions. We apply this conceptual model to our work on human-black bear conflicts in Colorado, USA to demonstrate its utility. Using this example, we identify key lessons and offer guidance to researchers and conservation practitioners for applying integrated approaches to other human-wildlife systems. 33 � https://cpw.cvlcollections.org/files/original/79dd3c63e5f783d321c9deca307a6ef3.pdf e6a8b673ffb6692fa88b8b75b5230b7c PDF Text Text Colorado Division of Parks and Wildlife July 2010–June 2011 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 0638 N/A Federal Aid Project No. N/A : : : : Division of Parks and Wildlife Mammals Research Wolverine Conservation Assessing the efficacy of monitoring wolverine on a regional scale using occupancy and abundance estimation Period Covered: July 1, 2010 – June 30, 2011 Author: J. S. Ivan Personnel: M. Schwartz, USFS Rocky Mountain Research Station; M. Ellis, University of Montana All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT The wolverine (Gulo gulo) has a circumpolar distribution comprised mostly of tundra and boreal forest. However, its current range extends southward in peninsular fashion to the Cascades and Rocky Mountains of the conterminous United States. Recently the U.S. Fish and Wildlife Service ruled that the North American wolverine in the contiguous U. S. is a candidate species for protection under the Endangered Species Act. Thus, there is considerable interest in identifying monitoring schemes capable of detecting declines in wolverine populations over a large scale. We used spatially explicit simulations in which wolverine were sampled on a virtual landscape to quantify our ability to detect declines using robust-design occupancy estimation. We systematically varied 1) the number of sample units surveyed, 2) the number of visits made to each unit in the sample, and 3) the rate of population decline and computed the power to detect declines under various scenarios. Initial results indicate that occupancy estimation may work well for detecting large declines (50% decline over 10 years), but power to detect less catastrophic declines was low. Approximately 100 sample units would need to be surveyed to have adequate power to detect a 50% decline over 10 years. A census (350 sample unit) would be needed to ensure decent power for detecting smaller declines. Power increases as number of visits to each sample unit increases from 2 to 3 per survey season, but making more than 3 visits does not increase power substantially. If confronted with design tradeoffs that lead to having a better detection probability vs. those that allow for more units to be sampled, it is better to increase detection probability and survey fewer units. Future simulations will address the power to detect increases in population size in addition to declines, and we will attempt to compare power to detect declines using abundance estimation with that obtained using occupancy estimation. 1 �WILDLIFE RESEARCH REPORT ASSESSING THE EFFICACY OF MONITORING WOLVERINE ON A REGIONAL SCALE USING OCCUPANCY AND ABUNDANCE ESTIMATION. JACOB S. IVAN P. N. OBJECTIVE Assess power for detecting trends in wolverine population growth using occupancy and abundance estimation. SEGMENT OBJECTIVES 1. Build code to simulate realistic distribution and space use of wolverine on the landscape. 2. Build code to realistically simulate sampling the wolverine population using an occupancy framework. 3. Build code to analyze data “collected” via occupancy surveys. 4. Summarize results of 100s of iterations of randomly generated wolverine distributions and subsequent occupancy surveys; plot power to detect trends against various scenarios intended to reflect the range of conditions expected for both the sampling and process portions of the simulation. INTRODUCTION The wolverine (Gulo gulo) has a circumpolar distribution comprised mostly of tundra and boreal forest. However, its current range also extends southward in peninsular fashion to the Cascades and Rocky Mountains of the conterminous United States. Recently the U.S. Fish and Wildlife Service ruled that the North American wolverine in the contiguous U. S. was a candidate for protection under the Endangered Species Act (U.S. Fish and Wildlife Service 2010). Therefore, considerable interest exists in identifying monitoring schemes capable of detecting declines in wolverine populations over a large scale. Colorado Parks and Wildlife (CPW) has expressed interest in potentially pursuing a wolverine reintroduction, and monitoring program would be an integral part of such an effort. Additionally, with minor modifications, the simulation approach outlined here could be used to inform current Canada lynx (Lynx canadensis) monitoring efforts in Colorado. Thus, the work described here holds benefits for wolverine conservation in general as well as current and future CPW projects. Estimating abundance or occupancy are 2 means around which a monitoring scheme for wolverines could be constructed. Within these general approaches, there are numerous sampling methods that could be employed in the field. For instance, individual identification necessary for abundance estimation can be obtained from pelage patterns (Royle et al. 2011), scat samples (Flagstad et al. 2004, Ulizio et al. 2006), hair snags (Mulders et al. 2007), or a combination of methods (Magoun et al. 2011). Similarly, occupancy information can be obtained via aerial track surveys (Magoun et al. 2007, Gardner et al. 2010), remote cameras (R. Inman, Wildlife Conservation Society, unpublished data) or any genetic sampling technique. In all cases, the models used to estimate abundance and/or occupancy are the same; field methods only change the probability of detecting (and potentially identifying an individual(s) and the cost of obtaining those detections. Our aim was to use simulation to generically estimate power for detecting population declines of interest in the Northern Rockies. Simulations are spatially explicit, sampling occurs randomly and we are currently using robust design occupancy models to look at power. Here we report only on our initial simulations using occupancy estimation. 2 �METHODS Simulated landscape and wolverine distribution All simulations were programmed in R (R Core Development Team 2011), with calls to C++ (Stroustrup 1997), RMARK (Laake and Rexstad 2011), and MARK (White and Burnham 1999) as necessary. The simulation landscape included Idaho, western Montana, and northwest Wyoming (Figure 1). We overlaid this landscape with a raster dataset depicting “persistent spring snow” as this layer adequately captures the bioclimatic niche of wolverines (Copeland et al. 2010). Each 500-m pixel in the raster could take values 1 to 7 depending on the number of years from 2000-2006 that snow was present between April 24 and May 15 in that pixel. At the beginning of each iteration of the simulation, we randomly dispersed home range centers across the landscape subject to the following constraints based on wolverine ecology (Figure 2): 1) 2) 3) 4) Home range centers (points) were required to fall within the spring snow layer. Male home range centers were required to be >12.5 km apart. Female home range centers were required to be >8.5 km apart. Female home range centers could fall within male buffers, and transient males could fall within resident male or female buffers. Once home range centers were distributed, we temporarily assigned each animal a bivariate normal utilization distribution scaled to match UD estimates from the literature. To impart more realism in these UDs, we multiplied the bivariate normal kernel for each animal by the underlying spring snow layer, then divided each pixel value in the resulting product by the total of all values for that animal to recreate a probability distribution. Functionally this process produces a center-weighted UD in which mass is piled up over pixels with higher values of persistent spring snow. Each animal’s UD was different depending on the underlying configuration of spring snow. We began each simulation with 200 males, 200 females, and 100 transients for a total of 500 wolverines in the Northern Rockies landscape. Our simulated population size was based on available wolverine abundance information and expert opinion. We then simulated a 10%, 20%, or 50% decline in this population over 10 years by randomly removing individuals from the landscape at each time step. Simulated Sampling To simulate collection of occupancy data, we overlaid a sampling grid of 225km2-cells (n = 385 total cells) across the landscape. This cell size corresponds roughly to the home range size of female wolverine. At the beginning of each year, we computed the probability of at least 1 wolverine being available to sample in each cell on any given occasion for each cell in sampling grid: p(2: 1 wolverine present in cell J) = 1 - [(}c1 - p (individual; is present ))] where w = total number of wolverines in the simulation. For each visit within a given year, we drew a random uniform number (i.e., U(0,1)) and compared this number to the product: p(≥1 wolverine available)p(wolverine detected | available). If the random number was less than this product, wolverine were detected in that cell on that visit (occasion) and we entered a “1” in the encounter history for that cell-occasion. Otherwise, we entered a “0.” We proceeded to sample in this manner for each visit to each cell for each year of the simulation. This results in a vector of 0s and 1s (i.e., an encounter history) for each cell that is 10x in length where “x” is the number of visits made during each of 10 years. For each unique landscape and declining wolverine population, we created several different datasets using this general sampling process. We specified detection probability, p(wolverine detected | available), to be 3 �either 0.2 or 0.8 and specified the number of visit to each cell in a year to be 2, 3, 4, 5, 6, or 7. This results in 2 × 6 = 12 datasets for each simulated population decline. We also considered the situation in which surveys could only be accomplished every other year, which resulted in another 12 datasets in which no data were collected during even years. Analysis of simulated data For each simulated dataset we used the R (R Development Core Team 2011) package RMARK (Laake and Rexstad 2011) to construct a robust design occupancy model (MacKenzie et al. 2006, p. 183224) for fitting in program MARK(White and Burnham 1999). We allowed the occupancy (use) parameter (ψt) as well as colonization (γt) and extinction (εt) to vary through time in an unconstrained manner, but constrained detection probability (p) to be constant to reflect how it was simulated. This resulted in 10 estimates of probability of occupancy, or use, from each dataset. We then fit a random effects trend model to these 10 data points (also using the RMARK interface for MARK to account for covariance between estimates; Figure 4), and retained the slope of the trend line along with 95% confidence interval for that slope. When the 95% confidence interval for the slope of the trend line did not include zero, we considered a trend detected, otherwise a trend was not detected. The number of times a trend was detected out of the total simulations is an estimate of the power of the approach to identify the specified declines given the number of visits and detection probability specified. RESULTS As expected, initial results indicate that occupancy estimation should work well for detecting large declines (50% decline over 10 years, λ = 0.933) when detection probability is high (p = 0.8). Under these conditions, power was 80% when sampling 50 units, regardless of the number of visits, and approached 100% when sampling 100 units (Figure 5, “continuous sampling” panels). Power declined some, but was still respectable, even when detection probability was low (p = 0.2). In that case a power of 0.8 could be achieved with 4-6 visits to 100 sample units. Power to detect a 20% decline over 10 years (λ = 0.977) was diminished, however, especially when detection probability was low. For instance, in order to achieve 80% power, even with high detection probability, would require surveys in an estimated 300 sample units. There is no realistic chance of detecting minor declines (e.g., 10% over 10 years, λ = 0.989) using occupancy estimation (Figure 5). Not surprisingly, power declines when sampling occurs every other year rather than annually (Figure 5, “gap sampling” panels). However, if detection probability is high, adequate power (0.8) can be achieved to detect a 50% decline over 10 years if such a scheme is implemented in a reasonable number of sample units (100), even with as few as 2-3 visits. Ability to detect smaller declines (20% or 10% over 10 years) is poor regardless of detection probability, number of sample units or number of visits (Figure 5, “gap sampling” panels). Generally, we found that when detection probability is high, power increases as number of visits to each sample unit increases from 2 to 3 per survey season, but making more than 3 visits does not increase power substantially. However, when detection probability is low, gains can be realized by making more visits. This result re-confirms a well-documented phenomenon unique to occupancy estimation (MacKenzie et al. 2006, p. 168). Also, if confronted with design tradeoffs that lead to having a better detection probability vs. those that allow for more units to be sampled, it is always better to increase detection probability and survey fewer units. DISCUSSION Our initial simulations suggest that occupancy estimation may work well in a monitoring context if the survey techniques employed have relatively high detection probability and interest lies only in 4 �detecting sharp declines in the population. Future work on this project will focus on determining the effects of varying the size of sample units, using alternate starting population sizes, detecting increasing trends rather than decreasing, and making sure that detection and occupancy estimates match well with recently collected pilot data (R. Inman, unpublished data). Additionally, we will incorporate cost functions into the modeling effort and investigate how well occupancy estimation compares to abundance estimation, which can be accomplished by sampling with hare snares or by photographing unique throat patch patterns via remote camera LITERATURE CITED Aubry, K. B., G. M. Koehler, and J. R. Squires. 2000. Ecology of Canada lynx in southern boreal forests. Pages 373-396 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, andJ. R. Squires, editors. Ecology and conservation of lynx in the United States. Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Collins, Colorado, USA. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd edition. Springer, New York. Copeland, J. P., K. S. McKelvey, K. B. Aubry, A. Landa, J. Persson, R. M. Inman, J. Krebs, E. Lofroth, H. Golden, J. R. Squires, A. Magoun, M. K. Schwartz, J. Wilmot, C. L. Copeland, R. E. Yates, I. Kojola, and R. May. 2010. The bioclimatic envelope of the wolverine (Gulo gulo): do climatic constraints limit its geographic distribution? Canadian Journal of Zoology-Revue Canadienne De Zoologie 88:233-246. Devineau, O., T. M. Shenk, G. C. White, P. F. Doherty, P. M. Lukacs, and R. H. Kahn. 2010. Evaluating the Canada lynx reintroduction programme in Colorado: patterns in mortality. Journal of Applied Ecology 47:524-531. Dolbeer, R. A., and W. R. Clark. 1975. Population ecology of snowshoe hares in the central Rocky Mountains. Journal of Wildlife Management 39:535-549. Flagstad, O., E. Hedmark, A. Landa, H. Broseth, J. Persson, R. Andersen, P. Segerstrom, and H. Ellegren. 2004. Colonization history and noninvasive monitoring of a reestablished wolverine population. Conservation Biology 18:676-688. Gardner, C. L., J. P. Lawler, J. M. Ver Hoef, A. J. Magoun, and K. A. Kellie. 2010. Coarse-Scale Distribution Surveys and Occurrence Probability Modeling for Wolverine in Interior Alaska. Journal of Wildlife Management 74:1894-1903. Getz, W. M., S. Fortmann-Roe, P. C. Cross, A. J. Lyons, S. J. Ryan, and C. C. Wilmers. 2007. LoCoH: Nonparameteric Kernel Methods for Constructing Home Ranges and Utilization Distributions. Plos One 2. Getz, W. M., and C. C. Wilmers. 2004. A local nearest-neighbor convex-hull construction of home ranges and utilization distributions. Ecography 27:489-505. Hodges, K. E. 2000a. The ecology of snowshoe hares in northern boreal forests. Pages 117-161 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, andJ. R. Squires, editors. Ecology and conservation of lynx in the United States. University Press of Colorado, Boulder, Colorado, USA. _____. 2000b. Ecology of snowshoe hares in southern boreal and montane forests. Pages 163-206 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, andJ. R. Squires, editors. Ecology and conservation of lynx in the United States. University Press of Colorado, Boulder, Colorado, USA. Ivan, J. S. 2011. Density, demography, and seasonal Movement of snowshoe hares in central Colorado. Dissertation, Colorado State University, Fort Collins, Colorado, USA. Laake, J. L., and E. Rexstad. 2011. RMark - an alternative approach to building linear models in MARK. in E. Cooch, andG. C. White, editors. Program MARK: A gentle introduction. 5 �MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. Academic Press, Oxford, UK. Magoun, A. J., C. D. Long, M. K. Schwartz, K. L. Pilgrim, R. E. Lowell, and P. Valkenburg. 2011. Integrating Motion-Detection Cameras and Hair Snags for Wolverine Identification. Journal of Wildlife Management 75:731-739. Magoun, A. J., J. C. Ray, D. S. Johnson, P. Valkenburg, F. N. Dawson, and J. Bowman. 2007. Modeling wolverine occurrence using aerial surveys of tracks in snow. Journal of Wildlife Management 71:2221-2229. McKelvey, K. S., K. B. Aubry, and Y. K. Ortega. 2000. History and distribution of lynx in the contiguous United States. Pages 207-264 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, andJ. R. Squires, editors. Ecology and conservation of lynx in the United States. University Press of Colorado, Boulder, Colorado, USA. Mulders, R., J. Boulanger, and D. Paetkau. 2007. Estimation of population size for wolverines Gulo gulo at Daring Lake, Northwest Territories, using DNA based mark-recapture methods. Wildlife Biology 13:38-51. Pebesma, E. J. 2004. Multivariable geostatistics in S: the gstat package. Computers & Geosciences 30:683-691. Royle, J. A., A. J. Magoun, B. Gardner, P. Valkenburg, and R. E. Lowell. 2011. Density Estimation in a Wolverine Population Using Spatial Capture-Recapture Models. Journal of Wildlife Management 75:604-611. Ruediger, B., J. Claar, S. Gniadek, B. Holt, L. Lyle, S. Mighton, B. Naney, G. Patton, T. Rinaldi, J. Trick, A. Vendehey, F. Wahl, N. Warren, D. Wenger, and A. Williamson. 2000. Canada lynx conservation assessment and strategy. 2nd edition. R1-00-53, U.S. Department of Agriculture, Forest Service, U.S. Department of Interior, Fish and Wildlife Service, Bureau of Land Management, National Park Service, Missoula, Montana, USA. Service, U. S. F. a. W. 2010. Endangered and threatened wildlife and plants: 12-month finding on a petition to list the North American wolverine as endangered or threatened. Federal Register. Shenk, T. M., and R. H. Kahn. 2010. The Colorado lynx reintroduction program. Colorado Division of Wildlife. Stroustrup, B. 1997. The C++ Programming Language. 3rd edition. Addison Wesley Longman, Reading, MA, USA. Team, R. D. C. 2011. R Foundation for Statistical Computing, Vienna, Austria. Theobald, D. M., and T. M. Shenk. 2011. Areas of high habitat use from 1999-2010 for radio-collared Canada lynx reintroduced to Colorado. Colorado State University. Ulizio, T. J., J. R. Squires, D. H. Pletscher, M. K. Schwartz, J. J. Claar, and L. F. Ruggiero. 2006. The efficacy of obtaining genetic-based identifications from putative wolverine snow tracks. Wildlife Society Bulletin 34:1326-1332. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 Supplement:120-138. Yee, T. W. 2010. The VGAM package for categorical data analysis. Journal of Statistical Software 32:134. _____. 2011. Zahratka, J. L., and T. M. Shenk. 2008. Population estimates of snowshoe hares in the southern Rocky Mountains. Journal of Wildlife Management 72:906-912. Zuur, A. F., E. N. Ieno, N. J. Walker, A. A. Saveliev, and G. M. Smith. 2009. Mixed Effects Models and Extensions in Ecology with R. Springer, New York, New York, USA. Prepared by ______________________________________ Jake S. Ivan, Wildlife Researcher 6 �Figure 1. Study area for simulation including montane regions of Idaho, western Montana, and northwest Wyoming. Black polygons indicate primary wolverine habitat defined as areas with snowcover between April 24 and May 15 during at least 1 year from 2000-2006. 7 �Females Resident Males Transient Males Figure 2. Example distribution of home range centers for male, female, and transient wolverines on the virtual landscape. Home range centers were required to fall within the spring snow layer, and intrasexual territorialty was enforced, except for transient individuals. The buffer around male home range centers was 12.5 km; female buffers were 8.5 km. 8 �Bivariate Normal UD x Persistent Snow Layer = Modified UD Figure 3. Simulated utilization distributions (UDs) for each individual were created by positioning a bivariate normal UD directly over each home range center (see Figure 2) then multiplying by the underlying persistent snow layer to form a modified, more realistic UD unique to each individual. 1 ] 0.8 V, ::I C 0 ·.g 0 .6 Q. ...0 Q. ] ....ro 0.4 E ·.;::; ~ 0 .2 0 1 2 3 4 5 6 7 8 9 10 Year Figure 4. Example output from a single simulation: estimates of occupancy over a 10-year period fitted with a linear random effects model. If the 95% confidence interval on the slope of the linear trend did not include zero, then we concluded that a trend had been detected. The percentage of iterations in which trends were detected out of the total iterations provided a measure of power. 9 �1 = 0.933 1 = 0.977 1 = 0.989 080 § 0.6C: ~ o 4(1) • II visits £ (J) - 2 - 3 C: 2 02- 0 © D E :::, ~ - 4 - 5 - G 7 (I) C: detP ~ o.su -0.2 (I) - - 0.8 2u m 0.60 0.4- 0.2- I I I I I I I 50 100 150 200 250 300 350 50 100 150 200 250 I I I I I 300 350 50 100 150 I I 200 250 I I 300 350 Number of grid eels sampled Figure 5. Power to detect population declines of 50% (λ=0.933), 20% (λ=0.977), and 10% (λ=0.989) using occupancy estimation. Curves represent 2 levels of detection probability (0.2 and 0.8) and varying number of visits annually to a sampled unit (2, 3, 4, 5, 6, 7). Top 3 panels depict estimates of power when occupancy surveys occur annually; bottom 3 panels depict power when surveys are conducted biannually. Note that the lowest power to detect a 50% decline with annual sampling is apparently realized with 7 visits to each sampling unit. This result is counterintuitive, and likely due to a coding error. It will be addressed in future simulations. 10 �Colorado Division of Parks and Wildlife July 2011–June 2012 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Colorado 3430 0638 N/A Federal Aid Project No. N/A : : : : Division of Parks and Wildlife Mammals Research Wolverine Conservation Assessing the efficacy of monitoring wolverine on a regional scale using occupancy and abundance estimation Period Covered: July 1, 2011 – June 30, 2012 Author: J. S. Ivan* Personnel: M. Schwartz, USFS Rocky Mountain Research Station; M. Ellis, University of Montana *(J. S. Ivan was the sole Colorado Parks and Wildlife contributor for this work and is thus listed as “author.” However, the draft manuscript included here was a collaborative effort and all personnel listed are co-authors on the manuscript. M. Ellis is the first author.) All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Conservation biologists and resource managers are often faced with the task of designing monitoring programs for species that are rare, diffuse, or patchily distributed across large landscapes. These efforts are frequently very expensive and seldom can be conducted by one entity. It is essential that a power analysis is undertaken to ensure stated goals are feasible. We developed a spatial-based simulation, which accounts for natural history, habitat use, and sampling scheme, to investigate power for monitoring wolverines in two areas of the U.S. Rocky Mountains. The first area is a well-established metapopulation of wolverine in the northern Rocky Mountain states of Montana, Idaho, and Wyoming, where the current population is approximately 350 individuals and there are concerns of population decline. Based on current population size estimates and detection probabilities in the northern U.S. Rockies, most sampling schemes are likely to only detect large declines in population sizes (i.e. 50% decline over 10 years). In general, increasing the number of grids sampled or the per visit detection probability had a much greater effect on power than increasing the number of visits per year. For small populations, we found very low power to detect declines. The second analysis was a forecast of the effort required to monitor an increasing population in the southern U.S. Rockies, given recolonization or reintroduction. Occupancy-based methods can only produce enough power to detect population trends if populations are increasing dramatically (i.e. doubling or tripling in 10 years), regardless of the sampling effort. In sum, our approach provides a spatially based framework to evaluate monitoring protocols and objectives by explicitly incorporating the link between changes in population size and estimated occupancy, all while accounting for natural history of the species in question. These analyses were specific to wolverines, but our approach could easily be adapted to other species. 6 �WILDLIFE RESEARCH REPORT ASSESSING THE EFFICACY OF MONITORING WOLVERINE ON A REGIONAL SCALE USING OCCUPANCY AND ABUNDANCE ESTIMATION. JACOB S. IVAN P. N. OBJECTIVE Assess power for detecting trends in wolverine population growth using occupancy. SEGMENT OBJECTIVES 1. Build code to simulate realistic distribution and space use of wolverine on the landscape. 2. Build code to realistically simulate sampling the wolverine population using an occupancy framework. 3. Build code to analyze data “collected” via occupancy surveys. 4. Summarize results of 1000s of iterations of randomly generated wolverine distributions and subsequent occupancy surveys; plot power to detect trends against various scenarios intended to reflect the range of conditions expected for both the sampling and process portions of the simulation. 5. Prepare manuscript for publication INTRODUCTION Wildlife populations worldwide have faced major population reductions in abundance and geographic range due to both natural and anthropogenic causes (Butchart et al. 2010, Hoffmann et al. 2010, Rands et al. 2010, Inman et al. 2011). Currently, many populations are facing multiple threats including habitat fragmentation and loss, climate change, direct and indirect exploitation, disease, invasive species, and the interaction among these threats (Primack 2006, Laurance et al. 2008, Povilitis and Suckling 2010). Responding to these major threats to wildlife and fish populations worldwide, many countries have adopted legislations aimed at affording protection to species of conservation concern (Hutchins et al. In Press, Waples et al. In Review). Two of the more powerful pieces of legislation are Canada’s Species at Risk Act (SARA) and the United States’ Endangered Species Act (ESA). These acts not only identify species at risk and aim to protect them from additional harm, but also stipulate and provide mechanisms for recovery. For example, in the United States approximately half of the annual budget spent on threatened and endangered species is designated for recovery (GAO 2005, Male and Bean 2005). However, determining when a species of concern is declining or subsequently recovering requires information about trend. The majority of studies that have examined trends in fish and wildlife were historically based in assessments of population abundance (Dennis et al. 1991, Bart et al. 2007, Foster et al. 2009, Broms et al. 2010). While estimates of abundance are important, other measures such as changes in genetic or demographic parameters or changes in geographic range size have been used to infer trend (Gaston 1991, Schwartz et al. 2007, Marucco et al. 2009, Broms et al. 2010). Recently, more attention has been placed on estimating changes in occupancy of a species geographic range (Joseph et al. 2006, MacKenzie et al. 2006). Occupancy estimation generally requires multiple surveys to a set of sample units, noting on each survey whether the species of interest was detected or not. Subsequently, these repeat-visit data are used to estimate the probability of detecting the species of interest if it was present, and then adjusting the raw presence-absence data in light of this probability to estimate the proportion of area occupied (MacKenzie 7 �et al. 2006). If occupancy estimation is conducted over multiple time intervals, trend in occupancy is obtained (Field et al. 2005, MacKenzie 2005, Marsh and Trenham 2008). Before launching an occupancy study, power analysis should be conducted to allocate monitoring effort efficiently (Field et al. 2005, MacKenzie 2005, Rhodes et al. 2006). Most studies base power analyses for occupancy estimation on detecting declines in occupancy over time; however, these simulations rarely consider spatial dynamics. Also, monitoring trends in occupancy is often used as a surrogate for trends in abundance, but this link is rarely evaluated (e.g. Field et al. 2005, Finley et al. 2005, Otto and Roloff 2011). Rhodes et al. (2006) and Rhodes and Jonzén (2011) modeled spatial and temporal correlations in population dynamics to account for spatial structure in populations and provide allocation recommendations in occupancy studies. They find, when spatial correlation is low and temporal correlation is high, it is most efficient to sample many sites infrequently. In the opposite situation, when spatial correlation among population dynamics is high and temporal correlation is low, they recommend sampling few sites often. Finally, when there is a decoupling of abundance and space, they suggest maximizing spatial replication (Rhodes and Jonzén 2011). Furthermore, if interest is in detecting declines in occupancy, they suggest sampling high quality habitats, whereas if the objective is to detect an increase, sampling intermediate-quality habitats is the best strategy. We extended their work by building a species-specific model of a population changing over time. We then sampled from this population using a multi-season occupancy framework to determine power to detect population trends under various scenarios. This approach allows us to optimally allocate scarce monitoring resources for designing an occupancy-based monitoring effort. Our model was designed to optimize sampling allocation for a large-scale wolverine monitoring effort. Wolverines are a Holarctic carnivore species known for their large home ranges, low densities, and occasional long distance movements (Lofroth and Krebs 2007, Squires et al. 2007, Inman et al. 2012). The species is currently under consideration for listing under the ESA (USFWS 2010) largely due to the fact that their numbers were greatly reduced (possibly eliminated) in the contiguous United States in the early 20th century. Wolverine populations have recolonized Idaho, Montana, Washington, and Wyoming and single male wolverines have recently dispersed to California and Colorado (Aubry et al. 2007, Moriarty et al. 2009). Yet, they are still absent from significant portions of their historical range and their current abundance in the contiguous United States is still likely to be at most 500 individuals. Recent research by Aubry et al. (2007) and Copeland et al. (2011) has shown that the historical distribution of wolverine was consistent with the distribution of persistent spring snow. Copeland et al. (2011) characterized persistent spring snow cover in the entire northern hemisphere based on a 21-day composite (24 April–15 May) of images from 2000-2006 at a 0.5km2 resolution using moderate resolution imaging spectroradiometer (MODIS) satellite images (Hall et al. 2006). They found that >99% of wolverine den sites and >89% year-round telemetry locations were located within areas that were classified as having persistent spring snow in at least one of the seven years for which data were available. Schwartz et al. (2009) demonstrated that wolverine gene flow was facilitated by areas with persistent spring snow compared to areas that were snow free. In this paper we use habitat (i.e., persistent spring snow), movement, and home range data to build a spatially based model for assessing the power for monitoring wolverine in their current range and in areas where they may eventually recolonize either naturally or through reintroduction. METHODS Study area There are two study areas for this project. The primary study area consists of the U.S. Rocky Mountains in northern and central Idaho, western Montana, and northwest Wyoming (“Northern 8 �Rockies”, Figure 1). The area is composed of individual mountain ranges each characterized by high alpine areas (maximum elevation 3900 m) and surrounded by wide areas of semiarid grasslands and irrigated agriculture (elevation ~1400 m). This area is known to be occupied by wolverines, with current population estimates ranging from 200-500 individuals (USFWS 2010). We removed from our analyses mountain ranges on the edge of this range, including the Wallowa Mountains of Eastern Oregon, the Bighorn Mountains of Eastern Montana and Wyoming, and the Bear River Range on the Idaho/Utah border; all three of which have no historical records of wolverines (Aubry et al. 2007) and do not contain continuous patches of persistent spring snow cover (Schwartz et al. 2009, Copeland et al. 2011). We allowed areas ‘used’ by simulated wolverines to extend up to 50 km into Alberta and British Columbia, Canada to account for continuous wolverine populations in the Northern Rockies, but excluded these areas from occupancy analyses. The second study area is the mountainous region of the Southern U.S. Rockies (“Southern Rockies”). This area is characterized by high, steep mountains (max elevation 4,400 m). As a result, there are strong gradients in physical attributes of the landscape, which lead to heavily dissected vegetation types. In the Southern Rockies, alpine and subalpine zones can be relatively narrow and give way to montane forests, montane shrublands, and semiarid grassland or sagebrush communities over relatively short distances. This area does not currently have a population of wolverines, although wolverines are thought to have occurred there historically (Aubry et al. 2007), and there seems to be adequate habitat, including persistent spring snow (Aubry et al. 2007, McKelvey et al. 2011). Areas of persistent spring snow are more patchily distributed in the Southern Rockies landscape, and separated from areas of persistent spring snow in the Northern U.S. Rockies by >200km. Most mountain ranges in this study area occur in Colorado, but we included the Medicine Bow and Sierra Madre ranges in southern Wyoming, as well as the southern San Juan Mountains in northern New Mexico. Individual Utilization Distributions We randomly selected points within areas of persistent spring snow (using Copeland et al. 2010) for the center of individual home ranges for adult female, adult male, and transient male wolverines. Among these three groups, locations were chosen independently to allow for overlapping home ranges (Copeland 1996, Inman et al. 2011); however, within each group, selection of home range centers was constrained to reflect territoriality. The buffer distances required between home ranges centers were at least 16 km for adult females, reflecting a 225 km2 home range, and at least 25.2 km for adult and transient males, reflecting 500 km2 home ranges (Banci 1994, Krebs et al. 2007, Schwartz et al. 2009). We also required that all home range centers were located in snow patches large enough to support at least one resident female wolverine (Krebs et al. 2007). Within each group (adult females, adult/resident males, adult/transient males), locations for home range centers were randomly selected in an iterative fashion until no additional individuals could be placed in the landscape or until the desired number of individuals was met. Once home range centers were established for a given simulated landscape, we assigned a bivariate normal utilization distribution for each individual. For resident females, we assumed that an individual spends 90% of her time within their 225 km2 home range (radius = 8.5 km). For resident males, we assumed individuals spend 90% of their time within their 12.6 km home range radius, but we allowed for larger sizes and greater overlap among transient male home ranges by assuming individuals only spend 70% of their time in the original 500 km2 home range. Each of these distributions produced a surface with decreasing probability of use with increasing distance from the home range center. To make these bivariate normal utilization distributions more realistic, we overlaid them on the persistent spring snow layer and multiplied the layers together. In the persistent spring snow layer, areas of non-snow were weighted as having 1/20 the probability of use compared to snow areas, based on resistance values found for models of genetic least cost paths (Schwartz et al. 2009). We standardized the product of the 9 �two layers to transform it back into a probability density. Thus, each individual utilization distribution takes a unique shape based on availability of snow. In this approach, it is possible for individuals to make short term, long distance movements during a given study period. The tails of the bivariate normal utilization distribution allow for a very small, but non-zero, probability of reaching any point on the landscape. In preliminary analyses, we tested for the effect of excluding these long distance movement events by cutting off the tails of the bivariate normal, such that the probability of an individual being more than 1-2 standard deviations away from its home range center was set to 0, compared to a situation with no limit on movement. Although allowing short-term, long-distance movements did affect the estimated occupancy of the landscape, the effect on power was minor. Occasional long-distance movements are possible in wolverine ecology, especially by males and transients (Moriarty et al. 2009). For territorial males and females, we would expect these movements to be less likely over the course of the relatively short survey period. Thus we based our power analyses on a ‘mixed’ scenario in which long distance movements were possible for transient males (i.e. no limit), resident males were allowed some larger movement events (limited to within 2 s.d. of home range center), and movements of females, which may have dens, were limited 1 sd from their home range center. Following the rules state above, our program, SPACE (Spatially-based Power Analyses for Conservation and Ecology), created 1000 surfaces for N=500 or N=200 individuals on the Northern Rockies landscape, reflecting high and low estimates of wolverine population size in the study area. We then simulated 10%, 20%, or 50% declines in our simulated populations over a decade (λ = 0.989, 0.977, 0.933) by randomly removing an appropriate number of individuals at each time step. We also simulated scenarios (nsim = 1000) for a hypothetical reintroduced or recolonizing population in the Southern Rockies. These populations were started with N=30 individuals then allowed to increase by 50%, 100%, or 200% over a decade (λ = 1.041, 1.072, 1.0116). We initiated all populations with a 2:1:2 ratio of females:resident males:transient males. Sampling To estimate occupancy, we sampled from our simulated landscapes during each time step or “year” of the simulation. We divided the study area into 225km2 sample units (cells), matching home range sizes for resident females, a strategy widely used for monitoring carnivores (e.g., Zielinski and Stauffer 1996). We excluded cells that did not overlap the persistent snow layer by ≥50%. This resulted in 388 cells for the main Northern Rockies study region, and 128 cells for the Southern Rockies. For each cell, the probability of at least one wolverine being present (hereafter, ‘probability of presence’ was Eqn 3): N   Ρ(# wolverines ≥ 1) j = 1 − Ρ( wolverines absent ) j = 1 − ∏ 1 − ∫∫ f i ( x, y ) dx dy    ( x , y )∈Ω i =1 j   where N is the number of wolverines in the simulated study area, fi(x,y) is the probability density function (i.e., utilization distribution) describing the use surface for the ith wolverine, and Ωj represents the area included in the jth grid. We approximated integral values by summing pixel values in the raster, assuming equal pixel areas. To construct a simulated encounter history (i.e., the data necessary for occupancy estimation) for cell j in year k, we assigned a 1 (present) or 0 (absent) for each visit by comparing a random draw from Uniform (0,1) with the probability of presence for that cell (draws less than the probability of presence resulted in a detection, and a 1 in the encounter history for that visit). Thus, a cell with simulated encounter history “010” indicates that 3 visits were made to the cell in a given year, and wolverines were 10 �detected on the second visit only. After initial construction, we used progressively reduced versions of the encounter histories to explore the effect of changes in parameters associated with sampling on power to detect population changes. For example, we omitted data from even numbered years (i.e., inserted “.” for each “0” or “1” of the omitted years) to examine the effect of sampling every other year; we tested the effects of smaller sample sizes by reducing the number of cells or visits included in the encounter histories; and we reduced the number of detections to simulate imperfect detection (See Table 1). To create encounter histories with lower detection probability, we randomly removed an appropriate proportion of 1s from each encounter history. Thus, to go from a detection probability of 1.0 to 0.8, we retained 0.8/1.0 = 80% of the 1s; for each 1 (wolverine detected) in a given encounter history, we conducted a random draw from uniform (0,1) and compared this draw against 0.8. We retained the 1 if the draw was ≤0.8, and changed it to a 0 (wolverine not detected) otherwise. Similarly, to go from encounter histories reflecting detection probability = 0.8 to detection probability = 0.2, we evaluated each 1 in a given history, retaining it if the random draw was ≤0.25 (0.2/0.8), changing it to 0 otherwise. We used these encounter histories to obtain annual estimates of occupancy and detection probability for each simulated landscape and parameter set. Note that the subject of our simulations is a mobile carnivore capable of moving freely between sample cells, and our simulation setup reflected this reality. Therefore, interpretation of estimated occupancy parameters was different than the usual context in which the status (occupied or not) of a given cell is assumed static over the course of a survey. Specifically, the estimate of occupancy (Ψ) generated under this context is the probability that any given cell is used rather than occupied, and any reference to Ψ or “occupancy” from here forward refers to probability of use. Furthermore, the estimate of detection probability generated in this context is actually the product of true detection probability (i.e., probability of detection given that the species of interest is present; this quantity is specified directly for any given simulation) and a landscape-wide probability that an individual is present and available for detection (i.e., probability of presence; see above). We refer to the detection probability estimated by the model as pest, and the actual detection probability specified for the simulations as psim, such that pest = psim × probability of presence. We used the R (R Development Core Team 2011) package RMark to input the encounter histories and construct models to fit in Program MARK (White and Burnham 1999). Specifically, we employed the ‘Robust Design Occupancy‘ data type (MacKenzie et al. 2006) in which colonization (γ) could vary through time but was constrained to be the complement of extinction (ε; i.e., changes in occupancy were considered random rather than Markovian or static) and detection probability (p) varied with time. This model structure is appropriate because: 1) we were interested primarily in the occupancy estimates themselves; we had no interest in modeling occupancy dynamics (colonization, extinction) explicitly, 2) the simulation specifications allowed “movement” in and out of adjacent cells, thus mimicking random changes in occupancy, and 3) “movement” between adjacent cells forced pest to be a function of probability of presence, which changed through time depending on the simulated landscape and birth/death of individuals. Thus, pest should have varied through time as well. We extracted the 10 occupancy estimates and the variance-covariance matrix for these estimates from each simulation, then used the variance components procedure in RMark to fit a linear random effects trend model to the estimates. A trend was ‘detected’ if the 95% confidence interval of the trend parameter (on the logit scale) from the random effects model excluded zero and was in the correct direction (e.g., <0 for declining trends; Tallmon et al. 2010). Thus, we computed the statistical power produced by a sampling scenario, i.e. the probability that we detect a significant trend given that there is a trend in the underlying data, as the percentage of simulations in which a trend was detected. For datasets in which we simulated sampling every other year, we fit models in which we fixed γ =γ , γ =γ4, etc. such that the product of these parameters were estimated, and we bridged years in which no data were collected to produce valid estimates of occupancy for those years where data were collected. 1 2 3 11 �In these scenarios, only 5 occupancy estimates were generated, and we fit random effects models to those 5 estimates. We repeated these analyses for each combination of population growth or decline, simulated detection probability (psim), number of visits, cell size, number of cells sampled, and annual or alternating year sampling schemes that were applied to the 1000 simulated landscapes of N = 30, 200, or 500 (Table 1). Where applicable, all sampling was cumulative to facilitate the most meaningful contrasts between levels of a parameter; for example, a sample of n = 50 cells would include the same cells as an n = 25 sample with 25 additional cells included. Similarly, an encounter history with 4 visits would include the same string of 0s and 1s as a 3-visit history, with one additional visit included. Our simulations were designed to be generalizations in that we do not attempt to define when a sampling season might begin, what the sampling mechanism was, or what constitutes a visit. Thus, these simulations could represent flying over selected cells in the study area to search for tracks in the snow, in which case a ‘visit’ is a single flight (Gardner et al. 2010), or they could reflect the use of hair snag devices in which a ‘visit’ is 1 month of continuous sampling (Magoun et al. 2011). We bracketed the sampling parameters (cell size, detection probability, visits) based on previous efforts described in the literature (Magoun et al. 2007, Gardner et al. 2010, Magoun et al. 2011). RESULTS Effects of home range parameters Due to the spacing rules among individuals that we used to reflect wolverine territoriality, the Northern Rockies landscape becomes saturated with approximately 850 individuals (420 ± 6 females, 219 ± 4 resident males, 219 ± 4 transient males; mean ± s.d. across 100 simulated landscapes). For N=800, the median probability of at least one wolverine per cell across the landscape was 0.47. This value reflects the availability of individuals on the landscape, yielding on average 280.4 cells in which wolverine were available for detection per sampling occasion across the 388 cells in the grid. As the population size decreased, the average probability of at least one wolverine per cell fell to 0.74 (212.4 detections per occasion) for N=500 and 0.05 (18.9 detections per occasion) for N=30 across simulations. With perfect detection associated with sampling (psim = 1), these cell-based probabilities for use translate to an estimated occupancy (Ψ) of 0.99±0.01 for the entire landscape for populations with N = 500 individuals and 0.06±0.01 for N = 30. Effects of population size and trend We investigated the upper limits of power with occupancy estimation by examining the ability to detect trends when the simulated detection probability was perfect (psim = 1) and with a large number of visits (5) to each unit. We focused these analyses on the U.S. Northern Rockies landscape and a quickly declining population (λ = 0.933). Even with perfect detection and intense sampling, detecting a large decline (50% over 10 years) in a large starting population (N = 500) with adequate power (>80% chance of detecting the trend) required a sample of 50 out of 388 cells (Figure 2). As the population size decreased, the amount of sampling needed to detect a 50% decline even under this ‘best case’ scenario with perfect detection increased dramatically. For example, when N=200, achieving 80% power required sampling approximately 75 to 150 cells. Detecting trends in small populations (N=30) was difficult; even if we included the entire grid (388 cells) in the sample and assumed perfect detection, we had less than 70% power to detect a trend. Regardless of the starting sample size, power to detect trends was lower for increasing populations compared to the decreasing scenarios described above. For example, to detect a 50% increase (λ = 1.041) with >80% confidence, the amount of the total sampling grid that would need to be included in the sample increased to ~25% of the grid (ncells≈60) for N=500 or ~ 50% (ncells≈125) for 12 �N=200. For N=30, sampling the entire grid, assuming perfect detection probability, and with an intense sampling effort (5 visits), we were able to detect a 50% increase in <40% of the simulations. With current population sizes (N=500) in the Northern Rockies, the ability to detect declines fell dramatically as the strength of the decline decreased (Figure 3). We found a reasonable chance ( ≥80%) of detecting a 50% decline in population size over a 10-year period, depending on the combination of sample size and detection probability. However, for a 10% decline in population size over the 10 year period, no amount of sampling could yield enough power to detect the trend. Similarly, even with a large sample size and high detection probability, a 20% decline was detected in <60% of the simulations (Figure 3). With either population increases or declines, sampling every other year substantially increased the number of cells and visits that would need to be sampled. Trade-offs in sampling methodology After the strength of the population decline or increase, the parameter that most influences power to detect change was the simulation detection probability (psim). In nearly all scenarios relatively large gains in power were realized when psim increased from 0.2 to 0.8. For instance, a monitoring scheme that called for 2 visits to each of 100 sample units would have ~25% chance of detecting a 50% decline over 10 years when psim = 0.2. Power for detecting that same decline under the same sampling regime increased to 80% when psim = 0.8 (Figure 3, upper left panel). By comparison, an increase in sample size from ncells = 50 to ncells = 300 resulted in only a doubling in power (25% to ~50%). In fact, when psim = 0.2, 80% power cannot be achieved even if the entire grid is sampled. Similar gains in power relative to simulation detection probability and sample were realized in other scenarios we simulated. The exceptions to this result were when the goal was to detect a 10% decline over 10 years or to detect a 20% decline when sampling was only conducted every other year. Both scenarios yield very low power and negligible improvement with increased psim or sample size (Figure 3, middle panels). The number of visits to each sample unit influenced power as well, although generally to a lesser degree than magnitude of population change, simulation detection probability, and sample size. Even with perfect simulation detection probability (psim = 1), the power to detect a trend increased with the number of visits at each grid cell due to the number of opportunities for an individual to be present. When simulation detection probability is high but imperfect (i.e., psim = 0.8), some gain in power could be realized by visiting each sampled cell 3 times vs. visiting them only twice (Figure 3, separation between the two lightest dotted lines). However, the gain realized for making 4 visits rather than 3 is small, and there is no appreciable difference in power for 4, 5, 6, or 7 visits under the scenarios we simulated. When simulated detection probability was low (i.e., psim = 0.2), potentially greater gains in power could be realized by making more visits, but it depends on the scenario (Figure 3, in some cases there is a moderate amount of separation in the solid lines, in other cases there is not). Note that at low detection probabilities (psim = 0.2), it is often inadvisable to make more visits to each sampled cell because such an approach actually decreases power (See discussion). Effect of Cell Size In order to achieve the threshold of 80% power to detect a 50% population decline, changing cell sizes in the grid had implications for both the number of cells and the total area that would need to be sampled. (Figure 5). Grids of 100km2 and 225km2 cells yielded similar power in terms of the percent of the grid that would need to be included in the sample, although the smaller cell size requires sampling more cells (i.e., the total grids were comprised of 887 100km2 cells versus 388 225km2 cells). Assuming 3 visits and high detection, getting 80% power for detecting a 50% decline required 120 cells (12,000km2) from the small grid versus 70 cells (15,750km2) for the medium sized grid. As the size of the grid increased, the power to detect trends in occupancy decreased. The 1000km2 grid produced very low power to detect population trends. In this case, the grid in the Northern Rockies comprised only 76 cells. Including every cell in the sample, with seven visits and high detection probability, we detected a 50% 13 �population decline in <20% of the runs. The phenomenon in which power is actually reduced with a high number of visits occurs for the 225km2 cell size at low psim, and for the 500km2 and 1000km2 size at high psim. Power to detect increases in small populations For small populations (N=30), power for detecting population trends was limited except for situations with large population increases and high detection probability (Figure 4). For the purposes of comparison, there was greater power for detecting trends in the Southern Rockies landscape than in the Northern Rockies, although the total sampling area in the Southern Rockies landscape is approximately only a third of the Northern Rockies. For both landscapes, a doubling of the population over ten years (λ = 1.072) could be detected with >80% power in scenarios where a large proportion of the landscape was included with relatively high capture probability. If simulation detection probability is low, then adequate power can only be achieved via sampling a large portion of the available landscape, and making a large number (≥5) of visits to each sampled cell. DISCUSSION Monitoring population trends over time is one of the most common goals for management of endangered species. Using a spatially explicit simulation for wolverine in the U.S. Rocky Mountains, we were able to test the ability of occupancy-based approaches to detect trends in population size under a range of monitoring scenarios. Even for large changes in population size (e.g. 50% declines over 10 years), we found that detecting population trends required large-scale, intensive sampling. In many scenarios, no amount of sampling could produce sufficient power to achieve monitoring goals. Our results highlight the importance of analyzing the statistical power of monitoring schemes and using approaches that incorporate the effect of sampling and power over the course of multiple steps in a monitoring protocol. In the case of the wolverine, work has commenced to evaluate the effectiveness of various approaches for detecting presence. These range from using fix-winged aircraft to find tracks in 100-km2 (Magoun et al. 2007) or 1000-km2 (Gardner et al. 2010) sampling cells, to using cameras at bait stations (Mulders et al. 2007, Magoun et al. 2011), to the use of non-invasive genetic sampling (Ulizio et al. 2006, Schwartz and Monfort 2008, Magoun et al. 2011). These efforts produce varying detection probabilities from 0.2 to 0.8 as bracketed in our simulations. However, matching estimates from field studies to our results, is not straightforward. It is important to note that detection probability estimated from pilot analysesis not the same as the psim input in our analyses. Due to the ‘mobile animal’ phenomenon, animals are capable of moving freely between sample cells and therefore can be detected in multiple cells during one sampling occasion. As a result, occupancy models cannot separate the effects of true detection probability (psim) and probability of presence (See Methods). Consequently, pest returned from pilot studies will be smaller than the detection probabilities used in our simulations (psim). For example, if pilot work indicates that pest = 0.2, power can be assumed to be slightly better than the curves shown for psim = 0.2 in our figures. The exact correspondence between pest and psim is dependent on cell size, population size, and home range size of the species in question. Thus, no rule of thumb holds for converting between the two. However, matching pest derived from pilot work to curves for psim can still be useful as it will result in conservative estimates of power, which would be a prudent way to design monitoring schemes. In the case of wolverines, pilot work specific to occupancy monitoring in the Northern Rockies has been carried out using camera stations (B. Inman, Wildlife Conservation Society, unpublished data) and hair snags (J. Waller, Glacier National Park, unpublished data) in 100-km2 sample units. Initial results from this work suggest pest is approximately 0.25 – 0.3, which in our simulations corresponded to 14 �psim ≈0.8 (i.e., pest = psim × probability of presence, where our mean probability of presence was 0.33; thus 0.25/0.33 ≈ 0.8). It’s important to note that the mean probability of presence depends on assumptions about the number of animals, the landscape, and home range configurations. Based on this estimate, and assuming 3-4 visits to each sample unit (sampling occurred during 3-4 months over winter for each pilot study), our research suggests that roughly 100-150 100-km2 cells would need to be sampled per year to attain an 80% probability of detecting a 50% decline in the Northern Rockies population (Figure 5). Thus, intensive sampling over a small area is unlikely to be a viable solution for detecting population trends. To accomplish anything meaningful, monitoring will require well-coordinated surveys across multiple entities and jurisdictions. Anything less than a large-scale, coordinated effort will likely be of limited or no value. The spatially explicit nature of our approach is especially important in linking changes in occupancy to population trends. Our results demonstrate that the underlying landscape can influence power to detect population changes. Specifically, in the comparison of power for populations with N=30 in the Northern versus Southern Rockies, power to detect trends in occupancy was similar in terms of percent of the total study area included in the sample, but very different in terms of the absolute area that needs to be sampled. For example, to detect a 3x increase of the N=30 populations with a 225km2 grid and >80% power required sampling ~20% of either landscape, which translates to sampling 16000km2 in the northern landscape versus 6000km2 in the south. Note, however, that the scenarios in this comparison, populations of N=30 in the Northern versus Southern U.S. Rockies, are intended to illustrate the effect of underlying landscape for a fixed population size. In reality, changing the size of a study area would generally also change the size of the population included, which we found to substantially affect power to detect trends. Previous recommendations for selecting cell sizes have been ad hoc. In some cases, our results indicate a relatively straightforward relationship between cell size and the number of cells needed or the total area sampled to achieve a given power threshold. Between a 100km2 grid and a 225km2 grid, with high detection probability, 80% power can be obtained either by sampling many small cells or fewer of the larger cells. However, by the time cell sizes reach 1000km2 for wolverine, the home ranges for multiple individuals are included in the cell, such that occupancy-based methods alone will only pick up changes once a much larger population decline has occurred. The point at which this switch occurs will likely depend on an interaction of the population size, landscape, home range sizes, and cell size. We also discovered a counterintuitive anomaly when computing power under scenarios in which cell size is equal to home range size, as is often advised for occupancy surveys of mobile carnivores. Specifically, we noted that when detection probability is low, power generally increases with increasing visits to each sample unit, but there is a point at which conducting more visits actually decreases power. We offer the following explanation for this phenomenon: When the cell size is equal to the home range size, the interplay between psim (i.e., 0.2) and availability is such that the pest is fairly low and makes a substantial upward adjustment on the count of cells (c) in which wolverines were actually detected. As we make more visits we detect wolverine use in cells that are seldom used, so c increases, but pest from the model does not (only the precision on pest improves). After about 6 or 7 visits c increases enough that the occupancy estimates resulting from upward adjustments on c approach 1.0. If estimates for all years are at or near 1.0, then there is no trend and we have no power to detect declines. This does not occur when cell sizes are small, because c will also be small, and any upward adjustments will not approach 1.0. A similar phenomenon occurs if cells are large and psim is high. In that case, most cells are used, and c will be large, especially with a large number of visits. Thus, even a small upward adjustment on c pushes the estimates to close to 1.0, which again makes detecting trends difficult. Thus, if maximizing power is a goal, increasing visits beyond a certain threshold may not be helpful depending on cell size, availability of animals, and the probability of detecting them given their presence. 15 �Our simulations currently do not include cost functions, so trade-offs between cell size, number of cells to sample, number of visits at each cell, and detection probability have been conducted absent an important real-world consideration. For instance, in a given situation, it may be easy to complete more visits to a site (e.g., leave camera sets out 1 more month), but extremely costly to improve capture probability (e.g, purchase an entire set of new cameras with improved functionality). Therefore, managers may opt to make more visits to improve power even though intensifying effort (visits) by a given percent may be inferior to improving detection probability by a similar percentage. Future simulation work should include cost as a factor in weighing the importance of the design factors we considered here. Most studies base power analyses for occupancy estimation solely on detecting various simulated declines in occupancy. Here, we employed a more mechanistic, spatially-based approach in which we simulated animals on a landscape, accounted for their natural history (territoriality, difference between sexes), tied their space use to key habitat features (persistent spring snow), and forced declines or increases in the real parameter of interest (abundance) to determine whether occupancy estimation could detect those changes. Thus, our approach is a direct test of the link between occupancy and abundance, providing a more meaningful examination of whether real-world changes of interest in population size can actually be detected using occupancy estimation. It also sets the stage for direct comparisons between occupancy and estimation of other metrics (e.g., abundance) that could potentially be used to monitor populations. That is, we have established the machinery necessary to simulate ‘truth’ (the configuration of animals on the landscape and changes in that configuration and/or number) and can then sample from that true population in various ways to simulate data gathering under different monitoring approaches. While results from this analysis can be used directly to guide the monitoring of wolverine or similar species, the largest contribution is the framework which can be used for making decisions about the design of a large scale monitoring effort provided information on movement and habitat use is available. Our goals were to establish this framework to encourage cost-effective decisions in designing monitoring programs and to inspire well-coordinated surveys across multiple entities and jurisdictions. Without such coordination our analyses convincingly show that most efforts for species like wolverine will be wasted. ACKNOWLEDGMENTS We thank Paul Lukacs, Gary White, and Larissa Bailey for providing invaluable technical advice, and Jeff Laake for implementing the “random occupancy dynamics” model into RMark so it could be used in this analysis. We thank the RMRS and a PECASE award to MKS for providing the initial funding for this effort. LITERATURE CITED Aubry, K. B., K. S. McKelvey, and J. P. Copeland. 2007. Distribution and broadscale habitat relations of the wolverine in the contiguous United States. Journal of Wildlife Management 71:2147–2158. Banci, V. 1994. Wolverine. Pages 99–122 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, L. J. Lyone, andW. J. Zielinski, editors. The scientific basis for conserving forest carnivores in the western United States. U.S. Forest Service, General Technical Report, RM-254. Bart, J., S. Brown, B. Harrington, and R. I. G. Morrison. 2007. Survey trends of North American shorebirds: Population declines or shifting distributions? . Journal of Avian Biology 38:73–82. Broms, K., J. R. Skalski, J. J. Millspaugh, C. A. Hagen, and J. H. Schulz. 2010. Using Statistical Population Reconstruction to Estimate Demographic Trends in Small Game Populations. The Journal of Wildlife Management 74:310–317. Butchart, S. H. M., M. Walpole, B. Collen, A. van Strien, J. P. W. Scharlemann, R. E. A. Almond, J. E. M. Baillie, B. Bomhard, C. Brown, J. Bruno, K. E. Carpenter, G. M. Carr, J. Chanson, A. M. Chenery, J. Csirke, N. C. Davidson, F. Dentener, M. Foster, A. Galli, J. N. Galloway, P. 16 �Genovesi, R. D. Gregory, M. Hockings, V. Kapos, J.-F. Lamarque, F. Leverington, J. Loh, M. A. McGeoch, L. McRae, A. Minasyan, M. H. Morcillo, T. E. E. Oldfield, D. Pauly, S. Quader, C. Revenga, J. R. Sauer, B. Skolnik, D. Spear, D. Stanwell-Smith, S. N. Stuart, A. Symes, M. Tierney, T. D. Tyrrell, J. Vié, and R. Watson. 2010. Global Biodiversity: Indicators of Recent Declines. Science 328:1164–1168. Copeland, J. P. 1996. Biology of wolverine in Central Idaho. University of Idaho, Moscow, ID. Copeland, J. P., K. S. McKelvey, K. B. Aubry, A. Landa, J. Persson, R. M. Inman, J. Krebs, E. Lofroth, H. Golden, J. R. Squires, A. Magoun, M. K. Schwartz, J. Wilmot, C. L. Copeland, R. E. Yates, I. Kojola, and R. May. 2010. The bioclimatic envelope of the wolverine (Gulo gulo): do climatic constraints limit its geographic distribution? Canadian Journal of Zoology-Revue Canadienne De Zoologie 88:233–246. _____. 2011. The bioclimatic envelope of the wolverine (Gulo gulo): do climatic constraints limit its geographic distribution? Canadian Journal of Zoology-Revue Canadienne De Zoologie 88:233– 246. Dennis, B., P. L. Munholland, and J. M. Scott. 1991. Estimation of growth and extinction parameters for endangered species. Ecological Monographs 61:115–143. Devineau, O., T. M. Shenk, G. C. White, P. F. Doherty, P. M. Lukacs, and R. H. Kahn. 2010. Evaluating the Canada lynx reintroduction programme in Colorado: patterns in mortality. Journal of Applied Ecology 47:524–531. Field, S. A., A. J. Tyre, and H. P. Possingham. 2005. Optimizing allocation of monitoring effort under economic and observational constraints. The Journal of Wildlife Management 69:473–482. Finley, D. J., G. C. White, J. P. Fitzgerald, and Vojta. 2005. Estimation of swift fox population size and occupancy rates in eastern Colorado. Journal of Wildlife Management 69:861–873. Foster, C. R., A. F. Amos, and L. A. Fuiman. 2009. Trends in Abundance of Coastal Birds and Human Activity on a Texas Barrier Island Over Three Decades. Estuaries and Coasts 32:1079–1089. GAO. 2005. Fish and Wildlife Service Generally Focuses Recovery Funding on High Priority Species, but Needs to Periodically Assess its Funding Decisions. US Government Accountability Office, Washington, DC, US-GAO Report No. GAO-05-211. GAO-05-211. Gardner, C. L., J. P. Lawler, J. M. V. Hoef, A. J. Magoun, and K. A. Kellie. 2010. Coarse-Scale Distribution Surveys and Occurrence Probability Modeling for Wolverine in Interior Alaska. Journal of Wildlife Management 74:1894–1903. Gaston, K. J. 1991. How large is a species' geographic range? Oikos 61:434–438. Getz, W. M., S. Fortmann-Roe, P. C. Cross, A. J. Lyons, S. J. Ryan, and C. C. Wilmers. 2007. LoCoH: Nonparameteric Kernel Methods for Constructing Home Ranges and Utilization Distributions. Plos One 2. Getz, W. M., and C. C. Wilmers. 2004. A local nearest-neighbor convex-hull construction of home ranges and utilization distributions. Ecography 27:489–505. Hall, D. K., G. A. Riggs, and V. V. Salomonson. 2006. MODIS/Terra snow cover daily L3 global 500m grid. Version 4, 24 April – 21 May 2000–2006. in National Snow and Ice Data Center, Boulder, CO. Hoffmann, M., C. Hilton-Taylor, A. Angulo, M. Böhm, T. M. Brooks, S. H. M. Butchart, K. E. Carpenter, J. Chanson, B. Collen, N. A. Cox, W. R. T. Darwall, N. K. Dulvy, L. R. Harrison, V. Katariya, C. M. Pollock, S. Quader, N. I. Richman, A. S. L. Rodrigues, M. F. Tognelli, J.-C. Vié, J. M. Aguiar, D. J. Allen, G. R. Allen, G. Amori, N. B. Ananjeva, F. Andreone, P. Andrew, A. L. A. Ortiz, J. E. M. Baillie, R. Baldi, B. D. Bell, S. D. Biju, J. P. Bird, P. Black-Decima, J. J. Blanc, F. Bolaños, W. Bolivar-G, I. J. Burfield, J. A. Burton, D. R. Capper, F. Castro, G. Catullo, R. D. Cavanagh, A. Channing, N. L. Chao, A. M. Chenery, F. Chiozza, V. Clausnitzer, N. J. Collar, L. C. Collett, B. B. Collette, C. F. C. Fernandez, M. T. Craig, M. J. Crosby, N. Cumberlidge, A. Cuttelod, A. E. Derocher, A. C. Diesmos, J. S. Donaldson, J. W. Duckworth, G. Dutson, S. K. Dutta, R. H. Emslie, A. Farjon, S. Fowler, J. Freyhof, D. L. Garshelis, J. Gerlach, D. J. Gower, T. D. Grant, G. A. Hammerson, R. B. Harris, L. R. Heaney, S. B. Hedges, J.-M. Hero, B. Hughes, S. 17 �A. Hussain, J. Icochea M, R. F. Inger, N. Ishii, D. T. Iskandar, R. K. B. Jenkins, Y. Kaneko, M. Kottelat, K. M. Kovacs, S. L. Kuzmin, E. La Marca, J. F. Lamoreux, M. W. N. Lau, E. O. Lavilla, K. Leus, R. L. Lewison, G. Lichtenstein, S. R. Livingstone, V. Lukoschek, A. M. Alonso, D. P. Mallon, P. J. K. McGowan, A. McIvor, P. D. Moehlman, S. Molur, A. M. Alonso, J. A. Musick, K. Nowell, R. A. Nussbaum, W. Olech, N. L. Orlov, T. J. Papenfuss, G. Parra-Olea, W. F. Perrin, B. A. Polidoro, M. Pourkazemi, P. A. Racey, J. S. Ragle, M. Ram, G. Rathbun, R. P. Reynolds, A. G. J. Rhodin, S. J. Richards, L. O. Rodríguez, S. R. Ron, C. Rondinini, A. B. Rylands, Y. Sadovy de Mitcheson, J. C. Sanciangco, K. L. Sanders, G. Santos-Barrera, J. Schipper, C. SelfSullivan, Y. Shi, A. Shoemaker, F. T. Short, C. Sillero-Zubiri, D. L. Silvano, K. G. Smith, A. T. Smith, J. Snoeks, A. J. Stattersfield, A. J. Symes, A. B. Taber, B. K. Talukdar, H. J. Temple, R. Timmins, J. A. Tobias, K. Tsytsulina, D. Tweddle, C. Ubeda, S. V. Valenti, P. Paul van Dijk, L. M. Veiga, A. Veloso, D. C. Wege, M. Wilkinson, E. A. Williamson, F. Xie, B. E. Young, H. R. Akçakaya, L. Bennun, T. M. Blackburn, L. Boitani, H. T. Dublin, G. A. B. da Fonseca, C. Gascon, T. E. Lacher, G. M. Mace, S. A. Mainka, J. A. McNeely, R. A. Mittermeier, G. M. Reid, J. P. Rodriguez, A. A. Rosenberg, M. J. Samways, J. Smart, B. A. Stein, and S. N. Stuart. 2010. The Impact of Conservation on the Status of the World's Vertebrates. Science 330:1503–1509. Inman, R. M., M. L. Packila, K. H. Inman, A. J. McCue, G. C. White, J. Persson, B. C. Aber, M. L. Orme, K. L. Alt, S. L. Cain, J. A. Fredrick, B. J. Oakleaf, and S. S. Sartorius. 2011. Spatial ecology of wolverines at the southern periphery of distribution. The Journal of Wildlife Management 76:778–792. Inman, R. M., M. L. Packila, K. H. Inman, A. J. Mccue, G. C. White, J. Persson, B. C. Aber, M. L. Orme, K. L. Alt, S. L. Cain, J. A. Fredrick, B. J. Oakleaf, and S. S. Sartorius. 2012. Spatial ecology of wolverines at the southern periphery of distribution. The Journal of Wildlife Management 76:778–792. Joseph, L. N., S. A. Field, C. Wilcox, and H. P. Possingham. 2006. Presence-absence versus abundance data for monitoring threatened species. Conservation Biology 20:1679–1687. Krebs, J., E. C. Lofroth, and I. Parfitt. 2007. Multiscale habitat use by wolverines in British Columbia, Canada. Journal of Wildlife Management 71:2180–2192. Laurance, W. F., S. G. Laurance, and D. W. Hilbert. 2008. Long-term dynamics of a fragmented rainforest mammal assemblage. Conservation Biology 22:1154–1164. Lofroth, E. C., and J. Krebs. 2007. The abundance and distribution of wolverines in British Columbia, Canada. Journal of Wildlife Management 71:2159–2169. MacKenzie, D. I. 2005. What Are the Issues with Presence-Absence Data for Wildlife Managers? The Journal of Wildlife Management 69:849–860. MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy Estimation and Modeling: Inferring Patterns and Dynamics of Species Occurrence. Elsevier, Amsterdam. Magoun, A. J., C. D. Long, M. K. Schwartz, K. L. Pilgrim, R. E. Lowell, and P. Valkenburg. 2011. Integrating motion-detection, cameras, and hair snags for wolverine identification. Journal of Wildlife Management 75:731–739. Magoun, A. J., J. C. Ray, D. S. Johnson, P. Valkenburg, F. N. Dawson, and J. Bowman. 2007. Modeling wolverine occurrence using aerial surveys of tracks in snow. Journal of Wildlife Management 71:2221–2229. Male, T. D., and M. J. Bean. 2005. Measuring progress in US endangered species conservation. Ecology Letters 8:986–992. Marsh, D. M., and P. C. Trenham. 2008. Current trends in plant and animal population monitoring. Conservation Biology 22:647–655. Marucco, F., D. H. Pletscher, L. Boitani, M. K. Schwartz, K. L. Pilgrim, and J.-D. Lebreton. 2009. Wolf survival and population trend using non-invasive capture-recapture techniques in the Western Alps. Journal of Applied Ecology 46:1003–1010. 18 �Moriarty, K. M., W. J. Zielinski, A. G. Gonzales, T. E. Dawson, K. M. Boatner, C. A. Wilson, F. V. Schlexer, K. L. Pilgrim, J. P. Copeland, and M. K. Schwartz. 2009. Wolverine Confirmation in California after Nearly a Century: Native or Long-Distance Immigrant? Northwest Science 83:154–162. Mulders, R., J. Boulanger, and D. Paetkau. 2007. Estimation of population size for wolverines Gulo gulo at Daring Lake, Northwest Territories, using DNA based mark-recapture methods. Wildlife Biology 13:38–51. Otto, C. R. V., and G. J. Roloff. 2011. Using multiple methods to assess detection probabilities of forestfloor wildlife. The Journal of Wildlife Management 75:423–431. Povilitis, A., and K. Suckling. 2010. Addressing Climate Change Threats to Endangered Species in U.S. Recovery Plans. . Conservation Biology 24:372–376. Primack, R. B. 2006. Essentials of Conservation Biology. Fourth edition. Sinauer Associates, Sunderland. R Development Core Team. 2011. R Foundation for Statistical Computing, Vienna, Austria. Rands, M. R. W., W. M. Adams, L. Bunnen, S. H. M. Butchart, A. Clements, D. Coomes, A. Entwistle, I. Hodge, V. Kapos, J. P. W. Scharlemann, W. J. Sutherland, and B. Vira. 2010. Biodiversity Conservation: Challenges Beyond 2010 Science 329:1298–1303. Rhodes, J. R., and N. Jonzén. 2011. Monitoring temporal trends in spatially structured populations: how should sampling effort be allocated between space and time?. . Ecography 34:1040–1048. Rhodes, J. R., A. J. Tyre, N. Jonzén, C. A. McAlpine, and H. P. Possingham. 2006. Optimizing PresenceAbsence Surveys For Detecting Population Trends. The Journal of Wildlife Management 70:8– 18. Schwartz, M. K., J. P. Copeland, N. J. Anderson, J. R. Squires, R. M. Inman, K. S. McKelvey, K. L. Pilgrim, L. P. Waits, and S. A. Cushman. 2009. Wolverine gene flow across a narrow climatic niche. Ecology 90:3222–3232. Schwartz, M. K., G. Luikart, and R. S. Waples. 2007. Genetic monitoring as a promising tool for conservation and management. Trends in Ecology & Evolution 22:25–33. Schwartz, M. K., and S. L. Monfort. 2008. Genetic and endocrine tools for carnivore surveys. in R. A. Long, P. MacKay, J. C. Ray, andW. J. Zielinski, editors. Noninvasive survey methods for North American carnivores. Island Press, Washington D.C. Shenk, T. M. 2009. Post–Release Monitoring of Lynx Reintroduced to Southwestern Colorado. Colorado Division of Wildlife. Shenk, T. M., and R. H. Kahn. 2010. The Colorado lynx reintroduction program. Colorado Division of Wildlife. Squires, J. R., J. Copeland, T. J. Ulizio, I. K. Schnvartz, and L. F. Ruggiero. 2007. Sources and patterns of wolverine mortality in western Montana. Journal of Wildlife Management 71:2213–2220. Theobald, D. M., and T. M. Shenk. 2011. Areas of high habitat use from 1999–2010 for radio-collared Canada lynx reintroduced to Colorado. Colorado State University. Ulizio, T. J., J. R. Squires, D. H. Pletscher, M. K. Schwartz, J. J. Claar, and L. F. Ruggiero. 2006. The Efficacy of Obtaining Genetic-Based Identifications from Putative Wolverine Snow Tracks. Wildlife Society Bulletin 34:1326–1332. USFWS. 2010. Endangered and threatened wildlife and plants; 12-month finding on a petition to list the North American wolverine as endangered or threatened. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation for populations of marked animals. Bird Study 46:120–138. Zielinski, W. J., and H. B. Stauffer. 1996. Monitoring Martes Populations in California: Survey Design and Power Analysis. Ecological Applications 6:1254–1267. Prepared by ___________________________ Jacob S. Ivan 19 �Table 1. A summary of variables and ranges of those variables tested in our simulations using program SPACE. Variable Population size Population growth rates Limit on movement Simulated detection probability Number of cells sampled Number of visits Cell size Sampling Values tested N = 30, 200, 500 0.933, 1.041, 1.072, 1.116 N = 30 λ = 0.933, 1.041 N=200 0.933, 0.977, 0.989, 1.041 N=500 none; 1, 2 s.d. from home range center 0.2, 0.8 10 - 90% of grid 2-7 100, 225, 500, 1000 km2 Annual or alternating years (every other year) 20 �■ ■ •Northern U.S. Rockies ■ ■ ■ 1 ■ ·~·~·~·~•_j • tJt • ••• •:Southern Rocki s --..i=~·~··; ; ·~ ·~·~·~·~· :··············· ■ •• ■ ■ ■ .. ■ • • • • ■ ■ ■ ■ ■ ■ ■ ■ •• ( ■ 0 90 180 350 1<11ometers Figure 1. Map of study area. Distribution of persistent spring snow in the U.S. Rocky Mountains. Two separate landscapes were included in this study: one corresponding to fairly continuous habitat in the U.S. Northern Rockies, which is currently occupied by wolverines, and a second area in the Southern Rockies, where wolverines may recolonize or be reintroduced. 21 �No = 30 No = 200 No= 500 1.0 1 0.8 0.6 ~ II 0 "'C "' <,) <,) 2 0.4 0 'Q) .D 0.2 ~ E :, z # visits - 2 - 3 - 4 -5 vi Q) .£ 0 Q) - 6 - 7 ~0.8 -i Q) P s1m ts Q) - m0.6 1 0 0.4 ., 0.2 50 100 150 200 250 300 350 50 100 150 200 250 300 350 50 100 150 200 250 300 350 Number of cells sampled Figure 2. Effect of population size. Effect of population size on power to detect trends in the Northern U.S. Rockies. Assumes perfect detection associated with sampling for a 50% decline (λ = 0.933) or a 50% increase (λ = 1.041) from initial population sizes of 30, 200, and 500 individuals in the Northern Rockies. Simulated populations were sampled using a grid of 225km2 cells overlaid on the landscape. 22 �1.0 1 •0.977 t• 0.989 t• 1.0,11 --=~~~::::::::::::::=-11---....c..._='-'----r ----"---'"-=---i '-------'C......C.'-"----h r , 0.8 0.6 1 -= 0.933 r -I ,I "' Porn C: 2 0.4 - 0.2 -- • 0.8 # visits 2 - 3 0.4 0.2 0.0 ■ :a ,-, 7! fft Ai '--,---,--,c---,---.--,--,-' •-..--,--.---,--,--,--,-,--,--,---,--,c---,---,--r'--,--..--,--.---,--,--,-'-' 50 100 150 200 250 300 350 50 100 150 200 250 300 350 50 100 150 200 250 300 350 50 100 150 200 250 300 350 Number of cells sampled Figure 3. Power for detecting trends in the U.S. Northern Rockies. Results from a power analysis for assessing the feasibility of using occupancy to monitor trend in the population of wolverines in the U.S. Northern Rockies, assuming N=500 individuals and a cells size of 225km2. Results are parsed by population growth rate (λ = 0.933, 0.977, 0.989, 1.041 corresponding to 50%, 20%, and 10% declines over 10 years or a 50% increase), sampling effort (whether sampling occurred annually or every other year), detection probability for sampling, number of visits per year, and number of grid cells sampled from a total of 388. Power is based on number of detected trends in 1,000 simulated populations. 23 �"iij J!l "- = 0.933 '". '"..,,' '' rf> ~u ' ' Q) :i2 Q a:: 0 C: .c Q) Cl) :i 0 '"'"' ' .,, ' ,,, ,,, ., '" 0 C> 00 C> <.O ' 1\\\ \ I \\\ \ I \\\ \ I "<I" C> C> N C> O ~ ~ ~ 0 0 ""': 0 0 C"! N ! o....: O M 1,. =1.041 \\ I \ I . I\ I . ,u au • a 111 an , a 111 O V LO Ill <.0 ~ ~ 0 0 I I I "' = 1.072 ...,''' ' '' ., ' ' ' •'..,'\ ' •' ""': 0 sunJ JO JaqwnN/Sau1pap papaiaa 1,. =1.116 \ 0 0 \ ,,, ,,, ' ' •'' ' ,',, ' ,,, •"•".., C> N 0 00 \ \ \ ' ... .. ,, , .,...' , ........., . C> 0 0 C> 00 (!) 0 .. O"O Q) v a. 0 N 0 0 ."'.,,.,,'',' 0 0 N LO 0 ,,...,,'............. '... .,. ... .... (!) 0 v C> 0 0 ~ LO 0 C> N C> 0 cu Cl) E Q) (/) LO'+- oO MQ .0 ME .... Q) o::J ~z 24 Figure 4. Comparison of N=30 populations in Northern and Southern Rockies. Power to detect population trends for populations of 30 wolverines in U.S. Northern Rockies compared to the same population size in the Southern Rockies landscape. Ability to detect a population decline depends on population growth rate (λ = 0.933, 1.041, 1.072, 1.116 corresponding to a 50% decline over 10 years or 50%, 2-fold, or 3-fold increases in population size over 10 years) and sampling effort (detection probability for sampling, number of visits per year, number of grid cells sampled from a total of 388 for the Northern Rockies or 128 for Colorado). Power is based on number of detected trends from 1,000 simulated populations. r-.. �500km2 1000km2 ,,j 1.0 0.6 "'20.4 Psm C - 0 0.2 - • 0.8 -Fz Wii? 7 C # visits - 2 - 3 - 4 ~ .,3 -s -< - ,; "'~ -6 7 {/) "'3 '2. 5· co 0.2 0.0 200 400 600 800 50 100 150 200 250 300 350 40 60 80 100 120 140 10 20 30 40 50 60 70 Number of cells sampled Figure 5. Effect of cell size on power. Effect of grid size on power to detect population trends in the wolverine population (N=500) in the Northern Rockies using occupancy. As grid size changes, the total number of grid cells on the landscape also changes from 887 for a 100km2 grid, 388 for a 225 km2 grid, to 76 for a 1000km2 grid. 25 � https://cpw.cvlcollections.org/files/original/9167cc61f07a79259299692f05e84e2d.pdf c23b922d1d1b7b9689588c8b9d5dd8d4 PDF Text Text Colorado Division of Parks and Wildlife July 2010 – June 2011 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Colorado 3430 3002 : : : : : Division of Parks andWildlife Mammals Research Elk Conservation Evaluating solutions to reduce elk and mule deer damage on agricultural resources Federal Aid Project No. Period Covered: July 1, 2010 – June 30, 2011 Author: H.E. Johnson; project cooperators, P. Dorsey, M. Hammond, C. Bishop, K. VerCauteren, D. Walter, and C. Anderson. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Elk and mule deer provide important recreational, ecological, and economic benefits, but they can also cause substantial damage to agricultural resources in rural environments. This situation has generated significant challenges for wildlife agencies that are responsible for maintaining viable ungulate populations while also minimizing crop damage. One of the most severe areas of ungulate damage in Colorado has been the sunflower fields around Dove Creek. In this region, roughly a quarter of million dollars were annually paid to farmers between 2007 and 2009, and kill permits, distribution hunts and private-land-only doe hunts have been routinely distributed to farmers. Pressure from local growers over damage, and frustration from the general public over kill permits, have generated the need for the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) to evaluate other management options for reducing elk and deer crop depredation. As a result, CPW partnered with wildlife damage researchers from the National Wildlife Research Center to find science-based solutions for reducing crop damage. Collaboratively, our goals are to 1) examine elk and mule deer distribution patterns to design public hunting opportunities to reduce depredation, 2) experimentally test a suite of non-lethal fencing techniques to minimize crop damage, and 3) map and model landscape characteristics associated with damage to specify more effective site-specific management practices. During FY10-11 we developed a research proposal for internal review, generated project funding, and initiated the construction of experimental fence plots. Other project activities (i.e., monitoring the effectiveness of the different fence types for minimizing elk and deer damage, deploying telemetry collars, and mapping and modeling ungulate damage) will be initiated between FY11-12 and FY13-14. 123 �WILDLIFE RESEARCH REPORT EVALUATING SOLUTIONS TO REDUCE ELK AND MULE DEER DAMAGE ON AGRICULTURAL RESOURCES HEATHER E. JOHNSON P.N. OBJECTIVES To conduct a study on elk and mule deer around the agricultural fields of Dove Creek that 1) examines wild ungulate distribution patterns to design public hunting opportunities to reduce crop damage, 2) experimentally tests a suite of non-lethal fencing techniques to minimize crop depredation, and 3) maps and models landscape characteristics associated with damage to specify more effective site-specific management practices. SEGMENT OBJECTIVES 1. Work with staff from CPW and the National Wildlife Research Center to develop a research proposal for internal CPW peer review and funding solicitation. 2. Implement the construction of experimental fence plots on sunflower fields south of Dove Creek, Colorado, including electric fences, temporary winged fences, and chemical repellent fences. INTRODUCTION Elk and deer provide important recreational, ecological, and economic benefits, but they can also cause substantial damage to agricultural resources in rural areas (Austin et al. 1998, Wisdom and Cook 2000). In Colorado, elk and deer damage of crops accounts for a majority of the wildlife damage claims in the state. CPW is obligated to pay eligible wildlife damage claims on agricultural resources, and in recent years, the agency has spent approximately $500,000 on an annual basis to compensate growers. One of the most significant hotspots of elk and mule deer damage has been in the vicinity of Dove Creek, in conjunction with a recent switch in the agricultural crops that are locally grown. Farmers traditionally grew crops such as dry beans, spring and winter wheat, oats, alfalfa and grass hay which had minimal damage by wild ungulates. In recent years, however, local growers have planted sunflowers, a high-value seed oil crop used to produce biofuels, and highly desirable by wild ungulates. The main management tool available to CPW to reduce ungulate sunflower damage has been to increase harvest through the use of kill permits, distribution hunts, and private land only doe hunts, however tolerance for these permits has been low among local sportsman and the general public. Given pressure by farmers over elk and deer crop damage, and frustration by sportsmen and the public over kill permits, CPW wildlife managers were interested in finding alternative management solutions for reducing sunflower depredation. As a result, CPW managers partnered with the CPW research branch and wildlife-damage researchers from the National Wildlife Research Center (NWRC) to find non-lethal science-based solutions for reducing sunflower damage. Collaboratively, our goals are to 1) identify public hunting strategies that reduce crop damage, 2) test a suite of non-lethal fencing techniques to minimize crop depredation, and 3) map and model landscape characteristics associated with damage behavior to specify more effective management practices. Results from this study should enable CPW and local growers to reduce ungulate crop depredation, leading to a decrease in compensation payments, a decrease in kill permits/distribution hunts, and an increase in public hunting opportunities. A detailed research proposal (Johnson et al. 2011) is provided in Appendix I. 124 �STUDY AREA The study will be conducted in the vicinity of Dove Creek, Colorado (Montezuma, San Miguel and Dolores Counties), which is comprised of a mixture of agricultural and public lands. The project will focus on the north half of CPW Game Management Unit 72 and the west half of 711 (the portion west of the Dolores River). The area is generally characterized as mountain shrubland interspersed with irrigated and dryland agricultural fields, ranging from 1,981 to 2,590 m in elevation. The mountain shrub habitat type is primarily composed of serviceberry (Amelanchier alnifolia), bitterbrush (Purshia tridentata), mountain mahogany (Cercocarpus montanus), squaw apple (Peraphyllum ramosissimum) and black sagebrush (Seriphidium novum). Sunflower fields around Dove Creek are spatially juxtaposed to deep canyons that provide refugia for elk, exacerbating ungulate damage on agricultural crops (Fig. 1). METHODS During winter and spring of FY10-11 project collaborators developed a research proposal for internal CPW review and for funding solicitation (Appendix I). We successfully obtained project funds from the Rocky Mountain Elk Foundation, Colorado Statewide Habitat Partnership Program (HPP), Montelores HPP, National Wildlife Research Center and CPW Auction/Raffle Grants. Project grants and in-kind contributions totaled ~$279,000 which was sufficient to fully finance the project. Once project funding was solidified we initiated field logistics: the acquisition of field materials (fencing materials, elk and deer GPS collars, etc), contracting a fence installation company to construct experimental fence plots, hiring a temporary employee to monitor elk and deer damage on experimental fence plots, and scheduling a helicopter capture to deploy elk and deer collars. During FY10-11 we constructed the experimental fence plots based on a randomized block design. We identified 5 replicate fields that have repeatedly suffered high ungulate crop damage. Within each field we specified 4 10-acre plots, one for each experimental fence treatment type (polyrope fence, temporary winged fence, chemical repellent fence) and a control (see Appendix I for detailed descriptions of the fence types and study design). The 4 plots were randomly assigned within each field, such that each field (block) contained one replicate of all treatments (Gotelli and Ellison 2004). Other scheduled project activities will be initiated during FY11-12 such as monitoring the effectiveness of the different fence types for minimizing elk and deer damage, deploying telemetry collars, and mapping and modeling ungulate damage. SUMMARY AND FUTURE PLANS During FY10-11 we successfully developed a research proposal, generated project funding, and constructed the experimental fence plots for the first year of fieldwork. Starting in FY11-12 we will monitor the efficiency of the experimental fence plots in reducing elk and deer damage (July – Oct 2011) and deploy 40 GPS collars; 20 collars on adult female elk and 20 on adult female deer (Oct 2011). Experimental fence plots will also be monitored for elk and deer damage during FY12-13 (July-Oct). Elk and deer collars will collect data for 2 years and then detach in Sept 2013. Once collars are retrieved we will analyze and model elk and deer location data relative to agricultural and wildland habitat (FY13-14) 125 �LITERATURE CITED Austin, D.D., P.J. Urness, and D. Duersch. 1998. Alfalfa hay crop loss due to mule deer depredation. Journal of Range Management 51:29-31. Gotelli, N.J., and A.M. Ellison. 2004. A primer of ecological statistics. Sinauer Associates, Sunderland, Massachusetts. Johnson, H.E., P.Dorsey, M. Hammond, C. Bishop, K. VerCauteren, D. Walter, and C. Anderson. 2011. Evaluating solutions to reduce elk and mule deer damage on agricultural resources. Study Proposal, Colorado Division of Parks and Wildlife, Fort Collins, USA. Wisdom, M.J., and J.G. Cook. 2000. North American Elk. Pages 694-735 in S. Demarais and P.R. Krausman, editors. Ecology and management of large mammals in North America. Prentice Hall, Upper Saddle River, New Jersey, USA. Prepared by _______________________________________ Heather E. Johnson, Wildlife Researcher 126 �Figure 1. Placement of experimental fence plots within the 5 replicate sunflower fields. Fields are located adjacent to wildland canyons. 127 �APPENDIX I PROGRAM NARRATIVE STUDY PLAN FOR MAMMALS RESEARCH FY 2010-11 Evaluating solutions to reduce elk and mule deer damage on agricultural resources A Research Proposal Submitted By Heather Johnson, Mammals Researcher, CPW Patt Dorsey, Area Wildlife Manager, CPW Matt Hammond, District Wildlife Manager, CPW Chad Bishop, Mammals Research Leader, CPW Kurt VerCauteren, Ungulate Damage and Disease Project Leader, National Wildlife Research Center David Walter, Post-Doctoral Researcher, National Wildlife Research Center Charles Anderson, Post-Doctoral Researcher, National Wildlife Research Center A. Need Elk and deer provide important recreational, ecological, and economic benefits, but they can also cause substantial damage to agricultural resources in rural environments (Austin et al. 1998, Wisdom and Cook 2000). Because crops are typically more digestible and contain higher levels of crude protein than native grasses and browse species, they are often preferentially selected and consumed by wild ungulates (Mould and Robbins 1981). Agricultural producers have reported more damage by elk and deer than any other wildlife species, and damage by deer alone has been projected to exceed 100 million dollars annually in the U.S. (Conover 2002). This situation has generated significant challenges for management agencies that are responsible for maintaining viable ungulate populations while also minimizing crop damage (Van Tassell et al. 1999, Wagner et al. 1997, Wilson et al. 2009, Hegel et al. 2009, Walter et al. 2010). Elk and deer crop depredation results from a combination of factors including the seasonal distribution and abundance of local forage resources, landscape configuration, and herd density patterns (Vecellio et al. 1994; Yoder 2002; Hegel et al. 2009). Damage can be highly variable within and among growing seasons, as local precipitation and temperatures will alter the availability of native forage and the motivation of ungulates to feed on agricultural fields (Walter et al. 2010). The juxtaposition of cropland and wildland has also been found to be particularly important in driving damage rates, as those cultivated fields closer to cover experience more damage (Nixon et al. 1989, Hegel et al. 2009). Additionally, studies have found that ungulate damage is often caused by only a subset of individuals in the population, depending on the spatial and social structuring of the herd. These observations have critical implications for wildlife managers, as 1) management practices may be differentially effective based on the variability of native forage and the spatial juxtaposition of other habitat features, and 2) management techniques targeted at specific animals may be more effective than implementing those techniques on the population at large (Blejwas et al. 2002, Hegel et al. 2009). As a result, an understanding of both the spatial configuration of seasonal resources and the resource selection patterns of different segments of local ungulate populations is important to successfully identify strategies to reduce elk and deer crop damage (Hegel et al. 2009). In Colorado, elk and deer damage of crops accounts for a majority of the wildlife damage claims in the state. The Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) is obligated to pay eligible wildlife damage claims on agricultural resources, and in recent years, the agency has spent 128 �approximately $500,000 on an annual basis to compensate growers. One of the most significant hotspots of elk and mule deer damage has been in the vicinity of Dove Creek (Montezuma, San Miguel and Dolores Counties; Fig. 1), where roughly a quarter of million dollars was annually paid to farmers between 2007 and 2009. These extraordinary reimbursements have resulted from a recent switch in the agricultural crops that are locally grown. Farmers around Dove Creek traditionally grew crops such as dry beans, spring and winter wheat, oats, alfalfa and grass hay which had minimal damage by wild ungulates. In recent years, however, local growers have planted sunflowers, a high-value seed oil crop used to produce biofuels. In 2009 growers were paid approximately $.43/lb for organically grown sunflowers and $.28/lb for conventionally grown sunflowers. In that same year, dry land yields averaged 800 lbs/acre in the region. Elk and deer have demonstrated a strong affinity for sunflowers, causing up to 100% crop loss on certain fields, and resulting in high damage claims. Ungulate damage around Dove Creek is exacerbated by the spatial distribution of sunflower fields in relation to the surrounding wildlands (e.g., sagebrush-mixed shrublands and piñon-juniper woodlands). The region is fractured with deep canyons that provide refugia for elk, and fields adjacent to the canyon rims experience the greatest amount of depredation. With the substantial increase in biofuel production in the U.S. (World Resources Institute 2008), the agricultural conversion observed around Dove Creek will likely become common, as highpriced, crops replace more traditionally-grown, lower-cost crops (Walter et al. 2009). The main management tool available to CPW to reduce ungulate sunflower damage has been to increase harvest through the use of kill permits (for males and females), distribution hunts, and private land only (PLO) doe hunts. In response to damage reports, CPW has been allocating these permits to local growers between June and October, with the intent of harvesting resident animals rather than migratory elk and deer. This management strategy has resulted in exceptionally high private land harvests. For example, in 2008, kill permits or distribution hunts were allocated on as many as 25 different fields, with approximately 300 deer and 30 elk harvested. On a single 140-acre sunflower field, 57 female mule deer were harvested, and still the annual damage claim for the field was approximately $40,000 in that year. The CPW, the Bureau of Land Management, Montelores Habitat Partnership Program (HPP) Committee, U.S. Forest Service and Rocky Mountain Elk Foundation have initiated several habitat enhancement projects in the region to draw elk and deer off of agricultural fields, but the benefits of these projects are expected to take several years to fully materialize. Although tolerance for elk and mule deer damage on sunflower crops is low among farmers, tolerance for kill permits and distribution hunts is also low among sportsmen, the general public and some farmers. A majority of the damage occurs during July and August when calves and fawns are still dependent on their mothers, reducing the acceptability of female hunts. Additionally, both elk and deer population numbers in the study area (DAUs D24 and E24) are below or near management objectives creating a paradox where CPW ultimately wants to increase ungulate herds, but reduce crop depredation. Finally, this region is popular with hunters, as large bulls and bucks have been harvested in recent years. Hunting is economically important to Dolores, Montezuma and San Miguel Counties, providing approximately 230 jobs and there is a strong desire to have increased public hunting opportunities. Pressure from local growers over damage, and frustration from the general public over kill permits, have generated the need for CPW to evaluate other management options for reducing elk and mule deer crop depredation. As a result, managers from CPW have partnered with wildlife-damage researchers from the National Wildlife Research Center (NWRC) to find science-based solutions for reducing sunflower damage. Collaboratively, the goals of our study are to design public hunting opportunities to reduce crop damage, test a suite of non-lethal techniques to minimize crop depredation, and map and model landscape characteristics associated with damage behavior to specify more effective management practices. Results from this study should ultimately enable CPW and local growers to reduce ungulate crop depredation, leading to a decrease in compensation payments and kill permits/distribution hunts, and an increase in public hunting opportunities and support from farmers and sportsmen. 129 �B. Objectives Objective 1: Examine the spatial structure, distribution, and migration patterns of local elk and mule deer around agricultural areas. This will enable CPW to design public hunting opportunities that better address crop damage while decreasing the need for kill permits/distribution hunts on private lands. Objective 2: Use treatment and control fields to experimentally test novel methods for reducing elk and mule deer damage to crops including a) the repellent “fence” Plantskydd, b) an electric polyrope fence, and c) a temporary “winged” fence. Objective 3: Map and model the spatial juxtaposition of crop fields, ungulate habitats, human infrastructure and topographic features to assess the predictors of elk and mule deer resource-use and damage. This will allow CPW to explicitly account for landscape configuration when working with landowners to identify best management strategies for reducing damage. C. Expected Results or Benefits Long-Term Benefits  Sustain healthy elk and mule deer populations on public and private lands where they do not cause agricultural damage and can provide quality hunting opportunities.  Reduce elk and deer game damage payments on sunflowers and other crops.  Allow sportsmen to harvest a greater proportion of elk and deer by strategically allocating the number of licenses, the location of those licenses, and the timing of hunts to target conflict animals, reducing the need for kill permits and distribution hunts.  The identification of alternative, non-lethal methods to prevent damage and reduce elk and deer use of crop fields.  The development of a modeling tool that can be used by CPW and growers to select the most appropriate management techniques to minimize damage based on field characteristics, ungulate distribution and landscape configuration.  The application of sound science to on-the-ground wildlife damage management. Short-Term Benefits  Gain knowledge about local elk and deer movements and distribution relative to the location of crop damage.  Help farmers with on-going damage by providing management tools and assistance.  Strengthen CPW’s relationship with the local community (farmers, sportsmen, and the general public) by reducing elk and deer crop damage and increasing public hunting opportunities. D. Approach Examining the spatial distribution of elk and mule deer: To understand ungulate movement patterns and more effectively address damage problems we will capture and collar 20 adult female elk and 20 adult female mule deer. Females cause a majority of the crop depredation and will provide the greatest insight into herd distributions. We will capture animals using a net-gun fired from a helicopter (Krausman et al. 1985), targeting elk and mule deer in the vicinity of high-damage crops. Captured animals will be fitted with global positioning system (GPS) collars, and locations of elk and mule deer will be remotely downloaded, collected once collars are retrieved, and 130 �recorded via ground or aerial telemetry. For both species, GPS collars will be programmed to collect ≥3 locations a day on a revolving schedule for 2 years. Elk and mule deer locations will be tracked yearround so that seasonal resource-use, migration patterns, and distributions can be clearly identified. Due to the elk herd’s close proximity to the Utah border, information on elk locations will be shared with Utah Division of Wildlife, as it is suspected that some animals travel across the Utah border during winter and forage on Colorado sunflower crops during summer. We will use elk and mule deer locations to map seasonal distributions and migration patterns, using kernel density analyses in ArcGIS mapping software (Worton 1989, Worton 1995). This will allow CPW to determine the best timing of special season hunts, kill permits and distribution hunts to avoid the private land harvest of migratory elk at the sportsman’s expense. CPW will also be able use distribution data to design public hunts that will target conflict elk and mule deer. For example, the Utah Division of Wildlife Resources is willing to consider special elk hunts south of Hwy 491 if we find that any or all of the resident elk herds (causing damage) spend portions of the year in Utah. Locations will also allow us to determine the amount of use of crop fields by elk and deer, and the proportion of animals using crop fields (whether it is only certain segments of the population, or the population at large). Testing 3 novel methods to reduce crop damage: In addition to implementing effective harvest practices to reduce crop damage, there is strong public interest in the application of nonlethal techniques for reducing ungulate depredation, generating a need for rigorous evaluation of such techniques by wildlife agencies. Most nonlethal techniques are designed to physically exclude offending animals or reduce the motivation of animals to access protected resources (Nolte 1999). We will test three exclusionary management tools for reducing elk and mule deer crop damage that can be easily implemented by farmers during the growing season: a repellent “fence”, a polyrope electric fence, and a temporary “winged” fence. To test the effectiveness of these methods we will initially select 5 replicate fields that have repeatedly suffered high ungulate crop damage (~160-200 acres), situated along the canyon rims. Within each of those fields we will identify 4 plots, one for each treatment type (repellent, polyrope fence, winged fence) and a control. The 4 plot types will be randomly assigned within each field, utilizing a randomized block study design where each field (block) contains one replicate of all treatments (Gotelli and Ellison 2004). This will allow us to statistically account for environmental heterogeneity, as we expect that damage will be variable among fields. Within the fields, study plots will be spaced as far apart as possible, to account for plot independence, and each plot will be 10 acres2 in size. All study plots will be placed along the agriculture/wildland boundary, where depredation is expected to be concentrated. Plots will be monitored from June through October during the growing seasons of 2011 and 2012. We will quantify the relative success of each nonlethal method by comparing crop depredation and ungulate incursion among treatment and control plots. Plantskydd - Repellents are nonlethal substances used to deter ungulates by decreasing a plant’s palatability, and have had mixed success in deterring ungulate foraging activity (Andelt et al. 1992; Baker et al. 1999). We will test the effectiveness of a relatively new product, Plantskydd, for reducing sunflower damage around Dove Creek. This product was developed in Sweden to reduce mammalian wildlife damage on commercial forests and works by emitting an odor that animals associate with predator activity, repelling the animal before it forages on crop plants. There is great interest in the success of such a technique as it can be easily applied to vegetation by ground and aerial spraying, used on both organic and conventionally grown sunflower crops, and is cost-effective for growers. That said, the effectiveness of Plantskydd has not been experimentally tested, only anecdotally reported. To test this method, the 5 Plantskydd treatment plots will be ground or aerial sprayed around field borders once germination has been begun. We will then re-apply Plantskydd to the treatment plots once/month throughout the 131 �sunflower growing season as the repellent may wash off or decompose over time and will need to be reapplied to new plant material. Polyrope electric fence – Fences provide an effective long-term, nonlethal tool for minimizing ungulate crop damage, providing both a physical and psychological barrier (Walter et al. 2010). While a permanent 2.4 meter woven-wire fence provides a true physical barrier to elk and deer, such a structure is can cost >$20/meter, prohibiting wide-spread use. We will test a novel design of a less expensive polyrope electric fence (approximately $8/meter), which acts primarily as psychological barrier based on learned behavioral, avoidance conditioning (Fig. 2; McKillop and Sibly 1988). These fences consist of conductive wires which are woven into synthetic electric “ropes” that are more durable, visible, and easy to install than traditional electric fences (Hygnstrom and Craven 1988, VerCauteren et al. 2006). Permanent fence posts are placed, and then the polyrope is strung between the posts to provide seasonal crop protection. Avoidance conditioning occurs when an animal contacts the fence, often with the nose or tongue, and receives a powerful electric shock. Training can be expedited by baiting the fence wire with peanut butter or molasses to create a negative stimuli when making contact with the electric charge (Porter 1983, Hygnstrom and Craven 1988, Jordan and Richmond 1992, USDA National Wildlife Research Center, unpublished data). Polyrope fences have had success in reducing deer damage (Hygnstrom and Craven 1988, Seamans and VerCauteren 2006), but have not been experimentally tested for reducing elk damage. For the 5 randomly selected polyrope treatment plots, we will construct a fence that is approximately 1.8 meter tall with 5 strands to discourage passage under, through, or over the fence. We will treat the polyrope with a sweet attractant, designed to facilitate avoidance learning, using a minimum charge of 3,000 volts (Curtis et al. 1994). The polyrope will be powered by a Speedrite™ 3000 energizer (Tru-Test Incorporated, San Antonio, Texas) which has a maximum pulse output of 3.0 joules and will be powered by a 12-volt deep-cycle battery with a solar-panel recharger. Temporary winged fence - For seasonal agricultural resources, such as sunflowers, temporary fences may provide reliable ungulate protection. Temporary fences are inexpensive, lightweight, and easy to erect and remove (Rosenberry et al. 2001, VerCauteren et al. 2006). Recently, investigators have been experimenting with a temporary “winged” fence made of polypropylene mesh. The fence is installed completely on one side of the target field, and partially installed on two other sides having 50-100 meter “wings” that extend perpendicular from the full fence line (see Fig. 3). This design was found to reduce deer damage in corn fields (Hildreth et al. In Review) while requiring limited costs for fence materials and installation. The effectiveness of such a fence has not yet been tested on elk or on deer with other crops than corn, but has potential to be an easily implemented management tool that could reduce crop depredation. On those plots randomly selected to be winged-fence treatments, we will install a fence with a similar design to Hildreth et al. (In Review), where the crop/wildland interface receives complete protection. For increased height and visual deterrence, the fence will be made of 2.4 meter tall orange barrier material (e.g., Guardian Orange Warning Barrier, Tenax Corporation, USA, Baltimore, Maryland). Monitoring the effectiveness of non-lethal treatments: All treatment and control plots will be monitored for 2 response variables: crop damage and elk/deer incursion. To measure damage to sunflower crops, we will monitor fields every 2 weeks between the time of germination and harvest. We will utilize the variable-area-transect (VAT) method for estimation of crop damage, which consists of both low and high intensity area monitoring (Engeman and Sterner 2002, Gilsdorf et al. 2004a, Gilsdorf et al. 2004b). We will randomly place 4 low-intensity sampling areas within each study plot. In each low-intensity sampling area, we will inspect a row of sunflowers, counting the total number of sunflowers including those that are damaged and undamaged. If 5 cervid-damaged sunflowers are tallied in 100 meters, we will record the distance traveled and the total number of sunflowers. If 5 cervid-damaged sunflowers were not tallied in 100 meters, the observer will record the total number of sunflowers and any cervid-damaged sunflowers observed in that distance. We will calculate the percentage of sunflowers damaged per sampling area using the equation ~ damage per area = [number of damaged sunflowers/(number of 132 �damaged sunflowers+number of undamaged sunflowers)] x 100 (Gilsdorf et al. 2004a, Gilsdorf et al. 2004b). We will also randomly locate 2 high-intensity sampling areas along every treatment and control plot edge to measure damage in proximity to places of high cervid pressure. Within the high-intensity sampling areas, we will use 5 VATs within each area. This will result in 12 total sampling areas (4 low intensity, 8 high intensity) per plot. Additionally, at the end of the season, we will evaluate game damage and year-end yields between treatment and control plots, the ultimate measure of success for each management technique. We will also quantify the level of incursion that occurs into treatment and control fields by elk and mule deer. To do this, we will use animal-activated cameras to record the number and frequency of elk and mule deer that pass through repellents or fence designs into sunflower fields. Cameras will be mounted on posts on the corners of treatment and control fields, capturing images of elk and mule deer inside and outside the field boundaries. Cameras will be positioned along field border that is closest to the agriculture/wildland boundary which is most likely to experience depredation. The Camera type is Moultrie® Game Spy Digital I-65 Infrared, 6.0 mega pixel (Moultry Products, LLC, Alabaster, AL, USA). Cameras can capture images up to 50 feet away, are weather-resistant with a built in solar panel and security box, and can wirelessly transmit images to a private web site for download by project personnel. Cameras will be activated for the duration of the growing season, and at the end of the season we will tally the number of elk and mule deer that penetrated each treatment and control plot boundary. Differences in elk and mule deer use of treatment/control fields will then be tested using repeated measures parametric statistics. This will allow us to evaluate the effectiveness of the repellent, polyrope fence, and winged fence in reducing crop depredation, relative to control plots. Mapping and modeling the spatial juxtaposition of ungulate damage within the landscape: To more effectively address ungulate damage problems we will use ArcGIS software to map crop fields, surrounding habitat types, human development, and topography. These variables have been important in explaining rates of ungulate depredation as damage tends to increase closer to cover, further from roads, and depending on crop palatability (Grover and Thompson 1986, Nixon et al. 1989, Hegel et al. 2009). Information about the location of a crop field in the context of the overall landscape will allow CPW to work with local growers to identify the most appropriate tools, and the timing of their implementation, to reduce damage. To meet this objective we will use satellite imagery to digitize agricultural fields and attribute those fields by crop type. We will use existing landcover, infrastructure, and digital elevation model (DEM) coverages to identify non-agricultural vegetation types, distance to human development, and topographic features (i.e. elevation, slope, aspect), respectively. We can then use landscape variables in conjunction with elk and mule deer location data (see Objective 1) to model the probability that a field is depredated by ungulates (Manly et al. 2002). This model can provide a powerful tool for CPW managers, as they will be able to predict the likelihood of depredation, depending on field location, the surrounding environment, and the crop type, and therefore help landowners specify crop choice or management actions that will reduce elk and deer damage. Timeline The study will take 3 years to complete. Non-lethal management techniques to reduce elk and deer damage will be implemented and monitored during the growing seasons of 2011 and 2012 (June – October), and the results of treatment/control plots will be analyzed thereafter. We will collar elk and deer during August or September 2011, and monitor animals for 2 years (the length of battery life of GPS collars). Once the GPS collars have been retrieved, we will analyze elk and deer location data. We will use that data to conduct damage mapping and modeling over the following ~6 months. 133 �Budget We obtained grants from the Colorado Statewide Habitat Partnership Program, the Montelores Committee Habitat Partnership Program, the Rocky Mountain Elk Foundation, the National Wildlife Research Center and from Colorado Division of Parks and Wildlife Auction/Raffle funds to conduct this work. Below is an itemized project budget. Item EQUIPMENT 20 Elk GPS Collars ($1,300 ea) Capture of Elk ($454 ea + per diem) 20 Deer GPS Collars ($2,500 ea) Capture of Mule Deer ($429 ea + per diem) Plantskydd Materials & Application Polyrope Materials & Installation Winged Fence Materials & Installation Animal-Activated Cameras (20 @ $750 ea) Camera Activation/Maintenance GIS Mapping Leased Truck (Jun-Oct/2 yrs) Gas for Leased Truck (Jun-Oct 2 yrs) PERSONNEL Technician for Monitoring (Jun-Oct/2 yrs) CPW Permanent Employee Salary (2 yrs) NWRC Post-doctoral Salary TOTAL Cost $26,000 $9,330 $50,000 $8,830 $16,710 $32,643 $19,042 $15,000 $4,180 $3,000 $12,000 $5,000 $25,583 $40,000 $12,000 $279,318 E. Location The study will be conducted in the vicinity of Dove Creek, Colorado (Montezuma, San Miguel and Dolores Counties), which is comprised of a mixture of agricultural and public lands. The project will focus on the north half of CPW Game Management Unit 72 and the west half of 711 (the portion west of the Dolores River). The area is generally characterized as mountain shrubland interspersed in irrigated and dryland agricultural areas. The mountain shrub habitat type, which occurs on both private and public lands, is composed primarily of serviceberry, antelope bitterbrush, mountain mahogany, squaw apple and black sagebrush. This habitat type is limited to elevations between 6,500 and 8,500 feet. F. Literature Cited Andelt, W.F., K.P. Burnham, and J.A. Manning. 1991. Relative effectiveness of repellents for reducing mule deer damage. Journal of Wildlife Management 55:341-347. Austin, D.D., P.J. Urness, and D. Duersch. 1998. Alfalfa hay crop loss due to mule deer depredation. Journal of Range Management 51:29-31. Baker, D.L., W.F. Andelt, K.P. Burnham, and W.D. Sheppard. 1999. Effectiveness of hot sauce and Deer Away repellents for deterring elk browsing of aspen sprouts. Journal of Wildlife Management 63:1327-1336. Blejwas, K.M., B.J. Sacks, M.M. Jaeger, and D.R. McCullough. 2002. The effectiveness of selective removal of breeding coyotes in reducing sheep predation. Journal of Wildlife Management 66:451-462. Conover, M.R. 2002. Resolving human-wildlife conflicts: the science of wildlife damage management. CRC Press LLC, Boca Raton, Florida, USA. 134 �Curtis, P.D., M.J. Farigone, and M.E. Richmond. 1994. Preventing deer damage with barrier, electrical, and behavioral fencing systems. Proceedings of the Vertebrate Pest Control Conference 16:223227. Engeman, R.M., R.T. Sterner. 2002. A comparison of potential labor-saving sampling methods for assessing large mammal damage in corn. Crop Protection 21:101-105. Gilsdorf, J.M., S.E. Hygnstrom, K.C. VerCauteren, G.M. Clements, E.E. Blankenship, and R.M. Engeman. 2004a. Evaluation of a deer-activated bio-acoustic frightening device for reducing deer damage in cornfields. Wildlife Society Bulletin 32:515-523. Gilsdorf, J.M., S.E. Hygnstrom, K.C. VerCauteren, G.M. Clements, E.E. Blankenship, and R.M. Engeman. 2004b. Propane exploders and Electronic Guards were ineffective at reducing deer damage in cornfields. Wildlife Society Bulletin 32:524-531. Gotelli, N.J., and A.M. Ellison. 2004. A primer of ecological statistics. Sinauer Associates, Sunderland, Massachusetts. Grover, K.E., and M.J. Thompson. 1986. Factors influencing spring feeding site selection by elk in the Elkhorn Mountains, Montana. Journal of Wildlife Management 50:466-470. Hegel, T.M., C.C. Gates, and D. Eslinger. 2009. The geography of conflict between elk and agricultural values in the Cypress Hills, Canada. Journal of Environmental Management 90:222-235. Hildreth, A.M., S.E. Hygnstrom, E.E. Blankenship, and K.C. VerCauteren. In Review. Efficacy of a partial poly-mesh fence with wings to reduce deer damage to corn. Hygnstrom, S.E. and S.R. Craven. 1988. Electric fences and commercial repellents for reducing deer damage in cornfields. Wildlife Society Bulletin 16:291-296. Jordon, D.M., and M.E. Richmond. 1992. Effectiveness of a vertical 3-wire electric fence modified with attractants or repellents as a deer exclosure. Proceedings of the Eastern Wildlife Damage Control Conference 5:44-47. Krausman, P. R., J. J. Hervert, and L. L. Ordway. 1985. Capturing deer and mountain sheep with a netgun. Wildlife Society Bulletin 13:71–73. McKillop, I.G., and R.M. Sibly. 1988. Animal behaviour at electric fences and implications for management. Mammal Review 18:91-103. Manly, B.F.J., L.L. McDonald, D.L. Thomas, T.L. McDonald, and W.P. Erickson. 2002. Resource selection by animals; statistical design and analysis for field studies. Second Edition. Kluwer Academic Publishers, Boston, Massachussetts. Mould, E.D., and C.T. Robbins. 1982. Digestive capabilities in elk compared to white-tailed deer. Journal of Wildlife Management 46:22-29. Nixon, C.M., L.P. Hansen, P.A. Brewer, J.E. Chelsvig. 1989. Ecology of white-tailed deer in an intensively farmed region of Illinois. Wildlife Monographs 118:1-77. Nolte, D. 1999. Behavioral approaches for limiting depredation by wild ungulates. Pages 60-69 in K.L. Launchbaugh, K.D. Sanders, and J.C. Mosley, editors. Grazing behavior of livestock and wildlife. Idahor Forest, Wildlife & Range Experimental Station Bulletin #70, University of Idaho, Moscow, USA. Porter, W.F. 1983. A baited electric fence for controlling deer damage to orchard seedlings. Wildlife Society Bulletin 11:325-329. Rosenberry, C.S., L.I. Muller, and M.C. Conner. 2001. Movable, deer-proof fencing. Wildlife Society Bulletin 29:754-757. Seamans, T.W., and K.C. VerCauteren. 2006. Evaluation of ElectroBraid™ fencing as a white-tailed deer barrier. Wildlife Society Bulletin 34:8-15. Van Tassell, L.W., C. Phillips, B. Yang. 1999. Depredation claim settlements in Wyoming. Wildlife Society Bulletin 25:886-894. Vecellio, G.M., R.H. Yahner, and G.L. Storm. 1994. Crop damage by deer at Gettysburg Park. Wildlife Society Bulletin 22:89-93. VerCauteren, K.C., M.J. Lavelle, and S. Hygnstrom. 2006. Fences and deer-damage management: a review of designs and efficacy. Wildlife Society Bulletin 34:191-200. 135 �Wagner, K.K., R.H. Schmidt, and M.R. Conover. 1997. Compensation programs for wildlife damage in North America. Wildlife Society Bulletin 25:312-319. Walter, W.D., M.J. Lavelle, J.W. Fischer, T.L. Johnson, S.E. Hygnstrom, and K.C. VerCauteren. 2010. Management of damage by elk (Cervus elaphus) in North America: a review. Wildlife Research 37:630-646. Walter, W.D., K.C. VerCauteren, J.M. Gilsdorf, and S.E. Hygnstrom. 2009. Crop, native vegetation, and biofuels: response of white-tailed deer to changing management priorities. Journal of Wildlife Management 73:339-344. World Resources Institute. 2008. WRI EarthTrends monthly update page. http://earthtrends.org/updates/node/180. Worton, B.J. 1989. Kernal methods for estimating the utilization distribution in home-range studies. Ecology 70:164-168. Worton, B.J. 1995. Using Monte Carlo simulation to evaluate kernel-based home range estimators. Journal of Wildlife Management 59:794-800. Yoder, J. 2002. Deer-inflicted crop damage and crop choice in Wisconsin. Human Dimensions of Wildlife 7:179-196. 136 �Figure 1. Pink areas delineate zones of high ungulate-crop depredation around Dove Creek, Colorado (Montezuma, San Miguel and Dolores Counties; figure from a Montelores HPP report). DOVE CORT 137 �Figure 2. Photo of a polyrope electric fence constructed in a sunflower field south of Dove Creek, CO. Figure 3. Photo along a wing of a temporary fence constructed in a sunflower field south of Dove Creek, CO. 138 �Colorado Division of Parks and Wildlife July 2011 – June 2012 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Colorado 3430 3002 : : : : Division of Parks andWildlife Mammals Research Elk Conservation Evaluating solutions to reduce elk and mule deer damage on agricultural resources. Federal Aid Project No. Period Covered: July 1, 2011 – June 30, 2012 Author: H.E. Johnson; project cooperators, P. Dorsey, M. Hammond, C. Bishop, K. VerCauteren, D. Walter, C. Anderson, and J. Fischer. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Elk and mule deer provide important recreational, ecological, and economic benefits, but they can also cause substantial damage to agricultural resources in rural environments. This situation has generated significant challenges for wildlife agencies that are responsible for maintaining viable ungulate populations while also minimizing crop damage. One of the most severe areas of ungulate damage in Colorado has been the sunflower fields around Dove Creek. In this region, roughly a quarter of million dollars were annually paid to farmers between 2007 and 2009 for depredation caused by elk and deer. The main management tool used by Colorado Parks and Wildlife (CPW) to reduce ungulate damage has been the allocation of kill permits, distribution hunts, and private land only doe/cow hunts; however, tolerance for these permits has been low among local sportsman and the general public. Pressure from local sunflower growers over crop damage, and frustration from the general public over kill permits, generated the need for CPW to evaluate other management options for reducing elk and deer crop depredation. As a result, CPW partnered with wildlife damage researchers from the National Wildlife Research Center to find science-based solutions for reducing crop damage. Collaboratively, our goals are to 1) examine elk and deer distribution and migration patterns around agricultural areas to design public hunting opportunities to reduce depredation, 2) experimentally test a suite of non-lethal fencing techniques to minimize crop damage, and 3) map and model landscape characteristics associated with ungulate damage to specify more effective site-specific management techniques to minimize depredation. During FY11-12 we focused on collecting field data to meet project objectives. Specifically, we constructed experimental fence plots and monitored their effectiveness in reducing elk and deer damage (objective 2) and collared elk and deer to collect information about local movement and distribution patterns (data required to meet objectives 1 and 3). 100 �WILDLIFE RESEARCH REPORT EVALUATING SOLUTIONS TO REDUCE ELK AND MULE DEER DAMAGE ON AGRICULTURAL RESOURCES HEATHER E. JOHNSON P.N. OBJECTIVES To conduct a study on elk and mule deer around the agricultural fields of Dove Creek that 1) examines wild ungulate distribution patterns to design public hunting opportunities to reduce crop damage, 2) experimentally tests a suite of non-lethal fencing techniques to minimize crop depredation, and 3) maps and models landscape characteristics associated with damage to specify more effective site-specific management practices. SEGMENT OBJECTIVES 1. Implement the construction of experimental fence plots on sunflower fields in the vicinity of Dove Creek, including electric fences, temporary winged fences, and chemical repellent fences. 2. Collect field data on elk and deer damage to sunflowers in experimental fence plots throughout the growing season. 3. Capture and collar adult female elk and mule deer around agricultural fields in the vicinity of Dove Creek. INTRODUCTION Elk and deer provide important recreational, ecological, and economic benefits, but they can also cause substantial damage to agricultural resources in rural areas (Austin et al. 1998, Wisdom and Cook 2000). In Colorado, elk and deer crop depredation accounts for a majority of the wildlife damage claims in the state, and CPW is obligated to pay for those lost resources. In recent years, the agency has spent approximately $500,000 on an annual statewide basis to compensate farmers for ungulate depredation. This situation has generated significant challenges for CPW and other wildlife agencies that are responsible for maintaining viable ungulate populations while also minimizing crop damage (Van Tassell et al. 1999, Wagner et al. 1997, Hegel et al. 2009, Walter et al. 2010). Elk and deer crop depredation results from a combination of factors including the seasonal distribution and abundance of local forage resources, landscape configuration, and herd density patterns (Vecellio et al. 1994; Yoder 2002; Hegel et al. 2009). Damage can be highly variable within and among growing seasons, as local patterns in precipitation and temperature will alter the availability of native forage and the motivation of ungulates to feed on agricultural fields (Walter et al. 2010). The juxtaposition of cropland and wildland has also been found to be particularly important in driving damage rates, as those cultivated fields closer to cover experience more damage (Nixon et al. 1989, Hegel et al. 2009). Additionally, studies have found that ungulate damage is often caused by only a subset of individuals in the population, depending on the spatial and social structuring of the herd. These observations have critical implications for wildlife managers, as 1) management practices may be differentially effective based on the variability of native forage conditions and the spatial juxtaposition of other habitat features, and 2) management techniques targeted at specific animals may be more effective than implementing those techniques on the population at large (Blejwas et al. 2002, Hegel et al. 2009). As a result, it is important to understand both the spatial configuration of seasonal resources and the resource selection patterns of different segments of local ungulate populations to successfully identify strategies to reduce elk and deer crop damage (Hegel et al. 2009). 101 �One of the most significant hotspots of elk and mule deer depredation in Colorado has been in the vicinity of Dove Creek, where CPW paid roughly a quarter of million dollars annually to farmers between 2007 and 2009. High damage in this region has been primarily attributed to a recent switch in the crops that are locally grown. Farmers traditionally grew beans, spring and winter wheat, oats, alfalfa and grass hay which had minimal damage by wild ungulates. In recent years, however, local growers have planted sunflowers, a high-value seed oil crop used to produce biofuels, and a crop that is highly desirable to wild ungulates. In addition to this recent switch in crops, ungulate damage around Dove Creek is exacerbated by the spatial distribution of sunflower fields in relation to the surrounding wildlands (e.g., sagebrushmixed shrublands and piñon-juniper woodlands). The region is fractured with deep canyons that provide refugia for elk and deer, and fields adjacent to the canyon rims experience the greatest amount of depredation. With the substantial increase in biofuel production in the U.S. (World Resources Institute 2008), the agricultural conversion observed around Dove Creek will likely become common, as highpriced, crops replace more traditionally-grown, lower-cost crops (Walter et al. 2009). The main management tool available to CPW to reduce ungulate sunflower damage has been to increase harvest through the use of kill permits, distribution hunts, and private land only (PLO) doe/cow hunts, however tolerance for these permits has been low among local sportsman and the general public. Permits are typically allocated to farmers between June and August, when calves and fawns are still dependent on their mothers, reducing the acceptability of female hunts. Additionally, local elk and deer populations are near or below management objectives, creating a paradox where CPW ultimately wants to increase ungulate herds, but reduce crop depredation. Hunting is also economically important around Dove Creek, so there is a strong desire in the local community to have increased public hunting opportunities and reduced PLO damage hunts. Given pressure by farmers over elk and deer sunflower damage, and frustration by sportsmen and the public over kill permits, CPW wildlife managers were interested in finding alternative solutions for reducing sunflower depredation. As a result, personnel from CPW partnered with wildlife-damage researchers from the National Wildlife Research Center (NWRC) to find non-lethal science-based solutions for reducing sunflower depredation. Collaboratively, we developed a proposal to 1) identify public hunting strategies that reduce crop damage, 2) test a suite of non-lethal fencing techniques to minimize crop depredation, and 3) map and model landscape characteristics associated with damage behavior to specify more effective management practices (Johnson et al. 2011). Results from this study should enable CPW and local growers to reduce ungulate crop depredation, leading to a decrease in compensation payments, a decrease in kill permits/distribution hunts, and an increase in public hunting opportunities. In FY11-12 we focused on collecting field data to meet project objectives. Specifically, we constructed experimental fence plots and monitored their effectiveness in reducing elk and deer damage (objective 2) and collared elk and deer to collect information about local movement and distribution patterns (data required to meet objectives 1 and 3). STUDY AREA The area around Dove Creek, Colorado (Montezuma, San Miguel and Dolores counties) is comprised of a mixture of agricultural and public lands. This project focuses on the north half of CPW Game Management Unit 72 and the west half of 711 (the portion west of the Dolores River). The area is generally characterized as mountain shrubland interspersed with irrigated and dryland agricultural fields, ranging from 1,981 to 2,590 m in elevation. The mountain shrub habitat type is primarily composed of serviceberry (Amelanchier alnifolia), bitterbrush (Purshia tridentata), mountain mahogany (Cercocarpus montanus), squaw apple (Peraphyllum ramosissimum) and black sagebrush (Seriphidium novum). 102 �Sunflower fields around Dove Creek are spatially juxtaposed to deep canyons that provide refugia for elk, exacerbating ungulate damage on agricultural crops (Figure 1). METHODS Testing the effectiveness of different fence types for reducing ungulate damage During FY11-12 we constructed experimental fence plots to test the effectiveness of three nonlethal exclusionary fences for reducing elk and deer damage: a polyrope electric fence, a temporary “winged” fence, and an organic repellent “fence.” These differ from traditional exclusionary fencing for elk and deer, in that they are cheaper to construct and can be easily moved among fields over time, as farmers grow sunflowers on a rotational basis. Each fence type is described below: • Polyrope electric fence – The polyrope electric fence acts primarily as psychological barrier for elk and deer based on learned behavioral, avoidance conditioning (McKillop and Sibly 1988). The fences consists of conductive wires which are woven into synthetic electric “ropes” that are more durable, visible, and easy to install than traditional electric fences (Figure 2; Hygnstrom and Craven 1988, VerCauteren et al. 2006). Avoidance conditioning occurs when an animal contacts the fence, often with the nose or tongue, and receives a powerful electric shock. Polyrope fences have had success in reducing deer damage (Hygnstrom and Craven 1988, Seamans and VerCauteren 2006), but have not been experimentally tested for reducing elk damage. For the 5 randomly selected polyrope treatment plots, we constructed a fence approximately 1.8 meter tall with 5 strands to discourage passage under, through, or over the fence. The polyrope was powered by a Speedrite™ 3000 energizer (Tru-Test Incorporated, San Antonio, Texas) using a 12-volt deep-cycle battery with a solar-panel recharger. • Temporary winged fence - For seasonal agricultural resources, such as sunflowers, temporary fences may be sufficient to provide protection from wild ungulates and are inexpensive, lightweight, and easy to erect and remove (Rosenberry et al. 2001, VerCauteren et al. 2006). We tested the effectiveness of a temporary “winged” fence made of polypropylene mesh (Figure 3). The fence is installed completely on one side of the target field, and partially installed on two other sides having 50-100 meter “wings” that extend perpendicular from the full fence line. This design was found to reduce deer damage in corn fields (Hildreth et al. In Review) but has not yet been tested on elk or on deer with crops other than corn. On those plots receiving winged-fence treatments, we installed the fence such that the side receiving complete protection was along the crop/wildland interface. The fence was made of 2.4 meter tall black barrier material (e.g., Guardian Warning Barrier, Tenax Corporation, USA, Baltimore, Maryland) for increased height and visual deterrence. • Plantskydd - Repellents are nonlethal substances that can be used to deter ungulates by decreasing a plant’s palatability (Walter et al. 2010). We will test the effectiveness of a relatively new product, Plantskydd, for reducing sunflower damage around Dove Creek. This product was developed in Sweden to decrease mammalian wildlife damage on commercial forests. It works by emitting an odor that animals associate with predator activity, repelling the animal before it forages on crop plants. There is great interest in the success of this product as it can be easily applied to vegetation by ground and aerial spraying, used on both organic and conventionally grown sunflowers, and is cost-effective for growers. That said, the effectiveness of Plantskydd has not been experimentally tested, only anecdotally reported. To test this method, 5 Plantskydd treatment plots were ground sprayed in a ~20 ft swath around the plot perimeters after germination had begun (as directed by the manufacturer). Plantskydd was reapplied to treatment plots once/month throughout the growing season as the repellent may wash off or decompose over time and needs to be reapplied to new plant material. We constructed the fence plots based on a randomized block design. We identified 5 different sunflower fields to serve as replicates (~160-200 acres in size); all fields had previously suffered high 103 �ungulate crop damage. Within each field we specified 4 10-acre plots, one for each experimental fence treatment type (polyrope fence, temporary winged fence, chemical repellent fence) and a control (Figure 1). The 4 plots were randomly assigned within each field, such that each field (block) contained one replicate of all treatments (Gotelli and Ellison 2004). This design allows us to statistically account for environmental heterogeneity, as we expect that damage will be variable among fields. Within the fields, study plots were spaced as far apart as possible, to account for plot independence. Plots were also placed along the agriculture/wildland boundary, where depredation is expected to be concentrated. Fences were installed by Dillon Fencing (Naturita, CO) during the end of June and early July 2011, after sunflowers had germinated. The 20 plots (experimental and control plots) were delineated were monitored from mid-July through mid-October (time of harvest). Treatment and control plots were examined for 2 key response variables: elk/deer incursion and sunflower damage. We quantified incursion by elk and deer into our plots on a biweekly basis, assessing the permeability of the different fence types. To do this, an observer walked the perimeter of each plot, counting the number of elk and deer tracks entering and exiting the field. Tracks were raked out between observations so they were not double-counted. Differences in the number of elk and mule deer tracks into treatment/control fields were tested using repeated measures ANOVA. In addition to quantifying incursion into experimental plots, we also quantified direct damage to sunflower plants. We assessed damage every 2 weeks using the variable-area-transect (VAT) method for estimation of crop depredation (Engeman and Sterner 2002, Gilsdorf et al. 2004a, Gilsdorf et al. 2004b). In each plot, we conducted 15 VAT transects at random starting points, inspecting a row of sunflowers, and counting the total number of sunflowers that were damaged and undamaged. If 5 cervid-damaged sunflowers were tallied in 100 meters, we recorded the distance traveled and the total number of sunflowers on the transect. If 5 cervid-damaged sunflowers were not tallied in 100 meters, the observer recorded the total number of sunflowers and any cervid-damaged sunflowers observed in that distance. We calculated the percentage of sunflowers damaged per transect using the equation ~ damage = (number of damaged sunflowers) / (number of damaged sunflowers+number of undamaged sunflowers) (Gilsdorf et al. 2004a, Gilsdorf et al. 2004b). Additionally, at the end of the season, we had an agricultural assessor evaluate game damage and year-end yields between treatment and control plots, the ultimate measure of success for each management technique. Just prior to the sunflower harvest in mid-October 2011, we removed all fencing materials from our study fields. The materials were stored over the winter by CPW and re-deployed to 4 different sunflower fields in June 2012 for the second year of testing. Collaring elk and deer to collect information on movement and distribution To obtain data on ungulate movement and distribution patterns we contracted Quicksilver Air to capture and collar 20 adult female elk and 20 adult female deer using a net gun from a helicopter (Krausman et al. 1985). Females were the target of collaring efforts because they cause a majority of the crop depredation and should provide valuable insight into herd distributions. Helicopter captures were scheduled from 11-13 October 2011, just prior to the start of first rifle season. There was a narrow window in which to capture animals, as helicopter operations could only occur after the heat of the summer had passed, but before rifle season had begun (to minimize impacts to hunters). Captured elk and deer were hobbled and blindfolded, fitted with a global positioning system (GPS) collar, aged, measured and released. GPS collars were programmed to collect a location every 4 hours for 2 years, and then drop off the animals in fall 2013. The collars are “store-on-board,” meaning that the data can only be downloaded once the collar is retrieved from the field. Until collars drop-off, we are conducting monthly aerial telemetry flights to monitor survival and obtain some general location information. 104 �Once GPS collar data has been retrieved, elk and mule deer locations will be used to map seasonal distribution and migration patterns in ArcGIS. This should allow CPW to design public hunts that will target conflict elk and mule deer, while minimizing the need for PLO hunts and kill permits. For example, the Utah Division of Wildlife Resources is willing to consider special elk hunts south of Hwy 491 if we find that any or all of the resident elk herds (causing damage) spend portions of the year in Utah. Locations will also allow us to determine the amount of use of crop fields by elk and deer, and the proportion of animals using crop fields (whether it is only certain segments of the population, or the population at large). Animal location data will also be used to model ungulate damage potential in relation to field locations, surrounding habitat types, human development, and topography. These variables have been important in explaining rates of ungulate depredation, as damage tends to increase closer to cover, further from roads, and depending on crop palatability (Grover and Thompson 1986, Nixon et al. 1989, Hegel et al. 2009). Information about the location of a crop field in the context of the overall landscape will allow CPW to work with local growers to identify appropriate management tools, and the timing of their implementation, to reduce game damage. Such a model will serve as a powerful tool for CPW managers, as they will be able to predict the likelihood of depredation for different fields, depending on location, the surrounding environment, and the crop type, and therefore help landowners specify crop choice or management actions to reduce damage. RESULTS AND DISCUSSION Between 20 July 2011 and 20 October 2011 we conducted 137 incursion surveys of the 20 fence plots and 2,100 sunflower ungulate damage assessments. We used repeated measures ANOVA to determine whether there were statistically significant differences in elk and deer incursion into each fence treatment type. We found that incursion varied by treatment for both elk and deer (Elk: F20, 116 = 6.84, P < 0.001; Deer: F20, 116 = 6.24, P < 0.001) such that the electric fence plots had the fewest elk and deer tracks entering the plot, followed by the winged fences, the Plantskydd repellant fences, and the control plots (Figure 4). Biweekly damage assessments of the sunflower fields showed that crop damage followed the same general trends as the frequency of elk and deer entering the treatment plots. Generally, the electric fence plots received the least elk and deer damage, followed by the winged fence, and then by the Plantskydd treatment; the control fields had the highest levels of ungulate damage (Table 1, Figure 5). As expected, the percentage of damaged plants/plot generally increased throughout the growing season, except during the last two assessments. This pattern may have resulted from differences in an observer’s ability to detect damage at different stages of sunflower growth. At the end of the growing season, when the plants are dry and sunflower heads are bent over, damage may be harder to detect than at earlier stages of growth (i.e., when the heads are upright and the leaves are erect). Damage, however, was generally minimal in 2011 across all fields and plots (<1% in plots with electric fences and ~4% in control plots). We suspect that minimal damage was the result of abundant natural forage for elk and deer, as late spring rains in 2011 generated more forage than is typically observed in the vicinity of Dove Creek during summer. Indeed, CPW did not pay out any damage claims to farmers for elk and deer crop depredation in 2011, as wild ungulates were not readily observed on fields. We plan on constructing and monitoring experimental fence plots for a second year in 2012, to test the effectiveness of these treatments when sunflower fields experience more typical rates of damage. Quicksilver Air captured and collared 20 adult female deer and 9 adult female elk. Although deer were readily available for capture throughout the study area, the helicopter crew had a difficult time finding elk in the study area. Wildlife managers suspect that the elk had already left the agriculture areas 105 �around Dove Creek, and had potentially crossed the Utah border by that point in the fall. We plan on trying to ground dart elk during summer 2012 to deploy the remaining elk collars. We conducted monthly aerial telemetry flights for collared animals to track survival and general movement patterns. Four deer died during winter 2011. One deer died in late October, likely due to capture related causes (D12). The other 3 mortalities occurred in February and March 2011, one from a vehicle collision (D4) and the other two from unknown causes (D8 and D19). GPS collars were retrieved from all mortalities so that the data could be downloaded and processed (Figure 6). This information will be used during FY13-14 to map and model seasonal ungulate distributions, game damage potential, and management options for farmers. SUMMARY AND FUTURE PLANS During FY11-12 we constructed the experimental fence plots for the first year of fieldwork, quantified elk and deer damage across our different fencing treatment types, and collared elk and deer in the study area. In FY12-13 we will conduct the fencing experiments for a second field season, and attempt to deploy our remaining elk collars via ground darting. We will continue to monitor the survival and movements of collared animals on a monthly basis using aerial telemetry, until collars detach from the animals in fall 2013. The benefits of this project include gaining knowledge about local elk and deer movements and distribution relative to agricultural fields, identifying non-lethal techniques for reducing ungulate damage to sunflowers and other crops, the development of models to identify areas highly susceptible to damage based on landscape characteristics, and the potential to redesign public hunting opportunities to increase opportunity while reducing those resident animals causing a majority of the damage. LITERATURE CITED Austin, D.D., P.J. Urness, and D. Duersch. 1998. Alfalfa hay crop loss due to mule deer depredation. Journal of Range Management 51:29-31. Blejwas, K.M., B.J. Sacks, M.M. Jaeger, and D.R. McCullough. 2002. The effectiveness of selective removal of breeding coyotes in reducing sheep predation. Journal of Wildlife Management 66:451-462. Engeman, R.M., R.T. Sterner. 2002. A comparison of potential labor-saving sampling methods for assessing large mammal damage in corn. Crop Protection 21:101-105. Gilsdorf, J.M., S.E. Hygnstrom, K.C. VerCauteren, G.M. Clements, E.E. Blankenship, and R.M. Engeman. 2004a. Evaluation of a deer-activated bio-acoustic frightening device for reducing deer damage in cornfields. Wildlife Society Bulletin 32:515-523. Gilsdorf, J.M., S.E. Hygnstrom, K.C. VerCauteren, G.M. Clements, E.E. Blankenship, and R.M. Engeman. 2004b. Propane exploders and Electronic Guards were ineffective at reducing deer damage in cornfields. Wildlife Society Bulletin 32:524-531. Gotelli, N.J., and A.M. Ellison. 2004. A primer of ecological statistics. Sinauer Associates, Sunderland, Massachusetts. Hegel, T.M., C.C. Gates, and D. Eslinger. 2009. The geography of conflict between elk and agricultural values in the Cypress Hills, Canada. Journal of Environmental Management 90:222-235. Hildreth, A.M., S.E. Hygnstrom, E.E. Blankenship, and K.C. VerCauteren. In Review. Efficacy of a partial poly-mesh fence with wings to reduce deer damage to corn. Hygnstrom, S.E. and S.R. Craven. 1988. Electric fences and commercial repellents for reducing deer damage in cornfields. Wildlife Society Bulletin 16:291-296. Johnson, H.E., P.Dorsey, M. Hammond, C. Bishop, K. VerCauteren, D. Walter, and C. Anderson. 2011. Evaluating solutions to reduce elk and mule deer damage on agricultural resources. Study Proposal, Colorado Division of Parks and Wildlife, Fort Collins, USA. 106 �Krausman, P. R., J. J. Hervert, and L. L. Ordway. 1985. Capturing deer and mountain sheep with a netgun. Wildlife Society Bulletin 13:71–73. McKillop, I.G., and R.M. Sibly. 1988. Animal behaviour at electric fences and implications for management. Mammal Review 18:91-103. Nixon, C.M., L.P. Hansen, P.A. Brewer, J.E. Chelsvig. 1989. Ecology of white-tailed deer in an intensively farmed region of Illinois. Wildlife Monographs 118:1-77. Rosenberry, C.S., L.I. Muller, and M.C. Conner. 2001. Movable, deer-proof fencing. Wildlife Society Bulletin 29:754-757. Seamans, T.W., and K.C. VerCauteren. 2006. Evaluation of ElectroBraid™ fencing as a white-tailed deer barrier. Wildlife Society Bulletin 34:8-15. Van Tassell, L.W., C. Phillips, B. Yang. 1999. Depredation claim settlements in Wyoming. Wildlife Society Bulletin 25:886-894. Vecellio, G.M., R.H. Yahner, and G.L. Storm. 1994. Crop damage by deer at Gettysburg Park. Wildlife Society Bulletin 22:89-93. VerCauteren, K.C., M.J. Lavelle, and S. Hygnstrom. 2006. Fences and deer-damage management: a review of designs and efficacy. Wildlife Society Bulletin 34:191-200. Wagner, K.K., R.H. Schmidt, and M.R. Conover. 1997. Compensation programs for wildlife damage in North America. Wildlife Society Bulletin 25:312-319. Walter, W.D., M.J. Lavelle, J.W. Fischer, T.L. Johnson, S.E. Hygnstrom, and K.C. VerCauteren. 2010. Management of damage by elk (Cervus elaphus) in North America: a review. Wildlife Research 37:630-646. Walter, W.D., K.C. VerCauteren, J.M. Gilsdorf, and S.E. Hygnstrom. 2009. Crop, native vegetation, and biofuels: response of white-tailed deer to changing management priorities. Journal of Wildlife Management 73:339-344. World Resources Institute. 2008. WRI EarthTrends monthly update page. http://earthtrends.org/updates/node/180. Wisdom, M.J., and J.G. Cook. 2000. North American Elk. Pages 694-735 in S. Demarais and P.R. Krausman, editors. Ecology and management of large mammals in North America. Prentice Hall, Upper Saddle River, New Jersey, USA. Yoder, J. 2002. Deer-inflicted crop damage and crop choice in Wisconsin. Human Dimensions of Wildlife 7:179-196. Prepared by _______________________________________ Heather E. Johnson, Wildlife Researcher 107 �Table 1. Average percentage of damaged plants/transect during successive damage assessments across the 2011 sunflower growing season. Averages are displayed by field and treatment plot. FIELD/PLOT Guynes Control Electric Plantskydd Winged Schear-Brewer Control Electric Plantskydd Winged Schear-Homestead Control Electric Plantskydd Winged Schear-Hudgeons Control Electric Plantskydd Winged Warren Control Electric Plantskydd Winged 1 2 3 ASSESSMENT 4 5 6 7 1.7 0.0 0.8 0.0 5.2 0.1 4.5 0.3 5.6 0.1 6.6 0.9 6.9 0.2 2.5 0.6 9.2 0.4 3.9 1.0 8.9 0.2 7.4 1.0 1.5 0.2 1.5 0.7 0.3 0.0 0.0 0.0 1.1 0.1 6.5 0.3 0.7 0.7 2.3 1.8 1.0 0.4 1.9 0.7 2.6 0.6 4.5 1.1 1.2 0.6 1.0 1.5 1.4 0.4 2.2 1.1 0.0 0.0 0.0 0.1 7.1 0.1 0.4 0.3 3.0 0.2 3.2 5.6 7.7 0.1 2.4 3.1 3.1 0.2 9.4 8.1 7.5 0.1 5.3 5.6 2.5 0.2 1.6 2.1 0.1 0.0 0.0 0.1 0.6 0.5 0.3 0.5 0.7 0.9 0.4 1.1 1.3 0.8 0.6 0.5 0.5 0.8 0.4 2.5 0.7 0.5 0.3 1.1 0.6 0.7 0.4 0.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.1 0.1 0.1 0.1 0.0 0.1 0.0 0.1 0.1 0.0 0.1 0.2 0.1 0.1 0.2 0.3 0.2 0.0 0.1 108 �Figure 1. Placement of experimental fence plots within the 5 replicate sunflower fields during the 2011 growing season (July – October). Fields are located adjacent to wildland canyons. 109 �Figure 2. Photo of a polyrope electric fence constructed in a sunflower field south of Dove Creek. Figure 3. Photo along a winged temporary fence constructed in a sunflower field south of Dove Creek. 110 �Figure 4. Mean number of deer and elk that crossed into experimental fence plots on a biweekly basis, by treatment type, during damage surveys throughout the growing season (results generated from a repeated measures ANOVA). Mean Number of Ungulates 7 Deer Elk 6 5 4 3 2 1 0 control electric plantskydd winged Treatment Figure 5. Average percentage of damaged plants/transect for biweekly assessments across all fields for different treatment types. 111 �Figure 6. GPS collar locations from deer mortalities during FY11-12 in the vicinity of Dove Creek, CO. 112 �Colorado Parks and Wildlife July 2012 – June 2013 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No. Colorado 3430 3002 Federal Aid Project No. : : : : : Parks and Wildlife Mammals Research Elk Conservation Evaluating solutions to reduce elk and mule deer damage on agricultural resources Period Covered: July 1, 2012 – June 30, 2013 Author: H.E. Johnson; project cooperators, P. Dorsey, M. Hammond, C. Bishop, K. VerCauteren, D. Walter, C. Anderson, and J. Fischer. All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT Elk and mule deer provide important recreational, ecological, and economic benefits, but they can also cause substantial damage to agricultural resources in rural environments. This situation has generated significant challenges for wildlife agencies that are responsible for maintaining viable ungulate populations while also minimizing crop damage. One of the most severe areas of ungulate damage in Colorado has been the sunflower fields around Dove Creek. In this region, roughly a quarter of million dollars were annually paid to farmers between 2007 and 2009 for depredation caused by elk and deer. The main management tool used by Colorado Parks and Wildlife (CPW) to reduce ungulate damage has been the allocation of kill permits, distribution hunts, and private land only doe/cow hunts; however, tolerance for these permits has been low among hunters and the general public. Pressure from local sunflower growers over crop damage, and frustration from the public over kill permits, generated the need for CPW to evaluate other management options for reducing elk and deer crop depredation. As a result, CPW partnered with wildlife damage researchers from the National Wildlife Research Center to find sciencebased solutions for reducing crop damage. Collaboratively, our goals are to 1) experimentally test a suite of non-lethal exclusion and repellent techniques to minimize crop damage, 2) examine elk and deer distribution and migration patterns around agricultural areas to design public hunting opportunities to reduce depredation, and 3) map and model landscape characteristics associated with ungulate damage to specify more effective site-specific management techniques. During FY12-13 we conducted the second and final year of an experiment to test exclusionary techniques for reducing elk and deer damage (objective 1), and submitted a scientific manuscript for publication on the results (Appendix 1). We also monitored collared elk and deer on a monthly basis to collect location information and to retrieve GPS collars from mortalities (data required to meet objectives 2 and 3). 67 �WILDLIFE RESEARCH REPORT EVALUATING SOLUTIONS TO REDUCE ELK AND MULE DEER DAMAGE ON AGRICULTURAL RESOURCES HEATHER E. JOHNSON PROJECT NARRITIVE OBJECTIVES To conduct a study on elk and mule deer around the agricultural fields of Dove Creek that 1) experimentally tests a suite of non-lethal exclusion and repellent techniques to minimize crop depredation, 2) examines wild ungulate distribution patterns to design public hunting opportunities to reduce crop damage, and 3) maps and models landscape characteristics associated with damage to specify more effective site-specific management practices. SEGMENT OBJECTIVES 1. Implement the second (and final) year of an experiment to assess non-lethal techniques to exclude or repel deer and elk from sunflower fields in the vicinity of Dove Creek. 2. Conduct data analysis on the exclusionary treatment experiment and submit a scientific manuscript with research results (Appendix 1). 3. Monitor collared deer and elk on a monthly basis for movements, survival and collar retrieval. INTRODUCTION Elk and deer provide important recreational, ecological, and economic benefits, but they can also cause substantial damage to agricultural resources in rural areas (Austin et al. 1998, Wisdom and Cook 2000). In Colorado, elk and deer crop depredation accounts for a majority of the wildlife damage claims in the state, and CPW is obligated to pay for those lost resources. In recent years, the agency has spent approximately $500,000 on an annual statewide basis to compensate farmers for ungulate depredation. This situation has generated significant challenges for CPW and other wildlife agencies that are responsible for maintaining viable ungulate populations while also minimizing crop damage (Wagner et al. 1997, Van Tassell et al. 1999, Hegel et al. 2009, Walter et al. 2010). Elk and deer crop depredation results from a combination of factors including the seasonal distribution and abundance of local forage resources, landscape configuration, and herd density patterns (Vecellio et al. 1994; Yoder 2002; Hegel et al. 2009). Damage can be highly variable within and among growing seasons, as local patterns in precipitation and temperature will alter the availability of native forage and the motivation of ungulates to feed on agricultural fields (Walter et al. 2010). The juxtaposition of cropland and wildland has also been found to be particularly important in driving damage rates, as those cultivated fields closer to cover experience more damage (Nixon et al. 1989, Hegel et al. 2009). Additionally, studies have found that ungulate damage is often caused by only a subset of individuals in the population, depending on the spatial and social structuring of the herd. These observations have critical implications for wildlife managers, as 1) management practices may be differentially effective based on the variability of native forage conditions and the spatial juxtaposition of other habitat features, and 2) management techniques targeted at specific animals may be more effective than implementing those techniques on the population at large (Blejwas et al. 2002, Hegel et al. 2009). As a result, it is important to understand both the spatial configuration of seasonal resources and the resource selection patterns of different segments of local ungulate populations to successfully identify strategies to reduce elk and deer crop damage (Hegel et al. 2009). 68 �One of the most significant hotspots of elk and mule deer depredation in Colorado has been in the vicinity of Dove Creek, where CPW paid roughly a quarter of million dollars annually to farmers between 2007 and 2009. High damage in this region has been primarily attributed to a recent switch in the crops that are locally grown. Farmers traditionally grew beans, spring and winter wheat, oats, alfalfa and grass hay which had minimal damage by wild ungulates. In recent years, however, local growers have planted sunflowers, a high-value seed oil crop used to produce biofuels, and a crop that is highly desirable to wild ungulates. In addition to this recent switch in crops, ungulate damage around Dove Creek is exacerbated by the spatial distribution of sunflower fields in relation to the surrounding wildlands (e.g., sagebrushmixed shrublands and piñon-juniper woodlands). The region is fractured with deep canyons that provide refugia for elk and deer, and fields adjacent to the canyon rims experience the greatest amount of depredation. With the substantial increase in biofuel production in the U.S. (World Resources Institute 2008), the agricultural conversion observed around Dove Creek will likely become common, as highpriced crops replace more traditionally-grown, lower-cost crops (Walter et al. 2009). The main management tool available to CPW to reduce ungulate sunflower damage has been to increase harvest through the use of kill permits, distribution hunts, and private land only (PLO) doe/cow hunts, however tolerance for these permits has been low among hunters and the general public. Permits are typically allocated to farmers between June and August, when calves and fawns are still dependent on their mothers, reducing the acceptability of female hunts. Additionally, local elk and deer populations are near or below management objectives, creating a paradox where CPW ultimately wants to increase ungulate herds, but reduce crop depredation. Hunting is also economically important around Dove Creek, so there is a strong desire in the local community to have increased public hunting opportunities and reduced PLO damage hunts. Given pressure by farmers over elk and deer sunflower damage, and frustration by hunters and the public over kill permits, CPW wildlife managers were interested in finding alternative solutions for reducing sunflower depredation. As a result, personnel from CPW partnered with wildlife-damage researchers from the National Wildlife Research Center (NWRC) to find non-lethal, science-based solutions for reducing sunflower depredation. Collaboratively, we developed a proposal to 1) test a suite of non-lethal exclusionary techniques to minimize crop depredation, 2) identify public hunting strategies that reduce crop damage, and 3) map and model landscape characteristics associated with damage behavior to specify more effective site-specific management practices (Johnson et al. 2011). Results from this study should enable CPW and local growers to reduce ungulate crop depredation, leading to a decrease in compensation payments, a decrease in kill permits/distribution hunts, and an increase in public hunting opportunities. In FY 2012-13 we completed objective 1, and continued to monitor collared elk and deer for objectives 2 and 3. Specifically, we conducted the second (and final) year of the experiment to assess non-lethal exclusion and repellent techniques for reducing elk and deer damage. We analyzed data from the experiment (collected during the growing seasons of 2011 and 2012) and submitted a scientific manuscript with research results for publication (currently in revision, Appendix 1). We also monitored collared elk and deer on a monthly basis to collect information about local movement and distribution patterns, and to retrieve collars from mortalities (GPS collar data will be used to meet objectives 2 and 3). STUDY AREA The area around Dove Creek, Colorado (Montezuma, San Miguel and Dolores counties) is comprised of a mixture of agricultural and public lands. This project focuses on the north half of CPW Game Management Unit 72 and the west half of 711 (the portion west of the Dolores River). The area is generally characterized as mountain shrubland interspersed with irrigated and dryland agricultural fields, ranging from 1,981 to 2,590 m in elevation. The mountain shrub habitat type is primarily composed of 69 �serviceberry (Amelanchier alnifolia), bitterbrush (Purshia tridentata), mountain mahogany (Cercocarpus montanus), squaw apple (Peraphyllum ramosissimum) and black sagebrush (Seriphidium novum). Sunflower fields around Dove Creek are spatially juxtaposed to deep canyons that provide refugia for elk, exacerbating ungulate damage on agricultural crops (see Appendix 1, Figure 1). METHODS Testing the effectiveness of different exclusionary treatment types for reducing ungulate damage During the sunflower growing seasons (Jul-Oct) of 2011 and 2012, we constructed experimental plots to test the effectiveness of three non-lethal exclusion and repellent techniques for reducing elk and deer damage: a polyrope electric fence, a temporary “winged” fence, and an organic repellent. These methods differ from traditional exclusionary fencing for elk and deer, in that they are cheaper to construct and can be easily moved among fields over time, as farmers grow sunflowers on a rotational basis. Each exclusionary treatment is described below: • • • Polyrope electric fence – The polyrope electric fence acts primarily as psychological barrier for elk and deer based on learned behavioral, avoidance conditioning (McKillop and Sibly 1988). The fences consists of conductive wires which are woven into synthetic electric “ropes” that are more durable, visible, and easy to install than traditional electric fences (Appendix 1, Figure 2; Hygnstrom and Craven 1988, VerCauteren et al. 2006). Avoidance conditioning occurs when an animal contacts the fence, often with the nose or tongue, and receives a powerful electric shock. Polyrope fences have had success in reducing deer damage (Hygnstrom and Craven 1988, Seamans and VerCauteren 2006), but have not been experimentally tested for reducing elk damage. We constructed polyrope fences approximately 1.8 meters tall with 5 strands to discourage passage under, through, or over the fence. The polyrope was powered by a Speedrite™ 3000 energizer (Tru-Test Incorporated, San Antonio, Texas) using a 12-volt deepcycle battery with a solar-panel recharger. Temporary winged fence - For seasonal agricultural resources, such as sunflowers, temporary fences may be sufficient to provide protection from wild ungulates and are inexpensive, lightweight, and easy to erect and remove (Rosenberry et al. 2001, VerCauteren et al. 2006). We tested the effectiveness of a temporary “winged” fence made of polypropylene mesh (Appendix 1, Figure 2). The fence is installed completely on one side of the target field, and partially installed on two other sides having 50-100 meter “wings” that extend perpendicular from the full fence line. This design was found to reduce deer damage in corn fields (Hildreth et al. 2012) but has not yet been tested on elk or on deer with crops other than corn. On those plots receiving winged-fence treatments, we installed the fence such that the side receiving complete protection was along the crop/wildland interface. The fence was made of 2.4 meter tall black barrier material (e.g., Guardian Warning Barrier, Tenax Corporation, USA, Baltimore, Maryland) for increased height and visual deterrence. Plantskydd - Repellents are nonlethal substances that can be used to deter ungulates by decreasing a plant’s palatability (Walter et al. 2010). We tested the effectiveness of a relatively new product, Plantskydd, for reducing sunflower damage around Dove Creek. This product was developed in Sweden to decrease mammalian wildlife damage on commercial forests. It works by emitting an odor that animals associate with predator presence, repelling the animal before it forages on crop plants. There is great interest in the success of this product as it can be easily applied to vegetation by ground and aerial spraying, used on both organic and conventionally grown sunflowers, and is cost-effective for growers. That said, the effectiveness of Plantskydd has not been experimentally tested, only anecdotally reported. To test this method, Plantskydd treatment plots were ground sprayed in a swath around the plot perimeters after germination had begun (as directed by the manufacturer). Plantskydd was reapplied to treatment plots once/month 70 �throughout the growing season as the repellent may wash off or decompose over time and needs to be reapplied to new plant material. We constructed treatment plots based on a randomized block design. We identified 5 sunflower fields to serve as replicates in 2011 and 4 fields in 2012 (~160-200 acres in size); all fields had previously suffered high ungulate crop damage. Within each field we specified 4 10-acre plots, one for each experimental treatment type (polyrope fence, temporary winged fence, chemical repellent fence) and a control. The 4 plots were randomly assigned within each field, such that each field (block) contained one replicate of all treatments (Gotelli and Ellison 2004). This design allowed us to statistically account for environmental heterogeneity, as we expected that damage would be variable among fields. Within the fields, study plots were spaced as far apart as possible, to account for plot independence. Plots were also placed along the agriculture/wildland boundary, where depredation was expected to be concentrated. Fences were installed during the end of June and early July after sunflowers had germinated. Experimental plots were monitored from mid-July through mid-October (just prior to harvest). Treatment and control plots were examined for 2 key response variables: sunflower damage and the number of elk and deer tracks that crossed plot perimeters. We used the variable-area-transect method for estimation of crop damage (Engeman and Sugihara 1998; Engeman and Sterner 2002; Gilsdorf et al. 2004a, b), conducting final damage assessments immediately before harvest (mid-Oct). In 2011 we assessed damage on 15 transects/plot, and in 2012 we increased the number to 30 transects/plot. For each transect, we randomly (with replacement) identified a starting location within the plot and inspected a row of sunflowers, counting the total number of sunflower plants, and the number of plants that were damaged by deer or elk. Typical damage was characterized by the removal of the terminal bud, consumption of the seed head and trampling of the plants, verified by accompanying cervid tracks. If 5 cervid-damaged sunflowers were tallied within 100 m, we recorded the distance traveled to the fifth damaged plant (<100 m) and the total number of sunflower plants observed within that distance. If 5 cervid-damaged sunflowers were not tallied within 100 m, the observer recorded the total number of sunflowers and the number of cervid-damaged plants counted within that distance. If the end of the sunflower row was reached before completing a transect, the observer would randomly select an adjacent row (i.e., right or left row) to complete the transect. We calculated mean proportion of end-of-season damage for each treatment and control plot, and mean number of elk and deer tracks traversing plot perimeters for each plot across the growing season. We used a generalized linear mixed model to identify whether exclusion or repellent treatment types were effective in reducing cervid damage to sunflower plots (Pinheiro and Bates 2000). Because damage data were recorded for each transect as the number of damaged plants/total plants, we used a binomial distribution with a logit link function. To evaluate the influence of exclusion techniques on deer and elk tracks traversing plot perimeters, we used generalized linear mixed models with Poisson distributions and log link functions. We generated separate models for predicting the number of tracks by deer and elk, as we hypothesized that treatments may vary in their effectiveness among cervid species. We used model coefficients to assess the direction and magnitude of different treatment types on cervid damage and plot use (95% confidence intervals non-overlapping zero). Collaring elk and deer to collect information on movement and distribution To obtain data on ungulate movement and distribution patterns, we captured and collared adult female elk and mule deer using a net gun from a helicopter in fall 2011 (Krausman et al. 1985). Females were the target of collaring efforts because they cause a majority of the crop depredation and should provide valuable insight into herd distributions. Captured elk and deer were hobbled and blindfolded, fitted with a global positioning system (GPS) collar, aged, measured and released. GPS collars were programmed to collect a location every 4 hours for 2 years, and then drop off the animals in fall 2013. The collars are “store-on-board,” meaning that the data can only be downloaded once the collar is 71 �retrieved from the field. Until collars drop-off, we are conducting monthly aerial telemetry flights to obtain some general location information and to monitor mortalities so collars can be retrieved from the field. Once GPS collar data has been retrieved, elk and mule deer locations will be used to map seasonal distribution and migration patterns in ArcGIS. This should allow CPW to design public hunts that will target conflict elk and mule deer, while minimizing the need for PLO hunts and kill permits. Animal location data will also be used to model ungulate damage potential in relation to field locations, surrounding habitat types, human development, and topography. These variables have been important in explaining rates of ungulate depredation, as damage tends to increase closer to cover, further from roads, and depending on crop palatability (Grover and Thompson 1986, Nixon et al. 1989, Hegel et al. 2009). Information about the location of a crop field in the context of the overall landscape will allow CPW to work with local growers to identify appropriate management tools, and the timing of their implementation, to reduce game damage. Data analysis for this portion of the project will be primarily conducted by collaborators at the USDA National Wildlife Research Center. RESULTS AND DISCUSSION Exclusion and Repellent Treatments - During summers 2011 and 2012, we conducted 3,288 damage transects and 233 track surveys across all treatment plots. The percentage of sunflowers damaged by cervids across plots and years ranged from 0.0% to 72.6% ( x = 8.3%, SE = 0.8). The mean bimonthly number of deer tracks crossing plot perimeters ranged from 0 to 149.8 ( x = 23.0, SE = 5.3) and the mean number of elk tracks ranged from 0 to 21.6 ( x = 5.3, SE = 1.1). Mean percentage sunflower damage and number of deer tracks were greater in 2012 than in 2011 (damage: t = -3.300, df = 29, P = 0.003 [Fig. 3a]; deer tracks: t = -4.512, df = 34, P < 0.001; Fig. 3b), but mean values for elk tracks were similar between years (t = 0.371, df = 34, P = 0.713). In 2011, treatment and control plots averaged 0.9% sunflower plant damage at the end of the growing season, and a bimonthly average of 6.0 deer and 5.7 elk tracks crossed plot boundaries. Conversely, 2012 plots had an average of 17.1% of plants damaged at harvest and an average of 44.4 deer tracks and 4.9 elk tracks crossed plot boundaries on a bimonthly basis. We found that electric fencing was the only treatment that significantly reduced damage and plot use by deer and elk (see Appendix 1 for details). Across years, the mean proportion of damaged plants on electric fence plots was 0.01 (95% CI: 0.00 – 0.03), on control plots was 0.05 (95% CI: 0.00 – 0.33), on repellent fences was 0.04 (95% CI: 0.01 – 0.15) and on winged fences was 0.04 (95% CI: 0.01 – 0.15). The average bimonthly number of deer tracks that crossed plot perimeters on plots with electric fencing was 0.6 (95% CI: 0.3 – 1.1), on control plots was 18.5 (95% CI: 3.8 – 91.5), on repellent fence plots was 18.4 (95% CI: 11.4 – 29.7), and on winged plots was 16.8 (95% CI: 10.4 – 27.0). Electric fences also reduced the number of elk that crossed plot perimeters on a bimonthly basis, but the effect was lesser than for deer. An average of only 0.1 elk tracks crossed electric fence plot boundaries (95% CI: 0.0 – 0.2), while 4.3 crossed control plots (95% CI: 1.8 – 10.3), 3.4 crossed repellent plots (95% CI: 2.2-5.2) and 3.7 crossed winged plots (95% CI: 2.4 – 5.7). For wildlife agencies seeking non-lethal management options for reducing elk and deer damage to high-value agricultural crops, we found that 5-strand polyrope electric fencing was effective. Polyrope is easy to assemble/disassemble, cost-effective relative to permanent fencing, and can be used on a temporary basis to minimize damage for certain crops grown on rotation or during years when natural forage for cervids is scarce. In areas where management agencies are working to maintain or increase deer and elk populations, but reduce cervid damage, the application of an effective exclusion technique like polyrope electric fencing could protect high-value crops, decrease the need for compensation payments and lethal cervid depredation permits, and increase satisfaction of producers and the public. Wildlife agencies will need to continue to work with producers to test and apply management techniques 72 �for crop protection based on the wildlife species present, population densities, crop types, landscape configuration, and abundance of local forage. Monitoring Collared Deer and Elk - We conducted monthly aerial telemetry flights for collared animals to track survival and general movement patterns. Three deer and one elk died during FY12-13 from unknown causes. GPS collars were retrieved from all mortalities so that the data could be downloaded and processed. This information will be used during FY13-14 to map and model seasonal ungulate distributions, game damage potential, and management options for sunflower producers. SUMMARY AND FUTURE PLANS During FY12-13 we completed the non-lethal, exclusionary treatment experiment, analyzed ungulate damage and use on plots by treatment type, submitted a scientific manuscript for publication with experiment results (Appendix 1), and monitored collared elk and deer in the study area. In FY13-14 we will continue to monitor the survival and movements of collared animals on a monthly basis using aerial telemetry, collect collars in the field once they drop off animals in fall 2013, and map and model GPS location data. The benefits of this project include the identification of non-lethal techniques for successfully reducing ungulate damage to sunflowers and other crops, gaining knowledge about local elk and deer movements and distribution relative to agricultural fields, and the development of models to identify areas highly susceptible to damage based on landscape characteristics. LITERATURE CITED Austin, D.D., P.J. Urness, and D. Duersch. 1998. Alfalfa hay crop loss due to mule deer depredation. Journal of Range Management 51:29-31. Blejwas, K.M., B.J. Sacks, M.M. Jaeger, and D.R. McCullough. 2002. The effectiveness of selective removal of breeding coyotes in reducing sheep predation. Journal of Wildlife Management 66:451-462. Engeman, R.M., R.T. Sterner. 2002. A comparison of potential labor-saving sampling methods for assessing large mammal damage in corn. Crop Protection 21:101-105. Engeman, R. M., and R. T. Sugihara. 1998. Optimization of variable area transect sampling using Monte Carlo simulation. Ecology 79:1425-1434. Gilsdorf, J.M., S.E. Hygnstrom, K.C. VerCauteren, G.M. Clements, E.E. Blankenship, and R.M. Engeman. 2004a. Evaluation of a deer-activated bio-acoustic frightening device for reducing deer damage in cornfields. Wildlife Society Bulletin 32:515-523. Gilsdorf, J.M., S.E. Hygnstrom, K.C. VerCauteren, G.M. Clements, E.E. Blankenship, and R.M. Engeman. 2004b. Propane exploders and Electronic Guards were ineffective at reducing deer damage in cornfields. Wildlife Society Bulletin 32:524-531. Gotelli, N.J., and A.M. Ellison. 2004. A primer of ecological statistics. Sinauer Associates, Sunderland, Massachusetts. Grover, K.E., and M.J. Thompson. 1986. Factors influencing spring feeding site selection by elk in the Elkhorn Mountains, Montana. Journal of Wildlife Management 50:466-470. Hegel, T.M., C.C. Gates, and D. Eslinger. 2009. The geography of conflict between elk and agricultural values in the Cypress Hills, Canada. Journal of Environmental Management 90:222-235. Hildreth, A.M., S.E. Hygnstrom, E.E. Blankenship, and K.C. VerCauteren. 2012. Efficacy of a partial poly-mesh fence with wings to reduce deer damage to corn. Wildlife Society Bulletin 36:199203. Hygnstrom, S.E. and S.R. Craven. 1988. Electric fences and commercial repellents for reducing deer damage in cornfields. Wildlife Society Bulletin 16:291-296. 73 �Johnson, H.E., P.Dorsey, M. Hammond, C. Bishop, K. VerCauteren, D. Walter, and C. Anderson. 2011. Evaluating solutions to reduce elk and mule deer damage on agricultural resources. Study Proposal, Colorado Division of Parks and Wildlife, Fort Collins, USA. Krausman, P. R., J. J. Hervert, and L. L. Ordway. 1985. Capturing deer and mountain sheep with a netgun. Wildlife Society Bulletin 13:71–73. McKillop, I.G., and R.M. Sibly. 1988. Animal behaviour at electric fences and implications for management. Mammal Review 18:91-103. Nixon, C.M., L.P. Hansen, P.A. Brewer, J.E. Chelsvig. 1989. Ecology of white-tailed deer in an intensively farmed region of Illinois. Wildlife Monographs 118:1-77. Pinheiro, J., and D. M. Bates. 2000. Mixed effects models in S and S-Plus. Springer-Verlag, New York, New York, USA. Rosenberry, C.S., L.I. Muller, and M.C. Conner. 2001. Movable, deer-proof fencing. Wildlife Society Bulletin 29:754-757. Seamans, T.W., and K.C. VerCauteren. 2006. Evaluation of ElectroBraid™ fencing as a white-tailed deer barrier. Wildlife Society Bulletin 34:8-15. Van Tassell, L.W., C. Phillips, B. Yang. 1999. Depredation claim settlements in Wyoming. Wildlife Society Bulletin 25:886-894. Vecellio, G.M., R.H. Yahner, and G.L. Storm. 1994. Crop damage by deer at Gettysburg Park. Wildlife Society Bulletin 22:89-93. VerCauteren, K.C., M.J. Lavelle, and S. Hygnstrom. 2006. Fences and deer-damage management: a review of designs and efficacy. Wildlife Society Bulletin 34:191-200. Wagner, K.K., R.H. Schmidt, and M.R. Conover. 1997. Compensation programs for wildlife damage in North America. Wildlife Society Bulletin 25:312-319. Walter, W.D., M.J. Lavelle, J.W. Fischer, T.L. Johnson, S.E. Hygnstrom, and K.C. VerCauteren. 2010. Management of damage by elk (Cervus elaphus) in North America: a review. Wildlife Research 37:630-646. Walter, W.D., K.C. VerCauteren, J.M. Gilsdorf, and S.E. Hygnstrom. 2009. Crop, native vegetation, and biofuels: response of white-tailed deer to changing management priorities. Journal of Wildlife Management 73:339-344. World Resources Institute. 2008. WRI EarthTrends monthly update page. http://earthtrends.org/updates/node/180. Wisdom, M.J., and J.G. Cook. 2000. North American Elk. Pages 694-735 in S. Demarais and P.R. Krausman, editors. Ecology and management of large mammals in North America. Prentice Hall, Upper Saddle River, New Jersey, USA. Yoder, J. 2002. Deer-inflicted crop damage and crop choice in Wisconsin. Human Dimensions of Wildlife 7:179-196. Prepared by _______________________________________ Heather E. Johnson, Wildlife Researcher 74 �Appendix 1 Evaluating techniques to reduce cervid damage Evaluation of techniques to reduce deer and elk damage to agricultural crops Colorado Parks and Wildlife and US Dept. of Agriculture, Heather Johnson Justin Fischer Mathew Hammond Patricia Dorsey W. David Walter Charles Anderson Kurt VerCauteren Research Report 75 �8/22/2013 Corresponding Author: Heather Johnson Colorado Parks and Wildlife 415 Turner Drive Durango, CO 81303 Phone: (970) 375-6715 FAX: (970) 375-6705 Email: Heather.Johnson@state.co.us Evaluation of techniques to reduce deer and elk damage to agricultural crops RH: Johnson et al. • Evaluating techniques to reduce cervid damage HEATHER E. JOHNSON,1 Colorado Parks and Wildlife, 415 Turner Drive, Durango CO 81303, USA JUSTIN W. FISCHER, United States Department of Agriculture, Animal and Plant Health Inspection Service, Wildlife Services, National Wildlife Research Center, 4101 LaPorte Avenue, Fort Collins, CO 80521, USA MATTHEW HAMMOND, Colorado Parks and Wildlife, 151 East 16th Street, Durango CO 81301, USA PATRICIA D. DORSEY, Colorado Parks and Wildlife, 415 Turner Drive, Durango CO 81303, USA W. DAVID WALTER2, United States Department of Agriculture, Animal and Plant Health Inspection Service, Wildlife Services, National Wildlife Research Center, 4101 LaPorte Avenue, Fort Collins, CO 80521, USA CHARLES ANDERSON3, United States Department of Agriculture, Animal and Plant Health Inspection Service, Wildlife Services, National Wildlife Research Center, 4101 LaPorte Avenue, Fort Collins, CO 80521, USA KURT C. VerCAUTEREN, United States Department of Agriculture, Animal and Plant Health Inspection Service, Wildlife Services, National Wildlife Research Center, 4101 LaPorte Avenue, Fort Collins, CO 80521, USA 1 E-mail: Heather.Johnson@state.co.us 2 Present addresses:United States Geological Survey, Pennsylvania Cooperative Fish and Wildlife Research Unit, 403 Forest Resources Building, University Park, PA 16802, USA (WDW), 3Missouri Department of Conservation, 2901 W. Truman Blvd., Jefferson City, MO 65109, USA (CA) KEY WORDS Cervus elaphus nelsoni, crop damage, electric fence, elk, mule deer, Odocoileus hemionus, repellent, sunflowers, wildlife damage management, winged fence. 76 �ABSTRACT Mule deer (Odocoileus hemionus) and Rocky Mountain elk (Cervus elaphus nelsoni) provide important recreational, ecological, and economic benefits, but can also cause substantial damage to agricultural crops. Cervid damage to agriculture creates challenges for wildlife agencies responsible for minimizing crop depredation while maintaining healthy deer and elk populations. Sunflower producers in southwestern Colorado have experienced high deer and elk damage and were interested in temporary methods to reduce damage that were cost-effective for rotational crops. To address this challenge we investigated three temporary, non-lethal exclusion and repellent techniques for reducing deer and elk damage to sunflowers: 1) a polyrope electric fence, 2) the chemical repellent Plantskydd™, and 3) a winged fence. During July through October 2011 and 2012, we used a randomized block design to test the efficacy of these techniques by quantifying cervid damage to sunflowers and the number of deer and elk tracks traversing treatment and control plot boundaries. Using generalized linear mixed models we found that polyrope electric fences reduced deer and elk damage and presence within plots, while the repellent and winged fences did not reduce ungulate activity. Polyrope electric fences may be a suitable tool in areas where wildlife management agencies want to maintain deer and elk populations but reduce seasonal damage by cervids to high value crops. In Colorado, use of an effective exclusion technique like polyrope electric fence could also decrease the need for lethal depredation permits and damage compensation payments, and increase satisfaction among producers and the public. Wildlife Society Bulletin: 00(0): 000-000, 201X Mule deer (Odocoileus hemionus) and Rocky Mountain elk (Cervus elaphus nelsoni) provide important recreational, ecological, and economic benefits, but also can cause substantial damage to agricultural crops (Austin et al. 1998, Wisdom and Cook 2000). Because crops are typically more digestible and contain higher levels of crude protein than native grasses and browse species, they are often selected and consumed by wild cervids (Mould and Robbins 1982). Agricultural producers have reported more damage by elk and deer (Odocoileus sp.) than any other wildlife species, and damage by deer alone has been projected to exceed 100 million dollars annually in the U.S. (Conover 2002). Cervid damage to crops has created significant challenges for wildlife management agencies, as agencies are often responsible for both maintaining cervid population sizes for recreation while minimizing damage to agriculture (Wagner et al. 1997, Hegel et al. 2009, Van Tassell et al. 1999, Walter et al. 2010). Agricultural producers often experience varying amounts of crop depredation caused by cervids depending on the seasonal distribution, abundance and landscape configuration of local food resources (Vecellio et al. 1994, Yoder 2002, Hegel et al. 2009). Damage also can be variable both within and among growing seasons, as local precipitation and temperatures will alter the availability of native forage and the motivation of deer and elk to feed on agricultural products (Walter et al. 2010). The proximity of cropland and wildland is also important in predicting patterns of damage, as cultivated fields closer to wildlife cover experience greater depredation (Nixon et al. 1989, Hegel et al. 2009). As a result, the effectiveness of management practices to reduce cervid damage may vary based on native forage availability, proximity of cover, and other habitat features (Hegel et al. 2009). Common management tools used to reduce cervid damage to crops include permanent fencing and lethal removal of animals through depredation permits (Walter et al. 2010); however, there are drawbacks to each approach. Permanent cervid-proof fencing is effective but often cost-prohibitive for producers that have large tracts of land (VerCauteren et al. 2006) or grow crops on a rotational basis where only one crop type experiences high rates of damage. Permanent fencing is also a concern as it can interfere with wildlife movements and reduce access to nearby habitat. Wildlife agencies use depredation permits to lethally remove animals causing damage, but tolerance for these permits is often low among hunters, some producers, and the general public (Colorado Parks and Wildlife [CPW], unpublished data). Hunters often perceive depredation permits as reducing hunting opportunity (Fritzell et al. 1995, Horton and Craven 1997), particularly when local deer and elk population sizes are below agency management 77 �objectives. Depredation permits are also often unpopular with the public, particularly when lethal removal includes female cervids with dependent young. Identifying cost-effective, non-lethal methods that reduce cervid damage to agricultural crops is of particular interest in Colorado. Deer and elk account for about 50% of wildlife damage claims on agriculture, and CPW is mandated to pay all eligible claims. These compensation payments are costly (i.e., $458,760 was paid in compensation for deer and elk damage in 2012; CPW 2012), thus, CPW is interested in methods to reduce cervid depredation and associated payments. While damage to agriculture is a management concern, many of Colorado’s deer and elk populations are at or below their management objectives, making depredation permits highly unpalatable to local hunters and the general public. Because deer and elk often depend upon private lands for habitat, finding cost-effective, non-lethal solutions to prevent cervid depredation is also essential to encourage private landowner tolerance of wildlife and to build effective agency-landowner partnerships. To identify cost-effective, non-lethal strategies for reducing deer and elk damage to crops, our objective was to experimentally test three temporary techniques: 1) a 5-strand polyrope electric fence (hereafter electric), 2) an organic chemical repellent (Plantskydd™; hereafter repellent), and 3) a winged or partial fence (hereafter winged). These methods are less expensive than permanent fencing and can be implemented on a temporary basis to account for crop rotation (VerCauteren et al. 2006, Walter et al. 2010). While these methods have received some testing on white- and black-tailed deer (Odocoileus virginianus, O. h. columbianus; Nolte 1998, Seamans and VerCauteren 2006, Hildreth et al. 2012), little is known about their effectiveness in reducing mule deer or elk damage to agriculture. STUDY AREA We tested temporary exclusion and repellent techniques for deer and elk near Dove Creek, Colorado, USA (Dolores County; 37⁰45’58.05” N, 108⁰54’21.10”W; Fig. 1). Experimental plots were placed in agricultural fields growing sunflowers that were spatially juxtaposed to native vegetation and wildland canyons, and which had previously experienced cervid damage (CPW, unpublished data). All sunflower fields were located on private property, but the region is generally comprised of a mix of private and public lands. Elevation in the study area ranges from 1,981 to 2,590 m, and vegetation is characterized as mountain shrub and pinyon-juniper woodlands, interspersed with irrigated and dryland agriculture. The native vegetation is primarily composed of serviceberry (Amelanchier alnifolia), bitterbrush (Purshia tridentata), mountain mahogany (Cercocarpus montanus), squaw apple (Peraphyllum ramosissimum), black sagebrush (Seriphidium novum), pinyon pine (Pinus edulis) and juniper (Juniperus osteoperma). Between 1996 and 2012 mean annual precipitation was 26.7 cm, which is typically received during late summer rains and as snow during winter (Weather Station DVCO1, Colorado Agricultural Meteorological Network 2012). Mean annual minimum and maximum temperatures were -0.4⁰C and 16.6⁰C, respectively (Colorado Agricultural Meteorological Network 2012). Since 1998 estimated deer population sizes have been consistently below CPW’s management objectives, while estimated elk population size has been above or within management objectives (CPW, unpublished data). The study area has experienced high rates of mule deer and elk agricultural damage in association with a recent switch in the types of crops that are grown. Farmers traditionally grew dry beans, spring and winter wheat, and grass hay which experienced minimal damage by cervids. Since 2007, however, many farmers started growing sunflowers on a rotational basis, a high-value seed oil crop used for biofuel, and have experienced up to 100% depredation on fields in some years. Sunflowers in the region are generally grown on a 3 to 4-year rotation with other crops (e.g., winter wheat, pinto beans) that experience minimal damage and thus, producers were interested in exclusion or repellent techniques that could be moved between fields in different years. Cervid damage in this area was also exacerbated by the spatial 78 �juxtaposition of agricultural fields alongside wildland canyons that provided refugia for deer and elk (Fig. 1). METHODS Exclusion and Repellent Methods Evaluated Electric – We tested a polyrope electric fence (ElectroBraid™ Fence Limited, Yarmouth, Nova Scotia, Canada; approximately $5-10/m for materials), which acts primarily as a psychological barrier based on learned behavioral and avoidance conditioning (Fig. 2a; McKillop and Sibly 1988, VerCauteren et al. 2012). The fence consisted of conductive copper wires woven into synthetic “ropes” that are more durable, visible and easier to install than traditional electric fence designs (Hygnstrom and Craven 1988, Seamans and VerCauteren 2006, VerCauteren et al. 2006, Fischer et al. 2011). We constructed fences 1.8 m high, with wooden h-brace assemblies placed approximately every 100 m and metal t-posts spaced every 15 m. Five polyrope lines were attached to the fence posts at 20, 56, 89, 135, and 183 cm above ground to discourage deer and elk incursions. Avoidance conditioning occurs when an animal contacts the fence, often with the nose or tongue, and receives an electric shock. Polyrope fences have reduced whitetailed deer damage to crops (Hygnstrom and Craven 1988, Seamans and VerCauteren 2006), but have not been experimentally tested for reducing mule deer or elk damage. The polyrope fence used a Speedrite™ 3000 energizer (Tru-Test Incorporated, San Antonio, Texas) which had a maximum pulse output of 3.0 joules and was operated from a 12-volt deep-cycle battery with a solar-panel recharger. Repellent - We tested the effectiveness of Plantskydd™ (Tree World Plant Care Products Inc, St. Joseph, Missouri, USA) for reducing deer and elk damage. This repellent can be used on conventional and organic crops and can be applied by ground or aerial spraying. Plantskydd™ was developed in Sweden for reducing mammalian wildlife damage on commercial forests. The active ingredient is dried bloodmeal, which the manufacturer asserts works by emitting an odor that wildlife associate with predator presence. We mixed Plantskydd™ powder with water following the manufacturer’s directions for severe damage (14.8 kg of Plantskydd™/plot perimeter). The manufacturer recommends spraying a swath ≥10 m around plot perimeters, and we sprayed an 18 m swath around treatment plot perimeters, the maximum distance that could be covered with an industrial ground sprayer (Model 4720, John Deere, Deere and Company, Moline, Illinois, USA). Given materials and application, this treatment cost ≤ $1/m of field perimeter spraying. Plantskydd™ was applied monthly throughout the growing season (Jul – Sept) to account for the repellent washing off or degrading, and to spray new plant growth. Plantskydd™ has reduced damage to tree seedlings caused by black-tailed deer (Nolte 1998, Wagner and Nolte 2001), but has been not been tested on mule deer or elk. Winged fence –Hildreth et al. (2012) recently experimented with “winged” or “partial” fences designed to reduce white-tailed deer access along field edges adjacent to cover. The fence is completely installed on the field side that borders native vegetation, and partially installed on the perpendicular sides, creating “wings” that extend around a portion of the field (Fig. 2b; approximately $6/m for materials). This fence is highly economical as only a portion of the field needs to be enclosed and materials can be easily erected and removed depending on crop rotation. We installed winged fences following Hildreth et al. (2012), where the side of the treatment plot closest to the crop/wildland interface received complete protection. We erected fences 2.1 m in height which consisted of UV-stable polypropylene high-strength mesh (Benner’s Gardens, Phoenixville, PA) secured to 3 m metal t-posts spaced every 7 m using cable ties. Two strands of 12.5 gauge high-tensile wire were placed 0.8 m and 2.1 m above ground, so the mesh could be suspended and anchored to the wire with circular staples along the length of the fence for support. The fence also had a 0.2 m apron extending outward from the field, secured with 0.3 m steel stakes, to further reduce elk and deer access. Corners and ends of the winged fence were supported with metal t-post angled h-brace assemblies. The fence wings extended 50 m along the two sides of the treatment plots that were adjacent to the fully installed side of the fence. 79 �Experimental Design We used a randomized block design (Gotelli and Ellison 2004) where each “block” was a sunflower field (~65-80 ha in size) that had previously experienced cervid crop damage, and which was directly adjacent to the wildland boundary where damage was expected to be greatest (Fig. 1). Within each field we delineated 4 4-ha treatment plots. Treatment plots were randomly assigned to receive one of the following treatments: no exclusion or repellent method (control), electric fence, repellent, or winged fence. We used this design to account for environmental heterogeneity, as we expected damage to vary among fields. We monitored 5 replicate fields during 2011 (Fields A-E) and 4 replicate fields in 2012 (Fields F-I); because sunflowers were grown on rotation the same fields were not tested in both years. Fences were constructed in late June and early July after sunflowers had germinated to ensure planting was successful, as pests or low soil moisture can cause failure in germination. The corners of all plots were marked with easily visible metal stakes to facilitate data collection. Monitoring Fence Effectiveness Plots in each field were monitored for two response variables: damage to sunflower plants and number of deer and elk tracks traversing plot boundaries (entry/exit into plots). We used the variablearea-transect method for estimation of crop damage (Engeman and Sugihara 1998; Engeman and Sterner 2002; Gilsdorf et al. 2004a, b), conducting final damage assessments immediately before harvest (midOct). In 2011 we assessed damage on 15 transects/plot, and in 2012 we increased the number to 30 transects/plot. For each transect, we randomly (and with replacement) identified a starting location within the plot and inspected a row of sunflowers, counting the total number of sunflower plants, and the number of plants that were damaged by deer or elk. Typical damage was characterized by the removal of the terminal bud, consumption of the seed head and trampling of the plants, verified by accompanying cervid tracks. If 5 cervid-damaged sunflowers were tallied within 100 m, we recorded the distance traveled to the fifth damaged plant (<100 m) and the total number of sunflower plants observed within that distance. If 5 cervid-damaged sunflowers were not tallied within 100 m, the observer recorded the total number of sunflowers and the number of cervid-damaged plants counted within that distance. If the end of the sunflower row was reached before completing a transect, the observer would randomly select an adjacent row (i.e., right or left row) for completing the transect. Each treatment and control plot was also monitored for deer and elk tracks that traversed plot boundaries on a bimonthly basis throughout the growing season (mid-Jul through mid-Oct). An observer would walk the perimeter of each plot, counting the total number of deer and elk tracks that crossed the plot perimeter. Cervid tracks were raked or stamped out after each observation to avoid double-counting in subsequent sampling periods. Statistical Approach We calculated mean proportion of end-of-season damage for each treatment and control plot, and mean number of elk and deer tracks traversing plot perimeters for each plot across the growing season. We also calculated mean values separately for fields monitored in 2011 and 2012, as cervid damage was uncharacteristically low in 2011. We did not include end-of-season damage values from the repellent plot of one field (Field F in 2012) because cervid damage occurred in that plot before the first application of the repellent. Similarly, end-of-season damage information from all treatment plots of a field in 2012 (Field I) were removed from data summaries and analyses because substantial depredation occurred after germination but before fence construction. We used a generalized linear mixed model to identify whether exclusion or repellent treatment types were effective in reducing cervid damage to sunflower plots (Pinheiro and Bates 2000). Because damage data were recorded for each transect as the number of damaged plants/total plants, we used a binomial distribution with a logit link function (Bolker et al. 2009). Treatment was included in the model as a categorical fixed effect (control plots were considered the reference class) and we nested plot within field within year for the random effects model structure. We used model coefficients to assess the 80 �direction and magnitude of different treatment types on cervid damage (95% confidence intervals nonoverlapping zero). To evaluate the influence of exclusion or repellent types on deer and elk tracks traversing plot perimeters, we used generalized linear mixed models with Poisson distributions and log link functions. As with the damage models, we included treatment type as a categorical fixed effect and nested plot within field within year for the random effects portion of the model. We generated separate models for predicting the number of tracks by deer and elk, as we hypothesized that treatments may vary in their effectiveness among cervid species (e.g., VerCauteren et al. 2006, Walter et al. 2010). As with the damage model, we used model coefficients, and their 95% confidence intervals, to assess the direction and magnitude of treatment effects on the number of tracks traversing plot boundaries. We used the package “lme4” in program R for all statistical modeling (R Core Team 2012). RESULTS Cervid damage and tracks varied across treatment and control plots. Just prior to harvest, the percentage of sunflowers damaged by cervids across plots and years ranged from 0.0% to 72.6% ( x = 8.3%, SE = 0.8). The mean bimonthly number of deer tracks crossing plot perimeters ranged from 0 to 149.8 ( x = 23.0, SE = 5.3) and the mean number of elk tracks ranged from 0 to 21.6 ( x = 5.3, SE = 1.1). Mean percentage sunflower damage and number of deer tracks were greater in 2012 than in 2011 (damage: t = -3.300, df = 29, P = 0.003 [Fig. 3a]; deer tracks: t = -4.512, df = 34, P < 0.001; Fig. 3b), but mean values for elk tracks were similar between years (t = 0.371, df = 34, P = 0.713). In 2011, treatment and control plots averaged 0.9% sunflower plant damage at the end of the growing season, and a bimonthly average of 6.0 deer and 5.7 elk tracks crossed plot boundaries. Conversely, 2012 plots had an average of 17.1% of plants damaged at harvest and an average of 44.4 deer tracks and 4.9 elk tracks crossed plot boundaries on a bimonthly basis. Despite differences in damage between years, plots protected with electric fencing consistently received the least amount of cervid damage and tracks (Fig. 3). The only treatment type that reduced damage to sunflowers was the electric fence (Table 1). Treatment effects on damage and plot use across both years, however, showed limited biological effect given that more data were collected in 2011 when minimal damage occurred. Across years, the mean proportion of damaged plants on electric fence plots was 0.01 (95% CI: 0.00 – 0.03), on control plots was 0.05 (95% CI: 0.00 – 0.33), on repellent fences was 0.04 (95% CI: 0.01 – 0.15) and on winged fences was 0.04 (95% CI: 0.01 – 0.15). Electric fencing was also the only treatment type that reduced cervid activity within sunflower plots (Table 1). The average bimonthly number of deer tracks that crossed plot perimeters on plots with electric fencing was 0.6 (95% CI: 0.3 – 1.1), on control plots was 18.5 (95% CI: 3.8 – 91.5), on repellent fence plots was 18.4 (95% CI: 11.4 – 29.7), and on winged plots was 16.8 (95% CI: 10.4 – 27.0). Electric fences also reduced the number of elk that crossed plot perimeters on a bimonthly basis, but the effect was lesser than for deer. An average of only 0.1 elk tracks crossed electric fence plot boundaries (95% CI: 0.0 – 0.2), while 4.3 crossed control plots (95% CI: 1.8 – 10.3), 3.4 crossed repellent plots (95% CI: 2.25.2) and 3.7 crossed winged plots (95% CI: 2.4 – 5.7). DISCUSSION As wildlife management agencies look for methods to reduce cervid damage to agricultural crops while maintaining deer and elk population sizes, non-lethal methods of crop protection will become increasingly important. We tested three methods for reducing deer and elk damage to sunflowers, a highvalue crop, but found that only polyrope electric fencing significantly reduced damage and use by deer and elk. Investigators have found different polyrope electric fence designs to be successful at reducing white-tailed deer damage to crops (Hygnstrom and Craven 1998, Seamans and VerCauteren 2006), but to 81 �our knowledge, this is the first study to test the 5-strand polyrope fence design on mule deer or elk. Polyrope appears to be effective at reducing deer and elk damage to sunflowers, providing a temporary and cost-effective option for producers to reduce depredation through non-lethal means. While the chemical repellent Plantskydd™ is advertised to imitate predator presence and induce fear in cervids, it was not consistently effective in our evaluation. Fear inducing repellents are generally more successful than repellents with other strategies (i.e., aversive taste or pain inducing; Wagner and Nolte 2001), and studies have found this repellent to reduce black-tailed deer damage to tree seedlings (Nolte 1998, Wagner and Nolte 2001). In our sunflower plots, however, the repellent did not reduce mule deer or elk damage or tracks, a result which may be influenced by numerous factors including: animal habituation, availability of native forage, local weather conditions, animal nutritional state, repellent concentration, or the frequency of repellant application (Kimball et al. 2009, Walter et al. 2010, Elmeros et al. 2011). Indeed, drought conditions in 2012 may have increased motivation by deer and elk to forage on sunflowers, despite the repellent odor. We applied repellent once/month to treatment plots. While >1 application/month may have increased effectiveness of the treatment, such a high frequency of applications would not be feasible for most sunflower producers, and therefore, not particularly useful as a routine damage management tool. The winged fence we used also did not decrease deer and elk damage and use of the plots. In contrast, Hildreth et al. (2012) found winged fencing reduced white-tailed deer depredation to corn by 13.5%. Based on profits from the yield of corn and the cost of fence construction, Hildreth et al. (2012) concluded that corn producers could save approximately $205/ha/annually by using a winged fence along the agriculture-wildland interface. In our experiment, damage in winged plots was lesser than control plots in 7 of 8 fields, but did not have a strong treatment effect. We often observed elk and deer tracks along the partial portion of the fence to cross into the plot at the termination of the wing. DeVault et al. (2008) reported similar results in which white-tailed deer (Odocoileus virginianus) traveled around partial fences at an airport runway to gain access to crop fields. Animal habituation and motivation, crop palatability, and wing length may all influence the success of this approach. We placed the fully fenced treatment side against the dominant wildland boundary, but the complex juxtaposition of agricultural fields and canyons in southwestern Colorado may reduce the utility of this approach in this region. This exclusionary method may perform better in a more homogenous landscape. Given that the number of elk tracks remained fairly consistent between years, while the number of deer tracks was greater in 2012, it appears that the greater damage rates in 2012 were primarily attributable to deer crop depredation. Elk in the vicinity of Dove Creek migrate seasonally, often arriving at agricultural areas during summer, and spending the remainder of the year in secluded, wildland canyons (CPW, unpublished data). In contrast, mule deer often inhabit agricultural areas year-round (CPW, unpublished data), potentially increasing their habituation to novel structures and odors. In the case of electric fencing, smaller bodied deer are more likely able to breach the strands of polyrope, an obstacle which may be more effective at inhibiting larger-bodied elk. Despite differences in habitat-use patterns, behavior and morphology of deer and elk, polyrope electric fences were effective at reducing crop damage for both species. We tested three techniques for reducing damage to sunflowers during 2011 and 2012, years when crop depredation was dramatically variable. In 2011, deer and elk damage to sunflowers averaged 1%, well within tolerance levels for farmers as evidenced by no damage claims filed by farmers that year (CPW, unpublished data). Spring and summer (Mar-Aug) precipitation was exceedingly high during 2011 (Weather Station DVCO1, Colorado Agricultural Meteorological Network 2012), ~153% of normal, and it appears that the availability of abundant natural forage likely reduced damage by deer and elk. In 2012, however, the Dove Creek region experienced a drought, receiving about 60% of spring and summer precipitation, and only 30% of average spring (Mar-Jun) rainfall, a critical time for dryland farming in 82 �southwest Colorado. Soil moisture was so low in 2012 that few producers planted sunflowers, and the majority of seeds planted in some fields never germinated. We suspect that observed differences in plot damage and use between 2011 and 2012 were largely driven by differences in weather, and the resulting effects on the native vegetation for deer and elk. High temporal and spatial variability in cervid damage, as observed in this study, is particularly challenging for producers and wildlife management agencies seeking solutions to reduce depredation. Such variability may reduce the motivation of producers to protect crops and alter priorities of wildlife managers, depending on whether cervid damage is severe or minimal in a particular year or area. This variability in damage also highlights the utility of a temporary method, like polyrope electric fence, for protecting crops when damage is expected to be high (e.g., in drought years). Ultimately, however, the decision to invest in a tool like polyrope electric fencing will depend on field size, expected amount of damage, crop prices, and the frequency and duration a producer will need to use the fencing, particularly for rotational crops. MANAGEMENT IMPLICATIONS For wildlife agencies seeking non-lethal management options for reducing deer and elk damage to high-value agricultural crops, we found that 5-strand polyrope electric fencing was effective. Polyrope is easy to assemble/disassemble, cost-effective relative to permanent fencing, and can be used on a temporary basis to minimize damage for certain crops grown on rotation or during years when natural forage for cervids is scarce. In areas where management agencies are working to maintain or increase deer and elk populations, but reduce cervid damage, the application of an effective exclusion technique like polyrope electric fencing could protect high-value crops, decrease the need for compensation payments and lethal cervid depredation permits, and increase satisfaction of producers and the public. Wildlife agencies will need to continue to work with producers to test and apply management techniques for crop protection based on the wildlife species present, population densities, crop types, landscape configuration, and abundance of local forage. ACKNOWLEDGEMENTS We thank M. Glow, C. Priest, M. Preisler, A. Brown, A. Hildreth, M. Lavelle, G. Martin, D. Sanders, and B. Beltran-Beck for collecting field data and helping with fence construction. We also thank T. Brown and G. Phillips for help in purchasing fencing supplies, and A. Berrada, P. White and D. Fernandez for project support. This project would not have been possible without the cooperation of landowners around Dove Creek. Funding was provided by the USDA National Wildlife Research Center, Colorado Habitat Partnership Program, Montelores Habitat Partnership Program, Rocky Mountain Elk Foundation, and Colorado Auction/Raffle Grant program. 83 �LITERATURE CITED Austin, D. D., P. J. Urness, and D. Duersch. 1998. Alfalfa hay crop loss due to mule deer depredation. Journal of Range Management 51:29-31. Bolker, B. M., M. E. Brooks, C. J. Clark, S. W. Geange, J. R. Poulsen, M. H. H. Stevens, and J.S. S. White. 2009. Generalized linear mixed models: a practical guide for ecology and evolution. Trends in Ecology and Evolution 24:127-135. Burnham, K. P., and D. R. Anderson. Model selection and multimodel inference: a practical informationtheoretic approach. Second Edition. Springer-Verlag, New York, New York USA. Colorado Agricultural Meteorological Network. 2012. COAgMet Homepage. <http://ccc.atmos.colostate.edu/~coagmet/index.php>. Accessed 10 Jan 2013. Colorado Parks and Wildlife. 2012. FY12 game damage claims annual report. Colorado Parks and Wildlife, Denver, Colorado, USA. Conover, M. R. 2002. Resolving human-wildlife conflicts: the science of wildlife damage management. CRC Press LLC, Boca Raton, Florida, USA. DeVault, T. L., J. E. Kubel, D. J. Glista, and O. E. Rhodes. 2008. Mammalian hazards at small airports in Indiana: impact of perimeter fencing. Human-Wildlife Conflicts 2:240-247. Elmeros, M., J. K. Winbladh, P. N. Andersen, A. B. Madsen, and J. T. Christensen. 2011. Effectiveness of odour repellents on red deer (Cervus elaphus) and roe deer (Capreolus capreolus): a field test. European Journal of Wildlife Research 57:1223-1226. Engeman, R. M., and R. T. Sterner. 2002. A comparison of potential labor-saving sampling methods for assessing large mammal damage in corn. Crop Protection 21:101-105. Engeman, R. M., and R. T. Sugihara. 1998. Optimization of variable area transect sampling using Monte Carlo simulation. Ecology 79:1425-1434. Fischer, J. W., G. E. Philips, D. M. Baasch, M. J. Lavelle, and K. C. VerCauteren. 2011. Modifying elk (Cervus elaphus) behavior with electric fencing at established fence-lines to reduce disease transmission potential. Wildlife Society Bulletin 35:9-14. Fritzell, P. A. Jr., D. L. Minnus, and R. B. Peyton. 1995. A comparison of deer hunter and farmer attitudes about crop damage abatement in Michigan: messages for hunters, farmers and managers. Seventh Eastern Wildlife Damage Management Conference, Paper 12. Gilsdorf, J. M., S. E. Hygnstrom, K. C. VerCauteren, G. M. Clements, E. E. Blankenship, and R. M. Engeman. 2004a. Evaluation of a deer-activated bio-acoustic frightening device for reducing deer damage in cornfields. Wildlife Society Bulletin 32:515-523. Gilsdorf, J. M., S. E. Hygnstrom, K. C. VerCauteren, G. M. Clements, E. E. Blankenship, and R. M. Engeman. 2004b. Propane exploders and Electronic Guards were ineffective at reducing deer damage in cornfields. Wildlife Society Bulletin 32:524-531. Gotelli, N.J., and A.M. Ellison. 2004. A primer of ecological statistics. Sinauer Associates, Sunderland, Massachusetts, USA. Hegel, T. M., C. C. Gates, and D. Eslinger. 2009. The geography of conflict between elk and agricultural values in the Cypress Hills, Canada. Journal of Environmental Management 90:222-235. Hildreth, A. M., S. E. Hygnstrom, E. E. Blankenship, and K. C. VerCauteren. 2012. Use of partially fenced fields to reduce deer damage to corn. Wildlife Society Bulletin 36:199-203. Horton, R. R., and S. R. Craven. 1997. Perceptions of shooting-permit use for deer damage abatement in Wisconsin. Wildlife Society Bulletin 25:330-336. Hygnstrom, S. E. and S. R. Craven. 1988. Electric fences and commercial repellents for reducing deer damage in cornfields. Wildlife Society Bulletin 16:291-296. Kimball, B. A., J. Taylor, K. R. Perry, and C. Capelli. 2009. Deer responses to repellent stimuli. Journal of Chemical Ecology 35:1461-1470. Lavelle, M. J., J. W. Fischer, S. E. Hygnstrom, J. J. White, A. M. Hildreth, G. E. Phillips, and K. C. VerCauteren. 2010. Response of deer to containment by a poly-mesh fence for mitigating disease outbreaks. Journal of Wildlife Management 74:1620-1625. McKillop, I.G., and R.M. Sibly. 1988. Animal behaviour at electric fences and implications for 84 �management. Mammal Review 18:91-103. Mould, E. D., and C. T. Robbins. 1982. Digestive capabilities in elk compared to white-tailed deer. Journal of Wildlife Management 46:22-29. Nixon, C. M., L. P. Hansen, P. A. Brewer, J. E. Chelsvig. 1989. Ecology of white-tailed deer in an intensively farmed region of Illinois. Wildlife Monographs 118:1-77. Nolte, D. L. 1998. Efficacy of selected repellents to deter deer browsing on conifer seedlings. International Biodeterioration and Biodegradation 42:101-107. Pinheiro, J., and D. M. Bates. 2000. Mixed effects models in S and S-Plus. Springer-Verlag, New York, New York, USA. R Core Team. 2012. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org/. Retamosa, M. I., L. A. Humberg, J. C. Beasley, and O. E. Rhodes Jr. 2008. Modeling wildlife damage to crops in northern Indiana. Human-Wildlife Conflicts 2:225-239. Seamans, T. W., and K. C. VerCauteren. 2006. Evaluation of Electrobraid™ fencing as a white-tailed deer barrier. Wildlife Society Bulletin 34:8-15. Van Tassell, L. W., C. Phillips, B. Yang. 1999. Depredation claim settlements in Wyoming. Wildlife Society Bulletin 25:886-894. Vecellio, G. M., R. H. Yahner, and G. L. Storm. 1994. Crop damage by deer at Gettysburg Park. Wildlife Society Bulletin 22:89-93. VerCauteren, K. C., M. J. Lavelle, and S. E. Hygnstrom. 2006. Fences and deer-damage management: a review of designs and efficacy. Wildlife Society Bulletin 34:191-200. VerCauteren, K. C., R. Dolbeer, and E. Gese. 2012. Identification and management of wildlife damage. Pages 232-269 in N. Silvy, editor. The Wildlife Techniques Manual. Seventh edition. The John Hopkins University Press, Baltimore, Maryland, USA. Wagner, K. K., and D. L. Nolte. 2001. Comparison of active ingredients and delivery systems in deer repellents. Wildlife Society Bulletin 29:322-330. Wagner, K. K., R. H. Schmidt, and M. R. Conover. 1997. Compensation programs for wildlife damage in North America. Wildlife Society Bulletin 25:312-319. Walter, W. D., M. J. Lavelle, J. W. Fischer, T. L. Johnson, S. E. Hygnstrom, and K. C. VerCauteren. 2010. Management of damage by elk (Cervus elaphus) in North America: a review. Wildlife Research 37:630-646. Wisdom, M. J., and J. G. Cook. 2000. North American Elk. Pages 694-735 in S. Demarais and P. R. Krausman, editors. Ecology and management of large mammals in North America. Prentice Hall, Upper Saddle River, New Jersey, USA. Yoder, J. 2002. Deer-inflicted crop damage and crop choice in Wisconsin. Human Dimensions of Wildlife 7:179-196. 85 �Table 1. Coefficients for fixed effects from generalized linear mixed models evaluating the effectiveness of different treatment types for reducing cervid sunflower damage and the number of deer and elk tracks traversing experimental plot boundaries. Model Damage β SE P L 95% CI U 95% CI -3.020 1.169 <0.010 -5.311 -0.729 Electric* -2.227 0.943 0.018 -4.075 -0.379 Repellent -0.296 0.806 0.713 -1.876 1.284 Winged -0.108 0.709 0.879 -1.498 1.282 2.919 0.815 <0.001 1.322 4.516 Electric* -3.451 0.302 <0.001 -4.043 -2.859 Repellent -0.005 0.244 0.982 -0.483 0.473 Winged -0.100 0.244 0.684 -0.578 0.378 1.468 0.441 <0.001 0.604 2.332 Electric* -4.052 0.416 <0.001 -4.867 -3.237 Repellent -0.249 0.222 0.262 -0.684 0.186 Winged -0.163 0.221 0.460 -0.596 0.270 Variable Intercept* Treatment Deer Tracks Intercept* Treatment Elk Tracks Intercept* Treatment *Statistically significant at α = 0.05 level. 86 �Figure 1. Location of experimental treatment fields near Dove Creek, Colorado where exclusion and repellent methods for cervids were evaluated. 87 �Figure 2. A polyrope electric fence (A) and a partial winged fence (B) for excluding deer and elk from agricultural fields. A B 88 �Figure 3. Proportion of sunflower plants damaged at time of harvest (A) and number of deer and elk tracks that crossed plot boundaries (B), summarized across plots ( x and SE) for each treatment type, Dove Creek, Colorado, 2011 and 2012 A Proportion of Plants Damaged 50 2011 2012 40 30 20 10 0 Control No. of Tracks Crossing Plot Boundaries 100 90 80 70 Electric Repellent Winged Electric Repellent Winged B 2011 Deer 2012 Deer 2011 Elk 2012 Elk 60 50 40 30 20 10 0 Control Treatment Type 89 � https://cpw.cvlcollections.org/files/original/acfed9c505d6f6c4b56632119157f26f.pdf c927932ce857bdce9ae2eb5708c184c8 PDF Text Text Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Mammal and breeding bird response to bark beetle outbreaks in Colorado Period Covered: July 1, 2013 − June 30, 2014 Principal Investigators: Jacob S. Ivan, Jake.Ivan@state.co.us; Amy Seglund, Amy.Seglund@state.co.us ; All information in this project summary is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the principal investigator. Manipulation of these data beyond that contained in this summary is discouraged. Mountain pine beetle (Dendroctonus ponderosae) and spruce beetle (Dendroctonus rufipennis) infestations have reached epidemic levels in Colorado, impacting approximately 3.7 million acres since the initial outbreak in 1996 (Figure 1). Though bark beetles are native to Colorado and periodic infestations are considered a natural ecological process, the geographic scale of their impact and simultaneous infestation within multiple forest systems has never been observed. This historic outbreak is having significant impacts on composition and structure of forest stands that will propagate for decades into the future. The widespread mortality of forested systems in Colorado is likely to have a dramatic, but poorly understood effect on wildlife species that depend on these habitats. The project described here uses occupancy estimation to determine which wildlife species (both species of conservation concern and game species) decrease their use of an area as bark beetles pass through, which increase their use, and which exhibit use similar to levels prior to infestation. Statewide sampling was conducted during the summers of 2013 and 2014 (Figure 2). We sampled 150 Engelmann spruce (Picea engelmanni)/subalpine fir (Abies lasiocarpa) sites and 150 sites consisting mostly of lodgepole pine (Pinus contorta) or lodgepole pine mixed with other conifers. For both strata, sampling covered conditions ranging from sites that have yet to be impacted by bark beetles to those that were impacted by beetles more than a decade ago. At each 1-km2 site, we sampled the breeding bird community using the Rocky Mountain Bird Observatory’s protocol for “Integrated Monitoring in Bird Conservation Regions” (Hanni et al. 2014). We sampled the mammal community by deploying a remote camera near the center of each sample unit. Fieldwork for this phase of the project is now complete. However, data entry for 2014 is ongoing. For the purposes of this interim document, we report preliminary results for 3 mammalian species of conservation concern based on 2013 data only: snowshoe hares (Lepus americanus) and red squirrels (Tamiasciurus hudsonicus), which together comprise nearly 100% of the diet of the federally listed Canada lynx (Lynx canadensis), and American marten (Martes americana), which is a USFS Region 2 sensitive species. We collected 197,092 photos of 25 species during summer 2013. Occupancy analyses of these data indicate that snowshoe hares are more likely to use spruce/fir stands than lodgepole stands, but in both cases, use of these stands declines as bark beetle infestations pass by. We expected use to increase dramatically at some point as the understory responds to increased light, but that response will apparently take longer than the decade or so that has passed since the earliest infestations. Unlike hares, red squirrel use is similar for spruce/fir and lodgepole stands, but similar to hares, use of these stands declined after bark beetle infestations. This may be related to significant mortality of cone-bearing trees that occurs with beetle infestations. Use of the 2 stand types by marten was similar, but in contrast to the previous 2 species, use is expected to increase following bark beetle infestations. We expect to complete a full analysis and report for this project by Fall 2015. 1 �Figure 1. Current (2013) extent of mountain pine beetle (red) and spruce beetle (purple) infestations in spruce/fir (blue-green) and lodgepole pine (bright green) forests in Colorado. Bark beetle data were collected via USFS aerial surveys. Figure 2. Sites sampled via point counts and remote cameras to assess impacts of bark beetle infestations on breeding bird and mammal species in spruce/fir (blue-green, N = 150) and lodgepole pine (bright green, N = 150) stands in Colorado, 2013−2014. 2 �1.0 0.9 0.8 - Lodgepole - Spruce/Fir 10 12 0.7 [,' o,6 C C'O §- 0.5 (.) (.) O 0.4 0.3 0.2 0.1 0.0 2 0 4 6 8 Years Since Initial Infestation Figure 3. Snowshoe hare occupancy (i.e., use) of stands in relation to the number of years since initial infestation by bark beetles. Note that “0” years since infestation represents stands that have not yet been impacted. Use of spruce/fir stands is generally higher than use of lodgepole stands, but in both strata, use is expected to decline through time as bark beetles pass over an area. 1.0 0.9 0.8 >, 0.7 g 0.6 co g-o.5 (.) (.) 0.4 0 0.3 0.2 0.1 0.0 0 2 4 6 8 10 12 Years Since Initial Infestation Figure 4. Red squirrel occupancy (i.e., use) of stands in relation to the number of years since initial infestation by bark beetles. Note that “0” years since infestation represents stands that have not yet been impacted. Use of spruce/fir and lodgepole stands is generally similar (only a single line here compared to 2 lines for snowshoe hares above) and is predicted to decline through time as bark beetles pass over an area. 3 �1.0 0.9 0.8 0.7 i:,'o,6 C: Cll g-o.s (.) (.) 0 0.4 0.3 0.2 0.1 0.0 0 2 4 6 8 10 12 Years Since Initial Infestation Figure 3. American marten occupancy (i.e., use) of stands in relation to the number of years since initial infestation by bark beetles. Note that “0” years since infestation represents stands that have not yet been impacted. Use of spruce/fir and lodgepole stands is generally similar (only a single line here compared to 2 lines for snowshoe hares above) and is predicted to increase through time as bark beetles pass over an area. 4 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Mammal and breeding bird response to bark beetle outbreaks in Colorado Period Covered: July 1, 2014 − June 30, 2015 Principal Investigators: Jacob S. Ivan, Jake.Ivan@state.co.us; Amy Seglund, Amy.Seglund@state.co.us ; All information in this project summary is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the principal investigator. Manipulation of these data beyond that contained in this summary is discouraged. Mountain pine beetle (Dendroctonus ponderosae) and spruce beetle (Dendroctonus rufipennis) infestations have reached epidemic levels in Colorado, impacting approximately 4 million acres since the initial outbreak in 1996 (Figure 1). Though bark beetles are native to Colorado and periodic infestations are considered a natural ecological process, the geographic scale of their impact and simultaneous infestation within multiple forest systems has never been observed. This historic outbreak is having significant impacts on composition and structure of forest stands that will propagate for decades into the future. The widespread mortality of forested systems in Colorado may have a dramatic, but poorly understood effect on wildlife species that depend on these habitats. The project described here uses occupancy estimation to determine which wildlife species (both species of conservation concern and game species) decrease their use of an area as bark beetles pass through, which increase their use, and which exhibit use similar to levels prior to infestation. Statewide sampling was conducted during the summers of 2013 and 2014 (Figure 2). We sampled 150 Engelmann spruce (Picea engelmanni)-subalpine fir (Abies lasiocarpa) sites and 150 sites consisting mostly of lodgepole pine (Pinus contorta) or lodgepole pine mixed with other conifers. For both strata, sampling covered conditions ranging from sites that have yet to be impacted by bark beetles to those that were impacted by beetles more than a decade ago. At each 1-km2 site, we sampled the breeding bird community using the Rocky Mountain Bird Observatory’s protocol for “Integrated Monitoring in Bird Conservation Regions” (Hanni et al. 2014). We sampled the mammal community by deploying a remote camera near the center of each sample unit. Avian data have not yet been analyzed. We collected 388,951 photos of 56 species (26 mammalian species). For the purposes of this interim document, we report preliminary results for 3 mammalian species of conservation concern: snowshoe hares (Lepus americanus) and red squirrels (Tamiasciurus hudsonicus), which together comprise nearly 100% of the diet of the federally listed Canada lynx, and American marten (Martes americana), which is a USFS Region 2 sensitive species. Using Program MARK (White and Burnham 1999), we fit standard occupancy models (MacKenzie et al. 2006) to data for each species in the following manner. First, we fit a base model with parameters for the spruce-fir or lodgepole stratum, percentage of aspen present at the site, canopy cover, shrub cover, amount of down wood, amount of bare ground, and three physiographic variables that collectively account for elevation, moisture accumulation, and solar radiation at each site. The purpose of this model was to account for basic occupancy patterns of each species in the state irrespective of bark beetles. Next, we fit additional parameters to the base model which allowed occupancy to change in a variety of patterns (e.g., linearly, quadratic, spline, change-point, etc.) in relation to time elapsed since a stand was initially impacted by beetles. We also explored whether there was any interaction between response to beetles and stratum and/or response to beetles and the severity of the impact (percent of trees that were killed). We used Akaike’s Information Criterion (Burnham and Anderson 2002) to assess fit of these various beetle response models, and model-averaged occupancy across the model set to provide a best estimate of response of each species to beetles. 1 �Results indicate that snowshoe hares are more likely to use spruce-fir stands than lodgepole stands (Figure 3). There may be a slight increase in use around the time needles drop, followed by a steady fall back to ‘green’ forest levels, but in general hare use remained relatively constant for the initial decade after bark beetle infestation. Red squirrel use was similar between the two stand types (Figure 4). However, best fitting models included an interaction between severity of the beetle outbreak and response of red squirrels. In areas of low severity, response was minimal (Figure 4a). However, in areas of high severity, red squirrel use was 25−35% lower (Figure 4b). Use of the two stand types by marten was similar and relatively invariant to beetle impact (Figure 5). Literature Cited Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd edition. Springer, New York. MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. Academic Press, Oxford, United Kingdom. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 Supplement:120-138. Figure 1. Current (2014) extent of mountain pine beetle (red) and spruce beetle (purple) infestations in spruce/fir (blue-green) and lodgepole pine (bright green) forests in Colorado. Bark beetle data were collected via USFS aerial surveys. 2 �Figure 2. Sites sampled via point counts and remotes cameras to assess impacts of bark beetle infestations on breeding bird and mammal species in spruce/fir (blue-green, N = 150) and lodgepole pine (bright green, N = 150) stands in Colorado, 2013−2014. 1.0 0.9 - Lodgepole Pine 0.8 - Spruce-Fir 0.7 {$' c: 0.6 co §- 0.5 8 0.4 0 :: ..~ - - -- -::===· -- - - ---------- - ~- - - - - -- 0.3 0.2 ------ -- - -- - - - - - ---- - - - - -- - - - ---- -- -- - - - - 0.1 0.0 - + - - . . - - . - - - - . - - - - r - - - , - - - r - - - , - - , - - - , - - - , - - - . - - - , Figure 3. Snowshoe hare occupancy (i.e., use) of stands in relation to stage of infestation by bark beetles. ‘Green’ forests are those that have not yet been impacted. ‘Red’ forests are recently impacted; dead needles remain on trees. ‘Silver’ forests were impacted more distantly in the past and needles have fallen. Numbers in parentheses approximately correspond to the number of years that have passed since trees initially turned red. 3 �a) 1.0 0.9 0.8 0.7 ~ 0.6 --- - - - ---- - - - -- ------ -- -------- ---------- -- -- - --- ----- --------- ---- ---- - ---- ---- ------------------ ------ ---- ffi 0.5 a. i3 0.4 --------------------- -------- --------- ,... - ---- - - - (.) O 0.3 - Lodgepole Pine, 25% dead 0.2 - spur ce-fir, 25%dead 0.1 0.0 + - - ~ , - - - , - - , - , - - , - ~ , - - , - - , - ~ , - - - , -, -- , b) 1.0 Lodgepole Pine, 75% dead 0.9 Spruce-fi r, 75%dead 0.8 ----------- -------- --- ------ 0.7 ~ 0.6 ffi 0.5 a. i3 0.4 ----------------- - - - - ...... _______ _ (.) O 0.3 0.2 -- -- - -- ----------- 0.1 0.0 + - - ~ , - - - , - - , ~ , ~ ~ , - - , - - , - ~ , - - - , - - , - - , ~ , Figure 4. Red squirrel occupancy (i.e., use) of stands in relation to the number of years since initial infestation by bark beetles. Use of spruce-fir and lodgepole stands is generally similar and remains stable for stands that are lightly impacted by beetles (a; 25% dead). However, red squirrel occupancy is reduced by 25-35% in stands that are heavily impacted (b; 75% dead). 4 �1.0 0.9 - Lodgepole Pine - spruce-Fir 0.8 --_-_-..---..-_............ - -_-_-_- 0.7 ~ 0.6 ffi 0.5 B o.4 a. Cl O 0.3 0.2 -- --- - - - - - ------ --- - - --- - --- - - - --- - - -- -- -- ----- ---- --- --- 0.1 0.0 +---.------,-----r----r-----,-----.----r---r---..-----,r-----.----, Figure 5. American marten occupancy (i.e., use) of stands in relation to stage of infestation by bark beetles. Use does not vary appreciably by stand type, and remains stable through time as bark beetles pass over an area. 5 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Mammal and breeding bird response to bark beetle outbreaks in Colorado Period Covered: July 1, 2015  June 30, 2016 Principal Investigators: Jacob S. Ivan, Jake.Ivan@state.co.us; Amy Seglund, Amy.Seglund@state.co.us ; All information in this project summary is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the principal investigator. Manipulation of these data beyond that contained in this summary is discouraged. ABSTRACT Mountain pine beetle (Dendroctonus ponderosae) and spruce beetle (Dendroctonus rufipennis) infestations have reached epidemic levels in Colorado, impacting approximately 4 million acres since the initial outbreak in 1996 (Figure 1). Though bark beetles are native to Colorado and periodic infestations are considered a natural ecological process, the geographic scale of their impact and simultaneous infestation within multiple forest systems has never been observed. This historic outbreak is having significant impacts on composition and structure of forest stands that will propagate for decades into the future, which in turn may have dramatic, but poorly understood effects on wildlife species that depend on these habitats. This project used occupancy estimation to determine statewide wildlife response to bark beetle outbreaks, as mediated by changes in forest structure. Surveys were conducted during the summers of 2013 and 2014. We randomly sampled 150 Engelmann spruce (Picea engelmanni)-subalpine fir (Abies lasiocarpa) sites and 150 sites consisting mostly of lodgepole pine (Pinus contorta) or lodgepole pine mixed with other conifers. For both strata, sampling covered conditions ranging from sites that were not impacted by bark beetles to those that were impacted by beetles more than a decade ago. At each 1-km2 site, we sampled the breeding bird community using the Rocky Mountain Bird Observatory’s protocol for “Integrated Monitoring in Bird Conservation Regions” (Hanni et al. 2014). We sampled the mammal community by deploying a remote camera near the center of each sample unit. Avian data have not yet been analyzed. We collected 388,951 photos of 56 species (26 mammalian species). Using Program MARK (White and Burnham 1999), we fit standard occupancy models (MacKenzie et al. 2006) to data for each species in the following manner. First, we fit a base model with parameters for the spruce-fir or lodgepole stratum, percentage of aspen present at the site, canopy cover, shrub cover, amount of down wood, amount of bare ground, and three physiographic variables that collectively account for elevation, moisture accumulation, and solar radiation at each site. The purpose of this model was to account for basic occupancy patterns of each species in the state irrespective of bark beetles. Next, we fit additional parameters to the base model which allowed occupancy to change in a variety of patterns (e.g., linear, quadratic, 3rd order polynomial, or change points when needles drop following an outbreak) in relation to time elapsed since a stand was initially impacted by beetles. We also explored whether there was any interaction between response to beetles and stratum or the severity of the outbreak (percent of trees that were killed). We used Akaike’s Information Criterion (Burnham and Anderson 2002) to assess fit of these various beetle response models, and model-averaged occupancy across the model set (i.e., ‘year since beetle outbreak’ was treated as a group such that parameters for each group could be averaged across all models in the set) to provide a best estimate of response of each species to beetles. Note that because we sampled mobile animals in a continuous landscape, ‘occupancy’ in this case refers to the probability that a species uses the forest stand which the camera was placed. 1 �Mean responses indicate that use of subalpine forest by elk (Figure 2) and mule deer (Figure 3) increased after stands were impacted by bark beetles (either through time or with increasing severity). Use by red squirrels (Figure 4) and coyotes (Figure 5) declined. Use by red fox (Figure 6) and black bears (Figure 7) was mixed as it increased through time after an outbreak, but may have declined with increasing severity. Snowshoe hares (Figure 9), and American martens (Figure 10) were largely unaffected by bark beetles in terms of their use of a stand as a function of beetle outbreaks. Both red squirrels and snowshoe hares used spruce-fir stands more heavily than lodgepole stands. Data from other species was too sparse to support fitting of the suite of models presented here. LITERATURE CITED Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd edition. Springer, New York. MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. Academic Press, Oxford, United Kingdom. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46 Supplement:120-138. Figure 1. Extent of mountain pine beetle (red) and spruce beetle (purple) infestations in spruce/fir (bluegreen) and lodgepole pine (bright green) forests in Colorado, 2014. Data were summarized from USFS Aerial Surveys. 2 �10% Dead , .o 0.9 o.a o.a 0.7 0.7 o.• g o.• il o.s ~ o.s g s •O O • • •-••••••••••• •• • •• ••• ••••• •••• ~ 0.◄ 0.l 90% Dead 1.0 0.9 i I O.◄ _j 0. l · --. 0.2 ~ 0 .2 0.1 0,1 0.0 0.0 10 11 10 11 Years Since Outbreak Years Since Outbreak Figure 2. Elk occupancy (use) of subalpine forest stands in relation to the number of years since initial infestation by bark beetles for lightly (left) and severely (right) impacted areas Dotted lines indicate 95% confidence intervals. Color bar indicates approximately when forest canopy changes from green to red (dead needles) to gray (no needles) following a bark beetle outbreak. 10% Dead 1.0 0.9 0.9 ---·· 0.8 0.8 0.7 g 90% Dead 1.0 0.7 g 0.b il , o.s 0.b il o.s s ~ o., 0.l ...-· • ••••·•··················- .. O. ◄ 0.l 0.2 • 0.2 0.1 o., 0.0 0.0 10 10 11 Years Since Outbreak 11 Years Since Outbreak Figure 3. Mule Deer occupancy occupancy (use) of subalpine forest stands in relation to the number of years since initial infestation by bark beetles for lightly (left) and severely (right) impacted areas. 10% Dead ,.o - ....= .. J.9 J.9 , .a ).8 J.7 ).7 r; ,,.s.• r; ~ ,.• 90% Dead 1.0 SpnJoo-Fir Lodgepol8 " - Spruce-Fir - Lodge~t-~ . ·- •• ·····--·--·-· ).6 ).5 ~ ,.• ,., ,., J.3 ,., , .1 ).1 ,.o ,.o 10 10 11 Years Since Outbreak 11 Years Since 0Ltbteak Figure 4. Red squirrel occupancy (use) of subalpine forest stands in relation to the number of years since initial infestation by bark beetles for lightly (left) and severely (right) impacted areas. 3 �HJ% Dead 1_0 0-1 g o_. I 90% Dead 1_0 o_, o_, 0.9 0,8 -.. ..... 0-7 g o.~ - a 0.5 o_, o_, o_, so., o_, 0_2 0.2 ••••·•··· 0_1 0_1 00 ••••••• ····· 0.0 11 10 11 10 ■■■■■■■.:::==v.=.=,,:s;:1nc =•=0u =,br:e:a:k=====--- Years Since Outbreak Figure 5. Coyote occupancy (use) of subalpine forest stands in relation to the number of years since initial infestation by bark beetles for lightly (left) and severely (right) impacted areas. 10% Dead 1_0 0,9 o_s 0,1 o_, o_, .. •·· .- g 0.6 ; g a 0. 5 05 so., 0,) 0.) 0,1 / 0.6 ~ o., 0_2 90% Dead 1_0 0_9 ·--~----·-···· ········ ..• - ... ·········-- 0.2 - - - - - :··········· : - -:~ -··· ••• ------------------ 0, 1 o_o o_o 11 10 10 Years Since Outbreak 11 Years Since Outbreak Figure 6. Red Fox occupancy (use) of subalpine forest stands in relation to the number of years since initial infestation by bark beetles for lightly (left) and severely (right) impacted areas. 10%Dead 1.0 90% Dead ,_o 09 09 0.8 0.8 0-1 ~ 0 .6 ~ 0 .6 g. 0.5 g. 0.5 ~ 0 .4 ~ 0.4 O.J o_, 0.2 0-2 0,1 ._, o.o 0 -0 10 11 10 YeArs Sioce Outbreak 11 Years Since Outbreak Figure 7. Black bear occupancy (use) of subalpine forest stands in relation to the number of years since initial infestation by bark beetles for lightly (left) and severely (right) impacted areas. 4 �HJ% Dead 1.0 0.9 0.8 o.a 0.7 0.7 g 0.6 g 0.6 ~ 0.5 ~ 0.4 90% Dead 1.0 0.9 0.5 ~ o., ··············,······························"···•············ ~ 0.3 0.l 0.2 0.2 0.1 0.1 $pruc:e-Fw - L- ..-•·· ····••••• ...•···· ••••••••• - ·-..................................................... ··-... 0.0 0.0 10 10 11 11 Years Since Outbreak Years Since Outbreak Figure 8. Snowshoe Hare occupancy (use) of subalpine forest stands in relation to the number of years since initial infestation by bark beetles for lightly (left) and severely (right) impacted areas. HJ% Dead 1.0 g 0.9 0.8 0.8 0.7 0,7 O.b g 0.6 a. a 0.5 0., ,a. 0.5 ~ o., 90% Dead 1.0 0.9 ···············-·········· ......... ·---···· •••••••••••• ~ 0.l 0.) 0.2 0.2 0,1 0,1 0.0 0.0 10 11 10 Years Since Outbreak 11 Years Since Outbreak Figure 9. American marten occupancy (use) of subalpine forest stands in relation to the number of years since initial infestation by bark beetles for lightly (left) and severely (right) impacted areas. 5 �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Mammal and breeding bird response to bark beetle outbreaks in Colorado Period Covered: July 1, 2016  June 30, 2017 Principal Investigators: Jacob S. Ivan, Jake.Ivan@state.co.us; Amy Seglund, Amy.Seglund@state.co.us ; All information in this project summary is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the principal investigator. Manipulation of these data beyond that contained in this summary is discouraged. ABSTRACT: Mountain pine beetle (Dendroctonus ponderosae) and spruce beetle (Dendroctonus rufipennis) infestations have reached epidemic levels in Colorado, impacting over 4 million acres since the initial outbreak in 1996. Though bark beetles are native to Colorado and periodic infestations are considered a natural ecological process, the geographic scale of their impact and simultaneous infestation within multiple forest systems has never been observed. This historic outbreak is having significant impacts on composition and structure of forest stands that will propagate for decades into the future. Here we used occupancy estimation to determine statewide wildlife response to bark beetle outbreaks, as mediated by changes in forest structure. Surveys were conducted during the summers of 2013 and 2014. We randomly sampled 150 Engelmann spruce (Picea engelmanni)-subalpine fir (Abies lasiocarpa) sites and 150 sites consisting mostly of lodgepole pine (Pinus contorta) or lodgepole pine mixed with other conifers. For both strata, sampling covered conditions ranging from sites that were not impacted by bark beetles to those that were impacted by beetles more than a decade ago. At each 1-km2 (0.4 mi2) site, we sampled the breeding bird community using the Rocky Mountain Bird Observatory’s protocol for “Integrated Monitoring in Bird Conservation Regions” (Hanni et al. 2014). We sampled the mammal community by deploying a remote camera near the center of each sample unit. Avian data have not yet been analyzed. We collected 388,951 photos of 56 species (25 mammalian species). Using Program MARK (White and Burnham 1999), we fit standard occupancy models (MacKenzie et al. 2006) to data for each species in the following manner. First, we identified the best-fitting ‘base’ model from among all combinations of 0-4 of the following variables: spruce-fir or lodgepole stratum, percentage of aspen present at the site, canopy cover, shrub cover due to deciduous species, shrub cover due to conifer species, shrub height, amount of down wood, amount of bare ground, and four physiographic variables that collectively account for elevation, topographic position (e.g., valley bottom, ridge top), moisture accumulation, and solar radiation at each site. The purpose of this model was to account for basic occupancy patterns of each species in the state irrespective of bark beetles. Next, we fit additional parameters to the base model which allowed occupancy to change in a variety of patterns (e.g., linear, quadratic, 3rd order polynomial, or change points when needles drop following an outbreak) in relation to time elapsed since a stand was initially impacted by beetles. We also explored whether there was any interaction between response to beetles and stratum or the severity of the outbreak (percent of trees that were killed). We used Akaike’s Information Criterion (Burnham and Anderson 2002) to assess fit of these various beetle response models, and model-averaged occupancy across the model set (i.e., ‘year since beetle outbreak’ was treated as a group such that parameters for each group could be averaged across all models in the set) to provide a best estimate of response of each species to beetles. As per our hypotheses, results suggest that ungulate species are positively associated with bark beetle outbreaks, although the shape and nature of their responses was variable (Fig. 1). Also not 2 �surprisingly, granivore species comprised the majority of species that were negatively associated with bark beetle outbreaks, although again the magnitude and shape of responses was variable (Fig. 2). We did not detect any response to bark beetles by American marten or black bears (Fig. 3). Snowshoe hares did not follow expectation either, as their use did not markedly increase through time with increasing development of a dense understory (Fig. 3). Both red squirrels and snowshoe hares used spruce-fir stands more heavily than lodgepole stands. 10% Dead ... ... ... ... •• Q) V, L< •• 1.1 1.1 ~ :.0 ro .0 0 ~ a... .r ---- . ... ~ .. .~ · ., ...~ .. • . •• • ::> .._ ,. 0 50% Dead • .. •• . ••... ..... •• ... • " ; " u ... • i u u ...., . ~ , , ~·.. • " ... .. u , r- ..." ~ •• , u ... ...•• 90% Dead • • , u u ...... , . , . ~ ... . . , •• I u • • " Years Since Outbreak Figure 1. Species that exhibited a positive association between use of forested stands and beetle activity (either years since the outbreak occurred, severity, or both). From left to right, panels indicate predicted model-averaged responses for cases where 10%, 50%, and 90% of the overstory in a stand is killed by beetle activity. From top to bottom, panels show response for elk, mule deer, and moose. Probably of use was estimated to vary little between the spruce-fir and lodgepole pine stands, so responses are pooled among strata for these species. Shaded areas represent model-averaged 95% confidence intervals. 3 �,.. 10% Dea<! •• •• ., ., u u • • • ,. .... => . - .. 0 ~ (1) .0 • • • , .. ... " u u •• ·- - - - - ,. • • u ... •• .. . u • ' .. • • ' " .. • • ' " .... . •• ... •• u " .. •'-----,. • •• u • ~ u ... ,, ---- .. " ..••. u • • OU .... • .... • u •• ...•• •• ,. u .. ., e ... a.. ... ,. . .. -----.. .. :a •• .... ~ .,.,,, ... •• rJ) 90% Dead •• ... •• Q) SO% Dead ..•• .. u L?• ,, " I Years Since Outbreak Figure 2. Species that exhibited a negative association between use of forested stands and beetle activity (either years since the outbreak occurred, severity, or both). From left to right, panels indicate predicted model-averaged responses for cases where 10%, 50%, and 90% of the overstory in a stand is killed by beetle activity. From top to bottom, panels show response for red squirrel, golden-mantled ground squirrel, chipmunk spp., and coyote. For red squirrels, use was estimated to vary between the spruce-fir (blue) and lodgepole pine stands (gray); for other species, habitat strata was less important and responses are pooled across habitat types. Shaded areas represent model-averaged 95% confidence intervals. 4 �,.. 10% Dea<! .. •• •• , . u .. • .,, ... e a.. • • • ,. ..., • • " ..... .. .. ---u . • u H u u .. • ,. ., .. • , .. •• ... ... ,. , .... ~ ., :0 ,. Ill .0 u tJ 0 ~ ... . . ,.-----..__ ... ... - .. rJ) ::) ..•• ... ... ., 90% Dead u .. • • , •• Q) 50% Dead .... ... .. u ,. .... , • " ,,; . , •• ... ,, .. .. . u • • " ...•• .. " • .. • • • " .. , u •• u .. U• .. ,, I Years Since Outbreak • • ,, Figure 3. Species that exhibited a little association between use of forested stands and beetle activity (either years since the outbreak occurred, severity, or both). From left to right, panels indicate predicted model-averaged responses for cases where 10%, 50%, and 90% of the overstory in a stand is killed by beetle activity. From top to bottom, panels show response for American marten, black bear, snowshoe hare, and porcupine. For snowshoe hares, and porcupine, use was estimated to vary between the sprucefir (blue) and lodgepole pine stands (gray); for other species, habitat strata was less important and responses are pooled across habitat types. Shaded areas represent model-averaged 95% confidence intervals. 5 �10% Dead .... .. Q) 0 ., .., u z- •• • ,. ... . .... .. •• :.0 (I) •• ------------- ...u •• ,, ' • ' I ' • ' ' •' ' ' ' ' I ..•• u 0 Cl.. ., Ll • .. • . , ....... 1-... ...., ~ .. . • • •• .0 •• •• ... ::::, •• 90% Dead .... ••• •• (JJ 50% Dead u ; ' I Years Since Outbreak Figure 4. Species that exhibited mixed associations between use of forested stands and beetle activity (positive association with YSO but negative with severity, or vice-versa). From left to right, panels indicate predicted model-averaged responses for cases where 10%, 50%, and 90% of the overstory in a stand is killed by beetle activity. From top to bottom, panels show responses for red fox and yellow-bellied marmot. Use was estimated to vary little between the spruce-fir and lodgepole pine stands, so responses are pooled among strata for these species. Shaded areas represent model-averaged 95% confidence intervals. LITERATURE CITED Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd edition. Springer, New York. MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. Academic Press, Oxford, United Kingdom. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populationsof marked animals. Bird Study 46 Supplement:120-138. 6 � https://cpw.cvlcollections.org/files/original/3535671bfccb7eff6f4fe4a7f8f1a314.pdf f704fcba8135a30c7870eba0c6aa7548 PDF Text Text ÿ ÿ ÿ ÿ 23435673ÿ965 ÿ67ÿ �474���ÿ ��������ÿ��������ÿ������ ÿ�!""��#ÿ ÿ $%&'()*+,ÿ('-./0-'ÿ/1ÿ-0/2-3/'ÿ3+('-ÿ+04ÿ('4ÿ-5%)((',-ÿ6/ÿ*3+07'4ÿ1/('-6ÿ-6(%*6%('ÿ('-%,6)07ÿ1(/&ÿ -.(%*'ÿ8''6,'ÿ/%68('+9-ÿ)0ÿ-/%632'-6ÿ�/,/(+4/:ÿ ÿ ;<=>?@ÿB?C<=<@DÿÿÿEFGHÿIJÿKLIMNEFO<ÿPLJÿKLIQÿ ÿ ;=>OR>STGÿUOC<VW>XTW?=VDÿÿÿETY<ÿUCTOJÿETY<ZUCTO[VWTW<ZR?ZFV\ÿ]=>Rÿ^<_Y>=YJÿ]=>RZ^<_Y>=Y[VWTW<ZR?ZFVÿÿ ÿ `44ÿ��35a6b�3ÿ�ÿbc�ÿd53e�fbÿgaa65hÿ�ÿd5�4�a�65hÿ67ÿgie�fbÿb3ÿ�g5bc�5ÿ�j64g6b�3kÿÿ l�35a6b�3ÿm`nÿopqÿrsÿ9trulvwsxÿpyÿztpqsxÿ{�bc3gbÿd�5a��3ÿ3�ÿbc�ÿd5�f�d64ÿ �j�b�|6b35kÿÿm6�dg46b�3ÿ3�ÿbc��ÿ76b6ÿi�h37ÿbc6bÿf3b6��7ÿ�ÿbc�ÿgaa65hÿ�ÿ7�f3g56|�7kÿÿÿ ÿ ÿ }~€‚ƒ€ÿ„ƒ€ÿ† ‡ÿˆ‰Š‹ŒŽŽ‹‘’ÿ”Ž‹ŒŠŽ’•Š–ÿ‚€—ÿ˜„™šÿ† ‡ÿˆ‰Š‹ŒŽŽ‹‘’ÿ‘›œ”Š‹‹œ’–ÿ ~†™‚˜ÿž‚Ÿÿ™‚šž—ÿ„ƒ— ƒšÿ‡Ÿ‡˜ÿƒ€ÿ¡~‡~™‚—~¢ÿƒ„‚šƒ€£ÿ~Ÿ™ÿ¤ÿƒ‡‡ƒ~€ÿ‚š™˜ÿ~¥ÿ‡~—£„~‡ÿ„ƒ€ÿ ˆ¦œ‹‘’ÿŽ‹Ž•–ÿ‚€—ÿ§€£‡‚€€ÿ˜„™šÿˆ¦œŠ•ÿŠ‹¨Š©ª•‹‹œ–«˜†‚‡„ƒ€ÿ¥ƒ™ÿˆ¬­œŠ’ÿ©•’œŽ•”•–ÿ˜®˜ ˜ÿ ˜ƒ€šÿ̄°°±²ÿÿ³ š€‡®¢ÿ́Ÿ‚€ÿÿ‚‡²ÿˆ1µ¯¶–ÿš~ „‡—ÿ‚ÿ‡‚™£«˜š‚‡ÿ„™~·šÿ~ÿ‚˜˜˜˜ÿƒ„‚š˜ÿ~¥ÿ† ‡ÿ ~†™‚˜ÿ~€ÿ~šš„‚€š®ÿ„‚™€˜ÿ~¥ÿ‚ÿ˜ƒÿ~¥ÿ̧ƒ‡—‡ƒ¥ÿ˜„šƒ˜ÿƒ€ÿ¡~‡~™‚—~²ÿÿ¹ ~€£ÿ~ž™ÿ™˜‡˜¢ÿžƒ˜ÿ ƒ€ƒƒ‚‡ÿ̧~™ÿƒ€—ƒš‚—ÿž‚ÿ˜€~¸˜ž~ÿž‚™ÿˆºŠ”‘’ÿ•ªŠœ•‹‘’–ÿ~šš„‚€š®ÿ™ ‚ƒ€—ÿ‡‚™£‡®ÿ€šž‚€£—ÿ —™ƒ€£ÿžÿ¥ƒ™˜ÿ—š‚—ÿ‚¥™ÿ† ‡˜ÿƒ„‚šÿ‚€ÿ‚™‚²ÿÿ»~¸ Ÿ™¢ÿ˜ÿ~¥ÿ† ‡ÿƒ„‚š—ÿ˜‚€—˜ÿ†®ÿ™—ÿ ˜¼ƒ™™‡˜ÿˆ½•ªœ•’œ‘‘’ÿ¾‘Œ’Ž‹œ‘’–ÿ—™~„„—ÿ‚™—‡®ÿ‚„„™~¿ƒ‚‡®ÿ¤ÿ®‚™˜ÿ„~˜«~†™‚ÿ‚€—ÿž€ÿ ™ ‚ƒ€—ÿ‡~¸¢ÿ˜„šƒ‚‡‡®ÿƒ€ÿ˜‚€—˜ÿž‚ÿž‚—ÿ† €ÿ˜Ÿ™‡®ÿƒ„‚š—²ÿ Àšš„‚€š®ÿ˜ƒ‚ƒ~€ÿƒ˜ÿ‚ÿ˜¥‡ÿ ™ƒšÿ¥~™ÿ—™ ƒ€ƒ€£ÿ†™~‚—ÿšž‚€£˜ÿƒ€ÿ—ƒ˜™ƒ†ƒ~€ÿ~™ÿž‚†ƒ‚ÿ ˜ÿ†®ÿ‚ÿ˜„šƒ˜ÿž™~£žÿƒ ÿ~™ÿ†®ÿ™‚ €²ÿÿ»~¸ Ÿ™¢ÿƒÿƒ˜ÿ‚‡˜~ÿš~‚™˜ÿ‚€—ÿš‚€ÿ†ÿƒ˜‡‚—ƒ€£ÿƒ€ÿ˜~ ÿ š‚˜˜²ÿÿÁ~™ÿƒ€˜‚€š¢ÿƒÿƒ˜ÿ„~˜˜ƒ†‡ÿž‚ÿ˜€~¸˜ž~ÿž‚™ÿ—€˜ƒ®ÿƒ€š™‚˜—ÿ~™ÿ—š™‚˜—ÿƒ€ÿ™˜„~€˜ÿ~ÿ † ‡˜¢ÿ®ÿ~šš„‚€š®ÿ™ ‚ƒ€—ÿ€šž‚€£—ÿ†š‚˜ÿž‚™˜ÿ—ƒ—ÿ€~ÿ¿„‚€—ÿ~™ÿ—š™‚˜ÿžƒ™ÿ—ƒ˜™ƒ†ƒ~€²ÿÿ  š‚˜ÿ˜€~¸˜ž~ÿž‚™˜ÿ‚€—ÿ™—ÿ˜¼ƒ™™‡˜ÿ~£ž™ÿš~ „™ƒ˜ÿ€‚™‡®ÿ̄µµÃÿ~¥ÿžÿ̧ƒ€™ÿ—ƒÿ~¥ÿ¥—™‚‡‡®ÿ ž™‚€—ÿ¡‚€‚—‚ÿ‡®€¿ÿƒ€ÿ¡~‡~™‚—~ÿˆ´Ÿ‚€ÿ‚€—ÿĞ€ÿ1µ¯±–¢ÿƒÿƒ˜ÿƒ„~™‚€ÿ€—™˜‚€—ÿžƒ™ÿ™˜„~€˜ÿ~ÿ ž˜ÿ~†™‚˜ÿ~™ÿž~™~£ž‡®²ÿÿ» ™ÿ̧ ÿ—˜š™ƒ†ÿ‚ÿ¥~‡‡~¸«„ÿ˜—®ÿ—˜ƒ£€—ÿ~ÿ„™~Ÿƒ—ÿ‚ÿ~™ÿ —‚ƒ‡—ÿ‚˜˜˜˜ €ÿ~¥ÿšž‚€£˜ÿƒ€ÿ—€˜ƒ®ÿ~¥ÿž˜ÿ¸~ÿƒ„~™‚€ÿ„™®ÿ˜„šƒ˜ÿƒ€ÿ™˜„~€˜ÿ~ÿ˜„™šÿ † ‡ÿ~†™‚˜ÿƒ€ÿš~™ÿ‡®€¿ÿž‚†ƒ‚ÿƒ€ÿžÿÀÿÅ‚€ÿ}~€‚ƒ€˜²ÿÿÀ™ÿ˜‚ „‡ƒ€£ÿ˜™‚£ƒš‚‡‡®ÿƒ€š‡——ÿ ˜ƒ˜ÿ̧ƒžÿ‚€—ÿ̧ƒž~ÿ‚ÿ˜™~€£ÿ¥ƒ™ÿ€—™˜~™®¢ÿƒ€ÿ‚——ƒƒ~€ÿ~ÿ˜„™š¢ÿƒ€ÿ~™—™ÿ~ÿ‚˜˜˜˜ÿžÿƒ„~™‚€šÿ~¥ÿ žƒ˜ÿ„‚™ƒš‡‚™ÿž‚†ƒ‚ÿ¥‚™ÿƒ€ÿ—ƒš‚ƒ€£ÿž‚™ÿ—€˜ƒ®²ÿÿ¹€š—~‚‡ÿŸƒ—€šÿ˜££˜˜ÿž‚ÿ˜†‚‡„ƒ€ÿ¥ƒ™ÿ ‚®ÿ†ÿ‚€ÿƒ„~™‚€ÿ—™ ƒ€‚€²ÿÿ¹——ƒƒ~€‚‡‡®¢ÿ̧ ÿš~€—ÿ˜€~¸˜ž~ÿž‚™ÿ„‡‡˜ÿ‚ÿš‡‚™—ÿ„‡~˜ÿ ƒ —ƒ‚‡®ÿ‚¥™ÿ˜ƒ‚ƒ€£ÿ̧ƒ€™ÿ—€˜ƒ®ÿƒ€ÿ~™—™ÿ~ÿ‚˜˜˜˜ÿ̧žž™ÿžÿ™‡‚ƒ~€˜žƒ„ÿ†¸ €ÿžÿ¸~ÿ ™ƒš˜ÿš~‡—ÿ†ÿ™‡ƒ—ÿ~€ÿ~ÿ~™ÿ¥¥ƒšƒ€‡®ÿ˜‚ „‡ÿ‚€—ÿ~€ƒ~™ÿž‚™ÿ—€˜ƒƒ˜ÿƒ€ÿ¡~‡~™‚—~²ÿÿÿÿÿ Æ ÿ˜—ÿ‚ÿ˜™ƒ˜ÿ~¥ÿÇ´Äÿ‡‚®™˜ÿ—„ƒšƒ€£ÿ¥~™˜ÿš~ „~˜ƒƒ~€¢ÿ˜‚€—ÿ˜™š™ÿ‚€—ÿ‚£¢ÿ˜„™šÿ † ‡ÿ¿€¢ÿ‚€—ÿ~™‚‡ƒ®ÿ~¥ÿ~Ÿ™˜~™®ÿ™ ˜ÿ~ÿƒ—€ƒ¥®ÿ̄Èÿ˜—®ÿ˜ƒ˜ÿž‚ÿš~‡‡šƒŸ‡®ÿš~Ÿ™—ÿ‚ÿ £™‚—ƒ€ÿ~¥ÿ®‚™˜ÿ‡‚„˜—ÿ˜ƒ€šÿƒ€ƒƒ‚‡ÿ† ‡ÿ~†™‚ÿˆÉÄÀ–ÿ‚˜ÿ̧ ‡‡ÿ‚˜ÿŸ‚™ƒ—ÿ™‚ƒ~˜ÿ~¥ÿ˜†‚‡„ƒ€ÿ¥ƒ™ÿ ˆ¹Âʹ–Ëÿ§€£‡‚€€ÿ˜„™šÿˆÌ´§Í–ÿ˜†š‚€~„®²ÿÿÆ ÿ̧ ™ÿš‚™¥‡ÿ~ÿš~€™~‡ÿ¥~™ÿ‚‡‡ÿ~ž™ÿ‚˜„š˜ÿ˜~ÿ‚˜ÿ~ÿ ƒ˜~‡‚ÿžÿ¥¥š˜ÿ~¥ÿÉÄÀÿ‚€—ÿ¥ƒ™ÿ˜†š‚€~„®ÿ~€ÿ„™®ÿ—€˜ƒ®²ÿÿΞ‚ÿƒ˜¢ÿžÿ˜‡š—ÿ˜ƒ˜ÿ̧ ™ÿš~ „‚™‚†‡ÿ ¸ƒžÿ™˜„šÿ~ÿ˜Ÿ™ƒ®ÿ~¥ÿ† ‡ÿƒ„‚šÿˆ¿š„ÿ¥~™ÿš~€™~‡ÿ˜ƒ˜–¢ÿ˜‡~„¢ÿ‚˜„šÿˆ€~™žÿ¥‚šƒ€£–¢ÿ‡Ÿ‚ƒ~€¢ÿ ˜™š™‚‡ÿ˜‚£¢ÿ‡‚šÿ~¥ÿ„™Ÿƒ~˜ÿ‚€‚£ €¢ÿš²ÿÿÿ Æ ÿ˜‚ „‡—ÿ˜€~¸˜ž~ÿž‚™˜ÿ‚ÿ‚šžÿ~¥ÿžÿ̄Èÿ˜—®ÿ˜ƒ˜ÿ˜ƒ€£ÿ‚ÿš~ †ƒ€‚ƒ~€ÿ~¥ÿ‚™«™š‚„™ÿ ‚€—ÿÇÌÄÿ‡ ™®ÿˆ´Ÿ‚€ÿÿ‚‡²ÿ1µ¯Ï¢ÿ́Ÿ‚€ÿÿ‚‡²ÿ1µ¯¤–ÿ¥™~ ÿÐ š †™ÿ1µ¯±N}‚™šžÿ1µ¯Ñ²ÿÿ ㇁‚€~˜‡®¢ÿ̧ ÿš~€—š—ÿ—ƒ˜‚€šÿ˜‚ „‡ƒ€£ÿ˜™Ÿ®˜ÿ~ÿ˜ƒ‚ÿ™—ÿ˜¼ƒ™™‡ÿ—€˜ƒ®ÿ‚ÿžÿ˜‚ ÿ ˜ƒ˜¢ÿ‚‡ž~£žÿž‚ÿ‚€‚‡®˜ƒ˜ÿƒ˜ÿ~€£~ƒ€£ÿ‚€—ÿ€~ÿ„™˜€—ÿž™²ÿÿÆ ÿ˜—ÿž<ÿÒÓ<OV>WHÿ_>WÔÿÕ<G<Ö<W=H×ÿ —‚‚ÿ®„ÿƒ€ÿ̙~£™‚ ÿ}¹³ØÿˆÆžƒÿ‚€—ÿ™€ž‚ ÿ̄°°°¢ÿ́Ÿ‚€ÿÿ‚‡²ÿ1µ¯Ï–ÿ~ÿ~†‚ƒ€ÿ—€˜ƒ®ÿ˜ƒ‚˜ÿ ¥™~ ÿ‚šžÿ~¥ÿžÿ̄Èÿ˜ƒ˜²ÿÿÆ ÿž€ÿ™‚—ÿž˜ÿ—€˜ƒ®ÿ˜ƒ‚˜ÿ‚˜ÿ™˜„~€˜ÿŸ‚™ƒ‚†‡˜ÿ‚€—ÿ¥ƒÿ‡ƒ€‚™ÿ 1ÿ ÿ �ÿ ÿ ÿ ÿ 234567ÿ893 9ÿ8952ÿ83ÿ8578ÿ3 ÿ9�3895757ÿ5�4��ÿ895ÿ56�8�3�79��ÿ�58�55�ÿ7�3�7935ÿ9�5ÿ45�7�8ÿ ��4ÿ���ÿ3ÿ7����3��ÿÿ��5������66�ÿ�5ÿ��8ÿ��ÿ��85�5�8ÿ3�6ÿ�25���ÿ23456ÿ�7ÿ�ÿ�3��8ÿ3�ÿ5�55��5�ÿ895�ÿ ��8ÿ����ÿ��� ÿ7����3��ÿ!"#!ÿ7����3��ÿ��4ÿ838�6ÿ7����3�ÿ7��6ÿ�63�ÿ��89ÿ���ÿ��ÿ�32����8�3�ÿ $%&'ÿ)*+',ÿÿ-)ÿ./)0ÿ12*%2)3/ÿ456789*&%75ÿ:8%&)8%75ÿ�!�;�ÿ# �9�2ÿ��4ÿ!�4573�ÿ<==<�ÿ��4ÿ�5��7�3�ÿ3�ÿ �35�����5�87ÿ83ÿ4�7�5�ÿ�9��9ÿ>����657ÿ�55ÿ2378ÿ�773���854ÿ��89ÿ9�5ÿ45�7�8�ÿÿ?5ÿ�673ÿ��8ÿ�ÿ7�2�65ÿ6��5�ÿ 23456ÿ83ÿ56�85ÿ25��ÿ�566587@�638ÿ��ÿ5��9ÿ78��4ÿ83ÿ578�2�854ÿ7�3�7935ÿ9�5ÿ45�7�8ÿ�3ÿ89�8ÿ78��4�ÿÿ?5ÿ��8ÿ 89575ÿ�56658A�638ÿ234567ÿ��89ÿ�3ÿ��85�5�8�ÿ897ÿ�3���ÿ�ÿ56�8�3�79��ÿ��ÿ�9��9ÿ�ÿ25��ÿ3�ÿB53ÿ�566587ÿ�5ÿ �638ÿ��4���857ÿ�ÿ45�7�8ÿ3�ÿ=ÿ9�57@9��ÿÿÿ C95ÿ�578ÿ��88��ÿ�55865A9���8�8ÿ23456ÿ���6454ÿ3�6ÿ838�6ÿ7����3�ÿ�7ÿ895ÿ7365ÿ5D�6���83 ÿ >����65�ÿÿC95ÿ23456ÿ���64��ÿ3�6ÿ!#"!ÿ7����3�ÿ��7ÿ75�3�4ÿ�578�ÿ��4ÿ895ÿ5��5�8ÿ3�ÿ!#"!ÿ 7��������86ÿ�37�8�>5�ÿÿE34567ÿ89�8ÿ���6454ÿ3�6ÿ���ÿ��4ÿ3�6ÿ��� ÿ7����3�ÿ�55ÿ��8�66ÿ�375�ÿ3ÿ 3�6ÿ76�986ÿ�5885ÿ89��ÿ895ÿ��85�5�8ÿ3�6ÿ23456�ÿ��4���8��ÿ�62378ÿ�3ÿ56�8�3�79��ÿ�58�55�ÿ7�3�7935ÿ 9�5ÿ45�7�8ÿ��4ÿ5�895ÿ3�ÿ89575ÿ258��7�ÿÿC97�ÿ7�85ÿ83ÿ7�85ÿ>���8�3�ÿ7����3��ÿ57�5���66ÿ��ÿ7����3�ÿ ��9��9ÿ�7ÿ6�F56ÿ�ÿ���8�3�ÿ3�ÿ73�67�ÿ23�78 5�ÿ��4ÿ3895ÿ���38��ÿ���837ÿ89�8ÿ45852��5ÿ7�85ÿ�385�8��6�ÿ�7ÿ�ÿ 235ÿ�2�38��8ÿ4�>5ÿ3�ÿ7�3�7935ÿ9�5ÿ45�7�8ÿ89��ÿ895ÿ�2�5ÿ3�ÿ5�7ÿ56��754ÿ7���5ÿ�558657ÿ�2���854ÿ ��ÿ�5�ÿ�G� 57ÿ<HI��ÿÿG���6ÿ���6757ÿ�5ÿ3�3��ÿ��4ÿ��66ÿ���645ÿJ�ÿ��88��ÿ235ÿ�32�6���854ÿ45�7�8ÿ 234567�ÿ<�ÿ��88��ÿ235ÿ��4ÿ4���55�8ÿ�3>���857ÿ5�575�8��ÿ3895ÿ�385�8��66ÿ�2�38��8ÿ�88��857ÿ3�ÿ895ÿ 7�2�654ÿ78��47�ÿ��4ÿ1�ÿ��88��ÿ�5�9854ÿ6��5�ÿ234567ÿ83ÿ��F�3�6545ÿ89�8ÿ895ÿ45�7�8ÿ57�3�75ÿ>����657ÿ �5ÿ8952756>57ÿ578�2�857ÿ��89ÿ5 3ÿ�3�4ÿ8952�ÿ��4ÿ578�2�857ÿ��89ÿ�5885ÿ�5��7�3�ÿ79364ÿ�� ÿ235ÿ �5�98ÿ89��ÿ89375ÿ��89ÿ235ÿ��58���8�ÿÿ C95ÿ�56658A�638ÿ23456ÿ793�54ÿ�ÿ783��ÿ�37�8�>5�ÿ6��5�ÿ�773���8�3�ÿ�58�55�ÿ895ÿ25��ÿ�2�5ÿ3�ÿ �566587ÿ�3�854ÿ�8ÿ�ÿ7�85ÿ���45D�ÿ��4ÿ895ÿ578�2�854ÿ7�3�7935ÿ9�5ÿ45�7�8ÿ�8ÿ89�8ÿ7�85ÿ����2585KÿG� 5ÿL��ÿ �7ÿ9�7ÿ�55�ÿ793��ÿ5675�955ÿ�5���ÿ#5 ÿ��4ÿM575ÿ<=J=�ÿ�ÿÿC9575ÿ57687ÿ7 578ÿ89�8ÿ7�2�6��ÿ7�3�7935ÿ 9�5ÿ�566587�ÿ�9��9ÿ�7ÿ��5D�5�7�>5ÿ��4ÿ5�����5�8ÿ�32��54ÿ83ÿ���8 ��ÿ��4ÿ�366���ÿ9�57�ÿ���ÿ�5ÿ�ÿ75�6ÿ 8336ÿ�3ÿ��45D��ÿ7�3�7935ÿ9�5ÿ45�7�8ÿ��ÿ;363�43�ÿÿG���6ÿ���6757ÿ�5ÿ�5�4��ÿ��4ÿ��66ÿ���645ÿ5A ��88��ÿ895ÿ23456ÿ��85ÿ78��4�4�B��ÿ�56658ÿ4�8�ÿ83ÿ�566587@23�89ÿ83ÿ���3�8ÿ�3ÿ4���5��ÿ�23�87ÿ3�ÿ8�25ÿ 56��754ÿ�58�55�ÿ�95�ÿ�6387ÿ�55ÿ�65�54ÿ��4ÿ�95�ÿ895ÿ�55ÿ�3�854�ÿÿÿÿÿÿÿÿÿÿÿÿ ÿ OPQRSTQUSRÿVPQRWXÿ ÿ #5 �ÿ�ÿY��ÿ��4ÿ��ÿE�ÿM575�ÿ<=J=�ÿZ56�8�3�79��ÿ#58�55�ÿG5��6ÿ�56658ÿ;3�87ÿ��4ÿ��3�7935ÿ[�5ÿ Y5�7�8ÿ��ÿ?5785�ÿ? 32���ÿ\3 ��6ÿ3�ÿ?�646��5ÿE���525�8ÿ]I^J]ILHJ]LJ�ÿ # �9�2�ÿ_�ÿ���ÿ��4ÿY�ÿZ�ÿ!�4573��ÿ<==<�ÿE3456ÿ7565�8�3�ÿ��4ÿ2 68�23456ÿ���55��5^ÿ�ÿ���8���6ÿ ���32�8�3�A895358��ÿ���3��9�ÿ<�4ÿ54�8�3��ÿ����5�ÿ5�ÿ�3F�ÿ5�ÿ�3F�ÿ̀�!�ÿ �>���ÿ\�ÿ���ÿ!�ÿ��ÿ�56�4�ÿZ�ÿ"�ÿC 5D�ÿ��4ÿ��ÿ��ÿ5�F�F�ÿ<=Ja�ÿE�22�6���ÿ57�3�757ÿ83ÿ�9��54ÿ�3578ÿ �3�4�8�3�7ÿ5768��ÿ�32ÿ��Fÿ�55865ÿ38�5�F7ÿ��ÿ895ÿ73895�ÿZ3�FÿE3�8���7�ÿ��37�955ÿ b^5=<1cb�ÿ �>���ÿ\�ÿ���ÿ��4ÿC�ÿE�ÿ�95�F�ÿ<=Jc�ÿ?��85ÿ4�58ÿ��4ÿ9�8��ÿ7��577ÿ3�ÿ;���4�ÿ6�Dÿ��ÿ;363�43�ÿC95ÿ \3 ��6ÿ3�ÿ?�646��5ÿE���525�8ÿa=^J=IbHJ=La�ÿ �>���ÿ\�ÿ���ÿM�ÿ;�ÿ?9�85�ÿ��4ÿC�ÿE�ÿ�95�F�ÿ<=J1�ÿ̀7��ÿ�D�6�� ÿ8565258 ÿ���32�8�3�ÿ83ÿ578�2�85ÿ���2�6ÿ 45�7�8ÿ�32ÿ���8 5A5���8 5ÿ4�8��ÿ��363 ÿbI^a=bHaJc�ÿ �>���ÿ\�ÿ���ÿM�ÿ;�ÿ?9�85�ÿ��4ÿC�ÿE�ÿ�95�F�ÿ<=JI�ÿY5�7�8ÿ��4ÿ4523 ��9ÿ3�ÿ7�3�7935ÿ9�57ÿ��ÿ�5�8�6ÿ ;363�43�ÿ\3 ��6ÿ3�ÿ?�646��5ÿE���525�8ÿ]a^La=HLbI�ÿ ?9�85�ÿM�ÿ;��ÿ��4ÿ_�ÿ��ÿ# �9�2�ÿJbbb�ÿ�3 �2ÿE!Z_^ÿ7 >�>�6ÿ578�2�8�3�ÿ�32ÿ�3�6�8�3�7ÿ3�ÿ 2�F54ÿ���2�67�ÿ#�4ÿ�84ÿIcÿ���6525�8^J<=HJ1a�ÿ ÿ 1ÿ ÿ �ÿ ÿ ÿ ÿ ~litlil * *~~ ;in;;&~ * ~~litlil * ÿ ÿ 234567ÿ89ÿÿ 5 ÿ�37�ÿ��ÿ�ÿ8��ÿ�7�7�7ÿ�ÿ���76ÿ�ÿ46�37�ÿ��ÿ7�6�ÿ�3��7ÿ��65�7ÿ�77�7ÿ�5�67��ÿ3�ÿ�7ÿ��ÿ �5��ÿ �5��3��!ÿ"���6��!ÿ#7�7$�76ÿ%&8'(���5�6ÿ%&8)9ÿÿ*�7ÿ+ÿ,7�76�-$��ÿ�37�ÿ,767ÿ4677�ÿ��67��ÿ ��ÿ,767ÿ��647�ÿ5�3$���7ÿ�ÿ�7ÿ3$7ÿ��ÿ��$��3�4.ÿ7��76�ÿ�37�ÿ3��5667ÿ��76��6ÿ$�6��3ÿ8(88ÿ 7�6�ÿ�63�6ÿ�ÿ��$��3�49ÿÿ ÿ 1.0 High Total Subcanopy Low Total Subcanopy 0.9 0.8 0.7 Ill 0.6 ....(I)Ill 0.5 .c ....... VI :r: 0.4 0.3 0.2 0.1 0.0 _ _ _ _2_ _ _ 3_ _ _ 4 _==5= = = 6= = ~ 7 ==~8= =~ 9= =~1;0= =11 0 ÿ ÿ 234567ÿ%9ÿ �7�7ÿ67��3����3�ÿ�7,77�ÿ���,���7ÿ��67ÿ7��3ÿ��ÿ7�6�ÿ7����7ÿ�3��7ÿ��65�7ÿ�77�7�ÿ �36�ÿ3$���7ÿ�ÿ�379ÿÿ237ÿ�673�3���ÿ��ÿ/�0ÿ"1�ÿ�67ÿ��6ÿ�37�ÿ,3�ÿ�7ÿ�34�7�ÿ�7949!ÿ25�76�ÿ3��7!ÿ 276$3ÿ47���ÿ��ÿ��,7�ÿ�7949!ÿ5�77�76ÿ67���43�ÿ767�!ÿ5���ÿ"677�ÿ4����ÿ���76�7ÿ�5������ÿ���769ÿÿ 8�7ÿ��ÿ�7ÿ3��767��7ÿ�7,77�ÿ��,ÿ��ÿ�34�ÿ�5������ÿ3�ÿ�����3�7ÿ,3�ÿ�ÿ�34�3�3���ÿ3��767��7ÿ3�ÿ ���,���7ÿ��67ÿ7��3ÿ�9&9&ÿ��67�:��ÿ�ÿ��,ÿ�5������.ÿ9&9�ÿ��67�:��ÿ�ÿ�34�ÿ�5�������!ÿ�5ÿ7�6�ÿ�3��7ÿ �5�67��ÿ$��7�ÿ�3�7ÿ3��767��7ÿ3�ÿ��67�:��9ÿÿÿÿÿ Years Since Outbreak 1ÿ ÿ �ÿ ÿ ÿ ÿ 1.0 High ABLA Subcanopy Low ABLA Subcanopy 0.9 0.8 0.7 "' 0.6 .s:: --- 0 .5 V, (IJ :0 I 0 .4 0 .3 0.2 0.1 0 .0 2 0 4 5 6 7 8 9 10 11 ÿ ÿ 234567ÿ89ÿ 77ÿ67��3���3�ÿ�7��77�ÿ�����7ÿ��67ÿ7��3��ÿ��ÿ�7�6�ÿ7���7ÿ�3��7ÿ��65�7ÿ�77�7�ÿ �36��ÿ3�����7ÿ�ÿ�3�79ÿÿ23��7ÿ�673��3��ÿ��ÿ�1�ÿ���ÿ�67ÿ�6ÿ�3�7�ÿ�3��ÿ��7ÿ�34�7��ÿ�7949�ÿ5��76�ÿ!�"7�ÿ 76�3�ÿ#7�"$ÿ��ÿ�7��ÿ�7949�ÿ%�7776ÿ&7 43�ÿ'67��ÿ% �ÿ�677"ÿ#���$ÿ��76(7ÿ�5������ÿ�(76ÿ�ÿ �5���3�7ÿ�36ÿ�')!'$9ÿÿ* �7ÿ����ÿ�5��ÿ�ÿ��7ÿ3��767��7ÿ7�3��7ÿ3�ÿ234567ÿ+ÿ���ÿ�7ÿ7,��3�7ÿ��ÿ')!'ÿ �5������9ÿÿ ÿÿÿÿ ÿ Years Since Outbreak 1.0 - High PIEN Subcanopy Low PIEN Subcanopy 0 .9 0.8 0 .7 - -~------------------- "' 0.6 - .s:: --- 0 .5 V, (IJ '- "' 0 .4 - I 0.3 0 .2 0 .1 - -- 0.0 I I I I I I I I I I I I 0 1 2 3 4 5 6 7 8 9 10 11 ÿ ÿ 234567ÿ.9ÿ 77ÿ67��3���3�ÿ�7��77�ÿ�����7ÿ��67ÿ7��3��ÿ��ÿ�7�6�ÿ7���7ÿ�3��7ÿ��65�7ÿ�77�7�ÿ �36��ÿ3�����7ÿ�ÿ�3�79ÿÿ23��7ÿ�673��3��ÿ��ÿ�1�ÿ���ÿ�67ÿ�6ÿ�3�7�ÿ�3��ÿ��7ÿ�34�7��ÿ��ÿ�7��ÿ��76(7ÿ �5������ÿ�(76ÿ�ÿ/�47����ÿ��65�7ÿ�#�/*$9ÿÿ* �7ÿ����ÿ��7ÿ3��767��7ÿ�7��77�ÿ�ÿ��ÿ�34�ÿ#�/*ÿ �5������ÿ3�ÿ��ÿ��6�4�ÿ����3��7ÿ�3��ÿ�ÿ�34�3�3����ÿ3��767��7ÿ3�ÿ�����7ÿ��67ÿ7��3��ÿ��7���ÿ�67ÿ �7�6�ÿ37��3���ÿ���ÿ(76��ÿ���376���$9ÿÿÿ ÿ Years Since Outbreak 1ÿ ÿ �ÿ ÿ ÿ ÿ 1.0 0.9 0.8 0.7 0.6 IV ..t::. .....VI 0.5 • Cl1 ,._ 0.4 IV ::c hares/ha = 0. 06 18(pellets/plot) 0.3 • • • • • •• • 0.2 0.1 0.0 0 2 R2 = 0.92 4 6 8 10 Pellets/ Cleared Plot 12 14 16 234567ÿ89ÿÿ23� 7ÿ37�6ÿ67�3���3�ÿ�7�77ÿ������7ÿ��67ÿ7�3�ÿ� ÿ�7�ÿ5��76ÿ��ÿ�77�ÿ�76ÿÿ �7�67ÿ��� ÿ��9ÿÿ��7ÿ�7ÿ�5 �34ÿ�37ÿ��5 76�ÿ��7!"ÿ��3��ÿ��7 3��ÿ���ÿ�343��3�� ÿ3�57�7ÿ�ÿ �7ÿ67467��3�9ÿÿ#7��$�ÿ��ÿ�3�ÿ�37ÿ�7�ÿ�ÿ���47ÿ�7ÿ���7"ÿ�5ÿ�7�ÿ675�7ÿ�7ÿ#ÿ�ÿ%9&%"ÿ��3��ÿ3�ÿ �3ÿ�ÿ4��ÿ�3ÿ� ÿ3 3��7�ÿ�ÿ�6�4ÿ�����3�3�ÿ�7�77ÿ�7ÿ3 7'ÿ��7�ÿ�77�(��!ÿ� ÿ7�3��7ÿ ��6��776ÿ�7�3�!9ÿ ÿ ÿ 1ÿ ÿ � https://cpw.cvlcollections.org/files/original/9798d67c32ff84289a24a3c9159f49d2.pdf 4a4f3fb5e2af076a46f6f9e6e6d85c6f PDF Text Text Colorado Division of Parks and Wildlife July 2010–June 2011 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Federal Aid Project No. Colorado 3430 0670 N/A : : : : Division of Parks and Wildlife Mammals Research Lynx Conservation Predicted lynx habitat in Colorado N/A Period Covered: July 1, 2010 – June 30, 2011 Author: J. S. Ivan Personnel: M. Rice, P. Lukacs, T. Shenk (National Park Service), D. Theobald (Colorado State University), E. Odell All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT In an effort to restore a viable population of federally threatened Canada lynx (Lynx canadensis) to the southern portion of their former range, 218 individuals were reintroduced into Colorado from 1999−2006 (Devineau et al. 2010). In 2010, the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) determined that the reintroduction effort met all benchmarks of success, and that a viable, self-sustaining population of Canada lynx had been established (Shenk and Kahn 2010). The purpose of this project was to develop a statewide predictive map of relative lynx use based upon location data collected during the reintroduction period. To build the map, we divided the state into 1.5 km × 1.5 km cells and tallied the number of locations in each cell. We then fit models to these count data using vegetation, elevation, slope, wetness, and degree of human development in each cell as predictor variables. We produced models for both summer and winter habitat use. We found that regardless of season, lynx were positively associated with spruce/fir (Picea engelmannii/Abies lasiocarpa), mixed spruce/fir, aspen (Populus tremuloides), elevation and slope; they were negatively associated with distance to large forest patches. During summer, lynx use of lodgepole pine (Pinus contorta) stands was predicted to increase. Lynx were predicted to avoid montane forest (Douglas-fir [Pseudotsuga menziesii], Ponderosa pine [Pinus ponderosa]), and areas near high traffic volume road segments, especially during summer. These maps of predicted lynx use should aid land managers in prioritizing areas for conservation, development, and resource extraction with respect to potential impacts to lynx and lynx habitat. 21 �WILDLIFE RESEARCH REPORT PREDICTED LYNX HABITAT IN COLORADO JACOB S. IVAN P. N. OBJECTIVE Use location data collected during Canada lynx (Lynx canadensis) reintroduction to build a model of relative use, then apply this model statewide to produce a predictive map of relative lynx use for Colorado. SEGMENT OBJECTIVES 1. Compile and filter raw location data to isolate highest quality lynx locations. 2. Compile spatial data for use as covariates for the model (e.g. vegetation type, elevation, etc). 3. Build a series of candidate models to explain variation on locations across the landscape using covariate data layers. 4. Model-average predictions from all candidate models to produce a maps of predicted relative use for Colorado. INTRODUCTION In an effort to restore a viable population of federally threatened Canada lynx (Lynx canadensis) to the southern portion of their former range, 218 individuals were reintroduced into Colorado from 1999−2006 by the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW], Devineau et al. 2010). In 2010, CPW determined that the reintroduction effort met all benchmarks of success, and that a viable, self-sustaining population of Canada lynx had been established (Shenk and Kahn 2010). Attainment of this goal is a conservation success, but it has also created a series of issues for land management agencies to consider as they plan changes to the landscape. These issues require knowledge of the types of landscapes and forest stands important for reproduction, movement, dispersal, and general home range use by lynx. As a first step toward providing this information, Theobald and Shenk (2011) conducted an analysis to describe the types of areas that were known to be used by re-introduced lynx. Specifically, they used LoCoH (Getz and Wilmers 2004, Getz et al. 2007) methods to create a population-level utilization distribution (UD, a probability surface of lynx occurrence) for lynx in Colorado. They then summarized landscape attributes within the 90% isopleth (i.e., polygon(s) containing 90% of the probability surface) of this UD. This work provides valuable information regarding the types of areas that were known to be used by lynx from 1999 to 2010. By nature of the data collection and research focus, most of this “use” information was derived from core areas in the San Juan Mountains of southwest Colorado and Sawatch Range in the central part of the state. The purpose of the current project is to extend the work of Theobald and Shenk (2011) by producing a map of predicted lynx use on a statewide scale. Such an exercise will identify areas within Colorado that should contain high quality lynx habitat, regardless of whether or not it was used by the sample of radio-telemetered individuals tracked during reintroduction research. Both works have strengths and weaknesses, but together they provide tools for prioritizing areas for conservation, development, and resource extraction with respect to potential impacts to lynx. 22 �METHODS Location Data Location data were collected from reintroduced lynx using 2 types of telemetry devices. All lynx released into Colorado, and those subsequently captured or re-captured, were fitted with a traditional VHF transmitter. VHF data were collected via telemetry from fixed-wing aircraft at approximately weekly intervals when research was ongoing during winter (approximately December – March) and reproductive seasons (May – June), but less often otherwise. Beginning in April 2000, released and captured lynx were outfitted with dual VHF-Argos satellite collars. In addition to sampling via fixed-wing aircraft, the satellite portion of these collars transmitted repeatedly for 12 hours, 1 day per week, year-round. Nearly 40,000 combined locations were collected between VHF and satellite sampling. These data were originally intended for assessing the success of the reintroduction and served CDOW well in estimating survival, productivity, and dispersal. They were not intended for use in constructing a predictive map of habitat use. We used only the best subset of these data following the filters applied by Theobald and Shenk (2011). Specifically, locations obtained during the first 6 months post-release were removed in order to exclude atypical movements made by animals that had not yet settled into home ranges. Next, poor precision satellite data (e.g., Argos location codes A, B, Z, 0 which do not have associated error estimates) were filtered out because they were too unreliable to be informative of lynx habitat use. We minimized dependence among locations (satellite collars transmitted several times per day, and a VHF location could have been obtained during the same day as well) by retaining only the most precise location for each lynx on a given day. When ties occurred, a single location was randomly selected from among the most precise locations. Finally, we discarded all data from lynx that were located fewer than 30 times over the course of the study. Predictor variables After filtering the location data, we assembled raw covariate data. We obtained housing density (HDENS, units per 1000 ha), road density (RDENS, km/km2 − all roads), slope (SLOPE), elevation (ELEV), topographic wetness (TW), distance to high-volume road segments (D10K, annual average daily traffic volume > 10,000 vehicles), and distance to mesic forest patches >50 ha (D50HA) from Theobald and Shenk (2011). We also downloaded vegetation data from the Colorado Vegetation Classification Project (CVCP, Colorado Division of Wildlife, U.S. Department of Interior Bureau of Land Management, U.S. Forest Service. http://ndis.nrel.colostate.edu/coveg/). CVCP is geographically limited to Colorado, but it accurately depicts many vegetation types that may be important to lynx including riparian zones and willow. Other vegetation data sources (i.e., LANDFIRE) have the advantage of a larger spatial extent, but classification of these non-forest vegetation types is not as detailed. We reclassified the 114 vegetation types in CVCP into 17 classes to simplify the number of covariates available for analysis (Appendix 1). Next, we divided the western portion of Colorado into 1.5 km × 1.5 km cells, which corresponds to 1 SD of the error distribution for the most imprecise (satellite) locations retained for analysis, as well as the smallest 90% UD observed for an individual lynx (Theobald and Shenk 2011). We computed the proportion of different vegetation types in each cell as well as mean SLOPE, ELEV, TW, HDENS, RDENS, D10K, and D50HA. We excluded cells with mean elevations <2,438m (8000 ft), assuming such cells do not provide habitat for lynx. This cutoff is consistent with previous literature (McKelvey et al. 2000, Ruediger et al. 2000), and over 99% of locations from our dataset were above 2,438m. We then standardized each covariate using all cells we intended to make predictions for. To maximize precision of parameter estimates and guard against erroneous predictions later on, we computed a correlation matrix between the potential explanatory variables but none were highly correlated (correlation coefficients were all <0.52 for covariates listed here). 23 �Analysis The response variable of interest for our models was the number of locations per individual in each cell, which we sought to predict using landscape attributes of the cells. We only used cells with ≥1 location for the purpose of constructing models. Excluding cells with no locations (zero counts) results in models that reflect relative use by lynx rather than resource selection. Thus in the generation of the model, we avoided delineation of what was available and suitable to lynx but never used (i.e., we avoided decisions regarding how many zero-count cells to include in the dataset and where they should come from on the landscape), which is a criticism of resource selection approaches. Furthermore, given ~10 years of work including weekly locations on hundreds of animals, we argue that nearly all cells in the Core Study Area that were suitable and available included ≥1 lynx location. This approach does, however, warrant the use of zero-truncated probability models to avoid possibly introducing bias in parameter estimates (Zuur et al. 2009, p. 269). In addition, we expected the data to be over-dispersed (variance of the counts was expected to be larger than the mean), we knew the number of locations collected per animal varied considerably, and we anticipated spatial autocorrelation in the residuals. To evaluate these assertions and determine the best model structure for our data, we successively compared the fits of a basic Poisson generalized linear model (GLM), negative binomial GLM, zero-truncated negative binomial (ZTNB), and ZTNB with an offset. We compared the fit of these alternate structures using Akaike’s Information Criterion (AIC, Burnham and Anderson 2002) and found that fitting a basic negative binomial GLM was an improvement over a Poisson (ΔAIC = 700.4), ZTNB was an improvement over the negative binomial (ΔAIC = 6463.0), and ZTNB with an offset provided the best fit (ΔAIC = 53.7). Thus, we used a ZTNB with an offset as the base model structure. We fit all models using the VGAM package (Yee 2010, 2011) in R (R Core Development Team 2011). To assess spatial autocorrelation we computed a variogram using the gstat package (Pebesma 2004) and standardized residuals from a highly parameterized model (including all covariates below; Figure 1). We found minimal autocorrelation, so we proceeded to build ZTNB models absent spatial structure in the error term. Within the general ZTNB model structure, we specified the candidate model set by including combinations of covariates for modeling the mean count for each cell as follows: 1) Lynx are associated with conifer forests and deep snow, and they rely heavily on snowshoe hares. In the Southern Rockies, lynx occur largely in conifer stands within the sub-alpine zone (Aubry et al. 2000). Therefore, we included proportion spruce/fir (SF, Picea engelmannii/Abies lasiocarpa,), mixed spruce/fir (MIXSF, spruce/fir mixed with Douglas-fir [Pseudotsuga menziesii], aspen [Populus tremuloides], and/or lodgepole pine [Pinus contorta], distance to forest patch >50ha (D50HA), ELEV, and SLOPE in every model. We expected positive associations with each of these covariates except D50HA, which we expected to be negative. 2) Research conducted during the reintroduction of lynx into Colorado focused primarily in the southern portion of the state. Lodgepole pine (LODGE) occurs only in the northern portion of the state, so we know relatively little regarding the importance of this vegetation type with respect to habitat use by lynx. Therefore, we included a LODGE effect in some models, but when LODGE entered as a covariate, we also included a LODGE × latitude (NORTH) interaction to attempt to account for the distribution of this forest type in Colorado. Thus, lodgepole pine was allowed to be an important predictor of lynx use (or not) depending on latitude. 3) Vegetation types other than spruce/fir occur in or adjacent to the subalpine zone. We know relatively little about how lynx use these types but they may be important intermittently and/or as travel corridors. Therefore, we also built models that included combinations of montane forest (MONFOR: Douglas-fir, Ponderosa pine [Pinus ponderosa], and mixed Doug-fir/ponderosa pine), aspen (ASPEN), willow (WILLOW), and montane shrub (MONSHB: Gambel oak 24 �[Quercus gambelii], serviceberry [Amelanchier utahensis], and snowberry [Symphoricarpos sp.]). 4) Though lynx are considered a high elevation species, we opted to exclude “alpine” in any model because lynx are forest-dwelling, and there are few opportunities to manage structure of alpine areas, which included both alpine tundra and rock/snow/ice. 5) Lynx are often considered reclusive. Thus, covariates representing human development might be important predictors of habitats used (or not used) by lynx, and we initially considered HDENS, RDENS, and D10K as potential covariates to include in the model set. However, initial modelfitting resulted in HDENS and RDENS having slightly positive effects on lynx locations (but confidence intervals on these slopes were largely centered on zero indicating the effect was negligible), which is probably an artifact of the trapping/collaring effort that often occurred near roads due to logistical considerations. Many cells outside of those used to construct the models had HDENS and RDENS scores that were orders of magnitude above those used to construct the models. Thus, when projected to the entire set of cells covering western Colorado, these models predicted the best lynx habitat in highly developed, urban areas with high road density. Given this implausible result, we excluded HDENS and RDENS from the analysis. We retained D10K because high volume road segments occurred throughout broad areas used by lynx (nearly every state highway has high volume segments) and it did not result in completely implausible results. We expected counts of lynx locations to be positively associated with distance to high traffic volume road segments. 6) TW was excluded from all models after initial model-fitting produced a result similar to HDENS and RDENS. TW was positively associated with lynx locations, which seems reasonable, but when projected to the expanse of western Colorado, the best lynx habitat was predicted in heavily irrigated agricultural areas, residential lawns, and lakes. These features had TW values that were orders of magnitude larger than any forest-dominated cell. Note that this phenomenon, predicting beyond the range of data used to build the model, can be risky, and it may have operated similarly on other variables but went undetected. 7) Lynx often make long-distance movements outside of the winter season, and these movements may include use of many types of vegetation. Therefore, we fit the model set to summer locations (April through October) and then to winter locations (November through March). Seasonal definitions were based on mean daily movement patterns of telemetered lynx (Theobald and Shenk unpublished data). We expected that the association between lynx locations and vegetation types other than SF and MIXSF would vary with season, with more use of these perceived secondary types during summer. In summary, our model set included all combinations of 5 vegetation types (LODGE, MONFOR, ASPEN, WILLOW, MONSHB) and D10K. Each combination was always paired with the base covariates (SF, MIXSF, ELEV, SLOPE, D50HA) listed in 1) above. This resulted in 26 = 64 models. We used Akaike’s Information Criterion (AIC, Burnham and Anderson 2002) to determine which model structures best explained variation in lynx locations, to assess the importance of each covariate, and to model-average predictions of lynx use for each cell across all models. Predictions were defined as the probability of observing at least 10 locations in a cell over a hypothetical 10-year sampling period, which corresponds to an average of 1 location per year over the time frame of the actual data generating process. We color-coded predictions into 10 quantiles for display such that each color represents 10% of the total (i.e., the darkest red represents the predicted best 10% of cells, dark red plus deep orange represent the predicted best 20% of cells, etc.) 25 �RESULTS The final winter dataset consisted of 3,915 locations from 68 individuals (min = 30 locations/lynx, max = 113, mean = 57.6). Winter cell counts ranged from 1 to 29 (mean = 2.3). Summer data consisted of 5,464 locations from 74 individuals (min = 30, max = 178, mean = 73.8). Summer cell counts ranged from 1 to 36 total lynx locations (mean = 2.8). Predicted Winter Use As expected, relative predicted use by lynx during winter months was negatively associated with D50HA and positively associated with SF, MIXSF, ELEV, and SLOPE (Table 1). Of these associations, SF was strongest (largest magnitude and 95% confidence interval [±2×SE] was well away from zero), followed by ELEV, MIXSF, and D50HA, respectively. The parameter estimate for SLOPE was small and its 95% CI substantially overlapped zero in all models. Thus it was not important in explaining variation in predicted habitat use. Of the covariates that were not included in every model, ASPEN was strongly, positively associated with use and was the only effect in this group that was clearly different from zero. MONSHB was negatively associated with predicted lynx use, but evidence for this effect was weak. WILLOW, MONFOR, and D10K were somewhat positively associated with lynx use, but evidence for these effects was relatively weak as well. LODGE and NORTH did not appear in any of the top models (cumulative AIC weights = 0.12). The winter predictive map reflects the strong effect of SF. Arbitrarily defining the top 20% of predictions as high quality lynx habitat, there are 1,869,975 ha of such habitat in Colorado. Most of this is predicted to occur in the southern part of the state in the San Juan, Culebra, and Wet Mountain Ranges (Figure 2). In the central portion of the state, high predicted use is expected in the northern Sawatch and West Elk Ranges, along with Grand Mesa. The Park Range and Flat Tops comprise the best predicted winter lynx habitat farther north (Figure 2). Predicted Summer Use Associations between relative predicted summer use and SF, MIXSF, ELEV, SLOPE, and D50HA were similar to those observed during winter (Table 2). However, the association with SLOPE was much stronger (larger effect and 95% CI indicated clear separation from zero) during summer, possibly due to den site selection and attendance during this time of year. The association with D50HA was slighter stronger as well. Of the covariates not included in every model, MONFOR and MONSHB were negatively associated with lynx locations; LODGE, NORTH, ASPEN, WILLOW, and D10K were positively associated. The effects of MONFOR, ASPEN, and D10K were substantially different from zero based on 95% CIs. Effects of other covariates were not clearly different from zero. The summer predictive map reflects more dispersed predicted use by lynx with LODGE, NORTH, and the LODGE × NORTH interaction playing a larger role (Figure 3). The central and southern Sawatch Range in central Colorado is predicted to have more use than during winter, whereas use on Grand Mesa is predicted to decline. In the northern part of the state, lynx use is predicted to shift more toward the Medicine Bow and Front Ranges. Using the same definition as before, we predict 1,791,675 ha of high quality summer habitat in Colorado. The overlap between high quality summer and winter cells (as arbitrarily defined above) is ~95%. DISCUSSION The data analyzed here were not collected for the purpose of constructing a predictive map and suffer from at least two shortcomings. First, the locations were not precise. We attempted to account for 26 �this imprecision by modeling at a 1.5 km scale, but matching covariates, response variables, and predictions at this scale reduces the clarity of relationships and weakens the modeling process. Second, the bulk of the reintroduction research effort, from which these data originated, was conducted in the southern and central portions of Colorado. Lodgepole pine only occurs in the northern 2/3 of the state, and is dominant there. Thus, predicting lynx habitat use in northern Colorado is difficult because the landscape is very different, yet we have little data available to help model lynx response to that landscape. That is, we are extrapolating beyond the range of covariates used to fit the models, which is tenuous. Caution should be exercised in interpreting results north of I-70. In addition to issues regarding the location data, we also lack important vegetation data that could be crucial in making accurate predictions. Snowshoe hares (Lepus americanus) are tied to forests with dense understory cover throughout their range (Hodges 2000a;b), including Colorado (Dolbeer and Clark 1975, Zahratka and Shenk 2008, Ivan 2011). Given the close tie between hares and lynx, habitat use of the latter should be strongly tied to understory cover as well. However, we have no covariate data for understory. Our models treat all spruce/fir, mixed spruce/fir, and lodgepole forests equally, but the quality of these forests likely varies considerably. Additionally, pine beetle (Dendroctonus ponderosae) and spruce beetle (Dendroctonus rufipennis) epidemics throughout the state are drastically changing the structure and composition of current and future forests. Our predictions are based on forest composition prior to these outbreaks. Despite these weaknesses, the predictive maps constructed here also have a distinct strength in that they were constructed objectively from rigorous mathematical models based on empirical data collected from wild lynx. They are the first such maps for Colorado. Results from this effort confirm relationships that were already known (e.g., lynx are strongly associated with high elevation spruce/fir and mixed spruce/fir forests but avoid lower elevation montane forests and montane shrublands), and highlight others that may be of interest. For instance, we found clear evidence that lynx use was positively associated with ASPEN during both summer and winter. It is unclear what the ecological relationship between the two might be and we have no causal evidence for ASPEN driving lynx use. However, this pattern is not a simple artifact of ASPEN occurring near SF or MIXSF − our preliminary vetting of potential covariates indicated that the correlation between ASPEN and SF or MIXSF was small and negative (-0.15 and -0.14, respectively). We also found evidence that lynx use of lodgepole forests may increase during summer, and that they tend to avoid areas near high traffic volume road segments, especially in summer. The strengths of this analysis and resulting maps merit their inclusion as a tool for making land management decisions. However, inherent weaknesses of the data require the reader to exercise caution when interpreting results. These maps should be viewed as a compliment to expert opinion and existing maps produced by other means. When assessing habitat quality for lynx at a given project site, it is imperative that managers consider current stand characteristics (especially understory) in formulating land use plans or specific management recommendations relative to lynx. LITERATURE CITED Aubry, K. B., G. M. Koehler, and J. R. Squires. 2000. Ecology of Canada lynx in southern boreal forests. Pages 373-396 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, andJ. R. Squires, editors. Ecology and conservation of lynx in the United States. Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Collins, Colorado, USA. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd edition. Springer, New York. 27 �Devineau, O., T. M. Shenk, G. C. White, P. F. Doherty, P. M. Lukacs, and R. H. Kahn. 2010. Evaluating the Canada lynx reintroduction programme in Colorado: patterns in mortality. Journal of Applied Ecology 47:524-531. Dolbeer, R. A., and W. R. Clark. 1975. Population ecology of snowshoe hares in the central Rocky Mountains. Journal of Wildlife Management 39:535-549. Getz, W. M., S. Fortmann-Roe, P. C. Cross, A. J. Lyons, S. J. Ryan, and C. C. Wilmers. 2007. LoCoH: Nonparameteric Kernel Methods for Constructing Home Ranges and Utilization Distributions. Plos One 2. Getz, W. M., and C. C. Wilmers. 2004. A local nearest-neighbor convex-hull construction of home ranges and utilization distributions. Ecography 27:489-505. Hodges, K. E. 2000a. The ecology of snowshoe hares in northern boreal forests. Pages 117-161 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, andJ. R. Squires, editors. Ecology and conservation of lynx in the United States. University Press of Colorado, Boulder, Colorado, USA. _____. 2000b. Ecology of snowshoe hares in southern boreal and montane forests. Pages 163-206 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, andJ. R. Squires, editors. Ecology and conservation of lynx in the United States. University Press of Colorado, Boulder, Colorado, USA. Ivan, J. S. 2011. Density, demography, and seasonal Movement of snowshoe hares in central Colorado. Dissertation, Colorado State University, Fort Collins, Colorado, USA. McKelvey, K. S., K. B. Aubry, and Y. K. Ortega. 2000. History and distribution of lynx in the contiguous United States. Pages 207-264 in L. F. Ruggiero, K. B. Aubry, S. W. Buskirk, G. M. Koehler, C. J. Krebs, K. S. McKelvey, andJ. R. Squires, editors. Ecology and conservation of lynx in the United States. University Press of Colorado, Boulder, Colorado, USA. Pebesma, E. J. 2004. Multivariable geostatistics in S: the gstat package. Computers & Geosciences 30:683-691. R Core Development Team. 2011. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3900051-07-0, URL http://www.R-project.org/. Ruediger, B., J. Claar, S. Gniadek, B. Holt, L. Lyle, S. Mighton, B. Naney, G. Patton, T. Rinaldi, J. Trick, A. Vendehey, F. Wahl, N. Warren, D. Wenger, and A. Williamson. 2000. Canada lynx conservation assessment and strategy. 2nd edition. R1-00-53, U.S. Department of Agriculture, Forest Service, U.S. Department of Interior, Fish and Wildlife Service, Bureau of Land Management, National Park Service, Missoula, Montana, USA. Shenk, T. M., and R. H. Kahn. 2010. The Colorado lynx reintroduction program. Colorado Division of Wildlife. Theobald, D. M., and T. M. Shenk. 2011. Areas of high habitat use from 1999-2010 for radio-collared Canada lynx reintroduced to Colorado. Colorado State University. Yee, T. W. 2010. The VGAM package for categorical data analysis. Journal of Statistical Software 32:134. _____. 2011. VGAM: Vector Generalized Linear and Additive Models. R package version 0.8-3. URL http://CRAN.R-project.org/package=VGAM. Zahratka, J. L., and T. M. Shenk. 2008. Population estimates of snowshoe hares in the southern Rocky Mountains. Journal of Wildlife Management 72:906-912. Zuur, A. F., E. N. Ieno, N. J. Walker, A. A. Saveliev, and G. M. Smith. 2009. Mixed Effects Models and Extensions in Ecology with R. Springer, New York, New York, USA. Prepared by ______________________________________ Jake S. Ivan, Wildlife Researcher 28 �D10K MONSHB WILLOW ASPEN MONFOR LODGE: NORTH NORTH LODGE SLOPE ELEV D50HA MIXSF SF Table 1. Model selection results (top 10 of 64) and parameter estimates (SE) for zero-truncated negative binomial models fit to cell counts of Canada lynx locations collected during winter (November – March) 1999-2010, southwest and central Colorado, USA. AIC ΔAIC AIC Wt. K 0.53 0.15 -1.1 0.24 0.07 (0.07) (0.08) (0.69) (0.11) (0.06) 0.28 0.06 (0.08) (0.04) 4672.1 0.0 0.15 9 0.48 0.13 -1.09 0.29 0.04 (0.06) (0.07) (0.69) (0.11) (0.05) 0.26 (0.08) 4672.9 0.8 0.10 8 0.52 0.14 -1.09 0.21 0.07 (0.07) (0.08) (0.69) (0.11) (0.06) 0.28 0.07 -0.33 (0.08) (0.04) (0.38) 4673.2 1.1 0.09 10 0.53 0.17 -1.12 0.25 0.07 (0.07) (0.08) (0.69) (0.11) (0.06) 0.27 0.06 (0.08) (0.04) 0.08 4673.2 (0.09) 1.1 0.09 10 0.48 0.15 -1.12 0.3 0.04 (0.06) (0.08) (0.69) (0.11) (0.05) 0.25 (0.08) 0.09 4673.8 (0.09) 1.7 0.06 9 0.54 0.16 -1.1 0.27 0.07 (0.07) (0.08) (0.69) (0.13) (0.06) 0.08 0.29 0.06 (0.22) (0.08) (0.04) 4673.9 1.9 0.06 10 0.47 0.12 -1.09 0.27 0.04 (0.07) (0.07) (0.69) (0.11) (0.05) 0.26 (0.08) 4674.1 2.1 0.05 9 0.52 0.16 -1.12 0.22 0.07 (0.07) (0.08) (0.69) (0.11) (0.06) 0.28 0.06 -0.32 0.08 4674.3 (0.08) (0.04) (0.38) (0.09) 2.2 0.05 11 4674.8 2.8 0.04 9 0.08 4675.0 (0.09) 3.0 0.03 11 0.49 0.14 -1.1 0.31 0.04 (0.07) (0.08) (0.69) (0.13) (0.05) 0.05 0.26 (0.22) (0.08) 0.54 0.18 -1.13 0.27 0.07 (0.08) (0.08) (0.69) (0.13) (0.06) 0.08 0.28 0.06 (0.22) (0.08) (0.04) 29 -0.29 (0.37) �-2.75 (0.7) 0.34 0.26 0.11 (0.13) (0.05) (0.11) 0.08 (0.1) 0.24 (0.12) AIC ΔAIC AIC Wt. K 0.2 (0.08) 6684.3 0.0 0.13 13 0.2 (0.08) -0.66 (0.5) 0.2 (0.08) 6684.4 0.1 0.13 14 0.39 0.11 -2.76 0.19 0.24 (0.06) (0.06) (0.67) (0.11) (0.05) -1.81 0.14 (0.39) (0.08) -0.87 (0.51) 0.15 (0.07) 6684.6 0.3 0.11 11 0.41 0.13 -2.77 0.23 0.25 (0.06) (0.06) (0.67) (0.11) (0.05) -1.82 0.13 (0.39) (0.08) 0.15 (0.07) 6685.9 1.6 0.06 10 0.34 0.07 -2.95 (0.05) (0.06) (0.67) -1.84 (0.39) -0.76 (0.49) 0.16 (0.07) 6686.0 1.7 0.06 10 -1.78 0.15 (0.39) (0.08) -0.85 (0.5) 6686.2 1.9 0.05 10 0.2 (0.08) 6686.3 2.0 0.05 14 -0.67 (0.5) 0.19 (0.08) 6686.3 2.0 0.05 15 0.39 0.11 -2.77 0.2 0.24 (0.06) (0.06) (0.67) (0.11) (0.05) -1.81 0.14 0 -0.86 (0.39) (0.08) (0.04) (0.51) 0.15 (0.07) 6686.6 2.3 0.04 12 0.36 0.09 -2.94 0.13 0.25 (0.05) (0.06) (0.67) (0.09) (0.05) -1.86 (0.38) 0.16 (0.07) 6686.8 2.4 0.04 9 0.09 (0.1) 0.25 (0.05) 0.4 0.08 -2.75 0.21 0.25 (0.06) (0.06) (0.67) (0.11) (0.05) -1.65 (0.4) D10K 0.45 0.11 (0.07) (0.07) MONSHB LODGE: NORTH 0.25 -1.65 0.2 (0.12) (0.39) (0.08) WILLOW NORTH 0.08 (0.1) ASPEN LODGE 0.38 0.27 0.13 (0.12) (0.05) (0.11) SLOPE -2.74 (0.7) ELEV D50HA 0.47 0.11 (0.07) (0.07) SF MIXSF MONFOR Table 2. Model selection results (top 10 of 64) and parameter estimates (SE) for zero-truncated negative binomial models fit to cell counts of Canada lynx locations collected during summer (April – October) 1999-2010, southwest and central Colorado, USA. 0.47 0.12 (0.07) (0.07) -2.74 (0.7) 0.37 0.27 0.13 (0.12) (0.05) (0.11) 0.08 (0.1) 0.25 -1.65 0.2 0.01 (0.12) (0.39) (0.08) (0.04) 0.46 0.11 (0.07) (0.07) -2.74 (0.7) 0.33 0.27 0.11 (0.13) (0.05) (0.11) 0.07 (0.1) 0.24 (0.12) -1.65 (0.4) 30 0.2 0.01 (0.08) (0.04) �i I 0.8 - I I • • • • • • • • • • • • • • • 0.6 - Q) 0 C: ro ·;:: ro ..:: E Q) 0.4 - (/) 0.2 - I I I I 10 20 30 40 Distance (km) Figure 1. Variogram contructured using standardizied residuals from a highly parameterized model fit to count data of lynx locations within 1.5km × 1.5km cells, 1999-2011, southwestern and central Colorado. Variance among pairs of points is similar regardless of the distance separating them, indicative of a lack of residual spatial autocorrelation after fitting important covariate effects. Strong evidence of spatial autocorrelation in residuals would result in a graph with small variance between pairs points that are near to each other, and larger variance at greater distances (i.e., a monotonically increasing pattern). 31 �Predicted Lynx Habitat Use (Winter) Probability of observing >10 locations) - 0.000000 - 0.000001 - 0.445108 - 0.445109 - 0 .523365 D D o.523366 - o.557893 o.557894 - o.582060 D o.582061 - 0 .601338 - 0.601339 - 0 .618958 - 0.618959 - 0 .635619 - 0.635620 - 0 .650491 - 0.650492 - 0 .674444 Figure 2. Predicted winter habitat use by Canada lynx in western Colorado. Predictions are probabilities of observing at least 10 locations within a 1.5 × 1.5km cell over a hypothetical 10-year sampling period. Predictions were averaged across 64 models constructed using all combinations of covariates of interest. 32 �Predicted Lynx Habitat Use (Summer) Probability of observing >10 locations) - 0.000000 - 0.000001 - 0.097487 - 0 .097488 - 0.362282 D D D o.362283 - o.518213 o.518214 - o.596715 o.596716- o.632416 - 0.632417 - 0.651318 - 0.651319 - 0.664720 - 0 .664721 - 0.677897 - 0.677898 - 0.712318 Figure 3. Predicted summer habitat use by Canada lynx in western Colorado. Predictions are probabilities of observing at least 10 locations within a 1.5 × 1.5km cell over a hypothetical 10-year sampling period. Predictions were averaged across 64 models constructed using all combinations of covariates of interest. 33 �Appendix I. Raster reclassification of CVCP dataset for use in lynx predictive map analysis. Lynx Reclass Null 2 2 2 1 1 1 1 4 4 8.2 4 4 4 4 4 14 4 4 4 8.2 8.2 8.2 8.2 8.2 8.2 8.1 8.1 8.2 8.2 8.2 8.2 8.2 4 8.2 4 4 4 4 8.2 10 10 8.1 8.2 8.1 8.1 3.1 8.2 10 10 10 CVCP Value 0 1 2 3 4 5 6 7 8 9 10 11 12 13 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 43 44 46 47 48 49 50 51 53 54 55 Description Unclassified Urban/Built Up Residential Commercial Agriculture Land Dryland Ag Irrigated Ag Orchard Rangeland Grass/Forb Rangeland Snakeweed/Shrub Mix Grass Dominated Forb Dominated Grass/Forb Mix Mid-grass Prairie Short-grass Prairie Sand Dune Complex Foothill and Mountain Grasses Disturbed Rangeland Sparse Grass (Blowouts) Shrub/Brush Rangeland Sagebrush Community Saltbush Community Greasewood Sagebrush/Gambel Oak Mix Snakeweed Snowberry Snowberry/Shrub Mix Bitterbrush Community Salt Desert Shrub Community Sagebrush/Greasewood Shrub/Grass/Forb Mix Sagebrush/Grass Mix Rabbitbrush/Grass Mix Sagebrush/Mesic Mtn Shrub Mix Grass/Misc. Cactus Mix Winterfat/Grass Mix Bitterbrush/Grass Mix Grass/Yucca Mix Sagebrush/Rabbitbrush Mix Pinon-Juniper Juniper Gambel Oak Xeric Mountain Shrub Mix Mesic Mountain Shrub Mix Serviceberry/Shrub Mix Upland Willow/Shrub Mix Manzanita PJ-Oak Mix PJ-Sagebrush Mix PJ-Mtn Shrub Mix 34 �Lynx Reclass 10 10 10 10 11 8.1 13 9.1 13 12 9.1 9.1 9.2 13 13 13 9.2 9.2 9.2 13 9.1 13 13 13 13 12 9.2 13 13 14 6 6 1 2 7 7 7 7 7 6 7 7 3.2 3.2 3.2 3.1 3.2 3.1 3.2 3.2 3.2 5 CVCP Value 56 57 58 59 62 63 65 66 67 68 69 70 71 72 73 75 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 96 97 98 99 100 101 102 103 104 105 106 108 109 110 111 112 113 114 Description Sparse PJ/Shrub/Rock Mix Sparse Juniper/Shrub/Rock Mix Juniper/Sagebrush Mix Juniper/Mtn Shrub Mix Aspen Aspen/Mesic Mountain Shrub Mix Ponderosa Pine Englemann Spruce/Fir Mix Douglas Fir Lodgepole Pine Sub-Alpine Fir Spruce/Fir Regeneration Spruce/Lodgepole Pine Mix Bristlecone Pine Ponderosa Pine/Douglas Fir Mix Limber Pine Lodgepole/Spruce/Fir Mix Fir/Lodgepole Pine Mix Douglas Fir/Englemann Spruce Mix Mixed Forest Land Spruce/Fir/Aspen Mix P. Pine/Gambel Oak Mix Ponderosa Pine/Aspen Mix Douglas Fir/Aspen Mix P. Pine/Aspen/Gamble Oak Mix Lodgepole Pine/Aspen Mix Spruce/Fir/Lodgepole/Aspen Mix Ponderosa Pine/Mesic Mtn. Shrub Ponderosa Pine/Aspen/Mesic Mtn. Barren Land Rock Talus Slopes & Rock Outcrops Soil Disturbed Soil Alpine Meadow Alpine Forb Dominated Alpine Grass Dominated Alpine Grass/Forb Mix SubAlpine Shrub Community Snow Subalpine Meadow Subalpine Grass/Forb Mix Riparian Forested Riparian Cottonwood Conifer Riparian Shrub Riparian Willow Exotic Riparian Shrubs Herbaceous Riparian Sedge Water 35 �Colorado Division of Parks and Wildlife July 2011–June 2012 WILDLIFE RESEARCH REPORT State of: Cost Center: Work Package: Task No.: Federal Aid Project No. Colorado 3430 0670 N/A : : : : Division of Parks and Wildlife Mammals Research Lynx Conservation Predicted lynx habitat in Colorado N/A Period Covered: July 1, 2011 – June 30, 2012 Author: J. S. Ivan Personnel: M. Rice, P. Lukacs, T. Shenk (National Park Service), D. Theobald (Colorado State University), E. Odell All information in this report is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the author. Manipulation of these data beyond that contained in this report is discouraged. ABSTRACT In an effort to restore a viable population of federally threatened Canada lynx (Lynx canadensis) to the southern portion of their former range, 218 individuals were reintroduced into Colorado from 1999−2006 (Devineau et al. 2010). In 2010, the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW]) determined that the reintroduction effort met all benchmarks of success, and that a viable, self-sustaining population of Canada lynx had been established (Shenk and Kahn 2010). The purpose of this project was to develop a statewide predictive map of relative lynx use based upon location data collected during the reintroduction period. To build the map, we divided the state into 1.5 km × 1.5 km cells and tallied the number of locations in each cell. We then fit models to these count data using vegetation, elevation, slope, wetness, and degree of human development in each cell as predictor variables. We produced models for both summer and winter habitat use. We found that regardless of season, lynx were positively associated with spruce/fir (Picea engelmannii/Abies lasiocarpa), mixed spruce/fir, aspen (Populus tremuloides), elevation and slope; they were negatively associated with distance to large forest patches. During summer, lynx use of lodgepole pine (Pinus contorta) stands was predicted to increase. Lynx were predicted to avoid montane forest (Douglas-fir [Pseudotsuga menziesii], Ponderosa pine [Pinus ponderosa]), and areas near high traffic volume road segments, especially during summer. These maps of predicted lynx use should aid land managers in prioritizing areas for conservation, development, and resource extraction with respect to potential impacts to lynx and lynx habitat. 36 �WILDLIFE RESEARCH REPORT PREDICTED LYNX HABITAT IN COLORADO JACOB S. IVAN P. N. OBJECTIVE Use location data collected during Canada lynx (Lynx canadensis) reintroduction to build a model of relative use, then apply this model statewide to produce a predictive map of relative lynx use for Colorado. SEGMENT OBJECTIVES 1. Prepare manuscript for submission to Journal of Wildlife Management. INTRODUCTION In an effort to restore a viable population of federally threatened Canada lynx (Lynx canadensis) to the southern portion of their former range, 218 individuals were reintroduced into Colorado from 1999−2006 by the Colorado Division of Wildlife (now Colorado Parks and Wildlife [CPW], Devineau et al. 2010). In 2010, CPW determined that the reintroduction effort met all benchmarks of success, and that a viable, self-sustaining population of Canada lynx had been established (Shenk and Kahn 2010). Attainment of this goal is a conservation success, but it has also created a series of issues for land management agencies to consider as they plan changes to the landscape. These issues require knowledge of the types of landscapes and forest stands important for reproduction, movement, dispersal, and general home range use by lynx. As a first step toward providing this information, Theobald and Shenk (2011) conducted an analysis to describe the types of areas that were known to be used by re-introduced lynx. Specifically, they used LoCoH (Getz and Wilmers 2004, Getz et al. 2007) methods to create a population-level utilization distribution (UD, a probability surface of lynx occurrence) for lynx in Colorado. They then summarized landscape attributes within the 90% isopleth (i.e., polygon(s) containing 90% of the probability surface) of this UD. This work provides valuable information regarding the types of areas that were known to be used by lynx from 1999 to 2010. By nature of the data collection and research focus, most of this “use” information was derived from core areas in the San Juan Mountains of southwest Colorado and Sawatch Range in the central part of the state. The purpose of the current project is to extend the work of Theobald and Shenk (2011) by producing a map of predicted lynx use on a statewide scale. Such an exercise will identify areas within Colorado that should contain high quality lynx habitat, regardless of whether or not it was used by the sample of radio-telemetered individuals tracked during reintroduction research. Both works have strengths and weaknesses, but together they provide tools for prioritizing areas for conservation, development, and resource extraction with respect to potential impacts to lynx. METHODS While this worked was completed in January 2012, the final report was included in revisions to the previous annual report and is not repeated here. We refer the reader to Ivan (2011) for details regarding methods and results from this work. Our intent is to work this report into a manuscript submission to Journal of Wildlife Management by Fall 2012. 37 �SUMMARY As expected, relative predicted use by lynx during winter months was negatively associated with distance to large patches of conifer (D50HA) and positively associated with spruce/fir (SF), mixed spruce/fir (MIXSF), elevation (ELEV) and slope . Of these associations, the relationship with spruce/fir was strongest. Predicted use was also positively associated with topographic wetness and aspen cover. We projected these associations (and other more minor associations included in competing models) onto a map of the state and arbitrarily defined the top 20% of predictions as high quality lynx habitat. There are 1,869,975 ha of such habitat in Colorado. Most of this high quality habitat was predicted to occur in the southern part of the state in the San Juan, Culebra, and Wet Mountain Ranges. In the central portion of the state, high predicted use is expected in the northern Sawatch and West Elk Ranges, along with Grand Mesa. The Park Range and Flat Tops comprised the best predicted winter lynx habitat farther north Associations between relative predicted summer use and SF, MIXSF, ELEV, slope, and D50HA were similar to those observed during winter. However, the associations with D50HA and slope were stronger during summer. We also found positive associations between lodgepole pine, aspen, and distance to high volume road segments. The summer predictive map reflects more dispersed predicted use by lynx with the lodgepole playing a larger role, especially farther north. The central and southern Sawatch Range in central Colorado is predicted to have more use than during winter, whereas use on Grand Mesa is predicted to decline. In the northern part of the state, lynx use was predicted to shift more toward the Medicine Bow and Front Ranges. LITERATURE CITED Ivan, J. S. 2011. Predicted lynx habitat in Colorado. Wildlife Research Report. Colorado Division of Parks and Wildlife, Fort Collins, CO, USA. Pages 21–35. Prepared by ___________________________ Jacob S. Ivan 38 � https://cpw.cvlcollections.org/files/original/12f296ca22f8138b094b66063aad8a4c.pdf 61237c58b17fe6f5592a656309257fac PDF Text Text Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Quantifying loss and degradation of mule deer habitat across western Colorado Period Covered: July 1, 2013 − June 30, 2014 Principal Investigator: Heather E. Johnson, Heather.Johnson@state.co.us Project Collaborators: Sarah E. Reed, Jessica R. Sushinsky, Andy Holland, Trevor Balzer, Jim Garner All information in this project summary is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the principal investigator. Manipulation of these data beyond that contained in this summary is discouraged. In recent decades, mule deer populations have declined across the western U.S., causing wildlife management agencies to seek factors limiting deer performance and strategies to increase their population sizes. The trend of declining mule deer populations has been primarily attributed to loss and degradation of deer habitat, through mechanisms such as urban/exurban development, resource extraction, agriculture, roads and vehicular traffic, fire suppression, and changing patterns in weather and plant productivity. While wildlife managers are well aware that these different factors can negatively affect deer populations, there is no information on their relative or cumulative impacts. In a report to the Colorado state legislature in 2001 titled, “Declining mule deer populations in Colorado: reasons and responses” Gill (2001) concluded that habitat factors had likely taken the greatest toll on deer populations but that there was no information quantifying the extent of habitat loss or deterioration across the state; critical information that is still lacking today. To address this issue, our objective is to conduct the first spatial and temporal analysis of landscape changes that have occurred to mule deer habitat across western Colorado (west of Interstate 25; Fig. 1). Specifically we are 1) mapping and quantifying changes to deer habitat that have occurred over the last ~40 years (in 5-10 year increments) related to residential development, energy development, fire, climate, and plant productivity, 2) calculating the amount of habitat that has been degraded and lost (directly and indirectly) due to these factors on an individual and cumulative basis for each deer data analysis unit (DAU) and within winter and summer ranges of each DAU, and 3) examining whether spatial and temporal changes to habitat conditions may be associated with observed trends in deer recruitment rates. During fiscal year 2013-2014 we completed the first two objectives of this project, and quantified the total area and proportion of deer habitat that was impacted by each land use land cover (LULC) factor, summarized by DAU. While we wanted to conduct these calculations across all LULC types for the past ~40 years, we were limited by the available data. We calculated metrics for climate and wildfire on an annual basis and in 5-year increments. Habitat loss due to residential development was summarized by decade because that is the finest temporal resolution available for the selected data source. Changes to deer habitat were determined on 5-year increments for energy development and annually for vegetation productivity, because collaborators agreed these were the most useful temporal resolutions for these LULC types. A brief summary of the data used to quantify each type of LULC change is described below: • Climate data were acquired from Parameter-elevation Regressions on Independent Slopes Models (PRISM) to quantify changes to precipitation and temperature. This dataset is considered to be one of the highest-quality historical climate datasets currently available, and was summarized at a 800 m spatial scale. From this dataset we calculated annual precipitation, June precipitation, summer precipitation, winter precipitation, and June minimum temperatures. 9 �• • • • Data on energy development were acquired from the Colorado Oil and Gas Conservation Commission. We obtained a spatial dataset representing the point locations of all oil and gas wells statewide and a tabular dataset representing years of well activity. We merged these datasets to produce a database which attributes all wells with the year the wells were drilled or first became active. At 5-year increments, we calculated the cumulative area affected by energy development at three distances: 200 m, 700 m, and 2,700 m. Changes to residential development were mapped and quantified using the Spatially Explicit Regional Growth Model (SERGoM) dataset. This nationwide dataset models housing density by decade at a spatial resolution of 100 m. Changes to deer habitat by DAU were calculated for urban, suburban, exurban, rural and undeveloped housing categories. We quantified plant productivity or “greenness” from the Normalized-Difference Vegetation Index (NDVI), which has been widely used to assess forage quality for deer and other large herbivores. We used NDVI metrics derived from 1 km Advanced Very High Resolution Radiometer (AVHRR) satellite imagery. For each DAU, on an annual basis, we determined the length of the growing season, time peak plant productivity, the rate of “green-up” across the season, and the cumulative area under the curve for the growing season. Data on fire history were obtained from the Monitoring Trends in Burn Severity (MTBS) project of the US Geological Survey and USDA Forest Service. This nationwide dataset maps the boundaries of wildfires as polygons on an annual basis between 1985 and 2010, on a 100 m spatial resolution. Information on changes to deer habitat due to climate, energy development, residential development, plant productivity and wildlife will be 1) distributed to biologists and relevant CPW staff in western Colorado to aid in future DAU planning, and 2) used to assess whether spatial and temporal changes to mule deer habitat are related to deer recruitment, a key measure of deer population performance. Results of this work will benefit wildlife professionals at statewide, regional, and local scales that will be able to use project results to help prioritize habitat enhancement efforts, connect deer population objectives to landscape conditions, identify key areas for habitat protection, provide comments on land-use proposals, develop policies related to land-use in critical deer ranges, and quantify general habitat impacts that are relevant to deer across western Colorado. Figure 1. The area of interest including all deer analysis units west of Interstate 25 in Colorado. ' Colorado state boundary - - 10unit (DAU) Deer analysis Interstate highway 25 - �Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Quantifying loss and degradation of mule deer habitat across western Colorado Period Covered: July 1, 2014 − June 30, 2015 Principal Investigator: Heather E. Johnson, Heather.Johnson@state.co.us Project Collaborators: Sarah E. Reed, Jessica R. Sushinsky, Andy Holland, Trevor Balzer, Jim Garner, Eric Bergman All information in this project summary is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the principal investigator. Manipulation of these data beyond that contained in this summary is discouraged. Background In recent decades, mule deer populations have declined across the western U.S., causing wildlife management agencies to seek factors limiting deer performance and strategies to increase their population sizes. The trend of declining mule deer populations has been primarily attributed to loss and degradation of deer habitat, through mechanisms such as urban/exurban development, resource extraction, roads and vehicular traffic, and changing patterns in weather and plant productivity. Other factors have also been implicated in contributing to deer declines, such as predation, interspecific competition with elk, and disease, but these factors have not been associated with much empirical support. While wildlife managers are well aware that different habitat factors can negatively affect deer populations, there is no information on their relative or cumulative impacts. In a report to the Colorado state legislature in 2001 titled, “Declining mule deer populations in Colorado: reasons and responses” Gill (2001) concluded that habitat factors had likely taken the greatest toll on deer populations but that there was no information quantifying the extent of habitat loss or deterioration across the state; critical information that is still lacking today. To address this issue, we conducted the first spatial and temporal analysis of landscape changes that have occurred to mule deer habitat across western Colorado (west of Interstate 25). Specifically, our objectives were to 1) quantify the annual changes that had occurred across the DAU and within winter and summer ranges relative to residential development, energy development, wildfire, plant productivity and weather conditions, and 2) test for associations between those changes in habitat conditions and deer recruitment. During FY2013-2014 we quantified changes that had occurred within mule deer ranges for each habitat factor (see Johnson et al. 2014), and in FY2014-15 we tested for associations between those factors and patterns in deer recruitment. In this summary we report findings with respect to residential development, energy development and climate conditions using data collected between 1980 and 2010. These habitat factors had consistent data available across this time period, years when major changes in both landscape conditions and deer populations occurred. Methods To quantify changes in residential development, energy development and climate conditions across western Colorado we were limited to coarse data types with high temporal and spatial extents. We tracked changes in residential development using the Spatially Explicit Regional Growth Model dataset (Bierwagen et al. 2010), which estimates changes in areas of rural, exurban, suburban and urban housing units over time (100m resolution). We obtained information on energy development from the Colorado Oil and Gas Conservation Commission, and used the date of first activity to monitor increases in the number of wells over the course of the study. Because the exact impact area for each well was unknown, we calculated areas within deer ranges that were within 2700m of oil and gas wells (100m resolution), 16 �based on Sawyer et al. (2006) that demonstrated mule deer avoidance within that distance. To assess climatic patterns that may influence deer recruitment, we used historic data from the Parameter-elevation Regressions on Independent Slopes Model (www.prism.oregonstate.edu). The model depicts precipitation and temperature on a monthly basis (800m resolution), which we used to calculate several metrics hypothesized to affect recruitment: average June minimum temperature, June precipitation, summer precipitation (May-Sep), average summer maximum temperature (Jun-Aug), winter precipitation (DecMar) and average winter minimum temperature (Dec-Mar). For more detailed information about the data types used in the analysis refer to Sushinsky et al. (2014). To examine the influence of development and climate factors on mule deer, we used recruitment as our response variable. We chose this demographic parameter because it exhibits high temporal and spatial variation, is sensitive to environmental conditions, is minimally influenced by harvest regulations, and is typically the most influential vital rate driving population growth. Our measure of fawn recruitment was fawn ratios collected annually by CPW personnel. Fawn ratios were observed with post-hunt helicopter surveys in each deer DAU in most years. Surveys occurred between 1 December and 15 January; survey data collected in January were considered data from the previous calendar year (the biological birth year of the fawns). During surveys, non-random paths were flown across the winter ranges with the purpose of encountering as many deer as possible. All observed deer were counted and classified as adult females, fawns or males based on body size and antler morphology. Annual ratios of the number of fawns/100 adult females (n = 904 ratios) and the number of males/100 adult females (n = 901 ratios) for each DAU were calculated from classification data. In conducting the analyses, we first estimated changes in habitat conditions across each DAU, winter and summer ranges by fitting linear mixed models with “year” as the explanatory variable and treating DAU as a random intercept to account for repeated measurements over time. We then tested univariate relationships between each habitat variable and recruitment rates (while also testing for lag effects), retaining those variables that had 80% confidence intervals non-overlapping zero. From the remaining variables, we then checked for multicollinearity. If two variables were highly correlated (r >|0.6|) we retained the variable with the higher univariate relationship with recruitment rates (based on tvalues). Our final variable set included total development across the DAU, exurban development on winter range, energy development on winter range, winter precipitation, June minimum temperature, June precipitation, summer precipitation, the male/female ratio, an interaction between June temperature and precipitation, and an interaction between energy development and precipitation on winter range. We used linear mixed models (DAU was the random intercept) to test all subsets of these habitat variables in predicting fawn recruitment. We used model selection to identify the top models and model averaging to estimate standardized and unstandardized coefficients. Results Increases in residential housing were significant for all development classes (rural, exurban, suburban and urban), particularly on mule deer winter ranges (Fig. 1). Between 1980 and 2010, across all DAUs, the proportion of winter range that was associated with residential development (all types) increased by an average 0.25%/year (SE=0.01, range = 0 – 0.85%/year), while on summer range it increased by an average of 0.18% (SE=0.01, range = 0.02 – 0.65%/year). Both winter and summer ranges experienced major increases in rural development, and winter ranges also experienced major increases in exurban development. On average, 23.8% of deer winter ranges overlapped with some form of residential development in 1980 and 31.2% overlapped with development in 2010; on average, 14.0% of deer summer ranges overlapped with development in 1980 and 19.5% in 2010. Changes in development were greatest in the Southwest and Southeast regions, driven by increases in the number of rural housing units. By 2010 between 0.7% (DAU 1) and 66.0% (DAU 29) of DAU winter ranges overlapped with residential development, while between 0.8% (DAU 41) and 46.0% (DAU 34) of summer ranges overlapped with development. On both winter and summer ranges, energy development significantly increased over time, although winter ranges experienced the greatest increase. Between 1980 and 2010, on average, the 17 �proportion of winter range associated with a well within 2700 m increased by an average of 0.24%/year (SE=0.01; range = 0.0 – 1.4%/year), while on summer range it increased by 0.18%/year (SE=0.01, range = 0.0 – 1.9%/year). Across all DAUs the average proportion of winter range within 2700 m of a well was 16.7% in 1980 and 23.8% in 2010. The average proportion of summer range within 2700 m of a well was 9.6% in 1980 and 15.6% in 2010. Rates of energy development differed among regions with the Northwest and Southeast experiencing the highest rates of increase. By 2010, the proportion of deer winter ranges within 2700 m of a well varied among DAUs between 0% (DAUs 14, 18, 25, and 53) and 79% (DAUs 11 and 12), while summer range varied between 0% (DAUs 18 and 25) and 68% (DAU 11). Seasonal temperature metrics significantly increased over time, while seasonal precipitation metrics significantly decreased, with the exception of winter precipitation which displayed no temporal trend. Between 1980 and 2010, models estimated that on average, June mean minimum temperatures increased from 3.91°C to 5.23°C, summer mean maximum temperatures increased from 21.98°C to 22.58°C, winter mean minimum temperatures increased from -10.72°C to -9.84°C, June precipitation decreased from 3.42 cm to 3.00 cm, and summer precipitation decreased from 26.29 cm to 21.42 cm. The only metric that showed a significant difference by region was the change in minimum temperatures in June, which were much higher in Southwest Colorado than any other region of the state (Table 1). The mean fawn ratio across all DAUs over the course of the study was 56.0 fawns/100 adult females (SE=13.6), with mean ratios in different DAUs ranging between 42.9 (SE=7.6; DAU 23) and 76.6 (SE=12.7, DAU 27). Across years, the mean ratio in the Southwest was 50.2 (SE=11.3), in the Southeast was 58.5 (SE=16.9), in the Northwest was 60.3 (SE=12.4) and in the Northeast was 64.6 (SE=14.6; Fig. 2A). Across all DAUs, in 1980 the modeled mean ratio was 65.4 (SE = 1.4) and in 2010 it was 50.4 (SE = 1.3). Over the course of the study, recruitment decreased by an average of 0.5 fawns/100 adult females/year, with the greatest rates of decline in Southwest (-0.66) and Northwest (-0.46) Colorado. Rates of change were highly variable among DAUs. Forty DAUs exhibited declining trends over time while 4 DAUs exhibited slightly increasing trends, with the rates of change varying between 8.50 to 0.15 fawns/100 adult females/year (Fig. 2B). In contrast to fawn ratios, the ratio of adult male/adult female mule deer significantly increased over the course of the study. In 1980 the mean was 13.5 adult males/100 adult females (SE = 1.1) and by 2010 the mean was 34.0 adult males/100 adult females (SE = 1.0). This increase was influenced by conservative buck harvest strategies implemented during the late 1990s. On average the number of adult males/100 adult females increased by 0.68 males/100 adult females/year (SE=0.03). There was no significant difference in the rate of change in male ratios among regions. Fawn ratios generally decreased in association with increasing residential development, energy development, June temperatures, winter precipitation, and male ratios. Fawn ratios increased in association with higher June precipitation, summer precipitation and winter precipitation in the previous year (lag effect). The interaction of June temperature and precipitation indicated that cold, dry weather had the greatest positive correlation with fawn recruitment, while warm, dry weather had the greatest negative correlation with recruitment. The interaction of energy development and precipitation on winter range suggested that winter severity had the strongest association with fawn recruitment when development was minimal. When a greater proportion of the winter range was impacted by energy development, the negative association with winter precipitation dampened. Fawn recruitment was predicted to be highest when both winter precipitation and energy development were low. Standardized coefficients of the main effects suggested that residential development had the strongest association with fawn recruitment (>2 times the magnitude of any other main effect), and fawn ratios were predicted to vary by 16 fawns/100 adult females across the observed range of development values. Energy development had the second strongest association with recruitment, followed closely by the climate variables. Conclusions Our results indicate that declining trends in mule deer recruitment are correlated with increasing residential and energy development on deer ranges, particularly within winter ranges. Recruitment is the 18 �primary demographic parameter responsible for ungulate population growth, and thus, factors that reduce deer productivity have long-term consequences for overall population performance. Comparing the relative magnitude of correlations of human development factors with climate factors, which are wellknown to be important drivers of juvenile survival, we found that residential housing had >2 times the magnitude of association of any other factor, and that the association with energy development was similar to key climate variables. We detected significant relationships between deer recruitment and habitat conditions, but it is important to acknowledge drawbacks of our analysis that limit our inference. For example, the correlations we detected between recruitment and habitat conditions do not demonstrate causation, as we could not experimentally manipulate levels of human development or climate metrics. Additionally, the data sources used in this analysis were coarse, limited to those that were available over extensive spatial and temporal scales. While development factors were associated with declining recruitment, the specific mechanisms responsible for these correlations are largely unknown and will require additional investigation. Finally, it is important to remember that this analysis only examined a few factors affecting deer habitat, but numerous factors have been associated with demographic trends in deer (i.e., predation, disease, competition with native and domestic ungulates, etc). Our findings have key implications for the conservation of mule deer across Colorado. Adequate, high quality winter range has been speculated to be the primary factor limiting mule deer in the state, and our findings generally corroborate this hypothesis. Indeed, development impacts on winter ranges were more strongly correlated with declining recruitment than impacts on summer ranges, and increases in both development types were greater on winter ranges. Our results suggest that expanding residential and energy development on mule deer ranges may not be compatible with the goal of maintaining highly productive deer populations, and that additional development may further reduce recruitment rates, and potentially, population sizes. Additionally, historic mule deer population objectives may be unrealistic given the increased development activity associated with declining fawn recruitment. While additional research is needed on the mechanisms driving the correlation between anthropogenic developments and declining deer recruitment, wildlife professionals should carefully consider changes to the human footprint when specifying long-term population objectives. If healthy mule deer populations are going to be maintained across the state, conservation practitioners, policy-makers, and land-use planners will need to collectively work to ensure that seasonal habitats, particularly winter ranges, are well preserved. Literature Cited Bierwagen, B.G., D.M. Theobald, C.R. Pykec, A. Choated, P. Crothd, J.V. Thomase, and P. Morefield. 2010. National housing and impervious surface scenarios for integrated climate impact assessments. Proceedings of the National Academy of Sciences of the United States of America 107:20887-20892. Johnson, H.E., S.E. Reed, J.R. Sushinsky, A. Holland, T. Balzer, J. Garner, and E. Bergman. 2014. Quantifying loss and degradation of mule deer habitat across western Colorado. Wildlife Research Project Summary. Colorado Parks and Wildlife, Fort Collins, Colorado. Sawyer, H., R.M. Nielson, F. Lindzey, and L.L. McDonald. 2006. Winter habitat selection of mule deer before and during development of a natural gas field. Journal of Wildlife Management 70:396- 403. Sushinsky, J.R., H.E. Johnson, A. Holland, T. Balzer, J. Garner, and S.E. Reed. 2014. Quantifying landuse and land-cover change in mule deer habitat across Western Colorado. Technical report to Colorado Parks and Wildlife. Wildlife Conservation Society, North America Program, Bozeman, Montana. 19 �D-1 .,_, N D-6 • ,, -- ' D-7 % Residential Develop 0.01 - 0.04 N 0.04 -0.12 A 0.12 - 0.14 0.14- 0.17 0.17 - 0 .26 0.26 - 0.29 0.29 - 0.39 0.39 - 0.56 '" D-29 Figure 1. Map of Colorado deer data analysis units (DAUs) and regions (heavy black lines) designated by Colorado Parks and Wildlife. DAU colors represent the average annual rate of increase in residential development between 1980 and 2010. 80.00 75.00 Northeast --------------------- - ........... 70.00 65.00 60.00 7S ............ ____ . . ----- 4S ro 40.00 40 C 35.00 1980 er: ~ 1990 2000 2010 ro u. 70 ss 50 4S 40 35 1980 75.00 Southwest 70,00 6S 60 B so 50.00 45.00 ·,;::; Northwest :: ::---------------------:: --------------------~:. 55.00 0 l 6S.00 ::----------................. --.... ... ......... ............ ...... 60.00 -- 90 80 70 60 1990 2000 2010 50 40 South east ........................................ --------- 30 20 5S,00 ... ... ... ...... ...... ... ...... .................. 100 Fawn Ratio A S0,00 10 45.00 40,00 3S + - - - - ~ - - - ~ - - - ~ 35.00 + - - - - ~ - - ~ ~ - - ~ 1980 1990 1990 2000 2010 1980 2000 2010 Yea r 0 1980 1990 2000 2010 Year Figure 2. Mean temporal trends between 1980 and 2010 in mule deer recruitment in Colorado by a) region and b) deer data analysis unit. 20 � https://cpw.cvlcollections.org/files/original/f1426445e0c41e397bce7e001c1825e4.pdf 9edb463492f30355f1fe80fb579ba6e9 PDF Text Text Colorado Parks and Wildlife WILDLIFE RESEARCH PROJECT SUMMARY Shifting perceptions of risk and reward: temporal and spatial variation in selection for human development by black bears around three urban systems Period Covered: July 1, 2013 − June 30, 2014 Principal Investigator: Heather E. Johnson, Heather.Johnson@state.co.us Project Collaborators: Stewart W. Breck, Sharon Baruch-Mordo, David L. Lewis, Carl W. Lackey, Kenneth R. Wilson, John Broderick, Julie S. Mao, and Jon P. Beckmann All information in this project summary is preliminary and subject to further evaluation. Information MAY NOT BE PUBLISHED OR QUOTED without permission of the principal investigator. Manipulation of these data beyond that contained in this summary is discouraged. As landscapes across the globe rapidly change due to increased human development, there is uncertainty about the behavioral responses of wildlife to these changes given associated shifts in resource availability and risk. Human development typically reduces native foods for animals, but introduces novel anthropogenic foods (crops, livestock, garbage, watered landscaping, etc) along with risks associated with foraging in human-dominated landscapes. The initial response of animals to human development is typically a change in behavior, as animals have been observed to alter patterns of habitat selection, vigilance, daily activities and foraging, often in highly diverse ways. These behavioral responses reflect perceived trade-offs between the benefits of acquiring key resources and the risks associated with human activity. While these trade-offs should be dynamic in space and time as a function of habitat quality, natural food conditions and the physiological states of individuals, little is known about how animals in human-altered landscapes behaviorally adapt to such variation, particularly under varying ecological conditions. Elucidating the behavioral responses of wildlife to human development is particularly important for large carnivores as their home ranges frequently overlap with human infrastructure and activities, and their interactions with people are often a major source of conflict. In many cases, large carnivores avoid people indicating they associate humans with risk. Some carnivores, however, forage within human development on their natural foods or on anthropogenic foods, exploiting resources associated with human infrastructure. Such behavior has been associated with increased human-carnivore conflicts, generating concern over human safety and property, and stymieing conservation efforts for some carnivore species. If wildlife managers are going to be successful at reducing human-carnivore conflicts and promoting public tolerance for these species, they need to understand how these animals are behaviorally responding to increased development, and the conditions that modify their behavior. These concerns are particularly relevant for black bears (Ursus americanus). Bears can readily exploit the wealth of reliable, high-calorie food resources available around residential development (i.e., garbage, fruit trees, livestock), but are also susceptible to increased mortality from vehicle collisions, conflict-related euthanasia, and other human-related factors. Although studies have found that bears perceive risk associated with human activity, human-bear conflicts have generally increased over time, albeit highly variable. As a long-lived species with relatively stable population dynamics, variation in conflict activity is likely a consequence of shifting foraging behavior, not shifting population sizes, as bears reassess trade-offs of using human foods. Factors such as natural food conditions, a bear’s gender, age, physiological state (e.g., reproductive status), or degree of exposure to human activity, may influence 19 �the benefits and risks of foraging in human-dominated landscapes, driving observed variation in conflict activity. To understand how a large carnivore weighs the benefits and risks of using human development, we examined patterns of black bear resource selection in three developed areas in the western US (Aspen [CO], Durango [CO], and Lake Tahoe [NV]). Using data from 109 bears, our objectives were to 1) examine temporal patterns of selection for development within and across years, 2) compare spatial patterns of selection for development across study systems, and 3) identify individual attributes (e.g., age, maternal status) associated with increased selection for development. Using mixed effects resource selection models we found that use of development by bears was similar across study sites, modifying their selection within and across seasons based on changing environmental and physiological conditions (Fig. 1). Results were based on 331851 locations collected May - October; 87,530 locations for Aspen females (14 different bears), 82,272 for Aspen males (29 bears), 152,365 for Durango females (50 bears), and 9,684 for Tahoe females (16 bears). Selection for human development was tied to nutritional demands, as bears increased their use of anthropogenic foods throughout the summer-fall and in years with poor natural food availability (Figs. 1 and 2). Selection also appeared to be related to bear experience, increasing with animal age. While there were general trends in how bears selected for human development across sites, there were also idiosyncratic differences between them. For example, Aspen males, Aspen females, and Tahoe females tended to select for intermediate development densities, while Durango females displayed a bimodal pattern of either selecting for very high or very low development densities (Fig. 1). In Aspen, males selected for intermediate densities of development in both good and poor natural food years (amplifying their selection for development in poor food years), while females avoided areas with high development densities in good natural food years and strongly selected for high development in poor years, particularly during hyperphagia (Fig. 1). Our findings illustrate that for three areas in the western US black bears selected positively for human development, increasing their use of development in years with poor natural food conditions, throughout the summer-fall, and as bears increased in age. These patterns were generally consistent across study systems and over numerous years of data collection, despite variation in individual bear behavior. Such patterns suggest that bears are similarly interpreting the shifting benefits and risks associated with foraging in human-dominated landscapes, as factors such as natural food conditions, physiological state (i.e., hyperphagia), and experience with anthropogenic foods, simultaneously shape their habitat selection decisions. Variation in bear use of development appeared to be primarily tied to nutritional demands, as the benefits of obtaining anthropogenic foods likely outweighed the risks of foraging around human activity when bears needed additional food resources. Results from this study have key implications for bear management. Wildlife agencies often assume that bears exposed to human food will consistently exhibit nuisance behavior, but our results suggest that bear behavior can be highly variable within and across years, and that bears may often use anthropogenic resources as a source of subsidy rather than relying on those resources outright. Because bear populations are notoriously difficult to monitor, wildlife agencies also often assume that increases in human-bear conflicts reflect increases in bear populations. Our work, however, suggests that bear selection for development may be increasing over time, particularly as individuals get older and gain experience with anthropogenic foods. This behavior may then be the source of additional conflicts without an associated increase in population size, a pattern that has been observed for polar bears. As human development continues to permeate bear habitat, and as changes in climate reduce natural foods for bears in some areas, we expect that bear exposure to development and anthropogenic foods will increase as will their selection for these resources. 20 �Figure 1. Black bear probabilities of selection for density of human development from May through October in Aspen (CO), Durango (CO), and Tahoe (NV), USA. Warm colors depict selection during poor natural food years and cooler colors depict selection in good natural food years. Data for bears in Tahoe were not available for years with different natural food conditions. Note: Durango experienced a maximum of 375 human structures/km2, while Aspen and Tahoe had maximum densities of 540 and 660 structures/km2, respectively. Aspen Males Aspen Females Durango Females Tahoe Females ~ 0.9 0.9 0.8 0.8 0.9 0.8 0 0.7 .:; u 0.7 0.7 0.1 0.6 0.6 C 0.9 0.8 41 0.6 0.6 Vl o.s 0.5 o.s i 0.4 0.4 0.4 :0 0.3 0.3 0.3 0.3 !? 0.2 0.2 0.2 0.1 0.1 0.1 0,1 0 0 0 di 0 "' .n 0.. 100 200 300 400 Density of Human Structures/km1 500 0 100 200 300 400 500 0 100 200 300 Density of Hu man Structures/km2 Density of Human Structures/km 2 - Sep! - Oct ! 100 400 100 300 soo Density of Human Structures/km2 Figure 2. Spatial predictions of resource selection from female black bears in Durango, Colorado, for a good (A) and poor (B) natural food year during fall (Oct 1st). 21 600 � Dublin Core The Dublin Core metadata element set is common to all Omeka records, including items, files, and collections. For more information see, http://dublincore.org/documents/dces/. Title A name given to the resource Mammals research pdfs Description An account of the resource Pdfs of mammals research projects by title . Pdfs are linked from this page. Includes federal aid progress reports and non federal aid progress reports. If reports cover more than one year they are combined into one pdf. Dublin Core The Dublin Core metadata element set is common to all Omeka records, including items, files, and collections. For more information see, http://dublincore.org/documents/dces/. Title A name given to the resource Mammals Research Projects pdfs (2011-2020) Rights Information about rights held in and over the resource <a href="http://rightsstatements.org/vocab/NoC-NC/1.0/">No Copyright - Non-Commercial Use Only</a>